diff options
Diffstat (limited to 'arch/x86/kernel')
151 files changed, 10337 insertions, 6783 deletions
diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index d87f09bc5a5..e77b2208372 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -47,8 +47,6 @@ obj-$(CONFIG_X86_TRAMPOLINE) += trampoline.o obj-y += process.o obj-y += i387.o xsave.o obj-y += ptrace.o -obj-$(CONFIG_X86_DS) += ds.o -obj-$(CONFIG_X86_DS_SELFTEST) += ds_selftest.o obj-$(CONFIG_X86_32) += tls.o obj-$(CONFIG_IA32_EMULATION) += tls.o obj-y += step.o @@ -87,6 +85,7 @@ obj-$(CONFIG_VM86) += vm86_32.o obj-$(CONFIG_EARLY_PRINTK) += early_printk.o obj-$(CONFIG_HPET_TIMER) += hpet.o +obj-$(CONFIG_APB_TIMER) += apb_timer.o obj-$(CONFIG_K8_NB) += k8.o obj-$(CONFIG_DEBUG_RODATA_TEST) += test_rodata.o diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index 036d28adf59..60cc4058ed5 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -31,10 +31,12 @@ #include <linux/module.h> #include <linux/dmi.h> #include <linux/irq.h> +#include <linux/slab.h> #include <linux/bootmem.h> #include <linux/ioport.h> #include <linux/pci.h> +#include <asm/pci_x86.h> #include <asm/pgtable.h> #include <asm/io_apic.h> #include <asm/apic.h> @@ -49,6 +51,7 @@ EXPORT_SYMBOL(acpi_disabled); #ifdef CONFIG_X86_64 # include <asm/proto.h> +# include <asm/numa_64.h> #endif /* X86 */ #define BAD_MADT_ENTRY(entry, end) ( \ @@ -60,7 +63,6 @@ EXPORT_SYMBOL(acpi_disabled); int acpi_noirq; /* skip ACPI IRQ initialization */ int acpi_pci_disabled; /* skip ACPI PCI scan and IRQ initialization */ EXPORT_SYMBOL(acpi_pci_disabled); -int acpi_ht __initdata = 1; /* enable HT */ int acpi_lapic; int acpi_ioapic; @@ -91,6 +93,53 @@ enum acpi_irq_model_id acpi_irq_model = ACPI_IRQ_MODEL_PIC; /* + * ISA irqs by default are the first 16 gsis but can be + * any gsi as specified by an interrupt source override. + */ +static u32 isa_irq_to_gsi[NR_IRQS_LEGACY] __read_mostly = { + 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 +}; + +static unsigned int gsi_to_irq(unsigned int gsi) +{ + unsigned int irq = gsi + NR_IRQS_LEGACY; + unsigned int i; + + for (i = 0; i < NR_IRQS_LEGACY; i++) { + if (isa_irq_to_gsi[i] == gsi) { + return i; + } + } + + /* Provide an identity mapping of gsi == irq + * except on truly weird platforms that have + * non isa irqs in the first 16 gsis. + */ + if (gsi >= NR_IRQS_LEGACY) + irq = gsi; + else + irq = gsi_end + 1 + gsi; + + return irq; +} + +static u32 irq_to_gsi(int irq) +{ + unsigned int gsi; + + if (irq < NR_IRQS_LEGACY) + gsi = isa_irq_to_gsi[irq]; + else if (irq <= gsi_end) + gsi = irq; + else if (irq <= (gsi_end + NR_IRQS_LEGACY)) + gsi = irq - gsi_end; + else + gsi = 0xffffffff; + + return gsi; +} + +/* * Temporarily use the virtual area starting from FIX_IO_APIC_BASE_END, * to map the target physical address. The problem is that set_fixmap() * provides a single page, and it is possible that the page is not @@ -310,7 +359,7 @@ acpi_parse_ioapic(struct acpi_subtable_header * header, const unsigned long end) /* * Parse Interrupt Source Override for the ACPI SCI */ -static void __init acpi_sci_ioapic_setup(u32 gsi, u16 polarity, u16 trigger) +static void __init acpi_sci_ioapic_setup(u8 bus_irq, u16 polarity, u16 trigger, u32 gsi) { if (trigger == 0) /* compatible SCI trigger is level */ trigger = 3; @@ -330,7 +379,7 @@ static void __init acpi_sci_ioapic_setup(u32 gsi, u16 polarity, u16 trigger) * If GSI is < 16, this will update its flags, * else it will create a new mp_irqs[] entry. */ - mp_override_legacy_irq(gsi, polarity, trigger, gsi); + mp_override_legacy_irq(bus_irq, polarity, trigger, gsi); /* * stash over-ride to indicate we've been here @@ -354,9 +403,10 @@ acpi_parse_int_src_ovr(struct acpi_subtable_header * header, acpi_table_print_madt_entry(header); if (intsrc->source_irq == acpi_gbl_FADT.sci_interrupt) { - acpi_sci_ioapic_setup(intsrc->global_irq, + acpi_sci_ioapic_setup(intsrc->source_irq, intsrc->inti_flags & ACPI_MADT_POLARITY_MASK, - (intsrc->inti_flags & ACPI_MADT_TRIGGER_MASK) >> 2); + (intsrc->inti_flags & ACPI_MADT_TRIGGER_MASK) >> 2, + intsrc->global_irq); return 0; } @@ -445,7 +495,21 @@ void __init acpi_pic_sci_set_trigger(unsigned int irq, u16 trigger) int acpi_gsi_to_irq(u32 gsi, unsigned int *irq) { - *irq = gsi; + *irq = gsi_to_irq(gsi); + +#ifdef CONFIG_X86_IO_APIC + if (acpi_irq_model == ACPI_IRQ_MODEL_IOAPIC) + setup_IO_APIC_irq_extra(gsi); +#endif + + return 0; +} + +int acpi_isa_irq_to_gsi(unsigned isa_irq, u32 *gsi) +{ + if (isa_irq >= 16) + return -1; + *gsi = irq_to_gsi(isa_irq); return 0; } @@ -473,7 +537,8 @@ int acpi_register_gsi(struct device *dev, u32 gsi, int trigger, int polarity) plat_gsi = mp_register_gsi(dev, gsi, trigger, polarity); } #endif - acpi_gsi_to_irq(plat_gsi, &irq); + irq = gsi_to_irq(plat_gsi); + return irq; } @@ -481,6 +546,26 @@ int acpi_register_gsi(struct device *dev, u32 gsi, int trigger, int polarity) * ACPI based hotplug support for CPU */ #ifdef CONFIG_ACPI_HOTPLUG_CPU +#include <acpi/processor.h> + +static void acpi_map_cpu2node(acpi_handle handle, int cpu, int physid) +{ +#ifdef CONFIG_ACPI_NUMA + int nid; + + nid = acpi_get_node(handle); + if (nid == -1 || !node_online(nid)) + return; +#ifdef CONFIG_X86_64 + apicid_to_node[physid] = nid; + numa_set_node(cpu, nid); +#else /* CONFIG_X86_32 */ + apicid_2_node[physid] = nid; + cpu_to_node_map[cpu] = nid; +#endif + +#endif +} static int __cpuinit _acpi_map_lsapic(acpi_handle handle, int *pcpu) { @@ -539,7 +624,10 @@ static int __cpuinit _acpi_map_lsapic(acpi_handle handle, int *pcpu) goto free_new_map; } + acpi_processor_set_pdc(handle); + cpu = cpumask_first(new_map); + acpi_map_cpu2node(handle, cpu, physid); *pcpu = cpu; retval = 0; @@ -834,29 +922,6 @@ static int __init acpi_parse_madt_lapic_entries(void) extern int es7000_plat; #endif -int __init acpi_probe_gsi(void) -{ - int idx; - int gsi; - int max_gsi = 0; - - if (acpi_disabled) - return 0; - - if (!acpi_ioapic) - return 0; - - max_gsi = 0; - for (idx = 0; idx < nr_ioapics; idx++) { - gsi = mp_gsi_routing[idx].gsi_end; - - if (gsi > max_gsi) - max_gsi = gsi; - } - - return max_gsi + 1; -} - static void assign_to_mp_irq(struct mpc_intsrc *m, struct mpc_intsrc *mp_irq) { @@ -914,13 +979,13 @@ void __init mp_override_legacy_irq(u8 bus_irq, u8 polarity, u8 trigger, u32 gsi) mp_irq.dstirq = pin; /* INTIN# */ save_mp_irq(&mp_irq); + + isa_irq_to_gsi[bus_irq] = gsi; } void __init mp_config_acpi_legacy_irqs(void) { int i; - int ioapic; - unsigned int dstapic; struct mpc_intsrc mp_irq; #if defined (CONFIG_MCA) || defined (CONFIG_EISA) @@ -941,19 +1006,27 @@ void __init mp_config_acpi_legacy_irqs(void) #endif /* - * Locate the IOAPIC that manages the ISA IRQs (0-15). - */ - ioapic = mp_find_ioapic(0); - if (ioapic < 0) - return; - dstapic = mp_ioapics[ioapic].apicid; - - /* * Use the default configuration for the IRQs 0-15. Unless * overridden by (MADT) interrupt source override entries. */ for (i = 0; i < 16; i++) { + int ioapic, pin; + unsigned int dstapic; int idx; + u32 gsi; + + /* Locate the gsi that irq i maps to. */ + if (acpi_isa_irq_to_gsi(i, &gsi)) + continue; + + /* + * Locate the IOAPIC that manages the ISA IRQ. + */ + ioapic = mp_find_ioapic(gsi); + if (ioapic < 0) + continue; + pin = mp_find_ioapic_pin(ioapic, gsi); + dstapic = mp_ioapics[ioapic].apicid; for (idx = 0; idx < mp_irq_entries; idx++) { struct mpc_intsrc *irq = mp_irqs + idx; @@ -963,7 +1036,7 @@ void __init mp_config_acpi_legacy_irqs(void) break; /* Do we already have a mapping for this IOAPIC pin */ - if (irq->dstapic == dstapic && irq->dstirq == i) + if (irq->dstapic == dstapic && irq->dstirq == pin) break; } @@ -978,7 +1051,7 @@ void __init mp_config_acpi_legacy_irqs(void) mp_irq.dstapic = dstapic; mp_irq.irqtype = mp_INT; mp_irq.srcbusirq = i; /* Identity mapped */ - mp_irq.dstirq = i; + mp_irq.dstirq = pin; save_mp_irq(&mp_irq); } @@ -1043,11 +1116,6 @@ int mp_register_gsi(struct device *dev, u32 gsi, int trigger, int polarity) ioapic_pin = mp_find_ioapic_pin(ioapic, gsi); -#ifdef CONFIG_X86_32 - if (ioapic_renumber_irq) - gsi = ioapic_renumber_irq(ioapic, gsi); -#endif - if (ioapic_pin > MP_MAX_IOAPIC_PIN) { printk(KERN_ERR "Invalid reference to IOAPIC pin " "%d-%d\n", mp_ioapics[ioapic].apicid, @@ -1061,7 +1129,7 @@ int mp_register_gsi(struct device *dev, u32 gsi, int trigger, int polarity) set_io_apic_irq_attr(&irq_attr, ioapic, ioapic_pin, trigger == ACPI_EDGE_SENSITIVE ? 0 : 1, polarity == ACPI_ACTIVE_HIGH ? 0 : 1); - io_apic_set_pci_routing(dev, gsi, &irq_attr); + io_apic_set_pci_routing(dev, gsi_to_irq(gsi), &irq_attr); return gsi; } @@ -1121,7 +1189,8 @@ static int __init acpi_parse_madt_ioapic_entries(void) * pretend we got one so we can set the SCI flags. */ if (!acpi_sci_override_gsi) - acpi_sci_ioapic_setup(acpi_gbl_FADT.sci_interrupt, 0, 0); + acpi_sci_ioapic_setup(acpi_gbl_FADT.sci_interrupt, 0, 0, + acpi_gbl_FADT.sci_interrupt); /* Fill in identity legacy mappings where no override */ mp_config_acpi_legacy_irqs(); @@ -1185,9 +1254,6 @@ static void __init acpi_process_madt(void) if (!error) { acpi_lapic = 1; -#ifdef CONFIG_X86_BIGSMP - generic_bigsmp_probe(); -#endif /* * Parse MADT IO-APIC entries */ @@ -1197,8 +1263,6 @@ static void __init acpi_process_madt(void) acpi_ioapic = 1; smp_found_config = 1; - if (apic->setup_apic_routing) - apic->setup_apic_routing(); } } if (error == -EINVAL) { @@ -1269,23 +1333,6 @@ static int __init dmi_disable_acpi(const struct dmi_system_id *d) } /* - * Limit ACPI to CPU enumeration for HT - */ -static int __init force_acpi_ht(const struct dmi_system_id *d) -{ - if (!acpi_force) { - printk(KERN_NOTICE "%s detected: force use of acpi=ht\n", - d->ident); - disable_acpi(); - acpi_ht = 1; - } else { - printk(KERN_NOTICE - "Warning: acpi=force overrules DMI blacklist: acpi=ht\n"); - } - return 0; -} - -/* * Force ignoring BIOS IRQ0 pin2 override */ static int __init dmi_ignore_irq0_timer_override(const struct dmi_system_id *d) @@ -1321,90 +1368,6 @@ static struct dmi_system_id __initdata acpi_dmi_table[] = { }, /* - * Boxes that need acpi=ht - */ - { - .callback = force_acpi_ht, - .ident = "FSC Primergy T850", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "FUJITSU SIEMENS"), - DMI_MATCH(DMI_PRODUCT_NAME, "PRIMERGY T850"), - }, - }, - { - .callback = force_acpi_ht, - .ident = "HP VISUALIZE NT Workstation", - .matches = { - DMI_MATCH(DMI_BOARD_VENDOR, "Hewlett-Packard"), - DMI_MATCH(DMI_PRODUCT_NAME, "HP VISUALIZE NT Workstation"), - }, - }, - { - .callback = force_acpi_ht, - .ident = "Compaq Workstation W8000", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "Compaq"), - DMI_MATCH(DMI_PRODUCT_NAME, "Workstation W8000"), - }, - }, - { - .callback = force_acpi_ht, - .ident = "ASUS P2B-DS", - .matches = { - DMI_MATCH(DMI_BOARD_VENDOR, "ASUSTeK Computer INC."), - DMI_MATCH(DMI_BOARD_NAME, "P2B-DS"), - }, - }, - { - .callback = force_acpi_ht, - .ident = "ASUS CUR-DLS", - .matches = { - DMI_MATCH(DMI_BOARD_VENDOR, "ASUSTeK Computer INC."), - DMI_MATCH(DMI_BOARD_NAME, "CUR-DLS"), - }, - }, - { - .callback = force_acpi_ht, - .ident = "ABIT i440BX-W83977", - .matches = { - DMI_MATCH(DMI_BOARD_VENDOR, "ABIT <http://www.abit.com>"), - DMI_MATCH(DMI_BOARD_NAME, "i440BX-W83977 (BP6)"), - }, - }, - { - .callback = force_acpi_ht, - .ident = "IBM Bladecenter", - .matches = { - DMI_MATCH(DMI_BOARD_VENDOR, "IBM"), - DMI_MATCH(DMI_BOARD_NAME, "IBM eServer BladeCenter HS20"), - }, - }, - { - .callback = force_acpi_ht, - .ident = "IBM eServer xSeries 360", - .matches = { - DMI_MATCH(DMI_BOARD_VENDOR, "IBM"), - DMI_MATCH(DMI_BOARD_NAME, "eServer xSeries 360"), - }, - }, - { - .callback = force_acpi_ht, - .ident = "IBM eserver xSeries 330", - .matches = { - DMI_MATCH(DMI_BOARD_VENDOR, "IBM"), - DMI_MATCH(DMI_BOARD_NAME, "eserver xSeries 330"), - }, - }, - { - .callback = force_acpi_ht, - .ident = "IBM eserver xSeries 440", - .matches = { - DMI_MATCH(DMI_BOARD_VENDOR, "IBM"), - DMI_MATCH(DMI_PRODUCT_NAME, "eserver xSeries 440"), - }, - }, - - /* * Boxes that need ACPI PCI IRQ routing disabled */ { @@ -1537,9 +1500,8 @@ void __init acpi_boot_table_init(void) /* * If acpi_disabled, bail out - * One exception: acpi=ht continues far enough to enumerate LAPICs */ - if (acpi_disabled && !acpi_ht) + if (acpi_disabled) return; /* @@ -1570,9 +1532,8 @@ int __init early_acpi_boot_init(void) { /* * If acpi_disabled, bail out - * One exception: acpi=ht continues far enough to enumerate LAPICs */ - if (acpi_disabled && !acpi_ht) + if (acpi_disabled) return 1; /* @@ -1590,9 +1551,8 @@ int __init acpi_boot_init(void) /* * If acpi_disabled, bail out - * One exception: acpi=ht continues far enough to enumerate LAPICs */ - if (acpi_disabled && !acpi_ht) + if (acpi_disabled) return 1; acpi_table_parse(ACPI_SIG_BOOT, acpi_parse_sbf); @@ -1609,6 +1569,9 @@ int __init acpi_boot_init(void) acpi_table_parse(ACPI_SIG_HPET, acpi_parse_hpet); + if (!acpi_noirq) + x86_init.pci.init = pci_acpi_init; + return 0; } @@ -1624,19 +1587,12 @@ static int __init parse_acpi(char *arg) /* acpi=force to over-ride black-list */ else if (strcmp(arg, "force") == 0) { acpi_force = 1; - acpi_ht = 1; acpi_disabled = 0; } /* acpi=strict disables out-of-spec workarounds */ else if (strcmp(arg, "strict") == 0) { acpi_strict = 1; } - /* Limit ACPI just to boot-time to enable HT */ - else if (strcmp(arg, "ht") == 0) { - if (!acpi_force) - disable_acpi(); - acpi_ht = 1; - } /* acpi=rsdt use RSDT instead of XSDT */ else if (strcmp(arg, "rsdt") == 0) { acpi_rsdt_forced = 1; @@ -1644,6 +1600,10 @@ static int __init parse_acpi(char *arg) /* "acpi=noirq" disables ACPI interrupt routing */ else if (strcmp(arg, "noirq") == 0) { acpi_noirq_set(); + } + /* "acpi=copy_dsdt" copys DSDT */ + else if (strcmp(arg, "copy_dsdt") == 0) { + acpi_gbl_copy_dsdt_locally = 1; } else { /* Core will printk when we return error. */ return -EINVAL; diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c index f9961034e55..82e508677b9 100644 --- a/arch/x86/kernel/acpi/sleep.c +++ b/arch/x86/kernel/acpi/sleep.c @@ -162,8 +162,6 @@ static int __init acpi_sleep_setup(char *str) #endif if (strncmp(str, "old_ordering", 12) == 0) acpi_old_suspend_ordering(); - if (strncmp(str, "sci_force_enable", 16) == 0) - acpi_set_sci_en_on_resume(); str = strchr(str, ','); if (str != NULL) str += strspn(str, ", \t"); diff --git a/arch/x86/kernel/alternative.c b/arch/x86/kernel/alternative.c index de7353c0ce9..70237732a6c 100644 --- a/arch/x86/kernel/alternative.c +++ b/arch/x86/kernel/alternative.c @@ -7,6 +7,8 @@ #include <linux/mm.h> #include <linux/vmalloc.h> #include <linux/memory.h> +#include <linux/stop_machine.h> +#include <linux/slab.h> #include <asm/alternative.h> #include <asm/sections.h> #include <asm/pgtable.h> @@ -192,7 +194,7 @@ static void __init_or_module add_nops(void *insns, unsigned int len) } extern struct alt_instr __alt_instructions[], __alt_instructions_end[]; -extern u8 *__smp_locks[], *__smp_locks_end[]; +extern s32 __smp_locks[], __smp_locks_end[]; static void *text_poke_early(void *addr, const void *opcode, size_t len); /* Replace instructions with better alternatives for this CPU type. @@ -205,7 +207,7 @@ void __init_or_module apply_alternatives(struct alt_instr *start, struct alt_instr *end) { struct alt_instr *a; - char insnbuf[MAX_PATCH_LEN]; + u8 insnbuf[MAX_PATCH_LEN]; DPRINTK("%s: alt table %p -> %p\n", __func__, start, end); for (a = start; a < end; a++) { @@ -223,6 +225,8 @@ void __init_or_module apply_alternatives(struct alt_instr *start, } #endif memcpy(insnbuf, a->replacement, a->replacementlen); + if (*insnbuf == 0xe8 && a->replacementlen == 5) + *(s32 *)(insnbuf + 1) += a->replacement - a->instr; add_nops(insnbuf + a->replacementlen, a->instrlen - a->replacementlen); text_poke_early(instr, insnbuf, a->instrlen); @@ -231,37 +235,41 @@ void __init_or_module apply_alternatives(struct alt_instr *start, #ifdef CONFIG_SMP -static void alternatives_smp_lock(u8 **start, u8 **end, u8 *text, u8 *text_end) +static void alternatives_smp_lock(const s32 *start, const s32 *end, + u8 *text, u8 *text_end) { - u8 **ptr; + const s32 *poff; mutex_lock(&text_mutex); - for (ptr = start; ptr < end; ptr++) { - if (*ptr < text) - continue; - if (*ptr > text_end) + for (poff = start; poff < end; poff++) { + u8 *ptr = (u8 *)poff + *poff; + + if (!*poff || ptr < text || ptr >= text_end) continue; /* turn DS segment override prefix into lock prefix */ - text_poke(*ptr, ((unsigned char []){0xf0}), 1); + if (*ptr == 0x3e) + text_poke(ptr, ((unsigned char []){0xf0}), 1); }; mutex_unlock(&text_mutex); } -static void alternatives_smp_unlock(u8 **start, u8 **end, u8 *text, u8 *text_end) +static void alternatives_smp_unlock(const s32 *start, const s32 *end, + u8 *text, u8 *text_end) { - u8 **ptr; + const s32 *poff; if (noreplace_smp) return; mutex_lock(&text_mutex); - for (ptr = start; ptr < end; ptr++) { - if (*ptr < text) - continue; - if (*ptr > text_end) + for (poff = start; poff < end; poff++) { + u8 *ptr = (u8 *)poff + *poff; + + if (!*poff || ptr < text || ptr >= text_end) continue; /* turn lock prefix into DS segment override prefix */ - text_poke(*ptr, ((unsigned char []){0x3E}), 1); + if (*ptr == 0xf0) + text_poke(ptr, ((unsigned char []){0x3E}), 1); }; mutex_unlock(&text_mutex); } @@ -272,8 +280,8 @@ struct smp_alt_module { char *name; /* ptrs to lock prefixes */ - u8 **locks; - u8 **locks_end; + const s32 *locks; + const s32 *locks_end; /* .text segment, needed to avoid patching init code ;) */ u8 *text; @@ -390,6 +398,27 @@ void alternatives_smp_switch(int smp) mutex_unlock(&smp_alt); } +/* Return 1 if the address range is reserved for smp-alternatives */ +int alternatives_text_reserved(void *start, void *end) +{ + struct smp_alt_module *mod; + const s32 *poff; + u8 *text_start = start; + u8 *text_end = end; + + list_for_each_entry(mod, &smp_alt_modules, next) { + if (mod->text > text_end || mod->text_end < text_start) + continue; + for (poff = mod->locks; poff < mod->locks_end; poff++) { + const u8 *ptr = (const u8 *)poff + *poff; + + if (text_start <= ptr && text_end > ptr) + return 1; + } + } + + return 0; +} #endif #ifdef CONFIG_PARAVIRT @@ -552,3 +581,62 @@ void *__kprobes text_poke(void *addr, const void *opcode, size_t len) local_irq_restore(flags); return addr; } + +/* + * Cross-modifying kernel text with stop_machine(). + * This code originally comes from immediate value. + */ +static atomic_t stop_machine_first; +static int wrote_text; + +struct text_poke_params { + void *addr; + const void *opcode; + size_t len; +}; + +static int __kprobes stop_machine_text_poke(void *data) +{ + struct text_poke_params *tpp = data; + + if (atomic_dec_and_test(&stop_machine_first)) { + text_poke(tpp->addr, tpp->opcode, tpp->len); + smp_wmb(); /* Make sure other cpus see that this has run */ + wrote_text = 1; + } else { + while (!wrote_text) + cpu_relax(); + smp_mb(); /* Load wrote_text before following execution */ + } + + flush_icache_range((unsigned long)tpp->addr, + (unsigned long)tpp->addr + tpp->len); + return 0; +} + +/** + * text_poke_smp - Update instructions on a live kernel on SMP + * @addr: address to modify + * @opcode: source of the copy + * @len: length to copy + * + * Modify multi-byte instruction by using stop_machine() on SMP. This allows + * user to poke/set multi-byte text on SMP. Only non-NMI/MCE code modifying + * should be allowed, since stop_machine() does _not_ protect code against + * NMI and MCE. + * + * Note: Must be called under get_online_cpus() and text_mutex. + */ +void *__kprobes text_poke_smp(void *addr, const void *opcode, size_t len) +{ + struct text_poke_params tpp; + + tpp.addr = addr; + tpp.opcode = opcode; + tpp.len = len; + atomic_set(&stop_machine_first, 1); + wrote_text = 0; + stop_machine(stop_machine_text_poke, (void *)&tpp, NULL); + return addr; +} + diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 23824fef789..fa5a1474cd1 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -18,8 +18,8 @@ */ #include <linux/pci.h> -#include <linux/gfp.h> #include <linux/bitmap.h> +#include <linux/slab.h> #include <linux/debugfs.h> #include <linux/scatterlist.h> #include <linux/dma-mapping.h> @@ -118,7 +118,7 @@ static bool check_device(struct device *dev) return false; /* No device or no PCI device */ - if (!dev || dev->bus != &pci_bus_type) + if (dev->bus != &pci_bus_type) return false; devid = get_device_id(dev); @@ -392,6 +392,7 @@ static int __iommu_queue_command(struct amd_iommu *iommu, struct iommu_cmd *cmd) u32 tail, head; u8 *target; + WARN_ON(iommu->cmd_buf_size & CMD_BUFFER_UNINITIALIZED); tail = readl(iommu->mmio_base + MMIO_CMD_TAIL_OFFSET); target = iommu->cmd_buf + tail; memcpy_toio(target, cmd, sizeof(*cmd)); @@ -730,18 +731,22 @@ static bool increase_address_space(struct protection_domain *domain, static u64 *alloc_pte(struct protection_domain *domain, unsigned long address, - int end_lvl, + unsigned long page_size, u64 **pte_page, gfp_t gfp) { + int level, end_lvl; u64 *pte, *page; - int level; + + BUG_ON(!is_power_of_2(page_size)); while (address > PM_LEVEL_SIZE(domain->mode)) increase_address_space(domain, gfp); - level = domain->mode - 1; - pte = &domain->pt_root[PM_LEVEL_INDEX(level, address)]; + level = domain->mode - 1; + pte = &domain->pt_root[PM_LEVEL_INDEX(level, address)]; + address = PAGE_SIZE_ALIGN(address, page_size); + end_lvl = PAGE_SIZE_LEVEL(page_size); while (level > end_lvl) { if (!IOMMU_PTE_PRESENT(*pte)) { @@ -751,6 +756,10 @@ static u64 *alloc_pte(struct protection_domain *domain, *pte = PM_LEVEL_PDE(level, virt_to_phys(page)); } + /* No level skipping support yet */ + if (PM_PTE_LEVEL(*pte) != level) + return NULL; + level -= 1; pte = IOMMU_PTE_PAGE(*pte); @@ -768,28 +777,47 @@ static u64 *alloc_pte(struct protection_domain *domain, * This function checks if there is a PTE for a given dma address. If * there is one, it returns the pointer to it. */ -static u64 *fetch_pte(struct protection_domain *domain, - unsigned long address, int map_size) +static u64 *fetch_pte(struct protection_domain *domain, unsigned long address) { int level; u64 *pte; - level = domain->mode - 1; - pte = &domain->pt_root[PM_LEVEL_INDEX(level, address)]; + if (address > PM_LEVEL_SIZE(domain->mode)) + return NULL; - while (level > map_size) { + level = domain->mode - 1; + pte = &domain->pt_root[PM_LEVEL_INDEX(level, address)]; + + while (level > 0) { + + /* Not Present */ if (!IOMMU_PTE_PRESENT(*pte)) return NULL; + /* Large PTE */ + if (PM_PTE_LEVEL(*pte) == 0x07) { + unsigned long pte_mask, __pte; + + /* + * If we have a series of large PTEs, make + * sure to return a pointer to the first one. + */ + pte_mask = PTE_PAGE_SIZE(*pte); + pte_mask = ~((PAGE_SIZE_PTE_COUNT(pte_mask) << 3) - 1); + __pte = ((unsigned long)pte) & pte_mask; + + return (u64 *)__pte; + } + + /* No level skipping support yet */ + if (PM_PTE_LEVEL(*pte) != level) + return NULL; + level -= 1; + /* Walk to the next level */ pte = IOMMU_PTE_PAGE(*pte); pte = &pte[PM_LEVEL_INDEX(level, address)]; - - if ((PM_PTE_LEVEL(*pte) == 0) && level != map_size) { - pte = NULL; - break; - } } return pte; @@ -806,44 +834,84 @@ static int iommu_map_page(struct protection_domain *dom, unsigned long bus_addr, unsigned long phys_addr, int prot, - int map_size) + unsigned long page_size) { u64 __pte, *pte; - - bus_addr = PAGE_ALIGN(bus_addr); - phys_addr = PAGE_ALIGN(phys_addr); - - BUG_ON(!PM_ALIGNED(map_size, bus_addr)); - BUG_ON(!PM_ALIGNED(map_size, phys_addr)); + int i, count; if (!(prot & IOMMU_PROT_MASK)) return -EINVAL; - pte = alloc_pte(dom, bus_addr, map_size, NULL, GFP_KERNEL); + bus_addr = PAGE_ALIGN(bus_addr); + phys_addr = PAGE_ALIGN(phys_addr); + count = PAGE_SIZE_PTE_COUNT(page_size); + pte = alloc_pte(dom, bus_addr, page_size, NULL, GFP_KERNEL); - if (IOMMU_PTE_PRESENT(*pte)) - return -EBUSY; + for (i = 0; i < count; ++i) + if (IOMMU_PTE_PRESENT(pte[i])) + return -EBUSY; + + if (page_size > PAGE_SIZE) { + __pte = PAGE_SIZE_PTE(phys_addr, page_size); + __pte |= PM_LEVEL_ENC(7) | IOMMU_PTE_P | IOMMU_PTE_FC; + } else + __pte = phys_addr | IOMMU_PTE_P | IOMMU_PTE_FC; - __pte = phys_addr | IOMMU_PTE_P; if (prot & IOMMU_PROT_IR) __pte |= IOMMU_PTE_IR; if (prot & IOMMU_PROT_IW) __pte |= IOMMU_PTE_IW; - *pte = __pte; + for (i = 0; i < count; ++i) + pte[i] = __pte; update_domain(dom); return 0; } -static void iommu_unmap_page(struct protection_domain *dom, - unsigned long bus_addr, int map_size) +static unsigned long iommu_unmap_page(struct protection_domain *dom, + unsigned long bus_addr, + unsigned long page_size) { - u64 *pte = fetch_pte(dom, bus_addr, map_size); + unsigned long long unmap_size, unmapped; + u64 *pte; + + BUG_ON(!is_power_of_2(page_size)); + + unmapped = 0; + + while (unmapped < page_size) { + + pte = fetch_pte(dom, bus_addr); + + if (!pte) { + /* + * No PTE for this address + * move forward in 4kb steps + */ + unmap_size = PAGE_SIZE; + } else if (PM_PTE_LEVEL(*pte) == 0) { + /* 4kb PTE found for this address */ + unmap_size = PAGE_SIZE; + *pte = 0ULL; + } else { + int count, i; + + /* Large PTE found which maps this address */ + unmap_size = PTE_PAGE_SIZE(*pte); + count = PAGE_SIZE_PTE_COUNT(unmap_size); + for (i = 0; i < count; i++) + pte[i] = 0ULL; + } + + bus_addr = (bus_addr & ~(unmap_size - 1)) + unmap_size; + unmapped += unmap_size; + } + + BUG_ON(!is_power_of_2(unmapped)); - if (pte) - *pte = 0; + return unmapped; } /* @@ -877,7 +945,7 @@ static int dma_ops_unity_map(struct dma_ops_domain *dma_dom, for (addr = e->address_start; addr < e->address_end; addr += PAGE_SIZE) { ret = iommu_map_page(&dma_dom->domain, addr, addr, e->prot, - PM_MAP_4k); + PAGE_SIZE); if (ret) return ret; /* @@ -980,7 +1048,7 @@ static int alloc_new_range(struct dma_ops_domain *dma_dom, { int index = dma_dom->aperture_size >> APERTURE_RANGE_SHIFT; struct amd_iommu *iommu; - int i; + unsigned long i; #ifdef CONFIG_IOMMU_STRESS populate = false; @@ -1005,7 +1073,7 @@ static int alloc_new_range(struct dma_ops_domain *dma_dom, u64 *pte, *pte_page; for (i = 0; i < num_ptes; ++i) { - pte = alloc_pte(&dma_dom->domain, address, PM_MAP_4k, + pte = alloc_pte(&dma_dom->domain, address, PAGE_SIZE, &pte_page, gfp); if (!pte) goto out_free; @@ -1041,7 +1109,7 @@ static int alloc_new_range(struct dma_ops_domain *dma_dom, for (i = dma_dom->aperture[index]->offset; i < dma_dom->aperture_size; i += PAGE_SIZE) { - u64 *pte = fetch_pte(&dma_dom->domain, i, PM_MAP_4k); + u64 *pte = fetch_pte(&dma_dom->domain, i); if (!pte || !IOMMU_PTE_PRESENT(*pte)) continue; @@ -1489,11 +1557,14 @@ static void __detach_device(struct device *dev) { struct iommu_dev_data *dev_data = get_dev_data(dev); struct iommu_dev_data *alias_data; + struct protection_domain *domain; unsigned long flags; BUG_ON(!dev_data->domain); - spin_lock_irqsave(&dev_data->domain->lock, flags); + domain = dev_data->domain; + + spin_lock_irqsave(&domain->lock, flags); if (dev_data->alias != dev) { alias_data = get_dev_data(dev_data->alias); @@ -1504,13 +1575,15 @@ static void __detach_device(struct device *dev) if (atomic_dec_and_test(&dev_data->bind)) do_detach(dev); - spin_unlock_irqrestore(&dev_data->domain->lock, flags); + spin_unlock_irqrestore(&domain->lock, flags); /* * If we run in passthrough mode the device must be assigned to the - * passthrough domain if it is detached from any other domain + * passthrough domain if it is detached from any other domain. + * Make sure we can deassign from the pt_domain itself. */ - if (iommu_pass_through && dev_data->domain == NULL) + if (iommu_pass_through && + (dev_data->domain == NULL && domain != pt_domain)) __attach_device(dev, pt_domain); } @@ -1706,7 +1779,7 @@ static u64* dma_ops_get_pte(struct dma_ops_domain *dom, pte = aperture->pte_pages[APERTURE_PAGE_INDEX(address)]; if (!pte) { - pte = alloc_pte(&dom->domain, address, PM_MAP_4k, &pte_page, + pte = alloc_pte(&dom->domain, address, PAGE_SIZE, &pte_page, GFP_ATOMIC); aperture->pte_pages[APERTURE_PAGE_INDEX(address)] = pte_page; } else @@ -2181,7 +2254,7 @@ static void prealloc_protection_domains(void) struct dma_ops_domain *dma_dom; u16 devid; - while ((dev = pci_get_device(PCI_ANY_ID, PCI_ANY_ID, dev)) != NULL) { + for_each_pci_dev(dev) { /* Do we handle this device? */ if (!check_device(&dev->dev)) @@ -2218,6 +2291,12 @@ static struct dma_map_ops amd_iommu_dma_ops = { /* * The function which clues the AMD IOMMU driver into dma_ops. */ + +void __init amd_iommu_init_api(void) +{ + register_iommu(&amd_iommu_ops); +} + int __init amd_iommu_init_dma_ops(void) { struct amd_iommu *iommu; @@ -2253,8 +2332,6 @@ int __init amd_iommu_init_dma_ops(void) /* Make the driver finally visible to the drivers */ dma_ops = &amd_iommu_dma_ops; - register_iommu(&amd_iommu_ops); - amd_iommu_stats_init(); return 0; @@ -2289,7 +2366,7 @@ static void cleanup_domain(struct protection_domain *domain) list_for_each_entry_safe(dev_data, next, &domain->dev_list, list) { struct device *dev = dev_data->dev; - do_detach(dev); + __detach_device(dev); atomic_set(&dev_data->bind, 0); } @@ -2318,6 +2395,7 @@ static struct protection_domain *protection_domain_alloc(void) return NULL; spin_lock_init(&domain->lock); + mutex_init(&domain->api_lock); domain->id = domain_id_alloc(); if (!domain->id) goto out_err; @@ -2370,9 +2448,7 @@ static void amd_iommu_domain_destroy(struct iommu_domain *dom) free_pagetable(domain); - domain_id_free(domain->id); - - kfree(domain); + protection_domain_free(domain); dom->priv = NULL; } @@ -2430,12 +2506,11 @@ static int amd_iommu_attach_device(struct iommu_domain *dom, return ret; } -static int amd_iommu_map_range(struct iommu_domain *dom, - unsigned long iova, phys_addr_t paddr, - size_t size, int iommu_prot) +static int amd_iommu_map(struct iommu_domain *dom, unsigned long iova, + phys_addr_t paddr, int gfp_order, int iommu_prot) { + unsigned long page_size = 0x1000UL << gfp_order; struct protection_domain *domain = dom->priv; - unsigned long i, npages = iommu_num_pages(paddr, size, PAGE_SIZE); int prot = 0; int ret; @@ -2444,53 +2519,50 @@ static int amd_iommu_map_range(struct iommu_domain *dom, if (iommu_prot & IOMMU_WRITE) prot |= IOMMU_PROT_IW; - iova &= PAGE_MASK; - paddr &= PAGE_MASK; - - for (i = 0; i < npages; ++i) { - ret = iommu_map_page(domain, iova, paddr, prot, PM_MAP_4k); - if (ret) - return ret; + mutex_lock(&domain->api_lock); + ret = iommu_map_page(domain, iova, paddr, prot, page_size); + mutex_unlock(&domain->api_lock); - iova += PAGE_SIZE; - paddr += PAGE_SIZE; - } - - return 0; + return ret; } -static void amd_iommu_unmap_range(struct iommu_domain *dom, - unsigned long iova, size_t size) +static int amd_iommu_unmap(struct iommu_domain *dom, unsigned long iova, + int gfp_order) { - struct protection_domain *domain = dom->priv; - unsigned long i, npages = iommu_num_pages(iova, size, PAGE_SIZE); + unsigned long page_size, unmap_size; - iova &= PAGE_MASK; + page_size = 0x1000UL << gfp_order; - for (i = 0; i < npages; ++i) { - iommu_unmap_page(domain, iova, PM_MAP_4k); - iova += PAGE_SIZE; - } + mutex_lock(&domain->api_lock); + unmap_size = iommu_unmap_page(domain, iova, page_size); + mutex_unlock(&domain->api_lock); iommu_flush_tlb_pde(domain); + + return get_order(unmap_size); } static phys_addr_t amd_iommu_iova_to_phys(struct iommu_domain *dom, unsigned long iova) { struct protection_domain *domain = dom->priv; - unsigned long offset = iova & ~PAGE_MASK; + unsigned long offset_mask; phys_addr_t paddr; - u64 *pte; + u64 *pte, __pte; - pte = fetch_pte(domain, iova, PM_MAP_4k); + pte = fetch_pte(domain, iova); if (!pte || !IOMMU_PTE_PRESENT(*pte)) return 0; - paddr = *pte & IOMMU_PAGE_MASK; - paddr |= offset; + if (PM_PTE_LEVEL(*pte) == 0) + offset_mask = PAGE_SIZE - 1; + else + offset_mask = PTE_PAGE_SIZE(*pte) - 1; + + __pte = *pte & PM_ADDR_MASK; + paddr = (__pte & ~offset_mask) | (iova & offset_mask); return paddr; } @@ -2506,8 +2578,8 @@ static struct iommu_ops amd_iommu_ops = { .domain_destroy = amd_iommu_domain_destroy, .attach_dev = amd_iommu_attach_device, .detach_dev = amd_iommu_detach_device, - .map = amd_iommu_map_range, - .unmap = amd_iommu_unmap_range, + .map = amd_iommu_map, + .unmap = amd_iommu_unmap, .iova_to_phys = amd_iommu_iova_to_phys, .domain_has_cap = amd_iommu_domain_has_cap, }; diff --git a/arch/x86/kernel/amd_iommu_init.c b/arch/x86/kernel/amd_iommu_init.c index fb490ce7dd5..3bacb4d0844 100644 --- a/arch/x86/kernel/amd_iommu_init.c +++ b/arch/x86/kernel/amd_iommu_init.c @@ -19,8 +19,8 @@ #include <linux/pci.h> #include <linux/acpi.h> -#include <linux/gfp.h> #include <linux/list.h> +#include <linux/slab.h> #include <linux/sysdev.h> #include <linux/interrupt.h> #include <linux/msi.h> @@ -120,6 +120,7 @@ struct ivmd_header { bool amd_iommu_dump; static int __initdata amd_iommu_detected; +static bool __initdata amd_iommu_disabled; u16 amd_iommu_last_bdf; /* largest PCI device id we have to handle */ @@ -138,9 +139,9 @@ int amd_iommus_present; bool amd_iommu_np_cache __read_mostly; /* - * Set to true if ACPI table parsing and hardware intialization went properly + * The ACPI table parsing functions set this variable on an error */ -static bool amd_iommu_initialized; +static int __initdata amd_iommu_init_err; /* * List of protection domains - used during resume @@ -391,9 +392,11 @@ static int __init find_last_devid_acpi(struct acpi_table_header *table) */ for (i = 0; i < table->length; ++i) checksum += p[i]; - if (checksum != 0) + if (checksum != 0) { /* ACPI table corrupt */ - return -ENODEV; + amd_iommu_init_err = -ENODEV; + return 0; + } p += IVRS_HEADER_LENGTH; @@ -436,7 +439,7 @@ static u8 * __init alloc_command_buffer(struct amd_iommu *iommu) if (cmd_buf == NULL) return NULL; - iommu->cmd_buf_size = CMD_BUFFER_SIZE; + iommu->cmd_buf_size = CMD_BUFFER_SIZE | CMD_BUFFER_UNINITIALIZED; return cmd_buf; } @@ -472,12 +475,13 @@ static void iommu_enable_command_buffer(struct amd_iommu *iommu) &entry, sizeof(entry)); amd_iommu_reset_cmd_buffer(iommu); + iommu->cmd_buf_size &= ~(CMD_BUFFER_UNINITIALIZED); } static void __init free_command_buffer(struct amd_iommu *iommu) { free_pages((unsigned long)iommu->cmd_buf, - get_order(iommu->cmd_buf_size)); + get_order(iommu->cmd_buf_size & ~(CMD_BUFFER_UNINITIALIZED))); } /* allocates the memory where the IOMMU will log its events to */ @@ -920,11 +924,16 @@ static int __init init_iommu_all(struct acpi_table_header *table) h->mmio_phys); iommu = kzalloc(sizeof(struct amd_iommu), GFP_KERNEL); - if (iommu == NULL) - return -ENOMEM; + if (iommu == NULL) { + amd_iommu_init_err = -ENOMEM; + return 0; + } + ret = init_iommu_one(iommu, h); - if (ret) - return ret; + if (ret) { + amd_iommu_init_err = ret; + return 0; + } break; default: break; @@ -934,8 +943,6 @@ static int __init init_iommu_all(struct acpi_table_header *table) } WARN_ON(p != end); - amd_iommu_initialized = true; - return 0; } @@ -1211,6 +1218,10 @@ static int __init amd_iommu_init(void) if (acpi_table_parse("IVRS", find_last_devid_acpi) != 0) return -ENODEV; + ret = amd_iommu_init_err; + if (ret) + goto out; + dev_table_size = tbl_size(DEV_TABLE_ENTRY_SIZE); alias_table_size = tbl_size(ALIAS_TABLE_ENTRY_SIZE); rlookup_table_size = tbl_size(RLOOKUP_TABLE_ENTRY_SIZE); @@ -1270,12 +1281,19 @@ static int __init amd_iommu_init(void) if (acpi_table_parse("IVRS", init_iommu_all) != 0) goto free; - if (!amd_iommu_initialized) + if (amd_iommu_init_err) { + ret = amd_iommu_init_err; goto free; + } if (acpi_table_parse("IVRS", init_memory_definitions) != 0) goto free; + if (amd_iommu_init_err) { + ret = amd_iommu_init_err; + goto free; + } + ret = sysdev_class_register(&amd_iommu_sysdev_class); if (ret) goto free; @@ -1288,16 +1306,19 @@ static int __init amd_iommu_init(void) if (ret) goto free; + enable_iommus(); + if (iommu_pass_through) ret = amd_iommu_init_passthrough(); else ret = amd_iommu_init_dma_ops(); + if (ret) goto free; - amd_iommu_init_notifier(); + amd_iommu_init_api(); - enable_iommus(); + amd_iommu_init_notifier(); if (iommu_pass_through) goto out; @@ -1312,6 +1333,7 @@ out: return ret; free: + disable_iommus(); amd_iommu_uninit_devices(); @@ -1351,6 +1373,9 @@ void __init amd_iommu_detect(void) if (no_iommu || (iommu_detected && !gart_iommu_aperture)) return; + if (amd_iommu_disabled) + return; + if (acpi_table_parse("IVRS", early_amd_iommu_detect) == 0) { iommu_detected = 1; amd_iommu_detected = 1; @@ -1380,6 +1405,8 @@ static int __init parse_amd_iommu_options(char *str) for (; *str; ++str) { if (strncmp(str, "fullflush", 9) == 0) amd_iommu_unmap_flush = true; + if (strncmp(str, "off", 3) == 0) + amd_iommu_disabled = true; } return 1; diff --git a/arch/x86/kernel/apb_timer.c b/arch/x86/kernel/apb_timer.c new file mode 100644 index 00000000000..a35347501d3 --- /dev/null +++ b/arch/x86/kernel/apb_timer.c @@ -0,0 +1,785 @@ +/* + * apb_timer.c: Driver for Langwell APB timers + * + * (C) Copyright 2009 Intel Corporation + * Author: Jacob Pan (jacob.jun.pan@intel.com) + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; version 2 + * of the License. + * + * Note: + * Langwell is the south complex of Intel Moorestown MID platform. There are + * eight external timers in total that can be used by the operating system. + * The timer information, such as frequency and addresses, is provided to the + * OS via SFI tables. + * Timer interrupts are routed via FW/HW emulated IOAPIC independently via + * individual redirection table entries (RTE). + * Unlike HPET, there is no master counter, therefore one of the timers are + * used as clocksource. The overall allocation looks like: + * - timer 0 - NR_CPUs for per cpu timer + * - one timer for clocksource + * - one timer for watchdog driver. + * It is also worth notice that APB timer does not support true one-shot mode, + * free-running mode will be used here to emulate one-shot mode. + * APB timer can also be used as broadcast timer along with per cpu local APIC + * timer, but by default APB timer has higher rating than local APIC timers. + */ + +#include <linux/clocksource.h> +#include <linux/clockchips.h> +#include <linux/delay.h> +#include <linux/errno.h> +#include <linux/init.h> +#include <linux/sysdev.h> +#include <linux/slab.h> +#include <linux/pm.h> +#include <linux/pci.h> +#include <linux/sfi.h> +#include <linux/interrupt.h> +#include <linux/cpu.h> +#include <linux/irq.h> + +#include <asm/fixmap.h> +#include <asm/apb_timer.h> + +#define APBT_MASK CLOCKSOURCE_MASK(32) +#define APBT_SHIFT 22 +#define APBT_CLOCKEVENT_RATING 150 +#define APBT_CLOCKSOURCE_RATING 250 +#define APBT_MIN_DELTA_USEC 200 + +#define EVT_TO_APBT_DEV(evt) container_of(evt, struct apbt_dev, evt) +#define APBT_CLOCKEVENT0_NUM (0) +#define APBT_CLOCKEVENT1_NUM (1) +#define APBT_CLOCKSOURCE_NUM (2) + +static unsigned long apbt_address; +static int apb_timer_block_enabled; +static void __iomem *apbt_virt_address; +static int phy_cs_timer_id; + +/* + * Common DW APB timer info + */ +static uint64_t apbt_freq; + +static void apbt_set_mode(enum clock_event_mode mode, + struct clock_event_device *evt); +static int apbt_next_event(unsigned long delta, + struct clock_event_device *evt); +static cycle_t apbt_read_clocksource(struct clocksource *cs); +static void apbt_restart_clocksource(struct clocksource *cs); + +struct apbt_dev { + struct clock_event_device evt; + unsigned int num; + int cpu; + unsigned int irq; + unsigned int tick; + unsigned int count; + unsigned int flags; + char name[10]; +}; + +int disable_apbt_percpu __cpuinitdata; + +static DEFINE_PER_CPU(struct apbt_dev, cpu_apbt_dev); + +#ifdef CONFIG_SMP +static unsigned int apbt_num_timers_used; +static struct apbt_dev *apbt_devs; +#endif + +static inline unsigned long apbt_readl_reg(unsigned long a) +{ + return readl(apbt_virt_address + a); +} + +static inline void apbt_writel_reg(unsigned long d, unsigned long a) +{ + writel(d, apbt_virt_address + a); +} + +static inline unsigned long apbt_readl(int n, unsigned long a) +{ + return readl(apbt_virt_address + a + n * APBTMRS_REG_SIZE); +} + +static inline void apbt_writel(int n, unsigned long d, unsigned long a) +{ + writel(d, apbt_virt_address + a + n * APBTMRS_REG_SIZE); +} + +static inline void apbt_set_mapping(void) +{ + struct sfi_timer_table_entry *mtmr; + + if (apbt_virt_address) { + pr_debug("APBT base already mapped\n"); + return; + } + mtmr = sfi_get_mtmr(APBT_CLOCKEVENT0_NUM); + if (mtmr == NULL) { + printk(KERN_ERR "Failed to get MTMR %d from SFI\n", + APBT_CLOCKEVENT0_NUM); + return; + } + apbt_address = (unsigned long)mtmr->phys_addr; + if (!apbt_address) { + printk(KERN_WARNING "No timer base from SFI, use default\n"); + apbt_address = APBT_DEFAULT_BASE; + } + apbt_virt_address = ioremap_nocache(apbt_address, APBT_MMAP_SIZE); + if (apbt_virt_address) { + pr_debug("Mapped APBT physical addr %p at virtual addr %p\n",\ + (void *)apbt_address, (void *)apbt_virt_address); + } else { + pr_debug("Failed mapping APBT phy address at %p\n",\ + (void *)apbt_address); + goto panic_noapbt; + } + apbt_freq = mtmr->freq_hz / USEC_PER_SEC; + sfi_free_mtmr(mtmr); + + /* Now figure out the physical timer id for clocksource device */ + mtmr = sfi_get_mtmr(APBT_CLOCKSOURCE_NUM); + if (mtmr == NULL) + goto panic_noapbt; + + /* Now figure out the physical timer id */ + phy_cs_timer_id = (unsigned int)(mtmr->phys_addr & 0xff) + / APBTMRS_REG_SIZE; + pr_debug("Use timer %d for clocksource\n", phy_cs_timer_id); + return; + +panic_noapbt: + panic("Failed to setup APB system timer\n"); + +} + +static inline void apbt_clear_mapping(void) +{ + iounmap(apbt_virt_address); + apbt_virt_address = NULL; +} + +/* + * APBT timer interrupt enable / disable + */ +static inline int is_apbt_capable(void) +{ + return apbt_virt_address ? 1 : 0; +} + +static struct clocksource clocksource_apbt = { + .name = "apbt", + .rating = APBT_CLOCKSOURCE_RATING, + .read = apbt_read_clocksource, + .mask = APBT_MASK, + .shift = APBT_SHIFT, + .flags = CLOCK_SOURCE_IS_CONTINUOUS, + .resume = apbt_restart_clocksource, +}; + +/* boot APB clock event device */ +static struct clock_event_device apbt_clockevent = { + .name = "apbt0", + .features = CLOCK_EVT_FEAT_PERIODIC | CLOCK_EVT_FEAT_ONESHOT, + .set_mode = apbt_set_mode, + .set_next_event = apbt_next_event, + .shift = APBT_SHIFT, + .irq = 0, + .rating = APBT_CLOCKEVENT_RATING, +}; + +/* + * if user does not want to use per CPU apb timer, just give it a lower rating + * than local apic timer and skip the late per cpu timer init. + */ +static inline int __init setup_x86_mrst_timer(char *arg) +{ + if (!arg) + return -EINVAL; + + if (strcmp("apbt_only", arg) == 0) + disable_apbt_percpu = 0; + else if (strcmp("lapic_and_apbt", arg) == 0) + disable_apbt_percpu = 1; + else { + pr_warning("X86 MRST timer option %s not recognised" + " use x86_mrst_timer=apbt_only or lapic_and_apbt\n", + arg); + return -EINVAL; + } + return 0; +} +__setup("x86_mrst_timer=", setup_x86_mrst_timer); + +/* + * start count down from 0xffff_ffff. this is done by toggling the enable bit + * then load initial load count to ~0. + */ +static void apbt_start_counter(int n) +{ + unsigned long ctrl = apbt_readl(n, APBTMR_N_CONTROL); + + ctrl &= ~APBTMR_CONTROL_ENABLE; + apbt_writel(n, ctrl, APBTMR_N_CONTROL); + apbt_writel(n, ~0, APBTMR_N_LOAD_COUNT); + /* enable, mask interrupt */ + ctrl &= ~APBTMR_CONTROL_MODE_PERIODIC; + ctrl |= (APBTMR_CONTROL_ENABLE | APBTMR_CONTROL_INT); + apbt_writel(n, ctrl, APBTMR_N_CONTROL); + /* read it once to get cached counter value initialized */ + apbt_read_clocksource(&clocksource_apbt); +} + +static irqreturn_t apbt_interrupt_handler(int irq, void *data) +{ + struct apbt_dev *dev = (struct apbt_dev *)data; + struct clock_event_device *aevt = &dev->evt; + + if (!aevt->event_handler) { + printk(KERN_INFO "Spurious APBT timer interrupt on %d\n", + dev->num); + return IRQ_NONE; + } + aevt->event_handler(aevt); + return IRQ_HANDLED; +} + +static void apbt_restart_clocksource(struct clocksource *cs) +{ + apbt_start_counter(phy_cs_timer_id); +} + +/* Setup IRQ routing via IOAPIC */ +#ifdef CONFIG_SMP +static void apbt_setup_irq(struct apbt_dev *adev) +{ + struct irq_chip *chip; + struct irq_desc *desc; + + /* timer0 irq has been setup early */ + if (adev->irq == 0) + return; + desc = irq_to_desc(adev->irq); + chip = get_irq_chip(adev->irq); + disable_irq(adev->irq); + desc->status |= IRQ_MOVE_PCNTXT; + irq_set_affinity(adev->irq, cpumask_of(adev->cpu)); + /* APB timer irqs are set up as mp_irqs, timer is edge triggerred */ + set_irq_chip_and_handler_name(adev->irq, chip, handle_edge_irq, "edge"); + enable_irq(adev->irq); + if (system_state == SYSTEM_BOOTING) + if (request_irq(adev->irq, apbt_interrupt_handler, + IRQF_TIMER | IRQF_DISABLED | IRQF_NOBALANCING, + adev->name, adev)) { + printk(KERN_ERR "Failed request IRQ for APBT%d\n", + adev->num); + } +} +#endif + +static void apbt_enable_int(int n) +{ + unsigned long ctrl = apbt_readl(n, APBTMR_N_CONTROL); + /* clear pending intr */ + apbt_readl(n, APBTMR_N_EOI); + ctrl &= ~APBTMR_CONTROL_INT; + apbt_writel(n, ctrl, APBTMR_N_CONTROL); +} + +static void apbt_disable_int(int n) +{ + unsigned long ctrl = apbt_readl(n, APBTMR_N_CONTROL); + + ctrl |= APBTMR_CONTROL_INT; + apbt_writel(n, ctrl, APBTMR_N_CONTROL); +} + + +static int __init apbt_clockevent_register(void) +{ + struct sfi_timer_table_entry *mtmr; + struct apbt_dev *adev = &__get_cpu_var(cpu_apbt_dev); + + mtmr = sfi_get_mtmr(APBT_CLOCKEVENT0_NUM); + if (mtmr == NULL) { + printk(KERN_ERR "Failed to get MTMR %d from SFI\n", + APBT_CLOCKEVENT0_NUM); + return -ENODEV; + } + + /* + * We need to calculate the scaled math multiplication factor for + * nanosecond to apbt tick conversion. + * mult = (nsec/cycle)*2^APBT_SHIFT + */ + apbt_clockevent.mult = div_sc((unsigned long) mtmr->freq_hz + , NSEC_PER_SEC, APBT_SHIFT); + + /* Calculate the min / max delta */ + apbt_clockevent.max_delta_ns = clockevent_delta2ns(0x7FFFFFFF, + &apbt_clockevent); + apbt_clockevent.min_delta_ns = clockevent_delta2ns( + APBT_MIN_DELTA_USEC*apbt_freq, + &apbt_clockevent); + /* + * Start apbt with the boot cpu mask and make it + * global if not used for per cpu timer. + */ + apbt_clockevent.cpumask = cpumask_of(smp_processor_id()); + adev->num = smp_processor_id(); + memcpy(&adev->evt, &apbt_clockevent, sizeof(struct clock_event_device)); + + if (disable_apbt_percpu) { + apbt_clockevent.rating = APBT_CLOCKEVENT_RATING - 100; + global_clock_event = &adev->evt; + printk(KERN_DEBUG "%s clockevent registered as global\n", + global_clock_event->name); + } + + if (request_irq(apbt_clockevent.irq, apbt_interrupt_handler, + IRQF_TIMER | IRQF_DISABLED | IRQF_NOBALANCING, + apbt_clockevent.name, adev)) { + printk(KERN_ERR "Failed request IRQ for APBT%d\n", + apbt_clockevent.irq); + } + + clockevents_register_device(&adev->evt); + /* Start APBT 0 interrupts */ + apbt_enable_int(APBT_CLOCKEVENT0_NUM); + + sfi_free_mtmr(mtmr); + return 0; +} + +#ifdef CONFIG_SMP +/* Should be called with per cpu */ +void apbt_setup_secondary_clock(void) +{ + struct apbt_dev *adev; + struct clock_event_device *aevt; + int cpu; + + /* Don't register boot CPU clockevent */ + cpu = smp_processor_id(); + if (cpu == boot_cpu_id) + return; + /* + * We need to calculate the scaled math multiplication factor for + * nanosecond to apbt tick conversion. + * mult = (nsec/cycle)*2^APBT_SHIFT + */ + printk(KERN_INFO "Init per CPU clockevent %d\n", cpu); + adev = &per_cpu(cpu_apbt_dev, cpu); + aevt = &adev->evt; + + memcpy(aevt, &apbt_clockevent, sizeof(*aevt)); + aevt->cpumask = cpumask_of(cpu); + aevt->name = adev->name; + aevt->mode = CLOCK_EVT_MODE_UNUSED; + + printk(KERN_INFO "Registering CPU %d clockevent device %s, mask %08x\n", + cpu, aevt->name, *(u32 *)aevt->cpumask); + + apbt_setup_irq(adev); + + clockevents_register_device(aevt); + + apbt_enable_int(cpu); + + return; +} + +/* + * this notify handler process CPU hotplug events. in case of S0i3, nonboot + * cpus are disabled/enabled frequently, for performance reasons, we keep the + * per cpu timer irq registered so that we do need to do free_irq/request_irq. + * + * TODO: it might be more reliable to directly disable percpu clockevent device + * without the notifier chain. currently, cpu 0 may get interrupts from other + * cpu timers during the offline process due to the ordering of notification. + * the extra interrupt is harmless. + */ +static int apbt_cpuhp_notify(struct notifier_block *n, + unsigned long action, void *hcpu) +{ + unsigned long cpu = (unsigned long)hcpu; + struct apbt_dev *adev = &per_cpu(cpu_apbt_dev, cpu); + + switch (action & 0xf) { + case CPU_DEAD: + apbt_disable_int(cpu); + if (system_state == SYSTEM_RUNNING) + pr_debug("skipping APBT CPU %lu offline\n", cpu); + else if (adev) { + pr_debug("APBT clockevent for cpu %lu offline\n", cpu); + free_irq(adev->irq, adev); + } + break; + default: + pr_debug(KERN_INFO "APBT notified %lu, no action\n", action); + } + return NOTIFY_OK; +} + +static __init int apbt_late_init(void) +{ + if (disable_apbt_percpu || !apb_timer_block_enabled) + return 0; + /* This notifier should be called after workqueue is ready */ + hotcpu_notifier(apbt_cpuhp_notify, -20); + return 0; +} +fs_initcall(apbt_late_init); +#else + +void apbt_setup_secondary_clock(void) {} + +#endif /* CONFIG_SMP */ + +static void apbt_set_mode(enum clock_event_mode mode, + struct clock_event_device *evt) +{ + unsigned long ctrl; + uint64_t delta; + int timer_num; + struct apbt_dev *adev = EVT_TO_APBT_DEV(evt); + + timer_num = adev->num; + pr_debug("%s CPU %d timer %d mode=%d\n", + __func__, first_cpu(*evt->cpumask), timer_num, mode); + + switch (mode) { + case CLOCK_EVT_MODE_PERIODIC: + delta = ((uint64_t)(NSEC_PER_SEC/HZ)) * apbt_clockevent.mult; + delta >>= apbt_clockevent.shift; + ctrl = apbt_readl(timer_num, APBTMR_N_CONTROL); + ctrl |= APBTMR_CONTROL_MODE_PERIODIC; + apbt_writel(timer_num, ctrl, APBTMR_N_CONTROL); + /* + * DW APB p. 46, have to disable timer before load counter, + * may cause sync problem. + */ + ctrl &= ~APBTMR_CONTROL_ENABLE; + apbt_writel(timer_num, ctrl, APBTMR_N_CONTROL); + udelay(1); + pr_debug("Setting clock period %d for HZ %d\n", (int)delta, HZ); + apbt_writel(timer_num, delta, APBTMR_N_LOAD_COUNT); + ctrl |= APBTMR_CONTROL_ENABLE; + apbt_writel(timer_num, ctrl, APBTMR_N_CONTROL); + break; + /* APB timer does not have one-shot mode, use free running mode */ + case CLOCK_EVT_MODE_ONESHOT: + ctrl = apbt_readl(timer_num, APBTMR_N_CONTROL); + /* + * set free running mode, this mode will let timer reload max + * timeout which will give time (3min on 25MHz clock) to rearm + * the next event, therefore emulate the one-shot mode. + */ + ctrl &= ~APBTMR_CONTROL_ENABLE; + ctrl &= ~APBTMR_CONTROL_MODE_PERIODIC; + + apbt_writel(timer_num, ctrl, APBTMR_N_CONTROL); + /* write again to set free running mode */ + apbt_writel(timer_num, ctrl, APBTMR_N_CONTROL); + + /* + * DW APB p. 46, load counter with all 1s before starting free + * running mode. + */ + apbt_writel(timer_num, ~0, APBTMR_N_LOAD_COUNT); + ctrl &= ~APBTMR_CONTROL_INT; + ctrl |= APBTMR_CONTROL_ENABLE; + apbt_writel(timer_num, ctrl, APBTMR_N_CONTROL); + break; + + case CLOCK_EVT_MODE_UNUSED: + case CLOCK_EVT_MODE_SHUTDOWN: + apbt_disable_int(timer_num); + ctrl = apbt_readl(timer_num, APBTMR_N_CONTROL); + ctrl &= ~APBTMR_CONTROL_ENABLE; + apbt_writel(timer_num, ctrl, APBTMR_N_CONTROL); + break; + + case CLOCK_EVT_MODE_RESUME: + apbt_enable_int(timer_num); + break; + } +} + +static int apbt_next_event(unsigned long delta, + struct clock_event_device *evt) +{ + unsigned long ctrl; + int timer_num; + + struct apbt_dev *adev = EVT_TO_APBT_DEV(evt); + + timer_num = adev->num; + /* Disable timer */ + ctrl = apbt_readl(timer_num, APBTMR_N_CONTROL); + ctrl &= ~APBTMR_CONTROL_ENABLE; + apbt_writel(timer_num, ctrl, APBTMR_N_CONTROL); + /* write new count */ + apbt_writel(timer_num, delta, APBTMR_N_LOAD_COUNT); + ctrl |= APBTMR_CONTROL_ENABLE; + apbt_writel(timer_num, ctrl, APBTMR_N_CONTROL); + return 0; +} + +/* + * APB timer clock is not in sync with pclk on Langwell, which translates to + * unreliable read value caused by sampling error. the error does not add up + * overtime and only happens when sampling a 0 as a 1 by mistake. so the time + * would go backwards. the following code is trying to prevent time traveling + * backwards. little bit paranoid. + */ +static cycle_t apbt_read_clocksource(struct clocksource *cs) +{ + unsigned long t0, t1, t2; + static unsigned long last_read; + +bad_count: + t1 = apbt_readl(phy_cs_timer_id, + APBTMR_N_CURRENT_VALUE); + t2 = apbt_readl(phy_cs_timer_id, + APBTMR_N_CURRENT_VALUE); + if (unlikely(t1 < t2)) { + pr_debug("APBT: read current count error %lx:%lx:%lx\n", + t1, t2, t2 - t1); + goto bad_count; + } + /* + * check against cached last read, makes sure time does not go back. + * it could be a normal rollover but we will do tripple check anyway + */ + if (unlikely(t2 > last_read)) { + /* check if we have a normal rollover */ + unsigned long raw_intr_status = + apbt_readl_reg(APBTMRS_RAW_INT_STATUS); + /* + * cs timer interrupt is masked but raw intr bit is set if + * rollover occurs. then we read EOI reg to clear it. + */ + if (raw_intr_status & (1 << phy_cs_timer_id)) { + apbt_readl(phy_cs_timer_id, APBTMR_N_EOI); + goto out; + } + pr_debug("APB CS going back %lx:%lx:%lx ", + t2, last_read, t2 - last_read); +bad_count_x3: + pr_debug(KERN_INFO "tripple check enforced\n"); + t0 = apbt_readl(phy_cs_timer_id, + APBTMR_N_CURRENT_VALUE); + udelay(1); + t1 = apbt_readl(phy_cs_timer_id, + APBTMR_N_CURRENT_VALUE); + udelay(1); + t2 = apbt_readl(phy_cs_timer_id, + APBTMR_N_CURRENT_VALUE); + if ((t2 > t1) || (t1 > t0)) { + printk(KERN_ERR "Error: APB CS tripple check failed\n"); + goto bad_count_x3; + } + } +out: + last_read = t2; + return (cycle_t)~t2; +} + +static int apbt_clocksource_register(void) +{ + u64 start, now; + cycle_t t1; + + /* Start the counter, use timer 2 as source, timer 0/1 for event */ + apbt_start_counter(phy_cs_timer_id); + + /* Verify whether apbt counter works */ + t1 = apbt_read_clocksource(&clocksource_apbt); + rdtscll(start); + + /* + * We don't know the TSC frequency yet, but waiting for + * 200000 TSC cycles is safe: + * 4 GHz == 50us + * 1 GHz == 200us + */ + do { + rep_nop(); + rdtscll(now); + } while ((now - start) < 200000UL); + + /* APBT is the only always on clocksource, it has to work! */ + if (t1 == apbt_read_clocksource(&clocksource_apbt)) + panic("APBT counter not counting. APBT disabled\n"); + + /* + * initialize and register APBT clocksource + * convert that to ns/clock cycle + * mult = (ns/c) * 2^APBT_SHIFT + */ + clocksource_apbt.mult = div_sc(MSEC_PER_SEC, + (unsigned long) apbt_freq, APBT_SHIFT); + clocksource_register(&clocksource_apbt); + + return 0; +} + +/* + * Early setup the APBT timer, only use timer 0 for booting then switch to + * per CPU timer if possible. + * returns 1 if per cpu apbt is setup + * returns 0 if no per cpu apbt is chosen + * panic if set up failed, this is the only platform timer on Moorestown. + */ +void __init apbt_time_init(void) +{ +#ifdef CONFIG_SMP + int i; + struct sfi_timer_table_entry *p_mtmr; + unsigned int percpu_timer; + struct apbt_dev *adev; +#endif + + if (apb_timer_block_enabled) + return; + apbt_set_mapping(); + if (apbt_virt_address) { + pr_debug("Found APBT version 0x%lx\n",\ + apbt_readl_reg(APBTMRS_COMP_VERSION)); + } else + goto out_noapbt; + /* + * Read the frequency and check for a sane value, for ESL model + * we extend the possible clock range to allow time scaling. + */ + + if (apbt_freq < APBT_MIN_FREQ || apbt_freq > APBT_MAX_FREQ) { + pr_debug("APBT has invalid freq 0x%llx\n", apbt_freq); + goto out_noapbt; + } + if (apbt_clocksource_register()) { + pr_debug("APBT has failed to register clocksource\n"); + goto out_noapbt; + } + if (!apbt_clockevent_register()) + apb_timer_block_enabled = 1; + else { + pr_debug("APBT has failed to register clockevent\n"); + goto out_noapbt; + } +#ifdef CONFIG_SMP + /* kernel cmdline disable apb timer, so we will use lapic timers */ + if (disable_apbt_percpu) { + printk(KERN_INFO "apbt: disabled per cpu timer\n"); + return; + } + pr_debug("%s: %d CPUs online\n", __func__, num_online_cpus()); + if (num_possible_cpus() <= sfi_mtimer_num) { + percpu_timer = 1; + apbt_num_timers_used = num_possible_cpus(); + } else { + percpu_timer = 0; + apbt_num_timers_used = 1; + adev = &per_cpu(cpu_apbt_dev, 0); + adev->flags &= ~APBT_DEV_USED; + } + pr_debug("%s: %d APB timers used\n", __func__, apbt_num_timers_used); + + /* here we set up per CPU timer data structure */ + apbt_devs = kzalloc(sizeof(struct apbt_dev) * apbt_num_timers_used, + GFP_KERNEL); + if (!apbt_devs) { + printk(KERN_ERR "Failed to allocate APB timer devices\n"); + return; + } + for (i = 0; i < apbt_num_timers_used; i++) { + adev = &per_cpu(cpu_apbt_dev, i); + adev->num = i; + adev->cpu = i; + p_mtmr = sfi_get_mtmr(i); + if (p_mtmr) { + adev->tick = p_mtmr->freq_hz; + adev->irq = p_mtmr->irq; + } else + printk(KERN_ERR "Failed to get timer for cpu %d\n", i); + adev->count = 0; + sprintf(adev->name, "apbt%d", i); + } +#endif + + return; + +out_noapbt: + apbt_clear_mapping(); + apb_timer_block_enabled = 0; + panic("failed to enable APB timer\n"); +} + +static inline void apbt_disable(int n) +{ + if (is_apbt_capable()) { + unsigned long ctrl = apbt_readl(n, APBTMR_N_CONTROL); + ctrl &= ~APBTMR_CONTROL_ENABLE; + apbt_writel(n, ctrl, APBTMR_N_CONTROL); + } +} + +/* called before apb_timer_enable, use early map */ +unsigned long apbt_quick_calibrate() +{ + int i, scale; + u64 old, new; + cycle_t t1, t2; + unsigned long khz = 0; + u32 loop, shift; + + apbt_set_mapping(); + apbt_start_counter(phy_cs_timer_id); + + /* check if the timer can count down, otherwise return */ + old = apbt_read_clocksource(&clocksource_apbt); + i = 10000; + while (--i) { + if (old != apbt_read_clocksource(&clocksource_apbt)) + break; + } + if (!i) + goto failed; + + /* count 16 ms */ + loop = (apbt_freq * 1000) << 4; + + /* restart the timer to ensure it won't get to 0 in the calibration */ + apbt_start_counter(phy_cs_timer_id); + + old = apbt_read_clocksource(&clocksource_apbt); + old += loop; + + t1 = __native_read_tsc(); + + do { + new = apbt_read_clocksource(&clocksource_apbt); + } while (new < old); + + t2 = __native_read_tsc(); + + shift = 5; + if (unlikely(loop >> shift == 0)) { + printk(KERN_INFO + "APBT TSC calibration failed, not enough resolution\n"); + return 0; + } + scale = (int)div_u64((t2 - t1), loop >> shift); + khz = (scale * apbt_freq * 1000) >> shift; + printk(KERN_INFO "TSC freq calculated by APB timer is %lu khz\n", khz); + return khz; +failed: + return 0; +} diff --git a/arch/x86/kernel/aperture_64.c b/arch/x86/kernel/aperture_64.c index f147a95fd84..b5d8b0bcf23 100644 --- a/arch/x86/kernel/aperture_64.c +++ b/arch/x86/kernel/aperture_64.c @@ -31,7 +31,6 @@ #include <asm/x86_init.h> int gart_iommu_aperture; -EXPORT_SYMBOL_GPL(gart_iommu_aperture); int gart_iommu_aperture_disabled __initdata; int gart_iommu_aperture_allowed __initdata; @@ -394,6 +393,7 @@ void __init gart_iommu_hole_init(void) for (i = 0; i < ARRAY_SIZE(bus_dev_ranges); i++) { int bus; int dev_base, dev_limit; + u32 ctl; bus = bus_dev_ranges[i].bus; dev_base = bus_dev_ranges[i].dev_base; @@ -407,7 +407,19 @@ void __init gart_iommu_hole_init(void) gart_iommu_aperture = 1; x86_init.iommu.iommu_init = gart_iommu_init; - aper_order = (read_pci_config(bus, slot, 3, AMD64_GARTAPERTURECTL) >> 1) & 7; + ctl = read_pci_config(bus, slot, 3, + AMD64_GARTAPERTURECTL); + + /* + * Before we do anything else disable the GART. It may + * still be enabled if we boot into a crash-kernel here. + * Reconfiguring the GART while it is enabled could have + * unknown side-effects. + */ + ctl &= ~GARTEN; + write_pci_config(bus, slot, 3, AMD64_GARTAPERTURECTL, ctl); + + aper_order = (ctl >> 1) & 7; aper_size = (32 * 1024 * 1024) << aper_order; aper_base = read_pci_config(bus, slot, 3, AMD64_GARTAPERTUREBASE) & 0x7fff; aper_base <<= 25; diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 3987e4408f7..c02cc692985 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -51,6 +51,7 @@ #include <asm/smp.h> #include <asm/mce.h> #include <asm/kvm_para.h> +#include <asm/tsc.h> unsigned int num_processors; @@ -581,7 +582,7 @@ calibrate_by_pmtimer(long deltapm, long *delta, long *deltatsc) res = (((u64)(*deltatsc)) * pm_100ms); do_div(res, deltapm); apic_printk(APIC_VERBOSE, "TSC delta adjusted to " - "PM-Timer: %lu (%ld) \n", + "PM-Timer: %lu (%ld)\n", (unsigned long)res, *deltatsc); *deltatsc = (long)res; } @@ -1151,8 +1152,13 @@ static void __cpuinit lapic_setup_esr(void) */ void __cpuinit setup_local_APIC(void) { - unsigned int value; - int i, j; + unsigned int value, queued; + int i, j, acked = 0; + unsigned long long tsc = 0, ntsc; + long long max_loops = cpu_khz; + + if (cpu_has_tsc) + rdtscll(tsc); if (disable_apic) { arch_disable_smp_support(); @@ -1204,13 +1210,32 @@ void __cpuinit setup_local_APIC(void) * the interrupt. Hence a vector might get locked. It was noticed * for timer irq (vector 0x31). Issue an extra EOI to clear ISR. */ - for (i = APIC_ISR_NR - 1; i >= 0; i--) { - value = apic_read(APIC_ISR + i*0x10); - for (j = 31; j >= 0; j--) { - if (value & (1<<j)) - ack_APIC_irq(); + do { + queued = 0; + for (i = APIC_ISR_NR - 1; i >= 0; i--) + queued |= apic_read(APIC_IRR + i*0x10); + + for (i = APIC_ISR_NR - 1; i >= 0; i--) { + value = apic_read(APIC_ISR + i*0x10); + for (j = 31; j >= 0; j--) { + if (value & (1<<j)) { + ack_APIC_irq(); + acked++; + } + } } - } + if (acked > 256) { + printk(KERN_ERR "LAPIC pending interrupts after %d EOI\n", + acked); + break; + } + if (cpu_has_tsc) { + rdtscll(ntsc); + max_loops = (cpu_khz << 10) - (ntsc - tsc); + } else + max_loops--; + } while (queued && max_loops > 0); + WARN_ON(max_loops <= 0); /* * Now that we are all set up, enable the APIC @@ -1390,7 +1415,7 @@ void __init enable_IR_x2apic(void) } local_irq_save(flags); - mask_8259A(); + legacy_pic->mask_all(); mask_IO_APIC_setup(ioapic_entries); if (dmar_table_init_ret) @@ -1422,7 +1447,7 @@ void __init enable_IR_x2apic(void) nox2apic: if (!ret) /* IR enabling failed */ restore_IO_APIC_setup(ioapic_entries); - unmask_8259A(); + legacy_pic->restore_mask(); local_irq_restore(flags); out: @@ -1640,8 +1665,8 @@ int __init APIC_init_uniprocessor(void) } #endif +#ifndef CONFIG_SMP enable_IR_x2apic(); -#ifdef CONFIG_X86_64 default_setup_apic_routing(); #endif @@ -1891,21 +1916,6 @@ void __cpuinit generic_processor_info(int apicid, int version) if (apicid > max_physical_apicid) max_physical_apicid = apicid; -#ifdef CONFIG_X86_32 - if (num_processors > 8) { - switch (boot_cpu_data.x86_vendor) { - case X86_VENDOR_INTEL: - if (!APIC_XAPIC(version)) { - def_to_bigsmp = 0; - break; - } - /* If P4 and above fall through */ - case X86_VENDOR_AMD: - def_to_bigsmp = 1; - } - } -#endif - #if defined(CONFIG_SMP) || defined(CONFIG_X86_64) early_per_cpu(x86_cpu_to_apicid, cpu) = apicid; early_per_cpu(x86_bios_cpu_apicid, cpu) = apicid; @@ -2035,7 +2045,7 @@ static int lapic_resume(struct sys_device *dev) } mask_IO_APIC_setup(ioapic_entries); - mask_8259A(); + legacy_pic->mask_all(); } if (x2apic_mode) @@ -2079,7 +2089,7 @@ static int lapic_resume(struct sys_device *dev) if (intr_remapping_enabled) { reenable_intr_remapping(x2apic_mode); - unmask_8259A(); + legacy_pic->restore_mask(); restore_IO_APIC_setup(ioapic_entries); free_ioapic_entries(ioapic_entries); } diff --git a/arch/x86/kernel/apic/apic_flat_64.c b/arch/x86/kernel/apic/apic_flat_64.c index e3c3d820c32..09d3b17ce0c 100644 --- a/arch/x86/kernel/apic/apic_flat_64.c +++ b/arch/x86/kernel/apic/apic_flat_64.c @@ -223,7 +223,7 @@ struct apic apic_flat = { }; /* - * Physflat mode is used when there are more than 8 CPUs on a AMD system. + * Physflat mode is used when there are more than 8 CPUs on a system. * We cannot use logical delivery in this case because the mask * overflows, so use physical mode. */ diff --git a/arch/x86/kernel/apic/es7000_32.c b/arch/x86/kernel/apic/es7000_32.c index dd2b5f26464..425e53a87fe 100644 --- a/arch/x86/kernel/apic/es7000_32.c +++ b/arch/x86/kernel/apic/es7000_32.c @@ -42,6 +42,7 @@ #include <linux/errno.h> #include <linux/acpi.h> #include <linux/init.h> +#include <linux/gfp.h> #include <linux/nmi.h> #include <linux/smp.h> #include <linux/io.h> @@ -130,24 +131,6 @@ int es7000_plat; static unsigned int base; -static int -es7000_rename_gsi(int ioapic, int gsi) -{ - if (es7000_plat == ES7000_ZORRO) - return gsi; - - if (!base) { - int i; - for (i = 0; i < nr_ioapics; i++) - base += nr_ioapic_registers[i]; - } - - if (!ioapic && (gsi < 16)) - gsi += base; - - return gsi; -} - static int __cpuinit wakeup_secondary_cpu_via_mip(int cpu, unsigned long eip) { unsigned long vect = 0, psaival = 0; @@ -189,7 +172,6 @@ static void setup_unisys(void) es7000_plat = ES7000_ZORRO; else es7000_plat = ES7000_CLASSIC; - ioapic_renumber_irq = es7000_rename_gsi; } /* diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c index 53243ca7816..33f3563a2a5 100644 --- a/arch/x86/kernel/apic/io_apic.c +++ b/arch/x86/kernel/apic/io_apic.c @@ -36,6 +36,7 @@ #include <linux/freezer.h> #include <linux/kthread.h> #include <linux/jiffies.h> /* time_after() */ +#include <linux/slab.h> #ifdef CONFIG_ACPI #include <acpi/acpi_bus.h> #endif @@ -73,8 +74,8 @@ */ int sis_apic_bug = -1; -static DEFINE_SPINLOCK(ioapic_lock); -static DEFINE_SPINLOCK(vector_lock); +static DEFINE_RAW_SPINLOCK(ioapic_lock); +static DEFINE_RAW_SPINLOCK(vector_lock); /* * # of IRQ routing registers @@ -88,14 +89,15 @@ int nr_ioapics; /* IO APIC gsi routing info */ struct mp_ioapic_gsi mp_gsi_routing[MAX_IO_APICS]; +/* The last gsi number used */ +u32 gsi_end; + /* MP IRQ source entries */ struct mpc_intsrc mp_irqs[MAX_IRQ_SOURCES]; /* # of MP IRQ source entries */ int mp_irq_entries; -/* Number of legacy interrupts */ -static int nr_legacy_irqs __read_mostly = NR_IRQS_LEGACY; /* GSI interrupts */ static int nr_irqs_gsi = NR_IRQS_LEGACY; @@ -140,33 +142,10 @@ static struct irq_pin_list *get_one_free_irq_2_pin(int node) /* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ #ifdef CONFIG_SPARSE_IRQ -static struct irq_cfg irq_cfgx[] = { +static struct irq_cfg irq_cfgx[NR_IRQS_LEGACY]; #else -static struct irq_cfg irq_cfgx[NR_IRQS] = { +static struct irq_cfg irq_cfgx[NR_IRQS]; #endif - [0] = { .vector = IRQ0_VECTOR, }, - [1] = { .vector = IRQ1_VECTOR, }, - [2] = { .vector = IRQ2_VECTOR, }, - [3] = { .vector = IRQ3_VECTOR, }, - [4] = { .vector = IRQ4_VECTOR, }, - [5] = { .vector = IRQ5_VECTOR, }, - [6] = { .vector = IRQ6_VECTOR, }, - [7] = { .vector = IRQ7_VECTOR, }, - [8] = { .vector = IRQ8_VECTOR, }, - [9] = { .vector = IRQ9_VECTOR, }, - [10] = { .vector = IRQ10_VECTOR, }, - [11] = { .vector = IRQ11_VECTOR, }, - [12] = { .vector = IRQ12_VECTOR, }, - [13] = { .vector = IRQ13_VECTOR, }, - [14] = { .vector = IRQ14_VECTOR, }, - [15] = { .vector = IRQ15_VECTOR, }, -}; - -void __init io_apic_disable_legacy(void) -{ - nr_legacy_irqs = 0; - nr_irqs_gsi = 0; -} int __init arch_early_irq_init(void) { @@ -176,6 +155,11 @@ int __init arch_early_irq_init(void) int node; int i; + if (!legacy_pic->nr_legacy_irqs) { + nr_irqs_gsi = 0; + io_apic_irqs = ~0UL; + } + cfg = irq_cfgx; count = ARRAY_SIZE(irq_cfgx); node= cpu_to_node(boot_cpu_id); @@ -185,8 +169,14 @@ int __init arch_early_irq_init(void) desc->chip_data = &cfg[i]; zalloc_cpumask_var_node(&cfg[i].domain, GFP_NOWAIT, node); zalloc_cpumask_var_node(&cfg[i].old_domain, GFP_NOWAIT, node); - if (i < nr_legacy_irqs) - cpumask_setall(cfg[i].domain); + /* + * For legacy IRQ's, start with assigning irq0 to irq15 to + * IRQ0_VECTOR to IRQ15_VECTOR on cpu 0. + */ + if (i < legacy_pic->nr_legacy_irqs) { + cfg[i].vector = IRQ0_VECTOR + i; + cpumask_set_cpu(0, cfg[i].domain); + } } return 0; @@ -406,7 +396,7 @@ static bool io_apic_level_ack_pending(struct irq_cfg *cfg) struct irq_pin_list *entry; unsigned long flags; - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); for_each_irq_pin(entry, cfg->irq_2_pin) { unsigned int reg; int pin; @@ -415,11 +405,11 @@ static bool io_apic_level_ack_pending(struct irq_cfg *cfg) reg = io_apic_read(entry->apic, 0x10 + pin*2); /* Is the remote IRR bit set? */ if (reg & IO_APIC_REDIR_REMOTE_IRR) { - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); return true; } } - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); return false; } @@ -433,10 +423,10 @@ static struct IO_APIC_route_entry ioapic_read_entry(int apic, int pin) { union entry_union eu; unsigned long flags; - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); eu.w1 = io_apic_read(apic, 0x10 + 2 * pin); eu.w2 = io_apic_read(apic, 0x11 + 2 * pin); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); return eu.entry; } @@ -459,9 +449,9 @@ __ioapic_write_entry(int apic, int pin, struct IO_APIC_route_entry e) void ioapic_write_entry(int apic, int pin, struct IO_APIC_route_entry e) { unsigned long flags; - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); __ioapic_write_entry(apic, pin, e); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); } /* @@ -474,10 +464,10 @@ static void ioapic_mask_entry(int apic, int pin) unsigned long flags; union entry_union eu = { .entry.mask = 1 }; - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); io_apic_write(apic, 0x10 + 2*pin, eu.w1); io_apic_write(apic, 0x11 + 2*pin, eu.w2); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); } /* @@ -604,9 +594,9 @@ static void mask_IO_APIC_irq_desc(struct irq_desc *desc) BUG_ON(!cfg); - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); __mask_IO_APIC_irq(cfg); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); } static void unmask_IO_APIC_irq_desc(struct irq_desc *desc) @@ -614,9 +604,9 @@ static void unmask_IO_APIC_irq_desc(struct irq_desc *desc) struct irq_cfg *cfg = desc->chip_data; unsigned long flags; - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); __unmask_IO_APIC_irq(cfg); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); } static void mask_IO_APIC_irq(unsigned int irq) @@ -865,7 +855,7 @@ static int __init find_isa_irq_apic(int irq, int type) */ static int EISA_ELCR(unsigned int irq) { - if (irq < nr_legacy_irqs) { + if (irq < legacy_pic->nr_legacy_irqs) { unsigned int port = 0x4d0 + (irq >> 3); return (inb(port) >> (irq & 7)) & 1; } @@ -1026,10 +1016,9 @@ static inline int irq_trigger(int idx) return MPBIOS_trigger(idx); } -int (*ioapic_renumber_irq)(int ioapic, int irq); static int pin_2_irq(int idx, int apic, int pin) { - int irq, i; + int irq; int bus = mp_irqs[idx].srcbus; /* @@ -1041,18 +1030,12 @@ static int pin_2_irq(int idx, int apic, int pin) if (test_bit(bus, mp_bus_not_pci)) { irq = mp_irqs[idx].srcbusirq; } else { - /* - * PCI IRQs are mapped in order - */ - i = irq = 0; - while (i < apic) - irq += nr_ioapic_registers[i++]; - irq += pin; - /* - * For MPS mode, so far only needed by ES7000 platform - */ - if (ioapic_renumber_irq) - irq = ioapic_renumber_irq(apic, irq); + u32 gsi = mp_gsi_routing[apic].gsi_base + pin; + + if (gsi >= NR_IRQS_LEGACY) + irq = gsi; + else + irq = gsi_end + 1 + gsi; } #ifdef CONFIG_X86_32 @@ -1140,12 +1123,12 @@ void lock_vector_lock(void) /* Used to the online set of cpus does not change * during assign_irq_vector. */ - spin_lock(&vector_lock); + raw_spin_lock(&vector_lock); } void unlock_vector_lock(void) { - spin_unlock(&vector_lock); + raw_spin_unlock(&vector_lock); } static int @@ -1162,7 +1145,8 @@ __assign_irq_vector(int irq, struct irq_cfg *cfg, const struct cpumask *mask) * Also, we've got to be careful not to trash gate * 0x80, because int 0x80 is hm, kind of importantish. ;) */ - static int current_vector = FIRST_DEVICE_VECTOR, current_offset = 0; + static int current_vector = FIRST_EXTERNAL_VECTOR + VECTOR_OFFSET_START; + static int current_offset = VECTOR_OFFSET_START % 8; unsigned int old_vector; int cpu, err; cpumask_var_t tmp_mask; @@ -1198,7 +1182,7 @@ next: if (vector >= first_system_vector) { /* If out of vectors on large boxen, must share them. */ offset = (offset + 1) % 8; - vector = FIRST_DEVICE_VECTOR + offset; + vector = FIRST_EXTERNAL_VECTOR + offset; } if (unlikely(current_vector == vector)) continue; @@ -1232,9 +1216,9 @@ int assign_irq_vector(int irq, struct irq_cfg *cfg, const struct cpumask *mask) int err; unsigned long flags; - spin_lock_irqsave(&vector_lock, flags); + raw_spin_lock_irqsave(&vector_lock, flags); err = __assign_irq_vector(irq, cfg, mask); - spin_unlock_irqrestore(&vector_lock, flags); + raw_spin_unlock_irqrestore(&vector_lock, flags); return err; } @@ -1268,14 +1252,27 @@ static void __clear_irq_vector(int irq, struct irq_cfg *cfg) void __setup_vector_irq(int cpu) { /* Initialize vector_irq on a new cpu */ - /* This function must be called with vector_lock held */ int irq, vector; struct irq_cfg *cfg; struct irq_desc *desc; + /* + * vector_lock will make sure that we don't run into irq vector + * assignments that might be happening on another cpu in parallel, + * while we setup our initial vector to irq mappings. + */ + raw_spin_lock(&vector_lock); /* Mark the inuse vectors */ for_each_irq_desc(irq, desc) { cfg = desc->chip_data; + + /* + * If it is a legacy IRQ handled by the legacy PIC, this cpu + * will be part of the irq_cfg's domain. + */ + if (irq < legacy_pic->nr_legacy_irqs && !IO_APIC_IRQ(irq)) + cpumask_set_cpu(cpu, cfg->domain); + if (!cpumask_test_cpu(cpu, cfg->domain)) continue; vector = cfg->vector; @@ -1291,6 +1288,7 @@ void __setup_vector_irq(int cpu) if (!cpumask_test_cpu(cpu, cfg->domain)) per_cpu(vector_irq, cpu)[vector] = -1; } + raw_spin_unlock(&vector_lock); } static struct irq_chip ioapic_chip; @@ -1440,6 +1438,14 @@ static void setup_IO_APIC_irq(int apic_id, int pin, unsigned int irq, struct irq cfg = desc->chip_data; + /* + * For legacy irqs, cfg->domain starts with cpu 0 for legacy + * controllers like 8259. Now that IO-APIC can handle this irq, update + * the cfg->domain. + */ + if (irq < legacy_pic->nr_legacy_irqs && cpumask_test_cpu(0, cfg->domain)) + apic->vector_allocation_domain(0, cfg->domain); + if (assign_irq_vector(irq, cfg, apic->target_cpus())) return; @@ -1461,8 +1467,8 @@ static void setup_IO_APIC_irq(int apic_id, int pin, unsigned int irq, struct irq } ioapic_register_intr(irq, desc, trigger); - if (irq < nr_legacy_irqs) - disable_8259A_irq(irq); + if (irq < legacy_pic->nr_legacy_irqs) + legacy_pic->chip->mask(irq); ioapic_write_entry(apic_id, pin, entry); } @@ -1473,7 +1479,7 @@ static struct { static void __init setup_IO_APIC_irqs(void) { - int apic_id = 0, pin, idx, irq; + int apic_id, pin, idx, irq; int notcon = 0; struct irq_desc *desc; struct irq_cfg *cfg; @@ -1481,14 +1487,7 @@ static void __init setup_IO_APIC_irqs(void) apic_printk(APIC_VERBOSE, KERN_DEBUG "init IO_APIC IRQs\n"); -#ifdef CONFIG_ACPI - if (!acpi_disabled && acpi_ioapic) { - apic_id = mp_find_ioapic(0); - if (apic_id < 0) - apic_id = 0; - } -#endif - + for (apic_id = 0; apic_id < nr_ioapics; apic_id++) for (pin = 0; pin < nr_ioapic_registers[apic_id]; pin++) { idx = find_irq_entry(apic_id, pin, mp_INT); if (idx == -1) { @@ -1510,6 +1509,9 @@ static void __init setup_IO_APIC_irqs(void) irq = pin_2_irq(idx, apic_id, pin); + if ((apic_id > 0) && (irq > 16)) + continue; + /* * Skip the timer IRQ if there's a quirk handler * installed and if it returns 1: @@ -1539,6 +1541,56 @@ static void __init setup_IO_APIC_irqs(void) } /* + * for the gsit that is not in first ioapic + * but could not use acpi_register_gsi() + * like some special sci in IBM x3330 + */ +void setup_IO_APIC_irq_extra(u32 gsi) +{ + int apic_id = 0, pin, idx, irq; + int node = cpu_to_node(boot_cpu_id); + struct irq_desc *desc; + struct irq_cfg *cfg; + + /* + * Convert 'gsi' to 'ioapic.pin'. + */ + apic_id = mp_find_ioapic(gsi); + if (apic_id < 0) + return; + + pin = mp_find_ioapic_pin(apic_id, gsi); + idx = find_irq_entry(apic_id, pin, mp_INT); + if (idx == -1) + return; + + irq = pin_2_irq(idx, apic_id, pin); +#ifdef CONFIG_SPARSE_IRQ + desc = irq_to_desc(irq); + if (desc) + return; +#endif + desc = irq_to_desc_alloc_node(irq, node); + if (!desc) { + printk(KERN_INFO "can not get irq_desc for %d\n", irq); + return; + } + + cfg = desc->chip_data; + add_pin_to_irq_node(cfg, node, apic_id, pin); + + if (test_bit(pin, mp_ioapic_routing[apic_id].pin_programmed)) { + pr_debug("Pin %d-%d already programmed\n", + mp_ioapics[apic_id].apicid, pin); + return; + } + set_bit(pin, mp_ioapic_routing[apic_id].pin_programmed); + + setup_IO_APIC_irq(apic_id, pin, irq, desc, + irq_trigger(idx), irq_polarity(idx)); +} + +/* * Set up the timer pin, possibly with the 8259A-master behind. */ static void __init setup_timer_IRQ0_pin(unsigned int apic_id, unsigned int pin, @@ -1601,14 +1653,14 @@ __apicdebuginit(void) print_IO_APIC(void) for (apic = 0; apic < nr_ioapics; apic++) { - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); reg_00.raw = io_apic_read(apic, 0); reg_01.raw = io_apic_read(apic, 1); if (reg_01.bits.version >= 0x10) reg_02.raw = io_apic_read(apic, 2); if (reg_01.bits.version >= 0x20) reg_03.raw = io_apic_read(apic, 3); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); printk("\n"); printk(KERN_DEBUG "IO APIC #%d......\n", mp_ioapics[apic].apicid); @@ -1647,7 +1699,7 @@ __apicdebuginit(void) print_IO_APIC(void) printk(KERN_DEBUG ".... IRQ redirection table:\n"); printk(KERN_DEBUG " NR Dst Mask Trig IRR Pol" - " Stat Dmod Deli Vect: \n"); + " Stat Dmod Deli Vect:\n"); for (i = 0; i <= reg_01.bits.entries; i++) { struct IO_APIC_route_entry entry; @@ -1825,12 +1877,12 @@ __apicdebuginit(void) print_PIC(void) unsigned int v; unsigned long flags; - if (!nr_legacy_irqs) + if (!legacy_pic->nr_legacy_irqs) return; printk(KERN_DEBUG "\nprinting PIC contents\n"); - spin_lock_irqsave(&i8259A_lock, flags); + raw_spin_lock_irqsave(&i8259A_lock, flags); v = inb(0xa1) << 8 | inb(0x21); printk(KERN_DEBUG "... PIC IMR: %04x\n", v); @@ -1844,7 +1896,7 @@ __apicdebuginit(void) print_PIC(void) outb(0x0a,0xa0); outb(0x0a,0x20); - spin_unlock_irqrestore(&i8259A_lock, flags); + raw_spin_unlock_irqrestore(&i8259A_lock, flags); printk(KERN_DEBUG "... PIC ISR: %04x\n", v); @@ -1894,22 +1946,10 @@ static struct { int pin, apic; } ioapic_i8259 = { -1, -1 }; void __init enable_IO_APIC(void) { - union IO_APIC_reg_01 reg_01; int i8259_apic, i8259_pin; int apic; - unsigned long flags; - /* - * The number of IO-APIC IRQ registers (== #pins): - */ - for (apic = 0; apic < nr_ioapics; apic++) { - spin_lock_irqsave(&ioapic_lock, flags); - reg_01.raw = io_apic_read(apic, 1); - spin_unlock_irqrestore(&ioapic_lock, flags); - nr_ioapic_registers[apic] = reg_01.bits.entries+1; - } - - if (!nr_legacy_irqs) + if (!legacy_pic->nr_legacy_irqs) return; for(apic = 0; apic < nr_ioapics; apic++) { @@ -1966,7 +2006,7 @@ void disable_IO_APIC(void) */ clear_IO_APIC(); - if (!nr_legacy_irqs) + if (!legacy_pic->nr_legacy_irqs) return; /* @@ -2045,9 +2085,9 @@ void __init setup_ioapic_ids_from_mpc(void) for (apic_id = 0; apic_id < nr_ioapics; apic_id++) { /* Read the register 0 value */ - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); reg_00.raw = io_apic_read(apic_id, 0); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); old_id = mp_ioapics[apic_id].apicid; @@ -2106,16 +2146,16 @@ void __init setup_ioapic_ids_from_mpc(void) mp_ioapics[apic_id].apicid); reg_00.bits.ID = mp_ioapics[apic_id].apicid; - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); io_apic_write(apic_id, 0, reg_00.raw); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); /* * Sanity check */ - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); reg_00.raw = io_apic_read(apic_id, 0); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); if (reg_00.bits.ID != mp_ioapics[apic_id].apicid) printk("could not set ID!\n"); else @@ -2198,15 +2238,15 @@ static unsigned int startup_ioapic_irq(unsigned int irq) unsigned long flags; struct irq_cfg *cfg; - spin_lock_irqsave(&ioapic_lock, flags); - if (irq < nr_legacy_irqs) { - disable_8259A_irq(irq); - if (i8259A_irq_pending(irq)) + raw_spin_lock_irqsave(&ioapic_lock, flags); + if (irq < legacy_pic->nr_legacy_irqs) { + legacy_pic->chip->mask(irq); + if (legacy_pic->irq_pending(irq)) was_pending = 1; } cfg = irq_cfg(irq); __unmask_IO_APIC_irq(cfg); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); return was_pending; } @@ -2217,9 +2257,9 @@ static int ioapic_retrigger_irq(unsigned int irq) struct irq_cfg *cfg = irq_cfg(irq); unsigned long flags; - spin_lock_irqsave(&vector_lock, flags); + raw_spin_lock_irqsave(&vector_lock, flags); apic->send_IPI_mask(cpumask_of(cpumask_first(cfg->domain)), cfg->vector); - spin_unlock_irqrestore(&vector_lock, flags); + raw_spin_unlock_irqrestore(&vector_lock, flags); return 1; } @@ -2312,14 +2352,14 @@ set_ioapic_affinity_irq_desc(struct irq_desc *desc, const struct cpumask *mask) irq = desc->irq; cfg = desc->chip_data; - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); ret = set_desc_affinity(desc, mask, &dest); if (!ret) { /* Only the high 8 bits are valid. */ dest = SET_APIC_LOGICAL_ID(dest); __target_IO_APIC_irq(irq, dest, cfg); } - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); return ret; } @@ -2489,6 +2529,9 @@ void irq_force_complete_move(int irq) struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg = desc->chip_data; + if (!cfg) + return; + __irq_complete_move(&desc, cfg->vector); } #else @@ -2554,9 +2597,9 @@ static void eoi_ioapic_irq(struct irq_desc *desc) irq = desc->irq; cfg = desc->chip_data; - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); __eoi_ioapic_irq(irq, cfg); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); } static void ack_apic_level(unsigned int irq) @@ -2734,8 +2777,8 @@ static inline void init_IO_APIC_traps(void) * so default to an old-fashioned 8259 * interrupt if we can.. */ - if (irq < nr_legacy_irqs) - make_8259A_irq(irq); + if (irq < legacy_pic->nr_legacy_irqs) + legacy_pic->make_irq(irq); else /* Strange. Oh, well.. */ desc->chip = &no_irq_chip; @@ -2892,7 +2935,7 @@ static inline void __init check_timer(void) /* * get/set the timer IRQ vector: */ - disable_8259A_irq(0); + legacy_pic->chip->mask(0); assign_irq_vector(0, cfg, apic->target_cpus()); /* @@ -2905,7 +2948,7 @@ static inline void __init check_timer(void) * automatically. */ apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_EXTINT); - init_8259A(1); + legacy_pic->init(1); #ifdef CONFIG_X86_32 { unsigned int ver; @@ -2964,7 +3007,7 @@ static inline void __init check_timer(void) if (timer_irq_works()) { if (nmi_watchdog == NMI_IO_APIC) { setup_nmi(); - enable_8259A_irq(0); + legacy_pic->chip->unmask(0); } if (disable_timer_pin_1 > 0) clear_IO_APIC_pin(0, pin1); @@ -2987,14 +3030,14 @@ static inline void __init check_timer(void) */ replace_pin_at_irq_node(cfg, node, apic1, pin1, apic2, pin2); setup_timer_IRQ0_pin(apic2, pin2, cfg->vector); - enable_8259A_irq(0); + legacy_pic->chip->unmask(0); if (timer_irq_works()) { apic_printk(APIC_QUIET, KERN_INFO "....... works.\n"); timer_through_8259 = 1; if (nmi_watchdog == NMI_IO_APIC) { - disable_8259A_irq(0); + legacy_pic->chip->mask(0); setup_nmi(); - enable_8259A_irq(0); + legacy_pic->chip->unmask(0); } goto out; } @@ -3002,7 +3045,7 @@ static inline void __init check_timer(void) * Cleanup, just in case ... */ local_irq_disable(); - disable_8259A_irq(0); + legacy_pic->chip->mask(0); clear_IO_APIC_pin(apic2, pin2); apic_printk(APIC_QUIET, KERN_INFO "....... failed.\n"); } @@ -3021,22 +3064,22 @@ static inline void __init check_timer(void) lapic_register_intr(0, desc); apic_write(APIC_LVT0, APIC_DM_FIXED | cfg->vector); /* Fixed mode */ - enable_8259A_irq(0); + legacy_pic->chip->unmask(0); if (timer_irq_works()) { apic_printk(APIC_QUIET, KERN_INFO "..... works.\n"); goto out; } local_irq_disable(); - disable_8259A_irq(0); + legacy_pic->chip->mask(0); apic_write(APIC_LVT0, APIC_LVT_MASKED | APIC_DM_FIXED | cfg->vector); apic_printk(APIC_QUIET, KERN_INFO "..... failed.\n"); apic_printk(APIC_QUIET, KERN_INFO "...trying to set up timer as ExtINT IRQ...\n"); - init_8259A(0); - make_8259A_irq(0); + legacy_pic->init(0); + legacy_pic->make_irq(0); apic_write(APIC_LVT0, APIC_DM_EXTINT); unlock_ExtINT_logic(); @@ -3078,7 +3121,7 @@ void __init setup_IO_APIC(void) /* * calling enable_IO_APIC() is moved to setup_local_APIC for BP */ - io_apic_irqs = nr_legacy_irqs ? ~PIC_IRQS : ~0UL; + io_apic_irqs = legacy_pic->nr_legacy_irqs ? ~PIC_IRQS : ~0UL; apic_printk(APIC_VERBOSE, "ENABLING IO-APIC IRQs\n"); /* @@ -3089,7 +3132,7 @@ void __init setup_IO_APIC(void) sync_Arb_IDs(); setup_IO_APIC_irqs(); init_IO_APIC_traps(); - if (nr_legacy_irqs) + if (legacy_pic->nr_legacy_irqs) check_timer(); } @@ -3138,13 +3181,13 @@ static int ioapic_resume(struct sys_device *dev) data = container_of(dev, struct sysfs_ioapic_data, dev); entry = data->entry; - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); reg_00.raw = io_apic_read(dev->id, 0); if (reg_00.bits.ID != mp_ioapics[dev->id].apicid) { reg_00.bits.ID = mp_ioapics[dev->id].apicid; io_apic_write(dev->id, 0, reg_00.raw); } - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); for (i = 0; i < nr_ioapic_registers[dev->id]; i++) ioapic_write_entry(dev->id, i, entry[i]); @@ -3207,7 +3250,7 @@ unsigned int create_irq_nr(unsigned int irq_want, int node) if (irq_want < nr_irqs_gsi) irq_want = nr_irqs_gsi; - spin_lock_irqsave(&vector_lock, flags); + raw_spin_lock_irqsave(&vector_lock, flags); for (new = irq_want; new < nr_irqs; new++) { desc_new = irq_to_desc_alloc_node(new, node); if (!desc_new) { @@ -3226,14 +3269,11 @@ unsigned int create_irq_nr(unsigned int irq_want, int node) irq = new; break; } - spin_unlock_irqrestore(&vector_lock, flags); + raw_spin_unlock_irqrestore(&vector_lock, flags); + + if (irq > 0) + dynamic_irq_init_keep_chip_data(irq); - if (irq > 0) { - dynamic_irq_init(irq); - /* restore it, in case dynamic_irq_init clear it */ - if (desc_new) - desc_new->chip_data = cfg_new; - } return irq; } @@ -3255,20 +3295,13 @@ int create_irq(void) void destroy_irq(unsigned int irq) { unsigned long flags; - struct irq_cfg *cfg; - struct irq_desc *desc; - /* store it, in case dynamic_irq_cleanup clear it */ - desc = irq_to_desc(irq); - cfg = desc->chip_data; - dynamic_irq_cleanup(irq); - /* connect back irq_cfg */ - desc->chip_data = cfg; + dynamic_irq_cleanup_keep_chip_data(irq); free_irte(irq); - spin_lock_irqsave(&vector_lock, flags); - __clear_irq_vector(irq, cfg); - spin_unlock_irqrestore(&vector_lock, flags); + raw_spin_lock_irqsave(&vector_lock, flags); + __clear_irq_vector(irq, get_irq_chip_data(irq)); + raw_spin_unlock_irqrestore(&vector_lock, flags); } /* @@ -3805,31 +3838,24 @@ int __init io_apic_get_redir_entries (int ioapic) union IO_APIC_reg_01 reg_01; unsigned long flags; - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); reg_01.raw = io_apic_read(ioapic, 1); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); - return reg_01.bits.entries; + /* The register returns the maximum index redir index + * supported, which is one less than the total number of redir + * entries. + */ + return reg_01.bits.entries + 1; } void __init probe_nr_irqs_gsi(void) { - int nr = 0; + int nr; - nr = acpi_probe_gsi(); - if (nr > nr_irqs_gsi) { + nr = gsi_end + 1 + NR_IRQS_LEGACY; + if (nr > nr_irqs_gsi) nr_irqs_gsi = nr; - } else { - /* for acpi=off or acpi is not compiled in */ - int idx; - - nr = 0; - for (idx = 0; idx < nr_ioapics; idx++) - nr += io_apic_get_redir_entries(idx) + 1; - - if (nr > nr_irqs_gsi) - nr_irqs_gsi = nr; - } printk(KERN_DEBUG "nr_irqs_gsi: %d\n", nr_irqs_gsi); } @@ -3890,7 +3916,7 @@ static int __io_apic_set_pci_routing(struct device *dev, int irq, /* * IRQs < 16 are already in the irq_2_pin[] map */ - if (irq >= nr_legacy_irqs) { + if (irq >= legacy_pic->nr_legacy_irqs) { cfg = desc->chip_data; if (add_pin_to_irq_node_nopanic(cfg, node, ioapic, pin)) { printk(KERN_INFO "can not add pin %d for irq %d\n", @@ -3969,9 +3995,9 @@ int __init io_apic_get_unique_id(int ioapic, int apic_id) if (physids_empty(apic_id_map)) apic->ioapic_phys_id_map(&phys_cpu_present_map, &apic_id_map); - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); reg_00.raw = io_apic_read(ioapic, 0); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); if (apic_id >= get_physical_broadcast()) { printk(KERN_WARNING "IOAPIC[%d]: Invalid apic_id %d, trying " @@ -4005,10 +4031,10 @@ int __init io_apic_get_unique_id(int ioapic, int apic_id) if (reg_00.bits.ID != apic_id) { reg_00.bits.ID = apic_id; - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); io_apic_write(ioapic, 0, reg_00.raw); reg_00.raw = io_apic_read(ioapic, 0); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); /* Sanity check */ if (reg_00.bits.ID != apic_id) { @@ -4029,29 +4055,34 @@ int __init io_apic_get_version(int ioapic) union IO_APIC_reg_01 reg_01; unsigned long flags; - spin_lock_irqsave(&ioapic_lock, flags); + raw_spin_lock_irqsave(&ioapic_lock, flags); reg_01.raw = io_apic_read(ioapic, 1); - spin_unlock_irqrestore(&ioapic_lock, flags); + raw_spin_unlock_irqrestore(&ioapic_lock, flags); return reg_01.bits.version; } -int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity) +int acpi_get_override_irq(u32 gsi, int *trigger, int *polarity) { - int i; + int ioapic, pin, idx; if (skip_ioapic_setup) return -1; - for (i = 0; i < mp_irq_entries; i++) - if (mp_irqs[i].irqtype == mp_INT && - mp_irqs[i].srcbusirq == bus_irq) - break; - if (i >= mp_irq_entries) + ioapic = mp_find_ioapic(gsi); + if (ioapic < 0) + return -1; + + pin = mp_find_ioapic_pin(ioapic, gsi); + if (pin < 0) + return -1; + + idx = find_irq_entry(ioapic, pin, mp_INT); + if (idx < 0) return -1; - *trigger = irq_trigger(i); - *polarity = irq_polarity(i); + *trigger = irq_trigger(idx); + *polarity = irq_polarity(idx); return 0; } @@ -4063,27 +4094,23 @@ int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity) #ifdef CONFIG_SMP void __init setup_ioapic_dest(void) { - int pin, ioapic = 0, irq, irq_entry; + int pin, ioapic, irq, irq_entry; struct irq_desc *desc; const struct cpumask *mask; if (skip_ioapic_setup == 1) return; -#ifdef CONFIG_ACPI - if (!acpi_disabled && acpi_ioapic) { - ioapic = mp_find_ioapic(0); - if (ioapic < 0) - ioapic = 0; - } -#endif - + for (ioapic = 0; ioapic < nr_ioapics; ioapic++) for (pin = 0; pin < nr_ioapic_registers[ioapic]; pin++) { irq_entry = find_irq_entry(ioapic, pin, mp_INT); if (irq_entry == -1) continue; irq = pin_2_irq(irq_entry, ioapic, pin); + if ((ioapic > 0) && (irq > 16)) + continue; + desc = irq_to_desc(irq); /* @@ -4196,7 +4223,7 @@ void __init ioapic_insert_resources(void) } } -int mp_find_ioapic(int gsi) +int mp_find_ioapic(u32 gsi) { int i = 0; @@ -4211,7 +4238,7 @@ int mp_find_ioapic(int gsi) return -1; } -int mp_find_ioapic_pin(int ioapic, int gsi) +int mp_find_ioapic_pin(int ioapic, u32 gsi) { if (WARN_ON(ioapic == -1)) return -1; @@ -4239,6 +4266,7 @@ static int bad_ioapic(unsigned long address) void __init mp_register_ioapic(int id, u32 address, u32 gsi_base) { int idx = 0; + int entries; if (bad_ioapic(address)) return; @@ -4257,9 +4285,17 @@ void __init mp_register_ioapic(int id, u32 address, u32 gsi_base) * Build basic GSI lookup table to facilitate gsi->io_apic lookups * and to prevent reprogramming of IOAPIC pins (PCI GSIs). */ + entries = io_apic_get_redir_entries(idx); mp_gsi_routing[idx].gsi_base = gsi_base; - mp_gsi_routing[idx].gsi_end = gsi_base + - io_apic_get_redir_entries(idx); + mp_gsi_routing[idx].gsi_end = gsi_base + entries - 1; + + /* + * The number of IO-APIC IRQ registers (== #pins): + */ + nr_ioapic_registers[idx] = entries; + + if (mp_gsi_routing[idx].gsi_end > gsi_end) + gsi_end = mp_gsi_routing[idx].gsi_end; printk(KERN_INFO "IOAPIC[%d]: apic_id %d, version %d, address 0x%x, " "GSI %d-%d\n", idx, mp_ioapics[idx].apicid, @@ -4268,3 +4304,24 @@ void __init mp_register_ioapic(int id, u32 address, u32 gsi_base) nr_ioapics++; } + +/* Enable IOAPIC early just for system timer */ +void __init pre_init_apic_IRQ0(void) +{ + struct irq_cfg *cfg; + struct irq_desc *desc; + + printk(KERN_INFO "Early APIC setup for system timer0\n"); +#ifndef CONFIG_SMP + phys_cpu_present_map = physid_mask_of_physid(boot_cpu_physical_apicid); +#endif + desc = irq_to_desc_alloc_node(0, 0); + + setup_local_APIC(); + + cfg = irq_cfg(0); + add_pin_to_irq_node(cfg, 0, 0, 0); + set_irq_chip_and_handler_name(0, &ioapic_chip, handle_edge_irq, "edge"); + + setup_IO_APIC_irq(0, 0, 0, desc, 0, 0); +} diff --git a/arch/x86/kernel/apic/nmi.c b/arch/x86/kernel/apic/nmi.c index 0159a69396c..1edaf15c0b8 100644 --- a/arch/x86/kernel/apic/nmi.c +++ b/arch/x86/kernel/apic/nmi.c @@ -18,6 +18,7 @@ #include <linux/delay.h> #include <linux/interrupt.h> #include <linux/module.h> +#include <linux/slab.h> #include <linux/sysdev.h> #include <linux/sysctl.h> #include <linux/percpu.h> @@ -177,7 +178,7 @@ int __init check_nmi_watchdog(void) error: if (nmi_watchdog == NMI_IO_APIC) { if (!timer_through_8259) - disable_8259A_irq(0); + legacy_pic->chip->mask(0); on_each_cpu(__acpi_nmi_disable, NULL, 1); } @@ -416,13 +417,13 @@ nmi_watchdog_tick(struct pt_regs *regs, unsigned reason) /* We can be called before check_nmi_watchdog, hence NULL check. */ if (cpumask_test_cpu(cpu, to_cpumask(backtrace_mask))) { - static DEFINE_SPINLOCK(lock); /* Serialise the printks */ + static DEFINE_RAW_SPINLOCK(lock); /* Serialise the printks */ - spin_lock(&lock); + raw_spin_lock(&lock); printk(KERN_WARNING "NMI backtrace for cpu %d\n", cpu); show_regs(regs); dump_stack(); - spin_unlock(&lock); + raw_spin_unlock(&lock); cpumask_clear_cpu(cpu, to_cpumask(backtrace_mask)); rc = 1; @@ -438,8 +439,8 @@ nmi_watchdog_tick(struct pt_regs *regs, unsigned reason) * Ayiee, looks like this CPU is stuck ... * wait a few IRQs (5 seconds) before doing the oops ... */ - __this_cpu_inc(per_cpu_var(alert_counter)); - if (__this_cpu_read(per_cpu_var(alert_counter)) == 5 * nmi_hz) + __this_cpu_inc(alert_counter); + if (__this_cpu_read(alert_counter) == 5 * nmi_hz) /* * die_nmi will return ONLY if NOTIFY_STOP happens.. */ @@ -447,7 +448,7 @@ nmi_watchdog_tick(struct pt_regs *regs, unsigned reason) regs, panic_on_timeout); } else { __get_cpu_var(last_irq_sum) = sum; - __this_cpu_write(per_cpu_var(alert_counter), 0); + __this_cpu_write(alert_counter, 0); } /* see if the nmi watchdog went off */ diff --git a/arch/x86/kernel/apic/numaq_32.c b/arch/x86/kernel/apic/numaq_32.c index 98c4665f251..3e28401f161 100644 --- a/arch/x86/kernel/apic/numaq_32.c +++ b/arch/x86/kernel/apic/numaq_32.c @@ -225,7 +225,7 @@ static void __init smp_read_mpc_oem(struct mpc_table *mpc) mpc_record = 0; printk(KERN_INFO - "Found an OEM MPC table at %8p - parsing it ... \n", oemtable); + "Found an OEM MPC table at %8p - parsing it...\n", oemtable); if (memcmp(oemtable->signature, MPC_OEM_SIGNATURE, 4)) { printk(KERN_WARNING @@ -277,6 +277,7 @@ static __init void early_check_numaq(void) x86_init.mpparse.mpc_oem_pci_bus = mpc_oem_pci_bus; x86_init.mpparse.mpc_oem_bus_info = mpc_oem_bus_info; x86_init.timers.tsc_pre_init = numaq_tsc_init; + x86_init.pci.init = pci_numaq_init; } } diff --git a/arch/x86/kernel/apic/probe_32.c b/arch/x86/kernel/apic/probe_32.c index 1a6559f6768..99d2fe01608 100644 --- a/arch/x86/kernel/apic/probe_32.c +++ b/arch/x86/kernel/apic/probe_32.c @@ -52,7 +52,32 @@ static int __init print_ipi_mode(void) } late_initcall(print_ipi_mode); -void default_setup_apic_routing(void) +void __init default_setup_apic_routing(void) +{ + int version = apic_version[boot_cpu_physical_apicid]; + + if (num_possible_cpus() > 8) { + switch (boot_cpu_data.x86_vendor) { + case X86_VENDOR_INTEL: + if (!APIC_XAPIC(version)) { + def_to_bigsmp = 0; + break; + } + /* If P4 and above fall through */ + case X86_VENDOR_AMD: + def_to_bigsmp = 1; + } + } + +#ifdef CONFIG_X86_BIGSMP + generic_bigsmp_probe(); +#endif + + if (apic->setup_apic_routing) + apic->setup_apic_routing(); +} + +static void setup_apic_flat_routing(void) { #ifdef CONFIG_X86_IO_APIC printk(KERN_INFO @@ -103,7 +128,7 @@ struct apic apic_default = { .init_apic_ldr = default_init_apic_ldr, .ioapic_phys_id_map = default_ioapic_phys_id_map, - .setup_apic_routing = default_setup_apic_routing, + .setup_apic_routing = setup_apic_flat_routing, .multi_timer_check = NULL, .apicid_to_node = default_apicid_to_node, .cpu_to_logical_apicid = default_cpu_to_logical_apicid, diff --git a/arch/x86/kernel/apic/probe_64.c b/arch/x86/kernel/apic/probe_64.c index 450fe2064a1..83e9be4778e 100644 --- a/arch/x86/kernel/apic/probe_64.c +++ b/arch/x86/kernel/apic/probe_64.c @@ -67,7 +67,7 @@ void __init default_setup_apic_routing(void) } #endif - if (apic == &apic_flat && num_processors > 8) + if (apic == &apic_flat && num_possible_cpus() > 8) apic = &apic_physflat; printk(KERN_INFO "Setting APIC routing to %s\n", apic->name); diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c index 21db3cbea7d..e46f98f36e3 100644 --- a/arch/x86/kernel/apic/x2apic_uv_x.c +++ b/arch/x86/kernel/apic/x2apic_uv_x.c @@ -5,7 +5,7 @@ * * SGI UV APIC functions (note: not an Intel compatible APIC) * - * Copyright (C) 2007-2008 Silicon Graphics, Inc. All rights reserved. + * Copyright (C) 2007-2009 Silicon Graphics, Inc. All rights reserved. */ #include <linux/cpumask.h> #include <linux/hardirq.h> @@ -17,9 +17,12 @@ #include <linux/ctype.h> #include <linux/sched.h> #include <linux/timer.h> +#include <linux/slab.h> #include <linux/cpu.h> #include <linux/init.h> #include <linux/io.h> +#include <linux/pci.h> +#include <linux/kdebug.h> #include <asm/uv/uv_mmrs.h> #include <asm/uv/uv_hub.h> @@ -34,10 +37,13 @@ DEFINE_PER_CPU(int, x2apic_extra_bits); +#define PR_DEVEL(fmt, args...) pr_devel("%s: " fmt, __func__, args) + static enum uv_system_type uv_system_type; static u64 gru_start_paddr, gru_end_paddr; int uv_min_hub_revision_id; EXPORT_SYMBOL_GPL(uv_min_hub_revision_id); +static DEFINE_SPINLOCK(uv_nmi_lock); static inline bool is_GRU_range(u64 start, u64 end) { @@ -71,6 +77,7 @@ static int __init uv_acpi_madt_oem_check(char *oem_id, char *oem_table_id) if (!strcmp(oem_id, "SGI")) { nodeid = early_get_nodeid(); x86_platform.is_untracked_pat_range = uv_is_untracked_pat_range; + x86_platform.nmi_init = uv_nmi_init; if (!strcmp(oem_table_id, "UVL")) uv_system_type = UV_LEGACY_APIC; else if (!strcmp(oem_table_id, "UVX")) @@ -114,11 +121,9 @@ EXPORT_SYMBOL_GPL(uv_possible_blades); unsigned long sn_rtc_cycles_per_second; EXPORT_SYMBOL(sn_rtc_cycles_per_second); -/* Start with all IRQs pointing to boot CPU. IRQ balancing will shift them. */ - static const struct cpumask *uv_target_cpus(void) { - return cpumask_of(0); + return cpu_online_mask; } static void uv_vector_allocation_domain(int cpu, struct cpumask *retmask) @@ -482,7 +487,7 @@ static void uv_heartbeat(unsigned long ignored) static void __cpuinit uv_heartbeat_enable(int cpu) { - if (!uv_cpu_hub_info(cpu)->scir.enabled) { + while (!uv_cpu_hub_info(cpu)->scir.enabled) { struct timer_list *timer = &uv_cpu_hub_info(cpu)->scir.timer; uv_set_cpu_scir_bits(cpu, SCIR_CPU_HEARTBEAT|SCIR_CPU_ACTIVITY); @@ -490,11 +495,10 @@ static void __cpuinit uv_heartbeat_enable(int cpu) timer->expires = jiffies + SCIR_CPU_HB_INTERVAL; add_timer_on(timer, cpu); uv_cpu_hub_info(cpu)->scir.enabled = 1; - } - /* check boot cpu */ - if (!uv_cpu_hub_info(0)->scir.enabled) - uv_heartbeat_enable(0); + /* also ensure that boot cpu is enabled */ + cpu = 0; + } } #ifdef CONFIG_HOTPLUG_CPU @@ -553,6 +557,30 @@ late_initcall(uv_init_heartbeat); #endif /* !CONFIG_HOTPLUG_CPU */ +/* Direct Legacy VGA I/O traffic to designated IOH */ +int uv_set_vga_state(struct pci_dev *pdev, bool decode, + unsigned int command_bits, bool change_bridge) +{ + int domain, bus, rc; + + PR_DEVEL("devfn %x decode %d cmd %x chg_brdg %d\n", + pdev->devfn, decode, command_bits, change_bridge); + + if (!change_bridge) + return 0; + + if ((command_bits & PCI_COMMAND_IO) == 0) + return 0; + + domain = pci_domain_nr(pdev->bus); + bus = pdev->bus->number; + + rc = uv_bios_set_legacy_vga_target(decode, domain, bus); + PR_DEVEL("vga decode %d %x:%x, rc: %d\n", decode, domain, bus, rc); + + return rc; +} + /* * Called on each cpu to initialize the per_cpu UV data area. * FIXME: hotplug not supported yet @@ -569,6 +597,46 @@ void __cpuinit uv_cpu_init(void) set_x2apic_extra_bits(uv_hub_info->pnode); } +/* + * When NMI is received, print a stack trace. + */ +int uv_handle_nmi(struct notifier_block *self, unsigned long reason, void *data) +{ + if (reason != DIE_NMI_IPI) + return NOTIFY_OK; + /* + * Use a lock so only one cpu prints at a time + * to prevent intermixed output. + */ + spin_lock(&uv_nmi_lock); + pr_info("NMI stack dump cpu %u:\n", smp_processor_id()); + dump_stack(); + spin_unlock(&uv_nmi_lock); + + return NOTIFY_STOP; +} + +static struct notifier_block uv_dump_stack_nmi_nb = { + .notifier_call = uv_handle_nmi +}; + +void uv_register_nmi_notifier(void) +{ + if (register_die_notifier(&uv_dump_stack_nmi_nb)) + printk(KERN_WARNING "UV NMI handler failed to register\n"); +} + +void uv_nmi_init(void) +{ + unsigned int value; + + /* + * Unmask NMI on all cpus + */ + value = apic_read(APIC_LVT1) | APIC_DM_NMI; + value &= ~APIC_LVT_MASKED; + apic_write(APIC_LVT1, value); +} void __init uv_system_init(void) { @@ -634,8 +702,8 @@ void __init uv_system_init(void) } uv_bios_init(); - uv_bios_get_sn_info(0, &uv_type, &sn_partition_id, - &sn_coherency_id, &sn_region_size); + uv_bios_get_sn_info(0, &uv_type, &sn_partition_id, &sn_coherency_id, + &sn_region_size, &system_serial_number); uv_rtc_init(); for_each_present_cpu(cpu) { @@ -667,9 +735,6 @@ void __init uv_system_init(void) uv_node_to_blade[nid] = blade; uv_cpu_to_blade[cpu] = blade; max_pnode = max(pnode, max_pnode); - - printk(KERN_DEBUG "UV: cpu %d, apicid 0x%x, pnode %d, nid %d, lcpu %d, blade %d\n", - cpu, apicid, pnode, nid, lcpu, blade); } /* Add blade/pnode info for nodes without cpus */ @@ -690,5 +755,9 @@ void __init uv_system_init(void) uv_cpu_init(); uv_scir_register_cpu_notifier(); + uv_register_nmi_notifier(); proc_mkdir("sgi_uv", NULL); + + /* register Legacy VGA I/O redirection handler */ + pci_register_set_vga_state(uv_set_vga_state); } diff --git a/arch/x86/kernel/apm_32.c b/arch/x86/kernel/apm_32.c index b5b6b23bce5..c4f9182ca3a 100644 --- a/arch/x86/kernel/apm_32.c +++ b/arch/x86/kernel/apm_32.c @@ -1224,7 +1224,7 @@ static void reinit_timer(void) #ifdef INIT_TIMER_AFTER_SUSPEND unsigned long flags; - spin_lock_irqsave(&i8253_lock, flags); + raw_spin_lock_irqsave(&i8253_lock, flags); /* set the clock to HZ */ outb_pit(0x34, PIT_MODE); /* binary, mode 2, LSB/MSB, ch 0 */ udelay(10); @@ -1232,7 +1232,7 @@ static void reinit_timer(void) udelay(10); outb_pit(LATCH >> 8, PIT_CH0); /* MSB */ udelay(10); - spin_unlock_irqrestore(&i8253_lock, flags); + raw_spin_unlock_irqrestore(&i8253_lock, flags); #endif } @@ -1992,8 +1992,8 @@ static int __init apm_is_horked_d850md(const struct dmi_system_id *d) apm_info.disabled = 1; printk(KERN_INFO "%s machine detected. " "Disabling APM.\n", d->ident); - printk(KERN_INFO "This bug is fixed in bios P15 which is available for \n"); - printk(KERN_INFO "download from support.intel.com \n"); + printk(KERN_INFO "This bug is fixed in bios P15 which is available for\n"); + printk(KERN_INFO "download from support.intel.com\n"); } return 0; } diff --git a/arch/x86/kernel/bios_uv.c b/arch/x86/kernel/bios_uv.c index b0206a211b0..8bc57baaa9a 100644 --- a/arch/x86/kernel/bios_uv.c +++ b/arch/x86/kernel/bios_uv.c @@ -15,8 +15,8 @@ * along with this program; if not, write to the Free Software * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA * - * Copyright (c) 2008 Silicon Graphics, Inc. All Rights Reserved. - * Copyright (c) Russ Anderson + * Copyright (c) 2008-2009 Silicon Graphics, Inc. All Rights Reserved. + * Copyright (c) Russ Anderson <rja@sgi.com> */ #include <linux/efi.h> @@ -30,6 +30,7 @@ static struct uv_systab uv_systab; s64 uv_bios_call(enum uv_bios_cmd which, u64 a1, u64 a2, u64 a3, u64 a4, u64 a5) { struct uv_systab *tab = &uv_systab; + s64 ret; if (!tab->function) /* @@ -37,9 +38,11 @@ s64 uv_bios_call(enum uv_bios_cmd which, u64 a1, u64 a2, u64 a3, u64 a4, u64 a5) */ return BIOS_STATUS_UNIMPLEMENTED; - return efi_call6((void *)__va(tab->function), - (u64)which, a1, a2, a3, a4, a5); + ret = efi_call6((void *)__va(tab->function), (u64)which, + a1, a2, a3, a4, a5); + return ret; } +EXPORT_SYMBOL_GPL(uv_bios_call); s64 uv_bios_call_irqsave(enum uv_bios_cmd which, u64 a1, u64 a2, u64 a3, u64 a4, u64 a5) @@ -73,11 +76,14 @@ long sn_coherency_id; EXPORT_SYMBOL_GPL(sn_coherency_id); long sn_region_size; EXPORT_SYMBOL_GPL(sn_region_size); +long system_serial_number; +EXPORT_SYMBOL_GPL(system_serial_number); int uv_type; +EXPORT_SYMBOL_GPL(uv_type); s64 uv_bios_get_sn_info(int fc, int *uvtype, long *partid, long *coher, - long *region) + long *region, long *ssn) { s64 ret; u64 v0, v1; @@ -97,8 +103,11 @@ s64 uv_bios_get_sn_info(int fc, int *uvtype, long *partid, long *coher, *coher = part.coherence_id; if (region) *region = part.region_size; + if (ssn) + *ssn = v1; return ret; } +EXPORT_SYMBOL_GPL(uv_bios_get_sn_info); int uv_bios_mq_watchlist_alloc(unsigned long addr, unsigned int mq_size, @@ -154,6 +163,25 @@ s64 uv_bios_freq_base(u64 clock_type, u64 *ticks_per_second) } EXPORT_SYMBOL_GPL(uv_bios_freq_base); +/* + * uv_bios_set_legacy_vga_target - Set Legacy VGA I/O Target + * @decode: true to enable target, false to disable target + * @domain: PCI domain number + * @bus: PCI bus number + * + * Returns: + * 0: Success + * -EINVAL: Invalid domain or bus number + * -ENOSYS: Capability not available + * -EBUSY: Legacy VGA I/O cannot be retargeted at this time + */ +int uv_bios_set_legacy_vga_target(bool decode, int domain, int bus) +{ + return uv_bios_call(UV_BIOS_SET_LEGACY_VGA_TARGET, + (u64)decode, (u64)domain, (u64)bus, 0, 0); +} +EXPORT_SYMBOL_GPL(uv_bios_set_legacy_vga_target); + #ifdef CONFIG_EFI void uv_bios_init(void) @@ -185,4 +213,3 @@ void uv_bios_init(void) void uv_bios_init(void) { } #endif - diff --git a/arch/x86/kernel/bootflag.c b/arch/x86/kernel/bootflag.c index 30f25a75fe2..5de7f4c5697 100644 --- a/arch/x86/kernel/bootflag.c +++ b/arch/x86/kernel/bootflag.c @@ -5,7 +5,6 @@ #include <linux/kernel.h> #include <linux/init.h> #include <linux/string.h> -#include <linux/slab.h> #include <linux/spinlock.h> #include <linux/acpi.h> #include <asm/io.h> diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile index c202b62f367..3a785da34b6 100644 --- a/arch/x86/kernel/cpu/Makefile +++ b/arch/x86/kernel/cpu/Makefile @@ -14,7 +14,7 @@ CFLAGS_common.o := $(nostackp) obj-y := intel_cacheinfo.o addon_cpuid_features.o obj-y += proc.o capflags.o powerflags.o common.o -obj-y += vmware.o hypervisor.o sched.o +obj-y += vmware.o hypervisor.o sched.o mshyperv.o obj-$(CONFIG_X86_32) += bugs.o cmpxchg.o obj-$(CONFIG_X86_64) += bugs_64.o diff --git a/arch/x86/kernel/cpu/addon_cpuid_features.c b/arch/x86/kernel/cpu/addon_cpuid_features.c index 468489b57aa..10fa5684a66 100644 --- a/arch/x86/kernel/cpu/addon_cpuid_features.c +++ b/arch/x86/kernel/cpu/addon_cpuid_features.c @@ -30,8 +30,14 @@ void __cpuinit init_scattered_cpuid_features(struct cpuinfo_x86 *c) const struct cpuid_bit *cb; static const struct cpuid_bit __cpuinitconst cpuid_bits[] = { - { X86_FEATURE_IDA, CR_EAX, 1, 0x00000006 }, - { X86_FEATURE_ARAT, CR_EAX, 2, 0x00000006 }, + { X86_FEATURE_IDA, CR_EAX, 1, 0x00000006 }, + { X86_FEATURE_ARAT, CR_EAX, 2, 0x00000006 }, + { X86_FEATURE_APERFMPERF, CR_ECX, 0, 0x00000006 }, + { X86_FEATURE_CPB, CR_EDX, 9, 0x80000007 }, + { X86_FEATURE_NPT, CR_EDX, 0, 0x8000000a }, + { X86_FEATURE_LBRV, CR_EDX, 1, 0x8000000a }, + { X86_FEATURE_SVML, CR_EDX, 2, 0x8000000a }, + { X86_FEATURE_NRIPS, CR_EDX, 3, 0x8000000a }, { 0, 0, 0, 0 } }; diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c index 01a26521239..c39576cb301 100644 --- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -86,7 +86,7 @@ static void __init check_fpu(void) static void __init check_hlt(void) { - if (paravirt_enabled()) + if (boot_cpu_data.x86 >= 5 || paravirt_enabled()) return; printk(KERN_INFO "Checking 'hlt' instruction... "); diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index 4868e4a951e..68e4a6f2211 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -1084,6 +1084,20 @@ static void clear_all_debug_regs(void) } } +#ifdef CONFIG_KGDB +/* + * Restore debug regs if using kgdbwait and you have a kernel debugger + * connection established. + */ +static void dbg_restore_debug_regs(void) +{ + if (unlikely(kgdb_connected && arch_kgdb_ops.correct_hw_break)) + arch_kgdb_ops.correct_hw_break(); +} +#else /* ! CONFIG_KGDB */ +#define dbg_restore_debug_regs() +#endif /* ! CONFIG_KGDB */ + /* * cpu_init() initializes state that is per-CPU. Some data is already * initialized (naturally) in the bootstrap process, such as the GDT @@ -1107,9 +1121,9 @@ void __cpuinit cpu_init(void) oist = &per_cpu(orig_ist, cpu); #ifdef CONFIG_NUMA - if (cpu != 0 && percpu_read(node_number) == 0 && - cpu_to_node(cpu) != NUMA_NO_NODE) - percpu_write(node_number, cpu_to_node(cpu)); + if (cpu != 0 && percpu_read(numa_node) == 0 && + early_cpu_to_node(cpu) != NUMA_NO_NODE) + set_numa_node(early_cpu_to_node(cpu)); #endif me = current; @@ -1174,18 +1188,8 @@ void __cpuinit cpu_init(void) load_TR_desc(); load_LDT(&init_mm.context); -#ifdef CONFIG_KGDB - /* - * If the kgdb is connected no debug regs should be altered. This - * is only applicable when KGDB and a KGDB I/O module are built - * into the kernel and you are using early debugging with - * kgdbwait. KGDB will control the kernel HW breakpoint registers. - */ - if (kgdb_connected && arch_kgdb_ops.correct_hw_break) - arch_kgdb_ops.correct_hw_break(); - else -#endif - clear_all_debug_regs(); + clear_all_debug_regs(); + dbg_restore_debug_regs(); fpu_init(); @@ -1239,14 +1243,12 @@ void __cpuinit cpu_init(void) #endif clear_all_debug_regs(); + dbg_restore_debug_regs(); /* * Force FPU initialization: */ - if (cpu_has_xsave) - current_thread_info()->status = TS_XSAVE; - else - current_thread_info()->status = 0; + current_thread_info()->status = 0; clear_used_math(); mxcsr_feature_mask_init(); diff --git a/arch/x86/kernel/cpu/cpufreq/Kconfig b/arch/x86/kernel/cpu/cpufreq/Kconfig index f138c6c389b..870e6cc6ad2 100644 --- a/arch/x86/kernel/cpu/cpufreq/Kconfig +++ b/arch/x86/kernel/cpu/cpufreq/Kconfig @@ -10,6 +10,20 @@ if CPU_FREQ comment "CPUFreq processor drivers" +config X86_PCC_CPUFREQ + tristate "Processor Clocking Control interface driver" + depends on ACPI && ACPI_PROCESSOR + help + This driver adds support for the PCC interface. + + For details, take a look at: + <file:Documentation/cpu-freq/pcc-cpufreq.txt>. + + To compile this driver as a module, choose M here: the + module will be called pcc-cpufreq. + + If in doubt, say N. + config X86_ACPI_CPUFREQ tristate "ACPI Processor P-States driver" select CPU_FREQ_TABLE diff --git a/arch/x86/kernel/cpu/cpufreq/Makefile b/arch/x86/kernel/cpu/cpufreq/Makefile index 509296df294..bd54bf67e6f 100644 --- a/arch/x86/kernel/cpu/cpufreq/Makefile +++ b/arch/x86/kernel/cpu/cpufreq/Makefile @@ -2,8 +2,9 @@ # K8 systems. ACPI is preferred to all other hardware-specific drivers. # speedstep-* is preferred over p4-clockmod. -obj-$(CONFIG_X86_POWERNOW_K8) += powernow-k8.o -obj-$(CONFIG_X86_ACPI_CPUFREQ) += acpi-cpufreq.o +obj-$(CONFIG_X86_POWERNOW_K8) += powernow-k8.o mperf.o +obj-$(CONFIG_X86_ACPI_CPUFREQ) += acpi-cpufreq.o mperf.o +obj-$(CONFIG_X86_PCC_CPUFREQ) += pcc-cpufreq.o obj-$(CONFIG_X86_POWERNOW_K6) += powernow-k6.o obj-$(CONFIG_X86_POWERNOW_K7) += powernow-k7.o obj-$(CONFIG_X86_LONGHAUL) += longhaul.o diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c index 1b1920fa7c8..1d3cddaa40e 100644 --- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c +++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c @@ -33,6 +33,7 @@ #include <linux/cpufreq.h> #include <linux/compiler.h> #include <linux/dmi.h> +#include <linux/slab.h> #include <trace/events/power.h> #include <linux/acpi.h> @@ -45,6 +46,7 @@ #include <asm/msr.h> #include <asm/processor.h> #include <asm/cpufeature.h> +#include "mperf.h" #define dprintk(msg...) cpufreq_debug_printk(CPUFREQ_DEBUG_DRIVER, \ "acpi-cpufreq", msg) @@ -70,8 +72,6 @@ struct acpi_cpufreq_data { static DEFINE_PER_CPU(struct acpi_cpufreq_data *, acfreq_data); -static DEFINE_PER_CPU(struct aperfmperf, acfreq_old_perf); - /* acpi_perf_data is a pointer to percpu data. */ static struct acpi_processor_performance *acpi_perf_data; @@ -239,45 +239,6 @@ static u32 get_cur_val(const struct cpumask *mask) return cmd.val; } -/* Called via smp_call_function_single(), on the target CPU */ -static void read_measured_perf_ctrs(void *_cur) -{ - struct aperfmperf *am = _cur; - - get_aperfmperf(am); -} - -/* - * Return the measured active (C0) frequency on this CPU since last call - * to this function. - * Input: cpu number - * Return: Average CPU frequency in terms of max frequency (zero on error) - * - * We use IA32_MPERF and IA32_APERF MSRs to get the measured performance - * over a period of time, while CPU is in C0 state. - * IA32_MPERF counts at the rate of max advertised frequency - * IA32_APERF counts at the rate of actual CPU frequency - * Only IA32_APERF/IA32_MPERF ratio is architecturally defined and - * no meaning should be associated with absolute values of these MSRs. - */ -static unsigned int get_measured_perf(struct cpufreq_policy *policy, - unsigned int cpu) -{ - struct aperfmperf perf; - unsigned long ratio; - unsigned int retval; - - if (smp_call_function_single(cpu, read_measured_perf_ctrs, &perf, 1)) - return 0; - - ratio = calc_aperfmperf_ratio(&per_cpu(acfreq_old_perf, cpu), &perf); - per_cpu(acfreq_old_perf, cpu) = perf; - - retval = (policy->cpuinfo.max_freq * ratio) >> APERFMPERF_SHIFT; - - return retval; -} - static unsigned int get_cur_freq_on_cpu(unsigned int cpu) { struct acpi_cpufreq_data *data = per_cpu(acfreq_data, cpu); @@ -701,7 +662,7 @@ static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy) /* Check for APERF/MPERF support in hardware */ if (cpu_has(c, X86_FEATURE_APERFMPERF)) - acpi_cpufreq_driver.getavg = get_measured_perf; + acpi_cpufreq_driver.getavg = cpufreq_get_measured_perf; dprintk("CPU%u - ACPI performance management activated.\n", cpu); for (i = 0; i < perf->state_count; i++) diff --git a/arch/x86/kernel/cpu/cpufreq/elanfreq.c b/arch/x86/kernel/cpu/cpufreq/elanfreq.c index 006b278b0d5..c587db472a7 100644 --- a/arch/x86/kernel/cpu/cpufreq/elanfreq.c +++ b/arch/x86/kernel/cpu/cpufreq/elanfreq.c @@ -20,7 +20,6 @@ #include <linux/module.h> #include <linux/init.h> -#include <linux/slab.h> #include <linux/delay.h> #include <linux/cpufreq.h> diff --git a/arch/x86/kernel/cpu/cpufreq/gx-suspmod.c b/arch/x86/kernel/cpu/cpufreq/gx-suspmod.c index ac27ec2264d..16e3483be9e 100644 --- a/arch/x86/kernel/cpu/cpufreq/gx-suspmod.c +++ b/arch/x86/kernel/cpu/cpufreq/gx-suspmod.c @@ -80,6 +80,7 @@ #include <linux/cpufreq.h> #include <linux/pci.h> #include <linux/errno.h> +#include <linux/slab.h> #include <asm/processor-cyrix.h> diff --git a/arch/x86/kernel/cpu/cpufreq/longrun.c b/arch/x86/kernel/cpu/cpufreq/longrun.c index da5f70fcb76..e7b559d74c5 100644 --- a/arch/x86/kernel/cpu/cpufreq/longrun.c +++ b/arch/x86/kernel/cpu/cpufreq/longrun.c @@ -9,7 +9,6 @@ #include <linux/kernel.h> #include <linux/module.h> #include <linux/init.h> -#include <linux/slab.h> #include <linux/cpufreq.h> #include <linux/timex.h> diff --git a/arch/x86/kernel/cpu/cpufreq/mperf.c b/arch/x86/kernel/cpu/cpufreq/mperf.c new file mode 100644 index 00000000000..911e193018a --- /dev/null +++ b/arch/x86/kernel/cpu/cpufreq/mperf.c @@ -0,0 +1,51 @@ +#include <linux/kernel.h> +#include <linux/smp.h> +#include <linux/module.h> +#include <linux/init.h> +#include <linux/cpufreq.h> +#include <linux/slab.h> + +#include "mperf.h" + +static DEFINE_PER_CPU(struct aperfmperf, acfreq_old_perf); + +/* Called via smp_call_function_single(), on the target CPU */ +static void read_measured_perf_ctrs(void *_cur) +{ + struct aperfmperf *am = _cur; + + get_aperfmperf(am); +} + +/* + * Return the measured active (C0) frequency on this CPU since last call + * to this function. + * Input: cpu number + * Return: Average CPU frequency in terms of max frequency (zero on error) + * + * We use IA32_MPERF and IA32_APERF MSRs to get the measured performance + * over a period of time, while CPU is in C0 state. + * IA32_MPERF counts at the rate of max advertised frequency + * IA32_APERF counts at the rate of actual CPU frequency + * Only IA32_APERF/IA32_MPERF ratio is architecturally defined and + * no meaning should be associated with absolute values of these MSRs. + */ +unsigned int cpufreq_get_measured_perf(struct cpufreq_policy *policy, + unsigned int cpu) +{ + struct aperfmperf perf; + unsigned long ratio; + unsigned int retval; + + if (smp_call_function_single(cpu, read_measured_perf_ctrs, &perf, 1)) + return 0; + + ratio = calc_aperfmperf_ratio(&per_cpu(acfreq_old_perf, cpu), &perf); + per_cpu(acfreq_old_perf, cpu) = perf; + + retval = (policy->cpuinfo.max_freq * ratio) >> APERFMPERF_SHIFT; + + return retval; +} +EXPORT_SYMBOL_GPL(cpufreq_get_measured_perf); +MODULE_LICENSE("GPL"); diff --git a/arch/x86/kernel/cpu/cpufreq/mperf.h b/arch/x86/kernel/cpu/cpufreq/mperf.h new file mode 100644 index 00000000000..5dbf2950dc2 --- /dev/null +++ b/arch/x86/kernel/cpu/cpufreq/mperf.h @@ -0,0 +1,9 @@ +/* + * (c) 2010 Advanced Micro Devices, Inc. + * Your use of this code is subject to the terms and conditions of the + * GNU general public license version 2. See "COPYING" or + * http://www.gnu.org/licenses/gpl.html + */ + +unsigned int cpufreq_get_measured_perf(struct cpufreq_policy *policy, + unsigned int cpu); diff --git a/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c b/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c index 86961519372..7b8a8ba67b0 100644 --- a/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c +++ b/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c @@ -25,7 +25,6 @@ #include <linux/init.h> #include <linux/smp.h> #include <linux/cpufreq.h> -#include <linux/slab.h> #include <linux/cpumask.h> #include <linux/timex.h> diff --git a/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c new file mode 100644 index 00000000000..ce7cde713e7 --- /dev/null +++ b/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c @@ -0,0 +1,621 @@ +/* + * pcc-cpufreq.c - Processor Clocking Control firmware cpufreq interface + * + * Copyright (C) 2009 Red Hat, Matthew Garrett <mjg@redhat.com> + * Copyright (C) 2009 Hewlett-Packard Development Company, L.P. + * Nagananda Chumbalkar <nagananda.chumbalkar@hp.com> + * + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; version 2 of the License. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or NON + * INFRINGEMENT. See the GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License along + * with this program; if not, write to the Free Software Foundation, Inc., + * 675 Mass Ave, Cambridge, MA 02139, USA. + * + * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + */ + +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/init.h> +#include <linux/smp.h> +#include <linux/sched.h> +#include <linux/cpufreq.h> +#include <linux/compiler.h> +#include <linux/slab.h> + +#include <linux/acpi.h> +#include <linux/io.h> +#include <linux/spinlock.h> +#include <linux/uaccess.h> + +#include <acpi/processor.h> + +#define PCC_VERSION "1.00.00" +#define POLL_LOOPS 300 + +#define CMD_COMPLETE 0x1 +#define CMD_GET_FREQ 0x0 +#define CMD_SET_FREQ 0x1 + +#define BUF_SZ 4 + +#define dprintk(msg...) cpufreq_debug_printk(CPUFREQ_DEBUG_DRIVER, \ + "pcc-cpufreq", msg) + +struct pcc_register_resource { + u8 descriptor; + u16 length; + u8 space_id; + u8 bit_width; + u8 bit_offset; + u8 access_size; + u64 address; +} __attribute__ ((packed)); + +struct pcc_memory_resource { + u8 descriptor; + u16 length; + u8 space_id; + u8 resource_usage; + u8 type_specific; + u64 granularity; + u64 minimum; + u64 maximum; + u64 translation_offset; + u64 address_length; +} __attribute__ ((packed)); + +static struct cpufreq_driver pcc_cpufreq_driver; + +struct pcc_header { + u32 signature; + u16 length; + u8 major; + u8 minor; + u32 features; + u16 command; + u16 status; + u32 latency; + u32 minimum_time; + u32 maximum_time; + u32 nominal; + u32 throttled_frequency; + u32 minimum_frequency; +}; + +static void __iomem *pcch_virt_addr; +static struct pcc_header __iomem *pcch_hdr; + +static DEFINE_SPINLOCK(pcc_lock); + +static struct acpi_generic_address doorbell; + +static u64 doorbell_preserve; +static u64 doorbell_write; + +static u8 OSC_UUID[16] = {0x63, 0x9B, 0x2C, 0x9F, 0x70, 0x91, 0x49, 0x1f, + 0xBB, 0x4F, 0xA5, 0x98, 0x2F, 0xA1, 0xB5, 0x46}; + +struct pcc_cpu { + u32 input_offset; + u32 output_offset; +}; + +static struct pcc_cpu *pcc_cpu_info; + +static int pcc_cpufreq_verify(struct cpufreq_policy *policy) +{ + cpufreq_verify_within_limits(policy, policy->cpuinfo.min_freq, + policy->cpuinfo.max_freq); + return 0; +} + +static inline void pcc_cmd(void) +{ + u64 doorbell_value; + int i; + + acpi_read(&doorbell_value, &doorbell); + acpi_write((doorbell_value & doorbell_preserve) | doorbell_write, + &doorbell); + + for (i = 0; i < POLL_LOOPS; i++) { + if (ioread16(&pcch_hdr->status) & CMD_COMPLETE) + break; + } +} + +static inline void pcc_clear_mapping(void) +{ + if (pcch_virt_addr) + iounmap(pcch_virt_addr); + pcch_virt_addr = NULL; +} + +static unsigned int pcc_get_freq(unsigned int cpu) +{ + struct pcc_cpu *pcc_cpu_data; + unsigned int curr_freq; + unsigned int freq_limit; + u16 status; + u32 input_buffer; + u32 output_buffer; + + spin_lock(&pcc_lock); + + dprintk("get: get_freq for CPU %d\n", cpu); + pcc_cpu_data = per_cpu_ptr(pcc_cpu_info, cpu); + + input_buffer = 0x1; + iowrite32(input_buffer, + (pcch_virt_addr + pcc_cpu_data->input_offset)); + iowrite16(CMD_GET_FREQ, &pcch_hdr->command); + + pcc_cmd(); + + output_buffer = + ioread32(pcch_virt_addr + pcc_cpu_data->output_offset); + + /* Clear the input buffer - we are done with the current command */ + memset_io((pcch_virt_addr + pcc_cpu_data->input_offset), 0, BUF_SZ); + + status = ioread16(&pcch_hdr->status); + if (status != CMD_COMPLETE) { + dprintk("get: FAILED: for CPU %d, status is %d\n", + cpu, status); + goto cmd_incomplete; + } + iowrite16(0, &pcch_hdr->status); + curr_freq = (((ioread32(&pcch_hdr->nominal) * (output_buffer & 0xff)) + / 100) * 1000); + + dprintk("get: SUCCESS: (virtual) output_offset for cpu %d is " + "0x%x, contains a value of: 0x%x. Speed is: %d MHz\n", + cpu, (pcch_virt_addr + pcc_cpu_data->output_offset), + output_buffer, curr_freq); + + freq_limit = (output_buffer >> 8) & 0xff; + if (freq_limit != 0xff) { + dprintk("get: frequency for cpu %d is being temporarily" + " capped at %d\n", cpu, curr_freq); + } + + spin_unlock(&pcc_lock); + return curr_freq; + +cmd_incomplete: + iowrite16(0, &pcch_hdr->status); + spin_unlock(&pcc_lock); + return -EINVAL; +} + +static int pcc_cpufreq_target(struct cpufreq_policy *policy, + unsigned int target_freq, + unsigned int relation) +{ + struct pcc_cpu *pcc_cpu_data; + struct cpufreq_freqs freqs; + u16 status; + u32 input_buffer; + int cpu; + + spin_lock(&pcc_lock); + cpu = policy->cpu; + pcc_cpu_data = per_cpu_ptr(pcc_cpu_info, cpu); + + dprintk("target: CPU %d should go to target freq: %d " + "(virtual) input_offset is 0x%x\n", + cpu, target_freq, + (pcch_virt_addr + pcc_cpu_data->input_offset)); + + freqs.new = target_freq; + freqs.cpu = cpu; + cpufreq_notify_transition(&freqs, CPUFREQ_PRECHANGE); + + input_buffer = 0x1 | (((target_freq * 100) + / (ioread32(&pcch_hdr->nominal) * 1000)) << 8); + iowrite32(input_buffer, + (pcch_virt_addr + pcc_cpu_data->input_offset)); + iowrite16(CMD_SET_FREQ, &pcch_hdr->command); + + pcc_cmd(); + + /* Clear the input buffer - we are done with the current command */ + memset_io((pcch_virt_addr + pcc_cpu_data->input_offset), 0, BUF_SZ); + + status = ioread16(&pcch_hdr->status); + if (status != CMD_COMPLETE) { + dprintk("target: FAILED for cpu %d, with status: 0x%x\n", + cpu, status); + goto cmd_incomplete; + } + iowrite16(0, &pcch_hdr->status); + + cpufreq_notify_transition(&freqs, CPUFREQ_POSTCHANGE); + dprintk("target: was SUCCESSFUL for cpu %d\n", cpu); + spin_unlock(&pcc_lock); + + return 0; + +cmd_incomplete: + iowrite16(0, &pcch_hdr->status); + spin_unlock(&pcc_lock); + return -EINVAL; +} + +static int pcc_get_offset(int cpu) +{ + acpi_status status; + struct acpi_buffer buffer = {ACPI_ALLOCATE_BUFFER, NULL}; + union acpi_object *pccp, *offset; + struct pcc_cpu *pcc_cpu_data; + struct acpi_processor *pr; + int ret = 0; + + pr = per_cpu(processors, cpu); + pcc_cpu_data = per_cpu_ptr(pcc_cpu_info, cpu); + + status = acpi_evaluate_object(pr->handle, "PCCP", NULL, &buffer); + if (ACPI_FAILURE(status)) + return -ENODEV; + + pccp = buffer.pointer; + if (!pccp || pccp->type != ACPI_TYPE_PACKAGE) { + ret = -ENODEV; + goto out_free; + }; + + offset = &(pccp->package.elements[0]); + if (!offset || offset->type != ACPI_TYPE_INTEGER) { + ret = -ENODEV; + goto out_free; + } + + pcc_cpu_data->input_offset = offset->integer.value; + + offset = &(pccp->package.elements[1]); + if (!offset || offset->type != ACPI_TYPE_INTEGER) { + ret = -ENODEV; + goto out_free; + } + + pcc_cpu_data->output_offset = offset->integer.value; + + memset_io((pcch_virt_addr + pcc_cpu_data->input_offset), 0, BUF_SZ); + memset_io((pcch_virt_addr + pcc_cpu_data->output_offset), 0, BUF_SZ); + + dprintk("pcc_get_offset: for CPU %d: pcc_cpu_data " + "input_offset: 0x%x, pcc_cpu_data output_offset: 0x%x\n", + cpu, pcc_cpu_data->input_offset, pcc_cpu_data->output_offset); +out_free: + kfree(buffer.pointer); + return ret; +} + +static int __init pcc_cpufreq_do_osc(acpi_handle *handle) +{ + acpi_status status; + struct acpi_object_list input; + struct acpi_buffer output = {ACPI_ALLOCATE_BUFFER, NULL}; + union acpi_object in_params[4]; + union acpi_object *out_obj; + u32 capabilities[2]; + u32 errors; + u32 supported; + int ret = 0; + + input.count = 4; + input.pointer = in_params; + input.count = 4; + input.pointer = in_params; + in_params[0].type = ACPI_TYPE_BUFFER; + in_params[0].buffer.length = 16; + in_params[0].buffer.pointer = OSC_UUID; + in_params[1].type = ACPI_TYPE_INTEGER; + in_params[1].integer.value = 1; + in_params[2].type = ACPI_TYPE_INTEGER; + in_params[2].integer.value = 2; + in_params[3].type = ACPI_TYPE_BUFFER; + in_params[3].buffer.length = 8; + in_params[3].buffer.pointer = (u8 *)&capabilities; + + capabilities[0] = OSC_QUERY_ENABLE; + capabilities[1] = 0x1; + + status = acpi_evaluate_object(*handle, "_OSC", &input, &output); + if (ACPI_FAILURE(status)) + return -ENODEV; + + if (!output.length) + return -ENODEV; + + out_obj = output.pointer; + if (out_obj->type != ACPI_TYPE_BUFFER) { + ret = -ENODEV; + goto out_free; + } + + errors = *((u32 *)out_obj->buffer.pointer) & ~(1 << 0); + if (errors) { + ret = -ENODEV; + goto out_free; + } + + supported = *((u32 *)(out_obj->buffer.pointer + 4)); + if (!(supported & 0x1)) { + ret = -ENODEV; + goto out_free; + } + + kfree(output.pointer); + capabilities[0] = 0x0; + capabilities[1] = 0x1; + + status = acpi_evaluate_object(*handle, "_OSC", &input, &output); + if (ACPI_FAILURE(status)) + return -ENODEV; + + if (!output.length) + return -ENODEV; + + out_obj = output.pointer; + if (out_obj->type != ACPI_TYPE_BUFFER) { + ret = -ENODEV; + goto out_free; + } + + errors = *((u32 *)out_obj->buffer.pointer) & ~(1 << 0); + if (errors) { + ret = -ENODEV; + goto out_free; + } + + supported = *((u32 *)(out_obj->buffer.pointer + 4)); + if (!(supported & 0x1)) { + ret = -ENODEV; + goto out_free; + } + +out_free: + kfree(output.pointer); + return ret; +} + +static int __init pcc_cpufreq_probe(void) +{ + acpi_status status; + struct acpi_buffer output = {ACPI_ALLOCATE_BUFFER, NULL}; + struct pcc_memory_resource *mem_resource; + struct pcc_register_resource *reg_resource; + union acpi_object *out_obj, *member; + acpi_handle handle, osc_handle; + int ret = 0; + + status = acpi_get_handle(NULL, "\\_SB", &handle); + if (ACPI_FAILURE(status)) + return -ENODEV; + + status = acpi_get_handle(handle, "_OSC", &osc_handle); + if (ACPI_SUCCESS(status)) { + ret = pcc_cpufreq_do_osc(&osc_handle); + if (ret) + dprintk("probe: _OSC evaluation did not succeed\n"); + /* Firmware's use of _OSC is optional */ + ret = 0; + } + + status = acpi_evaluate_object(handle, "PCCH", NULL, &output); + if (ACPI_FAILURE(status)) + return -ENODEV; + + out_obj = output.pointer; + if (out_obj->type != ACPI_TYPE_PACKAGE) { + ret = -ENODEV; + goto out_free; + } + + member = &out_obj->package.elements[0]; + if (member->type != ACPI_TYPE_BUFFER) { + ret = -ENODEV; + goto out_free; + } + + mem_resource = (struct pcc_memory_resource *)member->buffer.pointer; + + dprintk("probe: mem_resource descriptor: 0x%x," + " length: %d, space_id: %d, resource_usage: %d," + " type_specific: %d, granularity: 0x%llx," + " minimum: 0x%llx, maximum: 0x%llx," + " translation_offset: 0x%llx, address_length: 0x%llx\n", + mem_resource->descriptor, mem_resource->length, + mem_resource->space_id, mem_resource->resource_usage, + mem_resource->type_specific, mem_resource->granularity, + mem_resource->minimum, mem_resource->maximum, + mem_resource->translation_offset, + mem_resource->address_length); + + if (mem_resource->space_id != ACPI_ADR_SPACE_SYSTEM_MEMORY) { + ret = -ENODEV; + goto out_free; + } + + pcch_virt_addr = ioremap_nocache(mem_resource->minimum, + mem_resource->address_length); + if (pcch_virt_addr == NULL) { + dprintk("probe: could not map shared mem region\n"); + goto out_free; + } + pcch_hdr = pcch_virt_addr; + + dprintk("probe: PCCH header (virtual) addr: 0x%p\n", pcch_hdr); + dprintk("probe: PCCH header is at physical address: 0x%llx," + " signature: 0x%x, length: %d bytes, major: %d, minor: %d," + " supported features: 0x%x, command field: 0x%x," + " status field: 0x%x, nominal latency: %d us\n", + mem_resource->minimum, ioread32(&pcch_hdr->signature), + ioread16(&pcch_hdr->length), ioread8(&pcch_hdr->major), + ioread8(&pcch_hdr->minor), ioread32(&pcch_hdr->features), + ioread16(&pcch_hdr->command), ioread16(&pcch_hdr->status), + ioread32(&pcch_hdr->latency)); + + dprintk("probe: min time between commands: %d us," + " max time between commands: %d us," + " nominal CPU frequency: %d MHz," + " minimum CPU frequency: %d MHz," + " minimum CPU frequency without throttling: %d MHz\n", + ioread32(&pcch_hdr->minimum_time), + ioread32(&pcch_hdr->maximum_time), + ioread32(&pcch_hdr->nominal), + ioread32(&pcch_hdr->throttled_frequency), + ioread32(&pcch_hdr->minimum_frequency)); + + member = &out_obj->package.elements[1]; + if (member->type != ACPI_TYPE_BUFFER) { + ret = -ENODEV; + goto pcch_free; + } + + reg_resource = (struct pcc_register_resource *)member->buffer.pointer; + + doorbell.space_id = reg_resource->space_id; + doorbell.bit_width = reg_resource->bit_width; + doorbell.bit_offset = reg_resource->bit_offset; + doorbell.access_width = 64; + doorbell.address = reg_resource->address; + + dprintk("probe: doorbell: space_id is %d, bit_width is %d, " + "bit_offset is %d, access_width is %d, address is 0x%llx\n", + doorbell.space_id, doorbell.bit_width, doorbell.bit_offset, + doorbell.access_width, reg_resource->address); + + member = &out_obj->package.elements[2]; + if (member->type != ACPI_TYPE_INTEGER) { + ret = -ENODEV; + goto pcch_free; + } + + doorbell_preserve = member->integer.value; + + member = &out_obj->package.elements[3]; + if (member->type != ACPI_TYPE_INTEGER) { + ret = -ENODEV; + goto pcch_free; + } + + doorbell_write = member->integer.value; + + dprintk("probe: doorbell_preserve: 0x%llx," + " doorbell_write: 0x%llx\n", + doorbell_preserve, doorbell_write); + + pcc_cpu_info = alloc_percpu(struct pcc_cpu); + if (!pcc_cpu_info) { + ret = -ENOMEM; + goto pcch_free; + } + + printk(KERN_DEBUG "pcc-cpufreq: (v%s) driver loaded with frequency" + " limits: %d MHz, %d MHz\n", PCC_VERSION, + ioread32(&pcch_hdr->minimum_frequency), + ioread32(&pcch_hdr->nominal)); + kfree(output.pointer); + return ret; +pcch_free: + pcc_clear_mapping(); +out_free: + kfree(output.pointer); + return ret; +} + +static int pcc_cpufreq_cpu_init(struct cpufreq_policy *policy) +{ + unsigned int cpu = policy->cpu; + unsigned int result = 0; + + if (!pcch_virt_addr) { + result = -1; + goto pcch_null; + } + + result = pcc_get_offset(cpu); + if (result) { + dprintk("init: PCCP evaluation failed\n"); + goto free; + } + + policy->max = policy->cpuinfo.max_freq = + ioread32(&pcch_hdr->nominal) * 1000; + policy->min = policy->cpuinfo.min_freq = + ioread32(&pcch_hdr->minimum_frequency) * 1000; + policy->cur = pcc_get_freq(cpu); + + dprintk("init: policy->max is %d, policy->min is %d\n", + policy->max, policy->min); + + return 0; +free: + pcc_clear_mapping(); + free_percpu(pcc_cpu_info); +pcch_null: + return result; +} + +static int pcc_cpufreq_cpu_exit(struct cpufreq_policy *policy) +{ + return 0; +} + +static struct cpufreq_driver pcc_cpufreq_driver = { + .flags = CPUFREQ_CONST_LOOPS, + .get = pcc_get_freq, + .verify = pcc_cpufreq_verify, + .target = pcc_cpufreq_target, + .init = pcc_cpufreq_cpu_init, + .exit = pcc_cpufreq_cpu_exit, + .name = "pcc-cpufreq", + .owner = THIS_MODULE, +}; + +static int __init pcc_cpufreq_init(void) +{ + int ret; + + if (acpi_disabled) + return 0; + + ret = pcc_cpufreq_probe(); + if (ret) { + dprintk("pcc_cpufreq_init: PCCH evaluation failed\n"); + return ret; + } + + ret = cpufreq_register_driver(&pcc_cpufreq_driver); + + return ret; +} + +static void __exit pcc_cpufreq_exit(void) +{ + cpufreq_unregister_driver(&pcc_cpufreq_driver); + + pcc_clear_mapping(); + + free_percpu(pcc_cpu_info); +} + +MODULE_AUTHOR("Matthew Garrett, Naga Chumbalkar"); +MODULE_VERSION(PCC_VERSION); +MODULE_DESCRIPTION("Processor Clocking Control interface driver"); +MODULE_LICENSE("GPL"); + +late_initcall(pcc_cpufreq_init); +module_exit(pcc_cpufreq_exit); diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k6.c b/arch/x86/kernel/cpu/cpufreq/powernow-k6.c index cb01dac267d..b3379d6a5c5 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k6.c +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k6.c @@ -13,7 +13,6 @@ #include <linux/init.h> #include <linux/cpufreq.h> #include <linux/ioport.h> -#include <linux/slab.h> #include <linux/timex.h> #include <linux/io.h> diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c index f125e5c551c..7ec2123838e 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c @@ -1,6 +1,5 @@ - /* - * (c) 2003-2006 Advanced Micro Devices, Inc. + * (c) 2003-2010 Advanced Micro Devices, Inc. * Your use of this code is subject to the terms and conditions of the * GNU general public license version 2. See "COPYING" or * http://www.gnu.org/licenses/gpl.html @@ -46,6 +45,7 @@ #define PFX "powernow-k8: " #define VERSION "version 2.20.00" #include "powernow-k8.h" +#include "mperf.h" /* serialize freq changes */ static DEFINE_MUTEX(fidvid_mutex); @@ -54,6 +54,12 @@ static DEFINE_PER_CPU(struct powernow_k8_data *, powernow_data); static int cpu_family = CPU_OPTERON; +/* core performance boost */ +static bool cpb_capable, cpb_enabled; +static struct msr __percpu *msrs; + +static struct cpufreq_driver cpufreq_amd64_driver; + #ifndef CONFIG_SMP static inline const struct cpumask *cpu_core_mask(int cpu) { @@ -806,7 +812,7 @@ static int find_psb_table(struct powernow_k8_data *data) static void powernow_k8_acpi_pst_values(struct powernow_k8_data *data, unsigned int index) { - acpi_integer control; + u64 control; if (!data->acpi_data.state_count || (cpu_family == CPU_HW_PSTATE)) return; @@ -824,7 +830,7 @@ static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data) { struct cpufreq_frequency_table *powernow_table; int ret_val = -ENODEV; - acpi_integer control, status; + u64 control, status; if (acpi_processor_register_performance(&data->acpi_data, data->cpu)) { dprintk("register performance failed: bad ACPI data\n"); @@ -929,7 +935,8 @@ static int fill_powernow_table_pstate(struct powernow_k8_data *data, powernow_table[i].index = index; /* Frequency may be rounded for these */ - if (boot_cpu_data.x86 == 0x10 || boot_cpu_data.x86 == 0x11) { + if ((boot_cpu_data.x86 == 0x10 && boot_cpu_data.x86_model < 10) + || boot_cpu_data.x86 == 0x11) { powernow_table[i].frequency = freq_from_fid_did(lo & 0x3f, (lo >> 6) & 7); } else @@ -948,7 +955,7 @@ static int fill_powernow_table_fidvid(struct powernow_k8_data *data, u32 fid; u32 vid; u32 freq, index; - acpi_integer status, control; + u64 status, control; if (data->exttype) { status = data->acpi_data.states[i].status; @@ -1248,6 +1255,7 @@ static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol) struct powernow_k8_data *data; struct init_on_cpu init_on_cpu; int rc; + struct cpuinfo_x86 *c = &cpu_data(pol->cpu); if (!cpu_online(pol->cpu)) return -ENODEV; @@ -1322,6 +1330,10 @@ static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol) return -EINVAL; } + /* Check for APERF/MPERF support in hardware */ + if (cpu_has(c, X86_FEATURE_APERFMPERF)) + cpufreq_amd64_driver.getavg = cpufreq_get_measured_perf; + cpufreq_frequency_table_get_attr(data->powernow_table, pol->cpu); if (cpu_family == CPU_HW_PSTATE) @@ -1356,6 +1368,7 @@ static int __devexit powernowk8_cpu_exit(struct cpufreq_policy *pol) kfree(data->powernow_table); kfree(data); + per_cpu(powernow_data, pol->cpu) = NULL; return 0; } @@ -1375,7 +1388,7 @@ static unsigned int powernowk8_get(unsigned int cpu) int err; if (!data) - return -EINVAL; + return 0; smp_call_function_single(cpu, query_values_on_cpu, &err, true); if (err) @@ -1392,8 +1405,77 @@ out: return khz; } +static void _cpb_toggle_msrs(bool t) +{ + int cpu; + + get_online_cpus(); + + rdmsr_on_cpus(cpu_online_mask, MSR_K7_HWCR, msrs); + + for_each_cpu(cpu, cpu_online_mask) { + struct msr *reg = per_cpu_ptr(msrs, cpu); + if (t) + reg->l &= ~BIT(25); + else + reg->l |= BIT(25); + } + wrmsr_on_cpus(cpu_online_mask, MSR_K7_HWCR, msrs); + + put_online_cpus(); +} + +/* + * Switch on/off core performance boosting. + * + * 0=disable + * 1=enable. + */ +static void cpb_toggle(bool t) +{ + if (!cpb_capable) + return; + + if (t && !cpb_enabled) { + cpb_enabled = true; + _cpb_toggle_msrs(t); + printk(KERN_INFO PFX "Core Boosting enabled.\n"); + } else if (!t && cpb_enabled) { + cpb_enabled = false; + _cpb_toggle_msrs(t); + printk(KERN_INFO PFX "Core Boosting disabled.\n"); + } +} + +static ssize_t store_cpb(struct cpufreq_policy *policy, const char *buf, + size_t count) +{ + int ret = -EINVAL; + unsigned long val = 0; + + ret = strict_strtoul(buf, 10, &val); + if (!ret && (val == 0 || val == 1) && cpb_capable) + cpb_toggle(val); + else + return -EINVAL; + + return count; +} + +static ssize_t show_cpb(struct cpufreq_policy *policy, char *buf) +{ + return sprintf(buf, "%u\n", cpb_enabled); +} + +#define define_one_rw(_name) \ +static struct freq_attr _name = \ +__ATTR(_name, 0644, show_##_name, store_##_name) + +define_one_rw(cpb); + static struct freq_attr *powernow_k8_attr[] = { &cpufreq_freq_attr_scaling_available_freqs, + &cpb, NULL, }; @@ -1409,10 +1491,51 @@ static struct cpufreq_driver cpufreq_amd64_driver = { .attr = powernow_k8_attr, }; +/* + * Clear the boost-disable flag on the CPU_DOWN path so that this cpu + * cannot block the remaining ones from boosting. On the CPU_UP path we + * simply keep the boost-disable flag in sync with the current global + * state. + */ +static int cpb_notify(struct notifier_block *nb, unsigned long action, + void *hcpu) +{ + unsigned cpu = (long)hcpu; + u32 lo, hi; + + switch (action) { + case CPU_UP_PREPARE: + case CPU_UP_PREPARE_FROZEN: + + if (!cpb_enabled) { + rdmsr_on_cpu(cpu, MSR_K7_HWCR, &lo, &hi); + lo |= BIT(25); + wrmsr_on_cpu(cpu, MSR_K7_HWCR, lo, hi); + } + break; + + case CPU_DOWN_PREPARE: + case CPU_DOWN_PREPARE_FROZEN: + rdmsr_on_cpu(cpu, MSR_K7_HWCR, &lo, &hi); + lo &= ~BIT(25); + wrmsr_on_cpu(cpu, MSR_K7_HWCR, lo, hi); + break; + + default: + break; + } + + return NOTIFY_OK; +} + +static struct notifier_block cpb_nb = { + .notifier_call = cpb_notify, +}; + /* driver entry point for init */ static int __cpuinit powernowk8_init(void) { - unsigned int i, supported_cpus = 0; + unsigned int i, supported_cpus = 0, cpu; for_each_online_cpu(i) { int rc; @@ -1421,15 +1544,36 @@ static int __cpuinit powernowk8_init(void) supported_cpus++; } - if (supported_cpus == num_online_cpus()) { - printk(KERN_INFO PFX "Found %d %s " - "processors (%d cpu cores) (" VERSION ")\n", - num_online_nodes(), - boot_cpu_data.x86_model_id, supported_cpus); - return cpufreq_register_driver(&cpufreq_amd64_driver); + if (supported_cpus != num_online_cpus()) + return -ENODEV; + + printk(KERN_INFO PFX "Found %d %s (%d cpu cores) (" VERSION ")\n", + num_online_nodes(), boot_cpu_data.x86_model_id, supported_cpus); + + if (boot_cpu_has(X86_FEATURE_CPB)) { + + cpb_capable = true; + + register_cpu_notifier(&cpb_nb); + + msrs = msrs_alloc(); + if (!msrs) { + printk(KERN_ERR "%s: Error allocating msrs!\n", __func__); + return -ENOMEM; + } + + rdmsr_on_cpus(cpu_online_mask, MSR_K7_HWCR, msrs); + + for_each_cpu(cpu, cpu_online_mask) { + struct msr *reg = per_cpu_ptr(msrs, cpu); + cpb_enabled |= !(!!(reg->l & BIT(25))); + } + + printk(KERN_INFO PFX "Core Performance Boosting: %s.\n", + (cpb_enabled ? "on" : "off")); } - return -ENODEV; + return cpufreq_register_driver(&cpufreq_amd64_driver); } /* driver entry point for term */ @@ -1437,6 +1581,13 @@ static void __exit powernowk8_exit(void) { dprintk("exit\n"); + if (boot_cpu_has(X86_FEATURE_CPB)) { + msrs_free(msrs); + msrs = NULL; + + unregister_cpu_notifier(&cpb_nb); + } + cpufreq_unregister_driver(&cpufreq_amd64_driver); } diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h index 02ce824073c..df3529b1c02 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h @@ -5,7 +5,6 @@ * http://www.gnu.org/licenses/gpl.html */ - enum pstate { HW_PSTATE_INVALID = 0xff, HW_PSTATE_0 = 0, @@ -55,7 +54,6 @@ struct powernow_k8_data { struct cpumask *available_cores; }; - /* processor's cpuid instruction support */ #define CPUID_PROCESSOR_SIGNATURE 1 /* function 1 */ #define CPUID_XFAM 0x0ff00000 /* extended family */ diff --git a/arch/x86/kernel/cpu/cpufreq/speedstep-centrino.c b/arch/x86/kernel/cpu/cpufreq/speedstep-centrino.c index 8d672ef162c..9b1ff37de46 100644 --- a/arch/x86/kernel/cpu/cpufreq/speedstep-centrino.c +++ b/arch/x86/kernel/cpu/cpufreq/speedstep-centrino.c @@ -20,6 +20,7 @@ #include <linux/sched.h> /* current */ #include <linux/delay.h> #include <linux/compiler.h> +#include <linux/gfp.h> #include <asm/msr.h> #include <asm/processor.h> diff --git a/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c b/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c index 2ce8e0b5cc5..561758e9518 100644 --- a/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c +++ b/arch/x86/kernel/cpu/cpufreq/speedstep-ich.c @@ -23,7 +23,6 @@ #include <linux/init.h> #include <linux/cpufreq.h> #include <linux/pci.h> -#include <linux/slab.h> #include <linux/sched.h> #include "speedstep-lib.h" diff --git a/arch/x86/kernel/cpu/cpufreq/speedstep-lib.c b/arch/x86/kernel/cpu/cpufreq/speedstep-lib.c index ad0083abfa2..a94ec6be69f 100644 --- a/arch/x86/kernel/cpu/cpufreq/speedstep-lib.c +++ b/arch/x86/kernel/cpu/cpufreq/speedstep-lib.c @@ -13,7 +13,6 @@ #include <linux/moduleparam.h> #include <linux/init.h> #include <linux/cpufreq.h> -#include <linux/slab.h> #include <asm/msr.h> #include <asm/tsc.h> diff --git a/arch/x86/kernel/cpu/cpufreq/speedstep-smi.c b/arch/x86/kernel/cpu/cpufreq/speedstep-smi.c index 04d73c114e4..8abd869baab 100644 --- a/arch/x86/kernel/cpu/cpufreq/speedstep-smi.c +++ b/arch/x86/kernel/cpu/cpufreq/speedstep-smi.c @@ -17,7 +17,6 @@ #include <linux/moduleparam.h> #include <linux/init.h> #include <linux/cpufreq.h> -#include <linux/slab.h> #include <linux/delay.h> #include <linux/io.h> #include <asm/ist.h> diff --git a/arch/x86/kernel/cpu/hypervisor.c b/arch/x86/kernel/cpu/hypervisor.c index 08be922de33..dd531cc56a8 100644 --- a/arch/x86/kernel/cpu/hypervisor.c +++ b/arch/x86/kernel/cpu/hypervisor.c @@ -21,37 +21,55 @@ * */ +#include <linux/module.h> #include <asm/processor.h> -#include <asm/vmware.h> #include <asm/hypervisor.h> -static inline void __cpuinit -detect_hypervisor_vendor(struct cpuinfo_x86 *c) +/* + * Hypervisor detect order. This is specified explicitly here because + * some hypervisors might implement compatibility modes for other + * hypervisors and therefore need to be detected in specific sequence. + */ +static const __initconst struct hypervisor_x86 * const hypervisors[] = { - if (vmware_platform()) - c->x86_hyper_vendor = X86_HYPER_VENDOR_VMWARE; - else - c->x86_hyper_vendor = X86_HYPER_VENDOR_NONE; -} + &x86_hyper_vmware, + &x86_hyper_ms_hyperv, +}; -static inline void __cpuinit -hypervisor_set_feature_bits(struct cpuinfo_x86 *c) +const struct hypervisor_x86 *x86_hyper; +EXPORT_SYMBOL(x86_hyper); + +static inline void __init +detect_hypervisor_vendor(void) { - if (boot_cpu_data.x86_hyper_vendor == X86_HYPER_VENDOR_VMWARE) { - vmware_set_feature_bits(c); - return; + const struct hypervisor_x86 *h, * const *p; + + for (p = hypervisors; p < hypervisors + ARRAY_SIZE(hypervisors); p++) { + h = *p; + if (h->detect()) { + x86_hyper = h; + printk(KERN_INFO "Hypervisor detected: %s\n", h->name); + break; + } } } void __cpuinit init_hypervisor(struct cpuinfo_x86 *c) { - detect_hypervisor_vendor(c); - hypervisor_set_feature_bits(c); + if (x86_hyper && x86_hyper->set_cpu_features) + x86_hyper->set_cpu_features(c); } void __init init_hypervisor_platform(void) { + + detect_hypervisor_vendor(); + + if (!x86_hyper) + return; + init_hypervisor(&boot_cpu_data); - if (boot_cpu_data.x86_hyper_vendor == X86_HYPER_VENDOR_VMWARE) - vmware_platform_setup(); + + if (x86_hyper->init_platform) + x86_hyper->init_platform(); } diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c index 879666f4d87..85f69cdeae1 100644 --- a/arch/x86/kernel/cpu/intel.c +++ b/arch/x86/kernel/cpu/intel.c @@ -12,7 +12,6 @@ #include <asm/processor.h> #include <asm/pgtable.h> #include <asm/msr.h> -#include <asm/ds.h> #include <asm/bugs.h> #include <asm/cpu.h> @@ -47,6 +46,27 @@ static void __cpuinit early_init_intel(struct cpuinfo_x86 *c) (c->x86 == 0x6 && c->x86_model >= 0x0e)) set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC); + /* + * Atom erratum AAE44/AAF40/AAG38/AAH41: + * + * A race condition between speculative fetches and invalidating + * a large page. This is worked around in microcode, but we + * need the microcode to have already been loaded... so if it is + * not, recommend a BIOS update and disable large pages. + */ + if (c->x86 == 6 && c->x86_model == 0x1c && c->x86_mask <= 2) { + u32 ucode, junk; + + wrmsr(MSR_IA32_UCODE_REV, 0, 0); + sync_core(); + rdmsr(MSR_IA32_UCODE_REV, junk, ucode); + + if (ucode < 0x20e) { + printk(KERN_WARNING "Atom PSE erratum detected, BIOS microcode update recommended\n"); + clear_cpu_cap(c, X86_FEATURE_PSE); + } + } + #ifdef CONFIG_X86_64 set_cpu_cap(c, X86_FEATURE_SYSENTER32); #else @@ -70,7 +90,8 @@ static void __cpuinit early_init_intel(struct cpuinfo_x86 *c) if (c->x86_power & (1 << 8)) { set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC); set_cpu_cap(c, X86_FEATURE_NONSTOP_TSC); - sched_clock_stable = 1; + if (!check_tsc_unstable()) + sched_clock_stable = 1; } /* @@ -351,12 +372,6 @@ static void __cpuinit init_intel(struct cpuinfo_x86 *c) set_cpu_cap(c, X86_FEATURE_ARCH_PERFMON); } - if (c->cpuid_level > 6) { - unsigned ecx = cpuid_ecx(6); - if (ecx & 0x01) - set_cpu_cap(c, X86_FEATURE_APERFMPERF); - } - if (cpu_has_xmm2) set_cpu_cap(c, X86_FEATURE_LFENCE_RDTSC); if (cpu_has_ds) { @@ -366,7 +381,6 @@ static void __cpuinit init_intel(struct cpuinfo_x86 *c) set_cpu_cap(c, X86_FEATURE_BTS); if (!(l1 & (1<<12))) set_cpu_cap(c, X86_FEATURE_PEBS); - ds_init_intel(c); } if (c->x86 == 6 && c->x86_model == 29 && cpu_has_clflush) diff --git a/arch/x86/kernel/cpu/intel_cacheinfo.c b/arch/x86/kernel/cpu/intel_cacheinfo.c index fc6c8ef92dc..33eae2062cf 100644 --- a/arch/x86/kernel/cpu/intel_cacheinfo.c +++ b/arch/x86/kernel/cpu/intel_cacheinfo.c @@ -18,6 +18,7 @@ #include <asm/processor.h> #include <linux/smp.h> #include <asm/k8.h> +#include <asm/smp.h> #define LVL_1_INST 1 #define LVL_1_DATA 2 @@ -31,6 +32,8 @@ struct _cache_table { short size; }; +#define MB(x) ((x) * 1024) + /* All the cache descriptor types we care about (no TLB or trace cache entries) */ @@ -44,9 +47,9 @@ static const struct _cache_table __cpuinitconst cache_table[] = { 0x0d, LVL_1_DATA, 16 }, /* 4-way set assoc, 64 byte line size */ { 0x21, LVL_2, 256 }, /* 8-way set assoc, 64 byte line size */ { 0x22, LVL_3, 512 }, /* 4-way set assoc, sectored cache, 64 byte line size */ - { 0x23, LVL_3, 1024 }, /* 8-way set assoc, sectored cache, 64 byte line size */ - { 0x25, LVL_3, 2048 }, /* 8-way set assoc, sectored cache, 64 byte line size */ - { 0x29, LVL_3, 4096 }, /* 8-way set assoc, sectored cache, 64 byte line size */ + { 0x23, LVL_3, MB(1) }, /* 8-way set assoc, sectored cache, 64 byte line size */ + { 0x25, LVL_3, MB(2) }, /* 8-way set assoc, sectored cache, 64 byte line size */ + { 0x29, LVL_3, MB(4) }, /* 8-way set assoc, sectored cache, 64 byte line size */ { 0x2c, LVL_1_DATA, 32 }, /* 8-way set assoc, 64 byte line size */ { 0x30, LVL_1_INST, 32 }, /* 8-way set assoc, 64 byte line size */ { 0x39, LVL_2, 128 }, /* 4-way set assoc, sectored cache, 64 byte line size */ @@ -59,16 +62,16 @@ static const struct _cache_table __cpuinitconst cache_table[] = { 0x41, LVL_2, 128 }, /* 4-way set assoc, 32 byte line size */ { 0x42, LVL_2, 256 }, /* 4-way set assoc, 32 byte line size */ { 0x43, LVL_2, 512 }, /* 4-way set assoc, 32 byte line size */ - { 0x44, LVL_2, 1024 }, /* 4-way set assoc, 32 byte line size */ - { 0x45, LVL_2, 2048 }, /* 4-way set assoc, 32 byte line size */ - { 0x46, LVL_3, 4096 }, /* 4-way set assoc, 64 byte line size */ - { 0x47, LVL_3, 8192 }, /* 8-way set assoc, 64 byte line size */ - { 0x49, LVL_3, 4096 }, /* 16-way set assoc, 64 byte line size */ - { 0x4a, LVL_3, 6144 }, /* 12-way set assoc, 64 byte line size */ - { 0x4b, LVL_3, 8192 }, /* 16-way set assoc, 64 byte line size */ - { 0x4c, LVL_3, 12288 }, /* 12-way set assoc, 64 byte line size */ - { 0x4d, LVL_3, 16384 }, /* 16-way set assoc, 64 byte line size */ - { 0x4e, LVL_2, 6144 }, /* 24-way set assoc, 64 byte line size */ + { 0x44, LVL_2, MB(1) }, /* 4-way set assoc, 32 byte line size */ + { 0x45, LVL_2, MB(2) }, /* 4-way set assoc, 32 byte line size */ + { 0x46, LVL_3, MB(4) }, /* 4-way set assoc, 64 byte line size */ + { 0x47, LVL_3, MB(8) }, /* 8-way set assoc, 64 byte line size */ + { 0x49, LVL_3, MB(4) }, /* 16-way set assoc, 64 byte line size */ + { 0x4a, LVL_3, MB(6) }, /* 12-way set assoc, 64 byte line size */ + { 0x4b, LVL_3, MB(8) }, /* 16-way set assoc, 64 byte line size */ + { 0x4c, LVL_3, MB(12) }, /* 12-way set assoc, 64 byte line size */ + { 0x4d, LVL_3, MB(16) }, /* 16-way set assoc, 64 byte line size */ + { 0x4e, LVL_2, MB(6) }, /* 24-way set assoc, 64 byte line size */ { 0x60, LVL_1_DATA, 16 }, /* 8-way set assoc, sectored cache, 64 byte line size */ { 0x66, LVL_1_DATA, 8 }, /* 4-way set assoc, sectored cache, 64 byte line size */ { 0x67, LVL_1_DATA, 16 }, /* 4-way set assoc, sectored cache, 64 byte line size */ @@ -77,34 +80,34 @@ static const struct _cache_table __cpuinitconst cache_table[] = { 0x71, LVL_TRACE, 16 }, /* 8-way set assoc */ { 0x72, LVL_TRACE, 32 }, /* 8-way set assoc */ { 0x73, LVL_TRACE, 64 }, /* 8-way set assoc */ - { 0x78, LVL_2, 1024 }, /* 4-way set assoc, 64 byte line size */ - { 0x79, LVL_2, 128 }, /* 8-way set assoc, sectored cache, 64 byte line size */ - { 0x7a, LVL_2, 256 }, /* 8-way set assoc, sectored cache, 64 byte line size */ - { 0x7b, LVL_2, 512 }, /* 8-way set assoc, sectored cache, 64 byte line size */ - { 0x7c, LVL_2, 1024 }, /* 8-way set assoc, sectored cache, 64 byte line size */ - { 0x7d, LVL_2, 2048 }, /* 8-way set assoc, 64 byte line size */ - { 0x7f, LVL_2, 512 }, /* 2-way set assoc, 64 byte line size */ - { 0x82, LVL_2, 256 }, /* 8-way set assoc, 32 byte line size */ - { 0x83, LVL_2, 512 }, /* 8-way set assoc, 32 byte line size */ - { 0x84, LVL_2, 1024 }, /* 8-way set assoc, 32 byte line size */ - { 0x85, LVL_2, 2048 }, /* 8-way set assoc, 32 byte line size */ - { 0x86, LVL_2, 512 }, /* 4-way set assoc, 64 byte line size */ - { 0x87, LVL_2, 1024 }, /* 8-way set assoc, 64 byte line size */ - { 0xd0, LVL_3, 512 }, /* 4-way set assoc, 64 byte line size */ - { 0xd1, LVL_3, 1024 }, /* 4-way set assoc, 64 byte line size */ - { 0xd2, LVL_3, 2048 }, /* 4-way set assoc, 64 byte line size */ - { 0xd6, LVL_3, 1024 }, /* 8-way set assoc, 64 byte line size */ - { 0xd7, LVL_3, 2048 }, /* 8-way set assoc, 64 byte line size */ - { 0xd8, LVL_3, 4096 }, /* 12-way set assoc, 64 byte line size */ - { 0xdc, LVL_3, 2048 }, /* 12-way set assoc, 64 byte line size */ - { 0xdd, LVL_3, 4096 }, /* 12-way set assoc, 64 byte line size */ - { 0xde, LVL_3, 8192 }, /* 12-way set assoc, 64 byte line size */ - { 0xe2, LVL_3, 2048 }, /* 16-way set assoc, 64 byte line size */ - { 0xe3, LVL_3, 4096 }, /* 16-way set assoc, 64 byte line size */ - { 0xe4, LVL_3, 8192 }, /* 16-way set assoc, 64 byte line size */ - { 0xea, LVL_3, 12288 }, /* 24-way set assoc, 64 byte line size */ - { 0xeb, LVL_3, 18432 }, /* 24-way set assoc, 64 byte line size */ - { 0xec, LVL_3, 24576 }, /* 24-way set assoc, 64 byte line size */ + { 0x78, LVL_2, MB(1) }, /* 4-way set assoc, 64 byte line size */ + { 0x79, LVL_2, 128 }, /* 8-way set assoc, sectored cache, 64 byte line size */ + { 0x7a, LVL_2, 256 }, /* 8-way set assoc, sectored cache, 64 byte line size */ + { 0x7b, LVL_2, 512 }, /* 8-way set assoc, sectored cache, 64 byte line size */ + { 0x7c, LVL_2, MB(1) }, /* 8-way set assoc, sectored cache, 64 byte line size */ + { 0x7d, LVL_2, MB(2) }, /* 8-way set assoc, 64 byte line size */ + { 0x7f, LVL_2, 512 }, /* 2-way set assoc, 64 byte line size */ + { 0x82, LVL_2, 256 }, /* 8-way set assoc, 32 byte line size */ + { 0x83, LVL_2, 512 }, /* 8-way set assoc, 32 byte line size */ + { 0x84, LVL_2, MB(1) }, /* 8-way set assoc, 32 byte line size */ + { 0x85, LVL_2, MB(2) }, /* 8-way set assoc, 32 byte line size */ + { 0x86, LVL_2, 512 }, /* 4-way set assoc, 64 byte line size */ + { 0x87, LVL_2, MB(1) }, /* 8-way set assoc, 64 byte line size */ + { 0xd0, LVL_3, 512 }, /* 4-way set assoc, 64 byte line size */ + { 0xd1, LVL_3, MB(1) }, /* 4-way set assoc, 64 byte line size */ + { 0xd2, LVL_3, MB(2) }, /* 4-way set assoc, 64 byte line size */ + { 0xd6, LVL_3, MB(1) }, /* 8-way set assoc, 64 byte line size */ + { 0xd7, LVL_3, MB(2) }, /* 8-way set assoc, 64 byte line size */ + { 0xd8, LVL_3, MB(4) }, /* 12-way set assoc, 64 byte line size */ + { 0xdc, LVL_3, MB(2) }, /* 12-way set assoc, 64 byte line size */ + { 0xdd, LVL_3, MB(4) }, /* 12-way set assoc, 64 byte line size */ + { 0xde, LVL_3, MB(8) }, /* 12-way set assoc, 64 byte line size */ + { 0xe2, LVL_3, MB(2) }, /* 16-way set assoc, 64 byte line size */ + { 0xe3, LVL_3, MB(4) }, /* 16-way set assoc, 64 byte line size */ + { 0xe4, LVL_3, MB(8) }, /* 16-way set assoc, 64 byte line size */ + { 0xea, LVL_3, MB(12) }, /* 24-way set assoc, 64 byte line size */ + { 0xeb, LVL_3, MB(18) }, /* 24-way set assoc, 64 byte line size */ + { 0xec, LVL_3, MB(24) }, /* 24-way set assoc, 64 byte line size */ { 0x00, 0, 0} }; @@ -145,12 +148,19 @@ union _cpuid4_leaf_ecx { u32 full; }; +struct amd_l3_cache { + struct pci_dev *dev; + bool can_disable; + unsigned indices; + u8 subcaches[4]; +}; + struct _cpuid4_info { union _cpuid4_leaf_eax eax; union _cpuid4_leaf_ebx ebx; union _cpuid4_leaf_ecx ecx; unsigned long size; - unsigned long can_disable; + struct amd_l3_cache *l3; DECLARE_BITMAP(shared_cpu_map, NR_CPUS); }; @@ -160,7 +170,7 @@ struct _cpuid4_info_regs { union _cpuid4_leaf_ebx ebx; union _cpuid4_leaf_ecx ecx; unsigned long size; - unsigned long can_disable; + struct amd_l3_cache *l3; }; unsigned short num_cache_leaves; @@ -290,22 +300,213 @@ amd_cpuid4(int leaf, union _cpuid4_leaf_eax *eax, (ebx->split.ways_of_associativity + 1) - 1; } +struct _cache_attr { + struct attribute attr; + ssize_t (*show)(struct _cpuid4_info *, char *); + ssize_t (*store)(struct _cpuid4_info *, const char *, size_t count); +}; + +#ifdef CONFIG_CPU_SUP_AMD + +/* + * L3 cache descriptors + */ +static struct amd_l3_cache **__cpuinitdata l3_caches; + +static void __cpuinit amd_calc_l3_indices(struct amd_l3_cache *l3) +{ + unsigned int sc0, sc1, sc2, sc3; + u32 val = 0; + + pci_read_config_dword(l3->dev, 0x1C4, &val); + + /* calculate subcache sizes */ + l3->subcaches[0] = sc0 = !(val & BIT(0)); + l3->subcaches[1] = sc1 = !(val & BIT(4)); + l3->subcaches[2] = sc2 = !(val & BIT(8)) + !(val & BIT(9)); + l3->subcaches[3] = sc3 = !(val & BIT(12)) + !(val & BIT(13)); + + l3->indices = (max(max(max(sc0, sc1), sc2), sc3) << 10) - 1; +} + +static struct amd_l3_cache * __cpuinit amd_init_l3_cache(int node) +{ + struct amd_l3_cache *l3; + struct pci_dev *dev = node_to_k8_nb_misc(node); + + l3 = kzalloc(sizeof(struct amd_l3_cache), GFP_ATOMIC); + if (!l3) { + printk(KERN_WARNING "Error allocating L3 struct\n"); + return NULL; + } + + l3->dev = dev; + + amd_calc_l3_indices(l3); + + return l3; +} + static void __cpuinit amd_check_l3_disable(int index, struct _cpuid4_info_regs *this_leaf) { + int node; + + if (boot_cpu_data.x86 != 0x10) + return; + if (index < 3) return; - if (boot_cpu_data.x86 == 0x11) + /* see errata #382 and #388 */ + if (boot_cpu_data.x86_model < 0x8) return; - /* see erratum #382 */ - if ((boot_cpu_data.x86 == 0x10) && (boot_cpu_data.x86_model < 0x8)) + if ((boot_cpu_data.x86_model == 0x8 || + boot_cpu_data.x86_model == 0x9) + && + boot_cpu_data.x86_mask < 0x1) + return; + + /* not in virtualized environments */ + if (num_k8_northbridges == 0) return; - this_leaf->can_disable = 1; + /* + * Strictly speaking, the amount in @size below is leaked since it is + * never freed but this is done only on shutdown so it doesn't matter. + */ + if (!l3_caches) { + int size = num_k8_northbridges * sizeof(struct amd_l3_cache *); + + l3_caches = kzalloc(size, GFP_ATOMIC); + if (!l3_caches) + return; + } + + node = amd_get_nb_id(smp_processor_id()); + + if (!l3_caches[node]) { + l3_caches[node] = amd_init_l3_cache(node); + l3_caches[node]->can_disable = true; + } + + WARN_ON(!l3_caches[node]); + + this_leaf->l3 = l3_caches[node]; } +static ssize_t show_cache_disable(struct _cpuid4_info *this_leaf, char *buf, + unsigned int slot) +{ + struct pci_dev *dev = this_leaf->l3->dev; + unsigned int reg = 0; + + if (!this_leaf->l3 || !this_leaf->l3->can_disable) + return -EINVAL; + + if (!dev) + return -EINVAL; + + pci_read_config_dword(dev, 0x1BC + slot * 4, ®); + return sprintf(buf, "0x%08x\n", reg); +} + +#define SHOW_CACHE_DISABLE(slot) \ +static ssize_t \ +show_cache_disable_##slot(struct _cpuid4_info *this_leaf, char *buf) \ +{ \ + return show_cache_disable(this_leaf, buf, slot); \ +} +SHOW_CACHE_DISABLE(0) +SHOW_CACHE_DISABLE(1) + +static void amd_l3_disable_index(struct amd_l3_cache *l3, int cpu, + unsigned slot, unsigned long idx) +{ + int i; + + idx |= BIT(30); + + /* + * disable index in all 4 subcaches + */ + for (i = 0; i < 4; i++) { + u32 reg = idx | (i << 20); + + if (!l3->subcaches[i]) + continue; + + pci_write_config_dword(l3->dev, 0x1BC + slot * 4, reg); + + /* + * We need to WBINVD on a core on the node containing the L3 + * cache which indices we disable therefore a simple wbinvd() + * is not sufficient. + */ + wbinvd_on_cpu(cpu); + + reg |= BIT(31); + pci_write_config_dword(l3->dev, 0x1BC + slot * 4, reg); + } +} + + +static ssize_t store_cache_disable(struct _cpuid4_info *this_leaf, + const char *buf, size_t count, + unsigned int slot) +{ + struct pci_dev *dev = this_leaf->l3->dev; + int cpu = cpumask_first(to_cpumask(this_leaf->shared_cpu_map)); + unsigned long val = 0; + +#define SUBCACHE_MASK (3UL << 20) +#define SUBCACHE_INDEX 0xfff + + if (!this_leaf->l3 || !this_leaf->l3->can_disable) + return -EINVAL; + + if (!capable(CAP_SYS_ADMIN)) + return -EPERM; + + if (!dev) + return -EINVAL; + + if (strict_strtoul(buf, 10, &val) < 0) + return -EINVAL; + + /* do not allow writes outside of allowed bits */ + if ((val & ~(SUBCACHE_MASK | SUBCACHE_INDEX)) || + ((val & SUBCACHE_INDEX) > this_leaf->l3->indices)) + return -EINVAL; + + amd_l3_disable_index(this_leaf->l3, cpu, slot, val); + + return count; +} + +#define STORE_CACHE_DISABLE(slot) \ +static ssize_t \ +store_cache_disable_##slot(struct _cpuid4_info *this_leaf, \ + const char *buf, size_t count) \ +{ \ + return store_cache_disable(this_leaf, buf, count, slot); \ +} +STORE_CACHE_DISABLE(0) +STORE_CACHE_DISABLE(1) + +static struct _cache_attr cache_disable_0 = __ATTR(cache_disable_0, 0644, + show_cache_disable_0, store_cache_disable_0); +static struct _cache_attr cache_disable_1 = __ATTR(cache_disable_1, 0644, + show_cache_disable_1, store_cache_disable_1); + +#else /* CONFIG_CPU_SUP_AMD */ +static void __cpuinit +amd_check_l3_disable(int index, struct _cpuid4_info_regs *this_leaf) +{ +}; +#endif /* CONFIG_CPU_SUP_AMD */ + static int __cpuinit cpuid4_cache_lookup_regs(int index, struct _cpuid4_info_regs *this_leaf) @@ -317,8 +518,7 @@ __cpuinit cpuid4_cache_lookup_regs(int index, if (boot_cpu_data.x86_vendor == X86_VENDOR_AMD) { amd_cpuid4(index, &eax, &ebx, &ecx); - if (boot_cpu_data.x86 >= 0x10) - amd_check_l3_disable(index, this_leaf); + amd_check_l3_disable(index, this_leaf); } else { cpuid_count(4, index, &eax.full, &ebx.full, &ecx.full, &edx); } @@ -575,6 +775,7 @@ static void __cpuinit free_cache_attributes(unsigned int cpu) for (i = 0; i < num_cache_leaves; i++) cache_remove_shared_cpu_map(cpu, i); + kfree(per_cpu(ici_cpuid4_info, cpu)->l3); kfree(per_cpu(ici_cpuid4_info, cpu)); per_cpu(ici_cpuid4_info, cpu) = NULL; } @@ -711,82 +912,6 @@ static ssize_t show_type(struct _cpuid4_info *this_leaf, char *buf) #define to_object(k) container_of(k, struct _index_kobject, kobj) #define to_attr(a) container_of(a, struct _cache_attr, attr) -static ssize_t show_cache_disable(struct _cpuid4_info *this_leaf, char *buf, - unsigned int index) -{ - int cpu = cpumask_first(to_cpumask(this_leaf->shared_cpu_map)); - int node = cpu_to_node(cpu); - struct pci_dev *dev = node_to_k8_nb_misc(node); - unsigned int reg = 0; - - if (!this_leaf->can_disable) - return -EINVAL; - - if (!dev) - return -EINVAL; - - pci_read_config_dword(dev, 0x1BC + index * 4, ®); - return sprintf(buf, "%x\n", reg); -} - -#define SHOW_CACHE_DISABLE(index) \ -static ssize_t \ -show_cache_disable_##index(struct _cpuid4_info *this_leaf, char *buf) \ -{ \ - return show_cache_disable(this_leaf, buf, index); \ -} -SHOW_CACHE_DISABLE(0) -SHOW_CACHE_DISABLE(1) - -static ssize_t store_cache_disable(struct _cpuid4_info *this_leaf, - const char *buf, size_t count, unsigned int index) -{ - int cpu = cpumask_first(to_cpumask(this_leaf->shared_cpu_map)); - int node = cpu_to_node(cpu); - struct pci_dev *dev = node_to_k8_nb_misc(node); - unsigned long val = 0; - unsigned int scrubber = 0; - - if (!this_leaf->can_disable) - return -EINVAL; - - if (!capable(CAP_SYS_ADMIN)) - return -EPERM; - - if (!dev) - return -EINVAL; - - if (strict_strtoul(buf, 10, &val) < 0) - return -EINVAL; - - val |= 0xc0000000; - - pci_read_config_dword(dev, 0x58, &scrubber); - scrubber &= ~0x1f000000; - pci_write_config_dword(dev, 0x58, scrubber); - - pci_write_config_dword(dev, 0x1BC + index * 4, val & ~0x40000000); - wbinvd(); - pci_write_config_dword(dev, 0x1BC + index * 4, val); - return count; -} - -#define STORE_CACHE_DISABLE(index) \ -static ssize_t \ -store_cache_disable_##index(struct _cpuid4_info *this_leaf, \ - const char *buf, size_t count) \ -{ \ - return store_cache_disable(this_leaf, buf, count, index); \ -} -STORE_CACHE_DISABLE(0) -STORE_CACHE_DISABLE(1) - -struct _cache_attr { - struct attribute attr; - ssize_t (*show)(struct _cpuid4_info *, char *); - ssize_t (*store)(struct _cpuid4_info *, const char *, size_t count); -}; - #define define_one_ro(_name) \ static struct _cache_attr _name = \ __ATTR(_name, 0444, show_##_name, NULL) @@ -801,23 +926,28 @@ define_one_ro(size); define_one_ro(shared_cpu_map); define_one_ro(shared_cpu_list); -static struct _cache_attr cache_disable_0 = __ATTR(cache_disable_0, 0644, - show_cache_disable_0, store_cache_disable_0); -static struct _cache_attr cache_disable_1 = __ATTR(cache_disable_1, 0644, - show_cache_disable_1, store_cache_disable_1); +#define DEFAULT_SYSFS_CACHE_ATTRS \ + &type.attr, \ + &level.attr, \ + &coherency_line_size.attr, \ + &physical_line_partition.attr, \ + &ways_of_associativity.attr, \ + &number_of_sets.attr, \ + &size.attr, \ + &shared_cpu_map.attr, \ + &shared_cpu_list.attr static struct attribute *default_attrs[] = { - &type.attr, - &level.attr, - &coherency_line_size.attr, - &physical_line_partition.attr, - &ways_of_associativity.attr, - &number_of_sets.attr, - &size.attr, - &shared_cpu_map.attr, - &shared_cpu_list.attr, + DEFAULT_SYSFS_CACHE_ATTRS, + NULL +}; + +static struct attribute *default_l3_attrs[] = { + DEFAULT_SYSFS_CACHE_ATTRS, +#ifdef CONFIG_CPU_SUP_AMD &cache_disable_0.attr, &cache_disable_1.attr, +#endif NULL }; @@ -848,7 +978,7 @@ static ssize_t store(struct kobject *kobj, struct attribute *attr, return ret; } -static struct sysfs_ops sysfs_ops = { +static const struct sysfs_ops sysfs_ops = { .show = show, .store = store, }; @@ -908,6 +1038,7 @@ static int __cpuinit cache_add_dev(struct sys_device * sys_dev) unsigned int cpu = sys_dev->id; unsigned long i, j; struct _index_kobject *this_object; + struct _cpuid4_info *this_leaf; int retval; retval = cpuid4_cache_sysfs_init(cpu); @@ -926,6 +1057,14 @@ static int __cpuinit cache_add_dev(struct sys_device * sys_dev) this_object = INDEX_KOBJECT_PTR(cpu, i); this_object->cpu = cpu; this_object->index = i; + + this_leaf = CPUID4_INFO_IDX(cpu, i); + + if (this_leaf->l3 && this_leaf->l3->can_disable) + ktype_cache.default_attrs = default_l3_attrs; + else + ktype_cache.default_attrs = default_attrs; + retval = kobject_init_and_add(&(this_object->kobj), &ktype_cache, per_cpu(ici_cache_kobject, cpu), diff --git a/arch/x86/kernel/cpu/mcheck/Makefile b/arch/x86/kernel/cpu/mcheck/Makefile index 4ac6d48fe11..bb34b03af25 100644 --- a/arch/x86/kernel/cpu/mcheck/Makefile +++ b/arch/x86/kernel/cpu/mcheck/Makefile @@ -7,3 +7,5 @@ obj-$(CONFIG_X86_MCE_THRESHOLD) += threshold.o obj-$(CONFIG_X86_MCE_INJECT) += mce-inject.o obj-$(CONFIG_X86_THERMAL_VECTOR) += therm_throt.o + +obj-$(CONFIG_ACPI_APEI) += mce-apei.o diff --git a/arch/x86/kernel/cpu/mcheck/mce-apei.c b/arch/x86/kernel/cpu/mcheck/mce-apei.c new file mode 100644 index 00000000000..745b54f9be8 --- /dev/null +++ b/arch/x86/kernel/cpu/mcheck/mce-apei.c @@ -0,0 +1,138 @@ +/* + * Bridge between MCE and APEI + * + * On some machine, corrected memory errors are reported via APEI + * generic hardware error source (GHES) instead of corrected Machine + * Check. These corrected memory errors can be reported to user space + * through /dev/mcelog via faking a corrected Machine Check, so that + * the error memory page can be offlined by /sbin/mcelog if the error + * count for one page is beyond the threshold. + * + * For fatal MCE, save MCE record into persistent storage via ERST, so + * that the MCE record can be logged after reboot via ERST. + * + * Copyright 2010 Intel Corp. + * Author: Huang Ying <ying.huang@intel.com> + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include <linux/kernel.h> +#include <linux/acpi.h> +#include <linux/cper.h> +#include <acpi/apei.h> +#include <asm/mce.h> + +#include "mce-internal.h" + +void apei_mce_report_mem_error(int corrected, struct cper_sec_mem_err *mem_err) +{ + struct mce m; + + /* Only corrected MC is reported */ + if (!corrected) + return; + + mce_setup(&m); + m.bank = 1; + /* Fake a memory read corrected error with unknown channel */ + m.status = MCI_STATUS_VAL | MCI_STATUS_EN | MCI_STATUS_ADDRV | 0x9f; + m.addr = mem_err->physical_addr; + mce_log(&m); + mce_notify_irq(); +} +EXPORT_SYMBOL_GPL(apei_mce_report_mem_error); + +#define CPER_CREATOR_MCE \ + UUID_LE(0x75a574e3, 0x5052, 0x4b29, 0x8a, 0x8e, 0xbe, 0x2c, \ + 0x64, 0x90, 0xb8, 0x9d) +#define CPER_SECTION_TYPE_MCE \ + UUID_LE(0xfe08ffbe, 0x95e4, 0x4be7, 0xbc, 0x73, 0x40, 0x96, \ + 0x04, 0x4a, 0x38, 0xfc) + +/* + * CPER specification (in UEFI specification 2.3 appendix N) requires + * byte-packed. + */ +struct cper_mce_record { + struct cper_record_header hdr; + struct cper_section_descriptor sec_hdr; + struct mce mce; +} __packed; + +int apei_write_mce(struct mce *m) +{ + struct cper_mce_record rcd; + + memset(&rcd, 0, sizeof(rcd)); + memcpy(rcd.hdr.signature, CPER_SIG_RECORD, CPER_SIG_SIZE); + rcd.hdr.revision = CPER_RECORD_REV; + rcd.hdr.signature_end = CPER_SIG_END; + rcd.hdr.section_count = 1; + rcd.hdr.error_severity = CPER_SER_FATAL; + /* timestamp, platform_id, partition_id are all invalid */ + rcd.hdr.validation_bits = 0; + rcd.hdr.record_length = sizeof(rcd); + rcd.hdr.creator_id = CPER_CREATOR_MCE; + rcd.hdr.notification_type = CPER_NOTIFY_MCE; + rcd.hdr.record_id = cper_next_record_id(); + rcd.hdr.flags = CPER_HW_ERROR_FLAGS_PREVERR; + + rcd.sec_hdr.section_offset = (void *)&rcd.mce - (void *)&rcd; + rcd.sec_hdr.section_length = sizeof(rcd.mce); + rcd.sec_hdr.revision = CPER_SEC_REV; + /* fru_id and fru_text is invalid */ + rcd.sec_hdr.validation_bits = 0; + rcd.sec_hdr.flags = CPER_SEC_PRIMARY; + rcd.sec_hdr.section_type = CPER_SECTION_TYPE_MCE; + rcd.sec_hdr.section_severity = CPER_SER_FATAL; + + memcpy(&rcd.mce, m, sizeof(*m)); + + return erst_write(&rcd.hdr); +} + +ssize_t apei_read_mce(struct mce *m, u64 *record_id) +{ + struct cper_mce_record rcd; + ssize_t len; + + len = erst_read_next(&rcd.hdr, sizeof(rcd)); + if (len <= 0) + return len; + /* Can not skip other records in storage via ERST unless clear them */ + else if (len != sizeof(rcd) || + uuid_le_cmp(rcd.hdr.creator_id, CPER_CREATOR_MCE)) { + if (printk_ratelimit()) + pr_warning( + "MCE-APEI: Can not skip the unknown record in ERST"); + return -EIO; + } + + memcpy(m, &rcd.mce, sizeof(*m)); + *record_id = rcd.hdr.record_id; + + return sizeof(*m); +} + +/* Check whether there is record in ERST */ +int apei_check_mce(void) +{ + return erst_get_record_count(); +} + +int apei_clear_mce(u64 record_id) +{ + return erst_clear(record_id); +} diff --git a/arch/x86/kernel/cpu/mcheck/mce-inject.c b/arch/x86/kernel/cpu/mcheck/mce-inject.c index 73734baa50f..e7dbde7bfed 100644 --- a/arch/x86/kernel/cpu/mcheck/mce-inject.c +++ b/arch/x86/kernel/cpu/mcheck/mce-inject.c @@ -22,6 +22,7 @@ #include <linux/kdebug.h> #include <linux/cpu.h> #include <linux/sched.h> +#include <linux/gfp.h> #include <asm/mce.h> #include <asm/apic.h> diff --git a/arch/x86/kernel/cpu/mcheck/mce-internal.h b/arch/x86/kernel/cpu/mcheck/mce-internal.h index 32996f9fab6..fefcc69ee8b 100644 --- a/arch/x86/kernel/cpu/mcheck/mce-internal.h +++ b/arch/x86/kernel/cpu/mcheck/mce-internal.h @@ -28,3 +28,26 @@ extern int mce_ser; extern struct mce_bank *mce_banks; +#ifdef CONFIG_ACPI_APEI +int apei_write_mce(struct mce *m); +ssize_t apei_read_mce(struct mce *m, u64 *record_id); +int apei_check_mce(void); +int apei_clear_mce(u64 record_id); +#else +static inline int apei_write_mce(struct mce *m) +{ + return -EINVAL; +} +static inline ssize_t apei_read_mce(struct mce *m, u64 *record_id) +{ + return 0; +} +static inline int apei_check_mce(void) +{ + return 0; +} +static inline int apei_clear_mce(u64 record_id) +{ + return -EINVAL; +} +#endif diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c index a8aacd4b513..707165dbc20 100644 --- a/arch/x86/kernel/cpu/mcheck/mce.c +++ b/arch/x86/kernel/cpu/mcheck/mce.c @@ -26,6 +26,7 @@ #include <linux/sched.h> #include <linux/sysfs.h> #include <linux/types.h> +#include <linux/slab.h> #include <linux/init.h> #include <linux/kmod.h> #include <linux/poll.h> @@ -46,6 +47,13 @@ #include "mce-internal.h" +static DEFINE_MUTEX(mce_read_mutex); + +#define rcu_dereference_check_mce(p) \ + rcu_dereference_check((p), \ + rcu_read_lock_sched_held() || \ + lockdep_is_held(&mce_read_mutex)) + #define CREATE_TRACE_POINTS #include <trace/events/mce.h> @@ -158,7 +166,7 @@ void mce_log(struct mce *mce) mce->finished = 0; wmb(); for (;;) { - entry = rcu_dereference(mcelog.next); + entry = rcu_dereference_check_mce(mcelog.next); for (;;) { /* * When the buffer fills up discard new entries. @@ -256,7 +264,7 @@ static void wait_for_panic(void) static void mce_panic(char *msg, struct mce *final, char *exp) { - int i; + int i, apei_err = 0; if (!fake_panic) { /* @@ -279,8 +287,11 @@ static void mce_panic(char *msg, struct mce *final, char *exp) struct mce *m = &mcelog.entry[i]; if (!(m->status & MCI_STATUS_VAL)) continue; - if (!(m->status & MCI_STATUS_UC)) + if (!(m->status & MCI_STATUS_UC)) { print_mce(m); + if (!apei_err) + apei_err = apei_write_mce(m); + } } /* Now print uncorrected but with the final one last */ for (i = 0; i < MCE_LOG_LEN; i++) { @@ -289,11 +300,17 @@ static void mce_panic(char *msg, struct mce *final, char *exp) continue; if (!(m->status & MCI_STATUS_UC)) continue; - if (!final || memcmp(m, final, sizeof(struct mce))) + if (!final || memcmp(m, final, sizeof(struct mce))) { print_mce(m); + if (!apei_err) + apei_err = apei_write_mce(m); + } } - if (final) + if (final) { print_mce(final); + if (!apei_err) + apei_err = apei_write_mce(final); + } if (cpu_missing) printk(KERN_EMERG "Some CPUs didn't answer in synchronization\n"); print_mce_tail(); @@ -531,7 +548,7 @@ void machine_check_poll(enum mcp_flags flags, mce_banks_t *b) struct mce m; int i; - __get_cpu_var(mce_poll_count)++; + percpu_inc(mce_poll_count); mce_setup(&m); @@ -926,7 +943,7 @@ void do_machine_check(struct pt_regs *regs, long error_code) atomic_inc(&mce_entry); - __get_cpu_var(mce_exception_count)++; + percpu_inc(mce_exception_count); if (notify_die(DIE_NMI, "machine check", regs, error_code, 18, SIGKILL) == NOTIFY_STOP) @@ -1485,7 +1502,42 @@ static void collect_tscs(void *data) rdtscll(cpu_tsc[smp_processor_id()]); } -static DEFINE_MUTEX(mce_read_mutex); +static int mce_apei_read_done; + +/* Collect MCE record of previous boot in persistent storage via APEI ERST. */ +static int __mce_read_apei(char __user **ubuf, size_t usize) +{ + int rc; + u64 record_id; + struct mce m; + + if (usize < sizeof(struct mce)) + return -EINVAL; + + rc = apei_read_mce(&m, &record_id); + /* Error or no more MCE record */ + if (rc <= 0) { + mce_apei_read_done = 1; + return rc; + } + rc = -EFAULT; + if (copy_to_user(*ubuf, &m, sizeof(struct mce))) + return rc; + /* + * In fact, we should have cleared the record after that has + * been flushed to the disk or sent to network in + * /sbin/mcelog, but we have no interface to support that now, + * so just clear it to avoid duplication. + */ + rc = apei_clear_mce(record_id); + if (rc) { + mce_apei_read_done = 1; + return rc; + } + *ubuf += sizeof(struct mce); + + return 0; +} static ssize_t mce_read(struct file *filp, char __user *ubuf, size_t usize, loff_t *off) @@ -1500,16 +1552,20 @@ static ssize_t mce_read(struct file *filp, char __user *ubuf, size_t usize, return -ENOMEM; mutex_lock(&mce_read_mutex); - next = rcu_dereference(mcelog.next); - /* Only supports full reads right now */ - if (*off != 0 || usize < MCE_LOG_LEN*sizeof(struct mce)) { - mutex_unlock(&mce_read_mutex); - kfree(cpu_tsc); - - return -EINVAL; + if (!mce_apei_read_done) { + err = __mce_read_apei(&buf, usize); + if (err || buf != ubuf) + goto out; } + next = rcu_dereference_check_mce(mcelog.next); + + /* Only supports full reads right now */ + err = -EINVAL; + if (*off != 0 || usize < MCE_LOG_LEN*sizeof(struct mce)) + goto out; + err = 0; prev = 0; do { @@ -1556,16 +1612,23 @@ timeout: memset(&mcelog.entry[i], 0, sizeof(struct mce)); } } + + if (err) + err = -EFAULT; + +out: mutex_unlock(&mce_read_mutex); kfree(cpu_tsc); - return err ? -EFAULT : buf - ubuf; + return err ? err : buf - ubuf; } static unsigned int mce_poll(struct file *file, poll_table *wait) { poll_wait(file, &mce_wait, wait); - if (rcu_dereference(mcelog.next)) + if (rcu_dereference_check_mce(mcelog.next)) + return POLLIN | POLLRDNORM; + if (!mce_apei_read_done && apei_check_mce()) return POLLIN | POLLRDNORM; return 0; } @@ -2044,6 +2107,7 @@ static __init void mce_init_banks(void) struct mce_bank *b = &mce_banks[i]; struct sysdev_attribute *a = &b->attr; + sysfs_attr_init(&a->attr); a->attr.name = b->attrname; snprintf(b->attrname, ATTR_LEN, "bank%d", i); diff --git a/arch/x86/kernel/cpu/mcheck/mce_amd.c b/arch/x86/kernel/cpu/mcheck/mce_amd.c index 83a3d1f4efc..224392d8fe8 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_amd.c +++ b/arch/x86/kernel/cpu/mcheck/mce_amd.c @@ -21,6 +21,7 @@ #include <linux/errno.h> #include <linux/sched.h> #include <linux/sysfs.h> +#include <linux/slab.h> #include <linux/init.h> #include <linux/cpu.h> #include <linux/smp.h> @@ -388,7 +389,7 @@ static ssize_t store(struct kobject *kobj, struct attribute *attr, return ret; } -static struct sysfs_ops threshold_ops = { +static const struct sysfs_ops threshold_ops = { .show = show, .store = store, }; diff --git a/arch/x86/kernel/cpu/mcheck/mce_intel.c b/arch/x86/kernel/cpu/mcheck/mce_intel.c index 7c785634af2..62b48e40920 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_intel.c +++ b/arch/x86/kernel/cpu/mcheck/mce_intel.c @@ -5,6 +5,7 @@ * Author: Andi Kleen */ +#include <linux/gfp.h> #include <linux/init.h> #include <linux/interrupt.h> #include <linux/percpu.h> @@ -95,7 +96,7 @@ static void cmci_discover(int banks, int boot) /* Already owned by someone else? */ if (val & CMCI_EN) { - if (test_and_clear_bit(i, owned) || boot) + if (test_and_clear_bit(i, owned) && !boot) print_update("SHD", &hdr, i); __clear_bit(i, __get_cpu_var(mce_poll_banks)); continue; @@ -107,7 +108,7 @@ static void cmci_discover(int banks, int boot) /* Did the enable bit stick? -- the bank supports CMCI */ if (val & CMCI_EN) { - if (!test_and_set_bit(i, owned) || boot) + if (!test_and_set_bit(i, owned) && !boot) print_update("CMCI", &hdr, i); __clear_bit(i, __get_cpu_var(mce_poll_banks)); } else { diff --git a/arch/x86/kernel/cpu/mcheck/therm_throt.c b/arch/x86/kernel/cpu/mcheck/therm_throt.c index 81c499eceb2..e1a0a3bf971 100644 --- a/arch/x86/kernel/cpu/mcheck/therm_throt.c +++ b/arch/x86/kernel/cpu/mcheck/therm_throt.c @@ -190,7 +190,7 @@ thermal_throttle_cpu_callback(struct notifier_block *nfb, mutex_unlock(&therm_cpu_lock); break; } - return err ? NOTIFY_BAD : NOTIFY_OK; + return notifier_from_errno(err); } static struct notifier_block thermal_throttle_cpu_notifier __cpuinitdata = diff --git a/arch/x86/kernel/cpu/mshyperv.c b/arch/x86/kernel/cpu/mshyperv.c new file mode 100644 index 00000000000..16f41bbe46b --- /dev/null +++ b/arch/x86/kernel/cpu/mshyperv.c @@ -0,0 +1,55 @@ +/* + * HyperV Detection code. + * + * Copyright (C) 2010, Novell, Inc. + * Author : K. Y. Srinivasan <ksrinivasan@novell.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; version 2 of the License. + * + */ + +#include <linux/types.h> +#include <linux/module.h> +#include <asm/processor.h> +#include <asm/hypervisor.h> +#include <asm/hyperv.h> +#include <asm/mshyperv.h> + +struct ms_hyperv_info ms_hyperv; + +static bool __init ms_hyperv_platform(void) +{ + u32 eax; + u32 hyp_signature[3]; + + if (!boot_cpu_has(X86_FEATURE_HYPERVISOR)) + return false; + + cpuid(HYPERV_CPUID_VENDOR_AND_MAX_FUNCTIONS, + &eax, &hyp_signature[0], &hyp_signature[1], &hyp_signature[2]); + + return eax >= HYPERV_CPUID_MIN && + eax <= HYPERV_CPUID_MAX && + !memcmp("Microsoft Hv", hyp_signature, 12); +} + +static void __init ms_hyperv_init_platform(void) +{ + /* + * Extract the features and hints + */ + ms_hyperv.features = cpuid_eax(HYPERV_CPUID_FEATURES); + ms_hyperv.hints = cpuid_eax(HYPERV_CPUID_ENLIGHTMENT_INFO); + + printk(KERN_INFO "HyperV: features 0x%x, hints 0x%x\n", + ms_hyperv.features, ms_hyperv.hints); +} + +const __refconst struct hypervisor_x86 x86_hyper_ms_hyperv = { + .name = "Microsoft HyperV", + .detect = ms_hyperv_platform, + .init_platform = ms_hyperv_init_platform, +}; +EXPORT_SYMBOL(x86_hyper_ms_hyperv); diff --git a/arch/x86/kernel/cpu/mtrr/Makefile b/arch/x86/kernel/cpu/mtrr/Makefile index f4361b56f8e..ad9e5ed8118 100644 --- a/arch/x86/kernel/cpu/mtrr/Makefile +++ b/arch/x86/kernel/cpu/mtrr/Makefile @@ -1,3 +1,3 @@ -obj-y := main.o if.o generic.o state.o cleanup.o +obj-y := main.o if.o generic.o cleanup.o obj-$(CONFIG_X86_32) += amd.o cyrix.o centaur.o diff --git a/arch/x86/kernel/cpu/mtrr/amd.c b/arch/x86/kernel/cpu/mtrr/amd.c index 33af14110df..92ba9cd31c9 100644 --- a/arch/x86/kernel/cpu/mtrr/amd.c +++ b/arch/x86/kernel/cpu/mtrr/amd.c @@ -108,7 +108,7 @@ amd_validate_add_page(unsigned long base, unsigned long size, unsigned int type) return 0; } -static struct mtrr_ops amd_mtrr_ops = { +static const struct mtrr_ops amd_mtrr_ops = { .vendor = X86_VENDOR_AMD, .set = amd_set_mtrr, .get = amd_get_mtrr, diff --git a/arch/x86/kernel/cpu/mtrr/centaur.c b/arch/x86/kernel/cpu/mtrr/centaur.c index de89f14eff3..316fe3e60a9 100644 --- a/arch/x86/kernel/cpu/mtrr/centaur.c +++ b/arch/x86/kernel/cpu/mtrr/centaur.c @@ -110,7 +110,7 @@ centaur_validate_add_page(unsigned long base, unsigned long size, unsigned int t return 0; } -static struct mtrr_ops centaur_mtrr_ops = { +static const struct mtrr_ops centaur_mtrr_ops = { .vendor = X86_VENDOR_CENTAUR, .set = centaur_set_mcr, .get = centaur_get_mcr, diff --git a/arch/x86/kernel/cpu/mtrr/cleanup.c b/arch/x86/kernel/cpu/mtrr/cleanup.c index 09b1698e046..06130b52f01 100644 --- a/arch/x86/kernel/cpu/mtrr/cleanup.c +++ b/arch/x86/kernel/cpu/mtrr/cleanup.c @@ -22,10 +22,10 @@ #include <linux/pci.h> #include <linux/smp.h> #include <linux/cpu.h> -#include <linux/sort.h> #include <linux/mutex.h> #include <linux/uaccess.h> #include <linux/kvm_para.h> +#include <linux/range.h> #include <asm/processor.h> #include <asm/e820.h> @@ -34,11 +34,6 @@ #include "mtrr.h" -struct res_range { - unsigned long start; - unsigned long end; -}; - struct var_mtrr_range_state { unsigned long base_pfn; unsigned long size_pfn; @@ -56,7 +51,7 @@ struct var_mtrr_state { /* Should be related to MTRR_VAR_RANGES nums */ #define RANGE_NUM 256 -static struct res_range __initdata range[RANGE_NUM]; +static struct range __initdata range[RANGE_NUM]; static int __initdata nr_range; static struct var_mtrr_range_state __initdata range_state[RANGE_NUM]; @@ -64,152 +59,11 @@ static struct var_mtrr_range_state __initdata range_state[RANGE_NUM]; static int __initdata debug_print; #define Dprintk(x...) do { if (debug_print) printk(KERN_DEBUG x); } while (0) - -static int __init -add_range(struct res_range *range, int nr_range, - unsigned long start, unsigned long end) -{ - /* Out of slots: */ - if (nr_range >= RANGE_NUM) - return nr_range; - - range[nr_range].start = start; - range[nr_range].end = end; - - nr_range++; - - return nr_range; -} - -static int __init -add_range_with_merge(struct res_range *range, int nr_range, - unsigned long start, unsigned long end) -{ - int i; - - /* Try to merge it with old one: */ - for (i = 0; i < nr_range; i++) { - unsigned long final_start, final_end; - unsigned long common_start, common_end; - - if (!range[i].end) - continue; - - common_start = max(range[i].start, start); - common_end = min(range[i].end, end); - if (common_start > common_end + 1) - continue; - - final_start = min(range[i].start, start); - final_end = max(range[i].end, end); - - range[i].start = final_start; - range[i].end = final_end; - return nr_range; - } - - /* Need to add it: */ - return add_range(range, nr_range, start, end); -} - -static void __init -subtract_range(struct res_range *range, unsigned long start, unsigned long end) -{ - int i, j; - - for (j = 0; j < RANGE_NUM; j++) { - if (!range[j].end) - continue; - - if (start <= range[j].start && end >= range[j].end) { - range[j].start = 0; - range[j].end = 0; - continue; - } - - if (start <= range[j].start && end < range[j].end && - range[j].start < end + 1) { - range[j].start = end + 1; - continue; - } - - - if (start > range[j].start && end >= range[j].end && - range[j].end > start - 1) { - range[j].end = start - 1; - continue; - } - - if (start > range[j].start && end < range[j].end) { - /* Find the new spare: */ - for (i = 0; i < RANGE_NUM; i++) { - if (range[i].end == 0) - break; - } - if (i < RANGE_NUM) { - range[i].end = range[j].end; - range[i].start = end + 1; - } else { - printk(KERN_ERR "run of slot in ranges\n"); - } - range[j].end = start - 1; - continue; - } - } -} - -static int __init cmp_range(const void *x1, const void *x2) -{ - const struct res_range *r1 = x1; - const struct res_range *r2 = x2; - long start1, start2; - - start1 = r1->start; - start2 = r2->start; - - return start1 - start2; -} - -static int __init clean_sort_range(struct res_range *range, int az) -{ - int i, j, k = az - 1, nr_range = 0; - - for (i = 0; i < k; i++) { - if (range[i].end) - continue; - for (j = k; j > i; j--) { - if (range[j].end) { - k = j; - break; - } - } - if (j == i) - break; - range[i].start = range[k].start; - range[i].end = range[k].end; - range[k].start = 0; - range[k].end = 0; - k--; - } - /* count it */ - for (i = 0; i < az; i++) { - if (!range[i].end) { - nr_range = i; - break; - } - } - - /* sort them */ - sort(range, nr_range, sizeof(struct res_range), cmp_range, NULL); - - return nr_range; -} - #define BIOS_BUG_MSG KERN_WARNING \ "WARNING: BIOS bug: VAR MTRR %d contains strange UC entry under 1M, check with your system vendor!\n" static int __init -x86_get_mtrr_mem_range(struct res_range *range, int nr_range, +x86_get_mtrr_mem_range(struct range *range, int nr_range, unsigned long extra_remove_base, unsigned long extra_remove_size) { @@ -223,14 +77,14 @@ x86_get_mtrr_mem_range(struct res_range *range, int nr_range, continue; base = range_state[i].base_pfn; size = range_state[i].size_pfn; - nr_range = add_range_with_merge(range, nr_range, base, - base + size - 1); + nr_range = add_range_with_merge(range, RANGE_NUM, nr_range, + base, base + size); } if (debug_print) { printk(KERN_DEBUG "After WB checking\n"); for (i = 0; i < nr_range; i++) - printk(KERN_DEBUG "MTRR MAP PFN: %016lx - %016lx\n", - range[i].start, range[i].end + 1); + printk(KERN_DEBUG "MTRR MAP PFN: %016llx - %016llx\n", + range[i].start, range[i].end); } /* Take out UC ranges: */ @@ -252,19 +106,19 @@ x86_get_mtrr_mem_range(struct res_range *range, int nr_range, size -= (1<<(20-PAGE_SHIFT)) - base; base = 1<<(20-PAGE_SHIFT); } - subtract_range(range, base, base + size - 1); + subtract_range(range, RANGE_NUM, base, base + size); } if (extra_remove_size) - subtract_range(range, extra_remove_base, - extra_remove_base + extra_remove_size - 1); + subtract_range(range, RANGE_NUM, extra_remove_base, + extra_remove_base + extra_remove_size); if (debug_print) { printk(KERN_DEBUG "After UC checking\n"); for (i = 0; i < RANGE_NUM; i++) { if (!range[i].end) continue; - printk(KERN_DEBUG "MTRR MAP PFN: %016lx - %016lx\n", - range[i].start, range[i].end + 1); + printk(KERN_DEBUG "MTRR MAP PFN: %016llx - %016llx\n", + range[i].start, range[i].end); } } @@ -273,26 +127,22 @@ x86_get_mtrr_mem_range(struct res_range *range, int nr_range, if (debug_print) { printk(KERN_DEBUG "After sorting\n"); for (i = 0; i < nr_range; i++) - printk(KERN_DEBUG "MTRR MAP PFN: %016lx - %016lx\n", - range[i].start, range[i].end + 1); + printk(KERN_DEBUG "MTRR MAP PFN: %016llx - %016llx\n", + range[i].start, range[i].end); } - /* clear those is not used */ - for (i = nr_range; i < RANGE_NUM; i++) - memset(&range[i], 0, sizeof(range[i])); - return nr_range; } #ifdef CONFIG_MTRR_SANITIZER -static unsigned long __init sum_ranges(struct res_range *range, int nr_range) +static unsigned long __init sum_ranges(struct range *range, int nr_range) { unsigned long sum = 0; int i; for (i = 0; i < nr_range; i++) - sum += range[i].end + 1 - range[i].start; + sum += range[i].end - range[i].start; return sum; } @@ -621,7 +471,7 @@ static int __init parse_mtrr_spare_reg(char *arg) early_param("mtrr_spare_reg_nr", parse_mtrr_spare_reg); static int __init -x86_setup_var_mtrrs(struct res_range *range, int nr_range, +x86_setup_var_mtrrs(struct range *range, int nr_range, u64 chunk_size, u64 gran_size) { struct var_mtrr_state var_state; @@ -639,7 +489,7 @@ x86_setup_var_mtrrs(struct res_range *range, int nr_range, /* Write the range: */ for (i = 0; i < nr_range; i++) { set_var_mtrr_range(&var_state, range[i].start, - range[i].end - range[i].start + 1); + range[i].end - range[i].start); } /* Write the last range: */ @@ -742,7 +592,7 @@ mtrr_calc_range_state(u64 chunk_size, u64 gran_size, unsigned long x_remove_base, unsigned long x_remove_size, int i) { - static struct res_range range_new[RANGE_NUM]; + static struct range range_new[RANGE_NUM]; unsigned long range_sums_new; static int nr_range_new; int num_reg; @@ -869,10 +719,10 @@ int __init mtrr_cleanup(unsigned address_bits) * [0, 1M) should always be covered by var mtrr with WB * and fixed mtrrs should take effect before var mtrr for it: */ - nr_range = add_range_with_merge(range, nr_range, 0, - (1ULL<<(20 - PAGE_SHIFT)) - 1); + nr_range = add_range_with_merge(range, RANGE_NUM, nr_range, 0, + 1ULL<<(20 - PAGE_SHIFT)); /* Sort the ranges: */ - sort(range, nr_range, sizeof(struct res_range), cmp_range, NULL); + sort_range(range, nr_range); range_sums = sum_ranges(range, nr_range); printk(KERN_INFO "total RAM covered: %ldM\n", @@ -1089,9 +939,9 @@ int __init mtrr_trim_uncached_memory(unsigned long end_pfn) nr_range = 0; if (mtrr_tom2) { range[nr_range].start = (1ULL<<(32 - PAGE_SHIFT)); - range[nr_range].end = (mtrr_tom2 >> PAGE_SHIFT) - 1; - if (highest_pfn < range[nr_range].end + 1) - highest_pfn = range[nr_range].end + 1; + range[nr_range].end = mtrr_tom2 >> PAGE_SHIFT; + if (highest_pfn < range[nr_range].end) + highest_pfn = range[nr_range].end; nr_range++; } nr_range = x86_get_mtrr_mem_range(range, nr_range, 0, 0); @@ -1103,15 +953,15 @@ int __init mtrr_trim_uncached_memory(unsigned long end_pfn) /* Check the holes: */ for (i = 0; i < nr_range - 1; i++) { - if (range[i].end + 1 < range[i+1].start) - total_trim_size += real_trim_memory(range[i].end + 1, + if (range[i].end < range[i+1].start) + total_trim_size += real_trim_memory(range[i].end, range[i+1].start); } /* Check the top: */ i = nr_range - 1; - if (range[i].end + 1 < end_pfn) - total_trim_size += real_trim_memory(range[i].end + 1, + if (range[i].end < end_pfn) + total_trim_size += real_trim_memory(range[i].end, end_pfn); if (total_trim_size) { diff --git a/arch/x86/kernel/cpu/mtrr/cyrix.c b/arch/x86/kernel/cpu/mtrr/cyrix.c index 228d982ce09..68a3343e579 100644 --- a/arch/x86/kernel/cpu/mtrr/cyrix.c +++ b/arch/x86/kernel/cpu/mtrr/cyrix.c @@ -265,7 +265,7 @@ static void cyrix_set_all(void) post_set(); } -static struct mtrr_ops cyrix_mtrr_ops = { +static const struct mtrr_ops cyrix_mtrr_ops = { .vendor = X86_VENDOR_CYRIX, .set_all = cyrix_set_all, .set = cyrix_set_arr, diff --git a/arch/x86/kernel/cpu/mtrr/generic.c b/arch/x86/kernel/cpu/mtrr/generic.c index 55da0c5f68d..fd31a441c61 100644 --- a/arch/x86/kernel/cpu/mtrr/generic.c +++ b/arch/x86/kernel/cpu/mtrr/generic.c @@ -6,7 +6,6 @@ #include <linux/module.h> #include <linux/init.h> -#include <linux/slab.h> #include <linux/io.h> #include <linux/mm.h> @@ -464,7 +463,7 @@ static void generic_get_mtrr(unsigned int reg, unsigned long *base, tmp |= ~((1<<(hi - 1)) - 1); if (tmp != mask_lo) { - WARN_ONCE(1, KERN_INFO "mtrr: your BIOS has set up an incorrect mask, fixing it up.\n"); + printk(KERN_WARNING "mtrr: your BIOS has configured an incorrect mask, fixing it.\n"); mask_lo = tmp; } } @@ -570,7 +569,7 @@ static unsigned long set_mtrr_state(void) static unsigned long cr4; -static DEFINE_SPINLOCK(set_atomicity_lock); +static DEFINE_RAW_SPINLOCK(set_atomicity_lock); /* * Since we are disabling the cache don't allow any interrupts, @@ -590,7 +589,7 @@ static void prepare_set(void) __acquires(set_atomicity_lock) * changes to the way the kernel boots */ - spin_lock(&set_atomicity_lock); + raw_spin_lock(&set_atomicity_lock); /* Enter the no-fill (CD=1, NW=0) cache mode and flush caches. */ cr0 = read_cr0() | X86_CR0_CD; @@ -627,7 +626,7 @@ static void post_set(void) __releases(set_atomicity_lock) /* Restore value of CR4 */ if (cpu_has_pge) write_cr4(cr4); - spin_unlock(&set_atomicity_lock); + raw_spin_unlock(&set_atomicity_lock); } static void generic_set_all(void) @@ -752,7 +751,7 @@ int positive_have_wrcomb(void) /* * Generic structure... */ -struct mtrr_ops generic_mtrr_ops = { +const struct mtrr_ops generic_mtrr_ops = { .use_intel_if = 1, .set_all = generic_set_all, .get = generic_get_mtrr, diff --git a/arch/x86/kernel/cpu/mtrr/if.c b/arch/x86/kernel/cpu/mtrr/if.c index e006e56f699..79289632cb2 100644 --- a/arch/x86/kernel/cpu/mtrr/if.c +++ b/arch/x86/kernel/cpu/mtrr/if.c @@ -5,6 +5,7 @@ #include <linux/module.h> #include <linux/ctype.h> #include <linux/string.h> +#include <linux/slab.h> #include <linux/init.h> #define LINE_SIZE 80 diff --git a/arch/x86/kernel/cpu/mtrr/main.c b/arch/x86/kernel/cpu/mtrr/main.c index 84e83de5457..79556bd9b60 100644 --- a/arch/x86/kernel/cpu/mtrr/main.c +++ b/arch/x86/kernel/cpu/mtrr/main.c @@ -60,14 +60,14 @@ static DEFINE_MUTEX(mtrr_mutex); u64 size_or_mask, size_and_mask; static bool mtrr_aps_delayed_init; -static struct mtrr_ops *mtrr_ops[X86_VENDOR_NUM]; +static const struct mtrr_ops *mtrr_ops[X86_VENDOR_NUM]; -struct mtrr_ops *mtrr_if; +const struct mtrr_ops *mtrr_if; static void set_mtrr(unsigned int reg, unsigned long base, unsigned long size, mtrr_type type); -void set_mtrr_ops(struct mtrr_ops *ops) +void set_mtrr_ops(const struct mtrr_ops *ops) { if (ops->vendor && ops->vendor < X86_VENDOR_NUM) mtrr_ops[ops->vendor] = ops; @@ -145,6 +145,7 @@ struct set_mtrr_data { /** * ipi_handler - Synchronisation handler. Executed by "other" CPUs. + * @info: pointer to mtrr configuration data * * Returns nothing. */ diff --git a/arch/x86/kernel/cpu/mtrr/mtrr.h b/arch/x86/kernel/cpu/mtrr/mtrr.h index a501dee9a87..df5e41f31a2 100644 --- a/arch/x86/kernel/cpu/mtrr/mtrr.h +++ b/arch/x86/kernel/cpu/mtrr/mtrr.h @@ -32,7 +32,7 @@ extern int generic_get_free_region(unsigned long base, unsigned long size, extern int generic_validate_add_page(unsigned long base, unsigned long size, unsigned int type); -extern struct mtrr_ops generic_mtrr_ops; +extern const struct mtrr_ops generic_mtrr_ops; extern int positive_have_wrcomb(void); @@ -53,10 +53,10 @@ void fill_mtrr_var_range(unsigned int index, u32 base_lo, u32 base_hi, u32 mask_lo, u32 mask_hi); void get_mtrr_state(void); -extern void set_mtrr_ops(struct mtrr_ops *ops); +extern void set_mtrr_ops(const struct mtrr_ops *ops); extern u64 size_or_mask, size_and_mask; -extern struct mtrr_ops *mtrr_if; +extern const struct mtrr_ops *mtrr_if; #define is_cpu(vnd) (mtrr_if && mtrr_if->vendor == X86_VENDOR_##vnd) #define use_intel() (mtrr_if && mtrr_if->use_intel_if == 1) diff --git a/arch/x86/kernel/cpu/mtrr/state.c b/arch/x86/kernel/cpu/mtrr/state.c deleted file mode 100644 index dfc80b4e6b0..00000000000 --- a/arch/x86/kernel/cpu/mtrr/state.c +++ /dev/null @@ -1,94 +0,0 @@ -#include <linux/init.h> -#include <linux/io.h> -#include <linux/mm.h> - -#include <asm/processor-cyrix.h> -#include <asm/processor-flags.h> -#include <asm/mtrr.h> -#include <asm/msr.h> - -#include "mtrr.h" - -/* Put the processor into a state where MTRRs can be safely set */ -void set_mtrr_prepare_save(struct set_mtrr_context *ctxt) -{ - unsigned int cr0; - - /* Disable interrupts locally */ - local_irq_save(ctxt->flags); - - if (use_intel() || is_cpu(CYRIX)) { - - /* Save value of CR4 and clear Page Global Enable (bit 7) */ - if (cpu_has_pge) { - ctxt->cr4val = read_cr4(); - write_cr4(ctxt->cr4val & ~X86_CR4_PGE); - } - - /* - * Disable and flush caches. Note that wbinvd flushes the TLBs - * as a side-effect - */ - cr0 = read_cr0() | X86_CR0_CD; - wbinvd(); - write_cr0(cr0); - wbinvd(); - - if (use_intel()) { - /* Save MTRR state */ - rdmsr(MSR_MTRRdefType, ctxt->deftype_lo, ctxt->deftype_hi); - } else { - /* - * Cyrix ARRs - - * everything else were excluded at the top - */ - ctxt->ccr3 = getCx86(CX86_CCR3); - } - } -} - -void set_mtrr_cache_disable(struct set_mtrr_context *ctxt) -{ - if (use_intel()) { - /* Disable MTRRs, and set the default type to uncached */ - mtrr_wrmsr(MSR_MTRRdefType, ctxt->deftype_lo & 0xf300UL, - ctxt->deftype_hi); - } else { - if (is_cpu(CYRIX)) { - /* Cyrix ARRs - everything else were excluded at the top */ - setCx86(CX86_CCR3, (ctxt->ccr3 & 0x0f) | 0x10); - } - } -} - -/* Restore the processor after a set_mtrr_prepare */ -void set_mtrr_done(struct set_mtrr_context *ctxt) -{ - if (use_intel() || is_cpu(CYRIX)) { - - /* Flush caches and TLBs */ - wbinvd(); - - /* Restore MTRRdefType */ - if (use_intel()) { - /* Intel (P6) standard MTRRs */ - mtrr_wrmsr(MSR_MTRRdefType, ctxt->deftype_lo, - ctxt->deftype_hi); - } else { - /* - * Cyrix ARRs - - * everything else was excluded at the top - */ - setCx86(CX86_CCR3, ctxt->ccr3); - } - - /* Enable caches */ - write_cr0(read_cr0() & 0xbfffffff); - - /* Restore value of CR4 */ - if (cpu_has_pge) - write_cr4(ctxt->cr4val); - } - /* Re-enable interrupts locally (if enabled previously) */ - local_irq_restore(ctxt->flags); -} diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c index 8c1c07073cc..c77586061bc 100644 --- a/arch/x86/kernel/cpu/perf_event.c +++ b/arch/x86/kernel/cpu/perf_event.c @@ -7,6 +7,7 @@ * Copyright (C) 2009 Advanced Micro Devices, Inc., Robert Richter * Copyright (C) 2008-2009 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com> * Copyright (C) 2009 Intel Corporation, <markus.t.metzger@intel.com> + * Copyright (C) 2009 Google, Inc., Stephane Eranian * * For licencing details see kernel-base/COPYING */ @@ -20,215 +21,238 @@ #include <linux/kdebug.h> #include <linux/sched.h> #include <linux/uaccess.h> +#include <linux/slab.h> #include <linux/highmem.h> #include <linux/cpu.h> +#include <linux/bitops.h> #include <asm/apic.h> #include <asm/stacktrace.h> #include <asm/nmi.h> +#include <asm/compat.h> + +#if 0 +#undef wrmsrl +#define wrmsrl(msr, val) \ +do { \ + trace_printk("wrmsrl(%lx, %lx)\n", (unsigned long)(msr),\ + (unsigned long)(val)); \ + native_write_msr((msr), (u32)((u64)(val)), \ + (u32)((u64)(val) >> 32)); \ +} while (0) +#endif -static u64 perf_event_mask __read_mostly; +/* + * best effort, GUP based copy_from_user() that assumes IRQ or NMI context + */ +static unsigned long +copy_from_user_nmi(void *to, const void __user *from, unsigned long n) +{ + unsigned long offset, addr = (unsigned long)from; + int type = in_nmi() ? KM_NMI : KM_IRQ0; + unsigned long size, len = 0; + struct page *page; + void *map; + int ret; -/* The maximal number of PEBS events: */ -#define MAX_PEBS_EVENTS 4 + do { + ret = __get_user_pages_fast(addr, 1, 0, &page); + if (!ret) + break; -/* The size of a BTS record in bytes: */ -#define BTS_RECORD_SIZE 24 + offset = addr & (PAGE_SIZE - 1); + size = min(PAGE_SIZE - offset, n - len); -/* The size of a per-cpu BTS buffer in bytes: */ -#define BTS_BUFFER_SIZE (BTS_RECORD_SIZE * 2048) + map = kmap_atomic(page, type); + memcpy(to, map+offset, size); + kunmap_atomic(map, type); + put_page(page); -/* The BTS overflow threshold in bytes from the end of the buffer: */ -#define BTS_OVFL_TH (BTS_RECORD_SIZE * 128) + len += size; + to += size; + addr += size; + } while (len < n); -/* - * Bits in the debugctlmsr controlling branch tracing. - */ -#define X86_DEBUGCTL_TR (1 << 6) -#define X86_DEBUGCTL_BTS (1 << 7) -#define X86_DEBUGCTL_BTINT (1 << 8) -#define X86_DEBUGCTL_BTS_OFF_OS (1 << 9) -#define X86_DEBUGCTL_BTS_OFF_USR (1 << 10) + return len; +} -/* - * A debug store configuration. - * - * We only support architectures that use 64bit fields. - */ -struct debug_store { - u64 bts_buffer_base; - u64 bts_index; - u64 bts_absolute_maximum; - u64 bts_interrupt_threshold; - u64 pebs_buffer_base; - u64 pebs_index; - u64 pebs_absolute_maximum; - u64 pebs_interrupt_threshold; - u64 pebs_event_reset[MAX_PEBS_EVENTS]; +struct event_constraint { + union { + unsigned long idxmsk[BITS_TO_LONGS(X86_PMC_IDX_MAX)]; + u64 idxmsk64; + }; + u64 code; + u64 cmask; + int weight; +}; + +struct amd_nb { + int nb_id; /* NorthBridge id */ + int refcnt; /* reference count */ + struct perf_event *owners[X86_PMC_IDX_MAX]; + struct event_constraint event_constraints[X86_PMC_IDX_MAX]; }; +#define MAX_LBR_ENTRIES 16 + struct cpu_hw_events { - struct perf_event *events[X86_PMC_IDX_MAX]; - unsigned long used_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)]; + /* + * Generic x86 PMC bits + */ + struct perf_event *events[X86_PMC_IDX_MAX]; /* in counter order */ unsigned long active_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)]; - unsigned long interrupts; int enabled; + + int n_events; + int n_added; + int assign[X86_PMC_IDX_MAX]; /* event to counter assignment */ + u64 tags[X86_PMC_IDX_MAX]; + struct perf_event *event_list[X86_PMC_IDX_MAX]; /* in enabled order */ + + unsigned int group_flag; + + /* + * Intel DebugStore bits + */ struct debug_store *ds; -}; + u64 pebs_enabled; -struct event_constraint { - unsigned long idxmsk[BITS_TO_LONGS(X86_PMC_IDX_MAX)]; - int code; + /* + * Intel LBR bits + */ + int lbr_users; + void *lbr_context; + struct perf_branch_stack lbr_stack; + struct perf_branch_entry lbr_entries[MAX_LBR_ENTRIES]; + + /* + * AMD specific bits + */ + struct amd_nb *amd_nb; }; -#define EVENT_CONSTRAINT(c, m) { .code = (c), .idxmsk[0] = (m) } -#define EVENT_CONSTRAINT_END { .code = 0, .idxmsk[0] = 0 } +#define __EVENT_CONSTRAINT(c, n, m, w) {\ + { .idxmsk64 = (n) }, \ + .code = (c), \ + .cmask = (m), \ + .weight = (w), \ +} + +#define EVENT_CONSTRAINT(c, n, m) \ + __EVENT_CONSTRAINT(c, n, m, HWEIGHT(n)) -#define for_each_event_constraint(e, c) \ - for ((e) = (c); (e)->idxmsk[0]; (e)++) +/* + * Constraint on the Event code. + */ +#define INTEL_EVENT_CONSTRAINT(c, n) \ + EVENT_CONSTRAINT(c, n, ARCH_PERFMON_EVENTSEL_EVENT) +/* + * Constraint on the Event code + UMask + fixed-mask + * + * filter mask to validate fixed counter events. + * the following filters disqualify for fixed counters: + * - inv + * - edge + * - cnt-mask + * The other filters are supported by fixed counters. + * The any-thread option is supported starting with v3. + */ +#define FIXED_EVENT_CONSTRAINT(c, n) \ + EVENT_CONSTRAINT(c, (1ULL << (32+n)), X86_RAW_EVENT_MASK) + +/* + * Constraint on the Event code + UMask + */ +#define PEBS_EVENT_CONSTRAINT(c, n) \ + EVENT_CONSTRAINT(c, n, INTEL_ARCH_EVENT_MASK) + +#define EVENT_CONSTRAINT_END \ + EVENT_CONSTRAINT(0, 0, 0) + +#define for_each_event_constraint(e, c) \ + for ((e) = (c); (e)->weight; (e)++) + +union perf_capabilities { + struct { + u64 lbr_format : 6; + u64 pebs_trap : 1; + u64 pebs_arch_reg : 1; + u64 pebs_format : 4; + u64 smm_freeze : 1; + }; + u64 capabilities; +}; /* * struct x86_pmu - generic x86 pmu */ struct x86_pmu { + /* + * Generic x86 PMC bits + */ const char *name; int version; int (*handle_irq)(struct pt_regs *); void (*disable_all)(void); - void (*enable_all)(void); - void (*enable)(struct hw_perf_event *, int); - void (*disable)(struct hw_perf_event *, int); + void (*enable_all)(int added); + void (*enable)(struct perf_event *); + void (*disable)(struct perf_event *); + int (*hw_config)(struct perf_event *event); + int (*schedule_events)(struct cpu_hw_events *cpuc, int n, int *assign); unsigned eventsel; unsigned perfctr; u64 (*event_map)(int); - u64 (*raw_event)(u64); int max_events; - int num_events; - int num_events_fixed; - int event_bits; - u64 event_mask; + int num_counters; + int num_counters_fixed; + int cntval_bits; + u64 cntval_mask; int apic; u64 max_period; - u64 intel_ctrl; - void (*enable_bts)(u64 config); - void (*disable_bts)(void); - int (*get_event_idx)(struct cpu_hw_events *cpuc, - struct hw_perf_event *hwc); -}; - -static struct x86_pmu x86_pmu __read_mostly; + struct event_constraint * + (*get_event_constraints)(struct cpu_hw_events *cpuc, + struct perf_event *event); -static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = { - .enabled = 1, -}; - -static const struct event_constraint *event_constraints; - -/* - * Not sure about some of these - */ -static const u64 p6_perfmon_event_map[] = -{ - [PERF_COUNT_HW_CPU_CYCLES] = 0x0079, - [PERF_COUNT_HW_INSTRUCTIONS] = 0x00c0, - [PERF_COUNT_HW_CACHE_REFERENCES] = 0x0f2e, - [PERF_COUNT_HW_CACHE_MISSES] = 0x012e, - [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = 0x00c4, - [PERF_COUNT_HW_BRANCH_MISSES] = 0x00c5, - [PERF_COUNT_HW_BUS_CYCLES] = 0x0062, -}; + void (*put_event_constraints)(struct cpu_hw_events *cpuc, + struct perf_event *event); + struct event_constraint *event_constraints; + void (*quirks)(void); -static u64 p6_pmu_event_map(int hw_event) -{ - return p6_perfmon_event_map[hw_event]; -} + int (*cpu_prepare)(int cpu); + void (*cpu_starting)(int cpu); + void (*cpu_dying)(int cpu); + void (*cpu_dead)(int cpu); -/* - * Event setting that is specified not to count anything. - * We use this to effectively disable a counter. - * - * L2_RQSTS with 0 MESI unit mask. - */ -#define P6_NOP_EVENT 0x0000002EULL + /* + * Intel Arch Perfmon v2+ + */ + u64 intel_ctrl; + union perf_capabilities intel_cap; -static u64 p6_pmu_raw_event(u64 hw_event) -{ -#define P6_EVNTSEL_EVENT_MASK 0x000000FFULL -#define P6_EVNTSEL_UNIT_MASK 0x0000FF00ULL -#define P6_EVNTSEL_EDGE_MASK 0x00040000ULL -#define P6_EVNTSEL_INV_MASK 0x00800000ULL -#define P6_EVNTSEL_REG_MASK 0xFF000000ULL - -#define P6_EVNTSEL_MASK \ - (P6_EVNTSEL_EVENT_MASK | \ - P6_EVNTSEL_UNIT_MASK | \ - P6_EVNTSEL_EDGE_MASK | \ - P6_EVNTSEL_INV_MASK | \ - P6_EVNTSEL_REG_MASK) - - return hw_event & P6_EVNTSEL_MASK; -} + /* + * Intel DebugStore bits + */ + int bts, pebs; + int pebs_record_size; + void (*drain_pebs)(struct pt_regs *regs); + struct event_constraint *pebs_constraints; -static const struct event_constraint intel_p6_event_constraints[] = -{ - EVENT_CONSTRAINT(0xc1, 0x1), /* FLOPS */ - EVENT_CONSTRAINT(0x10, 0x1), /* FP_COMP_OPS_EXE */ - EVENT_CONSTRAINT(0x11, 0x1), /* FP_ASSIST */ - EVENT_CONSTRAINT(0x12, 0x2), /* MUL */ - EVENT_CONSTRAINT(0x13, 0x2), /* DIV */ - EVENT_CONSTRAINT(0x14, 0x1), /* CYCLES_DIV_BUSY */ - EVENT_CONSTRAINT_END + /* + * Intel LBR + */ + unsigned long lbr_tos, lbr_from, lbr_to; /* MSR base regs */ + int lbr_nr; /* hardware stack size */ }; -/* - * Intel PerfMon v3. Used on Core2 and later. - */ -static const u64 intel_perfmon_event_map[] = -{ - [PERF_COUNT_HW_CPU_CYCLES] = 0x003c, - [PERF_COUNT_HW_INSTRUCTIONS] = 0x00c0, - [PERF_COUNT_HW_CACHE_REFERENCES] = 0x4f2e, - [PERF_COUNT_HW_CACHE_MISSES] = 0x412e, - [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = 0x00c4, - [PERF_COUNT_HW_BRANCH_MISSES] = 0x00c5, - [PERF_COUNT_HW_BUS_CYCLES] = 0x013c, -}; +static struct x86_pmu x86_pmu __read_mostly; -static const struct event_constraint intel_core_event_constraints[] = -{ - EVENT_CONSTRAINT(0x10, 0x1), /* FP_COMP_OPS_EXE */ - EVENT_CONSTRAINT(0x11, 0x2), /* FP_ASSIST */ - EVENT_CONSTRAINT(0x12, 0x2), /* MUL */ - EVENT_CONSTRAINT(0x13, 0x2), /* DIV */ - EVENT_CONSTRAINT(0x14, 0x1), /* CYCLES_DIV_BUSY */ - EVENT_CONSTRAINT(0x18, 0x1), /* IDLE_DURING_DIV */ - EVENT_CONSTRAINT(0x19, 0x2), /* DELAYED_BYPASS */ - EVENT_CONSTRAINT(0xa1, 0x1), /* RS_UOPS_DISPATCH_CYCLES */ - EVENT_CONSTRAINT(0xcb, 0x1), /* MEM_LOAD_RETIRED */ - EVENT_CONSTRAINT_END +static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = { + .enabled = 1, }; -static const struct event_constraint intel_nehalem_event_constraints[] = -{ - EVENT_CONSTRAINT(0x40, 0x3), /* L1D_CACHE_LD */ - EVENT_CONSTRAINT(0x41, 0x3), /* L1D_CACHE_ST */ - EVENT_CONSTRAINT(0x42, 0x3), /* L1D_CACHE_LOCK */ - EVENT_CONSTRAINT(0x43, 0x3), /* L1D_ALL_REF */ - EVENT_CONSTRAINT(0x4e, 0x3), /* L1D_PREFETCH */ - EVENT_CONSTRAINT(0x4c, 0x3), /* LOAD_HIT_PRE */ - EVENT_CONSTRAINT(0x51, 0x3), /* L1D */ - EVENT_CONSTRAINT(0x52, 0x3), /* L1D_CACHE_PREFETCH_LOCK_FB_HIT */ - EVENT_CONSTRAINT(0x53, 0x3), /* L1D_CACHE_LOCK_FB_HIT */ - EVENT_CONSTRAINT(0xc5, 0x3), /* CACHE_LOCK_CYCLES */ - EVENT_CONSTRAINT_END -}; - -static u64 intel_pmu_event_map(int hw_event) -{ - return intel_perfmon_event_map[hw_event]; -} +static int x86_perf_event_set_period(struct perf_event *event); /* * Generalized hw caching related hw_event table, filled @@ -245,435 +269,18 @@ static u64 __read_mostly hw_cache_event_ids [PERF_COUNT_HW_CACHE_OP_MAX] [PERF_COUNT_HW_CACHE_RESULT_MAX]; -static __initconst u64 nehalem_hw_cache_event_ids - [PERF_COUNT_HW_CACHE_MAX] - [PERF_COUNT_HW_CACHE_OP_MAX] - [PERF_COUNT_HW_CACHE_RESULT_MAX] = -{ - [ C(L1D) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x0f40, /* L1D_CACHE_LD.MESI */ - [ C(RESULT_MISS) ] = 0x0140, /* L1D_CACHE_LD.I_STATE */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = 0x0f41, /* L1D_CACHE_ST.MESI */ - [ C(RESULT_MISS) ] = 0x0141, /* L1D_CACHE_ST.I_STATE */ - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0x014e, /* L1D_PREFETCH.REQUESTS */ - [ C(RESULT_MISS) ] = 0x024e, /* L1D_PREFETCH.MISS */ - }, - }, - [ C(L1I ) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x0380, /* L1I.READS */ - [ C(RESULT_MISS) ] = 0x0280, /* L1I.MISSES */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0x0, - [ C(RESULT_MISS) ] = 0x0, - }, - }, - [ C(LL ) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x0324, /* L2_RQSTS.LOADS */ - [ C(RESULT_MISS) ] = 0x0224, /* L2_RQSTS.LD_MISS */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = 0x0c24, /* L2_RQSTS.RFOS */ - [ C(RESULT_MISS) ] = 0x0824, /* L2_RQSTS.RFO_MISS */ - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0x4f2e, /* LLC Reference */ - [ C(RESULT_MISS) ] = 0x412e, /* LLC Misses */ - }, - }, - [ C(DTLB) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x0f40, /* L1D_CACHE_LD.MESI (alias) */ - [ C(RESULT_MISS) ] = 0x0108, /* DTLB_LOAD_MISSES.ANY */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = 0x0f41, /* L1D_CACHE_ST.MESI (alias) */ - [ C(RESULT_MISS) ] = 0x010c, /* MEM_STORE_RETIRED.DTLB_MISS */ - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0x0, - [ C(RESULT_MISS) ] = 0x0, - }, - }, - [ C(ITLB) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x01c0, /* INST_RETIRED.ANY_P */ - [ C(RESULT_MISS) ] = 0x20c8, /* ITLB_MISS_RETIRED */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - }, - [ C(BPU ) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x00c4, /* BR_INST_RETIRED.ALL_BRANCHES */ - [ C(RESULT_MISS) ] = 0x03e8, /* BPU_CLEARS.ANY */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - }, -}; - -static __initconst u64 core2_hw_cache_event_ids - [PERF_COUNT_HW_CACHE_MAX] - [PERF_COUNT_HW_CACHE_OP_MAX] - [PERF_COUNT_HW_CACHE_RESULT_MAX] = -{ - [ C(L1D) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x0f40, /* L1D_CACHE_LD.MESI */ - [ C(RESULT_MISS) ] = 0x0140, /* L1D_CACHE_LD.I_STATE */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = 0x0f41, /* L1D_CACHE_ST.MESI */ - [ C(RESULT_MISS) ] = 0x0141, /* L1D_CACHE_ST.I_STATE */ - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0x104e, /* L1D_PREFETCH.REQUESTS */ - [ C(RESULT_MISS) ] = 0, - }, - }, - [ C(L1I ) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x0080, /* L1I.READS */ - [ C(RESULT_MISS) ] = 0x0081, /* L1I.MISSES */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0, - [ C(RESULT_MISS) ] = 0, - }, - }, - [ C(LL ) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x4f29, /* L2_LD.MESI */ - [ C(RESULT_MISS) ] = 0x4129, /* L2_LD.ISTATE */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = 0x4f2A, /* L2_ST.MESI */ - [ C(RESULT_MISS) ] = 0x412A, /* L2_ST.ISTATE */ - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0, - [ C(RESULT_MISS) ] = 0, - }, - }, - [ C(DTLB) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x0f40, /* L1D_CACHE_LD.MESI (alias) */ - [ C(RESULT_MISS) ] = 0x0208, /* DTLB_MISSES.MISS_LD */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = 0x0f41, /* L1D_CACHE_ST.MESI (alias) */ - [ C(RESULT_MISS) ] = 0x0808, /* DTLB_MISSES.MISS_ST */ - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0, - [ C(RESULT_MISS) ] = 0, - }, - }, - [ C(ITLB) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x00c0, /* INST_RETIRED.ANY_P */ - [ C(RESULT_MISS) ] = 0x1282, /* ITLBMISSES */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - }, - [ C(BPU ) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x00c4, /* BR_INST_RETIRED.ANY */ - [ C(RESULT_MISS) ] = 0x00c5, /* BP_INST_RETIRED.MISPRED */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - }, -}; - -static __initconst u64 atom_hw_cache_event_ids - [PERF_COUNT_HW_CACHE_MAX] - [PERF_COUNT_HW_CACHE_OP_MAX] - [PERF_COUNT_HW_CACHE_RESULT_MAX] = -{ - [ C(L1D) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x2140, /* L1D_CACHE.LD */ - [ C(RESULT_MISS) ] = 0, - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = 0x2240, /* L1D_CACHE.ST */ - [ C(RESULT_MISS) ] = 0, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0x0, - [ C(RESULT_MISS) ] = 0, - }, - }, - [ C(L1I ) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x0380, /* L1I.READS */ - [ C(RESULT_MISS) ] = 0x0280, /* L1I.MISSES */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0, - [ C(RESULT_MISS) ] = 0, - }, - }, - [ C(LL ) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x4f29, /* L2_LD.MESI */ - [ C(RESULT_MISS) ] = 0x4129, /* L2_LD.ISTATE */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = 0x4f2A, /* L2_ST.MESI */ - [ C(RESULT_MISS) ] = 0x412A, /* L2_ST.ISTATE */ - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0, - [ C(RESULT_MISS) ] = 0, - }, - }, - [ C(DTLB) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x2140, /* L1D_CACHE_LD.MESI (alias) */ - [ C(RESULT_MISS) ] = 0x0508, /* DTLB_MISSES.MISS_LD */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = 0x2240, /* L1D_CACHE_ST.MESI (alias) */ - [ C(RESULT_MISS) ] = 0x0608, /* DTLB_MISSES.MISS_ST */ - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0, - [ C(RESULT_MISS) ] = 0, - }, - }, - [ C(ITLB) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x00c0, /* INST_RETIRED.ANY_P */ - [ C(RESULT_MISS) ] = 0x0282, /* ITLB.MISSES */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - }, - [ C(BPU ) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x00c4, /* BR_INST_RETIRED.ANY */ - [ C(RESULT_MISS) ] = 0x00c5, /* BP_INST_RETIRED.MISPRED */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - }, -}; - -static u64 intel_pmu_raw_event(u64 hw_event) -{ -#define CORE_EVNTSEL_EVENT_MASK 0x000000FFULL -#define CORE_EVNTSEL_UNIT_MASK 0x0000FF00ULL -#define CORE_EVNTSEL_EDGE_MASK 0x00040000ULL -#define CORE_EVNTSEL_INV_MASK 0x00800000ULL -#define CORE_EVNTSEL_REG_MASK 0xFF000000ULL - -#define CORE_EVNTSEL_MASK \ - (CORE_EVNTSEL_EVENT_MASK | \ - CORE_EVNTSEL_UNIT_MASK | \ - CORE_EVNTSEL_EDGE_MASK | \ - CORE_EVNTSEL_INV_MASK | \ - CORE_EVNTSEL_REG_MASK) - - return hw_event & CORE_EVNTSEL_MASK; -} - -static __initconst u64 amd_hw_cache_event_ids - [PERF_COUNT_HW_CACHE_MAX] - [PERF_COUNT_HW_CACHE_OP_MAX] - [PERF_COUNT_HW_CACHE_RESULT_MAX] = -{ - [ C(L1D) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x0040, /* Data Cache Accesses */ - [ C(RESULT_MISS) ] = 0x0041, /* Data Cache Misses */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = 0x0142, /* Data Cache Refills :system */ - [ C(RESULT_MISS) ] = 0, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0x0267, /* Data Prefetcher :attempts */ - [ C(RESULT_MISS) ] = 0x0167, /* Data Prefetcher :cancelled */ - }, - }, - [ C(L1I ) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x0080, /* Instruction cache fetches */ - [ C(RESULT_MISS) ] = 0x0081, /* Instruction cache misses */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0x014B, /* Prefetch Instructions :Load */ - [ C(RESULT_MISS) ] = 0, - }, - }, - [ C(LL ) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x037D, /* Requests to L2 Cache :IC+DC */ - [ C(RESULT_MISS) ] = 0x037E, /* L2 Cache Misses : IC+DC */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = 0x017F, /* L2 Fill/Writeback */ - [ C(RESULT_MISS) ] = 0, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0, - [ C(RESULT_MISS) ] = 0, - }, - }, - [ C(DTLB) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x0040, /* Data Cache Accesses */ - [ C(RESULT_MISS) ] = 0x0046, /* L1 DTLB and L2 DLTB Miss */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = 0, - [ C(RESULT_MISS) ] = 0, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = 0, - [ C(RESULT_MISS) ] = 0, - }, - }, - [ C(ITLB) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x0080, /* Instruction fecthes */ - [ C(RESULT_MISS) ] = 0x0085, /* Instr. fetch ITLB misses */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - }, - [ C(BPU ) ] = { - [ C(OP_READ) ] = { - [ C(RESULT_ACCESS) ] = 0x00c2, /* Retired Branch Instr. */ - [ C(RESULT_MISS) ] = 0x00c3, /* Retired Mispredicted BI */ - }, - [ C(OP_WRITE) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - [ C(OP_PREFETCH) ] = { - [ C(RESULT_ACCESS) ] = -1, - [ C(RESULT_MISS) ] = -1, - }, - }, -}; - -/* - * AMD Performance Monitor K7 and later. - */ -static const u64 amd_perfmon_event_map[] = -{ - [PERF_COUNT_HW_CPU_CYCLES] = 0x0076, - [PERF_COUNT_HW_INSTRUCTIONS] = 0x00c0, - [PERF_COUNT_HW_CACHE_REFERENCES] = 0x0080, - [PERF_COUNT_HW_CACHE_MISSES] = 0x0081, - [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = 0x00c4, - [PERF_COUNT_HW_BRANCH_MISSES] = 0x00c5, -}; - -static u64 amd_pmu_event_map(int hw_event) -{ - return amd_perfmon_event_map[hw_event]; -} - -static u64 amd_pmu_raw_event(u64 hw_event) -{ -#define K7_EVNTSEL_EVENT_MASK 0x7000000FFULL -#define K7_EVNTSEL_UNIT_MASK 0x00000FF00ULL -#define K7_EVNTSEL_EDGE_MASK 0x000040000ULL -#define K7_EVNTSEL_INV_MASK 0x000800000ULL -#define K7_EVNTSEL_REG_MASK 0x0FF000000ULL - -#define K7_EVNTSEL_MASK \ - (K7_EVNTSEL_EVENT_MASK | \ - K7_EVNTSEL_UNIT_MASK | \ - K7_EVNTSEL_EDGE_MASK | \ - K7_EVNTSEL_INV_MASK | \ - K7_EVNTSEL_REG_MASK) - - return hw_event & K7_EVNTSEL_MASK; -} - /* * Propagate event elapsed time into the generic event. * Can only be executed on the CPU where the event is active. * Returns the delta events processed. */ static u64 -x86_perf_event_update(struct perf_event *event, - struct hw_perf_event *hwc, int idx) +x86_perf_event_update(struct perf_event *event) { - int shift = 64 - x86_pmu.event_bits; + struct hw_perf_event *hwc = &event->hw; + int shift = 64 - x86_pmu.cntval_bits; u64 prev_raw_count, new_raw_count; + int idx = hwc->idx; s64 delta; if (idx == X86_PMC_IDX_FIXED_BTS) @@ -714,33 +321,32 @@ again: static atomic_t active_events; static DEFINE_MUTEX(pmc_reserve_mutex); +#ifdef CONFIG_X86_LOCAL_APIC + static bool reserve_pmc_hardware(void) { -#ifdef CONFIG_X86_LOCAL_APIC int i; if (nmi_watchdog == NMI_LOCAL_APIC) disable_lapic_nmi_watchdog(); - for (i = 0; i < x86_pmu.num_events; i++) { + for (i = 0; i < x86_pmu.num_counters; i++) { if (!reserve_perfctr_nmi(x86_pmu.perfctr + i)) goto perfctr_fail; } - for (i = 0; i < x86_pmu.num_events; i++) { + for (i = 0; i < x86_pmu.num_counters; i++) { if (!reserve_evntsel_nmi(x86_pmu.eventsel + i)) goto eventsel_fail; } -#endif return true; -#ifdef CONFIG_X86_LOCAL_APIC eventsel_fail: for (i--; i >= 0; i--) release_evntsel_nmi(x86_pmu.eventsel + i); - i = x86_pmu.num_events; + i = x86_pmu.num_counters; perfctr_fail: for (i--; i >= 0; i--) @@ -750,128 +356,36 @@ perfctr_fail: enable_lapic_nmi_watchdog(); return false; -#endif } static void release_pmc_hardware(void) { -#ifdef CONFIG_X86_LOCAL_APIC int i; - for (i = 0; i < x86_pmu.num_events; i++) { + for (i = 0; i < x86_pmu.num_counters; i++) { release_perfctr_nmi(x86_pmu.perfctr + i); release_evntsel_nmi(x86_pmu.eventsel + i); } if (nmi_watchdog == NMI_LOCAL_APIC) enable_lapic_nmi_watchdog(); -#endif -} - -static inline bool bts_available(void) -{ - return x86_pmu.enable_bts != NULL; -} - -static inline void init_debug_store_on_cpu(int cpu) -{ - struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds; - - if (!ds) - return; - - wrmsr_on_cpu(cpu, MSR_IA32_DS_AREA, - (u32)((u64)(unsigned long)ds), - (u32)((u64)(unsigned long)ds >> 32)); -} - -static inline void fini_debug_store_on_cpu(int cpu) -{ - if (!per_cpu(cpu_hw_events, cpu).ds) - return; - - wrmsr_on_cpu(cpu, MSR_IA32_DS_AREA, 0, 0); -} - -static void release_bts_hardware(void) -{ - int cpu; - - if (!bts_available()) - return; - - get_online_cpus(); - - for_each_online_cpu(cpu) - fini_debug_store_on_cpu(cpu); - - for_each_possible_cpu(cpu) { - struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds; - - if (!ds) - continue; - - per_cpu(cpu_hw_events, cpu).ds = NULL; - - kfree((void *)(unsigned long)ds->bts_buffer_base); - kfree(ds); - } - - put_online_cpus(); } -static int reserve_bts_hardware(void) -{ - int cpu, err = 0; - - if (!bts_available()) - return 0; - - get_online_cpus(); - - for_each_possible_cpu(cpu) { - struct debug_store *ds; - void *buffer; +#else - err = -ENOMEM; - buffer = kzalloc(BTS_BUFFER_SIZE, GFP_KERNEL); - if (unlikely(!buffer)) - break; - - ds = kzalloc(sizeof(*ds), GFP_KERNEL); - if (unlikely(!ds)) { - kfree(buffer); - break; - } - - ds->bts_buffer_base = (u64)(unsigned long)buffer; - ds->bts_index = ds->bts_buffer_base; - ds->bts_absolute_maximum = - ds->bts_buffer_base + BTS_BUFFER_SIZE; - ds->bts_interrupt_threshold = - ds->bts_absolute_maximum - BTS_OVFL_TH; - - per_cpu(cpu_hw_events, cpu).ds = ds; - err = 0; - } +static bool reserve_pmc_hardware(void) { return true; } +static void release_pmc_hardware(void) {} - if (err) - release_bts_hardware(); - else { - for_each_online_cpu(cpu) - init_debug_store_on_cpu(cpu); - } - - put_online_cpus(); +#endif - return err; -} +static int reserve_ds_buffers(void); +static void release_ds_buffers(void); static void hw_perf_event_destroy(struct perf_event *event) { if (atomic_dec_and_mutex_lock(&active_events, &pmc_reserve_mutex)) { release_pmc_hardware(); - release_bts_hardware(); + release_ds_buffers(); mutex_unlock(&pmc_reserve_mutex); } } @@ -914,88 +428,11 @@ set_ext_hw_attr(struct hw_perf_event *hwc, struct perf_event_attr *attr) return 0; } -static void intel_pmu_enable_bts(u64 config) -{ - unsigned long debugctlmsr; - - debugctlmsr = get_debugctlmsr(); - - debugctlmsr |= X86_DEBUGCTL_TR; - debugctlmsr |= X86_DEBUGCTL_BTS; - debugctlmsr |= X86_DEBUGCTL_BTINT; - - if (!(config & ARCH_PERFMON_EVENTSEL_OS)) - debugctlmsr |= X86_DEBUGCTL_BTS_OFF_OS; - - if (!(config & ARCH_PERFMON_EVENTSEL_USR)) - debugctlmsr |= X86_DEBUGCTL_BTS_OFF_USR; - - update_debugctlmsr(debugctlmsr); -} - -static void intel_pmu_disable_bts(void) -{ - struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); - unsigned long debugctlmsr; - - if (!cpuc->ds) - return; - - debugctlmsr = get_debugctlmsr(); - - debugctlmsr &= - ~(X86_DEBUGCTL_TR | X86_DEBUGCTL_BTS | X86_DEBUGCTL_BTINT | - X86_DEBUGCTL_BTS_OFF_OS | X86_DEBUGCTL_BTS_OFF_USR); - - update_debugctlmsr(debugctlmsr); -} - -/* - * Setup the hardware configuration for a given attr_type - */ -static int __hw_perf_event_init(struct perf_event *event) +static int x86_setup_perfctr(struct perf_event *event) { struct perf_event_attr *attr = &event->attr; struct hw_perf_event *hwc = &event->hw; u64 config; - int err; - - if (!x86_pmu_initialized()) - return -ENODEV; - - err = 0; - if (!atomic_inc_not_zero(&active_events)) { - mutex_lock(&pmc_reserve_mutex); - if (atomic_read(&active_events) == 0) { - if (!reserve_pmc_hardware()) - err = -EBUSY; - else - err = reserve_bts_hardware(); - } - if (!err) - atomic_inc(&active_events); - mutex_unlock(&pmc_reserve_mutex); - } - if (err) - return err; - - event->destroy = hw_perf_event_destroy; - - /* - * Generate PMC IRQs: - * (keep 'enabled' bit clear for now) - */ - hwc->config = ARCH_PERFMON_EVENTSEL_INT; - - hwc->idx = -1; - - /* - * Count user and OS events unless requested not to. - */ - if (!attr->exclude_user) - hwc->config |= ARCH_PERFMON_EVENTSEL_USR; - if (!attr->exclude_kernel) - hwc->config |= ARCH_PERFMON_EVENTSEL_OS; if (!hwc->sample_period) { hwc->sample_period = x86_pmu.max_period; @@ -1012,13 +449,8 @@ static int __hw_perf_event_init(struct perf_event *event) return -EOPNOTSUPP; } - /* - * Raw hw_event type provide the config in the hw_event structure - */ - if (attr->type == PERF_TYPE_RAW) { - hwc->config |= x86_pmu.raw_event(attr->config); + if (attr->type == PERF_TYPE_RAW) return 0; - } if (attr->type == PERF_TYPE_HW_CACHE) return set_ext_hw_attr(hwc, attr); @@ -1043,11 +475,11 @@ static int __hw_perf_event_init(struct perf_event *event) if ((attr->config == PERF_COUNT_HW_BRANCH_INSTRUCTIONS) && (hwc->sample_period == 1)) { /* BTS is not supported by this architecture. */ - if (!bts_available()) + if (!x86_pmu.bts) return -EOPNOTSUPP; /* BTS is currently only allowed for user-mode. */ - if (hwc->config & ARCH_PERFMON_EVENTSEL_OS) + if (!attr->exclude_kernel) return -EOPNOTSUPP; } @@ -1056,127 +488,122 @@ static int __hw_perf_event_init(struct perf_event *event) return 0; } -static void p6_pmu_disable_all(void) +static int x86_pmu_hw_config(struct perf_event *event) { - struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); - u64 val; + if (event->attr.precise_ip) { + int precise = 0; - if (!cpuc->enabled) - return; + /* Support for constant skid */ + if (x86_pmu.pebs) + precise++; - cpuc->enabled = 0; - barrier(); + /* Support for IP fixup */ + if (x86_pmu.lbr_nr) + precise++; + + if (event->attr.precise_ip > precise) + return -EOPNOTSUPP; + } + + /* + * Generate PMC IRQs: + * (keep 'enabled' bit clear for now) + */ + event->hw.config = ARCH_PERFMON_EVENTSEL_INT; + + /* + * Count user and OS events unless requested not to + */ + if (!event->attr.exclude_user) + event->hw.config |= ARCH_PERFMON_EVENTSEL_USR; + if (!event->attr.exclude_kernel) + event->hw.config |= ARCH_PERFMON_EVENTSEL_OS; - /* p6 only has one enable register */ - rdmsrl(MSR_P6_EVNTSEL0, val); - val &= ~ARCH_PERFMON_EVENTSEL0_ENABLE; - wrmsrl(MSR_P6_EVNTSEL0, val); + if (event->attr.type == PERF_TYPE_RAW) + event->hw.config |= event->attr.config & X86_RAW_EVENT_MASK; + + return x86_setup_perfctr(event); } -static void intel_pmu_disable_all(void) +/* + * Setup the hardware configuration for a given attr_type + */ +static int __hw_perf_event_init(struct perf_event *event) { - struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + int err; - if (!cpuc->enabled) - return; + if (!x86_pmu_initialized()) + return -ENODEV; - cpuc->enabled = 0; - barrier(); + err = 0; + if (!atomic_inc_not_zero(&active_events)) { + mutex_lock(&pmc_reserve_mutex); + if (atomic_read(&active_events) == 0) { + if (!reserve_pmc_hardware()) + err = -EBUSY; + else { + err = reserve_ds_buffers(); + if (err) + release_pmc_hardware(); + } + } + if (!err) + atomic_inc(&active_events); + mutex_unlock(&pmc_reserve_mutex); + } + if (err) + return err; + + event->destroy = hw_perf_event_destroy; - wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0); + event->hw.idx = -1; + event->hw.last_cpu = -1; + event->hw.last_tag = ~0ULL; - if (test_bit(X86_PMC_IDX_FIXED_BTS, cpuc->active_mask)) - intel_pmu_disable_bts(); + return x86_pmu.hw_config(event); } -static void amd_pmu_disable_all(void) +static void x86_pmu_disable_all(void) { struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); int idx; - if (!cpuc->enabled) - return; - - cpuc->enabled = 0; - /* - * ensure we write the disable before we start disabling the - * events proper, so that amd_pmu_enable_event() does the - * right thing. - */ - barrier(); - - for (idx = 0; idx < x86_pmu.num_events; idx++) { + for (idx = 0; idx < x86_pmu.num_counters; idx++) { u64 val; if (!test_bit(idx, cpuc->active_mask)) continue; - rdmsrl(MSR_K7_EVNTSEL0 + idx, val); - if (!(val & ARCH_PERFMON_EVENTSEL0_ENABLE)) + rdmsrl(x86_pmu.eventsel + idx, val); + if (!(val & ARCH_PERFMON_EVENTSEL_ENABLE)) continue; - val &= ~ARCH_PERFMON_EVENTSEL0_ENABLE; - wrmsrl(MSR_K7_EVNTSEL0 + idx, val); + val &= ~ARCH_PERFMON_EVENTSEL_ENABLE; + wrmsrl(x86_pmu.eventsel + idx, val); } } void hw_perf_disable(void) { - if (!x86_pmu_initialized()) - return; - return x86_pmu.disable_all(); -} - -static void p6_pmu_enable_all(void) -{ struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); - unsigned long val; - if (cpuc->enabled) + if (!x86_pmu_initialized()) return; - cpuc->enabled = 1; - barrier(); - - /* p6 only has one enable register */ - rdmsrl(MSR_P6_EVNTSEL0, val); - val |= ARCH_PERFMON_EVENTSEL0_ENABLE; - wrmsrl(MSR_P6_EVNTSEL0, val); -} - -static void intel_pmu_enable_all(void) -{ - struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); - - if (cpuc->enabled) + if (!cpuc->enabled) return; - cpuc->enabled = 1; + cpuc->n_added = 0; + cpuc->enabled = 0; barrier(); - wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, x86_pmu.intel_ctrl); - - if (test_bit(X86_PMC_IDX_FIXED_BTS, cpuc->active_mask)) { - struct perf_event *event = - cpuc->events[X86_PMC_IDX_FIXED_BTS]; - - if (WARN_ON_ONCE(!event)) - return; - - intel_pmu_enable_bts(event->hw.config); - } + x86_pmu.disable_all(); } -static void amd_pmu_enable_all(void) +static void x86_pmu_enable_all(int added) { struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); int idx; - if (cpuc->enabled) - return; - - cpuc->enabled = 1; - barrier(); - - for (idx = 0; idx < x86_pmu.num_events; idx++) { + for (idx = 0; idx < x86_pmu.num_counters; idx++) { struct perf_event *event = cpuc->events[idx]; u64 val; @@ -1184,88 +611,267 @@ static void amd_pmu_enable_all(void) continue; val = event->hw.config; - val |= ARCH_PERFMON_EVENTSEL0_ENABLE; - wrmsrl(MSR_K7_EVNTSEL0 + idx, val); + val |= ARCH_PERFMON_EVENTSEL_ENABLE; + wrmsrl(x86_pmu.eventsel + idx, val); } } -void hw_perf_enable(void) +static const struct pmu pmu; + +static inline int is_x86_event(struct perf_event *event) { - if (!x86_pmu_initialized()) - return; - x86_pmu.enable_all(); + return event->pmu == &pmu; } -static inline u64 intel_pmu_get_status(void) +static int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign) { - u64 status; + struct event_constraint *c, *constraints[X86_PMC_IDX_MAX]; + unsigned long used_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)]; + int i, j, w, wmax, num = 0; + struct hw_perf_event *hwc; - rdmsrl(MSR_CORE_PERF_GLOBAL_STATUS, status); + bitmap_zero(used_mask, X86_PMC_IDX_MAX); - return status; -} + for (i = 0; i < n; i++) { + c = x86_pmu.get_event_constraints(cpuc, cpuc->event_list[i]); + constraints[i] = c; + } -static inline void intel_pmu_ack_status(u64 ack) -{ - wrmsrl(MSR_CORE_PERF_GLOBAL_OVF_CTRL, ack); -} + /* + * fastpath, try to reuse previous register + */ + for (i = 0; i < n; i++) { + hwc = &cpuc->event_list[i]->hw; + c = constraints[i]; -static inline void x86_pmu_enable_event(struct hw_perf_event *hwc, int idx) -{ - (void)checking_wrmsrl(hwc->config_base + idx, - hwc->config | ARCH_PERFMON_EVENTSEL0_ENABLE); -} + /* never assigned */ + if (hwc->idx == -1) + break; -static inline void x86_pmu_disable_event(struct hw_perf_event *hwc, int idx) -{ - (void)checking_wrmsrl(hwc->config_base + idx, hwc->config); + /* constraint still honored */ + if (!test_bit(hwc->idx, c->idxmsk)) + break; + + /* not already used */ + if (test_bit(hwc->idx, used_mask)) + break; + + __set_bit(hwc->idx, used_mask); + if (assign) + assign[i] = hwc->idx; + } + if (i == n) + goto done; + + /* + * begin slow path + */ + + bitmap_zero(used_mask, X86_PMC_IDX_MAX); + + /* + * weight = number of possible counters + * + * 1 = most constrained, only works on one counter + * wmax = least constrained, works on any counter + * + * assign events to counters starting with most + * constrained events. + */ + wmax = x86_pmu.num_counters; + + /* + * when fixed event counters are present, + * wmax is incremented by 1 to account + * for one more choice + */ + if (x86_pmu.num_counters_fixed) + wmax++; + + for (w = 1, num = n; num && w <= wmax; w++) { + /* for each event */ + for (i = 0; num && i < n; i++) { + c = constraints[i]; + hwc = &cpuc->event_list[i]->hw; + + if (c->weight != w) + continue; + + for_each_set_bit(j, c->idxmsk, X86_PMC_IDX_MAX) { + if (!test_bit(j, used_mask)) + break; + } + + if (j == X86_PMC_IDX_MAX) + break; + + __set_bit(j, used_mask); + + if (assign) + assign[i] = j; + num--; + } + } +done: + /* + * scheduling failed or is just a simulation, + * free resources if necessary + */ + if (!assign || num) { + for (i = 0; i < n; i++) { + if (x86_pmu.put_event_constraints) + x86_pmu.put_event_constraints(cpuc, cpuc->event_list[i]); + } + } + return num ? -ENOSPC : 0; } -static inline void -intel_pmu_disable_fixed(struct hw_perf_event *hwc, int __idx) +/* + * dogrp: true if must collect siblings events (group) + * returns total number of events and error code + */ +static int collect_events(struct cpu_hw_events *cpuc, struct perf_event *leader, bool dogrp) { - int idx = __idx - X86_PMC_IDX_FIXED; - u64 ctrl_val, mask; + struct perf_event *event; + int n, max_count; + + max_count = x86_pmu.num_counters + x86_pmu.num_counters_fixed; - mask = 0xfULL << (idx * 4); + /* current number of events already accepted */ + n = cpuc->n_events; + + if (is_x86_event(leader)) { + if (n >= max_count) + return -ENOSPC; + cpuc->event_list[n] = leader; + n++; + } + if (!dogrp) + return n; + + list_for_each_entry(event, &leader->sibling_list, group_entry) { + if (!is_x86_event(event) || + event->state <= PERF_EVENT_STATE_OFF) + continue; - rdmsrl(hwc->config_base, ctrl_val); - ctrl_val &= ~mask; - (void)checking_wrmsrl(hwc->config_base, ctrl_val); + if (n >= max_count) + return -ENOSPC; + + cpuc->event_list[n] = event; + n++; + } + return n; } -static inline void -p6_pmu_disable_event(struct hw_perf_event *hwc, int idx) +static inline void x86_assign_hw_event(struct perf_event *event, + struct cpu_hw_events *cpuc, int i) { - struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); - u64 val = P6_NOP_EVENT; + struct hw_perf_event *hwc = &event->hw; - if (cpuc->enabled) - val |= ARCH_PERFMON_EVENTSEL0_ENABLE; + hwc->idx = cpuc->assign[i]; + hwc->last_cpu = smp_processor_id(); + hwc->last_tag = ++cpuc->tags[i]; - (void)checking_wrmsrl(hwc->config_base + idx, val); + if (hwc->idx == X86_PMC_IDX_FIXED_BTS) { + hwc->config_base = 0; + hwc->event_base = 0; + } else if (hwc->idx >= X86_PMC_IDX_FIXED) { + hwc->config_base = MSR_ARCH_PERFMON_FIXED_CTR_CTRL; + /* + * We set it so that event_base + idx in wrmsr/rdmsr maps to + * MSR_ARCH_PERFMON_FIXED_CTR0 ... CTR2: + */ + hwc->event_base = + MSR_ARCH_PERFMON_FIXED_CTR0 - X86_PMC_IDX_FIXED; + } else { + hwc->config_base = x86_pmu.eventsel; + hwc->event_base = x86_pmu.perfctr; + } +} + +static inline int match_prev_assignment(struct hw_perf_event *hwc, + struct cpu_hw_events *cpuc, + int i) +{ + return hwc->idx == cpuc->assign[i] && + hwc->last_cpu == smp_processor_id() && + hwc->last_tag == cpuc->tags[i]; } -static inline void -intel_pmu_disable_event(struct hw_perf_event *hwc, int idx) +static int x86_pmu_start(struct perf_event *event); +static void x86_pmu_stop(struct perf_event *event); + +void hw_perf_enable(void) { - if (unlikely(idx == X86_PMC_IDX_FIXED_BTS)) { - intel_pmu_disable_bts(); + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + struct perf_event *event; + struct hw_perf_event *hwc; + int i, added = cpuc->n_added; + + if (!x86_pmu_initialized()) return; - } - if (unlikely(hwc->config_base == MSR_ARCH_PERFMON_FIXED_CTR_CTRL)) { - intel_pmu_disable_fixed(hwc, idx); + if (cpuc->enabled) return; + + if (cpuc->n_added) { + int n_running = cpuc->n_events - cpuc->n_added; + /* + * apply assignment obtained either from + * hw_perf_group_sched_in() or x86_pmu_enable() + * + * step1: save events moving to new counters + * step2: reprogram moved events into new counters + */ + for (i = 0; i < n_running; i++) { + event = cpuc->event_list[i]; + hwc = &event->hw; + + /* + * we can avoid reprogramming counter if: + * - assigned same counter as last time + * - running on same CPU as last time + * - no other event has used the counter since + */ + if (hwc->idx == -1 || + match_prev_assignment(hwc, cpuc, i)) + continue; + + x86_pmu_stop(event); + } + + for (i = 0; i < cpuc->n_events; i++) { + event = cpuc->event_list[i]; + hwc = &event->hw; + + if (!match_prev_assignment(hwc, cpuc, i)) + x86_assign_hw_event(event, cpuc, i); + else if (i < n_running) + continue; + + x86_pmu_start(event); + } + cpuc->n_added = 0; + perf_events_lapic_init(); } - x86_pmu_disable_event(hwc, idx); + cpuc->enabled = 1; + barrier(); + + x86_pmu.enable_all(added); } -static inline void -amd_pmu_disable_event(struct hw_perf_event *hwc, int idx) +static inline void __x86_pmu_enable_event(struct hw_perf_event *hwc, + u64 enable_mask) { - x86_pmu_disable_event(hwc, idx); + wrmsrl(hwc->config_base + hwc->idx, hwc->config | enable_mask); +} + +static inline void x86_pmu_disable_event(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + + wrmsrl(hwc->config_base + hwc->idx, hwc->config); } static DEFINE_PER_CPU(u64 [X86_PMC_IDX_MAX], pmc_prev_left); @@ -1275,12 +881,12 @@ static DEFINE_PER_CPU(u64 [X86_PMC_IDX_MAX], pmc_prev_left); * To be called with the event disabled in hw: */ static int -x86_perf_event_set_period(struct perf_event *event, - struct hw_perf_event *hwc, int idx) +x86_perf_event_set_period(struct perf_event *event) { + struct hw_perf_event *hwc = &event->hw; s64 left = atomic64_read(&hwc->period_left); s64 period = hwc->sample_period; - int err, ret = 0; + int ret = 0, idx = hwc->idx; if (idx == X86_PMC_IDX_FIXED_BTS) return 0; @@ -1318,227 +924,81 @@ x86_perf_event_set_period(struct perf_event *event, */ atomic64_set(&hwc->prev_count, (u64)-left); - err = checking_wrmsrl(hwc->event_base + idx, - (u64)(-left) & x86_pmu.event_mask); + wrmsrl(hwc->event_base + idx, + (u64)(-left) & x86_pmu.cntval_mask); perf_event_update_userpage(event); return ret; } -static inline void -intel_pmu_enable_fixed(struct hw_perf_event *hwc, int __idx) -{ - int idx = __idx - X86_PMC_IDX_FIXED; - u64 ctrl_val, bits, mask; - int err; - - /* - * Enable IRQ generation (0x8), - * and enable ring-3 counting (0x2) and ring-0 counting (0x1) - * if requested: - */ - bits = 0x8ULL; - if (hwc->config & ARCH_PERFMON_EVENTSEL_USR) - bits |= 0x2; - if (hwc->config & ARCH_PERFMON_EVENTSEL_OS) - bits |= 0x1; - - /* - * ANY bit is supported in v3 and up - */ - if (x86_pmu.version > 2 && hwc->config & ARCH_PERFMON_EVENTSEL_ANY) - bits |= 0x4; - - bits <<= (idx * 4); - mask = 0xfULL << (idx * 4); - - rdmsrl(hwc->config_base, ctrl_val); - ctrl_val &= ~mask; - ctrl_val |= bits; - err = checking_wrmsrl(hwc->config_base, ctrl_val); -} - -static void p6_pmu_enable_event(struct hw_perf_event *hwc, int idx) +static void x86_pmu_enable_event(struct perf_event *event) { struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); - u64 val; - - val = hwc->config; if (cpuc->enabled) - val |= ARCH_PERFMON_EVENTSEL0_ENABLE; - - (void)checking_wrmsrl(hwc->config_base + idx, val); -} - - -static void intel_pmu_enable_event(struct hw_perf_event *hwc, int idx) -{ - if (unlikely(idx == X86_PMC_IDX_FIXED_BTS)) { - if (!__get_cpu_var(cpu_hw_events).enabled) - return; - - intel_pmu_enable_bts(hwc->config); - return; - } - - if (unlikely(hwc->config_base == MSR_ARCH_PERFMON_FIXED_CTR_CTRL)) { - intel_pmu_enable_fixed(hwc, idx); - return; - } - - x86_pmu_enable_event(hwc, idx); + __x86_pmu_enable_event(&event->hw, + ARCH_PERFMON_EVENTSEL_ENABLE); } -static void amd_pmu_enable_event(struct hw_perf_event *hwc, int idx) +/* + * activate a single event + * + * The event is added to the group of enabled events + * but only if it can be scehduled with existing events. + * + * Called with PMU disabled. If successful and return value 1, + * then guaranteed to call perf_enable() and hw_perf_enable() + */ +static int x86_pmu_enable(struct perf_event *event) { struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + struct hw_perf_event *hwc; + int assign[X86_PMC_IDX_MAX]; + int n, n0, ret; - if (cpuc->enabled) - x86_pmu_enable_event(hwc, idx); -} - -static int fixed_mode_idx(struct hw_perf_event *hwc) -{ - unsigned int hw_event; - - hw_event = hwc->config & ARCH_PERFMON_EVENT_MASK; - - if (unlikely((hw_event == - x86_pmu.event_map(PERF_COUNT_HW_BRANCH_INSTRUCTIONS)) && - (hwc->sample_period == 1))) - return X86_PMC_IDX_FIXED_BTS; + hwc = &event->hw; - if (!x86_pmu.num_events_fixed) - return -1; + n0 = cpuc->n_events; + n = collect_events(cpuc, event, false); + if (n < 0) + return n; /* - * fixed counters do not take all possible filters + * If group events scheduling transaction was started, + * skip the schedulability test here, it will be peformed + * at commit time(->commit_txn) as a whole */ - if (hwc->config & ARCH_PERFMON_EVENT_FILTER_MASK) - return -1; - - if (unlikely(hw_event == x86_pmu.event_map(PERF_COUNT_HW_INSTRUCTIONS))) - return X86_PMC_IDX_FIXED_INSTRUCTIONS; - if (unlikely(hw_event == x86_pmu.event_map(PERF_COUNT_HW_CPU_CYCLES))) - return X86_PMC_IDX_FIXED_CPU_CYCLES; - if (unlikely(hw_event == x86_pmu.event_map(PERF_COUNT_HW_BUS_CYCLES))) - return X86_PMC_IDX_FIXED_BUS_CYCLES; + if (cpuc->group_flag & PERF_EVENT_TXN_STARTED) + goto out; - return -1; -} - -/* - * generic counter allocator: get next free counter - */ -static int -gen_get_event_idx(struct cpu_hw_events *cpuc, struct hw_perf_event *hwc) -{ - int idx; - - idx = find_first_zero_bit(cpuc->used_mask, x86_pmu.num_events); - return idx == x86_pmu.num_events ? -1 : idx; -} - -/* - * intel-specific counter allocator: check event constraints - */ -static int -intel_get_event_idx(struct cpu_hw_events *cpuc, struct hw_perf_event *hwc) -{ - const struct event_constraint *event_constraint; - int i, code; - - if (!event_constraints) - goto skip; - - code = hwc->config & CORE_EVNTSEL_EVENT_MASK; - - for_each_event_constraint(event_constraint, event_constraints) { - if (code == event_constraint->code) { - for_each_bit(i, event_constraint->idxmsk, X86_PMC_IDX_MAX) { - if (!test_and_set_bit(i, cpuc->used_mask)) - return i; - } - return -1; - } - } -skip: - return gen_get_event_idx(cpuc, hwc); -} - -static int -x86_schedule_event(struct cpu_hw_events *cpuc, struct hw_perf_event *hwc) -{ - int idx; - - idx = fixed_mode_idx(hwc); - if (idx == X86_PMC_IDX_FIXED_BTS) { - /* BTS is already occupied. */ - if (test_and_set_bit(idx, cpuc->used_mask)) - return -EAGAIN; - - hwc->config_base = 0; - hwc->event_base = 0; - hwc->idx = idx; - } else if (idx >= 0) { - /* - * Try to get the fixed event, if that is already taken - * then try to get a generic event: - */ - if (test_and_set_bit(idx, cpuc->used_mask)) - goto try_generic; + ret = x86_pmu.schedule_events(cpuc, n, assign); + if (ret) + return ret; + /* + * copy new assignment, now we know it is possible + * will be used by hw_perf_enable() + */ + memcpy(cpuc->assign, assign, n*sizeof(int)); - hwc->config_base = MSR_ARCH_PERFMON_FIXED_CTR_CTRL; - /* - * We set it so that event_base + idx in wrmsr/rdmsr maps to - * MSR_ARCH_PERFMON_FIXED_CTR0 ... CTR2: - */ - hwc->event_base = - MSR_ARCH_PERFMON_FIXED_CTR0 - X86_PMC_IDX_FIXED; - hwc->idx = idx; - } else { - idx = hwc->idx; - /* Try to get the previous generic event again */ - if (idx == -1 || test_and_set_bit(idx, cpuc->used_mask)) { -try_generic: - idx = x86_pmu.get_event_idx(cpuc, hwc); - if (idx == -1) - return -EAGAIN; - - set_bit(idx, cpuc->used_mask); - hwc->idx = idx; - } - hwc->config_base = x86_pmu.eventsel; - hwc->event_base = x86_pmu.perfctr; - } +out: + cpuc->n_events = n; + cpuc->n_added += n - n0; - return idx; + return 0; } -/* - * Find a PMC slot for the freshly enabled / scheduled in event: - */ -static int x86_pmu_enable(struct perf_event *event) +static int x86_pmu_start(struct perf_event *event) { struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); - struct hw_perf_event *hwc = &event->hw; - int idx; - - idx = x86_schedule_event(cpuc, hwc); - if (idx < 0) - return idx; - - perf_events_lapic_init(); + int idx = event->hw.idx; - x86_pmu.disable(hwc, idx); + if (idx == -1) + return -EAGAIN; + x86_perf_event_set_period(event); cpuc->events[idx] = event; - set_bit(idx, cpuc->active_mask); - - x86_perf_event_set_period(event, hwc, idx); - x86_pmu.enable(hwc, idx); - + __set_bit(idx, cpuc->active_mask); + x86_pmu.enable(event); perf_event_update_userpage(event); return 0; @@ -1546,24 +1006,19 @@ static int x86_pmu_enable(struct perf_event *event) static void x86_pmu_unthrottle(struct perf_event *event) { - struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); - struct hw_perf_event *hwc = &event->hw; - - if (WARN_ON_ONCE(hwc->idx >= X86_PMC_IDX_MAX || - cpuc->events[hwc->idx] != event)) - return; - - x86_pmu.enable(hwc, hwc->idx); + int ret = x86_pmu_start(event); + WARN_ON_ONCE(ret); } void perf_event_print_debug(void) { u64 ctrl, status, overflow, pmc_ctrl, pmc_count, prev_left, fixed; + u64 pebs; struct cpu_hw_events *cpuc; unsigned long flags; int cpu, idx; - if (!x86_pmu.num_events) + if (!x86_pmu.num_counters) return; local_irq_save(flags); @@ -1576,16 +1031,18 @@ void perf_event_print_debug(void) rdmsrl(MSR_CORE_PERF_GLOBAL_STATUS, status); rdmsrl(MSR_CORE_PERF_GLOBAL_OVF_CTRL, overflow); rdmsrl(MSR_ARCH_PERFMON_FIXED_CTR_CTRL, fixed); + rdmsrl(MSR_IA32_PEBS_ENABLE, pebs); pr_info("\n"); pr_info("CPU#%d: ctrl: %016llx\n", cpu, ctrl); pr_info("CPU#%d: status: %016llx\n", cpu, status); pr_info("CPU#%d: overflow: %016llx\n", cpu, overflow); pr_info("CPU#%d: fixed: %016llx\n", cpu, fixed); + pr_info("CPU#%d: pebs: %016llx\n", cpu, pebs); } - pr_info("CPU#%d: used: %016llx\n", cpu, *(u64 *)cpuc->used_mask); + pr_info("CPU#%d: active: %016llx\n", cpu, *(u64 *)cpuc->active_mask); - for (idx = 0; idx < x86_pmu.num_events; idx++) { + for (idx = 0; idx < x86_pmu.num_counters; idx++) { rdmsrl(x86_pmu.eventsel + idx, pmc_ctrl); rdmsrl(x86_pmu.perfctr + idx, pmc_count); @@ -1598,7 +1055,7 @@ void perf_event_print_debug(void) pr_info("CPU#%d: gen-PMC%d left: %016llx\n", cpu, idx, prev_left); } - for (idx = 0; idx < x86_pmu.num_events_fixed; idx++) { + for (idx = 0; idx < x86_pmu.num_counters_fixed; idx++) { rdmsrl(MSR_ARCH_PERFMON_FIXED_CTR0 + idx, pmc_count); pr_info("CPU#%d: fixed-PMC%d count: %016llx\n", @@ -1607,257 +1064,50 @@ void perf_event_print_debug(void) local_irq_restore(flags); } -static void intel_pmu_drain_bts_buffer(struct cpu_hw_events *cpuc) -{ - struct debug_store *ds = cpuc->ds; - struct bts_record { - u64 from; - u64 to; - u64 flags; - }; - struct perf_event *event = cpuc->events[X86_PMC_IDX_FIXED_BTS]; - struct bts_record *at, *top; - struct perf_output_handle handle; - struct perf_event_header header; - struct perf_sample_data data; - struct pt_regs regs; - - if (!event) - return; - - if (!ds) - return; - - at = (struct bts_record *)(unsigned long)ds->bts_buffer_base; - top = (struct bts_record *)(unsigned long)ds->bts_index; - - if (top <= at) - return; - - ds->bts_index = ds->bts_buffer_base; - - - data.period = event->hw.last_period; - data.addr = 0; - data.raw = NULL; - regs.ip = 0; - - /* - * Prepare a generic sample, i.e. fill in the invariant fields. - * We will overwrite the from and to address before we output - * the sample. - */ - perf_prepare_sample(&header, &data, event, ®s); - - if (perf_output_begin(&handle, event, - header.size * (top - at), 1, 1)) - return; - - for (; at < top; at++) { - data.ip = at->from; - data.addr = at->to; - - perf_output_sample(&handle, &header, &data, event); - } - - perf_output_end(&handle); - - /* There's new data available. */ - event->hw.interrupts++; - event->pending_kill = POLL_IN; -} - -static void x86_pmu_disable(struct perf_event *event) +static void x86_pmu_stop(struct perf_event *event) { struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); struct hw_perf_event *hwc = &event->hw; int idx = hwc->idx; - /* - * Must be done before we disable, otherwise the nmi handler - * could reenable again: - */ - clear_bit(idx, cpuc->active_mask); - x86_pmu.disable(hwc, idx); + if (!__test_and_clear_bit(idx, cpuc->active_mask)) + return; - /* - * Make sure the cleared pointer becomes visible before we - * (potentially) free the event: - */ - barrier(); + x86_pmu.disable(event); /* * Drain the remaining delta count out of a event * that we are disabling: */ - x86_perf_event_update(event, hwc, idx); - - /* Drain the remaining BTS records. */ - if (unlikely(idx == X86_PMC_IDX_FIXED_BTS)) - intel_pmu_drain_bts_buffer(cpuc); + x86_perf_event_update(event); cpuc->events[idx] = NULL; - clear_bit(idx, cpuc->used_mask); - - perf_event_update_userpage(event); -} - -/* - * Save and restart an expired event. Called by NMI contexts, - * so it has to be careful about preempting normal event ops: - */ -static int intel_pmu_save_and_restart(struct perf_event *event) -{ - struct hw_perf_event *hwc = &event->hw; - int idx = hwc->idx; - int ret; - - x86_perf_event_update(event, hwc, idx); - ret = x86_perf_event_set_period(event, hwc, idx); - - if (event->state == PERF_EVENT_STATE_ACTIVE) - intel_pmu_enable_event(hwc, idx); - - return ret; -} - -static void intel_pmu_reset(void) -{ - struct debug_store *ds = __get_cpu_var(cpu_hw_events).ds; - unsigned long flags; - int idx; - - if (!x86_pmu.num_events) - return; - - local_irq_save(flags); - - printk("clearing PMU state on CPU#%d\n", smp_processor_id()); - - for (idx = 0; idx < x86_pmu.num_events; idx++) { - checking_wrmsrl(x86_pmu.eventsel + idx, 0ull); - checking_wrmsrl(x86_pmu.perfctr + idx, 0ull); - } - for (idx = 0; idx < x86_pmu.num_events_fixed; idx++) { - checking_wrmsrl(MSR_ARCH_PERFMON_FIXED_CTR0 + idx, 0ull); - } - if (ds) - ds->bts_index = ds->bts_buffer_base; - - local_irq_restore(flags); } -static int p6_pmu_handle_irq(struct pt_regs *regs) -{ - struct perf_sample_data data; - struct cpu_hw_events *cpuc; - struct perf_event *event; - struct hw_perf_event *hwc; - int idx, handled = 0; - u64 val; - - data.addr = 0; - data.raw = NULL; - - cpuc = &__get_cpu_var(cpu_hw_events); - - for (idx = 0; idx < x86_pmu.num_events; idx++) { - if (!test_bit(idx, cpuc->active_mask)) - continue; - - event = cpuc->events[idx]; - hwc = &event->hw; - - val = x86_perf_event_update(event, hwc, idx); - if (val & (1ULL << (x86_pmu.event_bits - 1))) - continue; - - /* - * event overflow - */ - handled = 1; - data.period = event->hw.last_period; - - if (!x86_perf_event_set_period(event, hwc, idx)) - continue; - - if (perf_event_overflow(event, 1, &data, regs)) - p6_pmu_disable_event(hwc, idx); - } - - if (handled) - inc_irq_stat(apic_perf_irqs); - - return handled; -} - -/* - * This handler is triggered by the local APIC, so the APIC IRQ handling - * rules apply: - */ -static int intel_pmu_handle_irq(struct pt_regs *regs) +static void x86_pmu_disable(struct perf_event *event) { - struct perf_sample_data data; - struct cpu_hw_events *cpuc; - int bit, loops; - u64 ack, status; - - data.addr = 0; - data.raw = NULL; - - cpuc = &__get_cpu_var(cpu_hw_events); - - perf_disable(); - intel_pmu_drain_bts_buffer(cpuc); - status = intel_pmu_get_status(); - if (!status) { - perf_enable(); - return 0; - } - - loops = 0; -again: - if (++loops > 100) { - WARN_ONCE(1, "perfevents: irq loop stuck!\n"); - perf_event_print_debug(); - intel_pmu_reset(); - perf_enable(); - return 1; - } + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + int i; - inc_irq_stat(apic_perf_irqs); - ack = status; - for_each_bit(bit, (unsigned long *)&status, X86_PMC_IDX_MAX) { - struct perf_event *event = cpuc->events[bit]; + x86_pmu_stop(event); - clear_bit(bit, (unsigned long *) &status); - if (!test_bit(bit, cpuc->active_mask)) - continue; + for (i = 0; i < cpuc->n_events; i++) { + if (event == cpuc->event_list[i]) { - if (!intel_pmu_save_and_restart(event)) - continue; + if (x86_pmu.put_event_constraints) + x86_pmu.put_event_constraints(cpuc, event); - data.period = event->hw.last_period; + while (++i < cpuc->n_events) + cpuc->event_list[i-1] = cpuc->event_list[i]; - if (perf_event_overflow(event, 1, &data, regs)) - intel_pmu_disable_event(&event->hw, bit); + --cpuc->n_events; + break; + } } - - intel_pmu_ack_status(ack); - - /* - * Repeat if there is more work to be done: - */ - status = intel_pmu_get_status(); - if (status) - goto again; - - perf_enable(); - - return 1; + perf_event_update_userpage(event); } -static int amd_pmu_handle_irq(struct pt_regs *regs) +static int x86_pmu_handle_irq(struct pt_regs *regs) { struct perf_sample_data data; struct cpu_hw_events *cpuc; @@ -1866,20 +1116,19 @@ static int amd_pmu_handle_irq(struct pt_regs *regs) int idx, handled = 0; u64 val; - data.addr = 0; - data.raw = NULL; + perf_sample_data_init(&data, 0); cpuc = &__get_cpu_var(cpu_hw_events); - for (idx = 0; idx < x86_pmu.num_events; idx++) { + for (idx = 0; idx < x86_pmu.num_counters; idx++) { if (!test_bit(idx, cpuc->active_mask)) continue; event = cpuc->events[idx]; hwc = &event->hw; - val = x86_perf_event_update(event, hwc, idx); - if (val & (1ULL << (x86_pmu.event_bits - 1))) + val = x86_perf_event_update(event); + if (val & (1ULL << (x86_pmu.cntval_bits - 1))) continue; /* @@ -1888,11 +1137,11 @@ static int amd_pmu_handle_irq(struct pt_regs *regs) handled = 1; data.period = event->hw.last_period; - if (!x86_perf_event_set_period(event, hwc, idx)) + if (!x86_perf_event_set_period(event)) continue; if (perf_event_overflow(event, 1, &data, regs)) - amd_pmu_disable_event(hwc, idx); + x86_pmu_stop(event); } if (handled) @@ -1922,7 +1171,6 @@ void set_perf_event_pending(void) void perf_events_lapic_init(void) { -#ifdef CONFIG_X86_LOCAL_APIC if (!x86_pmu.apic || !x86_pmu_initialized()) return; @@ -1930,7 +1178,6 @@ void perf_events_lapic_init(void) * Always use NMI for PMU */ apic_write(APIC_LVTPC, APIC_DM_NMI); -#endif } static int __kprobes @@ -1954,9 +1201,7 @@ perf_event_nmi_handler(struct notifier_block *self, regs = args->regs; -#ifdef CONFIG_X86_LOCAL_APIC apic_write(APIC_LVTPC, APIC_DM_NMI); -#endif /* * Can't rely on the handled return value to say it was our NMI, two * events could trigger 'simultaneously' raising two back-to-back NMIs. @@ -1975,193 +1220,64 @@ static __read_mostly struct notifier_block perf_event_nmi_notifier = { .priority = 1 }; -static __initconst struct x86_pmu p6_pmu = { - .name = "p6", - .handle_irq = p6_pmu_handle_irq, - .disable_all = p6_pmu_disable_all, - .enable_all = p6_pmu_enable_all, - .enable = p6_pmu_enable_event, - .disable = p6_pmu_disable_event, - .eventsel = MSR_P6_EVNTSEL0, - .perfctr = MSR_P6_PERFCTR0, - .event_map = p6_pmu_event_map, - .raw_event = p6_pmu_raw_event, - .max_events = ARRAY_SIZE(p6_perfmon_event_map), - .apic = 1, - .max_period = (1ULL << 31) - 1, - .version = 0, - .num_events = 2, - /* - * Events have 40 bits implemented. However they are designed such - * that bits [32-39] are sign extensions of bit 31. As such the - * effective width of a event for P6-like PMU is 32 bits only. - * - * See IA-32 Intel Architecture Software developer manual Vol 3B - */ - .event_bits = 32, - .event_mask = (1ULL << 32) - 1, - .get_event_idx = intel_get_event_idx, -}; - -static __initconst struct x86_pmu intel_pmu = { - .name = "Intel", - .handle_irq = intel_pmu_handle_irq, - .disable_all = intel_pmu_disable_all, - .enable_all = intel_pmu_enable_all, - .enable = intel_pmu_enable_event, - .disable = intel_pmu_disable_event, - .eventsel = MSR_ARCH_PERFMON_EVENTSEL0, - .perfctr = MSR_ARCH_PERFMON_PERFCTR0, - .event_map = intel_pmu_event_map, - .raw_event = intel_pmu_raw_event, - .max_events = ARRAY_SIZE(intel_perfmon_event_map), - .apic = 1, - /* - * Intel PMCs cannot be accessed sanely above 32 bit width, - * so we install an artificial 1<<31 period regardless of - * the generic event period: - */ - .max_period = (1ULL << 31) - 1, - .enable_bts = intel_pmu_enable_bts, - .disable_bts = intel_pmu_disable_bts, - .get_event_idx = intel_get_event_idx, -}; +static struct event_constraint unconstrained; +static struct event_constraint emptyconstraint; -static __initconst struct x86_pmu amd_pmu = { - .name = "AMD", - .handle_irq = amd_pmu_handle_irq, - .disable_all = amd_pmu_disable_all, - .enable_all = amd_pmu_enable_all, - .enable = amd_pmu_enable_event, - .disable = amd_pmu_disable_event, - .eventsel = MSR_K7_EVNTSEL0, - .perfctr = MSR_K7_PERFCTR0, - .event_map = amd_pmu_event_map, - .raw_event = amd_pmu_raw_event, - .max_events = ARRAY_SIZE(amd_perfmon_event_map), - .num_events = 4, - .event_bits = 48, - .event_mask = (1ULL << 48) - 1, - .apic = 1, - /* use highest bit to detect overflow */ - .max_period = (1ULL << 47) - 1, - .get_event_idx = gen_get_event_idx, -}; - -static __init int p6_pmu_init(void) +static struct event_constraint * +x86_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event) { - switch (boot_cpu_data.x86_model) { - case 1: - case 3: /* Pentium Pro */ - case 5: - case 6: /* Pentium II */ - case 7: - case 8: - case 11: /* Pentium III */ - event_constraints = intel_p6_event_constraints; - break; - case 9: - case 13: - /* Pentium M */ - event_constraints = intel_p6_event_constraints; - break; - default: - pr_cont("unsupported p6 CPU model %d ", - boot_cpu_data.x86_model); - return -ENODEV; - } + struct event_constraint *c; - x86_pmu = p6_pmu; + if (x86_pmu.event_constraints) { + for_each_event_constraint(c, x86_pmu.event_constraints) { + if ((event->hw.config & c->cmask) == c->code) + return c; + } + } - return 0; + return &unconstrained; } -static __init int intel_pmu_init(void) -{ - union cpuid10_edx edx; - union cpuid10_eax eax; - unsigned int unused; - unsigned int ebx; - int version; - - if (!cpu_has(&boot_cpu_data, X86_FEATURE_ARCH_PERFMON)) { - /* check for P6 processor family */ - if (boot_cpu_data.x86 == 6) { - return p6_pmu_init(); - } else { - return -ENODEV; - } - } +#include "perf_event_amd.c" +#include "perf_event_p6.c" +#include "perf_event_p4.c" +#include "perf_event_intel_lbr.c" +#include "perf_event_intel_ds.c" +#include "perf_event_intel.c" - /* - * Check whether the Architectural PerfMon supports - * Branch Misses Retired hw_event or not. - */ - cpuid(10, &eax.full, &ebx, &unused, &edx.full); - if (eax.split.mask_length <= ARCH_PERFMON_BRANCH_MISSES_RETIRED) - return -ENODEV; - - version = eax.split.version_id; - if (version < 2) - return -ENODEV; +static int __cpuinit +x86_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu) +{ + unsigned int cpu = (long)hcpu; + int ret = NOTIFY_OK; - x86_pmu = intel_pmu; - x86_pmu.version = version; - x86_pmu.num_events = eax.split.num_events; - x86_pmu.event_bits = eax.split.bit_width; - x86_pmu.event_mask = (1ULL << eax.split.bit_width) - 1; + switch (action & ~CPU_TASKS_FROZEN) { + case CPU_UP_PREPARE: + if (x86_pmu.cpu_prepare) + ret = x86_pmu.cpu_prepare(cpu); + break; - /* - * Quirk: v2 perfmon does not report fixed-purpose events, so - * assume at least 3 events: - */ - x86_pmu.num_events_fixed = max((int)edx.split.num_events_fixed, 3); + case CPU_STARTING: + if (x86_pmu.cpu_starting) + x86_pmu.cpu_starting(cpu); + break; - /* - * Install the hw-cache-events table: - */ - switch (boot_cpu_data.x86_model) { - case 15: /* original 65 nm celeron/pentium/core2/xeon, "Merom"/"Conroe" */ - case 22: /* single-core 65 nm celeron/core2solo "Merom-L"/"Conroe-L" */ - case 23: /* current 45 nm celeron/core2/xeon "Penryn"/"Wolfdale" */ - case 29: /* six-core 45 nm xeon "Dunnington" */ - memcpy(hw_cache_event_ids, core2_hw_cache_event_ids, - sizeof(hw_cache_event_ids)); - - pr_cont("Core2 events, "); - event_constraints = intel_core_event_constraints; + case CPU_DYING: + if (x86_pmu.cpu_dying) + x86_pmu.cpu_dying(cpu); break; - default: - case 26: - memcpy(hw_cache_event_ids, nehalem_hw_cache_event_ids, - sizeof(hw_cache_event_ids)); - event_constraints = intel_nehalem_event_constraints; - pr_cont("Nehalem/Corei7 events, "); + case CPU_UP_CANCELED: + case CPU_DEAD: + if (x86_pmu.cpu_dead) + x86_pmu.cpu_dead(cpu); break; - case 28: - memcpy(hw_cache_event_ids, atom_hw_cache_event_ids, - sizeof(hw_cache_event_ids)); - pr_cont("Atom events, "); + default: break; } - return 0; -} -static __init int amd_pmu_init(void) -{ - /* Performance-monitoring supported from K7 and later: */ - if (boot_cpu_data.x86 < 6) - return -ENODEV; - - x86_pmu = amd_pmu; - - /* Events are common for all AMDs */ - memcpy(hw_cache_event_ids, amd_hw_cache_event_ids, - sizeof(hw_cache_event_ids)); - - return 0; + return ret; } static void __init pmu_check_apic(void) @@ -2176,6 +1292,7 @@ static void __init pmu_check_apic(void) void __init init_hw_perf_events(void) { + struct event_constraint *c; int err; pr_info("Performance Events: "); @@ -2199,88 +1316,219 @@ void __init init_hw_perf_events(void) pr_cont("%s PMU driver.\n", x86_pmu.name); - if (x86_pmu.num_events > X86_PMC_MAX_GENERIC) { + if (x86_pmu.quirks) + x86_pmu.quirks(); + + if (x86_pmu.num_counters > X86_PMC_MAX_GENERIC) { WARN(1, KERN_ERR "hw perf events %d > max(%d), clipping!", - x86_pmu.num_events, X86_PMC_MAX_GENERIC); - x86_pmu.num_events = X86_PMC_MAX_GENERIC; + x86_pmu.num_counters, X86_PMC_MAX_GENERIC); + x86_pmu.num_counters = X86_PMC_MAX_GENERIC; } - perf_event_mask = (1 << x86_pmu.num_events) - 1; - perf_max_events = x86_pmu.num_events; + x86_pmu.intel_ctrl = (1 << x86_pmu.num_counters) - 1; + perf_max_events = x86_pmu.num_counters; - if (x86_pmu.num_events_fixed > X86_PMC_MAX_FIXED) { + if (x86_pmu.num_counters_fixed > X86_PMC_MAX_FIXED) { WARN(1, KERN_ERR "hw perf events fixed %d > max(%d), clipping!", - x86_pmu.num_events_fixed, X86_PMC_MAX_FIXED); - x86_pmu.num_events_fixed = X86_PMC_MAX_FIXED; + x86_pmu.num_counters_fixed, X86_PMC_MAX_FIXED); + x86_pmu.num_counters_fixed = X86_PMC_MAX_FIXED; } - perf_event_mask |= - ((1LL << x86_pmu.num_events_fixed)-1) << X86_PMC_IDX_FIXED; - x86_pmu.intel_ctrl = perf_event_mask; + x86_pmu.intel_ctrl |= + ((1LL << x86_pmu.num_counters_fixed)-1) << X86_PMC_IDX_FIXED; perf_events_lapic_init(); register_die_notifier(&perf_event_nmi_notifier); + unconstrained = (struct event_constraint) + __EVENT_CONSTRAINT(0, (1ULL << x86_pmu.num_counters) - 1, + 0, x86_pmu.num_counters); + + if (x86_pmu.event_constraints) { + for_each_event_constraint(c, x86_pmu.event_constraints) { + if (c->cmask != X86_RAW_EVENT_MASK) + continue; + + c->idxmsk64 |= (1ULL << x86_pmu.num_counters) - 1; + c->weight += x86_pmu.num_counters; + } + } + pr_info("... version: %d\n", x86_pmu.version); - pr_info("... bit width: %d\n", x86_pmu.event_bits); - pr_info("... generic registers: %d\n", x86_pmu.num_events); - pr_info("... value mask: %016Lx\n", x86_pmu.event_mask); + pr_info("... bit width: %d\n", x86_pmu.cntval_bits); + pr_info("... generic registers: %d\n", x86_pmu.num_counters); + pr_info("... value mask: %016Lx\n", x86_pmu.cntval_mask); pr_info("... max period: %016Lx\n", x86_pmu.max_period); - pr_info("... fixed-purpose events: %d\n", x86_pmu.num_events_fixed); - pr_info("... event mask: %016Lx\n", perf_event_mask); + pr_info("... fixed-purpose events: %d\n", x86_pmu.num_counters_fixed); + pr_info("... event mask: %016Lx\n", x86_pmu.intel_ctrl); + + perf_cpu_notifier(x86_pmu_notifier); } static inline void x86_pmu_read(struct perf_event *event) { - x86_perf_event_update(event, &event->hw, event->hw.idx); + x86_perf_event_update(event); +} + +/* + * Start group events scheduling transaction + * Set the flag to make pmu::enable() not perform the + * schedulability test, it will be performed at commit time + */ +static void x86_pmu_start_txn(const struct pmu *pmu) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + + cpuc->group_flag |= PERF_EVENT_TXN_STARTED; +} + +/* + * Stop group events scheduling transaction + * Clear the flag and pmu::enable() will perform the + * schedulability test. + */ +static void x86_pmu_cancel_txn(const struct pmu *pmu) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + + cpuc->group_flag &= ~PERF_EVENT_TXN_STARTED; +} + +/* + * Commit group events scheduling transaction + * Perform the group schedulability test as a whole + * Return 0 if success + */ +static int x86_pmu_commit_txn(const struct pmu *pmu) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + int assign[X86_PMC_IDX_MAX]; + int n, ret; + + n = cpuc->n_events; + + if (!x86_pmu_initialized()) + return -EAGAIN; + + ret = x86_pmu.schedule_events(cpuc, n, assign); + if (ret) + return ret; + + /* + * copy new assignment, now we know it is possible + * will be used by hw_perf_enable() + */ + memcpy(cpuc->assign, assign, n*sizeof(int)); + + return 0; } static const struct pmu pmu = { .enable = x86_pmu_enable, .disable = x86_pmu_disable, + .start = x86_pmu_start, + .stop = x86_pmu_stop, .read = x86_pmu_read, .unthrottle = x86_pmu_unthrottle, + .start_txn = x86_pmu_start_txn, + .cancel_txn = x86_pmu_cancel_txn, + .commit_txn = x86_pmu_commit_txn, }; -static int -validate_event(struct cpu_hw_events *cpuc, struct perf_event *event) +/* + * validate that we can schedule this event + */ +static int validate_event(struct perf_event *event) { - struct hw_perf_event fake_event = event->hw; + struct cpu_hw_events *fake_cpuc; + struct event_constraint *c; + int ret = 0; - if (event->pmu && event->pmu != &pmu) - return 0; + fake_cpuc = kmalloc(sizeof(*fake_cpuc), GFP_KERNEL | __GFP_ZERO); + if (!fake_cpuc) + return -ENOMEM; + + c = x86_pmu.get_event_constraints(fake_cpuc, event); + + if (!c || !c->weight) + ret = -ENOSPC; - return x86_schedule_event(cpuc, &fake_event) >= 0; + if (x86_pmu.put_event_constraints) + x86_pmu.put_event_constraints(fake_cpuc, event); + + kfree(fake_cpuc); + + return ret; } +/* + * validate a single event group + * + * validation include: + * - check events are compatible which each other + * - events do not compete for the same counter + * - number of events <= number of counters + * + * validation ensures the group can be loaded onto the + * PMU if it was the only group available. + */ static int validate_group(struct perf_event *event) { - struct perf_event *sibling, *leader = event->group_leader; - struct cpu_hw_events fake_pmu; + struct perf_event *leader = event->group_leader; + struct cpu_hw_events *fake_cpuc; + int ret, n; + + ret = -ENOMEM; + fake_cpuc = kmalloc(sizeof(*fake_cpuc), GFP_KERNEL | __GFP_ZERO); + if (!fake_cpuc) + goto out; - memset(&fake_pmu, 0, sizeof(fake_pmu)); + /* + * the event is not yet connected with its + * siblings therefore we must first collect + * existing siblings, then add the new event + * before we can simulate the scheduling + */ + ret = -ENOSPC; + n = collect_events(fake_cpuc, leader, true); + if (n < 0) + goto out_free; - if (!validate_event(&fake_pmu, leader)) - return -ENOSPC; + fake_cpuc->n_events = n; + n = collect_events(fake_cpuc, event, false); + if (n < 0) + goto out_free; - list_for_each_entry(sibling, &leader->sibling_list, group_entry) { - if (!validate_event(&fake_pmu, sibling)) - return -ENOSPC; - } + fake_cpuc->n_events = n; - if (!validate_event(&fake_pmu, event)) - return -ENOSPC; + ret = x86_pmu.schedule_events(fake_cpuc, n, NULL); - return 0; +out_free: + kfree(fake_cpuc); +out: + return ret; } const struct pmu *hw_perf_event_init(struct perf_event *event) { + const struct pmu *tmp; int err; err = __hw_perf_event_init(event); if (!err) { + /* + * we temporarily connect event to its pmu + * such that validate_group() can classify + * it as an x86 event using is_x86_event() + */ + tmp = event->pmu; + event->pmu = &pmu; + if (event->group_leader != event) err = validate_group(event); + else + err = validate_event(event); + + event->pmu = tmp; } if (err) { if (event->destroy) @@ -2304,7 +1552,6 @@ void callchain_store(struct perf_callchain_entry *entry, u64 ip) static DEFINE_PER_CPU(struct perf_callchain_entry, pmc_irq_entry); static DEFINE_PER_CPU(struct perf_callchain_entry, pmc_nmi_entry); -static DEFINE_PER_CPU(int, in_ignored_frame); static void @@ -2320,10 +1567,6 @@ static void backtrace_warning(void *data, char *msg) static int backtrace_stack(void *data, char *name) { - per_cpu(in_ignored_frame, smp_processor_id()) = - x86_is_stack_id(NMI_STACK, name) || - x86_is_stack_id(DEBUG_STACK, name); - return 0; } @@ -2331,11 +1574,7 @@ static void backtrace_address(void *data, unsigned long addr, int reliable) { struct perf_callchain_entry *entry = data; - if (per_cpu(in_ignored_frame, smp_processor_id())) - return; - - if (reliable) - callchain_store(entry, addr); + callchain_store(entry, addr); } static const struct stacktrace_ops backtrace_ops = { @@ -2357,49 +1596,42 @@ perf_callchain_kernel(struct pt_regs *regs, struct perf_callchain_entry *entry) dump_trace(NULL, regs, NULL, regs->bp, &backtrace_ops, entry); } -/* - * best effort, GUP based copy_from_user() that assumes IRQ or NMI context - */ -static unsigned long -copy_from_user_nmi(void *to, const void __user *from, unsigned long n) +#ifdef CONFIG_COMPAT +static inline int +perf_callchain_user32(struct pt_regs *regs, struct perf_callchain_entry *entry) { - unsigned long offset, addr = (unsigned long)from; - int type = in_nmi() ? KM_NMI : KM_IRQ0; - unsigned long size, len = 0; - struct page *page; - void *map; - int ret; - - do { - ret = __get_user_pages_fast(addr, 1, 0, &page); - if (!ret) - break; + /* 32-bit process in 64-bit kernel. */ + struct stack_frame_ia32 frame; + const void __user *fp; - offset = addr & (PAGE_SIZE - 1); - size = min(PAGE_SIZE - offset, n - len); + if (!test_thread_flag(TIF_IA32)) + return 0; - map = kmap_atomic(page, type); - memcpy(to, map+offset, size); - kunmap_atomic(map, type); - put_page(page); + fp = compat_ptr(regs->bp); + while (entry->nr < PERF_MAX_STACK_DEPTH) { + unsigned long bytes; + frame.next_frame = 0; + frame.return_address = 0; - len += size; - to += size; - addr += size; + bytes = copy_from_user_nmi(&frame, fp, sizeof(frame)); + if (bytes != sizeof(frame)) + break; - } while (len < n); + if (fp < compat_ptr(regs->sp)) + break; - return len; + callchain_store(entry, frame.return_address); + fp = compat_ptr(frame.next_frame); + } + return 1; } - -static int copy_stack_frame(const void __user *fp, struct stack_frame *frame) +#else +static inline int +perf_callchain_user32(struct pt_regs *regs, struct perf_callchain_entry *entry) { - unsigned long bytes; - - bytes = copy_from_user_nmi(frame, fp, sizeof(*frame)); - - return bytes == sizeof(*frame); + return 0; } +#endif static void perf_callchain_user(struct pt_regs *regs, struct perf_callchain_entry *entry) @@ -2415,11 +1647,16 @@ perf_callchain_user(struct pt_regs *regs, struct perf_callchain_entry *entry) callchain_store(entry, PERF_CONTEXT_USER); callchain_store(entry, regs->ip); + if (perf_callchain_user32(regs, entry)) + return; + while (entry->nr < PERF_MAX_STACK_DEPTH) { + unsigned long bytes; frame.next_frame = NULL; frame.return_address = 0; - if (!copy_stack_frame(fp, &frame)) + bytes = copy_from_user_nmi(&frame, fp, sizeof(frame)); + if (bytes != sizeof(frame)) break; if ((unsigned long)fp < regs->sp) @@ -2440,9 +1677,6 @@ perf_do_callchain(struct pt_regs *regs, struct perf_callchain_entry *entry) is_user = user_mode(regs); - if (!current || current->pid == 0) - return; - if (is_user && current->state != TASK_RUNNING) return; @@ -2457,6 +1691,11 @@ struct perf_callchain_entry *perf_callchain(struct pt_regs *regs) { struct perf_callchain_entry *entry; + if (perf_guest_cbs && perf_guest_cbs->is_in_guest()) { + /* TODO: We don't support guest os callchain now */ + return NULL; + } + if (in_nmi()) entry = &__get_cpu_var(pmc_nmi_entry); else @@ -2469,7 +1708,52 @@ struct perf_callchain_entry *perf_callchain(struct pt_regs *regs) return entry; } -void hw_perf_event_setup_online(int cpu) +void perf_arch_fetch_caller_regs(struct pt_regs *regs, unsigned long ip, int skip) +{ + regs->ip = ip; + /* + * perf_arch_fetch_caller_regs adds another call, we need to increment + * the skip level + */ + regs->bp = rewind_frame_pointer(skip + 1); + regs->cs = __KERNEL_CS; + /* + * We abuse bit 3 to pass exact information, see perf_misc_flags + * and the comment with PERF_EFLAGS_EXACT. + */ + regs->flags = 0; +} + +unsigned long perf_instruction_pointer(struct pt_regs *regs) { - init_debug_store_on_cpu(cpu); + unsigned long ip; + + if (perf_guest_cbs && perf_guest_cbs->is_in_guest()) + ip = perf_guest_cbs->get_guest_ip(); + else + ip = instruction_pointer(regs); + + return ip; +} + +unsigned long perf_misc_flags(struct pt_regs *regs) +{ + int misc = 0; + + if (perf_guest_cbs && perf_guest_cbs->is_in_guest()) { + if (perf_guest_cbs->is_user_mode()) + misc |= PERF_RECORD_MISC_GUEST_USER; + else + misc |= PERF_RECORD_MISC_GUEST_KERNEL; + } else { + if (user_mode(regs)) + misc |= PERF_RECORD_MISC_USER; + else + misc |= PERF_RECORD_MISC_KERNEL; + } + + if (regs->flags & PERF_EFLAGS_EXACT) + misc |= PERF_RECORD_MISC_EXACT_IP; + + return misc; } diff --git a/arch/x86/kernel/cpu/perf_event_amd.c b/arch/x86/kernel/cpu/perf_event_amd.c new file mode 100644 index 00000000000..611df11ba15 --- /dev/null +++ b/arch/x86/kernel/cpu/perf_event_amd.c @@ -0,0 +1,420 @@ +#ifdef CONFIG_CPU_SUP_AMD + +static DEFINE_RAW_SPINLOCK(amd_nb_lock); + +static __initconst const u64 amd_hw_cache_event_ids + [PERF_COUNT_HW_CACHE_MAX] + [PERF_COUNT_HW_CACHE_OP_MAX] + [PERF_COUNT_HW_CACHE_RESULT_MAX] = +{ + [ C(L1D) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0040, /* Data Cache Accesses */ + [ C(RESULT_MISS) ] = 0x0041, /* Data Cache Misses */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x0142, /* Data Cache Refills :system */ + [ C(RESULT_MISS) ] = 0, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0x0267, /* Data Prefetcher :attempts */ + [ C(RESULT_MISS) ] = 0x0167, /* Data Prefetcher :cancelled */ + }, + }, + [ C(L1I ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0080, /* Instruction cache fetches */ + [ C(RESULT_MISS) ] = 0x0081, /* Instruction cache misses */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0x014B, /* Prefetch Instructions :Load */ + [ C(RESULT_MISS) ] = 0, + }, + }, + [ C(LL ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x037D, /* Requests to L2 Cache :IC+DC */ + [ C(RESULT_MISS) ] = 0x037E, /* L2 Cache Misses : IC+DC */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x017F, /* L2 Fill/Writeback */ + [ C(RESULT_MISS) ] = 0, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0, + [ C(RESULT_MISS) ] = 0, + }, + }, + [ C(DTLB) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0040, /* Data Cache Accesses */ + [ C(RESULT_MISS) ] = 0x0046, /* L1 DTLB and L2 DLTB Miss */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0, + [ C(RESULT_MISS) ] = 0, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0, + [ C(RESULT_MISS) ] = 0, + }, + }, + [ C(ITLB) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0080, /* Instruction fecthes */ + [ C(RESULT_MISS) ] = 0x0085, /* Instr. fetch ITLB misses */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + }, + [ C(BPU ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x00c2, /* Retired Branch Instr. */ + [ C(RESULT_MISS) ] = 0x00c3, /* Retired Mispredicted BI */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + }, +}; + +/* + * AMD Performance Monitor K7 and later. + */ +static const u64 amd_perfmon_event_map[] = +{ + [PERF_COUNT_HW_CPU_CYCLES] = 0x0076, + [PERF_COUNT_HW_INSTRUCTIONS] = 0x00c0, + [PERF_COUNT_HW_CACHE_REFERENCES] = 0x0080, + [PERF_COUNT_HW_CACHE_MISSES] = 0x0081, + [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = 0x00c4, + [PERF_COUNT_HW_BRANCH_MISSES] = 0x00c5, +}; + +static u64 amd_pmu_event_map(int hw_event) +{ + return amd_perfmon_event_map[hw_event]; +} + +static int amd_pmu_hw_config(struct perf_event *event) +{ + int ret = x86_pmu_hw_config(event); + + if (ret) + return ret; + + if (event->attr.type != PERF_TYPE_RAW) + return 0; + + event->hw.config |= event->attr.config & AMD64_RAW_EVENT_MASK; + + return 0; +} + +/* + * AMD64 events are detected based on their event codes. + */ +static inline int amd_is_nb_event(struct hw_perf_event *hwc) +{ + return (hwc->config & 0xe0) == 0xe0; +} + +static inline int amd_has_nb(struct cpu_hw_events *cpuc) +{ + struct amd_nb *nb = cpuc->amd_nb; + + return nb && nb->nb_id != -1; +} + +static void amd_put_event_constraints(struct cpu_hw_events *cpuc, + struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + struct amd_nb *nb = cpuc->amd_nb; + int i; + + /* + * only care about NB events + */ + if (!(amd_has_nb(cpuc) && amd_is_nb_event(hwc))) + return; + + /* + * need to scan whole list because event may not have + * been assigned during scheduling + * + * no race condition possible because event can only + * be removed on one CPU at a time AND PMU is disabled + * when we come here + */ + for (i = 0; i < x86_pmu.num_counters; i++) { + if (nb->owners[i] == event) { + cmpxchg(nb->owners+i, event, NULL); + break; + } + } +} + + /* + * AMD64 NorthBridge events need special treatment because + * counter access needs to be synchronized across all cores + * of a package. Refer to BKDG section 3.12 + * + * NB events are events measuring L3 cache, Hypertransport + * traffic. They are identified by an event code >= 0xe00. + * They measure events on the NorthBride which is shared + * by all cores on a package. NB events are counted on a + * shared set of counters. When a NB event is programmed + * in a counter, the data actually comes from a shared + * counter. Thus, access to those counters needs to be + * synchronized. + * + * We implement the synchronization such that no two cores + * can be measuring NB events using the same counters. Thus, + * we maintain a per-NB allocation table. The available slot + * is propagated using the event_constraint structure. + * + * We provide only one choice for each NB event based on + * the fact that only NB events have restrictions. Consequently, + * if a counter is available, there is a guarantee the NB event + * will be assigned to it. If no slot is available, an empty + * constraint is returned and scheduling will eventually fail + * for this event. + * + * Note that all cores attached the same NB compete for the same + * counters to host NB events, this is why we use atomic ops. Some + * multi-chip CPUs may have more than one NB. + * + * Given that resources are allocated (cmpxchg), they must be + * eventually freed for others to use. This is accomplished by + * calling amd_put_event_constraints(). + * + * Non NB events are not impacted by this restriction. + */ +static struct event_constraint * +amd_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + struct amd_nb *nb = cpuc->amd_nb; + struct perf_event *old = NULL; + int max = x86_pmu.num_counters; + int i, j, k = -1; + + /* + * if not NB event or no NB, then no constraints + */ + if (!(amd_has_nb(cpuc) && amd_is_nb_event(hwc))) + return &unconstrained; + + /* + * detect if already present, if so reuse + * + * cannot merge with actual allocation + * because of possible holes + * + * event can already be present yet not assigned (in hwc->idx) + * because of successive calls to x86_schedule_events() from + * hw_perf_group_sched_in() without hw_perf_enable() + */ + for (i = 0; i < max; i++) { + /* + * keep track of first free slot + */ + if (k == -1 && !nb->owners[i]) + k = i; + + /* already present, reuse */ + if (nb->owners[i] == event) + goto done; + } + /* + * not present, so grab a new slot + * starting either at: + */ + if (hwc->idx != -1) { + /* previous assignment */ + i = hwc->idx; + } else if (k != -1) { + /* start from free slot found */ + i = k; + } else { + /* + * event not found, no slot found in + * first pass, try again from the + * beginning + */ + i = 0; + } + j = i; + do { + old = cmpxchg(nb->owners+i, NULL, event); + if (!old) + break; + if (++i == max) + i = 0; + } while (i != j); +done: + if (!old) + return &nb->event_constraints[i]; + + return &emptyconstraint; +} + +static struct amd_nb *amd_alloc_nb(int cpu, int nb_id) +{ + struct amd_nb *nb; + int i; + + nb = kmalloc(sizeof(struct amd_nb), GFP_KERNEL); + if (!nb) + return NULL; + + memset(nb, 0, sizeof(*nb)); + nb->nb_id = nb_id; + + /* + * initialize all possible NB constraints + */ + for (i = 0; i < x86_pmu.num_counters; i++) { + __set_bit(i, nb->event_constraints[i].idxmsk); + nb->event_constraints[i].weight = 1; + } + return nb; +} + +static int amd_pmu_cpu_prepare(int cpu) +{ + struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu); + + WARN_ON_ONCE(cpuc->amd_nb); + + if (boot_cpu_data.x86_max_cores < 2) + return NOTIFY_OK; + + cpuc->amd_nb = amd_alloc_nb(cpu, -1); + if (!cpuc->amd_nb) + return NOTIFY_BAD; + + return NOTIFY_OK; +} + +static void amd_pmu_cpu_starting(int cpu) +{ + struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu); + struct amd_nb *nb; + int i, nb_id; + + if (boot_cpu_data.x86_max_cores < 2) + return; + + nb_id = amd_get_nb_id(cpu); + WARN_ON_ONCE(nb_id == BAD_APICID); + + raw_spin_lock(&amd_nb_lock); + + for_each_online_cpu(i) { + nb = per_cpu(cpu_hw_events, i).amd_nb; + if (WARN_ON_ONCE(!nb)) + continue; + + if (nb->nb_id == nb_id) { + kfree(cpuc->amd_nb); + cpuc->amd_nb = nb; + break; + } + } + + cpuc->amd_nb->nb_id = nb_id; + cpuc->amd_nb->refcnt++; + + raw_spin_unlock(&amd_nb_lock); +} + +static void amd_pmu_cpu_dead(int cpu) +{ + struct cpu_hw_events *cpuhw; + + if (boot_cpu_data.x86_max_cores < 2) + return; + + cpuhw = &per_cpu(cpu_hw_events, cpu); + + raw_spin_lock(&amd_nb_lock); + + if (cpuhw->amd_nb) { + struct amd_nb *nb = cpuhw->amd_nb; + + if (nb->nb_id == -1 || --nb->refcnt == 0) + kfree(nb); + + cpuhw->amd_nb = NULL; + } + + raw_spin_unlock(&amd_nb_lock); +} + +static __initconst const struct x86_pmu amd_pmu = { + .name = "AMD", + .handle_irq = x86_pmu_handle_irq, + .disable_all = x86_pmu_disable_all, + .enable_all = x86_pmu_enable_all, + .enable = x86_pmu_enable_event, + .disable = x86_pmu_disable_event, + .hw_config = amd_pmu_hw_config, + .schedule_events = x86_schedule_events, + .eventsel = MSR_K7_EVNTSEL0, + .perfctr = MSR_K7_PERFCTR0, + .event_map = amd_pmu_event_map, + .max_events = ARRAY_SIZE(amd_perfmon_event_map), + .num_counters = 4, + .cntval_bits = 48, + .cntval_mask = (1ULL << 48) - 1, + .apic = 1, + /* use highest bit to detect overflow */ + .max_period = (1ULL << 47) - 1, + .get_event_constraints = amd_get_event_constraints, + .put_event_constraints = amd_put_event_constraints, + + .cpu_prepare = amd_pmu_cpu_prepare, + .cpu_starting = amd_pmu_cpu_starting, + .cpu_dead = amd_pmu_cpu_dead, +}; + +static __init int amd_pmu_init(void) +{ + /* Performance-monitoring supported from K7 and later: */ + if (boot_cpu_data.x86 < 6) + return -ENODEV; + + x86_pmu = amd_pmu; + + /* Events are common for all AMDs */ + memcpy(hw_cache_event_ids, amd_hw_cache_event_ids, + sizeof(hw_cache_event_ids)); + + return 0; +} + +#else /* CONFIG_CPU_SUP_AMD */ + +static int amd_pmu_init(void) +{ + return 0; +} + +#endif diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c new file mode 100644 index 00000000000..fdbc652d3fe --- /dev/null +++ b/arch/x86/kernel/cpu/perf_event_intel.c @@ -0,0 +1,1007 @@ +#ifdef CONFIG_CPU_SUP_INTEL + +/* + * Intel PerfMon, used on Core and later. + */ +static const u64 intel_perfmon_event_map[] = +{ + [PERF_COUNT_HW_CPU_CYCLES] = 0x003c, + [PERF_COUNT_HW_INSTRUCTIONS] = 0x00c0, + [PERF_COUNT_HW_CACHE_REFERENCES] = 0x4f2e, + [PERF_COUNT_HW_CACHE_MISSES] = 0x412e, + [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = 0x00c4, + [PERF_COUNT_HW_BRANCH_MISSES] = 0x00c5, + [PERF_COUNT_HW_BUS_CYCLES] = 0x013c, +}; + +static struct event_constraint intel_core_event_constraints[] = +{ + INTEL_EVENT_CONSTRAINT(0x11, 0x2), /* FP_ASSIST */ + INTEL_EVENT_CONSTRAINT(0x12, 0x2), /* MUL */ + INTEL_EVENT_CONSTRAINT(0x13, 0x2), /* DIV */ + INTEL_EVENT_CONSTRAINT(0x14, 0x1), /* CYCLES_DIV_BUSY */ + INTEL_EVENT_CONSTRAINT(0x19, 0x2), /* DELAYED_BYPASS */ + INTEL_EVENT_CONSTRAINT(0xc1, 0x1), /* FP_COMP_INSTR_RET */ + EVENT_CONSTRAINT_END +}; + +static struct event_constraint intel_core2_event_constraints[] = +{ + FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */ + FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */ + /* + * Core2 has Fixed Counter 2 listed as CPU_CLK_UNHALTED.REF and event + * 0x013c as CPU_CLK_UNHALTED.BUS and specifies there is a fixed + * ratio between these counters. + */ + /* FIXED_EVENT_CONSTRAINT(0x013c, 2), CPU_CLK_UNHALTED.REF */ + INTEL_EVENT_CONSTRAINT(0x10, 0x1), /* FP_COMP_OPS_EXE */ + INTEL_EVENT_CONSTRAINT(0x11, 0x2), /* FP_ASSIST */ + INTEL_EVENT_CONSTRAINT(0x12, 0x2), /* MUL */ + INTEL_EVENT_CONSTRAINT(0x13, 0x2), /* DIV */ + INTEL_EVENT_CONSTRAINT(0x14, 0x1), /* CYCLES_DIV_BUSY */ + INTEL_EVENT_CONSTRAINT(0x18, 0x1), /* IDLE_DURING_DIV */ + INTEL_EVENT_CONSTRAINT(0x19, 0x2), /* DELAYED_BYPASS */ + INTEL_EVENT_CONSTRAINT(0xa1, 0x1), /* RS_UOPS_DISPATCH_CYCLES */ + INTEL_EVENT_CONSTRAINT(0xc9, 0x1), /* ITLB_MISS_RETIRED (T30-9) */ + INTEL_EVENT_CONSTRAINT(0xcb, 0x1), /* MEM_LOAD_RETIRED */ + EVENT_CONSTRAINT_END +}; + +static struct event_constraint intel_nehalem_event_constraints[] = +{ + FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */ + FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */ + /* FIXED_EVENT_CONSTRAINT(0x013c, 2), CPU_CLK_UNHALTED.REF */ + INTEL_EVENT_CONSTRAINT(0x40, 0x3), /* L1D_CACHE_LD */ + INTEL_EVENT_CONSTRAINT(0x41, 0x3), /* L1D_CACHE_ST */ + INTEL_EVENT_CONSTRAINT(0x42, 0x3), /* L1D_CACHE_LOCK */ + INTEL_EVENT_CONSTRAINT(0x43, 0x3), /* L1D_ALL_REF */ + INTEL_EVENT_CONSTRAINT(0x48, 0x3), /* L1D_PEND_MISS */ + INTEL_EVENT_CONSTRAINT(0x4e, 0x3), /* L1D_PREFETCH */ + INTEL_EVENT_CONSTRAINT(0x51, 0x3), /* L1D */ + INTEL_EVENT_CONSTRAINT(0x63, 0x3), /* CACHE_LOCK_CYCLES */ + EVENT_CONSTRAINT_END +}; + +static struct event_constraint intel_westmere_event_constraints[] = +{ + FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */ + FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */ + /* FIXED_EVENT_CONSTRAINT(0x013c, 2), CPU_CLK_UNHALTED.REF */ + INTEL_EVENT_CONSTRAINT(0x51, 0x3), /* L1D */ + INTEL_EVENT_CONSTRAINT(0x60, 0x1), /* OFFCORE_REQUESTS_OUTSTANDING */ + INTEL_EVENT_CONSTRAINT(0x63, 0x3), /* CACHE_LOCK_CYCLES */ + EVENT_CONSTRAINT_END +}; + +static struct event_constraint intel_gen_event_constraints[] = +{ + FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */ + FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */ + /* FIXED_EVENT_CONSTRAINT(0x013c, 2), CPU_CLK_UNHALTED.REF */ + EVENT_CONSTRAINT_END +}; + +static u64 intel_pmu_event_map(int hw_event) +{ + return intel_perfmon_event_map[hw_event]; +} + +static __initconst const u64 westmere_hw_cache_event_ids + [PERF_COUNT_HW_CACHE_MAX] + [PERF_COUNT_HW_CACHE_OP_MAX] + [PERF_COUNT_HW_CACHE_RESULT_MAX] = +{ + [ C(L1D) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x010b, /* MEM_INST_RETIRED.LOADS */ + [ C(RESULT_MISS) ] = 0x0151, /* L1D.REPL */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x020b, /* MEM_INST_RETURED.STORES */ + [ C(RESULT_MISS) ] = 0x0251, /* L1D.M_REPL */ + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0x014e, /* L1D_PREFETCH.REQUESTS */ + [ C(RESULT_MISS) ] = 0x024e, /* L1D_PREFETCH.MISS */ + }, + }, + [ C(L1I ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0380, /* L1I.READS */ + [ C(RESULT_MISS) ] = 0x0280, /* L1I.MISSES */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0x0, + [ C(RESULT_MISS) ] = 0x0, + }, + }, + [ C(LL ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0324, /* L2_RQSTS.LOADS */ + [ C(RESULT_MISS) ] = 0x0224, /* L2_RQSTS.LD_MISS */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x0c24, /* L2_RQSTS.RFOS */ + [ C(RESULT_MISS) ] = 0x0824, /* L2_RQSTS.RFO_MISS */ + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0x4f2e, /* LLC Reference */ + [ C(RESULT_MISS) ] = 0x412e, /* LLC Misses */ + }, + }, + [ C(DTLB) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x010b, /* MEM_INST_RETIRED.LOADS */ + [ C(RESULT_MISS) ] = 0x0108, /* DTLB_LOAD_MISSES.ANY */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x020b, /* MEM_INST_RETURED.STORES */ + [ C(RESULT_MISS) ] = 0x010c, /* MEM_STORE_RETIRED.DTLB_MISS */ + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0x0, + [ C(RESULT_MISS) ] = 0x0, + }, + }, + [ C(ITLB) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x01c0, /* INST_RETIRED.ANY_P */ + [ C(RESULT_MISS) ] = 0x0185, /* ITLB_MISSES.ANY */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + }, + [ C(BPU ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x00c4, /* BR_INST_RETIRED.ALL_BRANCHES */ + [ C(RESULT_MISS) ] = 0x03e8, /* BPU_CLEARS.ANY */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + }, +}; + +static __initconst const u64 nehalem_hw_cache_event_ids + [PERF_COUNT_HW_CACHE_MAX] + [PERF_COUNT_HW_CACHE_OP_MAX] + [PERF_COUNT_HW_CACHE_RESULT_MAX] = +{ + [ C(L1D) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0f40, /* L1D_CACHE_LD.MESI */ + [ C(RESULT_MISS) ] = 0x0140, /* L1D_CACHE_LD.I_STATE */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x0f41, /* L1D_CACHE_ST.MESI */ + [ C(RESULT_MISS) ] = 0x0141, /* L1D_CACHE_ST.I_STATE */ + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0x014e, /* L1D_PREFETCH.REQUESTS */ + [ C(RESULT_MISS) ] = 0x024e, /* L1D_PREFETCH.MISS */ + }, + }, + [ C(L1I ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0380, /* L1I.READS */ + [ C(RESULT_MISS) ] = 0x0280, /* L1I.MISSES */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0x0, + [ C(RESULT_MISS) ] = 0x0, + }, + }, + [ C(LL ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0324, /* L2_RQSTS.LOADS */ + [ C(RESULT_MISS) ] = 0x0224, /* L2_RQSTS.LD_MISS */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x0c24, /* L2_RQSTS.RFOS */ + [ C(RESULT_MISS) ] = 0x0824, /* L2_RQSTS.RFO_MISS */ + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0x4f2e, /* LLC Reference */ + [ C(RESULT_MISS) ] = 0x412e, /* LLC Misses */ + }, + }, + [ C(DTLB) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0f40, /* L1D_CACHE_LD.MESI (alias) */ + [ C(RESULT_MISS) ] = 0x0108, /* DTLB_LOAD_MISSES.ANY */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x0f41, /* L1D_CACHE_ST.MESI (alias) */ + [ C(RESULT_MISS) ] = 0x010c, /* MEM_STORE_RETIRED.DTLB_MISS */ + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0x0, + [ C(RESULT_MISS) ] = 0x0, + }, + }, + [ C(ITLB) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x01c0, /* INST_RETIRED.ANY_P */ + [ C(RESULT_MISS) ] = 0x20c8, /* ITLB_MISS_RETIRED */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + }, + [ C(BPU ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x00c4, /* BR_INST_RETIRED.ALL_BRANCHES */ + [ C(RESULT_MISS) ] = 0x03e8, /* BPU_CLEARS.ANY */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + }, +}; + +static __initconst const u64 core2_hw_cache_event_ids + [PERF_COUNT_HW_CACHE_MAX] + [PERF_COUNT_HW_CACHE_OP_MAX] + [PERF_COUNT_HW_CACHE_RESULT_MAX] = +{ + [ C(L1D) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0f40, /* L1D_CACHE_LD.MESI */ + [ C(RESULT_MISS) ] = 0x0140, /* L1D_CACHE_LD.I_STATE */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x0f41, /* L1D_CACHE_ST.MESI */ + [ C(RESULT_MISS) ] = 0x0141, /* L1D_CACHE_ST.I_STATE */ + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0x104e, /* L1D_PREFETCH.REQUESTS */ + [ C(RESULT_MISS) ] = 0, + }, + }, + [ C(L1I ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0080, /* L1I.READS */ + [ C(RESULT_MISS) ] = 0x0081, /* L1I.MISSES */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0, + [ C(RESULT_MISS) ] = 0, + }, + }, + [ C(LL ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x4f29, /* L2_LD.MESI */ + [ C(RESULT_MISS) ] = 0x4129, /* L2_LD.ISTATE */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x4f2A, /* L2_ST.MESI */ + [ C(RESULT_MISS) ] = 0x412A, /* L2_ST.ISTATE */ + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0, + [ C(RESULT_MISS) ] = 0, + }, + }, + [ C(DTLB) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0f40, /* L1D_CACHE_LD.MESI (alias) */ + [ C(RESULT_MISS) ] = 0x0208, /* DTLB_MISSES.MISS_LD */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x0f41, /* L1D_CACHE_ST.MESI (alias) */ + [ C(RESULT_MISS) ] = 0x0808, /* DTLB_MISSES.MISS_ST */ + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0, + [ C(RESULT_MISS) ] = 0, + }, + }, + [ C(ITLB) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x00c0, /* INST_RETIRED.ANY_P */ + [ C(RESULT_MISS) ] = 0x1282, /* ITLBMISSES */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + }, + [ C(BPU ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x00c4, /* BR_INST_RETIRED.ANY */ + [ C(RESULT_MISS) ] = 0x00c5, /* BP_INST_RETIRED.MISPRED */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + }, +}; + +static __initconst const u64 atom_hw_cache_event_ids + [PERF_COUNT_HW_CACHE_MAX] + [PERF_COUNT_HW_CACHE_OP_MAX] + [PERF_COUNT_HW_CACHE_RESULT_MAX] = +{ + [ C(L1D) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x2140, /* L1D_CACHE.LD */ + [ C(RESULT_MISS) ] = 0, + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x2240, /* L1D_CACHE.ST */ + [ C(RESULT_MISS) ] = 0, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0x0, + [ C(RESULT_MISS) ] = 0, + }, + }, + [ C(L1I ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0380, /* L1I.READS */ + [ C(RESULT_MISS) ] = 0x0280, /* L1I.MISSES */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0, + [ C(RESULT_MISS) ] = 0, + }, + }, + [ C(LL ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x4f29, /* L2_LD.MESI */ + [ C(RESULT_MISS) ] = 0x4129, /* L2_LD.ISTATE */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x4f2A, /* L2_ST.MESI */ + [ C(RESULT_MISS) ] = 0x412A, /* L2_ST.ISTATE */ + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0, + [ C(RESULT_MISS) ] = 0, + }, + }, + [ C(DTLB) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x2140, /* L1D_CACHE_LD.MESI (alias) */ + [ C(RESULT_MISS) ] = 0x0508, /* DTLB_MISSES.MISS_LD */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x2240, /* L1D_CACHE_ST.MESI (alias) */ + [ C(RESULT_MISS) ] = 0x0608, /* DTLB_MISSES.MISS_ST */ + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = 0, + [ C(RESULT_MISS) ] = 0, + }, + }, + [ C(ITLB) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x00c0, /* INST_RETIRED.ANY_P */ + [ C(RESULT_MISS) ] = 0x0282, /* ITLB.MISSES */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + }, + [ C(BPU ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x00c4, /* BR_INST_RETIRED.ANY */ + [ C(RESULT_MISS) ] = 0x00c5, /* BP_INST_RETIRED.MISPRED */ + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + }, +}; + +static void intel_pmu_disable_all(void) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + + wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0); + + if (test_bit(X86_PMC_IDX_FIXED_BTS, cpuc->active_mask)) + intel_pmu_disable_bts(); + + intel_pmu_pebs_disable_all(); + intel_pmu_lbr_disable_all(); +} + +static void intel_pmu_enable_all(int added) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + + intel_pmu_pebs_enable_all(); + intel_pmu_lbr_enable_all(); + wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, x86_pmu.intel_ctrl); + + if (test_bit(X86_PMC_IDX_FIXED_BTS, cpuc->active_mask)) { + struct perf_event *event = + cpuc->events[X86_PMC_IDX_FIXED_BTS]; + + if (WARN_ON_ONCE(!event)) + return; + + intel_pmu_enable_bts(event->hw.config); + } +} + +/* + * Workaround for: + * Intel Errata AAK100 (model 26) + * Intel Errata AAP53 (model 30) + * Intel Errata BD53 (model 44) + * + * These chips need to be 'reset' when adding counters by programming + * the magic three (non counting) events 0x4300D2, 0x4300B1 and 0x4300B5 + * either in sequence on the same PMC or on different PMCs. + */ +static void intel_pmu_nhm_enable_all(int added) +{ + if (added) { + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + int i; + + wrmsrl(MSR_ARCH_PERFMON_EVENTSEL0 + 0, 0x4300D2); + wrmsrl(MSR_ARCH_PERFMON_EVENTSEL0 + 1, 0x4300B1); + wrmsrl(MSR_ARCH_PERFMON_EVENTSEL0 + 2, 0x4300B5); + + wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0x3); + wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0x0); + + for (i = 0; i < 3; i++) { + struct perf_event *event = cpuc->events[i]; + + if (!event) + continue; + + __x86_pmu_enable_event(&event->hw, + ARCH_PERFMON_EVENTSEL_ENABLE); + } + } + intel_pmu_enable_all(added); +} + +static inline u64 intel_pmu_get_status(void) +{ + u64 status; + + rdmsrl(MSR_CORE_PERF_GLOBAL_STATUS, status); + + return status; +} + +static inline void intel_pmu_ack_status(u64 ack) +{ + wrmsrl(MSR_CORE_PERF_GLOBAL_OVF_CTRL, ack); +} + +static void intel_pmu_disable_fixed(struct hw_perf_event *hwc) +{ + int idx = hwc->idx - X86_PMC_IDX_FIXED; + u64 ctrl_val, mask; + + mask = 0xfULL << (idx * 4); + + rdmsrl(hwc->config_base, ctrl_val); + ctrl_val &= ~mask; + wrmsrl(hwc->config_base, ctrl_val); +} + +static void intel_pmu_disable_event(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + + if (unlikely(hwc->idx == X86_PMC_IDX_FIXED_BTS)) { + intel_pmu_disable_bts(); + intel_pmu_drain_bts_buffer(); + return; + } + + if (unlikely(hwc->config_base == MSR_ARCH_PERFMON_FIXED_CTR_CTRL)) { + intel_pmu_disable_fixed(hwc); + return; + } + + x86_pmu_disable_event(event); + + if (unlikely(event->attr.precise_ip)) + intel_pmu_pebs_disable(event); +} + +static void intel_pmu_enable_fixed(struct hw_perf_event *hwc) +{ + int idx = hwc->idx - X86_PMC_IDX_FIXED; + u64 ctrl_val, bits, mask; + + /* + * Enable IRQ generation (0x8), + * and enable ring-3 counting (0x2) and ring-0 counting (0x1) + * if requested: + */ + bits = 0x8ULL; + if (hwc->config & ARCH_PERFMON_EVENTSEL_USR) + bits |= 0x2; + if (hwc->config & ARCH_PERFMON_EVENTSEL_OS) + bits |= 0x1; + + /* + * ANY bit is supported in v3 and up + */ + if (x86_pmu.version > 2 && hwc->config & ARCH_PERFMON_EVENTSEL_ANY) + bits |= 0x4; + + bits <<= (idx * 4); + mask = 0xfULL << (idx * 4); + + rdmsrl(hwc->config_base, ctrl_val); + ctrl_val &= ~mask; + ctrl_val |= bits; + wrmsrl(hwc->config_base, ctrl_val); +} + +static void intel_pmu_enable_event(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + + if (unlikely(hwc->idx == X86_PMC_IDX_FIXED_BTS)) { + if (!__get_cpu_var(cpu_hw_events).enabled) + return; + + intel_pmu_enable_bts(hwc->config); + return; + } + + if (unlikely(hwc->config_base == MSR_ARCH_PERFMON_FIXED_CTR_CTRL)) { + intel_pmu_enable_fixed(hwc); + return; + } + + if (unlikely(event->attr.precise_ip)) + intel_pmu_pebs_enable(event); + + __x86_pmu_enable_event(hwc, ARCH_PERFMON_EVENTSEL_ENABLE); +} + +/* + * Save and restart an expired event. Called by NMI contexts, + * so it has to be careful about preempting normal event ops: + */ +static int intel_pmu_save_and_restart(struct perf_event *event) +{ + x86_perf_event_update(event); + return x86_perf_event_set_period(event); +} + +static void intel_pmu_reset(void) +{ + struct debug_store *ds = __get_cpu_var(cpu_hw_events).ds; + unsigned long flags; + int idx; + + if (!x86_pmu.num_counters) + return; + + local_irq_save(flags); + + printk("clearing PMU state on CPU#%d\n", smp_processor_id()); + + for (idx = 0; idx < x86_pmu.num_counters; idx++) { + checking_wrmsrl(x86_pmu.eventsel + idx, 0ull); + checking_wrmsrl(x86_pmu.perfctr + idx, 0ull); + } + for (idx = 0; idx < x86_pmu.num_counters_fixed; idx++) + checking_wrmsrl(MSR_ARCH_PERFMON_FIXED_CTR0 + idx, 0ull); + + if (ds) + ds->bts_index = ds->bts_buffer_base; + + local_irq_restore(flags); +} + +/* + * This handler is triggered by the local APIC, so the APIC IRQ handling + * rules apply: + */ +static int intel_pmu_handle_irq(struct pt_regs *regs) +{ + struct perf_sample_data data; + struct cpu_hw_events *cpuc; + int bit, loops; + u64 ack, status; + + perf_sample_data_init(&data, 0); + + cpuc = &__get_cpu_var(cpu_hw_events); + + intel_pmu_disable_all(); + intel_pmu_drain_bts_buffer(); + status = intel_pmu_get_status(); + if (!status) { + intel_pmu_enable_all(0); + return 0; + } + + loops = 0; +again: + if (++loops > 100) { + WARN_ONCE(1, "perfevents: irq loop stuck!\n"); + perf_event_print_debug(); + intel_pmu_reset(); + goto done; + } + + inc_irq_stat(apic_perf_irqs); + ack = status; + + intel_pmu_lbr_read(); + + /* + * PEBS overflow sets bit 62 in the global status register + */ + if (__test_and_clear_bit(62, (unsigned long *)&status)) + x86_pmu.drain_pebs(regs); + + for_each_set_bit(bit, (unsigned long *)&status, X86_PMC_IDX_MAX) { + struct perf_event *event = cpuc->events[bit]; + + if (!test_bit(bit, cpuc->active_mask)) + continue; + + if (!intel_pmu_save_and_restart(event)) + continue; + + data.period = event->hw.last_period; + + if (perf_event_overflow(event, 1, &data, regs)) + x86_pmu_stop(event); + } + + intel_pmu_ack_status(ack); + + /* + * Repeat if there is more work to be done: + */ + status = intel_pmu_get_status(); + if (status) + goto again; + +done: + intel_pmu_enable_all(0); + return 1; +} + +static struct event_constraint * +intel_bts_constraints(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + unsigned int hw_event, bts_event; + + hw_event = hwc->config & INTEL_ARCH_EVENT_MASK; + bts_event = x86_pmu.event_map(PERF_COUNT_HW_BRANCH_INSTRUCTIONS); + + if (unlikely(hw_event == bts_event && hwc->sample_period == 1)) + return &bts_constraint; + + return NULL; +} + +static struct event_constraint * +intel_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event) +{ + struct event_constraint *c; + + c = intel_bts_constraints(event); + if (c) + return c; + + c = intel_pebs_constraints(event); + if (c) + return c; + + return x86_get_event_constraints(cpuc, event); +} + +static int intel_pmu_hw_config(struct perf_event *event) +{ + int ret = x86_pmu_hw_config(event); + + if (ret) + return ret; + + if (event->attr.type != PERF_TYPE_RAW) + return 0; + + if (!(event->attr.config & ARCH_PERFMON_EVENTSEL_ANY)) + return 0; + + if (x86_pmu.version < 3) + return -EINVAL; + + if (perf_paranoid_cpu() && !capable(CAP_SYS_ADMIN)) + return -EACCES; + + event->hw.config |= ARCH_PERFMON_EVENTSEL_ANY; + + return 0; +} + +static __initconst const struct x86_pmu core_pmu = { + .name = "core", + .handle_irq = x86_pmu_handle_irq, + .disable_all = x86_pmu_disable_all, + .enable_all = x86_pmu_enable_all, + .enable = x86_pmu_enable_event, + .disable = x86_pmu_disable_event, + .hw_config = x86_pmu_hw_config, + .schedule_events = x86_schedule_events, + .eventsel = MSR_ARCH_PERFMON_EVENTSEL0, + .perfctr = MSR_ARCH_PERFMON_PERFCTR0, + .event_map = intel_pmu_event_map, + .max_events = ARRAY_SIZE(intel_perfmon_event_map), + .apic = 1, + /* + * Intel PMCs cannot be accessed sanely above 32 bit width, + * so we install an artificial 1<<31 period regardless of + * the generic event period: + */ + .max_period = (1ULL << 31) - 1, + .get_event_constraints = intel_get_event_constraints, + .event_constraints = intel_core_event_constraints, +}; + +static void intel_pmu_cpu_starting(int cpu) +{ + init_debug_store_on_cpu(cpu); + /* + * Deal with CPUs that don't clear their LBRs on power-up. + */ + intel_pmu_lbr_reset(); +} + +static void intel_pmu_cpu_dying(int cpu) +{ + fini_debug_store_on_cpu(cpu); +} + +static __initconst const struct x86_pmu intel_pmu = { + .name = "Intel", + .handle_irq = intel_pmu_handle_irq, + .disable_all = intel_pmu_disable_all, + .enable_all = intel_pmu_enable_all, + .enable = intel_pmu_enable_event, + .disable = intel_pmu_disable_event, + .hw_config = intel_pmu_hw_config, + .schedule_events = x86_schedule_events, + .eventsel = MSR_ARCH_PERFMON_EVENTSEL0, + .perfctr = MSR_ARCH_PERFMON_PERFCTR0, + .event_map = intel_pmu_event_map, + .max_events = ARRAY_SIZE(intel_perfmon_event_map), + .apic = 1, + /* + * Intel PMCs cannot be accessed sanely above 32 bit width, + * so we install an artificial 1<<31 period regardless of + * the generic event period: + */ + .max_period = (1ULL << 31) - 1, + .get_event_constraints = intel_get_event_constraints, + + .cpu_starting = intel_pmu_cpu_starting, + .cpu_dying = intel_pmu_cpu_dying, +}; + +static void intel_clovertown_quirks(void) +{ + /* + * PEBS is unreliable due to: + * + * AJ67 - PEBS may experience CPL leaks + * AJ68 - PEBS PMI may be delayed by one event + * AJ69 - GLOBAL_STATUS[62] will only be set when DEBUGCTL[12] + * AJ106 - FREEZE_LBRS_ON_PMI doesn't work in combination with PEBS + * + * AJ67 could be worked around by restricting the OS/USR flags. + * AJ69 could be worked around by setting PMU_FREEZE_ON_PMI. + * + * AJ106 could possibly be worked around by not allowing LBR + * usage from PEBS, including the fixup. + * AJ68 could possibly be worked around by always programming + * a pebs_event_reset[0] value and coping with the lost events. + * + * But taken together it might just make sense to not enable PEBS on + * these chips. + */ + printk(KERN_WARNING "PEBS disabled due to CPU errata.\n"); + x86_pmu.pebs = 0; + x86_pmu.pebs_constraints = NULL; +} + +static __init int intel_pmu_init(void) +{ + union cpuid10_edx edx; + union cpuid10_eax eax; + unsigned int unused; + unsigned int ebx; + int version; + + if (!cpu_has(&boot_cpu_data, X86_FEATURE_ARCH_PERFMON)) { + switch (boot_cpu_data.x86) { + case 0x6: + return p6_pmu_init(); + case 0xf: + return p4_pmu_init(); + } + return -ENODEV; + } + + /* + * Check whether the Architectural PerfMon supports + * Branch Misses Retired hw_event or not. + */ + cpuid(10, &eax.full, &ebx, &unused, &edx.full); + if (eax.split.mask_length <= ARCH_PERFMON_BRANCH_MISSES_RETIRED) + return -ENODEV; + + version = eax.split.version_id; + if (version < 2) + x86_pmu = core_pmu; + else + x86_pmu = intel_pmu; + + x86_pmu.version = version; + x86_pmu.num_counters = eax.split.num_counters; + x86_pmu.cntval_bits = eax.split.bit_width; + x86_pmu.cntval_mask = (1ULL << eax.split.bit_width) - 1; + + /* + * Quirk: v2 perfmon does not report fixed-purpose events, so + * assume at least 3 events: + */ + if (version > 1) + x86_pmu.num_counters_fixed = max((int)edx.split.num_counters_fixed, 3); + + /* + * v2 and above have a perf capabilities MSR + */ + if (version > 1) { + u64 capabilities; + + rdmsrl(MSR_IA32_PERF_CAPABILITIES, capabilities); + x86_pmu.intel_cap.capabilities = capabilities; + } + + intel_ds_init(); + + /* + * Install the hw-cache-events table: + */ + switch (boot_cpu_data.x86_model) { + case 14: /* 65 nm core solo/duo, "Yonah" */ + pr_cont("Core events, "); + break; + + case 15: /* original 65 nm celeron/pentium/core2/xeon, "Merom"/"Conroe" */ + x86_pmu.quirks = intel_clovertown_quirks; + case 22: /* single-core 65 nm celeron/core2solo "Merom-L"/"Conroe-L" */ + case 23: /* current 45 nm celeron/core2/xeon "Penryn"/"Wolfdale" */ + case 29: /* six-core 45 nm xeon "Dunnington" */ + memcpy(hw_cache_event_ids, core2_hw_cache_event_ids, + sizeof(hw_cache_event_ids)); + + intel_pmu_lbr_init_core(); + + x86_pmu.event_constraints = intel_core2_event_constraints; + pr_cont("Core2 events, "); + break; + + case 26: /* 45 nm nehalem, "Bloomfield" */ + case 30: /* 45 nm nehalem, "Lynnfield" */ + case 46: /* 45 nm nehalem-ex, "Beckton" */ + memcpy(hw_cache_event_ids, nehalem_hw_cache_event_ids, + sizeof(hw_cache_event_ids)); + + intel_pmu_lbr_init_nhm(); + + x86_pmu.event_constraints = intel_nehalem_event_constraints; + x86_pmu.enable_all = intel_pmu_nhm_enable_all; + pr_cont("Nehalem events, "); + break; + + case 28: /* Atom */ + memcpy(hw_cache_event_ids, atom_hw_cache_event_ids, + sizeof(hw_cache_event_ids)); + + intel_pmu_lbr_init_atom(); + + x86_pmu.event_constraints = intel_gen_event_constraints; + pr_cont("Atom events, "); + break; + + case 37: /* 32 nm nehalem, "Clarkdale" */ + case 44: /* 32 nm nehalem, "Gulftown" */ + memcpy(hw_cache_event_ids, westmere_hw_cache_event_ids, + sizeof(hw_cache_event_ids)); + + intel_pmu_lbr_init_nhm(); + + x86_pmu.event_constraints = intel_westmere_event_constraints; + x86_pmu.enable_all = intel_pmu_nhm_enable_all; + pr_cont("Westmere events, "); + break; + + default: + /* + * default constraints for v2 and up + */ + x86_pmu.event_constraints = intel_gen_event_constraints; + pr_cont("generic architected perfmon, "); + } + return 0; +} + +#else /* CONFIG_CPU_SUP_INTEL */ + +static int intel_pmu_init(void) +{ + return 0; +} + +#endif /* CONFIG_CPU_SUP_INTEL */ diff --git a/arch/x86/kernel/cpu/perf_event_intel_ds.c b/arch/x86/kernel/cpu/perf_event_intel_ds.c new file mode 100644 index 00000000000..18018d1311c --- /dev/null +++ b/arch/x86/kernel/cpu/perf_event_intel_ds.c @@ -0,0 +1,641 @@ +#ifdef CONFIG_CPU_SUP_INTEL + +/* The maximal number of PEBS events: */ +#define MAX_PEBS_EVENTS 4 + +/* The size of a BTS record in bytes: */ +#define BTS_RECORD_SIZE 24 + +#define BTS_BUFFER_SIZE (PAGE_SIZE << 4) +#define PEBS_BUFFER_SIZE PAGE_SIZE + +/* + * pebs_record_32 for p4 and core not supported + +struct pebs_record_32 { + u32 flags, ip; + u32 ax, bc, cx, dx; + u32 si, di, bp, sp; +}; + + */ + +struct pebs_record_core { + u64 flags, ip; + u64 ax, bx, cx, dx; + u64 si, di, bp, sp; + u64 r8, r9, r10, r11; + u64 r12, r13, r14, r15; +}; + +struct pebs_record_nhm { + u64 flags, ip; + u64 ax, bx, cx, dx; + u64 si, di, bp, sp; + u64 r8, r9, r10, r11; + u64 r12, r13, r14, r15; + u64 status, dla, dse, lat; +}; + +/* + * A debug store configuration. + * + * We only support architectures that use 64bit fields. + */ +struct debug_store { + u64 bts_buffer_base; + u64 bts_index; + u64 bts_absolute_maximum; + u64 bts_interrupt_threshold; + u64 pebs_buffer_base; + u64 pebs_index; + u64 pebs_absolute_maximum; + u64 pebs_interrupt_threshold; + u64 pebs_event_reset[MAX_PEBS_EVENTS]; +}; + +static void init_debug_store_on_cpu(int cpu) +{ + struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds; + + if (!ds) + return; + + wrmsr_on_cpu(cpu, MSR_IA32_DS_AREA, + (u32)((u64)(unsigned long)ds), + (u32)((u64)(unsigned long)ds >> 32)); +} + +static void fini_debug_store_on_cpu(int cpu) +{ + if (!per_cpu(cpu_hw_events, cpu).ds) + return; + + wrmsr_on_cpu(cpu, MSR_IA32_DS_AREA, 0, 0); +} + +static void release_ds_buffers(void) +{ + int cpu; + + if (!x86_pmu.bts && !x86_pmu.pebs) + return; + + get_online_cpus(); + + for_each_online_cpu(cpu) + fini_debug_store_on_cpu(cpu); + + for_each_possible_cpu(cpu) { + struct debug_store *ds = per_cpu(cpu_hw_events, cpu).ds; + + if (!ds) + continue; + + per_cpu(cpu_hw_events, cpu).ds = NULL; + + kfree((void *)(unsigned long)ds->pebs_buffer_base); + kfree((void *)(unsigned long)ds->bts_buffer_base); + kfree(ds); + } + + put_online_cpus(); +} + +static int reserve_ds_buffers(void) +{ + int cpu, err = 0; + + if (!x86_pmu.bts && !x86_pmu.pebs) + return 0; + + get_online_cpus(); + + for_each_possible_cpu(cpu) { + struct debug_store *ds; + void *buffer; + int max, thresh; + + err = -ENOMEM; + ds = kzalloc(sizeof(*ds), GFP_KERNEL); + if (unlikely(!ds)) + break; + per_cpu(cpu_hw_events, cpu).ds = ds; + + if (x86_pmu.bts) { + buffer = kzalloc(BTS_BUFFER_SIZE, GFP_KERNEL); + if (unlikely(!buffer)) + break; + + max = BTS_BUFFER_SIZE / BTS_RECORD_SIZE; + thresh = max / 16; + + ds->bts_buffer_base = (u64)(unsigned long)buffer; + ds->bts_index = ds->bts_buffer_base; + ds->bts_absolute_maximum = ds->bts_buffer_base + + max * BTS_RECORD_SIZE; + ds->bts_interrupt_threshold = ds->bts_absolute_maximum - + thresh * BTS_RECORD_SIZE; + } + + if (x86_pmu.pebs) { + buffer = kzalloc(PEBS_BUFFER_SIZE, GFP_KERNEL); + if (unlikely(!buffer)) + break; + + max = PEBS_BUFFER_SIZE / x86_pmu.pebs_record_size; + + ds->pebs_buffer_base = (u64)(unsigned long)buffer; + ds->pebs_index = ds->pebs_buffer_base; + ds->pebs_absolute_maximum = ds->pebs_buffer_base + + max * x86_pmu.pebs_record_size; + /* + * Always use single record PEBS + */ + ds->pebs_interrupt_threshold = ds->pebs_buffer_base + + x86_pmu.pebs_record_size; + } + + err = 0; + } + + if (err) + release_ds_buffers(); + else { + for_each_online_cpu(cpu) + init_debug_store_on_cpu(cpu); + } + + put_online_cpus(); + + return err; +} + +/* + * BTS + */ + +static struct event_constraint bts_constraint = + EVENT_CONSTRAINT(0, 1ULL << X86_PMC_IDX_FIXED_BTS, 0); + +static void intel_pmu_enable_bts(u64 config) +{ + unsigned long debugctlmsr; + + debugctlmsr = get_debugctlmsr(); + + debugctlmsr |= DEBUGCTLMSR_TR; + debugctlmsr |= DEBUGCTLMSR_BTS; + debugctlmsr |= DEBUGCTLMSR_BTINT; + + if (!(config & ARCH_PERFMON_EVENTSEL_OS)) + debugctlmsr |= DEBUGCTLMSR_BTS_OFF_OS; + + if (!(config & ARCH_PERFMON_EVENTSEL_USR)) + debugctlmsr |= DEBUGCTLMSR_BTS_OFF_USR; + + update_debugctlmsr(debugctlmsr); +} + +static void intel_pmu_disable_bts(void) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + unsigned long debugctlmsr; + + if (!cpuc->ds) + return; + + debugctlmsr = get_debugctlmsr(); + + debugctlmsr &= + ~(DEBUGCTLMSR_TR | DEBUGCTLMSR_BTS | DEBUGCTLMSR_BTINT | + DEBUGCTLMSR_BTS_OFF_OS | DEBUGCTLMSR_BTS_OFF_USR); + + update_debugctlmsr(debugctlmsr); +} + +static void intel_pmu_drain_bts_buffer(void) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + struct debug_store *ds = cpuc->ds; + struct bts_record { + u64 from; + u64 to; + u64 flags; + }; + struct perf_event *event = cpuc->events[X86_PMC_IDX_FIXED_BTS]; + struct bts_record *at, *top; + struct perf_output_handle handle; + struct perf_event_header header; + struct perf_sample_data data; + struct pt_regs regs; + + if (!event) + return; + + if (!ds) + return; + + at = (struct bts_record *)(unsigned long)ds->bts_buffer_base; + top = (struct bts_record *)(unsigned long)ds->bts_index; + + if (top <= at) + return; + + ds->bts_index = ds->bts_buffer_base; + + perf_sample_data_init(&data, 0); + data.period = event->hw.last_period; + regs.ip = 0; + + /* + * Prepare a generic sample, i.e. fill in the invariant fields. + * We will overwrite the from and to address before we output + * the sample. + */ + perf_prepare_sample(&header, &data, event, ®s); + + if (perf_output_begin(&handle, event, header.size * (top - at), 1, 1)) + return; + + for (; at < top; at++) { + data.ip = at->from; + data.addr = at->to; + + perf_output_sample(&handle, &header, &data, event); + } + + perf_output_end(&handle); + + /* There's new data available. */ + event->hw.interrupts++; + event->pending_kill = POLL_IN; +} + +/* + * PEBS + */ + +static struct event_constraint intel_core_pebs_events[] = { + PEBS_EVENT_CONSTRAINT(0x00c0, 0x1), /* INSTR_RETIRED.ANY */ + PEBS_EVENT_CONSTRAINT(0xfec1, 0x1), /* X87_OPS_RETIRED.ANY */ + PEBS_EVENT_CONSTRAINT(0x00c5, 0x1), /* BR_INST_RETIRED.MISPRED */ + PEBS_EVENT_CONSTRAINT(0x1fc7, 0x1), /* SIMD_INST_RETURED.ANY */ + PEBS_EVENT_CONSTRAINT(0x01cb, 0x1), /* MEM_LOAD_RETIRED.L1D_MISS */ + PEBS_EVENT_CONSTRAINT(0x02cb, 0x1), /* MEM_LOAD_RETIRED.L1D_LINE_MISS */ + PEBS_EVENT_CONSTRAINT(0x04cb, 0x1), /* MEM_LOAD_RETIRED.L2_MISS */ + PEBS_EVENT_CONSTRAINT(0x08cb, 0x1), /* MEM_LOAD_RETIRED.L2_LINE_MISS */ + PEBS_EVENT_CONSTRAINT(0x10cb, 0x1), /* MEM_LOAD_RETIRED.DTLB_MISS */ + EVENT_CONSTRAINT_END +}; + +static struct event_constraint intel_nehalem_pebs_events[] = { + PEBS_EVENT_CONSTRAINT(0x00c0, 0xf), /* INSTR_RETIRED.ANY */ + PEBS_EVENT_CONSTRAINT(0xfec1, 0xf), /* X87_OPS_RETIRED.ANY */ + PEBS_EVENT_CONSTRAINT(0x00c5, 0xf), /* BR_INST_RETIRED.MISPRED */ + PEBS_EVENT_CONSTRAINT(0x1fc7, 0xf), /* SIMD_INST_RETURED.ANY */ + PEBS_EVENT_CONSTRAINT(0x01cb, 0xf), /* MEM_LOAD_RETIRED.L1D_MISS */ + PEBS_EVENT_CONSTRAINT(0x02cb, 0xf), /* MEM_LOAD_RETIRED.L1D_LINE_MISS */ + PEBS_EVENT_CONSTRAINT(0x04cb, 0xf), /* MEM_LOAD_RETIRED.L2_MISS */ + PEBS_EVENT_CONSTRAINT(0x08cb, 0xf), /* MEM_LOAD_RETIRED.L2_LINE_MISS */ + PEBS_EVENT_CONSTRAINT(0x10cb, 0xf), /* MEM_LOAD_RETIRED.DTLB_MISS */ + EVENT_CONSTRAINT_END +}; + +static struct event_constraint * +intel_pebs_constraints(struct perf_event *event) +{ + struct event_constraint *c; + + if (!event->attr.precise_ip) + return NULL; + + if (x86_pmu.pebs_constraints) { + for_each_event_constraint(c, x86_pmu.pebs_constraints) { + if ((event->hw.config & c->cmask) == c->code) + return c; + } + } + + return &emptyconstraint; +} + +static void intel_pmu_pebs_enable(struct perf_event *event) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + struct hw_perf_event *hwc = &event->hw; + + hwc->config &= ~ARCH_PERFMON_EVENTSEL_INT; + + cpuc->pebs_enabled |= 1ULL << hwc->idx; + WARN_ON_ONCE(cpuc->enabled); + + if (x86_pmu.intel_cap.pebs_trap && event->attr.precise_ip > 1) + intel_pmu_lbr_enable(event); +} + +static void intel_pmu_pebs_disable(struct perf_event *event) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + struct hw_perf_event *hwc = &event->hw; + + cpuc->pebs_enabled &= ~(1ULL << hwc->idx); + if (cpuc->enabled) + wrmsrl(MSR_IA32_PEBS_ENABLE, cpuc->pebs_enabled); + + hwc->config |= ARCH_PERFMON_EVENTSEL_INT; + + if (x86_pmu.intel_cap.pebs_trap && event->attr.precise_ip > 1) + intel_pmu_lbr_disable(event); +} + +static void intel_pmu_pebs_enable_all(void) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + + if (cpuc->pebs_enabled) + wrmsrl(MSR_IA32_PEBS_ENABLE, cpuc->pebs_enabled); +} + +static void intel_pmu_pebs_disable_all(void) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + + if (cpuc->pebs_enabled) + wrmsrl(MSR_IA32_PEBS_ENABLE, 0); +} + +#include <asm/insn.h> + +static inline bool kernel_ip(unsigned long ip) +{ +#ifdef CONFIG_X86_32 + return ip > PAGE_OFFSET; +#else + return (long)ip < 0; +#endif +} + +static int intel_pmu_pebs_fixup_ip(struct pt_regs *regs) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + unsigned long from = cpuc->lbr_entries[0].from; + unsigned long old_to, to = cpuc->lbr_entries[0].to; + unsigned long ip = regs->ip; + + /* + * We don't need to fixup if the PEBS assist is fault like + */ + if (!x86_pmu.intel_cap.pebs_trap) + return 1; + + /* + * No LBR entry, no basic block, no rewinding + */ + if (!cpuc->lbr_stack.nr || !from || !to) + return 0; + + /* + * Basic blocks should never cross user/kernel boundaries + */ + if (kernel_ip(ip) != kernel_ip(to)) + return 0; + + /* + * unsigned math, either ip is before the start (impossible) or + * the basic block is larger than 1 page (sanity) + */ + if ((ip - to) > PAGE_SIZE) + return 0; + + /* + * We sampled a branch insn, rewind using the LBR stack + */ + if (ip == to) { + regs->ip = from; + return 1; + } + + do { + struct insn insn; + u8 buf[MAX_INSN_SIZE]; + void *kaddr; + + old_to = to; + if (!kernel_ip(ip)) { + int bytes, size = MAX_INSN_SIZE; + + bytes = copy_from_user_nmi(buf, (void __user *)to, size); + if (bytes != size) + return 0; + + kaddr = buf; + } else + kaddr = (void *)to; + + kernel_insn_init(&insn, kaddr); + insn_get_length(&insn); + to += insn.length; + } while (to < ip); + + if (to == ip) { + regs->ip = old_to; + return 1; + } + + /* + * Even though we decoded the basic block, the instruction stream + * never matched the given IP, either the TO or the IP got corrupted. + */ + return 0; +} + +static int intel_pmu_save_and_restart(struct perf_event *event); + +static void __intel_pmu_pebs_event(struct perf_event *event, + struct pt_regs *iregs, void *__pebs) +{ + /* + * We cast to pebs_record_core since that is a subset of + * both formats and we don't use the other fields in this + * routine. + */ + struct pebs_record_core *pebs = __pebs; + struct perf_sample_data data; + struct pt_regs regs; + + if (!intel_pmu_save_and_restart(event)) + return; + + perf_sample_data_init(&data, 0); + data.period = event->hw.last_period; + + /* + * We use the interrupt regs as a base because the PEBS record + * does not contain a full regs set, specifically it seems to + * lack segment descriptors, which get used by things like + * user_mode(). + * + * In the simple case fix up only the IP and BP,SP regs, for + * PERF_SAMPLE_IP and PERF_SAMPLE_CALLCHAIN to function properly. + * A possible PERF_SAMPLE_REGS will have to transfer all regs. + */ + regs = *iregs; + regs.ip = pebs->ip; + regs.bp = pebs->bp; + regs.sp = pebs->sp; + + if (event->attr.precise_ip > 1 && intel_pmu_pebs_fixup_ip(®s)) + regs.flags |= PERF_EFLAGS_EXACT; + else + regs.flags &= ~PERF_EFLAGS_EXACT; + + if (perf_event_overflow(event, 1, &data, ®s)) + x86_pmu_stop(event); +} + +static void intel_pmu_drain_pebs_core(struct pt_regs *iregs) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + struct debug_store *ds = cpuc->ds; + struct perf_event *event = cpuc->events[0]; /* PMC0 only */ + struct pebs_record_core *at, *top; + int n; + + if (!ds || !x86_pmu.pebs) + return; + + at = (struct pebs_record_core *)(unsigned long)ds->pebs_buffer_base; + top = (struct pebs_record_core *)(unsigned long)ds->pebs_index; + + /* + * Whatever else happens, drain the thing + */ + ds->pebs_index = ds->pebs_buffer_base; + + if (!test_bit(0, cpuc->active_mask)) + return; + + WARN_ON_ONCE(!event); + + if (!event->attr.precise_ip) + return; + + n = top - at; + if (n <= 0) + return; + + /* + * Should not happen, we program the threshold at 1 and do not + * set a reset value. + */ + WARN_ON_ONCE(n > 1); + at += n - 1; + + __intel_pmu_pebs_event(event, iregs, at); +} + +static void intel_pmu_drain_pebs_nhm(struct pt_regs *iregs) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + struct debug_store *ds = cpuc->ds; + struct pebs_record_nhm *at, *top; + struct perf_event *event = NULL; + u64 status = 0; + int bit, n; + + if (!ds || !x86_pmu.pebs) + return; + + at = (struct pebs_record_nhm *)(unsigned long)ds->pebs_buffer_base; + top = (struct pebs_record_nhm *)(unsigned long)ds->pebs_index; + + ds->pebs_index = ds->pebs_buffer_base; + + n = top - at; + if (n <= 0) + return; + + /* + * Should not happen, we program the threshold at 1 and do not + * set a reset value. + */ + WARN_ON_ONCE(n > MAX_PEBS_EVENTS); + + for ( ; at < top; at++) { + for_each_set_bit(bit, (unsigned long *)&at->status, MAX_PEBS_EVENTS) { + event = cpuc->events[bit]; + if (!test_bit(bit, cpuc->active_mask)) + continue; + + WARN_ON_ONCE(!event); + + if (!event->attr.precise_ip) + continue; + + if (__test_and_set_bit(bit, (unsigned long *)&status)) + continue; + + break; + } + + if (!event || bit >= MAX_PEBS_EVENTS) + continue; + + __intel_pmu_pebs_event(event, iregs, at); + } +} + +/* + * BTS, PEBS probe and setup + */ + +static void intel_ds_init(void) +{ + /* + * No support for 32bit formats + */ + if (!boot_cpu_has(X86_FEATURE_DTES64)) + return; + + x86_pmu.bts = boot_cpu_has(X86_FEATURE_BTS); + x86_pmu.pebs = boot_cpu_has(X86_FEATURE_PEBS); + if (x86_pmu.pebs) { + char pebs_type = x86_pmu.intel_cap.pebs_trap ? '+' : '-'; + int format = x86_pmu.intel_cap.pebs_format; + + switch (format) { + case 0: + printk(KERN_CONT "PEBS fmt0%c, ", pebs_type); + x86_pmu.pebs_record_size = sizeof(struct pebs_record_core); + x86_pmu.drain_pebs = intel_pmu_drain_pebs_core; + x86_pmu.pebs_constraints = intel_core_pebs_events; + break; + + case 1: + printk(KERN_CONT "PEBS fmt1%c, ", pebs_type); + x86_pmu.pebs_record_size = sizeof(struct pebs_record_nhm); + x86_pmu.drain_pebs = intel_pmu_drain_pebs_nhm; + x86_pmu.pebs_constraints = intel_nehalem_pebs_events; + break; + + default: + printk(KERN_CONT "no PEBS fmt%d%c, ", format, pebs_type); + x86_pmu.pebs = 0; + break; + } + } +} + +#else /* CONFIG_CPU_SUP_INTEL */ + +static int reserve_ds_buffers(void) +{ + return 0; +} + +static void release_ds_buffers(void) +{ +} + +#endif /* CONFIG_CPU_SUP_INTEL */ diff --git a/arch/x86/kernel/cpu/perf_event_intel_lbr.c b/arch/x86/kernel/cpu/perf_event_intel_lbr.c new file mode 100644 index 00000000000..d202c1bece1 --- /dev/null +++ b/arch/x86/kernel/cpu/perf_event_intel_lbr.c @@ -0,0 +1,218 @@ +#ifdef CONFIG_CPU_SUP_INTEL + +enum { + LBR_FORMAT_32 = 0x00, + LBR_FORMAT_LIP = 0x01, + LBR_FORMAT_EIP = 0x02, + LBR_FORMAT_EIP_FLAGS = 0x03, +}; + +/* + * We only support LBR implementations that have FREEZE_LBRS_ON_PMI + * otherwise it becomes near impossible to get a reliable stack. + */ + +static void __intel_pmu_lbr_enable(void) +{ + u64 debugctl; + + rdmsrl(MSR_IA32_DEBUGCTLMSR, debugctl); + debugctl |= (DEBUGCTLMSR_LBR | DEBUGCTLMSR_FREEZE_LBRS_ON_PMI); + wrmsrl(MSR_IA32_DEBUGCTLMSR, debugctl); +} + +static void __intel_pmu_lbr_disable(void) +{ + u64 debugctl; + + rdmsrl(MSR_IA32_DEBUGCTLMSR, debugctl); + debugctl &= ~(DEBUGCTLMSR_LBR | DEBUGCTLMSR_FREEZE_LBRS_ON_PMI); + wrmsrl(MSR_IA32_DEBUGCTLMSR, debugctl); +} + +static void intel_pmu_lbr_reset_32(void) +{ + int i; + + for (i = 0; i < x86_pmu.lbr_nr; i++) + wrmsrl(x86_pmu.lbr_from + i, 0); +} + +static void intel_pmu_lbr_reset_64(void) +{ + int i; + + for (i = 0; i < x86_pmu.lbr_nr; i++) { + wrmsrl(x86_pmu.lbr_from + i, 0); + wrmsrl(x86_pmu.lbr_to + i, 0); + } +} + +static void intel_pmu_lbr_reset(void) +{ + if (!x86_pmu.lbr_nr) + return; + + if (x86_pmu.intel_cap.lbr_format == LBR_FORMAT_32) + intel_pmu_lbr_reset_32(); + else + intel_pmu_lbr_reset_64(); +} + +static void intel_pmu_lbr_enable(struct perf_event *event) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + + if (!x86_pmu.lbr_nr) + return; + + WARN_ON_ONCE(cpuc->enabled); + + /* + * Reset the LBR stack if we changed task context to + * avoid data leaks. + */ + + if (event->ctx->task && cpuc->lbr_context != event->ctx) { + intel_pmu_lbr_reset(); + cpuc->lbr_context = event->ctx; + } + + cpuc->lbr_users++; +} + +static void intel_pmu_lbr_disable(struct perf_event *event) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + + if (!x86_pmu.lbr_nr) + return; + + cpuc->lbr_users--; + WARN_ON_ONCE(cpuc->lbr_users < 0); + + if (cpuc->enabled && !cpuc->lbr_users) + __intel_pmu_lbr_disable(); +} + +static void intel_pmu_lbr_enable_all(void) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + + if (cpuc->lbr_users) + __intel_pmu_lbr_enable(); +} + +static void intel_pmu_lbr_disable_all(void) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + + if (cpuc->lbr_users) + __intel_pmu_lbr_disable(); +} + +static inline u64 intel_pmu_lbr_tos(void) +{ + u64 tos; + + rdmsrl(x86_pmu.lbr_tos, tos); + + return tos; +} + +static void intel_pmu_lbr_read_32(struct cpu_hw_events *cpuc) +{ + unsigned long mask = x86_pmu.lbr_nr - 1; + u64 tos = intel_pmu_lbr_tos(); + int i; + + for (i = 0; i < x86_pmu.lbr_nr; i++) { + unsigned long lbr_idx = (tos - i) & mask; + union { + struct { + u32 from; + u32 to; + }; + u64 lbr; + } msr_lastbranch; + + rdmsrl(x86_pmu.lbr_from + lbr_idx, msr_lastbranch.lbr); + + cpuc->lbr_entries[i].from = msr_lastbranch.from; + cpuc->lbr_entries[i].to = msr_lastbranch.to; + cpuc->lbr_entries[i].flags = 0; + } + cpuc->lbr_stack.nr = i; +} + +#define LBR_FROM_FLAG_MISPRED (1ULL << 63) + +/* + * Due to lack of segmentation in Linux the effective address (offset) + * is the same as the linear address, allowing us to merge the LIP and EIP + * LBR formats. + */ +static void intel_pmu_lbr_read_64(struct cpu_hw_events *cpuc) +{ + unsigned long mask = x86_pmu.lbr_nr - 1; + int lbr_format = x86_pmu.intel_cap.lbr_format; + u64 tos = intel_pmu_lbr_tos(); + int i; + + for (i = 0; i < x86_pmu.lbr_nr; i++) { + unsigned long lbr_idx = (tos - i) & mask; + u64 from, to, flags = 0; + + rdmsrl(x86_pmu.lbr_from + lbr_idx, from); + rdmsrl(x86_pmu.lbr_to + lbr_idx, to); + + if (lbr_format == LBR_FORMAT_EIP_FLAGS) { + flags = !!(from & LBR_FROM_FLAG_MISPRED); + from = (u64)((((s64)from) << 1) >> 1); + } + + cpuc->lbr_entries[i].from = from; + cpuc->lbr_entries[i].to = to; + cpuc->lbr_entries[i].flags = flags; + } + cpuc->lbr_stack.nr = i; +} + +static void intel_pmu_lbr_read(void) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + + if (!cpuc->lbr_users) + return; + + if (x86_pmu.intel_cap.lbr_format == LBR_FORMAT_32) + intel_pmu_lbr_read_32(cpuc); + else + intel_pmu_lbr_read_64(cpuc); +} + +static void intel_pmu_lbr_init_core(void) +{ + x86_pmu.lbr_nr = 4; + x86_pmu.lbr_tos = 0x01c9; + x86_pmu.lbr_from = 0x40; + x86_pmu.lbr_to = 0x60; +} + +static void intel_pmu_lbr_init_nhm(void) +{ + x86_pmu.lbr_nr = 16; + x86_pmu.lbr_tos = 0x01c9; + x86_pmu.lbr_from = 0x680; + x86_pmu.lbr_to = 0x6c0; +} + +static void intel_pmu_lbr_init_atom(void) +{ + x86_pmu.lbr_nr = 8; + x86_pmu.lbr_tos = 0x01c9; + x86_pmu.lbr_from = 0x40; + x86_pmu.lbr_to = 0x60; +} + +#endif /* CONFIG_CPU_SUP_INTEL */ diff --git a/arch/x86/kernel/cpu/perf_event_p4.c b/arch/x86/kernel/cpu/perf_event_p4.c new file mode 100644 index 00000000000..ae85d69644d --- /dev/null +++ b/arch/x86/kernel/cpu/perf_event_p4.c @@ -0,0 +1,858 @@ +/* + * Netburst Perfomance Events (P4, old Xeon) + * + * Copyright (C) 2010 Parallels, Inc., Cyrill Gorcunov <gorcunov@openvz.org> + * Copyright (C) 2010 Intel Corporation, Lin Ming <ming.m.lin@intel.com> + * + * For licencing details see kernel-base/COPYING + */ + +#ifdef CONFIG_CPU_SUP_INTEL + +#include <asm/perf_event_p4.h> + +#define P4_CNTR_LIMIT 3 +/* + * array indices: 0,1 - HT threads, used with HT enabled cpu + */ +struct p4_event_bind { + unsigned int opcode; /* Event code and ESCR selector */ + unsigned int escr_msr[2]; /* ESCR MSR for this event */ + char cntr[2][P4_CNTR_LIMIT]; /* counter index (offset), -1 on abscence */ +}; + +struct p4_cache_event_bind { + unsigned int metric_pebs; + unsigned int metric_vert; +}; + +#define P4_GEN_CACHE_EVENT_BIND(name) \ + [P4_CACHE__##name] = { \ + .metric_pebs = P4_PEBS__##name, \ + .metric_vert = P4_VERT__##name, \ + } + +static struct p4_cache_event_bind p4_cache_event_bind_map[] = { + P4_GEN_CACHE_EVENT_BIND(1stl_cache_load_miss_retired), + P4_GEN_CACHE_EVENT_BIND(2ndl_cache_load_miss_retired), + P4_GEN_CACHE_EVENT_BIND(dtlb_load_miss_retired), + P4_GEN_CACHE_EVENT_BIND(dtlb_store_miss_retired), +}; + +/* + * Note that we don't use CCCR1 here, there is an + * exception for P4_BSQ_ALLOCATION but we just have + * no workaround + * + * consider this binding as resources which particular + * event may borrow, it doesn't contain EventMask, + * Tags and friends -- they are left to a caller + */ +static struct p4_event_bind p4_event_bind_map[] = { + [P4_EVENT_TC_DELIVER_MODE] = { + .opcode = P4_OPCODE(P4_EVENT_TC_DELIVER_MODE), + .escr_msr = { MSR_P4_TC_ESCR0, MSR_P4_TC_ESCR1 }, + .cntr = { {4, 5, -1}, {6, 7, -1} }, + }, + [P4_EVENT_BPU_FETCH_REQUEST] = { + .opcode = P4_OPCODE(P4_EVENT_BPU_FETCH_REQUEST), + .escr_msr = { MSR_P4_BPU_ESCR0, MSR_P4_BPU_ESCR1 }, + .cntr = { {0, -1, -1}, {2, -1, -1} }, + }, + [P4_EVENT_ITLB_REFERENCE] = { + .opcode = P4_OPCODE(P4_EVENT_ITLB_REFERENCE), + .escr_msr = { MSR_P4_ITLB_ESCR0, MSR_P4_ITLB_ESCR1 }, + .cntr = { {0, -1, -1}, {2, -1, -1} }, + }, + [P4_EVENT_MEMORY_CANCEL] = { + .opcode = P4_OPCODE(P4_EVENT_MEMORY_CANCEL), + .escr_msr = { MSR_P4_DAC_ESCR0, MSR_P4_DAC_ESCR1 }, + .cntr = { {8, 9, -1}, {10, 11, -1} }, + }, + [P4_EVENT_MEMORY_COMPLETE] = { + .opcode = P4_OPCODE(P4_EVENT_MEMORY_COMPLETE), + .escr_msr = { MSR_P4_SAAT_ESCR0 , MSR_P4_SAAT_ESCR1 }, + .cntr = { {8, 9, -1}, {10, 11, -1} }, + }, + [P4_EVENT_LOAD_PORT_REPLAY] = { + .opcode = P4_OPCODE(P4_EVENT_LOAD_PORT_REPLAY), + .escr_msr = { MSR_P4_SAAT_ESCR0, MSR_P4_SAAT_ESCR1 }, + .cntr = { {8, 9, -1}, {10, 11, -1} }, + }, + [P4_EVENT_STORE_PORT_REPLAY] = { + .opcode = P4_OPCODE(P4_EVENT_STORE_PORT_REPLAY), + .escr_msr = { MSR_P4_SAAT_ESCR0 , MSR_P4_SAAT_ESCR1 }, + .cntr = { {8, 9, -1}, {10, 11, -1} }, + }, + [P4_EVENT_MOB_LOAD_REPLAY] = { + .opcode = P4_OPCODE(P4_EVENT_MOB_LOAD_REPLAY), + .escr_msr = { MSR_P4_MOB_ESCR0, MSR_P4_MOB_ESCR1 }, + .cntr = { {0, -1, -1}, {2, -1, -1} }, + }, + [P4_EVENT_PAGE_WALK_TYPE] = { + .opcode = P4_OPCODE(P4_EVENT_PAGE_WALK_TYPE), + .escr_msr = { MSR_P4_PMH_ESCR0, MSR_P4_PMH_ESCR1 }, + .cntr = { {0, -1, -1}, {2, -1, -1} }, + }, + [P4_EVENT_BSQ_CACHE_REFERENCE] = { + .opcode = P4_OPCODE(P4_EVENT_BSQ_CACHE_REFERENCE), + .escr_msr = { MSR_P4_BSU_ESCR0, MSR_P4_BSU_ESCR1 }, + .cntr = { {0, -1, -1}, {2, -1, -1} }, + }, + [P4_EVENT_IOQ_ALLOCATION] = { + .opcode = P4_OPCODE(P4_EVENT_IOQ_ALLOCATION), + .escr_msr = { MSR_P4_FSB_ESCR0, MSR_P4_FSB_ESCR1 }, + .cntr = { {0, -1, -1}, {2, -1, -1} }, + }, + [P4_EVENT_IOQ_ACTIVE_ENTRIES] = { /* shared ESCR */ + .opcode = P4_OPCODE(P4_EVENT_IOQ_ACTIVE_ENTRIES), + .escr_msr = { MSR_P4_FSB_ESCR1, MSR_P4_FSB_ESCR1 }, + .cntr = { {2, -1, -1}, {3, -1, -1} }, + }, + [P4_EVENT_FSB_DATA_ACTIVITY] = { + .opcode = P4_OPCODE(P4_EVENT_FSB_DATA_ACTIVITY), + .escr_msr = { MSR_P4_FSB_ESCR0, MSR_P4_FSB_ESCR1 }, + .cntr = { {0, -1, -1}, {2, -1, -1} }, + }, + [P4_EVENT_BSQ_ALLOCATION] = { /* shared ESCR, broken CCCR1 */ + .opcode = P4_OPCODE(P4_EVENT_BSQ_ALLOCATION), + .escr_msr = { MSR_P4_BSU_ESCR0, MSR_P4_BSU_ESCR0 }, + .cntr = { {0, -1, -1}, {1, -1, -1} }, + }, + [P4_EVENT_BSQ_ACTIVE_ENTRIES] = { /* shared ESCR */ + .opcode = P4_OPCODE(P4_EVENT_BSQ_ACTIVE_ENTRIES), + .escr_msr = { MSR_P4_BSU_ESCR1 , MSR_P4_BSU_ESCR1 }, + .cntr = { {2, -1, -1}, {3, -1, -1} }, + }, + [P4_EVENT_SSE_INPUT_ASSIST] = { + .opcode = P4_OPCODE(P4_EVENT_SSE_INPUT_ASSIST), + .escr_msr = { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 }, + .cntr = { {8, 9, -1}, {10, 11, -1} }, + }, + [P4_EVENT_PACKED_SP_UOP] = { + .opcode = P4_OPCODE(P4_EVENT_PACKED_SP_UOP), + .escr_msr = { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 }, + .cntr = { {8, 9, -1}, {10, 11, -1} }, + }, + [P4_EVENT_PACKED_DP_UOP] = { + .opcode = P4_OPCODE(P4_EVENT_PACKED_DP_UOP), + .escr_msr = { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 }, + .cntr = { {8, 9, -1}, {10, 11, -1} }, + }, + [P4_EVENT_SCALAR_SP_UOP] = { + .opcode = P4_OPCODE(P4_EVENT_SCALAR_SP_UOP), + .escr_msr = { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 }, + .cntr = { {8, 9, -1}, {10, 11, -1} }, + }, + [P4_EVENT_SCALAR_DP_UOP] = { + .opcode = P4_OPCODE(P4_EVENT_SCALAR_DP_UOP), + .escr_msr = { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 }, + .cntr = { {8, 9, -1}, {10, 11, -1} }, + }, + [P4_EVENT_64BIT_MMX_UOP] = { + .opcode = P4_OPCODE(P4_EVENT_64BIT_MMX_UOP), + .escr_msr = { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 }, + .cntr = { {8, 9, -1}, {10, 11, -1} }, + }, + [P4_EVENT_128BIT_MMX_UOP] = { + .opcode = P4_OPCODE(P4_EVENT_128BIT_MMX_UOP), + .escr_msr = { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 }, + .cntr = { {8, 9, -1}, {10, 11, -1} }, + }, + [P4_EVENT_X87_FP_UOP] = { + .opcode = P4_OPCODE(P4_EVENT_X87_FP_UOP), + .escr_msr = { MSR_P4_FIRM_ESCR0, MSR_P4_FIRM_ESCR1 }, + .cntr = { {8, 9, -1}, {10, 11, -1} }, + }, + [P4_EVENT_TC_MISC] = { + .opcode = P4_OPCODE(P4_EVENT_TC_MISC), + .escr_msr = { MSR_P4_TC_ESCR0, MSR_P4_TC_ESCR1 }, + .cntr = { {4, 5, -1}, {6, 7, -1} }, + }, + [P4_EVENT_GLOBAL_POWER_EVENTS] = { + .opcode = P4_OPCODE(P4_EVENT_GLOBAL_POWER_EVENTS), + .escr_msr = { MSR_P4_FSB_ESCR0, MSR_P4_FSB_ESCR1 }, + .cntr = { {0, -1, -1}, {2, -1, -1} }, + }, + [P4_EVENT_TC_MS_XFER] = { + .opcode = P4_OPCODE(P4_EVENT_TC_MS_XFER), + .escr_msr = { MSR_P4_MS_ESCR0, MSR_P4_MS_ESCR1 }, + .cntr = { {4, 5, -1}, {6, 7, -1} }, + }, + [P4_EVENT_UOP_QUEUE_WRITES] = { + .opcode = P4_OPCODE(P4_EVENT_UOP_QUEUE_WRITES), + .escr_msr = { MSR_P4_MS_ESCR0, MSR_P4_MS_ESCR1 }, + .cntr = { {4, 5, -1}, {6, 7, -1} }, + }, + [P4_EVENT_RETIRED_MISPRED_BRANCH_TYPE] = { + .opcode = P4_OPCODE(P4_EVENT_RETIRED_MISPRED_BRANCH_TYPE), + .escr_msr = { MSR_P4_TBPU_ESCR0 , MSR_P4_TBPU_ESCR0 }, + .cntr = { {4, 5, -1}, {6, 7, -1} }, + }, + [P4_EVENT_RETIRED_BRANCH_TYPE] = { + .opcode = P4_OPCODE(P4_EVENT_RETIRED_BRANCH_TYPE), + .escr_msr = { MSR_P4_TBPU_ESCR0 , MSR_P4_TBPU_ESCR1 }, + .cntr = { {4, 5, -1}, {6, 7, -1} }, + }, + [P4_EVENT_RESOURCE_STALL] = { + .opcode = P4_OPCODE(P4_EVENT_RESOURCE_STALL), + .escr_msr = { MSR_P4_ALF_ESCR0, MSR_P4_ALF_ESCR1 }, + .cntr = { {12, 13, 16}, {14, 15, 17} }, + }, + [P4_EVENT_WC_BUFFER] = { + .opcode = P4_OPCODE(P4_EVENT_WC_BUFFER), + .escr_msr = { MSR_P4_DAC_ESCR0, MSR_P4_DAC_ESCR1 }, + .cntr = { {8, 9, -1}, {10, 11, -1} }, + }, + [P4_EVENT_B2B_CYCLES] = { + .opcode = P4_OPCODE(P4_EVENT_B2B_CYCLES), + .escr_msr = { MSR_P4_FSB_ESCR0, MSR_P4_FSB_ESCR1 }, + .cntr = { {0, -1, -1}, {2, -1, -1} }, + }, + [P4_EVENT_BNR] = { + .opcode = P4_OPCODE(P4_EVENT_BNR), + .escr_msr = { MSR_P4_FSB_ESCR0, MSR_P4_FSB_ESCR1 }, + .cntr = { {0, -1, -1}, {2, -1, -1} }, + }, + [P4_EVENT_SNOOP] = { + .opcode = P4_OPCODE(P4_EVENT_SNOOP), + .escr_msr = { MSR_P4_FSB_ESCR0, MSR_P4_FSB_ESCR1 }, + .cntr = { {0, -1, -1}, {2, -1, -1} }, + }, + [P4_EVENT_RESPONSE] = { + .opcode = P4_OPCODE(P4_EVENT_RESPONSE), + .escr_msr = { MSR_P4_FSB_ESCR0, MSR_P4_FSB_ESCR1 }, + .cntr = { {0, -1, -1}, {2, -1, -1} }, + }, + [P4_EVENT_FRONT_END_EVENT] = { + .opcode = P4_OPCODE(P4_EVENT_FRONT_END_EVENT), + .escr_msr = { MSR_P4_CRU_ESCR2, MSR_P4_CRU_ESCR3 }, + .cntr = { {12, 13, 16}, {14, 15, 17} }, + }, + [P4_EVENT_EXECUTION_EVENT] = { + .opcode = P4_OPCODE(P4_EVENT_EXECUTION_EVENT), + .escr_msr = { MSR_P4_CRU_ESCR2, MSR_P4_CRU_ESCR3 }, + .cntr = { {12, 13, 16}, {14, 15, 17} }, + }, + [P4_EVENT_REPLAY_EVENT] = { + .opcode = P4_OPCODE(P4_EVENT_REPLAY_EVENT), + .escr_msr = { MSR_P4_CRU_ESCR2, MSR_P4_CRU_ESCR3 }, + .cntr = { {12, 13, 16}, {14, 15, 17} }, + }, + [P4_EVENT_INSTR_RETIRED] = { + .opcode = P4_OPCODE(P4_EVENT_INSTR_RETIRED), + .escr_msr = { MSR_P4_CRU_ESCR0, MSR_P4_CRU_ESCR1 }, + .cntr = { {12, 13, 16}, {14, 15, 17} }, + }, + [P4_EVENT_UOPS_RETIRED] = { + .opcode = P4_OPCODE(P4_EVENT_UOPS_RETIRED), + .escr_msr = { MSR_P4_CRU_ESCR0, MSR_P4_CRU_ESCR1 }, + .cntr = { {12, 13, 16}, {14, 15, 17} }, + }, + [P4_EVENT_UOP_TYPE] = { + .opcode = P4_OPCODE(P4_EVENT_UOP_TYPE), + .escr_msr = { MSR_P4_RAT_ESCR0, MSR_P4_RAT_ESCR1 }, + .cntr = { {12, 13, 16}, {14, 15, 17} }, + }, + [P4_EVENT_BRANCH_RETIRED] = { + .opcode = P4_OPCODE(P4_EVENT_BRANCH_RETIRED), + .escr_msr = { MSR_P4_CRU_ESCR2, MSR_P4_CRU_ESCR3 }, + .cntr = { {12, 13, 16}, {14, 15, 17} }, + }, + [P4_EVENT_MISPRED_BRANCH_RETIRED] = { + .opcode = P4_OPCODE(P4_EVENT_MISPRED_BRANCH_RETIRED), + .escr_msr = { MSR_P4_CRU_ESCR0, MSR_P4_CRU_ESCR1 }, + .cntr = { {12, 13, 16}, {14, 15, 17} }, + }, + [P4_EVENT_X87_ASSIST] = { + .opcode = P4_OPCODE(P4_EVENT_X87_ASSIST), + .escr_msr = { MSR_P4_CRU_ESCR2, MSR_P4_CRU_ESCR3 }, + .cntr = { {12, 13, 16}, {14, 15, 17} }, + }, + [P4_EVENT_MACHINE_CLEAR] = { + .opcode = P4_OPCODE(P4_EVENT_MACHINE_CLEAR), + .escr_msr = { MSR_P4_CRU_ESCR2, MSR_P4_CRU_ESCR3 }, + .cntr = { {12, 13, 16}, {14, 15, 17} }, + }, + [P4_EVENT_INSTR_COMPLETED] = { + .opcode = P4_OPCODE(P4_EVENT_INSTR_COMPLETED), + .escr_msr = { MSR_P4_CRU_ESCR0, MSR_P4_CRU_ESCR1 }, + .cntr = { {12, 13, 16}, {14, 15, 17} }, + }, +}; + +#define P4_GEN_CACHE_EVENT(event, bit, cache_event) \ + p4_config_pack_escr(P4_ESCR_EVENT(event) | \ + P4_ESCR_EMASK_BIT(event, bit)) | \ + p4_config_pack_cccr(cache_event | \ + P4_CCCR_ESEL(P4_OPCODE_ESEL(P4_OPCODE(event)))) + +static __initconst const u64 p4_hw_cache_event_ids + [PERF_COUNT_HW_CACHE_MAX] + [PERF_COUNT_HW_CACHE_OP_MAX] + [PERF_COUNT_HW_CACHE_RESULT_MAX] = +{ + [ C(L1D ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0, + [ C(RESULT_MISS) ] = P4_GEN_CACHE_EVENT(P4_EVENT_REPLAY_EVENT, NBOGUS, + P4_CACHE__1stl_cache_load_miss_retired), + }, + }, + [ C(LL ) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0, + [ C(RESULT_MISS) ] = P4_GEN_CACHE_EVENT(P4_EVENT_REPLAY_EVENT, NBOGUS, + P4_CACHE__2ndl_cache_load_miss_retired), + }, +}, + [ C(DTLB) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = 0x0, + [ C(RESULT_MISS) ] = P4_GEN_CACHE_EVENT(P4_EVENT_REPLAY_EVENT, NBOGUS, + P4_CACHE__dtlb_load_miss_retired), + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = 0x0, + [ C(RESULT_MISS) ] = P4_GEN_CACHE_EVENT(P4_EVENT_REPLAY_EVENT, NBOGUS, + P4_CACHE__dtlb_store_miss_retired), + }, + }, + [ C(ITLB) ] = { + [ C(OP_READ) ] = { + [ C(RESULT_ACCESS) ] = P4_GEN_CACHE_EVENT(P4_EVENT_ITLB_REFERENCE, HIT, + P4_CACHE__itlb_reference_hit), + [ C(RESULT_MISS) ] = P4_GEN_CACHE_EVENT(P4_EVENT_ITLB_REFERENCE, MISS, + P4_CACHE__itlb_reference_miss), + }, + [ C(OP_WRITE) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + [ C(OP_PREFETCH) ] = { + [ C(RESULT_ACCESS) ] = -1, + [ C(RESULT_MISS) ] = -1, + }, + }, +}; + +static u64 p4_general_events[PERF_COUNT_HW_MAX] = { + /* non-halted CPU clocks */ + [PERF_COUNT_HW_CPU_CYCLES] = + p4_config_pack_escr(P4_ESCR_EVENT(P4_EVENT_GLOBAL_POWER_EVENTS) | + P4_ESCR_EMASK_BIT(P4_EVENT_GLOBAL_POWER_EVENTS, RUNNING)), + + /* + * retired instructions + * in a sake of simplicity we don't use the FSB tagging + */ + [PERF_COUNT_HW_INSTRUCTIONS] = + p4_config_pack_escr(P4_ESCR_EVENT(P4_EVENT_INSTR_RETIRED) | + P4_ESCR_EMASK_BIT(P4_EVENT_INSTR_RETIRED, NBOGUSNTAG) | + P4_ESCR_EMASK_BIT(P4_EVENT_INSTR_RETIRED, BOGUSNTAG)), + + /* cache hits */ + [PERF_COUNT_HW_CACHE_REFERENCES] = + p4_config_pack_escr(P4_ESCR_EVENT(P4_EVENT_BSQ_CACHE_REFERENCE) | + P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_2ndL_HITS) | + P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_2ndL_HITE) | + P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_2ndL_HITM) | + P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_3rdL_HITS) | + P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_3rdL_HITE) | + P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_3rdL_HITM)), + + /* cache misses */ + [PERF_COUNT_HW_CACHE_MISSES] = + p4_config_pack_escr(P4_ESCR_EVENT(P4_EVENT_BSQ_CACHE_REFERENCE) | + P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_2ndL_MISS) | + P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, RD_3rdL_MISS) | + P4_ESCR_EMASK_BIT(P4_EVENT_BSQ_CACHE_REFERENCE, WR_2ndL_MISS)), + + /* branch instructions retired */ + [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = + p4_config_pack_escr(P4_ESCR_EVENT(P4_EVENT_RETIRED_BRANCH_TYPE) | + P4_ESCR_EMASK_BIT(P4_EVENT_RETIRED_BRANCH_TYPE, CONDITIONAL) | + P4_ESCR_EMASK_BIT(P4_EVENT_RETIRED_BRANCH_TYPE, CALL) | + P4_ESCR_EMASK_BIT(P4_EVENT_RETIRED_BRANCH_TYPE, RETURN) | + P4_ESCR_EMASK_BIT(P4_EVENT_RETIRED_BRANCH_TYPE, INDIRECT)), + + /* mispredicted branches retired */ + [PERF_COUNT_HW_BRANCH_MISSES] = + p4_config_pack_escr(P4_ESCR_EVENT(P4_EVENT_MISPRED_BRANCH_RETIRED) | + P4_ESCR_EMASK_BIT(P4_EVENT_MISPRED_BRANCH_RETIRED, NBOGUS)), + + /* bus ready clocks (cpu is driving #DRDY_DRV\#DRDY_OWN): */ + [PERF_COUNT_HW_BUS_CYCLES] = + p4_config_pack_escr(P4_ESCR_EVENT(P4_EVENT_FSB_DATA_ACTIVITY) | + P4_ESCR_EMASK_BIT(P4_EVENT_FSB_DATA_ACTIVITY, DRDY_DRV) | + P4_ESCR_EMASK_BIT(P4_EVENT_FSB_DATA_ACTIVITY, DRDY_OWN)) | + p4_config_pack_cccr(P4_CCCR_EDGE | P4_CCCR_COMPARE), +}; + +static struct p4_event_bind *p4_config_get_bind(u64 config) +{ + unsigned int evnt = p4_config_unpack_event(config); + struct p4_event_bind *bind = NULL; + + if (evnt < ARRAY_SIZE(p4_event_bind_map)) + bind = &p4_event_bind_map[evnt]; + + return bind; +} + +static u64 p4_pmu_event_map(int hw_event) +{ + struct p4_event_bind *bind; + unsigned int esel; + u64 config; + + config = p4_general_events[hw_event]; + bind = p4_config_get_bind(config); + esel = P4_OPCODE_ESEL(bind->opcode); + config |= p4_config_pack_cccr(P4_CCCR_ESEL(esel)); + + return config; +} + +static int p4_hw_config(struct perf_event *event) +{ + int cpu = get_cpu(); + int rc = 0; + unsigned int evnt; + u32 escr, cccr; + + /* + * the reason we use cpu that early is that: if we get scheduled + * first time on the same cpu -- we will not need swap thread + * specific flags in config (and will save some cpu cycles) + */ + + cccr = p4_default_cccr_conf(cpu); + escr = p4_default_escr_conf(cpu, event->attr.exclude_kernel, + event->attr.exclude_user); + event->hw.config = p4_config_pack_escr(escr) | + p4_config_pack_cccr(cccr); + + if (p4_ht_active() && p4_ht_thread(cpu)) + event->hw.config = p4_set_ht_bit(event->hw.config); + + if (event->attr.type == PERF_TYPE_RAW) { + + /* user data may have out-of-bound event index */ + evnt = p4_config_unpack_event(event->attr.config); + if (evnt >= ARRAY_SIZE(p4_event_bind_map)) { + rc = -EINVAL; + goto out; + } + + /* + * We don't control raw events so it's up to the caller + * to pass sane values (and we don't count the thread number + * on HT machine but allow HT-compatible specifics to be + * passed on) + * + * XXX: HT wide things should check perf_paranoid_cpu() && + * CAP_SYS_ADMIN + */ + event->hw.config |= event->attr.config & + (p4_config_pack_escr(P4_ESCR_MASK_HT) | + p4_config_pack_cccr(P4_CCCR_MASK_HT)); + } + + rc = x86_setup_perfctr(event); +out: + put_cpu(); + return rc; +} + +static inline int p4_pmu_clear_cccr_ovf(struct hw_perf_event *hwc) +{ + int overflow = 0; + u32 low, high; + + rdmsr(hwc->config_base + hwc->idx, low, high); + + /* we need to check high bit for unflagged overflows */ + if ((low & P4_CCCR_OVF) || !(high & (1 << 31))) { + overflow = 1; + (void)checking_wrmsrl(hwc->config_base + hwc->idx, + ((u64)low) & ~P4_CCCR_OVF); + } + + return overflow; +} + +static inline void p4_pmu_disable_event(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + + /* + * If event gets disabled while counter is in overflowed + * state we need to clear P4_CCCR_OVF, otherwise interrupt get + * asserted again and again + */ + (void)checking_wrmsrl(hwc->config_base + hwc->idx, + (u64)(p4_config_unpack_cccr(hwc->config)) & + ~P4_CCCR_ENABLE & ~P4_CCCR_OVF & ~P4_CCCR_RESERVED); +} + +static void p4_pmu_disable_all(void) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + int idx; + + for (idx = 0; idx < x86_pmu.num_counters; idx++) { + struct perf_event *event = cpuc->events[idx]; + if (!test_bit(idx, cpuc->active_mask)) + continue; + p4_pmu_disable_event(event); + } +} + +static void p4_pmu_enable_event(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + int thread = p4_ht_config_thread(hwc->config); + u64 escr_conf = p4_config_unpack_escr(p4_clear_ht_bit(hwc->config)); + unsigned int idx = p4_config_unpack_event(hwc->config); + unsigned int idx_cache = p4_config_unpack_cache_event(hwc->config); + struct p4_event_bind *bind; + struct p4_cache_event_bind *bind_cache; + u64 escr_addr, cccr; + + bind = &p4_event_bind_map[idx]; + escr_addr = (u64)bind->escr_msr[thread]; + + /* + * - we dont support cascaded counters yet + * - and counter 1 is broken (erratum) + */ + WARN_ON_ONCE(p4_is_event_cascaded(hwc->config)); + WARN_ON_ONCE(hwc->idx == 1); + + /* we need a real Event value */ + escr_conf &= ~P4_ESCR_EVENT_MASK; + escr_conf |= P4_ESCR_EVENT(P4_OPCODE_EVNT(bind->opcode)); + + cccr = p4_config_unpack_cccr(hwc->config); + + /* + * it could be Cache event so that we need to + * set metrics into additional MSRs + */ + BUILD_BUG_ON(P4_CACHE__MAX > P4_CCCR_CACHE_OPS_MASK); + if (idx_cache > P4_CACHE__NONE && + idx_cache < ARRAY_SIZE(p4_cache_event_bind_map)) { + bind_cache = &p4_cache_event_bind_map[idx_cache]; + (void)checking_wrmsrl(MSR_IA32_PEBS_ENABLE, (u64)bind_cache->metric_pebs); + (void)checking_wrmsrl(MSR_P4_PEBS_MATRIX_VERT, (u64)bind_cache->metric_vert); + } + + (void)checking_wrmsrl(escr_addr, escr_conf); + (void)checking_wrmsrl(hwc->config_base + hwc->idx, + (cccr & ~P4_CCCR_RESERVED) | P4_CCCR_ENABLE); +} + +static void p4_pmu_enable_all(int added) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + int idx; + + for (idx = 0; idx < x86_pmu.num_counters; idx++) { + struct perf_event *event = cpuc->events[idx]; + if (!test_bit(idx, cpuc->active_mask)) + continue; + p4_pmu_enable_event(event); + } +} + +static int p4_pmu_handle_irq(struct pt_regs *regs) +{ + struct perf_sample_data data; + struct cpu_hw_events *cpuc; + struct perf_event *event; + struct hw_perf_event *hwc; + int idx, handled = 0; + u64 val; + + data.addr = 0; + data.raw = NULL; + + cpuc = &__get_cpu_var(cpu_hw_events); + + for (idx = 0; idx < x86_pmu.num_counters; idx++) { + + if (!test_bit(idx, cpuc->active_mask)) + continue; + + event = cpuc->events[idx]; + hwc = &event->hw; + + WARN_ON_ONCE(hwc->idx != idx); + + /* it might be unflagged overflow */ + handled = p4_pmu_clear_cccr_ovf(hwc); + + val = x86_perf_event_update(event); + if (!handled && (val & (1ULL << (x86_pmu.cntval_bits - 1)))) + continue; + + /* event overflow for sure */ + data.period = event->hw.last_period; + + if (!x86_perf_event_set_period(event)) + continue; + if (perf_event_overflow(event, 1, &data, regs)) + p4_pmu_disable_event(event); + } + + if (handled) { + /* p4 quirk: unmask it again */ + apic_write(APIC_LVTPC, apic_read(APIC_LVTPC) & ~APIC_LVT_MASKED); + inc_irq_stat(apic_perf_irqs); + } + + return handled; +} + +/* + * swap thread specific fields according to a thread + * we are going to run on + */ +static void p4_pmu_swap_config_ts(struct hw_perf_event *hwc, int cpu) +{ + u32 escr, cccr; + + /* + * we either lucky and continue on same cpu or no HT support + */ + if (!p4_should_swap_ts(hwc->config, cpu)) + return; + + /* + * the event is migrated from an another logical + * cpu, so we need to swap thread specific flags + */ + + escr = p4_config_unpack_escr(hwc->config); + cccr = p4_config_unpack_cccr(hwc->config); + + if (p4_ht_thread(cpu)) { + cccr &= ~P4_CCCR_OVF_PMI_T0; + cccr |= P4_CCCR_OVF_PMI_T1; + if (escr & P4_ESCR_T0_OS) { + escr &= ~P4_ESCR_T0_OS; + escr |= P4_ESCR_T1_OS; + } + if (escr & P4_ESCR_T0_USR) { + escr &= ~P4_ESCR_T0_USR; + escr |= P4_ESCR_T1_USR; + } + hwc->config = p4_config_pack_escr(escr); + hwc->config |= p4_config_pack_cccr(cccr); + hwc->config |= P4_CONFIG_HT; + } else { + cccr &= ~P4_CCCR_OVF_PMI_T1; + cccr |= P4_CCCR_OVF_PMI_T0; + if (escr & P4_ESCR_T1_OS) { + escr &= ~P4_ESCR_T1_OS; + escr |= P4_ESCR_T0_OS; + } + if (escr & P4_ESCR_T1_USR) { + escr &= ~P4_ESCR_T1_USR; + escr |= P4_ESCR_T0_USR; + } + hwc->config = p4_config_pack_escr(escr); + hwc->config |= p4_config_pack_cccr(cccr); + hwc->config &= ~P4_CONFIG_HT; + } +} + +/* + * ESCR address hashing is tricky, ESCRs are not sequential + * in memory but all starts from MSR_P4_BSU_ESCR0 (0x03a0) and + * the metric between any ESCRs is laid in range [0xa0,0xe1] + * + * so we make ~70% filled hashtable + */ + +#define P4_ESCR_MSR_BASE 0x000003a0 +#define P4_ESCR_MSR_MAX 0x000003e1 +#define P4_ESCR_MSR_TABLE_SIZE (P4_ESCR_MSR_MAX - P4_ESCR_MSR_BASE + 1) +#define P4_ESCR_MSR_IDX(msr) (msr - P4_ESCR_MSR_BASE) +#define P4_ESCR_MSR_TABLE_ENTRY(msr) [P4_ESCR_MSR_IDX(msr)] = msr + +static const unsigned int p4_escr_table[P4_ESCR_MSR_TABLE_SIZE] = { + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_ALF_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_ALF_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_BPU_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_BPU_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_BSU_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_BSU_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_CRU_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_CRU_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_CRU_ESCR2), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_CRU_ESCR3), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_CRU_ESCR4), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_CRU_ESCR5), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_DAC_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_DAC_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_FIRM_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_FIRM_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_FLAME_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_FLAME_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_FSB_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_FSB_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_IQ_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_IQ_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_IS_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_IS_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_ITLB_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_ITLB_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_IX_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_IX_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_MOB_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_MOB_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_MS_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_MS_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_PMH_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_PMH_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_RAT_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_RAT_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_SAAT_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_SAAT_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_SSU_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_SSU_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_TBPU_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_TBPU_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_TC_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_TC_ESCR1), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_U2L_ESCR0), + P4_ESCR_MSR_TABLE_ENTRY(MSR_P4_U2L_ESCR1), +}; + +static int p4_get_escr_idx(unsigned int addr) +{ + unsigned int idx = P4_ESCR_MSR_IDX(addr); + + if (unlikely(idx >= P4_ESCR_MSR_TABLE_SIZE || + !p4_escr_table[idx] || + p4_escr_table[idx] != addr)) { + WARN_ONCE(1, "P4 PMU: Wrong address passed: %x\n", addr); + return -1; + } + + return idx; +} + +static int p4_next_cntr(int thread, unsigned long *used_mask, + struct p4_event_bind *bind) +{ + int i, j; + + for (i = 0; i < P4_CNTR_LIMIT; i++) { + j = bind->cntr[thread][i]; + if (j != -1 && !test_bit(j, used_mask)) + return j; + } + + return -1; +} + +static int p4_pmu_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign) +{ + unsigned long used_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)]; + unsigned long escr_mask[BITS_TO_LONGS(P4_ESCR_MSR_TABLE_SIZE)]; + int cpu = smp_processor_id(); + struct hw_perf_event *hwc; + struct p4_event_bind *bind; + unsigned int i, thread, num; + int cntr_idx, escr_idx; + + bitmap_zero(used_mask, X86_PMC_IDX_MAX); + bitmap_zero(escr_mask, P4_ESCR_MSR_TABLE_SIZE); + + for (i = 0, num = n; i < n; i++, num--) { + + hwc = &cpuc->event_list[i]->hw; + thread = p4_ht_thread(cpu); + bind = p4_config_get_bind(hwc->config); + escr_idx = p4_get_escr_idx(bind->escr_msr[thread]); + if (unlikely(escr_idx == -1)) + goto done; + + if (hwc->idx != -1 && !p4_should_swap_ts(hwc->config, cpu)) { + cntr_idx = hwc->idx; + if (assign) + assign[i] = hwc->idx; + goto reserve; + } + + cntr_idx = p4_next_cntr(thread, used_mask, bind); + if (cntr_idx == -1 || test_bit(escr_idx, escr_mask)) + goto done; + + p4_pmu_swap_config_ts(hwc, cpu); + if (assign) + assign[i] = cntr_idx; +reserve: + set_bit(cntr_idx, used_mask); + set_bit(escr_idx, escr_mask); + } + +done: + return num ? -ENOSPC : 0; +} + +static __initconst const struct x86_pmu p4_pmu = { + .name = "Netburst P4/Xeon", + .handle_irq = p4_pmu_handle_irq, + .disable_all = p4_pmu_disable_all, + .enable_all = p4_pmu_enable_all, + .enable = p4_pmu_enable_event, + .disable = p4_pmu_disable_event, + .eventsel = MSR_P4_BPU_CCCR0, + .perfctr = MSR_P4_BPU_PERFCTR0, + .event_map = p4_pmu_event_map, + .max_events = ARRAY_SIZE(p4_general_events), + .get_event_constraints = x86_get_event_constraints, + /* + * IF HT disabled we may need to use all + * ARCH_P4_MAX_CCCR counters simulaneously + * though leave it restricted at moment assuming + * HT is on + */ + .num_counters = ARCH_P4_MAX_CCCR, + .apic = 1, + .cntval_bits = 40, + .cntval_mask = (1ULL << 40) - 1, + .max_period = (1ULL << 39) - 1, + .hw_config = p4_hw_config, + .schedule_events = p4_pmu_schedule_events, +}; + +static __init int p4_pmu_init(void) +{ + unsigned int low, high; + + /* If we get stripped -- indexig fails */ + BUILD_BUG_ON(ARCH_P4_MAX_CCCR > X86_PMC_MAX_GENERIC); + + rdmsr(MSR_IA32_MISC_ENABLE, low, high); + if (!(low & (1 << 7))) { + pr_cont("unsupported Netburst CPU model %d ", + boot_cpu_data.x86_model); + return -ENODEV; + } + + memcpy(hw_cache_event_ids, p4_hw_cache_event_ids, + sizeof(hw_cache_event_ids)); + + pr_cont("Netburst events, "); + + x86_pmu = p4_pmu; + + return 0; +} + +#endif /* CONFIG_CPU_SUP_INTEL */ diff --git a/arch/x86/kernel/cpu/perf_event_p6.c b/arch/x86/kernel/cpu/perf_event_p6.c new file mode 100644 index 00000000000..34ba07be2cd --- /dev/null +++ b/arch/x86/kernel/cpu/perf_event_p6.c @@ -0,0 +1,142 @@ +#ifdef CONFIG_CPU_SUP_INTEL + +/* + * Not sure about some of these + */ +static const u64 p6_perfmon_event_map[] = +{ + [PERF_COUNT_HW_CPU_CYCLES] = 0x0079, + [PERF_COUNT_HW_INSTRUCTIONS] = 0x00c0, + [PERF_COUNT_HW_CACHE_REFERENCES] = 0x0f2e, + [PERF_COUNT_HW_CACHE_MISSES] = 0x012e, + [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = 0x00c4, + [PERF_COUNT_HW_BRANCH_MISSES] = 0x00c5, + [PERF_COUNT_HW_BUS_CYCLES] = 0x0062, +}; + +static u64 p6_pmu_event_map(int hw_event) +{ + return p6_perfmon_event_map[hw_event]; +} + +/* + * Event setting that is specified not to count anything. + * We use this to effectively disable a counter. + * + * L2_RQSTS with 0 MESI unit mask. + */ +#define P6_NOP_EVENT 0x0000002EULL + +static struct event_constraint p6_event_constraints[] = +{ + INTEL_EVENT_CONSTRAINT(0xc1, 0x1), /* FLOPS */ + INTEL_EVENT_CONSTRAINT(0x10, 0x1), /* FP_COMP_OPS_EXE */ + INTEL_EVENT_CONSTRAINT(0x11, 0x1), /* FP_ASSIST */ + INTEL_EVENT_CONSTRAINT(0x12, 0x2), /* MUL */ + INTEL_EVENT_CONSTRAINT(0x13, 0x2), /* DIV */ + INTEL_EVENT_CONSTRAINT(0x14, 0x1), /* CYCLES_DIV_BUSY */ + EVENT_CONSTRAINT_END +}; + +static void p6_pmu_disable_all(void) +{ + u64 val; + + /* p6 only has one enable register */ + rdmsrl(MSR_P6_EVNTSEL0, val); + val &= ~ARCH_PERFMON_EVENTSEL_ENABLE; + wrmsrl(MSR_P6_EVNTSEL0, val); +} + +static void p6_pmu_enable_all(int added) +{ + unsigned long val; + + /* p6 only has one enable register */ + rdmsrl(MSR_P6_EVNTSEL0, val); + val |= ARCH_PERFMON_EVENTSEL_ENABLE; + wrmsrl(MSR_P6_EVNTSEL0, val); +} + +static inline void +p6_pmu_disable_event(struct perf_event *event) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + struct hw_perf_event *hwc = &event->hw; + u64 val = P6_NOP_EVENT; + + if (cpuc->enabled) + val |= ARCH_PERFMON_EVENTSEL_ENABLE; + + (void)checking_wrmsrl(hwc->config_base + hwc->idx, val); +} + +static void p6_pmu_enable_event(struct perf_event *event) +{ + struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events); + struct hw_perf_event *hwc = &event->hw; + u64 val; + + val = hwc->config; + if (cpuc->enabled) + val |= ARCH_PERFMON_EVENTSEL_ENABLE; + + (void)checking_wrmsrl(hwc->config_base + hwc->idx, val); +} + +static __initconst const struct x86_pmu p6_pmu = { + .name = "p6", + .handle_irq = x86_pmu_handle_irq, + .disable_all = p6_pmu_disable_all, + .enable_all = p6_pmu_enable_all, + .enable = p6_pmu_enable_event, + .disable = p6_pmu_disable_event, + .hw_config = x86_pmu_hw_config, + .schedule_events = x86_schedule_events, + .eventsel = MSR_P6_EVNTSEL0, + .perfctr = MSR_P6_PERFCTR0, + .event_map = p6_pmu_event_map, + .max_events = ARRAY_SIZE(p6_perfmon_event_map), + .apic = 1, + .max_period = (1ULL << 31) - 1, + .version = 0, + .num_counters = 2, + /* + * Events have 40 bits implemented. However they are designed such + * that bits [32-39] are sign extensions of bit 31. As such the + * effective width of a event for P6-like PMU is 32 bits only. + * + * See IA-32 Intel Architecture Software developer manual Vol 3B + */ + .cntval_bits = 32, + .cntval_mask = (1ULL << 32) - 1, + .get_event_constraints = x86_get_event_constraints, + .event_constraints = p6_event_constraints, +}; + +static __init int p6_pmu_init(void) +{ + switch (boot_cpu_data.x86_model) { + case 1: + case 3: /* Pentium Pro */ + case 5: + case 6: /* Pentium II */ + case 7: + case 8: + case 11: /* Pentium III */ + case 9: + case 13: + /* Pentium M */ + break; + default: + pr_cont("unsupported p6 CPU model %d ", + boot_cpu_data.x86_model); + return -ENODEV; + } + + x86_pmu = p6_pmu; + + return 0; +} + +#endif /* CONFIG_CPU_SUP_INTEL */ diff --git a/arch/x86/kernel/cpu/perfctr-watchdog.c b/arch/x86/kernel/cpu/perfctr-watchdog.c index 898df9719af..fb329e9f849 100644 --- a/arch/x86/kernel/cpu/perfctr-watchdog.c +++ b/arch/x86/kernel/cpu/perfctr-watchdog.c @@ -115,17 +115,6 @@ int avail_to_resrv_perfctr_nmi_bit(unsigned int counter) return !test_bit(counter, perfctr_nmi_owner); } - -/* checks the an msr for availability */ -int avail_to_resrv_perfctr_nmi(unsigned int msr) -{ - unsigned int counter; - - counter = nmi_perfctr_msr_to_bit(msr); - BUG_ON(counter > NMI_MAX_COUNTER_BITS); - - return !test_bit(counter, perfctr_nmi_owner); -} EXPORT_SYMBOL(avail_to_resrv_perfctr_nmi_bit); int reserve_perfctr_nmi(unsigned int msr) @@ -691,7 +680,7 @@ static int setup_intel_arch_watchdog(unsigned nmi_hz) cpu_nmi_set_wd_enabled(); apic_write(APIC_LVTPC, APIC_DM_NMI); - evntsel |= ARCH_PERFMON_EVENTSEL0_ENABLE; + evntsel |= ARCH_PERFMON_EVENTSEL_ENABLE; wrmsr(evntsel_msr, evntsel, 0); intel_arch_wd_ops.checkbit = 1ULL << (eax.split.bit_width - 1); return 1; diff --git a/arch/x86/kernel/cpu/vmware.c b/arch/x86/kernel/cpu/vmware.c index 1cbed97b59c..b9d1ff58844 100644 --- a/arch/x86/kernel/cpu/vmware.c +++ b/arch/x86/kernel/cpu/vmware.c @@ -22,9 +22,10 @@ */ #include <linux/dmi.h> +#include <linux/module.h> #include <asm/div64.h> -#include <asm/vmware.h> #include <asm/x86_init.h> +#include <asm/hypervisor.h> #define CPUID_VMWARE_INFO_LEAF 0x40000000 #define VMWARE_HYPERVISOR_MAGIC 0x564D5868 @@ -64,7 +65,7 @@ static unsigned long vmware_get_tsc_khz(void) return tsc_hz; } -void __init vmware_platform_setup(void) +static void __init vmware_platform_setup(void) { uint32_t eax, ebx, ecx, edx; @@ -82,24 +83,21 @@ void __init vmware_platform_setup(void) * serial key should be enough, as this will always have a VMware * specific string when running under VMware hypervisor. */ -int vmware_platform(void) +static bool __init vmware_platform(void) { if (cpu_has_hypervisor) { - unsigned int eax, ebx, ecx, edx; - char hyper_vendor_id[13]; - - cpuid(CPUID_VMWARE_INFO_LEAF, &eax, &ebx, &ecx, &edx); - memcpy(hyper_vendor_id + 0, &ebx, 4); - memcpy(hyper_vendor_id + 4, &ecx, 4); - memcpy(hyper_vendor_id + 8, &edx, 4); - hyper_vendor_id[12] = '\0'; - if (!strcmp(hyper_vendor_id, "VMwareVMware")) - return 1; + unsigned int eax; + unsigned int hyper_vendor_id[3]; + + cpuid(CPUID_VMWARE_INFO_LEAF, &eax, &hyper_vendor_id[0], + &hyper_vendor_id[1], &hyper_vendor_id[2]); + if (!memcmp(hyper_vendor_id, "VMwareVMware", 12)) + return true; } else if (dmi_available && dmi_name_in_serial("VMware") && __vmware_platform()) - return 1; + return true; - return 0; + return false; } /* @@ -114,8 +112,16 @@ int vmware_platform(void) * so that the kernel could just trust the hypervisor with providing a * reliable virtual TSC that is suitable for timekeeping. */ -void __cpuinit vmware_set_feature_bits(struct cpuinfo_x86 *c) +static void __cpuinit vmware_set_cpu_features(struct cpuinfo_x86 *c) { set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC); set_cpu_cap(c, X86_FEATURE_TSC_RELIABLE); } + +const __refconst struct hypervisor_x86 x86_hyper_vmware = { + .name = "VMware", + .detect = vmware_platform, + .set_cpu_features = vmware_set_cpu_features, + .init_platform = vmware_platform_setup, +}; +EXPORT_SYMBOL(x86_hyper_vmware); diff --git a/arch/x86/kernel/cpuid.c b/arch/x86/kernel/cpuid.c index 83e5e628de7..1b7b31ab7d8 100644 --- a/arch/x86/kernel/cpuid.c +++ b/arch/x86/kernel/cpuid.c @@ -40,6 +40,7 @@ #include <linux/cpu.h> #include <linux/notifier.h> #include <linux/uaccess.h> +#include <linux/gfp.h> #include <asm/processor.h> #include <asm/msr.h> @@ -169,7 +170,7 @@ static int __cpuinit cpuid_class_cpu_callback(struct notifier_block *nfb, cpuid_device_destroy(cpu); break; } - return err ? NOTIFY_BAD : NOTIFY_OK; + return notifier_from_errno(err); } static struct notifier_block __refdata cpuid_class_cpu_notifier = diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c index a4849c10a77..ebd4c51d096 100644 --- a/arch/x86/kernel/crash.c +++ b/arch/x86/kernel/crash.c @@ -27,7 +27,6 @@ #include <asm/cpu.h> #include <asm/reboot.h> #include <asm/virtext.h> -#include <asm/x86_init.h> #if defined(CONFIG_SMP) && defined(CONFIG_X86_LOCAL_APIC) @@ -103,10 +102,5 @@ void native_machine_crash_shutdown(struct pt_regs *regs) #ifdef CONFIG_HPET_TIMER hpet_disable(); #endif - -#ifdef CONFIG_X86_64 - x86_platform.iommu_shutdown(); -#endif - crash_save_cpu(regs, safe_smp_processor_id()); } diff --git a/arch/x86/kernel/crash_dump_32.c b/arch/x86/kernel/crash_dump_32.c index cd97ce18c29..67414550c3c 100644 --- a/arch/x86/kernel/crash_dump_32.c +++ b/arch/x86/kernel/crash_dump_32.c @@ -5,6 +5,7 @@ * Copyright (C) IBM Corporation, 2004. All rights reserved */ +#include <linux/slab.h> #include <linux/errno.h> #include <linux/highmem.h> #include <linux/crash_dump.h> diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c deleted file mode 100644 index 1c47390dd0e..00000000000 --- a/arch/x86/kernel/ds.c +++ /dev/null @@ -1,1437 +0,0 @@ -/* - * Debug Store support - * - * This provides a low-level interface to the hardware's Debug Store - * feature that is used for branch trace store (BTS) and - * precise-event based sampling (PEBS). - * - * It manages: - * - DS and BTS hardware configuration - * - buffer overflow handling (to be done) - * - buffer access - * - * It does not do: - * - security checking (is the caller allowed to trace the task) - * - buffer allocation (memory accounting) - * - * - * Copyright (C) 2007-2009 Intel Corporation. - * Markus Metzger <markus.t.metzger@intel.com>, 2007-2009 - */ - -#include <linux/kernel.h> -#include <linux/string.h> -#include <linux/errno.h> -#include <linux/sched.h> -#include <linux/slab.h> -#include <linux/mm.h> -#include <linux/trace_clock.h> - -#include <asm/ds.h> - -#include "ds_selftest.h" - -/* - * The configuration for a particular DS hardware implementation: - */ -struct ds_configuration { - /* The name of the configuration: */ - const char *name; - - /* The size of pointer-typed fields in DS, BTS, and PEBS: */ - unsigned char sizeof_ptr_field; - - /* The size of a BTS/PEBS record in bytes: */ - unsigned char sizeof_rec[2]; - - /* The number of pebs counter reset values in the DS structure. */ - unsigned char nr_counter_reset; - - /* Control bit-masks indexed by enum ds_feature: */ - unsigned long ctl[dsf_ctl_max]; -}; -static struct ds_configuration ds_cfg __read_mostly; - - -/* Maximal size of a DS configuration: */ -#define MAX_SIZEOF_DS 0x80 - -/* Maximal size of a BTS record: */ -#define MAX_SIZEOF_BTS (3 * 8) - -/* BTS and PEBS buffer alignment: */ -#define DS_ALIGNMENT (1 << 3) - -/* Number of buffer pointers in DS: */ -#define NUM_DS_PTR_FIELDS 8 - -/* Size of a pebs reset value in DS: */ -#define PEBS_RESET_FIELD_SIZE 8 - -/* Mask of control bits in the DS MSR register: */ -#define BTS_CONTROL \ - ( ds_cfg.ctl[dsf_bts] | \ - ds_cfg.ctl[dsf_bts_kernel] | \ - ds_cfg.ctl[dsf_bts_user] | \ - ds_cfg.ctl[dsf_bts_overflow] ) - -/* - * A BTS or PEBS tracer. - * - * This holds the configuration of the tracer and serves as a handle - * to identify tracers. - */ -struct ds_tracer { - /* The DS context (partially) owned by this tracer. */ - struct ds_context *context; - /* The buffer provided on ds_request() and its size in bytes. */ - void *buffer; - size_t size; -}; - -struct bts_tracer { - /* The common DS part: */ - struct ds_tracer ds; - - /* The trace including the DS configuration: */ - struct bts_trace trace; - - /* Buffer overflow notification function: */ - bts_ovfl_callback_t ovfl; - - /* Active flags affecting trace collection. */ - unsigned int flags; -}; - -struct pebs_tracer { - /* The common DS part: */ - struct ds_tracer ds; - - /* The trace including the DS configuration: */ - struct pebs_trace trace; - - /* Buffer overflow notification function: */ - pebs_ovfl_callback_t ovfl; -}; - -/* - * Debug Store (DS) save area configuration (see Intel64 and IA32 - * Architectures Software Developer's Manual, section 18.5) - * - * The DS configuration consists of the following fields; different - * architetures vary in the size of those fields. - * - * - double-word aligned base linear address of the BTS buffer - * - write pointer into the BTS buffer - * - end linear address of the BTS buffer (one byte beyond the end of - * the buffer) - * - interrupt pointer into BTS buffer - * (interrupt occurs when write pointer passes interrupt pointer) - * - double-word aligned base linear address of the PEBS buffer - * - write pointer into the PEBS buffer - * - end linear address of the PEBS buffer (one byte beyond the end of - * the buffer) - * - interrupt pointer into PEBS buffer - * (interrupt occurs when write pointer passes interrupt pointer) - * - value to which counter is reset following counter overflow - * - * Later architectures use 64bit pointers throughout, whereas earlier - * architectures use 32bit pointers in 32bit mode. - * - * - * We compute the base address for the first 8 fields based on: - * - the field size stored in the DS configuration - * - the relative field position - * - an offset giving the start of the respective region - * - * This offset is further used to index various arrays holding - * information for BTS and PEBS at the respective index. - * - * On later 32bit processors, we only access the lower 32bit of the - * 64bit pointer fields. The upper halves will be zeroed out. - */ - -enum ds_field { - ds_buffer_base = 0, - ds_index, - ds_absolute_maximum, - ds_interrupt_threshold, -}; - -enum ds_qualifier { - ds_bts = 0, - ds_pebs -}; - -static inline unsigned long -ds_get(const unsigned char *base, enum ds_qualifier qual, enum ds_field field) -{ - base += (ds_cfg.sizeof_ptr_field * (field + (4 * qual))); - return *(unsigned long *)base; -} - -static inline void -ds_set(unsigned char *base, enum ds_qualifier qual, enum ds_field field, - unsigned long value) -{ - base += (ds_cfg.sizeof_ptr_field * (field + (4 * qual))); - (*(unsigned long *)base) = value; -} - - -/* - * Locking is done only for allocating BTS or PEBS resources. - */ -static DEFINE_SPINLOCK(ds_lock); - -/* - * We either support (system-wide) per-cpu or per-thread allocation. - * We distinguish the two based on the task_struct pointer, where a - * NULL pointer indicates per-cpu allocation for the current cpu. - * - * Allocations are use-counted. As soon as resources are allocated, - * further allocations must be of the same type (per-cpu or - * per-thread). We model this by counting allocations (i.e. the number - * of tracers of a certain type) for one type negatively: - * =0 no tracers - * >0 number of per-thread tracers - * <0 number of per-cpu tracers - * - * Tracers essentially gives the number of ds contexts for a certain - * type of allocation. - */ -static atomic_t tracers = ATOMIC_INIT(0); - -static inline int get_tracer(struct task_struct *task) -{ - int error; - - spin_lock_irq(&ds_lock); - - if (task) { - error = -EPERM; - if (atomic_read(&tracers) < 0) - goto out; - atomic_inc(&tracers); - } else { - error = -EPERM; - if (atomic_read(&tracers) > 0) - goto out; - atomic_dec(&tracers); - } - - error = 0; -out: - spin_unlock_irq(&ds_lock); - return error; -} - -static inline void put_tracer(struct task_struct *task) -{ - if (task) - atomic_dec(&tracers); - else - atomic_inc(&tracers); -} - -/* - * The DS context is either attached to a thread or to a cpu: - * - in the former case, the thread_struct contains a pointer to the - * attached context. - * - in the latter case, we use a static array of per-cpu context - * pointers. - * - * Contexts are use-counted. They are allocated on first access and - * deallocated when the last user puts the context. - */ -struct ds_context { - /* The DS configuration; goes into MSR_IA32_DS_AREA: */ - unsigned char ds[MAX_SIZEOF_DS]; - - /* The owner of the BTS and PEBS configuration, respectively: */ - struct bts_tracer *bts_master; - struct pebs_tracer *pebs_master; - - /* Use count: */ - unsigned long count; - - /* Pointer to the context pointer field: */ - struct ds_context **this; - - /* The traced task; NULL for cpu tracing: */ - struct task_struct *task; - - /* The traced cpu; only valid if task is NULL: */ - int cpu; -}; - -static DEFINE_PER_CPU(struct ds_context *, cpu_ds_context); - - -static struct ds_context *ds_get_context(struct task_struct *task, int cpu) -{ - struct ds_context **p_context = - (task ? &task->thread.ds_ctx : &per_cpu(cpu_ds_context, cpu)); - struct ds_context *context = NULL; - struct ds_context *new_context = NULL; - - /* Chances are small that we already have a context. */ - new_context = kzalloc(sizeof(*new_context), GFP_KERNEL); - if (!new_context) - return NULL; - - spin_lock_irq(&ds_lock); - - context = *p_context; - if (likely(!context)) { - context = new_context; - - context->this = p_context; - context->task = task; - context->cpu = cpu; - context->count = 0; - - *p_context = context; - } - - context->count++; - - spin_unlock_irq(&ds_lock); - - if (context != new_context) - kfree(new_context); - - return context; -} - -static void ds_put_context(struct ds_context *context) -{ - struct task_struct *task; - unsigned long irq; - - if (!context) - return; - - spin_lock_irqsave(&ds_lock, irq); - - if (--context->count) { - spin_unlock_irqrestore(&ds_lock, irq); - return; - } - - *(context->this) = NULL; - - task = context->task; - - if (task) - clear_tsk_thread_flag(task, TIF_DS_AREA_MSR); - - /* - * We leave the (now dangling) pointer to the DS configuration in - * the DS_AREA msr. This is as good or as bad as replacing it with - * NULL - the hardware would crash if we enabled tracing. - * - * This saves us some problems with having to write an msr on a - * different cpu while preventing others from doing the same for the - * next context for that same cpu. - */ - - spin_unlock_irqrestore(&ds_lock, irq); - - /* The context might still be in use for context switching. */ - if (task && (task != current)) - wait_task_context_switch(task); - - kfree(context); -} - -static void ds_install_ds_area(struct ds_context *context) -{ - unsigned long ds; - - ds = (unsigned long)context->ds; - - /* - * There is a race between the bts master and the pebs master. - * - * The thread/cpu access is synchronized via get/put_cpu() for - * task tracing and via wrmsr_on_cpu for cpu tracing. - * - * If bts and pebs are collected for the same task or same cpu, - * the same confiuration is written twice. - */ - if (context->task) { - get_cpu(); - if (context->task == current) - wrmsrl(MSR_IA32_DS_AREA, ds); - set_tsk_thread_flag(context->task, TIF_DS_AREA_MSR); - put_cpu(); - } else - wrmsr_on_cpu(context->cpu, MSR_IA32_DS_AREA, - (u32)((u64)ds), (u32)((u64)ds >> 32)); -} - -/* - * Call the tracer's callback on a buffer overflow. - * - * context: the ds context - * qual: the buffer type - */ -static void ds_overflow(struct ds_context *context, enum ds_qualifier qual) -{ - switch (qual) { - case ds_bts: - if (context->bts_master && - context->bts_master->ovfl) - context->bts_master->ovfl(context->bts_master); - break; - case ds_pebs: - if (context->pebs_master && - context->pebs_master->ovfl) - context->pebs_master->ovfl(context->pebs_master); - break; - } -} - - -/* - * Write raw data into the BTS or PEBS buffer. - * - * The remainder of any partially written record is zeroed out. - * - * context: the DS context - * qual: the buffer type - * record: the data to write - * size: the size of the data - */ -static int ds_write(struct ds_context *context, enum ds_qualifier qual, - const void *record, size_t size) -{ - int bytes_written = 0; - - if (!record) - return -EINVAL; - - while (size) { - unsigned long base, index, end, write_end, int_th; - unsigned long write_size, adj_write_size; - - /* - * Write as much as possible without producing an - * overflow interrupt. - * - * Interrupt_threshold must either be - * - bigger than absolute_maximum or - * - point to a record between buffer_base and absolute_maximum - * - * Index points to a valid record. - */ - base = ds_get(context->ds, qual, ds_buffer_base); - index = ds_get(context->ds, qual, ds_index); - end = ds_get(context->ds, qual, ds_absolute_maximum); - int_th = ds_get(context->ds, qual, ds_interrupt_threshold); - - write_end = min(end, int_th); - - /* - * If we are already beyond the interrupt threshold, - * we fill the entire buffer. - */ - if (write_end <= index) - write_end = end; - - if (write_end <= index) - break; - - write_size = min((unsigned long) size, write_end - index); - memcpy((void *)index, record, write_size); - - record = (const char *)record + write_size; - size -= write_size; - bytes_written += write_size; - - adj_write_size = write_size / ds_cfg.sizeof_rec[qual]; - adj_write_size *= ds_cfg.sizeof_rec[qual]; - - /* Zero out trailing bytes. */ - memset((char *)index + write_size, 0, - adj_write_size - write_size); - index += adj_write_size; - - if (index >= end) - index = base; - ds_set(context->ds, qual, ds_index, index); - - if (index >= int_th) - ds_overflow(context, qual); - } - - return bytes_written; -} - - -/* - * Branch Trace Store (BTS) uses the following format. Different - * architectures vary in the size of those fields. - * - source linear address - * - destination linear address - * - flags - * - * Later architectures use 64bit pointers throughout, whereas earlier - * architectures use 32bit pointers in 32bit mode. - * - * We compute the base address for the fields based on: - * - the field size stored in the DS configuration - * - the relative field position - * - * In order to store additional information in the BTS buffer, we use - * a special source address to indicate that the record requires - * special interpretation. - * - * Netburst indicated via a bit in the flags field whether the branch - * was predicted; this is ignored. - * - * We use two levels of abstraction: - * - the raw data level defined here - * - an arch-independent level defined in ds.h - */ - -enum bts_field { - bts_from, - bts_to, - bts_flags, - - bts_qual = bts_from, - bts_clock = bts_to, - bts_pid = bts_flags, - - bts_qual_mask = (bts_qual_max - 1), - bts_escape = ((unsigned long)-1 & ~bts_qual_mask) -}; - -static inline unsigned long bts_get(const char *base, unsigned long field) -{ - base += (ds_cfg.sizeof_ptr_field * field); - return *(unsigned long *)base; -} - -static inline void bts_set(char *base, unsigned long field, unsigned long val) -{ - base += (ds_cfg.sizeof_ptr_field * field); - (*(unsigned long *)base) = val; -} - - -/* - * The raw BTS data is architecture dependent. - * - * For higher-level users, we give an arch-independent view. - * - ds.h defines struct bts_struct - * - bts_read translates one raw bts record into a bts_struct - * - bts_write translates one bts_struct into the raw format and - * writes it into the top of the parameter tracer's buffer. - * - * return: bytes read/written on success; -Eerrno, otherwise - */ -static int -bts_read(struct bts_tracer *tracer, const void *at, struct bts_struct *out) -{ - if (!tracer) - return -EINVAL; - - if (at < tracer->trace.ds.begin) - return -EINVAL; - - if (tracer->trace.ds.end < (at + tracer->trace.ds.size)) - return -EINVAL; - - memset(out, 0, sizeof(*out)); - if ((bts_get(at, bts_qual) & ~bts_qual_mask) == bts_escape) { - out->qualifier = (bts_get(at, bts_qual) & bts_qual_mask); - out->variant.event.clock = bts_get(at, bts_clock); - out->variant.event.pid = bts_get(at, bts_pid); - } else { - out->qualifier = bts_branch; - out->variant.lbr.from = bts_get(at, bts_from); - out->variant.lbr.to = bts_get(at, bts_to); - - if (!out->variant.lbr.from && !out->variant.lbr.to) - out->qualifier = bts_invalid; - } - - return ds_cfg.sizeof_rec[ds_bts]; -} - -static int bts_write(struct bts_tracer *tracer, const struct bts_struct *in) -{ - unsigned char raw[MAX_SIZEOF_BTS]; - - if (!tracer) - return -EINVAL; - - if (MAX_SIZEOF_BTS < ds_cfg.sizeof_rec[ds_bts]) - return -EOVERFLOW; - - switch (in->qualifier) { - case bts_invalid: - bts_set(raw, bts_from, 0); - bts_set(raw, bts_to, 0); - bts_set(raw, bts_flags, 0); - break; - case bts_branch: - bts_set(raw, bts_from, in->variant.lbr.from); - bts_set(raw, bts_to, in->variant.lbr.to); - bts_set(raw, bts_flags, 0); - break; - case bts_task_arrives: - case bts_task_departs: - bts_set(raw, bts_qual, (bts_escape | in->qualifier)); - bts_set(raw, bts_clock, in->variant.event.clock); - bts_set(raw, bts_pid, in->variant.event.pid); - break; - default: - return -EINVAL; - } - - return ds_write(tracer->ds.context, ds_bts, raw, - ds_cfg.sizeof_rec[ds_bts]); -} - - -static void ds_write_config(struct ds_context *context, - struct ds_trace *cfg, enum ds_qualifier qual) -{ - unsigned char *ds = context->ds; - - ds_set(ds, qual, ds_buffer_base, (unsigned long)cfg->begin); - ds_set(ds, qual, ds_index, (unsigned long)cfg->top); - ds_set(ds, qual, ds_absolute_maximum, (unsigned long)cfg->end); - ds_set(ds, qual, ds_interrupt_threshold, (unsigned long)cfg->ith); -} - -static void ds_read_config(struct ds_context *context, - struct ds_trace *cfg, enum ds_qualifier qual) -{ - unsigned char *ds = context->ds; - - cfg->begin = (void *)ds_get(ds, qual, ds_buffer_base); - cfg->top = (void *)ds_get(ds, qual, ds_index); - cfg->end = (void *)ds_get(ds, qual, ds_absolute_maximum); - cfg->ith = (void *)ds_get(ds, qual, ds_interrupt_threshold); -} - -static void ds_init_ds_trace(struct ds_trace *trace, enum ds_qualifier qual, - void *base, size_t size, size_t ith, - unsigned int flags) { - unsigned long buffer, adj; - - /* - * Adjust the buffer address and size to meet alignment - * constraints: - * - buffer is double-word aligned - * - size is multiple of record size - * - * We checked the size at the very beginning; we have enough - * space to do the adjustment. - */ - buffer = (unsigned long)base; - - adj = ALIGN(buffer, DS_ALIGNMENT) - buffer; - buffer += adj; - size -= adj; - - trace->n = size / ds_cfg.sizeof_rec[qual]; - trace->size = ds_cfg.sizeof_rec[qual]; - - size = (trace->n * trace->size); - - trace->begin = (void *)buffer; - trace->top = trace->begin; - trace->end = (void *)(buffer + size); - /* - * The value for 'no threshold' is -1, which will set the - * threshold outside of the buffer, just like we want it. - */ - ith *= ds_cfg.sizeof_rec[qual]; - trace->ith = (void *)(buffer + size - ith); - - trace->flags = flags; -} - - -static int ds_request(struct ds_tracer *tracer, struct ds_trace *trace, - enum ds_qualifier qual, struct task_struct *task, - int cpu, void *base, size_t size, size_t th) -{ - struct ds_context *context; - int error; - size_t req_size; - - error = -EOPNOTSUPP; - if (!ds_cfg.sizeof_rec[qual]) - goto out; - - error = -EINVAL; - if (!base) - goto out; - - req_size = ds_cfg.sizeof_rec[qual]; - /* We might need space for alignment adjustments. */ - if (!IS_ALIGNED((unsigned long)base, DS_ALIGNMENT)) - req_size += DS_ALIGNMENT; - - error = -EINVAL; - if (size < req_size) - goto out; - - if (th != (size_t)-1) { - th *= ds_cfg.sizeof_rec[qual]; - - error = -EINVAL; - if (size <= th) - goto out; - } - - tracer->buffer = base; - tracer->size = size; - - error = -ENOMEM; - context = ds_get_context(task, cpu); - if (!context) - goto out; - tracer->context = context; - - /* - * Defer any tracer-specific initialization work for the context until - * context ownership has been clarified. - */ - - error = 0; - out: - return error; -} - -static struct bts_tracer *ds_request_bts(struct task_struct *task, int cpu, - void *base, size_t size, - bts_ovfl_callback_t ovfl, size_t th, - unsigned int flags) -{ - struct bts_tracer *tracer; - int error; - - /* Buffer overflow notification is not yet implemented. */ - error = -EOPNOTSUPP; - if (ovfl) - goto out; - - error = get_tracer(task); - if (error < 0) - goto out; - - error = -ENOMEM; - tracer = kzalloc(sizeof(*tracer), GFP_KERNEL); - if (!tracer) - goto out_put_tracer; - tracer->ovfl = ovfl; - - /* Do some more error checking and acquire a tracing context. */ - error = ds_request(&tracer->ds, &tracer->trace.ds, - ds_bts, task, cpu, base, size, th); - if (error < 0) - goto out_tracer; - - /* Claim the bts part of the tracing context we acquired above. */ - spin_lock_irq(&ds_lock); - - error = -EPERM; - if (tracer->ds.context->bts_master) - goto out_unlock; - tracer->ds.context->bts_master = tracer; - - spin_unlock_irq(&ds_lock); - - /* - * Now that we own the bts part of the context, let's complete the - * initialization for that part. - */ - ds_init_ds_trace(&tracer->trace.ds, ds_bts, base, size, th, flags); - ds_write_config(tracer->ds.context, &tracer->trace.ds, ds_bts); - ds_install_ds_area(tracer->ds.context); - - tracer->trace.read = bts_read; - tracer->trace.write = bts_write; - - /* Start tracing. */ - ds_resume_bts(tracer); - - return tracer; - - out_unlock: - spin_unlock_irq(&ds_lock); - ds_put_context(tracer->ds.context); - out_tracer: - kfree(tracer); - out_put_tracer: - put_tracer(task); - out: - return ERR_PTR(error); -} - -struct bts_tracer *ds_request_bts_task(struct task_struct *task, - void *base, size_t size, - bts_ovfl_callback_t ovfl, - size_t th, unsigned int flags) -{ - return ds_request_bts(task, 0, base, size, ovfl, th, flags); -} - -struct bts_tracer *ds_request_bts_cpu(int cpu, void *base, size_t size, - bts_ovfl_callback_t ovfl, - size_t th, unsigned int flags) -{ - return ds_request_bts(NULL, cpu, base, size, ovfl, th, flags); -} - -static struct pebs_tracer *ds_request_pebs(struct task_struct *task, int cpu, - void *base, size_t size, - pebs_ovfl_callback_t ovfl, size_t th, - unsigned int flags) -{ - struct pebs_tracer *tracer; - int error; - - /* Buffer overflow notification is not yet implemented. */ - error = -EOPNOTSUPP; - if (ovfl) - goto out; - - error = get_tracer(task); - if (error < 0) - goto out; - - error = -ENOMEM; - tracer = kzalloc(sizeof(*tracer), GFP_KERNEL); - if (!tracer) - goto out_put_tracer; - tracer->ovfl = ovfl; - - /* Do some more error checking and acquire a tracing context. */ - error = ds_request(&tracer->ds, &tracer->trace.ds, - ds_pebs, task, cpu, base, size, th); - if (error < 0) - goto out_tracer; - - /* Claim the pebs part of the tracing context we acquired above. */ - spin_lock_irq(&ds_lock); - - error = -EPERM; - if (tracer->ds.context->pebs_master) - goto out_unlock; - tracer->ds.context->pebs_master = tracer; - - spin_unlock_irq(&ds_lock); - - /* - * Now that we own the pebs part of the context, let's complete the - * initialization for that part. - */ - ds_init_ds_trace(&tracer->trace.ds, ds_pebs, base, size, th, flags); - ds_write_config(tracer->ds.context, &tracer->trace.ds, ds_pebs); - ds_install_ds_area(tracer->ds.context); - - /* Start tracing. */ - ds_resume_pebs(tracer); - - return tracer; - - out_unlock: - spin_unlock_irq(&ds_lock); - ds_put_context(tracer->ds.context); - out_tracer: - kfree(tracer); - out_put_tracer: - put_tracer(task); - out: - return ERR_PTR(error); -} - -struct pebs_tracer *ds_request_pebs_task(struct task_struct *task, - void *base, size_t size, - pebs_ovfl_callback_t ovfl, - size_t th, unsigned int flags) -{ - return ds_request_pebs(task, 0, base, size, ovfl, th, flags); -} - -struct pebs_tracer *ds_request_pebs_cpu(int cpu, void *base, size_t size, - pebs_ovfl_callback_t ovfl, - size_t th, unsigned int flags) -{ - return ds_request_pebs(NULL, cpu, base, size, ovfl, th, flags); -} - -static void ds_free_bts(struct bts_tracer *tracer) -{ - struct task_struct *task; - - task = tracer->ds.context->task; - - WARN_ON_ONCE(tracer->ds.context->bts_master != tracer); - tracer->ds.context->bts_master = NULL; - - /* Make sure tracing stopped and the tracer is not in use. */ - if (task && (task != current)) - wait_task_context_switch(task); - - ds_put_context(tracer->ds.context); - put_tracer(task); - - kfree(tracer); -} - -void ds_release_bts(struct bts_tracer *tracer) -{ - might_sleep(); - - if (!tracer) - return; - - ds_suspend_bts(tracer); - ds_free_bts(tracer); -} - -int ds_release_bts_noirq(struct bts_tracer *tracer) -{ - struct task_struct *task; - unsigned long irq; - int error; - - if (!tracer) - return 0; - - task = tracer->ds.context->task; - - local_irq_save(irq); - - error = -EPERM; - if (!task && - (tracer->ds.context->cpu != smp_processor_id())) - goto out; - - error = -EPERM; - if (task && (task != current)) - goto out; - - ds_suspend_bts_noirq(tracer); - ds_free_bts(tracer); - - error = 0; - out: - local_irq_restore(irq); - return error; -} - -static void update_task_debugctlmsr(struct task_struct *task, - unsigned long debugctlmsr) -{ - task->thread.debugctlmsr = debugctlmsr; - - get_cpu(); - if (task == current) - update_debugctlmsr(debugctlmsr); - put_cpu(); -} - -void ds_suspend_bts(struct bts_tracer *tracer) -{ - struct task_struct *task; - unsigned long debugctlmsr; - int cpu; - - if (!tracer) - return; - - tracer->flags = 0; - - task = tracer->ds.context->task; - cpu = tracer->ds.context->cpu; - - WARN_ON(!task && irqs_disabled()); - - debugctlmsr = (task ? - task->thread.debugctlmsr : - get_debugctlmsr_on_cpu(cpu)); - debugctlmsr &= ~BTS_CONTROL; - - if (task) - update_task_debugctlmsr(task, debugctlmsr); - else - update_debugctlmsr_on_cpu(cpu, debugctlmsr); -} - -int ds_suspend_bts_noirq(struct bts_tracer *tracer) -{ - struct task_struct *task; - unsigned long debugctlmsr, irq; - int cpu, error = 0; - - if (!tracer) - return 0; - - tracer->flags = 0; - - task = tracer->ds.context->task; - cpu = tracer->ds.context->cpu; - - local_irq_save(irq); - - error = -EPERM; - if (!task && (cpu != smp_processor_id())) - goto out; - - debugctlmsr = (task ? - task->thread.debugctlmsr : - get_debugctlmsr()); - debugctlmsr &= ~BTS_CONTROL; - - if (task) - update_task_debugctlmsr(task, debugctlmsr); - else - update_debugctlmsr(debugctlmsr); - - error = 0; - out: - local_irq_restore(irq); - return error; -} - -static unsigned long ds_bts_control(struct bts_tracer *tracer) -{ - unsigned long control; - - control = ds_cfg.ctl[dsf_bts]; - if (!(tracer->trace.ds.flags & BTS_KERNEL)) - control |= ds_cfg.ctl[dsf_bts_kernel]; - if (!(tracer->trace.ds.flags & BTS_USER)) - control |= ds_cfg.ctl[dsf_bts_user]; - - return control; -} - -void ds_resume_bts(struct bts_tracer *tracer) -{ - struct task_struct *task; - unsigned long debugctlmsr; - int cpu; - - if (!tracer) - return; - - tracer->flags = tracer->trace.ds.flags; - - task = tracer->ds.context->task; - cpu = tracer->ds.context->cpu; - - WARN_ON(!task && irqs_disabled()); - - debugctlmsr = (task ? - task->thread.debugctlmsr : - get_debugctlmsr_on_cpu(cpu)); - debugctlmsr |= ds_bts_control(tracer); - - if (task) - update_task_debugctlmsr(task, debugctlmsr); - else - update_debugctlmsr_on_cpu(cpu, debugctlmsr); -} - -int ds_resume_bts_noirq(struct bts_tracer *tracer) -{ - struct task_struct *task; - unsigned long debugctlmsr, irq; - int cpu, error = 0; - - if (!tracer) - return 0; - - tracer->flags = tracer->trace.ds.flags; - - task = tracer->ds.context->task; - cpu = tracer->ds.context->cpu; - - local_irq_save(irq); - - error = -EPERM; - if (!task && (cpu != smp_processor_id())) - goto out; - - debugctlmsr = (task ? - task->thread.debugctlmsr : - get_debugctlmsr()); - debugctlmsr |= ds_bts_control(tracer); - - if (task) - update_task_debugctlmsr(task, debugctlmsr); - else - update_debugctlmsr(debugctlmsr); - - error = 0; - out: - local_irq_restore(irq); - return error; -} - -static void ds_free_pebs(struct pebs_tracer *tracer) -{ - struct task_struct *task; - - task = tracer->ds.context->task; - - WARN_ON_ONCE(tracer->ds.context->pebs_master != tracer); - tracer->ds.context->pebs_master = NULL; - - ds_put_context(tracer->ds.context); - put_tracer(task); - - kfree(tracer); -} - -void ds_release_pebs(struct pebs_tracer *tracer) -{ - might_sleep(); - - if (!tracer) - return; - - ds_suspend_pebs(tracer); - ds_free_pebs(tracer); -} - -int ds_release_pebs_noirq(struct pebs_tracer *tracer) -{ - struct task_struct *task; - unsigned long irq; - int error; - - if (!tracer) - return 0; - - task = tracer->ds.context->task; - - local_irq_save(irq); - - error = -EPERM; - if (!task && - (tracer->ds.context->cpu != smp_processor_id())) - goto out; - - error = -EPERM; - if (task && (task != current)) - goto out; - - ds_suspend_pebs_noirq(tracer); - ds_free_pebs(tracer); - - error = 0; - out: - local_irq_restore(irq); - return error; -} - -void ds_suspend_pebs(struct pebs_tracer *tracer) -{ - -} - -int ds_suspend_pebs_noirq(struct pebs_tracer *tracer) -{ - return 0; -} - -void ds_resume_pebs(struct pebs_tracer *tracer) -{ - -} - -int ds_resume_pebs_noirq(struct pebs_tracer *tracer) -{ - return 0; -} - -const struct bts_trace *ds_read_bts(struct bts_tracer *tracer) -{ - if (!tracer) - return NULL; - - ds_read_config(tracer->ds.context, &tracer->trace.ds, ds_bts); - return &tracer->trace; -} - -const struct pebs_trace *ds_read_pebs(struct pebs_tracer *tracer) -{ - if (!tracer) - return NULL; - - ds_read_config(tracer->ds.context, &tracer->trace.ds, ds_pebs); - - tracer->trace.counters = ds_cfg.nr_counter_reset; - memcpy(tracer->trace.counter_reset, - tracer->ds.context->ds + - (NUM_DS_PTR_FIELDS * ds_cfg.sizeof_ptr_field), - ds_cfg.nr_counter_reset * PEBS_RESET_FIELD_SIZE); - - return &tracer->trace; -} - -int ds_reset_bts(struct bts_tracer *tracer) -{ - if (!tracer) - return -EINVAL; - - tracer->trace.ds.top = tracer->trace.ds.begin; - - ds_set(tracer->ds.context->ds, ds_bts, ds_index, - (unsigned long)tracer->trace.ds.top); - - return 0; -} - -int ds_reset_pebs(struct pebs_tracer *tracer) -{ - if (!tracer) - return -EINVAL; - - tracer->trace.ds.top = tracer->trace.ds.begin; - - ds_set(tracer->ds.context->ds, ds_pebs, ds_index, - (unsigned long)tracer->trace.ds.top); - - return 0; -} - -int ds_set_pebs_reset(struct pebs_tracer *tracer, - unsigned int counter, u64 value) -{ - if (!tracer) - return -EINVAL; - - if (ds_cfg.nr_counter_reset < counter) - return -EINVAL; - - *(u64 *)(tracer->ds.context->ds + - (NUM_DS_PTR_FIELDS * ds_cfg.sizeof_ptr_field) + - (counter * PEBS_RESET_FIELD_SIZE)) = value; - - return 0; -} - -static const struct ds_configuration ds_cfg_netburst = { - .name = "Netburst", - .ctl[dsf_bts] = (1 << 2) | (1 << 3), - .ctl[dsf_bts_kernel] = (1 << 5), - .ctl[dsf_bts_user] = (1 << 6), - .nr_counter_reset = 1, -}; -static const struct ds_configuration ds_cfg_pentium_m = { - .name = "Pentium M", - .ctl[dsf_bts] = (1 << 6) | (1 << 7), - .nr_counter_reset = 1, -}; -static const struct ds_configuration ds_cfg_core2_atom = { - .name = "Core 2/Atom", - .ctl[dsf_bts] = (1 << 6) | (1 << 7), - .ctl[dsf_bts_kernel] = (1 << 9), - .ctl[dsf_bts_user] = (1 << 10), - .nr_counter_reset = 1, -}; -static const struct ds_configuration ds_cfg_core_i7 = { - .name = "Core i7", - .ctl[dsf_bts] = (1 << 6) | (1 << 7), - .ctl[dsf_bts_kernel] = (1 << 9), - .ctl[dsf_bts_user] = (1 << 10), - .nr_counter_reset = 4, -}; - -static void -ds_configure(const struct ds_configuration *cfg, - struct cpuinfo_x86 *cpu) -{ - unsigned long nr_pebs_fields = 0; - - printk(KERN_INFO "[ds] using %s configuration\n", cfg->name); - -#ifdef __i386__ - nr_pebs_fields = 10; -#else - nr_pebs_fields = 18; -#endif - - /* - * Starting with version 2, architectural performance - * monitoring supports a format specifier. - */ - if ((cpuid_eax(0xa) & 0xff) > 1) { - unsigned long perf_capabilities, format; - - rdmsrl(MSR_IA32_PERF_CAPABILITIES, perf_capabilities); - - format = (perf_capabilities >> 8) & 0xf; - - switch (format) { - case 0: - nr_pebs_fields = 18; - break; - case 1: - nr_pebs_fields = 22; - break; - default: - printk(KERN_INFO - "[ds] unknown PEBS format: %lu\n", format); - nr_pebs_fields = 0; - break; - } - } - - memset(&ds_cfg, 0, sizeof(ds_cfg)); - ds_cfg = *cfg; - - ds_cfg.sizeof_ptr_field = - (cpu_has(cpu, X86_FEATURE_DTES64) ? 8 : 4); - - ds_cfg.sizeof_rec[ds_bts] = ds_cfg.sizeof_ptr_field * 3; - ds_cfg.sizeof_rec[ds_pebs] = ds_cfg.sizeof_ptr_field * nr_pebs_fields; - - if (!cpu_has(cpu, X86_FEATURE_BTS)) { - ds_cfg.sizeof_rec[ds_bts] = 0; - printk(KERN_INFO "[ds] bts not available\n"); - } - if (!cpu_has(cpu, X86_FEATURE_PEBS)) { - ds_cfg.sizeof_rec[ds_pebs] = 0; - printk(KERN_INFO "[ds] pebs not available\n"); - } - - printk(KERN_INFO "[ds] sizes: address: %u bit, ", - 8 * ds_cfg.sizeof_ptr_field); - printk("bts/pebs record: %u/%u bytes\n", - ds_cfg.sizeof_rec[ds_bts], ds_cfg.sizeof_rec[ds_pebs]); - - WARN_ON_ONCE(MAX_PEBS_COUNTERS < ds_cfg.nr_counter_reset); -} - -void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) -{ - /* Only configure the first cpu. Others are identical. */ - if (ds_cfg.name) - return; - - switch (c->x86) { - case 0x6: - switch (c->x86_model) { - case 0x9: - case 0xd: /* Pentium M */ - ds_configure(&ds_cfg_pentium_m, c); - break; - case 0xf: - case 0x17: /* Core2 */ - case 0x1c: /* Atom */ - ds_configure(&ds_cfg_core2_atom, c); - break; - case 0x1a: /* Core i7 */ - ds_configure(&ds_cfg_core_i7, c); - break; - default: - /* Sorry, don't know about them. */ - break; - } - break; - case 0xf: - switch (c->x86_model) { - case 0x0: - case 0x1: - case 0x2: /* Netburst */ - ds_configure(&ds_cfg_netburst, c); - break; - default: - /* Sorry, don't know about them. */ - break; - } - break; - default: - /* Sorry, don't know about them. */ - break; - } -} - -static inline void ds_take_timestamp(struct ds_context *context, - enum bts_qualifier qualifier, - struct task_struct *task) -{ - struct bts_tracer *tracer = context->bts_master; - struct bts_struct ts; - - /* Prevent compilers from reading the tracer pointer twice. */ - barrier(); - - if (!tracer || !(tracer->flags & BTS_TIMESTAMPS)) - return; - - memset(&ts, 0, sizeof(ts)); - ts.qualifier = qualifier; - ts.variant.event.clock = trace_clock_global(); - ts.variant.event.pid = task->pid; - - bts_write(tracer, &ts); -} - -/* - * Change the DS configuration from tracing prev to tracing next. - */ -void ds_switch_to(struct task_struct *prev, struct task_struct *next) -{ - struct ds_context *prev_ctx = prev->thread.ds_ctx; - struct ds_context *next_ctx = next->thread.ds_ctx; - unsigned long debugctlmsr = next->thread.debugctlmsr; - - /* Make sure all data is read before we start. */ - barrier(); - - if (prev_ctx) { - update_debugctlmsr(0); - - ds_take_timestamp(prev_ctx, bts_task_departs, prev); - } - - if (next_ctx) { - ds_take_timestamp(next_ctx, bts_task_arrives, next); - - wrmsrl(MSR_IA32_DS_AREA, (unsigned long)next_ctx->ds); - } - - update_debugctlmsr(debugctlmsr); -} - -static __init int ds_selftest(void) -{ - if (ds_cfg.sizeof_rec[ds_bts]) { - int error; - - error = ds_selftest_bts(); - if (error) { - WARN(1, "[ds] selftest failed. disabling bts.\n"); - ds_cfg.sizeof_rec[ds_bts] = 0; - } - } - - if (ds_cfg.sizeof_rec[ds_pebs]) { - int error; - - error = ds_selftest_pebs(); - if (error) { - WARN(1, "[ds] selftest failed. disabling pebs.\n"); - ds_cfg.sizeof_rec[ds_pebs] = 0; - } - } - - return 0; -} -device_initcall(ds_selftest); diff --git a/arch/x86/kernel/ds_selftest.c b/arch/x86/kernel/ds_selftest.c deleted file mode 100644 index 6bc7c199ab9..00000000000 --- a/arch/x86/kernel/ds_selftest.c +++ /dev/null @@ -1,408 +0,0 @@ -/* - * Debug Store support - selftest - * - * - * Copyright (C) 2009 Intel Corporation. - * Markus Metzger <markus.t.metzger@intel.com>, 2009 - */ - -#include "ds_selftest.h" - -#include <linux/kernel.h> -#include <linux/string.h> -#include <linux/smp.h> -#include <linux/cpu.h> - -#include <asm/ds.h> - - -#define BUFFER_SIZE 521 /* Intentionally chose an odd size. */ -#define SMALL_BUFFER_SIZE 24 /* A single bts entry. */ - -struct ds_selftest_bts_conf { - struct bts_tracer *tracer; - int error; - int (*suspend)(struct bts_tracer *); - int (*resume)(struct bts_tracer *); -}; - -static int ds_selftest_bts_consistency(const struct bts_trace *trace) -{ - int error = 0; - - if (!trace) { - printk(KERN_CONT "failed to access trace..."); - /* Bail out. Other tests are pointless. */ - return -1; - } - - if (!trace->read) { - printk(KERN_CONT "bts read not available..."); - error = -1; - } - - /* Do some sanity checks on the trace configuration. */ - if (!trace->ds.n) { - printk(KERN_CONT "empty bts buffer..."); - error = -1; - } - if (!trace->ds.size) { - printk(KERN_CONT "bad bts trace setup..."); - error = -1; - } - if (trace->ds.end != - (char *)trace->ds.begin + (trace->ds.n * trace->ds.size)) { - printk(KERN_CONT "bad bts buffer setup..."); - error = -1; - } - /* - * We allow top in [begin; end], since its not clear when the - * overflow adjustment happens: after the increment or before the - * write. - */ - if ((trace->ds.top < trace->ds.begin) || - (trace->ds.end < trace->ds.top)) { - printk(KERN_CONT "bts top out of bounds..."); - error = -1; - } - - return error; -} - -static int ds_selftest_bts_read(struct bts_tracer *tracer, - const struct bts_trace *trace, - const void *from, const void *to) -{ - const unsigned char *at; - - /* - * Check a few things which do not belong to this test. - * They should be covered by other tests. - */ - if (!trace) - return -1; - - if (!trace->read) - return -1; - - if (to < from) - return -1; - - if (from < trace->ds.begin) - return -1; - - if (trace->ds.end < to) - return -1; - - if (!trace->ds.size) - return -1; - - /* Now to the test itself. */ - for (at = from; (void *)at < to; at += trace->ds.size) { - struct bts_struct bts; - unsigned long index; - int error; - - if (((void *)at - trace->ds.begin) % trace->ds.size) { - printk(KERN_CONT - "read from non-integer index..."); - return -1; - } - index = ((void *)at - trace->ds.begin) / trace->ds.size; - - memset(&bts, 0, sizeof(bts)); - error = trace->read(tracer, at, &bts); - if (error < 0) { - printk(KERN_CONT - "error reading bts trace at [%lu] (0x%p)...", - index, at); - return error; - } - - switch (bts.qualifier) { - case BTS_BRANCH: - break; - default: - printk(KERN_CONT - "unexpected bts entry %llu at [%lu] (0x%p)...", - bts.qualifier, index, at); - return -1; - } - } - - return 0; -} - -static void ds_selftest_bts_cpu(void *arg) -{ - struct ds_selftest_bts_conf *conf = arg; - const struct bts_trace *trace; - void *top; - - if (IS_ERR(conf->tracer)) { - conf->error = PTR_ERR(conf->tracer); - conf->tracer = NULL; - - printk(KERN_CONT - "initialization failed (err: %d)...", conf->error); - return; - } - - /* We should meanwhile have enough trace. */ - conf->error = conf->suspend(conf->tracer); - if (conf->error < 0) - return; - - /* Let's see if we can access the trace. */ - trace = ds_read_bts(conf->tracer); - - conf->error = ds_selftest_bts_consistency(trace); - if (conf->error < 0) - return; - - /* If everything went well, we should have a few trace entries. */ - if (trace->ds.top == trace->ds.begin) { - /* - * It is possible but highly unlikely that we got a - * buffer overflow and end up at exactly the same - * position we started from. - * Let's issue a warning, but continue. - */ - printk(KERN_CONT "no trace/overflow..."); - } - - /* Let's try to read the trace we collected. */ - conf->error = - ds_selftest_bts_read(conf->tracer, trace, - trace->ds.begin, trace->ds.top); - if (conf->error < 0) - return; - - /* - * Let's read the trace again. - * Since we suspended tracing, we should get the same result. - */ - top = trace->ds.top; - - trace = ds_read_bts(conf->tracer); - conf->error = ds_selftest_bts_consistency(trace); - if (conf->error < 0) - return; - - if (top != trace->ds.top) { - printk(KERN_CONT "suspend not working..."); - conf->error = -1; - return; - } - - /* Let's collect some more trace - see if resume is working. */ - conf->error = conf->resume(conf->tracer); - if (conf->error < 0) - return; - - conf->error = conf->suspend(conf->tracer); - if (conf->error < 0) - return; - - trace = ds_read_bts(conf->tracer); - - conf->error = ds_selftest_bts_consistency(trace); - if (conf->error < 0) - return; - - if (trace->ds.top == top) { - /* - * It is possible but highly unlikely that we got a - * buffer overflow and end up at exactly the same - * position we started from. - * Let's issue a warning and check the full trace. - */ - printk(KERN_CONT - "no resume progress/overflow..."); - - conf->error = - ds_selftest_bts_read(conf->tracer, trace, - trace->ds.begin, trace->ds.end); - } else if (trace->ds.top < top) { - /* - * We had a buffer overflow - the entire buffer should - * contain trace records. - */ - conf->error = - ds_selftest_bts_read(conf->tracer, trace, - trace->ds.begin, trace->ds.end); - } else { - /* - * It is quite likely that the buffer did not overflow. - * Let's just check the delta trace. - */ - conf->error = - ds_selftest_bts_read(conf->tracer, trace, top, - trace->ds.top); - } - if (conf->error < 0) - return; - - conf->error = 0; -} - -static int ds_suspend_bts_wrap(struct bts_tracer *tracer) -{ - ds_suspend_bts(tracer); - return 0; -} - -static int ds_resume_bts_wrap(struct bts_tracer *tracer) -{ - ds_resume_bts(tracer); - return 0; -} - -static void ds_release_bts_noirq_wrap(void *tracer) -{ - (void)ds_release_bts_noirq(tracer); -} - -static int ds_selftest_bts_bad_release_noirq(int cpu, - struct bts_tracer *tracer) -{ - int error = -EPERM; - - /* Try to release the tracer on the wrong cpu. */ - get_cpu(); - if (cpu != smp_processor_id()) { - error = ds_release_bts_noirq(tracer); - if (error != -EPERM) - printk(KERN_CONT "release on wrong cpu..."); - } - put_cpu(); - - return error ? 0 : -1; -} - -static int ds_selftest_bts_bad_request_cpu(int cpu, void *buffer) -{ - struct bts_tracer *tracer; - int error; - - /* Try to request cpu tracing while task tracing is active. */ - tracer = ds_request_bts_cpu(cpu, buffer, BUFFER_SIZE, NULL, - (size_t)-1, BTS_KERNEL); - error = PTR_ERR(tracer); - if (!IS_ERR(tracer)) { - ds_release_bts(tracer); - error = 0; - } - - if (error != -EPERM) - printk(KERN_CONT "cpu/task tracing overlap..."); - - return error ? 0 : -1; -} - -static int ds_selftest_bts_bad_request_task(void *buffer) -{ - struct bts_tracer *tracer; - int error; - - /* Try to request cpu tracing while task tracing is active. */ - tracer = ds_request_bts_task(current, buffer, BUFFER_SIZE, NULL, - (size_t)-1, BTS_KERNEL); - error = PTR_ERR(tracer); - if (!IS_ERR(tracer)) { - error = 0; - ds_release_bts(tracer); - } - - if (error != -EPERM) - printk(KERN_CONT "task/cpu tracing overlap..."); - - return error ? 0 : -1; -} - -int ds_selftest_bts(void) -{ - struct ds_selftest_bts_conf conf; - unsigned char buffer[BUFFER_SIZE], *small_buffer; - unsigned long irq; - int cpu; - - printk(KERN_INFO "[ds] bts selftest..."); - conf.error = 0; - - small_buffer = (unsigned char *)ALIGN((unsigned long)buffer, 8) + 8; - - get_online_cpus(); - for_each_online_cpu(cpu) { - conf.suspend = ds_suspend_bts_wrap; - conf.resume = ds_resume_bts_wrap; - conf.tracer = - ds_request_bts_cpu(cpu, buffer, BUFFER_SIZE, - NULL, (size_t)-1, BTS_KERNEL); - ds_selftest_bts_cpu(&conf); - if (conf.error >= 0) - conf.error = ds_selftest_bts_bad_request_task(buffer); - ds_release_bts(conf.tracer); - if (conf.error < 0) - goto out; - - conf.suspend = ds_suspend_bts_noirq; - conf.resume = ds_resume_bts_noirq; - conf.tracer = - ds_request_bts_cpu(cpu, buffer, BUFFER_SIZE, - NULL, (size_t)-1, BTS_KERNEL); - smp_call_function_single(cpu, ds_selftest_bts_cpu, &conf, 1); - if (conf.error >= 0) { - conf.error = - ds_selftest_bts_bad_release_noirq(cpu, - conf.tracer); - /* We must not release the tracer twice. */ - if (conf.error < 0) - conf.tracer = NULL; - } - if (conf.error >= 0) - conf.error = ds_selftest_bts_bad_request_task(buffer); - smp_call_function_single(cpu, ds_release_bts_noirq_wrap, - conf.tracer, 1); - if (conf.error < 0) - goto out; - } - - conf.suspend = ds_suspend_bts_wrap; - conf.resume = ds_resume_bts_wrap; - conf.tracer = - ds_request_bts_task(current, buffer, BUFFER_SIZE, - NULL, (size_t)-1, BTS_KERNEL); - ds_selftest_bts_cpu(&conf); - if (conf.error >= 0) - conf.error = ds_selftest_bts_bad_request_cpu(0, buffer); - ds_release_bts(conf.tracer); - if (conf.error < 0) - goto out; - - conf.suspend = ds_suspend_bts_noirq; - conf.resume = ds_resume_bts_noirq; - conf.tracer = - ds_request_bts_task(current, small_buffer, SMALL_BUFFER_SIZE, - NULL, (size_t)-1, BTS_KERNEL); - local_irq_save(irq); - ds_selftest_bts_cpu(&conf); - if (conf.error >= 0) - conf.error = ds_selftest_bts_bad_request_cpu(0, buffer); - ds_release_bts_noirq(conf.tracer); - local_irq_restore(irq); - if (conf.error < 0) - goto out; - - conf.error = 0; - out: - put_online_cpus(); - printk(KERN_CONT "%s.\n", (conf.error ? "failed" : "passed")); - - return conf.error; -} - -int ds_selftest_pebs(void) -{ - return 0; -} diff --git a/arch/x86/kernel/ds_selftest.h b/arch/x86/kernel/ds_selftest.h deleted file mode 100644 index 2ba8745c666..00000000000 --- a/arch/x86/kernel/ds_selftest.h +++ /dev/null @@ -1,15 +0,0 @@ -/* - * Debug Store support - selftest - * - * - * Copyright (C) 2009 Intel Corporation. - * Markus Metzger <markus.t.metzger@intel.com>, 2009 - */ - -#ifdef CONFIG_X86_DS_SELFTEST -extern int ds_selftest_bts(void); -extern int ds_selftest_pebs(void); -#else -static inline int ds_selftest_bts(void) { return 0; } -static inline int ds_selftest_pebs(void) { return 0; } -#endif diff --git a/arch/x86/kernel/dumpstack.c b/arch/x86/kernel/dumpstack.c index 6d817554780..c89a386930b 100644 --- a/arch/x86/kernel/dumpstack.c +++ b/arch/x86/kernel/dumpstack.c @@ -224,11 +224,6 @@ unsigned __kprobes long oops_begin(void) int cpu; unsigned long flags; - /* notify the hw-branch tracer so it may disable tracing and - add the last trace to the trace buffer - - the earlier this happens, the more useful the trace. */ - trace_hw_branch_oops(); - oops_enter(); /* racy, but better than risking deadlock. */ diff --git a/arch/x86/kernel/dumpstack.h b/arch/x86/kernel/dumpstack.h index 4fd1420faff..e1a93be4fd4 100644 --- a/arch/x86/kernel/dumpstack.h +++ b/arch/x86/kernel/dumpstack.h @@ -14,6 +14,8 @@ #define get_bp(bp) asm("movq %%rbp, %0" : "=r" (bp) :) #endif +#include <linux/uaccess.h> + extern void show_trace_log_lvl(struct task_struct *task, struct pt_regs *regs, unsigned long *stack, unsigned long bp, char *log_lvl); @@ -29,4 +31,26 @@ struct stack_frame { struct stack_frame *next_frame; unsigned long return_address; }; + +struct stack_frame_ia32 { + u32 next_frame; + u32 return_address; +}; + +static inline unsigned long rewind_frame_pointer(int n) +{ + struct stack_frame *frame; + + get_bp(frame); + +#ifdef CONFIG_FRAME_POINTER + while (n--) { + if (probe_kernel_address(&frame->next_frame, frame)) + break; + } #endif + + return (unsigned long)frame; +} + +#endif /* DUMPSTACK_H */ diff --git a/arch/x86/kernel/dumpstack_32.c b/arch/x86/kernel/dumpstack_32.c index ae775ca47b2..11540a189d9 100644 --- a/arch/x86/kernel/dumpstack_32.c +++ b/arch/x86/kernel/dumpstack_32.c @@ -18,11 +18,6 @@ #include "dumpstack.h" -/* Just a stub for now */ -int x86_is_stack_id(int id, char *name) -{ - return 0; -} void dump_trace(struct task_struct *task, struct pt_regs *regs, unsigned long *stack, unsigned long bp, diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c index 0ad9597073f..272c9f1f05f 100644 --- a/arch/x86/kernel/dumpstack_64.c +++ b/arch/x86/kernel/dumpstack_64.c @@ -33,11 +33,6 @@ static char x86_stack_ids[][8] = { #endif }; -int x86_is_stack_id(int id, char *name) -{ - return x86_stack_ids[id - 1] == name; -} - static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack, unsigned *usedp, char **idp) { @@ -125,9 +120,15 @@ fixup_bp_irq_link(unsigned long bp, unsigned long *stack, { #ifdef CONFIG_FRAME_POINTER struct stack_frame *frame = (struct stack_frame *)bp; + unsigned long next; - if (!in_irq_stack(stack, irq_stack, irq_stack_end)) - return (unsigned long)frame->next_frame; + if (!in_irq_stack(stack, irq_stack, irq_stack_end)) { + if (!probe_kernel_address(&frame->next_frame, next)) + return next; + else + WARN_ONCE(1, "Perf: bad frame pointer = %p in " + "callchain\n", &frame->next_frame); + } #endif return bp; } @@ -207,7 +208,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, if (in_irq_stack(stack, irq_stack, irq_stack_end)) { if (ops->stack(data, "IRQ") < 0) break; - bp = print_context_stack(tinfo, stack, bp, + bp = ops->walk_stack(tinfo, stack, bp, ops, data, irq_stack_end, &graph); /* * We link to the next stack (which would be @@ -228,7 +229,7 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, /* * This handles the process stack: */ - bp = print_context_stack(tinfo, stack, bp, ops, data, NULL, &graph); + bp = ops->walk_stack(tinfo, stack, bp, ops, data, NULL, &graph); put_cpu(); } EXPORT_SYMBOL(dump_trace); @@ -291,6 +292,7 @@ void show_registers(struct pt_regs *regs) sp = regs->sp; printk("CPU %d ", cpu); + print_modules(); __show_regs(regs, 1); printk("Process %s (pid: %d, threadinfo %p, task %p)\n", cur->comm, cur->pid, task_thread_info(cur), cur); diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c index a1a7876cadc..7bca3c6a02f 100644 --- a/arch/x86/kernel/e820.c +++ b/arch/x86/kernel/e820.c @@ -12,21 +12,13 @@ #include <linux/types.h> #include <linux/init.h> #include <linux/bootmem.h> -#include <linux/ioport.h> -#include <linux/string.h> -#include <linux/kexec.h> -#include <linux/module.h> -#include <linux/mm.h> #include <linux/pfn.h> #include <linux/suspend.h> #include <linux/firmware-map.h> -#include <asm/pgtable.h> -#include <asm/page.h> #include <asm/e820.h> #include <asm/proto.h> #include <asm/setup.h> -#include <asm/trampoline.h> /* * The e820 map is the map that gets modified e.g. with command line parameters @@ -517,31 +509,55 @@ u64 __init e820_remove_range(u64 start, u64 size, unsigned old_type, int checktype) { int i; + u64 end; u64 real_removed_size = 0; if (size > (ULLONG_MAX - start)) size = ULLONG_MAX - start; + end = start + size; + printk(KERN_DEBUG "e820 remove range: %016Lx - %016Lx ", + (unsigned long long) start, + (unsigned long long) end); + if (checktype) + e820_print_type(old_type); + printk(KERN_CONT "\n"); + for (i = 0; i < e820.nr_map; i++) { struct e820entry *ei = &e820.map[i]; u64 final_start, final_end; + u64 ei_end; if (checktype && ei->type != old_type) continue; + + ei_end = ei->addr + ei->size; /* totally covered? */ - if (ei->addr >= start && - (ei->addr + ei->size) <= (start + size)) { + if (ei->addr >= start && ei_end <= end) { real_removed_size += ei->size; memset(ei, 0, sizeof(struct e820entry)); continue; } + + /* new range is totally covered? */ + if (ei->addr < start && ei_end > end) { + e820_add_region(end, ei_end - end, ei->type); + ei->size = start - ei->addr; + real_removed_size += size; + continue; + } + /* partially covered */ final_start = max(start, ei->addr); - final_end = min(start + size, ei->addr + ei->size); + final_end = min(end, ei_end); if (final_start >= final_end) continue; real_removed_size += final_end - final_start; + /* + * left range could be head or tail, so need to update + * size at first. + */ ei->size -= final_end - final_start; if (ei->addr < final_start) continue; @@ -722,319 +738,44 @@ core_initcall(e820_mark_nvs_memory); #endif /* - * Early reserved memory areas. - */ -#define MAX_EARLY_RES 32 - -struct early_res { - u64 start, end; - char name[16]; - char overlap_ok; -}; -static struct early_res early_res[MAX_EARLY_RES] __initdata = { - { 0, PAGE_SIZE, "BIOS data page", 1 }, /* BIOS data page */ -#if defined(CONFIG_X86_32) && defined(CONFIG_X86_TRAMPOLINE) - /* - * But first pinch a few for the stack/trampoline stuff - * FIXME: Don't need the extra page at 4K, but need to fix - * trampoline before removing it. (see the GDT stuff) - */ - { PAGE_SIZE, PAGE_SIZE + PAGE_SIZE, "EX TRAMPOLINE", 1 }, -#endif - - {} -}; - -static int __init find_overlapped_early(u64 start, u64 end) -{ - int i; - struct early_res *r; - - for (i = 0; i < MAX_EARLY_RES && early_res[i].end; i++) { - r = &early_res[i]; - if (end > r->start && start < r->end) - break; - } - - return i; -} - -/* - * Drop the i-th range from the early reservation map, - * by copying any higher ranges down one over it, and - * clearing what had been the last slot. - */ -static void __init drop_range(int i) -{ - int j; - - for (j = i + 1; j < MAX_EARLY_RES && early_res[j].end; j++) - ; - - memmove(&early_res[i], &early_res[i + 1], - (j - 1 - i) * sizeof(struct early_res)); - - early_res[j - 1].end = 0; -} - -/* - * Split any existing ranges that: - * 1) are marked 'overlap_ok', and - * 2) overlap with the stated range [start, end) - * into whatever portion (if any) of the existing range is entirely - * below or entirely above the stated range. Drop the portion - * of the existing range that overlaps with the stated range, - * which will allow the caller of this routine to then add that - * stated range without conflicting with any existing range. + * Find a free area with specified alignment in a specific range. */ -static void __init drop_overlaps_that_are_ok(u64 start, u64 end) +u64 __init find_e820_area(u64 start, u64 end, u64 size, u64 align) { int i; - struct early_res *r; - u64 lower_start, lower_end; - u64 upper_start, upper_end; - char name[16]; - for (i = 0; i < MAX_EARLY_RES && early_res[i].end; i++) { - r = &early_res[i]; + for (i = 0; i < e820.nr_map; i++) { + struct e820entry *ei = &e820.map[i]; + u64 addr; + u64 ei_start, ei_last; - /* Continue past non-overlapping ranges */ - if (end <= r->start || start >= r->end) + if (ei->type != E820_RAM) continue; - /* - * Leave non-ok overlaps as is; let caller - * panic "Overlapping early reservations" - * when it hits this overlap. - */ - if (!r->overlap_ok) - return; - - /* - * We have an ok overlap. We will drop it from the early - * reservation map, and add back in any non-overlapping - * portions (lower or upper) as separate, overlap_ok, - * non-overlapping ranges. - */ - - /* 1. Note any non-overlapping (lower or upper) ranges. */ - strncpy(name, r->name, sizeof(name) - 1); - - lower_start = lower_end = 0; - upper_start = upper_end = 0; - if (r->start < start) { - lower_start = r->start; - lower_end = start; - } - if (r->end > end) { - upper_start = end; - upper_end = r->end; - } - - /* 2. Drop the original ok overlapping range */ - drop_range(i); - - i--; /* resume for-loop on copied down entry */ - - /* 3. Add back in any non-overlapping ranges. */ - if (lower_end) - reserve_early_overlap_ok(lower_start, lower_end, name); - if (upper_end) - reserve_early_overlap_ok(upper_start, upper_end, name); - } -} - -static void __init __reserve_early(u64 start, u64 end, char *name, - int overlap_ok) -{ - int i; - struct early_res *r; - - i = find_overlapped_early(start, end); - if (i >= MAX_EARLY_RES) - panic("Too many early reservations"); - r = &early_res[i]; - if (r->end) - panic("Overlapping early reservations " - "%llx-%llx %s to %llx-%llx %s\n", - start, end - 1, name?name:"", r->start, - r->end - 1, r->name); - r->start = start; - r->end = end; - r->overlap_ok = overlap_ok; - if (name) - strncpy(r->name, name, sizeof(r->name) - 1); -} - -/* - * A few early reservtations come here. - * - * The 'overlap_ok' in the name of this routine does -not- mean it - * is ok for these reservations to overlap an earlier reservation. - * Rather it means that it is ok for subsequent reservations to - * overlap this one. - * - * Use this entry point to reserve early ranges when you are doing - * so out of "Paranoia", reserving perhaps more memory than you need, - * just in case, and don't mind a subsequent overlapping reservation - * that is known to be needed. - * - * The drop_overlaps_that_are_ok() call here isn't really needed. - * It would be needed if we had two colliding 'overlap_ok' - * reservations, so that the second such would not panic on the - * overlap with the first. We don't have any such as of this - * writing, but might as well tolerate such if it happens in - * the future. - */ -void __init reserve_early_overlap_ok(u64 start, u64 end, char *name) -{ - drop_overlaps_that_are_ok(start, end); - __reserve_early(start, end, name, 1); -} - -/* - * Most early reservations come here. - * - * We first have drop_overlaps_that_are_ok() drop any pre-existing - * 'overlap_ok' ranges, so that we can then reserve this memory - * range without risk of panic'ing on an overlapping overlap_ok - * early reservation. - */ -void __init reserve_early(u64 start, u64 end, char *name) -{ - if (start >= end) - return; - - drop_overlaps_that_are_ok(start, end); - __reserve_early(start, end, name, 0); -} - -void __init free_early(u64 start, u64 end) -{ - struct early_res *r; - int i; - - i = find_overlapped_early(start, end); - r = &early_res[i]; - if (i >= MAX_EARLY_RES || r->end != end || r->start != start) - panic("free_early on not reserved area: %llx-%llx!", - start, end - 1); - - drop_range(i); -} - -void __init early_res_to_bootmem(u64 start, u64 end) -{ - int i, count; - u64 final_start, final_end; - - count = 0; - for (i = 0; i < MAX_EARLY_RES && early_res[i].end; i++) - count++; - - printk(KERN_INFO "(%d early reservations) ==> bootmem [%010llx - %010llx]\n", - count, start, end); - for (i = 0; i < count; i++) { - struct early_res *r = &early_res[i]; - printk(KERN_INFO " #%d [%010llx - %010llx] %16s", i, - r->start, r->end, r->name); - final_start = max(start, r->start); - final_end = min(end, r->end); - if (final_start >= final_end) { - printk(KERN_CONT "\n"); - continue; - } - printk(KERN_CONT " ==> [%010llx - %010llx]\n", - final_start, final_end); - reserve_bootmem_generic(final_start, final_end - final_start, - BOOTMEM_DEFAULT); - } -} + ei_last = ei->addr + ei->size; + ei_start = ei->addr; + addr = find_early_area(ei_start, ei_last, start, end, + size, align); -/* Check for already reserved areas */ -static inline int __init bad_addr(u64 *addrp, u64 size, u64 align) -{ - int i; - u64 addr = *addrp; - int changed = 0; - struct early_res *r; -again: - i = find_overlapped_early(addr, addr + size); - r = &early_res[i]; - if (i < MAX_EARLY_RES && r->end) { - *addrp = addr = round_up(r->end, align); - changed = 1; - goto again; + if (addr != -1ULL) + return addr; } - return changed; + return -1ULL; } -/* Check for already reserved areas */ -static inline int __init bad_addr_size(u64 *addrp, u64 *sizep, u64 align) +u64 __init find_fw_memmap_area(u64 start, u64 end, u64 size, u64 align) { - int i; - u64 addr = *addrp, last; - u64 size = *sizep; - int changed = 0; -again: - last = addr + size; - for (i = 0; i < MAX_EARLY_RES && early_res[i].end; i++) { - struct early_res *r = &early_res[i]; - if (last > r->start && addr < r->start) { - size = r->start - addr; - changed = 1; - goto again; - } - if (last > r->end && addr < r->end) { - addr = round_up(r->end, align); - size = last - addr; - changed = 1; - goto again; - } - if (last <= r->end && addr >= r->start) { - (*sizep)++; - return 0; - } - } - if (changed) { - *addrp = addr; - *sizep = size; - } - return changed; + return find_e820_area(start, end, size, align); } -/* - * Find a free area with specified alignment in a specific range. - */ -u64 __init find_e820_area(u64 start, u64 end, u64 size, u64 align) +u64 __init get_max_mapped(void) { - int i; + u64 end = max_pfn_mapped; - for (i = 0; i < e820.nr_map; i++) { - struct e820entry *ei = &e820.map[i]; - u64 addr, last; - u64 ei_last; + end <<= PAGE_SHIFT; - if (ei->type != E820_RAM) - continue; - addr = round_up(ei->addr, align); - ei_last = ei->addr + ei->size; - if (addr < start) - addr = round_up(start, align); - if (addr >= ei_last) - continue; - while (bad_addr(&addr, size, align) && addr+size <= ei_last) - ; - last = addr + size; - if (last > ei_last) - continue; - if (last > end) - continue; - return addr; - } - return -1ULL; + return end; } - /* * Find next free range after *start */ @@ -1044,25 +785,19 @@ u64 __init find_e820_area_size(u64 start, u64 *sizep, u64 align) for (i = 0; i < e820.nr_map; i++) { struct e820entry *ei = &e820.map[i]; - u64 addr, last; - u64 ei_last; + u64 addr; + u64 ei_start, ei_last; if (ei->type != E820_RAM) continue; - addr = round_up(ei->addr, align); + ei_last = ei->addr + ei->size; - if (addr < start) - addr = round_up(start, align); - if (addr >= ei_last) - continue; - *sizep = ei_last - addr; - while (bad_addr_size(&addr, sizep, align) && - addr + *sizep <= ei_last) - ; - last = addr + *sizep; - if (last > ei_last) - continue; - return addr; + ei_start = ei->addr; + addr = find_early_area_size(ei_start, ei_last, start, + sizep, align); + + if (addr != -1ULL) + return addr; } return -1ULL; @@ -1421,6 +1156,8 @@ void __init e820_reserve_resources_late(void) end = MAX_RESOURCE_SIZE; if (start >= end) continue; + printk(KERN_DEBUG "reserve RAM buffer: %016llx - %016llx ", + start, end); reserve_region_with_split(&iomem_resource, start, end, "RAM buffer"); } diff --git a/arch/x86/kernel/early_printk.c b/arch/x86/kernel/early_printk.c index b9c830c12b4..fa99bae75ac 100644 --- a/arch/x86/kernel/early_printk.c +++ b/arch/x86/kernel/early_printk.c @@ -41,6 +41,14 @@ static void early_vga_write(struct console *con, const char *str, unsigned n) writew(0x720, VGABASE + 2*(max_xpos*j + i)); current_ypos = max_ypos-1; } +#ifdef CONFIG_KGDB_KDB + if (c == '\b') { + if (current_xpos > 0) + current_xpos--; + } else if (c == '\r') { + current_xpos = 0; + } else +#endif if (c == '\n') { current_xpos = 0; current_ypos++; diff --git a/arch/x86/kernel/efi.c b/arch/x86/kernel/efi.c index cdcfb122f25..c2fa9b8b497 100644 --- a/arch/x86/kernel/efi.c +++ b/arch/x86/kernel/efi.c @@ -362,7 +362,7 @@ void __init efi_init(void) printk(KERN_ERR PFX "Could not map the firmware vendor!\n"); early_iounmap(tmp, 2); - printk(KERN_INFO "EFI v%u.%.02u by %s \n", + printk(KERN_INFO "EFI v%u.%.02u by %s\n", efi.systab->hdr.revision >> 16, efi.systab->hdr.revision & 0xffff, vendor); diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S index 44a8e0dc673..cd49141cf15 100644 --- a/arch/x86/kernel/entry_32.S +++ b/arch/x86/kernel/entry_32.S @@ -53,6 +53,7 @@ #include <asm/processor-flags.h> #include <asm/ftrace.h> #include <asm/irq_vectors.h> +#include <asm/cpufeature.h> /* Avoid __ASSEMBLER__'ifying <linux/audit.h> just for this. */ #include <linux/elf-em.h> @@ -905,7 +906,25 @@ ENTRY(simd_coprocessor_error) RING0_INT_FRAME pushl $0 CFI_ADJUST_CFA_OFFSET 4 +#ifdef CONFIG_X86_INVD_BUG + /* AMD 486 bug: invd from userspace calls exception 19 instead of #GP */ +661: pushl $do_general_protection +662: +.section .altinstructions,"a" + .balign 4 + .long 661b + .long 663f + .byte X86_FEATURE_XMM + .byte 662b-661b + .byte 664f-663f +.previous +.section .altinstr_replacement,"ax" +663: pushl $do_simd_coprocessor_error +664: +.previous +#else pushl $do_simd_coprocessor_error +#endif CFI_ADJUST_CFA_OFFSET 4 jmp error_code CFI_ENDPROC diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c index 30968924543..cd37469b54e 100644 --- a/arch/x86/kernel/ftrace.c +++ b/arch/x86/kernel/ftrace.c @@ -30,14 +30,32 @@ #ifdef CONFIG_DYNAMIC_FTRACE +/* + * modifying_code is set to notify NMIs that they need to use + * memory barriers when entering or exiting. But we don't want + * to burden NMIs with unnecessary memory barriers when code + * modification is not being done (which is most of the time). + * + * A mutex is already held when ftrace_arch_code_modify_prepare + * and post_process are called. No locks need to be taken here. + * + * Stop machine will make sure currently running NMIs are done + * and new NMIs will see the updated variable before we need + * to worry about NMIs doing memory barriers. + */ +static int modifying_code __read_mostly; +static DEFINE_PER_CPU(int, save_modifying_code); + int ftrace_arch_code_modify_prepare(void) { set_kernel_text_rw(); + modifying_code = 1; return 0; } int ftrace_arch_code_modify_post_process(void) { + modifying_code = 0; set_kernel_text_ro(); return 0; } @@ -149,6 +167,11 @@ static void ftrace_mod_code(void) void ftrace_nmi_enter(void) { + __get_cpu_var(save_modifying_code) = modifying_code; + + if (!__get_cpu_var(save_modifying_code)) + return; + if (atomic_inc_return(&nmi_running) & MOD_CODE_WRITE_FLAG) { smp_rmb(); ftrace_mod_code(); @@ -160,6 +183,9 @@ void ftrace_nmi_enter(void) void ftrace_nmi_exit(void) { + if (!__get_cpu_var(save_modifying_code)) + return; + /* Finish all executions before clearing nmi_running */ smp_mb(); atomic_dec(&nmi_running); @@ -484,13 +510,3 @@ void prepare_ftrace_return(unsigned long *parent, unsigned long self_addr, } } #endif /* CONFIG_FUNCTION_GRAPH_TRACER */ - -#ifdef CONFIG_FTRACE_SYSCALLS - -extern unsigned long *sys_call_table; - -unsigned long __init arch_syscall_addr(int nr) -{ - return (unsigned long)(&sys_call_table)[nr]; -} -#endif diff --git a/arch/x86/kernel/head32.c b/arch/x86/kernel/head32.c index 5051b94c906..b2e24603739 100644 --- a/arch/x86/kernel/head32.c +++ b/arch/x86/kernel/head32.c @@ -7,6 +7,7 @@ #include <linux/init.h> #include <linux/start_kernel.h> +#include <linux/mm.h> #include <asm/setup.h> #include <asm/sections.h> @@ -29,14 +30,25 @@ static void __init i386_default_early_setup(void) void __init i386_start_kernel(void) { +#ifdef CONFIG_X86_TRAMPOLINE + /* + * But first pinch a few for the stack/trampoline stuff + * FIXME: Don't need the extra page at 4K, but need to fix + * trampoline before removing it. (see the GDT stuff) + */ + reserve_early_overlap_ok(PAGE_SIZE, PAGE_SIZE + PAGE_SIZE, + "EX TRAMPOLINE"); +#endif + reserve_early(__pa_symbol(&_text), __pa_symbol(&__bss_stop), "TEXT DATA BSS"); #ifdef CONFIG_BLK_DEV_INITRD /* Reserve INITRD */ if (boot_params.hdr.type_of_loader && boot_params.hdr.ramdisk_image) { + /* Assume only end is not page aligned */ u64 ramdisk_image = boot_params.hdr.ramdisk_image; u64 ramdisk_size = boot_params.hdr.ramdisk_size; - u64 ramdisk_end = ramdisk_image + ramdisk_size; + u64 ramdisk_end = PAGE_ALIGN(ramdisk_image + ramdisk_size); reserve_early(ramdisk_image, ramdisk_end, "RAMDISK"); } #endif diff --git a/arch/x86/kernel/head64.c b/arch/x86/kernel/head64.c index b5a9896ca1e..7147143fd61 100644 --- a/arch/x86/kernel/head64.c +++ b/arch/x86/kernel/head64.c @@ -103,9 +103,10 @@ void __init x86_64_start_reservations(char *real_mode_data) #ifdef CONFIG_BLK_DEV_INITRD /* Reserve INITRD */ if (boot_params.hdr.type_of_loader && boot_params.hdr.ramdisk_image) { + /* Assume only end is not page aligned */ unsigned long ramdisk_image = boot_params.hdr.ramdisk_image; unsigned long ramdisk_size = boot_params.hdr.ramdisk_size; - unsigned long ramdisk_end = ramdisk_image + ramdisk_size; + unsigned long ramdisk_end = PAGE_ALIGN(ramdisk_image + ramdisk_size); reserve_early(ramdisk_image, ramdisk_end, "RAMDISK"); } #endif diff --git a/arch/x86/kernel/head_32.S b/arch/x86/kernel/head_32.S index 7fd318bac59..37c3d4b17d8 100644 --- a/arch/x86/kernel/head_32.S +++ b/arch/x86/kernel/head_32.S @@ -442,8 +442,8 @@ is386: movl $2,%ecx # set MP */ cmpb $0,ready jne 1f - movl $per_cpu__gdt_page,%eax - movl $per_cpu__stack_canary,%ecx + movl $gdt_page,%eax + movl $stack_canary,%ecx movw %cx, 8 * GDT_ENTRY_STACK_CANARY + 2(%eax) shrl $16, %ecx movb %cl, 8 * GDT_ENTRY_STACK_CANARY + 4(%eax) @@ -706,7 +706,7 @@ idt_descr: .word 0 # 32 bit align gdt_desc.address ENTRY(early_gdt_descr) .word GDT_ENTRIES*8-1 - .long per_cpu__gdt_page /* Overwritten for secondary CPUs */ + .long gdt_page /* Overwritten for secondary CPUs */ /* * The boot_gdt must mirror the equivalent in setup.S and is diff --git a/arch/x86/kernel/head_64.S b/arch/x86/kernel/head_64.S index 2d8b5035371..3d1e6f16b7a 100644 --- a/arch/x86/kernel/head_64.S +++ b/arch/x86/kernel/head_64.S @@ -27,7 +27,7 @@ #define GET_CR2_INTO_RCX movq %cr2, %rcx #endif -/* we are not able to switch in one step to the final KERNEL ADRESS SPACE +/* we are not able to switch in one step to the final KERNEL ADDRESS SPACE * because we need identity-mapped pages. * */ diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index ad80a1c718c..a198b7c87a1 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -4,6 +4,7 @@ #include <linux/sysdev.h> #include <linux/delay.h> #include <linux/errno.h> +#include <linux/slab.h> #include <linux/hpet.h> #include <linux/init.h> #include <linux/cpu.h> @@ -35,6 +36,7 @@ unsigned long hpet_address; u8 hpet_blockid; /* OS timer block num */ u8 hpet_msi_disable; +u8 hpet_readback_cmp; #ifdef CONFIG_PCI_MSI static unsigned long hpet_num_timers; @@ -266,7 +268,7 @@ static void hpet_resume_device(void) force_hpet_resume(); } -static void hpet_resume_counter(void) +static void hpet_resume_counter(struct clocksource *cs) { hpet_resume_device(); hpet_restart_counter(); @@ -394,14 +396,24 @@ static int hpet_next_event(unsigned long delta, * at that point and we would wait for the next hpet interrupt * forever. We found out that reading the CMP register back * forces the transfer so we can rely on the comparison with - * the counter register below. If the read back from the - * compare register does not match the value we programmed - * then we might have a real hardware problem. We can not do - * much about it here, but at least alert the user/admin with - * a prominent warning. + * the counter register below. + * + * That works fine on those ATI chipsets, but on newer Intel + * chipsets (ICH9...) this triggers due to an erratum: Reading + * the comparator immediately following a write is returning + * the old value. + * + * We restrict the read back to the affected ATI chipsets (set + * by quirks) and also run it with hpet=verbose for debugging + * purposes. */ - WARN_ONCE(hpet_readl(HPET_Tn_CMP(timer)) != cnt, - KERN_WARNING "hpet: compare register read back failed.\n"); + if (hpet_readback_cmp || hpet_verbose) { + u32 cmp = hpet_readl(HPET_Tn_CMP(timer)); + + if (cmp != cnt) + printk_once(KERN_WARNING + "hpet: compare register read back failed.\n"); + } return (s32)(hpet_readl(HPET_COUNTER) - cnt) >= 0 ? -ETIME : 0; } @@ -1143,6 +1155,7 @@ int hpet_set_periodic_freq(unsigned long freq) do_div(clc, freq); clc >>= hpet_clockevent.shift; hpet_pie_delta = clc; + hpet_pie_limit = 0; } return 1; } diff --git a/arch/x86/kernel/hw_breakpoint.c b/arch/x86/kernel/hw_breakpoint.c index 05d5fec64a9..a8f1b803d2f 100644 --- a/arch/x86/kernel/hw_breakpoint.c +++ b/arch/x86/kernel/hw_breakpoint.c @@ -189,48 +189,20 @@ static int get_hbp_len(u8 hbp_len) } /* - * Check for virtual address in user space. - */ -int arch_check_va_in_userspace(unsigned long va, u8 hbp_len) -{ - unsigned int len; - - len = get_hbp_len(hbp_len); - - return (va <= TASK_SIZE - len); -} - -/* * Check for virtual address in kernel space. */ -static int arch_check_va_in_kernelspace(unsigned long va, u8 hbp_len) +int arch_check_bp_in_kernelspace(struct perf_event *bp) { unsigned int len; + unsigned long va; + struct arch_hw_breakpoint *info = counter_arch_bp(bp); - len = get_hbp_len(hbp_len); + va = info->address; + len = get_hbp_len(info->len); return (va >= TASK_SIZE) && ((va + len - 1) >= TASK_SIZE); } -/* - * Store a breakpoint's encoded address, length, and type. - */ -static int arch_store_info(struct perf_event *bp) -{ - struct arch_hw_breakpoint *info = counter_arch_bp(bp); - /* - * For kernel-addresses, either the address or symbol name can be - * specified. - */ - if (info->name) - info->address = (unsigned long) - kallsyms_lookup_name(info->name); - if (info->address) - return 0; - - return -EINVAL; -} - int arch_bp_generic_fields(int x86_len, int x86_type, int *gen_len, int *gen_type) { @@ -319,8 +291,7 @@ static int arch_build_bp_info(struct perf_event *bp) /* * Validate the arch-specific HW Breakpoint register settings */ -int arch_validate_hwbkpt_settings(struct perf_event *bp, - struct task_struct *tsk) +int arch_validate_hwbkpt_settings(struct perf_event *bp) { struct arch_hw_breakpoint *info = counter_arch_bp(bp); unsigned int align; @@ -333,16 +304,6 @@ int arch_validate_hwbkpt_settings(struct perf_event *bp, ret = -EINVAL; - if (info->type == X86_BREAKPOINT_EXECUTE) - /* - * Ptrace-refactoring code - * For now, we'll allow instruction breakpoint only for user-space - * addresses - */ - if ((!arch_check_va_in_userspace(info->address, info->len)) && - info->len != X86_BREAKPOINT_EXECUTE) - return ret; - switch (info->len) { case X86_BREAKPOINT_LEN_1: align = 0; @@ -362,10 +323,6 @@ int arch_validate_hwbkpt_settings(struct perf_event *bp, return ret; } - ret = arch_store_info(bp); - - if (ret < 0) - return ret; /* * Check that the low-order bits of the address are appropriate * for the alignment implied by len. @@ -373,15 +330,6 @@ int arch_validate_hwbkpt_settings(struct perf_event *bp, if (info->address & align) return -EINVAL; - /* Check that the virtual address is in the proper range */ - if (tsk) { - if (!arch_check_va_in_userspace(info->address, info->len)) - return -EFAULT; - } else { - if (!arch_check_va_in_kernelspace(info->address, info->len)) - return -EFAULT; - } - return 0; } @@ -502,8 +450,6 @@ static int __kprobes hw_breakpoint_handler(struct die_args *args) rcu_read_lock(); bp = per_cpu(bp_per_reg[i], cpu); - if (bp) - rc = NOTIFY_DONE; /* * Reset the 'i'th TRAP bit in dr6 to denote completion of * exception handling @@ -522,7 +468,13 @@ static int __kprobes hw_breakpoint_handler(struct die_args *args) rcu_read_unlock(); } - if (dr6 & (~DR_TRAP_BITS)) + /* + * Further processing in do_debug() is needed for a) user-space + * breakpoints (to generate signals) and b) when the system has + * taken exception due to multiple causes + */ + if ((current->thread.debugreg6 & DR_TRAP_BITS) || + (dr6 & (~DR_TRAP_BITS))) rc = NOTIFY_DONE; set_debugreg(dr7, 7); @@ -547,8 +499,3 @@ void hw_breakpoint_pmu_read(struct perf_event *bp) { /* TODO */ } - -void hw_breakpoint_pmu_unthrottle(struct perf_event *bp) -{ - /* TODO */ -} diff --git a/arch/x86/kernel/i387.c b/arch/x86/kernel/i387.c index f2f8540a7f3..86cef6b3225 100644 --- a/arch/x86/kernel/i387.c +++ b/arch/x86/kernel/i387.c @@ -8,6 +8,7 @@ #include <linux/module.h> #include <linux/regset.h> #include <linux/sched.h> +#include <linux/slab.h> #include <asm/sigcontext.h> #include <asm/processor.h> @@ -101,69 +102,71 @@ void __cpuinit fpu_init(void) mxcsr_feature_mask_init(); /* clean state in init */ - if (cpu_has_xsave) - current_thread_info()->status = TS_XSAVE; - else - current_thread_info()->status = 0; + current_thread_info()->status = 0; clear_used_math(); } #endif /* CONFIG_X86_64 */ -/* - * The _current_ task is using the FPU for the first time - * so initialize it and set the mxcsr to its default - * value at reset if we support XMM instructions and then - * remeber the current task has used the FPU. - */ -int init_fpu(struct task_struct *tsk) +static void fpu_finit(struct fpu *fpu) { - if (tsk_used_math(tsk)) { - if (HAVE_HWFP && tsk == current) - unlazy_fpu(tsk); - return 0; - } - - /* - * Memory allocation at the first usage of the FPU and other state. - */ - if (!tsk->thread.xstate) { - tsk->thread.xstate = kmem_cache_alloc(task_xstate_cachep, - GFP_KERNEL); - if (!tsk->thread.xstate) - return -ENOMEM; - } - #ifdef CONFIG_X86_32 if (!HAVE_HWFP) { - memset(tsk->thread.xstate, 0, xstate_size); - finit_task(tsk); - set_stopped_child_used_math(tsk); - return 0; + finit_soft_fpu(&fpu->state->soft); + return; } #endif if (cpu_has_fxsr) { - struct i387_fxsave_struct *fx = &tsk->thread.xstate->fxsave; + struct i387_fxsave_struct *fx = &fpu->state->fxsave; memset(fx, 0, xstate_size); fx->cwd = 0x37f; if (cpu_has_xmm) fx->mxcsr = MXCSR_DEFAULT; } else { - struct i387_fsave_struct *fp = &tsk->thread.xstate->fsave; + struct i387_fsave_struct *fp = &fpu->state->fsave; memset(fp, 0, xstate_size); fp->cwd = 0xffff037fu; fp->swd = 0xffff0000u; fp->twd = 0xffffffffu; fp->fos = 0xffff0000u; } +} + +/* + * The _current_ task is using the FPU for the first time + * so initialize it and set the mxcsr to its default + * value at reset if we support XMM instructions and then + * remeber the current task has used the FPU. + */ +int init_fpu(struct task_struct *tsk) +{ + int ret; + + if (tsk_used_math(tsk)) { + if (HAVE_HWFP && tsk == current) + unlazy_fpu(tsk); + return 0; + } + /* - * Only the device not available exception or ptrace can call init_fpu. + * Memory allocation at the first usage of the FPU and other state. */ + ret = fpu_alloc(&tsk->thread.fpu); + if (ret) + return ret; + + fpu_finit(&tsk->thread.fpu); + set_stopped_child_used_math(tsk); return 0; } +/* + * The xstateregs_active() routine is the same as the fpregs_active() routine, + * as the "regset->n" for the xstate regset will be updated based on the feature + * capabilites supported by the xsave. + */ int fpregs_active(struct task_struct *target, const struct user_regset *regset) { return tsk_used_math(target) ? regset->n : 0; @@ -188,7 +191,7 @@ int xfpregs_get(struct task_struct *target, const struct user_regset *regset, return ret; return user_regset_copyout(&pos, &count, &kbuf, &ubuf, - &target->thread.xstate->fxsave, 0, -1); + &target->thread.fpu.state->fxsave, 0, -1); } int xfpregs_set(struct task_struct *target, const struct user_regset *regset, @@ -204,22 +207,82 @@ int xfpregs_set(struct task_struct *target, const struct user_regset *regset, if (ret) return ret; - set_stopped_child_used_math(target); - ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, - &target->thread.xstate->fxsave, 0, -1); + &target->thread.fpu.state->fxsave, 0, -1); /* * mxcsr reserved bits must be masked to zero for security reasons. */ - target->thread.xstate->fxsave.mxcsr &= mxcsr_feature_mask; + target->thread.fpu.state->fxsave.mxcsr &= mxcsr_feature_mask; /* * update the header bits in the xsave header, indicating the * presence of FP and SSE state. */ if (cpu_has_xsave) - target->thread.xstate->xsave.xsave_hdr.xstate_bv |= XSTATE_FPSSE; + target->thread.fpu.state->xsave.xsave_hdr.xstate_bv |= XSTATE_FPSSE; + + return ret; +} + +int xstateregs_get(struct task_struct *target, const struct user_regset *regset, + unsigned int pos, unsigned int count, + void *kbuf, void __user *ubuf) +{ + int ret; + + if (!cpu_has_xsave) + return -ENODEV; + + ret = init_fpu(target); + if (ret) + return ret; + + /* + * Copy the 48bytes defined by the software first into the xstate + * memory layout in the thread struct, so that we can copy the entire + * xstateregs to the user using one user_regset_copyout(). + */ + memcpy(&target->thread.fpu.state->fxsave.sw_reserved, + xstate_fx_sw_bytes, sizeof(xstate_fx_sw_bytes)); + + /* + * Copy the xstate memory layout. + */ + ret = user_regset_copyout(&pos, &count, &kbuf, &ubuf, + &target->thread.fpu.state->xsave, 0, -1); + return ret; +} + +int xstateregs_set(struct task_struct *target, const struct user_regset *regset, + unsigned int pos, unsigned int count, + const void *kbuf, const void __user *ubuf) +{ + int ret; + struct xsave_hdr_struct *xsave_hdr; + + if (!cpu_has_xsave) + return -ENODEV; + + ret = init_fpu(target); + if (ret) + return ret; + + ret = user_regset_copyin(&pos, &count, &kbuf, &ubuf, + &target->thread.fpu.state->xsave, 0, -1); + + /* + * mxcsr reserved bits must be masked to zero for security reasons. + */ + target->thread.fpu.state->fxsave.mxcsr &= mxcsr_feature_mask; + + xsave_hdr = &target->thread.fpu.state->xsave.xsave_hdr; + + xsave_hdr->xstate_bv &= pcntxt_mask; + /* + * These bits must be zero. + */ + xsave_hdr->reserved1[0] = xsave_hdr->reserved1[1] = 0; return ret; } @@ -299,7 +362,7 @@ static inline u32 twd_fxsr_to_i387(struct i387_fxsave_struct *fxsave) static void convert_from_fxsr(struct user_i387_ia32_struct *env, struct task_struct *tsk) { - struct i387_fxsave_struct *fxsave = &tsk->thread.xstate->fxsave; + struct i387_fxsave_struct *fxsave = &tsk->thread.fpu.state->fxsave; struct _fpreg *to = (struct _fpreg *) &env->st_space[0]; struct _fpxreg *from = (struct _fpxreg *) &fxsave->st_space[0]; int i; @@ -339,7 +402,7 @@ static void convert_to_fxsr(struct task_struct *tsk, const struct user_i387_ia32_struct *env) { - struct i387_fxsave_struct *fxsave = &tsk->thread.xstate->fxsave; + struct i387_fxsave_struct *fxsave = &tsk->thread.fpu.state->fxsave; struct _fpreg *from = (struct _fpreg *) &env->st_space[0]; struct _fpxreg *to = (struct _fpxreg *) &fxsave->st_space[0]; int i; @@ -379,7 +442,7 @@ int fpregs_get(struct task_struct *target, const struct user_regset *regset, if (!cpu_has_fxsr) { return user_regset_copyout(&pos, &count, &kbuf, &ubuf, - &target->thread.xstate->fsave, 0, + &target->thread.fpu.state->fsave, 0, -1); } @@ -404,14 +467,12 @@ int fpregs_set(struct task_struct *target, const struct user_regset *regset, if (ret) return ret; - set_stopped_child_used_math(target); - if (!HAVE_HWFP) return fpregs_soft_set(target, regset, pos, count, kbuf, ubuf); if (!cpu_has_fxsr) { return user_regset_copyin(&pos, &count, &kbuf, &ubuf, - &target->thread.xstate->fsave, 0, -1); + &target->thread.fpu.state->fsave, 0, -1); } if (pos > 0 || count < sizeof(env)) @@ -426,7 +487,7 @@ int fpregs_set(struct task_struct *target, const struct user_regset *regset, * presence of FP. */ if (cpu_has_xsave) - target->thread.xstate->xsave.xsave_hdr.xstate_bv |= XSTATE_FP; + target->thread.fpu.state->xsave.xsave_hdr.xstate_bv |= XSTATE_FP; return ret; } @@ -437,7 +498,7 @@ int fpregs_set(struct task_struct *target, const struct user_regset *regset, static inline int save_i387_fsave(struct _fpstate_ia32 __user *buf) { struct task_struct *tsk = current; - struct i387_fsave_struct *fp = &tsk->thread.xstate->fsave; + struct i387_fsave_struct *fp = &tsk->thread.fpu.state->fsave; fp->status = fp->swd; if (__copy_to_user(buf, fp, sizeof(struct i387_fsave_struct))) @@ -448,7 +509,7 @@ static inline int save_i387_fsave(struct _fpstate_ia32 __user *buf) static int save_i387_fxsave(struct _fpstate_ia32 __user *buf) { struct task_struct *tsk = current; - struct i387_fxsave_struct *fx = &tsk->thread.xstate->fxsave; + struct i387_fxsave_struct *fx = &tsk->thread.fpu.state->fxsave; struct user_i387_ia32_struct env; int err = 0; @@ -483,7 +544,7 @@ static int save_i387_xsave(void __user *buf) * header as well as change any contents in the memory layout. * xrestore as part of sigreturn will capture all the changes. */ - tsk->thread.xstate->xsave.xsave_hdr.xstate_bv |= XSTATE_FPSSE; + tsk->thread.fpu.state->xsave.xsave_hdr.xstate_bv |= XSTATE_FPSSE; if (save_i387_fxsave(fx) < 0) return -1; @@ -535,7 +596,7 @@ static inline int restore_i387_fsave(struct _fpstate_ia32 __user *buf) { struct task_struct *tsk = current; - return __copy_from_user(&tsk->thread.xstate->fsave, buf, + return __copy_from_user(&tsk->thread.fpu.state->fsave, buf, sizeof(struct i387_fsave_struct)); } @@ -546,10 +607,10 @@ static int restore_i387_fxsave(struct _fpstate_ia32 __user *buf, struct user_i387_ia32_struct env; int err; - err = __copy_from_user(&tsk->thread.xstate->fxsave, &buf->_fxsr_env[0], + err = __copy_from_user(&tsk->thread.fpu.state->fxsave, &buf->_fxsr_env[0], size); /* mxcsr reserved bits must be masked to zero for security reasons */ - tsk->thread.xstate->fxsave.mxcsr &= mxcsr_feature_mask; + tsk->thread.fpu.state->fxsave.mxcsr &= mxcsr_feature_mask; if (err || __copy_from_user(&env, buf, sizeof(env))) return 1; convert_to_fxsr(tsk, &env); @@ -565,7 +626,7 @@ static int restore_i387_xsave(void __user *buf) struct i387_fxsave_struct __user *fx = (struct i387_fxsave_struct __user *) &fx_user->_fxsr_env[0]; struct xsave_hdr_struct *xsave_hdr = - ¤t->thread.xstate->xsave.xsave_hdr; + ¤t->thread.fpu.state->xsave.xsave_hdr; u64 mask; int err; diff --git a/arch/x86/kernel/i8253.c b/arch/x86/kernel/i8253.c index 23c167925a5..2dfd3159744 100644 --- a/arch/x86/kernel/i8253.c +++ b/arch/x86/kernel/i8253.c @@ -16,7 +16,7 @@ #include <asm/hpet.h> #include <asm/smp.h> -DEFINE_SPINLOCK(i8253_lock); +DEFINE_RAW_SPINLOCK(i8253_lock); EXPORT_SYMBOL(i8253_lock); /* @@ -33,7 +33,7 @@ struct clock_event_device *global_clock_event; static void init_pit_timer(enum clock_event_mode mode, struct clock_event_device *evt) { - spin_lock(&i8253_lock); + raw_spin_lock(&i8253_lock); switch (mode) { case CLOCK_EVT_MODE_PERIODIC: @@ -62,7 +62,7 @@ static void init_pit_timer(enum clock_event_mode mode, /* Nothing to do here */ break; } - spin_unlock(&i8253_lock); + raw_spin_unlock(&i8253_lock); } /* @@ -72,10 +72,10 @@ static void init_pit_timer(enum clock_event_mode mode, */ static int pit_next_event(unsigned long delta, struct clock_event_device *evt) { - spin_lock(&i8253_lock); + raw_spin_lock(&i8253_lock); outb_pit(delta & 0xff , PIT_CH0); /* LSB */ outb_pit(delta >> 8 , PIT_CH0); /* MSB */ - spin_unlock(&i8253_lock); + raw_spin_unlock(&i8253_lock); return 0; } @@ -130,7 +130,7 @@ static cycle_t pit_read(struct clocksource *cs) int count; u32 jifs; - spin_lock_irqsave(&i8253_lock, flags); + raw_spin_lock_irqsave(&i8253_lock, flags); /* * Although our caller may have the read side of xtime_lock, * this is now a seqlock, and we are cheating in this routine @@ -176,7 +176,7 @@ static cycle_t pit_read(struct clocksource *cs) old_count = count; old_jifs = jifs; - spin_unlock_irqrestore(&i8253_lock, flags); + raw_spin_unlock_irqrestore(&i8253_lock, flags); count = (LATCH - 1) - count; diff --git a/arch/x86/kernel/i8259.c b/arch/x86/kernel/i8259.c index df89102bef8..7c9f02c130f 100644 --- a/arch/x86/kernel/i8259.c +++ b/arch/x86/kernel/i8259.c @@ -5,7 +5,6 @@ #include <linux/ioport.h> #include <linux/interrupt.h> #include <linux/timex.h> -#include <linux/slab.h> #include <linux/random.h> #include <linux/init.h> #include <linux/kernel_stat.h> @@ -32,8 +31,14 @@ */ static int i8259A_auto_eoi; -DEFINE_SPINLOCK(i8259A_lock); +DEFINE_RAW_SPINLOCK(i8259A_lock); static void mask_and_ack_8259A(unsigned int); +static void mask_8259A(void); +static void unmask_8259A(void); +static void disable_8259A_irq(unsigned int irq); +static void enable_8259A_irq(unsigned int irq); +static void init_8259A(int auto_eoi); +static int i8259A_irq_pending(unsigned int irq); struct irq_chip i8259A_chip = { .name = "XT-PIC", @@ -63,51 +68,51 @@ unsigned int cached_irq_mask = 0xffff; */ unsigned long io_apic_irqs; -void disable_8259A_irq(unsigned int irq) +static void disable_8259A_irq(unsigned int irq) { unsigned int mask = 1 << irq; unsigned long flags; - spin_lock_irqsave(&i8259A_lock, flags); + raw_spin_lock_irqsave(&i8259A_lock, flags); cached_irq_mask |= mask; if (irq & 8) outb(cached_slave_mask, PIC_SLAVE_IMR); else outb(cached_master_mask, PIC_MASTER_IMR); - spin_unlock_irqrestore(&i8259A_lock, flags); + raw_spin_unlock_irqrestore(&i8259A_lock, flags); } -void enable_8259A_irq(unsigned int irq) +static void enable_8259A_irq(unsigned int irq) { unsigned int mask = ~(1 << irq); unsigned long flags; - spin_lock_irqsave(&i8259A_lock, flags); + raw_spin_lock_irqsave(&i8259A_lock, flags); cached_irq_mask &= mask; if (irq & 8) outb(cached_slave_mask, PIC_SLAVE_IMR); else outb(cached_master_mask, PIC_MASTER_IMR); - spin_unlock_irqrestore(&i8259A_lock, flags); + raw_spin_unlock_irqrestore(&i8259A_lock, flags); } -int i8259A_irq_pending(unsigned int irq) +static int i8259A_irq_pending(unsigned int irq) { unsigned int mask = 1<<irq; unsigned long flags; int ret; - spin_lock_irqsave(&i8259A_lock, flags); + raw_spin_lock_irqsave(&i8259A_lock, flags); if (irq < 8) ret = inb(PIC_MASTER_CMD) & mask; else ret = inb(PIC_SLAVE_CMD) & (mask >> 8); - spin_unlock_irqrestore(&i8259A_lock, flags); + raw_spin_unlock_irqrestore(&i8259A_lock, flags); return ret; } -void make_8259A_irq(unsigned int irq) +static void make_8259A_irq(unsigned int irq) { disable_irq_nosync(irq); io_apic_irqs &= ~(1<<irq); @@ -150,7 +155,7 @@ static void mask_and_ack_8259A(unsigned int irq) unsigned int irqmask = 1 << irq; unsigned long flags; - spin_lock_irqsave(&i8259A_lock, flags); + raw_spin_lock_irqsave(&i8259A_lock, flags); /* * Lightweight spurious IRQ detection. We do not want * to overdo spurious IRQ handling - it's usually a sign @@ -183,7 +188,7 @@ handle_real_irq: outb(cached_master_mask, PIC_MASTER_IMR); outb(0x60+irq, PIC_MASTER_CMD); /* 'Specific EOI to master */ } - spin_unlock_irqrestore(&i8259A_lock, flags); + raw_spin_unlock_irqrestore(&i8259A_lock, flags); return; spurious_8259A_irq: @@ -281,37 +286,37 @@ static int __init i8259A_init_sysfs(void) device_initcall(i8259A_init_sysfs); -void mask_8259A(void) +static void mask_8259A(void) { unsigned long flags; - spin_lock_irqsave(&i8259A_lock, flags); + raw_spin_lock_irqsave(&i8259A_lock, flags); outb(0xff, PIC_MASTER_IMR); /* mask all of 8259A-1 */ outb(0xff, PIC_SLAVE_IMR); /* mask all of 8259A-2 */ - spin_unlock_irqrestore(&i8259A_lock, flags); + raw_spin_unlock_irqrestore(&i8259A_lock, flags); } -void unmask_8259A(void) +static void unmask_8259A(void) { unsigned long flags; - spin_lock_irqsave(&i8259A_lock, flags); + raw_spin_lock_irqsave(&i8259A_lock, flags); outb(cached_master_mask, PIC_MASTER_IMR); /* restore master IRQ mask */ outb(cached_slave_mask, PIC_SLAVE_IMR); /* restore slave IRQ mask */ - spin_unlock_irqrestore(&i8259A_lock, flags); + raw_spin_unlock_irqrestore(&i8259A_lock, flags); } -void init_8259A(int auto_eoi) +static void init_8259A(int auto_eoi) { unsigned long flags; i8259A_auto_eoi = auto_eoi; - spin_lock_irqsave(&i8259A_lock, flags); + raw_spin_lock_irqsave(&i8259A_lock, flags); outb(0xff, PIC_MASTER_IMR); /* mask all of 8259A-1 */ outb(0xff, PIC_SLAVE_IMR); /* mask all of 8259A-2 */ @@ -356,5 +361,49 @@ void init_8259A(int auto_eoi) outb(cached_master_mask, PIC_MASTER_IMR); /* restore master IRQ mask */ outb(cached_slave_mask, PIC_SLAVE_IMR); /* restore slave IRQ mask */ - spin_unlock_irqrestore(&i8259A_lock, flags); + raw_spin_unlock_irqrestore(&i8259A_lock, flags); } + +/* + * make i8259 a driver so that we can select pic functions at run time. the goal + * is to make x86 binary compatible among pc compatible and non-pc compatible + * platforms, such as x86 MID. + */ + +static void legacy_pic_noop(void) { }; +static void legacy_pic_uint_noop(unsigned int unused) { }; +static void legacy_pic_int_noop(int unused) { }; + +static struct irq_chip dummy_pic_chip = { + .name = "dummy pic", + .mask = legacy_pic_uint_noop, + .unmask = legacy_pic_uint_noop, + .disable = legacy_pic_uint_noop, + .mask_ack = legacy_pic_uint_noop, +}; +static int legacy_pic_irq_pending_noop(unsigned int irq) +{ + return 0; +} + +struct legacy_pic null_legacy_pic = { + .nr_legacy_irqs = 0, + .chip = &dummy_pic_chip, + .mask_all = legacy_pic_noop, + .restore_mask = legacy_pic_noop, + .init = legacy_pic_int_noop, + .irq_pending = legacy_pic_irq_pending_noop, + .make_irq = legacy_pic_uint_noop, +}; + +struct legacy_pic default_legacy_pic = { + .nr_legacy_irqs = NR_IRQS_LEGACY, + .chip = &i8259A_chip, + .mask_all = mask_8259A, + .restore_mask = unmask_8259A, + .init = init_8259A, + .irq_pending = i8259A_irq_pending, + .make_irq = make_8259A_irq, +}; + +struct legacy_pic *legacy_pic = &default_legacy_pic; diff --git a/arch/x86/kernel/irqinit.c b/arch/x86/kernel/irqinit.c index d5932226614..990ae7cfc57 100644 --- a/arch/x86/kernel/irqinit.c +++ b/arch/x86/kernel/irqinit.c @@ -5,7 +5,6 @@ #include <linux/ioport.h> #include <linux/interrupt.h> #include <linux/timex.h> -#include <linux/slab.h> #include <linux/random.h> #include <linux/kprobes.h> #include <linux/init.h> @@ -61,7 +60,7 @@ static irqreturn_t math_error_irq(int cpl, void *dev_id) outb(0, 0xF0); if (ignore_fpu_irq || !boot_cpu_data.hard_math) return IRQ_NONE; - math_error((void __user *)get_irq_regs()->ip); + math_error(get_irq_regs(), 0, 16); return IRQ_HANDLED; } @@ -84,24 +83,7 @@ static struct irqaction irq2 = { }; DEFINE_PER_CPU(vector_irq_t, vector_irq) = { - [0 ... IRQ0_VECTOR - 1] = -1, - [IRQ0_VECTOR] = 0, - [IRQ1_VECTOR] = 1, - [IRQ2_VECTOR] = 2, - [IRQ3_VECTOR] = 3, - [IRQ4_VECTOR] = 4, - [IRQ5_VECTOR] = 5, - [IRQ6_VECTOR] = 6, - [IRQ7_VECTOR] = 7, - [IRQ8_VECTOR] = 8, - [IRQ9_VECTOR] = 9, - [IRQ10_VECTOR] = 10, - [IRQ11_VECTOR] = 11, - [IRQ12_VECTOR] = 12, - [IRQ13_VECTOR] = 13, - [IRQ14_VECTOR] = 14, - [IRQ15_VECTOR] = 15, - [IRQ15_VECTOR + 1 ... NR_VECTORS - 1] = -1 + [0 ... NR_VECTORS - 1] = -1, }; int vector_used_by_percpu_irq(unsigned int vector) @@ -123,12 +105,12 @@ void __init init_ISA_irqs(void) #if defined(CONFIG_X86_64) || defined(CONFIG_X86_LOCAL_APIC) init_bsp_APIC(); #endif - init_8259A(0); + legacy_pic->init(0); /* * 16 old-style INTA-cycle interrupts: */ - for (i = 0; i < NR_IRQS_LEGACY; i++) { + for (i = 0; i < legacy_pic->nr_legacy_irqs; i++) { struct irq_desc *desc = irq_to_desc(i); desc->status = IRQ_DISABLED; @@ -142,9 +124,44 @@ void __init init_ISA_irqs(void) void __init init_IRQ(void) { + int i; + + /* + * On cpu 0, Assign IRQ0_VECTOR..IRQ15_VECTOR's to IRQ 0..15. + * If these IRQ's are handled by legacy interrupt-controllers like PIC, + * then this configuration will likely be static after the boot. If + * these IRQ's are handled by more mordern controllers like IO-APIC, + * then this vector space can be freed and re-used dynamically as the + * irq's migrate etc. + */ + for (i = 0; i < legacy_pic->nr_legacy_irqs; i++) + per_cpu(vector_irq, 0)[IRQ0_VECTOR + i] = i; + x86_init.irqs.intr_init(); } +/* + * Setup the vector to irq mappings. + */ +void setup_vector_irq(int cpu) +{ +#ifndef CONFIG_X86_IO_APIC + int irq; + + /* + * On most of the platforms, legacy PIC delivers the interrupts on the + * boot cpu. But there are certain platforms where PIC interrupts are + * delivered to multiple cpu's. If the legacy IRQ is handled by the + * legacy PIC, for the new cpu that is coming online, setup the static + * legacy vector to irq mapping: + */ + for (irq = 0; irq < legacy_pic->nr_legacy_irqs; irq++) + per_cpu(vector_irq, cpu)[IRQ0_VECTOR + irq] = irq; +#endif + + __setup_vector_irq(cpu); +} + static void __init smp_intr_init(void) { #ifdef CONFIG_SMP diff --git a/arch/x86/kernel/k8.c b/arch/x86/kernel/k8.c index cbc4332a77b..0f7bc20cfcd 100644 --- a/arch/x86/kernel/k8.c +++ b/arch/x86/kernel/k8.c @@ -2,8 +2,8 @@ * Shared support code for AMD K8 northbridges and derivates. * Copyright 2006 Andi Kleen, SUSE Labs. Subject to GPLv2. */ -#include <linux/gfp.h> #include <linux/types.h> +#include <linux/slab.h> #include <linux/init.h> #include <linux/errno.h> #include <linux/module.h> @@ -121,3 +121,17 @@ void k8_flush_garts(void) } EXPORT_SYMBOL_GPL(k8_flush_garts); +static __init int init_k8_nbs(void) +{ + int err = 0; + + err = cache_k8_northbridges(); + + if (err < 0) + printk(KERN_NOTICE "K8 NB: Cannot enumerate AMD northbridges.\n"); + + return err; +} + +/* This has to go after the PCI subsystem */ +fs_initcall(init_k8_nbs); diff --git a/arch/x86/kernel/kdebugfs.c b/arch/x86/kernel/kdebugfs.c index e444357375c..8afd9f321f1 100644 --- a/arch/x86/kernel/kdebugfs.c +++ b/arch/x86/kernel/kdebugfs.c @@ -9,6 +9,7 @@ #include <linux/debugfs.h> #include <linux/uaccess.h> #include <linux/module.h> +#include <linux/slab.h> #include <linux/init.h> #include <linux/stat.h> #include <linux/io.h> diff --git a/arch/x86/kernel/kgdb.c b/arch/x86/kernel/kgdb.c index dd74fe7273b..4f4af75b948 100644 --- a/arch/x86/kernel/kgdb.c +++ b/arch/x86/kernel/kgdb.c @@ -42,24 +42,13 @@ #include <linux/init.h> #include <linux/smp.h> #include <linux/nmi.h> +#include <linux/hw_breakpoint.h> #include <asm/debugreg.h> #include <asm/apicdef.h> #include <asm/system.h> - #include <asm/apic.h> -/* - * Put the error code here just in case the user cares: - */ -static int gdb_x86errcode; - -/* - * Likewise, the vector number here (since GDB only gets the signal - * number through the usual means, and that's not very specific): - */ -static int gdb_x86vector = -1; - /** * pt_regs_to_gdb_regs - Convert ptrace regs to GDB regs * @gdb_regs: A pointer to hold the registers in the order GDB wants. @@ -204,40 +193,98 @@ void gdb_regs_to_pt_regs(unsigned long *gdb_regs, struct pt_regs *regs) static struct hw_breakpoint { unsigned enabled; - unsigned type; - unsigned len; unsigned long addr; + int len; + int type; + struct perf_event **pev; } breakinfo[4]; +static unsigned long early_dr7; + static void kgdb_correct_hw_break(void) { - unsigned long dr7; - int correctit = 0; - int breakbit; int breakno; - get_debugreg(dr7, 7); for (breakno = 0; breakno < 4; breakno++) { - breakbit = 2 << (breakno << 1); - if (!(dr7 & breakbit) && breakinfo[breakno].enabled) { - correctit = 1; - dr7 |= breakbit; - dr7 &= ~(0xf0000 << (breakno << 2)); - dr7 |= ((breakinfo[breakno].len << 2) | - breakinfo[breakno].type) << - ((breakno << 2) + 16); + struct perf_event *bp; + struct arch_hw_breakpoint *info; + int val; + int cpu = raw_smp_processor_id(); + if (!breakinfo[breakno].enabled) + continue; + if (dbg_is_early) { set_debugreg(breakinfo[breakno].addr, breakno); - - } else { - if ((dr7 & breakbit) && !breakinfo[breakno].enabled) { - correctit = 1; - dr7 &= ~breakbit; - dr7 &= ~(0xf0000 << (breakno << 2)); - } + early_dr7 |= encode_dr7(breakno, + breakinfo[breakno].len, + breakinfo[breakno].type); + set_debugreg(early_dr7, 7); + continue; } + bp = *per_cpu_ptr(breakinfo[breakno].pev, cpu); + info = counter_arch_bp(bp); + if (bp->attr.disabled != 1) + continue; + bp->attr.bp_addr = breakinfo[breakno].addr; + bp->attr.bp_len = breakinfo[breakno].len; + bp->attr.bp_type = breakinfo[breakno].type; + info->address = breakinfo[breakno].addr; + info->len = breakinfo[breakno].len; + info->type = breakinfo[breakno].type; + val = arch_install_hw_breakpoint(bp); + if (!val) + bp->attr.disabled = 0; } - if (correctit) - set_debugreg(dr7, 7); + if (!dbg_is_early) + hw_breakpoint_restore(); +} + +static int hw_break_reserve_slot(int breakno) +{ + int cpu; + int cnt = 0; + struct perf_event **pevent; + + if (dbg_is_early) + return 0; + + for_each_online_cpu(cpu) { + cnt++; + pevent = per_cpu_ptr(breakinfo[breakno].pev, cpu); + if (dbg_reserve_bp_slot(*pevent)) + goto fail; + } + + return 0; + +fail: + for_each_online_cpu(cpu) { + cnt--; + if (!cnt) + break; + pevent = per_cpu_ptr(breakinfo[breakno].pev, cpu); + dbg_release_bp_slot(*pevent); + } + return -1; +} + +static int hw_break_release_slot(int breakno) +{ + struct perf_event **pevent; + int cpu; + + if (dbg_is_early) + return 0; + + for_each_online_cpu(cpu) { + pevent = per_cpu_ptr(breakinfo[breakno].pev, cpu); + if (dbg_release_bp_slot(*pevent)) + /* + * The debugger is responisble for handing the retry on + * remove failure. + */ + return -1; + } + return 0; } static int @@ -251,6 +298,10 @@ kgdb_remove_hw_break(unsigned long addr, int len, enum kgdb_bptype bptype) if (i == 4) return -1; + if (hw_break_release_slot(i)) { + printk(KERN_ERR "Cannot remove hw breakpoint at %lx\n", addr); + return -1; + } breakinfo[i].enabled = 0; return 0; @@ -259,15 +310,27 @@ kgdb_remove_hw_break(unsigned long addr, int len, enum kgdb_bptype bptype) static void kgdb_remove_all_hw_break(void) { int i; + int cpu = raw_smp_processor_id(); + struct perf_event *bp; - for (i = 0; i < 4; i++) - memset(&breakinfo[i], 0, sizeof(struct hw_breakpoint)); + for (i = 0; i < 4; i++) { + if (!breakinfo[i].enabled) + continue; + bp = *per_cpu_ptr(breakinfo[i].pev, cpu); + if (bp->attr.disabled == 1) + continue; + if (dbg_is_early) + early_dr7 &= ~encode_dr7(i, breakinfo[i].len, + breakinfo[i].type); + else + arch_uninstall_hw_breakpoint(bp); + bp->attr.disabled = 1; + } } static int kgdb_set_hw_break(unsigned long addr, int len, enum kgdb_bptype bptype) { - unsigned type; int i; for (i = 0; i < 4; i++) @@ -278,27 +341,42 @@ kgdb_set_hw_break(unsigned long addr, int len, enum kgdb_bptype bptype) switch (bptype) { case BP_HARDWARE_BREAKPOINT: - type = 0; - len = 1; + len = 1; + breakinfo[i].type = X86_BREAKPOINT_EXECUTE; break; case BP_WRITE_WATCHPOINT: - type = 1; + breakinfo[i].type = X86_BREAKPOINT_WRITE; break; case BP_ACCESS_WATCHPOINT: - type = 3; + breakinfo[i].type = X86_BREAKPOINT_RW; break; default: return -1; } - - if (len == 1 || len == 2 || len == 4) - breakinfo[i].len = len - 1; - else + switch (len) { + case 1: + breakinfo[i].len = X86_BREAKPOINT_LEN_1; + break; + case 2: + breakinfo[i].len = X86_BREAKPOINT_LEN_2; + break; + case 4: + breakinfo[i].len = X86_BREAKPOINT_LEN_4; + break; +#ifdef CONFIG_X86_64 + case 8: + breakinfo[i].len = X86_BREAKPOINT_LEN_8; + break; +#endif + default: return -1; - - breakinfo[i].enabled = 1; + } breakinfo[i].addr = addr; - breakinfo[i].type = type; + if (hw_break_reserve_slot(i)) { + breakinfo[i].addr = 0; + return -1; + } + breakinfo[i].enabled = 1; return 0; } @@ -313,25 +391,26 @@ kgdb_set_hw_break(unsigned long addr, int len, enum kgdb_bptype bptype) */ void kgdb_disable_hw_debug(struct pt_regs *regs) { + int i; + int cpu = raw_smp_processor_id(); + struct perf_event *bp; + /* Disable hardware debugging while we are in kgdb: */ set_debugreg(0UL, 7); -} - -/** - * kgdb_post_primary_code - Save error vector/code numbers. - * @regs: Original pt_regs. - * @e_vector: Original error vector. - * @err_code: Original error code. - * - * This is needed on architectures which support SMP and KGDB. - * This function is called after all the slave cpus have been put - * to a know spin state and the primary CPU has control over KGDB. - */ -void kgdb_post_primary_code(struct pt_regs *regs, int e_vector, int err_code) -{ - /* primary processor is completely in the debugger */ - gdb_x86vector = e_vector; - gdb_x86errcode = err_code; + for (i = 0; i < 4; i++) { + if (!breakinfo[i].enabled) + continue; + if (dbg_is_early) { + early_dr7 &= ~encode_dr7(i, breakinfo[i].len, + breakinfo[i].type); + continue; + } + bp = *per_cpu_ptr(breakinfo[i].pev, cpu); + if (bp->attr.disabled == 1) + continue; + arch_uninstall_hw_breakpoint(bp); + bp->attr.disabled = 1; + } } #ifdef CONFIG_SMP @@ -378,7 +457,6 @@ int kgdb_arch_handle_exception(int e_vector, int signo, int err_code, struct pt_regs *linux_regs) { unsigned long addr; - unsigned long dr6; char *ptr; int newPC; @@ -404,20 +482,6 @@ int kgdb_arch_handle_exception(int e_vector, int signo, int err_code, raw_smp_processor_id()); } - get_debugreg(dr6, 6); - if (!(dr6 & 0x4000)) { - int breakno; - - for (breakno = 0; breakno < 4; breakno++) { - if (dr6 & (1 << breakno) && - breakinfo[breakno].type == 0) { - /* Set restore flag: */ - linux_regs->flags |= X86_EFLAGS_RF; - break; - } - } - } - set_debugreg(0UL, 6); kgdb_correct_hw_break(); return 0; @@ -485,8 +549,7 @@ static int __kgdb_notify(struct die_args *args, unsigned long cmd) break; case DIE_DEBUG: - if (atomic_read(&kgdb_cpu_doing_single_step) == - raw_smp_processor_id()) { + if (atomic_read(&kgdb_cpu_doing_single_step) != -1) { if (user_mode(regs)) return single_step_cont(regs, args); break; @@ -501,7 +564,7 @@ static int __kgdb_notify(struct die_args *args, unsigned long cmd) return NOTIFY_DONE; } - if (kgdb_handle_exception(args->trapnr, args->signr, args->err, regs)) + if (kgdb_handle_exception(args->trapnr, args->signr, cmd, regs)) return NOTIFY_DONE; /* Must touch watchdog before return to normal operation */ @@ -509,6 +572,26 @@ static int __kgdb_notify(struct die_args *args, unsigned long cmd) return NOTIFY_STOP; } +#ifdef CONFIG_KGDB_LOW_LEVEL_TRAP +int kgdb_ll_trap(int cmd, const char *str, + struct pt_regs *regs, long err, int trap, int sig) +{ + struct die_args args = { + .regs = regs, + .str = str, + .err = err, + .trapnr = trap, + .signr = sig, + + }; + + if (!kgdb_io_module_registered) + return NOTIFY_DONE; + + return __kgdb_notify(&args, cmd); +} +#endif /* CONFIG_KGDB_LOW_LEVEL_TRAP */ + static int kgdb_notify(struct notifier_block *self, unsigned long cmd, void *ptr) { @@ -542,6 +625,44 @@ int kgdb_arch_init(void) return register_die_notifier(&kgdb_notifier); } +void kgdb_arch_late(void) +{ + int i, cpu; + struct perf_event_attr attr; + struct perf_event **pevent; + + /* + * Pre-allocate the hw breakpoint structions in the non-atomic + * portion of kgdb because this operation requires mutexs to + * complete. + */ + hw_breakpoint_init(&attr); + attr.bp_addr = (unsigned long)kgdb_arch_init; + attr.bp_len = HW_BREAKPOINT_LEN_1; + attr.bp_type = HW_BREAKPOINT_W; + attr.disabled = 1; + for (i = 0; i < 4; i++) { + if (breakinfo[i].pev) + continue; + breakinfo[i].pev = register_wide_hw_breakpoint(&attr, NULL); + if (IS_ERR(breakinfo[i].pev)) { + printk(KERN_ERR "kgdb: Could not allocate hw" + "breakpoints\nDisabling the kernel debugger\n"); + breakinfo[i].pev = NULL; + kgdb_arch_exit(); + return; + } + for_each_online_cpu(cpu) { + pevent = per_cpu_ptr(breakinfo[i].pev, cpu); + pevent[0]->hw.sample_period = 1; + if (pevent[0]->destroy != NULL) { + pevent[0]->destroy = NULL; + release_bp_slot(*pevent); + } + } + } +} + /** * kgdb_arch_exit - Perform any architecture specific uninitalization. * @@ -550,6 +671,13 @@ int kgdb_arch_init(void) */ void kgdb_arch_exit(void) { + int i; + for (i = 0; i < 4; i++) { + if (breakinfo[i].pev) { + unregister_wide_hw_breakpoint(breakinfo[i].pev); + breakinfo[i].pev = NULL; + } + } unregister_die_notifier(&kgdb_notifier); } @@ -582,6 +710,11 @@ unsigned long kgdb_arch_pc(int exception, struct pt_regs *regs) return instruction_pointer(regs); } +void kgdb_arch_set_pc(struct pt_regs *regs, unsigned long ip) +{ + regs->ip = ip; +} + struct kgdb_arch arch_kgdb_ops = { /* Breakpoint instruction: */ .gdb_bpt_instr = { 0xcc }, diff --git a/arch/x86/kernel/kprobes.c b/arch/x86/kernel/kprobes.c index 5b8c7505b3b..345a4b1fe14 100644 --- a/arch/x86/kernel/kprobes.c +++ b/arch/x86/kernel/kprobes.c @@ -49,6 +49,7 @@ #include <linux/module.h> #include <linux/kdebug.h> #include <linux/kallsyms.h> +#include <linux/ftrace.h> #include <asm/cacheflush.h> #include <asm/desc.h> @@ -106,16 +107,22 @@ struct kretprobe_blackpoint kretprobe_blacklist[] = { }; const int kretprobe_blacklist_size = ARRAY_SIZE(kretprobe_blacklist); -/* Insert a jump instruction at address 'from', which jumps to address 'to'.*/ -static void __kprobes set_jmp_op(void *from, void *to) +static void __kprobes __synthesize_relative_insn(void *from, void *to, u8 op) { - struct __arch_jmp_op { - char op; + struct __arch_relative_insn { + u8 op; s32 raddr; - } __attribute__((packed)) * jop; - jop = (struct __arch_jmp_op *)from; - jop->raddr = (s32)((long)(to) - ((long)(from) + 5)); - jop->op = RELATIVEJUMP_INSTRUCTION; + } __attribute__((packed)) *insn; + + insn = (struct __arch_relative_insn *)from; + insn->raddr = (s32)((long)(to) - ((long)(from) + 5)); + insn->op = op; +} + +/* Insert a jump instruction at address 'from', which jumps to address 'to'.*/ +static void __kprobes synthesize_reljump(void *from, void *to) +{ + __synthesize_relative_insn(from, to, RELATIVEJUMP_OPCODE); } /* @@ -202,7 +209,7 @@ static int recover_probed_instruction(kprobe_opcode_t *buf, unsigned long addr) /* * Basically, kp->ainsn.insn has an original instruction. * However, RIP-relative instruction can not do single-stepping - * at different place, fix_riprel() tweaks the displacement of + * at different place, __copy_instruction() tweaks the displacement of * that instruction. In that case, we can't recover the instruction * from the kp->ainsn.insn. * @@ -284,21 +291,37 @@ static int __kprobes is_IF_modifier(kprobe_opcode_t *insn) } /* - * Adjust the displacement if the instruction uses the %rip-relative - * addressing mode. + * Copy an instruction and adjust the displacement if the instruction + * uses the %rip-relative addressing mode. * If it does, Return the address of the 32-bit displacement word. * If not, return null. * Only applicable to 64-bit x86. */ -static void __kprobes fix_riprel(struct kprobe *p) +static int __kprobes __copy_instruction(u8 *dest, u8 *src, int recover) { -#ifdef CONFIG_X86_64 struct insn insn; - kernel_insn_init(&insn, p->ainsn.insn); + int ret; + kprobe_opcode_t buf[MAX_INSN_SIZE]; + kernel_insn_init(&insn, src); + if (recover) { + insn_get_opcode(&insn); + if (insn.opcode.bytes[0] == BREAKPOINT_INSTRUCTION) { + ret = recover_probed_instruction(buf, + (unsigned long)src); + if (ret) + return 0; + kernel_insn_init(&insn, buf); + } + } + insn_get_length(&insn); + memcpy(dest, insn.kaddr, insn.length); + +#ifdef CONFIG_X86_64 if (insn_rip_relative(&insn)) { s64 newdisp; u8 *disp; + kernel_insn_init(&insn, dest); insn_get_displacement(&insn); /* * The copied instruction uses the %rip-relative addressing @@ -312,20 +335,23 @@ static void __kprobes fix_riprel(struct kprobe *p) * extension of the original signed 32-bit displacement would * have given. */ - newdisp = (u8 *) p->addr + (s64) insn.displacement.value - - (u8 *) p->ainsn.insn; + newdisp = (u8 *) src + (s64) insn.displacement.value - + (u8 *) dest; BUG_ON((s64) (s32) newdisp != newdisp); /* Sanity check. */ - disp = (u8 *) p->ainsn.insn + insn_offset_displacement(&insn); + disp = (u8 *) dest + insn_offset_displacement(&insn); *(s32 *) disp = (s32) newdisp; } #endif + return insn.length; } static void __kprobes arch_copy_kprobe(struct kprobe *p) { - memcpy(p->ainsn.insn, p->addr, MAX_INSN_SIZE * sizeof(kprobe_opcode_t)); - - fix_riprel(p); + /* + * Copy an instruction without recovering int3, because it will be + * put by another subsystem. + */ + __copy_instruction(p->ainsn.insn, p->addr, 0); if (can_boost(p->addr)) p->ainsn.boostable = 0; @@ -337,6 +363,9 @@ static void __kprobes arch_copy_kprobe(struct kprobe *p) int __kprobes arch_prepare_kprobe(struct kprobe *p) { + if (alternatives_text_reserved(p->addr, p->addr)) + return -EINVAL; + if (!can_probe((unsigned long)p->addr)) return -EILSEQ; /* insn: must be on special executable page on x86. */ @@ -393,26 +422,22 @@ static void __kprobes set_current_kprobe(struct kprobe *p, struct pt_regs *regs, static void __kprobes clear_btf(void) { - if (test_thread_flag(TIF_DEBUGCTLMSR)) - update_debugctlmsr(0); + if (test_thread_flag(TIF_BLOCKSTEP)) { + unsigned long debugctl = get_debugctlmsr(); + + debugctl &= ~DEBUGCTLMSR_BTF; + update_debugctlmsr(debugctl); + } } static void __kprobes restore_btf(void) { - if (test_thread_flag(TIF_DEBUGCTLMSR)) - update_debugctlmsr(current->thread.debugctlmsr); -} + if (test_thread_flag(TIF_BLOCKSTEP)) { + unsigned long debugctl = get_debugctlmsr(); -static void __kprobes prepare_singlestep(struct kprobe *p, struct pt_regs *regs) -{ - clear_btf(); - regs->flags |= X86_EFLAGS_TF; - regs->flags &= ~X86_EFLAGS_IF; - /* single step inline if the instruction is an int3 */ - if (p->opcode == BREAKPOINT_INSTRUCTION) - regs->ip = (unsigned long)p->addr; - else - regs->ip = (unsigned long)p->ainsn.insn; + debugctl |= DEBUGCTLMSR_BTF; + update_debugctlmsr(debugctl); + } } void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri, @@ -426,20 +451,50 @@ void __kprobes arch_prepare_kretprobe(struct kretprobe_instance *ri, *sara = (unsigned long) &kretprobe_trampoline; } +#ifdef CONFIG_OPTPROBES +static int __kprobes setup_detour_execution(struct kprobe *p, + struct pt_regs *regs, + int reenter); +#else +#define setup_detour_execution(p, regs, reenter) (0) +#endif + static void __kprobes setup_singlestep(struct kprobe *p, struct pt_regs *regs, - struct kprobe_ctlblk *kcb) + struct kprobe_ctlblk *kcb, int reenter) { -#if !defined(CONFIG_PREEMPT) || defined(CONFIG_FREEZER) + if (setup_detour_execution(p, regs, reenter)) + return; + +#if !defined(CONFIG_PREEMPT) if (p->ainsn.boostable == 1 && !p->post_handler) { /* Boost up -- we can execute copied instructions directly */ - reset_current_kprobe(); + if (!reenter) + reset_current_kprobe(); + /* + * Reentering boosted probe doesn't reset current_kprobe, + * nor set current_kprobe, because it doesn't use single + * stepping. + */ regs->ip = (unsigned long)p->ainsn.insn; preempt_enable_no_resched(); return; } #endif - prepare_singlestep(p, regs); - kcb->kprobe_status = KPROBE_HIT_SS; + if (reenter) { + save_previous_kprobe(kcb); + set_current_kprobe(p, regs, kcb); + kcb->kprobe_status = KPROBE_REENTER; + } else + kcb->kprobe_status = KPROBE_HIT_SS; + /* Prepare real single stepping */ + clear_btf(); + regs->flags |= X86_EFLAGS_TF; + regs->flags &= ~X86_EFLAGS_IF; + /* single step inline if the instruction is an int3 */ + if (p->opcode == BREAKPOINT_INSTRUCTION) + regs->ip = (unsigned long)p->addr; + else + regs->ip = (unsigned long)p->ainsn.insn; } /* @@ -453,11 +508,8 @@ static int __kprobes reenter_kprobe(struct kprobe *p, struct pt_regs *regs, switch (kcb->kprobe_status) { case KPROBE_HIT_SSDONE: case KPROBE_HIT_ACTIVE: - save_previous_kprobe(kcb); - set_current_kprobe(p, regs, kcb); kprobes_inc_nmissed_count(p); - prepare_singlestep(p, regs); - kcb->kprobe_status = KPROBE_REENTER; + setup_singlestep(p, regs, kcb, 1); break; case KPROBE_HIT_SS: /* A probe has been hit in the codepath leading up to, or just @@ -490,20 +542,6 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) struct kprobe_ctlblk *kcb; addr = (kprobe_opcode_t *)(regs->ip - sizeof(kprobe_opcode_t)); - if (*addr != BREAKPOINT_INSTRUCTION) { - /* - * The breakpoint instruction was removed right - * after we hit it. Another cpu has removed - * either a probepoint or a debugger breakpoint - * at this address. In either case, no further - * handling of this interrupt is appropriate. - * Back up over the (now missing) int3 and run - * the original instruction. - */ - regs->ip = (unsigned long)addr; - return 1; - } - /* * We don't want to be preempted for the entire * duration of kprobe processing. We conditionally @@ -532,13 +570,26 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) * more here. */ if (!p->pre_handler || !p->pre_handler(p, regs)) - setup_singlestep(p, regs, kcb); + setup_singlestep(p, regs, kcb, 0); return 1; } + } else if (*addr != BREAKPOINT_INSTRUCTION) { + /* + * The breakpoint instruction was removed right + * after we hit it. Another cpu has removed + * either a probepoint or a debugger breakpoint + * at this address. In either case, no further + * handling of this interrupt is appropriate. + * Back up over the (now missing) int3 and run + * the original instruction. + */ + regs->ip = (unsigned long)addr; + preempt_enable_no_resched(); + return 1; } else if (kprobe_running()) { p = __get_cpu_var(current_kprobe); if (p->break_handler && p->break_handler(p, regs)) { - setup_singlestep(p, regs, kcb); + setup_singlestep(p, regs, kcb, 0); return 1; } } /* else: not a kprobe fault; let the kernel handle it */ @@ -547,6 +598,69 @@ static int __kprobes kprobe_handler(struct pt_regs *regs) return 0; } +#ifdef CONFIG_X86_64 +#define SAVE_REGS_STRING \ + /* Skip cs, ip, orig_ax. */ \ + " subq $24, %rsp\n" \ + " pushq %rdi\n" \ + " pushq %rsi\n" \ + " pushq %rdx\n" \ + " pushq %rcx\n" \ + " pushq %rax\n" \ + " pushq %r8\n" \ + " pushq %r9\n" \ + " pushq %r10\n" \ + " pushq %r11\n" \ + " pushq %rbx\n" \ + " pushq %rbp\n" \ + " pushq %r12\n" \ + " pushq %r13\n" \ + " pushq %r14\n" \ + " pushq %r15\n" +#define RESTORE_REGS_STRING \ + " popq %r15\n" \ + " popq %r14\n" \ + " popq %r13\n" \ + " popq %r12\n" \ + " popq %rbp\n" \ + " popq %rbx\n" \ + " popq %r11\n" \ + " popq %r10\n" \ + " popq %r9\n" \ + " popq %r8\n" \ + " popq %rax\n" \ + " popq %rcx\n" \ + " popq %rdx\n" \ + " popq %rsi\n" \ + " popq %rdi\n" \ + /* Skip orig_ax, ip, cs */ \ + " addq $24, %rsp\n" +#else +#define SAVE_REGS_STRING \ + /* Skip cs, ip, orig_ax and gs. */ \ + " subl $16, %esp\n" \ + " pushl %fs\n" \ + " pushl %ds\n" \ + " pushl %es\n" \ + " pushl %eax\n" \ + " pushl %ebp\n" \ + " pushl %edi\n" \ + " pushl %esi\n" \ + " pushl %edx\n" \ + " pushl %ecx\n" \ + " pushl %ebx\n" +#define RESTORE_REGS_STRING \ + " popl %ebx\n" \ + " popl %ecx\n" \ + " popl %edx\n" \ + " popl %esi\n" \ + " popl %edi\n" \ + " popl %ebp\n" \ + " popl %eax\n" \ + /* Skip ds, es, fs, gs, orig_ax, and ip. Note: don't pop cs here*/\ + " addl $24, %esp\n" +#endif + /* * When a retprobed function returns, this code saves registers and * calls trampoline_handler() runs, which calls the kretprobe's handler. @@ -560,65 +674,16 @@ static void __used __kprobes kretprobe_trampoline_holder(void) /* We don't bother saving the ss register */ " pushq %rsp\n" " pushfq\n" - /* - * Skip cs, ip, orig_ax. - * trampoline_handler() will plug in these values - */ - " subq $24, %rsp\n" - " pushq %rdi\n" - " pushq %rsi\n" - " pushq %rdx\n" - " pushq %rcx\n" - " pushq %rax\n" - " pushq %r8\n" - " pushq %r9\n" - " pushq %r10\n" - " pushq %r11\n" - " pushq %rbx\n" - " pushq %rbp\n" - " pushq %r12\n" - " pushq %r13\n" - " pushq %r14\n" - " pushq %r15\n" + SAVE_REGS_STRING " movq %rsp, %rdi\n" " call trampoline_handler\n" /* Replace saved sp with true return address. */ " movq %rax, 152(%rsp)\n" - " popq %r15\n" - " popq %r14\n" - " popq %r13\n" - " popq %r12\n" - " popq %rbp\n" - " popq %rbx\n" - " popq %r11\n" - " popq %r10\n" - " popq %r9\n" - " popq %r8\n" - " popq %rax\n" - " popq %rcx\n" - " popq %rdx\n" - " popq %rsi\n" - " popq %rdi\n" - /* Skip orig_ax, ip, cs */ - " addq $24, %rsp\n" + RESTORE_REGS_STRING " popfq\n" #else " pushf\n" - /* - * Skip cs, ip, orig_ax and gs. - * trampoline_handler() will plug in these values - */ - " subl $16, %esp\n" - " pushl %fs\n" - " pushl %es\n" - " pushl %ds\n" - " pushl %eax\n" - " pushl %ebp\n" - " pushl %edi\n" - " pushl %esi\n" - " pushl %edx\n" - " pushl %ecx\n" - " pushl %ebx\n" + SAVE_REGS_STRING " movl %esp, %eax\n" " call trampoline_handler\n" /* Move flags to cs */ @@ -626,15 +691,7 @@ static void __used __kprobes kretprobe_trampoline_holder(void) " movl %edx, 52(%esp)\n" /* Replace saved flags with true return address. */ " movl %eax, 56(%esp)\n" - " popl %ebx\n" - " popl %ecx\n" - " popl %edx\n" - " popl %esi\n" - " popl %edi\n" - " popl %ebp\n" - " popl %eax\n" - /* Skip ds, es, fs, gs, orig_ax and ip */ - " addl $24, %esp\n" + RESTORE_REGS_STRING " popf\n" #endif " ret\n"); @@ -802,8 +859,8 @@ static void __kprobes resume_execution(struct kprobe *p, * These instructions can be executed directly if it * jumps back to correct address. */ - set_jmp_op((void *)regs->ip, - (void *)orig_ip + (regs->ip - copy_ip)); + synthesize_reljump((void *)regs->ip, + (void *)orig_ip + (regs->ip - copy_ip)); p->ainsn.boostable = 1; } else { p->ainsn.boostable = -1; @@ -1030,6 +1087,358 @@ int __kprobes longjmp_break_handler(struct kprobe *p, struct pt_regs *regs) return 0; } + +#ifdef CONFIG_OPTPROBES + +/* Insert a call instruction at address 'from', which calls address 'to'.*/ +static void __kprobes synthesize_relcall(void *from, void *to) +{ + __synthesize_relative_insn(from, to, RELATIVECALL_OPCODE); +} + +/* Insert a move instruction which sets a pointer to eax/rdi (1st arg). */ +static void __kprobes synthesize_set_arg1(kprobe_opcode_t *addr, + unsigned long val) +{ +#ifdef CONFIG_X86_64 + *addr++ = 0x48; + *addr++ = 0xbf; +#else + *addr++ = 0xb8; +#endif + *(unsigned long *)addr = val; +} + +void __kprobes kprobes_optinsn_template_holder(void) +{ + asm volatile ( + ".global optprobe_template_entry\n" + "optprobe_template_entry: \n" +#ifdef CONFIG_X86_64 + /* We don't bother saving the ss register */ + " pushq %rsp\n" + " pushfq\n" + SAVE_REGS_STRING + " movq %rsp, %rsi\n" + ".global optprobe_template_val\n" + "optprobe_template_val: \n" + ASM_NOP5 + ASM_NOP5 + ".global optprobe_template_call\n" + "optprobe_template_call: \n" + ASM_NOP5 + /* Move flags to rsp */ + " movq 144(%rsp), %rdx\n" + " movq %rdx, 152(%rsp)\n" + RESTORE_REGS_STRING + /* Skip flags entry */ + " addq $8, %rsp\n" + " popfq\n" +#else /* CONFIG_X86_32 */ + " pushf\n" + SAVE_REGS_STRING + " movl %esp, %edx\n" + ".global optprobe_template_val\n" + "optprobe_template_val: \n" + ASM_NOP5 + ".global optprobe_template_call\n" + "optprobe_template_call: \n" + ASM_NOP5 + RESTORE_REGS_STRING + " addl $4, %esp\n" /* skip cs */ + " popf\n" +#endif + ".global optprobe_template_end\n" + "optprobe_template_end: \n"); +} + +#define TMPL_MOVE_IDX \ + ((long)&optprobe_template_val - (long)&optprobe_template_entry) +#define TMPL_CALL_IDX \ + ((long)&optprobe_template_call - (long)&optprobe_template_entry) +#define TMPL_END_IDX \ + ((long)&optprobe_template_end - (long)&optprobe_template_entry) + +#define INT3_SIZE sizeof(kprobe_opcode_t) + +/* Optimized kprobe call back function: called from optinsn */ +static void __kprobes optimized_callback(struct optimized_kprobe *op, + struct pt_regs *regs) +{ + struct kprobe_ctlblk *kcb = get_kprobe_ctlblk(); + + preempt_disable(); + if (kprobe_running()) { + kprobes_inc_nmissed_count(&op->kp); + } else { + /* Save skipped registers */ +#ifdef CONFIG_X86_64 + regs->cs = __KERNEL_CS; +#else + regs->cs = __KERNEL_CS | get_kernel_rpl(); + regs->gs = 0; +#endif + regs->ip = (unsigned long)op->kp.addr + INT3_SIZE; + regs->orig_ax = ~0UL; + + __get_cpu_var(current_kprobe) = &op->kp; + kcb->kprobe_status = KPROBE_HIT_ACTIVE; + opt_pre_handler(&op->kp, regs); + __get_cpu_var(current_kprobe) = NULL; + } + preempt_enable_no_resched(); +} + +static int __kprobes copy_optimized_instructions(u8 *dest, u8 *src) +{ + int len = 0, ret; + + while (len < RELATIVEJUMP_SIZE) { + ret = __copy_instruction(dest + len, src + len, 1); + if (!ret || !can_boost(dest + len)) + return -EINVAL; + len += ret; + } + /* Check whether the address range is reserved */ + if (ftrace_text_reserved(src, src + len - 1) || + alternatives_text_reserved(src, src + len - 1)) + return -EBUSY; + + return len; +} + +/* Check whether insn is indirect jump */ +static int __kprobes insn_is_indirect_jump(struct insn *insn) +{ + return ((insn->opcode.bytes[0] == 0xff && + (X86_MODRM_REG(insn->modrm.value) & 6) == 4) || /* Jump */ + insn->opcode.bytes[0] == 0xea); /* Segment based jump */ +} + +/* Check whether insn jumps into specified address range */ +static int insn_jump_into_range(struct insn *insn, unsigned long start, int len) +{ + unsigned long target = 0; + + switch (insn->opcode.bytes[0]) { + case 0xe0: /* loopne */ + case 0xe1: /* loope */ + case 0xe2: /* loop */ + case 0xe3: /* jcxz */ + case 0xe9: /* near relative jump */ + case 0xeb: /* short relative jump */ + break; + case 0x0f: + if ((insn->opcode.bytes[1] & 0xf0) == 0x80) /* jcc near */ + break; + return 0; + default: + if ((insn->opcode.bytes[0] & 0xf0) == 0x70) /* jcc short */ + break; + return 0; + } + target = (unsigned long)insn->next_byte + insn->immediate.value; + + return (start <= target && target <= start + len); +} + +/* Decode whole function to ensure any instructions don't jump into target */ +static int __kprobes can_optimize(unsigned long paddr) +{ + int ret; + unsigned long addr, size = 0, offset = 0; + struct insn insn; + kprobe_opcode_t buf[MAX_INSN_SIZE]; + /* Dummy buffers for lookup_symbol_attrs */ + static char __dummy_buf[KSYM_NAME_LEN]; + + /* Lookup symbol including addr */ + if (!kallsyms_lookup(paddr, &size, &offset, NULL, __dummy_buf)) + return 0; + + /* Check there is enough space for a relative jump. */ + if (size - offset < RELATIVEJUMP_SIZE) + return 0; + + /* Decode instructions */ + addr = paddr - offset; + while (addr < paddr - offset + size) { /* Decode until function end */ + if (search_exception_tables(addr)) + /* + * Since some fixup code will jumps into this function, + * we can't optimize kprobe in this function. + */ + return 0; + kernel_insn_init(&insn, (void *)addr); + insn_get_opcode(&insn); + if (insn.opcode.bytes[0] == BREAKPOINT_INSTRUCTION) { + ret = recover_probed_instruction(buf, addr); + if (ret) + return 0; + kernel_insn_init(&insn, buf); + } + insn_get_length(&insn); + /* Recover address */ + insn.kaddr = (void *)addr; + insn.next_byte = (void *)(addr + insn.length); + /* Check any instructions don't jump into target */ + if (insn_is_indirect_jump(&insn) || + insn_jump_into_range(&insn, paddr + INT3_SIZE, + RELATIVE_ADDR_SIZE)) + return 0; + addr += insn.length; + } + + return 1; +} + +/* Check optimized_kprobe can actually be optimized. */ +int __kprobes arch_check_optimized_kprobe(struct optimized_kprobe *op) +{ + int i; + struct kprobe *p; + + for (i = 1; i < op->optinsn.size; i++) { + p = get_kprobe(op->kp.addr + i); + if (p && !kprobe_disabled(p)) + return -EEXIST; + } + + return 0; +} + +/* Check the addr is within the optimized instructions. */ +int __kprobes arch_within_optimized_kprobe(struct optimized_kprobe *op, + unsigned long addr) +{ + return ((unsigned long)op->kp.addr <= addr && + (unsigned long)op->kp.addr + op->optinsn.size > addr); +} + +/* Free optimized instruction slot */ +static __kprobes +void __arch_remove_optimized_kprobe(struct optimized_kprobe *op, int dirty) +{ + if (op->optinsn.insn) { + free_optinsn_slot(op->optinsn.insn, dirty); + op->optinsn.insn = NULL; + op->optinsn.size = 0; + } +} + +void __kprobes arch_remove_optimized_kprobe(struct optimized_kprobe *op) +{ + __arch_remove_optimized_kprobe(op, 1); +} + +/* + * Copy replacing target instructions + * Target instructions MUST be relocatable (checked inside) + */ +int __kprobes arch_prepare_optimized_kprobe(struct optimized_kprobe *op) +{ + u8 *buf; + int ret; + long rel; + + if (!can_optimize((unsigned long)op->kp.addr)) + return -EILSEQ; + + op->optinsn.insn = get_optinsn_slot(); + if (!op->optinsn.insn) + return -ENOMEM; + + /* + * Verify if the address gap is in 2GB range, because this uses + * a relative jump. + */ + rel = (long)op->optinsn.insn - (long)op->kp.addr + RELATIVEJUMP_SIZE; + if (abs(rel) > 0x7fffffff) + return -ERANGE; + + buf = (u8 *)op->optinsn.insn; + + /* Copy instructions into the out-of-line buffer */ + ret = copy_optimized_instructions(buf + TMPL_END_IDX, op->kp.addr); + if (ret < 0) { + __arch_remove_optimized_kprobe(op, 0); + return ret; + } + op->optinsn.size = ret; + + /* Copy arch-dep-instance from template */ + memcpy(buf, &optprobe_template_entry, TMPL_END_IDX); + + /* Set probe information */ + synthesize_set_arg1(buf + TMPL_MOVE_IDX, (unsigned long)op); + + /* Set probe function call */ + synthesize_relcall(buf + TMPL_CALL_IDX, optimized_callback); + + /* Set returning jmp instruction at the tail of out-of-line buffer */ + synthesize_reljump(buf + TMPL_END_IDX + op->optinsn.size, + (u8 *)op->kp.addr + op->optinsn.size); + + flush_icache_range((unsigned long) buf, + (unsigned long) buf + TMPL_END_IDX + + op->optinsn.size + RELATIVEJUMP_SIZE); + return 0; +} + +/* Replace a breakpoint (int3) with a relative jump. */ +int __kprobes arch_optimize_kprobe(struct optimized_kprobe *op) +{ + unsigned char jmp_code[RELATIVEJUMP_SIZE]; + s32 rel = (s32)((long)op->optinsn.insn - + ((long)op->kp.addr + RELATIVEJUMP_SIZE)); + + /* Backup instructions which will be replaced by jump address */ + memcpy(op->optinsn.copied_insn, op->kp.addr + INT3_SIZE, + RELATIVE_ADDR_SIZE); + + jmp_code[0] = RELATIVEJUMP_OPCODE; + *(s32 *)(&jmp_code[1]) = rel; + + /* + * text_poke_smp doesn't support NMI/MCE code modifying. + * However, since kprobes itself also doesn't support NMI/MCE + * code probing, it's not a problem. + */ + text_poke_smp(op->kp.addr, jmp_code, RELATIVEJUMP_SIZE); + return 0; +} + +/* Replace a relative jump with a breakpoint (int3). */ +void __kprobes arch_unoptimize_kprobe(struct optimized_kprobe *op) +{ + u8 buf[RELATIVEJUMP_SIZE]; + + /* Set int3 to first byte for kprobes */ + buf[0] = BREAKPOINT_INSTRUCTION; + memcpy(buf + 1, op->optinsn.copied_insn, RELATIVE_ADDR_SIZE); + text_poke_smp(op->kp.addr, buf, RELATIVEJUMP_SIZE); +} + +static int __kprobes setup_detour_execution(struct kprobe *p, + struct pt_regs *regs, + int reenter) +{ + struct optimized_kprobe *op; + + if (p->flags & KPROBE_FLAG_OPTIMIZED) { + /* This kprobe is really able to run optimized path. */ + op = container_of(p, struct optimized_kprobe, kp); + /* Detour through copied instructions */ + regs->ip = (unsigned long)op->optinsn.insn + TMPL_END_IDX; + if (!reenter) + reset_current_kprobe(); + preempt_enable_no_resched(); + return 1; + } + return 0; +} +#endif + int __init arch_init_kprobes(void) { return 0; diff --git a/arch/x86/kernel/kvmclock.c b/arch/x86/kernel/kvmclock.c index feaeb0d3aa4..eb9b76c716c 100644 --- a/arch/x86/kernel/kvmclock.c +++ b/arch/x86/kernel/kvmclock.c @@ -29,6 +29,8 @@ #define KVM_SCALE 22 static int kvmclock = 1; +static int msr_kvm_system_time = MSR_KVM_SYSTEM_TIME; +static int msr_kvm_wall_clock = MSR_KVM_WALL_CLOCK; static int parse_no_kvmclock(char *arg) { @@ -54,7 +56,8 @@ static unsigned long kvm_get_wallclock(void) low = (int)__pa_symbol(&wall_clock); high = ((u64)__pa_symbol(&wall_clock) >> 32); - native_write_msr(MSR_KVM_WALL_CLOCK, low, high); + + native_write_msr(msr_kvm_wall_clock, low, high); vcpu_time = &get_cpu_var(hv_clock); pvclock_read_wallclock(&wall_clock, vcpu_time, &ts); @@ -130,7 +133,8 @@ static int kvm_register_clock(char *txt) high = ((u64)__pa(&per_cpu(hv_clock, cpu)) >> 32); printk(KERN_INFO "kvm-clock: cpu %d, msr %x:%x, %s\n", cpu, high, low, txt); - return native_write_msr_safe(MSR_KVM_SYSTEM_TIME, low, high); + + return native_write_msr_safe(msr_kvm_system_time, low, high); } #ifdef CONFIG_X86_LOCAL_APIC @@ -165,14 +169,14 @@ static void __init kvm_smp_prepare_boot_cpu(void) #ifdef CONFIG_KEXEC static void kvm_crash_shutdown(struct pt_regs *regs) { - native_write_msr_safe(MSR_KVM_SYSTEM_TIME, 0, 0); + native_write_msr(msr_kvm_system_time, 0, 0); native_machine_crash_shutdown(regs); } #endif static void kvm_shutdown(void) { - native_write_msr_safe(MSR_KVM_SYSTEM_TIME, 0, 0); + native_write_msr(msr_kvm_system_time, 0, 0); native_machine_shutdown(); } @@ -181,27 +185,37 @@ void __init kvmclock_init(void) if (!kvm_para_available()) return; - if (kvmclock && kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE)) { - if (kvm_register_clock("boot clock")) - return; - pv_time_ops.sched_clock = kvm_clock_read; - x86_platform.calibrate_tsc = kvm_get_tsc_khz; - x86_platform.get_wallclock = kvm_get_wallclock; - x86_platform.set_wallclock = kvm_set_wallclock; + if (kvmclock && kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE2)) { + msr_kvm_system_time = MSR_KVM_SYSTEM_TIME_NEW; + msr_kvm_wall_clock = MSR_KVM_WALL_CLOCK_NEW; + } else if (!(kvmclock && kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE))) + return; + + printk(KERN_INFO "kvm-clock: Using msrs %x and %x", + msr_kvm_system_time, msr_kvm_wall_clock); + + if (kvm_register_clock("boot clock")) + return; + pv_time_ops.sched_clock = kvm_clock_read; + x86_platform.calibrate_tsc = kvm_get_tsc_khz; + x86_platform.get_wallclock = kvm_get_wallclock; + x86_platform.set_wallclock = kvm_set_wallclock; #ifdef CONFIG_X86_LOCAL_APIC - x86_cpuinit.setup_percpu_clockev = - kvm_setup_secondary_clock; + x86_cpuinit.setup_percpu_clockev = + kvm_setup_secondary_clock; #endif #ifdef CONFIG_SMP - smp_ops.smp_prepare_boot_cpu = kvm_smp_prepare_boot_cpu; + smp_ops.smp_prepare_boot_cpu = kvm_smp_prepare_boot_cpu; #endif - machine_ops.shutdown = kvm_shutdown; + machine_ops.shutdown = kvm_shutdown; #ifdef CONFIG_KEXEC - machine_ops.crash_shutdown = kvm_crash_shutdown; + machine_ops.crash_shutdown = kvm_crash_shutdown; #endif - kvm_get_preset_lpj(); - clocksource_register(&kvm_clock); - pv_info.paravirt_enabled = 1; - pv_info.name = "KVM"; - } + kvm_get_preset_lpj(); + clocksource_register(&kvm_clock); + pv_info.paravirt_enabled = 1; + pv_info.name = "KVM"; + + if (kvm_para_has_feature(KVM_FEATURE_CLOCKSOURCE_STABLE_BIT)) + pvclock_set_flags(PVCLOCK_TSC_STABLE_BIT); } diff --git a/arch/x86/kernel/ldt.c b/arch/x86/kernel/ldt.c index ec6ef60cbd1..ea697263b37 100644 --- a/arch/x86/kernel/ldt.c +++ b/arch/x86/kernel/ldt.c @@ -7,6 +7,7 @@ */ #include <linux/errno.h> +#include <linux/gfp.h> #include <linux/sched.h> #include <linux/string.h> #include <linux/mm.h> diff --git a/arch/x86/kernel/machine_kexec_64.c b/arch/x86/kernel/machine_kexec_64.c index 4a8bb82248a..035c8c52918 100644 --- a/arch/x86/kernel/machine_kexec_64.c +++ b/arch/x86/kernel/machine_kexec_64.c @@ -9,6 +9,7 @@ #include <linux/mm.h> #include <linux/kexec.h> #include <linux/string.h> +#include <linux/gfp.h> #include <linux/reboot.h> #include <linux/numa.h> #include <linux/ftrace.h> diff --git a/arch/x86/kernel/mca_32.c b/arch/x86/kernel/mca_32.c index 845d80ce1ef..63eaf659623 100644 --- a/arch/x86/kernel/mca_32.c +++ b/arch/x86/kernel/mca_32.c @@ -42,6 +42,7 @@ #include <linux/kernel.h> #include <linux/mca.h> #include <linux/kprobes.h> +#include <linux/slab.h> #include <asm/system.h> #include <asm/io.h> #include <linux/proc_fs.h> diff --git a/arch/x86/kernel/microcode_core.c b/arch/x86/kernel/microcode_core.c index cceb5bc3c3c..fa6551d36c1 100644 --- a/arch/x86/kernel/microcode_core.c +++ b/arch/x86/kernel/microcode_core.c @@ -201,9 +201,9 @@ static int do_microcode_update(const void __user *buf, size_t size) return error; } -static int microcode_open(struct inode *unused1, struct file *unused2) +static int microcode_open(struct inode *inode, struct file *file) { - return capable(CAP_SYS_RAWIO) ? 0 : -EPERM; + return capable(CAP_SYS_RAWIO) ? nonseekable_open(inode, file) : -EPERM; } static ssize_t microcode_write(struct file *file, const char __user *buf, @@ -260,6 +260,7 @@ static void microcode_dev_exit(void) } MODULE_ALIAS_MISCDEV(MICROCODE_MINOR); +MODULE_ALIAS("devname:cpu/microcode"); #else #define microcode_dev_init() 0 #define microcode_dev_exit() do { } while (0) diff --git a/arch/x86/kernel/microcode_intel.c b/arch/x86/kernel/microcode_intel.c index ebd193e476c..356170262a9 100644 --- a/arch/x86/kernel/microcode_intel.c +++ b/arch/x86/kernel/microcode_intel.c @@ -328,7 +328,7 @@ static int apply_microcode(int cpu) cpu_num, mc_intel->hdr.rev); return -1; } - pr_info("CPU%d updated to revision 0x%x, date = %04x-%02x-%02x \n", + pr_info("CPU%d updated to revision 0x%x, date = %04x-%02x-%02x\n", cpu_num, val[1], mc_intel->hdr.date & 0xffff, mc_intel->hdr.date >> 24, @@ -343,10 +343,11 @@ static enum ucode_state generic_load_microcode(int cpu, void *data, size_t size, int (*get_ucode_data)(void *, const void *, size_t)) { struct ucode_cpu_info *uci = ucode_cpu_info + cpu; - u8 *ucode_ptr = data, *new_mc = NULL, *mc; + u8 *ucode_ptr = data, *new_mc = NULL, *mc = NULL; int new_rev = uci->cpu_sig.rev; unsigned int leftover = size; enum ucode_state state = UCODE_OK; + unsigned int curr_mc_size = 0; while (leftover) { struct microcode_header_intel mc_header; @@ -361,9 +362,15 @@ static enum ucode_state generic_load_microcode(int cpu, void *data, size_t size, break; } - mc = vmalloc(mc_size); - if (!mc) - break; + /* For performance reasons, reuse mc area when possible */ + if (!mc || mc_size > curr_mc_size) { + if (mc) + vfree(mc); + mc = vmalloc(mc_size); + if (!mc) + break; + curr_mc_size = mc_size; + } if (get_ucode_data(mc, ucode_ptr, mc_size) || microcode_sanity_check(mc) < 0) { @@ -376,13 +383,16 @@ static enum ucode_state generic_load_microcode(int cpu, void *data, size_t size, vfree(new_mc); new_rev = mc_header.rev; new_mc = mc; - } else - vfree(mc); + mc = NULL; /* trigger new vmalloc */ + } ucode_ptr += mc_size; leftover -= mc_size; } + if (mc) + vfree(mc); + if (leftover) { if (new_mc) vfree(new_mc); diff --git a/arch/x86/kernel/mmconf-fam10h_64.c b/arch/x86/kernel/mmconf-fam10h_64.c index 712d15fdc41..71825806cd4 100644 --- a/arch/x86/kernel/mmconf-fam10h_64.c +++ b/arch/x86/kernel/mmconf-fam10h_64.c @@ -7,6 +7,8 @@ #include <linux/string.h> #include <linux/pci.h> #include <linux/dmi.h> +#include <linux/range.h> + #include <asm/pci-direct.h> #include <linux/sort.h> #include <asm/io.h> @@ -30,11 +32,6 @@ static struct pci_hostbridge_probe pci_probes[] __cpuinitdata = { { 0xff, 0, PCI_VENDOR_ID_AMD, 0x1200 }, }; -struct range { - u64 start; - u64 end; -}; - static int __cpuinit cmp_range(const void *x1, const void *x2) { const struct range *r1 = x1; diff --git a/arch/x86/kernel/module.c b/arch/x86/kernel/module.c index 89f386f044e..e0bc186d750 100644 --- a/arch/x86/kernel/module.c +++ b/arch/x86/kernel/module.c @@ -23,6 +23,7 @@ #include <linux/kernel.h> #include <linux/bug.h> #include <linux/mm.h> +#include <linux/gfp.h> #include <asm/system.h> #include <asm/page.h> diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c index 40b54ceb68b..5ae5d2426ed 100644 --- a/arch/x86/kernel/mpparse.c +++ b/arch/x86/kernel/mpparse.c @@ -115,21 +115,6 @@ static void __init MP_bus_info(struct mpc_bus *m) printk(KERN_WARNING "Unknown bustype %s - ignoring\n", str); } -static int bad_ioapic(unsigned long address) -{ - if (nr_ioapics >= MAX_IO_APICS) { - printk(KERN_ERR "ERROR: Max # of I/O APICs (%d) exceeded " - "(found %d)\n", MAX_IO_APICS, nr_ioapics); - panic("Recompile kernel with bigger MAX_IO_APICS!\n"); - } - if (!address) { - printk(KERN_ERR "WARNING: Bogus (zero) I/O APIC address" - " found in table, skipping!\n"); - return 1; - } - return 0; -} - static void __init MP_ioapic_info(struct mpc_ioapic *m) { if (!(m->flags & MPC_APIC_USABLE)) @@ -138,15 +123,7 @@ static void __init MP_ioapic_info(struct mpc_ioapic *m) printk(KERN_INFO "I/O APIC #%d Version %d at 0x%X.\n", m->apicid, m->apicver, m->apicaddr); - if (bad_ioapic(m->apicaddr)) - return; - - mp_ioapics[nr_ioapics].apicaddr = m->apicaddr; - mp_ioapics[nr_ioapics].apicid = m->apicid; - mp_ioapics[nr_ioapics].type = m->type; - mp_ioapics[nr_ioapics].apicver = m->apicver; - mp_ioapics[nr_ioapics].flags = m->flags; - nr_ioapics++; + mp_register_ioapic(m->apicid, m->apicaddr, gsi_end + 1); } static void print_MP_intsrc_info(struct mpc_intsrc *m) @@ -359,13 +336,6 @@ static int __init smp_read_mpc(struct mpc_table *mpc, unsigned early) x86_init.mpparse.mpc_record(1); } -#ifdef CONFIG_X86_BIGSMP - generic_bigsmp_probe(); -#endif - - if (apic->setup_apic_routing) - apic->setup_apic_routing(); - if (!num_processors) printk(KERN_ERR "MPTABLE: no processors registered!\n"); return num_processors; @@ -671,7 +641,7 @@ static void __init smp_reserve_memory(struct mpf_intel *mpf) { unsigned long size = get_mpc_size(mpf->physptr); - reserve_early(mpf->physptr, mpf->physptr+size, "MP-table mpc"); + reserve_early_overlap_ok(mpf->physptr, mpf->physptr+size, "MP-table mpc"); } static int __init smp_scan_config(unsigned long base, unsigned long length) @@ -700,7 +670,7 @@ static int __init smp_scan_config(unsigned long base, unsigned long length) mpf, (u64)virt_to_phys(mpf)); mem = virt_to_phys(mpf); - reserve_early(mem, mem + sizeof(*mpf), "MP-table mpf"); + reserve_early_overlap_ok(mem, mem + sizeof(*mpf), "MP-table mpf"); if (mpf->physptr) smp_reserve_memory(mpf); diff --git a/arch/x86/kernel/mrst.c b/arch/x86/kernel/mrst.c index 3b7078abc87..e796448f0eb 100644 --- a/arch/x86/kernel/mrst.c +++ b/arch/x86/kernel/mrst.c @@ -10,8 +10,211 @@ * of the License. */ #include <linux/init.h> +#include <linux/kernel.h> +#include <linux/sfi.h> +#include <linux/irq.h> +#include <linux/module.h> #include <asm/setup.h> +#include <asm/mpspec_def.h> +#include <asm/hw_irq.h> +#include <asm/apic.h> +#include <asm/io_apic.h> +#include <asm/mrst.h> +#include <asm/io.h> +#include <asm/i8259.h> +#include <asm/apb_timer.h> + +static u32 sfi_mtimer_usage[SFI_MTMR_MAX_NUM]; +static struct sfi_timer_table_entry sfi_mtimer_array[SFI_MTMR_MAX_NUM]; +int sfi_mtimer_num; + +struct sfi_rtc_table_entry sfi_mrtc_array[SFI_MRTC_MAX]; +EXPORT_SYMBOL_GPL(sfi_mrtc_array); +int sfi_mrtc_num; + +static inline void assign_to_mp_irq(struct mpc_intsrc *m, + struct mpc_intsrc *mp_irq) +{ + memcpy(mp_irq, m, sizeof(struct mpc_intsrc)); +} + +static inline int mp_irq_cmp(struct mpc_intsrc *mp_irq, + struct mpc_intsrc *m) +{ + return memcmp(mp_irq, m, sizeof(struct mpc_intsrc)); +} + +static void save_mp_irq(struct mpc_intsrc *m) +{ + int i; + + for (i = 0; i < mp_irq_entries; i++) { + if (!mp_irq_cmp(&mp_irqs[i], m)) + return; + } + + assign_to_mp_irq(m, &mp_irqs[mp_irq_entries]); + if (++mp_irq_entries == MAX_IRQ_SOURCES) + panic("Max # of irq sources exceeded!!\n"); +} + +/* parse all the mtimer info to a static mtimer array */ +static int __init sfi_parse_mtmr(struct sfi_table_header *table) +{ + struct sfi_table_simple *sb; + struct sfi_timer_table_entry *pentry; + struct mpc_intsrc mp_irq; + int totallen; + + sb = (struct sfi_table_simple *)table; + if (!sfi_mtimer_num) { + sfi_mtimer_num = SFI_GET_NUM_ENTRIES(sb, + struct sfi_timer_table_entry); + pentry = (struct sfi_timer_table_entry *) sb->pentry; + totallen = sfi_mtimer_num * sizeof(*pentry); + memcpy(sfi_mtimer_array, pentry, totallen); + } + + printk(KERN_INFO "SFI: MTIMER info (num = %d):\n", sfi_mtimer_num); + pentry = sfi_mtimer_array; + for (totallen = 0; totallen < sfi_mtimer_num; totallen++, pentry++) { + printk(KERN_INFO "timer[%d]: paddr = 0x%08x, freq = %dHz," + " irq = %d\n", totallen, (u32)pentry->phys_addr, + pentry->freq_hz, pentry->irq); + if (!pentry->irq) + continue; + mp_irq.type = MP_IOAPIC; + mp_irq.irqtype = mp_INT; +/* triggering mode edge bit 2-3, active high polarity bit 0-1 */ + mp_irq.irqflag = 5; + mp_irq.srcbus = 0; + mp_irq.srcbusirq = pentry->irq; /* IRQ */ + mp_irq.dstapic = MP_APIC_ALL; + mp_irq.dstirq = pentry->irq; + save_mp_irq(&mp_irq); + } + + return 0; +} + +struct sfi_timer_table_entry *sfi_get_mtmr(int hint) +{ + int i; + if (hint < sfi_mtimer_num) { + if (!sfi_mtimer_usage[hint]) { + pr_debug("hint taken for timer %d irq %d\n",\ + hint, sfi_mtimer_array[hint].irq); + sfi_mtimer_usage[hint] = 1; + return &sfi_mtimer_array[hint]; + } + } + /* take the first timer available */ + for (i = 0; i < sfi_mtimer_num;) { + if (!sfi_mtimer_usage[i]) { + sfi_mtimer_usage[i] = 1; + return &sfi_mtimer_array[i]; + } + i++; + } + return NULL; +} + +void sfi_free_mtmr(struct sfi_timer_table_entry *mtmr) +{ + int i; + for (i = 0; i < sfi_mtimer_num;) { + if (mtmr->irq == sfi_mtimer_array[i].irq) { + sfi_mtimer_usage[i] = 0; + return; + } + i++; + } +} + +/* parse all the mrtc info to a global mrtc array */ +int __init sfi_parse_mrtc(struct sfi_table_header *table) +{ + struct sfi_table_simple *sb; + struct sfi_rtc_table_entry *pentry; + struct mpc_intsrc mp_irq; + + int totallen; + + sb = (struct sfi_table_simple *)table; + if (!sfi_mrtc_num) { + sfi_mrtc_num = SFI_GET_NUM_ENTRIES(sb, + struct sfi_rtc_table_entry); + pentry = (struct sfi_rtc_table_entry *)sb->pentry; + totallen = sfi_mrtc_num * sizeof(*pentry); + memcpy(sfi_mrtc_array, pentry, totallen); + } + + printk(KERN_INFO "SFI: RTC info (num = %d):\n", sfi_mrtc_num); + pentry = sfi_mrtc_array; + for (totallen = 0; totallen < sfi_mrtc_num; totallen++, pentry++) { + printk(KERN_INFO "RTC[%d]: paddr = 0x%08x, irq = %d\n", + totallen, (u32)pentry->phys_addr, pentry->irq); + mp_irq.type = MP_IOAPIC; + mp_irq.irqtype = mp_INT; + mp_irq.irqflag = 0; + mp_irq.srcbus = 0; + mp_irq.srcbusirq = pentry->irq; /* IRQ */ + mp_irq.dstapic = MP_APIC_ALL; + mp_irq.dstirq = pentry->irq; + save_mp_irq(&mp_irq); + } + return 0; +} + +/* + * the secondary clock in Moorestown can be APBT or LAPIC clock, default to + * APBT but cmdline option can also override it. + */ +static void __cpuinit mrst_setup_secondary_clock(void) +{ + /* restore default lapic clock if disabled by cmdline */ + if (disable_apbt_percpu) + return setup_secondary_APIC_clock(); + apbt_setup_secondary_clock(); +} + +static unsigned long __init mrst_calibrate_tsc(void) +{ + unsigned long flags, fast_calibrate; + + local_irq_save(flags); + fast_calibrate = apbt_quick_calibrate(); + local_irq_restore(flags); + + if (fast_calibrate) + return fast_calibrate; + + return 0; +} + +void __init mrst_time_init(void) +{ + sfi_table_parse(SFI_SIG_MTMR, NULL, NULL, sfi_parse_mtmr); + pre_init_apic_IRQ0(); + apbt_time_init(); +} + +void __init mrst_rtc_init(void) +{ + sfi_table_parse(SFI_SIG_MRTC, NULL, NULL, sfi_parse_mrtc); +} + +/* + * if we use per cpu apb timer, the bootclock already setup. if we use lapic + * timer and one apbt timer for broadcast, we need to set up lapic boot clock. + */ +static void __init mrst_setup_boot_clock(void) +{ + pr_info("%s: per cpu apbt flag %d \n", __func__, disable_apbt_percpu); + if (disable_apbt_percpu) + setup_boot_APIC_clock(); +}; /* * Moorestown specific x86_init function overrides and early setup @@ -21,4 +224,22 @@ void __init x86_mrst_early_setup(void) { x86_init.resources.probe_roms = x86_init_noop; x86_init.resources.reserve_resources = x86_init_noop; + + x86_init.timers.timer_init = mrst_time_init; + x86_init.timers.setup_percpu_clockev = mrst_setup_boot_clock; + + x86_init.irqs.pre_vector_init = x86_init_noop; + + x86_cpuinit.setup_percpu_clockev = mrst_setup_secondary_clock; + + x86_platform.calibrate_tsc = mrst_calibrate_tsc; + x86_init.pci.init = pci_mrst_init; + x86_init.pci.fixup_irqs = x86_init_noop; + + legacy_pic = &null_legacy_pic; + + /* Avoid searching for BIOS MP tables */ + x86_init.mpparse.find_smp_config = x86_init_noop; + x86_init.mpparse.get_smp_config = x86_init_uint_noop; + } diff --git a/arch/x86/kernel/msr.c b/arch/x86/kernel/msr.c index 206735ac8cb..7bf2dc4c8f7 100644 --- a/arch/x86/kernel/msr.c +++ b/arch/x86/kernel/msr.c @@ -37,6 +37,7 @@ #include <linux/cpu.h> #include <linux/notifier.h> #include <linux/uaccess.h> +#include <linux/gfp.h> #include <asm/processor.h> #include <asm/msr.h> @@ -229,7 +230,7 @@ static int __cpuinit msr_class_cpu_callback(struct notifier_block *nfb, msr_device_destroy(cpu); break; } - return err ? NOTIFY_BAD : NOTIFY_OK; + return notifier_from_errno(err); } static struct notifier_block __refdata msr_class_cpu_notifier = { diff --git a/arch/x86/kernel/olpc.c b/arch/x86/kernel/olpc.c index 9d1d263f786..8297160c41b 100644 --- a/arch/x86/kernel/olpc.c +++ b/arch/x86/kernel/olpc.c @@ -17,7 +17,9 @@ #include <linux/spinlock.h> #include <linux/io.h> #include <linux/string.h> + #include <asm/geode.h> +#include <asm/setup.h> #include <asm/olpc.h> #ifdef CONFIG_OPEN_FIRMWARE @@ -243,9 +245,11 @@ static int __init olpc_init(void) olpc_ec_cmd(EC_FIRMWARE_REV, NULL, 0, (unsigned char *) &olpc_platform_info.ecver, 1); - /* check to see if the VSA exists */ - if (cs5535_has_vsa2()) - olpc_platform_info.flags |= OLPC_F_VSA; +#ifdef CONFIG_PCI_OLPC + /* If the VSA exists let it emulate PCI, if not emulate in kernel */ + if (!cs5535_has_vsa2()) + x86_init.pci.arch_init = pci_olpc_init; +#endif printk(KERN_INFO "OLPC board revision %s%X (EC=%x)\n", ((olpc_platform_info.boardrev & 0xf) < 8) ? "pre" : "", diff --git a/arch/x86/kernel/paravirt.c b/arch/x86/kernel/paravirt.c index 1b1739d1631..1db183ed7c0 100644 --- a/arch/x86/kernel/paravirt.c +++ b/arch/x86/kernel/paravirt.c @@ -428,10 +428,6 @@ struct pv_mmu_ops pv_mmu_ops = { .ptep_modify_prot_start = __ptep_modify_prot_start, .ptep_modify_prot_commit = __ptep_modify_prot_commit, -#ifdef CONFIG_HIGHPTE - .kmap_atomic_pte = kmap_atomic, -#endif - #if PAGETABLE_LEVELS >= 3 #ifdef CONFIG_X86_PAE .set_pte_atomic = native_set_pte_atomic, diff --git a/arch/x86/kernel/pci-calgary_64.c b/arch/x86/kernel/pci-calgary_64.c index 2bbde607814..fb99f7edb34 100644 --- a/arch/x86/kernel/pci-calgary_64.c +++ b/arch/x86/kernel/pci-calgary_64.c @@ -1309,7 +1309,7 @@ static void calgary_init_bitmap_from_tce_table(struct iommu_table *tbl) /* * get_tce_space_from_tar(): * Function for kdump case. Get the tce tables from first kernel - * by reading the contents of the base adress register of calgary iommu + * by reading the contents of the base address register of calgary iommu */ static void __init get_tce_space_from_tar(void) { diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c index 75e14e21f61..4b7e3d8b01d 100644 --- a/arch/x86/kernel/pci-dma.c +++ b/arch/x86/kernel/pci-dma.c @@ -2,6 +2,7 @@ #include <linux/dma-debug.h> #include <linux/dmar.h> #include <linux/bootmem.h> +#include <linux/gfp.h> #include <linux/pci.h> #include <linux/kmemleak.h> @@ -38,7 +39,7 @@ int iommu_detected __read_mostly = 0; * This variable becomes 1 if iommu=pt is passed on the kernel command line. * If this variable is 1, IOMMU implementations do no DMA translation for * devices and allow every device to access to whole physical memory. This is - * useful if a user want to use an IOMMU only for KVM device assignment to + * useful if a user wants to use an IOMMU only for KVM device assignment to * guests and not for driver dma translation. */ int iommu_pass_through __read_mostly; @@ -65,7 +66,7 @@ int dma_set_mask(struct device *dev, u64 mask) } EXPORT_SYMBOL(dma_set_mask); -#ifdef CONFIG_X86_64 +#if defined(CONFIG_X86_64) && !defined(CONFIG_NUMA) static __initdata void *dma32_bootmem_ptr; static unsigned long dma32_bootmem_size __initdata = (128ULL<<20); @@ -116,14 +117,21 @@ static void __init dma32_free_bootmem(void) dma32_bootmem_ptr = NULL; dma32_bootmem_size = 0; } +#else +void __init dma32_reserve_bootmem(void) +{ +} +static void __init dma32_free_bootmem(void) +{ +} + #endif void __init pci_iommu_alloc(void) { -#ifdef CONFIG_X86_64 /* free the range so iommu could get some range less than 4G */ dma32_free_bootmem(); -#endif + if (pci_swiotlb_detect()) goto out; diff --git a/arch/x86/kernel/pci-gart_64.c b/arch/x86/kernel/pci-gart_64.c index 34de53b46f8..0f7f130caa6 100644 --- a/arch/x86/kernel/pci-gart_64.c +++ b/arch/x86/kernel/pci-gart_64.c @@ -29,6 +29,7 @@ #include <linux/iommu-helper.h> #include <linux/sysdev.h> #include <linux/io.h> +#include <linux/gfp.h> #include <asm/atomic.h> #include <asm/mtrr.h> #include <asm/pgtable.h> @@ -564,6 +565,9 @@ static void enable_gart_translations(void) enable_gart_translation(dev, __pa(agp_gatt_table)); } + + /* Flush the GART-TLB to remove stale entries */ + k8_flush_garts(); } /* @@ -735,7 +739,7 @@ int __init gart_iommu_init(void) unsigned long scratch; long i; - if (cache_k8_northbridges() < 0 || num_k8_northbridges == 0) + if (num_k8_northbridges == 0) return 0; #ifndef CONFIG_AGP_AMD64 diff --git a/arch/x86/kernel/pci-nommu.c b/arch/x86/kernel/pci-nommu.c index 22be12b60a8..3af4af810c0 100644 --- a/arch/x86/kernel/pci-nommu.c +++ b/arch/x86/kernel/pci-nommu.c @@ -4,6 +4,7 @@ #include <linux/scatterlist.h> #include <linux/string.h> #include <linux/init.h> +#include <linux/gfp.h> #include <linux/pci.h> #include <linux/mm.h> diff --git a/arch/x86/kernel/pci-swiotlb.c b/arch/x86/kernel/pci-swiotlb.c index 7d2829dde20..a5bc528d432 100644 --- a/arch/x86/kernel/pci-swiotlb.c +++ b/arch/x86/kernel/pci-swiotlb.c @@ -31,8 +31,6 @@ static struct dma_map_ops swiotlb_dma_ops = { .free_coherent = swiotlb_free_coherent, .sync_single_for_cpu = swiotlb_sync_single_for_cpu, .sync_single_for_device = swiotlb_sync_single_for_device, - .sync_single_range_for_cpu = swiotlb_sync_single_range_for_cpu, - .sync_single_range_for_device = swiotlb_sync_single_range_for_device, .sync_sg_for_cpu = swiotlb_sync_sg_for_cpu, .sync_sg_for_device = swiotlb_sync_sg_for_device, .map_sg = swiotlb_map_sg_attrs, diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index c9b3522b6b4..e7e35219b32 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -20,7 +20,6 @@ #include <asm/idle.h> #include <asm/uaccess.h> #include <asm/i387.h> -#include <asm/ds.h> #include <asm/debugreg.h> unsigned long idle_halt; @@ -32,26 +31,22 @@ struct kmem_cache *task_xstate_cachep; int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src) { + int ret; + *dst = *src; - if (src->thread.xstate) { - dst->thread.xstate = kmem_cache_alloc(task_xstate_cachep, - GFP_KERNEL); - if (!dst->thread.xstate) - return -ENOMEM; - WARN_ON((unsigned long)dst->thread.xstate & 15); - memcpy(dst->thread.xstate, src->thread.xstate, xstate_size); + if (fpu_allocated(&src->thread.fpu)) { + memset(&dst->thread.fpu, 0, sizeof(dst->thread.fpu)); + ret = fpu_alloc(&dst->thread.fpu); + if (ret) + return ret; + fpu_copy(&dst->thread.fpu, &src->thread.fpu); } return 0; } void free_thread_xstate(struct task_struct *tsk) { - if (tsk->thread.xstate) { - kmem_cache_free(task_xstate_cachep, tsk->thread.xstate); - tsk->thread.xstate = NULL; - } - - WARN(tsk->thread.ds_ctx, "leaking DS context\n"); + fpu_free(&tsk->thread.fpu); } void free_thread_info(struct thread_info *ti) @@ -92,6 +87,13 @@ void exit_thread(void) } } +void show_regs(struct pt_regs *regs) +{ + show_registers(regs); + show_trace(NULL, regs, (unsigned long *)kernel_stack_pointer(regs), + regs->bp); +} + void show_regs_common(void) { const char *board, *product; @@ -191,11 +193,16 @@ void __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p, prev = &prev_p->thread; next = &next_p->thread; - if (test_tsk_thread_flag(next_p, TIF_DS_AREA_MSR) || - test_tsk_thread_flag(prev_p, TIF_DS_AREA_MSR)) - ds_switch_to(prev_p, next_p); - else if (next->debugctlmsr != prev->debugctlmsr) - update_debugctlmsr(next->debugctlmsr); + if (test_tsk_thread_flag(prev_p, TIF_BLOCKSTEP) ^ + test_tsk_thread_flag(next_p, TIF_BLOCKSTEP)) { + unsigned long debugctl = get_debugctlmsr(); + + debugctl &= ~DEBUGCTLMSR_BTF; + if (test_tsk_thread_flag(next_p, TIF_BLOCKSTEP)) + debugctl |= DEBUGCTLMSR_BTF; + + update_debugctlmsr(debugctl); + } if (test_tsk_thread_flag(prev_p, TIF_NOTSC) ^ test_tsk_thread_flag(next_p, TIF_NOTSC)) { @@ -519,21 +526,39 @@ static int __cpuinit mwait_usable(const struct cpuinfo_x86 *c) } /* - * Check for AMD CPUs, which have potentially C1E support + * Check for AMD CPUs, where APIC timer interrupt does not wake up CPU from C1e. + * For more information see + * - Erratum #400 for NPT family 0xf and family 0x10 CPUs + * - Erratum #365 for family 0x11 (not affected because C1e not in use) */ static int __cpuinit check_c1e_idle(const struct cpuinfo_x86 *c) { + u64 val; if (c->x86_vendor != X86_VENDOR_AMD) - return 0; - - if (c->x86 < 0x0F) - return 0; + goto no_c1e_idle; /* Family 0x0f models < rev F do not have C1E */ - if (c->x86 == 0x0f && c->x86_model < 0x40) - return 0; + if (c->x86 == 0x0F && c->x86_model >= 0x40) + return 1; - return 1; + if (c->x86 == 0x10) { + /* + * check OSVW bit for CPUs that are not affected + * by erratum #400 + */ + if (cpu_has(c, X86_FEATURE_OSVW)) { + rdmsrl(MSR_AMD64_OSVW_ID_LENGTH, val); + if (val >= 2) { + rdmsrl(MSR_AMD64_OSVW_STATUS, val); + if (!(val & BIT(1))) + goto no_c1e_idle; + } + } + return 1; + } + +no_c1e_idle: + return 0; } static cpumask_var_t c1e_mask; @@ -600,7 +625,7 @@ void __cpuinit select_idle_routine(const struct cpuinfo_x86 *c) { #ifdef CONFIG_SMP if (pm_idle == poll_idle && smp_num_siblings > 1) { - printk(KERN_WARNING "WARNING: polling idle and HT enabled," + printk_once(KERN_WARNING "WARNING: polling idle and HT enabled," " performance may degrade.\n"); } #endif diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c index 37ad1e046aa..8d128783af4 100644 --- a/arch/x86/kernel/process_32.c +++ b/arch/x86/kernel/process_32.c @@ -55,7 +55,6 @@ #include <asm/cpu.h> #include <asm/idle.h> #include <asm/syscalls.h> -#include <asm/ds.h> #include <asm/debugreg.h> asmlinkage void ret_from_fork(void) __asm__("ret_from_fork"); @@ -174,12 +173,6 @@ void __show_regs(struct pt_regs *regs, int all) d6, d7); } -void show_regs(struct pt_regs *regs) -{ - show_registers(regs); - show_trace(NULL, regs, ®s->sp, regs->bp); -} - void release_thread(struct task_struct *dead_task) { BUG_ON(dead_task->mm); @@ -244,13 +237,6 @@ int copy_thread(unsigned long clone_flags, unsigned long sp, kfree(p->thread.io_bitmap_ptr); p->thread.io_bitmap_max = 0; } - - clear_tsk_thread_flag(p, TIF_DS_AREA_MSR); - p->thread.ds_ctx = NULL; - - clear_tsk_thread_flag(p, TIF_DEBUGCTLMSR); - p->thread.debugctlmsr = 0; - return err; } @@ -323,7 +309,7 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p) /* we're going to use this soon, after a few expensive things */ if (preload_fpu) - prefetch(next->xstate); + prefetch(next->fpu.state); /* * Reload esp0. diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c index 41a26a82470..3c2422a99f1 100644 --- a/arch/x86/kernel/process_64.c +++ b/arch/x86/kernel/process_64.c @@ -49,7 +49,6 @@ #include <asm/ia32.h> #include <asm/idle.h> #include <asm/syscalls.h> -#include <asm/ds.h> #include <asm/debugreg.h> asmlinkage extern void ret_from_fork(void); @@ -211,12 +210,6 @@ void __show_regs(struct pt_regs *regs, int all) printk(KERN_DEFAULT "DR3: %016lx DR6: %016lx DR7: %016lx\n", d3, d6, d7); } -void show_regs(struct pt_regs *regs) -{ - show_registers(regs); - show_trace(NULL, regs, (void *)(regs + 1), regs->bp); -} - void release_thread(struct task_struct *dead_task) { if (dead_task->mm) { @@ -282,12 +275,12 @@ int copy_thread(unsigned long clone_flags, unsigned long sp, set_tsk_thread_flag(p, TIF_FORK); - p->thread.fs = me->thread.fs; - p->thread.gs = me->thread.gs; p->thread.io_bitmap_ptr = NULL; savesegment(gs, p->thread.gsindex); + p->thread.gs = p->thread.gsindex ? 0 : me->thread.gs; savesegment(fs, p->thread.fsindex); + p->thread.fs = p->thread.fsindex ? 0 : me->thread.fs; savesegment(es, p->thread.es); savesegment(ds, p->thread.ds); @@ -319,13 +312,6 @@ int copy_thread(unsigned long clone_flags, unsigned long sp, if (err) goto out; } - - clear_tsk_thread_flag(p, TIF_DS_AREA_MSR); - p->thread.ds_ctx = NULL; - - clear_tsk_thread_flag(p, TIF_DEBUGCTLMSR); - p->thread.debugctlmsr = 0; - err = 0; out: if (err && p->thread.io_bitmap_ptr) { @@ -402,7 +388,7 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p) /* we're going to use this soon, after a few expensive things */ if (preload_fpu) - prefetch(next->xstate); + prefetch(next->fpu.state); /* * Reload esp0, LDT and the page table pointer: @@ -527,6 +513,7 @@ void set_personality_ia32(void) /* Make sure to be in 32bit mode */ set_thread_flag(TIF_IA32); + current->personality |= force_personality32; /* Prepare the first "return" to user space */ current_thread_info()->status |= TS_COMPAT; diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c index 017d937639f..70c4872cd8a 100644 --- a/arch/x86/kernel/ptrace.c +++ b/arch/x86/kernel/ptrace.c @@ -2,9 +2,6 @@ /* * Pentium III FXSR, SSE support * Gareth Hughes <gareth@valinux.com>, May 2000 - * - * BTS tracing - * Markus Metzger <markus.t.metzger@intel.com>, Dec 2007 */ #include <linux/kernel.h> @@ -12,6 +9,7 @@ #include <linux/mm.h> #include <linux/smp.h> #include <linux/errno.h> +#include <linux/slab.h> #include <linux/ptrace.h> #include <linux/regset.h> #include <linux/tracehook.h> @@ -21,7 +19,6 @@ #include <linux/audit.h> #include <linux/seccomp.h> #include <linux/signal.h> -#include <linux/workqueue.h> #include <linux/perf_event.h> #include <linux/hw_breakpoint.h> @@ -35,7 +32,6 @@ #include <asm/desc.h> #include <asm/prctl.h> #include <asm/proto.h> -#include <asm/ds.h> #include <asm/hw_breakpoint.h> #include "tls.h" @@ -48,6 +44,7 @@ enum x86_regset { REGSET_FP, REGSET_XFP, REGSET_IOPERM64 = REGSET_XFP, + REGSET_XSTATE, REGSET_TLS, REGSET_IOPERM32, }; @@ -140,30 +137,6 @@ static const int arg_offs_table[] = { #endif }; -/** - * regs_get_argument_nth() - get Nth argument at function call - * @regs: pt_regs which contains registers at function entry. - * @n: argument number. - * - * regs_get_argument_nth() returns @n th argument of a function call. - * Since usually the kernel stack will be changed right after function entry, - * you must use this at function entry. If the @n th entry is NOT in the - * kernel stack or pt_regs, this returns 0. - */ -unsigned long regs_get_argument_nth(struct pt_regs *regs, unsigned int n) -{ - if (n < ARRAY_SIZE(arg_offs_table)) - return *(unsigned long *)((char *)regs + arg_offs_table[n]); - else { - /* - * The typical case: arg n is on the stack. - * (Note: stack[0] = return address, so skip it) - */ - n -= ARRAY_SIZE(arg_offs_table); - return regs_get_kernel_stack_nth(regs, 1 + n); - } -} - /* * does not yet catch signals sent when the child dies. * in exit.c or in signal.c. @@ -604,7 +577,7 @@ ptrace_modify_breakpoint(struct perf_event *bp, int len, int type, struct perf_event_attr attr; /* - * We shoud have at least an inactive breakpoint at this + * We should have at least an inactive breakpoint at this * slot. It means the user is writing dr7 without having * written the address register first */ @@ -702,7 +675,7 @@ static unsigned long ptrace_get_debugreg(struct task_struct *tsk, int n) } else if (n == 6) { val = thread->debugreg6; } else if (n == 7) { - val = ptrace_get_dr7(thread->ptrace_bps); + val = thread->ptrace_dr7; } return val; } @@ -715,7 +688,7 @@ static int ptrace_set_breakpoint_addr(struct task_struct *tsk, int nr, struct perf_event_attr attr; if (!t->ptrace_bps[nr]) { - hw_breakpoint_init(&attr); + ptrace_breakpoint_init(&attr); /* * Put stub len and type to register (reserve) an inactive but * correct bp @@ -778,8 +751,11 @@ int ptrace_set_debugreg(struct task_struct *tsk, int n, unsigned long val) return rc; } /* All that's left is DR7 */ - if (n == 7) + if (n == 7) { rc = ptrace_write_dr7(tsk, val); + if (!rc) + thread->ptrace_dr7 = val; + } ret_path: return rc; @@ -808,342 +784,6 @@ static int ioperm_get(struct task_struct *target, 0, IO_BITMAP_BYTES); } -#ifdef CONFIG_X86_PTRACE_BTS -/* - * A branch trace store context. - * - * Contexts may only be installed by ptrace_bts_config() and only for - * ptraced tasks. - * - * Contexts are destroyed when the tracee is detached from the tracer. - * The actual destruction work requires interrupts enabled, so the - * work is deferred and will be scheduled during __ptrace_unlink(). - * - * Contexts hold an additional task_struct reference on the traced - * task, as well as a reference on the tracer's mm. - * - * Ptrace already holds a task_struct for the duration of ptrace operations, - * but since destruction is deferred, it may be executed after both - * tracer and tracee exited. - */ -struct bts_context { - /* The branch trace handle. */ - struct bts_tracer *tracer; - - /* The buffer used to store the branch trace and its size. */ - void *buffer; - unsigned int size; - - /* The mm that paid for the above buffer. */ - struct mm_struct *mm; - - /* The task this context belongs to. */ - struct task_struct *task; - - /* The signal to send on a bts buffer overflow. */ - unsigned int bts_ovfl_signal; - - /* The work struct to destroy a context. */ - struct work_struct work; -}; - -static int alloc_bts_buffer(struct bts_context *context, unsigned int size) -{ - void *buffer = NULL; - int err = -ENOMEM; - - err = account_locked_memory(current->mm, current->signal->rlim, size); - if (err < 0) - return err; - - buffer = kzalloc(size, GFP_KERNEL); - if (!buffer) - goto out_refund; - - context->buffer = buffer; - context->size = size; - context->mm = get_task_mm(current); - - return 0; - - out_refund: - refund_locked_memory(current->mm, size); - return err; -} - -static inline void free_bts_buffer(struct bts_context *context) -{ - if (!context->buffer) - return; - - kfree(context->buffer); - context->buffer = NULL; - - refund_locked_memory(context->mm, context->size); - context->size = 0; - - mmput(context->mm); - context->mm = NULL; -} - -static void free_bts_context_work(struct work_struct *w) -{ - struct bts_context *context; - - context = container_of(w, struct bts_context, work); - - ds_release_bts(context->tracer); - put_task_struct(context->task); - free_bts_buffer(context); - kfree(context); -} - -static inline void free_bts_context(struct bts_context *context) -{ - INIT_WORK(&context->work, free_bts_context_work); - schedule_work(&context->work); -} - -static inline struct bts_context *alloc_bts_context(struct task_struct *task) -{ - struct bts_context *context = kzalloc(sizeof(*context), GFP_KERNEL); - if (context) { - context->task = task; - task->bts = context; - - get_task_struct(task); - } - - return context; -} - -static int ptrace_bts_read_record(struct task_struct *child, size_t index, - struct bts_struct __user *out) -{ - struct bts_context *context; - const struct bts_trace *trace; - struct bts_struct bts; - const unsigned char *at; - int error; - - context = child->bts; - if (!context) - return -ESRCH; - - trace = ds_read_bts(context->tracer); - if (!trace) - return -ESRCH; - - at = trace->ds.top - ((index + 1) * trace->ds.size); - if ((void *)at < trace->ds.begin) - at += (trace->ds.n * trace->ds.size); - - if (!trace->read) - return -EOPNOTSUPP; - - error = trace->read(context->tracer, at, &bts); - if (error < 0) - return error; - - if (copy_to_user(out, &bts, sizeof(bts))) - return -EFAULT; - - return sizeof(bts); -} - -static int ptrace_bts_drain(struct task_struct *child, - long size, - struct bts_struct __user *out) -{ - struct bts_context *context; - const struct bts_trace *trace; - const unsigned char *at; - int error, drained = 0; - - context = child->bts; - if (!context) - return -ESRCH; - - trace = ds_read_bts(context->tracer); - if (!trace) - return -ESRCH; - - if (!trace->read) - return -EOPNOTSUPP; - - if (size < (trace->ds.top - trace->ds.begin)) - return -EIO; - - for (at = trace->ds.begin; (void *)at < trace->ds.top; - out++, drained++, at += trace->ds.size) { - struct bts_struct bts; - - error = trace->read(context->tracer, at, &bts); - if (error < 0) - return error; - - if (copy_to_user(out, &bts, sizeof(bts))) - return -EFAULT; - } - - memset(trace->ds.begin, 0, trace->ds.n * trace->ds.size); - - error = ds_reset_bts(context->tracer); - if (error < 0) - return error; - - return drained; -} - -static int ptrace_bts_config(struct task_struct *child, - long cfg_size, - const struct ptrace_bts_config __user *ucfg) -{ - struct bts_context *context; - struct ptrace_bts_config cfg; - unsigned int flags = 0; - - if (cfg_size < sizeof(cfg)) - return -EIO; - - if (copy_from_user(&cfg, ucfg, sizeof(cfg))) - return -EFAULT; - - context = child->bts; - if (!context) - context = alloc_bts_context(child); - if (!context) - return -ENOMEM; - - if (cfg.flags & PTRACE_BTS_O_SIGNAL) { - if (!cfg.signal) - return -EINVAL; - - return -EOPNOTSUPP; - context->bts_ovfl_signal = cfg.signal; - } - - ds_release_bts(context->tracer); - context->tracer = NULL; - - if ((cfg.flags & PTRACE_BTS_O_ALLOC) && (cfg.size != context->size)) { - int err; - - free_bts_buffer(context); - if (!cfg.size) - return 0; - - err = alloc_bts_buffer(context, cfg.size); - if (err < 0) - return err; - } - - if (cfg.flags & PTRACE_BTS_O_TRACE) - flags |= BTS_USER; - - if (cfg.flags & PTRACE_BTS_O_SCHED) - flags |= BTS_TIMESTAMPS; - - context->tracer = - ds_request_bts_task(child, context->buffer, context->size, - NULL, (size_t)-1, flags); - if (unlikely(IS_ERR(context->tracer))) { - int error = PTR_ERR(context->tracer); - - free_bts_buffer(context); - context->tracer = NULL; - return error; - } - - return sizeof(cfg); -} - -static int ptrace_bts_status(struct task_struct *child, - long cfg_size, - struct ptrace_bts_config __user *ucfg) -{ - struct bts_context *context; - const struct bts_trace *trace; - struct ptrace_bts_config cfg; - - context = child->bts; - if (!context) - return -ESRCH; - - if (cfg_size < sizeof(cfg)) - return -EIO; - - trace = ds_read_bts(context->tracer); - if (!trace) - return -ESRCH; - - memset(&cfg, 0, sizeof(cfg)); - cfg.size = trace->ds.end - trace->ds.begin; - cfg.signal = context->bts_ovfl_signal; - cfg.bts_size = sizeof(struct bts_struct); - - if (cfg.signal) - cfg.flags |= PTRACE_BTS_O_SIGNAL; - - if (trace->ds.flags & BTS_USER) - cfg.flags |= PTRACE_BTS_O_TRACE; - - if (trace->ds.flags & BTS_TIMESTAMPS) - cfg.flags |= PTRACE_BTS_O_SCHED; - - if (copy_to_user(ucfg, &cfg, sizeof(cfg))) - return -EFAULT; - - return sizeof(cfg); -} - -static int ptrace_bts_clear(struct task_struct *child) -{ - struct bts_context *context; - const struct bts_trace *trace; - - context = child->bts; - if (!context) - return -ESRCH; - - trace = ds_read_bts(context->tracer); - if (!trace) - return -ESRCH; - - memset(trace->ds.begin, 0, trace->ds.n * trace->ds.size); - - return ds_reset_bts(context->tracer); -} - -static int ptrace_bts_size(struct task_struct *child) -{ - struct bts_context *context; - const struct bts_trace *trace; - - context = child->bts; - if (!context) - return -ESRCH; - - trace = ds_read_bts(context->tracer); - if (!trace) - return -ESRCH; - - return (trace->ds.top - trace->ds.begin) / trace->ds.size; -} - -/* - * Called from __ptrace_unlink() after the child has been moved back - * to its original parent. - */ -void ptrace_bts_untrace(struct task_struct *child) -{ - if (unlikely(child->bts)) { - free_bts_context(child->bts); - child->bts = NULL; - } -} -#endif /* CONFIG_X86_PTRACE_BTS */ - /* * Called by kernel/ptrace.c when detaching.. * @@ -1271,39 +911,6 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data) break; #endif - /* - * These bits need more cooking - not enabled yet: - */ -#ifdef CONFIG_X86_PTRACE_BTS - case PTRACE_BTS_CONFIG: - ret = ptrace_bts_config - (child, data, (struct ptrace_bts_config __user *)addr); - break; - - case PTRACE_BTS_STATUS: - ret = ptrace_bts_status - (child, data, (struct ptrace_bts_config __user *)addr); - break; - - case PTRACE_BTS_SIZE: - ret = ptrace_bts_size(child); - break; - - case PTRACE_BTS_GET: - ret = ptrace_bts_read_record - (child, data, (struct bts_struct __user *) addr); - break; - - case PTRACE_BTS_CLEAR: - ret = ptrace_bts_clear(child); - break; - - case PTRACE_BTS_DRAIN: - ret = ptrace_bts_drain - (child, data, (struct bts_struct __user *) addr); - break; -#endif /* CONFIG_X86_PTRACE_BTS */ - default: ret = ptrace_request(child, request, addr, data); break; @@ -1563,14 +1170,6 @@ long compat_arch_ptrace(struct task_struct *child, compat_long_t request, case PTRACE_GET_THREAD_AREA: case PTRACE_SET_THREAD_AREA: -#ifdef CONFIG_X86_PTRACE_BTS - case PTRACE_BTS_CONFIG: - case PTRACE_BTS_STATUS: - case PTRACE_BTS_SIZE: - case PTRACE_BTS_GET: - case PTRACE_BTS_CLEAR: - case PTRACE_BTS_DRAIN: -#endif /* CONFIG_X86_PTRACE_BTS */ return arch_ptrace(child, request, addr, data); default: @@ -1584,7 +1183,7 @@ long compat_arch_ptrace(struct task_struct *child, compat_long_t request, #ifdef CONFIG_X86_64 -static const struct user_regset x86_64_regsets[] = { +static struct user_regset x86_64_regsets[] __read_mostly = { [REGSET_GENERAL] = { .core_note_type = NT_PRSTATUS, .n = sizeof(struct user_regs_struct) / sizeof(long), @@ -1597,6 +1196,12 @@ static const struct user_regset x86_64_regsets[] = { .size = sizeof(long), .align = sizeof(long), .active = xfpregs_active, .get = xfpregs_get, .set = xfpregs_set }, + [REGSET_XSTATE] = { + .core_note_type = NT_X86_XSTATE, + .size = sizeof(u64), .align = sizeof(u64), + .active = xstateregs_active, .get = xstateregs_get, + .set = xstateregs_set + }, [REGSET_IOPERM64] = { .core_note_type = NT_386_IOPERM, .n = IO_BITMAP_LONGS, @@ -1622,7 +1227,7 @@ static const struct user_regset_view user_x86_64_view = { #endif /* CONFIG_X86_64 */ #if defined CONFIG_X86_32 || defined CONFIG_IA32_EMULATION -static const struct user_regset x86_32_regsets[] = { +static struct user_regset x86_32_regsets[] __read_mostly = { [REGSET_GENERAL] = { .core_note_type = NT_PRSTATUS, .n = sizeof(struct user_regs_struct32) / sizeof(u32), @@ -1641,6 +1246,12 @@ static const struct user_regset x86_32_regsets[] = { .size = sizeof(u32), .align = sizeof(u32), .active = xfpregs_active, .get = xfpregs_get, .set = xfpregs_set }, + [REGSET_XSTATE] = { + .core_note_type = NT_X86_XSTATE, + .size = sizeof(u64), .align = sizeof(u64), + .active = xstateregs_active, .get = xstateregs_get, + .set = xstateregs_set + }, [REGSET_TLS] = { .core_note_type = NT_386_TLS, .n = GDT_ENTRY_TLS_ENTRIES, .bias = GDT_ENTRY_TLS_MIN, @@ -1663,6 +1274,23 @@ static const struct user_regset_view user_x86_32_view = { }; #endif +/* + * This represents bytes 464..511 in the memory layout exported through + * the REGSET_XSTATE interface. + */ +u64 xstate_fx_sw_bytes[USER_XSTATE_FX_SW_WORDS]; + +void update_regset_xstate_info(unsigned int size, u64 xstate_mask) +{ +#ifdef CONFIG_X86_64 + x86_64_regsets[REGSET_XSTATE].n = size / sizeof(u64); +#endif +#if defined CONFIG_X86_32 || defined CONFIG_IA32_EMULATION + x86_32_regsets[REGSET_XSTATE].n = size / sizeof(u64); +#endif + xstate_fx_sw_bytes[USER_XSTATE_XCR0_WORD] = xstate_mask; +} + const struct user_regset_view *task_user_regset_view(struct task_struct *task) { #ifdef CONFIG_IA32_EMULATION diff --git a/arch/x86/kernel/pvclock.c b/arch/x86/kernel/pvclock.c index 03801f2f761..239427ca02a 100644 --- a/arch/x86/kernel/pvclock.c +++ b/arch/x86/kernel/pvclock.c @@ -31,8 +31,16 @@ struct pvclock_shadow_time { u32 tsc_to_nsec_mul; int tsc_shift; u32 version; + u8 flags; }; +static u8 valid_flags __read_mostly = 0; + +void pvclock_set_flags(u8 flags) +{ + valid_flags = flags; +} + /* * Scale a 64-bit delta by scaling and multiplying by a 32-bit fraction, * yielding a 64-bit result. @@ -91,6 +99,7 @@ static unsigned pvclock_get_time_values(struct pvclock_shadow_time *dst, dst->system_timestamp = src->system_time; dst->tsc_to_nsec_mul = src->tsc_to_system_mul; dst->tsc_shift = src->tsc_shift; + dst->flags = src->flags; rmb(); /* test version after fetching data */ } while ((src->version & 1) || (dst->version != src->version)); @@ -109,11 +118,14 @@ unsigned long pvclock_tsc_khz(struct pvclock_vcpu_time_info *src) return pv_tsc_khz; } +static atomic64_t last_value = ATOMIC64_INIT(0); + cycle_t pvclock_clocksource_read(struct pvclock_vcpu_time_info *src) { struct pvclock_shadow_time shadow; unsigned version; cycle_t ret, offset; + u64 last; do { version = pvclock_get_time_values(&shadow, src); @@ -123,6 +135,31 @@ cycle_t pvclock_clocksource_read(struct pvclock_vcpu_time_info *src) barrier(); } while (version != src->version); + if ((valid_flags & PVCLOCK_TSC_STABLE_BIT) && + (shadow.flags & PVCLOCK_TSC_STABLE_BIT)) + return ret; + + /* + * Assumption here is that last_value, a global accumulator, always goes + * forward. If we are less than that, we should not be much smaller. + * We assume there is an error marging we're inside, and then the correction + * does not sacrifice accuracy. + * + * For reads: global may have changed between test and return, + * but this means someone else updated poked the clock at a later time. + * We just need to make sure we are not seeing a backwards event. + * + * For updates: last_value = ret is not enough, since two vcpus could be + * updating at the same time, and one of them could be slightly behind, + * making the assumption that last_value always go forward fail to hold. + */ + last = atomic64_read(&last_value); + do { + if (ret < last) + return last; + last = atomic64_cmpxchg(&last_value, last, ret); + } while (unlikely(last != ret)); + return ret; } diff --git a/arch/x86/kernel/quirks.c b/arch/x86/kernel/quirks.c index 12e9feaa2f7..e72d3fc6547 100644 --- a/arch/x86/kernel/quirks.c +++ b/arch/x86/kernel/quirks.c @@ -495,10 +495,18 @@ void force_hpet_resume(void) /* * HPET MSI on some boards (ATI SB700/SB800) has side effect on * floppy DMA. Disable HPET MSI on such platforms. + * See erratum #27 (Misinterpreted MSI Requests May Result in + * Corrupted LPC DMA Data) in AMD Publication #46837, + * "SB700 Family Product Errata", Rev. 1.0, March 2010. + * + * Also force the read back of the CMP register in hpet_next_event() + * to work around the problem that the CMP register write seems to be + * delayed. See hpet_next_event() for details. */ static void force_disable_hpet_msi(struct pci_dev *unused) { hpet_msi_disable = 1; + hpet_readback_cmp = 1; } DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_SBX00_SMBUS, diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c index 1545bc0c984..8e1aac86b50 100644 --- a/arch/x86/kernel/reboot.c +++ b/arch/x86/kernel/reboot.c @@ -203,6 +203,15 @@ static struct dmi_system_id __initdata reboot_dmi_table[] = { DMI_MATCH(DMI_BOARD_NAME, "0T656F"), }, }, + { /* Handle problems with rebooting on Dell OptiPlex 760 with 0G919G*/ + .callback = set_bios_reboot, + .ident = "Dell OptiPlex 760", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), + DMI_MATCH(DMI_PRODUCT_NAME, "OptiPlex 760"), + DMI_MATCH(DMI_BOARD_NAME, "0G919G"), + }, + }, { /* Handle problems with rebooting on Dell 2400's */ .callback = set_bios_reboot, .ident = "Dell PowerEdge 2400", @@ -452,6 +461,14 @@ static struct dmi_system_id __initdata pci_reboot_dmi_table[] = { DMI_MATCH(DMI_PRODUCT_NAME, "Macmini3,1"), }, }, + { /* Handle problems with rebooting on the iMac9,1. */ + .callback = set_pci_reboot, + .ident = "Apple iMac9,1", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Apple Inc."), + DMI_MATCH(DMI_PRODUCT_NAME, "iMac9,1"), + }, + }, { } }; diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index f7b8b9894b2..b4ae4acbd03 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -55,7 +55,6 @@ #include <linux/stddef.h> #include <linux/unistd.h> #include <linux/ptrace.h> -#include <linux/slab.h> #include <linux/user.h> #include <linux/delay.h> @@ -121,7 +120,9 @@ unsigned long max_low_pfn_mapped; unsigned long max_pfn_mapped; +#ifdef CONFIG_DMI RESERVE_BRK(dmi_alloc, 65536); +#endif unsigned int boot_cpu_id __read_mostly; @@ -312,16 +313,17 @@ static void __init reserve_brk(void) #define MAX_MAP_CHUNK (NR_FIX_BTMAPS << PAGE_SHIFT) static void __init relocate_initrd(void) { - + /* Assume only end is not page aligned */ u64 ramdisk_image = boot_params.hdr.ramdisk_image; u64 ramdisk_size = boot_params.hdr.ramdisk_size; + u64 area_size = PAGE_ALIGN(ramdisk_size); u64 end_of_lowmem = max_low_pfn_mapped << PAGE_SHIFT; u64 ramdisk_here; unsigned long slop, clen, mapaddr; char *p, *q; /* We need to move the initrd down into lowmem */ - ramdisk_here = find_e820_area(0, end_of_lowmem, ramdisk_size, + ramdisk_here = find_e820_area(0, end_of_lowmem, area_size, PAGE_SIZE); if (ramdisk_here == -1ULL) @@ -330,7 +332,7 @@ static void __init relocate_initrd(void) /* Note: this includes all the lowmem currently occupied by the initrd, we rely on that fact to keep the data intact. */ - reserve_early(ramdisk_here, ramdisk_here + ramdisk_size, + reserve_early(ramdisk_here, ramdisk_here + area_size, "NEW RAMDISK"); initrd_start = ramdisk_here + PAGE_OFFSET; initrd_end = initrd_start + ramdisk_size; @@ -374,9 +376,10 @@ static void __init relocate_initrd(void) static void __init reserve_initrd(void) { + /* Assume only end is not page aligned */ u64 ramdisk_image = boot_params.hdr.ramdisk_image; u64 ramdisk_size = boot_params.hdr.ramdisk_size; - u64 ramdisk_end = ramdisk_image + ramdisk_size; + u64 ramdisk_end = PAGE_ALIGN(ramdisk_image + ramdisk_size); u64 end_of_lowmem = max_low_pfn_mapped << PAGE_SHIFT; if (!boot_params.hdr.type_of_loader || @@ -604,6 +607,16 @@ static int __init setup_elfcorehdr(char *arg) early_param("elfcorehdr", setup_elfcorehdr); #endif +static __init void reserve_ibft_region(void) +{ + unsigned long addr, size = 0; + + addr = find_ibft_region(&size); + + if (size) + reserve_early_overlap_ok(addr, addr + size, "ibft"); +} + #ifdef CONFIG_X86_RESERVE_LOW_64K static int __init dmi_low_memory_corruption(const struct dmi_system_id *d) { @@ -642,23 +655,59 @@ static struct dmi_system_id __initdata bad_bios_dmi_table[] = { DMI_MATCH(DMI_BIOS_VENDOR, "Phoenix/MSC"), }, }, - { /* - * AMI BIOS with low memory corruption was found on Intel DG45ID board. - * It hase different DMI_BIOS_VENDOR = "Intel Corp.", for now we will + * AMI BIOS with low memory corruption was found on Intel DG45ID and + * DG45FC boards. + * It has a different DMI_BIOS_VENDOR = "Intel Corp.", for now we will * match only DMI_BOARD_NAME and see if there is more bad products * with this vendor. */ + { .callback = dmi_low_memory_corruption, .ident = "AMI BIOS", .matches = { DMI_MATCH(DMI_BOARD_NAME, "DG45ID"), }, }, + { + .callback = dmi_low_memory_corruption, + .ident = "AMI BIOS", + .matches = { + DMI_MATCH(DMI_BOARD_NAME, "DG45FC"), + }, + }, + /* + * The Dell Inspiron Mini 1012 has DMI_BIOS_VENDOR = "Dell Inc.", so + * match on the product name. + */ + { + .callback = dmi_low_memory_corruption, + .ident = "Phoenix BIOS", + .matches = { + DMI_MATCH(DMI_PRODUCT_NAME, "Inspiron 1012"), + }, + }, #endif {} }; +static void __init trim_bios_range(void) +{ + /* + * A special case is the first 4Kb of memory; + * This is a BIOS owned area, not kernel ram, but generally + * not listed as such in the E820 table. + */ + e820_update_range(0, PAGE_SIZE, E820_RAM, E820_RESERVED); + /* + * special case: Some BIOSen report the PC BIOS + * area (640->1Mb) as ram even though it is not. + * take them out. + */ + e820_remove_range(BIOS_BEGIN, BIOS_END - BIOS_BEGIN, E820_RAM, 1); + sanitize_e820_map(e820.map, ARRAY_SIZE(e820.map), &e820.nr_map); +} + /* * Determine if we were loaded by an EFI loader. If so, then we have also been * passed the efi memmap, systab, etc., so we should use these data structures @@ -687,6 +736,7 @@ void __init setup_arch(char **cmdline_p) /* VMI may relocate the fixmap; do this before touching ioremap area */ vmi_init(); + early_trap_init(); early_cpu_init(); early_ioremap_init(); @@ -822,7 +872,7 @@ void __init setup_arch(char **cmdline_p) insert_resource(&iomem_resource, &data_resource); insert_resource(&iomem_resource, &bss_resource); - + trim_bios_range(); #ifdef CONFIG_X86_32 if (ppro_with_ram_bug()) { e820_update_range(0x70000000ULL, 0x40000ULL, E820_RAM, @@ -881,6 +931,8 @@ void __init setup_arch(char **cmdline_p) */ find_smp_config(); + reserve_ibft_region(); + reserve_trampoline_memory(); #ifdef CONFIG_ACPI_SLEEP @@ -942,17 +994,11 @@ void __init setup_arch(char **cmdline_p) #endif initmem_init(0, max_pfn, acpi, k8); - -#ifdef CONFIG_X86_64 - /* - * dma32_reserve_bootmem() allocates bootmem which may conflict - * with the crashkernel command line, so do that after - * reserve_crashkernel() - */ - dma32_reserve_bootmem(); +#ifndef CONFIG_NO_BOOTMEM + early_res_to_bootmem(0, max_low_pfn<<PAGE_SHIFT); #endif - reserve_ibft_region(); + dma32_reserve_bootmem(); #ifdef CONFIG_KVM_CLOCK kvmclock_init(); diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index 8c9f68ec97a..cf592767812 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -137,7 +137,13 @@ static void * __init pcpu_fc_alloc(unsigned int cpu, size_t size, size_t align) static void __init pcpu_fc_free(void *ptr, size_t size) { +#ifdef CONFIG_NO_BOOTMEM + u64 start = __pa(ptr); + u64 end = start + size; + free_early_partial(start, end); +#else free_bootmem(__pa(ptr), size); +#endif } static int __init pcpu_cpu_distance(unsigned int from, unsigned int to) @@ -259,10 +265,10 @@ void __init setup_per_cpu_areas(void) #if defined(CONFIG_X86_64) && defined(CONFIG_NUMA) /* - * make sure boot cpu node_number is right, when boot cpu is on the + * make sure boot cpu numa_node is right, when boot cpu is on the * node that doesn't have mem installed */ - per_cpu(node_number, boot_cpu_id) = cpu_to_node(boot_cpu_id); + set_cpu_numa_node(boot_cpu_id, early_cpu_to_node(boot_cpu_id)); #endif /* Setup node to cpumask map */ diff --git a/arch/x86/kernel/sfi.c b/arch/x86/kernel/sfi.c index 34e09938265..7ded57896c0 100644 --- a/arch/x86/kernel/sfi.c +++ b/arch/x86/kernel/sfi.c @@ -81,7 +81,6 @@ static int __init sfi_parse_cpus(struct sfi_table_header *table) #endif /* CONFIG_X86_LOCAL_APIC */ #ifdef CONFIG_X86_IO_APIC -static u32 gsi_base; static int __init sfi_parse_ioapic(struct sfi_table_header *table) { @@ -94,8 +93,7 @@ static int __init sfi_parse_ioapic(struct sfi_table_header *table) pentry = (struct sfi_apic_table_entry *)sb->pentry; for (i = 0; i < num; i++) { - mp_register_ioapic(i, pentry->phys_addr, gsi_base); - gsi_base += io_apic_get_redir_entries(i); + mp_register_ioapic(i, pentry->phys_addr, gsi_end + 1); pentry++; } diff --git a/arch/x86/kernel/smp.c b/arch/x86/kernel/smp.c index ec1de97600e..d801210945d 100644 --- a/arch/x86/kernel/smp.c +++ b/arch/x86/kernel/smp.c @@ -21,6 +21,7 @@ #include <linux/cache.h> #include <linux/interrupt.h> #include <linux/cpu.h> +#include <linux/gfp.h> #include <asm/mtrr.h> #include <asm/tlbflush.h> diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 678d0b8c26f..37462f1ddba 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -48,6 +48,8 @@ #include <linux/err.h> #include <linux/nmi.h> #include <linux/tboot.h> +#include <linux/stackprotector.h> +#include <linux/gfp.h> #include <asm/acpi.h> #include <asm/desc.h> @@ -67,6 +69,7 @@ #include <linux/mc146818rtc.h> #include <asm/smpboot_hooks.h> +#include <asm/i8259.h> #ifdef CONFIG_X86_32 u8 apicid_2_node[MAX_APICID]; @@ -240,7 +243,10 @@ static void __cpuinit smp_callin(void) end_local_APIC_setup(); map_cpu_to_logical_apicid(); - notify_cpu_starting(cpuid); + /* + * Need to setup vector mappings before we enable interrupts. + */ + setup_vector_irq(smp_processor_id()); /* * Get our bogomips. * @@ -257,6 +263,8 @@ static void __cpuinit smp_callin(void) */ smp_store_cpu_info(cpuid); + notify_cpu_starting(cpuid); + /* * Allow the master to continue. */ @@ -286,9 +294,9 @@ notrace static void __cpuinit start_secondary(void *unused) check_tsc_sync_target(); if (nmi_watchdog == NMI_IO_APIC) { - disable_8259A_irq(0); + legacy_pic->chip->mask(0); enable_NMI_through_LVT0(); - enable_8259A_irq(0); + legacy_pic->chip->unmask(0); } #ifdef CONFIG_X86_32 @@ -315,15 +323,18 @@ notrace static void __cpuinit start_secondary(void *unused) */ ipi_call_lock(); lock_vector_lock(); - __setup_vector_irq(smp_processor_id()); set_cpu_online(smp_processor_id(), true); unlock_vector_lock(); ipi_call_unlock(); per_cpu(cpu_state, smp_processor_id()) = CPU_ONLINE; + x86_platform.nmi_init(); /* enable local interrupts */ local_irq_enable(); + /* to prevent fake stack check failure in clock setup */ + boot_init_stack_canary(); + x86_cpuinit.setup_percpu_clockev(); wmb(); @@ -1083,9 +1094,7 @@ void __init native_smp_prepare_cpus(unsigned int max_cpus) set_cpu_sibling_map(0); enable_IR_x2apic(); -#ifdef CONFIG_X86_64 default_setup_apic_routing(); -#endif if (smp_sanity_check(max_cpus) < 0) { printk(KERN_INFO "SMP disabled\n"); @@ -1206,18 +1215,37 @@ __init void prefill_possible_map(void) if (!num_processors) num_processors = 1; - if (setup_possible_cpus == -1) - possible = num_processors + disabled_cpus; - else + i = setup_max_cpus ?: 1; + if (setup_possible_cpus == -1) { + possible = num_processors; +#ifdef CONFIG_HOTPLUG_CPU + if (setup_max_cpus) + possible += disabled_cpus; +#else + if (possible > i) + possible = i; +#endif + } else possible = setup_possible_cpus; total_cpus = max_t(int, possible, num_processors + disabled_cpus); - if (possible > CONFIG_NR_CPUS) { + /* nr_cpu_ids could be reduced via nr_cpus= */ + if (possible > nr_cpu_ids) { printk(KERN_WARNING "%d Processors exceeds NR_CPUS limit of %d\n", - possible, CONFIG_NR_CPUS); - possible = CONFIG_NR_CPUS; + possible, nr_cpu_ids); + possible = nr_cpu_ids; + } + +#ifdef CONFIG_HOTPLUG_CPU + if (!setup_max_cpus) +#endif + if (possible > i) { + printk(KERN_WARNING + "%d Processors exceeds max_cpus limit of %u\n", + possible, setup_max_cpus); + possible = i; } printk(KERN_INFO "SMP: Allowing %d CPUs, %d hotplug CPUs\n", @@ -1225,6 +1253,8 @@ __init void prefill_possible_map(void) for (i = 0; i < possible; i++) set_cpu_possible(i, true); + for (; i < NR_CPUS; i++) + set_cpu_possible(i, false); nr_cpu_ids = possible; } diff --git a/arch/x86/kernel/step.c b/arch/x86/kernel/step.c index 3149032ff10..58de45ee08b 100644 --- a/arch/x86/kernel/step.c +++ b/arch/x86/kernel/step.c @@ -158,22 +158,6 @@ static int enable_single_step(struct task_struct *child) } /* - * Install this value in MSR_IA32_DEBUGCTLMSR whenever child is running. - */ -static void write_debugctlmsr(struct task_struct *child, unsigned long val) -{ - if (child->thread.debugctlmsr == val) - return; - - child->thread.debugctlmsr = val; - - if (child != current) - return; - - update_debugctlmsr(val); -} - -/* * Enable single or block step. */ static void enable_step(struct task_struct *child, bool block) @@ -186,15 +170,17 @@ static void enable_step(struct task_struct *child, bool block) * that uses user-mode single stepping itself. */ if (enable_single_step(child) && block) { - set_tsk_thread_flag(child, TIF_DEBUGCTLMSR); - write_debugctlmsr(child, - child->thread.debugctlmsr | DEBUGCTLMSR_BTF); - } else { - write_debugctlmsr(child, - child->thread.debugctlmsr & ~DEBUGCTLMSR_BTF); - - if (!child->thread.debugctlmsr) - clear_tsk_thread_flag(child, TIF_DEBUGCTLMSR); + unsigned long debugctl = get_debugctlmsr(); + + debugctl |= DEBUGCTLMSR_BTF; + update_debugctlmsr(debugctl); + set_tsk_thread_flag(child, TIF_BLOCKSTEP); + } else if (test_tsk_thread_flag(child, TIF_BLOCKSTEP)) { + unsigned long debugctl = get_debugctlmsr(); + + debugctl &= ~DEBUGCTLMSR_BTF; + update_debugctlmsr(debugctl); + clear_tsk_thread_flag(child, TIF_BLOCKSTEP); } } @@ -213,11 +199,13 @@ void user_disable_single_step(struct task_struct *child) /* * Make sure block stepping (BTF) is disabled. */ - write_debugctlmsr(child, - child->thread.debugctlmsr & ~DEBUGCTLMSR_BTF); + if (test_tsk_thread_flag(child, TIF_BLOCKSTEP)) { + unsigned long debugctl = get_debugctlmsr(); - if (!child->thread.debugctlmsr) - clear_tsk_thread_flag(child, TIF_DEBUGCTLMSR); + debugctl &= ~DEBUGCTLMSR_BTF; + update_debugctlmsr(debugctl); + clear_tsk_thread_flag(child, TIF_BLOCKSTEP); + } /* Always clear TIF_SINGLESTEP... */ clear_tsk_thread_flag(child, TIF_SINGLESTEP); diff --git a/arch/x86/kernel/sys_i386_32.c b/arch/x86/kernel/sys_i386_32.c index dee1ff7cba5..196552bb412 100644 --- a/arch/x86/kernel/sys_i386_32.c +++ b/arch/x86/kernel/sys_i386_32.c @@ -25,191 +25,6 @@ #include <asm/syscalls.h> /* - * Perform the select(nd, in, out, ex, tv) and mmap() system - * calls. Linux/i386 didn't use to be able to handle more than - * 4 system call parameters, so these system calls used a memory - * block for parameter passing.. - */ - -struct mmap_arg_struct { - unsigned long addr; - unsigned long len; - unsigned long prot; - unsigned long flags; - unsigned long fd; - unsigned long offset; -}; - -asmlinkage int old_mmap(struct mmap_arg_struct __user *arg) -{ - struct mmap_arg_struct a; - int err = -EFAULT; - - if (copy_from_user(&a, arg, sizeof(a))) - goto out; - - err = -EINVAL; - if (a.offset & ~PAGE_MASK) - goto out; - - err = sys_mmap_pgoff(a.addr, a.len, a.prot, a.flags, - a.fd, a.offset >> PAGE_SHIFT); -out: - return err; -} - - -struct sel_arg_struct { - unsigned long n; - fd_set __user *inp, *outp, *exp; - struct timeval __user *tvp; -}; - -asmlinkage int old_select(struct sel_arg_struct __user *arg) -{ - struct sel_arg_struct a; - - if (copy_from_user(&a, arg, sizeof(a))) - return -EFAULT; - /* sys_select() does the appropriate kernel locking */ - return sys_select(a.n, a.inp, a.outp, a.exp, a.tvp); -} - -/* - * sys_ipc() is the de-multiplexer for the SysV IPC calls.. - * - * This is really horribly ugly. - */ -asmlinkage int sys_ipc(uint call, int first, int second, - int third, void __user *ptr, long fifth) -{ - int version, ret; - - version = call >> 16; /* hack for backward compatibility */ - call &= 0xffff; - - switch (call) { - case SEMOP: - return sys_semtimedop(first, (struct sembuf __user *)ptr, second, NULL); - case SEMTIMEDOP: - return sys_semtimedop(first, (struct sembuf __user *)ptr, second, - (const struct timespec __user *)fifth); - - case SEMGET: - return sys_semget(first, second, third); - case SEMCTL: { - union semun fourth; - if (!ptr) - return -EINVAL; - if (get_user(fourth.__pad, (void __user * __user *) ptr)) - return -EFAULT; - return sys_semctl(first, second, third, fourth); - } - - case MSGSND: - return sys_msgsnd(first, (struct msgbuf __user *) ptr, - second, third); - case MSGRCV: - switch (version) { - case 0: { - struct ipc_kludge tmp; - if (!ptr) - return -EINVAL; - - if (copy_from_user(&tmp, - (struct ipc_kludge __user *) ptr, - sizeof(tmp))) - return -EFAULT; - return sys_msgrcv(first, tmp.msgp, second, - tmp.msgtyp, third); - } - default: - return sys_msgrcv(first, - (struct msgbuf __user *) ptr, - second, fifth, third); - } - case MSGGET: - return sys_msgget((key_t) first, second); - case MSGCTL: - return sys_msgctl(first, second, (struct msqid_ds __user *) ptr); - - case SHMAT: - switch (version) { - default: { - ulong raddr; - ret = do_shmat(first, (char __user *) ptr, second, &raddr); - if (ret) - return ret; - return put_user(raddr, (ulong __user *) third); - } - case 1: /* iBCS2 emulator entry point */ - if (!segment_eq(get_fs(), get_ds())) - return -EINVAL; - /* The "(ulong *) third" is valid _only_ because of the kernel segment thing */ - return do_shmat(first, (char __user *) ptr, second, (ulong *) third); - } - case SHMDT: - return sys_shmdt((char __user *)ptr); - case SHMGET: - return sys_shmget(first, second, third); - case SHMCTL: - return sys_shmctl(first, second, - (struct shmid_ds __user *) ptr); - default: - return -ENOSYS; - } -} - -/* - * Old cruft - */ -asmlinkage int sys_uname(struct old_utsname __user *name) -{ - int err; - if (!name) - return -EFAULT; - down_read(&uts_sem); - err = copy_to_user(name, utsname(), sizeof(*name)); - up_read(&uts_sem); - return err? -EFAULT:0; -} - -asmlinkage int sys_olduname(struct oldold_utsname __user *name) -{ - int error; - - if (!name) - return -EFAULT; - if (!access_ok(VERIFY_WRITE, name, sizeof(struct oldold_utsname))) - return -EFAULT; - - down_read(&uts_sem); - - error = __copy_to_user(&name->sysname, &utsname()->sysname, - __OLD_UTS_LEN); - error |= __put_user(0, name->sysname + __OLD_UTS_LEN); - error |= __copy_to_user(&name->nodename, &utsname()->nodename, - __OLD_UTS_LEN); - error |= __put_user(0, name->nodename + __OLD_UTS_LEN); - error |= __copy_to_user(&name->release, &utsname()->release, - __OLD_UTS_LEN); - error |= __put_user(0, name->release + __OLD_UTS_LEN); - error |= __copy_to_user(&name->version, &utsname()->version, - __OLD_UTS_LEN); - error |= __put_user(0, name->version + __OLD_UTS_LEN); - error |= __copy_to_user(&name->machine, &utsname()->machine, - __OLD_UTS_LEN); - error |= __put_user(0, name->machine + __OLD_UTS_LEN); - - up_read(&uts_sem); - - error = error ? -EFAULT : 0; - - return error; -} - - -/* * Do a system call from kernel instead of calling sys_execve so we * end up with proper pt_regs. */ diff --git a/arch/x86/kernel/sys_x86_64.c b/arch/x86/kernel/sys_x86_64.c index 8aa2057efd1..ff14a5044ce 100644 --- a/arch/x86/kernel/sys_x86_64.c +++ b/arch/x86/kernel/sys_x86_64.c @@ -209,15 +209,3 @@ bottomup: return addr; } - - -SYSCALL_DEFINE1(uname, struct new_utsname __user *, name) -{ - int err; - down_read(&uts_sem); - err = copy_to_user(name, utsname(), sizeof(*name)); - up_read(&uts_sem); - if (personality(current->personality) == PER_LINUX32) - err |= copy_to_user(&name->machine, "i686", 5); - return err ? -EFAULT : 0; -} diff --git a/arch/x86/kernel/syscall_table_32.S b/arch/x86/kernel/syscall_table_32.S index 15228b5d3eb..8b372934121 100644 --- a/arch/x86/kernel/syscall_table_32.S +++ b/arch/x86/kernel/syscall_table_32.S @@ -81,7 +81,7 @@ ENTRY(sys_call_table) .long sys_settimeofday .long sys_getgroups16 /* 80 */ .long sys_setgroups16 - .long old_select + .long sys_old_select .long sys_symlink .long sys_lstat .long sys_readlink /* 85 */ @@ -89,7 +89,7 @@ ENTRY(sys_call_table) .long sys_swapon .long sys_reboot .long sys_old_readdir - .long old_mmap /* 90 */ + .long sys_old_mmap /* 90 */ .long sys_munmap .long sys_truncate .long sys_ftruncate diff --git a/arch/x86/kernel/tboot.c b/arch/x86/kernel/tboot.c index 86c9f91b48a..c2f1b26141e 100644 --- a/arch/x86/kernel/tboot.c +++ b/arch/x86/kernel/tboot.c @@ -46,6 +46,7 @@ /* Global pointer to shared data; NULL means no measured launch. */ struct tboot *tboot __read_mostly; +EXPORT_SYMBOL(tboot); /* timeout for APs (in secs) to enter wait-for-SIPI state during shutdown */ #define AP_WAIT_TIMEOUT 1 @@ -175,6 +176,9 @@ static void add_mac_region(phys_addr_t start, unsigned long size) struct tboot_mac_region *mr; phys_addr_t end = start + size; + if (tboot->num_mac_regions >= MAX_TB_MAC_REGIONS) + panic("tboot: Too many MAC regions\n"); + if (start && size) { mr = &tboot->mac_regions[tboot->num_mac_regions++]; mr->start = round_down(start, PAGE_SIZE); @@ -184,18 +188,17 @@ static void add_mac_region(phys_addr_t start, unsigned long size) static int tboot_setup_sleep(void) { + int i; + tboot->num_mac_regions = 0; - /* S3 resume code */ - add_mac_region(acpi_wakeup_address, WAKEUP_SIZE); + for (i = 0; i < e820.nr_map; i++) { + if ((e820.map[i].type != E820_RAM) + && (e820.map[i].type != E820_RESERVED_KERN)) + continue; -#ifdef CONFIG_X86_TRAMPOLINE - /* AP trampoline code */ - add_mac_region(virt_to_phys(trampoline_base), TRAMPOLINE_SIZE); -#endif - - /* kernel code + data + bss */ - add_mac_region(virt_to_phys(_text), _end - _text); + add_mac_region(e820.map[i].addr, e820.map[i].size); + } tboot->acpi_sinfo.kernel_s3_resume_vector = acpi_wakeup_address; diff --git a/arch/x86/kernel/time.c b/arch/x86/kernel/time.c index be2573448ed..fb5cc5e14cf 100644 --- a/arch/x86/kernel/time.c +++ b/arch/x86/kernel/time.c @@ -70,11 +70,11 @@ static irqreturn_t timer_interrupt(int irq, void *dev_id) * manually to deassert NMI lines for the watchdog if run * on an 82489DX-based system. */ - spin_lock(&i8259A_lock); + raw_spin_lock(&i8259A_lock); outb(0x0c, PIC_MASTER_OCW3); /* Ack the IRQ; AEOI will end it automatically. */ inb(PIC_MASTER_POLL); - spin_unlock(&i8259A_lock); + raw_spin_unlock(&i8259A_lock); } global_clock_event->event_handler(global_clock_event); diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c index 364d015efeb..7fea555929e 100644 --- a/arch/x86/kernel/tlb_uv.c +++ b/arch/x86/kernel/tlb_uv.c @@ -1,7 +1,7 @@ /* * SGI UltraViolet TLB flush routines. * - * (c) 2008 Cliff Wickman <cpw@sgi.com>, SGI. + * (c) 2008-2010 Cliff Wickman <cpw@sgi.com>, SGI. * * This code is released under the GNU General Public License version 2 or * later. @@ -9,6 +9,7 @@ #include <linux/seq_file.h> #include <linux/proc_fs.h> #include <linux/kernel.h> +#include <linux/slab.h> #include <asm/mmu_context.h> #include <asm/uv/uv.h> @@ -19,42 +20,67 @@ #include <asm/idle.h> #include <asm/tsc.h> #include <asm/irq_vectors.h> +#include <asm/timer.h> -static struct bau_control **uv_bau_table_bases __read_mostly; -static int uv_bau_retry_limit __read_mostly; +struct msg_desc { + struct bau_payload_queue_entry *msg; + int msg_slot; + int sw_ack_slot; + struct bau_payload_queue_entry *va_queue_first; + struct bau_payload_queue_entry *va_queue_last; +}; -/* base pnode in this partition */ -static int uv_partition_base_pnode __read_mostly; +#define UV_INTD_SOFT_ACK_TIMEOUT_PERIOD 0x000000000bUL + +static int uv_bau_max_concurrent __read_mostly; + +static int nobau; +static int __init setup_nobau(char *arg) +{ + nobau = 1; + return 0; +} +early_param("nobau", setup_nobau); -static unsigned long uv_mmask __read_mostly; +/* base pnode in this partition */ +static int uv_partition_base_pnode __read_mostly; +/* position of pnode (which is nasid>>1): */ +static int uv_nshift __read_mostly; +static unsigned long uv_mmask __read_mostly; static DEFINE_PER_CPU(struct ptc_stats, ptcstats); static DEFINE_PER_CPU(struct bau_control, bau_control); +static DEFINE_PER_CPU(cpumask_var_t, uv_flush_tlb_mask); + +struct reset_args { + int sender; +}; /* - * Determine the first node on a blade. + * Determine the first node on a uvhub. 'Nodes' are used for kernel + * memory allocation. */ -static int __init blade_to_first_node(int blade) +static int __init uvhub_to_first_node(int uvhub) { int node, b; for_each_online_node(node) { b = uv_node_to_blade_id(node); - if (blade == b) + if (uvhub == b) return node; } - return -1; /* shouldn't happen */ + return -1; } /* - * Determine the apicid of the first cpu on a blade. + * Determine the apicid of the first cpu on a uvhub. */ -static int __init blade_to_first_apicid(int blade) +static int __init uvhub_to_first_apicid(int uvhub) { int cpu; for_each_present_cpu(cpu) - if (blade == uv_cpu_to_blade_id(cpu)) + if (uvhub == uv_cpu_to_blade_id(cpu)) return per_cpu(x86_cpu_to_apicid, cpu); return -1; } @@ -67,195 +93,459 @@ static int __init blade_to_first_apicid(int blade) * clear of the Timeout bit (as well) will free the resource. No reply will * be sent (the hardware will only do one reply per message). */ -static void uv_reply_to_message(int resource, - struct bau_payload_queue_entry *msg, - struct bau_msg_status *msp) +static inline void uv_reply_to_message(struct msg_desc *mdp, + struct bau_control *bcp) { unsigned long dw; + struct bau_payload_queue_entry *msg; - dw = (1 << (resource + UV_SW_ACK_NPENDING)) | (1 << resource); + msg = mdp->msg; + if (!msg->canceled) { + dw = (msg->sw_ack_vector << UV_SW_ACK_NPENDING) | + msg->sw_ack_vector; + uv_write_local_mmr( + UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE_ALIAS, dw); + } msg->replied_to = 1; msg->sw_ack_vector = 0; - if (msp) - msp->seen_by.bits = 0; - uv_write_local_mmr(UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE_ALIAS, dw); } /* - * Do all the things a cpu should do for a TLB shootdown message. - * Other cpu's may come here at the same time for this message. + * Process the receipt of a RETRY message */ -static void uv_bau_process_message(struct bau_payload_queue_entry *msg, - int msg_slot, int sw_ack_slot) +static inline void uv_bau_process_retry_msg(struct msg_desc *mdp, + struct bau_control *bcp) { - unsigned long this_cpu_mask; - struct bau_msg_status *msp; - int cpu; + int i; + int cancel_count = 0; + int slot2; + unsigned long msg_res; + unsigned long mmr = 0; + struct bau_payload_queue_entry *msg; + struct bau_payload_queue_entry *msg2; + struct ptc_stats *stat; - msp = __get_cpu_var(bau_control).msg_statuses + msg_slot; - cpu = uv_blade_processor_id(); - msg->number_of_cpus = - uv_blade_nr_online_cpus(uv_node_to_blade_id(numa_node_id())); - this_cpu_mask = 1UL << cpu; - if (msp->seen_by.bits & this_cpu_mask) - return; - atomic_or_long(&msp->seen_by.bits, this_cpu_mask); + msg = mdp->msg; + stat = &per_cpu(ptcstats, bcp->cpu); + stat->d_retries++; + /* + * cancel any message from msg+1 to the retry itself + */ + for (msg2 = msg+1, i = 0; i < DEST_Q_SIZE; msg2++, i++) { + if (msg2 > mdp->va_queue_last) + msg2 = mdp->va_queue_first; + if (msg2 == msg) + break; + + /* same conditions for cancellation as uv_do_reset */ + if ((msg2->replied_to == 0) && (msg2->canceled == 0) && + (msg2->sw_ack_vector) && ((msg2->sw_ack_vector & + msg->sw_ack_vector) == 0) && + (msg2->sending_cpu == msg->sending_cpu) && + (msg2->msg_type != MSG_NOOP)) { + slot2 = msg2 - mdp->va_queue_first; + mmr = uv_read_local_mmr + (UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE); + msg_res = ((msg2->sw_ack_vector << 8) | + msg2->sw_ack_vector); + /* + * This is a message retry; clear the resources held + * by the previous message only if they timed out. + * If it has not timed out we have an unexpected + * situation to report. + */ + if (mmr & (msg_res << 8)) { + /* + * is the resource timed out? + * make everyone ignore the cancelled message. + */ + msg2->canceled = 1; + stat->d_canceled++; + cancel_count++; + uv_write_local_mmr( + UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE_ALIAS, + (msg_res << 8) | msg_res); + } else + printk(KERN_INFO "note bau retry: no effect\n"); + } + } + if (!cancel_count) + stat->d_nocanceled++; +} - if (msg->replied_to == 1) - return; +/* + * Do all the things a cpu should do for a TLB shootdown message. + * Other cpu's may come here at the same time for this message. + */ +static void uv_bau_process_message(struct msg_desc *mdp, + struct bau_control *bcp) +{ + int msg_ack_count; + short socket_ack_count = 0; + struct ptc_stats *stat; + struct bau_payload_queue_entry *msg; + struct bau_control *smaster = bcp->socket_master; + /* + * This must be a normal message, or retry of a normal message + */ + msg = mdp->msg; + stat = &per_cpu(ptcstats, bcp->cpu); if (msg->address == TLB_FLUSH_ALL) { local_flush_tlb(); - __get_cpu_var(ptcstats).alltlb++; + stat->d_alltlb++; } else { __flush_tlb_one(msg->address); - __get_cpu_var(ptcstats).onetlb++; + stat->d_onetlb++; } + stat->d_requestee++; + + /* + * One cpu on each uvhub has the additional job on a RETRY + * of releasing the resource held by the message that is + * being retried. That message is identified by sending + * cpu number. + */ + if (msg->msg_type == MSG_RETRY && bcp == bcp->uvhub_master) + uv_bau_process_retry_msg(mdp, bcp); - __get_cpu_var(ptcstats).requestee++; + /* + * This is a sw_ack message, so we have to reply to it. + * Count each responding cpu on the socket. This avoids + * pinging the count's cache line back and forth between + * the sockets. + */ + socket_ack_count = atomic_add_short_return(1, (struct atomic_short *) + &smaster->socket_acknowledge_count[mdp->msg_slot]); + if (socket_ack_count == bcp->cpus_in_socket) { + /* + * Both sockets dump their completed count total into + * the message's count. + */ + smaster->socket_acknowledge_count[mdp->msg_slot] = 0; + msg_ack_count = atomic_add_short_return(socket_ack_count, + (struct atomic_short *)&msg->acknowledge_count); + + if (msg_ack_count == bcp->cpus_in_uvhub) { + /* + * All cpus in uvhub saw it; reply + */ + uv_reply_to_message(mdp, bcp); + } + } - atomic_inc_short(&msg->acknowledge_count); - if (msg->number_of_cpus == msg->acknowledge_count) - uv_reply_to_message(sw_ack_slot, msg, msp); + return; } /* - * Examine the payload queue on one distribution node to see - * which messages have not been seen, and which cpu(s) have not seen them. + * Determine the first cpu on a uvhub. + */ +static int uvhub_to_first_cpu(int uvhub) +{ + int cpu; + for_each_present_cpu(cpu) + if (uvhub == uv_cpu_to_blade_id(cpu)) + return cpu; + return -1; +} + +/* + * Last resort when we get a large number of destination timeouts is + * to clear resources held by a given cpu. + * Do this with IPI so that all messages in the BAU message queue + * can be identified by their nonzero sw_ack_vector field. * - * Returns the number of cpu's that have not responded. + * This is entered for a single cpu on the uvhub. + * The sender want's this uvhub to free a specific message's + * sw_ack resources. */ -static int uv_examine_destination(struct bau_control *bau_tablesp, int sender) +static void +uv_do_reset(void *ptr) { - struct bau_payload_queue_entry *msg; - struct bau_msg_status *msp; - int count = 0; int i; - int j; + int slot; + int count = 0; + unsigned long mmr; + unsigned long msg_res; + struct bau_control *bcp; + struct reset_args *rap; + struct bau_payload_queue_entry *msg; + struct ptc_stats *stat; - for (msg = bau_tablesp->va_queue_first, i = 0; i < DEST_Q_SIZE; - msg++, i++) { - if ((msg->sending_cpu == sender) && (!msg->replied_to)) { - msp = bau_tablesp->msg_statuses + i; - printk(KERN_DEBUG - "blade %d: address:%#lx %d of %d, not cpu(s): ", - i, msg->address, msg->acknowledge_count, - msg->number_of_cpus); - for (j = 0; j < msg->number_of_cpus; j++) { - if (!((1L << j) & msp->seen_by.bits)) { - count++; - printk("%d ", j); - } + bcp = &per_cpu(bau_control, smp_processor_id()); + rap = (struct reset_args *)ptr; + stat = &per_cpu(ptcstats, bcp->cpu); + stat->d_resets++; + + /* + * We're looking for the given sender, and + * will free its sw_ack resource. + * If all cpu's finally responded after the timeout, its + * message 'replied_to' was set. + */ + for (msg = bcp->va_queue_first, i = 0; i < DEST_Q_SIZE; msg++, i++) { + /* uv_do_reset: same conditions for cancellation as + uv_bau_process_retry_msg() */ + if ((msg->replied_to == 0) && + (msg->canceled == 0) && + (msg->sending_cpu == rap->sender) && + (msg->sw_ack_vector) && + (msg->msg_type != MSG_NOOP)) { + /* + * make everyone else ignore this message + */ + msg->canceled = 1; + slot = msg - bcp->va_queue_first; + count++; + /* + * only reset the resource if it is still pending + */ + mmr = uv_read_local_mmr + (UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE); + msg_res = ((msg->sw_ack_vector << 8) | + msg->sw_ack_vector); + if (mmr & msg_res) { + stat->d_rcanceled++; + uv_write_local_mmr( + UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE_ALIAS, + msg_res); } - printk("\n"); } } - return count; + return; } /* - * Examine the payload queue on all the distribution nodes to see - * which messages have not been seen, and which cpu(s) have not seen them. - * - * Returns the number of cpu's that have not responded. + * Use IPI to get all target uvhubs to release resources held by + * a given sending cpu number. */ -static int uv_examine_destinations(struct bau_target_nodemask *distribution) +static void uv_reset_with_ipi(struct bau_target_uvhubmask *distribution, + int sender) { - int sender; - int i; - int count = 0; + int uvhub; + int cpu; + cpumask_t mask; + struct reset_args reset_args; + + reset_args.sender = sender; - sender = smp_processor_id(); - for (i = 0; i < sizeof(struct bau_target_nodemask) * BITSPERBYTE; i++) { - if (!bau_node_isset(i, distribution)) + cpus_clear(mask); + /* find a single cpu for each uvhub in this distribution mask */ + for (uvhub = 0; + uvhub < sizeof(struct bau_target_uvhubmask) * BITSPERBYTE; + uvhub++) { + if (!bau_uvhub_isset(uvhub, distribution)) continue; - count += uv_examine_destination(uv_bau_table_bases[i], sender); + /* find a cpu for this uvhub */ + cpu = uvhub_to_first_cpu(uvhub); + cpu_set(cpu, mask); } - return count; + /* IPI all cpus; Preemption is already disabled */ + smp_call_function_many(&mask, uv_do_reset, (void *)&reset_args, 1); + return; +} + +static inline unsigned long +cycles_2_us(unsigned long long cyc) +{ + unsigned long long ns; + unsigned long us; + ns = (cyc * per_cpu(cyc2ns, smp_processor_id())) + >> CYC2NS_SCALE_FACTOR; + us = ns / 1000; + return us; } /* - * wait for completion of a broadcast message - * - * return COMPLETE, RETRY or GIVEUP + * wait for all cpus on this hub to finish their sends and go quiet + * leaves uvhub_quiesce set so that no new broadcasts are started by + * bau_flush_send_and_wait() + */ +static inline void +quiesce_local_uvhub(struct bau_control *hmaster) +{ + atomic_add_short_return(1, (struct atomic_short *) + &hmaster->uvhub_quiesce); +} + +/* + * mark this quiet-requestor as done + */ +static inline void +end_uvhub_quiesce(struct bau_control *hmaster) +{ + atomic_add_short_return(-1, (struct atomic_short *) + &hmaster->uvhub_quiesce); +} + +/* + * Wait for completion of a broadcast software ack message + * return COMPLETE, RETRY(PLUGGED or TIMEOUT) or GIVEUP */ static int uv_wait_completion(struct bau_desc *bau_desc, - unsigned long mmr_offset, int right_shift) + unsigned long mmr_offset, int right_shift, int this_cpu, + struct bau_control *bcp, struct bau_control *smaster, long try) { - int exams = 0; - long destination_timeouts = 0; - long source_timeouts = 0; + int relaxes = 0; unsigned long descriptor_status; + unsigned long mmr; + unsigned long mask; + cycles_t ttime; + cycles_t timeout_time; + struct ptc_stats *stat = &per_cpu(ptcstats, this_cpu); + struct bau_control *hmaster; + + hmaster = bcp->uvhub_master; + timeout_time = get_cycles() + bcp->timeout_interval; + /* spin on the status MMR, waiting for it to go idle */ while ((descriptor_status = (((unsigned long) uv_read_local_mmr(mmr_offset) >> right_shift) & UV_ACT_STATUS_MASK)) != DESC_STATUS_IDLE) { - if (descriptor_status == DESC_STATUS_SOURCE_TIMEOUT) { - source_timeouts++; - if (source_timeouts > SOURCE_TIMEOUT_LIMIT) - source_timeouts = 0; - __get_cpu_var(ptcstats).s_retry++; - return FLUSH_RETRY; - } /* - * spin here looking for progress at the destinations + * Our software ack messages may be blocked because there are + * no swack resources available. As long as none of them + * has timed out hardware will NACK our message and its + * state will stay IDLE. */ - if (descriptor_status == DESC_STATUS_DESTINATION_TIMEOUT) { - destination_timeouts++; - if (destination_timeouts > DESTINATION_TIMEOUT_LIMIT) { - /* - * returns number of cpus not responding - */ - if (uv_examine_destinations - (&bau_desc->distribution) == 0) { - __get_cpu_var(ptcstats).d_retry++; - return FLUSH_RETRY; - } - exams++; - if (exams >= uv_bau_retry_limit) { - printk(KERN_DEBUG - "uv_flush_tlb_others"); - printk("giving up on cpu %d\n", - smp_processor_id()); + if (descriptor_status == DESC_STATUS_SOURCE_TIMEOUT) { + stat->s_stimeout++; + return FLUSH_GIVEUP; + } else if (descriptor_status == + DESC_STATUS_DESTINATION_TIMEOUT) { + stat->s_dtimeout++; + ttime = get_cycles(); + + /* + * Our retries may be blocked by all destination + * swack resources being consumed, and a timeout + * pending. In that case hardware returns the + * ERROR that looks like a destination timeout. + */ + if (cycles_2_us(ttime - bcp->send_message) < BIOS_TO) { + bcp->conseccompletes = 0; + return FLUSH_RETRY_PLUGGED; + } + + bcp->conseccompletes = 0; + return FLUSH_RETRY_TIMEOUT; + } else { + /* + * descriptor_status is still BUSY + */ + cpu_relax(); + relaxes++; + if (relaxes >= 10000) { + relaxes = 0; + if (get_cycles() > timeout_time) { + quiesce_local_uvhub(hmaster); + + /* single-thread the register change */ + spin_lock(&hmaster->masks_lock); + mmr = uv_read_local_mmr(mmr_offset); + mask = 0UL; + mask |= (3UL < right_shift); + mask = ~mask; + mmr &= mask; + uv_write_local_mmr(mmr_offset, mmr); + spin_unlock(&hmaster->masks_lock); + end_uvhub_quiesce(hmaster); + stat->s_busy++; return FLUSH_GIVEUP; } - /* - * delays can hang the simulator - udelay(1000); - */ - destination_timeouts = 0; } } - cpu_relax(); } + bcp->conseccompletes++; return FLUSH_COMPLETE; } +static inline cycles_t +sec_2_cycles(unsigned long sec) +{ + unsigned long ns; + cycles_t cyc; + + ns = sec * 1000000000; + cyc = (ns << CYC2NS_SCALE_FACTOR)/(per_cpu(cyc2ns, smp_processor_id())); + return cyc; +} + +/* + * conditionally add 1 to *v, unless *v is >= u + * return 0 if we cannot add 1 to *v because it is >= u + * return 1 if we can add 1 to *v because it is < u + * the add is atomic + * + * This is close to atomic_add_unless(), but this allows the 'u' value + * to be lowered below the current 'v'. atomic_add_unless can only stop + * on equal. + */ +static inline int atomic_inc_unless_ge(spinlock_t *lock, atomic_t *v, int u) +{ + spin_lock(lock); + if (atomic_read(v) >= u) { + spin_unlock(lock); + return 0; + } + atomic_inc(v); + spin_unlock(lock); + return 1; +} + /** * uv_flush_send_and_wait * - * Send a broadcast and wait for a broadcast message to complete. + * Send a broadcast and wait for it to complete. * - * The flush_mask contains the cpus the broadcast was sent to. + * The flush_mask contains the cpus the broadcast is to be sent to, plus + * cpus that are on the local uvhub. * - * Returns NULL if all remote flushing was done. The mask is zeroed. + * Returns NULL if all flushing represented in the mask was done. The mask + * is zeroed. * Returns @flush_mask if some remote flushing remains to be done. The - * mask will have some bits still set. + * mask will have some bits still set, representing any cpus on the local + * uvhub (not current cpu) and any on remote uvhubs if the broadcast failed. */ -const struct cpumask *uv_flush_send_and_wait(int cpu, int this_pnode, - struct bau_desc *bau_desc, - struct cpumask *flush_mask) +const struct cpumask *uv_flush_send_and_wait(struct bau_desc *bau_desc, + struct cpumask *flush_mask, + struct bau_control *bcp) { - int completion_status = 0; int right_shift; - int tries = 0; - int pnode; + int uvhub; int bit; + int completion_status = 0; + int seq_number = 0; + long try = 0; + int cpu = bcp->uvhub_cpu; + int this_cpu = bcp->cpu; + int this_uvhub = bcp->uvhub; unsigned long mmr_offset; unsigned long index; cycles_t time1; cycles_t time2; + struct ptc_stats *stat = &per_cpu(ptcstats, bcp->cpu); + struct bau_control *smaster = bcp->socket_master; + struct bau_control *hmaster = bcp->uvhub_master; + + /* + * Spin here while there are hmaster->max_concurrent or more active + * descriptors. This is the per-uvhub 'throttle'. + */ + if (!atomic_inc_unless_ge(&hmaster->uvhub_lock, + &hmaster->active_descriptor_count, + hmaster->max_concurrent)) { + stat->s_throttles++; + do { + cpu_relax(); + } while (!atomic_inc_unless_ge(&hmaster->uvhub_lock, + &hmaster->active_descriptor_count, + hmaster->max_concurrent)); + } + + while (hmaster->uvhub_quiesce) + cpu_relax(); if (cpu < UV_CPUS_PER_ACT_STATUS) { mmr_offset = UVH_LB_BAU_SB_ACTIVATION_STATUS_0; @@ -267,24 +557,108 @@ const struct cpumask *uv_flush_send_and_wait(int cpu, int this_pnode, } time1 = get_cycles(); do { - tries++; + /* + * Every message from any given cpu gets a unique message + * sequence number. But retries use that same number. + * Our message may have timed out at the destination because + * all sw-ack resources are in use and there is a timeout + * pending there. In that case, our last send never got + * placed into the queue and we need to persist until it + * does. + * + * Make any retry a type MSG_RETRY so that the destination will + * free any resource held by a previous message from this cpu. + */ + if (try == 0) { + /* use message type set by the caller the first time */ + seq_number = bcp->message_number++; + } else { + /* use RETRY type on all the rest; same sequence */ + bau_desc->header.msg_type = MSG_RETRY; + stat->s_retry_messages++; + } + bau_desc->header.sequence = seq_number; index = (1UL << UVH_LB_BAU_SB_ACTIVATION_CONTROL_PUSH_SHFT) | - cpu; + bcp->uvhub_cpu; + bcp->send_message = get_cycles(); + uv_write_local_mmr(UVH_LB_BAU_SB_ACTIVATION_CONTROL, index); + + try++; completion_status = uv_wait_completion(bau_desc, mmr_offset, - right_shift); - } while (completion_status == FLUSH_RETRY); + right_shift, this_cpu, bcp, smaster, try); + + if (completion_status == FLUSH_RETRY_PLUGGED) { + /* + * Our retries may be blocked by all destination swack + * resources being consumed, and a timeout pending. In + * that case hardware immediately returns the ERROR + * that looks like a destination timeout. + */ + udelay(TIMEOUT_DELAY); + bcp->plugged_tries++; + if (bcp->plugged_tries >= PLUGSB4RESET) { + bcp->plugged_tries = 0; + quiesce_local_uvhub(hmaster); + spin_lock(&hmaster->queue_lock); + uv_reset_with_ipi(&bau_desc->distribution, + this_cpu); + spin_unlock(&hmaster->queue_lock); + end_uvhub_quiesce(hmaster); + bcp->ipi_attempts++; + stat->s_resets_plug++; + } + } else if (completion_status == FLUSH_RETRY_TIMEOUT) { + hmaster->max_concurrent = 1; + bcp->timeout_tries++; + udelay(TIMEOUT_DELAY); + if (bcp->timeout_tries >= TIMEOUTSB4RESET) { + bcp->timeout_tries = 0; + quiesce_local_uvhub(hmaster); + spin_lock(&hmaster->queue_lock); + uv_reset_with_ipi(&bau_desc->distribution, + this_cpu); + spin_unlock(&hmaster->queue_lock); + end_uvhub_quiesce(hmaster); + bcp->ipi_attempts++; + stat->s_resets_timeout++; + } + } + if (bcp->ipi_attempts >= 3) { + bcp->ipi_attempts = 0; + completion_status = FLUSH_GIVEUP; + break; + } + cpu_relax(); + } while ((completion_status == FLUSH_RETRY_PLUGGED) || + (completion_status == FLUSH_RETRY_TIMEOUT)); time2 = get_cycles(); - __get_cpu_var(ptcstats).sflush += (time2 - time1); - if (tries > 1) - __get_cpu_var(ptcstats).retriesok++; - if (completion_status == FLUSH_GIVEUP) { + if ((completion_status == FLUSH_COMPLETE) && (bcp->conseccompletes > 5) + && (hmaster->max_concurrent < hmaster->max_concurrent_constant)) + hmaster->max_concurrent++; + + /* + * hold any cpu not timing out here; no other cpu currently held by + * the 'throttle' should enter the activation code + */ + while (hmaster->uvhub_quiesce) + cpu_relax(); + atomic_dec(&hmaster->active_descriptor_count); + + /* guard against cycles wrap */ + if (time2 > time1) + stat->s_time += (time2 - time1); + else + stat->s_requestor--; /* don't count this one */ + if (completion_status == FLUSH_COMPLETE && try > 1) + stat->s_retriesok++; + else if (completion_status == FLUSH_GIVEUP) { /* * Cause the caller to do an IPI-style TLB shootdown on - * the cpu's, all of which are still in the mask. + * the target cpu's, all of which are still in the mask. */ - __get_cpu_var(ptcstats).ptc_i++; + stat->s_giveup++; return flush_mask; } @@ -293,18 +667,17 @@ const struct cpumask *uv_flush_send_and_wait(int cpu, int this_pnode, * use the IPI method of shootdown on them. */ for_each_cpu(bit, flush_mask) { - pnode = uv_cpu_to_pnode(bit); - if (pnode == this_pnode) + uvhub = uv_cpu_to_blade_id(bit); + if (uvhub == this_uvhub) continue; cpumask_clear_cpu(bit, flush_mask); } if (!cpumask_empty(flush_mask)) return flush_mask; + return NULL; } -static DEFINE_PER_CPU(cpumask_var_t, uv_flush_tlb_mask); - /** * uv_flush_tlb_others - globally purge translation cache of a virtual * address or all TLB's @@ -321,8 +694,8 @@ static DEFINE_PER_CPU(cpumask_var_t, uv_flush_tlb_mask); * The caller has derived the cpumask from the mm_struct. This function * is called only if there are bits set in the mask. (e.g. flush_tlb_page()) * - * The cpumask is converted into a nodemask of the nodes containing - * the cpus. + * The cpumask is converted into a uvhubmask of the uvhubs containing + * those cpus. * * Note that this function should be called with preemption disabled. * @@ -334,52 +707,82 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask, struct mm_struct *mm, unsigned long va, unsigned int cpu) { - struct cpumask *flush_mask = __get_cpu_var(uv_flush_tlb_mask); - int i; - int bit; - int pnode; - int uv_cpu; - int this_pnode; + int remotes; + int tcpu; + int uvhub; int locals = 0; struct bau_desc *bau_desc; + struct cpumask *flush_mask; + struct ptc_stats *stat; + struct bau_control *bcp; - cpumask_andnot(flush_mask, cpumask, cpumask_of(cpu)); + if (nobau) + return cpumask; - uv_cpu = uv_blade_processor_id(); - this_pnode = uv_hub_info->pnode; - bau_desc = __get_cpu_var(bau_control).descriptor_base; - bau_desc += UV_ITEMS_PER_DESCRIPTOR * uv_cpu; + bcp = &per_cpu(bau_control, cpu); + /* + * Each sending cpu has a per-cpu mask which it fills from the caller's + * cpu mask. Only remote cpus are converted to uvhubs and copied. + */ + flush_mask = (struct cpumask *)per_cpu(uv_flush_tlb_mask, cpu); + /* + * copy cpumask to flush_mask, removing current cpu + * (current cpu should already have been flushed by the caller and + * should never be returned if we return flush_mask) + */ + cpumask_andnot(flush_mask, cpumask, cpumask_of(cpu)); + if (cpu_isset(cpu, *cpumask)) + locals++; /* current cpu was targeted */ - bau_nodes_clear(&bau_desc->distribution, UV_DISTRIBUTION_SIZE); + bau_desc = bcp->descriptor_base; + bau_desc += UV_ITEMS_PER_DESCRIPTOR * bcp->uvhub_cpu; - i = 0; - for_each_cpu(bit, flush_mask) { - pnode = uv_cpu_to_pnode(bit); - BUG_ON(pnode > (UV_DISTRIBUTION_SIZE - 1)); - if (pnode == this_pnode) { + bau_uvhubs_clear(&bau_desc->distribution, UV_DISTRIBUTION_SIZE); + remotes = 0; + for_each_cpu(tcpu, flush_mask) { + uvhub = uv_cpu_to_blade_id(tcpu); + if (uvhub == bcp->uvhub) { locals++; continue; } - bau_node_set(pnode - uv_partition_base_pnode, - &bau_desc->distribution); - i++; + bau_uvhub_set(uvhub, &bau_desc->distribution); + remotes++; } - if (i == 0) { + if (remotes == 0) { /* - * no off_node flushing; return status for local node + * No off_hub flushing; return status for local hub. + * Return the caller's mask if all were local (the current + * cpu may be in that mask). */ if (locals) - return flush_mask; + return cpumask; else return NULL; } - __get_cpu_var(ptcstats).requestor++; - __get_cpu_var(ptcstats).ntargeted += i; + stat = &per_cpu(ptcstats, cpu); + stat->s_requestor++; + stat->s_ntargcpu += remotes; + remotes = bau_uvhub_weight(&bau_desc->distribution); + stat->s_ntarguvhub += remotes; + if (remotes >= 16) + stat->s_ntarguvhub16++; + else if (remotes >= 8) + stat->s_ntarguvhub8++; + else if (remotes >= 4) + stat->s_ntarguvhub4++; + else if (remotes >= 2) + stat->s_ntarguvhub2++; + else + stat->s_ntarguvhub1++; bau_desc->payload.address = va; bau_desc->payload.sending_cpu = cpu; - return uv_flush_send_and_wait(uv_cpu, this_pnode, bau_desc, flush_mask); + /* + * uv_flush_send_and_wait returns null if all cpu's were messaged, or + * the adjusted flush_mask if any cpu's were not messaged. + */ + return uv_flush_send_and_wait(bau_desc, flush_mask, bcp); } /* @@ -388,87 +791,70 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask, * * We received a broadcast assist message. * - * Interrupts may have been disabled; this interrupt could represent + * Interrupts are disabled; this interrupt could represent * the receipt of several messages. * - * All cores/threads on this node get this interrupt. - * The last one to see it does the s/w ack. + * All cores/threads on this hub get this interrupt. + * The last one to see it does the software ack. * (the resource will not be freed until noninterruptable cpus see this - * interrupt; hardware will timeout the s/w ack and reply ERROR) + * interrupt; hardware may timeout the s/w ack and reply ERROR) */ void uv_bau_message_interrupt(struct pt_regs *regs) { - struct bau_payload_queue_entry *va_queue_first; - struct bau_payload_queue_entry *va_queue_last; - struct bau_payload_queue_entry *msg; - struct pt_regs *old_regs = set_irq_regs(regs); - cycles_t time1; - cycles_t time2; - int msg_slot; - int sw_ack_slot; - int fw; int count = 0; - unsigned long local_pnode; - - ack_APIC_irq(); - exit_idle(); - irq_enter(); - - time1 = get_cycles(); - - local_pnode = uv_blade_to_pnode(uv_numa_blade_id()); - - va_queue_first = __get_cpu_var(bau_control).va_queue_first; - va_queue_last = __get_cpu_var(bau_control).va_queue_last; - - msg = __get_cpu_var(bau_control).bau_msg_head; + cycles_t time_start; + struct bau_payload_queue_entry *msg; + struct bau_control *bcp; + struct ptc_stats *stat; + struct msg_desc msgdesc; + + time_start = get_cycles(); + bcp = &per_cpu(bau_control, smp_processor_id()); + stat = &per_cpu(ptcstats, smp_processor_id()); + msgdesc.va_queue_first = bcp->va_queue_first; + msgdesc.va_queue_last = bcp->va_queue_last; + msg = bcp->bau_msg_head; while (msg->sw_ack_vector) { count++; - fw = msg->sw_ack_vector; - msg_slot = msg - va_queue_first; - sw_ack_slot = ffs(fw) - 1; - - uv_bau_process_message(msg, msg_slot, sw_ack_slot); - + msgdesc.msg_slot = msg - msgdesc.va_queue_first; + msgdesc.sw_ack_slot = ffs(msg->sw_ack_vector) - 1; + msgdesc.msg = msg; + uv_bau_process_message(&msgdesc, bcp); msg++; - if (msg > va_queue_last) - msg = va_queue_first; - __get_cpu_var(bau_control).bau_msg_head = msg; + if (msg > msgdesc.va_queue_last) + msg = msgdesc.va_queue_first; + bcp->bau_msg_head = msg; } + stat->d_time += (get_cycles() - time_start); if (!count) - __get_cpu_var(ptcstats).nomsg++; + stat->d_nomsg++; else if (count > 1) - __get_cpu_var(ptcstats).multmsg++; - - time2 = get_cycles(); - __get_cpu_var(ptcstats).dflush += (time2 - time1); - - irq_exit(); - set_irq_regs(old_regs); + stat->d_multmsg++; + ack_APIC_irq(); } /* * uv_enable_timeouts * - * Each target blade (i.e. blades that have cpu's) needs to have + * Each target uvhub (i.e. a uvhub that has no cpu's) needs to have * shootdown message timeouts enabled. The timeout does not cause * an interrupt, but causes an error message to be returned to * the sender. */ static void uv_enable_timeouts(void) { - int blade; - int nblades; + int uvhub; + int nuvhubs; int pnode; unsigned long mmr_image; - nblades = uv_num_possible_blades(); + nuvhubs = uv_num_possible_blades(); - for (blade = 0; blade < nblades; blade++) { - if (!uv_blade_nr_possible_cpus(blade)) + for (uvhub = 0; uvhub < nuvhubs; uvhub++) { + if (!uv_blade_nr_possible_cpus(uvhub)) continue; - pnode = uv_blade_to_pnode(blade); + pnode = uv_blade_to_pnode(uvhub); mmr_image = uv_read_global_mmr64(pnode, UVH_LB_BAU_MISC_CONTROL); /* @@ -478,16 +864,16 @@ static void uv_enable_timeouts(void) * To program the period, the SOFT_ACK_MODE must be off. */ mmr_image &= ~((unsigned long)1 << - UV_ENABLE_INTD_SOFT_ACK_MODE_SHIFT); + UVH_LB_BAU_MISC_CONTROL_ENABLE_INTD_SOFT_ACK_MODE_SHFT); uv_write_global_mmr64 (pnode, UVH_LB_BAU_MISC_CONTROL, mmr_image); /* * Set the 4-bit period. */ mmr_image &= ~((unsigned long)0xf << - UV_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHIFT); + UVH_LB_BAU_MISC_CONTROL_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHFT); mmr_image |= (UV_INTD_SOFT_ACK_TIMEOUT_PERIOD << - UV_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHIFT); + UVH_LB_BAU_MISC_CONTROL_INTD_SOFT_ACK_TIMEOUT_PERIOD_SHFT); uv_write_global_mmr64 (pnode, UVH_LB_BAU_MISC_CONTROL, mmr_image); /* @@ -496,7 +882,7 @@ static void uv_enable_timeouts(void) * indicated in bits 2:0 (7 causes all of them to timeout). */ mmr_image |= ((unsigned long)1 << - UV_ENABLE_INTD_SOFT_ACK_MODE_SHIFT); + UVH_LB_BAU_MISC_CONTROL_ENABLE_INTD_SOFT_ACK_MODE_SHFT); uv_write_global_mmr64 (pnode, UVH_LB_BAU_MISC_CONTROL, mmr_image); } @@ -521,9 +907,20 @@ static void uv_ptc_seq_stop(struct seq_file *file, void *data) { } +static inline unsigned long long +millisec_2_cycles(unsigned long millisec) +{ + unsigned long ns; + unsigned long long cyc; + + ns = millisec * 1000; + cyc = (ns << CYC2NS_SCALE_FACTOR)/(per_cpu(cyc2ns, smp_processor_id())); + return cyc; +} + /* - * Display the statistics thru /proc - * data points to the cpu number + * Display the statistics thru /proc. + * 'data' points to the cpu number */ static int uv_ptc_seq_show(struct seq_file *file, void *data) { @@ -534,78 +931,155 @@ static int uv_ptc_seq_show(struct seq_file *file, void *data) if (!cpu) { seq_printf(file, - "# cpu requestor requestee one all sretry dretry ptc_i "); + "# cpu sent stime numuvhubs numuvhubs16 numuvhubs8 "); seq_printf(file, - "sw_ack sflush dflush sok dnomsg dmult starget\n"); + "numuvhubs4 numuvhubs2 numuvhubs1 numcpus dto "); + seq_printf(file, + "retries rok resetp resett giveup sto bz throt "); + seq_printf(file, + "sw_ack recv rtime all "); + seq_printf(file, + "one mult none retry canc nocan reset rcan\n"); } if (cpu < num_possible_cpus() && cpu_online(cpu)) { stat = &per_cpu(ptcstats, cpu); - seq_printf(file, "cpu %d %ld %ld %ld %ld %ld %ld %ld ", - cpu, stat->requestor, - stat->requestee, stat->onetlb, stat->alltlb, - stat->s_retry, stat->d_retry, stat->ptc_i); - seq_printf(file, "%lx %ld %ld %ld %ld %ld %ld\n", + /* source side statistics */ + seq_printf(file, + "cpu %d %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld ", + cpu, stat->s_requestor, cycles_2_us(stat->s_time), + stat->s_ntarguvhub, stat->s_ntarguvhub16, + stat->s_ntarguvhub8, stat->s_ntarguvhub4, + stat->s_ntarguvhub2, stat->s_ntarguvhub1, + stat->s_ntargcpu, stat->s_dtimeout); + seq_printf(file, "%ld %ld %ld %ld %ld %ld %ld %ld ", + stat->s_retry_messages, stat->s_retriesok, + stat->s_resets_plug, stat->s_resets_timeout, + stat->s_giveup, stat->s_stimeout, + stat->s_busy, stat->s_throttles); + /* destination side statistics */ + seq_printf(file, + "%lx %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld %ld\n", uv_read_global_mmr64(uv_cpu_to_pnode(cpu), UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE), - stat->sflush, stat->dflush, - stat->retriesok, stat->nomsg, - stat->multmsg, stat->ntargeted); + stat->d_requestee, cycles_2_us(stat->d_time), + stat->d_alltlb, stat->d_onetlb, stat->d_multmsg, + stat->d_nomsg, stat->d_retries, stat->d_canceled, + stat->d_nocanceled, stat->d_resets, + stat->d_rcanceled); } return 0; } /* + * -1: resetf the statistics * 0: display meaning of the statistics - * >0: retry limit + * >0: maximum concurrent active descriptors per uvhub (throttle) */ static ssize_t uv_ptc_proc_write(struct file *file, const char __user *user, size_t count, loff_t *data) { - long newmode; + int cpu; + long input_arg; char optstr[64]; + struct ptc_stats *stat; + struct bau_control *bcp; if (count == 0 || count > sizeof(optstr)) return -EINVAL; if (copy_from_user(optstr, user, count)) return -EFAULT; optstr[count - 1] = '\0'; - if (strict_strtoul(optstr, 10, &newmode) < 0) { + if (strict_strtol(optstr, 10, &input_arg) < 0) { printk(KERN_DEBUG "%s is invalid\n", optstr); return -EINVAL; } - if (newmode == 0) { + if (input_arg == 0) { printk(KERN_DEBUG "# cpu: cpu number\n"); + printk(KERN_DEBUG "Sender statistics:\n"); + printk(KERN_DEBUG + "sent: number of shootdown messages sent\n"); + printk(KERN_DEBUG + "stime: time spent sending messages\n"); + printk(KERN_DEBUG + "numuvhubs: number of hubs targeted with shootdown\n"); + printk(KERN_DEBUG + "numuvhubs16: number times 16 or more hubs targeted\n"); + printk(KERN_DEBUG + "numuvhubs8: number times 8 or more hubs targeted\n"); + printk(KERN_DEBUG + "numuvhubs4: number times 4 or more hubs targeted\n"); + printk(KERN_DEBUG + "numuvhubs2: number times 2 or more hubs targeted\n"); + printk(KERN_DEBUG + "numuvhubs1: number times 1 hub targeted\n"); + printk(KERN_DEBUG + "numcpus: number of cpus targeted with shootdown\n"); + printk(KERN_DEBUG + "dto: number of destination timeouts\n"); + printk(KERN_DEBUG + "retries: destination timeout retries sent\n"); + printk(KERN_DEBUG + "rok: : destination timeouts successfully retried\n"); + printk(KERN_DEBUG + "resetp: ipi-style resource resets for plugs\n"); + printk(KERN_DEBUG + "resett: ipi-style resource resets for timeouts\n"); + printk(KERN_DEBUG + "giveup: fall-backs to ipi-style shootdowns\n"); + printk(KERN_DEBUG + "sto: number of source timeouts\n"); + printk(KERN_DEBUG + "bz: number of stay-busy's\n"); + printk(KERN_DEBUG + "throt: number times spun in throttle\n"); + printk(KERN_DEBUG "Destination side statistics:\n"); printk(KERN_DEBUG - "requestor: times this cpu was the flush requestor\n"); + "sw_ack: image of UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE\n"); printk(KERN_DEBUG - "requestee: times this cpu was requested to flush its TLBs\n"); + "recv: shootdown messages received\n"); printk(KERN_DEBUG - "one: times requested to flush a single address\n"); + "rtime: time spent processing messages\n"); printk(KERN_DEBUG - "all: times requested to flush all TLB's\n"); + "all: shootdown all-tlb messages\n"); printk(KERN_DEBUG - "sretry: number of retries of source-side timeouts\n"); + "one: shootdown one-tlb messages\n"); printk(KERN_DEBUG - "dretry: number of retries of destination-side timeouts\n"); + "mult: interrupts that found multiple messages\n"); printk(KERN_DEBUG - "ptc_i: times UV fell through to IPI-style flushes\n"); + "none: interrupts that found no messages\n"); printk(KERN_DEBUG - "sw_ack: image of UVH_LB_BAU_INTD_SOFTWARE_ACKNOWLEDGE\n"); + "retry: number of retry messages processed\n"); printk(KERN_DEBUG - "sflush_us: cycles spent in uv_flush_tlb_others()\n"); + "canc: number messages canceled by retries\n"); printk(KERN_DEBUG - "dflush_us: cycles spent in handling flush requests\n"); - printk(KERN_DEBUG "sok: successes on retry\n"); - printk(KERN_DEBUG "dnomsg: interrupts with no message\n"); + "nocan: number retries that found nothing to cancel\n"); printk(KERN_DEBUG - "dmult: interrupts with multiple messages\n"); - printk(KERN_DEBUG "starget: nodes targeted\n"); + "reset: number of ipi-style reset requests processed\n"); + printk(KERN_DEBUG + "rcan: number messages canceled by reset requests\n"); + } else if (input_arg == -1) { + for_each_present_cpu(cpu) { + stat = &per_cpu(ptcstats, cpu); + memset(stat, 0, sizeof(struct ptc_stats)); + } } else { - uv_bau_retry_limit = newmode; - printk(KERN_DEBUG "timeout retry limit:%d\n", - uv_bau_retry_limit); + uv_bau_max_concurrent = input_arg; + bcp = &per_cpu(bau_control, smp_processor_id()); + if (uv_bau_max_concurrent < 1 || + uv_bau_max_concurrent > bcp->cpus_in_uvhub) { + printk(KERN_DEBUG + "Error: BAU max concurrent %d; %d is invalid\n", + bcp->max_concurrent, uv_bau_max_concurrent); + return -EINVAL; + } + printk(KERN_DEBUG "Set BAU max concurrent:%d\n", + uv_bau_max_concurrent); + for_each_present_cpu(cpu) { + bcp = &per_cpu(bau_control, cpu); + bcp->max_concurrent = uv_bau_max_concurrent; + } } return count; @@ -649,79 +1123,30 @@ static int __init uv_ptc_init(void) } /* - * begin the initialization of the per-blade control structures - */ -static struct bau_control * __init uv_table_bases_init(int blade, int node) -{ - int i; - struct bau_msg_status *msp; - struct bau_control *bau_tabp; - - bau_tabp = - kmalloc_node(sizeof(struct bau_control), GFP_KERNEL, node); - BUG_ON(!bau_tabp); - - bau_tabp->msg_statuses = - kmalloc_node(sizeof(struct bau_msg_status) * - DEST_Q_SIZE, GFP_KERNEL, node); - BUG_ON(!bau_tabp->msg_statuses); - - for (i = 0, msp = bau_tabp->msg_statuses; i < DEST_Q_SIZE; i++, msp++) - bau_cpubits_clear(&msp->seen_by, (int) - uv_blade_nr_possible_cpus(blade)); - - uv_bau_table_bases[blade] = bau_tabp; - - return bau_tabp; -} - -/* - * finish the initialization of the per-blade control structures - */ -static void __init -uv_table_bases_finish(int blade, - struct bau_control *bau_tablesp, - struct bau_desc *adp) -{ - struct bau_control *bcp; - int cpu; - - for_each_present_cpu(cpu) { - if (blade != uv_cpu_to_blade_id(cpu)) - continue; - - bcp = (struct bau_control *)&per_cpu(bau_control, cpu); - bcp->bau_msg_head = bau_tablesp->va_queue_first; - bcp->va_queue_first = bau_tablesp->va_queue_first; - bcp->va_queue_last = bau_tablesp->va_queue_last; - bcp->msg_statuses = bau_tablesp->msg_statuses; - bcp->descriptor_base = adp; - } -} - -/* * initialize the sending side's sending buffers */ -static struct bau_desc * __init +static void uv_activation_descriptor_init(int node, int pnode) { int i; + int cpu; unsigned long pa; unsigned long m; unsigned long n; - struct bau_desc *adp; - struct bau_desc *ad2; + struct bau_desc *bau_desc; + struct bau_desc *bd2; + struct bau_control *bcp; /* * each bau_desc is 64 bytes; there are 8 (UV_ITEMS_PER_DESCRIPTOR) - * per cpu; and up to 32 (UV_ADP_SIZE) cpu's per blade + * per cpu; and up to 32 (UV_ADP_SIZE) cpu's per uvhub */ - adp = (struct bau_desc *)kmalloc_node(sizeof(struct bau_desc)* + bau_desc = (struct bau_desc *)kmalloc_node(sizeof(struct bau_desc)* UV_ADP_SIZE*UV_ITEMS_PER_DESCRIPTOR, GFP_KERNEL, node); - BUG_ON(!adp); + BUG_ON(!bau_desc); - pa = uv_gpa(adp); /* need the real nasid*/ - n = uv_gpa_to_pnode(pa); + pa = uv_gpa(bau_desc); /* need the real nasid*/ + n = pa >> uv_nshift; m = pa & uv_mmask; uv_write_global_mmr64(pnode, UVH_LB_BAU_SB_DESCRIPTOR_BASE, @@ -730,96 +1155,188 @@ uv_activation_descriptor_init(int node, int pnode) /* * initializing all 8 (UV_ITEMS_PER_DESCRIPTOR) descriptors for each * cpu even though we only use the first one; one descriptor can - * describe a broadcast to 256 nodes. + * describe a broadcast to 256 uv hubs. */ - for (i = 0, ad2 = adp; i < (UV_ADP_SIZE*UV_ITEMS_PER_DESCRIPTOR); - i++, ad2++) { - memset(ad2, 0, sizeof(struct bau_desc)); - ad2->header.sw_ack_flag = 1; + for (i = 0, bd2 = bau_desc; i < (UV_ADP_SIZE*UV_ITEMS_PER_DESCRIPTOR); + i++, bd2++) { + memset(bd2, 0, sizeof(struct bau_desc)); + bd2->header.sw_ack_flag = 1; /* - * base_dest_nodeid is the first node in the partition, so - * the bit map will indicate partition-relative node numbers. - * note that base_dest_nodeid is actually a nasid. + * base_dest_nodeid is the nasid (pnode<<1) of the first uvhub + * in the partition. The bit map will indicate uvhub numbers, + * which are 0-N in a partition. Pnodes are unique system-wide. */ - ad2->header.base_dest_nodeid = uv_partition_base_pnode << 1; - ad2->header.dest_subnodeid = 0x10; /* the LB */ - ad2->header.command = UV_NET_ENDPOINT_INTD; - ad2->header.int_both = 1; + bd2->header.base_dest_nodeid = uv_partition_base_pnode << 1; + bd2->header.dest_subnodeid = 0x10; /* the LB */ + bd2->header.command = UV_NET_ENDPOINT_INTD; + bd2->header.int_both = 1; /* * all others need to be set to zero: * fairness chaining multilevel count replied_to */ } - return adp; + for_each_present_cpu(cpu) { + if (pnode != uv_blade_to_pnode(uv_cpu_to_blade_id(cpu))) + continue; + bcp = &per_cpu(bau_control, cpu); + bcp->descriptor_base = bau_desc; + } } /* * initialize the destination side's receiving buffers + * entered for each uvhub in the partition + * - node is first node (kernel memory notion) on the uvhub + * - pnode is the uvhub's physical identifier */ -static struct bau_payload_queue_entry * __init -uv_payload_queue_init(int node, int pnode, struct bau_control *bau_tablesp) +static void +uv_payload_queue_init(int node, int pnode) { - struct bau_payload_queue_entry *pqp; - unsigned long pa; int pn; + int cpu; char *cp; + unsigned long pa; + struct bau_payload_queue_entry *pqp; + struct bau_payload_queue_entry *pqp_malloc; + struct bau_control *bcp; pqp = (struct bau_payload_queue_entry *) kmalloc_node( (DEST_Q_SIZE + 1) * sizeof(struct bau_payload_queue_entry), GFP_KERNEL, node); BUG_ON(!pqp); + pqp_malloc = pqp; cp = (char *)pqp + 31; pqp = (struct bau_payload_queue_entry *)(((unsigned long)cp >> 5) << 5); - bau_tablesp->va_queue_first = pqp; + + for_each_present_cpu(cpu) { + if (pnode != uv_cpu_to_pnode(cpu)) + continue; + /* for every cpu on this pnode: */ + bcp = &per_cpu(bau_control, cpu); + bcp->va_queue_first = pqp; + bcp->bau_msg_head = pqp; + bcp->va_queue_last = pqp + (DEST_Q_SIZE - 1); + } /* * need the pnode of where the memory was really allocated */ pa = uv_gpa(pqp); - pn = uv_gpa_to_pnode(pa); + pn = pa >> uv_nshift; uv_write_global_mmr64(pnode, UVH_LB_BAU_INTD_PAYLOAD_QUEUE_FIRST, ((unsigned long)pn << UV_PAYLOADQ_PNODE_SHIFT) | uv_physnodeaddr(pqp)); uv_write_global_mmr64(pnode, UVH_LB_BAU_INTD_PAYLOAD_QUEUE_TAIL, uv_physnodeaddr(pqp)); - bau_tablesp->va_queue_last = pqp + (DEST_Q_SIZE - 1); uv_write_global_mmr64(pnode, UVH_LB_BAU_INTD_PAYLOAD_QUEUE_LAST, (unsigned long) - uv_physnodeaddr(bau_tablesp->va_queue_last)); + uv_physnodeaddr(pqp + (DEST_Q_SIZE - 1))); + /* in effect, all msg_type's are set to MSG_NOOP */ memset(pqp, 0, sizeof(struct bau_payload_queue_entry) * DEST_Q_SIZE); - - return pqp; } /* - * Initialization of each UV blade's structures + * Initialization of each UV hub's structures */ -static int __init uv_init_blade(int blade) +static void __init uv_init_uvhub(int uvhub, int vector) { int node; int pnode; - unsigned long pa; unsigned long apicid; - struct bau_desc *adp; - struct bau_payload_queue_entry *pqp; - struct bau_control *bau_tablesp; - - node = blade_to_first_node(blade); - bau_tablesp = uv_table_bases_init(blade, node); - pnode = uv_blade_to_pnode(blade); - adp = uv_activation_descriptor_init(node, pnode); - pqp = uv_payload_queue_init(node, pnode, bau_tablesp); - uv_table_bases_finish(blade, bau_tablesp, adp); + + node = uvhub_to_first_node(uvhub); + pnode = uv_blade_to_pnode(uvhub); + uv_activation_descriptor_init(node, pnode); + uv_payload_queue_init(node, pnode); /* * the below initialization can't be in firmware because the * messaging IRQ will be determined by the OS */ - apicid = blade_to_first_apicid(blade); - pa = uv_read_global_mmr64(pnode, UVH_BAU_DATA_CONFIG); + apicid = uvhub_to_first_apicid(uvhub); uv_write_global_mmr64(pnode, UVH_BAU_DATA_CONFIG, - ((apicid << 32) | UV_BAU_MESSAGE)); - return 0; + ((apicid << 32) | vector)); +} + +/* + * initialize the bau_control structure for each cpu + */ +static void uv_init_per_cpu(int nuvhubs) +{ + int i, j, k; + int cpu; + int pnode; + int uvhub; + short socket = 0; + struct bau_control *bcp; + struct uvhub_desc *bdp; + struct socket_desc *sdp; + struct bau_control *hmaster = NULL; + struct bau_control *smaster = NULL; + struct socket_desc { + short num_cpus; + short cpu_number[16]; + }; + struct uvhub_desc { + short num_sockets; + short num_cpus; + short uvhub; + short pnode; + struct socket_desc socket[2]; + }; + struct uvhub_desc *uvhub_descs; + + uvhub_descs = (struct uvhub_desc *) + kmalloc(nuvhubs * sizeof(struct uvhub_desc), GFP_KERNEL); + memset(uvhub_descs, 0, nuvhubs * sizeof(struct uvhub_desc)); + for_each_present_cpu(cpu) { + bcp = &per_cpu(bau_control, cpu); + memset(bcp, 0, sizeof(struct bau_control)); + spin_lock_init(&bcp->masks_lock); + bcp->max_concurrent = uv_bau_max_concurrent; + pnode = uv_cpu_hub_info(cpu)->pnode; + uvhub = uv_cpu_hub_info(cpu)->numa_blade_id; + bdp = &uvhub_descs[uvhub]; + bdp->num_cpus++; + bdp->uvhub = uvhub; + bdp->pnode = pnode; + /* time interval to catch a hardware stay-busy bug */ + bcp->timeout_interval = millisec_2_cycles(3); + /* kludge: assume uv_hub.h is constant */ + socket = (cpu_physical_id(cpu)>>5)&1; + if (socket >= bdp->num_sockets) + bdp->num_sockets = socket+1; + sdp = &bdp->socket[socket]; + sdp->cpu_number[sdp->num_cpus] = cpu; + sdp->num_cpus++; + } + socket = 0; + for_each_possible_blade(uvhub) { + bdp = &uvhub_descs[uvhub]; + for (i = 0; i < bdp->num_sockets; i++) { + sdp = &bdp->socket[i]; + for (j = 0; j < sdp->num_cpus; j++) { + cpu = sdp->cpu_number[j]; + bcp = &per_cpu(bau_control, cpu); + bcp->cpu = cpu; + if (j == 0) { + smaster = bcp; + if (i == 0) + hmaster = bcp; + } + bcp->cpus_in_uvhub = bdp->num_cpus; + bcp->cpus_in_socket = sdp->num_cpus; + bcp->socket_master = smaster; + bcp->uvhub_master = hmaster; + for (k = 0; k < DEST_Q_SIZE; k++) + bcp->socket_acknowledge_count[k] = 0; + bcp->uvhub_cpu = + uv_cpu_hub_info(cpu)->blade_processor_id; + } + socket++; + } + } + kfree(uvhub_descs); } /* @@ -827,38 +1344,54 @@ static int __init uv_init_blade(int blade) */ static int __init uv_bau_init(void) { - int blade; - int nblades; + int uvhub; + int pnode; + int nuvhubs; int cur_cpu; + int vector; + unsigned long mmr; if (!is_uv_system()) return 0; + if (nobau) + return 0; + for_each_possible_cpu(cur_cpu) zalloc_cpumask_var_node(&per_cpu(uv_flush_tlb_mask, cur_cpu), GFP_KERNEL, cpu_to_node(cur_cpu)); - uv_bau_retry_limit = 1; + uv_bau_max_concurrent = MAX_BAU_CONCURRENT; + uv_nshift = uv_hub_info->m_val; uv_mmask = (1UL << uv_hub_info->m_val) - 1; - nblades = uv_num_possible_blades(); + nuvhubs = uv_num_possible_blades(); - uv_bau_table_bases = (struct bau_control **) - kmalloc(nblades * sizeof(struct bau_control *), GFP_KERNEL); - BUG_ON(!uv_bau_table_bases); + uv_init_per_cpu(nuvhubs); uv_partition_base_pnode = 0x7fffffff; - for (blade = 0; blade < nblades; blade++) - if (uv_blade_nr_possible_cpus(blade) && - (uv_blade_to_pnode(blade) < uv_partition_base_pnode)) - uv_partition_base_pnode = uv_blade_to_pnode(blade); - for (blade = 0; blade < nblades; blade++) - if (uv_blade_nr_possible_cpus(blade)) - uv_init_blade(blade); - - alloc_intr_gate(UV_BAU_MESSAGE, uv_bau_message_intr1); + for (uvhub = 0; uvhub < nuvhubs; uvhub++) + if (uv_blade_nr_possible_cpus(uvhub) && + (uv_blade_to_pnode(uvhub) < uv_partition_base_pnode)) + uv_partition_base_pnode = uv_blade_to_pnode(uvhub); + + vector = UV_BAU_MESSAGE; + for_each_possible_blade(uvhub) + if (uv_blade_nr_possible_cpus(uvhub)) + uv_init_uvhub(uvhub, vector); + uv_enable_timeouts(); + alloc_intr_gate(vector, uv_bau_message_intr1); + + for_each_possible_blade(uvhub) { + pnode = uv_blade_to_pnode(uvhub); + /* INIT the bau */ + uv_write_global_mmr64(pnode, UVH_LB_BAU_SB_ACTIVATION_CONTROL, + ((unsigned long)1 << 63)); + mmr = 1; /* should be 1 to broadcast to both sockets */ + uv_write_global_mmr64(pnode, UVH_BAU_DATA_BROADCAST, mmr); + } return 0; } -__initcall(uv_bau_init); -__initcall(uv_ptc_init); +core_initcall(uv_bau_init); +core_initcall(uv_ptc_init); diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c index 33399176512..142d70c74b0 100644 --- a/arch/x86/kernel/traps.c +++ b/arch/x86/kernel/traps.c @@ -15,6 +15,7 @@ #include <linux/kprobes.h> #include <linux/uaccess.h> #include <linux/kdebug.h> +#include <linux/kgdb.h> #include <linux/kernel.h> #include <linux/module.h> #include <linux/ptrace.h> @@ -108,15 +109,6 @@ static inline void preempt_conditional_cli(struct pt_regs *regs) dec_preempt_count(); } -#ifdef CONFIG_X86_32 -static inline void -die_if_kernel(const char *str, struct pt_regs *regs, long err) -{ - if (!user_mode_vm(regs)) - die(str, regs, err); -} -#endif - static void __kprobes do_trap(int trapnr, int signr, char *str, struct pt_regs *regs, long error_code, siginfo_t *info) @@ -460,6 +452,11 @@ void restart_nmi(void) /* May run on IST stack. */ dotraplinkage void __kprobes do_int3(struct pt_regs *regs, long error_code) { +#ifdef CONFIG_KGDB_LOW_LEVEL_TRAP + if (kgdb_ll_trap(DIE_INT3, "int3", regs, error_code, 3, SIGTRAP) + == NOTIFY_STOP) + return; +#endif /* CONFIG_KGDB_LOW_LEVEL_TRAP */ #ifdef CONFIG_KPROBES if (notify_die(DIE_INT3, "int3", regs, error_code, 3, SIGTRAP) == NOTIFY_STOP) @@ -534,17 +531,20 @@ dotraplinkage void __kprobes do_debug(struct pt_regs *regs, long error_code) get_debugreg(dr6, 6); + /* Filter out all the reserved bits which are preset to 1 */ + dr6 &= ~DR6_RESERVED; + /* Catch kmemcheck conditions first of all! */ if ((dr6 & DR_STEP) && kmemcheck_trap(regs)) return; /* DR6 may or may not be cleared by the CPU */ set_debugreg(0, 6); + /* * The processor cleared BTF, so don't mark that we need it set. */ - clear_tsk_thread_flag(tsk, TIF_DEBUGCTLMSR); - tsk->thread.debugctlmsr = 0; + clear_tsk_thread_flag(tsk, TIF_BLOCKSTEP); /* Store the virtualized DR6 value */ tsk->thread.debugreg6 = dr6; @@ -582,55 +582,67 @@ dotraplinkage void __kprobes do_debug(struct pt_regs *regs, long error_code) return; } -#ifdef CONFIG_X86_64 -static int kernel_math_error(struct pt_regs *regs, const char *str, int trapnr) -{ - if (fixup_exception(regs)) - return 1; - - notify_die(DIE_GPF, str, regs, 0, trapnr, SIGFPE); - /* Illegal floating point operation in the kernel */ - current->thread.trap_no = trapnr; - die(str, regs, 0); - return 0; -} -#endif - /* * Note that we play around with the 'TS' bit in an attempt to get * the correct behaviour even in the presence of the asynchronous * IRQ13 behaviour */ -void math_error(void __user *ip) +void math_error(struct pt_regs *regs, int error_code, int trapnr) { - struct task_struct *task; + struct task_struct *task = current; siginfo_t info; - unsigned short cwd, swd, err; + unsigned short err; + char *str = (trapnr == 16) ? "fpu exception" : "simd exception"; + + if (notify_die(DIE_TRAP, str, regs, error_code, trapnr, SIGFPE) == NOTIFY_STOP) + return; + conditional_sti(regs); + + if (!user_mode_vm(regs)) + { + if (!fixup_exception(regs)) { + task->thread.error_code = error_code; + task->thread.trap_no = trapnr; + die(str, regs, error_code); + } + return; + } /* * Save the info for the exception handler and clear the error. */ - task = current; save_init_fpu(task); - task->thread.trap_no = 16; - task->thread.error_code = 0; + task->thread.trap_no = trapnr; + task->thread.error_code = error_code; info.si_signo = SIGFPE; info.si_errno = 0; - info.si_addr = ip; - /* - * (~cwd & swd) will mask out exceptions that are not set to unmasked - * status. 0x3f is the exception bits in these regs, 0x200 is the - * C1 reg you need in case of a stack fault, 0x040 is the stack - * fault bit. We should only be taking one exception at a time, - * so if this combination doesn't produce any single exception, - * then we have a bad program that isn't synchronizing its FPU usage - * and it will suffer the consequences since we won't be able to - * fully reproduce the context of the exception - */ - cwd = get_fpu_cwd(task); - swd = get_fpu_swd(task); + info.si_addr = (void __user *)regs->ip; + if (trapnr == 16) { + unsigned short cwd, swd; + /* + * (~cwd & swd) will mask out exceptions that are not set to unmasked + * status. 0x3f is the exception bits in these regs, 0x200 is the + * C1 reg you need in case of a stack fault, 0x040 is the stack + * fault bit. We should only be taking one exception at a time, + * so if this combination doesn't produce any single exception, + * then we have a bad program that isn't synchronizing its FPU usage + * and it will suffer the consequences since we won't be able to + * fully reproduce the context of the exception + */ + cwd = get_fpu_cwd(task); + swd = get_fpu_swd(task); - err = swd & ~cwd; + err = swd & ~cwd; + } else { + /* + * The SIMD FPU exceptions are handled a little differently, as there + * is only a single status/control register. Thus, to determine which + * unmasked exception was caught we must mask the exception mask bits + * at 0x1f80, and then use these to mask the exception bits at 0x3f. + */ + unsigned short mxcsr = get_fpu_mxcsr(task); + err = ~(mxcsr >> 7) & mxcsr; + } if (err & 0x001) { /* Invalid op */ /* @@ -659,97 +671,17 @@ void math_error(void __user *ip) dotraplinkage void do_coprocessor_error(struct pt_regs *regs, long error_code) { - conditional_sti(regs); - #ifdef CONFIG_X86_32 ignore_fpu_irq = 1; -#else - if (!user_mode(regs) && - kernel_math_error(regs, "kernel x87 math error", 16)) - return; #endif - math_error((void __user *)regs->ip); -} - -static void simd_math_error(void __user *ip) -{ - struct task_struct *task; - siginfo_t info; - unsigned short mxcsr; - - /* - * Save the info for the exception handler and clear the error. - */ - task = current; - save_init_fpu(task); - task->thread.trap_no = 19; - task->thread.error_code = 0; - info.si_signo = SIGFPE; - info.si_errno = 0; - info.si_code = __SI_FAULT; - info.si_addr = ip; - /* - * The SIMD FPU exceptions are handled a little differently, as there - * is only a single status/control register. Thus, to determine which - * unmasked exception was caught we must mask the exception mask bits - * at 0x1f80, and then use these to mask the exception bits at 0x3f. - */ - mxcsr = get_fpu_mxcsr(task); - switch (~((mxcsr & 0x1f80) >> 7) & (mxcsr & 0x3f)) { - case 0x000: - default: - break; - case 0x001: /* Invalid Op */ - info.si_code = FPE_FLTINV; - break; - case 0x002: /* Denormalize */ - case 0x010: /* Underflow */ - info.si_code = FPE_FLTUND; - break; - case 0x004: /* Zero Divide */ - info.si_code = FPE_FLTDIV; - break; - case 0x008: /* Overflow */ - info.si_code = FPE_FLTOVF; - break; - case 0x020: /* Precision */ - info.si_code = FPE_FLTRES; - break; - } - force_sig_info(SIGFPE, &info, task); + math_error(regs, error_code, 16); } dotraplinkage void do_simd_coprocessor_error(struct pt_regs *regs, long error_code) { - conditional_sti(regs); - -#ifdef CONFIG_X86_32 - if (cpu_has_xmm) { - /* Handle SIMD FPU exceptions on PIII+ processors. */ - ignore_fpu_irq = 1; - simd_math_error((void __user *)regs->ip); - return; - } - /* - * Handle strange cache flush from user space exception - * in all other cases. This is undocumented behaviour. - */ - if (regs->flags & X86_VM_MASK) { - handle_vm86_fault((struct kernel_vm86_regs *)regs, error_code); - return; - } - current->thread.trap_no = 19; - current->thread.error_code = error_code; - die_if_kernel("cache flush denied", regs, error_code); - force_sig(SIGSEGV, current); -#else - if (!user_mode(regs) && - kernel_math_error(regs, "kernel simd math error", 19)) - return; - simd_math_error((void __user *)regs->ip); -#endif + math_error(regs, error_code, 19); } dotraplinkage void @@ -876,6 +808,16 @@ dotraplinkage void do_iret_error(struct pt_regs *regs, long error_code) } #endif +/* Set of traps needed for early debugging. */ +void __init early_trap_init(void) +{ + set_intr_gate_ist(1, &debug, DEBUG_STACK); + /* int3 can be called from all */ + set_system_intr_gate_ist(3, &int3, DEBUG_STACK); + set_intr_gate(14, &page_fault); + load_idt(&idt_descr); +} + void __init trap_init(void) { int i; @@ -889,10 +831,7 @@ void __init trap_init(void) #endif set_intr_gate(0, ÷_error); - set_intr_gate_ist(1, &debug, DEBUG_STACK); set_intr_gate_ist(2, &nmi, NMI_STACK); - /* int3 can be called from all */ - set_system_intr_gate_ist(3, &int3, DEBUG_STACK); /* int4 can be called from all */ set_system_intr_gate(4, &overflow); set_intr_gate(5, &bounds); @@ -908,7 +847,6 @@ void __init trap_init(void) set_intr_gate(11, &segment_not_present); set_intr_gate_ist(12, &stack_segment, STACKFAULT_STACK); set_intr_gate(13, &general_protection); - set_intr_gate(14, &page_fault); set_intr_gate(15, &spurious_interrupt_bug); set_intr_gate(16, &coprocessor_error); set_intr_gate(17, &alignment_check); diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index 597683aa5ba..9faf91ae184 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -50,7 +50,7 @@ u64 native_sched_clock(void) * unstable. We do this because unlike Time Of Day, * the scheduler clock tolerates small errors and it's * very important for it to be as fast as the platform - * can achive it. ) + * can achieve it. ) */ if (unlikely(tsc_disabled)) { /* No locking but a rare wrong value is not a big deal: */ @@ -740,7 +740,7 @@ static cycle_t __vsyscall_fn vread_tsc(void) } #endif -static void resume_tsc(void) +static void resume_tsc(struct clocksource *cs) { clocksource_tsc.cycle_last = 0; } @@ -806,7 +806,7 @@ static void __init check_system_tsc_reliable(void) unsigned long res_low, res_high; rdmsr_safe(MSR_GEODE_BUSCONT_CONF0, &res_low, &res_high); - /* Geode_LX - the OLPC CPU has a possibly a very reliable TSC */ + /* Geode_LX - the OLPC CPU has a very reliable TSC */ if (res_low & RTSC_SUSP) tsc_clocksource_reliable = 1; #endif diff --git a/arch/x86/kernel/uv_irq.c b/arch/x86/kernel/uv_irq.c index ece73d8e324..1132129db79 100644 --- a/arch/x86/kernel/uv_irq.c +++ b/arch/x86/kernel/uv_irq.c @@ -10,6 +10,7 @@ #include <linux/module.h> #include <linux/rbtree.h> +#include <linux/slab.h> #include <linux/irq.h> #include <asm/apic.h> @@ -43,7 +44,7 @@ static void uv_ack_apic(unsigned int irq) ack_APIC_irq(); } -struct irq_chip uv_irq_chip = { +static struct irq_chip uv_irq_chip = { .name = "UV-CORE", .startup = uv_noop_ret, .shutdown = uv_noop, @@ -140,7 +141,7 @@ int uv_irq_2_mmr_info(int irq, unsigned long *offset, int *pnode) */ static int arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade, - unsigned long mmr_offset, int restrict) + unsigned long mmr_offset, int limit) { const struct cpumask *eligible_cpu = cpumask_of(cpu); struct irq_desc *desc = irq_to_desc(irq); @@ -159,7 +160,7 @@ arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade, if (err != 0) return err; - if (restrict == UV_AFFINITY_CPU) + if (limit == UV_AFFINITY_CPU) desc->status |= IRQ_NO_BALANCING; else desc->status |= IRQ_MOVE_PCNTXT; @@ -213,7 +214,7 @@ static int uv_set_irq_affinity(unsigned int irq, const struct cpumask *mask) unsigned long mmr_value; struct uv_IO_APIC_route_entry *entry; unsigned long mmr_offset; - unsigned mmr_pnode; + int mmr_pnode; if (set_desc_affinity(desc, mask, &dest)) return -1; @@ -247,7 +248,7 @@ static int uv_set_irq_affinity(unsigned int irq, const struct cpumask *mask) * interrupt is raised. */ int uv_setup_irq(char *irq_name, int cpu, int mmr_blade, - unsigned long mmr_offset, int restrict) + unsigned long mmr_offset, int limit) { int irq, ret; @@ -257,7 +258,7 @@ int uv_setup_irq(char *irq_name, int cpu, int mmr_blade, return -EBUSY; ret = arch_enable_uv_irq(irq_name, irq, cpu, mmr_blade, mmr_offset, - restrict); + limit); if (ret == irq) uv_set_irq_2_mmr_info(irq, mmr_offset, mmr_blade); else diff --git a/arch/x86/kernel/uv_sysfs.c b/arch/x86/kernel/uv_sysfs.c index 36afb98675a..309c70fb775 100644 --- a/arch/x86/kernel/uv_sysfs.c +++ b/arch/x86/kernel/uv_sysfs.c @@ -54,19 +54,19 @@ static int __init sgi_uv_sysfs_init(void) if (!sgi_uv_kobj) sgi_uv_kobj = kobject_create_and_add("sgi_uv", firmware_kobj); if (!sgi_uv_kobj) { - printk(KERN_WARNING "kobject_create_and_add sgi_uv failed \n"); + printk(KERN_WARNING "kobject_create_and_add sgi_uv failed\n"); return -EINVAL; } ret = sysfs_create_file(sgi_uv_kobj, &partition_id_attr.attr); if (ret) { - printk(KERN_WARNING "sysfs_create_file partition_id failed \n"); + printk(KERN_WARNING "sysfs_create_file partition_id failed\n"); return ret; } ret = sysfs_create_file(sgi_uv_kobj, &coherence_id_attr.attr); if (ret) { - printk(KERN_WARNING "sysfs_create_file coherence_id failed \n"); + printk(KERN_WARNING "sysfs_create_file coherence_id failed\n"); return ret; } diff --git a/arch/x86/kernel/uv_time.c b/arch/x86/kernel/uv_time.c index 3c84aa001c1..56e421bc379 100644 --- a/arch/x86/kernel/uv_time.c +++ b/arch/x86/kernel/uv_time.c @@ -19,6 +19,7 @@ * Copyright (c) Dimitri Sivanich */ #include <linux/clockchips.h> +#include <linux/slab.h> #include <asm/uv/uv_mmrs.h> #include <asm/uv/uv_hub.h> @@ -282,10 +283,21 @@ static int uv_rtc_unset_timer(int cpu, int force) /* * Read the RTC. + * + * Starting with HUB rev 2.0, the UV RTC register is replicated across all + * cachelines of it's own page. This allows faster simultaneous reads + * from a given socket. */ static cycle_t uv_read_rtc(struct clocksource *cs) { - return (cycle_t)uv_read_local_mmr(UVH_RTC); + unsigned long offset; + + if (uv_get_min_hub_revision_id() == 1) + offset = 0; + else + offset = (uv_blade_processor_id() * L1_CACHE_BYTES) % PAGE_SIZE; + + return (cycle_t)uv_read_local_mmr(UVH_RTC | offset); } /* diff --git a/arch/x86/kernel/visws_quirks.c b/arch/x86/kernel/visws_quirks.c index 34a279a7471..e680ea52db9 100644 --- a/arch/x86/kernel/visws_quirks.c +++ b/arch/x86/kernel/visws_quirks.c @@ -49,11 +49,6 @@ extern int no_broadcast; char visws_board_type = -1; char visws_board_rev = -1; -int is_visws_box(void) -{ - return visws_board_type >= 0; -} - static void __init visws_time_init(void) { printk(KERN_INFO "Starting Cobalt Timer system clock\n"); @@ -242,6 +237,8 @@ void __init visws_early_detect(void) x86_init.irqs.pre_vector_init = visws_pre_intr_init; x86_init.irqs.trap_init = visws_trap_init; x86_init.timers.timer_init = visws_time_init; + x86_init.pci.init = pci_visws_init; + x86_init.pci.init_irq = x86_init_noop; /* * Install reboot quirks: @@ -508,7 +505,7 @@ static struct irq_chip cobalt_irq_type = { */ static unsigned int startup_piix4_master_irq(unsigned int irq) { - init_8259A(0); + legacy_pic->init(0); return startup_cobalt_irq(irq); } @@ -532,9 +529,6 @@ static struct irq_chip piix4_master_irq_type = { static struct irq_chip piix4_virtual_irq_type = { .name = "PIIX4-virtual", - .shutdown = disable_8259A_irq, - .enable = enable_8259A_irq, - .disable = disable_8259A_irq, }; @@ -559,7 +553,7 @@ static irqreturn_t piix4_master_intr(int irq, void *dev_id) struct irq_desc *desc; unsigned long flags; - spin_lock_irqsave(&i8259A_lock, flags); + raw_spin_lock_irqsave(&i8259A_lock, flags); /* Find out what's interrupting in the PIIX4 master 8259 */ outb(0x0c, 0x20); /* OCW3 Poll command */ @@ -596,7 +590,7 @@ static irqreturn_t piix4_master_intr(int irq, void *dev_id) outb(0x60 + realirq, 0x20); } - spin_unlock_irqrestore(&i8259A_lock, flags); + raw_spin_unlock_irqrestore(&i8259A_lock, flags); desc = irq_to_desc(realirq); @@ -609,12 +603,12 @@ static irqreturn_t piix4_master_intr(int irq, void *dev_id) handle_IRQ_event(realirq, desc->action); if (!(desc->status & IRQ_DISABLED)) - enable_8259A_irq(realirq); + legacy_pic->chip->unmask(realirq); return IRQ_HANDLED; out_unlock: - spin_unlock_irqrestore(&i8259A_lock, flags); + raw_spin_unlock_irqrestore(&i8259A_lock, flags); return IRQ_NONE; } @@ -628,6 +622,12 @@ static struct irqaction cascade_action = { .name = "cascade", }; +static inline void set_piix4_virtual_irq_type(void) +{ + piix4_virtual_irq_type.shutdown = i8259A_chip.mask; + piix4_virtual_irq_type.enable = i8259A_chip.unmask; + piix4_virtual_irq_type.disable = i8259A_chip.mask; +} void init_VISWS_APIC_irqs(void) { @@ -653,6 +653,7 @@ void init_VISWS_APIC_irqs(void) desc->chip = &piix4_master_irq_type; } else if (i < CO_IRQ_APIC0) { + set_piix4_virtual_irq_type(); desc->chip = &piix4_virtual_irq_type; } else if (IS_CO_APIC(i)) { diff --git a/arch/x86/kernel/vmi_32.c b/arch/x86/kernel/vmi_32.c index d430e4c3019..ce9fbacb752 100644 --- a/arch/x86/kernel/vmi_32.c +++ b/arch/x86/kernel/vmi_32.c @@ -28,11 +28,13 @@ #include <linux/mm.h> #include <linux/highmem.h> #include <linux/sched.h> +#include <linux/gfp.h> #include <asm/vmi.h> #include <asm/io.h> #include <asm/fixmap.h> #include <asm/apicdef.h> #include <asm/apic.h> +#include <asm/pgalloc.h> #include <asm/processor.h> #include <asm/timer.h> #include <asm/vmi_time.h> @@ -266,30 +268,6 @@ static void vmi_nop(void) { } -#ifdef CONFIG_HIGHPTE -static void *vmi_kmap_atomic_pte(struct page *page, enum km_type type) -{ - void *va = kmap_atomic(page, type); - - /* - * Internally, the VMI ROM must map virtual addresses to physical - * addresses for processing MMU updates. By the time MMU updates - * are issued, this information is typically already lost. - * Fortunately, the VMI provides a cache of mapping slots for active - * page tables. - * - * We use slot zero for the linear mapping of physical memory, and - * in HIGHPTE kernels, slot 1 and 2 for KM_PTE0 and KM_PTE1. - * - * args: SLOT VA COUNT PFN - */ - BUG_ON(type != KM_PTE0 && type != KM_PTE1); - vmi_ops.set_linear_mapping((type - KM_PTE0)+1, va, 1, page_to_pfn(page)); - - return va; -} -#endif - static void vmi_allocate_pte(struct mm_struct *mm, unsigned long pfn) { vmi_ops.allocate_page(pfn, VMI_PAGE_L1, 0, 0, 0); @@ -640,6 +618,12 @@ static inline int __init activate_vmi(void) u64 reloc; const struct vmi_relocation_info *rel = (struct vmi_relocation_info *)&reloc; + /* + * Prevent page tables from being allocated in highmem, even if + * CONFIG_HIGHPTE is enabled. + */ + __userpte_alloc_gfp &= ~__GFP_HIGHMEM; + if (call_vrom_func(vmi_rom, vmi_init) != 0) { printk(KERN_ERR "VMI ROM failed to initialize!"); return 0; @@ -778,10 +762,6 @@ static inline int __init activate_vmi(void) /* Set linear is needed in all cases */ vmi_ops.set_linear_mapping = vmi_get_function(VMI_CALL_SetLinearMapping); -#ifdef CONFIG_HIGHPTE - if (vmi_ops.set_linear_mapping) - pv_mmu_ops.kmap_atomic_pte = vmi_kmap_atomic_pte; -#endif /* * These MUST always be patched. Don't support indirect jumps diff --git a/arch/x86/kernel/vmiclock_32.c b/arch/x86/kernel/vmiclock_32.c index 74c92bb194d..5e1ff66ecd7 100644 --- a/arch/x86/kernel/vmiclock_32.c +++ b/arch/x86/kernel/vmiclock_32.c @@ -79,11 +79,7 @@ unsigned long vmi_tsc_khz(void) static inline unsigned int vmi_get_timer_vector(void) { -#ifdef CONFIG_X86_IO_APIC - return FIRST_DEVICE_VECTOR; -#else - return FIRST_EXTERNAL_VECTOR; -#endif + return IRQ0_VECTOR; } /** vmi clockchip */ @@ -171,7 +167,7 @@ static int vmi_timer_next_event(unsigned long delta, { /* Unfortunately, set_next_event interface only passes relative * expiry, but we want absolute expiry. It'd be better if were - * were passed an aboslute expiry, since a bunch of time may + * were passed an absolute expiry, since a bunch of time may * have been stolen between the time the delta is computed and * when we set the alarm below. */ cycle_t now = vmi_timer_ops.get_cycle_counter(vmi_counter(VMI_ONESHOT)); diff --git a/arch/x86/kernel/vmlinux.lds.S b/arch/x86/kernel/vmlinux.lds.S index f3d77d729d5..d0bb52296fa 100644 --- a/arch/x86/kernel/vmlinux.lds.S +++ b/arch/x86/kernel/vmlinux.lds.S @@ -291,8 +291,8 @@ SECTIONS .smp_locks : AT(ADDR(.smp_locks) - LOAD_OFFSET) { __smp_locks = .; *(.smp_locks) - __smp_locks_end = .; . = ALIGN(PAGE_SIZE); + __smp_locks_end = .; } #ifdef CONFIG_X86_64 @@ -341,7 +341,7 @@ SECTIONS * Per-cpu symbols which need to be offset from __per_cpu_load * for the boot processor. */ -#define INIT_PER_CPU(x) init_per_cpu__##x = per_cpu__##x + __per_cpu_load +#define INIT_PER_CPU(x) init_per_cpu__##x = x + __per_cpu_load INIT_PER_CPU(gdt_page); INIT_PER_CPU(irq_stack_union); @@ -352,7 +352,7 @@ INIT_PER_CPU(irq_stack_union); "kernel image bigger than KERNEL_IMAGE_SIZE"); #ifdef CONFIG_SMP -. = ASSERT((per_cpu__irq_stack_union == 0), +. = ASSERT((irq_stack_union == 0), "irq_stack_union is not at start of per-cpu area"); #endif diff --git a/arch/x86/kernel/vsyscall_64.c b/arch/x86/kernel/vsyscall_64.c index 9055e5872ff..1c0c6ab9c60 100644 --- a/arch/x86/kernel/vsyscall_64.c +++ b/arch/x86/kernel/vsyscall_64.c @@ -301,7 +301,8 @@ static int __init vsyscall_init(void) register_sysctl_table(kernel_root_table2); #endif on_each_cpu(cpu_vsyscall_init, NULL, 1); - hotcpu_notifier(cpu_vsyscall_notifier, 0); + /* notifier priority > KVM */ + hotcpu_notifier(cpu_vsyscall_notifier, 30); return 0; } diff --git a/arch/x86/kernel/x8664_ksyms_64.c b/arch/x86/kernel/x8664_ksyms_64.c index 619f7f88b8c..1b950d151e5 100644 --- a/arch/x86/kernel/x8664_ksyms_64.c +++ b/arch/x86/kernel/x8664_ksyms_64.c @@ -26,7 +26,8 @@ EXPORT_SYMBOL(__put_user_2); EXPORT_SYMBOL(__put_user_4); EXPORT_SYMBOL(__put_user_8); -EXPORT_SYMBOL(copy_user_generic); +EXPORT_SYMBOL(copy_user_generic_string); +EXPORT_SYMBOL(copy_user_generic_unrolled); EXPORT_SYMBOL(__copy_user_nocache); EXPORT_SYMBOL(_copy_from_user); EXPORT_SYMBOL(_copy_to_user); @@ -53,7 +54,6 @@ EXPORT_SYMBOL(memcpy); EXPORT_SYMBOL(__memcpy); EXPORT_SYMBOL(empty_zero_page); -EXPORT_SYMBOL(init_level4_pgt); #ifndef CONFIG_PARAVIRT EXPORT_SYMBOL(native_load_gs_index); #endif diff --git a/arch/x86/kernel/x86_init.c b/arch/x86/kernel/x86_init.c index ccd179dec36..61a1e8c7e19 100644 --- a/arch/x86/kernel/x86_init.c +++ b/arch/x86/kernel/x86_init.c @@ -4,9 +4,11 @@ * For licencing details see kernel-base/COPYING */ #include <linux/init.h> +#include <linux/ioport.h> #include <asm/bios_ebda.h> #include <asm/paravirt.h> +#include <asm/pci_x86.h> #include <asm/mpspec.h> #include <asm/setup.h> #include <asm/apic.h> @@ -70,16 +72,25 @@ struct x86_init_ops x86_init __initdata = { .iommu = { .iommu_init = iommu_init_noop, }, + + .pci = { + .init = x86_default_pci_init, + .init_irq = x86_default_pci_init_irq, + .fixup_irqs = x86_default_pci_fixup_irqs, + }, }; struct x86_cpuinit_ops x86_cpuinit __cpuinitdata = { .setup_percpu_clockev = setup_secondary_APIC_clock, }; +static void default_nmi_init(void) { }; + struct x86_platform_ops x86_platform = { .calibrate_tsc = native_calibrate_tsc, .get_wallclock = mach_get_cmos_time, .set_wallclock = mach_set_rtc_mmss, .iommu_shutdown = iommu_shutdown_noop, .is_untracked_pat_range = is_ISA_range, + .nmi_init = default_nmi_init }; diff --git a/arch/x86/kernel/xsave.c b/arch/x86/kernel/xsave.c index c5ee17e8c6d..37e68fc5e24 100644 --- a/arch/x86/kernel/xsave.c +++ b/arch/x86/kernel/xsave.c @@ -99,7 +99,7 @@ int save_i387_xstate(void __user *buf) if (err) return err; - if (task_thread_info(tsk)->status & TS_XSAVE) + if (use_xsave()) err = xsave_user(buf); else err = fxsave_user(buf); @@ -109,14 +109,14 @@ int save_i387_xstate(void __user *buf) task_thread_info(tsk)->status &= ~TS_USEDFPU; stts(); } else { - if (__copy_to_user(buf, &tsk->thread.xstate->fxsave, + if (__copy_to_user(buf, &tsk->thread.fpu.state->fxsave, xstate_size)) return -1; } clear_used_math(); /* trigger finit */ - if (task_thread_info(tsk)->status & TS_XSAVE) { + if (use_xsave()) { struct _fpstate __user *fx = buf; struct _xstate __user *x = buf; u64 xstate_bv; @@ -225,7 +225,7 @@ int restore_i387_xstate(void __user *buf) clts(); task_thread_info(current)->status |= TS_USEDFPU; } - if (task_thread_info(tsk)->status & TS_XSAVE) + if (use_xsave()) err = restore_user_xstate(buf); else err = fxrstor_checking((__force struct i387_fxsave_struct *) @@ -337,6 +337,7 @@ void __ref xsave_cntxt_init(void) cpuid_count(0xd, 0, &eax, &ebx, &ecx, &edx); xstate_size = ebx; + update_regset_xstate_info(xstate_size, pcntxt_mask); prepare_fx_sw_frame(); setup_xstate_init(); |