From c0eaa4536f08b98fbcfa7fce5b7b0de1bebcb0e1 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sun, 12 Apr 2009 20:47:40 +0400 Subject: x86: apic - introduce imcr_ helpers Impact: cleanup Distinguish port writting magic into helpers with comments. Signed-off-by: Cyrill Gorcunov LKML-Reference: <20090412165058.535921550@openvz.org> Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic/apic.c | 29 +++++++++++++++++++++++++---- 1 file changed, 25 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 098ec84b8c0..c3be10f5773 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -98,6 +98,29 @@ early_param("lapic", parse_lapic); /* Local APIC was disabled by the BIOS and enabled by the kernel */ static int enabled_via_apicbase; +/* + * Handle interrupt mode configuration register (IMCR). + * This register controls whether the interrupt signals + * that reach the BSP come from the master PIC or from the + * local APIC. Before entering Symmetric I/O Mode, either + * the BIOS or the operating system must switch out of + * PIC Mode by changing the IMCR. + */ +static inline imcr_pic_to_apic(void) +{ + /* select IMCR register */ + outb(0x70, 0x22); + /* NMI and 8259 INTR go through APIC */ + outb(0x01, 0x23); +} + +static inline imcr_apic_to_pic(void) +{ + /* select IMCR register */ + outb(0x70, 0x22); + /* NMI and 8259 INTR go directly to BSP */ + outb(0x00, 0x23); +} #endif #ifdef CONFIG_X86_64 @@ -1727,8 +1750,7 @@ void __init connect_bsp_APIC(void) */ apic_printk(APIC_VERBOSE, "leaving PIC mode, " "enabling APIC mode.\n"); - outb(0x70, 0x22); - outb(0x01, 0x23); + imcr_pic_to_apic(); } #endif if (apic->enable_apic_mode) @@ -1756,8 +1778,7 @@ void disconnect_bsp_APIC(int virt_wire_setup) */ apic_printk(APIC_VERBOSE, "disabling APIC mode, " "entering PIC mode.\n"); - outb(0x70, 0x22); - outb(0x00, 0x23); + imcr_apic_to_pic(); return; } #endif -- cgit v1.2.3-70-g09d2 From 08306ce61d6848e6fbf74fa4cc693c3fb29e943f Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sun, 12 Apr 2009 20:47:41 +0400 Subject: x86: apic - introduce dummy apic operations Impact: refactor, speed up and robustize code In case if apic was disabled by kernel option or by hardware limits we can use dummy operations in apic->write to simplify the ack_APIC_irq() code. At the lame time the patch fixes the missed EOI in do_IRQ function (which has place if kernel is compiled as X86-32 and interrupt without handler happens where apic was not asked to be disabled via kernel option). Note that native_apic_write_dummy() consists of WARN_ON_ONCE to catch any buggy writes on enabled APICs. Could be removed after some time of testing. Signed-off-by: Cyrill Gorcunov LKML-Reference: <20090412165058.724788431@openvz.org> Signed-off-by: Ingo Molnar --- arch/x86/include/asm/apic.h | 3 ++- arch/x86/kernel/apic/apic.c | 24 ++++++++++++++++++++++++ arch/x86/kernel/irq.c | 10 ++-------- 3 files changed, 28 insertions(+), 9 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h index 42f2f837742..2bd5a463fd1 100644 --- a/arch/x86/include/asm/apic.h +++ b/arch/x86/include/asm/apic.h @@ -212,6 +212,7 @@ static inline void ack_x2APIC_irq(void) } #endif +extern void apic_disable(void); extern int lapic_get_maxlvt(void); extern void clear_local_APIC(void); extern void connect_bsp_APIC(void); @@ -252,7 +253,7 @@ static inline void lapic_shutdown(void) { } #define local_apic_timer_c2_ok 1 static inline void init_apic_mappings(void) { } static inline void disable_local_APIC(void) { } - +static inline void apic_disable(void) { } #endif /* !CONFIG_X86_LOCAL_APIC */ #ifdef CONFIG_X86_64 diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index c3be10f5773..9b849d4957d 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -232,6 +232,24 @@ static int modern_apic(void) return lapic_get_version() >= 0x14; } +/* + * bare function to substitute write operation + * and it's _that_ fast :) + */ +void native_apic_write_dummy(u32 reg, u32 v) +{ + WARN_ON_ONCE((cpu_has_apic || !disable_apic)); +} + +/* + * right after this call apic->write doesn't do anything + * note that there is no restore operation it works one way + */ +void apic_disable(void) +{ + apic->write = native_apic_write_dummy; +} + void native_apic_wait_icr_idle(void) { while (apic_read(APIC_ICR) & APIC_ICR_BUSY) @@ -1582,6 +1600,12 @@ void __init init_apic_mappings(void) */ if (boot_cpu_physical_apicid == -1U) boot_cpu_physical_apicid = read_apic_id(); + + /* lets check if we may to NOP'ify apic operations */ + if (!cpu_has_apic) { + pr_info("APIC: disable apic facility\n"); + apic_disable(); + } } /* diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c index 6603492e8b7..fd57bf35d0f 100644 --- a/arch/x86/kernel/irq.c +++ b/arch/x86/kernel/irq.c @@ -27,7 +27,6 @@ void ack_bad_irq(unsigned int irq) if (printk_ratelimit()) pr_err("unexpected IRQ trap at vector %02x\n", irq); -#ifdef CONFIG_X86_LOCAL_APIC /* * Currently unexpected vectors happen only on SMP and APIC. * We _must_ ack these because every local APIC has only N @@ -37,9 +36,7 @@ void ack_bad_irq(unsigned int irq) * completely. * But only ack when the APIC is enabled -AK */ - if (cpu_has_apic) - ack_APIC_irq(); -#endif + ack_APIC_irq(); } #define irq_stats(x) (&per_cpu(irq_stat, x)) @@ -214,10 +211,7 @@ unsigned int __irq_entry do_IRQ(struct pt_regs *regs) irq = __get_cpu_var(vector_irq)[vector]; if (!handle_irq(irq, regs)) { -#ifdef CONFIG_X86_64 - if (!disable_apic) - ack_APIC_irq(); -#endif + ack_APIC_irq(); if (printk_ratelimit()) pr_emerg("%s: %d.%d No irq handler for vector (irq %d)\n", -- cgit v1.2.3-70-g09d2 From 5cda395f4a262788d8ed79ac8a26a2b821e5f751 Mon Sep 17 00:00:00 2001 From: Alexander van Heukelum Date: Mon, 13 Apr 2009 17:39:24 +0200 Subject: x86: fix function definitions after: x86: apic - introduce imcr_ helpers The patch "introduce imcr_ helpers" introduced good comments, but also a few new compile warnings. This fixes the function definitions to have a 'void' return type. Signed-off-by: Alexander van Heukelum Acked-by: Cyrill Gorcunov LKML-Reference: <20090413153924.GA20287@mailshack.com> Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic/apic.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 9b849d4957d..4b48ff9163c 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -106,7 +106,7 @@ static int enabled_via_apicbase; * the BIOS or the operating system must switch out of * PIC Mode by changing the IMCR. */ -static inline imcr_pic_to_apic(void) +static inline void imcr_pic_to_apic(void) { /* select IMCR register */ outb(0x70, 0x22); @@ -114,7 +114,7 @@ static inline imcr_pic_to_apic(void) outb(0x01, 0x23); } -static inline imcr_apic_to_pic(void) +static inline void imcr_apic_to_pic(void) { /* select IMCR register */ outb(0x70, 0x22); -- cgit v1.2.3-70-g09d2 From 937582382c71b75b29fbb92615629494e1a05ac0 Mon Sep 17 00:00:00 2001 From: Weidong Han Date: Fri, 17 Apr 2009 16:42:14 +0800 Subject: x86, intr-remap: enable interrupt remapping early Currently, when x2apic is not enabled, interrupt remapping will be enabled in init_dmars(), where it is too late to remap ioapic interrupts, that is, ioapic interrupts are really in compatibility mode, not remappable mode. This patch always enables interrupt remapping before ioapic setup, it guarantees all interrupts will be remapped when interrupt remapping is enabled. Thus it doesn't need to set the compatibility interrupt bit. [ Impact: refactor intr-remap init sequence, enable fuller remap mode ] Signed-off-by: Suresh Siddha Signed-off-by: Weidong Han Acked-by: David Woodhouse Cc: iommu@lists.linux-foundation.org Cc: allen.m.kay@intel.com Cc: fenghua.yu@intel.com LKML-Reference: <1239957736-6161-4-git-send-email-weidong.han@intel.com> Signed-off-by: Ingo Molnar --- arch/x86/include/asm/apic.h | 7 ++-- arch/x86/kernel/apic/apic.c | 76 +++++++++++++++++++++----------------------- drivers/pci/intel-iommu.c | 9 ------ drivers/pci/intr_remapping.c | 28 ++++++++-------- include/linux/dmar.h | 1 + 5 files changed, 54 insertions(+), 67 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h index d4cb7e590c0..fbdd65446c7 100644 --- a/arch/x86/include/asm/apic.h +++ b/arch/x86/include/asm/apic.h @@ -169,7 +169,6 @@ static inline u64 native_x2apic_icr_read(void) extern int x2apic, x2apic_phys; extern void check_x2apic(void); extern void enable_x2apic(void); -extern void enable_IR_x2apic(void); extern void x2apic_icr_write(u32 low, u32 id); static inline int x2apic_enabled(void) { @@ -190,18 +189,18 @@ static inline void check_x2apic(void) static inline void enable_x2apic(void) { } -static inline void enable_IR_x2apic(void) -{ -} static inline int x2apic_enabled(void) { return 0; } #define x2apic 0 +#define x2apic_preenabled 0 #endif +extern void enable_IR_x2apic(void); + extern int get_physical_broadcast(void); extern void apic_disable(void); diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 83e47febcc8..0cf1eea750c 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -141,6 +141,8 @@ static int x2apic_preenabled; static int disable_x2apic; static __init int setup_nox2apic(char *str) { + if (x2apic_enabled()) + panic("Bios already enabled x2apic, can't enforce nox2apic"); disable_x2apic = 1; setup_clear_cpu_cap(X86_FEATURE_X2APIC); return 0; @@ -1345,6 +1347,7 @@ void enable_x2apic(void) wrmsr(MSR_IA32_APICBASE, msr | X2APIC_ENABLE, 0); } } +#endif /* CONFIG_X86_X2APIC */ void __init enable_IR_x2apic(void) { @@ -1353,32 +1356,21 @@ void __init enable_IR_x2apic(void) unsigned long flags; struct IO_APIC_route_entry **ioapic_entries = NULL; - if (!cpu_has_x2apic) - return; - - if (!x2apic_preenabled && disable_x2apic) { - pr_info("Skipped enabling x2apic and Interrupt-remapping " - "because of nox2apic\n"); - return; + ret = dmar_table_init(); + if (ret) { + pr_debug("dmar_table_init() failed with %d:\n", ret); + goto ir_failed; } - if (x2apic_preenabled && disable_x2apic) - panic("Bios already enabled x2apic, can't enforce nox2apic"); - - if (!x2apic_preenabled && skip_ioapic_setup) { - pr_info("Skipped enabling x2apic and Interrupt-remapping " - "because of skipping io-apic setup\n"); - return; + if (!intr_remapping_supported()) { + pr_debug("intr-remapping not supported\n"); + goto ir_failed; } - ret = dmar_table_init(); - if (ret) { - pr_info("dmar_table_init() failed with %d:\n", ret); - if (x2apic_preenabled) - panic("x2apic enabled by bios. But IR enabling failed"); - else - pr_info("Not enabling x2apic,Intr-remapping\n"); + if (!x2apic_preenabled && skip_ioapic_setup) { + pr_info("Skipped enabling intr-remap because of skipping " + "io-apic setup\n"); return; } @@ -1398,20 +1390,25 @@ void __init enable_IR_x2apic(void) mask_IO_APIC_setup(ioapic_entries); mask_8259A(); - ret = enable_intr_remapping(EIM_32BIT_APIC_ID); - - if (ret && x2apic_preenabled) { - local_irq_restore(flags); - panic("x2apic enabled by bios. But IR enabling failed"); - } +#ifdef CONFIG_X86_X2APIC + if (cpu_has_x2apic) + ret = enable_intr_remapping(EIM_32BIT_APIC_ID); + else +#endif + ret = enable_intr_remapping(EIM_8BIT_APIC_ID); if (ret) goto end_restore; - if (!x2apic) { + pr_info("Enabled Interrupt-remapping\n"); + +#ifdef CONFIG_X86_X2APIC + if (cpu_has_x2apic && !x2apic) { x2apic = 1; enable_x2apic(); + pr_info("Enabled x2apic\n"); } +#endif end_restore: if (ret) @@ -1426,30 +1423,29 @@ end_restore: local_irq_restore(flags); end: - if (!ret) { - if (!x2apic_preenabled) - pr_info("Enabled x2apic and interrupt-remapping\n"); - else - pr_info("Enabled Interrupt-remapping\n"); - } else - pr_err("Failed to enable Interrupt-remapping and x2apic\n"); if (ioapic_entries) free_ioapic_entries(ioapic_entries); + + if (!ret) + return; + +ir_failed: + if (x2apic_preenabled) + panic("x2apic enabled by bios. But IR enabling failed"); + else if (cpu_has_x2apic) + pr_info("Not enabling x2apic,Intr-remapping\n"); #else if (!cpu_has_x2apic) return; if (x2apic_preenabled) panic("x2apic enabled prior OS handover," - " enable CONFIG_INTR_REMAP"); - - pr_info("Enable CONFIG_INTR_REMAP for enabling intr-remapping " - " and x2apic\n"); + " enable CONFIG_X86_X2APIC, CONFIG_INTR_REMAP"); #endif return; } -#endif /* CONFIG_X86_X2APIC */ + #ifdef CONFIG_X86_64 /* diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c index 001b328adf8..9ce8f0764be 100644 --- a/drivers/pci/intel-iommu.c +++ b/drivers/pci/intel-iommu.c @@ -1968,15 +1968,6 @@ static int __init init_dmars(void) } } -#ifdef CONFIG_INTR_REMAP - if (!intr_remapping_enabled) { - ret = enable_intr_remapping(0); - if (ret) - printk(KERN_ERR - "IOMMU: enable interrupt remapping failed\n"); - } -#endif - /* * For each rmrr * for each dev attached to rmrr diff --git a/drivers/pci/intr_remapping.c b/drivers/pci/intr_remapping.c index f5e0ea724a6..5c2142656e9 100644 --- a/drivers/pci/intr_remapping.c +++ b/drivers/pci/intr_remapping.c @@ -423,20 +423,6 @@ static void iommu_set_intr_remapping(struct intel_iommu *iommu, int mode) readl, (sts & DMA_GSTS_IRTPS), sts); spin_unlock_irqrestore(&iommu->register_lock, flags); - if (mode == 0) { - spin_lock_irqsave(&iommu->register_lock, flags); - - /* enable comaptiblity format interrupt pass through */ - cmd = iommu->gcmd | DMA_GCMD_CFI; - iommu->gcmd |= DMA_GCMD_CFI; - writel(cmd, iommu->reg + DMAR_GCMD_REG); - - IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG, - readl, (sts & DMA_GSTS_CFIS), sts); - - spin_unlock_irqrestore(&iommu->register_lock, flags); - } - /* * global invalidation of interrupt entry cache before enabling * interrupt-remapping. @@ -516,6 +502,20 @@ end: spin_unlock_irqrestore(&iommu->register_lock, flags); } +int __init intr_remapping_supported(void) +{ + struct dmar_drhd_unit *drhd; + + for_each_drhd_unit(drhd) { + struct intel_iommu *iommu = drhd->iommu; + + if (!ecap_ir_support(iommu->ecap)) + return 0; + } + + return 1; +} + int __init enable_intr_remapping(int eim) { struct dmar_drhd_unit *drhd; diff --git a/include/linux/dmar.h b/include/linux/dmar.h index e397dc342cd..06f592a7f73 100644 --- a/include/linux/dmar.h +++ b/include/linux/dmar.h @@ -108,6 +108,7 @@ struct irte { }; #ifdef CONFIG_INTR_REMAP extern int intr_remapping_enabled; +extern int intr_remapping_supported(void); extern int enable_intr_remapping(int); extern void disable_intr_remapping(void); extern int reenable_intr_remapping(int); -- cgit v1.2.3-70-g09d2 From 9a2755c3569e4db92bd9b1daadeddb4045b0cccd Mon Sep 17 00:00:00 2001 From: Weidong Han Date: Fri, 17 Apr 2009 16:42:16 +0800 Subject: x86, intr-remap: fix x2apic/intr-remap resume Interrupt remapping was decoupled from x2apic. Shouldn't check x2apic before resume interrupt remapping. Otherwise, interrupt remapping won't be resumed when x2apic is not enabled. [ Impact: fix potential intr-remap resume hang on !x2apic ] Signed-off-by: Suresh Siddha Signed-off-by: Weidong Han Acked-by: David Woodhouse Cc: iommu@lists.linux-foundation.org Cc: allen.m.kay@intel.com Cc: fenghua.yu@intel.com LKML-Reference: <1239957736-6161-6-git-send-email-weidong.han@intel.com> Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic/apic.c | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 0cf1eea750c..7b41a32339e 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -2032,7 +2032,7 @@ static int lapic_resume(struct sys_device *dev) return 0; local_irq_save(flags); - if (x2apic) { + if (intr_remapping_enabled) { ioapic_entries = alloc_ioapic_entries(); if (!ioapic_entries) { WARN(1, "Alloc ioapic_entries in lapic resume failed."); @@ -2048,8 +2048,10 @@ static int lapic_resume(struct sys_device *dev) mask_IO_APIC_setup(ioapic_entries); mask_8259A(); - enable_x2apic(); } + + if (x2apic) + enable_x2apic(); #else if (!apic_pm_state.active) return 0; @@ -2097,10 +2099,12 @@ static int lapic_resume(struct sys_device *dev) apic_read(APIC_ESR); #ifdef CONFIG_INTR_REMAP - if (intr_remapping_enabled) - reenable_intr_remapping(EIM_32BIT_APIC_ID); + if (intr_remapping_enabled) { + if (x2apic) + reenable_intr_remapping(EIM_32BIT_APIC_ID); + else + reenable_intr_remapping(EIM_8BIT_APIC_ID); - if (x2apic) { unmask_8259A(); restore_IO_APIC_setup(ioapic_entries); free_ioapic_entries(ioapic_entries); @@ -2109,7 +2113,6 @@ static int lapic_resume(struct sys_device *dev) local_irq_restore(flags); - return 0; } -- cgit v1.2.3-70-g09d2 From fc1edaf9e7cc4d4696f83dee495b8f158d01c4eb Mon Sep 17 00:00:00 2001 From: Suresh Siddha Date: Mon, 20 Apr 2009 13:02:27 -0700 Subject: x86: x2apic, IR: Clean up X86_X2APIC and INTR_REMAP config checks Add x2apic_supported() to clean up CONFIG_X86_X2APIC checks. Fix CONFIG_INTR_REMAP checks. [ Impact: cleanup ] Signed-off-by: Suresh Siddha Cc: dwmw2@infradead.org Cc: Suresh Siddha Cc: Weidong Han LKML-Reference: <20090420200450.128993000@linux-os.sc.intel.com> Signed-off-by: Ingo Molnar --- arch/x86/include/asm/apic.h | 10 ++++---- arch/x86/include/asm/io_apic.h | 2 -- arch/x86/include/asm/irq_remapping.h | 2 +- arch/x86/kernel/apic/apic.c | 49 +++++++++--------------------------- arch/x86/kernel/apic/io_apic.c | 2 -- arch/x86/kernel/apic/probe_64.c | 2 +- include/linux/dmar.h | 2 ++ 7 files changed, 21 insertions(+), 48 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h index fbdd65446c7..3738438a91f 100644 --- a/arch/x86/include/asm/apic.h +++ b/arch/x86/include/asm/apic.h @@ -107,8 +107,7 @@ extern u32 native_safe_apic_wait_icr_idle(void); extern void native_apic_icr_write(u32 low, u32 id); extern u64 native_apic_icr_read(void); -#define EIM_8BIT_APIC_ID 0 -#define EIM_32BIT_APIC_ID 1 +extern int x2apic_mode; #ifdef CONFIG_X86_X2APIC /* @@ -166,7 +165,7 @@ static inline u64 native_x2apic_icr_read(void) return val; } -extern int x2apic, x2apic_phys; +extern int x2apic_phys; extern void check_x2apic(void); extern void enable_x2apic(void); extern void x2apic_icr_write(u32 low, u32 id); @@ -182,6 +181,8 @@ static inline int x2apic_enabled(void) return 1; return 0; } + +#define x2apic_supported() (cpu_has_x2apic) #else static inline void check_x2apic(void) { @@ -194,9 +195,8 @@ static inline int x2apic_enabled(void) return 0; } -#define x2apic 0 #define x2apic_preenabled 0 - +#define x2apic_supported() 0 #endif extern void enable_IR_x2apic(void); diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h index 9d826e43601..34eaa37f7ad 100644 --- a/arch/x86/include/asm/io_apic.h +++ b/arch/x86/include/asm/io_apic.h @@ -161,7 +161,6 @@ extern int io_apic_set_pci_routing(int ioapic, int pin, int irq, extern int (*ioapic_renumber_irq)(int ioapic, int irq); extern void ioapic_init_mappings(void); -#ifdef CONFIG_X86_64 extern struct IO_APIC_route_entry **alloc_ioapic_entries(void); extern void free_ioapic_entries(struct IO_APIC_route_entry **ioapic_entries); extern int save_IO_APIC_setup(struct IO_APIC_route_entry **ioapic_entries); @@ -169,7 +168,6 @@ extern void mask_IO_APIC_setup(struct IO_APIC_route_entry **ioapic_entries); extern int restore_IO_APIC_setup(struct IO_APIC_route_entry **ioapic_entries); extern void reinit_intr_remapped_IO_APIC(int intr_remapping, struct IO_APIC_route_entry **ioapic_entries); -#endif extern void probe_nr_irqs_gsi(void); diff --git a/arch/x86/include/asm/irq_remapping.h b/arch/x86/include/asm/irq_remapping.h index 0396760fccb..f275e224450 100644 --- a/arch/x86/include/asm/irq_remapping.h +++ b/arch/x86/include/asm/irq_remapping.h @@ -1,6 +1,6 @@ #ifndef _ASM_X86_IRQ_REMAPPING_H #define _ASM_X86_IRQ_REMAPPING_H -#define IRTE_DEST(dest) ((x2apic) ? dest : dest << 8) +#define IRTE_DEST(dest) ((x2apic_mode) ? dest : dest << 8) #endif /* _ASM_X86_IRQ_REMAPPING_H */ diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 7b41a32339e..2b30e520dce 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -134,8 +134,8 @@ static __init int setup_apicpmtimer(char *s) __setup("apicpmtimer", setup_apicpmtimer); #endif +int x2apic_mode; #ifdef CONFIG_X86_X2APIC -int x2apic; /* x2apic enabled before OS handover */ static int x2apic_preenabled; static int disable_x2apic; @@ -858,7 +858,7 @@ void clear_local_APIC(void) u32 v; /* APIC hasn't been mapped yet */ - if (!x2apic && !apic_phys) + if (!x2apic_mode && !apic_phys) return; maxlvt = lapic_get_maxlvt(); @@ -1330,7 +1330,7 @@ void check_x2apic(void) { if (x2apic_enabled()) { pr_info("x2apic enabled by BIOS, switching to x2apic ops\n"); - x2apic_preenabled = x2apic = 1; + x2apic_preenabled = x2apic_mode = 1; } } @@ -1338,7 +1338,7 @@ void enable_x2apic(void) { int msr, msr2; - if (!x2apic) + if (!x2apic_mode) return; rdmsr(MSR_IA32_APICBASE, msr, msr2); @@ -1390,25 +1390,17 @@ void __init enable_IR_x2apic(void) mask_IO_APIC_setup(ioapic_entries); mask_8259A(); -#ifdef CONFIG_X86_X2APIC - if (cpu_has_x2apic) - ret = enable_intr_remapping(EIM_32BIT_APIC_ID); - else -#endif - ret = enable_intr_remapping(EIM_8BIT_APIC_ID); - + ret = enable_intr_remapping(x2apic_supported()); if (ret) goto end_restore; pr_info("Enabled Interrupt-remapping\n"); -#ifdef CONFIG_X86_X2APIC - if (cpu_has_x2apic && !x2apic) { - x2apic = 1; + if (x2apic_supported() && !x2apic_mode) { + x2apic_mode = 1; enable_x2apic(); pr_info("Enabled x2apic\n"); } -#endif end_restore: if (ret) @@ -1576,7 +1568,7 @@ void __init early_init_lapic_mapping(void) */ void __init init_apic_mappings(void) { - if (x2apic) { + if (x2apic_mode) { boot_cpu_physical_apicid = read_apic_id(); return; } @@ -2010,10 +2002,10 @@ static int lapic_suspend(struct sys_device *dev, pm_message_t state) local_irq_save(flags); disable_local_APIC(); -#ifdef CONFIG_INTR_REMAP + if (intr_remapping_enabled) disable_intr_remapping(); -#endif + local_irq_restore(flags); return 0; } @@ -2023,8 +2015,6 @@ static int lapic_resume(struct sys_device *dev) unsigned int l, h; unsigned long flags; int maxlvt; - -#ifdef CONFIG_INTR_REMAP int ret; struct IO_APIC_route_entry **ioapic_entries = NULL; @@ -2050,17 +2040,8 @@ static int lapic_resume(struct sys_device *dev) mask_8259A(); } - if (x2apic) + if (x2apic_mode) enable_x2apic(); -#else - if (!apic_pm_state.active) - return 0; - - local_irq_save(flags); - if (x2apic) - enable_x2apic(); -#endif - else { /* * Make sure the APICBASE points to the right address @@ -2098,18 +2079,12 @@ static int lapic_resume(struct sys_device *dev) apic_write(APIC_ESR, 0); apic_read(APIC_ESR); -#ifdef CONFIG_INTR_REMAP if (intr_remapping_enabled) { - if (x2apic) - reenable_intr_remapping(EIM_32BIT_APIC_ID); - else - reenable_intr_remapping(EIM_8BIT_APIC_ID); - + reenable_intr_remapping(x2apic_mode); unmask_8259A(); restore_IO_APIC_setup(ioapic_entries); free_ioapic_entries(ioapic_entries); } -#endif local_irq_restore(flags); diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c index ea22a86e3cd..3a45d2ec974 100644 --- a/arch/x86/kernel/apic/io_apic.c +++ b/arch/x86/kernel/apic/io_apic.c @@ -736,7 +736,6 @@ static int __init ioapic_pirq_setup(char *str) __setup("pirq=", ioapic_pirq_setup); #endif /* CONFIG_X86_32 */ -#ifdef CONFIG_INTR_REMAP struct IO_APIC_route_entry **alloc_ioapic_entries(void) { int apic; @@ -857,7 +856,6 @@ void free_ioapic_entries(struct IO_APIC_route_entry **ioapic_entries) kfree(ioapic_entries); } -#endif /* * Find the IRQ entry number of a certain pin. diff --git a/arch/x86/kernel/apic/probe_64.c b/arch/x86/kernel/apic/probe_64.c index 1783652bb0e..bc3e880f9b8 100644 --- a/arch/x86/kernel/apic/probe_64.c +++ b/arch/x86/kernel/apic/probe_64.c @@ -50,7 +50,7 @@ static struct apic *apic_probe[] __initdata = { void __init default_setup_apic_routing(void) { #ifdef CONFIG_X86_X2APIC - if (x2apic && (apic != &apic_x2apic_phys && + if (x2apic_mode && (apic != &apic_x2apic_phys && #ifdef CONFIG_X86_UV apic != &apic_x2apic_uv_x && #endif diff --git a/include/linux/dmar.h b/include/linux/dmar.h index 06f592a7f73..10ff5c49882 100644 --- a/include/linux/dmar.h +++ b/include/linux/dmar.h @@ -158,6 +158,8 @@ static inline struct intel_iommu *map_ioapic_to_ir(int apic) } #define irq_remapped(irq) (0) #define enable_intr_remapping(mode) (-1) +#define disable_intr_remapping() (0) +#define reenable_intr_remapping(mode) (0) #define intr_remapping_enabled (0) #endif -- cgit v1.2.3-70-g09d2 From 39d83a5d684a457046aa2a6dac60f105966e78e9 Mon Sep 17 00:00:00 2001 From: Suresh Siddha Date: Mon, 20 Apr 2009 13:02:29 -0700 Subject: x86: x2apic, IR: Clean up panic() with nox2apic boot option Instead of panic() ignore the "nox2apic" boot option when BIOS has already enabled x2apic prior to OS handover. [ Impact: printk warning instead of panic() when BIOS has enabled x2apic already ] Signed-off-by: Suresh Siddha Cc: dwmw2@infradead.org Cc: Weidong Han LKML-Reference: <20090420200450.425091000@linux-os.sc.intel.com> Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic/apic.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 2b30e520dce..d32f5589f1d 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -141,8 +141,12 @@ static int x2apic_preenabled; static int disable_x2apic; static __init int setup_nox2apic(char *str) { - if (x2apic_enabled()) - panic("Bios already enabled x2apic, can't enforce nox2apic"); + if (x2apic_enabled()) { + pr_warning("Bios already enabled x2apic, " + "can't enforce nox2apic"); + return 0; + } + disable_x2apic = 1; setup_clear_cpu_cap(X86_FEATURE_X2APIC); return 0; -- cgit v1.2.3-70-g09d2 From ff166cb57a17124af75714a9c11f448f56f1a4a3 Mon Sep 17 00:00:00 2001 From: Suresh Siddha Date: Mon, 20 Apr 2009 13:02:30 -0700 Subject: x86: x2apic, IR: remove reinit_intr_remapped_IO_APIC() When interrupt-remapping is enabled, we are relying on setup_IO_APIC_irqs() to configure remapped entries in the IO-APIC, which comes little bit later after enabling interrupt-remapping. Meanwhile, restoration of old io-apic entries after enabling interrupt-remapping will not make the interrupts through io-apic functional anyway. So remove the unnecessary reinit_intr_remapped_IO_APIC() step. The longer story: When interrupt-remapping is enabled, IO-APIC entries need to be setup in the re-mappable format (pointing to interrupt-remapping table entries setup by the OS). This remapping configuration is happening in the same place where we traditionally configure IO-APIC (i.e., in setup_IO_APIC_irqs()). So when we enable interrupt-remapping successfully, there is no need to restore old io-apic RTE entries before we actually do a complete configuration shortly in setup_IO_APIC_irqs(). Old IO-APIC RTE's may be in traditional format (non re-mappable) or in re-mappable format pointing to interrupt-remapping table entries setup by BIOS. Restoring both of these will not make IO-APIC functional. We have to rely on setup_IO_APIC_irqs() for proper configuration by OS. So I am removing this unnecessary and broken step. [ Impact: remove unnecessary/broken IO-APIC setup step ] Signed-off-by: Suresh Siddha Acked-by: Weidong Han Cc: dwmw2@infradead.org LKML-Reference: <20090420200450.552359000@linux-os.sc.intel.com> Signed-off-by: Ingo Molnar --- arch/x86/include/asm/io_apic.h | 2 -- arch/x86/kernel/apic/apic.c | 2 -- arch/x86/kernel/apic/io_apic.c | 14 -------------- 3 files changed, 18 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h index 34eaa37f7ad..1cf145039ee 100644 --- a/arch/x86/include/asm/io_apic.h +++ b/arch/x86/include/asm/io_apic.h @@ -166,8 +166,6 @@ extern void free_ioapic_entries(struct IO_APIC_route_entry **ioapic_entries); extern int save_IO_APIC_setup(struct IO_APIC_route_entry **ioapic_entries); extern void mask_IO_APIC_setup(struct IO_APIC_route_entry **ioapic_entries); extern int restore_IO_APIC_setup(struct IO_APIC_route_entry **ioapic_entries); -extern void reinit_intr_remapped_IO_APIC(int intr_remapping, - struct IO_APIC_route_entry **ioapic_entries); extern void probe_nr_irqs_gsi(void); diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index d32f5589f1d..1386dbec552 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -1412,8 +1412,6 @@ end_restore: * IR enabling failed */ restore_IO_APIC_setup(ioapic_entries); - else - reinit_intr_remapped_IO_APIC(x2apic_preenabled, ioapic_entries); unmask_8259A(); local_irq_restore(flags); diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c index 4baa9cbd630..8aef5f9d947 100644 --- a/arch/x86/kernel/apic/io_apic.c +++ b/arch/x86/kernel/apic/io_apic.c @@ -833,20 +833,6 @@ int restore_IO_APIC_setup(struct IO_APIC_route_entry **ioapic_entries) return 0; } -void reinit_intr_remapped_IO_APIC(int intr_remapping, - struct IO_APIC_route_entry **ioapic_entries) - -{ - /* - * for now plain restore of previous settings. - * TBD: In the case of OS enabling interrupt-remapping, - * IO-APIC RTE's need to be setup to point to interrupt-remapping - * table entries. for now, do a plain restore, and wait for - * the setup_IO_APIC_irqs() to do proper initialization. - */ - restore_IO_APIC_setup(ioapic_entries); -} - void free_ioapic_entries(struct IO_APIC_route_entry **ioapic_entries) { int apic; -- cgit v1.2.3-70-g09d2 From e0e42142bab96404de535cceb85d6533d5ad7942 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sun, 26 Apr 2009 23:39:38 -0700 Subject: x86: Use dmi check in apic_is_clustered() on 64-bit to mark the TSC unstable We will have systems with 2 and more sockets 8cores/2thread, but we treat them as multi chassis - while they could have a stable TSC domain. Use DMI check instead. [ Impact: do not turn possibly stable TSCs off incorrectly ] Signed-off-by: Yinghai Lu Cc: Ravikiran Thirumalai LKML-Reference: <49F5532A.5000802@kernel.org> Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic/apic.c | 86 +++++++++++++++++++++++++++++++-------------- 1 file changed, 59 insertions(+), 27 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 1386dbec552..28f747d61d7 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -2138,31 +2138,14 @@ static void apic_pm_activate(void) { } #endif /* CONFIG_PM */ #ifdef CONFIG_X86_64 -/* - * apic_is_clustered_box() -- Check if we can expect good TSC - * - * Thus far, the major user of this is IBM's Summit2 series: - * - * Clustered boxes may have unsynced TSC problems if they are - * multi-chassis. Use available data to take a good guess. - * If in doubt, go HPET. - */ -__cpuinit int apic_is_clustered_box(void) + +static int __cpuinit apic_cluster_num(void) { int i, clusters, zeros; unsigned id; u16 *bios_cpu_apicid; DECLARE_BITMAP(clustermap, NUM_APIC_CLUSTERS); - /* - * there is not this kind of box with AMD CPU yet. - * Some AMD box with quadcore cpu and 8 sockets apicid - * will be [4, 0x23] or [8, 0x27] could be thought to - * vsmp box still need checking... - */ - if ((boot_cpu_data.x86_vendor == X86_VENDOR_AMD) && !is_vsmp_box()) - return 0; - bios_cpu_apicid = early_per_cpu_ptr(x86_bios_cpu_apicid); bitmap_zero(clustermap, NUM_APIC_CLUSTERS); @@ -2198,18 +2181,67 @@ __cpuinit int apic_is_clustered_box(void) ++zeros; } - /* ScaleMP vSMPowered boxes have one cluster per board and TSCs are - * not guaranteed to be synced between boards - */ - if (is_vsmp_box() && clusters > 1) + return clusters; +} + +static int __cpuinitdata multi_checked; +static int __cpuinitdata multi; + +static int __cpuinit set_multi(const struct dmi_system_id *d) +{ + if (multi) + return 0; + printk(KERN_INFO "APIC: %s detected, Multi Chassis\n", d->ident); + multi = 1; + return 0; +} + +static const __cpuinitconst struct dmi_system_id multi_dmi_table[] = { + { + .callback = set_multi, + .ident = "IBM System Summit2", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "IBM"), + DMI_MATCH(DMI_PRODUCT_NAME, "Summit2"), + }, + }, + {} +}; + +static void __cpuinit dmi_check_multi(void) +{ + if (multi_checked) + return; + + dmi_check_system(multi_dmi_table); + multi_checked = 1; +} + +/* + * apic_is_clustered_box() -- Check if we can expect good TSC + * + * Thus far, the major user of this is IBM's Summit2 series: + * Clustered boxes may have unsynced TSC problems if they are + * multi-chassis. + * Use DMI to check them + */ +__cpuinit int apic_is_clustered_box(void) +{ + dmi_check_multi(); + if (multi) return 1; + if (!is_vsmp_box()) + return 0; + /* - * If clusters > 2, then should be multi-chassis. - * May have to revisit this when multi-core + hyperthreaded CPUs come - * out, but AFAIK this will work even for them. + * ScaleMP vSMPowered boxes have one cluster per board and TSCs are + * not guaranteed to be synced between boards */ - return (clusters > 2); + if (apic_cluster_num() > 1) + return 1; + + return 0; } #endif -- cgit v1.2.3-70-g09d2 From 6f0aced639d346e5f54eea9fcb2784b633493d09 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Fri, 1 May 2009 23:54:25 +0400 Subject: x86, apic: use pr_ macro Replace recenly appeared printk with pr_ macro (the file already use a lot of them). [ Impact: cleanup ] Signed-off-by: Cyrill Gorcunov LKML-Reference: <20090501195425.GB4633@lenovo> Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic/apic.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 28f747d61d7..e258bedce7c 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -2191,7 +2191,7 @@ static int __cpuinit set_multi(const struct dmi_system_id *d) { if (multi) return 0; - printk(KERN_INFO "APIC: %s detected, Multi Chassis\n", d->ident); + pr_info("APIC: %s detected, Multi Chassis\n", d->ident); multi = 1; return 0; } -- cgit v1.2.3-70-g09d2 From 4401da6111ac58f94234417427d06a72c4048c74 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sat, 2 May 2009 10:40:57 -0700 Subject: x86: read apic ID in the !acpi_lapic case Ed found that on 32-bit, boot_cpu_physical_apicid is not read right, when the mptable is broken. Interestingly, actually three paths use/set it: 1. acpi: at that time that is already read from reg 2. mptable: only read from mptable 3. no madt, and no mptable, that use default apic id 0 for 64-bit, -1 for 32-bit so we could read the apic id for the 2/3 path. We trust the hardware register more than we trust a BIOS data structure (the mptable). We can also avoid the double set_fixmap() when acpi_lapic is used, and also need to move cpu_has_apic earlier and call apic_disable(). Also when need to update the apic id, we'd better read and set the apic version as well - so that quirks are applied precisely. v2: make path 3 with 64bit, use -1 as apic id, so could read it later. v3: fix whitespace problem pointed out by Ed Swierk [ Impact: get correct apic id for bsp other than acpi path ] Reported-by: Ed Swierk Signed-off-by: Yinghai Lu Acked-by: Cyrill Gorcunov LKML-Reference: <49FC85A9.2070702@kernel.org> [ v4: sanity-check in the ACPI case too ] Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic/apic.c | 31 +++++++++++++++++++++---------- 1 file changed, 21 insertions(+), 10 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index e258bedce7c..1ee966f4ae9 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -1456,7 +1456,6 @@ static int __init detect_init_APIC(void) } mp_lapic_addr = APIC_DEFAULT_PHYS_BASE; - boot_cpu_physical_apicid = 0; return 0; } #else @@ -1570,6 +1569,8 @@ void __init early_init_lapic_mapping(void) */ void __init init_apic_mappings(void) { + unsigned int new_apicid; + if (x2apic_mode) { boot_cpu_physical_apicid = read_apic_id(); return; @@ -1586,21 +1587,31 @@ void __init init_apic_mappings(void) } else apic_phys = mp_lapic_addr; - set_fixmap_nocache(FIX_APIC_BASE, apic_phys); + /* lets check if we may NOP'ify apic operations */ + if (!cpu_has_apic) { + pr_info("APIC: disable apic facility\n"); + apic_disable(); + return; + } + + /* + * acpi lapic path already maps that address in + * acpi_register_lapic_address() + */ + if (!acpi_lapic) + set_fixmap_nocache(FIX_APIC_BASE, apic_phys); + apic_printk(APIC_VERBOSE, "mapped APIC to %08lx (%08lx)\n", APIC_BASE, apic_phys); - /* * Fetch the APIC ID of the BSP in case we have a * default configuration (or the MP table is broken). */ - if (boot_cpu_physical_apicid == -1U) - boot_cpu_physical_apicid = read_apic_id(); - - /* lets check if we may to NOP'ify apic operations */ - if (!cpu_has_apic) { - pr_info("APIC: disable apic facility\n"); - apic_disable(); + new_apicid = read_apic_id(); + if (boot_cpu_physical_apicid != new_apicid) { + boot_cpu_physical_apicid = new_apicid; + apic_version[new_apicid] = + GET_APIC_VERSION(apic_read(APIC_LVR)); } } -- cgit v1.2.3-70-g09d2 From 97a52714658cd959a3cfa35c5b6f489859f0204b Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Fri, 8 May 2009 18:23:50 +0200 Subject: x86: display extended apic registers with print_local_APIC and cpu_debug code Both print_local_APIC (used when apic=debug kernel param is set) and cpu_debug code missed support for some extended APIC registers that I'd like to see. This adds support to show: - extended APIC feature register - extended APIC control register - extended LVT registers [ Impact: print more debug info ] Signed-off-by: Andreas Herrmann Cc: Jaswinder Singh Rajput Cc: Cyrill Gorcunov LKML-Reference: <20090508162350.GO29045@alberich.amd.com> Signed-off-by: Ingo Molnar --- arch/x86/include/asm/apicdef.h | 8 ++++---- arch/x86/kernel/apic/apic.c | 2 +- arch/x86/kernel/apic/io_apic.c | 14 +++++++++++++- arch/x86/kernel/cpu/cpu_debug.c | 14 +++++++++++++- 4 files changed, 31 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/include/asm/apicdef.h b/arch/x86/include/asm/apicdef.h index bc9514fb3b1..7ddb36ab933 100644 --- a/arch/x86/include/asm/apicdef.h +++ b/arch/x86/include/asm/apicdef.h @@ -22,6 +22,7 @@ # define APIC_INTEGRATED(x) (1) #endif #define APIC_XAPIC(x) ((x) >= 0x14) +#define APIC_EXT_SPACE(x) ((x) & 0x80000000) #define APIC_TASKPRI 0x80 #define APIC_TPRI_MASK 0xFFu #define APIC_ARBPRI 0x90 @@ -116,7 +117,9 @@ #define APIC_TDR_DIV_32 0x8 #define APIC_TDR_DIV_64 0x9 #define APIC_TDR_DIV_128 0xA -#define APIC_EILVT0 0x500 +#define APIC_EFEAT 0x400 +#define APIC_ECTRL 0x410 +#define APIC_EILVTn(n) (0x500 + 0x10 * n) #define APIC_EILVT_NR_AMD_K8 1 /* # of extended interrupts */ #define APIC_EILVT_NR_AMD_10H 4 #define APIC_EILVT_LVTOFF(x) (((x) >> 4) & 0xF) @@ -125,9 +128,6 @@ #define APIC_EILVT_MSG_NMI 0x4 #define APIC_EILVT_MSG_EXT 0x7 #define APIC_EILVT_MASKED (1 << 16) -#define APIC_EILVT1 0x510 -#define APIC_EILVT2 0x520 -#define APIC_EILVT3 0x530 #define APIC_BASE (fix_to_virt(FIX_APIC_BASE)) #define APIC_BASE_MSR 0x800 diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 1ee966f4ae9..0e6543fafb5 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -395,7 +395,7 @@ static void __setup_APIC_LVTT(unsigned int clocks, int oneshot, int irqen) static void setup_APIC_eilvt(u8 lvt_off, u8 vector, u8 msg_type, u8 mask) { - unsigned long reg = (lvt_off << 4) + APIC_EILVT0; + unsigned long reg = (lvt_off << 4) + APIC_EILVTn(0); unsigned int v = (mask << 16) | (msg_type << 8) | vector; apic_write(reg, v); diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c index 2afe145d277..65b824c9c4f 100644 --- a/arch/x86/kernel/apic/io_apic.c +++ b/arch/x86/kernel/apic/io_apic.c @@ -1739,7 +1739,7 @@ __apicdebuginit(void) print_APIC_bitfield(int base) __apicdebuginit(void) print_local_APIC(void *dummy) { - unsigned int v, ver, maxlvt; + unsigned int i, v, ver, maxlvt; u64 icr; if (apic_verbosity == APIC_QUIET) @@ -1827,6 +1827,18 @@ __apicdebuginit(void) print_local_APIC(void *dummy) printk(KERN_DEBUG "... APIC TMCCT: %08x\n", v); v = apic_read(APIC_TDCR); printk(KERN_DEBUG "... APIC TDCR: %08x\n", v); + + if (boot_cpu_has(X86_FEATURE_EXTAPIC)) { + v = apic_read(APIC_EFEAT); + maxlvt = (v >> 16) & 0xff; + printk(KERN_DEBUG "... APIC EFEAT: %08x\n", v); + v = apic_read(APIC_ECTRL); + printk(KERN_DEBUG "... APIC ECTRL: %08x\n", v); + for (i = 0; i < maxlvt; i++) { + v = apic_read(APIC_EILVTn(i)); + printk(KERN_DEBUG "... APIC EILVT%d: %08x\n", i, v); + } + } printk("\n"); } diff --git a/arch/x86/kernel/cpu/cpu_debug.c b/arch/x86/kernel/cpu/cpu_debug.c index 46e29ab96c6..2fc4f6bb9ca 100644 --- a/arch/x86/kernel/cpu/cpu_debug.c +++ b/arch/x86/kernel/cpu/cpu_debug.c @@ -588,8 +588,20 @@ static void print_apic(void *arg) seq_printf(seq, " TMICT\t\t: %08x\n", apic_read(APIC_TMICT)); seq_printf(seq, " TMCCT\t\t: %08x\n", apic_read(APIC_TMCCT)); seq_printf(seq, " TDCR\t\t: %08x\n", apic_read(APIC_TDCR)); -#endif /* CONFIG_X86_LOCAL_APIC */ + if (boot_cpu_has(X86_FEATURE_EXTAPIC)) { + unsigned int i, v, maxeilvt; + + v = apic_read(APIC_EFEAT); + maxeilvt = (v >> 16) & 0xff; + seq_printf(seq, " EFEAT\t\t: %08x\n", v); + seq_printf(seq, " ECTRL\t\t: %08x\n", apic_read(APIC_ECTRL)); + for (i = 0; i < maxeilvt; i++) { + v = apic_read(APIC_EILVTn(i)); + seq_printf(seq, " EILVT%d\t\t: %08x\n", i, v); + } + } +#endif /* CONFIG_X86_LOCAL_APIC */ seq_printf(seq, "\n MSR\t:\n"); } -- cgit v1.2.3-70-g09d2 From cec6be6d1069d697beb490bbb40a290d5ff554a2 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Mon, 11 May 2009 17:41:40 +0400 Subject: x86: apic: Fixmap apic address even if apic disabled In case if apic were disabled by boot option we still need read_apic operation. So fixmap a fake apic area if needed. [ Impact: fix boot crash ] Signed-off-by: Cyrill Gorcunov Cc: yinghai@kernel.org Cc: eswierk@aristanetworks.com LKML-Reference: <20090511134140.GH4624@lenovo> Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic/apic.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 0e6543fafb5..07cffc1214c 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -1587,13 +1587,6 @@ void __init init_apic_mappings(void) } else apic_phys = mp_lapic_addr; - /* lets check if we may NOP'ify apic operations */ - if (!cpu_has_apic) { - pr_info("APIC: disable apic facility\n"); - apic_disable(); - return; - } - /* * acpi lapic path already maps that address in * acpi_register_lapic_address() @@ -1602,7 +1595,15 @@ void __init init_apic_mappings(void) set_fixmap_nocache(FIX_APIC_BASE, apic_phys); apic_printk(APIC_VERBOSE, "mapped APIC to %08lx (%08lx)\n", - APIC_BASE, apic_phys); + APIC_BASE, apic_phys); + + /* lets check if we may NOP'ify apic operations */ + if (!cpu_has_apic) { + pr_info("APIC: disable apic facility\n"); + apic_disable(); + return; + } + /* * Fetch the APIC ID of the BSP in case we have a * default configuration (or the MP table is broken). -- cgit v1.2.3-70-g09d2 From 4797f6b021a3fa399942245d07a1feb30df81bb8 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Sat, 2 May 2009 10:40:57 -0700 Subject: x86: read apic ID in the !acpi_lapic case Ed found that on 32-bit, boot_cpu_physical_apicid is not read right, when the mptable is broken. Interestingly, actually three paths use/set it: 1. acpi: at that time that is already read from reg 2. mptable: only read from mptable 3. no madt, and no mptable, that use default apic id 0 for 64-bit, -1 for 32-bit so we could read the apic id for the 2/3 path. We trust the hardware register more than we trust a BIOS data structure (the mptable). We can also avoid the double set_fixmap() when acpi_lapic is used, and also need to move cpu_has_apic earlier and call apic_disable(). Also when need to update the apic id, we'd better read and set the apic version as well - so that quirks are applied precisely. v2: make path 3 with 64bit, use -1 as apic id, so could read it later. v3: fix whitespace problem pointed out by Ed Swierk v5: fix boot crash [ Impact: get correct apic id for bsp other than acpi path ] Reported-by: Ed Swierk Signed-off-by: Yinghai Lu Acked-by: Cyrill Gorcunov LKML-Reference: <49FC85A9.2070702@kernel.org> [ v4: sanity-check in the ACPI case too ] Signed-off-by: Ingo Molnar --- arch/x86/include/asm/smp.h | 2 +- arch/x86/kernel/apic/apic.c | 46 ++++++++++++++++++++---------------------- arch/x86/kernel/apic/io_apic.c | 5 +++++ 3 files changed, 28 insertions(+), 25 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/include/asm/smp.h b/arch/x86/include/asm/smp.h index 19e0d88b966..6a84ed166ae 100644 --- a/arch/x86/include/asm/smp.h +++ b/arch/x86/include/asm/smp.h @@ -180,7 +180,7 @@ extern int safe_smp_processor_id(void); static inline int logical_smp_processor_id(void) { /* we don't want to mark this access volatile - bad code generation */ - return GET_APIC_LOGICAL_ID(*(u32 *)(APIC_BASE + APIC_LDR)); + return GET_APIC_LOGICAL_ID(apic_read(APIC_LDR)); } #endif diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 07cffc1214c..b0fd26442c4 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -242,17 +242,24 @@ static int modern_apic(void) * bare function to substitute write operation * and it's _that_ fast :) */ -void native_apic_write_dummy(u32 reg, u32 v) +static void native_apic_write_dummy(u32 reg, u32 v) { WARN_ON_ONCE((cpu_has_apic || !disable_apic)); } +static u32 native_apic_read_dummy(u32 reg) +{ + WARN_ON_ONCE((cpu_has_apic || !disable_apic)); + return 0; +} + /* - * right after this call apic->write doesn't do anything + * right after this call apic->write/read doesn't do anything * note that there is no restore operation it works one way */ void apic_disable(void) { + apic->read = native_apic_read_dummy; apic->write = native_apic_write_dummy; } @@ -1576,32 +1583,23 @@ void __init init_apic_mappings(void) return; } - /* - * If no local APIC can be found then set up a fake all - * zeroes page to simulate the local APIC and another - * one for the IO-APIC. - */ + /* If no local APIC can be found return early */ if (!smp_found_config && detect_init_APIC()) { - apic_phys = (unsigned long) alloc_bootmem_pages(PAGE_SIZE); - apic_phys = __pa(apic_phys); - } else + /* lets NOP'ify apic operations */ + pr_info("APIC: disable apic facility\n"); + apic_disable(); + } else { apic_phys = mp_lapic_addr; - /* - * acpi lapic path already maps that address in - * acpi_register_lapic_address() - */ - if (!acpi_lapic) - set_fixmap_nocache(FIX_APIC_BASE, apic_phys); - - apic_printk(APIC_VERBOSE, "mapped APIC to %08lx (%08lx)\n", - APIC_BASE, apic_phys); + /* + * acpi lapic path already maps that address in + * acpi_register_lapic_address() + */ + if (!acpi_lapic) + set_fixmap_nocache(FIX_APIC_BASE, apic_phys); - /* lets check if we may NOP'ify apic operations */ - if (!cpu_has_apic) { - pr_info("APIC: disable apic facility\n"); - apic_disable(); - return; + apic_printk(APIC_VERBOSE, "mapped APIC to %08lx (%08lx)\n", + APIC_BASE, apic_phys); } /* diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c index 1f3d3669dae..74d2b480a20 100644 --- a/arch/x86/kernel/apic/io_apic.c +++ b/arch/x86/kernel/apic/io_apic.c @@ -1878,6 +1878,11 @@ __apicdebuginit(void) print_PIC(void) __apicdebuginit(int) print_all_ICs(void) { print_PIC(); + + /* don't print out if apic is not there */ + if (!cpu_has_apic || disable_apic) + return 0; + print_all_local_APICs(); print_IO_APIC(); -- cgit v1.2.3-70-g09d2 From a63eaf34ae60bdb067a354cc8def2e8f4a01f5f4 Mon Sep 17 00:00:00 2001 From: Paul Mackerras Date: Fri, 22 May 2009 14:17:31 +1000 Subject: perf_counter: Dynamically allocate tasks' perf_counter_context struct This replaces the struct perf_counter_context in the task_struct with a pointer to a dynamically allocated perf_counter_context struct. The main reason for doing is this is to allow us to transfer a perf_counter_context from one task to another when we do lazy PMU switching in a later patch. This has a few side-benefits: the task_struct becomes a little smaller, we save some memory because only tasks that have perf_counters attached get a perf_counter_context allocated for them, and we can remove the inclusion of in sched.h, meaning that we don't end up recompiling nearly everything whenever perf_counter.h changes. The perf_counter_context structures are reference-counted and freed when the last reference is dropped. A context can have references from its task and the counters on its task. Counters can outlive the task so it is possible that a context will be freed well after its task has exited. Contexts are allocated on fork if the parent had a context, or otherwise the first time that a per-task counter is created on a task. In the latter case, we set the context pointer in the task struct locklessly using an atomic compare-and-exchange operation in case we raced with some other task in creating a context for the subject task. This also removes the task pointer from the perf_counter struct. The task pointer was not used anywhere and would make it harder to move a context from one task to another. Anything that needed to know which task a counter was attached to was already using counter->ctx->task. The __perf_counter_init_context function moves up in perf_counter.c so that it can be called from find_get_context, and now initializes the refcount, but is otherwise unchanged. We were potentially calling list_del_counter twice: once from __perf_counter_exit_task when the task exits and once from __perf_counter_remove_from_context when the counter's fd gets closed. This adds a check in list_del_counter so it doesn't do anything if the counter has already been removed from the lists. Since perf_counter_task_sched_in doesn't do anything if the task doesn't have a context, and leaves cpuctx->task_ctx = NULL, this adds code to __perf_install_in_context to set cpuctx->task_ctx if necessary, i.e. in the case where the current task adds the first counter to itself and thus creates a context for itself. This also adds similar code to __perf_counter_enable to handle a similar situation which can arise when the counters have been disabled using prctl; that also leaves cpuctx->task_ctx = NULL. [ Impact: refactor counter context management to prepare for new feature ] Signed-off-by: Paul Mackerras Acked-by: Peter Zijlstra Cc: Corey Ashford Cc: Marcelo Tosatti Cc: Arnaldo Carvalho de Melo LKML-Reference: <18966.10075.781053.231153@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic/apic.c | 1 + include/linux/init_task.h | 13 --- include/linux/perf_counter.h | 4 +- include/linux/sched.h | 6 +- kernel/exit.c | 3 +- kernel/fork.c | 1 + kernel/perf_counter.c | 218 +++++++++++++++++++++++++++---------------- 7 files changed, 145 insertions(+), 101 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index e9021a90802..b4f64402a82 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -14,6 +14,7 @@ * Mikael Pettersson : PM converted to driver model. */ +#include #include #include #include diff --git a/include/linux/init_task.h b/include/linux/init_task.h index 503afaa0afa..d87247d2641 100644 --- a/include/linux/init_task.h +++ b/include/linux/init_task.h @@ -108,18 +108,6 @@ extern struct group_info init_groups; extern struct cred init_cred; -#ifdef CONFIG_PERF_COUNTERS -# define INIT_PERF_COUNTERS(tsk) \ - .perf_counter_ctx.counter_list = \ - LIST_HEAD_INIT(tsk.perf_counter_ctx.counter_list), \ - .perf_counter_ctx.event_list = \ - LIST_HEAD_INIT(tsk.perf_counter_ctx.event_list), \ - .perf_counter_ctx.lock = \ - __SPIN_LOCK_UNLOCKED(tsk.perf_counter_ctx.lock), -#else -# define INIT_PERF_COUNTERS(tsk) -#endif - /* * INIT_TASK is used to set up the first task table, touch at * your own risk!. Base=0, limit=0x1fffff (=2MB) @@ -183,7 +171,6 @@ extern struct cred init_cred; }, \ .dirties = INIT_PROP_LOCAL_SINGLE(dirties), \ INIT_IDS \ - INIT_PERF_COUNTERS(tsk) \ INIT_TRACE_IRQFLAGS \ INIT_LOCKDEP \ INIT_FTRACE_GRAPH \ diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h index f612941ef46..07130900546 100644 --- a/include/linux/perf_counter.h +++ b/include/linux/perf_counter.h @@ -449,7 +449,6 @@ struct perf_counter { struct hw_perf_counter hw; struct perf_counter_context *ctx; - struct task_struct *task; struct file *filp; struct perf_counter *parent; @@ -498,7 +497,6 @@ struct perf_counter { * Used as a container for task counters and CPU counters as well: */ struct perf_counter_context { -#ifdef CONFIG_PERF_COUNTERS /* * Protect the states of the counters in the list, * nr_active, and the list: @@ -516,6 +514,7 @@ struct perf_counter_context { int nr_counters; int nr_active; int is_active; + atomic_t refcount; struct task_struct *task; /* @@ -523,7 +522,6 @@ struct perf_counter_context { */ u64 time; u64 timestamp; -#endif }; /** diff --git a/include/linux/sched.h b/include/linux/sched.h index ff59d123151..9714d450f41 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -71,7 +71,6 @@ struct sched_param { #include #include #include -#include #include #include #include @@ -99,6 +98,7 @@ struct robust_list_head; struct bio; struct bts_tracer; struct fs_struct; +struct perf_counter_context; /* * List of flags we want to share for kernel threads, @@ -1387,7 +1387,9 @@ struct task_struct { struct list_head pi_state_list; struct futex_pi_state *pi_state_cache; #endif - struct perf_counter_context perf_counter_ctx; +#ifdef CONFIG_PERF_COUNTERS + struct perf_counter_context *perf_counter_ctxp; +#endif #ifdef CONFIG_NUMA struct mempolicy *mempolicy; short il_next; diff --git a/kernel/exit.c b/kernel/exit.c index f9dfedd94af..99ad4063ee4 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -48,6 +48,7 @@ #include #include #include +#include #include #include @@ -159,7 +160,7 @@ static void delayed_put_task_struct(struct rcu_head *rhp) struct task_struct *tsk = container_of(rhp, struct task_struct, rcu); #ifdef CONFIG_PERF_COUNTERS - WARN_ON_ONCE(!list_empty(&tsk->perf_counter_ctx.counter_list)); + WARN_ON_ONCE(tsk->perf_counter_ctxp); #endif trace_sched_process_free(tsk); put_task_struct(tsk); diff --git a/kernel/fork.c b/kernel/fork.c index d32fef4d38e..e72a09f5355 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -63,6 +63,7 @@ #include #include #include +#include #include #include diff --git a/kernel/perf_counter.c b/kernel/perf_counter.c index 08584c16049..06ea3eae886 100644 --- a/kernel/perf_counter.c +++ b/kernel/perf_counter.c @@ -97,6 +97,17 @@ void perf_enable(void) hw_perf_enable(); } +static void get_ctx(struct perf_counter_context *ctx) +{ + atomic_inc(&ctx->refcount); +} + +static void put_ctx(struct perf_counter_context *ctx) +{ + if (atomic_dec_and_test(&ctx->refcount)) + kfree(ctx); +} + static void list_add_counter(struct perf_counter *counter, struct perf_counter_context *ctx) { @@ -118,11 +129,17 @@ list_add_counter(struct perf_counter *counter, struct perf_counter_context *ctx) ctx->nr_counters++; } +/* + * Remove a counter from the lists for its context. + * Must be called with counter->mutex and ctx->mutex held. + */ static void list_del_counter(struct perf_counter *counter, struct perf_counter_context *ctx) { struct perf_counter *sibling, *tmp; + if (list_empty(&counter->list_entry)) + return; ctx->nr_counters--; list_del_init(&counter->list_entry); @@ -216,8 +233,6 @@ static void __perf_counter_remove_from_context(void *info) counter_sched_out(counter, cpuctx, ctx); - counter->task = NULL; - list_del_counter(counter, ctx); if (!ctx->task) { @@ -279,7 +294,6 @@ retry: */ if (!list_empty(&counter->list_entry)) { list_del_counter(counter, ctx); - counter->task = NULL; } spin_unlock_irq(&ctx->lock); } @@ -568,11 +582,17 @@ static void __perf_install_in_context(void *info) * If this is a task context, we need to check whether it is * the current task context of this cpu. If not it has been * scheduled out before the smp call arrived. + * Or possibly this is the right context but it isn't + * on this cpu because it had no counters. */ - if (ctx->task && cpuctx->task_ctx != ctx) - return; + if (ctx->task && cpuctx->task_ctx != ctx) { + if (cpuctx->task_ctx || ctx->task != current) + return; + cpuctx->task_ctx = ctx; + } spin_lock_irqsave(&ctx->lock, flags); + ctx->is_active = 1; update_context_time(ctx); /* @@ -653,7 +673,6 @@ perf_install_in_context(struct perf_counter_context *ctx, return; } - counter->task = task; retry: task_oncpu_function_call(task, __perf_install_in_context, counter); @@ -693,10 +712,14 @@ static void __perf_counter_enable(void *info) * If this is a per-task counter, need to check whether this * counter's task is the current task on this cpu. */ - if (ctx->task && cpuctx->task_ctx != ctx) - return; + if (ctx->task && cpuctx->task_ctx != ctx) { + if (cpuctx->task_ctx || ctx->task != current) + return; + cpuctx->task_ctx = ctx; + } spin_lock_irqsave(&ctx->lock, flags); + ctx->is_active = 1; update_context_time(ctx); counter->prev_state = counter->state; @@ -852,10 +875,10 @@ void __perf_counter_sched_out(struct perf_counter_context *ctx, void perf_counter_task_sched_out(struct task_struct *task, int cpu) { struct perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu); - struct perf_counter_context *ctx = &task->perf_counter_ctx; + struct perf_counter_context *ctx = task->perf_counter_ctxp; struct pt_regs *regs; - if (likely(!cpuctx->task_ctx)) + if (likely(!ctx || !cpuctx->task_ctx)) return; update_context_time(ctx); @@ -871,6 +894,8 @@ static void __perf_counter_task_sched_out(struct perf_counter_context *ctx) { struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + if (!cpuctx->task_ctx) + return; __perf_counter_sched_out(ctx, cpuctx); cpuctx->task_ctx = NULL; } @@ -969,8 +994,10 @@ __perf_counter_sched_in(struct perf_counter_context *ctx, void perf_counter_task_sched_in(struct task_struct *task, int cpu) { struct perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu); - struct perf_counter_context *ctx = &task->perf_counter_ctx; + struct perf_counter_context *ctx = task->perf_counter_ctxp; + if (likely(!ctx)) + return; __perf_counter_sched_in(ctx, cpuctx, cpu); cpuctx->task_ctx = ctx; } @@ -985,11 +1012,11 @@ static void perf_counter_cpu_sched_in(struct perf_cpu_context *cpuctx, int cpu) int perf_counter_task_disable(void) { struct task_struct *curr = current; - struct perf_counter_context *ctx = &curr->perf_counter_ctx; + struct perf_counter_context *ctx = curr->perf_counter_ctxp; struct perf_counter *counter; unsigned long flags; - if (likely(!ctx->nr_counters)) + if (!ctx || !ctx->nr_counters) return 0; local_irq_save(flags); @@ -1020,12 +1047,12 @@ int perf_counter_task_disable(void) int perf_counter_task_enable(void) { struct task_struct *curr = current; - struct perf_counter_context *ctx = &curr->perf_counter_ctx; + struct perf_counter_context *ctx = curr->perf_counter_ctxp; struct perf_counter *counter; unsigned long flags; int cpu; - if (likely(!ctx->nr_counters)) + if (!ctx || !ctx->nr_counters) return 0; local_irq_save(flags); @@ -1128,19 +1155,23 @@ void perf_counter_task_tick(struct task_struct *curr, int cpu) return; cpuctx = &per_cpu(perf_cpu_context, cpu); - ctx = &curr->perf_counter_ctx; + ctx = curr->perf_counter_ctxp; perf_adjust_freq(&cpuctx->ctx); - perf_adjust_freq(ctx); + if (ctx) + perf_adjust_freq(ctx); perf_counter_cpu_sched_out(cpuctx); - __perf_counter_task_sched_out(ctx); + if (ctx) + __perf_counter_task_sched_out(ctx); rotate_ctx(&cpuctx->ctx); - rotate_ctx(ctx); + if (ctx) + rotate_ctx(ctx); perf_counter_cpu_sched_in(cpuctx, cpu); - perf_counter_task_sched_in(curr, cpu); + if (ctx) + perf_counter_task_sched_in(curr, cpu); } /* @@ -1176,6 +1207,22 @@ static u64 perf_counter_read(struct perf_counter *counter) return atomic64_read(&counter->count); } +/* + * Initialize the perf_counter context in a task_struct: + */ +static void +__perf_counter_init_context(struct perf_counter_context *ctx, + struct task_struct *task) +{ + memset(ctx, 0, sizeof(*ctx)); + spin_lock_init(&ctx->lock); + mutex_init(&ctx->mutex); + INIT_LIST_HEAD(&ctx->counter_list); + INIT_LIST_HEAD(&ctx->event_list); + atomic_set(&ctx->refcount, 1); + ctx->task = task; +} + static void put_context(struct perf_counter_context *ctx) { if (ctx->task) @@ -1186,6 +1233,7 @@ static struct perf_counter_context *find_get_context(pid_t pid, int cpu) { struct perf_cpu_context *cpuctx; struct perf_counter_context *ctx; + struct perf_counter_context *tctx; struct task_struct *task; /* @@ -1225,15 +1273,36 @@ static struct perf_counter_context *find_get_context(pid_t pid, int cpu) if (!task) return ERR_PTR(-ESRCH); - ctx = &task->perf_counter_ctx; - ctx->task = task; - /* Reuse ptrace permission checks for now. */ if (!ptrace_may_access(task, PTRACE_MODE_READ)) { - put_context(ctx); + put_task_struct(task); return ERR_PTR(-EACCES); } + ctx = task->perf_counter_ctxp; + if (!ctx) { + ctx = kmalloc(sizeof(struct perf_counter_context), GFP_KERNEL); + if (!ctx) { + put_task_struct(task); + return ERR_PTR(-ENOMEM); + } + __perf_counter_init_context(ctx, task); + /* + * Make sure other cpus see correct values for *ctx + * once task->perf_counter_ctxp is visible to them. + */ + smp_wmb(); + tctx = cmpxchg(&task->perf_counter_ctxp, NULL, ctx); + if (tctx) { + /* + * We raced with some other task; use + * the context they set. + */ + kfree(ctx); + ctx = tctx; + } + } + return ctx; } @@ -1242,6 +1311,7 @@ static void free_counter_rcu(struct rcu_head *head) struct perf_counter *counter; counter = container_of(head, struct perf_counter, rcu_head); + put_ctx(counter->ctx); kfree(counter); } @@ -2247,7 +2317,7 @@ static void perf_counter_comm_event(struct perf_comm_event *comm_event) perf_counter_comm_ctx(&cpuctx->ctx, comm_event); put_cpu_var(perf_cpu_context); - perf_counter_comm_ctx(¤t->perf_counter_ctx, comm_event); + perf_counter_comm_ctx(current->perf_counter_ctxp, comm_event); } void perf_counter_comm(struct task_struct *task) @@ -2256,7 +2326,9 @@ void perf_counter_comm(struct task_struct *task) if (!atomic_read(&nr_comm_tracking)) return; - + if (!current->perf_counter_ctxp) + return; + comm_event = (struct perf_comm_event){ .task = task, .event = { @@ -2372,7 +2444,7 @@ got_name: perf_counter_mmap_ctx(&cpuctx->ctx, mmap_event); put_cpu_var(perf_cpu_context); - perf_counter_mmap_ctx(¤t->perf_counter_ctx, mmap_event); + perf_counter_mmap_ctx(current->perf_counter_ctxp, mmap_event); kfree(buf); } @@ -2384,6 +2456,8 @@ void perf_counter_mmap(unsigned long addr, unsigned long len, if (!atomic_read(&nr_mmap_tracking)) return; + if (!current->perf_counter_ctxp) + return; mmap_event = (struct perf_mmap_event){ .file = file, @@ -2985,6 +3059,7 @@ perf_counter_alloc(struct perf_counter_hw_event *hw_event, counter->group_leader = group_leader; counter->pmu = NULL; counter->ctx = ctx; + get_ctx(ctx); counter->state = PERF_COUNTER_STATE_INACTIVE; if (hw_event->disabled) @@ -3149,21 +3224,6 @@ err_put_context: goto out_fput; } -/* - * Initialize the perf_counter context in a task_struct: - */ -static void -__perf_counter_init_context(struct perf_counter_context *ctx, - struct task_struct *task) -{ - memset(ctx, 0, sizeof(*ctx)); - spin_lock_init(&ctx->lock); - mutex_init(&ctx->mutex); - INIT_LIST_HEAD(&ctx->counter_list); - INIT_LIST_HEAD(&ctx->event_list); - ctx->task = task; -} - /* * inherit a counter from parent task to child task: */ @@ -3195,7 +3255,6 @@ inherit_counter(struct perf_counter *parent_counter, /* * Link it up in the child's context: */ - child_counter->task = child; add_counter_to_ctx(child_counter, child_ctx); child_counter->parent = parent_counter; @@ -3294,40 +3353,15 @@ __perf_counter_exit_task(struct task_struct *child, struct perf_counter *parent_counter; /* - * If we do not self-reap then we have to wait for the - * child task to unschedule (it will happen for sure), - * so that its counter is at its final count. (This - * condition triggers rarely - child tasks usually get - * off their CPU before the parent has a chance to - * get this far into the reaping action) + * Protect against concurrent operations on child_counter + * due its fd getting closed, etc. */ - if (child != current) { - wait_task_inactive(child, 0); - update_counter_times(child_counter); - list_del_counter(child_counter, child_ctx); - } else { - struct perf_cpu_context *cpuctx; - unsigned long flags; - - /* - * Disable and unlink this counter. - * - * Be careful about zapping the list - IRQ/NMI context - * could still be processing it: - */ - local_irq_save(flags); - perf_disable(); - - cpuctx = &__get_cpu_var(perf_cpu_context); + mutex_lock(&child_counter->mutex); - group_sched_out(child_counter, cpuctx, child_ctx); - update_counter_times(child_counter); + update_counter_times(child_counter); + list_del_counter(child_counter, child_ctx); - list_del_counter(child_counter, child_ctx); - - perf_enable(); - local_irq_restore(flags); - } + mutex_unlock(&child_counter->mutex); parent_counter = child_counter->parent; /* @@ -3346,19 +3380,29 @@ __perf_counter_exit_task(struct task_struct *child, * * Note: we may be running in child context, but the PID is not hashed * anymore so new counters will not be added. + * (XXX not sure that is true when we get called from flush_old_exec. + * -- paulus) */ void perf_counter_exit_task(struct task_struct *child) { struct perf_counter *child_counter, *tmp; struct perf_counter_context *child_ctx; + unsigned long flags; WARN_ON_ONCE(child != current); - child_ctx = &child->perf_counter_ctx; + child_ctx = child->perf_counter_ctxp; - if (likely(!child_ctx->nr_counters)) + if (likely(!child_ctx)) return; + local_irq_save(flags); + __perf_counter_task_sched_out(child_ctx); + child->perf_counter_ctxp = NULL; + local_irq_restore(flags); + + mutex_lock(&child_ctx->mutex); + again: list_for_each_entry_safe(child_counter, tmp, &child_ctx->counter_list, list_entry) @@ -3371,6 +3415,10 @@ again: */ if (!list_empty(&child_ctx->counter_list)) goto again; + + mutex_unlock(&child_ctx->mutex); + + put_ctx(child_ctx); } /* @@ -3382,19 +3430,25 @@ void perf_counter_init_task(struct task_struct *child) struct perf_counter *counter; struct task_struct *parent = current; - child_ctx = &child->perf_counter_ctx; - parent_ctx = &parent->perf_counter_ctx; - - __perf_counter_init_context(child_ctx, child); + child->perf_counter_ctxp = NULL; /* * This is executed from the parent task context, so inherit - * counters that have been marked for cloning: + * counters that have been marked for cloning. + * First allocate and initialize a context for the child. */ - if (likely(!parent_ctx->nr_counters)) + child_ctx = kmalloc(sizeof(struct perf_counter_context), GFP_KERNEL); + if (!child_ctx) + return; + + parent_ctx = parent->perf_counter_ctxp; + if (likely(!parent_ctx || !parent_ctx->nr_counters)) return; + __perf_counter_init_context(child_ctx, child); + child->perf_counter_ctxp = child_ctx; + /* * Lock the parent list. No need to lock the child - not PID * hashed yet and not running, so nobody can access it. -- cgit v1.2.3-70-g09d2 From 48e22d56ecdeddd1ffb42a02fccba5c6ef42b133 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 25 May 2009 17:39:04 +0200 Subject: perf_counter: x86: Remove interrupt throttle remove the x86 specific interrupt throttle Signed-off-by: Peter Zijlstra Cc: Paul Mackerras Cc: Corey Ashford Cc: Arnaldo Carvalho de Melo Cc: John Kacur LKML-Reference: <20090525153931.616671838@chello.nl> Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic/apic.c | 2 -- arch/x86/kernel/cpu/perf_counter.c | 47 ++++---------------------------------- include/linux/perf_counter.h | 2 -- 3 files changed, 5 insertions(+), 46 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index b4f64402a82..89b63b5fad3 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -763,8 +763,6 @@ static void local_apic_timer_interrupt(void) inc_irq_stat(apic_timer_irqs); evt->event_handler(evt); - - perf_counter_unthrottle(); } /* diff --git a/arch/x86/kernel/cpu/perf_counter.c b/arch/x86/kernel/cpu/perf_counter.c index c14437faf5d..8c8177f859f 100644 --- a/arch/x86/kernel/cpu/perf_counter.c +++ b/arch/x86/kernel/cpu/perf_counter.c @@ -718,11 +718,6 @@ static void intel_pmu_save_and_restart(struct perf_counter *counter) intel_pmu_enable_counter(hwc, idx); } -/* - * Maximum interrupt frequency of 100KHz per CPU - */ -#define PERFMON_MAX_INTERRUPTS (100000/HZ) - /* * This handler is triggered by the local APIC, so the APIC IRQ handling * rules apply: @@ -775,15 +770,14 @@ again: if (status) goto again; - if (++cpuc->interrupts != PERFMON_MAX_INTERRUPTS) - perf_enable(); + perf_enable(); return 1; } static int amd_pmu_handle_irq(struct pt_regs *regs, int nmi) { - int cpu, idx, throttle = 0, handled = 0; + int cpu, idx, handled = 0; struct cpu_hw_counters *cpuc; struct perf_counter *counter; struct hw_perf_counter *hwc; @@ -792,16 +786,7 @@ static int amd_pmu_handle_irq(struct pt_regs *regs, int nmi) cpu = smp_processor_id(); cpuc = &per_cpu(cpu_hw_counters, cpu); - if (++cpuc->interrupts == PERFMON_MAX_INTERRUPTS) { - throttle = 1; - __perf_disable(); - cpuc->enabled = 0; - barrier(); - } - for (idx = 0; idx < x86_pmu.num_counters; idx++) { - int disable = 0; - if (!test_bit(idx, cpuc->active_mask)) continue; @@ -809,45 +794,23 @@ static int amd_pmu_handle_irq(struct pt_regs *regs, int nmi) hwc = &counter->hw; if (counter->hw_event.nmi != nmi) - goto next; + continue; val = x86_perf_counter_update(counter, hwc, idx); if (val & (1ULL << (x86_pmu.counter_bits - 1))) - goto next; + continue; /* counter overflow */ x86_perf_counter_set_period(counter, hwc, idx); handled = 1; inc_irq_stat(apic_perf_irqs); - disable = perf_counter_overflow(counter, nmi, regs, 0); - -next: - if (disable || throttle) + if (perf_counter_overflow(counter, nmi, regs, 0)) amd_pmu_disable_counter(hwc, idx); } return handled; } -void perf_counter_unthrottle(void) -{ - struct cpu_hw_counters *cpuc; - - if (!x86_pmu_initialized()) - return; - - cpuc = &__get_cpu_var(cpu_hw_counters); - if (cpuc->interrupts >= PERFMON_MAX_INTERRUPTS) { - /* - * Clear them before re-enabling irqs/NMIs again: - */ - cpuc->interrupts = 0; - perf_enable(); - } else { - cpuc->interrupts = 0; - } -} - void smp_perf_counter_interrupt(struct pt_regs *regs) { irq_enter(); diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h index d3e85de9bf1..0c160be2078 100644 --- a/include/linux/perf_counter.h +++ b/include/linux/perf_counter.h @@ -570,7 +570,6 @@ extern int perf_counter_init_task(struct task_struct *child); extern void perf_counter_exit_task(struct task_struct *child); extern void perf_counter_do_pending(void); extern void perf_counter_print_debug(void); -extern void perf_counter_unthrottle(void); extern void __perf_disable(void); extern bool __perf_enable(void); extern void perf_disable(void); @@ -635,7 +634,6 @@ static inline int perf_counter_init_task(struct task_struct *child) { } static inline void perf_counter_exit_task(struct task_struct *child) { } static inline void perf_counter_do_pending(void) { } static inline void perf_counter_print_debug(void) { } -static inline void perf_counter_unthrottle(void) { } static inline void perf_disable(void) { } static inline void perf_enable(void) { } static inline int perf_counter_task_disable(void) { return -EINVAL; } -- cgit v1.2.3-70-g09d2 From 4efc0670baf4b14bc95502e54a83ccf639146125 Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Tue, 28 Apr 2009 19:07:31 +0200 Subject: x86, mce: use 64bit machine check code on 32bit The 64bit machine check code is in many ways much better than the 32bit machine check code: it is more specification compliant, is cleaner, only has a single code base versus one per CPU, has better infrastructure for recovery, has a cleaner way to communicate with user space etc. etc. Use the 64bit code for 32bit too. This is the second attempt to do this. There was one a couple of years ago to unify this code for 32bit and 64bit. Back then this ran into some trouble with K7s and was reverted. I believe this time the K7 problems (and some others) are addressed. I went over the old handlers and was very careful to retain all quirks. But of course this needs a lot of testing on old systems. On newer 64bit capable systems I don't expect much problems because they have been already tested with the 64bit kernel. I made this a CONFIG for now that still allows to select the old machine check code. This is mostly to make testing easier, if someone runs into a problem we can ask them to try with the CONFIG switched. The new code is default y for more coverage. Once there is confidence the 64bit code works well on older hardware too the CONFIG_X86_OLD_MCE and the associated code can be easily removed. This causes a behaviour change for 32bit installations. They now have to install the mcelog package to be able to log corrected machine checks. The 64bit machine check code only handles CPUs which support the standard Intel machine check architecture described in the IA32 SDM. The 32bit code has special support for some older CPUs which have non standard machine check architectures, in particular WinChip C3 and Intel P5. I made those a separate CONFIG option and kept them for now. The WinChip variant could be probably removed without too much pain, it doesn't really do anything interesting. P5 is also disabled by default (like it was before) because many motherboards have it miswired, but according to Alan Cox a few embedded setups use that one. Forward ported/heavily changed version of old patch, original patch included review/fixes from Thomas Gleixner, Bert Wesarg. Signed-off-by: Andi Kleen Signed-off-by: H. Peter Anvin Signed-off-by: Hidetoshi Seto Signed-off-by: H. Peter Anvin --- arch/x86/Kconfig | 33 +++++++++++++++++++++++++++++++-- arch/x86/include/asm/entry_arch.h | 2 +- arch/x86/kernel/apic/apic.c | 4 ++-- arch/x86/kernel/apic/nmi.c | 2 +- arch/x86/kernel/cpu/mcheck/Makefile | 3 ++- arch/x86/kernel/cpu/mcheck/mce.c | 32 ++++++++++++++++++++++++++++---- arch/x86/kernel/cpu/mcheck/mce.h | 18 +++++++++++++++--- arch/x86/kernel/cpu/mcheck/p5.c | 5 +++++ arch/x86/kernel/irq.c | 4 ++-- arch/x86/kernel/irqinit_32.c | 2 +- arch/x86/kernel/signal.c | 4 ++-- arch/x86/kernel/traps.c | 4 ++-- 12 files changed, 92 insertions(+), 21 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index a6efe0a2e9a..c1c5ccd1937 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -789,6 +789,22 @@ config X86_MCE to disable it. MCE support simply ignores non-MCE processors like the 386 and 486, so nearly everyone can say Y here. +config X86_OLD_MCE + depends on X86_32 && X86_MCE + bool "Use legacy machine check code (will go away)" + default n + select X86_ANCIENT_MCE + ---help--- + Use the old i386 machine check code. This is merely intended for + testing in a transition period. Try this if you run into any machine + check related software problems, but report the problem to + linux-kernel. When in doubt say no. + +config X86_NEW_MCE + depends on X86_MCE + bool + default y if (!X86_OLD_MCE && X86_32) || X86_64 + config X86_MCE_INTEL def_bool y prompt "Intel MCE features" @@ -805,6 +821,15 @@ config X86_MCE_AMD Additional support for AMD specific MCE features such as the DRAM Error Threshold. +config X86_ANCIENT_MCE + def_bool n + depends on X86_32 + prompt "Support for old Pentium 5 / WinChip machine checks" + ---help--- + Include support for machine check handling on old Pentium 5 or WinChip + systems. These typically need to be enabled explicitely on the command + line. + config X86_MCE_THRESHOLD depends on X86_MCE_AMD || X86_MCE_INTEL bool @@ -812,7 +837,7 @@ config X86_MCE_THRESHOLD config X86_MCE_NONFATAL tristate "Check for non-fatal errors on AMD Athlon/Duron / Intel Pentium 4" - depends on X86_32 && X86_MCE + depends on X86_OLD_MCE ---help--- Enabling this feature starts a timer that triggers every 5 seconds which will look at the machine check registers to see if anything happened. @@ -825,11 +850,15 @@ config X86_MCE_NONFATAL config X86_MCE_P4THERMAL bool "check for P4 thermal throttling interrupt." - depends on X86_32 && X86_MCE && (X86_UP_APIC || SMP) + depends on X86_OLD_MCE && X86_MCE && (X86_UP_APIC || SMP) ---help--- Enabling this feature will cause a message to be printed when the P4 enters thermal throttling. +config X86_THERMAL_VECTOR + def_bool y + depends on X86_MCE_P4THERMAL || X86_MCE_INTEL + config VM86 bool "Enable VM86 support" if EMBEDDED default y diff --git a/arch/x86/include/asm/entry_arch.h b/arch/x86/include/asm/entry_arch.h index c2e6bedaf25..486c9e946f5 100644 --- a/arch/x86/include/asm/entry_arch.h +++ b/arch/x86/include/asm/entry_arch.h @@ -52,7 +52,7 @@ BUILD_INTERRUPT(spurious_interrupt,SPURIOUS_APIC_VECTOR) BUILD_INTERRUPT(perf_counter_interrupt, LOCAL_PERF_VECTOR) #endif -#ifdef CONFIG_X86_MCE_P4THERMAL +#ifdef CONFIG_X86_THERMAL_VECTOR BUILD_INTERRUPT(thermal_interrupt,THERMAL_APIC_VECTOR) #endif diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index f2870920f24..ad532289ef2 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -843,7 +843,7 @@ void clear_local_APIC(void) } /* lets not touch this if we didn't frob it */ -#if defined(CONFIG_X86_MCE_P4THERMAL) || defined(CONFIG_X86_MCE_INTEL) +#ifdef CONFIG_X86_THERMAL_VECTOR if (maxlvt >= 5) { v = apic_read(APIC_LVTTHMR); apic_write(APIC_LVTTHMR, v | APIC_LVT_MASKED); @@ -1962,7 +1962,7 @@ static int lapic_suspend(struct sys_device *dev, pm_message_t state) apic_pm_state.apic_lvterr = apic_read(APIC_LVTERR); apic_pm_state.apic_tmict = apic_read(APIC_TMICT); apic_pm_state.apic_tdcr = apic_read(APIC_TDCR); -#if defined(CONFIG_X86_MCE_P4THERMAL) || defined(CONFIG_X86_MCE_INTEL) +#ifdef CONFIG_X86_THERMAL_VECTOR if (maxlvt >= 5) apic_pm_state.apic_thmr = apic_read(APIC_LVTTHMR); #endif diff --git a/arch/x86/kernel/apic/nmi.c b/arch/x86/kernel/apic/nmi.c index ce4fbfa315a..c4762276c17 100644 --- a/arch/x86/kernel/apic/nmi.c +++ b/arch/x86/kernel/apic/nmi.c @@ -66,7 +66,7 @@ static inline unsigned int get_nmi_count(int cpu) static inline int mce_in_progress(void) { -#if defined(CONFIG_X86_64) && defined(CONFIG_X86_MCE) +#if defined(CONFIG_X86_NEW_MCE) return atomic_read(&mce_entry) > 0; #endif return 0; diff --git a/arch/x86/kernel/cpu/mcheck/Makefile b/arch/x86/kernel/cpu/mcheck/Makefile index 55f01b39a10..5f8b09425d3 100644 --- a/arch/x86/kernel/cpu/mcheck/Makefile +++ b/arch/x86/kernel/cpu/mcheck/Makefile @@ -1,6 +1,7 @@ obj-y = mce.o therm_throt.o -obj-$(CONFIG_X86_32) += k7.o p4.o p5.o p6.o winchip.o +obj-$(CONFIG_X86_OLD_MCE) += k7.o p4.o p6.o +obj-$(CONFIG_X86_ANCIENT_MCE) += winchip.o p5.o obj-$(CONFIG_X86_MCE_P4THERMAL) += mce_intel.o obj-$(CONFIG_X86_MCE_INTEL) += mce_intel_64.o mce_intel.o obj-$(CONFIG_X86_MCE_AMD) += mce_amd_64.o diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c index f4d6841d2bd..e193de44ef1 100644 --- a/arch/x86/kernel/cpu/mcheck/mce.c +++ b/arch/x86/kernel/cpu/mcheck/mce.c @@ -52,7 +52,7 @@ void (*machine_check_vector)(struct pt_regs *, long error_code) = int mce_disabled; -#ifdef CONFIG_X86_64 +#ifdef CONFIG_X86_NEW_MCE #define MISC_MCELOG_MINOR 227 @@ -662,6 +662,21 @@ static void mce_cpu_quirks(struct cpuinfo_x86 *c) } } +static void __cpuinit mce_ancient_init(struct cpuinfo_x86 *c) +{ + if (c->x86 != 5) + return; + switch (c->x86_vendor) { + case X86_VENDOR_INTEL: + if (mce_p5_enabled()) + intel_p5_mcheck_init(c); + break; + case X86_VENDOR_CENTAUR: + winchip_mcheck_init(c); + break; + } +} + static void mce_cpu_features(struct cpuinfo_x86 *c) { switch (c->x86_vendor) { @@ -695,6 +710,11 @@ static void mce_init_timer(void) */ void __cpuinit mcheck_init(struct cpuinfo_x86 *c) { + if (mce_disabled) + return; + + mce_ancient_init(c); + if (!mce_available(c)) return; @@ -893,6 +913,10 @@ static struct miscdevice mce_log_device = { */ static int __init mcheck_enable(char *str) { + if (*str == 0) + enable_p5_mce(); + if (*str == '=') + str++; if (!strcmp(str, "off")) mce_disabled = 1; else if (!strcmp(str, "bootlog") || !strcmp(str, "nobootlog")) @@ -900,13 +924,13 @@ static int __init mcheck_enable(char *str) else if (isdigit(str[0])) get_option(&str, &tolerant); else { - printk(KERN_INFO "mce= argument %s ignored. Please use /sys\n", + printk(KERN_INFO "mce argument %s ignored. Please use /sys\n", str); return 0; } return 1; } -__setup("mce=", mcheck_enable); +__setup("mce", mcheck_enable); /* * Sysfs support @@ -1259,7 +1283,7 @@ static __init int mce_init_device(void) device_initcall(mce_init_device); -#else /* CONFIG_X86_32: */ +#else /* CONFIG_X86_OLD_MCE: */ int nr_mce_banks; EXPORT_SYMBOL_GPL(nr_mce_banks); /* non-fatal.o */ diff --git a/arch/x86/kernel/cpu/mcheck/mce.h b/arch/x86/kernel/cpu/mcheck/mce.h index 966ae3c5cb1..84a552b458c 100644 --- a/arch/x86/kernel/cpu/mcheck/mce.h +++ b/arch/x86/kernel/cpu/mcheck/mce.h @@ -1,17 +1,29 @@ #include #include +#ifdef CONFIG_X86_OLD_MCE void amd_mcheck_init(struct cpuinfo_x86 *c); void intel_p4_mcheck_init(struct cpuinfo_x86 *c); -void intel_p5_mcheck_init(struct cpuinfo_x86 *c); void intel_p6_mcheck_init(struct cpuinfo_x86 *c); -void winchip_mcheck_init(struct cpuinfo_x86 *c); +#endif +#ifdef CONFIG_X86_ANCIENT_MCE +void intel_p5_mcheck_init(struct cpuinfo_x86 *c); +void winchip_mcheck_init(struct cpuinfo_x86 *c); +extern int mce_p5_enable; +static inline int mce_p5_enabled(void) { return mce_p5_enable; } +static inline void enable_p5_mce(void) { mce_p5_enable = 1; } +#else +static inline void intel_p5_mcheck_init(struct cpuinfo_x86 *c) {} +static inline void winchip_mcheck_init(struct cpuinfo_x86 *c) {} +static inline int mce_p5_enabled(void) { return 0; } +static inline void enable_p5_mce(void) { } +#endif /* Call the installed machine check handler for this CPU setup. */ extern void (*machine_check_vector)(struct pt_regs *, long error_code); -#ifdef CONFIG_X86_32 +#ifdef CONFIG_X86_OLD_MCE extern int nr_mce_banks; diff --git a/arch/x86/kernel/cpu/mcheck/p5.c b/arch/x86/kernel/cpu/mcheck/p5.c index 8812f544183..015f481ab1b 100644 --- a/arch/x86/kernel/cpu/mcheck/p5.c +++ b/arch/x86/kernel/cpu/mcheck/p5.c @@ -14,6 +14,9 @@ #include "mce.h" +/* By default disabled */ +int mce_p5_enable; + /* Machine check handler for Pentium class Intel CPUs: */ static void pentium_machine_check(struct pt_regs *regs, long error_code) { @@ -44,9 +47,11 @@ void intel_p5_mcheck_init(struct cpuinfo_x86 *c) if (!cpu_has(c, X86_FEATURE_MCE)) return; +#ifdef CONFIG_X86_OLD_MCE /* Default P5 to off as its often misconnected: */ if (mce_disabled != -1) return; +#endif machine_check_vector = pentium_machine_check; /* Make sure the vector pointer is visible before we enable MCEs: */ diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c index c3fe010d74c..35eddc9ec99 100644 --- a/arch/x86/kernel/irq.c +++ b/arch/x86/kernel/irq.c @@ -89,7 +89,7 @@ static int show_other_interrupts(struct seq_file *p, int prec) for_each_online_cpu(j) seq_printf(p, "%10u ", irq_stats(j)->irq_thermal_count); seq_printf(p, " Thermal event interrupts\n"); -# ifdef CONFIG_X86_64 +# ifdef CONFIG_X86_MCE_THRESHOLD seq_printf(p, "%*s: ", prec, "THR"); for_each_online_cpu(j) seq_printf(p, "%10u ", irq_stats(j)->irq_threshold_count); @@ -176,7 +176,7 @@ u64 arch_irq_stat_cpu(unsigned int cpu) #endif #ifdef CONFIG_X86_MCE sum += irq_stats(cpu)->irq_thermal_count; -# ifdef CONFIG_X86_64 +# ifdef CONFIG_X86_MCE_THRESHOLD sum += irq_stats(cpu)->irq_threshold_count; #endif #endif diff --git a/arch/x86/kernel/irqinit_32.c b/arch/x86/kernel/irqinit_32.c index 368b0a8836f..98846e03211 100644 --- a/arch/x86/kernel/irqinit_32.c +++ b/arch/x86/kernel/irqinit_32.c @@ -181,7 +181,7 @@ void __init native_init_IRQ(void) alloc_intr_gate(ERROR_APIC_VECTOR, error_interrupt); #endif -#if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86_MCE_P4THERMAL) +#ifdef CONFIG_X86_THERMAL_VECTOR /* thermal monitor LVT interrupt */ alloc_intr_gate(THERMAL_APIC_VECTOR, thermal_interrupt); #endif diff --git a/arch/x86/kernel/signal.c b/arch/x86/kernel/signal.c index 14425166b8e..d0851e3f77e 100644 --- a/arch/x86/kernel/signal.c +++ b/arch/x86/kernel/signal.c @@ -25,11 +25,11 @@ #include #include #include +#include #ifdef CONFIG_X86_64 #include #include -#include #endif /* CONFIG_X86_64 */ #include @@ -857,7 +857,7 @@ static void do_signal(struct pt_regs *regs) void do_notify_resume(struct pt_regs *regs, void *unused, __u32 thread_info_flags) { -#if defined(CONFIG_X86_64) && defined(CONFIG_X86_MCE) +#ifdef CONFIG_X86_NEW_MCE /* notify userspace of pending MCEs */ if (thread_info_flags & _TIF_MCE_NOTIFY) mce_notify_user(); diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c index a1d288327ff..ad771f15bdd 100644 --- a/arch/x86/kernel/traps.c +++ b/arch/x86/kernel/traps.c @@ -798,7 +798,8 @@ unsigned long patch_espfix_desc(unsigned long uesp, unsigned long kesp) return new_kesp; } -#else +#endif + asmlinkage void __attribute__((weak)) smp_thermal_interrupt(void) { } @@ -806,7 +807,6 @@ asmlinkage void __attribute__((weak)) smp_thermal_interrupt(void) asmlinkage void __attribute__((weak)) mce_threshold_interrupt(void) { } -#endif /* * 'math_state_restore()' saves the current math information in the -- cgit v1.2.3-70-g09d2 From c323d95fa4dbe0b6bf6d59e24a0b7db067dd08a7 Mon Sep 17 00:00:00 2001 From: Yong Wang Date: Fri, 29 May 2009 13:28:35 +0800 Subject: perf_counter/x86: Always use NMI for performance-monitoring interrupt Always use NMI for performance-monitoring interrupt as there could be racy situations if we switch between irq and nmi mode frequently. Signed-off-by: Yong Wang LKML-Reference: <20090529052835.GA13657@ywang-moblin2.bj.intel.com> Signed-off-by: Ingo Molnar --- arch/x86/include/asm/perf_counter.h | 4 ++-- arch/x86/kernel/apic/apic.c | 2 +- arch/x86/kernel/cpu/perf_counter.c | 19 +++++-------------- 3 files changed, 8 insertions(+), 17 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/include/asm/perf_counter.h b/arch/x86/include/asm/perf_counter.h index d08dd52cb8f..876ed97147b 100644 --- a/arch/x86/include/asm/perf_counter.h +++ b/arch/x86/include/asm/perf_counter.h @@ -91,10 +91,10 @@ extern void set_perf_counter_pending(void); #ifdef CONFIG_PERF_COUNTERS extern void init_hw_perf_counters(void); -extern void perf_counters_lapic_init(int nmi); +extern void perf_counters_lapic_init(void); #else static inline void init_hw_perf_counters(void) { } -static inline void perf_counters_lapic_init(int nmi) { } +static inline void perf_counters_lapic_init(void) { } #endif #endif /* _ASM_X86_PERF_COUNTER_H */ diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index 89b63b5fad3..60df2efd7c8 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -1135,7 +1135,7 @@ void __cpuinit setup_local_APIC(void) apic_write(APIC_ESR, 0); } #endif - perf_counters_lapic_init(0); + perf_counters_lapic_init(); preempt_disable(); diff --git a/arch/x86/kernel/cpu/perf_counter.c b/arch/x86/kernel/cpu/perf_counter.c index 2eeaa99add1..316b0c995f3 100644 --- a/arch/x86/kernel/cpu/perf_counter.c +++ b/arch/x86/kernel/cpu/perf_counter.c @@ -604,7 +604,7 @@ try_generic: hwc->counter_base = x86_pmu.perfctr; } - perf_counters_lapic_init(hwc->nmi); + perf_counters_lapic_init(); x86_pmu.disable(hwc, idx); @@ -863,24 +863,15 @@ void set_perf_counter_pending(void) apic->send_IPI_self(LOCAL_PENDING_VECTOR); } -void perf_counters_lapic_init(int nmi) +void perf_counters_lapic_init(void) { - u32 apic_val; - if (!x86_pmu_initialized()) return; /* - * Enable the performance counter vector in the APIC LVT: + * Always use NMI for PMU */ - apic_val = apic_read(APIC_LVTERR); - - apic_write(APIC_LVTERR, apic_val | APIC_LVT_MASKED); - if (nmi) - apic_write(APIC_LVTPC, APIC_DM_NMI); - else - apic_write(APIC_LVTPC, LOCAL_PERF_VECTOR); - apic_write(APIC_LVTERR, apic_val); + apic_write(APIC_LVTPC, APIC_DM_NMI); } static int __kprobes @@ -1054,7 +1045,7 @@ void __init init_hw_perf_counters(void) pr_info("... counter mask: %016Lx\n", perf_counter_mask); - perf_counters_lapic_init(0); + perf_counters_lapic_init(); register_die_notifier(&perf_counter_nmi_notifier); } -- cgit v1.2.3-70-g09d2 From 3d58829b0510244596079c1d2f1762c53aef2e97 Mon Sep 17 00:00:00 2001 From: Jiri Slaby Date: Thu, 28 May 2009 09:54:47 +0200 Subject: x86, apic: Restore irqs on fail paths lapic_resume forgets to restore interrupts on fail paths. Fix that. Signed-off-by: Jiri Slaby Acked-by: Cyrill Gorcunov LKML-Reference: <1243497289-18591-1-git-send-email-jirislaby@gmail.com> Signed-off-by: Ingo Molnar Cc: H. Peter Anvin --- arch/x86/kernel/apic/apic.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index b0fd26442c4..e82488d3f0b 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -2027,7 +2027,7 @@ static int lapic_resume(struct sys_device *dev) unsigned int l, h; unsigned long flags; int maxlvt; - int ret; + int ret = 0; struct IO_APIC_route_entry **ioapic_entries = NULL; if (!apic_pm_state.active) @@ -2038,14 +2038,15 @@ static int lapic_resume(struct sys_device *dev) ioapic_entries = alloc_ioapic_entries(); if (!ioapic_entries) { WARN(1, "Alloc ioapic_entries in lapic resume failed."); - return -ENOMEM; + ret = -ENOMEM; + goto restore; } ret = save_IO_APIC_setup(ioapic_entries); if (ret) { WARN(1, "Saving IO-APIC state failed: %d\n", ret); free_ioapic_entries(ioapic_entries); - return ret; + goto restore; } mask_IO_APIC_setup(ioapic_entries); @@ -2097,10 +2098,10 @@ static int lapic_resume(struct sys_device *dev) restore_IO_APIC_setup(ioapic_entries); free_ioapic_entries(ioapic_entries); } - +restore: local_irq_restore(flags); - return 0; + return ret; } /* -- cgit v1.2.3-70-g09d2 From 103428e57be323c3c5545db8ad12667099bc6005 Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Sun, 7 Jun 2009 16:48:40 +0400 Subject: x86, apic: Fix dummy apic read operation together with broken MP handling Ingo Molnar reported that read_apic is buggy novadays: [ 0.000000] Using APIC driver default [ 0.000000] SMP: Allowing 1 CPUs, 0 hotplug CPUs [ 0.000000] Local APIC disabled by BIOS -- you can enable it with "lapic" [ 0.000000] APIC: disable apic facility [ 0.000000] ------------[ cut here ]------------ [ 0.000000] WARNING: at arch/x86/kernel/apic/apic.c:254 native_apic_read_dummy+0x2d/0x3b() [ 0.000000] Hardware name: HP OmniBook PC Indeed we still rely on apic->read operation for SMP compiled kernel. And instead of disfigure the SMP code with #ifdef we allow to call apic->read. To capture any unexpected results we check for apic->read being called for sane reason via WARN_ON_ONCE but(!) instead of OR we should use AND logical operation (thanks Yinghai for spotting the root of the problem). Along with that we could be have bad MP table and we are to fix it that way no SMP started and no complains about BIOS bug if apic was just disabled via command line. Signed-off-by: Cyrill Gorcunov Cc: Yinghai Lu LKML-Reference: <20090607124840.GD4547@lenovo> Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic/apic.c | 9 ++++++++- arch/x86/kernel/smpboot.c | 8 +++++--- 2 files changed, 13 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel/apic/apic.c') diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index e82488d3f0b..a4c9cf0bf70 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -249,7 +249,7 @@ static void native_apic_write_dummy(u32 reg, u32 v) static u32 native_apic_read_dummy(u32 reg) { - WARN_ON_ONCE((cpu_has_apic || !disable_apic)); + WARN_ON_ONCE((cpu_has_apic && !disable_apic)); return 0; } @@ -1609,6 +1609,13 @@ void __init init_apic_mappings(void) new_apicid = read_apic_id(); if (boot_cpu_physical_apicid != new_apicid) { boot_cpu_physical_apicid = new_apicid; + /* + * yeah -- we lie about apic_version + * in case if apic was disabled via boot option + * but it's not a problem for SMP compiled kernel + * since smp_sanity_check is prepared for such a case + * and disable smp mode + */ apic_version[new_apicid] = GET_APIC_VERSION(apic_read(APIC_LVR)); } diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index d2e8de95815..7c80007ea5f 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -992,10 +992,12 @@ static int __init smp_sanity_check(unsigned max_cpus) */ if (APIC_INTEGRATED(apic_version[boot_cpu_physical_apicid]) && !cpu_has_apic) { - printk(KERN_ERR "BIOS bug, local APIC #%d not detected!...\n", - boot_cpu_physical_apicid); - printk(KERN_ERR "... forcing use of dummy APIC emulation." + if (!disable_apic) { + pr_err("BIOS bug, local APIC #%d not detected!...\n", + boot_cpu_physical_apicid); + pr_err("... forcing use of dummy APIC emulation." "(tell your hw vendor)\n"); + } smpboot_clear_io_apic(); arch_disable_smp_support(); return -1; -- cgit v1.2.3-70-g09d2