diff options
179 files changed, 6925 insertions, 1342 deletions
diff --git a/Documentation/cputopology.txt b/Documentation/cputopology.txt index 45932ec21ce..b41f3e58aef 100644 --- a/Documentation/cputopology.txt +++ b/Documentation/cputopology.txt @@ -18,11 +18,11 @@ For an architecture to support this feature, it must define some of these macros in include/asm-XXX/topology.h: #define topology_physical_package_id(cpu) #define topology_core_id(cpu) -#define topology_thread_siblings(cpu) -#define topology_core_siblings(cpu) +#define topology_thread_cpumask(cpu) +#define topology_core_cpumask(cpu) The type of **_id is int. -The type of siblings is cpumask_t. +The type of siblings is (const) struct cpumask *. To be consistent on all architectures, include/linux/topology.h provides default definitions for any of the above macros that are diff --git a/Documentation/perf-counters.txt b/Documentation/perf-counters.txt new file mode 100644 index 00000000000..fddd32189a5 --- /dev/null +++ b/Documentation/perf-counters.txt @@ -0,0 +1,147 @@ + +Performance Counters for Linux +------------------------------ + +Performance counters are special hardware registers available on most modern +CPUs. These registers count the number of certain types of hw events: such +as instructions executed, cachemisses suffered, or branches mis-predicted - +without slowing down the kernel or applications. These registers can also +trigger interrupts when a threshold number of events have passed - and can +thus be used to profile the code that runs on that CPU. + +The Linux Performance Counter subsystem provides an abstraction of these +hardware capabilities. It provides per task and per CPU counters, counter +groups, and it provides event capabilities on top of those. + +Performance counters are accessed via special file descriptors. +There's one file descriptor per virtual counter used. + +The special file descriptor is opened via the perf_counter_open() +system call: + + int sys_perf_counter_open(struct perf_counter_hw_event *hw_event_uptr, + pid_t pid, int cpu, int group_fd); + +The syscall returns the new fd. The fd can be used via the normal +VFS system calls: read() can be used to read the counter, fcntl() +can be used to set the blocking mode, etc. + +Multiple counters can be kept open at a time, and the counters +can be poll()ed. + +When creating a new counter fd, 'perf_counter_hw_event' is: + +/* + * Hardware event to monitor via a performance monitoring counter: + */ +struct perf_counter_hw_event { + s64 type; + + u64 irq_period; + u32 record_type; + + u32 disabled : 1, /* off by default */ + nmi : 1, /* NMI sampling */ + raw : 1, /* raw event type */ + __reserved_1 : 29; + + u64 __reserved_2; +}; + +/* + * Generalized performance counter event types, used by the hw_event.type + * parameter of the sys_perf_counter_open() syscall: + */ +enum hw_event_types { + /* + * Common hardware events, generalized by the kernel: + */ + PERF_COUNT_CYCLES = 0, + PERF_COUNT_INSTRUCTIONS = 1, + PERF_COUNT_CACHE_REFERENCES = 2, + PERF_COUNT_CACHE_MISSES = 3, + PERF_COUNT_BRANCH_INSTRUCTIONS = 4, + PERF_COUNT_BRANCH_MISSES = 5, + + /* + * Special "software" counters provided by the kernel, even if + * the hardware does not support performance counters. These + * counters measure various physical and sw events of the + * kernel (and allow the profiling of them as well): + */ + PERF_COUNT_CPU_CLOCK = -1, + PERF_COUNT_TASK_CLOCK = -2, + /* + * Future software events: + */ + /* PERF_COUNT_PAGE_FAULTS = -3, + PERF_COUNT_CONTEXT_SWITCHES = -4, */ +}; + +These are standardized types of events that work uniformly on all CPUs +that implements Performance Counters support under Linux. If a CPU is +not able to count branch-misses, then the system call will return +-EINVAL. + +More hw_event_types are supported as well, but they are CPU +specific and are enumerated via /sys on a per CPU basis. Raw hw event +types can be passed in under hw_event.type if hw_event.raw is 1. +For example, to count "External bus cycles while bus lock signal asserted" +events on Intel Core CPUs, pass in a 0x4064 event type value and set +hw_event.raw to 1. + +'record_type' is the type of data that a read() will provide for the +counter, and it can be one of: + +/* + * IRQ-notification data record type: + */ +enum perf_counter_record_type { + PERF_RECORD_SIMPLE = 0, + PERF_RECORD_IRQ = 1, + PERF_RECORD_GROUP = 2, +}; + +a "simple" counter is one that counts hardware events and allows +them to be read out into a u64 count value. (read() returns 8 on +a successful read of a simple counter.) + +An "irq" counter is one that will also provide an IRQ context information: +the IP of the interrupted context. In this case read() will return +the 8-byte counter value, plus the Instruction Pointer address of the +interrupted context. + +The parameter 'hw_event_period' is the number of events before waking up +a read() that is blocked on a counter fd. Zero value means a non-blocking +counter. + +The 'pid' parameter allows the counter to be specific to a task: + + pid == 0: if the pid parameter is zero, the counter is attached to the + current task. + + pid > 0: the counter is attached to a specific task (if the current task + has sufficient privilege to do so) + + pid < 0: all tasks are counted (per cpu counters) + +The 'cpu' parameter allows a counter to be made specific to a full +CPU: + + cpu >= 0: the counter is restricted to a specific CPU + cpu == -1: the counter counts on all CPUs + +(Note: the combination of 'pid == -1' and 'cpu == -1' is not valid.) + +A 'pid > 0' and 'cpu == -1' counter is a per task counter that counts +events of that task and 'follows' that task to whatever CPU the task +gets schedule to. Per task counters can be created by any user, for +their own tasks. + +A 'pid == -1' and 'cpu == x' counter is a per CPU counter that counts +all events on CPU-x. Per CPU counters need CAP_SYS_ADMIN privilege. + +Group counters are created by passing in a group_fd of another counter. +Groups are scheduled at once and can be used with PERF_RECORD_GROUP +to record multi-dimensional timestamps. + diff --git a/arch/alpha/kernel/irq.c b/arch/alpha/kernel/irq.c index 703731accda..7bc7489223f 100644 --- a/arch/alpha/kernel/irq.c +++ b/arch/alpha/kernel/irq.c @@ -55,7 +55,7 @@ int irq_select_affinity(unsigned int irq) cpu = (cpu < (NR_CPUS-1) ? cpu + 1 : 0); last_cpu = cpu; - irq_desc[irq].affinity = cpumask_of_cpu(cpu); + cpumask_copy(irq_desc[irq].affinity, cpumask_of(cpu)); irq_desc[irq].chip->set_affinity(irq, cpumask_of(cpu)); return 0; } diff --git a/arch/arm/kernel/irq.c b/arch/arm/kernel/irq.c index 7141cee1fab..4bb723eadad 100644 --- a/arch/arm/kernel/irq.c +++ b/arch/arm/kernel/irq.c @@ -104,6 +104,11 @@ static struct irq_desc bad_irq_desc = { .lock = SPIN_LOCK_UNLOCKED }; +#ifdef CONFIG_CPUMASK_OFFSTACK +/* We are not allocating bad_irq_desc.affinity or .pending_mask */ +#error "ARM architecture does not support CONFIG_CPUMASK_OFFSTACK." +#endif + /* * do_IRQ handles all hardware IRQ's. Decoded IRQs should not * come via this function. Instead, they should provide their @@ -161,7 +166,7 @@ void __init init_IRQ(void) irq_desc[irq].status |= IRQ_NOREQUEST | IRQ_NOPROBE; #ifdef CONFIG_SMP - bad_irq_desc.affinity = CPU_MASK_ALL; + cpumask_setall(bad_irq_desc.affinity); bad_irq_desc.cpu = smp_processor_id(); #endif init_arch_irq(); @@ -191,15 +196,16 @@ void migrate_irqs(void) struct irq_desc *desc = irq_desc + i; if (desc->cpu == cpu) { - unsigned int newcpu = any_online_cpu(desc->affinity); - - if (newcpu == NR_CPUS) { + unsigned int newcpu = cpumask_any_and(desc->affinity, + cpu_online_mask); + if (newcpu >= nr_cpu_ids) { if (printk_ratelimit()) printk(KERN_INFO "IRQ%u no longer affine to CPU%u\n", i, cpu); - cpus_setall(desc->affinity); - newcpu = any_online_cpu(desc->affinity); + cpumask_setall(desc->affinity); + newcpu = cpumask_any_and(desc->affinity, + cpu_online_mask); } route_irq(desc, i, newcpu); diff --git a/arch/arm/kernel/vmlinux.lds.S b/arch/arm/kernel/vmlinux.lds.S index 00216071eaf..85598f7da40 100644 --- a/arch/arm/kernel/vmlinux.lds.S +++ b/arch/arm/kernel/vmlinux.lds.S @@ -65,6 +65,7 @@ SECTIONS #endif . = ALIGN(4096); __per_cpu_start = .; + *(.data.percpu.page_aligned) *(.data.percpu) *(.data.percpu.shared_aligned) __per_cpu_end = .; diff --git a/arch/arm/oprofile/op_model_mpcore.c b/arch/arm/oprofile/op_model_mpcore.c index 6d6bd589924..853d42bb868 100644 --- a/arch/arm/oprofile/op_model_mpcore.c +++ b/arch/arm/oprofile/op_model_mpcore.c @@ -263,7 +263,7 @@ static void em_route_irq(int irq, unsigned int cpu) const struct cpumask *mask = cpumask_of(cpu); spin_lock_irq(&desc->lock); - desc->affinity = *mask; + cpumask_copy(desc->affinity, mask); desc->chip->set_affinity(irq, mask); spin_unlock_irq(&desc->lock); } diff --git a/arch/blackfin/kernel/irqchip.c b/arch/blackfin/kernel/irqchip.c index ab8209cbbad..5780d6df154 100644 --- a/arch/blackfin/kernel/irqchip.c +++ b/arch/blackfin/kernel/irqchip.c @@ -69,6 +69,11 @@ static struct irq_desc bad_irq_desc = { #endif }; +#ifdef CONFIG_CPUMASK_OFFSTACK +/* We are not allocating a variable-sized bad_irq_desc.affinity */ +#error "Blackfin architecture does not support CONFIG_CPUMASK_OFFSTACK." +#endif + int show_interrupts(struct seq_file *p, void *v) { int i = *(loff_t *) v, j; diff --git a/arch/ia64/kernel/iosapic.c b/arch/ia64/kernel/iosapic.c index 5cfd3d91001..006ad366a45 100644 --- a/arch/ia64/kernel/iosapic.c +++ b/arch/ia64/kernel/iosapic.c @@ -880,7 +880,7 @@ iosapic_unregister_intr (unsigned int gsi) if (iosapic_intr_info[irq].count == 0) { #ifdef CONFIG_SMP /* Clear affinity */ - cpus_setall(idesc->affinity); + cpumask_setall(idesc->affinity); #endif /* Clear the interrupt information */ iosapic_intr_info[irq].dest = 0; diff --git a/arch/ia64/kernel/irq.c b/arch/ia64/kernel/irq.c index a58f64ca9f0..226233a6fa1 100644 --- a/arch/ia64/kernel/irq.c +++ b/arch/ia64/kernel/irq.c @@ -103,7 +103,7 @@ static char irq_redir [NR_IRQS]; // = { [0 ... NR_IRQS-1] = 1 }; void set_irq_affinity_info (unsigned int irq, int hwid, int redir) { if (irq < NR_IRQS) { - cpumask_copy(&irq_desc[irq].affinity, + cpumask_copy(irq_desc[irq].affinity, cpumask_of(cpu_logical_id(hwid))); irq_redir[irq] = (char) (redir & 0xff); } @@ -148,7 +148,7 @@ static void migrate_irqs(void) if (desc->status == IRQ_PER_CPU) continue; - if (cpumask_any_and(&irq_desc[irq].affinity, cpu_online_mask) + if (cpumask_any_and(irq_desc[irq].affinity, cpu_online_mask) >= nr_cpu_ids) { /* * Save it for phase 2 processing diff --git a/arch/ia64/kernel/irq_ia64.c b/arch/ia64/kernel/irq_ia64.c index 28d3d483db9..927ad027820 100644 --- a/arch/ia64/kernel/irq_ia64.c +++ b/arch/ia64/kernel/irq_ia64.c @@ -493,11 +493,13 @@ ia64_handle_irq (ia64_vector vector, struct pt_regs *regs) saved_tpr = ia64_getreg(_IA64_REG_CR_TPR); ia64_srlz_d(); while (vector != IA64_SPURIOUS_INT_VECTOR) { + struct irq_desc *desc = irq_to_desc(vector); + if (unlikely(IS_LOCAL_TLB_FLUSH(vector))) { smp_local_flush_tlb(); - kstat_this_cpu.irqs[vector]++; + kstat_incr_irqs_this_cpu(vector, desc); } else if (unlikely(IS_RESCHEDULE(vector))) - kstat_this_cpu.irqs[vector]++; + kstat_incr_irqs_this_cpu(vector, desc); else { int irq = local_vector_to_irq(vector); @@ -551,11 +553,13 @@ void ia64_process_pending_intr(void) * Perform normal interrupt style processing */ while (vector != IA64_SPURIOUS_INT_VECTOR) { + struct irq_desc *desc = irq_to_desc(vector); + if (unlikely(IS_LOCAL_TLB_FLUSH(vector))) { smp_local_flush_tlb(); - kstat_this_cpu.irqs[vector]++; + kstat_incr_irqs_this_cpu(vector, desc); } else if (unlikely(IS_RESCHEDULE(vector))) - kstat_this_cpu.irqs[vector]++; + kstat_incr_irqs_this_cpu(vector, desc); else { struct pt_regs *old_regs = set_irq_regs(NULL); int irq = local_vector_to_irq(vector); diff --git a/arch/ia64/kernel/msi_ia64.c b/arch/ia64/kernel/msi_ia64.c index 89033933903..dcb6b7c51ea 100644 --- a/arch/ia64/kernel/msi_ia64.c +++ b/arch/ia64/kernel/msi_ia64.c @@ -75,7 +75,7 @@ static void ia64_set_msi_irq_affinity(unsigned int irq, msg.data = data; write_msi_msg(irq, &msg); - irq_desc[irq].affinity = cpumask_of_cpu(cpu); + cpumask_copy(irq_desc[irq].affinity, cpumask_of(cpu)); } #endif /* CONFIG_SMP */ @@ -187,7 +187,7 @@ static void dmar_msi_set_affinity(unsigned int irq, const struct cpumask *mask) msg.address_lo |= MSI_ADDR_DESTID_CPU(cpu_physical_id(cpu)); dmar_msi_write(irq, &msg); - irq_desc[irq].affinity = *mask; + cpumask_copy(irq_desc[irq].affinity, mask); } #endif /* CONFIG_SMP */ diff --git a/arch/ia64/kernel/vmlinux.lds.S b/arch/ia64/kernel/vmlinux.lds.S index 10a7d47e851..f45e4e508ec 100644 --- a/arch/ia64/kernel/vmlinux.lds.S +++ b/arch/ia64/kernel/vmlinux.lds.S @@ -219,6 +219,7 @@ SECTIONS .data.percpu PERCPU_ADDR : AT(__phys_per_cpu_start - LOAD_OFFSET) { __per_cpu_start = .; + *(.data.percpu.page_aligned) *(.data.percpu) *(.data.percpu.shared_aligned) __per_cpu_end = .; diff --git a/arch/ia64/sn/kernel/msi_sn.c b/arch/ia64/sn/kernel/msi_sn.c index ca553b0429c..81e428943d7 100644 --- a/arch/ia64/sn/kernel/msi_sn.c +++ b/arch/ia64/sn/kernel/msi_sn.c @@ -205,7 +205,7 @@ static void sn_set_msi_irq_affinity(unsigned int irq, msg.address_lo = (u32)(bus_addr & 0x00000000ffffffff); write_msi_msg(irq, &msg); - irq_desc[irq].affinity = *cpu_mask; + cpumask_copy(irq_desc[irq].affinity, cpu_mask); } #endif /* CONFIG_SMP */ diff --git a/arch/mips/include/asm/irq.h b/arch/mips/include/asm/irq.h index abc62aa744a..3214ade02d1 100644 --- a/arch/mips/include/asm/irq.h +++ b/arch/mips/include/asm/irq.h @@ -66,7 +66,7 @@ extern void smtc_forward_irq(unsigned int irq); */ #define IRQ_AFFINITY_HOOK(irq) \ do { \ - if (!cpu_isset(smp_processor_id(), irq_desc[irq].affinity)) { \ + if (!cpumask_test_cpu(smp_processor_id(), irq_desc[irq].affinity)) {\ smtc_forward_irq(irq); \ irq_exit(); \ return; \ diff --git a/arch/mips/kernel/irq-gic.c b/arch/mips/kernel/irq-gic.c index 494a49a317e..87deb8f6c45 100644 --- a/arch/mips/kernel/irq-gic.c +++ b/arch/mips/kernel/irq-gic.c @@ -187,7 +187,7 @@ static void gic_set_affinity(unsigned int irq, const struct cpumask *cpumask) set_bit(irq, pcpu_masks[first_cpu(tmp)].pcpu_mask); } - irq_desc[irq].affinity = *cpumask; + cpumask_copy(irq_desc[irq].affinity, cpumask); spin_unlock_irqrestore(&gic_lock, flags); } diff --git a/arch/mips/kernel/smtc.c b/arch/mips/kernel/smtc.c index b6cca01ff82..5f5af7d4c89 100644 --- a/arch/mips/kernel/smtc.c +++ b/arch/mips/kernel/smtc.c @@ -686,7 +686,7 @@ void smtc_forward_irq(unsigned int irq) * and efficiency, we just pick the easiest one to find. */ - target = first_cpu(irq_desc[irq].affinity); + target = cpumask_first(irq_desc[irq].affinity); /* * We depend on the platform code to have correctly processed @@ -921,11 +921,13 @@ void ipi_decode(struct smtc_ipi *pipi) struct clock_event_device *cd; void *arg_copy = pipi->arg; int type_copy = pipi->type; + int irq = MIPS_CPU_IRQ_BASE + 1; + smtc_ipi_nq(&freeIPIq, pipi); switch (type_copy) { case SMTC_CLOCK_TICK: irq_enter(); - kstat_this_cpu.irqs[MIPS_CPU_IRQ_BASE + 1]++; + kstat_incr_irqs_this_cpu(irq, irq_to_desc(irq)); cd = &per_cpu(mips_clockevent_device, cpu); cd->event_handler(cd); irq_exit(); diff --git a/arch/mips/mti-malta/malta-smtc.c b/arch/mips/mti-malta/malta-smtc.c index aabd7274507..5ba31888fef 100644 --- a/arch/mips/mti-malta/malta-smtc.c +++ b/arch/mips/mti-malta/malta-smtc.c @@ -116,7 +116,7 @@ struct plat_smp_ops msmtc_smp_ops = { void plat_set_irq_affinity(unsigned int irq, const struct cpumask *affinity) { - cpumask_t tmask = *affinity; + cpumask_t tmask; int cpu = 0; void smtc_set_irq_affinity(unsigned int irq, cpumask_t aff); @@ -139,11 +139,12 @@ void plat_set_irq_affinity(unsigned int irq, const struct cpumask *affinity) * be made to forward to an offline "CPU". */ + cpumask_copy(&tmask, affinity); for_each_cpu(cpu, affinity) { if ((cpu_data[cpu].vpe_id != 0) || !cpu_online(cpu)) cpu_clear(cpu, tmask); } - irq_desc[irq].affinity = tmask; + cpumask_copy(irq_desc[irq].affinity, &tmask); if (cpus_empty(tmask)) /* diff --git a/arch/mips/sgi-ip22/ip22-int.c b/arch/mips/sgi-ip22/ip22-int.c index f8b18af141a..0ecd5fe9486 100644 --- a/arch/mips/sgi-ip22/ip22-int.c +++ b/arch/mips/sgi-ip22/ip22-int.c @@ -155,7 +155,7 @@ static void indy_buserror_irq(void) int irq = SGI_BUSERR_IRQ; irq_enter(); - kstat_this_cpu.irqs[irq]++; + kstat_incr_irqs_this_cpu(irq, irq_to_desc(irq)); ip22_be_interrupt(irq); irq_exit(); } diff --git a/arch/mips/sgi-ip22/ip22-time.c b/arch/mips/sgi-ip22/ip22-time.c index 3dcb27ec0c5..c8f7d2328b2 100644 --- a/arch/mips/sgi-ip22/ip22-time.c +++ b/arch/mips/sgi-ip22/ip22-time.c @@ -122,7 +122,7 @@ void indy_8254timer_irq(void) char c; irq_enter(); - kstat_this_cpu.irqs[irq]++; + kstat_incr_irqs_this_cpu(irq, irq_to_desc(irq)); printk(KERN_ALERT "Oops, got 8254 interrupt.\n"); ArcRead(0, &c, 1, &cnt); ArcEnterInteractiveMode(); diff --git a/arch/mips/sibyte/bcm1480/smp.c b/arch/mips/sibyte/bcm1480/smp.c index dddfda8e829..314691648c9 100644 --- a/arch/mips/sibyte/bcm1480/smp.c +++ b/arch/mips/sibyte/bcm1480/smp.c @@ -178,9 +178,10 @@ struct plat_smp_ops bcm1480_smp_ops = { void bcm1480_mailbox_interrupt(void) { int cpu = smp_processor_id(); + int irq = K_BCM1480_INT_MBOX_0_0; unsigned int action; - kstat_this_cpu.irqs[K_BCM1480_INT_MBOX_0_0]++; + kstat_incr_irqs_this_cpu(irq, irq_to_desc(irq)); /* Load the mailbox register to figure out what we're supposed to do */ action = (__raw_readq(mailbox_0_regs[cpu]) >> 48) & 0xffff; diff --git a/arch/mips/sibyte/sb1250/smp.c b/arch/mips/sibyte/sb1250/smp.c index 5950a288a7d..cad14003b84 100644 --- a/arch/mips/sibyte/sb1250/smp.c +++ b/arch/mips/sibyte/sb1250/smp.c @@ -166,9 +166,10 @@ struct plat_smp_ops sb_smp_ops = { void sb1250_mailbox_interrupt(void) { int cpu = smp_processor_id(); + int irq = K_INT_MBOX_0; unsigned int action; - kstat_this_cpu.irqs[K_INT_MBOX_0]++; + kstat_incr_irqs_this_cpu(irq, irq_to_desc(irq)); /* Load the mailbox register to figure out what we're supposed to do */ action = (____raw_readq(mailbox_regs[cpu]) >> 48) & 0xffff; diff --git a/arch/mn10300/kernel/mn10300-watchdog.c b/arch/mn10300/kernel/mn10300-watchdog.c index 10811e981d2..2e370d88a87 100644 --- a/arch/mn10300/kernel/mn10300-watchdog.c +++ b/arch/mn10300/kernel/mn10300-watchdog.c @@ -130,6 +130,7 @@ void watchdog_interrupt(struct pt_regs *regs, enum exception_code excep) * the stack NMI-atomically, it's safe to use smp_processor_id(). */ int sum, cpu = smp_processor_id(); + int irq = NMIIRQ; u8 wdt, tmp; wdt = WDCTR & ~WDCTR_WDCNE; @@ -138,7 +139,7 @@ void watchdog_interrupt(struct pt_regs *regs, enum exception_code excep) NMICR = NMICR_WDIF; nmi_count(cpu)++; - kstat_this_cpu.irqs[NMIIRQ]++; + kstat_incr_irqs_this_cpu(irq, irq_to_desc(irq)); sum = irq_stat[cpu].__irq_count; if (last_irq_sums[cpu] == sum) { diff --git a/arch/parisc/kernel/irq.c b/arch/parisc/kernel/irq.c index ac2c822928c..49482806863 100644 --- a/arch/parisc/kernel/irq.c +++ b/arch/parisc/kernel/irq.c @@ -120,7 +120,7 @@ int cpu_check_affinity(unsigned int irq, cpumask_t *dest) if (CHECK_IRQ_PER_CPU(irq)) { /* Bad linux design decision. The mask has already * been set; we must reset it */ - irq_desc[irq].affinity = CPU_MASK_ALL; + cpumask_setall(irq_desc[irq].affinity); return -EINVAL; } @@ -136,7 +136,7 @@ static void cpu_set_affinity_irq(unsigned int irq, const struct cpumask *dest) if (cpu_check_affinity(irq, dest)) return; - irq_desc[irq].affinity = *dest; + cpumask_copy(irq_desc[irq].affinity, dest); } #endif @@ -295,7 +295,7 @@ int txn_alloc_irq(unsigned int bits_wide) unsigned long txn_affinity_addr(unsigned int irq, int cpu) { #ifdef CONFIG_SMP - irq_desc[irq].affinity = cpumask_of_cpu(cpu); + cpumask_copy(irq_desc[irq].affinity, cpumask_of(cpu)); #endif return per_cpu(cpu_data, cpu).txn_addr; @@ -352,7 +352,7 @@ void do_cpu_irq_mask(struct pt_regs *regs) irq = eirr_to_irq(eirr_val); #ifdef CONFIG_SMP - dest = irq_desc[irq].affinity; + cpumask_copy(&dest, irq_desc[irq].affinity); if (CHECK_IRQ_PER_CPU(irq_desc[irq].status) && !cpu_isset(smp_processor_id(), dest)) { int cpu = first_cpu(dest); diff --git a/arch/powerpc/include/asm/hw_irq.h b/arch/powerpc/include/asm/hw_irq.h index f75a5fc64d2..e10f151c3db 100644 --- a/arch/powerpc/include/asm/hw_irq.h +++ b/arch/powerpc/include/asm/hw_irq.h @@ -131,5 +131,36 @@ static inline int irqs_disabled_flags(unsigned long flags) */ struct hw_interrupt_type; +#ifdef CONFIG_PERF_COUNTERS +static inline unsigned long get_perf_counter_pending(void) +{ + unsigned long x; + + asm volatile("lbz %0,%1(13)" + : "=r" (x) + : "i" (offsetof(struct paca_struct, perf_counter_pending))); + return x; +} + +static inline void set_perf_counter_pending(int x) +{ + asm volatile("stb %0,%1(13)" : : + "r" (x), + "i" (offsetof(struct paca_struct, perf_counter_pending))); +} + +extern void perf_counter_do_pending(void); + +#else + +static inline unsigned long get_perf_counter_pending(void) +{ + return 0; +} + +static inline void set_perf_counter_pending(int x) {} +static inline void perf_counter_do_pending(void) {} +#endif /* CONFIG_PERF_COUNTERS */ + #endif /* __KERNEL__ */ #endif /* _ASM_POWERPC_HW_IRQ_H */ diff --git a/arch/powerpc/include/asm/paca.h b/arch/powerpc/include/asm/paca.h index 082b3aedf14..6ef05572301 100644 --- a/arch/powerpc/include/asm/paca.h +++ b/arch/powerpc/include/asm/paca.h @@ -99,6 +99,7 @@ struct paca_struct { u8 soft_enabled; /* irq soft-enable flag */ u8 hard_enabled; /* set if irqs are enabled in MSR */ u8 io_sync; /* writel() needs spin_unlock sync */ + u8 perf_counter_pending; /* PM interrupt while soft-disabled */ /* Stuff for accurate time accounting */ u64 user_time; /* accumulated usermode TB ticks */ diff --git a/arch/powerpc/include/asm/perf_counter.h b/arch/powerpc/include/asm/perf_counter.h new file mode 100644 index 00000000000..9d7ff6d7fb5 --- /dev/null +++ b/arch/powerpc/include/asm/perf_counter.h @@ -0,0 +1,72 @@ +/* + * Performance counter support - PowerPC-specific definitions. + * + * Copyright 2008-2009 Paul Mackerras, IBM Corporation. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version + * 2 of the License, or (at your option) any later version. + */ +#include <linux/types.h> + +#define MAX_HWCOUNTERS 8 +#define MAX_EVENT_ALTERNATIVES 8 + +/* + * This struct provides the constants and functions needed to + * describe the PMU on a particular POWER-family CPU. + */ +struct power_pmu { + int n_counter; + int max_alternatives; + u64 add_fields; + u64 test_adder; + int (*compute_mmcr)(unsigned int events[], int n_ev, + unsigned int hwc[], u64 mmcr[]); + int (*get_constraint)(unsigned int event, u64 *mskp, u64 *valp); + int (*get_alternatives)(unsigned int event, unsigned int alt[]); + void (*disable_pmc)(unsigned int pmc, u64 mmcr[]); + int n_generic; + int *generic_events; +}; + +extern struct power_pmu *ppmu; + +/* + * The power_pmu.get_constraint function returns a 64-bit value and + * a 64-bit mask that express the constraints between this event and + * other events. + * + * The value and mask are divided up into (non-overlapping) bitfields + * of three different types: + * + * Select field: this expresses the constraint that some set of bits + * in MMCR* needs to be set to a specific value for this event. For a + * select field, the mask contains 1s in every bit of the field, and + * the value contains a unique value for each possible setting of the + * MMCR* bits. The constraint checking code will ensure that two events + * that set the same field in their masks have the same value in their + * value dwords. + * + * Add field: this expresses the constraint that there can be at most + * N events in a particular class. A field of k bits can be used for + * N <= 2^(k-1) - 1. The mask has the most significant bit of the field + * set (and the other bits 0), and the value has only the least significant + * bit of the field set. In addition, the 'add_fields' and 'test_adder' + * in the struct power_pmu for this processor come into play. The + * add_fields value contains 1 in the LSB of the field, and the + * test_adder contains 2^(k-1) - 1 - N in the field. + * + * NAND field: this expresses the constraint that you may not have events + * in all of a set of classes. (For example, on PPC970, you can't select + * events from the FPU, ISU and IDU simultaneously, although any two are + * possible.) For N classes, the field is N+1 bits wide, and each class + * is assigned one bit from the least-significant N bits. The mask has + * only the most-significant bit set, and the value has only the bit + * for the event's class set. The test_adder has the least significant + * bit set in the field. + * + * If an event is not subject to the constraint expressed by a particular + * field, then it will have 0 in both the mask and value for that field. + */ diff --git a/arch/powerpc/include/asm/systbl.h b/arch/powerpc/include/asm/systbl.h index 72353f6070a..4c8095f6bec 100644 --- a/arch/powerpc/include/asm/systbl.h +++ b/arch/powerpc/include/asm/systbl.h @@ -322,3 +322,4 @@ SYSCALL_SPU(epoll_create1) SYSCALL_SPU(dup3) SYSCALL_SPU(pipe2) SYSCALL(inotify_init1) +SYSCALL(perf_counter_open) diff --git a/arch/powerpc/include/asm/unistd.h b/arch/powerpc/include/asm/unistd.h index e07d0c76ed7..7cef5afe89d 100644 --- a/arch/powerpc/include/asm/unistd.h +++ b/arch/powerpc/include/asm/unistd.h @@ -341,10 +341,11 @@ #define __NR_dup3 316 #define __NR_pipe2 317 #define __NR_inotify_init1 318 +#define __NR_perf_counter_open 319 #ifdef __KERNEL__ -#define __NR_syscalls 319 +#define __NR_syscalls 320 #define __NR__exit __NR_exit #define NR_syscalls __NR_syscalls diff --git a/arch/powerpc/kernel/Makefile b/arch/powerpc/kernel/Makefile index 8d1a419df35..7c941ec3b23 100644 --- a/arch/powerpc/kernel/Makefile +++ b/arch/powerpc/kernel/Makefile @@ -94,6 +94,7 @@ obj-$(CONFIG_AUDIT) += audit.o obj64-$(CONFIG_AUDIT) += compat_audit.o obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o +obj-$(CONFIG_PERF_COUNTERS) += perf_counter.o ppc970-pmu.o power6-pmu.o obj-$(CONFIG_8XX_MINIMAL_FPEMU) += softemu8xx.o diff --git a/arch/powerpc/kernel/asm-offsets.c b/arch/powerpc/kernel/asm-offsets.c index 19ee491e9e2..3734973f739 100644 --- a/arch/powerpc/kernel/asm-offsets.c +++ b/arch/powerpc/kernel/asm-offsets.c @@ -131,6 +131,7 @@ int main(void) DEFINE(PACAKMSR, offsetof(struct paca_struct, kernel_msr)); DEFINE(PACASOFTIRQEN, offsetof(struct paca_struct, soft_enabled)); DEFINE(PACAHARDIRQEN, offsetof(struct paca_struct, hard_enabled)); + DEFINE(PACAPERFPEND, offsetof(struct paca_struct, perf_counter_pending)); DEFINE(PACASLBCACHE, offsetof(struct paca_struct, slb_cache)); DEFINE(PACASLBCACHEPTR, offsetof(struct paca_struct, slb_cache_ptr)); DEFINE(PACACONTEXTID, offsetof(struct paca_struct, context.id)); diff --git a/arch/powerpc/kernel/entry_64.S b/arch/powerpc/kernel/entry_64.S index 383ed6eb008..f30b4e553c5 100644 --- a/arch/powerpc/kernel/entry_64.S +++ b/arch/powerpc/kernel/entry_64.S @@ -526,6 +526,15 @@ ALT_FW_FTR_SECTION_END_IFCLR(FW_FEATURE_ISERIES) 2: TRACE_AND_RESTORE_IRQ(r5); +#ifdef CONFIG_PERF_COUNTERS + /* check paca->perf_counter_pending if we're enabling ints */ + lbz r3,PACAPERFPEND(r13) + and. r3,r3,r5 + beq 27f + bl .perf_counter_do_pending +27: +#endif /* CONFIG_PERF_COUNTERS */ + /* extract EE bit and use it to restore paca->hard_enabled */ ld r3,_MSR(r1) rldicl r4,r3,49,63 /* r0 = (r3 >> 15) & 1 */ diff --git a/arch/powerpc/kernel/irq.c b/arch/powerpc/kernel/irq.c index 23b8b5e36f9..7f8e6a92c5a 100644 --- a/arch/powerpc/kernel/irq.c +++ b/arch/powerpc/kernel/irq.c @@ -104,6 +104,13 @@ static inline notrace void set_soft_enabled(unsigned long enable) : : "r" (enable), "i" (offsetof(struct paca_struct, soft_enabled))); } +#ifdef CONFIG_PERF_COUNTERS +notrace void __weak perf_counter_do_pending(void) +{ + set_perf_counter_pending(0); +} +#endif + notrace void raw_local_irq_restore(unsigned long en) { /* @@ -135,6 +142,9 @@ notrace void raw_local_irq_restore(unsigned long en) iseries_handle_interrupts(); } + if (get_perf_counter_pending()) + perf_counter_do_pending(); + /* * if (get_paca()->hard_enabled) return; * But again we need to take care that gcc gets hard_enabled directly @@ -231,7 +241,7 @@ void fixup_irqs(cpumask_t map) if (irq_desc[irq].status & IRQ_PER_CPU) continue; - cpus_and(mask, irq_desc[irq].affinity, map); + cpumask_and(&mask, irq_desc[irq].affinity, &map); if (any_online_cpu(mask) == NR_CPUS) { printk("Breaking affinity for irq %i\n", irq); mask = map; diff --git a/arch/powerpc/kernel/perf_counter.c b/arch/powerpc/kernel/perf_counter.c new file mode 100644 index 00000000000..5b0211348c7 --- /dev/null +++ b/arch/powerpc/kernel/perf_counter.c @@ -0,0 +1,785 @@ +/* + * Performance counter support - powerpc architecture code + * + * Copyright 2008-2009 Paul Mackerras, IBM Corporation. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version + * 2 of the License, or (at your option) any later version. + */ +#include <linux/kernel.h> +#include <linux/sched.h> +#include <linux/perf_counter.h> +#include <linux/percpu.h> +#include <linux/hardirq.h> +#include <asm/reg.h> +#include <asm/pmc.h> +#include <asm/machdep.h> + +struct cpu_hw_counters { + int n_counters; + int n_percpu; + int disabled; + int n_added; + struct perf_counter *counter[MAX_HWCOUNTERS]; + unsigned int events[MAX_HWCOUNTERS]; + u64 mmcr[3]; + u8 pmcs_enabled; +}; +DEFINE_PER_CPU(struct cpu_hw_counters, cpu_hw_counters); + +struct power_pmu *ppmu; + +void perf_counter_print_debug(void) +{ +} + +/* + * Read one performance monitor counter (PMC). + */ +static unsigned long read_pmc(int idx) +{ + unsigned long val; + + switch (idx) { + case 1: + val = mfspr(SPRN_PMC1); + break; + case 2: + val = mfspr(SPRN_PMC2); + break; + case 3: + val = mfspr(SPRN_PMC3); + break; + case 4: + val = mfspr(SPRN_PMC4); + break; + case 5: + val = mfspr(SPRN_PMC5); + break; + case 6: + val = mfspr(SPRN_PMC6); + break; + case 7: + val = mfspr(SPRN_PMC7); + break; + case 8: + val = mfspr(SPRN_PMC8); + break; + default: + printk(KERN_ERR "oops trying to read PMC%d\n", idx); + val = 0; + } + return val; +} + +/* + * Write one PMC. + */ +static void write_pmc(int idx, unsigned long val) +{ + switch (idx) { + case 1: + mtspr(SPRN_PMC1, val); + break; + case 2: + mtspr(SPRN_PMC2, val); + break; + case 3: + mtspr(SPRN_PMC3, val); + break; + case 4: + mtspr(SPRN_PMC4, val); + break; + case 5: + mtspr(SPRN_PMC5, val); + break; + case 6: + mtspr(SPRN_PMC6, val); + break; + case 7: + mtspr(SPRN_PMC7, val); + break; + case 8: + mtspr(SPRN_PMC8, val); + break; + default: + printk(KERN_ERR "oops trying to write PMC%d\n", idx); + } +} + +/* + * Check if a set of events can all go on the PMU at once. + * If they can't, this will look at alternative codes for the events + * and see if any combination of alternative codes is feasible. + * The feasible set is returned in event[]. + */ +static int power_check_constraints(unsigned int event[], int n_ev) +{ + u64 mask, value, nv; + unsigned int alternatives[MAX_HWCOUNTERS][MAX_EVENT_ALTERNATIVES]; + u64 amasks[MAX_HWCOUNTERS][MAX_EVENT_ALTERNATIVES]; + u64 avalues[MAX_HWCOUNTERS][MAX_EVENT_ALTERNATIVES]; + u64 smasks[MAX_HWCOUNTERS], svalues[MAX_HWCOUNTERS]; + int n_alt[MAX_HWCOUNTERS], choice[MAX_HWCOUNTERS]; + int i, j; + u64 addf = ppmu->add_fields; + u64 tadd = ppmu->test_adder; + + if (n_ev > ppmu->n_counter) + return -1; + + /* First see if the events will go on as-is */ + for (i = 0; i < n_ev; ++i) { + alternatives[i][0] = event[i]; + if (ppmu->get_constraint(event[i], &amasks[i][0], + &avalues[i][0])) + return -1; + choice[i] = 0; + } + value = mask = 0; + for (i = 0; i < n_ev; ++i) { + nv = (value | avalues[i][0]) + (value & avalues[i][0] & addf); + if ((((nv + tadd) ^ value) & mask) != 0 || + (((nv + tadd) ^ avalues[i][0]) & amasks[i][0]) != 0) + break; + value = nv; + mask |= amasks[i][0]; + } + if (i == n_ev) + return 0; /* all OK */ + + /* doesn't work, gather alternatives... */ + if (!ppmu->get_alternatives) + return -1; + for (i = 0; i < n_ev; ++i) { + n_alt[i] = ppmu->get_alternatives(event[i], alternatives[i]); + for (j = 1; j < n_alt[i]; ++j) + ppmu->get_constraint(alternatives[i][j], + &amasks[i][j], &avalues[i][j]); + } + + /* enumerate all possibilities and see if any will work */ + i = 0; + j = -1; + value = mask = nv = 0; + while (i < n_ev) { + if (j >= 0) { + /* we're backtracking, restore context */ + value = svalues[i]; + mask = smasks[i]; + j = choice[i]; + } + /* + * See if any alternative k for event i, + * where k > j, will satisfy the constraints. + */ + while (++j < n_alt[i]) { + nv = (value | avalues[i][j]) + + (value & avalues[i][j] & addf); + if ((((nv + tadd) ^ value) & mask) == 0 && + (((nv + tadd) ^ avalues[i][j]) + & amasks[i][j]) == 0) + break; + } + if (j >= n_alt[i]) { + /* + * No feasible alternative, backtrack + * to event i-1 and continue enumerating its + * alternatives from where we got up to. + */ + if (--i < 0) + return -1; + } else { + /* + * Found a feasible alternative for event i, + * remember where we got up to with this event, + * go on to the next event, and start with + * the first alternative for it. + */ + choice[i] = j; + svalues[i] = value; + smasks[i] = mask; + value = nv; + mask |= amasks[i][j]; + ++i; + j = -1; + } + } + + /* OK, we have a feasible combination, tell the caller the solution */ + for (i = 0; i < n_ev; ++i) + event[i] = alternatives[i][choice[i]]; + return 0; +} + +static void power_perf_read(struct perf_counter *counter) +{ + long val, delta, prev; + + if (!counter->hw.idx) + return; + /* + * Performance monitor interrupts come even when interrupts + * are soft-disabled, as long as interrupts are hard-enabled. + * Therefore we treat them like NMIs. + */ + do { + prev = atomic64_read(&counter->hw.prev_count); + barrier(); + val = read_pmc(counter->hw.idx); + } while (atomic64_cmpxchg(&counter->hw.prev_count, prev, val) != prev); + + /* The counters are only 32 bits wide */ + delta = (val - prev) & 0xfffffffful; + atomic64_add(delta, &counter->count); + atomic64_sub(delta, &counter->hw.period_left); +} + +/* + * Disable all counters to prevent PMU interrupts and to allow + * counters to be added or removed. + */ +u64 hw_perf_save_disable(void) +{ + struct cpu_hw_counters *cpuhw; + unsigned long ret; + unsigned long flags; + + local_irq_save(flags); + cpuhw = &__get_cpu_var(cpu_hw_counters); + + ret = cpuhw->disabled; + if (!ret) { + cpuhw->disabled = 1; + cpuhw->n_added = 0; + + /* + * Check if we ever enabled the PMU on this cpu. + */ + if (!cpuhw->pmcs_enabled) { + if (ppc_md.enable_pmcs) + ppc_md.enable_pmcs(); + cpuhw->pmcs_enabled = 1; + } + + /* + * Set the 'freeze counters' bit. + * The barrier is to make sure the mtspr has been + * executed and the PMU has frozen the counters + * before we return. + */ + mtspr(SPRN_MMCR0, mfspr(SPRN_MMCR0) | MMCR0_FC); + mb(); + } + local_irq_restore(flags); + return ret; +} + +/* + * Re-enable all counters if disable == 0. + * If we were previously disabled and counters were added, then + * put the new config on the PMU. + */ +void hw_perf_restore(u64 disable) +{ + struct perf_counter *counter; + struct cpu_hw_counters *cpuhw; + unsigned long flags; + long i; + unsigned long val; + s64 left; + unsigned int hwc_index[MAX_HWCOUNTERS]; + + if (disable) + return; + local_irq_save(flags); + cpuhw = &__get_cpu_var(cpu_hw_counters); + cpuhw->disabled = 0; + + /* + * If we didn't change anything, or only removed counters, + * no need to recalculate MMCR* settings and reset the PMCs. + * Just reenable the PMU with the current MMCR* settings + * (possibly updated for removal of counters). + */ + if (!cpuhw->n_added) { + mtspr(SPRN_MMCRA, cpuhw->mmcr[2]); + mtspr(SPRN_MMCR1, cpuhw->mmcr[1]); + mtspr(SPRN_MMCR0, cpuhw->mmcr[0]); + if (cpuhw->n_counters == 0) + get_lppaca()->pmcregs_in_use = 0; + goto out; + } + + /* + * Compute MMCR* values for the new set of counters + */ + if (ppmu->compute_mmcr(cpuhw->events, cpuhw->n_counters, hwc_index, + cpuhw->mmcr)) { + /* shouldn't ever get here */ + printk(KERN_ERR "oops compute_mmcr failed\n"); + goto out; + } + + /* + * Write the new configuration to MMCR* with the freeze + * bit set and set the hardware counters to their initial values. + * Then unfreeze the counters. + */ + get_lppaca()->pmcregs_in_use = 1; + mtspr(SPRN_MMCRA, cpuhw->mmcr[2]); + mtspr(SPRN_MMCR1, cpuhw->mmcr[1]); + mtspr(SPRN_MMCR0, (cpuhw->mmcr[0] & ~(MMCR0_PMC1CE | MMCR0_PMCjCE)) + | MMCR0_FC); + + /* + * Read off any pre-existing counters that need to move + * to another PMC. + */ + for (i = 0; i < cpuhw->n_counters; ++i) { + counter = cpuhw->counter[i]; + if (counter->hw.idx && counter->hw.idx != hwc_index[i] + 1) { + power_perf_read(counter); + write_pmc(counter->hw.idx, 0); + counter->hw.idx = 0; + } + } + + /* + * Initialize the PMCs for all the new and moved counters. + */ + for (i = 0; i < cpuhw->n_counters; ++i) { + counter = cpuhw->counter[i]; + if (counter->hw.idx) + continue; + val = 0; + if (counter->hw_event.irq_period) { + left = atomic64_read(&counter->hw.period_left); + if (left < 0x80000000L) + val = 0x80000000L - left; + } + atomic64_set(&counter->hw.prev_count, val); + counter->hw.idx = hwc_index[i] + 1; + write_pmc(counter->hw.idx, val); + } + mb(); + cpuhw->mmcr[0] |= MMCR0_PMXE | MMCR0_FCECE; + mtspr(SPRN_MMCR0, cpuhw->mmcr[0]); + + out: + local_irq_restore(flags); +} + +static int collect_events(struct perf_counter *group, int max_count, + struct perf_counter *ctrs[], unsigned int *events) +{ + int n = 0; + struct perf_counter *counter; + + if (!is_software_counter(group)) { + if (n >= max_count) + return -1; + ctrs[n] = group; + events[n++] = group->hw.config; + } + list_for_each_entry(counter, &group->sibling_list, list_entry) { + if (!is_software_counter(counter) && + counter->state != PERF_COUNTER_STATE_OFF) { + if (n >= max_count) + return -1; + ctrs[n] = counter; + events[n++] = counter->hw.config; + } + } + return n; +} + +static void counter_sched_in(struct perf_counter *counter, int cpu) +{ + counter->state = PERF_COUNTER_STATE_ACTIVE; + counter->oncpu = cpu; + if (is_software_counter(counter)) + counter->hw_ops->enable(counter); +} + +/* + * Called to enable a whole group of counters. + * Returns 1 if the group was enabled, or -EAGAIN if it could not be. + * Assumes the caller has disabled interrupts and has + * frozen the PMU with hw_perf_save_disable. + */ +int hw_perf_group_sched_in(struct perf_counter *group_leader, + struct perf_cpu_context *cpuctx, + struct perf_counter_context *ctx, int cpu) +{ + struct cpu_hw_counters *cpuhw; + long i, n, n0; + struct perf_counter *sub; + + cpuhw = &__get_cpu_var(cpu_hw_counters); + n0 = cpuhw->n_counters; + n = collect_events(group_leader, ppmu->n_counter - n0, + &cpuhw->counter[n0], &cpuhw->events[n0]); + if (n < 0) + return -EAGAIN; + if (power_check_constraints(cpuhw->events, n + n0)) + return -EAGAIN; + cpuhw->n_counters = n0 + n; + cpuhw->n_added += n; + + /* + * OK, this group can go on; update counter states etc., + * and enable any software counters + */ + for (i = n0; i < n0 + n; ++i) + cpuhw->counter[i]->hw.config = cpuhw->events[i]; + cpuctx->active_oncpu += n; + n = 1; + counter_sched_in(group_leader, cpu); + list_for_each_entry(sub, &group_leader->sibling_list, list_entry) { + if (sub->state != PERF_COUNTER_STATE_OFF) { + counter_sched_in(sub, cpu); + ++n; + } + } + ctx->nr_active += n; + + return 1; +} + +/* + * Add a counter to the PMU. + * If all counters are not already frozen, then we disable and + * re-enable the PMU in order to get hw_perf_restore to do the + * actual work of reconfiguring the PMU. + */ +static int power_perf_enable(struct perf_counter *counter) +{ + struct cpu_hw_counters *cpuhw; + unsigned long flags; + u64 pmudis; + int n0; + int ret = -EAGAIN; + + local_irq_save(flags); + pmudis = hw_perf_save_disable(); + + /* + * Add the counter to the list (if there is room) + * and check whether the total set is still feasible. + */ + cpuhw = &__get_cpu_var(cpu_hw_counters); + n0 = cpuhw->n_counters; + if (n0 >= ppmu->n_counter) + goto out; + cpuhw->counter[n0] = counter; + cpuhw->events[n0] = counter->hw.config; + if (power_check_constraints(cpuhw->events, n0 + 1)) + goto out; + + counter->hw.config = cpuhw->events[n0]; + ++cpuhw->n_counters; + ++cpuhw->n_added; + + ret = 0; + out: + hw_perf_restore(pmudis); + local_irq_restore(flags); + return ret; +} + +/* + * Remove a counter from the PMU. + */ +static void power_perf_disable(struct perf_counter *counter) +{ + struct cpu_hw_counters *cpuhw; + long i; + u64 pmudis; + unsigned long flags; + + local_irq_save(flags); + pmudis = hw_perf_save_disable(); + + power_perf_read(counter); + + cpuhw = &__get_cpu_var(cpu_hw_counters); + for (i = 0; i < cpuhw->n_counters; ++i) { + if (counter == cpuhw->counter[i]) { + while (++i < cpuhw->n_counters) + cpuhw->counter[i-1] = cpuhw->counter[i]; + --cpuhw->n_counters; + ppmu->disable_pmc(counter->hw.idx - 1, cpuhw->mmcr); + write_pmc(counter->hw.idx, 0); + counter->hw.idx = 0; + break; + } + } + if (cpuhw->n_counters == 0) { + /* disable exceptions if no counters are running */ + cpuhw->mmcr[0] &= ~(MMCR0_PMXE | MMCR0_FCECE); + } + + hw_perf_restore(pmudis); + local_irq_restore(flags); +} + +struct hw_perf_counter_ops power_perf_ops = { + .enable = power_perf_enable, + .disable = power_perf_disable, + .read = power_perf_read +}; + +const struct hw_perf_counter_ops * +hw_perf_counter_init(struct perf_counter *counter) +{ + unsigned long ev; + struct perf_counter *ctrs[MAX_HWCOUNTERS]; + unsigned int events[MAX_HWCOUNTERS]; + int n; + + if (!ppmu) + return NULL; + if ((s64)counter->hw_event.irq_period < 0) + return NULL; + ev = counter->hw_event.type; + if (!counter->hw_event.raw) { + if (ev >= ppmu->n_generic || + ppmu->generic_events[ev] == 0) + return NULL; + ev = ppmu->generic_events[ev]; + } + counter->hw.config_base = ev; + counter->hw.idx = 0; + + /* + * If this is in a group, check if it can go on with all the + * other hardware counters in the group. We assume the counter + * hasn't been linked into its leader's sibling list at this point. + */ + n = 0; + if (counter->group_leader != counter) { + n = collect_events(counter->group_leader, ppmu->n_counter - 1, + ctrs, events); + if (n < 0) + return NULL; + } + events[n++] = ev; + if (power_check_constraints(events, n)) + return NULL; + + counter->hw.config = events[n - 1]; + atomic64_set(&counter->hw.period_left, counter->hw_event.irq_period); + return &power_perf_ops; +} + +/* + * Handle wakeups. + */ +void perf_counter_do_pending(void) +{ + int i; + struct cpu_hw_counters *cpuhw = &__get_cpu_var(cpu_hw_counters); + struct perf_counter *counter; + + set_perf_counter_pending(0); + for (i = 0; i < cpuhw->n_counters; ++i) { + counter = cpuhw->counter[i]; + if (counter && counter->wakeup_pending) { + counter->wakeup_pending = 0; + wake_up(&counter->waitq); + } + } +} + +/* + * Record data for an irq counter. + * This function was lifted from the x86 code; maybe it should + * go in the core? + */ +static void perf_store_irq_data(struct perf_counter *counter, u64 data) +{ + struct perf_data *irqdata = counter->irqdata; + + if (irqdata->len > PERF_DATA_BUFLEN - sizeof(u64)) { + irqdata->overrun++; + } else { + u64 *p = (u64 *) &irqdata->data[irqdata->len]; + + *p = data; + irqdata->len += sizeof(u64); + } +} + +/* + * Record all the values of the counters in a group + */ +static void perf_handle_group(struct perf_counter *counter) +{ + struct perf_counter *leader, *sub; + + leader = counter->group_leader; + list_for_each_entry(sub, &leader->sibling_list, list_entry) { + if (sub != counter) + sub->hw_ops->read(sub); + perf_store_irq_data(counter, sub->hw_event.type); + perf_store_irq_data(counter, atomic64_read(&sub->count)); + } +} + +/* + * A counter has overflowed; update its count and record + * things if requested. Note that interrupts are hard-disabled + * here so there is no possibility of being interrupted. + */ +static void record_and_restart(struct perf_counter *counter, long val, + struct pt_regs *regs) +{ + s64 prev, delta, left; + int record = 0; + + /* we don't have to worry about interrupts here */ + prev = atomic64_read(&counter->hw.prev_count); + delta = (val - prev) & 0xfffffffful; + atomic64_add(delta, &counter->count); + + /* + * See if the total period for this counter has expired, + * and update for the next period. + */ + val = 0; + left = atomic64_read(&counter->hw.period_left) - delta; + if (counter->hw_event.irq_period) { + if (left <= 0) { + left += counter->hw_event.irq_period; + if (left <= 0) + left = counter->hw_event.irq_period; + record = 1; + } + if (left < 0x80000000L) + val = 0x80000000L - left; + } + write_pmc(counter->hw.idx, val); + atomic64_set(&counter->hw.prev_count, val); + atomic64_set(&counter->hw.period_left, left); + + /* + * Finally record data if requested. + */ + if (record) { + switch (counter->hw_event.record_type) { + case PERF_RECORD_SIMPLE: + break; + case PERF_RECORD_IRQ: + perf_store_irq_data(counter, instruction_pointer(regs)); + counter->wakeup_pending = 1; + break; + case PERF_RECORD_GROUP: + perf_handle_group(counter); + counter->wakeup_pending = 1; + break; + } + } +} + +/* + * Performance monitor interrupt stuff + */ +static void perf_counter_interrupt(struct pt_regs *regs) +{ + int i; + struct cpu_hw_counters *cpuhw = &__get_cpu_var(cpu_hw_counters); + struct perf_counter *counter; + long val; + int need_wakeup = 0, found = 0; + + for (i = 0; i < cpuhw->n_counters; ++i) { + counter = cpuhw->counter[i]; + val = read_pmc(counter->hw.idx); + if ((int)val < 0) { + /* counter has overflowed */ + found = 1; + record_and_restart(counter, val, regs); + if (counter->wakeup_pending) + need_wakeup = 1; + } + } + + /* + * In case we didn't find and reset the counter that caused + * the interrupt, scan all counters and reset any that are + * negative, to avoid getting continual interrupts. + * Any that we processed in the previous loop will not be negative. + */ + if (!found) { + for (i = 0; i < ppmu->n_counter; ++i) { + val = read_pmc(i + 1); + if ((int)val < 0) + write_pmc(i + 1, 0); + } + } + + /* + * Reset MMCR0 to its normal value. This will set PMXE and + * clear FC (freeze counters) and PMAO (perf mon alert occurred) + * and thus allow interrupts to occur again. + * XXX might want to use MSR.PM to keep the counters frozen until + * we get back out of this interrupt. + */ + mtspr(SPRN_MMCR0, cpuhw->mmcr[0]); + + /* + * If we need a wakeup, check whether interrupts were soft-enabled + * when we took the interrupt. If they were, we can wake stuff up + * immediately; otherwise we'll have to set a flag and do the + * wakeup when interrupts get soft-enabled. + */ + if (need_wakeup) { + if (regs->softe) { + irq_enter(); + perf_counter_do_pending(); + irq_exit(); + } else { + set_perf_counter_pending(1); + } + } +} + +void hw_perf_counter_setup(int cpu) +{ + struct cpu_hw_counters *cpuhw = &per_cpu(cpu_hw_counters, cpu); + + memset(cpuhw, 0, sizeof(*cpuhw)); + cpuhw->mmcr[0] = MMCR0_FC; +} + +extern struct power_pmu ppc970_pmu; +extern struct power_pmu power6_pmu; + +static int init_perf_counters(void) +{ + unsigned long pvr; + + if (reserve_pmc_hardware(perf_counter_interrupt)) { + printk(KERN_ERR "Couldn't init performance monitor subsystem\n"); + return -EBUSY; + } + + /* XXX should get this from cputable */ + pvr = mfspr(SPRN_PVR); + switch (PVR_VER(pvr)) { + case PV_970: + case PV_970FX: + case PV_970MP: + ppmu = &ppc970_pmu; + break; + case 0x3e: + ppmu = &power6_pmu; + break; + } + return 0; +} + +arch_initcall(init_perf_counters); diff --git a/arch/powerpc/kernel/power6-pmu.c b/arch/powerpc/kernel/power6-pmu.c new file mode 100644 index 00000000000..b1f61f3c97b --- /dev/null +++ b/arch/powerpc/kernel/power6-pmu.c @@ -0,0 +1,283 @@ +/* + * Performance counter support for POWER6 processors. + * + * Copyright 2008-2009 Paul Mackerras, IBM Corporation. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version + * 2 of the License, or (at your option) any later version. + */ +#include <linux/kernel.h> +#include <linux/perf_counter.h> +#include <asm/reg.h> + +/* + * Bits in event code for POWER6 + */ +#define PM_PMC_SH 20 /* PMC number (1-based) for direct events */ +#define PM_PMC_MSK 0x7 +#define PM_PMC_MSKS (PM_PMC_MSK << PM_PMC_SH) +#define PM_UNIT_SH 16 /* Unit event comes (TTMxSEL encoding) */ +#define PM_UNIT_MSK 0xf +#define PM_UNIT_MSKS (PM_UNIT_MSK << PM_UNIT_SH) +#define PM_LLAV 0x8000 /* Load lookahead match value */ +#define PM_LLA 0x4000 /* Load lookahead match enable */ +#define PM_BYTE_SH 12 /* Byte of event bus to use */ +#define PM_BYTE_MSK 3 +#define PM_SUBUNIT_SH 8 /* Subunit event comes from (NEST_SEL enc.) */ +#define PM_SUBUNIT_MSK 7 +#define PM_SUBUNIT_MSKS (PM_SUBUNIT_MSK << PM_SUBUNIT_SH) +#define PM_PMCSEL_MSK 0xff /* PMCxSEL value */ +#define PM_BUSEVENT_MSK 0xf3700 + +/* + * Bits in MMCR1 for POWER6 + */ +#define MMCR1_TTM0SEL_SH 60 +#define MMCR1_TTMSEL_SH(n) (MMCR1_TTM0SEL_SH - (n) * 4) +#define MMCR1_TTMSEL_MSK 0xf +#define MMCR1_TTMSEL(m, n) (((m) >> MMCR1_TTMSEL_SH(n)) & MMCR1_TTMSEL_MSK) +#define MMCR1_NESTSEL_SH 45 +#define MMCR1_NESTSEL_MSK 0x7 +#define MMCR1_NESTSEL(m) (((m) >> MMCR1_NESTSEL_SH) & MMCR1_NESTSEL_MSK) +#define MMCR1_PMC1_LLA ((u64)1 << 44) +#define MMCR1_PMC1_LLA_VALUE ((u64)1 << 39) +#define MMCR1_PMC1_ADDR_SEL ((u64)1 << 35) +#define MMCR1_PMC1SEL_SH 24 +#define MMCR1_PMCSEL_SH(n) (MMCR1_PMC1SEL_SH - (n) * 8) +#define MMCR1_PMCSEL_MSK 0xff + +/* + * Assign PMC numbers and compute MMCR1 value for a set of events + */ +static int p6_compute_mmcr(unsigned int event[], int n_ev, + unsigned int hwc[], u64 mmcr[]) +{ + u64 mmcr1 = 0; + int i; + unsigned int pmc, ev, b, u, s, psel; + unsigned int ttmset = 0; + unsigned int pmc_inuse = 0; + + if (n_ev > 4) + return -1; + for (i = 0; i < n_ev; ++i) { + pmc = (event[i] >> PM_PMC_SH) & PM_PMC_MSK; + if (pmc) { + if (pmc_inuse & (1 << (pmc - 1))) + return -1; /* collision! */ + pmc_inuse |= 1 << (pmc - 1); + } + } + for (i = 0; i < n_ev; ++i) { + ev = event[i]; + pmc = (ev >> PM_PMC_SH) & PM_PMC_MSK; + if (pmc) { + --pmc; + } else { + /* can go on any PMC; find a free one */ + for (pmc = 0; pmc < 4; ++pmc) + if (!(pmc_inuse & (1 << pmc))) + break; + pmc_inuse |= 1 << pmc; + } + hwc[i] = pmc; + psel = ev & PM_PMCSEL_MSK; + if (ev & PM_BUSEVENT_MSK) { + /* this event uses the event bus */ + b = (ev >> PM_BYTE_SH) & PM_BYTE_MSK; + u = (ev >> PM_UNIT_SH) & PM_UNIT_MSK; + /* check for conflict on this byte of event bus */ + if ((ttmset & (1 << b)) && MMCR1_TTMSEL(mmcr1, b) != u) + return -1; + mmcr1 |= (u64)u << MMCR1_TTMSEL_SH(b); + ttmset |= 1 << b; + if (u == 5) { + /* Nest events have a further mux */ + s = (ev >> PM_SUBUNIT_SH) & PM_SUBUNIT_MSK; + if ((ttmset & 0x10) && + MMCR1_NESTSEL(mmcr1) != s) + return -1; + ttmset |= 0x10; + mmcr1 |= (u64)s << MMCR1_NESTSEL_SH; + } + if (0x30 <= psel && psel <= 0x3d) { + /* these need the PMCx_ADDR_SEL bits */ + if (b >= 2) + mmcr1 |= MMCR1_PMC1_ADDR_SEL >> pmc; + } + /* bus select values are different for PMC3/4 */ + if (pmc >= 2 && (psel & 0x90) == 0x80) + psel ^= 0x20; + } + if (ev & PM_LLA) { + mmcr1 |= MMCR1_PMC1_LLA >> pmc; + if (ev & PM_LLAV) + mmcr1 |= MMCR1_PMC1_LLA_VALUE >> pmc; + } + mmcr1 |= (u64)psel << MMCR1_PMCSEL_SH(pmc); + } + mmcr[0] = 0; + if (pmc_inuse & 1) + mmcr[0] = MMCR0_PMC1CE; + if (pmc_inuse & 0xe) + mmcr[0] |= MMCR0_PMCjCE; + mmcr[1] = mmcr1; + mmcr[2] = 0; + return 0; +} + +/* + * Layout of constraint bits: + * + * 0-1 add field: number of uses of PMC1 (max 1) + * 2-3, 4-5, 6-7: ditto for PMC2, 3, 4 + * 8-10 select field: nest (subunit) event selector + * 16-19 select field: unit on byte 0 of event bus + * 20-23, 24-27, 28-31 ditto for bytes 1, 2, 3 + */ +static int p6_get_constraint(unsigned int event, u64 *maskp, u64 *valp) +{ + int pmc, byte, sh; + unsigned int mask = 0, value = 0; + + pmc = (event >> PM_PMC_SH) & PM_PMC_MSK; + if (pmc) { + if (pmc > 4) + return -1; + sh = (pmc - 1) * 2; + mask |= 2 << sh; + value |= 1 << sh; + } + if (event & PM_BUSEVENT_MSK) { + byte = (event >> PM_BYTE_SH) & PM_BYTE_MSK; + sh = byte * 4; + mask |= PM_UNIT_MSKS << sh; + value |= (event & PM_UNIT_MSKS) << sh; + if ((event & PM_UNIT_MSKS) == (5 << PM_UNIT_SH)) { + mask |= PM_SUBUNIT_MSKS; + value |= event & PM_SUBUNIT_MSKS; + } + } + *maskp = mask; + *valp = value; + return 0; +} + +#define MAX_ALT 4 /* at most 4 alternatives for any event */ + +static const unsigned int event_alternatives[][MAX_ALT] = { + { 0x0130e8, 0x2000f6, 0x3000fc }, /* PM_PTEG_RELOAD_VALID */ + { 0x080080, 0x10000d, 0x30000c, 0x4000f0 }, /* PM_LD_MISS_L1 */ + { 0x080088, 0x200054, 0x3000f0 }, /* PM_ST_MISS_L1 */ + { 0x10000a, 0x2000f4 }, /* PM_RUN_CYC */ + { 0x10000b, 0x2000f5 }, /* PM_RUN_COUNT */ + { 0x10000e, 0x400010 }, /* PM_PURR */ + { 0x100010, 0x4000f8 }, /* PM_FLUSH */ + { 0x10001a, 0x200010 }, /* PM_MRK_INST_DISP */ + { 0x100026, 0x3000f8 }, /* PM_TB_BIT_TRANS */ + { 0x100054, 0x2000f0 }, /* PM_ST_FIN */ + { 0x100056, 0x2000fc }, /* PM_L1_ICACHE_MISS */ + { 0x1000f0, 0x40000a }, /* PM_INST_IMC_MATCH_CMPL */ + { 0x1000f8, 0x200008 }, /* PM_GCT_EMPTY_CYC */ + { 0x1000fc, 0x400006 }, /* PM_LSU_DERAT_MISS_CYC */ + { 0x20000e, 0x400007 }, /* PM_LSU_DERAT_MISS */ + { 0x200012, 0x300012 }, /* PM_INST_DISP */ + { 0x2000f2, 0x3000f2 }, /* PM_INST_DISP */ + { 0x2000f8, 0x300010 }, /* PM_EXT_INT */ + { 0x2000fe, 0x300056 }, /* PM_DATA_FROM_L2MISS */ + { 0x2d0030, 0x30001a }, /* PM_MRK_FPU_FIN */ + { 0x30000a, 0x400018 }, /* PM_MRK_INST_FIN */ + { 0x3000f6, 0x40000e }, /* PM_L1_DCACHE_RELOAD_VALID */ + { 0x3000fe, 0x400056 }, /* PM_DATA_FROM_L3MISS */ +}; + +/* + * This could be made more efficient with a binary search on + * a presorted list, if necessary + */ +static int find_alternatives_list(unsigned int event) +{ + int i, j; + unsigned int alt; + + for (i = 0; i < ARRAY_SIZE(event_alternatives); ++i) { + if (event < event_alternatives[i][0]) + return -1; + for (j = 0; j < MAX_ALT; ++j) { + alt = event_alternatives[i][j]; + if (!alt || event < alt) + break; + if (event == alt) + return i; + } + } + return -1; +} + +static int p6_get_alternatives(unsigned int event, unsigned int alt[]) +{ + int i, j; + unsigned int aevent, psel, pmc; + unsigned int nalt = 1; + + alt[0] = event; + + /* check the alternatives table */ + i = find_alternatives_list(event); + if (i >= 0) { + /* copy out alternatives from list */ + for (j = 0; j < MAX_ALT; ++j) { + aevent = event_alternatives[i][j]; + if (!aevent) + break; + if (aevent != event) + alt[nalt++] = aevent; + } + + } else { + /* Check for alternative ways of computing sum events */ + /* PMCSEL 0x32 counter N == PMCSEL 0x34 counter 5-N */ + psel = event & (PM_PMCSEL_MSK & ~1); /* ignore edge bit */ + pmc = (event >> PM_PMC_SH) & PM_PMC_MSK; + if (pmc && (psel == 0x32 || psel == 0x34)) + alt[nalt++] = ((event ^ 0x6) & ~PM_PMC_MSKS) | + ((5 - pmc) << PM_PMC_SH); + + /* PMCSEL 0x38 counter N == PMCSEL 0x3a counter N+/-2 */ + if (pmc && (psel == 0x38 || psel == 0x3a)) + alt[nalt++] = ((event ^ 0x2) & ~PM_PMC_MSKS) | + ((pmc > 2? pmc - 2: pmc + 2) << PM_PMC_SH); + } + + return nalt; +} + +static void p6_disable_pmc(unsigned int pmc, u64 mmcr[]) +{ + /* Set PMCxSEL to 0 to disable PMCx */ + mmcr[1] &= ~(0xffUL << MMCR1_PMCSEL_SH(pmc)); +} + +static int power6_generic_events[] = { + [PERF_COUNT_CPU_CYCLES] = 0x1e, + [PERF_COUNT_INSTRUCTIONS] = 2, + [PERF_COUNT_CACHE_REFERENCES] = 0x280030, /* LD_REF_L1 */ + [PERF_COUNT_CACHE_MISSES] = 0x30000c, /* LD_MISS_L1 */ + [PERF_COUNT_BRANCH_INSTRUCTIONS] = 0x410a0, /* BR_PRED */ + [PERF_COUNT_BRANCH_MISSES] = 0x400052, /* BR_MPRED */ +}; + +struct power_pmu power6_pmu = { + .n_counter = 4, + .max_alternatives = MAX_ALT, + .add_fields = 0x55, + .test_adder = 0, + .compute_mmcr = p6_compute_mmcr, + .get_constraint = p6_get_constraint, + .get_alternatives = p6_get_alternatives, + .disable_pmc = p6_disable_pmc, + .n_generic = ARRAY_SIZE(power6_generic_events), + .generic_events = power6_generic_events, +}; diff --git a/arch/powerpc/kernel/ppc970-pmu.c b/arch/powerpc/kernel/ppc970-pmu.c new file mode 100644 index 00000000000..c3256580be1 --- /dev/null +++ b/arch/powerpc/kernel/ppc970-pmu.c @@ -0,0 +1,375 @@ +/* + * Performance counter support for PPC970-family processors. + * + * Copyright 2008-2009 Paul Mackerras, IBM Corporation. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version + * 2 of the License, or (at your option) any later version. + */ +#include <linux/string.h> +#include <linux/perf_counter.h> +#include <asm/reg.h> + +/* + * Bits in event code for PPC970 + */ +#define PM_PMC_SH 12 /* PMC number (1-based) for direct events */ +#define PM_PMC_MSK 0xf +#define PM_UNIT_SH 8 /* TTMMUX number and setting - unit select */ +#define PM_UNIT_MSK 0xf +#define PM_BYTE_SH 4 /* Byte number of event bus to use */ +#define PM_BYTE_MSK 3 +#define PM_PMCSEL_MSK 0xf + +/* Values in PM_UNIT field */ +#define PM_NONE 0 +#define PM_FPU 1 +#define PM_VPU 2 +#define PM_ISU 3 +#define PM_IFU 4 +#define PM_IDU 5 +#define PM_STS 6 +#define PM_LSU0 7 +#define PM_LSU1U 8 +#define PM_LSU1L 9 +#define PM_LASTUNIT 9 + +/* + * Bits in MMCR0 for PPC970 + */ +#define MMCR0_PMC1SEL_SH 8 +#define MMCR0_PMC2SEL_SH 1 +#define MMCR_PMCSEL_MSK 0x1f + +/* + * Bits in MMCR1 for PPC970 + */ +#define MMCR1_TTM0SEL_SH 62 +#define MMCR1_TTM1SEL_SH 59 +#define MMCR1_TTM3SEL_SH 53 +#define MMCR1_TTMSEL_MSK 3 +#define MMCR1_TD_CP_DBG0SEL_SH 50 +#define MMCR1_TD_CP_DBG1SEL_SH 48 +#define MMCR1_TD_CP_DBG2SEL_SH 46 +#define MMCR1_TD_CP_DBG3SEL_SH 44 +#define MMCR1_PMC1_ADDER_SEL_SH 39 +#define MMCR1_PMC2_ADDER_SEL_SH 38 +#define MMCR1_PMC6_ADDER_SEL_SH 37 +#define MMCR1_PMC5_ADDER_SEL_SH 36 +#define MMCR1_PMC8_ADDER_SEL_SH 35 +#define MMCR1_PMC7_ADDER_SEL_SH 34 +#define MMCR1_PMC3_ADDER_SEL_SH 33 +#define MMCR1_PMC4_ADDER_SEL_SH 32 +#define MMCR1_PMC3SEL_SH 27 +#define MMCR1_PMC4SEL_SH 22 +#define MMCR1_PMC5SEL_SH 17 +#define MMCR1_PMC6SEL_SH 12 +#define MMCR1_PMC7SEL_SH 7 +#define MMCR1_PMC8SEL_SH 2 + +static short mmcr1_adder_bits[8] = { + MMCR1_PMC1_ADDER_SEL_SH, + MMCR1_PMC2_ADDER_SEL_SH, + MMCR1_PMC3_ADDER_SEL_SH, + MMCR1_PMC4_ADDER_SEL_SH, + MMCR1_PMC5_ADDER_SEL_SH, + MMCR1_PMC6_ADDER_SEL_SH, + MMCR1_PMC7_ADDER_SEL_SH, + MMCR1_PMC8_ADDER_SEL_SH +}; + +/* + * Bits in MMCRA + */ + +/* + * Layout of constraint bits: + * 6666555555555544444444443333333333222222222211111111110000000000 + * 3210987654321098765432109876543210987654321098765432109876543210 + * <><>[ >[ >[ >< >< >< >< ><><><><><><><><> + * T0T1 UC PS1 PS2 B0 B1 B2 B3 P1P2P3P4P5P6P7P8 + * + * T0 - TTM0 constraint + * 46-47: TTM0SEL value (0=FPU, 2=IFU, 3=VPU) 0xC000_0000_0000 + * + * T1 - TTM1 constraint + * 44-45: TTM1SEL value (0=IDU, 3=STS) 0x3000_0000_0000 + * + * UC - unit constraint: can't have all three of FPU|IFU|VPU, ISU, IDU|STS + * 43: UC3 error 0x0800_0000_0000 + * 42: FPU|IFU|VPU events needed 0x0400_0000_0000 + * 41: ISU events needed 0x0200_0000_0000 + * 40: IDU|STS events needed 0x0100_0000_0000 + * + * PS1 + * 39: PS1 error 0x0080_0000_0000 + * 36-38: count of events needing PMC1/2/5/6 0x0070_0000_0000 + * + * PS2 + * 35: PS2 error 0x0008_0000_0000 + * 32-34: count of events needing PMC3/4/7/8 0x0007_0000_0000 + * + * B0 + * 28-31: Byte 0 event source 0xf000_0000 + * Encoding as for the event code + * + * B1, B2, B3 + * 24-27, 20-23, 16-19: Byte 1, 2, 3 event sources + * + * P1 + * 15: P1 error 0x8000 + * 14-15: Count of events needing PMC1 + * + * P2..P8 + * 0-13: Count of events needing PMC2..PMC8 + */ + +/* Masks and values for using events from the various units */ +static u64 unit_cons[PM_LASTUNIT+1][2] = { + [PM_FPU] = { 0xc80000000000ull, 0x040000000000ull }, + [PM_VPU] = { 0xc80000000000ull, 0xc40000000000ull }, + [PM_ISU] = { 0x080000000000ull, 0x020000000000ull }, + [PM_IFU] = { 0xc80000000000ull, 0x840000000000ull }, + [PM_IDU] = { 0x380000000000ull, 0x010000000000ull }, + [PM_STS] = { 0x380000000000ull, 0x310000000000ull }, +}; + +static int p970_get_constraint(unsigned int event, u64 *maskp, u64 *valp) +{ + int pmc, byte, unit, sh; + u64 mask = 0, value = 0; + int grp = -1; + + pmc = (event >> PM_PMC_SH) & PM_PMC_MSK; + if (pmc) { + if (pmc > 8) + return -1; + sh = (pmc - 1) * 2; + mask |= 2 << sh; + value |= 1 << sh; + grp = ((pmc - 1) >> 1) & 1; + } + unit = (event >> PM_UNIT_SH) & PM_UNIT_MSK; + if (unit) { + if (unit > PM_LASTUNIT) + return -1; + mask |= unit_cons[unit][0]; + value |= unit_cons[unit][1]; + byte = (event >> PM_BYTE_SH) & PM_BYTE_MSK; + /* + * Bus events on bytes 0 and 2 can be counted + * on PMC1/2/5/6; bytes 1 and 3 on PMC3/4/7/8. + */ + if (!pmc) + grp = byte & 1; + /* Set byte lane select field */ + mask |= 0xfULL << (28 - 4 * byte); + value |= (u64)unit << (28 - 4 * byte); + } + if (grp == 0) { + /* increment PMC1/2/5/6 field */ + mask |= 0x8000000000ull; + value |= 0x1000000000ull; + } else if (grp == 1) { + /* increment PMC3/4/7/8 field */ + mask |= 0x800000000ull; + value |= 0x100000000ull; + } + *maskp = mask; + *valp = value; + return 0; +} + +static int p970_get_alternatives(unsigned int event, unsigned int alt[]) +{ + alt[0] = event; + + /* 2 alternatives for LSU empty */ + if (event == 0x2002 || event == 0x3002) { + alt[1] = event ^ 0x1000; + return 2; + } + + return 1; +} + +static int p970_compute_mmcr(unsigned int event[], int n_ev, + unsigned int hwc[], u64 mmcr[]) +{ + u64 mmcr0 = 0, mmcr1 = 0, mmcra = 0; + unsigned int pmc, unit, byte, psel; + unsigned int ttm, grp; + unsigned int pmc_inuse = 0; + unsigned int pmc_grp_use[2]; + unsigned char busbyte[4]; + unsigned char unituse[16]; + unsigned char unitmap[] = { 0, 0<<3, 3<<3, 1<<3, 2<<3, 0|4, 3|4 }; + unsigned char ttmuse[2]; + unsigned char pmcsel[8]; + int i; + + if (n_ev > 8) + return -1; + + /* First pass to count resource use */ + pmc_grp_use[0] = pmc_grp_use[1] = 0; + memset(busbyte, 0, sizeof(busbyte)); + memset(unituse, 0, sizeof(unituse)); + for (i = 0; i < n_ev; ++i) { + pmc = (event[i] >> PM_PMC_SH) & PM_PMC_MSK; + if (pmc) { + if (pmc_inuse & (1 << (pmc - 1))) + return -1; + pmc_inuse |= 1 << (pmc - 1); + /* count 1/2/5/6 vs 3/4/7/8 use */ + ++pmc_grp_use[((pmc - 1) >> 1) & 1]; + } + unit = (event[i] >> PM_UNIT_SH) & PM_UNIT_MSK; + byte = (event[i] >> PM_BYTE_SH) & PM_BYTE_MSK; + if (unit) { + if (unit > PM_LASTUNIT) + return -1; + if (!pmc) + ++pmc_grp_use[byte & 1]; + if (busbyte[byte] && busbyte[byte] != unit) + return -1; + busbyte[byte] = unit; + unituse[unit] = 1; + } + } + if (pmc_grp_use[0] > 4 || pmc_grp_use[1] > 4) + return -1; + + /* + * Assign resources and set multiplexer selects. + * + * PM_ISU can go either on TTM0 or TTM1, but that's the only + * choice we have to deal with. + */ + if (unituse[PM_ISU] & + (unituse[PM_FPU] | unituse[PM_IFU] | unituse[PM_VPU])) + unitmap[PM_ISU] = 2 | 4; /* move ISU to TTM1 */ + /* Set TTM[01]SEL fields. */ + ttmuse[0] = ttmuse[1] = 0; + for (i = PM_FPU; i <= PM_STS; ++i) { + if (!unituse[i]) + continue; + ttm = unitmap[i]; + ++ttmuse[(ttm >> 2) & 1]; + mmcr1 |= (u64)(ttm & ~4) << MMCR1_TTM1SEL_SH; + } + /* Check only one unit per TTMx */ + if (ttmuse[0] > 1 || ttmuse[1] > 1) + return -1; + + /* Set byte lane select fields and TTM3SEL. */ + for (byte = 0; byte < 4; ++byte) { + unit = busbyte[byte]; + if (!unit) + continue; + if (unit <= PM_STS) + ttm = (unitmap[unit] >> 2) & 1; + else if (unit == PM_LSU0) + ttm = 2; + else { + ttm = 3; + if (unit == PM_LSU1L && byte >= 2) + mmcr1 |= 1ull << (MMCR1_TTM3SEL_SH + 3 - byte); + } + mmcr1 |= (u64)ttm << (MMCR1_TD_CP_DBG0SEL_SH - 2 * byte); + } + + /* Second pass: assign PMCs, set PMCxSEL and PMCx_ADDER_SEL fields */ + memset(pmcsel, 0x8, sizeof(pmcsel)); /* 8 means don't count */ + for (i = 0; i < n_ev; ++i) { + pmc = (event[i] >> PM_PMC_SH) & PM_PMC_MSK; + unit = (event[i] >> PM_UNIT_SH) & PM_UNIT_MSK; + byte = (event[i] >> PM_BYTE_SH) & PM_BYTE_MSK; + psel = event[i] & PM_PMCSEL_MSK; + if (!pmc) { + /* Bus event or any-PMC direct event */ + if (unit) + psel |= 0x10 | ((byte & 2) << 2); + else + psel |= 8; + for (pmc = 0; pmc < 8; ++pmc) { + if (pmc_inuse & (1 << pmc)) + continue; + grp = (pmc >> 1) & 1; + if (unit) { + if (grp == (byte & 1)) + break; + } else if (pmc_grp_use[grp] < 4) { + ++pmc_grp_use[grp]; + break; + } + } + pmc_inuse |= 1 << pmc; + } else { + /* Direct event */ + --pmc; + if (psel == 0 && (byte & 2)) + /* add events on higher-numbered bus */ + mmcr1 |= 1ull << mmcr1_adder_bits[pmc]; + } + pmcsel[pmc] = psel; + hwc[i] = pmc; + } + for (pmc = 0; pmc < 2; ++pmc) + mmcr0 |= pmcsel[pmc] << (MMCR0_PMC1SEL_SH - 7 * pmc); + for (; pmc < 8; ++pmc) + mmcr1 |= (u64)pmcsel[pmc] << (MMCR1_PMC3SEL_SH - 5 * (pmc - 2)); + if (pmc_inuse & 1) + mmcr0 |= MMCR0_PMC1CE; + if (pmc_inuse & 0xfe) + mmcr0 |= MMCR0_PMCjCE; + + mmcra |= 0x2000; /* mark only one IOP per PPC instruction */ + + /* Return MMCRx values */ + mmcr[0] = mmcr0; + mmcr[1] = mmcr1; + mmcr[2] = mmcra; + return 0; +} + +static void p970_disable_pmc(unsigned int pmc, u64 mmcr[]) +{ + int shift, i; + + if (pmc <= 1) { + shift = MMCR0_PMC1SEL_SH - 7 * pmc; + i = 0; + } else { + shift = MMCR1_PMC3SEL_SH - 5 * (pmc - 2); + i = 1; + } + /* + * Setting the PMCxSEL field to 0x08 disables PMC x. + */ + mmcr[i] = (mmcr[i] & ~(0x1fUL << shift)) | (0x08UL << shift); +} + +static int ppc970_generic_events[] = { + [PERF_COUNT_CPU_CYCLES] = 7, + [PERF_COUNT_INSTRUCTIONS] = 1, + [PERF_COUNT_CACHE_REFERENCES] = 0x8810, /* PM_LD_REF_L1 */ + [PERF_COUNT_CACHE_MISSES] = 0x3810, /* PM_LD_MISS_L1 */ + [PERF_COUNT_BRANCH_INSTRUCTIONS] = 0x431, /* PM_BR_ISSUED */ + [PERF_COUNT_BRANCH_MISSES] = 0x327, /* PM_GRP_BR_MPRED */ +}; + +struct power_pmu ppc970_pmu = { + .n_counter = 8, + .max_alternatives = 2, + .add_fields = 0x001100005555ull, + .test_adder = 0x013300000000ull, + .compute_mmcr = p970_compute_mmcr, + .get_constraint = p970_get_constraint, + .get_alternatives = p970_get_alternatives, + .disable_pmc = p970_disable_pmc, + .n_generic = ARRAY_SIZE(ppc970_generic_events), + .generic_events = ppc970_generic_events, +}; diff --git a/arch/powerpc/kernel/vmlinux.lds.S b/arch/powerpc/kernel/vmlinux.lds.S index 161b9b9691f..295ccc5e86b 100644 --- a/arch/powerpc/kernel/vmlinux.lds.S +++ b/arch/powerpc/kernel/vmlinux.lds.S @@ -184,6 +184,7 @@ SECTIONS . = ALIGN(PAGE_SIZE); .data.percpu : AT(ADDR(.data.percpu) - LOAD_OFFSET) { __per_cpu_start = .; + *(.data.percpu.page_aligned) *(.data.percpu) *(.data.percpu.shared_aligned) __per_cpu_end = .; diff --git a/arch/powerpc/platforms/Kconfig.cputype b/arch/powerpc/platforms/Kconfig.cputype index e868b5c5072..dc0f3c93351 100644 --- a/arch/powerpc/platforms/Kconfig.cputype +++ b/arch/powerpc/platforms/Kconfig.cputype @@ -1,6 +1,7 @@ config PPC64 bool "64-bit kernel" default n + select HAVE_PERF_COUNTERS help This option selects whether a 32-bit or a 64-bit kernel will be built. diff --git a/arch/powerpc/platforms/pseries/xics.c b/arch/powerpc/platforms/pseries/xics.c index 84e058f1e1c..80b513449f4 100644 --- a/arch/powerpc/platforms/pseries/xics.c +++ b/arch/powerpc/platforms/pseries/xics.c @@ -153,9 +153,10 @@ static int get_irq_server(unsigned int virq, unsigned int strict_check) { int server; /* For the moment only implement delivery to all cpus or one cpu */ - cpumask_t cpumask = irq_desc[virq].affinity; + cpumask_t cpumask; cpumask_t tmp = CPU_MASK_NONE; + cpumask_copy(&cpumask, irq_desc[virq].affinity); if (!distribute_irqs) return default_server; @@ -869,7 +870,7 @@ void xics_migrate_irqs_away(void) virq, cpu); /* Reset affinity to all cpus */ - irq_desc[virq].affinity = CPU_MASK_ALL; + cpumask_setall(irq_desc[virq].affinity); desc->chip->set_affinity(virq, cpu_all_mask); unlock: spin_unlock_irqrestore(&desc->lock, flags); diff --git a/arch/powerpc/sysdev/mpic.c b/arch/powerpc/sysdev/mpic.c index a35297dbac2..532e205303a 100644 --- a/arch/powerpc/sysdev/mpic.c +++ b/arch/powerpc/sysdev/mpic.c @@ -566,9 +566,10 @@ static void __init mpic_scan_ht_pics(struct mpic *mpic) #ifdef CONFIG_SMP static int irq_choose_cpu(unsigned int virt_irq) { - cpumask_t mask = irq_desc[virt_irq].affinity; + cpumask_t mask; int cpuid; + cpumask_copy(&mask, irq_desc[virt_irq].affinity); if (cpus_equal(mask, CPU_MASK_ALL)) { static int irq_rover; static DEFINE_SPINLOCK(irq_rover_lock); diff --git a/arch/sparc/kernel/irq_64.c b/arch/sparc/kernel/irq_64.c index cab8e028687..4ac5c651e00 100644 --- a/arch/sparc/kernel/irq_64.c +++ b/arch/sparc/kernel/irq_64.c @@ -247,9 +247,10 @@ struct irq_handler_data { #ifdef CONFIG_SMP static int irq_choose_cpu(unsigned int virt_irq) { - cpumask_t mask = irq_desc[virt_irq].affinity; + cpumask_t mask; int cpuid; + cpumask_copy(&mask, irq_desc[virt_irq].affinity); if (cpus_equal(mask, CPU_MASK_ALL)) { static int irq_rover; static DEFINE_SPINLOCK(irq_rover_lock); @@ -854,7 +855,7 @@ void fixup_irqs(void) !(irq_desc[irq].status & IRQ_PER_CPU)) { if (irq_desc[irq].chip->set_affinity) irq_desc[irq].chip->set_affinity(irq, - &irq_desc[irq].affinity); + irq_desc[irq].affinity); } spin_unlock_irqrestore(&irq_desc[irq].lock, flags); } diff --git a/arch/sparc/kernel/time_64.c b/arch/sparc/kernel/time_64.c index 2db3c2229b9..db310aa0018 100644 --- a/arch/sparc/kernel/time_64.c +++ b/arch/sparc/kernel/time_64.c @@ -729,7 +729,7 @@ void timer_interrupt(int irq, struct pt_regs *regs) irq_enter(); - kstat_this_cpu.irqs[0]++; + kstat_incr_irqs_this_cpu(0, irq_to_desc(0)); if (unlikely(!evt->event_handler)) { printk(KERN_WARNING diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 73f7fe8fd4d..1f484450576 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -685,6 +685,7 @@ config X86_UP_IOAPIC config X86_LOCAL_APIC def_bool y depends on X86_64 || (X86_32 && (X86_UP_APIC || (SMP && !X86_VOYAGER) || X86_GENERICARCH)) + select HAVE_PERF_COUNTERS if (!M386 && !M486) config X86_IO_APIC def_bool y diff --git a/arch/x86/ia32/ia32entry.S b/arch/x86/ia32/ia32entry.S index 256b00b6189..01e7c4c5c7f 100644 --- a/arch/x86/ia32/ia32entry.S +++ b/arch/x86/ia32/ia32entry.S @@ -112,8 +112,8 @@ ENTRY(ia32_sysenter_target) CFI_DEF_CFA rsp,0 CFI_REGISTER rsp,rbp SWAPGS_UNSAFE_STACK - movq %gs:pda_kernelstack, %rsp - addq $(PDA_STACKOFFSET),%rsp + movq PER_CPU_VAR(kernel_stack), %rsp + addq $(KERNEL_STACK_OFFSET),%rsp /* * No need to follow this irqs on/off section: the syscall * disabled irqs, here we enable it straight after entry: @@ -273,13 +273,13 @@ ENDPROC(ia32_sysenter_target) ENTRY(ia32_cstar_target) CFI_STARTPROC32 simple CFI_SIGNAL_FRAME - CFI_DEF_CFA rsp,PDA_STACKOFFSET + CFI_DEF_CFA rsp,KERNEL_STACK_OFFSET CFI_REGISTER rip,rcx /*CFI_REGISTER rflags,r11*/ SWAPGS_UNSAFE_STACK movl %esp,%r8d CFI_REGISTER rsp,r8 - movq %gs:pda_kernelstack,%rsp + movq PER_CPU_VAR(kernel_stack),%rsp /* * No need to follow this irqs on/off section: the syscall * disabled irqs and here we enable it straight after entry: @@ -823,7 +823,8 @@ ia32_sys_call_table: .quad compat_sys_signalfd4 .quad sys_eventfd2 .quad sys_epoll_create1 - .quad sys_dup3 /* 330 */ + .quad sys_dup3 /* 330 */ .quad sys_pipe2 .quad sys_inotify_init1 + .quad sys_perf_counter_open ia32_syscall_end: diff --git a/arch/x86/include/asm/apicnum.h b/arch/x86/include/asm/apicnum.h new file mode 100644 index 00000000000..82f613c607c --- /dev/null +++ b/arch/x86/include/asm/apicnum.h @@ -0,0 +1,12 @@ +#ifndef _ASM_X86_APICNUM_H +#define _ASM_X86_APICNUM_H + +/* define MAX_IO_APICS */ +#ifdef CONFIG_X86_32 +# define MAX_IO_APICS 64 +#else +# define MAX_IO_APICS 128 +# define MAX_LOCAL_APIC 32768 +#endif + +#endif /* _ASM_X86_APICNUM_H */ diff --git a/arch/x86/include/asm/atomic_32.h b/arch/x86/include/asm/atomic_32.h index 85b46fba422..977250ed8b8 100644 --- a/arch/x86/include/asm/atomic_32.h +++ b/arch/x86/include/asm/atomic_32.h @@ -247,5 +247,223 @@ static inline int atomic_add_unless(atomic_t *v, int a, int u) #define smp_mb__before_atomic_inc() barrier() #define smp_mb__after_atomic_inc() barrier() +/* An 64bit atomic type */ + +typedef struct { + unsigned long long counter; +} atomic64_t; + +#define ATOMIC64_INIT(val) { (val) } + +/** + * atomic64_read - read atomic64 variable + * @v: pointer of type atomic64_t + * + * Atomically reads the value of @v. + * Doesn't imply a read memory barrier. + */ +#define __atomic64_read(ptr) ((ptr)->counter) + +static inline unsigned long long +cmpxchg8b(unsigned long long *ptr, unsigned long long old, unsigned long long new) +{ + asm volatile( + + LOCK_PREFIX "cmpxchg8b (%[ptr])\n" + + : "=A" (old) + + : [ptr] "D" (ptr), + "A" (old), + "b" (ll_low(new)), + "c" (ll_high(new)) + + : "memory"); + + return old; +} + +static inline unsigned long long +atomic64_cmpxchg(atomic64_t *ptr, unsigned long long old_val, + unsigned long long new_val) +{ + return cmpxchg8b(&ptr->counter, old_val, new_val); +} + +/** + * atomic64_set - set atomic64 variable + * @ptr: pointer to type atomic64_t + * @new_val: value to assign + * + * Atomically sets the value of @ptr to @new_val. + */ +static inline void atomic64_set(atomic64_t *ptr, unsigned long long new_val) +{ + unsigned long long old_val; + + do { + old_val = atomic_read(ptr); + } while (atomic64_cmpxchg(ptr, old_val, new_val) != old_val); +} + +/** + * atomic64_read - read atomic64 variable + * @ptr: pointer to type atomic64_t + * + * Atomically reads the value of @ptr and returns it. + */ +static inline unsigned long long atomic64_read(atomic64_t *ptr) +{ + unsigned long long curr_val; + + do { + curr_val = __atomic64_read(ptr); + } while (atomic64_cmpxchg(ptr, curr_val, curr_val) != curr_val); + + return curr_val; +} + +/** + * atomic64_add_return - add and return + * @delta: integer value to add + * @ptr: pointer to type atomic64_t + * + * Atomically adds @delta to @ptr and returns @delta + *@ptr + */ +static inline unsigned long long +atomic64_add_return(unsigned long long delta, atomic64_t *ptr) +{ + unsigned long long old_val, new_val; + + do { + old_val = atomic_read(ptr); + new_val = old_val + delta; + + } while (atomic64_cmpxchg(ptr, old_val, new_val) != old_val); + + return new_val; +} + +static inline long atomic64_sub_return(unsigned long long delta, atomic64_t *ptr) +{ + return atomic64_add_return(-delta, ptr); +} + +static inline long atomic64_inc_return(atomic64_t *ptr) +{ + return atomic64_add_return(1, ptr); +} + +static inline long atomic64_dec_return(atomic64_t *ptr) +{ + return atomic64_sub_return(1, ptr); +} + +/** + * atomic64_add - add integer to atomic64 variable + * @delta: integer value to add + * @ptr: pointer to type atomic64_t + * + * Atomically adds @delta to @ptr. + */ +static inline void atomic64_add(unsigned long long delta, atomic64_t *ptr) +{ + atomic64_add_return(delta, ptr); +} + +/** + * atomic64_sub - subtract the atomic64 variable + * @delta: integer value to subtract + * @ptr: pointer to type atomic64_t + * + * Atomically subtracts @delta from @ptr. + */ +static inline void atomic64_sub(unsigned long long delta, atomic64_t *ptr) +{ + atomic64_add(-delta, ptr); +} + +/** + * atomic64_sub_and_test - subtract value from variable and test result + * @delta: integer value to subtract + * @ptr: pointer to type atomic64_t + * + * Atomically subtracts @delta from @ptr and returns + * true if the result is zero, or false for all + * other cases. + */ +static inline int +atomic64_sub_and_test(unsigned long long delta, atomic64_t *ptr) +{ + unsigned long long old_val = atomic64_sub_return(delta, ptr); + + return old_val == 0; +} + +/** + * atomic64_inc - increment atomic64 variable + * @ptr: pointer to type atomic64_t + * + * Atomically increments @ptr by 1. + */ +static inline void atomic64_inc(atomic64_t *ptr) +{ + atomic64_add(1, ptr); +} + +/** + * atomic64_dec - decrement atomic64 variable + * @ptr: pointer to type atomic64_t + * + * Atomically decrements @ptr by 1. + */ +static inline void atomic64_dec(atomic64_t *ptr) +{ + atomic64_sub(1, ptr); +} + +/** + * atomic64_dec_and_test - decrement and test + * @ptr: pointer to type atomic64_t + * + * Atomically decrements @ptr by 1 and + * returns true if the result is 0, or false for all other + * cases. + */ +static inline int atomic64_dec_and_test(atomic64_t *ptr) +{ + return atomic64_sub_and_test(1, ptr); +} + +/** + * atomic64_inc_and_test - increment and test + * @ptr: pointer to type atomic64_t + * + * Atomically increments @ptr by 1 + * and returns true if the result is zero, or false for all + * other cases. + */ +static inline int atomic64_inc_and_test(atomic64_t *ptr) +{ + return atomic64_sub_and_test(-1, ptr); +} + +/** + * atomic64_add_negative - add and test if negative + * @delta: integer value to add + * @ptr: pointer to type atomic64_t + * + * Atomically adds @delta to @ptr and returns true + * if the result is negative, or false when + * result is greater than or equal to zero. + */ +static inline int +atomic64_add_negative(unsigned long long delta, atomic64_t *ptr) +{ + long long old_val = atomic64_add_return(delta, ptr); + + return old_val < 0; +} + #include <asm-generic/atomic.h> #endif /* _ASM_X86_ATOMIC_32_H */ diff --git a/arch/x86/include/asm/bitops.h b/arch/x86/include/asm/bitops.h index e02a359d2aa..02b47a603fc 100644 --- a/arch/x86/include/asm/bitops.h +++ b/arch/x86/include/asm/bitops.h @@ -3,6 +3,9 @@ /* * Copyright 1992, Linus Torvalds. + * + * Note: inlines with more than a single statement should be marked + * __always_inline to avoid problems with older gcc's inlining heuristics. */ #ifndef _LINUX_BITOPS_H @@ -53,7 +56,8 @@ * Note that @nr may be almost arbitrarily large; this function is not * restricted to acting on a single-word quantity. */ -static inline void set_bit(unsigned int nr, volatile unsigned long *addr) +static __always_inline void +set_bit(unsigned int nr, volatile unsigned long *addr) { if (IS_IMMEDIATE(nr)) { asm volatile(LOCK_PREFIX "orb %1,%0" @@ -90,7 +94,8 @@ static inline void __set_bit(int nr, volatile unsigned long *addr) * you should call smp_mb__before_clear_bit() and/or smp_mb__after_clear_bit() * in order to ensure changes are visible on other processors. */ -static inline void clear_bit(int nr, volatile unsigned long *addr) +static __always_inline void +clear_bit(int nr, volatile unsigned long *addr) { if (IS_IMMEDIATE(nr)) { asm volatile(LOCK_PREFIX "andb %1,%0" @@ -204,7 +209,8 @@ static inline int test_and_set_bit(int nr, volatile unsigned long *addr) * * This is the same as test_and_set_bit on x86. */ -static inline int test_and_set_bit_lock(int nr, volatile unsigned long *addr) +static __always_inline int +test_and_set_bit_lock(int nr, volatile unsigned long *addr) { return test_and_set_bit(nr, addr); } @@ -300,7 +306,7 @@ static inline int test_and_change_bit(int nr, volatile unsigned long *addr) return oldbit; } -static inline int constant_test_bit(unsigned int nr, const volatile unsigned long *addr) +static __always_inline int constant_test_bit(unsigned int nr, const volatile unsigned long *addr) { return ((1UL << (nr % BITS_PER_LONG)) & (((unsigned long *)addr)[nr / BITS_PER_LONG])) != 0; diff --git a/arch/x86/include/asm/cpu.h b/arch/x86/include/asm/cpu.h index bae482df603..f03b23e3286 100644 --- a/arch/x86/include/asm/cpu.h +++ b/arch/x86/include/asm/cpu.h @@ -7,6 +7,20 @@ #include <linux/nodemask.h> #include <linux/percpu.h> +#ifdef CONFIG_SMP + +extern void prefill_possible_map(void); + +#else /* CONFIG_SMP */ + +static inline void prefill_possible_map(void) {} + +#define cpu_physical_id(cpu) boot_cpu_physical_apicid +#define safe_smp_processor_id() 0 +#define stack_smp_processor_id() 0 + +#endif /* CONFIG_SMP */ + struct x86_cpu { struct cpu cpu; }; @@ -17,4 +31,11 @@ extern void arch_unregister_cpu(int); #endif DECLARE_PER_CPU(int, cpu_state); + +#ifdef CONFIG_X86_HAS_BOOT_CPU_ID +extern unsigned char boot_cpu_id; +#else +#define boot_cpu_id 0 +#endif + #endif /* _ASM_X86_CPU_H */ diff --git a/arch/x86/include/asm/cpumask.h b/arch/x86/include/asm/cpumask.h new file mode 100644 index 00000000000..26c6dad9047 --- /dev/null +++ b/arch/x86/include/asm/cpumask.h @@ -0,0 +1,28 @@ +#ifndef _ASM_X86_CPUMASK_H +#define _ASM_X86_CPUMASK_H +#ifndef __ASSEMBLY__ +#include <linux/cpumask.h> + +#ifdef CONFIG_X86_64 + +extern cpumask_var_t cpu_callin_mask; +extern cpumask_var_t cpu_callout_mask; +extern cpumask_var_t cpu_initialized_mask; +extern cpumask_var_t cpu_sibling_setup_mask; + +#else /* CONFIG_X86_32 */ + +extern cpumask_t cpu_callin_map; +extern cpumask_t cpu_callout_map; +extern cpumask_t cpu_initialized; +extern cpumask_t cpu_sibling_setup_map; + +#define cpu_callin_mask ((struct cpumask *)&cpu_callin_map) +#define cpu_callout_mask ((struct cpumask *)&cpu_callout_map) +#define cpu_initialized_mask ((struct cpumask *)&cpu_initialized) +#define cpu_sibling_setup_mask ((struct cpumask *)&cpu_sibling_setup_map) + +#endif /* CONFIG_X86_32 */ + +#endif /* __ASSEMBLY__ */ +#endif /* _ASM_X86_CPUMASK_H */ diff --git a/arch/x86/include/asm/current.h b/arch/x86/include/asm/current.h index 0930b4f8d67..c68c361697e 100644 --- a/arch/x86/include/asm/current.h +++ b/arch/x86/include/asm/current.h @@ -1,39 +1,21 @@ #ifndef _ASM_X86_CURRENT_H #define _ASM_X86_CURRENT_H -#ifdef CONFIG_X86_32 #include <linux/compiler.h> #include <asm/percpu.h> +#ifndef __ASSEMBLY__ struct task_struct; DECLARE_PER_CPU(struct task_struct *, current_task); -static __always_inline struct task_struct *get_current(void) -{ - return x86_read_percpu(current_task); -} - -#else /* X86_32 */ - -#ifndef __ASSEMBLY__ -#include <asm/pda.h> - -struct task_struct; static __always_inline struct task_struct *get_current(void) { - return read_pda(pcurrent); + return percpu_read(current_task); } -#else /* __ASSEMBLY__ */ - -#include <asm/asm-offsets.h> -#define GET_CURRENT(reg) movq %gs:(pda_pcurrent),reg +#define current get_current() #endif /* __ASSEMBLY__ */ -#endif /* X86_32 */ - -#define current get_current() - #endif /* _ASM_X86_CURRENT_H */ diff --git a/arch/x86/include/asm/hardirq_32.h b/arch/x86/include/asm/hardirq_32.h index cf7954d1405..7838276bfe5 100644 --- a/arch/x86/include/asm/hardirq_32.h +++ b/arch/x86/include/asm/hardirq_32.h @@ -9,6 +9,7 @@ typedef struct { unsigned long idle_timestamp; unsigned int __nmi_count; /* arch dependent */ unsigned int apic_timer_irqs; /* arch dependent */ + unsigned int apic_perf_irqs; /* arch dependent */ unsigned int irq0_irqs; unsigned int irq_resched_count; unsigned int irq_call_count; @@ -19,6 +20,9 @@ typedef struct { DECLARE_PER_CPU(irq_cpustat_t, irq_stat); +/* We can have at most NR_VECTORS irqs routed to a cpu at a time */ +#define MAX_HARDIRQS_PER_CPU NR_VECTORS + #define __ARCH_IRQ_STAT #define __IRQ_STAT(cpu, member) (per_cpu(irq_stat, cpu).member) diff --git a/arch/x86/include/asm/hardirq_64.h b/arch/x86/include/asm/hardirq_64.h index b5a6b5d5670..42930b27921 100644 --- a/arch/x86/include/asm/hardirq_64.h +++ b/arch/x86/include/asm/hardirq_64.h @@ -3,22 +3,37 @@ #include <linux/threads.h> #include <linux/irq.h> -#include <asm/pda.h> #include <asm/apic.h> +typedef struct { + unsigned int __softirq_pending; + unsigned int __nmi_count; /* arch dependent */ + unsigned int apic_timer_irqs; /* arch dependent */ + unsigned int apic_perf_irqs; /* arch dependent */ + unsigned int irq0_irqs; + unsigned int irq_resched_count; + unsigned int irq_call_count; + unsigned int irq_tlb_count; + unsigned int irq_thermal_count; + unsigned int irq_spurious_count; + unsigned int irq_threshold_count; +} ____cacheline_aligned irq_cpustat_t; + +DECLARE_PER_CPU(irq_cpustat_t, irq_stat); + /* We can have at most NR_VECTORS irqs routed to a cpu at a time */ #define MAX_HARDIRQS_PER_CPU NR_VECTORS #define __ARCH_IRQ_STAT 1 -#define inc_irq_stat(member) add_pda(member, 1) +#define inc_irq_stat(member) percpu_add(irq_stat.member, 1) -#define local_softirq_pending() read_pda(__softirq_pending) +#define local_softirq_pending() percpu_read(irq_stat.__softirq_pending) #define __ARCH_SET_SOFTIRQ_PENDING 1 -#define set_softirq_pending(x) write_pda(__softirq_pending, (x)) -#define or_softirq_pending(x) or_pda(__softirq_pending, (x)) +#define set_softirq_pending(x) percpu_write(irq_stat.__softirq_pending, (x)) +#define or_softirq_pending(x) percpu_or(irq_stat.__softirq_pending, (x)) extern void ack_bad_irq(unsigned int irq); diff --git a/arch/x86/include/asm/hw_irq.h b/arch/x86/include/asm/hw_irq.h index 8de644b6b95..aa93e53b85e 100644 --- a/arch/x86/include/asm/hw_irq.h +++ b/arch/x86/include/asm/hw_irq.h @@ -30,6 +30,8 @@ /* Interrupt handlers registered during init_IRQ */ extern void apic_timer_interrupt(void); extern void error_interrupt(void); +extern void perf_counter_interrupt(void); + extern void spurious_interrupt(void); extern void thermal_interrupt(void); extern void reschedule_interrupt(void); diff --git a/arch/x86/include/asm/intel_arch_perfmon.h b/arch/x86/include/asm/intel_arch_perfmon.h deleted file mode 100644 index fa0fd068bc2..00000000000 --- a/arch/x86/include/asm/intel_arch_perfmon.h +++ /dev/null @@ -1,31 +0,0 @@ -#ifndef _ASM_X86_INTEL_ARCH_PERFMON_H -#define _ASM_X86_INTEL_ARCH_PERFMON_H - -#define MSR_ARCH_PERFMON_PERFCTR0 0xc1 -#define MSR_ARCH_PERFMON_PERFCTR1 0xc2 - -#define MSR_ARCH_PERFMON_EVENTSEL0 0x186 -#define MSR_ARCH_PERFMON_EVENTSEL1 0x187 - -#define ARCH_PERFMON_EVENTSEL0_ENABLE (1 << 22) -#define ARCH_PERFMON_EVENTSEL_INT (1 << 20) -#define ARCH_PERFMON_EVENTSEL_OS (1 << 17) -#define ARCH_PERFMON_EVENTSEL_USR (1 << 16) - -#define ARCH_PERFMON_UNHALTED_CORE_CYCLES_SEL (0x3c) -#define ARCH_PERFMON_UNHALTED_CORE_CYCLES_UMASK (0x00 << 8) -#define ARCH_PERFMON_UNHALTED_CORE_CYCLES_INDEX (0) -#define ARCH_PERFMON_UNHALTED_CORE_CYCLES_PRESENT \ - (1 << (ARCH_PERFMON_UNHALTED_CORE_CYCLES_INDEX)) - -union cpuid10_eax { - struct { - unsigned int version_id:8; - unsigned int num_counters:8; - unsigned int bit_width:8; - unsigned int mask_length:8; - } split; - unsigned int full; -}; - -#endif /* _ASM_X86_INTEL_ARCH_PERFMON_H */ diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h index 7a1f44ac1f1..08ec793aa04 100644 --- a/arch/x86/include/asm/io_apic.h +++ b/arch/x86/include/asm/io_apic.h @@ -114,38 +114,16 @@ struct IR_IO_APIC_route_entry { extern int nr_ioapics; extern int nr_ioapic_registers[MAX_IO_APICS]; -/* - * MP-BIOS irq configuration table structures: - */ - #define MP_MAX_IOAPIC_PIN 127 -struct mp_config_ioapic { - unsigned long mp_apicaddr; - unsigned int mp_apicid; - unsigned char mp_type; - unsigned char mp_apicver; - unsigned char mp_flags; -}; - -struct mp_config_intsrc { - unsigned int mp_dstapic; - unsigned char mp_type; - unsigned char mp_irqtype; - unsigned short mp_irqflag; - unsigned char mp_srcbus; - unsigned char mp_srcbusirq; - unsigned char mp_dstirq; -}; - /* I/O APIC entries */ -extern struct mp_config_ioapic mp_ioapics[MAX_IO_APICS]; +extern struct mpc_ioapic mp_ioapics[MAX_IO_APICS]; /* # of MP IRQ source entries */ extern int mp_irq_entries; /* MP IRQ source entries */ -extern struct mp_config_intsrc mp_irqs[MAX_IRQ_SOURCES]; +extern struct mpc_intsrc mp_irqs[MAX_IRQ_SOURCES]; /* non-0 if default (table-less) MP configuration */ extern int mpc_default_type; diff --git a/arch/x86/include/asm/irq_regs_32.h b/arch/x86/include/asm/irq_regs_32.h index 86afd747345..d7ed33ee94e 100644 --- a/arch/x86/include/asm/irq_regs_32.h +++ b/arch/x86/include/asm/irq_regs_32.h @@ -15,7 +15,7 @@ DECLARE_PER_CPU(struct pt_regs *, irq_regs); static inline struct pt_regs *get_irq_regs(void) { - return x86_read_percpu(irq_regs); + return percpu_read(irq_regs); } static inline struct pt_regs *set_irq_regs(struct pt_regs *new_regs) @@ -23,7 +23,7 @@ static inline struct pt_regs *set_irq_regs(struct pt_regs *new_regs) struct pt_regs *old_regs; old_regs = get_irq_regs(); - x86_write_percpu(irq_regs, new_regs); + percpu_write(irq_regs, new_regs); return old_regs; } diff --git a/arch/x86/include/asm/irq_vectors.h b/arch/x86/include/asm/irq_vectors.h index f7ff65032b9..1554d0236e0 100644 --- a/arch/x86/include/asm/irq_vectors.h +++ b/arch/x86/include/asm/irq_vectors.h @@ -87,6 +87,11 @@ #define LOCAL_TIMER_VECTOR 0xef /* + * Performance monitoring interrupt vector: + */ +#define LOCAL_PERF_VECTOR 0xee + +/* * First APIC vector available to drivers: (vectors 0x30-0xee) we * start at 0x31(0x41) to spread out vectors evenly between priority * levels. (0x80 is the syscall vector) @@ -105,6 +110,8 @@ #if defined(CONFIG_X86_IO_APIC) && !defined(CONFIG_X86_VOYAGER) +#include <asm/apicnum.h> /* need MAX_IO_APICS */ + #ifndef CONFIG_SPARSE_IRQ # if NR_CPUS < MAX_IO_APICS # define NR_IRQS (NR_VECTORS + (32 * NR_CPUS)) @@ -112,11 +119,12 @@ # define NR_IRQS (NR_VECTORS + (32 * MAX_IO_APICS)) # endif #else -# if (8 * NR_CPUS) > (32 * MAX_IO_APICS) -# define NR_IRQS (NR_VECTORS + (8 * NR_CPUS)) -# else -# define NR_IRQS (NR_VECTORS + (32 * MAX_IO_APICS)) -# endif + +# define NR_IRQS \ + ((8 * NR_CPUS) > (32 * MAX_IO_APICS) ? \ + (NR_VECTORS + (8 * NR_CPUS)) : \ + (NR_VECTORS + (32 * MAX_IO_APICS))) \ + #endif #elif defined(CONFIG_X86_VOYAGER) diff --git a/arch/x86/include/asm/mach-default/entry_arch.h b/arch/x86/include/asm/mach-default/entry_arch.h index 6b1add8e31d..ad31e5d90e9 100644 --- a/arch/x86/include/asm/mach-default/entry_arch.h +++ b/arch/x86/include/asm/mach-default/entry_arch.h @@ -25,10 +25,15 @@ BUILD_INTERRUPT(irq_move_cleanup_interrupt,IRQ_MOVE_CLEANUP_VECTOR) * a much simpler SMP time architecture: */ #ifdef CONFIG_X86_LOCAL_APIC + BUILD_INTERRUPT(apic_timer_interrupt,LOCAL_TIMER_VECTOR) BUILD_INTERRUPT(error_interrupt,ERROR_APIC_VECTOR) BUILD_INTERRUPT(spurious_interrupt,SPURIOUS_APIC_VECTOR) +#ifdef CONFIG_PERF_COUNTERS +BUILD_INTERRUPT(perf_counter_interrupt, LOCAL_PERF_VECTOR) +#endif + #ifdef CONFIG_X86_MCE_P4THERMAL BUILD_INTERRUPT(thermal_interrupt,THERMAL_APIC_VECTOR) #endif diff --git a/arch/x86/include/asm/mmu_context_32.h b/arch/x86/include/asm/mmu_context_32.h index 7e98ce1d2c0..08b53454f83 100644 --- a/arch/x86/include/asm/mmu_context_32.h +++ b/arch/x86/include/asm/mmu_context_32.h @@ -4,8 +4,8 @@ static inline void enter_lazy_tlb(struct mm_struct *mm, struct task_struct *tsk) { #ifdef CONFIG_SMP - if (x86_read_percpu(cpu_tlbstate.state) == TLBSTATE_OK) - x86_write_percpu(cpu_tlbstate.state, TLBSTATE_LAZY); + if (percpu_read(cpu_tlbstate.state) == TLBSTATE_OK) + percpu_write(cpu_tlbstate.state, TLBSTATE_LAZY); #endif } @@ -19,8 +19,8 @@ static inline void switch_mm(struct mm_struct *prev, /* stop flush ipis for the previous mm */ cpu_clear(cpu, prev->cpu_vm_mask); #ifdef CONFIG_SMP - x86_write_percpu(cpu_tlbstate.state, TLBSTATE_OK); - x86_write_percpu(cpu_tlbstate.active_mm, next); + percpu_write(cpu_tlbstate.state, TLBSTATE_OK); + percpu_write(cpu_tlbstate.active_mm, next); #endif cpu_set(cpu, next->cpu_vm_mask); @@ -35,8 +35,8 @@ static inline void switch_mm(struct mm_struct *prev, } #ifdef CONFIG_SMP else { - x86_write_percpu(cpu_tlbstate.state, TLBSTATE_OK); - BUG_ON(x86_read_percpu(cpu_tlbstate.active_mm) != next); + percpu_write(cpu_tlbstate.state, TLBSTATE_OK); + BUG_ON(percpu_read(cpu_tlbstate.active_mm) != next); if (!cpu_test_and_set(cpu, next->cpu_vm_mask)) { /* We were in lazy tlb mode and leave_mm disabled diff --git a/arch/x86/include/asm/mmu_context_64.h b/arch/x86/include/asm/mmu_context_64.h index 677d36e9540..c4572505ab3 100644 --- a/arch/x86/include/asm/mmu_context_64.h +++ b/arch/x86/include/asm/mmu_context_64.h @@ -1,13 +1,11 @@ #ifndef _ASM_X86_MMU_CONTEXT_64_H #define _ASM_X86_MMU_CONTEXT_64_H -#include <asm/pda.h> - static inline void enter_lazy_tlb(struct mm_struct *mm, struct task_struct *tsk) { #ifdef CONFIG_SMP - if (read_pda(mmu_state) == TLBSTATE_OK) - write_pda(mmu_state, TLBSTATE_LAZY); + if (percpu_read(cpu_tlbstate.state) == TLBSTATE_OK) + percpu_write(cpu_tlbstate.state, TLBSTATE_LAZY); #endif } @@ -19,8 +17,8 @@ static inline void switch_mm(struct mm_struct *prev, struct mm_struct *next, /* stop flush ipis for the previous mm */ cpu_clear(cpu, prev->cpu_vm_mask); #ifdef CONFIG_SMP - write_pda(mmu_state, TLBSTATE_OK); - write_pda(active_mm, next); + percpu_write(cpu_tlbstate.state, TLBSTATE_OK); + percpu_write(cpu_tlbstate.active_mm, next); #endif cpu_set(cpu, next->cpu_vm_mask); load_cr3(next->pgd); @@ -30,9 +28,9 @@ static inline void switch_mm(struct mm_struct *prev, struct mm_struct *next, } #ifdef CONFIG_SMP else { - write_pda(mmu_state, TLBSTATE_OK); - if (read_pda(active_mm) != next) - BUG(); + percpu_write(cpu_tlbstate.state, TLBSTATE_OK); + BUG_ON(percpu_read(cpu_tlbstate.active_mm) != next); + if (!cpu_test_and_set(cpu, next->cpu_vm_mask)) { /* We were in lazy tlb mode and leave_mm disabled * tlb flush IPI delivery. We must reload CR3 diff --git a/arch/x86/include/asm/mpspec_def.h b/arch/x86/include/asm/mpspec_def.h index 59568bc4767..4a7f96d7c18 100644 --- a/arch/x86/include/asm/mpspec_def.h +++ b/arch/x86/include/asm/mpspec_def.h @@ -24,17 +24,18 @@ # endif #endif -struct intel_mp_floating { - char mpf_signature[4]; /* "_MP_" */ - unsigned int mpf_physptr; /* Configuration table address */ - unsigned char mpf_length; /* Our length (paragraphs) */ - unsigned char mpf_specification;/* Specification version */ - unsigned char mpf_checksum; /* Checksum (makes sum 0) */ - unsigned char mpf_feature1; /* Standard or configuration ? */ - unsigned char mpf_feature2; /* Bit7 set for IMCR|PIC */ - unsigned char mpf_feature3; /* Unused (0) */ - unsigned char mpf_feature4; /* Unused (0) */ - unsigned char mpf_feature5; /* Unused (0) */ +/* Intel MP Floating Pointer Structure */ +struct mpf_intel { + char signature[4]; /* "_MP_" */ + unsigned int physptr; /* Configuration table address */ + unsigned char length; /* Our length (paragraphs) */ + unsigned char specification; /* Specification version */ + unsigned char checksum; /* Checksum (makes sum 0) */ + unsigned char feature1; /* Standard or configuration ? */ + unsigned char feature2; /* Bit7 set for IMCR|PIC */ + unsigned char feature3; /* Unused (0) */ + unsigned char feature4; /* Unused (0) */ + unsigned char feature5; /* Unused (0) */ }; #define MPC_SIGNATURE "PCMP" diff --git a/arch/x86/include/asm/page_64.h b/arch/x86/include/asm/page_64.h index 5ebca29f44f..e27fdbe5f9e 100644 --- a/arch/x86/include/asm/page_64.h +++ b/arch/x86/include/asm/page_64.h @@ -13,8 +13,8 @@ #define DEBUG_STACK_ORDER (EXCEPTION_STACK_ORDER + 1) #define DEBUG_STKSZ (PAGE_SIZE << DEBUG_STACK_ORDER) -#define IRQSTACK_ORDER 2 -#define IRQSTACKSIZE (PAGE_SIZE << IRQSTACK_ORDER) +#define IRQ_STACK_ORDER 2 +#define IRQ_STACK_SIZE (PAGE_SIZE << IRQ_STACK_ORDER) #define STACKFAULT_STACK 1 #define DOUBLEFAULT_STACK 2 diff --git a/arch/x86/include/asm/paravirt.h b/arch/x86/include/asm/paravirt.h index ba3e2ff6aed..c26c6bf4da0 100644 --- a/arch/x86/include/asm/paravirt.h +++ b/arch/x86/include/asm/paravirt.h @@ -244,7 +244,8 @@ struct pv_mmu_ops { void (*flush_tlb_user)(void); void (*flush_tlb_kernel)(void); void (*flush_tlb_single)(unsigned long addr); - void (*flush_tlb_others)(const cpumask_t *cpus, struct mm_struct *mm, + void (*flush_tlb_others)(const struct cpumask *cpus, + struct mm_struct *mm, unsigned long va); /* Hooks for allocating and freeing a pagetable top-level */ @@ -984,10 +985,11 @@ static inline void __flush_tlb_single(unsigned long addr) PVOP_VCALL1(pv_mmu_ops.flush_tlb_single, addr); } -static inline void flush_tlb_others(cpumask_t cpumask, struct mm_struct *mm, +static inline void flush_tlb_others(const struct cpumask *cpumask, + struct mm_struct *mm, unsigned long va) { - PVOP_VCALL3(pv_mmu_ops.flush_tlb_others, &cpumask, mm, va); + PVOP_VCALL3(pv_mmu_ops.flush_tlb_others, cpumask, mm, va); } static inline int paravirt_pgd_alloc(struct mm_struct *mm) diff --git a/arch/x86/include/asm/pda.h b/arch/x86/include/asm/pda.h index 2fbfff88df3..c31ca048a90 100644 --- a/arch/x86/include/asm/pda.h +++ b/arch/x86/include/asm/pda.h @@ -5,133 +5,41 @@ #include <linux/stddef.h> #include <linux/types.h> #include <linux/cache.h> +#include <linux/threads.h> #include <asm/page.h> +#include <asm/percpu.h> /* Per processor datastructure. %gs points to it while the kernel runs */ struct x8664_pda { - struct task_struct *pcurrent; /* 0 Current process */ - unsigned long data_offset; /* 8 Per cpu data offset from linker - address */ - unsigned long kernelstack; /* 16 top of kernel stack for current */ - unsigned long oldrsp; /* 24 user rsp for system call */ - int irqcount; /* 32 Irq nesting counter. Starts -1 */ - unsigned int cpunumber; /* 36 Logical CPU number */ + unsigned long unused1; + unsigned long unused2; + unsigned long unused3; + unsigned long unused4; + int unused5; + unsigned int unused6; /* 36 was cpunumber */ #ifdef CONFIG_CC_STACKPROTECTOR unsigned long stack_canary; /* 40 stack canary value */ /* gcc-ABI: this canary MUST be at offset 40!!! */ #endif - char *irqstackptr; - short nodenumber; /* number of current node (32k max) */ short in_bootmem; /* pda lives in bootmem */ - unsigned int __softirq_pending; - unsigned int __nmi_count; /* number of NMI on this CPUs */ - short mmu_state; - short isidle; - struct mm_struct *active_mm; - unsigned apic_timer_irqs; - unsigned irq0_irqs; - unsigned irq_resched_count; - unsigned irq_call_count; - unsigned irq_tlb_count; - unsigned irq_thermal_count; - unsigned irq_threshold_count; - unsigned irq_spurious_count; } ____cacheline_aligned_in_smp; -extern struct x8664_pda **_cpu_pda; +DECLARE_PER_CPU(struct x8664_pda, __pda); extern void pda_init(int); -#define cpu_pda(i) (_cpu_pda[i]) +#define cpu_pda(cpu) (&per_cpu(__pda, cpu)) -/* - * There is no fast way to get the base address of the PDA, all the accesses - * have to mention %fs/%gs. So it needs to be done this Torvaldian way. - */ -extern void __bad_pda_field(void) __attribute__((noreturn)); - -/* - * proxy_pda doesn't actually exist, but tell gcc it is accessed for - * all PDA accesses so it gets read/write dependencies right. - */ -extern struct x8664_pda _proxy_pda; - -#define pda_offset(field) offsetof(struct x8664_pda, field) - -#define pda_to_op(op, field, val) \ -do { \ - typedef typeof(_proxy_pda.field) T__; \ - if (0) { T__ tmp__; tmp__ = (val); } /* type checking */ \ - switch (sizeof(_proxy_pda.field)) { \ - case 2: \ - asm(op "w %1,%%gs:%c2" : \ - "+m" (_proxy_pda.field) : \ - "ri" ((T__)val), \ - "i"(pda_offset(field))); \ - break; \ - case 4: \ - asm(op "l %1,%%gs:%c2" : \ - "+m" (_proxy_pda.field) : \ - "ri" ((T__)val), \ - "i" (pda_offset(field))); \ - break; \ - case 8: \ - asm(op "q %1,%%gs:%c2": \ - "+m" (_proxy_pda.field) : \ - "ri" ((T__)val), \ - "i"(pda_offset(field))); \ - break; \ - default: \ - __bad_pda_field(); \ - } \ -} while (0) - -#define pda_from_op(op, field) \ -({ \ - typeof(_proxy_pda.field) ret__; \ - switch (sizeof(_proxy_pda.field)) { \ - case 2: \ - asm(op "w %%gs:%c1,%0" : \ - "=r" (ret__) : \ - "i" (pda_offset(field)), \ - "m" (_proxy_pda.field)); \ - break; \ - case 4: \ - asm(op "l %%gs:%c1,%0": \ - "=r" (ret__): \ - "i" (pda_offset(field)), \ - "m" (_proxy_pda.field)); \ - break; \ - case 8: \ - asm(op "q %%gs:%c1,%0": \ - "=r" (ret__) : \ - "i" (pda_offset(field)), \ - "m" (_proxy_pda.field)); \ - break; \ - default: \ - __bad_pda_field(); \ - } \ - ret__; \ -}) - -#define read_pda(field) pda_from_op("mov", field) -#define write_pda(field, val) pda_to_op("mov", field, val) -#define add_pda(field, val) pda_to_op("add", field, val) -#define sub_pda(field, val) pda_to_op("sub", field, val) -#define or_pda(field, val) pda_to_op("or", field, val) +#define read_pda(field) percpu_read(__pda.field) +#define write_pda(field, val) percpu_write(__pda.field, val) +#define add_pda(field, val) percpu_add(__pda.field, val) +#define sub_pda(field, val) percpu_sub(__pda.field, val) +#define or_pda(field, val) percpu_or(__pda.field, val) /* This is not atomic against other CPUs -- CPU preemption needs to be off */ #define test_and_clear_bit_pda(bit, field) \ -({ \ - int old__; \ - asm volatile("btr %2,%%gs:%c3\n\tsbbl %0,%0" \ - : "=r" (old__), "+m" (_proxy_pda.field) \ - : "dIr" (bit), "i" (pda_offset(field)) : "memory");\ - old__; \ -}) + x86_test_and_clear_bit_percpu(bit, __pda.field) #endif -#define PDA_STACKOFFSET (5*8) - #endif /* _ASM_X86_PDA_H */ diff --git a/arch/x86/include/asm/percpu.h b/arch/x86/include/asm/percpu.h index ece72053ba6..165d5272ece 100644 --- a/arch/x86/include/asm/percpu.h +++ b/arch/x86/include/asm/percpu.h @@ -2,53 +2,12 @@ #define _ASM_X86_PERCPU_H #ifdef CONFIG_X86_64 -#include <linux/compiler.h> - -/* Same as asm-generic/percpu.h, except that we store the per cpu offset - in the PDA. Longer term the PDA and every per cpu variable - should be just put into a single section and referenced directly - from %gs */ - -#ifdef CONFIG_SMP -#include <asm/pda.h> - -#define __per_cpu_offset(cpu) (cpu_pda(cpu)->data_offset) -#define __my_cpu_offset read_pda(data_offset) - -#define per_cpu_offset(x) (__per_cpu_offset(x)) - +#define __percpu_seg gs +#define __percpu_mov_op movq +#else +#define __percpu_seg fs +#define __percpu_mov_op movl #endif -#include <asm-generic/percpu.h> - -DECLARE_PER_CPU(struct x8664_pda, pda); - -/* - * These are supposed to be implemented as a single instruction which - * operates on the per-cpu data base segment. x86-64 doesn't have - * that yet, so this is a fairly inefficient workaround for the - * meantime. The single instruction is atomic with respect to - * preemption and interrupts, so we need to explicitly disable - * interrupts here to achieve the same effect. However, because it - * can be used from within interrupt-disable/enable, we can't actually - * disable interrupts; disabling preemption is enough. - */ -#define x86_read_percpu(var) \ - ({ \ - typeof(per_cpu_var(var)) __tmp; \ - preempt_disable(); \ - __tmp = __get_cpu_var(var); \ - preempt_enable(); \ - __tmp; \ - }) - -#define x86_write_percpu(var, val) \ - do { \ - preempt_disable(); \ - __get_cpu_var(var) = (val); \ - preempt_enable(); \ - } while(0) - -#else /* CONFIG_X86_64 */ #ifdef __ASSEMBLY__ @@ -65,47 +24,26 @@ DECLARE_PER_CPU(struct x8664_pda, pda); * PER_CPU(cpu_gdt_descr, %ebx) */ #ifdef CONFIG_SMP -#define PER_CPU(var, reg) \ - movl %fs:per_cpu__##this_cpu_off, reg; \ +#define PER_CPU(var, reg) \ + __percpu_mov_op %__percpu_seg:per_cpu__this_cpu_off, reg; \ lea per_cpu__##var(reg), reg -#define PER_CPU_VAR(var) %fs:per_cpu__##var +#define PER_CPU_VAR(var) %__percpu_seg:per_cpu__##var #else /* ! SMP */ -#define PER_CPU(var, reg) \ - movl $per_cpu__##var, reg +#define PER_CPU(var, reg) \ + __percpu_mov_op $per_cpu__##var, reg #define PER_CPU_VAR(var) per_cpu__##var #endif /* SMP */ #else /* ...!ASSEMBLY */ -/* - * PER_CPU finds an address of a per-cpu variable. - * - * Args: - * var - variable name - * cpu - 32bit register containing the current CPU number - * - * The resulting address is stored in the "cpu" argument. - * - * Example: - * PER_CPU(cpu_gdt_descr, %ebx) - */ -#ifdef CONFIG_SMP - -#define __my_cpu_offset x86_read_percpu(this_cpu_off) - -/* fs segment starts at (positive) offset == __per_cpu_offset[cpu] */ -#define __percpu_seg "%%fs:" - -#else /* !SMP */ +#include <linux/stringify.h> -#define __percpu_seg "" - -#endif /* SMP */ - -#include <asm-generic/percpu.h> - -/* We can use this directly for local CPU (faster). */ -DECLARE_PER_CPU(unsigned long, this_cpu_off); +#ifdef CONFIG_SMP +#define __percpu_arg(x) "%%"__stringify(__percpu_seg)":%P" #x +#define __my_cpu_offset percpu_read(this_cpu_off) +#else +#define __percpu_arg(x) "%" #x +#endif /* For arch-specific code, we can use direct single-insn ops (they * don't give an lvalue though). */ @@ -120,20 +58,25 @@ do { \ } \ switch (sizeof(var)) { \ case 1: \ - asm(op "b %1,"__percpu_seg"%0" \ + asm(op "b %1,"__percpu_arg(0) \ : "+m" (var) \ : "ri" ((T__)val)); \ break; \ case 2: \ - asm(op "w %1,"__percpu_seg"%0" \ + asm(op "w %1,"__percpu_arg(0) \ : "+m" (var) \ : "ri" ((T__)val)); \ break; \ case 4: \ - asm(op "l %1,"__percpu_seg"%0" \ + asm(op "l %1,"__percpu_arg(0) \ : "+m" (var) \ : "ri" ((T__)val)); \ break; \ + case 8: \ + asm(op "q %1,"__percpu_arg(0) \ + : "+m" (var) \ + : "r" ((T__)val)); \ + break; \ default: __bad_percpu_size(); \ } \ } while (0) @@ -143,17 +86,22 @@ do { \ typeof(var) ret__; \ switch (sizeof(var)) { \ case 1: \ - asm(op "b "__percpu_seg"%1,%0" \ + asm(op "b "__percpu_arg(1)",%0" \ : "=r" (ret__) \ : "m" (var)); \ break; \ case 2: \ - asm(op "w "__percpu_seg"%1,%0" \ + asm(op "w "__percpu_arg(1)",%0" \ : "=r" (ret__) \ : "m" (var)); \ break; \ case 4: \ - asm(op "l "__percpu_seg"%1,%0" \ + asm(op "l "__percpu_arg(1)",%0" \ + : "=r" (ret__) \ + : "m" (var)); \ + break; \ + case 8: \ + asm(op "q "__percpu_arg(1)",%0" \ : "=r" (ret__) \ : "m" (var)); \ break; \ @@ -162,13 +110,36 @@ do { \ ret__; \ }) -#define x86_read_percpu(var) percpu_from_op("mov", per_cpu__##var) -#define x86_write_percpu(var, val) percpu_to_op("mov", per_cpu__##var, val) -#define x86_add_percpu(var, val) percpu_to_op("add", per_cpu__##var, val) -#define x86_sub_percpu(var, val) percpu_to_op("sub", per_cpu__##var, val) -#define x86_or_percpu(var, val) percpu_to_op("or", per_cpu__##var, val) +#define percpu_read(var) percpu_from_op("mov", per_cpu__##var) +#define percpu_write(var, val) percpu_to_op("mov", per_cpu__##var, val) +#define percpu_add(var, val) percpu_to_op("add", per_cpu__##var, val) +#define percpu_sub(var, val) percpu_to_op("sub", per_cpu__##var, val) +#define percpu_and(var, val) percpu_to_op("and", per_cpu__##var, val) +#define percpu_or(var, val) percpu_to_op("or", per_cpu__##var, val) +#define percpu_xor(var, val) percpu_to_op("xor", per_cpu__##var, val) + +/* This is not atomic against other CPUs -- CPU preemption needs to be off */ +#define x86_test_and_clear_bit_percpu(bit, var) \ +({ \ + int old__; \ + asm volatile("btr %2,"__percpu_arg(1)"\n\tsbbl %0,%0" \ + : "=r" (old__), "+m" (per_cpu__##var) \ + : "dIr" (bit)); \ + old__; \ +}) + +#include <asm-generic/percpu.h> + +/* We can use this directly for local CPU (faster). */ +DECLARE_PER_CPU(unsigned long, this_cpu_off); + +#ifdef CONFIG_X86_64 +extern void load_pda_offset(int cpu); +#else +static inline void load_pda_offset(int cpu) { } +#endif + #endif /* !__ASSEMBLY__ */ -#endif /* !CONFIG_X86_64 */ #ifdef CONFIG_SMP @@ -195,9 +166,9 @@ do { \ #define early_per_cpu_ptr(_name) (_name##_early_ptr) #define early_per_cpu_map(_name, _idx) (_name##_early_map[_idx]) #define early_per_cpu(_name, _cpu) \ - (early_per_cpu_ptr(_name) ? \ - early_per_cpu_ptr(_name)[_cpu] : \ - per_cpu(_name, _cpu)) + *(early_per_cpu_ptr(_name) ? \ + &early_per_cpu_ptr(_name)[_cpu] : \ + &per_cpu(_name, _cpu)) #else /* !CONFIG_SMP */ #define DEFINE_EARLY_PER_CPU(_type, _name, _initvalue) \ diff --git a/arch/x86/include/asm/perf_counter.h b/arch/x86/include/asm/perf_counter.h new file mode 100644 index 00000000000..2e08ed73664 --- /dev/null +++ b/arch/x86/include/asm/perf_counter.h @@ -0,0 +1,95 @@ +#ifndef _ASM_X86_PERF_COUNTER_H +#define _ASM_X86_PERF_COUNTER_H + +/* + * Performance counter hw details: + */ + +#define X86_PMC_MAX_GENERIC 8 +#define X86_PMC_MAX_FIXED 3 + +#define X86_PMC_IDX_GENERIC 0 +#define X86_PMC_IDX_FIXED 32 +#define X86_PMC_IDX_MAX 64 + +#define MSR_ARCH_PERFMON_PERFCTR0 0xc1 +#define MSR_ARCH_PERFMON_PERFCTR1 0xc2 + +#define MSR_ARCH_PERFMON_EVENTSEL0 0x186 +#define MSR_ARCH_PERFMON_EVENTSEL1 0x187 + +#define ARCH_PERFMON_EVENTSEL0_ENABLE (1 << 22) +#define ARCH_PERFMON_EVENTSEL_INT (1 << 20) +#define ARCH_PERFMON_EVENTSEL_OS (1 << 17) +#define ARCH_PERFMON_EVENTSEL_USR (1 << 16) + +/* + * Includes eventsel and unit mask as well: + */ +#define ARCH_PERFMON_EVENT_MASK 0xffff + +#define ARCH_PERFMON_UNHALTED_CORE_CYCLES_SEL 0x3c +#define ARCH_PERFMON_UNHALTED_CORE_CYCLES_UMASK (0x00 << 8) +#define ARCH_PERFMON_UNHALTED_CORE_CYCLES_INDEX 0 +#define ARCH_PERFMON_UNHALTED_CORE_CYCLES_PRESENT \ + (1 << (ARCH_PERFMON_UNHALTED_CORE_CYCLES_INDEX)) + +#define ARCH_PERFMON_BRANCH_MISSES_RETIRED 6 + +/* + * Intel "Architectural Performance Monitoring" CPUID + * detection/enumeration details: + */ +union cpuid10_eax { + struct { + unsigned int version_id:8; + unsigned int num_counters:8; + unsigned int bit_width:8; + unsigned int mask_length:8; + } split; + unsigned int full; +}; + +union cpuid10_edx { + struct { + unsigned int num_counters_fixed:4; + unsigned int reserved:28; + } split; + unsigned int full; +}; + + +/* + * Fixed-purpose performance counters: + */ + +/* + * All 3 fixed-mode PMCs are configured via this single MSR: + */ +#define MSR_ARCH_PERFMON_FIXED_CTR_CTRL 0x38d + +/* + * The counts are available in three separate MSRs: + */ + +/* Instr_Retired.Any: */ +#define MSR_ARCH_PERFMON_FIXED_CTR0 0x309 +#define X86_PMC_IDX_FIXED_INSTRUCTIONS (X86_PMC_IDX_FIXED + 0) + +/* CPU_CLK_Unhalted.Core: */ +#define MSR_ARCH_PERFMON_FIXED_CTR1 0x30a +#define X86_PMC_IDX_FIXED_CPU_CYCLES (X86_PMC_IDX_FIXED + 1) + +/* CPU_CLK_Unhalted.Ref: */ +#define MSR_ARCH_PERFMON_FIXED_CTR2 0x30b +#define X86_PMC_IDX_FIXED_BUS_CYCLES (X86_PMC_IDX_FIXED + 2) + +#ifdef CONFIG_PERF_COUNTERS +extern void init_hw_perf_counters(void); +extern void perf_counters_lapic_init(int nmi); +#else +static inline void init_hw_perf_counters(void) { } +static inline void perf_counters_lapic_init(int nmi) { } +#endif + +#endif /* _ASM_X86_PERF_COUNTER_H */ diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h index 091cd8855f2..f511246fa6c 100644 --- a/arch/x86/include/asm/processor.h +++ b/arch/x86/include/asm/processor.h @@ -378,6 +378,9 @@ union thread_xstate { #ifdef CONFIG_X86_64 DECLARE_PER_CPU(struct orig_ist, orig_ist); + +DECLARE_PER_CPU(char[IRQ_STACK_SIZE], irq_stack); +DECLARE_PER_CPU(char *, irq_stack_ptr); #endif extern void print_cpu_info(struct cpuinfo_x86 *); diff --git a/arch/x86/include/asm/setup.h b/arch/x86/include/asm/setup.h index ebe858cdc8a..536949749bc 100644 --- a/arch/x86/include/asm/setup.h +++ b/arch/x86/include/asm/setup.h @@ -100,7 +100,6 @@ extern unsigned long init_pg_tables_start; extern unsigned long init_pg_tables_end; #else -void __init x86_64_init_pda(void); void __init x86_64_start_kernel(char *real_mode); void __init x86_64_start_reservations(char *real_mode_data); diff --git a/arch/x86/include/asm/smp.h b/arch/x86/include/asm/smp.h index 19953df61c5..68636e767a9 100644 --- a/arch/x86/include/asm/smp.h +++ b/arch/x86/include/asm/smp.h @@ -17,32 +17,7 @@ #endif #include <asm/pda.h> #include <asm/thread_info.h> - -#ifdef CONFIG_X86_64 - -extern cpumask_var_t cpu_callin_mask; -extern cpumask_var_t cpu_callout_mask; -extern cpumask_var_t cpu_initialized_mask; -extern cpumask_var_t cpu_sibling_setup_mask; - -#else /* CONFIG_X86_32 */ - -extern cpumask_t cpu_callin_map; -extern cpumask_t cpu_callout_map; -extern cpumask_t cpu_initialized; -extern cpumask_t cpu_sibling_setup_map; - -#define cpu_callin_mask ((struct cpumask *)&cpu_callin_map) -#define cpu_callout_mask ((struct cpumask *)&cpu_callout_map) -#define cpu_initialized_mask ((struct cpumask *)&cpu_initialized) -#define cpu_sibling_setup_mask ((struct cpumask *)&cpu_sibling_setup_map) - -#endif /* CONFIG_X86_32 */ - -extern void (*mtrr_hook)(void); -extern void zap_low_mappings(void); - -extern int __cpuinit get_local_pda(int cpu); +#include <asm/cpumask.h> extern int smp_num_siblings; extern unsigned int num_processors; @@ -50,9 +25,7 @@ extern unsigned int num_processors; DECLARE_PER_CPU(cpumask_t, cpu_sibling_map); DECLARE_PER_CPU(cpumask_t, cpu_core_map); DECLARE_PER_CPU(u16, cpu_llc_id); -#ifdef CONFIG_X86_32 DECLARE_PER_CPU(int, cpu_number); -#endif static inline struct cpumask *cpu_sibling_mask(int cpu) { @@ -167,8 +140,6 @@ void play_dead_common(void); void native_send_call_func_ipi(const struct cpumask *mask); void native_send_call_func_single_ipi(int cpu); -extern void prefill_possible_map(void); - void smp_store_cpu_info(int id); #define cpu_physical_id(cpu) per_cpu(x86_cpu_to_apicid, cpu) @@ -177,10 +148,6 @@ static inline int num_booting_cpus(void) { return cpumask_weight(cpu_callout_mask); } -#else -static inline void prefill_possible_map(void) -{ -} #endif /* CONFIG_SMP */ extern unsigned disabled_cpus __cpuinitdata; @@ -191,11 +158,11 @@ extern unsigned disabled_cpus __cpuinitdata; * from the initial startup. We map APIC_BASE very early in page_setup(), * so this is correct in the x86 case. */ -#define raw_smp_processor_id() (x86_read_percpu(cpu_number)) +#define raw_smp_processor_id() (percpu_read(cpu_number)) extern int safe_smp_processor_id(void); #elif defined(CONFIG_X86_64_SMP) -#define raw_smp_processor_id() read_pda(cpunumber) +#define raw_smp_processor_id() (percpu_read(cpu_number)) #define stack_smp_processor_id() \ ({ \ @@ -205,10 +172,6 @@ extern int safe_smp_processor_id(void); }) #define safe_smp_processor_id() smp_processor_id() -#else /* !CONFIG_X86_32_SMP && !CONFIG_X86_64_SMP */ -#define cpu_physical_id(cpu) boot_cpu_physical_apicid -#define safe_smp_processor_id() 0 -#define stack_smp_processor_id() 0 #endif #ifdef CONFIG_X86_LOCAL_APIC @@ -251,11 +214,5 @@ static inline int hard_smp_processor_id(void) #endif /* CONFIG_X86_LOCAL_APIC */ -#ifdef CONFIG_X86_HAS_BOOT_CPU_ID -extern unsigned char boot_cpu_id; -#else -#define boot_cpu_id 0 -#endif - #endif /* __ASSEMBLY__ */ #endif /* _ASM_X86_SMP_H */ diff --git a/arch/x86/include/asm/system.h b/arch/x86/include/asm/system.h index 8e626ea33a1..d1dc27dba36 100644 --- a/arch/x86/include/asm/system.h +++ b/arch/x86/include/asm/system.h @@ -94,7 +94,7 @@ do { \ "call __switch_to\n\t" \ ".globl thread_return\n" \ "thread_return:\n\t" \ - "movq %%gs:%P[pda_pcurrent],%%rsi\n\t" \ + "movq "__percpu_arg([current_task])",%%rsi\n\t" \ "movq %P[thread_info](%%rsi),%%r8\n\t" \ LOCK_PREFIX "btr %[tif_fork],%P[ti_flags](%%r8)\n\t" \ "movq %%rax,%%rdi\n\t" \ @@ -106,7 +106,7 @@ do { \ [ti_flags] "i" (offsetof(struct thread_info, flags)), \ [tif_fork] "i" (TIF_FORK), \ [thread_info] "i" (offsetof(struct task_struct, stack)), \ - [pda_pcurrent] "i" (offsetof(struct x8664_pda, pcurrent)) \ + [current_task] "m" (per_cpu_var(current_task)) \ : "memory", "cc" __EXTRA_CLOBBER) #endif diff --git a/arch/x86/include/asm/thread_info.h b/arch/x86/include/asm/thread_info.h index 98789647baa..f38488989db 100644 --- a/arch/x86/include/asm/thread_info.h +++ b/arch/x86/include/asm/thread_info.h @@ -82,6 +82,7 @@ struct thread_info { #define TIF_SYSCALL_AUDIT 7 /* syscall auditing active */ #define TIF_SECCOMP 8 /* secure computing */ #define TIF_MCE_NOTIFY 10 /* notify userspace of an MCE */ +#define TIF_PERF_COUNTERS 11 /* notify perf counter work */ #define TIF_NOTSC 16 /* TSC is not accessible in userland */ #define TIF_IA32 17 /* 32bit process */ #define TIF_FORK 18 /* ret_from_fork */ @@ -104,6 +105,7 @@ struct thread_info { #define _TIF_SYSCALL_AUDIT (1 << TIF_SYSCALL_AUDIT) #define _TIF_SECCOMP (1 << TIF_SECCOMP) #define _TIF_MCE_NOTIFY (1 << TIF_MCE_NOTIFY) +#define _TIF_PERF_COUNTERS (1 << TIF_PERF_COUNTERS) #define _TIF_NOTSC (1 << TIF_NOTSC) #define _TIF_IA32 (1 << TIF_IA32) #define _TIF_FORK (1 << TIF_FORK) @@ -135,7 +137,7 @@ struct thread_info { /* Only used for 64 bit */ #define _TIF_DO_NOTIFY_MASK \ - (_TIF_SIGPENDING|_TIF_MCE_NOTIFY|_TIF_NOTIFY_RESUME) + (_TIF_SIGPENDING|_TIF_MCE_NOTIFY|_TIF_PERF_COUNTERS|_TIF_NOTIFY_RESUME) /* flags to check in __switch_to() */ #define _TIF_WORK_CTXSW \ @@ -194,25 +196,21 @@ static inline struct thread_info *current_thread_info(void) #else /* X86_32 */ -#include <asm/pda.h> +#include <asm/percpu.h> +#define KERNEL_STACK_OFFSET (5*8) /* * macros/functions for gaining access to the thread information structure * preempt_count needs to be 1 initially, until the scheduler is functional. */ #ifndef __ASSEMBLY__ -static inline struct thread_info *current_thread_info(void) -{ - struct thread_info *ti; - ti = (void *)(read_pda(kernelstack) + PDA_STACKOFFSET - THREAD_SIZE); - return ti; -} +DECLARE_PER_CPU(unsigned long, kernel_stack); -/* do not use in interrupt context */ -static inline struct thread_info *stack_thread_info(void) +static inline struct thread_info *current_thread_info(void) { struct thread_info *ti; - asm("andq %%rsp,%0; " : "=r" (ti) : "0" (~(THREAD_SIZE - 1))); + ti = (void *)(percpu_read(kernel_stack) + + KERNEL_STACK_OFFSET - THREAD_SIZE); return ti; } @@ -220,8 +218,8 @@ static inline struct thread_info *stack_thread_info(void) /* how to get the thread information struct from ASM */ #define GET_THREAD_INFO(reg) \ - movq %gs:pda_kernelstack,reg ; \ - subq $(THREAD_SIZE-PDA_STACKOFFSET),reg + movq PER_CPU_VAR(kernel_stack),reg ; \ + subq $(THREAD_SIZE-KERNEL_STACK_OFFSET),reg #endif diff --git a/arch/x86/include/asm/tlbflush.h b/arch/x86/include/asm/tlbflush.h index 0e7bbb54911..d3539f998f8 100644 --- a/arch/x86/include/asm/tlbflush.h +++ b/arch/x86/include/asm/tlbflush.h @@ -113,7 +113,7 @@ static inline void flush_tlb_range(struct vm_area_struct *vma, __flush_tlb(); } -static inline void native_flush_tlb_others(const cpumask_t *cpumask, +static inline void native_flush_tlb_others(const struct cpumask *cpumask, struct mm_struct *mm, unsigned long va) { @@ -142,31 +142,28 @@ static inline void flush_tlb_range(struct vm_area_struct *vma, flush_tlb_mm(vma->vm_mm); } -void native_flush_tlb_others(const cpumask_t *cpumask, struct mm_struct *mm, - unsigned long va); +void native_flush_tlb_others(const struct cpumask *cpumask, + struct mm_struct *mm, unsigned long va); #define TLBSTATE_OK 1 #define TLBSTATE_LAZY 2 -#ifdef CONFIG_X86_32 struct tlb_state { struct mm_struct *active_mm; int state; - char __cacheline_padding[L1_CACHE_BYTES-8]; }; DECLARE_PER_CPU(struct tlb_state, cpu_tlbstate); -void reset_lazy_tlbstate(void); -#else static inline void reset_lazy_tlbstate(void) { + percpu_write(cpu_tlbstate.state, 0); + percpu_write(cpu_tlbstate.active_mm, &init_mm); } -#endif #endif /* SMP */ #ifndef CONFIG_PARAVIRT -#define flush_tlb_others(mask, mm, va) native_flush_tlb_others(&mask, mm, va) +#define flush_tlb_others(mask, mm, va) native_flush_tlb_others(mask, mm, va) #endif static inline void flush_tlb_kernel_range(unsigned long start, @@ -175,4 +172,6 @@ static inline void flush_tlb_kernel_range(unsigned long start, flush_tlb_all(); } +extern void zap_low_mappings(void); + #endif /* _ASM_X86_TLBFLUSH_H */ diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h index 4e2f2e0aab2..ffea1fe03a9 100644 --- a/arch/x86/include/asm/topology.h +++ b/arch/x86/include/asm/topology.h @@ -83,7 +83,8 @@ extern cpumask_t *node_to_cpumask_map; DECLARE_EARLY_PER_CPU(int, x86_cpu_to_node_map); /* Returns the number of the current Node. */ -#define numa_node_id() read_pda(nodenumber) +DECLARE_PER_CPU(int, node_number); +#define numa_node_id() percpu_read(node_number) #ifdef CONFIG_DEBUG_PER_CPU_MAPS extern int cpu_to_node(int cpu); @@ -102,10 +103,7 @@ static inline int cpu_to_node(int cpu) /* Same function but used if called before per_cpu areas are setup */ static inline int early_cpu_to_node(int cpu) { - if (early_per_cpu_ptr(x86_cpu_to_node_map)) - return early_per_cpu_ptr(x86_cpu_to_node_map)[cpu]; - - return per_cpu(x86_cpu_to_node_map, cpu); + return early_per_cpu(x86_cpu_to_node_map, cpu); } /* Returns a pointer to the cpumask of CPUs on Node 'node'. */ diff --git a/arch/x86/include/asm/trampoline.h b/arch/x86/include/asm/trampoline.h index 780ba0ab94f..90f06c25221 100644 --- a/arch/x86/include/asm/trampoline.h +++ b/arch/x86/include/asm/trampoline.h @@ -13,6 +13,7 @@ extern unsigned char *trampoline_base; extern unsigned long init_rsp; extern unsigned long initial_code; +extern unsigned long initial_gs; #define TRAMPOLINE_SIZE roundup(trampoline_end - trampoline_data, PAGE_SIZE) #define TRAMPOLINE_BASE 0x6000 diff --git a/arch/x86/include/asm/unistd_32.h b/arch/x86/include/asm/unistd_32.h index f2bba78430a..7e47658b0a6 100644 --- a/arch/x86/include/asm/unistd_32.h +++ b/arch/x86/include/asm/unistd_32.h @@ -338,6 +338,7 @@ #define __NR_dup3 330 #define __NR_pipe2 331 #define __NR_inotify_init1 332 +#define __NR_perf_counter_open 333 #ifdef __KERNEL__ diff --git a/arch/x86/include/asm/unistd_64.h b/arch/x86/include/asm/unistd_64.h index d2e415e6666..53025feaf88 100644 --- a/arch/x86/include/asm/unistd_64.h +++ b/arch/x86/include/asm/unistd_64.h @@ -653,7 +653,8 @@ __SYSCALL(__NR_dup3, sys_dup3) __SYSCALL(__NR_pipe2, sys_pipe2) #define __NR_inotify_init1 294 __SYSCALL(__NR_inotify_init1, sys_inotify_init1) - +#define __NR_perf_counter_open 295 +__SYSCALL(__NR_perf_counter_open, sys_perf_counter_open) #ifndef __NO_STUBS #define __ARCH_WANT_OLD_READDIR diff --git a/arch/x86/include/asm/uv/uv_bau.h b/arch/x86/include/asm/uv/uv_bau.h index 50423c7b56b..74e6393bfdd 100644 --- a/arch/x86/include/asm/uv/uv_bau.h +++ b/arch/x86/include/asm/uv/uv_bau.h @@ -325,7 +325,8 @@ static inline void bau_cpubits_clear(struct bau_local_cpumask *dstp, int nbits) #define cpubit_isset(cpu, bau_local_cpumask) \ test_bit((cpu), (bau_local_cpumask).bits) -extern int uv_flush_tlb_others(cpumask_t *, struct mm_struct *, unsigned long); +extern int uv_flush_tlb_others(struct cpumask *, + struct mm_struct *, unsigned long); extern void uv_bau_message_intr1(void); extern void uv_bau_timeout_intr1(void); diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index d37593c2f43..4cb5964f149 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -912,8 +912,8 @@ static u8 __init uniq_ioapic_id(u8 id) DECLARE_BITMAP(used, 256); bitmap_zero(used, 256); for (i = 0; i < nr_ioapics; i++) { - struct mp_config_ioapic *ia = &mp_ioapics[i]; - __set_bit(ia->mp_apicid, used); + struct mpc_ioapic *ia = &mp_ioapics[i]; + __set_bit(ia->apicid, used); } if (!test_bit(id, used)) return id; @@ -945,47 +945,47 @@ void __init mp_register_ioapic(int id, u32 address, u32 gsi_base) idx = nr_ioapics; - mp_ioapics[idx].mp_type = MP_IOAPIC; - mp_ioapics[idx].mp_flags = MPC_APIC_USABLE; - mp_ioapics[idx].mp_apicaddr = address; + mp_ioapics[idx].type = MP_IOAPIC; + mp_ioapics[idx].flags = MPC_APIC_USABLE; + mp_ioapics[idx].apicaddr = address; set_fixmap_nocache(FIX_IO_APIC_BASE_0 + idx, address); - mp_ioapics[idx].mp_apicid = uniq_ioapic_id(id); + mp_ioapics[idx].apicid = uniq_ioapic_id(id); #ifdef CONFIG_X86_32 - mp_ioapics[idx].mp_apicver = io_apic_get_version(idx); + mp_ioapics[idx].apicver = io_apic_get_version(idx); #else - mp_ioapics[idx].mp_apicver = 0; + mp_ioapics[idx].apicver = 0; #endif /* * Build basic GSI lookup table to facilitate gsi->io_apic lookups * and to prevent reprogramming of IOAPIC pins (PCI GSIs). */ - mp_ioapic_routing[idx].apic_id = mp_ioapics[idx].mp_apicid; + mp_ioapic_routing[idx].apic_id = mp_ioapics[idx].apicid; mp_ioapic_routing[idx].gsi_base = gsi_base; mp_ioapic_routing[idx].gsi_end = gsi_base + io_apic_get_redir_entries(idx); - printk(KERN_INFO "IOAPIC[%d]: apic_id %d, version %d, address 0x%lx, " - "GSI %d-%d\n", idx, mp_ioapics[idx].mp_apicid, - mp_ioapics[idx].mp_apicver, mp_ioapics[idx].mp_apicaddr, + printk(KERN_INFO "IOAPIC[%d]: apic_id %d, version %d, address 0x%x, " + "GSI %d-%d\n", idx, mp_ioapics[idx].apicid, + mp_ioapics[idx].apicver, mp_ioapics[idx].apicaddr, mp_ioapic_routing[idx].gsi_base, mp_ioapic_routing[idx].gsi_end); nr_ioapics++; } -static void assign_to_mp_irq(struct mp_config_intsrc *m, - struct mp_config_intsrc *mp_irq) +static void assign_to_mp_irq(struct mpc_intsrc *m, + struct mpc_intsrc *mp_irq) { - memcpy(mp_irq, m, sizeof(struct mp_config_intsrc)); + memcpy(mp_irq, m, sizeof(struct mpc_intsrc)); } -static int mp_irq_cmp(struct mp_config_intsrc *mp_irq, - struct mp_config_intsrc *m) +static int mp_irq_cmp(struct mpc_intsrc *mp_irq, + struct mpc_intsrc *m) { - return memcmp(mp_irq, m, sizeof(struct mp_config_intsrc)); + return memcmp(mp_irq, m, sizeof(struct mpc_intsrc)); } -static void save_mp_irq(struct mp_config_intsrc *m) +static void save_mp_irq(struct mpc_intsrc *m) { int i; @@ -1003,7 +1003,7 @@ void __init mp_override_legacy_irq(u8 bus_irq, u8 polarity, u8 trigger, u32 gsi) { int ioapic; int pin; - struct mp_config_intsrc mp_irq; + struct mpc_intsrc mp_irq; /* * Convert 'gsi' to 'ioapic.pin'. @@ -1021,13 +1021,13 @@ void __init mp_override_legacy_irq(u8 bus_irq, u8 polarity, u8 trigger, u32 gsi) if ((bus_irq == 0) && (trigger == 3)) trigger = 1; - mp_irq.mp_type = MP_INTSRC; - mp_irq.mp_irqtype = mp_INT; - mp_irq.mp_irqflag = (trigger << 2) | polarity; - mp_irq.mp_srcbus = MP_ISA_BUS; - mp_irq.mp_srcbusirq = bus_irq; /* IRQ */ - mp_irq.mp_dstapic = mp_ioapics[ioapic].mp_apicid; /* APIC ID */ - mp_irq.mp_dstirq = pin; /* INTIN# */ + mp_irq.type = MP_INTSRC; + mp_irq.irqtype = mp_INT; + mp_irq.irqflag = (trigger << 2) | polarity; + mp_irq.srcbus = MP_ISA_BUS; + mp_irq.srcbusirq = bus_irq; /* IRQ */ + mp_irq.dstapic = mp_ioapics[ioapic].apicid; /* APIC ID */ + mp_irq.dstirq = pin; /* INTIN# */ save_mp_irq(&mp_irq); } @@ -1037,7 +1037,7 @@ void __init mp_config_acpi_legacy_irqs(void) int i; int ioapic; unsigned int dstapic; - struct mp_config_intsrc mp_irq; + struct mpc_intsrc mp_irq; #if defined (CONFIG_MCA) || defined (CONFIG_EISA) /* @@ -1062,7 +1062,7 @@ void __init mp_config_acpi_legacy_irqs(void) ioapic = mp_find_ioapic(0); if (ioapic < 0) return; - dstapic = mp_ioapics[ioapic].mp_apicid; + dstapic = mp_ioapics[ioapic].apicid; /* * Use the default configuration for the IRQs 0-15. Unless @@ -1072,16 +1072,14 @@ void __init mp_config_acpi_legacy_irqs(void) int idx; for (idx = 0; idx < mp_irq_entries; idx++) { - struct mp_config_intsrc *irq = mp_irqs + idx; + struct mpc_intsrc *irq = mp_irqs + idx; /* Do we already have a mapping for this ISA IRQ? */ - if (irq->mp_srcbus == MP_ISA_BUS - && irq->mp_srcbusirq == i) + if (irq->srcbus == MP_ISA_BUS && irq->srcbusirq == i) break; /* Do we already have a mapping for this IOAPIC pin */ - if (irq->mp_dstapic == dstapic && - irq->mp_dstirq == i) + if (irq->dstapic == dstapic && irq->dstirq == i) break; } @@ -1090,13 +1088,13 @@ void __init mp_config_acpi_legacy_irqs(void) continue; /* IRQ already used */ } - mp_irq.mp_type = MP_INTSRC; - mp_irq.mp_irqflag = 0; /* Conforming */ - mp_irq.mp_srcbus = MP_ISA_BUS; - mp_irq.mp_dstapic = dstapic; - mp_irq.mp_irqtype = mp_INT; - mp_irq.mp_srcbusirq = i; /* Identity mapped */ - mp_irq.mp_dstirq = i; + mp_irq.type = MP_INTSRC; + mp_irq.irqflag = 0; /* Conforming */ + mp_irq.srcbus = MP_ISA_BUS; + mp_irq.dstapic = dstapic; + mp_irq.irqtype = mp_INT; + mp_irq.srcbusirq = i; /* Identity mapped */ + mp_irq.dstirq = i; save_mp_irq(&mp_irq); } @@ -1207,22 +1205,22 @@ int mp_config_acpi_gsi(unsigned char number, unsigned int devfn, u8 pin, u32 gsi, int triggering, int polarity) { #ifdef CONFIG_X86_MPPARSE - struct mp_config_intsrc mp_irq; + struct mpc_intsrc mp_irq; int ioapic; if (!acpi_ioapic) return 0; /* print the entry should happen on mptable identically */ - mp_irq.mp_type = MP_INTSRC; - mp_irq.mp_irqtype = mp_INT; - mp_irq.mp_irqflag = (triggering == ACPI_EDGE_SENSITIVE ? 4 : 0x0c) | + mp_irq.type = MP_INTSRC; + mp_irq.irqtype = mp_INT; + mp_irq.irqflag = (triggering == ACPI_EDGE_SENSITIVE ? 4 : 0x0c) | (polarity == ACPI_ACTIVE_HIGH ? 1 : 3); - mp_irq.mp_srcbus = number; - mp_irq.mp_srcbusirq = (((devfn >> 3) & 0x1f) << 2) | ((pin - 1) & 3); + mp_irq.srcbus = number; + mp_irq.srcbusirq = (((devfn >> 3) & 0x1f) << 2) | ((pin - 1) & 3); ioapic = mp_find_ioapic(gsi); - mp_irq.mp_dstapic = mp_ioapic_routing[ioapic].apic_id; - mp_irq.mp_dstirq = gsi - mp_ioapic_routing[ioapic].gsi_base; + mp_irq.dstapic = mp_ioapic_routing[ioapic].apic_id; + mp_irq.dstirq = gsi - mp_ioapic_routing[ioapic].gsi_base; save_mp_irq(&mp_irq); #endif diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c index 707c1f6f95f..4abff454c55 100644 --- a/arch/x86/kernel/acpi/sleep.c +++ b/arch/x86/kernel/acpi/sleep.c @@ -101,6 +101,7 @@ int acpi_save_state_mem(void) stack_start.sp = temp_stack + sizeof(temp_stack); early_gdt_descr.address = (unsigned long)get_cpu_gdt_table(smp_processor_id()); + initial_gs = per_cpu_offset(smp_processor_id()); #endif initial_code = (unsigned long)wakeup_long64; saved_magic = 0x123456789abcdef0; diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 0f830e4f567..e9af14f748e 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -35,6 +35,7 @@ #include <linux/nmi.h> #include <linux/timex.h> +#include <asm/perf_counter.h> #include <asm/atomic.h> #include <asm/mtrr.h> #include <asm/mpspec.h> @@ -895,6 +896,10 @@ void disable_local_APIC(void) { unsigned int value; + /* APIC hasn't been mapped yet */ + if (!apic_phys) + return; + clear_local_APIC(); /* @@ -1126,6 +1131,11 @@ void __cpuinit setup_local_APIC(void) unsigned int value; int i, j; + if (disable_apic) { + disable_ioapic_setup(); + return; + } + #ifdef CONFIG_X86_32 /* Pound the ESR really hard over the head with a big hammer - mbligh */ if (lapic_is_integrated() && esr_disable) { @@ -1135,6 +1145,7 @@ void __cpuinit setup_local_APIC(void) apic_write(APIC_ESR, 0); } #endif + perf_counters_lapic_init(0); preempt_disable(); @@ -1566,11 +1577,11 @@ int apic_version[MAX_APICS]; int __init APIC_init_uniprocessor(void) { -#ifdef CONFIG_X86_64 if (disable_apic) { pr_info("Apic disabled\n"); return -1; } +#ifdef CONFIG_X86_64 if (!cpu_has_apic) { disable_apic = 1; pr_info("Apic disabled by BIOS\n"); @@ -1868,17 +1879,8 @@ void __cpuinit generic_processor_info(int apicid, int version) #endif #if defined(CONFIG_X86_SMP) || defined(CONFIG_X86_64) - /* are we being called early in kernel startup? */ - if (early_per_cpu_ptr(x86_cpu_to_apicid)) { - u16 *cpu_to_apicid = early_per_cpu_ptr(x86_cpu_to_apicid); - u16 *bios_cpu_apicid = early_per_cpu_ptr(x86_bios_cpu_apicid); - - cpu_to_apicid[cpu] = apicid; - bios_cpu_apicid[cpu] = apicid; - } else { - per_cpu(x86_cpu_to_apicid, cpu) = apicid; - per_cpu(x86_bios_cpu_apicid, cpu) = apicid; - } + early_per_cpu(x86_cpu_to_apicid, cpu) = apicid; + early_per_cpu(x86_bios_cpu_apicid, cpu) = apicid; #endif set_cpu_possible(cpu, true); diff --git a/arch/x86/kernel/asm-offsets_64.c b/arch/x86/kernel/asm-offsets_64.c index 1d41d3f1edb..64c834a39aa 100644 --- a/arch/x86/kernel/asm-offsets_64.c +++ b/arch/x86/kernel/asm-offsets_64.c @@ -49,13 +49,7 @@ int main(void) BLANK(); #undef ENTRY #define ENTRY(entry) DEFINE(pda_ ## entry, offsetof(struct x8664_pda, entry)) - ENTRY(kernelstack); - ENTRY(oldrsp); - ENTRY(pcurrent); - ENTRY(irqcount); - ENTRY(cpunumber); - ENTRY(irqstackptr); - ENTRY(data_offset); + DEFINE(pda_size, sizeof(struct x8664_pda)); BLANK(); #undef ENTRY #ifdef CONFIG_PARAVIRT diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile index 82db7f45e2d..c3813306e0b 100644 --- a/arch/x86/kernel/cpu/Makefile +++ b/arch/x86/kernel/cpu/Makefile @@ -1,5 +1,5 @@ # -# Makefile for x86-compatible CPU details and quirks +# Makefile for x86-compatible CPU details, features and quirks # # Don't trace early stages of a secondary CPU boot @@ -22,11 +22,13 @@ obj-$(CONFIG_CPU_SUP_CENTAUR_64) += centaur_64.o obj-$(CONFIG_CPU_SUP_TRANSMETA_32) += transmeta.o obj-$(CONFIG_CPU_SUP_UMC_32) += umc.o -obj-$(CONFIG_X86_MCE) += mcheck/ -obj-$(CONFIG_MTRR) += mtrr/ -obj-$(CONFIG_CPU_FREQ) += cpufreq/ +obj-$(CONFIG_PERF_COUNTERS) += perf_counter.o -obj-$(CONFIG_X86_LOCAL_APIC) += perfctr-watchdog.o +obj-$(CONFIG_X86_MCE) += mcheck/ +obj-$(CONFIG_MTRR) += mtrr/ +obj-$(CONFIG_CPU_FREQ) += cpufreq/ + +obj-$(CONFIG_X86_LOCAL_APIC) += perfctr-watchdog.o quiet_cmd_mkcapflags = MKCAP $@ cmd_mkcapflags = $(PERL) $(srctree)/$(src)/mkcapflags.pl $< $@ diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index 83492b1f93b..95eb30e1e67 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -17,10 +17,13 @@ #include <asm/mmu_context.h> #include <asm/mtrr.h> #include <asm/mce.h> +#include <asm/perf_counter.h> #include <asm/pat.h> #include <asm/asm.h> #include <asm/numa.h> #include <asm/smp.h> +#include <asm/cpu.h> +#include <asm/cpumask.h> #ifdef CONFIG_X86_LOCAL_APIC #include <asm/mpspec.h> #include <asm/apic.h> @@ -772,6 +775,7 @@ void __init identify_boot_cpu(void) #else vgetcpu_set_mode(); #endif + init_hw_perf_counters(); } void __cpuinit identify_secondary_cpu(struct cpuinfo_x86 *c) @@ -877,54 +881,34 @@ static __init int setup_disablecpuid(char *arg) __setup("clearcpuid=", setup_disablecpuid); #ifdef CONFIG_X86_64 -struct x8664_pda **_cpu_pda __read_mostly; -EXPORT_SYMBOL(_cpu_pda); - struct desc_ptr idt_descr = { 256 * 16 - 1, (unsigned long) idt_table }; -static char boot_cpu_stack[IRQSTACKSIZE] __page_aligned_bss; +DEFINE_PER_CPU_PAGE_ALIGNED(char[IRQ_STACK_SIZE], irq_stack); +#ifdef CONFIG_SMP +DEFINE_PER_CPU(char *, irq_stack_ptr); /* will be set during per cpu init */ +#else +DEFINE_PER_CPU(char *, irq_stack_ptr) = + per_cpu_var(irq_stack) + IRQ_STACK_SIZE - 64; +#endif + +DEFINE_PER_CPU(unsigned long, kernel_stack) = + (unsigned long)&init_thread_union - KERNEL_STACK_OFFSET + THREAD_SIZE; +EXPORT_PER_CPU_SYMBOL(kernel_stack); + +DEFINE_PER_CPU(unsigned int, irq_count) = -1; void __cpuinit pda_init(int cpu) { - struct x8664_pda *pda = cpu_pda(cpu); - /* Setup up data that may be needed in __get_free_pages early */ loadsegment(fs, 0); loadsegment(gs, 0); - /* Memory clobbers used to order PDA accessed */ - mb(); - wrmsrl(MSR_GS_BASE, pda); - mb(); - - pda->cpunumber = cpu; - pda->irqcount = -1; - pda->kernelstack = (unsigned long)stack_thread_info() - - PDA_STACKOFFSET + THREAD_SIZE; - pda->active_mm = &init_mm; - pda->mmu_state = 0; - - if (cpu == 0) { - /* others are initialized in smpboot.c */ - pda->pcurrent = &init_task; - pda->irqstackptr = boot_cpu_stack; - pda->irqstackptr += IRQSTACKSIZE - 64; - } else { - if (!pda->irqstackptr) { - pda->irqstackptr = (char *) - __get_free_pages(GFP_ATOMIC, IRQSTACK_ORDER); - if (!pda->irqstackptr) - panic("cannot allocate irqstack for cpu %d", - cpu); - pda->irqstackptr += IRQSTACKSIZE - 64; - } - if (pda->nodenumber == 0 && cpu_to_node(cpu) != NUMA_NO_NODE) - pda->nodenumber = cpu_to_node(cpu); - } + load_pda_offset(cpu); } -static char boot_exception_stacks[(N_EXCEPTION_STACKS - 1) * EXCEPTION_STKSZ + - DEBUG_STKSZ] __page_aligned_bss; +static DEFINE_PER_CPU_PAGE_ALIGNED(char, exception_stacks + [(N_EXCEPTION_STACKS - 1) * EXCEPTION_STKSZ + DEBUG_STKSZ]) + __aligned(PAGE_SIZE); extern asmlinkage void ignore_sysret(void); @@ -982,15 +966,18 @@ void __cpuinit cpu_init(void) struct tss_struct *t = &per_cpu(init_tss, cpu); struct orig_ist *orig_ist = &per_cpu(orig_ist, cpu); unsigned long v; - char *estacks = NULL; struct task_struct *me; int i; /* CPU 0 is initialised in head64.c */ if (cpu != 0) pda_init(cpu); - else - estacks = boot_exception_stacks; + +#ifdef CONFIG_NUMA + if (cpu != 0 && percpu_read(node_number) == 0 && + cpu_to_node(cpu) != NUMA_NO_NODE) + percpu_write(node_number, cpu_to_node(cpu)); +#endif me = current; @@ -1024,18 +1011,13 @@ void __cpuinit cpu_init(void) * set up and load the per-CPU TSS */ if (!orig_ist->ist[0]) { - static const unsigned int order[N_EXCEPTION_STACKS] = { - [0 ... N_EXCEPTION_STACKS - 1] = EXCEPTION_STACK_ORDER, - [DEBUG_STACK - 1] = DEBUG_STACK_ORDER + static const unsigned int sizes[N_EXCEPTION_STACKS] = { + [0 ... N_EXCEPTION_STACKS - 1] = EXCEPTION_STKSZ, + [DEBUG_STACK - 1] = DEBUG_STKSZ }; + char *estacks = per_cpu(exception_stacks, cpu); for (v = 0; v < N_EXCEPTION_STACKS; v++) { - if (cpu) { - estacks = (char *)__get_free_pages(GFP_ATOMIC, order[v]); - if (!estacks) - panic("Cannot allocate exception " - "stack %ld %d\n", v, cpu); - } - estacks += PAGE_SIZE << order[v]; + estacks += sizes[v]; orig_ist->ist[v] = t->x86_tss.ist[v] = (unsigned long)estacks; } diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c index 6f11e029e8c..8f3c95c7e61 100644 --- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c +++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c @@ -235,8 +235,6 @@ static u32 get_cur_val(const struct cpumask *mask) return 0; } - cpumask_copy(cmd.mask, mask); - drv_read(&cmd); dprintk("get_cur_val = %u\n", cmd.val); diff --git a/arch/x86/kernel/cpu/intel_cacheinfo.c b/arch/x86/kernel/cpu/intel_cacheinfo.c index 48533d77be7..58527a9fc40 100644 --- a/arch/x86/kernel/cpu/intel_cacheinfo.c +++ b/arch/x86/kernel/cpu/intel_cacheinfo.c @@ -132,7 +132,16 @@ struct _cpuid4_info { union _cpuid4_leaf_ecx ecx; unsigned long size; unsigned long can_disable; - cpumask_t shared_cpu_map; /* future?: only cpus/node is needed */ + DECLARE_BITMAP(shared_cpu_map, NR_CPUS); +}; + +/* subset of above _cpuid4_info w/o shared_cpu_map */ +struct _cpuid4_info_regs { + union _cpuid4_leaf_eax eax; + union _cpuid4_leaf_ebx ebx; + union _cpuid4_leaf_ecx ecx; + unsigned long size; + unsigned long can_disable; }; #ifdef CONFIG_PCI @@ -263,7 +272,7 @@ amd_cpuid4(int leaf, union _cpuid4_leaf_eax *eax, } static void __cpuinit -amd_check_l3_disable(int index, struct _cpuid4_info *this_leaf) +amd_check_l3_disable(int index, struct _cpuid4_info_regs *this_leaf) { if (index < 3) return; @@ -271,7 +280,8 @@ amd_check_l3_disable(int index, struct _cpuid4_info *this_leaf) } static int -__cpuinit cpuid4_cache_lookup(int index, struct _cpuid4_info *this_leaf) +__cpuinit cpuid4_cache_lookup_regs(int index, + struct _cpuid4_info_regs *this_leaf) { union _cpuid4_leaf_eax eax; union _cpuid4_leaf_ebx ebx; @@ -299,6 +309,15 @@ __cpuinit cpuid4_cache_lookup(int index, struct _cpuid4_info *this_leaf) return 0; } +static int +__cpuinit cpuid4_cache_lookup(int index, struct _cpuid4_info *this_leaf) +{ + struct _cpuid4_info_regs *leaf_regs = + (struct _cpuid4_info_regs *)this_leaf; + + return cpuid4_cache_lookup_regs(index, leaf_regs); +} + static int __cpuinit find_num_cache_leaves(void) { unsigned int eax, ebx, ecx, edx; @@ -338,11 +357,10 @@ unsigned int __cpuinit init_intel_cacheinfo(struct cpuinfo_x86 *c) * parameters cpuid leaf to find the cache details */ for (i = 0; i < num_cache_leaves; i++) { - struct _cpuid4_info this_leaf; - + struct _cpuid4_info_regs this_leaf; int retval; - retval = cpuid4_cache_lookup(i, &this_leaf); + retval = cpuid4_cache_lookup_regs(i, &this_leaf); if (retval >= 0) { switch(this_leaf.eax.split.level) { case 1: @@ -491,17 +509,20 @@ static void __cpuinit cache_shared_cpu_map_setup(unsigned int cpu, int index) num_threads_sharing = 1 + this_leaf->eax.split.num_threads_sharing; if (num_threads_sharing == 1) - cpu_set(cpu, this_leaf->shared_cpu_map); + cpumask_set_cpu(cpu, to_cpumask(this_leaf->shared_cpu_map)); else { index_msb = get_count_order(num_threads_sharing); for_each_online_cpu(i) { if (cpu_data(i).apicid >> index_msb == c->apicid >> index_msb) { - cpu_set(i, this_leaf->shared_cpu_map); + cpumask_set_cpu(i, + to_cpumask(this_leaf->shared_cpu_map)); if (i != cpu && per_cpu(cpuid4_info, i)) { - sibling_leaf = CPUID4_INFO_IDX(i, index); - cpu_set(cpu, sibling_leaf->shared_cpu_map); + sibling_leaf = + CPUID4_INFO_IDX(i, index); + cpumask_set_cpu(cpu, to_cpumask( + sibling_leaf->shared_cpu_map)); } } } @@ -513,9 +534,10 @@ static void __cpuinit cache_remove_shared_cpu_map(unsigned int cpu, int index) int sibling; this_leaf = CPUID4_INFO_IDX(cpu, index); - for_each_cpu_mask_nr(sibling, this_leaf->shared_cpu_map) { + for_each_cpu(sibling, to_cpumask(this_leaf->shared_cpu_map)) { sibling_leaf = CPUID4_INFO_IDX(sibling, index); - cpu_clear(cpu, sibling_leaf->shared_cpu_map); + cpumask_clear_cpu(cpu, + to_cpumask(sibling_leaf->shared_cpu_map)); } } #else @@ -620,8 +642,9 @@ static ssize_t show_shared_cpu_map_func(struct _cpuid4_info *this_leaf, int n = 0; if (len > 1) { - cpumask_t *mask = &this_leaf->shared_cpu_map; + const struct cpumask *mask; + mask = to_cpumask(this_leaf->shared_cpu_map); n = type? cpulist_scnprintf(buf, len-2, mask) : cpumask_scnprintf(buf, len-2, mask); @@ -684,7 +707,8 @@ static struct pci_dev *get_k8_northbridge(int node) static ssize_t show_cache_disable(struct _cpuid4_info *this_leaf, char *buf) { - int node = cpu_to_node(first_cpu(this_leaf->shared_cpu_map)); + const struct cpumask *mask = to_cpumask(this_leaf->shared_cpu_map); + int node = cpu_to_node(cpumask_first(mask)); struct pci_dev *dev = NULL; ssize_t ret = 0; int i; @@ -718,7 +742,8 @@ static ssize_t store_cache_disable(struct _cpuid4_info *this_leaf, const char *buf, size_t count) { - int node = cpu_to_node(first_cpu(this_leaf->shared_cpu_map)); + const struct cpumask *mask = to_cpumask(this_leaf->shared_cpu_map); + int node = cpu_to_node(cpumask_first(mask)); struct pci_dev *dev = NULL; unsigned int ret, index, val; @@ -863,7 +888,7 @@ err_out: return -ENOMEM; } -static cpumask_t cache_dev_map = CPU_MASK_NONE; +static DECLARE_BITMAP(cache_dev_map, NR_CPUS); /* Add/Remove cache interface for CPU device */ static int __cpuinit cache_add_dev(struct sys_device * sys_dev) @@ -903,7 +928,7 @@ static int __cpuinit cache_add_dev(struct sys_device * sys_dev) } kobject_uevent(&(this_object->kobj), KOBJ_ADD); } - cpu_set(cpu, cache_dev_map); + cpumask_set_cpu(cpu, to_cpumask(cache_dev_map)); kobject_uevent(per_cpu(cache_kobject, cpu), KOBJ_ADD); return 0; @@ -916,9 +941,9 @@ static void __cpuinit cache_remove_dev(struct sys_device * sys_dev) if (per_cpu(cpuid4_info, cpu) == NULL) return; - if (!cpu_isset(cpu, cache_dev_map)) + if (!cpumask_test_cpu(cpu, to_cpumask(cache_dev_map))) return; - cpu_clear(cpu, cache_dev_map); + cpumask_clear_cpu(cpu, to_cpumask(cache_dev_map)); for (i = 0; i < num_cache_leaves; i++) kobject_put(&(INDEX_KOBJECT_PTR(cpu,i)->kobj)); diff --git a/arch/x86/kernel/cpu/mcheck/mce_amd_64.c b/arch/x86/kernel/cpu/mcheck/mce_amd_64.c index 8ae8c4ff094..4772e91e824 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_amd_64.c +++ b/arch/x86/kernel/cpu/mcheck/mce_amd_64.c @@ -67,7 +67,7 @@ static struct threshold_block threshold_defaults = { struct threshold_bank { struct kobject *kobj; struct threshold_block *blocks; - cpumask_t cpus; + cpumask_var_t cpus; }; static DEFINE_PER_CPU(struct threshold_bank *, threshold_banks[NR_BANKS]); @@ -481,7 +481,7 @@ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank) #ifdef CONFIG_SMP if (cpu_data(cpu).cpu_core_id && shared_bank[bank]) { /* symlink */ - i = first_cpu(per_cpu(cpu_core_map, cpu)); + i = cpumask_first(&per_cpu(cpu_core_map, cpu)); /* first core not up yet */ if (cpu_data(i).cpu_core_id) @@ -501,7 +501,7 @@ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank) if (err) goto out; - b->cpus = per_cpu(cpu_core_map, cpu); + cpumask_copy(b->cpus, &per_cpu(cpu_core_map, cpu)); per_cpu(threshold_banks, cpu)[bank] = b; goto out; } @@ -512,15 +512,20 @@ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank) err = -ENOMEM; goto out; } + if (!alloc_cpumask_var(&b->cpus, GFP_KERNEL)) { + kfree(b); + err = -ENOMEM; + goto out; + } b->kobj = kobject_create_and_add(name, &per_cpu(device_mce, cpu).kobj); if (!b->kobj) goto out_free; #ifndef CONFIG_SMP - b->cpus = CPU_MASK_ALL; + cpumask_setall(b->cpus); #else - b->cpus = per_cpu(cpu_core_map, cpu); + cpumask_copy(b->cpus, &per_cpu(cpu_core_map, cpu)); #endif per_cpu(threshold_banks, cpu)[bank] = b; @@ -529,7 +534,7 @@ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank) if (err) goto out_free; - for_each_cpu_mask_nr(i, b->cpus) { + for_each_cpu(i, b->cpus) { if (i == cpu) continue; @@ -545,6 +550,7 @@ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank) out_free: per_cpu(threshold_banks, cpu)[bank] = NULL; + free_cpumask_var(b->cpus); kfree(b); out: return err; @@ -619,7 +625,7 @@ static void threshold_remove_bank(unsigned int cpu, int bank) #endif /* remove all sibling symlinks before unregistering */ - for_each_cpu_mask_nr(i, b->cpus) { + for_each_cpu(i, b->cpus) { if (i == cpu) continue; @@ -632,6 +638,7 @@ static void threshold_remove_bank(unsigned int cpu, int bank) free_out: kobject_del(b->kobj); kobject_put(b->kobj); + free_cpumask_var(b->cpus); kfree(b); per_cpu(threshold_banks, cpu)[bank] = NULL; } diff --git a/arch/x86/kernel/cpu/perf_counter.c b/arch/x86/kernel/cpu/perf_counter.c new file mode 100644 index 00000000000..9376771f757 --- /dev/null +++ b/arch/x86/kernel/cpu/perf_counter.c @@ -0,0 +1,695 @@ +/* + * Performance counter x86 architecture code + * + * Copyright(C) 2008 Thomas Gleixner <tglx@linutronix.de> + * Copyright(C) 2008 Red Hat, Inc., Ingo Molnar + * + * For licencing details see kernel-base/COPYING + */ + +#include <linux/perf_counter.h> +#include <linux/capability.h> +#include <linux/notifier.h> +#include <linux/hardirq.h> +#include <linux/kprobes.h> +#include <linux/module.h> +#include <linux/kdebug.h> +#include <linux/sched.h> + +#include <asm/perf_counter.h> +#include <asm/apic.h> + +static bool perf_counters_initialized __read_mostly; + +/* + * Number of (generic) HW counters: + */ +static int nr_counters_generic __read_mostly; +static u64 perf_counter_mask __read_mostly; +static u64 counter_value_mask __read_mostly; + +static int nr_counters_fixed __read_mostly; + +struct cpu_hw_counters { + struct perf_counter *counters[X86_PMC_IDX_MAX]; + unsigned long used[BITS_TO_LONGS(X86_PMC_IDX_MAX)]; +}; + +/* + * Intel PerfMon v3. Used on Core2 and later. + */ +static DEFINE_PER_CPU(struct cpu_hw_counters, cpu_hw_counters); + +static const int intel_perfmon_event_map[] = +{ + [PERF_COUNT_CPU_CYCLES] = 0x003c, + [PERF_COUNT_INSTRUCTIONS] = 0x00c0, + [PERF_COUNT_CACHE_REFERENCES] = 0x4f2e, + [PERF_COUNT_CACHE_MISSES] = 0x412e, + [PERF_COUNT_BRANCH_INSTRUCTIONS] = 0x00c4, + [PERF_COUNT_BRANCH_MISSES] = 0x00c5, + [PERF_COUNT_BUS_CYCLES] = 0x013c, +}; + +static const int max_intel_perfmon_events = ARRAY_SIZE(intel_perfmon_event_map); + +/* + * Propagate counter elapsed time into the generic counter. + * Can only be executed on the CPU where the counter is active. + * Returns the delta events processed. + */ +static void +x86_perf_counter_update(struct perf_counter *counter, + struct hw_perf_counter *hwc, int idx) +{ + u64 prev_raw_count, new_raw_count, delta; + + /* + * Careful: an NMI might modify the previous counter value. + * + * Our tactic to handle this is to first atomically read and + * exchange a new raw count - then add that new-prev delta + * count to the generic counter atomically: + */ +again: + prev_raw_count = atomic64_read(&hwc->prev_count); + rdmsrl(hwc->counter_base + idx, new_raw_count); + + if (atomic64_cmpxchg(&hwc->prev_count, prev_raw_count, + new_raw_count) != prev_raw_count) + goto again; + + /* + * Now we have the new raw value and have updated the prev + * timestamp already. We can now calculate the elapsed delta + * (counter-)time and add that to the generic counter. + * + * Careful, not all hw sign-extends above the physical width + * of the count, so we do that by clipping the delta to 32 bits: + */ + delta = (u64)(u32)((s32)new_raw_count - (s32)prev_raw_count); + + atomic64_add(delta, &counter->count); + atomic64_sub(delta, &hwc->period_left); +} + +/* + * Setup the hardware configuration for a given hw_event_type + */ +static int __hw_perf_counter_init(struct perf_counter *counter) +{ + struct perf_counter_hw_event *hw_event = &counter->hw_event; + struct hw_perf_counter *hwc = &counter->hw; + + if (unlikely(!perf_counters_initialized)) + return -EINVAL; + + /* + * Count user events, and generate PMC IRQs: + * (keep 'enabled' bit clear for now) + */ + hwc->config = ARCH_PERFMON_EVENTSEL_USR | ARCH_PERFMON_EVENTSEL_INT; + + /* + * If privileged enough, count OS events too, and allow + * NMI events as well: + */ + hwc->nmi = 0; + if (capable(CAP_SYS_ADMIN)) { + hwc->config |= ARCH_PERFMON_EVENTSEL_OS; + if (hw_event->nmi) + hwc->nmi = 1; + } + + hwc->irq_period = hw_event->irq_period; + /* + * Intel PMCs cannot be accessed sanely above 32 bit width, + * so we install an artificial 1<<31 period regardless of + * the generic counter period: + */ + if ((s64)hwc->irq_period <= 0 || hwc->irq_period > 0x7FFFFFFF) + hwc->irq_period = 0x7FFFFFFF; + + atomic64_set(&hwc->period_left, hwc->irq_period); + + /* + * Raw event type provide the config in the event structure + */ + if (hw_event->raw) { + hwc->config |= hw_event->type; + } else { + if (hw_event->type >= max_intel_perfmon_events) + return -EINVAL; + /* + * The generic map: + */ + hwc->config |= intel_perfmon_event_map[hw_event->type]; + } + counter->wakeup_pending = 0; + + return 0; +} + +u64 hw_perf_save_disable(void) +{ + u64 ctrl; + + if (unlikely(!perf_counters_initialized)) + return 0; + + rdmsrl(MSR_CORE_PERF_GLOBAL_CTRL, ctrl); + wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0); + + return ctrl; +} +EXPORT_SYMBOL_GPL(hw_perf_save_disable); + +void hw_perf_restore(u64 ctrl) +{ + if (unlikely(!perf_counters_initialized)) + return; + + wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, ctrl); +} +EXPORT_SYMBOL_GPL(hw_perf_restore); + +static inline void +__pmc_fixed_disable(struct perf_counter *counter, + struct hw_perf_counter *hwc, unsigned int __idx) +{ + int idx = __idx - X86_PMC_IDX_FIXED; + u64 ctrl_val, mask; + int err; + + mask = 0xfULL << (idx * 4); + + rdmsrl(hwc->config_base, ctrl_val); + ctrl_val &= ~mask; + err = checking_wrmsrl(hwc->config_base, ctrl_val); +} + +static inline void +__pmc_generic_disable(struct perf_counter *counter, + struct hw_perf_counter *hwc, unsigned int idx) +{ + if (unlikely(hwc->config_base == MSR_ARCH_PERFMON_FIXED_CTR_CTRL)) + __pmc_fixed_disable(counter, hwc, idx); + else + wrmsr_safe(hwc->config_base + idx, hwc->config, 0); +} + +static DEFINE_PER_CPU(u64, prev_left[X86_PMC_IDX_MAX]); + +/* + * Set the next IRQ period, based on the hwc->period_left value. + * To be called with the counter disabled in hw: + */ +static void +__hw_perf_counter_set_period(struct perf_counter *counter, + struct hw_perf_counter *hwc, int idx) +{ + s64 left = atomic64_read(&hwc->period_left); + s32 period = hwc->irq_period; + int err; + + /* + * If we are way outside a reasoable range then just skip forward: + */ + if (unlikely(left <= -period)) { + left = period; + atomic64_set(&hwc->period_left, left); + } + + if (unlikely(left <= 0)) { + left += period; + atomic64_set(&hwc->period_left, left); + } + + per_cpu(prev_left[idx], smp_processor_id()) = left; + + /* + * The hw counter starts counting from this counter offset, + * mark it to be able to extra future deltas: + */ + atomic64_set(&hwc->prev_count, (u64)-left); + + err = checking_wrmsrl(hwc->counter_base + idx, + (u64)(-left) & counter_value_mask); +} + +static inline void +__pmc_fixed_enable(struct perf_counter *counter, + struct hw_perf_counter *hwc, unsigned int __idx) +{ + int idx = __idx - X86_PMC_IDX_FIXED; + u64 ctrl_val, bits, mask; + int err; + + /* + * Enable IRQ generation (0x8) and ring-3 counting (0x2), + * and enable ring-0 counting if allowed: + */ + bits = 0x8ULL | 0x2ULL; + if (hwc->config & ARCH_PERFMON_EVENTSEL_OS) + bits |= 0x1; + bits <<= (idx * 4); + mask = 0xfULL << (idx * 4); + + rdmsrl(hwc->config_base, ctrl_val); + ctrl_val &= ~mask; + ctrl_val |= bits; + err = checking_wrmsrl(hwc->config_base, ctrl_val); +} + +static void +__pmc_generic_enable(struct perf_counter *counter, + struct hw_perf_counter *hwc, int idx) +{ + if (unlikely(hwc->config_base == MSR_ARCH_PERFMON_FIXED_CTR_CTRL)) + __pmc_fixed_enable(counter, hwc, idx); + else + wrmsr(hwc->config_base + idx, + hwc->config | ARCH_PERFMON_EVENTSEL0_ENABLE, 0); +} + +static int +fixed_mode_idx(struct perf_counter *counter, struct hw_perf_counter *hwc) +{ + unsigned int event; + + if (unlikely(hwc->nmi)) + return -1; + + event = hwc->config & ARCH_PERFMON_EVENT_MASK; + + if (unlikely(event == intel_perfmon_event_map[PERF_COUNT_INSTRUCTIONS])) + return X86_PMC_IDX_FIXED_INSTRUCTIONS; + if (unlikely(event == intel_perfmon_event_map[PERF_COUNT_CPU_CYCLES])) + return X86_PMC_IDX_FIXED_CPU_CYCLES; + if (unlikely(event == intel_perfmon_event_map[PERF_COUNT_BUS_CYCLES])) + return X86_PMC_IDX_FIXED_BUS_CYCLES; + + return -1; +} + +/* + * Find a PMC slot for the freshly enabled / scheduled in counter: + */ +static int pmc_generic_enable(struct perf_counter *counter) +{ + struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters); + struct hw_perf_counter *hwc = &counter->hw; + int idx; + + idx = fixed_mode_idx(counter, hwc); + if (idx >= 0) { + /* + * Try to get the fixed counter, if that is already taken + * then try to get a generic counter: + */ + if (test_and_set_bit(idx, cpuc->used)) + goto try_generic; + + hwc->config_base = MSR_ARCH_PERFMON_FIXED_CTR_CTRL; + /* + * We set it so that counter_base + idx in wrmsr/rdmsr maps to + * MSR_ARCH_PERFMON_FIXED_CTR0 ... CTR2: + */ + hwc->counter_base = + MSR_ARCH_PERFMON_FIXED_CTR0 - X86_PMC_IDX_FIXED; + hwc->idx = idx; + } else { + idx = hwc->idx; + /* Try to get the previous generic counter again */ + if (test_and_set_bit(idx, cpuc->used)) { +try_generic: + idx = find_first_zero_bit(cpuc->used, nr_counters_generic); + if (idx == nr_counters_generic) + return -EAGAIN; + + set_bit(idx, cpuc->used); + hwc->idx = idx; + } + hwc->config_base = MSR_ARCH_PERFMON_EVENTSEL0; + hwc->counter_base = MSR_ARCH_PERFMON_PERFCTR0; + } + + perf_counters_lapic_init(hwc->nmi); + + __pmc_generic_disable(counter, hwc, idx); + + cpuc->counters[idx] = counter; + /* + * Make it visible before enabling the hw: + */ + smp_wmb(); + + __hw_perf_counter_set_period(counter, hwc, idx); + __pmc_generic_enable(counter, hwc, idx); + + return 0; +} + +void perf_counter_print_debug(void) +{ + u64 ctrl, status, overflow, pmc_ctrl, pmc_count, prev_left, fixed; + struct cpu_hw_counters *cpuc; + int cpu, idx; + + if (!nr_counters_generic) + return; + + local_irq_disable(); + + cpu = smp_processor_id(); + cpuc = &per_cpu(cpu_hw_counters, cpu); + + rdmsrl(MSR_CORE_PERF_GLOBAL_CTRL, ctrl); + rdmsrl(MSR_CORE_PERF_GLOBAL_STATUS, status); + rdmsrl(MSR_CORE_PERF_GLOBAL_OVF_CTRL, overflow); + rdmsrl(MSR_ARCH_PERFMON_FIXED_CTR_CTRL, fixed); + + printk(KERN_INFO "\n"); + printk(KERN_INFO "CPU#%d: ctrl: %016llx\n", cpu, ctrl); + printk(KERN_INFO "CPU#%d: status: %016llx\n", cpu, status); + printk(KERN_INFO "CPU#%d: overflow: %016llx\n", cpu, overflow); + printk(KERN_INFO "CPU#%d: fixed: %016llx\n", cpu, fixed); + printk(KERN_INFO "CPU#%d: used: %016llx\n", cpu, *(u64 *)cpuc->used); + + for (idx = 0; idx < nr_counters_generic; idx++) { + rdmsrl(MSR_ARCH_PERFMON_EVENTSEL0 + idx, pmc_ctrl); + rdmsrl(MSR_ARCH_PERFMON_PERFCTR0 + idx, pmc_count); + + prev_left = per_cpu(prev_left[idx], cpu); + + printk(KERN_INFO "CPU#%d: gen-PMC%d ctrl: %016llx\n", + cpu, idx, pmc_ctrl); + printk(KERN_INFO "CPU#%d: gen-PMC%d count: %016llx\n", + cpu, idx, pmc_count); + printk(KERN_INFO "CPU#%d: gen-PMC%d left: %016llx\n", + cpu, idx, prev_left); + } + for (idx = 0; idx < nr_counters_fixed; idx++) { + rdmsrl(MSR_ARCH_PERFMON_FIXED_CTR0 + idx, pmc_count); + + printk(KERN_INFO "CPU#%d: fixed-PMC%d count: %016llx\n", + cpu, idx, pmc_count); + } + local_irq_enable(); +} + +static void pmc_generic_disable(struct perf_counter *counter) +{ + struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters); + struct hw_perf_counter *hwc = &counter->hw; + unsigned int idx = hwc->idx; + + __pmc_generic_disable(counter, hwc, idx); + + clear_bit(idx, cpuc->used); + cpuc->counters[idx] = NULL; + /* + * Make sure the cleared pointer becomes visible before we + * (potentially) free the counter: + */ + smp_wmb(); + + /* + * Drain the remaining delta count out of a counter + * that we are disabling: + */ + x86_perf_counter_update(counter, hwc, idx); +} + +static void perf_store_irq_data(struct perf_counter *counter, u64 data) +{ + struct perf_data *irqdata = counter->irqdata; + + if (irqdata->len > PERF_DATA_BUFLEN - sizeof(u64)) { + irqdata->overrun++; + } else { + u64 *p = (u64 *) &irqdata->data[irqdata->len]; + + *p = data; + irqdata->len += sizeof(u64); + } +} + +/* + * Save and restart an expired counter. Called by NMI contexts, + * so it has to be careful about preempting normal counter ops: + */ +static void perf_save_and_restart(struct perf_counter *counter) +{ + struct hw_perf_counter *hwc = &counter->hw; + int idx = hwc->idx; + + x86_perf_counter_update(counter, hwc, idx); + __hw_perf_counter_set_period(counter, hwc, idx); + + if (counter->state == PERF_COUNTER_STATE_ACTIVE) + __pmc_generic_enable(counter, hwc, idx); +} + +static void +perf_handle_group(struct perf_counter *sibling, u64 *status, u64 *overflown) +{ + struct perf_counter *counter, *group_leader = sibling->group_leader; + + /* + * Store sibling timestamps (if any): + */ + list_for_each_entry(counter, &group_leader->sibling_list, list_entry) { + + x86_perf_counter_update(counter, &counter->hw, counter->hw.idx); + perf_store_irq_data(sibling, counter->hw_event.type); + perf_store_irq_data(sibling, atomic64_read(&counter->count)); + } +} + +/* + * This handler is triggered by the local APIC, so the APIC IRQ handling + * rules apply: + */ +static void __smp_perf_counter_interrupt(struct pt_regs *regs, int nmi) +{ + int bit, cpu = smp_processor_id(); + u64 ack, status, saved_global; + struct cpu_hw_counters *cpuc; + + rdmsrl(MSR_CORE_PERF_GLOBAL_CTRL, saved_global); + + /* Disable counters globally */ + wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0); + ack_APIC_irq(); + + cpuc = &per_cpu(cpu_hw_counters, cpu); + + rdmsrl(MSR_CORE_PERF_GLOBAL_STATUS, status); + if (!status) + goto out; + +again: + ack = status; + for_each_bit(bit, (unsigned long *)&status, X86_PMC_IDX_MAX) { + struct perf_counter *counter = cpuc->counters[bit]; + + clear_bit(bit, (unsigned long *) &status); + if (!counter) + continue; + + perf_save_and_restart(counter); + + switch (counter->hw_event.record_type) { + case PERF_RECORD_SIMPLE: + continue; + case PERF_RECORD_IRQ: + perf_store_irq_data(counter, instruction_pointer(regs)); + break; + case PERF_RECORD_GROUP: + perf_handle_group(counter, &status, &ack); + break; + } + /* + * From NMI context we cannot call into the scheduler to + * do a task wakeup - but we mark these generic as + * wakeup_pending and initate a wakeup callback: + */ + if (nmi) { + counter->wakeup_pending = 1; + set_tsk_thread_flag(current, TIF_PERF_COUNTERS); + } else { + wake_up(&counter->waitq); + } + } + + wrmsrl(MSR_CORE_PERF_GLOBAL_OVF_CTRL, ack); + + /* + * Repeat if there is more work to be done: + */ + rdmsrl(MSR_CORE_PERF_GLOBAL_STATUS, status); + if (status) + goto again; +out: + /* + * Restore - do not reenable when global enable is off: + */ + wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, saved_global); +} + +void smp_perf_counter_interrupt(struct pt_regs *regs) +{ + irq_enter(); + inc_irq_stat(apic_perf_irqs); + apic_write(APIC_LVTPC, LOCAL_PERF_VECTOR); + __smp_perf_counter_interrupt(regs, 0); + + irq_exit(); +} + +/* + * This handler is triggered by NMI contexts: + */ +void perf_counter_notify(struct pt_regs *regs) +{ + struct cpu_hw_counters *cpuc; + unsigned long flags; + int bit, cpu; + + local_irq_save(flags); + cpu = smp_processor_id(); + cpuc = &per_cpu(cpu_hw_counters, cpu); + + for_each_bit(bit, cpuc->used, X86_PMC_IDX_MAX) { + struct perf_counter *counter = cpuc->counters[bit]; + + if (!counter) + continue; + + if (counter->wakeup_pending) { + counter->wakeup_pending = 0; + wake_up(&counter->waitq); + } + } + + local_irq_restore(flags); +} + +void __cpuinit perf_counters_lapic_init(int nmi) +{ + u32 apic_val; + + if (!perf_counters_initialized) + return; + /* + * Enable the performance counter vector in the APIC LVT: + */ + 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); +} + +static int __kprobes +perf_counter_nmi_handler(struct notifier_block *self, + unsigned long cmd, void *__args) +{ + struct die_args *args = __args; + struct pt_regs *regs; + + if (likely(cmd != DIE_NMI_IPI)) + return NOTIFY_DONE; + + regs = args->regs; + + apic_write(APIC_LVTPC, APIC_DM_NMI); + __smp_perf_counter_interrupt(regs, 1); + + return NOTIFY_STOP; +} + +static __read_mostly struct notifier_block perf_counter_nmi_notifier = { + .notifier_call = perf_counter_nmi_handler +}; + +void __init init_hw_perf_counters(void) +{ + union cpuid10_eax eax; + unsigned int ebx; + unsigned int unused; + union cpuid10_edx edx; + + if (!cpu_has(&boot_cpu_data, X86_FEATURE_ARCH_PERFMON)) + return; + + /* + * Check whether the Architectural PerfMon supports + * Branch Misses Retired Event or not. + */ + cpuid(10, &eax.full, &ebx, &unused, &edx.full); + if (eax.split.mask_length <= ARCH_PERFMON_BRANCH_MISSES_RETIRED) + return; + + printk(KERN_INFO "Intel Performance Monitoring support detected.\n"); + + printk(KERN_INFO "... version: %d\n", eax.split.version_id); + printk(KERN_INFO "... num counters: %d\n", eax.split.num_counters); + nr_counters_generic = eax.split.num_counters; + if (nr_counters_generic > X86_PMC_MAX_GENERIC) { + nr_counters_generic = X86_PMC_MAX_GENERIC; + WARN(1, KERN_ERR "hw perf counters %d > max(%d), clipping!", + nr_counters_generic, X86_PMC_MAX_GENERIC); + } + perf_counter_mask = (1 << nr_counters_generic) - 1; + perf_max_counters = nr_counters_generic; + + printk(KERN_INFO "... bit width: %d\n", eax.split.bit_width); + counter_value_mask = (1ULL << eax.split.bit_width) - 1; + printk(KERN_INFO "... value mask: %016Lx\n", counter_value_mask); + + printk(KERN_INFO "... mask length: %d\n", eax.split.mask_length); + + nr_counters_fixed = edx.split.num_counters_fixed; + if (nr_counters_fixed > X86_PMC_MAX_FIXED) { + nr_counters_fixed = X86_PMC_MAX_FIXED; + WARN(1, KERN_ERR "hw perf counters fixed %d > max(%d), clipping!", + nr_counters_fixed, X86_PMC_MAX_FIXED); + } + printk(KERN_INFO "... fixed counters: %d\n", nr_counters_fixed); + + perf_counter_mask |= ((1LL << nr_counters_fixed)-1) << X86_PMC_IDX_FIXED; + + printk(KERN_INFO "... counter mask: %016Lx\n", perf_counter_mask); + perf_counters_initialized = true; + + perf_counters_lapic_init(0); + register_die_notifier(&perf_counter_nmi_notifier); +} + +static void pmc_generic_read(struct perf_counter *counter) +{ + x86_perf_counter_update(counter, &counter->hw, counter->hw.idx); +} + +static const struct hw_perf_counter_ops x86_perf_counter_ops = { + .enable = pmc_generic_enable, + .disable = pmc_generic_disable, + .read = pmc_generic_read, +}; + +const struct hw_perf_counter_ops * +hw_perf_counter_init(struct perf_counter *counter) +{ + int err; + + err = __hw_perf_counter_init(counter); + if (err) + return NULL; + + return &x86_perf_counter_ops; +} diff --git a/arch/x86/kernel/cpu/perfctr-watchdog.c b/arch/x86/kernel/cpu/perfctr-watchdog.c index 9abd48b2267..d6f5b9fbde3 100644 --- a/arch/x86/kernel/cpu/perfctr-watchdog.c +++ b/arch/x86/kernel/cpu/perfctr-watchdog.c @@ -20,7 +20,7 @@ #include <linux/kprobes.h> #include <asm/apic.h> -#include <asm/intel_arch_perfmon.h> +#include <asm/perf_counter.h> struct nmi_watchdog_ctlblk { unsigned int cccr_msr; diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c index c689d19e35a..11b93cabdf7 100644 --- a/arch/x86/kernel/crash.c +++ b/arch/x86/kernel/crash.c @@ -24,7 +24,7 @@ #include <asm/apic.h> #include <asm/hpet.h> #include <linux/kdebug.h> -#include <asm/smp.h> +#include <asm/cpu.h> #include <asm/reboot.h> #include <asm/virtext.h> diff --git a/arch/x86/kernel/dumpstack_64.c b/arch/x86/kernel/dumpstack_64.c index c302d070704..d35db5993fd 100644 --- a/arch/x86/kernel/dumpstack_64.c +++ b/arch/x86/kernel/dumpstack_64.c @@ -106,7 +106,8 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, const struct stacktrace_ops *ops, void *data) { const unsigned cpu = get_cpu(); - unsigned long *irqstack_end = (unsigned long *)cpu_pda(cpu)->irqstackptr; + unsigned long *irq_stack_end = + (unsigned long *)per_cpu(irq_stack_ptr, cpu); unsigned used = 0; struct thread_info *tinfo; int graph = 0; @@ -160,23 +161,23 @@ void dump_trace(struct task_struct *task, struct pt_regs *regs, stack = (unsigned long *) estack_end[-2]; continue; } - if (irqstack_end) { - unsigned long *irqstack; - irqstack = irqstack_end - - (IRQSTACKSIZE - 64) / sizeof(*irqstack); + if (irq_stack_end) { + unsigned long *irq_stack; + irq_stack = irq_stack_end - + (IRQ_STACK_SIZE - 64) / sizeof(*irq_stack); - if (stack >= irqstack && stack < irqstack_end) { + if (stack >= irq_stack && stack < irq_stack_end) { if (ops->stack(data, "IRQ") < 0) break; bp = print_context_stack(tinfo, stack, bp, - ops, data, irqstack_end, &graph); + ops, data, irq_stack_end, &graph); /* * We link to the next stack (which would be * the process stack normally) the last * pointer (index -1 to end) in the IRQ stack: */ - stack = (unsigned long *) (irqstack_end[-1]); - irqstack_end = NULL; + stack = (unsigned long *) (irq_stack_end[-1]); + irq_stack_end = NULL; ops->stack(data, "EOI"); continue; } @@ -199,10 +200,10 @@ show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs, unsigned long *stack; int i; const int cpu = smp_processor_id(); - unsigned long *irqstack_end = - (unsigned long *) (cpu_pda(cpu)->irqstackptr); - unsigned long *irqstack = - (unsigned long *) (cpu_pda(cpu)->irqstackptr - IRQSTACKSIZE); + unsigned long *irq_stack_end = + (unsigned long *)(per_cpu(irq_stack_ptr, cpu)); + unsigned long *irq_stack = + (unsigned long *)(per_cpu(irq_stack_ptr, cpu) - IRQ_STACK_SIZE); /* * debugging aid: "show_stack(NULL, NULL);" prints the @@ -218,9 +219,9 @@ show_stack_log_lvl(struct task_struct *task, struct pt_regs *regs, stack = sp; for (i = 0; i < kstack_depth_to_print; i++) { - if (stack >= irqstack && stack <= irqstack_end) { - if (stack == irqstack_end) { - stack = (unsigned long *) (irqstack_end[-1]); + if (stack >= irq_stack && stack <= irq_stack_end) { + if (stack == irq_stack_end) { + stack = (unsigned long *) (irq_stack_end[-1]); printk(" <EOI> "); } } else { @@ -241,7 +242,7 @@ void show_registers(struct pt_regs *regs) int i; unsigned long sp; const int cpu = smp_processor_id(); - struct task_struct *cur = cpu_pda(cpu)->pcurrent; + struct task_struct *cur = current; sp = regs->sp; printk("CPU %d ", cpu); diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index e28c7a98779..c092e7d2686 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -52,6 +52,7 @@ #include <asm/irqflags.h> #include <asm/paravirt.h> #include <asm/ftrace.h> +#include <asm/percpu.h> /* Avoid __ASSEMBLER__'ifying <linux/audit.h> just for this. */ #include <linux/elf-em.h> @@ -209,7 +210,7 @@ ENTRY(native_usergs_sysret64) /* %rsp:at FRAMEEND */ .macro FIXUP_TOP_OF_STACK tmp offset=0 - movq %gs:pda_oldrsp,\tmp + movq PER_CPU_VAR(old_rsp),\tmp movq \tmp,RSP+\offset(%rsp) movq $__USER_DS,SS+\offset(%rsp) movq $__USER_CS,CS+\offset(%rsp) @@ -220,7 +221,7 @@ ENTRY(native_usergs_sysret64) .macro RESTORE_TOP_OF_STACK tmp offset=0 movq RSP+\offset(%rsp),\tmp - movq \tmp,%gs:pda_oldrsp + movq \tmp,PER_CPU_VAR(old_rsp) movq EFLAGS+\offset(%rsp),\tmp movq \tmp,R11+\offset(%rsp) .endm @@ -336,15 +337,15 @@ ENTRY(save_args) je 1f SWAPGS /* - * irqcount is used to check if a CPU is already on an interrupt stack + * irq_count is used to check if a CPU is already on an interrupt stack * or not. While this is essentially redundant with preempt_count it is * a little cheaper to use a separate counter in the PDA (short of * moving irq_enter into assembly, which would be too much work) */ -1: incl %gs:pda_irqcount +1: incl PER_CPU_VAR(irq_count) jne 2f popq_cfi %rax /* move return address... */ - mov %gs:pda_irqstackptr,%rsp + mov PER_CPU_VAR(irq_stack_ptr),%rsp EMPTY_FRAME 0 pushq_cfi %rax /* ... to the new stack */ /* @@ -467,7 +468,7 @@ END(ret_from_fork) ENTRY(system_call) CFI_STARTPROC simple CFI_SIGNAL_FRAME - CFI_DEF_CFA rsp,PDA_STACKOFFSET + CFI_DEF_CFA rsp,KERNEL_STACK_OFFSET CFI_REGISTER rip,rcx /*CFI_REGISTER rflags,r11*/ SWAPGS_UNSAFE_STACK @@ -478,8 +479,8 @@ ENTRY(system_call) */ ENTRY(system_call_after_swapgs) - movq %rsp,%gs:pda_oldrsp - movq %gs:pda_kernelstack,%rsp + movq %rsp,PER_CPU_VAR(old_rsp) + movq PER_CPU_VAR(kernel_stack),%rsp /* * No need to follow this irqs off/on section - it's straight * and short: @@ -522,7 +523,7 @@ sysret_check: CFI_REGISTER rip,rcx RESTORE_ARGS 0,-ARG_SKIP,1 /*CFI_REGISTER rflags,r11*/ - movq %gs:pda_oldrsp, %rsp + movq PER_CPU_VAR(old_rsp), %rsp USERGS_SYSRET64 CFI_RESTORE_STATE @@ -832,11 +833,11 @@ common_interrupt: XCPT_FRAME addq $-0x80,(%rsp) /* Adjust vector to [-256,-1] range */ interrupt do_IRQ - /* 0(%rsp): oldrsp-ARGOFFSET */ + /* 0(%rsp): old_rsp-ARGOFFSET */ ret_from_intr: DISABLE_INTERRUPTS(CLBR_NONE) TRACE_IRQS_OFF - decl %gs:pda_irqcount + decl PER_CPU_VAR(irq_count) leaveq CFI_DEF_CFA_REGISTER rsp CFI_ADJUST_CFA_OFFSET -8 @@ -1024,6 +1025,11 @@ apicinterrupt ERROR_APIC_VECTOR \ apicinterrupt SPURIOUS_APIC_VECTOR \ spurious_interrupt smp_spurious_interrupt +#ifdef CONFIG_PERF_COUNTERS +apicinterrupt LOCAL_PERF_VECTOR \ + perf_counter_interrupt smp_perf_counter_interrupt +#endif + /* * Exception entry points. */ @@ -1072,10 +1078,10 @@ ENTRY(\sym) TRACE_IRQS_OFF movq %rsp,%rdi /* pt_regs pointer */ xorl %esi,%esi /* no error code */ - movq %gs:pda_data_offset, %rbp - subq $EXCEPTION_STKSZ, per_cpu__init_tss + TSS_ist + (\ist - 1) * 8(%rbp) + PER_CPU(init_tss, %rbp) + subq $EXCEPTION_STKSZ, TSS_ist + (\ist - 1) * 8(%rbp) call \do_sym - addq $EXCEPTION_STKSZ, per_cpu__init_tss + TSS_ist + (\ist - 1) * 8(%rbp) + addq $EXCEPTION_STKSZ, TSS_ist + (\ist - 1) * 8(%rbp) jmp paranoid_exit /* %ebx: no swapgs flag */ CFI_ENDPROC END(\sym) @@ -1259,14 +1265,14 @@ ENTRY(call_softirq) CFI_REL_OFFSET rbp,0 mov %rsp,%rbp CFI_DEF_CFA_REGISTER rbp - incl %gs:pda_irqcount - cmove %gs:pda_irqstackptr,%rsp + incl PER_CPU_VAR(irq_count) + cmove PER_CPU_VAR(irq_stack_ptr),%rsp push %rbp # backlink for old unwinder call __do_softirq leaveq CFI_DEF_CFA_REGISTER rsp CFI_ADJUST_CFA_OFFSET -8 - decl %gs:pda_irqcount + decl PER_CPU_VAR(irq_count) ret CFI_ENDPROC END(call_softirq) @@ -1296,15 +1302,15 @@ ENTRY(xen_do_hypervisor_callback) # do_hypervisor_callback(struct *pt_regs) movq %rdi, %rsp # we don't return, adjust the stack frame CFI_ENDPROC DEFAULT_FRAME -11: incl %gs:pda_irqcount +11: incl PER_CPU_VAR(irq_count) movq %rsp,%rbp CFI_DEF_CFA_REGISTER rbp - cmovzq %gs:pda_irqstackptr,%rsp + cmovzq PER_CPU_VAR(irq_stack_ptr),%rsp pushq %rbp # backlink for old unwinder call xen_evtchn_do_upcall popq %rsp CFI_DEF_CFA_REGISTER rsp - decl %gs:pda_irqcount + decl PER_CPU_VAR(irq_count) jmp error_exit CFI_ENDPROC END(do_hypervisor_callback) diff --git a/arch/x86/kernel/head64.c b/arch/x86/kernel/head64.c index b9a4d8c4b93..af67d3227ea 100644 --- a/arch/x86/kernel/head64.c +++ b/arch/x86/kernel/head64.c @@ -26,27 +26,6 @@ #include <asm/bios_ebda.h> #include <asm/trampoline.h> -/* boot cpu pda */ -static struct x8664_pda _boot_cpu_pda; - -#ifdef CONFIG_SMP -/* - * We install an empty cpu_pda pointer table to indicate to early users - * (numa_set_node) that the cpu_pda pointer table for cpus other than - * the boot cpu is not yet setup. - */ -static struct x8664_pda *__cpu_pda[NR_CPUS] __initdata; -#else -static struct x8664_pda *__cpu_pda[NR_CPUS] __read_mostly; -#endif - -void __init x86_64_init_pda(void) -{ - _cpu_pda = __cpu_pda; - cpu_pda(0) = &_boot_cpu_pda; - pda_init(0); -} - static void __init zap_identity_mappings(void) { pgd_t *pgd = pgd_offset_k(0UL); @@ -112,7 +91,7 @@ void __init x86_64_start_kernel(char * real_mode_data) if (console_loglevel == 10) early_printk("Kernel alive\n"); - x86_64_init_pda(); + pda_init(0); x86_64_start_reservations(real_mode_data); } diff --git a/arch/x86/kernel/head_64.S b/arch/x86/kernel/head_64.S index 0e275d49556..c8ace880661 100644 --- a/arch/x86/kernel/head_64.S +++ b/arch/x86/kernel/head_64.S @@ -19,6 +19,7 @@ #include <asm/msr.h> #include <asm/cache.h> #include <asm/processor-flags.h> +#include <asm/percpu.h> #ifdef CONFIG_PARAVIRT #include <asm/asm-offsets.h> @@ -204,6 +205,23 @@ ENTRY(secondary_startup_64) pushq $0 popfq +#ifdef CONFIG_SMP + /* + * early_gdt_base should point to the gdt_page in static percpu init + * data area. Computing this requires two symbols - __per_cpu_load + * and per_cpu__gdt_page. As linker can't do no such relocation, do + * it by hand. As early_gdt_descr is manipulated by C code for + * secondary CPUs, this should be done only once for the boot CPU + * when early_gdt_descr_base contains zero. + */ + movq early_gdt_descr_base(%rip), %rax + testq %rax, %rax + jnz 1f + movq $__per_cpu_load, %rax + addq $per_cpu__gdt_page, %rax + movq %rax, early_gdt_descr_base(%rip) +1: +#endif /* * We must switch to a new descriptor in kernel space for the GDT * because soon the kernel won't have access anymore to the userspace @@ -226,12 +244,18 @@ ENTRY(secondary_startup_64) movl %eax,%fs movl %eax,%gs - /* - * Setup up a dummy PDA. this is just for some early bootup code - * that does in_interrupt() - */ + /* Set up %gs. + * + * On SMP, %gs should point to the per-cpu area. For initial + * boot, make %gs point to the init data section. For a + * secondary CPU,initial_gs should be set to its pda address + * before the CPU runs this code. + * + * On UP, initial_gs points to PER_CPU_VAR(__pda) and doesn't + * change. + */ movl $MSR_GS_BASE,%ecx - movq $empty_zero_page,%rax + movq initial_gs(%rip),%rax movq %rax,%rdx shrq $32,%rdx wrmsr @@ -257,6 +281,12 @@ ENTRY(secondary_startup_64) .align 8 ENTRY(initial_code) .quad x86_64_start_kernel + ENTRY(initial_gs) +#ifdef CONFIG_SMP + .quad __per_cpu_load +#else + .quad PER_CPU_VAR(__pda) +#endif __FINITDATA ENTRY(stack_start) @@ -401,7 +431,12 @@ NEXT_PAGE(level2_spare_pgt) .globl early_gdt_descr early_gdt_descr: .word GDT_ENTRIES*8-1 - .quad per_cpu__gdt_page +#ifdef CONFIG_SMP +early_gdt_descr_base: + .quad 0x0000000000000000 +#else + .quad per_cpu__gdt_page +#endif ENTRY(phys_base) /* This must match the first entry in level2_kernel_pgt */ diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 1c4a1302536..f7966039072 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -46,6 +46,7 @@ #include <asm/idle.h> #include <asm/io.h> #include <asm/smp.h> +#include <asm/cpu.h> #include <asm/desc.h> #include <asm/proto.h> #include <asm/acpi.h> @@ -82,11 +83,11 @@ static DEFINE_SPINLOCK(vector_lock); int nr_ioapic_registers[MAX_IO_APICS]; /* I/O APIC entries */ -struct mp_config_ioapic mp_ioapics[MAX_IO_APICS]; +struct mpc_ioapic mp_ioapics[MAX_IO_APICS]; int nr_ioapics; /* MP IRQ source entries */ -struct mp_config_intsrc mp_irqs[MAX_IRQ_SOURCES]; +struct mpc_intsrc mp_irqs[MAX_IRQ_SOURCES]; /* # of MP IRQ source entries */ int mp_irq_entries; @@ -356,7 +357,7 @@ set_extra_move_desc(struct irq_desc *desc, const struct cpumask *mask) if (!cfg->move_in_progress) { /* it means that domain is not changed */ - if (!cpumask_intersects(&desc->affinity, mask)) + if (!cpumask_intersects(desc->affinity, mask)) cfg->move_desc_pending = 1; } } @@ -386,7 +387,7 @@ struct io_apic { static __attribute_const__ struct io_apic __iomem *io_apic_base(int idx) { return (void __iomem *) __fix_to_virt(FIX_IO_APIC_BASE_0 + idx) - + (mp_ioapics[idx].mp_apicaddr & ~PAGE_MASK); + + (mp_ioapics[idx].apicaddr & ~PAGE_MASK); } static inline unsigned int io_apic_read(unsigned int apic, unsigned int reg) @@ -579,9 +580,9 @@ set_desc_affinity(struct irq_desc *desc, const struct cpumask *mask) if (assign_irq_vector(irq, cfg, mask)) return BAD_APICID; - cpumask_and(&desc->affinity, cfg->domain, mask); + cpumask_and(desc->affinity, cfg->domain, mask); set_extra_move_desc(desc, mask); - return cpu_mask_to_apicid_and(&desc->affinity, cpu_online_mask); + return cpu_mask_to_apicid_and(desc->affinity, cpu_online_mask); } static void @@ -944,10 +945,10 @@ static int find_irq_entry(int apic, int pin, int type) int i; for (i = 0; i < mp_irq_entries; i++) - if (mp_irqs[i].mp_irqtype == type && - (mp_irqs[i].mp_dstapic == mp_ioapics[apic].mp_apicid || - mp_irqs[i].mp_dstapic == MP_APIC_ALL) && - mp_irqs[i].mp_dstirq == pin) + if (mp_irqs[i].irqtype == type && + (mp_irqs[i].dstapic == mp_ioapics[apic].apicid || + mp_irqs[i].dstapic == MP_APIC_ALL) && + mp_irqs[i].dstirq == pin) return i; return -1; @@ -961,13 +962,13 @@ static int __init find_isa_irq_pin(int irq, int type) int i; for (i = 0; i < mp_irq_entries; i++) { - int lbus = mp_irqs[i].mp_srcbus; + int lbus = mp_irqs[i].srcbus; if (test_bit(lbus, mp_bus_not_pci) && - (mp_irqs[i].mp_irqtype == type) && - (mp_irqs[i].mp_srcbusirq == irq)) + (mp_irqs[i].irqtype == type) && + (mp_irqs[i].srcbusirq == irq)) - return mp_irqs[i].mp_dstirq; + return mp_irqs[i].dstirq; } return -1; } @@ -977,17 +978,17 @@ static int __init find_isa_irq_apic(int irq, int type) int i; for (i = 0; i < mp_irq_entries; i++) { - int lbus = mp_irqs[i].mp_srcbus; + int lbus = mp_irqs[i].srcbus; if (test_bit(lbus, mp_bus_not_pci) && - (mp_irqs[i].mp_irqtype == type) && - (mp_irqs[i].mp_srcbusirq == irq)) + (mp_irqs[i].irqtype == type) && + (mp_irqs[i].srcbusirq == irq)) break; } if (i < mp_irq_entries) { int apic; for(apic = 0; apic < nr_ioapics; apic++) { - if (mp_ioapics[apic].mp_apicid == mp_irqs[i].mp_dstapic) + if (mp_ioapics[apic].apicid == mp_irqs[i].dstapic) return apic; } } @@ -1012,23 +1013,23 @@ int IO_APIC_get_PCI_irq_vector(int bus, int slot, int pin) return -1; } for (i = 0; i < mp_irq_entries; i++) { - int lbus = mp_irqs[i].mp_srcbus; + int lbus = mp_irqs[i].srcbus; for (apic = 0; apic < nr_ioapics; apic++) - if (mp_ioapics[apic].mp_apicid == mp_irqs[i].mp_dstapic || - mp_irqs[i].mp_dstapic == MP_APIC_ALL) + if (mp_ioapics[apic].apicid == mp_irqs[i].dstapic || + mp_irqs[i].dstapic == MP_APIC_ALL) break; if (!test_bit(lbus, mp_bus_not_pci) && - !mp_irqs[i].mp_irqtype && + !mp_irqs[i].irqtype && (bus == lbus) && - (slot == ((mp_irqs[i].mp_srcbusirq >> 2) & 0x1f))) { - int irq = pin_2_irq(i,apic,mp_irqs[i].mp_dstirq); + (slot == ((mp_irqs[i].srcbusirq >> 2) & 0x1f))) { + int irq = pin_2_irq(i, apic, mp_irqs[i].dstirq); if (!(apic || IO_APIC_IRQ(irq))) continue; - if (pin == (mp_irqs[i].mp_srcbusirq & 3)) + if (pin == (mp_irqs[i].srcbusirq & 3)) return irq; /* * Use the first all-but-pin matching entry as a @@ -1071,7 +1072,7 @@ static int EISA_ELCR(unsigned int irq) * EISA conforming in the MP table, that means its trigger type must * be read in from the ELCR */ -#define default_EISA_trigger(idx) (EISA_ELCR(mp_irqs[idx].mp_srcbusirq)) +#define default_EISA_trigger(idx) (EISA_ELCR(mp_irqs[idx].srcbusirq)) #define default_EISA_polarity(idx) default_ISA_polarity(idx) /* PCI interrupts are always polarity one level triggered, @@ -1088,13 +1089,13 @@ static int EISA_ELCR(unsigned int irq) static int MPBIOS_polarity(int idx) { - int bus = mp_irqs[idx].mp_srcbus; + int bus = mp_irqs[idx].srcbus; int polarity; /* * Determine IRQ line polarity (high active or low active): */ - switch (mp_irqs[idx].mp_irqflag & 3) + switch (mp_irqs[idx].irqflag & 3) { case 0: /* conforms, ie. bus-type dependent polarity */ if (test_bit(bus, mp_bus_not_pci)) @@ -1130,13 +1131,13 @@ static int MPBIOS_polarity(int idx) static int MPBIOS_trigger(int idx) { - int bus = mp_irqs[idx].mp_srcbus; + int bus = mp_irqs[idx].srcbus; int trigger; /* * Determine IRQ trigger mode (edge or level sensitive): */ - switch ((mp_irqs[idx].mp_irqflag>>2) & 3) + switch ((mp_irqs[idx].irqflag>>2) & 3) { case 0: /* conforms, ie. bus-type dependent */ if (test_bit(bus, mp_bus_not_pci)) @@ -1214,16 +1215,16 @@ int (*ioapic_renumber_irq)(int ioapic, int irq); static int pin_2_irq(int idx, int apic, int pin) { int irq, i; - int bus = mp_irqs[idx].mp_srcbus; + int bus = mp_irqs[idx].srcbus; /* * Debugging check, we are in big trouble if this message pops up! */ - if (mp_irqs[idx].mp_dstirq != pin) + if (mp_irqs[idx].dstirq != pin) printk(KERN_ERR "broken BIOS or MPTABLE parser, ayiee!!\n"); if (test_bit(bus, mp_bus_not_pci)) { - irq = mp_irqs[idx].mp_srcbusirq; + irq = mp_irqs[idx].srcbusirq; } else { /* * PCI IRQs are mapped in order @@ -1566,14 +1567,14 @@ static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, struct irq_de apic_printk(APIC_VERBOSE,KERN_DEBUG "IOAPIC[%d]: Set routing entry (%d-%d -> 0x%x -> " "IRQ %d Mode:%i Active:%i)\n", - apic, mp_ioapics[apic].mp_apicid, pin, cfg->vector, + apic, mp_ioapics[apic].apicid, pin, cfg->vector, irq, trigger, polarity); - if (setup_ioapic_entry(mp_ioapics[apic].mp_apicid, irq, &entry, + if (setup_ioapic_entry(mp_ioapics[apic].apicid, irq, &entry, dest, trigger, polarity, cfg->vector)) { printk("Failed to setup ioapic entry for ioapic %d, pin %d\n", - mp_ioapics[apic].mp_apicid, pin); + mp_ioapics[apic].apicid, pin); __clear_irq_vector(irq, cfg); return; } @@ -1604,12 +1605,10 @@ static void __init setup_IO_APIC_irqs(void) notcon = 1; apic_printk(APIC_VERBOSE, KERN_DEBUG " %d-%d", - mp_ioapics[apic].mp_apicid, - pin); + mp_ioapics[apic].apicid, pin); } else apic_printk(APIC_VERBOSE, " %d-%d", - mp_ioapics[apic].mp_apicid, - pin); + mp_ioapics[apic].apicid, pin); continue; } if (notcon) { @@ -1699,7 +1698,7 @@ __apicdebuginit(void) print_IO_APIC(void) printk(KERN_DEBUG "number of MP IRQ sources: %d.\n", mp_irq_entries); for (i = 0; i < nr_ioapics; i++) printk(KERN_DEBUG "number of IO-APIC #%d registers: %d.\n", - mp_ioapics[i].mp_apicid, nr_ioapic_registers[i]); + mp_ioapics[i].apicid, nr_ioapic_registers[i]); /* * We are a bit conservative about what we expect. We have to @@ -1719,7 +1718,7 @@ __apicdebuginit(void) print_IO_APIC(void) spin_unlock_irqrestore(&ioapic_lock, flags); printk("\n"); - printk(KERN_DEBUG "IO APIC #%d......\n", mp_ioapics[apic].mp_apicid); + printk(KERN_DEBUG "IO APIC #%d......\n", mp_ioapics[apic].apicid); printk(KERN_DEBUG ".... register #00: %08X\n", reg_00.raw); printk(KERN_DEBUG "....... : physical APIC id: %02X\n", reg_00.bits.ID); printk(KERN_DEBUG "....... : Delivery Type: %X\n", reg_00.bits.delivery_type); @@ -2121,14 +2120,14 @@ static void __init setup_ioapic_ids_from_mpc(void) reg_00.raw = io_apic_read(apic, 0); spin_unlock_irqrestore(&ioapic_lock, flags); - old_id = mp_ioapics[apic].mp_apicid; + old_id = mp_ioapics[apic].apicid; - if (mp_ioapics[apic].mp_apicid >= get_physical_broadcast()) { + if (mp_ioapics[apic].apicid >= get_physical_broadcast()) { printk(KERN_ERR "BIOS bug, IO-APIC#%d ID is %d in the MPC table!...\n", - apic, mp_ioapics[apic].mp_apicid); + apic, mp_ioapics[apic].apicid); printk(KERN_ERR "... fixing up to %d. (tell your hw vendor)\n", reg_00.bits.ID); - mp_ioapics[apic].mp_apicid = reg_00.bits.ID; + mp_ioapics[apic].apicid = reg_00.bits.ID; } /* @@ -2137,9 +2136,9 @@ static void __init setup_ioapic_ids_from_mpc(void) * 'stuck on smp_invalidate_needed IPI wait' messages. */ if (check_apicid_used(phys_id_present_map, - mp_ioapics[apic].mp_apicid)) { + mp_ioapics[apic].apicid)) { printk(KERN_ERR "BIOS bug, IO-APIC#%d ID %d is already used!...\n", - apic, mp_ioapics[apic].mp_apicid); + apic, mp_ioapics[apic].apicid); for (i = 0; i < get_physical_broadcast(); i++) if (!physid_isset(i, phys_id_present_map)) break; @@ -2148,13 +2147,13 @@ static void __init setup_ioapic_ids_from_mpc(void) printk(KERN_ERR "... fixing up to %d. (tell your hw vendor)\n", i); physid_set(i, phys_id_present_map); - mp_ioapics[apic].mp_apicid = i; + mp_ioapics[apic].apicid = i; } else { physid_mask_t tmp; - tmp = apicid_to_cpu_present(mp_ioapics[apic].mp_apicid); + tmp = apicid_to_cpu_present(mp_ioapics[apic].apicid); apic_printk(APIC_VERBOSE, "Setting %d in the " "phys_id_present_map\n", - mp_ioapics[apic].mp_apicid); + mp_ioapics[apic].apicid); physids_or(phys_id_present_map, phys_id_present_map, tmp); } @@ -2163,11 +2162,11 @@ static void __init setup_ioapic_ids_from_mpc(void) * We need to adjust the IRQ routing table * if the ID changed. */ - if (old_id != mp_ioapics[apic].mp_apicid) + if (old_id != mp_ioapics[apic].apicid) for (i = 0; i < mp_irq_entries; i++) - if (mp_irqs[i].mp_dstapic == old_id) - mp_irqs[i].mp_dstapic - = mp_ioapics[apic].mp_apicid; + if (mp_irqs[i].dstapic == old_id) + mp_irqs[i].dstapic + = mp_ioapics[apic].apicid; /* * Read the right value from the MPC table and @@ -2175,9 +2174,9 @@ static void __init setup_ioapic_ids_from_mpc(void) */ apic_printk(APIC_VERBOSE, KERN_INFO "...changing IO-APIC physical APIC ID to %d ...", - mp_ioapics[apic].mp_apicid); + mp_ioapics[apic].apicid); - reg_00.bits.ID = mp_ioapics[apic].mp_apicid; + reg_00.bits.ID = mp_ioapics[apic].apicid; spin_lock_irqsave(&ioapic_lock, flags); io_apic_write(apic, 0, reg_00.raw); spin_unlock_irqrestore(&ioapic_lock, flags); @@ -2188,7 +2187,7 @@ static void __init setup_ioapic_ids_from_mpc(void) spin_lock_irqsave(&ioapic_lock, flags); reg_00.raw = io_apic_read(apic, 0); spin_unlock_irqrestore(&ioapic_lock, flags); - if (reg_00.bits.ID != mp_ioapics[apic].mp_apicid) + if (reg_00.bits.ID != mp_ioapics[apic].apicid) printk("could not set ID!\n"); else apic_printk(APIC_VERBOSE, " ok.\n"); @@ -2383,7 +2382,7 @@ migrate_ioapic_irq_desc(struct irq_desc *desc, const struct cpumask *mask) if (cfg->move_in_progress) send_cleanup_vector(cfg); - cpumask_copy(&desc->affinity, mask); + cpumask_copy(desc->affinity, mask); } static int migrate_irq_remapped_level_desc(struct irq_desc *desc) @@ -2405,11 +2404,11 @@ static int migrate_irq_remapped_level_desc(struct irq_desc *desc) } /* everthing is clear. we have right of way */ - migrate_ioapic_irq_desc(desc, &desc->pending_mask); + migrate_ioapic_irq_desc(desc, desc->pending_mask); ret = 0; desc->status &= ~IRQ_MOVE_PENDING; - cpumask_clear(&desc->pending_mask); + cpumask_clear(desc->pending_mask); unmask: unmask_IO_APIC_irq_desc(desc); @@ -2434,7 +2433,7 @@ static void ir_irq_migration(struct work_struct *work) continue; } - desc->chip->set_affinity(irq, &desc->pending_mask); + desc->chip->set_affinity(irq, desc->pending_mask); spin_unlock_irqrestore(&desc->lock, flags); } } @@ -2448,7 +2447,7 @@ static void set_ir_ioapic_affinity_irq_desc(struct irq_desc *desc, { if (desc->status & IRQ_LEVEL) { desc->status |= IRQ_MOVE_PENDING; - cpumask_copy(&desc->pending_mask, mask); + cpumask_copy(desc->pending_mask, mask); migrate_irq_remapped_level_desc(desc); return; } @@ -2516,7 +2515,7 @@ static void irq_complete_move(struct irq_desc **descp) /* domain has not changed, but affinity did */ me = smp_processor_id(); - if (cpu_isset(me, desc->affinity)) { + if (cpumask_test_cpu(me, desc->affinity)) { *descp = desc = move_irq_desc(desc, me); /* get the new one */ cfg = desc->chip_data; @@ -3117,8 +3116,8 @@ static int ioapic_resume(struct sys_device *dev) spin_lock_irqsave(&ioapic_lock, flags); reg_00.raw = io_apic_read(dev->id, 0); - if (reg_00.bits.ID != mp_ioapics[dev->id].mp_apicid) { - reg_00.bits.ID = mp_ioapics[dev->id].mp_apicid; + if (reg_00.bits.ID != mp_ioapics[dev->id].apicid) { + reg_00.bits.ID = mp_ioapics[dev->id].apicid; io_apic_write(dev->id, 0, reg_00.raw); } spin_unlock_irqrestore(&ioapic_lock, flags); @@ -3183,7 +3182,7 @@ unsigned int create_irq_nr(unsigned int irq_want) irq = 0; spin_lock_irqsave(&vector_lock, flags); - for (new = irq_want; new < NR_IRQS; new++) { + for (new = irq_want; new < nr_irqs; new++) { if (platform_legacy_irq(new)) continue; @@ -3258,6 +3257,9 @@ static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_ms int err; unsigned dest; + if (disable_apic) + return -ENXIO; + cfg = irq_cfg(irq); err = assign_irq_vector(irq, cfg, TARGET_CPUS); if (err) @@ -3726,6 +3728,9 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) struct irq_cfg *cfg; int err; + if (disable_apic) + return -ENXIO; + cfg = irq_cfg(irq); err = assign_irq_vector(irq, cfg, TARGET_CPUS); if (!err) { @@ -3850,6 +3855,22 @@ void __init probe_nr_irqs_gsi(void) nr_irqs_gsi = nr; } +#ifdef CONFIG_SPARSE_IRQ +int __init arch_probe_nr_irqs(void) +{ + int nr; + + nr = ((8 * nr_cpu_ids) > (32 * nr_ioapics) ? + (NR_VECTORS + (8 * nr_cpu_ids)) : + (NR_VECTORS + (32 * nr_ioapics))); + + if (nr < nr_irqs && nr > nr_irqs_gsi) + nr_irqs = nr; + + return 0; +} +#endif + /* -------------------------------------------------------------------------- ACPI-based IOAPIC Configuration -------------------------------------------------------------------------- */ @@ -3984,8 +4005,8 @@ int acpi_get_override_irq(int bus_irq, int *trigger, int *polarity) return -1; for (i = 0; i < mp_irq_entries; i++) - if (mp_irqs[i].mp_irqtype == mp_INT && - mp_irqs[i].mp_srcbusirq == bus_irq) + if (mp_irqs[i].irqtype == mp_INT && + mp_irqs[i].srcbusirq == bus_irq) break; if (i >= mp_irq_entries) return -1; @@ -4039,7 +4060,7 @@ void __init setup_ioapic_dest(void) */ if (desc->status & (IRQ_NO_BALANCING | IRQ_AFFINITY_SET)) - mask = &desc->affinity; + mask = desc->affinity; else mask = TARGET_CPUS; @@ -4100,7 +4121,7 @@ void __init ioapic_init_mappings(void) ioapic_res = ioapic_setup_resources(); for (i = 0; i < nr_ioapics; i++) { if (smp_found_config) { - ioapic_phys = mp_ioapics[i].mp_apicaddr; + ioapic_phys = mp_ioapics[i].apicaddr; #ifdef CONFIG_X86_32 if (!ioapic_phys) { printk(KERN_ERR diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c index 3973e2df7f8..a6bca1d33a8 100644 --- a/arch/x86/kernel/irq.c +++ b/arch/x86/kernel/irq.c @@ -36,11 +36,7 @@ void ack_bad_irq(unsigned int irq) #endif } -#ifdef CONFIG_X86_32 -# define irq_stats(x) (&per_cpu(irq_stat, x)) -#else -# define irq_stats(x) cpu_pda(x) -#endif +#define irq_stats(x) (&per_cpu(irq_stat, x)) /* * /proc/interrupts printing: */ @@ -57,6 +53,10 @@ static int show_other_interrupts(struct seq_file *p) for_each_online_cpu(j) seq_printf(p, "%10u ", irq_stats(j)->apic_timer_irqs); seq_printf(p, " Local timer interrupts\n"); + seq_printf(p, "CNT: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", irq_stats(j)->apic_perf_irqs); + seq_printf(p, " Performance counter interrupts\n"); #endif #ifdef CONFIG_SMP seq_printf(p, "RES: "); @@ -164,6 +164,7 @@ u64 arch_irq_stat_cpu(unsigned int cpu) #ifdef CONFIG_X86_LOCAL_APIC sum += irq_stats(cpu)->apic_timer_irqs; + sum += irq_stats(cpu)->apic_perf_irqs; #endif #ifdef CONFIG_SMP sum += irq_stats(cpu)->irq_resched_count; diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index 74b9ff7341e..e0f29be8ab0 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -248,7 +248,7 @@ void fixup_irqs(void) if (irq == 2) continue; - affinity = &desc->affinity; + affinity = desc->affinity; if (cpumask_any_and(affinity, cpu_online_mask) >= nr_cpu_ids) { printk("Breaking affinity for irq %i\n", irq); affinity = cpu_all_mask; diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 63c88e6ec02..1db05247b47 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -19,6 +19,9 @@ #include <asm/io_apic.h> #include <asm/idle.h> +DEFINE_PER_CPU_SHARED_ALIGNED(irq_cpustat_t, irq_stat); +EXPORT_PER_CPU_SYMBOL(irq_stat); + /* * Probabilistic stack overflow check: * @@ -100,7 +103,7 @@ void fixup_irqs(void) /* interrupt's are disabled at this point */ spin_lock(&desc->lock); - affinity = &desc->affinity; + affinity = desc->affinity; if (!irq_has_action(irq) || cpumask_equal(affinity, cpu_online_mask)) { spin_unlock(&desc->lock); diff --git a/arch/x86/kernel/irqinit_32.c b/arch/x86/kernel/irqinit_32.c index 1507ad4e674..0bef6280f30 100644 --- a/arch/x86/kernel/irqinit_32.c +++ b/arch/x86/kernel/irqinit_32.c @@ -171,6 +171,9 @@ void __init native_init_IRQ(void) /* IPI vectors for APIC spurious and error interrupts */ alloc_intr_gate(SPURIOUS_APIC_VECTOR, spurious_interrupt); alloc_intr_gate(ERROR_APIC_VECTOR, error_interrupt); +# ifdef CONFIG_PERF_COUNTERS + alloc_intr_gate(LOCAL_PERF_VECTOR, perf_counter_interrupt); +# endif #endif #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86_MCE_P4THERMAL) diff --git a/arch/x86/kernel/irqinit_64.c b/arch/x86/kernel/irqinit_64.c index da481a1e3f3..6a71bfc51e5 100644 --- a/arch/x86/kernel/irqinit_64.c +++ b/arch/x86/kernel/irqinit_64.c @@ -150,6 +150,11 @@ static void __init apic_intr_init(void) /* IPI vectors for APIC spurious and error interrupts */ alloc_intr_gate(SPURIOUS_APIC_VECTOR, spurious_interrupt); alloc_intr_gate(ERROR_APIC_VECTOR, error_interrupt); + + /* Performance monitoring interrupt: */ +#ifdef CONFIG_PERF_COUNTERS + alloc_intr_gate(LOCAL_PERF_VECTOR, perf_counter_interrupt); +#endif } void __init native_init_IRQ(void) diff --git a/arch/x86/kernel/microcode_intel.c b/arch/x86/kernel/microcode_intel.c index b7f4c929e61..5e9f4fc5138 100644 --- a/arch/x86/kernel/microcode_intel.c +++ b/arch/x86/kernel/microcode_intel.c @@ -87,9 +87,9 @@ #include <linux/cpu.h> #include <linux/firmware.h> #include <linux/platform_device.h> +#include <linux/uaccess.h> #include <asm/msr.h> -#include <asm/uaccess.h> #include <asm/processor.h> #include <asm/microcode.h> @@ -196,7 +196,7 @@ static inline int update_match_cpu(struct cpu_signature *csig, int sig, int pf) return (!sigmatch(sig, csig->sig, pf, csig->pf)) ? 0 : 1; } -static inline int +static inline int update_match_revision(struct microcode_header_intel *mc_header, int rev) { return (mc_header->rev <= rev) ? 0 : 1; @@ -442,8 +442,8 @@ static int request_microcode_fw(int cpu, struct device *device) return ret; } - ret = generic_load_microcode(cpu, (void*)firmware->data, firmware->size, - &get_ucode_fw); + ret = generic_load_microcode(cpu, (void *)firmware->data, + firmware->size, &get_ucode_fw); release_firmware(firmware); @@ -460,7 +460,7 @@ static int request_microcode_user(int cpu, const void __user *buf, size_t size) /* We should bind the task to the CPU */ BUG_ON(cpu != raw_smp_processor_id()); - return generic_load_microcode(cpu, (void*)buf, size, &get_ucode_user); + return generic_load_microcode(cpu, (void *)buf, size, &get_ucode_user); } static void microcode_fini_cpu(int cpu) diff --git a/arch/x86/kernel/module_32.c b/arch/x86/kernel/module_32.c index 3db0a5442eb..0edd819050e 100644 --- a/arch/x86/kernel/module_32.c +++ b/arch/x86/kernel/module_32.c @@ -42,7 +42,7 @@ void module_free(struct module *mod, void *module_region) { vfree(module_region); /* FIXME: If module_region == mod->init_region, trim exception - table entries. */ + table entries. */ } /* We don't need anything special. */ @@ -113,13 +113,13 @@ int module_finalize(const Elf_Ehdr *hdr, *para = NULL; char *secstrings = (void *)hdr + sechdrs[hdr->e_shstrndx].sh_offset; - for (s = sechdrs; s < sechdrs + hdr->e_shnum; s++) { + for (s = sechdrs; s < sechdrs + hdr->e_shnum; s++) { if (!strcmp(".text", secstrings + s->sh_name)) text = s; if (!strcmp(".altinstructions", secstrings + s->sh_name)) alt = s; if (!strcmp(".smp_locks", secstrings + s->sh_name)) - locks= s; + locks = s; if (!strcmp(".parainstructions", secstrings + s->sh_name)) para = s; } diff --git a/arch/x86/kernel/module_64.c b/arch/x86/kernel/module_64.c index 6ba87830d4b..c23880b90b5 100644 --- a/arch/x86/kernel/module_64.c +++ b/arch/x86/kernel/module_64.c @@ -30,14 +30,14 @@ #include <asm/page.h> #include <asm/pgtable.h> -#define DEBUGP(fmt...) +#define DEBUGP(fmt...) #ifndef CONFIG_UML void module_free(struct module *mod, void *module_region) { vfree(module_region); /* FIXME: If module_region == mod->init_region, trim exception - table entries. */ + table entries. */ } void *module_alloc(unsigned long size) @@ -77,7 +77,7 @@ int apply_relocate_add(Elf64_Shdr *sechdrs, Elf64_Rela *rel = (void *)sechdrs[relsec].sh_addr; Elf64_Sym *sym; void *loc; - u64 val; + u64 val; DEBUGP("Applying relocate section %u to %u\n", relsec, sechdrs[relsec].sh_info); @@ -91,11 +91,11 @@ int apply_relocate_add(Elf64_Shdr *sechdrs, sym = (Elf64_Sym *)sechdrs[symindex].sh_addr + ELF64_R_SYM(rel[i].r_info); - DEBUGP("type %d st_value %Lx r_addend %Lx loc %Lx\n", - (int)ELF64_R_TYPE(rel[i].r_info), - sym->st_value, rel[i].r_addend, (u64)loc); + DEBUGP("type %d st_value %Lx r_addend %Lx loc %Lx\n", + (int)ELF64_R_TYPE(rel[i].r_info), + sym->st_value, rel[i].r_addend, (u64)loc); - val = sym->st_value + rel[i].r_addend; + val = sym->st_value + rel[i].r_addend; switch (ELF64_R_TYPE(rel[i].r_info)) { case R_X86_64_NONE: @@ -113,16 +113,16 @@ int apply_relocate_add(Elf64_Shdr *sechdrs, if ((s64)val != *(s32 *)loc) goto overflow; break; - case R_X86_64_PC32: + case R_X86_64_PC32: val -= (u64)loc; *(u32 *)loc = val; #if 0 if ((s64)val != *(s32 *)loc) - goto overflow; + goto overflow; #endif break; default: - printk(KERN_ERR "module %s: Unknown rela relocation: %Lu\n", + printk(KERN_ERR "module %s: Unknown rela relocation: %llu\n", me->name, ELF64_R_TYPE(rel[i].r_info)); return -ENOEXEC; } @@ -130,7 +130,7 @@ int apply_relocate_add(Elf64_Shdr *sechdrs, return 0; overflow: - printk(KERN_ERR "overflow in relocation type %d val %Lx\n", + printk(KERN_ERR "overflow in relocation type %d val %Lx\n", (int)ELF64_R_TYPE(rel[i].r_info), val); printk(KERN_ERR "`%s' likely not compiled with -mcmodel=kernel\n", me->name); @@ -143,13 +143,13 @@ int apply_relocate(Elf_Shdr *sechdrs, unsigned int relsec, struct module *me) { - printk("non add relocation not supported\n"); + printk(KERN_ERR "non add relocation not supported\n"); return -ENOSYS; -} +} int module_finalize(const Elf_Ehdr *hdr, - const Elf_Shdr *sechdrs, - struct module *me) + const Elf_Shdr *sechdrs, + struct module *me) { const Elf_Shdr *s, *text = NULL, *alt = NULL, *locks = NULL, *para = NULL; @@ -161,7 +161,7 @@ int module_finalize(const Elf_Ehdr *hdr, if (!strcmp(".altinstructions", secstrings + s->sh_name)) alt = s; if (!strcmp(".smp_locks", secstrings + s->sh_name)) - locks= s; + locks = s; if (!strcmp(".parainstructions", secstrings + s->sh_name)) para = s; } diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c index a649a4ccad4..fa6bb263892 100644 --- a/arch/x86/kernel/mpparse.c +++ b/arch/x86/kernel/mpparse.c @@ -144,11 +144,11 @@ static void __init MP_ioapic_info(struct mpc_ioapic *m) if (bad_ioapic(m->apicaddr)) return; - mp_ioapics[nr_ioapics].mp_apicaddr = m->apicaddr; - mp_ioapics[nr_ioapics].mp_apicid = m->apicid; - mp_ioapics[nr_ioapics].mp_type = m->type; - mp_ioapics[nr_ioapics].mp_apicver = m->apicver; - mp_ioapics[nr_ioapics].mp_flags = m->flags; + mp_ioapics[nr_ioapics].apicaddr = m->apicaddr; + mp_ioapics[nr_ioapics].apicid = m->apicid; + mp_ioapics[nr_ioapics].type = m->type; + mp_ioapics[nr_ioapics].apicver = m->apicver; + mp_ioapics[nr_ioapics].flags = m->flags; nr_ioapics++; } @@ -160,55 +160,55 @@ static void print_MP_intsrc_info(struct mpc_intsrc *m) m->srcbusirq, m->dstapic, m->dstirq); } -static void __init print_mp_irq_info(struct mp_config_intsrc *mp_irq) +static void __init print_mp_irq_info(struct mpc_intsrc *mp_irq) { apic_printk(APIC_VERBOSE, "Int: type %d, pol %d, trig %d, bus %02x," " IRQ %02x, APIC ID %x, APIC INT %02x\n", - mp_irq->mp_irqtype, mp_irq->mp_irqflag & 3, - (mp_irq->mp_irqflag >> 2) & 3, mp_irq->mp_srcbus, - mp_irq->mp_srcbusirq, mp_irq->mp_dstapic, mp_irq->mp_dstirq); + mp_irq->irqtype, mp_irq->irqflag & 3, + (mp_irq->irqflag >> 2) & 3, mp_irq->srcbus, + mp_irq->srcbusirq, mp_irq->dstapic, mp_irq->dstirq); } static void __init assign_to_mp_irq(struct mpc_intsrc *m, - struct mp_config_intsrc *mp_irq) + struct mpc_intsrc *mp_irq) { - mp_irq->mp_dstapic = m->dstapic; - mp_irq->mp_type = m->type; - mp_irq->mp_irqtype = m->irqtype; - mp_irq->mp_irqflag = m->irqflag; - mp_irq->mp_srcbus = m->srcbus; - mp_irq->mp_srcbusirq = m->srcbusirq; - mp_irq->mp_dstirq = m->dstirq; + mp_irq->dstapic = m->dstapic; + mp_irq->type = m->type; + mp_irq->irqtype = m->irqtype; + mp_irq->irqflag = m->irqflag; + mp_irq->srcbus = m->srcbus; + mp_irq->srcbusirq = m->srcbusirq; + mp_irq->dstirq = m->dstirq; } -static void __init assign_to_mpc_intsrc(struct mp_config_intsrc *mp_irq, +static void __init assign_to_mpc_intsrc(struct mpc_intsrc *mp_irq, struct mpc_intsrc *m) { - m->dstapic = mp_irq->mp_dstapic; - m->type = mp_irq->mp_type; - m->irqtype = mp_irq->mp_irqtype; - m->irqflag = mp_irq->mp_irqflag; - m->srcbus = mp_irq->mp_srcbus; - m->srcbusirq = mp_irq->mp_srcbusirq; - m->dstirq = mp_irq->mp_dstirq; + m->dstapic = mp_irq->dstapic; + m->type = mp_irq->type; + m->irqtype = mp_irq->irqtype; + m->irqflag = mp_irq->irqflag; + m->srcbus = mp_irq->srcbus; + m->srcbusirq = mp_irq->srcbusirq; + m->dstirq = mp_irq->dstirq; } -static int __init mp_irq_mpc_intsrc_cmp(struct mp_config_intsrc *mp_irq, +static int __init mp_irq_mpc_intsrc_cmp(struct mpc_intsrc *mp_irq, struct mpc_intsrc *m) { - if (mp_irq->mp_dstapic != m->dstapic) + if (mp_irq->dstapic != m->dstapic) return 1; - if (mp_irq->mp_type != m->type) + if (mp_irq->type != m->type) return 2; - if (mp_irq->mp_irqtype != m->irqtype) + if (mp_irq->irqtype != m->irqtype) return 3; - if (mp_irq->mp_irqflag != m->irqflag) + if (mp_irq->irqflag != m->irqflag) return 4; - if (mp_irq->mp_srcbus != m->srcbus) + if (mp_irq->srcbus != m->srcbus) return 5; - if (mp_irq->mp_srcbusirq != m->srcbusirq) + if (mp_irq->srcbusirq != m->srcbusirq) return 6; - if (mp_irq->mp_dstirq != m->dstirq) + if (mp_irq->dstirq != m->dstirq) return 7; return 0; @@ -417,7 +417,7 @@ static void __init construct_default_ioirq_mptable(int mpc_default_type) intsrc.type = MP_INTSRC; intsrc.irqflag = 0; /* conforming */ intsrc.srcbus = 0; - intsrc.dstapic = mp_ioapics[0].mp_apicid; + intsrc.dstapic = mp_ioapics[0].apicid; intsrc.irqtype = mp_INT; @@ -570,14 +570,14 @@ static inline void __init construct_default_ISA_mptable(int mpc_default_type) } } -static struct intel_mp_floating *mpf_found; +static struct mpf_intel *mpf_found; /* * Scan the memory blocks for an SMP configuration block. */ static void __init __get_smp_config(unsigned int early) { - struct intel_mp_floating *mpf = mpf_found; + struct mpf_intel *mpf = mpf_found; if (!mpf) return; @@ -598,9 +598,9 @@ static void __init __get_smp_config(unsigned int early) } printk(KERN_INFO "Intel MultiProcessor Specification v1.%d\n", - mpf->mpf_specification); + mpf->specification); #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86_32) - if (mpf->mpf_feature2 & (1 << 7)) { + if (mpf->feature2 & (1 << 7)) { printk(KERN_INFO " IMCR and PIC compatibility mode.\n"); pic_mode = 1; } else { @@ -611,7 +611,7 @@ static void __init __get_smp_config(unsigned int early) /* * Now see if we need to read further. */ - if (mpf->mpf_feature1 != 0) { + if (mpf->feature1 != 0) { if (early) { /* * local APIC has default address @@ -621,16 +621,16 @@ static void __init __get_smp_config(unsigned int early) } printk(KERN_INFO "Default MP configuration #%d\n", - mpf->mpf_feature1); - construct_default_ISA_mptable(mpf->mpf_feature1); + mpf->feature1); + construct_default_ISA_mptable(mpf->feature1); - } else if (mpf->mpf_physptr) { + } else if (mpf->physptr) { /* * Read the physical hardware table. Anything here will * override the defaults. */ - if (!smp_read_mpc(phys_to_virt(mpf->mpf_physptr), early)) { + if (!smp_read_mpc(phys_to_virt(mpf->physptr), early)) { #ifdef CONFIG_X86_LOCAL_APIC smp_found_config = 0; #endif @@ -688,19 +688,19 @@ static int __init smp_scan_config(unsigned long base, unsigned long length, unsigned reserve) { unsigned int *bp = phys_to_virt(base); - struct intel_mp_floating *mpf; + struct mpf_intel *mpf; apic_printk(APIC_VERBOSE, "Scan SMP from %p for %ld bytes.\n", bp, length); BUILD_BUG_ON(sizeof(*mpf) != 16); while (length > 0) { - mpf = (struct intel_mp_floating *)bp; + mpf = (struct mpf_intel *)bp; if ((*bp == SMP_MAGIC_IDENT) && - (mpf->mpf_length == 1) && + (mpf->length == 1) && !mpf_checksum((unsigned char *)bp, 16) && - ((mpf->mpf_specification == 1) - || (mpf->mpf_specification == 4))) { + ((mpf->specification == 1) + || (mpf->specification == 4))) { #ifdef CONFIG_X86_LOCAL_APIC smp_found_config = 1; #endif @@ -713,7 +713,7 @@ static int __init smp_scan_config(unsigned long base, unsigned long length, return 1; reserve_bootmem_generic(virt_to_phys(mpf), PAGE_SIZE, BOOTMEM_DEFAULT); - if (mpf->mpf_physptr) { + if (mpf->physptr) { unsigned long size = PAGE_SIZE; #ifdef CONFIG_X86_32 /* @@ -722,14 +722,14 @@ static int __init smp_scan_config(unsigned long base, unsigned long length, * the bottom is mapped now. * PC-9800's MPC table places on the very last * of physical memory; so that simply reserving - * PAGE_SIZE from mpg->mpf_physptr yields BUG() + * PAGE_SIZE from mpf->physptr yields BUG() * in reserve_bootmem. */ unsigned long end = max_low_pfn * PAGE_SIZE; - if (mpf->mpf_physptr + size > end) - size = end - mpf->mpf_physptr; + if (mpf->physptr + size > end) + size = end - mpf->physptr; #endif - reserve_bootmem_generic(mpf->mpf_physptr, size, + reserve_bootmem_generic(mpf->physptr, size, BOOTMEM_DEFAULT); } @@ -809,15 +809,15 @@ static int __init get_MP_intsrc_index(struct mpc_intsrc *m) /* not legacy */ for (i = 0; i < mp_irq_entries; i++) { - if (mp_irqs[i].mp_irqtype != mp_INT) + if (mp_irqs[i].irqtype != mp_INT) continue; - if (mp_irqs[i].mp_irqflag != 0x0f) + if (mp_irqs[i].irqflag != 0x0f) continue; - if (mp_irqs[i].mp_srcbus != m->srcbus) + if (mp_irqs[i].srcbus != m->srcbus) continue; - if (mp_irqs[i].mp_srcbusirq != m->srcbusirq) + if (mp_irqs[i].srcbusirq != m->srcbusirq) continue; if (irq_used[i]) { /* already claimed */ @@ -922,10 +922,10 @@ static int __init replace_intsrc_all(struct mpc_table *mpc, if (irq_used[i]) continue; - if (mp_irqs[i].mp_irqtype != mp_INT) + if (mp_irqs[i].irqtype != mp_INT) continue; - if (mp_irqs[i].mp_irqflag != 0x0f) + if (mp_irqs[i].irqflag != 0x0f) continue; if (nr_m_spare > 0) { @@ -1001,7 +1001,7 @@ static int __init update_mp_table(void) { char str[16]; char oem[10]; - struct intel_mp_floating *mpf; + struct mpf_intel *mpf; struct mpc_table *mpc, *mpc_new; if (!enable_update_mptable) @@ -1014,19 +1014,19 @@ static int __init update_mp_table(void) /* * Now see if we need to go further. */ - if (mpf->mpf_feature1 != 0) + if (mpf->feature1 != 0) return 0; - if (!mpf->mpf_physptr) + if (!mpf->physptr) return 0; - mpc = phys_to_virt(mpf->mpf_physptr); + mpc = phys_to_virt(mpf->physptr); if (!smp_check_mpc(mpc, oem, str)) return 0; printk(KERN_INFO "mpf: %lx\n", virt_to_phys(mpf)); - printk(KERN_INFO "mpf_physptr: %x\n", mpf->mpf_physptr); + printk(KERN_INFO "physptr: %x\n", mpf->physptr); if (mpc_new_phys && mpc->length > mpc_new_length) { mpc_new_phys = 0; @@ -1047,23 +1047,23 @@ static int __init update_mp_table(void) } printk(KERN_INFO "use in-positon replacing\n"); } else { - mpf->mpf_physptr = mpc_new_phys; + mpf->physptr = mpc_new_phys; mpc_new = phys_to_virt(mpc_new_phys); memcpy(mpc_new, mpc, mpc->length); mpc = mpc_new; /* check if we can modify that */ - if (mpc_new_phys - mpf->mpf_physptr) { - struct intel_mp_floating *mpf_new; + if (mpc_new_phys - mpf->physptr) { + struct mpf_intel *mpf_new; /* steal 16 bytes from [0, 1k) */ printk(KERN_INFO "mpf new: %x\n", 0x400 - 16); mpf_new = phys_to_virt(0x400 - 16); memcpy(mpf_new, mpf, 16); mpf = mpf_new; - mpf->mpf_physptr = mpc_new_phys; + mpf->physptr = mpc_new_phys; } - mpf->mpf_checksum = 0; - mpf->mpf_checksum -= mpf_checksum((unsigned char *)mpf, 16); - printk(KERN_INFO "mpf_physptr new: %x\n", mpf->mpf_physptr); + mpf->checksum = 0; + mpf->checksum -= mpf_checksum((unsigned char *)mpf, 16); + printk(KERN_INFO "physptr new: %x\n", mpf->physptr); } /* diff --git a/arch/x86/kernel/msr.c b/arch/x86/kernel/msr.c index 726266695b2..3cf3413ec62 100644 --- a/arch/x86/kernel/msr.c +++ b/arch/x86/kernel/msr.c @@ -35,10 +35,10 @@ #include <linux/device.h> #include <linux/cpu.h> #include <linux/notifier.h> +#include <linux/uaccess.h> #include <asm/processor.h> #include <asm/msr.h> -#include <asm/uaccess.h> #include <asm/system.h> static struct class *msr_class; diff --git a/arch/x86/kernel/nmi.c b/arch/x86/kernel/nmi.c index 7228979f1e7..23b6d9e6e4f 100644 --- a/arch/x86/kernel/nmi.c +++ b/arch/x86/kernel/nmi.c @@ -61,11 +61,7 @@ static int endflag __initdata; static inline unsigned int get_nmi_count(int cpu) { -#ifdef CONFIG_X86_64 - return cpu_pda(cpu)->__nmi_count; -#else - return nmi_count(cpu); -#endif + return per_cpu(irq_stat, cpu).__nmi_count; } static inline int mce_in_progress(void) @@ -82,12 +78,8 @@ static inline int mce_in_progress(void) */ static inline unsigned int get_timer_irqs(int cpu) { -#ifdef CONFIG_X86_64 - return read_pda(apic_timer_irqs) + read_pda(irq0_irqs); -#else return per_cpu(irq_stat, cpu).apic_timer_irqs + per_cpu(irq_stat, cpu).irq0_irqs; -#endif } #ifdef CONFIG_SMP diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c index a546f55c77b..2c00a57ccb9 100644 --- a/arch/x86/kernel/process_32.c +++ b/arch/x86/kernel/process_32.c @@ -66,9 +66,6 @@ asmlinkage void ret_from_fork(void) __asm__("ret_from_fork"); DEFINE_PER_CPU(struct task_struct *, current_task) = &init_task; EXPORT_PER_CPU_SYMBOL(current_task); -DEFINE_PER_CPU(int, cpu_number); -EXPORT_PER_CPU_SYMBOL(cpu_number); - /* * Return saved PC of a blocked thread. */ @@ -591,7 +588,7 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p) if (prev->gs | next->gs) loadsegment(gs, next->gs); - x86_write_percpu(current_task, next_p); + percpu_write(current_task, next_p); return prev_p; } diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c index 416fb9282f4..4523ff88a69 100644 --- a/arch/x86/kernel/process_64.c +++ b/arch/x86/kernel/process_64.c @@ -57,6 +57,12 @@ asmlinkage extern void ret_from_fork(void); +DEFINE_PER_CPU(struct task_struct *, current_task) = &init_task; +EXPORT_PER_CPU_SYMBOL(current_task); + +DEFINE_PER_CPU(unsigned long, old_rsp); +static DEFINE_PER_CPU(unsigned char, is_idle); + unsigned long kernel_thread_flags = CLONE_VM | CLONE_UNTRACED; static ATOMIC_NOTIFIER_HEAD(idle_notifier); @@ -75,13 +81,13 @@ EXPORT_SYMBOL_GPL(idle_notifier_unregister); void enter_idle(void) { - write_pda(isidle, 1); + percpu_write(is_idle, 1); atomic_notifier_call_chain(&idle_notifier, IDLE_START, NULL); } static void __exit_idle(void) { - if (test_and_clear_bit_pda(0, isidle) == 0) + if (x86_test_and_clear_bit_percpu(0, is_idle) == 0) return; atomic_notifier_call_chain(&idle_notifier, IDLE_END, NULL); } @@ -392,7 +398,7 @@ start_thread(struct pt_regs *regs, unsigned long new_ip, unsigned long new_sp) load_gs_index(0); regs->ip = new_ip; regs->sp = new_sp; - write_pda(oldrsp, new_sp); + percpu_write(old_rsp, new_sp); regs->cs = __USER_CS; regs->ss = __USER_DS; regs->flags = 0x200; @@ -613,13 +619,13 @@ __switch_to(struct task_struct *prev_p, struct task_struct *next_p) /* * Switch the PDA and FPU contexts. */ - prev->usersp = read_pda(oldrsp); - write_pda(oldrsp, next->usersp); - write_pda(pcurrent, next_p); + prev->usersp = percpu_read(old_rsp); + percpu_write(old_rsp, next->usersp); + percpu_write(current_task, next_p); - write_pda(kernelstack, + percpu_write(kernel_stack, (unsigned long)task_stack_page(next_p) + - THREAD_SIZE - PDA_STACKOFFSET); + THREAD_SIZE - KERNEL_STACK_OFFSET); #ifdef CONFIG_CC_STACKPROTECTOR write_pda(stack_canary, next_p->stack_canary); /* diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c index 2b46eb41643..f8536fee5c1 100644 --- a/arch/x86/kernel/reboot.c +++ b/arch/x86/kernel/reboot.c @@ -14,6 +14,7 @@ #include <asm/reboot.h> #include <asm/pci_x86.h> #include <asm/virtext.h> +#include <asm/cpu.h> #ifdef CONFIG_X86_32 # include <linux/dmi.h> diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index ae0d8042cf6..f41c4486c27 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -89,7 +89,7 @@ #include <asm/system.h> #include <asm/vsyscall.h> -#include <asm/smp.h> +#include <asm/cpu.h> #include <asm/desc.h> #include <asm/dma.h> #include <asm/iommu.h> diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index 55c46074eba..efbafbbff58 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -13,6 +13,23 @@ #include <asm/mpspec.h> #include <asm/apicdef.h> #include <asm/highmem.h> +#include <asm/proto.h> +#include <asm/cpumask.h> + +#ifdef CONFIG_DEBUG_PER_CPU_MAPS +# define DBG(x...) printk(KERN_DEBUG x) +#else +# define DBG(x...) +#endif + +/* + * Could be inside CONFIG_HAVE_SETUP_PER_CPU_AREA with other stuff but + * voyager wants cpu_number too. + */ +#ifdef CONFIG_SMP +DEFINE_PER_CPU(int, cpu_number); +EXPORT_PER_CPU_SYMBOL(cpu_number); +#endif #ifdef CONFIG_X86_LOCAL_APIC unsigned int num_processors; @@ -26,31 +43,84 @@ unsigned int max_physical_apicid; physid_mask_t phys_cpu_present_map; #endif -/* map cpu index to physical APIC ID */ +/* + * Map cpu index to physical APIC ID + */ DEFINE_EARLY_PER_CPU(u16, x86_cpu_to_apicid, BAD_APICID); DEFINE_EARLY_PER_CPU(u16, x86_bios_cpu_apicid, BAD_APICID); EXPORT_EARLY_PER_CPU_SYMBOL(x86_cpu_to_apicid); EXPORT_EARLY_PER_CPU_SYMBOL(x86_bios_cpu_apicid); #if defined(CONFIG_NUMA) && defined(CONFIG_X86_64) -#define X86_64_NUMA 1 +#define X86_64_NUMA 1 /* (used later) */ +DEFINE_PER_CPU(int, node_number) = 0; +EXPORT_PER_CPU_SYMBOL(node_number); -/* map cpu index to node index */ +/* + * Map cpu index to node index + */ DEFINE_EARLY_PER_CPU(int, x86_cpu_to_node_map, NUMA_NO_NODE); EXPORT_EARLY_PER_CPU_SYMBOL(x86_cpu_to_node_map); -/* which logical CPUs are on which nodes */ +/* + * Which logical CPUs are on which nodes + */ cpumask_t *node_to_cpumask_map; EXPORT_SYMBOL(node_to_cpumask_map); -/* setup node_to_cpumask_map */ +/* + * Setup node_to_cpumask_map + */ static void __init setup_node_to_cpumask_map(void); #else static inline void setup_node_to_cpumask_map(void) { } #endif -#if defined(CONFIG_HAVE_SETUP_PER_CPU_AREA) && defined(CONFIG_X86_SMP) +/* + * Define load_pda_offset() and per-cpu __pda for x86_64. + * load_pda_offset() is responsible for loading the offset of pda into + * %gs. + * + * On SMP, pda offset also duals as percpu base address and thus it + * should be at the start of per-cpu area. To achieve this, it's + * preallocated in vmlinux_64.lds.S directly instead of using + * DEFINE_PER_CPU(). + */ +#ifdef CONFIG_X86_64 +void __cpuinit load_pda_offset(int cpu) +{ + /* Memory clobbers used to order pda/percpu accesses */ + mb(); + wrmsrl(MSR_GS_BASE, cpu_pda(cpu)); + mb(); +} +#ifndef CONFIG_SMP +DEFINE_PER_CPU(struct x8664_pda, __pda); +#endif +EXPORT_PER_CPU_SYMBOL(__pda); +#endif /* CONFIG_SMP && CONFIG_X86_64 */ + +#ifdef CONFIG_X86_64 + +/* correctly size the local cpu masks */ +static void setup_cpu_local_masks(void) +{ + alloc_bootmem_cpumask_var(&cpu_initialized_mask); + alloc_bootmem_cpumask_var(&cpu_callin_mask); + alloc_bootmem_cpumask_var(&cpu_callout_mask); + alloc_bootmem_cpumask_var(&cpu_sibling_setup_mask); +} + +#else /* CONFIG_X86_32 */ + +static inline void setup_cpu_local_masks(void) +{ +} + +#endif /* CONFIG_X86_32 */ + +#ifdef CONFIG_HAVE_SETUP_PER_CPU_AREA /* * Copy data used in early init routines from the initial arrays to the * per cpu data areas. These arrays then become expendable and the @@ -79,78 +149,14 @@ static void __init setup_per_cpu_maps(void) #endif } -#ifdef CONFIG_X86_32 -/* - * Great future not-so-futuristic plan: make i386 and x86_64 do it - * the same way - */ +#ifdef CONFIG_X86_64 +unsigned long __per_cpu_offset[NR_CPUS] __read_mostly = { + [0] = (unsigned long)__per_cpu_load, +}; +#else unsigned long __per_cpu_offset[NR_CPUS] __read_mostly; +#endif EXPORT_SYMBOL(__per_cpu_offset); -static inline void setup_cpu_pda_map(void) { } - -#elif !defined(CONFIG_SMP) -static inline void setup_cpu_pda_map(void) { } - -#else /* CONFIG_SMP && CONFIG_X86_64 */ - -/* - * Allocate cpu_pda pointer table and array via alloc_bootmem. - */ -static void __init setup_cpu_pda_map(void) -{ - char *pda; - struct x8664_pda **new_cpu_pda; - unsigned long size; - int cpu; - - size = roundup(sizeof(struct x8664_pda), cache_line_size()); - - /* allocate cpu_pda array and pointer table */ - { - unsigned long tsize = nr_cpu_ids * sizeof(void *); - unsigned long asize = size * (nr_cpu_ids - 1); - - tsize = roundup(tsize, cache_line_size()); - new_cpu_pda = alloc_bootmem(tsize + asize); - pda = (char *)new_cpu_pda + tsize; - } - - /* initialize pointer table to static pda's */ - for_each_possible_cpu(cpu) { - if (cpu == 0) { - /* leave boot cpu pda in place */ - new_cpu_pda[0] = cpu_pda(0); - continue; - } - new_cpu_pda[cpu] = (struct x8664_pda *)pda; - new_cpu_pda[cpu]->in_bootmem = 1; - pda += size; - } - - /* point to new pointer table */ - _cpu_pda = new_cpu_pda; -} - -#endif /* CONFIG_SMP && CONFIG_X86_64 */ - -#ifdef CONFIG_X86_64 - -/* correctly size the local cpu masks */ -static void setup_cpu_local_masks(void) -{ - alloc_bootmem_cpumask_var(&cpu_initialized_mask); - alloc_bootmem_cpumask_var(&cpu_callin_mask); - alloc_bootmem_cpumask_var(&cpu_callout_mask); - alloc_bootmem_cpumask_var(&cpu_sibling_setup_mask); -} - -#else /* CONFIG_X86_32 */ - -static inline void setup_cpu_local_masks(void) -{ -} - -#endif /* CONFIG_X86_32 */ /* * Great future plan: @@ -164,9 +170,6 @@ void __init setup_per_cpu_areas(void) int cpu; unsigned long align = 1; - /* Setup cpu_pda map */ - setup_cpu_pda_map(); - /* Copy section for each CPU (we discard the original) */ old_size = PERCPU_ENOUGH_ROOM; align = max_t(unsigned long, PAGE_SIZE, align); @@ -197,8 +200,25 @@ void __init setup_per_cpu_areas(void) cpu, node, __pa(ptr)); } #endif + + memcpy(ptr, __per_cpu_load, __per_cpu_end - __per_cpu_start); per_cpu_offset(cpu) = ptr - __per_cpu_start; - memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start); + per_cpu(this_cpu_off, cpu) = per_cpu_offset(cpu); + per_cpu(cpu_number, cpu) = cpu; +#ifdef CONFIG_X86_64 + per_cpu(irq_stack_ptr, cpu) = + (char *)per_cpu(irq_stack, cpu) + IRQ_STACK_SIZE - 64; + /* + * CPU0 modified pda in the init data area, reload pda + * offset for CPU0 and clear the area for others. + */ + if (cpu == 0) + load_pda_offset(0); + else + memset(cpu_pda(cpu), 0, sizeof(*cpu_pda(cpu))); +#endif + + DBG("PERCPU: cpu %4d %p\n", cpu, ptr); } /* Setup percpu data maps */ @@ -220,6 +240,7 @@ void __init setup_per_cpu_areas(void) * Requires node_possible_map to be valid. * * Note: node_to_cpumask() is not valid until after this is done. + * (Use CONFIG_DEBUG_PER_CPU_MAPS to check this.) */ static void __init setup_node_to_cpumask_map(void) { @@ -235,6 +256,7 @@ static void __init setup_node_to_cpumask_map(void) /* allocate the map */ map = alloc_bootmem_low(nr_node_ids * sizeof(cpumask_t)); + DBG("node_to_cpumask_map at %p for %d nodes\n", map, nr_node_ids); pr_debug("Node to cpumask map at %p for %d nodes\n", map, nr_node_ids); @@ -247,17 +269,23 @@ void __cpuinit numa_set_node(int cpu, int node) { int *cpu_to_node_map = early_per_cpu_ptr(x86_cpu_to_node_map); - if (cpu_pda(cpu) && node != NUMA_NO_NODE) - cpu_pda(cpu)->nodenumber = node; - - if (cpu_to_node_map) + /* early setting, no percpu area yet */ + if (cpu_to_node_map) { cpu_to_node_map[cpu] = node; + return; + } - else if (per_cpu_offset(cpu)) - per_cpu(x86_cpu_to_node_map, cpu) = node; +#ifdef CONFIG_DEBUG_PER_CPU_MAPS + if (cpu >= nr_cpu_ids || !per_cpu_offset(cpu)) { + printk(KERN_ERR "numa_set_node: invalid cpu# (%d)\n", cpu); + dump_stack(); + return; + } +#endif + per_cpu(x86_cpu_to_node_map, cpu) = node; - else - pr_debug("Setting node for non-present cpu %d\n", cpu); + if (node != NUMA_NO_NODE) + per_cpu(node_number, cpu) = node; } void __cpuinit numa_clear_node(int cpu) @@ -274,7 +302,7 @@ void __cpuinit numa_add_cpu(int cpu) void __cpuinit numa_remove_cpu(int cpu) { - cpu_clear(cpu, node_to_cpumask_map[cpu_to_node(cpu)]); + cpu_clear(cpu, node_to_cpumask_map[early_cpu_to_node(cpu)]); } #else /* CONFIG_DEBUG_PER_CPU_MAPS */ @@ -284,7 +312,7 @@ void __cpuinit numa_remove_cpu(int cpu) */ static void __cpuinit numa_set_cpumask(int cpu, int enable) { - int node = cpu_to_node(cpu); + int node = early_cpu_to_node(cpu); cpumask_t *mask; char buf[64]; diff --git a/arch/x86/kernel/signal.c b/arch/x86/kernel/signal.c index 89bb7668041..4fa5243c206 100644 --- a/arch/x86/kernel/signal.c +++ b/arch/x86/kernel/signal.c @@ -6,7 +6,7 @@ * 2000-06-20 Pentium III FXSR, SSE support by Gareth Hughes * 2000-2002 x86-64 support by Andi Kleen */ - +#include <linux/perf_counter.h> #include <linux/sched.h> #include <linux/mm.h> #include <linux/smp.h> @@ -886,6 +886,11 @@ do_notify_resume(struct pt_regs *regs, void *unused, __u32 thread_info_flags) tracehook_notify_resume(regs); } + if (thread_info_flags & _TIF_PERF_COUNTERS) { + clear_thread_flag(TIF_PERF_COUNTERS); + perf_counter_notify(regs); + } + #ifdef CONFIG_X86_32 clear_thread_flag(TIF_IRET); #endif /* CONFIG_X86_32 */ diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index bb1a3b1fc87..869b98840fd 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -53,7 +53,6 @@ #include <asm/nmi.h> #include <asm/irq.h> #include <asm/idle.h> -#include <asm/smp.h> #include <asm/trampoline.h> #include <asm/cpu.h> #include <asm/numa.h> @@ -745,52 +744,6 @@ static void __cpuinit do_fork_idle(struct work_struct *work) complete(&c_idle->done); } -#ifdef CONFIG_X86_64 - -/* __ref because it's safe to call free_bootmem when after_bootmem == 0. */ -static void __ref free_bootmem_pda(struct x8664_pda *oldpda) -{ - if (!after_bootmem) - free_bootmem((unsigned long)oldpda, sizeof(*oldpda)); -} - -/* - * Allocate node local memory for the AP pda. - * - * Must be called after the _cpu_pda pointer table is initialized. - */ -int __cpuinit get_local_pda(int cpu) -{ - struct x8664_pda *oldpda, *newpda; - unsigned long size = sizeof(struct x8664_pda); - int node = cpu_to_node(cpu); - - if (cpu_pda(cpu) && !cpu_pda(cpu)->in_bootmem) - return 0; - - oldpda = cpu_pda(cpu); - newpda = kmalloc_node(size, GFP_ATOMIC, node); - if (!newpda) { - printk(KERN_ERR "Could not allocate node local PDA " - "for CPU %d on node %d\n", cpu, node); - - if (oldpda) - return 0; /* have a usable pda */ - else - return -1; - } - - if (oldpda) { - memcpy(newpda, oldpda, size); - free_bootmem_pda(oldpda); - } - - newpda->in_bootmem = 0; - cpu_pda(cpu) = newpda; - return 0; -} -#endif /* CONFIG_X86_64 */ - static int __cpuinit do_boot_cpu(int apicid, int cpu) /* * NOTE - on most systems this is a PHYSICAL apic ID, but on multiquad @@ -808,16 +761,6 @@ static int __cpuinit do_boot_cpu(int apicid, int cpu) }; INIT_WORK(&c_idle.work, do_fork_idle); -#ifdef CONFIG_X86_64 - /* Allocate node local memory for AP pdas */ - if (cpu > 0) { - boot_error = get_local_pda(cpu); - if (boot_error) - goto restore_state; - /* if can't get pda memory, can't start cpu */ - } -#endif - alternatives_smp_switch(1); c_idle.idle = get_idle_for_cpu(cpu); @@ -847,14 +790,17 @@ static int __cpuinit do_boot_cpu(int apicid, int cpu) set_idle_for_cpu(cpu, c_idle.idle); do_rest: -#ifdef CONFIG_X86_32 per_cpu(current_task, cpu) = c_idle.idle; +#ifdef CONFIG_X86_32 init_gdt(cpu); /* Stack for startup_32 can be just as for start_secondary onwards */ irq_ctx_init(cpu); #else - cpu_pda(cpu)->pcurrent = c_idle.idle; clear_tsk_thread_flag(c_idle.idle, TIF_FORK); + initial_gs = per_cpu_offset(cpu); + per_cpu(kernel_stack, cpu) = + (unsigned long)task_stack_page(c_idle.idle) - + KERNEL_STACK_OFFSET + THREAD_SIZE; #endif early_gdt_descr.address = (unsigned long)get_cpu_gdt_table(cpu); initial_code = (unsigned long)start_secondary; @@ -931,9 +877,7 @@ do_rest: inquire_remote_apic(apicid); } } -#ifdef CONFIG_X86_64 -restore_state: -#endif + if (boot_error) { /* Try to put things back the way they were before ... */ numa_remove_cpu(cpu); /* was set by numa_add_cpu */ @@ -1125,6 +1069,7 @@ static int __init smp_sanity_check(unsigned max_cpus) printk(KERN_ERR "... forcing use of dummy APIC emulation." "(tell your hw vendor)\n"); smpboot_clear_io_apic(); + disable_ioapic_setup(); return -1; } diff --git a/arch/x86/kernel/smpcommon.c b/arch/x86/kernel/smpcommon.c index 397e309839d..add36b4e37c 100644 --- a/arch/x86/kernel/smpcommon.c +++ b/arch/x86/kernel/smpcommon.c @@ -3,11 +3,16 @@ */ #include <linux/module.h> #include <asm/smp.h> +#include <asm/sections.h> -#ifdef CONFIG_X86_32 +#ifdef CONFIG_X86_64 +DEFINE_PER_CPU(unsigned long, this_cpu_off) = (unsigned long)__per_cpu_load; +#else DEFINE_PER_CPU(unsigned long, this_cpu_off); +#endif EXPORT_PER_CPU_SYMBOL(this_cpu_off); +#ifdef CONFIG_X86_32 /* * Initialize the CPU's GDT. This is either the boot CPU doing itself * (still using the master per-cpu area), or a CPU doing it for a @@ -23,8 +28,5 @@ __cpuinit void init_gdt(int cpu) write_gdt_entry(get_cpu_gdt_table(cpu), GDT_ENTRY_PERCPU, &gdt, DESCTYPE_S); - - per_cpu(this_cpu_off, cpu) = __per_cpu_offset[cpu]; - per_cpu(cpu_number, cpu) = cpu; } #endif diff --git a/arch/x86/kernel/syscall_table_32.S b/arch/x86/kernel/syscall_table_32.S index e2e86a08f31..0c4d601bc85 100644 --- a/arch/x86/kernel/syscall_table_32.S +++ b/arch/x86/kernel/syscall_table_32.S @@ -332,3 +332,4 @@ ENTRY(sys_call_table) .long sys_dup3 /* 330 */ .long sys_pipe2 .long sys_inotify_init1 + .long sys_perf_counter_open diff --git a/arch/x86/kernel/tlb_32.c b/arch/x86/kernel/tlb_32.c index ce505464224..abf0808d6fc 100644 --- a/arch/x86/kernel/tlb_32.c +++ b/arch/x86/kernel/tlb_32.c @@ -4,8 +4,8 @@ #include <asm/tlbflush.h> -DEFINE_PER_CPU(struct tlb_state, cpu_tlbstate) - ____cacheline_aligned = { &init_mm, 0, }; +DEFINE_PER_CPU_SHARED_ALIGNED(struct tlb_state, cpu_tlbstate) + = { &init_mm, 0, }; /* must come after the send_IPI functions above for inlining */ #include <mach_ipi.h> @@ -20,7 +20,7 @@ DEFINE_PER_CPU(struct tlb_state, cpu_tlbstate) * Optimizations Manfred Spraul <manfred@colorfullife.com> */ -static cpumask_t flush_cpumask; +static cpumask_var_t flush_cpumask; static struct mm_struct *flush_mm; static unsigned long flush_va; static DEFINE_SPINLOCK(tlbstate_lock); @@ -34,8 +34,8 @@ static DEFINE_SPINLOCK(tlbstate_lock); */ void leave_mm(int cpu) { - BUG_ON(x86_read_percpu(cpu_tlbstate.state) == TLBSTATE_OK); - cpu_clear(cpu, x86_read_percpu(cpu_tlbstate.active_mm)->cpu_vm_mask); + BUG_ON(percpu_read(cpu_tlbstate.state) == TLBSTATE_OK); + cpu_clear(cpu, percpu_read(cpu_tlbstate.active_mm)->cpu_vm_mask); load_cr3(swapper_pg_dir); } EXPORT_SYMBOL_GPL(leave_mm); @@ -92,7 +92,7 @@ void smp_invalidate_interrupt(struct pt_regs *regs) cpu = get_cpu(); - if (!cpu_isset(cpu, flush_cpumask)) + if (!cpumask_test_cpu(cpu, flush_cpumask)) goto out; /* * This was a BUG() but until someone can quote me the @@ -103,8 +103,8 @@ void smp_invalidate_interrupt(struct pt_regs *regs) * BUG(); */ - if (flush_mm == x86_read_percpu(cpu_tlbstate.active_mm)) { - if (x86_read_percpu(cpu_tlbstate.state) == TLBSTATE_OK) { + if (flush_mm == percpu_read(cpu_tlbstate.active_mm)) { + if (percpu_read(cpu_tlbstate.state) == TLBSTATE_OK) { if (flush_va == TLB_FLUSH_ALL) local_flush_tlb(); else @@ -114,35 +114,22 @@ void smp_invalidate_interrupt(struct pt_regs *regs) } ack_APIC_irq(); smp_mb__before_clear_bit(); - cpu_clear(cpu, flush_cpumask); + cpumask_clear_cpu(cpu, flush_cpumask); smp_mb__after_clear_bit(); out: put_cpu_no_resched(); inc_irq_stat(irq_tlb_count); } -void native_flush_tlb_others(const cpumask_t *cpumaskp, struct mm_struct *mm, - unsigned long va) +void native_flush_tlb_others(const struct cpumask *cpumask, + struct mm_struct *mm, unsigned long va) { - cpumask_t cpumask = *cpumaskp; - /* - * A couple of (to be removed) sanity checks: - * - * - current CPU must not be in mask * - mask must exist :) */ - BUG_ON(cpus_empty(cpumask)); - BUG_ON(cpu_isset(smp_processor_id(), cpumask)); + BUG_ON(cpumask_empty(cpumask)); BUG_ON(!mm); -#ifdef CONFIG_HOTPLUG_CPU - /* If a CPU which we ran on has gone down, OK. */ - cpus_and(cpumask, cpumask, cpu_online_map); - if (unlikely(cpus_empty(cpumask))) - return; -#endif - /* * i'm not happy about this global shared spinlock in the * MM hot path, but we'll see how contended it is. @@ -150,9 +137,17 @@ void native_flush_tlb_others(const cpumask_t *cpumaskp, struct mm_struct *mm, */ spin_lock(&tlbstate_lock); + cpumask_andnot(flush_cpumask, cpumask, cpumask_of(smp_processor_id())); +#ifdef CONFIG_HOTPLUG_CPU + /* If a CPU which we ran on has gone down, OK. */ + cpumask_and(flush_cpumask, flush_cpumask, cpu_online_mask); + if (unlikely(cpumask_empty(flush_cpumask))) { + spin_unlock(&tlbstate_lock); + return; + } +#endif flush_mm = mm; flush_va = va; - cpus_or(flush_cpumask, cpumask, flush_cpumask); /* * Make the above memory operations globally visible before @@ -163,9 +158,9 @@ void native_flush_tlb_others(const cpumask_t *cpumaskp, struct mm_struct *mm, * We have to send the IPI only to * CPUs affected. */ - send_IPI_mask(&cpumask, INVALIDATE_TLB_VECTOR); + send_IPI_mask(flush_cpumask, INVALIDATE_TLB_VECTOR); - while (!cpus_empty(flush_cpumask)) + while (!cpumask_empty(flush_cpumask)) /* nothing. lockup detection does not belong here */ cpu_relax(); @@ -177,25 +172,19 @@ void native_flush_tlb_others(const cpumask_t *cpumaskp, struct mm_struct *mm, void flush_tlb_current_task(void) { struct mm_struct *mm = current->mm; - cpumask_t cpu_mask; preempt_disable(); - cpu_mask = mm->cpu_vm_mask; - cpu_clear(smp_processor_id(), cpu_mask); local_flush_tlb(); - if (!cpus_empty(cpu_mask)) - flush_tlb_others(cpu_mask, mm, TLB_FLUSH_ALL); + if (cpumask_any_but(&mm->cpu_vm_mask, smp_processor_id()) < nr_cpu_ids) + flush_tlb_others(&mm->cpu_vm_mask, mm, TLB_FLUSH_ALL); preempt_enable(); } void flush_tlb_mm(struct mm_struct *mm) { - cpumask_t cpu_mask; preempt_disable(); - cpu_mask = mm->cpu_vm_mask; - cpu_clear(smp_processor_id(), cpu_mask); if (current->active_mm == mm) { if (current->mm) @@ -203,8 +192,8 @@ void flush_tlb_mm(struct mm_struct *mm) else leave_mm(smp_processor_id()); } - if (!cpus_empty(cpu_mask)) - flush_tlb_others(cpu_mask, mm, TLB_FLUSH_ALL); + if (cpumask_any_but(&mm->cpu_vm_mask, smp_processor_id()) < nr_cpu_ids) + flush_tlb_others(&mm->cpu_vm_mask, mm, TLB_FLUSH_ALL); preempt_enable(); } @@ -212,11 +201,8 @@ void flush_tlb_mm(struct mm_struct *mm) void flush_tlb_page(struct vm_area_struct *vma, unsigned long va) { struct mm_struct *mm = vma->vm_mm; - cpumask_t cpu_mask; preempt_disable(); - cpu_mask = mm->cpu_vm_mask; - cpu_clear(smp_processor_id(), cpu_mask); if (current->active_mm == mm) { if (current->mm) @@ -225,9 +211,8 @@ void flush_tlb_page(struct vm_area_struct *vma, unsigned long va) leave_mm(smp_processor_id()); } - if (!cpus_empty(cpu_mask)) - flush_tlb_others(cpu_mask, mm, va); - + if (cpumask_any_but(&mm->cpu_vm_mask, smp_processor_id()) < nr_cpu_ids) + flush_tlb_others(&mm->cpu_vm_mask, mm, va); preempt_enable(); } EXPORT_SYMBOL(flush_tlb_page); @@ -237,7 +222,7 @@ static void do_flush_tlb_all(void *info) unsigned long cpu = smp_processor_id(); __flush_tlb_all(); - if (x86_read_percpu(cpu_tlbstate.state) == TLBSTATE_LAZY) + if (percpu_read(cpu_tlbstate.state) == TLBSTATE_LAZY) leave_mm(cpu); } @@ -246,11 +231,9 @@ void flush_tlb_all(void) on_each_cpu(do_flush_tlb_all, NULL, 1); } -void reset_lazy_tlbstate(void) +static int init_flush_cpumask(void) { - int cpu = raw_smp_processor_id(); - - per_cpu(cpu_tlbstate, cpu).state = 0; - per_cpu(cpu_tlbstate, cpu).active_mm = &init_mm; + alloc_cpumask_var(&flush_cpumask, GFP_KERNEL); + return 0; } - +early_initcall(init_flush_cpumask); diff --git a/arch/x86/kernel/tlb_64.c b/arch/x86/kernel/tlb_64.c index f8be6f1d2e4..e64a32c4882 100644 --- a/arch/x86/kernel/tlb_64.c +++ b/arch/x86/kernel/tlb_64.c @@ -18,6 +18,9 @@ #include <asm/uv/uv_hub.h> #include <asm/uv/uv_bau.h> +DEFINE_PER_CPU_SHARED_ALIGNED(struct tlb_state, cpu_tlbstate) + = { &init_mm, 0, }; + #include <mach_ipi.h> /* * Smarter SMP flushing macros. @@ -43,10 +46,10 @@ union smp_flush_state { struct { - cpumask_t flush_cpumask; struct mm_struct *flush_mm; unsigned long flush_va; spinlock_t tlbstate_lock; + DECLARE_BITMAP(flush_cpumask, NR_CPUS); }; char pad[SMP_CACHE_BYTES]; } ____cacheline_aligned; @@ -62,9 +65,9 @@ static DEFINE_PER_CPU(union smp_flush_state, flush_state); */ void leave_mm(int cpu) { - if (read_pda(mmu_state) == TLBSTATE_OK) + if (percpu_read(cpu_tlbstate.state) == TLBSTATE_OK) BUG(); - cpu_clear(cpu, read_pda(active_mm)->cpu_vm_mask); + cpu_clear(cpu, percpu_read(cpu_tlbstate.active_mm)->cpu_vm_mask); load_cr3(swapper_pg_dir); } EXPORT_SYMBOL_GPL(leave_mm); @@ -131,7 +134,7 @@ asmlinkage void smp_invalidate_interrupt(struct pt_regs *regs) sender = ~regs->orig_ax - INVALIDATE_TLB_VECTOR_START; f = &per_cpu(flush_state, sender); - if (!cpu_isset(cpu, f->flush_cpumask)) + if (!cpumask_test_cpu(cpu, to_cpumask(f->flush_cpumask))) goto out; /* * This was a BUG() but until someone can quote me the @@ -142,8 +145,8 @@ asmlinkage void smp_invalidate_interrupt(struct pt_regs *regs) * BUG(); */ - if (f->flush_mm == read_pda(active_mm)) { - if (read_pda(mmu_state) == TLBSTATE_OK) { + if (f->flush_mm == percpu_read(cpu_tlbstate.active_mm)) { + if (percpu_read(cpu_tlbstate.state) == TLBSTATE_OK) { if (f->flush_va == TLB_FLUSH_ALL) local_flush_tlb(); else @@ -153,19 +156,15 @@ asmlinkage void smp_invalidate_interrupt(struct pt_regs *regs) } out: ack_APIC_irq(); - cpu_clear(cpu, f->flush_cpumask); + cpumask_clear_cpu(cpu, to_cpumask(f->flush_cpumask)); inc_irq_stat(irq_tlb_count); } -void native_flush_tlb_others(const cpumask_t *cpumaskp, struct mm_struct *mm, - unsigned long va) +static void flush_tlb_others_ipi(const struct cpumask *cpumask, + struct mm_struct *mm, unsigned long va) { int sender; union smp_flush_state *f; - cpumask_t cpumask = *cpumaskp; - - if (is_uv_system() && uv_flush_tlb_others(&cpumask, mm, va)) - return; /* Caller has disabled preemption */ sender = smp_processor_id() % NUM_INVALIDATE_TLB_VECTORS; @@ -180,7 +179,8 @@ void native_flush_tlb_others(const cpumask_t *cpumaskp, struct mm_struct *mm, f->flush_mm = mm; f->flush_va = va; - cpus_or(f->flush_cpumask, cpumask, f->flush_cpumask); + cpumask_andnot(to_cpumask(f->flush_cpumask), + cpumask, cpumask_of(smp_processor_id())); /* * Make the above memory operations globally visible before @@ -191,9 +191,10 @@ void native_flush_tlb_others(const cpumask_t *cpumaskp, struct mm_struct *mm, * We have to send the IPI only to * CPUs affected. */ - send_IPI_mask(&cpumask, INVALIDATE_TLB_VECTOR_START + sender); + send_IPI_mask(to_cpumask(f->flush_cpumask), + INVALIDATE_TLB_VECTOR_START + sender); - while (!cpus_empty(f->flush_cpumask)) + while (!cpumask_empty(to_cpumask(f->flush_cpumask))) cpu_relax(); f->flush_mm = NULL; @@ -201,6 +202,25 @@ void native_flush_tlb_others(const cpumask_t *cpumaskp, struct mm_struct *mm, spin_unlock(&f->tlbstate_lock); } +void native_flush_tlb_others(const struct cpumask *cpumask, + struct mm_struct *mm, unsigned long va) +{ + if (is_uv_system()) { + /* FIXME: could be an percpu_alloc'd thing */ + static DEFINE_PER_CPU(cpumask_t, flush_tlb_mask); + struct cpumask *after_uv_flush = &get_cpu_var(flush_tlb_mask); + + cpumask_andnot(after_uv_flush, cpumask, + cpumask_of(smp_processor_id())); + if (!uv_flush_tlb_others(after_uv_flush, mm, va)) + flush_tlb_others_ipi(after_uv_flush, mm, va); + + put_cpu_var(flush_tlb_uv_cpumask); + return; + } + flush_tlb_others_ipi(cpumask, mm, va); +} + static int __cpuinit init_smp_flush(void) { int i; @@ -215,25 +235,18 @@ core_initcall(init_smp_flush); void flush_tlb_current_task(void) { struct mm_struct *mm = current->mm; - cpumask_t cpu_mask; preempt_disable(); - cpu_mask = mm->cpu_vm_mask; - cpu_clear(smp_processor_id(), cpu_mask); local_flush_tlb(); - if (!cpus_empty(cpu_mask)) - flush_tlb_others(cpu_mask, mm, TLB_FLUSH_ALL); + if (cpumask_any_but(&mm->cpu_vm_mask, smp_processor_id()) < nr_cpu_ids) + flush_tlb_others(&mm->cpu_vm_mask, mm, TLB_FLUSH_ALL); preempt_enable(); } void flush_tlb_mm(struct mm_struct *mm) { - cpumask_t cpu_mask; - preempt_disable(); - cpu_mask = mm->cpu_vm_mask; - cpu_clear(smp_processor_id(), cpu_mask); if (current->active_mm == mm) { if (current->mm) @@ -241,8 +254,8 @@ void flush_tlb_mm(struct mm_struct *mm) else leave_mm(smp_processor_id()); } - if (!cpus_empty(cpu_mask)) - flush_tlb_others(cpu_mask, mm, TLB_FLUSH_ALL); + if (cpumask_any_but(&mm->cpu_vm_mask, smp_processor_id()) < nr_cpu_ids) + flush_tlb_others(&mm->cpu_vm_mask, mm, TLB_FLUSH_ALL); preempt_enable(); } @@ -250,11 +263,8 @@ void flush_tlb_mm(struct mm_struct *mm) void flush_tlb_page(struct vm_area_struct *vma, unsigned long va) { struct mm_struct *mm = vma->vm_mm; - cpumask_t cpu_mask; preempt_disable(); - cpu_mask = mm->cpu_vm_mask; - cpu_clear(smp_processor_id(), cpu_mask); if (current->active_mm == mm) { if (current->mm) @@ -263,8 +273,8 @@ void flush_tlb_page(struct vm_area_struct *vma, unsigned long va) leave_mm(smp_processor_id()); } - if (!cpus_empty(cpu_mask)) - flush_tlb_others(cpu_mask, mm, va); + if (cpumask_any_but(&mm->cpu_vm_mask, smp_processor_id()) < nr_cpu_ids) + flush_tlb_others(&mm->cpu_vm_mask, mm, va); preempt_enable(); } @@ -274,7 +284,7 @@ static void do_flush_tlb_all(void *info) unsigned long cpu = smp_processor_id(); __flush_tlb_all(); - if (read_pda(mmu_state) == TLBSTATE_LAZY) + if (percpu_read(cpu_tlbstate.state) == TLBSTATE_LAZY) leave_mm(cpu); } diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c index f885023167e..690dcf1a27d 100644 --- a/arch/x86/kernel/tlb_uv.c +++ b/arch/x86/kernel/tlb_uv.c @@ -212,11 +212,11 @@ static int uv_wait_completion(struct bau_desc *bau_desc, * The cpumaskp mask contains the cpus the broadcast was sent to. * * Returns 1 if all remote flushing was done. The mask is zeroed. - * Returns 0 if some remote flushing remains to be done. The mask is left - * unchanged. + * Returns 0 if some remote flushing remains to be done. The mask will have + * some bits still set. */ int uv_flush_send_and_wait(int cpu, int this_blade, struct bau_desc *bau_desc, - cpumask_t *cpumaskp) + struct cpumask *cpumaskp) { int completion_status = 0; int right_shift; @@ -263,13 +263,13 @@ int uv_flush_send_and_wait(int cpu, int this_blade, struct bau_desc *bau_desc, * Success, so clear the remote cpu's from the mask so we don't * use the IPI method of shootdown on them. */ - for_each_cpu_mask(bit, *cpumaskp) { + for_each_cpu(bit, cpumaskp) { blade = uv_cpu_to_blade_id(bit); if (blade == this_blade) continue; - cpu_clear(bit, *cpumaskp); + cpumask_clear_cpu(bit, cpumaskp); } - if (!cpus_empty(*cpumaskp)) + if (!cpumask_empty(cpumaskp)) return 0; return 1; } @@ -296,7 +296,7 @@ int uv_flush_send_and_wait(int cpu, int this_blade, struct bau_desc *bau_desc, * Returns 1 if all remote flushing was done. * Returns 0 if some remote flushing remains to be done. */ -int uv_flush_tlb_others(cpumask_t *cpumaskp, struct mm_struct *mm, +int uv_flush_tlb_others(struct cpumask *cpumaskp, struct mm_struct *mm, unsigned long va) { int i; @@ -315,7 +315,7 @@ int uv_flush_tlb_others(cpumask_t *cpumaskp, struct mm_struct *mm, bau_nodes_clear(&bau_desc->distribution, UV_DISTRIBUTION_SIZE); i = 0; - for_each_cpu_mask(bit, *cpumaskp) { + for_each_cpu(bit, cpumaskp) { blade = uv_cpu_to_blade_id(bit); BUG_ON(blade > (UV_DISTRIBUTION_SIZE - 1)); if (blade == this_blade) { diff --git a/arch/x86/kernel/vmlinux_32.lds.S b/arch/x86/kernel/vmlinux_32.lds.S index 82c67559dde..3eba7f7bac0 100644 --- a/arch/x86/kernel/vmlinux_32.lds.S +++ b/arch/x86/kernel/vmlinux_32.lds.S @@ -178,14 +178,7 @@ SECTIONS __initramfs_end = .; } #endif - . = ALIGN(PAGE_SIZE); - .data.percpu : AT(ADDR(.data.percpu) - LOAD_OFFSET) { - __per_cpu_start = .; - *(.data.percpu.page_aligned) - *(.data.percpu) - *(.data.percpu.shared_aligned) - __per_cpu_end = .; - } + PERCPU(PAGE_SIZE) . = ALIGN(PAGE_SIZE); /* freed after init ends here */ diff --git a/arch/x86/kernel/vmlinux_64.lds.S b/arch/x86/kernel/vmlinux_64.lds.S index 1a614c0e6be..a09abb8fb97 100644 --- a/arch/x86/kernel/vmlinux_64.lds.S +++ b/arch/x86/kernel/vmlinux_64.lds.S @@ -5,6 +5,7 @@ #define LOAD_OFFSET __START_KERNEL_map #include <asm-generic/vmlinux.lds.h> +#include <asm/asm-offsets.h> #include <asm/page.h> #undef i386 /* in case the preprocessor is a 32bit one */ @@ -13,12 +14,14 @@ OUTPUT_FORMAT("elf64-x86-64", "elf64-x86-64", "elf64-x86-64") OUTPUT_ARCH(i386:x86-64) ENTRY(phys_startup_64) jiffies_64 = jiffies; -_proxy_pda = 1; PHDRS { text PT_LOAD FLAGS(5); /* R_E */ data PT_LOAD FLAGS(7); /* RWE */ user PT_LOAD FLAGS(7); /* RWE */ data.init PT_LOAD FLAGS(7); /* RWE */ +#ifdef CONFIG_SMP + percpu PT_LOAD FLAGS(7); /* RWE */ +#endif note PT_NOTE FLAGS(0); /* ___ */ } SECTIONS @@ -208,14 +211,29 @@ SECTIONS __initramfs_end = .; #endif +#ifdef CONFIG_SMP + /* + * percpu offsets are zero-based on SMP. PERCPU_VADDR() changes the + * output PHDR, so the next output section - __data_nosave - should + * switch it back to data.init. Also, pda should be at the head of + * percpu area. Preallocate it and define the percpu offset symbol + * so that it can be accessed as a percpu variable. + */ + . = ALIGN(PAGE_SIZE); + PERCPU_VADDR_PREALLOC(0, :percpu, pda_size) + per_cpu____pda = __per_cpu_start; +#else PERCPU(PAGE_SIZE) +#endif . = ALIGN(PAGE_SIZE); __init_end = .; . = ALIGN(PAGE_SIZE); __nosave_begin = .; - .data_nosave : AT(ADDR(.data_nosave) - LOAD_OFFSET) { *(.data.nosave) } + .data_nosave : AT(ADDR(.data_nosave) - LOAD_OFFSET) { + *(.data.nosave) + } :data.init /* switch back to data.init, see PERCPU_VADDR() above */ . = ALIGN(PAGE_SIZE); __nosave_end = .; diff --git a/arch/x86/kernel/x8664_ksyms_64.c b/arch/x86/kernel/x8664_ksyms_64.c index 695e426aa35..3909e3ba5ce 100644 --- a/arch/x86/kernel/x8664_ksyms_64.c +++ b/arch/x86/kernel/x8664_ksyms_64.c @@ -58,5 +58,3 @@ EXPORT_SYMBOL(__memcpy); EXPORT_SYMBOL(empty_zero_page); EXPORT_SYMBOL(init_level4_pgt); EXPORT_SYMBOL(load_gs_index); - -EXPORT_SYMBOL(_proxy_pda); diff --git a/arch/x86/mach-voyager/setup.c b/arch/x86/mach-voyager/setup.c index a580b9562e7..0ade62555ff 100644 --- a/arch/x86/mach-voyager/setup.c +++ b/arch/x86/mach-voyager/setup.c @@ -9,6 +9,7 @@ #include <asm/e820.h> #include <asm/io.h> #include <asm/setup.h> +#include <asm/cpu.h> void __init pre_intr_init_hook(void) { diff --git a/arch/x86/mach-voyager/voyager_smp.c b/arch/x86/mach-voyager/voyager_smp.c index 9840b7ec749..96f15b09a4c 100644 --- a/arch/x86/mach-voyager/voyager_smp.c +++ b/arch/x86/mach-voyager/voyager_smp.c @@ -402,7 +402,7 @@ void __init find_smp_config(void) VOYAGER_SUS_IN_CONTROL_PORT); current_thread_info()->cpu = boot_cpu_id; - x86_write_percpu(cpu_number, boot_cpu_id); + percpu_write(cpu_number, boot_cpu_id); } /* @@ -531,6 +531,7 @@ static void __init do_boot_cpu(__u8 cpu) stack_start.sp = (void *)idle->thread.sp; init_gdt(cpu); + per_cpu(this_cpu_off, cpu) = __per_cpu_offset[cpu]; per_cpu(current_task, cpu) = idle; early_gdt_descr.address = (unsigned long)get_cpu_gdt_table(cpu); irq_ctx_init(cpu); @@ -1748,6 +1749,7 @@ static void __init voyager_smp_prepare_cpus(unsigned int max_cpus) static void __cpuinit voyager_smp_prepare_boot_cpu(void) { init_gdt(smp_processor_id()); + per_cpu(this_cpu_off, cpu) = __per_cpu_offset[cpu]; switch_to_new_gdt(); cpu_set(smp_processor_id(), cpu_online_map); @@ -1780,7 +1782,7 @@ static void __init voyager_smp_cpus_done(unsigned int max_cpus) void __init smp_setup_processor_id(void) { current_thread_info()->cpu = hard_smp_processor_id(); - x86_write_percpu(cpu_number, hard_smp_processor_id()); + percpu_write(cpu_number, hard_smp_processor_id()); } static void voyager_send_call_func(cpumask_t callmask) diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c index 88f1b10de3b..4a6989e47a5 100644 --- a/arch/x86/mm/init_32.c +++ b/arch/x86/mm/init_32.c @@ -49,7 +49,6 @@ #include <asm/paravirt.h> #include <asm/setup.h> #include <asm/cacheflush.h> -#include <asm/smp.h> unsigned int __VMALLOC_RESERVE = 128 << 20; diff --git a/arch/x86/mm/pat.c b/arch/x86/mm/pat.c index 8b08fb95527..c9488513fd7 100644 --- a/arch/x86/mm/pat.c +++ b/arch/x86/mm/pat.c @@ -333,11 +333,20 @@ int reserve_memtype(u64 start, u64 end, unsigned long req_type, req_type & _PAGE_CACHE_MASK); } - is_range_ram = pagerange_is_ram(start, end); - if (is_range_ram == 1) - return reserve_ram_pages_type(start, end, req_type, new_type); - else if (is_range_ram < 0) - return -EINVAL; + /* + * For legacy reasons, some parts of the physical address range in the + * legacy 1MB region is treated as non-RAM (even when listed as RAM in + * the e820 tables). So we will track the memory attributes of this + * legacy 1MB region using the linear memtype_list always. + */ + if (end >= ISA_END_ADDRESS) { + is_range_ram = pagerange_is_ram(start, end); + if (is_range_ram == 1) + return reserve_ram_pages_type(start, end, req_type, + new_type); + else if (is_range_ram < 0) + return -EINVAL; + } new = kmalloc(sizeof(struct memtype), GFP_KERNEL); if (!new) @@ -437,11 +446,19 @@ int free_memtype(u64 start, u64 end) if (is_ISA_range(start, end - 1)) return 0; - is_range_ram = pagerange_is_ram(start, end); - if (is_range_ram == 1) - return free_ram_pages_type(start, end); - else if (is_range_ram < 0) - return -EINVAL; + /* + * For legacy reasons, some parts of the physical address range in the + * legacy 1MB region is treated as non-RAM (even when listed as RAM in + * the e820 tables). So we will track the memory attributes of this + * legacy 1MB region using the linear memtype_list always. + */ + if (end >= ISA_END_ADDRESS) { + is_range_ram = pagerange_is_ram(start, end); + if (is_range_ram == 1) + return free_ram_pages_type(start, end); + else if (is_range_ram < 0) + return -EINVAL; + } spin_lock(&memtype_lock); list_for_each_entry(entry, &memtype_list, nd) { diff --git a/arch/x86/oprofile/op_model_ppro.c b/arch/x86/oprofile/op_model_ppro.c index e9f80c744cf..07c914555a5 100644 --- a/arch/x86/oprofile/op_model_ppro.c +++ b/arch/x86/oprofile/op_model_ppro.c @@ -18,7 +18,7 @@ #include <asm/msr.h> #include <asm/apic.h> #include <asm/nmi.h> -#include <asm/intel_arch_perfmon.h> +#include <asm/perf_counter.h> #include "op_x86_model.h" #include "op_counter.h" diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c index bea215230b2..75b94139e1f 100644 --- a/arch/x86/xen/enlighten.c +++ b/arch/x86/xen/enlighten.c @@ -634,35 +634,27 @@ static void xen_flush_tlb_single(unsigned long addr) preempt_enable(); } -static void xen_flush_tlb_others(const cpumask_t *cpus, struct mm_struct *mm, - unsigned long va) +static void xen_flush_tlb_others(const struct cpumask *cpus, + struct mm_struct *mm, unsigned long va) { struct { struct mmuext_op op; - cpumask_t mask; + DECLARE_BITMAP(mask, NR_CPUS); } *args; - cpumask_t cpumask = *cpus; struct multicall_space mcs; - /* - * A couple of (to be removed) sanity checks: - * - * - current CPU must not be in mask - * - mask must exist :) - */ - BUG_ON(cpus_empty(cpumask)); - BUG_ON(cpu_isset(smp_processor_id(), cpumask)); + BUG_ON(cpumask_empty(cpus)); BUG_ON(!mm); - /* If a CPU which we ran on has gone down, OK. */ - cpus_and(cpumask, cpumask, cpu_online_map); - if (cpus_empty(cpumask)) - return; - mcs = xen_mc_entry(sizeof(*args)); args = mcs.args; - args->mask = cpumask; - args->op.arg2.vcpumask = &args->mask; + args->op.arg2.vcpumask = to_cpumask(args->mask); + + /* Remove us, and any offline CPUS. */ + cpumask_and(to_cpumask(args->mask), cpus, cpu_online_mask); + cpumask_clear_cpu(smp_processor_id(), to_cpumask(args->mask)); + if (unlikely(cpumask_empty(to_cpumask(args->mask)))) + goto issue; if (va == TLB_FLUSH_ALL) { args->op.cmd = MMUEXT_TLB_FLUSH_MULTI; @@ -673,6 +665,7 @@ static void xen_flush_tlb_others(const cpumask_t *cpus, struct mm_struct *mm, MULTI_mmuext_op(mcs.mc, &args->op, 1, NULL, DOMID_SELF); +issue: xen_mc_issue(PARAVIRT_LAZY_MMU); } @@ -702,17 +695,17 @@ static void xen_write_cr0(unsigned long cr0) static void xen_write_cr2(unsigned long cr2) { - x86_read_percpu(xen_vcpu)->arch.cr2 = cr2; + percpu_read(xen_vcpu)->arch.cr2 = cr2; } static unsigned long xen_read_cr2(void) { - return x86_read_percpu(xen_vcpu)->arch.cr2; + return percpu_read(xen_vcpu)->arch.cr2; } static unsigned long xen_read_cr2_direct(void) { - return x86_read_percpu(xen_vcpu_info.arch.cr2); + return percpu_read(xen_vcpu_info.arch.cr2); } static void xen_write_cr4(unsigned long cr4) @@ -725,12 +718,12 @@ static void xen_write_cr4(unsigned long cr4) static unsigned long xen_read_cr3(void) { - return x86_read_percpu(xen_cr3); + return percpu_read(xen_cr3); } static void set_current_cr3(void *v) { - x86_write_percpu(xen_current_cr3, (unsigned long)v); + percpu_write(xen_current_cr3, (unsigned long)v); } static void __xen_write_cr3(bool kernel, unsigned long cr3) @@ -755,7 +748,7 @@ static void __xen_write_cr3(bool kernel, unsigned long cr3) MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF); if (kernel) { - x86_write_percpu(xen_cr3, cr3); + percpu_write(xen_cr3, cr3); /* Update xen_current_cr3 once the batch has actually been submitted. */ @@ -771,7 +764,7 @@ static void xen_write_cr3(unsigned long cr3) /* Update while interrupts are disabled, so its atomic with respect to ipis */ - x86_write_percpu(xen_cr3, cr3); + percpu_write(xen_cr3, cr3); __xen_write_cr3(true, cr3); @@ -1652,7 +1645,7 @@ asmlinkage void __init xen_start_kernel(void) #ifdef CONFIG_X86_64 /* Disable until direct per-cpu data access. */ have_vcpu_info_placement = 0; - x86_64_init_pda(); + pda_init(0); #endif xen_smp_init(); diff --git a/arch/x86/xen/irq.c b/arch/x86/xen/irq.c index bb042608c60..2e8271431e1 100644 --- a/arch/x86/xen/irq.c +++ b/arch/x86/xen/irq.c @@ -39,7 +39,7 @@ static unsigned long xen_save_fl(void) struct vcpu_info *vcpu; unsigned long flags; - vcpu = x86_read_percpu(xen_vcpu); + vcpu = percpu_read(xen_vcpu); /* flag has opposite sense of mask */ flags = !vcpu->evtchn_upcall_mask; @@ -62,7 +62,7 @@ static void xen_restore_fl(unsigned long flags) make sure we're don't switch CPUs between getting the vcpu pointer and updating the mask. */ preempt_disable(); - vcpu = x86_read_percpu(xen_vcpu); + vcpu = percpu_read(xen_vcpu); vcpu->evtchn_upcall_mask = flags; preempt_enable_no_resched(); @@ -83,7 +83,7 @@ static void xen_irq_disable(void) make sure we're don't switch CPUs between getting the vcpu pointer and updating the mask. */ preempt_disable(); - x86_read_percpu(xen_vcpu)->evtchn_upcall_mask = 1; + percpu_read(xen_vcpu)->evtchn_upcall_mask = 1; preempt_enable_no_resched(); } @@ -96,7 +96,7 @@ static void xen_irq_enable(void) the caller is confused and is trying to re-enable interrupts on an indeterminate processor. */ - vcpu = x86_read_percpu(xen_vcpu); + vcpu = percpu_read(xen_vcpu); vcpu->evtchn_upcall_mask = 0; /* Doesn't matter if we get preempted here, because any diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c index 503c240e26c..98cb9869eb2 100644 --- a/arch/x86/xen/mmu.c +++ b/arch/x86/xen/mmu.c @@ -1063,18 +1063,14 @@ static void drop_other_mm_ref(void *info) struct mm_struct *mm = info; struct mm_struct *active_mm; -#ifdef CONFIG_X86_64 - active_mm = read_pda(active_mm); -#else - active_mm = __get_cpu_var(cpu_tlbstate).active_mm; -#endif + active_mm = percpu_read(cpu_tlbstate.active_mm); if (active_mm == mm) leave_mm(smp_processor_id()); /* If this cpu still has a stale cr3 reference, then make sure it has been flushed. */ - if (x86_read_percpu(xen_current_cr3) == __pa(mm->pgd)) { + if (percpu_read(xen_current_cr3) == __pa(mm->pgd)) { load_cr3(swapper_pg_dir); arch_flush_lazy_cpu_mode(); } diff --git a/arch/x86/xen/multicalls.h b/arch/x86/xen/multicalls.h index 85893824161..e786fa7f261 100644 --- a/arch/x86/xen/multicalls.h +++ b/arch/x86/xen/multicalls.h @@ -39,7 +39,7 @@ static inline void xen_mc_issue(unsigned mode) xen_mc_flush(); /* restore flags saved in xen_mc_batch */ - local_irq_restore(x86_read_percpu(xen_mc_irq_flags)); + local_irq_restore(percpu_read(xen_mc_irq_flags)); } /* Set up a callback to be called when the current batch is flushed */ diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c index c44e2069c7c..72c2eb9b64c 100644 --- a/arch/x86/xen/smp.c +++ b/arch/x86/xen/smp.c @@ -50,11 +50,7 @@ static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id); */ static irqreturn_t xen_reschedule_interrupt(int irq, void *dev_id) { -#ifdef CONFIG_X86_32 - __get_cpu_var(irq_stat).irq_resched_count++; -#else - add_pda(irq_resched_count, 1); -#endif + inc_irq_stat(irq_resched_count); return IRQ_HANDLED; } @@ -78,7 +74,7 @@ static __cpuinit void cpu_bringup(void) xen_setup_cpu_clockevents(); cpu_set(cpu, cpu_online_map); - x86_write_percpu(cpu_state, CPU_ONLINE); + percpu_write(cpu_state, CPU_ONLINE); wmb(); /* We can take interrupts now: we're officially "up". */ @@ -283,22 +279,11 @@ static int __cpuinit xen_cpu_up(unsigned int cpu) struct task_struct *idle = idle_task(cpu); int rc; -#ifdef CONFIG_X86_64 - /* Allocate node local memory for AP pdas */ - WARN_ON(cpu == 0); - if (cpu > 0) { - rc = get_local_pda(cpu); - if (rc) - return rc; - } -#endif - + per_cpu(current_task, cpu) = idle; #ifdef CONFIG_X86_32 init_gdt(cpu); - per_cpu(current_task, cpu) = idle; irq_ctx_init(cpu); #else - cpu_pda(cpu)->pcurrent = idle; clear_tsk_thread_flag(idle, TIF_FORK); #endif xen_setup_timer(cpu); @@ -445,11 +430,7 @@ static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id) { irq_enter(); generic_smp_call_function_interrupt(); -#ifdef CONFIG_X86_32 - __get_cpu_var(irq_stat).irq_call_count++; -#else - add_pda(irq_call_count, 1); -#endif + inc_irq_stat(irq_call_count); irq_exit(); return IRQ_HANDLED; @@ -459,11 +440,7 @@ static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id) { irq_enter(); generic_smp_call_function_single_interrupt(); -#ifdef CONFIG_X86_32 - __get_cpu_var(irq_stat).irq_call_count++; -#else - add_pda(irq_call_count, 1); -#endif + inc_irq_stat(irq_call_count); irq_exit(); return IRQ_HANDLED; diff --git a/arch/x86/xen/xen-asm_64.S b/arch/x86/xen/xen-asm_64.S index 05794c566e8..d6fc51f4ce8 100644 --- a/arch/x86/xen/xen-asm_64.S +++ b/arch/x86/xen/xen-asm_64.S @@ -17,6 +17,7 @@ #include <asm/processor-flags.h> #include <asm/errno.h> #include <asm/segment.h> +#include <asm/percpu.h> #include <xen/interface/xen.h> @@ -28,12 +29,10 @@ #if 1 /* - x86-64 does not yet support direct access to percpu variables - via a segment override, so we just need to make sure this code - never gets used + FIXME: x86_64 now can support direct access to percpu variables + via a segment override. Update xen accordingly. */ #define BUG ud2a -#define PER_CPU_VAR(var, off) 0xdeadbeef #endif /* @@ -45,14 +44,14 @@ ENTRY(xen_irq_enable_direct) BUG /* Unmask events */ - movb $0, PER_CPU_VAR(xen_vcpu_info, XEN_vcpu_info_mask) + movb $0, PER_CPU_VAR(xen_vcpu_info) + XEN_vcpu_info_mask /* Preempt here doesn't matter because that will deal with any pending interrupts. The pending check may end up being run on the wrong CPU, but that doesn't hurt. */ /* Test for pending */ - testb $0xff, PER_CPU_VAR(xen_vcpu_info, XEN_vcpu_info_pending) + testb $0xff, PER_CPU_VAR(xen_vcpu_info) + XEN_vcpu_info_pending jz 1f 2: call check_events @@ -69,7 +68,7 @@ ENDPATCH(xen_irq_enable_direct) ENTRY(xen_irq_disable_direct) BUG - movb $1, PER_CPU_VAR(xen_vcpu_info, XEN_vcpu_info_mask) + movb $1, PER_CPU_VAR(xen_vcpu_info) + XEN_vcpu_info_mask ENDPATCH(xen_irq_disable_direct) ret ENDPROC(xen_irq_disable_direct) @@ -87,7 +86,7 @@ ENDPATCH(xen_irq_disable_direct) ENTRY(xen_save_fl_direct) BUG - testb $0xff, PER_CPU_VAR(xen_vcpu_info, XEN_vcpu_info_mask) + testb $0xff, PER_CPU_VAR(xen_vcpu_info) + XEN_vcpu_info_mask setz %ah addb %ah,%ah ENDPATCH(xen_save_fl_direct) @@ -107,13 +106,13 @@ ENTRY(xen_restore_fl_direct) BUG testb $X86_EFLAGS_IF>>8, %ah - setz PER_CPU_VAR(xen_vcpu_info, XEN_vcpu_info_mask) + setz PER_CPU_VAR(xen_vcpu_info) + XEN_vcpu_info_mask /* Preempt here doesn't matter because that will deal with any pending interrupts. The pending check may end up being run on the wrong CPU, but that doesn't hurt. */ /* check for unmasked and pending */ - cmpw $0x0001, PER_CPU_VAR(xen_vcpu_info, XEN_vcpu_info_pending) + cmpw $0x0001, PER_CPU_VAR(xen_vcpu_info) + XEN_vcpu_info_pending jz 1f 2: call check_events 1: @@ -195,11 +194,11 @@ RELOC(xen_sysexit, 1b+1) ENTRY(xen_sysret64) /* We're already on the usermode stack at this point, but still with the kernel gs, so we can easily switch back */ - movq %rsp, %gs:pda_oldrsp - movq %gs:pda_kernelstack,%rsp + movq %rsp, PER_CPU_VAR(old_rsp) + movq PER_CPU_VAR(kernel_stack),%rsp pushq $__USER_DS - pushq %gs:pda_oldrsp + pushq PER_CPU_VAR(old_rsp) pushq %r11 pushq $__USER_CS pushq %rcx @@ -212,11 +211,11 @@ RELOC(xen_sysret64, 1b+1) ENTRY(xen_sysret32) /* We're already on the usermode stack at this point, but still with the kernel gs, so we can easily switch back */ - movq %rsp, %gs:pda_oldrsp - movq %gs:pda_kernelstack, %rsp + movq %rsp, PER_CPU_VAR(old_rsp) + movq PER_CPU_VAR(kernel_stack), %rsp pushq $__USER32_DS - pushq %gs:pda_oldrsp + pushq PER_CPU_VAR(old_rsp) pushq %r11 pushq $__USER32_CS pushq %rcx diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c index 66a9d814556..7acb23f830c 100644 --- a/drivers/acpi/processor_idle.c +++ b/drivers/acpi/processor_idle.c @@ -271,8 +271,11 @@ static atomic_t c3_cpu_count; /* Common C-state entry for C2, C3, .. */ static void acpi_cstate_enter(struct acpi_processor_cx *cstate) { + u64 perf_flags; + /* Don't trace irqs off for idle */ stop_critical_timings(); + perf_flags = hw_perf_save_disable(); if (cstate->entry_method == ACPI_CSTATE_FFH) { /* Call into architectural FFH based C-state */ acpi_processor_ffh_cstate_enter(cstate); @@ -285,6 +288,7 @@ static void acpi_cstate_enter(struct acpi_processor_cx *cstate) gets asserted in time to freeze execution properly. */ unused = inl(acpi_gbl_FADT.xpm_timer_block.address); } + hw_perf_restore(perf_flags); start_critical_timings(); } #endif /* !CONFIG_CPU_IDLE */ @@ -1426,8 +1430,11 @@ static inline void acpi_idle_update_bm_rld(struct acpi_processor *pr, */ static inline void acpi_idle_do_entry(struct acpi_processor_cx *cx) { + u64 pctrl; + /* Don't trace irqs off for idle */ stop_critical_timings(); + pctrl = hw_perf_save_disable(); if (cx->entry_method == ACPI_CSTATE_FFH) { /* Call into architectural FFH based C-state */ acpi_processor_ffh_cstate_enter(cx); @@ -1442,6 +1449,7 @@ static inline void acpi_idle_do_entry(struct acpi_processor_cx *cx) gets asserted in time to freeze execution properly. */ unused = inl(acpi_gbl_FADT.xpm_timer_block.address); } + hw_perf_restore(pctrl); start_critical_timings(); } diff --git a/drivers/base/cpu.c b/drivers/base/cpu.c index 719ee5c1c8d..5b257a57bc5 100644 --- a/drivers/base/cpu.c +++ b/drivers/base/cpu.c @@ -107,7 +107,7 @@ static SYSDEV_ATTR(crash_notes, 0400, show_crash_notes, NULL); /* * Print cpu online, possible, present, and system maps */ -static ssize_t print_cpus_map(char *buf, cpumask_t *map) +static ssize_t print_cpus_map(char *buf, const struct cpumask *map) { int n = cpulist_scnprintf(buf, PAGE_SIZE-2, map); diff --git a/drivers/base/topology.c b/drivers/base/topology.c index a778fb52b11..bf6b13206d0 100644 --- a/drivers/base/topology.c +++ b/drivers/base/topology.c @@ -31,7 +31,10 @@ #include <linux/hardirq.h> #include <linux/topology.h> -#define define_one_ro(_name) \ +#define define_one_ro_named(_name, _func) \ +static SYSDEV_ATTR(_name, 0444, _func, NULL) + +#define define_one_ro(_name) \ static SYSDEV_ATTR(_name, 0444, show_##_name, NULL) #define define_id_show_func(name) \ @@ -42,8 +45,8 @@ static ssize_t show_##name(struct sys_device *dev, \ return sprintf(buf, "%d\n", topology_##name(cpu)); \ } -#if defined(topology_thread_siblings) || defined(topology_core_siblings) -static ssize_t show_cpumap(int type, cpumask_t *mask, char *buf) +#if defined(topology_thread_cpumask) || defined(topology_core_cpumask) +static ssize_t show_cpumap(int type, const struct cpumask *mask, char *buf) { ptrdiff_t len = PTR_ALIGN(buf + PAGE_SIZE - 1, PAGE_SIZE) - buf; int n = 0; @@ -65,7 +68,7 @@ static ssize_t show_##name(struct sys_device *dev, \ struct sysdev_attribute *attr, char *buf) \ { \ unsigned int cpu = dev->id; \ - return show_cpumap(0, &(topology_##name(cpu)), buf); \ + return show_cpumap(0, topology_##name(cpu), buf); \ } #define define_siblings_show_list(name) \ @@ -74,7 +77,7 @@ static ssize_t show_##name##_list(struct sys_device *dev, \ char *buf) \ { \ unsigned int cpu = dev->id; \ - return show_cpumap(1, &(topology_##name(cpu)), buf); \ + return show_cpumap(1, topology_##name(cpu), buf); \ } #else @@ -82,9 +85,7 @@ static ssize_t show_##name##_list(struct sys_device *dev, \ static ssize_t show_##name(struct sys_device *dev, \ struct sysdev_attribute *attr, char *buf) \ { \ - unsigned int cpu = dev->id; \ - cpumask_t mask = topology_##name(cpu); \ - return show_cpumap(0, &mask, buf); \ + return show_cpumap(0, topology_##name(dev->id), buf); \ } #define define_siblings_show_list(name) \ @@ -92,9 +93,7 @@ static ssize_t show_##name##_list(struct sys_device *dev, \ struct sysdev_attribute *attr, \ char *buf) \ { \ - unsigned int cpu = dev->id; \ - cpumask_t mask = topology_##name(cpu); \ - return show_cpumap(1, &mask, buf); \ + return show_cpumap(1, topology_##name(dev->id), buf); \ } #endif @@ -107,13 +106,13 @@ define_one_ro(physical_package_id); define_id_show_func(core_id); define_one_ro(core_id); -define_siblings_show_func(thread_siblings); -define_one_ro(thread_siblings); -define_one_ro(thread_siblings_list); +define_siblings_show_func(thread_cpumask); +define_one_ro_named(thread_siblings, show_thread_cpumask); +define_one_ro_named(thread_siblings_list, show_thread_cpumask_list); -define_siblings_show_func(core_siblings); -define_one_ro(core_siblings); -define_one_ro(core_siblings_list); +define_siblings_show_func(core_cpumask); +define_one_ro_named(core_siblings, show_core_cpumask); +define_one_ro_named(core_siblings_list, show_core_cpumask_list); static struct attribute *default_attrs[] = { &attr_physical_package_id.attr, diff --git a/drivers/char/sysrq.c b/drivers/char/sysrq.c index 33a9351c896..fa71b84f217 100644 --- a/drivers/char/sysrq.c +++ b/drivers/char/sysrq.c @@ -25,6 +25,7 @@ #include <linux/kbd_kern.h> #include <linux/proc_fs.h> #include <linux/quotaops.h> +#include <linux/perf_counter.h> #include <linux/kernel.h> #include <linux/module.h> #include <linux/suspend.h> @@ -244,6 +245,7 @@ static void sysrq_handle_showregs(int key, struct tty_struct *tty) struct pt_regs *regs = get_irq_regs(); if (regs) show_regs(regs); + perf_counter_print_debug(); } static struct sysrq_key_op sysrq_showregs_op = { .handler = sysrq_handle_showregs, diff --git a/drivers/firmware/dcdbas.c b/drivers/firmware/dcdbas.c index 777fba48d2d..3009e0171e5 100644 --- a/drivers/firmware/dcdbas.c +++ b/drivers/firmware/dcdbas.c @@ -244,7 +244,7 @@ static ssize_t host_control_on_shutdown_store(struct device *dev, */ int dcdbas_smi_request(struct smi_cmd *smi_cmd) { - cpumask_t old_mask; + cpumask_var_t old_mask; int ret = 0; if (smi_cmd->magic != SMI_CMD_MAGIC) { @@ -254,8 +254,11 @@ int dcdbas_smi_request(struct smi_cmd *smi_cmd) } /* SMI requires CPU 0 */ - old_mask = current->cpus_allowed; - set_cpus_allowed_ptr(current, &cpumask_of_cpu(0)); + if (!alloc_cpumask_var(&old_mask, GFP_KERNEL)) + return -ENOMEM; + + cpumask_copy(old_mask, ¤t->cpus_allowed); + set_cpus_allowed_ptr(current, cpumask_of(0)); if (smp_processor_id() != 0) { dev_dbg(&dcdbas_pdev->dev, "%s: failed to get CPU 0\n", __func__); @@ -275,7 +278,8 @@ int dcdbas_smi_request(struct smi_cmd *smi_cmd) ); out: - set_cpus_allowed_ptr(current, &old_mask); + set_cpus_allowed_ptr(current, old_mask); + free_cpumask_var(old_mask); return ret; } diff --git a/drivers/misc/sgi-xp/xpc_main.c b/drivers/misc/sgi-xp/xpc_main.c index 89218f7cfaa..6576170de96 100644 --- a/drivers/misc/sgi-xp/xpc_main.c +++ b/drivers/misc/sgi-xp/xpc_main.c @@ -318,7 +318,7 @@ xpc_hb_checker(void *ignore) /* this thread was marked active by xpc_hb_init() */ - set_cpus_allowed_ptr(current, &cpumask_of_cpu(XPC_HB_CHECK_CPU)); + set_cpus_allowed_ptr(current, cpumask_of(XPC_HB_CHECK_CPU)); /* set our heartbeating to other partitions into motion */ xpc_hb_check_timeout = jiffies + (xpc_hb_check_interval * HZ); diff --git a/drivers/net/sfc/efx.c b/drivers/net/sfc/efx.c index 7673fd92eaf..101c00a7bb7 100644 --- a/drivers/net/sfc/efx.c +++ b/drivers/net/sfc/efx.c @@ -854,20 +854,27 @@ static void efx_fini_io(struct efx_nic *efx) * interrupts across them. */ static int efx_wanted_rx_queues(void) { - cpumask_t core_mask; + cpumask_var_t core_mask; int count; int cpu; - cpus_clear(core_mask); + if (!alloc_cpumask_var(&core_mask, GFP_KERNEL)) { + printk(KERN_WARNING + "efx.c: allocation failure, irq balancing hobbled\n"); + return 1; + } + + cpumask_clear(core_mask); count = 0; for_each_online_cpu(cpu) { - if (!cpu_isset(cpu, core_mask)) { + if (!cpumask_test_cpu(cpu, core_mask)) { ++count; - cpus_or(core_mask, core_mask, - topology_core_siblings(cpu)); + cpumask_or(core_mask, core_mask, + topology_core_cpumask(cpu)); } } + free_cpumask_var(core_mask); return count; } diff --git a/drivers/oprofile/buffer_sync.c b/drivers/oprofile/buffer_sync.c index 9da5a4b8113..c3ea5fa7d05 100644 --- a/drivers/oprofile/buffer_sync.c +++ b/drivers/oprofile/buffer_sync.c @@ -38,7 +38,7 @@ static LIST_HEAD(dying_tasks); static LIST_HEAD(dead_tasks); -static cpumask_t marked_cpus = CPU_MASK_NONE; +static cpumask_var_t marked_cpus; static DEFINE_SPINLOCK(task_mortuary); static void process_task_mortuary(void); @@ -456,10 +456,10 @@ static void mark_done(int cpu) { int i; - cpu_set(cpu, marked_cpus); + cpumask_set_cpu(cpu, marked_cpus); for_each_online_cpu(i) { - if (!cpu_isset(i, marked_cpus)) + if (!cpumask_test_cpu(i, marked_cpus)) return; } @@ -468,7 +468,7 @@ static void mark_done(int cpu) */ process_task_mortuary(); - cpus_clear(marked_cpus); + cpumask_clear(marked_cpus); } @@ -565,6 +565,20 @@ void sync_buffer(int cpu) mutex_unlock(&buffer_mutex); } +int __init buffer_sync_init(void) +{ + if (!alloc_cpumask_var(&marked_cpus, GFP_KERNEL)) + return -ENOMEM; + + cpumask_clear(marked_cpus); + return 0; +} + +void __exit buffer_sync_cleanup(void) +{ + free_cpumask_var(marked_cpus); +} + /* The function can be used to add a buffer worth of data directly to * the kernel buffer. The buffer is assumed to be a circular buffer. * Take the entries from index start and end at index end, wrapping diff --git a/drivers/oprofile/buffer_sync.h b/drivers/oprofile/buffer_sync.h index 3110732c183..0ebf5db6267 100644 --- a/drivers/oprofile/buffer_sync.h +++ b/drivers/oprofile/buffer_sync.h @@ -19,4 +19,8 @@ void sync_stop(void); /* sync the given CPU's buffer */ void sync_buffer(int cpu); +/* initialize/destroy the buffer system. */ +int buffer_sync_init(void); +void buffer_sync_cleanup(void); + #endif /* OPROFILE_BUFFER_SYNC_H */ diff --git a/drivers/oprofile/oprof.c b/drivers/oprofile/oprof.c index 3cffce90f82..ced39f60229 100644 --- a/drivers/oprofile/oprof.c +++ b/drivers/oprofile/oprof.c @@ -183,6 +183,10 @@ static int __init oprofile_init(void) { int err; + err = buffer_sync_init(); + if (err) + return err; + err = oprofile_arch_init(&oprofile_ops); if (err < 0 || timer) { @@ -191,8 +195,10 @@ static int __init oprofile_init(void) } err = oprofilefs_register(); - if (err) + if (err) { oprofile_arch_exit(); + buffer_sync_cleanup(); + } return err; } @@ -202,6 +208,7 @@ static void __exit oprofile_exit(void) { oprofilefs_unregister(); oprofile_arch_exit(); + buffer_sync_cleanup(); } diff --git a/drivers/pci/intr_remapping.c b/drivers/pci/intr_remapping.c index f78371b2252..5a57753ea9f 100644 --- a/drivers/pci/intr_remapping.c +++ b/drivers/pci/intr_remapping.c @@ -6,6 +6,7 @@ #include <linux/irq.h> #include <asm/io_apic.h> #include <asm/smp.h> +#include <asm/cpu.h> #include <linux/intel-iommu.h> #include "intr_remapping.h" diff --git a/drivers/xen/events.c b/drivers/xen/events.c index eb0dfdeaa94..3141e149d59 100644 --- a/drivers/xen/events.c +++ b/drivers/xen/events.c @@ -26,6 +26,7 @@ #include <linux/irq.h> #include <linux/module.h> #include <linux/string.h> +#include <linux/bootmem.h> #include <asm/ptrace.h> #include <asm/irq.h> @@ -75,7 +76,14 @@ enum { static int evtchn_to_irq[NR_EVENT_CHANNELS] = { [0 ... NR_EVENT_CHANNELS-1] = -1 }; -static unsigned long cpu_evtchn_mask[NR_CPUS][NR_EVENT_CHANNELS/BITS_PER_LONG]; +struct cpu_evtchn_s { + unsigned long bits[NR_EVENT_CHANNELS/BITS_PER_LONG]; +}; +static struct cpu_evtchn_s *cpu_evtchn_mask_p; +static inline unsigned long *cpu_evtchn_mask(int cpu) +{ + return cpu_evtchn_mask_p[cpu].bits; +} static u8 cpu_evtchn[NR_EVENT_CHANNELS]; /* Reference counts for bindings to IRQs. */ @@ -115,7 +123,7 @@ static inline unsigned long active_evtchns(unsigned int cpu, unsigned int idx) { return (sh->evtchn_pending[idx] & - cpu_evtchn_mask[cpu][idx] & + cpu_evtchn_mask(cpu)[idx] & ~sh->evtchn_mask[idx]); } @@ -125,11 +133,11 @@ static void bind_evtchn_to_cpu(unsigned int chn, unsigned int cpu) BUG_ON(irq == -1); #ifdef CONFIG_SMP - irq_to_desc(irq)->affinity = cpumask_of_cpu(cpu); + cpumask_copy(irq_to_desc(irq)->affinity, cpumask_of(cpu)); #endif - __clear_bit(chn, cpu_evtchn_mask[cpu_evtchn[chn]]); - __set_bit(chn, cpu_evtchn_mask[cpu]); + __clear_bit(chn, cpu_evtchn_mask(cpu_evtchn[chn])); + __set_bit(chn, cpu_evtchn_mask(cpu)); cpu_evtchn[chn] = cpu; } @@ -142,12 +150,12 @@ static void init_evtchn_cpu_bindings(void) /* By default all event channels notify CPU#0. */ for_each_irq_desc(i, desc) { - desc->affinity = cpumask_of_cpu(0); + cpumask_copy(desc->affinity, cpumask_of(0)); } #endif memset(cpu_evtchn, 0, sizeof(cpu_evtchn)); - memset(cpu_evtchn_mask[0], ~0, sizeof(cpu_evtchn_mask[0])); + memset(cpu_evtchn_mask(0), ~0, sizeof(cpu_evtchn_mask(0))); } static inline unsigned int cpu_from_evtchn(unsigned int evtchn) @@ -822,6 +830,10 @@ static struct irq_chip xen_dynamic_chip __read_mostly = { void __init xen_init_IRQ(void) { int i; + size_t size = nr_cpu_ids * sizeof(struct cpu_evtchn_s); + + cpu_evtchn_mask_p = alloc_bootmem(size); + BUG_ON(cpu_evtchn_mask_p == NULL); init_evtchn_cpu_bindings(); diff --git a/drivers/xen/manage.c b/drivers/xen/manage.c index 9b91617b958..e7e83b65c18 100644 --- a/drivers/xen/manage.c +++ b/drivers/xen/manage.c @@ -100,7 +100,7 @@ static void do_suspend(void) /* XXX use normal device tree? */ xenbus_suspend(); - err = stop_machine(xen_suspend, &cancelled, &cpumask_of_cpu(0)); + err = stop_machine(xen_suspend, &cancelled, cpumask_of(0)); if (err) { printk(KERN_ERR "failed to start xen_suspend: %d\n", err); goto out; diff --git a/fs/exec.c b/fs/exec.c index 0dd60a01f1b..97f9c225dd6 100644 --- a/fs/exec.c +++ b/fs/exec.c @@ -33,6 +33,7 @@ #include <linux/string.h> #include <linux/init.h> #include <linux/pagemap.h> +#include <linux/perf_counter.h> #include <linux/highmem.h> #include <linux/spinlock.h> #include <linux/key.h> @@ -1010,6 +1011,13 @@ int flush_old_exec(struct linux_binprm * bprm) current->personality &= ~bprm->per_clear; + /* + * Flush performance counters when crossing a + * security domain: + */ + if (!get_dumpable(current->mm)) + perf_counter_exit_task(current); + /* An exec changes our domain. We are no longer part of the thread group */ diff --git a/include/asm-generic/bitops/__ffs.h b/include/asm-generic/bitops/__ffs.h index 9a3274aecf8..937d7c43557 100644 --- a/include/asm-generic/bitops/__ffs.h +++ b/include/asm-generic/bitops/__ffs.h @@ -9,7 +9,7 @@ * * Undefined if no bit exists, so code should check against 0 first. */ -static inline unsigned long __ffs(unsigned long word) +static __always_inline unsigned long __ffs(unsigned long word) { int num = 0; diff --git a/include/asm-generic/bitops/__fls.h b/include/asm-generic/bitops/__fls.h index be24465403d..a60a7ccb678 100644 --- a/include/asm-generic/bitops/__fls.h +++ b/include/asm-generic/bitops/__fls.h @@ -9,7 +9,7 @@ * * Undefined if no set bit exists, so code should check against 0 first. */ -static inline unsigned long __fls(unsigned long word) +static __always_inline unsigned long __fls(unsigned long word) { int num = BITS_PER_LONG - 1; diff --git a/include/asm-generic/bitops/fls.h b/include/asm-generic/bitops/fls.h index 850859bc506..0576d1f42f4 100644 --- a/include/asm-generic/bitops/fls.h +++ b/include/asm-generic/bitops/fls.h @@ -9,7 +9,7 @@ * Note fls(0) = 0, fls(1) = 1, fls(0x80000000) = 32. */ -static inline int fls(int x) +static __always_inline int fls(int x) { int r = 32; diff --git a/include/asm-generic/bitops/fls64.h b/include/asm-generic/bitops/fls64.h index 86d403f8b25..b097cf8444e 100644 --- a/include/asm-generic/bitops/fls64.h +++ b/include/asm-generic/bitops/fls64.h @@ -15,7 +15,7 @@ * at position 64. */ #if BITS_PER_LONG == 32 -static inline int fls64(__u64 x) +static __always_inline int fls64(__u64 x) { __u32 h = x >> 32; if (h) @@ -23,7 +23,7 @@ static inline int fls64(__u64 x) return fls(x); } #elif BITS_PER_LONG == 64 -static inline int fls64(__u64 x) +static __always_inline int fls64(__u64 x) { if (x == 0) return 0; diff --git a/include/asm-generic/percpu.h b/include/asm-generic/percpu.h index b0e63c672eb..00f45ff081a 100644 --- a/include/asm-generic/percpu.h +++ b/include/asm-generic/percpu.h @@ -80,4 +80,56 @@ extern void setup_per_cpu_areas(void); #define DECLARE_PER_CPU(type, name) extern PER_CPU_ATTRIBUTES \ __typeof__(type) per_cpu_var(name) +/* + * Optional methods for optimized non-lvalue per-cpu variable access. + * + * @var can be a percpu variable or a field of it and its size should + * equal char, int or long. percpu_read() evaluates to a lvalue and + * all others to void. + * + * These operations are guaranteed to be atomic w.r.t. preemption. + * The generic versions use plain get/put_cpu_var(). Archs are + * encouraged to implement single-instruction alternatives which don't + * require preemption protection. + */ +#ifndef percpu_read +# define percpu_read(var) \ + ({ \ + typeof(per_cpu_var(var)) __tmp_var__; \ + __tmp_var__ = get_cpu_var(var); \ + put_cpu_var(var); \ + __tmp_var__; \ + }) +#endif + +#define __percpu_generic_to_op(var, val, op) \ +do { \ + get_cpu_var(var) op val; \ + put_cpu_var(var); \ +} while (0) + +#ifndef percpu_write +# define percpu_write(var, val) __percpu_generic_to_op(var, (val), =) +#endif + +#ifndef percpu_add +# define percpu_add(var, val) __percpu_generic_to_op(var, (val), +=) +#endif + +#ifndef percpu_sub +# define percpu_sub(var, val) __percpu_generic_to_op(var, (val), -=) +#endif + +#ifndef percpu_and +# define percpu_and(var, val) __percpu_generic_to_op(var, (val), &=) +#endif + +#ifndef percpu_or +# define percpu_or(var, val) __percpu_generic_to_op(var, (val), |=) +#endif + +#ifndef percpu_xor +# define percpu_xor(var, val) __percpu_generic_to_op(var, (val), ^=) +#endif + #endif /* _ASM_GENERIC_PERCPU_H_ */ diff --git a/include/asm-generic/sections.h b/include/asm-generic/sections.h index 79a7ff925bf..4ce48e87853 100644 --- a/include/asm-generic/sections.h +++ b/include/asm-generic/sections.h @@ -9,7 +9,7 @@ extern char __bss_start[], __bss_stop[]; extern char __init_begin[], __init_end[]; extern char _sinittext[], _einittext[]; extern char _end[]; -extern char __per_cpu_start[], __per_cpu_end[]; +extern char __per_cpu_load[], __per_cpu_start[], __per_cpu_end[]; extern char __kprobes_text_start[], __kprobes_text_end[]; extern char __initdata_begin[], __initdata_end[]; extern char __start_rodata[], __end_rodata[]; diff --git a/include/asm-generic/vmlinux.lds.h b/include/asm-generic/vmlinux.lds.h index c61fab1dd2f..aa6b9b1b30b 100644 --- a/include/asm-generic/vmlinux.lds.h +++ b/include/asm-generic/vmlinux.lds.h @@ -430,12 +430,75 @@ *(.initcall7.init) \ *(.initcall7s.init) -#define PERCPU(align) \ - . = ALIGN(align); \ - VMLINUX_SYMBOL(__per_cpu_start) = .; \ - .data.percpu : AT(ADDR(.data.percpu) - LOAD_OFFSET) { \ +#define PERCPU_PROLOG(vaddr) \ + VMLINUX_SYMBOL(__per_cpu_load) = .; \ + .data.percpu vaddr : AT(VMLINUX_SYMBOL(__per_cpu_load) \ + - LOAD_OFFSET) { \ + VMLINUX_SYMBOL(__per_cpu_start) = .; + +#define PERCPU_EPILOG(phdr) \ + VMLINUX_SYMBOL(__per_cpu_end) = .; \ + } phdr \ + . = VMLINUX_SYMBOL(__per_cpu_load) + SIZEOF(.data.percpu); + +/** + * PERCPU_VADDR_PREALLOC - define output section for percpu area with prealloc + * @vaddr: explicit base address (optional) + * @phdr: destination PHDR (optional) + * @prealloc: the size of prealloc area + * + * Macro which expands to output section for percpu area. If @vaddr + * is not blank, it specifies explicit base address and all percpu + * symbols will be offset from the given address. If blank, @vaddr + * always equals @laddr + LOAD_OFFSET. + * + * @phdr defines the output PHDR to use if not blank. Be warned that + * output PHDR is sticky. If @phdr is specified, the next output + * section in the linker script will go there too. @phdr should have + * a leading colon. + * + * If @prealloc is non-zero, the specified number of bytes will be + * reserved at the start of percpu area. As the prealloc area is + * likely to break alignment, this macro puts areas in increasing + * alignment order. + * + * This macro defines three symbols, __per_cpu_load, __per_cpu_start + * and __per_cpu_end. The first one is the vaddr of loaded percpu + * init data. __per_cpu_start equals @vaddr and __per_cpu_end is the + * end offset. + */ +#define PERCPU_VADDR_PREALLOC(vaddr, segment, prealloc) \ + PERCPU_PROLOG(vaddr) \ + . += prealloc; \ + *(.data.percpu) \ + *(.data.percpu.shared_aligned) \ + *(.data.percpu.page_aligned) \ + PERCPU_EPILOG(segment) + +/** + * PERCPU_VADDR - define output section for percpu area + * @vaddr: explicit base address (optional) + * @phdr: destination PHDR (optional) + * + * Macro which expands to output section for percpu area. Mostly + * identical to PERCPU_VADDR_PREALLOC(@vaddr, @phdr, 0) other than + * using slighly different layout. + */ +#define PERCPU_VADDR(vaddr, phdr) \ + PERCPU_PROLOG(vaddr) \ *(.data.percpu.page_aligned) \ *(.data.percpu) \ *(.data.percpu.shared_aligned) \ - } \ - VMLINUX_SYMBOL(__per_cpu_end) = .; + PERCPU_EPILOG(phdr) + +/** + * PERCPU - define output section for percpu area, simple version + * @align: required alignment + * + * Align to @align and outputs output section for percpu area. This + * macro doesn't maniuplate @vaddr or @phdr and __per_cpu_load and + * __per_cpu_start will be identical. + */ +#define PERCPU(align) \ + . = ALIGN(align); \ + PERCPU_VADDR( , ) diff --git a/include/linux/init_task.h b/include/linux/init_task.h index 2f3c2d4ef73..49a40fbc806 100644 --- a/include/linux/init_task.h +++ b/include/linux/init_task.h @@ -115,6 +115,16 @@ 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.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) @@ -179,6 +189,7 @@ extern struct cred init_cred; INIT_IDS \ INIT_TRACE_IRQFLAGS \ INIT_LOCKDEP \ + INIT_PERF_COUNTERS(tsk) \ } diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h index 9127f6b51a3..472f11765f6 100644 --- a/include/linux/interrupt.h +++ b/include/linux/interrupt.h @@ -467,6 +467,7 @@ int show_interrupts(struct seq_file *p, void *v); struct irq_desc; extern int early_irq_init(void); +extern int arch_probe_nr_irqs(void); extern int arch_early_irq_init(void); extern int arch_init_chip_data(struct irq_desc *desc, int cpu); diff --git a/include/linux/irq.h b/include/linux/irq.h index f899b502f18..27a67536511 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -182,11 +182,11 @@ struct irq_desc { unsigned int irqs_unhandled; spinlock_t lock; #ifdef CONFIG_SMP - cpumask_t affinity; + cpumask_var_t affinity; unsigned int cpu; -#endif #ifdef CONFIG_GENERIC_PENDING_IRQ - cpumask_t pending_mask; + cpumask_var_t pending_mask; +#endif #endif #ifdef CONFIG_PROC_FS struct proc_dir_entry *dir; @@ -422,4 +422,84 @@ extern int set_irq_msi(unsigned int irq, struct msi_desc *entry); #endif /* !CONFIG_S390 */ +#ifdef CONFIG_SMP +/** + * init_alloc_desc_masks - allocate cpumasks for irq_desc + * @desc: pointer to irq_desc struct + * @cpu: cpu which will be handling the cpumasks + * @boot: true if need bootmem + * + * Allocates affinity and pending_mask cpumask if required. + * Returns true if successful (or not required). + * Side effect: affinity has all bits set, pending_mask has all bits clear. + */ +static inline bool init_alloc_desc_masks(struct irq_desc *desc, int cpu, + bool boot) +{ + int node; + + if (boot) { + alloc_bootmem_cpumask_var(&desc->affinity); + cpumask_setall(desc->affinity); + +#ifdef CONFIG_GENERIC_PENDING_IRQ + alloc_bootmem_cpumask_var(&desc->pending_mask); + cpumask_clear(desc->pending_mask); +#endif + return true; + } + + node = cpu_to_node(cpu); + + if (!alloc_cpumask_var_node(&desc->affinity, GFP_ATOMIC, node)) + return false; + cpumask_setall(desc->affinity); + +#ifdef CONFIG_GENERIC_PENDING_IRQ + if (!alloc_cpumask_var_node(&desc->pending_mask, GFP_ATOMIC, node)) { + free_cpumask_var(desc->affinity); + return false; + } + cpumask_clear(desc->pending_mask); +#endif + return true; +} + +/** + * init_copy_desc_masks - copy cpumasks for irq_desc + * @old_desc: pointer to old irq_desc struct + * @new_desc: pointer to new irq_desc struct + * + * Insures affinity and pending_masks are copied to new irq_desc. + * If !CONFIG_CPUMASKS_OFFSTACK the cpumasks are embedded in the + * irq_desc struct so the copy is redundant. + */ + +static inline void init_copy_desc_masks(struct irq_desc *old_desc, + struct irq_desc *new_desc) +{ +#ifdef CONFIG_CPUMASKS_OFFSTACK + cpumask_copy(new_desc->affinity, old_desc->affinity); + +#ifdef CONFIG_GENERIC_PENDING_IRQ + cpumask_copy(new_desc->pending_mask, old_desc->pending_mask); +#endif +#endif +} + +#else /* !CONFIG_SMP */ + +static inline bool init_alloc_desc_masks(struct irq_desc *desc, int cpu, + bool boot) +{ + return true; +} + +static inline void init_copy_desc_masks(struct irq_desc *old_desc, + struct irq_desc *new_desc) +{ +} + +#endif /* CONFIG_SMP */ + #endif /* _LINUX_IRQ_H */ diff --git a/include/linux/irqnr.h b/include/linux/irqnr.h index 86af92e9e84..887477bc2ab 100644 --- a/include/linux/irqnr.h +++ b/include/linux/irqnr.h @@ -20,6 +20,7 @@ # define for_each_irq_desc_reverse(irq, desc) \ for (irq = nr_irqs - 1; irq >= 0; irq--) + #else /* CONFIG_GENERIC_HARDIRQS */ extern int nr_irqs; diff --git a/include/linux/kernel_stat.h b/include/linux/kernel_stat.h index 570d2041311..ecfa6681763 100644 --- a/include/linux/kernel_stat.h +++ b/include/linux/kernel_stat.h @@ -78,7 +78,15 @@ static inline unsigned int kstat_irqs(unsigned int irq) return sum; } + +/* + * Lock/unlock the current runqueue - to extract task statistics: + */ +extern void curr_rq_lock_irq_save(unsigned long *flags); +extern void curr_rq_unlock_irq_restore(unsigned long *flags); +extern unsigned long long __task_delta_exec(struct task_struct *tsk, int update); extern unsigned long long task_delta_exec(struct task_struct *); + extern void account_user_time(struct task_struct *, cputime_t, cputime_t); extern void account_system_time(struct task_struct *, int, cputime_t, cputime_t); extern void account_steal_time(cputime_t); diff --git a/include/linux/perf_counter.h b/include/linux/perf_counter.h new file mode 100644 index 00000000000..33ba9fe0a78 --- /dev/null +++ b/include/linux/perf_counter.h @@ -0,0 +1,290 @@ +/* + * Performance counters: + * + * Copyright(C) 2008, Thomas Gleixner <tglx@linutronix.de> + * Copyright(C) 2008, Red Hat, Inc., Ingo Molnar + * + * Data type definitions, declarations, prototypes. + * + * Started by: Thomas Gleixner and Ingo Molnar + * + * For licencing details see kernel-base/COPYING + */ +#ifndef _LINUX_PERF_COUNTER_H +#define _LINUX_PERF_COUNTER_H + +#include <asm/atomic.h> +#include <asm/ioctl.h> + +#ifdef CONFIG_PERF_COUNTERS +# include <asm/perf_counter.h> +#endif + +#include <linux/list.h> +#include <linux/mutex.h> +#include <linux/rculist.h> +#include <linux/rcupdate.h> +#include <linux/spinlock.h> + +struct task_struct; + +/* + * User-space ABI bits: + */ + +/* + * Generalized performance counter event types, used by the hw_event.type + * parameter of the sys_perf_counter_open() syscall: + */ +enum hw_event_types { + /* + * Common hardware events, generalized by the kernel: + */ + PERF_COUNT_CPU_CYCLES = 0, + PERF_COUNT_INSTRUCTIONS = 1, + PERF_COUNT_CACHE_REFERENCES = 2, + PERF_COUNT_CACHE_MISSES = 3, + PERF_COUNT_BRANCH_INSTRUCTIONS = 4, + PERF_COUNT_BRANCH_MISSES = 5, + PERF_COUNT_BUS_CYCLES = 6, + + PERF_HW_EVENTS_MAX = 7, + + /* + * Special "software" counters provided by the kernel, even if + * the hardware does not support performance counters. These + * counters measure various physical and sw events of the + * kernel (and allow the profiling of them as well): + */ + PERF_COUNT_CPU_CLOCK = -1, + PERF_COUNT_TASK_CLOCK = -2, + PERF_COUNT_PAGE_FAULTS = -3, + PERF_COUNT_CONTEXT_SWITCHES = -4, + PERF_COUNT_CPU_MIGRATIONS = -5, + + PERF_SW_EVENTS_MIN = -6, +}; + +/* + * IRQ-notification data record type: + */ +enum perf_counter_record_type { + PERF_RECORD_SIMPLE = 0, + PERF_RECORD_IRQ = 1, + PERF_RECORD_GROUP = 2, +}; + +/* + * Hardware event to monitor via a performance monitoring counter: + */ +struct perf_counter_hw_event { + s64 type; + + u64 irq_period; + u32 record_type; + + u32 disabled : 1, /* off by default */ + nmi : 1, /* NMI sampling */ + raw : 1, /* raw event type */ + inherit : 1, /* children inherit it */ + pinned : 1, /* must always be on PMU */ + exclusive : 1, /* only counter on PMU */ + + __reserved_1 : 26; + + u64 __reserved_2; +}; + +/* + * Ioctls that can be done on a perf counter fd: + */ +#define PERF_COUNTER_IOC_ENABLE _IO('$', 0) +#define PERF_COUNTER_IOC_DISABLE _IO('$', 1) + +/* + * Kernel-internal data types: + */ + +/** + * struct hw_perf_counter - performance counter hardware details: + */ +struct hw_perf_counter { +#ifdef CONFIG_PERF_COUNTERS + u64 config; + unsigned long config_base; + unsigned long counter_base; + int nmi; + unsigned int idx; + atomic64_t prev_count; + u64 irq_period; + atomic64_t period_left; +#endif +}; + +/* + * Hardcoded buffer length limit for now, for IRQ-fed events: + */ +#define PERF_DATA_BUFLEN 2048 + +/** + * struct perf_data - performance counter IRQ data sampling ... + */ +struct perf_data { + int len; + int rd_idx; + int overrun; + u8 data[PERF_DATA_BUFLEN]; +}; + +struct perf_counter; + +/** + * struct hw_perf_counter_ops - performance counter hw ops + */ +struct hw_perf_counter_ops { + int (*enable) (struct perf_counter *counter); + void (*disable) (struct perf_counter *counter); + void (*read) (struct perf_counter *counter); +}; + +/** + * enum perf_counter_active_state - the states of a counter + */ +enum perf_counter_active_state { + PERF_COUNTER_STATE_ERROR = -2, + PERF_COUNTER_STATE_OFF = -1, + PERF_COUNTER_STATE_INACTIVE = 0, + PERF_COUNTER_STATE_ACTIVE = 1, +}; + +struct file; + +/** + * struct perf_counter - performance counter kernel representation: + */ +struct perf_counter { +#ifdef CONFIG_PERF_COUNTERS + struct list_head list_entry; + struct list_head sibling_list; + struct perf_counter *group_leader; + const struct hw_perf_counter_ops *hw_ops; + + enum perf_counter_active_state state; + atomic64_t count; + + struct perf_counter_hw_event hw_event; + struct hw_perf_counter hw; + + struct perf_counter_context *ctx; + struct task_struct *task; + struct file *filp; + + struct perf_counter *parent; + struct list_head child_list; + + /* + * Protect attach/detach and child_list: + */ + struct mutex mutex; + + int oncpu; + int cpu; + + /* read() / irq related data */ + wait_queue_head_t waitq; + /* optional: for NMIs */ + int wakeup_pending; + struct perf_data *irqdata; + struct perf_data *usrdata; + struct perf_data data[2]; +#endif +}; + +/** + * struct perf_counter_context - counter context structure + * + * 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: + */ + spinlock_t lock; + /* + * Protect the list of counters. Locking either mutex or lock + * is sufficient to ensure the list doesn't change; to change + * the list you need to lock both the mutex and the spinlock. + */ + struct mutex mutex; + + struct list_head counter_list; + int nr_counters; + int nr_active; + int is_active; + struct task_struct *task; +#endif +}; + +/** + * struct perf_counter_cpu_context - per cpu counter context structure + */ +struct perf_cpu_context { + struct perf_counter_context ctx; + struct perf_counter_context *task_ctx; + int active_oncpu; + int max_pertask; + int exclusive; +}; + +/* + * Set by architecture code: + */ +extern int perf_max_counters; + +#ifdef CONFIG_PERF_COUNTERS +extern const struct hw_perf_counter_ops * +hw_perf_counter_init(struct perf_counter *counter); + +extern void perf_counter_task_sched_in(struct task_struct *task, int cpu); +extern void perf_counter_task_sched_out(struct task_struct *task, int cpu); +extern void perf_counter_task_tick(struct task_struct *task, int cpu); +extern void perf_counter_init_task(struct task_struct *child); +extern void perf_counter_exit_task(struct task_struct *child); +extern void perf_counter_notify(struct pt_regs *regs); +extern void perf_counter_print_debug(void); +extern u64 hw_perf_save_disable(void); +extern void hw_perf_restore(u64 ctrl); +extern int perf_counter_task_disable(void); +extern int perf_counter_task_enable(void); +extern int hw_perf_group_sched_in(struct perf_counter *group_leader, + struct perf_cpu_context *cpuctx, + struct perf_counter_context *ctx, int cpu); + +/* + * Return 1 for a software counter, 0 for a hardware counter + */ +static inline int is_software_counter(struct perf_counter *counter) +{ + return !counter->hw_event.raw && counter->hw_event.type < 0; +} + +#else +static inline void +perf_counter_task_sched_in(struct task_struct *task, int cpu) { } +static inline void +perf_counter_task_sched_out(struct task_struct *task, int cpu) { } +static inline void +perf_counter_task_tick(struct task_struct *task, int cpu) { } +static inline void perf_counter_init_task(struct task_struct *child) { } +static inline void perf_counter_exit_task(struct task_struct *child) { } +static inline void perf_counter_notify(struct pt_regs *regs) { } +static inline void perf_counter_print_debug(void) { } +static inline void hw_perf_restore(u64 ctrl) { } +static inline u64 hw_perf_save_disable(void) { return 0; } +static inline int perf_counter_task_disable(void) { return -EINVAL; } +static inline int perf_counter_task_enable(void) { return -EINVAL; } +#endif + +#endif /* _LINUX_PERF_COUNTER_H */ diff --git a/include/linux/prctl.h b/include/linux/prctl.h index 48d887e3c6e..b00df4c79c6 100644 --- a/include/linux/prctl.h +++ b/include/linux/prctl.h @@ -85,4 +85,7 @@ #define PR_SET_TIMERSLACK 29 #define PR_GET_TIMERSLACK 30 +#define PR_TASK_PERF_COUNTERS_DISABLE 31 +#define PR_TASK_PERF_COUNTERS_ENABLE 32 + #endif /* _LINUX_PRCTL_H */ diff --git a/include/linux/sched.h b/include/linux/sched.h index 4cae9b81a1f..f134a0f7080 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -71,6 +71,7 @@ struct sched_param { #include <linux/fs_struct.h> #include <linux/compiler.h> #include <linux/completion.h> +#include <linux/perf_counter.h> #include <linux/pid.h> #include <linux/percpu.h> #include <linux/topology.h> @@ -1031,6 +1032,8 @@ struct sched_entity { u64 last_wakeup; u64 avg_overlap; + u64 nr_migrations; + #ifdef CONFIG_SCHEDSTATS u64 wait_start; u64 wait_max; @@ -1046,7 +1049,6 @@ struct sched_entity { u64 exec_max; u64 slice_max; - u64 nr_migrations; u64 nr_migrations_cold; u64 nr_failed_migrations_affine; u64 nr_failed_migrations_running; @@ -1349,6 +1351,7 @@ 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_NUMA struct mempolicy *mempolicy; short il_next; @@ -2322,6 +2325,13 @@ static inline void inc_syscw(struct task_struct *tsk) #define TASK_SIZE_OF(tsk) TASK_SIZE #endif +/* + * Call the function if the target task is executing on a CPU right now: + */ +extern void task_oncpu_function_call(struct task_struct *p, + void (*func) (void *info), void *info); + + #ifdef CONFIG_MM_OWNER extern void mm_update_next_owner(struct mm_struct *mm); extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p); diff --git a/include/linux/syscalls.h b/include/linux/syscalls.h index 16875f89e6a..fc81937009f 100644 --- a/include/linux/syscalls.h +++ b/include/linux/syscalls.h @@ -55,6 +55,7 @@ struct compat_timeval; struct robust_list_head; struct getcpu_cache; struct old_linux_dirent; +struct perf_counter_hw_event; #include <linux/types.h> #include <linux/aio_abi.h> @@ -689,4 +690,11 @@ asmlinkage long sys_pipe(int __user *); int kernel_execve(const char *filename, char *const argv[], char *const envp[]); + +asmlinkage int sys_perf_counter_open( + + struct perf_counter_hw_event *hw_event_uptr __user, + pid_t pid, + int cpu, + int group_fd); #endif diff --git a/include/linux/topology.h b/include/linux/topology.h index e632d29f054..a16b9e06f2e 100644 --- a/include/linux/topology.h +++ b/include/linux/topology.h @@ -193,5 +193,11 @@ int arch_update_cpu_topology(void); #ifndef topology_core_siblings #define topology_core_siblings(cpu) cpumask_of_cpu(cpu) #endif +#ifndef topology_thread_cpumask +#define topology_thread_cpumask(cpu) cpumask_of(cpu) +#endif +#ifndef topology_core_cpumask +#define topology_core_cpumask(cpu) cpumask_of(cpu) +#endif #endif /* _LINUX_TOPOLOGY_H */ diff --git a/init/Kconfig b/init/Kconfig index 2af83825634..6af96b7e223 100644 --- a/init/Kconfig +++ b/init/Kconfig @@ -776,6 +776,36 @@ config AIO by some high performance threaded applications. Disabling this option saves about 7k. +config HAVE_PERF_COUNTERS + bool + +menu "Performance Counters" + +config PERF_COUNTERS + bool "Kernel Performance Counters" + depends on HAVE_PERF_COUNTERS + default y + select ANON_INODES + help + Enable kernel support for performance counter hardware. + + Performance counters are special hardware registers available + on most modern CPUs. These registers count the number of certain + types of hw events: such as instructions executed, cachemisses + suffered, or branches mis-predicted - without slowing down the + kernel or applications. These registers can also trigger interrupts + when a threshold number of events have passed - and can thus be + used to profile the code that runs on that CPU. + + The Linux Performance Counter subsystem provides an abstraction of + these hardware capabilities, available via a system call. It + provides per task and per CPU counters, and it provides event + capabilities on top of those. + + Say Y if unsure. + +endmenu + config VM_EVENT_COUNTERS default y bool "Enable VM event counters for /proc/vmstat" if EMBEDDED diff --git a/kernel/Makefile b/kernel/Makefile index 170a9213c1b..5537554ed80 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -92,6 +92,7 @@ obj-$(CONFIG_HAVE_GENERIC_DMA_COHERENT) += dma-coherent.o obj-$(CONFIG_FUNCTION_TRACER) += trace/ obj-$(CONFIG_TRACING) += trace/ obj-$(CONFIG_SMP) += sched_cpupri.o +obj-$(CONFIG_PERF_COUNTERS) += perf_counter.o ifneq ($(CONFIG_SCHED_OMIT_FRAME_POINTER),y) # According to Alan Modra <alan@linuxcare.com.au>, the -fno-omit-frame-pointer is diff --git a/kernel/exit.c b/kernel/exit.c index f80dec3f187..29f4b790751 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -159,6 +159,9 @@ 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)); +#endif trace_sched_process_free(tsk); put_task_struct(tsk); } @@ -1093,10 +1096,6 @@ NORET_TYPE void do_exit(long code) tsk->mempolicy = NULL; #endif #ifdef CONFIG_FUTEX - /* - * This must happen late, after the PID is not - * hashed anymore: - */ if (unlikely(!list_empty(&tsk->pi_state_list))) exit_pi_state_list(tsk); if (unlikely(current->pi_state_cache)) @@ -1363,6 +1362,12 @@ static int wait_task_zombie(struct task_struct *p, int options, */ read_unlock(&tasklist_lock); + /* + * Flush inherited counters to the parent - before the parent + * gets woken up by child-exit notifications. + */ + perf_counter_exit_task(p); + retval = ru ? getrusage(p, RUSAGE_BOTH, ru) : 0; status = (p->signal->flags & SIGNAL_GROUP_EXIT) ? p->signal->group_exit_code : p->exit_code; diff --git a/kernel/fork.c b/kernel/fork.c index bf0cef8bbdf..70ca1852a0e 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -985,6 +985,7 @@ static struct task_struct *copy_process(unsigned long clone_flags, goto fork_out; rt_mutex_init_task(p); + perf_counter_init_task(p); #ifdef CONFIG_PROVE_LOCKING DEBUG_LOCKS_WARN_ON(!p->hardirqs_enabled); diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index f63c706d25e..c248eba98b4 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -46,7 +46,10 @@ void dynamic_irq_init(unsigned int irq) desc->irq_count = 0; desc->irqs_unhandled = 0; #ifdef CONFIG_SMP - cpumask_setall(&desc->affinity); + cpumask_setall(desc->affinity); +#ifdef CONFIG_GENERIC_PENDING_IRQ + cpumask_clear(desc->pending_mask); +#endif #endif spin_unlock_irqrestore(&desc->lock, flags); } diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index c20db0be917..375d68cd5bf 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -17,6 +17,7 @@ #include <linux/kernel_stat.h> #include <linux/rculist.h> #include <linux/hash.h> +#include <linux/bootmem.h> #include "internals.h" @@ -57,6 +58,7 @@ int nr_irqs = NR_IRQS; EXPORT_SYMBOL_GPL(nr_irqs); #ifdef CONFIG_SPARSE_IRQ + static struct irq_desc irq_desc_init = { .irq = -1, .status = IRQ_DISABLED, @@ -64,9 +66,6 @@ static struct irq_desc irq_desc_init = { .handle_irq = handle_bad_irq, .depth = 1, .lock = __SPIN_LOCK_UNLOCKED(irq_desc_init.lock), -#ifdef CONFIG_SMP - .affinity = CPU_MASK_ALL -#endif }; void init_kstat_irqs(struct irq_desc *desc, int cpu, int nr) @@ -101,6 +100,10 @@ static void init_one_irq_desc(int irq, struct irq_desc *desc, int cpu) printk(KERN_ERR "can not alloc kstat_irqs\n"); BUG_ON(1); } + if (!init_alloc_desc_masks(desc, cpu, false)) { + printk(KERN_ERR "can not alloc irq_desc cpumasks\n"); + BUG_ON(1); + } arch_init_chip_data(desc, cpu); } @@ -109,7 +112,7 @@ static void init_one_irq_desc(int irq, struct irq_desc *desc, int cpu) */ DEFINE_SPINLOCK(sparse_irq_lock); -struct irq_desc *irq_desc_ptrs[NR_IRQS] __read_mostly; +struct irq_desc **irq_desc_ptrs __read_mostly; static struct irq_desc irq_desc_legacy[NR_IRQS_LEGACY] __cacheline_aligned_in_smp = { [0 ... NR_IRQS_LEGACY-1] = { @@ -119,14 +122,10 @@ static struct irq_desc irq_desc_legacy[NR_IRQS_LEGACY] __cacheline_aligned_in_sm .handle_irq = handle_bad_irq, .depth = 1, .lock = __SPIN_LOCK_UNLOCKED(irq_desc_init.lock), -#ifdef CONFIG_SMP - .affinity = CPU_MASK_ALL -#endif } }; -/* FIXME: use bootmem alloc ...*/ -static unsigned int kstat_irqs_legacy[NR_IRQS_LEGACY][NR_CPUS]; +static unsigned int *kstat_irqs_legacy; int __init early_irq_init(void) { @@ -134,18 +133,30 @@ int __init early_irq_init(void) int legacy_count; int i; + /* initialize nr_irqs based on nr_cpu_ids */ + arch_probe_nr_irqs(); + printk(KERN_INFO "NR_IRQS:%d nr_irqs:%d\n", NR_IRQS, nr_irqs); + desc = irq_desc_legacy; legacy_count = ARRAY_SIZE(irq_desc_legacy); + /* allocate irq_desc_ptrs array based on nr_irqs */ + irq_desc_ptrs = alloc_bootmem(nr_irqs * sizeof(void *)); + + /* allocate based on nr_cpu_ids */ + /* FIXME: invert kstat_irgs, and it'd be a per_cpu_alloc'd thing */ + kstat_irqs_legacy = alloc_bootmem(NR_IRQS_LEGACY * nr_cpu_ids * + sizeof(int)); + for (i = 0; i < legacy_count; i++) { desc[i].irq = i; - desc[i].kstat_irqs = kstat_irqs_legacy[i]; + desc[i].kstat_irqs = kstat_irqs_legacy + i * nr_cpu_ids; lockdep_set_class(&desc[i].lock, &irq_desc_lock_class); - + init_alloc_desc_masks(&desc[i], 0, true); irq_desc_ptrs[i] = desc + i; } - for (i = legacy_count; i < NR_IRQS; i++) + for (i = legacy_count; i < nr_irqs; i++) irq_desc_ptrs[i] = NULL; return arch_early_irq_init(); @@ -153,7 +164,10 @@ int __init early_irq_init(void) struct irq_desc *irq_to_desc(unsigned int irq) { - return (irq < NR_IRQS) ? irq_desc_ptrs[irq] : NULL; + if (irq_desc_ptrs && irq < nr_irqs) + return irq_desc_ptrs[irq]; + + return NULL; } struct irq_desc *irq_to_desc_alloc_cpu(unsigned int irq, int cpu) @@ -162,10 +176,9 @@ struct irq_desc *irq_to_desc_alloc_cpu(unsigned int irq, int cpu) unsigned long flags; int node; - if (irq >= NR_IRQS) { - printk(KERN_WARNING "irq >= NR_IRQS in irq_to_desc_alloc: %d %d\n", - irq, NR_IRQS); - WARN_ON(1); + if (irq >= nr_irqs) { + WARN(1, "irq (%d) >= nr_irqs (%d) in irq_to_desc_alloc\n", + irq, nr_irqs); return NULL; } @@ -207,9 +220,6 @@ struct irq_desc irq_desc[NR_IRQS] __cacheline_aligned_in_smp = { .handle_irq = handle_bad_irq, .depth = 1, .lock = __SPIN_LOCK_UNLOCKED(irq_desc->lock), -#ifdef CONFIG_SMP - .affinity = CPU_MASK_ALL -#endif } }; @@ -219,12 +229,15 @@ int __init early_irq_init(void) int count; int i; + printk(KERN_INFO "NR_IRQS:%d\n", NR_IRQS); + desc = irq_desc; count = ARRAY_SIZE(irq_desc); - for (i = 0; i < count; i++) + for (i = 0; i < count; i++) { desc[i].irq = i; - + init_alloc_desc_masks(&desc[i], 0, true); + } return arch_early_irq_init(); } diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index e6d0a43cc12..40416a81a0f 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -16,7 +16,14 @@ extern int __irq_set_trigger(struct irq_desc *desc, unsigned int irq, extern struct lock_class_key irq_desc_lock_class; extern void init_kstat_irqs(struct irq_desc *desc, int cpu, int nr); extern spinlock_t sparse_irq_lock; + +#ifdef CONFIG_SPARSE_IRQ +/* irq_desc_ptrs allocated at boot time */ +extern struct irq_desc **irq_desc_ptrs; +#else +/* irq_desc_ptrs is a fixed size array */ extern struct irq_desc *irq_desc_ptrs[NR_IRQS]; +#endif #ifdef CONFIG_PROC_FS extern void register_irq_proc(unsigned int irq, struct irq_desc *desc); diff --git a/kernel/irq/manage.c b/kernel/irq/manage.c index cd0cd8dcb34..b98739af455 100644 --- a/kernel/irq/manage.c +++ b/kernel/irq/manage.c @@ -98,14 +98,14 @@ int irq_set_affinity(unsigned int irq, const struct cpumask *cpumask) #ifdef CONFIG_GENERIC_PENDING_IRQ if (desc->status & IRQ_MOVE_PCNTXT || desc->status & IRQ_DISABLED) { - cpumask_copy(&desc->affinity, cpumask); + cpumask_copy(desc->affinity, cpumask); desc->chip->set_affinity(irq, cpumask); } else { desc->status |= IRQ_MOVE_PENDING; - cpumask_copy(&desc->pending_mask, cpumask); + cpumask_copy(desc->pending_mask, cpumask); } #else - cpumask_copy(&desc->affinity, cpumask); + cpumask_copy(desc->affinity, cpumask); desc->chip->set_affinity(irq, cpumask); #endif desc->status |= IRQ_AFFINITY_SET; @@ -127,16 +127,16 @@ int do_irq_select_affinity(unsigned int irq, struct irq_desc *desc) * one of the targets is online. */ if (desc->status & (IRQ_AFFINITY_SET | IRQ_NO_BALANCING)) { - if (cpumask_any_and(&desc->affinity, cpu_online_mask) + if (cpumask_any_and(desc->affinity, cpu_online_mask) < nr_cpu_ids) goto set_affinity; else desc->status &= ~IRQ_AFFINITY_SET; } - cpumask_and(&desc->affinity, cpu_online_mask, irq_default_affinity); + cpumask_and(desc->affinity, cpu_online_mask, irq_default_affinity); set_affinity: - desc->chip->set_affinity(irq, &desc->affinity); + desc->chip->set_affinity(irq, desc->affinity); return 0; } diff --git a/kernel/irq/migration.c b/kernel/irq/migration.c index bd72329e630..e05ad9be43b 100644 --- a/kernel/irq/migration.c +++ b/kernel/irq/migration.c @@ -18,7 +18,7 @@ void move_masked_irq(int irq) desc->status &= ~IRQ_MOVE_PENDING; - if (unlikely(cpumask_empty(&desc->pending_mask))) + if (unlikely(cpumask_empty(desc->pending_mask))) return; if (!desc->chip->set_affinity) @@ -38,13 +38,13 @@ void move_masked_irq(int irq) * For correct operation this depends on the caller * masking the irqs. */ - if (likely(cpumask_any_and(&desc->pending_mask, cpu_online_mask) + if (likely(cpumask_any_and(desc->pending_mask, cpu_online_mask) < nr_cpu_ids)) { - cpumask_and(&desc->affinity, - &desc->pending_mask, cpu_online_mask); - desc->chip->set_affinity(irq, &desc->affinity); + cpumask_and(desc->affinity, + desc->pending_mask, cpu_online_mask); + desc->chip->set_affinity(irq, desc->affinity); } - cpumask_clear(&desc->pending_mask); + cpumask_clear(desc->pending_mask); } void move_native_irq(int irq) diff --git a/kernel/irq/numa_migrate.c b/kernel/irq/numa_migrate.c index ecf765c6a77..666260e4c06 100644 --- a/kernel/irq/numa_migrate.c +++ b/kernel/irq/numa_migrate.c @@ -38,15 +38,22 @@ static void free_kstat_irqs(struct irq_desc *old_desc, struct irq_desc *desc) old_desc->kstat_irqs = NULL; } -static void init_copy_one_irq_desc(int irq, struct irq_desc *old_desc, +static bool init_copy_one_irq_desc(int irq, struct irq_desc *old_desc, struct irq_desc *desc, int cpu) { memcpy(desc, old_desc, sizeof(struct irq_desc)); + if (!init_alloc_desc_masks(desc, cpu, false)) { + printk(KERN_ERR "irq %d: can not get new irq_desc cpumask " + "for migration.\n", irq); + return false; + } spin_lock_init(&desc->lock); desc->cpu = cpu; lockdep_set_class(&desc->lock, &irq_desc_lock_class); init_copy_kstat_irqs(old_desc, desc, cpu, nr_cpu_ids); + init_copy_desc_masks(old_desc, desc); arch_init_copy_chip_data(old_desc, desc, cpu); + return true; } static void free_one_irq_desc(struct irq_desc *old_desc, struct irq_desc *desc) @@ -76,12 +83,18 @@ static struct irq_desc *__real_move_irq_desc(struct irq_desc *old_desc, node = cpu_to_node(cpu); desc = kzalloc_node(sizeof(*desc), GFP_ATOMIC, node); if (!desc) { - printk(KERN_ERR "irq %d: can not get new irq_desc for migration.\n", irq); + printk(KERN_ERR "irq %d: can not get new irq_desc " + "for migration.\n", irq); + /* still use old one */ + desc = old_desc; + goto out_unlock; + } + if (!init_copy_one_irq_desc(irq, old_desc, desc, cpu)) { /* still use old one */ + kfree(desc); desc = old_desc; goto out_unlock; } - init_copy_one_irq_desc(irq, old_desc, desc, cpu); irq_desc_ptrs[irq] = desc; diff --git a/kernel/irq/proc.c b/kernel/irq/proc.c index aae3f742bce..692363dd591 100644 --- a/kernel/irq/proc.c +++ b/kernel/irq/proc.c @@ -20,11 +20,11 @@ static struct proc_dir_entry *root_irq_dir; static int irq_affinity_proc_show(struct seq_file *m, void *v) { struct irq_desc *desc = irq_to_desc((long)m->private); - const struct cpumask *mask = &desc->affinity; + const struct cpumask *mask = desc->affinity; #ifdef CONFIG_GENERIC_PENDING_IRQ if (desc->status & IRQ_MOVE_PENDING) - mask = &desc->pending_mask; + mask = desc->pending_mask; #endif seq_cpumask(m, mask); seq_putc(m, '\n'); diff --git a/kernel/perf_counter.c b/kernel/perf_counter.c new file mode 100644 index 00000000000..1ac18daa424 --- /dev/null +++ b/kernel/perf_counter.c @@ -0,0 +1,2169 @@ +/* + * Performance counter core code + * + * Copyright(C) 2008 Thomas Gleixner <tglx@linutronix.de> + * Copyright(C) 2008 Red Hat, Inc., Ingo Molnar + * + * For licencing details see kernel-base/COPYING + */ + +#include <linux/fs.h> +#include <linux/cpu.h> +#include <linux/smp.h> +#include <linux/file.h> +#include <linux/poll.h> +#include <linux/sysfs.h> +#include <linux/ptrace.h> +#include <linux/percpu.h> +#include <linux/uaccess.h> +#include <linux/syscalls.h> +#include <linux/anon_inodes.h> +#include <linux/kernel_stat.h> +#include <linux/perf_counter.h> + +/* + * Each CPU has a list of per CPU counters: + */ +DEFINE_PER_CPU(struct perf_cpu_context, perf_cpu_context); + +int perf_max_counters __read_mostly = 1; +static int perf_reserved_percpu __read_mostly; +static int perf_overcommit __read_mostly = 1; + +/* + * Mutex for (sysadmin-configurable) counter reservations: + */ +static DEFINE_MUTEX(perf_resource_mutex); + +/* + * Architecture provided APIs - weak aliases: + */ +extern __weak const struct hw_perf_counter_ops * +hw_perf_counter_init(struct perf_counter *counter) +{ + return NULL; +} + +u64 __weak hw_perf_save_disable(void) { return 0; } +void __weak hw_perf_restore(u64 ctrl) { barrier(); } +void __weak hw_perf_counter_setup(int cpu) { barrier(); } +int __weak hw_perf_group_sched_in(struct perf_counter *group_leader, + struct perf_cpu_context *cpuctx, + struct perf_counter_context *ctx, int cpu) +{ + return 0; +} + +void __weak perf_counter_print_debug(void) { } + +static void +list_add_counter(struct perf_counter *counter, struct perf_counter_context *ctx) +{ + struct perf_counter *group_leader = counter->group_leader; + + /* + * Depending on whether it is a standalone or sibling counter, + * add it straight to the context's counter list, or to the group + * leader's sibling list: + */ + if (counter->group_leader == counter) + list_add_tail(&counter->list_entry, &ctx->counter_list); + else + list_add_tail(&counter->list_entry, &group_leader->sibling_list); +} + +static void +list_del_counter(struct perf_counter *counter, struct perf_counter_context *ctx) +{ + struct perf_counter *sibling, *tmp; + + list_del_init(&counter->list_entry); + + /* + * If this was a group counter with sibling counters then + * upgrade the siblings to singleton counters by adding them + * to the context list directly: + */ + list_for_each_entry_safe(sibling, tmp, + &counter->sibling_list, list_entry) { + + list_del_init(&sibling->list_entry); + list_add_tail(&sibling->list_entry, &ctx->counter_list); + sibling->group_leader = sibling; + } +} + +static void +counter_sched_out(struct perf_counter *counter, + struct perf_cpu_context *cpuctx, + struct perf_counter_context *ctx) +{ + if (counter->state != PERF_COUNTER_STATE_ACTIVE) + return; + + counter->state = PERF_COUNTER_STATE_INACTIVE; + counter->hw_ops->disable(counter); + counter->oncpu = -1; + + if (!is_software_counter(counter)) + cpuctx->active_oncpu--; + ctx->nr_active--; + if (counter->hw_event.exclusive || !cpuctx->active_oncpu) + cpuctx->exclusive = 0; +} + +static void +group_sched_out(struct perf_counter *group_counter, + struct perf_cpu_context *cpuctx, + struct perf_counter_context *ctx) +{ + struct perf_counter *counter; + + if (group_counter->state != PERF_COUNTER_STATE_ACTIVE) + return; + + counter_sched_out(group_counter, cpuctx, ctx); + + /* + * Schedule out siblings (if any): + */ + list_for_each_entry(counter, &group_counter->sibling_list, list_entry) + counter_sched_out(counter, cpuctx, ctx); + + if (group_counter->hw_event.exclusive) + cpuctx->exclusive = 0; +} + +/* + * Cross CPU call to remove a performance counter + * + * We disable the counter on the hardware level first. After that we + * remove it from the context list. + */ +static void __perf_counter_remove_from_context(void *info) +{ + struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + struct perf_counter *counter = info; + struct perf_counter_context *ctx = counter->ctx; + unsigned long flags; + u64 perf_flags; + + /* + * 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. + */ + if (ctx->task && cpuctx->task_ctx != ctx) + return; + + curr_rq_lock_irq_save(&flags); + spin_lock(&ctx->lock); + + counter_sched_out(counter, cpuctx, ctx); + + counter->task = NULL; + ctx->nr_counters--; + + /* + * Protect the list operation against NMI by disabling the + * counters on a global level. NOP for non NMI based counters. + */ + perf_flags = hw_perf_save_disable(); + list_del_counter(counter, ctx); + hw_perf_restore(perf_flags); + + if (!ctx->task) { + /* + * Allow more per task counters with respect to the + * reservation: + */ + cpuctx->max_pertask = + min(perf_max_counters - ctx->nr_counters, + perf_max_counters - perf_reserved_percpu); + } + + spin_unlock(&ctx->lock); + curr_rq_unlock_irq_restore(&flags); +} + + +/* + * Remove the counter from a task's (or a CPU's) list of counters. + * + * Must be called with counter->mutex and ctx->mutex held. + * + * CPU counters are removed with a smp call. For task counters we only + * call when the task is on a CPU. + */ +static void perf_counter_remove_from_context(struct perf_counter *counter) +{ + struct perf_counter_context *ctx = counter->ctx; + struct task_struct *task = ctx->task; + + if (!task) { + /* + * Per cpu counters are removed via an smp call and + * the removal is always sucessful. + */ + smp_call_function_single(counter->cpu, + __perf_counter_remove_from_context, + counter, 1); + return; + } + +retry: + task_oncpu_function_call(task, __perf_counter_remove_from_context, + counter); + + spin_lock_irq(&ctx->lock); + /* + * If the context is active we need to retry the smp call. + */ + if (ctx->nr_active && !list_empty(&counter->list_entry)) { + spin_unlock_irq(&ctx->lock); + goto retry; + } + + /* + * The lock prevents that this context is scheduled in so we + * can remove the counter safely, if the call above did not + * succeed. + */ + if (!list_empty(&counter->list_entry)) { + ctx->nr_counters--; + list_del_counter(counter, ctx); + counter->task = NULL; + } + spin_unlock_irq(&ctx->lock); +} + +/* + * Cross CPU call to disable a performance counter + */ +static void __perf_counter_disable(void *info) +{ + struct perf_counter *counter = info; + struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + struct perf_counter_context *ctx = counter->ctx; + unsigned long flags; + + /* + * 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; + + curr_rq_lock_irq_save(&flags); + spin_lock(&ctx->lock); + + /* + * If the counter is on, turn it off. + * If it is in error state, leave it in error state. + */ + if (counter->state >= PERF_COUNTER_STATE_INACTIVE) { + if (counter == counter->group_leader) + group_sched_out(counter, cpuctx, ctx); + else + counter_sched_out(counter, cpuctx, ctx); + counter->state = PERF_COUNTER_STATE_OFF; + } + + spin_unlock(&ctx->lock); + curr_rq_unlock_irq_restore(&flags); +} + +/* + * Disable a counter. + */ +static void perf_counter_disable(struct perf_counter *counter) +{ + struct perf_counter_context *ctx = counter->ctx; + struct task_struct *task = ctx->task; + + if (!task) { + /* + * Disable the counter on the cpu that it's on + */ + smp_call_function_single(counter->cpu, __perf_counter_disable, + counter, 1); + return; + } + + retry: + task_oncpu_function_call(task, __perf_counter_disable, counter); + + spin_lock_irq(&ctx->lock); + /* + * If the counter is still active, we need to retry the cross-call. + */ + if (counter->state == PERF_COUNTER_STATE_ACTIVE) { + spin_unlock_irq(&ctx->lock); + goto retry; + } + + /* + * Since we have the lock this context can't be scheduled + * in, so we can change the state safely. + */ + if (counter->state == PERF_COUNTER_STATE_INACTIVE) + counter->state = PERF_COUNTER_STATE_OFF; + + spin_unlock_irq(&ctx->lock); +} + +/* + * Disable a counter and all its children. + */ +static void perf_counter_disable_family(struct perf_counter *counter) +{ + struct perf_counter *child; + + perf_counter_disable(counter); + + /* + * Lock the mutex to protect the list of children + */ + mutex_lock(&counter->mutex); + list_for_each_entry(child, &counter->child_list, child_list) + perf_counter_disable(child); + mutex_unlock(&counter->mutex); +} + +static int +counter_sched_in(struct perf_counter *counter, + struct perf_cpu_context *cpuctx, + struct perf_counter_context *ctx, + int cpu) +{ + if (counter->state <= PERF_COUNTER_STATE_OFF) + return 0; + + counter->state = PERF_COUNTER_STATE_ACTIVE; + counter->oncpu = cpu; /* TODO: put 'cpu' into cpuctx->cpu */ + /* + * The new state must be visible before we turn it on in the hardware: + */ + smp_wmb(); + + if (counter->hw_ops->enable(counter)) { + counter->state = PERF_COUNTER_STATE_INACTIVE; + counter->oncpu = -1; + return -EAGAIN; + } + + if (!is_software_counter(counter)) + cpuctx->active_oncpu++; + ctx->nr_active++; + + if (counter->hw_event.exclusive) + cpuctx->exclusive = 1; + + return 0; +} + +/* + * Return 1 for a group consisting entirely of software counters, + * 0 if the group contains any hardware counters. + */ +static int is_software_only_group(struct perf_counter *leader) +{ + struct perf_counter *counter; + + if (!is_software_counter(leader)) + return 0; + list_for_each_entry(counter, &leader->sibling_list, list_entry) + if (!is_software_counter(counter)) + return 0; + return 1; +} + +/* + * Work out whether we can put this counter group on the CPU now. + */ +static int group_can_go_on(struct perf_counter *counter, + struct perf_cpu_context *cpuctx, + int can_add_hw) +{ + /* + * Groups consisting entirely of software counters can always go on. + */ + if (is_software_only_group(counter)) + return 1; + /* + * If an exclusive group is already on, no other hardware + * counters can go on. + */ + if (cpuctx->exclusive) + return 0; + /* + * If this group is exclusive and there are already + * counters on the CPU, it can't go on. + */ + if (counter->hw_event.exclusive && cpuctx->active_oncpu) + return 0; + /* + * Otherwise, try to add it if all previous groups were able + * to go on. + */ + return can_add_hw; +} + +/* + * Cross CPU call to install and enable a performance counter + */ +static void __perf_install_in_context(void *info) +{ + struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + struct perf_counter *counter = info; + struct perf_counter_context *ctx = counter->ctx; + struct perf_counter *leader = counter->group_leader; + int cpu = smp_processor_id(); + unsigned long flags; + u64 perf_flags; + int err; + + /* + * 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. + */ + if (ctx->task && cpuctx->task_ctx != ctx) + return; + + curr_rq_lock_irq_save(&flags); + spin_lock(&ctx->lock); + + /* + * Protect the list operation against NMI by disabling the + * counters on a global level. NOP for non NMI based counters. + */ + perf_flags = hw_perf_save_disable(); + + list_add_counter(counter, ctx); + ctx->nr_counters++; + + /* + * Don't put the counter on if it is disabled or if + * it is in a group and the group isn't on. + */ + if (counter->state != PERF_COUNTER_STATE_INACTIVE || + (leader != counter && leader->state != PERF_COUNTER_STATE_ACTIVE)) + goto unlock; + + /* + * An exclusive counter can't go on if there are already active + * hardware counters, and no hardware counter can go on if there + * is already an exclusive counter on. + */ + if (!group_can_go_on(counter, cpuctx, 1)) + err = -EEXIST; + else + err = counter_sched_in(counter, cpuctx, ctx, cpu); + + if (err) { + /* + * This counter couldn't go on. If it is in a group + * then we have to pull the whole group off. + * If the counter group is pinned then put it in error state. + */ + if (leader != counter) + group_sched_out(leader, cpuctx, ctx); + if (leader->hw_event.pinned) + leader->state = PERF_COUNTER_STATE_ERROR; + } + + if (!err && !ctx->task && cpuctx->max_pertask) + cpuctx->max_pertask--; + + unlock: + hw_perf_restore(perf_flags); + + spin_unlock(&ctx->lock); + curr_rq_unlock_irq_restore(&flags); +} + +/* + * Attach a performance counter to a context + * + * First we add the counter to the list with the hardware enable bit + * in counter->hw_config cleared. + * + * If the counter is attached to a task which is on a CPU we use a smp + * call to enable it in the task context. The task might have been + * scheduled away, but we check this in the smp call again. + * + * Must be called with ctx->mutex held. + */ +static void +perf_install_in_context(struct perf_counter_context *ctx, + struct perf_counter *counter, + int cpu) +{ + struct task_struct *task = ctx->task; + + counter->ctx = ctx; + if (!task) { + /* + * Per cpu counters are installed via an smp call and + * the install is always sucessful. + */ + smp_call_function_single(cpu, __perf_install_in_context, + counter, 1); + return; + } + + counter->task = task; +retry: + task_oncpu_function_call(task, __perf_install_in_context, + counter); + + spin_lock_irq(&ctx->lock); + /* + * we need to retry the smp call. + */ + if (ctx->is_active && list_empty(&counter->list_entry)) { + spin_unlock_irq(&ctx->lock); + goto retry; + } + + /* + * The lock prevents that this context is scheduled in so we + * can add the counter safely, if it the call above did not + * succeed. + */ + if (list_empty(&counter->list_entry)) { + list_add_counter(counter, ctx); + ctx->nr_counters++; + } + spin_unlock_irq(&ctx->lock); +} + +/* + * Cross CPU call to enable a performance counter + */ +static void __perf_counter_enable(void *info) +{ + struct perf_counter *counter = info; + struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + struct perf_counter_context *ctx = counter->ctx; + struct perf_counter *leader = counter->group_leader; + unsigned long flags; + int err; + + /* + * 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; + + curr_rq_lock_irq_save(&flags); + spin_lock(&ctx->lock); + + if (counter->state >= PERF_COUNTER_STATE_INACTIVE) + goto unlock; + counter->state = PERF_COUNTER_STATE_INACTIVE; + + /* + * If the counter is in a group and isn't the group leader, + * then don't put it on unless the group is on. + */ + if (leader != counter && leader->state != PERF_COUNTER_STATE_ACTIVE) + goto unlock; + + if (!group_can_go_on(counter, cpuctx, 1)) + err = -EEXIST; + else + err = counter_sched_in(counter, cpuctx, ctx, + smp_processor_id()); + + if (err) { + /* + * If this counter can't go on and it's part of a + * group, then the whole group has to come off. + */ + if (leader != counter) + group_sched_out(leader, cpuctx, ctx); + if (leader->hw_event.pinned) + leader->state = PERF_COUNTER_STATE_ERROR; + } + + unlock: + spin_unlock(&ctx->lock); + curr_rq_unlock_irq_restore(&flags); +} + +/* + * Enable a counter. + */ +static void perf_counter_enable(struct perf_counter *counter) +{ + struct perf_counter_context *ctx = counter->ctx; + struct task_struct *task = ctx->task; + + if (!task) { + /* + * Enable the counter on the cpu that it's on + */ + smp_call_function_single(counter->cpu, __perf_counter_enable, + counter, 1); + return; + } + + spin_lock_irq(&ctx->lock); + if (counter->state >= PERF_COUNTER_STATE_INACTIVE) + goto out; + + /* + * If the counter is in error state, clear that first. + * That way, if we see the counter in error state below, we + * know that it has gone back into error state, as distinct + * from the task having been scheduled away before the + * cross-call arrived. + */ + if (counter->state == PERF_COUNTER_STATE_ERROR) + counter->state = PERF_COUNTER_STATE_OFF; + + retry: + spin_unlock_irq(&ctx->lock); + task_oncpu_function_call(task, __perf_counter_enable, counter); + + spin_lock_irq(&ctx->lock); + + /* + * If the context is active and the counter is still off, + * we need to retry the cross-call. + */ + if (ctx->is_active && counter->state == PERF_COUNTER_STATE_OFF) + goto retry; + + /* + * Since we have the lock this context can't be scheduled + * in, so we can change the state safely. + */ + if (counter->state == PERF_COUNTER_STATE_OFF) + counter->state = PERF_COUNTER_STATE_INACTIVE; + out: + spin_unlock_irq(&ctx->lock); +} + +/* + * Enable a counter and all its children. + */ +static void perf_counter_enable_family(struct perf_counter *counter) +{ + struct perf_counter *child; + + perf_counter_enable(counter); + + /* + * Lock the mutex to protect the list of children + */ + mutex_lock(&counter->mutex); + list_for_each_entry(child, &counter->child_list, child_list) + perf_counter_enable(child); + mutex_unlock(&counter->mutex); +} + +void __perf_counter_sched_out(struct perf_counter_context *ctx, + struct perf_cpu_context *cpuctx) +{ + struct perf_counter *counter; + u64 flags; + + spin_lock(&ctx->lock); + ctx->is_active = 0; + if (likely(!ctx->nr_counters)) + goto out; + + flags = hw_perf_save_disable(); + if (ctx->nr_active) { + list_for_each_entry(counter, &ctx->counter_list, list_entry) + group_sched_out(counter, cpuctx, ctx); + } + hw_perf_restore(flags); + out: + spin_unlock(&ctx->lock); +} + +/* + * Called from scheduler to remove the counters of the current task, + * with interrupts disabled. + * + * We stop each counter and update the counter value in counter->count. + * + * This does not protect us against NMI, but disable() + * sets the disabled bit in the control field of counter _before_ + * accessing the counter control register. If a NMI hits, then it will + * not restart the counter. + */ +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; + + if (likely(!cpuctx->task_ctx)) + return; + + __perf_counter_sched_out(ctx, cpuctx); + + cpuctx->task_ctx = NULL; +} + +static void perf_counter_cpu_sched_out(struct perf_cpu_context *cpuctx) +{ + __perf_counter_sched_out(&cpuctx->ctx, cpuctx); +} + +static int +group_sched_in(struct perf_counter *group_counter, + struct perf_cpu_context *cpuctx, + struct perf_counter_context *ctx, + int cpu) +{ + struct perf_counter *counter, *partial_group; + int ret; + + if (group_counter->state == PERF_COUNTER_STATE_OFF) + return 0; + + ret = hw_perf_group_sched_in(group_counter, cpuctx, ctx, cpu); + if (ret) + return ret < 0 ? ret : 0; + + if (counter_sched_in(group_counter, cpuctx, ctx, cpu)) + return -EAGAIN; + + /* + * Schedule in siblings as one group (if any): + */ + list_for_each_entry(counter, &group_counter->sibling_list, list_entry) { + if (counter_sched_in(counter, cpuctx, ctx, cpu)) { + partial_group = counter; + goto group_error; + } + } + + return 0; + +group_error: + /* + * Groups can be scheduled in as one unit only, so undo any + * partial group before returning: + */ + list_for_each_entry(counter, &group_counter->sibling_list, list_entry) { + if (counter == partial_group) + break; + counter_sched_out(counter, cpuctx, ctx); + } + counter_sched_out(group_counter, cpuctx, ctx); + + return -EAGAIN; +} + +static void +__perf_counter_sched_in(struct perf_counter_context *ctx, + struct perf_cpu_context *cpuctx, int cpu) +{ + struct perf_counter *counter; + u64 flags; + int can_add_hw = 1; + + spin_lock(&ctx->lock); + ctx->is_active = 1; + if (likely(!ctx->nr_counters)) + goto out; + + flags = hw_perf_save_disable(); + + /* + * First go through the list and put on any pinned groups + * in order to give them the best chance of going on. + */ + list_for_each_entry(counter, &ctx->counter_list, list_entry) { + if (counter->state <= PERF_COUNTER_STATE_OFF || + !counter->hw_event.pinned) + continue; + if (counter->cpu != -1 && counter->cpu != cpu) + continue; + + if (group_can_go_on(counter, cpuctx, 1)) + group_sched_in(counter, cpuctx, ctx, cpu); + + /* + * If this pinned group hasn't been scheduled, + * put it in error state. + */ + if (counter->state == PERF_COUNTER_STATE_INACTIVE) + counter->state = PERF_COUNTER_STATE_ERROR; + } + + list_for_each_entry(counter, &ctx->counter_list, list_entry) { + /* + * Ignore counters in OFF or ERROR state, and + * ignore pinned counters since we did them already. + */ + if (counter->state <= PERF_COUNTER_STATE_OFF || + counter->hw_event.pinned) + continue; + + /* + * Listen to the 'cpu' scheduling filter constraint + * of counters: + */ + if (counter->cpu != -1 && counter->cpu != cpu) + continue; + + if (group_can_go_on(counter, cpuctx, can_add_hw)) { + if (group_sched_in(counter, cpuctx, ctx, cpu)) + can_add_hw = 0; + } + } + hw_perf_restore(flags); + out: + spin_unlock(&ctx->lock); +} + +/* + * Called from scheduler to add the counters of the current task + * with interrupts disabled. + * + * We restore the counter value and then enable it. + * + * This does not protect us against NMI, but enable() + * sets the enabled bit in the control field of counter _before_ + * accessing the counter control register. If a NMI hits, then it will + * keep the counter running. + */ +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; + + __perf_counter_sched_in(ctx, cpuctx, cpu); + cpuctx->task_ctx = ctx; +} + +static void perf_counter_cpu_sched_in(struct perf_cpu_context *cpuctx, int cpu) +{ + struct perf_counter_context *ctx = &cpuctx->ctx; + + __perf_counter_sched_in(ctx, cpuctx, cpu); +} + +int perf_counter_task_disable(void) +{ + struct task_struct *curr = current; + struct perf_counter_context *ctx = &curr->perf_counter_ctx; + struct perf_counter *counter; + unsigned long flags; + u64 perf_flags; + int cpu; + + if (likely(!ctx->nr_counters)) + return 0; + + curr_rq_lock_irq_save(&flags); + cpu = smp_processor_id(); + + /* force the update of the task clock: */ + __task_delta_exec(curr, 1); + + perf_counter_task_sched_out(curr, cpu); + + spin_lock(&ctx->lock); + + /* + * Disable all the counters: + */ + perf_flags = hw_perf_save_disable(); + + list_for_each_entry(counter, &ctx->counter_list, list_entry) { + if (counter->state != PERF_COUNTER_STATE_ERROR) + counter->state = PERF_COUNTER_STATE_OFF; + } + + hw_perf_restore(perf_flags); + + spin_unlock(&ctx->lock); + + curr_rq_unlock_irq_restore(&flags); + + return 0; +} + +int perf_counter_task_enable(void) +{ + struct task_struct *curr = current; + struct perf_counter_context *ctx = &curr->perf_counter_ctx; + struct perf_counter *counter; + unsigned long flags; + u64 perf_flags; + int cpu; + + if (likely(!ctx->nr_counters)) + return 0; + + curr_rq_lock_irq_save(&flags); + cpu = smp_processor_id(); + + /* force the update of the task clock: */ + __task_delta_exec(curr, 1); + + perf_counter_task_sched_out(curr, cpu); + + spin_lock(&ctx->lock); + + /* + * Disable all the counters: + */ + perf_flags = hw_perf_save_disable(); + + list_for_each_entry(counter, &ctx->counter_list, list_entry) { + if (counter->state > PERF_COUNTER_STATE_OFF) + continue; + counter->state = PERF_COUNTER_STATE_INACTIVE; + counter->hw_event.disabled = 0; + } + hw_perf_restore(perf_flags); + + spin_unlock(&ctx->lock); + + perf_counter_task_sched_in(curr, cpu); + + curr_rq_unlock_irq_restore(&flags); + + return 0; +} + +/* + * Round-robin a context's counters: + */ +static void rotate_ctx(struct perf_counter_context *ctx) +{ + struct perf_counter *counter; + u64 perf_flags; + + if (!ctx->nr_counters) + return; + + spin_lock(&ctx->lock); + /* + * Rotate the first entry last (works just fine for group counters too): + */ + perf_flags = hw_perf_save_disable(); + list_for_each_entry(counter, &ctx->counter_list, list_entry) { + list_del(&counter->list_entry); + list_add_tail(&counter->list_entry, &ctx->counter_list); + break; + } + hw_perf_restore(perf_flags); + + spin_unlock(&ctx->lock); +} + +void perf_counter_task_tick(struct task_struct *curr, int cpu) +{ + struct perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu); + struct perf_counter_context *ctx = &curr->perf_counter_ctx; + const int rotate_percpu = 0; + + if (rotate_percpu) + perf_counter_cpu_sched_out(cpuctx); + perf_counter_task_sched_out(curr, cpu); + + if (rotate_percpu) + rotate_ctx(&cpuctx->ctx); + rotate_ctx(ctx); + + if (rotate_percpu) + perf_counter_cpu_sched_in(cpuctx, cpu); + perf_counter_task_sched_in(curr, cpu); +} + +/* + * Cross CPU call to read the hardware counter + */ +static void __read(void *info) +{ + struct perf_counter *counter = info; + unsigned long flags; + + curr_rq_lock_irq_save(&flags); + counter->hw_ops->read(counter); + curr_rq_unlock_irq_restore(&flags); +} + +static u64 perf_counter_read(struct perf_counter *counter) +{ + /* + * If counter is enabled and currently active on a CPU, update the + * value in the counter structure: + */ + if (counter->state == PERF_COUNTER_STATE_ACTIVE) { + smp_call_function_single(counter->oncpu, + __read, counter, 1); + } + + return atomic64_read(&counter->count); +} + +/* + * Cross CPU call to switch performance data pointers + */ +static void __perf_switch_irq_data(void *info) +{ + struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + struct perf_counter *counter = info; + struct perf_counter_context *ctx = counter->ctx; + struct perf_data *oldirqdata = counter->irqdata; + + /* + * 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. + */ + if (ctx->task) { + if (cpuctx->task_ctx != ctx) + return; + spin_lock(&ctx->lock); + } + + /* Change the pointer NMI safe */ + atomic_long_set((atomic_long_t *)&counter->irqdata, + (unsigned long) counter->usrdata); + counter->usrdata = oldirqdata; + + if (ctx->task) + spin_unlock(&ctx->lock); +} + +static struct perf_data *perf_switch_irq_data(struct perf_counter *counter) +{ + struct perf_counter_context *ctx = counter->ctx; + struct perf_data *oldirqdata = counter->irqdata; + struct task_struct *task = ctx->task; + + if (!task) { + smp_call_function_single(counter->cpu, + __perf_switch_irq_data, + counter, 1); + return counter->usrdata; + } + +retry: + spin_lock_irq(&ctx->lock); + if (counter->state != PERF_COUNTER_STATE_ACTIVE) { + counter->irqdata = counter->usrdata; + counter->usrdata = oldirqdata; + spin_unlock_irq(&ctx->lock); + return oldirqdata; + } + spin_unlock_irq(&ctx->lock); + task_oncpu_function_call(task, __perf_switch_irq_data, counter); + /* Might have failed, because task was scheduled out */ + if (counter->irqdata == oldirqdata) + goto retry; + + return counter->usrdata; +} + +static void put_context(struct perf_counter_context *ctx) +{ + if (ctx->task) + put_task_struct(ctx->task); +} + +static struct perf_counter_context *find_get_context(pid_t pid, int cpu) +{ + struct perf_cpu_context *cpuctx; + struct perf_counter_context *ctx; + struct task_struct *task; + + /* + * If cpu is not a wildcard then this is a percpu counter: + */ + if (cpu != -1) { + /* Must be root to operate on a CPU counter: */ + if (!capable(CAP_SYS_ADMIN)) + return ERR_PTR(-EACCES); + + if (cpu < 0 || cpu > num_possible_cpus()) + return ERR_PTR(-EINVAL); + + /* + * We could be clever and allow to attach a counter to an + * offline CPU and activate it when the CPU comes up, but + * that's for later. + */ + if (!cpu_isset(cpu, cpu_online_map)) + return ERR_PTR(-ENODEV); + + cpuctx = &per_cpu(perf_cpu_context, cpu); + ctx = &cpuctx->ctx; + + return ctx; + } + + rcu_read_lock(); + if (!pid) + task = current; + else + task = find_task_by_vpid(pid); + if (task) + get_task_struct(task); + rcu_read_unlock(); + + 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); + return ERR_PTR(-EACCES); + } + + return ctx; +} + +/* + * Called when the last reference to the file is gone. + */ +static int perf_release(struct inode *inode, struct file *file) +{ + struct perf_counter *counter = file->private_data; + struct perf_counter_context *ctx = counter->ctx; + + file->private_data = NULL; + + mutex_lock(&ctx->mutex); + mutex_lock(&counter->mutex); + + perf_counter_remove_from_context(counter); + put_context(ctx); + + mutex_unlock(&counter->mutex); + mutex_unlock(&ctx->mutex); + + kfree(counter); + + return 0; +} + +/* + * Read the performance counter - simple non blocking version for now + */ +static ssize_t +perf_read_hw(struct perf_counter *counter, char __user *buf, size_t count) +{ + u64 cntval; + + if (count != sizeof(cntval)) + return -EINVAL; + + /* + * Return end-of-file for a read on a counter that is in + * error state (i.e. because it was pinned but it couldn't be + * scheduled on to the CPU at some point). + */ + if (counter->state == PERF_COUNTER_STATE_ERROR) + return 0; + + mutex_lock(&counter->mutex); + cntval = perf_counter_read(counter); + mutex_unlock(&counter->mutex); + + return put_user(cntval, (u64 __user *) buf) ? -EFAULT : sizeof(cntval); +} + +static ssize_t +perf_copy_usrdata(struct perf_data *usrdata, char __user *buf, size_t count) +{ + if (!usrdata->len) + return 0; + + count = min(count, (size_t)usrdata->len); + if (copy_to_user(buf, usrdata->data + usrdata->rd_idx, count)) + return -EFAULT; + + /* Adjust the counters */ + usrdata->len -= count; + if (!usrdata->len) + usrdata->rd_idx = 0; + else + usrdata->rd_idx += count; + + return count; +} + +static ssize_t +perf_read_irq_data(struct perf_counter *counter, + char __user *buf, + size_t count, + int nonblocking) +{ + struct perf_data *irqdata, *usrdata; + DECLARE_WAITQUEUE(wait, current); + ssize_t res, res2; + + irqdata = counter->irqdata; + usrdata = counter->usrdata; + + if (usrdata->len + irqdata->len >= count) + goto read_pending; + + if (nonblocking) + return -EAGAIN; + + spin_lock_irq(&counter->waitq.lock); + __add_wait_queue(&counter->waitq, &wait); + for (;;) { + set_current_state(TASK_INTERRUPTIBLE); + if (usrdata->len + irqdata->len >= count) + break; + + if (signal_pending(current)) + break; + + if (counter->state == PERF_COUNTER_STATE_ERROR) + break; + + spin_unlock_irq(&counter->waitq.lock); + schedule(); + spin_lock_irq(&counter->waitq.lock); + } + __remove_wait_queue(&counter->waitq, &wait); + __set_current_state(TASK_RUNNING); + spin_unlock_irq(&counter->waitq.lock); + + if (usrdata->len + irqdata->len < count && + counter->state != PERF_COUNTER_STATE_ERROR) + return -ERESTARTSYS; +read_pending: + mutex_lock(&counter->mutex); + + /* Drain pending data first: */ + res = perf_copy_usrdata(usrdata, buf, count); + if (res < 0 || res == count) + goto out; + + /* Switch irq buffer: */ + usrdata = perf_switch_irq_data(counter); + res2 = perf_copy_usrdata(usrdata, buf + res, count - res); + if (res2 < 0) { + if (!res) + res = -EFAULT; + } else { + res += res2; + } +out: + mutex_unlock(&counter->mutex); + + return res; +} + +static ssize_t +perf_read(struct file *file, char __user *buf, size_t count, loff_t *ppos) +{ + struct perf_counter *counter = file->private_data; + + switch (counter->hw_event.record_type) { + case PERF_RECORD_SIMPLE: + return perf_read_hw(counter, buf, count); + + case PERF_RECORD_IRQ: + case PERF_RECORD_GROUP: + return perf_read_irq_data(counter, buf, count, + file->f_flags & O_NONBLOCK); + } + return -EINVAL; +} + +static unsigned int perf_poll(struct file *file, poll_table *wait) +{ + struct perf_counter *counter = file->private_data; + unsigned int events = 0; + unsigned long flags; + + poll_wait(file, &counter->waitq, wait); + + spin_lock_irqsave(&counter->waitq.lock, flags); + if (counter->usrdata->len || counter->irqdata->len) + events |= POLLIN; + spin_unlock_irqrestore(&counter->waitq.lock, flags); + + return events; +} + +static long perf_ioctl(struct file *file, unsigned int cmd, unsigned long arg) +{ + struct perf_counter *counter = file->private_data; + int err = 0; + + switch (cmd) { + case PERF_COUNTER_IOC_ENABLE: + perf_counter_enable_family(counter); + break; + case PERF_COUNTER_IOC_DISABLE: + perf_counter_disable_family(counter); + break; + default: + err = -ENOTTY; + } + return err; +} + +static const struct file_operations perf_fops = { + .release = perf_release, + .read = perf_read, + .poll = perf_poll, + .unlocked_ioctl = perf_ioctl, + .compat_ioctl = perf_ioctl, +}; + +static int cpu_clock_perf_counter_enable(struct perf_counter *counter) +{ + int cpu = raw_smp_processor_id(); + + atomic64_set(&counter->hw.prev_count, cpu_clock(cpu)); + return 0; +} + +static void cpu_clock_perf_counter_update(struct perf_counter *counter) +{ + int cpu = raw_smp_processor_id(); + s64 prev; + u64 now; + + now = cpu_clock(cpu); + prev = atomic64_read(&counter->hw.prev_count); + atomic64_set(&counter->hw.prev_count, now); + atomic64_add(now - prev, &counter->count); +} + +static void cpu_clock_perf_counter_disable(struct perf_counter *counter) +{ + cpu_clock_perf_counter_update(counter); +} + +static void cpu_clock_perf_counter_read(struct perf_counter *counter) +{ + cpu_clock_perf_counter_update(counter); +} + +static const struct hw_perf_counter_ops perf_ops_cpu_clock = { + .enable = cpu_clock_perf_counter_enable, + .disable = cpu_clock_perf_counter_disable, + .read = cpu_clock_perf_counter_read, +}; + +/* + * Called from within the scheduler: + */ +static u64 task_clock_perf_counter_val(struct perf_counter *counter, int update) +{ + struct task_struct *curr = counter->task; + u64 delta; + + delta = __task_delta_exec(curr, update); + + return curr->se.sum_exec_runtime + delta; +} + +static void task_clock_perf_counter_update(struct perf_counter *counter, u64 now) +{ + u64 prev; + s64 delta; + + prev = atomic64_read(&counter->hw.prev_count); + + atomic64_set(&counter->hw.prev_count, now); + + delta = now - prev; + + atomic64_add(delta, &counter->count); +} + +static void task_clock_perf_counter_read(struct perf_counter *counter) +{ + u64 now = task_clock_perf_counter_val(counter, 1); + + task_clock_perf_counter_update(counter, now); +} + +static int task_clock_perf_counter_enable(struct perf_counter *counter) +{ + u64 now = task_clock_perf_counter_val(counter, 0); + + atomic64_set(&counter->hw.prev_count, now); + + return 0; +} + +static void task_clock_perf_counter_disable(struct perf_counter *counter) +{ + u64 now = task_clock_perf_counter_val(counter, 0); + + task_clock_perf_counter_update(counter, now); +} + +static const struct hw_perf_counter_ops perf_ops_task_clock = { + .enable = task_clock_perf_counter_enable, + .disable = task_clock_perf_counter_disable, + .read = task_clock_perf_counter_read, +}; + +static u64 get_page_faults(void) +{ + struct task_struct *curr = current; + + return curr->maj_flt + curr->min_flt; +} + +static void page_faults_perf_counter_update(struct perf_counter *counter) +{ + u64 prev, now; + s64 delta; + + prev = atomic64_read(&counter->hw.prev_count); + now = get_page_faults(); + + atomic64_set(&counter->hw.prev_count, now); + + delta = now - prev; + + atomic64_add(delta, &counter->count); +} + +static void page_faults_perf_counter_read(struct perf_counter *counter) +{ + page_faults_perf_counter_update(counter); +} + +static int page_faults_perf_counter_enable(struct perf_counter *counter) +{ + /* + * page-faults is a per-task value already, + * so we dont have to clear it on switch-in. + */ + + return 0; +} + +static void page_faults_perf_counter_disable(struct perf_counter *counter) +{ + page_faults_perf_counter_update(counter); +} + +static const struct hw_perf_counter_ops perf_ops_page_faults = { + .enable = page_faults_perf_counter_enable, + .disable = page_faults_perf_counter_disable, + .read = page_faults_perf_counter_read, +}; + +static u64 get_context_switches(void) +{ + struct task_struct *curr = current; + + return curr->nvcsw + curr->nivcsw; +} + +static void context_switches_perf_counter_update(struct perf_counter *counter) +{ + u64 prev, now; + s64 delta; + + prev = atomic64_read(&counter->hw.prev_count); + now = get_context_switches(); + + atomic64_set(&counter->hw.prev_count, now); + + delta = now - prev; + + atomic64_add(delta, &counter->count); +} + +static void context_switches_perf_counter_read(struct perf_counter *counter) +{ + context_switches_perf_counter_update(counter); +} + +static int context_switches_perf_counter_enable(struct perf_counter *counter) +{ + /* + * ->nvcsw + curr->nivcsw is a per-task value already, + * so we dont have to clear it on switch-in. + */ + + return 0; +} + +static void context_switches_perf_counter_disable(struct perf_counter *counter) +{ + context_switches_perf_counter_update(counter); +} + +static const struct hw_perf_counter_ops perf_ops_context_switches = { + .enable = context_switches_perf_counter_enable, + .disable = context_switches_perf_counter_disable, + .read = context_switches_perf_counter_read, +}; + +static inline u64 get_cpu_migrations(void) +{ + return current->se.nr_migrations; +} + +static void cpu_migrations_perf_counter_update(struct perf_counter *counter) +{ + u64 prev, now; + s64 delta; + + prev = atomic64_read(&counter->hw.prev_count); + now = get_cpu_migrations(); + + atomic64_set(&counter->hw.prev_count, now); + + delta = now - prev; + + atomic64_add(delta, &counter->count); +} + +static void cpu_migrations_perf_counter_read(struct perf_counter *counter) +{ + cpu_migrations_perf_counter_update(counter); +} + +static int cpu_migrations_perf_counter_enable(struct perf_counter *counter) +{ + /* + * se.nr_migrations is a per-task value already, + * so we dont have to clear it on switch-in. + */ + + return 0; +} + +static void cpu_migrations_perf_counter_disable(struct perf_counter *counter) +{ + cpu_migrations_perf_counter_update(counter); +} + +static const struct hw_perf_counter_ops perf_ops_cpu_migrations = { + .enable = cpu_migrations_perf_counter_enable, + .disable = cpu_migrations_perf_counter_disable, + .read = cpu_migrations_perf_counter_read, +}; + +static const struct hw_perf_counter_ops * +sw_perf_counter_init(struct perf_counter *counter) +{ + const struct hw_perf_counter_ops *hw_ops = NULL; + + switch (counter->hw_event.type) { + case PERF_COUNT_CPU_CLOCK: + hw_ops = &perf_ops_cpu_clock; + break; + case PERF_COUNT_TASK_CLOCK: + hw_ops = &perf_ops_task_clock; + break; + case PERF_COUNT_PAGE_FAULTS: + hw_ops = &perf_ops_page_faults; + break; + case PERF_COUNT_CONTEXT_SWITCHES: + hw_ops = &perf_ops_context_switches; + break; + case PERF_COUNT_CPU_MIGRATIONS: + hw_ops = &perf_ops_cpu_migrations; + break; + default: + break; + } + return hw_ops; +} + +/* + * Allocate and initialize a counter structure + */ +static struct perf_counter * +perf_counter_alloc(struct perf_counter_hw_event *hw_event, + int cpu, + struct perf_counter *group_leader, + gfp_t gfpflags) +{ + const struct hw_perf_counter_ops *hw_ops; + struct perf_counter *counter; + + counter = kzalloc(sizeof(*counter), gfpflags); + if (!counter) + return NULL; + + /* + * Single counters are their own group leaders, with an + * empty sibling list: + */ + if (!group_leader) + group_leader = counter; + + mutex_init(&counter->mutex); + INIT_LIST_HEAD(&counter->list_entry); + INIT_LIST_HEAD(&counter->sibling_list); + init_waitqueue_head(&counter->waitq); + + INIT_LIST_HEAD(&counter->child_list); + + counter->irqdata = &counter->data[0]; + counter->usrdata = &counter->data[1]; + counter->cpu = cpu; + counter->hw_event = *hw_event; + counter->wakeup_pending = 0; + counter->group_leader = group_leader; + counter->hw_ops = NULL; + + counter->state = PERF_COUNTER_STATE_INACTIVE; + if (hw_event->disabled) + counter->state = PERF_COUNTER_STATE_OFF; + + hw_ops = NULL; + if (!hw_event->raw && hw_event->type < 0) + hw_ops = sw_perf_counter_init(counter); + if (!hw_ops) + hw_ops = hw_perf_counter_init(counter); + + if (!hw_ops) { + kfree(counter); + return NULL; + } + counter->hw_ops = hw_ops; + + return counter; +} + +/** + * sys_perf_task_open - open a performance counter, associate it to a task/cpu + * + * @hw_event_uptr: event type attributes for monitoring/sampling + * @pid: target pid + * @cpu: target cpu + * @group_fd: group leader counter fd + */ +asmlinkage int +sys_perf_counter_open(struct perf_counter_hw_event *hw_event_uptr __user, + pid_t pid, int cpu, int group_fd) +{ + struct perf_counter *counter, *group_leader; + struct perf_counter_hw_event hw_event; + struct perf_counter_context *ctx; + struct file *counter_file = NULL; + struct file *group_file = NULL; + int fput_needed = 0; + int fput_needed2 = 0; + int ret; + + if (copy_from_user(&hw_event, hw_event_uptr, sizeof(hw_event)) != 0) + return -EFAULT; + + /* + * Get the target context (task or percpu): + */ + ctx = find_get_context(pid, cpu); + if (IS_ERR(ctx)) + return PTR_ERR(ctx); + + /* + * Look up the group leader (we will attach this counter to it): + */ + group_leader = NULL; + if (group_fd != -1) { + ret = -EINVAL; + group_file = fget_light(group_fd, &fput_needed); + if (!group_file) + goto err_put_context; + if (group_file->f_op != &perf_fops) + goto err_put_context; + + group_leader = group_file->private_data; + /* + * Do not allow a recursive hierarchy (this new sibling + * becoming part of another group-sibling): + */ + if (group_leader->group_leader != group_leader) + goto err_put_context; + /* + * Do not allow to attach to a group in a different + * task or CPU context: + */ + if (group_leader->ctx != ctx) + goto err_put_context; + /* + * Only a group leader can be exclusive or pinned + */ + if (hw_event.exclusive || hw_event.pinned) + goto err_put_context; + } + + ret = -EINVAL; + counter = perf_counter_alloc(&hw_event, cpu, group_leader, GFP_KERNEL); + if (!counter) + goto err_put_context; + + ret = anon_inode_getfd("[perf_counter]", &perf_fops, counter, 0); + if (ret < 0) + goto err_free_put_context; + + counter_file = fget_light(ret, &fput_needed2); + if (!counter_file) + goto err_free_put_context; + + counter->filp = counter_file; + mutex_lock(&ctx->mutex); + perf_install_in_context(ctx, counter, cpu); + mutex_unlock(&ctx->mutex); + + fput_light(counter_file, fput_needed2); + +out_fput: + fput_light(group_file, fput_needed); + + return ret; + +err_free_put_context: + kfree(counter); + +err_put_context: + put_context(ctx); + + 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); + ctx->task = task; +} + +/* + * inherit a counter from parent task to child task: + */ +static struct perf_counter * +inherit_counter(struct perf_counter *parent_counter, + struct task_struct *parent, + struct perf_counter_context *parent_ctx, + struct task_struct *child, + struct perf_counter *group_leader, + struct perf_counter_context *child_ctx) +{ + struct perf_counter *child_counter; + + /* + * Instead of creating recursive hierarchies of counters, + * we link inherited counters back to the original parent, + * which has a filp for sure, which we use as the reference + * count: + */ + if (parent_counter->parent) + parent_counter = parent_counter->parent; + + child_counter = perf_counter_alloc(&parent_counter->hw_event, + parent_counter->cpu, group_leader, + GFP_KERNEL); + if (!child_counter) + return NULL; + + /* + * Link it up in the child's context: + */ + child_counter->ctx = child_ctx; + child_counter->task = child; + list_add_counter(child_counter, child_ctx); + child_ctx->nr_counters++; + + child_counter->parent = parent_counter; + /* + * inherit into child's child as well: + */ + child_counter->hw_event.inherit = 1; + + /* + * Get a reference to the parent filp - we will fput it + * when the child counter exits. This is safe to do because + * we are in the parent and we know that the filp still + * exists and has a nonzero count: + */ + atomic_long_inc(&parent_counter->filp->f_count); + + /* + * Link this into the parent counter's child list + */ + mutex_lock(&parent_counter->mutex); + list_add_tail(&child_counter->child_list, &parent_counter->child_list); + + /* + * Make the child state follow the state of the parent counter, + * not its hw_event.disabled bit. We hold the parent's mutex, + * so we won't race with perf_counter_{en,dis}able_family. + */ + if (parent_counter->state >= PERF_COUNTER_STATE_INACTIVE) + child_counter->state = PERF_COUNTER_STATE_INACTIVE; + else + child_counter->state = PERF_COUNTER_STATE_OFF; + + mutex_unlock(&parent_counter->mutex); + + return child_counter; +} + +static int inherit_group(struct perf_counter *parent_counter, + struct task_struct *parent, + struct perf_counter_context *parent_ctx, + struct task_struct *child, + struct perf_counter_context *child_ctx) +{ + struct perf_counter *leader; + struct perf_counter *sub; + + leader = inherit_counter(parent_counter, parent, parent_ctx, + child, NULL, child_ctx); + if (!leader) + return -ENOMEM; + list_for_each_entry(sub, &parent_counter->sibling_list, list_entry) { + if (!inherit_counter(sub, parent, parent_ctx, + child, leader, child_ctx)) + return -ENOMEM; + } + return 0; +} + +static void sync_child_counter(struct perf_counter *child_counter, + struct perf_counter *parent_counter) +{ + u64 parent_val, child_val; + + parent_val = atomic64_read(&parent_counter->count); + child_val = atomic64_read(&child_counter->count); + + /* + * Add back the child's count to the parent's count: + */ + atomic64_add(child_val, &parent_counter->count); + + /* + * Remove this counter from the parent's list + */ + mutex_lock(&parent_counter->mutex); + list_del_init(&child_counter->child_list); + mutex_unlock(&parent_counter->mutex); + + /* + * Release the parent counter, if this was the last + * reference to it. + */ + fput(parent_counter->filp); +} + +static void +__perf_counter_exit_task(struct task_struct *child, + struct perf_counter *child_counter, + struct perf_counter_context *child_ctx) +{ + struct perf_counter *parent_counter; + struct perf_counter *sub, *tmp; + + /* + * 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) + */ + if (child != current) { + wait_task_inactive(child, 0); + list_del_init(&child_counter->list_entry); + } else { + struct perf_cpu_context *cpuctx; + unsigned long flags; + u64 perf_flags; + + /* + * Disable and unlink this counter. + * + * Be careful about zapping the list - IRQ/NMI context + * could still be processing it: + */ + curr_rq_lock_irq_save(&flags); + perf_flags = hw_perf_save_disable(); + + cpuctx = &__get_cpu_var(perf_cpu_context); + + group_sched_out(child_counter, cpuctx, child_ctx); + + list_del_init(&child_counter->list_entry); + + child_ctx->nr_counters--; + + hw_perf_restore(perf_flags); + curr_rq_unlock_irq_restore(&flags); + } + + parent_counter = child_counter->parent; + /* + * It can happen that parent exits first, and has counters + * that are still around due to the child reference. These + * counters need to be zapped - but otherwise linger. + */ + if (parent_counter) { + sync_child_counter(child_counter, parent_counter); + list_for_each_entry_safe(sub, tmp, &child_counter->sibling_list, + list_entry) { + if (sub->parent) + sync_child_counter(sub, sub->parent); + kfree(sub); + } + } + + kfree(child_counter); +} + +/* + * When a child task exits, feed back counter values to parent counters. + * + * Note: we may be running in child context, but the PID is not hashed + * anymore so new counters will not be added. + */ +void perf_counter_exit_task(struct task_struct *child) +{ + struct perf_counter *child_counter, *tmp; + struct perf_counter_context *child_ctx; + + child_ctx = &child->perf_counter_ctx; + + if (likely(!child_ctx->nr_counters)) + return; + + list_for_each_entry_safe(child_counter, tmp, &child_ctx->counter_list, + list_entry) + __perf_counter_exit_task(child, child_counter, child_ctx); +} + +/* + * Initialize the perf_counter context in task_struct + */ +void perf_counter_init_task(struct task_struct *child) +{ + struct perf_counter_context *child_ctx, *parent_ctx; + 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); + + /* + * This is executed from the parent task context, so inherit + * counters that have been marked for cloning: + */ + + if (likely(!parent_ctx->nr_counters)) + return; + + /* + * Lock the parent list. No need to lock the child - not PID + * hashed yet and not running, so nobody can access it. + */ + mutex_lock(&parent_ctx->mutex); + + /* + * We dont have to disable NMIs - we are only looking at + * the list, not manipulating it: + */ + list_for_each_entry(counter, &parent_ctx->counter_list, list_entry) { + if (!counter->hw_event.inherit) + continue; + + if (inherit_group(counter, parent, + parent_ctx, child, child_ctx)) + break; + } + + mutex_unlock(&parent_ctx->mutex); +} + +static void __cpuinit perf_counter_init_cpu(int cpu) +{ + struct perf_cpu_context *cpuctx; + + cpuctx = &per_cpu(perf_cpu_context, cpu); + __perf_counter_init_context(&cpuctx->ctx, NULL); + + mutex_lock(&perf_resource_mutex); + cpuctx->max_pertask = perf_max_counters - perf_reserved_percpu; + mutex_unlock(&perf_resource_mutex); + + hw_perf_counter_setup(cpu); +} + +#ifdef CONFIG_HOTPLUG_CPU +static void __perf_counter_exit_cpu(void *info) +{ + struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context); + struct perf_counter_context *ctx = &cpuctx->ctx; + struct perf_counter *counter, *tmp; + + list_for_each_entry_safe(counter, tmp, &ctx->counter_list, list_entry) + __perf_counter_remove_from_context(counter); +} +static void perf_counter_exit_cpu(int cpu) +{ + struct perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu); + struct perf_counter_context *ctx = &cpuctx->ctx; + + mutex_lock(&ctx->mutex); + smp_call_function_single(cpu, __perf_counter_exit_cpu, NULL, 1); + mutex_unlock(&ctx->mutex); +} +#else +static inline void perf_counter_exit_cpu(int cpu) { } +#endif + +static int __cpuinit +perf_cpu_notify(struct notifier_block *self, unsigned long action, void *hcpu) +{ + unsigned int cpu = (long)hcpu; + + switch (action) { + + case CPU_UP_PREPARE: + case CPU_UP_PREPARE_FROZEN: + perf_counter_init_cpu(cpu); + break; + + case CPU_DOWN_PREPARE: + case CPU_DOWN_PREPARE_FROZEN: + perf_counter_exit_cpu(cpu); + break; + + default: + break; + } + + return NOTIFY_OK; +} + +static struct notifier_block __cpuinitdata perf_cpu_nb = { + .notifier_call = perf_cpu_notify, +}; + +static int __init perf_counter_init(void) +{ + perf_cpu_notify(&perf_cpu_nb, (unsigned long)CPU_UP_PREPARE, + (void *)(long)smp_processor_id()); + register_cpu_notifier(&perf_cpu_nb); + + return 0; +} +early_initcall(perf_counter_init); + +static ssize_t perf_show_reserve_percpu(struct sysdev_class *class, char *buf) +{ + return sprintf(buf, "%d\n", perf_reserved_percpu); +} + +static ssize_t +perf_set_reserve_percpu(struct sysdev_class *class, + const char *buf, + size_t count) +{ + struct perf_cpu_context *cpuctx; + unsigned long val; + int err, cpu, mpt; + + err = strict_strtoul(buf, 10, &val); + if (err) + return err; + if (val > perf_max_counters) + return -EINVAL; + + mutex_lock(&perf_resource_mutex); + perf_reserved_percpu = val; + for_each_online_cpu(cpu) { + cpuctx = &per_cpu(perf_cpu_context, cpu); + spin_lock_irq(&cpuctx->ctx.lock); + mpt = min(perf_max_counters - cpuctx->ctx.nr_counters, + perf_max_counters - perf_reserved_percpu); + cpuctx->max_pertask = mpt; + spin_unlock_irq(&cpuctx->ctx.lock); + } + mutex_unlock(&perf_resource_mutex); + + return count; +} + +static ssize_t perf_show_overcommit(struct sysdev_class *class, char *buf) +{ + return sprintf(buf, "%d\n", perf_overcommit); +} + +static ssize_t +perf_set_overcommit(struct sysdev_class *class, const char *buf, size_t count) +{ + unsigned long val; + int err; + + err = strict_strtoul(buf, 10, &val); + if (err) + return err; + if (val > 1) + return -EINVAL; + + mutex_lock(&perf_resource_mutex); + perf_overcommit = val; + mutex_unlock(&perf_resource_mutex); + + return count; +} + +static SYSDEV_CLASS_ATTR( + reserve_percpu, + 0644, + perf_show_reserve_percpu, + perf_set_reserve_percpu + ); + +static SYSDEV_CLASS_ATTR( + overcommit, + 0644, + perf_show_overcommit, + perf_set_overcommit + ); + +static struct attribute *perfclass_attrs[] = { + &attr_reserve_percpu.attr, + &attr_overcommit.attr, + NULL +}; + +static struct attribute_group perfclass_attr_group = { + .attrs = perfclass_attrs, + .name = "perf_counters", +}; + +static int __init perf_counter_sysfs_init(void) +{ + return sysfs_create_group(&cpu_sysdev_class.kset.kobj, + &perfclass_attr_group); +} +device_initcall(perf_counter_sysfs_init); diff --git a/kernel/sched.c b/kernel/sched.c index 52bbf1c842a..40d70d9c0af 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -668,7 +668,7 @@ static inline int cpu_of(struct rq *rq) #define task_rq(p) cpu_rq(task_cpu(p)) #define cpu_curr(cpu) (cpu_rq(cpu)->curr) -static inline void update_rq_clock(struct rq *rq) +inline void update_rq_clock(struct rq *rq) { rq->clock = sched_clock_cpu(cpu_of(rq)); } @@ -979,6 +979,26 @@ static struct rq *task_rq_lock(struct task_struct *p, unsigned long *flags) } } +void curr_rq_lock_irq_save(unsigned long *flags) + __acquires(rq->lock) +{ + struct rq *rq; + + local_irq_save(*flags); + rq = cpu_rq(smp_processor_id()); + spin_lock(&rq->lock); +} + +void curr_rq_unlock_irq_restore(unsigned long *flags) + __releases(rq->lock) +{ + struct rq *rq; + + rq = cpu_rq(smp_processor_id()); + spin_unlock(&rq->lock); + local_irq_restore(*flags); +} + void task_rq_unlock_wait(struct task_struct *p) { struct rq *rq = task_rq(p); @@ -1885,12 +1905,14 @@ void set_task_cpu(struct task_struct *p, unsigned int new_cpu) p->se.sleep_start -= clock_offset; if (p->se.block_start) p->se.block_start -= clock_offset; +#endif if (old_cpu != new_cpu) { - schedstat_inc(p, se.nr_migrations); + p->se.nr_migrations++; +#ifdef CONFIG_SCHEDSTATS if (task_hot(p, old_rq->clock, NULL)) schedstat_inc(p, se.nr_forced2_migrations); - } #endif + } p->se.vruntime -= old_cfsrq->min_vruntime - new_cfsrq->min_vruntime; @@ -2242,6 +2264,27 @@ static int sched_balance_self(int cpu, int flag) #endif /* CONFIG_SMP */ +/** + * task_oncpu_function_call - call a function on the cpu on which a task runs + * @p: the task to evaluate + * @func: the function to be called + * @info: the function call argument + * + * Calls the function @func when the task is currently running. This might + * be on the current CPU, which just calls the function directly + */ +void task_oncpu_function_call(struct task_struct *p, + void (*func) (void *info), void *info) +{ + int cpu; + + preempt_disable(); + cpu = task_cpu(p); + if (task_curr(p)) + smp_call_function_single(cpu, func, info, 1); + preempt_enable(); +} + /*** * try_to_wake_up - wake up a thread * @p: the to-be-woken-up thread @@ -2384,6 +2427,7 @@ static void __sched_fork(struct task_struct *p) p->se.exec_start = 0; p->se.sum_exec_runtime = 0; p->se.prev_sum_exec_runtime = 0; + p->se.nr_migrations = 0; p->se.last_wakeup = 0; p->se.avg_overlap = 0; @@ -2604,6 +2648,7 @@ static void finish_task_switch(struct rq *rq, struct task_struct *prev) */ prev_state = prev->state; finish_arch_switch(prev); + perf_counter_task_sched_in(current, cpu_of(rq)); finish_lock_switch(rq, prev); #ifdef CONFIG_SMP if (current->sched_class->post_schedule) @@ -4132,6 +4177,29 @@ EXPORT_PER_CPU_SYMBOL(kstat); * Return any ns on the sched_clock that have not yet been banked in * @p in case that task is currently running. */ +unsigned long long __task_delta_exec(struct task_struct *p, int update) +{ + s64 delta_exec; + struct rq *rq; + + rq = task_rq(p); + WARN_ON_ONCE(!runqueue_is_locked()); + WARN_ON_ONCE(!task_current(rq, p)); + + if (update) + update_rq_clock(rq); + + delta_exec = rq->clock - p->se.exec_start; + + WARN_ON_ONCE(delta_exec < 0); + + return delta_exec; +} + +/* + * Return any ns on the sched_clock that have not yet been banked in + * @p in case that task is currently running. + */ unsigned long long task_delta_exec(struct task_struct *p) { unsigned long flags; @@ -4391,6 +4459,7 @@ void scheduler_tick(void) update_rq_clock(rq); update_cpu_load(rq); curr->sched_class->task_tick(rq, curr, 0); + perf_counter_task_tick(curr, cpu); spin_unlock(&rq->lock); #ifdef CONFIG_SMP @@ -4586,6 +4655,7 @@ need_resched_nonpreemptible: if (likely(prev != next)) { sched_info_switch(prev, next); + perf_counter_task_sched_out(prev, cpu); rq->nr_switches++; rq->curr = next; diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index 954e1a81b79..da932f4c852 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -960,16 +960,17 @@ static struct task_struct *pick_next_highest_task_rt(struct rq *rq, int cpu) static DEFINE_PER_CPU(cpumask_var_t, local_cpu_mask); -static inline int pick_optimal_cpu(int this_cpu, cpumask_t *mask) +static inline int pick_optimal_cpu(int this_cpu, + const struct cpumask *mask) { int first; /* "this_cpu" is cheaper to preempt than a remote processor */ - if ((this_cpu != -1) && cpu_isset(this_cpu, *mask)) + if ((this_cpu != -1) && cpumask_test_cpu(this_cpu, mask)) return this_cpu; - first = first_cpu(*mask); - if (first != NR_CPUS) + first = cpumask_first(mask); + if (first < nr_cpu_ids) return first; return -1; @@ -981,6 +982,7 @@ static int find_lowest_rq(struct task_struct *task) struct cpumask *lowest_mask = __get_cpu_var(local_cpu_mask); int this_cpu = smp_processor_id(); int cpu = task_cpu(task); + cpumask_var_t domain_mask; if (task->rt.nr_cpus_allowed == 1) return -1; /* No other targets possible */ @@ -1013,19 +1015,25 @@ static int find_lowest_rq(struct task_struct *task) if (this_cpu == cpu) this_cpu = -1; /* Skip this_cpu opt if the same */ - for_each_domain(cpu, sd) { - if (sd->flags & SD_WAKE_AFFINE) { - cpumask_t domain_mask; - int best_cpu; + if (alloc_cpumask_var(&domain_mask, GFP_ATOMIC)) { + for_each_domain(cpu, sd) { + if (sd->flags & SD_WAKE_AFFINE) { + int best_cpu; - cpumask_and(&domain_mask, sched_domain_span(sd), - lowest_mask); + cpumask_and(domain_mask, + sched_domain_span(sd), + lowest_mask); - best_cpu = pick_optimal_cpu(this_cpu, - &domain_mask); - if (best_cpu != -1) - return best_cpu; + best_cpu = pick_optimal_cpu(this_cpu, + domain_mask); + + if (best_cpu != -1) { + free_cpumask_var(domain_mask); + return best_cpu; + } + } } + free_cpumask_var(domain_mask); } /* diff --git a/kernel/softirq.c b/kernel/softirq.c index bdbe9de9cd8..0365b4899a3 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -795,6 +795,11 @@ int __init __weak early_irq_init(void) return 0; } +int __init __weak arch_probe_nr_irqs(void) +{ + return 0; +} + int __init __weak arch_early_irq_init(void) { return 0; diff --git a/kernel/sys.c b/kernel/sys.c index e7dc0e10a48..87ca037dc03 100644 --- a/kernel/sys.c +++ b/kernel/sys.c @@ -14,6 +14,7 @@ #include <linux/prctl.h> #include <linux/highuid.h> #include <linux/fs.h> +#include <linux/perf_counter.h> #include <linux/resource.h> #include <linux/kernel.h> #include <linux/kexec.h> @@ -1799,6 +1800,12 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3, case PR_SET_TSC: error = SET_TSC_CTL(arg2); break; + case PR_TASK_PERF_COUNTERS_DISABLE: + error = perf_counter_task_disable(); + break; + case PR_TASK_PERF_COUNTERS_ENABLE: + error = perf_counter_task_enable(); + break; case PR_GET_TIMERSLACK: error = current->timer_slack_ns; break; diff --git a/kernel/sys_ni.c b/kernel/sys_ni.c index 27dad296738..68320f6b07b 100644 --- a/kernel/sys_ni.c +++ b/kernel/sys_ni.c @@ -175,3 +175,6 @@ cond_syscall(compat_sys_timerfd_settime); cond_syscall(compat_sys_timerfd_gettime); cond_syscall(sys_eventfd); cond_syscall(sys_eventfd2); + +/* performance counters: */ +cond_syscall(sys_perf_counter_open); diff --git a/lib/smp_processor_id.c b/lib/smp_processor_id.c index 0f8fc22ed10..4689cb073da 100644 --- a/lib/smp_processor_id.c +++ b/lib/smp_processor_id.c @@ -22,7 +22,7 @@ notrace unsigned int debug_smp_processor_id(void) * Kernel threads bound to a single CPU can safely use * smp_processor_id(): */ - if (cpus_equal(current->cpus_allowed, cpumask_of_cpu(this_cpu))) + if (cpumask_equal(¤t->cpus_allowed, cpumask_of(this_cpu))) goto out; /* |