From 5e219b3c671b34b2d79468fe89c44c0460c0f02b Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Wed, 1 May 2013 17:25:41 +0200 Subject: x86/signals: Propagate RF EFLAGS bit through the signal restore call While porting Vince's perf overflow tests I found perf event breakpoint overflow does not work properly. I found the x86 RF EFLAG bit not being set when returning from debug exception after triggering signal handler. Which is exactly what you get when you set perf breakpoint overflow SIGIO handler. This patch and the next two patches fix the underlying bugs. This patch adds the RF EFLAGS bit to be restored on return from signal from the original register context before the signal was entered. This will prevent the RF flag to disappear when returning from exception due to the signal handler being executed. Signed-off-by: Jiri Olsa Tested-by: Oleg Nesterov Reviewed-by: Frederic Weisbecker Originally-Reported-by: Vince Weaver Cc: H. Peter Anvin Cc: Andi Kleen Cc: Oleg Nesterov Cc: Arnaldo Carvalho de Melo Cc: Ingo Molnar Cc: Paul Mackerras Cc: Corey Ashford Cc: Frederic Weisbecker Cc: Vince Weaver Cc: Stephane Eranian Cc: Linus Torvalds Cc: Andrew Morton Cc: Peter Zijlstra Cc: Thomas Gleixner Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/r/1367421944-19082-2-git-send-email-jolsa@redhat.com Signed-off-by: Ingo Molnar --- arch/x86/ia32/ia32_signal.c | 2 -- arch/x86/include/asm/sighandling.h | 4 ++-- arch/x86/kernel/signal.c | 6 ------ 3 files changed, 2 insertions(+), 10 deletions(-) diff --git a/arch/x86/ia32/ia32_signal.c b/arch/x86/ia32/ia32_signal.c index cf1a471a18a2..bccfca68430e 100644 --- a/arch/x86/ia32/ia32_signal.c +++ b/arch/x86/ia32/ia32_signal.c @@ -34,8 +34,6 @@ #include #include -#define FIX_EFLAGS __FIX_EFLAGS - int copy_siginfo_to_user32(compat_siginfo_t __user *to, siginfo_t *from) { int err = 0; diff --git a/arch/x86/include/asm/sighandling.h b/arch/x86/include/asm/sighandling.h index beff97f7df37..7a958164088c 100644 --- a/arch/x86/include/asm/sighandling.h +++ b/arch/x86/include/asm/sighandling.h @@ -7,10 +7,10 @@ #include -#define __FIX_EFLAGS (X86_EFLAGS_AC | X86_EFLAGS_OF | \ +#define FIX_EFLAGS (X86_EFLAGS_AC | X86_EFLAGS_OF | \ X86_EFLAGS_DF | X86_EFLAGS_TF | X86_EFLAGS_SF | \ X86_EFLAGS_ZF | X86_EFLAGS_AF | X86_EFLAGS_PF | \ - X86_EFLAGS_CF) + X86_EFLAGS_CF | X86_EFLAGS_RF) void signal_fault(struct pt_regs *regs, void __user *frame, char *where); diff --git a/arch/x86/kernel/signal.c b/arch/x86/kernel/signal.c index 69562992e457..9df4c0b5ecac 100644 --- a/arch/x86/kernel/signal.c +++ b/arch/x86/kernel/signal.c @@ -43,12 +43,6 @@ #include -#ifdef CONFIG_X86_32 -# define FIX_EFLAGS (__FIX_EFLAGS | X86_EFLAGS_RF) -#else -# define FIX_EFLAGS __FIX_EFLAGS -#endif - #define COPY(x) do { \ get_user_ex(regs->x, &sc->x); \ } while (0) -- cgit v1.2.2 From 24cda10996f5420ab962f91cd03c15869a3a94b1 Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Wed, 1 May 2013 17:25:42 +0200 Subject: x86/signals: Clear RF EFLAGS bit for signal handler Clearing RF EFLAGS bit for signal handler. The reason is that this flag is set by debug exception code to prevent the recursive exception entry. Leaving it set for signal handler might prevent debug exception of the signal handler itself. Signed-off-by: Jiri Olsa Tested-by: Oleg Nesterov Reviewed-by: Frederic Weisbecker Originally-Reported-by: Vince Weaver Cc: H. Peter Anvin Cc: Andi Kleen Cc: Oleg Nesterov Cc: Arnaldo Carvalho de Melo Cc: Ingo Molnar Cc: Paul Mackerras Cc: Corey Ashford Cc: Frederic Weisbecker Cc: Vince Weaver Cc: Stephane Eranian Cc: Linus Torvalds Cc: Andrew Morton Cc: Peter Zijlstra Cc: Thomas Gleixner Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/r/1367421944-19082-3-git-send-email-jolsa@redhat.com Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/arch/x86/kernel/signal.c b/arch/x86/kernel/signal.c index 9df4c0b5ecac..cb12fc9c0642 100644 --- a/arch/x86/kernel/signal.c +++ b/arch/x86/kernel/signal.c @@ -664,6 +664,12 @@ handle_signal(struct ksignal *ksig, struct pt_regs *regs) * Clear the direction flag as per the ABI for function entry. */ regs->flags &= ~X86_EFLAGS_DF; + /* + * Clear RF when entering the signal handler, because + * it might disable possible debug exception from the + * signal handler. + */ + regs->flags &= ~X86_EFLAGS_RF; /* * Clear TF when entering the signal handler, but * notify any tracer that was single-stepping it. -- cgit v1.2.2 From ddd40da4ccbabdd2e941837aa987e08dfa4396b4 Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Wed, 1 May 2013 17:25:43 +0200 Subject: x86/signals: Merge EFLAGS bit clearing into a single statement Merging EFLAGS bit clearing into a single statement, to ensure EFLAGS bits are being cleared in a single instruction. Signed-off-by: Jiri Olsa Tested-by: Oleg Nesterov Reviewed-by: Frederic Weisbecker Originally-Reported-by: Vince Weaver Cc: H. Peter Anvin Cc: Andi Kleen Cc: Oleg Nesterov Cc: Arnaldo Carvalho de Melo Cc: Ingo Molnar Cc: Paul Mackerras Cc: Corey Ashford Cc: Frederic Weisbecker Cc: Vince Weaver Cc: Stephane Eranian Cc: Linus Torvalds Cc: Andrew Morton Cc: Peter Zijlstra Cc: Thomas Gleixner Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/r/1367421944-19082-4-git-send-email-jolsa@redhat.com Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal.c | 10 +++------- 1 file changed, 3 insertions(+), 7 deletions(-) diff --git a/arch/x86/kernel/signal.c b/arch/x86/kernel/signal.c index cb12fc9c0642..cf913587d4dd 100644 --- a/arch/x86/kernel/signal.c +++ b/arch/x86/kernel/signal.c @@ -662,21 +662,17 @@ handle_signal(struct ksignal *ksig, struct pt_regs *regs) if (!failed) { /* * Clear the direction flag as per the ABI for function entry. - */ - regs->flags &= ~X86_EFLAGS_DF; - /* + * * Clear RF when entering the signal handler, because * it might disable possible debug exception from the * signal handler. - */ - regs->flags &= ~X86_EFLAGS_RF; - /* + * * Clear TF when entering the signal handler, but * notify any tracer that was single-stepping it. * The tracer may want to single-step inside the * handler too. */ - regs->flags &= ~X86_EFLAGS_TF; + regs->flags &= ~(X86_EFLAGS_DF|X86_EFLAGS_RF|X86_EFLAGS_TF); } signal_setup_done(failed, ksig, test_thread_flag(TIF_SINGLESTEP)); } -- cgit v1.2.2 From ab573844e3058eef2788803d373019f8bebead57 Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Wed, 1 May 2013 17:25:44 +0200 Subject: perf: Fix hw breakpoints overflow period sampling The hw breakpoint pmu 'add' function is missing the period_left update needed for SW events. The perf HW breakpoint events use the SW events framework to process the overflow, so it needs to be properly initialized in the PMU 'add' method. Signed-off-by: Jiri Olsa Reviewed-by: Peter Zijlstra Cc: H. Peter Anvin Cc: Oleg Nesterov Cc: Arnaldo Carvalho de Melo Cc: Ingo Molnar Cc: Paul Mackerras Cc: Corey Ashford Cc: Frederic Weisbecker Cc: Vince Weaver Cc: Stephane Eranian Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/r/1367421944-19082-5-git-send-email-jolsa@redhat.com Signed-off-by: Ingo Molnar --- include/linux/perf_event.h | 2 ++ kernel/events/core.c | 2 +- kernel/events/hw_breakpoint.c | 5 +++++ 3 files changed, 8 insertions(+), 1 deletion(-) diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index f463a46424e2..fa38612d70b6 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -743,6 +743,7 @@ extern unsigned int perf_output_skip(struct perf_output_handle *handle, unsigned int len); extern int perf_swevent_get_recursion_context(void); extern void perf_swevent_put_recursion_context(int rctx); +extern u64 perf_swevent_set_period(struct perf_event *event); extern void perf_event_enable(struct perf_event *event); extern void perf_event_disable(struct perf_event *event); extern int __perf_event_disable(void *info); @@ -782,6 +783,7 @@ static inline void perf_event_fork(struct task_struct *tsk) { } static inline void perf_event_init(void) { } static inline int perf_swevent_get_recursion_context(void) { return -1; } static inline void perf_swevent_put_recursion_context(int rctx) { } +static inline u64 perf_swevent_set_period(struct perf_event *event) { return 0; } static inline void perf_event_enable(struct perf_event *event) { } static inline void perf_event_disable(struct perf_event *event) { } static inline int __perf_event_disable(void *info) { return -1; } diff --git a/kernel/events/core.c b/kernel/events/core.c index 9dc297faf7c0..e0dcced282e4 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -4961,7 +4961,7 @@ static DEFINE_PER_CPU(struct swevent_htable, swevent_htable); * sign as trigger. */ -static u64 perf_swevent_set_period(struct perf_event *event) +u64 perf_swevent_set_period(struct perf_event *event) { struct hw_perf_event *hwc = &event->hw; u64 period = hwc->last_period; diff --git a/kernel/events/hw_breakpoint.c b/kernel/events/hw_breakpoint.c index a64f8aeb5c1f..966a241e8616 100644 --- a/kernel/events/hw_breakpoint.c +++ b/kernel/events/hw_breakpoint.c @@ -612,6 +612,11 @@ static int hw_breakpoint_add(struct perf_event *bp, int flags) if (!(flags & PERF_EF_START)) bp->hw.state = PERF_HES_STOPPED; + if (is_sampling_event(bp)) { + bp->hw.last_period = bp->hw.sample_period; + perf_swevent_set_period(bp); + } + return arch_install_hw_breakpoint(bp); } -- cgit v1.2.2 From 9e6302056f8029f438e853432a856b9f13de26a6 Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Wed, 3 Apr 2013 14:21:33 +0200 Subject: perf: Use hrtimers for event multiplexing The current scheme of using the timer tick was fine for per-thread events. However, it was causing bias issues in system-wide mode (including for uncore PMUs). Event groups would not get their fair share of runtime on the PMU. With tickless kernels, if a core is idle there is no timer tick, and thus no event rotation (multiplexing). However, there are events (especially uncore events) which do count even though cores are asleep. This patch changes the timer source for multiplexing. It introduces a per-PMU per-cpu hrtimer. The advantage is that even when a core goes idle, it will come back to service the hrtimer, thus multiplexing on system-wide events works much better. The per-PMU implementation (suggested by PeterZ) enables adjusting the multiplexing interval per PMU. The preferred interval is stashed into the struct pmu. If not set, it will be forced to the default interval value. In order to minimize the impact of the hrtimer, it is turned on and off on demand. When the PMU on a CPU is overcommited, the hrtimer is activated. It is stopped when the PMU is not overcommitted. In order for this to work properly, we had to change the order of initialization in start_kernel() such that hrtimer_init() is run before perf_event_init(). The default interval in milliseconds is set to a timer tick just like with the old code. We will provide a sysctl to tune this in another patch. Signed-off-by: Stephane Eranian Signed-off-by: Peter Zijlstra Cc: Frederic Weisbecker Cc: Arnaldo Carvalho de Melo Link: http://lkml.kernel.org/r/1364991694-5876-2-git-send-email-eranian@google.com Signed-off-by: Ingo Molnar --- include/linux/perf_event.h | 3 +- init/main.c | 2 +- kernel/events/core.c | 114 +++++++++++++++++++++++++++++++++++++++++---- 3 files changed, 109 insertions(+), 10 deletions(-) diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index fa38612d70b6..72138d75a60a 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -501,8 +501,9 @@ struct perf_cpu_context { struct perf_event_context *task_ctx; int active_oncpu; int exclusive; + struct hrtimer hrtimer; + ktime_t hrtimer_interval; struct list_head rotation_list; - int jiffies_interval; struct pmu *unique_pmu; struct perf_cgroup *cgrp; }; diff --git a/init/main.c b/init/main.c index 9484f4ba88d0..ec549581d732 100644 --- a/init/main.c +++ b/init/main.c @@ -542,7 +542,6 @@ asmlinkage void __init start_kernel(void) if (WARN(!irqs_disabled(), "Interrupts were enabled *very* early, fixing it\n")) local_irq_disable(); idr_init_cache(); - perf_event_init(); rcu_init(); tick_nohz_init(); radix_tree_init(); @@ -555,6 +554,7 @@ asmlinkage void __init start_kernel(void) softirq_init(); timekeeping_init(); time_init(); + perf_event_init(); profile_init(); call_function_init(); WARN(!irqs_disabled(), "Interrupts were enabled early\n"); diff --git a/kernel/events/core.c b/kernel/events/core.c index e0dcced282e4..97bfac7e6f45 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -170,6 +170,8 @@ int sysctl_perf_event_sample_rate __read_mostly = DEFAULT_MAX_SAMPLE_RATE; static int max_samples_per_tick __read_mostly = DIV_ROUND_UP(DEFAULT_MAX_SAMPLE_RATE, HZ); +static int perf_rotate_context(struct perf_cpu_context *cpuctx); + int perf_proc_update_handler(struct ctl_table *table, int write, void __user *buffer, size_t *lenp, loff_t *ppos) @@ -658,6 +660,98 @@ perf_cgroup_mark_enabled(struct perf_event *event, } #endif +/* + * set default to be dependent on timer tick just + * like original code + */ +#define PERF_CPU_HRTIMER (1000 / HZ) +/* + * function must be called with interrupts disbled + */ +static enum hrtimer_restart perf_cpu_hrtimer_handler(struct hrtimer *hr) +{ + struct perf_cpu_context *cpuctx; + enum hrtimer_restart ret = HRTIMER_NORESTART; + int rotations = 0; + + WARN_ON(!irqs_disabled()); + + cpuctx = container_of(hr, struct perf_cpu_context, hrtimer); + + rotations = perf_rotate_context(cpuctx); + + /* + * arm timer if needed + */ + if (rotations) { + hrtimer_forward_now(hr, cpuctx->hrtimer_interval); + ret = HRTIMER_RESTART; + } + + return ret; +} + +/* CPU is going down */ +void perf_cpu_hrtimer_cancel(int cpu) +{ + struct perf_cpu_context *cpuctx; + struct pmu *pmu; + unsigned long flags; + + if (WARN_ON(cpu != smp_processor_id())) + return; + + local_irq_save(flags); + + rcu_read_lock(); + + list_for_each_entry_rcu(pmu, &pmus, entry) { + cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); + + if (pmu->task_ctx_nr == perf_sw_context) + continue; + + hrtimer_cancel(&cpuctx->hrtimer); + } + + rcu_read_unlock(); + + local_irq_restore(flags); +} + +static void __perf_cpu_hrtimer_init(struct perf_cpu_context *cpuctx, int cpu) +{ + struct hrtimer *hr = &cpuctx->hrtimer; + struct pmu *pmu = cpuctx->ctx.pmu; + + /* no multiplexing needed for SW PMU */ + if (pmu->task_ctx_nr == perf_sw_context) + return; + + cpuctx->hrtimer_interval = + ns_to_ktime(NSEC_PER_MSEC * PERF_CPU_HRTIMER); + + hrtimer_init(hr, CLOCK_MONOTONIC, HRTIMER_MODE_REL_PINNED); + hr->function = perf_cpu_hrtimer_handler; +} + +static void perf_cpu_hrtimer_restart(struct perf_cpu_context *cpuctx) +{ + struct hrtimer *hr = &cpuctx->hrtimer; + struct pmu *pmu = cpuctx->ctx.pmu; + + /* not for SW PMU */ + if (pmu->task_ctx_nr == perf_sw_context) + return; + + if (hrtimer_active(hr)) + return; + + if (!hrtimer_callback_running(hr)) + __hrtimer_start_range_ns(hr, cpuctx->hrtimer_interval, + 0, HRTIMER_MODE_REL_PINNED, 0); +} + void perf_pmu_disable(struct pmu *pmu) { int *count = this_cpu_ptr(pmu->pmu_disable_count); @@ -1506,6 +1600,7 @@ group_sched_in(struct perf_event *group_event, if (event_sched_in(group_event, cpuctx, ctx)) { pmu->cancel_txn(pmu); + perf_cpu_hrtimer_restart(cpuctx); return -EAGAIN; } @@ -1552,6 +1647,8 @@ group_error: pmu->cancel_txn(pmu); + perf_cpu_hrtimer_restart(cpuctx); + return -EAGAIN; } @@ -1807,8 +1904,10 @@ static int __perf_event_enable(void *info) * If this event can't go on and it's part of a * group, then the whole group has to come off. */ - if (leader != event) + if (leader != event) { group_sched_out(leader, cpuctx, ctx); + perf_cpu_hrtimer_restart(cpuctx); + } if (leader->attr.pinned) { update_group_times(leader); leader->state = PERF_EVENT_STATE_ERROR; @@ -2555,7 +2654,7 @@ static void rotate_ctx(struct perf_event_context *ctx) * because they're strictly cpu affine and rotate_start is called with IRQs * disabled, while rotate_context is called from IRQ context. */ -static void perf_rotate_context(struct perf_cpu_context *cpuctx) +static int perf_rotate_context(struct perf_cpu_context *cpuctx) { struct perf_event_context *ctx = NULL; int rotate = 0, remove = 1; @@ -2594,6 +2693,8 @@ static void perf_rotate_context(struct perf_cpu_context *cpuctx) done: if (remove) list_del_init(&cpuctx->rotation_list); + + return rotate; } #ifdef CONFIG_NO_HZ_FULL @@ -2625,10 +2726,6 @@ void perf_event_task_tick(void) ctx = cpuctx->task_ctx; if (ctx) perf_adjust_freq_unthr_context(ctx, throttled); - - if (cpuctx->jiffies_interval == 1 || - !(jiffies % cpuctx->jiffies_interval)) - perf_rotate_context(cpuctx); } } @@ -6001,7 +6098,9 @@ skip_type: lockdep_set_class(&cpuctx->ctx.lock, &cpuctx_lock); cpuctx->ctx.type = cpu_context; cpuctx->ctx.pmu = pmu; - cpuctx->jiffies_interval = 1; + + __perf_cpu_hrtimer_init(cpuctx, cpu); + INIT_LIST_HEAD(&cpuctx->rotation_list); cpuctx->unique_pmu = pmu; } @@ -7387,7 +7486,6 @@ perf_cpu_notify(struct notifier_block *self, unsigned long action, void *hcpu) case CPU_DOWN_PREPARE: perf_event_exit_cpu(cpu); break; - default: break; } -- cgit v1.2.2 From 62b8563979273424d6ebe9201e34d1acc133ad4f Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Wed, 3 Apr 2013 14:21:34 +0200 Subject: perf: Add sysfs entry to adjust multiplexing interval per PMU This patch adds /sys/device/xxx/perf_event_mux_interval_ms to ajust the multiplexing interval per PMU. The unit is milliseconds. Value has to be >= 1. In the 4th version, we renamed the sysfs file to be more consistent with the other /proc/sys/kernel entries for perf_events. In the 5th version, we handle the reprogramming of the hrtimer using hrtimer_forward_now(). That way, we sync up to new timer value quickly (suggested by Jiri Olsa). Signed-off-by: Stephane Eranian Signed-off-by: Peter Zijlstra Cc: Frederic Weisbecker Cc: Arnaldo Carvalho de Melo Link: http://lkml.kernel.org/r/1364991694-5876-3-git-send-email-eranian@google.com Signed-off-by: Ingo Molnar --- include/linux/perf_event.h | 1 + kernel/events/core.c | 63 +++++++++++++++++++++++++++++++++++++++++++--- 2 files changed, 60 insertions(+), 4 deletions(-) diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 72138d75a60a..6fddac1b27cb 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -194,6 +194,7 @@ struct pmu { int * __percpu pmu_disable_count; struct perf_cpu_context * __percpu pmu_cpu_context; int task_ctx_nr; + int hrtimer_interval_ms; /* * Fully disable/enable this PMU, can be used to protect from the PMI diff --git a/kernel/events/core.c b/kernel/events/core.c index 97bfac7e6f45..53d1b300116a 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -723,13 +723,21 @@ static void __perf_cpu_hrtimer_init(struct perf_cpu_context *cpuctx, int cpu) { struct hrtimer *hr = &cpuctx->hrtimer; struct pmu *pmu = cpuctx->ctx.pmu; + int timer; /* no multiplexing needed for SW PMU */ if (pmu->task_ctx_nr == perf_sw_context) return; - cpuctx->hrtimer_interval = - ns_to_ktime(NSEC_PER_MSEC * PERF_CPU_HRTIMER); + /* + * check default is sane, if not set then force to + * default interval (1/tick) + */ + timer = pmu->hrtimer_interval_ms; + if (timer < 1) + timer = pmu->hrtimer_interval_ms = PERF_CPU_HRTIMER; + + cpuctx->hrtimer_interval = ns_to_ktime(NSEC_PER_MSEC * timer); hrtimer_init(hr, CLOCK_MONOTONIC, HRTIMER_MODE_REL_PINNED); hr->function = perf_cpu_hrtimer_handler; @@ -6001,9 +6009,56 @@ type_show(struct device *dev, struct device_attribute *attr, char *page) return snprintf(page, PAGE_SIZE-1, "%d\n", pmu->type); } +static ssize_t +perf_event_mux_interval_ms_show(struct device *dev, + struct device_attribute *attr, + char *page) +{ + struct pmu *pmu = dev_get_drvdata(dev); + + return snprintf(page, PAGE_SIZE-1, "%d\n", pmu->hrtimer_interval_ms); +} + +static ssize_t +perf_event_mux_interval_ms_store(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t count) +{ + struct pmu *pmu = dev_get_drvdata(dev); + int timer, cpu, ret; + + ret = kstrtoint(buf, 0, &timer); + if (ret) + return ret; + + if (timer < 1) + return -EINVAL; + + /* same value, noting to do */ + if (timer == pmu->hrtimer_interval_ms) + return count; + + pmu->hrtimer_interval_ms = timer; + + /* update all cpuctx for this PMU */ + for_each_possible_cpu(cpu) { + struct perf_cpu_context *cpuctx; + cpuctx = per_cpu_ptr(pmu->pmu_cpu_context, cpu); + cpuctx->hrtimer_interval = ns_to_ktime(NSEC_PER_MSEC * timer); + + if (hrtimer_active(&cpuctx->hrtimer)) + hrtimer_forward_now(&cpuctx->hrtimer, cpuctx->hrtimer_interval); + } + + return count; +} + +#define __ATTR_RW(attr) __ATTR(attr, 0644, attr##_show, attr##_store) + static struct device_attribute pmu_dev_attrs[] = { - __ATTR_RO(type), - __ATTR_NULL, + __ATTR_RO(type), + __ATTR_RW(perf_event_mux_interval_ms), + __ATTR_NULL, }; static int pmu_bus_running; -- cgit v1.2.2 From 13acac307528c9cd7dd9fa9c577419401527b464 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Sat, 18 May 2013 21:34:52 +0300 Subject: perf/x86/intel: Prevent some shift wrapping bugs in the Intel uncore driver We're trying to use 64 bit masks but the shifts wrap so we can't use the high 32 bits. I've fixed this by changing several types to unsigned long long. This is a static checker fix. The one change which is clearly needed is "mask = 0xff << (idx * 8);" where the author obviously intended to use all 64 bits. The other changes are mostly to silence my static checker. Signed-off-by: Dan Carpenter Signed-off-by: Peter Zijlstra Cc: Stephane Eranian Cc: Arnaldo Carvalho de Melo Cc: Jiri Olsa Link: http://lkml.kernel.org/r/20130518183452.GA14587@elgon.mountain Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/perf_event_intel_uncore.c | 4 ++-- arch/x86/kernel/cpu/perf_event_intel_uncore.h | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.c b/arch/x86/kernel/cpu/perf_event_intel_uncore.c index 52441a2af538..c0e356da7408 100644 --- a/arch/x86/kernel/cpu/perf_event_intel_uncore.c +++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.c @@ -644,7 +644,7 @@ snbep_pcu_get_constraint(struct intel_uncore_box *box, struct perf_event *event) (!uncore_box_is_fake(box) && reg1->alloc)) return NULL; again: - mask = 0xff << (idx * 8); + mask = 0xffULL << (idx * 8); raw_spin_lock_irqsave(&er->lock, flags); if (!__BITS_VALUE(atomic_read(&er->ref), idx, 8) || !((config1 ^ er->config) & mask)) { @@ -1923,7 +1923,7 @@ static u64 nhmex_mbox_alter_er(struct perf_event *event, int new_idx, bool modif { struct hw_perf_event *hwc = &event->hw; struct hw_perf_event_extra *reg1 = &hwc->extra_reg; - int idx, orig_idx = __BITS_VALUE(reg1->idx, 0, 8); + u64 idx, orig_idx = __BITS_VALUE(reg1->idx, 0, 8); u64 config = reg1->config; /* get the non-shared control bits and shift them */ diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.h b/arch/x86/kernel/cpu/perf_event_intel_uncore.h index f9528917f6e8..47b3d00c9d89 100644 --- a/arch/x86/kernel/cpu/perf_event_intel_uncore.h +++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.h @@ -337,10 +337,10 @@ NHMEX_M_PMON_CTL_SET_FLAG_SEL_MASK) #define NHMEX_M_PMON_ZDP_CTL_FVC_MASK (((1 << 11) - 1) | (1 << 23)) -#define NHMEX_M_PMON_ZDP_CTL_FVC_EVENT_MASK(n) (0x7 << (11 + 3 * (n))) +#define NHMEX_M_PMON_ZDP_CTL_FVC_EVENT_MASK(n) (0x7ULL << (11 + 3 * (n))) #define WSMEX_M_PMON_ZDP_CTL_FVC_MASK (((1 << 12) - 1) | (1 << 24)) -#define WSMEX_M_PMON_ZDP_CTL_FVC_EVENT_MASK(n) (0x7 << (12 + 3 * (n))) +#define WSMEX_M_PMON_ZDP_CTL_FVC_EVENT_MASK(n) (0x7ULL << (12 + 3 * (n))) /* * use the 9~13 bits to select event If the 7th bit is not set, -- cgit v1.2.2 From 2b923c8f5de6722393e614b096d5040b6d4eaf98 Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Tue, 21 May 2013 12:53:37 +0200 Subject: perf/x86: Check branch sampling priv level in generic code This patch moves commit 7cc23cd to the generic code: perf/x86/intel/lbr: Demand proper privileges for PERF_SAMPLE_BRANCH_KERNEL The check is now implemented in generic code instead of x86 specific code. That way we do not have to repeat the test in each arch supporting branch sampling. Signed-off-by: Stephane Eranian Signed-off-by: Peter Zijlstra Cc: Frederic Weisbecker Cc: Jiri Olsa Cc: Arnaldo Carvalho de Melo Link: http://lkml.kernel.org/r/20130521105337.GA2879@quad Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/perf_event_intel_lbr.c | 13 +++---------- kernel/events/core.c | 9 ++++----- 2 files changed, 7 insertions(+), 15 deletions(-) diff --git a/arch/x86/kernel/cpu/perf_event_intel_lbr.c b/arch/x86/kernel/cpu/perf_event_intel_lbr.c index d978353c939b..de341d4ec92a 100644 --- a/arch/x86/kernel/cpu/perf_event_intel_lbr.c +++ b/arch/x86/kernel/cpu/perf_event_intel_lbr.c @@ -310,7 +310,7 @@ void intel_pmu_lbr_read(void) * - in case there is no HW filter * - in case the HW filter has errata or limitations */ -static int intel_pmu_setup_sw_lbr_filter(struct perf_event *event) +static void intel_pmu_setup_sw_lbr_filter(struct perf_event *event) { u64 br_type = event->attr.branch_sample_type; int mask = 0; @@ -318,11 +318,8 @@ static int intel_pmu_setup_sw_lbr_filter(struct perf_event *event) if (br_type & PERF_SAMPLE_BRANCH_USER) mask |= X86_BR_USER; - if (br_type & PERF_SAMPLE_BRANCH_KERNEL) { - if (perf_paranoid_kernel() && !capable(CAP_SYS_ADMIN)) - return -EACCES; + if (br_type & PERF_SAMPLE_BRANCH_KERNEL) mask |= X86_BR_KERNEL; - } /* we ignore BRANCH_HV here */ @@ -342,8 +339,6 @@ static int intel_pmu_setup_sw_lbr_filter(struct perf_event *event) * be used by fixup code for some CPU */ event->hw.branch_reg.reg = mask; - - return 0; } /* @@ -391,9 +386,7 @@ int intel_pmu_setup_lbr_filter(struct perf_event *event) /* * setup SW LBR filter */ - ret = intel_pmu_setup_sw_lbr_filter(event); - if (ret) - return ret; + intel_pmu_setup_sw_lbr_filter(event); /* * setup HW LBR filter, if any diff --git a/kernel/events/core.c b/kernel/events/core.c index 53d1b300116a..a0780b3a3d50 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -6481,11 +6481,6 @@ static int perf_copy_attr(struct perf_event_attr __user *uattr, if (!(mask & ~PERF_SAMPLE_BRANCH_PLM_ALL)) return -EINVAL; - /* kernel level capture: check permissions */ - if ((mask & PERF_SAMPLE_BRANCH_PERM_PLM) - && perf_paranoid_kernel() && !capable(CAP_SYS_ADMIN)) - return -EACCES; - /* propagate priv level, when not set for branch */ if (!(mask & PERF_SAMPLE_BRANCH_PLM_ALL)) { @@ -6503,6 +6498,10 @@ static int perf_copy_attr(struct perf_event_attr __user *uattr, */ attr->branch_sample_type = mask; } + /* kernel level capture: check permissions */ + if ((mask & PERF_SAMPLE_BRANCH_KERNEL) + && perf_paranoid_kernel() && !capable(CAP_SYS_ADMIN)) + return -EACCES; } if (attr->sample_type & PERF_SAMPLE_REGS_USER) { -- cgit v1.2.2 From 1b45adcd9a503428e6de6b39bc6892d86c9c1d41 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 21 May 2013 13:05:37 +0200 Subject: perf/x86/amd: Rework AMD PMU init code Josh reported that his QEMU is a bad hardware emulator and trips a WARN in the AMD PMU init code. He requested the WARN be turned into a pr_err() or similar. While there, rework the code a little. Reported-by: Josh Boyer Acked-by: Robert Richter Acked-by: Jacob Shin Cc: Stephane Eranian Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/r/20130521110537.GG26912@twins.programming.kicks-ass.net Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/perf_event_amd.c | 34 +++++++++++++++++----------------- 1 file changed, 17 insertions(+), 17 deletions(-) diff --git a/arch/x86/kernel/cpu/perf_event_amd.c b/arch/x86/kernel/cpu/perf_event_amd.c index 7e28d9467bb4..4cbe03287b08 100644 --- a/arch/x86/kernel/cpu/perf_event_amd.c +++ b/arch/x86/kernel/cpu/perf_event_amd.c @@ -648,48 +648,48 @@ static __initconst const struct x86_pmu amd_pmu = { .cpu_dead = amd_pmu_cpu_dead, }; -static int setup_event_constraints(void) +static int __init amd_core_pmu_init(void) { - if (boot_cpu_data.x86 == 0x15) + if (!cpu_has_perfctr_core) + return 0; + + switch (boot_cpu_data.x86) { + case 0x15: + pr_cont("Fam15h "); x86_pmu.get_event_constraints = amd_get_event_constraints_f15h; - return 0; -} + break; -static int setup_perfctr_core(void) -{ - if (!cpu_has_perfctr_core) { - WARN(x86_pmu.get_event_constraints == amd_get_event_constraints_f15h, - KERN_ERR "Odd, counter constraints enabled but no core perfctrs detected!"); + default: + pr_err("core perfctr but no constraints; unknown hardware!\n"); return -ENODEV; } - WARN(x86_pmu.get_event_constraints == amd_get_event_constraints, - KERN_ERR "hw perf events core counters need constraints handler!"); - /* * If core performance counter extensions exists, we must use * MSR_F15H_PERF_CTL/MSR_F15H_PERF_CTR msrs. See also - * x86_pmu_addr_offset(). + * amd_pmu_addr_offset(). */ x86_pmu.eventsel = MSR_F15H_PERF_CTL; x86_pmu.perfctr = MSR_F15H_PERF_CTR; x86_pmu.num_counters = AMD64_NUM_COUNTERS_CORE; - printk(KERN_INFO "perf: AMD core performance counters detected\n"); - + pr_cont("core perfctr, "); return 0; } __init int amd_pmu_init(void) { + int ret; + /* Performance-monitoring supported from K7 and later: */ if (boot_cpu_data.x86 < 6) return -ENODEV; x86_pmu = amd_pmu; - setup_event_constraints(); - setup_perfctr_core(); + ret = amd_core_pmu_init(); + if (ret) + return ret; /* Events are common for all AMDs */ memcpy(hw_cache_event_ids, amd_hw_cache_event_ids, -- cgit v1.2.2 From a6572f84c5b135d9b6df279ed3c8de028bd1edd9 Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Fri, 17 May 2013 10:31:04 +0800 Subject: watchdog: Disallow setting watchdog_thresh to -1 In old kernels, it's allowed to set softlockup_thresh to -1 or 0 to disable softlockup detection. However watchdog_thresh only uses 0 to disable detection, and setting it to -1 just froze my box and nothing I can do but reboot. Signed-off-by: Li Zefan Acked-by: Don Zickus Link: http://lkml.kernel.org/r/51959668.9040106@huawei.com Signed-off-by: Ingo Molnar --- kernel/sysctl.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 9edcf456e0fc..b0a1f99907f3 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -120,7 +120,6 @@ extern int blk_iopoll_enabled; /* Constants used for minimum and maximum */ #ifdef CONFIG_LOCKUP_DETECTOR static int sixty = 60; -static int neg_one = -1; #endif static int zero; @@ -814,7 +813,7 @@ static struct ctl_table kern_table[] = { .maxlen = sizeof(int), .mode = 0644, .proc_handler = proc_dowatchdog, - .extra1 = &neg_one, + .extra1 = &zero, .extra2 = &sixty, }, { -- cgit v1.2.2 From 08825c90af6e4bb902b3a51abb0ae6530199f682 Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Fri, 17 May 2013 10:31:20 +0800 Subject: watchdog: Document watchdog_thresh sysctl Signed-off-by: Li Zefan Acked-by: Don Zickus Link: http://lkml.kernel.org/r/51959678.6000802@huawei.com Signed-off-by: Ingo Molnar --- Documentation/sysctl/kernel.txt | 14 ++++++++++++++ 1 file changed, 14 insertions(+) diff --git a/Documentation/sysctl/kernel.txt b/Documentation/sysctl/kernel.txt index ccd42589e124..e8fabd6cda35 100644 --- a/Documentation/sysctl/kernel.txt +++ b/Documentation/sysctl/kernel.txt @@ -76,6 +76,7 @@ show up in /proc/sys/kernel: - tainted - threads-max - unknown_nmi_panic +- watchdog_thresh - version ============================================================== @@ -648,3 +649,16 @@ that time, kernel debugging information is displayed on console. NMI switch that most IA32 servers have fires unknown NMI up, for example. If a system hangs up, try pressing the NMI switch. + +============================================================== + +watchdog_thresh: + +This value can be used to control the frequency of hrtimer and NMI +events and the soft and hard lockup thresholds. The default threshold +is 10 seconds. + +The softlockup threshold is (2 * watchdog_thresh). Setting this +tunable to zero will disable lockup detection altogether. + +============================================================== -- cgit v1.2.2 From c0ffaf3655fab1909a920c8f30ba1722932d01bb Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Fri, 17 May 2013 10:31:35 +0800 Subject: watchdog: Remove softlockup_thresh from Documentation The old softlockup detector has been replaced with new lockup detector long ago. Signed-off-by: Li Zefan Acked-by: Don Zickus Link: http://lkml.kernel.org/r/51959687.9090305@huawei.com Signed-off-by: Ingo Molnar --- Documentation/sysctl/kernel.txt | 10 ---------- 1 file changed, 10 deletions(-) diff --git a/Documentation/sysctl/kernel.txt b/Documentation/sysctl/kernel.txt index e8fabd6cda35..bcff3f9de550 100644 --- a/Documentation/sysctl/kernel.txt +++ b/Documentation/sysctl/kernel.txt @@ -70,7 +70,6 @@ show up in /proc/sys/kernel: - shmall - shmmax [ sysv ipc ] - shmmni -- softlockup_thresh - stop-a [ SPARC only ] - sysrq ==> Documentation/sysrq.txt - tainted @@ -605,15 +604,6 @@ without users and with a dead originative process will be destroyed. ============================================================== -softlockup_thresh: - -This value can be used to lower the softlockup tolerance threshold. The -default threshold is 60 seconds. If a cpu is locked up for 60 seconds, -the kernel complains. Valid values are 1-60 seconds. Setting this -tunable to zero will disable the softlockup detection altogether. - -============================================================== - tainted: Non-zero if the kernel has been tainted. Numeric values, which -- cgit v1.2.2 From 6956664a5c4c32d5aa48fe96d5e2421a3e3f72d5 Mon Sep 17 00:00:00 2001 From: Sukadev Bhattiprolu Date: Fri, 29 Mar 2013 12:14:43 -0700 Subject: perf tools: Fix bug in isupper() and islower() One of the reasons 'perf test' is failing on Power appears to be due to a bug in isupper(). isupper(c) and islower(c) should be checking 'c' against the mask 0x20. Instead they are checking sane_ctype[c] which causes isupper() to be true for lower case letters. Signed-off-by: Sukadev Bhattiprolu Acked-by: Namhyung Kim Link: http://lkml.kernel.org/r/20130329192950.GA9312@us.ibm.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/util.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/tools/perf/util/util.h b/tools/perf/util/util.h index a45710b70a55..7a484c97e500 100644 --- a/tools/perf/util/util.h +++ b/tools/perf/util/util.h @@ -221,8 +221,8 @@ extern unsigned char sane_ctype[256]; #define isalpha(x) sane_istest(x,GIT_ALPHA) #define isalnum(x) sane_istest(x,GIT_ALPHA | GIT_DIGIT) #define isprint(x) sane_istest(x,GIT_PRINT) -#define islower(x) (sane_istest(x,GIT_ALPHA) && sane_istest(x,0x20)) -#define isupper(x) (sane_istest(x,GIT_ALPHA) && !sane_istest(x,0x20)) +#define islower(x) (sane_istest(x,GIT_ALPHA) && (x & 0x20)) +#define isupper(x) (sane_istest(x,GIT_ALPHA) && !(x & 0x20)) #define tolower(x) sane_case((unsigned char)(x), 0x20) #define toupper(x) sane_case((unsigned char)(x), 0) -- cgit v1.2.2 From 26353a61b977e57b58dd3555bc0422fea46c5ad6 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Mon, 1 Apr 2013 20:35:17 +0900 Subject: perf hists: Fix an invalid memory free on he->branch_info The branch info was allocated for the whole stack and passed matching hist entry for each level during processing samples. Thus when a hist entry tries to free its branch info like in hists__collapse_insert_entry it'll face following error. *** glibc detected *** perf: munmap_chunk(): invalid pointer: 0x00000000014e9d20 *** ======= Backtrace: ========= /lib64/libc.so.6[0x387d47ae16] perf[0x4923bd] perf(cmd_report+0xd68)[0x432a08] perf[0x41a663] perf(main+0x58f)[0x419eaf] /lib64/libc.so.6(__libc_start_main+0xf5)[0x387d421735] perf[0x419f95] Fix it by allocating and copying branch info for each new hist entry. Signed-off-by: Namhyung Kim Cc: Andi Kleen Cc: David Ahern Cc: Ingo Molnar Cc: Jiri Olsa Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1364816125-12212-2-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-report.c | 9 ++++++--- tools/perf/util/hist.c | 14 ++++++++++++++ 2 files changed, 20 insertions(+), 3 deletions(-) diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index bd0ca81eeaca..d9f2de3e81fe 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -187,6 +187,9 @@ static int perf_report__add_branch_hist_entry(struct perf_tool *tool, for (i = 0; i < sample->branch_stack->nr; i++) { if (rep->hide_unresolved && !(bi[i].from.sym && bi[i].to.sym)) continue; + + err = -ENOMEM; + /* * The report shows the percentage of total branches captured * and not events sampled. Thus we use a pseudo period of 1. @@ -195,7 +198,6 @@ static int perf_report__add_branch_hist_entry(struct perf_tool *tool, &bi[i], 1, 1); if (he) { struct annotation *notes; - err = -ENOMEM; bx = he->branch_info; if (bx->from.sym && use_browser == 1 && sort__has_sym) { notes = symbol__annotation(bx->from.sym); @@ -226,11 +228,12 @@ static int perf_report__add_branch_hist_entry(struct perf_tool *tool, } evsel->hists.stats.total_period += 1; hists__inc_nr_events(&evsel->hists, PERF_RECORD_SAMPLE); - err = 0; } else - return -ENOMEM; + goto out; } + err = 0; out: + free(bi); return err; } diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 6b32721f829a..9438d576459d 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -292,6 +292,20 @@ static struct hist_entry *hist_entry__new(struct hist_entry *template) he->ms.map->referenced = true; if (he->branch_info) { + /* + * This branch info is (a part of) allocated from + * machine__resolve_bstack() and will be freed after + * adding new entries. So we need to save a copy. + */ + he->branch_info = malloc(sizeof(*he->branch_info)); + if (he->branch_info == NULL) { + free(he); + return NULL; + } + + memcpy(he->branch_info, template->branch_info, + sizeof(*he->branch_info)); + if (he->branch_info->from.map) he->branch_info->from.map->referenced = true; if (he->branch_info->to.map) -- cgit v1.2.2 From ceb2acbc2c1387c8785b3c98b482f5a2b89447c3 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Mon, 1 Apr 2013 20:35:18 +0900 Subject: perf hists: Free unused mem info of a matched hist entry The mem info is shared between matched entries so one should be freed. Signed-off-by: Namhyung Kim Cc: Andi Kleen Cc: David Ahern Cc: Ingo Molnar Cc: Jiri Olsa Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1364816125-12212-3-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/hist.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 9438d576459d..514fc0470e38 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -374,6 +374,12 @@ static struct hist_entry *add_hist_entry(struct hists *hists, if (!cmp) { he_stat__add_period(&he->stat, period, weight); + /* + * This mem info was allocated from machine__resolve_mem + * and will not be used anymore. + */ + free(entry->mem_info); + /* If the map of an existing hist_entry has * become out-of-date due to an exec() or * similar, update it. Otherwise we will -- cgit v1.2.2 From ded19d57a621e92a27a05972949ad3230f84d0b0 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Mon, 1 Apr 2013 20:35:19 +0900 Subject: perf report: Fix alignment of symbol column when -v is given When -v option is given, the symbol sort key prints its address also but it wasn't properly aligned since hists__calc_col_len() misses the additional part. Also it missed 2 spaces for 0x prefix when printing. $ perf report --stdio -v -s sym # Samples: 133 of event 'cycles' # Event count (approx.): 50536717 # # Overhead Symbol # ........ .............................. # 12.20% 0xffffffff81384c50 v [k] intel_idle 7.62% 0xffffffff8170976a v [k] ftrace_caller 7.02% 0x2d986d B [.] 0x00000000002d986d Signed-off-by: Namhyung Kim Cc: Andi Kleen Cc: David Ahern Cc: Ingo Molnar Cc: Jiri Olsa Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1364816125-12212-4-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/hist.c | 26 +++++++++++++++----------- tools/perf/util/sort.c | 2 +- 2 files changed, 16 insertions(+), 12 deletions(-) diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 514fc0470e38..72b4eec820c3 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -70,9 +70,17 @@ void hists__calc_col_len(struct hists *hists, struct hist_entry *h) int symlen; u16 len; - if (h->ms.sym) - hists__new_col_len(hists, HISTC_SYMBOL, h->ms.sym->namelen + 4); - else { + /* + * +4 accounts for '[x] ' priv level info + * +2 accounts for 0x prefix on raw addresses + * +3 accounts for ' y ' symtab origin info + */ + if (h->ms.sym) { + symlen = h->ms.sym->namelen + 4; + if (verbose) + symlen += BITS_PER_LONG / 4 + 2 + 3; + hists__new_col_len(hists, HISTC_SYMBOL, symlen); + } else { symlen = unresolved_col_width + 4 + 2; hists__new_col_len(hists, HISTC_SYMBOL, symlen); hists__set_unres_dso_col_len(hists, HISTC_DSO); @@ -91,12 +99,10 @@ void hists__calc_col_len(struct hists *hists, struct hist_entry *h) hists__new_col_len(hists, HISTC_PARENT, h->parent->namelen); if (h->branch_info) { - /* - * +4 accounts for '[x] ' priv level info - * +2 account of 0x prefix on raw addresses - */ if (h->branch_info->from.sym) { symlen = (int)h->branch_info->from.sym->namelen + 4; + if (verbose) + symlen += BITS_PER_LONG / 4 + 2 + 3; hists__new_col_len(hists, HISTC_SYMBOL_FROM, symlen); symlen = dso__name_len(h->branch_info->from.map->dso); @@ -109,6 +115,8 @@ void hists__calc_col_len(struct hists *hists, struct hist_entry *h) if (h->branch_info->to.sym) { symlen = (int)h->branch_info->to.sym->namelen + 4; + if (verbose) + symlen += BITS_PER_LONG / 4 + 2 + 3; hists__new_col_len(hists, HISTC_SYMBOL_TO, symlen); symlen = dso__name_len(h->branch_info->to.map->dso); @@ -121,10 +129,6 @@ void hists__calc_col_len(struct hists *hists, struct hist_entry *h) } if (h->mem_info) { - /* - * +4 accounts for '[x] ' priv level info - * +2 account of 0x prefix on raw addresses - */ if (h->mem_info->daddr.sym) { symlen = (int)h->mem_info->daddr.sym->namelen + 4 + unresolved_col_width + 2; diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c index 5f52d492590c..16d5e38befe5 100644 --- a/tools/perf/util/sort.c +++ b/tools/perf/util/sort.c @@ -194,7 +194,7 @@ static int _hist_entry__sym_snprintf(struct map *map, struct symbol *sym, if (verbose) { char o = map ? dso__symtab_origin(map->dso) : '!'; ret += repsep_snprintf(bf, size, "%-#*llx %c ", - BITS_PER_LONG / 4, ip, o); + BITS_PER_LONG / 4 + 2, ip, o); } ret += repsep_snprintf(bf + ret, size - ret, "[%c] ", level); -- cgit v1.2.2 From 55369fc179b0572d0b4a06a9be1d2779b3ac22e0 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Mon, 1 Apr 2013 20:35:20 +0900 Subject: perf sort: Introduce sort__mode variable It's used for determining current sort mode which can be one of NORMAL, BRANCH and new MEMORY. Signed-off-by: Namhyung Kim Cc: Andi Kleen Cc: David Ahern Cc: Ingo Molnar Cc: Jiri Olsa Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1364816125-12212-5-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-report.c | 23 +++++++++++++---------- tools/perf/ui/browsers/hists.c | 4 ++-- tools/perf/util/sort.c | 4 ++-- tools/perf/util/sort.h | 8 +++++++- 4 files changed, 24 insertions(+), 15 deletions(-) diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index d9f2de3e81fe..c877982a64d3 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -311,7 +311,7 @@ static int process_sample_event(struct perf_tool *tool, if (rep->cpu_list && !test_bit(sample->cpu, rep->cpu_bitmap)) return 0; - if (sort__branch_mode == 1) { + if (sort__mode == SORT_MODE__BRANCH) { if (perf_report__add_branch_hist_entry(tool, &al, sample, evsel, machine)) { pr_debug("problem adding lbr entry, skipping event\n"); @@ -387,7 +387,7 @@ static int perf_report__setup_sample_type(struct perf_report *rep) } } - if (sort__branch_mode == 1) { + if (sort__mode == SORT_MODE__BRANCH) { if (!self->fd_pipe && !(sample_type & PERF_SAMPLE_BRANCH_STACK)) { ui__error("Selected -b but no branch data. " @@ -694,7 +694,9 @@ static int parse_branch_mode(const struct option *opt __maybe_unused, const char *str __maybe_unused, int unset) { - sort__branch_mode = !unset; + int *branch_mode = opt->value; + + *branch_mode = !unset; return 0; } @@ -703,6 +705,7 @@ int cmd_report(int argc, const char **argv, const char *prefix __maybe_unused) struct perf_session *session; struct stat st; bool has_br_stack = false; + int branch_mode = -1; int ret = -1; char callchain_default_opt[] = "fractal,0.5,callee"; const char * const report_usage[] = { @@ -799,7 +802,7 @@ int cmd_report(int argc, const char **argv, const char *prefix __maybe_unused) "Show a column with the sum of periods"), OPT_BOOLEAN(0, "group", &symbol_conf.event_group, "Show event group information together"), - OPT_CALLBACK_NOOPT('b', "branch-stack", &sort__branch_mode, "", + OPT_CALLBACK_NOOPT('b', "branch-stack", &branch_mode, "", "use branch records for histogram filling", parse_branch_mode), OPT_STRING(0, "objdump", &objdump_path, "path", "objdump binary to use for disassembly and annotations"), @@ -849,11 +852,11 @@ repeat: has_br_stack = perf_header__has_feat(&session->header, HEADER_BRANCH_STACK); - if (sort__branch_mode == -1 && has_br_stack) - sort__branch_mode = 1; + if (branch_mode == -1 && has_br_stack) + sort__mode = SORT_MODE__BRANCH; - /* sort__branch_mode could be 0 if --no-branch-stack */ - if (sort__branch_mode == 1) { + /* sort__mode could be NORMAL if --no-branch-stack */ + if (sort__mode == SORT_MODE__BRANCH) { /* * if no sort_order is provided, then specify * branch-mode specific order @@ -864,7 +867,7 @@ repeat: } if (report.mem_mode) { - if (sort__branch_mode == 1) { + if (sort__mode == SORT_MODE__BRANCH) { fprintf(stderr, "branch and mem mode incompatible\n"); goto error; } @@ -934,7 +937,7 @@ repeat: sort_entry__setup_elide(&sort_comm, symbol_conf.comm_list, "comm", stdout); - if (sort__branch_mode == 1) { + if (sort__mode == SORT_MODE__BRANCH) { sort_entry__setup_elide(&sort_dso_from, symbol_conf.dso_from_list, "dso_from", stdout); sort_entry__setup_elide(&sort_dso_to, symbol_conf.dso_to_list, "dso_to", stdout); sort_entry__setup_elide(&sort_sym_from, symbol_conf.sym_from_list, "sym_from", stdout); diff --git a/tools/perf/ui/browsers/hists.c b/tools/perf/ui/browsers/hists.c index d88a2d0acb6d..cad8e37f05d9 100644 --- a/tools/perf/ui/browsers/hists.c +++ b/tools/perf/ui/browsers/hists.c @@ -1155,7 +1155,7 @@ static struct hist_browser *hist_browser__new(struct hists *hists) browser->b.refresh = hist_browser__refresh; browser->b.seek = ui_browser__hists_seek; browser->b.use_navkeypressed = true; - if (sort__branch_mode == 1) + if (sort__mode == SORT_MODE__BRANCH) browser->has_symbols = sort_sym_from.list.next != NULL; else browser->has_symbols = sort_sym.list.next != NULL; @@ -1488,7 +1488,7 @@ static int perf_evsel__hists_browse(struct perf_evsel *evsel, int nr_events, if (!browser->has_symbols) goto add_exit_option; - if (sort__branch_mode == 1) { + if (sort__mode == SORT_MODE__BRANCH) { bi = browser->he_selection->branch_info; if (browser->selection != NULL && bi && diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c index 16d5e38befe5..a6ddad41d57a 100644 --- a/tools/perf/util/sort.c +++ b/tools/perf/util/sort.c @@ -9,7 +9,7 @@ const char *sort_order = default_sort_order; int sort__need_collapse = 0; int sort__has_parent = 0; int sort__has_sym = 0; -int sort__branch_mode = -1; /* -1 = means not set */ +enum sort_mode sort__mode = SORT_MODE__NORMAL; enum sort_type sort__first_dimension; @@ -943,7 +943,7 @@ int sort_dimension__add(const char *tok) if (strncasecmp(tok, sd->name, strlen(tok))) continue; - if (sort__branch_mode != 1) + if (sort__mode != SORT_MODE__BRANCH) return -EINVAL; if (sd->entry == &sort_sym_from || sd->entry == &sort_sym_to) diff --git a/tools/perf/util/sort.h b/tools/perf/util/sort.h index f24bdf64238c..39ff4b86ae84 100644 --- a/tools/perf/util/sort.h +++ b/tools/perf/util/sort.h @@ -32,7 +32,7 @@ extern const char default_sort_order[]; extern int sort__need_collapse; extern int sort__has_parent; extern int sort__has_sym; -extern int sort__branch_mode; +extern enum sort_mode sort__mode; extern struct sort_entry sort_comm; extern struct sort_entry sort_dso; extern struct sort_entry sort_sym; @@ -123,6 +123,12 @@ static inline void hist_entry__add_pair(struct hist_entry *he, list_add_tail(&he->pairs.head, &pair->pairs.node); } +enum sort_mode { + SORT_MODE__NORMAL, + SORT_MODE__BRANCH, + SORT_MODE__MEMORY, +}; + enum sort_type { /* common sort keys */ SORT_PID, -- cgit v1.2.2 From 2f532d09fa3a7eaf7cf1c23de9767eab8c8c0e7e Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 3 Apr 2013 21:26:10 +0900 Subject: perf sort: Factor out common code in sort_dimension__add() Let's remove duplicate code. Suggested-by: Jiri Olsa Signed-off-by: Namhyung Kim Cc: Andi Kleen Cc: David Ahern Cc: Ingo Molnar Cc: Jiri Olsa Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1364991979-3008-2-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/sort.c | 41 +++++++++++++++++------------------------ 1 file changed, 17 insertions(+), 24 deletions(-) diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c index a6ddad41d57a..a997955085eb 100644 --- a/tools/perf/util/sort.c +++ b/tools/perf/util/sort.c @@ -895,6 +895,21 @@ static struct sort_dimension bstack_sort_dimensions[] = { #undef DIM +static void __sort_dimension__add(struct sort_dimension *sd, enum sort_type idx) +{ + if (sd->taken) + return; + + if (sd->entry->se_collapse) + sort__need_collapse = 1; + + if (list_empty(&hist_entry__sort_list)) + sort__first_dimension = idx; + + list_add_tail(&sd->entry->list, &hist_entry__sort_list); + sd->taken = 1; +} + int sort_dimension__add(const char *tok) { unsigned int i; @@ -922,18 +937,7 @@ int sort_dimension__add(const char *tok) sort__has_sym = 1; } - if (sd->taken) - return 0; - - if (sd->entry->se_collapse) - sort__need_collapse = 1; - - if (list_empty(&hist_entry__sort_list)) - sort__first_dimension = i; - - list_add_tail(&sd->entry->list, &hist_entry__sort_list); - sd->taken = 1; - + __sort_dimension__add(sd, i); return 0; } @@ -949,18 +953,7 @@ int sort_dimension__add(const char *tok) if (sd->entry == &sort_sym_from || sd->entry == &sort_sym_to) sort__has_sym = 1; - if (sd->taken) - return 0; - - if (sd->entry->se_collapse) - sort__need_collapse = 1; - - if (list_empty(&hist_entry__sort_list)) - sort__first_dimension = i + __SORT_BRANCH_STACK; - - list_add_tail(&sd->entry->list, &hist_entry__sort_list); - sd->taken = 1; - + __sort_dimension__add(sd, i + __SORT_BRANCH_STACK); return 0; } -- cgit v1.2.2 From afab87b91f3f331d55664172dad8e476e6ffca9d Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 3 Apr 2013 21:26:11 +0900 Subject: perf sort: Separate out memory-specific sort keys Since they're used only for perf mem, separate out them to a different dimension so that normal user cannot access them by any chance. For global/local weights, I'm not entirely sure to place them into the memory dimension. But it's the only user at this time. Signed-off-by: Namhyung Kim Cc: Andi Kleen Cc: David Ahern Cc: Ingo Molnar Cc: Jiri Olsa Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1364991979-3008-3-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-report.c | 2 ++ tools/perf/util/sort.c | 39 +++++++++++++++++++++++++++++++-------- tools/perf/util/sort.h | 19 +++++++++++-------- 3 files changed, 44 insertions(+), 16 deletions(-) diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index c877982a64d3..669405c9b8a2 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -871,6 +871,8 @@ repeat: fprintf(stderr, "branch and mem mode incompatible\n"); goto error; } + sort__mode = SORT_MODE__MEMORY; + /* * if no sort_order is provided, then specify * branch-mode specific order diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c index a997955085eb..1dbf16949250 100644 --- a/tools/perf/util/sort.c +++ b/tools/perf/util/sort.c @@ -871,14 +871,6 @@ static struct sort_dimension common_sort_dimensions[] = { DIM(SORT_PARENT, "parent", sort_parent), DIM(SORT_CPU, "cpu", sort_cpu), DIM(SORT_SRCLINE, "srcline", sort_srcline), - DIM(SORT_LOCAL_WEIGHT, "local_weight", sort_local_weight), - DIM(SORT_GLOBAL_WEIGHT, "weight", sort_global_weight), - DIM(SORT_MEM_DADDR_SYMBOL, "symbol_daddr", sort_mem_daddr_sym), - DIM(SORT_MEM_DADDR_DSO, "dso_daddr", sort_mem_daddr_dso), - DIM(SORT_MEM_LOCKED, "locked", sort_mem_locked), - DIM(SORT_MEM_TLB, "tlb", sort_mem_tlb), - DIM(SORT_MEM_LVL, "mem", sort_mem_lvl), - DIM(SORT_MEM_SNOOP, "snoop", sort_mem_snoop), }; #undef DIM @@ -895,6 +887,21 @@ static struct sort_dimension bstack_sort_dimensions[] = { #undef DIM +#define DIM(d, n, func) [d - __SORT_MEMORY_MODE] = { .name = n, .entry = &(func) } + +static struct sort_dimension memory_sort_dimensions[] = { + DIM(SORT_LOCAL_WEIGHT, "local_weight", sort_local_weight), + DIM(SORT_GLOBAL_WEIGHT, "weight", sort_global_weight), + DIM(SORT_MEM_DADDR_SYMBOL, "symbol_daddr", sort_mem_daddr_sym), + DIM(SORT_MEM_DADDR_DSO, "dso_daddr", sort_mem_daddr_dso), + DIM(SORT_MEM_LOCKED, "locked", sort_mem_locked), + DIM(SORT_MEM_TLB, "tlb", sort_mem_tlb), + DIM(SORT_MEM_LVL, "mem", sort_mem_lvl), + DIM(SORT_MEM_SNOOP, "snoop", sort_mem_snoop), +}; + +#undef DIM + static void __sort_dimension__add(struct sort_dimension *sd, enum sort_type idx) { if (sd->taken) @@ -957,6 +964,22 @@ int sort_dimension__add(const char *tok) return 0; } + for (i = 0; i < ARRAY_SIZE(memory_sort_dimensions); i++) { + struct sort_dimension *sd = &memory_sort_dimensions[i]; + + if (strncasecmp(tok, sd->name, strlen(tok))) + continue; + + if (sort__mode != SORT_MODE__MEMORY) + return -EINVAL; + + if (sd->entry == &sort_mem_daddr_sym) + sort__has_sym = 1; + + __sort_dimension__add(sd, i + __SORT_MEMORY_MODE); + return 0; + } + return -ESRCH; } diff --git a/tools/perf/util/sort.h b/tools/perf/util/sort.h index 39ff4b86ae84..0232d476da87 100644 --- a/tools/perf/util/sort.h +++ b/tools/perf/util/sort.h @@ -138,14 +138,6 @@ enum sort_type { SORT_PARENT, SORT_CPU, SORT_SRCLINE, - SORT_LOCAL_WEIGHT, - SORT_GLOBAL_WEIGHT, - SORT_MEM_DADDR_SYMBOL, - SORT_MEM_DADDR_DSO, - SORT_MEM_LOCKED, - SORT_MEM_TLB, - SORT_MEM_LVL, - SORT_MEM_SNOOP, /* branch stack specific sort keys */ __SORT_BRANCH_STACK, @@ -154,6 +146,17 @@ enum sort_type { SORT_SYM_FROM, SORT_SYM_TO, SORT_MISPREDICT, + + /* memory mode specific sort keys */ + __SORT_MEMORY_MODE, + SORT_LOCAL_WEIGHT = __SORT_MEMORY_MODE, + SORT_GLOBAL_WEIGHT, + SORT_MEM_DADDR_SYMBOL, + SORT_MEM_DADDR_DSO, + SORT_MEM_LOCKED, + SORT_MEM_TLB, + SORT_MEM_LVL, + SORT_MEM_SNOOP, }; /* -- cgit v1.2.2 From 08e71542fd0f4a0e30b4e3794329d63ae891e0c0 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 3 Apr 2013 21:26:19 +0900 Subject: perf sort: Consolidate sort_entry__setup_elide() The same code was duplicate to places, factor them out to common sort__setup_elide(). Signed-off-by: Namhyung Kim Cc: Andi Kleen Cc: David Ahern Cc: Ingo Molnar Cc: Jiri Olsa Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1364991979-3008-11-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/builtin-diff.c | 4 +--- tools/perf/builtin-report.c | 20 +------------------- tools/perf/builtin-top.c | 4 +--- tools/perf/util/sort.c | 45 +++++++++++++++++++++++++++++++++++++++++++-- tools/perf/util/sort.h | 3 +-- 5 files changed, 47 insertions(+), 29 deletions(-) diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c index 2d0462d89a97..cabbea5f0bc2 100644 --- a/tools/perf/builtin-diff.c +++ b/tools/perf/builtin-diff.c @@ -611,9 +611,7 @@ int cmd_diff(int argc, const char **argv, const char *prefix __maybe_unused) setup_pager(); - sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "dso", NULL); - sort_entry__setup_elide(&sort_comm, symbol_conf.comm_list, "comm", NULL); - sort_entry__setup_elide(&sort_sym, symbol_conf.sym_list, "symbol", NULL); + sort__setup_elide(NULL); return __cmd_diff(); } diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index 669405c9b8a2..d45bf9b0361d 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -937,25 +937,7 @@ repeat: report.symbol_filter_str = argv[0]; } - sort_entry__setup_elide(&sort_comm, symbol_conf.comm_list, "comm", stdout); - - if (sort__mode == SORT_MODE__BRANCH) { - sort_entry__setup_elide(&sort_dso_from, symbol_conf.dso_from_list, "dso_from", stdout); - sort_entry__setup_elide(&sort_dso_to, symbol_conf.dso_to_list, "dso_to", stdout); - sort_entry__setup_elide(&sort_sym_from, symbol_conf.sym_from_list, "sym_from", stdout); - sort_entry__setup_elide(&sort_sym_to, symbol_conf.sym_to_list, "sym_to", stdout); - } else { - if (report.mem_mode) { - sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "symbol_daddr", stdout); - sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "dso_daddr", stdout); - sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "mem", stdout); - sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "local_weight", stdout); - sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "tlb", stdout); - sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "snoop", stdout); - } - sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "dso", stdout); - sort_entry__setup_elide(&sort_sym, symbol_conf.sym_list, "symbol", stdout); - } + sort__setup_elide(stdout); ret = __cmd_report(&report); if (ret == K_SWITCH_INPUT_DATA) { diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 67bdb9f14ad6..2eb272d8753c 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -1200,9 +1200,7 @@ int cmd_top(int argc, const char **argv, const char *prefix __maybe_unused) if (symbol__init() < 0) return -1; - sort_entry__setup_elide(&sort_dso, symbol_conf.dso_list, "dso", stdout); - sort_entry__setup_elide(&sort_comm, symbol_conf.comm_list, "comm", stdout); - sort_entry__setup_elide(&sort_sym, symbol_conf.sym_list, "symbol", stdout); + sort__setup_elide(stdout); /* * Avoid annotation data structures overhead when symbols aren't on the diff --git a/too