From a7636d9ecfa3ab7800a7c04c1f89378229eff609 Mon Sep 17 00:00:00 2001 From: Martin KaFai Lau Date: Wed, 3 Feb 2016 12:28:28 -0800 Subject: kprobes: Optimize hot path by using percpu counter to collect 'nhit' statistics When doing ebpf+kprobe on some hot TCP functions (e.g. tcp_rcv_established), the kprobe_dispatcher() function shows up in 'perf report'. In kprobe_dispatcher(), there is a lot of cache bouncing on 'tk->nhit++'. 'tk->nhit' and 'tk->tp.flags' also share the same cacheline. perf report (cycles:pp): 8.30% ipv4_dst_check 4.74% copy_user_enhanced_fast_string 3.93% dst_release 2.80% tcp_v4_rcv 2.31% queued_spin_lock_slowpath 2.30% _raw_spin_lock 1.88% mlx4_en_process_rx_cq 1.84% eth_get_headlen 1.81% ip_rcv_finish ~~~~ 1.71% kprobe_dispatcher ~~~~ 1.55% mlx4_en_xmit 1.09% __probe_kernel_read perf report after patch: 9.15% ipv4_dst_check 5.00% copy_user_enhanced_fast_string 4.12% dst_release 2.96% tcp_v4_rcv 2.50% _raw_spin_lock 2.39% queued_spin_lock_slowpath 2.11% eth_get_headlen 2.03% mlx4_en_process_rx_cq 1.69% mlx4_en_xmit 1.19% ip_rcv_finish 1.12% __probe_kernel_read 1.02% ehci_hcd_cleanup Signed-off-by: Martin KaFai Lau Acked-by: Masami Hiramatsu Acked-by: Steven Rostedt Cc: Alexei Starovoitov Cc: Josef Bacik Cc: Kernel Team Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1454531308-2441898-1-git-send-email-kafai@fb.com Signed-off-by: Ingo Molnar --- kernel/trace/trace_kprobe.c | 19 +++++++++++++++---- 1 file changed, 15 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c index c9956440d0e6..21b81a41dae5 100644 --- a/kernel/trace/trace_kprobe.c +++ b/kernel/trace/trace_kprobe.c @@ -30,7 +30,7 @@ struct trace_kprobe { struct list_head list; struct kretprobe rp; /* Use rp.kp for kprobe use */ - unsigned long nhit; + unsigned long __percpu *nhit; const char *symbol; /* symbol name */ struct trace_probe tp; }; @@ -274,6 +274,10 @@ static struct trace_kprobe *alloc_trace_kprobe(const char *group, if (!tk) return ERR_PTR(ret); + tk->nhit = alloc_percpu(unsigned long); + if (!tk->nhit) + goto error; + if (symbol) { tk->symbol = kstrdup(symbol, GFP_KERNEL); if (!tk->symbol) @@ -313,6 +317,7 @@ static struct trace_kprobe *alloc_trace_kprobe(const char *group, error: kfree(tk->tp.call.name); kfree(tk->symbol); + free_percpu(tk->nhit); kfree(tk); return ERR_PTR(ret); } @@ -327,6 +332,7 @@ static void free_trace_kprobe(struct trace_kprobe *tk) kfree(tk->tp.call.class->system); kfree(tk->tp.call.name); kfree(tk->symbol); + free_percpu(tk->nhit); kfree(tk); } @@ -874,9 +880,14 @@ static const struct file_operations kprobe_events_ops = { static int probes_profile_seq_show(struct seq_file *m, void *v) { struct trace_kprobe *tk = v; + unsigned long nhit = 0; + int cpu; + + for_each_possible_cpu(cpu) + nhit += *per_cpu_ptr(tk->nhit, cpu); seq_printf(m, " %-44s %15lu %15lu\n", - trace_event_name(&tk->tp.call), tk->nhit, + trace_event_name(&tk->tp.call), nhit, tk->rp.kp.nmissed); return 0; @@ -1225,7 +1236,7 @@ static int kprobe_dispatcher(struct kprobe *kp, struct pt_regs *regs) { struct trace_kprobe *tk = container_of(kp, struct trace_kprobe, rp.kp); - tk->nhit++; + raw_cpu_inc(*tk->nhit); if (tk->tp.flags & TP_FLAG_TRACE) kprobe_trace_func(tk, regs); @@ -1242,7 +1253,7 @@ kretprobe_dispatcher(struct kretprobe_instance *ri, struct pt_regs *regs) { struct trace_kprobe *tk = container_of(ri->rp, struct trace_kprobe, rp); - tk->nhit++; + raw_cpu_inc(*tk->nhit); if (tk->tp.flags & TP_FLAG_TRACE) kretprobe_trace_func(tk, ri, regs); -- cgit v1.2.3 From 3b364d7b587db0f0eeafde0f271e0698187de776 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 9 Feb 2016 20:11:40 +0000 Subject: perf/core: Remove unused arguments from a bunch of functions No functional change, just less confusing to read. Signed-off-by: Thomas Gleixner Cc: Arnaldo Carvalho de Melo Cc: Jiri Olsa Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Vince Weaver Link: http://lkml.kernel.org/r/20160209201007.921540566@linutronix.de Signed-off-by: Ingo Molnar --- kernel/events/core.c | 21 ++++++++++----------- 1 file changed, 10 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/kernel/events/core.c b/kernel/events/core.c index 0d58522103cd..94c47e3f9a0a 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -6710,7 +6710,7 @@ static void swevent_hlist_release(struct swevent_htable *swhash) kfree_rcu(hlist, rcu_head); } -static void swevent_hlist_put_cpu(struct perf_event *event, int cpu) +static void swevent_hlist_put_cpu(int cpu) { struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu); @@ -6722,15 +6722,15 @@ static void swevent_hlist_put_cpu(struct perf_event *event, int cpu) mutex_unlock(&swhash->hlist_mutex); } -static void swevent_hlist_put(struct perf_event *event) +static void swevent_hlist_put(void) { int cpu; for_each_possible_cpu(cpu) - swevent_hlist_put_cpu(event, cpu); + swevent_hlist_put_cpu(cpu); } -static int swevent_hlist_get_cpu(struct perf_event *event, int cpu) +static int swevent_hlist_get_cpu(int cpu) { struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu); int err = 0; @@ -6753,14 +6753,13 @@ exit: return err; } -static int swevent_hlist_get(struct perf_event *event) +static int swevent_hlist_get(void) { - int err; - int cpu, failed_cpu; + int err, cpu, failed_cpu; get_online_cpus(); for_each_possible_cpu(cpu) { - err = swevent_hlist_get_cpu(event, cpu); + err = swevent_hlist_get_cpu(cpu); if (err) { failed_cpu = cpu; goto fail; @@ -6773,7 +6772,7 @@ fail: for_each_possible_cpu(cpu) { if (cpu == failed_cpu) break; - swevent_hlist_put_cpu(event, cpu); + swevent_hlist_put_cpu(cpu); } put_online_cpus(); @@ -6789,7 +6788,7 @@ static void sw_perf_event_destroy(struct perf_event *event) WARN_ON(event->parent); static_key_slow_dec(&perf_swevent_enabled[event_id]); - swevent_hlist_put(event); + swevent_hlist_put(); } static int perf_swevent_init(struct perf_event *event) @@ -6820,7 +6819,7 @@ static int perf_swevent_init(struct perf_event *event) if (!event->parent) { int err; - err = swevent_hlist_get(event); + err = swevent_hlist_get(); if (err) return err; -- cgit v1.2.3 From 675965b00d734c985e4285f5bec7e524d15fc4e1 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 22 Feb 2016 22:19:27 +0000 Subject: perf: Export perf_event_sysfs_show() Required to use it in modular perf drivers. Signed-off-by: Thomas Gleixner Signed-off-by: Peter Zijlstra (Intel) Cc: Andi Kleen Cc: Arnaldo Carvalho de Melo Cc: Borislav Petkov Cc: Harish Chegondi Cc: Jacob Pan Cc: Jiri Olsa Cc: Kan Liang Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Vince Weaver Cc: linux-kernel@vger.kernel.org Link: http://lkml.kernel.org/r/20160222221012.930735780@linutronix.de Signed-off-by: Ingo Molnar --- kernel/events/core.c | 1 + 1 file changed, 1 insertion(+) (limited to 'kernel') diff --git a/kernel/events/core.c b/kernel/events/core.c index df5589fa8bf0..5dcc0bd08d11 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -9446,6 +9446,7 @@ ssize_t perf_event_sysfs_show(struct device *dev, struct device_attribute *attr, return 0; } +EXPORT_SYMBOL_GPL(perf_event_sysfs_show); static int __init perf_event_sysfs_init(void) { -- cgit v1.2.3 From 026842d148b920dc28f0499ede4950dcb098d4d5 Mon Sep 17 00:00:00 2001 From: Taeung Song Date: Fri, 26 Feb 2016 13:23:01 -0500 Subject: tracing/syscalls: Rename "/format" tracepoint field name "nr" to "__syscall_nr: Some tracepoint have multiple fields with the same name, "nr", the first one is a unique syscall ID, the other is a syscall argument: # cat /sys/kernel/debug/tracing/events/syscalls/sys_enter_io_getevents/format name: sys_enter_io_getevents ID: 747 format: field:unsigned short common_type; offset:0; size:2; signed:0; field:unsigned char common_flags; offset:2; size:1; signed:0; field:unsigned char common_preempt_count; offset:3; size:1; signed:0; field:int common_pid; offset:4; size:4; signed:1; field:int nr; offset:8; size:4; signed:1; field:aio_context_t ctx_id; offset:16; size:8; signed:0; field:long min_nr; offset:24; size:8; signed:0; field:long nr; offset:32; size:8; signed:0; field:struct io_event * events; offset:40; size:8; signed:0; field:struct timespec * timeout; offset:48; size:8; signed:0; print fmt: "ctx_id: 0x%08lx, min_nr: 0x%08lx, nr: 0x%08lx, events: 0x%08lx, timeout: 0x%08lx", ((unsigned long)(REC->ctx_id)), ((unsigned long)(REC->min_nr)), ((unsigned long)(REC->nr)), ((unsigned long)(REC->events)), ((unsigned long)(REC->timeout)) # Fix it by renaming the "/format" common tracepoint field "nr" to "__syscall_nr". Signed-off-by: Taeung Song [ Do not rename the struct member, just the '/format' field name ] Signed-off-by: Steven Rostedt Acked-by: Peter Zijlstra Cc: Jiri Olsa Cc: Lai Jiangshan Cc: Namhyung Kim Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/20160226132301.3ae065a4@gandalf.local.home Signed-off-by: Arnaldo Carvalho de Melo --- kernel/trace/trace_syscalls.c | 16 +++++++++------- 1 file changed, 9 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_syscalls.c b/kernel/trace/trace_syscalls.c index 0655afbea83f..d1663083d903 100644 --- a/kernel/trace/trace_syscalls.c +++ b/kernel/trace/trace_syscalls.c @@ -186,11 +186,11 @@ print_syscall_exit(struct trace_iterator *iter, int flags, extern char *__bad_type_size(void); -#define SYSCALL_FIELD(type, name) \ - sizeof(type) != sizeof(trace.name) ? \ +#define SYSCALL_FIELD(type, field, name) \ + sizeof(type) != sizeof(trace.field) ? \ __bad_type_size() : \ - #type, #name, offsetof(typeof(trace), name), \ - sizeof(trace.name), is_signed_type(type) + #type, #name, offsetof(typeof(trace), field), \ + sizeof(trace.field), is_signed_type(type) static int __init __set_enter_print_fmt(struct syscall_metadata *entry, char *buf, int len) @@ -261,7 +261,8 @@ static int __init syscall_enter_define_fields(struct trace_event_call *call) int i; int offset = offsetof(typeof(trace), args); - ret = trace_define_field(call, SYSCALL_FIELD(int, nr), FILTER_OTHER); + ret = trace_define_field(call, SYSCALL_FIELD(int, nr, __syscall_nr), + FILTER_OTHER); if (ret) return ret; @@ -281,11 +282,12 @@ static int __init syscall_exit_define_fields(struct trace_event_call *call) struct syscall_trace_exit trace; int ret; - ret = trace_define_field(call, SYSCALL_FIELD(int, nr), FILTER_OTHER); + ret = trace_define_field(call, SYSCALL_FIELD(int, nr, __syscall_nr), + FILTER_OTHER); if (ret) return ret; - ret = trace_define_field(call, SYSCALL_FIELD(long, ret), + ret = trace_define_field(call, SYSCALL_FIELD(long, ret, ret), FILTER_OTHER); return ret; -- cgit v1.2.3 From 927a5570855836e5d5859a80ce7e91e963545e8f Mon Sep 17 00:00:00 2001 From: Alexander Shishkin Date: Wed, 2 Mar 2016 13:24:14 +0200 Subject: perf/core: Fix perf_sched_count derailment The error path in perf_event_open() is such that asking for a sampling event on a PMU that doesn't generate interrupts will end up in dropping the perf_sched_count even though it hasn't been incremented for this event yet. Given a sufficient amount of these calls, we'll end up disabling scheduler's jump label even though we'd still have active events in the system, thereby facilitating the arrival of the infernal regions upon us. I'm fixing this by moving account_event() inside perf_event_alloc(). Signed-off-by: Alexander Shishkin Signed-off-by: Peter Zijlstra (Intel) Cc: Cc: Arnaldo Carvalho de Melo Cc: Arnaldo Carvalho de Melo Cc: Jiri Olsa Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Thomas Gleixner Cc: Vince Weaver Cc: vince@deater.net Link: http://lkml.kernel.org/r/1456917854-29427-1-git-send-email-alexander.shishkin@linux.intel.com Signed-off-by: Ingo Molnar --- kernel/events/core.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/events/core.c b/kernel/events/core.c index 5dcc0bd08d11..b7231498de47 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -8000,6 +8000,9 @@ perf_event_alloc(struct perf_event_attr *attr, int cpu, } } + /* symmetric to unaccount_event() in _free_event() */ + account_event(event); + return event; err_per_task: @@ -8363,8 +8366,6 @@ SYSCALL_DEFINE5(perf_event_open, } } - account_event(event); - /* * Special case software events and allow them to be part of * any hardware group. @@ -8661,8 +8662,6 @@ perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu, /* Mark owner so we could distinguish it from user events. */ event->owner = TASK_TOMBSTONE; - account_event(event); - ctx = find_get_context(event->pmu, task, event); if (IS_ERR(ctx)) { err = PTR_ERR(ctx); -- cgit v1.2.3