kernel_optimize_test/kernel/trace
Frederic Weisbecker c530665c31 perf: Take a hot regs snapshot for trace events
We are taking a wrong regs snapshot when a trace event triggers.
Either we use get_irq_regs(), which gives us the interrupted
registers if we are in an interrupt, or we use task_pt_regs()
which gives us the state before we entered the kernel, assuming
we are lucky enough to be no kernel thread, in which case
task_pt_regs() returns the initial set of regs when the kernel
thread was started.

What we want is different. We need a hot snapshot of the regs,
so that we can get the instruction pointer to record in the
sample, the frame pointer for the callchain, and some other
things.

Let's use the new perf_fetch_caller_regs() for that.

Comparison with perf record -e lock: -R -a -f -g
Before:

        perf  [kernel]                   [k] __do_softirq
               |
               --- __do_softirq
                  |
                  |--55.16%-- __open
                  |
                   --44.84%-- __write_nocancel

After:

            perf  [kernel]           [k] perf_tp_event
               |
               --- perf_tp_event
                  |
                  |--41.07%-- lock_acquire
                  |          |
                  |          |--39.36%-- _raw_spin_lock
                  |          |          |
                  |          |          |--7.81%-- hrtimer_interrupt
                  |          |          |          smp_apic_timer_interrupt
                  |          |          |          apic_timer_interrupt

The old case was producing unreliable callchains. Now having
right frame and instruction pointers, we have the trace we
want.

Also syscalls and kprobe events already have the right regs,
let's use them instead of wasting a retrieval.

v2: Follow the rename perf_save_regs() -> perf_fetch_caller_regs()

Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Jason Baron <jbaron@redhat.com>
Cc: Archs <linux-arch@vger.kernel.org>
2010-03-10 14:40:38 +01:00
..
blktrace.c blktrace: perform cleanup after setup error 2010-02-28 19:47:19 +01:00
ftrace.c Merge branch 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip 2010-02-28 10:20:25 -08:00
Kconfig Merge branch 'tracing/core' of git://git.kernel.org/pub/scm/linux/kernel/git/frederic/random-tracing into tracing/core 2010-02-27 10:06:10 +01:00
kmemtrace.c kmemtrace: Fix up tracer registration 2009-10-01 11:53:44 +02:00
Makefile perf events: Remove CONFIG_EVENT_PROFILE 2009-12-28 10:33:06 +01:00
power-traces.c tracing/power: Remove two exports 2009-12-13 18:37:28 +01:00
ring_buffer_benchmark.c local_t: Move local.h include to ringbuffer.c and ring_buffer_benchmark.c 2010-01-05 15:34:50 +09:00
ring_buffer.c Merge branch 'master' into percpu 2010-02-02 14:38:15 +09:00
trace_boot.c tracing: add filter event logic to special, mmiotrace and boot tracers 2009-09-12 23:34:04 -04:00
trace_branch.c tracing: Add correct/incorrect to sort keys for branch annotation output 2010-02-09 21:35:05 -05:00
trace_clock.c locking: Convert __raw_spin* functions to arch_spin* 2009-12-14 23:55:32 +01:00
trace_entries.h hw-breakpoints: Rewrite the hw-breakpoints layer on top of perf events 2009-11-08 15:34:42 +01:00
trace_event_profile.c perf: Take a hot regs snapshot for trace events 2010-03-10 14:40:38 +01:00
trace_events_filter.c Merge branch 'perf/urgent' into perf/core 2010-01-29 10:36:22 +01:00
trace_events.c tracing: Simplify memory recycle of trace_define_field 2010-02-25 10:42:55 -05:00
trace_export.c tracing: Remove show_format and related macros from TRACE_EVENT 2010-01-06 12:08:46 -05:00
trace_functions_graph.c Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu 2010-03-03 07:34:18 -08:00
trace_functions.c tracing: switch function prints from %pf to %ps 2009-09-17 15:53:40 -04:00
trace_hw_branches.c Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu 2009-12-14 09:58:24 -08:00
trace_irqsoff.c tracing: Add stack trace to irqsoff tracer 2009-12-11 13:19:51 -05:00
trace_kprobe.c perf: Take a hot regs snapshot for trace events 2010-03-10 14:40:38 +01:00
trace_ksym.c ksym_tracer: Remove trace_stat 2009-12-30 07:50:50 +01:00
trace_mmiotrace.c tracing: add filter event logic to special, mmiotrace and boot tracers 2009-09-12 23:34:04 -04:00
trace_nop.c
trace_output.c tracing: Add full state to trace_seq 2009-12-09 14:05:49 -05:00
trace_output.h tracing: consolidate code between trace_output.c and trace_function_graph.c 2009-09-11 14:24:13 -04:00
trace_printk.c tracing: Remove markers 2009-09-18 21:22:08 +02:00
trace_sched_switch.c tracing: pass around ring buffer instead of tracer 2009-09-04 18:59:39 -04:00
trace_sched_wakeup.c locking: Convert __raw_spin* functions to arch_spin* 2009-12-14 23:55:32 +01:00
trace_selftest_dynamic.c
trace_selftest.c locking: Convert __raw_spin* functions to arch_spin* 2009-12-14 23:55:32 +01:00
trace_stack.c tracing: Fix circular dead lock in stack trace 2010-02-02 10:20:18 -05:00
trace_stat.c trace_stat: Fix missing entry in stat file 2009-08-17 11:25:09 +02:00
trace_stat.h tracing/stat: Add stat_release() callback 2009-07-10 12:14:05 +02:00
trace_syscalls.c perf: Take a hot regs snapshot for trace events 2010-03-10 14:40:38 +01:00
trace_sysprof.c perf events, x86/stacktrace: Make stack walking optional 2009-12-17 09:56:19 +01:00
trace_workqueue.c tracing/workqueues: Add refcnt to struct cpu_workqueue_stats 2009-07-10 12:14:07 +02:00
trace.c Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu 2010-03-03 07:34:18 -08:00
trace.h tracing: Fix ftrace_event_call alignment for use with gcc 4.5 2010-02-25 09:38:11 -05:00