linux/tools/perf
Kan Liang ff165628d7 perf callchain: Stitch LBR call stack
In LBR call stack mode, the depth of reconstructed LBR call stack limits
to the number of LBR registers.

  For example, on skylake, the depth of reconstructed LBR call stack is
  always <= 32.

  # To display the perf.data header info, please use
  # --header/--header-only options.
  #
  #
  # Total Lost Samples: 0
  #
  # Samples: 6K of event 'cycles'
  # Event count (approx.): 6487119731
  #
  # Children      Self  Command          Shared Object       Symbol
  # ........  ........  ...............  ..................
  # ................................

    99.97%    99.97%  tchain_edit      tchain_edit        [.] f43
            |
             --99.64%--f11
                       f12
                       f13
                       f14
                       f15
                       f16
                       f17
                       f18
                       f19
                       f20
                       f21
                       f22
                       f23
                       f24
                       f25
                       f26
                       f27
                       f28
                       f29
                       f30
                       f31
                       f32
                       f33
                       f34
                       f35
                       f36
                       f37
                       f38
                       f39
                       f40
                       f41
                       f42
                       f43

For a call stack which is deeper than LBR limit, HW will overwrite the
LBR register with oldest branch. Only partial call stacks can be
reconstructed.

However, the overwritten LBRs may still be retrieved from previous
sample. At that moment, HW hasn't overwritten the LBR registers yet.
Perf tools can stitch those overwritten LBRs on current call stacks to
get a more complete call stack.

To determine if LBRs can be stitched, perf tools need to compare current
sample with previous sample.

- They should have identical LBR records (Same from, to and flags
  values, and the same physical index of LBR registers).

- The searching starts from the base-of-stack of current sample.

Once perf determines to stitch the previous LBRs, the corresponding LBR
cursor nodes will be copied to 'lists'.  The 'lists' is to track the LBR
cursor nodes which are going to be stitched.

When the stitching is over, the nodes will not be freed immediately.
They will be moved to 'free_lists'. Next stitching may reuse the space.
Both 'lists' and 'free_lists' will be freed when all samples are
processed.

Committer notes:

Fix the intel-pt.c initialization of the union with 'struct
branch_flags', that breaks the build with its unnamed union on older gcc
versions.

Uninline thread__free_stitch_list(), as it grew big and started dragging
includes to thread.h, so move it to thread.c where what it needs in
terms of headers are already there.

This fixes the build in several systems such as debian:experimental when
cross building to the MIPS32 architecture, i.e. in the other cases what
was needed was being included by sheer luck.

  In file included from builtin-sched.c:11:
  util/thread.h: In function 'thread__free_stitch_list':
  util/thread.h:169:3: error: implicit declaration of function 'free' [-Werror=implicit-function-declaration]
    169 |   free(pos);
        |   ^~~~
  util/thread.h:169:3: error: incompatible implicit declaration of built-in function 'free' [-Werror]
  util/thread.h:19:1: note: include '<stdlib.h>' or provide a declaration of 'free'
     18 | #include "callchain.h"
    +++ |+#include <stdlib.h>
     19 |
  util/thread.h:174:3: error: incompatible implicit declaration of built-in function 'free' [-Werror]
    174 |   free(pos);
        |   ^~~~
  util/thread.h:174:3: note: include '<stdlib.h>' or provide a declaration of 'free'

Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexey Budankov <alexey.budankov@linux.intel.com>
Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Pavel Gerasimov <pavel.gerasimov@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@linux.ibm.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Vitaly Slobodskoy <vitaly.slobodskoy@intel.com>
Link: http://lore.kernel.org/lkml/20200319202517.23423-13-kan.liang@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2020-04-18 09:05:01 -03:00
..
arch tools headers: Update x86's syscall_64.tbl with the kernel sources 2020-04-14 11:02:52 -03:00
bench perf bench: Add event synthesis benchmark 2020-04-16 12:19:12 -03:00
Documentation perf header: Support CPU PMU capabilities 2020-04-18 09:05:00 -03:00
examples/bpf Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next 2020-01-28 16:02:33 -08:00
include/bpf perf bpf: Remove bpf/ subdir from bpf.h headers used to build bpf events 2020-02-18 10:13:28 -03:00
jvmti perf jvmti: Link against tools/lib/ctype.h to have weak strlcpy() 2019-10-15 11:47:38 -03:00
pmu-events perf pmu-events x86: Use CPU_CLK_UNHALTED.THREAD in Kernel_Utilization metric 2020-04-03 09:37:56 -03:00
python
scripts perf script: Add flamegraph.py script 2020-04-16 12:19:14 -03:00
tests perf parser: Add support to specify rXXX event with pmu 2020-04-18 09:05:00 -03:00
trace tools headers UAPI: Sync linux/mman.h with the kernel 2020-04-14 09:04:53 -03:00
ui perf report/top TUI: Fix title line formatting 2020-04-03 09:37:55 -03:00
util perf callchain: Stitch LBR call stack 2020-04-18 09:05:01 -03:00
.gitignore .gitignore: add SPDX License Identifier 2020-03-25 11:50:48 +01:00
Build
builtin-annotate.c perf annotate: Prefer cmdline option over default config 2020-02-27 10:45:08 -03:00
builtin-bench.c perf bench: Add event synthesis benchmark 2020-04-16 12:19:12 -03:00
builtin-buildid-cache.c perf session: Return error code for perf_session__new() function on failure 2019-09-20 15:58:11 -03:00
builtin-buildid-list.c perf session: Return error code for perf_session__new() function on failure 2019-09-20 15:58:11 -03:00
builtin-c2c.c perf c2c: Fix return type for histogram sorting comparision functions 2020-01-14 13:29:21 -03:00
builtin-config.c perf tools: Remove util.h from where it is not needed 2019-09-20 09:19:20 -03:00
builtin-data.c perf debug: Remove needless include directives from debug.h 2019-08-31 19:10:19 -03:00
builtin-diff.c perf tools: Basic support for CGROUP event 2020-04-03 09:37:55 -03:00
builtin-evlist.c perf evsel: Introduce evsel_fprintf.h 2019-09-25 16:26:34 -03:00
builtin-ftrace.c perf tools: Support CAP_PERFMON capability 2020-04-16 12:19:08 -03:00
builtin-help.c perf debug: Remove needless include directives from debug.h 2019-08-31 19:10:19 -03:00
builtin-inject.c perf inject: Fix processing of ID index for injected instruction tracing 2019-12-04 12:39:53 -03:00
builtin-kallsyms.c perf dsos: Move the dsos struct and its methods to separate source files 2019-08-31 22:24:10 -03:00
builtin-kmem.c perf callchain: Use 'struct map_symbol' in 'struct callchain_cursor_node' 2019-11-12 08:20:53 -03:00
builtin-kvm.c perf kvm: Use evlist layer api when possible 2019-11-06 15:43:05 -03:00
builtin-list.c perf list: Hide deprecated events by default 2019-10-19 15:35:01 -03:00
builtin-lock.c perf session: Return error code for perf_session__new() function on failure 2019-09-20 15:58:11 -03:00
builtin-mem.c perf session: Return error code for perf_session__new() function on failure 2019-09-20 15:58:11 -03:00
builtin-probe.c perf probe: Check return value of strlist__add() for -ENOMEM 2020-02-27 11:03:13 -03:00
builtin-record.c perf record: Add --all-cgroups option 2020-04-03 09:37:55 -03:00
builtin-report.c perf auxtrace: Add an option to synthesize callchains for regular events 2020-04-16 12:19:15 -03:00
builtin-sched.c perf sched timehist: Add support for filtering on CPU 2020-01-06 11:46:09 -03:00
builtin-script.c perf auxtrace: Add an option to synthesize callchains for regular events 2020-04-16 12:19:15 -03:00
builtin-stat.c perf stat: Honour --timeout for forked workloads 2020-04-16 12:17:41 -03:00
builtin-timechart.c perf session: Return error code for perf_session__new() function on failure 2019-09-20 15:58:11 -03:00
builtin-top.c perf top: Support hotkey to change sort order 2020-04-03 09:37:55 -03:00
builtin-trace.c perf trace: Resolve prctl's 'option' arg strings to numbers 2020-02-11 16:41:50 -03:00
builtin-version.c perf symbols: Move mem_info and branch_info out of symbol.h 2019-08-31 22:27:48 -03:00
builtin.h perf tools: Remove needless util.h include from builtin.h 2019-08-28 17:19:34 -03:00
check-headers.sh tools headers: Synchronize linux/bits.h with the kernel sources 2020-04-14 11:40:05 -03:00
command-list.txt
CREDITS
design.txt perf tools: Support CAP_PERFMON capability 2020-04-16 12:19:08 -03:00
Makefile tools: Let O= makes handle a relative path with -C option 2020-03-06 17:08:28 -03:00
Makefile.config perf tools: Support Python 3.8+ in Makefile 2020-04-03 10:03:44 -03:00
Makefile.perf perf: Normalize gcc parameter when generating arch errno table 2020-03-26 11:04:01 -03:00
MANIFEST libperf: Move to tools/lib/perf 2020-01-06 11:46:09 -03:00
perf-archive.sh
perf-completion.sh
perf-read-vdso.c
perf-sys.h perf tools: Make usage of test_attr__* optional for perf-sys.h 2019-10-31 21:38:41 +01:00
perf-with-kcore.sh
perf.c libperf: Merge libperf_set_print() into libperf_init() 2019-09-25 09:51:49 -03:00
perf.h perf time-utils: Adopt rdclock() from perf.h 2019-08-29 17:38:32 -03:00