perf disasm: Don't include C files from the arch directory

Move the arch instructions.c files into appropriately named files in
annotate-arch in the util directory.

Don't #include to compile the code, switch to building the files and fix
up the #includes accordingly.

Move powerpc specific disasm code out of disasm.c and into
annotate-powerpc.c.

Declarations and static removed as appropriate for the code to compile
as separate compilation units.

The e_machine and e_flags set up is moved to the disasm.c architectures
array so that later patches can sort by them.

Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Bill Wendling <morbo@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Julia Lawall <Julia.Lawall@inria.fr>
Cc: Justin Stitt <justinstitt@google.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <nick.desaulniers+lkml@gmail.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Suchit Karunakaran <suchitkarunakaran@gmail.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tianyou Li <tianyou.li@intel.com>
Cc: Will Deacon <will@kernel.org>
Cc: Zecheng Li <zecheng@google.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf disasm: Rework the string arch__is to use the ELF machine

Add new arch__is_x86 and arch__is_powerpc functions that avoid string
comparisons and use the ELF machine.

Remove arch__is() that is no longer used.

Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Bill Wendling <morbo@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Julia Lawall <Julia.Lawall@inria.fr>
Cc: Justin Stitt <justinstitt@google.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <nick.desaulniers+lkml@gmail.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Suchit Karunakaran <suchitkarunakaran@gmail.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tianyou Li <tianyou.li@intel.com>
Cc: Will Deacon <will@kernel.org>
Cc: Zecheng Li <zecheng@google.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf disasm: Constify use of 'struct ins'

The 'struct ins' holds variables that are read but not written, except
during some initialization.

Change most uses to be for a "const struct ins *" version to capture
this immutability.

So the x86__instructions can be const pre-sort it and make the sorted
variable true.

Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Bill Wendling <morbo@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Julia Lawall <Julia.Lawall@inria.fr>
Cc: Justin Stitt <justinstitt@google.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <nick.desaulniers+lkml@gmail.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Suchit Karunakaran <suchitkarunakaran@gmail.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tianyou Li <tianyou.li@intel.com>
Cc: Will Deacon <will@kernel.org>
Cc: Zecheng Li <zecheng@google.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf disasm: Constify use of 'struct ins_op'

The 'struct ins_op' holds variables to function pointers that are read
but not written. Change uses to be for a "const struct ins_op *"
version to capture this immutability.

Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Bill Wendling <morbo@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Julia Lawall <Julia.Lawall@inria.fr>
Cc: Justin Stitt <justinstitt@google.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <nick.desaulniers+lkml@gmail.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Suchit Karunakaran <suchitkarunakaran@gmail.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tianyou Li <tianyou.li@intel.com>
Cc: Will Deacon <will@kernel.org>
Cc: Zecheng Li <zecheng@google.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf disasm: Constify use of 'struct arch'

The 'struct arch' holds variables that are read but not written, except
during some initialization.

Change most uses to be for a "const struct arch *" version to capture
this immutability.

Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Bill Wendling <morbo@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Julia Lawall <Julia.Lawall@inria.fr>
Cc: Justin Stitt <justinstitt@google.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <nick.desaulniers+lkml@gmail.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Suchit Karunakaran <suchitkarunakaran@gmail.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tianyou Li <tianyou.li@intel.com>
Cc: Will Deacon <will@kernel.org>
Cc: Zecheng Li <zecheng@google.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf maps: Fix reference count leak in maps__find_ams()

ams and so ams->ms.map is an in argument, however, it is also
overwritten. As a map is reference counted, ensure a map__put() is done
before overwriting it.

Fixes: 42fd623b58dbcc48 ("perf maps: Get map before returning in maps__find")
Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Bill Wendling <morbo@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Julia Lawall <Julia.Lawall@inria.fr>
Cc: Justin Stitt <justinstitt@google.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <nick.desaulniers+lkml@gmail.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Suchit Karunakaran <suchitkarunakaran@gmail.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tianyou Li <tianyou.li@intel.com>
Cc: Will Deacon <will@kernel.org>
Cc: Zecheng Li <zecheng@google.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf annotate: Fix args leak of map_symbol

map_symbol__exit() needs calling on an annotate_args.ms, however, rather
than introduce proper reference count handling to symbol__annotate()
just switch to passing the map_symbol pointer parameter around, making
the puts the caller's responsibility.

Fix a number of cases to ensure the map in a map_symbol has a
reference count increment and add the then necessary map_symbol_exits.

Fixes: 56e144fe98260a0f ("perf mem_info: Add and use map_symbol__exit and addr_map_symbol__exit")
Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Bill Wendling <morbo@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Julia Lawall <Julia.Lawall@inria.fr>
Cc: Justin Stitt <justinstitt@google.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-csky@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <nick.desaulniers+lkml@gmail.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Suchit Karunakaran <suchitkarunakaran@gmail.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tianyou Li <tianyou.li@intel.com>
Cc: Will Deacon <will@kernel.org>
Cc: Zecheng Li <zecheng@google.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf header: Fix memory leaks in process_cpu_domain_info()

do_read_string() returns a string in allocated memory, for some reason
there was unused memory allocations and unnecessary strdups.

Remove these and make the "perf annotate basic tests" leak sanitizer
clean.

Fixes: d40c68a49f69c9bd ("perf header: Support CPU DOMAIN relation info")
Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Bill Wendling <morbo@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Julia Lawall <Julia.Lawall@inria.fr>
Cc: Justin Stitt <justinstitt@google.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-csky@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <nick.desaulniers+lkml@gmail.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Suchit Karunakaran <suchitkarunakaran@gmail.com>
Cc: Swapnil Sapkal <swapnil.sapkal@amd.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tianyou Li <tianyou.li@intel.com>
Cc: Will Deacon <will@kernel.org>
Cc: Zecheng Li <zecheng@google.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf sched stats: Add details in man page

Document 'perf sched stats' purpose, usage examples and guide on how to
interpret the report data in the perf-sched man page.

Signed-off-by: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Swapnil Sapkal <swapnil.sapkal@amd.com>
Tested-by: Chen Yu <yu.c.chen@intel.com>
Acked-by: Ian Rogers <irogers@google.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Anubhav Shelat <ashelat@redhat.com>
Cc: Ben Gainey <ben.gainey@arm.com>
Cc: Blake Jones <blakejones@google.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: David Vernet <void@manifault.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Gautham Shenoy <gautham.shenoy@amd.com>
Cc: Graham Woodward <graham.woodward@arm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: K Prateek Nayak <kprateek.nayak@amd.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Madadi Vineeth Reddy <vineethr@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Santosh Shukla <santosh.shukla@amd.com>
Cc: Shrikanth Hegde <sshegde@linux.ibm.com>
Cc: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Yang Jihong <yangjihong@bytedance.com>
Cc: Yujie Liu <yujie.liu@intel.com>
Cc: Zhongqiu Han <quic_zhonhan@quicinc.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf sched stats: Add basic 'perf sched stats' test

Add basic test for 'perf sched stats {record|report|diff}' subcommand.

Signed-off-by: Swapnil Sapkal <swapnil.sapkal@amd.com>
Tested-by: Chen Yu <yu.c.chen@intel.com>
Acked-by: Ian Rogers <irogers@google.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Anubhav Shelat <ashelat@redhat.com>
Cc: Ben Gainey <ben.gainey@arm.com>
Cc: Blake Jones <blakejones@google.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: David Vernet <void@manifault.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Gautham Shenoy <gautham.shenoy@amd.com>
Cc: Graham Woodward <graham.woodward@arm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: K Prateek Nayak <kprateek.nayak@amd.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Madadi Vineeth Reddy <vineethr@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Santosh Shukla <santosh.shukla@amd.com>
Cc: Shrikanth Hegde <sshegde@linux.ibm.com>
Cc: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Yang Jihong <yangjihong@bytedance.com>
Cc: Yujie Liu <yujie.liu@intel.com>
Cc: Zhongqiu Han <quic_zhonhan@quicinc.com>
Link: https://lore.kernel.org/r/20260119175833.340369-10-swapnil.sapkal@amd.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf sched stats: Add support for diff subcommand

`perf sched stats diff` subcommand will take two perf.data files as an
input and it will print the diff between the two perf.data files. The
default input to this subcommnd is perf.data.old and perf.data.

Example usage:

 # perf sched stats diff sample1.data sample2.data
 Description
 ----------------------------------------------------------------------------------------------------
 DESC                          -> Description of the field
 COUNT                         -> Value of the field
 PCT_CHANGE                    -> Percent change with corresponding base value
 AVG_JIFFIES                   -> Avg time in jiffies between two consecutive occurrence of event
 ----------------------------------------------------------------------------------------------------

 Time elapsed (in jiffies)                                        :           1,          1
 ----------------------------------------------------------------------------------------------------
 CPU: <ALL CPUS SUMMARY>
 ----------------------------------------------------------------------------------------------------
 DESC                                                                    COUNT1      COUNT2   PCT_CHANGE    PCT_CHANGE1 PCT_CHANGE2
 ----------------------------------------------------------------------------------------------------
 yld_count                                                        :           0,          0  |     0.00% |
 array_exp                                                        :           0,          0  |     0.00% |
 sched_count                                                      :           0,          0  |     0.00% |
 sched_goidle                                                     :           0,          0  |     0.00% |  (     0.00%,      0.00% )
 ttwu_count                                                       :           0,          0  |     0.00% |
 ttwu_local                                                       :           0,          0  |     0.00% |  (     0.00%,      0.00% )
 rq_cpu_time                                                      :       32565,      33525  |     2.95% |
 run_delay                                                        :           0,        436  |     0.00% |  (     0.00%,      1.30% )
 pcount                                                           :           0,          0  |     0.00% |
 ----------------------------------------------------------------------------------------------------
 CPU: <ALL CPUS SUMMARY> | DOMAIN: SMT
 ----------------------------------------------------------------------------------------------------
 DESC                                                                    COUNT1      COUNT2   PCT_CHANGE     AVG_JIFFIES1 AVG_JIFFIES2
 ----------------------------------------- <Category busy> ------------------------------------------
 busy_lb_count                                                    :           0,          0  |     0.00% |  $        0.00,        0.00 $
 busy_lb_balanced                                                 :           0,          0  |     0.00% |  $        0.00,        0.00 $
 busy_lb_failed                                                   :           0,          0  |     0.00% |  $        0.00,        0.00 $
 busy_lb_imbalance_load                                           :           0,          0  |     0.00% |
 busy_lb_imbalance_util                                           :           0,          0  |     0.00% |
 busy_lb_imbalance_task                                           :           0,          0  |     0.00% |
 busy_lb_imbalance_misfit                                         :           0,          0  |     0.00% |
 busy_lb_gained                                                   :           0,          0  |     0.00% |
 busy_lb_hot_gained                                               :           0,          0  |     0.00% |
 busy_lb_nobusyq                                                  :           0,          0  |     0.00% |  $        0.00,        0.00 $
 busy_lb_nobusyg                                                  :           0,          0  |     0.00% |  $        0.00,        0.00 $
 *busy_lb_success_count                                           :           0,          0  |     0.00% |
 *busy_lb_avg_pulled                                              :        0.00,       0.00  |     0.00% |

 ... and so on. Output contains the diff of aggregated data of all the
busy, idle and newidle categories for all the sched domains in the system.

Signed-off-by: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Swapnil Sapkal <swapnil.sapkal@amd.com>
Tested-by: Chen Yu <yu.c.chen@intel.com>
Acked-by: Ian Rogers <irogers@google.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Anubhav Shelat <ashelat@redhat.com>
Cc: Ben Gainey <ben.gainey@arm.com>
Cc: Blake Jones <blakejones@google.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: David Vernet <void@manifault.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Gautham Shenoy <gautham.shenoy@amd.com>
Cc: Graham Woodward <graham.woodward@arm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: K Prateek Nayak <kprateek.nayak@amd.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Madadi Vineeth Reddy <vineethr@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Santosh Shukla <santosh.shukla@amd.com>
Cc: Shrikanth Hegde <sshegde@linux.ibm.com>
Cc: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Yang Jihong <yangjihong@bytedance.com>
Cc: Yujie Liu <yujie.liu@intel.com>
Cc: Zhongqiu Han <quic_zhonhan@quicinc.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf sched stats: Add support for live mode

The live mode works similar to simple `perf stat` command, by profiling
the target and printing results on the terminal as soon as the target
finishes.

Example usage:

  # perf sched stats -- true
  Description
  ----------------------------------------------------------------------------------------------------
  DESC                          -> Description of the field
  COUNT                         -> Value of the field
  PCT_CHANGE                    -> Percent change with corresponding base value
  AVG_JIFFIES                   -> Avg time in jiffies between two consecutive occurrence of event
  ----------------------------------------------------------------------------------------------------

  Time elapsed (in jiffies)                                        :           1
  ----------------------------------------------------------------------------------------------------
  CPU: <ALL CPUS SUMMARY>
  ----------------------------------------------------------------------------------------------------
  DESC                                                                     COUNT   PCT_CHANGE
  ----------------------------------------------------------------------------------------------------
  yld_count                                                        :           0
  array_exp                                                        :           0
  sched_count                                                      :           0
  sched_goidle                                                     :           0  (     0.00% )
  ttwu_count                                                       :           0
  ttwu_local                                                       :           0  (     0.00% )
  rq_cpu_time                                                      :       27875
  run_delay                                                        :           0  (     0.00% )
  pcount                                                           :           0
  ----------------------------------------------------------------------------------------------------
  CPU: <ALL CPUS SUMMARY> | DOMAIN: SMT
  ----------------------------------------------------------------------------------------------------
  DESC                                                                     COUNT    AVG_JIFFIES
  ----------------------------------------- <Category busy> ------------------------------------------
  busy_lb_count                                                    :           0  $        0.00 $
  busy_lb_balanced                                                 :           0  $        0.00 $
  busy_lb_failed                                                   :           0  $        0.00 $
  busy_lb_imbalance_load                                           :           0
  busy_lb_imbalance_util                                           :           0
  busy_lb_imbalance_task                                           :           0
  busy_lb_imbalance_misfit                                         :           0
  busy_lb_gained                                                   :           0
  busy_lb_hot_gained                                               :           0
  busy_lb_nobusyq                                                  :           0  $        0.00 $
  busy_lb_nobusyg                                                  :           0  $        0.00 $
  *busy_lb_success_count                                           :           0
  *busy_lb_avg_pulled                                              :        0.00

  ... and so on. Output will show similar data for all the cpus in the
system.

Co-developed-by: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Swapnil Sapkal <swapnil.sapkal@amd.com>
Tested-by: Chen Yu <yu.c.chen@intel.com>
Tested-by: James Clark <james.clark@linaro.org>
Acked-by: Ian Rogers <irogers@google.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Anubhav Shelat <ashelat@redhat.com>
Cc: Ben Gainey <ben.gainey@arm.com>
Cc: Blake Jones <blakejones@google.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: David Vernet <void@manifault.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Gautham Shenoy <gautham.shenoy@amd.com>
Cc: Graham Woodward <graham.woodward@arm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: K Prateek Nayak <kprateek.nayak@amd.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Madadi Vineeth Reddy <vineethr@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Santosh Shukla <santosh.shukla@amd.com>
Cc: Shrikanth Hegde <sshegde@linux.ibm.com>
Cc: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Yang Jihong <yangjihong@bytedance.com>
Cc: Yujie Liu <yujie.liu@intel.com>
Cc: Zhongqiu Han <quic_zhonhan@quicinc.com>
[ Avoid potentially using 'sv' uninitialized by calling free_cpu_domain_info() only when build_cpu_domain_map() is called ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf sched stats: Add support for report subcommand

`perf sched stats record` captures two sets of samples. For workload
profile, first set right before workload starts and second set after
workload finishes. For the systemwide profile, first set at the
beginning of profile and second set on receiving SIGINT signal.

Add `perf sched stats report` subcommand that will read both the set
of samples, get the diff and render a final report. Final report prints
scheduler stat at cpu granularity as well as sched domain granularity.

Example usage:

  # ./perf sched stats record -- true
  [ perf sched stats: Wrote samples to perf.data ]

  # perf sched stats report
  Description
  ----------------------------------------------------------------------------------------------------
  DESC                          -> Description of the field
  COUNT                         -> Value of the field
  PCT_CHANGE                    -> Percent change with corresponding base value
  AVG_JIFFIES                   -> Avg time in jiffies between two consecutive occurrence of event
  ----------------------------------------------------------------------------------------------------

  Time elapsed (in jiffies)                                        :           1
  ----------------------------------------------------------------------------------------------------
  CPU: <ALL CPUS SUMMARY>
  ----------------------------------------------------------------------------------------------------
  DESC                                                                     COUNT   PCT_CHANGE
  ----------------------------------------------------------------------------------------------------
  yld_count                                                        :           0
  array_exp                                                        :           0
  sched_count                                                      :           0
  sched_goidle                                                     :           0  (     0.00% )
  ttwu_count                                                       :           0
  ttwu_local                                                       :           0  (     0.00% )
  rq_cpu_time                                                      :       33525
  run_delay                                                        :         436  (     1.30% )
  pcount                                                           :           0
  ----------------------------------------------------------------------------------------------------
  CPU: <ALL CPUS SUMMARY> | DOMAIN: SMT
  ----------------------------------------------------------------------------------------------------
  DESC                                                                     COUNT    AVG_JIFFIES
  ----------------------------------------- <Category busy> ------------------------------------------
  busy_lb_count                                                    :           0  $        0.00 $
  busy_lb_balanced                                                 :           0  $        0.00 $
  busy_lb_failed                                                   :           0  $        0.00 $
  busy_lb_imbalance_load                                           :           0
  busy_lb_imbalance_util                                           :           0
  busy_lb_imbalance_task                                           :           0
  busy_lb_imbalance_misfit                                         :           0
  busy_lb_gained                                                   :           0
  busy_lb_hot_gained                                               :           0
  busy_lb_nobusyq                                                  :           0  $        0.00 $
  busy_lb_nobusyg                                                  :           0  $        0.00 $
  *busy_lb_success_count                                           :           0
  *busy_lb_avg_pulled                                              :        0.00

  ... and so on. Output shows similar data for all the cpus in the
system.

Co-developed-by: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Swapnil Sapkal <swapnil.sapkal@amd.com>
Tested-by: Chen Yu <yu.c.chen@intel.com>
Tested-by: James Clark <james.clark@linaro.org>
Acked-by: Ian Rogers <irogers@google.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Anubhav Shelat <ashelat@redhat.com>
Cc: Ben Gainey <ben.gainey@arm.com>
Cc: Blake Jones <blakejones@google.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: David Vernet <void@manifault.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Gautham Shenoy <gautham.shenoy@amd.com>
Cc: Graham Woodward <graham.woodward@arm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: K Prateek Nayak <kprateek.nayak@amd.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Madadi Vineeth Reddy <vineethr@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Santosh Shukla <santosh.shukla@amd.com>
Cc: Shrikanth Hegde <sshegde@linux.ibm.com>
Cc: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Yang Jihong <yangjihong@bytedance.com>
Cc: Yujie Liu <yujie.liu@intel.com>
Cc: Zhongqiu Han <quic_zhonhan@quicinc.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf sched stats: Add schedstat v17 support

The /proc/schedstat file output is standardized with version number.

Add support to record and raw dump v17 version layout.

Version 17 of schedstats removed 'lb_imbalance' field as it has no
significance anymore and instead added more relevant fields namely
'lb_imbalance_load', 'lb_imbalance_util', 'lb_imbalance_task' and
'lb_imbalance_misfit'.

The domain field prints the name of the corresponding sched domain from
this version onwards.

Co-developed-by: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Swapnil Sapkal <swapnil.sapkal@amd.com>
Tested-by: Chen Yu <yu.c.chen@intel.com>
Acked-by: Ian Rogers <irogers@google.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Anubhav Shelat <ashelat@redhat.com>
Cc: Ben Gainey <ben.gainey@arm.com>
Cc: Blake Jones <blakejones@google.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: David Vernet <void@manifault.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Gautham Shenoy <gautham.shenoy@amd.com>
Cc: Graham Woodward <graham.woodward@arm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: K Prateek Nayak <kprateek.nayak@amd.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Madadi Vineeth Reddy <vineethr@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Santosh Shukla <santosh.shukla@amd.com>
Cc: Shrikanth Hegde <sshegde@linux.ibm.com>
Cc: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Yang Jihong <yangjihong@bytedance.com>
Cc: Yujie Liu <yujie.liu@intel.com>
Cc: Zhongqiu Han <quic_zhonhan@quicinc.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf sched stats: Add schedstat v16 support

The /proc/schedstat file output is standardized with version number.

Add support to record and raw dump v16 version layout.

Version 16 of schedstats changed the order of definitions within 'enum
cpu_idle_type', which changed the order of [CPU_MAX_IDLE_TYPES] columns
in show_schedstat().

In particular the position of CPU_IDLE and __CPU_NOT_IDLE changed
places.

Co-developed-by: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Swapnil Sapkal <swapnil.sapkal@amd.com>
Tested-by: Chen Yu <yu.c.chen@intel.com>
Tested-by: James Clark <james.clark@linaro.org>
Acked-by: Ian Rogers <irogers@google.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Anubhav Shelat <ashelat@redhat.com>
Cc: Ben Gainey <ben.gainey@arm.com>
Cc: Blake Jones <blakejones@google.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: David Vernet <void@manifault.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Gautham Shenoy <gautham.shenoy@amd.com>
Cc: Graham Woodward <graham.woodward@arm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: K Prateek Nayak <kprateek.nayak@amd.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Madadi Vineeth Reddy <vineethr@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Santosh Shukla <santosh.shukla@amd.com>
Cc: Shrikanth Hegde <sshegde@linux.ibm.com>
Cc: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Yang Jihong <yangjihong@bytedance.com>
Cc: Yujie Liu <yujie.liu@intel.com>
Cc: Zhongqiu Han <quic_zhonhan@quicinc.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf sched stats: Add record and rawdump support

Define new, perf tool only, sample types and their layouts. Add logic
to parse /proc/schedstat, convert it to perf sample format and save
samples to perf.data file with `perf sched stats record` command.

Also add logic to read perf.data file, interpret schedstat samples and
print rawdump of samples with `perf script -D`.

Note that, /proc/schedstat file output is standardized with version
number. The patch supports v15 but older or newer version can be added
easily.

Co-developed-by: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Ravi Bangoria <ravi.bangoria@amd.com>
Signed-off-by: Swapnil Sapkal <swapnil.sapkal@amd.com>
Tested-by: Chen Yu <yu.c.chen@intel.com>
Tested-by: James Clark <james.clark@linaro.org>
Acked-by: Ian Rogers <irogers@google.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Anubhav Shelat <ashelat@redhat.com>
Cc: Ben Gainey <ben.gainey@arm.com>
Cc: Blake Jones <blakejones@google.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: David Vernet <void@manifault.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Gautham Shenoy <gautham.shenoy@amd.com>
Cc: Graham Woodward <graham.woodward@arm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: K Prateek Nayak <kprateek.nayak@amd.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Madadi Vineeth Reddy <vineethr@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Santosh Shukla <santosh.shukla@amd.com>
Cc: Shrikanth Hegde <sshegde@linux.ibm.com>
Cc: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Yang Jihong <yangjihong@bytedance.com>
Cc: Yujie Liu <yujie.liu@intel.com>
Cc: Zhongqiu Han <quic_zhonhan@quicinc.com>
[ PRIu64 needs uint64_t, not 'unsigned long' to work on both 32-bit and 64-bit ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf header: Support CPU DOMAIN relation info

The '/proc/schedstat' file gives info about load balancing statistics
within a given domain.

It also contains the cpu_mask giving information about the sibling cpus
and domain names after schedstat version 17.

Storing this information in perf header will help tools like `perf sched
stats` for better analysis.

Signed-off-by: Swapnil Sapkal <swapnil.sapkal@amd.com>
Tested-by: Chen Yu <yu.c.chen@intel.com>
Acked-by: Ian Rogers <irogers@google.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Anubhav Shelat <ashelat@redhat.com>
Cc: Ben Gainey <ben.gainey@arm.com>
Cc: Blake Jones <blakejones@google.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: David Vernet <void@manifault.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Gautham Shenoy <gautham.shenoy@amd.com>
Cc: Graham Woodward <graham.woodward@arm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: K Prateek Nayak <kprateek.nayak@amd.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Madadi Vineeth Reddy <vineethr@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Santosh Shukla <santosh.shukla@amd.com>
Cc: Shrikanth Hegde <sshegde@linux.ibm.com>
Cc: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Yang Jihong <yangjihong@bytedance.com>
Cc: Yujie Liu <yujie.liu@intel.com>
Cc: Zhongqiu Han <quic_zhonhan@quicinc.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools/lib: Add list_is_first()

Add list_is_first() to check whether @list is the first entry in list @head

Signed-off-by: Swapnil Sapkal <swapnil.sapkal@amd.com>
Tested-by: Chen Yu <yu.c.chen@intel.com>
Acked-by: Ian Rogers <irogers@google.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Anubhav Shelat <ashelat@redhat.com>
Cc: Ben Gainey <ben.gainey@arm.com>
Cc: Blake Jones <blakejones@google.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: David Vernet <void@manifault.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Gautham Shenoy <gautham.shenoy@amd.com>
Cc: Graham Woodward <graham.woodward@arm.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: K Prateek Nayak <kprateek.nayak@amd.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Madadi Vineeth Reddy <vineethr@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Santosh Shukla <santosh.shukla@amd.com>
Cc: Shrikanth Hegde <sshegde@linux.ibm.com>
Cc: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Yang Jihong <yangjihong@bytedance.com>
Cc: Yujie Liu <yujie.liu@intel.com>
Cc: Zhongqiu Han <quic_zhonhan@quicinc.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf list: Don't write to const memory

Something now detected on fedora 44, where strchr() returns const if it
is passed a const pointer:

  util/print-events.c: In function 'print_sdt_events':
  util/print-events.c:89:29: error: initialization discards 'const' qualifier from pointer target type [-Werror=discarded-qualifiers]
     89 |                 char *bid = strchr(sdt_name->s, '@');
        |                             ^~~~~~

Fix it by using strchrnul() + strncmp() instead of temporarily scrubbing
it with '\0'.

Reviewed-by: Ian Rogers <irogers@google.com>
Suggested-by: David Laight <david.laight.linux@gmail.com>
Link: https://lore.kernel.org/r/20260121112536.27fd5d11@pumpkin
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf list: Signal changing const memory is ok

In this case its a temp list that is created just for listing events and
will be deleted at the end, so just cast it to get rid of the compiler
warning.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf trace: Deal with compiler const checks

The strchr() function these days return const/non-const based on the arg
it receives, and sometimes we need to use casts when we're dealing with
variables that are used in code that needs to safely change the returned
value and sometimes not (as it points to really const areas).

Tweak one such case.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tests sw-clock: Mark the volatile tmp variable as __maybe_unused

As it is just used to waste some cycles, not being used as all, to
silence some compilers.

Noticed with gcc version 16.0.1 20260115 on fedora 44:

    tests/sw-clock.c: In function '__test__sw_clock_freq':
    tests/sw-clock.c:31:22: error: variable 'tmp' set but not used [-Werror=unused-but-set-variable=]
       31 |         volatile int tmp = 0;
          |                      ^~~

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf cs-etm: Test sparse CPU maps

We only currently test with default (all CPUs) or --per-thread mode.

Different permutations of the "-C" option can affect decoding so add
tests for some of them.

Signed-off-by: James Clark <james.clark@linaro.org>
Tested-by: Leo Yan <leo.yan@arm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf cs-etm: Fix decoding for sparse CPU maps

The ETM decoder incorrectly assumed that auxtrace queue indices were
equivalent to CPU number. This assumption is used for inserting records
into the queue, and for fetching queues when given a CPU number. This
assumption held when Perf always opened a dummy event on every CPU, even
if the user provided a subset of CPUs on the commandline, resulting in
the indices aligning.

For example:

  # event : name = cs_etm//u, , id = { 2451, 2452 }, type = 11 (cs_etm), size = 136, config = 0x4010, { sample_period, samp>
  # event : name = dummy:u, , id = { 2453, 2454, 2455, 2456 }, type = 1 (PERF_TYPE_SOFTWARE), size = 136, config = 0x9 (PER>

  0 0 0x200 [0xd0]: PERF_RECORD_ID_INDEX nr: 6
  ... id: 2451  idx: 2  cpu: 2  tid: -1
  ... id: 2452  idx: 3  cpu: 3  tid: -1
  ... id: 2453  idx: 0  cpu: 0  tid: -1
  ... id: 2454  idx: 1  cpu: 1  tid: -1
  ... id: 2455  idx: 2  cpu: 2  tid: -1
  ... id: 2456  idx: 3  cpu: 3  tid: -1

Since commit 811082e4b668 ("perf parse-events: Support user CPUs mixed
with threads/processes") the dummy event no longer behaves in this way,
making the ETM event indices start from 0 on the first CPU recorded
regardless of its ID:

  # event : name = cs_etm//u, , id = { 771, 772 }, type = 11 (cs_etm), size = 144, config = 0x4010, { sample_period, sample>
  # event : name = dummy:u, , id = { 773, 774 }, type = 1 (PERF_TYPE_SOFTWARE), size = 144, config = 0x9 (PERF_COUNT_SW_DUM>

  0 0 0x200 [0x90]: PERF_RECORD_ID_INDEX nr: 4
  ... id: 771  idx: 0  cpu: 2  tid: -1
  ... id: 772  idx: 1  cpu: 3  tid: -1
  ... id: 773  idx: 0  cpu: 2  tid: -1
  ... id: 774  idx: 1  cpu: 3  tid: -1

This causes the following segfault when decoding:

  $ perf record -e cs_etm//u -C 2,3 -- true
  $ perf report

  perf: Segmentation fault
  -------- backtrace --------
  #0 0xaaaabf9fd020 in ui__signal_backtrace setup.c:110
  #1 0xffffab5c7930 in __kernel_rt_sigreturn [vdso][930]
  #2 0xaaaabfb68d30 in cs_etm_decoder__reset cs-etm-decoder.c:85
  #3 0xaaaabfb65930 in cs_etm__get_data_block cs-etm.c:2032
  #4 0xaaaabfb666fc in cs_etm__run_per_cpu_timeless_decoder cs-etm.c:2551
  #5 0xaaaabfb6692c in (cs_etm__process_timeless_queues cs-etm.c:2612
  #6 0xaaaabfb63390 in cs_etm__flush_events cs-etm.c:921
  #7 0xaaaabfb324c0 in auxtrace__flush_events auxtrace.c:2915
  #8 0xaaaabfaac378 in __perf_session__process_events session.c:2285
  #9 0xaaaabfaacc9c in perf_session__process_events session.c:2442
  #10 0xaaaabf8d3d90 in __cmd_report builtin-report.c:1085
  #11 0xaaaabf8d6944 in cmd_report builtin-report.c:1866
  #12 0xaaaabf95ebfc in run_builtin perf.c:351
  #13 0xaaaabf95eeb0 in handle_internal_command perf.c:404
  #14 0xaaaabf95f068 in run_argv perf.c:451
  #15 0xaaaabf95f390 in main perf.c:558
  #16 0xffffaab97400 in __libc_start_call_main libc_start_call_main.h:74
  #17 0xffffaab974d8 in __libc_start_main@@GLIBC_2.34 libc-start.c:128
  #18 0xaaaabf8aa8f0 in _start perf[7a8f0]

Fix it by inserting into the queues based on CPU number, rather than
using the index.

Fixes: 811082e4b668db96 ("perf parse-events: Support user CPUs mixed with threads/processes")
Signed-off-by: James Clark <james.clark@linaro.org>
Tested-by: Leo Yan <leo.yan@arm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: coresight@lists.linaro.org
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf test: Add missing newlines in debug messages

These debug messages bleed into the next log line. Fix it by adding the
missing newlines.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Always uniquify event names

evlist__uniquify_evsel_names() only gets called in __parse_events() if
verbose is > 0. This means that the auto added "slots" events stay as
"slots" rather than being expanded to "cpu_core/slots/" unless Perf is
run in verbose mode. This is invisible to users when running Perf stat
because evlist__print_counters() always calls it regardless of verbose
mode before displaying.

The only thing this seems to affect is the test "Parsing of all PMU
events from sysfs" which fails when not run in verbose mode.
test__checkevent_pmu_events() always expects event names to be prefixed
with the pmu name, but this only happens for "slots" events after
evlist__uniquify_evsel_names() is called.

One fix could be to relax the test to accept the non prefixed name in
normal mode. But seeing as Perf stat uniquifies unconditionally, make
parse_events() do the same.

This fixes the following test failure:

  $ perf test "Parsing of all PMU events from sysfs"
  5.2: Parsing of all PMU events from sysfs                    : FAILED!
  $

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf c2c: Update documentation for adding memory event table

Users may occasionally need to see which options are applied to memory
events.

This helps to understand the behavior of "perf c2c" and "perf mem", and
provides guidance for configuring memory event options directly.

Add a table to track memory events and their corresponding options, and
include the Arm SPE events in it.

Suggested-by: Al Grant <al.grant@arm.com>
Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Leo Yan <leo.yan@arm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf mem: Simplify Arm SPE event config

Since configuration fields default to zero, the zero assignments are
redundant, remove them.

Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Leo Yan <leo.yan@arm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Al Grant <al.grant@arm.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tests kallsyms: Fix missed map__put()

Issue was caught by leak sanitizer and the test robot.

Fixes: 34e271ae55382fbd ("perf test: Add kallsyms split test")
Reported-by: kernel test robot <oliver.sang@intel.com>
Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Closes: https://lore.kernel.org/oe-lkp/202512101502.f3819cd3-lkp@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Add the legacy-cache.json to .gitignore

The commit 0012e0fa221bf9cc ("perf jevents: Add legacy-hardware and
legacy-cache json") will auto-generate: "pmu-events/arch/common/common/legacy-cache.json".

Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Haiyue Wang <haiyuewa@163.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Get debug info of DSO properly

The dso__debuginfo() just used the path name to open the file but it may
be outdated.  It should check build-ID and use the file in the build-ID
cache if available rather than just using the path name.

Let's factor out dso__get_filename() to avoid code duplicate.

Fixes: 53a61a6ca279165d ("perf annotate: Add dso__debuginfo() helper")
Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf test: Add DWARF callchain conversion test

  $ perf test -vv "DWARF callchain"
   87: perf inject to convert DWARF callchains to regular ones:
  --- start ---
  test child forked, pid 1560328
  recording data with DWARF callchain
  [ perf record: Woken up 4 times to write data ]
  [ perf record: Captured and wrote 0.908 MB /tmp/perf-test.nM3WoW (105 samples) ]
  convert DWARF callchain using perf inject
  compare the both result excluding inlined functions
  ---- end(0) ----
   87: perf inject to convert DWARF callchains to regular ones         : Ok
  $

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf inject: Add --convert-callchain option

There are applications not built with frame pointers, so DWARF is needed
to get the stack traces.

`perf record --call-graph dwarf` saves the stack and register data for
each sample to get the stacktrace offline.  But sometimes this data may
have sensitive information and we don't want to keep them in the file.

This new 'perf inject --convert-callchain' option creates the callchains
and discards the stack and register after that.

This saves storage space and processing time for the new data file.

Of course, users should remove the original data file to not keep
sensitive data around.  :)

The down side is that it cannot handle inlined callchain entries as they
all have the same IPs.

Maybe we can add an option to 'perf report' to look up inlined functions
using DWARF - IIUC it doesn't require stack and register data.

This is an example.

  $ perf record --call-graph dwarf -- perf test -w noploop

  $ perf report --stdio --no-children --percent-limit=0 > output-prev

  $ perf inject -i perf.data --convert-callchain -o perf.data.out

  $ perf report --stdio --no-children --percent-limit=0 -i perf.data.out > output-next

  $ diff -u output-prev output-next
  ...
        0.23%  perf          ld-linux-x86-64.so.2  [.] _dl_relocate_object_no_relro
               |
  -            ---elf_dynamic_do_Rela (inlined)
  -               _dl_relocate_object_no_relro
  +            ---_dl_relocate_object_no_relro
                  _dl_relocate_object
                  dl_main
                  _dl_sysdep_start
  -               _dl_start_final (inlined)
                  _dl_start
                  _start

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf machine: Add inline information to frame pointer and LBR callchains

Use append_inlines() in frame pointer and LBR cases.

Update the addr2line test to also test frame pointers.

LBR is also updated but inaccuracy in the branched to IP means the
inline information is missing in the leaf.

Leave LBR callchains untested for now.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf unwind-libdw: Don't discard loaded ELF/DWARF after every unwind

The unwind-libdw dwfl has ELF binaries associated with mmap
addresses. Experimenting with using the per dso dwfl it is required to
alter the address to be 0 based variant. Unfortunately libdwfl doesn't
allow a single unwind and then an update to the return address to be 0
based as there are assertions that registers aren't updated once an
unwind has started, etc.

As removing the dwfl didn't prove possible, an alternative is to just
not discard the dwfl when the unwind ends. The dwfl is valid for a
process unless a dso is loaded at the same address as a previous
one. So keep the dwfl with the maps, invalidate it if a map is removed
(in case a new map replaces it) and recycle the dwfl in the unwinding
code. A wrinkly in the implementation of this is that the attached
thread argument is remembered by the dwfl and so it needs to be a
pointer to memory that also persists with the dwfl (struct
dwfl_ui_thread_info in the code).

Recording 10 seconds of system wide data with --call-graph=dwarf and
then processing with perf report shows a total runtime improvement
from 41.583s to 2.279s (an 18x speedup).

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf build: Remove NO_LIBDW_DWARF_UNWIND option

Libdw unwinding support is present for every architecture that has a
perf_regs.h - perf registers are needed for the initial frame to
unwind.

Elfutils also supports SPARC, ARC and m68k but there is no support in
the Linux kernel for perf registers on these architectures.

As the perf supported DWARF unwinding architectures are a subset of the
elfutils ones, remove NO_LIBDW_DWARF_UNWIND as there isn't a case of
elfutils lacking the support need for perf.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dwarf-regs: Add MIPS perf to DWARF register number mapping functions

Despite an unused function declaration, there was no unwind-libdw for
MIPS but there is a perf_regs.h and a libdw implementation.

Fill in the pieces so hopefully MIPS unwinding with libdw works.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dwarf-regs: Add S390 perf to DWARF register number mapping functions

These functions allow the generic initial register state code in
unwind-libdw to be used.

Now the non-generic code in unwind-libdw has no uses remove it.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dwarf-regs: Add RISC-V perf to DWARF register number mapping functions

These functions allow the generic initial register state code in
unwind-libdw to be used.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dwarf-regs: Add powerpc perf to DWARF register number mapping functions

These functions allow the generic initial register state code in
unwind-libdw to be used.

Note, the link register was being coped to DWARF register 65 that the
SysV ABI spec claims is FPSCR. It is corrected here to 108, but this is
unlikely to matter as FPSCR has little to no impact on unwinding.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dwarf-regs: Add loongarch perf to DWARF register number mapping functions

These functions allow the generic initial register state code in
unwind-libdw to be used.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dwarf-regs: Add csky perf to dwarf register number mapping functions

These functions allow the generic initial register state code in
unwind-libdw to be used.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dwarf-regs: Add ARM perf to dwarf register number mapping functions

These functions allow the generic initial register state code in
unwind-libdw to be used.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dwarf-regs: Add basic get_dwarf_regnum() for most architectures

Add a basic get_dwarf_regnum() implementation for most architectures by
using the get_dwarf_regstr() tables and returning the index of the name
within the table.

Some minor name and constification clean up for csky.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dwarf-regs: Add get_dwarf_regnum_for_perf_regnum() and use for x86 unwinding

Add a utility to map a perf register number to a DWARF register number
for a particular ELF machine type.

Create a generic unwind-libdw initial register initialization routine
that uses this function and thereby avoids arch specific
initialization. The unwind-libdw code does:

1) compute the maximum DWARF register from the set of sampled user registers,
2) allocates a set of DWARF registers,
3) copies the sample registers into the appropriate DWARF registers.

This generic solution is initially implemented for use with x86 as
only get_dwarf_regnum_for_perf_regnum() support for x86 is currently present.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dwarf-regs: Clean up x86 dwarf_regnum code

The i386 and x86-64 register numbers differ on x86, but previously x86
was a single arch string and so this couldn't be handled. The
transition to using ELF EM_* values means we can translate x86
registers correctly for either the x86-64 dwarf register mappings
(from the System V ABI) or i386 register mappings. Correct the mappings.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dwarf-regs: Remove get_arch_regnum()

Except in dwarf-regs the function is never called. The weak function
has no strong arch implementations.

Remove so that the fall-through case applies.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dwarf-regs: Add util/dwarf-regs-arch for consistency with perf-regs

perf_regs.h has cross architecture functions for operating with the
differing perf register constants. dwarf-regs.h is similar but for
cross architecture dwarf notions of registers.

For consistency move the arch parts of dwarf-regs out of util and into
its own directory.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf perf_regs: Switch from arch string to int e_machine

The arch string requires multiple strcmp to identify things like the
IP and SP.

Switch to passing in an e_machine that in the bulk of cases is computed
using a current thread load.

The e_machine also allows identification of 32-bit vs 64-bit processes.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
[ Include dwarf-regs.h to get conditional defines for EM_CSKY and EM_LOONGARCH, not available in old distros ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf powerpc: Unify the skip-callchain-idx libdw with that for addr2line

Rather than have 2 Dwfl unify the Dwfl in skip-callchain-idx with that
is used by libdw__addr2line().

Rename that variable in 'struct dso' from 'a2l_libdw' to just 'libdw' as
it is now used in more than addr2line.

The Dwfl in skip-callchain-idx uses a map address when being read with
dwfl_report_elf (rather than dwfl_report_offline that addr2line
uses).

skip-callchain-idx is wrong as the map address can vary between
processes because of ASLR, ie it should need a different Dwfl per
process.

In the code after this patch the base address becomes 0 and the mapped
PC is used with the dwfl functions.

This should increase the accuracy of skip-callchain-idx, but the impact
has only been build tested.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf unwind-libdw: Correct argument to dwfl_attach_state()

Argument is a pointer but EM_NONE (0) was being passed. Correct by
passing NULL.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf libdw_addr2line: Fixes to srcline memory allocation

Some irregular stack traces are causing double frees and memory
leaks. Make the code robust by proactively freeing and being more
careful with the memory management of the leaf_srcline.

Fixes: 88c51002d06f9a68 ("perf addr2line: Add a libdw implementation")
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf unwind-libdw: Fix a cross-arch unwinding bug

The set_initial_registers field of Dwfl_Thread_Callbacks needs to be set
according to the arch of the stack samples being analyzed, not the arch
that perf itself is built for.

Currently perf fails to unwind stack samples collected from archs
different from that of the host perf is running on.

This patch moves the arch-specific implementations of set_initial_registers
from tools/perf/arch to tools/perf/utli/unwind-libdw-arch, similar to the
way the perf-regs-arch folder contains arch-specific functions related to
registers, and chooses the implementation based on the arch of the data
being processed.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Shimin Guo <shimin.guo@skydio.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf record: Disable inline frames when marking build IDs

Marking DSOs doesn't need inline frames traversing as the inline
frames are all part of the same DSO. Disable to improve performance
and also to avoid potential issues with dwarf information.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dso: Extra validity checks that e_machine is valid

Better ensure a read e_machine is valid by checking the file appears
like an ELF file and the read e_machine value is less than EM_NUM.

This better avoids spurious e_machine values when looking for an
e_machine in say a thread.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf symbol-elf: Fix leak of ELF files with GNU debugdata

The processing of DSO_BINARY_TYPE__GNU_DEBUGDATA in symsrc__init happens
with an open ELF file but the error path only closes the associate fd.

Fix the goto so that the ELF file is also ended and memory released.

Fixes: b10f74308e130527 ("perf symbol: Support .gnu_debugdata for symbols")
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Aditya Bodkhe <aditya.b1@linux.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.ibm.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Guo Ren <guoren@kernel.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Krzysztof Łopatowski <krzysztof.m.lopatowski@gmail.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Wielaard <mark@klomp.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sergei Trofimovich <slyich@gmail.com>
Cc: Shimin Guo <shimin.guo@skydio.com>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf arm-spe: Don't hard code config attribute

Use the config attribute that's published by the driver instead of
hard coding "attr.config".

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf cs-etm: Don't hard code config attribute when configuring the event

These instances of hard coded config attributes are used for configuring
and validating the event options. Use the config attribute that's
published by the driver by replacing the open coded operations with
evsel__get_config_val() and evsel__set_config_if_unset().

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf cs-etm: Don't use hard coded config bits when setting up TRCCONFIGR

Perf only looks at attr.config when determining what was programmed into
TRCCONFIGR. These bits could theoretically be in any of the config
fields. Use the evsel__get_config_val() helper so it's agnostic to
which config field they are in.

The kernel will also stop publishing the TRCCONFIGR register bits in a
header [1] so preempt that by defining them here.

[1]: https://lore.kernel.org/linux-arm-kernel/20251128-james-cs-syncfreq-v8-10-4d319764cc58@linaro.org/

Reviewed-by: Ian Rogers <irogers@google.com>
Reviewed-by: Leo Yan <leo.yan@arm.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf cs-etm: Don't use hard coded config bits when setting up ETMCR

Perf only looks at attr.config when determining what was programmed into
ETMCR. These bits could theoretically be in any of the config fields.
Add a generic helper to find the value of any named format field in any
config field and then use it to get the attributes relevant to ETMCR.

The kernel will also stop publishing the ETMCR register bits in a header
[1] so preempt that by defining them here.

Move field_prep() to util.h so we can define it along side field_get().
Unfortunately FIELD_PREP() and FIELD_GET() from the kernel can't be used
as they require the mask to be a compile time constant.

[1]: https://lore.kernel.org/linux-arm-kernel/20251128-james-cs-syncfreq-v8-10-4d319764cc58@linaro.org/

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf cs-etm: Make a helper to find the Coresight evsel

This pattern occurs a few times and we'll add another one later, so add
a helper function for it.

Reviewed-by: Ian Rogers <irogers@google.com>
Reviewed-by: Leo Yan <leo.yan@arm.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tests: Test evsel__set_config_if_unset() and config change tracking

Test that evsel__set_config_if_unset() behaves as expected. This also
tests the user config change tracking mechanism as it depends on it.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf parse-events: Always track user config changes

Requiring the 'pmu->perf_event_attr_init_default' callback to be set to
track user changes is a bit of a trap to fall in. It's hard to see that
this is required when depending on the user change tracking.

It's possible to want all 0 defaults so not set it, but at the same time
still do some programmatic setting of configs with
evsel__set_config_if_unset(). Also if a PMU reverts to 0 defaults and
deletes its existing callback, it will silently break existing uses of
evsel__set_config_if_unset().

One way to fix this would be to assert in evsel__set_config_if_unset()
if the changes weren't tracked, but that would be a possibly untested
runtime failure. Instead, always track it as it's harmless and
simplifies testing too.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf evsel: Add a helper to get the value of a config field

This will be used by aux PMUs to read an already written value for
configuring their events and for also testing.

Its helper perf_pmu__format_unpack() does the opposite of the existing
pmu_format_value() so rename that one to perf_pmu__format_pack() so it's
clear how they are related.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf evsel: apply evsel__set_config_if_unset() to all config fields

Misleadingly, evsel__set_config_if_unset() only works with the config
field and not config1, config2, etc. This is fine at the moment because
all users of it happen to operate on bits that are in that config field.
Fix it before there are any new users of the function which operate on
bits in different config fields.

In theory it's also possible for a driver to move an existing bit to
another config field and this fixes that scenario too, although this
hasn't happened yet either.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf parse-events: Track all user changed config bits

Currently we only track which bits were set by the user in attr->config.
But all configN fields should be treated equally as they can all have
default and user overridden values.

Track them all by making get_config_chgs() generic and calling it once
for each config value.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf evsel: Support sparse fields in evsel__set_config_if_unset()

Sparse config fields are technically supported although currently
unused. field_prep() only works for contiguous bitfields so replace it
with pmu_format_value().

pmu_format_value() also takes a bitmap rather than a u64 so replace
'u64 bits' with format->bits.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf evsel: Move evsel__* functions to evsel.c

At least one of these were put here to avoid a Python binding linking
issue which is no longer present. Put them back in their correct
location to avoid confusion about which file to add a new evsel__*
function to later.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/all/ZEbAS2yx2fguW60w@kernel.org/
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf evsel: Refactor evsel__set_config_if_unset() arguments

Make the evsel argument first to match the other evsel__* functions
and remove the redundant pmu argument, which can be accessed via evsel.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf parse-events: Refactor get_config_terms() to remove macros

The ADD_CONFIG_TERM() macros build the __type argument out of a partial
EVSEL__CONFIG_TERM_x enum name. This means that they can't be called
from a function where __type is a variable and it's also impossible to
grep the codebase to find usages of these enums as they're never typed
in full.

Fix this by removing the macros and replacing them with an
add_config_term() function. It seems the main reason these existed in
the first place was to avoid type punning and to write to a specific
field in the union, but the same thing can be achieved with a single
write to a u64 'val' field.

Running the Perf tests with "-fsanitize=undefined -fno-sanitize-recover"
results in no new issues as a result of this change.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Switch printf("...%s", strerror(errno)) to printf("...%m")

strerror() has thread safety issues, strerror_r() requires stack
allocated buffers.

Code in perf has already been using the "%m" formatting flag that is a
widely support glibc extension to print the current errno's description.

Expand the usage of this formatting flag and remove usage of
strerror()/strerror_r().

Signed-off-by: Ian Rogers <irogers@google.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexandre Ghiti <alexghiti@rivosinc.com>
Cc: Blake Jones <blakejones@google.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Dr. David Alan Gilbert <linux@treblig.org>
Cc: Haibo Xu <haibo1.xu@intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Yunseong Kim <ysk@kzalloc.com>
Cc: Zhongqiu Han <quic_zhonhan@quicinc.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf help: Move common_cmds into builtin-help

There's a lot of infrastructure for generating a relatively simple
array used by one function.

Move the array into the function and remove the supporting build logic.

At the same time opportunistically const-ify the array.

Signed-off-by: Ian Rogers <irogers@google.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Charlie Jenkins <charlie@rivosinc.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events amd: Add Zen 6 metrics

Add metrics taken from Section 1.2 "Performance Measurement" of the
Performance Monitor Counters for AMD Family 1Ah Model 50h-57h Processors
document available at the link below.

The recommended metrics are sourced from Table 1 "Guidance for Common
Performance Statistics with Complex Event Selects".

The pipeline utilization metrics are sourced from Table 2 "Guidance
for Pipeline Utilization Analysis Statistics". These are useful for
finding performance bottlenecks by analyzing activity at different
stages of the pipeline. There are metric groups available for Level 1
and Level 2 analysis.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Sandipan Das <sandipan.das@amd.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ananth Narayan <ananth.narayan@amd.com>
Cc: Caleb Biggers <caleb.biggers@intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Stephane Eranian <eranian@google.com>
Link: https://bugzilla.kernel.org/attachment.cgi?id=309149
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events amd: Add Zen 6 uncore events

Add uncore events taken from Section 1.6 "L3 Cache Performance Monitor
Counters" and Section 2.2 "UMC Performance Monitor Events" of the
Performance Monitor Counters for AMD Family 1Ah Model 50h-57h Processors
document available at the link below.

This constitutes events which capture L3 cache and UMC command activity.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Sandipan Das <sandipan.das@amd.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ananth Narayan <ananth.narayan@amd.com>
Cc: Caleb Biggers <caleb.biggers@intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Stephane Eranian <eranian@google.com>
Link: https://bugzilla.kernel.org/attachment.cgi?id=309149
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events amd: Add Zen 6 core events

Add core events taken from Section 1.5 "Core Performance Monitor
Counters" of the Performance Monitor Counters for AMD Family 1Ah Model
50h-57h Processors document available at the link below.

This constitutes events which capture information on op dispatch,
execution and retirement, branch prediction, L1 and L2 cache activity,
TLB activity, etc.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Sandipan Das <sandipan.das@amd.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ananth Narayan <ananth.narayan@amd.com>
Cc: Caleb Biggers <caleb.biggers@intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Stephane Eranian <eranian@google.com>
Link: https://bugzilla.kernel.org/attachment.cgi?id=309149
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events amd: Add Zen 6 mapping

Add a regular expression in the map file so that appropriate JSON event
files are used for AMD Zen 6 processors. Restrict the regular expression
for AMD Zen 5 processors to known model ranges since they also belong to
Family 1Ah.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Sandipan Das <sandipan.das@amd.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ananth Narayan <ananth.narayan@amd.com>
Cc: Caleb Biggers <caleb.biggers@intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Stephane Eranian <eranian@google.com>
[ Moved this one to the front of the series to keep the tree bisectable, as per Ian Rogers suggestion ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events riscv: Add CVA6 JSON file

This patch add the OpenHW Core-V CVA6 Risc-V JSON file.

For more info:

https://openhwfoundation.org/news/2023/11/07/openhw-group-announces-core-v-cva6-platform-project-for-risc-v-software-development-and-testing/

Signed-off-by: Manuel Hernández Méndez <manuel.hernandez@openchip.com>
Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf addr_location: Update outdated comment

The function addr_location__put() was renamed addr_location__exit() in
commit 0dd5041c9a0eaf8c ("perf addr_location: Add init/exit/copy
functions").  Make the comment preceding the function consistent with
the function itself.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Julia Lawall <Julia.Lawall@inria.fr>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kexin Sun <kexinsun@smail.nju.edu.cn>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ratnadira Widyasari <ratnadiraw@smu.edu.sg>
Cc: Xutong Ma <xutong.ma@inria.fr>
Cc: Yumbo Lyu <yunbolyu@smu.edu.sg>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Dump callchain context marker names

These are hard to interpret in the raw output because they are printed
as hex but are defined in perf_event.h as decimal. Make it much easier
to read the raw callchains by just printing their names.

For example:

  $ perf report -D

  1798195372321 0x4638 [0xb0]: PERF_RECORD_SAMPLE(IP, 0x4002): 44922/44922: 0x7c8046dd3400 period: 120218 addr: 0
  ... FP chain: nr:12
  .....  0: fffffffffffffe00 (PERF_CONTEXT_USER)
  .....  1: 00007c8046dd3400
  .....  2: 00007c8046db86d3

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
[ Add PERF_CONTEXT_USER_DEFERRED too, as per Namhyung's review comment ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events arm64: Remove uncountable events

These events are never countable by the PMU and are only intended to
be used as external inputs to trace. Therefore showing them in 'perf
list' is misleading so remove them.

The generator script doesn't emit these events when used with the new
telemetry-solution input files [1].

'perf list' should only show countable events because there are events
that are sometimes implemented, sometimes countable and sometimes not,
for example TRB_TRIG. If we always include any implemented events
whether they are countable or not then it's not possible to tell whether
they are usable in perf without going to the docs, defeating the point
of 'perf list'.

It's also not useful yet to display implemented events that are not
countable (for help in using trace rather than perf stat), because
PMU_OVFS and PMU_HOVFS are practically always implemented and TRB_TRIG
is always implemented when there is TRBE.

[1]: https://gitlab.arm.com/telemetry-solution/telemetry-solution/-/tree/main/data/pmu/cpu

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: James Clark <james.clark@linaro.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Akio Kakuno <fj3333bs@aa.jp.fujitsu.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Leo Yan <leo.yan@linux.dev>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will@kernel.org>
Cc: Yoshihiro Furudera <fj5100bi@fujitsu.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf Documentation: Correct branch stack sampling call-stack option

The correct call-stack option for branch stack sampling should be "stack"
instead of "call_stack". Correct it.

$perf record -e instructions -j call_stack -- sleep 1
unknown branch filter call_stack, check man page

 Usage: perf record [<options>] [<command>]
    or: perf record [<options>] -- <command> [<options>]

    -j, --branch-filter <branch filter mask>
                          branch stack filter modes

Fixes: 955f6def5590ce6c ("perf record: Add remaining branch filters: "no_cycles", "no_flags" & "hw_index"")
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Dapeng Mi <dapeng1.mi@linux.intel.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Falcon <thomas.falcon@intel.com>
Cc: Xudong Hao <xudong.hao@intel.com>
Cc: Zide Chen <zide.chen@intel.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf test: Do not skip when some metric-group tests succeed

I think the return value of SKIP (2) should be used when it skipped the
entire test suite rather than a few of them.  While the FAIL should be
reserved if any of test failed.

  $ perf test -vv 109
  109: perf all metricgroups test:
  --- start ---
  test child forked, pid 2493003
  Testing Backend
  Testing Bad
  Testing BadSpec
  Testing BigFootprint
  Testing BrMispredicts
  Testing Branches
  Testing BvBC
  Testing BvBO
  Testing BvCB
  Testing BvFB
  Testing BvIO
  Testing BvMB
  Testing BvML
  Testing BvMP
  Testing BvMS
  Testing BvMT
  Testing BvOB
  Testing BvUW
  Testing CacheHits
  Testing CacheMisses
  Testing CodeGen
  Testing Compute
  Testing Cor
  Testing DSB
  Testing DSBmiss
  Testing DataSharing
  Testing Default
  Testing Default2
  Testing Default3
  Testing Default4
  Ignoring failures in Default4 that may contain unsupported legacy events
  Testing Fed
  Testing FetchBW
  Testing FetchLat
  Testing Flops
  Testing FpScalar
  Testing FpVector
  Testing Frontend
  Testing HPC
  Testing IcMiss
  Testing InsType
  Testing LSD
  Testing LockCont
  Testing MachineClears
  Testing Machine_Clears
  Testing Mem
  Testing MemOffcore
  Testing MemoryBW
  Testing MemoryBound
  Testing MemoryLat
  Testing MemoryTLB
  Testing Memory_BW
  Testing Memory_Lat
  Testing MicroSeq
  Testing OS
  Testing Offcore
  Testing PGO
  Testing Pipeline
  Testing PortsUtil
  Testing Power
  Testing Prefetches
  Testing Ret
  Testing Retire
  Testing SMT
  Testing Snoop
  Testing SoC
  Testing Summary
  Testing TmaL1
  Testing TmaL2
  Testing TmaL3mem
  Testing TopdownL1
  Testing TopdownL2
  Testing TopdownL3
  Testing TopdownL4
  Testing TopdownL5
  Testing TopdownL6
  Testing smi
  Testing tma_L1_group
  Testing tma_L2_group
  Testing tma_L3_group
  Testing tma_L4_group
  Testing tma_L5_group
  Testing tma_L6_group
  Testing tma_alu_op_utilization_group
  Testing tma_assists_group
  Testing tma_backend_bound_group
  Testing tma_bad_speculation_group
  Testing tma_branch_mispredicts_group
  Testing tma_branch_resteers_group
  Testing tma_code_stlb_miss_group
  Testing tma_core_bound_group
  Testing tma_divider_group
  Testing tma_dram_bound_group
  Testing tma_dtlb_load_group
  Testing tma_dtlb_store_group
  Testing tma_fetch_bandwidth_group
  Testing tma_fetch_latency_group
  Testing tma_fp_arith_group
  Testing tma_fp_vector_group
  Testing tma_frontend_bound_group
  Testing tma_heavy_operations_group
  Testing tma_icache_misses_group
  Testing tma_issue2P
  Testing tma_issueBM
  Testing tma_issueBW
  Testing tma_issueComp
  Testing tma_issueD0
  Testing tma_issueFB
  Testing tma_issueFL
  Testing tma_issueL1
  Testing tma_issueLat
  Testing tma_issueMC
  Testing tma_issueMS
  Testing tma_issueMV
  Testing tma_issueRFO
  Testing tma_issueSL
  Testing tma_issueSO
  Testing tma_issueSmSt
  Testing tma_issueSpSt
  Testing tma_issueSyncxn
  Testing tma_issueTLB
  Testing tma_itlb_misses_group
  Testing tma_l1_bound_group
  Testing tma_l2_bound_group
  Testing tma_l3_bound_group
  Testing tma_light_operations_group
  Testing tma_load_stlb_miss_group
  Testing tma_machine_clears_group
  Testing tma_memory_bound_group
  Testing tma_microcode_sequencer_group
  Testing tma_mite_group
  Testing tma_other_light_ops_group
  Testing tma_ports_utilization_group
  Testing tma_ports_utilized_0_group
  Testing tma_ports_utilized_3m_group
  Testing tma_retiring_group
  Testing tma_serializing_operation_group
  Testing tma_store_bound_group
  Testing tma_store_stlb_miss_group
  Testing transaction
  ---- end(0) ----
  109: perf all metricgroups test                                      : Ok

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf test: Do not skip when some metrics tests succeeded

I think the return value of SKIP (2) should be used when it skipped the
entire test suite rather than a few of them.  While the FAIL should be
reserved if any of test failed.

  $ perf test -vv 110
  110: perf all metrics test:
  --- start ---
  test child forked, pid 2496399
  Testing tma_core_bound
  Testing tma_info_core_ilp
  Testing tma_info_memory_l2mpki
  Testing tma_memory_bound
  Testing tma_bottleneck_irregular_overhead
  Testing tma_bottleneck_mispredictions
  Testing tma_info_bad_spec_branch_misprediction_cost
  Testing tma_info_bad_spec_ipmisp_cond_ntaken
  Testing tma_info_bad_spec_ipmisp_cond_taken
  Testing tma_info_bad_spec_ipmisp_indirect
  Testing tma_info_bad_spec_ipmisp_ret
  Testing tma_info_bad_spec_ipmispredict
  Testing tma_info_branches_callret
  Testing tma_info_branches_cond_nt
  Testing tma_info_branches_cond_tk
  Testing tma_info_branches_jump
  Testing tma_info_branches_other_branches
  Testing tma_branch_mispredicts
  Testing tma_clears_resteers
  Testing tma_machine_clears
  Testing tma_mispredicts_resteers
  Testing tma_bottleneck_big_code
  Testing tma_icache_misses
  Testing tma_itlb_misses
  Testing tma_unknown_branches
  Testing tma_info_bad_spec_spec_clears_ratio
  Testing tma_other_mispredicts
  Testing tma_branch_instructions
  Testing tma_info_frontend_tbpc
  Testing tma_info_inst_mix_bptkbranch
  Testing tma_info_inst_mix_ipbranch
  Testing tma_info_inst_mix_ipcall
  Testing tma_info_inst_mix_iptb
  Testing tma_info_system_ipfarbranch
  Testing tma_info_thread_uptb
  Testing tma_bottleneck_branching_overhead
  Testing tma_nop_instructions
  Testing tma_bottleneck_compute_bound_est
  Testing tma_divider
  Testing tma_ports_utilized_3m
  Testing tma_bottleneck_instruction_fetch_bw
  Testing tma_frontend_bound
  Testing tma_assists
  Testing tma_other_nukes
  Testing tma_serializing_operation
  Testing tma_bottleneck_data_cache_memory_bandwidth
  Testing tma_fb_full
  Testing tma_mem_bandwidth
  Testing tma_sq_full
  Testing tma_bottleneck_data_cache_memory_latency
  Testing tma_l1_latency_dependency
  Testing tma_l2_bound
  Testing tma_l3_hit_latency
  Testing tma_mem_latency
  Testing tma_store_latency
  Testing tma_bottleneck_memory_synchronization
  Testing tma_contested_accesses
  Testing tma_data_sharing
  Testing tma_false_sharing
  Testing tma_bottleneck_memory_data_tlbs
  Testing tma_dtlb_load
  Testing tma_dtlb_store
  Testing tma_backend_bound
  Testing tma_bottleneck_other_bottlenecks
  Testing tma_bottleneck_useful_work
  Testing tma_retiring
  Testing tma_info_memory_fb_hpki
  Testing tma_info_memory_l1mpki
  Testing tma_info_memory_l1mpki_load
  Testing tma_info_memory_l2hpki_all
  Testing tma_info_memory_l2hpki_load
  Testing tma_info_memory_l2mpki_all
  Testing tma_info_memory_l2mpki_load
  Testing tma_l1_bound
  Testing tma_l3_bound
  Testing tma_info_memory_l2mpki_rfo
  Testing tma_fp_scalar
  Testing tma_fp_vector
  Testing tma_fp_vector_128b
  Testing tma_fp_vector_256b
  Testing tma_fp_vector_512b
  Testing tma_port_0
  Testing tma_x87_use
  Testing tma_info_botlnk_l0_core_bound_likely
  Testing tma_info_core_fp_arith_utilization
  Testing tma_info_pipeline_execute
  Testing tma_info_system_gflops
  Testing tma_info_thread_execute_per_issue
  Testing tma_dsb
  Testing tma_info_botlnk_l2_dsb_bandwidth
  Testing tma_info_frontend_dsb_coverage
  Testing tma_decoder0_alone
  Testing tma_dsb_switches
  Testing tma_info_botlnk_l2_dsb_misses
  Testing tma_info_frontend_dsb_switch_cost
  Testing tma_info_frontend_ipdsb_miss_ret
  Testing tma_mite
  Testing tma_mite_4wide
  Testing CPUs_utilized
  Testing backend_cycles_idle
  [Ignored backend_cycles_idle] failed but as a Default metric this can be expected
  Performance counter stats for 'perf test -w noploop': <not counted> cpu-cycles:u <not supported> stalled-cycles-backend:u 1.014051473 seconds time elapsed 1.005718000 seconds user 0.008013000 seconds sys
  Testing branch_frequency
  Testing branch_miss_rate
  Testing cs_per_second
  Testing cycles_frequency
  Testing frontend_cycles_idle
  [Ignored frontend_cycles_idle] failed but as a Default metric this can be expected
  Performance counter stats for 'perf test -w noploop': <not counted> cpu-cycles:u <not supported> stalled-cycles-frontend:u 1.012813656 seconds time elapsed 1.004603000 seconds user 0.008004000 seconds sys
  Testing insn_per_cycle
  Testing migrations_per_second
  Testing page_faults_per_second
  Testing stalled_cycles_per_instruction
  [Ignored stalled_cycles_per_instruction] failed but as a Default metric this can be expected
  Error: No supported events found. The stalled-cycles-backend:u event is not supported.
  Testing tma_bad_speculation
  Testing l1d_miss_rate
  Testing llc_miss_rate
  Testing dtlb_miss_rate
  Testing itlb_miss_rate
  [Ignored itlb_miss_rate] failed but as a Default metric this can be expected
  Performance counter stats for 'perf test -w noploop': <not supported> iTLB-loads:u 3,097 iTLB-load-misses:u 1.012766732 seconds time elapsed 1.004318000 seconds user 0.008002000 seconds sys
  Testing l1i_miss_rate
  [Ignored l1i_miss_rate] failed but as a Default metric this can be expected
  Performance counter stats for 'perf test -w noploop': <not counted> L1-icache-load-misses:u <not supported> L1-icache-loads:u 1.013606395 seconds time elapsed 1.001371000 seconds user 0.011968000 seconds sys
  Testing l1_prefetch_miss_rate
  [Ignored l1_prefetch_miss_rate] failed but as a Default metric this can be expected
  Error: No supported events found. The L1-dcache-prefetches:u event is not supported.
  Testing tma_info_botlnk_l2_ic_misses
  Testing tma_info_frontend_fetch_upc
  Testing tma_info_frontend_icache_miss_latency
  Testing tma_info_frontend_ipunknown_branch
  Testing tma_info_frontend_lsd_coverage
  Testing tma_info_memory_tlb_code_stlb_mpki
  Testing tma_info_pipeline_fetch_dsb
  Testing tma_info_pipeline_fetch_lsd
  Testing tma_info_pipeline_fetch_mite
  Testing tma_info_pipeline_fetch_ms
  Testing tma_fetch_bandwidth
  Testing tma_lsd
  Testing tma_branch_resteers
  Testing tma_code_l2_hit
  Testing tma_code_l2_miss
  Testing tma_code_stlb_hit
  Testing tma_code_stlb_miss
  Testing tma_code_stlb_miss_2m
  Testing tma_code_stlb_miss_4k
  Testing tma_lcp
  Testing tma_ms_switches
  Testing tma_info_core_flopc
  Testing tma_info_inst_mix_iparith
  Testing tma_info_inst_mix_iparith_avx128
  Testing tma_info_inst_mix_iparith_avx256
  Testing tma_info_inst_mix_iparith_avx512
  Testing tma_info_inst_mix_iparith_scalar_dp
  Testing tma_info_inst_mix_iparith_scalar_sp
  Testing tma_info_inst_mix_ipflop
  Testing tma_info_inst_mix_ippause
  Testing tma_fetch_latency
  Testing tma_fp_arith
  Testing tma_fp_assists
  Testing tma_info_system_cpu_utilization
  Testing tma_info_system_dram_bw_use
  [Skipped tma_info_system_dram_bw_use] Not supported events
  Performance counter stats for 'perf test -w noploop': <not supported> UNC_ARB_TRK_REQUESTS.ALL:u <not supported> UNC_ARB_COH_TRK_REQUESTS.ALL:u 1,013,554,749 duration_time 1.013527265 seconds time elapsed 1.005417000 seconds user 0.008011000 seconds sys
  Testing tma_info_frontend_l2mpki_code
  Testing tma_info_frontend_l2mpki_code_all
  Testing tma_info_inst_mix_ipload
  Testing tma_info_inst_mix_ipstore
  Testing tma_info_memory_latency_load_l2_miss_latency
  Testing tma_lock_latency
  Testing tma_info_memory_core_l1d_cache_fill_bw_2t
  Testing tma_info_memory_core_l2_cache_fill_bw_2t
  Testing tma_info_memory_core_l3_cache_access_bw_2t
  Testing tma_info_memory_core_l3_cache_fill_bw_2t
  Testing tma_info_memory_l1d_cache_fill_bw
  Testing tma_info_memory_l2_cache_fill_bw
  Testing tma_info_memory_l3_cache_access_bw
  Testing tma_info_memory_l3_cache_fill_bw
  Testing tma_info_memory_l3mpki
  Testing tma_info_memory_load_miss_real_latency
  Testing tma_info_memory_mix_bus_lock_pki
  Testing tma_info_memory_mix_uc_load_pki
  Testing tma_info_memory_mlp
  Testing tma_info_memory_tlb_load_stlb_mpki
  Testing tma_info_memory_tlb_page_walks_utilization
  Testing tma_info_memory_tlb_store_stlb_mpki
  Testing tma_info_system_mem_parallel_reads
  [Skipped tma_info_system_mem_parallel_reads] Not supported events
  Performance counter stats for 'perf test -w noploop': <not supported> UNC_ARB_DAT_OCCUPANCY.RD:u <not counted> UNC_ARB_DAT_OCCUPANCY.RD/cmask=1/ 1.013354884 seconds time elapsed 1.009239000 seconds user 0.004004000 seconds sys
  Testing tma_info_system_mem_read_latency
  [Skipped tma_info_system_mem_read_latency] Not supported events
  Performance counter stats for 'perf test -w noploop': <not supported> UNC_ARB_DAT_OCCUPANCY.RD:u <not counted> UNC_ARB_TRK_OCCUPANCY.RD <not counted> UNC_ARB_TRK_REQUESTS.RD 1.012882143 seconds time elapsed 1.004600000 seconds user 0.008036000 seconds sys
  Testing tma_info_thread_cpi
  Testing tma_streaming_stores
  Testing tma_dram_bound
  Testing tma_store_bound
  Testing tma_l2_hit_latency
  Testing tma_load_stlb_hit
  Testing tma_load_stlb_miss
  Testing tma_load_stlb_miss_1g
  Testing tma_load_stlb_miss_2m
  Testing tma_load_stlb_miss_4k
  Testing tma_store_stlb_hit
  Testing tma_store_stlb_miss
  Testing tma_store_stlb_miss_1g
  Testing tma_store_stlb_miss_2m
  Testing tma_store_stlb_miss_4k
  Testing tma_info_memory_latency_data_l2_mlp
  Testing tma_info_memory_latency_load_l2_mlp
  Testing tma_info_pipeline_ipassist
  Testing tma_microcode_sequencer
  Testing tma_ms
  Testing tma_info_system_kernel_cpi
  [Failed tma_info_system_kernel_cpi] Metric contains missing events
  Error: No supported events found. Access to performance monitoring and observability operations is limited. Consider adjusting /proc/sys/kernel/perf_event_paranoid setting to open access to performance monitoring and observability operations for processes without CAP_PERFMON, CAP_SYS_PTRACE or CAP_SYS_ADMIN Linux capability. More information can be found at 'Perf events and tool security' document: https://www.kernel.org/doc/html/latest/admin-guide/perf-security.html perf_event_paranoid setting is 2: -1: Allow use of (almost) all events by all users Ignore mlock limit after perf_event_mlock_kb without CAP_IPC_LOCK >= 0: Disallow raw and ftrace function tracepoint access >= 1: Disallow CPU event access >= 2: Disallow kernel profiling To make the adjusted perf_event_paranoid setting permanent preserve it in /etc/sysctl.conf (e.g. kernel.perf_event_paranoid = <setting>)
  Testing tma_info_system_kernel_utilization
  [Failed tma_info_system_kernel_utilization] Metric contains missing events
  Error: No supported events found. Access to performance monitoring and observability operations is limited. Consider adjusting /proc/sys/kernel/perf_event_paranoid setting to open access to performance monitoring and observability operations for processes without CAP_PERFMON, CAP_SYS_PTRACE or CAP_SYS_ADMIN Linux capability. More information can be found at 'Perf events and tool security' document: https://www.kernel.org/doc/html/latest/admin-guide/perf-security.html perf_event_paranoid setting is 2: -1: Allow use of (almost) all events by all users Ignore mlock limit after perf_event_mlock_kb without CAP_IPC_LOCK >= 0: Disallow raw and ftrace function tracepoint access >= 1: Disallow CPU event access >= 2: Disallow kernel profiling To make the adjusted perf_event_paranoid setting permanent preserve it in /etc/sysctl.conf (e.g. kernel.perf_event_paranoid = <setting>)
  Testing tma_info_pipeline_retire
  Testing tma_info_thread_clks
  Testing tma_info_thread_uoppi
  Testing tma_memory_operations
  Testing tma_other_light_ops
  Testing tma_ports_utilization
  Testing tma_ports_utilized_0
  Testing tma_ports_utilized_1
  Testing tma_ports_utilized_2
  Testing C10_Pkg_Residency
  [Failed C10_Pkg_Residency] Metric contains missing events
  WARNING: grouped events cpus do not match. Events with CPUs not matching the leader will be removed from the group. anon group { cstate_pkg/c10-residency/, msr/tsc/ } Error: No supported events found. Invalid event (cstate_pkg/c10-residency/u) in per-thread mode, enable system wide with '-a'.
  Testing C2_Pkg_Residency
  [Failed C2_Pkg_Residency] Metric contains missing events
  WARNING: grouped events cpus do not match. Events with CPUs not matching the leader will be removed from the group. anon group { cstate_pkg/c2-residency/, msr/tsc/ } Error: No supported events found. Invalid event (cstate_pkg/c2-residency/u) in per-thread mode, enable system wide with '-a'.
  Testing C3_Pkg_Residency
  [Failed C3_Pkg_Residency] Metric contains missing events
  WARNING: grouped events cpus do not match. Events with CPUs not matching the leader will be removed from the group. anon group { msr/tsc/, cstate_pkg/c3-residency/ } Error: No supported events found. Invalid event (msr/tsc/u) in per-thread mode, enable system wide with '-a'.
  Testing C6_Core_Residency
  [Failed C6_Core_Residency] Metric contains missing events
  WARNING: grouped events cpus do not match. Events with CPUs not matching the leader will be removed from the group. anon group { cstate_core/c6-residency/, msr/tsc/ } Error: No supported events found. Invalid event (cstate_core/c6-residency/u) in per-thread mode, enable system wide with '-a'.
  Testing C6_Pkg_Residency
  [Failed C6_Pkg_Residency] Metric contains missing events
  WARNING: grouped events cpus do not match. Events with CPUs not matching the leader will be removed from the group. anon group { cstate_pkg/c6-residency/, msr/tsc/ } Error: No supported events found. Invalid event (cstate_pkg/c6-residency/u) in per-thread mode, enable system wide with '-a'.
  Testing C7_Core_Residency
  [Failed C7_Core_Residency] Metric contains missing events
  WARNING: grouped events cpus do not match. Events with CPUs not matching the leader will be removed from the group. anon group { cstate_core/c7-residency/, msr/tsc/ } Error: No supported events found. Invalid event (cstate_core/c7-residency/u) in per-thread mode, enable system wide with '-a'.
  Testing C7_Pkg_Residency
  [Failed C7_Pkg_Residency] Metric contains missing events
  WARNING: grouped events cpus do not match. Events with CPUs not matching the leader will be removed from the group. anon group { cstate_pkg/c7-residency/, msr/tsc/ } Error: No supported events found. Invalid event (cstate_pkg/c7-residency/u) in per-thread mode, enable system wide with '-a'.
  Testing C8_Pkg_Residency
  [Failed C8_Pkg_Residency] Metric contains missing events
  WARNING: grouped events cpus do not match. Events with CPUs not matching the leader will be removed from the group. anon group { cstate_pkg/c8-residency/, msr/tsc/ } Error: No supported events found. Invalid event (cstate_pkg/c8-residency/u) in per-thread mode, enable system wide with '-a'.
  Testing C9_Pkg_Residency
  [Failed C9_Pkg_Residency] Metric contains missing events
  WARNING: grouped events cpus do not match. Events with CPUs not matching the leader will be removed from the group. anon group { cstate_pkg/c9-residency/, msr/tsc/ } Error: No supported events found. Invalid event (cstate_pkg/c9-residency/u) in per-thread mode, enable system wide with '-a'.
  Testing tma_info_core_epc
  Testing tma_info_system_core_frequency
  Testing tma_info_system_power
  [Skipped tma_info_system_power] Not supported events
  Performance counter stats for 'perf test -w noploop': <not supported> Joules power/energy-pkg/u 1,013,238,256 duration_time 1.013223072 seconds time elapsed 0.995924000 seconds user 0.011903000 seconds sys
  Testing tma_info_system_power_license0_utilization
  Testing tma_info_system_power_license1_utilization
  Testing tma_info_system_power_license2_utilization
  Testing tma_info_system_turbo_utilization
  Testing tma_info_inst_mix_ipswpf
  Testing tma_info_memory_prefetches_useless_hwpf
  Testing tma_info_core_coreipc
  Testing tma_info_thread_ipc
  Testing tma_heavy_operations
  Testing tma_light_operations
  Testing tma_info_core_core_clks
  Testing tma_info_system_smt_2t_utilization
  Testing tma_info_thread_slots_utilization
  Testing UNCORE_FREQ
  [Skipped UNCORE_FREQ] Not supported events
  Performance counter stats for 'perf test -w noploop': <not supported> UNC_CLOCK.SOCKET:u 1,015,993,466 duration_time 1.015949387 seconds time elapsed 1.007676000 seconds user 0.008029000 seconds sys
  Testing tma_info_system_socket_clks
  [Failed tma_info_system_socket_clks] Metric contains missing events
  Error: No supported events found. Invalid event (UNC_CLOCK.SOCKET:u) in per-thread mode, enable system wide with '-a'.
  Testing tma_info_inst_mix_instructions
  Testing tma_info_system_cpus_utilized
  Testing tma_info_system_mux
  Testing tma_info_system_time
  Testing tma_info_thread_slots
  Testing tma_few_uops_instructions
  Testing tma_4k_aliasing
  Testing tma_cisc
  Testing tma_fp_divider
  Testing tma_int_divider
  Testing tma_slow_pause
  Testing tma_split_loads
  Testing tma_split_stores
  Testing tma_store_fwd_blk
  Testing tma_alu_op_utilization
  Testing tma_load_op_utilization
  Testing tma_mixing_vectors
  Testing tma_store_op_utilization
  Testing tma_port_1
  Testing tma_port_5
  Testing tma_port_6
  Testing smi_cycles
  [Skipped smi_cycles] Not supported events
  Performance counter stats for 'perf test -w noploop': <not supported> msr/smi/u <not supported> msr/aperf/u 3,965,789,327 cycles:u 1.012779591 seconds time elapsed 1.004579000 seconds user 0.007972000 seconds sys
  Testing smi_num
  [Failed smi_num] Metric contains missing events
  Error: No supported events found. Invalid event (msr/smi/u) in per-thread mode, enable system wide with '-a'.
  Testing tsx_aborted_cycles
  Testing tsx_cycles_per_elision
  Testing tsx_cycles_per_transaction
  Testing tsx_transactional_cycles
  ---- end(-1) ----
  110: perf all metrics test                                           : FAILED!

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf test: Use shelldir to refer perf source location

It uses tools/perf/include which assumes it's running from the root of
the linux kernel source tree.  But you can run perf from other places
like tools/perf, then the include path won't match.  We can use the
shelldir variable to locate the test script in the tree.

  $ cd tools/perf

  $ ./perf test dlfilter
   63: dlfilter C API                                                  : Ok
  101: perf script --dlfilter tests                                    : Ok

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf test: Skip dlfilter test for build failures

For some reason, it may fail to build the dlfilter.  Let's skip the test
as it's not an error in the perf.  This can happen when you run the perf
test without source code or in a different directory.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf inject: Keep build-ID data if no option is used

The keep_feat() determines which header features will be kept or
discarded.  Usually 'perf inject' will add build-IDs based on -b, -B or
other related options.  But it lose build-ID when none of those options
are used.  This is meaningful only when --buildid-mmap is not used.

The following example shows the impact of this change.

  $ perf record --no-buildid-mmap true
  [ perf record: Woken up 1 times to write data ]
  [ perf record: Captured and wrote 0.037 MB perf.data (5 samples) ]

  $ perf inject -i perf.data -o perf.data.inject

  $ perf buildid-list -i perf.data
  08cccc2a9388d5247ccb3e864f3063b975b0a15d /usr/lib/x86_64-linux-gnu/ld-linux-x86-64.so.2
  fd5c4d5673256cd6bda51725dba048dabb0f854e [kernel.kallsyms]
  97a36ce1140071be5c36b147fa0bed173e05a602 [vdso]

  $ perf buildid-list -i perf.data.inject
  97a36ce1140071be5c36b147fa0bed173e05a602 [vdso]

With this change, perf.data.inject would show the same list (of course,
you need to run perf inject again).

Reported-by: Gabriel Marin <gmx@google.com>
Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf util: Remove SHA-1 code

Now that the SHA-1 code is no longer used, remove it.

Signed-off-by: Eric Biggers <ebiggers@kernel.org>
Tested-by: Ian Rogers <irogers@google.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Fangrui Song <maskray@sourceware.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jason A. Donenfeld <Jason@zx2c4.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Pablo Galindo <pablogsal@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf genelf: Switch from SHA-1 to BLAKE2s for build ID generation

Recent patches [1] [2] added an implementation of SHA-1 to perf and made
it be used for build ID generation.

I had understood the choice of SHA-1, which is a legacy algorithm, to be
for backwards compatibility.

It turns out, though, that there's no backwards compatibility
requirement here other than the size of the build ID field, which is
fixed at 20 bytes.  Not only did the hash algorithm already change (from
MD5 to SHA-1), but the inputs to the hash changed too: from 'load_addr
|| code' to just 'code', and now again to 'code || symtab || strsym'
[3].  Different linkers generate different build IDs, with the LLVM
linker using BLAKE3 hashes for example [4].

Therefore, we might as well switch to a more modern algorithm.  Let's go
with BLAKE2s.  It's faster than SHA-1, isn't cryptographically broken,
is easier to implement than BLAKE3, and the kernel's implementation in
lib/crypto/blake2s.c is easily borrowed.  It also natively supports
variable-length hashes, so it can directly produce the needed 20 bytes.

Also make the following additional improvements:

- Hash the three inputs incrementally, so they don't all have to be
  concatenated into one buffer.

- Add tag/length prefixes to each of the three inputs, so that distinct
  input tuples reliably result in distinct hashes.

[1] https://lore.kernel.org/linux-perf-users/20250521225307.743726-1-yuzhuo@google.com/
[2] https://lore.kernel.org/linux-perf-users/20250625202311.23244-1-ebiggers@kernel.org/
[3] https://lore.kernel.org/linux-perf-users/20251125080748.461014-1-namhyung@kernel.org/
[4] https://github.com/llvm/llvm-project/commit/d3e5b6f7539b86995aef6e2075c1edb3059385ce

Signed-off-by: Eric Biggers <ebiggers@kernel.org>
Tested-by: Ian Rogers <irogers@google.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Fangrui Song <maskray@sourceware.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jason A. Donenfeld <Jason@zx2c4.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Pablo Galindo <pablogsal@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf util: Add BLAKE2s support

Add BLAKE2s support to the perf utility library.  The code is borrowed
from the kernel.  This will replace the use of SHA-1 in genelf.c.

Signed-off-by: Eric Biggers <ebiggers@kernel.org>
Tested-by: Ian Rogers <irogers@google.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Fangrui Song <maskray@sourceware.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jason A. Donenfeld <Jason@zx2c4.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Pablo Galindo <pablogsal@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

libsubcmd: Fix null intersection case in exclude_cmds()

When there is no exclusion occurring from the cmds list - for example -
cmds contains ["read-vdso32"] and excludes contains ["archive"] - the
main loop completes with ci == cj == 0. In the original code the loop
processing the remaining elements in the list was conditional:

    if (ci != cj) { ...}

So we end up in the assertion loop since ci < cmds->cnt and we
incorrectly try to assert the list elements to be NULL and fail with
the following error

   help.c:104: exclude_cmds: Assertion `cmds->names[ci] == NULL' failed.

Fix this by moving the if (ci != cj) check inside of a broader loop.
If ci != cj, left shift the list elements, as before, and then
unconditionally advance the ci and cj indicies which also covers the
ci == cj case.

Fixes: 1fdf938168c4d26f ("perf tools: Fix use-after-free in help_unknown_cmd()")
Reviewed-by: Guilherme Amadio <amadio@gentoo.org>
Signed-off-by: Sri Jayaramappa <sjayaram@akamai.com>
Tested-by: Guilherme Amadio <amadio@gentoo.org>
Tested-by: Ian Rogers <irogers@google.com>
Cc: Joshua Hunt <johunt@akamai.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/r/20251202213632.2873731-1-sjayaram@akamai.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf test: Test addr2line unwinding works with inline functions

Add a test that seeks to see inline functions correctly displayed in
'perf script' from the inlineloop workload.

Committer testing:

  # perf test 'addr2line inline unwinding'
   76: test addr2line inline unwinding                                 : Ok
  # perf test -vv 'addr2line inline unwinding'
   76: test addr2line inline unwinding:
  --- start ---
  test child forked, pid 1508628
  Inline unwinding verification test
  [ perf record: Woken up 129 times to write data ]
  [ perf record: Captured and wrote 32.282 MB /tmp/perf-test-inline-addr2line.L4Sz8QtADJ/perf.data (4014 samples) ]
  Inline unwinding verification test [Success]
  ---- end(0) ----
   76: test addr2line inline unwinding                                 : Ok
  #

Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Tony Jones <tonyj@suse.de>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf callchain: Fix srcline printing with inlines

sample__fprintf_callchain() was using map__fprintf_srcline() which won't
report inline line numbers.

Fix by using the srcline from the callchain and falling back to the map
variant.

Fixes: 25da4fab5f66e659 ("perf evsel: Move fprintf methods to separate source file")
Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Tony Jones <tonyj@suse.de>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf srcline: Add configuration support for the addr2line style

Allow the addr2line style to be specified on the `perf report` command
line or in the .perfconfig file.

Committer testing:

The methods:

  # perf probe -x ~/bin/perf -F *__addr2line
  cmd__addr2line
  libbfd__addr2line
  libdw__addr2line
  llvm__addr2line
  #

So if we configure one of them, say 'addr2line':

  # perf config addr2line.style=addr2line
  # perf config addr2line.style
  addr2line.style=addr2line
  #

And have probes on all of them:

  # perf probe -x ~/bin/perf *__addr2line
  Added new events:
    probe_perf:cmd__addr2line (on *__addr2line in /home/acme/bin/perf)
    probe_perf:llvm__addr2line (on *__addr2line in /home/acme/bin/perf)
    probe_perf:libbfd__addr2line (on *__addr2line in /home/acme/bin/perf)
    probe_perf:libdw__addr2line (on *__addr2line in /home/acme/bin/perf)

  You can now use it in all perf tools, such as:

   perf record -e probe_perf:libdw__addr2line -aR sleep 1

  #

Only the selected method should be used:

  # perf stat -e probe_perf:*_addr2line perf report -f --dso perf --stdio -s srcfile,srcline
  # Total Lost Samples: 0
  #
  # Samples: 4K of event 'cpu/cycles/Pu'
  # Event count (approx.): 5535180842
  #
  # Overhead  Source File   Source:Line
  # ........  ............  ...............
  #
      99.04%  inlineloop.c  inlineloop.c:21
       0.46%  inlineloop.c  inlineloop.c:20

  #
  # (Tip: For hierarchical output, try: perf report --hierarchy)
  #

   Performance counter stats for 'perf report -f --dso perf --stdio -s srcfile,srcline':

                  44      probe_perf:cmd__addr2line
                   0      probe_perf:llvm__addr2line
                   0      probe_perf:libbfd__addr2line
                   0      probe_perf:libdw__addr2line

         0.035915611 seconds time elapsed

         0.028008000 seconds user
         0.009051000 seconds sys
  #

I checked and that is the case for the other methods.

Also when using:

  # perf config addr2line.style=libdw,llvm

 Performance counter stats for 'perf report -f --dso perf --stdio -s srcfile,srcline':

                 0      probe_perf:cmd__addr2line
                23      probe_perf:llvm__addr2line
                 0      probe_perf:libbfd__addr2line
                44      probe_perf:libdw__addr2line

Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Tony Jones <tonyj@suse.de>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf addr2line.c: Rename a2l_style to cmd_a2l_style

The a2l_style is only relevant to the command line version, so rename
to make this clearer.

Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Tony Jones <tonyj@suse.de>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf addr2line: Add a libdw implementation

Add an implementation of addr2line that uses libdw.

Other addr2line implementations are slow, particularly in the case of
forking addr2line.

Add an implementation that caches the libdw information in the dso and
uses it to find the file and line number information.

Inline information is supported but because cu_walk_functions_at visits
the leaf function last add a inline_list__append_tail to reverse the
lists order.

Committer testing:

  # perf probe -x ~/bin/perf libdw__addr2line
  Added new event:
    probe_perf:libdw_addr2line (on libdw__addr2line in /home/acme/bin/perf)

  You can now use it in all perf tools, such as:

   perf record -e probe_perf:libdw_addr2line -aR sleep 1

  #
  # perf stat -e probe_perf:libdw_addr2line perf report -f --dso perf --stdio -s srcfile,srcline
  # To display the perf.data header info, please use --header/--header-only options.
  #
  #
  # Total Lost Samples: 0
  #
  # Samples: 4K of event 'cpu/cycles/Pu'
  # Event count (approx.): 5535180842
  #
  # Overhead  Source File   Source:Line
  # ........  ............  ...............
  #
      99.04%  inlineloop.c  inlineloop.c:21
       0.46%  inlineloop.c  inlineloop.c:20

  #
  # (Tip: For tracepoint events, try: perf report -s trace_fields)
  #

   Performance counter stats for 'perf report -f --dso perf --stdio -s srcfile,srcline':

                  44      probe_perf:libdw_addr2line

         0.037260744 seconds time elapsed

         0.025299000 seconds user
         0.011918000 seconds sys
  #

Adding probes to the other addr2line implementations (llvm__addr2line,
libbfd__addr2line and cmd__addr2line) I noticed some fallbacks to the
llvm one:

 Performance counter stats for 'perf report -f --dso perf --stdio -s srcfile,srcline':

                44      probe_perf:libdw_addr2line
                23      probe_perf:llvm_addr2line
                 0 probe_perf:libbfd_addr2line
                 0      probe_perf:cmd_addr2line

Something to investigate further, but at least we don't fallback to the
cmd based one :-)

Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Tony Jones <tonyj@suse.de>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf test workload: Add inlineloop test workload

The purpose of this workload is to gather samples in an inlined
function. This can be used to test whether inlined addr2line works
correctly.

Committer testing:

  $ perf record perf test -w inlineloop 1
  [ perf record: Woken up 2 times to write data ]
  [ perf record: Captured and wrote 0.161 MB perf.data (4005 samples) ]
  $ perf report --stdio --dso perf -s srcfile,srcline
  #
  # Total Lost Samples: 0
  #
  # Samples: 4K of event 'cpu/cycles/Pu'
  # Event count (approx.): 5535180842
  #
  # Overhead  Source File   Source:Line
  # ........  ............  ...............
  #
      99.04%  inlineloop.c  inlineloop.c:21
       0.46%  inlineloop.c  inlineloop.c:20
  #
  $

Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Tony Jones <tonyj@suse.de>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf unwind-libdw: Fix invalid reference counts

The addition of addr_location__exit() causes use-after put on the maps
and map references in the unwind info. Add the gets and then add the
map_symbol__exit() calls.

Fixes: 0dd5041c9a0eaf8c ("perf addr_location: Add init/exit/copy functions")
Reviewed-by: James Clark <james.clark@linaro.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Howard Chu <howardchu95@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephen Brennan <stephen.s.brennan@oracle.com>
Cc: Tony Jones <tonyj@suse.de>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf test subcmd help: Add exclude disjoint subcmd names

The test is based on an error/fix posted to linux-perf-users.

Reported-by: Sri Jayaramappa <sjayaram@akamai.com>
Reviewed-by: Sri Jayaramappa <sjayaram@akamai.com>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Guilherme Amadio <amadio@gentoo.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Closes: https://lore.kernel.org/linux-perf-users/20251202213632.2873731-1-sjayaram@akamai.com/
Closes: https://urldefense.com/v3/__https://lore.kernel.org/linux-perf-users/20251202213632.2873731-1-sjayaram@akamai.com/__;!!GjvTz_vk!XehekKNUE4Ib_tvqIH6PMIIhly4X3BZ-Y40RC1HKMQ-6OdYEFvUPQhyWv_gk9vsRRN4_RcOLS2Bh0CQ$
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf stat display: Make %f precision consistent

Commit bc22de9bcdb22491 ("perf stat: Display time in precision based on
std deviation") added multirun workload elapsed time. There was an
effort to make the precision in the output most useful for the user,
however, when gathering over runs it means the formatting varies. This
change just makes the output format fixed.

Before:
```
$ while :; do perf stat --null --repeat 3 sleep 0.1 2>&1 | grep elapsed; done
          0.101140 +- 0.000149 seconds time elapsed  ( +-  0.15% )
         0.1011396 +- 0.0000218 seconds time elapsed  ( +-  0.02% )
          0.101331 +- 0.000124 seconds time elapsed  ( +-  0.12% )
^C
$ while :; do perf stat --null --repeat 3 sleep 1 2>&1 | grep elapsed; done
          1.001317 +- 0.000146 seconds time elapsed  ( +-  0.01% )
          1.001377 +- 0.000172 seconds time elapsed  ( +-  0.02% )
           1.00253 +- 0.00131 seconds time elapsed  ( +-  0.13% )
```

After:
```
$ while :; do perf stat --null --repeat 3 sleep 0.1 2>&1 | grep elapsed; done
       0.101406408 +- 0.000064778 seconds time elapsed  ( +-  0.06% )
       0.101367315 +- 0.000027253 seconds time elapsed  ( +-  0.03% )
       0.101434164 +- 0.000084750 seconds time elapsed  ( +-  0.08% )
^C
$ while :; do perf stat --null --repeat 3 sleep 1 2>&1 | grep elapsed; done
       1.001525467 +- 0.000051703 seconds time elapsed  ( +-  0.01% )
       1.001375093 +- 0.000116200 seconds time elapsed  ( +-  0.01% )
       1.001141025 +- 0.000046361 seconds time elapsed  ( +-  0.00% )
```

Closes: https://lore.kernel.org/lkml/aTQRgAOpKyI53TEq@gmail.com/
Suggested-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Ian Rogers <irogers@google.com>
Tested-by: Ingo Molnar <mingo@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Chun-Tse Shao <ctshao@google.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf build: Raise minimum shellcheck version to 0.7.2

Raise the minimum shellcheck version for perf builds to 0.7.2, so that
systems with shellcheck versions below 0.7.2 will automatically skip the
shell script checking, even if NO_SHELLCHECK is unset.

Since commit 241f21be7d0fdf3c ("perf test perftool_testsuite: Use
absolute paths"), shellcheck versions before 0.7.2 break the perf build
with several SC1090 [2] warnings due to its too strict dynamic source
handling [1], e.g.:

  In tests/shell/base_probe/test_line_semantics.sh line 20:
  . "$DIR_PATH/../common/init.sh"
    ^---------------------------^ SC1090: Can't follow non-constant source. Use a directive to specify location.

Fixes: 241f21be7d0fdf3c ("perf test perftool_testsuite: Use absolute paths")
Signed-off-by: Nicolas Schier <n.schier@avm.de>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jakub Brnak <jbrnak@redhat.com>
Cc: James Clark <james.clark@linaro.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Nicolas Schier <nsc@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Philipp Hahn <p.hahn@avm.de>
Cc: Veronika Molnarova <vmolnaro@redhat.com>
Link: https://github.com/koalaman/shellcheck/issues/1998
Link: https://www.shellcheck.net/wiki/SC1090
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>