This is a breakdown of perf_mem_data_src.mem_lvl_num. But it's also
divided into two parts because the combination is bigger than 8.
Since there are many entries for different cache levels, 'cache' field
focuses on them. I generalized buffers like LFB, MAB and MHB to L1-buf
and L2-buf.
The rest goes to 'memory' field which can be RAM, CXL, PMEM, IO, etc.
$ perf mem report -F cache,mem,dso --stdio
...
#
# -------------- Cache -------------- --- Memory ---
# L1 L2 L3 L1-buf Other RAM Other Shared Object
# ................................... .............. ....................................
#
53.9% 3.6% 16.2% 21.6% 4.8% 4.8% 95.2% [kernel.kallsyms]
64.7% 1.7% 3.5% 17.4% 12.8% 12.8% 87.2% chrome (deleted)
78.3% 2.8% 0.0% 1.0% 17.9% 17.9% 82.1% libc.so.6
39.6% 1.5% 0.0% 5.7% 53.2% 53.2% 46.8% libxul.so
26.2% 0.0% 0.0% 0.0% 73.8% 73.8% 26.2% [unknown]
85.5% 0.0% 0.0% 14.5% 0.0% 0.0% 100.0% libspa-audioconvert.so
66.3% 4.4% 0.0% 29.4% 0.0% 0.0% 100.0% libglib-2.0.so.0.8200.1 (deleted)
1.9% 0.0% 0.0% 0.0% 98.1% 98.1% 1.9% libmutter-cogl-15.so.0.0.0 (deleted)
10.6% 0.0% 0.0% 89.4% 0.0% 0.0% 100.0% libpulsecommon-16.1.so
0.0% 0.0% 0.0% 100.0% 0.0% 0.0% 100.0% libfreeblpriv3.so (deleted)
...
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: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Link: https://lore.kernel.org/r/20250430205548.789750-10-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
This is an actual example of the he_mem_stat based sample breakdown. It
uses 'mem_op' field of union perf_mem_data_src which means memory
operations.
It'd have basically 'load' or 'store' which can be useful if PMU doesn't
have separate events for them like IBS or SPE. In addition, there's an
entry in case load and store happen at the same time. Also adds entries
for prefetching and execution.
$ perf mem report -F +op -s comm --stdio
# To display the perf.data header info, please use --header/--header-only options.
#
#
# Total Lost Samples: 0
#
# Samples: 4K of event 'ibs_op//'
# Total weight : 9559
# Sort order : comm
#
# --------------------- Mem Op ----------------------
# Overhead Samples Load Store Ld+St Pfetch Exec Other N/A N/A Command
# ........ ....... ................................................... ...............
#
44.85% 4077 21.1% 30.7% 0.0% 0.0% 0.0% 48.3% 0.0% 0.0% swapper
26.82% 45 98.8% 0.3% 0.0% 0.0% 0.0% 0.9% 0.0% 0.0% netsli-prober
7.19% 442 51.7% 13.7% 0.0% 0.0% 0.0% 34.6% 0.0% 0.0% perf
5.81% 75 89.7% 2.2% 0.0% 0.0% 0.0% 8.1% 0.0% 0.0% qemu-system-ppc
4.77% 1 100.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% notifications_c
1.77% 10 95.9% 1.2% 0.0% 0.0% 0.0% 3.0% 0.0% 0.0% MemoryReleaser
0.77% 32 71.6% 4.1% 0.0% 0.0% 0.0% 24.3% 0.0% 0.0% DefaultEventMan
0.19% 10 66.7% 22.2% 0.0% 0.0% 0.0% 11.1% 0.0% 0.0% gnome-shell
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Link: https://lore.kernel.org/r/20250430205548.789750-8-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
This is a preparation for later changes to support mem_stat output. The
new fields will need two lines for the header - the first line will show
type of mem stat and the second line will show the name of each item
which is returned by mem_stat_name().
Each element in the mem_stat array will be printed in percentage for the
hist_entry and their sum would be 100%.
Add new output field dimension only for SORT_MODE__MEM using mem_stat.
To handle possible name conflict with existing sort keys, move the order
of checking output field dimensions after the sort dimensions when it
looks for sort keys.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Link: https://lore.kernel.org/r/20250430205548.789750-7-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
When preparing the mem events for the argv copies are intentionally
made. These copies are leaked and cause runs of perf using address
sanitizer to fail. Rather than leak the memory allocate a chunk of
memory for the mem event names upfront and build the strings in this -
the storage is sized larger than the previous buffer size. The caller
is then responsible for clearing up this memory. As part of this
change, remove the mem_loads_name and mem_stores_name global buffers
then change the perf_pmu__mem_events_name to write to an out argument
buffer.
Signed-off-by: Ian Rogers <irogers@google.com>
Tested-by: Thomas Falcon <thomas.falcon@intel.com>
Reviewed-by: Leo Yan <leo.yan@arm.com>
Link: https://lore.kernel.org/r/20250308012853.1384762-1-irogers@google.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
In sysfs, the perf events are all located in
/sys/bus/event_source/devices/ but some places ended up hard-coding the
location to be at the root of /sys/devices/ which could be very risky as
you do not exactly know what type of device you are accessing in sysfs
at that location.
So fix this all up by properly pointing everything at the bus device
list instead of the root of the sysfs devices/ tree.
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Kan Liang <kan.liang@linux.intel.com>
Link: https://lore.kernel.org/r/2025021955-implant-excavator-179d@gregkh
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
The perf_cpu_map__merge() function has two arguments, 'orig' and
'other'. The function definition might cause confusion as it could give
the impression that the CPU maps in the two arguments are copied into a
new allocated structure, which is then returned as the result.
The purpose of the function is to merge the CPU map 'other' into the CPU
map 'orig'. This commit changes the 'orig' argument to a pointer to
pointer, so the new result will be updated into 'orig'.
The return value is changed to an int type, as an error number or 0 for
success.
Update callers and tests for the new function definition.
Reviewed-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Leo Yan <leo.yan@arm.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.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>
Link: https://lore.kernel.org/r/20241107125308.41226-2-leo.yan@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
With commit 8ec9497d3e ("tools/include: Sync uapi/linux/perf.h
with the kernel sources"), 'perf mem report' gives an incorrect memory
access string.
...
0.02% 1 3644 L5 hit [.] 0x0000000000009b0e mlc [.] 0x00007fce43f59480
...
This occurs because, if no entry exists in mem_lvlnum, perf_mem__lvl_scnprintf
will default to 'L%d, lvl', which in this case for PERF_MEM_LVLNUM_L2_MHB is 0x05.
Add entries for PERF_MEM_LVLNUM_L2_MHB and PERF_MEM_LVLNUM_MSC to mem_lvlnum,
so that the correct strings are printed.
...
0.02% 1 3644 L2 MHB hit [.] 0x0000000000009b0e mlc [.] 0x00007fce43f59480
...
Fixes: 8ec9497d3e ("tools/include: Sync uapi/linux/perf.h with the kernel sources")
Suggested-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Thomas Falcon <thomas.falcon@intel.com>
Reviewed-by: Leo Yan <leo.yan@arm.com>
Link: https://lore.kernel.org/r/20240926144040.77897-1-thomas.falcon@intel.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
The p-core mem events are missed when launching 'perf mem record' on ADL
and RPL.
root@number:~# perf mem record sleep 1
Memory events are enabled on a subset of CPUs: 16-27
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.032 MB perf.data ]
root@number:~# perf evlist
cpu_atom/mem-loads,ldlat=30/P
cpu_atom/mem-stores/P
dummy:u
A variable 'record' in the 'struct perf_mem_event' is to indicate
whether a mem event in a mem_events[] should be recorded. The current
code only configure the variable for the first eligible PMU.
It's good enough for a non-hybrid machine or a hybrid machine which has
the same mem_events[].
However, if a different mem_events[] is used for different PMUs on a
hybrid machine, e.g., ADL or RPL, the 'record' for the second PMU never
get a chance to be set.
The mem_events[] of the second PMU are always ignored.
'perf mem' doesn't support the per-PMU configuration now. A per-PMU
mem_events[] 'record' variable doesn't make sense. Make it global.
That could also avoid searching for the per-PMU mem_events[] via
perf_pmu__mem_events_ptr every time.
Committer testing:
root@number:~# perf evlist -g
cpu_atom/mem-loads,ldlat=30/P
cpu_atom/mem-stores/P
{cpu_core/mem-loads-aux/,cpu_core/mem-loads,ldlat=30/}
cpu_core/mem-stores/P
dummy:u
root@number:~#
The :S for '{cpu_core/mem-loads-aux/,cpu_core/mem-loads,ldlat=30/}' is
not being added by 'perf evlist -g', to be checked.
Fixes: abbdd79b78 ("perf mem: Clean up perf_mem_events__name()")
Reported-by: Arnaldo Carvalho de Melo <acme@kernel.org>
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Closes: https://lore.kernel.org/lkml/Zthu81fA3kLC2CS2@x1/
Link: https://lore.kernel.org/r/20240905170737.4070743-2-kan.liang@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The current perf_pmu__mem_events_init() only checks the availability of
the mem_events for the first eligible PMU. It works for non-hybrid
machines and hybrid machines that have the same mem_events.
However, it may bring issues if a hybrid machine has a different
mem_events on different PMU, e.g., Alder Lake and Raptor Lake. A
mem-loads-aux event is only required for the p-core. The mem_events on
both e-core and p-core should be checked and marked.
The issue was not found, because it's hidden by another bug, which only
records the mem-events for the e-core. The wrong check for the p-core
events didn't yell.
Fixes: abbdd79b78 ("perf mem: Clean up perf_mem_events__name()")
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20240905170737.4070743-1-kan.liang@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Guilherme reported a crash in perf mem record. It's because the
perf_mem_event->name was NULL on his machine. It should just return
a NULL string when it has no format string in the name.
The backtrace at the crash is below:
Program received signal SIGSEGV, Segmentation fault.
__strchrnul_avx2 () at ../sysdeps/x86_64/multiarch/strchr-avx2.S:67
67 vmovdqu (%rdi), %ymm2
(gdb) bt
#0 __strchrnul_avx2 () at ../sysdeps/x86_64/multiarch/strchr-avx2.S:67
#1 0x00007ffff6c982de in __find_specmb (format=0x0) at printf-parse.h:82
#2 __printf_buffer (buf=buf@entry=0x7fffffffc760, format=format@entry=0x0, ap=ap@entry=0x7fffffffc880,
mode_flags=mode_flags@entry=0) at vfprintf-internal.c:649
#3 0x00007ffff6cb7840 in __vsnprintf_internal (string=<optimized out>, maxlen=<optimized out>, format=0x0,
args=0x7fffffffc880, mode_flags=mode_flags@entry=0) at vsnprintf.c:96
#4 0x00007ffff6cb787f in ___vsnprintf (string=<optimized out>, maxlen=<optimized out>, format=<optimized out>,
args=<optimized out>) at vsnprintf.c:103
#5 0x00005555557b9391 in scnprintf (buf=0x555555fe9320 <mem_loads_name> "", size=100, fmt=0x0)
at ../lib/vsprintf.c:21
#6 0x00005555557b74c3 in perf_pmu__mem_events_name (i=0, pmu=0x555556832180) at util/mem-events.c:106
#7 0x00005555557b7ab9 in perf_mem_events__record_args (rec_argv=0x55555684c000, argv_nr=0x7fffffffca20)
at util/mem-events.c:252
#8 0x00005555555e370d in __cmd_record (argc=3, argv=0x7fffffffd760, mem=0x7fffffffcd80) at builtin-mem.c:156
#9 0x00005555555e49c4 in cmd_mem (argc=4, argv=0x7fffffffd760) at builtin-mem.c:514
#10 0x000055555569716c in run_builtin (p=0x555555fcde80 <commands+672>, argc=8, argv=0x7fffffffd760) at perf.c:349
#11 0x0000555555697402 in handle_internal_command (argc=8, argv=0x7fffffffd760) at perf.c:402
#12 0x0000555555697560 in run_argv (argcp=0x7fffffffd59c, argv=0x7fffffffd590) at perf.c:446
#13 0x00005555556978a6 in main (argc=8, argv=0x7fffffffd760) at perf.c:562
Reported-by: Guilherme Amadio <amadio@cern.ch>
Reviewed-by: Kan Liang <kan.liang@linux.intel.com>
Closes: https://lore.kernel.org/linux-perf-users/Zlns_o_IE5L28168@cern.ch
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20240621170528.608772-5-namhyung@kernel.org
The below error can be triggered on a hybrid machine.
$ perf mem record -t load sleep 1
event syntax error: 'breakpoint/mem-loads,ldlat=30/P'
\___ Bad event or PMU
Unable to find PMU or event on a PMU of 'breakpoint'
In the perf_mem_events__record_args(), the current perf never checks the
availability of a mem event on a given PMU. All the PMUs will be added
to the perf mem event list. Perf errors out for the unsupported PMU.
Extend perf_mem_event__supported() and take a PMU into account. Check
the mem event for each PMU before adding it to the perf mem event list.
Optimize the perf_mem_events__init() a little bit. The function is to
check whether the mem events are supported in the system. It doesn't
need to scan all PMUs. Just return with the first supported PMU is good
enough.
Fixes: 5752c20f37 ("perf mem: Scan all PMUs instead of just core ones")
Reported-by: Ammy Yi <ammy.yi@intel.com>
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Tested-by: Ammy Yi <ammy.yi@intel.com>
Acked-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.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>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Link: https://lore.kernel.org/r/20231128203940.3964287-1-kan.liang@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Interpretation of 'union perf_mem_data_src' by perf_mem__lvl_scnprintf()
is non-intuitive. For ex, it ignores 'mem_lvl' when 'mem_hops' is set
but considers it otherwise. It prints both 'mem_lvl_num' and 'mem_lvl'
when 'mem_hops' is not set.
Refactor this function such that it behaves more intuitively: Use new
API 'mem_lvl_num'|'mem_remote'|'mem_hops' if 'mem_lvl_num' contains
value other than PERF_MEM_LVLNUM_NA. Otherwise, fallback to old API
'mem_lvl'. Since new API has no way to indicate MISS, use it from old
api, otherwise don't club old and new APIs while parsing as well as
printing.
Before:
$ sudo ./perf mem report -F sample,mem --stdio
# Samples Memory access
# ............ ........................
#
250097 N/A
188907 L1 hit
4116 L2 hit
3496 Remote Cache (1 hop) hit
3271 Remote Cache (2 hops) hit
873 L3 hit
598 Local RAM hit
438 Remote RAM (1 hop) hit
1 Uncached hit
After:
$ sudo ./perf mem report -F sample,mem --stdio
# Samples Memory access
# ............ .......................................
#
255517 N/A
189989 L1 hit
4541 L2 hit
3363 Remote core, same node Any cache hit
3336 Remote node, same socket Any cache hit
1275 L3 hit
743 RAM hit
545 Remote node, same socket RAM hit
4 Uncached hit
Signed-off-by: Ravi Bangoria <ravi.bangoria@amd.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ananth Narayan <ananth.narayan@amd.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Santosh Shukla <santosh.shukla@amd.com>
Cc: Stephane Eranian <eranian@google.com>
Link: https://lore.kernel.org/r/20230407112459.548-5-ravi.bangoria@amd.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Going forward, future generation systems can have more hierarchy
within the node/package level but currently we don't have any data source
encoding field in perf, which can be used to represent this level of data.
Add a new field called 'mem_hops' in the perf_mem_data_src structure
which can be used to represent intra-node/package or inter-node/off-package
details. This field is of size 3 bits where PERF_MEM_HOPS_{NA, 0..6} value
can be used to present different hop levels data.
Also add corresponding macros to define mem_hop field values
and shift value.
Currently we define macro for HOPS_0 which corresponds
to data coming from another core but same node.
Add functionality to represent mem_hop field data in
perf_mem__lvl_scnprintf function with the help of added string
array called mem_hops.
For ex: Encodings for mem_hops fields with L2 cache:
L2 - local L2
L2 | REMOTE | HOPS_0 - remote core, same node L2
Since with the addition of HOPS field, now remote can be used to
denote cache access from the same node but different core, a check
is added in the c2c_decode_stats function to set mrem only when HOPS
is zero along with set remote field.
Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20211006140654.298352-4-kjain@linux.ibm.com
The perf_mem_events__name() can generate the mem-load event name.
It uses a variable 'mem_loads_name__init' to avoid generating the
event name every time (because perf_pmu__scan takes some time).
The perf_mem_events__name() assumes the pmu is "cpu" but it's not
correct for hybrid platform. For Alderlake, the pmu is "cpu_core" or
"cpu_atom"
Introduce a new parameter 'pmu_name' in perf_mem_events__name
to let the caller specify a pmu name.
Considering such event name is x86 specific, so move
perf_mem_events[] to arch/x86/util/mem-events.c.
We still keep the variable 'mem_loads_name__init' but it's only
used when pmu_name is NULL (compatible for original behavior). When
pmu_name is not NULL (e.g. "cpu_core"), this patch doesn't have
optimization. That can be implemented in follow up patch.
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/r/20210527001610.10553-3-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
