Add flag that disables affinity behavior. Using sched_setaffinity to
place a perf thread on a CPU can avoid certain interprocessor
interrupts but may introduce a delay due to the scheduling,
particularly on loaded machines. Add a command line option to disable
the behavior. This behavior is less present in other tools like `perf
record`, as it uses a ring buffer and doesn't make repeated system
calls.
Signed-off-by: Ian Rogers <irogers@google.com>
---
tools/perf/Documentation/perf-stat.txt | 4 ++++
tools/perf/builtin-stat.c | 6 ++++++
tools/perf/util/evlist.c | 6 +-----
tools/perf/util/evlist.h | 1 +
4 files changed, 12 insertions(+), 5 deletions(-)
diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt
index 1a766d4a2233..1ffb510606af 100644
--- a/tools/perf/Documentation/perf-stat.txt
+++ b/tools/perf/Documentation/perf-stat.txt
@@ -382,6 +382,10 @@ color the metric's computed value.
Don't print output, warnings or messages. This is useful with perf stat
record below to only write data to the perf.data file.
+--no-affinity::
+Don't change scheduler affinities when iterating over CPUs. Disables
+an optimization aimed at minimizing interprocessor interrupts.
+
STAT RECORD
-----------
Stores stat data into perf data file.
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index c1bb40b99176..8bbdea44c3ba 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -2426,6 +2426,7 @@ static int parse_tpebs_mode(const struct option *opt, const char *str,
int cmd_stat(int argc, const char **argv)
{
struct opt_aggr_mode opt_mode = {};
+ bool affinity = true, affinity_set = false;
struct option stat_options[] = {
OPT_BOOLEAN('T', "transaction", &transaction_run,
"hardware transaction statistics"),
@@ -2554,6 +2555,8 @@ int cmd_stat(int argc, const char **argv)
"don't print 'summary' for CSV summary output"),
OPT_BOOLEAN(0, "quiet", &quiet,
"don't print any output, messages or warnings (useful with record)"),
+ OPT_BOOLEAN_SET(0, "affinity", &affinity, &affinity_set,
+ "don't allow affinity optimizations aimed at reducing IPIs"),
OPT_CALLBACK(0, "cputype", &evsel_list, "hybrid cpu type",
"Only enable events on applying cpu with this type "
"for hybrid platform (e.g. core or atom)",
@@ -2611,6 +2614,9 @@ int cmd_stat(int argc, const char **argv)
} else
stat_config.csv_sep = DEFAULT_SEPARATOR;
+ if (affinity_set)
+ evsel_list->no_affinity = !affinity;
+
if (argc && strlen(argv[0]) > 2 && strstarts("record", argv[0])) {
argc = __cmd_record(stat_options, &opt_mode, argc, argv);
if (argc < 0)
diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
index 45833244daf3..591bdf0b3e2a 100644
--- a/tools/perf/util/evlist.c
+++ b/tools/perf/util/evlist.c
@@ -369,11 +369,7 @@ static bool evlist__use_affinity(struct evlist *evlist)
struct perf_cpu_map *used_cpus = NULL;
bool ret = false;
- /*
- * With perf record core.user_requested_cpus is usually NULL.
- * Use the old method to handle this for now.
- */
- if (!evlist->core.user_requested_cpus ||
+ if (evlist->no_affinity || !evlist->core.user_requested_cpus ||
cpu_map__is_dummy(evlist->core.user_requested_cpus))
return false;
diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h
index 30dff7484d3c..d17c3b57a409 100644
--- a/tools/perf/util/evlist.h
+++ b/tools/perf/util/evlist.h
@@ -59,6 +59,7 @@ struct event_enable_timer;
struct evlist {
struct perf_evlist core;
bool enabled;
+ bool no_affinity;
int id_pos;
int is_pos;
int nr_br_cntr;
--
2.53.0.rc2.204.g2597b5adb4-googOn Fri, Feb 06, 2026 at 02:25:09PM -0800, Ian Rogers wrote:
> Add flag that disables affinity behavior. Using sched_setaffinity to
> place a perf thread on a CPU can avoid certain interprocessor
> interrupts but may introduce a delay due to the scheduling,
> particularly on loaded machines. Add a command line option to disable
> the behavior. This behavior is less present in other tools like `perf
> record`, as it uses a ring buffer and doesn't make repeated system
> calls.
This is confusing:
⬢ [acme@toolbx perf-tools-next]$ perf stat -h affinity
Usage: perf stat [<options>] [<command>]
--affinity don't allow affinity optimizations aimed at reducing IPIs
⬢ [acme@toolbx perf-tools-next]$
The way it is presented in the -h output it looks as if one has to use:
perf stat --affinity
To disable affinity setting, when used that way it looks as if the user
is asking for affinity to be used.
We have things like:
⬢ [acme@toolbx perf-tools-next]$ grep -A2 OPT_.*no- tools/perf/builtin-record.c
OPT_BOOLEAN(0, "no-buffering", &record.opts.no_buffering,
"collect data without buffering"),
OPT_BOOLEAN('R', "raw-samples", &record.opts.raw_samples,
--
OPT_BOOLEAN_SET('i', "no-inherit", &record.opts.no_inherit,
&record.opts.no_inherit_set,
"child tasks do not inherit counters"),
--
OPT_BOOLEAN(0, "no-bpf-event", &record.opts.no_bpf_event, "do not record bpf events"),
OPT_BOOLEAN(0, "strict-freq", &record.opts.strict_freq,
"Fail if the specified frequency can't be used"),
--
OPT_BOOLEAN('n', "no-samples", &record.opts.no_samples,
"don't sample"),
OPT_BOOLEAN_SET('N', "no-buildid-cache", &record.no_buildid_cache,
&record.no_buildid_cache_set,
"do not update the buildid cache"),
OPT_BOOLEAN_SET('B', "no-buildid", &record.no_buildid,
&record.no_buildid_set,
"do not collect buildids in perf.data"),
⬢ [acme@toolbx perf-tools-next]$
Probably this needs to be that way?
- Arnaldo
> Signed-off-by: Ian Rogers <irogers@google.com>
> ---
> tools/perf/Documentation/perf-stat.txt | 4 ++++
> tools/perf/builtin-stat.c | 6 ++++++
> tools/perf/util/evlist.c | 6 +-----
> tools/perf/util/evlist.h | 1 +
> 4 files changed, 12 insertions(+), 5 deletions(-)
>
> diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt
> index 1a766d4a2233..1ffb510606af 100644
> --- a/tools/perf/Documentation/perf-stat.txt
> +++ b/tools/perf/Documentation/perf-stat.txt
> @@ -382,6 +382,10 @@ color the metric's computed value.
> Don't print output, warnings or messages. This is useful with perf stat
> record below to only write data to the perf.data file.
>
> +--no-affinity::
> +Don't change scheduler affinities when iterating over CPUs. Disables
> +an optimization aimed at minimizing interprocessor interrupts.
> +
> STAT RECORD
> -----------
> Stores stat data into perf data file.
> diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
> index c1bb40b99176..8bbdea44c3ba 100644
> --- a/tools/perf/builtin-stat.c
> +++ b/tools/perf/builtin-stat.c
> @@ -2426,6 +2426,7 @@ static int parse_tpebs_mode(const struct option *opt, const char *str,
> int cmd_stat(int argc, const char **argv)
> {
> struct opt_aggr_mode opt_mode = {};
> + bool affinity = true, affinity_set = false;
> struct option stat_options[] = {
> OPT_BOOLEAN('T', "transaction", &transaction_run,
> "hardware transaction statistics"),
> @@ -2554,6 +2555,8 @@ int cmd_stat(int argc, const char **argv)
> "don't print 'summary' for CSV summary output"),
> OPT_BOOLEAN(0, "quiet", &quiet,
> "don't print any output, messages or warnings (useful with record)"),
> + OPT_BOOLEAN_SET(0, "affinity", &affinity, &affinity_set,
> + "don't allow affinity optimizations aimed at reducing IPIs"),
> OPT_CALLBACK(0, "cputype", &evsel_list, "hybrid cpu type",
> "Only enable events on applying cpu with this type "
> "for hybrid platform (e.g. core or atom)",
> @@ -2611,6 +2614,9 @@ int cmd_stat(int argc, const char **argv)
> } else
> stat_config.csv_sep = DEFAULT_SEPARATOR;
>
> + if (affinity_set)
> + evsel_list->no_affinity = !affinity;
> +
> if (argc && strlen(argv[0]) > 2 && strstarts("record", argv[0])) {
> argc = __cmd_record(stat_options, &opt_mode, argc, argv);
> if (argc < 0)
> diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
> index 45833244daf3..591bdf0b3e2a 100644
> --- a/tools/perf/util/evlist.c
> +++ b/tools/perf/util/evlist.c
> @@ -369,11 +369,7 @@ static bool evlist__use_affinity(struct evlist *evlist)
> struct perf_cpu_map *used_cpus = NULL;
> bool ret = false;
>
> - /*
> - * With perf record core.user_requested_cpus is usually NULL.
> - * Use the old method to handle this for now.
> - */
> - if (!evlist->core.user_requested_cpus ||
> + if (evlist->no_affinity || !evlist->core.user_requested_cpus ||
> cpu_map__is_dummy(evlist->core.user_requested_cpus))
> return false;
>
> diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h
> index 30dff7484d3c..d17c3b57a409 100644
> --- a/tools/perf/util/evlist.h
> +++ b/tools/perf/util/evlist.h
> @@ -59,6 +59,7 @@ struct event_enable_timer;
> struct evlist {
> struct perf_evlist core;
> bool enabled;
> + bool no_affinity;
> int id_pos;
> int is_pos;
> int nr_br_cntr;
> --
> 2.53.0.rc2.204.g2597b5adb4-goog
>
On Sat, Feb 7, 2026 at 4:51 AM Arnaldo Carvalho de Melo <acme@kernel.org> wrote:
>
> On Fri, Feb 06, 2026 at 02:25:09PM -0800, Ian Rogers wrote:
> > Add flag that disables affinity behavior. Using sched_setaffinity to
> > place a perf thread on a CPU can avoid certain interprocessor
> > interrupts but may introduce a delay due to the scheduling,
> > particularly on loaded machines. Add a command line option to disable
> > the behavior. This behavior is less present in other tools like `perf
> > record`, as it uses a ring buffer and doesn't make repeated system
> > calls.
>
> This is confusing:
>
> ⬢ [acme@toolbx perf-tools-next]$ perf stat -h affinity
>
> Usage: perf stat [<options>] [<command>]
>
> --affinity don't allow affinity optimizations aimed at reducing IPIs
>
> ⬢ [acme@toolbx perf-tools-next]$
>
> The way it is presented in the -h output it looks as if one has to use:
>
> perf stat --affinity
>
> To disable affinity setting, when used that way it looks as if the user
> is asking for affinity to be used.
>
> We have things like:
>
> ⬢ [acme@toolbx perf-tools-next]$ grep -A2 OPT_.*no- tools/perf/builtin-record.c
> OPT_BOOLEAN(0, "no-buffering", &record.opts.no_buffering,
> "collect data without buffering"),
> OPT_BOOLEAN('R', "raw-samples", &record.opts.raw_samples,
> --
> OPT_BOOLEAN_SET('i', "no-inherit", &record.opts.no_inherit,
> &record.opts.no_inherit_set,
> "child tasks do not inherit counters"),
> --
> OPT_BOOLEAN(0, "no-bpf-event", &record.opts.no_bpf_event, "do not record bpf events"),
> OPT_BOOLEAN(0, "strict-freq", &record.opts.strict_freq,
> "Fail if the specified frequency can't be used"),
> --
> OPT_BOOLEAN('n', "no-samples", &record.opts.no_samples,
> "don't sample"),
> OPT_BOOLEAN_SET('N', "no-buildid-cache", &record.no_buildid_cache,
> &record.no_buildid_cache_set,
> "do not update the buildid cache"),
> OPT_BOOLEAN_SET('B', "no-buildid", &record.no_buildid,
> &record.no_buildid_set,
> "do not collect buildids in perf.data"),
> ⬢ [acme@toolbx perf-tools-next]$
>
> Probably this needs to be that way?
So it was that way in the v4 patch set but Namhyung asked for it to be
the other way around:
https://lore.kernel.org/lkml/aRvcuMfbDRSBU87k@google.com/
The comment on the option didn't get changed and that's a mistake.
Let me know what's preferred and I can send a patch.
Thanks,
Ian
> - Arnaldo
>
> > Signed-off-by: Ian Rogers <irogers@google.com>
> > ---
> > tools/perf/Documentation/perf-stat.txt | 4 ++++
> > tools/perf/builtin-stat.c | 6 ++++++
> > tools/perf/util/evlist.c | 6 +-----
> > tools/perf/util/evlist.h | 1 +
> > 4 files changed, 12 insertions(+), 5 deletions(-)
> >
> > diff --git a/tools/perf/Documentation/perf-stat.txt b/tools/perf/Documentation/perf-stat.txt
> > index 1a766d4a2233..1ffb510606af 100644
> > --- a/tools/perf/Documentation/perf-stat.txt
> > +++ b/tools/perf/Documentation/perf-stat.txt
> > @@ -382,6 +382,10 @@ color the metric's computed value.
> > Don't print output, warnings or messages. This is useful with perf stat
> > record below to only write data to the perf.data file.
> >
> > +--no-affinity::
> > +Don't change scheduler affinities when iterating over CPUs. Disables
> > +an optimization aimed at minimizing interprocessor interrupts.
> > +
> > STAT RECORD
> > -----------
> > Stores stat data into perf data file.
> > diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
> > index c1bb40b99176..8bbdea44c3ba 100644
> > --- a/tools/perf/builtin-stat.c
> > +++ b/tools/perf/builtin-stat.c
> > @@ -2426,6 +2426,7 @@ static int parse_tpebs_mode(const struct option *opt, const char *str,
> > int cmd_stat(int argc, const char **argv)
> > {
> > struct opt_aggr_mode opt_mode = {};
> > + bool affinity = true, affinity_set = false;
> > struct option stat_options[] = {
> > OPT_BOOLEAN('T', "transaction", &transaction_run,
> > "hardware transaction statistics"),
> > @@ -2554,6 +2555,8 @@ int cmd_stat(int argc, const char **argv)
> > "don't print 'summary' for CSV summary output"),
> > OPT_BOOLEAN(0, "quiet", &quiet,
> > "don't print any output, messages or warnings (useful with record)"),
> > + OPT_BOOLEAN_SET(0, "affinity", &affinity, &affinity_set,
> > + "don't allow affinity optimizations aimed at reducing IPIs"),
> > OPT_CALLBACK(0, "cputype", &evsel_list, "hybrid cpu type",
> > "Only enable events on applying cpu with this type "
> > "for hybrid platform (e.g. core or atom)",
> > @@ -2611,6 +2614,9 @@ int cmd_stat(int argc, const char **argv)
> > } else
> > stat_config.csv_sep = DEFAULT_SEPARATOR;
> >
> > + if (affinity_set)
> > + evsel_list->no_affinity = !affinity;
> > +
> > if (argc && strlen(argv[0]) > 2 && strstarts("record", argv[0])) {
> > argc = __cmd_record(stat_options, &opt_mode, argc, argv);
> > if (argc < 0)
> > diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
> > index 45833244daf3..591bdf0b3e2a 100644
> > --- a/tools/perf/util/evlist.c
> > +++ b/tools/perf/util/evlist.c
> > @@ -369,11 +369,7 @@ static bool evlist__use_affinity(struct evlist *evlist)
> > struct perf_cpu_map *used_cpus = NULL;
> > bool ret = false;
> >
> > - /*
> > - * With perf record core.user_requested_cpus is usually NULL.
> > - * Use the old method to handle this for now.
> > - */
> > - if (!evlist->core.user_requested_cpus ||
> > + if (evlist->no_affinity || !evlist->core.user_requested_cpus ||
> > cpu_map__is_dummy(evlist->core.user_requested_cpus))
> > return false;
> >
> > diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h
> > index 30dff7484d3c..d17c3b57a409 100644
> > --- a/tools/perf/util/evlist.h
> > +++ b/tools/perf/util/evlist.h
> > @@ -59,6 +59,7 @@ struct event_enable_timer;
> > struct evlist {
> > struct perf_evlist core;
> > bool enabled;
> > + bool no_affinity;
> > int id_pos;
> > int is_pos;
> > int nr_br_cntr;
> > --
> > 2.53.0.rc2.204.g2597b5adb4-goog
> >
© 2016 - 2026 Red Hat, Inc.