The perf benchmark for =
collections: numa, futex and epoll
hits failure in system configu=
ration with CPU's more than 1024.
These benchmarks uses "sched_ge=
taffinity" and "sched_setaffinity"
in the code to work with affin=
ity.

Example snippet from numa benchmark:
perf: bench/numa.c:302: bind_to_node: Assertio=
n `!(ret)' failed.
Aborted (core dumped)
<<>&g=
t;

bind_to_node function uses "sched_getaffinity" =
to save the cpumask.
This fails with EINVAL because the default m=
ask size in glibc is 1024.

Similarly in futex and =
epoll benchmark, uses sched_setaffinity during
pthread_create wit=
h affinity. And since it returns EINVAL in such system
configurat=
ion, benchmark doesn't run.

To overcome this 1024 =
CPUs mask size limitation of cpu_set_t,
change the mask size usin=
g the CPU_*_S macros ie, use CPU_ALLOC to
allocate cpumask, CPU_A=
LLOC_SIZE for size, CPU_SET_S to set mask bit.

Fix=
 all the relevant places in the code to use mask size which is large
<=
br>
Fix parse_setup_cpu_list function in numa bench to check if i=
nput CPU
is online before binding task to that CPU. This is to fi=
x failures where,
though CPU number is within max CPU, it could h=
appen that CPU is offline.
Here, sched_setaffinity will result in=
 failure when using cpumask having
that cpu bit set in the mask.<=
/pre>
Patch 1 and Patch 2 address fix for perf bench fut=
ex and perf bench
epoll benchmark. Patch 3 and Patch 4 address fi=
x in perf bench numa
benchmark

Athira Ra=
jeev (4):
  tools/perf: Fix perf bench futex to correct usage of =
affinity for
    machines with #CPUs > 1K
  tools/pe=
rf: Fix perf bench epoll to correct usage of affinity for
    mac=
hines with #CPUs > 1K
  tools/perf: Fix perf numa bench to fix=
 usage of affinity for machines
    with #CPUs > 1K
=
  tools/perf: Fix perf bench numa testcase to check if CPU used to bind    task is online

Tesed the patches on pow=
erpc with CPU > 1K and other configurations as well, verified the perf b=
enchmark numa, futex and epoll for whole series with the patch set.
Tested-by: Disha Goel <=
disgoel@linux.vnet.ibm.com>
=

 tools/perf/bench/epoll-ctl.c           |  25 ++++--<=
/pre> tools/perf/bench/epoll-wait.c          |  25 ++++--
 t=
ools/perf/bench/futex-hash.c          |  26 ++++--
 tools/perf/be=
nch/futex-lock-pi.c       |  21 +++--
 tools/perf/bench/futex-req=
ueue.c       |  21 +++--
 tools/perf/bench/futex-wake-parallel.c =
|  21 +++--
 tools/perf/bench/futex-wake.c          |  22 +++-- tools/perf/bench/numa.c                | 117 ++++++++++++++++++--=
-----
 tools/perf/util/header.c               |  43 +++++++++ tools/perf/util/header.h               |   1 +
 10 files =
changed, 252 insertions(+), 70 deletions(-)

--=-ofgAJE5RZSyPp+LpVbqp--