[MSAN] mark test as unsupported for powerpc.

[kda]

Summary

The PowerPC implementation seems to return EINVAL (22) for pthread_getaffinity_np.
Without access to source code, it is hard to know why, as parameters appear to be valid.

Drops changes added in these commits:
https://reviews.llvm.org/D127359
https://reviews.llvm.org/D127346
https://reviews.llvm.org/D127320

Depends on: https://reviews.llvm.org/D127327

[vitalybuka]

compiler-rt/test/sanitizer_common/TestCases/Linux/pthread_getaffinity_np.cpp

9

typo

9

"no access to source code" -> "no access to the bot"?

code should be just regular glibc

[MaskRay]

compiler-rt/test/sanitizer_common/TestCases/Linux/pthread_getaffinity_np.cpp

9

I tested this. It passed on a 5.10.0-14-powerpc64le kernel. Do you have a link to the failure?

[vitalybuka]

https://lab.llvm.org/buildbot/#/builders/105/builds/26496
and
https://lab.llvm.org/buildbot/#/builders/19/builds/11181

[nemanjai]

In D127364#3568733, @vitalybuka wrote:

https://lab.llvm.org/buildbot/#/builders/105/builds/26496
and
https://lab.llvm.org/buildbot/#/builders/19/builds/11181

These are running on a machine that has 1536 CPUs:

$ lscpu
Architecture:        ppc64le
Byte Order:          Little Endian
CPU(s):              1536
On-line CPU(s) list: 0-1535
Thread(s) per core:  8
Core(s) per socket:  12
Socket(s):           16
NUMA node(s):        16

and the man page mentions:
EINVAL (pthread_getaffinity_np()) cpusetsize is smaller than the size of the affinity mask used by the kernel.
Could that have something to do with it?

[MaskRay]

In D127364#3568851, @nemanjai wrote:

In D127364#3568733, @vitalybuka wrote:

https://lab.llvm.org/buildbot/#/builders/105/builds/26496
and
https://lab.llvm.org/buildbot/#/builders/19/builds/11181

These are running on a machine that has 1536 CPUs:

$ lscpu
Architecture:        ppc64le
Byte Order:          Little Endian
CPU(s):              1536
On-line CPU(s) list: 0-1535
Thread(s) per core:  8
Core(s) per socket:  12
Socket(s):           16
NUMA node(s):        16

and the man page mentions:
EINVAL (pthread_getaffinity_np()) cpusetsize is smaller than the size of the affinity mask used by the kernel.
Could that have something to do with it?

I read a bit of Linux kernel code. D127368 should have fixed the test :)

[kda]

@MaskRay @nemanjai : Thanks for all the help. I'm dropping this.

compiler-rt/test/sanitizer_common/TestCases/Linux/pthread_getaffinity_np.cpp

 // RUN: %clangxx -O0 %s -o %t && %run %t
 
 // Android does not implement pthread_getaffinity_np.
 // (Note: libresolv is integrated with libc, but apparently only
 // sched_getaffinity).
 // UNSUPPORTED: android
 
-#ifndef _GNU_SOURCE
-#  define _GNU_SOURCE
-#endif
+// PowerPC implementation of pthread_getaffinity_np returns EINVAL (22).
+// Why? Uknown.  (no access to source code)
+// UNSUPPORTED: powerpc64, powerpc64le
 
 #include <assert.h>
 #include <pthread.h>
 #include <stdio.h>
 #include <sys/sysinfo.h>
 
 #include <sanitizer/msan_interface.h>
 
 int main() {
   cpu_set_t set_x;
-  pthread_t tid = pthread_self();
-  int res = pthread_getaffinity_np(tid, sizeof(set_x), &set_x);
-  if (res != 0) {
-    fprintf(stderr, "tid: %lu\n", tid);
-    fprintf(stderr, "sizeof(set_x): %d\n", sizeof(set_x));
-    fprintf(stderr, "res: %d\n", res);
-  }
+  int res = pthread_getaffinity_np(pthread_self(), sizeof(set_x), &set_x);
   assert(res == 0);
   assert(CPU_COUNT_S(sizeof(set_x), &set_x) == get_nprocs());
 
   return 0;
 }