Merge tag 'perf-tools-fixes-for-v5.18-2022-04-09' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux

Pull perf tools fixes from Arnaldo Carvalho de Melo:

 - Fix the clang command line option probing and remove some options to
   filter out, fixing the build with the latest clang versions

 - Fix 'perf bench' futex and epoll benchmarks to deal with machines
   with more than 1K CPUs

 - Fix 'perf test tsc' error message when not supported

 - Remap perf ring buffer if there is no space for event, fixing perf
   usage in 32-bit ChromeOS

 - Drop objdump stderr to avoid getting stuck waiting for stdout output
   in 'perf annotate'

 - Fix up garbled output by now showing unwind error messages when
   augmenting frame in best effort mode

 - Fix perf's libperf_print callback, use the va_args eprintf() variant

 - Sync vhost and arm64 cputype headers with the kernel sources

 - Fix 'perf report --mem-mode' with ARM SPE

 - Add missing external commands ('iiostat', etc) to 'perf --list-cmds'

* tag 'perf-tools-fixes-for-v5.18-2022-04-09' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux:
  perf annotate: Drop objdump stderr to avoid getting stuck waiting for stdout output
  perf tools: Add external commands to list-cmds
  perf docs: Add perf-iostat link to manpages
  perf session: Remap buf if there is no space for event
  perf bench: Fix epoll bench to correct usage of affinity for machines with #CPUs > 1K
  perf bench: Fix futex bench to correct usage of affinity for machines with #CPUs > 1K
  perf tools: Fix perf's libperf_print callback
  perf: arm-spe: Fix perf report --mem-mode
  perf unwind: Don't show unwind error messages when augmenting frame pointer stack
  tools headers arm64: Sync arm64's cputype.h with the kernel sources
  perf test tsc: Fix error message when not supported
  perf build: Don't use -ffat-lto-objects in the python feature test when building with clang-13
  perf python: Fix probing for some clang command line options
  tools build: Filter out options and warnings not supported by clang
  tools build: Use $(shell ) instead of `` to get embedded libperl's ccopts
  tools include UAPI: Sync linux/vhost.h with the kernel sources

1 files changed, 16 insertions, 5 deletions

diff --git a/tools/perf/bench/futex-lock-pi.c b/tools/perf/bench/futex-lock-pi.c
index 6fc9a3d55c1f..0abb3f7ee24f 100644
--- a/tools/perf/bench/futex-lock-pi.c
+++ b/tools/perf/bench/futex-lock-pi.c
@@ -120,11 +120,17 @@ static void *workerfn(void *arg)
 static void create_threads(struct worker *w, pthread_attr_t thread_attr,
 			   struct perf_cpu_map *cpu)
 {
-	cpu_set_t cpuset;
+	cpu_set_t *cpuset;
 	unsigned int i;
+	int nrcpus =  perf_cpu_map__nr(cpu);
+	size_t size;
 
 	threads_starting = params.nthreads;
 
+	cpuset = CPU_ALLOC(nrcpus);
+	BUG_ON(!cpuset);
+	size = CPU_ALLOC_SIZE(nrcpus);
+
 	for (i = 0; i < params.nthreads; i++) {
 		worker[i].tid = i;
 
@@ -135,15 +141,20 @@ static void create_threads(struct worker *w, pthread_attr_t thread_attr,
 		} else
 			worker[i].futex = &global_futex;
 
-		CPU_ZERO(&cpuset);
-		CPU_SET(perf_cpu_map__cpu(cpu, i % perf_cpu_map__nr(cpu)).cpu, &cpuset);
+		CPU_ZERO_S(size, cpuset);
+		CPU_SET_S(perf_cpu_map__cpu(cpu, i % perf_cpu_map__nr(cpu)).cpu, size, cpuset);
 
-		if (pthread_attr_setaffinity_np(&thread_attr, sizeof(cpu_set_t), &cpuset))
+		if (pthread_attr_setaffinity_np(&thread_attr, size, cpuset)) {
+			CPU_FREE(cpuset);
 			err(EXIT_FAILURE, "pthread_attr_setaffinity_np");
+		}
 
-		if (pthread_create(&w[i].thread, &thread_attr, workerfn, &worker[i]))
+		if (pthread_create(&w[i].thread, &thread_attr, workerfn, &worker[i])) {
+			CPU_FREE(cpuset);
 			err(EXIT_FAILURE, "pthread_create");
+		}
 	}
+	CPU_FREE(cpuset);
 }
 
 int bench_futex_lock_pi(int argc, const char **argv)