perf record: Disable profiling before draining the buffer

I noticed that perf-record continues profiling itself after the
child terminated and we're draining the buffer.

This can cause a _lot_ of overhead with --all recording - we keep
and keep recording, which produces new and new events.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

1 files changed, 12 insertions, 1 deletions

diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 5f3127e7a61..2459e5a22ed 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -524,6 +524,7 @@ static int __cmd_record(int argc, const char **argv)
         pid_t pid = 0;
         int flags;
         int ret;
+        unsigned long waking = 0;
 
         page_size = sysconf(_SC_PAGE_SIZE);
         nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
@@ -634,10 +635,20 @@ static int __cmd_record(int argc, const char **argv)
                 if (hits == samples) {
                         if (done)
                                 break;
-                        ret = poll(event_array, nr_poll, 100);
+                        ret = poll(event_array, nr_poll, -1);
+                        waking++;
+                }
+
+                if (done) {
+                        for (i = 0; i < nr_cpu; i++) {
+                                for (counter = 0; counter < nr_counters; counter++)
+                                        ioctl(fd[i][counter], PERF_COUNTER_IOC_DISABLE);
+                        }
                 }
         }
 
+        fprintf(stderr, "[ perf record: Woken up %ld times to write data ]\n", waking);
+
         /*
          * Approximate RIP event size: 24 bytes.
          */