perf tools: Fix multi-counter stat bug caused by incorrect reading of perf.data file...

Brice Goglin reported that only the first result from a
multi-counter perf record --stat run is accurate, the
rest looks bogus.

A silly mistake made us re-read the first attribute for
every recorded attribute.

Reported-by: Brice Goglin <[email protected]>
Signed-off-by: Peter Zijlstra <[email protected]>
Tested-by: Brice Goglin <[email protected]>
Cc: [email protected]
Signed-off-by: Ingo Molnar <[email protected]>

diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
index 450384b3bbe53605db82ec2dc676aac0405ee531..95a44bcfc2dc5c7042f57dfe3ed49e6d5b1398a2 100644 (file)
--- a/tools/perf/util/header.c
+++ b/tools/perf/util/header.c
@@ -213,9 +213,10 @@ struct perf_header *perf_header__read(int fd)
 
        for (i = 0; i < nr_attrs; i++) {
                struct perf_header_attr *attr;
-               off_t tmp = lseek(fd, 0, SEEK_CUR);
+               off_t tmp;
 
                do_read(fd, &f_attr, sizeof(f_attr));
+               tmp = lseek(fd, 0, SEEK_CUR);
 
                attr = perf_header_attr__new(&f_attr.attr);