perf intel-pt: Fix PEBS sample for XMM registers

The condition to add XMM registers was missing, the regs array needed to
be in the outer scope, and the size of the regs array was too small.

Fixes: 143d34a6b387b ("perf intel-pt: Add XMM registers to synthesized PEBS sample")
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Luwei Kang <luwei.kang@intel.com>
Link: http://lore.kernel.org/lkml/20200630133935.11150-4-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

diff --git a/tools/perf/util/intel-pt.c b/tools/perf/util/intel-pt.c
index e4dd8bf..cb3c1e5 100644 (file)
--- a/tools/perf/util/intel-pt.c
+++ b/tools/perf/util/intel-pt.c
@@ -1735,6 +1735,7 @@ static int intel_pt_synth_pebs_sample(struct intel_pt_queue *ptq)
        u64 sample_type = evsel->core.attr.sample_type;
        u64 id = evsel->core.id[0];
        u8 cpumode;
+       u64 regs[8 * sizeof(sample.intr_regs.mask)];
 
        if (intel_pt_skip_event(pt))
                return 0;
@@ -1784,8 +1785,8 @@ static int intel_pt_synth_pebs_sample(struct intel_pt_queue *ptq)
        }
 
        if (sample_type & PERF_SAMPLE_REGS_INTR &&
-           items->mask[INTEL_PT_GP_REGS_POS]) {
-               u64 regs[sizeof(sample.intr_regs.mask)];
+           (items->mask[INTEL_PT_GP_REGS_POS] ||
+            items->mask[INTEL_PT_XMM_POS])) {
                u64 regs_mask = evsel->core.attr.sample_regs_intr;
                u64 *pos;