Messages in this thread | | | Date | Mon, 14 Jul 2014 21:05:42 -0700 | From | Andi Kleen <> | Subject | Re: [PATCH 2/2] perf, x86: Don't mark DataLA addresses as store |
| |
On Tue, Jul 15, 2014 at 12:50:27AM +0200, Stephane Eranian wrote: > On Tue, Jul 15, 2014 at 12:49 AM, Stephane Eranian <eranian@google.com> wrote: > > On Tue, Jul 15, 2014 at 12:39 AM, Andi Kleen <ak@linux.intel.com> wrote: > >>> I have a problem with this patch. > >>> > >>> It makes: perf mem -t store rec record OP_NA for the store. > >>> It was recording OP_STORE before. > >>> > >>> I think we need to keep LD/ST info. This is useful for analysis > >>> especially if we collect loads/stores simultaneously. > >>> > >>> Was working before for the mem-loads, mem-stores events. > >> > >> Ok. Would it be enough if it only worked for "mem-stores" and not > >> all PEBS events? > >> > > Ok, do that at a minimum. > > > But if I recall the PEBS stores events were not that many to begin with.
Yes, there are only three store events:
MEM_UOPS_RETIRED.STLB_MISS_STORES MEM_UOPS_RETIRED.SPLIT_STORES MEM_UOPS_RETIRED.ALL_STORES
These can be added.
But most others are loads, so if you wanted loads too (besides mem-loads) it would be nearly a full list.
-Andi
| |