Re: [PATCH 4/4] perf mem amd: Scan all PMUs instead of just core ones
From: Ravi Bangoria
Date: Tue Jun 13 2023 - 23:58:45 EST
>> @@ -165,7 +173,7 @@ static void perf_mem_events__print_unsupport_hybrid(struct perf_mem_event *e,
>> char sysfs_name[100];
>> struct perf_pmu *pmu = NULL;
>>
>> - while ((pmu = perf_pmus__scan_core(pmu)) != NULL) {
>> + while ((pmu = perf_mem_scan_next_pmu(pmu)) != NULL) {
>
> It was my mistake to optimize this,
Not really. I mean, there was already a bug which just got exacerbated.
> I think we can just go back to:
> perf_pmus__scan(pmu)
> which would remove a lot of the weak/macros etc. here. We can have a
> comment as to why this is scan not scan_core, because of AMD. I plan
> to further improve overhead of PMUs so I'm not worried about losing
> the small performance win from this.
Sure. Let me do that.
Thanks,
Ravi