> On 22 Jul 2026, at 2:30 am, Kyrylo Tkachov <[email protected]> wrote:
>
>
>
>> On 21 Jul 2026, at 18:03, Andi Kleen <[email protected]> wrote:
>>
>> On Tue, Jul 21, 2026 at 04:29:15PM +0200, [email protected] wrote:
>>> From: Kyrylo Tkachov <[email protected]>
>>>
>>> The autoprofile bootstrap support probes for Arm SPE from gcc/configure,
>>> gcc-auto-profile, and the profile-optimization testsuite support. Each
>>> bare "perf list" scans every event class even though arm_spe is a PMU.
>>> This can make the probe unnecessarily slow on systems with many events.
>>>
>>> Use "perf list pmu" at all three call sites. The PMU event class filter
>>> has long been supported by perf and still reports arm_spe. Stubbed BRBE,
>>> SPE, and fallback paths select the same profiler as before.
>>>
>>> This patch actually saves a few minutes of end-to-end bootstrap time on my
>>> SPE
>>> AArch64 many-core system.
>>
>> Or just -d /sys/devices/arm_spe ? (or what the right PMU name is)
>> But yes it's a clear improvement.
>
> I think it has to be /sys/bus/event_source/devices/arm_spe_0 after
> https://github.com/torvalds/linux/commit/4052ce07d5d7
test -d /sys/devices/arm_spe_0/ also seems to work
>
> Tbh "perf list pmu” looks easier to me unless there are concerns with perf
> availability or something?
>
>
>>
>> BTW do you actually see a faster compiler out of autoprofiledbootstrap?
>> I currently don't on x86.
>
> I haven’t done thorough evaluation of that myself (isolated compute time on
> my big build machines is hard to get exclusively at the moment), I was just
> picking some low hanging fruit on improving bootstrap times on aarch64.
>
> Thanks for looking at this,
> Kyrill
>
>>
>> -Andi
