Re: [PATCH] perf/x86/core: Simplify hybrid PMU registration failure path

From: Mi, Dapeng

Date: Tue Sep 08 2026 - 21:39:05 EST


LGTM. Thanks.

Reviewed-by: Dapeng Mi <dapeng1.mi@xxxxxxxxxxxxxxx>


On 9/5/2026 4:13 AM, Thorsten Blum wrote:
> Handle perf_pmu_register() errors directly instead of breaking out of
> the registration loop and checking the index separately. Reuse the loop
> index i to unregister any previously registered PMUs in reverse order
> and drop the extra index j.
>
> Also log the failing PMU's name and error code.
>
> Signed-off-by: Thorsten Blum <blum@xxxxxxxxxx>
> ---
> arch/x86/events/core.c | 18 ++++++++----------
> 1 file changed, 8 insertions(+), 10 deletions(-)
>
> diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c
> index 8eb14add44c8..0a8fcfd52ad0 100644
> --- a/arch/x86/events/core.c
> +++ b/arch/x86/events/core.c
> @@ -2256,7 +2256,7 @@ static int __init init_hw_perf_events(void)
> goto out2;
> } else {
> struct x86_hybrid_pmu *hybrid_pmu;
> - int i, j;
> + int i;
>
> for (i = 0; i < x86_pmu.num_hybrid_pmus; i++) {
> hybrid_pmu = &x86_pmu.hybrid_pmu[i];
> @@ -2268,15 +2268,13 @@ static int __init init_hw_perf_events(void)
>
> err = perf_pmu_register(&hybrid_pmu->pmu, hybrid_pmu->name,
> (hybrid_pmu->pmu_type == hybrid_big) ? PERF_TYPE_RAW : -1);
> - if (err)
> - break;
> - }
> -
> - if (i < x86_pmu.num_hybrid_pmus) {
> - for (j = 0; j < i; j++)
> - perf_pmu_unregister(&x86_pmu.hybrid_pmu[j].pmu);
> - pr_warn("Failed to register hybrid PMUs\n");
> - goto out2;
> + if (err) {
> + while (i--)
> + perf_pmu_unregister(&x86_pmu.hybrid_pmu[i].pmu);
> + pr_warn("Failed to register hybrid PMU %s: %d\n",
> + hybrid_pmu->name, err);
> + goto out2;
> + }
> }
> }
>