Re: [PATCH] perf/x86/amd: Run AMD BRS code only on supported hw

From: Stephane Eranian
Date: Mon May 16 2022 - 21:00:03 EST


On Mon, May 16, 2022 at 8:48 AM Borislav Petkov <bp@xxxxxxxxx> wrote:
>
> From: Borislav Petkov <bp@xxxxxxx>
>
> This fires on a Fam16h machine here:
>
> unchecked MSR access error: WRMSR to 0xc000010f (tried to write 0x0000000000000018) \
> at rIP: 0xffffffff81007db1 (amd_brs_reset+0x11/0x50)
> Call Trace:
> <TASK>
> amd_pmu_cpu_starting
> ? x86_pmu_dead_cpu
> x86_pmu_starting_cpu
> cpuhp_invoke_callback
> ? x86_pmu_starting_cpu
> ? x86_pmu_dead_cpu
> cpuhp_issue_call
> ? x86_pmu_starting_cpu
> __cpuhp_setup_state_cpuslocked
> ? x86_pmu_dead_cpu
> ? x86_pmu_starting_cpu
> __cpuhp_setup_state
> ? map_vsyscall
> init_hw_perf_events
> ? map_vsyscall
> do_one_initcall
> ? _raw_spin_unlock_irqrestore
> ? try_to_wake_up
> kernel_init_freeable
> ? rest_init
> kernel_init
> ret_from_fork
>
> because that CPU hotplug callback gets executed on any AMD CPU - not
> only on the BRS-enabled ones. Check the BRS feature bit properly.
>
Thanks for catching this. I tested on Zen2 (Fam17h) and it did not
trigger the problem, yet I think
the problem in amd_brs_reset() is there.

> Signed-off-by: Borislav Petkov <bp@xxxxxxx>
Acked-By: Stephane Eranian <eranian@xxxxxxxxxx>

> ---
> arch/x86/events/amd/brs.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/arch/x86/events/amd/brs.c b/arch/x86/events/amd/brs.c
> index 895c82165d85..bee8765a1e9b 100644
> --- a/arch/x86/events/amd/brs.c
> +++ b/arch/x86/events/amd/brs.c
> @@ -57,7 +57,7 @@ static inline u64 get_debug_extn_cfg(void)
>
> static bool __init amd_brs_detect(void)
> {
> - if (!boot_cpu_has(X86_FEATURE_BRS))
> + if (!cpu_feature_enabled(X86_FEATURE_BRS))
> return false;
>
> switch (boot_cpu_data.x86) {
> @@ -112,6 +112,9 @@ static inline int amd_brs_get_tos(union amd_debug_extn_cfg *cfg)
> */
> void amd_brs_reset(void)
> {
> + if (!cpu_feature_enabled(X86_FEATURE_BRS))
> + return;
> +
> /*
> * Reset config
> */
> --
> 2.35.1
>