[PATCH v3 02/10] perf: arm_spe: Support FEAT_SPEv1p4 filters
James Clark
james.clark at linaro.org
Tue Jul 15 04:23:08 PDT 2025
On 14/07/2025 2:26 pm, Will Deacon wrote:
> On Thu, Jun 05, 2025 at 11:49:00AM +0100, James Clark wrote:
>> FEAT_SPEv1p4 (optional from Armv8.8) adds some new filter bits, so
>> remove them from the previous version's RES0 bits using
>> PMSEVFR_EL1_RES0_V1P4_EXCL. It also makes some previously available bits
>> unavailable again, so add those back using PMSEVFR_EL1_RES0_V1P4_INCL.
>> E.g:
>>
>> E[30], bit [30]
>> When FEAT_SPEv1p4 is _not_ implemented ...
>>
>> FEAT_SPE_V1P3 has the same filters as V1P2 so explicitly add it to the
>> switch.
>>
>> Reviewed-by: Leo Yan <leo.yan at arm.com>
>> Tested-by: Leo Yan <leo.yan at arm.com>
>> Signed-off-by: James Clark <james.clark at linaro.org>
>> ---
>> arch/arm64/include/asm/sysreg.h | 7 +++++++
>> drivers/perf/arm_spe_pmu.c | 5 ++++-
>> 2 files changed, 11 insertions(+), 1 deletion(-)
>>
>> diff --git a/arch/arm64/include/asm/sysreg.h b/arch/arm64/include/asm/sysreg.h
>> index f1bb0d10c39a..880090df3efc 100644
>> --- a/arch/arm64/include/asm/sysreg.h
>> +++ b/arch/arm64/include/asm/sysreg.h
>> @@ -358,6 +358,13 @@
>> (PMSEVFR_EL1_RES0_IMP & ~(BIT_ULL(18) | BIT_ULL(17) | BIT_ULL(11)))
>> #define PMSEVFR_EL1_RES0_V1P2 \
>> (PMSEVFR_EL1_RES0_V1P1 & ~BIT_ULL(6))
>> +#define PMSEVFR_EL1_RES0_V1P4_EXCL \
>> + (BIT_ULL(2) | BIT_ULL(4) | GENMASK_ULL(10, 8) | GENMASK_ULL(23, 19))
>> +#define PMSEVFR_EL1_RES0_V1P4_INCL \
>> + (GENMASK_ULL(31, 26))
>> +#define PMSEVFR_EL1_RES0_V1P4 \
>> + (PMSEVFR_EL1_RES0_V1P4_INCL | \
>> + (PMSEVFR_EL1_RES0_V1P2 & ~PMSEVFR_EL1_RES0_V1P4_EXCL))
>>
>> /* Buffer error reporting */
>> #define PMBSR_EL1_FAULT_FSC_SHIFT PMBSR_EL1_MSS_SHIFT
>> diff --git a/drivers/perf/arm_spe_pmu.c b/drivers/perf/arm_spe_pmu.c
>> index 3efed8839a4e..d9f6d229dce8 100644
>> --- a/drivers/perf/arm_spe_pmu.c
>> +++ b/drivers/perf/arm_spe_pmu.c
>> @@ -701,9 +701,12 @@ static u64 arm_spe_pmsevfr_res0(u16 pmsver)
>> case ID_AA64DFR0_EL1_PMSVer_V1P1:
>> return PMSEVFR_EL1_RES0_V1P1;
>> case ID_AA64DFR0_EL1_PMSVer_V1P2:
>> + case ID_AA64DFR0_EL1_PMSVer_V1P3:
>> + return PMSEVFR_EL1_RES0_V1P2;
>> + case ID_AA64DFR0_EL1_PMSVer_V1P4:
>> /* Return the highest version we support in default */
>> default:
>> - return PMSEVFR_EL1_RES0_V1P2;
>> + return PMSEVFR_EL1_RES0_V1P4;
>
> See my reply [1] to Leo about this function, but I think we should just
> remove it.
>
> Will
>
> [1] https://lore.kernel.org/all/20250707-arm_spe_support_hitm_overhead_v1_public-v3-0-33ea82da3280@arm.com/
We're only refusing filters that we know for sure are RES0. Unless
there's a mistake, the ones that are maybes are still up to userspace to
decide whether it wants to use them or not.
I think it could be quite useful for some automated tool to fall back to
another behavior if it needs an event that isn't implemented.
If they were _all_ defined as maybes like "When FEAT_SPEv1p4 is
implemented or filtering on event 2 is optionally supported" then I
would agree it's not definite enough to bother restricting them. But a
lot of them are known for sure like "When FEAT_SPEv1p4 is not
implemented ...", so I don't see the harm in preventing use of those.
Or as I mentioned in the other thread if we think we can probe the valid
filters that would be even better.
Thanks
James
More information about the linux-arm-kernel
mailing list