Re: [PATCH v1] KVM: arm64: selftests: Handle feature fields with nonzero minimum value correctly

From: Itaru Kitayama <itaru.kitayama@linux.dev>
To: Jing Zhang <jingzhangos@google.com>
Cc: KVM <kvm@vger.kernel.org>, KVMARM <kvmarm@lists.linux.dev>,
	ARMLinux <linux-arm-kernel@lists.infradead.org>,
	Marc Zyngier <maz@kernel.org>,
	Oliver Upton <oliver.upton@linux.dev>,
	Paolo Bonzini <pbonzini@redhat.com>,
	James Morse <james.morse@arm.com>,
	Suzuki K Poulose <suzuki.poulose@arm.com>,
	Zenghui Yu <yuzenghui@huawei.com>
Subject: Re: [PATCH v1] KVM: arm64: selftests: Handle feature fields with nonzero minimum value correctly
Date: Wed, 10 Jan 2024 07:27:28 +0900	[thread overview]
Message-ID: <ZZ3IUPwhLKNv98QZ@vm3> (raw)
In-Reply-To: <20240109165622.4104387-1-jingzhangos@google.com>

On Tue, Jan 09, 2024 at 08:56:21AM -0800, Jing Zhang wrote:
> There are some feature fields with nonzero minimum valid value. Make
> sure get_safe_value() won't return invalid field values for them.
> Also fix a bug that wrongly uses the feature bits type as the feature
> bits sign causing all fields as signed in the get_safe_value() and
> get_invalid_value().
> 
> Fixes: 54a9ea73527d ("KVM: arm64: selftests: Test for setting ID register from usersapce")
> Reported-by: Zenghui Yu <yuzenghui@huawei.com>
> Reported-by: Itaru Kitayama <itaru.kitayama@linux.dev>
> Signed-off-by: Jing Zhang <jingzhangos@google.com>
> ---
>  .../selftests/kvm/aarch64/set_id_regs.c       | 20 +++++++++++++++----
>  1 file changed, 16 insertions(+), 4 deletions(-)
> 
> diff --git a/tools/testing/selftests/kvm/aarch64/set_id_regs.c b/tools/testing/selftests/kvm/aarch64/set_id_regs.c
> index bac05210b539..f17454dc6d9e 100644
> --- a/tools/testing/selftests/kvm/aarch64/set_id_regs.c
> +++ b/tools/testing/selftests/kvm/aarch64/set_id_regs.c
> @@ -224,13 +224,20 @@ uint64_t get_safe_value(const struct reg_ftr_bits *ftr_bits, uint64_t ftr)
>  {
>  	uint64_t ftr_max = GENMASK_ULL(ARM64_FEATURE_FIELD_BITS - 1, 0);
>  
> -	if (ftr_bits->type == FTR_UNSIGNED) {
> +	if (ftr_bits->sign == FTR_UNSIGNED) {
>  		switch (ftr_bits->type) {
>  		case FTR_EXACT:
>  			ftr = ftr_bits->safe_val;
>  			break;
>  		case FTR_LOWER_SAFE:
> -			if (ftr > 0)
> +			uint64_t min_safe = 0;
> +
> +			if (!strcmp(ftr_bits->name, "ID_AA64DFR0_EL1_DebugVer"))
> +				min_safe = ID_AA64DFR0_EL1_DebugVer_IMP;
> +			else if (!strcmp(ftr_bits->name, "ID_DFR0_EL1_CopDbg"))
> +				min_safe = ID_DFR0_EL1_CopDbg_Armv8;
> +
> +			if (ftr > min_safe)
>  				ftr--;
>  			break;
>  		case FTR_HIGHER_SAFE:
> @@ -252,7 +259,12 @@ uint64_t get_safe_value(const struct reg_ftr_bits *ftr_bits, uint64_t ftr)
>  			ftr = ftr_bits->safe_val;
>  			break;
>  		case FTR_LOWER_SAFE:
> -			if (ftr > 0)
> +			uint64_t min_safe = 0;
> +
> +			if (!strcmp(ftr_bits->name, "ID_DFR0_EL1_PerfMon"))
> +				min_safe = ID_DFR0_EL1_PerfMon_PMUv3;
> +
> +			if (ftr > min_safe)
>  				ftr--;
>  			break;
>  		case FTR_HIGHER_SAFE:
> @@ -276,7 +288,7 @@ uint64_t get_invalid_value(const struct reg_ftr_bits *ftr_bits, uint64_t ftr)
>  {
>  	uint64_t ftr_max = GENMASK_ULL(ARM64_FEATURE_FIELD_BITS - 1, 0);
>  
> -	if (ftr_bits->type == FTR_UNSIGNED) {
> +	if (ftr_bits->sign == FTR_UNSIGNED) {
>  		switch (ftr_bits->type) {
>  		case FTR_EXACT:
>  			ftr = max((uint64_t)ftr_bits->safe_val + 1, ftr + 1);
> 

This fixes the issue seen on an AEM RevC FVP launched via the shrinkwrap
ns-edk2.yaml config.

[...]
# ok 79 ID_AA64ZFR0_EL1_SVEver
# # Totals: pass:79 fail:0 xfail:0 xpass:0 skip:0 error:0
ok 1 selftests: kvm: set_id_regs

Tested-by: Itaru Kitayama <itaru.kitayama@fujitsu.com>

Thanks,
Itaru.

> base-commit: 0dd3ee31125508cd67f7e7172247f05b7fd1753a
> -- 
> 2.43.0.472.g3155946c3a-goog
> 