diff mbox series

[v2,1/6] arm64: Provide basic EL2 setup for FEAT_{LS64, LS64_V} usage at EL0/1

Message ID 20250331094320.35226-2-yangyicong@huawei.com
State New
Headers show
Series Add support for FEAT_{LS64, LS64_V} and related tests | expand

Commit Message

Yicong Yang March 31, 2025, 9:43 a.m. UTC
From: Yicong Yang <yangyicong@hisilicon.com>

Instructions introduced by FEAT_{LS64, LS64_V} is controlled by
HCRX_EL2.{EnALS, EnASR}. Configure all of these to allow usage
at EL0/1.

This doesn't mean these instructions are always available in
EL0/1 if provided. The hypervisor still have the control at
runtime.

Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
---
 arch/arm64/include/asm/el2_setup.h | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

Comments

Yicong Yang April 7, 2025, 3:50 a.m. UTC | #1
On 2025/4/3 17:04, Suzuki K Poulose wrote:
> On 31/03/2025 10:43, Yicong Yang wrote:
>> From: Yicong Yang <yangyicong@hisilicon.com>
>>
>> Instructions introduced by FEAT_{LS64, LS64_V} is controlled by
>> HCRX_EL2.{EnALS, EnASR}. Configure all of these to allow usage
>> at EL0/1.
>>
>> This doesn't mean these instructions are always available in
>> EL0/1 if provided. The hypervisor still have the control at
>> runtime.
>>
>> Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
>> ---
>>   arch/arm64/include/asm/el2_setup.h | 12 +++++++++++-
>>   1 file changed, 11 insertions(+), 1 deletion(-)
>>
>> diff --git a/arch/arm64/include/asm/el2_setup.h b/arch/arm64/include/asm/el2_setup.h
>> index ebceaae3c749..0259941602c4 100644
>> --- a/arch/arm64/include/asm/el2_setup.h
>> +++ b/arch/arm64/include/asm/el2_setup.h
>> @@ -57,9 +57,19 @@
>>           /* Enable GCS if supported */
>>       mrs_s    x1, SYS_ID_AA64PFR1_EL1
>>       ubfx    x1, x1, #ID_AA64PFR1_EL1_GCS_SHIFT, #4
>> -    cbz    x1, .Lset_hcrx_\@
>> +    cbz    x1, .Lskip_gcs_hcrx_\@
>>       orr    x0, x0, #HCRX_EL2_GCSEn
>>   +.Lskip_gcs_hcrx_\@:
> 
> minor nit: For consistency, could we rename this "set_ls64", similar to "set_hcrx" ?
> 

IIUC, set_xxx really touches the registers and skip_xxx should just check and prepare
the feature bits. so here using .Lskip_gcs_hrcx_\@ should be more proper and consistent
with other places in el2_setup.h, like __init_el2_debug/__init_el2_fgt which also use
.Lskip_xxx for skipping an unsupported feature?

Thanks.

>> +    /* Enable LS64, LS64_V if supported */
>> +    mrs_s    x1, SYS_ID_AA64ISAR1_EL1
>> +    ubfx    x1, x1, #ID_AA64ISAR1_EL1_LS64_SHIFT, #4
>> +    cbz    x1, .Lset_hcrx_\@
>> +    orr    x0, x0, #HCRX_EL2_EnALS
>> +    cmp    x1, #ID_AA64ISAR1_EL1_LS64_LS64_V
>> +    b.lt    .Lset_hcrx_\@
>> +    orr    x0, x0, #HCRX_EL2_EnASR
>> +
>>   .Lset_hcrx_\@:
>>       msr_s    SYS_HCRX_EL2, x0
>>   .Lskip_hcrx_\@:
> 
> Suzuki
> 
> .
Will Deacon April 29, 2025, 2:47 p.m. UTC | #2
On Mon, Mar 31, 2025 at 05:43:15PM +0800, Yicong Yang wrote:
> From: Yicong Yang <yangyicong@hisilicon.com>
> 
> Instructions introduced by FEAT_{LS64, LS64_V} is controlled by
> HCRX_EL2.{EnALS, EnASR}. Configure all of these to allow usage
> at EL0/1.
> 
> This doesn't mean these instructions are always available in
> EL0/1 if provided. The hypervisor still have the control at
> runtime.
> 
> Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
> ---
>  arch/arm64/include/asm/el2_setup.h | 12 +++++++++++-
>  1 file changed, 11 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/arm64/include/asm/el2_setup.h b/arch/arm64/include/asm/el2_setup.h
> index ebceaae3c749..0259941602c4 100644
> --- a/arch/arm64/include/asm/el2_setup.h
> +++ b/arch/arm64/include/asm/el2_setup.h
> @@ -57,9 +57,19 @@
>          /* Enable GCS if supported */
>  	mrs_s	x1, SYS_ID_AA64PFR1_EL1
>  	ubfx	x1, x1, #ID_AA64PFR1_EL1_GCS_SHIFT, #4
> -	cbz	x1, .Lset_hcrx_\@
> +	cbz	x1, .Lskip_gcs_hcrx_\@
>  	orr	x0, x0, #HCRX_EL2_GCSEn
>  
> +.Lskip_gcs_hcrx_\@:
> +	/* Enable LS64, LS64_V if supported */
> +	mrs_s	x1, SYS_ID_AA64ISAR1_EL1
> +	ubfx	x1, x1, #ID_AA64ISAR1_EL1_LS64_SHIFT, #4
> +	cbz	x1, .Lset_hcrx_\@
> +	orr	x0, x0, #HCRX_EL2_EnALS
> +	cmp	x1, #ID_AA64ISAR1_EL1_LS64_LS64_V
> +	b.lt	.Lset_hcrx_\@
> +	orr	x0, x0, #HCRX_EL2_EnASR

How do these HCRX settings survive a world-switch to and from a KVM
guest? AFAICT, the trap deactivation path there restores HCRX_HOST_FLAGS
when returning to the host. That would mean that the GCS code is broken
too, unless I'm missing something :/

Will
diff mbox series

Patch

diff --git a/arch/arm64/include/asm/el2_setup.h b/arch/arm64/include/asm/el2_setup.h
index ebceaae3c749..0259941602c4 100644
--- a/arch/arm64/include/asm/el2_setup.h
+++ b/arch/arm64/include/asm/el2_setup.h
@@ -57,9 +57,19 @@ 
         /* Enable GCS if supported */
 	mrs_s	x1, SYS_ID_AA64PFR1_EL1
 	ubfx	x1, x1, #ID_AA64PFR1_EL1_GCS_SHIFT, #4
-	cbz	x1, .Lset_hcrx_\@
+	cbz	x1, .Lskip_gcs_hcrx_\@
 	orr	x0, x0, #HCRX_EL2_GCSEn
 
+.Lskip_gcs_hcrx_\@:
+	/* Enable LS64, LS64_V if supported */
+	mrs_s	x1, SYS_ID_AA64ISAR1_EL1
+	ubfx	x1, x1, #ID_AA64ISAR1_EL1_LS64_SHIFT, #4
+	cbz	x1, .Lset_hcrx_\@
+	orr	x0, x0, #HCRX_EL2_EnALS
+	cmp	x1, #ID_AA64ISAR1_EL1_LS64_LS64_V
+	b.lt	.Lset_hcrx_\@
+	orr	x0, x0, #HCRX_EL2_EnASR
+
 .Lset_hcrx_\@:
 	msr_s	SYS_HCRX_EL2, x0
 .Lskip_hcrx_\@: