[PATCH] drm/amdkfd: set precise mem ops caps to disabled for gfx 11 and 12

[PATCH] drm/amdkfd: set precise mem ops caps to disabled for gfx 11 and 12

[Jonathan Kim]

Clause instructions with precise memory enabled currently hang the
shader so set capabilities flag to disabled since it's unsafe to use
for debugging.

Signed-off-by: Jonathan Kim <jonathan.kim@amd.com>
---
 drivers/gpu/drm/amd/amdkfd/kfd_topology.c | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
index 27e7356eed6f..e477d7509646 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
@@ -2006,10 +2006,6 @@ static void kfd_topology_set_capabilities(struct kfd_topology_device *dev)
 		dev->node_props.debug_prop |= HSA_DBG_WATCH_ADDR_MASK_LO_BIT_GFX10 |
 					HSA_DBG_WATCH_ADDR_MASK_HI_BIT;
 
-		if (KFD_GC_VERSION(dev->gpu) >= IP_VERSION(11, 0, 0))
-			dev->node_props.capability |=
-				HSA_CAP_TRAP_DEBUG_PRECISE_MEMORY_OPERATIONS_SUPPORTED;
-
 		if (KFD_GC_VERSION(dev->gpu) >= IP_VERSION(12, 0, 0))
 			dev->node_props.capability |=
 				HSA_CAP_TRAP_DEBUG_PRECISE_ALU_OPERATIONS_SUPPORTED;
-- 
2.34.1

Re: [PATCH] drm/amdkfd: set precise mem ops caps to disabled for gfx 11 and 12

[Lancelot SIX]

On 14/03/2025 15:14, Jonathan Kim wrote:
> Clause instructions with precise memory enabled currently hang the
> shader so set capabilities flag to disabled since it's unsafe to use
> for debugging.
> 
> Signed-off-by: Jonathan Kim <jonathan.kim@amd.com>

Hi,

I have tested this for gfx11.
For what it is worth, this patch looks to me.

Tested-by: Lancelot Six <lancelot.six@amd.com>

Best,
Lancelot.

> ---
>   drivers/gpu/drm/amd/amdkfd/kfd_topology.c | 4 ----
>   1 file changed, 4 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
> index 27e7356eed6f..e477d7509646 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
> @@ -2006,10 +2006,6 @@ static void kfd_topology_set_capabilities(struct kfd_topology_device *dev)
>   		dev->node_props.debug_prop |= HSA_DBG_WATCH_ADDR_MASK_LO_BIT_GFX10 |
>   					HSA_DBG_WATCH_ADDR_MASK_HI_BIT;
>   
> -		if (KFD_GC_VERSION(dev->gpu) >= IP_VERSION(11, 0, 0))
> -			dev->node_props.capability |=
> -				HSA_CAP_TRAP_DEBUG_PRECISE_MEMORY_OPERATIONS_SUPPORTED;
> -
>   		if (KFD_GC_VERSION(dev->gpu) >= IP_VERSION(12, 0, 0))
>   			dev->node_props.capability |=
>   				HSA_CAP_TRAP_DEBUG_PRECISE_ALU_OPERATIONS_SUPPORTED;

RE: [PATCH] drm/amdkfd: set precise mem ops caps to disabled for gfx 11 and 12

[Kasiviswanathan, Harish]

[Public]

Reviewed-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>



-----Original Message-----
From: amd-gfx <amd-gfx-bounces@lists.freedesktop.org> On Behalf Of Lancelot SIX
Sent: Monday, March 17, 2025 7:02 AM
To: Kim, Jonathan <Jonathan.Kim@amd.com>; amd-gfx@lists.freedesktop.org
Subject: Re: [PATCH] drm/amdkfd: set precise mem ops caps to disabled for gfx 11 and 12



On 14/03/2025 15:14, Jonathan Kim wrote:
> Clause instructions with precise memory enabled currently hang the
> shader so set capabilities flag to disabled since it's unsafe to use
> for debugging.
>
> Signed-off-by: Jonathan Kim <jonathan.kim@amd.com>

Hi,

I have tested this for gfx11.
For what it is worth, this patch looks to me.

Tested-by: Lancelot Six <lancelot.six@amd.com>

Best,
Lancelot.

> ---
>   drivers/gpu/drm/amd/amdkfd/kfd_topology.c | 4 ----
>   1 file changed, 4 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
> index 27e7356eed6f..e477d7509646 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c
> @@ -2006,10 +2006,6 @@ static void kfd_topology_set_capabilities(struct kfd_topology_device *dev)
>               dev->node_props.debug_prop |= HSA_DBG_WATCH_ADDR_MASK_LO_BIT_GFX10 |
>                                       HSA_DBG_WATCH_ADDR_MASK_HI_BIT;
>
> -             if (KFD_GC_VERSION(dev->gpu) >= IP_VERSION(11, 0, 0))
> -                     dev->node_props.capability |=
> -                             HSA_CAP_TRAP_DEBUG_PRECISE_MEMORY_OPERATIONS_SUPPORTED;
> -
>               if (KFD_GC_VERSION(dev->gpu) >= IP_VERSION(12, 0, 0))
>                       dev->node_props.capability |=
>                               HSA_CAP_TRAP_DEBUG_PRECISE_ALU_OPERATIONS_SUPPORTED;