[PATCH] amdkfd: Do nto wait for queue op response during reset

[PATCH] amdkfd: Do nto wait for queue op response during reset

[Ahmad Rehman]

This patch adds the condition to not wait for
the queue response for unmap, if the gpu is in reset.

Signed-off-by: Ahmad Rehman <Ahmad.Rehman@amd.com>
---
 drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index 4fbe865ff279..0904c36192c7 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -2094,7 +2094,8 @@ int amdkfd_fence_wait_timeout(struct device_queue_manager *dqm,
 
 	while (*fence_addr != fence_value) {
 		/* Fatal err detected, this response won't come */
-		if (amdgpu_amdkfd_is_fed(dqm->dev->adev))
+		if (amdgpu_amdkfd_is_fed(dqm->dev->adev) ||
+				amdgpu_in_reset(dqm->dev->adev))
 			return -EIO;
 
 		if (time_after(jiffies, end_jiffies)) {
-- 
2.34.1

Re: [PATCH] amdkfd: Do nto wait for queue op response during reset

[Lazar, Lijo]

On 11/4/2025 11:03 PM, Ahmad Rehman wrote:
> This patch adds the condition to not wait for
> the queue response for unmap, if the gpu is in reset.
> 
> Signed-off-by: Ahmad Rehman <Ahmad.Rehman@amd.com>

Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>

Thanks,
Lijo

> ---
>   drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> index 4fbe865ff279..0904c36192c7 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> @@ -2094,7 +2094,8 @@ int amdkfd_fence_wait_timeout(struct device_queue_manager *dqm,
>   
>   	while (*fence_addr != fence_value) {
>   		/* Fatal err detected, this response won't come */
> -		if (amdgpu_amdkfd_is_fed(dqm->dev->adev))
> +		if (amdgpu_amdkfd_is_fed(dqm->dev->adev) ||
> +				amdgpu_in_reset(dqm->dev->adev))
>   			return -EIO;
>   
>   		if (time_after(jiffies, end_jiffies)) {

Re: [PATCH] amdkfd: Do nto wait for queue op response during reset

[Michel Dänzer]

On 11/4/25 18:33, Ahmad Rehman wrote:
> This patch adds the condition to not wait for
> the queue response for unmap, if the gpu is in reset.
> 
> Signed-off-by: Ahmad Rehman <Ahmad.Rehman@amd.com>
> ---
>  drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> index 4fbe865ff279..0904c36192c7 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
> @@ -2094,7 +2094,8 @@ int amdkfd_fence_wait_timeout(struct device_queue_manager *dqm,
>  
>  	while (*fence_addr != fence_value) {
>  		/* Fatal err detected, this response won't come */
> -		if (amdgpu_amdkfd_is_fed(dqm->dev->adev))
> +		if (amdgpu_amdkfd_is_fed(dqm->dev->adev) ||
> +				amdgpu_in_reset(dqm->dev->adev))
>  			return -EIO;
>  
>  		if (time_after(jiffies, end_jiffies)) {

Indentation of the second new line looks wrong, should be aligned to the opening parenthesis.


-- 
Earthling Michel Dänzer       \        GNOME / Xwayland / Mesa developer
https://redhat.com             \               Libre software enthusiast