On 3/18/26 14:31, Ruijing Dong wrote:
> amdgpu_device_get_job_timeout_settings() passes a pointer directly
> to the global amdgpu_lockup_timeout[] buffer into strsep().
> strsep() destructively replaces delimiter characters with '\0'
> in-place.
> 
> On multi-GPU systems, this function is called once per device.
> When a multi-value setting like "0,0,0,-1" is used, the first
> GPU's call transforms the global buffer into "0\00\00\0-1". The
> second GPU then sees only "0" (terminated at the first '\0'),
> parses a single value, hits the single-value fallthrough
> (index == 1), and applies timeout=0 to all rings — causing
> immediate false job timeouts.
> 
> Fix this by copying into a stack-local array before calling
> strsep(), so the global module parameter buffer remains intact
> across calls. The buffer is AMDGPU_MAX_TIMEOUT_PARAM_LENGTH
> (256) bytes, which is safe for the stack.
> 
> v2: wrap commit message to 72 columns, add Assisted-by tag.
> v3: use stack array with strscpy() instead of kstrdup()/kfree()
>     to avoid unnecessary heap allocation (Christian).
> 
> Assisted-by: Claude:claude-opus-4-6
> Signed-off-by: Ruijing Dong <[email protected]>

Reviewed-by: Christian König <[email protected]>

> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 13 +++++++++++--
>  1 file changed, 11 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c 
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> index dcae77b6c272..fbe553c38583 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> @@ -3498,7 +3498,8 @@ static void amdgpu_device_xgmi_reset_func(struct 
> work_struct *__work)
>  
>  static int amdgpu_device_get_job_timeout_settings(struct amdgpu_device *adev)
>  {
> -     char *input = amdgpu_lockup_timeout;
> +     char buf[AMDGPU_MAX_TIMEOUT_PARAM_LENGTH];
> +     char *input = buf;
>       char *timeout_setting = NULL;
>       int index = 0;
>       long timeout;
> @@ -3508,9 +3509,17 @@ static int 
> amdgpu_device_get_job_timeout_settings(struct amdgpu_device *adev)
>       adev->gfx_timeout = adev->compute_timeout = adev->sdma_timeout =
>               adev->video_timeout = msecs_to_jiffies(2000);
>  
> -     if (!strnlen(input, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH))
> +     if (!strnlen(amdgpu_lockup_timeout, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH))
>               return 0;
>  
> +     /*
> +      * strsep() destructively modifies its input by replacing delimiters
> +      * with '\0'. Use a stack copy so the global module parameter buffer
> +      * remains intact for multi-GPU systems where this function is called
> +      * once per device.
> +      */
> +     strscpy(buf, amdgpu_lockup_timeout, sizeof(buf));
> +
>       while ((timeout_setting = strsep(&input, ",")) &&
>              strnlen(timeout_setting, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH)) {
>               ret = kstrtol(timeout_setting, 0, &timeout);

Reply via email to