On 12/1/25 14:41, Natalie Vock wrote:
> Otherwise userspace may be fooled into believing it has a reserved VMID
> when in reality it doesn't, ultimately leading to GPU hangs when SPM is
> used.
Good catch!
> Fixes: 80e709ee6ecc ("drm/amdgpu: add option params to enforce process
> isolation between graphics and compute")
> Cc: [email protected]
> Signed-off-by: Natalie Vock <[email protected]>
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 7 ++++---
> 1 file changed, 4 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> index 61820166efbf6..52f8038125530 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
> @@ -2913,6 +2913,7 @@ int amdgpu_vm_ioctl(struct drm_device *dev, void *data,
> struct drm_file *filp)
> struct amdgpu_device *adev = drm_to_adev(dev);
> struct amdgpu_fpriv *fpriv = filp->driver_priv;
> struct amdgpu_vm *vm = &fpriv->vm;
> + int r = 0;
Initializing local variables used as return code is usually seen as bad coding
style, but see below.
>
> /* No valid flags defined yet */
> if (args->in.flags)
> @@ -2921,16 +2922,16 @@ int amdgpu_vm_ioctl(struct drm_device *dev, void
> *data, struct drm_file *filp)
> switch (args->in.op) {
> case AMDGPU_VM_OP_RESERVE_VMID:
> /* We only have requirement to reserve vmid from gfxhub */
> - amdgpu_vmid_alloc_reserved(adev, vm, AMDGPU_GFXHUB(0));
> + r = amdgpu_vmid_alloc_reserved(adev, vm, AMDGPU_GFXHUB(0));
> break;
You can just use return amdgpu_vmid_alloc_reserved((..) here, no need for the
local variables.
Apart from that looks good to me.
Regards,
Christian.
> case AMDGPU_VM_OP_UNRESERVE_VMID:
> amdgpu_vmid_free_reserved(adev, vm, AMDGPU_GFXHUB(0));
> break;
> default:
> - return -EINVAL;
> + r = -EINVAL;
> }
>
> - return 0;
> + return r;
> }
>
> /**