Re: [PATCH] drm/amdgpu: Fix a deadlock if previous GEM object allocation fails

[Andrey Grodzovsky <andrey.grodzovsky@amd.com>]

On 2021-08-30 11:24 p.m., Pan, Xinhui wrote:
> [AMD Official Use Only]
>
> [AMD Official Use Only]
>
> Unreserve root BO before return otherwise next allocation got deadlock.
>
> Signed-off-by: xinhui pan <xinhui.pan@amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 11 +++++------
>   1 file changed, 5 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
> index 85b292ed5c43..c9db7d2c5816 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
> @@ -355,19 +355,18 @@ int amdgpu_gem_create_ioctl(struct drm_device *dev, void *data,
>                          DRM_DEBUG("Failed to allocate GEM object (%llu, %d, %llu, %d)\n",
>                                    size, initial_domain, args->in.alignment, r);
>                  }
> +
> +               if (flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID)
> +                       amdgpu_bo_unreserve(vm->root.bo);
>                  return r;
>          }
>
>          if (flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID) {
> -               if (!r) {
> -                       struct amdgpu_bo *abo = gem_to_amdgpu_bo(gobj);
> +               struct amdgpu_bo *abo = gem_to_amdgpu_bo(gobj);
>
> -                       abo->parent = amdgpu_bo_ref(vm->root.bo);
> -               }
> +               abo->parent = amdgpu_bo_ref(vm->root.bo);
>                  amdgpu_bo_unreserve(vm->root.bo);
>          }
> -       if (r)
> -               return r;


The above early return seems to be needed for -ERESTARTSYS case.

Andrey


>
>          r = drm_gem_handle_create(filp, gobj, &handle);
>          /* drop reference from allocate - handle holds it now */
> --
> 2.25.1