Am 04.08.2016 um 17:42 schrieb Amitoj Kaur Chawla:

To indicate an error, debugfs_create_file can return an ERR_PTR for !CONFIG_DEBUG_FS and NULL otherwise, so in case the result is dereferenced there should be a previous IS_ERR and a NULL check.

The Coccinelle semantic patch used to find the issue is as follows: @@ expression e; identifier f; @@

• e = debugfs_create_file(...);

... when != e == NULL e->f

Signed-off-by: Amitoj Kaur Chawla amitoj1606@gmail.com

The whole section is surrounded by CONFIG_DEBUG_FS, so I suggest to just replace the check or IS_ERR with a NULL check.

Regards, Christian.

---

drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 4 ++++ 1 file changed, 4 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index 8a1752f..9ff70a7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -1198,6 +1198,8 @@ static int amdgpu_ttm_debugfs_init(struct amdgpu_device *adev) adev, &amdgpu_ttm_vram_fops); if (IS_ERR(ent)) return PTR_ERR(ent);

• if (!ent)

• return -ENOMEM;
  

  i_size_write(ent->d_inode, adev->mc.mc_vram_size); adev->mman.vram = ent;

@@ -1205,6 +1207,8 @@ static int amdgpu_ttm_debugfs_init(struct amdgpu_device *adev) adev, &amdgpu_ttm_gtt_fops); if (IS_ERR(ent)) return PTR_ERR(ent);

• if (!ent)

• return -ENOMEM;
  

  i_size_write(ent->d_inode, adev->mc.gtt_size); adev->mman.gtt = ent;