On Tue, Jul 23, 2024 at 06:59:07PM +0000, Matthew Brost wrote: > On Tue, Jul 23, 2024 at 01:52:46PM -0500, Dan Carpenter wrote: > > tree: https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master > > head: dee7f101b64219f512bb2f842227bd04c14efe30 > > commit: 61ac035361ae555ee5a17a7667fe96afdde3d59a [13298/13495] drm/xe: Drop xe_gt_tlb_invalidation_wait > > config: i386-randconfig-141-20240722 (https://download.01.org/0day-ci/archive/20240723/202407231049.esig0Fkb-lkp@xxxxxxxxx/config) > > compiler: clang version 18.1.5 (https://github.com/llvm/llvm-project 617a15a9eac96088ae5e9134248d8236e34b91b1) > > > > If you fix the issue in a separate patch/commit (i.e. not just a new version of > > the same patch/commit), kindly add following tags > > | Reported-by: kernel test robot <lkp@xxxxxxxxx> > > | Reported-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx> > > | Closes: https://lore.kernel.org/r/202407231049.esig0Fkb-lkp@xxxxxxxxx/ > > > > New smatch warnings: > > drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c:183 send_tlb_invalidation() warn: variable dereferenced before check 'fence' (see line 178) > > > > vim +/fence +183 drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c > > > > fc108a8b759f52 Matthew Brost 2023-01-17 159 static int send_tlb_invalidation(struct xe_guc *guc, > > 332dd0116c82a7 Matthew Brost 2023-01-24 160 struct xe_gt_tlb_invalidation_fence *fence, > > 332dd0116c82a7 Matthew Brost 2023-01-24 161 u32 *action, int len) > > a9351846d94568 Matthew Brost 2023-01-17 162 { > > a9351846d94568 Matthew Brost 2023-01-17 163 struct xe_gt *gt = guc_to_gt(guc); > > 501c4255c40935 Radhakrishna Sripada 2024-06-07 164 struct xe_device *xe = gt_to_xe(gt); > > a9351846d94568 Matthew Brost 2023-01-17 165 int seqno; > > a9351846d94568 Matthew Brost 2023-01-17 166 int ret; > > a9351846d94568 Matthew Brost 2023-01-17 167 > > 61ac035361ae55 Matthew Brost 2024-07-19 168 xe_gt_assert(gt, fence); > > We assert the fence is not NULL here. It is invalid call with a NULL fence. > This assert just does a drm_WARN(). It doesn't call called BUG() or do anything that affects flow analysis. These type of asserts are deliberately ignored by Smatch. With Smatch, I generally try not to print warnings about inconsistent NULL checking when I know the extra checks are harmless. If you had cross function analysis enabled then possibly that would silence the warning. (This code is too new for me to have tested it on my own system yet so I can't be sure if Smatch parses the callers correctly). Unfortunately, cross function stuff doesn't scale well enough so we can't use it on in the kbuild bot. regards, dan carpenter
