On Thu, Nov 27, 2025 at 11:35:32AM +0530, Borah, Chaitanya Kumar wrote:
> Hello Christian,
>
> Hope you are doing well. I am Chaitanya from the linux graphics team in
> Intel.
>
> This mail is regarding a regression we are seeing in our CI runs[1] on
> linux-next repository.
>
> Since the version next-20251125 [2], we are seeing the following regression
>
> `````````````````````````````````````````````````````````````````````````````````
> (kms_busy:5818) sw_sync-CRITICAL: Test assertion failure function
> sw_sync_timeline_create_fence, file ../lib/sw_sync.c:117:
> (kms_busy:5818) sw_sync-CRITICAL: Failed assertion:
> sw_sync_fd_is_valid(fence)
> (kms_busy:5818) sw_sync-CRITICAL: Last errno: 2, No such file or directory
> (kms_busy:5818) sw_sync-CRITICAL: Created invalid fence
> (kms_busy:5818) igt_core-INFO: Stack trace:
> (kms_busy:5818) igt_core-INFO: #0 ../lib/igt_core.c:2075
> __igt_fail_assert()
> (kms_busy:5818) igt_core-INFO: #1 [sw_sync_timeline_create_fence+0x5f]
> (kms_busy:5818) igt_core-INFO: #2 ../tests/intel/kms_busy.c:122
> flip_to_fb()
> (kms_busy:5818) igt_core-INFO: #3 ../tests/intel/kms_busy.c:220
> test_flip()
> (kms_busy:5818) igt_core-INFO: #4 ../tests/intel/kms_busy.c:459
> __igt_unique____real_main411()
> (kms_busy:5818) igt_core-INFO: #5 ../tests/intel/kms_busy.c:411 main()
> (kms_busy:5818) igt_core-INFO: #6 [__libc_init_first+0x8a]
> (kms_busy:5818) igt_core-INFO: #7 [__libc_start_main+0x8b]
> (kms_busy:5818) igt_core-INFO: #8 [_start+0x25]
> `````````````````````````````````````````````````````````````````````````````````
> Details log can be found in [3].
>
> After bisecting the tree, the following patch [4] seems to be the first
> "bad" commit
>
> `````````````````````````````````````````````````````````````````````````````````````````````````````````
> commit 8459303c886151b71e8de08b73e384fd2bb7499c
> Author: Christian Brauner [email protected]
> Date: Sun Nov 23 17:33:55 2025 +0100
>
> dma: port sw_sync_ioctl_create_fence() to FD_PREPARE()
> `````````````````````````````````````````````````````````````````````````````````````````````````````````
>
> We also verified that if we revert the patch the issue is not seen.
>
> Could you please check why the patch causes this regression and provide a
> fix if necessary?
Gah, sorry about this:
diff --git a/drivers/dma-buf/sync_file.c b/drivers/dma-buf/sync_file.c
index dc2e79a1b196..8d827b03e84c 100644
--- a/drivers/dma-buf/sync_file.c
+++ b/drivers/dma-buf/sync_file.c
@@ -244,7 +244,8 @@ static long sync_file_ioctl_merge(struct sync_file
*sync_file,
if (copy_to_user((void __user *)arg, &data, sizeof(data)))
return -EFAULT;
- return fd_publish(fdf);
+ fd_publish(fdf);
+ return 0;
}
static int sync_fill_fence_info(struct dma_fence *fence,
Pushing out the fix now. Can I trigger a new test myself somehow?
