nv-tegra-mirror/linux-nvgpu
1
0
Fork 0
mirror of git://nv-tegra.nvidia.com/linux-nvgpu.git synced 2025-12-23 01:50:07 +03:00
Code Issues Packages Projects Releases Wiki Activity

gpu: nvgpu: disallow invalid syncpoint wait ids

Browse Source 
Instead of ignoring a wait when a raw syncpoint prefence has an invalid
id, reject the submit with -EINVAL just like with syncpoints in syncfds.

Change-Id: I9b5c417bd1c7cd081c79659d088ac2c915de8c0e
Signed-off-by: Konsta Holtta <kholtta@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1680281
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
This commit is contained in:
Konsta Holtta
2018-03-22 13:35:01 +02:00
committed by mobile promotions
parent bac51e8081
commit cb8d8337a6
2 changed files with 4 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
drivers/gpu/nvgpu/common/linux/channel.c
View File

@@ -462,11 +462,8 @@ static int gk20a_submit_prepare_syncs(struct channel_gk20a *c,
}
/*
* Optionally insert syncpt wait in the beginning of gpfifo submission
* when user requested and the wait hasn't expired. Validate that the id
* makes sense, elide if not. The only reason this isn't being
* unceremoniously killed is to keep running some tests which trigger
* this condition.
* Optionally insert syncpt/semaphore wait in the beginning of gpfifo
* submission when user requested and the wait hasn't expired.
*/
if (flags & NVGPU_SUBMIT_GPFIFO_FLAGS_FENCE_WAIT) {
int max_wait_cmds = c->deterministic ? 1 : 0;