gpu: nvgpu: Syncpoint invalid value to ~0.

As qnx syncpoint's invalid value is ~0, change the code
to handle this.

Bug 200603716

Change-Id: I5ec79688cd9e60066725781f1effe57692ec0c27
Signed-off-by: Dinesh <dt@nvidia.com>
(cherry picked from commit 705260565a75bc90683841c4c08e4c857bda39f0)
Reviewed-on: https://git-master.nvidia.com/r/c/linux-nvgpu/+/2331208
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
This commit is contained in:
Dinesh
2020-04-06 16:23:24 +05:30
committed by Alex Waterman
parent c1521a7bba
commit 1c1da3d6b4
6 changed files with 19 additions and 5 deletions

View File

@@ -354,7 +354,13 @@ nvgpu_channel_sync_syncpt_create(struct nvgpu_channel *c)
sp->id = nvgpu_nvhost_get_syncpt_host_managed(sp->nvhost,
c->chid, syncpt_name);
if (sp->id == 0U) {
/**
* This is a WAR to handle invalid value of a syncpt.
* Once nvhost update the return value as NVGPU_INVALID_SYNCPT_ID,
* we can remove the zero check.
*/
if ((sp->id == 0U) ||
(sp->id == NVGPU_INVALID_SYNCPT_ID)) {
nvgpu_kfree(c->g, sp);
nvgpu_err(c->g, "failed to get free syncpt");
return NULL;