pthread mutex lock when include cuGraphicsEGLRegisterImage call

Autonomous Machines Jetson & Embedded Systems Jetson TX2
21

Hi,

Thanks for keeping update the status.
We will discuss this issue in detail today. Will update with you if there is any progress.

Thanks.

22

Hi,

There is a possible issue when we apply optimization in the copy_from_user.
We will try to revert the optimization to check if this fix the issue.

If time is important to you, you can also revert the change by your own.
Just update the following file in the kernel from the upstream:

arch/arm64/lib/copy_from_user.S
arch/arm64/lib/copy_in_user.S
arch/arm64/lib/copy_to_user.S

Thanks.

23

Hi

Can you provide a temporary files to me, I find three files copy_from_user.S copy_in_user.S copy_to_user.S
replace in the kernel, But kernel compile faile

24

Hi,

We will check this issue this week.
Will update information with you later.

Thanks for your patience.

25

Hi,

We found this error may be caused by other issue.

The error shared in #18 is a known bug found in kernel 3.18.
But the same issue doesn’t occur on kernel 4.4. since the copy_*_user.S are all same as upstream.
There is no NV optimization anymore.

We will keep investigating this issue and sorry for the inconvenience.
Thanks.

26

Hi

We are waiting for the solution, Our production have got into hot water due to the issue

27

Hi,

If this problem has any solution ?

Thanks.

28

Hi,

Thanks for your patience.
Current status is that we have found the root cause and fix it internally.

Thanks.

29

Hi,

How do we use the revised version ?

Thanks.

30

Hi,

If We can use the newest version?

Thanks.

31

Hi,

The fix should be available in the rel-28.4.
Please pay attention to our announcement for the future release.

Thanks.

32

Hi

How about the other problem [url]https://devtalk.nvidia.com/default/topic/1045889/cueglstreamconsumerconnect-and-cueglstreamconsumerdisconnect-function-would-stuck-in-multithread-frequent-call-/#5307106[/url]? do it fix together in the rel-28.4 ?

33

Hi,

We will check this and update information with you.
Thanks.

34

Hi,

The issue of topic 1045889 is also fixed in the rel-28.4.
Thanks.

35

Hi

when we can use rel-28.4? do it have time schedule?

36

Hi,

Sorry that we cannot disclose our future plan here.
Please pay attention to our announcement to get the software update.

Thanks.

37

Hi,

if this issue was fixed in the R32.1 ?

38

Hi,

Sorry that the fix is merged into rel-28.4.
Thanks.

39

As one of the groups affected by this bug we are very interested in getting the fix. Currently we copy image files to get around the bug, but that adds a significant lag to our processing, so we would like to transition back to using mapping as soon as possible.
We have locked our system at 28.2.1 to provide a stable development environment for our engineering team. Would you please make the correction available as a patch to the 28.2.1 LTS kernel?

40

Hi,

Sorry that the fix is in the GPU driver rather than kernel.
You will need to wait for our next release to fix this issue.

Good news is that this fix is catching up the release of rel-28.3 branch.
You should be able to get the fix soon.
Sorry that you will still need to wait for rel-28.4 for the fix.

Thanks.