-
Notifications
You must be signed in to change notification settings - Fork 788
[SYCL] Ensure correct errc thrown for unassociated placeholder accessor (4.7.6.9) #5737
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
…placeholder accessor (4.7.6.9) Signed-off-by: Chris Perkins <[email protected]>
/verify with intel/llvm-test-suite#890 |
@cperkinsintel It seems like this change can be covered by a unit test. Could you please add one? |
@sergey-semenov - I forgot to update the description. I already made a test for it here, this is also waiting for review: intel/llvm-test-suite#890 |
Right, I meant that since handling this error case is limited to the DPCPP runtime component (i.e. does not require running any device code), a unit test would probably be a better way to verify this as opposed to a full on E2E test. |
Signed-off-by: Chris Perkins <[email protected]>
Signed-off-by: Chris Perkins <[email protected]>
Signed-off-by: Chris Perkins <[email protected]>
Signed-off-by: Chris Perkins <[email protected]>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Since the feature needs to report at the time of enqueuing a kernel, E2E tests should be included anyway. As such I am fine to not have a unittest for it.
/verify with intel/llvm-test-suite#890 |
This got overlooked. All tests are passing. Needs merging. |
Update test to check for correct errc when attempting to use an unassociated placeholder accessor. Test for PR intel/llvm#5737 Signed-off-by: Chris Perkins [email protected]
…#890) Update test to check for correct errc when attempting to use an unassociated placeholder accessor. Test for PR intel#5737 Signed-off-by: Chris Perkins [email protected]
a simple fix for ensuring correct errc is thrown for an unassociated placeholder accessor (4.7.6.9)
Signed-off-by: Chris Perkins [email protected]
Note that this fix is not complete. It only addresses the most common cases. There are still cases where an unassociated placholder accessor could get by undetected and crash in the kernel. Unfortunately, the fix for those cases is not trivial. I have the outlines of a fix, but it would be disruptive, too disruptive for such a low priority fix. I will open a ticket about it and discuss it with various stakeholders.
Expanded testing to cover this here: intel/llvm-test-suite#890