[SYCL] xfail on zedebug #707
Conversation
Signed-off-by: Byoungro So <[email protected]>
|
I believe the two failures in pre-ci-windows are irrelevant to my changeset. |
| @@ -2,6 +2,7 @@ | |||
| // RUN: env SYCL_PI_TRACE=-1 SYCL_PROGRAM_COMPILE_OPTIONS=-DENV_COMPILE_OPTS SYCL_PROGRAM_LINK_OPTIONS=-DENV_LINK_OPTS SYCL_DEVICE_FILTER=%sycl_be %t.out | FileCheck %s | |||
| // REQUIRES: gpu | |||
| // UNSUPPORTED: cuda || hip | |||
| // XFAIL: ze_debug-1,ze_debug4 | |||
There was a problem hiding this comment.
@bso-intel - The XFAIL directive means the test will still be run, just expected to fail. Whereas the UNSUPPORTED directive prevents the test from ever being run. In this PR you are trying to prevent a memory leak on the CI system, no? If so, wouldn't using the UNSUPPORTED mechanism be better?
Also, if this is a temporary change while we await a fix, it might be best to also add a comment saying that to the test itself, rather than just the PR, so it doesn't get overlooked in the future.
There was a problem hiding this comment.
I know we could use //UNSUPPORTED, but we decided to use //XFAIL.
//XFAIL is exactly for this purpose to run it and find out if it passes some day in the future.
ze_debug-1 and ze_debug4 are all temporary markers.
There was a problem hiding this comment.
@cperkinsintel , please refer to #704.
I just add two more tests that are missed from PR-704.
Signed-off-by: Byoungro So <[email protected]>
Signed-off-by: Byoungro So <[email protected]>
Two test cases show memory leak on github.
So, we need to disable reporting failures tentatively.
Signed-off-by: Byoungro So [email protected]