[ADT test failure] linux: ubuntu_qrt_apparmor.test-apparmor.py -- ONEXEC - check current 'unconfined' != expected
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
AppArmor |
Fix Released
|
Medium
|
Tyler Hicks | ||
QA Regression Testing |
Invalid
|
Undecided
|
Unassigned | ||
apparmor (Ubuntu) |
Fix Released
|
Medium
|
Tyler Hicks | ||
Xenial |
Fix Released
|
Medium
|
Tyler Hicks |
Bug Description
[Impact]
The AppArmor regression tests are ran as part of the kernel SRU verification process. One of the tests, onexec.sh, contains a racy section that introduces intermittent failures. Such failures introduce doubt in the kernel SRU process and generate additional work to verify that the failures can be ignored.
[Test Case]
Since the fix is to the onexec.sh test itself, the best test case is to simply run the test. The QRT test-apparmor.py script runs this test, among many others, and should be used for verification.
[Regression Potential]
None. The fix is to the test and does not affect code that is user-facing.
[Original Report]
New ADT test failure in Vivid:
running onexec
ONEXEC - check current 'unconfined' != expected '/tmp/testlibIU
Fatal Error (onexec): Unable to run test sub-executable
description: | updated |
Changed in linux (Ubuntu): | |
status: | New → Confirmed |
tags: | added: apparmor |
tags: | added: qart |
tags: | removed: qart |
description: | updated |
no longer affects: | linux (Ubuntu Xenial) |
no longer affects: | linux (Ubuntu) |
Changed in apparmor: | |
status: | Fix Committed → Fix Released |
I think this was likely a problem (race condition) in the upstream test itself. I think I've been able to eliminate any races in the test with upstream commits r3488 and r3489.