Filed by: btara [at] mozilla.com https://treeherder.mozilla.org/logviewer.html#?job_id=215087705&repo=mozilla-inbound https://queue.taskcluster.net/v1/task/RYHbYIY4QIySvxMJu4rU9g/runs/0/artifacts/public/logs/live_backing.log 16:15:28 INFO - TEST-PASS | tests/jit-test/jit-test/tests/self-test/assertRecoveredOnBailout-1.js | Success (code 0, args "--ion-eager --ion-offthread-compile=off") [0.3 s] 16:15:28 INFO - {"action": "test_start", "jitflags": "--ion-eager --ion-offthread-compile=off", "pid": 477, "source": "jittests", "test": "self-test/assertRecoveredOnBailout-1.js", "thread": "main", "time": 1543680928.2161071} 16:15:28 INFO - {"action": "test_end", "extra": {"jitflags": "--ion-eager --ion-offthread-compile=off"}, "jitflags": "--ion-eager --ion-offthread-compile=off", "message": "Success", "pid": 477, "source": "jittests", "status": "PASS", "test": "self-test/assertRecoveredOnBailout-1.js", "thread": "main", "time": 1543680928.519289} 16:15:28 INFO - exception output: Segmentation fault 16:15:28 INFO - Segmentation faultSegmentation faultExit code: 139 16:15:28 INFO - FAIL - self-test/assertRecoveredOnBailout-1.js 16:15:28 WARNING - TEST-UNEXPECTED-FAIL | tests/jit-test/jit-test/tests/self-test/assertRecoveredOnBailout-1.js | Segmentation fault (code 139, args "--ion-eager --ion-offthread-compile=off --ion-check-range-analysis --ion-extra-checks --no-sse3 --no-threads") [0.2 s] 16:15:28 INFO - {"action": "test_start", "jitflags": "--ion-eager --ion-offthread-compile=off --ion-check-range-analysis --ion-extra-checks --no-sse3 --no-threads", "pid": 477, "source": "jittests", "test": "self-test/assertRecoveredOnBailout-1.js", "thread": "main", "time": 1543680928.520023} 16:15:28 INFO - {"action": "test_end", "extra": {"jitflags": "--ion-eager --ion-offthread-compile=off --ion-check-range-analysis --ion-extra-checks --no-sse3 --no-threads"}, "jitflags": "--ion-eager --ion-offthread-compile=off --ion-check-range-analysis --ion-extra-checks --no-sse3 --no-threads", "message": "Segmentation fault", "pid": 477, "source": "jittests", "status": "FAIL", "test": "self-test/assertRecoveredOnBailout-1.js", "thread": "main", "time": 1543680928.723086} 16:15:28 INFO - INFO exit-status : 139 16:15:28 INFO - INFO timed-out : False 16:15:28 INFO - INFO stdout > Segmentation fault 16:15:28 INFO - INFO stderr 2> Segmentation fault 16:15:28 INFO - TEST-PASS | tests/jit-test/jit-test/tests/self-test/assertRecoveredOnBailout-1.js | Success (code 0, args "--baseline-eager") [0.2 s]

Retriggers ended up in this merge https://treeherder.mozilla.org/#/jobs?repo=mozilla-inbound&resultStatus=success%2Cpending%2Crunning%2Ctestfailed%2Cbusted%2Cexception%2Crunnable&classifiedState=unclassified&tochange=6478fc93b0ce471c39e2f4efc8eea34135e0f6d7&fromchange=1395bbfd01b8b5ac84862684026f59ffa45cd46a&searchStr=android%2C8.0%2Cpixel2%2Caarch64%2Copt%2Ctest-android-hw-p2-8-0-android-aarch64%2Fopt-jittest-10%2C%28jit10%29&selectedJob=215134615 however since this is from the 10th of November, i think retriggers won't run further back. Aryx can you help here?

Those were failing before, just as tier 3: https://treeherder.mozilla.org/#/jobs?repo=mozilla-inbound&resultStatus=success%2Cpending%2Crunning%2Ctestfailed%2Cbusted%2Cexception%2Crunnable&classifiedState=unclassified&tier=1%2C2%2C3&tochange=6478fc93b0ce471c39e2f4efc8eea34135e0f6d7&fromchange=1395bbfd01b8b5ac84862684026f59ffa45cd46a&searchStr=android%2Cjit

I saw these run on inbound over the weekend but I didn't understand why they would do so since they are only enabled for try. They were classified as infra so I assumed it was some extraneous issue which caused them to run. These failures are known and tracked in bug 1475648 and bug 1454918. This should be marked invalid or duped to one of the other jit bugs for android-hw.

This test failure affects both ARMv7 and ARM64 and is blocking us from enabling ARM jit-tests in automation.

This is checking that we doing some optimization in IonMonkey, until we get IonMonkey working on ARM64, I will suggest to disable the test on ARM64 until we get IonMonkey working there.

Comment 6 suggests this also applies to ARM 32-bit FWIW...

Testing locally with patches to re-enable ARM64 IonMonkey backend, I am unable to reproduce this issue both on ARM64 and simulator. However, what is weird is: (In reply to Treeherder Bug Filer from comment #0) > 16:15:28 INFO - TEST-PASS | > tests/jit-test/jit-test/tests/self-test/assertRecoveredOnBailout-1.js | > Success (code 0, args "--ion-eager --ion-offthread-compile=off") [0.3 s] This should be the failure! An exit code of 0 on a test which is expected to crash is a failure. > 16:15:28 WARNING - TEST-UNEXPECTED-FAIL | > tests/jit-test/jit-test/tests/self-test/assertRecoveredOnBailout-1.js | > Segmentation fault (code 139, args "--ion-eager --ion-offthread-compile=off > --ion-check-range-analysis --ion-extra-checks --no-sse3 --no-threads") [0.2 > s] While this should be a successful run.

(In reply to Nicolas B. Pierron [:nbp] from comment #9) > Testing locally with patches to re-enable ARM64 IonMonkey backend Testing locally without these patches, I am still unable to reproduce on a simulator build. I will try on a native build tomorrow.

Trying with no patches, I still cannot reproduce natively on ARM64. I wonder if this is not just an error with running tests remotely.

Dan, would you be able to run the jit-test remotely, only on the self-test directory, and report if you are able to reproduce this issue? Or point me to the documentation to do the cross-compilation and remote testing procedure. I failed to reproduce this issue when testing locally (on ARM64, and x64 ARM64-simulator), and I do not have anything setup for testing remotely.

I'm sorry, I haven't worked on automation and testing for a few years and I'm no longer very familiar with the test harness or the remote testing procedure. I'm redirecting to Joel, I think someone on his team would be in a better position to help out.

:bc, can you look at this when finished with the battery test- I think we are getting closer to something actionable here and maybe there is something to fix in the remote setup or output handling.

64=wontfix because we probably don't need to uplift the fix to the GV 64 relbranch.

I've reproduced the issue with the exit code on a test with an expected failure. Investigating. Worst case, I can work around it.

Pushed by bclary@mozilla.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/2bdcac07076a js/src/tests/lib/jittests.py check_output should detect rc 139 as a crash, r=jmaher.