New issue
Advanced search Search tips

Issue 864908 link

Starred by 2 users

Issue metadata

Status: WontFix
Owner:
Closed: Aug 16
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 1
Type: Bug



Sign in to add a comment

binaryen_wasm_fuzzer set up incorrectly?

Project Member Reported by clemensh@chromium.org, Jul 18

Issue description

The new binaryen_wasm_fuzzer found two bugs:
1) https://clusterfuzz.com/v2/testcase-detail/5286025870704640?noredirect=1
2) https://clusterfuzz.com/v2/testcase-detail/6612060516122624?noredirect=1

But then it failed to reproduce them since it used the "linux_d8_dbg" bot instead of the original "linux_asan_d8_v8_arm_dbg". These are totally different bots, in particular one is executing 32-bit code, the other 64-bit.

Re-uploading the same test cases, and selecting "linux_asan_d8_v8_arm_dbg" manually worked, and we now get regression ranges and everything:
1) https://clusterfuzz.com/v2/testcase-detail/5189487907569664
2) https://clusterfuzz.com/v2/testcase-detail/6347432926117888

Is this a configuration error on the binaryen_wasm_fuzzer? Or what would cause it to try reproducing on another bot than the one which found the bug?
 
Status: WontFix (was: Assigned)
No it is not using the linux_db_dbg. linux_db_dbg is just used to provide a second stacktrace, in case any assert is interesting. that section shows "SECOND STACKTRACE". we always reproduce this with its job type which is linux_asan_d8_v8_arm_dbg

SECOND STACKTRACE ON REVISION 54504 (1 LINES)
No crash found using linux_d8_dbg job.

Reason it stopped reproducing was it was flaky.
[2018-07-17 20:31:57 UTC] clusterfuzz-linux-high-end-21wl: Progression task started: r54504.
[2018-07-17 20:32:34 UTC] clusterfuzz-linux-high-end-21wl: Progression task errored out: Known crash revision 53772 did not crash.
[2018-07-17 20:32:34 UTC] clusterfuzz-linux-high-end-21wl: Progression task errored out: Testcase appears to be flaky.
Thanks for clarifying!

Sign in to add a comment