Linux x86_64 (Fedora 23) on opt and debug mozilla-central builds. 1. https://treeherder.allizom.org/#/jobs?repo=mozilla-inbound&exclusion_profile=false&filter-tier=1&filter-tier=2&filter-tier=3&fromchange=ac21d6f878e6&filter-searchStr=android&selectedJob=18125735 2. Click Repos and middle click mozilla-central to open mozilla-central in a new tab. 3. You may need to refresh or enter search terms in the filter. Using a local build from http://hg.mozilla.org/mozilla-central/rev/211a4c710fb6af2cad10102c4cabc7cb525998b8 and running the output through fix_linux_stack.py I hit this twice [Child 29164] ###!!! ABORT: constructor for actor failed: file /mozilla/builds/nightly/mozilla/firefox-debug/ipc/ipdl/PLayerTransactionChild.cpp, line 226 Hit MOZ_CRASH() at /mozilla/builds/nightly/mozilla/memory/mozalloc/mozalloc_abort.cpp:33 Without fix_linux_stack.py I hit [Child 13195] ###!!! ABORT: file /mozilla/builds/nightly/mozilla/ipc/chromium/src/base/shared_memory_posix.cc, line 221 [Child 14445] ###!!! ABORT: constructor for actor failed: file /mozilla/builds/nightly/mozilla/firefox-debug/ipc/ipdl/PLayerTransactionChild.cpp, line 177 Using mozregression on debug builds I found: https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=30a3de56668481084c10b898f6d4c4e0c7a5e208&tochange=3624090bdcf26024a075f9357f76537a01e7f838 82:06.65 LOG: MainThread main INFO Looks like the following bug has the changes which introduced the regression: https://bugzilla.mozilla.org/show_bug.cgi?id=1229437 I can't get good or any crash reports really and this generated a system core dump on my system. It makes Treeherder completely unusable on Nightly and I've been forced to use Chrome. :-(

This bug doesn't seem to relate to ruby at all. Why do you think it was bug 1229437 introduced this issue? I would rather suspect bug 1153156. masayuki, any idea about this crash?

Just from the bisection which point to the bug.

The commit range you showed includes three bugs... That was mainly because my patches for bug 1229437 broke the build until my followup 2 there.

Ah, sorry. I just went by mozregression's message that bug 1229437 was the regressor and didn't look at the range specifically. I'll remove the blocking and I see you've already ni the appropriate person. Thanks and sorry for the spam.

running stdout/stderr through fix_linux_stack.py

running stdout/stderr through fix_linux_stack.py

ABORT: file /mozilla/builds/nightly/mozilla/ipc/chromium/src/base/shared_memory_posix.cc, line 221

I have no idea... I'm not familiar with layers. If this is reproduced after my landing, sounds like my patch(es) made a hidden bug shown. My APZC patch did just modify its wheel event handling code. And other patches works only with Windows actually. If you can reproduce this without wheel operation, my patch shouldn't be a cause of this bug even on Windows.

I don't understand this. I attempted building debug manually and bisecting and found another range which doesn't make any sense. With today's Nightly I can't reproduce this the way I could earlier this week, so I'm going to mark this WFM. If I ever reproduce this with current builds and can get reliable regression ranges, I'll either reopen or refile. Sorry for the spam. If anyone knows which bug this may have been and if it was fixed, please let me know. Thanks.