1367544 - crashtest crash in e10s just hangs the crashtest harness

Status: NEW

(Testing :: Reftest, defect, P3)

Description

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I was locally testing the crashtest for bug 1367413 -- i.e., testing with the crashtest but without the code change -- and I got a crash in the crashtest harness from the fatal assertion, but then the harness just hung.  It seems like it ought to report a crash and then either exit or (per bug 1344991) continue.

The end of the output was:


Assertion failure: kidOverflowBEnd >= kidBEnd, at /home/dbaron/builds/mozilla-central/mozilla/layout/generic/nsAbsoluteContainingBlock.cpp:154 in Reflow
#01: nsAbsoluteContainingBlock::Reflow(nsContainerFrame*, nsPresContext*, mozilla::ReflowInput const&, nsReflowStatus&, nsRect const&, nsAbsoluteContainingBlock::AbsPosReflowFlags, nsOverflowAreas*) (/home/dbaron/builds/mozilla-central/mozilla/layout/generic/nsAbsoluteContainingBlock.cpp:154 (discriminator 1))
...
#64: _start (/home/dbaron/builds/mozilla-central/obj/firefox-debugopt/dist/bin/firefox)
#65: ??? (???:???)
[Parent 11253] WARNING: pipe error (70): Connection reset by peer: file /home/dbaron/builds/mozilla-central/mozilla/ipc/chromium/src/chrome/common/ipc_channel_posix.cc, line 353
[Parent 11253] WARNING: pipe error (71): Connection reset by peer: file /home/dbaron/builds/mozilla-central/mozilla/ipc/chromium/src/chrome/common/ipc_channel_posix.cc, line 353
###!!! [Parent][MessageChannel] Error: (msgtype=0x2C0071,name=PBrowser::Msg_RealMouseButtonEvent) Channel error: cannot send/recv
###!!! [Parent][RunMessage] Error: Channel error: cannot send/recv
###!!! [Parent][MessageChannel] Error: (msgtype=0x2C0071,name=PBrowser::Msg_RealMouseButtonEvent) Channel error: cannot send/recv
###!!! [Parent][MessageChannel] Error: (msgtype=0x2C006F,name=PBrowser::Msg_RealMouseMoveEvent) Channel error: cannot send/recv
###!!! [Parent][RunMessage] Error: Channel error: cannot send/recv
###!!! [Parent][MessageChannel] Error: (msgtype=0x2C0071,name=PBrowser::Msg_RealMouseButtonEvent) Channel error: cannot send/recv
###!!! [Parent][MessageChannel] Error: (msgtype=0x2C0071,name=PBrowser::Msg_RealMouseButtonEvent) Channel error: cannot send/recv
###!!! [Parent][RunMessage] Error: Channel error: cannot send/recv
###!!! [Parent][RunMessage] Error: Channel error: cannot send/recv
###!!! [Parent][RunMessage] Error: Channel error: cannot send/recv
###!!! [Parent][RunMessage] Error: Channel error: cannot send/recv
###!!! [Parent][RunMessage] Error: Channel error: cannot send/recv
###!!! [Parent][RunMessage] Error: Channel error: cannot send/recv
###!!! [Parent][RunMessage] Error: Channel error: cannot send/recv
###!!! [Parent][MessageChannel] Error: (msgtype=0x2C0081,name=PBrowser::Msg_Destroy) Channel error: cannot send/recv
JavaScript error: resource:///modules/ContentCrashHandlers.jsm, line 135: TypeError: WeakMap key must be an object, got undefined
--DOCSHELL 0x55bbc0d12ea0 == 4 [pid = 11253] [id = {50e0a5f5-f511-43b1-b5a6-33b9b3b17098}]
--DOCSHELL 0x55bbc00d6500 == 3 [pid = 11253] [id = {a04e7afd-106d-4d08-81fb-04d8e78d1504}]
--DOMWINDOW == 14 (0x55bbbfb48f00) [pid = 11253] [serial = 2] [outer = (nil)] [url = about:blank]
--DOMWINDOW == 13 (0x55bbc16d1a00) [pid = 11253] [serial = 8] [outer = (nil)] [url = about:blank]
--DOMWINDOW == 12 (0x55bbc0ced530) [pid = 11253] [serial = 6] [outer = (nil)] [url = about:blank]
--DOMWINDOW == 11 (0x55bbc0d13540) [pid = 11253] [serial = 7] [outer = (nil)] [url = about:blank]
--DOMWINDOW == 10 (0x55bbc1d1c940) [pid = 11253] [serial = 10] [outer = (nil)] [url = about:blank]
--DOMWINDOW == 9 (0x55bbc1d247b0) [pid = 11253] [serial = 11] [outer = (nil)] [url = about:blank]
--DOMWINDOW == 8 (0x55bbc12b9130) [pid = 11253] [serial = 9] [outer = (nil)] [url = about:blank]
JavaScript error: file:///home/dbaron/builds/mozilla-central/obj/firefox-debugopt/dist/bin/browser/components/nsBrowserGlue.js, line 983: TypeError: win is null


It then just paused, and then repeated this message every minute or so:

JavaScript error: file:///home/dbaron/builds/mozilla-central/obj/firefox-debugopt/dist/bin/browser/components/nsBrowserGlue.js, line 983: TypeError: win is null

until I stopped it by pressing Ctrl+C.

My tree is based on 093eab57d504, with local patches.

Comment 2

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Given that there are 2 reports of this in a week, I suspect this is a recent regression.

Shing, is it possible this is a regression from bug 1359288, bug 1344991, or similar?

Comment 3

[Nils Ohlmeier [:drno]]

FYI when I ran crashtest with a whole directory of tests it hang with the same JS error messages as in the original report. But the hanging window finally got killed with the infamous "330 seconds with no output" by the test framework, which then appeared to proceed to run something like self-tests (?). Eventually ./mach exited with TEST-PASS, as if it had not detected the test failures. Although the output obviously contained a couple of UNEXPECTED, so these should hopefully get the attention of test framework on treeherder.

Comment 4

[(not currently active) Ted Mielczarek]

That JS error is from this line:
https://dxr.mozilla.org/mozilla-central/rev/39d5cc0fda5e16c49a59d29d4ca186a5534cc88b/browser/modules/ContentCrashHandlers.jsm#135

It looks like `permanentKey` is managed by tabbrowser:
https://dxr.mozilla.org/mozilla-central/rev/39d5cc0fda5e16c49a59d29d4ca186a5534cc88b/browser/base/content/tabbrowser.xml#2007

...which reftest doesn't use, so that code probably would never have worked in reftest. That code is several months old, however.

Comment 5

[Kartikaya Gupta (email:[e-mail])]

Dropping stale needinfo to get this bug off the triage dashboard. It's unclear if this bug still happens, but should be easy to try and reproduce it. I can mentor anybody who's looking for a beginner-type bug to work on.