1192300 - [e10s] Many crashes with varying stacks while loading specific page

Status: RESOLVED DUPLICATE of bug 1189277

(Core :: General, defect)

Description

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

This bug was filed from the Socorro interface and is 
report bp-e98474a5-5e7a-4cf9-a680-043cd2150807.
=============================================================

When loading the page at http://www.rollingstone.com/tv/features/david-letterman-famous-fans-say-goodbye-20150520?page=2 in the latest Nightly on Windows 8.1 I'm consistently getting a child process crash. The stacks are all over the place it looks like. Some reports (the top few are from the latest nightly, the next lot are from a slightly older nightly):

bp-507bdc9e-f5d8-47a2-b13f-cce222150807
bp-eacdba79-0d6e-44ab-a5cf-14d6c2150807
bp-e98474a5-5e7a-4cf9-a680-043cd2150807
bp-4670f68b-e7f1-4d4e-9a12-99ad02150807
bp-5af0c474-7ca2-4cef-a893-2ca842150807
bp-008b6060-07c9-4779-bfc4-c97c92150807

Comment 1

[Matthias Versen [:Matti]]

I'm pretty sure that this is the same crash as bug 1192300.
this is a recent regression according to that reporter

Comment 2

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

(In reply to Matthias Versen [:Matti] from comment #1)
> I'm pretty sure that this is the same crash as bug 1192300.

Wrong bug number?

Comment 3

[Matthias Versen [:Matti]]

sorry, i meant bug 1191661

Comment 4

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

I ran mozregression-gui and bisected this down to the nightly of 2014-11-07. regression window:

https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=2114ef80f6ae&tochange=b62ccf3228ba

I noticed during bisection that all the e10s builds had this crash and the non-e10s builds did not, so it's quite likely that this crash was introduced on nightly when e10s was turned on by default on nightly.

Comment 5

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

On the latest nightly with e10s disabled I can't reproduce this crash. With e10s enabled it reproduces 100%.

Comment 6

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

I re-ran mozregression but this time always opened an e10s window to do the test. This is the regression range I got where the crashes started (nightly of october 22, 2014).

https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=29fbfc1b31aa&tochange=ae4d9b4ff2ee

In nightlies prior to october 22, 2014 in my range, loading the URL in an e10s window would not crash, but it would cause firefox to lock up. (i.e. I would open an e10s window, load the URL, scroll around, and close the window - at this point the original non-e10s window would be "Not Responding" and i would have to get windows to terminate the process).

Comment 7

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

(In reply to Kartikaya Gupta (email:[e-mail]) from comment #6)
> In nightlies prior to october 22, 2014 in my range, loading the URL in an
> e10s window would not crash, but it would cause firefox to lock up. (i.e. I
> would open an e10s window, load the URL, scroll around, and close the window
> - at this point the original non-e10s window would be "Not Responding" and i
> would have to get windows to terminate the process).

... and this behaviour started in the 2014-08-18 nightly. regression range:

https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=94ba78a42305&tochange=0aaa2d3d15cc

Prior to this with OMTC force-enabled and loading the URL in an e10s window there were no issues.