Closed Bug 749705 Opened 13 years ago Closed 13 years ago

Very frequent crashes when using JavaScript

Categories

(Core :: Widget: Gtk, defect)

12 Branch
x86
Linux
defect
Not set
normal

Tracking

()

RESOLVED DUPLICATE of bug 654877

People

(Reporter: a_geek, Unassigned)

Details

I can only say it for FF, but since scripts are mentioned in the description of "Core", I felt that I should report the problem here. After an awfully frequently crashing FF10, I upgraded to FF11 (Help -> About, then upgrade), and FF11 was stable again. Then I upgraded to FF12, and it keeps crashing _very_ frequently - eg. 3 times today so far. Here are the crash IDs from this week: bp-f5745fcb-884a-478d-b246-b48742120423 bp-1a8f2899-78e0-4389-b64b-9f57b2120424 bp-21d683ea-601c-40a6-834e-aa0422120424 bp-31a7b549-401f-4ed5-9bdc-4effd2120424 bp-a0526d03-5602-4259-81b2-a6d0a2120424 bp-d90654c9-3802-4170-9478-960ca2120425 bp-72972903-0575-4644-aaf0-f39d32120426 bp-a3fdd2d3-b56b-4605-bb69-3ce062120426 bp-a093aeb2-627b-4c42-a109-9f52f2120427 bp-c84cfdca-8e0b-4fdd-9907-ae53f2120427 bp-9f9ee8c5-6714-4c31-9d5d-363672120427 I have not discovered any other irregularity on this machine. These days, I usually have some 40-70 open in up to 4 windows, plus the usual assortment of NoScript, RequestPolicy and friends that make surfing half-way safe. The crashes typically happen when I activate JavaScript for a tab - but usually, only for a part of the desired scripts (eg. not for google-analytics). On activating scripts, or enabling another target in RequestPolicy, the page will reload, and this is where most of the crashes happen. But some simply happen out of the blue, like while I'm reading a web page, while probably some script is working behind my back and brings the browser to its knees. I have no resource problems on this computer.
A large number of those crashes are X11Error being called from libbonoboui-2.so. Most of them are that, in fact. The crash reports include this note: X_ShmPutImage: BadMatch (invalid parameter attributes); 763 requests agoxpcom_runtime_abort(###!!! ABORT: X_ShmPutImage: BadMatch (invalid parameter attributes); 763 requests ago: file /builds/slave/rel-m-rel-lnx-bld/build/toolkit/xre/nsX11ErrorHandler.cpp, line 190)
Component: General → Widget: Gtk
QA Contact: general → gtk
You should set dom.ipc.plugins.* preferences to their default values.
Status: UNCONFIRMED → RESOLVED
Closed: 13 years ago
Resolution: --- → DUPLICATE
I'm largely unfamiliar with this, but found at least two crashes mentioning JSProtoObjectFinalized. The machine is basically a stock Debian Squeeze (i386). The crashes are all SIGSEGV which usually hints at bad memory, but the machine has ECC memory, so I'm not yet convinced that it is this - and so far, FF is the only app that really keeps crashing. I'm running the openvz variant of the kernel. Maybe the problem is related to this stuff? Do you want me to test something specific?
@Scoobidiver: I've just checked and see this: dom.ipc.plugins.enabled was set to 'false', everything else was at their default values. Can't even remember having touched these, as I have no idea about what they do. Now I've switched this to 'true' (the default value) and see how it goes.
Oh, and fwiw, the problem also occurs if I don't have anything with flash running.
If you still get crashes after that pref flip, please either reopen this bug with the new crash IDs or file a new one? There _were_ a few IDs above that were not obviously the X11Error thing, but they didn't seem to have much info. :(
I don't know how to properly re-open a bug, but here you have a new crash id from today: bp-81c063b1-2bbb-401c-b061-7f2be2120428
Status: RESOLVED → UNCONFIRMED
Resolution: DUPLICATE → ---
(In reply to a_geek from comment #7) > I don't know how to properly re-open a bug, but here you have a new crash id > from today: > bp-81c063b1-2bbb-401c-b061-7f2be2120428 This one is a dupe of bug 654877. If you have reliable steps to reproduce, provide them.
Status: UNCONFIRMED → RESOLVED
Closed: 13 years ago13 years ago
Resolution: --- → DUPLICATE
For what it's worth, you reopened the bug exactly the right way.
You need to log in before you can comment on or make changes to this bug.