Closed Bug 1207766 Opened 9 years ago Closed 9 years ago

Firefox 41 hangs (not responding) on tweakers.net/nieuws

Categories

(Core Graveyard :: Plug-ins, defect)

41 Branch
defect
Not set
critical

Tracking

(firefox41- fixed)

RESOLVED FIXED
mozilla41
Tracking Status
firefox41 - fixed

People

(Reporter: dusty-2011, Assigned: bobowen)

References

()

Details

(Keywords: hang, reproducible)

User Agent: Mozilla/5.0 (Windows NT 6.3; WOW64; rv:41.0) Gecko/20100101 Firefox/41.0
Build ID: 20150917150946

Steps to reproduce:

This bug is easy to reproduce:

1) Make a new Firefox profile (also reproducible without step 1)
2) visit tweakers.net/nieuws (no crash)
3) click on tweakers logo to visit tweakers.net (no crash)
4) Click on 'Nieuws' to visit tweakers.net/nieuws (possibly a crash)
5) If it hasn't crashed yet, repeat steps 3 and 4 until it crashes.

Every next visit of tweakers.net/nieuws leads to a crash. This is only solvable by deleting the entire browsing history including cookies etc. Then you can visit the page once or twice again without a crash, after that it will keep crashing again.


Actual results:

See above.


Expected results:

The page shouldn't crash. In previous Firefox versions it doesn't crash. In other browsers it doesn't crash.

I've also reported the bug to tweakers: http://gathering.tweakers.net/forum/list_messages/1651352
FF41 doesn't crash on my side, but hangs (not responding). In about:crashes, do you have some crash reports to provide? (bp-...)

In addition, did you test with Nightly to see if it was reproducible too? (sometimes, it's already fixed in dev versions)
Flags: needinfo?(dusty-2011)
(In reply to Loic from comment #1)
> FF41 doesn't crash on my side, but hangs (not responding). In about:crashes,
> do you have some crash reports to provide? (bp-...)
> 
> In addition, did you test with Nightly to see if it was reproducible too?
> (sometimes, it's already fixed in dev versions)

My bad, Firefox hangs here too. The program doesn't respond anymore, and I have to end it with Task Manager. There is indeed no crash, and there is nothing in about:crashes. It hangs.

I haven't tested with Nightly. But this bug deserves a hotfix for Firefox 41 in my opinion.

For the moment, I have downgraded to Firefox 40.0.3. to avoid the hang.

Here is a permalink for the page which causes Firefox to hang: http://tweakers.net/nieuws/list/20150923
Flags: needinfo?(dusty-2011)
Summary: Firefox 41 crashes on tweakers.net/nieuws → Firefox 41 hangs (not responding) on tweakers.net/nieuws
Yes, refreshing the link http://tweakers.net/nieuws/list/20150923 make FF hang sometimes, but it's intermittent.
Severity: normal → critical
Keywords: hang
See Also: → 1197077
From previous release of Fx 41(beta or aurora) was getting the browser hanging, after a refresh and disabled  oop for flash in mms.cfg and the plugin setting was resolved.

With the release of Fx 41(release channel) the problem got back, I have restricted to the process plugincontainer, kill it and the browser got responsive again.

The hang of plugincontainer is intermitent and don't get any report(doesn't crash).

In related websites say is related to dom.ipc.plugins.asyncInit setted to true, but in my case is false already. Was the flash plugin whitelisted to async comunication?

Perharps i need another option to complete disable the async options for the addons.

The workaround was revert back to Fx 40.0.3
(In reply to dusty-2011 from comment #2)
> (In reply to Loic from comment #1)
> > FF41 doesn't crash on my side, but hangs (not responding). In about:crashes,
> > do you have some crash reports to provide? (bp-...)
> > 
> > In addition, did you test with Nightly to see if it was reproducible too?
> > (sometimes, it's already fixed in dev versions)
> 
> My bad, Firefox hangs here too. The program doesn't respond anymore, and I
> have to end it with Task Manager. There is indeed no crash, and there is
> nothing in about:crashes. It hangs.
> 
> I haven't tested with Nightly. But this bug deserves a hotfix for Firefox 41
> in my opinion.
> 
> For the moment, I have downgraded to Firefox 40.0.3. to avoid the hang.
> 
> Here is a permalink for the page which causes Firefox to hang:
> http://tweakers.net/nieuws/list/20150923

When crash/hang is the proccess plugincontainer the culprit?
Plugin container seems to use 25+% cpu on my 4-core system, so I'd say yes. Killing plugin container or a (the?) flash process directly allows the load to continue.
I forgot to mention; that link is a list of our news items of a specific day. Those occasionally include a embedded video (with a flash-based player via a iframe), there are 3 of those news items on that day. When opening any of the news items seperately, they load just fine, like:

http://tweakers.net/nieuws/105436/pebble-presenteert-ronde-smartwatch.html
http://tweakers.net/nieuws/105429/capcom-maakt-systeemeisen-street-fighter-v-bekend.html
http://tweakers.net/nieuws/105423/io-interactive-stelt-release-hitman-uit-tot-maart-2016.html

Apart from those video players, there may be flash-based advertisements. But the advertisements on the 'day list' should be similar to those on the specific news item pages, so I wouldn't directly suspect those.

My guess is the three video players in Firefox 41 (40 was ok, so are Chrome and Edge) might be the causing this issue.
Firefox hangs with fresh profile
stack using crashfirefox.exe: bp-2f28956f-4df9-47d2-a1cf-a12d32150927

Reproducible: almost 100% on Firefox41 release build.
              However it seems that 42.0b1, Aurora43.0a2 and Nightly44.0a1 are not affected.

Steps To Reproduce:
1. Open http://tweakers.net/nieuws/list/20150923 in 1st tab
2. Quit browser and restart
3. Restore previous session
   --- observe the hang
5. Repeat From step1 if necessary

Regression range
with force dom.ipc.plugins.asyncInit.enabled/dom.ipc.plugins.asyncInit = false(Firefox41 release default value):

https://hg.mozilla.org/releases/mozilla-beta/pushloghtml?fromchange=0aa0af046d24&tochange=6fbcbf05900f

Suspect: Bug 1182411
I could not find the range in m-central and  m-aurora. (i.e., m-central and m-aurora do not hang even if the patch Bug 1182411 was included)


In my case, dom.ipc.plugins.asyncInit.enabled=true seems to fix the hang on Firefox41beta.
Status: UNCONFIRMED → NEW
Ever confirmed: true
Keywords: reproducible
> 
> In my case, dom.ipc.plugins.asyncInit.enabled=true seems to fix the hang on
> Firefox41beta.

Perhaps the item was incorrect named enabled instead of disabled
(In reply to besrml from comment #9)
> > 
> > In my case, dom.ipc.plugins.asyncInit.enabled=true seems to fix the hang on
> > Firefox41beta.
> 
> Perhaps the item was incorrect named enabled instead of disabled

You're misunderstanding something.

The name of comment#8 is correct.
Blocks: 1182411
It looks like this is my change to the initialisation of windowless plugins that is causing the problem.

When the parent is being set in the chrome process, it is causing a sync native Windows message to the child, which is sometimes causing flash to make a sync call back to the parent causing a hang.
NPN_Evaluate is the call from the child that I've seen.

I'm building a debug version of Release to see if I can work out more details.
[Tracking Requested - why for this release]:
flash plugin hang
Assignee: nobody → bobowen.code
Status: NEW → ASSIGNED
Component: Untriaged → Plug-ins
Product: Firefox → Core
Target Milestone: --- → mozilla41
Discussed with Bob, we think that bug 1185639 should be uplifted to 41.
Are you going to do a chemspill for this? I can barely use Firefox 41. I create a lot of new profiles and I can't go around setting that pref every time.
(In reply to Mike Kaply [:mkaply] from comment #15)
> Are you going to do a chemspill for this? I can barely use Firefox 41. I
> create a lot of new profiles and I can't go around setting that pref every
> time.

The uplift mentioned in comment 14 fixes this, we'll have to wait to see if the request is accepted.
I've had several strange "FF not responding" situations since FF 40, and I was hoping it was fixed in the newly installed FF 41.  I saw somewhere (don't recall where) during the recent buzz over Flash vulnerabilities that FF may (or does) disable Flash, and that this may cause problems certain Flash-dependent pages.  Is this perhaps the same situation here?

The problematic sites for me are generally news & info sites (known lots of Flash there), but also various garden-variety sites that don't obviously/prominently use Flash but of course might do so.  But there is no rhyme or reason to the hangs -- just sometimes they happen, other times not.  And re-visit doesn't seem to make a difference -- sometimes a hang, sometimes not.  For example, a click-bait sequence of pages invariably results in a hang, where closing and restarting FF resumes and continues for a few more pages before hanging again.  I don't recall that clearing history made any difference.

Occasionally, if I wait long enough, FF might eventually respond, but usually, I end up with the standard Windows "close program" / "wait for respond" popup, and close the program.  (Then usually cancel the Windows search for solution because nothing comes up anyway.)
Further to my Comment 17:
I recently encountered FF non-responses on two PC machines (tower and laptop, both running Windows 7), including today.  It so happens that a routine update of my Norton Security program needed to apply a patch on each machine and waited for certain programs to close on each.  Specifically, Firefox and something called "ncolow", neither of which were obviously open.

On the laptop, Task Manager showed two FF stubs (typically about 2700 kb), with a single instance of "ncolow".  On the tower, seven or eight FF stubs and one "ncolow".  In both cases, I manually END-TASK'd all FF stubs, and in both cases, "ncolow" quit itself when the last FF was ended.  [The Norton Update then proceeded automatically and applied the patches successfully.]

I understand from problems with FF37 through FF39 that the FF stubs should NOT exist after FF is closed.  Makes sense to presume they were not removed because the closures were forced (ie- symptom), but could be an indication that something else was wrong (ie- cause).  As for "ncolow", I don't recall seeing it before, and not sure what this is -- Google is quite cryptic about it.
hi, could people in the bug who were able to reproduce the issue somewhat reliable test if it is fixed with the build from https://archive.mozilla.org/pub/firefox/candidates/41.0.1-candidates/build2/ and give feedback if this works?
I can verify that 41.0.1-candidates/build2 fixes the hang.

With STR of comment%8,
I can reproduce the hang on Firefox40.0
https://hg.mozilla.org/releases/mozilla-release/rev/78c82e5cd777
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:41.0) Gecko/20100101 Firefox/41.0 ID:20150917150946.

And, I can not reproduce the hang on Firefox40.0.1
https://hg.mozilla.org/releases/mozilla-release/rev/76aad04b2a72
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:41.0) Gecko/20100101 Firefox/41.0 ID:20150929144111.
Fix from bug 1185639 was uplifted to 41. Marking this bug as resolved to reflect that.
Status: ASSIGNED → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
In this version https://archive.mozilla.org/pub/firefox/candidates/41.0.1-candidates/build2/win32/en-US/ the bug seems fixed. So lets just wait for it to become released. Or is it already fixed in the 41 version? Earlier today I have had the same issues with the 41 version and 19x version of Flashplayer on more than 10 machines in my company
41.0.1 build2 is in the process of being released today - at least to the people that already have 41.0, which shows this issue with Flash 19.
I have upgraded to the latest stable Firefox version (41.0.1) and the bug is fixed!! Thanks a lot for this very swift fix! Awesome.
(In reply to dusty-2011 from comment #25)
> I have upgraded to the latest stable Firefox version (41.0.1) and the bug is
> fixed!! Thanks a lot for this very swift fix! Awesome.

Thanks for confirming that the fix worked! \o/
Aparently the fix is working in Fx 41.0.1
(In reply to besrml from comment #28)
> Aparently the fix is working in Fx 41.0.1

Not for me >.<
(In reply to gumballisthegreatest from comment #29)
> (In reply to besrml from comment #28)
> > Aparently the fix is working in Fx 41.0.1
> 
> Not for me >.<

Hi, are you still getting a hang with flash?

If you are, what sites are you seeing this with?

Thanks
Flags: needinfo?(gumballisthegreatest)
Bob Owen, yes, and I get the hang on here:
http://totaljerkface.com/happy_wheels.tjf
I really hope this issue is fixed in Firefox 41.0.2.
Flags: needinfo?(gumballisthegreatest)
(In reply to gumballisthegreatest from comment #31)
> Bob Owen, yes, and I get the hang on here:
> http://totaljerkface.com/happy_wheels.tjf
> I really hope this issue is fixed in Firefox 41.0.2.

That's not a Firefox hang. It's a different bug. The page loads fine in Firefox, but when you start playing the game Adobe Flash hangs. And then Firefox displays a message allowing you to close down Flash. And after closing Flash Firefox works again.
(In reply to gumballisthegreatest from comment #31)
> Bob Owen, yes, and I get the hang on here:
> http://totaljerkface.com/happy_wheels.tjf
> I really hope this issue is fixed in Firefox 41.0.2.

This is a different hang, I've filed a separate bug 1210799 and copied you in.

If you have any reported crashes that you know are related to this hang (in about:crashes), it would be great if you would post them in that bug, thanks.
Bob, I confirm Gumball's basic page loads, runs, and exits OK in FF 41.0.1.  (I did not play the game, nor did I allow FF to forward to another page.)  Running Windows 7/32 on a PC.

Also, over the past few days I've gone a few times to one of my sites which hung regularly during the problem -- now works fine in 41.0.1.
For example:  
http://www.motherearthliving.com/cooking-methods/savor-the-flavors-of-fall.aspx?newsletter=1&utm_source=Sailthru&utm_medium=email&utm_campaign=10.01.15%20MEL&utm_term=MEL%20eNews
...and similar article pages.
(In reply to Dan Pernokis from comment #34)

> Also, over the past few days I've gone a few times to one of my sites which
> hung regularly during the problem -- now works fine in 41.0.1.

Thanks for testing and confirming this.
Product: Core → Core Graveyard
You need to log in before you can comment on or make changes to this bug.