Firefox 52.0.2 crashes

kikinovak · 04-19-2017, 03:01 PM

Quote:

Originally Posted by SlowLearner

Thank you Niki is very sound advice..
.
Can you detect any kinda pattern or similarities in the type of errors causing the crashes?

Yes. My phone ringing and users complaining about Firefox crashing when they try to access $WEBSITE.

Didier Spaier · 04-19-2017, 10:45 PM

Quote:

Originally Posted by kikinovak

I'll see how I can get my hands on the previous 45.8.0 version (which unfortunately disappeared from all the Slackware mirrors) and install that while the Firefox devs get their act together.

George uploaded packages for 45.9.0 yesterday. As usual, this is a repackaging of official binaries with lot of locales bundled:

You know where to find them, but for other readers:
http://slackware.uk/salix/x86_64/14.1/
http://slackware.uk/salix/i486/14.1/

This reminds me that I should upgrade TB in a similar fashion in the Slint repos.

orbea · 04-19-2017, 11:52 PM

Quote:

Originally Posted by kikinovak

Sorry, but I don't have time to potty-train the new Firefox version. I'll see how I can get my hands on the previous 45.8.0 version (which unfortunately disappeared from all the Slackware mirrors) and install that while the Firefox devs get their act together.

Cheers,

Niki

I just want to make it clear I was replying to this.

Quote:

Originally Posted by SlowLearner

It would be FUN to find where exactly this is occurring or even just to eliminate where it is NOT occurring. But my debugging knowledge is pretty slim.

Any advice for the debugging, SlowLearner?

SlowLearner · 04-20-2017, 11:14 AM

Quote:

build firefox with debugging symbols, run it in gdb, reproduce the crash, get a backtrace and finally share it with the firefox devs.

Thanks @orbea.. I think i can do this! I know next to nothing about gdb but i'm sure i can figure it out with time. Seems like a good time waiting to happen.

@kikinovak

Quote:

Sorry, but I don't have time to potty-train the new Firefox version.

lol is ok- no need for apologies :-)
I enjoy making the time to play with inner workings and LEARN (however slowly). I'm just addicted to the learning process itself. This addiction/obsession with learning new skills and discovering how things work together is a big reason WHY i originally chose and still prefer using the slackware platform.

But yes- i do always wish i had much MORE time to play/learn or... you know, perhaps potty train troublesome glitches.

the old standard "..that's not a bug, that's a FEATURE!" lol
or my personal favorite:

There are no problems, there are only opportunities to succeed.

@Didier Spaier Thank You - Thank You !

Quote:

George uploaded packages for 45.9.0 yesterday. As usual, this is a repackaging of official binaries with lot of locales bundled:

http://slackware.uk/salix/x86_64/14.1/
http://slackware.uk/salix/i486/14.1/

I am just soooo thankful that this board exists as it makes it so much easier for us all to succeed in our own way.

orbea · 04-20-2017, 11:31 AM

Quote:

Originally Posted by SlowLearner

Thanks @orbea.. I think i can do this! I know next to nothing about gdb but i'm sure i can figure it out with time. Seems like a good time waiting to happen.

To be honest, it doesn't sound very fun to do it with firefox, but if you really want to debug it that is what needs to be done. Building firefox with debugging symbols (Be sure to not strip them as many slackbuilds do by default) will probably require much more space to build than it already does... GDB will be the easy part.

Code:

gdb --args firefox 2>&1 | tee firefox.log
(in the gdb prompt)
run
(reproduce the issue now and when it crashes...)
bt
bt full
t a a f

And then you will have the entire log complete with the backtrace, a much more verbose output of the backtrace and a list all of the running threads (Which sometimes may be useful) in your new firefox.log file. You can then give this log file to a firefox developer who is interested in fixing the issue.

SlowLearner · 04-20-2017, 12:13 PM

@orbea
Ya SEE what i mean !
I adore this community
THANKS AGAIN for the further much more detailed guidance.

Incidentally, the rob.rice on another LQ thread greatly narrowed down the cause of the abrupt seg fault mozilla firefox crashes occuring after the latest updated of:

Quote:

Slackware 14.1 32bit
mozilla-firefox-52.0.2esr-i486-2_slack14.1.txz

Basically, not surprisingly it's a javascript issue.

http://www.linuxquestions.org/questions/showthread.php?p=5699802#post5699802

is the post where I explain how to disable javascript from about:config in firefox.

I'm assuming this'll make the firefox gdb process easier...

Here is to knowing where to start!

orbea · 04-20-2017, 12:24 PM

I would suggest noscript or an equivalent addon which will disable javascript by default effectively masking this issue. At least until you enable the wrong bit of javascript on a poorly made webpage...

SlowLearner · 04-22-2017, 11:07 PM

That didn't take long!

"...switching back to the 45.x ESR branch due to instabilty of the
52.x ESR branch on Slackware 14.1."

http://www.slackware.com/security/vi...ecurity.320508

Here are the details from the Slackware 14.1 ChangeLog:
+--------------------------+
patches/packages/mozilla-firefox-45.9.0esr-i486-1_slack14.1.txz: Upgraded.
This release contains security fixes and improvements.
Also, switching back to the 45.x ESR branch due to instabilty of the
52.x ESR branch on Slackware 14.1.
For more information, see:
https://www.mozilla.org/security/kno...irefoxESR.html
(* Security fix *)
+--------------------------+

AlleyTrotter · 04-23-2017, 07:51 AM

Patrick is one amazing dude.

onebuck · 04-23-2017, 08:43 AM

Hi,

I was having issues with; mozilla-firefox-52.0.2-x86_64-1ro as being very slow and locking. Not a ESR package but generated by latest-firefox.sh script by Ruari Oedegaard. I did a upgradepkg using ./mozilla-firefox-53.0-x86_64-1ro.tgz. Still no fun in the sun! I did some trouble-shooting and found the add-on 'IP and Domain Information' to be at fault. I simply disabled that extension and restart to find that was indeed the problem. Further checking found that add-on was replacing my DNS with a totally different address that were my preferred nameservers in '/etc/resolv.conf'. Unsure as to the locking but since disabling that add-on I have no locks using Firefox.

So to my fellow Slackers, be aware of what you are using and what are the potential issues. Check thoroughly! Even signed extensions or whatever can cause issues beyond the installed package.

Check out this post for some useful information for ESR and future Firefox releases; http://www.linuxquestions.org/questi...ml#post5699909

Hope this helps.
Have fun & enjoy!

haary · 04-23-2017, 10:37 AM

Firefox 52.1.0 ESR is out. Tried to build it with my own and the standard SlackBuild which led to an error. It is a problem with FileSystemDirectoryReader.h and I think that the Mozilla developers forgot something, but I am not a C/C++ programmer. However, with this patch applied FF 52.1.0 builds fine:

Code:

diff -urN firefox-52.1.0esr-orig/dom/filesystem/compat/FileSystemDirectoryReader.h firefox-52.1.0esr/dom/filesystem/compat/FileSystemDirectoryReader.h
--- firefox-52.1.0esr-orig/dom/filesystem/compat/FileSystemDirectoryReader.h    2017-04-11 04:13:18.000000000 +0200
+++ firefox-52.1.0esr/dom/filesystem/compat/FileSystemDirectoryReader.h 2017-04-20 11:47:29.514979404 +0200
@@ -10,6 +10,7 @@
 #include "mozilla/Attributes.h"
 #include "mozilla/ErrorResult.h"
 #include "mozilla/dom/BindingDeclarations.h"
+#include "mozilla/dom/FileSystemDirectoryEntry.h"
 #include "nsCycleCollectionParticipant.h"
 #include "nsWrapperCache.h"

Didier Spaier · 04-23-2017, 10:46 AM

Well, for Slackware stable at least maybe we could wait the end of the qualification period i.e. FF 52.2.0 ESR to switch to the 52 branch, as only then we can expect the same level of stability as for the 49 branch. Only my humble opinion, of course.