66164 - Various crashes visiting a particular sequence of URLs

Status: RESOLVED FIXED

(SeaMonkey :: General, defect)

Description

[Scott Gifford (old account)]

By visiting a list of URLs culled from various bugs, I can consistently get
Mozilla to crash.  As I have tracked this bug down, I have consistently traced
the crash to a specific bug, found a patch for that  bug, applied it, and then
crashed in a different place.

I'm opening up this bug so that I don't have to keep moving this list of URLs
from bug to bug as various layers of the bug are stripped away.

Currently, I'm testing on a CVS build from 1/19/2001, with patches for several
bugs applied.

Here is my list (mostly from bug 63243):

http://osdn.com/
http://www.msnbc.com
http://search.perl.com/
http://www.zdnet.com/eweek
http://home.excite.com
http://www.kinigadner.com/intro.html
http://www.siemens.com
http://www.kiss.fi/
http://www.bestmat.be/en/occasbe.html
http://www.xdrive.com/target.html?path=/company/main_download.html
http://www.scee.com
http://www.real.com
http://www.vizzavi.co.uk
http://www.lge.com.au/
http://www.maximonline.com
http://www.m-w.com
http://www.tellium.com
http://cnnfn.cnn.com/2001/01/08/technology/wires/ibm_translate_wg
http://www.thecounter.com/stats/2000/
http://ntv.ru

Comment 1

[Scott Gifford (old account)]

Attachment: First stack trace

Comment 2

[Scott Gifford (old account)]

Attachment: Stack trace after applying patch from bug 61386

Comment 3

[Scott Gifford (old account)]

Attachment: Stack trace after applying patch from bug 65800

Comment 4

[Scott Gifford (old account)]

Attachment: Mozilla talkback crashes.

Comment 5

[Scott Gifford (old account)]

As another note, I'm currently testing by visiting each page, refusing to
install Java or Flash if I'm asked, hitting a link, hitting back, then closing
the window.

My build doesn't have Flash installed or Java.

Comment 6

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

I have to wonder what's strange about your system.  Mozilla is quite stable for
many people.  Any ideas why it crashes so much for you?  Is there something
strange about your machine?  How fast is your Internet connection?

Comment 7

[Scott Gifford (old account)]

Attachment: Another stack backtrace

Comment 8

[Scott Gifford (old account)]

I don't think that Mozilla is unstable for me in general...I regularly keep it
up for 2-3 full 8+ hour days of normal browsing use.  But this particular set of
URLs can consistently get it to crash for me.  That's why I opened up this bug,
so other people can try these URLs and hopefully reliably reproduce the crashes
I'm seeing.

Note that these patches have made Mozilla much more stable with these URLs; I
now have to go through the list 2-3 times, and click on some links inside each
of the basic URLs in order to get it to crash consistently.

My system is a fairly fast 750 MHz Athlon, on a Linux 2.2.16 kernel, on an
up-to-date RedHat 6.2 distribution, with some components upgraded to RedHat 7.0
to work with my video card.  I connect to the Internet via an Ethernet card,
connected to a Squid Web proxy, connected to a NAT router, connected to a 128K
dual-channel ISDN line, connected to my regular dial-up ISP.  I'm compiling with
"gcc version 2.96 20000731 (Red Hat Linux 7.0)", although about half of these
stack traces are from a nightly build, so the compiler certainly isn't the root
cause.

Comment 9

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

Did you see that stack with the patch from bug 65800?  It looks like you did.

If you see that again with the patch, could you go to the frame with
FrameManager::HandlePLEvent (e.g., with the command "frame 1"), and then try:

p frameManager
p *frameManager
p frameManager->mPresShell
p *(PresShell*)(frameManager->mPresShell)

Comment 10

[Scott Gifford (old account)]

Attachment: Another stack trace

Comment 11

[Scott Gifford (old account)]

Yes, that was with the patch.

Here's another one, not in exactly the same place...one end of the stack looks like:

	#0  0x6f0073 in ?? () at ../../../dist/include/nsIPageSequenceFrame.h:112
	#1  0x40127d30 in PL_HandleEvent (self=0x9823198) at plevent.c:576
	#2  0x40127b49 in PL_ProcessPendingEvents (self=0x80a51d8) at 		
		plevent.c:509
	[ ... ]

Since it wasn't in quite the same spot, I wasn't able to get the values of the
variables you asked for; but I made the assumption that self->owner was a
(*frameManager), cast it, and provided the values I got.  I have no idea if that
was a correct assumption; hopefully you'll be able to tell if the data is gibberish.

Comment 12

[buster]

I'll bet most, if not all, of these problems stem from bug 65243.  A fix for
that bug was checked in 1/22/01.  Please retest with the patch from that bug,
and let's see how many of these problems go away.

Comment 13

[Scott Gifford (old account)]

Attachment: Different crash stack trace

Comment 14

[Scott Gifford (old account)]

I just crashed on what looks like a different bug than the one from bug 65800 or
but 65243, by visiting these URLs, and some others from the bugs this bug
depends on.

I'll see if I can figure out what this crash is from, add the URLs I used to
this page, and continue seeing if I can reproduce 65800/65243.

Comment 15

[Scott Gifford (old account)]

Looks like that last crash was bug 61756.  Adding that as a dependency...

Comment 16

[Scott Gifford (old account)]

Here is a second set of URLs that triggered my most recent crash:

http://www.iex.net/rouyer/guru_test
http://nirvana.media3.net/
http://www.telepolis.com
http://www.msn.comwww.demonews.com ww4.janus.com www.netscape.com/webmail
http://www.iseekpasswords.com
http://www.vgmusic.com
http://www.yahoo.com
http://register.com
http://www.java.sun.com
http://www.nyt.com (tech section)
http://www.realplayer.com

Comment 17

[Scott Gifford (old account)]

Added dependancy (I thought I did that?...)

Comment 18

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

Get the patch from bug 65243 and un-apply the one from bug 65800.  That should
help the event crashes.

Comment 19

[Scott Gifford (old account)]

I removed your patches and updated my CVS tree.  The updated tree seems to
contain the patch from 65243.

I actually had updated my tree (and therefore was using the patch from 65243)
last night when I crashed, but I crashed in a completely different part of the
code.

If your patches are just good "defensive programming", though --- making sure
that we don't crash even if bug 65243 crops up again --- I'd rather leave them
in, in addition to the 65243 patch.

Comment 20

[Scott Gifford (old account)]

OK, I can no longer make Mozilla crash.  I've been through this list of URLs 4
times over 2 days, and I'm composing this comment in the same browser instance
(yay!).  I'll test again when all of the patches I used are integrated into the
tree.

I currently have a CVS build from 1/22/2001, with the patches from bug 61386 and
bug 61756.

I will add this info to other relevant bugs.

Comment 21

[Scott Gifford (old account)]

Wow, I've been up for over a week!  I've been using Mozilla for all of my daily
browsing (which is quite a bit) for a week and a day, with zero crashes.  Very
impressive; for the first time, I'm seeing better reliability than NS4.  It
looks like all of the patches I applied are now in CVS, so I'm going to try
removing everything custom, and running with a trunk build for awhile.

[ Also, removing dependency on 51384, since that seems to be another victim of
the same bugs, and not something that needs to be fixed on its own. ]

Comment 22

[Scott Gifford (old account)]

Up for two weeks now, and I just went through the URL list one last time with no
problems.  I'm going to shut down this copy of Mozilla, get a nightly with no
hand-sewn patches, and see how it does.  If it does as well as this build has,
I'll go ahead and close out this bug.

Comment 23

[Scott Gifford (old account)]

OK, I haven't had problems with this in a while, and I just tried it again with
the 2001032614 build, with libpr0n and the new cache, so I'm comfortable that
this issue is resolved.

Thanks, dbaron, for your help!