193103 - [Gtk2 xft] crash immediately before displaying document on linux

Status: RESOLVED DUPLICATE of bug 175108

(SeaMonkey :: General, defect)

Description

[©tìpán Li¹ka]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3b) Gecko/20030211
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3b) Gecko/20030211

I don't see anything from target page, all mozilla windows ends immediately.

it starts (I think) after installing some plugins and add-ons (spellchecker and
flash). It doesnt crash for mozilla 1.0.1 from RH8.0 original RPMs, but crashes
for 1.3a and 1.3b from mozilla.org's RPMs.

I work on RH8.0 in remote X-session.

Reproducible: Always

Steps to Reproduce:
1.to the URL edit write "www.pavouczech.net"
2.press [Enter]


Actual Results:  
mozilla closes all windows and writes no message, even if runs from xterm

Comment 1

[©tìpán Li¹ka]

Attachment: page on which mozilla crashes

This is a part of page www.pavouczech.net, which i "wget"ed and this local copy
makes my mozilla to crash also.

I've tried some links from "flash crash" bugs and mozilla doesn't crash.

Comment 2

[Matthias Versen [:Matti]]

which flash version do you installed ?
HAve you copied the Java plugon or correctly symlinked it ?

Comment 3

[Andrew Schultz]

worksforme with linux trunk build 20030212 at URL and attached page.  I'm not
seeing a flash object in either.  I have flash 6.0 r68

Comment 4

[Artur Polaczynski]

wfm 
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3b ; ..:: Ar't ::..) Gecko/20030212

mayby that remote X mode ?
try local sesion if posible !

Comment 5

[R.K.Aa.]

WFM, current CVS, Linux.
Are you using a default or experimental build?
(The experimental rmps being the Gtk2 and Xft enabled ones)

Comment 6

[©tìpán Li¹ka]

I've tried flash 5 (from netscape 7 xpi) and 6 (from mozilla.org/start/1.0 ..
plugins - MM's install)

I'm using mozilla-*-1.3b_-0_gtk2_xft.i386.rpm and I've installed it with "rpm
-Uvh *" on RH 8.0

It works ok on local session. Mozilla crashes in remote XSession on many more
pages (e.g. swnet.cz, www.ihned.cz). From some of them I can see drawing the
page, on some of them it crashes immediately.

Comment 7

[©tìpán Li¹ka]

Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3b) Gecko/20030211

Now I did installation of mozilla-*-1.3b-0_xft.i386.rpm without any additional
plugins. It crashes on about:plugins page (only remote X-session) also.
I can't find "vanilla" RPMs as were in 1.3a release. Should I try 1.3a vanilla
packages?

If I do "rpm -e plugger" the about:plugins page is ok, but www.pavouczech.net
crashes. Do you need my about:plugins page with plugger installed?

Another thing (I don't know if its connected with this bug) is, that when I do
"rpm -ivvvh " rpm frozens on calling ldconfig for package mozilla-nss-1.3b* and
I must "kill -KILL" it.

Comment 8

[Andrew Schultz]

> It works ok on local session. Mozilla crashes in remote XSession

Flash 5 will crash consistently on remote display, but flash 6 should be ok. 
but it sounds like you have bigger problems than flash.

> If I do "rpm -e plugger" the about:plugins page is ok, but www.pavouczech.net
> crashes. Do you need my about:plugins page with plugger installed?

what version of plugger do you have installed?

rpm having problems is probably rpm's bug.

you might uninstalling Mozilla, completely clobbering the mozilla directory
(/usr/lib/mozilla-xxx) (save any plugins), and then reinstalling with RPM.  RPM
sometimes leaves stuff behind.

other than that you would need to grab a stacktrace with gdb or try out one of
the installer (non-rpm) builds which have talkback and use that when it crashes.

Comment 9

[©tìpán Li¹ka]

>> If I do "rpm -e plugger" the about:plugins page is ok, but www.pavouczech.net
>> crashes. Do you need my about:plugins page with plugger installed?
>
>what version of plugger do you have installed?
plugger-4.0-17 from original RH 8.0

>
>rpm having problems is probably rpm's bug.
probably, but why only mozilla-nss and why in ldconfig?

>
>you might uninstalling Mozilla, completely clobbering the mozilla directory
>(/usr/lib/mozilla-xxx) (save any plugins), and then reinstalling with RPM.  RPM
>sometimes leaves stuff behind.
I did "find / | grep -i mozilla | less " and removed all mozilla stuff. I tested
it also from new account - so new profile too.

>
>other than that you would need to grab a stacktrace with gdb or try out one of
>the installer (non-rpm) builds which have talkback and use that when it >crashes.
It seems there's no core-file.

It work for all of you with remote X-session?


Trace with gdb:
After "about:plugins" page:
--------------------------------------------------------
Program received signal SIGPIPE, Broken pipe.
[Switching to Thread 8192 (LWP 24147)]
0x420cdb84 in write () from /lib/i686/libc.so.6
(gdb) cont
Continuing.
--------------------------------------------------------


It continues normally. And then page "www.pavouczech.net":
--------------------------------------------------------
[New Thread 180247 (LWP 24207)]
[New Thread 188440 (LWP 24208)]
[New Thread 196633 (LWP 24209)]
[New Thread 204826 (LWP 24210)]
Gdk-ERROR **: BadValue (integer parameter out of range for operation)
  serial 11643 error_code 2 request_code 53 minor_code 0
Gdk-ERROR **: BadDrawable (invalid Pixmap or Window parameter)
  serial 11644 error_code 9 request_code 55 minor_code 0

Program exited with code 01.
--------------------------------------------------------
no stack is there, because it end regularly. W/O gdb it says nothing.

Talkback release would be useless, i think, because it doesnt crash with signal.
Or am I wrong?

P.S. Weekend starts and I have no internet at home, so I can continue at Monday.

Comment 10

[Andrew Schultz]

Program received signal SIGPIPE, Broken pipe.
[Switching to Thread 8192 (LWP 24147)]
0x420cdb84 in write () from /lib/i686/libc.so.6
(gdb) cont

don't continue... grab the stack there (bt)

to grab the gtk "crash" (which talkback would not catch) you'll need to do this
sometime before the crash:

(gdb) b exit
Breakpoint 1 at 0x8063814
(gdb) b abort
Breakpoint 2 at 0x8063534

Comment 11

[R.K.Aa.]

Reassigning to Blizzard

Comment 12

[Christopher Blizzard (:blizzard)]

This worksforme remotely.  No problems.  Might be bug 173204.

Comment 13

[©tìpán Li¹ka]

I think it's not connected with gtk2, I'm now working with xft-only build, and
it behaves identically.

I think it's not connected with <a
href="http://bugzilla.mozilla.org/show_bug.cgi?id=173204"> bug 173204 </a> also,
because it worksforme.

>Program received signal SIGPIPE, Broken pipe.
>don't continue... grab the stack there (bt)

I can't reproduce SIGPIPE today. In gdb it shows page "about:plugins" without
plugger's entries (no crash, no message), without gdb it crashes. Strange.


>to grab the gtk "crash" (which talkback would not catch) you'll need to do this
>sometime before the crash:
>
>(gdb) b exit
>Breakpoint 1 at 0x8063814
>(gdb) b abort
>Breakpoint 2 at 0x8063534

I have "no debugging symbols" in gdb. 
Which is the easiest way for me to get not-stripped version? Can I install
browser.xpi and talkback.xpi from linux-xpi dir? Or should I make it from
sources (release or nightly)?


Another possibility: My X-server (on which I have remote X-session) is old
XFree86-3.3.5 on RH 6.1. Sometimes it says:
Xlib:  extension "RENDER" missing on display "fcc1-135.localnet:0.0".
Could it be the reason?

Comment 14

[©tìpán Li¹ka]

Attachment: gdb bt results

So I did the bt. Sorry for questions in previous comment. I can do "break exit"
only after run. 

The SIGPIPE signal seems to be masked, its only because of lack of soundcard
driver on my computer, I guess.

Whats wrong with these gdk errors I don't know. Look at the bt.txt attached,
please.

Comment 15

[Christopher Blizzard (:blizzard)]

Also might be bug #175108.  The fact that you're running a really old X server
is an important clue.

Comment 16

[©tìpán Li¹ka]

Yes, it seems to be the same bug as in bug #175108. It crashes on page
http://www.cs.umu.se/~stric/tmp/x12a0.html with similar trace:

Gdk-ERROR **: BadValue (integer parameter out of range for operation) 
  serial 32919 error_code 2 request_code 53 minor_code 0 
[Switching to Thread 8192 (LWP 9579)] 
 
Breakpoint 1, 0x4202b0f6 in exit () from /lib/i686/libc.so.6 
(gdb) bt 
#0  0x4202b0f6 in exit () from /lib/i686/libc.so.6 
#1  0x4026d455 in gdk_x_error () from /usr/lib/libgdk-1.2.so.0 
#2  0x4030f831 in _XError () from /usr/X11R6/lib/libX11.so.6 
#3  0x4030ddcf in _XReply () from /usr/X11R6/lib/libX11.so.6 
#4  0x402f6be5 in XGetImage () from /usr/X11R6/lib/libX11.so.6 
#5  0x414bc643 in XftGlyphFontSpecCore () from /usr/lib/libXft.so.2 
#6  0x414bed67 in XftDrawGlyphFontSpec () from /usr/lib/libXft.so.2 
#7  0x414bef1f in XftDrawCharFontSpec () from /usr/lib/libXft.so.2 
#8  0x4149a7d7 in nsFontMetricsXft::DrawString () from
/usr/lib/mozilla-1.3b/components/libgfx_gtk.so 
#9  0x41486271 in nsRenderingContextGTK::DrawString () from
/usr/lib/mozilla-1.3b/components/libgfx_gtk.so 

Comment 17

[Christopher Blizzard (:blizzard)]

Marking as a dup.

*** This bug has been marked as a duplicate of 175108 ***