161967 - Crash on printing

Status: VERIFIED DUPLICATE of bug 159358

(Core :: Printing: Output, defect)

Description

[Brian Foddy]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; SunOS sun4u; en-US; rv:1.1b) Gecko/20020806
BuildID:    2002080610

Mozilla 1.1beta.
While trying to test bugs for landscape printing and
140122 I encountered an often repeatable crash while
trying to print the enclosed page.  

This version was compiled myself on Solaris 2.8 using gcc295
and following the distribution build instructions.  No Solaris
1.1b version was available precompiled.  The seems to work noramlly
otherwise.

After the crash, on restart of mozilla, the following error message
is encountered:

LoadPlugin: failed to initialize shared library libXm.so [ld.so.1:
/apps/soc/mozilla/mozilla-bin: fatal: relocation error: file
/usr/local/lib/libXm.so: symbol __eprintf: referenced symbol not found]

Mozilla will then not be able to read my profile.  It complains
the directory associated with my profile cannot be found (even
tho the dir is there and it shows the profile in the dialog).  I
can create a new profile, or reload my .mozilla directory from a 
backup.

Selected printer is HP 2100 postscript printer in landscape mode.
Print command is "lp".

Reproducible: Sometimes
Steps to Reproduce:
1.  I will attach HTML page with example.
2.
3.

Comment 1

[Brian Foddy]

Attachment: Example HTML that crashes while printing.

Comment 2

[Roland Mainz]

Uhm, this part is confusing:
> LoadPlugin: failed to initialize shared library libXm.so [ld.so.1:
> /apps/soc/mozilla/mozilla-bin: fatal: relocation error: file
> /usr/local/lib/libXm.so: symbol __eprintf: referenced symbol not found]

It sounds you have a custom Motif library in /usr/local/lib which was compiled
with an older version of gcc. I guess that shared library looks for libstdc++.so

Weired.

Reporter:
- Can you try to start Mozilla with en empty LD_LIBRARY_PATH env var ?
- Can you try to rename the shared library ?
- Can you try to rebuild the shared library with Sun Workshop that it stops
depending on gcc-specific libs ?

Comment 3

[Sam J. Fleet]

Confirmed using BuildID# 2002080904 using Windows 2000.

Need to change the O/S & Platform setting to all.  (To include Sun and Windows)

Discover one more thing that also cause the crash.  When I use the "Print
Preview" to preview it, it crashed the Mozilla browser.  So, the printer is not
the issue here.  

Comment 4

[Sam J. Fleet]

Additional Notice: The Talkback agent was brought up and it was submitted.

Will try to create a simplier html file to see if the cause can be narrowed
down.  Will post it if it work or won't post it if it won't work.

Comment 5

[Brian Foddy]

Thank you for looking at this so fast...

You were correct, I had an old (version unknown) copy of Lesstif
in /usr/local/lib.  When I removed those libraries from /usr/local/lib,
this part of the error (the LoadPlugin) went away.  Seems they were
only getting pulled in after a crash, becuase from a working 1.1b
a fuser on these files showed nothing using them.  So one part of
the mystery solved.  Thanks.

However, the printing crash was still there, as was the .mozilla
profile corruption.

Comment 6

[Roland Mainz]

Yes, I am getting a crash with my 2002-08-04-08-trunk build on Solaris/SPARC,
too.

Stack trace looks like this:
-- snip --
t@1 (l@1) signal SEGV (no mapping at the fault address) in nsVoidArray::Count
dbx: warning: can't find file
"/shared/bigtmp2/mozilla/2002-08-04-08-trunk/objdir_ws7_gtk/layout/build/nsTableFrame.o"
dbx: warning: see `help finding-files'
at 0xfb0e4e64
0xfb0e4e64: Count+0x002c:       ld      [%l0 + 0x4], %l0
dbx: warning: can't find file
"/shared/bigtmp2/mozilla/2002-08-04-08-trunk/objdir_ws7_gtk/layout/build/nsContainerFrame.o"
dbx: warning: can't find file
"/shared/bigtmp2/mozilla/2002-08-04-08-trunk/objdir_ws7_gtk/layout/build/nsTableOuterFrame.o"
dbx: warning: can't find file
"/shared/bigtmp2/mozilla/2002-08-04-08-trunk/objdir_ws7_gtk/layout/build/nsBlockFrame.o"
dbx: warning: can't find file
"/shared/bigtmp2/mozilla/2002-08-04-08-trunk/objdir_ws7_gtk/layout/build/nsPageContentFrame.o"
dbx: warning: can't find file
"/shared/bigtmp2/mozilla/2002-08-04-08-trunk/objdir_ws7_gtk/layout/build/nsPresShell.o"
Current function is nsView::Paint
  278           observer->Paint((nsIView *)this, rc, rect);
(dbx) where
current thread: t@1
  [1] nsVoidArray::Count(0x0, 0x7, 0xffbeceec, 0x67b338, 0x1, 0xa74a50), at
0xfb0e4e64
  [2] nsVoidArray::ElementAt(0x0, 0x0, 0xffbecf60, 0x0, 0x0, 0x0), at 0xfb0e4d60
  [3] BCMapBorderIterator::SetNewData(0xffbed4d8, 0x0, 0x0, 0xbc9800,
0xffbed59c, 0x2400), at 0xfb0dab08
  [4] BCMapBorderIterator::First(0xffbed4d8, 0xbc973c, 0xbc97c4, 0xbc9800,
0xffbed59c, 0xbcf0d4), at 0xfb0db3c4
  [5] nsTableFrame::PaintBCBorders(0xbc973c, 0x67a2c0, 0xbd5868, 0xffbed754,
0x0, 0x0), at 0xfb0dc97c
  [6] nsTableFrame::Paint(0xbc973c, 0x67a2c0, 0xbd5868, 0xffbed754, 0x0, 0x0),
at 0xfb0c5a08
  [7] nsContainerFrame::PaintChild(0xbc96f0, 0x67a2c0, 0xbd5868, 0xffbed86c,
0xbc973c, 0x0), at 0xfadbd684
  [8] nsTableOuterFrame::Paint(0xbc96f0, 0x67a2c0, 0xbd5868, 0xffbed86c, 0x0,
0x0), at 0xfb0ea008
  [9] nsContainerFrame::PaintChild(0xbca508, 0x67a2c0, 0xbd5868, 0xffbedc6c,
0xbc96f0, 0x0), at 0xfadbd684
  [10] nsBlockFrame::PaintChildren(0xbca508, 0x67a2c0, 0xbd5868, 0xffbedc6c,
0x0, 0x0), at 0xfad94118
  [11] nsBlockFrame::Paint(0xba5d08, 0xbcbe80, 0xbd5868, 0xffbedc6c, 0x0, 0x0),
at 0xfad93b60
  [12] nsContainerFrame::PaintChild(0xbca584, 0x67a2c0, 0xbd5868, 0xffbee06c,
0xbca508, 0x0), at 0xfadbd684
  [13] nsBlockFrame::PaintChildren(0xbca584, 0x67a2c0, 0xbd5868, 0xffbee06c,
0x0, 0x0), at 0xfad94118
  [14] nsBlockFrame::Paint(0xba5d08, 0xbcbe80, 0xbd5868, 0xffbee06c, 0x0, 0x0),
at 0xfad93b60
  [15] nsContainerFrame::PaintChild(0xbca6b4, 0x67a2c0, 0xbd5868, 0xffbee46c,
0xbca584, 0x0), at 0xfadbd684
  [16] nsBlockFrame::PaintChildren(0xbca6b4, 0x67a2c0, 0xbd5868, 0xffbee46c,
0x0, 0x0), at 0xfad94118
  [17] nsBlockFrame::Paint(0xba5d08, 0xbcbe80, 0xbd5868, 0xffbee46c, 0x0, 0x0),
at 0xfad93b60
  [18] nsContainerFrame::PaintChild(0xbca66c, 0x67a2c0, 0xbd5868, 0xffbee7c0,
0xbca6b4, 0x0), at 0xfadbd684
  [19] nsContainerFrame::PaintChildren(0xfadbd3b0, 0x67a2c0, 0xbd5868,
0xffbee7c0, 0x0, 0x0), at 0xfadbd374
  [20] nsContainerFrame::Paint(0xbca66c, 0x67a2c0, 0xbd5868, 0xffbee7c0, 0x0,
0x0), at 0xfadbd2a0
  [21] nsPageContentFrame::Paint(0xbca66c, 0x67a2c0, 0xbd5868, 0xffbee7c0, 0x0,
0x0), at 0xfaeb5a90
  [22] PresShell::Paint(0x67a7d0, 0xbcad30, 0xbd5868, 0xffbee7c0, 0xfa5796e0,
0x3), at 0xfaed2d6c
=>[23] nsView::Paint(this = 0xbcad30, rc = CLASS, rect = STRUCT, aPaintFlags =
0, aResult = 0), line 278 in "nsView.cpp"
  [24] nsViewManager::RenderDisplayListElement(this = 0xa6dd60, element =
0x6d6ef8, aRC = CLASS), line 1207 in "nsViewManager.cpp"
  [25] nsViewManager::RenderViews(this = 0xa6dd60, aRootView = 0xbcacc0, aRC =
CLASS, aRect = STRUCT, aResult = 0), line 1155 in "nsViewManager.cpp"
  [26] nsViewManager::Display(this = 0xa6dd60, aView = 0xbcacc0, aX = 0, aY = 0,
aClipRect = STRUCT), line 3094 in "nsViewManager.cpp"
dbx: warning: can't find file
"/shared/bigtmp2/mozilla/2002-08-04-08-trunk/objdir_ws7_gtk/layout/build/nsSimplePageSequence.o"
  [27] nsSimplePageSequenceFrame::PrintNextPage(0xbb56d0, 0x67a2c0, 0x4, 0x0,
0x0, 0x0), at 0xfaf11b28
dbx: warning: can't find file
"/shared/bigtmp2/mozilla/2002-08-04-08-trunk/objdir_ws7_gtk/content/build/nsDocumentViewer.o"
  [28] DocumentViewerImpl::PrintPage(0x917c00, 0x67a2c0, 0x9016c0, 0x98f008,
0xffbeec1c, 0x10), at 0xfc204c48
  [29] nsPagePrintTimer::Notify(0xbcd708, 0x679b80, 0xa, 0x425e0ec9, 0xff087800,
0x21), at 0xfc247cec
dbx: warning: can't find file
"/shared/bigtmp2/mozilla/2002-08-04-08-trunk/objdir_ws7_gtk/xpcom/build/nsTimerImpl.o"
  [30] nsTimerImpl::Fire(0x679b80, 0x2acc, 0x1, 0xff1757f0, 0xff1674a8, 0x2800),
at 0xff084be8
  [31] handleTimerEvent(0xbb46c8, 0xff1757e8, 0xe27e0, 0xfef3d4a8, 0x0,
0xff1674a8), at 0xff084e54
dbx: warning: can't find file
"/shared/bigtmp2/mozilla/2002-08-04-08-trunk/objdir_ws7_gtk/xpcom/build/plevent.o"
  [32] PL_HandleEvent(0xbb46c8, 0xff084cbc, 0xc68e4, 0x1, 0x0, 0xbb46c8), at
0xff07a924
  [33] PL_ProcessPendingEvents(0xc68e0, 0xff1674a8, 0x1, 0xff152bd0, 0xff152bb8,
0xff175494), at 0xff07a75c
dbx: warning: can't find file
"/shared/bigtmp2/mozilla/2002-08-04-08-trunk/objdir_ws7_gtk/xpcom/build/nsEventQueue.o"
  [34] nsEventQueueImpl::ProcessPendingEvents(0xc68b0, 0x0, 0xeaad8, 0x0, 0x0,
0x80004000), at 0xff07ca30
  [35] event_processor_callback(data = ???, source = ???, condition = ???)
(optimized), at 0xfd3a77b4 (line ~184) in "nsAppShell.cpp"
  [36] our_gdk_io_invoke(source = ???, condition = ???, data = ???) (optimized),
at 0xfd3a7374 (line ~76) in "nsAppShell.cpp"
dbx: warning: can't find file
"/home/gisburn/package-builds/glib/glib-1.2.8/objdir/giounix.lo"
  [37] g_io_unix_dispatch(0x31c618, 0xffbef078, 0x30f7d0, 0x0, 0x0, 0xffbeefe0),
at 0xfe7b2dc8
dbx: warning: can't find file
"/home/gisburn/package-builds/glib/glib-1.2.8/objdir/gmain.lo"
  [38] g_main_dispatch(0xffbef078, 0xb9df8, 0x1, 0x30f7d0, 0xfe85155b, 0x378),
at 0xfe7b6dc8
  [39] g_main_iterate(0x1, 0x1, 0x5, 0xff3e4270, 0xfd39923c, 0x18), at
0xfe7b7bcc
  [40] g_main_run(0x318208, 0x318208, 0x1, 0xfd42adb8, 0xfd42adbc, 0x11bbdc), at
0xfe7b7f64
dbx: warning: can't find file
"/home/gisburn/package-builds/gtk+/gtk+-1.2.8/objdir/gtk/gtkmain.lo"
  [41] gtk_main(0xc68b0, 0xc84b0, 0xffbef154, 0xffbef158, 0x0, 0xfd3a7d24), at
0xfe9d60a0
  [42] nsAppShell::Run(this = ???) (optimized), at 0xfd3a7c78 (line ~332) in
"nsAppShell.cpp"
  [43] nsAppShellService::Run(this = ???) (optimized), at 0xfd4aed6c (line ~451)
in "nsAppShellService.cpp"
  [44] main1(argc = ???, argv = ???, nativeApp = ???) (optimized), at 0x1a2c0
(line ~1518) in "nsAppRunner.cpp"
  [45] main(argc = ???, argv = ???) (optimized), at 0x1aca8 (line ~1878) in
"nsAppRunner.cpp"
-- snip --

Comment 7

[Sam J. Fleet]

Found the problem!  Give me 30 minutes to make the test html file and see if
this cause you the same problem.  It turned out that this line of html script
somehow cause the browser to crash.  When I removed this line of code, the
browser no longer crashed.

--clip--
<thead>
  <tr>
   <th>FltNum</th>
   <th>FltDate</th>
   <th>Origin</th>
   <th>Dest</th>
   <th>ACNum</th>
   <th>Reason</th>
   <th>sort</th>
   <th>Pax<br>Tot/Lost</th>
   <th>User Name</th>
   <th>Last Update</th>
  </tr>
</thead>
--clip

Comment 8

[Sam J. Fleet]

Attachment: Test #1 (This one cause the crash)

(This one cause the crash)

Comment 9

[Sam J. Fleet]

Attachment: Test #2 (This one doesn't cause the crash)

This one work fine, no crash was found

Comment 10

[Sam J. Fleet]

Attachment: Test #2 (This one doesn't cause the crash)

This one work fine, no crash was found

Comment 11

[Sam J. Fleet]

Attachment: Test #3 (Same as Test #1 but without the thead tag)(No crash)

The "<thead>" tag was removed and there was no crash.

Comment 12

[Brian Foddy]

I agree with all your additional examples, both those that
crash and those that don't.

Comment 13

[Sam J. Fleet]

I'm so sorry for the comment #10 and the duplicate test file, Test #2.  Got
mixed up when walking back and forth with the back/forward button and got mixed up.

Okay!  I'll need you to test it to confirm if this problem is what we think is
happening.  Then I'll explain what the problem is that cause the crash in the
first place.

1) Run Test #1 
    --> Do Test Print.  (Result:  Browser Crash)
    --> Do Test Print Preview & scroll to the 2nd page.  (Result: Browser Crash)

2) Run Test #2 
    --> Do Test Print.  (Result:  No Crash)
    --> Do Test Print Preview (It is one page only).  (Result: No Crash)

3) Run Test #3
    --> Do Test Print.  (Result:  No Crash)
    --> Do Test Print Preview & scroll to the 2nd page.  (Result: No Crash).

Does your test result turned out as expected on Test #1, #2 & #3?  This test I
did is based on Win2000.  I have confidence that this will will perform the same
result as it will on Solaris.  (Hopefully).

Those test files use only one html table.  Based on all 3 tests I perform.  It
should that html tag "<thead>" is responsible for the browser crash.  (Test #1)
This tag along work great if the webpage is only 1 page long.  (Test #2) If the
data on the webpage is longer than 1 page, making it 2 pages or more.  This
cause the browser to crash.  (Test #3) To confirm my theory, I took out the html
tag, "<thead>" and use the page that is 2 pages or more.  The browser does not
crash.  

So, using the html tag, "<thead>" in the html table and if the table itself is
longer than 1 page, this somehow make the web browser become unstable for some
reason.  Hope that help!

Comment 14

[Brian Foddy]

Clarification on my test agreements.  I actually only tested
with a real print, not a preview.  But your description makes
sense  that it seems common to both, and needs to be 2 pages long.

Comment 15

[rods (gone)]

crash in tables

Comment 16

[Sam J. Fleet]

Understandable!  Not a problem with that!  Seem the Print Preview & Printing
only affected Win2000.  Win2000 and Sun Solaris use different architecture, so
that is understandable.  I have the confidence that Mozilla Engineer will find
the solution to the problem.  

Let's hope the Mozilla Engineer or any volunteers who would love to take this
assignment and work on this problem.  :-)  Let's vote for this bug.

Comment 17

[Brian Foddy]

I never actually tested the preview as a crash.  Its rather painful every
crash because I have to reload my .mozilla directory.  At least I have a 
backup copy on disk now.

I am amazed how fast this has been looked at, broken down and identified
as a reproduceable crash element.  Thanks to everyone for helping with this.
The example crash page is a major page that our office needs to use so
to us its a show-stopper.

Brian

Comment 18

[Sam J. Fleet]

Can somebody change the platform setting to all instead of Sun?  Again, can
somebody change the O/S setting to all instead of Solaris?  There was a crash in
Windows on PC also.  Thanks!  I don't have the authority to do this.  :-)

Comment 19

[greer]

The stack for this crash (see incident #9145117) matches that in bug 159358,
which already has a patch and a review.

-> DUPE

*** This bug has been marked as a duplicate of 159358 ***

Comment 20

[sujay]

verified.