28380 - After profile window closes, no browser window appears

Status: VERIFIED FIXED

(SeaMonkey :: General, defect, P3)

Description

[Jason L. Grace]

No coring, just no browser window appearing.  Here is the last printouts of the
run (including an automatic stack-trace I make with an interrupt):

ProfileManager : GetProfileDir
ProfileName : mozProfile
ProfileDir  : /home/h9/jgrace/.mozilla/mozProfile
ProfileManager : GetProfileDir
nsWidget::~nsWidget() of toplevel: 4 widgets still exist.

( 0)  0x00013f90   ae_handler__Fi + 0x1c  [././mozilla-bin]
( 1)  0xc013ece0   _sigreturn  [/usr/lib/libc.1]
( 2)  0xc013eac0   __syscall + 0x50  [/usr/lib/libc.1]
( 3)  0xc0149284   syscall + 0x4c  [/usr/lib/libc.1]
( 4)  0xc1c8e524   poll + 0x50 
[/work/gnu/mozilla/obj-hppa1.1-hp-hpux10.20/dist/bin/libnspr3.sl]
( 5)  0xc1e94a1c   g_main_poll + 0x304  [/work/gnu/common/lib/libglib-1.2.sl.0]
( 6)  0xc1e93cb8   g_main_iterate + 0x71c 
[/work/gnu/common/lib/libglib-1.2.sl.0]
( 7)  0xc1e944bc   g_main_run + 0xfc  [/work/gnu/common/lib/libglib-1.2.sl.0]
( 8)  0xc206bc14   gtk_main + 0x218  [/work/gnu/common/lib/libgtk-1.2.sl.5]
( 9)  0xc3b55f08   Run__10nsAppShellFv + 0xa4 
[/work/gnu/mozilla/obj-hppa1.1-hp-hpux10.20/dist/bin/libwidget_gtk.sl]
(10)  0xc3896f74   Run__17nsAppShellServiceFv + 0x30 
[/work/gnu/mozilla/obj-hppa1.1-hp-hpux10.20/dist/bin/components/libnsappshell.sl]
(11)  0xc3807a24   LoadDefaultProfileDir__9nsProfileFR9nsCString + 0x6b8 
[/work/gnu/mozilla/obj-hppa1.1-hp-hpux10.20/dist/bin/components/libprofile.sl]
(12)  0xc38072d8   StartupWithArgs__9nsProfileFP17nsICmdLineService + 0x84 
[/work/gnu/mozilla/obj-hppa1.1-hp-hpux10.20/dist/bin/components/libprofile.sl]
(13)  0x0000fee0   main1__FiPPcP15nsISplashScreen + 0x830  [././mozilla-bin]
(14)  0x00011c54   main + 0x2cc  [././mozilla-bin]
(15)  0xc00656bc   _start + 0x8c  [/usr/lib/libc.1]
(16)  0x00009d14   $START$ + 0x134  [././mozilla-bin]

Comment 1

[Christine Begle]

Hi, jgracer.  Would you let us know the last time you pulled and built?  That 
kind of information is good to include in any bug you file.

jdunn, i'm reassinging to you since it's my understanding you're the HPUX 
expert around here.  let me know if that's wrong and who i should be assigning 
these to instead.

I looked at http://tinderbox.mozilla.org/showbuilds.cgi?tree=SeaMonkey-Ports,
and HPUX 10.20 has been green for the last few days i looked at.  However, it 
doesn't look like the tests are really being run -- there's no Mozilla console 
output in the logs.  Might be a good idea to get these working.

Comment 2

[Jason L. Grace]

Seems to be a problem in the profile's nsAppShellService::Run, which may be
hanging.  Occasionally it DOES complete, perhaps when my mouse is moved in some
specific way.  Should that happen, it goes on to the following error:

initialized appshell
GFX: dpi=96 t2p=0.0666667 p2t=15 depth=8
WEBSHELL+ = 1
XXX WARNING: Number of webshells being leaked: 1 
~nsProfile 
###!!! ASSERTION: NS_ENSURE_TRUE(docShellElement) failed: 'docShellElement',
file ../../../../xpfe/appshell/src/nsXULWindow.cpp, line 710
###!!! Break: at file ../../../../xpfe/appshell/src/nsXULWindow.cpp, line 710
###!!! ASSERTION: NS_ENSURE_TRUE(docShellElement) failed: 'docShellElement',
file ../../../../xpfe/appshell/src/nsXULWindow.cpp, line 639
###!!! Break: at file ../../../../xpfe/appshell/src/nsXULWindow.cpp, line 639
nsWidget::~nsWidget() of toplevel: 1 widgets still exist.
CanUnload_enumerate: skipping native

I want to add, in case it's some thread problem, that I'm using classic threads,
not posix or dce stuff.  These problems are rather new (I have the latest code I
could get).  I use ansi C and hpux 10.20 as always, and run with "mozilla" and
no arguments.

Comment 3

[Jason L. Grace]

I had nspr logging at its highest; when I turned the logging off, the problem
is gone, but the XUL problem is still there.

Comment 4

[Jason L. Grace]

Somehow it got fixed; maybe the Shaver stuff with threads?

Comment 5

[Jim Dunn]

marking verified... no qa u know.