85370 - Typing URL often causes segfault (0.9.1/SunOS5.7/PSM2)

Status: VERIFIED WORKSFORME

(SeaMonkey :: Location Bar, defect)

Description

[david.butts]

Typing a URL[1] in the URL bar can cause a segfault.  I have found no site which
is consistently safe or consistently causes a crash.  The segfault can occur
before or after the protocol declaration (either http or https), or with a bare
hostname, so it doesn't seem to be related to the content being typed, but
rather to the act of typing.  I have also not yet had any problems following links.

The crashes seem more frequent w/ Google as the search tool, instead of the
default settings, but my data set is not particularly large.  In any case, I
have not yet managed to type five URLs without a crash.

This isn't an official build, but is built off of unmodified 0.9.1 source.

GTK+ v1.2.10
glib v1.2.10
gdk_pixbuf 0.11.0
libIDL 0.6.5

Based on the stack trace it looks to my ill-trained eye that it may also be a
GTK issue, but the URL bar is the only place the symptom asserts itself.

A sample stack trace (retyped, hopefully with few typos):
#0 gdk_window_ref (window=0x0) at gdkwindow.c:718
#1 0xfefb4a2c in gdk_event_copy (event=0x2d9048) at gdkevents.c:672
#2 0xfefb4944 in gdk_event_put (event=0x2d9048) at gdkevents.c:620
#3 0xfde37560 in handle_key_release_event () from components/libwidget_gtk.so
#4 0xfde37ca4 in handle_gdk_event() from components/libwidget_gtk.so
#5 0xfde377d4 in handle_gdk_event() from components/libwidget_gtk.so
#6 0xfefb63cc in gdk_event_dispatch (source_data=0x2d8fe8,      
current_time=0xffbef6e0m user_data=0x0) at gdkevents.c:2139
#7 0xfee2ea9c in g_main_dispatch (dispatch_time=0xffbef6e0) at gmain.c:656
#8 0xfee2f2c0 in g_main_iterate (block=-18557660, dispatch=1) at gmain.c:877
#9 0xfee2f478 in g_main_run (loop=0x2b8ff0) at gmain.c:935
#10 0xfef07d9c in gtk_main () at gtkmain.c:524
#11 0xfde2ab84 in nsAppShell::Run () from components/libwidget_gtk.so
#12 0xfe88b2c8 in nsAppShellServiceLLRun () from components/libnsappshell.so
#13 0x1d9fc in NS_CreateNativeAppSupport ()
#14 0x1e950 in main ()

If I have submitted this or a similar bug more than once, I apologize.  There
does not seem to be much indication that the report was accepted, and I don't
know how long to expect it to take for a new report to show up in a query. 
FWIW, I have read more of TFM in the interim, so this is hopefully the most
useful version.

Thanks-
David

Comment 1

[Alec Flett]

I've seen a lot of talkback reports on the nsSliderFrame bug causing crashes
when typing in the url bar...probably a dupe of that..

Comment 2

[david.butts]

Having looked at a couple of the nsSliderFrame bugs, this still feels
different to me.  For one thing, I haven't managed to get any other
typing to generate a segfault (witness like this comment), so I still
suspect that it is somehow related to the URL bar.

The presence of history (via the drop-down from the URL bar in my
case, or the back or refresh buttons in others) is interstingly
consistent across several of the nsSliderFrame bugs, but the stack
traces have essentially nothing in common with mine.

Unfortunately, I'm sufficiently short on disk space on my other boxes
that I can't try to duplicate this on other hardware/OS combos.

Comment 3

[Keyser Sose]

Marking NEW.

Comment 4

[Paul Chen]

nav triage team:

reassigning to [e-mail] since it's reported as a solaris issue

Comment 5

[Jim Dunn]

reassigning to rich cuz i can't figure out margaret's id

Comment 6

[david.butts]

Hello-

My apologies for not updating this sooner.

I've been using 0.9.2 since a matter of hours after its release, and haven't
seen a single segfault, so I think it's safe to consider this bug fixed.  I'd be
curious to know whether it was actually a corollary of the ns_slider_frame, but,
(a) You folks are undoubtedly still ridiculously busy, and (b) My persistant
neglect of this ticket once the symptom went away is probably not going to earn
me any points :)

I am tempted to mark this fixed myself, but don't want to step on any toes or
short-circuit any auditing that may be in place.  Of course, given that there
are currently no votes for it, I'd probably be safe...

All around, I have to say that 0.9.2 is just excellent, and, though I'll build
and test the new versions when they come out, it will probably be a while before
I actually package another one to replace this version as my default install. 
Great work!

Comment 7

[Rich Burridge]

Reassigning to Ed to do with as he wants.

Comment 8

[edburns]

Marking WORKSFORME.  Qa, please reopen

Comment 9

[Alexei V. Mokeev]

tryed several times, but 0.9.1 for Sparc Solaris 2.8 do not crash.
Mark verified WORKSFORME