53517 - crash on startup in nsImageBoxFrame::UpdateImageFrame

Status: VERIFIED FIXED

(SeaMonkey :: General, defect, P3)

Description

[Ben Goodger (use ben at mozilla dot org for email)]

Using the 8pm Mac optimised mozilla build from sweetlou, start Mozilla. 
Profile manager comes up fine, however when you select a profile and 
start with it, shortly after the Navigator window is displayed the browser 
crashes. I'll attach a stack trace. 

This happens erratically. In some runs it is reproducable time after time... 
although I've also been able to get this build to start, although on a 
different reboot-cycle.

Comment 1

[Ben Goodger (use ben at mozilla dot org for email)]

Attachment: stack trace

Comment 2

[Kathleen :Brade]

reassign to evaughan since he has a checkin to nsHTMLImageLoader.cpp

I'm running into this at startup; I can't test/debug/fix if I can't run the 
product.  Upgrading severity to blocker.

Comment 3

[Blake Ross]

really reassigning to evaughan

Comment 4

[Jon Granrose]

This didn't show up in this morning's smoketests, but I'm holding the tree until
I see some progress on this.

Comment 5

[Tracy Walker [:tracy]]

I did see this happen this morning on the mac smoketest (comm build 
2000-09-21-04-m18)...but it happened just the first time I attempted to launch. 
 After that it ran fine.  Though the profile manager did not come up upon later 
launches.

Comment 6

[buster]

FWIW: WinNT debug build bits from around 11:00pm last night, no crash after
repeated attempts to reproduce.

Comment 7

[Asa Dotzler [:asa]]

I did not see this on the first launch but did see it every launch after that on
mozilla Mac Installer build from this morning.

Comment 8

[Asa Dotzler [:asa]]

This is happening on both the Installer and the sea.bin Mozilla builds.

Comment 9

[Kathleen :Brade]

update: I backed out evaughan's checkins and I no longer can reproduce the crash.

Comment 10

[Jon Granrose]

sounds like a debug vs opt problem.  eric's looking at this.  if we get win32
and linux builds and smoketesting complete before he has a fix, we'll back him out.

Comment 11

[Kathleen :Brade]

This appears to be Mac-specific.  I don't see it on Linux (at least not so far). 



Everyone--please be sure to confirm that the stack trace you are seeing is 

similar to the one Ben attached (and not the crash having to do with the MRJ 

Plugin or another crash on launch bug).

Comment 12

[Kathleen :Brade]

granrose--no, I don't think it's a debug vs optimized.  My debug build exhibits 
the crash and Asa sees it in (presumably) optimized builds.

Also note:  I have seen this crash (same stack crawl) while running the product 
(assuming I can get it to start at all which is only about 20% of the time on my 
Mac).

Comment 13

[Asa Dotzler [:asa]]

seeing this on mozilla win32 talkback builds this morning as well.  stack looks 
like this

 nsHTMLImageLoader::Update
[d:\builds\seamonkey\mozilla\layout\html\base\src\nsHTMLImageLoader.cpp,line149]
nsHTMLImageLoader::ImageLoadCB
[d:\builds\seamonkey\mozilla\layout\html\base\src\nsHTMLImageLoader.cpp,line134]
nsFrameImageLoader::NotifyFrames
[d:\builds\seamonkey\mozilla\layout\base\src\nsFrameImageLoader.cpp,line574]
nsFrameImageLoader::Notify
[d:\builds\seamonkey\mozilla\layout\base\src\nsFrameImageLoader.cpp,line544]
ns_observer_proc
[d:\builds\seamonkey\mozilla\gfx\src\nsImageRequest.cpp,line135]
XP_NotifyObservers
[d:\builds\seamonkey\mozilla\modules\libutil\src\obs.c,line260]
il_image_complete_notify
[d:\builds\seamonkey\mozilla\modules\libimg\src\if.cpp,line329]
il_image_complete
[d:\builds\seamonkey\mozilla\modules\libimg\src\if.cpp,line1653]
ImgDCallbk::ImgDCBHaveImageAll
[d:\builds\seamonkey\mozilla\modules\libimg\src\if.cpp,line191]
process_buffered_gif_input_data
[d:\builds\seamonkey\mozilla\modules\libimg\gifcom\gif.cpp,line694]
gif_delay_time_callback
[d:\builds\seamonkey\mozilla\modules\libimg\gifcom\gif.cpp,line728]
timer_callback
[d:\builds\seamonkey\mozilla\gfx\src\nsImageSystemServices.cpp,line72]
nsTimer::Fire
[d:\builds\seamonkey\mozilla\widget\timer\src\windows\nsTimer.cpp,line196]
nsTimerManager::FireNextReadyTimer
[d:\builds\seamonkey\mozilla\widget\timer\src\windows\nsTimerManager.cpp,line117
]
nsAppShell::Run
[d:\builds\seamonkey\mozilla\widget\src\windows\nsAppShell.cpp,line116]
0x01be220c

Comment 14

[Asa Dotzler [:asa]]

I sent in 4 talkback reports that all had this same crash at starutp but I had 
no problems with the non-talkback mozilla zipped build.

Comment 15

[Kathleen :Brade]

*** Bug 53583 has been marked as a duplicate of this bug. ***

Comment 16

[Kathleen :Brade]

evaughan sent me a fix; I'm rebuilding with it to test on Mac.
Changing platforms/os to all since waterson and asa have seen problems on 
Windows.

Comment 17

[Mike Pinkerton (not reading bugmail)]

i see this on both mac & win32 from builds from last night.

Comment 18

[Mike Pinkerton (not reading bugmail)]

*** Bug 53515 has been marked as a duplicate of this bug. ***

Comment 19

[Asa Dotzler [:asa]]

after repeated attempts I finally got my mac build to start and ran through most 
of the smoketests.  Attempting to switch from teh default classic theme to the 
Modern theme crashed me with the same stack trace as the startup crash.

Comment 20

[Kathleen :Brade]

evaughan's fix works for me on MacP

Comment 21

[Asa Dotzler [:asa]]

so is this getting checked in?

Comment 22

[Eric Vaughan]

Yep my changes are in.

Comment 23

[Jim Roskind]

Putting on dogfood-plus radar.

Please mark this as fixed if the correction has landed.
Thx,
Jim

Comment 24

[Doron Rosenberg (IBM)]

asa as qa contact, as he sees this on win32 (I don't), plus he has a mac

Comment 25

[Eric Vaughan]

fixed

Comment 26

[pnunn]

*** Bug 53317 has been marked as a duplicate of this bug. ***

Comment 27

[Kathleen :Brade]

verifying; I can't reproduce this anymore

Comment 28

[John Morrison]

*** Bug 53751 has been marked as a duplicate of this bug. ***