394160 - crash @ nsDeviceContextSpecGTK::EndDocument when printing

Status: RESOLVED FIXED

(Core :: Printing: Output, defect)

Description

[Robert Kaiser]

I just crashed my self-compiled debug build of "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9a8pre) Gecko/2007082815 SeaMonkey/2.0a1pre" when trying to print a mail message.
Looking at the stacktrace and searching crash-stats gives results from Firefox that look the same as my crash, so I'll point to those here instead of my own one.

The top ~5 frames from those stacks are the same as mine:
bp-0ce702ae-561b-11dc-a2e4-001a4bd43ef6
bp-eb9d0f75-561a-11dc-951b-001a4bd43ef6
bp-1ae02bee-559c-11dc-bc87-001a4bd43ef6
bp-c5e5ff8c-5597-11dc-b7fd-001a4bd43ef6
Looks like all of http://crash-stats.mozilla.com/report/list?query=nsDeviceContextSpecGTK%3A%3AEndDocument have that signature.

Filing this in printing component as it happens when printing, but it might as well be a widget or thebes bug, looking at the stack.

Comment 1

[Kenneth Herron]

Could this be from bug 367829? mPrintJob would be NULL if MOZ_ENABLE_POSTSCRIPT is false.

Comment 2

[Nickolay_Ponomarev]

I've only checked that patch in. It was romaxa's.

Can MOZ_ENABLE_POSTSCRIPT be 0 though? As I understand, trunk simply wouldn't build with MOZ_ENABLE_POSTSCRIPT=0 before bug 367829 was fixed. So unless Robert (or whoever produced the build) manually tweaked enable-postscript in the build config recently, 367829 shouldn't be the reason for the crash.

Comment 3

[Oleg Romashin (:romaxa)]

Attachment: Crash fix

According this 
http://mxr.mozilla.org/seamonkey/source/widget/src/gtk2/Makefile.in#161

we have to use USE_POSTSCRIPT instead MOZ_ENABLE_POSTSCRIPT

Also on destroy we have to check mPrintJob for null...

Comment 4

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 278825 [details] [diff] [review]
Crash fix

sr=me

Comment 5

[Stuart Parmenter]

Comment on attachment 278825 [details] [diff] [review]
Crash fix

this should just remove the ifdefs entirely

Comment 6

[Kenneth Herron]

The patch for bug 367829 was backed out on 8/31 and crash-stats.mozilla.com doesn't show any crashes here since then. I think this could be closed; bug 394413 is about removing the conditional compilation option.

Comment 7

[Robert Kaiser]

OK, then. Fixed by the backout of bug 367829 (actually it might technically even be INVA now, but the main point is it's RESOLVED in some way).