128120 - Printing fails after first invocation

Status: VERIFIED FIXED

(Core :: Printing: Output, defect)

Description

[Peter Janes]

On Linux nightly -sea build 2002022708, I am only able to print once. 
Subsequent attempts report "You cannot print while in print preview." even if
I've never entered print preview since starting the browser.

I've tried going into print preview mode (which has a "Print" button that
displays the same error dialog) and exiting, but the behaviour is the same.

Comment 1

[sujay]

twalker, please check in tomorrow's smoke builds...

Comment 2

[sujay]

Boris, do you see this problem ?

Comment 3

[Boris Zbarsky [:bzbarsky]]

Yes, I do (build 2002-02-27-08).

Comment 4

[Boris Zbarsky [:bzbarsky]]

This is caused by the patch for bug 127627.  Backing it out fixes the problem. 
Are we forgetting to decrement instance_counter somewhere?

Comment 5

[rods (gone)]

That's what I was wondering about (the decrement), I think the best would be to
back it out unless there is an easy fix.

Comment 6

[Roland Mainz]

The decrement for the instance counter is "done" in the destructor... just
wondering... is this maybe a duplicate of bug 121620 ("Closing TABs causes leak
of print device context") or another devicecontext leak ?

If we leak the device context (or better: all device contexts) and only use them
in the order we created them (which seems to be not the case for the PP
crashers) - then we can back out the patch for bug 127627 for now. Or better:
make the if()-checks assertions...

Comment 7

[Roland Mainz]

Accepting bug...

Comment 8

[Roland Mainz]

Some results:
1. I can print multiple times with the Xprint module
2. I can only print once with the PostScript module (the |instance_counter| gets
never decremented)

Looks we leak something within the gfx/src/ps/ area which finally ends in
leakage of the nsDeviceContextPS object... xx@@@@!!... ;-( 

Comment 9

[Roland Mainz]

Attachment: Patch for 2002-02-27-08-trunk

Fix was sponsored by the "stop-leaking-objects"-campaign ...

:)

Comment 10

[Roland Mainz]

Requesting r=/sr=/a= ...

Comment 11

[rods (gone)]

Comment on attachment 71875 [details] [diff] [review]
Patch for 2002-02-27-08-trunk

Yes, this is exactly it!

r=rods

Comment 12

[Suresh]

a=suresh (sheriff for thursday). Get a sr and checkin.

Comment 13

[Christopher Blizzard (:blizzard)]

Comment on attachment 71875 [details] [diff] [review]
Patch for 2002-02-27-08-trunk

sr=blizzard

Get it in.

Comment 14

[timeless]

Comment on attachment 71875 [details] [diff] [review]
Patch for 2002-02-27-08-trunk

checked in

Comment 15

[Suresh]

marking fixed. Thanks Timeless.

Comment 16

[sujay]

Peter  can you print successively now ?


twalker please try printing a few times in a row in today's linux
build....

thanks.

Comment 17

[Tracy Walker [:tracy]]

three in a row, no problem
 marking verified on linux commercial build 2002-03-01-06-trunk

Comment 18

[Peter Janes]

Ditto, no problems here.