Firefox crashes

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Normal installation with Windows Service method
2. Open firefox, use normally.

What is the expected output? What do you see instead?
Some random websites are not loaded properly at random times. Today logging
into Facebook caused firefox to corrupt. A few weeks ago Wikipedia caused
the same problem, but only in a specific page (eszet article). This page
does not cause problems today. I am attaching a screenshot of firefox. I
cannot change tabs or click on anything. If try to close firefox, "Close
Firefox" dialog does not have any buttons. Restarting firefox does not
solve the problem. I observe the same problem with ezgdi also.

What version of gdipp are you using? On what operating system (with
bitness)?
Latest snapshot from the repository, I built it myself.

Please provide any additional information below.

Original issue reported on code.google.com by ibrahimk...@gmail.com on 22 Mar 2010 at 8:06

Attachments:

• firefox.png

[GoogleCodeExporter]

I can visit Facebook without any problem. I have two recommendations for you. 
1) wait
the upcoming 0.7 release, which is a major improvement. Check if the problem 
still
exists. 2) If 1) fails, build the debug version, check which assertion fails. 
Report
me back and I will take a look at it. Thanks.

Original comment by crendk...@gmail.com on 22 Mar 2010 at 9:29

[GoogleCodeExporter]

0.7.0 still has the problem. I'll try the debug version. Where is the log file 
located?

Original comment by ibrahimk...@gmail.com on 22 Mar 2010 at 10:46

[GoogleCodeExporter]

You are quick. No log file is created. I simply make assertion after each 
important
function call. You should get a assertion fail dialog with error file and line 
number
if such function call returns incorrect code.

Original comment by crendk...@gmail.com on 22 Mar 2010 at 10:54

[GoogleCodeExporter]

I built the debug version, but I could not get the executables to work. Debug
versions of loader, service and demo do not work. Debug version of gdimm.dll 
works
with release versions of the executables. What am I doing wrong?

Original comment by ibrahimk...@gmail.com on 22 Mar 2010 at 11:16

[GoogleCodeExporter]

I do not know. They work fine in my case. You can debug the executables if you 
want.
However, what you need is only the debug version gdimm.dll. Substitute the 
release
version and visit Facebook.

Original comment by crendk...@gmail.com on 22 Mar 2010 at 11:43

[GoogleCodeExporter]

There are some suggestions to debug gdipp: 
http://code.google.com/p/gdipp/wiki/HowToBuild

Original comment by crendk...@gmail.com on 23 Mar 2010 at 2:14

[GoogleCodeExporter]

Thanks for the instructions. MS Linker segfaults when building gdipp_loader, 
demo and
svc. It would be great if you can upload the debug build.

Original comment by ibrahimk...@gmail.com on 23 Mar 2010 at 3:18

[GoogleCodeExporter]

Find the debug version in the download tab.

Original comment by crendk...@gmail.com on 23 Mar 2010 at 4:59

[GoogleCodeExporter]

I use a clean Firefox with no addon. Currently no problem. However, my chrome 
with 9
addons crashed every-time it starts.

I think it may be related to addons in these two browsers.

Original comment by DCatcher...@gmail.com on 23 Mar 2010 at 8:34

[GoogleCodeExporter]

Firefox was crashing in safe mode also. Latest snapshot from the repository 
solves
the problem.

Original comment by ibrahimk...@gmail.com on 24 Mar 2010 at 10:16

[GoogleCodeExporter]

No, now forums.adobe.com crashes firefox. Interesting.

Original comment by ibrahimk...@gmail.com on 25 Mar 2010 at 4:27

[GoogleCodeExporter]

Solved. The forums.adobe.com problem is different than multi-threading. If you 
can
debug, you can clearly see that the lpDx array in that call contains negative
distances, therefore the calculated bitmap width is negative, leading to a 
invalid
BITMAPINFO for CreateDIBSection. Solution is easy, abs(bmp_width).

Yes, there is no restriction of using negative distances in lpDx, per MSDN
(http://msdn.microsoft.com/en-us/library/dd162713%28VS.85%29.aspx). It is just 
not
orthodox.

Anyway, thanks for catching this bug.

Original comment by crendk...@gmail.com on 25 Mar 2010 at 6:35

• Changed state: Fixed