WINE Regression due to patch applied Feb 22, 2005 10:50 CDT

[Arren Lex]

I've just finished regression testing WINE using the instructions at 
http://winehq.com/site/docs/wine-devel/x1375

One of my favourite applications, the open-source and freeware KeyNote, 
available at http://www.tranglos.com/free/keynote.html , experienced a 
regression between WINE 20050211 and 20050310. I've been too lazy to 
regression test it for a long time, preferring simply to use an older 
version of wine, but in honour of WINE's pending non-alpha release I went 
ahead and tested it.

Keynote used to run perfectly under WINE <=20050211, but after 20050310, it 
doesn't launch: it spits out the error message "Exception EAccessViolation 
in module kernel32.dll at 0004FF5D. Access violation at address 404DFF5D in 
module 'kernel32.dll'. Read of address 00000000." and quits. This regression 
still exists in the latest version of WINE.

I found WINE's ability to run the application was broken between 20050222 
10:50:10 CDT and 20050222 10:50:15 CDT. Watching the output of 'cvs update', 
it seems three files were modified during these five seconds:
U dlls/kernel/string.c
U dlls/shell32/shell32.spec
U include/winbase.h

I don't know which one is the culprit; I'm not a programmer. I just hope 
I've given you guys enough information to fix it!

Keep WINE alive. It's an awesome utility.

P.S. Is there some sort of list of patches and when they were applied? The 
regression testing page above doesn't link to one (maybe it should!) and I 
wasn't able to find one I could check against.

_________________________________________________________________
Don't just Search. Find! http://search.sympatico.msn.ca/default.aspx The new 
MSN Search! Check it out!

[paul]

Arren Lex wrote:
> I've just finished regression testing WINE using the instructions at 
> http://winehq.com/site/docs/wine-devel/x1375
>
> One of my favourite applications, the open-source and freeware KeyNote, 
> available at http://www.tranglos.com/free/keynote.html , experienced a 
> regression between WINE 20050211 and 20050310. I've been too lazy to 
> regression test it for a long time, preferring simply to use an older 
> version of wine, but in honour of WINE's pending non-alpha release I 
> went ahead and tested it.
>
> Keynote used to run perfectly under WINE <=20050211, but after 20050310, 
> it doesn't launch: it spits out the error message "Exception 
> EAccessViolation in module kernel32.dll at 0004FF5D. Access violation at 
> address 404DFF5D in module 'kernel32.dll'. Read of address 00000000." 
> and quits. This regression still exists in the latest version of WINE.
>
> I found WINE's ability to run the application was broken between 
> 20050222 10:50:10 CDT and 20050222 10:50:15 CDT. Watching the output of 
> 'cvs update', it seems three files were modified during these five seconds:
> U dlls/kernel/string.c
> U dlls/shell32/shell32.spec
> U include/winbase.h
>
> I don't know which one is the culprit; I'm not a programmer. I just hope 
> I've given you guys enough information to fix it!
>
> Keep WINE alive. It's an awesome utility.
>
> P.S. Is there some sort of list of patches and when they were applied? 
> The regression testing page above doesn't link to one (maybe it should!) 
> and I wasn't able to find one I could check against.
>
> Don't just Search. Find! http://search.sympatico.msn.ca/default.aspx The 
> new MSN Search! Check it out!

Go to gmane.comp.emulators.wine.devel.cvs and look at date you have 
found and you will find this link to the patch. 
http://cvs.winehq.org/patch.py?id=16245
and the log message Make lstr* functions inline inside Wine.

It would be best to also file a bug and include Regression and patch id. 
Download keyword url for the application.

A starting point to move things along and narrow down the location of 
the problem is to with current CVS run

WINEDEBUG=+relay wine pathtoyourapp/app.exe

a 100 or so lines before the exception along with the backtrace should 
be added to the bug. At that point additional traces may be requested.

Paul

[James Hawkins]

On 10/8/05, Arren Lex  wrote:
> I've just finished regression testing WINE using the instructions at
> http://winehq.com/site/docs/wine-devel/x1375
>
> One of my favourite applications, the open-source and freeware KeyNote,
> available at http://www.tranglos.com/free/keynote.html , experienced a
> regression between WINE 20050211 and 20050310. I've been too lazy to
> regression test it for a long time, preferring simply to use an older
> version of wine, but in honour of WINE's pending non-alpha release I went
> ahead and tested it.
>
> Keynote used to run perfectly under WINE <=20050211, but after 20050310, it
> doesn't launch: it spits out the error message "Exception EAccessViolation
> in module kernel32.dll at 0004FF5D. Access violation at address 404DFF5D in
> module 'kernel32.dll'. Read of address 00000000." and quits. This regression
> still exists in the latest version of WINE.
>
> I found WINE's ability to run the application was broken between 20050222
> 10:50:10 CDT and 20050222 10:50:15 CDT. Watching the output of 'cvs update',
> it seems three files were modified during these five seconds:
> U dlls/kernel/string.c
> U dlls/shell32/shell32.spec
> U include/winbase.h
>
> I don't know which one is the culprit; I'm not a programmer. I just hope
> I've given you guys enough information to fix it!
>
> Keep WINE alive. It's an awesome utility.
>
> P.S. Is there some sort of list of patches and when they were applied? The
> regression testing page above doesn't link to one (maybe it should!) and I
> wasn't able to find one I could check against.

The patch that changed these files is:

ChangeSet ID:   16245
CVSROOT:        /opt/cvs-commit
Module name:    wine
Changes by:     [e-mail]   2005/02/22 09:50:14

Modified files:
       include        : winbase.h
       dlls/shell32   : shell32.spec
       dlls/kernel    : string.c

Log message:
       Mike McCormack 
       Make lstr* functions inline inside Wine.

The way I check what patch changed certain files at certain time
intervals is cvs.winehq.org/cvsweb/wine and then the path to the file
in wine relative to the wine directory.

--
James Hawkins

[Mike McCormack]

James Hawkins wrote:
> > I found WINE's ability to run the application was broken between 20050222
> > 10:50:10 CDT and 20050222 10:50:15 CDT. Watching the output of 'cvs update',
> > it seems three files were modified during these five seconds:
> Log message:
>        Mike McCormack 
>        Make lstr* functions inline inside Wine.

This patch means that internally in Wine, there's no exception handlers 
around the lstr* functions, so passing a invalid pointer to a lstrcpyW 
(for example) will crash (and so it should).

KeyNote is causing a NULL pointer to be used with one of the lstr* 
functions in Wine, and this is causing the crash.

I tried installing KeyNote, and it seems to have a richedit problem... 
it crashes in EM_GETTEXTRANGE around line 1579 in editor.c... making the 
  allocation at line 1573 bigger seems to solve that.

Following that, it fails when trying to create directories:

000b:Call kernel32.CreateDirectoryA(7f0d5740 "C:\Program 
Files\KeyNote",00000000) ret=0043ab97
000b:Ret  kernel32.CreateDirectoryA() retval=00000001 ret=0043ab97
000b:Call kernel32.SetErrorMode(00000001) ret=00422fa1
000b:Ret  kernel32.SetErrorMode() retval=00000000 ret=00422fa1
000b:Call kernel32.GetFileAttributesA(7f0e573c "C:\Program 
Files\KeyNote") ret=00422fca
000b:Ret  kernel32.GetFileAttributesA() retval=00000010 ret=00422fca

...

000b:Call kernel32.CreateFileA(7fbcf6f0 "C:\Program 
Files\KeyNote",c0000000,00000000,00000000,00000002,000
00080,00000000) ret=00404820
000b:Ret  kernel32.CreateFileA() retval=ffffffff ret=00404820
000b:Call kernel32.GetLastError() ret=0040485a
000b:Ret  kernel32.GetLastError() retval=00000005 ret=0040485a
000b:Call user32.LoadStringA(00400000,0000ff8c,7fbcf250,00000400) 
ret=0040613e
000b:Ret  user32.LoadStringA() retval=00000012 ret=0040613e
000b:Call kernel32.RaiseException(0eedface,00000001,00000007,7fbcf684) 
ret=0043df7d

Don't know what it's expecting to happen there...

Anyway, if you send me a backtrace of where it's crashing when you try 
start it up, it should be easy enough to fix...

Mike

[Krzysztof Foltman]

Mike McCormack wrote:
> I tried installing KeyNote, and it seems to have a richedit problem... 
> it crashes in EM_GETTEXTRANGE around line 1579 in editor.c... making the 
>  allocation at line 1573 bigger seems to solve that.

Can you test it with the latest uncommitted patches applied? 
(particularly, the recent 1.0 emulation "1/3" patch - ME_GetTextW fix - 
seems to be relevant). The same patch has fixed a similar problem with 
heap corruption in a different application (Revelation Patch).

Krzysztof

[Mike McCormack]

Krzysztof Foltman wrote:
> Can you test it with the latest uncommitted patches applied? 
> (particularly, the recent 1.0 emulation "1/3" patch - ME_GetTextW fix - 
> seems to be relevant). The same patch has fixed a similar problem with 
> heap corruption in a different application (Revelation Patch).

Hmm. Still crashes.  The download isn't too big...

Mike