November 2007 - Wine-Devel - list.winehq.org

Re: widl [1/7]: Output working exception handling macros. [take 3]
by Alexandre Julliard Nov. 29, 2007

Nov. 29, 2007

Dan Hipschman <dsh(a)linux.ucla.edu> writes: > In this revision, I leave setting permissions on page faults to the > unhandled exception filter and simply ignore page faults on execution > access (read access exceptions should be caught since they translate > into bad stub data, but these don't pose a problem anyway). I also > fixed a bug regarding how the mask field of __widl_except_frame was > set. There's no reason to ignore execute access errors, those are bugs too. You are still not filtering the exception properly though, the condition checking has to be done directly in the handler, you can't longjmp back to the code and then re-raise the exception. That's doubly true for the finally handling. This means you have to generate different handlers for the different exception conditions. -- Alexandre Julliard julliard(a)winehq.org

2 2

Interesting Behavior of OS X
by Steven Edwards Nov. 29, 2007

Nov. 29, 2007

When tracking down a crash in the kernel32 loader test, Dmitry found a bug in the Mac OS loader when Wine tried to load his dummy PE file. Upon further research I found that the Mac loader seems to have its own undocumented PE loader built in. I did some further testing with a windows binary and got some really interesting results.... #include <stdio.h> #include <dlfcn.h> int main(int argc, char **argv) { void *handle; const char *error; // handle = dlopen("./ldr330e.tmp", RTLD_NOW | RTLD_FIRST ); handle = dlopen("./procexp.exe", RTLD_NOW | RTLD_FIRST ); if (!handle) { fputs (dlerror(), stderr); printf("I guess it didn't load\n"); exit(1); } printf("I guess it loaded\n"); dlclose(handle); } Its trying to load the dlls. I get this output steven-edwardss-imac:temp sedwards$ ./a.out dlopen(./procexp.exe, 258): Library not loaded: WS2_32.dll Referenced from: /Users/sedwards/Library/Application Support/CrossOver/Bottles/winetest/drive_c/windows/temp/procexp.exe Reason: image not foundI guess it didn't load steven-edwardss-imac:temp sedwards$ file procexp.exe procexp.exe: MS-DOS executable PE for MS Windows (GUI) Intel 80386 32-bit steven-edwardss-imac:temp sedwards$ open . steven-edwardss-imac:temp sedwards$ ./a.out dlopen(./procexp.exe, 258): Library not loaded: MPR.dll Referenced from: /Users/sedwards/Library/Application Support/CrossOver/Bottles/winetest/drive_c/windows/temp/procexp.exe Reason: image not foundI guess it didn't load steven-edwardss-imac:temp sedwards$ ./a.out dlopen(./procexp.exe, 258): Library not loaded: MPR.dll So this leads to the question. Whats going on? Is this a hold over from EFI which is PE by default? Why would the OS need to load the EFI files? Maybe just for easy of development and testing? Or is something else going on? Is Apple going to be adding a win32 compatibility layer to OS X? Is having a native PE loader of any use to us? Thanks Steven -- Steven Edwards "There is one thing stronger than all the armies in the world, and that is an idea whose time has come." - Victor Hugo

1 2

re: wininet [2/2]: Fix error checking for FtpGetCurrentDirectoryW
by Dan Kegel Nov. 29, 2007

Nov. 29, 2007

Hi Zac, you wrote * Fixed errors in error checking in FtpGetCurrentDirectoryW in wininet/ftp.c It's probably a good idea to explain what real-world problem this fixes. One good way to do that is to submit a test first that fails, with todo(wine) around the failing part, and then submit a second patch which fixes the problem and removes the todo(wine) from the test. - if (NULL == lpwfs || WH_HFTPSESSION != lpwfs->hdr.htype) + if ( lpwfs == NULL || WH_HFTPSESSION != lpwfs->hdr.htype ) Why did you make this change? It doesn't fix anything, and it removes an old coder's anti-typo-measure of putting the const on the left hand of an equality test. Please remove this hunk. - if (lpwfs->download_in_progress != NULL) + if ( lpszCurrentDirectory == NULL || *lpdwCurrentDirectory == NULL ) You switched to tabs, tsk. Also, you changed whitespace style. Always try to match the file's existing style. - Dan -- Wine for Windows ISVs: http://kegel.com/wine/isv

1 0

re: wininet [1/2]: Add test for FtpGetCurrentDirectoryA
by Dan Kegel Nov. 29, 2007

Nov. 29, 2007

Nice to see you contributing tests! Three comments: 1. you seem to have ragged indentation. Did you remember to remove all tabs? (Hmm. Our doc on the tab issue is a bit weak: neither http://winehq.org/site/sending_patches http://winehq.org/site/docs/winedev-guide/style-notes really capture the current anti-tab mood.) 2. You have a nice testDir variable, but then you hard-code "pub" somewhere. Use the variable everywhere. 3. Did you make sure the new tests pass on Windows? -- Wine for Windows ISVs: http://kegel.com/wine/isv

1 0

Adding Bugzilla keyword for Copy-protection/Debugger issues
by Ben Hodgetts (Enverex) Nov. 29, 2007

Nov. 29, 2007

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Just a thought but it may be a good idea to add a keyword to Bugzilla for issues related to debuggers or copy-protection, that would help group them all together as at the moment there seem to be many bugs related to breakages from obscure debugger protections or things like GameGuard. It's much cleaner than the hated meta bug idea and is easier to implement. Just a thought. Ben H. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) iD8DBQFHTrfQuqCUopk11kIRAhDnAJ4iGkr6bUk1/FPUFrrdzlsnqC5dDwCfd2r5 WDylDKMPprC22RtEwYG01o4= =fLJB -----END PGP SIGNATURE-----

2 1

Re: [server] set serial port fd back to blocking if necessary
by Alexandre Julliard Nov. 29, 2007

Nov. 29, 2007

"Damjan Jovanovic" <damjan.jov(a)gmail.com> writes: > This fixes a regression caused by commit > f85437c57fb44d511db97edbaa5b1f8abfe75fd3 (bug 9356) where serial ports > break because they are not made blocking. > > Changelog: > * Set serial port fd back to blocking if necessary This will break timeout handling. There shouldn't be any reason for the fd to be blocking, we should handle EAGAIN properly. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: [PATCH 1/3] advpack/tests: Rename the wrappers around HeapAlloc() &Co to use the new standard naming.
by James Hawkins Nov. 28, 2007

Nov. 28, 2007

On Nov 28, 2007 3:39 PM, Michael Stefaniuc <mstefani(a)redhat.de> wrote: > Alexandre, > > i'm unsure about this patch series. Yes this are simple wrappers around > HeapAlloc() that are the same as the "standard" wrappers used in Wine. > But they are used only as callbacks aka different usage. > I don't think this particular series is beneficial. Your script won't be able to use these names because they're not called directly. -- James Hawkins

2 1

d3dx9__xx dlls
by David.Adam＠math.cnrs.fr Nov. 28, 2007

Nov. 28, 2007

Following Alexandre (and I agree with him :D ), one only needs to export the functions of d3dx9_36 (the latest one) to the older d3dx9_xx dlls. So, we just just need to implement the functions in the d3dx9_36 dll repertory. No wine_d3dx9 is useful. David

5 5

Re: [try2 1/1] msi/tests: automation: Fix pointer check before calling lstrcpyW.
by Alexandre Julliard Nov. 28, 2007

Nov. 28, 2007

Misha Koshelev <mk144210(a)bcm.edu> writes: > The previous patch was last week I think. This fixes a Valgrind error in the test. This is > correct as vtResult is just hte _expected_ return type, and this way we check the actual return type > (new this version) and then the validity of the pointer (same as old patch) before calling lstrcpyW > so we don't pass NULL to lstrcpyW. It seems to me what you really want to do is check that the call succeeded, instead of relying on the variant being cleared on error. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: [PATCH] svchost: Implementation of svchost (revised submission 2)
by Alexandre Julliard Nov. 28, 2007

Nov. 28, 2007

Roy Shea <roy(a)cs.hmc.edu> writes: > This is another revised and standalone version of the svchost patch. > Changes in this revision include: > > - Only using HEAP_ZERO_MEMORY in calls to HeapReAlloc. Calls to > HeapAlloc explicitly set the NULL terminator when required. The realloc is overkill too, you can just allocate one extra char from the start, and set it to zero unconditionally. -- Alexandre Julliard julliard(a)winehq.org

1 0