June 2014 - Wine-Devel - list.winehq.org

Re: kernel32: Raise DBG_PRINTEXCEPTION_C exception in OutputDebugStringA.

by Marvin

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at https://testbot.winehq.org/JobDetails.pl?Key=7510 Your paranoid android. === w2000pro (32 bit exception) === exception.c:406: Test failed: 42: Wrong exception address 00950000/00950001 exception: unhandled exception 80000004 at 00950000 exception.c:406: Test failed: 42: Wrong exception address 7C5C20B0/00950001 === wxppro (32 bit exception) === exception.c:406: Test failed: 42: Wrong exception address 00A30000/00A30001 exception: unhandled exception 80000004 at 00A30000 exception.c:406: Test failed: 42: Wrong exception address 7C80DF3C/00A30001 === wvistau64 (32 bit exception) === exception.c:406: Test failed: 42: Wrong exception address 00340000/00340001 exception: unhandled exception 80000004 at 00340000 exception.c:406: Test failed: 42: Wrong exception address 7751ED97/00340001 === w2008s64 (32 bit exception) === exception.c:406: Test failed: 42: Wrong exception address 003C0000/003C0001 exception: unhandled exception 80000004 at 003C0000 exception.c:406: Test failed: 42: Wrong exception address 77E5ED97/003C0001 === w7u (32 bit exception) === exception.c:406: Test failed: 42: Wrong exception address 003F0000/003F0001 exception: unhandled exception 80000004 at 003F0000 exception.c:406: Test failed: 42: Wrong exception address 77CDE34F/003F0001 === w8 (32 bit exception) === exception.c:406: Test failed: 42: Wrong exception address 004F0000/004F0001 exception: unhandled exception 80000004 at 004F0000 exception.c:406: Test failed: 42: Wrong exception address 77174D17/004F0001

June 15, 2014

Re: windowscodecs/tests: Fixed some leaks on error path (Valgrind)

by Vincent Povirk

+ if (FAILED(hr)) + { + IStream_Release(stream); + IWICComponentFactory_Release(factory); + return; + } The rest of the function is already has code to handle failure, so returning early is unnecessary.

June 15, 2014

Re: shell32: Fix SHFileOperation when deleting a nonexistent directory. (try 2)

by Zhenbo Li

2014-05-21 9:23 GMT+08:00 Zhenbo Li <litimetal(a)gmail.com>: > Thanks Alexandre > > try 2: > 1. Let SHELL_DeleteDirectoryW return DWORD > 2. Get the right return value by > ret = SHNotifyRemoveDirectoryW(pszDir) > origin: > Thanks Bruno Jesus > This patch would fix bug 34324, by returning correct > error value of SHFileOperation. > > --- > dlls/shell32/shlfileop.c | 55 > +++++++++++++++++++++--------------------- > dlls/shell32/tests/shlfileop.c | 4 --- > 2 files changed, 27 insertions(+), 32 deletions(-) Hi, Could someone have a look with my patch after three weeks :-) Thank you very much -- Have a nice day! Zhenbo Li

June 14, 2014

Re: ws2_32/socket.c: Fix compile for linux systems without net/ipx.h

by Felix Janda

Bruno Jesus wrote: > On Fri, Jun 13, 2014 at 2:14 PM, Felix Janda <felix.janda(a)posteo.de> wrote: > > ... > > Not defining HAS_IPX would also fix the compilation but I don't see > > why one has to disable IPX because of 2 ipx specific constants not > > declared in linux/ipx.h. (The 2 constants are also declared in an > > kernel header not exported to userspace.) > > > > I'm not sure what's the best solution. I guess that one of the main > > remaining uses of IPX in linux is to play old network games in wine. > > Other possible solutions are for musl to include a netipx/ipx.h > > header (unlikely, the headers are tried to be kept as clean as > > possible) or declare the constants in sys/socket.h (maybe) or for > > the kernel to include the constants in linux/ipx.h. > > If the system does not #define the required variables to support IPX > then I don't think it's right to manually define then and assume they > will work. I'll download it and give it a try later. The constants are only used for the [sg]etsockopt() system calls. They are hardcoded in the linux kernel. As stated above I'm open to alternative solutions than the patch. Thanks for trying to reproduce the problem. Felix

June 13, 2014

Re: ws2_32/socket.c: Fix compile for linux systems without net/ipx.h

by Bruno Jesus

On Fri, Jun 13, 2014 at 2:14 PM, Felix Janda <felix.janda(a)posteo.de> wrote: > ... > Not defining HAS_IPX would also fix the compilation but I don't see > why one has to disable IPX because of 2 ipx specific constants not > declared in linux/ipx.h. (The 2 constants are also declared in an > kernel header not exported to userspace.) > > I'm not sure what's the best solution. I guess that one of the main > remaining uses of IPX in linux is to play old network games in wine. > Other possible solutions are for musl to include a netipx/ipx.h > header (unlikely, the headers are tried to be kept as clean as > possible) or declare the constants in sys/socket.h (maybe) or for > the kernel to include the constants in linux/ipx.h. If the system does not #define the required variables to support IPX then I don't think it's right to manually define then and assume they will work. I'll download it and give it a try later. Bruno

June 13, 2014

Re: ws2_32/socket.c: Fix compile for linux systems without net/ipx.h

by Felix Janda

Bruno Jesus wrote: > On Mon, Jun 2, 2014 at 3:01 PM, Felix Janda <felix.janda(a)posteo.de> wrote: > > --- > > dlls/ws2_32/socket.c | 2 ++ > > 1 file changed, 2 insertions(+) > > > > diff --git a/dlls/ws2_32/socket.c b/dlls/ws2_32/socket.c > > index e31f396..cf5473d 100644 > > --- a/dlls/ws2_32/socket.c > > +++ b/dlls/ws2_32/socket.c > > @@ -109,6 +109,8 @@ > > # include <linux/types.h> > > # endif > > # include <linux/ipx.h> > > +# define SOL_IPX 256 > > +# define IPX_TYPE 1 > > # define HAS_IPX > > #endif > > Hi, can you share the details about how you are compiling this? There > may be a better solution like not defining HAS_IPX if SOL_IPX is not > present. If you help me reproduce the issue I can help to fix it. Thanks for your review. All my patches stem from issues when trying to compile wine on a musl libc[1] based system. The wiki[2] contains information on how to get a musl based distro but in order to reproduce the issue it should be enough to rename the netipx/ipx.h header file. The situation is that there is no netipx/ipx.h in musls headers but if linux headers are installed then there will be linux/ipx.h. So except for the definitions of SOL_IPX and IPX_TYPE everything is declared in order to make dlls/ws2_32/socket.c compile with HAS_IPX. Without the patch HAS_IPX will be declared anyway since linux/ipx.h exists. Then there are passages in socket.c behind #ifdef HAS_IPX where #ifdef SOL_IPX is used in order to determine whether to take linux or BSD specific code. musl pulls in the BSD specific code and so the compilation will fail. Not defining HAS_IPX would also fix the compilation but I don't see why one has to disable IPX because of 2 ipx specific constants not declared in linux/ipx.h. (The 2 constants are also declared in an kernel header not exported to userspace.) I'm not sure what's the best solution. I guess that one of the main remaining uses of IPX in linux is to play old network games in wine. Other possible solutions are for musl to include a netipx/ipx.h header (unlikely, the headers are tried to be kept as clean as possible) or declare the constants in sys/socket.h (maybe) or for the kernel to include the constants in linux/ipx.h. Felix [1]: http://musl-libc.org [2]: http://wiki.musl-libc.org

June 13, 2014

Re: [1/4] msvcrt/tests: Add a test for fwrite flushing behavior. (try 2)

by Piotr Caban

On 06/13/14 18:11, Grazvydas Ignotas wrote: > On Fri, Jun 13, 2014 at 12:00 PM, Piotr Caban <piotr.caban(a)gmail.com> wrote: >> Hi, >> >> >> On 06/13/14 01:27, Grazvydas Ignotas wrote: >>> >>> +static void test_write_flush(void) >>> +{ >>> + char iobuf[1024]; >>> + char *tempf; >>> + FILE *file; >>> + >>> + tempf = _tempnam(".","wne"); >>> + file = fopen(tempf, "wb+"); >>> + ok(file != NULL, "unable to create test file\n"); >>> + /* the default should be 4K */ >>> + test_write_flush_size(file, 4096); >> >> It would be nice to read the size from the file->_bufsiz instead of >> hard-coding 4096. This value is not initialized while the file is opened so >> you will need to call something before reading it's value. > > One of the goals of this test was to test if the default buffer size > is really 4K, after reading file->_bufsiz that will no longer be > tested. Also no test currently is accessing FILE members, so I'd > prefer to keep it as-is. I don't see any problem with accessing FILE members in tests. If you want to make sure it's set to 4096 just add following test: ok(file->_bufsiz == 4096, ...); You are generally testing behavior that depends on _bufsiz so I think it's cleaner to pass it there. >>> + unlink(tempf); >>> + free(tempf); >> >> _tempnam returns internal buffer, free should not be called on it. > > Are you sure? From what I see wine does MSVCRT__strdup(tmpbuf), unix > manpage also says it should be freed, many tests are already freeing > it. You're right, this buffer needs to be freed, sorry.

June 13, 2014

Re: [1/4] msvcrt/tests: Add a test for fwrite flushing behavior. (try 2)

by Grazvydas Ignotas

On Fri, Jun 13, 2014 at 12:00 PM, Piotr Caban <piotr.caban(a)gmail.com> wrote: > Hi, > > > On 06/13/14 01:27, Grazvydas Ignotas wrote: >> >> +static void test_write_flush(void) >> +{ >> + char iobuf[1024]; >> + char *tempf; >> + FILE *file; >> + >> + tempf = _tempnam(".","wne"); >> + file = fopen(tempf, "wb+"); >> + ok(file != NULL, "unable to create test file\n"); >> + /* the default should be 4K */ >> + test_write_flush_size(file, 4096); > > It would be nice to read the size from the file->_bufsiz instead of > hard-coding 4096. This value is not initialized while the file is opened so > you will need to call something before reading it's value. One of the goals of this test was to test if the default buffer size is really 4K, after reading file->_bufsiz that will no longer be tested. Also no test currently is accessing FILE members, so I'd prefer to keep it as-is. > >> + setvbuf(file, iobuf, _IOFBF, sizeof(iobuf)); >> + test_write_flush_size(file, sizeof(iobuf)); >> + fclose(file); > > Current implemententation of setvbuf is broken in wine. This code will cause > free(iobuf) to be called. I think that setvbuf needs to be fixed before this > test can be added. ok > >> + unlink(tempf); >> + free(tempf); > > _tempnam returns internal buffer, free should not be called on it. Are you sure? From what I see wine does MSVCRT__strdup(tmpbuf), unix manpage also says it should be freed, many tests are already freeing it. -- Gražvydas

June 13, 2014

Re: vbscript: Reset error number on error mode change.

by Marvin

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at https://testbot.winehq.org/JobDetails.pl?Key=7476 Your paranoid android. === w2000pro (32 bit) === No test summary line found === wxppro (32 bit) === No test summary line found === wvistau64 (32 bit) === No test summary line found === w2008s64 (32 bit) === No test summary line found === w7u (32 bit) === No test summary line found === w8 (32 bit) === No test summary line found === wvistau64 (64 bit) === No test summary line found === w2008s64 (64 bit) === No test summary line found

June 13, 2014

Re: [PATCH 1/2] riched20: Implement ITextSelection::GetText.

by Jactry Zeng

Hello, Could someone take a look at this patch? Any comment will be appreciate! Thanks. 2014-05-28 15:17 GMT+08:00 Jactry Zeng <wine(a)jactry.com>: > ChangeLog: > - skip a test in native 64bit riched20 > - implementing without helper function > > --- > dlls/riched20/richole.c | 28 +++++++++++++- > dlls/riched20/tests/richole.c | 88 ++++++++++++++++++++++++++++++ > ++++++++++++- > 2 files changed, 113 insertions(+), 3 deletions(-) > > > > > > -- Regards, Jactry Zeng

June 13, 2014