November 2010 - Wine-Devel - list.winehq.org

Re: wined3d: Set alpha channel to 0 for pixels with color key
by Henri Verbeet Nov. 16, 2010

Nov. 16, 2010

On 16 November 2010 12:46, Ilya Shpigor <shpigor(a)etersoft.ru> wrote: > Hello, > > This is patch for bug: > http://bugs.winehq.org/show_bug.cgi?id=23750 > > Current DirectDraw implementation doesn't process color key correctly. > I have add manually pixels color checking on surface loading because don't > able to find GL function to do this. > This patch is wrong. Color key processing happens mostly in d3dfmt_convert_surface(), you should try to figure out why it isn't happening for this application. Some more general points are that perform_color_key() would need to be static, casting void pointer is redundant, the color key is a range, not a single value, and try to use consistent formatting.

2 1

Re: [PATCH 5/5] shell32/tests: Added SHPropStg* tests
by Marvin Nov. 16, 2010

Nov. 16, 2010

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 http://testbot.winehq.org/JobDetails.pl?Key=7088 Your paranoid android. === WVISTAADM (32 bit misc) === misc: unhandled exception c0000005 at 767CA659 === W2K8SE (32 bit misc) === misc: unhandled exception c0000005 at 75A29E4D === W7PRO (32 bit misc) === misc: unhandled exception c0000005 at 76C59DAE === W7PROX64 (32 bit misc) === misc: unhandled exception c0000005 at 74D79DAE === W7PROX64 (64 bit misc) === Timeout

1 0

Developing on 64 bit Ubuntu? Add the Wine PPA for new ia32-libs that supports gstreamer codecs
by Scott Ritchie Nov. 15, 2010

Nov. 15, 2010

The ia32-libs that ships with Ubuntu doesn't contain 32 bit libraries for all the useful codecs of gstreamer. There's enough to build Wine, but not enough to actually show a video. I've fixed this in the ia32-libs on the Wine PPA by adding _73_ new libraries to it. This is something you'll want even if you're not using the wine packages and just doing Wine development. Thankfully, ia32-libs should be finally obsolete by 11.04 with multiarch, but in the meantime we can use the PPA. I'll work on porting the ia32-libs changes to 10.04 and 9.10, but it's less of a priority. On those we have other problems (such as old GCC versions causing problems). Thanks, Scott Ritchie

1 0

Re: [PATCH] console: fix output for console handles opened with GENERIC_WRITE access
by Martin Wilck Nov. 15, 2010

Nov. 15, 2010

Hi Eric, hi everyone, as you may already know, my patch is obsoleted by your commit 91e3a40e412a6c04cfaf8f6d0d3d16a1a4c4bfdb. Thanks! Martin -- Dr. Martin Wilck <Martin.Wilck(a)particle-dynamics.de> Physicist and Software Engineer particle dynamics GmbH, Leipzig Geschäftsführer (Director): Markus Fait Handelsregister (Trade Register): HRB 19578 AG Leipzig Richterstr. 1, 04105 Leipzig, Germany Tel. ++49-341-9849387 Fax. ++49-341-9849377 http://www.particle-dynamics.de

1 0

Re: [PATCH 1/4] msvcp90: Added basic_string<char>::swap implementation. (try 2)
by Piotr Caban Nov. 15, 2010

Nov. 15, 2010

On 11/15/10 18:26, Vincas Miliūnas wrote: > The way I see this is that the code above handles the case when lhs string is in the local buffer and rhs is allocated. The exchange requires copying (lhs->data->buf into rhs->data->buf and rhs->data->ptr into lhs->data->ptr), since setting one's ptr to other's buf would be illegal. Also, the ptr and buf are in an union construct and overlap, therefore only one can be stored at a time. You can copy whole union without knowing what field is used. This way you can implement it with only one else statement. + if (this == str) + { Please try to follow coding style used in modified file (no space after if, new block bracket in the same line).

1 0

how to use wcslen() in wine?
by 飞翔 Nov. 15, 2010

Nov. 15, 2010

hello, I'm developing under Wine environment. I'm using Unicode and I want to be able to use standard library call wcslen, but I don't know how to change my C library from the native glibc C library to the msvcrt C library. As it is described in "Winelib_User's_Guide":"In these cases you should use msvcrt to provide your C runtime calls. import msvcrt.dll to your applications .spec file." I'm using wineg++ to compile the .cpp & .h files, there is no .spec file, how can I import the DLL? May you please help me? Thanks very much! ------------------ Where there is a will, there is a way! ------------------ 丁耿佳 Gavin-Ding Computer Science and Technology， South China University of Technology Email: gj.ding(a)mail.scut.edu.cn gengjia.ding(a)gmail.com gengjia.ding(a)qq.com Mailing Address: 339, Building C12, SCUT, Guangzhou Higher Education Mega Center, Guangzhou, China Postcode:510006

2 1

Kernel APC in Wine?
by Joerg-Cyril.Hoehle＠t-systems.com Nov. 15, 2010

Nov. 15, 2010

Hi, my WINMM callback tests (cf. bug #3930, comments 37+38) showed that most MOM_ callbacks were executed in my own thread! The only way I know how to achieve this is to use an APC as in QueueUserAPC(). However, reading MSDN again, it seems like user-mode APC are never invoked unless the app uses the Wait*Ex() functions instead of the regular ones likes WaitForMultipleObjects() or Sleep(). As I can't expect a random app to use these, I assume that either - something confused me and I'm on the wrong track, or - native's winmm uses kernel APC, not user ones, and kernel APC are invoked even with plain Wait* calls, or - user APC are executed with plain Sleep/Wait*() (non ~Ex) too. Does Wine have something like kernel APC? What other mechanism exist that could allow a player loop inside WINMM to invoke callbacks in the application's thread context? BTW, I find it's a good idea from MS to execute callbacks in the caller's thread. This avoids common pitfalls and nasty bugs. Is there potential trouble with APC in Wine on 64 bit? MSDN recommends against posting an APC to a thread in another process. Thank you for your help, Jörg Höhle

1 0

Re: winecoreaudio: OSStatus is not necessarily a FOURCC.
by Michael Stefaniuc Nov. 15, 2010

Nov. 15, 2010

Hi! Joerg-Cyril.Hoehle(a)t-systems.com wrote: > This fixes the string from midi.ok mentioned in bug #21589 > err:wave:AudioUnit_GetVolume > AudioUnitGetParameter return an error "\xff\xff\xd5\x8e" > Now it will print a signed integer instead: > -10866 aka. kAudioUnitErr_InvalidScope > > As to whether that function should stay "inline", I don't know. Doesn't really matter as "inline" is just a hint to the compiler and the compiler is free to ignore it. bye michael

1 0

Re: [PATCH 1/4] msvcp90: Added basic_string<char>::swap implementation.
by Piotr Caban Nov. 14, 2010

Nov. 14, 2010

On 11/11/10 18:55, Vincas Miliūnas wrote: > This implementation handles cases when basic_strings are allocated using different allocators, however WINE's implementation seems to ignore allocators. It's not ignoring allocators (but it's not setting allocator field). There's no legal way of accessing it by end user, so there's no point of setting it, unless it's needed internally (and it's not needed for correct implementation of basic_string_char_swap function). +/* Helper function for MSVCP_basic_string_char_swap */ +void basic_string_char_swap_bufxptr_same_alloc(basic_string_char * lhs, basic_string_char * rhs) +{ + char * const ptr = rhs->data.ptr; + const size_t size = rhs->size; + const size_t res = rhs->res; + + memcpy(rhs->data.buf, lhs->data.buf, lhs->size * sizeof(char)); This function should work in constant time. It should only swap pointers and lengths. +/* Helper function for MSVCP_basic_string_char_swap */ +MSVCP_BOOL basic_string_char_swap_bufxptr_diff_alloc(basic_string_char * lhs, basic_string_char * rhs) +{ + char tmp[BUF_SIZE_CHAR]; + const size_t size = lhs->size; + const size_t res = lhs->res; + memcpy(tmp, lhs->data.buf, lhs->size * sizeof(char)); + + char * const ptr = MSVCP_allocator_char_allocate(lhs->allocator, rhs->res); You can't pass unknown data to MSVCP_allocator_char_allocate, this is the implementation of default char allocator. When other allocators are in use compiler will use templates. The function you're implementing is only used when two basic_string_char are swapped, because there's no legal way of modifying allocator field there's only one case - the same allocators are used. Cheers, Piotr

1 0

Re: [PATCH 2/2] ntdll: Check for case-insensitive volumes. (try 6
by James McKenzie Nov. 14, 2010

Nov. 14, 2010

On 11/5/10 8:17 AM, Charles Davis wrote: > Changes since try 5: > o All case-sensitivity logic has been absorbed into find_file_in_dir(). > o Eliminated one more stat(2) that I missed last time. > o Eliminated a statfs(2) call on Linux. > > This is a far as I think I can take it without gutting it completely. If > any of you have better ideas, I'd like to hear them. > --- > dlls/ntdll/directory.c | 233 +++++++++++++++++++++++++++++++++++++++++++++++- > 1 files changed, 229 insertions(+), 4 deletions(-) > > diff --git a/dlls/ntdll/directory.c b/dlls/ntdll/directory.c > index 865c2fa..2891792 100644 > --- a/dlls/ntdll/directory.c > +++ b/dlls/ntdll/directory.c > @@ -41,6 +41,9 @@ > #ifdef HAVE_SYS_STAT_H > # include<sys/stat.h> > #endif > +#ifdef HAVE_SYS_ATTR_H > +#include<sys/attr.h> > +#endif > #ifdef HAVE_SYS_IOCTL_H > #include<sys/ioctl.h> > #endif > @@ -56,6 +59,9 @@ > #ifdef HAVE_SYS_MOUNT_H > #include<sys/mount.h> > #endif > +#ifdef HAVE_SYS_STATFS_H > +#include<sys/statfs.h> > +#endif > #include<time.h> > #ifdef HAVE_UNISTD_H > # include<unistd.h> > @@ -130,6 +136,35 @@ static inline int getdents64( int fd, char *de, unsigned int size ) > > #endif /* linux */ > > +#if defined(HAVE_GETATTRLIST) > + > +struct get_fsid > +{ > + ULONG size; > + dev_t dev; > + fsid_t fsid; > +}; > + > +struct fs_cache > +{ > + dev_t dev; > + fsid_t fsid; > + BOOLEAN case_sensitive; > +} fs_cache[64]; > + > +#if defined(ATTR_VOL_CAPABILITIES)&& defined(VOL_CAPABILITIES_FORMAT)&& \ > + defined(VOL_CAP_FMT_CASE_SENSITIVE) > + > +struct vol_caps > +{ > + ULONG size; > + vol_capabilities_attr_t caps; > +}; > + > +#endif /* ATTR_VOL_CAPABILITIES */ > + > +#endif /* HAVE_GETATTRLIST */ > + > #define IS_OPTION_TRUE(ch) ((ch) == 'y' || (ch) == 'Y' || (ch) == 't' || (ch) == 'T' || (ch) == '1') > #define IS_SEPARATOR(ch) ((ch) == '\\' || (ch) == '/') > > @@ -783,6 +818,192 @@ static char *get_device_mount_point( dev_t dev ) > } > > > + > +#ifdef HAVE_GETATTRLIST > +/*********************************************************************** > + * look_up_fs_cache > + * > + * Checks if the specified file system is in the cache. > + */ > +static struct fs_cache *look_up_fs_cache( dev_t dev ) > +{ > + int i; > + for (i = 0; i< sizeof(fs_cache)/sizeof(fs_cache[0]); i++) > + if (fs_cache[i].dev == dev) > + return fs_cache+i; > + return NULL; > +} > + > +/*********************************************************************** > + * add_fs_cache > + * > + * Adds the specified file system to the cache. > + */ > +static void add_fs_cache( dev_t dev, fsid_t fsid, BOOLEAN case_sensitive ) > +{ > + int i; > + struct fs_cache *entry = look_up_fs_cache( dev ); > + static int once = 0; > + if (entry) > + { > + /* Update the cache */ > + entry->fsid = fsid; > + entry->case_sensitive = case_sensitive; > + return; > + } > + > + /* Add a new entry */ > + for (i = 0; i< sizeof(fs_cache)/sizeof(fs_cache[0]); i++) > + if (fs_cache[i].dev == 0) > + { > + /* This entry is empty, use it */ > + fs_cache[i].dev = dev; > + fs_cache[i].fsid = fsid; > + fs_cache[i].case_sensitive = case_sensitive; > + return; > + } > + > + /* Cache is out of space, warn */ > + if (once++) > + WARN( "FS cache is out of space, expect performance problems\n" ); > +} > +#endif > + > +/*********************************************************************** > + * get_dir_case_sensitivity > + * > + * Checks if the volume containing the specified directory is case > + * sensitive or not. > + */ > +static BOOLEAN get_dir_case_sensitivity( const char *dir ) > +{ > +#if defined(__APPLE__) || defined(__FreeBSD__) || defined(__FreeBSD_kernel__) > + struct statfs stfs; > +#elif defined(__NetBSD__) > + struct statvfs stfs; > +#elif defined(__linux__) > + struct stat st; > + char *cifile; > +#endif > + > +#if defined(HAVE_GETATTRLIST)&& defined(ATTR_VOL_CAPABILITIES)&& \ > + defined(VOL_CAPABILITIES_FORMAT)&& defined(VOL_CAP_FMT_CASE_SENSITIVE) > + char *mntpoint = NULL; > + struct attrlist attr; > + struct vol_caps caps; > + struct get_fsid get_fsid; > + struct fs_cache *entry; > + > + /* First get the FS ID of the volume */ > + attr.bitmapcount = ATTR_BIT_MAP_COUNT; > + attr.reserved = 0; > + attr.commonattr = ATTR_CMN_DEVID|ATTR_CMN_FSID; > + attr.volattr = attr.dirattr = attr.fileattr = attr.forkattr = 0; > + get_fsid.size = 0; > + if (getattrlist( dir,&attr,&get_fsid, sizeof(get_fsid), 0 ) != 0 || > + get_fsid.size != sizeof(get_fsid)) > + return TRUE; > + /* Try to look it up in the cache */ > + entry = look_up_fs_cache( get_fsid.dev ); > + if (entry&& !memcmp(&entry->fsid,&get_fsid.fsid, sizeof(fsid_t) )) > + /* Cache lookup succeeded */ > + return entry->case_sensitive; > + /* Cache is stale at this point, we have to update it */ > + > + mntpoint = get_device_mount_point( get_fsid.dev ); > + /* Now look up the case-sensitivity */ > + attr.commonattr = 0; > + attr.volattr = ATTR_VOL_INFO|ATTR_VOL_CAPABILITIES; > + if (getattrlist( mntpoint,&attr,&caps, sizeof(caps), 0 ) == 0) > + { > + if (caps.size == sizeof(caps)&& > + (caps.caps.valid[VOL_CAPABILITIES_FORMAT]& > + (VOL_CAP_FMT_CASE_SENSITIVE | VOL_CAP_FMT_CASE_PRESERVING)) == > + (VOL_CAP_FMT_CASE_SENSITIVE | VOL_CAP_FMT_CASE_PRESERVING)) > + { > + BOOLEAN ret; > + > + RtlFreeHeap( GetProcessHeap(), 0, mntpoint ); > + if ((caps.caps.capabilities[VOL_CAPABILITIES_FORMAT]& > + VOL_CAP_FMT_CASE_SENSITIVE) != VOL_CAP_FMT_CASE_SENSITIVE) > + ret = FALSE; > + ret = TRUE; > + /* Update the cache */ > + add_fs_cache( get_fsid.dev, get_fsid.fsid, ret ); > + return ret; > + } > + } > + RtlFreeHeap( GetProcessHeap(), 0, mntpoint ); > + /* Fall through */ > +#endif > +#if defined(__APPLE__) || defined(__FreeBSD__) || defined(__FreeBSD_kernel__) > + statfs( dir,&stfs ); > + /* Assume these file systems are always case insensitive on Mac OS. Don't do this. Quake2 and quake2 turn out to be different (and I tried to move to a directory called Quake2 using cd $HOME/wine.../drive_c/quake2 and received an error this morning [Yes, I'm playing Quake2 on my Mac, I really, really need the stress relief, it is really hard to explain.}) > + * For FreeBSD, only assume CIOPFS is case insensitive (AFAIK, Mac OS > + * is the only UNIX that supports case-insensitive lookup). > + */

2 1