- Flusher lock fixes: wrong kind of lock,total mess (and the wrong kind of lock). Properly fixed throughout cmapi.c, but still missing in many other places.
- Add support for detecting loading of an already loaded hive.
- Start adding calls to CmpReportNotify to support registry callbacks.
- Do work needed to flush notifications for a deleted node (but CmpFlushNotify not yet implemented).
- Add support for adding each newly loaded hive to the HiveList key in the registry (but CmpAddHiveToFileList not yet implemented).
- Add some ViewLock acquire/releases where needed.
- Load the key in a faster way (Ob vs Zw)
- Add checks everywhere for HvMarkCellDirty success. In future (when log/backup file is enabled), it can return FALSE (e.g. when we are out of space).
- Change logic in CmpDoFlushAll to only flush a hive if it won't shrink (in the future, flushing may lead to hive shrinkage for efficiency).
- Add SEH2 protection to all CmApis that may deal with user-mode data.
- Add HvTrackCellRef/HvReleaseCellRef for tracking cell references in scenarios where we might need many GetCell/ReleaseCell calls. For now stubbed to only work with up to 4 static references.
- Properly unlock/free in some failure paths in some of the CM APIs.
- Add some missing HvReleaseCell in paths where it was missing.
- Try to fix hack in enumerate key.
- Fix wrong usage of KeQuerySystemTime. It was called twice to save it in 2 different places. Instead, there should be only one call, and then duplicate the value across.
- Fix logic in CmpSetValueExistingData/Key.
Tested with winetests and .NET framework 1.1 installation which fully completes.
svn path=/trunk/; revision=46702
44467, 44468, 44470, 44471, 46524
[OLDNAMES]
Update moldname-msvcrt.def
Add a proper oldnames lib, instead of implicitly including it with mingw.
46526
[FTP/DHCP]
- remove the oldnames hack, as we now have a proper oldnames lib (included in mingw_common)
- protect struct timezone from redefinition
svn path=/trunk/; revision=46592
46483 (tkreuzer)
[CRT]
- add _wctype table, it's not identical to _ctype as assumed before
- point _pwctype to _wctype instead of _ctype
- implement __pctype_func and __pwctype_func
46502 (tkreuzer)
[MSVCRT]
- enable exports for __pctype_func, __pwctype_func, _wctype
svn path=/trunk/; revision=46504
- add _wctype table, it's not identical to _ctype as assumed before
- point _pwctype to _wctype instead of _ctype
- implement __pctype_func and __pwctype_func
svn path=/branches/ros-amd64-bringup/; revision=46483
- define _CRTBLD
- rename float_t/double_t/long_double_t to float_s/double_s/long_double_s to avoid name conflicts
svn path=/branches/ros-amd64-bringup/; revision=46478
Since we have switched to non-prefixed symbols, we need to deal with 2 kinds of symbols that the compiler doesn't handle for us. 1.) __image_base__ is still defined the same way by ld, but since gcc doesn't give us an underscore anymore, we need to specify it on our own. 2.) __MINGW_IMP_SYMBOL, which is a hacky way of declaring a symbol dllimport, by prefixing it with __imp_ also needs special treatment. Fix these issues by adding appropriate macros to _mingw.h and removing the broken and duplicate definitions from multiple .c files.
svn path=/branches/ros-amd64-bringup/; revision=46460
Import x87 asm version of pow() from glibc. Replaces our broken implementation, fixing a bunch of winetests (oleaut32 vartest)
svn path=/trunk/; revision=46283
When adding new glyph cache entry convert the glyph bitmap with proper alignment to comply with the new code in EngCreateBitmap and remove the corresponding hack from SURFMEM_bCreateDib.
[FREETYPE]
When doing 1BPP -> 8BPP conversion set all 8 bits not only the LSB. (Freetype devs were notified of this issue)
See issue #5244 for more details.
svn path=/trunk/; revision=46246
Reintegrate header-work branch. Thanks to hours of boring work on the ddk headers (Most of it done by Amine Khaldi), we have now a decent set, which is still incomplete but already relatively close to latest WDK. A clean rebuild is recommended.
svn path=/trunk/; revision=46155
- Fix Samuel's mistake of assuming that xmlstrs are zero-terminated by adding a xmlstr2unicode function, which returns a UNICODE_STRING representation of xmlstr suitable for printing. No debug log garbage now (disadvantage: more code changes compared to the original code).
svn path=/trunk/; revision=45845
- Whilst this is non-standard and potentially makes the code less portable, it's supported by both gcc and msvc and should bringg increased compilation speed due to its optimized behaviour (doesn't need to invoke the preprocessor)
- Patch by Amine Khaldi [[email protected]]
svn path=/trunk/; revision=45685
- Replace mixer / wave support code by mmixer library
[MMIXER]
- Compare device name from 2 character onwards as there are difference in paths in user mode / kernel mode
- Invoke create pin callback if the caller provides it
[WDMAUD.DRV]
- Apply changes from mmixer library
-- Tested in Vbox with Winamp / ReactOS mplay32
svn path=/trunk/; revision=45579
- Set the minimum GCC version to 4.4.0 and minimum Binutils version to 20091016.
- Remove the -fno-unit-at-a-time hacks.
- Remove dgorbachev's temporary hacks for user-mode C++ applications.
- Merge ntoskrnl's "precomp.h" and "ntoskrnl.h" again, the PCH issues have been fixed.
- Use our "ofmt_stub.a" to be compatible with newer MinGW Runtime versions.
- Use LD's new -exclude-all-symbols option to not auto-export all symbols in case no DEF file or __declspec(dllexport) was specified.
We used to patch our LD version to work around this problem.
- Use LD's --start-group and --end-group parameters to allow a more flexible linking order.
- Use GCC's -fno-set-stack-executable for amd64 and i386 targets to disable adding the __enable_execute_stack code.
- Change the makefile's name to "makefile-$(ARCH).auto" for all architectures.
- Remove some -Wno-error flags.
See issue #4810 for more details.
svn path=/trunk/; revision=45535
Gregor Schneider