It isn't possible to access memory between changing the stack and setting the cr3 register.
A page fault needs a valid kmode stack which is possible not available.
svn path=/trunk/; revision=15129
them new are STRING_CMD_INVALIDDRIVE, STRING_CMD_SHELLINFO, STRING_VERSION_RUNVER
remove some #define strings that was not in use.
update ver so it output svn version number and update copyright note
it take the copyright and buildversion from <reactos/resource.h> and <reactos/buildno.h>
so we only need change it on one place
svn path=/trunk/; revision=15123
update reactos specify patch by Magnus Olsen
markout better what is reactos specify code
but not all code are not mark
svn path=/trunk/; revision=15119
- The status results are always set before an event is signaled.
- Some checks for ASYNC irp's were wrong (missing brackets, now replaced with SyncIrp).
- Don't signal the FO event if it is the same as the UserEvent.
- Set the IOSB on error for async irp's with a sync FO (fix bug #609).
IMHO, the old schematics is wrong, because on error it is possible that an event
isn't signaled if the driver has previous returned STATUS_PENDING.
In this case, the caller will wait forever.
svn path=/trunk/; revision=15117
<Fireball> Imported WINE's RtlUnicodeStringToInteger and RtlCharToInteger - they were seriously broken in our revision. Thanks to GreatLord for his help pointing out this bug with me :)
svn path=/trunk/; revision=15092
it take less memmory now, and it is bit faster
CMD_ModuleHandle
contain GetModuleHandle and it is only execute in main() in cmd
it optimze for lite more speed in cmd
Langues changes
in En.rc follow new strings have been add and can be translate
STRING_ERROR_WRONG_DATE
STRING_SET_ERROR
svn path=/trunk/; revision=15081
60 are the dec value but it have not be convort to hex that is 0x3c
legmet value are, e.g., 10, 20, 40, 60, 80, 100, 200.'
in dec in hex 0x0a, 0x14, 0x28, 0x3c, 0x50, 0x64, 0xc8
svn path=/trunk/; revision=15060
Robert Reif <[email protected]>
- Stub out DRVM_MAPPER_RECONFIGURE support.
- Correctly handle where waveOutGetPosition changes timepos.wType
because the requested type is not supported.
- Added Jeremy White's waveOutGetPosition fix to waveInGetPosition.
- Fix memory leak in error path.
- Provide default implementation of waveInPrepareHeader and
waveInUnprepareHeader if driver doesn't support them.
Jose Manuel Ferrer Ortiz <[email protected]>
- Spanish translations updated.
Peter Berg Larsen <[email protected]>
- Assorted memleak fixes. Found on Michael Stefaniuc smatch list.
Jeremy White <[email protected]>
- Do not fallback to defaults if a driver, mapper, or midi is specified
in the registry; consolidate MMDRV_Init() into a single function.
Robert Reif <[email protected]>
- Fix memory leak when there are too many drivers.
Vincent Beron <[email protected]>
- Correct and complete some api documentation.
Filip Navara <[email protected]>
- Specify correct buffer size in GetPrivateProfileStringW calls.
Jakob Eriksson <[email protected]>
- Get rid of HeapAlloc casts.
Jason Edmeades <[email protected]>
- Avoid trap in mixerGetLineControlsA when cControls is uninitialized
and MIXER_GETLINECONTROLSSF_ONEBYTYPE requested.
svn path=/trunk/; revision=15056
- prevent completion port notification by not passing an APC context to the kernel when the low-order bit of the event handle is set (for asynchronous operations)
svn path=/trunk/; revision=15051
Hartmut Birr