- Partly implement HidP_TranslateUsageAndPagesToI8042ScanCodes
- Needs key modifier state function to work correctly
- KBDHID should now receive translated keys (without key modifiers)
svn path=/branches/usb-bringup/; revision=54858
- Fix few leaks and possible buffer overflows in New menu code
- Add icons of folder and shortcut to New menu
- Display message if file can't be created
svn path=/trunk/; revision=54848
- Rewrite Open With menu and dialog
- Load all applications
- Use treeview with Recommended and Other programs
- Don't allow duplicates
- Load registry keys in proper way
svn path=/trunk/; revision=54844
- Implement quering compatible ids
- Fix result status for quering instance id
- Fix bug in msmouse.inf
svn path=/branches/usb-bringup/; revision=54843
- Add new style SAL definitions (only dummys for now)
- Move old style SAL definitions to sal_old.h
- Convert annotations in sspi.h to new style
- Allow new style annotations for C++ again, old style annotations are not possible with C++ and gcc headers
svn path=/trunk/; revision=54839
- Add a ReactOS hack to let the hid stack initialize
- Add input.inf
- HID Mouse initializes and starts, now win32k needs some work to open the mice
svn path=/branches/usb-bringup/; revision=54838
- Fix creating instance id
- Print debug message
- Sync msmouse with head and add hid installation
- Needs input.inf written
- HID stack starts to initializes and then fails the to enable the device interface (/drivers/hid/hidclass/pdo.c:511), Logs show that IoGetDeviceObjectPointer() fails with STATUS_UNSUCCESSFUL
svn path=/branches/usb-bringup/; revision=54837
- Create a pesudo random instance id
- Fixes assert when plugin usb devices whose device descriptor don't provide a serial number
svn path=/branches/usb-bringup/; revision=54836
- Implement a HID parsing library, which is based on Haiku`s exellent HID Stack
- Implement HidP_FreeCollection, HidP_GetCaps, HidP_GetCollectionDescription, HidP_MaxUsageListLength, HidP_GetSpecificValueCaps, HidP_GetUsages, HidP_GetScaledUsageValue
- hidparse driver is now implemented enough to support a hid mouse
- Tested in VBox 4.1.4 + WinXP + ReactOS usbohci+ mouhid+ hidclass+ hidparse+hidusb
svn path=/branches/usb-bringup/; revision=54834
Complete RtlpDosPathNameToRelativeNtPathName_Ustr() rewrite by properly handling relative names.
This fixes several regressions brought by r54804 (see winetests).
This also fixes file system corruption when using directory functions.
This may also fix general file system corruption, so, please retest bug #6720.
svn path=/trunk/; revision=54820
- add module module type, which is basically a DLL without entry point and use it for typelibs
- get rid of custom targets for generating .tlb files and use source file dependencies instead
svn path=/trunk/; revision=54813
Comment out two series of tests untill a proper fix is available. Those lead to memory misuse in shell32.
This has been revealed by r54804.
Should fix both test bots.
svn path=/trunk/; revision=54811
- Create PDO for each top level collection found
- Use the specified collection index instead of the first
- add collection id to hardware id when the device has more than one top level collection, see http://msdn.microsoft.com/en-us/windows/hardware/gg487473 for details
svn path=/branches/usb-bringup/; revision=54808
- Fake status success for unimplemented IRP_MN_REMOVE_DEVICE
- Fix tons of bugs in the read report completion routine
- Implement function to re-use a complete report irp
- Use correct device object when invoking the mini driver's dispatch routine in the read function
- silence a few traces
- Mouse now works with with ReactOS USB stack + ReactOS HID stack (TBD: implement hidparse.sys)
svn path=/branches/usb-bringup/; revision=54806
Reimplement CreateDirectoryA, CreateDirectoryExA, CreateDirectoryW, CreateDirectoryExW, RemoveDirectoryA, RemoveDirectoryW.
This reimplementation now matches the w2k3 one, properly handles reparse points (especially mount points) and also brings several fixes.
svn path=/trunk/; revision=54804
- Partly implement IRP_MN_START_DEVICE for PDO
- PDO now completely initializes
- Remaining IOCTL to implement IRP_MJ_CREATE, IRP_MJ_CLOSE, IRP_MJ_FLUSH, IRP_MJ_READ for partial mouse support in hidclass
svn path=/branches/usb-bringup/; revision=54792
CreateDirectory: handle the case when the FS doesn't support EAs.
This should fix VBox and is mandatory if we want to support MS FSD with FAT32
svn path=/trunk/; revision=54789
- Remove an unneeded (and prone to race) RtlZeroMemory (found by Thomas Faber)
- Fix misuse of PAGE_MASK macro and broken PTE flag comparisons
svn path=/trunk/; revision=54782
- Implement AddDevice routine for hidusb
- Fix bug in Hid_DispatchUrb, which returned not initialized status code
- Implement IOCTL_HID_GET_REPORT_DESCRIPTOR, IOCTL_HID_READ_REPORT
- Add support routines for reseting / aborting pipe, not yet used
- TBD: implement error handling
- hidusb now works in combination with a usb mouse (other hid devices such as keyboards will require more ioctl implemented)
- Tested in Vbox 4.1.4 + WinXP + ReactOS usbhid.sys + ReactOS usbmou.sys
- Testing procedure -> inject usb mouse after startup (needs mouse integration disabled -> thanks to timo for pointing this out)
svn path=/branches/usb-bringup/; revision=54781
- Silence a few traces
- Add a hack for handling URB_FUNCTION_CLASS_INTERFACE. The hid bus driver performs this request with zero buffer length, which is not valid. Needs more investigation
svn path=/branches/usb-bringup/; revision=54777
- Usage page of zero is not used, increment, the MouHid_ButtonUpFlags / MouHid_ButtonDownFlags array
- Mouse clicks now work
- Implement flushing of the hid report queue
- Driver now works in VBox 4.1.4 + WinXP
- Secondary mouse pointer does not move in VBox (either a Vbox bug / WinXP), tested with ms driver and the same results were reveiled
svn path=/branches/usb-bringup/; revision=54776
- Fix up / down button detection
- Fix move detection. Mouse changes are now detected
- Mouse does not yet work in Vbox, as mouclass driver is for unknown reasons not connecting to the mouhid, needs to be investigated
- Tested in VBox 4.1.4 + ReactOS mouhid.sys
svn path=/branches/usb-bringup/; revision=54775
- Fix a few bugs & race condition in the read report routine
- mouhid initializes and is able to read input reports
- button press / wheel state change is detected
- mouse move detection not yet working
svn path=/branches/usb-bringup/; revision=54774
- Tweak the balancer to prevent an excessive buildup of user pages while cache is paged out completely each run
- Bugcheck if we've trimmed everything as much as possibly but we are still in need of pages
svn path=/trunk/; revision=54771
- Implement mouse button change detection
- Implement support routine for dispatching mouse input data
- Allocate mdl for input report, not yet used
svn path=/branches/usb-bringup/; revision=54767
- Don't flush the TLB in MmEnableVirtualMapping and MmCreateVirtualMapping because the PTE is already invalid so it can't possibly be in the TLB per x86 spec
- Added explanations of why flushes are needed or not needed in certain places
- A few small code improvements
- Should speed up ROS a bit by avoiding unnecessary TLB entry flushing
svn path=/trunk/; revision=54766
- Replace the log file critical section by a resource and use shared locks for read actions and exclusive locks for write actions.
- Add stubs for clear and backup actions.
svn path=/trunk/; revision=54763
- Reference the cache segment when flushing to avoid a race between deleting the file cache and the balancer flushing pages to disk which causes random cache segments to be freed while we're holding a lock on it
- Fixes a crash in KeReleaseMutex (used to be a crash in ExReleasePushLock) during periods of high memory load
svn path=/trunk/; revision=54745
- Samuel Serapion: Implement RtlIpv4AddressToStringEx A/W and minor enhancements RtlIpv4AddressToStringA/W. Fixes some ntdll:rtl winetests.
See issue #6779 for more details.
svn path=/trunk/; revision=54742
- Wait to acquire locks in CcRosFlushDirtyPages
- This is needed because this path is used during critical times like shutdown and very low memory situations when flushes cannot fail
svn path=/trunk/; revision=54725
- Remove "Explore from here" from context menu. It's useless, does not exist in Windows and can't be translated in current form.
svn path=/trunk/; revision=54723
- Fix broken code in drive.cpp. Spotted by r3ddr4g0n.
- General code improvements
- Don't try to create static open handler in registry when file properties are opened. Fixes two "Open" items in shortcuts context menu after opening properties window.
See issue #6746 for more details.
svn path=/trunk/; revision=54722
- Fix SHCreateSessionKey
[EXPLORER]
- Don't run startup applications if StartupHasBeenRun key exists. If it doesn't, create new key.
- Make startup code readable
See issue #1801 for more details.
svn path=/trunk/; revision=54715
- A number of fixes to CmpQueryKeyName():
* Properly check the provided buffer size against needed size.
* Don't overwrite user provided buffer.
* Write as much data as could fit into the buffer (this is normal behaviour for any query function in the kernel), returning STATUS_INFO_LENGTH_MISMATCH if not all data were written.
Thanks to r3ddr4g0n for identifying the problem, testing with DPH and testing this patch.
svn path=/trunk/; revision=54711
- Properly close Open With dialog window (EndDialog works only for modal dialogs, modeless dialogs should be destroyed with DestroyWindow instead). Fixes folder window hang if Open With windows was opened twice (and also a window leak).
svn path=/trunk/; revision=54701
- "Hyperspace is local!": OK, system PTE space isn't…
- Change the function name to reflect the real location of the mapping and match its partner MiUnmapPagesInZeroSpace()
svn path=/trunk/; revision=54700
- Lock the registry before flushing hives to avoid, among other things, use of an ERESOURCE without being inside a critical region
svn path=/trunk/; revision=54694
- Tweak the balancer a bit to favor RosMm allocations in low memory scenarios
- Fix the balancer out of memory warning
svn path=/trunk/; revision=54693
- Only unmap pages if they were mapped
- Make sure the page reference counts are sane when freeing the cache memory area
svn path=/trunk/; revision=54690
- Fix getting name for context menu item. Fixes two "find" items in My Computer context menu.
- Minor code and comments improvements
svn path=/trunk/; revision=54688
- Fail if display name device is given to EnumerateDisplayDevices and index is not set to 0. Fixes hang in Display Settings -> Advanced -> Monitor.
svn path=/trunk/; revision=54679
- Revert 54075 hack (registration from old shell32 based on WINE), so things does not get registered twice. Two Control Panels in MyComputer are no longer visible
- Fix all regressions, which was hack-fixed before.
- Add more debug output in case of errors
svn path=/trunk/; revision=54673
- Implement ElfrReadELA.
- Improve LogfReadEvent in such a way that it can return ANSI and UNICODE log entries.
- Add a hack to ElfrReadELW in order to return a proper status code in case of an end-of-file situation.
svn path=/trunk/; revision=54672
EngAllocMem:
- Respect the FL_NONPAGED_MEMORY flag (actually allocate the memory from non-paged pool)
- Fix an improper flag comparison that caused memory allocated with both FL_NONPAGED_MEMORY and FL_ZERO_MEMORY set to not be zeroed as requested
svn path=/trunk/; revision=54669
- Sync translated resource files (DeskADP and most of CPL) with English files.
- Small Czech update of DXDiag and TaskMgr translations by Radek Liska (bug #6774).
svn path=/trunk/; revision=54663
Follow original design/style of printf module
Always use dosmaperr to convert win32 errors to sane crt errors
Enable forgotten _get_doserrno, get_errno, set_doserrno and _gmtime32_s
Small fixes to pass more wine tests.
Patch by Samuel Serapión.
svn path=/trunk/; revision=54661
- Fix SxS path calculation, now it returns proper full path to the DLL instead of just "C:\ReactOS\winsxs". Spotted by AbiWord 2.9.2.
svn path=/trunk/; revision=54659
- inetcpl romanian translation update, resource file update for all languages by Stefan Fulea (bug no. 6766);
- polish translation update of inetcpl and resources changed by revision 54644;
svn path=/trunk/; revision=54657
- Implement ElfrClearELFA and ElfrClearELFW.
- Fix the module name conversion in ElfrOpenELA and ElfrRegisterEventSourceA.
svn path=/trunk/; revision=54656
Nothing calls this code yet (SearchPathW soon will), this is just an implementation commit. Code was tested to boot to 3rd stage after SearchPath modification, which is the next target (step 2).
After which, the only code in the loader paths that uses legacy path APIs will be the create process path, the third and final step.
svn path=/trunk/; revision=54646
- Handle SO_SNDBUF in WSPSetSockOpt so that it doesn't fall through to wshtcpip and result in "unknown level: 65536".
- Added more trace messages.
See issue #5477 for more details.
svn path=/trunk/; revision=54643
ReportEventA/W: Strings must be passed to ElfrReportEventA/W as a pointer to an array of PANSI_STRINGs instead of a pointer to an array of ANSI_STRINGs. Now, strings appear in the event viewer.
svn path=/trunk/; revision=54642
- Fix Get/SetWindowPlacement and everything related. Which exposed the scroll bar over write issue, it's not refreshing the client window when it is maximized. Next thing to fix.
- Fix ShowWindowAsync, now restore and minimize all works. This is based on wine but in reality it's based on Windows queuing event messages used in server side internal calls. Also this patch includes the server side hooking points.
- Removed one ugly API.
- Fixes bug 6239, 6739 and client window restore from maximize and the incremental crawling down issue when parent window is minimized then restored.
svn path=/trunk/; revision=54637
- Fixups to set focus and show window support, added notes. Add is window visible. Fix a binary bit return, 1 is TRUE. Fix get client rectangle. Minor fixups.
svn path=/trunk/; revision=54634
- Cache the last data block and offset so we don't have to search from the start each time
- This avoids needless swapping of cabinet data in and out of RAM
- File copy in text-mode setup on 32 MB RAM is down from 8 minutes to 20 seconds (in my testing)
svn path=/trunk/; revision=54632
- Fix invalid errors about "system DLL kernel32.dll was relocated in memory". This reverts my previous fix suggested by rafalh and puts a better explanation there in comments.
svn path=/trunk/; revision=54629
- Improve and rewrite a few active context support functions in RTL, making them more close to their real functionality and throwing away wineisms.
- Uncomment a line in RtlAllocateActivationContextStack() which was causing it to always allocate a new activation context stack even if there was one. Thanks to my previous commit in kernel32 counterpart, this hack is not needed anymore and memory won't be wasted.
- Since we have mostly Wine's code in our activation context implementation, it makes sense to use it properly until it's rewritten, so first of all:
- Properly initialize current process's activation context support by calling actctx_init() from LdrpInitializeProcess instead of calling it all the time when a new context is created.
- Bring in create_module_activation_context and find_actctx_dll (newer version from Wine which supports manifests/assmeblies in local directories) and use it. Every place where it's used is marked as a HACK so it will be easy to detach it later.
- Remove actctx-related hack LdrpInitializeThread, now it works properly.
- Fix a typo in LdrpMapDll found by rafalh. Now the error-detection condition will be correctly evaluated.
See issue #6611 for more details.
svn path=/trunk/; revision=54623
- Zero-initialize ActivationContextStack pointer before calling RtlAllocateActivationContextStack(). This allows to get rid of some hacks in the existing ActCtx support in the ntdll loader and actual implementation.
svn path=/trunk/; revision=54618
- Don't strip relocations for some sections. Linker knows better if they are needed. Fixes removed relocations for delayed imports in idata section. Bug traced by me, Timo and Amine.
See issue #6618 for more details.
svn path=/trunk/; revision=54613
- Massive IPPacket usage overhaul (fix missing frees, initializations, and double frees)
- Make IP receive completely agnostic of the buffers and data location (part 2 of 2)
- Use paged pool for the majority of allocations during receive and assembly
- Implement ProtocolReceivePacket for faster performance with 3rd party NIC drivers
- Fix a buffer overflow if an IPv4 packet is received with a header larger than 20 bytes
- Many more improvements and bug fixes
svn path=/trunk/; revision=54597
- Fix more set active window API tests. The window postion change message with fixed lparam is erroneous. How can the address be the same since it originates from kernel space?
svn path=/trunk/; revision=54596
- Avoid an extra copy operation when receiving packets
- Optimize loopback code a bit
- Make IP receive independent of the location of the data and continuity of buffers in the NDIS packet for ProtocolReceivePacket support (part 1 of x)
svn path=/trunk/; revision=54593
[KERNEL32]: Replace BasepGetDllPath with BasepComputeProcessDllPath. Paths now work right. This is step 1. SearchPatchW is next (BasepComputeProcessPath), followed by CreateProcessInternalW (BasepComputeProcessExePath). Then all paths will work right.
svn path=/trunk/; revision=54592
- Fix a logic error
- Signal the balancer in the !CanWait case if we're too low on pages since it runs asynchronously as of recently
svn path=/trunk/; revision=54591
- Prevent buffer overflow when creating the display value for a value of type REG_MULTI_SZ.
- Prevent processing excess data when preparing a value of type REG_MULTI_SZ for editing.
svn path=/trunk/; revision=54590
- Print a warning and signal RosMm to trim pages if we drop below MmMinimumFreePages in MiRemovePageByColor and MiUnlinkFreeOrZeroedPage
svn path=/trunk/; revision=54589
- Depending on how big the full path is, either the application path or the full path should be used. Size the buffers correctly, too.
- If the caller did not specify a window title, should use the app path as title.
- Clear the current directory handle if InheritHandles is false.
- Correctly handle CREATE_NEW_PROCESS_GROUP and CREATE_NEW_CONSOLE.
- Check the failure codes of many APIs that we just assumed returned success.
- STARTF_USEHOTKEY | STARTF_SHELLPRIVATE should be checked when inheriting console handles.
- Handle 1MB and 16MB reservation requests.
- Bugs remain: ImageSubsystem not written, and failure will leak. To be fixed later.
svn path=/trunk/; revision=54587
- Convert the remainder of PL strings to UTF-8 to fix build, since that RC is declared as Unicode. Convert italian resources to UTF-8 and place all rc in a single section;
svn path=/trunk/; revision=54582
- Release the address space lock before entering SEH because an exception would result in recursive acquisition of the address space lock when the fault handler tries to grab it
svn path=/trunk/; revision=54579
Remove class2 and scsiport service keys from the registry because these drivers are loaded automatically when they are needed.
svn path=/trunk/; revision=54575
- store property item when there are custom specific properties
- Fix bug where the property handle routine received an invalid pointer to the data
svn path=/trunk/; revision=54572
add wine dll registration functions to libwine (taken from winecrt0)
TODO: support for linking multiple res files
rbuild/cmake: generating .rgs files from typelibs
svn path=/trunk/; revision=54566
NtFreeVirtualMemory:
- Handle the case where a region size of 0 is passed in
- Return the correct error status for failure
- Copy back the rounded values
- Add checks and prints to catch callers doing nasty things (one is commented out because csrsrv triggers it trying to release the first 1 MB of RAM during video init, not sure what to do there)
- There is a heap bug where calling RtlReAllocateHeap on a large block allocation (HEAP_ENTRY_VIRTUAL_ALLOC is set) will cause a call to NtFreeVirtualMemory with an offset base (illegal) and a length smaller than the total pages reserved in the VM region (also illegal). This bug is exposed by setupapi when it parses large INF files (like the PRO/1000 driver INF).
svn path=/trunk/; revision=54565
Normalize all crt dll entry points to be identical.
Implement a single per thread data structure, instead of ancient and incomplete
"wine port" hack and use it everywhere.
Remove completely unused HeapCreate calls that just wasted memory.
Update cpp.c and cppexcept.c to wine 1.3.33
Fix up strtok, wcstok, rand, and _beginthread_trampoline according to wine.
Add place holders for future locale stuff.
Patch by Samuel Serapion.
svn path=/trunk/; revision=54563
- Implement NdisReturnPackets and do proper reference tracking of packets sent to ProtocolReceivePacket
- Plug a massive memory leak that resulted in leaking every packet descriptor that any deserialized miniport indicated to TCP/IP
svn path=/trunk/; revision=54559
- Call FreeCommonBuffer() synchronously if we're running at PASSIVE_LEVEL to avoid cases where the miniport could be freed before the work item runs
svn path=/trunk/; revision=54558
- Fix regressions, running RegEdit. Adding more support for scroll bar controls. Pass all but 7 tests including the two wine todos.
svn path=/trunk/; revision=54556
- Fix scrollbar class support. Sync ports from wine. Added the window scroll structure to the class. Pass all but 15 tests, two are wine todos lines 304 and 343.
svn path=/trunk/; revision=54555
- Optimize CcRosTrimCache by eliminating an extra loop of all segments
- Allow swapping of dirty segments by flushing them first
svn path=/trunk/; revision=54549
- Add NtAllocateVirtualMemory test, which is a small stress test for virtual memory allocation/freeing/reuse
- Can be used to reproduce bug 5857 in third stage (install with VT-x enabled, then disable for the test)
- Shows that the bug is a race condition in Mm, and not heap's fault
- I've put ASSERTs instead of ok's for easier debugging
svn path=/trunk/; revision=54548
- Many balancer fixes for concurrency, improved swapping efficiency, and reduced code duplication
- Move the low memory case back under the PFN lock
- Debugging is on for now because I don't trust the paging code all that much (it was not used very much until recently)
svn path=/trunk/; revision=54544
- Revert part of r54530
- Try to rebalance RosMm's pages if we're low on memory in ARM3
- ROS runs with 48 MB RAM again (but setup is very slow due to horrendous paging caused by cache pollution)
svn path=/trunk/; revision=54539
- Don't signal the balancer from within the balancer
- Optimize usage of the balancer a bit so we don't have to wait on it as much
svn path=/trunk/; revision=54537
MSPaint:
- Add Swedish translation by Michael Wu (bug #6695).
- Sync all translation files with American English file (bug #6700).
svn path=/trunk/; revision=54533
- Display file size in bytes in file properties dialog
- Don't use ULONGLONG if ULONG is enough
- Change LARGE_INTEGER to ULARGE_INTEGER where it makes sense
- Fix calculation of file size in CountFolderAndFiles
- WM_GETTEXT receives size in bytes (not characters count)
- Use Get(Set)DlgItemTextW instead of SendDlgItemMessageW
Patch by Carlo Bramix. Improvements by me.
See issue #6294 for more details.
svn path=/trunk/; revision=54529
All .lnk files in Windows has structure called "id list". This is something like a path (we can't use environment variables there) and it is relative to Desktop. Usually first element is My Computer folder. In our shortcuts it points to livecd_start.cmd in My Documents folder, which is simple batch file. It's functionality is to start application given in argument. Because command line parameter of shell links can have environment variables, our shortcuts does not depend on patch to system directory (especially the drive number). It's a bit complicated but I have not found a better way.
Now we have working shortcuts on Desktop and Start Menu. Control Panel will benefit in future because Admin Tools depend on Start Menu.
svn path=/trunk/; revision=54514
Switch to the 'new' disk geometry (63 sectors per track and 255 tracks per cylinder). This geometry is compatible with Windows, Linux and other Operating Systems.
ATTENTION: You must delete and re-create the partitions on your harddisk.
svn path=/trunk/; revision=54511
- Enable translations of the language page. Patch by Stefan Fulea.
- Simplified the clear text code.
- Fixed the german language page.
See issue #6637 for more details.
svn path=/trunk/; revision=54502
- Fix errno in dialog.c by linking to msvcrt with rbuild. Cmake does not have this issue, already links to msvcrt. Gdi32 rbuild does link to crtdll instead of msvcrt. Should we fix this too, before the Cmake switch over to see anything is different?
svn path=/trunk/; revision=54498
- Fix a typo in NtUserSetClipboardData that tried to print a struct as a pointer (cid 15051)
- Clean up clipboard.c a bit. note: this cleanup "fixes" cid 14599 and 14598, but even though they are false positives, this cleanup was needed
- Add missing break in NtUserQueryWindow (cid 13486)
svn path=/trunk/; revision=54483
- Fix dwServiceType and dwServiceState checks in REnumServicesStatusW (see r54373)
- Fix a memory leak in RQueryServiceConfig2A
svn path=/trunk/; revision=54481
- Also round down offset we're given when creating a page op, so we don't miss the one that could be running at the time
svn path=/trunk/; revision=54480
- create a page op when changing page protection attribute
- properly handle PAGE_NOACCESS protection change in NtProtectVirtualMemory
- Round down the address we're given when creating a page op, so we're sure there is o,ly one page op for running for one page
- Add a small hack
Now ReatOS boots with DPH enabled on low memory conditions
svn path=/trunk/; revision=54477
- Fix process reference leak in case of GuiConsoleOpenUserRegistryPathPerProcessId failing to open registry key. Fixes console processes in LiveCD being ghosts.
svn path=/trunk/; revision=54475
- Do not repeat messages into serial port even when using other debug methods.
- Allow to specify debug log file name.
- When using the "Screen" method, show blue screen when entering KDB.
- Disable the repetition of previous KDB command with long many-page output.
[USETUP]
- Add "ReactOS (Screen)" entry to boot menu.
svn path=/trunk/; revision=54468
- Print debugging information if a mismatch occurs between requested and returned parameter types
- Print debugging information if a registry read returns a type that NDIS doesn't handle
- Ensure that all digits are less than the base during conversion from string to integer (otherwise it's a string)
svn path=/trunk/; revision=54464
Break ReactOS CRT:
- Add back crash on NULL pointer arg. This is the behaviour on w2k3.
- Don't skip \\?\ prefix any longer. This is the behaviour on w2k3.
svn path=/trunk/; revision=54452
- Display all hardware profiles
- Add the hardware profile copy dialog to most languages.
- Disable the Up, Down, Property, Copy, Rename and Delete buttons in the Hardware Profiles dialog.
svn path=/trunk/; revision=54448
- Bring more awesomeness to the heap manager by zefklop's request:
* Add support for settable user values and flags in RtlAllocateHeap and RtlReAllocateHeap.
* Return error if RtlSetUserValueHeap was unable to set the value.
* Fixes remaining 2 failures of the kernel32_winetest alloc test.
svn path=/trunk/; revision=54434
Fix _fpclass prototype
Fix headers included
Remove internal defines that were useless and even not matching real defines.
This fixes 4 tests
svn path=/trunk/; revision=54432
- Updating MsgiAnsiToUnicodeReply fixed bug 4856 and 6650. Mirrored with it's counterpart. More work is needed.
- Adding some future DBCS char support.
- Miscellaneous changes.
svn path=/trunk/; revision=54428
Add test suite for IEEE floatting-point functions.
It's been tested on w2k3 and it has 0 failed tests.
Some are failing on ReactOS.
svn path=/trunk/; revision=54425
- Implement support for hot plugging PS/2 mice if one was present at boot (same requirement as Windows)
- Fixes bug 1395
svn path=/trunk/; revision=54415
- Make sure GetEnvironmentVariableW does not use uninitialized variable
- Fix GetDllLoadPath not terminating string with NULL if PATH env variable is not defined. Fixes hang on manual INF selection in New Device wizard.
See issue #6480 for more details.
svn path=/trunk/; revision=54406
- Fix a critical bug in registry entry parsing that was causing all REG_SZ values to be reported as NdisParameterTypeInteger
- Distinguish between NdisParameterTypeInteger and NdisParameterTypeHexInteger
- 3Com NICs are fully functional now [bug 4330]
svn path=/trunk/; revision=54402
- Fix running the test with a relative path. Already Sent Upstream(TM)
- Do the rest of that tiny Winesync, too, while we're at it. Now at 1.3.33
- Remove the spooler hack in the process. We don't actually need it
svn path=/trunk/; revision=54400
- Romanian translation update (bug no. 6654) by Stefan Fulea;
- Simplified Chineese translation of base/shell and system (bug no. 6655) by Song Fuchang;
- Slovak translation fixes (bug no. 6658) by Mario Kacmar;
svn path=/trunk/; revision=54399
- Some more cleanup.
- Work with page rounded down addresses in SectionView page faults.
- You should not create a virtual mapping on a non page-aligned address range : ASSERT about that.
- Flush Tlb each time we change it
- Fix logic I broke in MmCreateVirtualMapping (thanks to Thomas Faber)
svn path=/trunk/; revision=54392
- ensure that all callers of MM providing an user land address also provide a process.
- ASSERT about that
- Fix broken check
svn path=/trunk/; revision=54390
- call mm functions with a process, when we have one.
- Fix potential rounding issues
- Add some sanity ASSERTs
- You never use enough brackets ;-)
svn path=/trunk/; revision=54386
- Remove broken map register "prefetcher" code
- Check that we have enough space to grow our map registers
- Fix a bad hint to RtlFindClearBitsAndSet
- Queuing the grow work item is non-vital; treat it as such
- Fixes BSOD with CM3738 and VIA1723 sound cards [bug 6657]
svn path=/trunk/; revision=54382
- Always return ERROR_SUCCESS when ScmWaitForServiceConnect completes successfully instead of sometimes returning ERROR_PIPE_CONNECTED because the callers of this helper just look for ERROR_SUCCESS to indicate a successful call
svn path=/trunk/; revision=54381
- Export IContextMenu interface in RecycleBin object. Fixes recycle bin context menu.
- Export IShellPropSheetExt interface too and fix order.
See issue #6543 for more details.
svn path=/trunk/; revision=54379
- Fix a critical bug in NdisMGetDmaAlignment which was causing every NIC driver calling this function to crash because we thought we were passed an NDIS_DMA_BLOCK handle when in fact we receive an NDIS_MINIPORT_BLOCK handle
- Fixes RTL8169 and 3Com NICs (and probably several more) [bug 6651 and bug 4330]
svn path=/trunk/; revision=54372
- Fix DMA on the EISA bus
- Fix a critical calculation error in map register size resulting in the map register allocation worker not allocating enough map registers for a dynamic resize
svn path=/trunk/; revision=54371
- base/setup,shell and system conversion patch, by UTF-8 conversion patch by Elton Chung, see bug 6333;
- Romanian translation fixes and UTF-8 conversion by Stefan Fulea, see bug 6603;
svn path=/trunk/; revision=54364
[KERNEL32]: Add notes on how DeviceIoControl and Beep should support TS scenarios in the future (nod to Ged).
[KERNEL32]: Make Beep call NotifySoundSentry, and implement the latter, which sends a message to CSRSS w.r.t the previous commit. Obviously we are not anywhere close real sound sentry support, but at least part of the path is there now.
svn path=/trunk/; revision=54358
- Allow changing the image path with ChangeServiceConfig. Fixes timeout that shouldn't actually be a timeout in advapi32:service test.
See issue #6646 for more details.
svn path=/trunk/; revision=54348
- Disable usbdrv
- I've heard too many reports of it breaking boot and too few reports of it actually working on real hardware (although it does work perfectly in QEMU)
svn path=/trunk/; revision=54347
- Skip service start timeout tests. They cannot time out until services uses asynchronous I/O.
See issue #6646 for more details.
svn path=/trunk/; revision=54346
- Do not use local memory functions (and possibly free the result using HeapFree!)
- Do not use double percent signs with _wcsnicmp
- DPRINT and style fixes
svn path=/trunk/; revision=54344
[KERNEL32]: Rewrite the tape APIs to use a single generic function, and in the process support pending I/O now, which wasn't handled before.
[KERNEL32]: Moved the interlocked APIs to synch.c instead of dllmain.c
svn path=/trunk/; revision=54334
- Fix a potential CCB state inconsistency if the IRP is canceled before adding it to the pending list
- Don't wait if the IRP has been canceled
- Preserve the return status of NpfsAddListeningServerInstance
svn path=/trunk/; revision=54328
[KERNEL32]: Cleanup files a bit.
- Bintype.c is gone and moved into VDM (these are functions for VDM support).
- Npipe.c and pipe.c are merged.
- Dosdev is moved into client, these are not quite File APIs.
- Base(8Bit/Unicode) APIs are moved into utils.c
- Wow64 APIs are moved into utils.c
- The mess that was file.c has the Replace* file APIs going into move.c (similar behaviors), the Open* APIs into create.c (similar) and the Read/Write Scatter/Gather APIs into rw.c (where they belong). What remained in file.c were mostly Set/Query functions, so it becomes fileinfo.c
- The remaining APIs in file.c were all about paths -- at the file level, specifically filenames, so they go into filename.c
- Move some File API functions from certain files into other files where they belong more.
svn path=/trunk/; revision=54326
- Make the initial NpfsRead wait uninterruptible since it's really being used like a mutex (avoid the kernel stack being paged out during the wait causing memory corruption)
- Handle interrupted waits that were missed in certain cases
- Fix a potential deadlock that would happen if more I/O was done on a file object after the previous operation was interrupted by an APC or alert
svn path=/trunk/; revision=54325
[KERNEL32]: Reimplement GetFullPathNameA to implement similar functionality and checks as the other reimplemented A functions, especially w.r.t ANSI/OEM and DBCS strings.
svn path=/trunk/; revision=54321
- Fix wait parameters so we can break out of a wait if a user APC is delivered and the file object is opened with FILE_SYNCHRONOUS_IO_ALERT
- Fixes hanging ntdll:pipe winetest (only 1 failure out of 53 tests now)
svn path=/trunk/; revision=54319
- Windows allows the two input buffers to overlap. This means we have to be very careful with copying (and always use RtlMoveMemory) -- the old function was not handling this at all.
- We also have to handle cases where we need to make our own local buffer copy.
- Length validation is more stringent now.
- Checking for short/long path names wasn't correct w.r.t ANSI file mode, as it was calling Rtl which assumes OEM.
- Shortcuts were taken while parsing slashes and path separators. We now call into Rtl to support this, and also support unlimited slashes (note that \\??\c:\\\windows\\\\system32 is actully a valid path, for example).
- ErrorMode is now correctly set for the thread, to avoid "Insert floppy" or "Close CDROM bay door" errors as we are using the FindFile API.
- Correct LastError is set where appropriate.
- An application compatibility flag is now supported.
svn path=/trunk/; revision=54318
- Use the correct buffer length in GetDllDirectoryA
- Do not always return FALSE from BasepIsCurDirAllowedForPlainExeNames
- Some cosmetic fixes
svn path=/trunk/; revision=54317
- Use FIELD_OFFSET for variable-length structure sizes
- Handle an invalid parameter condition in RCreateServiceW. Fixes an advapi32:service test
- Do not dereference a NULL-pointer on out-of-memory
svn path=/trunk/; revision=54315
[KERNEL32]: Fix SetDllDirectoryW:
- Fail if ";" is in path
- Use RtlCreateUnicodeString instead of manual string management -- if the allocation fails, do not free the current base directory string! Older API would always free the current directory string, even if setting up the new one failed.
- Don't assume lpPathName is filled out, it can be NULL, and this means the DLL directory should be cleared.
[KERNEL32]: Fix SetDllDirectoryA as above, additionally use RtlInitAnsiStringEx to protect against overflow, and use Basep8BitStringToUnicodeString, to add support for OEM paths instead of assuming ANSI.
[KERNEL32]: GetDllDirectoryW: correctly NULL-terminate and check lengths
[KERNEL32]: GetDllDirectoryA: same as above, plus use BasepUnicodeString* APIs to correctly support OEM paths instead of assuming ANSI.
[KERNEL32]: NeedCurrentDirectoryForExePath(A/W): call internal BasepIsCurDirAllowedForPlainExeNames instead of converting from A->W, this isn't needed.
[KERNEL32]: GetFullPathNameW just calls RtlGetFullPathName_U, no extra checks/etc are needed.
Next steps are GetFullPathNameA, GetShort/LongPathNameA, GetShort/LongPathNameW, hoping to bring the wine test failures closer to zero.
svn path=/trunk/; revision=54309
- Initialize OLEINPLACEFRAMEINFO.cb for IOleInPlaceSite::GetWindowContext. Fixes explorer "Web" button assert. Already applied upstream.
See issue #6011 for more details.
svn path=/trunk/; revision=54304
- Fix ParseDisplayName based on wine by checking for paths that are parsable through IEParseDisplayNameWithBCW. Score 2 failed wine tests in shlfolder
svn path=/trunk/; revision=54300
- Sync heap.c and undname.c with Wine 1.3.32
- Remove incorrect casts. Undo an incorrect warning fix. Fixes hang in msvcrt:cpp test
svn path=/trunk/; revision=54297
- Fixed Dialog Ansi and color control brush acquisition.
- Added stack protection for message recursion. Due to wine message bug when in dialog initialization. See bug 6126.
- Miscellaneous fixups.
svn path=/trunk/; revision=54293
[KERNEL32]: Acquire SE_INCREASE_WORKING_SET_PRIVILEGE in SetInformationJobObject if needed.
[KERNEL32]: Only do conversion of Process Priority Class in SetInformationJobObject if the limit flags request it.
[KERNEL32]: Add missing size checks in QueryInformationJobObject.
[KERNEL32]: Use RtlEncode/DecodePointer to protect the top level exception handler pointer, as done on NT 5.2+
[KERNEL32]: Fix IsBadRead/WritePtr to use the correct page size from the CSRSS shared section.
[KERNEL32]: Fix IsBad(String)Read/WritePtr loops
[KERNEL32]: SetErrorMode now handles SEM_NOALIGNMENTFAULTEXCEPT.
[KERNEL32]: Set/GetErrorMode do not set LastError on failure.
[KERNEL32]: RaiseException can simply use memcpy instead of copying argument by argument.
[KERNEL32]: Minor formatting changes.
svn path=/trunk/; revision=54291
- StackLimit was set incorrectly.
- Code was not using BaseStaticServerdata, but querying NT instead.
- Fix memory leak in failure case.
- StackCommit and StackReserved values were not aligned correctly.
- Windows Server 2003+ feature of "Guaranteed Stack Commit Size" was not respected.
- Some math was screwy.
- Failure to get NT headers was not handled.
svn path=/trunk/; revision=54285
[KERNEL32]: Fix some error cases in the helper string functions.
[KERNEL32]: Implement BasepIsRealtimeAllowed.
[KERNEL32]: Don't need RPL_MASK when setting up the CONTEXT, this now matches what Windows sets up (and also helps us validate the kernel is doing the right thing).
svn path=/trunk/; revision=54284
- Small performance improvement : Directly reference dll when getting its handle instead of just checking it's there and then load it
svn path=/trunk/; revision=54283
- Use a separate function, InstallFat12BootCodeToFloppy(), to install the floppy boot code
- Format the floppy to FAT12 before writing to it
- Booting FreeLoader from floppy works again (bug #6495)
svn path=/trunk/; revision=54269
- Prevent a deadlock on very old controllers by sending the configure and version commands before resetting the controller
svn path=/trunk/; revision=54267
- Improve the floppy controller initialization code to fix several cases where the controller could deadlock or otherwise malfunction on us
svn path=/trunk/; revision=54266
- Icon title, menu and scroll bar do not set the death bit. Menu and scroll bar set the id only on specific window messages.
svn path=/trunk/; revision=54261
- Fix function ids, setting of the death bit when in NC destroy.
- Implement get control brush and color.
- Patch by Alexander LAW, Replicate Windows behavior of WM_SETTEXT handler regarding WM_CTLCOLOR*
svn path=/trunk/; revision=54259
- Pass a placeholder trap information, indicating that we are unlocked, to MmAccessFault in MmProbeAndLockPages as temporary hack until Mm locking is fixed
- Fixes "Assertion 'GuardedMutex->Owner == KeGetCurrentThread()' failed" when faulting in pages via MmProbeAndLockPages
- Dedicated to Jim :)
svn path=/trunk/; revision=54256
- Up Dialog.c to 1.3.31, sync port and minimized differences.
- Sync Information:
Alexandre Julliard <julliard@winehq.org> : Don't overwrite the default button id when creating the dialog structure.
Michael Stefaniuc <mstefani@redhat.de>: Avoid using long. Avoid using HIWORD() on atoms/resources.
svn path=/trunk/; revision=54255
- Fix class window procedure checks, prevents misbehaving applications from calling wrong class procs.
- Fixed callout for static control color brush.
svn path=/trunk/; revision=54249
- Use SetWindowTextA/W so that it passes through defwnd to be processed for themes.
- Fix validation of the dialog window and procedure.
svn path=/trunk/; revision=54248
- Fix SetWindowTextA/W so that it passes through defwnd to be processed for themes.
- Fix desktop checks.
- Add create window flags.
svn path=/trunk/; revision=54247
- Fix a major off-by-one bug in resource conflict checking
- Respect requested alignment when checking for valid port/memory ranges
svn path=/trunk/; revision=54238
- Report legacy resources to the PnP manager via IRP_MN_FILTER_RESOURCE_REQUIREMENTS
- Report legacy access ranges in response to VideoPortGetAccessRanges
- TODO: Handle an optional resource requirements list in VideoPortGetAccessRanges
svn path=/trunk/; revision=54237
- Slot is an output parameter not an input parameter. Treat it as such.
- Don't use magic constants
- Remove a duplicate check
- Set RangeShareable for port resources based on the resource descriptor
- Initialize and set RangeShareable for memory and port access ranges instead of leaving garbage in there
svn path=/trunk/; revision=54235
- Fix ME_CharFromPoint relying on Wine's incorrect behavior of GetTextExtentExPointW. Already sent upstream(TM)
See issue #6196,5784,6037,6095,6187,6513 for more details.
svn path=/trunk/; revision=54234
Implement argument processing. Patch by Lee Schroeder.
The original patch was modified to fix the following issues:
- Indentation and coding style.
- Remove obsolete code: 1. The if-statement around the for-loop is not needed. 2. The call to wcsicmp() is already case insensitive (it is the 'i' in wcsicmp).
- Check the existence of a script file name and a timeout value before trying to process them.
- Call run_script() after the argument list has been processed. This fixes the argument order issue.
See issue #6552 for more details.
svn path=/trunk/; revision=54233
- Move playing log on sound to HandleLogon
- Don't use HKEY_CURRENT_USER for accessing user registry key. This key points to settings of user which started Winlogon process (SYSTEM) instead of logged user. Instead use HKU\SID.
See issue #5436 for more details.
svn path=/trunk/; revision=54232
- Mouse code cleanup
- Support MOUSEEVENTF_VIRTUALDESK flag in MOUSEINPUT properly
- Support MOUSE_VIRTUAL_DESKTOP flag in MOUSE_INPUT_DATA properly
- Do not ignore MK_SHIFT and MK_CONTROL flags in some mouse messages
svn path=/trunk/; revision=54227
- Don't set every loaded layout as default. Instead implement SPI_SETDEFAULTINPUTLANG. Fixes some minor problems with wrong layout loaded.
- Fix removing hotkey not associated with window (fixes winetest).
svn path=/trunk/; revision=54223
Mark the VGA driver as VgaCompatible too so it doesn't take precedence over the VBE driver in case both are installed.
See issue #6578 for more details.
svn path=/trunk/; revision=54221
- SysSetup: add a very tiny enhancement to Spanish translation (Elhoir, bug #6596).
- USetup: change encoding of Brazilian Portuguese file to CP 850.
- USetup: fix a few bugs in German and Italian translations.
svn path=/trunk/; revision=54219
- Use a Yes/No message box for "Quit the current game?" instead of Ok/Cancel. Patch by Lee Schroeder
See issue #6589 for more details.
svn path=/trunk/; revision=54218
- Fix profiles path. FIXES ICONS IN LIVECD!
- Add few interfaces to livecd registry so it doesn't spam the log so much
svn path=/trunk/; revision=54216
* Don't even create the stlport static lib. No working C++ module links to it, and until we have a proper STL support the other modules won't compile either.
svn path=/trunk/; revision=54215
- If no keyboard layout can be loaded, load US layout
- Try to load all layouts from Preloaded key even if some of them fail
svn path=/trunk/; revision=54212
- Fix path support, now all wine path tests pass.
- Need a PolyDraw test for non-path functions! The non-path code was synced ported, but still failed 16 path tests, then path was moved out. Purged missing patches from the last trunk freeze, tested with the program Area.
- Sync Information:
Evan Stade <estade@gmail.com> : Improved PolyDraw in path closed case. Added PATH_PolyDraw.
svn path=/trunk/; revision=54210
- Fix crash in PATH_ExtTextOut. See bug 6587.
- Sync Information:
Huw Davies <huw@codeweavers.com> : Add support for ETO_PDY and improve world transform support.
Massimo Del Fedele <max@veneto.com> : Fix PATH_add_outline when mapping mode != MM_TEXT. PATH_ExtTextOut remove incorrect shift to DC origin. Correctly handle space char on Path_ExtTextOut().
Dmitry Timoshkov <dmitry@codeweavers.com> : The MAT2 parameter of GetGlyphOutline is mandatory.
svn path=/trunk/; revision=54209
- Make keyboard layouts code more compatible
- Layouts and kbd files are user objects
- Preloaded layouts are added by Winlogon, not win32k
- Support Keyboard Layout substitutes in registry
svn path=/trunk/; revision=54203
- When themes are active and the text of a window changes, we should send the WM_NCUAHDRAWCAPTION message instead of drawing the window directly ( which will result painting a classic caption on the themed window)
- Fixes explorer window that is repainted with a classic style when a subfolder is opened
svn path=/trunk/; revision=54201
- Simplify monitors code
- fix few minor bugs (ie. not updating last error)
- Make monitor object more compatible to windows
svn path=/trunk/; revision=54197
- Probe input / output parameters when requests are coming from user mode
- Fix KspPinPropertyHandler to use the correct output parameter location
- Fix KsHandleSizedListQuery to use the correct output parameter location
- Tested in Vbox 4.1.4 r74291
- May fix a few sound drivers who are having user mode extra utility apps
svn path=/trunk/; revision=54191
- Apply a wallpaper and desktop background only when they have changed
- Fixes changing desktop background to the color defined by the theme
svn path=/trunk/; revision=54189
-Force all top level windows to update the size and position of their non client area when SPI_SETNONCLIENTMETRICS is used
-Should fix changing the non client metrics like caption height for open windows (before this we needed to close and reopen existing windows to let them be painted properly)
svn path=/trunk/; revision=54186
- Do not send message WM_NCCALCSIZE asynchronously because we need its return value from the receiver. Should fix resizing or moving the non-client area of windows
svn path=/trunk/; revision=54185
- Change time zone information for Russia.
- Add Russian translation of country list.
- Recode Swedish translation of USetup from CP 1252 to 850. Bug #6574.
- Fix Dutch tr. of USetup. Bug #6510, patch by <congruwer.AT.yahoo.DOT.co.DOT.uk>.
svn path=/trunk/; revision=54178
- Undefine HAVE_GETADDRINFO
- Even though this is fully implemented (and used other parts of ROS), it seems to enable broken code paths in some WINE code
- Fixes the rapps regression
svn path=/trunk/; revision=54174
- Fix TME. Not complete. These changes will allow theme highlighting to work.
- Passes wine test_TrackMouseEvent.
- Adamopoulos TME tests are proving to be difficult in locking down thread issues. On a slow system in CMD not redirected the test passes all but 15. Slow systems in CMD redirected to a output file or on real hardware in CMD not redirected, the results vary from 15 when the CMD buffer is full to 35 when it is not. When isolating each test, it passes. The QS bit issue will be handled later.
svn path=/trunk/; revision=54172
- Differentiate between incomplete and complete NCEs with regard to the timeout
- Change the IP timer to fire only once each second
- Remove unneeded completion functions
svn path=/trunk/; revision=54171
- Fix timer resolution so it will run every 10 ms +/-1ms. This is the default setting when the user provides less than 10 ms for the period to wait for the timer event. Pass wine test_timers_no_wnd.
svn path=/trunk/; revision=54170
- Do not transmit fragments recursively since it can cause a kernel stack overflow with large packets
- Big thanks to hto for his work on this bug which has eluded me since last year
See issue #5796 for more details.
svn path=/trunk/; revision=54169
- Romanian translation of userenv, winmm, ntvdm, and win32csr (bug no. 6580) and inf (bug no. 6577) by Stefan Fulea (minor corrections by me);
- Polish translation update/cleanup of inf by me;
- Spanish inetcpl translation (bug no. 6472) and French subst translation (bug no. 6166) by Jason Genie;
svn path=/trunk/; revision=54168
- Prevent corruption of the search context list using a combination of references and broader spin lock usage
- Fixes bug 6506
svn path=/trunk/; revision=54167
- Remove ReadRegistryValue in kbdlayout.c as we already implement RegQueryValue for reading from registry in W32k. This function additionally checks value type.
- Rename KBL to KL based on Techwiki
- HKCU\Keyboard Layout\Preload value is REG_SZ (not REG_EXPAND_SZ)
- NtUserLoadKeyboardLayoutEx should set last error if flags are invalid
- Minor keyboard layout code cleanup
svn path=/trunk/; revision=54163
- Reapply the ConvertThreadToFiberEx changes from r52804, they were lost in r52805. Confirmed by Alex.
See issue #6394 for more details.
svn path=/trunk/; revision=54162
- Fix insertion of the GRAPHICS_DEVICE into the global list. Fixes mode enumeration when the 2nd adapter is used.
- Remove a hack.
svn path=/trunk/; revision=54156
- Increase size of the homepage link control to correctly show the complete focus rectangle. Patch by Elton Chung (elton328 at gmail dot com)
See issue #5038 for more details.
svn path=/trunk/; revision=54154
Mark the VBE driver as VgaCompatible. This is an outrageous lie, but it lets us use 3rd party drivers, like VBoxVideo, while VBE is installed. Consider this a hack, which is to be fixed bei either making vbemp really vga compatible or by fixing whatever keeps the VBoxVideo from replacing vbemp.
svn path=/trunk/; revision=54152
- Fix a bug when iterating through the DEVMODE list
- Handle VgaCompatible flag differently, by reading this value from the registry for every installed driver.
- Priorize non-vga compatible devices over vga compatible, unless /BASEVIDEO is requested
- Fall back to vga compatible driver when no other is present
svn path=/trunk/; revision=54151
- don't define _cdecl and __cdecl to nothing on MSVC builds.
- define cdecl and CDECL to nothing
- add #pragma once
- this file is full of hacks
svn path=/trunk/; revision=54149
* Sync config.h with recent Wine and remove some now unneeded defines. By Samuel Serapion.
See issue #6174 for more details.
svn path=/trunk/; revision=54146
* Move the wctype table from kernel32 to libwine in preparations for implementing crt locale support. By Samuel Serapion.
See issue #6538 for more details.
svn path=/trunk/; revision=54144
- Properly set the timer event id's to match tests and add one more bit for track.
- Restore TME user API (now in mouse.c) and Hover Timer, these are not the Droids you are looking for.
- Restore TME cancellation point in co_MsqInsertMouseMessage and duplicate the the same in co_IntProcessMouseMessage. First thing, pDesk->spwndTrack is never NULL. The Theme TME patch code was setting the wrong QS bit and sending to the wrong queue instead of the pDesk->spwndTrack queue. Surprising this was working at all.
- Tested with Themes, passes wine test_TrackMouseEvent and fixes bug (6257 <-- !).
Notes:
It's all in the sequencing. Locate cancellation points, tracking mouse moves and set the hit test and track window. When the window message queue does not match the current thread, check it again and restore the mouse hover window if it is the same window as the track window and the hover bit set. Track cancellation points and clear the bits at the same time. Not when in WM_MOUSEMOVE. Guess when posting or sending to another thread? co_MsqInsertMouseMessage?
svn path=/trunk/; revision=54142
- Get rid of RETURN macro in input functions
- Don't check functions parameters when USER lock is held
- Improve formatting
svn path=/trunk/; revision=54141
Fix the confusion between mode index 12 and video mode 12h introduced more than six years ago. ReactOS should now boot again when VGA is selected in the 1st stage.
svn path=/trunk/; revision=54139
* Fix a copy-paste error in EngBitBlt() and improve DIB_32BPP_ColorFill() behaviour in case when input rectangle is invalid (RECTL structure fields are signed, so "jle" instruction should be used instead of "jbe").
* Patch by Igor Sudarikov <4seev3 at gmail [dot] com>
See issue #4423 for more details.
svn path=/trunk/; revision=54138
Remove an old hack, that caused an inconsistency between the display's device object name and the name of its symbolic link / registry key. The hack isn't needed anymore, because win32k now loads the next device if one isn't available. Fixes VBoxVideo being loaded together with framebuf.dll instead of VBoxDisp.dll and now the mouse pointer integration works completely.
svn path=/trunk/; revision=54135
- Keyboard state is update when msg is peeked from queue (not posted) and before virtual keys are simplified. Fixes 2 user32:input winetests.
svn path=/trunk/; revision=54134
- Since some display drivers don't do as they should - return the drisplay driver name in the DEVMODE's dmDisplayName field - but return the miniport driver name, fix that name when creating the mode list. This fixes loading of VBox driver, when it is installed properly, which needs to be fixed as well.
svn path=/trunk/; revision=54133
- Remove _fltused from alldiv_asm.s / mscmain.c and use fltused.c instead.
- Remove amd64/alldiv.S
- Implement amd64 ceil, ceilf with sse
[NTOSKNRL]
- Fix amd64 exports
[ASM]
- Fix amd64 asm files to conform to the new syntax
svn path=/trunk/; revision=54131
- Remove outdated comments
- Make Raw Input Thread CSRSS thread too
Thanks to Smiley and ThFabba for investigating TEB issue
svn path=/trunk/; revision=54128
- Added Print Screen support in previous comit accidentally...
- Fix build and warning about global object being removed
svn path=/trunk/; revision=54126
- Fix and simplify retrieving the current process window station. All processes that enter win32k have a window station (including csrss)
svn path=/trunk/; revision=54124
- Enable usbdrv.sys in 1st stage setup and the system hive
- USB keyboards and mice work now (file bugs if you have problems)
svn path=/trunk/; revision=54123
- Add a small hack to bind all keyboard port devices to the same keyboard class device when no default setting is present (circumvents win32k not supporting multiple keyboard class FDOs)
svn path=/trunk/; revision=54120
- serialui, setupapi, shimgvw and tapiui romanian translation, patch by Stefan Fulea;
See issue #6558 for more details.
svn path=/trunk/; revision=54119
- Fix two bugs revealed by the newly added test in r54114.
* Original length of the passed string should be used when checking for out-of-buffer access in the loop
* PathChars should be updated if a DOS device name was found in the given string.
See issue #6390 for more details.
svn path=/trunk/; revision=54116
Improve .proc / .endproc syntax, so that the full symbolic name is passed, which allows to create fastcall symbols. Remove the need to pass the function name to .endp
svn path=/trunk/; revision=54112
Bail out of compare_emf_bits earlier. This way we only record one skipped test, but it allows us to actually complete the metafile tests. Restore formatting to its previous glory.
svn path=/trunk/; revision=54111
- Convert wParam in WM_*CHAR messages to ANSI in GetMessageA instead of win32k. Fixes wrong characters used in cmake notepad which is ANSI.
svn path=/trunk/; revision=54109
- Right ALT on some keyboard sends LCONTROL and RALT key events
- Shift has never extended bit in key messages
- Fixes about 32 user32:input winetest
- Don't allow to register hotkeys with special keys combinations (eg. WIN key)
- Set last error in RegisterHotKey if hotkey is already registered
svn path=/trunk/; revision=54107
Allow creating of shortcuts with icon indexes != 0.
[EXPLORER]
Shuffle the resources a bit so icons with indexes 0 and 1 match those in windows explorer.
See issue #6544 for more details.
svn path=/trunk/; revision=54105
- Restore global class operations.
[ComCtl32]
- Use local class since this has higher priority over all global classes.
[Themes]
- Tested with XP Luna and two others. Rune and another Luna.
svn path=/trunk/; revision=54104
- Fix duplicated entries in screensaver selector. Patch by Carlo Bramix <carlo dot bramix at libero dot it>.
See issue #6533 for more details.
svn path=/trunk/; revision=54098
On x86 builds use the asm versions of the byte swap functions. This is neccessary for RtlUlonglongByteswap, because the function breaks the calling convention on Windows, not cleaning up the stack. Make the asm compatible to ML and improve it a bit. Investigated by Thomas Faber.
svn path=/trunk/; revision=54097
Do not use image base addresses above MmHighestUserAddress in MmMapViewOfSection, because it would first succeed when checking for conflicting memory areas, and thus not try to find a better address, but later fail in MmMapViewOfSegment, which cannot create a memory area in system space (only possible with MmKernelAddressSpace / PsIdleProcess).
See issue #5680 for more details.
svn path=/trunk/; revision=54096
- Do not keep the process handle of a started process open
- when sending a STOP command to a process failed, then succeed and mark the process as stopped.
Fixes ghost processes, when terminating a service with taskmgr and allows to restart services that got terminated or possibly crashed.
svn path=/trunk/; revision=54095
- Handle a NULL buffer in RtlIpv4AddressToStringA as Windows does. Fixes crash in ntdll:rtl test
[CRT]
- Fail on qsort with 0 size. Fixes hang in ntdll:string test
svn path=/trunk/; revision=54093
* Use the gcc toolchain-provided STL.
* Use STLport only on MSVC builds.
[KSPROXY]
* Set imagebase, properly set its entry point, use set_cpp() instead of hardcoded definition/includes and add no exceptions/rtti gcc flags.
[GLU32]
* Use set_cpp() instead of the hardcoded target_link_libraries().
[PORTCLS]
* Use set_module_type() with proper entry point, instead of expanded macro calls, and get rid of hardcoded target_link_libraries().
* Commit Dedicated to Olaf Siejka (Caemyr).
svn path=/trunk/; revision=54082
Implement SystemTimeAdjustmentInformation case in NtSetSystemInformation. This silences the annoying messages caused by VBoxService
svn path=/trunk/; revision=54081
Implement support for hotkeys for run, show desktop and logoff
currently it also causes the startmenu to open, like with win-E, but this is a different bug
See issue #1512 for more details.
svn path=/trunk/; revision=54076
- Properly copy image name in TH32CreateSnapshotSectionInitialize. Fixes truncated file names in kernel32:toolhelp test
svn path=/trunk/; revision=54071
- Hackfix RtlGenRandom to be slightly less dependent on the current time, while staying very un-random.
- Should fix services problems (no joke) due to non-unique RPC UUIDs
See issue #6542 for more details.
svn path=/trunk/; revision=54069
- Use 64 byte bitmap for keystate instead of 256 elements array (like Windows)
- Fix GetAsyncKeyState lowest bit meaning. See MSDN
- Call low level keyboard hook before setting async key state for keys which distinguish between left and right hand. Fixes user32:GetKeyState apitest failure
- Improve IsHotKey and delete GetHotKey function to avoid code duplication
- Support WIN key in hotkeys code instead of keyboard.c
svn path=/trunk/; revision=54063
set CF, ZF and SF based on REGS structure, when doing int's. Fixes slowness during 2nd stage boot in QEMU. Dedicated to Samuel.
svn path=/trunk/; revision=54062
- SetupDiCreateDeviceInfoW: return the proper error code. Fixes a winetest.
- Implements parts of SetupDiRegisterDeviceInfo.
svn path=/trunk/; revision=54059
- Implement proper security cookie initialization code (same algorithm as in sdk/crt, however using as much data directly available as possible).
- Don't initialize security cookie as a 32bit value when 16bit security cookie was requested. This fixes msi-related crashes (or any other executables which has 16bit security cookie set).
See issue #6341 for more details.
svn path=/trunk/; revision=54057
* Temporarily go back to using the ret^H^H^H previous way of using debug info. Build will, for example, be slower, take double the disk space it needs right now, but...
* Introduce __ROS_DWARF__ for the meaning __ROS_CMAKE__ implied previously.
svn path=/trunk/; revision=54055
- Fix possible integer overflow in RtlpInitializeHeapSegment
- Simplify loop in RtlIsDosDeviceName_Ustr
- Make MonthLengths an array of UCHARs instead of ints
- Remove pointless loops in RtlTimeToTimeFields
- Fix MSVC warnings
svn path=/trunk/; revision=54054
* Use the previous registration mechanism. Fixes several shell32 regressions.
* Dedicated to Claudiu Mihail and Johannes Anderwald.
svn path=/trunk/; revision=54053
- Do not call UserSetCursor in MsqCleanupMessageQueue because it uses current thread queue instead of queue in cleanup (they can be different).
See issue #6530 for more details.
svn path=/trunk/; revision=54051
- Rename KTHREAD.DispatchHeader to Header
- Remove KeInitializeDispatcherHeader and initialize dispatcher headers depending on object type. Fixes all ntos:KeEvent and KeTimer kmtests.
svn path=/trunk/; revision=54048
- Fix symbolic link creation with a reference string
- Audio devices start successfully again (sound still seems to be regressed however)
svn path=/trunk/; revision=54040
- Do not fail in SetupDiInstallDevice if the inf file already exists. Fixes VirtualBox 4.1.x additions
See issue #6522 for more details.
svn path=/trunk/; revision=54037
Fix memory leaks
Fix broken if condition
Use delete[] operator for allocations made with new[] operator
Properly check for failed allocations by not throwing exception
svn path=/trunk/; revision=54034
- Save special values in F/Blink when Ex(f)InterlockedRemoveHeadList is called on checked build. Fixes all ntos:ExDoubleList kmtests
svn path=/trunk/; revision=54026
- Add a couple of tests for the messages sent when changing the z-oder of windows with DeferWindowPos (this test ignores owners z-order)
svn path=/trunk/; revision=54025
- remove all previous SetActiveWindow tests as they were completely wrong. This is because changing the z-order of the owner windows is done by SetWindowPos as tested in the same test suite. However SetActiveWindow calls SetWindowPos without SWP_NOOWNERZORDER parameter which leads to changing the zorder of owner windows
svn path=/trunk/; revision=54023
- Add some more tests for SetWindowPos with or without SWP_NOOWNERZORDER. This clears the misunderstanding on changing owner window z-order once and for all
svn path=/trunk/; revision=54022
Fix handling of RLE compressed bitmaps in NtGdiStretchDIBitsInternal, by using similar code as in NtGdiSetDIBitsToDeviceInternal calling GreCreateBitmapEx instead of NtGdiCreateDIBitmapInternal.
Fixes bug 6168.
svn path=/trunk/; revision=54019
- Post-r52687 implementation of RtlIsDosDeviceName_U is more advanced than pre-r52687 one and covers more corner cases which may not be tested by existing winetests.
- Thus bring back the post-r52687 and also introduce fixes for the following bugs in it:
* Incorrect extension/space-at-the-end truncation
* Incorrect encoding of returned offset/length pair
* Incorrect condition for for-loop exit
* Incorrect calculation of the return length in case of COMx/AUX devices (it should not include the terminating 0 character)
- After those trivial fixes (mostly oneliners and typos), the function works as expected and passes all ntdll_winetest path tests. Sorry for any inconviniences!
svn path=/trunk/; revision=54018
- Implement support for scatter/gather DMA
- Tested with Intel's NDIS 5.1 PRO/1000 driver (but more testing would be helpful to weed out bugs)
[NDIS]
- Report NDIS 5.1 now since S/G DMA is supported
- Stop warning when S/G DMA is used
svn path=/trunk/; revision=54016
- Do not treat all extended keys as power keys in HandlePowerKeys. Fixes WM_KEYDOWN not being send right after extended key has been pressed.
See issue #3212 for more details.
svn path=/trunk/; revision=54007
- Rewrite keyboard input processing to match windows behavior
- Deduplicate code, add comments, mark functions static
- Support keyboard LEDs in SendInput API
- Set last error in more places
- Make keyboard layouts compatible with Windows
- Add some defines from keyboard layouts to kbd.h and use names from MS file
- Move kbd.h from ntoskrnl to NDK as it has nothing to do with kernel
- More fixes which I don't remember
- Please test CAPSLOCK key with AltGr characters (RALT+X) for layouts which support it. If something fails, report it to me on IRC
svn path=/trunk/; revision=54006
- Restore a ros-diff lost during wine sync. Should fix kernel32:volume crash
- Better mark ros diffs in testlist.c and also bring it somewhat into sync
svn path=/trunk/; revision=54003
- netid, vmwinst romanian translation, see bug number:6531;
- resource updates and corrections in multiple romanian translations, see bug number: 6532;
- Patches by Stefan Fulea;
[ACLUI]
- convert module to ReactOS resource style (separate lang directory);
svn path=/trunk/; revision=53998
Improve the new heap to use a freelist, which boosts allocation performance by a factor of 7. Its now even slightly faster then bget.
svn path=/trunk/; revision=53989
Fix a bug, when the memory for a file cluster chain record was freed from the heap although it was still in use. Instead free the buffer when the file is closed.
svn path=/trunk/; revision=53982
- Do not access an object after destruction.
- Remove the remaining mentions of a function which Sir Richard defeated a year ago.
svn path=/trunk/; revision=53978
- Add a flag to distinguish Ansi and Unicode message tables. Keep them Ansi for now pending further testing
- Apply this to MSVC build as well. Fixes ntoskrnl messages incorrectly being Unicode
svn path=/trunk/; revision=53971
- Clean up some headers
- Add DPRINT_HEAP debug channel mask
- Make ExAllocatePool a function instead of a macro
- Add a cleanup function to the new heap code
svn path=/trunk/; revision=53967
- Implement CM_Get_DevNode_Custom_Property[_Ex]A/W.
- Fix the return value of CM_Open_Class_Key_ExA.
- Handle REG_MULTI_SZ values properly in CM_Get_DevNode_Registry_Property_ExA.
svn path=/trunk/; revision=53965
- Correctly handle an arbitrarily large sockaddr in WSPConnect. Fixes mIRC 7.1x crash on connect
See issue #6005 for more details.
svn path=/trunk/; revision=53962
* Work around an issue with clang and __debugbreak(). They mark __debugbreak as built-in (so we can't use ours) but it's not supported yet (so if we use nothing we get an error).
svn path=/trunk/; revision=53958
Implement new heap code. The code is much less complex than bget and allows to use allocation tags and releasing unused memory. Its not yet active.
svn path=/trunk/; revision=53956
- Pass the module list head to a number of pe loader functions instead of the loader block.
- use static for some functions
- remove unneeded prototypes
svn path=/trunk/; revision=53947
- Fix RECTL_bPointInRect to work like PtInRect
- Use the fixed RECTL_bPointInRect in DIB_XXBPP_FloodFillSolid to fix a bugchek
See issue #6437 for more details.
svn path=/trunk/; revision=53946
* Reduce the scope of a variable.
* Comment out some unused code.
* Initialize some member variables in the respective constructors.
svn path=/trunk/; revision=53944
- use RtlStringCbPrintfA instead of manually calculating required length and checking if the buffer is large enough
- Use %wZ as format specifier for a UNICODE_STRING, instead of using %S and making assumptions about zero termination.
- Don't "while (TRUE);" on buffer overflow!
svn path=/trunk/; revision=53942
Fix remaining issues, bootsector compiles with GAS and ML. Sadly GAs sucks a bit and does neither allow to specify byte offsets, nor does it chose this itself to save space. As a result the code gets several bytes larger and I had to trim the messages even more.
svn path=/trunk/; revision=53938
- Somehow our shell now expects the icon of My Computer to be the resource icon %SystemRoot%\Explorer.exe,0. Change the resource id of My computer icon to 0 to fix the wrong icon on the desktop
svn path=/trunk/; revision=53927
* Add a new Starfield screensaver. By Carlo Bramini (carlo DOT bramix AT libero DOT it).
* Adapted the build system files to get it to compile with current trunk.
See issue #6270 for more details.
svn path=/trunk/; revision=53920
use a dll instead of an exe for the freeldr_pe and setupldr_pe targets and add a spec file to export scsiport stuff. This is the first step in cleaning up the scsiport import hack in freeldr
svn path=/trunk/; revision=53914
- Enable all interrupts at the same time to avoid a condition where the controller stops responding to KBD_READ_MODE on some systems
- See bug #6271 for details
svn path=/trunk/; revision=53911
- Implement MiInitializePfnAndMakePteValid() which is basically MiInitializePfn() combined with making PTE valid and some other changes.
- Don't initialize PTE and pass valid PTE for initialization to MiInitializePfn() when allocating a page in special pool, this is wrong. Instead call a function designed to do this - MiInitializePfnAndMakePteValid(). It performs correct checks and PFN reference counting.
svn path=/trunk/; revision=53909
Remove DTEs for ntbootdd.sys and freeldr.sys after they are not used any longer. freeldr one will even point to invalid stack data. It only works currently, because the LoadOrderListHead is reset later. More fixes are coming.
svn path=/trunk/; revision=53908
- properly cleanup when tests are completed
- Add some more tests for SetActiveWindow that prove that bringing owner/owned windows to front is done by SetActiveWindow and not by SetWindowPos
svn path=/trunk/; revision=53905
- Implement SHPropStgCreate, SHPropStgReadMultiple, SHPropStgWriteMultiple based on wine
- This should fix the shellole tests
svn path=/trunk/; revision=53903
- implement helper functions MiIsMemoryTypeFree and MiIsMemoryTypeInvisible
- use symbolic names instead of hex values
- Don't ASSERT on mising large page support, just warn
- add some more mm macros for amd64
svn path=/trunk/; revision=53902
- Report the device's lockable status in the device capabilities request
- Only report the device as disabled if it is both not enabled and not functional
svn path=/trunk/; revision=53899
- Move some disk related stuff that is unrelated to the registry data into a new file, hwdisk.c
- Don't get the disk count from the size value of a structure that was previously calculated from the disk count, but instead save it in a global variable.
- Initialize certain data in a better place
svn path=/trunk/; revision=53896
- Update ACPICA from 20091214 to 20110922
- Rewrite a large portion of the OSL code
- Perform a full initialization of ACPI objects (don't skip devices and events)
- Please retest ACPI bugs after this commit
svn path=/trunk/; revision=53895
- Add a hack to skip calling methods of the HPET object because it causes a hardlock on VMWare
- Attempt #2 at fixing VMware
svn path=/trunk/; revision=53892
- Do not force the hard error dialog to appear. As a result, calling SetErrorMode now really prevents the hard error dialog from appearing.
svn path=/trunk/; revision=53889
- Minor fixes to ScmAssignNewTag
- Correctly handle an invalid parameter case in RCreateServiceW. Fixes second stage boot under certain circumstances.
svn path=/trunk/; revision=53886
- Remove pointless NTKRNLAPI
- remove unneeded include
- delete deprecated file amd64/loader.c
- add an #ifef __REACTOS__ around public header includes to make it possible to use an alternative header set from ntsup.h
svn path=/trunk/; revision=53885
- The width parameter in AcpiOsReadPciConfiguration and AcpiOsWritePciConfiguration was in bits but we were treating it as a width in bytes
- This caused overreads, memory corruption, and crashes when these functions were called (VMWare was particularly picky about bad accesses to the PCI configuration space)
- A hack was (unknowingly) added which prevented some crashes but had a side-effect of causing the partial disruption of ACPI's PCI configuration space accesses while the others that went through wrote bad data to the PCI config space or corrupted kernel memory
svn path=/trunk/; revision=53880
- Modify some memory types to match what windows uses
- Modify PcMemGetBiosMemoryMap, so that is crops free memory to page alignment, while extending non-free memory to page alignment. This way we don't loose small firmware descriptors like the one for the extended bios data area enumerated by the bios. Fixes a warning from win 2003 about not owned memory.
- Modify MempAddMemoryBlock to map LoaderFirmwarePermanent into kernel space like ntldr, but don't map page 0! ntldr doesn't do that either. Also map LoaderXIPRom.
- After generating memory descriptors from the page lookup table, add any descriptor from the bios memory table, that is higher than MmHigestPhysical address, so we don't loose high rom descriptors. Do not map beyond LoaderPagesSpanned.
svn path=/trunk/; revision=53878
- copy ebp into REGS structure on Int386
- Write pointer and size of extended bios data area in physical page 0 location 0x740 and pointers to vga rom fonts in location 0x700. This is required by windows 2003's videoprt. Now freeldr is more Windows 2003 ready then ever!
svn path=/trunk/; revision=53876
- Fix auto complete init method based on Wine. Autocomplete shell tests should no longer crash now.
- Brought to you by Claudiu Mihail.
svn path=/trunk/; revision=53871
* Implement CDefView::SetCurrentViewMode based on Wine, along with some other bugfixes. Shell32 shlview winetest doesn't crash anymore as a result.
* Brought to you by Claudiu Mihail.
svn path=/trunk/; revision=53869
- Introduce two MM_EMPTY_LIST definitions - one for PTE as a whole and one for MMPTE_LIST NextEntry field only and move them to the actual architecture specific headers.
- Implement special pool. This is a special (Capt. Obvious!) kernel pool which is used to catch overruns or underruns, freed memory access and perform additional validation for paged/nonpaged allocations access.
svn path=/trunk/; revision=53868
- Do not subclass user32 controls when themes are disabled.This is a temporary hack bacause it turns out that subclassing causes problems. When it is fixed this will be reverted. This change means that in order to make themes work properly, the user has to reboot after enabling.
svn path=/trunk/; revision=53866
- Fix memory type for hal and referenced dlls
- Don't allocate the LPB, MADs and all the other stuff from the heap, instead allocate one block of memory for the LPB, extension, etc and strings (fits into one page) and another page for the MADs limit to 200 MADs, 1024 is way over the top)
- Map only those pages into kernel mode, that the kernel will need, this way we don't need to unmap anything later.
svn path=/trunk/; revision=53864
- Fix a typo in HalpAllocPhysicalMemory, that caused the function to remove MADs that still had pages rather than removing those who are empty. Fixes an assertion on VMWare.
Kudos go to Kamil for tracking it down.
- Fix another bug in HalpAllocPhysicalMemory, where the size of the newly allocated MAD was set to the alignment value instead of the original MAD, this lead to conflicting MADs and possible reuse of hal memory by the kernel. This seems to fix a bugcheck 0x19 with halacpi.
svn path=/trunk/; revision=53861
- Get rid of MmAllocateMemory, freeing about 1MB of low physical memory for the kernel
- verify that MmMarkPagesInLookupTable is not called with invalid page regions
- add maximum allocation to heap statistics
<long explanation>
There was a function called MmAllocateMemory, what was commented as "// Temporary forwarder..." since January 2008. This function allocated one page of memory and marked it as LoaderOsLoaderHeap. This function was used by the fat filesystem code (and linuxboot.c) which allocated and freed memory in small chunks all the time. Since MmFreeMemory() is not implemented at all (obviously someone removed it) we were allocating one full page for allocations as small as 8 bytes and never free them. This accumulated to a total of 240 pages, almost 1MB, split into into 14 chunks. This memory was never freed by the kernel (the kernel keeps the loader heap memory for some reason) and fragmented the low memory region.
Remove MmAllocateMemory completely and replace references in the fat code with MmHeapAlloc. The maximum heap usage after this was 184 KB, heap size is 4MB.
</long explanation>
svn path=/trunk/; revision=53857
- In TrackMouseEvent tests, use mouse_event function to move the mouse cause this causes low level hooks to be called and SetCursorPos doesn't
- Test low level mouse hooks for TrackMouseEvent
- Test ways to block WM_MOUSEHOVER and WM_MOUSELEAVE. This is useful to detect where each message is generated
- clean up the code a bit by adding FLUSH_MESSAGES macro that checks for expected message queue status and flushes the message queue
svn path=/trunk/; revision=53855
- Improve tracing
- Fix memory corruption caused by marking pages outside of page lookup table (when table is allocated only free pages was counted but MmMarkPagesInLookupTable was used for all types of memory like firmware). Fixes bugcheck in ACPI HAL which happened for me
svn path=/trunk/; revision=53848
- Undo disabling of systemcalls, which was disabled in r53646 for testing purpose.
- Sillence "Support for SYSENTER detected." debug print, since its the normal case, only keep the disabled and not present cases.
- Silence "xxx support detected but not yet taken advantage of" debug prints. CPU features are already reported and there is no point in printing every single feature we do not support.
- Invert if/else in KiRestoreFastSyscallReturnState
svn path=/trunk/; revision=53846
Add FREELDR_MEMORY_DESCRIPTOR structure using TYPE_OF_MEMORY instead of
MEMORY_TYPE. Use it in place of MEMORY_DESCRIPTOR. Undo r53837 changes.
svn path=/trunk/; revision=53841
* Add _dbg versions of freeldr_pe and setupldr_pe per Dmitry's request. They retain debug info, and they're compiled on-demand (not included in the normal build).
svn path=/trunk/; revision=53840
Fix wrong behavior of the "native applications startup at boot time" feature in SMSS.
Modifications:
- Added missing buffer allocation checks.
- Check for presence of default path.
- Minor clean-ups.
- Fix bad indentation/coding style.
Original patch by Hermès BÉLUSCA.
See issue #6180 for more details.
svn path=/trunk/; revision=53839
- Cleanup amd64 mm.h header
- Call MiScanMemoryDescriptors earlier, since it doesn't depend on anything except the loader block.
- MiComputeColorInformation() and the calculation of the pfn database size are portable, so move them out of MiInitMachineDependent
svn path=/trunk/; revision=53838
- Implement logging post messages or winevents
- For SetActiveWindow test, also test the parameters of WM_WINDOWPOSCHANGING and WM_WINDOWPOSCHANGED messages
- For TrackMouseEvent test, also test the WM_SYSTIMER that is possible to be logged now
svn path=/trunk/; revision=53830
The following modifications were made:
- Removed the help texts because they were copied from Windows Diskpart.
- Used a single string resource per help text.
- Used only Unicode strings and Unicode string functions.
- Replaced the giant if-statement in the command dispatcher in interpreter.c by a command-table and a little loop.
- Improved the command-line parser. Argument count and argument vector are passed to the command and help functions for easy evaluation.
- Moved the help function for commands into the command specific file.
TODO:
- Implement all commands as the utility is only an empty frame.
- Add help texts.
- Translate help texts.
See issue #6402 for more details.
svn path=/trunk/; revision=53826
- Session space address layout is architecture specific, move its initialization into MiInitializeSesseionSpaceLayout() in architecture specific file
- Use dedicated constants for setting MmSystemRangeStart, MmUserProbeAddress and MmHighestUserAddress isnetad of making assumptions.
svn path=/trunk/; revision=53825
Add MiScanMemoryDescriptors, that combines the work of MiPagesInLoaderBlock, the loop in MmInitializeMemoryLimits and the loop in MiInitMachineDependent.
svn path=/trunk/; revision=53822
* Bring-in some idl changes from Wine. These changes (and the related changes below) are to allow us to sync shell32 tests to Wine 1.3.29.
[UUID]
* CLSID_TaskbarList is no longer needed in here.
[ACTXPRXY]
* Sync with Wine 1.3.29.
svn path=/trunk/; revision=53820
Converting MmLargeStackSize from registry setting to byte value is portable code, move it out of MiInitMachineDependent
svn path=/trunk/; revision=53819
- Use TYPE_OF_MEMORY from the beginning instead of MEMORY_TYPE and later converting it.
- Disable tracing, accidentally comitted in previous revision
svn path=/trunk/; revision=53818
Fix some bugs in the code that deals with memory descriptors:
ArcGetMemoryDescriptor was enumerating the bios generated entries first and then the static entries. The latter were conflicting with the former and took precedence when initializing the page lookup table. The problem was, that MmFindLocationForPageLookupTable would use the highest range of pages available. If that range would be conflicting with a following static descriptor, the static would have been ignored. It only worked because on x86 the highest bios descriptor has enough free pages, before it conflicts with a static entry (page 0xfff, marked as unusable), so that the page lookup table could be created.
MmGetAddressablePageCountIncludingHoles enumerated all memory descriptors to find MmLowestPhysicalPage, but it was only counting FreeMemory, thus skipping all other memory ranges. This only worked, due to the previous bug, so that the bios pages shown first took precedence over the following static descriptors. Without the former bug MmLowestPhysicalPage would be 0x100 which would tigger in the next bug:
MmAreMemoryPagesAvailable took the passed address range and looked up all pages in the lookup table to see whether all are free. Now the function didn't check, whether the passed address was below MmLowestPhysicalPage and in case it was, happily accessed the memory below the lookup table. This would result in hal being loaded at 0x40000 overwriting the loader itself.
This is all fixe by implementing a new way of creating the memory map. First there is a static memory map, which has enough free entries to add dynamic ranges. Then AddMemoryDescriptor will allow you to add a new range,
while the already existing ranges will take precedence and the new ranges will be properly split and inserted, so that the resulting map does not contain any overlapping regions and is sorted from low to high pages. The static memory descriptor exists in the architecture specific file.
The code that enumerates the bios information now uses this function to add the ranges on top of the static descriptor.
More cleanup work to follow.
svn path=/trunk/; revision=53816
- Move shared code used to log window messages in helper.c
- Implement logging hooks
- Fix comparing the logged messages to show the correct line of the test
- Add tests for hooks in TrackMouseEvent test
svn path=/trunk/; revision=53812
- Fix several typos. Patch by Elton Chung (elton328 at gmail dot com). Slight compatibility fix by me
- Add missing NTSYSAPI to some externs
- remove Se*DefaultDacl externs from ntifs, as they don't belong there, and are already correctly found in ndk/setypes.h
svn path=/trunk/; revision=53809
- Revert part of r53788 (MAXIMUM_IDTVECTOR is a valid vector number)
- The buffer overrun in the IDT code was already fixed by r53420
svn path=/trunk/; revision=53806
- Multimodule translation and UTF-8 conversion to Portugese Brazilian. Patch by mkbu95. Minor fixes by me;
See issue #6467 for more details.
svn path=/trunk/; revision=53805
- Fix a bug, where the bios memory map could have a bogus entry: according to R.B.'s Interrupt list, in some bioses the function for enumerating the memory can return ebx != 0 for the last descriptor, and fail in the following call returning CF = 1. In that case we were previously counting that last descriptor as well, but its content wouldn't describe a valid memory region.
- Use a static buffer for the bios memory map, stored in the achitecture specific file and return a pointer to that variable from PcMemGetMemoryMap. Reuse the static map from DetectAcpiBios instead of enumerating again. On arm, we don't even need a buffer, we return the pointer from the arm block.
svn path=/trunk/; revision=53803
Patch by Brian Palmer:
Fix problems that could occur, when the BIOS returns memory regions that are not page aligned, by fixing up the bios memory map before using it.
svn path=/trunk/; revision=53797
- Implement the LSA object database.
- Implement the information classes PolicyPrimaryDomainInformation and PolicyAccountDomainInformation of LsarSetInformationPolicy().
svn path=/trunk/; revision=53783
- Keep APC delivery disabled while holding a Heap Lock, not just while acquiring/releasing it (Dedicated to Timo and Ged).
- Store the Heap Lock in non-paged pool, the only place where executive resources belong (Dedicated to patient people).
[RTL]
- Restructure/simplify/correct Heap and Heap Segment initialization (partially to cope with the changed Heap Lock interface).
- Restrict the location of Heap Segment headers to the base address of the Heap Segment (which frees up a whopping 60 bytes per Heap!).
- Cater for acquiring the Heap Lock exclusively or shared (the latter is only available in kernel-mode); only exclusive locking is used for now.
- Use a plain critical section to guard the Process Heap list, no reason to disguise it as a Heap Lock (saves us a handful more bytes).
svn path=/trunk/; revision=53761
Fix CsrProbeForRead and CsrProbeForWrite to actually access the memory, by using volatile. The compiler optimized the access away previously.
svn path=/trunk/; revision=53760
- Fix a number of MSVC/amd64 warnings/problems
- Remove code duplication in RtlCharToInteger and RtlUnicodeStringToInteger
- Fixup too large string length in RtlInitAnsiString, RtlInitUnicodeString
- Use symbolic constants instead of hardcoded values
- Fix overflow check in RtlInitUnicodeStringEx
svn path=/trunk/; revision=53746
- Merge changes from the themes branch. These changes will introduce test regressions for class functions and for TrackMouseEvent. The changes in the class functions are needed in order to let comctl32 register its own versions of built in controls. This is a hack and will be removed when we properly support class versions. The second change that causes test regressions is in TrackMouseEvent and it is needed in order to fix WM_MOUSELEAVE messages that were completely broken and not tested at all by wine's test suite.
svn path=/trunk/; revision=53745
- Make Vector parameter in HalEnableSystemInterrupt, HalDisableSystemInterrupt and HalBeginSystemInterrupt an ULONG, not an UCHAR
[NDK]
- 64bit fixes for HANDLE_TABLE, KPROCESS, SECTION_IMAGE_INFORMATION, MMADDRESS_LIST, MMVAD_FLAGS, MMVAD, MMVAD_LONG, MMVAD_SHORT, MEMORY_DESCRIPTOR, MEMORY_ALLOCATION_DESCRIPTOR, LdrVerifyMappedImageMatchesChecksum
- KDPC_DATA::DpcQueueDepth is signed on amd64, unsigned on x86
[NTOSKRNL]
- Fix hundreds of MSVC and amd64 warnings
- add a pragma message to FstubFixupEfiPartition, since it looks broken
- Move portable Ke constants from <arch>/cpu.c to krnlinit.c
- Fixed a bug in amd64 KiGeneralProtectionFaultHandler
svn path=/trunk/; revision=53734
Fix the overly complicated and buggy implementation of RtlIsDosDeviceName_Ustr based on the simple pre-r52687 implementation of RtlIsDosDeviceName_U. This fixes 14(!) wine tests.
svn path=/trunk/; revision=53726
RtlGenerate8dot3Name: Ignore spaces in long file names instead of converting them to underscores.
See issue #6385 for more details.
svn path=/trunk/; revision=53725
- Don't check for leading "." char in a filename, because it's valid in case of LFN. The driver is already quite broken, so it can't get more broken if this assert is commented out.
Fixes assertion in "kernel32_winetest module".
svn path=/trunk/; revision=53722
- Fix KiGeneralProtectionFault
- Don't bugcheck in KiUnexpectedInterrupt
- Remove debugprint in KiEndInterrupt, and instead assert, its called from the clock interrupt, since thats the only instance we handle right now
svn path=/trunk/; revision=53720
- add common amd64 asm trap macros
- fix amd64 version of HackEoi
- send EOI in amd64 trap exit code, so we avoid using HackEoi
- modify HalpClockInterruptHandler for better portability
- fix amd64 TscCalibrationISR
svn path=/trunk/; revision=53719
- get version.h and buildno.h out of precompiled headers, to avoid rebuilding the whole module each time we reconfigure the build
svn path=/trunk/; revision=53716
- Fix some stuff that gcc complains about. It doesn't like to use static data in non-static inline functions for some unknown reason.
- Implement support for level triggered interrupts, when using lazy irql
svn path=/trunk/; revision=53711
- Improve List_PaToVa, making the code cleaner
- Dereference LoaderBlock->Extension only after WinLdrSetProcessorContext, since its in kernel address space
svn path=/trunk/; revision=53710
- Move some shared constants (HYPERSPACE_BASE, MM_HAL_VA_START, APIC_BASE) to NDK to avoid duplicating them, also fix wrong values for amd64
- Map the hal address space in amd64 freeldr
- Fix use of uninitialized variable (HighPart of PHYSICAL_ADDRESS) in HalpAcpiCopyBiosTable
- Fix some MSVC and amd64 warnings
svn path=/trunk/; revision=53709
* Merge r53700. Point 3 in bug 6481 was fixed by fixing the parser (at line 414) to use only ',' as the delimiter, to match Windows behavior.
svn path=/trunk/; revision=53702
Implement lazy irql for APIC. This is useful for VMs, since APIC usually has high overhead due to the need of invoking the hypervisor on every irql raise and lower. With lazy irql we avoid that until absolutely neccessary. Note that we misuse the PCR's IRR field to save the current hardware irql. Its a huge performance boost (some parts take half the time), making APIC performance close to PIC performance on VBox. This is something that Windows doesn't have :)
svn path=/trunk/; revision=53695
- Create a volatile device tree during 1st stage so the new HAL that's loaded in 2nd stage can do proper enumeration without phantom devices from the old HAL
svn path=/trunk/; revision=53694
- Fix incorrect loop boundaries in the ARM version of MmCreateProcessAddressSpace. Spotted by Vincenzo Cotugno and cppcheck
- Correct outdated wineimport.cmd URL in README.WINE
- Assorted spelling and readability fixes
svn path=/trunk/; revision=53688
- Add support for guarded allocations, that can be used to detect buffer overruns
- Add GuardedMemory example test
- Add MmSection test with some basic parameter checks for MmCreateSection
- Fix some x64 build problems
- Update readme
svn path=/trunk/; revision=53687
- Add optional asynchronous io code for service control pipes. This is disabled by default due to bugs in NPFS.
- Read service pipe timeout value from the registry.
svn path=/trunk/; revision=53686
- End the interrupt in HalpApcInterruptHandler and HalpDispatchInterruptHandler before calling the kernel, otherwise we would be stuck at high processor irql
- Improve HalEnableSystemInterrupt
- disable interrupts in HalpInitializeClock
=> APIC hal boots to desktop!
svn path=/trunk/; revision=53685
The handlers for unexpected interrupts that we generate from asm macros use a push instruction with an 8 bit operand to push the interrupt number on the stack. Now the 8 bit push has the unfortunate behaviour of sign extending to 32 bit. But since the assembler is smart (or at least thinks so) it will not generate such a push instruction for values larger than 0x7f. This resulted in 7 bytes long stubs in the range of 0x30 .. 0x7f and 10 bytes long stubs after that. To fix this 128 is substracted from the value and later readded in the common code path. A second issue arose, because the assembler would start to use 8 bit relative short jumps as soon as the target was less than 127 bytes away, which happened for some of the higher interrupts, which then had a smaller stub then the others. Fix this by moving the stubs up in the code further away from the target label, so that always 32bit relative jumps will be used. These problems didn't show up so far, since we used the PIC and interrupts in the range of 0x30 .. 0x40 and the code that relied on the stubs all having the same length was working well. This changes with an APIC based hal, where interrupts of much higher number are likely to be used.
svn path=/trunk/; revision=53679
- Use a global variable HalpBuildType instead of a define when checking the BUILD (we don't recompile this code for SMP hal)
- Move registering IRQ2 out of the generic init code into HalInitializePICs
- Modify legacy HalpGetRootInterruptVector to use HalpIrqToVector and HalpVectorToIrql so it works with APIC, too
svn path=/trunk/; revision=53675
- Implement HalpApcInterruptHandler, fix HalpDispatchInterruptHandler
- Insert a read to the APIC version register in ApicGetCurrentIrql(). This fixes inconsistencies between the internal APIC state and register reads/writes.
- Remove old hacks in ApicGetCurrentIrql() and ApicSetCurrentIrql
- Fix HalpVectorToIrql()
- read/write cr8 for the value of the TPR on amd64 builds
- Fix amd64 build
svn path=/trunk/; revision=53673
- Skip the remaining tests that result in assertions
- Hide KeProcessor test, as it takes very long
svn path=/branches/GSoC_2011/KMTestSuite/; revision=53669
Add a hack to work around a bug in VBox: The APIC emulation requires an iret instruction following closely after writing the EOI register. Since we return from kernel mode traps with a jmp (yes you can return from an interrupt with a jmp!) the EOI is never triggered, making VBox believe we are still serving the interrupt and keeping the PPR on high level, preventing following interrupts. A small asm stub now does the work of both writing the EOI and doing an iret.
svn path=/trunk/; revision=53665
- Add a hack to ApicGetCurrentIrql and ApicSetCurrentIrql to work around a VBox bug
- call KeSetTimeIncrement after initializing the rtc clock
- calculate TSC frequency from the samples
- Fix a bug in KeStallExecutionProcessor
svn path=/trunk/; revision=53658
- Rename gQueueKeyStateTable to gKeyStateTable as it has nothing to do with msg queues. They have their own table
- Use KS_*_BIT defines instead of magic values
svn path=/trunk/; revision=53655
* Reintegrate the c++ shell32 branch. Exemplary team-work.. kudos !
* Better code quality, more tests run with less failures... and more.
* Dedicated to everyone who helped ;)
svn path=/trunk/; revision=53653
[FORMATTING]
- Second wave of formatting. At least now we won't have any tabs and spaces mixed in.
svn path=/branches/shell32_new-bringup/; revision=53651
- Fix bug 6336 of shell32_new. The bug was caused by not properly verifying the return value of ShellExecuteExW.
[FORMATTING]
- First wave of accidental formatting (sorry again about mixing code and formatting heh).
svn path=/branches/shell32_new-bringup/; revision=53650
- Start a new device manager to test the reactos ATL code.
- Also, the old one sucks big time and was quickly hacked together.
svn path=/trunk/; revision=53648
- Rewrite clipboard to match Windows implementation
- Clipboard uses window stations instead of global variables
- Data is saved in clipboard data user objects
- Memory is no longer leaked when winsta is destroyed
- Data is synthesized on demand
- Make internal functions static
- Fix possible memory corruption in IntGetAtomName
- More winetests are passed
svn path=/trunk/; revision=53644
- Simplify accelerators code a bit
- Set last error in NtUserCopyAcceleratorTable
- Fix user32:resource winetest regression
svn path=/trunk/; revision=53642
- Make sure the ParseDisplayName functions make the ppidle parameter NULL whenever that's possible. Scores a wine test. Spotted by Victor Martinez.
svn path=/branches/shell32_new-bringup/; revision=53641
- Remove kmtest_old and kmtloader as all their functionality is included in the new test suite
svn path=/branches/GSoC_2011/KMTestSuite/; revision=53639
- Fix I/O APIC register access
- set APIC logical id based on Cpu (currently flat model with up to 8 cpus supported)
- In HalpInitializeTsc, setup the RTC clock, since the timer is initialized later
- in the TSC calibration ISR, send EOI and read RTC register C to get the next interrupt
svn path=/trunk/; revision=53634
- Initialize uninitialized variables.
- Add code to guard against potential NULL pointer dereferencing.
Thanks to Amine Khaldi for pointing out all these.
svn path=/branches/shell32_new-bringup/; revision=53633
- Don't use debug prints before the debugger is even initialized
- acquire cmos spinlock when accessing the cmos registers
- Fix amd64 build
svn path=/trunk/; revision=53631
- Take NetworkAddr into account when a named pipe client is opened.
- Return RPC_S_SERVER_UNAVAILABLE when an attempt to create a named pipe client fails with an ERROR_BAD_NETPATH error.
This fixes the first test failure in the advapi32 service winetest.
svn path=/trunk/; revision=53630
- More uninitialized value fixing. Again Amine Khaldi spotting them with his magical powers.
- I took the liberty to convert tabs I found to spaces. Normally I shouldn't mix formatting with coding changes but I forgot. Sorry.
svn path=/branches/shell32_new-bringup/; revision=53629
- Attempt at fixing bug 6315 (it doesn't completely work yet). Original patch by Edijs Kolesnikovics.
svn path=/branches/shell32_new-bringup/; revision=53628
- give apic hal its own halinit_apic
- give minihal an own halinit_mini.c containing function stubs
- move HalInitializeProcessor to halinit.c and add hal specific function HalpInitProcessor
- get rid of all the #ifndef _MINIHAL_ in halinit.c and move the file to lib_hal_generic
- improve the hal version/build check
- Update spec file for amd64
svn path=/trunk/; revision=53625
- Add a test case for SetActiveWindow that shows its behavior related to owner windows. It should point the bug observed in bug 1239
svn path=/trunk/; revision=53621
Move the internal headers below the redefinition of types, because gcc instantiates inline function where they are defined, using the original types. Fixes build.
svn path=/trunk/; revision=53620
Make KeNumberProcessors internally an UCHAR and move it into krnlinit.c instead of having it once per architecture. Fixes a bunch of MSVC warnings.
svn path=/trunk/; revision=53618
- No code changes. Make the code more readable in preparation for investigating some failing wine tests.
svn path=/branches/shell32_new-bringup/; revision=53617
- move v86 code out of trap.S into v86.S (APIC uses a different trap entry file), compile it directly in the dll instead of a lib to be able to silence linker warnings
- Group files in HAL_PIC_SOURCE and HAL_APIC_SOURCE, you just need to replace PIC with APIC to build the hal with APIC support (we can later use it to build the other hals: halapic, halaacpi, halmacpi, halmps)
- use apic on amd64 builds
- give halacpi its own resource file
- silence MSVC linker warnings
svn path=/trunk/; revision=53616
- Fix max key name length, max value name length, max value data length properties of registry key not being set correctly
- Fixes#6042 - regedit on LiveCD shows keys and values now
svn path=/trunk/; revision=53614
Start implementing APIC support, which is needed for both SMP and x64. It will use the local APIC + I/O APIC for interrupt control, the RTC instead of the PIT for the timer interrupt (PIT doesn't always work with I/O APIC), the APIC timer for profiling and finally the TSC for the performance counter and KeStallExecutionProcessor.
The code is incomplete and doesn't work yet
svn path=/trunk/; revision=53611
- Add missing initialization. Fixes the problem where explorer fails to show the child window at the first attempt to open a folder
svn path=/branches/shell32_new-bringup/; revision=53610
- Any attempt to open or create a file on a remote machine will return STATUS_BAD_NETWORK_PATH. This means the server is not available. Trying to create \\testserver\testshare\testdir\test.txt fails just like on Windows XP.
- Add service key and values for the MUP driver. Add the driver to the File System group because adding it to the Network group, like on Windows XP, does not load the driver.
svn path=/trunk/; revision=53609
- Use functions instead of macros for irq<->vector<->irql translation, this will be neccessary, once apic support is there
- Disable mp hal in rbuild for now, its broken anyway
svn path=/trunk/; revision=53600
- Remove unnecessary assignments. Spotted by PVS and Dmitry Chapyshev. This may change the behaviour of that codepath, so test results are going to be observed.
svn path=/trunk/; revision=53596
- Fixed over check conditions for deleting regions in set window region.
- Disabled offsets with window region. Fixes wine tests..
svn path=/trunk/; revision=53595
RStartServiceA: Convert the ANSI start arguments vector to a UNICODE vector and pass it to ScmStartService. Calls to StartServiceA can now be used to pass arguments to a starting service.
svn path=/trunk/; revision=53579
- add flags to ASM as well instead of hacking the compile rule
- remove -fPIC flag from ASM compile flags, (it's a cmake bug)
- enable languages before including compiler specific files.
svn path=/trunk/; revision=53567
- Add ObReference to build. Was missing for some reason
- Use the new CMake macros in preparation for PCH
- Use kmtest_.exe as a file name for testcd to please rosautotest. Perhaps there's a better solution later
svn path=/branches/GSoC_2011/KMTestSuite/; revision=53562
- Replace add_linkerflag and add_compiler_flags with their new equivalents add_target_link_flags/add_compile_flags
- Remove those now obsolete macros
- Use the new set_module_type to get rid of evil hacks for drivers without entry point (wmilib, drmk)
svn path=/trunk/; revision=53558
- merge set_module_type function into the global CMakeMacros.cmake file, leaving compiler particular bits to set_module_type_toolchain function.
- Add the ENTRYPOINT and BASEADDRESS options to this function. Not used yet
svn path=/trunk/; revision=53553
- merge r53215 and r53221 from themes branch
- Fixes drawing apps that use SetWindowRgn and also fixes a gdi object leak
See issue #1239 for more details.
svn path=/trunk/; revision=53548
- Add add_target_property macro and add wrapper macros to easily set important properties
- This makes macros more consistent, follows CMake property names, favors per-target instead of per-directory property settings and favors adding to properties instead of replacing them
- Convert the global settings to use the new add_compile_flags and add_target_link_flags - no functional change
- Fix add_importlibs to add to compile definitions instead of replacing them
- Big thanks to Amine
svn path=/trunk/; revision=53545
- base/application UTF-8 conversion patch by Elton Chung;
- minor corrections by me;
See issue #6333 for more details.
svn path=/trunk/; revision=53544
- Fix bug related to correct error code returning in delete_files. The value of 1026 was revealed to be returned by windows 2003 server. Score several passed winetests.
- Fix a bug in ShellLink::SetShowCmd and score one more passed winetest
svn path=/branches/shell32_new-bringup/; revision=53543
- Introduce WINE-like debug macros ERR, FIXME, WARN and TRACE
- Always show ERR and FIXME information
- WARN and TRACE are enabled by DebugPrintMask in debug.c as before
- Thanks to new macros find bug in CMake livecd creation and fix it
svn path=/trunk/; revision=53542
- Don't do some parameter checks, even if MSDN says they're incompatible. Winetests with windows 2003 server as reference tells us msdn lies. (Spotted by Victor Martinez)
svn path=/branches/shell32_new-bringup/; revision=53541
- Romanian syssetup and usetup translation by Stefan Fulea <fulea.stefan> at <gmail.com>;
- Fix ipconfig build under MSVC, use "" instead of \" as escape character;
svn path=/trunk/; revision=53539
[FORMATTING]
- Format the code to a more acceptable level. This is just for my sanity while sifting through it.
svn path=/branches/shell32_new-bringup/; revision=53537
- rewrite the parser to handle command line argument parsing
- include basic description of what the rules for parsing are
- some minor formatting here and there (couldn't be helped)
svn path=/branches/shell32_new-bringup/; revision=53533
- Move cleaning the shared user data to after WinLdrSetProcessorContext, as we need the new page tables to be active
- Fix indentation/coding style in registry.c
- Fix a few 64 bit warnings
- delete <arch>/ntsetup.c and rename <arch>/wlmemory.c to <arch>/winldr.c
svn path=/trunk/; revision=53526
Don't use enums as bitfields. MSVC doesn't handle it the way gcc does. Fixes clock initialization on MSVC builds. You would wonder how the MSVC build even worked without a clock interrupt :D
svn path=/trunk/; revision=53524
- Massively refactor the winldr code
- move common parts of LoadReactOSSetup and LoadAndBootWindows into LoadAndBootWindowsCommon
- Combine architecture specific code into 2 functions: WinLdrSetupMachineDependent prepares the main stuff and WinLdrSetProcessorContext is the last thing done before transferring control to the kernel.
- rename WinLdrTunOnPaging to WinLdrSetupMemoryLayout
- Stop wasting stack space by decreasing the number of huge text buffers used
- Don't handle x86 specific data like Tss in portable code
- Add the progressbar for reactos setup as well
- Add missing DPRINT_PELOADER to DEBUG_ALL
svn path=/trunk/; revision=53510
This will enforce use of reactos msvcrt when apps provide one.
Its required because cmake dlls currently depend on msvcrt for their startup code.
svn path=/trunk/; revision=53505
- use /h to specify header file name when producing rpc proxy file.
Don't ask me why /header doesn't work, I have no idea.
svn path=/trunk/; revision=53504
Until now, rpcrt4 winetests where badly broken : client calls remotely forwarded to serverside functions which called... client side functions. One has to wonder how testbot survived the operation
- Cmake : fix that
- rbuild : add a hack keeping the old behaviour, because I don't know how to make this right
svn path=/trunk/; revision=53503
- Add stub hidusb driver
- Hidusb driver is reponsible to connect with the hidclass driver (not yet present) and connect hid framework with usb hid devices
svn path=/branches/usb-bringup/; revision=53500
We cannot make any assumptions about the latency whith which the timer interrupt fires after a rollover, since VBox (other VMs probably as well) doesn't always meet this. Add another check to KeQueryPerformanceCounter that gracefully handles missing interrupts. Also raise to DISPATCH_LEVEL, since the function is not reentrant.
svn path=/trunk/; revision=53496
- revert r53355.
- it's changes are not needed as wine's spy channel is better, and this time we will get rid of all unwanted changes ( my previous commit did not fix all unwanted changes)
svn path=/trunk/; revision=53492
* Revert r53487 (and as a result 53490) as per Giannis (the committer) suggestion. It broke testbot and he couldn't look into it at the moment.
svn path=/trunk/; revision=53491
- reenable -Oif flag, it's too good to be disabled
With this commit, advapi32:service winetest suffers only 2 failures.
svn path=/trunk/; revision=53481
- duplicate type entry for parameters with the range attribute.
This fixes the problem that Eric Kohl detected.
This has already been sent to wine.
svn path=/trunk/; revision=53480
- Report a unique instance ID that can be used to identify different processors
- Device Manager now displays one processor device per CPU
svn path=/trunk/; revision=53478
Revert parts of r53171:
Remove the -Oif flag for WIDL because the NT4 stub code generated by the current WIDL is heavily broken with respect to the range attribute and others. Please enable the -Oif feature only after these bugs are fixed in WIDL!
svn path=/trunk/; revision=53469
- Clean up win32k headers even more
- remove INTERNAL_CALL and use NTAPI instead
- remove several unneeded definitions
- remove several unneeded header inclusions
svn path=/trunk/; revision=53467
- fix a bug in x86 PnpBiosGetDeviceNodeCount
- enable pxe for amd64 (not yet implemented though)
- store boot drive and boot partition in shared memory isnetad of relying on a register, which seems to be erased on switch to long mode
64 bit freeldr is back on track!
svn path=/trunk/; revision=53459
- Remove /nologo option from RC again - WDK's version doesn't support it
- Add /nologo to MIDL's flags
- Remove redundant IDL compiler declaration for GCC build
svn path=/trunk/; revision=53447
- Fix wrong loop condition which would often lead to heap underread.
- Fix wrong subkey string length calculation, which would result in an incorrect string being used to open an image specific key.
- Don't close the key handle in LdrQueryImageFileKeyOption() because it's used more than once and closed after it's not needed anymore.
- Use Zw* functions instead of Nt* where necessary in LdrQueryImageFileKeyOption().
- Per-image execution options work now (e.g. enabling DPH).
svn path=/trunk/; revision=53446
- Implement real mode entry point for amd64 (amd64 freeldr starts but dies later, it cannot yet switch back to real mode, which is a bit more complicated)
- delete empty folders
svn path=/trunk/; revision=53443
- Resize controls so text fits in
- Improve Polish translations
[MSPAINT]
- Use GET_X/Y_LPARAM macros instead of LOWORD/HIWORD in mouse messages because coordinates can be negative
svn path=/trunk/; revision=53441
- According to MSDN IOCTL_VIDEO_QUERY_POINTER_CAPABILITIES doesn't have an in-parameter. This code was copied from MS sample code, but it doesn't make sense, since the size field doesn't even match the in-buffer size. VBox driver also doesn't use these parameters and we don't handle the ioctl anyway.
svn path=/trunk/; revision=53434
- Use sizeof(PVOID) instead of sizeof(PFOO) just for clarity that we really want a pointer size
[SCSIPORT]
- Don't initialize a pointer with RtlZeromemory
svn path=/trunk/; revision=53429
Add PCI_HAL device for non-ACPI machines. Code taken from the ACPI-HAL. This is just a quick hack and needs some more improvements.
svn path=/trunk/; revision=53425
- Handle IRP_MN_QUERY_CAPABILITIES for the ACPI_HAL device object
[PNPMGR]
- Don't bring up a device if a required IRP has failed
- Don't perform operations on a child if it is not ready yet
svn path=/trunk/; revision=53421
- Do not return an NTSTATUS from KeSynchronizeExecution, which returns BOOLEAN
[KERNEL32]
- Do not return an NTSTATUS from BasepGetModuleHandleExW
Found by MSVC.
svn path=/trunk/; revision=53413
- fix usage of set_entrypoint
- link libpng to ntdll on amd64
- build crtdll and nmidebug only on x86
- silence some linker warnings on MSVC
- compile riched20/msvc-thiscall.c only on x86
- Add WIN64 definition to acpica
- Add ehandler.c to libcntpr
svn path=/trunk/; revision=53396
- Fix copy paste error in DIALOG_IsAccelerator, fixes a crash.
- Up Dialog.c to 1.3.26, sync port and minimized differences.
svn path=/trunk/; revision=53391
- Implement FlashWindowEx without the state verification test via global atom and window properties list. Pass all but two in win:test_FlashWindowEx.
- Add missing atom's and global thread information pointer, based on patch from bug 5655.
svn path=/trunk/; revision=53385
- Up win to 1.3.26, keeping original patch plus skipping test_SetForegroundWindow, it hangs on WaitForSingleObject line 2771, needs regression testing.
svn path=/trunk/; revision=53384
- Print an debug message if IRP_MN_FILTER_RESOURCE_REQUIREMENTS fails
- Don't let a broken driver remove our resource requirements list
svn path=/trunk/; revision=53379
* CreateFile: If the specified file or device does not exist, the function fails and the last-error code is set to ERROR_FILE_NOT_FOUND. Spotted by Claudiu and Vic.
svn path=/trunk/; revision=53373
- Mark ExTimer test as hidden
- Mark some more tests as interactive in ExHardError
- Now an automated test run on ROS runs smoothly, aside from assertions
svn path=/branches/GSoC_2011/KMTestSuite/; revision=53365
- Add current source directory to MS RC include paths
- Add include directories to WRC include paths
- Now they should behave identically enough. Thanks to Amine.
svn path=/trunk/; revision=53361
Rewrite the command packet transfer code between the service control manager (services.exe) and the service control dispatcher (advapi32.dll). The service name is no longer stored in an array of WCHARs at the end of the command packet. The service name is now stored behind the fixed size command packet header using an offset value in the header to determine the start of the string. The same method will be used to store start arguments to the service dispatcher, but it is not fully implemented yet.
svn path=/trunk/; revision=53358
QueryServiceLockStatusA/W: If lpLockStatus is NULL or cbBufSize is less than sizeof(QUERY_SERVICE_LOCK_STATUSA/W) pass a pointer to an buffer to RQueryServiceLockStatusA/W.
svn path=/trunk/; revision=53356
- Remove some unneeded and misleading debug prints when parsing debug channels
- Switch NtUser to the new debug system
svn path=/trunk/; revision=53354
- Don't mess with the device interface string
- Fixes receiving power button events
- The "ACPI Shutdown" option in the VirtualBox menu works now
svn path=/trunk/; revision=53351
Fix completely non-standard, broken and retarded definition of GET_UNALIGNED_W for 64 bit architectures
The function is supposed to read an unaligned little endian USHORT value. The old version copied the 2 bytes with memcpy to an aligned stack variable and then (bug bug) read the unaligned value directly (discarding the result of the previous operation) and passed it to bswap if neccessary.
The new function simply reads the 2 bytes and combines them using << and |.
svn path=/trunk/; revision=53348
- Add __C_specific_handler stub
- Add aliases for C++ code on amd64 builds
- Fix fmod symbol
- Merge .CRT section into .data on amd64
- remove an old hack
svn path=/trunk/; revision=53342
- Fix acquiring image path from process parameters block (the if condition was inverted so that normalization didn't occur when it should).
- Reenable the image options initialization code.
svn path=/trunk/; revision=53327
- Rename memgen.c to byteswap.c and merge with the implementations from largeint.c, that were using intrinsics
- Fix amd64 build
svn path=/trunk/; revision=53321
- Implement themes support. Even though it doesn't show a preview and doesn't support choosing a theme size, it can configure themes fine
svn path=/branches/GSoC_2011/ThemesSupport/; revision=53315
- Use GdiGradientFill and GdiTransparentBlt instead of GradientFill and TransparentBlt
- Do not link to msimg32
svn path=/branches/GSoC_2011/ThemesSupport/; revision=53309
- Implement a special case copy-in for IRP_MN_FILTER_RESOURCE_REQUIREMENTS to prevent NULL pointer access in function drivers
svn path=/trunk/; revision=53304
- Fix a typo in a remove lock timeout check
- Fixes breaking into kdbg randomly when a caller specifies no timeout on their remove lock
svn path=/trunk/; revision=53302
- Add an awful hack to prevent creating a LOGFONTW with a positive height that makes ros nuts
svn path=/branches/GSoC_2011/ThemesSupport/; revision=53295
- Make IRP major function name table public
- IoHelper: catch all IRPs and add debug messages
svn path=/branches/GSoC_2011/KMTestSuite/; revision=53293
- ObType: take advantage of r53211
- IoDriverObject: add the remaining functions from the old drvobj and devobj tests
- fix some rbuild files
svn path=/branches/GSoC_2011/KMTestSuite/; revision=53280
- Properly disable themes when theme hooks are removed
- Add WND_CONTEXT a per window struct that will hold several theme specific information. For now it contains only theme region related info. It will be stored as a window property
- Implement destroying WND_CONTEXT for every window when theme hooks are removed or when the window is destroyed
- Hook the messages that are sent to dialogs, like they are hooked for normal windows
- Hook SetWindowRgn call from user32. When an application calls it, uxtheme will stop setting window region on its own region
- Implement setting a custom window region for windows after receiving WM_WINDOWPOSCHANGED message using GetThemeBackgroundRegion
svn path=/branches/GSoC_2011/ThemesSupport/; revision=53262
- Fix a bug in IopCreateResourceListFromRequirements which was triggering a sanity ASSERT because it forgot to skip conflicting entries
svn path=/trunk/; revision=53261
- Autodetect ACPI like we autodetect a multiprocessor configuration
- This will make the ACPI HAL the default on ACPI-compliant machines while the standard HAL will continue to be the default on older machines
- Please file bugs for any issues encountered when running in ACPI mode and assign them to me (I think it's very important to have working ACPI for 0.4.0 so I'd really like to get everything fixed now)
[TXTSETUP.SIF]
- Add a MP configuration for ACPI HAL
- Slightly modify the names of each HAL configuration
svn path=/trunk/; revision=53257
- Hack around a strange registry-related bug (we created the whole registry path previously but it won't let us open it without the correct volatile option) to fix devices enumerated via IoReportDetectedDevice
svn path=/trunk/; revision=53255
- Rewrite IopIsAcpiComputer (and rename to IopIsFirmwareMapperDisabled) to eliminate the need for the ENABLE_ACPI define
- IopIsFirmwareMapperDisabled now reads the registry entry that HAL sets to indicate that the firmware mapper should be disabled
- Partially rewrite IopUpdateRootKey so it does not interfere with HAL's PnP driver
- ACPI is automatically enabled when the ACPI HAL is loaded and disabled when the standard HAL is loaded
svn path=/trunk/; revision=53249
- Fix NULL termination of strings in IoGetDeviceProperty
- Fixes garbage displayed in the Enumerator field of the device manager property page
svn path=/trunk/; revision=53232
- Unfortunately ExtCreateRegion does not seem to work, so for the time being use a simpler (and slower) implementation of UXTHEME_RegionFromDibBits that uses multiple calls to CombineRgn
svn path=/branches/GSoC_2011/ThemesSupport/; revision=53224
- Delete the region passed to NtUserSetWindowRgn. Fixes a horrible region leak that happens in every window that uses SetWindowRgn
svn path=/branches/GSoC_2011/ThemesSupport/; revision=53221
* Don't include the whole NDK for just one header.
* This allows us to depend on 49 headers less, and it shrinks the PCH by 2MB.
svn path=/trunk/; revision=53217
Fix and EnumDependentServices[A/W]:
- If lpServices is NULL or cbBufSize is less than sizeof(ENUM_SERVICE_STATUS/W) pass a pointer to an internal status buffer to REnumDependentServicesA/W.
svn path=/trunk/; revision=53214
- Store the correct buffer size for requests that aren't satisfied via the registry
- Fixes displaying the enumerator on the device manager property page
svn path=/trunk/; revision=53212
- OkToClose procedure (OB_OKAYTOCLOSE_METHOD) returns BOOLEAN, not NTSTATUS
- Check ppi for NULL before dereferencing it in IntWinstaOkToClose (it is NULL, if we do our job of cleaning up the win32 process properly)
- Set the win32 process info to NULL on process cleanup
- Free THREADINFO and PROCESSINFO instead of leaking them
- Set NULL cursor, when the last thread is cleaned up in win32k, so we don't call UserSetCursor() in the process cleanup, when we don't have a THREADINFO anymore. This should hopefully fix a crash of MSVC builds.
- Remove unused UserGetNextHandle function
- Fix some warnings
svn path=/trunk/; revision=53211
Fix EnumServiceGroupW and EnumServiceStatus[A/W]:
- If lpServices is NULL or cbBufSize is less than sizeof(ENUM_SERVICE_STATUS/W) pass a pointer to an internal status buffer to REnumServiceGrouW or REnumServiceStatusA/W.
svn path=/trunk/; revision=53210
Partly sync static.c with wine, reduce diff between ros and wine code. The leak fix is taken care of. Also go back from Get/SetWindowLongPtr to Get/SetWindowLong for LONG values. Using the *Ptr version is wrong for LONG values!
svn path=/trunk/; revision=53204
- Display the loading message for drivers loaded by the PnP manager during boot
- Maybe now NDIS.SYS won't get all the blame for boot hangs ;)
svn path=/trunk/; revision=53203
Fix QueryServiceConfigEx[A/W]:
- If lpServiceConfig is NULL or cbBufSize is less than sizeof(QUERY_SERVICE_CONFIGA/W) pass a pointer to an internal status buffer to RQueryServiceConfigA/W.
- Revert r53153 and r53154. Adding 'in' and 'unique' attributes is NOT an option because this is not compatible with Windows.
svn path=/trunk/; revision=53202
Fix EnumServicesStatusEx[A/W]:
- If lpServices is NULL or cbBufSize is less than sizeof(ENUM_SERVICE_STATUS_PROCESS) pass a pointer to an internal status buffer to REnumServicesStatusExA/W. Adding 'in' and 'unique' attributes in the idl file is NOT an option because this is not compatible with Windows.
- Check the InfoLevel.
svn path=/trunk/; revision=53201
- Add the themes service. For now it will only call ThemeHooksInstall and ThemeHooksRemove. From now on we do not need apihooksrv in order to have themes
svn path=/branches/GSoC_2011/ThemesSupport/; revision=53195
- Fix a massive memory leak of our entire TDI entity ID database after each call to getInterfaceInfoSet
- Fix another massive leak of our entire TDI interface table after each call to getInterfaceInfoByIndex and getInterfaceInfoByName
- Fix some uninitialized variable warnings
- Explorer no longer leaks horrendous amounts of memory (about 1 MB per minute even without network activity) when the network status icon is in the system tray
svn path=/trunk/; revision=53190
- Rewrite receive code to make it much simpler, handle some corner cases that weren't treated correctly, and fix the data corruption bug
- Downloaded successfully (no data corruption): OpenOffice 2.4, OpenOffice 3.3, BitTorrent 7.2.1, Firefox 5, Firefox 3.6, Firefox 3, Firefox 2, Abyss Web Server, Opera 9.64, and Opera 11.01
svn path=/trunk/; revision=53189
- Fix a buffer overflow when the packet queue has more packets than the receive request can take
- Remove an extra variable
svn path=/trunk/; revision=53188
- Reapply the switchtype patch from widl-ros.diff. This patch should be sent upstream to wine.
See issue #6406 for more details.
svn path=/trunk/; revision=53187
- Add a dummy IoHelper driver that can be loaded by driver/device tests
- Minor changes to other standalone drivers
svn path=/branches/GSoC_2011/KMTestSuite/; revision=53177
- Implement SO_LINGER and SO_DONTLINGER for getsockopt and setsockopt
- Fix and add buffer size checks
- Yes, we've gone this long without SO_LINGER support
svn path=/trunk/; revision=53174
- use -Oif flag to produce modern (!) NT4 stubs, with Microsoft extensions to IDL language.
[ADVAPI32]
- Initialize *pcbBytesNeeded, as it's marked as [in], it needs to be in the 0-8k range before being passed to RPC.
Now 2nd stage got his mouse and keyboard again.
svn path=/trunk/; revision=53171
- More reworking of ObType test. Now it works reliably on Windows.
- Split out testing of object referencing into separate ObReference test
svn path=/branches/GSoC_2011/KMTestSuite/; revision=53169
- Remove breakpoint on buffer overflow
- Use a compromise for ok_* macros for SIZE_T and [U]LONG_PTR values
svn path=/branches/GSoC_2011/KMTestSuite/; revision=53167
Fix MSVC warning about signed/unsigned mismatch. In this case it was not a bug, because the compiler implicitly casts to unsigned, when signed and unsigned are compared, so (-1 < 0x23U) is false.
svn path=/trunk/; revision=53163
- fix confusion on buffer size between string length and number of bytes.
Might fix some infamous heap corruption bug during install. Please test.
svn path=/trunk/; revision=53160
- addendum to r53153 : [unique] and [ptr] can't be applied to [out] only parameters, so mark them as [in] too.
Now MSVC build gets as far as 3rd stage.
svn path=/trunk/; revision=53154
memory barriers
- implement a fixed version of __invlpg for MSVC, since the intrinsic is broken and can generate wrong instructions, when optimization is enabled
- add MSVC versions of _mm_mfence, _mm_lfence, _mm_sfence and __faststorefence
- add _mm_mfence() for gcc
- give _mm_lfence, _mm_sfence, __invlpg and __wbinvd proper
svn path=/trunk/; revision=53151
- Implement a simple buffer allocation algorithm which backs off the buffer size until all buffer allocations succeed
See issue #6284 for more details.
svn path=/trunk/; revision=53149
- Allocate receive and transmit buffers (each 49152 bytes) as cached shared memory to increase the likelihood of a successful allocation in a low memory scenario
See issue #6284 for more details.
svn path=/trunk/; revision=53148
- Implement NdisMDeregisterIoPortRange
[PCNET]
- Don't call NdisMFreeMapRegisters if no map registers were allocated
- Call MiFreeSharedMemory before freeing map registers to avoid a crash because freeing map registers deallocates the adapter object
svn path=/trunk/; revision=53147
- Add co_MsqSendMessageAsync in order to send asynchronous normal and internal messages
- Use co_MsqSendMessageAsync to load and unload api hook module
- This hopefully fixes the last problems with api hooks implementation
svn path=/branches/GSoC_2011/ThemesSupport/; revision=53139
- Resize controls to allow wider strings of polish translation to fit in;
- Assorted translation fixes, added missing strings;
- Patch by Adam Stachowicz (Saibamen) and wojo664, minor corrections by me;
svn path=/trunk/; revision=53127
- UXTHEME_LoadTheme call in theme hook callback is still needed in order to load the active theme in an application that starts after themes are enabled
svn path=/branches/GSoC_2011/ThemesSupport/; revision=53125
- Broadcast WM_THEMECHANGED when hook themes are activated
- Do not call UXTHEME_LoadTheme in the hook theme callback. It will now be called when the application gets WM_THEMECHANGED
svn path=/branches/GSoC_2011/ThemesSupport/; revision=53124
- Store ghmodUserApiHook before calling ClearUserApiHook because it will be zeroed during this call
- Fix a typo in ClearUserApiHook
svn path=/branches/GSoC_2011/ThemesSupport/; revision=53123
- Merge the Google Summer of Code TcpIpDriver project to replace our current TCP implementation
- Remove oskittcp
- Big thanks to Claudiu Mihail for his work on the project
svn path=/trunk/; revision=53122
- When a service gets started, read the start parameters from the edit control and pass them to StartService.
- Disable the start parameter edit control while a service is running.
svn path=/trunk/; revision=53121
- Hack cmake build file to get wininet.dll delay importing.
msvc build reactos now gets as far as showing 2nd stage setup wizard.
Mouse/keyboard don't respond yet.
svn path=/trunk/; revision=53120
- turn most macros in function 2/x.
- separate rpc server and client stubs, and generate them directly instead of using a library, so they got the flags that the module uses.
- commit forgotten file from r53088
- get rid of winesdk target, it's useless now
svn path=/trunk/; revision=53112
- turn most macros in function 1/x.
This avoids defining a truckload of variables in the global space, and some of these variables would collapse themselves in certain cases.
svn path=/trunk/; revision=53110
- Rename ObCreate test to ObType, and apply some major refactoring. Now _almost_ works on Windows....
- Add a readme file
svn path=/branches/GSoC_2011/KMTestSuite/; revision=53103
- make generated files depend on their generator
It seems stupid, but I removed this quite some time ago, don't ask me why.
Now, you just have to build the tools and do an incremental build each time a tool is updated.
svn path=/trunk/; revision=53088
- do not redefine UNICODE_STRING and NTSTATUS if wintrnl.h has already been included
- define PNTSTATUS in wintrnl.h
- add RPC_PROTSEC_VECTOR and RPC_MGMT_AUTHORIZATION_FN definitions
svn path=/trunk/; revision=53087
- Add back apparently lost line of r52098.
Patch by Alex Ionescu, tested by Igor Paliychuk.
See issue #6292 for more details.
svn path=/trunk/; revision=53085
- Add function types, not only function pointer types, for splay/avl tree functions
- RTL_SPLAY_LINKS needs to be defined regardless of RTL_USE_AVL_TABLES
[NDK]
- Add missing splay/avl tree function declarations
- Add the function types as well. These are also not in ntifs anymore
svn path=/trunk/; revision=53073
- Define NDEBUG only once
- Don't define helper macro in the source
- DPRINT fixes
[NTDLL/LDR]
- More verbose DPRINT for initialization failure
svn path=/trunk/; revision=53072
[KERNEL32]: Rename hProcessHeap to BaseHeap, and only allow the heapmem.c access to this (change find.c to use RtlGetProcessHeap() like everyone else).
svn path=/trunk/; revision=53068
- Comment out the timeout parameter until I implement disconnect timeouts for lwIP
- Branch repair (part 5 of x)
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=53067
- Remove oskittcp.h from the PCH
[LWIP]
- Don't define LWIP_PROVIDE_ERRNO because we get those from our CRT
- Branch repair (part 4 of x)
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=53066
- Actually, there is no need to crawl the entire tree, just going past the limit is enough. Thanks Timo.
- Fix MSVC warning, thanks Ged.
Now you have correct AND optimized code, Alex AND Timo should be happy.
I'm only the committer of these fix, the work was mostly done by Thomas Faber and Alex Ionescu.
svn path=/trunk/; revision=53064
[NTOSKRNL]: Handle failure of NLS and PEB mapping correctly.
[NTOSKRNL]: Only set Peb->OsCSDVersion if NtHeader->Win32VersionValue is set.
svn path=/trunk/; revision=53059
- use functions instead of macros.
This avoids adding flags multiple times.
See the PARENT_SCOPE option of cmake's command "set" for details.
svn path=/trunk/; revision=53048
- add IoDriverObject test, partly ported from old kmtests. Should possibly be merged with another test
- ExDoubleList: add another small testcase
svn path=/branches/GSoC_2011/KMTestSuite/; revision=53047
- add RtlSplayTree test. Patch by arty (hacks by me ;])
- add RtlAvlTree test using the same code
- add kmtest_drivers and kmtest_all build targets
svn path=/branches/GSoC_2011/KMTestSuite/; revision=53046
[CSRSRV/KERNEL32]: Have CSRSS fill out the base server static data. Still a bit of a hack since we should have a basesrv to do this, but it'll fix the "can't install any drivers" bug as well as increase performance since we now cache a bunch of data system-wide.
[KERNEL32]: Remove all related hacks in kernel32.
[CSRSRV]: Only accept the incoming connection if we found CSR process data about it.
svn path=/trunk/; revision=53040
- Correct the error number comments for lwIP 1.4.0
- Translate ERR_WOULDBLOCK
- Translate ERR_IF and ERR_CONN to a more accurate status
[LWIP]
- Remove another debug print
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=53039
- If a service will be started and another service is already running in the same image, send a start command instead of starting the process again.
This enables us to run multiple services within a single process, like svchost.exe.
svn path=/trunk/; revision=53036
- Remove debug prints
- Modify error code translations
[lwIP]
- Fix race conditions caused by checking the SocketContexts of connections from outside the lwIP main thread context
- Don't explicitly remove callbacks from pcbs (no reason to do apperently)
- Use pbuf_free_callback (which can be safely called from outside the lwIP main thread context) instead of pbuf_free
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=53033
- Make UserDrawCaption much more simple and compatible with Windows. It supports now icon given in parameter places text and icon in valid position. Fixes Start button icon in explorer_new.
- Uncomment two function in win32k. I used one of them in UserDrawCaptionText and seems to work
- Add debug print to NtGdiPatBlt
svn path=/trunk/; revision=53031
- HACK RegisterClass in order to let comctl32 re-register built in classes. This is needed in order to fix themes and will keep it till we find a way to do it properly. (of course this breaks class wine tests as well)
svn path=/branches/GSoC_2011/ThemesSupport/; revision=53030
- Register themed built-in controls even if themes are not enabled. This allows built in controls to use themes without restarting the application
- Register built in controls in the same way it done in windows (as global classes that belong to comctl32). This is needed in order to have themed controls in dialogs. Unfortunately our user32/win32k does not support it yet and this breaks themed built in controls
svn path=/branches/GSoC_2011/ThemesSupport/; revision=53029
- Do not try to load the current theme when uxtheme loads. Instead, load the current theme when theme hooks are installed or when the application receives a WM_THEMECHANGED message. This allows us to change themes without restarting the running applications.
- Use the default behaviour of DefWindowProc from user32 when themes are not active. This allows us to have api hooks enabled when themes are disabled.
svn path=/branches/GSoC_2011/ThemesSupport/; revision=53026
- Fix possible thread reference leak when calling hook
- Fix possible memory corruption if hook is unexpectedly removed
- Cleanup hooks a bit
- Fixes bug #1567 (explorer ghost in taskmgr)
svn path=/trunk/; revision=53022
- add kmt_platform.h that includes user or kernel headers as appropriate and allows Rtl tests to run in user mode without modification
- include kmt_platform.h from kmt_test.h, so that tests don't have to include separate headers. This also allows for a PCH
svn path=/branches/GSoC_2011/KMTestSuite/; revision=53021
Implemenet SwapSplayLinks, return 'NewElement' correctly when inserting.
Thanks to Alex Ionescu for helping out with this patch.
svn path=/trunk/; revision=53015
- Force the connection to be closed immediately
- Hopefully will fix the random bind issues caused by oskit becoming out of sync with the rest of the IP library
svn path=/trunk/; revision=53014
Fix OBTRACE definition.
the macro used __VA_ARGS__ and passes that to DPRINT(fmt, ...)
The preprocessor interpretes all arguments of __VA_ARGS__ as one token. This token was now passed as the first parameter of DPRINT1. This caused wrong parameters being used, resulting in a crash.
svn path=/trunk/; revision=53008
- Report recalibration failure if the EC flag is set after seeking to fix a hang while trying to read
- Thanks to igorko for testing
svn path=/trunk/; revision=53006
- ScmCreateOrReferenceServiceImage: Link the SERVICE_IMAGE struct to the SERVICE struct in the create and the reference case. Otherwise only the first service started from an image gets a valid pImage pointer.
- Add some debug code.
svn path=/trunk/; revision=53004
Use a high segment and 0 offset when loading setupldr.
This fixes problems with an award bios.
Thanks to Igor Paliychuk (igorko) for his help and patience burning / testing dozens of isos
Fixes boot regression of cmake builds on real hw.
svn path=/trunk/; revision=53003
- When ClearUserApiHook returns false, this means that the library will be freed in EndUserApiHook. In this case return success to let win32k know that the process has removed the api hook. Fixes reloading the api hook after unloading it
svn path=/branches/GSoC_2011/ThemesSupport/; revision=52949
- probe user address in user mode
- don't free the result buffer before the driver unlocks it. Fixes assertion in ROS for large buffers
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52947
* Introduce a PCH.
* BTW, the recent and upcoming changes are two-fold: First, having a precompiled header will increase build speed. Furthermore, PCH support allows CMake dependency tracking files to shrink drastically. For example it's (depend.make) 95% smaller for this module than it used to be. This means an even better build system compared to rbuild, not to mention it's now faster than before.
* This work is dedicated to Jerome.
svn path=/trunk/; revision=52904
- Revert the timer hack. It's no longer needed for speed boosts. Now lwIP is clean of any foreign code.
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52900
[KERNEL32]: VirtualAllocEx should build a SEH frame.
[KERNEL32]: VirtualAllocEx should make sure the address given is not lower than the system's allocation granularity.
svn path=/trunk/; revision=52899
[KERNEL32]: Add support for terminal services timezone redirection (one more for GedMurphy's pipe dream) in all the time functions.
[KERNEL32]: Make the time functions acquire SE_SYSTEMTIME_PRIVILEGE when needed (and release it).
[KERNEL32]: Fix the DOS time conversion routines.
[KERNEL32]: Add missing validations to some time functions.
[KERNEL32]: Implement IsTimeZoneRedirectionEnabled.
[KERNEL32]: Reimplement GetSystemTimes to support MP, and treat parameters as optional.
svn path=/trunk/; revision=52898
- Fix completing IRPs with 0 bytes received when getting the data from the connection's packet queue
- Eliminate memory leaks caused by unreleased pbufs
- Get rid of some commented code
In principle the speed issue with lwIP should be pretty much solved now. There's still some minor things to iron out that testing will reveal probably. Initial tests like running opera, downloading stuff etc seem to be very encouraging however.
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52894
- Fix#6262: each new applied color scheme increase the window title height by one
- Don't read REG_QWORD directly to DWORD variable
svn path=/trunk/; revision=52892
- Implement support for debug channels. When a process starts, win32k will parse DEBUGCHANNEL environment variable in order to enable or disable specific debug channels or levels for this application. This gives us the ability to activate from user mode extended debug output for one application, for one component, for one level. In order to use the new debugging system we have to use the macros ERR, FIXME, WARN and TRACE. These are not used yet but shortly every DPRINT/DPRINT1 call will be converted to use the new macros
- Define several win32k channels that will be used in the future
- Remove some hacky definitions of the new macros
svn path=/trunk/; revision=52876
FF4 was removed, FF5 was added instead.
Thunderbird 5 was added, too.
Rest was updated to the newest version.
Maybe we should add http://source.winehq.org/patches/data/74067 from wine to get ftp support fixed.
svn path=/trunk/; revision=52874
* Update our asm support thanks to Jerome.
* Bring in CMakeDetermineCompilerId.cmake to make sure this will work for pretty much all systems.
* Disable the use of response files when you enable ccache.
svn path=/trunk/; revision=52873
[KERNEL32]: Same goes for BaseCachedSysInfo. Use the BaseStaticServerData->SysInfo instead now.
[KERNEL32]: Don't assume that the BNO directory is called BaseNamedObjects (this won't be true once GedMurphy's dream of multi-session support comes true) -- instead, use the one from BaseStaticServerData.
svn path=/trunk/; revision=52872
- add KeSpinLock test
- KeIrql: also test the actual stdcall versions of KeRaiseIrql/KeLowerIrql
- ExHardError: take advantage of r52847
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52869
* Introduce a temporary cmake variable (PCH) to enable/disable cmake pch support as this requires our patched binary.
* Wrap pch macros around this variable.
* Implement the pch marco accordingly. By Jerome.
svn path=/trunk/; revision=52865
- Add queue for connections objects in order to buffer packets that have arrived but have no corresponding receive requests to be assigned to. We buffer these packets to avoid giving a timeout that could cause throughput slowdowns.
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52864
- Convert KiCallUserMode from asm to C (with a small asm wrapper)
- Convert KiGetUserModeStackAddress into a C inline function
svn path=/trunk/; revision=52855
- Remove Resource functions, as they are in wdm.h
- add KeTryToAcquireQueuedSpinLockRaiseToSynch, ExRaiseHardError & enum members
Patch by Thomas Faber
svn path=/trunk/; revision=52847
- get rid of atexit handling, it's arlready handled by msvcrt.
The funny thing is that the only module using this is msvcrt_winetest, which tested the mingw implementation, not the msvcrt one.
svn path=/trunk/; revision=52828
- link to static crt instead of forwarding function to msvcrt.
As msvcrt, msvcrt20 exports some variable, marked as DATA in the def file. Forwarding them is not a good idea, since msvcrt would be initialized only if one of the forwarded function is called. We must keep those variables in our module. And as the forwarded functions might use the said variables, we can't really forward them, as they would not know which instance they should reference.
svn path=/trunk/; revision=52826
- *Temporarily* back out some rossym/kdbg features until bugs 6372 and 6373 are fixed.
- Traces should be reliable again.
See issue #6372 for more details.
svn path=/trunk/; revision=52825
[KERNEL32]: Bug: QueryMemoryResourceNotification should set ERROR_INVALID_PARAMETER and fail if the Handle is NULL or INVALID_HANDLE_VALUE.
[KERNEL32]: Fix ULONG<->BOOL cast in QueryMemoryResourceNotification as well.
svn path=/trunk/; revision=52817
[KERNEL32]: SetLastError should only set the error code if it's different from the one already set. This is not about optimization: it's compatibility with hardware data breakpoint behavior.
svn path=/trunk/; revision=52816
Bug: RemoveHandles was checking for process AND thread handle match, instead of OR.
Bug: CloseAllProcessHandles and RemoveHandles's processing loop was buggy with its unlinking and re-use of nested ThreadData.
Bug: ProcessIdToHandle was requesting more process rights than needed.
Bug: Some ULONG<->BOOL conversions were dirty (as in, could result in values of 2 or higher).
Bug: In WaitForDebugEvent, During CREATE_PROCESS_DEBUG_EVENT, the wrong handle was saved as the thread handle.
Bug: In WaitForDebugEvent, all events returned TRUE. Instead, only valid events should return TRUE, and non-existing events should return FALSE.
Add some asserts.
Simplify MarkThreadHandle and MarkProcessHandle.
Simplify OutputDebugStringW.
svn path=/trunk/; revision=52813
- move BaseFiberStartup to fiber.S
- add FPO debug info for MSVC builds
- remove excessive END directives (fixes msvc build)
- add a stub for amd64
svn path=/trunk/; revision=52810
[KERNEL32]: BasepInitializeContext was not creating a correct CONTEXT record for fibers: the stack return address was not set (EIP was being used instead), and support for FPU-compatible Fibers was non-existent.
[KERNEL32]: CreateFiberEx was not passing the correct context flags to BasepInitializeContext to notify it that this is an FPU-fiber.
[KERNEL32]: SwitchToFiber was using some weird "FXSR" constant that maps to checking of PowerPC 64-bit Move instructions are available. We actually want to check for XMMI.
svn path=/trunk/; revision=52809
Bug #50: SwitchToFiber needs to check if FXSR is *NOT* present in order to skip using ldmxcsr/stmxcsr. Previously, it would check if it's unsupported, and jump past the instruction if it was (resulting in invalid opcode instructions on older systems)
50 bugs. Penance has been paid.
svn path=/trunk/; revision=52807
Bug #47: CreateFiberEx should return ERROR_INVALID_PARAMETER for invalid flags.
Bug #48: DeleteFiber should also check if HasFiberData is TRUE in TEB before calling ExitThread.
Bug #49: DeleteFiber should only delete the FIBER data structure after cleanup is complete, not before.
Also, enable the calls to RtlAllocateActivationContextStack since SxS kind-of exists now. However, we leak the context since RtlFreeActivationContextStack is not implemented.
Also, add a call to BaseRundownFls so we don't forget to cleanup FLS data once FLS is implemented.
svn path=/trunk/; revision=52805
Bug #45: ConvertThreadToFiberEx should return ERROR_INVALID_PARAMETER if an invalid flag is passed.
Bug #46: ConvertThreadToFiberEx should return ERROR_ALREADY_FIBER if the thread is already a fiber, not return the current fiber data.
svn path=/trunk/; revision=52804
Bug #43: ConvertFiberToThread should return ERROR_ALREADY_THREAD, not ERROR_INVALID_PARAMETER in case of failure.
Bug #44: ConvertFiberToThread should set FiberData to NULL in the TEB, after doing the conversion.
Thanks, Winetests, for saying "0 failures", when this API was broken.
svn path=/trunk/; revision=52803
Bug #39: BaseThreadStartup needs to call CsrNewThread (which registers the termination port with CSRSS) for Win32 threads.
Bug #40: If a crash happens during thread execution, ExitProcess should be called instead of ExitThread (unless this is a CSR thread).
Bug #41: The ExitThread/Process APIs should be called still from within the context of the SEH frame, not outside of it.
Use BaseRunningInServerProcess global instead of local IsServer local when connecting to CSRSS. This is needed for things such as the above.
svn path=/trunk/; revision=52801
Bug #35: TlsSetValue should allocate the TlsExpansionSlots under the PEB Lock.
Bug #36: TlsGetValue needs to set the last error to ERROR_SUCCESS when the index is valid (amd even if no expansion slots exist!).
Bug #37: TlsFree did not check the return of NtSetInformationThread(ThreadZeroTlsCell).
Bug #38: TlsAlloc was setting error to ERROR_NO_MORE_ITEMS instead of STATUS_NO_MEMORY. In principle the former is more accurate, but that's not how API compatibility works.
Optimize TLS functions not to call NtQueryTeb/Peb all the time. Cache the TEB/PEB value instead.
svn path=/trunk/; revision=52800
[KERNEL32]: Fix bugs #26, #27: RegisterWait(Ex) was not calling GetConsoleInputWaitHandle, so waits on console handles were given straight to the kernel (which obvioulsy doesn't grok them).
[KERNEL32]: Fix bugs #28, #29: UnRegisterWait(Ex) was not checking for STATUS_PENDING, which is a "warning" in NT, but an error as far as this API is concerned, so FALSE should be returned.
svn path=/trunk/; revision=52797
[KERNEL32]: Update Event APIs to use the new macros. Fixes bugs #14, #15: the create/open ANSI APIs were not returning the right error in case of object names that were too large.
svn path=/trunk/; revision=52791
[KERNEL32]: Fix bugs #11, #12, #13: OpenJobObjectA did not correctly set the error code in case the string was too long, nor did it use the TEB's static unicode cache, nor did it enforce a 260 character object name limit. Fixed by using the new macro.
svn path=/trunk/; revision=52789
[KERNEL32]: Fix bugs #7, #8, #9, #10: CreateJobObjectW did not add OBJ_OPENIF to named jobs, it did not add named objects in the BaseNamedObjects directory, it did not correctly set ERROR_ALREADY_EXISTS when collisions happened, and it did not set ERROR_SUCCESS when the object was created. All this was fixed by using the new CreateNtObjectFromWin32Api macro.
svn path=/trunk/; revision=52788
- KeIrql: make use of the new build type features
- KeApc: test internals of enabling/disable APCs
[KMTESTS/EX]
- better portability for ExInterlocked test
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52782
- small structure improvement in kmt_test.h
- add globals to allow easy checking for UP/MP & Free/Checked builds, as that is required for some tests
- add KernelType example test demonstrating the above
- add a function to restore the irql (allows some failing tests to exit gracefully)
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52781
- put driver files into 'bin' with the other tests instead of system32\drivers
- globally define NTDDI_VERSION=NTDDI_WS03_SP1 instead of doing it in each test
- fix testlist sorting (ASCII-order, but ignore leading minus)
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52780
[KERNEL32]: Fix bug #1: CreateFileMappingA was not returning the right error if the file mapping name was too long. By making it use the new ConvertWin32AnsiObjectApiToUnicodeApi macro, it now does.
svn path=/trunk/; revision=52776
[KERNEL32]: And at last, the final stroke. Move toolhelp.c into client, and split time.c into timezone.c (wineisms) and time.c (incorrect/buggy ntisms), which goes into client. Now "misc" truly only contains Wine-code (along with winnls), and the true NT base API that needs to be focused on is in "client".
svn path=/trunk/; revision=52772
- Modify the signatures of the lwIP wrapper interfaces. This is the first step toward fixing some nasty crashes related to race conditions.
[TCPIP]
- Call the new lwIP wrapper interfaces appropriately.
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52767
[KERNEL32]: Move more Vista APIs out of the way and un-export some functions. Reset the WINNT_VERSION to 0x502 instead of 0x600, except for synch.c (TBD).
[KERNEL32]: Fix casts and math in GetTickCount and GetTickCount64. GetTickCount64 is Vista, but Wine-synched DLLs need it, so leave it exported.
svn path=/trunk/; revision=52761
[KERNEL32]: Remove GetLocaleInfoEx again.
[KERNEL32]: It seems CMAKE build doesn't like kernel32 code calling "HeapAlloc" because HeapAlloc doesn't exist -- it's a forward to RtlAllocateHeap. As a hack, add compiler defines to do the forwarding in C code as well.
svn path=/trunk/; revision=52760
Add GetLocaleInfoEx back to the exports, idndl needs it.
Should fix default (rbuild) build. No idea (and don't care) about the cmake build, the cmake boys will have to deal with that if it breaks.
svn path=/trunk/; revision=52757
In theory, this code is "better" than before, and it is closer to Wine (which arguably has better compatibility). It also resets things in sync with Wine however, and may lose and "fixes" ReactOS may have added over the years. But this is a good thing, since these fixes have been "lost" (they obviously never made it into Wine), and if regressions are now found due to this, actual upstream patches can be sent and picked up on the next sync. This avoids maintaining duplicate code, at the expenses of some potential short-term regressions in i18n.
Finally, note that much of /string seems to be taken from Wine's Unicode library (which a host "unicode" already exists in ReactOS' tools/. It may be better (for someone with more experience as to these wine-isms) to simply just pull-in whatever winelib files are not currently present in ReactOS, and have kernel32 and tools/unicode use winelib, instead of having 2 or 3 copies of the code.
svn path=/trunk/; revision=52754
[KERNEL32]: Final part of the re-structure: move the last few files from "misc" into "client". Now "misc" is mostly composed of Wine-based code (which will soon by Winesynched) that provides functionality such as INI, .rsrc parsing and LZip expansion, as well as the RS232-based APIs. The rest of "client" is NT-layer glue, much of it buggy and in need of help.
[KERNEL32]: Move "Beep" from environ.c to deviceio.c... seems to make more sense there.
[KERNEL32]: Set kernel32_handle since Wine-synched code will depend on this.
[KERNEL32]: Link with winelib, don't link with normaliz anymore.
svn path=/trunk/; revision=52752
- Add some enums which were accidentally removed.
- Add the rather strange Wineisms which some of their dlls rely on.
- Should fix the build for this branch
svn path=/branches/shell32_new-bringup/; revision=52750
Patch by Dmitry Gorbachev:
Don't overwrite the OemName field in the bootsector. Although this field could theoretically contain any string, MS recommends using "MSWIN4.0" and MSDOS does even rely on it being this (or having higher last 2 characters)
This way we preserve what is being put there when the disk is formatted, which is "MSWIN4.0"
See issue #6386 for more details.
svn path=/trunk/; revision=52742
Patch by Thomas Faber
ExAcquireSharedStarveExclusive and ExAcquireSharedWaitForExclusive should
return FALSE when they cannot immediately acquire the resource, and the Wait parameter is false. They return TRUE instead, leading the caller to believe it acquired the resource, which will result in a bugcheck when it tries to release it.
See issue #6375 for more details.
svn path=/trunk/; revision=52741
[KERNEL32]: Reorganize the source tree into a "client" section (The Win32 API proper and its wrapping of the NT base system -- unsynchable with Wine until they have a proper NtDll and choose to use our code) and a "winnls" section (string/language/etc stuff easily synchable with Wine).
[KERNEL32]: Note: this is not perfect yet, but a step in the right direction.
svn path=/trunk/; revision=52738
[KERNEL32]: Reorganize certain functions in different files, and move files into more appropriate locations. Will lead to a simplification of the directory structure.
svn path=/trunk/; revision=52737
[NDK]: Fix certain SxS functions.
[RTL]: "Implement" RtlActivateActivationContextEx which lets you specify a TEB. RtlActivateActivationContext uses the current TEB. NOTE: All these functions are broken because they copy-paste Win32 wine code into NT Native functions (with different parameters/structures).
[NDK]: Fix pstypes.h -- it was a #define mess, and worked by luck because we are a < Vista OS.
[NDK]: Add RtlInitEmptyAnsiString.
svn path=/trunk/; revision=52736
[RTL]: Reimplement RtlDosPathNameToNtPathName_U to use UNICODE_STRING semantics.
[RTL]: Fix RtlGetFullPathName_Ustr.
[RTL]: RtlGetFullPathName_U remains based on the legacy non-UNICODE_STRING mechanism, but it's too complex to attempt changing for now.
svn path=/trunk/; revision=52726
- Fix a nasty datagram corruption bug that would result in an uninitialized buffer data being returned instead of packet data if the client read buffer was smaller than the datagram received
- Fix broken user-mode send datagram IRP completion code which didn't set the completion status
- Implement disabling/enabling event select triggers
svn path=/trunk/; revision=52723
- Skip test_enhmetafile to prevent the oleaut_winetest:olepicture crash.
See issue #5000 for more details. Test skipping must be diffed, else will be reverted again at winesync.
svn path=/trunk/; revision=52721
- Fix a race condition that occurs when the socket is closed by the remote host while waiting on OSKLock to perform a socket operation and results in accessing freed memory
svn path=/trunk/; revision=52718
Implement service image records. This will enable us to run executables that contain multiple services and control them individually.
svn path=/trunk/; revision=52717
- Sync comctl32 with wine. Among it's improvements are reduced flicker in the listview and hot tracking support for themed buttons
svn path=/branches/GSoC_2011/ThemesSupport/; revision=52714
[RTL]: Simplify RtlDosSearchPath_U by not using a temporary "char" value, and just reading straight from the string buffer.
svn path=/trunk/; revision=52713
- fix crash caused by a race condition when trying to close a socket
- move timer hack from sys_arch.c to timers.c because there it gives best performance (will have to get rid of it at some point though)
- when closing a socket only mask out the receive and accept callbacks (as they can cause of a crash when closing a socket). The rest of the callbacks are left alone because they may be used.
[TCPIP]
- merge r52698
- remove some DbgPrint calls that are now useless since the bug they were introduced to help discover has been fixed
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52712
Generate MSVC debugging info for KiUserExceptionDispatcher, KiIntSystemCall, KiFastSystemCall, KiFastSystemCallRet, KiSystemService, KiFastCallEntry, all systemcall stubs
Now usermode backtraces look as they should in WinDbg
svn path=/trunk/; revision=52705
[RTL]: Re-implement RtlDosSearchPath_U using new path functions from past commits. Uses correct behavior with RtlDoesFileExists (for relative paths, return success when file is locked -- for absolute paths, do not!). Also uses a more optimized path loop that does not alloc/realloc/free heap continously (only one heap allocation is used).
svn path=/trunk/; revision=52699
- Fix the broken TDI_QUERY_CONNECTION_INFO implementation
- Perform buffer size checks on TDI_QUERY_MAX_DATAGRAM_INFO requests
[AFD]
- Greatly simplify AfdGetPeerName by using the remote address stored while connecting
svn path=/trunk/; revision=52698
- Fix translation crediting by Elton Chung;
- Solitaire and Spider swedish translation by Ronny Svensson;
- Multiple applications translated to simplified chinese by Song Fuchang;
svn path=/trunk/; revision=52695
[RTL]: Make RtlDoesFileExists_UStrEx use the new RTL_RELATIVE_NAME structure and also support whether or not sharing violations should return success or not.
[RTL]: For now, use the old RtlDosPathNameToNtPathName API instead of the newer one.
svn path=/trunk/; revision=52685
[NDK]: Add RTL_RELATIVE_NAME_U which is the Windows 2003+ structure used instead of CURDIR for all relative path APIs.
svn path=/trunk/; revision=52681
- Fix the size of the caption buttons. Now they have the correct size with themes like Luna Inspirat or clear looks
svn path=/branches/GSoC_2011/ThemesSupport/; revision=52679
- ExResource part 2: test concurrent acquisition of a resource in multiple threads
- ExInterlocked part 2: check calling convention correctness
- these show quite a number of bugs in ReactOS ;)
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52674
- Disable Server to server calls in csr because they do not work yet as our csr is not compatible with windows.
[user32]
- Call CsrClientConnectToServer when the dll is initialized in order to connect with the server and determine if user32 is running in the server process.
- Do not load user api hook if we are in the server process
- Fixes drawing the console window
svn path=/branches/GSoC_2011/ThemesSupport/; revision=52672
- Store the address of an outgoing NIC properly (gethostname() fix)
- Don't do an explicit bind if we don't have to (ws2_32_winetest sock hang fix)
svn path=/trunk/; revision=52667
- Commit a test utility that can install user api hooks and give for the first time real themes support. This is a temporary test program till the same functionality is integrated to winlogon. To use it someone has to activate themes with themecfg and then run "apihook uxtheme.dll". Also note that it is possible to use this tool in windows for testing purposes
svn path=/branches/GSoC_2011/ThemesSupport/; revision=52665
- Add a test utility called themecfg that can configure themes. This is a small part of winecfg from wine. With this utility we can test themes support now. In the future this functionality will become part of desk.cpl
svn path=/branches/GSoC_2011/ThemesSupport/; revision=52664
- Fix broken handling for duplicate loads in IopLoadUnloadDriver
- Fixes duplicate loading messages being displayed during boot and duplicate legacy device nodes being created for the same driver
- NOTE: These fixes do not affect drivers loaded by FreeLoader which are handled by a separate function IopInitializeBuiltinDriver
svn path=/trunk/; revision=52660
- Call UserUnregisterUserApiHook from the process cleanup routine if the process has registered the user api hook
svn path=/branches/GSoC_2011/ThemesSupport/; revision=52652
- Convert LOADUSERAPIHOOK macro to an inline function
- Implement ClientLoadLibrary callback from win32k
svn path=/branches/GSoC_2011/ThemesSupport/; revision=52648
- Implement drawing non client scrollbars. This code is heavily based on code from wine and modified properly to use themes
svn path=/branches/GSoC_2011/ThemesSupport/; revision=52645
- Increase the default and overflow size of the TCP/IP packet pool to allow for periods of very high network load (such as using BitTorrent or hosting high load services on a network) without dropping outgoing or incoming packets
- Increase the size of the TCP/IP buffer pool for the same reason
svn path=/trunk/; revision=52641
- After doing any drawing operation for the non client area, restore the previously opened theme. Fixes GetWindowTheme
svn path=/branches/GSoC_2011/ThemesSupport/; revision=52639
- add a small hack in sys_arch.c to boost throughput; we forcefully make the timeout value a foxed constant to boost speed
- add small language optimizations
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52637
- Do not close hKey in query_reg_path. Noticed while testing our uxtheme in windows. Should be sent to wine
svn path=/branches/GSoC_2011/ThemesSupport/; revision=52634
-Add a hack to GreStretchBltMask to return TRUE when a dimension is 0. After some investigation it seems that windows returns TRUE in such a case but couldn't investigate more if this change is really correct. This hack is needed because a drawing operation in uxtheme is aborted when StretchBlt returns FALSE
svn path=/branches/GSoC_2011/ThemesSupport/; revision=52631
- Optimize the way Send and Close handlers work by executing executing the raw lwIP functions directly instead of using lwIP's callback system. This works because the handlers are executed from lwIP's master thread context
- small language optimizations
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52630
Implement the "SLIST hack". It checks whether the access to the Next member of the first list item of an iSList caused the exception, this can happen, when a concurrent thread pops the first entry and frees its memory between the point where the address of the entry is loaded and the access to it. This is done before the call to MmAccessFault to handle the theoretical scenario of a guard page exception being triggered by the fault, which we don't want to be handled.
Currently only kernel mode is handled.
svn path=/trunk/; revision=52628
- Fix CSR hard error messages. Parameters are now converted to ANSI string which are expected by message format string.
svn path=/trunk/; revision=52619
[RTL]: Implement RtlImageNtHeaderEx and make RtlImageNtHeader call it. This will result in much stricter (and correct) PE loading by default, and enable a flag which can do even stricter checks.
[NTDLL]: Export RtlImageNtHeaderEx (maybe NTOS should export it too, not sure).
svn path=/trunk/; revision=52618
- fix broken logic
- simplify calculation of remainig space to alter
- call ExFreePoolWithTag instead of ExFreePool
- return error on failure
svn path=/trunk/; revision=52614
- Don't use -1 for USHORT LDR_DATA_TABLE_ENTRY::LoadCount. When comparing ((USHORT)-1) == 0xFFFF vs (int)-1 it ends in comparison 0xFFFF vs 0xFFFFFFFF with is wrong. Fixes shutdown from shell
See issue #6345 for more details.
svn path=/trunk/; revision=52601
- Fix broken TDI address functions which were causing subtle memory corruptions, wasted memory, and incorrect or uninitialized values of the address length field
- Remove some buggy code that wasn't used anywhere
- General refactoring to reduce code duplication
- Handle an invalid address type properly
svn path=/trunk/; revision=52600
- Probe and lock pages with the KernelMode access mode instead of Irp->RequestorMode (UserMode) since we're dealing with memory that we allocated in kernel-mode
- Fixes getpeername() and HTTPS browsing in Firefox
svn path=/trunk/; revision=52597
[NTDLL]: Use original symbolic function names instead of modified ones.
[NTDLL]: Fixup some incorrect flag and typedef definitions with correct ones.
[NTDLL]: Tag functions which seem broken/lacking in functionality/very different from the rest of the code the LDR rewrite was based on.
svn path=/trunk/; revision=52595
- decrease log buffer size because ROS has a problem with unlocking it otherwise ;)
- add test for critical/guarded regions
- add test for executive resources (part 1)
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52592
- Fix CMake build
- Don't override failure status in LdrGetDllHandleEx
- Minor improvements
Fixes loading of mshtml.tlb in 2nd stage. Patch by Thomas Faber
svn path=/trunk/; revision=52591
- Don't build normaliz.dll - it's Win Vista library. It didn't even work because exports was redirected to commented out kernel32 exports. GCC ignored it but MSVC returns error in such case.
svn path=/trunk/; revision=52587
Update TRAP_ENTRY macro to generate debugging information. Now WinDbg knows how to deal with the trap functions and also shows the user mode back traces.
svn path=/trunk/; revision=52586
[NTDLL]: Fix LdrpInitializeProcess to use new loader relocation API instead of LdrRelocateImage. Warn if an EXE gets relocated since this was never supported in ReactOS.
[NTDLL]: Move APIs into ldrapi.c and delete 3 leftover cruft files.
This is the extent of my review for the moment. There are still many problems with the loader. It should've been tested/put in a branch. I found these 25 bugs in less than a day. Other problems remaining are missing SEH around loader lock, not too sure if the cookie generation is correct, cookies are ULONG_PTR not ULONGs, and haven't checked the PE loading code yet (only reviewed ldrapi.c).
svn path=/trunk/; revision=52585
[NTDLL]: LdrVerifyImageMatchesCheckSum does not actually do the check if the FileHandle is ORed with 1 (this is an optimization for KnownDLLs). Also, use EndOfFile.LowPart for the checksum, not ViewSize.
[NTDLL]: LdrpGetProcedureAddress: fix overflows, incorrect constants, incorrect memcopy call, and only run init routines if the thunk was actually snapped.
svn path=/trunk/; revision=52583
In MmDbgCopyMemory do physical memory copy, if the virtual target address is valid, but not writable. Fixes Step-Over on user mode addresses with WinDbg (can write break points now)
svn path=/trunk/; revision=52582
- Comment out the actctx redirection check because it's placed wrongly. This should fix cases when a DLL is being loaded twice or more times for the same process. Spotted by Rafal Harabien.
svn path=/trunk/; revision=52577
- Use \\KnownDlls\\KnownDllPath instead of \\KnownDlls\\KnownDlls to check known dlls patch (spotted by Thomas Faber)
- Assume known dlls are in system32 if \\KnownDlls exists but KnownDllPath not
svn path=/trunk/; revision=52574
- Signal the event immediately if the select triggers have already been satisfied
- This fixes hangs with applications that wait on the select event before performing any socket operations
svn path=/trunk/; revision=52562
- Let WSPSend handle WSPSendTo on a non-datagram socket (mswsock_new does this)
- Fixes sendto on connection-oriented sockets
svn path=/trunk/; revision=52560
- Return the amount of data sent in XXXSendDatagram() functions
- Fixes send() and sendto() return values for datagram sockets (bug exposed by the AFD send user-mode IRP rewrite)
svn path=/trunk/; revision=52559
- Move read-only data into data section (allows to boot with GRUB again).
- Discard .drectve sections.
- Silence "set but not used" warnings.
svn path=/trunk/; revision=52557
- Let WSPRecv handle WSPRecvFrom on a non-datagram socket (mswsock_new does this)
- Fixes recvfrom on connection-oriented sockets
svn path=/trunk/; revision=52553
- Rewrite user-mode send IRP handling so send errors get returned to the user properly
- Fixes OskitTCP: ESHUTDOWN/EPIPE debug log spam when a chargen telnet session is terminated by the remote side
svn path=/trunk/; revision=52552
Fix 2nd stage pool corruption.
CCBs are organized in a linked list in the corresponding Fcb, which is protected by Fcb->CcbListLock. They are also linked together server <-> client. This way they can be referenced without holding the lock. This lead to a race condition where a CCB's link was modified after the CCB was deleted. Fix this by using a reference counter and adding appropriate functions. Also make use of pool tags.
svn path=/trunk/; revision=52547
- Fill in WINDOWINFO.cbSize before calling GetWindowInfo
- Handle invalid regions passed to WM_NCPAINT
- Handle WM_NCUAHDRAWCAPTION and WM_NCUAHDRAWFRAME without doing anything there for now
svn path=/branches/GSoC_2011/ThemesSupport/; revision=52542
- prepare the device object for being opened multiple times (required for communication with special-purpose drivers)
- more verbose DPRINTs
- implement utility functions for app<->driver communication
- misc fixes
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52541
- Remove formatting and partitioning hacks which are unnecessary after fixing the invalid signature bug in the FAT formatting code
svn path=/trunk/; revision=52540
- Add support for AFD_INFO_INLINING_MODE
- Add support for setting AFD_INFO_RECEIVE_WINDOW_SIZE and AFD_INFO_SEND_WINDOW_SIZE
- None of this is used by our current msafd but mswsock_new uses it
svn path=/trunk/; revision=52538
- Pass the non-blocking information as a boolean (mswsock_new does this too)
[AFD]
- Don't block waiting for a connection to accept on a non-blocking listener
svn path=/trunk/; revision=52537
- Fix RtlFindSetBits to search for bits before the HintIndex as well
- Remove a wrong (and commented out) ASSERT
- Fix MSVC warnings
svn path=/trunk/; revision=52536
- rework the kmtest loader application to follow everything else's coding style and to allow user-mode test parts
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52519
apply changes r52501 for lwIP branch
- Fix binding to an unspecified port on a connect so that it works reliably by asking the TCP library for a free port instead of assuming that one we have is free
- Fix binding to an unspecified port on a listen which previously would result in the address file not having information stored about the port number assigned
- Fix a nasty bug which resulted in us binding to an arbitrary port during a connect even when the client wanted a specific port
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52517
merge r52497
- Rewrite disconnect to work asynchronously
merge r52492
- Fix several bugs related to the FD_WRITE event
- We would not set the sendable poll state if the socket had data waiting to be sent even if there was buffer space
- We did not set the poll state after performing a send
- We did not clear the sendable poll state if we ran out of buffer space
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52516
- Fix KdComponentTableSize
- Make assembly instruction in KiCallUserMode MASM compatible
- silence some KD internal debug prints
The MSVC bootcd boots to desktop, but crashes with BAD_POOL_HEADER before 2nd stage setup starts.
svn path=/trunk/; revision=52513
- Only wait to bind to a port if INADDR_ANY is specified
- Fixes EADDRNOTAVAIL when running Firefox and ws2_32_winetest sock (which now gets further but still doesn't complete)
svn path=/trunk/; revision=52511
- We really do need to count the outstanding send IRP in our pending send count (with a detailed explanation of the reason included in the code)
- Wait on an outstanding send IRP to dispatch the disconnect
- When we receive a FIN from the other side and our receive comes back with 0 data, only close receiving on the socket because sending is still legal in this state
- Fixes many bugs on the ws2_32_winetest sock and likely lots of other partial disconnect related stuff
- Retest network related hanging bugs after this please (ftp dir hang is fixed)
svn path=/trunk/; revision=52510
Fix KiConvertToGuiThread on MSVC builds, by using a raw assembly function instead of a (broken) inline asembly function
MSVC built kernel boots further now.
svn path=/trunk/; revision=52509
- Change an ASSERT to a KeBugCheck, since the assertion can fail for any invalid memory access and this is not an internal Mm failure.
- Remove 2 cases, that "Should NEVER happen on ARM3!!!", but can very well happen.
- Do NOT make the code cleaner, by releasing the PFN lock in the same function that acquires it, but keep it 2 functions down. This is because it *SHOULD* be that way, since some internal undocumented functions, that we do not implement but that are (theoretically) called from here, also do release the PFN lock. Thanks Alex for explaining this.
svn path=/trunk/; revision=52507
- Remove socket state checks from OskitTCPRecv and OskitTCPSend
- Oskit already returns ESHUTDOWN when sosend is called on a socket with SS_CANTSENDMORE so it is unnecessary
- The check in OskitTCPRecv broke returning success and 0 bytes on a graceful shutdown (We got the FIN, set SS_CANTRCVMORE, signaled the connection, tried to read and failed with ESHUTDOWN instead of succeeding with 0 bytes)
svn path=/trunk/; revision=52505
- Signal readable when read is closed and writeable when write is closed
- This seems counterintuitive on the surface but signaling the socket in this way ensures that pending reads and writes are cancelled if either side shuts down send/receive
- Graceful close (recv() returns 0 bytes and FD_CLOSE is signaled when we receive a FIN) works much better
svn path=/trunk/; revision=52504
- Implement support for timing out disconnects
- Add debug prints for oskittcp errors
[MSAFD]
- Set a disconnect timeout value that actually makes sense (1 second vs 100 nanoseconds)
svn path=/trunk/; revision=52503
- Fix binding to an unspecified port on a connect so that it works reliably by asking the TCP library for a free port instead of assuming that one we have is free
- Fix binding to an unspecified port on a listen which previously would result in the address file not having information stored about the port number assigned
- Fix a nasty bug which resulted in us binding to an arbitrary port during a connect even when the client wanted a specific port
- Revert the hack that partially fixed this before
svn path=/trunk/; revision=52501
- Really fix loading the MSVC built kernel, by moving the FAT sector buffer beyond the area for freeldr.sys
- Load segment only in one place
svn path=/trunk/; revision=52500
- Set SO_REUSEPORT on new sockets to allow binding to ports shared by sockets which are closed according to upper layers of tcpip but still pending a close in oskittcp (see the BSD try-10-times hack)
- Fill all members of the sockaddr_in struct when fulfilling a query accept request
svn path=/trunk/; revision=52498
Add some code that does ... nothing, but fixes 2nd stage boot with msvc. To figure out why this is needed, is left as an excercise to the reader :)
svn path=/trunk/; revision=52496
- Fix several bugs related to the FD_WRITE event
- We would not set the sendable poll state if the socket had data waiting to be sent even if there was buffer space
- We did not set the poll state after performing a send
- We did not clear the sendable poll state if we ran out of buffer space
svn path=/trunk/; revision=52492
Remove the old boot method.
Its not needed anymore, since booting with more than 4GB works fine with the new method now. It didn't work anyway since the neccessary code in the kernel has been #if'ed out in r49445
svn path=/trunk/; revision=52491
- add skip() functionality (with more useful syntax than Wine's)
- add format attributes for GCC
- do not show the full path of the source file for failed tests
- general cleanup
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52489
Revert r52482, it didn't really work. Instead grab the StallScaleFactor from the PCR and use this as a base for the repeat count.
svn path=/trunk/; revision=52487
- fix system crash when failing to connect to a remote host
- Set address length in the TDI address struct
- Simplify copying the address
- Zero the sin_zero member
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52485
Use KeStallExecutionProcessor, once its ready (after KdDebuggerInitialize1 was called) for the receive timeout. Repeat 100 times with 1 ms wait. This seems to work quite well, but might be tuned for better values.
svn path=/trunk/; revision=52482
On amd64 builds, where we don't have underscored symbols, create an alias fabs for the function _fabs, since we cannot create fabs as a label directly
svn path=/trunk/; revision=52481
change the way import libs are created for MSVC. Since we don't use underscores in symbols on other architectures than x86, its not possible to create labels like "div", since it conflicts with the opcode. Therefore add a prefix and redirect all exports to the prefixed version when building an importlib
svn path=/trunk/; revision=52476
- unify spec2def and add_importlib_target, thanks to cmake_parse_arguments.
- also generate stub files only when necessary.
- better precompiled header support, commented out as it wasn't tested on any other machine than mine, and it depends on cmake features which don't exist yet.
For now only for gcc toolchain
svn path=/trunk/; revision=52460
- Fix Zw-systemcall stubs on MSVC
- Fix systemcall return stubs on MSVC
-> MSVC kernel boots and usetup works!
Didn't test anymore, because there is a massive atapi debug spam
svn path=/trunk/; revision=52457
- Don't try to load referenced dlls in any other dir than system32. A second attempt would fail on the first dll that was already loaded and the function would bail out before any missing dll could be loaded. This worked only by chance on gcc builds, because the 1st import was bootvid, which was always loaded in the 2nd attempt from system32, and the function failed after that, since hal and kdcom were already loaded.
svn path=/trunk/; revision=52452
- The long awaited LDR rewrite, a commit for testing the new implementation. In case of serious problems it may be reverted (revert should be approved by me).
- Its features include:
* A proper ...everything. Process, thread initialization codes, DLL loading (including compatible path lookup, and compatible/proper loading order of the dependent DLLs, including their initialization) and mapping and section creation, reference counting, relocations, good and understandable PE code for walking import descriptor, snapping, etc etc. Hacks--; GoodCode++;
* Activation contexts operations are now being performed compatible to how Windows performs them (though the actual actctx implementation is still Wine's, it was modified to be compatible). Previously, actctx stuff was added to the ldr code like a pepper is added to the soup: in different places until it starts to work.
* Partial DLL redirection implementation.
* Possibility to support Shim engine and app compat stuff in future.
* More cool stuff, just browse the code.
- I fixed all regressions I could find but one (hang during shutdown of the 3rd stage). The purpose of this commit is to seek and destroy the regressions I couldn't find (if there are any).
- Some of the old rarely called ldr code still remains in startup.c and utils.c. They are subject to be rewritten/removed soon, and every remaining old function is marked with a respective DPRINT1 to see when it's being called.
svn path=/trunk/; revision=52446
- merge r52438
- Fail SIO_GET_EXTENSION_FUNCTION_POINTER
- Fixes the crash during ws2_32_winetest sock
[AFD]
- merge r52441
- Queue the user-mode connect IRP before calling TdiConnect to avoid a race if the TDI_CONNECT IRP is completed before we get a chance to queue the user-mode IRP to accept the connection
- Change the non-blocking TDI helper functions to always return STATUS_PENDING if the completion function will be called to avoid duplicate handling of IRPs
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52443
- Queue the user-mode connect IRP before calling TdiConnect to avoid a race if the TDI_CONNECT IRP is completed before we get a chance to queue the user-mode IRP to accept the connection
- Change the non-blocking TDI helper functions to always return STATUS_PENDING if the completion function will be called to avoid duplicate handling of IRPs
svn path=/trunk/; revision=52441
* Add a way to use WDK headers (psdk, crt and ddk) instead of ours. To use it, just alter USE_WDK_HEADERS in configure.cmd to be 1 instead of 0.
svn path=/trunk/; revision=52439
- Fail SIO_GET_EXTENSION_FUNCTION_POINTER
- Fixes the crash during ws2_32_winetest sock
- In my testing, "sock: 4374916 tests executed (0 marked as todo, 93 failures), 6 skipped."
svn path=/trunk/; revision=52438
- Fix copying of the remote address when satisfying an accept request
- accept() and WSAAccept() no longer report strange peer addresses
- Fixes a few ws2_32_winetest sock tests
svn path=/trunk/; revision=52437
- Lock the volume before formatting to prevent concurrent access during formatting
- This should (in theory) fix our formatting issue on mounted devices but currently we don't support FSCTL_LOCK_VOLUME in our FS drivers
svn path=/trunk/; revision=52435
Fix mshtml.tlb build with Visual Studio.
With mshtml_tlb and mshtml.tlb targets, the solution contained 2 mshtml_tlb targets.
svn path=/trunk/; revision=52433
Optimize KiSystemCallTrampoline inline assembly for MSVC. save one register, use eax for the handler address and don't assign to an intermediate variable before returning (hint by Alex)
svn path=/trunk/; revision=52430
- Hold on to the disconnect IRP until all pending sending IRPs are sent to the transport driver if a controlled disconnect was requested (merge of r52415)
[TCPIP]
- add some mroe debug statements to track new current sync problem
- remove some old debug statements that are useless now
[lwIP]
- add some mroe debug statements to track new current sync problem
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52424
- Register for asynchronous buffer notifications from oskittcp which will fire when a packet is sent
- Fixes sending pending packets that can't fit in the window and fixes the graceful shutdown hang
svn path=/trunk/; revision=52422
* Skip some modules that do not compile, in MSVC build.
* Welcome, MSVC BootCD. I'm sure ros devs will fix the s.it out of you.. Right ?
svn path=/trunk/; revision=52419
- Wait until the all pending sends are serviced before shutting down the socket on a graceful disconnect
- Cancel pending sends and receives on abortive disconnect
- Remove the nasty hack that was the completion queue and replace it with async completions like the lwIP implementation does
- There is a bug with the graceful disconnects which occurs if the graceful disconnect cannot be serviced immediately (a very rare case in my testing) and results in the shutdown() call stalling forever because oskittcp never indicates that send is possible again (which would allow pending send IRPs to be serviced and the shutdown IRP to be completed)
svn path=/trunk/; revision=52416
- Hold on to the disconnect IRP until all pending sending IRPs are sent to the transport driver if a controlled disconnect was requested
svn path=/trunk/; revision=52415
[NTOSKRNL]: Also, fix locking order such that the current PRCB is only read after the IRQL is raised to SYNCH_LEVEL. I'm too tired to tell if this is a requirement, but the PDF shows it being done that way in Windows, so I've re-ordered the operations to match. Intuition would indicate that this shouldn't matter, as fs:1C shouldn't change from under us in the middle of execution, but why argue with public source code?
vicmarcal said it's okay if this breaks the build (I don't have any way to build ReactOS anymore).
svn path=/trunk/; revision=52412
- Revert part of r52239: Fix support for \r and do not handle backspace
- Simplify it a bit
[NTOSKRNL]
- Add backspace support to KdpScreenPrint
- Do not draw boot text bitmap if InbvDisplayString is called with "" or "." (matches NT behavior)
svn path=/trunk/; revision=52409
Igor Paliychuk:
* Convert UKRAINIAN and RUSSIAN localization files to utf-8 in the dll folder.
* Add UKRAINIAN translation for msports.
* Some Wine modules also were converted (already sent upstream).
svn path=/trunk/; revision=52407
- Don't attempt to service a request if the only flag specified was SEL_FIN
- Handle TDI_DISCONNECT_RELEASE properly
[OSKITTCP]
- Verify that the socket is in a legal state to do a send or receive
- Don't indicate send and receive events when the socket is not connected or has been been closed in that particular direction
- Add a small hack to soshutdown so shutdown in the send direction works correctly (ie. doesn't shutdown receive)
- Fixes the hangs in ws2_32_apitest ioctlsocket and ws2_32_apitest recv
svn path=/trunk/; revision=52405
- Display boot blue bitmap when native application or kernel calls NtDisplayString with string displaying enabled
- Fix drawing order in Phase1InitializationDiscard
svn path=/trunk/; revision=52404
- add assembly version of call_method for MSVC
- fix a syntax error (sizeof(int[3]) is not valid C)
- move definitions to the top of the CMakeLists.txt
svn path=/trunk/; revision=52401
- Add msvc.S containing call_stubless_func in assembly
- Fix spec file, 2 stub functions are already implemented
- fix use of GetExceptionCode(), it cannot be used in that context.
svn path=/trunk/; revision=52399
- get rid of unused OskitDumpBuffer
- make the code in AFD better conform to ROS coding guidelines
- fix some MSVC compilation errors (still not fully compiling with MSVC yet)
[IP]
- Add a sanity check in the TCPAcceptEventHandler
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52398
- Remove foreign identifier field from the tcp_pcb structure. This was added for some debugging purposes sometime ago but now it's useless and it was just polluting lwIP code.
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52396
- Abort the connection if the socket is not in LISTEN state, instead of closing it
[TCPIP]
- move redundant socket closure where so we don't try to free uninitialized variable
- in case our socket got closed or for some reason doesn't exist return a default success
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52395
- define _MIDL_USE_GUIDDEF_ when building iid libraries, so that the guids have declspec(selectany)
- add xml_uuids library
[MSXML3]
- Add an include for MSVC to define missing IIDs
- add CLSID_DOMDocument2
fixes build of msxml3 with MSVC
svn path=/trunk/; revision=52391
- Fix crash bug when disconnecting after accepting a connection from a client.
- Move the send and receive callback assignments to make the code more clear
- Change variable types to avoid excessive casting
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52387
Fix the nasty crash on socket closure bug. The bug was due to corrupting memory by wrongly assuming the LISTEN pcb had send, receive and error callbacks.
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52385
- don't forward _fileinfo_dll to MSVCRT, this doesn't work and its not what MS crtdll does.
- don't define _CRTIMP to _declspec(dllexport)
- link to MSVCRT when compiling with MSVC
- remove def file
svn path=/trunk/; revision=52380
- Convert format, fdebug, ntvdm and win32csr polish locale to UTF-8;
- Clean up remainder of #pragma definitions in individual language rc files. This should prevent bugs like one with chineese locale in calc;
- Move bulgarian and romanian locales, already converted to UTF-8 to proper section in module rc file;
svn path=/trunk/; revision=52376
- Move verification code to the BlockDev* functions (like CDFS does) so functions that call those directly will have verification handled for them
- Add verification handling for IOCTL requests
svn path=/trunk/; revision=52374
- Handle STATUS_VERIFY_REQUIRED for write requests
- Clear the device to verify after retrieving it
- Fixes write failure after changing floppy media
[CDFS]
- Reissue the I/O request if IoVerifyVolume was successful
- Fixes read after a media change
svn path=/trunk/; revision=52372
- add START_TEST macro in case test functions ever have to be passed an argument
- add a few ok_* convenience macros
- remove unnecessary C library function renaming macros
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52371
- set PCH file after importing libraries for modules that link to msvcrt.
Otherwise it wouldn't get the added definitions and become useless.
svn path=/trunk/; revision=52369
- [TRANSLATIONS] add Simplified Chinese translation for calc
- #pragma code_page(65001) added in zh-CN.rc (needs to be reviewed)
- See issue #6319 for details.
svn path=/trunk/; revision=52368
- Fix incorrect KiUserTrap which returned 3 for UserMode traps. Kind-of spotted by Jerome, but fixed in the wrong place...ReactOS-style, as always.
-Patch by Alex Ionescu
svn path=/trunk/; revision=52366
- Set the prototypes to be long instead of LONG. This only worked in rbuild previously because WIDL incorrectly generates a LONG for long types.
- define SHANDLE_PTR and use it in IWebBrowserApp::get_HWND
- Along with fixing WIDL, __WINESRC__ and __ROS_LONG64__ should be removed as part of a future fix.
- Should now build with both rbuild and msvc
svn path=/trunk/; revision=52364
No objections on the mailing list, so I'm excluding "usbdriver" from build.
It should still stay as a reference in the tree till the new USB drivers are finalized and merged back to Trunk.
svn path=/trunk/; revision=52361
- Fix MSVC linkerflags for setupldr
- Move stack from 78000 to 98000 to fix conflicts with freeldr code, which is larger when built with MSVC
- MSVC compiled setupldr works now!
svn path=/trunk/; revision=52360
- add the proper testing framework functions for winetest-like feel
- remove thereby obsolete Log* functions
- update example test to show usage of the new macros
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52359
- Send IRP_MN_QUERY_POWER and IRP_MN_SET_POWER to all enumerated devices prior to a system power state change
- We don't mind failure right now since our drivers suck
svn path=/trunk/; revision=52358
* Temporarily add liboldnames.a, I mean, oldnames.lib, to the importlibs folder, until we get one for msvc.
* Link crtdll to oldnames in msvc build.
svn path=/trunk/; revision=52350
- Change base addresses to 0xF800 (realmode code) and 0x10000 (PE file)
- use FREELDR_BASE constant in bootsectors
- Use es segment to address PE file from realmode
svn path=/trunk/; revision=52346
- Remove commented out lines added in r52332 - clip children should not be taken into account here
[SHELL32]
- Use COPYRIGHT_YEAR definition instead of current year
svn path=/trunk/; revision=52343
- do not use RtlpCoalesceFreeBlocks to reinsert a block resulting from a split.
This bad idea was mine, introduced in r51757
svn path=/trunk/; revision=52340
Cleanup the CMakeLists.txt file. Create one shared library containing most files and only compile a few files for both freeldr and setupldr, which reduces build time.
There is only one file, that takes different compile time definitions, but since LD is stupid and cannot bidirectionally link between seperate libraries, we have to add 3 more files to each build.
svn path=/trunk/; revision=52338
- [TRANSLATIONS] convert UKRAINIAN and RUSSIAN localization strings to UTF-8.
- convert files in base, boot and subsystems folders(fontview and logon skipped)
- add Ukrainian localization for subst
- fix filename for Ukrainian resource of dxdiag
svn path=/trunk/; revision=52336
- Don't clip siblings in co_WinPosSetWindowPos when window has no WS_CLIPSIBLINGS style. Fixes ReactOS about box.
- Commented out line with WS_CLIPCHILDREN support needs more work
- Simplify co_WinPosSetWindowPos a bit
svn path=/trunk/; revision=52332
- reference streamout.c and provide a wctomb stub to have an any-IRQL vsnprintf implementation for the test framework
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52330
- Convert Boot*LinuxKernel to new realmode call mechanism
- Remove switch to realmode when multiboot fails
- remove switch_to_prot and switch_to_real completely
svn path=/trunk/; revision=52329
- Use .586P for 16 bit code to allow privileged instructions, like lidt/lgdt. Found by ThFabba
- Add a 0 byte at the end to make ML fill the file up to the required length
svn path=/trunk/; revision=52327
- Instead of using the far jmp when switching to protected mode to go to the entry point, use a relative absolute jmp instruction, manually encoded
- Remove the END from fathelp.inc, that was terminating the code
- fix code for ML
svn path=/trunk/; revision=52325
- Make .asciz macro accept a varaiable number of arguments
- add CR and LF definitions, since MASM doesn't like "\r\n" codes
svn path=/trunk/; revision=52323
[SACDRV]: Implement DriverEntry point which connects to NTOS HeadLess Module (HDL). If HDL enabled through FreeLDR (/EMSPORT), connection should work.
[SACDRV]: Fix typo's.
svn path=/trunk/; revision=52320
[NTOSKRNL]: Export HeadlessDispatch.
[NTOSKRNL]: Add more missing definitions to internal Headless header. Alex: Move to NDK?
[NTOSKRNL]: Headless header uses SYSTEM_TIMEOFDAY_INFORMATION, which is only in NDK. So hdl.h needs NDK.
svn path=/trunk/; revision=52319
- Call the completion handler asynchronously to avoid a deadlock if AFD issues a synchronous request in its completion function
- Complete all pending requests before closing the socket
- Fixes send and receive on accepted sockets
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52307
* Don't mark DllRegisterServer and DllUnregisterServer as private, since rsabase forwards to them. Fixes rsabase msvc build.
svn path=/trunk/; revision=52303
* Fix exports. Thank you, gcc toolchain, for letting this slip without any problems.
* Brought to you by the msvc toolchain.
svn path=/trunk/; revision=52298
- Revert r52289 (I guess 0 is a valid minor function)
- Add a comment to prevent anyone else from making the same mistake I did
svn path=/trunk/; revision=52290
- Fix a critical bug in IopDeviceFsIoControl which caused all FSCTL requests sent via NtFsControlFile and ZwFsControlFile to fail
svn path=/trunk/; revision=52289
- Set the BIOS drive number depending on the media type (independent of the actual number of drives) because it's only relevant for the boot drive
- See http://support.microsoft.com/kb/140418 for details
svn path=/trunk/; revision=52286
- Write the boot sector signature to the new boot sector after formatting
- Volumes formatted after setup are now mountable
svn path=/trunk/; revision=52285
- Use the new realmode code for PxeCallApi, SoftReboot and ChainLoadBiosBootSectorCode
- Remove o386cpi.S from build, the functions are not used
- Use constants for function ids
- Fix a warning
svn path=/trunk/; revision=52284
- Determine the correct FAT type based on the size and offset of the partition if the current type is not a FAT type (another FS type or unformatted)
- Formatting a floppy, blank partition, or other non-FAT formatted partition from within ROS no longer fails
svn path=/trunk/; revision=52278
- Setup the IDT in C code instead of using 16 bit assembly and relying on trap handlers being below 64k
- Make the trap handler code MSVC compatible
- Add back multiboot code
svn path=/trunk/; revision=52256
- Add underscore to names of trap handlers and add C prototypes (will be needed later)
- Move multiboot code to seperate file multiboot.S, not compiled atm, the code wasn't in use anyway
- Remove EnableA20 dummy
svn path=/trunk/; revision=52253
* Rename _math.h to math.h and merge r50732 from the cmake branch.
* Remove msvchelper.h as it's not needed anymore for d3d8 and d3d9.
* Skip ftp, libxslt and ksproxy from msvc build.
* Dedicated to igorko.
svn path=/trunk/; revision=52251
use baseaddress of ßx10000 on MSVC builds. This doesn't yet work on gcc, because we still have 16 bit code inside the PE, that must reside below 64k
svn path=/trunk/; revision=52249
- Finally say goodbye to my "wcsfuncs" experiment started in 2008.
We have (and always had) Wine's "unicode" library for this, which does the same and is more complete.
Now that host tools only use include/host, we only need to include <wine/unicode.h> and it will choose the right file in either include/host or include/reactos.
- Fix some include guards in PSDK headers to match the MS ones.
- Always define WINE_UNICODE_API to an empty string instead of doing this separately for every module.
- Cleanup some empty directories or non-existing include pathes.
svn path=/trunk/; revision=52248
- Disable displaying string by bootvid after successful kernel phase 1 initialization, so there won't be graphical glitches at the top of screen if system has screen debugging enabled
- Make screen black right after phase 1 initialization finish instead of waiting 5 seconds. It removes black rectangle which appeared for me in explorer.
- Initialize some global variables in inbv.c (it was done by compiler before)
svn path=/trunk/; revision=52244
- Fix scrolling if ScrollRegion contain whole screen.
- Add backspace support
- Remove unneeded code
- Now screen debugging in BSOD works much better :)
svn path=/trunk/; revision=52239
- Return const pointers instead of converting them
- Fix MemoryMapSizeInPages calculation
- Don't use pages after 4GB limit on x86 (PAE is not supported yet)
- Should fix starting ROS on 4GB machines
See issue #6031 for more details.
svn path=/trunk/; revision=52234
* Skip ftp, libxslt and ksproxy from msvc build.
* Remove msvchelper.h as it's not needed anymore for d3d8 and d3d9.
* Merge r50731 from the cmake branch.
* Merge r50732 from the cmake branch and apply the same fix to state.c.
svn path=/trunk/; revision=52232
- Fix a bug on line 191, where the pending send requests for datagrams weren't canceled (the receive list was being used instead
- Also added some formatting
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52229
- Only indicate that receive is possible if there are no pending receive IRPs waiting for data
- Don't attempt to receive on a connection which has been closed in the receive direction
svn path=/trunk/; revision=52224
- FD_CLOSE is only for remote disconnects so don't send it when the app closes its own socket
- Validate the user-mode event handle properly
svn path=/trunk/; revision=52223
Start moving the 16 bit code into the raw binary chunk.
We now switch to protected mode before jumping to the PE entry point
svn path=/trunk/; revision=52221
- use a shared memory buffer for storing test results to provide seamless communication between all test parts
- Now user mode code can easily add messages to the buffer
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52216
- add a minimal example test so there's something to see when running kmtest
- remove _DLL and __USE_CRTIMP in accordance with r52213
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52215
* Automatically add _DLL and __USE_CRTIMP definitions when linking to msvcrt.
* Separate some flags from definitions (using add_compiler_flags).
svn path=/trunk/; revision=52213
* Move cmake files to the cmake folder.
* Rename msc.cmake to msvc.cmake and toolchain-mingw32.cmake to toolchain-gcc.cmake.
* Add compilerflags.cmake to group macros that handle different compiler flags.
* Move add_linkerflag and set_unicode macros to compilerflags.cmake. They were duplicated in gcc.cmake and msvc.cmake.
* Add add_compiler_flags macro to handle CMAKE_C_FLAGS and CMAKE_CXX_FLAGS properly, and replace add_definitions calls with add_compiler_flags calls when we're adding compiler flags, not definitions.
* Reorganize gcc.cmake to have compiler flags then linking and compiling rules.
* Move CMAKE_ASM_COMPILE_OBJECT out of toolchain-gcc.cmake into gcc.cmake and add ${CMAKE_C_FLAGS} to it, now that flags are properly set.
* Don't pass <FLAGS> (compiler flags) when linking.
* Without the (commented out) linker script flag, CMAKE_SHARED_LINKER_FLAGS_INIT is the same for both i386 and amd64. Deduplicate it.
* Fix CMAKE_C_STANDARD_LIBRARIES comment.
* Use <FLAGS> in msvc CMAKE_ASM_COMPILE_OBJECT instead of manual includes.
* Don't set ntdllsys linker language to C, it's an asm lib.
* Remove ros_cd.cmake as it's not needed anymore.
* Define _CRT_SECURE_NO_WARNINGS when compiling host tools with msvc.
svn path=/trunk/; revision=52210
Make freeldr/setupldr composed of a PE file and prepended raw 16bit code. The 16 bit code starts with the usual fathelp code. This brings back support for fragmented fat12/16 disks.
Later all 16 bit code is supposed to go into the raw binary chunk to be able to build freeldr with MSVC.
svn path=/trunk/; revision=52204
- Fix a memory leak during graceful socket closure
- Print a message if not all data is taken in a receive request
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52201
- Implement LibTCPShutdown
[IP]
- Replace an incredibly broken implementation of TCPDisconnect which caused memory corruption with a working one using LibTCPShutdown
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52200
- Add iexplore.exe to bootcd (it's being copied to its localized location in 3rd stage via runonce due to lack of support for additional paths in the installer)
- Register default system web browser, fixes launching URLs from the run dialog.
- Firefox 3.X/4.X still don't register as default web browsers (bug 4444)
See issue #1461 for more details.
svn path=/trunk/; revision=52196
- Fix linger and graceful disconnect
- Fix a crash in WSPGetSockName and WSPGetPeerName exposed by ws2_32 winetest sock
svn path=/trunk/; revision=52193
* Move CMAKE_RC_COMPILE_OBJECT and CMAKE_ASM_COMPILE_OBJECT to msc.cmake.
* Fix CMAKE_ASM_COMPILE_OBJECT to account for the needed includes and defines.
* Add CMAKE_ASM_CREATE_STATIC_LIBRARY to handle our asm static lib.
* Don't include chkstk_ms.s into chkstk and libcntpr for msvc.
* Add missing chkstk -> asm dependency.
* Don't make an assumption about the location of genincdata.dll.
* Move definition on top of the included cmake files in the root cmake file.
* This brings the VS solution to a working state, many modules can compile now.
svn path=/trunk/; revision=52190
* Copy CMakeDetermineASMCompiler.cmake from CMake 2.8.4 so that it's used with previous versions too. Fixes ASM compiler ID issue with MSVC.
svn path=/trunk/; revision=52187
- Rewrite user-mode request locking to fix several bugs
- IRP_MJ_READ and IRP_MJ_WRITE doesn't crash
- Revert the IoModifyAccess change (the Write parameter was relevant to the buffer not the request)
- Use GetLockedData to retrieve a locked buffer instead of accessing it manually
- Lock several requests which were not locked and could cause a crash
- ASSERT that the in flight request is the IRP being completed
- Fix a socket closure bug so shutdown(SO_SEND) will not prevent packets being received
- Don't count bytes used in the receive content
- Don't free the datagram buffer if it is a peek request
- Requeue the datagram buffer at the head of the list if it is a peek request
- Fix a bug which could cause corrupted data to be received if multiple datagrams come in before the app receives them
- Make sure that we're not overwriting an existing in flight request when creating a new one
- Perform an implicit bind in AfdPacketSocketWriteData if it is not already bound
- Implement batching several user-mode send IRPs into one TDI request (if there is already one pending)
- Fix a potential crash if a connectionless send IRP is cancelled
svn path=/trunk/; revision=52186
- Do not create/open the CurrentVersion key in a single call to NtCreateKey because its parent key might not exist yet.
See issue #6302 for more details.
svn path=/trunk/; revision=52184
Bind issue seems solved now. Server app can be restarted without failing at binding. More information is at r52166. Patch by cgutman.
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52170
- Fix a bug in the new fat32 bootsector code. The old code was using a 32bit push, the new code uses 2 16 bit pushes
- enable new fat32 and isobtrt bootsectors
svn path=/trunk/; revision=52165
- Fix hot tracking but does not clear it properly. See bug 6257.
- Ported NC_HandleNCHitTest from wine, will sync User32 later.
svn path=/trunk/; revision=52163
- add simple loader application that can list/run tests from a driver via DeviceIoControl, and prints output to the console
svn path=/branches/GSoC_2011/KMTestSuite/; revision=52157
- export obj2bin on gcc builds, too
- Add new macro CreateBootSectorTarget2, which uses portable assembly and use it with isoboot.S. I will replace all bootsectors with the new code one at a time, and in the end we can eventually drop nmake
- add wrapper isobtrt.S, which defines ROS_REGTEST and includes isoboot.S
svn path=/trunk/; revision=52156
Fix portable bootsector and make it gas compatible (its opcode identical to the nasm compiled one and works correctly)
svn path=/trunk/; revision=52153
- Fix a bug in some completion routines that resulted in failures being treated as successes
- Fix various cancellation bugs
- Fixes termination of programs that open listening sockets (Firefox, server.exe, tcpsvcs.exe, telnetd.exe, etc) and should increase stability in these programs as well
svn path=/trunk/; revision=52152
Fix the "portable" version of the fat32 bootsector
Its now mostly identical to the one we currently use, only a few encoding differences
svn path=/trunk/; revision=52147
- Add new tool obj2bin, that converts a coff object file into a raw binary file
- Move pe/coff types into a common header file pecoff.h
svn path=/trunk/; revision=52133
Implement support for SafeMode boot
Patch by Adam Kachwalla (geekdundee at gmail dot com)
See issue #6275 for more details.
svn path=/trunk/; revision=52132
* Temporarily disable PCH support, as it breaks build for windows users with ccache (most importantly our windows buildbot).
svn path=/trunk/; revision=52131
Remove useless casts to volatile ULONG. It doesn't make sense to cast something to volatile, after the pointer is already dereferenced. But we don't need this here anyway.
svn path=/trunk/; revision=52118
Fixes to r52065:
- In IopInterlockedDecrementUlong() and IopInterlockedIncrementUlong(), cast received value to volatile before de/incrementing it so ensure strict ordering with spin locks acquisition.
This was brought to you by Alex Ionescu.
svn path=/trunk/; revision=52117
* Fix several modules to compile with the recent gcc versions.
* ReactOS now compiles with mingw-w64 gcc 4.6.0 and boots to shell.
* Dedicated to dreimer.
svn path=/trunk/; revision=52114
Timo Kreuzer:
* Change inline assembly constraint from "r" to "q". "r" means any kind of register, including *si and *di, these cannot be used for 8bit operations on x86, only x64 has sil, dil registers. But gcc is stupid and doesn't get that if we don't tell it explicitly, by using "q", which will only allow a, b, c, d registers. Fixes an assembler error messages that can occur under certain circumstances.
svn path=/trunk/; revision=52100
NOTE: Windows does not do this, opting instead to force manufacturers/use pull-up resistors/reconfigure the ARM Bus to map RAM at 0x00000000. For wider portability, I believe it makes more sense to simply do this "trick" in the boot loader.
svn path=/trunk/; revision=52098
- Create an empty TRANSPORT_ADDRESS for the local address when doing an implicit bind so TCP/IP can choose its own local port and address instead of trying to bind the remote address and port
- Fixes another critical networking bug
svn path=/trunk/; revision=52092
- Add a Next member to CONNECTION_ENDPOINT to allow multiple connections to be associated with a single address file while not overwriting pointers, dereferencing other people's connection objects, and causing horrific amounts of memory corruption
- Add several sanity checks to prevent this from happening again
- Don't try dereference address files and connection endpoints in the free functions; there should be none associated since r52083 (sanity checks ensure this)
- Don't hold an extra reference to the address file when creating a listener; this reference is implicit
- This should greatly increase reliability of activities that open lots of sockets such as web browsing and running servers on ROS
- This also fixes most issues of not releasing a server port when the listener is closed
svn path=/trunk/; revision=52086
* Fix unix build. We need to quote the file paths in reactos.dff for cabman to parse them correctly. Dedicated to Arty.
svn path=/branches/cmake-bringup/; revision=52085
- server.exe and client.exe do a simple exchange of messages (each sends one message to the other) and then terminate.
- server_multi.exe and client_multi.exe exchange 3 message with one another before termination.
- client_delayed.exe waits one second before sending any message to the server after a connection ahs been made (otherwise identical to client.exe)
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52082
- add definition for KeMemoryBarrierWithoutFence
- add x64 version of KeMemoryBarrier and some related defines
- Use KeMemoryBarrierWithoutFence in spinlocks as suggested by Alex
svn path=/trunk/; revision=52079
- Add explicit memory barriers to KxAcquireSpinLock, KxReleaseSpinLock inline functions and KeTryToAcquireQueuedSpinLock, KeTryToAcquireQueuedSpinLockRaiseToSynch in the UP case. This will prevent the compiler from reordering memory access instructions across the boundaries of these functions, even when being inlined.
- Use the inline functions in x64 spinlock functions, too
svn path=/trunk/; revision=52078
Hackfix to r52065:
- Comment out call to IopDecrementDeviceObjectRef in IopLoadFileSystemDriver, on Pierre's request. Needs more research. Should fix boot and testcd ASSERT in 2nd stage.
svn path=/trunk/; revision=52075
Fixes to r52065:
- Acquire resources within a critical region
- Simplified IoEnumerateRegisteredFiltersList and made its behaviour match the one described in MSDN: http://msdn.microsoft.com/en-us/library/ff548348%28v=vs.85%29.aspx
This was brought to you by Alex Ionescu.
svn path=/trunk/; revision=52073
- Don't leave DISPATCH_LEVEL while holding a spin lock acquired at DISPATCH_LEVEL
- Synchronize cancellation checking by acquiring the cancel spin lock
svn path=/branches/usb-bringup/; revision=52071
- Fix a bug in device removal which resulted in an IRP_MN_CANCEL_REMOVE_DEVICE request not being sent if the initial IRP_MN_QUERY_REMOVE_DEVICE failed
svn path=/trunk/; revision=52066
- renamed Io volumes global to match Windows names (in case of debug)
- renamed IopDereferenceVpb() to IopDereferenceVpbAndFree(), IopReferenceVpbForVerify() to IopReferenceVerifyVpb(), IopInitializeVpbForMount() to IopMountInitializeVpb(), IopLoadFileSystem() to IopLoadFileSystemDriver()
- implemented IopDecrementDeviceObjectHandleCount(), IopInterlockedIncrementUlong(), IopInterlockedDecrementUlong(), IopNotifyAlreadyRegisteredFileSystems() and IoEnumerateRegisteredFiltersList()
- halfplemented IopDecrementDeviceObjectRef()
- implemented check for already registrered notification in IoRegisterFsRegistrationChange()
- implemented sending notifications for already registered drivers to registrant in IoRegisterFsRegistrationChange()
- implemented VPB freeing in IopDereferenceVpbAndFree()
- acquire Io volumes lists once and forever for system shutdown, instead of keeping acquiring and releasing
- reference device object in IopShutdownBaseFileSystems() before sending it the shutdown IRP. To ensure to keep it valid till the end
- added a FS driver registration operations counter
- use this counter to handle failed mounts
- fixed: release locks before calling driver for mounting and reacquire them after
- fixed check for boot partition failure (and associated bugcheck): check we are in boot phase 0 or 1
- simplified lock process by using only one lock (ie removed mutex). Also use only critical region where needed
- fixed: ensure that locks are properly released when quitting a function
- fixed wrong return in IopCreateVpb()
- minor fixes around
svn path=/trunk/; revision=52065
- Reset KdbNumberOfRowsPrinted and KdbNumberOfColsPrinted before reading new command in kernel debugger so if previous command printed nearly full page, new command won't show "Press q to abort" message before processing
svn path=/trunk/; revision=52064
- Comment out getting usb device interface. It is currently unused and usbhub does not yet support it
- Forward pnp capabilities request to lower device object
- Fix compilation by Usurp
svn path=/branches/usb-bringup/; revision=52063
"According to gcc doc, if header.h is precompiled, the output should be named "header.h.gch" and not anything else."
Knowing it was the first step before actually doing it :-/
svn path=/trunk/; revision=52061
fix pch support, with proper dependancies.
Use cmake built-in functions instead of doing weird "magic" with tabs and spaces.
According to gcc doc, if header.h is precompiled, the output should be named "header.h.gch" and not anything else.
svn path=/trunk/; revision=52058
Merge r52051 and r52052 from cmake branch:
* Improve static libs creation in *nix. Dedicated to arty.
* Add a missing dependency.
svn path=/trunk/; revision=52053
Merge r52027 from cmake branch:
* Several improvements to debug symbols handling.
* A new (killer/awesome/...etc) feature has been introduced to kdbg: argument values support. Now backtraces contain not only usermode and kernelmode addresses translated, but also the argument values passed to the functions along the trace.
* Brought to you by the Arty.
* works only with cmake builds
svn path=/trunk/; revision=52050
replace FILE IO with Memory IO functions for batch files
Patch by HansH (hans at atbas dot org)
See issue #5807 for more details.
svn path=/trunk/; revision=52047
- Massive rework of device removal
- IRP_MN_QUERY_REMOVE_DEVICE is sent to all devices that will be removed in response to a removal/ejection
- If any of the IRP_MN_QUERY_REMOVE_DEVICE requests fail, all devices which received a IRP_MN_QUERY_REMOVE_DEVICE will receive IRP_MN_CANCEL_REMOVE_DEVICE
- If everything approves the remove request, IRP_MN_REMOVE_DEVICE will be sent to children, ejection and/or removal relations, and finally the device itself
- Handle removal relations for all devices not just 1st level devices
- Remove child devices before removing the parent
- Move GUID_DEVICE_REMOVE_PENDING and GUID_DEVICE_REMOVAL_VETOED notifications to IopQueryRemoveDevice
- Implement IopCancelRemoveDevice which sends IRP_MN_CANCEL_REMOVE_DEVICE to the driver stack and sends GUID_TARGET_DEVICE_REMOVE_CANCELLED to any drivers waiting on target device change
- Use the IopRemoveDevice function to remove devices instead of calling IopQueryRemoveDevice and IopSendRemoveDevice manually so removal and/or ejection relations and children are processed
- IRPs and PnP notifications sent upon device removal (surprise and safe) should now be compatible with XP/Vista/7
svn path=/trunk/; revision=52046
- Use tags when allocation and freeing memory and define them in tag.h
- Fix some wrongly used tags when freeing
- Our new memory manager doesn't check tags when ExFreePoolWithTag is used. It will be fixed after testing
svn path=/trunk/; revision=52043
[FORMATTING]
This is just a code formatting phase. I reordered some of the code and put some debugging DbgPrint-s in order to help understand the control and data flow. Working on trying to stop the system from crashing when server socket gets closed.
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=52042
in _flsbuf, always reset _cnt to 0 and don't allocate/use a buffer when _IONBUF is set. Fixes a regression in msvcrt_winetest file.
See issue #5829 for more details.
svn path=/trunk/; revision=52040
- Copy disk.inf from trunk
- Add upper layer filter partmgr when usb mass storage device is plugged in
- Usbstor now loads and starts up
- Currently stops at loading partmgr(not yet present)
svn path=/branches/usb-bringup/; revision=52039
- Fix bug in PnpRegSzToString. The caller expects the result length to be the length of the string in bytes, not the string length
- Introduced in rev 46690
svn path=/trunk/; revision=52037
- Check if the port is usb 2.0 controller. In that case call the Usb2.0 initialize function
- Prevents crash when used with usbohci
svn path=/branches/usb-bringup/; revision=52036
- Remove the device node when the device object is destroyed to prevent PnP manager from crashing when manipulating device nodes with device objects that no longer exist
- Move removal notifications into IopRemoveDevice and call it from IopPrepareDeviceForRemoval and IoRequestDeviceEject to manage all of the removal relations for each device removed
svn path=/trunk/; revision=52034
- Fix a bug in device removal and ejection which caused the request to fail when EjectionRelations or RemovalRelations was not handled by a driver in the stack
svn path=/trunk/; revision=52033
- Implement IopRemoveDevice and IopStopDevice
- Do a graceful remove before installing new drivers (like Windows does)
- Debug log warnings abound (don't be alarmed ;)) because drivers aren't used to receiving IRP_MN_QUERY_REMOVE_DEVICE, IRP_MN_REMOVE_DEVICE and IRP_MN_QUERY_PNP_DEVICE_STATE from our (formly) crippled PnP manager
svn path=/trunk/; revision=52029
- Handle IRP_MN_QUERY_REMOVE_DEVICE
- Handle IRP_MN_QUERY_PNP_DEVICE_STATE to report that we are not disableable to the PnP manager so we don't get removed during driver installation
svn path=/trunk/; revision=52028
* Several improvements to debug symbols handling.
* A new (killer/awesome/...etc) feature has been introduced to kdbg: argument values support. Now backtraces contain not only usermode and kernelmode addresses translated, but also the argument values passed to the functions along the trace.
* Brought to you by the Arty.
svn path=/branches/cmake-bringup/; revision=52027
- Set the DNF_DISABLED flag if a driver returns with the HardwareDisabled flag set to true (but only BEFORE starting the device)
- Don't try to start a device with the DNF_DISABLED flag set
- Send IRP_MN_QUERY_CAPABILITIES (again) and IRP_MN_QUERY_PNP_DEVICE_STATE after the device has started
- Write the UINumber and Capabilities registry entries in IopQueryDeviceCapabilities so they will be updated after every IRP_MN_QUERY_CAPABILITIES is sent
- Set the DNUF_DONT_SHOW_IN_UI flag in response to the NoDisplayInUI flag set in DEVICE_CAPABILITIES or PNP_DEVICE_DONT_DISPLAY_IN_UI set after sending IRP_MN_QUERY_PNP_DEVICE_STATE
- Set DNUF_NBOT_DISABLEABLE if PNP_DEVICE_NOT_DISABLEABLE is set after sending IRP_MN_QUERY_PNP_DEVICE_STATE
- Remove duplicate code
- Fix sending GUID_DEVICE_ARRIVAL (patch by Eric Kohl)
- IRPs sent to PnP devices on enumeration and start are now completely compatible with Windows XP/2003/Vista
svn path=/trunk/; revision=52026
- Launch a new listen IRP right after the current one completes instead of waiting until the current pending connection is accepted
- Reduces the chance of AFD missing two connection requests that are issued very close together and increases network performance by allowing our connection queue to do its job
svn path=/trunk/; revision=52013
- Implement sending GUID_TARGET_DEVICE_QUERY_REMOVE and GUID_TARGET_DEVICE_REMOVE_COMPLETE notifications
- Fix sending EventCategoryTargetDeviceChange notifications
- Remove some incorrect checks (IRP_MN_START_DEVICE CAN be called for a device that is already "started")
- Check the final status of the IRP_MN_START_DEVICE request sent after resource rebalancing and send IRP_MN_REMOVE_DEVICE if it fails
svn path=/trunk/; revision=52003
- Simplify remove lock tracking block list handling.
- Correct check for unlimited locking time being allowed.
- Eliminate use-after-free after removing a tracking block.
svn path=/trunk/; revision=52002
- dll/win32 and dll/shellext converted
- currently, only shell32 is left, due to incoming change to new version, and kernel32 - need to research how it should be done.
Tested on rbuild/cmake
svn path=/trunk/; revision=52000
- Fix OHCI_ISO_TD structure. Fixes host system crashes in Windows XP
- Fix multiple bugs in isochronous transfer implementation
- Still not yet working, as the interrupt completion is not fired yet
svn path=/branches/usb-bringup/; revision=51998
- Fix a reference leak which prevented driver objects passed to IoRegisterPlugPlayNotification from being able to unload after the notification was unregistered
- Fix a non-paged pool leak
svn path=/trunk/; revision=51997
- Add stubs for CM_Set_Class_Registry_PropertyA/W.
- Add support for missing properties to CM_Set_DevNode_Registry_Property_ExA/W.
svn path=/trunk/; revision=51996
* Improve reactos.cab creation a bit: create it using a custom command instead of a custom target.. marking it as output this way ensures a proper cleanup with make clean and co.
svn path=/trunk/; revision=51987
- Get screen saver running. Need more work on WinLogon register issues and problems.
- Miscellaneous changes and cleanups.
svn path=/trunk/; revision=51971
- Load other driver sections from txtsetup.sif
- Enable a debug print
[TXTSETUP]
- Add PnP entries for storage devices
[PCIIDE/PCIIDEX]
- Add to bootcd
- Both usetup's PnP manager and the user-mode PnP manager now recognize PCI IDE channels
svn path=/trunk/; revision=51969
- PCIIDE should be installed for each PCI IDE controller and UNIATA should be installed for each IDE channel
- Remove an extra comma in the compatible adapters entries
svn path=/trunk/; revision=51968
- Don't report raw device support to allow the IDE channel driver (uniata) to take over control of the enumerated devices
- Ideally the PnP manager should deal with a driver collision like this but it doesn't currently
svn path=/trunk/; revision=51967
- Add a special case in IopLoadServiceModule for 1st stage setup when we have no service entries so we can still load drivers
- Fixes an old regression with 1st stage PnP which now can load drivers again
svn path=/trunk/; revision=51966
- Determine whether the device is really installed by reading the ClassGUID key instead of looking at the device flags
- Add a debug print for device installation (shouldn't be too spammy except and will be helpful for hotplugging)
[USETUP]
- Reenable the 1st stage PnP manager
svn path=/trunk/; revision=51963
Closing a socket while in listening state (that has not accepted any connections) no longer certainly crashes the system. There's still a tendency to crash the system sometimes, but it's related to possibly not canceling any outstanding IRPs for the listening socket.
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=51962
- Implement a simple property sheet provider for serial ports. The property sheet does not show yet because the dialog resource is not found because a call to FindResourceW fails.
svn path=/trunk/; revision=51960
- SetupDiGetClassDevPropertySheetsW: Improve the way property sheet pages are added to a property sheet. This change enables us to retrieve the number of property sheet pages provided by a property sheet provider without leaking property sheets.
svn path=/trunk/; revision=51959
- Implement function to retrieve isochronous head endpoint descriptor
- Implement function to retrieve current frame number
- Set isochronous descriptor flag for head endpoint
- Implement retrieve device status for devices
- Implement retrieve class status from device
- Partly implement isochronous descriptor handling in usb queue
- Start implementing isochronous transfers in ISUBRequest
- Code currently not tested as the Virtual Machine with XP + ReactOS usbohci driver brings down the host system when starting the iso transfers. Ironically it crashes in MS usbohci driver
svn path=/branches/usb-bringup/; revision=51957
- Disable the 1st stage user-mode PnP manager because it is creating incomplete registry keys which is confusing the real user-mode PnP manager and preventing devices installed in 1st stage from appearing in device manager
- Fixes displaying PCI bus controllers in device manager
- I need to discuss with Herve what should be done with this code
svn path=/trunk/; revision=51947
- Merge 51783
- The final revision to merge from the existing shell32 code.
- The C++ shell32 module is now fully up to date and ready for testing.
svn path=/branches/shell32_new-bringup/; revision=51941
- Merge 51436, 51448 and 51526
- Properly support WM_SYSCOLORCHANGE in ShellView window. Changing system colors in ReactOS works perfectly now :)
svn path=/branches/shell32_new-bringup/; revision=51937
- Try to read the PortName registry value and get the port number from the name.
- Determine the port number for parallel ports the same way we determine the port number for serial ports, except for the use of the com port database.
svn path=/trunk/; revision=51935
- Merge 51423, 51426 and 51435
- Remove "Play" and "Preview" translations - they shouldn't be in shell32
- Pass an argument array to FormatMessage to insert the filename details into error messages (instead of failing)
svn path=/branches/shell32_new-bringup/; revision=51934
- Add Start Menu shell32 class to registry so LiveCD has it. It's needed for explorer-new.
[FREELDR]
- Fix rbuild file
svn path=/trunk/; revision=51931
- Implement the kernel side of surprise device removal including sending the appropriate IRPs and notifying the user-mode PnP manager
- Fix the completely broken GUID_DEVICE_ARRIVAL notification (we were sending it for the parent not the children and we were sending regardless of whether any device had actually arrived)
svn path=/trunk/; revision=51930
- Inform the PnP manager that we don't support stopping devices by failing the IRP_MN_QUERY_STOP_DEVICE IRP instead of the IRP_MN_STOP_DEVICE IRP (which is not allowed)
svn path=/trunk/; revision=51928
- Fix handling of some PnP IRPs
- Handle IRP_MN_STOP_DEVICE as it travels down the stack not up
- Only complete the IRP_MN_QUERY_XXX_DEVICE IRPs if we fail them otherwise pass them down the device stack
- Handle IRP_MN_CANCEL_XXX_DEVICE as it travels back up the stack not on its way down
- Append our flags when handling IRP_MN_QUERY_PNP_DEVICE_STATE and pass it down the device stack
svn path=/trunk/; revision=51927
- Removed hack from lwIP code. Now the listening pcb is obtained in a way as to not affect pllute lwIP code.
- Also there's still a little foreign code left in lwIP, but this can be easily removed as it contains no functional purpose, except debugging.
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=51924
- Set up periodic threshold (90 %)
- Add function to retrieve interrupt endpoints
- Add function to retrieve specific descriptors from interface. These are used by HID devices such as mice / keyboards
- Add function to retrieve interrupt interval
- Enqueue all endpoint descriptors at the end of the associated queue
- Only notify hardware of insertion when it is an bulk / control request
- Scan interrupt endpoint list to find the completed transfer descriptor
- Add debugging function to print out linked endpoint descriptors
- Interrupt transfers are now implemented.
- Tested in Windows XP SP3 + Vbox 4.04 + Microsoft 5-Button Mouse. The HID mouse installs, initializes and starts up. Unfortunately the mouse does not work as expected yet
svn path=/branches/usb-bringup/; revision=51922
- Remove dead code
- Silence traces
- Pass status & DoneHead as parameters to dpc routine
- Move IUSBRequest cleanup code into new function and use it for bulk / control transfer cleanup
- Fix bugs in AllocateEndpointDescriptor. It did not take the device address into account. It also did not respect the direction of the descriptor
- Implement support for bulk transfer requests
- Handle irp completion in CompletionCallback
- Tested in Windows XP SP3 + Vbox 4.04 + USB2.0 disabled + ReactOS usbstor + USB mass storage device
- OHCI Mass storage support is now also ready
- Next interrupt transfers
svn path=/branches/usb-bringup/; revision=51917
Store the PortName value in the registry.
Please note that SetupDiCreateDevRegKey and SetupDiOpenDevRegKey are broken because they do not open or create the 'Device Parameters' subkey of a device instance key but the device instance key itself.
svn path=/trunk/; revision=51916
- Merge 51322 and 51323
- Return 1 in WM_ERASEBKGND when desktop is painted
- Fix background for desktop icons and their labels
- Moreover fixes explorer_new background
- Properly arrange icons in explorer shell view window. This also fixes navigating in files with keyboard arrows
svn path=/branches/shell32_new-bringup/; revision=51915
- Merge 50899, 50941, 50957, 50964, 51053
- Check the correct pidl instead of the free one
- Found by the amazing DPH!
- Return the icon index, not the icon id
-Fix missing "Open" menu entry on right click for All Users desktop shortcuts.
svn path=/branches/shell32_new-bringup/; revision=51910
- Merge 50114 and 50144
- Add NULL checks in IShellView and IDropTarget implementations. Should fix bug #5780.
- Bug 5756: TRANSLATION: multiple modules: Bulgarian by CCTAHEB
svn path=/branches/shell32_new-bringup/; revision=51909
- Merge 49969, refactored to use CComPtr's thanks to the new C++ model (it's sooo much cleaner now)
- Implement copy / paste for files when tree view is present
- Don't add shortcuts / executables to recent file list
svn path=/branches/shell32_new-bringup/; revision=51908
- Start to put together the logon / logoff user interface process
- It currently paints an exact pixel replica of WinXP's welcome screen in a logging off state
- Just a bit of fun at the moment, but will be part of the UI revamp we're all working towards.
svn path=/trunk/; revision=51906
- Implement ComDBClaimNextFreePort and ComDBReleasePort.
- Add stubs for ComDBGetCurrentPortUsage and ComDBResizeDatabase.
- Do not keep the port bitmap in a buffer between calls but use a read-modify-write strategy instead.
- Replace pointer-based access to the port bitmap by index-based access.
- Use ComDBClaimNextFreePort in the serial port installer.
See issue #6272 for more details.
svn path=/trunk/; revision=51901
- Add glue code for supporting iso transfers
- Remove dead from Handle
- Remove broken asserts
- Add support for string descriptors
- Add support for class specific endpoint requests (Needs be ported to usbehci)
- Link to usbd driver
- Add support for retrieving string descriptor
- Rewrite configuration descriptor handling in IUSBDevice. New code is smaller, smarter and handles a lot more cases. Needs to be ported to usbehci
- Wrap usbdlib.h include in extern c macro, it has c linkage
- Control transfers should now work
- Need to implement support for isochronous / bulk / interrupt transfers until functional
- Tested with Bluetooth USB Stick (multi function, interrupt / bulk / control / isochronous) / USB Microphone (isochronous & control) in XP SP3
svn path=/branches/usb-bringup/; revision=51900
- Fix remove locks with debug block initialization in IoInitializeRemoveLockEx(). Before only their debug block was initialized, and not the lock itself...
- Implemented support remove locks debug blocks in IoAcquireRemoveLockEx(), IoReleaseRemoveLockEx() and IoReleaseRemoveLockAndWaitEx()
This will help debugging in storage stack and is required since partmgr is using them.
svn path=/trunk/; revision=51898
- Add many missing (or previously hidden) shell COM interfaces to the public headers.
- The new C++ shell32 dll should now build.
svn path=/branches/shell32_new-bringup/; revision=51896
- Merge 48786. (This is why we moved to C++, so much cleaner than the previous C code.)
- Improve parameter checks for IShellLinkA interface.
- Fix heap corruption when an invalid pointer is passed.
- Implement IPersistFile_fnGetCurFile.
- TODO: Add same checks to IShellLinkW interface and fix IShellLink[A|W]::GetPath / SetPath / SetIdList / GetIdList logic.
svn path=/trunk/; revision=51891
- Merge 48254, 48371, 48463, 48501
- Don't worry about trying to build this dll, it needs the new PSDK interfaces which won't be committed until the very end as they'll break the existing shell32. It make take a day or two as everything has to be merged by hand.
svn path=/trunk/; revision=51890
Fixed an error in the accepting code (in rostcp.c line 576). tcp_accepted was being called for the wrong pcb. Instead of the pcb belonging to the listening socket it was being called for the pcb belonging to the newly created connection socket. In order to fix this I added an extra field to the tcp_pcb structure to hold a reference to the listening socket. This is a crude method and it will be replaced by something more elegant.
Right now however there's a slight nondeterminism regarding the connection. One it's established the message either gets through to the other side or not, randomly. This could be due to a race condition of some sorts.
Also another problem is that the server side brings down the system when closing.
svn path=/branches/GSoC_2011/TcpIpDriver/; revision=51889
- Add sanity checks
- Preserve command status value when notifying the hc that a control / bulk endpoint was added
- Re-enable root hub notification interrupt when reset port has been completed
- Dispatch processing to USBQeueu when DoneHead event arrives
- Scan endpoints to find the logical endpoint which was completed by the hardware
- Signal completion to IUSBRequest by calling CompletionCallback
- Create a final transfer descriptor which is linked to the endpoint descriptor
- Control transfers now appear to be working (Device retrieves device descriptor / configuration descriptor and succeeds in setting device address)
- Configuration parsing code needs more work now (currently fails there due to invalid / unexpected configuration descriptor from device)
svn path=/branches/usb-bringup/; revision=51888
- Enable global power mode
- Wait untill reset is complete
- Clear reset complete bit when reset is done
- Enable port
- Reset port now works
- USBOHCI still hangs after adding first control request, needs more investigation
svn path=/branches/usb-bringup/; revision=51883
Addenum. This change slipped past in commit 51873. Fix name of irot header include, which should be irot.h and not irot_s.h. Fixes rpcss build in MSVC.
svn path=/trunk/; revision=51874
- Add msports.h header file.
- Add the initial version of the com port database and use it to track the used com ports.
svn path=/trunk/; revision=51866
- Start implementing control transfers
- Move initialization of usb queue after the bulk / control head endpoints have been created
- Add interface functions to retrieve bulk / control head endpoint descriptors
- Add macros for setting transfer descriptors field (taken from Haiku)
- Partly implement IUSBQueue::AddUSBRequest for control / bulk transfers
- Create endpoint descriptor for control request and link setup descriptor to it
- Link setup descriptors to data descriptor / status descriptor
- WIP (needs fixes)
svn path=/branches/usb-bringup/; revision=51864
- Create a new library msvcrtex, which will be automatically linked, when msvcrt is linked. This is yet a minimal version which replaces the different mingw libs.
- Use proper unicode entrypoints
- remove unneeded libcoldname.a
svn path=/trunk/; revision=51860
- Fix bug in calculating offset of status register
- Start implementing deferred processing routine
- Signal status change when new device is arrived
- Implement basic GetPortStatus
- Start implementing SetPortFeature, ClearPortStatus
- ResetPort not yet fully working, WIP
svn path=/branches/usb-bringup/; revision=51858
- Add check that controller is functional state
- enable interrupts after the controller has been started
- partly implement isr routine
svn path=/branches/usb-bringup/; revision=51852
- Start of OHCI support
- Based on new usbehci driver and Haiku USB Stack
- Driver initializes the controller and retrieve number of ports
svn path=/branches/usb-bringup/; revision=51851
- Implement handling WM_NCMOUSEMOVE, WM_NCMOUSELEAVE and WM_NCLBUTTONDOWN, needed for painting the caption buttons when the user hovers or clicks them
svn path=/branches/GSoC_2011/ThemesSupport/; revision=51846
Arch Blackmann should know not to remove empty lines at the end of resource/rsrc files in base\applications. Should fix MS Linker fatal error RC1004 for modules affected.
svn path=/trunk/; revision=51844
Determine serial port names COM1-4 and parallel port names LPT1-2 from their resources. Other serial and parallel are still named COMx and LPTx.
svn path=/trunk/; revision=51843
- base/applications. pl-PL.rc conversion;
- add PRAGMA, required by MSVC to rsrc/modulename/resource.rc, instead of individual localisation files. Localisation files already converted to UTF-8 will be placed at the end, after PRAGMA.
Tested on rbuild/cmake/nmake
svn path=/trunk/; revision=51841
- Move creation of the scrollbar Gray brush from user32 to win32k.
- Noticed a color and pattern (patchy white pattern) difference when using GreCreateBitmap and it did not work the same as the main function. Correcting this fixed the Gray brush. Someone from YAROTOWS needs to look into this.
svn path=/trunk/; revision=51835
- Add Switch Class support. See bug 5378. Original patch by David Quintana. Alt-Tab works the same as windows, Alt-Esc needs more work.
svn path=/trunk/; revision=51831
All dlls and drivers are in their normal directories instead of having all in reactos\.
Original implementation by zefklop in cmake branch.
svn path=/trunk/; revision=51830
- Pass memory manager to USBQueue object
- Fix bug in memory manager which did not check if an allocation equals page size
- Implement interrupt queue head array with frequencys of 1ms up to 32ms
- Store queue heads in the sync schedule array
- WIP
svn path=/branches/usb-bringup/; revision=51826
- Implement a simple class installer that assigns the friendly name "Serial Port (COMx)" to all serial ports.
- Enable the class installer in ports.inf.
- Add parallel port installer configuration to ports.inf (disabled).
svn path=/trunk/; revision=51813
- Load freeldr.ini first chance, boot.ini on second chance
- Make freeldr list box scrollable
- Fix some FIXME's
[msgconfig/system]
- Implement parsing and displaying system.ini contents
See issue #1986 for more details.
svn path=/trunk/; revision=51811
Fix SetupDiCreateDevRegKeyW: When trying to open a devices driver key (DIREG_DRV) the function always created a new driver instance key. The fixed version checks for an existing driver value. If the driver value exists, it opens or creates the driver instance key. If the driver value does not exist, the function generates a new driver instance key and sets a new driver value for the device instance.
svn path=/trunk/; revision=51809
- Add error handling to USBSTOR_CSWCompletionRoutine for errors that can be recovered via reset of pipe/port/controller. Only pipe reset is attempted. WIP.
svn path=/branches/usb-bringup/; revision=51808
- SetupDiCreateDeviceInfoW: Really create a new device instance instead and return its device info data instead of returning the device info data for an existing device.
- Implement SetupDiDeleteDeviceInfo.
svn path=/trunk/; revision=51807
- Add forwarders for functions that have been implemented in setupapi.dll.
- Remove forwarders which are not exported by cfgmgr32.dll.
svn path=/trunk/; revision=51803
- Add stubs CM_Get_Class_Registry_PropertyA/W.
- Move CMP_RegisterNotification, CMP_UnregisterNotification and CMP_WaitNoPendingInstallEvents.
- Remove stubs TPWriteProfileString and suErrorToIds because they are part of setupx.dll (16 bit DLL).
svn path=/trunk/; revision=51802
- Fix the fix for IopStoreSystemPartitionInformation(). The sizeof wasn't on an unicode string...
I blame the guy who first wrote that function!
Thanks to "rs" on IRC for having spotted the error.
svn path=/trunk/; revision=51797
- Implemented IoSetSystemPartition(). This is required by MountMgr.
- Fixed buffer size in IopStoreSystemPartitionInformation(). No need to have a buffer twice bigger than maximum use.
svn path=/trunk/; revision=51796
- generate source files and header files in one pass by passing right arguments to widl.
Why this didn't work before is left as an exercise for the reader.
svn path=/branches/cmake-bringup/; revision=51772
- refactor generation of rpcproxy file to avoid recompiling
Get rid of unnecessary libraries in the process.
Tested with both mingw32-make+widl/nmake+midl
- do not rebuild livecd hives if not necessary
svn path=/branches/cmake-bringup/; revision=51771
- Implement retrieving serial number from device
- Use serial number to format instance id and retrieve in StorageDeviceProperty requests
- Silence traces
svn path=/branches/usb-bringup/; revision=51770
- Add the shell32 code which has been converted into C++ from the existing C code.
- This is a straight drop using the code Andrew sent to me, it's currently unchanged and untested
- All C++ conversion work to date was carried out by Andrew Hill <ash77 at reactos dot org>
svn path=/trunk/; revision=51768
- Properly fix compiler warning, by initializing the integer array with zeros. Remove GNUC #ifdef, introduced in rev 42970. Patch by Markus (thechief at praise dash land dot de)
See issue #6179 for more details.
svn path=/trunk/; revision=51767
- Add asserts
- Rewrite queue methods to accept the FDO only
- Fix multiple synchronization bugs
- Don't start next irp before current irp has completed
- Use contiouslogicalbytes blocks from srb
- Should fix race conditions
svn path=/branches/usb-bringup/; revision=51764
- Add DECLARE_NO_REGISTRY
- Fix invalid pointer use and add an overloaded assignment operator in CComDynamicUnkArray
- Add UpdateWindow to the CWindow class
- Add missing error checks
Changes by Andrew Hill <ash77 at reactos dot org>
svn path=/trunk/; revision=51763
- Silence warnings
- Silence traces
- Add asserts
- Add code to detect usb reset in progress. In that case cancel the request
- Add more code to synchronize urb action
- WIP, still hangs are observed
svn path=/branches/usb-bringup/; revision=51759
Commit some stuff reviewed by aleksey long ago
- get rid of LastEntryInSegment in HEAP_SEGMENT struct
- Simplify code for page commit
- Simplify code for splitting entries
svn path=/trunk/; revision=51757
- Start implementing error handling.
- On errors use a worker thread to attempt a recovery. If successful restart the operation that failed.
- If STALL_PID is received reset the pipe of the device.
- Not entered into code path yet. WIP.
svn path=/branches/usb-bringup/; revision=51754
- USBSTOR_SendRequest: The buffer for read/write may not be NonPagedPool, which is documented as a requirement for using MmBuildMdlForNonPagedPool. Also locking the buffers pages is also not an option as the routine is called at DISPATCH_LEVEL.
It so happens that Irp->MdlAddress is valid for read/write operations. Use it instead of procedure above.
- Add a sanity check to make sure the Mdl does describe the transfer buffer.
Fixes IRQL_NOT_LESS_OR_EQUAL bugcheck in windows. These changes also may fixed writing to device.
svn path=/branches/usb-bringup/; revision=51752
Patch by Paolo Bonzini <bonzini [at] gnu [dot] org>
Fix implementation of ExInterlockedAddLargeStatistic
The old version wasn't really atomic.
See issue #6223 for more details.
svn path=/trunk/; revision=51748
- Implement __chkstk_ms for x86 and amd64
- amd64 code is based on code by Kai Tietz and Richard Henderson
- x86 code is based on old __chkstk implementation and code by Kai Tietz and Richard Henderson
- Function is required by newer gcc versions
svn path=/trunk/; revision=51747
- Begin implementing handling WM_NCPAINT
- Evert time a draw operation is performed on the non client area, a DRAW_CONTEXT will be initialised in order to keep most used information about the drawing opoeration
svn path=/branches/GSoC_2011/ThemesSupport/; revision=51746
- Sync to 0.41b2
- Fixed bug with unnecessary call to AtapiEnableInterrupts() inside AtapiSendCommand() when DRQ is never asserted
- Added registry option AtapiSendDisableIntr to manage ATAPI command sending behavior. AFAIU, under VirtualBox it is unnecessary to disable interrupts around sending ATAPI commands.
- Implemented registry options for BUSY and DRQ timing adjustments.
- Added VirtualBox tuning option - VirtualBox value in registry. Sets timings and ATAPI (CD/DVD) related options to the most compatible values.
- Fixed bug with UDMA2 support on Intel ICH1/2/3/4/5 chips.
- Made some experimental changes in automatic slow-down algorithm. This code is rather old, but was not tested properly.
- Fixed bug with accessing SATA registers on Intel chips
- Many other fixes
svn path=/trunk/; revision=51736
Completely refactor bootcd generation
- one macro (à la cmake's install) to rule them all, to make your life simpler and have build files cleaner
- Completely rehaul the bootcd tree : now you have in it some minimal reactos system
- reflect those changes in setupldr
No livecd yet!
svn path=/branches/cmake-bringup/; revision=51731
- Set the focus for the treeview when the main window gets the focus.
- Select the root item after the devices tree has been built.
These changes enable the user to navigate the device tree using the keyboard.
svn path=/trunk/; revision=51715
Merge r51273 from cmake branch:
The crt library must provide a and w version of the functions and not rely to _UNICODE being set or not
svn path=/trunk/; revision=51713
Update widl_ros.diff
The change was done in r49282 in cmake branch and mysteriously made it into trunk in r50063, which is a wine sync, but wine doesn't have that code.
svn path=/trunk/; revision=51704
* Partially revert r50224 as the current resource handling doesn't suffer from the iconv issue on unix.
svn path=/branches/cmake-bringup/; revision=51703
CM_Create_DevNode_ExW: Copy the device id string into a local buffer before passing it to PNP_CreateDevInst because its 2nd argument is an 'in out' string. Using a writable string buffer prevents exceptions in case the device id passed to CM_Create_DevNode_ExW is a string constant.
svn path=/trunk/; revision=51697
- Fix bug in read command, which uses wrong format. LBA address is stored in the CDB10 struct
- Don't build the mdl for nonpaged pool if the driver passes an mdl (i/o paging requests)
- Usbstor can now list disk contents and browse files when used in Windows XP
- Need to implement SCSIOP_WRITE to write files
svn path=/branches/usb-bringup/; revision=51694
- Handle SCSIOP_MEDIUM_REMOVAL
- Handle paging i/o requests. These request don't provide a data buffer in the srb, but the buffer is stored in the irp' mdl
- Fix the hack for unimplemented Mode Sense command
- Usbstor now completes initializes and receives read requests
- Still need to fix read request, as Windows XP fails to recognize the disk format, WIP
svn path=/branches/usb-bringup/; revision=51693
- Add a check at USBSTOR_FdoHandleStartDevice to make sure the device uses Bulk Transfers.
- Use macros for initializing the Urbs. The macros ensure that the proper fields of the URB are set correctly.
Fixes failing of getting device descriptor and getting devices BlockLength.
svn path=/branches/usb-bringup/; revision=51691
- Only get the string descriptors if they are available. Fixes crash when using usb mouse and keyboards in windows.
svn path=/branches/usb-bringup/; revision=51688
- Modify BuildBulkTransferQueueHead to support TransferBufferLengths larger than PAGE_SIZE * 5.
- Acquire a SpinLock before adding QueueHeads to AsyncList and PendingList.
- Dont request a new QueueHead for incomplete transfers in QueueHeadCompletion, as the memory for the just completed QueueHead has not been released yet. Doing so overwrites the m_TransferDescriptor[x] members with new address resulting in memory leaks. Instead request a new QueueHead after the QueueHead has been freed in QueueHeadCleanup.
- Fix a bug where a QueueHead was removed from the m_CompletedRequestAsyncList instead of the m_PendingRequestAsyncList.
- Temporary hackfix InternalCalculateTransferLength to return the TransferBufferLength. This hack will be removed as soon as possible.
- With these changes the hub and ehci driver allow viewing content of and transfers to/from usb disks.
svn path=/branches/usb-bringup/; revision=51684
- DispatchMessage does not allow pointers to be passed forward so need to check for non synchronous messages. Added the same check for the other functions in User32 leaving the rest to be checked in Win32k.
- Add the timer callback validation check.
svn path=/trunk/; revision=51676
- Save PortNumber in child device extensions as its needed when PNP request device capabilities.
- Handle IOCTL_INTERNAL_USB_SUBMIT_URB request by creating a new Irp and passing the Urb down to the miniport.
- Now working with our ehci driver, we at least see a usb driver and can browse directory in Windows, though doing so results in assert as ehci doesnt support transfers over 5 Pages yet. Fix coming soon.
svn path=/branches/usb-bringup/; revision=51672
- Result of LdrQueryImageFileExecutionOptions() shouldn't be used as a return status. Always reset it to STATUS_SUCCESS whatever the function returns.
See bug #6231 for more details.
svn path=/trunk/; revision=51668
[WIN32CSR.DLL]
- Removed previous hack in CsrDefineDosDevice() that was stripping off extra trailing path separators.
[SUBST.EXE]
- The help message was not show aligned properly, fixing that.
See issue #993 for more details.
svn path=/trunk/; revision=51667
- Sam Arun Raj Seeniraj: ObpLookupObjectName() used to reparse endlessly when a symlink points to itself, preventing this by limiting number of reparse attemtps.
See issue #993 for more details.
svn path=/trunk/; revision=51666
- Make ObpParseSymbolicLink properly handle the case when target path has a trailing path separator. Based on findings by Samarunraj.
See issue #993 for more details.
svn path=/trunk/; revision=51665
- Run the dialog window in a new thread patch by Mikhail Denisenko <denisenkom AT gmail DOT com>
See issue #3339 for more details.
svn path=/trunk/; revision=51656
- When clearing feature port reset, remove the flag from Change and if the port is hardware enabled set the enable flag in Status.
- When checking port status, if Change has any flags set an SCE request most be completed. Verified while writing usbhub driver in xp.
- Probably fixes a case where an endless loop of completing SCE requests.
svn path=/branches/usb-bringup/; revision=51642
- Implement constructing CompatibleIds string using device descriptor. However, if the Class of the device descriptor is 0, then Class, SubClass and Prot which make up this string must come from the Interface.
- Change Ids and Descriptions to Unicode Strings.
- Move the creating of Id strings to one place, CreateDeviceIds, and called it at device creation.
svn path=/branches/usb-bringup/; revision=51641
- Some devices implement a default Language, LangId of 0. Some do not. Use LangId 0x0409 (English) until LangId is implemented.
- Remove an unused variable.
svn path=/branches/usb-bringup/; revision=51622
- Prompt before importing from the command line.
- Inform the user about successful import from the command line and the menu.
- Updated Spanish & Italian translations.
See issue #6230 for more details.
svn path=/trunk/; revision=51621
- The lower level driver completes the SCE IRP whenever the change state has been modified for a port. When handling the SCE IRP completion only check for flags in change state. The Change flags must be cleared else the hub driver will get non stop SCE IRP completions.
- Allocate IRP from Pool vice calling IoAllocateIrp. Dont free the IRP in the completion routine as the port driver, oddly, frees them.
- GetUsbDeviceDescriptor: Remove use of MDL and use only buffer instead to make our new usbehci happy.
- When calling the interface routines, correctly use the BusContext returned from the interface instead of the RootHubPdo. This worked on windows as it just happened they were the same.
- Implement RootHubInitCallbackFunction, which only job currently is to send the first SCE IRP.
- For Start Device for child device objects fake success for now. Will be implemented later.
- Implement returning IRP_MN_QUERY_IDs, IRP_MN_QUERY_DEVICE_TEXTs and IRP_MN_QUERY_BUS_INFORMATION.
- Add basic handling for IRP_MJ_POWER.
- Misc code changes.
svn path=/branches/usb-bringup/; revision=51620
* Sync nls files to wine 1.3.19
* Escape some characters to fix MSVC build
* Patch by Igor Paliychuk <mansonigor at gmail com>
svn path=/trunk/; revision=51616
- Implement retrieving drive capacity
- Use constant from scsi.h instead of defining own format
- Don't reset returned parameters length to zero, as ioctl handler might set it
svn path=/branches/usb-bringup/; revision=51608
* Igor Paliychuk <mansonigor at gmail dot com>:
- Remove duplicate string and add some missing resources
- Increase buffer sizes in FolderOptionsFileTypesDlg
- Update RU and UA translations
* Myself: update German translation
svn path=/trunk/; revision=51596
- Update Dejavu fonts & documentation to version 2.33 (by Jon Voris, bug #6106)
- Update GNU FreeFonts & documentation to version 2010-09-19 (by Jon Voris, bug #6115)
- Update 3rdparty version info including recent Freetype sync
svn path=/trunk/; revision=51594
- Standardize keyboard flags for the input thread and send input. Next round will update the rest of user32 defwnd.c, combo.c and appswitch.c.
- Add the missing keyboard flags needed for menu operations for in-user applications.
- Setup a function to be used to get, set and clear queue flags of special use for User32 functions.
svn path=/trunk/; revision=51587
Folder options:
- Load text from language file.
- Fill in the extension and description.
Patch by Edijs Kolesnikovics <terminedijs at yahoo dot com>
- Fix an Italian string.
See issue #6208 for more details.
svn path=/trunk/; revision=51586
- Move hook call close to the keyboard message post. This allows the registering of Aly keys before the hook call. This fixes all but one GetKeyState Api Test.
- Fix GetAsync/KeyState error codes and the return for GetAsyncKeyState. See http://appdb.winehq.org/objectManager.php?sClass=version&iId=8516&iTestingId=13644
- Fix TranslateMessage, check for window and use the window pti for re-posting to message queue.
- Peeking hardware queue should be the same as other Peek queue.
svn path=/trunk/; revision=51583
- Make DeviceId generation to match ms usbstor
- Don't hardcode the device type in USBSTOR_PdoHandleQueryHardwareId
- Don't hardcode device type in USBSTOR_PdoHandleQueryCompatibleId
- Hackfix generation of instance id
- Add comments
- Cleanup code
- Usbstor now receives ioctl from disk.sys
svn path=/branches/usb-bringup/; revision=51582
- Add UsbDeviceHandle to child device extensions structure to be passed with URB requests.
- Reimplement the Worker Thread DeviceStatusChangeThread called on SCE completion. Each SetFeature and ClearFeature resend SCE request.
Only create the usb child device if the PortStatusChange member of the device extensions was modified.
- Implement CreateUsbChildDeviceObject used for creating a deviceobject for the USB device and initializing it.
- Implement GetUsbStringDescriptor for getting DeviceId, TextDescription and InstanceId from USB Device.
- Move the initial QueryStatusChangeEndpoint to after the hub device receives start pnp.
- USBHUB driver is now back to the same functionality of previous version. Showing the device wizard for connected devices.
svn path=/branches/usb-bringup/; revision=51581
- Rewrite hardware id / instance id matching
- Implement support for compatible id
- UsbStor still not attached to disk.sys, needs more research on how to get this done
svn path=/branches/usb-bringup/; revision=51580
- Update to version 2.7.8. Patch by Carlo Bramini.
- Installed and booted successfully to the desktop. Tested
abiword 2.6.4 and FF 2.0.0.20.
See issue #4988 for more details.
svn path=/trunk/; revision=51577
- Change parameter name for DeviceObject in functions to clearify which device object is to received IRPs.
- Fix a typo in GetPortStatusAndChange that caused the function to return nothing for status and change.
- Call QueryStatusChangeEndpoint to send the first SCE request in Query Bus Releations.
- Complete implementation of StatusChangeEndpointCompletion. Queue a work item for DeviceStatusChangeThread, which will be used to initialize the usb device and create related device objects.
svn path=/branches/usb-bringup/; revision=51576
- Don't free device descriptor if it failed to acquire it
- Don't move out of usb configuration array. Found by special pool
- Fix bug when storing interface information which lead to a overwrite
- Implement retrieving generic type for device
- Fix bugs in QueryDeviceId
- Implement querying hardware ids
- Implement querying instance ids
- With these changes, usbstor now initializes and create pdo, which is also enumerated
- Needs more code / fixes so that disk.sys attaches to usbstor
- WIP
svn path=/branches/usb-bringup/; revision=51575
- Invoke status change callback at passive level
- Fill out usb descriptor struct instead passing everything in char array
- Use 1 byte interrupt endpoint packet size for hub for now
- Fill in interface information when SelectConfiguration request is passed. Fixes sce callback and pool corruptions
- Return device descriptor in USBHI_QueryDeviceInformation
- Tested in WinXP with special pool enabled & overrun support
svn path=/branches/usb-bringup/; revision=51572
- Move up the chain to convert F10 to system key mode before going into the queue, so when in menu tracking, the message can be removed and processed, see bug 1009.
svn path=/trunk/; revision=51571
- Start work on reimplementing UsbHub driver.
- Implement SubmitRequestToRootHub, QueryStatusChangeEndpoint, QueryInterface, GetPortStatusAndChange, SetPortFeature and ClearPortFeature.Handle QueryBusRelations.
- Partially implement PNP handling for FDO. When starting the device get all the information about the RootHub the driver will need and enable Power on all ports.
- Partially implement PNP handling for PDO, QueryId and QueryDeviceText.
- WIP
svn path=/branches/usb-bringup/; revision=51569
- Handle font height == 0 also in TextIntGetTextExtentPoint (bug 5305)
- Remove hardcoded font height (11) and replace it with
the default font height from the device context by Timo.
svn path=/trunk/; revision=51566
Fix a double off-by-one bug:
- The eventlog service was reporting one event more than was available (+1).
- The event viewer did not display the latest event from the eventlog service (-1).
See issue #6182 for more details.
svn path=/trunk/; revision=51558
sync with trunk (r51543)
(I had to comment out sv-SE.rc in rapps, there are problems with characters)
svn path=/branches/cmake-bringup/; revision=51553
* Fix ShowingCursor counter interpretation. Values >= 0 means cursor is shown, values <0 means cursor is hidden. Fixes few user32:cursoricon tests.
svn path=/trunk/; revision=51552
- Initialize ViewSize to 0 when mapping view of section. Fixes DLLs registration problems in 2nd stage (happening all the time with the ldr rewrite).
svn path=/trunk/; revision=51539
- Carlo Bramini: implement a skeleton for the charmap advanced view, Italian translation included
- German translation added by myself
See issue #6065 for more details.
svn path=/trunk/; revision=51537
* Fix a bug when after changing color scheme Apply and then OK is clicked. Background color was restored to the previous one in this case. Spotted by Gabriel Ilardi.
svn path=/trunk/; revision=51531
- Add event sources "EventLog" and "Service Control Manager" to the registry.
- Implement an internal event reporting function and report the successful start of the event logging service.
svn path=/trunk/; revision=51529
* Do not reference current thread before sending message to it. It was causing thread to be a ghost if ExitThread was called in WndProc.
* Hold only UserRefObjectCo reference on windows when sending messages to them.
svn path=/trunk/; revision=51527
- Do not try to initialize thread info in UserGetWindowObject. Do it in thread callout. Removing this hack is needed in order to use UserPostMessage in ProcessCallout when a process is terminating
- Notify winlogon that the shell is exiting in order to restart the shell
svn path=/trunk/; revision=51519
Implement an event source list and use it to find the event log file for a given event source when an event was reported.
svn path=/trunk/; revision=51513
Fix for the loading of the "KnownDlls" in SMSS. Patch by Hermès BÉLUSCA - MAÏTO.
The patch was applied without changes to comments, indentation and coding style.
See issue #6159 for more details.
svn path=/trunk/; revision=51511
- Clear and set send erase background bit when sending WM_ERASEBKGND and if return zero reset the bit. This prevents a message loop in riched20 when calling GetUpdateRect. Spotted by Michael Martin.
svn path=/trunk/; revision=51509
- Store number of bytes transferred in the transfer descriptors
- Perform queue head completion when the door bell ring has been acknowledged. Fixes race condition between multiple irps in the async list
- Fix calculation of transfer length when the request is an bulk in operation
- Use EndPointDescriptor member to access transfer type / pid direction
- Use MmGetSystemAddressForMdlSafe to retrieve system address for urb buffer
- Fix check if first transfer buffer finishes on first size if the size is of page_size
- With these changes and little luck and good weather, usb mass storage devices have been seen to work in Windows XP SP3
- Code inspired of mjmartin usbehci driver and Haiku's usb stack
svn path=/branches/usb-bringup/; revision=51506
- Port URB_FUNCTION_CLASS_INTERFACE from mjmartin usbehci driver
- Fix typo in interface declaration
- Disable assert for now
- Return irp status pending when irp is added to the queue
- Fix bug in InternalGetPidDirection
- More work on the bulk transfers, not yet working
svn path=/branches/usb-bringup/; revision=51501
- Return 1 when explorer terminates. This value lets winlogon know that explorer terminated without errors and it doesn't need to restart
svn path=/trunk/; revision=51499
- When terminating an application use 1 as exit code. This is used by Task Manager in windows. It also prevents explorer from restarting in windows
- Disable redrawing the list of processes while the list is updated. Fixes flicker in windows
svn path=/trunk/; revision=51498
Patch by Edijs Kolesnikovics: Write default font, weight and point size when run for the first time. That's what ms notepad does too.
See issue #5305 for more details.
svn path=/trunk/; revision=51497
- UserSetProcessWindowStation: Use PsGetProcessWin32WindowStation and PsSetProcessWindowStation. Close the prevoious window station handle only when the window station was not set by UserSetProcessWindowStation (should be the case for console apps)
svn path=/trunk/; revision=51496
Jeisson Toscano: French update + some Spanish translation updates (with corrections from vicmarcal and me)
See issue #6012 for more details.
svn path=/trunk/; revision=51494
- Fix a type causing CommitIrp to fail.
- Add a sanity check in StatusChangeEndpointCallBack to check that a pending SCE irp is queued.
- Start Implementing Bulk Transfers. Needs more work.
svn path=/branches/usb-bringup/; revision=51493
- Set the NakCountReload to 0 in the QueueHead that is set in the Async register, as this is a dead QueueHead marked as halted and does not need it.
- Set the Async register in StartController. Note: The Async and Periodic registers can only be modified when the controller is stopped.
- Driver now works in Qemu.
svn path=/branches/usb-bringup/; revision=51491
- hub_controller.cpp: Change RtlSetBit and RtlClearBit to RtlSetBtis and RtlClearBits for Win2k compatibility.
svn path=/branches/usb-bringup/; revision=51488
* Use current directory in GetDriveType if no path is given
* Check if path is properly ended
* Remove slash from end when calling NtCrateFile function
* Fixes few kernel32_apitest tests and floppy icon in My Computer if no floppy is inserted
See issue #6148 for more details.
svn path=/trunk/; revision=51485
- Implement retrieving device / configuration descriptor for usb devices
- Based on mjmartin usbehci driver
- Mass storage device found wizard now pops up (fails to install - WIP)
svn path=/branches/usb-bringup/; revision=51478
- Move error checking if controller reported an error inside the door bell ring check block
- Extend interface of IUSBRequest to include DeviceAddress, which is needed to retrieve configuration descriptor of newly set address
- Allocate CtrlSetup from nonpaged pool
- Release setup packet on cleanup
- Devices now receive an device address
- Currently asserts while retrieving configuration descriptor as this code path is not yet existant
*** Assertion failed: Urb->UrbHeader.UsbdDeviceHandle == NULL
*** Source File: d:\usb-bringup\drivers\usb\usbehci_new\hub_controller.cpp, line 1118
svn path=/branches/usb-bringup/; revision=51477
- Change interface to return real async queue head
- Set the async queue head register after the controller has been started
- Enable async queue in StartController
- Port DumpDeviceDescriptor from mjmartin usbehci driver
- Remove pseudo queue head from usb queue, instead use the real async queue head exported from IUSBHardwareDevice
- Get physical address for transfer data in BuildControlTransferQueueHead
- Retrieving device descriptor now ~works, currently stops at setting device address (needs more work)
svn path=/branches/usb-bringup/; revision=51476
- Don't use exclusive locks for pens
- Don't try to delete the window update region
- Don't ignore pid in GreIsHandleValid
- Check if handle is valid before deleting it in DC_Cleanup
- Silence a DPRINT
svn path=/trunk/; revision=51475
- Check if the current request type is supported (currently only control is supported, bulk will be implemented soon)
- Fix bug in IUSBRequest::CreateQueueHead
- Code runs now until first device descriptor get request (async / periodic queue not yet enabled)
- mjmartin usbehci status not yet reached
- Let the fun now begin ;)
svn path=/branches/usb-bringup/; revision=51474
- Cache port status & port change settings because it appears that some ehci controllers do not store the status reset register bit correctly
- This commit is a temporary solutions, needs to be reworked
- Return correct status from HandleClassOther
- Handle IOCTL_INTERNAL_USB_SUBMIT_IDLE_NOTIFICATION to return status success
svn path=/branches/usb-bringup/; revision=51473
Rewrite of the GDI handle manager
- The old handle manager used a completely retarded spinlock in combination with KeDelayExecutionThread() for both exclusive
and shared locks. This is probably the most uneffective algorithm possible. It was also duplicating code everywhere and it was a overall mess It
is now replaced with a lock-free reference counter for shared locks and a pushlock for exclusive locks. -> Better performance and scalability.
- Allocate user mode object attributes from the new gdi pool. This way, we don't need any caching, since the pool serves as a cache. Its also
much faster and uses much less memory.
- Allow object allocations of different size, instead of fixed size from a table. This way a single allocation can take care of actual needs.
- Allow allcoating objects without a handle and insert them into the handle table later
- Properly synchronize the process GDIHandleCount. Now gdiview and taskmanager show the correct number of gdi handles.
- Implement a new event tracking system, that is capable of tracking all reverences and locks of objects and pool allocations to help track
possible leaks
- Make sure that all objects of a process are deleted in cleanup
- Make sure all usermode memory allocations are freed, when cleaning up the process pool.
- Make sure that each object type is using the correct type of lock (either shared or exclusive, not a mixture)
- Fix some object / reference leaks
- Lots of inferface improvements
- Use global variables for certain things instead of members in the mapped gdi handle table
- Make IntSysCreateRectpRgn create a region without a handle
- Fix detection od source and mask use in GreStretchBltMask
- Use GDIOBJ_bLockMultipleObjects in NtGdiCombineRegion to avoid possible deadlocks
- Fix NtGdiAbortPath to reset DCPATH_ACTIVE flag in the dc and only bail out on error, instead of always
- Replace DC_AllocateDcAttr and DC_AllocDcAttr with DC_bAllocDcAttr using the new user mode pool
- Remove DCU_SyncDcAttrtoUser and DCU_SynchDcAttrtoUser. Those functions were unused and didn't do anything useful anyway,
- Replace IntGdiSetDCOwnerEx and DC_SetOwnership with GreSetDCOwner, remove unused NoSetBrush parameter
- Replace GDIOBJ_bValidateHandle and IsObjectDead with GreIsHandleValid
- Chage GDIOBJ_bLockMultipleObjects: pass object type, return a BOOL, whether all objects could be locked, cleanup on failure
svn path=/trunk/; revision=51470
- Return the physical address of the async queue head in CUSBHardwareDevice::GetAsyncListRegister
- Remove superflous entries from queue head structure, they are processed within IUSBRequest class
- Remove USBHI_GetPortHackFlags, this function has been deprecated
- Add interface functions for IUSBRequest / IUSBQueue
- Callback into IUSBQueue when a hardware interrupt arrives
- Implement callback function to check for completed queue heads / free completed queue head depending on the Async Advance interrupt bit
svn path=/branches/usb-bringup/; revision=51466
-Fix a case when a window is created and we don't have a monitor created yet. (This can happen when we may attempt to use gui on demand)
svn path=/trunk/; revision=51464
- Sync up menu without the Right to Left support. Still missing needed changes for adding it.
- Patches by Alexandre Julliard and Dmitry Timoshkov.
svn path=/trunk/; revision=51461
- Merge 50674 to trunk:
- Allocate memory for the logical pin array
- Use the correct counter variable for logical pin array
- Found by msvc
See issue #6143 for more details.
svn path=/trunk/; revision=51455
[NTOSKRNL.EXE]
- Complemented existing stub routines ObpDeleteSymbolicLinkName() and ObpCreateSymbolicLinkName(). Now when symbolic links are created, they are checked for dos drive mapping and ObSystemDeviceMap is updated accordingly.
- Updates to ObpDeviceMapLock are protected by a guarded mutex (changed from fast mutex by Aleksey Bragin).
- Aleksey Bragin: Fix concurrent access issues in ObDereferenceDeviceMap. The DeviceMap read/write operation must be atomic (within the ObpDeviceMapLock region).
[SUBST.EXE]
- Removed hard coded strings and moved it to resource file.
[WIN32CSR.DLL]
- Added a workaround, when target path end with a trailing '\', the substituted drive fails to parse to the target, this is a bug in the object manager, but working around it in CsrDefineDosDevice() by stripping trailing '\'.
See issue #993 for more details.
svn path=/trunk/; revision=51452
- Isolated Alt-Tab/Esc issues. Placed update_input_key_state when posting to a queue like wine. This fixed the shift key checks. The Alt-Space is handled after a Translate Message call (WM_SYSCHAR), developers need to check application code to help support it (ConCtrl32). Will move AltSwitch.c from win32csr and port patch from bug 5378 to User32. ReactOS registers the Switch Class in Win32k and it is not used.
- Implemented the rest of HotKey Support. ReactOS Explorer does not save hotkey actions.
- See related bugs 1091.
svn path=/trunk/; revision=51450
* Set IMAGE_SCN_MEM_NOT_PAGED flag for read-only data section (.rdata). Patch by Pavel Pisa.
See issue #5618 for more details.
svn path=/trunk/; revision=51447
- Correctly return whether there is a kernel APC pending in KiSwapContextExit. Fixes IRQL asserts seen in bugs 5955 and 5777. Michael Martin's help in finding and fixing this bug was invaluable. Also thanks to Art Yerkes and testing team for additional information.
See issue #5955 for more details.
svn path=/trunk/; revision=51445
- Swedish translation adjustments of calc and find
- Fix comments in magnify
Patch by Jan Blomqvist Kinander (JaixBly [at] spray [dot] se)
svn path=/trunk/; revision=51444
- Create member for storing the request object
- Add interface function which is invoked when the queue head should be freed
- Add support functions which deal with completed queue heads
svn path=/branches/usb-bringup/; revision=51440
Prepare ReactOS for the change from the ancient 32 sectors / 64 tracks disk layout (used until Windows NT4) to the modern 63 sectors / 255 tracks disk layout (used since Windows 2000).
svn path=/trunk/; revision=51437
* Use EDITTEXT instead of LTEXT. Patch by Edijs Kolesnikovics <terminedijs at yahoo dot com>
See issue #6129 for more details.
svn path=/trunk/; revision=51428
* explorer-new: Fix some wrong UK and RU translations.
* shell32: Update UK and RU translations.
* Patch by Igor Paliychuk <mansonigor at gmail dot com>
svn path=/trunk/; revision=51426
* Move "Play" string from shell32. It will be used in audio files context menu in explorer
[SHIMGVW]
* Move "Preview" string from shell32. It will be used in image files context menu in explorer
* Add Polish translation
[SHELL32]
* Remove "Play" and "Preview" translations - they shouldn't be in shell32
* Add "explore" translation for folder context menu
svn path=/trunk/; revision=51423
* Properly check if buffer given to GetComputerName is too small. Fixes hostname.exe if computer name is MAX_COMPUTERNAME_LENGTH long (the default for bootcd since it's generated randomly in Setup).
* Simplify it a bit
svn path=/trunk/; revision=51412
Increase floppy drives count in global ConfigurationInformation. Fixes not working floppies in explorer.
See issue #4236 for more details.
svn path=/trunk/; revision=51409
hub_controller:
- Implement StatusChangeEndpointCallBack called by HardwareDevice class when port status has changed on controller.
- Move Status Change Endpoint query code to IQueryStatusChageEndpoint as it is also needed by StatusChangeEndpointCallBack.
usb_request:
- Implement InternalCreateUsbRequest.
usb_queue:
- Implement CreateUsbRequest and AddUsbRequest
hardware:
- Implement GetAsyncListRegister and GetPeriodicListRegister.
- Implement SetStatusChangeEndpointCallBack for setting callback and context. Call the callback when a port status changes.
- Initialize the UsbQueue after creating the AsyncQueueHead, as the UsbQueue calls will call back with GetAsyncListRegister.
svn path=/branches/usb-bringup/; revision=51407
- Fix warning because comparing signed vs unsigned
- Implement function for retrieving the device descriptor
- Implement function for retrieving a configuration descriptor (including contained interface descriptor and endpoint descriptor)
- Cleanup interface for IUSBRequest:
- When caller initializes IUSBRequest with an IRP, then the operation mode is asynchronously. Therefore when the request is completed, IUSBRequest::ShouldReleaseRequestAfterCompletion will return true, which makes IUSBQueue call Release on IUSBRequest
- When the caller initializes IUSBRequest with an setup packet, the operation mode is synchronously. After submitting the IUSBRequest to queue, the caller should call IUSBRequest::GetResultStatus. This function will then block untill the operation has been completed. However, the caller needs to call Release() as those requests are not cleaned up by the IUSBQueue
svn path=/branches/usb-bringup/; revision=51406
* The cmake branch is so awesome you don't even need to manually translate addresses. Automagic usermode address translation brought to you by the Arty.
svn path=/branches/cmake-bringup/; revision=51404
- Initialize the MemoryManager only after allocating a CommonBuffer to pass to it.
- Create a QueueHead to put into the AsyncList Register.
svn path=/branches/usb-bringup/; revision=51402
- hub_controller: Implement Status Change Endpoint for RootHub.
Handling the URB can return STATUS_PENDING, as in the new SCE code. Check for this before completing the Irp.
Uncomment calls to PortStatus, SetPortFeature, and ClearPortStatus now that they are implemented.
- For function receiving a port number check that its not larger than the actual number of ports on the controller.
svn path=/branches/usb-bringup/; revision=51401
- Move USB_REQUEST_SET_FEATURE / USB_REQUEST_GET_FEATURE flags to header file.
- Remove QueueHead and Descriptor creation as its in UsbRequest.
- Move DMACommon Buffer allocation back to UsbHardwareDevice class.
- Implement Port Handling functions GetPortStatus, ClearPortStatus and SetPortFeature.
- Implement GetUSBQueue for returning pointer to UsbQueue and GetDMA to return DMAMemoryManager.
- Fix a typo in one of the defines for Port Register Flags.
svn path=/branches/usb-bringup/; revision=51397
- Sam Arun Raj Seeniraj: Added new test cases to QueryDosDevice(). Should be sent to Wine (tm).
See issue #993 for more details.
svn path=/trunk/; revision=51394
[KERNEL32]
- DefineDosDeviceW() is implemented and calls into csrss.exe CsrDefineDosDevice().
- Fixed a minor bug in DefineDosDeviceA(), as calling RtlCreateUnicodeStringFromAsciiz() prevented NULL lpTargetPath to be passed down to DefineDosDeviceW().
- Fixed a minor bug in QueryDosDeviceW() that causes lpTargetPath buffer to be returned with NULL string terminator placed at the wrong point in the buffer.
[WIN32CSR.DLL]
- Implemented CsrDefineDosDevice() in win32csr.dll, the symbolic links are created in global name space currently.
[SUBST.EXE]
- Implemented a subst.exe clone.
See issue #993 for more details.
svn path=/trunk/; revision=51393
- Use PQUEUE_HEAD as the Head of pending list and Async (Active) list vice using PLIST_ENTRY.
- Implement linking/unlinking a QueueHead to these Head QueueHeads.
- Implement chain linking/unlinking of QueueHeads. Thr plan being to used these functions during DMA operations to pull a number of QueueHeads from the pending list to place into the Async list for execution.
- Will look for optimization in these functions later.
svn path=/branches/usb-bringup/; revision=51390
- Use LIST_ENTRY in QueueHeads and Descriptors vice Next and Previous Pointers.
- Add functions to interface for setting AsyncListRegister and PeriodicListRegister.
- USBHardwareDevice: Initialize USBQueue when handling PNPStart.
- USBQueue: Allocate Common Buffer and use it to Create and Initialize DmaMemoryManager object.
- USBQueue: Implement CreateQueueHead and CreateDescriptor.
svn path=/branches/usb-bringup/; revision=51388
- Update interfaces with new methods needed for UsbHardWare class.
- Remove GetDmaMemoryManager as the DmaMemory will only be needed by UsbQueue class.
- GetDeviceDetails: Only fill in a parameter if its not null.
- Start implementing UsbQueue class.
svn path=/branches/usb-bringup/; revision=51386
- Implement IRP_MN_QUERY_DEVICE_RELATIONS for hub controller object
- Implement stub handler for URB_FUNCTION_CLASS_OTHER, URB_FUNCTION_SELECT_CONFIGURATION
- Partly implement URB_FUNCTION_GET_STATUS_FROM_DEVICE
- Return success from unimplemented USBHI_Initialize20Hub routine
- based on mjmartin usbehci driver
- Tested in Windows XP SP2
svn path=/branches/usb-bringup/; revision=51385
- Setup a default device descriptor for the root hub and initialize it with vendor & product id
- Partly implement URB_FUNCTION_GET_DESCRIPTOR_FROM_DEVICE, URB_FUNCTION_CLASS_DEVICE
- based on mjmartin usbehci
svn path=/branches/usb-bringup/; revision=51384
* Adopt the retar, I mean, actual way of handling resource files. This is a temporary measure until a windres issue is fixed.
svn path=/branches/cmake-bringup/; revision=51381
- Check for the correct, just created, registry key. Fix
a wrong comparison. Patch by Thomas Faber.
See issue #6123 for more details.
svn path=/trunk/; revision=51375
* Inform compiler about encoding in resource files using pragma. No pragmas caused problem because *.mc parser in cmake branch inserts pragma in its output files, which was included before other rc files
* Change RC files encoding to UTF-8 (without BOM)
* Fixes broken non-latin characters in rapps
svn path=/branches/cmake-bringup/; revision=51373
- Fix bug in CDMAMemoryManager initialization, which calculated the bitmap length wrong
- Create interface IUSBDevice, which will be used to abstract connected usb devices
- Implement support functions for the device interface.
- Implement USBHI_CreateUsbDevice, USBHI_InitializeUsbDevice, USBHI_GetUsbDescriptors, USBHI_RemoveUsbDevice, USBHI_GetExtendedHubInformation, USBHI_RootHubInitNotification, USBHI_SetDeviceHandleData, USBDI_GetUSBDIVersion, USBDI_IsDeviceHighSpeed
- Partly implement USBHI_QueryDeviceInformation
- Based on mjmartin usbehci
svn path=/branches/usb-bringup/; revision=51372
- Implement IDMAMemoryManager interface, with IDMAMemoryManager::Allocate, IDMAMemoryManager::Free
- Fix interface of IDMAMemoryManager
- Will be used by IUSBHardwareDevice
svn path=/branches/usb-bringup/; revision=51369
- Add reference count when the hub controller is created. Necessary to prevent the controller from getting deleted when the hub driver performs interface reference removes
- Implement IRP_MN_QUERY_ID, based on mjmartin ehci driver
- Fix bug in IRP_MN_QUERY_BUS_INFORMATION, it is supposed to return a PNP_BUS_INFORMATION not a simple guid
- Implement IOCTL_INTERNAL_USB_GET_DEVICE_HANDLE, IOCTL_INTERNAL_USB_GET_ROOTHUB_PDO, IOCTL_INTERNAL_USB_GET_HUB_COUNT based on mjmartin ehci driver
- Add interfaces for communicating with the hub driver
- Dispatch internal device requests not wmi requests (EHCI_Dispatch)
svn path=/branches/usb-bringup/; revision=51368
- Change VendorId and DeviceId to the same size as PciConfig members, PUSHORT.
- Add support function GetBusInterface.
- Implement GetDeviceDetails and ResetPort.
svn path=/branches/usb-bringup/; revision=51367
- Add flags for Port Status Control Register and remove structiures for Port Status Control Register.
- Remove USB Status Register and use the flags instead.
- Remove functions that are not used and never will be.
- Implement InterruptServiceRoutine and Deffered Procedure Routine.
- Partially implement port status change to detect device insert and removal.
svn path=/branches/usb-bringup/; revision=51366
- Add flags and structures needed for communicating with controller and handling schedules.
- Add support routines for modifying operational registers on controller.
- Implement getting the controller capabilties, starting and stopping the controller.
svn path=/branches/usb-bringup/; revision=51365
* Remove inflib from build. It's been superseded by newinflib.
* Rename newinflibhost and newinflib targets into inflibhost and inflib respectively.
* Update mkhive and usetup to link to the new targets.
svn path=/branches/cmake-bringup/; revision=51362
* Back out r51360. The export in question is in ole32 (DllGetClassObject). This fixes comcat msvc build.
svn path=/branches/cmake-bringup/; revision=51361
* Generate random computer name (REACTOS-xxxxxxx) in Setup. Patch by David Quintana <gigaherz at gmail dot com> modified by me
See issue #6099 for more details.
svn path=/trunk/; revision=51358
Implement gdi pool. An allocator for user mode gdi object attributes.
The old method allocated a 4k page for every object, wasting 4k physical memory and 64k address space (allcoation granularity)
The new allocator creates a per process pool for each object attribute type. Allocations are done from "sections" that start with 1 page and grow dynamically up to 64k, if neccessary a new section is allocated. This will use about 1/10 of memory for dc attributes and 1/512 for brush attributes. Also allocation is way faster. Caching object attributes is not neccessary anymore.
svn path=/trunk/; revision=51357
* Revert part of 37148. One FontGDI is used for fonts with different sizes. Do not cache font metrics then and instead calculate them every time when needed. It fixes spacing in menus if some application uses default GUI font with size other than 11.
See issue #4106 for more details.
svn path=/trunk/; revision=51356
- Create Interface IDispatchIrp which is used to handle ioctl requests
- Partly implement IHubController interface
svn path=/branches/usb-bringup/; revision=51355
- Define tag for allocations
- Fixes assert hit in ReactOS
- Fix typo in IHubController interface declaration
svn path=/branches/usb-bringup/; revision=51353
- Fix IOCTL_GET_HCD_DRIVERKEY_NAME implementation
- Now needs IOCTL_USB_GET_ROOT_HUB_NAME to get further in usbview
- Tested in Windows XP SP2
svn path=/branches/usb-bringup/; revision=51350
* Computer name length is limited to 15 characters, not 63. Fixes possible buffer overflow
* Spotted by Victor Martinez
svn path=/trunk/; revision=51349
- Build fix#1 for mingw
- Define cxa_pure_virtual which is required that gcc is aware that it should really use the libc++ instead of falling back to libc
svn path=/branches/usb-bringup/; revision=51348
- Move dereferencing of the MessageQueue from co_UserDestroyWindow to co_UserFreeWindow.
A reference is added to the MessageQueue for all windows created.co_UserFreeWindow is called for all windows and needs to be where the dereferencing of MessageQueue occurs.
This fixes NonPagedPool memory leak. See bug 6112 for more info.
svn path=/trunk/; revision=51345
- KbdDispatch: Set Status vice IoStatus.Status for all branches of switch. Use break instead of goto intcontfailure.
The previous code would set Status to what ever was in the IoStatus.Status of the IRP and return that Status.
Some times this would be STATUS_PENDING, which IIRC is a no-no.
This caused the kbdclass to enter into a wait that would never be satisifed hence blocking the keyboard thread.
This should fix the failure of keyboard input bugs.
svn path=/trunk/; revision=51344
- Implement CHCDController::Initialize, CHCDController::HandlePnp for PDO / FDO,
- Partly implement CHCDController::HandleDeviceControl
- Based on mjmartin usbehci sources
- Driver now startups and acquires pnp resources
- Fails in HandleDeviceControl with IOCTL_GET_HCD_DRIVERKEY_NAME when usbview is launched
- Tested in Windows XP SP2 + Vmware Server 2.0.2
svn path=/branches/usb-bringup/; revision=51343
* Sync with trunk r51339.
* Move the static lib creation command from the toolchain file to gcc.cmake
svn path=/branches/cmake-bringup/; revision=51340
- Reflect the recent user32_apitest changes. Please consider updating the cmake files when you alter the rbuild files in the future.
svn path=/trunk/; revision=51336
- Create empty USB EHCI driver in c++
- Design & review the interfaces with mjmartin
- Create stubs for creating IUSBHardwareDevice / IHCDController classes
- Implement DriverEntry, EHCI_AddDevice and EHCI_Dispatch
svn path=/branches/usb-bringup/; revision=51335
- Properly check parameters in RtlFindActivationContextSectionString. Almost the same as Wine did, but factored out into a standalone function, improved undocumented flags checks and allowing null data pointer in certain cases.
svn path=/trunk/; revision=51333
- Use the shell default auto-arrangement positioning. Aloow using custom positioning when configured. This fixes the positioning of icons of the desktop to look like windows
svn path=/trunk/; revision=51330
* DPtoLP/LPtoDP functions should return TRUE even if they get 0 or less points or if pointer to points is invalid. See gdi32:DPtoLP apitest.
* Fixes 24 gdi32:DPtoLP apitests.
svn path=/trunk/; revision=51328
- Implement ExecuteTransfer to replace SubmitTransfer.
- Start implentation for BulkTransfers and needed DMA. This needs much more work to be done correctly. We will get to it.
- Implement URB_FUNCTION_CLASS_INTERFACE.
- Add comments.
Change 3 of 3.
svn path=/trunk/; revision=51327
- For common buffer memory, only allocate one page and allocate more as needed. This implementation will most likely be replaced with janderwalds.
- Use a safecount while attempting to stop and start the controller, so that an error can be displayed vice doing endless loop for controllers that are misbehaving.
- Reorganize a few structuers and remove members of structures that are not used.
- Add comments for some of the DirectCall Interface Functions.
Change 2 of 3.
svn path=/trunk/; revision=51326
- Fix linking of QueueHeads. Implement LinkQueueHeadToCompletedList and CleanupAsyncList. These functions are used to software link the completed queueheads to a completed list. Only after the driver handshakes with the controller can this memory be freed.
Fix a few incorrect values for QueueHead.
- EhciDefferedRoutine: Implement handling of completed QueueHeads by removing them from the AnsycList and setting any errors for failure.
- Reserved a QueueHead that will always be in the AsyncList Address Register. By setting it as the Head of Reclamation the controller can know when it has reached the end of the QueueHead list.
- Remove all code from FdoDispatchInternalDeviceControl. This should never be called by upper level drivers.
Change 1 of 3.
svn path=/trunk/; revision=51325
- Re-implement DMA buffer routines. It is now faster and consumes less memory overhead per allocated memory block.
- IoGetDeviceProperty needs a PDO
- Add few comments
- No need to clear buffer twice in CreateQueueHead / CreateDescriptor
svn path=/branches/cmake-bringup/; revision=51324
* Return 1 in WM_ERASEBKGND when desktop is painted
* Fix background for desktop icons and their labels
* Moreover fixes explorer_new background
svn path=/trunk/; revision=51322
* Sync listview to Wine 1.3.17
* Set background and text color before calling DrawFocusRect. Those colors are used by monochrome brush during painting. Tested this function in Windows and our implementation is correct. Wine has DrawFocusRect not correctly implemented and that's why they don't see this bug. Should be sent to wine (tm).
svn path=/trunk/; revision=51321
* Copy RegQueryValueExA from Wine. Fixes 24 winetests fails which appeared after r51222. They are caused by calling RegQueryValueExA with buffer size set to large value. Old implementation was allocating new buffer with specified size (it was failing) and using unicode function. New implementation use given buffer size only when coping data.
See issue #6105 for more details.
svn path=/trunk/; revision=51314
NetUserModalsGet failed if the domain SID returned by LsaQueryInformationPolicy was NULL. Handle the NULL SID case correctly.
See issue #6102 for more details.
svn path=/trunk/; revision=51313
- Revert parts of 51283, thanks to Giannis for setting me straight on SetCursorPos and without breaking setting the Track window for TrackMouseEvent. Pass the new SetCursorPos, and wine test_TrackMouseEvent and input test_mouse_ll_hook tests.
svn path=/trunk/; revision=51305
- Rewrite GetKeyState based on wine. Now it's results are based on whether or nor the message queue has processed a keyboard message and not the actual state of the keyboard.c
- TODO: ensure it works corectly and use this implementation in most other places in win32k when we need to check the state of a key. To ensure nothing breaks, this change affects only GetKeyState
svn path=/trunk/; revision=51295
- Fix more subtle bugs:
- Check with long line name if the current connection is activated when adding lines to dialog or when adding lines to the preferences dialog
- Use the temporary preferences context when writing the line settings
- Don't destroy the mixer when the preferences dialog is closed. the new mixer will be used for the application when the preferences are changed
- Pass the current selected line to the preferences dialog
svn path=/trunk/; revision=51292
- Only call hooks and check if Win32Thread is in cleanup if the windows procedure is being called.
- Set DispatchingListEntry.Flink to NULL. The real fix for bug 5580. This bug may be the cause of other crashes experienced by users just before loading desktop. Found by rafalh.
svn path=/trunk/; revision=51291
- Added more missed mouse injection flag and extra data being passed in and from set cursor position. Wine mouse low level hook msg tests now see it and check the validity of the hook message.
svn path=/trunk/; revision=51290
- Always write line states configuration values in one step
- Fixes the problem when old line state settings were not present
- Advantage is that it is faster and sndvol32 now works in ReactOS
svn path=/trunk/; revision=51286
- Polish translation by Adam "Saibman" Stachowicz
[SNDREC32]
- Polish translation by Adam "Saibman" Stachowicz
[SNDVOL32]
- Polish translation update by me
[TRANSLATION]
- Polish translation fixes (spelling, typos, grammar) and updates on multiple modules by Adam "Saibman" Stachowicz, few small corrections by me.
See issue #6087 for more details.
svn path=/trunk/; revision=51285
- Fix TrackMouseEvent! Keyboard and Mouse pass all the right flags. Update to accelerator code. Fixed recursion in global hooks. Hack-implement GetMouseMovePointsEx due to slack time and boredom, it passes the input tests but does nothing. Pass all the test_TrackMouseEvent!!! test_accelerators and test_menu_messages work too, so far no hangs.
svn path=/trunk/; revision=51283
- Fix incorrect prototypes and partially implement fast unsafe versions of activation context activation/deactivation, however disable them for now, until the new loader is in place.
- Fix their usage in the new loader code, which by mistake (which got copypasted into three other places) was passing a totally empty activation context to the activate function.
svn path=/trunk/; revision=51281
ExInitializePushLock takes a PEX_PUSH_LOCK as parameter, do it the same in the inline version instead of casting to PULONG_PTR everywhere
svn path=/trunk/; revision=51278
- Zero-initialize base address in BasepLoadLibraryAsDatafile. Should fix 2nd stage problems.
- Add a trace debugprint, silented by default.
See issue #6092 for more details.
svn path=/trunk/; revision=51277
- co_IntSendMessageWithCallBack: Insert the message into the messagequeue before attempting to wake the queue. Probably fixes bug 5580.
- Remove the use of MSQ_SENTNOWAIT that was used to determine if messagequeues needed dereferencing. Instead use the SenderQueue member of the message.
- Do not close the NewMessagesHandle member of the messagequeue in IntDereferenceMessageQueue.
The last reference on a messagequeue could be released by a thread belonging to a process other than the on that created the handle.
Also the creating process could already be long gone. Instead close this handle in MsqDestroyMessageQueue which is in the thread/process that created the handle. Problem noticed by smiley.
- Revert change done in 51254. James you are indeed correct.
Testing shows that the WM_PARENTNOTIFY message is not sent when the program is exiting.
svn path=/trunk/; revision=51275
- Implement retrieving current volume level and mute state and set the dialog controls on startup
- Remove last line separator from GUI
- Implement updating slider / switch control when another application modifies volume settings
- Implement writing current selected lines settings into registry when the preference dialog is closed
- Implement helper functions which receive / set volume level
- Sndvol32 is now fully functional and has been tested in Windows XP SP3
- TODO: implement support for setting volume balance (left - right slider)
- TODO: Resources have not yet been commited
svn path=/trunk/; revision=51274
We provide everything (headers, libraries, executables...), so tell it to cmake
[CRT]
The crt library must provide a and w version of the functions and not rely to _UNICODE being set or not
svn path=/branches/cmake-bringup/; revision=51273
- Finish the ldr.c rewrite. Properly implement LoadLibraryExW (which contained some good code pieces added by Alex). Implement BasepLoadLibraryAsDatafile, however it doesn't support redirection yet and a call to load alternate resource module is commented out (it's totally absent in the existing ntdll/ldr API now).
- Mark GetDllLoadPath as deprecated, should be removed when SearchPathW is rewritten (or at least, reviewed). The new function to use is BasepGetDllPath.
svn path=/trunk/; revision=51272
- Implement loading of resources and parsing the dialog template in memory
- Store the created windows in an array
- Implement destroying of controls when mixer is changed / line selection is altered
- Implement generating layout for each new line
- Implement setting line name
- Resources have not yet been commited
- Sndvol32 is not yet functional
svn path=/trunk/; revision=51271
- Rewrite most of kernel32 module loading APIs (except for LoadLibraryExW, for testing purposes it's gonna be committed in a separate commit). Rewritten parts include:
* Better definition of LOADPARMS32 structure for LoadModule, and rewritten LoadModule which is now a well structured function with proper parameters validation, proper work algorithm and setting proper last error values in failure cases. Optimization of allocating MAX_PATH buffer on stack instead of in the heap is saved from the old version of the function.
* LoadLibraryA, added a path-hack (present at least up to Windows 2003) for twain_32.dll loading (not that important, but still a compatibility issue).
* Make GetProcAddress() use BasepMapModuleHandle for mapping the passed-in handle (including properly handling NULL module handle which was incorrectly handled by the old implementation), and check for a special failure case when LdrGetProcedureAddress returns pointer to the image base address.
svn path=/trunk/; revision=51270
- Pass all the wine test_capture_1/2/3/4.
- Fix the menu tracking pop up hang issue by looking for deceased windows that pass on while in the start of tracking. This could be a fix for other wine tests.
- ReactOS is not Linux and wine is not enough!
svn path=/trunk/; revision=51265
- Implement LsaLookupNames2 in advapi32.dll.
- Implement LsarLookupNames3 in lsasrv.dll as a mock-up which returns a hard-coded user account (Administrator).
This fixes several wine tests.
svn path=/trunk/; revision=51257
- co_MsqDispatchOneSentMessage: After calling the windows procedure check if the message was a callback. If so place the message into the calling MessageQueue. Also check if the current MessageQueue is the calling MessageQueue in which case dont call the Windows procedure but call the callback routine. SendMessageCallback is now properly implemented.
- Remove the use of MSQ_SENTNOWAIT and instead use the SenderQueue and CallBackSenderQueue members to determine if the MessageQueues need to be dereferenced.
- Modify co_MsqSendMessage to accept more parameters so that it can handle Notification, Callback and Internal messages. These changes make this function more complex but removes duplicate code in messages.c and move the handling for all queued messages in one location.
- co_IntSendMessageWithCallBack: If the callback is for the same Message Queue, call it vice queuing it.
Insert the message into the MessageQueue before waking the thread to handle the message. Should fix bug 5580.
svn path=/trunk/; revision=51255
- Minor cleanup, better flag names (thanks to ProcessHacker team for the good names and values).
- Return error in failure path of BasepGetModuleHandleExW.
- Optimize GetModuleHandleExA so that it calls the internal routine directly, without going through GetModuleHandleExW first and thus validating parameters second time.
svn path=/trunk/; revision=51253
- Set Control variable to null, which prevents that mixer controls being assigned to another line when the current line does not have any controls. Bug was highlighted when opening the property window. In that case sndvol32 frees the previous setting, so all mixer lines are also getting freed including their controls, which caused a double free in the case the line did not have any controls assigned.
svn path=/trunk/; revision=51252
- Working on wine win test_capture_4, pass all the tests in test_capture_4_proc but still waiting forever in MenuTrackMenu:WaitMessage!
svn path=/trunk/; revision=51248
- Implement BasepMapModuleHandle and use it in GetModuleFileNameW. Later, it'll be used by other module and loader related functions in kernel32.
- Wrap a call to RtlImageNtHeader into SEH, because winetest and for some reason FF installer call it with incorrect handle and expect it to not crash (Windows also has SEH, but inside RtlImageNtHeader).
- Fix null-termination problems in GetModuleFileNameA, and incorrect calculation of characters returned in GetModuleFileNameW.
- Various smaller improvements to the rewritten code (missing parenthesis, better variable names, use pretty LDR_ macros from http://msdn.microsoft.com/en-us/library/ms684179(v=vs.85).aspx ).
- Thanks for your feedback!
See issue #6079 for more details.
svn path=/trunk/; revision=51246
- Add HAVE_GETPAGESIZE to host config.h. should fix linux build (fix by Sylvain)
- uncomment @echo off in configure script
- Move a DECLSPEC_NORETURN
svn path=/branches/cmake-bringup/; revision=51242
Fix typos in lsa.idl (LsarLookupNames2, LsarLookupNames3 and LsarLookupNames4) and fix resulting errors in lsasrv.dll.
svn path=/trunk/; revision=51240
- Rewrite large part of ldr.c code, including FreeLibrary, FreeLibraryAndExitThread, GetModuleFileNameA/W, GetModuleHandleA/W, ExA/W. Now they properly validate parameters, common code for getting module handles is factored into a standalone function which is reused for all 4 APIs instead of duplicating almost the same code four times. GetModuleFileNameA became a typical A->W wrapper around GetModuleFileNameW so duplication is also removed there.
- Getting module handle also became a correct implementation because now it takes process's dll path into account instead of always assuming a "default" one (which is incompatible with a new ntdll loader).
- Fixed a typo in BasepGetProcessPath which led to incorrect value being returned. Downgrade two debugprints.
svn path=/trunk/; revision=51238
- Fix the return checks for co_MsqWaitForNewMessages, now the wait_idle test no longer randomly crash, also tested the testkbd check from bug 3212. Setup co_MsqWaitForNewMessages to act like get message, checking QS bits and dispatching send messages if queued.
- Added desktop hit test data point, now the new track mouse timer code works. Not yet committed.
svn path=/trunk/; revision=51237
- Add RTL_PERTHREAD_CURDIR structure definition.
- Fix RTL_DRIVE_LETTER_CURDIR type vs. structure definition typo (adding more underscores is a truly universal fix!).
svn path=/trunk/; revision=51235
- Fix warnings for both msc and rbuild
- These fixes will be applied to the FullFat source too, so I've not bothered with a local ros.diff
svn path=/trunk/; revision=51234
- Update FullFat to the latest SVN code on James' recommendation.
- This release has many advantages over the previous, including full UTF-8 and UTF-16 support and the ability to modify attributes and timestamps. All of which are important for ros.
- It currently has a few warnings (in both gcc and msc), so I've had to turn allow warnings on.
svn path=/trunk/; revision=51229
- Update FullFat to the latest stable release (1.0.5)
- I'm awaiting a reply from James as to when 1.1.0 will be released as it have some nice improvements.
svn path=/trunk/; revision=51227
- Simplify co_IntWaitMessage
- In co_IntGetPeekMessage, call co_MsqWaitForNewMessages instead of co_IntWaitMessage when we need to wait for a new message. co_IntWaitMessage calls co_IntPeekMessage again without removing messages. The problem is that processing some messages produce other messages or hooks even though the message must not be removed. As a result some messages like WM_NCHITTEST were received twice by the application
- Should fix bug 3232
svn path=/trunk/; revision=51224
- Fix a typo which led to incorrect logic of TLS allocation. This repairs those installers broken by r51051.
See issue #6009 for more details.
svn path=/trunk/; revision=51223
* Fix bug in RegQueryValueExA, which causes buffer overflow
* Fixes systeminfo in ReactOS
See issue #6050 for more details.
svn path=/trunk/; revision=51222
Don't translate System Type
Don't mix TCHARs and chars
Fix Time Zone for not English systems
Use proper registry key for Install Date
Format dates using system settings
Format numbers
svn path=/trunk/; revision=51221
- Do not make an artificial target for gendib generated files.
GENERATED property for source files is there for that and is automatically set by cmake for custom command OUTPUT files.
svn path=/branches/cmake-bringup/; revision=51219
- Fix wine post message test, corrected the peek message window selection.
- Add a check for dead windows and new message queue flags.
- Other miscellaneous fix ups.
svn path=/trunk/; revision=51216
Add System Manufacturer and System Model to resources
Convert TABs to spaces in resources
Sort strings in resources a bit
svn path=/trunk/; revision=51215
Convert TABs to spaces
Print output to stdout instead of stderr
Close opened registry keys
Simplify a bit
Add System Manufacturer and System Model
Add System Up Time (patch by Edijs Kolesnikovics <terminedijs at yahoo dot com>, see issue 6055 for more details)
svn path=/trunk/; revision=51212
- Translation patch by Adam "Saibamen" Stachowicz. See issue #6057 for more details.
[EXPLORER]
- Translation patch by Adam "Saibamen" Stachowicz.
[SERVMAN]
- Translation patch by Adam "Saibamen" Stachowicz.
[SHELL32]
- Translation patch by Adam "Saibamen" Stachowicz. See issue #6060 for more details.
svn path=/trunk/; revision=51206
- Rename global vars to some better names (some of the names seen in Windows 2003 asserts and text messages).
- Add BaseDefaultPath initialization, which will be needed but the new kernel32/ldr code, which is in turn required by the new ntdll/ldr code.
- Add some beginnings of BasepGetDllPath(), but it returns NULL anyway now so no change in execution.
svn path=/trunk/; revision=51197
- Fix compile warning (uninitialized use). Based on patch by Edijs Kolesnikovics <terminedijs at yahoo dot com> slightly modified by Aleksiej
See issue #6058 for more details.
svn path=/trunk/; revision=51196
- Fix compilation warning. Patch by Edijs Kolesnikovics <terminedijs at yahoo dot com>
See issue #6059 for more details.
svn path=/trunk/; revision=51195
- Remove IMG_SCN_TYPE_NOLOAD support which was firstly removed by Alex, then reenabled by Herve. Hopefully it works this time.
- Return proper image size after successfully mapping an image section (previously, 0 was always returned). The change above is necessary for returning correct size.
svn path=/trunk/; revision=51192
- Properly implement RealGetWindowClass, fixes the ApiTest. Keeping the Ansi support for now, will use it as a reference.
svn path=/trunk/; revision=51187
- Finish object attribute support part 2.
- Removed DC from the list, ReactOS handles this in a different way. Comment out Brush and Pen switches until proper handling is implemented.
svn path=/trunk/; revision=51185
Fix a bug causing "Attempted to lock object, wrong reuse counter" messages when CMD was started in LiveCD. Once got desktop HDC was used in different threads. When thread, which created handle terminated, DC handle was removed and next calls to CreateIcon* API was using invalid hDC.
svn path=/trunk/; revision=51184
Move utf-8 resources back into rsrc.rc, move avifile_Bg.rc to the bottom. Fixes compilation with MSVC
svn path=/branches/cmake-bringup/; revision=51183
Fix crash if option, which needs number is on the end of command line
Get rid of GetULONG and GetULONG2 functions. MS doesn't check input so precisely
Command line options can be preceded by "/"
See issue #6047 for more details.
svn path=/trunk/; revision=51177
Set HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\InstallDate to install date.
Patch by Edijs Kolesnikovics <terminedijs at yahoo dot com> slightly modified by me
See issue #6044 for more details.
svn path=/trunk/; revision=51176
ScmWriteDependencies:
- Fixed an off-by-one bug. This bug caused a wrong calculation of the 'DependOnGroup' value data length. Group dependencies were not written properly to the registry.
- Write 'DependOnService' and 'DependOnService' values to the registry only if the dependency strings are not empty. Delete a value if the corresponding dependency string is empty.
This fixes another winetest.
svn path=/trunk/; revision=51172
- Return STATUS_IMAGE_NOT_AT_BASE when mapping an image section to a different base address. New ntdll loader depends on this and fails to relocate the image if this status is not returned.
svn path=/trunk/; revision=51171
Rename IntResetMonitorSize to IntUpdateMonitorSize
Use IntUpdateMonitorSize instead of setting monitor size manually when attaching monitor
Change display device name from VIDEOx to DISPLAYx as it is in Windows (fixes user32:monitor winetest)
[USER32]
Return FALSE in EnumDisplayMonitors if EnumProc returned FALSE (fixes user32:monitor winetest)
svn path=/trunk/; revision=51169
- Add DoubleCommander to app list. Russian and Ukrainian translation by Igorko, Polish translation by me.
[CMD]
- Fix Russian translation of cmd. Patch by seven_ro. See bugreport no.#6043
svn path=/trunk/; revision=51168
- Enable the desktop class and set the ReactOS variant of csrss back to local desktop class. Now all class tests pass, see bug 4778.
svn path=/trunk/; revision=51145
Change strncpy calls to RtlStringSbCopyA (PART 1/x)
Fix bug in MmLoadSystemImage which caused FileName parameter to be freed
svn path=/trunk/; revision=51138
Fix bug causing no NULL termination of strings after conversion in EnumDisplayDevicesA. Fixes one usr32:monitor winetest
svn path=/trunk/; revision=51137
Import strsafe.h from mingw-w64. It's more complete compared to our headers
[DDK]
Import ntstrsafe.h from mingw-w64 (converted from strsafe.h). It's more complete compared to our headers
[NTOSKRNL]
Use sizeof instead of magic numbers
Let's use strsafe functions now instead of strncpy/wcsncpy, which doesn't always NULL terminate :)
svn path=/trunk/; revision=51135
Sync RunFileDlg to wine. It was crashing for one char input.
Add "All files (*.*)" filter to RunDlg and OpenWith dialogs.
Improved polish translation.
Translators: please localize rest...
svn path=/trunk/; revision=51127
Revert r51108. ProbeForRead doesn't check if memory is readable.
Improve comments and replace trick raising exception with call to ExRaiseAccessViolation.
[WIN32K]
Remove ASSERTs in IntSetSysColors: this function is only called in NtUserSetSysColors in SEH block. It shouldn't crash for NULL pointers.
svn path=/trunk/; revision=51126
- Fix a few bugs (wrong variable usage, wrong variable initialization) which led to incorrect snapping of import address table.
- Wrap LdrpSnapThunk() invocations into SEH.
svn path=/trunk/; revision=51123
- Remove VirtualBox SYSENTER-hack added in r50991.
- Bring back NT-compatible SYSENTER support detection changed in r50991.
svn path=/trunk/; revision=51120
- Implement LdrpInitializeProcess(). Currently unused.
- Add a whole bunch of supportive PE related code. Used only by the new code, which is not enabled yet in trunk.
- Implement LdrpCheckForLoadedDll().
- Move on to usage of various global vars defined in the new code, rename some of them to match.
- Almost no changes for the codepath in trunk.
svn path=/trunk/; revision=51117
- Implement calling OkayToCloseProcedure callouts to win32k for desktop and window station objects
- Fix a bug that caused ObpCloseHandle to return success even when OkayToCloseProcedure failed
[win32k]
- Rewrite SetProcessWindowStation to actually set the current window station and close the previous one
- Implement OkayToCloseProcedure callouts from the kernel to prevent closing the current desktop or window station
svn path=/trunk/; revision=51115
- Danish translation by Wilhelm Wiens
[RAPPS]
- Polish translations of SumatraPDF entry by Adam Stachowicz
[SETUP]
- Polish translation of 'VBR only' in the reactos.exe
svn path=/trunk/; revision=51098
- Finishing with setting window class types from the window proc. Next step will be checking it in the User32 message exchange instead of using the class structure. This was recognized with the MDI tests, this too could fix the combo listbox issue.
- Minor static test fixes and miscellaneous notes.
svn path=/trunk/; revision=51095
- Move the appropriate code from window.c to winpos.c, decreasing the size in window.c. Adding proper checks for maximums and desktop/message windows.
- Started MinMaximize, experimental, based on an idea and boredom, needing a good test case. Also away of troubleshooting our window positioning with minimize and maximizing windows.
svn path=/trunk/; revision=51094
RQueryServiceConfig2A/W must return ERROR_SUCCESS if the Description value of a service does not exist. This fixes a winetest failure.
svn path=/trunk/; revision=51092
- Activate "Show window contents while dragging" button for all languages present. Spotted by Piotr_hh
See issue #6002 for more details.
svn path=/trunk/; revision=51091
Remove HAL entry point (HalInitSystem) from generic lib, and move it to generic libs that handle different build flags.
This fixes HAL MP entry point, for example
svn path=/trunk/; revision=51090
- Implement the set of DeferWindowPos functions. Regedit (one of many application) uses it and allocated three when it should be four. This was a good test to verify the batch list growing routine. Instead of drawing per DeferWindowPos call, now a real list is created and run down as a batch.
- ReactOS should handle this the correct way and a good test case is located here: http://bugs.winehq.org/show_bug.cgi?id=23187
- The code is from wine and modified for the use in ReactOS.
svn path=/trunk/; revision=51089
- Rewrite the very first initialization routine: LdrpInit().
- Rename LdrpInit2() to LdrpInitializeProcess(), and LdrpAttachThread() to LdrpInitializeThread(). Still old code there.
svn path=/trunk/; revision=51088
- Use LdrpRunInitializeRoutines() instead of the old LdrpAttachThread(). This adds calls to unimplemented context functions, but since they were not called at all by the old code it shouldn't change the behaviour significantly. Besides of that, it's start of significant changes in the "mainstream" codepaths, so regressions might occur.
svn path=/trunk/; revision=51085
- Implement fetching address of a security cookie in an image. Still not used even if it's present there.
- Silence some debug prints.
svn path=/trunk/; revision=51084
- Limit unimplemented warnings about RtlActivateActivationContextUnsafeFast calls for now so that they don't flood the log too much.
svn path=/trunk/; revision=51083
- Fix a typo in LdrQueryProcessModuleInformationEx().
- Stub LdrpEnsureLoaderLockIsHeld() for now.
- Improve LdrpRunInitializeRoutines() and enable it (just compiling, it's not in the running path of the code yet).
- Fix an uninit var warning in ldrpe.c
- Implement LdrpClearLoadInProgress().
- Rename "ExeModule" to a more consistent LdrpImageEntry.
svn path=/trunk/; revision=51079
Patch by Rafal Harabien (rafalh1992 at o2 dot pl)
- Change () to (VOID) in function declarators. Part 6/XX.
[USETUP]
Patch by Rafal Harabien (rafalh1992 at o2 dot pl)
- Initialize Offset field of partition list. Spotted using DPH.
svn path=/trunk/; revision=51072
Thomas Faber
- Fix memory leak in RenderHDROP.
Modifications by me to have just a single return statement in the function.
See issue #5998 for more details.
svn path=/trunk/; revision=51069
- Don't alter the IoStatus.Information field with the number of bytes written
- Fixes sndrec32 crash (now also working in ReactOS)
- Should be merged to 0.3.13 ;)
svn path=/trunk/; revision=51067
- Rewrite LdrQueryProcessModuleInformation.
- Start committing PE loader related code (quite a lot of it, and still not all parts yet), unused right now so won't break anything.
svn path=/trunk/; revision=51065
Change all ftp:// links to http:// ones. The changes to the fragile Mozilla links should be reverted as soon as our/Wine's URLDownloadToFileW function finally supports FTP URLs.
Vote for Wine Bug 26445 (http://bugs.winehq.org/show_bug.cgi?id=26445) to get this fixed!
svn path=/trunk/; revision=51063
- Hack away LdrVerifyMappedImageMatchesChecksum() invocations with zero ImageSize until MM is fixed.
- Implement a helper ChkSum() routine to be used later by LdrVerifyMappedImageMatchesChecksum().
svn path=/trunk/; revision=51061
- Revert 38185 which synced fputwc to the Wine version for the sake of fixing Winetests.
This removed the implicit ANSI conversion, which was honestly checking the wrong flags. The new version correctly checks for WX_TEXT and only does the conversion in this case. This way we also don't fail any additional Winetests.
- Simplify streamout_char in the new printf implementation and use _fputtc instead of _flsbuf.
This way, we correctly do an implicit ANSI conversion when required.
- Fix fgetwc declaration.
Thanks to Timo for many hints and assistance. Most of this code is actually his idea :-)
Thanks to Olaf for providing a build server to speed up my code tests.
See issue #6007 for more details.
svn path=/trunk/; revision=51058
Define HAL_IRQ_TRANSLATOR_VERSION constant
[NTOSKRNL]
When needed, use HAL_IRQ_TRANSLATOR_VERSION from NDK and don't define it locally
svn path=/trunk/; revision=51057
- Mario Kacmar: Added 'VBR only' as possible option to the reactos.exe - it is necessary to add it to other languages (I would not mess up codepages).
See issue #5734 for more details.
svn path=/trunk/; revision=51055
Fix missing "Open" menu entry on right click for All Users desktop shortcuts.
Patch by Thomas Faber <thfabba at gmx dot de>
See issue #5998 for more details.
svn path=/trunk/; revision=51053
- Rewrite TLS part of the ntdll loader. Old TLS-related messy, duplicated and hard to maintain code is gone.
- Disclaimer: All new ldr-rewrite code is almost entirely based on Alex's patch made in 2006, with my improvements, bugfixes and adapting existing codepaths.
- File naming convention: Files of the ldr*.c mask are considered as a new rewritten loader code. All other files in ntdll/ldr directory are old code which soon may be deprecated.
svn path=/trunk/; revision=51051
Improve configure script. It now defaults to nmake for MSVC, if you want a "VSSolution", pass exactly this as 1st paramter
Patch by Samuel Serapion
svn path=/branches/cmake-bringup/; revision=50857
Add support for VS 8-10 and SDK to configure script, add error checks. Patch by Samuel Serapion with small modifications by me.
svn path=/branches/cmake-bringup/; revision=50810
- Remove include/reactos and include/reactos/wine include directories for host tools
- Add include/host/wine directory
- Move wpp.h to include/host/wine
- Put a seperate config.h into include/host
- Add a dummy debug.h and guiddef.h, wine/list.h, wine/port.h and wine/rpcfc.h that link to the original ones to include/host
svn path=/branches/cmake-bringup/; revision=50763
Fix compilation of nsiface.idl with midl. Only pointers to interfaces are allowed in typedefs.
Also define nsISupports to IUnknown, because thats what it is and nothing else is allowed.
Please someone pass this to wine.
svn path=/branches/cmake-bringup/; revision=50702
- Drop support for MSVC versions below 1300
- use the same flags for WDK / host as normal
- Disable optimization on Debug builds, its incompatible with /ZI
svn path=/branches/cmake-bringup/; revision=50689
- Get rid of the redundancy in setting msvc flags, and set /ZI instead of /Zi for both host and target modules.
- Get rid of the redundancy in setting the default behaviour of the FIND_XXX() commands between toolchain files.
svn path=/branches/cmake-bringup/; revision=50688
- Allocate memory for the logical pin array
- Use the correct counter variable for logical pin array
- Found by msvc
svn path=/branches/cmake-bringup/; revision=50674
- Sort exports like on Windows
- Implement some functions as proper stubs
[SECURITY]
- Cleanup exports (remove 31 of 77 functions)
- properly forward all exports
- Link to secur32 for forwarded exports
svn path=/branches/cmake-bringup/; revision=50597
mscms needs to link to advapi32, currently gcc builds optimize the calls to Reg* apis away, because of unimplemented fucntions.
svn path=/branches/cmake-bringup/; revision=50586
(M)IDL doesn't allow definiting multiple identifiers with the same name and different case. Define MSIRUNMODE to MSIRUNMODE_T so it doesn't conflict with MsiRunMode anymore.
svn path=/branches/cmake-bringup/; revision=50580
Don't automatically add stub files to the sources, instead add them manually where neccessary. also fix a number of spec files to resemble actual state of implementation.
svn path=/branches/cmake-bringup/; revision=50574
Define _MIDL_USE_GUIDDEF_, this causes __decspec(selectany) to be used on the iids. This way we don't need any linker flags to silence warnings and it also fixes errors when linking later.
svn path=/branches/cmake-bringup/; revision=50573
The old uuid library was one file containing all the uuids, generated from the psdk headers, which results in all GUIDs being linked, as soon as a single one is used. Also while widl creates DEFINE_GUID() entries in the header files, midl only creates "extern GUID", so this didn't work on MSVC. The new version uses iid files generated from the idl files and some extra C files. This works with both midl and widl, reduces overhead when linking uuids and is much closer to MS uuid lib.
svn path=/branches/cmake-bringup/; revision=50569
- Add a macro add_iid_library to generate a static lib containing iids
- Fix get_includes and get_defines
svn path=/branches/cmake-bringup/; revision=50559
Convert checksum.S to new ML compatible syntax. Resulting obj was compared and is identical to trunk (both GAS and ML)
svn path=/branches/cmake-bringup/; revision=50545
Fix crt/bootcdregtest regression, due to _USER32_WSPRINTF.
add_definitions does not apply to target, but current directory.
svn path=/branches/cmake-bringup/; revision=50535
_Interlocked(Compare)ExchangePointer is an intrinsic that doesn't exist on x86, use portable version.
svn path=/branches/cmake-bringup/; revision=50522
On MSVC mark abs, labs and strcmp as intrinsics (needed when compiling with /O2+) and disable a warning that they are not intrinsics (when compiled with /O1)
svn path=/branches/cmake-bringup/; revision=50494
On MSVC builds, compile native SEH support into PSEH library.
This should be part of the crt, but our crt is too disorganized to handle that. The empty C file is for cmake to get the library right.
svn path=/branches/cmake-bringup/; revision=50486
Refactor IDL compilation:
- Move all macros to idl-support.cmake
- Make rpc proxy a static library, instead of fiddeling with SOURCE variable
- Merge client and server rpc libraries into one library
- Reorder global makelist file a little
- Improve configure script
- Improve naming, use add_* when a target is added, generate_* when only a file is generated
- Don't compile wbemcli.idl on MSVC (temp hack)
- Don't compile rossym on MSVC, we don't use it
- Use ml64 as assembler when compiling for amd64
- fix spec2def invocation
svn path=/branches/cmake-bringup/; revision=50478
- _mingw.h: Fix definition of _CRT_ALIGN for MSVC
- intrin.h: _InterlockedAddLargeStatistic is not an intrinsic on MSVC for x64
- intrin.h: Fix return type of _InterlockedOr64
- wdm.h: Fix warnings in KfRaiseIrql and RtlExtendedMagicDivide inline implementations
- winnt.h: Simplify definition of GetCurrentFiber
svn path=/branches/cmake-bringup/; revision=50470
The function dexec uses the varable p only in one location where its checked against nil, It neither initialized, not used anywhere else. This is probably a typo and b->p was meant. Yes I'm completely guessing, but this code doesn't give any other chance then guessing and it will probably not be worse then using an uninitialized variable...
@original author: please review.
svn path=/branches/cmake-bringup/; revision=50450
Move secure api functions from external file into string.h and use standard macros. Fixes 2 warnings.
svn path=/branches/cmake-bringup/; revision=50431
Provide the asm offset constants that are used throughout the code instead the official ones. We'll have to decide which way to go later.
svn path=/branches/cmake-bringup/; revision=50401
- Fix wmi and drmk entry points.
- Don't use both reactos and wine debug headers in iphlpapi. Fixes the macro redefinition warnings.
svn path=/branches/cmake-bringup/; revision=50353
- Prevent cmake from escaping preprocessor definition values added via add_definitions.
- Fix a FIXME in oleaut32 and urlmon builds.
svn path=/branches/cmake-bringup/; revision=50333
- Delete an unneeded lib.mak
- Don't build bin2c, we don't seem to need it.
- A minor cleanup, no functionality change intended.
svn path=/branches/cmake-bringup/; revision=50308
Add generation of a depencency graph for shared libraries.
The output is a graphml file. Can be enabled with GENERATE_DEPENDENCY_GRAPH switch.
svn path=/branches/cmake-bringup/; revision=50243
- Disable some resource files (mostly japanese ones) with a FIXME. They prevent us from compiling in unix due to an iconv issue.
- Lower the minimum required cmake version to 2.6
- Fix browseui resource paths.
svn path=/branches/cmake-bringup/; revision=50224
LD is stupid and doesn't handle stdcall decoration as proper as dlltool does (after we provided a patch). Passing --kill-at, also kills C++ mangled names and exports with stdcall decoration are imposible. In trunk we use dlltool to generate an exp file that we link with LD, but in the cmake branch we pass the def file to LD directly. Luckily we have a tool called spec2def that can handle these things. We now generate 2 different .def files, one for LD, containing the undecorated export name forwarded to the decorated symbol name (FooFunc=FooFunc@12), while the 2nd def file which is passed to dlltool for exportlib generation has full stdcall decorations. --kill-at is now passed to dlltool only. This commit might break msvc, but should be pretty easy to fix.
svn path=/branches/cmake-bringup/; revision=50217
- Add MiPdeToAddress and MI_IS_PAGE_LARGE for amd64
- Fix a pragma message
- Add some missing globals for amd64
svn path=/branches/cmake-bringup/; revision=50209
- Fix set_entrypoint for other architectures than x86, add an optional 3rd parameter for stdcall stackbytes
- don't add stdcall decoration to def files for other architectures then x86
svn path=/branches/cmake-bringup/; revision=50188
- Add rostests to build when the folder exists.
- Alter the importlib target macro to handle definition files along with spec files.
svn path=/branches/cmake-bringup/; revision=50101
Add a simple configure script for windows hosts. Run it from any path, where you want to build both the tools and reactos in 2 subfolders.
svn path=/branches/cmake-bringup/; revision=50054
- Fix msacm32.dll entry point. This fixes the World Peace issue, I mean, bootcd and livecd userinit exception issue. We get to the shell now in both. Thanks to Johannes for helping me investigate this.
svn path=/branches/cmake-bringup/; revision=50050
- Improve some modules.
- Convert notifyhook def into a spec, and add the import lib target.
- Get rid of the import libs that we already generate (the rest are still just temporary)
svn path=/branches/cmake-bringup/; revision=49937
- Group some cross compiling related sections.
- Implement add_importlib_target() for gcc.
- Improve the msvc version.
svn path=/branches/cmake-bringup/; revision=49903
On MSVC builds link eventvwr, dwnl, ftp, ping, telnet, tracert to ntdll to resolve _allmul and friends
link ftp to oldnames
svn path=/branches/cmake-bringup/; revision=49902
Add stubs file to source automatically.
Gcc: fix set_rc_compiler() so that it doesn't conflict with custom_incdefs().
svn path=/branches/cmake-bringup/; revision=49893
- Link wine dx, that do forwards to the respective import libs (MSVC requires that)
- add import lib for d3dx9_24
- add IID_IUnknown to d3dx9_36_main.c for MSVC
svn path=/branches/cmake-bringup/; revision=49891
- add support for stubs generation
- add support for "proper" stubs, with calling convention and parameter dump
- handle -private
- change invokation a bit
[CMAKE]
- remove the old VARIANT hack
svn path=/branches/cmake-bringup/; revision=49862
Don't use stdole32.tlb.
Seems that this is the problem that made midl fail. Maybe not. Yesterday it was all about VARIANT, which has no effect today. Let's see what tomorrow will bring :-?
svn path=/branches/cmake-bringup/; revision=49860
remove /GR from c++ options, it's on by default and we get a warning when additionally specifiying /GR-
svn path=/branches/cmake-bringup/; revision=49857
Don't compile CRT_fp8.c, CRT_fp10.c and xtxtmode.c. we don't need those and MSVC complains about duplicate symbols, as it's exported from msvcrt already.
svn path=/branches/cmake-bringup/; revision=49854
Reapply the midl hack. Someone wants to trick me or something. I could swear it just worked one minute ago.
svn path=/branches/cmake-bringup/; revision=49847
- Remove the dxsdk hack. It works now without. Why? I don't know. Why didn't it work before? I have no idea. Maybe the newlines. Anyway it works. If someone breaks it, I'll ki^Z^Z revert it ;-)
svn path=/branches/cmake-bringup/; revision=49845
Add a hack for midl. Don't ask me why, but renaming IEnumVARIANT to something different globally fixes the compilation of dxsdk. someone with more idl knowledge should check that.
svn path=/branches/cmake-bringup/; revision=49843
Add a number of GUIDs. They are duplicated from idl generated headers, because of a small difference between widl and midl. widl generates "DEFINE_GUID" entries for each interface, while midl generates "extern GUID" only. The latter matching MS SDK headers. If someone can think of a more elegant way of solving this, let me know.
svn path=/branches/cmake-bringup/; revision=49837
- Link libxslt to ntdll for MSVC
- Don't compile some directx stuff with MSVC atm
- Add importlib target for dinput
- add importlib target for wined3d and use autogenerated def
- Improve quartz
svn path=/branches/cmake-bringup/; revision=49833
- Fix forwarded export of GetLocaleInfoW
- Rename _inp & co to MSVCRT__inp, because they are only available as intrinsics on MSVC
- Implement them using the intrinsic
svn path=/branches/cmake-bringup/; revision=49822
Merge .CRT section into .rdata when linking to mingw lib. This silences the warnings about uninitialized constructors.
svn path=/branches/cmake-bringup/; revision=49799
No mesa32 for MSVC for now. It has lots of assembly that would need to be ported. It also talkes ages to compile.
svn path=/branches/cmake-bringup/; revision=49796
Fix rc syntax (multiline strings with \ at the end). This is to compile with MS reasource compiler, we can hope wine will adopt this, because latest versions of gcc's preprocessor also don't support this anymore.
svn path=/branches/cmake-bringup/; revision=49788
- add #pragma code_page(1252) at the end of some resource files to help msvc's resource compiler
- "Untranslate" a few Japanese strings, as they contain illegal characters. Someone speaking Japanese should retranslate them using proper escape sequences where neccessary.
- Don't include headers twice
svn path=/branches/cmake-bringup/; revision=49776
- Add a workaround to dispex.idl to prevent sal annotations to be generated. A better solution would be to have definitions in pur headers, but they conflict with external c++ headers.
- Add a number of missing interfaces
- Fix MIDL compilation of mlang, by adding adding a library definition
svn path=/branches/cmake-bringup/; revision=49762
Use spec file for ntoskrnl and set image base to 0x00400000 on MSVC builds. This is the value that Windows has and the ms linker doesn't like to use the kernel address.
svn path=/branches/cmake-bringup/; revision=49709
Don't include initguid.h in a global header. It causes the guids to be generated in every object file, using 16 bytes per guid. We were lucky that the linker seemed to detect duplicates, so that the resulting image size isn't affected. Add a new file guid.c, that only contains the guids.
svn path=/branches/cmake-bringup/; revision=49708
MIDL doesn't like if interfaces are not defined. Add some empty interfaces to activedbg.idl make it compile.
svn path=/branches/cmake-bringup/; revision=49703
Fix HTMLFormElement uuid and add missing retval in IHTMLDocument6
Add back some previously disabled idl files for msvc
svn path=/branches/cmake-bringup/; revision=49702
- Add a number of stub functions to fix build
- Comment out some exports that are probably 2k/nt4 only
- don't set -u flag for MSVC
svn path=/branches/cmake-bringup/; revision=49684
- Add /nologo, when preprocessing asm files
- Move add_pch to compiler specific files (maybe someone can fix it for MSVC)
- Update set_module_type for MSVC
svn path=/branches/cmake-bringup/; revision=49683
- Use portable interlocked functions in code, define them to intrinsics for x86 and x64 in the header
svn path=/branches/cmake-bringup/; revision=49680
- Use ___readcr4 instead of __readcr4, because MSVC doesn't like this intrinsic to be redefined.
- Add _ftol2
svn path=/branches/cmake-bringup/; revision=49679
Get rid of stub entries in spec file, either replacing them with a real export or commenting them out in case of win2k only exports and 1 Win7 export
svn path=/branches/cmake-bringup/; revision=49652
- Convert dispatch.S to new asm syntax
- Add exports for RtlAddVectoredContinueHandler and RtlRemoveVectoredContinueHandler
svn path=/branches/cmake-bringup/; revision=49651
Add a new macro set_export_spec(), that replaces the use of spec2def + adding the generated def file to the sources
Remove EXTERNAL_OBJECT TRUE property for the generated def files.
svn path=/branches/cmake-bringup/; revision=49629
Add a workaround for __readcr4, MSVC suffers an internal error when using the intrinsic and optimization is on.
svn path=/branches/cmake-bringup/; revision=49628
- use correct link script file for freeldr and setupldr
- use correct definition file for advapi32
- don't specify name to add_bootcd_target if not needed
- portcls is not a kernel mode driver
- Disable stcall fixup, except for mesa32 which requires it
- remove an useless file from minihal, add a forgotten one to freeldr
svn path=/branches/cmake-bringup/; revision=49615
Add a new file realmode.S for MSVC currently only containing some stubs to resolve references. The 16 bit code needs some work to be compilable with ML.
Delete some amd64 files that only duplicated x86 16 bit code. Remove unused _alloca.S. Improve freeldr rules, add some experimental linking flags for MSVC, it links with a few warnings.
svn path=/branches/cmake-bringup/; revision=49596
Convert fathelp.S into ML compatible code. The result is identical to the old version when compiled with gcc, but slightly different when compiled with ml. Especially the code starts at position 6 instead of 0, with a one byte text section at the top, this needs more investigation. also there is a large reg move, but that doesn't hurt. Finally the padding at the end doesn't work at all.
svn path=/branches/cmake-bringup/; revision=49588
- Use macro __ASM__ instead of ASM
- guard #pragma once against inclusion in asm files
- Use the HEX() macro for constants for the asm files
svn path=/branches/cmake-bringup/; revision=49583
- move stlport headers to the c++ directory
- add a specific _reactos.h configuration file for stlport
- add cerrno to c++ headers
- __declspec(nothrow) must be placed before calling convention with msvc. GCC doesn't care.
We now use our headers with both msvc and gcc.
svn path=/branches/cmake-bringup/; revision=49565
add genincdata, a module that will be compiled as a dll to provide the data for geninc to be converted into an .inc file.
svn path=/branches/cmake-bringup/; revision=49561
Create importlibs from spec files for kernel32, advapi32, gdi32, user32 and msvcrt
calc now compiles with MSVC and works (slightly buggy)
svn path=/branches/cmake-bringup/; revision=49554
- start using our own c++ headers and forward stlport ones to them in msvc build.
- fix fpecode declaration for MSVC.
[CMAKE]
- cardlib is a cpp library.
Now stlport compiles with msvc.
svn path=/branches/cmake-bringup/; revision=49552
- Move spec2def into comppiler specific files, use spec2pdef tool on MSVC builds
- Add a ridiculously complex macro to create the importlibs for MSVC. (It was hard for me to figure this out, so be it for you ;-))
svn path=/branches/cmake-bringup/; revision=49545
Add a hack to fix compilation of irot.idl with MSVC. If someone knows how to correctly fix this, please let me know.
svn path=/branches/cmake-bringup/; revision=49538
- Add #pragma function to a number of intrisics that we implement to avoid a compiler error of MSVC
- Add a workaround to prevent some functions from being inlined
- Move sqrtf out of i386 directory
- Convert a number of inline assembly functions to raw assembly
svn path=/branches/cmake-bringup/; revision=49534
- add new macro add_linkerflag to avoid code duplication
- add new macro set_pdef_file, replacing the use of pdef2def, which is neccessary to resolve issues with MSVC. cmake doesn't handle def files as source files very well, when they are not in the current source directory.
svn path=/branches/cmake-bringup/; revision=49524
- Add a preprocessed definition file for acledit.
- Add the pdef2def macro.
- Include wpp into build, widl links to it.
- Fix libmpg123.
svn path=/branches/cmake-bringup/; revision=49509
#if 0 _matherr(). We already have it in our crt and it's exported from msvcrt. MSVC doesn't like when it's there twice.
svn path=/branches/cmake-bringup/; revision=49506
- Explicitly disable auto imports, and fix the modules that were missed out by the previous related commit.
- Dedicated to Usurp.
svn path=/branches/cmake-bringup/; revision=49497
MSVC has no sys/time.h
[DDK]
Directly include <wingdi.h>
[UNICODE]
Fix build on msvc
All tools now build with msvc.
svn path=/branches/cmake-bringup/; revision=49461
allow to specify build tools directory with -DREACTOS_BUILD_TOOLS_DIR=Y:\you\path
So now you're able to use multiple compilers with one tree without too much worry.
svn path=/branches/cmake-bringup/; revision=49459
- fix add_importlib_target
- fix win32dll entrypoint (we wouldn't need to specify one, if the one in mingw lib had enough underscores)
svn path=/branches/cmake-bringup/; revision=49453
- Make all x86assembly ML compatible
- Remove memcpy, it was duplicated from memmove, copy the label instead
- Guard some code against compilation on msvc, as these functions are intrinsics on MSVC and cannot be implemented
- Fix some x64 assembly (don't modify stack below rsp)
svn path=/branches/cmake-bringup/; revision=49421
Make systimer.S ML compatible. Replace most hardcoded numerical values with proper symbolic constants.
svn path=/branches/cmake-bringup/; revision=49408
When PreLoad.cmake is run, the CMAKE_CURRENT_SOURCE_DIR points to the directory, where the cmake command was executed (bug?). To get the root folder of reactos, use CMAKE_CURRENT_LIST_FILE, which points to PreLoad.cmake in the root folder.
svn path=/branches/cmake-bringup/; revision=49405
- cmake handles gracefully def files if you tell him to do ;-)
- partial revert of an (accidental?) kernel32 change
svn path=/branches/cmake-bringup/; revision=49380
Include English resource file first, escape certain characters for MSVC compatibility.
"Should be sent to wine"(tm)
svn path=/branches/cmake-bringup/; revision=49379
Implement add_importlib_target macro to build importlibs for MSVC. We can't use MSVC's libraries, as they are incomplete.
gcc version does nothing.
svn path=/branches/cmake-bringup/; revision=49377
- "cmake gracefully handles def files as source files" You mean it gracefully ignores them, right ?
- Fix the kernel and class2 exports. More to come.
- Fix imaadp32.acm
svn path=/branches/cmake-bringup/; revision=49367
- rewrite spec2def macro because
1. cmake gracefully handles def files as source files
2. cmake gracefully handles autogenerated files as source
3. it did not take into account the fact that all shared libraries haven't the .dll extension
svn path=/branches/cmake-bringup/; revision=49358
- improve reactos.dff generation
- rename reactos.dff reactos.dff.in to mark this is an input file.
This is THE file you'd have to edit for slipstreaming, adding binary files, whatever you usually like to do with reactos.dff
svn path=/branches/cmake-bringup/; revision=49349
- add macros add_cab and add_cab_target to add files to reactos.dff
- add bootcd target
- make use of these macros to create bootcd and livecd
Bootcd works and install things, but 2nd stage fails booting. We need debug symbols!!!
svn path=/branches/cmake-bringup/; revision=49347
- fix a whole bunch of function prototype. (how does t even work in trunk?)
[CMAKE]
- add classpnp.sys to build
svn path=/branches/cmake-bringup/; revision=49345
Add each win32dll module to livecd and remove global temporary add-on from gcc.cmake.
Fix pch use in fmifs.
Enable remaining dlls in dll/win32/CMakeLists.txt.
svn path=/branches/cmake-bringup/; revision=49230
Add win32dll modules to livecd for now.
For better flexibility, this will be needed in each module file.
svn path=/branches/cmake-bringup/; revision=49221
Add mkhive to build.
Add livecd target.
Make filename optional in add_mini_target, using the target one as default.
Move MACRO_IDL_FILES to CMakeMacros.cmake.
Add rpcproxy macro.
Build interface definitions headers for epm and irot.
svn path=/branches/cmake-bringup/; revision=49218
- include stlport directory in set_cpp macro
- move set_cpp macro where it belongs : it's generic enough to go to CMakeMacros.cmake
svn path=/branches/cmake-bringup/; revision=49148
- add kernelmodedriver module type
- add set_cpp macro, which takes care of all c++ stuff for you
- hackfix stlport headers for c++ kernel mode modules.
- complain about explorer
- solitaire, telnet and uniata now build happily against stlport
svn path=/branches/cmake-bringup/; revision=49141
- add Windows.cmake, copied from cmake, but with command line file disabled for debugging puposes
- for now use /MTd (static crt) for C++ apps compiled with MSVC
svn path=/branches/cmake-bringup/; revision=49088
Add minicd target.
This partial bootcd target produces a minicd iso,
able to start the install process to the cabinet installation page.
svn path=/branches/cmake-bringup/; revision=49072
- More work on the dlls.
- libgcc is already a standard C lib, don't relink to it for executables.
- ICU isn't needed for now, disable it.
svn path=/branches/cmake-bringup/; revision=49059
- link executables against gcc and mingw_common
- remove link dependencies from mingw_common
This puts C executables into a good shape again.
svn path=/branches/cmake-bringup/; revision=49054
- Use stlport specific includes only for c++ source files
- remove linking to ntdll, kernel32 and oldnames for mingw_common
svn path=/branches/cmake-bringup/; revision=49050
- add gcc dependencies to executables, not to mingw_(w)main
[STLPORT]
- move features definitions where it belongs.
[SDK]
- errno _must_ be defined in errno.h, nowhere else, or stlport headers choke.
svn path=/branches/cmake-bringup/; revision=49049
Jerome Gardou:
- Add STLport 5.2.1 (yes, STLport, the c++ stl implementation) to build. For now, it works only in user mode.
- Link some c++ executables to it
- sol.exe : one step towards a complete and modern OS.
- Dedicated to Amine for his patience and his help.
- Might Break Things! (tm)
svn path=/branches/cmake-bringup/; revision=49046
- Add almost all base applications (some are included but disabled too). This is the work of a human being, so no bots BS, no rbuildism, every single module was tested, built with the new build system mentality (no defines and linked libs redundancy, use of the the available macros, the linked libs are carefully selected, the dependencies are correctly tracked... etc)
- Brought to you by Alexey Komarov, with my humble review.
svn path=/branches/cmake-bringup/; revision=48973
- Improve the unicode situation.
- Apply a consistent formatting.
- Improve the unicode modules we have so far.
- Get rid of some rbuildism.
- Overall improvements.
svn path=/branches/cmake-bringup/; revision=48964
- Make static libs out of the idl files.
- Improve wshirda, wshtcpip and wsock32.
- Include bmfd into build.
svn path=/branches/cmake-bringup/; revision=48956
Add a set_rc_compiler macro,
its a workaround for lack of mingw RC support in cmake.
Enable winemine build.
Improve user32 CMakeLists.txt.
Thanks to AmineKhaldi for initial support of resources in user32.
svn path=/branches/cmake-bringup/; revision=48906
- Introduce proper set_entrypoint and set_subsystem macros for both mingw and msvc.
- Fix the subsystem for console apps.
svn path=/branches/cmake-bringup/; revision=48893
- Don't call KeSetEvent while holding a spin lock (thanks to lassy for pointing it out)
- Don't advance to the next list entry because we are using ExInterlockedRemoveHeadList
- A couple more small code improvements
svn path=/branches/tcp-rewrite-branch/; revision=48841
- Rewrite most of the lower TCP layer code to work with lwIP
[LWIP]
- Write all of the implementation-specific code and headers
[TCPIP]
- Add assertions in the object reference functions
- Remove the completion queue hack
This is still isn't totally finished yet. There are still two main issues that need to be overcome: listen/accept don't work at all and the performance is horrible.
I hope that once I tune some of the options, I can get it to perform reasonably well. If any curious testers want to try this out, be my guest but realize that not
a lot of apps will work yet. Suggestions are welcome! :)
svn path=/branches/tcp-rewrite-branch/; revision=48840
- Remove a superfluous slash.
- Don't link to any standard libraries, we explicitly link to what's needed.
svn path=/branches/cmake-bringup/; revision=48837
__MINGW_ATTRIB_DEPRECATED aka __declspec(deprecated) must be put before the function declaration for MSVC.
svn path=/branches/cmake-bringup/; revision=48828
- Improved MSVC toolchain support, now working with WDK, as well
(Thanks to Amine for solving the problems)
- Add compiler specific macro set_subsystem
- add nativecui module type
svn path=/branches/cmake-bringup/; revision=48824
- Introduce MSVC toolchain support. Just use -DCMAKE_TOOLCHAIN_FILE=toolchain-msvc.cmake for the toolchain.
- Sync setjmp.h from trunk.
- Improve mingw targets (mingw_common, mingw_wmain, oldnames... etc) a bit to make them compile and link with msvc.
- Dedicated to Timo ;)
svn path=/branches/cmake-bringup/; revision=48809
Generalize CMAKE_C_CREATE_SHARED_LIBRARY and remove per project instances.
Thanks to Amine for regexxer tool discovery ;)
svn path=/branches/cmake-bringup/; revision=48804
- Generalize CMAKE_C_LINK_EXECUTABLE and CMAKE_EXE_LINKER_FLAGS, and use additional flags per module.
- Fix usetup: make it depend on psdk and buildno_header, and also fix the entry point (_NtProcessStartup -> _NtProcessStartup@4)
- Make ext2lib depend on psdk.
svn path=/branches/cmake-bringup/; revision=48802
- Introduce support for the amd64 toolchain. Using it is as simple as adding -DARCH=amd64 to the cmake line (next to the toolchain file define).
- Dedicated to Sylvain (Usurp) ;)
svn path=/branches/cmake-bringup/; revision=48763
- Use fathelp.S from r43775 (amd64 branch by timo)
[CMAKE]
- Improve hal, ntoskrnl and freeldr/setupldr build.
svn path=/branches/cmake-bringup/; revision=48688
- Convert fat12/16 boot sector helper code to gas syntax. Brought to you by the Arty.
[CMAKE]
- Add freeldr and setupldr to build.
svn path=/branches/cmake-bringup/; revision=48687
- Manually add source file (as suggested by Colin).
- I've filed a feature request to have the same effect when still using GLOB/GLOB_RECURSE : http://public.kitware.com/Bug/view.php?id=11156
svn path=/branches/cmake-bringup/; revision=48592
- Now that we have c++ flags set globally, no need to mention them on the respective targets.
- debug.h : leftover from the kernel32 sync.
svn path=/branches/cmake-bringup/; revision=48543
- I forgot to link to the wine lib instead of its compiled lib (I have it locally to avoid recompiling it).
svn path=/branches/cmake-bringup/; revision=48540
- Mark DbgUiWaitStateChange as NTSYSAPI.
[CMAKE]
- Don't create an import library for ntdll, we'll ship with one.
- Temporarily add the ntdll import library (until RosBE ships with one).
- Make the definition file MSVC compatible.
- Include every object file in librtl when linking.
- NTDLL now compiles successfully with the MSVC compatible definition file.
svn path=/branches/cmake-bringup/; revision=48529
- Automatically choose the image base for DLLs.
- Improve RC compiler invocation.
- Add and apply the global linker script.
svn path=/branches/cmake-bringup/; revision=48528
- Move generated precompiled headers to binary folders.
- Enable auto image base for ntdll.
- Update the location of generated PCH files for rtl, ntdll and crt.
svn path=/branches/cmake-bringup/; revision=48459
[CMAKE]: Set global linker flags we need for all libraries.
[CMAKE]: Set specific NTDLL linker flag.
Thanks to Amine for some of his help.
svn path=/branches/cmake-bringup/; revision=48457
- We won't be using mkdir to create folders since cmake has a command for it.
- Split nci generated files into separate targets, so that modules can depend on each one of them separately if needed.
- Add win32ksys to build.
svn path=/branches/cmake-bringup/; revision=48329
- Rename ncitool to nci and mark it as an exported native tool.
- Add nci generated intermediate files to build.
svn path=/branches/cmake-bringup/; revision=48325
- Separate the resources creation into several targets. With this we're able to set proper dependencies on them selectively, and not as a whole.
- Make chew and csq, as a result, depend only on bugcodes.
svn path=/branches/cmake-bringup/; revision=48297
- Rename MACRO_ADD_INTERFACES to MACRO_ADD_PSDK_INTERFACES, HEADERS to PSDK_HEADERS.
- Change the psdk target name from build_headers to psdk.
svn path=/branches/cmake-bringup/; revision=48294
[CMAKE]: Add a makelist to build the mc->headers as one target.
[CMAKE]: Fix the IDL rule to build all IDLs as one target.
[CMAKE]: Build a whole lot more libraries.
svn path=/branches/cmake-bringup/; revision=48267
[CMAKE]: Next steps should be to bring all the compiler settings/defines that aren't yet being used, to add per-architecture file tracking, and to enable precompiled header support.
svn path=/branches/cmake-bringup/; revision=48262
[CMAKE]: Add a mignw32 toolchain file, right now it's pretty hacked for my system only, but it works.
svn path=/branches/cmake-bringup/; revision=48256
IDS_STRING_LICENSE "Kalkulator ReactOS jest darmowym programem, publikowanym zgodnie z licencj<63> GNU GPL.\r\n\r\nMo<4D>esz otrzyma<6D> kopi<70> licencji GNU GPL tutaj:\r\nhttp://www.gnu.org/licenses/gpl.html\r\n\r\nT<6E>umaczenie licencji GNU GPL znajdziesz tutaj:\r\nhttp://www.gnu.org/licenses/translations.html"
IDS_LICENSE "This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.\r\n\r\nThis program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.\r\n\r\nYou should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA."
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
