More forwards to LocalSpl and LocalMon. At sometime will be merged together.
Bug fixes.
Printer Driver code is a wine hack. (WIP)
Added information for shell tray icon notifications.
Sync wine WinSpool driver tests. Unplugged from build.
RtlGetNtProductType comes into two variants: one in user-mode that is exported for use from NTDLL layer and the kernel-mode that is used exclusively by the NT kernel. The kernel-mode variant of the function is not exported.
..\ntoskrnl\vdm\vdmmain.c(91,34): warning: expression does not compute the number of elements in this array; element type is 'unsigned short', not 'char' [-Wsizeof-array-div]
UNICODE_STRING PhysMemName = RTL_CONSTANT_STRING(L"\\Device\\PhysicalMemory");
^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~
C:\ROS\reactos\output-clang\sdk\include\psdk\ntdef.h(1542,15): note: expanded from macro 'RTL_CONSTANT_STRING'
sizeof(s) / sizeof(_RTL_CONSTANT_STRING_type_check(s)), \
~ ^
..\ntoskrnl\vdm\vdmmain.c(91,34): note: place parentheses around the 'sizeof (_RTL_CONSTANT_STRING_type_check(L"\\Device\\PhysicalMemory"))' expression to silence this warning
C:\ROS\reactos\output-clang\sdk\include\psdk\ntdef.h(1542,15): note: expanded from macro 'RTL_CONSTANT_STRING'
sizeof(s) / sizeof(_RTL_CONSTANT_STRING_type_check(s)), \
^
1 warning generated.
* Add the remaining functions & annotate.
* Add the RTLP_RANGE_LIST_ENTRY structure, which is present in symbols
Note: range list declarations should really be !NTOS_MODE_USER, since ntdll
doesn't export these functions; however, our RTL itself is built with
NTOS_MODE_USER, so that's not an option at this point.
These has been first added to DDK 10.0.19041.0 and made backwards-compatible
with all Windows versions since Windows 2000.
Our implementation does not use POOL_ZERO_ALLOCATION flag because it's not
supported by ReactOS yet, and thus erases the memory unconditionally.
Also update the includes.
It will allow to use those declarations globally where they are required (e. g. for SvchostPushServiceGlobals functions in some services).
No functional changes.
Required by vssapi stubs CORE-9511.
Our current vswriter.h has no some needed definitions, but simple adding them in the current header is not an option, since Wine's vsbackup.idl (which is required too) depends on it.
Then it will not need to add ROS-specific diff(s) in Wine's vsbackup.idl.
- Make "ReactOS Server" the default product option again instead of "ReactOS Workstation".
- Write "Service Pack" info onto registry.
- Add ProductOption option to bootcd unattend.inf.
- Delete IDC_PRODUCT_SUITE and IDC_PRODUCT_TYPE controls.
CORE-17028
This version (after .rsrc) works different than the
proper version I used for 0.4.12 (after .reloc).
Inserting after .rsrc is actually not correct, but Thomas believes it can
be used as a temporary trick to avoid random memory corruption upon
relocations of the kernel, caused by ROSBE-154.
I follow his advice, although when judging from practical tests only:
as long as we limit this script to NTOSKRNL like I do for releases
there have no negative consequences been observed in real life yet
even with the proper version of 0.4.12.
Up to now those problems have only been observed when used for drivers
MODULE TYPE sdk/cmake/gcc.cmake as well, like
it was tried for a moment in master 0.4.13-dev-609-g
c4d8e2a6e9
Using for drivers immediately did lead to BSODs like CORE-16183 and therefore was
mitigated in master by total disabling of the scripts for both,
kernel and drivers in
0.4.13-dev-621-g
36e9a6f8dd
To allow installing DVDWritenow without BSOD,
we need the script at least for ntoskrnl!
I committed this patch (after .rsrc) already into 0.4.13RC and 0.4.14RC.
- Add boot video color constants
- Refactor palette initialization
- Move some common stuff in right place
- Get rid of some magic constants and hardcoded values
- Get rid of TopDelta variable (calculated at compile time)
- Update SAL annotations
Addendum to 5f2ca473. CORE-16216 CORE-16219
- Print message on failed memory allocations.
- Functions return error code instead of exit code.
- Convert error code to exit code in the main function only.
- Always free the allocated memory.
- Fix bug from previous commit: Use _tprintf instead _putts to print messages.
This is a partial sync of the CRT library with wcsrtombs_l and _mbstowcs_l functions from WINE. The _wctomb_s_l implementation of WINE which is used by _wctomb_s, _wctomb_l and wctomb brings failed results of the wctomb unit testcase and at the same time it crashes the whole testcase after.
Therefore I will not address the wctomb function for the moment being.
- Move translations to netmsg.dll.
- Keep italian translation as a comment because netmsg.dll does not have an italian translation yet.
- Remove obsolete resource files.
The current _mbtowc_l code directly passes the string count size argument to cbMultiByte argument parameter to MultiByteToWideChar(). As the size of lpMultiByteStr expressed in bytes by cbMultiByte is dependent of the actual given string, the size of lpWideCharStr expressed in characters by cchWideChar is always 1.
The resulting parameters to MultiByteToWideChar() will lead to a failure as lpWideCharStr cannot accommodate the converted string if the size pointed by lpMultiByteStr is larger than the buffer wide string to receive the converted output, a Win32 error code of ERROR_INSUFFICIENT_BUFFER in other words. The fact that _mbtowc_l never fails without WINE Gecko package installed is because the codepage pointed by lc_codepage is CP_ACP and the converted output is directly given to the destination wide string parameter thus never calling MultiByteToWideChar(). Installing WINE Gecko merely unhides this problem.
As _mbtowc_l is imported from WINE and that they have updated their code (which at the same time this fixes the aforementioned problem), the following commit syncs the updated code from WINE.
CORE-16885
Incoming r4 contains the number of dwords to allocate, converting to bytes then return.
This makes ReactOS applications compiled for ARM running on true ARM Windows.
Reference: 2b095beace
Native CRT on ARM & ARM64 Windows doesn't have these functions.
For compatibility, it's mandatory to not have it at all. Otherwise,
ARM executables built for ReactOS do not run on true ARM Windows.
Consider doing full sync with mingw-w64 CRT in future.
This small win32 library provides time-zone utility wrappers around
Win32 functions, that are used by different ReactOS modules such as
timedate.cpl, syssetup.dll, and a possible future 'tzutil' tool.
The code has been extracted from the common code found in both
timedate.cpl and syssetup.dll.
Since binutils can not add this, we re-introduce pefixup.
It searches for the exported symbol '_load_config_used',
and uses that to fill out the LOAD_CONFIG directory in the PE header
DiskGeometryGetPartition() and DiskGeometryGetDetect() are guarded in NTDDI_VERSION for version checking as they differ between Windows XP and Server 2003. In conjunction to that, the PARTITION_INFORMATION_EX structure has "IsServicePartition" as member which is missing in the current NT IOCTL Disk interface header.
Shell change notification has been implemented in #2432. But as @yagoulas said, source code structure is in mess. We improved simplicity and human readability of our source code.
- Move wine/changenotify.c code into changenotify.c and shelldesktop/CChangeNotify.cpp.
- Simplify code and rename the identifiers and add many comments. CORE-13950
Need a way to send WM_DEVICECHANGE from the Service manager application.
Next step is to process it in the Client/Server Run-time Subsystem.
See CORE-16492.
Remove RTC switch from configure.cmd, now MSVC runtime checks are
supposed to be enabled/disabled via CMake define (-DRUNTIME_CHECKS=1)
Prepare for adding more dynamic analysis options
- No need to use a PCH when compiling a 1-file library.
- Add the _NTSYSTEM_ definition.
- Since the arbiter.h file is to be included by the users of this
library, namely only kernel-mode NT components (incl. NTOS), there
is no need to include the standard kernel-mode headers in it.
Include these headers instead in the private implementation of the
library, arbiter.c.
This adds missing features like using events and APCs within IcmpSendEcho2
functions and others.
CORE-10742 CORE-14411
Co-authored-by: Tim Crawford <crawfxrd@gmail.com>
Required by Wine's qcap.dll CORE-16350
[DXSDK] Restore all deleted code in axextend.idl to propely fix compilation
Also regularize some parts of added code and replace IAMStreamControl interface
with Wine's one, because otherwise compilation fails.
Pierre recommended this workaround for 0.4.8rls before.
Avoids "GetVolumeInformation now fails on NFS volume"
This workaround was recurrently applied for all releases
0.4.8, 0.4.9, 0.4.10, 0.4.11, 0.4.12, 0.4.13.
I never got any reply in the regression-ticket and recurrently
applying this over and over again is a waste of time.
So I decided to commit to master today, but will leave
the ticket unresolved, so when a proper fix will arrive in the future,
the existing ticket will remind us to undo this workaround.
Please note that I replaced #if 0 with #if 1
as discussed with Pierre. That's different to the patch in ticket.
which implements the required functionality.
ntdll and ntoskrnl now have a wrapper for this, with SEH.
This protects the function against malformed / bad images,
whilst still being able to use the code in freeldr et al.
Idea from Thomas.
CORE-14857
CORE-16757
After doing investigations on the priority class structure alignment, it's been revealed that in Windows XP and Server 2003 this PROCESS_PRIORITY_CLASS structure is aligned as a 4-bytes of size hence NtQueryInformationProcess() probes the alignment of user mode arguments buffer output and buffer length with requirement of a ULONG.
As PROCESS_PRIORITY_CLASS was initially aligned as a 1-byte size because both BOOLEAN and UCHAR are just unsigned characters, the compiler may not align such structure and gracefully let the default alignment of such structure as is, 1-byte because an unsigned char has a size of 1 byte. Setting an align attribute to this structure fixes the problem of a potential datatype misalignment which caused GetPriorityClass() to not retrieve the process' priority class properly.
This not only makes PAE possible, but also allows to make the definitions and macros in the memory manager header files more human-readable.
CORE-16702
* [SDK][NDK] Make the _MMPTE_HARDWARE structure more compact and more human-readable.
* [SDK][NDK] Add definitions a struct _MMPTE for PAE mode.
NOP_FUNCTION is defined to "(void)0". Calling :
RTL_PAGED_CODE();
in user mode was changed to:
(void)0();
As "0" is not callable, this was leading to a compilation error.
Most of them are unaffected by packing, the ones that are affected have pointers, and packing them will only misalign them, but not making the structures 32 bit compatible.
- Add ARC-emulation support for NEC PC-98 series
- Add global definition for PC-98 port into CMakeLists.txt
- Add floppy verison of freeldr.ini for PC-98 CD boot
* Add missing enums to MSV1_0_PROTOCOL_MESSAGE_TYPE and SECPKG_EXTENDED_INFORMATION_CLASS
* Add missing function entries to LSA_SECPKG_FUNCTION_TABLE and add needed definitions.
* Fix params of PLSA_AP_CALL_PACKAGE
* Fix SpInitializeFn (+ NTAPI)
A non optimized mesh does not prevent rendering as long as we return valid data to the application.
In our case we provided an identity remapping array when no vertices reordering is done.
Avencast demo works perfectly well (using native effects functions).
wine-staging patch by Christian Costa <titan.costa@gmail.com>
This patch fixes vertex processing issue of bug 33770.
The problem comes from the fact that even if the call succeeds,
the game interprets a non null error_messages pointer as an error.
By calling D3DCompile we use a newer version of the compiler which is more
strict and generates the following warning.
- warning X3206: 'dot': implicit truncation of vector type
- warning X3206: implicit truncation of vector type
- warning X3206: 'mul': implicit truncation of vector type
D3DCompileShader does not generate such warnings.
These is confirmed in the DX SDK release note:
New Warning X3206: Implicit Truncation of Vector Type
Beginning in the August 2009 release of the DirectX SDK, the compiler will warn
when an implicit truncation of a vector type occurs.
The warnings cannot be disable so this patch filters out these strings in D3DCompileShader
and reset the error messages pointer if the resulting buffer is empty.
Try 2:
- only filter out lines containing "X3206:" in case d3dcompiler_43 has localization
Try 3:
- use move in place instead of copying the buffer
Try 4:
- filter simplification by Sebastian and remove 'mul' testing left-out in search string
wine-staging patch by Christian Costa <titan.costa@gmail.com>
Fix remaining text issues in Air Strike.
Should fix text placement in Stronghold Kingdoms.
wine-staging patch by Christian Costa <titan.costa@gmail.com>
Changes by Sebastian Lackner <sebastian@fds-team.de>:
* Use pitch value for locked buffer instead of assuming that pitch = width * bytesperpixel
* Avoid one for loop to simplify code
* Ensure that DrawText doesn't dereference a NULL pointer when count != 0.
Changes by Christian Costa <titan.costa@gmail.com>
* Use dedicated variables for text width & height instead of reusing rect.right and rect.bottom
* Remove useless test in pixel conversion
* Remove left over 'partial stub' in fixme
Changes by Sebastian Lackner <sebastian@fds-team.de>:
* Replace code to convert text from ascii to widechar
* Strip terminating NULL chars before drawing text
wine-staging patch by Christian Costa <titan.costa@gmail.com>
Signed-off-by: Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
wine commit id f72204abd6037e59ba1bf500a3a9d0316e65a024 by Paul Gofman <gofmanp@gmail.com>
The issue is spotted in https://bugs.winehq.org/show_bug.cgi?id=46553,
while fixing it alone is reported not to affect the bug.
Signed-off-by: Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
wine commit id b3e91c3408797da72fdd5066e983d8861d9271ea by Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
wine commit id 6fe2728b215fcc22940c556f195fe223c5ac2cff by Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
wine commit id cf992bebb056814a05fe3a82011c4de701c144c4 by Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Michael Stefaniuc <mstefani@winehq.org>
Signed-off-by: Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
wine commit id 7e01ec6018c062d11b29ba2b9a2056d98e8bc406 by Michael Stefaniuc <mstefani@winehq.org>
This requires the pygit2 module
usage : ./winesync.py module <wine-tag> <winestaging-tag>
for instance ./winesync.py d3dx9 wine-4.1 v4.1
This requires to have a wine git checkout and a wine-staging checkout
configuration is done through yaml file named <module>.cfg specifying
the following:
- file mappings
- directory mappings
- latest wine version the module is synced with
it then creates a local branch in the wine checkout, based on the given
tag, and then the staging script is ran on top of it.
Thanks to the mappings defined in the module configuration file, it then
create individual commits in the reactos git checkout with reworked
path.
In case of problem, it stops and lets you amend the latest commit and go
along with the process once this is done. (Makefile.in modified, new or
removed files, patches not cleanly applied)
Staging patches are added into the <module>_staging directory for the
ease of reverting them later. (TODO)
See previous [WINESYNC] commits to have an overview of the result
A non optimized mesh does not prevent rendering as long as we return valid data to the application.
In our case we provided an identity remapping array when no vertices reordering is done.
Avencast demo works perfectly well (using native effects functions).
wine-staging patch by Christian Costa <titan.costa@gmail.com>
This patch fixes vertex processing issue of bug 33770.
The problem comes from the fact that even if the call succeeds,
the game interprets a non null error_messages pointer as an error.
By calling D3DCompile we use a newer version of the compiler which is more
strict and generates the following warning.
- warning X3206: 'dot': implicit truncation of vector type
- warning X3206: implicit truncation of vector type
- warning X3206: 'mul': implicit truncation of vector type
D3DCompileShader does not generate such warnings.
These is confirmed in the DX SDK release note:
New Warning X3206: Implicit Truncation of Vector Type
Beginning in the August 2009 release of the DirectX SDK, the compiler will warn
when an implicit truncation of a vector type occurs.
The warnings cannot be disable so this patch filters out these strings in D3DCompileShader
and reset the error messages pointer if the resulting buffer is empty.
Try 2:
- only filter out lines containing "X3206:" in case d3dcompiler_43 has localization
Try 3:
- use move in place instead of copying the buffer
Try 4:
- filter simplification by Sebastian and remove 'mul' testing left-out in search string
wine-staging patch by Christian Costa <titan.costa@gmail.com>
Fix remaining text issues in Air Strike.
Should fix text placement in Stronghold Kingdoms.
wine-staging patch by Christian Costa <titan.costa@gmail.com>
Changes by Sebastian Lackner <sebastian@fds-team.de>:
* Use pitch value for locked buffer instead of assuming that pitch = width * bytesperpixel
* Avoid one for loop to simplify code
* Ensure that DrawText doesn't dereference a NULL pointer when count != 0.
Changes by Christian Costa <titan.costa@gmail.com>
* Use dedicated variables for text width & height instead of reusing rect.right and rect.bottom
* Remove useless test in pixel conversion
* Remove left over 'partial stub' in fixme
Changes by Sebastian Lackner <sebastian@fds-team.de>:
* Replace code to convert text from ascii to widechar
* Strip terminating NULL chars before drawing text
wine-staging patch by Christian Costa <titan.costa@gmail.com>
Most of those new tests were written by Rico Schüller.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=33456
Signed-off-by: Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
wine commit id 956d7356e0559a07aac68bcaf73a5005e0600fad by David Adam <david.adam.cnrs@gmail.com>
Signed-off-by: Nikolay Sivov <nsivov@codeweavers.com>
Signed-off-by: Vincent Povirk <vincent@codeweavers.com>
Signed-off-by: Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
wine commit id 77c3761ad4e557f9fa311299e77746bf225a575a by Nikolay Sivov <nsivov@codeweavers.com>
Inspired by a patch from David Adam.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=33456
Signed-off-by: Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
wine commit id f54260a789387ab40de3d80291ad9000176dabe1 by Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
wine commit id 65956ae50a136fe8467a0332c7495f36617a16dc by Paul Gofman <gofmanp@gmail.com>
Fixes a regression triggered by commit
949dbbd31f450178c90ea8267097a975b77c3219.
Signed-off-by: Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
wine commit id 092c14b9d86ee1e96e465908cc524ec85988d0ff by Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
wine commit id ce090114340fe6ce9314f543a1389de9bdc6df14 by Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
wine commit id 61e7e6ec1bf5d13d4cafa883ac74fe987c339740 by Paul Gofman <gofmanp@gmail.com>
Fixes a regression triggered by commit 949dbbd31f450178c90ea8267097a975b77c3219.
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=46453
Signed-off-by: Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
wine commit id 4e44d74afb12b40f673e3be0a91cd523013544cd by Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Paul Gofman <gofmanp@gmail.com>
Signed-off-by: Matteo Bruni <mbruni@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
wine commit id 33be8439fc95ca8f005bd3a2b6790c101842c02e by Paul Gofman <gofmanp@gmail.com>
Author: Puetz Kevin A <PuetzKevinA@JohnDeere.com>
atlbase.h: Fix some declarations on win64.
Signed-off-by: Jacek Caban <jacek@codeweavers.com>
Signed-off-by: Alexandre Julliard <julliard@winehq.org>
- Rename parameters according to [MS_PNPR] (no longer available for download).
- Remove unsupported PNP_DEVINST_MOVE and PNP_DEVINST_DISABLE actions.
- Implement most of the PNP_DEVINST_SETUP action.
Reject invalid input filename characters by using shell32!SHLimitInputEdit function and IItemNameLimits interface. Improve SHLimitInputEdit to sanitize paste.
CORE-11701
Surprisingly this also happens to "fix" random "Invalid Opcode" exceptions in XQEMU.
(But I think it's more like a coincidence... --hbelusca)
CORE-16627 CORE-16216
- Add "Product Options" wizard page into ReactOS Setup.
- Implement CSIDL_Type_InMyDocuments CSIDL type.
- If the product type is workstation, then some special folders will be in My Documents.
CORE-13795
The code is mostly unchanged. This includes the following changes:
* Move all wine code to crt/wine to keep it separated from our own code
* Add a minimal winternl.h
* Remove the asm macros from wine/config.h
* Include wine/asm.h where required
* Fix the names of the exported functions (GCC uses thiscall now and no wrappers are used anymore)
The "TypeOffset" thing was just an informative comment to tell that the
data that follows after the IMAGE_BASE_RELOCATION header is an arbitrary
array of WORDs describing packed (Type + Offset)s.
Having the header structure containing that spurious "TypeOffset" was
breaking all the code that was basing on expected size of the
IMAGE_BASE_RELOCATION structure in order to apply relocations (typically
this would mean that the first 2 relocations described by it would not
be applied).
Hopefully this bug only hit the host-tools, and not the OS itself :)
The first part of PC-98 Port - https://reactos.org/wiki/PC-98
- Add FAT12 file system boot sector for NEC PC-98 series.
- Add a new build target for a PC-98 bootable floppy disk.
- Add a new sub-architecture into config.cmake.
- Make the format actually MS-compatible: "ATL" followed by colon,
followed by hexadecimal digits of pointer.
- The MS counterpart of this DLL was delivered with Visual C++ 6.0 and
Windows 98+, so obviously it always was 32-bit and they never had a
64-bit version for it. But we do. So make the size of the m_szAutoName
buffer cross-compatible.
- See previous commit dbddd878 and the discussion in PR #2010.
The autogenerated name has the format:
"ATL:<hexadecimal_digits_of_pointer><NULL-terminator>"
and the number of hex digits in 0xABCD1234 (for 32-bit == 4-byte)
pointers (without the '0x') is 8 == 4*2, and for 64-bit == 8-byte
pointers (e.g. 0xABCDEF0123456789) is 16 == 8*2.
Also convert all sizes and positions of CONSOLE_DRAW to USHORT since
this is the standard type for all console buffer positions & sizes
(minimum value 0, maximum value 0xFFFF == 65535).
The reason is that dlltool orders the exports differently than MSVC builds (MSVC orders the exports by symbol name, rather than by export name), so we rely on sorting in the spec file, which was only respected, when ordinals were put into the def file.
On MSVC builds it is left to the linker to determine the correct order, which helps to get the differences between architectures right (different symbol decoration, difference between order for functions like NtLoadKey vs NtLoadKey2, which results from the stdcall decoration on x86, which is missing on other architectures.
TODO: To correctly handle non-x86 architectures with GCC builds, spec2def would need to reorder the export list based on symbol names, which would work for C functions, by taking the calling convention into account, but would require an extra c++-stdcall calling convention to be added to know the corresponding symbol starts with "?".
