Fix codegen on x86-64
Part 1 of the changes being worked on in #3451
Makes x86emitter emit the x86-64 machine code you would expect it to
Also adds some unit tests to verify that things are working
* macOS compile
* Fix memprotect error on macOS
* Fix semaphore wait + thread cancel on macOS
* Fix timedlock timeout calculation
* spu2-x macOS
* onepad macOS support
* Add MacOS game controller db
* Disable onepad_legacy on macOS
* Fix spu2-x configuration crashes
* Make recompiler work on 32-bit macOS
* Use dylib extension for plugins on macOS
* Make app bundle on macOS
* Use git info for Info.plist version
* Change the minimum cmake version to a range. Uses the new policies of the newer versions if you are on them, taking care of CMP0054.
* Grab a newer version of FindHarfBuzz.cmake from Webkit that fixes the cmake warning the earlier one was giving.
* Add in -ftime-trace as an option on build.sh.
* Precompile PrecompiledHeader.h if you are using cmake 3.16+ and GCC. (Currently disabled on clang due to compilation issues.)
* Check if target_precompile_headers exists rather than by version.
Co-authored-by: scribam <scribam@users.noreply.github.com>
* spu2-x: Make portaudio an optional dependency.
v2: Fix crash in the audio settings with SDL1 and without portaudio.
* cmake: Add PORTAUDIO_API to disable portaudio support.
The warnings issue should be resolved now thanks to arcum.
Fixes compiling issues on some systems/distros using gcc 8.2+
Idea by turtleli.
Also add -mfxsr flag suggested by Gregory.
gcc complained about "Wno-packed-not-aligned" and "Wno-class-memaccess"
flags so I won't bother with it, the warnings can be fixed properly
but for now I'll just revert the previous commit.
A couple of users reported compilation issues using gcc 8.2+ on some
machines/distros on github and discord and adding the -mxsave flag
suggested by turtleli seems to fix the issue.
Fixes#2669
in the top-level source directory. The build folder should NOT be
transferred between computers when PGO is used, though I don't
see why anyone would be doing so anyway.
Also adds support for PGO and LTO to the build.sh script.
Warning can be reenabled on GCC
A warning isn't fixed as potentially the code is wrong
../pcsx2/gui/MemoryCardFolder.cpp: In member function ‘void FolderMemoryCard::FlushFileEntries(u32, u32, const wxString&, MemoryCardFileMetadataReference*)’:
../pcsx2/gui/MemoryCardFolder.cpp:1027:10: warning: unused variable ‘filenameCleaned’ [-Wunused-variable]
bool filenameCleaned = FileAccessHelper::CleanMemcardFilename( cleanName );
Add 3 new warnings (first one must be fixed I think)
pcsx2/gui/MessageBoxes.cpp: In copy constructor ‘BaseMessageBoxEvent::BaseMessageBoxEvent(const BaseMessageBoxEvent&)’:
pcsx2/gui/MessageBoxes.cpp:62:1: warning: base class ‘class pxActionEvent’ should be explicitly initialized in the copy constructor [-Wextra]
BaseMessageBoxEvent::BaseMessageBoxEvent( const BaseMessageBoxEvent& event )
plugins/GSdx/GSPng.cpp: In function ‘bool GSPng::SaveFile(const string&, GSPng::Format, uint8*, uint8*, int, int, int, int, bool, bool)’:
/home/gregory/playstation/emulateur/pcsx2_merge/plugins/GSdx/GSPng.cpp:64:14: warning: variable ‘success’ might be clobbered by ‘longjmp’ or ‘vfork’ [-Wclobbered]
bool success = false;
^
plugins/GSdx/GSPng.cpp:44:58: warning: argument ‘image’ might be clobbered by ‘longjmp’ or ‘vfork’ [-Wclobbered]
bool SaveFile(const string& file, Format fmt, uint8* image, uint8* row,
For older distribution you can still use SDL1.2
./build.sh ... --sdl12 ...
or
cmake ... -DSDL2_API=FALSE ...
Note: there is a hard dependency between WxWidget and SDL. If Wx is linked against
SDL1.2, you must use SDL1.2. Crashes are expected otherwise.
Fixes a cmake detection problem when compiling with clang (3.7.0 in my
case). clang outputs the error "Invalid argument '-std=c++11' not allowed
with 'C/ObjC'", which causes lzma, Xext and other things to be
considered "not found".
Helpful for dropping dependencies package dependencies.
- Git only needed if there is a .git folder
- bzip2 only needed by CDVDiso
- CG/GLEW was for zerogs and ZZogl
- jpeg was for ZZogl
.
I also added that EGL_API is also for GSdx in option comment.
Debian has a goal to make reproducible builds therefore make it an
option instead of distro specific.
.
I added an "OR openSusE" to not "break" the old openSUSE behavior but ideally
they should just use -DDISABLE_BUILD_DATE=TRUE instead.
.
The old -DopenSUSE is not used for anything else so I removed it.
The goal is to reduce the burden for new architecture port.
Patch is mostly inspired from 3kinox initial patch. The diff are
*/ used ifdef instead of raw removal
*/ gui don't rely on UseMicroVU* option
*/ completely remove sVU_micro.* file
User may forget or not know about -DUSE_CLANG=TRUE. It could probably
be always autodetected instead of requiring user input.
.
Trivial wording changes to compiler_version.
Tell the user to use CMAKE_TOOLCHAIN_FILE upon failure.
.
Cleanup detectOperatingSystem. Should be the same but adding
GNU and kFreeBSD from Debian and filtering out pre OS X Apple.
.
libaio is linux only. Also check for the correct header since
aio.h is the POSIX one. Both are in /usr/include/ anyway.
Only build core if (Linux AND NOT AIO_FOUND) == false.
.
Use Unix for GTK and X11. Macs has gtk-quartzs and xquartz or native X11.
*BSD, linux, etc should have both.
The obtained binaries before and after this commit are identical (sha1sum)
when compiled in Debian/Ubuntu/Fedora/ArchLinux.
.
The linker will always pick the 32bit libraries the only thing this does is
make sure we have all the 32bit dependencies installed. Basically we avoid
detecting the 64bit libraries and telling the users the 32bit libraries were
found. We always link with 32bit libraries therefore this avoids having to
wait 5-10min to just be told -lXXX is missing.
.
The only thing really needed are
set(CMAKE_LIBRARY_ARCHITECTURE "../lib32")
set(CMAKE_LIBRARY_ARCHITECTURE ".")
.
which basically ensures we don't pick 64bit headers since
CMAKE_LIBRARY_ARCHITECTURE always gets tested first and for some reason
FindGTK2 test searches lib64 first then lib32/lib. These values are hardcoded.
Right now these arch specific headers are not used but can't say this will
always be true.
.
FIND_LIBRARY_USE_LIB64_PATHS is not needed for native builds and it's covered
by CMAKE_SYSTEM_IGNORE_PATH.
.
NOTE:
We filter out lib32 because multilib is not compatible with multiarch.
- Fedora only needs --arch if cross compiling.
+ It's only used to select libdir so i386=i686=i986
- Messages are nice to debug build logs.
.
Everything seems to work unless some other distro broke. Other
distros need to add CMAKE_TOOLCHAIN_FILE or --cross-multilib if
they cross compiled amd64 -> i686.
http://www.vtk.org/Wiki/CMake_Cross_Compilinghttp://www.cmake.org/cmake/help/v3.0/manual/cmake-toolchains.7.html
.
The official way that cmake does cross compiling is via the use of a
CMAKE_TOOLCHAIN_FILE. This has to be given by the user and can't be
included from within a Cmake file since setting up the toolchain has
to be the first thing that happens.
.
After the file is given and validated cmake behaves nicely and all the
workarounds and hacks are not really needed anymore.
.
The consequence of this change is that without this file cmake will
try to build for the HOST architecture as expected and with the file
it will build for the TARGET architecture of the given toolchain. Due to
this remove 64BIT_BUILD_DONT_WORK and just ERROR out if the user tries
in the same way as before.
Currently crosscompiling was broken since we detected the cpu of the host
not of the target. Building arm -> i386, i386->amd64, etc resulted in
interesting stuff.
.
CMAKE_HOST_SYSTEM_PROCESSOR and CMAKE_SYSTEM_PROCESSOR should have done this
but they are incredibly broken and unreliable. I use
CMAKE_HOST_SYSTEM_PROCESSOR just to retain the format of the old message
and is not used for critical stuff so it does not matter.
.
This code also allows scalability so that if one day in the very distant
and unlikely future the x86/x86_64 specific code gets made portable
by replacing asm/MMX/SSE*/AVX/etc code with generic/portable code then
it also would work on all architectures.
.
For *BSD/OS X it probably would need to use POSIX AIO instead of linux
specific libaio but that is a different issue.
.
Error out the x86_64 builds.
Also the 64BIT_BUILD_DONT_WORK option became obsolete more info in following
commit.
Namely GTK3_API and 64BIT_BUILD_DONT_WORK
The former was barely tested and doesn't compile with standard wx
The latter doesn't work at all
Note: I was very close to disable SDL2 too. If wxWidget was built with
SDL1 support it will crash
Description: Building with SDL2_API=TRUE is incomplete
SDL_FOUND does not imply SDL2_FOUND
Use check_libs for detection
Only include ${SDL_INCLUDE_DIR} when needed
Use SDL2_LIBRARIES
Author: Miguel A. Colón Vélez
Gregory: add back SDL_BUILDING_LIBRARY
+ Wx3.0 fixes various issue on linux (better sizing of box)
+ Debian doesn't provide wx2.8 anymore. Ubuntu will probably follow soon (close issue #342)
If you want to use wx2.8:
* Either you remove wx3.0 from your system
* Either you use the cmake option -DWX28_API=TRUE
Please don't hesitate to report any regresions.
I need to check carefully the consequence of ABI change. So far wx is very unhappy!
Fatal Error: Mismatch between the program and library build versions detected.
The library used 2.8 (no debug,Unicode,compiler with C++ ABI 1002,wx containers,compatible with 2.6),
and your program used 2.8 (no debug,Unicode,compiler with C++ ABI 1006,wx containers,compatible with 2.6).
* disable unused value warning (don't like the syntax 0&&)
* remove -Wno-ignored-attributes (c option). It replaces gcc warning
by an unsupported option warning...
GCC doesn't support pentium4 as a arch target for x86_64. It complains that the architecture doesn't support x86_64
Which to be fair the first few models didn't support 64bit.
Just remove the architecture setting since there isn't a need for it.
Move build configuration from CMakeLists to cmake/BuildParameters.cmake where it belongs
Use option syntax for on/off option
Complete a bit previous commit to force all 64bits path detections (likely done by cmake)
By default the cmake build will still cross compile a 32bit binary and spout a message about it if not enabled.
This doesn't fix the 64bit build issues, just makes it easier for someone to test 64bit builds in the future.
Look towards a bright future instead of a dark and gloomy past
lightningterror