This commit adds support for compiling Dolphin for ARM on MacOS so that it can
run natively on the M1 processors without running through Rosseta2 emulation
providing a 30-50% performance speedup and less hitches from Rosseta2.
It consists of several key changes:
- Adding support for W^X allocation(MAP_JIT) for the ARM JIT
- Adding the machine context and config info to identify the M1 processor
- Additions to the build system and docs to support building universal binaries
- Adding code signing entitlements to access the MAP_JIT functionality
- Updating the MoltenVK libvulkan.dylib to a newer version with M1 support
LTO is supposed to be enabled by default for VS Release builds
according to the VS prop files but a build log from JMC reveals
that /GL and /LTCG are not actually passed to cl.exe/link.exe
for some reason...
LTO also leads to *extremely* and unacceptably slow build times
when using link.exe, so let's disable it by default to actually
match the project files.
This adds a CMake option (DOLPHIN_DEFAULT_UPDATE_TRACK) to allow
configuring SCM_UPDATE_TRACK_STR. This is needed to enable auto-updates
in Windows CMake builds by default.
Xcode is configured by default to turn on the strict aliasing optimization
through a separate project setting. This would cause the compiler to be called
with both -fno-strict-aliasing and -fstrict-aliasing. This change turns off
that project setting causing only the -fno-strict-aliasing flag to be provided
to the compiler.
Modify the CMakeLists.txt so that it doesn't try to use a shared zstd library
that doesn't have header files. This was a support issue on Macs because
homebrew was installing headerless zstd libraries with Qt.
Add a check to define FMT_EXCEPTIONS = 0 when compiling on MacOS.
Replaces -fno-objc-exceptions, which also disabled fmt exceptions but
generated warnings when compiling c++ files.
While the gdbstub is somewhat broken at the moment, building it by
default should keep us from breaking the build (which has happened
several times in the past: #2614, #5904, #8262).
It isn't very intrusive and it does nothing unless it is enabled
in the config file.
CMake's CMAKE_SYSTEM_PROCESSOR call returns "arm64" on Apple Silicon. This patch makes sure that arm64 and aarch64 are treated as the same architecture.
Having extra warnings enabled for everything including external
libraries produces an overwhelming amount of warnings in code that
isn't even part of our codebase.
Move the various warning flags to Source/CMakeLists.txt to get rid
of those useless warnings.
Note that the Source CMakeLists.txt is already where the MSVC warnings
are defined, so this commit improves consistency as well.
While manually capturing constexpr variables used in lambda
expressions does work, it's really easy to forget doing so since
we don't have a Windows CMake builder and the workaround isn't
necessary anywhere else. Fortunately, MSVC has a flag that fixes
the constexpr capture behavior, so let's use that instead.
Disable by default, to be used by distributions who care to try and
prefer system-wide libraries when available. It makes sense for us to
keep using vendored libs by default when possible to make it easier for
users to compile, but we should provide appropriate tools for distro to
figure out which dependencies they can share with the rest of the
system.
We must not provide the /Externals directory as global include directory.
Here, this yield a crash because of external minizip header and system library mismatch.
Soundtouch itself recormends to include it with <SoundTouch.h> and -I/usr/include/soundtouch, so this should fit better.
Minizip 1.x is the old broken version, while 2.x is the newer fork we
use. Many distros ship the old version which is causing stack smashing
in RedumpVerifier, so we'll only use the shared lib if it's the newer
version.
Removed conditional use of std::mutex instead of std::shared_mutex on MacOS.
Because MacOS < 10.12 did not support std::shared_mutex, a previous commit
naïvely substituted std::mutex, which does not have the same behavior.
Reverses PR #8273, which substitues std::mutex for std::shared_mutex on
macOS, and results in several bugs that seem to only affect MacOS
- https://bugs.dolphin-emu.org/issues/11919
- https://bugs.dolphin-emu.org/issues/11842
- https://bugs.dolphin-emu.org/issues/11845
This change eliminates conditional code for MacOS in the core configuration
layer code and enables the use of modern language features that are more
secure and thread-safe.
This allows the user to set this to LEGACY. This can be useful for
people under Linux using optirun/primusrun with "old" nvidia drivers
that don't support GLVND properly.
Allows us to migrate off of printf specifiers and have more type-safe
formatting facilities. It also allows for custom type support as well.
fmt is also on track to have part of it standardized within C++2a, so
this will also lessen the transitional work necessary later on by
allowing new code to use it.
This simply adds the library but doesn't do anything with it yet.
MemoryWatcher only works on Linux and affects emulation determinism due
to scheduling additional events, which causes NetPlay to desync.
Considering that this interface is a rather specialized use case, the
communication with it is kinda crappy *and* it's affecting emulation, I
think it's best to just axe it and come up with a better implementation
of the functionality.
The documentation for setting cache entries dicates that the type must
either be BOOL, FILEPATH, PATH, STRING, or INTERNAL (with those exact
casings). Also, given we properly case it in other places, this is just
consistent.
The general convention for CMake is to use lowercase for commands, and
given we also follow that convention through most CMake files, this just
makes it more consistent.
Thin archives contain pathnames pointing to the object files instead of
full copies of the object files. This significantly reduces the disk
usage when building Dolphin.
Size of *.a files: (gcc-8.1.0, Linux amd64)
- Before: 83,876 KB
- After: 1,876 KB
- Diff: -82,000 KB
The resulting binaries are the same as before.
A similar change was implemented in the Linux kernel v4.8:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=a5967db9af51a84f5e181600954714a9e4c69f1f
We have to decide if we want to link to
libGL.so -- legacy
libOpenGL.so -- GLVND
As we use xxxGetProcAddress, we don't care about feature
levels of the linked library, so we're very fine with the
new way.
Allows us to bring includes and relevant libraries into scope by explicitly declaring linkage against the target
as opposed to using a variable. Also removes the dumping of OProfile includes into the top-level directory.
Skyler Saleh