cda5a530e6
:yakkie: The build process for the Zig compiler is more involved than most of the other ports, because the Zig compiler is mostly self-hosting. In order to build it, the zig-bootstrap build system is used, which does the following: 1) Build LLVM for the host OS; 2) Build Zig for the host OS with the SerenityOS target enabled; 3) Build zlib, zstd and LLVM for SerenityOS using `zig cc` as the C/C++ compiler; 4) Build Zig for SerenityOS using the host Zig. A few hacks are required in order to tell `zig cc` and zig about what Serenity's libc looks like in the build process, but other than that it's fairly straightforward. All of the patches that are included with this commit are Zig-upstream ready once the LLVM patches are upstreamed. |
||
---|---|---|
.. | ||
0001-Add-support-for-building-LLVM-on-SerenityOS.patch | ||
0002-Add-triple-for-SerenityOS.patch | ||
0003-Add-support-for-SerenityOS.patch | ||
0004-Default-to-ftls-model-initial-exec-on-SerenityOS.patch | ||
0005-Add-support-for-SerenityOS.patch | ||
0006-Allow-undefined-symbols-on-SerenityOS.patch | ||
0007-Support-building-shared-libLLVM-and-libClang-for-Ser.patch | ||
0008-Add-SerenityOS-to-config.guess.patch | ||
0009-Prevent-the-use-of-POSIX-shm-on-SerenityOS.patch | ||
0010-llvm-Implement-bigint-to-LLVM-int-for-32-bit-compile.patch | ||
0011-Add-SerenityOS-target.patch | ||
0012-Implement-SerenityOS-support-in-std.patch | ||
0013-build-Adjust-build-process-for-SerenityOS.patch | ||
ReadMe.md |
Patches for zig on SerenityOS
0001-Add-support-for-building-LLVM-on-SerenityOS.patch
Add support for building LLVM on SerenityOS
Adds SerenityOS #ifdef
s for platform-specific code.
We stub out wait4, as SerenityOS doesn't support querying a child process's resource usage information.
0002-Add-triple-for-SerenityOS.patch
Add triple for SerenityOS
0003-Add-support-for-SerenityOS.patch
Add support for SerenityOS
Adds support for the $arch-pc-serenity
target to the Clang front end.
This makes the compiler look for libraries and headers in the right
places, and enables some security mitigations like stack-smashing
protection and position-independent code by default.
0004-Default-to-ftls-model-initial-exec-on-SerenityOS.patch
Default to -ftls-model=initial-exec on SerenityOS
This is a hack to make Clang use the initial-exec TLS model instead of the default local-exec when building code for Serenity.
This patch should be removed when we implement proper TLS support.
0005-Add-support-for-SerenityOS.patch
Add support for SerenityOS
This commit teaches libc++ about what features are available in our LibC, namely:
- We do not have locale support, so no-op shims should be used in place of the C locale API.
- The number of errno constants defined by us is given by the value of
the
ELAST
macro. - Multithreading is implemented though the pthread library.
- Use libc++'s builtin character type table instead of the one provided by LibC as there's a lot of extra porting work to convince the rest of locale.cpp to use our character type table properly.
This commit is an adaptation of the LLVM patch by Daniel Bertalan to fit the layout of the zig-bootstrap project.
0006-Allow-undefined-symbols-on-SerenityOS.patch
Allow undefined symbols on SerenityOS
Allow undefined symbols in LLVM libraries, which is needed because only stubs are available for SerenityOS libraries when libc++ and libunwind are built.
0007-Support-building-shared-libLLVM-and-libClang-for-Ser.patch
Support building shared libLLVM and libClang for SerenityOS
This patch tells CMake that the --whole-archive linker option should be used for specifying the archives whose members will constitute these shared libraries.
Symbol versioning is disabled, as the SerenityOS loader doesn't support it, and the ELF sections that store version data would just waste space.
0008-Add-SerenityOS-to-config.guess.patch
Add SerenityOS to config.guess
0009-Prevent-the-use-of-POSIX-shm-on-SerenityOS.patch
Prevent the use of POSIX shm on SerenityOS
POSIX shm is not supported by SerenityOS yet, so this causes a compilation error.
0010-llvm-Implement-bigint-to-LLVM-int-for-32-bit-compile.patch
llvm: Implement bigint-to-LLVM int for 32-bit compiler builds
The conversion to DoubleLimb is necessary due to LLVM only accepting 64-bit limbs for big integers. Since we need some space to store it, we also have to allocate. This is an unfortunate penalty that 32-bit compiler builds have to take.
0011-Add-SerenityOS-target.patch
Add SerenityOS target
Named "serenity" within the code to match what LLVM says.
0012-Implement-SerenityOS-support-in-std.patch
Implement SerenityOS support in std
0013-build-Adjust-build-process-for-SerenityOS.patch
build: Adjust build process for SerenityOS