previously zig would search up directory parents for
lib/zig/std/std.zig. It still does that, but now it also accepts
lib/std/std.zig. This will probably be how we start shipping windows
.zip files, since the extra directory is pointless.
It also means you can run zig binaries from subdirectories of the zig
source tree without lib files being installed.
From: Szabolcs Nagy <szabolcs.nagy@arm.com>
armv8 removed the coprocessor instructions other than cp14, so
on an armv8 system the related hwcaps should never be set.
new llvm complains about the use of coprocessor instructions in
armv8-a mode (even though they are never executed at runtime),
so ifdef them out when musl is built for armv8.
<dalias> i think the patch looks ok
Until ability to specify target CPU features (#2883) is done, this
commit gives riscv target better default features.
This side-steps #3275 which is a deficiency in compiler-rt when features
do not include 32 bit integer division.
With this commit, RISC-V compiler-rt tests pass and Hello World works
both pure-zig and with musl libc.
Previously if the type parameter was a pointer, it would assert that the
size of the type was resolved. It used to be that the size of pointers was
always resolved, however with lazy values, pointers gained the
possibility of not having their size resolved.
Now, type_allowed_in_extern triggers the resolution of whether a pointer
is zero bits, and returns a possible error if the resolution fails.
This fixes a compiler assertion when building the
[zootdeck project](https://github.com/donpdonp/zootdeck). I do not have
a test case reduction for the issue.
* add zig build option `-Dskip-libc` to skip tests that build libc
(e.g. if you don't want to wait for musl to build)
* add `-Denable-wine` option which uses wine to run cross compiled
windows tests on non-windows hosts
* add `-Denable-qemu` option which uses qemu to run cross compiled
foreign architecture tests
* add `-Denable-foreign-glibc=path` option which combined with
`-Denable-qemu` enables running cross compiled tests that link
against glibc. See
https://github.com/ziglang/zig/wiki/Updating-libc#glibc for how to
produce this directory.
* the test matrix is done manually. release test builds are only
enabled by default for the native target. this should save us some CI
time, while still providing decent coverage of release builds.
- add test coverage for `x86_64-linux-musl -lc` (building musl libc)
- add test coverage for `x86_64-linux-gnu -lc` (building glibc)
- add test coverage for `aarch64v8_5a-linux-none`
- add test coverage for `aarch64v8_5a-linux-musl -lc` (building musl libc)
- add test coverage for `aarch64v8_5a-linux-gnu -lc` (building glibc)
- add test coverage for `arm-linux-none`
- test coverage for `arm-linux-musleabihf -lc` (building musl libc) is
disabled due to #3286
- test coverage for `arm-linux-gnueabihf -lc` (building glibc) is disabled
due to #3287
- test coverage for `x86_64-windows-gnu -lc` (building mingw-w64) is
disabled due to #3285
* enable qemu testing on the Linux CI job. There's not really a good
reason to enable wine, since we have a Windows CI job as well.
* remove the no longer needed `--build-file ../build.zig` from CI
scripts
* fix bug in glibc compilation where it wasn't properly reading the abi
list txt files, resulting in "key not found" error.
* std.build.Target gains:
- isNetBSD
- isLinux
- osRequiresLibC
- getArchPtrBitWidth
- getExternalExecutor
* zig build system gains support for enabling wine and enabling qemu.
`artifact.enable_wine = true;`, `artifact.enable_qemu = true;`. This
communicates that the system has these tools installed and the build
system will use them to run tests.
* zig build system gains support for overriding the dynamic linker of
an executable artifact.
* fix std.c.lseek prototype. makes behavior tests for
arm-linux-musleabihf pass.
* disable std lib tests that are failing on ARM. See #3288, #3289
* provide `std.os.off_t`.
* disable some of the compiler_rt symbols for arm 32 bit. Fixes
compiler_rt tests for arm 32 bit
* add __stack_chk_guard when linking against glibc. Fixes std lib tests
for aarch64-linux-gnu
* workaround for "unable to inline function" using `@inlineCall`. Fixes
compiler_rt tests for arm 32 bit.
- bug presented on FreeBSD when `/proc` filesystem is not mounted
- bogus `cc` was used as exename, causing incorrect executable spawn
- llvm::sys::fs::getMainExecutable() has platform-specific code
to get exename and fallback is to use param argv0
- linux fallback is rare because `/proc` is usually mounted
- *BSD fallback is not rare because `/proc` is often not mounted
- macOS doesn't ever fallback: bug cannot present
- windows doesn't ever fallback: bug cannot present
- other Posix will always present
Currently, slices are passed via reference, even though it would be
better to pass the ptr and len as separate arguments (#561). This means
that any function call with a slice parameter cannot be a tail call,
because according to LLVM spec:
> Both [tail,musttail] markers imply that the callee does not access
> allocas from the caller
There was one other place we were setting `tail` and I made that
conditional on whether or not the argument referenced allocas in the
caller.
This was causing undefined behavior in the compiler when it hit asserts,
causing it to print garbage memory to the terminal. See #3262 for
example.
* Fix codegen for splat - instead of giving vectors of length N
to shufflevector for both of the operands, it gives vectors of length
1. The mask vector is the only one that needs N elements.
* Separate Splat into SplatSrc and SplatGen; the `len` is not needed
once it gets to codegen since it is redundant with the result type.
* Refactor compile error for wrong vector element type so that the
compile error message is not duplicated in zig source code
* Improve implementation to correctly handle comptime values such as
undefined and lazy values.
* Improve compile error for bad vector element type to point to the
correct place.
* Delete dead code.
* Modify behavior test to use an array cast instead of vector element
indexing since I'm merging this splat commit out-of-order from
Shawn's patch set.
* update docs for `@byteSwap`.
* fix hash & eql functions for ZigLLVMFnIdBswap not updated to
include vector len. this was causing incorrect bswap function
being called in unrelated code
* fix `@byteSwap` behavior tests only testing comptime and not
runtime operations
* implement runtime `@byteSwap`
* fix incorrect logic in ir_render_vector_to_array and
ir_render_array_to_vector with regards to whether or not to bitcast
* `@byteSwap` accepts an array operand which it will cast to vector
* simplify `@byteSwap` semantic analysis code and various fixes
* update documentation
- move `@shuffle` to be sorted alphabetically
- remove mention of LLVM
- minor clarifications & rewording
* introduce ir_resolve_vector_elem_type to avoid duplicate compile
error message and duplicate vector element checking logic
* rework ir_analyze_shuffle_vector to solve various issues
* improve `@shuffle` to allow implicit cast of arrays
* the shuffle tests weren't being run
I change the semantics of the mask operand, to make it a little more
flexible. There is no real danger in this because it is a compile-error
if you do it the LLVM way (and there is an appropiate error to tell you
this).
v2: avoid problems with double-free
The question was:
> // TODO do we need lazy values on vector comparisons?
Nope, in fact the existing code already was returning ErrorNotLazy
for that particular type, and would already goto
never_mind_just_calculate_it_normally. So the explicit check for
ZigTypeIdVector is not needed. I appreciate the caution though.
* bitcasting is still better when the size_in_bits aligns with the ABI
size of the element type. Logic is reworked to do bitcasting when
possible
* rather than using insertelement/extractelement to work with arrays,
store/load elements directly. This matches codegen for arrays
elsewhere.
