Future improvement: make plain error notes actually render as notes
rather than errors, but keep them as errors for the case of
sub-compilation errors, e.g. when compiler-rt has compilation errors.
MachO linker now handles `-needed-l<name>`, `-needed_library=<name>`
and `-needed_framework=<name>`. While on macOS `-l` is equivalent
to `-needed-l`, and `-framework` to `-needed_framework`, it can be
used to the same effect as on Linux if combined with `-dead_strip_dylibs`.
This commit also adds handling for `-needed_library` which is macOS
specific flag only (in addition to `-needed-l`).
Finally, in order to leverage new linker testing harness, this commit
added ability to specify lowering to those flags via `build.zig`:
`linkSystemLibraryNeeded` (and related), and `linkFrameworkNeeded`.
Includes both traditiona and incremental codepaths with one caveat that
in incremental case, the requested size cannot be smaller than the
default padding size due to prealloc required due to incremental nature
of linking.
Also parse `-headerpad_max_install_names`, however, not actionable just yet -
missing implementation.
This adds clarification to the getGlobalSymbol doc comments,
as well as renames the `addExternFn` function for MachO to `getGlobalSymbol`.
This function will now be called from 'src/link.zig' as well.
Finally, this also enables compiling zig's libc using LLVM even though
the `fno-LLVM` flag is given.
Rather than checking if the user wants to use LLVM for the current compilation,
check for the existance of LLVM as part of the compiler. This is temporarily,
until other backends gain the ability to compiler LLVM themselves.
This means that when a user passed `-fno-LLVM` we will use the native
backend for the user's code, but use LLVM for compiler-rt.
This also fixes emitting names for symbols in the Wasm linker,
by deduplicating symbol names when multiple symbols point the same object.
If page aligned requested pagezero size is 0, skip generating
__PAGEZERO segment.
Add misc improvements to the pipeline, and correctly transfer the
requested __PAGEZERO size to the linker.
Pass `-pagezero_size` to the MachO linker. This is the final
"unsupported linker arg" that I could chase that CGo uses. After this
and #11874 we may be able to fail on an "unsupported linker arg" instead
of emiting a warning.
Test case:
zig=/code/zig/build/zig
CGO_ENABLED=1 GOOS=darwin GOARCH=amd64 CC="$zig cc -target x86_64-macos" CXX="$zig c++ -target x86_64-macos" go build -a -ldflags "-s -w" cgo.go
I compiled a trivial CGo program and executed it on an amd64 Darwin
host.
To be honest, I am not entirely sure what this is doing. This feels
right after reading what this argument does in LLVM sources, but I am by
no means qualified to make MachO pull requests. Will take feedback.
After doing performance testing, it seems that multi-compilation-unit
compiler-rt did not bring the performance improvements that we expected
it to. The idea is that it makes linking faster, however, it incurred a
cost in the frontend that was not offset by any gains in linking.
Furthermore, the single-object compiler-rt (with -ffunction-sections and
--gc-sections) ends up being fewer bytes on disk and so it's actually
the same or faster linking speed than the multi-compilation-unit
version.
So we are planning to keep using single-compilation-unit compiler-rt for
the foreseeable future, but may experiment with this again in the
future, in which case this commit can be reverted.
compiler_rt_lib and compiler_rt_obj are extracted from the generic
JobQueue into simple boolean flags, and then handled explicitly inside
performAllTheWork().
Introduced generic handling of allocation failure and made
setMiscFailure not return a possible error.
Building the compiler-rt static library now takes advantage of
Compilation's ThreadPool. This introduced a problem, however, because
now each of the object files of compiler-rt all perform AstGen for the
full standard library and compiler-rt files. Even though all of them end
up being cache hits except for the first ones, this is wasteful - O(N*M)
where N is number of compilation units inside compiler-rt and M is the
number of .zig files in the standard library and compiler-rt combined.
More importantly, however, it causes a deadlock, because each thread
interacts with a file system lock for doing AstGen on files, and threads
end up waiting for each other. This will need to be handled with a
process-level file caching system, or some other creative solution.
Note that the current documentation for the `-z noexecstack` is
incorrect. This indicates that an object *does not* require an
executable stack.
This is actually the default of LLD, and there has never been a way to
override this default by passing `-z execstack` to LLD.
This commit removes the redundant `-z noexecstack` option from
zig build-exe/build-lib/build-obj and ignores the option if passed
to zig cc for compatibility.
As far as I can tell, there is no reason for code to require an
executable stack. This option only exists because the stack was
originally executable by default and some programs came to depend
on that behavior. Instead, mprotect(2) may be used to make memory
pages executable.
Full RELRO is a hardening feature that makes it impossible to perform
certian attacks involving overwriting parts of the Global Offset Table
to invoke arbitrary code.
It requires all symbols to be resolved before execution of the program
starts which may have an impact on startup time. However most if
not all popular Linux distributions enable full RELRO by default for
all binaries and this does not seem to make a noticeable difference
in practice.
"Partial RELRO" is equivalent to `-z relro -z lazy`.
"Full RELRO" is equivalent to `-z relro -z now`.
LLD defaults to `-z relro -z lazy`, which means Zig's current `-z relro`
option has no effect on LLD's behavior.
The changes made by this commit are as follows:
- Document that `-z relro` is the default and add `-z norelro`.
- Pass `-z now` to LLD by default to enable full RELRO by default.
- Add `-z lazy` to disable passing `-z now`.
Prior to this change we would assume the ABI for Apple targets to
be GNU which could result in subtle errors in LLVM emitting calls
to non-existent system libc provided functions such as `_sincosf`
which is a GNU extension and as such is not provided by macOS for example.
This would result in linker errors where the linker would not be
able to find the said symbol in `libSystem.tbd`.
With this change, we now correctly identify macOS (and other Apple
platforms) as having ABI `unknown` which translates to unspecified
in LLVM under-the-hood:
```
// main.ll
target triple = "aarch64-unknown-macos-unknown"
```
Note however that we never suffix the target OS with target version
such as `macos11` or `macos12` which means we fail to instruct LLVM
of potential optimisations provided by the OS such as the availability
of function `___sincosf_stret`. I suggest we investigate that in a
follow-up commit.
Warnings about non-implemented `-z nocopyreloc` are common when
compiling go code (including Go's tests themselves). Let's just
make it stop complaining.
Originally I thought interleaving AIR with MIR will be useful, however
as it stands, I have used it very sporadically, and recently, not at
all, and I do not think anyone else is actually using it. If there is
a simple error such as a wrong instruction emitted,
`objdump` is perfectly capable of narrowing it down, while if there's
something more subtle happening, regardless of having `--verbose-mir`
functionality or not, you still gotta go via the debugger which
offers a better view at interleaved source program with the emitted
machine code. Finally, I believe `-femit-asm` when we add it will offer a
more generic substitute.
Rename all references of sparcv9 to sparc64, to make Zig align more with
other projects. Also, added new function to convert glibc arch name to Zig
arch name, since it refers to the architecture as sparcv9.
This is based on the suggestion by @kubkon in PR 11847.
(https://github.com/ziglang/zig/pull/11487#pullrequestreview-963761757)
* back out the changes to RunStep
* move the disabled test to the .cpp code and avoid a confusing
name-collision with the _LIBCPP macro prefix
* fix merge conflict with the edits to the same test that ensure global
initializers are called.
Now this branch is only concerned with single-threaded targets and
passing the correct macro defines to libc++.
stage2: change logic for detecting whether the main package is inside
the std package. Previously it relied on realpath() which is not portable.
This uses resolve() which is how imports already work.
* stage2: fix cleanup bug when creating Module
* flatten lib/std/special/* to lib/*
- this was motivated by making main_pkg_is_inside_std false for
compiler_rt & friends.
* rename "mini libc" to "universal libc"
Instead, just return ChildProcess directly. This structure does not
require a stable address, so we can put it on the stack just fine. If
someone wants it on the heap they should do.
const proc = try allocator.create(ChildProcess);
proc.* = ChildProcess.init(args, allocator);
This reverts commit a430630002bf02162ccbf8d3eb10fd73e490cefd.
Wait a minute, I'm sorry, I need to revert this. The whole premise
of this change is broken because the point of the hash is that it tells
whether the same compilation has been done before. This requires items
to be added to the hash in the same sequence every time. This means that
introducing a lock is fundamentally broken because the order needs to be
the same in future runs of the compiler, and not decided by threads
racing against each other.
The proper solution to this is to, in whole cache mode, append the hash
inputs to some data structure, and then after the compilation is
complete, do some kind of sorting on the hash inputs so that they will
be the same order every time, then apply them in sequence. No lock on
the Cache object is needed for this scheme.
This makes stage2 and stage3 have different cache namespaces, so that
building something with stage3 does not try to reuse the same cached
artifacts as were produced by stage2. This makes sense since the code
of stage3 is produced by the self-hosted compiler, whereas the code of
stage2 is produced by the bootstrap compiler. Note also that stage4 and
stage3 will share the same zig_backend, end hence cache namespace.
Ideally stage4 and stage3 are identical binaries, so this checks out.
Rather than allocating Decl objects with an Allocator, we instead allocate
them with a SegmentedList. This provides four advantages:
* Stable memory so that one thread can access a Decl object while another
thread allocates additional Decl objects from this list.
* It allows us to use u32 indexes to reference Decl objects rather than
pointers, saving memory in Type, Value, and dependency sets.
* Using integers to reference Decl objects rather than pointers makes
serialization trivial.
* It provides a unique integer to be used for anonymous symbol names,
avoiding multi-threaded contention on an atomic counter.
This flag is used when building stage1 to omit the stage2 backends from
the compiler to save memory on the CI server. It regressed with the
merging of e8813b296bc55a13b534bd9b2a03e1f6af366915 because Value
functions started calling into Sema functions.
The end goal for this build option is to eliminate it.
With this change, we can now bake in entitlements into the binary.
Additionally, I see this as the first step towards full code signature
support which includes baking in Apple issued certificates for
redistribution, etc.
