According to a comment in mold, this is the expected (and desired)
condition by the linkers, except for some architectures (RISCV and
Loongarch) where this condition does not have to upheld.
If you follow the changes in this patch and in particular doc comments
I have linked the comment/code in mold that explains and implements
this.
I have also modified `testEhFrameRelocatable` test to now test both
cases such that `zig ld -r a.o b.o -o c.o` and `zig ld -r b.o a.o -o
d.o`. In both cases, `c.o` and `d.o` should produce valid object
files which was not the case before this patch.
Make shared_objects a StringArrayHashMap so that deduping does not
need to happen in flush. That deduping code also was using an O(N^2)
algorithm, which is not allowed in this codebase. There is another
violation of this rule in resolveSymbols but this commit does not
address it.
This required reworking shared object parsing, breaking it into
independent components so that we could access soname earlier.
Shared object parsing had a few problems that I noticed and fixed in
this commit:
* Many instances of incorrect use of align(1).
* `shnum * @sizeOf(elf.Elf64_Shdr)` can overflow based on user data.
* `@divExact` can cause illegal behavior based on user data.
* Strange versyms logic that wasn't present in mold nor lld. The logic
was not commented and there is no git blame information in ziglang/zig
nor kubkon/zld. I changed it to match mold and lld instead.
* Use of ArrayList for slices of memory that are never resized.
* finding DT_VERDEFNUM in a different loop than finding DT_SONAME.
Ultimately I think we should follow mold's lead and ignore this
integer, relying on null termination instead.
* Doing logic based on VER_FLG_BASE rather than ignoring it like mold
and LLD do. No comment explaining why the behavior is different.
* Mutating the original ELF symbols rather than only storing the mangled
name on the new Symbol struct.
I noticed something that I didn't try to address in this commit: Symbol
stores a lot of redundant information that is already present in the ELF
symbols. I suspect that the codebase could benefit from reworking Symbol
to not store redundant information.
Additionally:
* Add some type safety to std.elf.
* Eliminate 1-3 file system reads for determining the kind of input
files, by taking advantage of file name extension and handling error
codes properly.
* Move more error handling methods to link.Diags and make them
infallible and thread-safe
* Make the data dependencies obvious in the parameters of
parseSharedObject. It's now clear that the first two steps (Header and
Parsed) can be done during the main Compilation pipeline, rather than
waiting for flush().
By organizing linker diagnostics into this struct, it becomes possible
to share more code between linker backends, and more importantly it
becomes possible to pass only the Diag struct to some functions, rather
than passing the entire linker state object in. This makes data
dependencies more obvious, making it easier to rearrange code and to
multithread.
Also fix MachO code abusing an atomic variable. Not only was it using
the wrong atomic operation, it is unnecessary additional state since
the state is already being protected by a mutex.
Embrace the Path abstraction, doing more operations based on directory
handles rather than absolute file paths. Most of the diff noise here
comes from this one.
Fix sorting of crtbegin/crtend atoms. Previously it would look at all
path components for those strings.
Make the C runtime path detection partially a pure function, and move
some logic to glibc.zig where it belongs.
The initAtoms function now only uses the `elf_file` parameter for
reporting linker error messages, making it easier to see that the
function has no data dependencies other than the Object struct itself,
making it easier to parallelize or otherwise move that logic around.
Also removed an indirect call via `addExtra` since we already know the
atom's file is the current Object instance. All calls to `Atom.addExtra`
should be audited for similar reasons.
Also removed unjustified use of `inline fn`.
flush() must not do anything more than necessary. Determining the type
of input files must be done only once, before flush. Fortunately, we
don't even need any file system accesses to do this since that
information is statically known in most cases, and in the rest of the
cases can be determined by file extension alone.
This commit also updates the nearby code to conform to the convention
for error handling where there is exactly one error code to represent
the fact that error messages have already been emitted. This had the
side effect of improving the error message for a linker script parse
error.
"positionals" is not a linker concept; it is a command line interface
concept. Zig's linker implementation should not mention "positionals".
This commit deletes that array list in favor of directly making function
calls, eliminating that heap allocation during flush().
The compiler actually doesn't need any functional changes for this: Sema
does reification based on the tag indices of `std.builtin.Type` already!
So, no zig1.wasm update is necessary.
This change is necessary to disallow name clashes between fields and
decls on a type, which is a prerequisite of #9938.
These are fundamentally incapable of producing accurate information for reasons
I've laid out in #20771. Since our only use of these functions is to check that
object files have the correct machine type, and since #21020 made
`std.Target.to{Coff,Elf}Machine()` more accurate, just switch these checks over
to that and compare the machine type tags instead.
Closes#20771.
