| Age | Commit message (Collapse) | Author |
|---|
|
Initially, internally within the linker.
|
|
|
|
Without branch islands, it is impossible to link self-hosted using
the common linker path.
|
|
This is particularly relevant for x86_64 and C++ when relocating
StaticInit sections containing static initializers machine code.
Then, in case of SIGNED_X relocations, it is necessary to have the
full image of the VM address layout of the sections in the object
file as this is how the addend needs to be adjusted for non-extern
relocations.
|
|
Also, calculate non-extern, section offset based addends for SIGNED
and UNSIGNED relocations on x86_64 upfront as an offset wrt to the
target symbol representing position of the section/atom within the
final artifact.
|
|
Otherwise, Apple's tooling goes mental and reports that the executable
is malformed/fails strict validation. We absolutely have to get it
right to support tools such `codesign` which are required to
successfully launch an app on an iOS device for instance. When Zig
matures enough so that we can ditch any Apple tooling and still be
able to successfully codesign for iOS and other, we can revisit this
area. Until then however, we are stuck in having to rewrite the LINKEDIT
segment at every update run of the self-hosted.
FYI, the strict layout for the MachO binary apparently is (please,
read this with a pinch of salt as this is inferred by me):
* __TEXT segment
* __DATA_CONST segment
* __DATA segment
* __LINKEDIT segment
* dyld info (rebase, bind, weak bind, lazy bind, export)
* symbol table
* dynamic symbol table
* string table
* code signature (if expected)
|
|
MachO, why are doing this to me?
|
|
Similarly, dirty rebase info when adding a GOT atom.
|
|
Instead of checking for stage1 at every callsite, move the logic
inside `allocateAtom`. This is fine since this logic will disappear
anyhow once I add expanding and shifting segments and sections.
|
|
This commit makes it possible to combine self-hosted with a pre-compiled
C object file, e.g.:
```
zig-out/bin/zig build-exe hello.zig add.o
```
where `add.o` is a pre-compiled C object file.
|
|
|
|
|
|
|
|
instead of globally storing unresolved and tentative defs,
store indices to actual symbols in the functions that are responsible
for symbol resolution.
|
|
Instead of storing a two-way relation (seg,sect) <=> ordinal
we get the latter with `getIndex((seg, sect))`.
|
|
instead, pass it in functions that require it. Also, when parsing
relocs, make Object part of the context struct where we pass in
additional goodies such as `*MachO` or `*Allocator`.
|
|
However, make it default only when building in release modes since
it's a prelude to advanced dead code stripping not very useful in
debug.
|
|
then, when sorting sections within segments, clear and redo the
ordinals since we re-apply them to symbols anyway. It is vital
to have the ordinals consistent with parsing and resolving relocs
however.
|
|
This happens on every call to `TextBlock.empty` by the `Module`.
|
|
Thanks to this, we no longer need to do allocs per symbol name
landing in the symbol resolver, plus we do not need to actively
track if the string was already inserted into the string table.
|
|
in self-hosted.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
It makes sense to have them as a dependent type since they only ever
deal with TextBlocks. Simplify Relocations to rely on symbol indices
and symbol resolver rather than pointers.
|
|
|
|
which should make managing the logic of parsing and resolving relocs
that much simpler to parse.
|
