We now do not allocate memory for headers and other metadata unless
requested by the caller. Instead, we read-in the entire contents
of the image into memory and operate on pointers and casts wherever
possible. I have a left a TODO to hook up Windows' memory-mapped API
here in-place of standard `readToEndAlloc` which should be more memory
proof on memory constrained hosts.
This commit also supplements our `std.coff` with a lot missing basic
extern structs required to make our COFF linker.
* Added support for stroffsetsptr class in Dwarf stdlib
* Proper initializion of debug_str_offsets in DwarfInfo
* Added missing null initializer to DwarfInfo in Macho
* Added missing is_64 field to getAttrString in DwarfInfo
* Fixed formatting
* Added missing is_64 param to getAttrString
* Added required cast to usize
* Adding missing .debug_str_offsets initialization
* getAttrString now uses the str_offsets_base attr
When an object file is being parsed from within an archive
file, we provide the object file size to ensure we do not
read past the object file. This is because follow up object
files can exist there, as well as an LF character to notate
the end of the file was reached. Such a character is invalid
within the object file.
This also fixes a bug in getting the function/global type
for defined globals/functions from object files as it was missing
the substraction with the import count of the respective type.
Wasm archive files are encoded the same way as GNU.
This means that the header notates the character index within
the long file name list rather than the length of the name.
The entire name is then delimited by an LF character (0x0a).
This also makes a cosmetic update to remove the `self` name,
and rather label it as `archive` instead.
Firstly, opening a file handle is not really needed since we won't even
use it, and secondly, this can cause AccessDenied errors on Windows
when trying to move a directory from zig-cache/tmp/ to zig-cache/o/
since, without POSIX semantics, it is illegal to move directories
with open handles to any of its resources.
This adds additional checks during symbol resolution:
- Ensures function signatures match when a symbol will be replaced.
- Ensures global types match when the symbol is being replaced.
- When both symbols are undefined, ensures they have a matching module name.
Those changes ensure the result will pass the validator when
the runtime compiles the Wasm module.
Additionally, this also slightly changes the behavior when both
the existing symbol and new symbol are both defined. Rather than
always resulting in a collision, it only results in a collision
when both are also weak. Else, the non-weak symbol will be picked.
Add handling for these additional `MCValue`s:
* `.immediate` - lower to `DW.OP.consts` or `DW.OP.constu` depending
on signedness followed by popping off the DWARF stack with
`DW.OP.stack_value`
* `.undef` - lower to `DW.OP.implicit_value`
* `.none` - lower to `DW.OP.lit0` followed by popping off the DWARF
stack with `DW.OP.stack_value`
For any remaining unhandled case, we generate `DW.OP.nop` in order
not to mess up remaining DWARF info.
This way, tracking segment-to-section mapping becomes a lot easier
since it's effectively just start index plus number of sections
defined within the segment. If a section becomes empty however
care needs to be taken to remove the header upon committing to the
final binary.
Fix incorrect writing of symtab and strtab in dSYM bundle in incremental
context.
Fix incorrectly navigating unnamed consts (freeing) in incremental context.
This is currently hard-coded to require all consts to land in `__TEXT,__const`,
which is wrong and needs a rewrite.
Instead of generating sections upfront, allow generation by scanning
the object files for input -> output sections mapping. Next, always
strive to keep output sections in the final container sorted as they
appear in the final binary. This makes the linker less messy wrt
handling of output sections sort order for dyld/macOS not to complain.
There's still more work to be done for incremental context though
to make this work but looks promising already.
Shared libraries can be provided on the command line as if they were
objects, as a path to the ".so" file. The "each-lib-rpath" functionality
was ignoring these shared libraries accidentally, causing missing rpaths
in the output executable.
Now instead of zig.h being baked into the compiler binary, it is a
header file distributed along with all the other header files
distributed with Zig.
Closes#11643
At least on Linux, the pwritev syscall checks the pointer and returns
EFAULT before it checks if the length is nonzero.
Perhaps this should be fixed in the standard library, however, these are
still improvements since they make the kernel do less work within the
syscall.
This reverts commit 7cbd586ace46a8e8cebab660ebca3cfc049305d9.
This is causing a fail to build from source:
```
./lib/std/fmt.zig:492:17: error: cannot format optional without a specifier (i.e. {?} or {any})
@compileError("cannot format optional without a specifier (i.e. {?} or {any})");
^
./src/link/MachO/Atom.zig:544:26: note: called from here
log.debug(" RELA({s}) @ {x} => %{d} in object({d})", .{
^
```
I looked at the code to fix it but none of those args are optionals.
Previously, we would get a pointer to a slot in the symbol table,
apply changes to the symbol, and return the pointer. This however
didn't take into account that the symbol table may be moved in memory
in-between the modification and return from the function (`fn placeDecl`).
Prior to my rewrite, this was not possible within the body of the said
function. However, my rewrite revamped how we allocate GOT atoms and
their matching symtab indexes, which now may cause a move in memory
of the container.
