* Advance line and PC prior to ending sequence in debug line program
for a fn_decl. This is equivalent to closing scope in the debugger
and without it, the debugger will not map source-to-address info
as a result will not print the source when breaking at a symbol.
* Fix debug aranges sentinels to be of the size as the actual tuple
descriptor (assuming segment selector to be ommitted). In summary,
the sentinels were 32bit 0s, whereas they ought to be 64bit 0s.
* Make naming of symbols in the binary more consistent by prefixing
each symbol name with an underscore '_'.
* for cross testing stage2 tests, we use `darling shell` command
since the path to the tested binary is relative to cwd
* for the `zig test` command, we simply use `darling` since the
path to the binary is absolute
There are some small problems here and there, mostly due to the pointers
having the lsb set and disrupting the fn alignment tests and the
`@FrameSize` implementation.
The @ptrCast(X, @alignCast(@alignOf(T), Y)) pattern is only correct if T
is not a function type or a pointer, in that case the @alignOf refers to
the pointer itself and not to the pointee type.
The code would previously assume every function would start at addresses
being multiples of 16, this is not true beside some specific cases.
Moreover LLVM picks different alignment values depending on whether it's
trying to generate dense or fast code.
Let's use the minimum guaranteed alignment as base value, computed
according to how big the opcodes are.
The alignment of function pointers is always 1, a safe value that won't
cause any error at runtime. Note that this was already the case before
this commit, here we're making this choice explicit.
Let the 'alignment' field for TypeInfo of fn types reflect the ABI
alignment used by the compiler, make this field behave similarly to the
'alignment' one for pointers.
This was also an experiment to see if it were easier to implement a new
feature when using the instruction encoder.
Verdict: It's not that much easier, but I think it's certainly much more
readable, because the description of the Instruction annotates what each
field means. Right now, precise knowledge of x86_64 instructions is
still required because things like when to set the 64-bit flag, how to
read x86_64 instruction references, etc. are still not automatically
done for you.
In the future, this interface might make it sligtly easier to write an
assembler for x86_64, by abstracting the bit-fiddling aspects of
instruction encoding.
From my very cursory reading, it seems that the register manager doesn't
distinguish between registers that are physically the same but have
different sizes.
In that case, this means that during codegen, we can't rely on
`reg.size()` when determining the width of the operations we have to
perform. Instead, we must use some form of `ty.abiSize(self.target.*)`
to determine the size of the type we're operating with. If this size is
64 bits, then we should enable 64-bit operation.
This fixed a bug in the codegen for spilling instructions, which was
overwriting the previous stack entry with zeroes. See the modified test
case in this commit.
In std.meta.cast when casting to an enum type from an integer type, first
do a C-style cast from the source value to the tag type of the enum.
This ensures that we don't get an error due to the source value not being
representable by the enum.
In transCCast() use std.meta.cast instead of directly emitting the cast
operation since the enum's underlying type may not be known at translation
time due to an MSVC bug, see https://github.com/ziglang/zig/issues/8003Fixes#6011
Happy LLVM 12 Release Day.
Please note that the llvm 12 tag does not include the latest commit in
the release/12.x branch, which is in fact a bug fix for a regression
that is causing a failure in Zig's test suite.
Zig master branch is tracking release/12.x, and will be enabling the
test that is fixed by that commit.
This allows `break` statements to be directly translated from the original C.
Add a break statement as the last statement of the while loop to ensure we
don't have an infinite loop if no breaks / returns are hit in the switch.
Fixes#8387
Tests a scenario where the linker line has the following:
```
main.o libA.a libB.a
```
where `main.o` pulls a symbol from `libB.a`, which in turn is
dependent on a symbol from `libA.a`.
The code initializes twice `t` instead of `t1`, leaving the latter
uninitialized. The problem manifested itself by corrupting the LSBs of
the result in unpredictable ways.
- limit expected-output to main source file;
ie. tolerate changes to start.zig
- when mode != .Debug the function name is now symbolically represented;
ie. tolerate changes in llvm optimizer effects on the callstack
- cleanup how test cases are specified
- add test case predicates for excluding by arch, os or custom fn
