Get rid of `std.math.F80Repr`. Instead of trying to match the memory
layout of f80, we treat it as a value, same as the other floating point
types. The functions `make_f80` and `break_f80` are introduced to
compose an f80 value out of its parts, and the inverse operation.
stage2 LLVM backend: fix pointer to zero length array tripping LLVM
assertion. It now checks for when the element type is a zero-bit type
and lowers such thing the same way that pointers to other zero-bit types
are lowered.
Both stage1 and stage2 LLVM backends are adjusted so that f80 is lowered
as x86_fp80 on x86_64 and i386 architectures, and identical to a u80 on
others. LLVM constants are lowered in a less hacky way now that #10860
is fixed, by using the expression `(exp << 64) | fraction` using llvm
constants.
Sema is improved to handle c_longdouble by recursively handling it
correctly for whatever the float bit width is. In both stage1 and
stage2.
* F80Repr extern struct needs no explicit padding; let's match the
target padding.
* stage2: fix lowering of f80 constants.
* stage1: decide ABI size and alignment of f80 based on alignment of
u64. x86 has alignof u64 equal to 4 but arm has it as 8.
* stage2: fix Value.floatReadFromMemory to use F80Repr
replaceAllUsesWith requires the type to be unchanged. So we bitcast
the new global to the old type and use that as the thing to replace
old uses.
Fixes an LLVM assertion found while troubleshooting #10837.
Match changes required to `Elf` linker, which enable lowering
of const slices on `MachO` targets.
Expand `Mir` instructions requiring the knowledge of the containing
atom - pass the symbol index into the linker's table from codegen
via mir to emitter, to then utilise it in the linker.
In `getDeclVAddr`, it may happen that the target `Decl` has not
been allocated space in virtual memory. In this case, we store a
relocation in the linker-global table which we will iterate over
when flushing the module, and fill in any missing address in the
final binary. Note that for optimisation, if the address was resolved
at the time of a call to `getDeclVAddr`, we skip relocating this
atom.
This commit also adds the glue code for lowering const slices in
the ARM backend.
This implements the `field_ptr` value for pointers. As the value only provides us with the index,
we must calculate the offset from the container type using said index. (i.e. the offset from a struct field at index 2).
Besides this, small miscellaneous fixes/updates were done to get remaining behavior tests passing:
- We start the function table index at 1, so unresolved function pointers don't can be null-checked properly.
- Implement genTypedValue for floats up to f64.
- Fix zero-sized arguments by only creating `args` for non-zero-sized types.
- lowerConstant now works for all decl_ref's.
- lowerConstant properly lowers optional pointers, so `null` pointers are lowered to `0`.
LLVM union globals have to be lowered as unnamed structs if the
non-most-aligned field is the active tag. In this case it bubbles up so
that structs containing unions have the same restriction.
This fix needs to be applied to optionals and other callsites of
createNamedStruct.
The bug fixed in this commit was revealed in searching for
the cause of #10837.
We need to pad out the file to the required maximum size equal the
final section's offset plus the section's size. We only need to
this when populating initial metadata and only when section header
was updated.
For some projects, they can't help themselves, -lgcc_s ends up on the
compiler command line even though it does not belong there. In Zig we
know what -lgcc_s does. It's an alternative to compiler-rt. With this
commit we emit a warning telling that it is unnecessary to put such
thing on the command line, and happily ignore it, since we will fulfill
the dependency with compiler-rt.
When Sema sees a store_node instruction, it now checks for
the possibility of this pattern:
%a = ret_ptr
%b = store(%a, %c)
Where %c is an error union. In such case we need to add to the
current function's inferred error set, if any.
Coercion from error union to error union will be handled ideally if the
operand is comptime known. In such case it does the appropriate
unwrapping, then wraps again.
In the future, coercion from error union to error union should do the
same thing for a runtime value; emitting a runtime branch to check if
the value is an error or not.
`Value.arrayLen` for structs returns the number of fields. This is so
that Liveness can use it for the `vector_init` instruction (soon to be
renamed to `aggregate_init`).
`const` declarations inside comptime blocks were not getting properly
evaluated at compile-time. To accomplish this there is a new ZIR
instruction, `alloc_inferred_comptime`. Actually we already had one
named that, but it got renamed to `alloc_inferred_comptime_mut` to match
the naming convention with the other similar instructions.
Rather than using runtime to perform pointer arithmetic to set the stack offset as
a pointer into a local, we now store the offset as a WValue from the bottom of the stack.
This has the benefit of less instructions, few locals, and less performance impact when
we allocate a value on the virtual stack.
* pass more x64 behavior tests
* return with a TODO error when lowering a decl with no runtime bits
* insert some debug logs for tracing recursive descent down the
type-value tree when lowering types
* print `Decl`'s name when print debugging `decl_ref` value
This updates the test runner for stage2 to emit to stdout with the passed, skipped and failed tests
similar to the LLVM backend.
Another change to this is the start function, as it's now more in line with stage1's.
The stage2 test infrastructure for wasm/wasi has been updated to reflect this as well.
* pass air_tag instead of zir_tag
* also pass eval function so that the branch only happens once and the
body of zirUnaryMath is simplified
* Value.sqrt: update to handle f80 and f128 in the normalized way that
includes handling c_longdouble.
Semi-related change: fix incorrect sqrt builtin name for f80 in stage1.
Support for f128, comptime_float, and c_longdouble require improvements
to compiler_rt and will implemented in a later PR. Some of the code in
this commit could be made more generic, for instance `llvm.airSqrt`
could probably be `llvm.airUnaryMath`, but let's cross that
bridge when we get to it.
* link: add a virtual function `lowerUnnamedConsts`, similar to
`updateFunc` or `updateDecl` which needs to be implemented by the
linker backend in order to be used with the `CodeGen` code
* elf: implement `lowerUnnamedConsts` specialization where we
lower unnamed constants to `.rodata` section. We keep track of the
atoms encompassing the lowered unnamed consts in a global table
indexed by parent `Decl`. When the `Decl` is updated or destroyed,
we clear the unnamed consts referenced within the `Decl`.
* macho: implement `lowerUnnamedConsts` specialization where we
lower unnamed constants to `__TEXT,__const` section. We keep track of the
atoms encompassing the lowered unnamed consts in a global table
indexed by parent `Decl`. When the `Decl` is updated or destroyed,
we clear the unnamed consts referenced within the `Decl`.
* x64: change `MCValue.linker_sym_index` into two `MCValue`s: `.got_load` and
`.direct_load`. The former signifies to the emitter that it should
emit a GOT load relocation, while the latter that it should emit
a direct load (`SIGNED`) relocation.
* x64: lower `struct` instantiations
