If the hw doesn't have support for exotic floating-point types such
as `f80`, we lower the call to a compiler-rt function call instead.
I've added a behavior test specifically targeting this use case which
now passes on `aarch64-macos`.
Additionally, this commit makes it possible to successfully build
stage3 on `aarch64-macos`. We can print the compiler's help message,
however, building with it needs a little bit more love still.
So far it's supported by the LLVM backend only. I recommend for the
other backends to wait for the resolution of #10761 before adding
support for this feature.
These targets now have a similar disagreement with LLVM about the
alignment of 128-bit integers as x86_64:
* riscv64
* powerpc64
* powerpc64le
* mips64
* mips64el
* sparcv9
See #2987
For x86_64, LLVMABIAlignmentOfType(i128) reports 8. However I think 16
is a better number for two reasons:
1. Better machine code when loading into SIMD register.
2. The C ABI wants 16 for extern structs.
ZIR instructions updated: atomic_load, atomic_rmw, atomic_store, cmpxchg
These no longer construct a pointer type as the result location. This
solves a TODO that was preventing the pointer from possibly being
volatile, as well as properly handling allowzero and addrspace.
It also allows the pointer to be over-aligned, which may be needed
depending on the target. As a consequence, the element type needs to be
communicated in the ZIR. This is done by strategically making one of the
operands be ResultLoc.ty instead of ResultLoc.coerced_ty if possible, or
otherwise explicitly adding elem_type into the ZIR encoding, such as in
the case of atomic_load.
The pointer type of atomic operations is now checked in Sema by coercing
it to an expected pointer type, that maybe over-aligned according to
target requirements.
Together with the previous commit, Zig now has smaller alignment for
large integers, depending on the target, and yet still has type safety
for atomic operations that specially require higher alignment.
Prior to this commit, the logic for ABI size and ABI alignment for
integers was naive and incorrect. This results in wasted hardware as
well as undefined behavior in the LLVM backend when we memset an
incorrect number of bytes to 0xaa due to disagreeing with LLVM about the
ABI size of integers.
This commit introduces a "max int align" value which is different per
Target. This value is used to derive the ABI size and alignment of all
integers.
This commit makes an interesting change from stage1, which treats
128-bit integers as 16-bytes aligned for x86_64-linux. stage1 is
incorrect. The maximum integer alignment on this system is only 8 bytes.
This change breaks the behavior test called "128-bit cmpxchg" because on
that target, 128-bit cmpxchg does require a 16-bytes aligned pointer to
a 128 bit integer. However, this alignment property does not belong on
*all* 128 bit integers - only on the pointer type in the `@cmpxchg`
builtin function prototype. The user can then use an alignment override
annotation on a 128-bit integer variable or struct field to obtain such
a pointer.
Just like for Struct in 8238d4b33585a715c58ab559cd001dd3ea1db55b, in the
case of ErrorUnion struct we need to return a compound literal "(T){...}"
instead of just "{}", which is invalid code when used in e.g. a "return"
expression.
* Sema: Correctly determine whether array_cat lhs and rhs are single ptrs
Many-pointers are also not single-pointers and wouldn't be considered
here. This commit makes the conditions use the appropriately-named
isSinglePointer instead.
* Sema: Correctly obtain ArrayInfo for many-pointer concatenation
Many-pointers at comptime have a known size like slices and can be used
in array concatenation. This fixes a stage1 regression.
* test: Add comptime manyptr concatenation test
Co-authored-by: sin-ack <sin-ack@users.noreply.github.com>
This is to account for the small differences in math functions of
different libcs. For example, if the compiler links against glibc,
but the target is musl libc, then these values might be
slightly different.
Arguably, this is a bug in the compiler because comptime should
emulate the target, including rounding errors in libc math
functions. However that behavior is not what this particular test
is intended to cover.
The reason for having `@tan` is that we already have `@sin` and `@cos`
because some targets have machine code instructions for them, but in the
case that the implementation needs to go into compiler-rt, sin, cos, and
tan all share a common dependency which includes a table of data. To
avoid duplicating this table of data, we promote tan to become a builtin
alongside sin and cos.
ZIR: The tag enum is at capacity so this commit moves
`field_call_bind_named` to be `extended`. I measured this as one of
the least used tags in the zig codebase.
Fix libc math suffix for `f32` being wrong in both stage1 and stage2.
stage1: add missing libc prefix for float functions.
* unify the logic for exporting math functions from compiler-rt,
with the appropriate suffixes and prefixes.
- add all missing f128 and f80 exports. Functions with missing
implementations call other functions and have TODO comments.
- also add f16 functions
* move math functions from freestanding libc to compiler-rt (#7265)
* enable all the f128 and f80 code in the stage2 compiler and behavior
tests (#11161).
* update std lib to use builtins rather than `std.math`.
Split big test into the two separate things it is testing.
Add missing checks to the test which revealed the test is not actually
passing yet for the C backend.
According to Apple docs, the long double type is a double precision
IEEE754 binary floating-point type, which makes it identical to the
double type. This behavior contrasts to the standard specification,
in which a long double is a quad-precision, IEEE754 binary,
floating-point type.
Thus, we need to take this into account when using the compiler
intrinsics so that we select the correct function version for
FloatMulAdd.
* The `@bitCast` workaround is removed in favor of `@ptrCast` properly
doing element casting for slice element types. This required an
enhancement both to stage1 and stage2.
* stage1 incorrectly accepts `.{}` instead of `{}`. stage2 code that
abused this is fixed.
* Make some parameters comptime to support functions in switch
expressions (as opposed to making them function pointers).
* Avoid relying on local temporaries being mutable.
* Workarounds for when stage1 and stage2 disagree on function pointer
types.
* Workaround recursive formatting bug with a `@panic("TODO")`.
* Remove unreachable `else` prongs for some inferred error sets.
All in effort towards #89.
The problem was that types of non-anytype parameters were being included
as part of the check to see if generic function instantiations were
equal. Now, Module.Fn additionally stores the information for whether each
parameter is anytype or not. `generic_poison` cannot be used to signal
this because the type is still needed for comptime arguments; in such
case the type will not be present in the newly generated function
prototype.
This presented one additional challenge: we need to compare equality of
two values where one of them is post-coercion and the other is not. So
we make some minor adjustments to `Type.eql` to support this. I think
this small complexity tradeoff is worth it because it means the compiler
does much less work on the hot path that a generic function is called
and there is already an existing matching instantiation.
closes#11146
