Rather than lowering float negation as `0.0 - x`.
* Add AIR instruction for float negation.
* Add compiler-rt functions for f128, f80 negation
closes#11853
Implements semantic analysis for the new try/try_inline ZIR
instruction. Adds the new try/try_ptr AIR instructions and implements
them for the LLVM backend.
Fixes not calling rvalue() for tryExpr in AstGen.
This is part of an effort to implement #11772.
Generally, the load instruction may need to make a copy of an
isByRef=true value, such as in the case of the following code:
```zig
pub fn swap(comptime T: type, a: *T, b: *T) void {
const tmp = a.*;
a.* = b.*;
b.* = tmp;
}
```
However, it only needs to do so if there are any instructions which can
possibly write to memory. When calling functions with isByRef=true
parameters, the AIR code that is generated looks like loads followed
directly by call.
This allows for a peephole optimization when lowering loads: if the load
instruction operates on an isByRef=true type and dies before any side effects
occur, then we can safely lower the load as a no-op that returns its
operand.
This is one out of three changes I intend to make to address #11498.
However I will put these changes in separate branches and merge them
separately so that we can have three independent points on the perf
charts.
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.
* 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.
Prior to this, Liveness encoded `asm`, `call`, and `aggregate_init` with
a single 32-bit integer, allowing up to 35 operands (3 are provided by
the regular tomb_bits). However, the Zig language allows function calls
with more than 35 arguments, inline assembly with more than 35 inputs,
and anonymous tuples with more than 35 elements.
The new encoding stores an index to the extra array instead of the bits
directly, and then as many extra elements as needed to encode all the
operands. The MSB is used as a flag to tell which element is the last
one, allowing for 31 bits per element.
Prior to this, print_air did not bother correctly printing tombstones
for these instructions; now it does.
In addition to updating the BigTomb iteration logic in the machine code
backends, this commit extracts the common logic into the Liveness namespace.
This commit introduces a new AIR instruction `cmp_lt_errors_len`. It's
specific to this use case for two reasons:
* The total number of errors is not stable during semantic analysis; it
can only be reliably checked when flush() is called. So the backend
that is lowering the instruction must emit a relocation of some kind
and then populate it during flush().
* The fewer AIR instructions in memory, the better for compiler
performance, so we squish complex meanings into AIR tags without
hesitation.
The instruction is implemented only in the LLVM backend so far. It does
this by creating a simple function which is gutted and re-populated
with each flush().
AstGen now uses ResultLoc.coerced_ty for `@intToError` and Sema does the
coercion.
The existing `cmp_*` instructions get their result type from `lhs`, but
vector comparison will always return a vector of bools with only the
length derived from its operands. This necessitates the creation of a
new AIR instruction.
Notably, Value.eql and Value.hash are improved to treat NaN as equal to
itself, so that Type/Value can be hash map keys. Likewise float hashing
normalizes the float value before computing the hash.
Adds 2 new AIR instructions:
* dbg_var_ptr
* dbg_var_val
Sema no longer emits dbg_stmt AIR instructions when strip=true.
LLVM backend: fixed lowerPtrToVoid when calling ptrAlignment on
the element type is problematic.
LLVM backend: fixed alloca instructions improperly getting debug
location annotated, causing chaotic debug info behavior.
zig_llvm.cpp: fixed incorrect bindings for a function that should use
unsigned integers for line and column.
A bunch of C test cases regressed because the new dbg_var AIR
instructions caused their operands to be alive, exposing latent bugs.
Mostly it's just a problem that the C backend lowers mutable
and const slices to the same C type, so we need to represent that in the
C backend instead of printing two duplicate typedefs.
* use the real start code for LLVM backend with x86_64-linux
- there is still a check for zig_backend after initializing the TLS
area to skip some stuff.
* introduce new AIR instructions and implement them for the LLVM
backend. They are the same as `call` except with a modifier.
- call_always_tail
- call_never_tail
- call_never_inline
* LLVM backend calls hasRuntimeBitsIgnoringComptime in more places to
avoid unnecessarily depending on comptimeOnly being resolved for some
types.
* LLVM backend: remove duplicate code for setting linkage and value
name. The canonical place for this is in `updateDeclExports`.
* LLVM backend: do some assembly template massaging to make `%%`
rendered as `%`. More hacks will be needed to make inline assembly
catch up with stage1.
* mul_add AIR instruction: use `pl_op` instead of `ty_pl`. The type is
always the same as the operand; no need to waste bytes redundantly
storing the type.
* AstGen: use coerced_ty for all the operands except for one which we
use to communicate the type.
* Sema: use the correct source location for requireRuntimeBlock in
handling of `@mulAdd`.
* native backends: handle liveness even for the functions that are
TODO.
* C backend: implement `@mulAdd`. It lowers to libc calls.
* LLVM backend: make `@mulAdd` handle all float types.
- improved fptrunc and fpext to handle f80 with compiler-rt calls.
* Value.mulAdd: handle all float types and use the `@mulAdd` builtin.
* behavior tests: revert the changes to testing `@mulAdd`. These
changes broke the test coverage, making it only tested at
compile-time.
Improved f80 support:
* std.math.fma handles f80
* move fma functions from freestanding libc to compiler-rt
- add __fmax and fmal
- make __fmax and fmaq only exported when they don't alias fmal.
- make their linkage weak just like the rest of compiler-rt symbols.
* removed `longDoubleIsF128` and replaced it with `longDoubleIs` which
takes a type as a parameter. The implementation is now more accurate
and handles more targets. Similarly, in stage2 the function
CTypes.sizeInBits is more accurate for long double for more targets.
Similarly to the other wasm builtin, this implements the grow variation where the memory
index is a comptime known value. The operand as well as the result are runtime values.
This also verifies during semantic analysis the target we're building for is wasm, or else
emits a compilation error. This means that other backends do not have to handle this AIR instruction,
other than the wasm and LLVM backends.
This implements the `wasmMemorySize` builtin, in Sema and the Wasm backend.
The Stage2 implementation differs from stage1 in the way that `index` must be a comptime value.
The stage1 variant is incorrect, as the index is part of the instruction encoding, and therefore,
cannot be a runtime value.
The ZIR instruction `union_init_ptr` is renamed to `union_init`.
I made it always use by-value semantics for now, not taking the time to
invest in result location semantics, in case we decide to change the
rules for unions. This way is much simpler.
There is a new AIR instruction: union_init. This is for a comptime known
tag, runtime-known field value.
vector_init is renamed to aggregate_init, which solves a TODO comment.
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.
* AIR instruction vector_init gains the ability to init arrays and
tuples in addition to vectors. This will probably also gain the
ability to initialize structs and be renamed to `aggregate_init`.
* AstGen prefers to use an `anon_array_init` ZIR instruction for
local variables when the init expr is an array literal and there is
no type.
This reverts commit d48e4245b68bf25c7f41804a5012ac157a5ee546.
I have no idea why this is failing Drone CI, but in a branch, reverting
this commit solved the problem.
AIR:
* `array_elem_val` is now allowed to be used with a vector as the array
type.
* New instructions: splat, vector_init
AstGen:
* The splat ZIR instruction uses coerced_ty for the ResultLoc, avoiding
an unnecessary `as` instruction, since the coercion will be performed
in Sema.
* Builtins that accept vectors now ignore the type parameter. Comment
from this commit reproduced here:
The accepted proposal #6835 tells us to remove the type parameter from
these builtins. To stay source-compatible with stage1, we still observe
the parameter here, but we do not encode it into the ZIR. To implement
this proposal in stage2, only AstGen code will need to be changed.
Sema:
* `clz` and `ctz` ZIR instructions are now handled by the same function
which accept AIR tag and comptime eval function pointer to
differentiate.
* `@typeInfo` for vectors is implemented.
* `@splat` is implemented. It takes advantage of `Value.Tag.repeated` 😎
* `elemValue` is implemented for vectors, when the index is a scalar.
Handling a vector index is still TODO.
* Element-wise coercion is implemented for vectors. It could probably
be optimized a bit, but it is at least complete & correct.
* `Type.intInfo` supports vectors, returning int info for the element.
* `Value.ctz` initial implementation. Needs work.
* `Value.eql` is implemented for arrays and vectors.
LLVM backend:
* Implement vector support when lowering `array_elem_val`.
* Implement vector support when lowering `ctz` and `clz`.
* Implement `splat` and `vector_init`.
Introduced a new AIR instruction: `tag_name`. Reasons to do this
instead of lowering it in Sema to a switch, function call, array
lookup, or if-else tower:
* Sema is a bottleneck; do less work in Sema whenever possible.
* If any optimization passes run, and the operand to becomes
comptime-known, then it could change to have a comptime result
value instead of lowering to a function or array or something which
would then have to be garbage-collected.
* Backends may want to choose to use a function and a switch branch,
or they may want to use a different strategy.
Codegen for `@tagName` is implemented for the LLVM backend but not any
others yet.
Introduced some new `Type` tags:
* `const_slice_u8_sentinel_0`
* `manyptr_const_u8_sentinel_0`
The motivation for this was to make typeof() on the tag_name AIR
instruction non-allocating.
A bunch more enum tests are passing now.
