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.
* Sema: fix `zirTypeInfo` allocating with the wrong arenas for some
stuff.
* LLVM: split `airDbgInline` into two functions, one for each AIR tag.
- remove the redundant copy to type_map_arena. This is the first
thing that lowerDebugType does so this hack was probably just
accidentally avoiding UB (which is still present prior to this
commit).
- don't store an inline fn inst into the di_map for the generic
decl.
- use a dummy function type for the debug info to avoid whatever UB
is happening.
- we are now ignoring the function type passed in with the
dbg_inline_begin and dbg_inline_end.
* behavior tests: prepare the vector tests to be enabled one at a time.
Mitigates #11199.
`testing.expect` is better than `testing.expectEqual` for behavior
tests. Better for behavior tests to stick to only testing the limited
behavior they are meant to test and avoid functions such as
`expectEqual` that drag in too much of the standard library (in this
case to print helpful diffs about why a value is not equal to another).
Adds the sentinel element to the type name to avoid ambiguous
declarations, and outputs the sentinel element (if needed) even in what
would otherwise be empty arrays.
This fixes one of the major issues plaguing the `std.sort` comptime tests.
The high level issue is that at comptime, we need to know whether `elem_ptr` is
being used to subslice an array-like pointer or access a child value. High-level
example:
var x: [2][2]i32 = undefined;
var a = &x[0]; // elem_ptr, type *[2]i32
var y: [5]i32 = undefined;
var b = y[1..3]; // elem_ptr, type *[2]i32
`a` is pointing directly to the 0th element of `x`. But `b` is
subslicing the 1st and 2nd element of `y`. At runtime with a well
defined memory layout, this is an inconsequential detail. At comptime,
the values aren't laid out exactly in-memory so we need to know the
difference.
This becomes an issue specifically in this case:
var c: []i32 = a;
var d: []i32 = b;
When converting the `*[N]T` to `[]T` we need to know what array to point
to. For runtime, its all the same. For comptime, we need to know if its
the parent array or the child value.
See the behavior tests for more details.
This commit fixes this by adding a boolean to track this on the
`elem_ptr`. We can't just immediately deref the child for `&x[0]`
because it is legal to ptrCast it to a many-pointer, do arithmetic, and
then cast it back (see behavior test) so we need to retain access to the
"parent" indexable.
Prior to these, the return type was non-null but the value was generic
poison which wasn't usable in user-space. This sets the value to null.
This also adds a behavior test for this.
Co-authored-by: InKryption <inkryption07@gmail.com>
This resolves https://github.com/ziglang/zig/issues/11159
The problem was that:
1. We were not correctly deleting the field stores after recognizing
that an array initializer was a comptime-known value.
2. LLVM was not checking that the final type had no runtime bits, and
so would generate an invalid store.
This also adds several test cases for related bugs, just to check these
in for later work.
We must resolve the type fully so that pointer children (i.e. slices)
are resolved. Additionally, we must resolve even if we can know the
value at comptime because the `alloc_inferred` ZIR always produces a
constant in the AIR.
Fixes#11181
A const local which had its init expression write to the result pointer,
but then gets elided to directly initialize, was missing the coercion to
the type annotation.
* don't store `has_well_defined_layout` in memory.
* remove struct `hasWellDefinedLayout` logic. it's just
`layout != .Auto`. This means we only need one implementation, in
Type.
* fix some of the cases being wrong in `hasWellDefinedLayout`, such as
optional pointers.
* move `tag_ty_inferred` field into a position that makes it more
obvious how the struct layout will be done. Also we don't have a
compiler that intelligently moves fields around so this layout is
better.
* Sema: don't `resolveTypeLayout` in `zirCoerceResultPtr` unless
necessary.
* Rename `ComptimePtrLoadKit` `target` field to `pointee` to avoid
confusion with `target`.
We need to make sure that we bitcast our pointers correctly before
we use get_element_ptr to compute the offset for the parent
pointer.
This also includes a small fix-up for a problem where ptrs to const
i64/u64 were not using the correct type in >1-level decl chains
(where we call lowerParentPtr recursively)
It is possible for the value length to be longer than the type because
we allow in-memory coercing of types such as `[5:0]u8` to `[5]u8`. In
such a case, the value length is 6 but the type length if 5.
The `.repeated` value type already got this right, so this is extending
similar logic out to `.aggregate` and `.bytes`. Both scenarios are
tested in behavior tests.
Fixes#11165
This also surfaces the fact that clz, ctz and popCount didn't actually
support 128 bit integers, despite what was claimed by
226fcd7c709ec664c5d883042cf7beb3026f66cb. This was partially hidden by
the fact that the test code for popCount only exercised 128 bit integers
in a comptime context. This commit duplicates that test case for runtime
ints too.
Implements `@Type` for structs, anon structs, and tuples. This is another place that would probably benefit from a `.reified_struct` type tag but will defer for later in the interest of getting tests passing first.
Detect if we are storing an array operand to a bitcasted vector pointer.
If so, we instead reach through the bitcasted pointer to the vector pointer,
bitcast the array operand to a vector, and then lower this as a store of
a vector value to a vector pointer. This generally results in better code,
as well as working around an LLVM bug.
See #11154
This folds the airCountZeroes() code from
226fcd7c709ec664c5d883042cf7beb3026f66cb back into airBuiltinCall(),
since most of these builtins happen to require the same arguments and
can be unified under a common function signature.
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.
In stage1, this behavior was allowed (by accident?) and also
accidentally exercised by the behavior test changed in this commit. In
discussion on Discord, Andrew decided this should not be allowed in
stage2 since there is currently on real world reason to allow this
strange edge case.
I've added the compiler test to solidify that this behavior should NOT
occur and updated the behavior test to the new valid semantics.
Introduce `Module.ensureFuncBodyAnalyzed` and corresponding `Sema`
function. This mirrors `ensureDeclAnalyzed` except also waits until the
function body has been semantically analyzed, meaning that inferred
error sets will have been populated.
Resolving error sets can now emit a "unable to resolve inferred error
set" error instead of producing an incorrect error set type. Resolving
error sets now calls `ensureFuncBodyAnalyzed`. Closes#11046.
`coerceInMemoryAllowedErrorSets` now does a lot more work to avoid
resolving an inferred error set if possible. Same with
`wrapErrorUnionSet`.
Inferred error set types no longer check the `func` field to determine if
they are equal. That was incorrect because an inline or comptime function
call produces a unique error set which has the same `*Module.Fn` value for
this field. Instead we use the `*Module.Fn.InferredErrorSet` pointers to
test equality of inferred error sets.
