There will be more call sites to `preparePanicId` as we transition away
from safety checks in Sema towards safety checked instructions; it's
silly for them to all have this clunky usage.
This safety check was completely broken; it triggered unchecked illegal
behavior *in order to implement the safety check*. You definitely can't
do that! Instead, we must explicitly check the boundaries. This is a
tiny bit fiddly, because we need to make sure we do floating-point
rounding in the correct direction, and also handle the fact that the
operation truncates so the boundary works differently for min vs max.
Instead of implementing this safety check in Sema, there are now
dedicated AIR instructions for safety-checked intfromfloat (two
instructions; which one is used depends on the float mode). Currently,
no backend directly implements them; instead, a `Legalize.Feature` is
added which expands the safety check, and this feature is enabled for
all backends we currently test, including the LLVM backend.
The `u0` case is still handled in Sema, because Sema needs to check for
that anyway due to the comptime-known result. The old safety check here
was also completely broken and has therefore been rewritten. In that
case, we just check for 'abs(input) < 1.0'.
I've added a bunch of test coverage for the boundary cases of
`@intFromFloat`, both for successes (in `test/behavior/cast.zig`) and
failures (in `test/cases/safety/`).
Resolves: #24161
These conversion routines accept a `round` argument to control how the
result is rounded and return whether the result is exact. Most callers
wanted this functionality and had hacks around it being missing.
Also delete `std.math.big.rational` because it was only being used for
float conversion, and using rationals for that is a lot more complex
than necessary. It also required an allocator, whereas the new integer
routines only need to be passed enough memory to store the result.
If you write an if expression in mem.doNotOptimizeAway like
doNotOptimizeAway(if (ix < 0x00100000) x / 0x1p120 else x + 0x1p120);,
FCSEL instruction is used on AArch64.
FCSEL instruction selects one of the two registers according to
the condition and copies its value.
In this example, `x / 0x1p120` and `x + 0x1p120` are expressions
that raise different floating-point exceptions.
However, since both are actually evaluated before the FCSEL
instruction, the exception not intended by the programmer may
also be raised.
To prevent FCSEL instruction from being used here, this commit
splits doNotOptimizeAway in two.
and also rename `advancedPrint` to `bufferedPrint` in the zig init templates
These are left overs from my previous changes to zig init.
The new templating system removes LITNAME because the new restrictions on package names make it redundant with NAME, and the use of underscores for marking templated identifiers lets us template variable names while still keeping zig fmt happy.
Did you know that allocators reuse addresses? If not, then don't feel
bad, because apparently I don't either! This dumb mistake was probably
responsible for the CI failures on `master` yesterday.
I messed up atomic orderings on this variable because they changed in a
local refactor at some point. We need to always release on the store and
acquire on the loads so that a linker thread observing `.ready` sees the
stored MIR.
Because any `LazyPath` might be resolved to a relative path, it's
incorrect to pass that directly to a child process whose cwd might
differ. Instead, if the child has an overriden cwd, we need to convert
such paths to be relative to the child cwd using `std.fs.path.relative`.
File arguments added to `std.Build.Step.Run` with e.g. `addFileArg` are
not necessarily passed as absolute paths. It used to be the case that
they were as a consequence of an unnecessary path conversion done by the
frontend, but this no longer happens, at least not always, so these
tests were sometimes failing when run locally. Therefore, the standalone
tests must handle cwd-relative CLI paths correctly.
* Sema: allow binary operations and boolean not on vectors of bool
* langref: Clarify use of operators on vectors (`and` and `or` not allowed)
closes#24093
Looking at a compilation of 'test/behavior/x86_64/unary.zig' in
callgrind showed that a full 30% of the compiler runtime was spent in
this `stringToEnum` call, so optimizing it was low-hanging fruit.
We tried replacing it with nested `switch` statements using
`inline else`, but that generated too much code; it didn't emit huge
binaries or anything, but LLVM used a *ridiculous* amount of memory
compiling it in some cases. The core problem here is that only a small
subset of the cases are actually used (the rest fell through to an
"error" path), but that subset is computed at comptime, so we must rely
on the optimizer to eliminate the thousands of redundant cases. This
would be solved by #21507.
Instead, we pre-compute a lookup table at comptime. This table is pretty
big (I guess a couple hundred k?), but only the "valid" subset of
entries will be accessed in practice (unless a bug in the backend is
hit), so it's not too awful on the cache; and it performs much better
than the old `std.meta.stringToEnum` call.
Update the estimated total items for the codegen and link progress nodes
earlier. Rather than waiting for the main thread to dispatch the tasks,
we can add the item to the estimated total as soon as we queue the main
task. The only difference is we need to complete it even in error cases.
Without this cap, unlucky scheduling and/or details of what pipeline
stages perform best on the host machine could cause many gigabytes of
MIR to be stuck in the queue. At a certain point, pause the main thread
until some of the functions in flight have been processed.
This isn't really coherent to model as a `Feature`; this makes sense
because of zig1's specific environment. As such, I opted to check
`dev.env` directly.
Previously, `PerThread.populateTestFunctions` was analyzing the
`test_functions` declaration if it hadn't already been analyzed, so that
it could then populate it. However, the logic for doing this wasn't
actually correct, because it didn't trigger the necessary type
resolution. I could have tried to fix this, but there's actually a
simpler solution! If the `test_functions` declaration isn't referenced
or has a compile error, then we simply don't need to update it; either
it's unreferenced so its value doesn't matter, or we're going to get a
compile error anyway. Either way, we can just give up early. This avoids
doing semantic analysis after `performAllTheWork` finishes.
Also, get rid of the "Code Generation" progress node while updating the
test decl: this is a linking task.
The name of the ZCU object file emitted by the LLVM backend has been
changed in this branch from e.g. `foo.obj` to `foo_zcu.obj`. This is to
avoid name clashes. This commit just updates the stack trace tests which
started failing on windows because of the object name change.
The name of the ZCU object file emitted by the LLVM backend has been
changed in this branch from e.g. `foo.o` to `foo_zcu.o`. This is to
avoid name clashes. This commit just updates a link test which started
failing because the object name in a linker error changed.
glibc, freebsd, and netbsd all do caching manually, because of the fact
that they emit multiple files which they want to cache as a block.
Therefore, the individual sub-compilation on a cache miss should be
using `CacheMode.none` so that we can specify the output paths for each
sub-compilation as being in the shared output directory.
* "Flush" nodes ("LLVM Emit Object", "ELF Flush") appear under "Linking"
* "Code Generation" disappears when all analysis and codegen is done
* We only show one node under "Semantic Analysis" to accurately convey
that analysis isn't happening in parallel, but rather that we're
pausing one task to do another
Previously, various doc comments heavily disagreed with the
implementation on both what lives where on the filesystem at what time,
and how that was represented in code. Notably, the combination of emit
paths outside the cache and `disable_lld_caching` created a kind of
ad-hoc "cache disable" mechanism -- which didn't actually *work* very
well, 'most everything still ended up in this cache. There was also a
long-standing issue where building using the LLVM backend would put a
random object file in your cwd.
This commit reworks how emit paths are specified in
`Compilation.CreateOptions`, how they are represented internally, and
how the cache usage is specified.
There are now 3 options for `Compilation.CacheMode`:
* `.none`: do not use the cache. The paths we have to emit to are
relative to the compiler cwd (they're either user-specified, or
defaults inferred from the root name). If we create any temporary
files (e.g. the ZCU object when using the LLVM backend) they are
emitted to a directory in `local_cache/tmp/`, which is deleted once
the update finishes.
* `.whole`: cache the compilation based on all inputs, including file
contents. All emit paths are computed by the compiler (and will be
stored as relative to the local cache directory); it is a CLI error to
specify an explicit emit path. Artifacts (including temporary files)
are written to a directory under `local_cache/tmp/`, which is later
renamed to an appropriate `local_cache/o/`. The caller (who is using
`--listen`; e.g. the build system) learns the name of this directory,
and can get the artifacts from it.
* `.incremental`: similar to `.whole`, but Zig source file contents, and
anything else which incremental compilation can handle changes for, is
not included in the cache manifest. We don't need to do the dance
where the output directory is initially in `tmp/`, because our digest
is computed entirely from CLI inputs.
To be clear, the difference between `CacheMode.whole` and
`CacheMode.incremental` is unchanged. `CacheMode.none` is new
(previously it was sort of poorly imitated with `CacheMode.whole`). The
defined behavior for temporary/intermediate files is new.
`.none` is used for direct CLI invocations like `zig build-exe foo.zig`.
The other cache modes are reserved for `--listen`, and the cache mode in
use is currently just based on the presence of the `-fincremental` flag.
There are two cases in which `CacheMode.whole` is used despite there
being no `--listen` flag: `zig test` and `zig run`. Unless an explicit
`-femit-bin=xxx` argument is passed on the CLI, these subcommands will
use `CacheMode.whole`, so that they can put the output somewhere without
polluting the cwd (plus, caching is potentially more useful for direct
usage of these subcommands).
Users of `--listen` (such as the build system) can now use
`std.zig.EmitArtifact.cacheName` to find out what an output will be
named. This avoids having to synchronize logic between the compiler and
all users of `--listen`.
