As a side effect this should add raw-dylib support to cg_gcc as the
default ArchiveBuilderBuilder that is used implements
create_dll_import_lib. I haven't tested if the raw-dylib support
actually works however.
Set "symbol name" in raw-dylib import libraries to the decorated name
`windows-rs` received a bug report that mixing raw-dylib generated and the Windows SDK import libraries was causing linker failures: <https://github.com/microsoft/windows-rs/issues/3285>
The root cause turned out to be #124958, that is we are not including the decorated name in the import library and so the import name type is also not being correctly set.
This change modifies the generation of import libraries to set the "symbol name" to the fully decorated name and correctly marks the import as being data vs function.
Note that this also required some changes to how the symbol is named within Rust: for MSVC we now need to use the decorated name but for MinGW we still need to use partially decorated (or undecorated) name.
Fixes#124958
Passing i686 MSVC and MinGW build: <https://github.com/rust-lang/rust/actions/runs/11000433888?pr=130586>
r? `@ChrisDenton`
mark some target features as 'forbidden' so they cannot be (un)set with -Ctarget-feature
The context for this is https://github.com/rust-lang/rust/issues/116344: some target features change the way floats are passed between functions. Changing those target features is unsound as code compiled for the same target may now use different ABIs.
So this introduces a new concept of "forbidden" target features (on top of the existing "stable " and "unstable" categories), and makes it a hard error to (un)set such a target feature. For now, the x86 and ARM feature `soft-float` is on that list. We'll have to make some effort to collect more relevant features, and similar features from other targets, but that can happen after the basic infrastructure for this landed. (These features are being collected in https://github.com/rust-lang/rust/issues/131799.)
I've made this a warning for now to give people some time to speak up if this would break something.
MCP: https://github.com/rust-lang/compiler-team/issues/780
Support clobber_abi and vector registers (clobber-only) in PowerPC inline assembly
This supports `clobber_abi` which is one of the requirements of stabilization mentioned in #93335.
This basically does a similar thing I did in https://github.com/rust-lang/rust/pull/130630 to implement `clobber_abi` for s390x, but for powerpc/powerpc64/powerpc64le.
- This also supports vector registers (as `vreg`) as clobber-only, which need to support clobbering of them to implement `clobber_abi`.
- `vreg` should be able to accept `#[repr(simd)]` types as input/output if the unstable `altivec` target feature is enabled, but `core::arch::{powerpc,powerpc64}` vector types, `#[repr(simd)]`, and `core::simd` are all unstable, so the fact that this is currently a clobber-only should not be considered a blocker of clobber_abi implementation or stabilization. So I have not implemented it in this PR.
- See https://github.com/rust-lang/rust/pull/131551 (which is based on this PR) for a PR to implement this.
- (I'm not sticking to whether that PR should be a separate PR or part of this PR, so I can merge that PR into this PR if needed.)
Refs:
- PPC32 SysV: Section "Function Calling Sequence" in [System V Application Binary Interface PowerPC Processor Supplement](https://refspecs.linuxfoundation.org/elf/elfspec_ppc.pdf)
- PPC64 ELFv1: Section 3.2 "Function Calling Sequence" in [64-bit PowerPC ELF Application Binary Interface Supplement](https://refspecs.linuxfoundation.org/ELF/ppc64/PPC-elf64abi.html#FUNC-CALL)
- PPC64 ELFv2: Section 2.2 "Function Calling Sequence" in [64-Bit ELF V2 ABI Specification](https://openpowerfoundation.org/specifications/64bitelfabi/)
- AIX: [Register usage and conventions](https://www.ibm.com/docs/en/aix/7.3?topic=overview-register-usage-conventions), [Special registers in the PowerPC®](https://www.ibm.com/docs/en/aix/7.3?topic=overview-special-registers-in-powerpc), [AIX vector programming](https://www.ibm.com/docs/en/aix/7.3?topic=concepts-aix-vector-programming)
- Register definition in LLVM: https://github.com/llvm/llvm-project/blob/llvmorg-19.1.0/llvm/lib/Target/PowerPC/PPCRegisterInfo.td#L189
If I understand the above four ABI documentations correctly, except for the PPC32 SysV's VR (Vector Registers) and 32-bit AIX (currently not supported by rustc)'s r13, there does not appear to be important differences in terms of implementing `clobber_abi`:
- The above four ABIs are consistent about FPR (0-13: volatile, 14-31: nonvolatile), CR (0-1,5-7: volatile, 2-4: nonvolatile), XER (volatile), and CTR (volatile).
- As for GPR, only the registers we are treating as reserved are slightly different
- r0, r3-r12 are volatile
- r1(sp, reserved), r14-31 are nonvolatile
- r2(reserved) is TOC pointer in PPC64 ELF/AIX, system-reserved register in PPC32 SysV (AFAIK used as thread pointer in Linux/BSDs)
- r13(reserved for non-32-bit-AIX) is thread pointer in PPC64 ELF, small data area pointer register in PPC32 SysV, "reserved under 64-bit environment; not restored across system calls[^r13]" in AIX)
- As for FPSCR, volatile in PPC64 ELFv1/AIX, some fields are volatile only in certain situations (rest are volatile) in PPC32 SysV/PPC64 ELFv2.
- As for VR (Vector Registers), it is not mentioned in PPC32 SysV, v0-v19 are volatile in both in PPC64 ELF/AIX, v20-v31 are nonvolatile in PPC64 ELF, reserved or nonvolatile depending on the ABI ([vec-extabi vs vec-default in LLVM](https://reviews.llvm.org/D89684), we are [using vec-extabi](https://github.com/rust-lang/rust/pull/131341#discussion_r1797693299)) in AIX:
> When the default Vector enabled mode is used, these registers are reserved and must not be used.
> In the extended ABI vector enabled mode, these registers are nonvolatile and their values are preserved across function calls
I left [FIXME comment about PPC32 SysV](https://github.com/rust-lang/rust/pull/131341#discussion_r1790496095) and added ABI check for AIX.
- As for VRSAVE, it is not mentioned in PPC32 SysV, nonvolatile in PPC64 ELFv1, reserved in PPC64 ELFv2/AIX
- As for VSCR, it is not mentioned in PPC32 SysV/PPC64 ELFv1, some fields are volatile only in certain situations (rest are volatile) in PPC64 ELFv2, volatile in AIX
We are currently treating r1-r2, r13 (non-32-bit-AIX), r29-r31, LR, CTR, and VRSAVE as reserved.
We are currently not processing anything about FPSCR and VSCR, but I feel those are things that should be processed by `preserves_flags` rather than `clobber_abi` if we need to do something about them. (However, PPCRegisterInfo.td in LLVM does not seem to define anything about them.)
Replaces #111335 and #124279
cc `@ecnelises` `@bzEq` `@lu-zero`
r? `@Amanieu`
`@rustbot` label +O-PowerPC +A-inline-assembly
[^r13]: callee-saved, according to [LLVM](6a6af0246b/llvm/lib/Target/PowerPC/PPCCallingConv.td (L322)) and [GCC](a9173a50e7/gcc/config/rs6000/rs6000.h (L859)).
The target name can be anything with custom target specs. Matching on
fields inside the target spec is much more robust than matching on the
target name.
- removed extra bits from predicates queries that are no longer needed in the new system
- removed the need for `non_erasable_generics` to take in tcx and DefId, removed unused arguments in callers
Rename Receiver -> LegacyReceiver
As part of the "arbitrary self types v2" project, we are going to replace the current `Receiver` trait with a new mechanism based on a new, different `Receiver` trait.
This PR renames the old trait to get it out the way. Naming is hard. Options considered included:
* HardCodedReceiver (because it should only be used for things in the standard library, and hence is sort-of hard coded)
* LegacyReceiver
* TargetLessReceiver
* OldReceiver
These are all bad names, but fortunately this will be temporary. Assuming the new mechanism proceeds to stabilization as intended, the legacy trait will be removed altogether.
Although we expect this trait to be used only in the standard library, we suspect it may be in use elsehwere, so we're landing this change separately to identify any surprising breakages.
It's known that this trait is used within the Rust for Linux project; a patch is in progress to remove their dependency.
This is a part of the arbitrary self types v2 project,
https://github.com/rust-lang/rfcs/pull/3519https://github.com/rust-lang/rust/issues/44874
r? `@wesleywiser`
As part of the "arbitrary self types v2" project, we are going to
replace the current `Receiver` trait with a new mechanism based on a
new, different `Receiver` trait.
This PR renames the old trait to get it out the way. Naming is hard.
Options considered included:
* HardCodedReceiver (because it should only be used for things in the
standard library, and hence is sort-of hard coded)
* LegacyReceiver
* TargetLessReceiver
* OldReceiver
These are all bad names, but fortunately this will be temporary.
Assuming the new mechanism proceeds to stabilization as intended, the
legacy trait will be removed altogether.
Although we expect this trait to be used only in the standard library,
we suspect it may be in use elsehwere, so we're landing this change
separately to identify any surprising breakages.
It's known that this trait is used within the Rust for Linux project; a
patch is in progress to remove their dependency.
This is a part of the arbitrary self types v2 project,
https://github.com/rust-lang/rfcs/pull/3519https://github.com/rust-lang/rust/issues/44874
r? @wesleywiser
rust_for_linux: -Zregparm=<N> commandline flag for X86 (#116972)
Command line flag `-Zregparm=<N>` for X86 (32-bit) for rust-for-linux: https://github.com/rust-lang/rust/issues/116972
Implemented in the similar way as fastcall/vectorcall support (args are marked InReg if fit).
Add intrinsics `fmuladd{f16,f32,f64,f128}`. This computes `(a * b) +
c`, to be fused if the code generator determines that (i) the target
instruction set has support for a fused operation, and (ii) that the
fused operation is more efficient than the equivalent, separate pair
of `mul` and `add` instructions.
https://llvm.org/docs/LangRef.html#llvm-fmuladd-intrinsic
MIRI support is included for f32 and f64.
The codegen_cranelift uses the `fma` function from libc, which is a
correct implementation, but without the desired performance semantic. I
think this requires an update to cranelift to expose a suitable
instruction in its IR.
I have not tested with codegen_gcc, but it should behave the same
way (using `fma` from libc).
codegen_ssa: consolidate tied target checks
Fixes#105110.
Fixes#105111.
`rustc_codegen_llvm` and `rustc_codegen_gcc` duplicated logic for checking if tied target features were partially enabled. This PR consolidates these checks into `rustc_codegen_ssa` in the `codegen_fn_attrs` query, which also is run pre-monomorphisation for each function, which ensures that this check is run for unused functions, as would be expected.
Also adds a test confirming that enabling one tied feature doesn't imply another - the appropriate error for this was already being emitted. I did a bisect and narrowed it down to two patches it was likely to be - something in #128796, probably #128221 or #128679.
`rustc_codegen_llvm` and `rustc_codegen_gcc` duplicated logic for
checking if tied target features were partially enabled. This commit
consolidates these checks into `rustc_codegen_ssa` in the
`codegen_fn_attrs` query, which also is run pre-monomorphisation for
each function, which ensures that this check is run for unused functions,
as would be expected.
llvm: replace some deprecated functions
`LLVMMDStringInContext` and `LLVMMDNodeInContext` are deprecated, replace them with `LLVMMDStringInContext2` and `LLVMMDNodeInContext2`.
Also replace `Value` with `Metadata` in some function signatures for better consistency.
It's crazy to have the integer methods in something close to random
order.
The reordering makes the gaps clear: `const_i64`, `const_i128`,
`const_isize`, and `const_u16`. I guess they just aren't needed.
Supertraits of `BuilderMethods` are all called `XyzBuilderMethods`.
Supertraits of `CodegenMethods` are all called `XyzMethods`. This commit
changes the latter to `XyzCodegenMethods`, for consistency.
It has `Backend` and `Deref` boudns, plus an associated type
`CodegenCx`, and it has a single use. This commit "inlines" it into
`BuilderMethods`, which makes the complicated backend trait situation a
little simpler.
Don't leave debug locations for constants sitting on the builder indefinitely
Because constants are currently emitted *before* the prologue, leaving the debug location on the IRBuilder spills onto other instructions in the prologue and messes up both line numbers as well as the point LLVM chooses to be the prologue end.
Example LLVM IR (irrelevant IR elided):
Before:
```
define internal { i64, i64 } `@_ZN3tmp3Foo18var_return_opt_try17he02116165b0fc08cE(ptr` align 8 %self) !dbg !347 { start:
%self.dbg.spill = alloca [8 x i8], align 8
%_0 = alloca [16 x i8], align 8
%residual.dbg.spill = alloca [0 x i8], align 1
#dbg_declare(ptr %residual.dbg.spill, !353, !DIExpression(), !357)
store ptr %self, ptr %self.dbg.spill, align 8, !dbg !357
#dbg_declare(ptr %self.dbg.spill, !350, !DIExpression(), !358)
```
After:
```
define internal { i64, i64 } `@_ZN3tmp3Foo18var_return_opt_try17h00b17d08874ddd90E(ptr` align 8 %self) !dbg !347 { start:
%self.dbg.spill = alloca [8 x i8], align 8
%_0 = alloca [16 x i8], align 8
%residual.dbg.spill = alloca [0 x i8], align 1
#dbg_declare(ptr %residual.dbg.spill, !353, !DIExpression(), !357)
store ptr %self, ptr %self.dbg.spill, align 8
#dbg_declare(ptr %self.dbg.spill, !350, !DIExpression(), !358)
```
Note in particular how !357 from %residual.dbg.spill's dbg_declare no longer falls through onto the store to %self.dbg.spill. This fixes argument values at entry when the constant is a ZST (e.g. `<Option as Try>::Residual`). This fixes#130003 (but note that it does *not* fix issues with argument values and non-ZST constants, which emit their own stores that have debug info on them, like #128945).
r? `@michaelwoerister`
Because constants are currently emitted *before* the prologue, leaving the
debug location on the IRBuilder spills onto other instructions in the prologue
and messes up both line numbers as well as the point LLVM chooses to be the
prologue end.
Example LLVM IR (irrelevant IR elided):
Before:
define internal { i64, i64 } @_ZN3tmp3Foo18var_return_opt_try17he02116165b0fc08cE(ptr align 8 %self) !dbg !347 {
start:
%self.dbg.spill = alloca [8 x i8], align 8
%_0 = alloca [16 x i8], align 8
%residual.dbg.spill = alloca [0 x i8], align 1
#dbg_declare(ptr %residual.dbg.spill, !353, !DIExpression(), !357)
store ptr %self, ptr %self.dbg.spill, align 8, !dbg !357
#dbg_declare(ptr %self.dbg.spill, !350, !DIExpression(), !358)
After:
define internal { i64, i64 } @_ZN3tmp3Foo18var_return_opt_try17h00b17d08874ddd90E(ptr align 8 %self) !dbg !347 {
start:
%self.dbg.spill = alloca [8 x i8], align 8
%_0 = alloca [16 x i8], align 8
%residual.dbg.spill = alloca [0 x i8], align 1
#dbg_declare(ptr %residual.dbg.spill, !353, !DIExpression(), !357)
store ptr %self, ptr %self.dbg.spill, align 8
#dbg_declare(ptr %self.dbg.spill, !350, !DIExpression(), !358)
Note in particular how !357 from %residual.dbg.spill's dbg_declare no longer
falls through onto the store to %self.dbg.spill. This fixes argument values
at entry when the constant is a ZST (e.g. <Option as Try>::Residual). This
fixes#130003 (but note that it does *not* fix issues with argument values and
non-ZST constants, which emit their own stores that have debug info on them,
like #128945).
This shrinks `compiler/rustc_codegen_gcc/Cargo.lock` quite a bit. The
only remaining dependencies in `compiler/rustc_codegen_gcc/Cargo.lock`
are `gccjit`, `lang_tester`, and `boml`, all of which aren't used in any
other compiler crates.
The commit also reorders and adds comments to the `extern crate` items
so they match those in miri.
simd_shuffle intrinsic: allow argument to be passed as vector
See https://github.com/rust-lang/rust/issues/128738 for context.
I'd like to get rid of [this hack](6c0b89dfac/compiler/rustc_codegen_ssa/src/mir/block.rs (L922-L935)). https://github.com/rust-lang/rust/pull/128537 almost lets us do that since constant SIMD vectors will then be passed as immediate arguments. However, simd_shuffle for some reason actually takes an *array* as argument, not a vector, so the hack is still required to ensure that the array becomes an immediate (which then later stages of codegen convert into a vector, as that's what LLVM needs).
This PR prepares simd_shuffle to also support a vector as the `idx` argument. Once this lands, stdarch can hopefully be updated to pass `idx` as a vector, and then support for arrays can be removed, which finally lets us get rid of that hack.
Shrink `TyKind::FnPtr`.
By splitting the `FnSig` within `TyKind::FnPtr` into `FnSigTys` and `FnHeader`, which can be packed more efficiently. This reduces the size of the hot `TyKind` type from 32 bytes to 24 bytes on 64-bit platforms. This reduces peak memory usage by a few percent on some benchmarks. It also reduces cache misses and page faults similarly, though this doesn't translate to clear cycles or wall-time improvements on CI.
r? `@compiler-errors`
bootstrap: don't use rustflags for `--rustc-args`
r? `@onur-ozkan`
This is going to require a bit of context.
https://github.com/rust-lang/rust/pull/47558 has added `--rustc-args` to `./x test` to allow passing flags when building `compiletest` tests. It was made specifically because using `RUSTFLAGS` would rebuild the compiler/stdlib, which would in turn require the flag you want to build tests with to successfully bootstrap.
#113178 made the request that it also works for other tests and doctests. This is not trivial to support across the board for `library`/`compiler` unit-tests/doctests and across stages. This issue was closed in #113948 by using `RUSTFLAGS`, seemingly incorrectly since https://github.com/rust-lang/rust/pull/123489 fixed that part to make it work.
Unfortunately #123489/#113948 have regressed the goals of `--rustc-args`:
- now we can't use rustc args that don't bootstrap, to run the UI tests: we can't test incomplete features. The new trait solver doesn't bootstrap, in-progress borrowck/polonius changes don't bootstrap, some other features are similarly incomplete, etc.
- using the flag now rebuilds everything from scratch: stage0 stdlib, stage1 compiler, stage1 stdlib. You don't need to re-do all this to compile UI tests, you only need the latter to run stdlib tests with a new flag, etc. This happens for contributors, but also on CI today. (Not to mention that in doing that it will rebuild things with flags that are not meant to be used, e.g. stdlib cfgs that don't exist in the compiler; or you could also imagine that this silently enables flags that were not meant to be enabled in this way).
Since then, bd71c71ea0 has started using it to test a stdlib feature, relying on the fact that it now rebuilds everything. So #125011 also regressed CI times more than necessary because it rebuilds everything instead of just stage 1 stdlib.
It's not easy for me to know how to properly fix#113178 in bootstrap, but #113948/#123489 are not it since they regress the initial intent. I'd think bootstrap would have to know from the list of test targets that are passed that the `library` or `compiler` paths that are passed could require rebuilding these crates with different rustflags, probably also depending on stages. Therefore I would not be able to fix it, and will just try in this PR to unregress the situation to unblock the initial use-case.
It seems miri now also uses `./x miri --rustc-args` in this incorrect meaning to rebuild the `library` paths they support to run with the new args. I've not made any bootstrap changes related to `./x miri` in this PR, so `--rustc-args` wouldn't work there anymore. I'd assume this would need to use rustflags again but I don't know how to make that work properly in bootstrap, hence opening as draft, so you can tell me how to do that. I assume we don't want to break their use-case again now that it exists, even though there are ways to use `./x test` to do exactly that.
`RUSTFLAGS_NOT_BOOTSTRAP=flag ./x test library/std` is a way to run unit tests with a new flag without rebuilding everything, while with #123489 there is no way anymore to run tests with a flag that doesn't bootstrap.
---
edit: after review, this PR:
- renames `./x test --rustc-args` to `./x test --compiletest-rustc-args` as it only applies there, and cannot use rustflags for this purpose.
- fixes the regression that using these args rebuilt everything from scratch
- speeds up some CI jobs via the above point
- removes `./x miri --rustc-args` as only library tests are supported, needs to rebuild libstd, and `./x miri --compiletest-rustc-args` wouldn't work since compiletests are not supported.
const vector passed through to codegen
This allows constant vectors using a repr(simd) type to be propagated
through to the backend by reusing the functionality used to do a similar
thing for the simd_shuffle intrinsic
#118209
r? RalfJung
nontemporal_store: make sure that the intrinsic is truly just a hint
The `!nontemporal` flag for stores in LLVM *sounds* like it is just a hint, but actually, it is not -- at least on x86, non-temporal stores need very special treatment by the programmer or else the Rust memory model breaks down. LLVM still treats these stores as-if they were normal stores for optimizations, which is [highly dubious](https://github.com/llvm/llvm-project/issues/64521). Let's avoid all that dubiousness by making our own non-temporal stores be truly just a hint, which is possible on some targets (e.g. ARM). On all other targets, non-temporal stores become regular stores.
~~Blocked on https://github.com/rust-lang/stdarch/pull/1541 propagating to the rustc repo, to make sure the `_mm_stream` intrinsics are unaffected by this change.~~
Fixes https://github.com/rust-lang/rust/issues/114582
Cc `@Amanieu` `@workingjubilee`
By splitting the `FnSig` within `TyKind::FnPtr` into `FnSigTys` and
`FnHeader`, which can be packed more efficiently. This reduces the size
of the hot `TyKind` type from 32 bytes to 24 bytes on 64-bit platforms.
This reduces peak memory usage by a few percent on some benchmarks. It
also reduces cache misses and page faults similarly, though this doesn't
translate to clear cycles or wall-time improvements on CI.
Update compiler_builtins to 0.1.114
The `weak-intrinsics` feature was removed from compiler_builtins in https://github.com/rust-lang/compiler-builtins/pull/598, so dropped the `compiler-builtins-weak-intrinsics` feature from alloc/std/sysroot.
In https://github.com/rust-lang/compiler-builtins/pull/593, some builtins for f16/f128 were added. These don't work for all compiler backends, so add a `compiler-builtins-no-f16-f128` feature and disable it for cranelift and gcc.
The `weak-intrinsics` feature was removed from compiler_builtins in
https://github.com/rust-lang/compiler-builtins/pull/598, so dropped the
`compiler-builtins-weak-intrinsics` feature from alloc/std/sysroot.
In https://github.com/rust-lang/compiler-builtins/pull/593, some
builtins for f16/f128 were added. These don't work for all compiler
backends, so add a `compiler-builtins-no-f16-f128` feature and disable
it for cranelift and gcc. Also disable it for LLVM targets that don't
support it.
Sync ar_archive_writer to LLVM 18.1.3
From LLVM 15.0.0-rc3. This adds support for COFF archives containing Arm64EC object files and has various fixes for AIX big archive files.
Miri function identity hack: account for possible inlining
Having a non-lifetime generic is not the only reason a function can be duplicated. Another possibility is that the function may be eligible for cross-crate inlining. So also take into account the inlining attribute in this Miri hack for function pointer identity.
That said, `cross_crate_inlinable` will still sometimes return true even for `inline(never)` functions:
- when they are `DefKind::Ctor(..) | DefKind::Closure` -- I assume those cannot be `InlineAttr::Never` anyway?
- when `cross_crate_inline_threshold == InliningThreshold::Always`
so maybe this is still not quite the right criterion to use for function pointer identity.
Most modules have such a blank line, but some don't. Inserting the blank
line makes it clearer that the `//!` comments are describing the entire
module, rather than the `use` declaration(s) that immediately follows.
rustc_codegen_llvm: add support for writing summary bitcode
Typical uses of ThinLTO don't have any use for this as a standalone file, but distributed ThinLTO uses this to make the linker phase more efficient. With clang you'd do something like `clang -flto=thin -fthin-link-bitcode=foo.indexing.o -c foo.c` and then get both foo.o (full of bitcode) and foo.indexing.o (just the summary or index part of the bitcode). That's then usable by a two-stage linking process that's more friendly to distributed build systems like bazel, which is why I'm working on this area.
I talked some to `@teresajohnson` about naming in this area, as things seem to be a little confused between various blog posts and build systems. "bitcode index" and "bitcode summary" tend to be a little too ambiguous, and she tends to use "thin link bitcode" and "minimized bitcode" (which matches the descriptions in LLVM). Since the clang option is thin-link-bitcode, I went with that to try and not add a new spelling in the world.
Per `@dtolnay,` you can work around the lack of this by using `lld --thinlto-index-only` to do the indexing on regular .o files of bitcode, but that is a bit wasteful on actions when we already have all the information in rustc and could just write out the matching minimized bitcode. I didn't test that at all in our infrastructure, because by the time I learned that I already had this patch largely written.
Typical uses of ThinLTO don't have any use for this as a standalone
file, but distributed ThinLTO uses this to make the linker phase more
efficient. With clang you'd do something like `clang -flto=thin
-fthin-link-bitcode=foo.indexing.o -c foo.c` and then get both foo.o
(full of bitcode) and foo.indexing.o (just the summary or index part of
the bitcode). That's then usable by a two-stage linking process that's
more friendly to distributed build systems like bazel, which is why I'm
working on this area.
I talked some to @teresajohnson about naming in this area, as things
seem to be a little confused between various blog posts and build
systems. "bitcode index" and "bitcode summary" tend to be a little too
ambiguous, and she tends to use "thin link bitcode" and "minimized
bitcode" (which matches the descriptions in LLVM). Since the clang
option is thin-link-bitcode, I went with that to try and not add a new
spelling in the world.
Per @dtolnay, you can work around the lack of this by using `lld
--thinlto-index-only` to do the indexing on regular .o files of
bitcode, but that is a bit wasteful on actions when we already have all
the information in rustc and could just write out the matching minimized
bitcode. I didn't test that at all in our infrastructure, because by the
time I learned that I already had this patch largely written.
Rollup of 5 pull requests
Successful merges:
- #124615 (coverage: Further simplify extraction of mapping info from MIR)
- #124778 (Fix parse error message for meta items)
- #124797 (Refactor float `Primitive`s to a separate `Float` type)
- #124888 (Migrate `run-make/rustdoc-output-path` to rmake)
- #124957 (Make `Ty::builtin_deref` just return a `Ty`)
r? `@ghost`
`@rustbot` modify labels: rollup
Refactor float `Primitive`s to a separate `Float` type
Now there are 4 of them, it makes sense to refactor `F16`, `F32`, `F64` and `F128` out of `Primitive` and into a separate `Float` type (like integers already are). This allows patterns like `F16 | F32 | F64 | F128` to be simplified into `Float(_)`, and is consistent with `ty::FloatTy`.
As a side effect, this PR also makes the `Ty::primitive_size` method work with `f16` and `f128`.
Tracking issue: #116909
`@rustbot` label +F-f16_and_f128
Stabilize the size of incr comp object file names
The current implementation does not produce stable-length paths, and we create the paths in a way that makes our allocation behavior is nondeterministic. I think `@eddyb` fixed a number of other cases like this in the past, and this PR fixes another one. Whether that actually matters I have no idea, but we still have bimodal behavior in rustc-perf and the non-uniformity in `find` and `ls` was bothering me.
I've also removed the truncation of the mangled CGU names. Before this PR incr comp paths look like this:
```
target/debug/incremental/scratch-38izrrq90cex7/s-gux6gz0ow8-1ph76gg-ewe1xj434l26w9up5bedsojpd/261xgo1oqnd90ry5.o
```
And after, they look like this:
```
target/debug/incremental/scratch-035omutqbfkbw/s-gux6borni0-16r3v1j-6n64tmwqzchtgqzwwim5amuga/55v2re42sztc8je9bva6g8ft3.o
```
On the one hand, I'm sure this will break some people's builds because they're on Windows and only a few bytes from the path length limit. But if we're that seriously worried about the length of our file names, I have some other ideas on how to make them smaller. And last time I deleted some hash truncations from the compiler, there was a huge drop in the number if incremental compilation ICEs that were reported: https://github.com/rust-lang/rust/pull/110367https://github.com/rust-lang/rust/pull/110367
---
Upon further reading, this PR actually fixes a bug. This comment says the CGU names are supposed to be a fixed-length hash, and before this PR they aren't: ca7d34efa9/compiler/rustc_monomorphize/src/partitioning.rs (L445-L448)
Use `tcx.types.unit` instead of `Ty::new_unit(tcx)`
I don't think there is any need for the function, given that we can just access the `.types`, similarly to all other primitives?
Stop using LLVM struct types for alloca
The alloca type has no semantic meaning, only the size (and alignment, but we specify it explicitly) matter. Using `[N x i8]` is a more direct way to specify that we want `N` bytes, and avoids relying on LLVM's struct layout. It is likely that a future LLVM version will change to an untyped alloca representation.
Split out from #121577.
r? `@ghost`
Dellvmize some intrinsics (use `u32` instead of `Self` in some integer intrinsics)
This implements https://github.com/rust-lang/compiler-team/issues/693 minus what was implemented in #123226.
Note: I decided to _not_ change `shl`/... builder methods, as it just doesn't seem worth it.
r? ``@scottmcm``
Introduce perma-unstable `wasm-c-abi` flag
Now that `wasm-bindgen` v0.2.88 supports the spec-compliant C ABI, the idea is to switch to that in a future version of Rust. In the meantime it would be good to let people test and play around with it.
This PR introduces a new perma-unstable `-Zwasm-c-abi` compiler flag, which switches to the new spec-compliant C ABI when targeting `wasm32-unknown-unknown`.
Alternatively, we could also stabilize this and then deprecate it when we switch. I will leave this to the Rust maintainers to decide.
This is a companion PR to #117918, but they could be merged independently.
MCP: https://github.com/rust-lang/compiler-team/issues/703
Tracking issue: https://github.com/rust-lang/rust/issues/122532
Add `Ord::cmp` for primitives as a `BinOp` in MIR
Update: most of this OP was written months ago. See https://github.com/rust-lang/rust/pull/118310#issuecomment-2016940014 below for where we got to recently that made it ready for review.
---
There are dozens of reasonable ways to implement `Ord::cmp` for integers using comparison, bit-ops, and branches. Those differences are irrelevant at the rust level, however, so we can make things better by adding `BinOp::Cmp` at the MIR level:
1. Exactly how to implement it is left up to the backends, so LLVM can use whatever pattern its optimizer best recognizes and cranelift can use whichever pattern codegens the fastest.
2. By not inlining those details for every use of `cmp`, we drastically reduce the amount of MIR generated for `derive`d `PartialOrd`, while also making it more amenable to MIR-level optimizations.
Having extremely careful `if` ordering to μoptimize resource usage on broadwell (#63767) is great, but it really feels to me like libcore is the wrong place to put that logic. Similarly, using subtraction [tricks](https://graphics.stanford.edu/~seander/bithacks.html#CopyIntegerSign) (#105840) is arguably even nicer, but depends on the optimizer understanding it (https://github.com/llvm/llvm-project/issues/73417) to be practical. Or maybe [bitor is better than add](https://discourse.llvm.org/t/representing-in-ir/67369/2?u=scottmcm)? But maybe only on a future version that [has `or disjoint` support](https://discourse.llvm.org/t/rfc-add-or-disjoint-flag/75036?u=scottmcm)? And just because one of those forms happens to be good for LLVM, there's no guarantee that it'd be the same form that GCC or Cranelift would rather see -- especially given their very different optimizers. Not to mention that if LLVM gets a spaceship intrinsic -- [which it should](https://rust-lang.zulipchat.com/#narrow/stream/131828-t-compiler/topic/Suboptimal.20inlining.20in.20std.20function.20.60binary_search.60/near/404250586) -- we'll need at least a rustc intrinsic to be able to call it.
As for simplifying it in Rust, we now regularly inline `{integer}::partial_cmp`, but it's quite a large amount of IR. The best way to see that is with 8811efa88b (diff-d134c32d028fbe2bf835fef2df9aca9d13332dd82284ff21ee7ebf717bfa4765R113) -- I added a new pre-codegen MIR test for a simple 3-tuple struct, and this PR change it from 36 locals and 26 basic blocks down to 24 locals and 8 basic blocks. Even better, as soon as the construct-`Some`-then-match-it-in-same-BB noise is cleaned up, this'll expose the `Cmp == 0` branches clearly in MIR, so that an InstCombine (#105808) can simplify that to just a `BinOp::Eq` and thus fix some of our generated code perf issues. (Tracking that through today's `if a < b { Less } else if a == b { Equal } else { Greater }` would be *much* harder.)
---
r? `@ghost`
But first I should check that perf is ok with this
~~...and my true nemesis, tidy.~~
Codegen const panic messages as function calls
This skips emitting extra arguments at every callsite (of which there
can be many). For a librustc_driver build with overflow checks enabled,
this cuts 0.7MB from the resulting shared library (see [perf]).
A sample improvement from nightly:
```
leaq str.0(%rip), %rdi
leaq .Lalloc_d6aeb8e2aa19de39a7f0e861c998af13(%rip), %rdx
movl $25, %esi
callq *_ZN4core9panicking5panic17h17cabb89c5bcc999E@GOTPCREL(%rip)
```
to this PR:
```
leaq .Lalloc_d6aeb8e2aa19de39a7f0e861c998af13(%rip), %rdi
callq *_RNvNtNtCsduqIKoij8JB_4core9panicking11panic_const23panic_const_div_by_zero@GOTPCREL(%rip)
```
[perf]: https://perf.rust-lang.org/compare.html?start=a7e4de13c1785819f4d61da41f6704ed69d5f203&end=64fbb4f0b2d621ff46d559d1e9f5ad89a8d7789b&stat=instructions:u
Unbox and unwrap the contents of `StatementKind::Coverage`
The payload of coverage statements was historically a structure with several fields, so it was boxed to avoid bloating `StatementKind`.
Now that the payload is a single relatively-small enum, we can replace `Box<Coverage>` with just `CoverageKind`.
This patch also adds a size assertion for `StatementKind`, to avoid accidentally bloating it in the future.
``@rustbot`` label +A-code-coverage
We already use `Instance` at declaration sites when available to glean
additional information about possible abstractions of the type in use.
This does the same when possible at callsites as well.
The primary purpose of this change is to allow CFI to alter how it
generates type information for indirect calls through `Virtual`
instances.
The payload of coverage statements was historically a structure with several
fields, so it was boxed to avoid bloating `StatementKind`.
Now that the payload is a single relatively-small enum, we can replace
`Box<Coverage>` with just `CoverageKind`.
This patch also adds a size assertion for `StatementKind`, to avoid
accidentally bloating it in the future.
This skips emitting extra arguments at every callsite (of which there
can be many). For a librustc_driver build with overflow checks enabled,
this cuts 0.7MB from the resulting binary.
Stabilize associated type bounds (RFC 2289)
This PR stabilizes associated type bounds, which were laid out in [RFC 2289]. This gives us a shorthand to express nested type bounds that would otherwise need to be expressed with nested `impl Trait` or broken into several `where` clauses.
### What are we stabilizing?
We're stabilizing the associated item bounds syntax, which allows us to put bounds in associated type position within other bounds, i.e. `T: Trait<Assoc: Bounds...>`. See [RFC 2289] for motivation.
In all position, the associated type bound syntax expands into a set of two (or more) bounds, and never anything else (see "How does this differ[...]" section for more info).
Associated type bounds are stabilized in four positions:
* **`where` clauses (and APIT)** - This is equivalent to breaking up the bound into two (or more) `where` clauses. For example, `where T: Trait<Assoc: Bound>` is equivalent to `where T: Trait, <T as Trait>::Assoc: Bound`.
* **Supertraits** - Similar to above, `trait CopyIterator: Iterator<Item: Copy> {}`. This is almost equivalent to breaking up the bound into two (or more) `where` clauses; however, the bound on the associated item is implied whenever the trait is used. See #112573/#112629.
* **Associated type item bounds** - This allows constraining the *nested* rigid projections that are associated with a trait's associated types. e.g. `trait Trait { type Assoc: Trait2<Assoc2: Copy>; }`.
* **opaque item bounds (RPIT, TAIT)** - This allows constraining associated types that are associated with the opaque without having to *name* the opaque. For example, `impl Iterator<Item: Copy>` defines an iterator whose item is `Copy` without having to actually name that item bound.
The latter three are not expressible in surface Rust (though for associated type item bounds, this will change in #120752, which I don't believe should block this PR), so this does represent a slight expansion of what can be expressed in trait bounds.
### How does this differ from the RFC?
Compared to the RFC, the current implementation *always* desugars associated type bounds to sets of `ty::Clause`s internally. Specifically, it does *not* introduce a position-dependent desugaring as laid out in [RFC 2289], and in particular:
* It does *not* desugar to anonymous associated items in associated type item bounds.
* It does *not* desugar to nested RPITs in RPIT bounds, nor nested TAITs in TAIT bounds.
This position-dependent desugaring laid out in the RFC existed simply to side-step limitations of the trait solver, which have mostly been fixed in #120584. The desugaring laid out in the RFC also added unnecessary complication to the design of the feature, and introduces its own limitations to, for example:
* Conditionally lowering to nested `impl Trait` in certain positions such as RPIT and TAIT means that we inherit the limitations of RPIT/TAIT, namely lack of support for higher-ranked opaque inference. See this code example: https://github.com/rust-lang/rust/pull/120752#issuecomment-1979412531.
* Introducing anonymous associated types makes traits no longer object safe, since anonymous associated types are not nameable, and all associated types must be named in `dyn` types.
This last point motivates why this PR is *not* stabilizing support for associated type bounds in `dyn` types, e.g, `dyn Assoc<Item: Bound>`. Why? Because `dyn` types need to have *concrete* types for all associated items, this would necessitate a distinct lowering for associated type bounds, which seems both complicated and unnecessary compared to just requiring the user to write `impl Trait` themselves. See #120719.
### Implementation history:
Limited to the significant behavioral changes and fixes and relevant PRs, ping me if I left something out--
* #57428
* #108063
* #110512
* #112629
* #120719
* #120584Closes#52662
[RFC 2289]: https://rust-lang.github.io/rfcs/2289-associated-type-bounds.html
Expose the Freeze trait again (unstably) and forbid implementing it manually
non-emoji version of https://github.com/rust-lang/rust/pull/121501
cc #60715
This trait is useful for generic constants (associated consts of generic traits). See the test (`tests/ui/associated-consts/freeze.rs`) added in this PR for a usage example. The builtin `Freeze` trait is the only way to do it, users cannot work around this issue.
It's also a useful trait for building some very specific abstrations, as shown by the usage by the `zerocopy` crate: https://github.com/google/zerocopy/issues/941
cc ```@RalfJung```
T-lang signed off on reexposing this unstably: https://github.com/rust-lang/rust/pull/121501#issuecomment-1969827742
Add asm goto support to `asm!`
Tracking issue: #119364
This PR implements asm-goto support, using the syntax described in "future possibilities" section of [RFC2873](https://rust-lang.github.io/rfcs/2873-inline-asm.html#asm-goto).
Currently I have only implemented the `label` part, not the `fallthrough` part (i.e. fallthrough is implicit). This doesn't reduce the expressive though, since you can use label-break to get arbitrary control flow or simply set a value and rely on jump threading optimisation to get the desired control flow. I can add that later if deemed necessary.
r? ``@Amanieu``
cc ``@ojeda``
