Commit Graph

243 Commits

Author SHA1 Message Date
DianQK
cc08749df2
Add the no-builtins attribute to functions when no_builtins is applied at the crate level.
When `no_builtins` is applied at the crate level, we should add the
`no-builtins` attribute to each function to ensure it takes effect in LTO.
2023-07-18 22:15:47 +08:00
bors
7a17f577b3 Auto merge of #112157 - erikdesjardins:align, r=nikic
Resurrect: rustc_target: Add alignment to indirectly-passed by-value types, correcting the alignment of byval on x86 in the process.

Same as #111551, which I [accidentally closed](https://github.com/rust-lang/rust/pull/111551#issuecomment-1571222612) :/

---

This resurrects PR #103830, which has sat idle for a while.

Beyond #103830, this also:
- fixes byval alignment for types containing vectors on Darwin (see `tests/codegen/align-byval-vector.rs`)
- fixes byval alignment for overaligned types on x86 Windows (see `tests/codegen/align-byval.rs`)
- fixes ABI for types with 128bit requested alignment on ARM64 Linux (see `tests/codegen/aarch64-struct-align-128.rs`)

r? `@nikic`

---

`@pcwalton's` original PR description is reproduced below:

Commit 88e4d2c from five years ago removed
support for alignment on indirectly-passed arguments because of problems with
the `i686-pc-windows-msvc` target. Unfortunately, the `memcpy` optimizations I
recently added to LLVM 16 depend on this to forward `memcpy`s. This commit
attempts to fix the problems with `byval` parameters on that target and now
correctly adds the `align` attribute.

The problem is summarized in [this comment] by `@eddyb.` Briefly, 32-bit x86 has
special alignment rules for `byval` parameters: for the most part, their
alignment is forced to 4. This is not well-documented anywhere but in the Clang
source. I looked at the logic in Clang `TargetInfo.cpp` and tried to replicate
it here. The relevant methods in that file are
`X86_32ABIInfo::getIndirectResult()` and
`X86_32ABIInfo::getTypeStackAlignInBytes()`. The `align` parameter attribute
for `byval` parameters in LLVM must match the platform ABI, or miscompilations
will occur. Note that this doesn't use the approach suggested by eddyb, because
I felt it was overkill to store the alignment in `on_stack` when special
handling is really only needed for 32-bit x86.

As a side effect, this should fix #80127, because it will make the `align`
parameter attribute for `byval` parameters match the platform ABI on LLVM
x86-64.

[this comment]: #80822 (comment)
2023-07-15 15:39:53 +00:00
Erik Desjardins
2daacf5af9 i686-windows: make requested alignment > 4 special case apply transitively 2023-07-14 17:48:13 -04:00
bors
079e544174 Auto merge of #109025 - cjgillot:refprop-dbg, r=JakobDegen
Enable MIR reference propagation by default
2023-07-14 17:32:59 +00:00
Camille GILLOT
f993c6d73e Bless codegen test. 2023-07-14 12:37:29 +00:00
Jubilee Young
7dc049c378 Reenable all cases of simd-wide-sum 2023-07-11 20:21:32 -07:00
Jubilee Young
571aac9fba Add mir-opt3 rev to simd-wide-sum test 2023-07-11 19:07:39 -07:00
Erik Desjardins
d1e764cb3b aarch64-linux: properly handle 128bit aligned aggregates 2023-07-10 19:19:40 -04:00
Erik Desjardins
7e933b4e26 repr(align) <= 4 should still be byval 2023-07-10 19:19:40 -04:00
Erik Desjardins
f704396c0e align-byval test: add cases for lower requested alignment, wrapped, and repr(transparent) 2023-07-10 19:19:39 -04:00
Erik Desjardins
0e76446a9f ensure byval allocas are sufficiently aligned 2023-07-10 19:19:38 -04:00
Erik Desjardins
209ed071ba align-byval test: add cases for <= align 4 2023-07-10 19:19:38 -04:00
Erik Desjardins
8ec90f6f14 align-byval test: add cases distinguishing natural vs forced/requested alignment 2023-07-10 19:19:37 -04:00
Erik Desjardins
08d18929fb align-byval test: add x86
x86 Windows also should not use byval since the struct is
overaligned, see https://reviews.llvm.org/D72114
2023-07-10 19:19:37 -04:00
Erik Desjardins
fdaaf86cc0 add align attr to addr-of-mutate test 2023-07-10 19:19:36 -04:00
Erik Desjardins
a07eb0abbd implement vector-containing aggregate alignment for x86 darwin 2023-07-10 19:19:36 -04:00
Erik Desjardins
be1d4e3e0b update array-map test for removed alloca 2023-07-10 19:19:35 -04:00
Erik Desjardins
102292655b align-byval test: use revisions to test different targets 2023-07-10 19:19:35 -04:00
Patrick Walton
0becc89d4a rustc_target: Add alignment to indirectly-passed by-value types, correcting the
alignment of `byval` on x86 in the process.

Commit 88e4d2c291 from five years ago removed
support for alignment on indirectly-passed arguments because of problems with
the `i686-pc-windows-msvc` target. Unfortunately, the `memcpy` optimizations I
recently added to LLVM 16 depend on this to forward `memcpy`s. This commit
attempts to fix the problems with `byval` parameters on that target and now
correctly adds the `align` attribute.

The problem is summarized in [this comment] by @eddyb. Briefly, 32-bit x86 has
special alignment rules for `byval` parameters: for the most part, their
alignment is forced to 4. This is not well-documented anywhere but in the Clang
source. I looked at the logic in Clang `TargetInfo.cpp` and tried to replicate
it here. The relevant methods in that file are
`X86_32ABIInfo::getIndirectResult()` and
`X86_32ABIInfo::getTypeStackAlignInBytes()`. The `align` parameter attribute
for `byval` parameters in LLVM must match the platform ABI, or miscompilations
will occur. Note that this doesn't use the approach suggested by eddyb, because
I felt it was overkill to store the alignment in `on_stack` when special
handling is really only needed for 32-bit x86.

As a side effect, this should fix #80127, because it will make the `align`
parameter attribute for `byval` parameters match the platform ABI on LLVM
x86-64.

[this comment]: https://github.com/rust-lang/rust/pull/80822#issuecomment-829985417
2023-07-10 19:19:30 -04:00
Camille GILLOT
4de2d8fb66 Perform reference propagation earlier. 2023-07-10 16:01:18 +00:00
Camille GILLOT
341e0f78d1 Fortify codegen test. 2023-07-10 16:01:18 +00:00
Matthias Krüger
6a20f681d5
Rollup merge of #113247 - mirkootter:test-wasm-exceptions-nostd, r=Mark-Simulacrum
Add Tests for native wasm exceptions

### Motivation
In PR #111322, I added support for native WASM exceptions. I was asked by ``@davidtwco`` to add some tests for it in a follow up PR, which seems like a very good idea.

This PR adds three tests for this feature:
* codegen: ensure the correct LLVM instructions are used
* assembly: ensure the correct WASM instructions are used
* run-make: ensure the exception handling works; the WASM code is run using a small nodejs script which demonstrates the exception handling

### Complications
There are a few changes beside adding the tests, which were necessary
* Tests for the wasm32-unknown-unknown target are (as far as I know) only run on `test-various`. Its docker image uses nodejs-15, which is very old. Experimental support for wasm-exceptions was added in nodejs16. In nodejs 18.12 (LTS), they are stable.
  - --> increase nodejs to 18.12 in `test-various`
* codegen/assembly tests are not performed for the wasm32-unknown-unknown target yet
  - --> add those to `test-various` as well

Due to the last point, some tests are run which have not run before (assembly+codegen tests for wasm32-unknown-unknown). I added `// ignore wasm32-bare` for those which failed

### Local testing
I run all tests locally using both `test-various` and `wasm32`. As far as I know, none of the other systems run any test for wasm32 targets.
2023-07-09 16:33:36 +02:00
Camille GILLOT
d7983a2f23 Always name the return place. 2023-07-08 15:38:40 +02:00
Jan-Mirko Otter
a0bd381052 test-various: run codegen and assembly tests 2023-07-02 02:22:22 +02:00
Jan-Mirko Otter
db76154817 add codegen-test: wasm-exceptions 2023-07-01 20:24:15 +02:00
bors
3c554f5cb4 Auto merge of #112516 - erikdesjardins:loop, r=davidtwco
cg_llvm: use index-based loop in write_operand_repeatedly

This should be easier for LLVM to analyze.

Fixes #111603

This needs a perf run.

[cc](https://github.com/rust-lang/rust/issues/111603#issuecomment-1567531178) `@caojoshua`
2023-06-27 15:01:56 +00:00
bors
1d67eba687 Auto merge of #112891 - oli-obk:impl_trait_in_assoc_tys_cleanup, r=compiler-errors
Various impl trait in assoc tys cleanups

r? `@compiler-errors`

All commits except for the last are pure refactorings. 274dab5bd658c97886a8987340bf50ae57900c39 allows struct fields to participate in deciding whether a function has an opaque in its signature.

best reviewed commit by commit
2023-06-23 23:26:38 +00:00
bors
22e9fe644e Auto merge of #112974 - matthiaskrgr:rollup-hnk7ans, r=matthiaskrgr
Rollup of 8 pull requests

Successful merges:

 - #112616 (Improve tests on targets without unwinding)
 - #112643 (Always register sized obligation for argument)
 - #112740 (Add link to rustdoc book search chapter in help popover)
 - #112810 (Don't ICE on unnormalized struct tail in layout computation)
 - #112870 (Migrate `item_bounds` to `ty::Clause`)
 - #112925 (Stop hiding const eval limit in external macros)
 - #112960 ([tests/rustdoc] Add `@files` command)
 - #112962 (Fix rustdoc gui tester)

r? `@ghost`
`@rustbot` modify labels: rollup
2023-06-23 20:45:23 +00:00
Matthias Krüger
ff596144be
Rollup merge of #112616 - ferrocene:pa-more-test-suite-fixes, r=Nilstrieb
Improve tests on targets without unwinding

This PR makes more miscellaneous changes to tests, to make it work on targets without unwinding support.
2023-06-23 19:39:57 +02:00
Augie Fackler
52d50fba2a tests: be even more permissive on attributes in one test 2023-06-23 09:48:00 -04:00
Augie Fackler
d94d17c0bb tests: be more permissive on attributes in one test 2023-06-22 18:03:23 -04:00
Oli Scherer
d6e1b20623 Fix a codegen test 2023-06-22 15:36:46 +00:00
bors
8d1fa473dd Auto merge of #112724 - scottmcm:simpler-unchecked-shifts, r=Mark-Simulacrum
[libs] Simplify `unchecked_{shl,shr}`

There's no need for the `const_eval_select` dance here.  And while I originally wrote the `.try_into().unwrap_unchecked()` implementation here, it's kinda a mess in MIR -- this new one is substantially simpler, as shown by the old one being above the inlining threshold but the new one being below it in the `mir-opt/inline/unchecked_shifts` tests.

We don't need `u32::checked_shl` doing a dance through both `Result` *and* `Option` 🙃
2023-06-19 04:48:35 +00:00
Scott McMurray
3ec4eeddef [libs] Simplify unchecked_{shl,shr}
There's no need for the `const_eval_select` dance here.  And while I originally wrote the `.try_into().unwrap_unchecked()` implementation here, it's kinda a mess in MIR -- this new one is substantially simpler, as shown by the old one being above the inlining threshold but the new one being below it.
2023-06-16 16:03:19 -07:00
Michael Goulet
3eb8c2ae10
Rollup merge of #112474 - ldm0:ldm_enum_debuginfo_128_support, r=compiler-errors
Support 128-bit enum variant in debuginfo codegen

fixes #111600
2023-06-16 12:53:22 -07:00
The 8472
d90508f761 use indexed loop instead of ptr bumping
this seems to produce less IR
2023-06-14 22:22:41 +02:00
Pietro Albini
f5d3de2abc
update codegent test to be independent of panic strategy 2023-06-14 12:52:02 +02:00
The 8472
ba5b2f0b4b add codegen test for slice::Iter::fold 2023-06-12 23:47:08 +02:00
Erik Desjardins
bd0aae92dc cg_llvm: use index-based loop in write_operand_repeatedly
This is easier for LLVM to analyze.
2023-06-11 00:04:53 -04:00
ldm0
ea19243d15 Add test 2023-06-10 16:58:51 +00:00
ldm0
859bac7af8 Fix codegen tests 2023-06-10 08:31:06 +00:00
Scott McMurray
e1b020df9f Use load-store instead of memcpy for short integer arrays 2023-06-04 00:51:49 -07:00
Scott McMurray
cce0b52e7b Add a codegen test for manually swapping a small Copy type
To confirm we're not just helping `mem::swap`
2023-06-04 00:50:55 -07:00
Matthias Krüger
bdf9ed497a
Rollup merge of #111878 - ferrocene:pa-codegen-tests, r=Mark-Simulacrum
Fix codegen test suite for bare-metal-like targets

For Ferrocene I needed to run the test suite for custom target with no unwinding and static relocation. Running the tests uncovered ~20 failures due to the test suite not accounting for these options. This PR fixes them by:

* Fixing `CHECK`s to account for functions having extra LLVM IR attributes (in this case `nounwind`).
* Fixing `CHECK`s to account for the `dso_local` LLVM IR modifier, which is [added to every item when relocation is static](f3d597b31c/compiler/rustc_codegen_llvm/src/mono_item.rs (L139-L142)).
* Fixing `CHECK`s to account for missing `uwtables` attributes.
* Added the `needs-unwind` attributes for tests that are designed to check unwinding.

There is no part of Rust CI that checks this unfortunately, and testing whether the PR works locally is kinda hard because you need a target with std enabled but no unwinding and static relocations. Still, this works in my local testing, and if future PRs accidentally break this Ferrocene will take care of sending followup PRs.
2023-06-03 20:38:11 +02:00
Matthias Krüger
ec51b15b81
Rollup merge of #112182 - rcvalle:rust-cfi-fix-111185, r=compiler-errors
CFI: Fix cfi with repr(transparent): transform_ty: unexpected Alias(Proj

Fixes https://github.com/rust-lang/rust/issues/111185 by normalizing ty::Alias before encoding.
2023-06-02 18:12:45 +02:00
Ramon de C Valle
faf31b5052 CFI: Fix cfi with repr(transparent): transform_ty: unexpected Alias(Proj
Fixes #111185 by normalizing ty::Alias before encoding.
2023-06-01 19:56:36 +00:00
bors
642c92e630 Auto merge of #112002 - saethlin:enable-sroa, r=oli-obk,scottmcm
Enable ScalarReplacementOfAggregates in optimized builds

Like MatchBranchSimplification, this pass is known to produce significant runtime improvements in Cranelift artifacts, and I believe based on the perf runs here that the primary effect of this pass is to empower MatchBranchSimplification. ScalarReplacementOfAggregates on its own has little effect on anything, but when this was rebased up to include https://github.com/rust-lang/rust/pull/112001 we started seeing significant and majority-positive results.

Based on the fact that we see most of the regressions in debug builds (https://github.com/rust-lang/rust/pull/112002#issuecomment-1566270144) and some rather significant ones in cycles and wall time, I'm only enabling this in optimized builds at the moment.
2023-06-01 10:47:14 +00:00
Scott McMurray
bf36193ef6 Add a distinct OperandValue::ZeroSized variant for ZSTs
These tend to have special handling in a bunch of places anyway, so the variant helps remember that.  And I think it's easier to grok than non-Scalar Aggregates sometimes being `Immediates` (like I got wrong and caused 109992).  As a minor bonus, it means we don't need to generate poison LLVM values for them to pass around in `OperandValue::Immediate`s.
2023-05-31 19:10:28 -07:00
Ben Kimock
79ba7b307d Enable ScalarReplacementOfAggregates 2023-05-31 19:18:16 -04:00
bors
871b595202 Auto merge of #111913 - oli-obk:valtrees2, r=lcnr
Only rewrite valtree-constants to patterns and keep other constants opaque

Now that we can reliably fall back to comparing constants with `PartialEq::eq` to the match scrutinee, we can

1. eagerly try to convert constants to valtrees
2. then deeply convert the valtree to a pattern
3. if the to-valtree conversion failed, create an "opaque constant" pattern.

This PR specifically avoids any behavioral changes or major cleanups. What we can now do as follow ups is

* move the two remaining call sites to `destructure_mir_constant` off that query
* make valtree to pattern conversion infallible
    * this needs to be done after careful analysis of the effects. There may be user visible changes from that.

based on https://github.com/rust-lang/rust/pull/111768
2023-05-31 16:36:51 +00:00