Issue-125323: ICE non-ADT in struct pattern when long time constant evaluation is in for loop
This PR fixes#125323
## Context
According to the issue, the ICE happens since #121206.
In the PR, some error methods were reorganized. For example, has_errors() was renamed to has_errors_exclude_lint_errors(). However, some codes which used the original has_errors() were not switched to has_errors_exclude_lint_errors(). I finally found that report_error() in writeback.rs causes this ICE. Currently the method uses tainted_by_errors() to get guar (ErrorGuaranteed), but originally it used dcx().has_errors() but it wasn't changed to has_errors_exclude_lint_errors() when changes in #121206 were merged. I don't think I fully understand how an error is propagated, but I suppose that the error from long time constant evaluation is unexpectedly propagated other parts (in this ICE, for loop), then cause the non-ADT in struct pattern ICE.
## Change
- Fix report_error() in writeback.rs: use dcx().has_errors_exclude_lint_errors() instead of tainted_by_errors() to prevent error propagation from constant evaluation.
- Add test for the ICE
- Modify some tests to align the change: Due to this fix, E0282 error happens (or not happen anymore) in some tests.
## NOTE
The 4th commit aims to revert the fix in #123516 because I confirmed that the ICE solved by the PR doesn't happen if I modify report_error(). I think the root cause of that ICE is the same as #125323 . But I can discard this commit since we can fix#125323 without it.
Result/Option layout guarantee clarifications
- It seems worth spelling out that this guarantee allows particular transmutes.
- After the `Result` section was written, the `Option` section it referenced gained *more* guarantees, saying that `None` is represented as `[0u8; N]`. Those guarantees were not meant to apply to `Result`. Make the `Result` section more self-contained to make this more clear.
- "Type has no fields" is unclear since there is no general definition of what the fields of some arbitrary type are. Replace that by a more accurate description of the actual check implemented [here](https://github.com/rust-lang/rust/blob/e379c7758667f900aaf5551c4553c7d4c121e3e1/compiler/rustc_lint/src/types.rs#L828-L838).
r? `@traviscross`
deduced_param_attrs: check Freeze on monomorphic types.
`deduced_param_attrs` currently checks `Freeze` bound on polymorphic MIR. This pessimizes the deduction, as generic types are not `Freeze` by default.
This moves the check to the ABI adjustment.
Simplify trivial constants in SimplifyConstCondition
After `InstSimplify-after-simplifycfg` with `-Zub_checks=false`, there are many of the following patterns.
```
_13 = const false;
assume(copy _13);
_12 = unreachable_unchecked::precondition_check() -> [return: bb1, unwind unreachable];
```
Simplifying them to unreachable early should make CFG simpler.
Rollup of 8 pull requests
Successful merges:
- rust-lang/rust#140153 (Implement `Debug` for `EncodeWide`)
- rust-lang/rust#145724 (the `#[track_caller]` shim should not inherit `#[no_mangle]`)
- rust-lang/rust#147258 (iter repeat: panic on last)
- rust-lang/rust#147454 (Fix backtraces with `-C panic=abort` on qnx; emit unwind tables by default)
- rust-lang/rust#147468 (Implement fs api set_times and set_times_nofollow)
- rust-lang/rust#147764 (Undo CopyForDeref assertion in const qualif)
- rust-lang/rust#147805 (use module_child index as disambiguator for external items)
- rust-lang/rust#147824 (docs: update Motor OS target docs)
r? `@ghost`
`@rustbot` modify labels: rollup
use module_child index as disambiguator for external items
When defining the items of an external module, if that item is an underscore we use it's index as the disambiguator. This is needed for parallel import resolution, which is being worked on in rust-lang/rust#145108.
r? `@petrochenkov`
Fix backtraces with `-C panic=abort` on qnx; emit unwind tables by default
While syncing https://github.com/rust-lang/rust/pull/143613 into Ferrocene as part of https://github.com/ferrocene/ferrocene/pull/1803, we noted a failure on our QNX targets:
```
---- [ui] tests/ui/panics/panic-abort-backtrace-without-debuginfo.rs stdout ----
error: test did not exit with success! code=Some(134) so test would pass with `run-crash`
status: exit status: 134
command: RUSTC="/home/ci/project/build/x86_64-unknown-linux-gnu/stage2/bin/rustc" RUST_TEST_THREADS="1" "/home/ci/project/build/x86_64-unknown-linux-gnu/stage1-tools-bin/remote-test-client" "run" "0" "/home/ci/project/build/x86_64-unknown-linux-gnu/test/ui/panics/panic-abort-backtrace-without-debuginfo/a"
--- stdout -------------------------------
uploaded "/home/ci/project/build/x86_64-unknown-linux-gnu/test/ui/panics/panic-abort-backtrace-without-debuginfo/a", waiting for result
died due to signal 6
------------------------------------------
--- stderr -------------------------------
thread 'main' (1) panicked at /home/ci/project/tests/ui/panics/panic-abort-backtrace-without-debuginfo.rs:39:9:
ERROR: no `this_function_must_be_in_the_backtrace` in stderr! actual stderr:
thread 'main' (1) panicked at /home/ci/project/tests/ui/panics/panic-abort-backtrace-without-debuginfo.rs:27:5:
generate panic backtrace
stack backtrace:
0: 0x4e66a53643 - <std::sys::backtrace::BacktraceLock::print::DisplayBacktrace as core::fmt::Display>::fmt::h55e010263b1e3169
1: 0x4e66a68cd2 - core::fmt::write::h0d6e2e8752abc333
2: 0x4e66a16919 - std::io::Write::write_fmt::h71c4c024d832b384
3: 0x4e66a1f8e2 - std::sys::backtrace::BacktraceLock::print::hdd80dfdf90bb7100
4: 0x4e66a221e0 - std::panicking::default_hook::{{closure}}::h77758f25a686500f
5: 0x4e66a21f69 - std::panicking::default_hook::ha63f7d476af6c267
6: 0x4e66a22999 - std::panicking::panic_with_hook::h3a36a8a0f0dd8ccd
7: 0x4e66a21cac - std::panicking::begin_panic::{{closure}}::h570dedb92e232392
8: 0x4e66a1fa69 - std::sys::backtrace::__rust_end_short_backtrace::h5366eec354f92733
9: 0x4e669f9589 - std::panicking::begin_panic::h04a4bd4c33dd4056
10: 0x4e66a00aca - panic_abort_backtrace_without_debuginfo::and_this_function_too::h5b034b94cbe9c3d3
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
------------------------------------------
---- [ui] tests/ui/panics/panic-abort-backtrace-without-debuginfo.rs stdout end ----
failures:
[ui] tests/ui/panics/panic-abort-backtrace-without-debuginfo.rs
test result: FAILED. 19958 passed; 1 failed; 328 ignored; 0 measured; 0 filtered out; finished in 1827.71s
Some tests failed in compiletest suite=ui mode=ui host=x86_64-unknown-linux-gnu target=x86_64-pc-nto-qnx710
Build completed unsuccessfully in 0:43:28
Exited with code exit status 1
```
This patch applies the same fix as the one found in https://github.com/rust-lang/rust/pull/143613 of adding the `default_uwtable: true` to the target.
I've run it locally, when https://github.com/ferrocene/ferrocene/pull/1803 merges we'll know it has passed within our CI, which is about a close an analog as I can offer to Rust.
the `#[track_caller]` shim should not inherit `#[no_mangle]`
fixes https://github.com/rust-lang/rust/issues/143162
builds on https://github.com/rust-lang/rust/pull/143293 which introduced a mechanism to strip attributes from shims.
cc `@Jules-Bertholet` `@workingjubilee` `@bjorn3`
---
Summary:
This PR fixes an interaction between `#[track_caller]`, `#[no_mangle]`, and casting to a function pointer.
A function annotated with `#[track_caller]` internally has a hidden extra argument for the panic location. The `#[track_caller]` attribute is only allowed on `extern "Rust"` functions. When a function is annotated with both `#[no_mangle]` and `#[track_caller]`, the exported symbol has the signature that includes the extra panic location argument. This works on stable rust today:
```rust
extern "Rust" {
#[track_caller]
fn rust_track_caller_ffi_test_tracked() -> &'static Location<'static>;
}
mod provides {
use std::panic::Location;
#[track_caller] // UB if we did not have this!
#[no_mangle]
fn rust_track_caller_ffi_test_tracked() -> &'static Location<'static> {
Location::caller()
}
}
```
When a `#[track_caller]` function is converted to a function pointer, a shim is added to drop the additional argument. So this is a valid program:
```rust
#[track_caller]
fn foo() {}
fn main() {
let f = foo as fn();
f();
}
```
The issue arises when `foo` is additionally annotated with `#[no_mangle]`, the generated shim currently inherits this attribute, also exporting a symbol named `foo`, but one without the hidden panic location argument. The linker rightfully complains about a duplicate symbol.
The solution of this PR is to have the generated shim drop the `#[no_mangle]` attribute.
mismatched_lifetime_syntax lint refactors and optimizations
I found several opportunities to return early so I'm hoping those will have a perf improvement. Otherwise, it's various refactors for simplicity.
Limit impl_trait_header query to only trait impls
Changes `impl_trait_header` to panic on inherent impls intstead of returning None. A few downstream functions are split into option and non-option returning functions. This gets rid of a lot of unwraps where we know we have a trait impl, while there are still some cases where the Option is helpful.
Summary of changes to tcx methods:
* `impl_is_of_trait` (new)
* `impl_trait_header` -> `impl_trait_header`/`impl_opt_trait_header`
* `impl_trait_ref` -> `impl_trait_ref`/`impl_opt_trait_ref`
* `trait_id_of_impl` -> `impl_trait_id`/`impl_opt_trait_id`
Pre-compute MIR CFG caches for borrowck and other analyses
I was puzzled that https://github.com/rust-lang/rust/pull/142390 introduces additional computations of CFG traversals: borrowck computes them, right?
It turns out that borrowck clones the MIR body, so doesn't share its cache with other analyses.
This PR:
- forces the computation of all caches in `mir_promoted` query;
- modifies region renumbering to avoid dropping that cache.
error from const eval lint causes ICE at check_pat in late_lint, because the function expects the typeck result isn't tainted by error but it is.
To avoid the ICE, check_pat returns earlier if the typeck_result is tainted.
check_mod_deathness also has an issue from the same reason. visit_body for making live symbols expects the typeck result has no error.
So this commit adds a check in visit_nested_body to avoid the ICE.
However, if visit_nested_body just returns without doing anything, all codes with the error are marked as dead, because live_symbols is empty.
To avoid this side effect, visit_nested_body and other visit_* functions in MarkSymbolVistior should return appropriate error.
If a function returns ControlFlow::Break, live_symbols_and_ignore_derived_traits returns earlier with error,
then check_mod_deathness, the caller of the function returns earlier without pushing everything into dead_codes.
When -Z autodiff=Enable is used, the compiler automatically enables
fat-lto for all crates. However, proc-macro crates cannot use fat-lto
without the -Zdylib-lto flag, causing compilation errors.
This commit modifies the lto() method in Session to exclude proc-macro
crates from fat-lto when autodiff is enabled, while preserving the
existing behavior for all other crate types.
The fix ensures that:
- Non-proc-macro crates still get fat-lto when autodiff is enabled
- Proc-macro crates are excluded from fat-lto when autodiff is enabled
- Existing autodiff functionality remains unchanged for regular crates
Signed-off-by: Osama Abdelkader <osama.abdelkader@gmail.com>
Use `bit_set::Word` in a couple more places.
It's a synonym for `u64` and there are a couple of places where we use `u64` where we should use `Word`, which this commit fixes.
I found this when I tried changing `Word` to `u128` (which made performance worse).
r? `````@Zalathar`````
Fix ICE on offsetted ZST pointer
I'm not sure this is the *right* fix, but it's simple enough and does roughly what I'd expect. Like with the previous optimization to codegen usize rather than a zero-sized static, there's no guarantee that we continue returning a particular value from the offsetting.
A grep for `const_usize.*align` found the same code copied to rustc_codegen_gcc and cranelift but a quick skim didn't find other cases of similar 'optimization'. That said, I'm not convinced I caught everything, it's not trivial to search for this.
Closesrust-lang/rust#147516
Guard HIR lowered contracts with `contract_checks`
Refactor contract HIR lowering to ensure no contract code is executed when contract-checks are disabled.
The call to `contract_checks` is moved to inside the lowered fn body, and contract closures are built conditionally, ensuring no side-effects present in contracts occur when those are disabled. This partially addresses rust-lang/rust#139548, i.e. the bad behavior no longer happens with contract checks disabled (`-Zcontract-checks=no`).
The change is made in preparation for adding contract variable declarations - variables declared before the `requires` assertion, and accessible from both `requires` and `ensures`, but not in the function body (PR rust-lang/rust#144444). As those declarations may also have side-effects, it's good to guard them with `contract_checks` - the new lowering approach allows for this to be done easily.
Contracts tracking issue: rust-lang/rust#128044
**Known limiatations**:
- It is still possible to early return from the *function* from within a contract, e.g.
```rust
#[ensures({if x > 0 { return 0 }; |_| true})]
fn foo(x: u32) -> i32 {
42
}
```
When `foo` is called with an argument greater than 0, instead of `42`, `0` will be returned.
As this is not a regression, it is not addressed in this PR. However, it may be worth revisiting later down the line, as users may expect a form of early return from *contract specifications*, and so returning from the entire *function* could cause confusion.
- ~Contracts are still not optimised out when disabled. Currently, even when contracts are disabled, the code generated causes existing optimisations to fail, meaning even disabled contracts could impact runtime performance. This issue is blocking rust-lang/rust#136578, and has not been addressed in this PR, i.e. the `mir-opt` and `codegen` tests that fail in rust-lang/rust#136578 still fail with these new HIR lowering changes.~ Contracts should now be optimised out when disabled, however some regressions tests still need to be added to be sure that is indeed the case.
It's a synonym for `u64` and there are a couple of places where we use
`u64` where we should use `Word`, which this commit fixes.
I found this when I tried changing `Word` to `u128` (which made
performance worse).
Rollup of 11 pull requests
Successful merges:
- rust-lang/rust#143191 (Stabilize `rwlock_downgrade` library feature)
- rust-lang/rust#147444 (Allow printing a fully-qualified path in `def_path_str`)
- rust-lang/rust#147527 (Update t-compiler beta nomination Zulip msg)
- rust-lang/rust#147670 (some `ErrorGuaranteed` cleanups)
- rust-lang/rust#147676 (Return spans out of `is_doc_comment` to reduce reliance on `.span()` on attributes)
- rust-lang/rust#147708 (const `mem::drop`)
- rust-lang/rust#147710 (Fix ICE when using contracts on async functions)
- rust-lang/rust#147716 (Fix some comments)
- rust-lang/rust#147718 (miri: use allocator_shim_contents codegen helper)
- rust-lang/rust#147729 (ignore boring locals when explaining why a borrow contains a point due to drop of a live local under polonius)
- rust-lang/rust#147742 (Revert unintentional whitespace changes to rustfmt-excluded file)
r? `@ghost`
`@rustbot` modify labels: rollup
Kivooeo