Update invalid atomic ordering lint
The restriction that success ordering must be at least as strong as its
failure ordering in compare-exchange operations was lifted in #98383.
Mention first and last macro in backtrace
Slight improvement to diagnostic mentioning what macro an error originates from. Not sure if it's worthwhile.
wf-check generators
fixes#90409
We should not rely on generators being well formed by construction now that they can get used via type alias impl trait (and thus users can choose generic arguments that are invalid). This can cause surprising behaviour if (definitely unsound) transmutes are used, and it's generally saner to just check for well formedness.
proc_macro/bridge: stop using a remote object handle for proc_macro Ident and Literal
This is the fourth part of https://github.com/rust-lang/rust/pull/86822, split off as requested in https://github.com/rust-lang/rust/pull/86822#pullrequestreview-1008655452. This patch transforms the `Ident` and `Group` types into structs serialized over IPC rather than handles.
Symbol values are interned on both the client and server when deserializing, to avoid unnecessary string copies and keep the size of `TokenTree` down. To do the interning efficiently on the client, the proc-macro crate is given a vendored version of the fxhash hasher, as `SipHash` appeared to cause performance issues. This was done rather than depending on `rustc_hash` as it is unfortunately difficult to depend on crates from within `proc_macro` due to it being built at the same time as `std`.
In addition, a custom arena allocator and symbol store was also added, inspired by those in `rustc_arena` and `rustc_span`. To prevent symbol re-use across multiple invocations of a macro on the same thread, a new range of `Symbol` names are used for each invocation of the macro, and symbols from previous invocations are cleaned-up.
In order to keep `Ident` creation efficient, a special ASCII-only case was added to perform ident validation without using RPC for simple identifiers. Full identifier validation couldn't be easily added, as it would require depending on the `rustc_lexer` and `unicode-normalization` crates from within `proc_macro`. Unicode identifiers are validated and normalized using RPC.
See the individual commit messages for more details on trade-offs and design decisions behind these patches.
Rollup of 7 pull requests
Successful merges:
- #98839 (Add assertion that `transmute_copy`'s U is not larger than T)
- #98998 (Remove branch target prologues from `#[naked] fn`)
- #99198 (add missing null ptr check in alloc example)
- #99344 (rustdoc: avoid inlining items with duplicate `(type, name)`)
- #99351 (Use `typeck_results` to get accurate qpath res for arg mismatch error)
- #99378 (interpret/visitor: add missing early return)
- #99394 (Add regression test for #95230)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Doing this for all unicode identifiers would require a dependency on
`unicode-normalization` and `rustc_lexer`, which is currently not
possible for `proc_macro` due to it being built concurrently with `std`
and `core`. Instead, ASCII identifiers are validated locally, and an RPC
message is used to validate unicode identifiers when needed.
String values are interned on the both the server and client when
deserializing, to avoid unnecessary copies and keep Ident cheap to copy and
move. This appears to be important for performance.
The client-side interner is based roughly on the one from rustc_span, and uses
an arena inspired by rustc_arena.
RPC messages passing symbols always include the full value. This could
potentially be optimized in the future if it is revealed to be a
performance bottleneck.
Despite now having a relevant implementaion of Display for Ident, ToString is
still specialized, as it is a hot-path for this object.
The symbol infrastructure will also be used for literals in the next
part.
Use `typeck_results` to get accurate qpath res for arg mismatch error
Improves error message from "function" to actually what we're calling (e.g. enum variant constrcutor) in a few cases 😸
Remove branch target prologues from `#[naked] fn`
This patch hacks around rust-lang/rust#98768 for now via injecting appropriate attributes into the LLVMIR we emit for naked functions. I intend to pursue this upstream so that these attributes can be removed in general, but it's slow going wading through C++ for me.
Don't pass InferCtxt to WfPredicates
Simple cleanup. Infer vars will get passed up as obligations and shallowed resolved later. This actually improves one test output.
Rollup of 6 pull requests
Successful merges:
- #98383 (Remove restrictions on compare-exchange memory ordering.)
- #99350 (Be more precise when suggesting removal of parens on unit ctor)
- #99356 (Do not constraint TAITs when checking impl/trait item compatibility)
- #99360 (Do not ICE when we have `-Zunpretty=expanded` with invalid ABI)
- #99373 (Fix source code sidebar tree auto-expand)
- #99374 (Fix doc for `rchunks_exact`)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Fix source code sidebar tree auto-expand
Here is the bug:
It was happening because as soon as we found the file (from the URL), every item following it was then opened, even if it wasn't supposed to.
The GUI test ensures that it doesn't happen by adding two nested levels and ensuring only this path is "open".
r? ``@notriddle``
Do not constraint TAITs when checking impl/trait item compatibility
Check out the UI test for the example.
Open to other approaches to fix this issue -- ideally we _would_ be able to collect this opaque type constraint in a way to use it in `find_opaque_ty_constraints`, so we can report a better mismatch error in the incompatible case, and just allow it in the compatible case. But that seems like a bigger refactor, so I wouldn't want to start it unless someone else thought it was a good idea.
cc #99348
r? ``@oli-obk``
Be more precise when suggesting removal of parens on unit ctor
* Fixes#99240 by only suggesting to remove parens on path exprs, not arbitrary expressions with enum type
* Generalizes by suggesting removal of parens on unit struct, too, because why not?
Use ICF (identical code folding) for building rustc
It seems that ICF (identical code folding) is able to remove duplicated functions created by monomorphization from binaries, resulting in smaller binary size and better i-cache utilization. Let's see if it helps for `rustc`.
I'm not sure if `lld` is even used for linking `rustc` on the Linux `dist` builder, let's see.
Use constant eval to do strict mem::uninit/zeroed validity checks
I'm not sure about the code organisation here, I just dumped the check in rustc_const_eval at the root. Not hard to move it elsewhere, in any case.
Also, this means cranelift codegen intrinsics lose the strict checks, since they don't seem to depend on rustc_const_eval, and I didn't see a point in keeping around two copies.
I also left comments in the is_zero_valid methods about "uhhh help how do i do this", those apply to both methods equally.
Also rustc_codegen_ssa now depends on rustc_const_eval... is this okay?
Pinging `@RalfJung` since you were the one who mentioned this to me, so I'm assuming you're interested.
Haven't had a chance to run full tests on this since it's really warm, and it's 1AM, I'll check out any failures/comments in the morning :)
Add a special case for align_offset /w stride != 1
This generalizes the previous `stride == 1` special case to apply to any
situation where the requested alignment is divisible by the stride. This
in turn allows the test case from #98809 produce ideal assembly, along
the lines of:
leaq 15(%rdi), %rax
andq $-16, %rax
This also produces pretty high quality code for situations where the
alignment of the input pointer isn’t known:
pub unsafe fn ptr_u32(slice: *const u32) -> *const u32 {
slice.offset(slice.align_offset(16) as isize)
}
// =>
movl %edi, %eax
andl $3, %eax
leaq 15(%rdi), %rcx
andq $-16, %rcx
subq %rdi, %rcx
shrq $2, %rcx
negq %rax
sbbq %rax, %rax
orq %rcx, %rax
leaq (%rdi,%rax,4), %rax
Here LLVM is smart enough to replace the `usize::MAX` special case with
a branch-less bitwise-OR approach, where the mask is constructed using
the neg and sbb instructions. This appears to work across various
architectures I’ve tried.
This change ends up introducing more branches and code in situations
where there is less knowledge of the arguments. For example when the
requested alignment is entirely unknown. This use-case was never really
a focus of this function, so I’m not particularly worried, especially
since llvm-mca is saying that the new code is still appreciably faster,
despite all the new branching.
Fixes#98809.
Sadly, this does not help with #72356.
This generalizes the previous `stride == 1` special case to apply to any
situation where the requested alignment is divisible by the stride. This
in turn allows the test case from #98809 produce ideal assembly, along
the lines of:
leaq 15(%rdi), %rax
andq $-16, %rax
This also produces pretty high quality code for situations where the
alignment of the input pointer isn’t known:
pub unsafe fn ptr_u32(slice: *const u32) -> *const u32 {
slice.offset(slice.align_offset(16) as isize)
}
// =>
movl %edi, %eax
andl $3, %eax
leaq 15(%rdi), %rcx
andq $-16, %rcx
subq %rdi, %rcx
shrq $2, %rcx
negq %rax
sbbq %rax, %rax
orq %rcx, %rax
leaq (%rdi,%rax,4), %rax
Here LLVM is smart enough to replace the `usize::MAX` special case with
a branch-less bitwise-OR approach, where the mask is constructed using
the neg and sbb instructions. This appears to work across various
architectures I’ve tried.
This change ends up introducing more branches and code in situations
where there is less knowledge of the arguments. For example when the
requested alignment is entirely unknown. This use-case was never really
a focus of this function, so I’m not particularly worried, especially
since llvm-mca is saying that the new code is still appreciably faster,
despite all the new branching.
Fixes#98809.
Sadly, this does not help with #72356.
