-
Notifications
You must be signed in to change notification settings - Fork 12.5k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
nontemporal_store: make sure that the intrinsic is truly just a hint #128149
Conversation
Some changes occurred in compiler/rustc_codegen_cranelift cc @bjorn3 Some changes occurred in compiler/rustc_codegen_gcc |
This comment has been minimized.
This comment has been minimized.
05e50ef
to
8d0d26e
Compare
e2efa65
to
d1e92af
Compare
It's a bit sad if this becomes an actual no-op on x86, nontemporal stores can be (very rarely) very useful. Is this usable through anything else than mm_stream_* and friends? These functions already contain a safety note that a fence has to be used after them. |
Those intrinsic functions are, in the latest commits to |
So mm_stream will still emit nontemporal instructions on x86? Thanks for clarifying! |
Yes it will, that's why rust-lang/stdarch#1541 needs to propagate first. |
Ok, great, then I misunderstood this change, sorry! |
Thank you, @RalfJung. |
0135bb1
to
697787a
Compare
rust-lang/stdarch#1541 has propagated, so this should be good to go. We haven't heard from @fmease though so let's re-roll the reviewer dice. r? compiler |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks! The changes themselves look good based on the prior discussions. Just a question about the codegen test with regards to target coverage.
const WELL_BEHAVED_NONTEMPORAL_ARCHS: &[&str] = | ||
&["aarch64", "arm", "riscv32", "riscv64"]; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Question [REV1 (2/2)]: If I understand it correctly, the codegen test is modified such that it only codegens for aarch64-unknown-linux-gnu
target, but here we have 4 "well-behaved" architectures that we do emit !nontemporal
for.
r? jieyouxu |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks, feel free to r=me,Amanieu,Jubilee after maybe fixing the nit :3
b1eca7b
to
5be6c7d
Compare
@bors r=jieyouxu,Amanieu,Jubilee |
5be6c7d
to
75743dc
Compare
@bors r=jieyouxu,Amanieu,Jubilee |
…llaumeGomez Rollup of 10 pull requests Successful merges: - rust-lang#128149 (nontemporal_store: make sure that the intrinsic is truly just a hint) - rust-lang#128394 (Unify run button display with "copy code" button and with mdbook buttons) - rust-lang#128537 (const vector passed through to codegen) - rust-lang#128632 (std: do not overwrite style in `get_backtrace_style`) - rust-lang#128878 (Slightly refactor `Flags` in bootstrap) - rust-lang#128886 (Get rid of some `#[allow(rustc::untranslatable_diagnostic)]`) - rust-lang#128929 (Fix codegen-units tests that were disabled 8 years ago) - rust-lang#128937 (Fix warnings in rmake tests on `x86_64-unknown-linux-gnu`) - rust-lang#128978 (Use `assert_matches` around the compiler more) - rust-lang#128994 (Fix bug in `Parser::look_ahead`.) r? `@ghost` `@rustbot` modify labels: rollup
Rollup merge of rust-lang#128149 - RalfJung:nontemporal_store, r=jieyouxu,Amanieu,Jubilee 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](llvm/llvm-project#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 rust-lang/stdarch#1541 propagating to the rustc repo, to make sure the `_mm_stream` intrinsics are unaffected by this change.~~ Fixes rust-lang#114582 Cc `@Amanieu` `@workingjubilee`
febaf22f00e Rollup merge of rust-lang#129592 - saethlin:core-cfg-test, r=tgross35 77a1318f7f7 Rollup merge of rust-lang#129588 - hermit-os:sleep-micros, r=workingjubilee 12fe23bd5dc Rollup merge of rust-lang#129539 - oconnor663:poll_link, r=tgross35 864e465be7b Rollup merge of rust-lang#129377 - chorman0773:unbounded-shifts-impl, r=scottmcm d958260763a Auto merge of rust-lang#129595 - matthiaskrgr:rollup-4udn7nn, r=matthiaskrgr 8dd3363de6a Remove cfg(test) from library/core cd554e2b4ea Rollup merge of rust-lang#129544 - mu001999-contrib:dead-code/clean, r=compiler-errors ff769eef88a Rollup merge of rust-lang#129525 - notriddle:notriddle/fake-variadic-tuple-array, r=GuillaumeGomez 4d22c1c6b37 Auto merge of rust-lang#129488 - saethlin:alignment-precondition, r=workingjubilee c688deff96a pal/hermit: saturate `usleep` microseconds at `u64::MAX` 8ea71ae5647 Auto merge of rust-lang#129563 - matthiaskrgr:rollup-t6bai2d, r=matthiaskrgr fdb5fc1fca6 pal/hermit: correctly round up microseconds in `Thread::sleep` 22ec8977a9a link to Future::poll from the Poll docs a994fbbca83 Rollup merge of rust-lang#129487 - GrigorenkoPV:repr_transparent_external_private_fields, r=compiler-errors 3a339226a95 Rollup merge of rust-lang#129416 - workingjubilee:partial-move-from-stabilization, r=dtolnay 3a8de952989 Rollup merge of rust-lang#129091 - RalfJung:box_as_ptr, r=Amanieu 4de4debd1eb Auto merge of rust-lang#129295 - Zalathar:profiler-builtins, r=Kobzol 8dafd337e12 Auto merge of rust-lang#129521 - matthiaskrgr:rollup-uigv77m, r=matthiaskrgr d9e489b48ce Removes dead code from the compiler c14cf57e404 Rollup merge of rust-lang#129481 - scottmcm:update-cb, r=tgross35 acf6f03b6fa Rollup merge of rust-lang#129449 - coolreader18:pin-as_deref_mut-signature, r=dtolnay 112ebc4d5a9 Rollup merge of rust-lang#128735 - jieyouxu:pr-120176-revive, r=cjgillot 49aa496c5bb rustdoc: clean up tuple <-> primitive conversion docs 0fe374666ac Rollup merge of rust-lang#129501 - RalfJung:miri-rust-backtrace, r=Noratrieb 7d5cf38931b Rollup merge of rust-lang#129500 - fee1-dead-contrib:fxrel, r=compiler-errors e91d825f826 Rollup merge of rust-lang#129323 - Urgau:ptr_fn_addr_eq, r=Mark-Simulacrum f6470795864 Rollup merge of rust-lang#128596 - RalfJung:const_fn_floating_point_arithmetic, r=nnethercote f965950f05e New `#[rustc_pub_transparent]` attribute a6ea125cb0e panicking: improve hint for Miri's RUST_BACKTRACE behavior a437005e6f3 Build `library/profiler_builtins` from `ci-llvm` if appropriate 693477a3f65 remove invalid `TyCompat` relation for effects 82fc74f6f1a library: Move unstable API of new_uninit to new features 3ee2e18c8dc Enable Alignment::new_unchecked precondition check 0803686e7fb Change `f16` doctests in core to run on x86-64 linux 9359a126d41 Update `compiler_builtins` to `0.1.121` da02e8b5609 Enable `f16` tests on x86 and x86-64 976fb4aeefc Move into_inner_unchecked back to the bottom of the impl block 2741e8dacb4 Put Pin::as_deref_mut in impl Pin<Ptr> f670207d0f4 Auto merge of rust-lang#129464 - GuillaumeGomez:rollup-ckfqd7h, r=GuillaumeGomez 5bf661cc64f Rollup merge of rust-lang#129276 - eduardosm:stabilize-char_indices_offset, r=Amanieu e2614f24b27 Rollup merge of rust-lang#129400 - Amjad50:update-compiler-builtins, r=tgross35 2c06146be10 Rollup merge of rust-lang#127623 - lolbinarycat:fix_remove_dir_all, r=Amanieu eb747e53dd0 Check that `library/profiler_builtins` actually found some source files fe33d2c256c fix: fs::remove_dir_all: treat ENOENT as success 3fd591ebdef feat(core): Make `unbounded_shl{l,r}` unstably const and remove `rustc_allow_const_fn_unstable` 2168ce32967 Auto merge of rust-lang#129398 - matthiaskrgr:rollup-50l01ry, r=matthiaskrgr 12944c76047 Update `compiler_builtins` to `0.1.120` 7496478b7a5 stabilize const_fn_floating_point_arithmetic 6f534f94217 Rollup merge of rust-lang#129382 - tgross35:once-cell-const-into-inner, r=Noratrieb 2535017098e Rollup merge of rust-lang#129376 - ChaiTRex:assert_unsafe_precondition_check_language_ub, r=workingjubilee,the8472 4ec19afe669 Rollup merge of rust-lang#129374 - ChaiTRex:digit_unchecked_assert_unsafe_precondition, r=scottmcm 024ec3c0f62 Rollup merge of rust-lang#128432 - g0djan:godjan/wasi_prohibit_implicit_unsafe, r=tgross35 f671c1129b9 Auto merge of rust-lang#129365 - matthiaskrgr:rollup-ebwx6ya, r=matthiaskrgr 5299ef149b1 fix(core): Use correct operations/values in `unbounded_shr` doctests 84230062104 chore: `x fmt` cbe7338e1f3 fix(core): Add `#![feature(unbounded_shifts)]` to doctests for `unbounded_shr`/`unbounded_shl` 863123bd7c4 Add `const_cell_into_inner` to `OnceCell` 6fd539327d2 chore: `x fmt` and hopefully fix the tidy issue 9d2bb976994 Change `assert_unsafe_precondition` docs to refer to `check_language_ub` 32bd5dfb369 chore: Also format the control flow 5f8cf71d7d6 Manually format functions and use `rhs` instead of `v` from my CE testing 700af565751 feat(core): Add implementations for `unbounded_shl`/`unbounded_shr` a9ad57eb6a1 Use `assert_unsafe_precondition!` in `AsciiChar::digit_unchecked` 77bd65fdedc Rollup merge of rust-lang#129321 - krtab:float_sum, r=workingjubilee cc219788b51 Rollup merge of rust-lang#129232 - ivmarkov:master, r=workingjubilee c9cf844ccd3 Rollup merge of rust-lang#127945 - tgross35:debug-more-non-exhaustive, r=Noratrieb d37ebfea900 Rollup merge of rust-lang#129332 - cuviper:cstr-cast, r=compiler-errors 6d01ed8b3bd Rollup merge of rust-lang#129312 - tbu-:pr_str_not_impl_error, r=Noratrieb 93319c80754 Fix stability attribute of `impl !Error for &str` 7f8bdd574b6 Auto merge of rust-lang#126556 - saethlin:layout-precondition, r=joboet 9e9141f54eb Auto merge of rust-lang#128866 - scottmcm:update-stdarch, r=tgross35 d47cfba89b7 Update stdarch submodule b507a8bfeb9 Try to golf down the amount of code in Layout 32b574e848f Avoid extra `cast()`s after `CStr::as_ptr()` 9d4113ff24d Rollup merge of rust-lang#129294 - scottmcm:stabilize-repeat-n, r=Noratrieb 62d240d9b6a Implement `ptr::fn_addr_eq` 529e33acb80 Change neutral element of <fNN as iter::Sum> to neg_zero 126935f7257 Stabilize `iter::repeat_n` 91439ce7b58 Auto merge of rust-lang#129226 - RalfJung:libc, r=Mark-Simulacrum bef7be0e71e Add a precondition check for Layout::from_size_align_unchecked a55ab85ad47 Stabilize feature `char_indices_offset` 7f45dcfa195 library: bump libc dependency ebe99f3b8b6 Rollup merge of rust-lang#128902 - evanj:evan.jones/env-var-doc, r=workingjubilee 8bdd95ba4da soft-deprecate the addr_of macros 23b0aadc2ce code review improvements 0b0dad4af6f Fix for issue rust-lang#129212 for the ESP-IDF bd7aa576572 Auto merge of rust-lang#126877 - GrigorenkoPV:clone_to_uninit, r=dtolnay d3c08f8f8ac Auto merge of rust-lang#128598 - RalfJung:float-comments, r=workingjubilee dc5fed53253 Auto merge of rust-lang#106943 - mina86:exact_size_take_repeat, r=dtolnay 88927ac26eb Auto merge of rust-lang#116528 - daxpedda:stabilize-ready-into-inner, r=dtolnay 9952947d86b Rollup merge of rust-lang#129161 - dtolnay:spawnunck, r=Noratrieb db3abec9727 Rollup merge of rust-lang#129086 - slanterns:is_none_or, r=dtolnay 44a558dc7dc Stabilize std::thread::Builder::spawn_unchecked 5c553c41134 float to/from bits and classify: update comments regarding non-conformant hardware 9704e2df60c Rollup merge of rust-lang#128064 - ijackson:noop-waker-doc, r=workingjubilee 0497f0c6c91 Add cautionary paragraph about noop wakers. 16dd42669a2 Rollup merge of rust-lang#128946 - orlp:faster-ip-hash, r=joboet 383c4db14b0 Rollup merge of rust-lang#128925 - dingxiangfei2009:smart-ptr-helper-attr, r=compiler-errors ba3a942d5de Rollup merge of rust-lang#125970 - RalfJung:before_exec, r=m-ou-se 0dbf8cff9de Rollup merge of rust-lang#128954 - zachs18:fromresidual-no-default, r=scottmcm 4f0959927f2 Rollup merge of rust-lang#128570 - folkertdev:stabilize-asm-const, r=Amanieu b6c9e44d2a6 add Box::as_ptr and Box::as_mut_ptr methods 23d1309b02e CommandExt::before_exec: deprecate safety in edition 2024 9858d49b168 stabilize `is_none_or` fd2b339c5a6 Auto merge of rust-lang#129060 - matthiaskrgr:rollup-s72gpif, r=matthiaskrgr 3b8aab7df81 Rollup merge of rust-lang#129001 - cblh:fix/128713, r=Noratrieb 16edf695130 Rollup merge of rust-lang#128873 - ChrisDenton:windows-targets, r=Mark-Simulacrum 0199b00c91f Rollup merge of rust-lang#128759 - notriddle:notriddle/spec-to-string, r=workingjubilee,compiler-errors c6dc243b917 stabilize `asm_const` b4bfc215048 Rollup merge of rust-lang#129034 - henryksloan:coroutine-must-use, r=joboet b56fdcb2730 Rollup merge of rust-lang#127857 - tbu-:pr_deprecated_safe_todo, r=petrochenkov 77f462da866 Rollup merge of rust-lang#122884 - mzabaluev:pow-remove-exit-branch, r=Amanieu 0a6a74bce1a Reduce merged doctest source code size a83dde61642 Mark location doctest as standalone since file information will not work in merged doctest file 7334c7178ce Auto merge of rust-lang#129046 - matthiaskrgr:rollup-9x4xgak, r=matthiaskrgr 9ed72103664 Rollup merge of rust-lang#128745 - dtolnay:spawnunchecked, r=workingjubilee c39d90e4d51 Rollup merge of rust-lang#128655 - joboet:play_with_the_dice, r=ChrisDenton f81c96a863e `#[deprecated_safe_2024]`: Also use the `// TODO:` hint in the compiler error 23a19685c9b Allow to customize `// TODO:` comment for deprecated safe autofix 37017c0f6f6 Auto merge of rust-lang#128962 - devnexen:fs_get_mode_haiku, r=workingjubilee 8a2671a2889 Revert to original loop for const pow exponents c5e81895dfb Auto merge of rust-lang#128742 - RalfJung:miri-vtable-uniqueness, r=saethlin ac682f19873 Add must_use attribute to Coroutine trait 658904d1a9a chore(lib): fmt core::fmt::Formatter's write_fmt method 7eb73762bb3 trying common codepath for every unixes 5fabf93c765 std::fs: get_mode implementation for haiku. e3da824e62c Rollup merge of rust-lang#129017 - its-the-shrimp:core_fmt_from_fn, r=Noratrieb b247d9a7a9a derive(SmartPointer): register helper attributes aa854485cea Explicitly specify type parameter on FromResidual impls in stdlib. 262a4f6b641 std::fmt::FormatterFn -> std::fmt::FromFn ceceae30ced Rollup merge of rust-lang#128632 - joboet:dont_overwrite_style, r=Amanieu e8f7afeb117 Rollup merge of rust-lang#128149 - RalfJung:nontemporal_store, r=jieyouxu,Amanieu,Jubilee 7dd208356e1 chore(lib): Enhance documentation for core::fmt::Formatter's write_fmt method 048efd0bcec ignore some vtable/fn ptr equality tests in Miri, their result is not fully predictable a367a12df0a std: use `/scheme/rand` on Redox 1ca6b42583f std: do not overwrite style in `get_backtrace_style` 91477777de1 Auto merge of rust-lang#128862 - cblh:fix/128855, r=scottmcm 56e1afe0810 Auto merge of rust-lang#126793 - saethlin:mono-rawvec, r=scottmcm ec7a585087c Do not use unnecessary endian conversion. f48facfed72 Rollup merge of rust-lang#128882 - RalfJung:local-waker-will-wake, r=cuviper b581949746c Rollup merge of rust-lang#120314 - mina86:i, r=Mark-Simulacrum 451feca66ac Fix stability annotation and expand comment 2e34ac388e0 Hash Ipv*Addr as an integer b8b61e1e931 Auto merge of rust-lang#128927 - GuillaumeGomez:rollup-ei2lr0f, r=GuillaumeGomez 44f5b4fe515 Rollup merge of rust-lang#128273 - Voultapher:improve-ord-violation-help, r=workingjubilee 3d7afa0e721 Update std and compiler 971df1c2948 Stabilize `min_exhaustive_patterns` c37c6665b9b Add an optimizer hint for the capacity that with_capacity_in returns c8cbd5c499c Hoist IS_ZST check out of RawVecInner::from_*_in e843f7103a0 Polymorphize RawVec dc39cbf9234 core: optimise Debug impl for ascii::Char 9668691af5d doc: std::env::var: Returns None for names with '=' or NUL byte 5d5d8bc73a9 Rollup merge of rust-lang#128859 - MinxuanZ:mips-sig, r=Amanieu 825def017bc Rollup merge of rust-lang#128817 - biabbas:vxworks_update, r=tgross35 6e933a82c90 make LocalWaker::will_wake consistent with Waker::will_wake 118c71296c8 Fix linkchecker issue b1460b93704 Exclude windows-targets from the workspace a3a6a9856c2 Add windows-targets crate to std's sysroot f74940d94c2 Rollup merge of rust-lang#128824 - GuillaumeGomez:update-compiler-builtins, r=Amanieu 39b1eafc08c VxWorks: Add safety comment for vxCpuEnabledGet 8b0a25df983 fix: Ensure `Guard`'s `drop` method is removed at `opt-level=s` for `Copy` types c54958c5dad delete space dadbd585cb3 fix format 7c34ebf93de [SPARC] fix the name of signal 19 in sparc arch b75648a7515 [MIPS] fix the name of signal 19 in mips 3840b09aae3 Rollup merge of rust-lang#128818 - RalfJung:std-miri-floats, r=tgross35 d03bb5e33a9 Rollup merge of rust-lang#128640 - RalfJung:rwlock-macos-miri, r=joboet 7680a3c7598 Rollup merge of rust-lang#128749 - tgross35:float-inline, r=scottmcm 9df61adfaa1 Rollup merge of rust-lang#128306 - WiktorPrzetacznik:WiktorPrzetacznik-nonnull-alignoffset-update, r=Amanieu 39860ad52d1 Update compiler-builtins version to 0.1.118 42811859e46 std float tests: special-case Miri in feature detection 4d6b36adfe6 Vxworks: Extern taskNameSet and fix build errors e24a6ca11fa rwlock: disable 'frob' test in Miri on macOS c21ba971a8a Fix VxWorks available parallelism: Move nonzero::uncheked into unsafe block 249541802ec Rollup merge of rust-lang#128800 - clarfonthey:core-pattern-type, r=compiler-errors 79cd72af482 Rollup merge of rust-lang#128691 - tgross35:update-builtins, r=Amanieu 8f840157d66 Add tracking issue to core-pattern-type b8f7f384f75 Stabilize `Ready::into_inner()` 62ccdeb315d Rollup merge of rust-lang#128261 - clarfonthey:iter-default, r=dtolnay b4e53303f07 alloc: make `to_string_str!` a bit less complex ec74467d64c Mark `{f32,f64}::{next_up,next_down,midpoint}` inline b90a026d6f8 Rollup merge of rust-lang#128766 - Monadic-Cat:patch-1, r=tgross35 5d7906c0270 Rollup merge of rust-lang#128417 - tgross35:f16-f128-math, r=dtolnay 83d1d167737 Trivial grammar fix in const keyword docs 97384fa701b Update `compiler-builtins` to 0.1.117 6dc79bb6235 Rollup merge of rust-lang#128751 - devnexen:vxworks_set_thread_name, r=tgross35 432425d28f7 Rollup merge of rust-lang#128539 - biabbas:deny_unsafe, r=workingjubilee 1bd5338eadf Rollup merge of rust-lang#128406 - lolbinarycat:bufreader_peek, r=Mark-Simulacrum e20aa6430f1 Rollup merge of rust-lang#125048 - dingxiangfei2009:stable-deref, r=amanieu bc13c6ca57a alloc: add ToString specialization for `&&str` 14fe723f6b9 std::thread: set_name implementation proposal for vxWorks. 67fa603356d Remove unused lifetime parameter from spawn_unchecked 4a3da122172 Add a special case for CStr/CString in the improper_ctypes lint 51ec2bb7ea2 implement BufReader::peek e6aede2233f nontemporal_store: make sure that the intrinsic is truly just a hint a300df74d13 WASI fixing unsafe_op_in_unsafe_fn for std::{os, sys} 59436fcc0b1 std: refactor UNIX random data generation 6fafc6b5d92 Apply review comments to PartialOrd section 7850a64f5bb Forbid unsafe_op_in_unsafe_fn in vxworks specific os and sys files e844efffe8f Add a disclaimer about x86 `f128` math functions 21d297b29ad Update comments for `{f16, f32, f64, f128}::midpoint` ad27d08e73e Add `core` functions for `f16` and `f128` that require math routines c6407b0bfa7 Add math functions for `f16` and `f128` d9b1de5180d Add math intrinsics for `f16` and `f128` 3c1586b3ce8 Hide internal sort module 21887129721 Apply review comments 2ebe00aa0ba PinCoerceUnsized trait into core 569ab6a3a03 CloneToUninit: use a private specialization trait 26874cc98cc Sparkle some attributes over `CloneToUninit` stuff e8c37187b60 impl CloneToUninit for Path and OsStr ef8c591ec02 impl CloneToUninit for str and CStr 65c6173bfe1 Update NonNull::align_offset quarantees b014b0d7b74 Improve panic sections for sort*, sort_unstable* and select_nth_unstable* 9bcfe84e72b Improve panic message and surrounding documentation for Ord violations 7e55abb1837 Okay, I guess I have to give these a different feature name bdc18e2ea2b impl Default for collection iterators that don't already have it 971aa37f27b LocalWaker docs: Make long-ago omitted but probably intended changes c4fdac9fe60 Docs for Waker and LocalWaker: Add cross-refs in comment 9c299bc6b10 Implement `debug_more_non_exhaustive` b405024dc09 Make use of raw strings in `core::fmt::builders` 20e64bd6cd3 Use is_val_statically_known to optimize pow 05ee32298cb Explicitly unroll integer pow for small exponents 4cfe24a3555 Optimize integer pow by removing exit branch 7c219da2111 Implement DoubleEnded and ExactSize for Take<Repeat> and Take<RepeatWith> git-subtree-dir: library git-subtree-split: febaf22f00e7290f42d3f6f50c4a9dbe8b4cbe5f
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. 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 rust-lang/stdarch#1541 propagating to the rustc repo, to make sure the_mm_stream
intrinsics are unaffected by this change.Fixes #114582
Cc @Amanieu @workingjubilee