-
Notifications
You must be signed in to change notification settings - Fork 12.8k
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
Experimental JIT compiler #3292
Conversation
This looks great. Linking rustllvm to morestack turns out to be problematic because morestack expects two symbols from the Rust runtime to exist. Apparently this isn't a problem on linux, but on mac it doesn't link. See the errors on mac here: http://bot.rust-lang.org/logs/2012/07/28/2012-08-28T18:37:31Z-f5a3db41-389e-47f0-ab45-e35d9374bac0.html We can possibly work around this further by linking rustllvm to rustrt. |
For the licensing issue, probably what we should do is add a comment near the code saying where it is copied from, then in LICENSE.txt add some text like 'Portions of the code for integrating LLVM, under src/rustllvm, is derived from the LLVM codebase, under the following license.' then list the whole license. |
Here's a patch that makes the address of __morestack available through the 'morestack_addr' intrinsic. It looks to me like it works correctly. You could use from within rustc to grab the correct address and pass to rustllvm. |
This commit adds JIT testing support to the main test harness: Use it with |
OK, this commit runs the run-fail tests through the JIT as well, and they all fail with the 'terminate called after throwing ...' error. |
We're going to need to think a bit about what it is that the JIT should actually do. What we're doing here effictively starts an entirely new runtime by calling "main". This will have some subtle consequences, partly because the Rust runtime isn't entirely threadsafe when there are multiple instances, and also because the TLS usage of the second runtime will, in some cases, end up clobbering the TLS of the original runtime. More importantly though, what we really want is to run the JIT, then get the address of the Rust main function, then begin executing that function from within the Rust stack. This is all for later though. |
I understand, I didn't think about how there would be two instances of the runtime. Do you think starting two runtimes could have something to do with the terminate error? WRT spawning it inside the compiler's runtime, I don't know how to do that, so if you want to do it / point me to the correct way I'll be happy to. Also can you explain the significance of the intrinsic commit? I don't know what it does or has to be done to make the symbol available. EDIT: I think I understand. The intrinsic is a custom ABI that I can use to pass the address to rustllvm. |
"rust-intrinsic" is a special ABI for native functions that refers to a few built-in functions that the compiler can emit on demand. It's mostly for very low-level stuff like getting the alignment of types. The 'morestack_addr' intrinsic tells rustc to create a function that just returns the address of I do think that the extra runtime could be related to the terminate error, though I haven't imagined exactly how. It looks to me that the Rust ABI main function is called "_rust_main", so instead of asking for the address of "main", then executing it inside of rustllvm, we should be able to ask for "_rust_main", return it to rustc, then rustc can use it to synthesize a closure without an environment and call it. This commit has an example of synthesizing a closure: 6e20ffe |
…king in libmorestack and return _rust_main and call it from rustc
Seems to be working fine! If someone can test this fixed the build issues on OSX, that would be great. I'm trying to get crate loading working and then everything should be good. EDIT: The error message is |
I ran your latest revision through the try branch on the bots and they were mostly happy. The windows bot failed but I suspect it's unrelated to your changes. Unfortunately I've been busy with other things and won't be able to merge this until tomorrow. |
Pushed to incoming. |
Automatic Rustup
Update Rust toolchain from nightly-2024-06-24 to nightly-2024-06-25 without any other source changes. This is an automatically generated pull request. If any of the CI checks fail, manual intervention is required. In such a case, review the changes at https://github.com/rust-lang/rust from rust-lang@bcf94de up to rust-lang@6b0f4b5. The log for this commit range is: rust-lang@6b0f4b5ec3 Auto merge of rust-lang#126914 - compiler-errors:rollup-zx0hchm, r=compiler-errors rust-lang@16bd6e25e1 Rollup merge of rust-lang#126911 - oli-obk:do_not_count_errors, r=compiler-errors rust-lang@59c258f51f Rollup merge of rust-lang#126909 - onur-ozkan:add-kobzol, r=matthiaskrgr rust-lang@85eb835a14 Rollup merge of rust-lang#126904 - GrigorenkoPV:nonzero-fixme, r=joboet rust-lang@a7721a0373 Rollup merge of rust-lang#126899 - GrigorenkoPV:suggest-const-block, r=davidtwco rust-lang@9ce2a070b3 Rollup merge of rust-lang#126682 - Zalathar:coverage-attr, r=lcnr rust-lang@49bdf460a2 Rollup merge of rust-lang#126673 - oli-obk:dont_rely_on_err_reporting, r=compiler-errors rust-lang@46e43984d1 Rollup merge of rust-lang#126413 - matthiaskrgr:crshmsg, r=oli-obk rust-lang@ed460d2eaa Rollup merge of rust-lang#125575 - dingxiangfei2009:derive-smart-ptr, r=davidtwco rust-lang@c77dc28f87 Rollup merge of rust-lang#125082 - kpreid:const-uninit, r=dtolnay rust-lang@faa28be2f1 Rollup merge of rust-lang#124712 - Enselic:deprecate-inline-threshold, r=pnkfelix rust-lang@00e5f5886a Rollup merge of rust-lang#124460 - long-long-float:show-notice-about-enum-with-debug, r=pnkfelix rust-lang@d8d5732456 Auto merge of rust-lang#126784 - scottmcm:smaller-terminator, r=compiler-errors rust-lang@13fca73f49 Replace `MaybeUninit::uninit_array()` with array repeat expression. rust-lang@5a3e2a4e92 Auto merge of rust-lang#126523 - joboet:the_great_big_tls_refactor, r=Mark-Simulacrum rust-lang@45261ff2ec add @Kobzol to bootstrap team for triagebot rust-lang@84474a25a4 Small fixme in core now that NonZero is generic rust-lang@50a02ed789 std: fix wasm builds rust-lang@8fc6b3de19 Separate the mir body lifetime from the other lifetimes rust-lang@1c4d0ced58 Separate the lifetimes of the `BorrowckInferCtxt` from the other borrowed items rust-lang@d371d17496 Auto merge of rust-lang#126900 - matthiaskrgr:rollup-24ah97b, r=matthiaskrgr rust-lang@8ffb5f936a compiletest: make the crash test error message abit more informative rust-lang@a80ee9159b Rollup merge of rust-lang#126882 - estebank:multiline-order, r=WaffleLapkin rust-lang@8bfde609e2 Rollup merge of rust-lang#126414 - ChrisDenton:target-known, r=Nilstrieb rust-lang@94b9ea417d Rollup merge of rust-lang#126213 - zachs18:atomicbool-u8-i8-from-ptr-alignment, r=Nilstrieb rust-lang@9d24ecc37b Rollup merge of rust-lang#125241 - Veykril:tool-rust-analyzer, r=davidtwco rust-lang@ba5ec1fc5c Suggest inline const blocks for array initialization rust-lang@06c072f158 Auto merge of rust-lang#126788 - GuillaumeGomez:migrate-rustdoc-tests-syntax, r=fmease,oli-obk rust-lang@1852141219 coverage: Bless coverage attribute tests rust-lang@b7c057c9b2 coverage: Always error on `#[coverage(..)]` in unexpected places rust-lang@a000fa8b54 coverage: Tighten validation of `#[coverage(off)]` and `#[coverage(on)]` rust-lang@b5dfeba0e1 coverage: Forbid multiple `#[coverage(..)]` attributes rust-lang@6909feab8e Allow numbers in rustdoc tests commands rust-lang@4e258bb4c3 Fix tidy issue for rustdoc tests commands rust-lang@51fedf65ff Remove commands duplication between `compiletest` and `tests/rustdoc` rust-lang@1b67035579 Update `tests/rustdoc` to new test syntax rust-lang@d3ec92e16e Move `tests/rustdoc` testsuite to `//@` syntax rust-lang@2c243d9570 Auto merge of rust-lang#126891 - matthiaskrgr:rollup-p6dl1gk, r=matthiaskrgr rust-lang@b94d2754b5 Rollup merge of rust-lang#126888 - compiler-errors:oops-debug-printing, r=dtolnay rust-lang@9892b3e9fe Rollup merge of rust-lang#126854 - devnexen:std_unix_os_fallback_upd, r=Mark-Simulacrum rust-lang@3108dfaced Rollup merge of rust-lang#126849 - workingjubilee:correctly-classify-arm-low-dregs, r=Amanieu rust-lang@dcace866f0 Rollup merge of rust-lang#126845 - rust-lang:cargo_update, r=Mark-Simulacrum rust-lang@21850f5bd8 Rollup merge of rust-lang#126807 - devnexen:copy_file_macos_simpl, r=Mark-Simulacrum rust-lang@b24e3df0df Rollup merge of rust-lang#126754 - compiler-errors:use-rustfmt, r=calebcartwright rust-lang@ad0531ae0d Rollup merge of rust-lang#126455 - surechen:fix_126222, r=estebank rust-lang@7babf99ea9 Rollup merge of rust-lang#126298 - heiher:loongarch64-musl-ci, r=Mark-Simulacrum rust-lang@9a591ea1ce Rollup merge of rust-lang#126177 - carbotaniuman:unsafe_attr_errors, r=jieyouxu rust-lang@25446c25fc Remove stray println from rustfmt rust-lang@d49994b060 Auto merge of rust-lang#126023 - amandasystems:you-dropped-this-again, r=nikomatsakis rust-lang@a23917cfd0 Add hard error and migration lint for unsafe attrs rust-lang@284437d434 Special case when a code line only has multiline span starts rust-lang@f1be59fa72 SmartPointer derive-macro rust-lang@a426d6fdf0 Implement use<> formatting in rustfmt rust-lang@16fef40896 Promote loongarch64-unknown-linux-musl to Tier 2 with host tools rust-lang@03d73fa6ba ci: Add support for dist-loongarch64-musl rust-lang@fc50acae90 fix build rust-lang@bd9ce3e074 std::unix::os::home_dir: fallback's optimisation. rust-lang@0d8f734172 compiler: Fix arm32 asm issues by hierarchically sorting reg classes rust-lang@e8b5ba1111 For [E0308]: mismatched types, when expr is in an arm's body, not add semicolon ';' at the end of it. rust-lang@990535723d cargo update rust-lang@b28efb11af Save 2 pointers in `TerminatorKind` (96 → 80 bytes) rust-lang@65530ba100 std::unix::fs: copy simplification for apple. rust-lang@339015920d Add `rust_analyzer` as a predefined tool rust-lang@3f2f8438b4 Ensure we don't accidentally succeed when we want to report an error rust-lang@32f9b8bf76 std: rename module for clarity rust-lang@35f050b8da std: update TLS module documentation rust-lang@b2f29edc81 std: use the `c_int` from `core::ffi` instead of `libc` rust-lang@d70f071392 std: simplify `#[cfg]`s for TLS rust-lang@d630f5da7a Show notice about "never used" for enum rust-lang@f3facf1175 std: refactor the TLS implementation rust-lang@f5f067bf9d Deprecate no-op codegen option `-Cinline-threshold=...` rust-lang@651ff643ae Fix typo in `-Cno-stack-check` deprecation warning rust-lang@3af624272a rustc_codegen_ssa: Remove unused ModuleConfig::inline_threshold rust-lang@34e6ea1446 Tier 2 std support must always be known rust-lang@2d4cb7aa5a Update docs for AtomicU8/I8. rust-lang@7885c7b7b2 Update safety docs for AtomicBool::from_ptr. rust-lang@7b5b7a7010 Remove confusing `use_polonius` flag and do less cloning Co-authored-by: tautschnig <[email protected]>
This is my 3rd attempt to get this working over around 4 months. It's working pretty well compared to last attempts.
I've found that it runs perfectly (slow parsing/compiling times, fast executing times) with most tests running fine. Most benchmarks actually run faster than statically compiled executables! Of course, it's not perfect yet and there are some issues:
llvm::sys::DynamicLibrary::LoadLibraryPermanently
is segfaulting when I try and load crates. The code to do this (pretty simple) is commented out and ready to be fixed (I imagine the fix is something simple)std
andcore
are the only crates available because of above - they're only available because the JIT compiler usesdlsym
to resolve symbols from the main program and any loaded crates - hence it loads thestd
andcore
crates used byrustc
. This is not a good idea, as users may want to use specific versions of the core libraries, so it has to be fixed with above--test
results interminate called after throwing an instance of 'rust_task*'
- I do not know enough to understand what's causing thatThese issues can easily be fixed in patches (I'll investigate the first two issues) but I'm not sure about the 3rd one. #rust were pretty keen to see this pulled upstream and I think it's ready for that.
Also important to mention is I based the memory manager source code on LLVM's
lli
tool (most of the original comments remain) because it worked well with what was needed and it had fixed issues that I knew were going to be hit (also it was the only good example of JIT memory managers that I could find - MCJIT managers require them in order to run). Not sure what sort of things need to be done there in terms of licensing / credits / etc.