We also sever syntax's dependency on rustc_target as a result.
This should slightly improve pipe-lining.
Moreover, some cleanup is done in related code.
Configure LLVM module PIC level
As of LLVM 9, this is required for 32-bit PowerPC to properly generate
PLT references. Previously, only BigPIC was supported; now LLVM supports
both BigPIC and SmallPIC, and there is no default value provided.
Rename `LocalInternedString` and more
This PR renames `LocalInternedString` as `SymbolStr`, removes an unnecessary `impl` from it, improves comments, and cleans up some `SymbolStr` uses.
r? @estebank
Allow specifying LLVM's MCTargetOptions::ABIName in target specification files
This addresses #65024, as it allows RISC-V target specification files to set `"llvm-abiname": "lp64d"`.
Other languages (read: C) usually expose this codegen parameter under a compiler argument like `-mabi=<XYZ>`.
Rollup of 10 pull requests
Successful merges:
- #65136 (Update codegen option documentation.)
- #65574 (docs: improve disclaimer regarding LinkedList)
- #65720 (Add FFI bindings for LLVM's Module::getInstructionCount())
- #65905 ([doc] fixes for unix/vxworks `OpenOptionsExt::mode`)
- #65962 (Fix logic in example.)
- #66019 (Improved std::iter::Chain documentation)
- #66038 (doc(str): show example of chars().count() under len())
- #66042 (Suggest correct code when encountering an incorrect trait bound referencing the current trait)
- #66073 (Do not needlessly write-lock)
- #66096 (Add a failing UI test for multiple loops of all kinds in a `const`)
Failed merges:
r? @ghost
Add FFI bindings for LLVM's Module::getInstructionCount()
Just to make it usable for profiling and such inside
rustc itself. It was vaguely useful in
https://wiki.alopex.li/WhereRustcSpendsItsTime and I figured
I might as well upstream it; I may or may not ever get around
to doing more with it (hopefully I will), but it may be useful
for others.
rustc_target: rename {Fn,Arg}Type to {Fn,Arg}Abi.
I was trying to tweak the API of `FnType` (now `FnAbi`) and the name kept bothering me.
`FnAbi` is to a function signature a bit like a layout is to a type, so the name still isn't perfect yet, but at least it doesn't have the misleading `Type` in it anymore.
If this can't land I think I can continue my original refactor without it, so I'm not strongly attached to it.
r? @nagisa cc @oli-obk
As of LLVM 9, this is required for 32-bit PowerPC to properly generate
PLT references. Previously, only BigPIC was supported; now LLVM supports
both BigPIC and SmallPIC, and there is no default value provided.
Just to make it useable for profiling and such inside
rustc itself. It was vaguely useful in
https://wiki.alopex.li/WhereRustcSpendsItsTime and I figured
I might as well upstream it; I may or may not ever get around
to doing more with it (hopefully I will), but it may be useful
for others.
This addresses #65024, as it allows RISC-V target specification
files to set "llvm-abiname": "lp64d". In general, it is useful
for the programmer to be able to set this codegen parameter,
which other languages usually expose under a compiler argument
like "-mabi=<XYZ>".
Re-enable Emscripten's exception handling support
Passes LLVM codegen and Emscripten link-time flags for exception
handling if and only if the panic strategy is `unwind`. Sets the
default panic strategy for Emscripten targets to `unwind`. Re-enables
tests that depend on unwinding support for Emscripten, including
`should_panic` tests.
r? @alexcrichton
Fix#64153
This PR changes how the compiler detects if an object file from an upstream crate is a Rust object file or not. Instead of checking if the name starts with the crate name and ends with `.o` (which is not always the case, as described in #64153), it now just checks if the filename ends with `.rcgu.o`.
This fixes#64153. However, ideally we'd clean up the code around filename generation some more. Then this check could be made more robust.
r? @alexcrichton
This allows us to remove `static_panic_msg` from the SSA<->LLVM
boundary, along with its fat pointer representation for &str.
Also changes the signature of PanicInfo::internal_contructor to
avoid copying.
Closes#65856.
Passes LLVM codegen and Emscripten link-time flags for exception
handling if and only if the panic strategy is `unwind`. Sets the
default panic strategy for Emscripten targets to `unwind`. Re-enables
tests that depend on unwinding support for Emscripten, including
`should_panic` tests.
Remove `InternedString`
This PR removes `InternedString` by converting all occurrences to `Symbol`. There are a handful of places that need to use the symbol chars instead of the symbol index, e.g. for stable sorting; local conversions `LocalInternedString` is used in those places.
r? @eddyb
Remove `src/llvm-emscripten` submodule
With #65251 landed there's no need to build two LLVM backends and ship
them with rustc, every target we have now uses the same LLVM backend!
This removes the `src/llvm-emscripten` submodule and additionally
removes all support from rustbuild for building the emscripten LLVM
backend. Multiple codegen backend support is left in place for now, and
this is intended to be an easy 10-15 minute win on CI times by avoiding
having to build LLVM twice.
With #65251 landed there's no need to build two LLVM backends and ship
them with rustc, every target we have now uses the same LLVM backend!
This removes the `src/llvm-emscripten` submodule and additionally
removes all support from rustbuild for building the emscripten LLVM
backend. Multiple codegen backend support is left in place for now, and
this is intended to be an easy 10-15 minute win on CI times by avoiding
having to build LLVM twice.
It's a full conversion, except in `DefKey::compute_stable_hash()` where
a `Symbol` now is converted to an `InternedString` before being hashed.
This was necessary to avoid test failures.
- Compatible with Emscripten 1.38.46-upstream or later upstream.
- Refactors the Emscripten target spec to share code with other wasm
targets.
- Replaces the old incorrect wasm32 C call ABI with the correct one,
preserving the old one as wasm32_bindgen_compat for wasm-bindgen
compatibility.
- Updates the varargs ABI used by Emscripten and deletes the old one.
- Removes the obsolete wasm32-experimental-emscripten target.
- Uses EMCC_CFLAGS on CI to avoid the timeout problems with #63649.
Several changes to the codegen backend organization
* Split functions from values in cg_ssa `BackendTypes`.
* Remove `is_const_integral` function from `ConstMethods`.
* Actually register the invalid monomorphization of intrinsic long diagnostic and remove the `diagnostics` method from `CodegenBackends`, as it was unused.
* Add cg_ssa and cg_utils provided methods to `default_provide`, so codegen backend don't have to do it themself.
Always mark rust and rust-call abi's as unwind
PR #63909 identified a bug that had been injected by PR #55982. As discussed on https://github.com/rust-lang/rust/issues/64655#issuecomment-537517428 , we started marking extern items as nounwind, *even* extern items that said they were using "Rust" or "rust-call" ABI.
This is a more targeted variant of PR #63909 that fixes the above bug.
Fix#64655
----
I personally suspect we will want PR #63909 to land in the long-term
But:
* it is not certain that PR #63909 *will* land,
* more importantly, PR #63909 almost certainly will not be backported to beta/stable.
The identified bug was more severe than I think anyone realized (apart from perhaps @gnzlbg, as noted [here](https://github.com/rust-lang/rust/pull/63909#issuecomment-524818838)).
Thus, I was motivated to write this PR, which fixes *just* the issue with extern rust/rust-call functions, and deliberately avoids injecting further deviation from current behavior (you can see further notes on this in the comments of the code added here).
Add llvm.sideeffect to potential infinite loops and recursions
LLVM assumes that a thread will eventually cause side effect. This is
not true in Rust if a loop or recursion does nothing in its body,
causing undefined behavior even in common cases like `loop {}`.
Inserting llvm.sideeffect fixes the undefined behavior.
As a micro-optimization, only insert llvm.sideeffect when jumping back
in blocks or calling a function.
A patch for LLVM is expected to allow empty non-terminate code by
default and fix this issue from LLVM side.
https://github.com/rust-lang/rust/issues/28728
**UPDATE:** [Mentoring instructions here](https://github.com/rust-lang/rust/pull/59546#issuecomment-515072429) to unstall this PR
Disable stack probe when thread sanitizer is enabled
When thread sanitizer instrumentation is enabled during compilation of
stack probe function, the function will be miscompiled and trigger
segmentation fault at runtime. Disable stack probes when tsan is
enabled.
Revert #63649 - "Upgrade Emscripten targets to use upstream LLVM backend"
This change caused the runtime of the linux-asmjs builder to nearly double from 2+ hours to about 4 hours, which happens to be the bors timeout. (It made it in barely under 4 hours when it was merged.) This is causing timeouts on all new changes.
This reverts commit 7870050796, reversing
changes made to 2e7244807a.
Upgrade Emscripten targets to use upstream LLVM backend
- Refactors the Emscripten target spec to share code with other wasm
targets.
- Replaces the incorrect wasm32 C call ABI with the old asmjs
version, which is correct for both wasm32 and JS.
- Updates the varargs ABI used by Emscripten and deletes the old one.
- Removes the obsolete wasm32-experimental-emscripten target.
- Temporarily makes Emscripten targets use panic=abort by default
because supporting unwinding will require an LLVM patch.
When thread sanitizer instrumentation is enabled during compilation of
stack probe function, the function will be miscompiled and trigger
segmentation fault at runtime. Disable stack probes when tsan is
enabled.
- Refactors the Emscripten target spec to share code with other wasm
targets.
- Replaces the incorrect wasm32 C call ABI with the old asmjs
version, which is correct for both wasm32 and JS.
- Updates the varargs ABI used by Emscripten and deletes the old one.
- Removes the obsolete wasm32-experimental-emscripten target.
- Temporarily makes Emscripten targets use panic=abort by default
because supporting unwinding will require an LLVM patch.
SelfProfiler API refactoring and part one of event review
This PR refactors the `SelfProfiler` a little bit so that most profiling methods are RAII-based. The codegen backend code already had something similar, this refactoring pulls this functionality up into `SelfProfiler` itself, for general use.
The second commit of this PR is a review and update of the existing events we are already recording. Names have been made more consistent. CGU names have been removed from event names. They will be added back in when function parameter recording is implemented.
There is still some work to be done for adding new events, especially around trait resolution and the incremental system.
r? @wesleywiser
No StableHasherResult everywhere
This removes the generic parameter on `StableHasher`, instead moving it to the call to `finish`. This has the side-effect of making all `HashStable` impls nicer, since we no longer need the verbose `<W: StableHasherResult>` that previously existed -- often forcing line wrapping.
This is done for two reasons:
* we should avoid false "generic" dependency on the result of StableHasher
* we don't need to codegen two/three copies of all the HashStable impls when they're transitively used to produce a fingerprint, u64, or u128. I haven't measured, but this might actually make our artifacts somewhat smaller too.
* Easier to understand/read/write code -- the result of the stable hasher is irrelevant when writing a hash impl.
LLVM assumes that a thread will eventually cause side effect. This is
not true in Rust if a loop or recursion does nothing in its body,
causing undefined behavior even in common cases like `loop {}`.
Inserting llvm.sideeffect fixes the undefined behavior.
As a micro-optimization, only insert llvm.sideeffect when jumping back
in blocks or calling a function.
A patch for LLVM is expected to allow empty non-terminate code by
default and fix this issue from LLVM side.
https://github.com/rust-lang/rust/issues/28728
Rename `*.node` to `*.kind`, and `hair::Pattern*` to `hair::Pat*`
In both `ast::Expr` and `hir::Expr`:
- Rename `Expr.node` to `Expr.kind`.
- Rename `Pat.node` to `Pat.kind`.
- Rename `ImplItem.node` to `ImplItem.kind`.
- Rename `Lit.node` to `Lit.kind`.
- Rename `TraitItem.node` to `TraitItem.kind`.
- Rename `Ty.node` to `Ty.kind`.
- Rename `Stmt.node` to `Stmt.kind`.
- Rename `Item.node` to `Item.kind`.
- Rename `ForeignItem.node` to `ForeignItem.kind`.
- Rename `MetaItem.node` to `MetaItem.kind`.
Also:
- Rename `hair::FieldPattern` to `hair::FieldPat`.
- Rename `hair::PatternKind` to `hair::PatKind`.
- Rename `hair::PatternRange` to `hair::PatRange`.
- Rename `PatternContext` to `PatCtxt`.
- Rename `PatternTypeProjection` to `PatTyProj`.
- Rename `hair::Pattern` to `hair::Pat`.
These two sets of changes are grouped together to aid with merging. The only changes are renamings.
r? @petrochenkov
Remove tx_to_llvm_workers from TyCtxt
This can be kept within the codegen backend crates entirely -- there's no reason for us to create it outside and attempt to hold it in the (global) context.
Changes here aren't really too easily reviewable I suspect -- not sure if they can be cleaned up by splitting into more commits though, it's just hard to reason about `Box<Any>` in general. If there are thoughts though I'd be happy to hear them.
The primary goal of this PR is to get rid of the field on `rustc_interface::Queries`.
Not doing this leads to building two copies of e.g. num_cpus in the
sysroot and _llvm deps, leading to conflicts between the two when
compiling librustc_codegen_llvm. It's not entirely clear why this is the
case after the changes in this PR but likely has something to do with a
subtle difference in ordering or similar.
place: Passing `align` = `layout.align.abi`, when also passing `layout`
Of the calls changed:
7/12 use `align` = `layout.align.abi`.
`from_const_alloc` uses `alloc.align`, but that is `assert_eq!` to `layout.align.abi`.
only 4/11 use something interesting for `align`.
Minimize uses of `LocalInternedString`
`LocalInternedString` is described as "An alternative to `Symbol` and `InternedString`, useful when the chars within the symbol need to be accessed. It is best used for temporary values."
This PR makes the code match that comment, by removing all non-local uses of `LocalInternedString`. This allows the removal of a number of operations on `LocalInternedString` and a couple of uses of `unsafe`.
Make Allocation::bytes private
Fixes#62931.
Direct immutable access to the bytes is still possible but redirected through the new method `raw_bytes_with_undef_and_ptr`, similar to `get_bytes_with_undef_and_ptr` but without requiring an interpretation context and not doing *any* relocation or bounds checks. The `size` of the allocation is stored separately which makes access as `Size` and `usize` more ergonomic.
cc: @RalfJung
so rename it `new_sized_aligned`.
6/11 use `align` = `layout.align.abi`.
`from_const_alloc` uses `alloc.align`, but that is `assert_eq!` to `layout.align.abi`.
only 4/11 use something interesting for `align`.
rustc: Handle modules in "fat" LTO more robustly
When performing a "fat" LTO the compiler has a whole mess of codegen
units that it links together. To do this it needs to select one module
as a "base" module and then link everything else into this module.
Previously LTO passes assume that there's at least one module in-memory
to link into, but nowadays that's not always true! With incremental
compilation modules may actually largely be cached and it may be
possible that there's no in-memory modules to work with.
This commit updates the logic of the LTO backend to handle modules a bit
more uniformly during a fat LTO. This commit immediately splits them
into two lists, one serialized and one in-memory. The in-memory list is
then searched for the largest module and failing that we simply
deserialize the first serialized module and link into that. This
refactoring avoids juggling three lists, two of which are serialized
modules and one of which is half serialized and half in-memory.
Closes#63349
When performing a "fat" LTO the compiler has a whole mess of codegen
units that it links together. To do this it needs to select one module
as a "base" module and then link everything else into this module.
Previously LTO passes assume that there's at least one module in-memory
to link into, but nowadays that's not always true! With incremental
compilation modules may actually largely be cached and it may be
possible that there's no in-memory modules to work with.
This commit updates the logic of the LTO backend to handle modules a bit
more uniformly during a fat LTO. This commit immediately splits them
into two lists, one serialized and one in-memory. The in-memory list is
then searched for the largest module and failing that we simply
deserialize the first serialized module and link into that. This
refactoring avoids juggling three lists, two of which are serialized
modules and one of which is half serialized and half in-memory.
Closes#63349
Closure types have been moved to the namespace where they
are defined, and both closure and generator type names now
include the disambiguator.
This fixes an exception when lldb prints nested closures.
Fixes#57822
Do not generate allocations for zero sized allocations
Alternative to https://github.com/rust-lang/rust/issues/62487
r? @eddyb
There are other places where we could do this, too, but that would cause `static FOO: () = ();` to not have a unique address
Deduplicate rustc_demangle in librustc_codegen_llvm
This commit removes the crates.io dependency of `rustc-demangle` from
`rustc_codegen_llvm`. This crate is actually already pulled in to part
of the `librustc_driver` build and with the upcoming pipelining
implementation in Cargo it causes build issues if `rustc-demangle` is
left to its own devices.
This is not currently required, but once pipelining is enabled for
rustc's own build it will be required to build correctly.
