| Age | Commit message (Collapse) | Author | Lines |
|---|
|
modifiers with custom consistency check function
|
|
This is intended to be used for Linux kernel RETPOLINE builds.
Signed-off-by: Miguel Ojeda <ojeda@kernel.org>
|
|
|
|
|
|
|
|
This flag turned out to be less useful than anticipated, and interferes with
work towards expansion support.
|
|
Nobody seems to actually use this, while still adding some extra
complexity to the already rather complex codegen coordinator code.
It is also not supported by any backend other than the LLVM backend.
|
|
Use serde for target spec json deserialize
The previous manual parsing of `serde_json::Value` was a lot of complicated code and extremely error-prone. It was full of janky behavior like sometimes ignoring type errors, sometimes erroring for type errors, sometimes warning for type errors, and sometimes just ICEing for type errors (the icing on the top).
Additionally, many of the error messages about allowed values were out of date because they were in a completely different place than the FromStr impls. Overall, the system caused confusion for users.
I also found the old deserialization code annoying to read. Whenever a `key!` invocation was found, one had to first look for the right macro arm, and no go to definition could help.
This PR replaces all this manual parsing with a 2-step process involving serde.
First, the string is parsed into a `TargetSpecJson` struct. This struct is a 1:1 representation of the spec JSON. It already parses all the enums and is very simple to read and write.
Then, the fields from this struct are copied into the actual `Target`. The reason for this two-step process instead of just serializing into a `Target` is because of a few reasons
1. There are a few transformations performed between the two formats
2. The default logic is implemented this way. Otherwise all the default field values would have to be spelled out again, which is suboptimal. With this logic, they fall out naturally, because everything in the json struct is an `Option`.
Overall, the mapping is pretty simple, with the vast majority of fields just doing a 1:1 mapping that is captured by two macros. I have deliberately avoided making the macros generic to keep them simple.
All the `FromStr` impls now have the error message right inside them, which increases the chance of it being up to date. Some "`from_str`" impls were turned into proper `FromStr` impls to support this.
The new code is much less involved, delegating all the JSON parsing logic to serde, without any manual type matching.
This change introduces a few breaking changes for consumers. While it is possible to use this format on stable, it is very much subject to change, so breaking changes are expected. The hope is also that because of the way stricter behavior, breaking changes are easier to deal with, as they come with clearer error messages.
1. Invalid types now always error, everywhere. Previously, they would sometimes error, and sometimes just be ignored (which meant the users JSON was still broken, just silently!)
2. This now makes use of `deny_unknown_fields` instead of just warning on unused fields, which was done previously. Serde doesn't make it easy to get such warning behavior, which was the primary reason that this now changed. But I think error behavior is very reasonable too. If someone has random stale fields in their JSON, it is likely because these fields did something at some point but no longer do, and the user likely wants to be informed of this so they can figure out what to do.
This is also relevant for the future. If we remove a field but someone has it set, it probably makes sense for them to take a look whether they need this and should look for alternatives, or whether they can just delete it. Overall, the JSON is made more explicit.
This is the only expected breakage, but there could also be small breakage from small mistakes. All targets roundtrip though, so it can't be anything too major.
fixes rust-lang/rust#144153
|
|
The previous manual parsing of `serde_json::Value` was a lot of
complicated code and extremely error-prone. It was full of janky
behavior like sometimes ignoring type errors, sometimes erroring for
type errors, sometimes warning for type errors, and sometimes just
ICEing for type errors (the icing on the top).
Additionally, many of the error messages about allowed values were out
of date because they were in a completely different place than the
FromStr impls. Overall, the system caused confusion for users.
I also found the old deserialization code annoying to read. Whenever a
`key!` invocation was found, one had to first look for the right macro
arm, and no go to definition could help.
This PR replaces all this manual parsing with a 2-step process involving
serde.
First, the string is parsed into a `TargetSpecJson` struct. This struct
is a 1:1 representation of the spec JSON. It already parses all the
enums and is very simple to read and write.
Then, the fields from this struct are copied into the actual `Target`.
The reason for this two-step process instead of just serializing into a
`Target` is because of a few reasons
1. There are a few transformations performed between the two formats
2. The default logic is implemented this way. Otherwise all the default
field values would have to be spelled out again, which is
suboptimal. With this logic, they fall out naturally, because
everything in the json struct is an `Option`.
Overall, the mapping is pretty simple, with the vast majority of fields
just doing a 1:1 mapping that is captured by two macros. I have
deliberately avoided making the macros generic to keep them simple.
All the `FromStr` impls now have the error message right inside them,
which increases the chance of it being up to date. Some "`from_str`"
impls were turned into proper `FromStr` impls to support this.
The new code is much less involved, delegating all the JSON parsing
logic to serde, without any manual type matching.
This change introduces a few breaking changes for consumers. While it is
possible to use this format on stable, it is very much subject to
change, so breaking changes are expected. The hope is also that because
of the way stricter behavior, breaking changes are easier to deal with,
as they come with clearer error messages.
1. Invalid types now always error, everywhere. Previously, they would
sometimes error, and sometimes just be ignored (which meant the users
JSON was still broken, just silently!)
2. This now makes use of `deny_unknown_fields` instead of just warning
on unused fields, which was done previously. Serde doesn't make it
easy to get such warning behavior, which was the primary reason that
this now changed. But I think error behavior is very reasonable too.
If someone has random stale fields in their JSON, it is likely
because these fields did something at some point but no longer do,
and the user likely wants to be informed of this so they can figure
out what to do.
This is also relevant for the future. If we remove a field but
someone has it set, it probably makes sense for them to take a look
whether they need this and should look for alternatives, or whether
they can just delete it. Overall, the JSON is made more explicit.
This is the only expected breakage, but there could also be small
breakage from small mistakes. All targets roundtrip though, so it can't
be anything too major.
|
|
code generation
|
|
|
|
This stabilizes a subset of the `-Clinker-features` components on x64 linux:
the lld opt-out.
The opt-in is not stabilized, as interactions with other stable flags require
more internal work, but are not needed for stabilizing using rust-lld by default.
Similarly, since we only switch to rust-lld on x64 linux, the opt-out is
only stabilized there. Other targets still require `-Zunstable-options`
to use it.
|
|
Add PrintTAFn flag for targeted type analysis printing
## Summary
This PR adds a new `PrintTAFn` flag to the `-Z autodiff` option that allows printing type analysis information for a specific function, rather than all functions.
## Changes
### New Flag
- Added `PrintTAFn=<function_name>` option to `-Z autodiff`
- Usage: `-Z autodiff=Enable,PrintTAFn=my_function_name`
### Implementation Details
- **Rust side**: Added `PrintTAFn(String)` variant to `AutoDiff` enum
- **Parser**: Updated `parse_autodiff` to handle `PrintTAFn=<function_name>` syntax with proper error handling
- **FFI**: Added `set_print_type_fun` function to interface with Enzyme's `FunctionToAnalyze` command line option
- **Documentation**: Updated help text and documentation for the new flag
### Files Modified
- `compiler/rustc_session/src/config.rs`: Added `PrintTAFn(String)` variant
- `compiler/rustc_session/src/options.rs`: Updated parser and help text (now shows `PrintTAFn` in the list)
- `compiler/rustc_codegen_llvm/src/llvm/enzyme_ffi.rs`: Added FFI function and static variable
- `compiler/rustc_codegen_llvm/src/back/lto.rs`: Added handling for new flag
- `src/doc/rustc-dev-guide/src/autodiff/flags.md`: Updated documentation
- `src/doc/unstable-book/src/compiler-flags/autodiff.md`: Updated documentation
## Testing
The flag can be tested with:
```bash
rustc +enzyme -Z autodiff=Enable,PrintTAFn=square test.rs
```
This will print type analysis information only for the function named "square" instead of all functions.
## Error Handling
The parser includes proper error handling:
- Missing argument: `PrintTAFn` without `=<function_name>` will show an error
- Unknown options: Invalid autodiff options will be reported
r? ```@ZuseZ4```
|
|
Signed-off-by: Karan Janthe <karanjanthe@gmail.com>
|
|
Also avoid creating and cloning sysroot unnecessarily.
|
|
|
|
Try unremapping compiler sources
See [#t-compiler/help > Span pointing to wrong file location (`rustc-dev` component)](https://rust-lang.zulipchat.com/#narrow/channel/182449-t-compiler.2Fhelp/topic/Span.20pointing.20to.20wrong.20file.20location.20.28.60rustc-dev.60.20component.29/with/521087083).
This PR is a follow-up to rust-lang/rust#141751 regarding the compiler side.
Specifically we now take into account the `CFG_VIRTUAL_RUSTC_DEV_SOURCE_BASE_DIR` env from rust-lang/rust#141751 when trying to unremap sources from `$sysroot/lib/rustlib/rustc-src/rust` (the `rustc-dev` component install directory).
Best reviewed commit by commit.
cc ``@samueltardieu``
r? ``@jieyouxu``
|
|
Implement initial support for timing sections (`--json=timings`)
This PR implements initial support for emitting high-level compilation section timings. The idea is to provide a very lightweight way of emitting durations of various compilation sections (frontend, backend, linker, or on a more granular level macro expansion, typeck, borrowck, etc.). The ultimate goal is to stabilize this output (in some form), make Cargo pass `--json=timings` and then display this information in the HTML output of `cargo build --timings`, to make it easier to quickly profile "what takes so long" during the compilation of a Cargo project. I would personally also like if Cargo printed some of this information in the interactive `cargo build` output, but the `build --timings` use-case is the main one.
Now, this information is already available with several other sources, but I don't think that we can just use them as they are, which is why I proposed a new way of outputting this data (`--json=timings`):
- This data is available under `-Zself-profile`, but that is very expensive and forever unstable. It's just a too big of a hammer to tell us the duration it took to run the linker.
- It could also be extracted with `-Ztime-passes`. That is pretty much "for free" in terms of performance, and it can be emitted in a structured form to JSON via `-Ztime-passes-format=json`. I guess that one alternative might be to stabilize this flag in some form, but that form might just be `--json=timings`? I guess what we could do in theory is take the already emitted time passes and reuse them for `--json=timings`. Happy to hear suggestions!
I'm sending this PR mostly for a vibeck, to see if the way I implemented it is passable. There are some things to figure out:
- How do we represent the sections? Originally I wanted to output `{ section, duration }`, but then I realized that it might be more useful to actually emit `start` and `end` events. Both because it enables to see the output incrementally (in case compilation takes a long time and you read the outputs directly, or Cargo decides to show this data in `cargo build` some day in the future), and because it makes it simpler to represent hierarchy (see below). The timestamps currently emit microseconds elapsed from a predetermined point in time (~start of rustc), but otherwise they are fully opaque, and should be only ever used to calculate the duration using `end - start`. We could also precompute the duration for the user in the `end` event, but that would require doing more work in rustc, which I would ideally like to avoid :P
- Do we want to have some form of hierarchy? I think that it would be nice to show some more granular sections rather than just frontend/backend/linker (e.g. macro expansion, typeck and borrowck as a part of the frontend). But for that we would need some way of representing hierarchy. A simple way would be something like `{ parent: "frontend" }`, but I realized that with start/end timestamps we get the hierarchy "for free", only the client will need to reconstruct it from the order of start/end events (e.g. `start A`, `start B` means that `B` is a child of `A`).
- What exactly do we want to stabilize? This is probably a question for later. I think that we should definitely stabilize the format of the emitted JSON objects, and *maybe* some specific section names (but we should also make it clear that they can be missing, e.g. you don't link everytime you invoke `rustc`).
The PR be tested e.g. with `rustc +stage1 src/main.rs --json=timings --error-format=json -Zunstable-options` on a crate without dependencies (it is not easy to use `--json` with stock Cargo, because it also passes this flag to `rustc`, so this will later need Cargo integration to be usable with it).
Zulip discussions: [#t-compiler > Outputting time spent in various compiler sections](https://rust-lang.zulipchat.com/#narrow/channel/131828-t-compiler/topic/Outputting.20time.20spent.20in.20various.20compiler.20sections/with/518850162)
MCP: https://github.com/rust-lang/compiler-team/issues/873
r? ``@nnethercote``
|
|
Add `-Z hint-mostly-unused` to tell rustc that most of a crate will go unused
This hint allows the compiler to optimize its operation based on this assumption, in order to compile faster. This is a hint, and does not guarantee any particular behavior.
This option can substantially speed up compilation if applied to a large dependency where the majority of the dependency does not get used. This flag may slow down compilation in other cases.
Currently, this option makes the compiler defer as much code generation as possible from functions in the crate, until later crates invoke those functions. Functions that never get invoked will never have code generated for them. For instance, if a crate provides thousands of functions, but only a few of them will get called, this flag will result in the compiler only doing code generation for the called functions. (This uses the same mechanisms as cross-crate inlining of functions.) This does not affect `extern` functions, or functions marked as `#[inline(never)]`.
This option has already existed in nightly as `-Zcross-crate-inline-threshold=always` for some time, and has gotten testing in that form. However, this option is still unstable, to give an opportunity for wider testing in this form.
Some performance numbers, based on a crate with many dependencies having just *one* large dependency set to `-Z hint-mostly-unused` (using Cargo's `profile-rustflags` option):
A release build went from 4m07s to 2m04s.
A non-release build went from 2m26s to 1m28s.
|
|
This allows a more gradual transition path for projects that need to use
use the spec-complaint C ABI both with older and newer rustc versions.
|
|
|
|
|
|
|
|
|
|
|
|
Rollup of 9 pull requests
Successful merges:
- rust-lang/rust#128425 (Make `missing_fragment_specifier` an unconditional error)
- rust-lang/rust#135927 (retpoline and retpoline-external-thunk flags (target modifiers) to enable retpoline-related target features)
- rust-lang/rust#140770 (add `extern "custom"` functions)
- rust-lang/rust#142176 (tests: Split dont-shuffle-bswaps along opt-levels and arches)
- rust-lang/rust#142248 (Add supported asm types for LoongArch32)
- rust-lang/rust#142267 (assert more in release in `rustc_ast_lowering`)
- rust-lang/rust#142274 (Update the stdarch submodule)
- rust-lang/rust#142276 (Update dependencies in `library/Cargo.lock`)
- rust-lang/rust#142308 (Upgrade `object`, `addr2line`, and `unwinding` in the standard library)
Failed merges:
- rust-lang/rust#140920 (Extract some shared code from codegen backend target feature handling)
r? `@ghost`
`@rustbot` modify labels: rollup
try-job: aarch64-apple
try-job: x86_64-msvc-1
try-job: x86_64-gnu
try-job: dist-i586-gnu-i586-i686-musl
try-job: test-various
|
|
It collects data about macro expansions and prints them in a table after
expansion finishes. It's very useful for detecting macro bloat,
especially for proc macros.
Details:
- It measures code snippets by pretty-printing them and then measuring
lines and bytes. This required a bunch of additional pretty-printing
plumbing, in `rustc_ast_pretty` and `rustc_expand`.
- The measurement is done in `MacroExpander::expand_invoc`.
- The measurements are stored in `ExtCtxt::macro_stats`.
|
|
enable retpoline-related target features
|
|
This hint allows the compiler to optimize its operation based on this
assumption, in order to compile faster. This is a hint, and does not
guarantee any particular behavior.
This option can substantially speed up compilation if applied to a large
dependency where the majority of the dependency does not get used. This flag
may slow down compilation in other cases.
Currently, this option makes the compiler defer as much code generation as
possible from functions in the crate, until later crates invoke those
functions. Functions that never get invoked will never have code generated for
them. For instance, if a crate provides thousands of functions, but only a few
of them will get called, this flag will result in the compiler only doing code
generation for the called functions. (This uses the same mechanisms as
cross-crate inlining of functions.) This does not affect `extern` functions, or
functions marked as `#[inline(never)]`.
Some performance numbers, based on a crate with many dependencies having
just *one* large dependency set to `-Z hint-mostly-unused` (using
Cargo's `profile-rustflags` option):
A release build went from 4m07s to 2m04s.
A non-release build went from 2m26s to 1m28s.
|
|
|
|
This reverts commit 3b22c21dd8c30f499051fe7a758ca0e5d81eb638, reversing
changes made to 5f292eea6d63abbd26f1e6e00a0b8cf21d828d7d.
|
|
coverage: Detect unused local file IDs to avoid an LLVM assertion
Each function's coverage metadata contains a *local file table* that maps local file IDs (used by the function's mapping regions) to global file IDs (shared by all functions in the same CGU).
LLVM requires all local file IDs to have at least one mapping region, and has an assertion that will fail if it detects a local file ID with no regions. To make sure that assertion doesn't fire, we need to detect and skip functions whose metadata would trigger it.
(This can't actually happen yet, because currently all of a function's spans must belong to the same file and expansion. But this will be an important edge case when adding expansion region support.)
|
|
This case can't actually happen yet (other than via a testing flag), because
currently all of a function's spans must belong to the same file and expansion.
But this will be an important edge case when adding expansion region support.
|
|
Previously `-Zprint-mono-items` would override the mono item collection
strategy. When debugging one doesn't want to change the behaviour, so
this was counter productive. Additionally, the produced behaviour was
artificial and might never arise without using the option in the first
place (`-Zprint-mono-items=eager` without `-Clink-dead-code`). Finally,
the option was incorrectly marked as `UNTRACKED`.
Resolve those issues, by turning `-Zprint-mono-items` into a boolean
flag that prints results of mono item collection without changing the
behaviour of mono item collection.
For codegen-units test incorporate `-Zprint-mono-items` flag directly
into compiletest tool.
Test changes are mechanical. `-Zprint-mono-items=lazy` was removed
without additional changes, and `-Zprint-mono-items=eager` was turned
into `-Clink-dead-code`. Linking dead code disables internalization, so
tests have been updated accordingly.
|
|
Make `-Zfixed-x18` into a target modifier
As part of #136966, the `-Zfixed-x18` flag should be turned into a target modifier. This is a blocker to stabilization of the flag. The flag was originally added in #124655 and the MCP for its addition is [MCP#748](https://github.com/rust-lang/compiler-team/issues/748).
On some aarch64 targets, the x18 register is used as a temporary caller-saved register by default. When the `-Zfixed-x18` flag is passed, this is turned off so that the compiler doesn't use the x18 register. This allows end-users to use the x18 register for other purposes. For example, by accessing it with inline asm you can use the register as a very efficient thread-local variable. Another common use-case is to store the stack pointer needed by the shadow-call-stack sanitizer. There are also some aarch64 targets where not using x18 is the default – in those cases the flag is a no-op.
Note that this flag does not *on its own* cause an ABI mismatch. What actually causes an ABI mismatch is when you have different compilation units that *disagree* on what it should be used for. But having a CU that uses it and another CU that doesn't normally isn't enough to trigger an ABI problem. However, we still consider the flag to be a target modifier in all cases, since it is assumed that you are passing the flag because you intend to assign some other meaning to the register. Rejecting all flag mismatches even if not all are unsound is consistent with [RFC#3716](https://rust-lang.github.io/rfcs/3716-target-modifiers.html). See the headings "not all mismatches are unsound" and "cases that are not caught" for additional discussion of this.
On aarch64 targets where `-Zfixed-x18` is not a no-op, it is an error to pass `-Zsanitizer=shadow-call-stack` without also passing `-Zfixed-x18`.
|
|
|
|
|
|
Autodiff flags
Interestingly, it seems that some other projects have conflicts with exactly the same LLVM optimization passes as autodiff.
At least `LLVMRustOptimize` has exactly the flags that we need to disable problematic opt passes.
This PR enables us to compile code where users differentiate two identical functions in the same module. This has been especially common in test cases, but it's not impossible to encounter in the wild.
It also enables two new flags for testing/debugging. I consider writing an MCP to upgrade PrintPasses to be a standalone -Z flag, since it is *not* the same as `-Z print-llvm-passes`, which IMHO gives less useful output. A discussion can be found here: [#t-compiler/llvm > Print llvm passes. @ 💬](https://rust-lang.zulipchat.com/#narrow/channel/187780-t-compiler.2Fllvm/topic/Print.20llvm.20passes.2E/near/511533038)
Finally, it improves `PrintModBefore` and `PrintModAfter`. They used to work reliable, but now we just schedule enzyme as part of an existing ModulePassManager (MPM). Since Enzyme is last in the MPM scheduling, PrintModBefore became very inaccurate. It used to print the input module, which we gave to the Enzyme and was great to create llvm-ir reproducer. However, lately the MPM would run the whole `default<O3>` pipeline, which heavily modifies the llvm module, before we pass it to Enzyme. That made it impossible to use the flag to create llvm-ir reproducers for Enzyme bugs. We now schedule a PrintModule pass just before Enzyme, solving this problem.
Based on the PrintPass output, it also _seems_ like changing `registerEnzymeAndPassPipeline(PB, true);` to `registerEnzymeAndPassPipeline(PB, false);` has no effect. In theory, the bool should tell Enzyme to schedule some helpful passes in the PassBuilder. However, since it doesn't do anything and I'm not 100% sure anymore on whether we really need it, I'll just disable it for now and postpone investigations.
r? ``@oli-obk``
closes #139471
Tracking:
- https://github.com/rust-lang/rust/issues/124509
|
|
Do not remove trivial `SwitchInt` in analysis MIR
This PR ensures that we don't prematurely remove trivial `SwitchInt` terminators which affects both the borrow-checking and runtime semantics (i.e. UB) of the code. Previously the `SimplifyCfg` optimization was removing `SwitchInt` terminators when they was "trivial", i.e. when all arms branched to the same basic block, even if that `SwitchInt` terminator had the side-effect of reading an operand which (for example) may not be initialized or may point to an invalid place in memory.
This behavior is unlike all other optimizations, which are only applied after "analysis" (i.e. borrow-checking) is finished, and which Miri disables to make sure the compiler doesn't silently remove UB.
Fixing this code "breaks" (i.e. unmasks) code that used to borrow-check but no longer does, like:
```rust
fn foo() {
let x;
let (0 | _) = x;
}
```
This match expression should perform a read because `_` does not shadow the `0` literal pattern, and the compiler should have to read the match scrutinee to compare it to 0. I've checked that this behavior does not actually manifest in practice via a crater run which came back clean: https://github.com/rust-lang/rust/pull/139042#issuecomment-2767436367
As a side-note, it may be tempting to suggest that this is actually a good thing or that we should preserve this behavior. If we wanted to make this work (i.e. trivially optimize out reads from matches that are redundant like `0 | _`), then we should be enabling this behavior *after* fixing this. However, I think it's kinda unprincipled, and for example other variations of the code don't even work today, e.g.:
```rust
fn foo() {
let x;
let (0.. | _) = x;
}
```
|
|
Add unstable parsing of `--extern foo::bar=libbar.rlib` command line options
This is a tiny step towards implementing the rustc side of support for implementing packages as optional namespaces (#122349). We add support for parsing command line options like `--extern foo::bar=libbar.rlib` when the `-Z namespaced-crates` option is present.
We don't do anything further with them. The next step is to plumb this down to the name resolver.
This PR also generally refactors the extern argument parsing code and adds some unit tests to make it clear what forms should be accepted with and without the flag.
cc ```@epage``` ```@ehuss```
|
|
Also refactors some of the crate name parsing code and adds unit tests
Issue #122349
Co-authored-by: León Orell Valerian Liehr <me@fmease.dev>
|
|
|
|
|
|
|
|
KCFI: Add KCFI arity indicator support
Adds KCFI arity indicator support to the Rust compiler (see https://github.com/rust-lang/rust/issues/138311, https://github.com/llvm/llvm-project/pull/121070, and https://lore.kernel.org/lkml/CANiq72=3ghFxy8E=AU9p+0imFxKr5iU3sd0hVUXed5BA+KjdNQ@mail.gmail.com/).
|
|
Adds KCFI arity indicator support to the Rust compiler (see rust-lang/rust#138311,
https://github.com/llvm/llvm-project/pull/121070, and
https://lore.kernel.org/lkml/CANiq72=3ghFxy8E=AU9p+0imFxKr5iU3sd0hVUXed5BA+KjdNQ@mail.gmail.com/).
|
|
Autodiff batching
Enzyme supports batching, which is especially known from the ML side when training neural networks.
There we would normally have a training loop, where in each iteration we would pass in some data (e.g. an image), and a target vector. Based on how close we are with our prediction we compute our loss, and then use backpropagation to compute the gradients and update our weights.
That's quite inefficient, so what you normally do is passing in a batch of 8/16/.. images and targets, and compute the gradients for those all at once, allowing better optimizations.
Enzyme supports batching in two ways, the first one (which I implemented here) just accepts a Batch size,
and then each Dual/Duplicated argument has not one, but N shadow arguments. So instead of
```rs
for i in 0..100 {
df(x[i], y[i], 1234);
}
```
You can now do
```rs
for i in 0..100.step_by(4) {
df(x[i+0],x[i+1],x[i+2],x[i+3], y[i+0], y[i+1], y[i+2], y[i+3], 1234);
}
```
which will give the same results, but allows better compiler optimizations. See the testcase for details.
There is a second variant, where we can mark certain arguments and instead of having to pass in N shadow arguments, Enzyme assumes that the argument is N times longer. I.e. instead of accepting 4 slices with 12 floats each, we would accept one slice with 48 floats. I'll implement this over the next days.
I will also add more tests for both modes.
For any one preferring some more interactive explanation, here's a video of Tim's llvm dev talk, where he presents his work. https://www.youtube.com/watch?v=edvaLAL5RqU
I'll also add some other docs to the dev guide and user docs in another PR.
r? ghost
Tracking:
- https://github.com/rust-lang/rust/issues/124509
- https://github.com/rust-lang/rust/issues/135283
|
|
add `TypingMode::Borrowck`
Shares the first commit with #138499, doesn't really matter which PR to land first :blush: :grin:
Introduces `TypingMode::Borrowck` which unlike `TypingMode::Analysis`, uses the hidden type computed by HIR typeck as the initial value of opaques instead of an unconstrained infer var. This is a part of https://github.com/rust-lang/types-team/issues/129.
Using this new `TypingMode` is unfortunately a breaking change for now, see tests/ui/impl-trait/non-defining-uses/as-projection-term.rs. Using an inference variable as the initial value results in non-defining uses in the defining scope. We therefore only enable it if with `-Znext-solver=globally` or `-Ztyping-mode-borrowck`
To do that the PR contains the following changes:
- `TypeckResults::concrete_opaque_type` are already mapped to the definition of the opaque type
- writeback now checks that the non-lifetime parameters of the opaque are universal
- for this, `fn check_opaque_type_parameter_valid` is moved from `rustc_borrowck` to `rustc_trait_selection`
- we add a new `query type_of_opaque_hir_typeck` which, using the same visitors as MIR typeck, attempts to merge the hidden types from HIR typeck from all defining scopes
- done by adding a `DefiningScopeKind` flag to toggle between using borrowck and HIR typeck
- the visitors stop checking that the MIR type matches the HIR type. This is trivial as the HIR type are now used as the initial hidden types of the opaque. This check is useful as a safeguard when not using `TypingMode::Borrowck`, but adding it to the new structure is annoying and it's not soundness critical, so I intend to not add it back.
- add a `TypingMode::Borrowck` which behaves just like `TypingMode::Analysis` except when normalizing opaque types
- it uses `type_of_opaque_hir_typeck(opaque)` as the initial value after replacing its regions with new inference vars
- it uses structural lookup in the new solver
fixes #112201, fixes #132335, fixes #137751
r? `@compiler-errors` `@oli-obk`
|
|
|
|
|
