indirect immutable freeze by-value function parameters.
Right now, `rustc` only examines function signatures and the platform ABI when
determining the LLVM attributes to apply to parameters. This results in missed
optimizations, because there are some attributes that can be determined via
analysis of the MIR making up the function body. In particular, `readonly`
could be applied to most indirectly-passed by-value function arguments
(specifically, those that are freeze and are observed not to be mutated), but
it currently is not.
This patch introduces the machinery that allows `rustc` to determine those
attributes. It consists of a query, `deduced_param_attrs`, that, when
evaluated, analyzes the MIR of the function to determine supplementary
attributes. The results of this query for each function are written into the
crate metadata so that the deduced parameter attributes can be applied to
cross-crate functions. In this patch, we simply check the parameter for
mutations to determine whether the `readonly` attribute should be applied to
parameters that are indirect immutable freeze by-value. More attributes could
conceivably be deduced in the future: `nocapture` and `noalias` come to mind.
Adding `readonly` to indirect function parameters where applicable enables some
potential optimizations in LLVM that are discussed in [issue 103103] and [PR
103070] around avoiding stack-to-stack memory copies that appear in functions
like `core::fmt::Write::write_fmt` and `core::panicking::assert_failed`. These
functions pass a large structure unchanged by value to a subfunction that also
doesn't mutate it. Since the structure in this case is passed as an indirect
parameter, it's a pointer from LLVM's perspective. As a result, the
intermediate copy of the structure that our codegen emits could be optimized
away by LLVM's MemCpyOptimizer if it knew that the pointer is `readonly
nocapture noalias` in both the caller and callee. We already pass `nocapture
noalias`, but we're missing `readonly`, as we can't determine whether a
by-value parameter is mutated by examining the signature in Rust. I didn't have
much success with having LLVM infer the `readonly` attribute, even with fat
LTO; it seems that deducing it at the MIR level is necessary.
No large benefits should be expected from this optimization *now*; LLVM needs
some changes (discussed in [PR 103070]) to more aggressively use the `noalias
nocapture readonly` combination in its alias analysis. I have some LLVM patches
for these optimizations and have had them looked over. With all the patches
applied locally, I enabled LLVM to remove all the `memcpy`s from the following
code:
```rust
fn main() {
println!("Hello {}", 3);
}
```
which is a significant codegen improvement over the status quo. I expect that
if this optimization kicks in in multiple places even for such a simple
program, then it will apply to Rust code all over the place.
[issue 103103]: https://github.com/rust-lang/rust/issues/103103
[PR 103070]: https://github.com/rust-lang/rust/pull/103070
Queries can provide an arbitrary expression for their description and
their caching behavior. Before, these expressions where stored in a
`rustc_query_description` macro emitted by the `rustc_queries` macro,
and then used in `rustc_query_impl` to fill out the methods for the
`QueryDescription` trait.
Instead, we now emit two new modules from `rustc_queries` containing the
functions with the expressions. `rustc_query_impl` calls these functions
now instead of invoking the macro.
Since we are now defining some of the functions in
`rustc_middle::query`, we now need all the imports for the key types
there as well.
The compiler currently has `-Ztime` and `-Ztime-passes`. I've used
`-Ztime-passes` for years but only recently learned about `-Ztime`.
What's the difference? Let's look at the `-Zhelp` output:
```
-Z time=val -- measure time of rustc processes (default: no)
-Z time-passes=val -- measure time of each rustc pass (default: no)
```
The `-Ztime-passes` description is clear, but the `-Ztime` one is less so.
Sounds like it measures the time for the entire process?
No. The real difference is that `-Ztime-passes` prints out info about passes,
and `-Ztime` does the same, but only for a subset of those passes. More
specifically, there is a distinction in the profiling code between a "verbose
generic activity" and an "extra verbose generic activity". `-Ztime-passes`
prints both kinds, while `-Ztime` only prints the first one. (It took me
a close reading of the source code to determine this difference.)
In practice this distinction has low value. Perhaps in the past the "extra
verbose" output was more voluminous, but now that we only print stats for a
pass if it exceeds 5ms or alters the RSS, `-Ztime-passes` is less spammy. Also,
a lot of the "extra verbose" cases are for individual lint passes, and you need
to also use `-Zno-interleave-lints` to see those anyway.
Therefore, this commit removes `-Ztime` and the associated machinery. One thing
to note is that the existing "extra verbose" activities all have an extra
string argument, so the commit adds the ability to accept an extra argument to
the "verbose" activities.
Use function pointers instead of macro-unrolled loops in rustc_query_impl
By making these standalone functions, we
a) allow making them extensible in the future with a new `QueryStruct`
b) greatly decrease the amount of code in each individual function, avoiding exponential blowup in llvm
Helps with https://github.com/rust-lang/rust/issues/96524. Based on https://github.com/rust-lang/rust/pull/101173; only the last commit is relevant.
r? `@cjgillot`
By making these standalone functions, we
a) allow making them extensible in the future with a new `QueryStruct`
b) greatly decrease the amount of code in each individual function, avoiding exponential blowup in llvm
fix a ui test
use `into`
fix clippy ui test
fix a run-make-fulldeps test
implement `IntoQueryParam<DefId>` for `OwnerId`
use `OwnerId` for more queries
change the type of `ParentOwnerIterator::Item` to `(OwnerId, OwnerNode)`
add note for `layout_of` when query depth overflows
Fixes#101747
Added `try_find_layout_root` function to add a note for `layout_of` when query depth overflows. This would make the error in #101747 look like this:
```
error: queries overflow the depth limit!
|
note: Query depth increased by 66 when computing layout of `core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<core::option::Option<alloc::boxed::Box<alloc::string::String>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>`!
--> D:\rust-backup\parallel_rust\query_depth.rs:40:1
|
40 | fn main() {
| ^^^^^^^^^
error: aborting due to previous error
```
cc ``@semicoleon``
Further simplify the macros generated by `rustc_queries`
This doesn't actually move anything outside the macros, but it makes them simpler to read.
- Add a new `rustc_query_names` macro. This allows a much simpler syntax for the matchers in the macros passed to it as a callback.
- Convert `define_dep_nodes` and `alloc_once` to use `rustc_query_names`. This is possible because they only use the names
(despite the quite complicated matchers in `define_dep_nodes`, none of the other arguments are used).
- Get rid of `rustc_dep_node_append`.
r? `@cjgillot`
Simplify caching and storage for queries
I highly recommend reviewing commit-by-commit; each individual commit is quite small but it can be hard to see looking at the overall diff that the behavior is the same. Each commit depends on the previous.
r? `@cjgillot`
We want to refer to `crate::plumbing::try_load_from_disk` in the const, but hard-coding it in
rustc_queries, where we don't yet know the crate this macro will be called in, seems kind of hacky.
Do it in query_impl instead.
- Add a new `rustc_query_names` macro. This allows a much simpler syntax for the matchers in the macros passed to it as a callback.
- Convert `define_dep_nodes` and `alloc_once` to use `rustc_query_names`. This is possible because they only use the names
(despite the quite complicated matchers in `define_dep_nodes`, none of the other arguments are used).
- Get rid of `rustc_dep_node_append`.
- Add a `HandleCycleError` enum to rustc_query_system, along with a `handle_cycle_error` function
- Move `Value` to rustc_query_system, so `handle_cycle_error` can use it
- Move the `Value` impls from rustc_query_impl to rustc_middle. This is necessary due to orphan rules.
- Parameterize DepKindStruct over `'tcx`
This allows passing in an invariant function pointer in `query_callback`,
rather than having to try and make it work for any lifetime.
- Add a new `execute_query` function to `QueryDescription` so we can call `tcx.$name` without needing to be in a macro context
`rustc_data_structures::thin_vec::ThinVec` looks like this:
```
pub struct ThinVec<T>(Option<Box<Vec<T>>>);
```
It's just a zero word if the vector is empty, but requires two
allocations if it is non-empty. So it's only usable in cases where the
vector is empty most of the time.
This commit removes it in favour of `thin_vec::ThinVec`, which is also
word-sized, but stores the length and capacity in the same allocation as
the elements. It's good in a wider variety of situation, e.g. in enum
variants where the vector is usually/always non-empty.
The commit also:
- Sorts some `Cargo.toml` dependency lists, to make additions easier.
- Sorts some `use` item lists, to make additions easier.
- Changes `clean_trait_ref_with_bindings` to take a
`ThinVec<TypeBinding>` rather than a `&[TypeBinding]`, because this
avoid some unnecessary allocations.
