During incremental ThinLTO compilation, we attempt to re-use the
optimized (post-ThinLTO) bitcode file for a module if it is 'safe' to do
so.
Up until now, 'safe' has meant that the set of modules that our current
modules imports from/exports to is unchanged from the previous
compilation session. See PR #67020 and PR #71131 for more details.
However, this turns out be insufficient to guarantee that it's safe
to reuse the post-LTO module (i.e. that optimizing the pre-LTO module
would produce the same result). When LLVM optimizes a module during
ThinLTO, it may look at other information from the 'module index', such
as whether a (non-imported!) global variable is used. If this
information changes between compilation runs, we may end up re-using an
optimized module that (for example) had dead-code elimination run on a
function that is now used by another module.
Fortunately, LLVM implements its own ThinLTO module cache, which is used
when ThinLTO is performed by a linker plugin (e.g. when clang is used to
compile a C proect). Using this cache directly would require extensive
refactoring of our code - but fortunately for us, LLVM provides a
function that does exactly what we need.
The function `llvm::computeLTOCacheKey` is used to compute a SHA-1 hash
from all data that might influence the result of ThinLTO on a module.
In addition to the module imports/exports that we manually track, it
also hashes information about global variables (e.g. their liveness)
which might be used during optimization. By using this function, we
shouldn't have to worry about new LLVM passes breaking our module re-use
behavior.
In LLVM, the output of this function forms part of the filename used to
store the post-ThinLTO module. To keep our current filename structure
intact, this PR just writes out the mapping 'CGU name -> Hash' to a
file. To determine if a post-LTO module should be reused, we compare
hashes from the previous session.
This should unblock PR #75199 - by sheer chance, it seems to have hit
this issue due to the particular CGU partitioning and optimization
decisions that end up getting made.
A significant amount of intrinsics do not actually need backend-specific
behaviors to be implemented, instead relying on methods already in
rustc_codegen_ssa. So, extract those methods out to rustc_codegen_ssa,
so that each backend doesn't need to reimplement the same code.
Move `rustllvm` into `compiler/rustc_llvm`
The `rustllvm` directory is not self-contained, it contains C++ code built by a build script of the `rustc_llvm` crate which is then linked into that crate.
So it makes sense to make `rustllvm` a part of `rustc_llvm` and move it into its directory.
I replaced `rustllvm` with more obvious `llvm-wrapper` as the subdirectory name, but something like `llvm-adapter` would work as well, other suggestions are welcome.
To make things more confusing, the Rust side of FFI functions defined in `rustllvm` can be found in `rustc_codegen_llvm` rather than in `rustc_llvm`. Perhaps they need to be moved as well, but this PR doesn't do that.
The presence of multiple LLVM-related directories in `src` (`llvm-project`, `rustllvm`, `librustc_llvm`, `librustc_codegen_llvm` and their predecessors) historically confused me and made me wonder about their purpose.
With this PR we will have LLVM itself (`llvm-project`), a FFI crate (`rustc_llvm`, kind of `llvm-sys`) and a codegen backend crate using LLVM through the FFI crate (`rustc_codegen_llvm`).
If a symbol name can only be imported from one place for a type, and
as long as it was not glob-imported anywhere in the current crate, we
can trim its printed path and print only the name.
This has wide implications on error messages with types, for example,
shortening `std::vec::Vec` to just `Vec`, as long as there is no other
`Vec` importable anywhere.
This adds a new '-Z trim-diagnostic-paths=false' option to control this
feature.
On the good path, with no diagnosis printed, we should try to avoid
issuing this query, so we need to prevent trimmed_def_paths query on
several cases.
This change also relies on a previous commit that differentiates
between `Debug` and `Display` on various rustc types, where the latter
is trimmed and presented to the user and the former is not.
Make to_immediate/from_immediate configurable by backends
`librustc_codegen_ssa` has the concept of an immediate vs. memory type, and `librustc_codegen_llvm` uses this distinction to implement `bool`s being `i8` in memory, and `i1` in immediate contexts. However, some of that implementation leaked into `codegen_ssa` when converting to/from immediate values. So, move those methods into builder traits, so that behavior can be configured by backends.
This is useful if a backend is able to keep bools as bools, or, needs to do more trickery than just bools to bytes.
(Note that there's already a large amount of things abstracted with "immediate types" - this is just bringing this particular thing in line to be abstracted as well)
---
Pinging @eddyb since that's who I was talking about this change with when they suggested I submit a PR.
This commit changes `uncached_llvm_type` so that a named struct type
(with an empty name) is always created when the `fewer_names` option
is enabled. By skipping the generation of names, we can improve perf.
Giving `LLVMStructCreateNamed` an empty name works because LLVM will
perform random renames to avoid collisions.
Signed-off-by: David Wood <david@davidtw.co>
This commit removes the obsolete printer and replaces all uses of it
with `FmtPrinter`. Of the replaced uses, all but one use was in `debug!`
logging, two cases were notable:
- `MonoItem::to_string` is used in `-Z print-mono-items` and therefore
affects the output of all codegen-units tests.
- `DefPathBasedNames` was used in `librustc_codegen_llvm/type_of.rs`
with `LLVMStructCreateNamed` and that'll now get different values, but
this should result in no functional change.
Signed-off-by: David Wood <david@davidtw.co>
