Simplify the `tcx.alloc_map` API
This PR changes all functions that require manually locking the `alloc_map` to functions on `TyCtxt` that lock the map internally. In the same step we make the `TyCtxt::alloc_map` field private.
r? @RalfJung
rustllvm: Use .init_array rather than .ctors
LLVM TargetMachines default to using the (now-legacy) .ctors
representation of init functions. Mixing .ctors and .init_array
representations can cause issues when linking with lld.
This happens in practice for:
* Our profiling runtime which is currently implicitly built with
.init_array since it is built by clang, which sets this field.
* External C/C++ code that may be linked into the same process.
Fixes: #71233
The App Store performs certain sanity checks on bitcode, including that
an acceptable set of command line arguments was used when compiling a
given module. For Rust code to be distributed on the app store with
bitcode rustc must pretend to have the same command line arguments.
Add Option to Force Unwind Tables
When panic != unwind, `nounwind` is added to all functions for a target.
This can cause issues when a panic happens with RUST_BACKTRACE=1, as
there needs to be a way to reconstruct the backtrace. There are three
possible sources of this information: forcing frame pointers (for which
an option exists already), debug info (for which an option exists), or
unwind tables.
Especially for embedded devices, forcing frame pointers can have code
size overheads (RISC-V sees ~10% overheads, ARM sees ~2-3% overheads).
In production code, it can be the case that debug info is not kept, so it is useful
to provide this third option, unwind tables, that users can use to
reconstruct the call stack. Reconstructing this stack is harder than
with frame pointers, but it is still possible.
---
This came up in discussion on #69890, and turned out to be a fairly simple addition.
r? @hanna-kruppe
When panic != unwind, `nounwind` is added to all functions for a target.
This can cause issues when a panic happens with RUST_BACKTRACE=1, as
there needs to be a way to reconstruct the backtrace. There are three
possible sources of this information: forcing frame pointers (for which
an option exists already), debug info (for which an option exists), or
unwind tables.
Especially for embedded devices, forcing frame pointers can have code
size overheads (RISC-V sees ~10% overheads, ARM sees ~2-3% overheads).
In code, it can be the case that debug info is not kept, so it is useful
to provide this third option, unwind tables, that users can use to
reconstruct the call stack. Reconstructing this stack is harder than
with frame pointers, but it is still possible.
This commit adds a compiler option which allows a user to force the
addition of unwind tables. Unwind tables cannot be disabled on targets
that require them for correctness, or when using `-C panic=unwind`.
Have the per-query caches store the results on arenas
This PR leverages the cache for each query to serve as storage area for the query results.
It introduces a new cache `ArenaCache`, which moves the result to an arena,
and only stores the reference in the hash map.
This allows to remove a sizeable part of the usage of the global `TyCtxt` arena.
I only migrated queries that already used arenas before.
Remove -Z no-landing-pads flag
Since #67502, `-Z no-landing-pads` will cause all attempted unwinds to abort since we don't generate a `try` / `catch`. This previously worked because `__rust_try` was located in libpanic_unwind which is always compiled with `-C panic=unwind`, but `__rust_try` is now directly inline into the crate that uses `catch_unwind`.
As such, `-Z no-landing-pads` is now mostly useless and people should use `-C panic=abort` instead.
Store LLVM bitcode in object files, not compressed
This commit is an attempted resurrection of #70458 where LLVM bitcode
emitted by rustc into rlibs is stored into object file sections rather
than in a separate file. The main rationale for doing this is that when
rustc emits bitcode it will no longer use a custom compression scheme
which makes it both easier to interoperate with existing tools and also
cuts down on compile time since this compression isn't happening.
The blocker for this in #70458 turned out to be that native linkers
didn't handle the new sections well, causing the sections to either
trigger bugs in the linker or actually end up in the final linked
artifact. This commit attempts to address these issues by ensuring that
native linkers ignore the new sections by inserting custom flags with
module-level inline assembly.
Note that this does not currently change the API of the compiler at all.
The pre-existing `-C bitcode-in-rlib` flag is co-opted to indicate
whether the bitcode should be present in the object file or not.
Finally, note that an important consequence of this commit, which is also
one of its primary purposes, is to enable rustc's `-Clto` bitcode
loading to load rlibs produced with `-Clinker-plugin-lto`. The goal here
is that when you're building with LTO Cargo will tell rustc to skip
codegen of all intermediate crates and only generate LLVM IR. Today
rustc will generate both object code and LLVM IR, but the object code is
later simply thrown away, wastefully.
LLVM TargetMachines default to using the (now-legacy) .ctors
representation of init functions. Mixing .ctors and .init_array
representations can cause issues when linking with lld.
This happens in practice for:
* Our profiling runtime which is currently implicitly built with
.init_array since it is built by clang, which sets this field.
* External C/C++ code that may be linked into the same process.
To support legacy systems which may use .ctors, targets may now specify
that they use .ctors via the use_ctors attribute which defaults to
false.
For debugging and manual control, -Z use-ctors-section=yes/no will allow
manual override.
Fixes: #71233
This commit is an attempted resurrection of #70458 where LLVM bitcode
emitted by rustc into rlibs is stored into object file sections rather
than in a separate file. The main rationale for doing this is that when
rustc emits bitcode it will no longer use a custom compression scheme
which makes it both easier to interoperate with existing tools and also
cuts down on compile time since this compression isn't happening.
The blocker for this in #70458 turned out to be that native linkers
didn't handle the new sections well, causing the sections to either
trigger bugs in the linker or actually end up in the final linked
artifact. This commit attempts to address these issues by ensuring that
native linkers ignore the new sections by inserting custom flags with
module-level inline assembly.
Note that this does not currently change the API of the compiler at all.
The pre-existing `-C bitcode-in-rlib` flag is co-opted to indicate
whether the bitcode should be present in the object file or not.
Finally, note that an important consequence of this commit, which is also
one of its primary purposes, is to enable rustc's `-Clto` bitcode
loading to load rlibs produced with `-Clinker-plugin-lto`. The goal here
is that when you're building with LTO Cargo will tell rustc to skip
codegen of all intermediate crates and only generate LLVM IR. Today
rustc will generate both object code and LLVM IR, but the object code is
later simply thrown away, wastefully.
Detect mistyped associated consts in `Instance::resolve`.
*Based on #71049 to prevent redundant/misleading downstream errors.*
Fixes#70942 by refusing to resolve an associated `const` if it doesn't have the same type in the `impl` that it does in the `trait` (which we assume had errored, and `delay_span_bug` guards against bugs).
attempt to recover perf by removing `exports_all_green`
attempt to recover perf by removing `exports_all_green` flag.
cc #71248
(My hypothesis is that my use of this flag was an overly conservative generalization of PR #67020.)
This lets us specify the default at the options declaration point,
instead of using `.unwrap(default)` or `None | Some(default)` at some
use point far away. It also makes the code more concise.
Renamed the struct to make it a little clearer that it doesn't just hold one
imports map. (I couldn't bring myself to write it as `ThinLTOImportsExports`
though, mainly since the exports map is literally derived from the imports map
data.) Added some doc to the struct too.
Revised comments to add link to the newer issue that discusses why the exports
are relevant.
Renamed a few of the methods so that the two character difference is more
apparent (because 1. the method name is shorter and, perhaps more importantly,
the changed characters now lie at the beginning of the method name.)
incremental compilation.
This is symmetric to PR #67020, which handled the case where the LLVM module's
*imports* changed. This commit builds upon the infrastructure added there; the
export map is just the inverse of the import map, so we can build the export map
at the same time that we load the serialized import map.
Fix#69798
Don't import integer and float modules, use assoc consts
Stop importing the standard library integer and float modules to reach the `MIN`, `MAX` and other constants. They are available directly on the primitive types now.
This PR is a follow up of #69860 which made sure we use the new constants in documentation.
This type of change touches a lot of files, and previously all my assoc int consts PRs had collisions and were accepted only after a long delay. So I'd prefer to do it in smaller steps now. Just removing these imports seem like a good next step.
r? @dtolnay
