trans: Internalize symbols without relying on LLVM
This PR makes the compiler use the information gather by the trans collector in order to determine which symbols/trans-items can be made internal. This has the advantages:
+ of being LLVM independent,
+ of also working in incremental mode, and
+ of allowing to not keep all LLVM modules in memory at the same time.
This is in preparation for fixing issue #39280.
cc @rust-lang/compiler
Add support for full RELRO
This commit adds support for full RELRO, and enables it for the
platforms I know have support for it.
Full RELRO makes the PLT+GOT data read-only on startup, preventing it
from being overwritten.
http://tk-blog.blogspot.com/2009/02/relro-not-so-well-known-memory.htmlFixesrust-lang/rust#29877.
---
I'm not entirely certain if this is the best way to do it, but I figured mimicking the way it's done for PIE seemed like a good start at least. I'm not sure whether we want to have it enabled by default globally and then disabling it explicitly for targets that don't support it though. I'm also not sure whether the `full_relro` function should call `bug!()` or something like it for linkers that don't support it rather than no-opping.
Refactor: {Lvalue,Rvalue,Operand}::ty only need the locals' types, not the full &Mir
I am writing code that needs to call these `ty` methods while mutating MIR -- which is impossible with the current API.
Even with the refactoring the situation is not great: I am cloning the `local_decls` and then passing the clone to the `ty` methods. I have to clone because `Mir::basic_blocks_mut` borrows the entire `Mir` including the `local_decls`. But even that is better than not being able to get these types at all...
Cc @nikomatsakis
incr.comp.: Don't include span information in the ICH of type definitions
This should improve some of the `regex` tests on perf.rlo. Not including spans into the ICH is harmless until we also cache warnings. To really solve the problem, we need to do more refactoring (see #43088).
r? @nikomatsakis
Use similar compression settings as before updating to use flate2
Fixes#42879
(My first PR to rust-lang yay)
This changes the compression settings back to how they were before the change to use the flate2 crate rather than the in-tree flate library. The specific changes are to use the `Fast` compression level (which should be equivialent to what was used before), and use a raw deflate stream rather than wrapping the stream in a zlib wrapper. The [zlib](https://tools.ietf.org/html/rfc1950) wrapper adds an extra 2 bytes of header data, and 4 bytes for a checksum at the end. The change to use a faster compression level did give some compile speedups in the past (see #37298). Having to calculate a checksum also added a small overhead, which didn't exist before the change to flate2.
r? @alexcrichton
This commit implements stack probes on x86/x86_64 using the freshly landed
support upstream in LLVM. The purpose of stack probes here are to guarantee a
segfault on stack overflow rather than having a chance of running over the guard
page already present on all threads by accident.
At this time there's no support for any other architecture because LLVM itself
does not have support for other architectures.
Switch to rust-lang-nursery/compiler-builtins
This commit migrates the in-tree `libcompiler_builtins` to the upstream version
at https://github.com/rust-lang-nursery/compiler-builtins. The upstream version
has a number of intrinsics written in Rust and serves as an in-progress rewrite
of compiler-rt into Rust. Additionally it also contains all the existing
intrinsics defined in `libcompiler_builtins` for 128-bit integers.
It's been the intention since the beginning to make this transition but
previously it just lacked the manpower to get done. As this PR likely shows it
wasn't a trivial integration! Some highlight changes are:
* The PR rust-lang-nursery/compiler-builtins#166 contains a number of fixes
across platforms and also some refactorings to make the intrinsics easier to
read. The additional testing added there also fixed a number of integration
issues when pulling the repository into this tree.
* LTO with the compiler-builtins crate was fixed to link in the entire crate
after the LTO process as these intrinsics are excluded from LTO.
* Treatment of hidden symbols was updated as previously the
`#![compiler_builtins]` crate would mark all symbol *imports* as hidden
whereas it was only intended to mark *exports* as hidden.
This PR is an implementation of [RFC 1974] which specifies a new method of
defining a global allocator for a program. This obsoletes the old
`#![allocator]` attribute and also removes support for it.
[RFC 1974]: https://github.com/rust-lang/rfcs/pull/197
The new `#[global_allocator]` attribute solves many issues encountered with the
`#![allocator]` attribute such as composition and restrictions on the crate
graph itself. The compiler now has much more control over the ABI of the
allocator and how it's implemented, allowing much more freedom in terms of how
this feature is implemented.
cc #27389
This commit migrates the in-tree `libcompiler_builtins` to the upstream version
at https://github.com/rust-lang-nursery/compiler-builtins. The upstream version
has a number of intrinsics written in Rust and serves as an in-progress rewrite
of compiler-rt into Rust. Additionally it also contains all the existing
intrinsics defined in `libcompiler_builtins` for 128-bit integers.
It's been the intention since the beginning to make this transition but
previously it just lacked the manpower to get done. As this PR likely shows it
wasn't a trivial integration! Some highlight changes are:
* The PR rust-lang-nursery/compiler-builtins#166 contains a number of fixes
across platforms and also some refactorings to make the intrinsics easier to
read. The additional testing added there also fixed a number of integration
issues when pulling the repository into this tree.
* LTO with the compiler-builtins crate was fixed to link in the entire crate
after the LTO process as these intrinsics are excluded from LTO.
* Treatment of hidden symbols was updated as previously the
`#![compiler_builtins]` crate would mark all symbol *imports* as hidden
whereas it was only intended to mark *exports* as hidden.
Make wasm32 buildbot test LLVM backend
This adds the experimental targets option to configure so it can be used
by the builders and changes the wasm32 Dockerfile accordingly. Instead
of using LLVM from the emsdk, the builder's emscripten tools now uses
the Rust in-tree LLVM, since this is the one built with wasm support.
Print -Zincremental-info to stderr instead of stdout.
Fixes#42583.
The [cargo-incremental](https://github.com/nikomatsakis/cargo-incremental) tool probably does not need to be updated. It already merges stdout and stderr before parsing the compiler's output.
r? @alexcrichton
Replaced by adding extra imports, adding hidden code (`# ...`), modifying
examples to be runnable (sorry Homura), specifying non-Rust code, and
converting to should_panic, no_run, or compile_fail.
Remaining "```ignore"s received an explanation why they are being ignored.
This is used in wasm32-experimental-emscripten to ensure that emscripten
links against the libc bitcode files produced by the wasm LLVM backend,
instead of using fastcomp.
Integrate jobserver support to parallel codegen
This commit integrates the `jobserver` crate into the compiler. The crate was
previously integrated in to Cargo as part of rust-lang/cargo#4110. The purpose
here is to two-fold:
* Primarily the compiler can cooperate with Cargo on parallelism. When you run
`cargo build -j4` then this'll make sure that the entire build process between
Cargo/rustc won't use more than 4 cores, whereas today you'd get 4 rustc
instances which may all try to spawn lots of threads.
* Secondarily rustc/Cargo can now integrate with a foreign GNU `make` jobserver.
This means that if you call cargo/rustc from `make` or another
jobserver-compatible implementation it'll use foreign parallelism settings
instead of creating new ones locally.
As the number of parallel codegen instances in the compiler continues to grow
over time with the advent of incremental compilation it's expected that this'll
become more of a problem, so this is intended to nip concurrent concerns in the
bud by having all the tools to cooperate!
Note that while rustc has support for itself creating a jobserver it's far more
likely that rustc will always use the jobserver configured by Cargo. Cargo today
will now set a jobserver unconditionally for rustc to use.
mark calls in the unwind path as !noinline
The unwind path is always cold, so that should not have bad performance
implications. This avoids catastrophic exponential inlining, and also
decreases the size of librustc.so by 1.5% (OTOH, the size of `libstd.so`
increased by 0.5% for some reason).
Fixes#41696.
r? @nagisa
This commit integrates the `jobserver` crate into the compiler. The crate was
previously integrated in to Cargo as part of rust-lang/cargo#4110. The purpose
here is to two-fold:
* Primarily the compiler can cooperate with Cargo on parallelism. When you run
`cargo build -j4` then this'll make sure that the entire build process between
Cargo/rustc won't use more than 4 cores, whereas today you'd get 4 rustc
instances which may all try to spawn lots of threads.
* Secondarily rustc/Cargo can now integrate with a foreign GNU `make` jobserver.
This means that if you call cargo/rustc from `make` or another
jobserver-compatible implementation it'll use foreign parallelism settings
instead of creating new ones locally.
As the number of parallel codegen instances in the compiler continues to grow
over time with the advent of incremental compilation it's expected that this'll
become more of a problem, so this is intended to nip concurrent concerns in the
bud by having all the tools to cooperate!
Note that while rustc has support for itself creating a jobserver it's far more
likely that rustc will always use the jobserver configured by Cargo. Cargo today
will now set a jobserver unconditionally for rustc to use.
