-
Notifications
You must be signed in to change notification settings - Fork 12.8k
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
rustc: Enable LTO and multiple codegen units
This commit is a refactoring of the LTO backend in Rust to support compilations with multiple codegen units. The immediate result of this PR is to remove the artificial error emitted by rustc about `-C lto -C codegen-units-8`, but longer term this is intended to lay the groundwork for LTO with incremental compilation and ultimately be the underpinning of ThinLTO support. The problem here that needed solving is that when rustc is producing multiple codegen units in one compilation LTO needs to merge them all together. Previously only upstream dependencies were merged and it was inherently relied on that there was only one local codegen unit. Supporting this involved refactoring the optimization backend architecture for rustc, namely splitting the `optimize_and_codegen` function into `optimize` and `codegen`. After an LLVM module has been optimized it may be blocked and queued up for LTO, and only after LTO are modules code generated. Non-LTO compilations should look the same as they do today backend-wise, we'll spin up a thread for each codegen unit and optimize/codegen in that thread. LTO compilations will, however, send the LLVM module back to the coordinator thread once optimizations have finished. When all LLVM modules have finished optimizing the coordinator will invoke the LTO backend, producing a further list of LLVM modules. Currently this is always a list of one LLVM module. The coordinator then spawns further work to run LTO and code generation passes over each module. In the course of this refactoring a number of other pieces were refactored: * Management of the bytecode encoding in rlibs was centralized into one module instead of being scattered across LTO and linking. * Some internal refactorings on the link stage of the compiler was done to work directly from `CompiledModule` structures instead of lists of paths. * The trans time-graph output was tweaked a little to include a name on each bar and inflate the size of the bars a little
- Loading branch information
1 parent
d514263
commit ded38db
Showing
13 changed files
with
803 additions
and
430 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,160 @@ | ||
| // Copyright 2017 The Rust Project Developers. See the COPYRIGHT | ||
| // file at the top-level directory of this distribution and at | ||
| // http://rust-lang.org/COPYRIGHT. | ||
| // | ||
| // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or | ||
| // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license | ||
| // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your | ||
| // option. This file may not be copied, modified, or distributed | ||
| // except according to those terms. | ||
|
|
||
| //! Management of the encoding of LLVM bytecode into rlibs | ||
| //! | ||
| //! This module contains the management of encoding LLVM bytecode into rlibs, | ||
| //! primarily for the usage in LTO situations. Currently the compiler will | ||
| //! unconditionally encode LLVM-IR into rlibs regardless of what's happening | ||
| //! elsewhere, so we currently compress the bytecode via deflate to avoid taking | ||
| //! up too much space on disk. | ||
| //! | ||
| //! After compressing the bytecode we then have the rest of the format to | ||
| //! basically deal with various bugs in various archive implementations. The | ||
| //! format currently is: | ||
| //! | ||
| //! RLIB LLVM-BYTECODE OBJECT LAYOUT | ||
| //! Version 2 | ||
| //! Bytes Data | ||
| //! 0..10 "RUST_OBJECT" encoded in ASCII | ||
| //! 11..14 format version as little-endian u32 | ||
| //! 15..19 the length of the module identifier string | ||
| //! 20..n the module identifier string | ||
| //! n..n+8 size in bytes of deflate compressed LLVM bitcode as | ||
| //! little-endian u64 | ||
| //! n+9.. compressed LLVM bitcode | ||
| //! ? maybe a byte to make this whole thing even length | ||
|
|
||
| use std::io::{Read, Write}; | ||
| use std::ptr; | ||
| use std::str; | ||
|
|
||
| use flate2::Compression; | ||
| use flate2::read::DeflateDecoder; | ||
| use flate2::write::DeflateEncoder; | ||
|
|
||
| // This is the "magic number" expected at the beginning of a LLVM bytecode | ||
| // object in an rlib. | ||
| pub const RLIB_BYTECODE_OBJECT_MAGIC: &'static [u8] = b"RUST_OBJECT"; | ||
|
|
||
| // The version number this compiler will write to bytecode objects in rlibs | ||
| pub const RLIB_BYTECODE_OBJECT_VERSION: u8 = 2; | ||
|
|
||
| pub const RLIB_BYTECODE_EXTENSION: &str = "bytecode.encoded"; | ||
|
|
||
| pub fn encode(identifier: &str, bytecode: &[u8]) -> Vec<u8> { | ||
| let mut encoded = Vec::new(); | ||
|
|
||
| // Start off with the magic string | ||
| encoded.extend_from_slice(RLIB_BYTECODE_OBJECT_MAGIC); | ||
|
|
||
| // Next up is the version | ||
| encoded.extend_from_slice(&[RLIB_BYTECODE_OBJECT_VERSION, 0, 0, 0]); | ||
|
|
||
| // Next is the LLVM module identifier length + contents | ||
| let identifier_len = identifier.len(); | ||
| encoded.extend_from_slice(&[ | ||
| (identifier_len >> 0) as u8, | ||
| (identifier_len >> 8) as u8, | ||
| (identifier_len >> 16) as u8, | ||
| (identifier_len >> 24) as u8, | ||
| ]); | ||
| encoded.extend_from_slice(identifier.as_bytes()); | ||
|
|
||
| // Next is the LLVM module deflate compressed, prefixed with its length. We | ||
| // don't know its length yet, so fill in 0s | ||
| let deflated_size_pos = encoded.len(); | ||
| encoded.extend_from_slice(&[0, 0, 0, 0, 0, 0, 0, 0]); | ||
|
|
||
| let before = encoded.len(); | ||
| DeflateEncoder::new(&mut encoded, Compression::Fast) | ||
| .write_all(bytecode) | ||
| .unwrap(); | ||
| let after = encoded.len(); | ||
|
|
||
| // Fill in the length we reserved space for before | ||
| let bytecode_len = (after - before) as u64; | ||
| encoded[deflated_size_pos + 0] = (bytecode_len >> 0) as u8; | ||
| encoded[deflated_size_pos + 1] = (bytecode_len >> 8) as u8; | ||
| encoded[deflated_size_pos + 2] = (bytecode_len >> 16) as u8; | ||
| encoded[deflated_size_pos + 3] = (bytecode_len >> 24) as u8; | ||
| encoded[deflated_size_pos + 4] = (bytecode_len >> 32) as u8; | ||
| encoded[deflated_size_pos + 5] = (bytecode_len >> 40) as u8; | ||
| encoded[deflated_size_pos + 6] = (bytecode_len >> 48) as u8; | ||
| encoded[deflated_size_pos + 7] = (bytecode_len >> 56) as u8; | ||
|
|
||
| // If the number of bytes written to the object so far is odd, add a | ||
| // padding byte to make it even. This works around a crash bug in LLDB | ||
| // (see issue #15950) | ||
| if encoded.len() % 2 == 1 { | ||
| encoded.push(0); | ||
| } | ||
|
|
||
| return encoded | ||
| } | ||
|
|
||
| pub struct DecodedBytecode<'a> { | ||
| identifier: &'a str, | ||
| encoded_bytecode: &'a [u8], | ||
| } | ||
|
|
||
| impl<'a> DecodedBytecode<'a> { | ||
| pub fn new(data: &'a [u8]) -> Result<DecodedBytecode<'a>, String> { | ||
| if !data.starts_with(RLIB_BYTECODE_OBJECT_MAGIC) { | ||
| return Err(format!("magic bytecode prefix not found")) | ||
| } | ||
| let data = &data[RLIB_BYTECODE_OBJECT_MAGIC.len()..]; | ||
| if !data.starts_with(&[RLIB_BYTECODE_OBJECT_VERSION, 0, 0, 0]) { | ||
| return Err(format!("wrong version prefix found in bytecode")) | ||
| } | ||
| let data = &data[4..]; | ||
| if data.len() < 4 { | ||
| return Err(format!("bytecode corrupted")) | ||
| } | ||
| let identifier_len = unsafe { | ||
| u32::from_le(ptr::read_unaligned(data.as_ptr() as *const u32)) as usize | ||
| }; | ||
| let data = &data[4..]; | ||
| if data.len() < identifier_len { | ||
| return Err(format!("bytecode corrupted")) | ||
| } | ||
| let identifier = match str::from_utf8(&data[..identifier_len]) { | ||
| Ok(s) => s, | ||
| Err(_) => return Err(format!("bytecode corrupted")) | ||
| }; | ||
| let data = &data[identifier_len..]; | ||
| if data.len() < 8 { | ||
| return Err(format!("bytecode corrupted")) | ||
| } | ||
| let bytecode_len = unsafe { | ||
| u64::from_le(ptr::read_unaligned(data.as_ptr() as *const u64)) as usize | ||
| }; | ||
| let data = &data[8..]; | ||
| if data.len() < bytecode_len { | ||
| return Err(format!("bytecode corrupted")) | ||
| } | ||
| let encoded_bytecode = &data[..bytecode_len]; | ||
|
|
||
| Ok(DecodedBytecode { | ||
| identifier, | ||
| encoded_bytecode, | ||
| }) | ||
| } | ||
|
|
||
| pub fn bytecode(&self) -> Vec<u8> { | ||
| let mut data = Vec::new(); | ||
| DeflateDecoder::new(self.encoded_bytecode).read_to_end(&mut data).unwrap(); | ||
| return data | ||
| } | ||
|
|
||
| pub fn identifier(&self) -> &'a str { | ||
| self.identifier | ||
| } | ||
| } |
Oops, something went wrong.