rust/src/librustc_codegen_llvm/back/linker.rs

1081 lines
36 KiB
Rust
Raw Normal View History

// Copyright 2015 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.
rustc: Implement custom derive (macros 1.1) This commit is an implementation of [RFC 1681] which adds support to the compiler for first-class user-define custom `#[derive]` modes with a far more stable API than plugins have today. [RFC 1681]: https://github.com/rust-lang/rfcs/blob/master/text/1681-macros-1.1.md The main features added by this commit are: * A new `rustc-macro` crate-type. This crate type represents one which will provide custom `derive` implementations and perhaps eventually flower into the implementation of macros 2.0 as well. * A new `rustc_macro` crate in the standard distribution. This crate will provide the runtime interface between macro crates and the compiler. The API here is particularly conservative right now but has quite a bit of room to expand into any manner of APIs required by macro authors. * The ability to load new derive modes through the `#[macro_use]` annotations on other crates. All support added here is gated behind the `rustc_macro` feature gate, both for the library support (the `rustc_macro` crate) as well as the language features. There are a few minor differences from the implementation outlined in the RFC, such as the `rustc_macro` crate being available as a dylib and all symbols are `dlsym`'d directly instead of having a shim compiled. These should only affect the implementation, however, not the public interface. This commit also ended up touching a lot of code related to `#[derive]`, making a few notable changes: * Recognized derive attributes are no longer desugared to `derive_Foo`. Wasn't sure how to keep this behavior and *not* expose it to custom derive. * Derive attributes no longer have access to unstable features by default, they have to opt in on a granular level. * The `derive(Copy,Clone)` optimization is now done through another "obscure attribute" which is just intended to ferry along in the compiler that such an optimization is possible. The `derive(PartialEq,Eq)` optimization was also updated to do something similar. --- One part of this PR which needs to be improved before stabilizing are the errors and exact interfaces here. The error messages are relatively poor quality and there are surprising spects of this such as `#[derive(PartialEq, Eq, MyTrait)]` not working by default. The custom attributes added by the compiler end up becoming unstable again when going through a custom impl. Hopefully though this is enough to start allowing experimentation on crates.io! syntax-[breaking-change]
2016-08-22 17:07:11 -07:00
use std::collections::HashMap;
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
use std::ffi::{OsStr, OsString};
use std::fs::{self, File};
use std::io::prelude::*;
rustc: Implement custom derive (macros 1.1) This commit is an implementation of [RFC 1681] which adds support to the compiler for first-class user-define custom `#[derive]` modes with a far more stable API than plugins have today. [RFC 1681]: https://github.com/rust-lang/rfcs/blob/master/text/1681-macros-1.1.md The main features added by this commit are: * A new `rustc-macro` crate-type. This crate type represents one which will provide custom `derive` implementations and perhaps eventually flower into the implementation of macros 2.0 as well. * A new `rustc_macro` crate in the standard distribution. This crate will provide the runtime interface between macro crates and the compiler. The API here is particularly conservative right now but has quite a bit of room to expand into any manner of APIs required by macro authors. * The ability to load new derive modes through the `#[macro_use]` annotations on other crates. All support added here is gated behind the `rustc_macro` feature gate, both for the library support (the `rustc_macro` crate) as well as the language features. There are a few minor differences from the implementation outlined in the RFC, such as the `rustc_macro` crate being available as a dylib and all symbols are `dlsym`'d directly instead of having a shim compiled. These should only affect the implementation, however, not the public interface. This commit also ended up touching a lot of code related to `#[derive]`, making a few notable changes: * Recognized derive attributes are no longer desugared to `derive_Foo`. Wasn't sure how to keep this behavior and *not* expose it to custom derive. * Derive attributes no longer have access to unstable features by default, they have to opt in on a granular level. * The `derive(Copy,Clone)` optimization is now done through another "obscure attribute" which is just intended to ferry along in the compiler that such an optimization is possible. The `derive(PartialEq,Eq)` optimization was also updated to do something similar. --- One part of this PR which needs to be improved before stabilizing are the errors and exact interfaces here. The error messages are relatively poor quality and there are surprising spects of this such as `#[derive(PartialEq, Eq, MyTrait)]` not working by default. The custom attributes added by the compiler end up becoming unstable again when going through a custom impl. Hopefully though this is enough to start allowing experimentation on crates.io! syntax-[breaking-change]
2016-08-22 17:07:11 -07:00
use std::io::{self, BufWriter};
use std::path::{Path, PathBuf};
trans: Use LLVM's writeArchive to modify archives We have previously always relied upon an external tool, `ar`, to modify archives that the compiler produces (staticlibs, rlibs, etc). This approach, however, has a number of downsides: * Spawning a process is relatively expensive for small compilations * Encoding arguments across process boundaries often incurs unnecessary overhead or lossiness. For example `ar` has a tough time dealing with files that have the same name in archives, and the compiler copies many files around to ensure they can be passed to `ar` in a reasonable fashion. * Most `ar` programs found do **not** have the ability to target arbitrary platforms, so this is an extra tool which needs to be found/specified when cross compiling. The LLVM project has had a tool called `llvm-ar` for quite some time now, but it wasn't available in the standard LLVM libraries (it was just a standalone program). Recently, however, in LLVM 3.7, this functionality has been moved to a library and is now accessible by consumers of LLVM via the `writeArchive` function. This commit migrates our archive bindings to no longer invoke `ar` by default but instead make a library call to LLVM to do various operations. This solves all of the downsides listed above: * Archive management is now much faster, for example creating a "hello world" staticlib is now 6x faster (50ms => 8ms). Linking dynamic libraries also recently started requiring modification of rlibs, and linking a hello world dynamic library is now 2x faster. * The compiler is now one step closer to "hassle free" cross compilation because no external tool is needed for managing archives, LLVM does the right thing! This commit does not remove support for calling a system `ar` utility currently. We will continue to maintain compatibility with LLVM 3.5 and 3.6 looking forward (so the system LLVM can be used wherever possible), and in these cases we must shell out to a system utility. All nightly builds of Rust, however, will stop needing a system `ar`.
2015-07-09 00:14:20 -07:00
use back::archive;
use back::command::Command;
use back::symbol_export;
use rustc::hir::def_id::{LOCAL_CRATE, CrateNum};
use rustc::middle::dependency_format::Linkage;
use rustc::session::Session;
use rustc::session::config::{self, CrateType, OptLevel, DebugInfoLevel,
CrossLangLto};
use rustc::ty::TyCtxt;
use rustc_target::spec::{LinkerFlavor, LldFlavor};
use serialize::{json, Encoder};
/// For all the linkers we support, and information they might
/// need out of the shared crate context before we get rid of it.
pub struct LinkerInfo {
rustc: Implement custom derive (macros 1.1) This commit is an implementation of [RFC 1681] which adds support to the compiler for first-class user-define custom `#[derive]` modes with a far more stable API than plugins have today. [RFC 1681]: https://github.com/rust-lang/rfcs/blob/master/text/1681-macros-1.1.md The main features added by this commit are: * A new `rustc-macro` crate-type. This crate type represents one which will provide custom `derive` implementations and perhaps eventually flower into the implementation of macros 2.0 as well. * A new `rustc_macro` crate in the standard distribution. This crate will provide the runtime interface between macro crates and the compiler. The API here is particularly conservative right now but has quite a bit of room to expand into any manner of APIs required by macro authors. * The ability to load new derive modes through the `#[macro_use]` annotations on other crates. All support added here is gated behind the `rustc_macro` feature gate, both for the library support (the `rustc_macro` crate) as well as the language features. There are a few minor differences from the implementation outlined in the RFC, such as the `rustc_macro` crate being available as a dylib and all symbols are `dlsym`'d directly instead of having a shim compiled. These should only affect the implementation, however, not the public interface. This commit also ended up touching a lot of code related to `#[derive]`, making a few notable changes: * Recognized derive attributes are no longer desugared to `derive_Foo`. Wasn't sure how to keep this behavior and *not* expose it to custom derive. * Derive attributes no longer have access to unstable features by default, they have to opt in on a granular level. * The `derive(Copy,Clone)` optimization is now done through another "obscure attribute" which is just intended to ferry along in the compiler that such an optimization is possible. The `derive(PartialEq,Eq)` optimization was also updated to do something similar. --- One part of this PR which needs to be improved before stabilizing are the errors and exact interfaces here. The error messages are relatively poor quality and there are surprising spects of this such as `#[derive(PartialEq, Eq, MyTrait)]` not working by default. The custom attributes added by the compiler end up becoming unstable again when going through a custom impl. Hopefully though this is enough to start allowing experimentation on crates.io! syntax-[breaking-change]
2016-08-22 17:07:11 -07:00
exports: HashMap<CrateType, Vec<String>>,
}
impl LinkerInfo {
pub fn new(tcx: TyCtxt) -> LinkerInfo {
LinkerInfo {
exports: tcx.sess.crate_types.borrow().iter().map(|&c| {
(c, exported_symbols(tcx, c))
rustc: Implement custom derive (macros 1.1) This commit is an implementation of [RFC 1681] which adds support to the compiler for first-class user-define custom `#[derive]` modes with a far more stable API than plugins have today. [RFC 1681]: https://github.com/rust-lang/rfcs/blob/master/text/1681-macros-1.1.md The main features added by this commit are: * A new `rustc-macro` crate-type. This crate type represents one which will provide custom `derive` implementations and perhaps eventually flower into the implementation of macros 2.0 as well. * A new `rustc_macro` crate in the standard distribution. This crate will provide the runtime interface between macro crates and the compiler. The API here is particularly conservative right now but has quite a bit of room to expand into any manner of APIs required by macro authors. * The ability to load new derive modes through the `#[macro_use]` annotations on other crates. All support added here is gated behind the `rustc_macro` feature gate, both for the library support (the `rustc_macro` crate) as well as the language features. There are a few minor differences from the implementation outlined in the RFC, such as the `rustc_macro` crate being available as a dylib and all symbols are `dlsym`'d directly instead of having a shim compiled. These should only affect the implementation, however, not the public interface. This commit also ended up touching a lot of code related to `#[derive]`, making a few notable changes: * Recognized derive attributes are no longer desugared to `derive_Foo`. Wasn't sure how to keep this behavior and *not* expose it to custom derive. * Derive attributes no longer have access to unstable features by default, they have to opt in on a granular level. * The `derive(Copy,Clone)` optimization is now done through another "obscure attribute" which is just intended to ferry along in the compiler that such an optimization is possible. The `derive(PartialEq,Eq)` optimization was also updated to do something similar. --- One part of this PR which needs to be improved before stabilizing are the errors and exact interfaces here. The error messages are relatively poor quality and there are surprising spects of this such as `#[derive(PartialEq, Eq, MyTrait)]` not working by default. The custom attributes added by the compiler end up becoming unstable again when going through a custom impl. Hopefully though this is enough to start allowing experimentation on crates.io! syntax-[breaking-change]
2016-08-22 17:07:11 -07:00
}).collect(),
}
}
pub fn to_linker<'a>(&'a self,
cmd: Command,
sess: &'a Session) -> Box<dyn Linker+'a> {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
match sess.linker_flavor() {
LinkerFlavor::Lld(LldFlavor::Link) |
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
LinkerFlavor::Msvc => {
Box::new(MsvcLinker {
cmd,
sess,
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
info: self
}) as Box<dyn Linker>
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
}
LinkerFlavor::Em => {
Box::new(EmLinker {
cmd,
sess,
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
info: self
}) as Box<dyn Linker>
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
}
LinkerFlavor::Gcc => {
Box::new(GccLinker {
cmd,
sess,
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
info: self,
hinted_static: false,
is_ld: false,
}) as Box<dyn Linker>
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
}
LinkerFlavor::Lld(LldFlavor::Ld) |
LinkerFlavor::Lld(LldFlavor::Ld64) |
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
LinkerFlavor::Ld => {
Box::new(GccLinker {
cmd,
sess,
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
info: self,
hinted_static: false,
is_ld: true,
}) as Box<dyn Linker>
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
}
rust: Import LLD for linking wasm objects This commit imports the LLD project from LLVM to serve as the default linker for the `wasm32-unknown-unknown` target. The `binaryen` submoule is consequently removed along with "binaryen linker" support in rustc. Moving to LLD brings with it a number of benefits for wasm code: * LLD is itself an actual linker, so there's no need to compile all wasm code with LTO any more. As a result builds should be *much* speedier as LTO is no longer forcibly enabled for all builds of the wasm target. * LLD is quickly becoming an "official solution" for linking wasm code together. This, I believe at least, is intended to be the main supported linker for native code and wasm moving forward. Picking up support early on should help ensure that we can help LLD identify bugs and otherwise prove that it works great for all our use cases! * Improvements to the wasm toolchain are currently primarily focused around LLVM and LLD (from what I can tell at least), so it's in general much better to be on this bandwagon for bugfixes and new features. * Historical "hacks" like `wasm-gc` will soon no longer be necessary, LLD will [natively implement][gc] `--gc-sections` (better than `wasm-gc`!) which means a postprocessor is no longer needed to show off Rust's "small wasm binary size". LLD is added in a pretty standard way to rustc right now. A new rustbuild target was defined for building LLD, and this is executed when a compiler's sysroot is being assembled. LLD is compiled against the LLVM that we've got in tree, which means we're currently on the `release_60` branch, but this may get upgraded in the near future! LLD is placed into rustc's sysroot in a `bin` directory. This is similar to where `gcc.exe` can be found on Windows. This directory is automatically added to `PATH` whenever rustc executes the linker, allowing us to define a `WasmLd` linker which implements the interface that `wasm-ld`, LLD's frontend, expects. Like Emscripten the LLD target is currently only enabled for Tier 1 platforms, notably OSX/Windows/Linux, and will need to be installed manually for compiling to wasm on other platforms. LLD is by default turned off in rustbuild, and requires a `config.toml` option to be enabled to turn it on. Finally the unstable `#![wasm_import_memory]` attribute was also removed as LLD has a native option for controlling this. [gc]: https://reviews.llvm.org/D42511
2017-08-26 18:30:12 -07:00
LinkerFlavor::Lld(LldFlavor::Wasm) => {
Box::new(WasmLd {
cmd,
Upgrade to LLVM's master branch (LLVM 7) This commit upgrades the main LLVM submodule to LLVM's current master branch. The LLD submodule is updated in tandem as well as compiler-builtins. Along the way support was also added for LLVM 7's new features. This primarily includes the support for custom section concatenation natively in LLD so we now add wasm custom sections in LLVM IR rather than having custom support in rustc itself for doing so. Some other miscellaneous changes are: * We now pass `--gc-sections` to `wasm-ld` * The optimization level is now passed to `wasm-ld` * A `--stack-first` option is passed to LLD to have stack overflow always cause a trap instead of corrupting static data * The wasm target for LLVM switched to `wasm32-unknown-unknown`. * The syntax for aligned pointers has changed in LLVM IR and tests are updated to reflect this. * The `thumbv6m-none-eabi` target is disabled due to an [LLVM bug][llbug] Nowadays we've been mostly only upgrading whenever there's a major release of LLVM but enough changes have been happening on the wasm target that there's been growing motivation for quite some time now to upgrade out version of LLD. To upgrade LLD, however, we need to upgrade LLVM to avoid needing to build yet another version of LLVM on the builders. The revision of LLVM in use here is arbitrarily chosen. We will likely need to continue to update it over time if and when we discover bugs. Once LLVM 7 is fully released we can switch to that channel as well. [llbug]: https://bugs.llvm.org/show_bug.cgi?id=37382
2018-06-01 10:20:00 -07:00
sess,
}) as Box<dyn Linker>
std: Add a new wasm32-unknown-unknown target This commit adds a new target to the compiler: wasm32-unknown-unknown. This target is a reimagining of what it looks like to generate WebAssembly code from Rust. Instead of using Emscripten which can bring with it a weighty runtime this instead is a target which uses only the LLVM backend for WebAssembly and a "custom linker" for now which will hopefully one day be direct calls to lld. Notable features of this target include: * There is zero runtime footprint. The target assumes nothing exists other than the wasm32 instruction set. * There is zero toolchain footprint beyond adding the target. No custom linker is needed, rustc contains everything. * Very small wasm modules can be generated directly from Rust code using this target. * Most of the standard library is stubbed out to return an error, but anything related to allocation works (aka `HashMap`, `Vec`, etc). * Naturally, any `#[no_std]` crate should be 100% compatible with this new target. This target is currently somewhat janky due to how linking works. The "linking" is currently unconditional whole program LTO (aka LLVM is being used as a linker). Naturally that means compiling programs is pretty slow! Eventually though this target should have a linker. This target is also intended to be quite experimental. I'm hoping that this can act as a catalyst for further experimentation in Rust with WebAssembly. Breaking changes are very likely to land to this target, so it's not recommended to rely on it in any critical capacity yet. We'll let you know when it's "production ready". --- Currently testing-wise this target is looking pretty good but isn't complete. I've got almost the entire `run-pass` test suite working with this target (lots of tests ignored, but many passing as well). The `core` test suite is still getting LLVM bugs fixed to get that working and will take some time. Relatively simple programs all seem to work though! --- It's worth nothing that you may not immediately see the "smallest possible wasm module" for the input you feed to rustc. For various reasons it's very difficult to get rid of the final "bloat" in vanilla rustc (again, a real linker should fix all this). For now what you'll have to do is: cargo install --git https://github.com/alexcrichton/wasm-gc wasm-gc foo.wasm bar.wasm And then `bar.wasm` should be the smallest we can get it! --- In any case for now I'd love feedback on this, particularly on the various integration points if you've got better ideas of how to approach them!
2017-10-22 20:01:00 -07:00
}
}
}
}
/// Linker abstraction used by back::link to build up the command to invoke a
/// linker.
///
/// This trait is the total list of requirements needed by `back::link` and
/// represents the meaning of each option being passed down. This trait is then
/// used to dispatch on whether a GNU-like linker (generally `ld.exe`) or an
/// MSVC linker (e.g. `link.exe`) is being used.
pub trait Linker {
fn link_dylib(&mut self, lib: &str);
fn link_rust_dylib(&mut self, lib: &str, path: &Path);
fn link_framework(&mut self, framework: &str);
fn link_staticlib(&mut self, lib: &str);
fn link_rlib(&mut self, lib: &Path);
fn link_whole_rlib(&mut self, lib: &Path);
fn link_whole_staticlib(&mut self, lib: &str, search_path: &[PathBuf]);
fn include_path(&mut self, path: &Path);
fn framework_path(&mut self, path: &Path);
fn output_filename(&mut self, path: &Path);
fn add_object(&mut self, path: &Path);
fn gc_sections(&mut self, keep_metadata: bool);
fn position_independent_executable(&mut self);
fn no_position_independent_executable(&mut self);
fn full_relro(&mut self);
fn partial_relro(&mut self);
fn no_relro(&mut self);
fn optimize(&mut self);
fn pgo_gen(&mut self);
fn debuginfo(&mut self);
fn no_default_libraries(&mut self);
fn build_dylib(&mut self, out_filename: &Path);
fn build_static_executable(&mut self);
fn args(&mut self, args: &[String]);
fn export_symbols(&mut self, tmpdir: &Path, crate_type: CrateType);
fn subsystem(&mut self, subsystem: &str);
fn group_start(&mut self);
fn group_end(&mut self);
fn cross_lang_lto(&mut self);
// Should have been finalize(self), but we don't support self-by-value on trait objects (yet?).
fn finalize(&mut self) -> Command;
}
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
pub struct GccLinker<'a> {
cmd: Command,
sess: &'a Session,
info: &'a LinkerInfo,
hinted_static: bool, // Keeps track of the current hinting mode.
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
// Link as ld
is_ld: bool,
}
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
impl<'a> GccLinker<'a> {
/// Argument that must be passed *directly* to the linker
///
/// These arguments need to be prepended with '-Wl,' when a gcc-style linker is used
fn linker_arg<S>(&mut self, arg: S) -> &mut Self
where S: AsRef<OsStr>
{
if !self.is_ld {
let mut os = OsString::from("-Wl,");
os.push(arg.as_ref());
self.cmd.arg(os);
} else {
self.cmd.arg(arg);
}
self
}
fn takes_hints(&self) -> bool {
!self.sess.target.target.options.is_like_osx
}
// Some platforms take hints about whether a library is static or dynamic.
// For those that support this, we ensure we pass the option if the library
// was flagged "static" (most defaults are dynamic) to ensure that if
// libfoo.a and libfoo.so both exist that the right one is chosen.
fn hint_static(&mut self) {
if !self.takes_hints() { return }
if !self.hinted_static {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
self.linker_arg("-Bstatic");
self.hinted_static = true;
}
}
fn hint_dynamic(&mut self) {
if !self.takes_hints() { return }
if self.hinted_static {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
self.linker_arg("-Bdynamic");
self.hinted_static = false;
}
}
fn push_cross_lang_lto_args(&mut self, plugin_path: Option<&OsStr>) {
if let Some(plugin_path) = plugin_path {
let mut arg = OsString::from("-plugin=");
arg.push(plugin_path);
self.linker_arg(&arg);
}
let opt_level = match self.sess.opts.optimize {
config::OptLevel::No => "O0",
config::OptLevel::Less => "O1",
config::OptLevel::Default => "O2",
config::OptLevel::Aggressive => "O3",
config::OptLevel::Size => "Os",
config::OptLevel::SizeMin => "Oz",
};
self.linker_arg(&format!("-plugin-opt={}", opt_level));
self.linker_arg(&format!("-plugin-opt=mcpu={}", self.sess.target_cpu()));
match self.sess.opts.cg.lto {
config::Lto::Thin |
config::Lto::ThinLocal => {
self.linker_arg(&format!("-plugin-opt=thin"));
}
config::Lto::Fat |
config::Lto::Yes |
config::Lto::No => {
// default to regular LTO
}
}
}
}
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
impl<'a> Linker for GccLinker<'a> {
fn link_dylib(&mut self, lib: &str) { self.hint_dynamic(); self.cmd.arg("-l").arg(lib); }
fn link_staticlib(&mut self, lib: &str) { self.hint_static(); self.cmd.arg("-l").arg(lib); }
fn link_rlib(&mut self, lib: &Path) { self.hint_static(); self.cmd.arg(lib); }
fn include_path(&mut self, path: &Path) { self.cmd.arg("-L").arg(path); }
fn framework_path(&mut self, path: &Path) { self.cmd.arg("-F").arg(path); }
fn output_filename(&mut self, path: &Path) { self.cmd.arg("-o").arg(path); }
fn add_object(&mut self, path: &Path) { self.cmd.arg(path); }
fn position_independent_executable(&mut self) { self.cmd.arg("-pie"); }
fn no_position_independent_executable(&mut self) { self.cmd.arg("-no-pie"); }
fn full_relro(&mut self) { self.linker_arg("-z,relro,-z,now"); }
fn partial_relro(&mut self) { self.linker_arg("-z,relro"); }
fn no_relro(&mut self) { self.linker_arg("-z,norelro"); }
fn build_static_executable(&mut self) { self.cmd.arg("-static"); }
fn args(&mut self, args: &[String]) { self.cmd.args(args); }
fn link_rust_dylib(&mut self, lib: &str, _path: &Path) {
self.hint_dynamic();
self.cmd.arg("-l").arg(lib);
}
fn link_framework(&mut self, framework: &str) {
self.hint_dynamic();
self.cmd.arg("-framework").arg(framework);
}
// Here we explicitly ask that the entire archive is included into the
// result artifact. For more details see #15460, but the gist is that
// the linker will strip away any unused objects in the archive if we
// don't otherwise explicitly reference them. This can occur for
// libraries which are just providing bindings, libraries with generic
// functions, etc.
fn link_whole_staticlib(&mut self, lib: &str, search_path: &[PathBuf]) {
self.hint_static();
let target = &self.sess.target.target;
if !target.options.is_like_osx {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
self.linker_arg("--whole-archive").cmd.arg("-l").arg(lib);
self.linker_arg("--no-whole-archive");
} else {
// -force_load is the macOS equivalent of --whole-archive, but it
// involves passing the full path to the library to link.
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
let mut v = OsString::from("-force_load,");
trans: Use LLVM's writeArchive to modify archives We have previously always relied upon an external tool, `ar`, to modify archives that the compiler produces (staticlibs, rlibs, etc). This approach, however, has a number of downsides: * Spawning a process is relatively expensive for small compilations * Encoding arguments across process boundaries often incurs unnecessary overhead or lossiness. For example `ar` has a tough time dealing with files that have the same name in archives, and the compiler copies many files around to ensure they can be passed to `ar` in a reasonable fashion. * Most `ar` programs found do **not** have the ability to target arbitrary platforms, so this is an extra tool which needs to be found/specified when cross compiling. The LLVM project has had a tool called `llvm-ar` for quite some time now, but it wasn't available in the standard LLVM libraries (it was just a standalone program). Recently, however, in LLVM 3.7, this functionality has been moved to a library and is now accessible by consumers of LLVM via the `writeArchive` function. This commit migrates our archive bindings to no longer invoke `ar` by default but instead make a library call to LLVM to do various operations. This solves all of the downsides listed above: * Archive management is now much faster, for example creating a "hello world" staticlib is now 6x faster (50ms => 8ms). Linking dynamic libraries also recently started requiring modification of rlibs, and linking a hello world dynamic library is now 2x faster. * The compiler is now one step closer to "hassle free" cross compilation because no external tool is needed for managing archives, LLVM does the right thing! This commit does not remove support for calling a system `ar` utility currently. We will continue to maintain compatibility with LLVM 3.5 and 3.6 looking forward (so the system LLVM can be used wherever possible), and in these cases we must shell out to a system utility. All nightly builds of Rust, however, will stop needing a system `ar`.
2015-07-09 00:14:20 -07:00
v.push(&archive::find_library(lib, search_path, &self.sess));
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
self.linker_arg(&v);
}
}
fn link_whole_rlib(&mut self, lib: &Path) {
self.hint_static();
if self.sess.target.target.options.is_like_osx {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
let mut v = OsString::from("-force_load,");
v.push(lib);
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
self.linker_arg(&v);
} else {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
self.linker_arg("--whole-archive").cmd.arg(lib);
self.linker_arg("--no-whole-archive");
}
}
fn gc_sections(&mut self, keep_metadata: bool) {
// The dead_strip option to the linker specifies that functions and data
// unreachable by the entry point will be removed. This is quite useful
// with Rust's compilation model of compiling libraries at a time into
// one object file. For example, this brings hello world from 1.7MB to
// 458K.
//
// Note that this is done for both executables and dynamic libraries. We
// won't get much benefit from dylibs because LLVM will have already
// stripped away as much as it could. This has not been seen to impact
// link times negatively.
//
// -dead_strip can't be part of the pre_link_args because it's also used
// for partial linking when using multiple codegen units (-r). So we
// insert it here.
if self.sess.target.target.options.is_like_osx {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
self.linker_arg("-dead_strip");
2016-01-28 14:02:31 +03:00
} else if self.sess.target.target.options.is_like_solaris {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
self.linker_arg("-z");
self.linker_arg("ignore");
// If we're building a dylib, we don't use --gc-sections because LLVM
// has already done the best it can do, and we also don't want to
// eliminate the metadata. If we're building an executable, however,
// --gc-sections drops the size of hello world from 1.8MB to 597K, a 67%
// reduction.
} else if !keep_metadata {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
self.linker_arg("--gc-sections");
}
}
fn optimize(&mut self) {
if !self.sess.target.target.options.linker_is_gnu { return }
// GNU-style linkers support optimization with -O. GNU ld doesn't
// need a numeric argument, but other linkers do.
if self.sess.opts.optimize == config::OptLevel::Default ||
self.sess.opts.optimize == config::OptLevel::Aggressive {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
self.linker_arg("-O1");
}
}
fn pgo_gen(&mut self) {
if !self.sess.target.target.options.linker_is_gnu { return }
// If we're doing PGO generation stuff and on a GNU-like linker, use the
// "-u" flag to properly pull in the profiler runtime bits.
//
// This is because LLVM otherwise won't add the needed initialization
// for us on Linux (though the extra flag should be harmless if it
// does).
//
// See https://reviews.llvm.org/D14033 and https://reviews.llvm.org/D14030.
//
// Though it may be worth to try to revert those changes upstream, since
// the overhead of the initialization should be minor.
self.cmd.arg("-u");
self.cmd.arg("__llvm_profile_runtime");
}
fn debuginfo(&mut self) {
match self.sess.opts.debuginfo {
DebugInfoLevel::NoDebugInfo => {
// If we are building without debuginfo enabled and we were called with
// `-Zstrip-debuginfo-if-disabled=yes`, tell the linker to strip any debuginfo
// found when linking to get rid of symbols from libstd.
match self.sess.opts.debugging_opts.strip_debuginfo_if_disabled {
Some(true) => { self.linker_arg("-S"); },
_ => {},
}
},
_ => {},
};
}
fn no_default_libraries(&mut self) {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
if !self.is_ld {
self.cmd.arg("-nodefaultlibs");
}
}
fn build_dylib(&mut self, out_filename: &Path) {
// On mac we need to tell the linker to let this library be rpathed
if self.sess.target.target.options.is_like_osx {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
self.cmd.arg("-dynamiclib");
self.linker_arg("-dylib");
rustbuild: Fix LC_ID_DYLIB directives on OSX Currently libraries installed by rustbuild on OSX have an incorrect `LC_ID_DYLIB` directive located in the dynamic libraries that are installed. The directive we expect looks like: @rpath/libstd.dylib Which means that if you want to find that dynamic library you should look at the dylib's other `@rpath` directives. Typically our `@rpath` directives look like `@loader_path/../lib` for the compiler as that's where the installed libraries will be located. Currently, though, rustbuild produces dylibs with the directive that looks like: /Users/rustbuild/src/rust-buildbot/slave/nightly-dist-rustc-mac/build/build/x86_64-apple-darwin/stage1-std/x86_64-apple-darwin/release/deps/libstd-713ad88203512705.dylib In other words, the build directory is encoded erroneously. The compiler already [knows how] to change this directive, but it only passes that argument when `-C rpath` is also passed. The rustbuild system, however, explicitly [does not pass] this option explicitly and instead bakes its own. This logic then also erroneously didn't pass `-Wl,-install_name` like the compiler. [knows how]: https://github.com/rust-lang/rust/blob/4a008cccaabc8b3fe65ccf5868b9d16319c9ac58/src/librustc_trans/back/linker.rs#L210-L214 [does not pass]: https://github.com/rust-lang/rust/blob/4a008cccaabc8b3fe65ccf5868b9d16319c9ac58/src/bootstrap/bin/rustc.rs#L133-L158 To fix this regression this patch introduces a new `-Z` flag, `-Z osx-rpath-install-name` which basically just forces the compiler to take the previous `-install_name` branch when creating a dynamic library. Hopefully we can sort out a better rpath story in the future, but for now this "hack" should suffice in getting our nightly builds back to the same state as before. Closes #38430
2016-12-17 14:11:02 -08:00
// Note that the `osx_rpath_install_name` option here is a hack
// purely to support rustbuild right now, we should get a more
// principled solution at some point to force the compiler to pass
// the right `-Wl,-install_name` with an `@rpath` in it.
if self.sess.opts.cg.rpath ||
self.sess.opts.debugging_opts.osx_rpath_install_name {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
let mut v = OsString::from("-install_name,@rpath/");
v.push(out_filename.file_name().unwrap());
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
self.linker_arg(&v);
}
} else {
self.cmd.arg("-shared");
}
}
fn export_symbols(&mut self, tmpdir: &Path, crate_type: CrateType) {
// If we're compiling a dylib, then we let symbol visibility in object
// files to take care of whether they're exported or not.
//
// If we're compiling a cdylib, however, we manually create a list of
// exported symbols to ensure we don't expose any more. The object files
// have far more public symbols than we actually want to export, so we
// hide them all here.
rustc: Implement custom derive (macros 1.1) This commit is an implementation of [RFC 1681] which adds support to the compiler for first-class user-define custom `#[derive]` modes with a far more stable API than plugins have today. [RFC 1681]: https://github.com/rust-lang/rfcs/blob/master/text/1681-macros-1.1.md The main features added by this commit are: * A new `rustc-macro` crate-type. This crate type represents one which will provide custom `derive` implementations and perhaps eventually flower into the implementation of macros 2.0 as well. * A new `rustc_macro` crate in the standard distribution. This crate will provide the runtime interface between macro crates and the compiler. The API here is particularly conservative right now but has quite a bit of room to expand into any manner of APIs required by macro authors. * The ability to load new derive modes through the `#[macro_use]` annotations on other crates. All support added here is gated behind the `rustc_macro` feature gate, both for the library support (the `rustc_macro` crate) as well as the language features. There are a few minor differences from the implementation outlined in the RFC, such as the `rustc_macro` crate being available as a dylib and all symbols are `dlsym`'d directly instead of having a shim compiled. These should only affect the implementation, however, not the public interface. This commit also ended up touching a lot of code related to `#[derive]`, making a few notable changes: * Recognized derive attributes are no longer desugared to `derive_Foo`. Wasn't sure how to keep this behavior and *not* expose it to custom derive. * Derive attributes no longer have access to unstable features by default, they have to opt in on a granular level. * The `derive(Copy,Clone)` optimization is now done through another "obscure attribute" which is just intended to ferry along in the compiler that such an optimization is possible. The `derive(PartialEq,Eq)` optimization was also updated to do something similar. --- One part of this PR which needs to be improved before stabilizing are the errors and exact interfaces here. The error messages are relatively poor quality and there are surprising spects of this such as `#[derive(PartialEq, Eq, MyTrait)]` not working by default. The custom attributes added by the compiler end up becoming unstable again when going through a custom impl. Hopefully though this is enough to start allowing experimentation on crates.io! syntax-[breaking-change]
2016-08-22 17:07:11 -07:00
if crate_type == CrateType::CrateTypeDylib ||
crate_type == CrateType::CrateTypeProcMacro {
return
}
let mut arg = OsString::new();
let path = tmpdir.join("list");
debug!("EXPORTED SYMBOLS:");
if self.sess.target.target.options.is_like_osx {
// Write a plain, newline-separated list of symbols
let res = (|| -> io::Result<()> {
let mut f = BufWriter::new(File::create(&path)?);
rustc: Implement custom derive (macros 1.1) This commit is an implementation of [RFC 1681] which adds support to the compiler for first-class user-define custom `#[derive]` modes with a far more stable API than plugins have today. [RFC 1681]: https://github.com/rust-lang/rfcs/blob/master/text/1681-macros-1.1.md The main features added by this commit are: * A new `rustc-macro` crate-type. This crate type represents one which will provide custom `derive` implementations and perhaps eventually flower into the implementation of macros 2.0 as well. * A new `rustc_macro` crate in the standard distribution. This crate will provide the runtime interface between macro crates and the compiler. The API here is particularly conservative right now but has quite a bit of room to expand into any manner of APIs required by macro authors. * The ability to load new derive modes through the `#[macro_use]` annotations on other crates. All support added here is gated behind the `rustc_macro` feature gate, both for the library support (the `rustc_macro` crate) as well as the language features. There are a few minor differences from the implementation outlined in the RFC, such as the `rustc_macro` crate being available as a dylib and all symbols are `dlsym`'d directly instead of having a shim compiled. These should only affect the implementation, however, not the public interface. This commit also ended up touching a lot of code related to `#[derive]`, making a few notable changes: * Recognized derive attributes are no longer desugared to `derive_Foo`. Wasn't sure how to keep this behavior and *not* expose it to custom derive. * Derive attributes no longer have access to unstable features by default, they have to opt in on a granular level. * The `derive(Copy,Clone)` optimization is now done through another "obscure attribute" which is just intended to ferry along in the compiler that such an optimization is possible. The `derive(PartialEq,Eq)` optimization was also updated to do something similar. --- One part of this PR which needs to be improved before stabilizing are the errors and exact interfaces here. The error messages are relatively poor quality and there are surprising spects of this such as `#[derive(PartialEq, Eq, MyTrait)]` not working by default. The custom attributes added by the compiler end up becoming unstable again when going through a custom impl. Hopefully though this is enough to start allowing experimentation on crates.io! syntax-[breaking-change]
2016-08-22 17:07:11 -07:00
for sym in self.info.exports[&crate_type].iter() {
debug!(" _{}", sym);
writeln!(f, "_{}", sym)?;
}
Ok(())
})();
if let Err(e) = res {
self.sess.fatal(&format!("failed to write lib.def file: {}", e));
}
} else {
// Write an LD version script
let res = (|| -> io::Result<()> {
let mut f = BufWriter::new(File::create(&path)?);
writeln!(f, "{{\n global:")?;
rustc: Implement custom derive (macros 1.1) This commit is an implementation of [RFC 1681] which adds support to the compiler for first-class user-define custom `#[derive]` modes with a far more stable API than plugins have today. [RFC 1681]: https://github.com/rust-lang/rfcs/blob/master/text/1681-macros-1.1.md The main features added by this commit are: * A new `rustc-macro` crate-type. This crate type represents one which will provide custom `derive` implementations and perhaps eventually flower into the implementation of macros 2.0 as well. * A new `rustc_macro` crate in the standard distribution. This crate will provide the runtime interface between macro crates and the compiler. The API here is particularly conservative right now but has quite a bit of room to expand into any manner of APIs required by macro authors. * The ability to load new derive modes through the `#[macro_use]` annotations on other crates. All support added here is gated behind the `rustc_macro` feature gate, both for the library support (the `rustc_macro` crate) as well as the language features. There are a few minor differences from the implementation outlined in the RFC, such as the `rustc_macro` crate being available as a dylib and all symbols are `dlsym`'d directly instead of having a shim compiled. These should only affect the implementation, however, not the public interface. This commit also ended up touching a lot of code related to `#[derive]`, making a few notable changes: * Recognized derive attributes are no longer desugared to `derive_Foo`. Wasn't sure how to keep this behavior and *not* expose it to custom derive. * Derive attributes no longer have access to unstable features by default, they have to opt in on a granular level. * The `derive(Copy,Clone)` optimization is now done through another "obscure attribute" which is just intended to ferry along in the compiler that such an optimization is possible. The `derive(PartialEq,Eq)` optimization was also updated to do something similar. --- One part of this PR which needs to be improved before stabilizing are the errors and exact interfaces here. The error messages are relatively poor quality and there are surprising spects of this such as `#[derive(PartialEq, Eq, MyTrait)]` not working by default. The custom attributes added by the compiler end up becoming unstable again when going through a custom impl. Hopefully though this is enough to start allowing experimentation on crates.io! syntax-[breaking-change]
2016-08-22 17:07:11 -07:00
for sym in self.info.exports[&crate_type].iter() {
debug!(" {};", sym);
writeln!(f, " {};", sym)?;
}
writeln!(f, "\n local:\n *;\n}};")?;
Ok(())
})();
if let Err(e) = res {
self.sess.fatal(&format!("failed to write version script: {}", e));
}
}
if self.sess.target.target.options.is_like_osx {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
if !self.is_ld {
arg.push("-Wl,")
}
arg.push("-exported_symbols_list,");
} else if self.sess.target.target.options.is_like_solaris {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
if !self.is_ld {
arg.push("-Wl,")
}
arg.push("-M,");
} else {
-Z linker-flavor This patch adds a `-Z linker-flavor` flag to rustc which can be used to invoke the linker using a different interface. For example, by default rustc assumes that all the Linux targets will be linked using GCC. This makes it impossible to use LLD as a linker using just `-C linker=ld.lld` because that will invoke LLD with invalid command line arguments. (e.g. rustc will pass -Wl,--gc-sections to LLD but LLD doesn't understand that; --gc-sections would be the right argument) With this patch one can pass `-Z linker-flavor=ld` to rustc to invoke the linker using a LD-like interface. This way, `rustc -C linker=ld.lld -Z linker-flavor=ld` will invoke LLD with the right arguments. `-Z linker-flavor` accepts 4 different arguments: `em` (emcc), `ld`, `gcc`, `msvc` (link.exe). `em`, `gnu` and `msvc` cover all the existing linker interfaces. `ld` is a new flavor for interfacing GNU's ld and LLD. This patch also changes target specifications. `linker-flavor` is now a mandatory field that specifies the *default* linker flavor that the target will use. This change also makes the linker interface *explicit*; before, it used to be derived from other fields like linker-is-gnu, is-like-msvc, is-like-emscripten, etc. Another change to target specifications is that the fields `pre-link-args`, `post-link-args` and `late-link-args` now expect a map from flavor to linker arguments. ``` diff - "pre-link-args": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "pre-link-args": { + "gcc": ["-Wl,--as-needed", "-Wl,-z,-noexecstack"], + "ld": ["--as-needed", "-z,-noexecstack"], + }, ``` [breaking-change] for users of custom targets specifications
2017-02-21 14:47:15 -05:00
if !self.is_ld {
arg.push("-Wl,")
}
arg.push("--version-script=");
}
arg.push(&path);
self.cmd.arg(arg);
}
fn subsystem(&mut self, subsystem: &str) {
self.linker_arg(&format!("--subsystem,{}", subsystem));
}
fn finalize(&mut self) -> Command {
self.hint_dynamic(); // Reset to default before returning the composed command line.
let mut cmd = Command::new("");
::std::mem::swap(&mut cmd, &mut self.cmd);
cmd
}
fn group_start(&mut self) {
if !self.sess.target.target.options.is_like_osx {
self.linker_arg("--start-group");
}
}
fn group_end(&mut self) {
if !self.sess.target.target.options.is_like_osx {
self.linker_arg("--end-group");
}
}
fn cross_lang_lto(&mut self) {
match self.sess.opts.debugging_opts.cross_lang_lto {
CrossLangLto::Disabled => {
// Nothing to do
}
CrossLangLto::LinkerPluginAuto => {
self.push_cross_lang_lto_args(None);
}
CrossLangLto::LinkerPlugin(ref path) => {
self.push_cross_lang_lto_args(Some(path.as_os_str()));
}
}
}
}
pub struct MsvcLinker<'a> {
cmd: Command,
sess: &'a Session,
info: &'a LinkerInfo
}
impl<'a> Linker for MsvcLinker<'a> {
fn link_rlib(&mut self, lib: &Path) { self.cmd.arg(lib); }
fn add_object(&mut self, path: &Path) { self.cmd.arg(path); }
fn args(&mut self, args: &[String]) { self.cmd.args(args); }
fn build_dylib(&mut self, out_filename: &Path) {
self.cmd.arg("/DLL");
let mut arg: OsString = "/IMPLIB:".into();
arg.push(out_filename.with_extension("dll.lib"));
self.cmd.arg(arg);
}
fn build_static_executable(&mut self) {
// noop
}
fn gc_sections(&mut self, _keep_metadata: bool) {
// MSVC's ICF (Identical COMDAT Folding) link optimization is
// slow for Rust and thus we disable it by default when not in
// optimization build.
if self.sess.opts.optimize != config::OptLevel::No {
self.cmd.arg("/OPT:REF,ICF");
} else {
// It is necessary to specify NOICF here, because /OPT:REF
// implies ICF by default.
self.cmd.arg("/OPT:REF,NOICF");
}
}
fn link_dylib(&mut self, lib: &str) {
self.cmd.arg(&format!("{}.lib", lib));
}
fn link_rust_dylib(&mut self, lib: &str, path: &Path) {
// When producing a dll, the MSVC linker may not actually emit a
// `foo.lib` file if the dll doesn't actually export any symbols, so we
// check to see if the file is there and just omit linking to it if it's
// not present.
let name = format!("{}.dll.lib", lib);
if fs::metadata(&path.join(&name)).is_ok() {
self.cmd.arg(name);
}
}
fn link_staticlib(&mut self, lib: &str) {
self.cmd.arg(&format!("{}.lib", lib));
}
fn position_independent_executable(&mut self) {
// noop
}
fn no_position_independent_executable(&mut self) {
// noop
}
fn full_relro(&mut self) {
// noop
}
fn partial_relro(&mut self) {
// noop
}
fn no_relro(&mut self) {
// noop
}
fn no_default_libraries(&mut self) {
// Currently we don't pass the /NODEFAULTLIB flag to the linker on MSVC
// as there's been trouble in the past of linking the C++ standard
// library required by LLVM. This likely needs to happen one day, but
// in general Windows is also a more controlled environment than
// Unix, so it's not necessarily as critical that this be implemented.
//
// Note that there are also some licensing worries about statically
// linking some libraries which require a specific agreement, so it may
// not ever be possible for us to pass this flag.
}
fn include_path(&mut self, path: &Path) {
let mut arg = OsString::from("/LIBPATH:");
arg.push(path);
self.cmd.arg(&arg);
}
fn output_filename(&mut self, path: &Path) {
let mut arg = OsString::from("/OUT:");
arg.push(path);
self.cmd.arg(&arg);
}
fn framework_path(&mut self, _path: &Path) {
bug!("frameworks are not supported on windows")
}
fn link_framework(&mut self, _framework: &str) {
bug!("frameworks are not supported on windows")
}
fn link_whole_staticlib(&mut self, lib: &str, _search_path: &[PathBuf]) {
// not supported?
self.link_staticlib(lib);
}
fn link_whole_rlib(&mut self, path: &Path) {
// not supported?
self.link_rlib(path);
}
fn optimize(&mut self) {
// Needs more investigation of `/OPT` arguments
}
fn pgo_gen(&mut self) {
// Nothing needed here.
}
fn debuginfo(&mut self) {
// This will cause the Microsoft linker to generate a PDB file
// from the CodeView line tables in the object files.
self.cmd.arg("/DEBUG");
// This will cause the Microsoft linker to embed .natvis info into the the PDB file
let sysroot = self.sess.sysroot();
let natvis_dir_path = sysroot.join("lib\\rustlib\\etc");
if let Ok(natvis_dir) = fs::read_dir(&natvis_dir_path) {
// LLVM 5.0.0's lld-link frontend doesn't yet recognize, and chokes
// on, the /NATVIS:... flags. LLVM 6 (or earlier) should at worst ignore
// them, eventually mooting this workaround, per this landed patch:
// https://github.com/llvm-mirror/lld/commit/27b9c4285364d8d76bb43839daa100
if let Some(ref linker_path) = self.sess.opts.cg.linker {
if let Some(linker_name) = Path::new(&linker_path).file_stem() {
if linker_name.to_str().unwrap().to_lowercase() == "lld-link" {
self.sess.warn("not embedding natvis: lld-link may not support the flag");
return;
}
}
}
for entry in natvis_dir {
match entry {
Ok(entry) => {
let path = entry.path();
if path.extension() == Some("natvis".as_ref()) {
let mut arg = OsString::from("/NATVIS:");
arg.push(path);
self.cmd.arg(arg);
}
},
Err(err) => {
self.sess.warn(&format!("error enumerating natvis directory: {}", err));
},
}
}
}
}
// Currently the compiler doesn't use `dllexport` (an LLVM attribute) to
// export symbols from a dynamic library. When building a dynamic library,
// however, we're going to want some symbols exported, so this function
// generates a DEF file which lists all the symbols.
//
// The linker will read this `*.def` file and export all the symbols from
// the dynamic library. Note that this is not as simple as just exporting
2018-05-08 16:10:16 +03:00
// all the symbols in the current crate (as specified by `codegen.reachable`)
// but rather we also need to possibly export the symbols of upstream
// crates. Upstream rlibs may be linked statically to this dynamic library,
// in which case they may continue to transitively be used and hence need
// their symbols exported.
fn export_symbols(&mut self,
tmpdir: &Path,
crate_type: CrateType) {
let path = tmpdir.join("lib.def");
let res = (|| -> io::Result<()> {
let mut f = BufWriter::new(File::create(&path)?);
// Start off with the standard module name header and then go
// straight to exports.
writeln!(f, "LIBRARY")?;
writeln!(f, "EXPORTS")?;
rustc: Implement custom derive (macros 1.1) This commit is an implementation of [RFC 1681] which adds support to the compiler for first-class user-define custom `#[derive]` modes with a far more stable API than plugins have today. [RFC 1681]: https://github.com/rust-lang/rfcs/blob/master/text/1681-macros-1.1.md The main features added by this commit are: * A new `rustc-macro` crate-type. This crate type represents one which will provide custom `derive` implementations and perhaps eventually flower into the implementation of macros 2.0 as well. * A new `rustc_macro` crate in the standard distribution. This crate will provide the runtime interface between macro crates and the compiler. The API here is particularly conservative right now but has quite a bit of room to expand into any manner of APIs required by macro authors. * The ability to load new derive modes through the `#[macro_use]` annotations on other crates. All support added here is gated behind the `rustc_macro` feature gate, both for the library support (the `rustc_macro` crate) as well as the language features. There are a few minor differences from the implementation outlined in the RFC, such as the `rustc_macro` crate being available as a dylib and all symbols are `dlsym`'d directly instead of having a shim compiled. These should only affect the implementation, however, not the public interface. This commit also ended up touching a lot of code related to `#[derive]`, making a few notable changes: * Recognized derive attributes are no longer desugared to `derive_Foo`. Wasn't sure how to keep this behavior and *not* expose it to custom derive. * Derive attributes no longer have access to unstable features by default, they have to opt in on a granular level. * The `derive(Copy,Clone)` optimization is now done through another "obscure attribute" which is just intended to ferry along in the compiler that such an optimization is possible. The `derive(PartialEq,Eq)` optimization was also updated to do something similar. --- One part of this PR which needs to be improved before stabilizing are the errors and exact interfaces here. The error messages are relatively poor quality and there are surprising spects of this such as `#[derive(PartialEq, Eq, MyTrait)]` not working by default. The custom attributes added by the compiler end up becoming unstable again when going through a custom impl. Hopefully though this is enough to start allowing experimentation on crates.io! syntax-[breaking-change]
2016-08-22 17:07:11 -07:00
for symbol in self.info.exports[&crate_type].iter() {
debug!(" _{}", symbol);
writeln!(f, " {}", symbol)?;
}
Ok(())
})();
if let Err(e) = res {
self.sess.fatal(&format!("failed to write lib.def file: {}", e));
}
let mut arg = OsString::from("/DEF:");
arg.push(path);
self.cmd.arg(&arg);
}
fn subsystem(&mut self, subsystem: &str) {
// Note that previous passes of the compiler validated this subsystem,
// so we just blindly pass it to the linker.
self.cmd.arg(&format!("/SUBSYSTEM:{}", subsystem));
// Windows has two subsystems we're interested in right now, the console
// and windows subsystems. These both implicitly have different entry
// points (starting symbols). The console entry point starts with
// `mainCRTStartup` and the windows entry point starts with
// `WinMainCRTStartup`. These entry points, defined in system libraries,
// will then later probe for either `main` or `WinMain`, respectively to
// start the application.
//
// In Rust we just always generate a `main` function so we want control
// to always start there, so we force the entry point on the windows
// subsystem to be `mainCRTStartup` to get everything booted up
// correctly.
//
// For more information see RFC #1665
if subsystem == "windows" {
self.cmd.arg("/ENTRY:mainCRTStartup");
}
}
fn finalize(&mut self) -> Command {
let mut cmd = Command::new("");
::std::mem::swap(&mut cmd, &mut self.cmd);
cmd
}
// MSVC doesn't need group indicators
fn group_start(&mut self) {}
fn group_end(&mut self) {}
fn cross_lang_lto(&mut self) {
// Do nothing
}
}
pub struct EmLinker<'a> {
cmd: Command,
sess: &'a Session,
info: &'a LinkerInfo
}
impl<'a> Linker for EmLinker<'a> {
fn include_path(&mut self, path: &Path) {
self.cmd.arg("-L").arg(path);
}
fn link_staticlib(&mut self, lib: &str) {
self.cmd.arg("-l").arg(lib);
}
fn output_filename(&mut self, path: &Path) {
self.cmd.arg("-o").arg(path);
}
fn add_object(&mut self, path: &Path) {
self.cmd.arg(path);
}
fn link_dylib(&mut self, lib: &str) {
// Emscripten always links statically
self.link_staticlib(lib);
}
fn link_whole_staticlib(&mut self, lib: &str, _search_path: &[PathBuf]) {
// not supported?
self.link_staticlib(lib);
}
fn link_whole_rlib(&mut self, lib: &Path) {
// not supported?
self.link_rlib(lib);
}
fn link_rust_dylib(&mut self, lib: &str, _path: &Path) {
self.link_dylib(lib);
}
fn link_rlib(&mut self, lib: &Path) {
self.add_object(lib);
}
fn position_independent_executable(&mut self) {
// noop
}
fn no_position_independent_executable(&mut self) {
// noop
}
fn full_relro(&mut self) {
// noop
}
fn partial_relro(&mut self) {
// noop
}
fn no_relro(&mut self) {
// noop
}
fn args(&mut self, args: &[String]) {
self.cmd.args(args);
}
fn framework_path(&mut self, _path: &Path) {
bug!("frameworks are not supported on Emscripten")
}
fn link_framework(&mut self, _framework: &str) {
bug!("frameworks are not supported on Emscripten")
}
fn gc_sections(&mut self, _keep_metadata: bool) {
// noop
}
fn optimize(&mut self) {
// Emscripten performs own optimizations
self.cmd.arg(match self.sess.opts.optimize {
OptLevel::No => "-O0",
OptLevel::Less => "-O1",
OptLevel::Default => "-O2",
OptLevel::Aggressive => "-O3",
OptLevel::Size => "-Os",
OptLevel::SizeMin => "-Oz"
});
// Unusable until https://github.com/rust-lang/rust/issues/38454 is resolved
self.cmd.args(&["--memory-init-file", "0"]);
}
fn pgo_gen(&mut self) {
// noop, but maybe we need something like the gnu linker?
}
fn debuginfo(&mut self) {
// Preserve names or generate source maps depending on debug info
self.cmd.arg(match self.sess.opts.debuginfo {
DebugInfoLevel::NoDebugInfo => "-g0",
DebugInfoLevel::LimitedDebugInfo => "-g3",
DebugInfoLevel::FullDebugInfo => "-g4"
});
}
fn no_default_libraries(&mut self) {
self.cmd.args(&["-s", "DEFAULT_LIBRARY_FUNCS_TO_INCLUDE=[]"]);
}
fn build_dylib(&mut self, _out_filename: &Path) {
bug!("building dynamic library is unsupported on Emscripten")
}
fn build_static_executable(&mut self) {
// noop
}
fn export_symbols(&mut self, _tmpdir: &Path, crate_type: CrateType) {
let symbols = &self.info.exports[&crate_type];
debug!("EXPORTED SYMBOLS:");
self.cmd.arg("-s");
let mut arg = OsString::from("EXPORTED_FUNCTIONS=");
let mut encoded = String::new();
{
let mut encoder = json::Encoder::new(&mut encoded);
let res = encoder.emit_seq(symbols.len(), |encoder| {
for (i, sym) in symbols.iter().enumerate() {
encoder.emit_seq_elt(i, |encoder| {
encoder.emit_str(&("_".to_string() + sym))
})?;
}
Ok(())
});
if let Err(e) = res {
self.sess.fatal(&format!("failed to encode exported symbols: {}", e));
}
}
debug!("{}", encoded);
arg.push(encoded);
self.cmd.arg(arg);
}
fn subsystem(&mut self, _subsystem: &str) {
// noop
}
fn finalize(&mut self) -> Command {
let mut cmd = Command::new("");
::std::mem::swap(&mut cmd, &mut self.cmd);
cmd
}
// Appears not necessary on Emscripten
fn group_start(&mut self) {}
fn group_end(&mut self) {}
fn cross_lang_lto(&mut self) {
// Do nothing
}
}
fn exported_symbols(tcx: TyCtxt, crate_type: CrateType) -> Vec<String> {
let mut symbols = Vec::new();
let export_threshold = symbol_export::crates_export_threshold(&[crate_type]);
2018-02-27 17:52:07 +01:00
for &(symbol, level) in tcx.exported_symbols(LOCAL_CRATE).iter() {
if level.is_below_threshold(export_threshold) {
2018-02-27 17:52:07 +01:00
symbols.push(symbol.symbol_name(tcx).to_string());
}
}
let formats = tcx.sess.dependency_formats.borrow();
let deps = formats[&crate_type].iter();
for (index, dep_format) in deps.enumerate() {
let cnum = CrateNum::new(index + 1);
// For each dependency that we are linking to statically ...
if *dep_format == Linkage::Static {
// ... we add its symbol list to our export list.
2018-02-27 17:52:07 +01:00
for &(symbol, level) in tcx.exported_symbols(cnum).iter() {
if level.is_below_threshold(export_threshold) {
2018-02-27 17:52:07 +01:00
symbols.push(symbol.symbol_name(tcx).to_string());
}
}
}
}
symbols
}
rust: Import LLD for linking wasm objects This commit imports the LLD project from LLVM to serve as the default linker for the `wasm32-unknown-unknown` target. The `binaryen` submoule is consequently removed along with "binaryen linker" support in rustc. Moving to LLD brings with it a number of benefits for wasm code: * LLD is itself an actual linker, so there's no need to compile all wasm code with LTO any more. As a result builds should be *much* speedier as LTO is no longer forcibly enabled for all builds of the wasm target. * LLD is quickly becoming an "official solution" for linking wasm code together. This, I believe at least, is intended to be the main supported linker for native code and wasm moving forward. Picking up support early on should help ensure that we can help LLD identify bugs and otherwise prove that it works great for all our use cases! * Improvements to the wasm toolchain are currently primarily focused around LLVM and LLD (from what I can tell at least), so it's in general much better to be on this bandwagon for bugfixes and new features. * Historical "hacks" like `wasm-gc` will soon no longer be necessary, LLD will [natively implement][gc] `--gc-sections` (better than `wasm-gc`!) which means a postprocessor is no longer needed to show off Rust's "small wasm binary size". LLD is added in a pretty standard way to rustc right now. A new rustbuild target was defined for building LLD, and this is executed when a compiler's sysroot is being assembled. LLD is compiled against the LLVM that we've got in tree, which means we're currently on the `release_60` branch, but this may get upgraded in the near future! LLD is placed into rustc's sysroot in a `bin` directory. This is similar to where `gcc.exe` can be found on Windows. This directory is automatically added to `PATH` whenever rustc executes the linker, allowing us to define a `WasmLd` linker which implements the interface that `wasm-ld`, LLD's frontend, expects. Like Emscripten the LLD target is currently only enabled for Tier 1 platforms, notably OSX/Windows/Linux, and will need to be installed manually for compiling to wasm on other platforms. LLD is by default turned off in rustbuild, and requires a `config.toml` option to be enabled to turn it on. Finally the unstable `#![wasm_import_memory]` attribute was also removed as LLD has a native option for controlling this. [gc]: https://reviews.llvm.org/D42511
2017-08-26 18:30:12 -07:00
Upgrade to LLVM's master branch (LLVM 7) This commit upgrades the main LLVM submodule to LLVM's current master branch. The LLD submodule is updated in tandem as well as compiler-builtins. Along the way support was also added for LLVM 7's new features. This primarily includes the support for custom section concatenation natively in LLD so we now add wasm custom sections in LLVM IR rather than having custom support in rustc itself for doing so. Some other miscellaneous changes are: * We now pass `--gc-sections` to `wasm-ld` * The optimization level is now passed to `wasm-ld` * A `--stack-first` option is passed to LLD to have stack overflow always cause a trap instead of corrupting static data * The wasm target for LLVM switched to `wasm32-unknown-unknown`. * The syntax for aligned pointers has changed in LLVM IR and tests are updated to reflect this. * The `thumbv6m-none-eabi` target is disabled due to an [LLVM bug][llbug] Nowadays we've been mostly only upgrading whenever there's a major release of LLVM but enough changes have been happening on the wasm target that there's been growing motivation for quite some time now to upgrade out version of LLD. To upgrade LLD, however, we need to upgrade LLVM to avoid needing to build yet another version of LLVM on the builders. The revision of LLVM in use here is arbitrarily chosen. We will likely need to continue to update it over time if and when we discover bugs. Once LLVM 7 is fully released we can switch to that channel as well. [llbug]: https://bugs.llvm.org/show_bug.cgi?id=37382
2018-06-01 10:20:00 -07:00
pub struct WasmLd<'a> {
rust: Import LLD for linking wasm objects This commit imports the LLD project from LLVM to serve as the default linker for the `wasm32-unknown-unknown` target. The `binaryen` submoule is consequently removed along with "binaryen linker" support in rustc. Moving to LLD brings with it a number of benefits for wasm code: * LLD is itself an actual linker, so there's no need to compile all wasm code with LTO any more. As a result builds should be *much* speedier as LTO is no longer forcibly enabled for all builds of the wasm target. * LLD is quickly becoming an "official solution" for linking wasm code together. This, I believe at least, is intended to be the main supported linker for native code and wasm moving forward. Picking up support early on should help ensure that we can help LLD identify bugs and otherwise prove that it works great for all our use cases! * Improvements to the wasm toolchain are currently primarily focused around LLVM and LLD (from what I can tell at least), so it's in general much better to be on this bandwagon for bugfixes and new features. * Historical "hacks" like `wasm-gc` will soon no longer be necessary, LLD will [natively implement][gc] `--gc-sections` (better than `wasm-gc`!) which means a postprocessor is no longer needed to show off Rust's "small wasm binary size". LLD is added in a pretty standard way to rustc right now. A new rustbuild target was defined for building LLD, and this is executed when a compiler's sysroot is being assembled. LLD is compiled against the LLVM that we've got in tree, which means we're currently on the `release_60` branch, but this may get upgraded in the near future! LLD is placed into rustc's sysroot in a `bin` directory. This is similar to where `gcc.exe` can be found on Windows. This directory is automatically added to `PATH` whenever rustc executes the linker, allowing us to define a `WasmLd` linker which implements the interface that `wasm-ld`, LLD's frontend, expects. Like Emscripten the LLD target is currently only enabled for Tier 1 platforms, notably OSX/Windows/Linux, and will need to be installed manually for compiling to wasm on other platforms. LLD is by default turned off in rustbuild, and requires a `config.toml` option to be enabled to turn it on. Finally the unstable `#![wasm_import_memory]` attribute was also removed as LLD has a native option for controlling this. [gc]: https://reviews.llvm.org/D42511
2017-08-26 18:30:12 -07:00
cmd: Command,
Upgrade to LLVM's master branch (LLVM 7) This commit upgrades the main LLVM submodule to LLVM's current master branch. The LLD submodule is updated in tandem as well as compiler-builtins. Along the way support was also added for LLVM 7's new features. This primarily includes the support for custom section concatenation natively in LLD so we now add wasm custom sections in LLVM IR rather than having custom support in rustc itself for doing so. Some other miscellaneous changes are: * We now pass `--gc-sections` to `wasm-ld` * The optimization level is now passed to `wasm-ld` * A `--stack-first` option is passed to LLD to have stack overflow always cause a trap instead of corrupting static data * The wasm target for LLVM switched to `wasm32-unknown-unknown`. * The syntax for aligned pointers has changed in LLVM IR and tests are updated to reflect this. * The `thumbv6m-none-eabi` target is disabled due to an [LLVM bug][llbug] Nowadays we've been mostly only upgrading whenever there's a major release of LLVM but enough changes have been happening on the wasm target that there's been growing motivation for quite some time now to upgrade out version of LLD. To upgrade LLD, however, we need to upgrade LLVM to avoid needing to build yet another version of LLVM on the builders. The revision of LLVM in use here is arbitrarily chosen. We will likely need to continue to update it over time if and when we discover bugs. Once LLVM 7 is fully released we can switch to that channel as well. [llbug]: https://bugs.llvm.org/show_bug.cgi?id=37382
2018-06-01 10:20:00 -07:00
sess: &'a Session,
rust: Import LLD for linking wasm objects This commit imports the LLD project from LLVM to serve as the default linker for the `wasm32-unknown-unknown` target. The `binaryen` submoule is consequently removed along with "binaryen linker" support in rustc. Moving to LLD brings with it a number of benefits for wasm code: * LLD is itself an actual linker, so there's no need to compile all wasm code with LTO any more. As a result builds should be *much* speedier as LTO is no longer forcibly enabled for all builds of the wasm target. * LLD is quickly becoming an "official solution" for linking wasm code together. This, I believe at least, is intended to be the main supported linker for native code and wasm moving forward. Picking up support early on should help ensure that we can help LLD identify bugs and otherwise prove that it works great for all our use cases! * Improvements to the wasm toolchain are currently primarily focused around LLVM and LLD (from what I can tell at least), so it's in general much better to be on this bandwagon for bugfixes and new features. * Historical "hacks" like `wasm-gc` will soon no longer be necessary, LLD will [natively implement][gc] `--gc-sections` (better than `wasm-gc`!) which means a postprocessor is no longer needed to show off Rust's "small wasm binary size". LLD is added in a pretty standard way to rustc right now. A new rustbuild target was defined for building LLD, and this is executed when a compiler's sysroot is being assembled. LLD is compiled against the LLVM that we've got in tree, which means we're currently on the `release_60` branch, but this may get upgraded in the near future! LLD is placed into rustc's sysroot in a `bin` directory. This is similar to where `gcc.exe` can be found on Windows. This directory is automatically added to `PATH` whenever rustc executes the linker, allowing us to define a `WasmLd` linker which implements the interface that `wasm-ld`, LLD's frontend, expects. Like Emscripten the LLD target is currently only enabled for Tier 1 platforms, notably OSX/Windows/Linux, and will need to be installed manually for compiling to wasm on other platforms. LLD is by default turned off in rustbuild, and requires a `config.toml` option to be enabled to turn it on. Finally the unstable `#![wasm_import_memory]` attribute was also removed as LLD has a native option for controlling this. [gc]: https://reviews.llvm.org/D42511
2017-08-26 18:30:12 -07:00
}
Upgrade to LLVM's master branch (LLVM 7) This commit upgrades the main LLVM submodule to LLVM's current master branch. The LLD submodule is updated in tandem as well as compiler-builtins. Along the way support was also added for LLVM 7's new features. This primarily includes the support for custom section concatenation natively in LLD so we now add wasm custom sections in LLVM IR rather than having custom support in rustc itself for doing so. Some other miscellaneous changes are: * We now pass `--gc-sections` to `wasm-ld` * The optimization level is now passed to `wasm-ld` * A `--stack-first` option is passed to LLD to have stack overflow always cause a trap instead of corrupting static data * The wasm target for LLVM switched to `wasm32-unknown-unknown`. * The syntax for aligned pointers has changed in LLVM IR and tests are updated to reflect this. * The `thumbv6m-none-eabi` target is disabled due to an [LLVM bug][llbug] Nowadays we've been mostly only upgrading whenever there's a major release of LLVM but enough changes have been happening on the wasm target that there's been growing motivation for quite some time now to upgrade out version of LLD. To upgrade LLD, however, we need to upgrade LLVM to avoid needing to build yet another version of LLVM on the builders. The revision of LLVM in use here is arbitrarily chosen. We will likely need to continue to update it over time if and when we discover bugs. Once LLVM 7 is fully released we can switch to that channel as well. [llbug]: https://bugs.llvm.org/show_bug.cgi?id=37382
2018-06-01 10:20:00 -07:00
impl<'a> Linker for WasmLd<'a> {
rust: Import LLD for linking wasm objects This commit imports the LLD project from LLVM to serve as the default linker for the `wasm32-unknown-unknown` target. The `binaryen` submoule is consequently removed along with "binaryen linker" support in rustc. Moving to LLD brings with it a number of benefits for wasm code: * LLD is itself an actual linker, so there's no need to compile all wasm code with LTO any more. As a result builds should be *much* speedier as LTO is no longer forcibly enabled for all builds of the wasm target. * LLD is quickly becoming an "official solution" for linking wasm code together. This, I believe at least, is intended to be the main supported linker for native code and wasm moving forward. Picking up support early on should help ensure that we can help LLD identify bugs and otherwise prove that it works great for all our use cases! * Improvements to the wasm toolchain are currently primarily focused around LLVM and LLD (from what I can tell at least), so it's in general much better to be on this bandwagon for bugfixes and new features. * Historical "hacks" like `wasm-gc` will soon no longer be necessary, LLD will [natively implement][gc] `--gc-sections` (better than `wasm-gc`!) which means a postprocessor is no longer needed to show off Rust's "small wasm binary size". LLD is added in a pretty standard way to rustc right now. A new rustbuild target was defined for building LLD, and this is executed when a compiler's sysroot is being assembled. LLD is compiled against the LLVM that we've got in tree, which means we're currently on the `release_60` branch, but this may get upgraded in the near future! LLD is placed into rustc's sysroot in a `bin` directory. This is similar to where `gcc.exe` can be found on Windows. This directory is automatically added to `PATH` whenever rustc executes the linker, allowing us to define a `WasmLd` linker which implements the interface that `wasm-ld`, LLD's frontend, expects. Like Emscripten the LLD target is currently only enabled for Tier 1 platforms, notably OSX/Windows/Linux, and will need to be installed manually for compiling to wasm on other platforms. LLD is by default turned off in rustbuild, and requires a `config.toml` option to be enabled to turn it on. Finally the unstable `#![wasm_import_memory]` attribute was also removed as LLD has a native option for controlling this. [gc]: https://reviews.llvm.org/D42511
2017-08-26 18:30:12 -07:00
fn link_dylib(&mut self, lib: &str) {
self.cmd.arg("-l").arg(lib);
}
fn link_staticlib(&mut self, lib: &str) {
self.cmd.arg("-l").arg(lib);
}
fn link_rlib(&mut self, lib: &Path) {
self.cmd.arg(lib);
}
fn include_path(&mut self, path: &Path) {
self.cmd.arg("-L").arg(path);
}
fn framework_path(&mut self, _path: &Path) {
panic!("frameworks not supported")
}
fn output_filename(&mut self, path: &Path) {
self.cmd.arg("-o").arg(path);
}
fn add_object(&mut self, path: &Path) {
self.cmd.arg(path);
}
fn position_independent_executable(&mut self) {
}
fn full_relro(&mut self) {
}
rust: Import LLD for linking wasm objects This commit imports the LLD project from LLVM to serve as the default linker for the `wasm32-unknown-unknown` target. The `binaryen` submoule is consequently removed along with "binaryen linker" support in rustc. Moving to LLD brings with it a number of benefits for wasm code: * LLD is itself an actual linker, so there's no need to compile all wasm code with LTO any more. As a result builds should be *much* speedier as LTO is no longer forcibly enabled for all builds of the wasm target. * LLD is quickly becoming an "official solution" for linking wasm code together. This, I believe at least, is intended to be the main supported linker for native code and wasm moving forward. Picking up support early on should help ensure that we can help LLD identify bugs and otherwise prove that it works great for all our use cases! * Improvements to the wasm toolchain are currently primarily focused around LLVM and LLD (from what I can tell at least), so it's in general much better to be on this bandwagon for bugfixes and new features. * Historical "hacks" like `wasm-gc` will soon no longer be necessary, LLD will [natively implement][gc] `--gc-sections` (better than `wasm-gc`!) which means a postprocessor is no longer needed to show off Rust's "small wasm binary size". LLD is added in a pretty standard way to rustc right now. A new rustbuild target was defined for building LLD, and this is executed when a compiler's sysroot is being assembled. LLD is compiled against the LLVM that we've got in tree, which means we're currently on the `release_60` branch, but this may get upgraded in the near future! LLD is placed into rustc's sysroot in a `bin` directory. This is similar to where `gcc.exe` can be found on Windows. This directory is automatically added to `PATH` whenever rustc executes the linker, allowing us to define a `WasmLd` linker which implements the interface that `wasm-ld`, LLD's frontend, expects. Like Emscripten the LLD target is currently only enabled for Tier 1 platforms, notably OSX/Windows/Linux, and will need to be installed manually for compiling to wasm on other platforms. LLD is by default turned off in rustbuild, and requires a `config.toml` option to be enabled to turn it on. Finally the unstable `#![wasm_import_memory]` attribute was also removed as LLD has a native option for controlling this. [gc]: https://reviews.llvm.org/D42511
2017-08-26 18:30:12 -07:00
fn partial_relro(&mut self) {
}
fn no_relro(&mut self) {
rust: Import LLD for linking wasm objects This commit imports the LLD project from LLVM to serve as the default linker for the `wasm32-unknown-unknown` target. The `binaryen` submoule is consequently removed along with "binaryen linker" support in rustc. Moving to LLD brings with it a number of benefits for wasm code: * LLD is itself an actual linker, so there's no need to compile all wasm code with LTO any more. As a result builds should be *much* speedier as LTO is no longer forcibly enabled for all builds of the wasm target. * LLD is quickly becoming an "official solution" for linking wasm code together. This, I believe at least, is intended to be the main supported linker for native code and wasm moving forward. Picking up support early on should help ensure that we can help LLD identify bugs and otherwise prove that it works great for all our use cases! * Improvements to the wasm toolchain are currently primarily focused around LLVM and LLD (from what I can tell at least), so it's in general much better to be on this bandwagon for bugfixes and new features. * Historical "hacks" like `wasm-gc` will soon no longer be necessary, LLD will [natively implement][gc] `--gc-sections` (better than `wasm-gc`!) which means a postprocessor is no longer needed to show off Rust's "small wasm binary size". LLD is added in a pretty standard way to rustc right now. A new rustbuild target was defined for building LLD, and this is executed when a compiler's sysroot is being assembled. LLD is compiled against the LLVM that we've got in tree, which means we're currently on the `release_60` branch, but this may get upgraded in the near future! LLD is placed into rustc's sysroot in a `bin` directory. This is similar to where `gcc.exe` can be found on Windows. This directory is automatically added to `PATH` whenever rustc executes the linker, allowing us to define a `WasmLd` linker which implements the interface that `wasm-ld`, LLD's frontend, expects. Like Emscripten the LLD target is currently only enabled for Tier 1 platforms, notably OSX/Windows/Linux, and will need to be installed manually for compiling to wasm on other platforms. LLD is by default turned off in rustbuild, and requires a `config.toml` option to be enabled to turn it on. Finally the unstable `#![wasm_import_memory]` attribute was also removed as LLD has a native option for controlling this. [gc]: https://reviews.llvm.org/D42511
2017-08-26 18:30:12 -07:00
}
fn build_static_executable(&mut self) {
}
fn args(&mut self, args: &[String]) {
self.cmd.args(args);
}
fn link_rust_dylib(&mut self, lib: &str, _path: &Path) {
self.cmd.arg("-l").arg(lib);
}
fn link_framework(&mut self, _framework: &str) {
panic!("frameworks not supported")
}
fn link_whole_staticlib(&mut self, lib: &str, _search_path: &[PathBuf]) {
self.cmd.arg("-l").arg(lib);
}
fn link_whole_rlib(&mut self, lib: &Path) {
self.cmd.arg(lib);
}
fn gc_sections(&mut self, _keep_metadata: bool) {
Upgrade to LLVM's master branch (LLVM 7) This commit upgrades the main LLVM submodule to LLVM's current master branch. The LLD submodule is updated in tandem as well as compiler-builtins. Along the way support was also added for LLVM 7's new features. This primarily includes the support for custom section concatenation natively in LLD so we now add wasm custom sections in LLVM IR rather than having custom support in rustc itself for doing so. Some other miscellaneous changes are: * We now pass `--gc-sections` to `wasm-ld` * The optimization level is now passed to `wasm-ld` * A `--stack-first` option is passed to LLD to have stack overflow always cause a trap instead of corrupting static data * The wasm target for LLVM switched to `wasm32-unknown-unknown`. * The syntax for aligned pointers has changed in LLVM IR and tests are updated to reflect this. * The `thumbv6m-none-eabi` target is disabled due to an [LLVM bug][llbug] Nowadays we've been mostly only upgrading whenever there's a major release of LLVM but enough changes have been happening on the wasm target that there's been growing motivation for quite some time now to upgrade out version of LLD. To upgrade LLD, however, we need to upgrade LLVM to avoid needing to build yet another version of LLVM on the builders. The revision of LLVM in use here is arbitrarily chosen. We will likely need to continue to update it over time if and when we discover bugs. Once LLVM 7 is fully released we can switch to that channel as well. [llbug]: https://bugs.llvm.org/show_bug.cgi?id=37382
2018-06-01 10:20:00 -07:00
self.cmd.arg("--gc-sections");
rust: Import LLD for linking wasm objects This commit imports the LLD project from LLVM to serve as the default linker for the `wasm32-unknown-unknown` target. The `binaryen` submoule is consequently removed along with "binaryen linker" support in rustc. Moving to LLD brings with it a number of benefits for wasm code: * LLD is itself an actual linker, so there's no need to compile all wasm code with LTO any more. As a result builds should be *much* speedier as LTO is no longer forcibly enabled for all builds of the wasm target. * LLD is quickly becoming an "official solution" for linking wasm code together. This, I believe at least, is intended to be the main supported linker for native code and wasm moving forward. Picking up support early on should help ensure that we can help LLD identify bugs and otherwise prove that it works great for all our use cases! * Improvements to the wasm toolchain are currently primarily focused around LLVM and LLD (from what I can tell at least), so it's in general much better to be on this bandwagon for bugfixes and new features. * Historical "hacks" like `wasm-gc` will soon no longer be necessary, LLD will [natively implement][gc] `--gc-sections` (better than `wasm-gc`!) which means a postprocessor is no longer needed to show off Rust's "small wasm binary size". LLD is added in a pretty standard way to rustc right now. A new rustbuild target was defined for building LLD, and this is executed when a compiler's sysroot is being assembled. LLD is compiled against the LLVM that we've got in tree, which means we're currently on the `release_60` branch, but this may get upgraded in the near future! LLD is placed into rustc's sysroot in a `bin` directory. This is similar to where `gcc.exe` can be found on Windows. This directory is automatically added to `PATH` whenever rustc executes the linker, allowing us to define a `WasmLd` linker which implements the interface that `wasm-ld`, LLD's frontend, expects. Like Emscripten the LLD target is currently only enabled for Tier 1 platforms, notably OSX/Windows/Linux, and will need to be installed manually for compiling to wasm on other platforms. LLD is by default turned off in rustbuild, and requires a `config.toml` option to be enabled to turn it on. Finally the unstable `#![wasm_import_memory]` attribute was also removed as LLD has a native option for controlling this. [gc]: https://reviews.llvm.org/D42511
2017-08-26 18:30:12 -07:00
}
fn optimize(&mut self) {
Upgrade to LLVM's master branch (LLVM 7) This commit upgrades the main LLVM submodule to LLVM's current master branch. The LLD submodule is updated in tandem as well as compiler-builtins. Along the way support was also added for LLVM 7's new features. This primarily includes the support for custom section concatenation natively in LLD so we now add wasm custom sections in LLVM IR rather than having custom support in rustc itself for doing so. Some other miscellaneous changes are: * We now pass `--gc-sections` to `wasm-ld` * The optimization level is now passed to `wasm-ld` * A `--stack-first` option is passed to LLD to have stack overflow always cause a trap instead of corrupting static data * The wasm target for LLVM switched to `wasm32-unknown-unknown`. * The syntax for aligned pointers has changed in LLVM IR and tests are updated to reflect this. * The `thumbv6m-none-eabi` target is disabled due to an [LLVM bug][llbug] Nowadays we've been mostly only upgrading whenever there's a major release of LLVM but enough changes have been happening on the wasm target that there's been growing motivation for quite some time now to upgrade out version of LLD. To upgrade LLD, however, we need to upgrade LLVM to avoid needing to build yet another version of LLVM on the builders. The revision of LLVM in use here is arbitrarily chosen. We will likely need to continue to update it over time if and when we discover bugs. Once LLVM 7 is fully released we can switch to that channel as well. [llbug]: https://bugs.llvm.org/show_bug.cgi?id=37382
2018-06-01 10:20:00 -07:00
self.cmd.arg(match self.sess.opts.optimize {
OptLevel::No => "-O0",
OptLevel::Less => "-O1",
OptLevel::Default => "-O2",
OptLevel::Aggressive => "-O3",
// Currently LLD doesn't support `Os` and `Oz`, so pass through `O2`
// instead.
OptLevel::Size => "-O2",
OptLevel::SizeMin => "-O2"
});
rust: Import LLD for linking wasm objects This commit imports the LLD project from LLVM to serve as the default linker for the `wasm32-unknown-unknown` target. The `binaryen` submoule is consequently removed along with "binaryen linker" support in rustc. Moving to LLD brings with it a number of benefits for wasm code: * LLD is itself an actual linker, so there's no need to compile all wasm code with LTO any more. As a result builds should be *much* speedier as LTO is no longer forcibly enabled for all builds of the wasm target. * LLD is quickly becoming an "official solution" for linking wasm code together. This, I believe at least, is intended to be the main supported linker for native code and wasm moving forward. Picking up support early on should help ensure that we can help LLD identify bugs and otherwise prove that it works great for all our use cases! * Improvements to the wasm toolchain are currently primarily focused around LLVM and LLD (from what I can tell at least), so it's in general much better to be on this bandwagon for bugfixes and new features. * Historical "hacks" like `wasm-gc` will soon no longer be necessary, LLD will [natively implement][gc] `--gc-sections` (better than `wasm-gc`!) which means a postprocessor is no longer needed to show off Rust's "small wasm binary size". LLD is added in a pretty standard way to rustc right now. A new rustbuild target was defined for building LLD, and this is executed when a compiler's sysroot is being assembled. LLD is compiled against the LLVM that we've got in tree, which means we're currently on the `release_60` branch, but this may get upgraded in the near future! LLD is placed into rustc's sysroot in a `bin` directory. This is similar to where `gcc.exe` can be found on Windows. This directory is automatically added to `PATH` whenever rustc executes the linker, allowing us to define a `WasmLd` linker which implements the interface that `wasm-ld`, LLD's frontend, expects. Like Emscripten the LLD target is currently only enabled for Tier 1 platforms, notably OSX/Windows/Linux, and will need to be installed manually for compiling to wasm on other platforms. LLD is by default turned off in rustbuild, and requires a `config.toml` option to be enabled to turn it on. Finally the unstable `#![wasm_import_memory]` attribute was also removed as LLD has a native option for controlling this. [gc]: https://reviews.llvm.org/D42511
2017-08-26 18:30:12 -07:00
}
fn pgo_gen(&mut self) {
}
rust: Import LLD for linking wasm objects This commit imports the LLD project from LLVM to serve as the default linker for the `wasm32-unknown-unknown` target. The `binaryen` submoule is consequently removed along with "binaryen linker" support in rustc. Moving to LLD brings with it a number of benefits for wasm code: * LLD is itself an actual linker, so there's no need to compile all wasm code with LTO any more. As a result builds should be *much* speedier as LTO is no longer forcibly enabled for all builds of the wasm target. * LLD is quickly becoming an "official solution" for linking wasm code together. This, I believe at least, is intended to be the main supported linker for native code and wasm moving forward. Picking up support early on should help ensure that we can help LLD identify bugs and otherwise prove that it works great for all our use cases! * Improvements to the wasm toolchain are currently primarily focused around LLVM and LLD (from what I can tell at least), so it's in general much better to be on this bandwagon for bugfixes and new features. * Historical "hacks" like `wasm-gc` will soon no longer be necessary, LLD will [natively implement][gc] `--gc-sections` (better than `wasm-gc`!) which means a postprocessor is no longer needed to show off Rust's "small wasm binary size". LLD is added in a pretty standard way to rustc right now. A new rustbuild target was defined for building LLD, and this is executed when a compiler's sysroot is being assembled. LLD is compiled against the LLVM that we've got in tree, which means we're currently on the `release_60` branch, but this may get upgraded in the near future! LLD is placed into rustc's sysroot in a `bin` directory. This is similar to where `gcc.exe` can be found on Windows. This directory is automatically added to `PATH` whenever rustc executes the linker, allowing us to define a `WasmLd` linker which implements the interface that `wasm-ld`, LLD's frontend, expects. Like Emscripten the LLD target is currently only enabled for Tier 1 platforms, notably OSX/Windows/Linux, and will need to be installed manually for compiling to wasm on other platforms. LLD is by default turned off in rustbuild, and requires a `config.toml` option to be enabled to turn it on. Finally the unstable `#![wasm_import_memory]` attribute was also removed as LLD has a native option for controlling this. [gc]: https://reviews.llvm.org/D42511
2017-08-26 18:30:12 -07:00
fn debuginfo(&mut self) {
}
fn no_default_libraries(&mut self) {
}
fn build_dylib(&mut self, _out_filename: &Path) {
}
fn export_symbols(&mut self, _tmpdir: &Path, _crate_type: CrateType) {
}
fn subsystem(&mut self, _subsystem: &str) {
}
fn no_position_independent_executable(&mut self) {
}
fn finalize(&mut self) -> Command {
// There have been reports in the wild (rustwasm/wasm-bindgen#119) of
// using threads causing weird hangs and bugs. Disable it entirely as
// this isn't yet the bottleneck of compilation at all anyway.
self.cmd.arg("--no-threads");
Upgrade to LLVM's master branch (LLVM 7) This commit upgrades the main LLVM submodule to LLVM's current master branch. The LLD submodule is updated in tandem as well as compiler-builtins. Along the way support was also added for LLVM 7's new features. This primarily includes the support for custom section concatenation natively in LLD so we now add wasm custom sections in LLVM IR rather than having custom support in rustc itself for doing so. Some other miscellaneous changes are: * We now pass `--gc-sections` to `wasm-ld` * The optimization level is now passed to `wasm-ld` * A `--stack-first` option is passed to LLD to have stack overflow always cause a trap instead of corrupting static data * The wasm target for LLVM switched to `wasm32-unknown-unknown`. * The syntax for aligned pointers has changed in LLVM IR and tests are updated to reflect this. * The `thumbv6m-none-eabi` target is disabled due to an [LLVM bug][llbug] Nowadays we've been mostly only upgrading whenever there's a major release of LLVM but enough changes have been happening on the wasm target that there's been growing motivation for quite some time now to upgrade out version of LLD. To upgrade LLD, however, we need to upgrade LLVM to avoid needing to build yet another version of LLVM on the builders. The revision of LLVM in use here is arbitrarily chosen. We will likely need to continue to update it over time if and when we discover bugs. Once LLVM 7 is fully released we can switch to that channel as well. [llbug]: https://bugs.llvm.org/show_bug.cgi?id=37382
2018-06-01 10:20:00 -07:00
// By default LLD only gives us one page of stack (64k) which is a
// little small. Default to a larger stack closer to other PC platforms
// (1MB) and users can always inject their own link-args to override this.
self.cmd.arg("-z").arg("stack-size=1048576");
rust: Import LLD for linking wasm objects This commit imports the LLD project from LLVM to serve as the default linker for the `wasm32-unknown-unknown` target. The `binaryen` submoule is consequently removed along with "binaryen linker" support in rustc. Moving to LLD brings with it a number of benefits for wasm code: * LLD is itself an actual linker, so there's no need to compile all wasm code with LTO any more. As a result builds should be *much* speedier as LTO is no longer forcibly enabled for all builds of the wasm target. * LLD is quickly becoming an "official solution" for linking wasm code together. This, I believe at least, is intended to be the main supported linker for native code and wasm moving forward. Picking up support early on should help ensure that we can help LLD identify bugs and otherwise prove that it works great for all our use cases! * Improvements to the wasm toolchain are currently primarily focused around LLVM and LLD (from what I can tell at least), so it's in general much better to be on this bandwagon for bugfixes and new features. * Historical "hacks" like `wasm-gc` will soon no longer be necessary, LLD will [natively implement][gc] `--gc-sections` (better than `wasm-gc`!) which means a postprocessor is no longer needed to show off Rust's "small wasm binary size". LLD is added in a pretty standard way to rustc right now. A new rustbuild target was defined for building LLD, and this is executed when a compiler's sysroot is being assembled. LLD is compiled against the LLVM that we've got in tree, which means we're currently on the `release_60` branch, but this may get upgraded in the near future! LLD is placed into rustc's sysroot in a `bin` directory. This is similar to where `gcc.exe` can be found on Windows. This directory is automatically added to `PATH` whenever rustc executes the linker, allowing us to define a `WasmLd` linker which implements the interface that `wasm-ld`, LLD's frontend, expects. Like Emscripten the LLD target is currently only enabled for Tier 1 platforms, notably OSX/Windows/Linux, and will need to be installed manually for compiling to wasm on other platforms. LLD is by default turned off in rustbuild, and requires a `config.toml` option to be enabled to turn it on. Finally the unstable `#![wasm_import_memory]` attribute was also removed as LLD has a native option for controlling this. [gc]: https://reviews.llvm.org/D42511
2017-08-26 18:30:12 -07:00
Upgrade to LLVM's master branch (LLVM 7) This commit upgrades the main LLVM submodule to LLVM's current master branch. The LLD submodule is updated in tandem as well as compiler-builtins. Along the way support was also added for LLVM 7's new features. This primarily includes the support for custom section concatenation natively in LLD so we now add wasm custom sections in LLVM IR rather than having custom support in rustc itself for doing so. Some other miscellaneous changes are: * We now pass `--gc-sections` to `wasm-ld` * The optimization level is now passed to `wasm-ld` * A `--stack-first` option is passed to LLD to have stack overflow always cause a trap instead of corrupting static data * The wasm target for LLVM switched to `wasm32-unknown-unknown`. * The syntax for aligned pointers has changed in LLVM IR and tests are updated to reflect this. * The `thumbv6m-none-eabi` target is disabled due to an [LLVM bug][llbug] Nowadays we've been mostly only upgrading whenever there's a major release of LLVM but enough changes have been happening on the wasm target that there's been growing motivation for quite some time now to upgrade out version of LLD. To upgrade LLD, however, we need to upgrade LLVM to avoid needing to build yet another version of LLVM on the builders. The revision of LLVM in use here is arbitrarily chosen. We will likely need to continue to update it over time if and when we discover bugs. Once LLVM 7 is fully released we can switch to that channel as well. [llbug]: https://bugs.llvm.org/show_bug.cgi?id=37382
2018-06-01 10:20:00 -07:00
// By default LLD's memory layout is:
//
// 1. First, a blank page
// 2. Next, all static data
// 3. Finally, the main stack (which grows down)
//
// This has the unfortunate consequence that on stack overflows you
// corrupt static data and can cause some exceedingly weird bugs. To
// help detect this a little sooner we instead request that the stack is
// placed before static data.
//
// This means that we'll generate slightly larger binaries as references
// to static data will take more bytes in the ULEB128 encoding, but
// stack overflow will be guaranteed to trap as it underflows instead of
// corrupting static data.
self.cmd.arg("--stack-first");
rust: Import LLD for linking wasm objects This commit imports the LLD project from LLVM to serve as the default linker for the `wasm32-unknown-unknown` target. The `binaryen` submoule is consequently removed along with "binaryen linker" support in rustc. Moving to LLD brings with it a number of benefits for wasm code: * LLD is itself an actual linker, so there's no need to compile all wasm code with LTO any more. As a result builds should be *much* speedier as LTO is no longer forcibly enabled for all builds of the wasm target. * LLD is quickly becoming an "official solution" for linking wasm code together. This, I believe at least, is intended to be the main supported linker for native code and wasm moving forward. Picking up support early on should help ensure that we can help LLD identify bugs and otherwise prove that it works great for all our use cases! * Improvements to the wasm toolchain are currently primarily focused around LLVM and LLD (from what I can tell at least), so it's in general much better to be on this bandwagon for bugfixes and new features. * Historical "hacks" like `wasm-gc` will soon no longer be necessary, LLD will [natively implement][gc] `--gc-sections` (better than `wasm-gc`!) which means a postprocessor is no longer needed to show off Rust's "small wasm binary size". LLD is added in a pretty standard way to rustc right now. A new rustbuild target was defined for building LLD, and this is executed when a compiler's sysroot is being assembled. LLD is compiled against the LLVM that we've got in tree, which means we're currently on the `release_60` branch, but this may get upgraded in the near future! LLD is placed into rustc's sysroot in a `bin` directory. This is similar to where `gcc.exe` can be found on Windows. This directory is automatically added to `PATH` whenever rustc executes the linker, allowing us to define a `WasmLd` linker which implements the interface that `wasm-ld`, LLD's frontend, expects. Like Emscripten the LLD target is currently only enabled for Tier 1 platforms, notably OSX/Windows/Linux, and will need to be installed manually for compiling to wasm on other platforms. LLD is by default turned off in rustbuild, and requires a `config.toml` option to be enabled to turn it on. Finally the unstable `#![wasm_import_memory]` attribute was also removed as LLD has a native option for controlling this. [gc]: https://reviews.llvm.org/D42511
2017-08-26 18:30:12 -07:00
// FIXME we probably shouldn't pass this but instead pass an explicit
// whitelist of symbols we'll allow to be undefined. Unfortunately
// though we can't handle symbols like `log10` that LLVM injects at a
// super late date without actually parsing object files. For now let's
// stick to this and hopefully fix it before stabilization happens.
self.cmd.arg("--allow-undefined");
// For now we just never have an entry symbol
self.cmd.arg("--no-entry");
let mut cmd = Command::new("");
::std::mem::swap(&mut cmd, &mut self.cmd);
cmd
}
// Not needed for now with LLD
fn group_start(&mut self) {}
fn group_end(&mut self) {}
fn cross_lang_lto(&mut self) {
// Do nothing for now
}
rust: Import LLD for linking wasm objects This commit imports the LLD project from LLVM to serve as the default linker for the `wasm32-unknown-unknown` target. The `binaryen` submoule is consequently removed along with "binaryen linker" support in rustc. Moving to LLD brings with it a number of benefits for wasm code: * LLD is itself an actual linker, so there's no need to compile all wasm code with LTO any more. As a result builds should be *much* speedier as LTO is no longer forcibly enabled for all builds of the wasm target. * LLD is quickly becoming an "official solution" for linking wasm code together. This, I believe at least, is intended to be the main supported linker for native code and wasm moving forward. Picking up support early on should help ensure that we can help LLD identify bugs and otherwise prove that it works great for all our use cases! * Improvements to the wasm toolchain are currently primarily focused around LLVM and LLD (from what I can tell at least), so it's in general much better to be on this bandwagon for bugfixes and new features. * Historical "hacks" like `wasm-gc` will soon no longer be necessary, LLD will [natively implement][gc] `--gc-sections` (better than `wasm-gc`!) which means a postprocessor is no longer needed to show off Rust's "small wasm binary size". LLD is added in a pretty standard way to rustc right now. A new rustbuild target was defined for building LLD, and this is executed when a compiler's sysroot is being assembled. LLD is compiled against the LLVM that we've got in tree, which means we're currently on the `release_60` branch, but this may get upgraded in the near future! LLD is placed into rustc's sysroot in a `bin` directory. This is similar to where `gcc.exe` can be found on Windows. This directory is automatically added to `PATH` whenever rustc executes the linker, allowing us to define a `WasmLd` linker which implements the interface that `wasm-ld`, LLD's frontend, expects. Like Emscripten the LLD target is currently only enabled for Tier 1 platforms, notably OSX/Windows/Linux, and will need to be installed manually for compiling to wasm on other platforms. LLD is by default turned off in rustbuild, and requires a `config.toml` option to be enabled to turn it on. Finally the unstable `#![wasm_import_memory]` attribute was also removed as LLD has a native option for controlling this. [gc]: https://reviews.llvm.org/D42511
2017-08-26 18:30:12 -07:00
}