rust/src/librustc_codegen_llvm/attributes.rs

308 lines
11 KiB
Rust
Raw Normal View History

// Copyright 2012-2015 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.
//! Set and unset common attributes on LLVM values.
use std::ffi::CString;
use rustc::hir::{CodegenFnAttrFlags, CodegenFnAttrs};
use rustc::hir::def_id::{DefId, LOCAL_CRATE};
use rustc::session::Session;
use rustc::session::config::Sanitizer;
use rustc::ty::TyCtxt;
2018-06-13 13:44:43 +00:00
use rustc::ty::query::Providers;
2018-02-27 16:11:14 +00:00
use rustc_data_structures::sync::Lrc;
rustc: Add a `#[wasm_import_module]` attribute This commit adds a new attribute to the Rust compiler specific to the wasm target (and no other targets). The `#[wasm_import_module]` attribute is used to specify the module that a name is imported from, and is used like so: #[wasm_import_module = "./foo.js"] extern { fn some_js_function(); } Here the import of the symbol `some_js_function` is tagged with the `./foo.js` module in the wasm output file. Wasm-the-format includes two fields on all imports, a module and a field. The field is the symbol name (`some_js_function` above) and the module has historically unconditionally been `"env"`. I'm not sure if this `"env"` convention has asm.js or LLVM roots, but regardless we'd like the ability to configure it! The proposed ES module integration with wasm (aka a wasm module is "just another ES module") requires that the import module of wasm imports is interpreted as an ES module import, meaning that you'll need to encode paths, NPM packages, etc. As a result, we'll need this to be something other than `"env"`! Unfortunately neither our version of LLVM nor LLD supports custom import modules (aka anything not `"env"`). My hope is that by the time LLVM 7 is released both will have support, but in the meantime this commit adds some primitive encoding/decoding of wasm files to the compiler. This way rustc postprocesses the wasm module that LLVM emits to ensure it's got all the imports we'd like to have in it. Eventually I'd ideally like to unconditionally require this attribute to be placed on all `extern { ... }` blocks. For now though it seemed prudent to add it as an unstable attribute, so for now it's not required (as that'd force usage of a feature gate). Hopefully it doesn't take too long to "stabilize" this! cc rust-lang-nursery/rust-wasm#29
2018-02-10 22:28:17 +00:00
use rustc_data_structures::fx::FxHashMap;
use rustc_target::spec::PanicStrategy;
use attributes;
use llvm::{self, Attribute};
2016-08-02 21:25:19 +00:00
use llvm::AttributePlace::Function;
use llvm_util;
pub use syntax::attr::{self, InlineAttr};
use context::CodegenCx;
use value::Value;
/// Mark LLVM function to use provided inline heuristic.
#[inline]
pub fn inline(val: &'ll Value, inline: InlineAttr) {
use self::InlineAttr::*;
match inline {
2016-08-02 21:25:19 +00:00
Hint => Attribute::InlineHint.apply_llfn(Function, val),
Always => Attribute::AlwaysInline.apply_llfn(Function, val),
Never => Attribute::NoInline.apply_llfn(Function, val),
None => {
Attribute::InlineHint.unapply_llfn(Function, val);
Attribute::AlwaysInline.unapply_llfn(Function, val);
Attribute::NoInline.unapply_llfn(Function, val);
},
};
}
/// Tell LLVM to emit or not emit the information necessary to unwind the stack for the function.
#[inline]
pub fn emit_uwtable(val: &'ll Value, emit: bool) {
2016-08-02 21:25:19 +00:00
Attribute::UWTable.toggle_llfn(Function, val, emit);
}
/// Tell LLVM whether the function can or cannot unwind.
#[inline]
pub fn unwind(val: &'ll Value, can_unwind: bool) {
2016-08-02 21:25:19 +00:00
Attribute::NoUnwind.toggle_llfn(Function, val, !can_unwind);
}
2017-08-15 19:45:21 +00:00
/// Tell LLVM whether it should optimize function for size.
#[inline]
#[allow(dead_code)] // possibly useful function
pub fn set_optimize_for_size(val: &'ll Value, optimize: bool) {
2016-08-02 21:25:19 +00:00
Attribute::OptimizeForSize.toggle_llfn(Function, val, optimize);
}
2016-03-21 20:01:08 +00:00
/// Tell LLVM if this function should be 'naked', i.e. skip the epilogue and prologue.
#[inline]
pub fn naked(val: &'ll Value, is_naked: bool) {
2016-08-02 21:25:19 +00:00
Attribute::Naked.toggle_llfn(Function, val, is_naked);
2016-03-21 20:01:08 +00:00
}
pub fn set_frame_pointer_elimination(cx: &CodegenCx<'ll, '_>, llfn: &'ll Value) {
2018-01-05 05:04:08 +00:00
if cx.sess().must_not_eliminate_frame_pointers() {
2016-08-02 21:25:19 +00:00
llvm::AddFunctionAttrStringValue(
llfn, llvm::AttributePlace::Function,
const_cstr!("no-frame-pointer-elim"), const_cstr!("true"));
}
}
pub fn set_probestack(cx: &CodegenCx<'ll, '_>, llfn: &'ll Value) {
// Only use stack probes if the target specification indicates that we
// should be using stack probes
2018-01-05 05:04:08 +00:00
if !cx.sess().target.target.options.stack_probes {
return
}
// Currently stack probes seem somewhat incompatible with the address
// sanitizer. With asan we're already protected from stack overflow anyway
// so we don't really need stack probes regardless.
2018-01-05 05:04:08 +00:00
match cx.sess().opts.debugging_opts.sanitizer {
Some(Sanitizer::Address) => return,
_ => {}
}
// probestack doesn't play nice either with pgo-gen.
if cx.sess().opts.debugging_opts.pgo_gen.is_some() {
return;
}
// probestack doesn't play nice either with gcov profiling.
if cx.sess().opts.debugging_opts.profile {
return;
}
// Flag our internal `__rust_probestack` function as the stack probe symbol.
// This is defined in the `compiler-builtins` crate for each architecture.
llvm::AddFunctionAttrStringValue(
llfn, llvm::AttributePlace::Function,
const_cstr!("probe-stack"), const_cstr!("__rust_probestack"));
}
pub fn llvm_target_features(sess: &Session) -> impl Iterator<Item = &str> {
const RUSTC_SPECIFIC_FEATURES: &[&str] = &[
"crt-static",
];
let cmdline = sess.opts.cg.target_feature.split(',')
.filter(|f| !RUSTC_SPECIFIC_FEATURES.iter().any(|s| f.contains(s)));
sess.target.target.options.features.split(',')
.chain(cmdline)
.filter(|l| !l.is_empty())
}
pub fn apply_target_cpu_attr(cx: &CodegenCx<'ll, '_>, llfn: &'ll Value) {
let target_cpu = CString::new(cx.tcx.sess.target_cpu().to_string()).unwrap();
llvm::AddFunctionAttrStringValue(
llfn,
llvm::AttributePlace::Function,
const_cstr!("target-cpu"),
target_cpu.as_c_str());
}
/// Composite function which sets LLVM attributes for function depending on its AST (#[attribute])
/// attributes.
pub fn from_fn_attrs(
cx: &CodegenCx<'ll, '_>,
llfn: &'ll Value,
id: Option<DefId>,
) {
let codegen_fn_attrs = id.map(|id| cx.tcx.codegen_fn_attrs(id))
.unwrap_or(CodegenFnAttrs::new());
2018-05-08 13:10:16 +00:00
inline(llfn, codegen_fn_attrs.inline);
// The `uwtable` attribute according to LLVM is:
//
// This attribute indicates that the ABI being targeted requires that an
// unwind table entry be produced for this function even if we can show
// that no exceptions passes by it. This is normally the case for the
// ELF x86-64 abi, but it can be disabled for some compilation units.
//
// Typically when we're compiling with `-C panic=abort` (which implies this
// `no_landing_pads` check) we don't need `uwtable` because we can't
// generate any exceptions! On Windows, however, exceptions include other
// events such as illegal instructions, segfaults, etc. This means that on
// Windows we end up still needing the `uwtable` attribute even if the `-C
// panic=abort` flag is passed.
//
// You can also find more info on why Windows is whitelisted here in:
// https://bugzilla.mozilla.org/show_bug.cgi?id=1302078
if !cx.sess().no_landing_pads() ||
cx.sess().target.target.options.requires_uwtable {
attributes::emit_uwtable(llfn, true);
}
2018-01-05 05:04:08 +00:00
set_frame_pointer_elimination(cx, llfn);
set_probestack(cx, llfn);
2018-05-08 13:10:16 +00:00
if codegen_fn_attrs.flags.contains(CodegenFnAttrFlags::COLD) {
Attribute::Cold.apply_llfn(Function, llfn);
}
2018-05-08 13:10:16 +00:00
if codegen_fn_attrs.flags.contains(CodegenFnAttrFlags::NAKED) {
naked(llfn, true);
}
2018-05-08 13:10:16 +00:00
if codegen_fn_attrs.flags.contains(CodegenFnAttrFlags::ALLOCATOR) {
Attribute::NoAlias.apply_llfn(
llvm::AttributePlace::ReturnValue, llfn);
}
let can_unwind = if codegen_fn_attrs.flags.contains(CodegenFnAttrFlags::UNWIND) {
Some(true)
} else if codegen_fn_attrs.flags.contains(CodegenFnAttrFlags::RUSTC_ALLOCATOR_NOUNWIND) {
Some(false)
// Perhaps questionable, but we assume that anything defined
// *in Rust code* may unwind. Foreign items like `extern "C" {
// fn foo(); }` are assumed not to unwind **unless** they have
// a `#[unwind]` attribute.
} else if id.map(|id| !cx.tcx.is_foreign_item(id)).unwrap_or(false) {
Some(true)
} else {
None
};
match can_unwind {
Some(false) => attributes::unwind(llfn, false),
Some(true) if cx.tcx.sess.panic_strategy() == PanicStrategy::Unwind => {
attributes::unwind(llfn, true);
}
Some(true) | None => {}
}
// Always annotate functions with the target-cpu they are compiled for.
// Without this, ThinLTO won't inline Rust functions into Clang generated
// functions (because Clang annotates functions this way too).
// NOTE: For now we just apply this if -Zcross-lang-lto is specified, since
// it introduce a little overhead and isn't really necessary otherwise.
if cx.tcx.sess.opts.debugging_opts.cross_lang_lto.enabled() {
apply_target_cpu_attr(cx, llfn);
}
let features = llvm_target_features(cx.tcx.sess)
.map(|s| s.to_string())
.chain(
2018-05-08 13:10:16 +00:00
codegen_fn_attrs.target_features
.iter()
.map(|f| {
let feature = &*f.as_str();
format!("+{}", llvm_util::to_llvm_feature(cx.tcx.sess, feature))
})
)
.collect::<Vec<String>>()
.join(",");
if !features.is_empty() {
let val = CString::new(features).unwrap();
llvm::AddFunctionAttrStringValue(
llfn, llvm::AttributePlace::Function,
const_cstr!("target-features"), &val);
}
rustc: Add a `#[wasm_import_module]` attribute This commit adds a new attribute to the Rust compiler specific to the wasm target (and no other targets). The `#[wasm_import_module]` attribute is used to specify the module that a name is imported from, and is used like so: #[wasm_import_module = "./foo.js"] extern { fn some_js_function(); } Here the import of the symbol `some_js_function` is tagged with the `./foo.js` module in the wasm output file. Wasm-the-format includes two fields on all imports, a module and a field. The field is the symbol name (`some_js_function` above) and the module has historically unconditionally been `"env"`. I'm not sure if this `"env"` convention has asm.js or LLVM roots, but regardless we'd like the ability to configure it! The proposed ES module integration with wasm (aka a wasm module is "just another ES module") requires that the import module of wasm imports is interpreted as an ES module import, meaning that you'll need to encode paths, NPM packages, etc. As a result, we'll need this to be something other than `"env"`! Unfortunately neither our version of LLVM nor LLD supports custom import modules (aka anything not `"env"`). My hope is that by the time LLVM 7 is released both will have support, but in the meantime this commit adds some primitive encoding/decoding of wasm files to the compiler. This way rustc postprocesses the wasm module that LLVM emits to ensure it's got all the imports we'd like to have in it. Eventually I'd ideally like to unconditionally require this attribute to be placed on all `extern { ... }` blocks. For now though it seemed prudent to add it as an unstable attribute, so for now it's not required (as that'd force usage of a feature gate). Hopefully it doesn't take too long to "stabilize" this! cc rust-lang-nursery/rust-wasm#29
2018-02-10 22:28:17 +00:00
// Note that currently the `wasm-import-module` doesn't do anything, but
// eventually LLVM 7 should read this and ferry the appropriate import
// module to the output file.
if let Some(id) = id {
if cx.tcx.sess.target.target.arch == "wasm32" {
if let Some(module) = wasm_import_module(cx.tcx, id) {
llvm::AddFunctionAttrStringValue(
llfn,
llvm::AttributePlace::Function,
const_cstr!("wasm-import-module"),
&module,
);
}
rustc: Add a `#[wasm_import_module]` attribute This commit adds a new attribute to the Rust compiler specific to the wasm target (and no other targets). The `#[wasm_import_module]` attribute is used to specify the module that a name is imported from, and is used like so: #[wasm_import_module = "./foo.js"] extern { fn some_js_function(); } Here the import of the symbol `some_js_function` is tagged with the `./foo.js` module in the wasm output file. Wasm-the-format includes two fields on all imports, a module and a field. The field is the symbol name (`some_js_function` above) and the module has historically unconditionally been `"env"`. I'm not sure if this `"env"` convention has asm.js or LLVM roots, but regardless we'd like the ability to configure it! The proposed ES module integration with wasm (aka a wasm module is "just another ES module") requires that the import module of wasm imports is interpreted as an ES module import, meaning that you'll need to encode paths, NPM packages, etc. As a result, we'll need this to be something other than `"env"`! Unfortunately neither our version of LLVM nor LLD supports custom import modules (aka anything not `"env"`). My hope is that by the time LLVM 7 is released both will have support, but in the meantime this commit adds some primitive encoding/decoding of wasm files to the compiler. This way rustc postprocesses the wasm module that LLVM emits to ensure it's got all the imports we'd like to have in it. Eventually I'd ideally like to unconditionally require this attribute to be placed on all `extern { ... }` blocks. For now though it seemed prudent to add it as an unstable attribute, so for now it's not required (as that'd force usage of a feature gate). Hopefully it doesn't take too long to "stabilize" this! cc rust-lang-nursery/rust-wasm#29
2018-02-10 22:28:17 +00:00
}
}
}
pub fn provide(providers: &mut Providers) {
providers.target_features_whitelist = |tcx, cnum| {
assert_eq!(cnum, LOCAL_CRATE);
if tcx.sess.opts.actually_rustdoc {
// rustdoc needs to be able to document functions that use all the features, so
// whitelist them all
Lrc::new(llvm_util::all_known_features()
.map(|(a, b)| (a.to_string(), b.map(|s| s.to_string())))
.collect())
} else {
Lrc::new(llvm_util::target_feature_whitelist(tcx.sess)
.iter()
.map(|&(a, b)| (a.to_string(), b.map(|s| s.to_string())))
.collect())
}
};
rustc: Add a `#[wasm_import_module]` attribute This commit adds a new attribute to the Rust compiler specific to the wasm target (and no other targets). The `#[wasm_import_module]` attribute is used to specify the module that a name is imported from, and is used like so: #[wasm_import_module = "./foo.js"] extern { fn some_js_function(); } Here the import of the symbol `some_js_function` is tagged with the `./foo.js` module in the wasm output file. Wasm-the-format includes two fields on all imports, a module and a field. The field is the symbol name (`some_js_function` above) and the module has historically unconditionally been `"env"`. I'm not sure if this `"env"` convention has asm.js or LLVM roots, but regardless we'd like the ability to configure it! The proposed ES module integration with wasm (aka a wasm module is "just another ES module") requires that the import module of wasm imports is interpreted as an ES module import, meaning that you'll need to encode paths, NPM packages, etc. As a result, we'll need this to be something other than `"env"`! Unfortunately neither our version of LLVM nor LLD supports custom import modules (aka anything not `"env"`). My hope is that by the time LLVM 7 is released both will have support, but in the meantime this commit adds some primitive encoding/decoding of wasm files to the compiler. This way rustc postprocesses the wasm module that LLVM emits to ensure it's got all the imports we'd like to have in it. Eventually I'd ideally like to unconditionally require this attribute to be placed on all `extern { ... }` blocks. For now though it seemed prudent to add it as an unstable attribute, so for now it's not required (as that'd force usage of a feature gate). Hopefully it doesn't take too long to "stabilize" this! cc rust-lang-nursery/rust-wasm#29
2018-02-10 22:28:17 +00:00
provide_extern(providers);
}
rustc: Add a `#[wasm_import_module]` attribute This commit adds a new attribute to the Rust compiler specific to the wasm target (and no other targets). The `#[wasm_import_module]` attribute is used to specify the module that a name is imported from, and is used like so: #[wasm_import_module = "./foo.js"] extern { fn some_js_function(); } Here the import of the symbol `some_js_function` is tagged with the `./foo.js` module in the wasm output file. Wasm-the-format includes two fields on all imports, a module and a field. The field is the symbol name (`some_js_function` above) and the module has historically unconditionally been `"env"`. I'm not sure if this `"env"` convention has asm.js or LLVM roots, but regardless we'd like the ability to configure it! The proposed ES module integration with wasm (aka a wasm module is "just another ES module") requires that the import module of wasm imports is interpreted as an ES module import, meaning that you'll need to encode paths, NPM packages, etc. As a result, we'll need this to be something other than `"env"`! Unfortunately neither our version of LLVM nor LLD supports custom import modules (aka anything not `"env"`). My hope is that by the time LLVM 7 is released both will have support, but in the meantime this commit adds some primitive encoding/decoding of wasm files to the compiler. This way rustc postprocesses the wasm module that LLVM emits to ensure it's got all the imports we'd like to have in it. Eventually I'd ideally like to unconditionally require this attribute to be placed on all `extern { ... }` blocks. For now though it seemed prudent to add it as an unstable attribute, so for now it's not required (as that'd force usage of a feature gate). Hopefully it doesn't take too long to "stabilize" this! cc rust-lang-nursery/rust-wasm#29
2018-02-10 22:28:17 +00:00
pub fn provide_extern(providers: &mut Providers) {
providers.wasm_import_module_map = |tcx, cnum| {
// Build up a map from DefId to a `NativeLibrary` structure, where
// `NativeLibrary` internally contains information about
// `#[link(wasm_import_module = "...")]` for example.
let native_libs = tcx.native_libraries(cnum);
let mut def_id_to_native_lib = FxHashMap();
for lib in native_libs.iter() {
if let Some(id) = lib.foreign_module {
def_id_to_native_lib.insert(id, lib);
}
}
rustc: Add a `#[wasm_import_module]` attribute This commit adds a new attribute to the Rust compiler specific to the wasm target (and no other targets). The `#[wasm_import_module]` attribute is used to specify the module that a name is imported from, and is used like so: #[wasm_import_module = "./foo.js"] extern { fn some_js_function(); } Here the import of the symbol `some_js_function` is tagged with the `./foo.js` module in the wasm output file. Wasm-the-format includes two fields on all imports, a module and a field. The field is the symbol name (`some_js_function` above) and the module has historically unconditionally been `"env"`. I'm not sure if this `"env"` convention has asm.js or LLVM roots, but regardless we'd like the ability to configure it! The proposed ES module integration with wasm (aka a wasm module is "just another ES module") requires that the import module of wasm imports is interpreted as an ES module import, meaning that you'll need to encode paths, NPM packages, etc. As a result, we'll need this to be something other than `"env"`! Unfortunately neither our version of LLVM nor LLD supports custom import modules (aka anything not `"env"`). My hope is that by the time LLVM 7 is released both will have support, but in the meantime this commit adds some primitive encoding/decoding of wasm files to the compiler. This way rustc postprocesses the wasm module that LLVM emits to ensure it's got all the imports we'd like to have in it. Eventually I'd ideally like to unconditionally require this attribute to be placed on all `extern { ... }` blocks. For now though it seemed prudent to add it as an unstable attribute, so for now it's not required (as that'd force usage of a feature gate). Hopefully it doesn't take too long to "stabilize" this! cc rust-lang-nursery/rust-wasm#29
2018-02-10 22:28:17 +00:00
let mut ret = FxHashMap();
for lib in tcx.foreign_modules(cnum).iter() {
let module = def_id_to_native_lib
.get(&lib.def_id)
.and_then(|s| s.wasm_import_module);
rustc: Add a `#[wasm_import_module]` attribute This commit adds a new attribute to the Rust compiler specific to the wasm target (and no other targets). The `#[wasm_import_module]` attribute is used to specify the module that a name is imported from, and is used like so: #[wasm_import_module = "./foo.js"] extern { fn some_js_function(); } Here the import of the symbol `some_js_function` is tagged with the `./foo.js` module in the wasm output file. Wasm-the-format includes two fields on all imports, a module and a field. The field is the symbol name (`some_js_function` above) and the module has historically unconditionally been `"env"`. I'm not sure if this `"env"` convention has asm.js or LLVM roots, but regardless we'd like the ability to configure it! The proposed ES module integration with wasm (aka a wasm module is "just another ES module") requires that the import module of wasm imports is interpreted as an ES module import, meaning that you'll need to encode paths, NPM packages, etc. As a result, we'll need this to be something other than `"env"`! Unfortunately neither our version of LLVM nor LLD supports custom import modules (aka anything not `"env"`). My hope is that by the time LLVM 7 is released both will have support, but in the meantime this commit adds some primitive encoding/decoding of wasm files to the compiler. This way rustc postprocesses the wasm module that LLVM emits to ensure it's got all the imports we'd like to have in it. Eventually I'd ideally like to unconditionally require this attribute to be placed on all `extern { ... }` blocks. For now though it seemed prudent to add it as an unstable attribute, so for now it's not required (as that'd force usage of a feature gate). Hopefully it doesn't take too long to "stabilize" this! cc rust-lang-nursery/rust-wasm#29
2018-02-10 22:28:17 +00:00
let module = match module {
Some(s) => s,
None => continue,
};
for id in lib.foreign_items.iter() {
assert_eq!(id.krate, cnum);
ret.insert(*id, module.to_string());
}
}
Lrc::new(ret)
};
rustc: Add a `#[wasm_import_module]` attribute This commit adds a new attribute to the Rust compiler specific to the wasm target (and no other targets). The `#[wasm_import_module]` attribute is used to specify the module that a name is imported from, and is used like so: #[wasm_import_module = "./foo.js"] extern { fn some_js_function(); } Here the import of the symbol `some_js_function` is tagged with the `./foo.js` module in the wasm output file. Wasm-the-format includes two fields on all imports, a module and a field. The field is the symbol name (`some_js_function` above) and the module has historically unconditionally been `"env"`. I'm not sure if this `"env"` convention has asm.js or LLVM roots, but regardless we'd like the ability to configure it! The proposed ES module integration with wasm (aka a wasm module is "just another ES module") requires that the import module of wasm imports is interpreted as an ES module import, meaning that you'll need to encode paths, NPM packages, etc. As a result, we'll need this to be something other than `"env"`! Unfortunately neither our version of LLVM nor LLD supports custom import modules (aka anything not `"env"`). My hope is that by the time LLVM 7 is released both will have support, but in the meantime this commit adds some primitive encoding/decoding of wasm files to the compiler. This way rustc postprocesses the wasm module that LLVM emits to ensure it's got all the imports we'd like to have in it. Eventually I'd ideally like to unconditionally require this attribute to be placed on all `extern { ... }` blocks. For now though it seemed prudent to add it as an unstable attribute, so for now it's not required (as that'd force usage of a feature gate). Hopefully it doesn't take too long to "stabilize" this! cc rust-lang-nursery/rust-wasm#29
2018-02-10 22:28:17 +00:00
}
fn wasm_import_module(tcx: TyCtxt, id: DefId) -> Option<CString> {
tcx.wasm_import_module_map(id.krate)
.get(&id)
.map(|s| CString::new(&s[..]).unwrap())
}