rust/compiler/rustc_session/src/filesearch.rs

158 lines
5.7 KiB
Rust
Raw Normal View History

2021-08-31 16:31:29 +00:00
//! A module for searching for libraries
pub use self::FileMatch::*;
std: Add a new `env` module This is an implementation of [RFC 578][rfc] which adds a new `std::env` module to replace most of the functionality in the current `std::os` module. More details can be found in the RFC itself, but as a summary the following methods have all been deprecated: [rfc]: https://github.com/rust-lang/rfcs/pull/578 * `os::args_as_bytes` => `env::args` * `os::args` => `env::args` * `os::consts` => `env::consts` * `os::dll_filename` => no replacement, use `env::consts` directly * `os::page_size` => `env::page_size` * `os::make_absolute` => use `env::current_dir` + `join` instead * `os::getcwd` => `env::current_dir` * `os::change_dir` => `env::set_current_dir` * `os::homedir` => `env::home_dir` * `os::tmpdir` => `env::temp_dir` * `os::join_paths` => `env::join_paths` * `os::split_paths` => `env::split_paths` * `os::self_exe_name` => `env::current_exe` * `os::self_exe_path` => use `env::current_exe` + `pop` * `os::set_exit_status` => `env::set_exit_status` * `os::get_exit_status` => `env::get_exit_status` * `os::env` => `env::vars` * `os::env_as_bytes` => `env::vars` * `os::getenv` => `env::var` or `env::var_string` * `os::getenv_as_bytes` => `env::var` * `os::setenv` => `env::set_var` * `os::unsetenv` => `env::remove_var` Many function signatures have also been tweaked for various purposes, but the main changes were: * `Vec`-returning APIs now all return iterators instead * All APIs are now centered around `OsString` instead of `Vec<u8>` or `String`. There is currently on convenience API, `env::var_string`, which can be used to get the value of an environment variable as a unicode `String`. All old APIs are `#[deprecated]` in-place and will remain for some time to allow for migrations. The semantics of the APIs have been tweaked slightly with regard to dealing with invalid unicode (panic instead of replacement). The new `std::env` module is all contained within the `env` feature, so crates must add the following to access the new APIs: #![feature(env)] [breaking-change]
2015-01-27 20:20:58 +00:00
use std::env;
use std::fs;
use std::path::{Path, PathBuf};
use crate::search_paths::{PathKind, SearchPath, SearchPathFile};
2019-12-22 22:42:04 +00:00
use rustc_fs_util::fix_windows_verbatim_for_gcc;
2020-08-14 06:05:01 +00:00
use tracing::debug;
2015-03-30 13:38:44 +00:00
#[derive(Copy, Clone)]
pub enum FileMatch {
FileMatches,
FileDoesntMatch,
}
#[derive(Clone)]
2014-03-09 12:24:58 +00:00
pub struct FileSearch<'a> {
sysroot: &'a Path,
triple: &'a str,
search_paths: &'a [SearchPath],
tlib_path: &'a SearchPath,
kind: PathKind,
}
2014-03-09 12:24:58 +00:00
impl<'a> FileSearch<'a> {
pub fn search_paths(&self) -> impl Iterator<Item = &'a SearchPath> {
let kind = self.kind;
2019-12-22 22:42:04 +00:00
self.search_paths
.iter()
.filter(move |sp| sp.kind.matches(kind))
.chain(std::iter::once(self.tlib_path))
}
pub fn get_lib_path(&self) -> PathBuf {
make_target_lib_path(self.sysroot, self.triple)
}
2020-06-22 17:56:56 +00:00
pub fn get_self_contained_lib_path(&self) -> PathBuf {
self.get_lib_path().join("self-contained")
}
rustc: Fix a leak in dependency= paths With the addition of separate search paths to the compiler, it was intended that applications such as Cargo could require a `--extern` flag per `extern crate` directive in the source. The system can currently be subverted, however, due to the `existing_match()` logic in the crate loader. When loading crates we first attempt to match an `extern crate` directive against all previously loaded crates to avoid reading metadata twice. This "hit the cache if possible" step was erroneously leaking crates across the search path boundaries, however. For example: extern crate b; extern crate a; If `b` depends on `a`, then it will load crate `a` when the `extern crate b` directive is being processed. When the compiler reaches `extern crate a` it will use the previously loaded version no matter what. If the compiler was not invoked with `-L crate=path/to/a`, it will still succeed. This behavior is allowing `extern crate` declarations in Cargo without a corresponding declaration in the manifest of a dependency, which is considered a bug. This commit fixes this problem by keeping track of the origin search path for a crate. Crates loaded from the dependency search path are not candidates for crates which are loaded from the crate search path. As a result of this fix, this is a likely a breaking change for a number of Cargo packages. If the compiler starts informing that a crate can no longer be found, it likely means that the dependency was forgotten in your Cargo.toml. [breaking-change]
2015-01-06 16:46:07 +00:00
pub fn search<F>(&self, mut pick: F)
2019-12-22 22:42:04 +00:00
where
F: FnMut(&SearchPathFile, PathKind) -> FileMatch,
rustc: Fix a leak in dependency= paths With the addition of separate search paths to the compiler, it was intended that applications such as Cargo could require a `--extern` flag per `extern crate` directive in the source. The system can currently be subverted, however, due to the `existing_match()` logic in the crate loader. When loading crates we first attempt to match an `extern crate` directive against all previously loaded crates to avoid reading metadata twice. This "hit the cache if possible" step was erroneously leaking crates across the search path boundaries, however. For example: extern crate b; extern crate a; If `b` depends on `a`, then it will load crate `a` when the `extern crate b` directive is being processed. When the compiler reaches `extern crate a` it will use the previously loaded version no matter what. If the compiler was not invoked with `-L crate=path/to/a`, it will still succeed. This behavior is allowing `extern crate` declarations in Cargo without a corresponding declaration in the manifest of a dependency, which is considered a bug. This commit fixes this problem by keeping track of the origin search path for a crate. Crates loaded from the dependency search path are not candidates for crates which are loaded from the crate search path. As a result of this fix, this is a likely a breaking change for a number of Cargo packages. If the compiler starts informing that a crate can no longer be found, it likely means that the dependency was forgotten in your Cargo.toml. [breaking-change]
2015-01-06 16:46:07 +00:00
{
for search_path in self.search_paths() {
debug!("searching {}", search_path.dir.display());
fn is_rlib(spf: &SearchPathFile) -> bool {
if let Some(f) = &spf.file_name_str { f.ends_with(".rlib") } else { false }
}
// Reading metadata out of rlibs is faster, and if we find both
// an rlib and a dylib we only read one of the files of
// metadata, so in the name of speed, bring all rlib files to
// the front of the search list.
let files1 = search_path.files.iter().filter(|spf| is_rlib(&spf));
let files2 = search_path.files.iter().filter(|spf| !is_rlib(&spf));
for spf in files1.chain(files2) {
debug!("testing {}", spf.path.display());
let maybe_picked = pick(spf, search_path.kind);
match maybe_picked {
FileMatches => {
debug!("picked {}", spf.path.display());
}
FileDoesntMatch => {
debug!("rejected {}", spf.path.display());
}
}
}
}
}
2019-12-22 22:42:04 +00:00
pub fn new(
sysroot: &'a Path,
triple: &'a str,
search_paths: &'a [SearchPath],
2019-12-22 22:42:04 +00:00
tlib_path: &'a SearchPath,
kind: PathKind,
) -> FileSearch<'a> {
debug!("using sysroot = {}, triple = {}", sysroot.display(), triple);
2019-12-22 22:42:04 +00:00
FileSearch { sysroot, triple, search_paths, tlib_path, kind }
}
2021-08-31 16:31:29 +00:00
/// Returns just the directories within the search paths.
pub fn search_path_dirs(&self) -> Vec<PathBuf> {
2019-12-22 22:42:04 +00:00
self.search_paths().map(|sp| sp.dir.to_path_buf()).collect()
}
}
pub fn make_target_lib_path(sysroot: &Path, target_triple: &str) -> PathBuf {
let rustlib_path = rustc_target::target_rustlib_path(sysroot, target_triple);
std::array::IntoIter::new([sysroot, Path::new(&rustlib_path), Path::new("lib")])
.collect::<PathBuf>()
}
2021-08-31 16:31:29 +00:00
/// This function checks if sysroot is found using env::args().next(), and if it
/// is not found, uses env::current_exe() to imply sysroot.
pub fn get_or_default_sysroot() -> PathBuf {
// Follow symlinks. If the resolved path is relative, make it absolute.
fn canonicalize(path: PathBuf) -> PathBuf {
let path = fs::canonicalize(&path).unwrap_or(path);
// See comments on this target function, but the gist is that
// gcc chokes on verbatim paths which fs::canonicalize generates
// so we try to avoid those kinds of paths.
fix_windows_verbatim_for_gcc(&path)
}
// Use env::current_exe() to get the path of the executable following
// symlinks/canonicalizing components.
fn from_current_exe() -> PathBuf {
match env::current_exe() {
Ok(exe) => {
let mut p = canonicalize(exe);
p.pop();
p.pop();
p
}
Err(e) => panic!("failed to get current_exe: {}", e),
}
}
// Use env::args().next() to get the path of the executable without
// following symlinks/canonicalizing any component. This makes the rustc
// binary able to locate Rust libraries in systems using content-addressable
// storage (CAS).
fn from_env_args_next() -> Option<PathBuf> {
match env::args_os().next() {
Some(first_arg) => {
let mut p = PathBuf::from(first_arg);
// Check if sysroot is found using env::args().next() only if the rustc in argv[0]
// is a symlink (see #79253). We might want to change/remove it to conform with
// https://www.gnu.org/prep/standards/standards.html#Finding-Program-Files in the
// future.
if fs::read_link(&p).is_err() {
// Path is not a symbolic link or does not exist.
return None;
}
// Pop off `bin/rustc`, obtaining the suspected sysroot.
p.pop();
p.pop();
// Look for the target rustlib directory in the suspected sysroot.
let mut rustlib_path = rustc_target::target_rustlib_path(&p, "dummy");
rustlib_path.pop(); // pop off the dummy target.
if rustlib_path.exists() { Some(p) } else { None }
}
None => None,
}
}
// Check if sysroot is found using env::args().next(), and if is not found,
// use env::current_exe() to imply sysroot.
2021-02-25 01:13:42 +00:00
from_env_args_next().unwrap_or_else(from_current_exe)
2011-11-10 16:41:42 +00:00
}