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 07:14:20 +00:00
|
|
|
//! A helper class for dealing with static archives
|
|
|
|
|
2019-12-22 22:42:04 +00:00
|
|
|
use std::ffi::{CStr, CString};
|
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 07:14:20 +00:00
|
|
|
use std::io;
|
|
|
|
use std::mem;
|
|
|
|
use std::path::{Path, PathBuf};
|
2018-07-17 12:02:11 +00:00
|
|
|
use std::ptr;
|
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 07:14:20 +00:00
|
|
|
use std::str;
|
|
|
|
|
2019-02-17 18:58:58 +00:00
|
|
|
use crate::llvm::archive_ro::{ArchiveRO, Child};
|
2021-03-08 20:42:54 +00:00
|
|
|
use crate::llvm::{self, ArchiveKind, LLVMMachineType, LLVMRustCOFFShortExport};
|
2019-12-22 22:42:04 +00:00
|
|
|
use rustc_codegen_ssa::back::archive::{find_library, ArchiveBuilder};
|
2020-04-23 18:45:55 +00:00
|
|
|
use rustc_codegen_ssa::{looks_like_rust_object_file, METADATA_FILENAME};
|
2021-03-08 20:42:54 +00:00
|
|
|
use rustc_data_structures::temp_dir::MaybeTempDir;
|
2021-06-08 20:56:06 +00:00
|
|
|
use rustc_middle::middle::cstore::{DllCallingConvention, DllImport};
|
2020-03-11 11:49:08 +00:00
|
|
|
use rustc_session::Session;
|
2020-01-01 18:30:57 +00:00
|
|
|
use rustc_span::symbol::Symbol;
|
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 07:14:20 +00:00
|
|
|
|
2019-03-30 11:59:40 +00:00
|
|
|
struct ArchiveConfig<'a> {
|
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 07:14:20 +00:00
|
|
|
pub sess: &'a Session,
|
|
|
|
pub dst: PathBuf,
|
|
|
|
pub src: Option<PathBuf>,
|
|
|
|
pub lib_search_paths: Vec<PathBuf>,
|
|
|
|
}
|
|
|
|
|
2017-10-08 08:41:12 +00:00
|
|
|
/// Helper for adding many files to an archive.
|
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 07:14:20 +00:00
|
|
|
#[must_use = "must call build() to finish building the archive"]
|
2019-03-30 11:59:40 +00:00
|
|
|
pub struct LlvmArchiveBuilder<'a> {
|
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 07:14:20 +00:00
|
|
|
config: ArchiveConfig<'a>,
|
|
|
|
removals: Vec<String>,
|
|
|
|
additions: Vec<Addition>,
|
|
|
|
should_update_symbols: bool,
|
|
|
|
src_archive: Option<Option<ArchiveRO>>,
|
|
|
|
}
|
|
|
|
|
|
|
|
enum Addition {
|
2019-12-22 22:42:04 +00:00
|
|
|
File { path: PathBuf, name_in_archive: String },
|
|
|
|
Archive { path: PathBuf, archive: ArchiveRO, skip: Box<dyn FnMut(&str) -> bool> },
|
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 07:14:20 +00:00
|
|
|
}
|
|
|
|
|
2019-07-02 00:28:10 +00:00
|
|
|
impl Addition {
|
|
|
|
fn path(&self) -> &Path {
|
|
|
|
match self {
|
|
|
|
Addition::File { path, .. } | Addition::Archive { path, .. } => path,
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-02-25 07:40:18 +00:00
|
|
|
fn is_relevant_child(c: &Child<'_>) -> bool {
|
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 07:14:20 +00:00
|
|
|
match c.name() {
|
|
|
|
Some(name) => !name.contains("SYMDEF"),
|
|
|
|
None => false,
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-12-22 22:42:04 +00:00
|
|
|
fn archive_config<'a>(sess: &'a Session, output: &Path, input: Option<&Path>) -> ArchiveConfig<'a> {
|
2019-03-30 11:59:40 +00:00
|
|
|
use rustc_codegen_ssa::back::link::archive_search_paths;
|
|
|
|
ArchiveConfig {
|
|
|
|
sess,
|
|
|
|
dst: output.to_path_buf(),
|
|
|
|
src: input.map(|p| p.to_path_buf()),
|
|
|
|
lib_search_paths: archive_search_paths(sess),
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2021-03-08 20:42:54 +00:00
|
|
|
/// Map machine type strings to values of LLVM's MachineTypes enum.
|
|
|
|
fn llvm_machine_type(cpu: &str) -> LLVMMachineType {
|
|
|
|
match cpu {
|
|
|
|
"x86_64" => LLVMMachineType::AMD64,
|
|
|
|
"x86" => LLVMMachineType::I386,
|
|
|
|
"aarch64" => LLVMMachineType::ARM64,
|
|
|
|
"arm" => LLVMMachineType::ARM,
|
|
|
|
_ => panic!("unsupported cpu type {}", cpu),
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-03-30 11:59:40 +00:00
|
|
|
impl<'a> ArchiveBuilder<'a> for LlvmArchiveBuilder<'a> {
|
2019-02-08 13:53:55 +00:00
|
|
|
/// Creates a new static archive, ready for modifying the archive specified
|
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 07:14:20 +00:00
|
|
|
/// by `config`.
|
2019-12-22 22:42:04 +00:00
|
|
|
fn new(sess: &'a Session, output: &Path, input: Option<&Path>) -> LlvmArchiveBuilder<'a> {
|
2019-03-30 11:59:40 +00:00
|
|
|
let config = archive_config(sess, output, input);
|
|
|
|
LlvmArchiveBuilder {
|
2017-08-07 05:54:09 +00:00
|
|
|
config,
|
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 07:14:20 +00:00
|
|
|
removals: Vec::new(),
|
|
|
|
additions: Vec::new(),
|
|
|
|
should_update_symbols: false,
|
|
|
|
src_archive: None,
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Removes a file from this archive
|
2019-03-30 11:59:40 +00:00
|
|
|
fn remove_file(&mut self, file: &str) {
|
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 07:14:20 +00:00
|
|
|
self.removals.push(file.to_string());
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Lists all files in an archive
|
2019-03-30 11:59:40 +00:00
|
|
|
fn src_files(&mut self) -> Vec<String> {
|
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 07:14:20 +00:00
|
|
|
if self.src_archive().is_none() {
|
2019-12-22 22:42:04 +00:00
|
|
|
return Vec::new();
|
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 07:14:20 +00:00
|
|
|
}
|
2018-10-06 09:37:28 +00:00
|
|
|
|
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 07:14:20 +00:00
|
|
|
let archive = self.src_archive.as_ref().unwrap().as_ref().unwrap();
|
2018-10-06 09:46:46 +00:00
|
|
|
|
2019-12-22 22:42:04 +00:00
|
|
|
archive
|
|
|
|
.iter()
|
|
|
|
.filter_map(|child| child.ok())
|
|
|
|
.filter(is_relevant_child)
|
|
|
|
.filter_map(|child| child.name())
|
|
|
|
.filter(|name| !self.removals.iter().any(|x| x == name))
|
|
|
|
.map(|name| name.to_owned())
|
|
|
|
.collect()
|
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 07:14:20 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/// Adds all of the contents of a native library to this archive. This will
|
|
|
|
/// search in the relevant locations for a library named `name`.
|
2021-03-25 04:45:09 +00:00
|
|
|
fn add_native_library(&mut self, name: Symbol, verbatim: bool) {
|
|
|
|
let location =
|
|
|
|
find_library(name, verbatim, &self.config.lib_search_paths, self.config.sess);
|
2016-06-10 21:27:19 +00:00
|
|
|
self.add_archive(&location, |_| false).unwrap_or_else(|e| {
|
2019-12-22 22:42:04 +00:00
|
|
|
self.config.sess.fatal(&format!(
|
|
|
|
"failed to add native library {}: {}",
|
|
|
|
location.to_string_lossy(),
|
|
|
|
e
|
|
|
|
));
|
2015-09-25 04:46:33 +00:00
|
|
|
});
|
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 07:14:20 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/// Adds all of the contents of the rlib at the specified path to this
|
|
|
|
/// archive.
|
|
|
|
///
|
|
|
|
/// This ignores adding the bytecode from the rlib, and if LTO is enabled
|
|
|
|
/// then the object file also isn't added.
|
2019-12-22 22:42:04 +00:00
|
|
|
fn add_rlib(
|
|
|
|
&mut self,
|
|
|
|
rlib: &Path,
|
|
|
|
name: &str,
|
|
|
|
lto: bool,
|
|
|
|
skip_objects: bool,
|
|
|
|
) -> io::Result<()> {
|
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 07:14:20 +00:00
|
|
|
// Ignoring obj file starting with the crate name
|
|
|
|
// as simple comparison is not enough - there
|
|
|
|
// might be also an extra name suffix
|
2018-07-27 09:11:18 +00:00
|
|
|
let obj_start = name.to_owned();
|
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 07:14:20 +00:00
|
|
|
|
2016-06-10 21:27:19 +00:00
|
|
|
self.add_archive(rlib, move |fname: &str| {
|
2020-04-23 18:45:55 +00:00
|
|
|
// Ignore metadata files, no matter the name.
|
|
|
|
if fname == METADATA_FILENAME {
|
2019-12-22 22:42:04 +00:00
|
|
|
return true;
|
rustc: Implement #[link(cfg(..))] and crt-static
This commit is an implementation of [RFC 1721] which adds a new target feature
to the compiler, `crt-static`, which can be used to select how the C runtime for
a target is linked. Most targets dynamically linke the C runtime by default with
the notable exception of some of the musl targets.
[RFC 1721]: https://github.com/rust-lang/rfcs/blob/master/text/1721-crt-static.md
This commit first adds the new target-feature, `crt-static`. If enabled, then
the `cfg(target_feature = "crt-static")` will be available. Targets like musl
will have this enabled by default. This feature can be controlled through the
standard target-feature interface, `-C target-feature=+crt-static` or
`-C target-feature=-crt-static`.
Next this adds an gated and unstable `#[link(cfg(..))]` feature to enable the
`crt-static` semantics we want with libc. The exact behavior of this attribute
is a little squishy, but it's intended to be a forever-unstable
implementation detail of the liblibc crate.
Specifically the `#[link(cfg(..))]` annotation means that the `#[link]`
directive is only active in a compilation unit if that `cfg` value is satisfied.
For example when compiling an rlib, these directives are just encoded and
ignored for dylibs, and all staticlibs are continued to be put into the rlib as
usual. When placing that rlib into a staticlib, executable, or dylib, however,
the `cfg` is evaluated *as if it were defined in the final artifact* and the
library is decided to be linked or not.
Essentially, what'll happen is:
* On MSVC with `-C target-feature=-crt-static`, the `msvcrt.lib` library will be
linked to.
* On MSVC with `-C target-feature=+crt-static`, the `libcmt.lib` library will be
linked to.
* On musl with `-C target-feature=-crt-static`, the object files in liblibc.rlib
are removed and `-lc` is passed instead.
* On musl with `-C target-feature=+crt-static`, the object files in liblibc.rlib
are used and `-lc` is not passed.
This commit does **not** include an update to the liblibc module to implement
these changes. I plan to do that just after the 1.14.0 beta release is cut to
ensure we get ample time to test this feature.
cc #37406
2016-10-31 23:40:13 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
// Don't include Rust objects if LTO is enabled
|
2019-10-15 11:42:27 +00:00
|
|
|
if lto && looks_like_rust_object_file(fname) {
|
2019-12-22 22:42:04 +00:00
|
|
|
return true;
|
rustc: Implement #[link(cfg(..))] and crt-static
This commit is an implementation of [RFC 1721] which adds a new target feature
to the compiler, `crt-static`, which can be used to select how the C runtime for
a target is linked. Most targets dynamically linke the C runtime by default with
the notable exception of some of the musl targets.
[RFC 1721]: https://github.com/rust-lang/rfcs/blob/master/text/1721-crt-static.md
This commit first adds the new target-feature, `crt-static`. If enabled, then
the `cfg(target_feature = "crt-static")` will be available. Targets like musl
will have this enabled by default. This feature can be controlled through the
standard target-feature interface, `-C target-feature=+crt-static` or
`-C target-feature=-crt-static`.
Next this adds an gated and unstable `#[link(cfg(..))]` feature to enable the
`crt-static` semantics we want with libc. The exact behavior of this attribute
is a little squishy, but it's intended to be a forever-unstable
implementation detail of the liblibc crate.
Specifically the `#[link(cfg(..))]` annotation means that the `#[link]`
directive is only active in a compilation unit if that `cfg` value is satisfied.
For example when compiling an rlib, these directives are just encoded and
ignored for dylibs, and all staticlibs are continued to be put into the rlib as
usual. When placing that rlib into a staticlib, executable, or dylib, however,
the `cfg` is evaluated *as if it were defined in the final artifact* and the
library is decided to be linked or not.
Essentially, what'll happen is:
* On MSVC with `-C target-feature=-crt-static`, the `msvcrt.lib` library will be
linked to.
* On MSVC with `-C target-feature=+crt-static`, the `libcmt.lib` library will be
linked to.
* On musl with `-C target-feature=-crt-static`, the object files in liblibc.rlib
are removed and `-lc` is passed instead.
* On musl with `-C target-feature=+crt-static`, the object files in liblibc.rlib
are used and `-lc` is not passed.
This commit does **not** include an update to the liblibc module to implement
these changes. I plan to do that just after the 1.14.0 beta release is cut to
ensure we get ample time to test this feature.
cc #37406
2016-10-31 23:40:13 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
// Otherwise if this is *not* a rust object and we're skipping
|
|
|
|
// objects then skip this file
|
|
|
|
if skip_objects && (!fname.starts_with(&obj_start) || !fname.ends_with(".o")) {
|
2019-12-22 22:42:04 +00:00
|
|
|
return true;
|
rustc: Implement #[link(cfg(..))] and crt-static
This commit is an implementation of [RFC 1721] which adds a new target feature
to the compiler, `crt-static`, which can be used to select how the C runtime for
a target is linked. Most targets dynamically linke the C runtime by default with
the notable exception of some of the musl targets.
[RFC 1721]: https://github.com/rust-lang/rfcs/blob/master/text/1721-crt-static.md
This commit first adds the new target-feature, `crt-static`. If enabled, then
the `cfg(target_feature = "crt-static")` will be available. Targets like musl
will have this enabled by default. This feature can be controlled through the
standard target-feature interface, `-C target-feature=+crt-static` or
`-C target-feature=-crt-static`.
Next this adds an gated and unstable `#[link(cfg(..))]` feature to enable the
`crt-static` semantics we want with libc. The exact behavior of this attribute
is a little squishy, but it's intended to be a forever-unstable
implementation detail of the liblibc crate.
Specifically the `#[link(cfg(..))]` annotation means that the `#[link]`
directive is only active in a compilation unit if that `cfg` value is satisfied.
For example when compiling an rlib, these directives are just encoded and
ignored for dylibs, and all staticlibs are continued to be put into the rlib as
usual. When placing that rlib into a staticlib, executable, or dylib, however,
the `cfg` is evaluated *as if it were defined in the final artifact* and the
library is decided to be linked or not.
Essentially, what'll happen is:
* On MSVC with `-C target-feature=-crt-static`, the `msvcrt.lib` library will be
linked to.
* On MSVC with `-C target-feature=+crt-static`, the `libcmt.lib` library will be
linked to.
* On musl with `-C target-feature=-crt-static`, the object files in liblibc.rlib
are removed and `-lc` is passed instead.
* On musl with `-C target-feature=+crt-static`, the object files in liblibc.rlib
are used and `-lc` is not passed.
This commit does **not** include an update to the liblibc module to implement
these changes. I plan to do that just after the 1.14.0 beta release is cut to
ensure we get ample time to test this feature.
cc #37406
2016-10-31 23:40:13 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
// ok, don't skip this
|
2020-03-20 14:03:11 +00:00
|
|
|
false
|
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 07:14:20 +00:00
|
|
|
})
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Adds an arbitrary file to this archive
|
2019-03-30 11:59:40 +00:00
|
|
|
fn add_file(&mut self, file: &Path) {
|
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 07:14:20 +00:00
|
|
|
let name = file.file_name().unwrap().to_str().unwrap();
|
2019-12-22 22:42:04 +00:00
|
|
|
self.additions
|
|
|
|
.push(Addition::File { path: file.to_path_buf(), name_in_archive: name.to_owned() });
|
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 07:14:20 +00:00
|
|
|
}
|
|
|
|
|
2017-10-08 08:41:12 +00:00
|
|
|
/// Indicate that the next call to `build` should update all symbols in
|
|
|
|
/// the archive (equivalent to running 'ar s' over it).
|
2019-03-30 11:59:40 +00:00
|
|
|
fn update_symbols(&mut self) {
|
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 07:14:20 +00:00
|
|
|
self.should_update_symbols = true;
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Combine the provided files, rlibs, and native libraries into a single
|
|
|
|
/// `Archive`.
|
2019-03-30 11:59:40 +00:00
|
|
|
fn build(mut self) {
|
2019-12-22 22:42:04 +00:00
|
|
|
let kind = self.llvm_archive_kind().unwrap_or_else(|kind| {
|
|
|
|
self.config.sess.fatal(&format!("Don't know how to build archive of type: {}", kind))
|
|
|
|
});
|
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 07:14:20 +00:00
|
|
|
|
2016-06-10 21:27:19 +00:00
|
|
|
if let Err(e) = self.build_with_llvm(kind) {
|
|
|
|
self.config.sess.fatal(&format!("failed to build archive: {}", e));
|
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 07:14:20 +00:00
|
|
|
}
|
|
|
|
}
|
2021-03-08 20:42:54 +00:00
|
|
|
|
|
|
|
fn inject_dll_import_lib(
|
|
|
|
&mut self,
|
|
|
|
lib_name: &str,
|
|
|
|
dll_imports: &[DllImport],
|
|
|
|
tmpdir: &MaybeTempDir,
|
|
|
|
) {
|
|
|
|
let output_path = {
|
|
|
|
let mut output_path: PathBuf = tmpdir.as_ref().to_path_buf();
|
|
|
|
output_path.push(format!("{}_imports", lib_name));
|
|
|
|
output_path.with_extension("lib")
|
|
|
|
};
|
|
|
|
|
|
|
|
// we've checked for \0 characters in the library name already
|
|
|
|
let dll_name_z = CString::new(lib_name).unwrap();
|
|
|
|
// All import names are Rust identifiers and therefore cannot contain \0 characters.
|
|
|
|
// FIXME: when support for #[link_name] implemented, ensure that import.name values don't
|
|
|
|
// have any \0 characters
|
|
|
|
let import_name_vector: Vec<CString> = dll_imports
|
|
|
|
.iter()
|
2021-06-08 20:56:06 +00:00
|
|
|
.map(|import: &DllImport| {
|
|
|
|
if self.config.sess.target.arch == "x86" {
|
|
|
|
LlvmArchiveBuilder::i686_decorated_name(import)
|
|
|
|
} else {
|
|
|
|
CString::new(import.name.to_string()).unwrap()
|
|
|
|
}
|
2021-03-08 20:42:54 +00:00
|
|
|
})
|
|
|
|
.collect();
|
|
|
|
|
|
|
|
let output_path_z = rustc_fs_util::path_to_c_string(&output_path);
|
|
|
|
|
|
|
|
tracing::trace!("invoking LLVMRustWriteImportLibrary");
|
|
|
|
tracing::trace!(" dll_name {:#?}", dll_name_z);
|
|
|
|
tracing::trace!(" output_path {}", output_path.display());
|
|
|
|
tracing::trace!(
|
|
|
|
" import names: {}",
|
|
|
|
dll_imports.iter().map(|import| import.name.to_string()).collect::<Vec<_>>().join(", "),
|
|
|
|
);
|
|
|
|
|
|
|
|
let ffi_exports: Vec<LLVMRustCOFFShortExport> = import_name_vector
|
|
|
|
.iter()
|
|
|
|
.map(|name_z| LLVMRustCOFFShortExport::from_name(name_z.as_ptr()))
|
|
|
|
.collect();
|
|
|
|
let result = unsafe {
|
|
|
|
crate::llvm::LLVMRustWriteImportLibrary(
|
|
|
|
dll_name_z.as_ptr(),
|
|
|
|
output_path_z.as_ptr(),
|
|
|
|
ffi_exports.as_ptr(),
|
|
|
|
ffi_exports.len(),
|
|
|
|
llvm_machine_type(&self.config.sess.target.arch) as u16,
|
|
|
|
!self.config.sess.target.is_like_msvc,
|
|
|
|
)
|
|
|
|
};
|
|
|
|
|
|
|
|
if result == crate::llvm::LLVMRustResult::Failure {
|
|
|
|
self.config.sess.fatal(&format!(
|
|
|
|
"Error creating import library for {}: {}",
|
|
|
|
lib_name,
|
|
|
|
llvm::last_error().unwrap_or("unknown LLVM error".to_string())
|
|
|
|
));
|
|
|
|
}
|
|
|
|
|
|
|
|
self.add_archive(&output_path, |_| false).unwrap_or_else(|e| {
|
|
|
|
self.config.sess.fatal(&format!(
|
|
|
|
"failed to add native library {}: {}",
|
|
|
|
output_path.display(),
|
|
|
|
e
|
|
|
|
));
|
|
|
|
});
|
|
|
|
}
|
2019-03-30 11:59:40 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
impl<'a> LlvmArchiveBuilder<'a> {
|
|
|
|
fn src_archive(&mut self) -> Option<&ArchiveRO> {
|
|
|
|
if let Some(ref a) = self.src_archive {
|
2019-12-22 22:42:04 +00:00
|
|
|
return a.as_ref();
|
2019-03-30 11:59:40 +00:00
|
|
|
}
|
|
|
|
let src = self.config.src.as_ref()?;
|
|
|
|
self.src_archive = Some(ArchiveRO::open(src).ok());
|
|
|
|
self.src_archive.as_ref().unwrap().as_ref()
|
|
|
|
}
|
|
|
|
|
2019-12-22 22:42:04 +00:00
|
|
|
fn add_archive<F>(&mut self, archive: &Path, skip: F) -> io::Result<()>
|
|
|
|
where
|
|
|
|
F: FnMut(&str) -> bool + 'static,
|
2019-03-30 11:59:40 +00:00
|
|
|
{
|
2019-07-02 00:28:10 +00:00
|
|
|
let archive_ro = match ArchiveRO::open(archive) {
|
2019-03-30 11:59:40 +00:00
|
|
|
Ok(ar) => ar,
|
|
|
|
Err(e) => return Err(io::Error::new(io::ErrorKind::Other, e)),
|
|
|
|
};
|
2019-07-02 00:28:10 +00:00
|
|
|
if self.additions.iter().any(|ar| ar.path() == archive) {
|
2019-12-22 22:42:04 +00:00
|
|
|
return Ok(());
|
2019-07-02 00:28:10 +00:00
|
|
|
}
|
2019-03-30 11:59:40 +00:00
|
|
|
self.additions.push(Addition::Archive {
|
2019-07-02 00:28:10 +00:00
|
|
|
path: archive.to_path_buf(),
|
|
|
|
archive: archive_ro,
|
2019-03-30 11:59:40 +00:00
|
|
|
skip: Box::new(skip),
|
|
|
|
});
|
|
|
|
Ok(())
|
|
|
|
}
|
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 07:14:20 +00:00
|
|
|
|
2016-06-10 21:27:19 +00:00
|
|
|
fn llvm_archive_kind(&self) -> Result<ArchiveKind, &str> {
|
2020-11-08 11:27:51 +00:00
|
|
|
let kind = &*self.config.sess.target.archive_format;
|
2016-06-10 21:27:19 +00:00
|
|
|
kind.parse().map_err(|_| kind)
|
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 07:14:20 +00:00
|
|
|
}
|
|
|
|
|
2015-07-16 07:11:09 +00:00
|
|
|
fn build_with_llvm(&mut self, kind: ArchiveKind) -> io::Result<()> {
|
2019-06-30 18:30:01 +00:00
|
|
|
let removals = mem::take(&mut self.removals);
|
|
|
|
let mut additions = mem::take(&mut self.additions);
|
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 07:14:20 +00:00
|
|
|
let mut strings = Vec::new();
|
|
|
|
let mut members = Vec::new();
|
2018-07-17 13:00:10 +00:00
|
|
|
|
|
|
|
let dst = CString::new(self.config.dst.to_str().unwrap())?;
|
|
|
|
let should_update_symbols = self.should_update_symbols;
|
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 07:14:20 +00:00
|
|
|
|
|
|
|
unsafe {
|
|
|
|
if let Some(archive) = self.src_archive() {
|
|
|
|
for child in archive.iter() {
|
2016-03-23 03:01:37 +00:00
|
|
|
let child = child.map_err(string_to_io_error)?;
|
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 07:14:20 +00:00
|
|
|
let child_name = match child.name() {
|
|
|
|
Some(s) => s,
|
|
|
|
None => continue,
|
|
|
|
};
|
|
|
|
if removals.iter().any(|r| r == child_name) {
|
2019-12-22 22:42:04 +00:00
|
|
|
continue;
|
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 07:14:20 +00:00
|
|
|
}
|
|
|
|
|
2016-03-23 03:01:37 +00:00
|
|
|
let name = CString::new(child_name)?;
|
2019-12-22 22:42:04 +00:00
|
|
|
members.push(llvm::LLVMRustArchiveMemberNew(
|
|
|
|
ptr::null(),
|
|
|
|
name.as_ptr(),
|
|
|
|
Some(child.raw),
|
|
|
|
));
|
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 07:14:20 +00:00
|
|
|
strings.push(name);
|
|
|
|
}
|
|
|
|
}
|
2018-07-17 13:00:10 +00:00
|
|
|
for addition in &mut additions {
|
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 07:14:20 +00:00
|
|
|
match addition {
|
|
|
|
Addition::File { path, name_in_archive } => {
|
2016-03-23 03:01:37 +00:00
|
|
|
let path = CString::new(path.to_str().unwrap())?;
|
2018-07-17 13:00:10 +00:00
|
|
|
let name = CString::new(name_in_archive.clone())?;
|
2019-12-22 22:42:04 +00:00
|
|
|
members.push(llvm::LLVMRustArchiveMemberNew(
|
|
|
|
path.as_ptr(),
|
|
|
|
name.as_ptr(),
|
|
|
|
None,
|
|
|
|
));
|
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 07:14:20 +00:00
|
|
|
strings.push(path);
|
|
|
|
strings.push(name);
|
|
|
|
}
|
2019-07-02 00:28:10 +00:00
|
|
|
Addition::Archive { archive, skip, .. } => {
|
2015-10-23 05:07:19 +00:00
|
|
|
for child in archive.iter() {
|
2016-03-23 03:01:37 +00:00
|
|
|
let child = child.map_err(string_to_io_error)?;
|
2015-10-23 05:07:19 +00:00
|
|
|
if !is_relevant_child(&child) {
|
2019-12-22 22:42:04 +00:00
|
|
|
continue;
|
2015-10-23 05:07:19 +00:00
|
|
|
}
|
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 07:14:20 +00:00
|
|
|
let child_name = child.name().unwrap();
|
2015-10-23 05:07:19 +00:00
|
|
|
if skip(child_name) {
|
2019-12-22 22:42:04 +00:00
|
|
|
continue;
|
2015-10-23 05:07:19 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
// It appears that LLVM's archive writer is a little
|
|
|
|
// buggy if the name we pass down isn't just the
|
|
|
|
// filename component, so chop that off here and
|
|
|
|
// pass it in.
|
|
|
|
//
|
|
|
|
// See LLVM bug 25877 for more info.
|
2019-12-22 22:42:04 +00:00
|
|
|
let child_name =
|
|
|
|
Path::new(child_name).file_name().unwrap().to_str().unwrap();
|
2016-03-23 03:01:37 +00:00
|
|
|
let name = CString::new(child_name)?;
|
2019-12-22 22:42:04 +00:00
|
|
|
let m = llvm::LLVMRustArchiveMemberNew(
|
|
|
|
ptr::null(),
|
|
|
|
name.as_ptr(),
|
|
|
|
Some(child.raw),
|
|
|
|
);
|
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 07:14:20 +00:00
|
|
|
members.push(m);
|
|
|
|
strings.push(name);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-12-22 22:42:04 +00:00
|
|
|
let r = llvm::LLVMRustWriteArchive(
|
|
|
|
dst.as_ptr(),
|
|
|
|
members.len() as libc::size_t,
|
|
|
|
members.as_ptr() as *const &_,
|
|
|
|
should_update_symbols,
|
|
|
|
kind,
|
|
|
|
);
|
2016-08-01 21:16:16 +00:00
|
|
|
let ret = if r.into_result().is_err() {
|
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 07:14:20 +00:00
|
|
|
let err = llvm::LLVMRustGetLastError();
|
|
|
|
let msg = if err.is_null() {
|
2018-10-06 09:42:14 +00:00
|
|
|
"failed to write archive".into()
|
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 07:14:20 +00:00
|
|
|
} else {
|
|
|
|
String::from_utf8_lossy(CStr::from_ptr(err).to_bytes())
|
|
|
|
};
|
|
|
|
Err(io::Error::new(io::ErrorKind::Other, msg))
|
|
|
|
} else {
|
|
|
|
Ok(())
|
|
|
|
};
|
|
|
|
for member in members {
|
|
|
|
llvm::LLVMRustArchiveMemberFree(member);
|
|
|
|
}
|
2018-10-06 09:46:46 +00:00
|
|
|
ret
|
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 07:14:20 +00:00
|
|
|
}
|
|
|
|
}
|
2021-06-08 20:56:06 +00:00
|
|
|
|
|
|
|
fn i686_decorated_name(import: &DllImport) -> CString {
|
|
|
|
let name = import.name;
|
|
|
|
// We verified during construction that `name` does not contain any NULL characters, so the
|
|
|
|
// conversion to CString is guaranteed to succeed.
|
|
|
|
CString::new(match import.calling_convention {
|
|
|
|
DllCallingConvention::C => format!("_{}", name),
|
|
|
|
DllCallingConvention::Stdcall(arg_list_size) => format!("_{}@{}", name, arg_list_size),
|
|
|
|
DllCallingConvention::Fastcall(arg_list_size) => format!("@{}@{}", name, arg_list_size),
|
|
|
|
DllCallingConvention::Vectorcall(arg_list_size) => {
|
|
|
|
format!("{}@@{}", name, arg_list_size)
|
|
|
|
}
|
|
|
|
})
|
|
|
|
.unwrap()
|
|
|
|
}
|
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 07:14:20 +00:00
|
|
|
}
|
2015-10-23 05:07:19 +00:00
|
|
|
|
2015-10-24 01:18:44 +00:00
|
|
|
fn string_to_io_error(s: String) -> io::Error {
|
2015-10-23 05:07:19 +00:00
|
|
|
io::Error::new(io::ErrorKind::Other, format!("bad archive: {}", s))
|
|
|
|
}
|