1
Fork 0
rust/src/libstd/sys_common/backtrace.rs

189 lines
5.5 KiB
Rust
Raw Normal View History

/// Common code for printing the backtrace in the same way across the different
/// supported platforms.
2019-02-11 04:23:21 +09:00
use crate::env;
use crate::fmt;
2019-02-11 04:23:21 +09:00
use crate::io;
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
use crate::io::prelude::*;
2019-02-11 04:23:21 +09:00
use crate::path::{self, Path};
use crate::sync::atomic::{self, Ordering};
use crate::sys::mutex::Mutex;
use backtrace::{BacktraceFmt, BytesOrWideString, PrintFmt};
/// Max number of frames to print.
const MAX_NB_FRAMES: usize = 100;
/// Prints the current backtrace.
pub fn print(w: &mut dyn Write, format: PrintFmt) -> io::Result<()> {
static LOCK: Mutex = Mutex::new();
// There are issues currently linking libbacktrace into tests, and in
// general during libstd's own unit tests we're not testing this path. In
// test mode immediately return here to optimize away any references to the
// libbacktrace symbols
if cfg!(test) {
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
return Ok(());
}
// Use a lock to prevent mixed output in multithreading context.
// Some platforms also requires it, like `SymFromAddr` on Windows.
unsafe {
LOCK.lock();
let res = _print(w, format);
LOCK.unlock();
res
}
}
fn _print(w: &mut dyn Write, format: PrintFmt) -> io::Result<()> {
struct DisplayBacktrace {
format: PrintFmt,
}
impl fmt::Display for DisplayBacktrace {
fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result {
_print_fmt(fmt, self.format)
}
}
write!(w, "{}", DisplayBacktrace { format })
}
fn _print_fmt(fmt: &mut fmt::Formatter<'_>, print_fmt: PrintFmt) -> fmt::Result {
let mut print_path = move |fmt: &mut fmt::Formatter<'_>, bows: BytesOrWideString<'_>| {
output_filename(fmt, bows, print_fmt)
};
let mut bt_fmt = BacktraceFmt::new(fmt, print_fmt, &mut print_path);
bt_fmt.add_context()?;
let mut skipped = false;
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
unsafe {
let mut idx = 0;
let mut res = Ok(());
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
backtrace::trace_unsynchronized(|frame| {
if print_fmt == PrintFmt::Short && idx > MAX_NB_FRAMES {
skipped = true;
return false;
}
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
let mut hit = false;
let mut stop = false;
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
backtrace::resolve_frame_unsynchronized(frame, |symbol| {
hit = true;
if print_fmt == PrintFmt::Short {
if let Some(sym) = symbol.name().and_then(|s| s.as_str()) {
if sym.contains("__rust_begin_short_backtrace") {
skipped = true;
stop = true;
return;
}
}
}
res = bt_fmt.frame().symbol(frame, symbol);
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
});
if stop {
return false;
}
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
if !hit {
res = bt_fmt.frame().print_raw(frame.ip(), None, None, None);
}
idx += 1;
res.is_ok()
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
});
res?;
}
bt_fmt.finish()?;
if skipped {
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
writeln!(
fmt,
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
"note: Some details are omitted, \
run with `RUST_BACKTRACE=full` for a verbose backtrace."
)?;
}
Ok(())
}
/// Fixed frame used to clean the backtrace with `RUST_BACKTRACE=1`.
#[inline(never)]
pub fn __rust_begin_short_backtrace<F, T>(f: F) -> T
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
where
F: FnOnce() -> T,
F: Send,
T: Send,
{
f()
}
// For now logging is turned off by default, and this function checks to see
// whether the magical environment variable is present to see if it's turned on.
pub fn log_enabled() -> Option<PrintFmt> {
static ENABLED: atomic::AtomicIsize = atomic::AtomicIsize::new(0);
match ENABLED.load(Ordering::SeqCst) {
0 => {}
1 => return None,
2 => return Some(PrintFmt::Short),
_ => return Some(PrintFmt::Full),
}
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
let val = env::var_os("RUST_BACKTRACE").and_then(|x| {
if &x == "0" {
None
} else if &x == "full" {
Some(PrintFmt::Full)
} else {
Some(PrintFmt::Short)
}
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
});
ENABLED.store(
match val {
Some(v) => v as isize,
None => 1,
},
Ordering::SeqCst,
);
val
}
/// Prints the filename of the backtrace frame.
///
/// See also `output`.
fn output_filename(
fmt: &mut fmt::Formatter<'_>,
bows: BytesOrWideString<'_>,
print_fmt: PrintFmt,
) -> fmt::Result {
#[cfg(windows)]
let path_buf;
let file = match bows {
#[cfg(unix)]
BytesOrWideString::Bytes(bytes) => {
use crate::os::unix::prelude::*;
Path::new(crate::ffi::OsStr::from_bytes(bytes))
}
#[cfg(not(unix))]
BytesOrWideString::Bytes(bytes) => {
Path::new(crate::str::from_utf8(bytes).unwrap_or("<unknown>"))
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
}
#[cfg(windows)]
BytesOrWideString::Wide(wide) => {
use crate::os::windows::prelude::*;
path_buf = crate::ffi::OsString::from_wide(wide);
Path::new(&path_buf)
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
}
#[cfg(not(windows))]
BytesOrWideString::Wide(_wide) => {
Path::new("<unknown>")
}
};
if print_fmt == PrintFmt::Short && file.is_absolute() {
if let Ok(cwd) = env::current_dir() {
if let Ok(stripped) = file.strip_prefix(&cwd) {
if let Some(s) = stripped.to_str() {
return write!(fmt, ".{}{}", path::MAIN_SEPARATOR, s);
std: Depend on `backtrace` crate from crates.io This commit removes all in-tree support for generating backtraces in favor of depending on the `backtrace` crate on crates.io. This resolves a very longstanding piece of duplication where the standard library has long contained the ability to generate a backtrace on panics, but the code was later extracted and duplicated on crates.io with the `backtrace` crate. Since that fork each implementation has seen various improvements one way or another, but typically `backtrace`-the-crate has lagged behind libstd in one way or another. The goal here is to remove this duplication of a fairly critical piece of code and ensure that there's only one source of truth for generating backtraces between the standard library and the crate on crates.io. Recently I've been working to bring the `backtrace` crate on crates.io up to speed with the support in the standard library which includes: * Support for `StackWalkEx` on MSVC to recover inline frames with debuginfo. * Using `libbacktrace` by default on MinGW targets. * Supporting `libbacktrace` on OSX as an option. * Ensuring all the requisite support in `backtrace`-the-crate compiles with `#![no_std]`. * Updating the `libbacktrace` implementation in `backtrace`-the-crate to initialize the global state with the correct filename where necessary. After reviewing the code in libstd the `backtrace` crate should be at exact feature parity with libstd today. The backtraces generated should have the same symbols and same number of frames in general, and there's not known divergence from libstd currently. Note that one major difference between libstd's backtrace support and the `backtrace` crate is that on OSX the crates.io crate enables the `coresymbolication` feature by default. This feature, however, uses private internal APIs that aren't published for OSX. While they provide more accurate backtraces this isn't appropriate for libstd distributed as a binary, so libstd's dependency on the `backtrace` crate explicitly disables this feature and forces OSX to use `libbacktrace` as a symbolication strategy. The long-term goal of this refactoring is to eventually move us towards a world where we can drop `libbacktrace` entirely and simply use Gimli and the surrounding crates for backtrace support. That's still aways off but hopefully will much more easily enabled by having the source of truth for backtraces live in crates.io! Procedurally if we go forward with this I'd like to transfer the `backtrace-rs` crate to the rust-lang GitHub organization as well, but I figured I'd hold off on that until we get closer to merging.
2019-05-15 07:30:15 -07:00
}
}
}
}
fmt::Display::fmt(&file.display(), fmt)
}