NetBSD: backtrace support broken

[0-wiz-0]

I tried this code:

fn main() { panic!("explode!") }                                                                                                                

I expected to see this happen:

$ rustc t.rs
$ ./t
thread 'main' panicked at 'explode!', t.rs:1:13
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
$ RUST_BACKTRACE=1 ./t
thread 'main' panicked at 'explode!', t.rs:1:13
stack backtrace:
   0: std::panicking::begin_panic
   1: t::main
   2: core::ops::function::FnOnce::call_once
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.

(provided by a friendly Linux user)

Instead, this happened (on NetBSD)

I see this instead:

$ ./t    
thread 'main' panicked at 'explode!', t.rs:1:13
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
$ RUST_BACKTRACE=1 ./t
thread 'main' panicked at 'explode!', t.rs:1:13
stack backtrace:
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
$ RUST_BACKTRACE=full ./t
thread 'main' panicked at 'explode!', t.rs:1:13
stack backtrace:
   0:         0x6dc13726 - <unknown>
   1:         0x6dc3790c - <unknown>
   2:         0x6dc25ea5 - <unknown>
   3:         0x6dc0c981 - <unknown>
   4:         0x6dc0c581 - <unknown>
   5:         0x6dc0ced1 - <unknown>
   6:         0x6dc07a9e - <unknown>
   7:         0x6dc07d85 - <unknown>
   8:         0x6dc07a22 - <unknown>
   9:         0x6dc07e79 - <unknown>
  10:         0x6dc07f63 - <unknown>
  11:         0x6dc07d29 - <unknown>
  12:         0x6dc08749 - <unknown>
  13:         0x6dc209a1 - <unknown>
  14:         0x6dc08731 - <unknown>
  15:         0x6dc07f03 - <unknown>
  16:         0x6dc05b7d - <unknown>

Meta

rustc --version --verbose:

rustc 1.59.0
binary: rustc
commit-hash: unknown
commit-date: unknown
host: x86_64-unknown-netbsd
release: 1.59.0
LLVM version: 13.0.1

Backtrace

Yes, exactly that's the problem.

On the C level, I know that Linux provides the backtrace() family of functions in libc.
NetBSD provides all the functions too, but you need to link against libexecinfo, see
https://man.netbsd.org/backtrace.3

So I guess it's probably not too hard to add, I hope someone can take this on.

[ijackson]

@rustbot modify labels +O-netbsd +A-runtime

[0-wiz-0]

Patches fixing this now exist:
https://mail-index.netbsd.org/pkgsrc-changes/2022/05/18/msg254732.html
thanks to Håvard Eidnes. What's the best way to get them integrated?

[workingjubilee]

If you submit the backtrace stuff to https://github.com/rust-lang/backtrace-rs then I will try to get it merged.

[workingjubilee]

Oh, it was submitted to backtrace in rust-lang/backtrace-rs#512

Is there anything left?

[0-wiz-0]

Thank you for looking at this, @workingjubilee !
I've just tried 0.3.73 on NetBSD-10.99.11/amd64, and it seemed to work fine, here's the output:

backtrace-rs-0.3.73> RUST_BACKTRACE=1 cargo test
   Compiling backtrace v0.3.73 (/home/wiz/a/backtrace-rs-0.3.73)
    Finished `test` profile [unoptimized + debuginfo] target(s) in 0.56s
     Running unittests src/lib.rs (target/debug/deps/backtrace-ed5435a466558a04)

running 3 tests
test symbolize::gimli::parse_running_mmaps::check_maps_entry_parsing_32bit ... ok
test symbolize::gimli::parse_running_mmaps::check_maps_entry_parsing_64bit ... ok
test capture::tests::test_frame_conversion ... ok

test result: ok. 3 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.03s

     Running tests/accuracy/main.rs (target/debug/deps/accuracy-b042c1723a62a800)

running 0 tests

test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 0.00s

     Running tests/concurrent-panics.rs (target/debug/deps/concurrent_panics-2be74b7ee972732b)
test result: ok
     Running tests/current-exe-mismatch.rs (target/debug/deps/current_exe_mismatch-e7ad37dd6f4cbaa7)
stdout:

stderr:

code: signal: 11 (SIGSEGV) (core dumped)
thread 'main' panicked at tests/current-exe-mismatch.rs:72:5:
explicit panic
stack backtrace:
   0: rust_begin_unwind
   1: core::panicking::panic_fmt
   2: core::panicking::panic_display
   3: core::panicking::panic_explicit
   4: current_exe_mismatch::parent::panic_cold_explicit
             at /scratch/lang/rust/work/rustc-1.79.0-src/library/core/src/panic.rs:87:13
   5: current_exe_mismatch::parent
             at ./tests/current-exe-mismatch.rs:72:5
   6: current_exe_mismatch::main
             at ./tests/current-exe-mismatch.rs:15:15
   7: core::ops::function::FnOnce::call_once
             at /scratch/lang/rust/work/rustc-1.79.0-src/library/core/src/ops/function.rs:250:5
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
error: test failed, to rerun pass `--test current-exe-mismatch`

I also tried my original test program with rust 1.79:

> RUST_BACKTRACE=1 cargo run
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.02s
     Running `target/debug/rust-backtracetest`
thread 'main' panicked at src/main.rs:1:13:
explode!
stack backtrace:
   0: rust_begin_unwind
   1: core::panicking::panic_fmt
   2: rust_backtracetest::main
             at ./src/main.rs:1:13
   3: core::ops::function::FnOnce::call_once
             at /scratch/lang/rust/work/rustc-1.79.0-src/library/core/src/ops/function.rs:250:5
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.

That's what it should look like, right?
If yes, please feel free to close this! Thanks.

[workingjubilee]

SIGSEGV

uh

that seems incorrect.

[0-wiz-0]

Oh, true. I was just happy that the backtrace had function names instead of <unknown>

[0-wiz-0]

Should I file a new ticket with backtrace-rs about the segfault?

[workingjubilee]

Yeees, let's do that and call this issue good.