Skipping too many frames

[smessmer]

I set up a sample panic somewhere in my code base, ran with cargo run --release, and caught it with the following custom panic handler:

pub fn show_backtrace_on_panic() {
    std::panic::set_hook(Box::new(|panic_info| {
        let backtrace = std::backtrace::Backtrace::force_capture();
        eprintln!("{panic_info}");
        eprintln!("\nBacktrace:\n{backtrace}");
    }));
}

Which showed me this backtrace:

Backtrace:
   0: cryfs_cli_utils::panic_handler::show_backtrace_on_panic::{{closure}}
   1: std::panicking::rust_panic_with_hook
   2: std::panicking::begin_panic_handler::{{closure}}
   3: std::sys_common::backtrace::__rust_end_short_backtrace
   4: rust_begin_unwind
   5: core::panicking::panic_fmt
   6: cryfs_cryfs::config::encryption::decrypt
   7: cryfs_cryfs::config::loader::load_or_create
   8: <cryfs_cli::cli::Cli as cryfs_cli_utils::application::Application>::main
   9: cryfs_cli_utils::application::run
  10: std::sys_common::backtrace::__rust_begin_short_backtrace
  11: main
  12: __libc_start_call_main
             at ./csu/../sysdeps/nptl/libc_start_call_main.h:58:16
  13: __libc_start_main_impl
             at ./csu/../csu/libc-start.c:360:3
  14: _start

Frames 0..=5 are useless here, but starting from frame 6, the frames are relevant.

However, using human-panic, I got this backtrace, which skips frame 6:

"backtrace" = """
   0: 0x57a214f25d51 - cryfs_cryfs::config::loader::load_or_create::h0164e15f3c7e3b2f
   1: 0x57a214eba0f8 - <cryfs_cli::cli::Cli as cryfs_cli_utils::application::Application>::main::h857c34904cb9b5d6
   2: 0x57a214d42d65 - cryfs_cli_utils::application::run::h524b32ee2ef4b9b3
   3: 0x57a214d399c3 - std::sys_common::backtrace::__rust_begin_short_backtrace::ha12c550ae26c7977
   4: 0x57a214d3a260 - main
   5: 0x79c6fac2a1ca - __libc_start_call_main
                at ./csu/../sysdeps/nptl/libc_start_call_main.h:58
   6: 0x79c6fac2a28b - __libc_start_main_impl
                at ./csu/../csu/libc-start.c:360
   7: 0x57a214c81455 - _start
   8:        0x0 - <unresolved>"""

Seems like it's skipping one frame too many.

[epage]

158 made improvements to what gets skipped and has been released in v2.0.2. Could you test it out?