debuginfo: Debugger pretty printers cannot handle all enum memory layouts

[michaelwoerister]

Since @luqmana's cool enum memory layout optimizations landed, GDB and LLDB pretty printers are slightly out of date, i.e. they cannot handle cases where the discriminant isn't stored at the top-level. The necessary data is already encoded in DWARF, but the pretty printers don't use it yet.

[cactorium]

Mind if I take a stab at this?

[cactorium]

Okay, so I'm having trouble figuring out how to load the pretty printers into gdb, and it doesn't seem to be documented in any of the Rust docs. Any advice you can give me? Sorry!

[michaelwoerister]

Whoops, sorry, I meant to respond to this earlier.
Assuming that you are using a current version of rustc, the easiest way to do that is to use the rust-gdb script. It will use the pretty printers coming with the rustc that is in your PATH. So, if you call...

LD_LIBRARY_PATH=<your-rust-directory>/x86_64-unknown-linux-gnu/stage1/lib \
PATH=<your-rust-directory>/x86_64-unknown-linux-gnu/stage1/bin:$PATH \
rust-gdb <regular gdb args>

... then GDB will be started with the pretty printers located in <your-rust-directory>/x86_64-unknown-linux-gnu/stage1/lib/rustlib. Note though that these are copies of the 'real' pretty printer source files in src/etc. You might want to modify the files in stage1/lib/rustlib for quick testing but beware of losing those modifications when rebuilding rustc because that will overwrite the files in stage1/lib/rustlib with those in src/etc again.

Don't hesitate to post any further questions here as they arise :)

[cactorium]

Ah, that is exactly what I needed! Thank you very much :)

[cactorium]

I think this can be closed now?

[alexcrichton]

Thanks @cactorium!