fitzgen (he/him) (Jul 26 2021 at 17:01): fitzgen (he/him) (Jul 26 2021 at 17:01):Anyone know why we might not symbolicate stack traces when we otherwise "should"?
I'm setting config.wasm_backtrace_details(wasmtime::WasmBacktraceDetails::Enable).debug_info(true) and my wasm has both DWARF and name custom sections.
any ideas?
fitzgen (he/him) (Jul 26 2021 at 17:03):ah wait, I'm using module linking and my inner wasm module has those things, but my outer module does not. do we only look at the outer module for this data?
Alex Crichton (Jul 26 2021 at 17:22):in theory we should be handling each module independently
Alex Crichton (Jul 26 2021 at 17:22):but I wouldn't be surprised if there was some bug around this
fitzgen (he/him) (Jul 26 2021 at 17:34):hm I think this is unrelated to module linking; I'm getting a symbolication context, and it is returning Ok(frames), but the frames are empty
fitzgen (he/him) (Jul 26 2021 at 18:25):bah, the .debug_line tables are empty (the metadata is there that establishes filename <--> index mappings and sets up the opcodes and everything, but the programs taht generate the line tables are all empty)
fitzgen (he/him) (Jul 26 2021 at 18:26):and it is like this coming out of wasi-sdk. maybe wasm-opt messing things up here
Last updated: Oct 23 2024 at 20:03 UTC