Stream: git-wasmtime

Topic: wasmtime / issue #6960 Support referencing stack slots in...


view this post on Zulip Wasmtime GitHub notifications bot (Sep 05 2023 at 20:35):

alexcrichton commented on issue #6960:

Oh interesting! So there's no way for the dwarf sections to actually refer to .eh_frame? It's required at the binary level that the information is duplicated?

view this post on Zulip Wasmtime GitHub notifications bot (Sep 05 2023 at 20:48):

SingleAccretion commented on issue #6960:

So there's no way for the dwarf sections to actually refer to .eh_frame? It's required at the binary level that the information is duplicated?

I am no great Unix expert myself, so do not know for certain. I _suspect_ that it is indeed possible for the debugger to use .eh_frame alone, though the two sections are a little different in format details as well as usage (.eh_frame is like .pdata on Windows in that it is "mandatory").

FWIW, there are much larger (pun intended) size problems with -g-produced object files at the moment. For one, all the location lists for variables with are very fragmented, it would help a lot if they could always refer to the one and only stack location, at least for unoptimized code.

view this post on Zulip Wasmtime GitHub notifications bot (Sep 05 2023 at 21:04):

bjorn3 commented on issue #6960:

Rustc doesn't ever emit .debug_frame. The debugger has to use .eh_frame for this information for rust code.


Last updated: Nov 22 2024 at 17:03 UTC