On Wed, Feb 05, 2025 at 01:46:02PM +0100, Richard Biener wrote:
> > Please let me know any comments on this?
>
> I think I've seen this elsewhere -- the issue is the unwind register API does
> not allow for failures but I also think calling abort() is bad.
>
> Are you calling this from a JIT context or so? We're assuming that at program
> start malloc() will not fail.
>
> The proper solution is probably to add an alternate ABI which has a way to
> fail
> during registration.
Note, e.g. the __register_frame_info_table_bases changes look just wrong.
This function is normally called by code created by collect2.cc and
that actually guarantees that it is only called if the table is not empty.
If it is some JIT and it is trying to register something that is empty or
NULL, just don't call that.
Jakub
