gdb.mi/mi-cli.exp failures

[Ian Lance Taylor]

Daniel Jacobowitz <drow@mvista.com> writes:

> Note that this still isn't ideal, because the symbol table is not
> clearly cached in the BFD anywhere; so we'll get a new one each time we
> relocate a section.  What we really need is to cache the canonicalized
> symbol table, I suppose.

Can we use the existing bfd_free_cached_info() interface in some way?
That function is probably only implemented for a.out, but it could be
implemented for other targets as well.

Actually, I see one issue, which is that calling this function will
leave information lying around in memory which you might prefer to
free.  And there is the converse issue, which is that every time you
call this function you need to canonicalize the symbol table, and that
is a pain if you call it a lot for the same BFD.  You can solve one or
the other issue, but I don't think you can solve both, since for most
targets the canonical symbol table will point into the internal symbol
information.

Ian