Re: Simple but crucial bug fix to gdb

[Charlie Mills <cmills@synopsys.com>]

At 02:25 PM 6/1/01 -0400, Fernando Nasser wrote:
>Daniel Berlin wrote:
>> 
>> Michael Snyder <msnyder@cygnus.com> writes:
>> 
>> > "Frank Ch. Eigler" wrote:
>> > >
>> > > Daniel Berlin <dan@cgsoftware.com> writes:
>> > >
>> > > : [...]
>> > > : However, we should *never* see a case where pst is NULL, and
>> > > : textlow_not_set is 1, at the point we see a function.
>> > > : [...]
>> > >
>> > > Would a gdb_assert() to this effect satisfy all sides?
>> >
>> > gdb_assert causes an abort if the conditional fails.
>> > I generally think it's better if the debugger doesn't abort
>> > (unles it's believed to be in an unrecoverable state).
>> 
>> To be honest, i'd consider it an unrecoverable state.
>> This is because if the compiler is producing such broken debug info
>> that we see functions outside of where we should, it's likely your
>> debug info is so screwed up as to be worthless, and just cause you to
>> think GDB is broken.
>> 
>
>As I understand from Charles Mills message, he can use the debugger if it
is allowed to proceed.  So things are not that bad.
>
>Charles, can you confirm that?  Thanks.
>
>
>-- 
>Fernando Nasser
>Red Hat - Toronto                       E-Mail:  fnasser@redhat.com
>2323 Yonge Street, Suite #300
>Toronto, Ontario   M4P 2C9
>
>

Well, yes and no!  I can and do use gdb with my simple
patch, it's certainly a lot better than a crash,
but it is true that gdb then cannot find line numbers in
*some* of my functions, just as Jim Blandy surmises, so I am
still going to have to figure out how to fix my executable to
be more friendly to gdb.

You guys are excellent!  Thanks.

-- Charlie Mills
   cmills@synopsys.com  ..work
   mills@q7.com         ..personal
   (503)748-2665        ..at work today