Subject: Re: Kernel support for ELF-format core files
To: None <firstname.lastname@example.org>
From: Andrew Cagney <email@example.com>
Date: 12/11/2001 10:32:07
> On Mon, Dec 10, 2001 at 08:51:11AM -0800, Andrew Cagney wrote:
> > Ah, thanks, sorry didn't realize this. Might be time to teach GDB a
> > little bit more about core files.
> Err, personally, I think the current arrangement (libbfd doing the
> section work) is good -- it allows you to pull them apart with e.g.
> objcopy :-)
> GDB really should be working with "bfd sections", which do not secessarily
> map 1:1 with actual sections in the file. The .reg and .reg2 bfd sections
> are actually in one file section, even in Solaris.
True. The abstraction is fine. Sometimes, though, it can at least
appear to get in the way - for instance when trying to implement code
that writes core files. That is outside the scope of this discussion
> > > Yes, I know. Those pseudo-sections are put together by libbfd. The
> > > number after the / is actually ((lwpid << 16) | pid).
> > > > Hmm, definitly time to do this.
> Err... gdb already deals with this properly. As far as I can tell, anyway.
Back in '93 a 64k process table was considered huge, now, well, you know
:-) GDB can handle very large process id's but I'm not so sure about
some of the interfaces.