This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: Committed, MMIX: Fix valgrind-diagnosed elf64-mmix.c bugs
- From: Hans-Peter Nilsson <hp at bitrange dot com>
- To: Alan Modra <amodra at bigpond dot net dot au>
- Cc: binutils at sources dot redhat dot com
- Date: Sun, 17 Nov 2002 18:49:47 -0500 (EST)
- Subject: Re: Committed, MMIX: Fix valgrind-diagnosed elf64-mmix.c bugs
On Mon, 18 Nov 2002, Alan Modra wrote:
> On Wed, Nov 13, 2002 at 08:33:56PM -0500, Hans-Peter Nilsson wrote:
> > + /* We need to have a _raw_size contents even though there's only
> > + _cooked_size worth of data, since the generic relocation machinery
> > + will allocate and copy that much temporarily. */
>
> Hmm, so should you really be fixing the "generic relocation machinery"
> instead?
No, this is before relocation (after relaxation), so _raw_size
is the right field by definition. See the comment at the field
definitions: field _cooked_size isn't valid until after
relocation. Field _raw_size is the one that may be tweaked by
relaxation. (I guess I should fix the comment above, sorry.)
Well by *current* definition. Maybe we should change the
definition? If relocation shrinks a section then it sounds to
me like something that should happen at relaxation. Perhaps
that was my confusion. The comment at _raw_size seems
confusing: it's "the original size on disk" and may be bigger
after relaxation?
brgds, H-P