This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Use (void) in no-arguments function definitions
- From: "Joseph S. Myers" <joseph at codesourcery dot com>
- To: Roland McGrath <roland at hack dot frob dot com>
- Cc: <libc-alpha at sourceware dot org>
- Date: Sat, 8 Jun 2013 00:28:17 +0000
- Subject: Re: Use (void) in no-arguments function definitions
- References: <Pine dot LNX dot 4 dot 64 dot 1306072249350 dot 9729 at digraph dot polyomino dot org dot uk> <20130607231209 dot ACF692C078 at topped-with-meat dot com>
On Fri, 7 Jun 2013, Roland McGrath wrote:
> > Tested x86_64; the only differences in disassembly of glibc's .so
> > files are, as for the "register" patch, changed address offsets.
>
> I wonder why even those changed, since no non-code allocated sections (like
> .dynsym) should have changed either, just debug info (if that). I'm
> curious enough about that to hope you bother investigating, but not enough
> to investigate it myself, and not enough to give me pause about approving
> the change as is. I think it's fine.
Actually in this case it seems it isn't even offsets changing, just names
of internal symbols appearing in disassembly (e.g.
__PRETTY_FUNCTION__.9279 changing to __PRETTY_FUNCTION__.9280). In
general just about any change to the token sequence can change the
compiler's internal uids for various objects like that (e.g. affecting
declaration merging or what types get created in what order).
--
Joseph S. Myers
joseph@codesourcery.com