This is the mail archive of the
mailing list for the binutils project.
Re: Change ld "notice" interface for better handling of indirect symbols
- From: Alan Modra <amodra at gmail dot com>
- To: Kyle McMartin <kmcmarti at redhat dot com>
- Cc: binutils at sourceware dot org
- Date: Thu, 28 Aug 2014 11:41:19 +0930
- Subject: Re: Change ld "notice" interface for better handling of indirect symbols
- Authentication-results: sourceware.org; auth=none
- References: <20140812123911 dot GJ7047 at bubble dot grove dot modra dot org> <20140827140213 dot GD23212 at redacted dot bos dot redhat dot com>
On Wed, Aug 27, 2014 at 10:02:13AM -0400, Kyle McMartin wrote:
> On Tue, Aug 12, 2014 at 10:09:11PM +0930, Alan Modra wrote:
> > The main aim of this change was to have non_ir_ref set correctly on
> > new indirect symbols. I could have added a "copy" param to the "notice"
> > function, so that indirect symbols could be created in plugin_notice,
> > but it seemed cleaner to create indirect syms earlier and pass them
> > rather than "string" to "notice".
> Happily, this seems to resolve an issue building with LTO and ld.bfd...
> We had a bug report where building systemd on aarch64 and s390 were
> failing with multiply defined symbols (everyone else has gold ported,
> which was, for some as yet undebugged by me reason, didn't do this.)
That's nice to know. I'd been looking at LTO and ld.bfd problems in
general, and warning symbols specifically (PR16746) when I saw the
potential problem with non_ir_ref and lack of following a warning sym
link in plugin_notice. So I guess it isn't surprising that this fixed
Australia Development Lab, IBM