This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [PATCH] Fix PR ld/17277: bogus dynamic relocs and TEXTREL for ARM PC-relative relocs
- From: Alan Modra <amodra at gmail dot com>
- To: Roland McGrath <mcgrathr at google dot com>
- Cc: "binutils at sourceware dot org" <binutils at sourceware dot org>
- Date: Wed, 20 Aug 2014 22:40:13 +0930
- Subject: Re: [PATCH] Fix PR ld/17277: bogus dynamic relocs and TEXTREL for ARM PC-relative relocs
- Authentication-results: sourceware.org; auth=none
- References: <CAB=4xhqoJPa-v25zJok+Ajo9BaeoUKNFt0DhBaB3PzyDn1gU_g at mail dot gmail dot com> <CAB=4xhp+zi+MNkkUTjNB6K+RmtJBPUC8yMq-V__GLpT3aekSBQ at mail dot gmail dot com> <20140817233035 dot GW7047 at bubble dot grove dot modra dot org> <CAB=4xho1-Hi35vFC49+VvGfVeh2LxtzZgOmiAU8qyy85HKmn_Q at mail dot gmail dot com>
On Tue, Aug 19, 2014 at 01:52:28PM -0700, Roland McGrath wrote:
> On Sun, Aug 17, 2014 at 4:30 PM, Alan Modra <amodra@gmail.com> wrote:
> > It seems to me the problem here is that may_become_dynamic_p should
> > not be set for pc_relative relocations against a local symbol.
>
> Thanks for the pointer. That makes lots of sense: the tests controlling
> that were the other places that had been explicitly checking for just
> R_ARM_REL32{,_NOI} and it seems right that all those tests should be
> consistent (now all changed to check howto->pc_relative).
>
> I've extended my test case to cover the local-symbol case as well. The new
> version of the change passes that new test and still has no regressions.
> Likewise for the 2.24 backport.
>
> OK for trunk and 2.24 now?
Looks like a step in the right direction. OK for trunk, and 2.24 in a
few days if there are no objections.
--
Alan Modra
Australia Development Lab, IBM