svn commit: r293226 - head/libexec/rtld-elf

[Warner Losh]

On Wed, Jan 6, 2016 at 9:10 AM, Konstantin Belousov <kostikbel at gmail.com>
wrote:

> On Tue, Jan 05, 2016 at 09:20:46PM +0000, Warner Losh wrote:
> > Author: imp
> > Date: Tue Jan  5 21:20:46 2016
> > New Revision: 293226
> > URL: https://svnweb.freebsd.org/changeset/base/293226
> >
> > Log:
> >   Disable abi variant hook until strangeness with packages can be sorted
> >   out.
> >
> > Modified:
> >   head/libexec/rtld-elf/rtld.c
> >
> > Modified: head/libexec/rtld-elf/rtld.c
> >
> ==============================================================================
> > --- head/libexec/rtld-elf/rtld.c      Tue Jan  5 21:12:49 2016
> (r293225)
> > +++ head/libexec/rtld-elf/rtld.c      Tue Jan  5 21:20:46 2016
> (r293226)
> > @@ -435,7 +435,7 @@ _rtld(Elf_Addr *sp, func_ptr_type *exit_
> >
> >      trust = !issetugid();
> >
> > -    md_abi_variant_hook(aux_info);
> > +/*  md_abi_variant_hook(aux_info); */
> >
> >      ld_bind_now = getenv(_LD("BIND_NOW"));
> >      /*
> The ARM hook resets the default paths to the libsoft variants.  It cannot
> work while /lib, /usr/lib and /usr/local/lib are populated with soft-fp
> libraries.  I thought that you are going to switch the default target_arch
> to armv6hf immediately after the commit to rtld.
>

That analysis isn't quite right. I'll throw the replacement switch very
soon.
The problem is that it highlighted was things were incomplete, not that
I'd left soft float libraries where I did for a short period of time.


> And then, looking into the issue, IMO the way forward is to leave softfp
> libraries where they are, but install v6hf somewhere in /lib/armv6hf etc.
>

No. That's even worse. We're not doing that.

Warner