Re: [PATCH 14/17 v5] Avoid stack-protecting signal-handling functions sibcalled from assembly.

[Florian Weimer <fweimer at redhat dot com>]

On 05/15/2016 05:49 PM, Andreas Schwab wrote:
> Nix <nix@esperi.org.uk> writes:
>
> > On 15 May 2016, Andreas Schwab verbalised:
> >
> > > Nix <nix@esperi.org.uk> writes:
> > >
> > > > I'll have a hunt for whatever it is that's causing sigjmp.c to not be
> > > > rebuilt after this test cycle, because that's the underlying bug here,
> > > > really.
> > >
> > > Just changing CFLAGS-foo doesn't trigger a rebuild of dependent objects.
> >
> > Well, no, but sigjmp.c should be rebuilt if it's incorporated in ld.so
> > anyway: we should be getting an rtld-sigjmp.o, and we're not.
>
> Most architectures implement setjmp in assembler anyway.

But a lot (all?) of them use a C tail for some of the work, and the 
patch is about how that is compiled.

Florian