Re: Fix ____longjmp_chk for ppc(64)

[Richard Henderson <rth at twiddle dot net>]

On 06/24/2009 08:46 AM, Andreas Schwab wrote:
> Richard Henderson<rth@twiddle.net>  writes:
>
> > On 06/24/2009 05:27 AM, Andreas Schwab wrote:
> > > +	mflr	r0;					\
> > > +	std	r0,16(r1);				\
> > > +	stdu	r1,-112(r1);				\
> > > +	cfi_adjust_cfa_offset (112);			\
> > > +	cfi_offset (lr, 16);				\
> > >    	ld	r3,.LC1@toc(2);				\
> > >    	bl	HIDDEN_JUMPTARGET (__fortify_fail);	\
> > > +	addi	r1,r1,112;				\
> > > +	cfi_adjust_cfa_offset (-112);			\
> > > +	cfi_same_value (lr);				\
> > >    .Lok:
> > __fortify_fail doesn't return, does it?  So, why are you
> > adding the stack adjustment after the return?
>
> Just to be able to attach the cfi to it.  It would be wrong to attach it
> to the branch.

Ah.  Then perhaps a trap insn might be more
appropriate (and slightly less confusing)?


r~