Re: [PATCH v3 18/39] x86/ibt,ftrace: Make function-graph play nice

[Josh Poimboeuf]

On Fri, Mar 04, 2022 at 10:44:54PM +0100, Peter Zijlstra wrote:
> On Fri, Mar 04, 2022 at 01:03:34PM -0800, Josh Poimboeuf wrote:
> > On Fri, Mar 04, 2022 at 08:48:53PM +0100, Peter Zijlstra wrote:
> > > On Fri, Mar 04, 2022 at 09:51:54AM -0800, Josh Poimboeuf wrote:
> > > > On Thu, Mar 03, 2022 at 12:23:39PM +0100, Peter Zijlstra wrote:
> > > > > +
> > > > > +	addq $16, %rsp
> > > > > +	ANNOTATE_INTRA_FUNCTION_CALL
> > > > > +	call .Ldo_rop
> > > > > +	int3
> > > > > +.Ldo_rop:
> > > > > +	mov %rdi, (%rsp)
> > > > > +	UNWIND_HINT_FUNC
> > > > > +	RET
> > > > 
> > > > Why the int3?
> > > 
> > > Speculation trap :-) Either I'm too paranoid or not paranoid enough; but
> > > without it it's just too close to a retpoline and it doesn't feel right.
> > 
> > Um, it *is* a retpoline :-)
> > 
> > Can you just use the RETPOLINE macro?  Along with a comment stating why
> > it can't just do a JMP_NOSPEC?
> 
> There is no RETPOLINE macro; or rather it is completely contained in
> lib/retpoline.S and I'd sorta like to keep it that way.
> 
> That said, I can stick a comment on.

The only reason it's in retpoline.S is because nobody else needed it.

It just seems weird to reinvent the wheel, especially with a slightly
different method of stopping speculation.

And I could envision other cases where we might want an unconditional
retpoline.

Your call though...

-- 
Josh