Re: x86 entry perf unwinding failure (missing IRET_REGS annotation on stack switch?)

[Josh Poimboeuf]

On Tue, Apr 28, 2020 at 04:16:14PM +0200, Peter Zijlstra wrote:
> On Tue, Apr 28, 2020 at 02:46:27PM +0200, Peter Zijlstra wrote:
> > On Tue, Apr 28, 2020 at 02:04:50AM -0500, Josh Poimboeuf wrote:
> 
> > > I'm thinking something like this should fix it.  Peter, does this look
> > > ok?
> > 
> > Unfortunate. But also, I fear, insufficient. Specifically consider
> > things like:
> > 
> > 	ALTERNATIVE "jmp 1f",
> > 		"alt...
> > 		"..."
> > 		"...insn", X86_FEAT_foo
> > 	1:
> > 
> > This results in something like:
> > 
> > 
> > 	.text	.altinstr_replacement
> > 	e8 xx	...
> > 	90
> > 	90
> > 	...
> > 	90
> > 
> > Where all our normal single byte nops (0x90) are unreachable with
> > undefined CFI, but the alternative might have CFI, which is never
> > propagated.
> > 
> > We ran into this with the validate_alternative stuff from Alexandre.
> 
> > So rather than hacking around this issue, should we not make
> > create_orc() smarter?
> > 
> > I'm trying to come up with something, but so far I'm just making a mess.
> 
> Like this, it's horrid, but it seems to work.
> 
> What do you think of the approach? I'll work on cleaning it up if you
> don't hate it too much ;-)

How'd you know I'd hate it ;-)

That's quite the monstrosity, and I still don't see the point.  I
thought we decided to just disallow CFI changes in alternatives anyway?
That can be done much simpler.

-- 
Josh