Re: [PATCH v11 04/12] scs: disable when function graph tracing is enabled

[Peter Zijlstra]

On Fri, Apr 17, 2020 at 03:46:21PM +0100, Mark Rutland wrote:
> > > diff --git a/arch/Kconfig b/arch/Kconfig
> > > index 691a552c2cc3..c53cb9025ad2 100644
> > > --- a/arch/Kconfig
> > > +++ b/arch/Kconfig
> > > @@ -542,6 +542,7 @@ config ARCH_SUPPORTS_SHADOW_CALL_STACK
> > >  
> > >  config SHADOW_CALL_STACK
> > >  	bool "Clang Shadow Call Stack"
> > > +	depends on DYNAMIC_FTRACE_WITH_REGS || !FUNCTION_GRAPH_TRACER
> > >  	depends on ARCH_SUPPORTS_SHADOW_CALL_STACK
> > >  	help
> > >  	  This option enables Clang's Shadow Call Stack, which uses a
>  
> > AFAICT you also need to kill KRETPROBES, which plays similar games.
> 
> Hmm... how does KREPROBES work? If you can only mess with the return
> address when probing the first instruction in the function, it'll just
> work for SCS or pointer authentication, as the LR is used at that
> instant. If KRETPROBES tries to mess with the return address elsewhere
> it'd be broken today...

To be fair, I've not looked at the arm64 implementation. x86 does gross
things like ftrace does. On x86 ftrace_graph and kretprobe also can't
be on at the same time for the same function, there's some yuck around
there.

Rostedt was recently talking about cleaning some of that up.

But if kretprobe can work on arm64, then ftrace_graph can too, but I
think that links back to what you said earlier, you didn't want more
ftrace variants or something.

> > And doesn't BPF also do stuff like this?
> 
> Can BPF mess with return addresses now!?

At least on x86 I think it does. But what do I know, I can't operate
that stuff. Rostedt might know.