Re: [RFC][PATCH v3] ftrace/x86_64: Emulate call function while updating in breakpoint handler

[Steven Rostedt]

On Wed, 1 May 2019 20:57:26 +0200
Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:

> On Wed, May 01, 2019 at 11:01:07AM -0700, Linus Torvalds wrote:
> > This looks sane to me, although I'm surprised that we didn't already
> > have an annotation for the nonstandard stack frame for asm files. That
> > probably would be cleaner in a separate commit, but I guess it doesn't
> > matter.
> > 
> > Anyway, I'm willing to consider the entry code version if it looks a
> > _lot_ simpler than this (so I'd like to see them side-by-side), but
> > it's not like this looks all that complicated to me either.  
> 
> So I posted one earlier today:
> 
>   https://lkml.kernel.org/r/20190501131117.GW2623@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
> 
> it's about a 1/3rd the number of lines and has 32bit support. It also
> provides all the bits required to implement static_call().

That's the patch I started with.

> 
> That is; I think I'm firmly in favour of the entry variant -- provided
> it actually works of course.

And it works. I ran it through tools/testing/selftests/ftrace/ftracetest
and it passed as good as without that patch.

I haven't ran it through my full test suite. I can do that and see how
it makes out.

-- Steve