2019-10-29 16:20:04

by Heiko Carstens

Subject: Re: [PATCH v2 3/3] s390/livepatch: Implement reliable stack tracing for the consistency model

Hi Miroslav,

> +bool unwind_next_frame_reliable(struct unwind_state *state)
> +{
...
> +}
> +
> void __unwind_start(struct unwind_state *state, struct task_struct *task,
> struct pt_regs *regs, unsigned long sp,
> bool unwind_reliable)

Did you send the wrong version of your patch series? This patch does
not integrate your new function into the existing one. Also the new
parameter you added with the second patch isn't used at all.

2019-10-30 10:07:02

by Miroslav Benes

Subject: Re: [PATCH v2 3/3] s390/livepatch: Implement reliable stack tracing for the consistency model

On Tue, 29 Oct 2019, Heiko Carstens wrote:

> Hi Miroslav,
>
> > +bool unwind_next_frame_reliable(struct unwind_state *state)
> > +{
> ...
> > +}
> > +
> > void __unwind_start(struct unwind_state *state, struct task_struct *task,
> > struct pt_regs *regs, unsigned long sp,
> > bool unwind_reliable)
>
> Did you send the wrong version of your patch series? This patch does
> not integrate your new function into the existing one. Also the new
> parameter you added with the second patch isn't used at all.

No, the version should be correct. Only __unwind_start_reliable() was
integrated. The new parameter is used in arch_stack_walk_reliable()
(unwind_reliable is set to true) and it is propagated to get_stack_info()
where it is used to simplify things for the case.

Miroslav