Re: [RFC PATCH 13/13] m68k: mvme16x: Convert to clocksource APIy

[Thomas Gleixner]

Finn,

On Wed, 14 Nov 2018, Finn Thain wrote:
> On Tue, 13 Nov 2018, Thomas Gleixner wrote:
> > Urgh. Then you have more serious trouble. If the interrupting handler 
> > calls any of the time accessor functions then you can actually live lock 
> > when the interrupt happens in the middle of the write locked section of 
> > the core timekeeping update.
> 
> Does this apply to arch_gettimeoffset also? If so, that would mean another 
> bug fix patch for -stable...

arch_gettimeoffset() can be called from any context. The only interrupt
which can affect it is the timer interrupt. As m68k are all UP machines
this is not a problem because:

  ktime_get()
     do {
        seq = read_seqcount_begin(tk.seq);
	nsec = .... + arch_gettimeoffset();
     } while (seqcount_retry(tk.seq);

and the timer interrupt does:

  arch_timeoffset += value;
  ...
    xtime_update()
      ...
        write_seqcount_begin(tk.seq);
        ....
        write_seqcount_end(tk.seq);

So where ever the timer interrupt hits inside of ktime_get() or any other
accessor the seqcount will make it retry, so the result is always
consistent.

For SMP that would be a different story.

> > So you really want to disable interrupts across the whole timer 
> > interrupt function or make sure that the timer interrupt is the highest 
> > priority one on the system.
> > 
> 
> The timer interrupt priority isn't always what one would hope. 
> 
> But I can certainly disable interrupts for timer_interrupt() execution 
> (that is, xtime_update() etc.)

Oh yes, you should.

Thanks,

	tglx