Re: [PATCH 1/8] perf: Allow to block process in syscall tracepoints

[Peter Zijlstra]

On Thu, Dec 06, 2018 at 09:10:28AM +0100, Peter Zijlstra wrote:
> On Wed, Dec 05, 2018 at 05:05:02PM +0100, Jiri Olsa wrote:
> > +static void trace_block_syscall(struct pt_regs *regs, bool enter)
> > +{
> > +	current->perf_blocked = true;
> > +
> > +	do {
> > +		schedule_timeout(100 * HZ);
> > +		current->perf_blocked_cnt = 0;
> > +
> > +		if (enter) {
> > +			/* perf syscalls:* enter */
> > +			perf_trace_syscall_enter(regs);
> > +
> > +			/* perf raw_syscalls:* enter */
> > +			perf_trace_sys_enter(&event_sys_enter, regs, regs->orig_ax);
> > +		} else {
> > +			/* perf syscalls:* enter */
> > +			perf_trace_syscall_exit(regs);
> > +
> > +			/* perf raw_syscalls:* enter */
> > +			perf_trace_sys_exit(&event_sys_exit, regs, regs->ax);
> > +		}
> > +	} while (current->perf_blocked_cnt);
> > +
> > +	current->perf_blocked = false;
> > +}
> 
> I don't understand this.. why are we using schedule_timeout() and all
> that?

Urgh.. in fact, the more I look at this the more I hate it.

We want to block in __perf_output_begin(), but we cannot because both
tracepoints and perf will have preemptability disabled down there.

So what we do is fail the event, fake the lost count and go all the way
up that callstack, detect the failure and then poll-wait and retry.

And only do this for a few special events...  *yuck*