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

[Peter Zijlstra]

On Thu, Dec 06, 2018 at 09:24:55AM +0100, Jiri Olsa wrote:
> 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?
> 
> the idea is to block the process and try to deliver the event later
> 
> the ring buffer space is freed by user space process moving the tail
> pointer, so I can't see doing this other way than polling

Right, figured that out.

I really really hate this stuff. poll-waiting fundamentally stinks, but
also perf is designed to not/minimally interfere with the observed
tasks, as is ftrace.

And bolting, as it really doesn't fit in the fundamental design, this
fugly thing on top for this special case.. urgh.