Re: [PATCH v2 2/6] perf test: Handle metric reuse in pmu-events parsing test

From: Jiri Olsa <jolsa@redhat.com>
To: John Garry <john.garry@huawei.com>
Cc: will@kernel.org, mathieu.poirier@linaro.org, leo.yan@linaro.org,
	peterz@infradead.org, mingo@redhat.com, acme@kernel.org,
	mark.rutland@arm.com, alexander.shishkin@linux.intel.com,
	namhyung@kernel.org, irogers@google.com, linuxarm@huawei.com,
	kjain@linux.ibm.com, kan.liang@linux.intel.com,
	linux-kernel@vger.kernel.org,
	linux-arm-kernel@lists.infradead.org, zhangshaokun@hisilicon.com,
	pc@us.ibm.com
Subject: Re: [PATCH v2 2/6] perf test: Handle metric reuse in pmu-events parsing test
Date: Tue, 6 Apr 2021 15:34:16 +0200	[thread overview]
Message-ID: <YGxjWNdZGqWqL87r@krava> (raw)
In-Reply-To: <49025439-f9e3-0d32-b0a3-ff9f9ff71835@huawei.com>

On Tue, Apr 06, 2021 at 02:21:11PM +0100, John Garry wrote:
> On 06/04/2021 13:55, Jiri Olsa wrote:
> > > > > So expr__find_other() may add a new item to pctx->ids, and we always iterate
> > > > > again, and try to lookup any pmu_events, *, above. If none exist, then we
> > > > hm, I don't see that.. so, what you do is:
> > > > 
> > > > 	hashmap__for_each_entry_safe((&pctx->ids) ....) {
> > > > 
> > > > 		rc = expr__find_other(pe->metric_expr, NULL, pctx, 0);
> > > > 	}
> > > > 
> > > > and what I think we need to do is:
> > > > 
> > > > 	hashmap__for_each_entry_safe((&pctx->ids) ....) {
> > > > 
> > > > 		rc = expr__find_other(pe->metric_expr, NULL, pctx, 0);
> > > > 
> > > > 		break;	
> > > > 	}
> > > > 
> > > > each time you resolve another metric, you need to restart
> > > > the pctx->ids iteration, because there will be new items,
> > > > and we are in the middle of it
> > > Sure, but we will restart anyway.
> > hum, where? you call expr__find_other and continue to next
> > pctx->ids item
> 
> We have:
> 
> resolve_metric_simple()
> {
> 	bool all;
> 
> 	do {
> 		all = true;
> 
> 		hashmap__for_each_entry_safe(&pctx->ids, ...) {
> 
> 			pe = metricgroup_find_metric(cur->key, map);
> 			if (!pe)
> 				continue;
> 
> 			...
> 			all = false;
> 
> 			expr_del_id(pctx, cur->key);
> 
> 			...
> 			rc = expr__find_other(pe->metric_expr, pctx);
> 			if (rc)
> 				goto out_err;
> 		}
> 
> 	} while (!all);
> 
> }
> 
> So once we evaluate a pmu_event in pctx->ids in @pe, @all is set false, and
> we would loop again in the do-while loop, regardless of what
> expr__find_other() does (apart from erroring), and so call
> hashmap__for_each_entry_safe(&pctx->ids, ) again.

ah ok, so it finishes the hash iteration first and
then restarts it.. ok, I missed that, then it's fine

> 
> This is really what is done in __resolve_metric() - indeed, I would use that
> function directly, but it looks hard to extract that from metricgroup.c .

yea, it's another world ;-) it's better to keep it separated

thanks,
jirka

> 
> Thanks,
> John
> 
> > 
> > > Regardless of this, I don't think what I am doing is safe, i.e. adding new
> > > items in the middle of the iter, so I will change in the way you suggest.
> > it'll always add items in the middle of the iteration
> 

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel