[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20160307125934.GD22896@danjae.kornet>
Date: Mon, 7 Mar 2016 21:59:34 +0900
From: Namhyung Kim <namhyung@...nel.org>
To: Jiri Olsa <jolsa@...hat.com>
Cc: Arnaldo Carvalho de Melo <acme@...nel.org>,
Ingo Molnar <mingo@...nel.org>,
Peter Zijlstra <peterz@...radead.org>,
Jiri Olsa <jolsa@...nel.org>,
LKML <linux-kernel@...r.kernel.org>,
David Ahern <dsahern@...il.com>,
Andi Kleen <andi@...stfloor.org>,
Stephane Eranian <eranian@...gle.com>,
Wang Nan <wangnan0@...wei.com>
Subject: Re: [PATCH v2 2/8] perf tools: Introduce
perf_hpp__setup_hists_formats()
On Sun, Mar 06, 2016 at 07:43:25PM +0100, Jiri Olsa wrote:
> On Fri, Mar 04, 2016 at 11:59:36PM +0900, Namhyung Kim wrote:
>
> SNIP
>
> >
> > @@ -2133,8 +2134,19 @@ static void hists__delete_all_entries(struct hists *hists)
> > static void hists_evsel__exit(struct perf_evsel *evsel)
> > {
> > struct hists *hists = evsel__hists(evsel);
> > + struct perf_hpp_fmt *fmt, *pos;
> > + struct perf_hpp_list_node *node;
> >
> > hists__delete_all_entries(hists);
> > +
> > + list_for_each_entry(node, &hists->hpp_formats, list) {
> > + perf_hpp_list__for_each_format_safe(&node->hpp, fmt, pos) {
> > + list_del(&fmt->list);
> > + free(fmt);
> > + }
> > + list_del(&node->list);
> > + free(node);
> > + }
> > }
> >
> > static int hists_evsel__init(struct perf_evsel *evsel)
> > diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h
> > index f4ef513527ba..3cab9dc20822 100644
> > --- a/tools/perf/util/hist.h
> > +++ b/tools/perf/util/hist.h
> > @@ -78,6 +78,7 @@ struct hists {
> > u16 col_len[HISTC_NR_COLS];
> > int socket_filter;
> > struct perf_hpp_list *hpp_list;
> > + struct list_head hpp_formats;
>
> I've been thinking.. should hpp_formats and hpp_list be merged? something like:
>
> struct perf_hpp_list {
> struct list_head nodes;
> int levels;
> }
>
> sturct perf_hpp_list_node {
> int level;
> struct list_head fields;
> struct list_head sorts;
> int nr_sort_keys;
> };
>
> it seems wrong to me that your hierarchy code and current one
> got in separate paths.. we could have the new hierarchy support
> struct above being used in current non-hierarchy code just
> by using single level
>
> I haven't thought all this through.. just an idea ;-)
Yes, I also think that we need to converge on the same data structure
eventually. But I think it's too intrusive to be merged at once.
Also keeping up with the latest perf changes is a PITA. ;-)
So I decided to use it on the hierarchy first, and get tested in the
wild for some time. After that we can convert the non-hierarchy code
to use same path and data structure gradually IMHO.
Thanks,
Namhyung
Powered by blists - more mailing lists