Re: [PATCH] perf report: Create option to disable raw event ordering

[Arnaldo Carvalho de Melo <acme@kernel.org>]

Em Mon, Feb 22, 2021 at 10:32:40PM +0100, Jiri Olsa escreveu:
> On Fri, Feb 19, 2021 at 03:00:05PM +0800, Jin Yao wrote:
> > Warning "dso not found" is reported when using "perf report -D".
> > 
> >  66702781413407 0x32c0 [0x30]: PERF_RECORD_SAMPLE(IP, 0x2): 28177/28177: 0x55e493e00563 period: 106578 addr: 0
> >   ... thread: perf:28177
> >   ...... dso: <not found>
> > 
> >  66702727832429 0x9dd8 [0x38]: PERF_RECORD_COMM exec: triad_loop:28177/28177
> > 
> > The PERF_RECORD_SAMPLE event (timestamp: 66702781413407) should be after the
> > PERF_RECORD_COMM event (timestamp: 66702727832429), but it's early processed.
> > 
> > So for most of cases, it makes sense to keep the event ordered even for dump
> > mode. But it would be also useful to disable ordered_events for reporting raw
> > dump to see events as they are stored in the perf.data file.
> > 
> > So now, set ordered_events by default to true and add a new option
> > 'disable-order' to disable it. For example,
> > 
> > perf report -D --disable-order
> > 
> > Fixes: 977f739b7126b ("perf report: Disable ordered_events for raw dump")
> > Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
> 
> Acked-by: Jiri Olsa <jolsa@redhat.com>

Thanks, applied to perf/core.

- Arnaldo

 
> thanks,
> jirka
> 
> 
> > ---
> >  tools/perf/Documentation/perf-report.txt | 3 +++
> >  tools/perf/builtin-report.c              | 5 ++++-
> >  2 files changed, 7 insertions(+), 1 deletion(-)
> > 
> > diff --git a/tools/perf/Documentation/perf-report.txt b/tools/perf/Documentation/perf-report.txt
> > index f546b5e9db05..87112e8d904e 100644
> > --- a/tools/perf/Documentation/perf-report.txt
> > +++ b/tools/perf/Documentation/perf-report.txt
> > @@ -224,6 +224,9 @@ OPTIONS
> >  --dump-raw-trace::
> >          Dump raw trace in ASCII.
> >  
> > +--disable-order::
> > +	Disable raw trace ordering.
> > +
> >  -g::
> >  --call-graph=<print_type,threshold[,print_limit],order,sort_key[,branch],value>::
> >          Display call chains using type, min percent threshold, print limit,
> > diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
> > index 2a845d6cac09..0d65c98794a8 100644
> > --- a/tools/perf/builtin-report.c
> > +++ b/tools/perf/builtin-report.c
> > @@ -84,6 +84,7 @@ struct report {
> >  	bool			nonany_branch_mode;
> >  	bool			group_set;
> >  	bool			stitch_lbr;
> > +	bool			disable_order;
> >  	int			max_stack;
> >  	struct perf_read_values	show_threads_values;
> >  	struct annotation_options annotation_opts;
> > @@ -1296,6 +1297,8 @@ int cmd_report(int argc, const char **argv)
> >  	OPTS_EVSWITCH(&report.evswitch),
> >  	OPT_BOOLEAN(0, "total-cycles", &report.total_cycles_mode,
> >  		    "Sort all blocks by 'Sampled Cycles%'"),
> > +	OPT_BOOLEAN(0, "disable-order", &report.disable_order,
> > +		    "Disable raw trace ordering"),
> >  	OPT_END()
> >  	};
> >  	struct perf_data data = {
> > @@ -1329,7 +1332,7 @@ int cmd_report(int argc, const char **argv)
> >  	if (report.mmaps_mode)
> >  		report.tasks_mode = true;
> >  
> > -	if (dump_trace)
> > +	if (dump_trace && report.disable_order)
> >  		report.tool.ordered_events = false;
> >  
> >  	if (quiet)
> > -- 
> > 2.17.1
> > 
> 

-- 

- Arnaldo