Re: [RFC/PATCH 0/4] perf report: Support folded callchain output (v2)

From: Brendan Gregg
Date: Mon Nov 02 2015 - 17:43:41 EST


On Mon, Nov 2, 2015 at 2:12 PM, Namhyung Kim <namhyung@xxxxxxxxxx> wrote:
> Hi Arnaldo,
>
> On Mon, Nov 02, 2015 at 06:30:21PM -0300, Arnaldo Carvalho de Melo wrote:
>> Em Mon, Nov 02, 2015 at 12:37:28PM -0800, Brendan Gregg escreveu:
>> > G'Day Namhyung,
>> >
>> > On Mon, Nov 2, 2015 at 4:57 AM, Namhyung Kim <namhyung@xxxxxxxxxx> wrote:
>> > > Hello,
>> > >
>> > > This is what Brendan requested on the perf-users mailing list [1] to
>> > > support FlameGraphs [2] more efficiently. This patchset adds a few
>> > > more callchain options to adjust the output for it.
>> > >
>> > > At first, 'folded' output mode was added. The folded output puts all
>> > > calchain nodes in a line separated by semicolons, a space and the
>> > > value. Now it only supports --stdio as other UI provides some way of
>> > > folding/expanding callchains dynamically.
>> > >
>> > > The value is now can be one of 'percent', 'period', or 'count'. The
>> > > percent is current default output and the period is the raw number of
>> > > sample periods. The count is the number of samples for each callchain.
>> > >
>> > > Here's an example:
>> > >
>> > > $ perf report --no-children --show-nr-samples --stdio -g folded,count
>> > > ...
>> > > 39.93% 80 swapper [kernel.vmlinux] [k] intel_idel
>> > > intel_idle;cpuidle_enter_state;cpuidle_enter;call_cpuidle;cpu_startup_entry;start_secondary 57
>> > > intel_idle;cpuidle_enter_state;cpuidle_enter;call_cpuidle;cpu_startup_entry;rest_init;... 23
>> >
>> > Thanks!
>> >
>> > So for the folded output I don't need the summary line (the row of
>> > columns printed by hist_entry__snprintf()), and don't need anything
>> > except folded stacks and the counts. If working with the existing
>> > stdio interface is making it harder than it needs to be, might it be
>>
>> I don't think it so, just add some flag asking for that
>> hist_entry__snprintf() to be supressed, ideas for a long option name?
>>
>> Having it as Namhyung did may have value for some people as a more
>> compact way to show the callchains together with the hist_entry line.
>
> Yeah, I'd keep the hist entry line unless it's too hard to
> parse/filter. IMHO it's just a way to show callchains, so no need to
> have separate output mode..

Ok, good point, it can be thought of as a different stack representation format.

>
> Brendan, I guess you still need to know other info like cpu or pid, no?
>

Yes, I just realized that I either include the process name (Command
column) or name-PID, as the first folded element. Eg, output can be:

mkdir;getopt_long;page_fault;do_page_fault;__do_page_fault;filemap_map_pages 3

Or:

mkdir-21918;getopt_long;page_fault;do_page_fault;__do_page_fault;filemap_map_pages
2

Usually the first, but sometimes it's helpful to split on PID as well.

As for what to call such options (which may be a follow on patch
anyway) ... maybe something like:

"folded": fold stacks as single lines
"nameonly,folded": suppress summary line and include process name in
the folded stack
"pidonly,folded": suppress summary line and include process_name-PID
in the folded stack

> And I feel like it'd be better to put the count before the callchains
> for consistency like below. Is it OK to you?
>
> $ perf report --no-children --show-nr-samples --stdio -g folded,count
> ...
> 39.93% 80 swapper [kernel.vmlinux] [k] intel_idel
> 57 intel_idle;cpuidle_enter_state;cpuidle_enter;call_cpuidle;cpu_startup_entry;start_secondary
> 23 intel_idle;cpuidle_enter_state;cpuidle_enter;call_cpuidle;cpu_startup_entry;rest_init;...
>

If it was printing with the perf report summary, sure, but if we have
a way to only emit folded output, then counts last would be perfect
and maybe a bit more intuitive (key then value).

>
>>
>> With this in mind, do you have any other issues with Namhyung's
>> patchkit? An acked-by/tested-by you would be nice to have, and then we
>> could work out the new option to suppress that hist_entry__snprintf()
>> in a follow up patch.

Acked and tested, yes.

Looks like I'd be using caller ordering, eg, to get lines like this:

__GI___libc_read;entry_SYSCALL_64_fastpath;sys_read;vfs_read;__vfs_read;urandom_read;extract_entropy_user;extract_buf;check_events;xen_hypercall_xen_version
91

Which I can do just by using "-g folded,count,caller".

>>
>> > easier to make it a separate interface (ui/folded), that just emitted
>> > the folded output? Just an idea. This existing patchset is working for
>> > me, I'd just be filtering the output.
>> >
>> > Having the option for percentages and periods is nice. I can envisage
>> > using periods (for latency flame graphs).
>
> Glad to see you like it. :)
>
> Thanks,
> Namhyung
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/