Re: Scrolling down broken with "perf top --hierarchy"

[Namhyung Kim]

On Mon, Oct 24, 2016 at 07:53:12AM +0200, Markus Trippelsdorf wrote:
> On 2016.10.24 at 13:55 +0900, Namhyung Kim wrote:
> > Hi,
> > 
> > Sorry for late reply.
> > 
> > On Mon, Oct 10, 2016 at 07:54:27PM +0200, Markus Trippelsdorf wrote:
> > > On 2016.10.08 at 13:21 +0200, Markus Trippelsdorf wrote:
> > > > On 2016.10.07 at 07:09 +0200, Markus Trippelsdorf wrote:
> > > > > On 2016.10.07 at 06:56 +0200, Markus Trippelsdorf wrote:
> > > > > > On 2016.10.07 at 06:32 +0200, Markus Trippelsdorf wrote:
> > > > > > And BTW symbols are currently always cut off at 60 characters in
> > > > > > expanded entries.
> > > > > 
> > > > > Hmm, no. Sometimes they are cut off, sometimes they are not. I haven't
> > > > > figured out what triggered this strange behavior.
> > > > 
> > > > Here is an example:
> > > > 
> > > >  % echo $COLUMNS
> > > >  179
> > > >  % perf top --hierarchy
> > > > +  34.81%     [kernel]
> > > > -  20.89%     chrome
> > > >        0.51%     [.] v8::internal::IncrementalMarking:
> > > >        0.43%     [.] tc_malloc
> > > >        0.29%     [.] sqlite3BtreeMovetoUnpacked
> > > >        0.28%     [.] tc_free
> > > >        0.24%     [.] v8::internal::BodyDescriptorBase:
> > > >        0.24%     [.] sqlite3VdbeExec
> > > >        0.22%     [.] v8::internal::MarkCompactCollecto
> > > >        0.19%     [.] blink::SelectorChecker::checkOne
> > > >        0.19%     [.] SkBlitRow::Color32
> > > >        0.18%     [.] SkBlitLCD16OpaqueRow_SSE2
> > > >        0.17%     [.] btreeInitPage.part.366
> > > >        0.16%     [.] blink::SelectorChecker::matchSele
> > > >        0.15%     [.] blink::ElementRuleCollector::coll
> > > >        0.15%     [.] blink::CSSTokenizer::consumeName
> > > >        0.14%     [.] sqlite3GetVarint
> > > >        0.13%     [.] operator new[]
> > > >        0.12%     [.] FPDFAPI_inflate_fast
> > > >        0.11%     [.] v8::internal::HeapObject::SizeFro
> > > >        0.09%     [.] tracked_objects::ThreadData::Tall
> > > > ...
> > > 
> > > To continue this monologue, perf doesn't even look at these entries. So some
> > > hists__calc_col_len() calls seem to be missing for the "perf top --hierarchy"
> > > case or hists__reset_col_len() is called too early or too often.
> > 
> > Right, it missed to call the function for leaf entries.  Could you
> > please check below patch fixes the problem?
> 
> Yes, it does. Thank you.
> 
> Another issue: all entries vanish if one scrolls to the left two times.

Hmm.. Did you mean pressing RIGHT key two times?

Thanks,
Namhyung