Messages in this thread | | | Date | Wed, 25 Nov 2015 10:26:08 +0900 | From | Namhyung Kim <> | Subject | Re: [PATCH 34/37] perf hists browser: Support flat callchains |
| |
Hi Arnaldo,
On Tue, Nov 24, 2015 at 12:45:51PM -0200, Arnaldo Carvalho de Melo wrote: > Em Tue, Nov 24, 2015 at 02:27:08PM +0900, Namhyung Kim escreveu: > > On Mon, Nov 23, 2015 at 04:16:48PM +0100, Frederic Weisbecker wrote: > > > On Thu, Nov 19, 2015 at 02:53:20PM -0300, Arnaldo Carvalho de Melo wrote: > > > > From: Namhyung Kim <namhyung@kernel.org> > > > [...] > > > > > > > +int callchain_node__make_parent_list(struct callchain_node *node) > > > > +{ > > > > + struct callchain_node *parent = node->parent; > > > > + struct callchain_list *chain, *new; > > > > + LIST_HEAD(head); > > > > + > > > > + while (parent) { > > > > + list_for_each_entry_reverse(chain, &parent->val, list) { > > > > + new = malloc(sizeof(*new)); > > > > + if (new == NULL) > > > > + goto out; > > > > + *new = *chain; > > > > + new->has_children = false; > > > > + list_add_tail(&new->list, &head); > > > > + } > > > > + parent = parent->parent; > > > > + } > > > > + > > > > + list_for_each_entry_safe_reverse(chain, new, &head, list) > > > > + list_move_tail(&chain->list, &node->parent_val); > > > > + > > > > + if (!list_empty(&node->parent_val)) { > > > > + chain = list_first_entry(&node->parent_val, struct callchain_list, list); > > > > + chain->has_children = rb_prev(&node->rb_node) || rb_next(&node->rb_node); > > > > + > > > > + chain = list_first_entry(&node->val, struct callchain_list, list); > > > > + chain->has_children = false; > > > > > > I'm a bit puzzled with this, can't we rewind through the parents on printing or adding > > > to the flat rbtree instead of having this parent_val field? > > > > Yes, this code is to simplify things on parent nodes. Maybe we could > > go up to parents and print the callchain list there as you said. > > > > However, problem I think is how to handle 'has_children' information > > on parents. That info controls folding status of each callchain. As > > the info is in the struct callchain_list and flat or folded callchain > > mode require the info should be in the top-most entry, I cannot share > > entries in parent nodes. > > > > Thus I simply copied callchain lists in parents to leaf nodes. Yes, > > it will consume some memory but can simplify the code. > > I haven't done any measuring, but I'm noticing that 'perf top -g' is > showing more warnings about not being able to process events fast enough > and so ends up losing events, I tried with --max-stack 16 and it helped, > this is just a heads up.
OK, but it seems that it's not related to this patch since this patch only affects flat or folded callchain mode.
> > Perhaps my workstation workloads are gettning deeper callchains over > time, but perhaps this is the cost of processing callchains that is > increasing, I need to stop and try to quantify this. > > We really need to look at reducing the overhead of processing > callchains.
Right, but with my multi-thread work, I realized that perf is getting heavier recently. I guess it's mostly due to the atomic refcount work. I need to get back to the multi-thread work..
Anyway I made a initial multi-thread support for perf top too. I think I posted it to the list, but I cannot find the link. You can take a look at it on 'perf/top-threaded-v1' branch in my tree.
git.kernel.org/pub/scm/linux/kernel/git/namhyung/linux-perf.git
Thanks, Namhyung
| |