Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1327663 > unrolled thread

[PATCHSET 00/23] perf tools: Add support for hierachy view (v5)

Started byNamhyung Kim <namhyung@kernel.org>
First post2016-02-05 14:20 +0100
Last post2016-02-05 14:20 +0100
Articles 5 — 1 participant

Back to article view | Back to linux.kernel

Contents

  [PATCHSET 00/23] perf tools: Add support for hierachy view (v5) Namhyung Kim <namhyung@kernel.org> - 2016-02-05 14:20 +0100
    [PATCH 06/23] perf callchain: Check return value of append_chain_children() Namhyung Kim <namhyung@kernel.org> - 2016-02-05 14:20 +0100
    [PATCH 01/23] perf hists browser: Fix percentage update on key press Namhyung Kim <namhyung@kernel.org> - 2016-02-05 14:20 +0100
    [PATCH 04/23] perf callchain: Add enum match_result for match_chain() Namhyung Kim <namhyung@kernel.org> - 2016-02-05 14:20 +0100
    [PATCH 05/23] perf callchain: Check return value of split_add_child() Namhyung Kim <namhyung@kernel.org> - 2016-02-05 14:20 +0100

#1327663 — [PATCHSET 00/23] perf tools: Add support for hierachy view (v5)

FromNamhyung Kim <namhyung@kernel.org>
Date2016-02-05 14:20 +0100
Subject[PATCHSET 00/23] perf tools: Add support for hierachy view (v5)
Message-ID<qYOj8-ED-9@gated-at.bofh.it>
Hello,

This patchset implements a new feature that collects hist entries in a
hierachical manner.  That means lower-level entries belong to an
upper-level entry.  The entry hierachy is built on the sort keys
given, so users can set it whatever they want.  It only shows
top-level entries first, and user can expand/collapse it dynamically.

 * Changes from v4)
  - rebased onto the current acme/perf/core
  - fix memory leak on callchian_merge error path  (Arnaldo)
  - fix a bug on perf-top regarding percent calculation
  - split hierarchy filtering code
 
 * Changes from v3)
  - rebased onto the percent limit patchset v2

 * Changes from v2)
  - check memory allocation failure in hists__hierarchy_insert_entry  (Jiri)
  - remove unused rb_hierarchy_first()  (Arnaldo)
  - support callchain percent limit  (Andi)
  - break TUI context menu cleanup  (Arnaldo)
  

This time I implemented it for every output browser including TUI.
A screenshot on TUI looks like below:

For normal output:

  $ perf report --tui
  Samples: 3K of event 'cycles:pp', Event count (approx.): 1695979674
    Overhead  Command        Shared Object         Symbol
  ------------------------------------------------------------------------
  -    7.57%  swapper        [kernel.vmlinux]      [k] intel_idle
       intel_idle
       cpuidle_enter_state
       cpuidle_enter
       call_cpuidle
     + cpu_startup_entry
  +    1.16   firefox        firefox               [.] 0x00000000000019433
  +    0.97%  firefox        libpthread-2.22.so    [.] pthread_mutex_lock
  ...


With hierarchy view,

  $ perf report --tui --hierarchy
  Samples: 3K of event 'cycles:pp', Event count (approx.): 1695979674
   Overhead        Command / Shared Object / Symbol
  -------------------------------------------------------------------
  +  76.30%        firefox
  -   9.95%        swapper
     -   9.51%        [kernel.vmlinux]
        -   7.57         [k] intel_idle
	     intel_idle
	     cpuidle_enter_state
	     cpuidle_enter
	     call_cpuidle
	   + cpu_startup_entry
	+   0.15%        [k] __schedule
	+   0.12%        [k] menu_select
	...
     +   0.34%        [sdhci]
     +   0.06%        [e1000e]
     ...
 +    5.65%        Xorg
 +    5.42%        Socket Thread
 ...

As you can see, overhead of an upper level entry is the sum of
overhead of lower level entries.  The entries are aligned by its order
of matching sort keys.

This is available from 'perf/hierarchy-v5' branch in my tree:

  git://git.kernel.org/pub/scm/linux/kernel/git/namhyung/linux-perf.git


Any comments are welcome, thanks!
Namhyung


Cc: Don Zickus <dzickus@redhat.com>
Cc: Pekka Enberg <penberg@kernel.org>
Cc: Moinuddin Quadri <moin18@gmail.com>


[1] https://lkml.org/lkml/2013/5/21/24


Namhyung Kim (23):
  perf hists browser: Fix percentage update on key press
  perf callchain: Check return value of add_child()
  perf callchain: Check return value of fill_node()
  perf callchain: Add enum match_result for match_chain()
  perf callchain: Check return value of split_add_child()
  perf callchain: Check return value of append_chain_children()
  perf hists: Return error from hists__collapse_resort()
  perf report: Check error during report__collapse_hists()
  perf hists: Basic support of hierarchical report view
  perf hists: Resort hist entries with hierarchy
  perf hists: Add helper functions for hierarchy mode
  perf hists: Introduce hist_entry__filter()
  perf hists: Support filtering in hierarchy mode
  perf ui/stdio: Implement hierarchy output mode
  perf ui/stdio: Align column header for hierarchy output
  perf hists browser: Count number of hierarchy entries
  perf hists browser: Support collapsing/expanding whole entries in
    hierarchy
  perf hists browser: Implement hierarchy output
  perf hists browser: Align column header in hierarchy mode
  perf ui/gtk: Implement hierarchy output mode
  perf report: Add --hierarchy option
  perf hists: Support decaying in hierarchy mode
  perf top: Add --hierarchy option

 tools/perf/Documentation/perf-report.txt |   3 +
 tools/perf/Documentation/perf-top.txt    |   3 +
 tools/perf/Documentation/tips.txt        |   1 +
 tools/perf/builtin-report.c              |  31 +-
 tools/perf/builtin-top.c                 |  15 +
 tools/perf/ui/browsers/hists.c           | 488 ++++++++++++++++++++++++++++---
 tools/perf/ui/gtk/hists.c                | 161 +++++++++-
 tools/perf/ui/hist.c                     |  14 +
 tools/perf/ui/stdio/hist.c               | 184 +++++++++++-
 tools/perf/util/callchain.c              | 102 +++++--
 tools/perf/util/hist.c                   | 463 ++++++++++++++++++++++++++---
 tools/perf/util/hist.h                   |  16 +-
 tools/perf/util/sort.c                   | 113 +++++++
 tools/perf/util/sort.h                   |  14 +-
 tools/perf/util/symbol.h                 |   3 +-
 15 files changed, 1493 insertions(+), 118 deletions(-)

-- 
2.7.0

[toc] | [next] | [standalone]

#1327665 — [PATCH 06/23] perf callchain: Check return value of append_chain_children()

FromNamhyung Kim <namhyung@kernel.org>
Date2016-02-05 14:20 +0100
Subject[PATCH 06/23] perf callchain: Check return value of append_chain_children()
Message-ID<qYOsO-I2-3@gated-at.bofh.it>
In reply to#1327663
Now it can check the error case, so check and pass it to the caller.

Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
 tools/perf/util/callchain.c | 18 ++++++++++++------
 1 file changed, 12 insertions(+), 6 deletions(-)

diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c
index 5259379892e1..24b4bd0d7754 100644
--- a/tools/perf/util/callchain.c
+++ b/tools/perf/util/callchain.c
@@ -586,7 +586,7 @@ append_chain(struct callchain_node *root,
 	     struct callchain_cursor *cursor,
 	     u64 period);
 
-static void
+static int
 append_chain_children(struct callchain_node *root,
 		      struct callchain_cursor *cursor,
 		      u64 period)
@@ -598,7 +598,7 @@ append_chain_children(struct callchain_node *root,
 
 	node = callchain_cursor_current(cursor);
 	if (!node)
-		return;
+		return -1;
 
 	/* lookup in childrens */
 	while (*p) {
@@ -611,6 +611,8 @@ append_chain_children(struct callchain_node *root,
 		ret = append_chain(rnode, cursor, period);
 		if (ret == MATCH_EQ)
 			goto inc_children_hit;
+		if (ret == MATCH_ERROR)
+			return -1;
 
 		if (ret == MATCH_LT)
 			p = &parent->rb_left;
@@ -620,7 +622,7 @@ append_chain_children(struct callchain_node *root,
 	/* nothing in children, add to the current node */
 	rnode = add_child(root, cursor, period);
 	if (rnode == NULL)
-		return;
+		return -1;
 
 	rb_link_node(&rnode->rb_node_in, parent, p);
 	rb_insert_color(&rnode->rb_node_in, &root->rb_root_in);
@@ -628,6 +630,7 @@ append_chain_children(struct callchain_node *root,
 inc_children_hit:
 	root->children_hit += period;
 	root->children_count++;
+	return 0;
 }
 
 static enum match_result
@@ -688,7 +691,8 @@ append_chain(struct callchain_node *root,
 	}
 
 	/* We match the node and still have a part remaining */
-	append_chain_children(root, cursor, period);
+	if (append_chain_children(root, cursor, period) < 0)
+		return MATCH_ERROR;
 
 	return MATCH_EQ;
 }
@@ -702,7 +706,8 @@ int callchain_append(struct callchain_root *root,
 
 	callchain_cursor_commit(cursor);
 
-	append_chain_children(&root->node, cursor, period);
+	if (append_chain_children(&root->node, cursor, period) < 0)
+		return -1;
 
 	if (cursor->nr > root->max_depth)
 		root->max_depth = cursor->nr;
@@ -730,7 +735,8 @@ merge_chain_branch(struct callchain_cursor *cursor,
 
 	if (src->hit) {
 		callchain_cursor_commit(cursor);
-		append_chain_children(dst, cursor, src->hit);
+		if (append_chain_children(dst, cursor, src->hit) < 0)
+			return -1;
 	}
 
 	n = rb_first(&src->rb_root_in);
-- 
2.7.0

[toc] | [prev] | [next] | [standalone]

#1327666 — [PATCH 01/23] perf hists browser: Fix percentage update on key press

FromNamhyung Kim <namhyung@kernel.org>
Date2016-02-05 14:20 +0100
Subject[PATCH 01/23] perf hists browser: Fix percentage update on key press
Message-ID<qYOsO-I2-7@gated-at.bofh.it>
In reply to#1327663
Currently 'perf top --tui' decrements percentage of all entries on any
key press.  This is because it adds total period as new samples are
added to hists.  As perf-top does it currently but added samples are not
passed to the display thread, the percentages are decresing
continuously.

So separate total period stat into a different variable so that it
cannot affect the output total period.  This new total period stats are
used only for calcualating callchain percent limit.

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
 tools/perf/util/hist.c | 26 +++++++++++++++++++-------
 tools/perf/util/hist.h |  2 ++
 2 files changed, 21 insertions(+), 7 deletions(-)

diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
index 12f2d794dc28..f6bdb827e6fb 100644
--- a/tools/perf/util/hist.c
+++ b/tools/perf/util/hist.c
@@ -405,6 +405,16 @@ static u8 symbol__parent_filter(const struct symbol *parent)
 	return 0;
 }
 
+static void hist_entry__add_callchain_period(struct hist_entry *he, u64 period)
+{
+	if (!symbol_conf.use_callchain)
+		return;
+
+	he->hists->callchain_period += period;
+	if (!he->filtered)
+		he->hists->callchain_non_filtered_period += period;
+}
+
 static struct hist_entry *hists__findnew_entry(struct hists *hists,
 					       struct hist_entry *entry,
 					       struct addr_location *al,
@@ -434,9 +444,7 @@ static struct hist_entry *hists__findnew_entry(struct hists *hists,
 		if (!cmp) {
 			if (sample_self) {
 				he_stat__add_period(&he->stat, period, weight);
-				hists->stats.total_period += period;
-				if (!he->filtered)
-					hists->stats.total_non_filtered_period += period;
+				hist_entry__add_callchain_period(he, period);
 			}
 			if (symbol_conf.cumulate_callchain)
 				he_stat__add_period(he->stat_acc, period, weight);
@@ -471,9 +479,8 @@ static struct hist_entry *hists__findnew_entry(struct hists *hists,
 		return NULL;
 
 	if (sample_self)
-		hists__inc_stats(hists, he);
-	else
-		hists->nr_entries++;
+		hist_entry__add_callchain_period(he, period);
+	hists->nr_entries++;
 
 	rb_link_node(&he->rb_node_in, parent, p);
 	rb_insert_color(&he->rb_node_in, hists->entries_in);
@@ -1206,9 +1213,14 @@ static void output_resort(struct hists *hists, struct ui_progress *prog,
 	struct rb_root *root;
 	struct rb_node *next;
 	struct hist_entry *n;
+	u64 callchain_total;
 	u64 min_callchain_hits;
 
-	min_callchain_hits = hists__total_period(hists) * (callchain_param.min_percent / 100);
+	callchain_total = hists->callchain_period;
+	if (symbol_conf.filter_relative)
+		callchain_total = hists->callchain_non_filtered_period;
+
+	min_callchain_hits = callchain_total * (callchain_param.min_percent / 100);
 
 	if (sort__need_collapse)
 		root = &hists->entries_collapsed;
diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h
index 1c7544a8fe1a..21c1bc4661e3 100644
--- a/tools/perf/util/hist.h
+++ b/tools/perf/util/hist.h
@@ -66,6 +66,8 @@ struct hists {
 	struct rb_root		entries_collapsed;
 	u64			nr_entries;
 	u64			nr_non_filtered_entries;
+	u64			callchain_period;
+	u64			callchain_non_filtered_period;
 	struct thread		*thread_filter;
 	const struct dso	*dso_filter;
 	const char		*uid_filter_str;
-- 
2.7.0

[toc] | [prev] | [next] | [standalone]

#1327668 — [PATCH 04/23] perf callchain: Add enum match_result for match_chain()

FromNamhyung Kim <namhyung@kernel.org>
Date2016-02-05 14:20 +0100
Subject[PATCH 04/23] perf callchain: Add enum match_result for match_chain()
Message-ID<qYOsO-I2-11@gated-at.bofh.it>
In reply to#1327663
The append_chain() might return either result of match_chain() or
other (error) code.  But match_chain() can return any value in s64 type
so it's hard to check the error case.  Add new enum match_result and
make match_chain() return non-negative values only so that we can check
the error cases.

Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
 tools/perf/util/callchain.c | 52 +++++++++++++++++++++++++++++----------------
 1 file changed, 34 insertions(+), 18 deletions(-)

diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c
index a82ea6f6fc0f..dab2c1f1e86b 100644
--- a/tools/perf/util/callchain.c
+++ b/tools/perf/util/callchain.c
@@ -475,16 +475,32 @@ add_child(struct callchain_node *parent,
 	return new;
 }
 
-static s64 match_chain(struct callchain_cursor_node *node,
-		      struct callchain_list *cnode)
+enum match_result {
+	MATCH_ERROR  = -1,
+	MATCH_EQ,
+	MATCH_LT,
+	MATCH_GT,
+};
+
+static enum match_result match_chain(struct callchain_cursor_node *node,
+				     struct callchain_list *cnode)
 {
 	struct symbol *sym = node->sym;
+	u64 left, right;
 
 	if (cnode->ms.sym && sym &&
-	    callchain_param.key == CCKEY_FUNCTION)
-		return cnode->ms.sym->start - sym->start;
-	else
-		return cnode->ip - node->ip;
+	    callchain_param.key == CCKEY_FUNCTION) {
+		left = cnode->ms.sym->start;
+		right = sym->start;
+	} else {
+		left = cnode->ip;
+		right = node->ip;
+	}
+
+	if (left == right)
+		return MATCH_EQ;
+
+	return left > right ? MATCH_GT : MATCH_LT;
 }
 
 /*
@@ -549,7 +565,7 @@ split_add_child(struct callchain_node *parent,
 		cnode = list_first_entry(&first->val, struct callchain_list,
 					 list);
 
-		if (match_chain(node, cnode) < 0)
+		if (match_chain(node, cnode) == MATCH_LT)
 			pp = &p->rb_left;
 		else
 			pp = &p->rb_right;
@@ -562,7 +578,7 @@ split_add_child(struct callchain_node *parent,
 	}
 }
 
-static int
+static enum match_result
 append_chain(struct callchain_node *root,
 	     struct callchain_cursor *cursor,
 	     u64 period);
@@ -583,17 +599,17 @@ append_chain_children(struct callchain_node *root,
 
 	/* lookup in childrens */
 	while (*p) {
-		s64 ret;
+		enum match_result ret;
 
 		parent = *p;
 		rnode = rb_entry(parent, struct callchain_node, rb_node_in);
 
 		/* If at least first entry matches, rely to children */
 		ret = append_chain(rnode, cursor, period);
-		if (ret == 0)
+		if (ret == MATCH_EQ)
 			goto inc_children_hit;
 
-		if (ret < 0)
+		if (ret == MATCH_LT)
 			p = &parent->rb_left;
 		else
 			p = &parent->rb_right;
@@ -611,7 +627,7 @@ inc_children_hit:
 	root->children_count++;
 }
 
-static int
+static enum match_result
 append_chain(struct callchain_node *root,
 	     struct callchain_cursor *cursor,
 	     u64 period)
@@ -620,7 +636,7 @@ append_chain(struct callchain_node *root,
 	u64 start = cursor->pos;
 	bool found = false;
 	u64 matches;
-	int cmp = 0;
+	enum match_result cmp = MATCH_ERROR;
 
 	/*
 	 * Lookup in the current node
@@ -636,7 +652,7 @@ append_chain(struct callchain_node *root,
 			break;
 
 		cmp = match_chain(node, cnode);
-		if (cmp)
+		if (cmp != MATCH_EQ)
 			break;
 
 		found = true;
@@ -646,7 +662,7 @@ append_chain(struct callchain_node *root,
 
 	/* matches not, relay no the parent */
 	if (!found) {
-		WARN_ONCE(!cmp, "Chain comparison error\n");
+		WARN_ONCE(cmp == MATCH_ERROR, "Chain comparison error\n");
 		return cmp;
 	}
 
@@ -655,20 +671,20 @@ append_chain(struct callchain_node *root,
 	/* we match only a part of the node. Split it and add the new chain */
 	if (matches < root->val_nr) {
 		split_add_child(root, cursor, cnode, start, matches, period);
-		return 0;
+		return MATCH_EQ;
 	}
 
 	/* we match 100% of the path, increment the hit */
 	if (matches == root->val_nr && cursor->pos == cursor->nr) {
 		root->hit += period;
 		root->count++;
-		return 0;
+		return MATCH_EQ;
 	}
 
 	/* We match the node and still have a part remaining */
 	append_chain_children(root, cursor, period);
 
-	return 0;
+	return MATCH_EQ;
 }
 
 int callchain_append(struct callchain_root *root,
-- 
2.7.0

[toc] | [prev] | [next] | [standalone]

#1327669 — [PATCH 05/23] perf callchain: Check return value of split_add_child()

FromNamhyung Kim <namhyung@kernel.org>
Date2016-02-05 14:20 +0100
Subject[PATCH 05/23] perf callchain: Check return value of split_add_child()
Message-ID<qYOsO-I2-19@gated-at.bofh.it>
In reply to#1327663
Now create_child() and add_child() return errors so check and pass it
to the caller.

Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
---
 tools/perf/util/callchain.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c
index dab2c1f1e86b..5259379892e1 100644
--- a/tools/perf/util/callchain.c
+++ b/tools/perf/util/callchain.c
@@ -508,7 +508,7 @@ static enum match_result match_chain(struct callchain_cursor_node *node,
  * give a part of its callchain to the created child.
  * Then create another child to host the given callchain of new branch
  */
-static void
+static int
 split_add_child(struct callchain_node *parent,
 		struct callchain_cursor *cursor,
 		struct callchain_list *to_split,
@@ -520,6 +520,8 @@ split_add_child(struct callchain_node *parent,
 
 	/* split */
 	new = create_child(parent, true);
+	if (new == NULL)
+		return -1;
 
 	/* split the callchain and move a part to the new child */
 	old_tail = parent->val.prev;
@@ -554,7 +556,7 @@ split_add_child(struct callchain_node *parent,
 		node = callchain_cursor_current(cursor);
 		new = add_child(parent, cursor, period);
 		if (new == NULL)
-			return;
+			return -1;
 
 		/*
 		 * This is second child since we moved parent's children
@@ -576,6 +578,7 @@ split_add_child(struct callchain_node *parent,
 		parent->hit = period;
 		parent->count = 1;
 	}
+	return 0;
 }
 
 static enum match_result
@@ -670,7 +673,10 @@ append_chain(struct callchain_node *root,
 
 	/* we match only a part of the node. Split it and add the new chain */
 	if (matches < root->val_nr) {
-		split_add_child(root, cursor, cnode, start, matches, period);
+		if (split_add_child(root, cursor, cnode, start, matches,
+				    period) < 0)
+			return MATCH_ERROR;
+
 		return MATCH_EQ;
 	}
 
-- 
2.7.0

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web