[PATCH 16/29] perf sched: Use cached rbtrees

From: Arnaldo Carvalho de Melo
Date: Fri Jan 25 2019 - 18:21:24 EST


From: Davidlohr Bueso <dave@xxxxxxxxxxxx>

At the cost of an extra pointer, we can avoid the O(logN) cost of
finding the first element in the tree (smallest node), which is
something heavily required for perf-sched.

Signed-off-by: Davidlohr Bueso <dbueso@xxxxxxx>
Cc: Jiri Olsa <jolsa@xxxxxxxxxx>
Cc: Namhyung Kim <namhyung@xxxxxxxxxx>
Link: http://lkml.kernel.org/r/20181206191819.30182-8-dave@xxxxxxxxxxxx
Signed-off-by: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
---
tools/perf/builtin-sched.c | 45 +++++++++++++++++++++-----------------
1 file changed, 25 insertions(+), 20 deletions(-)

diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
index 2e0f0c65964a..640558e9352e 100644
--- a/tools/perf/builtin-sched.c
+++ b/tools/perf/builtin-sched.c
@@ -213,7 +213,7 @@ struct perf_sched {
u64 all_runtime;
u64 all_count;
u64 cpu_last_switched[MAX_CPUS];
- struct rb_root atom_root, sorted_atom_root, merged_atom_root;
+ struct rb_root_cached atom_root, sorted_atom_root, merged_atom_root;
struct list_head sort_list, cmp_pid;
bool force;
bool skip_merge;
@@ -271,7 +271,7 @@ struct evsel_runtime {
struct idle_thread_runtime {
struct thread_runtime tr;
struct thread *last_thread;
- struct rb_root sorted_root;
+ struct rb_root_cached sorted_root;
struct callchain_root callchain;
struct callchain_cursor cursor;
};
@@ -950,10 +950,10 @@ thread_lat_cmp(struct list_head *list, struct work_atoms *l, struct work_atoms *
}

static struct work_atoms *
-thread_atoms_search(struct rb_root *root, struct thread *thread,
+thread_atoms_search(struct rb_root_cached *root, struct thread *thread,
struct list_head *sort_list)
{
- struct rb_node *node = root->rb_node;
+ struct rb_node *node = root->rb_root.rb_node;
struct work_atoms key = { .thread = thread };

while (node) {
@@ -976,10 +976,11 @@ thread_atoms_search(struct rb_root *root, struct thread *thread,
}

static void
-__thread_latency_insert(struct rb_root *root, struct work_atoms *data,
+__thread_latency_insert(struct rb_root_cached *root, struct work_atoms *data,
struct list_head *sort_list)
{
- struct rb_node **new = &(root->rb_node), *parent = NULL;
+ struct rb_node **new = &(root->rb_root.rb_node), *parent = NULL;
+ bool leftmost = true;

while (*new) {
struct work_atoms *this;
@@ -992,12 +993,14 @@ __thread_latency_insert(struct rb_root *root, struct work_atoms *data,

if (cmp > 0)
new = &((*new)->rb_left);
- else
+ else {
new = &((*new)->rb_right);
+ leftmost = false;
+ }
}

rb_link_node(&data->node, parent, new);
- rb_insert_color(&data->node, root);
+ rb_insert_color_cached(&data->node, root, leftmost);
}

static int thread_atoms_insert(struct perf_sched *sched, struct thread *thread)
@@ -1447,15 +1450,15 @@ static int sort_dimension__add(const char *tok, struct list_head *list)
static void perf_sched__sort_lat(struct perf_sched *sched)
{
struct rb_node *node;
- struct rb_root *root = &sched->atom_root;
+ struct rb_root_cached *root = &sched->atom_root;
again:
for (;;) {
struct work_atoms *data;
- node = rb_first(root);
+ node = rb_first_cached(root);
if (!node)
break;

- rb_erase(node, root);
+ rb_erase_cached(node, root);
data = rb_entry(node, struct work_atoms, node);
__thread_latency_insert(&sched->sorted_atom_root, data, &sched->sort_list);
}
@@ -2762,12 +2765,12 @@ static size_t callchain__fprintf_folded(FILE *fp, struct callchain_node *node)
return ret;
}

-static size_t timehist_print_idlehist_callchain(struct rb_root *root)
+static size_t timehist_print_idlehist_callchain(struct rb_root_cached *root)
{
size_t ret = 0;
FILE *fp = stdout;
struct callchain_node *chain;
- struct rb_node *rb_node = rb_first(root);
+ struct rb_node *rb_node = rb_first_cached(root);

printf(" %16s %8s %s\n", "Idle time (msec)", "Count", "Callchains");
printf(" %.16s %.8s %.50s\n", graph_dotted_line, graph_dotted_line,
@@ -2868,7 +2871,7 @@ static void timehist_print_summary(struct perf_sched *sched,
if (itr == NULL)
continue;

- callchain_param.sort(&itr->sorted_root, &itr->callchain,
+ callchain_param.sort(&itr->sorted_root.rb_root, &itr->callchain,
0, &callchain_param);

printf(" CPU %2d:", i);
@@ -3074,11 +3077,12 @@ static void print_bad_events(struct perf_sched *sched)
}
}

-static void __merge_work_atoms(struct rb_root *root, struct work_atoms *data)
+static void __merge_work_atoms(struct rb_root_cached *root, struct work_atoms *data)
{
- struct rb_node **new = &(root->rb_node), *parent = NULL;
+ struct rb_node **new = &(root->rb_root.rb_node), *parent = NULL;
struct work_atoms *this;
const char *comm = thread__comm_str(data->thread), *this_comm;
+ bool leftmost = true;

while (*new) {
int cmp;
@@ -3092,6 +3096,7 @@ static void __merge_work_atoms(struct rb_root *root, struct work_atoms *data)
new = &((*new)->rb_left);
} else if (cmp < 0) {
new = &((*new)->rb_right);
+ leftmost = false;
} else {
this->num_merged++;
this->total_runtime += data->total_runtime;
@@ -3109,7 +3114,7 @@ static void __merge_work_atoms(struct rb_root *root, struct work_atoms *data)

data->num_merged++;
rb_link_node(&data->node, parent, new);
- rb_insert_color(&data->node, root);
+ rb_insert_color_cached(&data->node, root, leftmost);
}

static void perf_sched__merge_lat(struct perf_sched *sched)
@@ -3120,8 +3125,8 @@ static void perf_sched__merge_lat(struct perf_sched *sched)
if (sched->skip_merge)
return;

- while ((node = rb_first(&sched->atom_root))) {
- rb_erase(node, &sched->atom_root);
+ while ((node = rb_first_cached(&sched->atom_root))) {
+ rb_erase_cached(node, &sched->atom_root);
data = rb_entry(node, struct work_atoms, node);
__merge_work_atoms(&sched->merged_atom_root, data);
}
@@ -3143,7 +3148,7 @@ static int perf_sched__lat(struct perf_sched *sched)
printf(" Task | Runtime ms | Switches | Average delay ms | Maximum delay ms | Maximum delay at |\n");
printf(" -----------------------------------------------------------------------------------------------------------------\n");

- next = rb_first(&sched->sorted_atom_root);
+ next = rb_first_cached(&sched->sorted_atom_root);

while (next) {
struct work_atoms *work_list;
--
2.20.1