* [GIT PULL 0/5] perf/core improvements
@ 2011-01-31 21:29 Arnaldo Carvalho de Melo
2011-01-31 21:29 ` [PATCH 1/5] perf evlist: Move evlist methods to evlist.c Arnaldo Carvalho de Melo
` (5 more replies)
0 siblings, 6 replies; 8+ messages in thread
From: Arnaldo Carvalho de Melo @ 2011-01-31 21:29 UTC (permalink / raw)
To: Ingo Molnar
Cc: linux-kernel, Arnaldo Carvalho de Melo, Frederic Weisbecker,
Ingo Molnar, Mike Galbraith, Paul Mackerras, Peter Zijlstra,
Stephane Eranian, Tom Zanussi, Arnaldo Carvalho de Melo
Hi Ingo,
Please consider pulling from:
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux-2.6 perf/core
Regards,
- Arnaldo
Arnaldo Carvalho de Melo (5):
perf evlist: Move evlist methods to evlist.c
perf evlist: Store pointer to the cpu and thread maps
perf top: Move display agnostic routines to util/top.[ch]
perf tools: Don't fallback to setup_pager unconditionally
perf top: Introduce slang based TUI
tools/perf/Makefile | 6 +
tools/perf/builtin-annotate.c | 2 +-
tools/perf/builtin-record.c | 44 ++--
tools/perf/builtin-report.c | 2 +-
tools/perf/builtin-stat.c | 45 ++--
tools/perf/builtin-test.c | 6 +-
tools/perf/builtin-top.c | 530 ++++++++++++-------------------------
tools/perf/python/twatch.py | 4 +-
tools/perf/util/cache.h | 7 +-
tools/perf/util/evlist.c | 185 +++++++++++++-
tools/perf/util/evlist.h | 28 ++-
tools/perf/util/evsel.c | 144 +----------
tools/perf/util/evsel.h | 4 -
tools/perf/util/python.c | 25 +-
tools/perf/util/top.c | 217 +++++++++++++++
tools/perf/util/top.h | 80 ++++++
tools/perf/util/ui/browsers/top.c | 136 ++++++++++
tools/perf/util/ui/setup.c | 5 +-
18 files changed, 887 insertions(+), 583 deletions(-)
create mode 100644 tools/perf/util/top.c
create mode 100644 tools/perf/util/top.h
create mode 100644 tools/perf/util/ui/browsers/top.c
^ permalink raw reply [flat|nested] 8+ messages in thread
* [PATCH 1/5] perf evlist: Move evlist methods to evlist.c
2011-01-31 21:29 [GIT PULL 0/5] perf/core improvements Arnaldo Carvalho de Melo
@ 2011-01-31 21:29 ` Arnaldo Carvalho de Melo
2011-01-31 21:29 ` [PATCH 2/5] perf evlist: Store pointer to the cpu and thread maps Arnaldo Carvalho de Melo
` (4 subsequent siblings)
5 siblings, 0 replies; 8+ messages in thread
From: Arnaldo Carvalho de Melo @ 2011-01-31 21:29 UTC (permalink / raw)
To: Ingo Molnar
Cc: linux-kernel, Arnaldo Carvalho de Melo, Frederic Weisbecker,
Ingo Molnar, Mike Galbraith, Paul Mackerras, Peter Zijlstra,
Stephane Eranian, Tom Zanussi
From: Arnaldo Carvalho de Melo <acme@redhat.com>
They were on evsel.c because they came from refactoring existing evsel
methods, so, to make reviewing the changes easier, I kept it there, now
its a plain move.
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/util/evlist.c | 142 +++++++++++++++++++++++++++++++++++++++++++++
tools/perf/util/evlist.h | 7 ++
tools/perf/util/evsel.c | 144 +++-------------------------------------------
tools/perf/util/evsel.h | 4 -
4 files changed, 158 insertions(+), 139 deletions(-)
diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
index 917fc18..dcd5932 100644
--- a/tools/perf/util/evlist.c
+++ b/tools/perf/util/evlist.c
@@ -1,11 +1,26 @@
+/*
+ * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com>
+ *
+ * Parts came from builtin-{top,stat,record}.c, see those files for further
+ * copyright notes.
+ *
+ * Released under the GPL v2. (and only v2, not any later version)
+ */
#include <poll.h>
+#include "cpumap.h"
+#include "thread_map.h"
#include "evlist.h"
#include "evsel.h"
#include "util.h"
+#include <sys/mman.h>
+
#include <linux/bitops.h>
#include <linux/hash.h>
+#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y))
+#define SID(e, x, y) xyarray__entry(e->id, x, y)
+
void perf_evlist__init(struct perf_evlist *evlist)
{
int i;
@@ -88,6 +103,30 @@ void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd)
evlist->nr_fds++;
}
+static int perf_evlist__id_hash(struct perf_evlist *evlist, struct perf_evsel *evsel,
+ int cpu, int thread, int fd)
+{
+ struct perf_sample_id *sid;
+ u64 read_data[4] = { 0, };
+ int hash, id_idx = 1; /* The first entry is the counter value */
+
+ if (!(evsel->attr.read_format & PERF_FORMAT_ID) ||
+ read(fd, &read_data, sizeof(read_data)) == -1)
+ return -1;
+
+ if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_ENABLED)
+ ++id_idx;
+ if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_RUNNING)
+ ++id_idx;
+
+ sid = SID(evsel, cpu, thread);
+ sid->id = read_data[id_idx];
+ sid->evsel = evsel;
+ hash = hash_64(sid->id, PERF_EVLIST__HLIST_BITS);
+ hlist_add_head(&sid->node, &evlist->heads[hash]);
+ return 0;
+}
+
struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id)
{
struct hlist_head *head;
@@ -173,3 +212,106 @@ union perf_event *perf_evlist__read_on_cpu(struct perf_evlist *evlist, int cpu)
return event;
}
+
+void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus)
+{
+ int cpu;
+
+ for (cpu = 0; cpu < ncpus; cpu++) {
+ if (evlist->mmap[cpu].base != NULL) {
+ munmap(evlist->mmap[cpu].base, evlist->mmap_len);
+ evlist->mmap[cpu].base = NULL;
+ }
+ }
+}
+
+int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus)
+{
+ evlist->mmap = zalloc(ncpus * sizeof(struct perf_mmap));
+ return evlist->mmap != NULL ? 0 : -ENOMEM;
+}
+
+static int __perf_evlist__mmap(struct perf_evlist *evlist, int cpu, int prot,
+ int mask, int fd)
+{
+ evlist->mmap[cpu].prev = 0;
+ evlist->mmap[cpu].mask = mask;
+ evlist->mmap[cpu].base = mmap(NULL, evlist->mmap_len, prot,
+ MAP_SHARED, fd, 0);
+ if (evlist->mmap[cpu].base == MAP_FAILED)
+ return -1;
+
+ perf_evlist__add_pollfd(evlist, fd);
+ return 0;
+}
+
+/** perf_evlist__mmap - Create per cpu maps to receive events
+ *
+ * @evlist - list of events
+ * @cpus - cpu map being monitored
+ * @threads - threads map being monitored
+ * @pages - map length in pages
+ * @overwrite - overwrite older events?
+ *
+ * If overwrite is false the user needs to signal event consuption using:
+ *
+ * struct perf_mmap *m = &evlist->mmap[cpu];
+ * unsigned int head = perf_mmap__read_head(m);
+ *
+ * perf_mmap__write_tail(m, head)
+ */
+int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus,
+ struct thread_map *threads, int pages, bool overwrite)
+{
+ unsigned int page_size = sysconf(_SC_PAGE_SIZE);
+ int mask = pages * page_size - 1, cpu;
+ struct perf_evsel *first_evsel, *evsel;
+ int thread, prot = PROT_READ | (overwrite ? 0 : PROT_WRITE);
+
+ if (evlist->mmap == NULL &&
+ perf_evlist__alloc_mmap(evlist, cpus->nr) < 0)
+ return -ENOMEM;
+
+ if (evlist->pollfd == NULL &&
+ perf_evlist__alloc_pollfd(evlist, cpus->nr, threads->nr) < 0)
+ return -ENOMEM;
+
+ evlist->overwrite = overwrite;
+ evlist->mmap_len = (pages + 1) * page_size;
+ first_evsel = list_entry(evlist->entries.next, struct perf_evsel, node);
+
+ list_for_each_entry(evsel, &evlist->entries, node) {
+ if ((evsel->attr.read_format & PERF_FORMAT_ID) &&
+ evsel->id == NULL &&
+ perf_evsel__alloc_id(evsel, cpus->nr, threads->nr) < 0)
+ return -ENOMEM;
+
+ for (cpu = 0; cpu < cpus->nr; cpu++) {
+ for (thread = 0; thread < threads->nr; thread++) {
+ int fd = FD(evsel, cpu, thread);
+
+ if (evsel->idx || thread) {
+ if (ioctl(fd, PERF_EVENT_IOC_SET_OUTPUT,
+ FD(first_evsel, cpu, 0)) != 0)
+ goto out_unmap;
+ } else if (__perf_evlist__mmap(evlist, cpu, prot, mask, fd) < 0)
+ goto out_unmap;
+
+ if ((evsel->attr.read_format & PERF_FORMAT_ID) &&
+ perf_evlist__id_hash(evlist, evsel, cpu, thread, fd) < 0)
+ goto out_unmap;
+ }
+ }
+ }
+
+ return 0;
+
+out_unmap:
+ for (cpu = 0; cpu < cpus->nr; cpu++) {
+ if (evlist->mmap[cpu].base != NULL) {
+ munmap(evlist->mmap[cpu].base, evlist->mmap_len);
+ evlist->mmap[cpu].base = NULL;
+ }
+ }
+ return -1;
+}
diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h
index 022ae40..85aca6e 100644
--- a/tools/perf/util/evlist.h
+++ b/tools/perf/util/evlist.h
@@ -6,6 +6,8 @@
#include "event.h"
struct pollfd;
+struct thread_map;
+struct cpu_map;
#define PERF_EVLIST__HLIST_BITS 8
#define PERF_EVLIST__HLIST_SIZE (1 << PERF_EVLIST__HLIST_BITS)
@@ -39,4 +41,9 @@ struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id);
union perf_event *perf_evlist__read_on_cpu(struct perf_evlist *self, int cpu);
+int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus);
+int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus,
+ struct thread_map *threads, int pages, bool overwrite);
+void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus);
+
#endif /* __PERF_EVLIST_H */
diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
index fddeb08..2720bc1 100644
--- a/tools/perf/util/evsel.c
+++ b/tools/perf/util/evsel.c
@@ -1,18 +1,19 @@
+/*
+ * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com>
+ *
+ * Parts came from builtin-{top,stat,record}.c, see those files for further
+ * copyright notes.
+ *
+ * Released under the GPL v2. (and only v2, not any later version)
+ */
+
#include "evsel.h"
#include "evlist.h"
-#include "../perf.h"
#include "util.h"
#include "cpumap.h"
#include "thread_map.h"
-#include <unistd.h>
-#include <sys/mman.h>
-
-#include <linux/bitops.h>
-#include <linux/hash.h>
-
#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y))
-#define SID(e, x, y) xyarray__entry(e->id, x, y)
void perf_evsel__init(struct perf_evsel *evsel,
struct perf_event_attr *attr, int idx)
@@ -74,24 +75,6 @@ void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads)
}
}
-void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus)
-{
- int cpu;
-
- for (cpu = 0; cpu < ncpus; cpu++) {
- if (evlist->mmap[cpu].base != NULL) {
- munmap(evlist->mmap[cpu].base, evlist->mmap_len);
- evlist->mmap[cpu].base = NULL;
- }
- }
-}
-
-int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus)
-{
- evlist->mmap = zalloc(ncpus * sizeof(struct perf_mmap));
- return evlist->mmap != NULL ? 0 : -ENOMEM;
-}
-
void perf_evsel__exit(struct perf_evsel *evsel)
{
assert(list_empty(&evsel->node));
@@ -258,115 +241,6 @@ int perf_evsel__open_per_thread(struct perf_evsel *evsel,
return __perf_evsel__open(evsel, &empty_cpu_map.map, threads, group, inherit);
}
-static int __perf_evlist__mmap(struct perf_evlist *evlist, int cpu, int prot,
- int mask, int fd)
-{
- evlist->mmap[cpu].prev = 0;
- evlist->mmap[cpu].mask = mask;
- evlist->mmap[cpu].base = mmap(NULL, evlist->mmap_len, prot,
- MAP_SHARED, fd, 0);
- if (evlist->mmap[cpu].base == MAP_FAILED)
- return -1;
-
- perf_evlist__add_pollfd(evlist, fd);
- return 0;
-}
-
-static int perf_evlist__id_hash(struct perf_evlist *evlist, struct perf_evsel *evsel,
- int cpu, int thread, int fd)
-{
- struct perf_sample_id *sid;
- u64 read_data[4] = { 0, };
- int hash, id_idx = 1; /* The first entry is the counter value */
-
- if (!(evsel->attr.read_format & PERF_FORMAT_ID) ||
- read(fd, &read_data, sizeof(read_data)) == -1)
- return -1;
-
- if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_ENABLED)
- ++id_idx;
- if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_RUNNING)
- ++id_idx;
-
- sid = SID(evsel, cpu, thread);
- sid->id = read_data[id_idx];
- sid->evsel = evsel;
- hash = hash_64(sid->id, PERF_EVLIST__HLIST_BITS);
- hlist_add_head(&sid->node, &evlist->heads[hash]);
- return 0;
-}
-
-/** perf_evlist__mmap - Create per cpu maps to receive events
- *
- * @evlist - list of events
- * @cpus - cpu map being monitored
- * @threads - threads map being monitored
- * @pages - map length in pages
- * @overwrite - overwrite older events?
- *
- * If overwrite is false the user needs to signal event consuption using:
- *
- * struct perf_mmap *m = &evlist->mmap[cpu];
- * unsigned int head = perf_mmap__read_head(m);
- *
- * perf_mmap__write_tail(m, head)
- */
-int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus,
- struct thread_map *threads, int pages, bool overwrite)
-{
- unsigned int page_size = sysconf(_SC_PAGE_SIZE);
- int mask = pages * page_size - 1, cpu;
- struct perf_evsel *first_evsel, *evsel;
- int thread, prot = PROT_READ | (overwrite ? 0 : PROT_WRITE);
-
- if (evlist->mmap == NULL &&
- perf_evlist__alloc_mmap(evlist, cpus->nr) < 0)
- return -ENOMEM;
-
- if (evlist->pollfd == NULL &&
- perf_evlist__alloc_pollfd(evlist, cpus->nr, threads->nr) < 0)
- return -ENOMEM;
-
- evlist->overwrite = overwrite;
- evlist->mmap_len = (pages + 1) * page_size;
- first_evsel = list_entry(evlist->entries.next, struct perf_evsel, node);
-
- list_for_each_entry(evsel, &evlist->entries, node) {
- if ((evsel->attr.read_format & PERF_FORMAT_ID) &&
- evsel->id == NULL &&
- perf_evsel__alloc_id(evsel, cpus->nr, threads->nr) < 0)
- return -ENOMEM;
-
- for (cpu = 0; cpu < cpus->nr; cpu++) {
- for (thread = 0; thread < threads->nr; thread++) {
- int fd = FD(evsel, cpu, thread);
-
- if (evsel->idx || thread) {
- if (ioctl(fd, PERF_EVENT_IOC_SET_OUTPUT,
- FD(first_evsel, cpu, 0)) != 0)
- goto out_unmap;
- } else if (__perf_evlist__mmap(evlist, cpu, prot, mask, fd) < 0)
- goto out_unmap;
-
- if ((evsel->attr.read_format & PERF_FORMAT_ID) &&
- perf_evlist__id_hash(evlist, evsel, cpu, thread, fd) < 0)
- goto out_unmap;
- }
- }
- }
-
- return 0;
-
-out_unmap:
- for (cpu = 0; cpu < cpus->nr; cpu++) {
- if (evlist->mmap[cpu].base != NULL) {
- munmap(evlist->mmap[cpu].base, evlist->mmap_len);
- evlist->mmap[cpu].base = NULL;
- }
- }
- return -1;
-}
-
static int perf_event__parse_id_sample(const union perf_event *event, u64 type,
struct perf_sample *sample)
{
diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h
index 7962e75..eecdc3a 100644
--- a/tools/perf/util/evsel.h
+++ b/tools/perf/util/evsel.h
@@ -60,7 +60,6 @@ void perf_evsel__delete(struct perf_evsel *evsel);
int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads);
int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads);
int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus);
-int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus);
void perf_evsel__free_fd(struct perf_evsel *evsel);
void perf_evsel__free_id(struct perf_evsel *evsel);
void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads);
@@ -71,9 +70,6 @@ int perf_evsel__open_per_thread(struct perf_evsel *evsel,
struct thread_map *threads, bool group, bool inherit);
int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus,
struct thread_map *threads, bool group, bool inherit);
-int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus,
- struct thread_map *threads, int pages, bool overwrite);
-void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus);
#define perf_evsel__match(evsel, t, c) \
(evsel->attr.type == PERF_TYPE_##t && \
--
1.6.2.5
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH 2/5] perf evlist: Store pointer to the cpu and thread maps
2011-01-31 21:29 [GIT PULL 0/5] perf/core improvements Arnaldo Carvalho de Melo
2011-01-31 21:29 ` [PATCH 1/5] perf evlist: Move evlist methods to evlist.c Arnaldo Carvalho de Melo
@ 2011-01-31 21:29 ` Arnaldo Carvalho de Melo
2011-01-31 21:29 ` [PATCH 3/5] perf top: Move display agnostic routines to util/top.[ch] Arnaldo Carvalho de Melo
` (3 subsequent siblings)
5 siblings, 0 replies; 8+ messages in thread
From: Arnaldo Carvalho de Melo @ 2011-01-31 21:29 UTC (permalink / raw)
To: Ingo Molnar
Cc: linux-kernel, Arnaldo Carvalho de Melo, Frederic Weisbecker,
Ingo Molnar, Mike Galbraith, Paul Mackerras, Peter Zijlstra,
Stephane Eranian, Tom Zanussi
From: Arnaldo Carvalho de Melo <acme@redhat.com>
So that we don't have to pass it around to the several methods that
needs it, simplifying usage.
There is one case where we don't have the thread/cpu map in advance,
which is in the parsing routines used by top, stat, record, that we have
to wait till all options are parsed to know if a cpu or thread list was
passed to then create those maps.
For that case consolidate the cpu and thread map creation via
perf_evlist__create_maps() out of the code in top and record, while also
providing a perf_evlist__set_maps() for cases where multiple evlists
share maps or for when maps that represent CPU sockets, for instance,
get crafted out of topology information or subsets of threads in a
particular application are to be monitored, providing more granularity
in specifying which cpus and threads to monitor.
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/builtin-record.c | 44 ++++++++++------------------
tools/perf/builtin-stat.c | 45 ++++++++++++++---------------
tools/perf/builtin-test.c | 6 ++--
tools/perf/builtin-top.c | 47 ++++++++++++------------------
tools/perf/python/twatch.py | 4 +-
tools/perf/util/evlist.c | 67 ++++++++++++++++++++++++++++++++-----------
tools/perf/util/evlist.h | 29 ++++++++++++++----
tools/perf/util/python.c | 25 +++++++++-------
8 files changed, 148 insertions(+), 119 deletions(-)
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index edc3555..07f8d6d 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -42,7 +42,6 @@ static u64 user_interval = ULLONG_MAX;
static u64 default_interval = 0;
static u64 sample_type;
-static struct cpu_map *cpus;
static unsigned int page_size;
static unsigned int mmap_pages = 128;
static unsigned int user_freq = UINT_MAX;
@@ -58,7 +57,6 @@ static bool sample_id_all_avail = true;
static bool system_wide = false;
static pid_t target_pid = -1;
static pid_t target_tid = -1;
-static struct thread_map *threads;
static pid_t child_pid = -1;
static bool no_inherit = false;
static enum write_mode_t write_mode = WRITE_FORCE;
@@ -189,7 +187,7 @@ static void create_counter(struct perf_evsel *evsel, int cpu)
int thread_index;
int ret;
- for (thread_index = 0; thread_index < threads->nr; thread_index++) {
+ for (thread_index = 0; thread_index < evsel_list->threads->nr; thread_index++) {
h_attr = get_header_attr(attr, evsel->idx);
if (h_attr == NULL)
die("nomem\n");
@@ -317,7 +315,8 @@ static void open_counters(struct perf_evlist *evlist)
retry_sample_id:
attr->sample_id_all = sample_id_all_avail ? 1 : 0;
try_again:
- if (perf_evsel__open(pos, cpus, threads, group, !no_inherit) < 0) {
+ if (perf_evsel__open(pos, evlist->cpus, evlist->threads, group,
+ !no_inherit) < 0) {
int err = errno;
if (err == EPERM || err == EACCES)
@@ -368,10 +367,10 @@ try_again:
}
}
- if (perf_evlist__mmap(evlist, cpus, threads, mmap_pages, false) < 0)
+ if (perf_evlist__mmap(evlist, mmap_pages, false) < 0)
die("failed to mmap with %d (%s)\n", errno, strerror(errno));
- for (cpu = 0; cpu < cpus->nr; ++cpu) {
+ for (cpu = 0; cpu < evsel_list->cpus->nr; ++cpu) {
list_for_each_entry(pos, &evlist->entries, node)
create_counter(pos, cpu);
}
@@ -450,7 +449,7 @@ static void mmap_read_all(void)
{
int i;
- for (i = 0; i < cpus->nr; i++) {
+ for (i = 0; i < evsel_list->cpus->nr; i++) {
if (evsel_list->mmap[i].base)
mmap_read(&evsel_list->mmap[i]);
}
@@ -584,7 +583,7 @@ static int __cmd_record(int argc, const char **argv)
}
if (!system_wide && target_tid == -1 && target_pid == -1)
- threads->map[0] = child_pid;
+ evsel_list->threads->map[0] = child_pid;
close(child_ready_pipe[1]);
close(go_pipe[0]);
@@ -718,12 +717,12 @@ static int __cmd_record(int argc, const char **argv)
}
if (done) {
- for (i = 0; i < cpus->nr; i++) {
+ for (i = 0; i < evsel_list->cpus->nr; i++) {
struct perf_evsel *pos;
list_for_each_entry(pos, &evsel_list->entries, node) {
for (thread = 0;
- thread < threads->nr;
+ thread < evsel_list->threads->nr;
thread++)
ioctl(FD(pos, i, thread),
PERF_EVENT_IOC_DISABLE);
@@ -816,7 +815,7 @@ int cmd_record(int argc, const char **argv, const char *prefix __used)
int err = -ENOMEM;
struct perf_evsel *pos;
- evsel_list = perf_evlist__new();
+ evsel_list = perf_evlist__new(NULL, NULL);
if (evsel_list == NULL)
return -ENOMEM;
@@ -850,28 +849,19 @@ int cmd_record(int argc, const char **argv, const char *prefix __used)
if (target_pid != -1)
target_tid = target_pid;
- threads = thread_map__new(target_pid, target_tid);
- if (threads == NULL) {
- pr_err("Problems finding threads of monitor\n");
- usage_with_options(record_usage, record_options);
- }
-
- if (target_tid != -1)
- cpus = cpu_map__dummy_new();
- else
- cpus = cpu_map__new(cpu_list);
-
- if (cpus == NULL)
+ if (perf_evlist__create_maps(evsel_list, target_pid,
+ target_tid, cpu_list) < 0)
usage_with_options(record_usage, record_options);
list_for_each_entry(pos, &evsel_list->entries, node) {
- if (perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0)
+ if (perf_evsel__alloc_fd(pos, evsel_list->cpus->nr,
+ evsel_list->threads->nr) < 0)
goto out_free_fd;
if (perf_header__push_event(pos->attr.config, event_name(pos)))
goto out_free_fd;
}
- if (perf_evlist__alloc_pollfd(evsel_list, cpus->nr, threads->nr) < 0)
+ if (perf_evlist__alloc_pollfd(evsel_list) < 0)
goto out_free_fd;
if (user_interval != ULLONG_MAX)
@@ -893,10 +883,8 @@ int cmd_record(int argc, const char **argv, const char *prefix __used)
}
err = __cmd_record(argc, argv);
-
out_free_fd:
- thread_map__delete(threads);
- threads = NULL;
+ perf_evlist__delete_maps(evsel_list);
out_symbol_exit:
symbol__exit();
return err;
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 8906adf..e0f9575 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -76,7 +76,6 @@ static struct perf_event_attr default_attrs[] = {
struct perf_evlist *evsel_list;
static bool system_wide = false;
-static struct cpu_map *cpus;
static int run_idx = 0;
static int run_count = 1;
@@ -85,7 +84,6 @@ static bool scale = true;
static bool no_aggr = false;
static pid_t target_pid = -1;
static pid_t target_tid = -1;
-static struct thread_map *threads;
static pid_t child_pid = -1;
static bool null_run = false;
static bool big_num = true;
@@ -170,7 +168,7 @@ static int create_perf_stat_counter(struct perf_evsel *evsel)
PERF_FORMAT_TOTAL_TIME_RUNNING;
if (system_wide)
- return perf_evsel__open_per_cpu(evsel, cpus, false, false);
+ return perf_evsel__open_per_cpu(evsel, evsel_list->cpus, false, false);
attr->inherit = !no_inherit;
if (target_pid == -1 && target_tid == -1) {
@@ -178,7 +176,7 @@ static int create_perf_stat_counter(struct perf_evsel *evsel)
attr->enable_on_exec = 1;
}
- return perf_evsel__open_per_thread(evsel, threads, false, false);
+ return perf_evsel__open_per_thread(evsel, evsel_list->threads, false, false);
}
/*
@@ -203,7 +201,8 @@ static int read_counter_aggr(struct perf_evsel *counter)
u64 *count = counter->counts->aggr.values;
int i;
- if (__perf_evsel__read(counter, cpus->nr, threads->nr, scale) < 0)
+ if (__perf_evsel__read(counter, evsel_list->cpus->nr,
+ evsel_list->threads->nr, scale) < 0)
return -1;
for (i = 0; i < 3; i++)
@@ -236,7 +235,7 @@ static int read_counter(struct perf_evsel *counter)
u64 *count;
int cpu;
- for (cpu = 0; cpu < cpus->nr; cpu++) {
+ for (cpu = 0; cpu < evsel_list->cpus->nr; cpu++) {
if (__perf_evsel__read_on_cpu(counter, cpu, 0, scale) < 0)
return -1;
@@ -301,7 +300,7 @@ static int run_perf_stat(int argc __used, const char **argv)
}
if (target_tid == -1 && target_pid == -1 && !system_wide)
- threads->map[0] = child_pid;
+ evsel_list->threads->map[0] = child_pid;
/*
* Wait for the child to be ready to exec.
@@ -353,12 +352,13 @@ static int run_perf_stat(int argc __used, const char **argv)
if (no_aggr) {
list_for_each_entry(counter, &evsel_list->entries, node) {
read_counter(counter);
- perf_evsel__close_fd(counter, cpus->nr, 1);
+ perf_evsel__close_fd(counter, evsel_list->cpus->nr, 1);
}
} else {
list_for_each_entry(counter, &evsel_list->entries, node) {
read_counter_aggr(counter);
- perf_evsel__close_fd(counter, cpus->nr, threads->nr);
+ perf_evsel__close_fd(counter, evsel_list->cpus->nr,
+ evsel_list->threads->nr);
}
}
@@ -386,7 +386,7 @@ static void nsec_printout(int cpu, struct perf_evsel *evsel, double avg)
if (no_aggr)
sprintf(cpustr, "CPU%*d%s",
csv_output ? 0 : -4,
- cpus->map[cpu], csv_sep);
+ evsel_list->cpus->map[cpu], csv_sep);
fprintf(stderr, fmt, cpustr, msecs, csv_sep, event_name(evsel));
@@ -414,7 +414,7 @@ static void abs_printout(int cpu, struct perf_evsel *evsel, double avg)
if (no_aggr)
sprintf(cpustr, "CPU%*d%s",
csv_output ? 0 : -4,
- cpus->map[cpu], csv_sep);
+ evsel_list->cpus->map[cpu], csv_sep);
else
cpu = 0;
@@ -500,14 +500,14 @@ static void print_counter(struct perf_evsel *counter)
u64 ena, run, val;
int cpu;
- for (cpu = 0; cpu < cpus->nr; cpu++) {
+ for (cpu = 0; cpu < evsel_list->cpus->nr; cpu++) {
val = counter->counts->cpu[cpu].val;
ena = counter->counts->cpu[cpu].ena;
run = counter->counts->cpu[cpu].run;
if (run == 0 || ena == 0) {
fprintf(stderr, "CPU%*d%s%*s%s%-24s",
csv_output ? 0 : -4,
- cpus->map[cpu], csv_sep,
+ evsel_list->cpus->map[cpu], csv_sep,
csv_output ? 0 : 18,
"<not counted>", csv_sep,
event_name(counter));
@@ -652,7 +652,7 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used)
setlocale(LC_ALL, "");
- evsel_list = perf_evlist__new();
+ evsel_list = perf_evlist__new(NULL, NULL);
if (evsel_list == NULL)
return -ENOMEM;
@@ -701,18 +701,18 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used)
if (target_pid != -1)
target_tid = target_pid;
- threads = thread_map__new(target_pid, target_tid);
- if (threads == NULL) {
+ evsel_list->threads = thread_map__new(target_pid, target_tid);
+ if (evsel_list->threads == NULL) {
pr_err("Problems finding threads of monitor\n");
usage_with_options(stat_usage, options);
}
if (system_wide)
- cpus = cpu_map__new(cpu_list);
+ evsel_list->cpus = cpu_map__new(cpu_list);
else
- cpus = cpu_map__dummy_new();
+ evsel_list->cpus = cpu_map__dummy_new();
- if (cpus == NULL) {
+ if (evsel_list->cpus == NULL) {
perror("failed to parse CPUs map");
usage_with_options(stat_usage, options);
return -1;
@@ -720,8 +720,8 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used)
list_for_each_entry(pos, &evsel_list->entries, node) {
if (perf_evsel__alloc_stat_priv(pos) < 0 ||
- perf_evsel__alloc_counts(pos, cpus->nr) < 0 ||
- perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0)
+ perf_evsel__alloc_counts(pos, evsel_list->cpus->nr) < 0 ||
+ perf_evsel__alloc_fd(pos, evsel_list->cpus->nr, evsel_list->threads->nr) < 0)
goto out_free_fd;
}
@@ -750,7 +750,6 @@ out_free_fd:
perf_evsel__free_stat_priv(pos);
perf_evlist__delete(evsel_list);
out:
- thread_map__delete(threads);
- threads = NULL;
+ perf_evlist__delete_maps(evsel_list);
return status;
}
diff --git a/tools/perf/builtin-test.c b/tools/perf/builtin-test.c
index 845b9bd..1b2106c 100644
--- a/tools/perf/builtin-test.c
+++ b/tools/perf/builtin-test.c
@@ -509,7 +509,7 @@ static int test__basic_mmap(void)
goto out_free_cpus;
}
- evlist = perf_evlist__new();
+ evlist = perf_evlist__new(cpus, threads);
if (evlist == NULL) {
pr_debug("perf_evlist__new\n");
goto out_free_cpus;
@@ -537,7 +537,7 @@ static int test__basic_mmap(void)
}
}
- if (perf_evlist__mmap(evlist, cpus, threads, 128, true) < 0) {
+ if (perf_evlist__mmap(evlist, 128, true) < 0) {
pr_debug("failed to mmap events: %d (%s)\n", errno,
strerror(errno));
goto out_close_fd;
@@ -579,7 +579,7 @@ static int test__basic_mmap(void)
err = 0;
out_munmap:
- perf_evlist__munmap(evlist, 1);
+ perf_evlist__munmap(evlist);
out_close_fd:
for (i = 0; i < nsyscalls; ++i)
perf_evsel__close_fd(evsels[i], 1, threads->nr);
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index 2f4d1f2..599036b 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -73,9 +73,7 @@ static int print_entries;
static int target_pid = -1;
static int target_tid = -1;
-static struct thread_map *threads;
static bool inherit = false;
-static struct cpu_map *cpus;
static int realtime_prio = 0;
static bool group = false;
static unsigned int page_size;
@@ -567,12 +565,13 @@ static void print_sym_table(struct perf_session *session)
printf(" (all");
if (cpu_list)
- printf(", CPU%s: %s)\n", cpus->nr > 1 ? "s" : "", cpu_list);
+ printf(", CPU%s: %s)\n", evsel_list->cpus->nr > 1 ? "s" : "", cpu_list);
else {
if (target_tid != -1)
printf(")\n");
else
- printf(", %d CPU%s)\n", cpus->nr, cpus->nr > 1 ? "s" : "");
+ printf(", %d CPU%s)\n", evsel_list->cpus->nr,
+ evsel_list->cpus->nr > 1 ? "s" : "");
}
printf("%-*.*s\n", win_width, win_width, graph_dotted_line);
@@ -1124,7 +1123,7 @@ static void perf_session__mmap_read(struct perf_session *self)
{
int i;
- for (i = 0; i < cpus->nr; i++)
+ for (i = 0; i < evsel_list->cpus->nr; i++)
perf_session__mmap_read_cpu(self, i);
}
@@ -1150,7 +1149,8 @@ static void start_counters(struct perf_evlist *evlist)
attr->mmap = 1;
try_again:
- if (perf_evsel__open(counter, cpus, threads, group, inherit) < 0) {
+ if (perf_evsel__open(counter, evsel_list->cpus,
+ evsel_list->threads, group, inherit) < 0) {
int err = errno;
if (err == EPERM || err == EACCES)
@@ -1181,7 +1181,7 @@ try_again:
}
}
- if (perf_evlist__mmap(evlist, cpus, threads, mmap_pages, false) < 0)
+ if (perf_evlist__mmap(evlist, mmap_pages, false) < 0)
die("failed to mmap with %d (%s)\n", errno, strerror(errno));
}
@@ -1296,7 +1296,7 @@ int cmd_top(int argc, const char **argv, const char *prefix __used)
struct perf_evsel *pos;
int status = -ENOMEM;
- evsel_list = perf_evlist__new();
+ evsel_list = perf_evlist__new(NULL, NULL);
if (evsel_list == NULL)
return -ENOMEM;
@@ -1306,15 +1306,6 @@ int cmd_top(int argc, const char **argv, const char *prefix __used)
if (argc)
usage_with_options(top_usage, options);
- if (target_pid != -1)
- target_tid = target_pid;
-
- threads = thread_map__new(target_pid, target_tid);
- if (threads == NULL) {
- pr_err("Problems finding threads of monitor\n");
- usage_with_options(top_usage, options);
- }
-
/* CPU and PID are mutually exclusive */
if (target_tid > 0 && cpu_list) {
printf("WARNING: PID switch overriding CPU\n");
@@ -1322,6 +1313,13 @@ int cmd_top(int argc, const char **argv, const char *prefix __used)
cpu_list = NULL;
}
+ if (target_pid != -1)
+ target_tid = target_pid;
+
+ if (perf_evlist__create_maps(evsel_list, target_pid,
+ target_tid, cpu_list) < 0)
+ usage_with_options(top_usage, options);
+
if (!evsel_list->nr_entries &&
perf_evlist__add_default(evsel_list) < 0) {
pr_err("Not enough memory for event selector list\n");
@@ -1343,16 +1341,9 @@ int cmd_top(int argc, const char **argv, const char *prefix __used)
exit(EXIT_FAILURE);
}
- if (target_tid != -1)
- cpus = cpu_map__dummy_new();
- else
- cpus = cpu_map__new(cpu_list);
-
- if (cpus == NULL)
- usage_with_options(top_usage, options);
-
list_for_each_entry(pos, &evsel_list->entries, node) {
- if (perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0)
+ if (perf_evsel__alloc_fd(pos, evsel_list->cpus->nr,
+ evsel_list->threads->nr) < 0)
goto out_free_fd;
/*
* Fill in the ones not specifically initialized via -c:
@@ -1363,8 +1354,8 @@ int cmd_top(int argc, const char **argv, const char *prefix __used)
pos->attr.sample_period = default_interval;
}
- if (perf_evlist__alloc_pollfd(evsel_list, cpus->nr, threads->nr) < 0 ||
- perf_evlist__alloc_mmap(evsel_list, cpus->nr) < 0)
+ if (perf_evlist__alloc_pollfd(evsel_list) < 0 ||
+ perf_evlist__alloc_mmap(evsel_list) < 0)
goto out_free_fd;
sym_evsel = list_entry(evsel_list->entries.next, struct perf_evsel, node);
diff --git a/tools/perf/python/twatch.py b/tools/perf/python/twatch.py
index 5e9f3b7..df638c4 100755
--- a/tools/perf/python/twatch.py
+++ b/tools/perf/python/twatch.py
@@ -23,9 +23,9 @@ def main():
sample_id_all = 1,
sample_type = perf.SAMPLE_PERIOD | perf.SAMPLE_TID | perf.SAMPLE_CPU | perf.SAMPLE_TID)
evsel.open(cpus = cpus, threads = threads);
- evlist = perf.evlist()
+ evlist = perf.evlist(cpus, threads)
evlist.add(evsel)
- evlist.mmap(cpus = cpus, threads = threads)
+ evlist.mmap()
while True:
evlist.poll(timeout = -1)
for cpu in cpus:
diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
index dcd5932..95b21fe 100644
--- a/tools/perf/util/evlist.c
+++ b/tools/perf/util/evlist.c
@@ -21,21 +21,24 @@
#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y))
#define SID(e, x, y) xyarray__entry(e->id, x, y)
-void perf_evlist__init(struct perf_evlist *evlist)
+void perf_evlist__init(struct perf_evlist *evlist, struct cpu_map *cpus,
+ struct thread_map *threads)
{
int i;
for (i = 0; i < PERF_EVLIST__HLIST_SIZE; ++i)
INIT_HLIST_HEAD(&evlist->heads[i]);
INIT_LIST_HEAD(&evlist->entries);
+ perf_evlist__set_maps(evlist, cpus, threads);
}
-struct perf_evlist *perf_evlist__new(void)
+struct perf_evlist *perf_evlist__new(struct cpu_map *cpus,
+ struct thread_map *threads)
{
struct perf_evlist *evlist = zalloc(sizeof(*evlist));
if (evlist != NULL)
- perf_evlist__init(evlist);
+ perf_evlist__init(evlist, cpus, threads);
return evlist;
}
@@ -88,9 +91,9 @@ int perf_evlist__add_default(struct perf_evlist *evlist)
return 0;
}
-int perf_evlist__alloc_pollfd(struct perf_evlist *evlist, int ncpus, int nthreads)
+int perf_evlist__alloc_pollfd(struct perf_evlist *evlist)
{
- int nfds = ncpus * nthreads * evlist->nr_entries;
+ int nfds = evlist->cpus->nr * evlist->threads->nr * evlist->nr_entries;
evlist->pollfd = malloc(sizeof(struct pollfd) * nfds);
return evlist->pollfd != NULL ? 0 : -ENOMEM;
}
@@ -213,11 +216,11 @@ union perf_event *perf_evlist__read_on_cpu(struct perf_evlist *evlist, int cpu)
return event;
}
-void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus)
+void perf_evlist__munmap(struct perf_evlist *evlist)
{
int cpu;
- for (cpu = 0; cpu < ncpus; cpu++) {
+ for (cpu = 0; cpu < evlist->cpus->nr; cpu++) {
if (evlist->mmap[cpu].base != NULL) {
munmap(evlist->mmap[cpu].base, evlist->mmap_len);
evlist->mmap[cpu].base = NULL;
@@ -225,9 +228,9 @@ void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus)
}
}
-int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus)
+int perf_evlist__alloc_mmap(struct perf_evlist *evlist)
{
- evlist->mmap = zalloc(ncpus * sizeof(struct perf_mmap));
+ evlist->mmap = zalloc(evlist->cpus->nr * sizeof(struct perf_mmap));
return evlist->mmap != NULL ? 0 : -ENOMEM;
}
@@ -248,8 +251,6 @@ static int __perf_evlist__mmap(struct perf_evlist *evlist, int cpu, int prot,
/** perf_evlist__mmap - Create per cpu maps to receive events
*
* @evlist - list of events
- * @cpus - cpu map being monitored
- * @threads - threads map being monitored
* @pages - map length in pages
* @overwrite - overwrite older events?
*
@@ -259,21 +260,22 @@ static int __perf_evlist__mmap(struct perf_evlist *evlist, int cpu, int prot,
* unsigned int head = perf_mmap__read_head(m);
*
* perf_mmap__write_tail(m, head)
+ *
+ * Using perf_evlist__read_on_cpu does this automatically.
*/
-int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus,
- struct thread_map *threads, int pages, bool overwrite)
+int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite)
{
unsigned int page_size = sysconf(_SC_PAGE_SIZE);
int mask = pages * page_size - 1, cpu;
struct perf_evsel *first_evsel, *evsel;
+ const struct cpu_map *cpus = evlist->cpus;
+ const struct thread_map *threads = evlist->threads;
int thread, prot = PROT_READ | (overwrite ? 0 : PROT_WRITE);
- if (evlist->mmap == NULL &&
- perf_evlist__alloc_mmap(evlist, cpus->nr) < 0)
+ if (evlist->mmap == NULL && perf_evlist__alloc_mmap(evlist) < 0)
return -ENOMEM;
- if (evlist->pollfd == NULL &&
- perf_evlist__alloc_pollfd(evlist, cpus->nr, threads->nr) < 0)
+ if (evlist->pollfd == NULL && perf_evlist__alloc_pollfd(evlist) < 0)
return -ENOMEM;
evlist->overwrite = overwrite;
@@ -315,3 +317,34 @@ out_unmap:
}
return -1;
}
+
+int perf_evlist__create_maps(struct perf_evlist *evlist, pid_t target_pid,
+ pid_t target_tid, const char *cpu_list)
+{
+ evlist->threads = thread_map__new(target_pid, target_tid);
+
+ if (evlist->threads == NULL)
+ return -1;
+
+ if (target_tid != -1)
+ evlist->cpus = cpu_map__dummy_new();
+ else
+ evlist->cpus = cpu_map__new(cpu_list);
+
+ if (evlist->cpus == NULL)
+ goto out_delete_threads;
+
+ return 0;
+
+out_delete_threads:
+ thread_map__delete(evlist->threads);
+ return -1;
+}
+
+void perf_evlist__delete_maps(struct perf_evlist *evlist)
+{
+ cpu_map__delete(evlist->cpus);
+ thread_map__delete(evlist->threads);
+ evlist->cpus = NULL;
+ evlist->threads = NULL;
+}
diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h
index 85aca6e..c988405 100644
--- a/tools/perf/util/evlist.h
+++ b/tools/perf/util/evlist.h
@@ -22,28 +22,43 @@ struct perf_evlist {
union perf_event event_copy;
struct perf_mmap *mmap;
struct pollfd *pollfd;
+ struct thread_map *threads;
+ struct cpu_map *cpus;
};
struct perf_evsel;
-struct perf_evlist *perf_evlist__new(void);
-void perf_evlist__init(struct perf_evlist *evlist);
+struct perf_evlist *perf_evlist__new(struct cpu_map *cpus,
+ struct thread_map *threads);
+void perf_evlist__init(struct perf_evlist *evlist, struct cpu_map *cpus,
+ struct thread_map *threads);
void perf_evlist__exit(struct perf_evlist *evlist);
void perf_evlist__delete(struct perf_evlist *evlist);
void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry);
int perf_evlist__add_default(struct perf_evlist *evlist);
-int perf_evlist__alloc_pollfd(struct perf_evlist *evlist, int ncpus, int nthreads);
+int perf_evlist__alloc_pollfd(struct perf_evlist *evlist);
void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd);
struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id);
union perf_event *perf_evlist__read_on_cpu(struct perf_evlist *self, int cpu);
-int perf_evlist__alloc_mmap(struct perf_evlist *evlist, int ncpus);
-int perf_evlist__mmap(struct perf_evlist *evlist, struct cpu_map *cpus,
- struct thread_map *threads, int pages, bool overwrite);
-void perf_evlist__munmap(struct perf_evlist *evlist, int ncpus);
+int perf_evlist__alloc_mmap(struct perf_evlist *evlist);
+int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite);
+void perf_evlist__munmap(struct perf_evlist *evlist);
+
+static inline void perf_evlist__set_maps(struct perf_evlist *evlist,
+ struct cpu_map *cpus,
+ struct thread_map *threads)
+{
+ evlist->cpus = cpus;
+ evlist->threads = threads;
+}
+
+int perf_evlist__create_maps(struct perf_evlist *evlist, pid_t target_pid,
+ pid_t target_tid, const char *cpu_list);
+void perf_evlist__delete_maps(struct perf_evlist *evlist);
#endif /* __PERF_EVLIST_H */
diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c
index 88d4789..d2d5217 100644
--- a/tools/perf/util/python.c
+++ b/tools/perf/util/python.c
@@ -553,7 +553,16 @@ struct pyrf_evlist {
static int pyrf_evlist__init(struct pyrf_evlist *pevlist,
PyObject *args, PyObject *kwargs)
{
- perf_evlist__init(&pevlist->evlist);
+ PyObject *pcpus = NULL, *pthreads = NULL;
+ struct cpu_map *cpus;
+ struct thread_map *threads;
+
+ if (!PyArg_ParseTuple(args, "OO", &pcpus, &pthreads))
+ return -1;
+
+ threads = ((struct pyrf_thread_map *)pthreads)->threads;
+ cpus = ((struct pyrf_cpu_map *)pcpus)->cpus;
+ perf_evlist__init(&pevlist->evlist, cpus, threads);
return 0;
}
@@ -567,21 +576,15 @@ static PyObject *pyrf_evlist__mmap(struct pyrf_evlist *pevlist,
PyObject *args, PyObject *kwargs)
{
struct perf_evlist *evlist = &pevlist->evlist;
- PyObject *pcpus = NULL, *pthreads = NULL;
- struct cpu_map *cpus = NULL;
- struct thread_map *threads = NULL;
- static char *kwlist[] = {"cpus", "threads", "pages", "overwrite",
+ static char *kwlist[] = {"pages", "overwrite",
NULL, NULL};
int pages = 128, overwrite = false;
- if (!PyArg_ParseTupleAndKeywords(args, kwargs, "OO|ii", kwlist,
- &pcpus, &pthreads, &pages, &overwrite))
+ if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|ii", kwlist,
+ &pages, &overwrite))
return NULL;
- threads = ((struct pyrf_thread_map *)pthreads)->threads;
- cpus = ((struct pyrf_cpu_map *)pcpus)->cpus;
-
- if (perf_evlist__mmap(evlist, cpus, threads, pages, overwrite) < 0) {
+ if (perf_evlist__mmap(evlist, pages, overwrite) < 0) {
PyErr_SetFromErrno(PyExc_OSError);
return NULL;
}
--
1.6.2.5
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH 3/5] perf top: Move display agnostic routines to util/top.[ch]
2011-01-31 21:29 [GIT PULL 0/5] perf/core improvements Arnaldo Carvalho de Melo
2011-01-31 21:29 ` [PATCH 1/5] perf evlist: Move evlist methods to evlist.c Arnaldo Carvalho de Melo
2011-01-31 21:29 ` [PATCH 2/5] perf evlist: Store pointer to the cpu and thread maps Arnaldo Carvalho de Melo
@ 2011-01-31 21:29 ` Arnaldo Carvalho de Melo
2011-01-31 21:29 ` [PATCH 4/5] perf tools: Don't fallback to setup_pager unconditionally Arnaldo Carvalho de Melo
` (2 subsequent siblings)
5 siblings, 0 replies; 8+ messages in thread
From: Arnaldo Carvalho de Melo @ 2011-01-31 21:29 UTC (permalink / raw)
To: Ingo Molnar
Cc: linux-kernel, Arnaldo Carvalho de Melo, Frederic Weisbecker,
Ingo Molnar, Mike Galbraith, Paul Mackerras, Peter Zijlstra,
Stephane Eranian, Tom Zanussi
From: Arnaldo Carvalho de Melo <acme@redhat.com>
Paving the way for a slang browser a la 'perf report --tui'.
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/Makefile | 2 +
tools/perf/builtin-top.c | 482 +++++++++++++---------------------------------
tools/perf/util/top.c | 212 ++++++++++++++++++++
tools/perf/util/top.h | 67 +++++++
4 files changed, 418 insertions(+), 345 deletions(-)
create mode 100644 tools/perf/util/top.c
create mode 100644 tools/perf/util/top.h
diff --git a/tools/perf/Makefile b/tools/perf/Makefile
index 36ff73c..edc660e 100644
--- a/tools/perf/Makefile
+++ b/tools/perf/Makefile
@@ -437,6 +437,7 @@ LIB_H += util/probe-finder.h
LIB_H += util/probe-event.h
LIB_H += util/pstack.h
LIB_H += util/cpumap.h
+LIB_H += util/top.h
LIB_H += $(ARCH_INCLUDE)
LIB_OBJS += $(OUTPUT)util/abspath.o
@@ -464,6 +465,7 @@ LIB_OBJS += $(OUTPUT)util/strbuf.o
LIB_OBJS += $(OUTPUT)util/string.o
LIB_OBJS += $(OUTPUT)util/strlist.o
LIB_OBJS += $(OUTPUT)util/strfilter.o
+LIB_OBJS += $(OUTPUT)util/top.o
LIB_OBJS += $(OUTPUT)util/usage.o
LIB_OBJS += $(OUTPUT)util/wrapper.o
LIB_OBJS += $(OUTPUT)util/sigchain.o
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index 599036b..3c9ba94 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -27,6 +27,7 @@
#include "util/symbol.h"
#include "util/thread.h"
#include "util/thread_map.h"
+#include "util/top.h"
#include "util/util.h"
#include <linux/rbtree.h>
#include "util/parse-options.h"
@@ -47,7 +48,6 @@
#include <errno.h>
#include <time.h>
#include <sched.h>
-#include <pthread.h>
#include <sys/syscall.h>
#include <sys/ioctl.h>
@@ -62,75 +62,35 @@
#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y))
-struct perf_evlist *evsel_list;
+static struct perf_top top = {
+ .count_filter = 5,
+ .delay_secs = 2,
+ .display_weighted = -1,
+ .target_pid = -1,
+ .target_tid = -1,
+ .active_symbols = LIST_HEAD_INIT(top.active_symbols),
+ .active_symbols_lock = PTHREAD_MUTEX_INITIALIZER,
+ .freq = 1000, /* 1 KHz */
+};
static bool system_wide = false;
static int default_interval = 0;
-static int count_filter = 5;
-static int print_entries;
-
-static int target_pid = -1;
-static int target_tid = -1;
static bool inherit = false;
static int realtime_prio = 0;
static bool group = false;
static unsigned int page_size;
static unsigned int mmap_pages = 128;
-static int freq = 1000; /* 1 KHz */
-static int delay_secs = 2;
-static bool zero = false;
static bool dump_symtab = false;
-static bool hide_kernel_symbols = false;
-static bool hide_user_symbols = false;
static struct winsize winsize;
-/*
- * Source
- */
-
-struct source_line {
- u64 eip;
- unsigned long count[MAX_COUNTERS];
- char *line;
- struct source_line *next;
-};
-
static const char *sym_filter = NULL;
struct sym_entry *sym_filter_entry = NULL;
struct sym_entry *sym_filter_entry_sched = NULL;
static int sym_pcnt_filter = 5;
-static int sym_counter = 0;
-static struct perf_evsel *sym_evsel = NULL;
-static int display_weighted = -1;
-static const char *cpu_list;
-
-/*
- * Symbols
- */
-
-struct sym_entry_source {
- struct source_line *source;
- struct source_line *lines;
- struct source_line **lines_tail;
- pthread_mutex_t lock;
-};
-
-struct sym_entry {
- struct rb_node rb_node;
- struct list_head node;
- unsigned long snap_count;
- double weight;
- int skip;
- u16 name_len;
- u8 origin;
- struct map *map;
- struct sym_entry_source *src;
- unsigned long count[0];
-};
/*
* Source functions
@@ -165,10 +125,10 @@ void get_term_dimensions(struct winsize *ws)
static void update_print_entries(struct winsize *ws)
{
- print_entries = ws->ws_row;
+ top.print_entries = ws->ws_row;
- if (print_entries > 9)
- print_entries -= 9;
+ if (top.print_entries > 9)
+ top.print_entries -= 9;
}
static void sig_winch_handler(int sig __used)
@@ -269,7 +229,7 @@ static void __zero_source_counters(struct sym_entry *syme)
line = syme->src->lines;
while (line) {
- for (i = 0; i < evsel_list->nr_entries; i++)
+ for (i = 0; i < top.evlist->nr_entries; i++)
line->count[i] = 0;
line = line->next;
}
@@ -331,9 +291,9 @@ static void show_lines(struct source_line *queue, int count, int total)
line = queue;
for (i = 0; i < count; i++) {
- float pcnt = 100.0*(float)line->count[sym_counter]/(float)total;
+ float pcnt = 100.0*(float)line->count[top.sym_counter]/(float)total;
- printf("%8li %4.1f%%\t%s\n", line->count[sym_counter], pcnt, line->line);
+ printf("%8li %4.1f%%\t%s\n", line->count[top.sym_counter], pcnt, line->line);
line = line->next;
}
}
@@ -358,13 +318,13 @@ static void show_details(struct sym_entry *syme)
return;
symbol = sym_entry__symbol(syme);
- printf("Showing %s for %s\n", event_name(sym_evsel), symbol->name);
+ printf("Showing %s for %s\n", event_name(top.sym_evsel), symbol->name);
printf(" Events Pcnt (>=%d%%)\n", sym_pcnt_filter);
pthread_mutex_lock(&syme->src->lock);
line = syme->src->source;
while (line) {
- total += line->count[sym_counter];
+ total += line->count[top.sym_counter];
line = line->next;
}
@@ -376,10 +336,10 @@ static void show_details(struct sym_entry *syme)
line_queue = line;
line_queue_count++;
- if (line->count[sym_counter])
- pcnt = 100.0 * line->count[sym_counter] / (float)total;
+ if (line->count[top.sym_counter])
+ pcnt = 100.0 * line->count[top.sym_counter] / (float)total;
if (pcnt >= (float)sym_pcnt_filter) {
- if (displayed <= print_entries)
+ if (displayed <= top.print_entries)
show_lines(line_queue, line_queue_count, total);
else more++;
displayed += line_queue_count;
@@ -390,7 +350,7 @@ static void show_details(struct sym_entry *syme)
line_queue_count--;
}
- line->count[sym_counter] = zero ? 0 : line->count[sym_counter] * 7 / 8;
+ line->count[top.sym_counter] = top.zero ? 0 : line->count[top.sym_counter] * 7 / 8;
line = line->next;
}
pthread_mutex_unlock(&syme->src->lock);
@@ -398,181 +358,30 @@ static void show_details(struct sym_entry *syme)
printf("%d lines not displayed, maybe increase display entries [e]\n", more);
}
-/*
- * Symbols will be added here in perf_event__process_sample and will get out
- * after decayed.
- */
-static LIST_HEAD(active_symbols);
-static pthread_mutex_t active_symbols_lock = PTHREAD_MUTEX_INITIALIZER;
-
-/*
- * Ordering weight: count-1 * count-2 * ... / count-n
- */
-static double sym_weight(const struct sym_entry *sym)
-{
- double weight = sym->snap_count;
- int counter;
-
- if (!display_weighted)
- return weight;
-
- for (counter = 1; counter < evsel_list->nr_entries - 1; counter++)
- weight *= sym->count[counter];
-
- weight /= (sym->count[counter] + 1);
-
- return weight;
-}
-
-static long samples;
-static long kernel_samples, us_samples;
-static long exact_samples;
-static long guest_us_samples, guest_kernel_samples;
static const char CONSOLE_CLEAR[] = "^[[H^[[2J";
static void __list_insert_active_sym(struct sym_entry *syme)
{
- list_add(&syme->node, &active_symbols);
-}
-
-static void list_remove_active_sym(struct sym_entry *syme)
-{
- pthread_mutex_lock(&active_symbols_lock);
- list_del_init(&syme->node);
- pthread_mutex_unlock(&active_symbols_lock);
-}
-
-static void rb_insert_active_sym(struct rb_root *tree, struct sym_entry *se)
-{
- struct rb_node **p = &tree->rb_node;
- struct rb_node *parent = NULL;
- struct sym_entry *iter;
-
- while (*p != NULL) {
- parent = *p;
- iter = rb_entry(parent, struct sym_entry, rb_node);
-
- if (se->weight > iter->weight)
- p = &(*p)->rb_left;
- else
- p = &(*p)->rb_right;
- }
-
- rb_link_node(&se->rb_node, parent, p);
- rb_insert_color(&se->rb_node, tree);
+ list_add(&syme->node, &top.active_symbols);
}
static void print_sym_table(struct perf_session *session)
{
- int printed = 0, j;
- struct perf_evsel *counter;
- int snap = !display_weighted ? sym_counter : 0;
- float samples_per_sec = samples/delay_secs;
- float ksamples_per_sec = kernel_samples/delay_secs;
- float us_samples_per_sec = (us_samples)/delay_secs;
- float guest_kernel_samples_per_sec = (guest_kernel_samples)/delay_secs;
- float guest_us_samples_per_sec = (guest_us_samples)/delay_secs;
- float esamples_percent = (100.0*exact_samples)/samples;
- float sum_ksamples = 0.0;
- struct sym_entry *syme, *n;
- struct rb_root tmp = RB_ROOT;
+ char bf[160];
+ int printed = 0;
struct rb_node *nd;
- int sym_width = 0, dso_width = 0, dso_short_width = 0;
+ struct sym_entry *syme;
+ struct rb_root tmp = RB_ROOT;
const int win_width = winsize.ws_col - 1;
-
- samples = us_samples = kernel_samples = exact_samples = 0;
- guest_kernel_samples = guest_us_samples = 0;
-
- /* Sort the active symbols */
- pthread_mutex_lock(&active_symbols_lock);
- syme = list_entry(active_symbols.next, struct sym_entry, node);
- pthread_mutex_unlock(&active_symbols_lock);
-
- list_for_each_entry_safe_from(syme, n, &active_symbols, node) {
- syme->snap_count = syme->count[snap];
- if (syme->snap_count != 0) {
-
- if ((hide_user_symbols &&
- syme->origin == PERF_RECORD_MISC_USER) ||
- (hide_kernel_symbols &&
- syme->origin == PERF_RECORD_MISC_KERNEL)) {
- list_remove_active_sym(syme);
- continue;
- }
- syme->weight = sym_weight(syme);
- rb_insert_active_sym(&tmp, syme);
- sum_ksamples += syme->snap_count;
-
- for (j = 0; j < evsel_list->nr_entries; j++)
- syme->count[j] = zero ? 0 : syme->count[j] * 7 / 8;
- } else
- list_remove_active_sym(syme);
- }
+ int sym_width, dso_width, dso_short_width;
+ float sum_ksamples = perf_top__decay_samples(&top, &tmp);
puts(CONSOLE_CLEAR);
- if (!perf_guest) {
- printf(" PerfTop:%8.0f irqs/sec kernel:%4.1f%%"
- " exact: %4.1f%% [",
- samples_per_sec,
- 100.0 - (100.0 * ((samples_per_sec - ksamples_per_sec) /
- samples_per_sec)),
- esamples_percent);
- } else {
- printf(" PerfTop:%8.0f irqs/sec kernel:%4.1f%% us:%4.1f%%"
- " guest kernel:%4.1f%% guest us:%4.1f%%"
- " exact: %4.1f%% [",
- samples_per_sec,
- 100.0 - (100.0 * ((samples_per_sec-ksamples_per_sec) /
- samples_per_sec)),
- 100.0 - (100.0 * ((samples_per_sec-us_samples_per_sec) /
- samples_per_sec)),
- 100.0 - (100.0 * ((samples_per_sec -
- guest_kernel_samples_per_sec) /
- samples_per_sec)),
- 100.0 - (100.0 * ((samples_per_sec -
- guest_us_samples_per_sec) /
- samples_per_sec)),
- esamples_percent);
- }
+ perf_top__header_snprintf(&top, bf, sizeof(bf));
+ printf("%s\n", bf);
- if (evsel_list->nr_entries == 1 || !display_weighted) {
- struct perf_evsel *first;
- first = list_entry(evsel_list->entries.next, struct perf_evsel, node);
- printf("%" PRIu64, (uint64_t)first->attr.sample_period);
- if (freq)
- printf("Hz ");
- else
- printf(" ");
- }
-
- if (!display_weighted)
- printf("%s", event_name(sym_evsel));
- else list_for_each_entry(counter, &evsel_list->entries, node) {
- if (counter->idx)
- printf("/");
-
- printf("%s", event_name(counter));
- }
-
- printf( "], ");
-
- if (target_pid != -1)
- printf(" (target_pid: %d", target_pid);
- else if (target_tid != -1)
- printf(" (target_tid: %d", target_tid);
- else
- printf(" (all");
-
- if (cpu_list)
- printf(", CPU%s: %s)\n", evsel_list->cpus->nr > 1 ? "s" : "", cpu_list);
- else {
- if (target_tid != -1)
- printf(")\n");
- else
- printf(", %d CPU%s)\n", evsel_list->cpus->nr,
- evsel_list->cpus->nr > 1 ? "s" : "");
- }
+ perf_top__reset_sample_counters(&top);
printf("%-*.*s\n", win_width, win_width, graph_dotted_line);
@@ -587,26 +396,8 @@ static void print_sym_table(struct perf_session *session)
return;
}
- /*
- * Find the longest symbol name that will be displayed
- */
- for (nd = rb_first(&tmp); nd; nd = rb_next(nd)) {
- syme = rb_entry(nd, struct sym_entry, rb_node);
- if (++printed > print_entries ||
- (int)syme->snap_count < count_filter)
- continue;
-
- if (syme->map->dso->long_name_len > dso_width)
- dso_width = syme->map->dso->long_name_len;
-
- if (syme->map->dso->short_name_len > dso_short_width)
- dso_short_width = syme->map->dso->short_name_len;
-
- if (syme->name_len > sym_width)
- sym_width = syme->name_len;
- }
-
- printed = 0;
+ perf_top__find_widths(&top, &tmp, &dso_width, &dso_short_width,
+ &sym_width);
if (sym_width + dso_width > winsize.ws_col - 29) {
dso_width = dso_short_width;
@@ -614,7 +405,7 @@ static void print_sym_table(struct perf_session *session)
sym_width = winsize.ws_col - dso_width - 29;
}
putchar('\n');
- if (evsel_list->nr_entries == 1)
+ if (top.evlist->nr_entries == 1)
printf(" samples pcnt");
else
printf(" weight samples pcnt");
@@ -623,7 +414,7 @@ static void print_sym_table(struct perf_session *session)
printf(" RIP ");
printf(" %-*.*s DSO\n", sym_width, sym_width, "function");
printf(" %s _______ _____",
- evsel_list->nr_entries == 1 ? " " : "______");
+ top.evlist->nr_entries == 1 ? " " : "______");
if (verbose)
printf(" ________________");
printf(" %-*.*s", sym_width, sym_width, graph_line);
@@ -636,13 +427,14 @@ static void print_sym_table(struct perf_session *session)
syme = rb_entry(nd, struct sym_entry, rb_node);
sym = sym_entry__symbol(syme);
- if (++printed > print_entries || (int)syme->snap_count < count_filter)
+ if (++printed > top.print_entries ||
+ (int)syme->snap_count < top.count_filter)
continue;
pcnt = 100.0 - (100.0 * ((sum_ksamples - syme->snap_count) /
sum_ksamples));
- if (evsel_list->nr_entries == 1 || !display_weighted)
+ if (top.evlist->nr_entries == 1 || !top.display_weighted)
printf("%20.2f ", syme->weight);
else
printf("%9.1f %10ld ", syme->weight, syme->snap_count);
@@ -715,11 +507,11 @@ static void prompt_symbol(struct sym_entry **target, const char *msg)
if (p)
*p = 0;
- pthread_mutex_lock(&active_symbols_lock);
- syme = list_entry(active_symbols.next, struct sym_entry, node);
- pthread_mutex_unlock(&active_symbols_lock);
+ pthread_mutex_lock(&top.active_symbols_lock);
+ syme = list_entry(top.active_symbols.next, struct sym_entry, node);
+ pthread_mutex_unlock(&top.active_symbols_lock);
- list_for_each_entry_safe_from(syme, n, &active_symbols, node) {
+ list_for_each_entry_safe_from(syme, n, &top.active_symbols, node) {
struct symbol *sym = sym_entry__symbol(syme);
if (!strcmp(buf, sym->name)) {
@@ -749,28 +541,28 @@ static void print_mapped_keys(void)
}
fprintf(stdout, "\nMapped keys:\n");
- fprintf(stdout, "\t[d] display refresh delay. \t(%d)\n", delay_secs);
- fprintf(stdout, "\t[e] display entries (lines). \t(%d)\n", print_entries);
+ fprintf(stdout, "\t[d] display refresh delay. \t(%d)\n", top.delay_secs);
+ fprintf(stdout, "\t[e] display entries (lines). \t(%d)\n", top.print_entries);
- if (evsel_list->nr_entries > 1)
- fprintf(stdout, "\t[E] active event counter. \t(%s)\n", event_name(sym_evsel));
+ if (top.evlist->nr_entries > 1)
+ fprintf(stdout, "\t[E] active event counter. \t(%s)\n", event_name(top.sym_evsel));
- fprintf(stdout, "\t[f] profile display filter (count). \t(%d)\n", count_filter);
+ fprintf(stdout, "\t[f] profile display filter (count). \t(%d)\n", top.count_filter);
fprintf(stdout, "\t[F] annotate display filter (percent). \t(%d%%)\n", sym_pcnt_filter);
fprintf(stdout, "\t[s] annotate symbol. \t(%s)\n", name?: "NULL");
fprintf(stdout, "\t[S] stop annotation.\n");
- if (evsel_list->nr_entries > 1)
- fprintf(stdout, "\t[w] toggle display weighted/count[E]r. \t(%d)\n", display_weighted ? 1 : 0);
+ if (top.evlist->nr_entries > 1)
+ fprintf(stdout, "\t[w] toggle display weighted/count[E]r. \t(%d)\n", top.display_weighted ? 1 : 0);
fprintf(stdout,
"\t[K] hide kernel_symbols symbols. \t(%s)\n",
- hide_kernel_symbols ? "yes" : "no");
+ top.hide_kernel_symbols ? "yes" : "no");
fprintf(stdout,
"\t[U] hide user symbols. \t(%s)\n",
- hide_user_symbols ? "yes" : "no");
- fprintf(stdout, "\t[z] toggle sample zeroing. \t(%d)\n", zero ? 1 : 0);
+ top.hide_user_symbols ? "yes" : "no");
+ fprintf(stdout, "\t[z] toggle sample zeroing. \t(%d)\n", top.zero ? 1 : 0);
fprintf(stdout, "\t[qQ] quit.\n");
}
@@ -791,7 +583,7 @@ static int key_mapped(int c)
return 1;
case 'E':
case 'w':
- return evsel_list->nr_entries > 1 ? 1 : 0;
+ return top.evlist->nr_entries > 1 ? 1 : 0;
default:
break;
}
@@ -826,47 +618,47 @@ static void handle_keypress(struct perf_session *session, int c)
switch (c) {
case 'd':
- prompt_integer(&delay_secs, "Enter display delay");
- if (delay_secs < 1)
- delay_secs = 1;
+ prompt_integer(&top.delay_secs, "Enter display delay");
+ if (top.delay_secs < 1)
+ top.delay_secs = 1;
break;
case 'e':
- prompt_integer(&print_entries, "Enter display entries (lines)");
- if (print_entries == 0) {
+ prompt_integer(&top.print_entries, "Enter display entries (lines)");
+ if (top.print_entries == 0) {
sig_winch_handler(SIGWINCH);
signal(SIGWINCH, sig_winch_handler);
} else
signal(SIGWINCH, SIG_DFL);
break;
case 'E':
- if (evsel_list->nr_entries > 1) {
+ if (top.evlist->nr_entries > 1) {
fprintf(stderr, "\nAvailable events:");
- list_for_each_entry(sym_evsel, &evsel_list->entries, node)
- fprintf(stderr, "\n\t%d %s", sym_evsel->idx, event_name(sym_evsel));
+ list_for_each_entry(top.sym_evsel, &top.evlist->entries, node)
+ fprintf(stderr, "\n\t%d %s", top.sym_evsel->idx, event_name(top.sym_evsel));
- prompt_integer(&sym_counter, "Enter details event counter");
+ prompt_integer(&top.sym_counter, "Enter details event counter");
- if (sym_counter >= evsel_list->nr_entries) {
- sym_evsel = list_entry(evsel_list->entries.next, struct perf_evsel, node);
- sym_counter = 0;
- fprintf(stderr, "Sorry, no such event, using %s.\n", event_name(sym_evsel));
+ if (top.sym_counter >= top.evlist->nr_entries) {
+ top.sym_evsel = list_entry(top.evlist->entries.next, struct perf_evsel, node);
+ top.sym_counter = 0;
+ fprintf(stderr, "Sorry, no such event, using %s.\n", event_name(top.sym_evsel));
sleep(1);
break;
}
- list_for_each_entry(sym_evsel, &evsel_list->entries, node)
- if (sym_evsel->idx == sym_counter)
+ list_for_each_entry(top.sym_evsel, &top.evlist->entries, node)
+ if (top.sym_evsel->idx == top.sym_counter)
break;
- } else sym_counter = 0;
+ } else top.sym_counter = 0;
break;
case 'f':
- prompt_integer(&count_filter, "Enter display event count filter");
+ prompt_integer(&top.count_filter, "Enter display event count filter");
break;
case 'F':
prompt_percent(&sym_pcnt_filter, "Enter details display event filter (percent)");
break;
case 'K':
- hide_kernel_symbols = !hide_kernel_symbols;
+ top.hide_kernel_symbols = !top.hide_kernel_symbols;
break;
case 'q':
case 'Q':
@@ -890,13 +682,13 @@ static void handle_keypress(struct perf_session *session, int c)
}
break;
case 'U':
- hide_user_symbols = !hide_user_symbols;
+ top.hide_user_symbols = !top.hide_user_symbols;
break;
case 'w':
- display_weighted = ~display_weighted;
+ top.display_weighted = ~top.display_weighted;
break;
case 'z':
- zero = !zero;
+ top.zero = !top.zero;
break;
default:
break;
@@ -917,7 +709,7 @@ static void *display_thread(void *arg __used)
tc.c_cc[VTIME] = 0;
repeat:
- delay_msecs = delay_secs * 1000;
+ delay_msecs = top.delay_secs * 1000;
tcsetattr(0, TCSANOW, &tc);
/* trash return*/
getc(stdin);
@@ -1005,27 +797,27 @@ static void perf_event__process_sample(const union perf_event *event,
struct machine *machine;
u8 origin = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK;
- ++samples;
+ ++top.samples;
switch (origin) {
case PERF_RECORD_MISC_USER:
- ++us_samples;
- if (hide_user_symbols)
+ ++top.us_samples;
+ if (top.hide_user_symbols)
return;
machine = perf_session__find_host_machine(session);
break;
case PERF_RECORD_MISC_KERNEL:
- ++kernel_samples;
- if (hide_kernel_symbols)
+ ++top.kernel_samples;
+ if (top.hide_kernel_symbols)
return;
machine = perf_session__find_host_machine(session);
break;
case PERF_RECORD_MISC_GUEST_KERNEL:
- ++guest_kernel_samples;
+ ++top.guest_kernel_samples;
machine = perf_session__find_machine(session, event->ip.pid);
break;
case PERF_RECORD_MISC_GUEST_USER:
- ++guest_us_samples;
+ ++top.guest_us_samples;
/*
* TODO: we don't process guest user from host side
* except simple counting.
@@ -1042,7 +834,7 @@ static void perf_event__process_sample(const union perf_event *event,
}
if (event->header.misc & PERF_RECORD_MISC_EXACT_IP)
- exact_samples++;
+ top.exact_samples++;
if (perf_event__preprocess_sample(event, session, &al, sample,
symbol_filter) < 0 ||
@@ -1093,14 +885,14 @@ static void perf_event__process_sample(const union perf_event *event,
struct perf_evsel *evsel;
syme->origin = origin;
- evsel = perf_evlist__id2evsel(evsel_list, sample->id);
+ evsel = perf_evlist__id2evsel(top.evlist, sample->id);
assert(evsel != NULL);
syme->count[evsel->idx]++;
record_precise_ip(syme, evsel->idx, ip);
- pthread_mutex_lock(&active_symbols_lock);
+ pthread_mutex_lock(&top.active_symbols_lock);
if (list_empty(&syme->node) || !syme->node.next)
__list_insert_active_sym(syme);
- pthread_mutex_unlock(&active_symbols_lock);
+ pthread_mutex_unlock(&top.active_symbols_lock);
}
}
@@ -1109,7 +901,7 @@ static void perf_session__mmap_read_cpu(struct perf_session *self, int cpu)
struct perf_sample sample;
union perf_event *event;
- while ((event = perf_evlist__read_on_cpu(evsel_list, cpu)) != NULL) {
+ while ((event = perf_evlist__read_on_cpu(top.evlist, cpu)) != NULL) {
perf_session__parse_sample(self, event, &sample);
if (event->header.type == PERF_RECORD_SAMPLE)
@@ -1123,7 +915,7 @@ static void perf_session__mmap_read(struct perf_session *self)
{
int i;
- for (i = 0; i < evsel_list->cpus->nr; i++)
+ for (i = 0; i < top.evlist->cpus->nr; i++)
perf_session__mmap_read_cpu(self, i);
}
@@ -1136,10 +928,10 @@ static void start_counters(struct perf_evlist *evlist)
attr->sample_type = PERF_SAMPLE_IP | PERF_SAMPLE_TID;
- if (freq) {
+ if (top.freq) {
attr->sample_type |= PERF_SAMPLE_PERIOD;
attr->freq = 1;
- attr->sample_freq = freq;
+ attr->sample_freq = top.freq;
}
if (evlist->nr_entries > 1) {
@@ -1149,8 +941,8 @@ static void start_counters(struct perf_evlist *evlist)
attr->mmap = 1;
try_again:
- if (perf_evsel__open(counter, evsel_list->cpus,
- evsel_list->threads, group, inherit) < 0) {
+ if (perf_evsel__open(counter, top.evlist->cpus,
+ top.evlist->threads, group, inherit) < 0) {
int err = errno;
if (err == EPERM || err == EACCES)
@@ -1198,18 +990,18 @@ static int __cmd_top(void)
if (session == NULL)
return -ENOMEM;
- if (target_tid != -1)
- perf_event__synthesize_thread(target_tid, perf_event__process,
+ if (top.target_tid != -1)
+ perf_event__synthesize_thread(top.target_tid, perf_event__process,
session);
else
perf_event__synthesize_threads(perf_event__process, session);
- start_counters(evsel_list);
- first = list_entry(evsel_list->entries.next, struct perf_evsel, node);
+ start_counters(top.evlist);
+ first = list_entry(top.evlist->entries.next, struct perf_evsel, node);
perf_session__set_sample_type(session, first->attr.sample_type);
/* Wait for a minimal set of events before starting the snapshot */
- poll(evsel_list->pollfd, evsel_list->nr_fds, 100);
+ poll(top.evlist->pollfd, top.evlist->nr_fds, 100);
perf_session__mmap_read(session);
@@ -1229,12 +1021,12 @@ static int __cmd_top(void)
}
while (1) {
- int hits = samples;
+ u64 hits = top.samples;
perf_session__mmap_read(session);
- if (hits == samples)
- ret = poll(evsel_list->pollfd, evsel_list->nr_fds, 100);
+ if (hits == top.samples)
+ ret = poll(top.evlist->pollfd, top.evlist->nr_fds, 100);
}
return 0;
@@ -1246,31 +1038,31 @@ static const char * const top_usage[] = {
};
static const struct option options[] = {
- OPT_CALLBACK('e', "event", &evsel_list, "event",
+ OPT_CALLBACK('e', "event", &top.evlist, "event",
"event selector. use 'perf list' to list available events",
parse_events),
OPT_INTEGER('c', "count", &default_interval,
"event period to sample"),
- OPT_INTEGER('p', "pid", &target_pid,
+ OPT_INTEGER('p', "pid", &top.target_pid,
"profile events on existing process id"),
- OPT_INTEGER('t', "tid", &target_tid,
+ OPT_INTEGER('t', "tid", &top.target_tid,
"profile events on existing thread id"),
OPT_BOOLEAN('a', "all-cpus", &system_wide,
"system-wide collection from all CPUs"),
- OPT_STRING('C', "cpu", &cpu_list, "cpu",
+ OPT_STRING('C', "cpu", &top.cpu_list, "cpu",
"list of cpus to monitor"),
OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name,
"file", "vmlinux pathname"),
- OPT_BOOLEAN('K', "hide_kernel_symbols", &hide_kernel_symbols,
+ OPT_BOOLEAN('K', "hide_kernel_symbols", &top.hide_kernel_symbols,
"hide kernel symbols"),
OPT_UINTEGER('m', "mmap-pages", &mmap_pages, "number of mmap data pages"),
OPT_INTEGER('r', "realtime", &realtime_prio,
"collect data with this RT SCHED_FIFO priority"),
- OPT_INTEGER('d', "delay", &delay_secs,
+ OPT_INTEGER('d', "delay", &top.delay_secs,
"number of seconds to delay between refreshes"),
OPT_BOOLEAN('D', "dump-symtab", &dump_symtab,
"dump the symbol table used for profiling"),
- OPT_INTEGER('f', "count-filter", &count_filter,
+ OPT_INTEGER('f', "count-filter", &top.count_filter,
"only display functions with more events than this"),
OPT_BOOLEAN('g', "group", &group,
"put the counters into a counter group"),
@@ -1278,13 +1070,13 @@ static const struct option options[] = {
"child tasks inherit counters"),
OPT_STRING('s', "sym-annotate", &sym_filter, "symbol name",
"symbol to annotate"),
- OPT_BOOLEAN('z', "zero", &zero,
+ OPT_BOOLEAN('z', "zero", &top.zero,
"zero history across updates"),
- OPT_INTEGER('F', "freq", &freq,
+ OPT_INTEGER('F', "freq", &top.freq,
"profile at this frequency"),
- OPT_INTEGER('E', "entries", &print_entries,
+ OPT_INTEGER('E', "entries", &top.print_entries,
"display this many functions"),
- OPT_BOOLEAN('U', "hide_user_symbols", &hide_user_symbols,
+ OPT_BOOLEAN('U', "hide_user_symbols", &top.hide_user_symbols,
"hide user symbols"),
OPT_INCR('v', "verbose", &verbose,
"be more verbose (show counter open errors, etc)"),
@@ -1296,8 +1088,8 @@ int cmd_top(int argc, const char **argv, const char *prefix __used)
struct perf_evsel *pos;
int status = -ENOMEM;
- evsel_list = perf_evlist__new(NULL, NULL);
- if (evsel_list == NULL)
+ top.evlist = perf_evlist__new(NULL, NULL);
+ if (top.evlist == NULL)
return -ENOMEM;
page_size = sysconf(_SC_PAGE_SIZE);
@@ -1307,43 +1099,43 @@ int cmd_top(int argc, const char **argv, const char *prefix __used)
usage_with_options(top_usage, options);
/* CPU and PID are mutually exclusive */
- if (target_tid > 0 && cpu_list) {
+ if (top.target_tid > 0 && top.cpu_list) {
printf("WARNING: PID switch overriding CPU\n");
sleep(1);
- cpu_list = NULL;
+ top.cpu_list = NULL;
}
- if (target_pid != -1)
- target_tid = target_pid;
+ if (top.target_pid != -1)
+ top.target_tid = top.target_pid;
- if (perf_evlist__create_maps(evsel_list, target_pid,
- target_tid, cpu_list) < 0)
+ if (perf_evlist__create_maps(top.evlist, top.target_pid,
+ top.target_tid, top.cpu_list) < 0)
usage_with_options(top_usage, options);
- if (!evsel_list->nr_entries &&
- perf_evlist__add_default(evsel_list) < 0) {
+ if (!top.evlist->nr_entries &&
+ perf_evlist__add_default(top.evlist) < 0) {
pr_err("Not enough memory for event selector list\n");
return -ENOMEM;
}
- if (delay_secs < 1)
- delay_secs = 1;
+ if (top.delay_secs < 1)
+ top.delay_secs = 1;
/*
* User specified count overrides default frequency.
*/
if (default_interval)
- freq = 0;
- else if (freq) {
- default_interval = freq;
+ top.freq = 0;
+ else if (top.freq) {
+ default_interval = top.freq;
} else {
fprintf(stderr, "frequency and count are zero, aborting\n");
exit(EXIT_FAILURE);
}
- list_for_each_entry(pos, &evsel_list->entries, node) {
- if (perf_evsel__alloc_fd(pos, evsel_list->cpus->nr,
- evsel_list->threads->nr) < 0)
+ list_for_each_entry(pos, &top.evlist->entries, node) {
+ if (perf_evsel__alloc_fd(pos, top.evlist->cpus->nr,
+ top.evlist->threads->nr) < 0)
goto out_free_fd;
/*
* Fill in the ones not specifically initialized via -c:
@@ -1354,28 +1146,28 @@ int cmd_top(int argc, const char **argv, const char *prefix __used)
pos->attr.sample_period = default_interval;
}
- if (perf_evlist__alloc_pollfd(evsel_list) < 0 ||
- perf_evlist__alloc_mmap(evsel_list) < 0)
+ if (perf_evlist__alloc_pollfd(top.evlist) < 0 ||
+ perf_evlist__alloc_mmap(top.evlist) < 0)
goto out_free_fd;
- sym_evsel = list_entry(evsel_list->entries.next, struct perf_evsel, node);
+ top.sym_evsel = list_entry(top.evlist->entries.next, struct perf_evsel, node);
symbol_conf.priv_size = (sizeof(struct sym_entry) +
- (evsel_list->nr_entries + 1) * sizeof(unsigned long));
+ (top.evlist->nr_entries + 1) * sizeof(unsigned long));
symbol_conf.try_vmlinux_path = (symbol_conf.vmlinux_name == NULL);
if (symbol__init() < 0)
return -1;
get_term_dimensions(&winsize);
- if (print_entries == 0) {
+ if (top.print_entries == 0) {
update_print_entries(&winsize);
signal(SIGWINCH, sig_winch_handler);
}
status = __cmd_top();
out_free_fd:
- perf_evlist__delete(evsel_list);
+ perf_evlist__delete(top.evlist);
return status;
}
diff --git a/tools/perf/util/top.c b/tools/perf/util/top.c
new file mode 100644
index 0000000..c06cc53
--- /dev/null
+++ b/tools/perf/util/top.c
@@ -0,0 +1,212 @@
+/*
+ * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com>
+ *
+ * Refactored from builtin-top.c, see that files for further copyright notes.
+ *
+ * Released under the GPL v2. (and only v2, not any later version)
+ */
+
+#include "cpumap.h"
+#include "event.h"
+#include "evlist.h"
+#include "evsel.h"
+#include "parse-events.h"
+#include "symbol.h"
+#include "top.h"
+#include <inttypes.h>
+
+/*
+ * Ordering weight: count-1 * count-2 * ... / count-n
+ */
+static double sym_weight(const struct sym_entry *sym, struct perf_top *top)
+{
+ double weight = sym->snap_count;
+ int counter;
+
+ if (!top->display_weighted)
+ return weight;
+
+ for (counter = 1; counter < top->evlist->nr_entries - 1; counter++)
+ weight *= sym->count[counter];
+
+ weight /= (sym->count[counter] + 1);
+
+ return weight;
+}
+
+static void perf_top__remove_active_sym(struct perf_top *top, struct sym_entry *syme)
+{
+ pthread_mutex_lock(&top->active_symbols_lock);
+ list_del_init(&syme->node);
+ pthread_mutex_unlock(&top->active_symbols_lock);
+}
+
+static void rb_insert_active_sym(struct rb_root *tree, struct sym_entry *se)
+{
+ struct rb_node **p = &tree->rb_node;
+ struct rb_node *parent = NULL;
+ struct sym_entry *iter;
+
+ while (*p != NULL) {
+ parent = *p;
+ iter = rb_entry(parent, struct sym_entry, rb_node);
+
+ if (se->weight > iter->weight)
+ p = &(*p)->rb_left;
+ else
+ p = &(*p)->rb_right;
+ }
+
+ rb_link_node(&se->rb_node, parent, p);
+ rb_insert_color(&se->rb_node, tree);
+}
+
+size_t perf_top__header_snprintf(struct perf_top *top, char *bf, size_t size)
+{
+ struct perf_evsel *counter;
+ float samples_per_sec = top->samples / top->delay_secs;
+ float ksamples_per_sec = top->kernel_samples / top->delay_secs;
+ float esamples_percent = (100.0 * top->exact_samples) / top->samples;
+ size_t ret = 0;
+
+ if (!perf_guest) {
+ ret = snprintf(bf, size,
+ " PerfTop:%8.0f irqs/sec kernel:%4.1f%%"
+ " exact: %4.1f%% [", samples_per_sec,
+ 100.0 - (100.0 * ((samples_per_sec - ksamples_per_sec) /
+ samples_per_sec)),
+ esamples_percent);
+ } else {
+ float us_samples_per_sec = top->us_samples / top->delay_secs;
+ float guest_kernel_samples_per_sec = top->guest_kernel_samples / top->delay_secs;
+ float guest_us_samples_per_sec = top->guest_us_samples / top->delay_secs;
+
+ ret = snprintf(bf, size,
+ " PerfTop:%8.0f irqs/sec kernel:%4.1f%% us:%4.1f%%"
+ " guest kernel:%4.1f%% guest us:%4.1f%%"
+ " exact: %4.1f%% [", samples_per_sec,
+ 100.0 - (100.0 * ((samples_per_sec - ksamples_per_sec) /
+ samples_per_sec)),
+ 100.0 - (100.0 * ((samples_per_sec - us_samples_per_sec) /
+ samples_per_sec)),
+ 100.0 - (100.0 * ((samples_per_sec -
+ guest_kernel_samples_per_sec) /
+ samples_per_sec)),
+ 100.0 - (100.0 * ((samples_per_sec -
+ guest_us_samples_per_sec) /
+ samples_per_sec)),
+ esamples_percent);
+ }
+
+ if (top->evlist->nr_entries == 1 || !top->display_weighted) {
+ struct perf_evsel *first;
+ first = list_entry(top->evlist->entries.next, struct perf_evsel, node);
+ ret += snprintf(bf + ret, size - ret, "%" PRIu64 "%s ",
+ (uint64_t)first->attr.sample_period,
+ top->freq ? "Hz" : "");
+ }
+
+ if (!top->display_weighted) {
+ ret += snprintf(bf + ret, size - ret, "%s",
+ event_name(top->sym_evsel));
+ } else list_for_each_entry(counter, &top->evlist->entries, node) {
+ ret += snprintf(bf + ret, size - ret, "%s%s",
+ counter->idx ? "/" : "", event_name(counter));
+ }
+
+ ret += snprintf(bf + ret, size - ret, "], ");
+
+ if (top->target_pid != -1)
+ ret += snprintf(bf + ret, size - ret, " (target_pid: %d",
+ top->target_pid);
+ else if (top->target_tid != -1)
+ ret += snprintf(bf + ret, size - ret, " (target_tid: %d",
+ top->target_tid);
+ else
+ ret += snprintf(bf + ret, size - ret, " (all");
+
+ if (top->cpu_list)
+ ret += snprintf(bf + ret, size - ret, ", CPU%s: %s)",
+ top->evlist->cpus->nr > 1 ? "s" : "", top->cpu_list);
+ else {
+ if (top->target_tid != -1)
+ ret += snprintf(bf + ret, size - ret, ")");
+ else
+ ret += snprintf(bf + ret, size - ret, ", %d CPU%s)",
+ top->evlist->cpus->nr,
+ top->evlist->cpus->nr > 1 ? "s" : "");
+ }
+
+ return ret;
+}
+
+void perf_top__reset_sample_counters(struct perf_top *top)
+{
+ top->samples = top->us_samples = top->kernel_samples =
+ top->exact_samples = top->guest_kernel_samples =
+ top->guest_us_samples = 0;
+}
+
+float perf_top__decay_samples(struct perf_top *top, struct rb_root *root)
+{
+ struct sym_entry *syme, *n;
+ float sum_ksamples = 0.0;
+ int snap = !top->display_weighted ? top->sym_counter : 0, j;
+
+ /* Sort the active symbols */
+ pthread_mutex_lock(&top->active_symbols_lock);
+ syme = list_entry(top->active_symbols.next, struct sym_entry, node);
+ pthread_mutex_unlock(&top->active_symbols_lock);
+
+ list_for_each_entry_safe_from(syme, n, &top->active_symbols, node) {
+ syme->snap_count = syme->count[snap];
+ if (syme->snap_count != 0) {
+
+ if ((top->hide_user_symbols &&
+ syme->origin == PERF_RECORD_MISC_USER) ||
+ (top->hide_kernel_symbols &&
+ syme->origin == PERF_RECORD_MISC_KERNEL)) {
+ perf_top__remove_active_sym(top, syme);
+ continue;
+ }
+ syme->weight = sym_weight(syme, top);
+ rb_insert_active_sym(root, syme);
+ sum_ksamples += syme->snap_count;
+
+ for (j = 0; j < top->evlist->nr_entries; j++)
+ syme->count[j] = top->zero ? 0 : syme->count[j] * 7 / 8;
+ } else
+ perf_top__remove_active_sym(top, syme);
+ }
+
+ return sum_ksamples;
+}
+
+/*
+ * Find the longest symbol name that will be displayed
+ */
+void perf_top__find_widths(struct perf_top *top, struct rb_root *root,
+ int *dso_width, int *dso_short_width, int *sym_width)
+{
+ struct rb_node *nd;
+ int printed = 0;
+
+ *sym_width = *dso_width = *dso_short_width = 0;
+
+ for (nd = rb_first(root); nd; nd = rb_next(nd)) {
+ struct sym_entry *syme = rb_entry(nd, struct sym_entry, rb_node);
+
+ if (++printed > top->print_entries ||
+ (int)syme->snap_count < top->count_filter)
+ continue;
+
+ if (syme->map->dso->long_name_len > *dso_width)
+ *dso_width = syme->map->dso->long_name_len;
+
+ if (syme->map->dso->short_name_len > *dso_short_width)
+ *dso_short_width = syme->map->dso->short_name_len;
+
+ if (syme->name_len > *sym_width)
+ *sym_width = syme->name_len;
+ }
+}
diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h
new file mode 100644
index 0000000..0467b26
--- /dev/null
+++ b/tools/perf/util/top.h
@@ -0,0 +1,67 @@
+#ifndef __PERF_TOP_H
+#define __PERF_TOP_H 1
+
+#include "types.h"
+#include "../perf.h"
+#include <stddef.h>
+#include <pthread.h>
+#include <linux/list.h>
+#include <linux/rbtree.h>
+
+struct perf_evlist;
+struct perf_evsel;
+
+struct source_line {
+ u64 eip;
+ unsigned long count[MAX_COUNTERS]; /* FIXME */
+ char *line;
+ struct source_line *next;
+};
+
+struct sym_entry_source {
+ struct source_line *source;
+ struct source_line *lines;
+ struct source_line **lines_tail;
+ pthread_mutex_t lock;
+};
+
+struct sym_entry {
+ struct rb_node rb_node;
+ struct list_head node;
+ unsigned long snap_count;
+ double weight;
+ int skip;
+ u16 name_len;
+ u8 origin;
+ struct map *map;
+ struct sym_entry_source *src;
+ unsigned long count[0];
+};
+
+struct perf_top {
+ struct perf_evlist *evlist;
+ /*
+ * Symbols will be added here in perf_event__process_sample and will
+ * get out after decayed.
+ */
+ struct list_head active_symbols;
+ pthread_mutex_t active_symbols_lock;
+ u64 samples;
+ u64 kernel_samples, us_samples;
+ u64 exact_samples;
+ u64 guest_us_samples, guest_kernel_samples;
+ int print_entries, count_filter, delay_secs;
+ int display_weighted, freq;
+ int sym_counter, target_pid, target_tid;
+ bool hide_kernel_symbols, hide_user_symbols, zero;
+ const char *cpu_list;
+ struct perf_evsel *sym_evsel;
+};
+
+size_t perf_top__header_snprintf(struct perf_top *top, char *bf, size_t size);
+void perf_top__reset_sample_counters(struct perf_top *top);
+float perf_top__decay_samples(struct perf_top *top, struct rb_root *root);
+void perf_top__find_widths(struct perf_top *top, struct rb_root *root,
+ int *dso_width, int *dso_short_width, int *sym_width);
+
+#endif /* __PERF_TOP_H */
--
1.6.2.5
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH 4/5] perf tools: Don't fallback to setup_pager unconditionally
2011-01-31 21:29 [GIT PULL 0/5] perf/core improvements Arnaldo Carvalho de Melo
` (2 preceding siblings ...)
2011-01-31 21:29 ` [PATCH 3/5] perf top: Move display agnostic routines to util/top.[ch] Arnaldo Carvalho de Melo
@ 2011-01-31 21:29 ` Arnaldo Carvalho de Melo
2011-01-31 21:29 ` [PATCH 5/5] perf top: Introduce slang based TUI Arnaldo Carvalho de Melo
2011-02-01 9:14 ` [GIT PULL 0/5] perf/core improvements Ingo Molnar
5 siblings, 0 replies; 8+ messages in thread
From: Arnaldo Carvalho de Melo @ 2011-01-31 21:29 UTC (permalink / raw)
To: Ingo Molnar
Cc: linux-kernel, Arnaldo Carvalho de Melo, Frederic Weisbecker,
Ingo Molnar, Mike Galbraith, Paul Mackerras, Peter Zijlstra,
Stephane Eranian, Tom Zanussi
From: Arnaldo Carvalho de Melo <acme@redhat.com>
Because in tools like 'top' we don't want the pager.
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/builtin-annotate.c | 2 +-
tools/perf/builtin-report.c | 2 +-
tools/perf/util/cache.h | 7 ++++---
tools/perf/util/ui/setup.c | 5 +++--
4 files changed, 9 insertions(+), 7 deletions(-)
diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c
index 7006786..cd9dec4 100644
--- a/tools/perf/builtin-annotate.c
+++ b/tools/perf/builtin-annotate.c
@@ -452,7 +452,7 @@ int cmd_annotate(int argc, const char **argv, const char *prefix __used)
else if (use_tui)
use_browser = 1;
- setup_browser();
+ setup_browser(true);
symbol_conf.priv_size = sizeof(struct sym_priv);
symbol_conf.try_vmlinux_path = true;
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index a6a4e54..080937c 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -499,7 +499,7 @@ int cmd_report(int argc, const char **argv, const char *prefix __used)
use_browser = 1;
if (strcmp(input_name, "-") != 0)
- setup_browser();
+ setup_browser(true);
else
use_browser = 0;
/*
diff --git a/tools/perf/util/cache.h b/tools/perf/util/cache.h
index a772979..fc5e5a0 100644
--- a/tools/perf/util/cache.h
+++ b/tools/perf/util/cache.h
@@ -34,13 +34,14 @@ extern int pager_use_color;
extern int use_browser;
#ifdef NO_NEWT_SUPPORT
-static inline void setup_browser(void)
+static inline void setup_browser(bool fallback_to_pager)
{
- setup_pager();
+ if (fallback_to_pager)
+ setup_pager();
}
static inline void exit_browser(bool wait_for_ok __used) {}
#else
-void setup_browser(void);
+void setup_browser(bool fallback_to_pager);
void exit_browser(bool wait_for_ok);
#endif
diff --git a/tools/perf/util/ui/setup.c b/tools/perf/util/ui/setup.c
index 6620850..fbf1a14 100644
--- a/tools/perf/util/ui/setup.c
+++ b/tools/perf/util/ui/setup.c
@@ -14,11 +14,12 @@ static void newt_suspend(void *d __used)
newtResume();
}
-void setup_browser(void)
+void setup_browser(bool fallback_to_pager)
{
if (!isatty(1) || !use_browser || dump_trace) {
use_browser = 0;
- setup_pager();
+ if (fallback_to_pager)
+ setup_pager();
return;
}
--
1.6.2.5
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH 5/5] perf top: Introduce slang based TUI
2011-01-31 21:29 [GIT PULL 0/5] perf/core improvements Arnaldo Carvalho de Melo
` (3 preceding siblings ...)
2011-01-31 21:29 ` [PATCH 4/5] perf tools: Don't fallback to setup_pager unconditionally Arnaldo Carvalho de Melo
@ 2011-01-31 21:29 ` Arnaldo Carvalho de Melo
2011-02-01 9:14 ` [GIT PULL 0/5] perf/core improvements Ingo Molnar
5 siblings, 0 replies; 8+ messages in thread
From: Arnaldo Carvalho de Melo @ 2011-01-31 21:29 UTC (permalink / raw)
To: Ingo Molnar
Cc: linux-kernel, Arnaldo Carvalho de Melo, Frederic Weisbecker,
Ingo Molnar, Mike Galbraith, Paul Mackerras, Peter Zijlstra,
Stephane Eranian, Tom Zanussi
From: Arnaldo Carvalho de Melo <acme@redhat.com>
Disabled by default as there are features found in the stdio based one
that aren't implemented, like live annotation, filtering knobs data
entry.
Annotation hopefully will get somehow merged with the 'perf annotate'
code.
To use it:
perf top --tui
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/Makefile | 4 +
tools/perf/builtin-top.c | 35 ++++++++--
tools/perf/util/top.c | 7 ++-
tools/perf/util/top.h | 15 ++++-
tools/perf/util/ui/browsers/top.c | 136 +++++++++++++++++++++++++++++++++++++
5 files changed, 189 insertions(+), 8 deletions(-)
create mode 100644 tools/perf/util/ui/browsers/top.c
diff --git a/tools/perf/Makefile b/tools/perf/Makefile
index edc660e..67a9f4d 100644
--- a/tools/perf/Makefile
+++ b/tools/perf/Makefile
@@ -621,6 +621,7 @@ else
LIB_OBJS += $(OUTPUT)util/ui/browsers/annotate.o
LIB_OBJS += $(OUTPUT)util/ui/browsers/hists.o
LIB_OBJS += $(OUTPUT)util/ui/browsers/map.o
+ LIB_OBJS += $(OUTPUT)util/ui/browsers/top.o
LIB_OBJS += $(OUTPUT)util/ui/helpline.o
LIB_OBJS += $(OUTPUT)util/ui/progress.o
LIB_OBJS += $(OUTPUT)util/ui/util.o
@@ -1050,6 +1051,9 @@ $(OUTPUT)util/ui/browser.o: util/ui/browser.c $(OUTPUT)PERF-CFLAGS
$(OUTPUT)util/ui/browsers/annotate.o: util/ui/browsers/annotate.c $(OUTPUT)PERF-CFLAGS
$(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DENABLE_SLFUTURE_CONST $<
+$(OUTPUT)util/ui/browsers/top.o: util/ui/browsers/top.c $(OUTPUT)PERF-CFLAGS
+ $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DENABLE_SLFUTURE_CONST $<
+
$(OUTPUT)util/ui/browsers/hists.o: util/ui/browsers/hists.c $(OUTPUT)PERF-CFLAGS
$(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DENABLE_SLFUTURE_CONST $<
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index 3c9ba94..104de9a 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -20,6 +20,7 @@
#include "perf.h"
+#include "util/cache.h"
#include "util/color.h"
#include "util/evlist.h"
#include "util/evsel.h"
@@ -75,6 +76,8 @@ static struct perf_top top = {
static bool system_wide = false;
+static bool use_tui, use_stdio;
+
static int default_interval = 0;
static bool inherit = false;
@@ -96,11 +99,6 @@ static int sym_pcnt_filter = 5;
* Source functions
*/
-static inline struct symbol *sym_entry__symbol(struct sym_entry *self)
-{
- return ((void *)self) + symbol_conf.priv_size;
-}
-
void get_term_dimensions(struct winsize *ws)
{
char *s = getenv("LINES");
@@ -695,6 +693,14 @@ static void handle_keypress(struct perf_session *session, int c)
}
}
+static void *display_thread_tui(void *arg __used)
+{
+ perf_top__tui_browser(&top);
+ exit_browser(0);
+ exit(0);
+ return NULL;
+}
+
static void *display_thread(void *arg __used)
{
struct pollfd stdin_poll = { .fd = 0, .events = POLLIN };
@@ -1005,7 +1011,8 @@ static int __cmd_top(void)
perf_session__mmap_read(session);
- if (pthread_create(&thread, NULL, display_thread, session)) {
+ if (pthread_create(&thread, NULL, (use_browser > 0 ? display_thread_tui :
+ display_thread), session)) {
printf("Could not create display thread.\n");
exit(-1);
}
@@ -1078,6 +1085,8 @@ static const struct option options[] = {
"display this many functions"),
OPT_BOOLEAN('U', "hide_user_symbols", &top.hide_user_symbols,
"hide user symbols"),
+ OPT_BOOLEAN(0, "tui", &use_tui, "Use the TUI interface"),
+ OPT_BOOLEAN(0, "stdio", &use_stdio, "Use the stdio interface"),
OPT_INCR('v', "verbose", &verbose,
"be more verbose (show counter open errors, etc)"),
OPT_END()
@@ -1098,6 +1107,20 @@ int cmd_top(int argc, const char **argv, const char *prefix __used)
if (argc)
usage_with_options(top_usage, options);
+ /*
+ * XXX For now start disabled, only using TUI if explicitely asked for.
+ * Change that when handle_keys equivalent gets written, live annotation
+ * done, etc.
+ */
+ use_browser = 0;
+
+ if (use_stdio)
+ use_browser = 0;
+ else if (use_tui)
+ use_browser = 1;
+
+ setup_browser(false);
+
/* CPU and PID are mutually exclusive */
if (top.target_tid > 0 && top.cpu_list) {
printf("WARNING: PID switch overriding CPU\n");
diff --git a/tools/perf/util/top.c b/tools/perf/util/top.c
index c06cc53..1d2e265 100644
--- a/tools/perf/util/top.c
+++ b/tools/perf/util/top.c
@@ -158,6 +158,7 @@ float perf_top__decay_samples(struct perf_top *top, struct rb_root *root)
syme = list_entry(top->active_symbols.next, struct sym_entry, node);
pthread_mutex_unlock(&top->active_symbols_lock);
+ top->rb_entries = 0;
list_for_each_entry_safe_from(syme, n, &top->active_symbols, node) {
syme->snap_count = syme->count[snap];
if (syme->snap_count != 0) {
@@ -170,7 +171,11 @@ float perf_top__decay_samples(struct perf_top *top, struct rb_root *root)
continue;
}
syme->weight = sym_weight(syme, top);
- rb_insert_active_sym(root, syme);
+
+ if ((int)syme->snap_count >= top->count_filter) {
+ rb_insert_active_sym(root, syme);
+ ++top->rb_entries;
+ }
sum_ksamples += syme->snap_count;
for (j = 0; j < top->evlist->nr_entries; j++)
diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h
index 0467b26..611370f 100644
--- a/tools/perf/util/top.h
+++ b/tools/perf/util/top.h
@@ -38,6 +38,11 @@ struct sym_entry {
unsigned long count[0];
};
+static inline struct symbol *sym_entry__symbol(struct sym_entry *self)
+{
+ return ((void *)self) + symbol_conf.priv_size;
+}
+
struct perf_top {
struct perf_evlist *evlist;
/*
@@ -51,7 +56,7 @@ struct perf_top {
u64 exact_samples;
u64 guest_us_samples, guest_kernel_samples;
int print_entries, count_filter, delay_secs;
- int display_weighted, freq;
+ int display_weighted, freq, rb_entries;
int sym_counter, target_pid, target_tid;
bool hide_kernel_symbols, hide_user_symbols, zero;
const char *cpu_list;
@@ -64,4 +69,12 @@ float perf_top__decay_samples(struct perf_top *top, struct rb_root *root);
void perf_top__find_widths(struct perf_top *top, struct rb_root *root,
int *dso_width, int *dso_short_width, int *sym_width);
+#ifdef NO_NEWT_SUPPORT
+static inline int perf_top__tui_browser(struct perf_top *top __used)
+{
+ return 0;
+}
+#else
+int perf_top__tui_browser(struct perf_top *top);
+#endif
#endif /* __PERF_TOP_H */
diff --git a/tools/perf/util/ui/browsers/top.c b/tools/perf/util/ui/browsers/top.c
new file mode 100644
index 0000000..ca60624
--- /dev/null
+++ b/tools/perf/util/ui/browsers/top.c
@@ -0,0 +1,136 @@
+/*
+ * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com>
+ *
+ * Parts came from builtin-{top,stat,record}.c, see those files for further
+ * copyright notes.
+ *
+ * Released under the GPL v2. (and only v2, not any later version)
+ */
+#include "../browser.h"
+#include "../helpline.h"
+#include "../libslang.h"
+#include "../../evlist.h"
+#include "../../hist.h"
+#include "../../sort.h"
+#include "../../symbol.h"
+#include "../../top.h"
+
+struct perf_top_browser {
+ struct ui_browser b;
+ struct rb_root root;
+ float sum_ksamples;
+ int dso_width;
+ int dso_short_width;
+ int sym_width;
+};
+
+static void perf_top_browser__write(struct ui_browser *browser, void *entry, int row)
+{
+ struct perf_top_browser *top_browser = container_of(browser, struct perf_top_browser, b);
+ struct sym_entry *syme = rb_entry(entry, struct sym_entry, rb_node);
+ bool current_entry = ui_browser__is_current_entry(browser, row);
+ struct symbol *symbol = sym_entry__symbol(syme);
+ struct perf_top *top = browser->priv;
+ int width = browser->width;
+ double pcnt;
+
+ pcnt = 100.0 - (100.0 * ((top_browser->sum_ksamples - syme->snap_count) /
+ top_browser->sum_ksamples));
+ ui_browser__set_percent_color(browser, pcnt, current_entry);
+
+ if (top->evlist->nr_entries == 1 || !top->display_weighted) {
+ slsmg_printf("%20.2f ", syme->weight);
+ width -= 24;
+ } else {
+ slsmg_printf("%9.1f %10ld ", syme->weight, syme->snap_count);
+ width -= 23;
+ }
+
+ slsmg_printf("%4.1f%%", pcnt);
+ width -= 7;
+
+ if (verbose) {
+ slsmg_printf(" %016" PRIx64, symbol->start);
+ width -= 17;
+ }
+
+ slsmg_printf(" %-*.*s ", top_browser->sym_width, top_browser->sym_width,
+ symbol->name);
+ width -= top_browser->sym_width;
+ slsmg_write_nstring(width >= syme->map->dso->long_name_len ?
+ syme->map->dso->long_name :
+ syme->map->dso->short_name, width);
+}
+
+static void perf_top_browser__update_rb_tree(struct perf_top_browser *browser)
+{
+ struct perf_top *top = browser->b.priv;
+
+ browser->root = RB_ROOT;
+ browser->b.top = NULL;
+ browser->sum_ksamples = perf_top__decay_samples(top, &browser->root);
+ perf_top__find_widths(top, &browser->root, &browser->dso_width,
+ &browser->dso_short_width,
+ &browser->sym_width);
+ if (browser->sym_width + browser->dso_width > browser->b.width - 29) {
+ browser->dso_width = browser->dso_short_width;
+ if (browser->sym_width + browser->dso_width > browser->b.width - 29)
+ browser->sym_width = browser->b.width - browser->dso_width - 29;
+ }
+ browser->b.nr_entries = top->rb_entries;
+}
+
+static int perf_top_browser__run(struct perf_top_browser *browser)
+{
+ int key;
+ char title[160];
+ struct perf_top *top = browser->b.priv;
+ int delay_msecs = top->delay_secs * 1000;
+
+ perf_top_browser__update_rb_tree(browser);
+ perf_top__header_snprintf(top, title, sizeof(title));
+ perf_top__reset_sample_counters(top);
+
+ if (ui_browser__show(&browser->b, title, "ESC: exit") < 0)
+ return -1;
+
+ newtFormSetTimer(browser->b.form, delay_msecs);
+
+ while (1) {
+ key = ui_browser__run(&browser->b);
+
+ switch (key) {
+ case -1:
+ /* FIXME we need to check if it was es.reason == NEWT_EXIT_TIMER */
+ perf_top_browser__update_rb_tree(browser);
+ perf_top__header_snprintf(top, title, sizeof(title));
+ perf_top__reset_sample_counters(top);
+ ui_browser__set_color(&browser->b, NEWT_COLORSET_ROOT);
+ SLsmg_gotorc(0, 0);
+ slsmg_write_nstring(title, browser->b.width);
+ break;
+ case NEWT_KEY_TAB:
+ default:
+ goto out;
+ }
+ }
+out:
+ ui_browser__hide(&browser->b);
+ return key;
+}
+
+int perf_top__tui_browser(struct perf_top *top)
+{
+ struct perf_top_browser browser = {
+ .b = {
+ .entries = &browser.root,
+ .refresh = ui_browser__rb_tree_refresh,
+ .seek = ui_browser__rb_tree_seek,
+ .write = perf_top_browser__write,
+ .priv = top,
+ },
+ };
+
+ ui_helpline__push("Press <- or ESC to exit");
+ return perf_top_browser__run(&browser);
+}
--
1.6.2.5
^ permalink raw reply related [flat|nested] 8+ messages in thread
* Re: [GIT PULL 0/5] perf/core improvements
2011-01-31 21:29 [GIT PULL 0/5] perf/core improvements Arnaldo Carvalho de Melo
` (4 preceding siblings ...)
2011-01-31 21:29 ` [PATCH 5/5] perf top: Introduce slang based TUI Arnaldo Carvalho de Melo
@ 2011-02-01 9:14 ` Ingo Molnar
2011-02-01 14:28 ` Arnaldo Carvalho de Melo
5 siblings, 1 reply; 8+ messages in thread
From: Ingo Molnar @ 2011-02-01 9:14 UTC (permalink / raw)
To: Arnaldo Carvalho de Melo
Cc: linux-kernel, Frederic Weisbecker, Mike Galbraith,
Paul Mackerras, Peter Zijlstra, Stephane Eranian, Tom Zanussi,
Arnaldo Carvalho de Melo
* Arnaldo Carvalho de Melo <acme@infradead.org> wrote:
> Hi Ingo,
>
> Please consider pulling from:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux-2.6 perf/core
>
> Regards,
>
> - Arnaldo
>
> Arnaldo Carvalho de Melo (5):
> perf evlist: Move evlist methods to evlist.c
> perf evlist: Store pointer to the cpu and thread maps
> perf top: Move display agnostic routines to util/top.[ch]
> perf tools: Don't fallback to setup_pager unconditionally
> perf top: Introduce slang based TUI
>
> tools/perf/Makefile | 6 +
> tools/perf/builtin-annotate.c | 2 +-
> tools/perf/builtin-record.c | 44 ++--
> tools/perf/builtin-report.c | 2 +-
> tools/perf/builtin-stat.c | 45 ++--
> tools/perf/builtin-test.c | 6 +-
> tools/perf/builtin-top.c | 530 ++++++++++++-------------------------
> tools/perf/python/twatch.py | 4 +-
> tools/perf/util/cache.h | 7 +-
> tools/perf/util/evlist.c | 185 +++++++++++++-
> tools/perf/util/evlist.h | 28 ++-
> tools/perf/util/evsel.c | 144 +----------
> tools/perf/util/evsel.h | 4 -
> tools/perf/util/python.c | 25 +-
> tools/perf/util/top.c | 217 +++++++++++++++
> tools/perf/util/top.h | 80 ++++++
> tools/perf/util/ui/browsers/top.c | 136 ++++++++++
> tools/perf/util/ui/setup.c | 5 +-
> 18 files changed, 887 insertions(+), 583 deletions(-)
> create mode 100644 tools/perf/util/top.c
> create mode 100644 tools/perf/util/top.h
> create mode 100644 tools/perf/util/ui/browsers/top.c
Pulled, thanks Arnaldo!
Note, there's a beauty wart i noticed, while building perf on 32-bit Fedora: the build output
includes new, unintended lines:
CC util/scripting-engines/trace-event-perl.o
CC scripts/perl/Perf-Trace-Util/Context.o
CC util/scripting-engines/trace-event-python.o
CC scripts/python/Perf-Trace-Util/Context.o
CC perf.o
CC builtin-help.o
gcc -pthread -fno-strict-aliasing -DNDEBUG -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i686 -mtune=atom -fasynchronous-unwind-tables -D_GNU_SOURCE -fPIC -fwrapv -fPIC -Iutil/include -I/usr/include/python2.6 -c util/evlist.c -o temp.linux-i686-2.6/util/evlist.o -fno-strict-aliasing -Wno-write-strings
gcc -pthread -fno-strict-aliasing -DNDEBUG -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i686 -mtune=atom -fasynchronous-unwind-tables -D_GNU_SOURCE -fPIC -fwrapv -fPIC -Iutil/include -I/usr/include/python2.6 -c util/evsel.c -o temp.linux-i686-2.6/util/evsel.o -fno-strict-aliasing -Wno-write-strings
gcc -pthread -fno-strict-aliasing -DNDEBUG -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i686 -mtune=atom -fasynchronous-unwind-tables -D_GNU_SOURCE -fPIC -fwrapv -fPIC -Iutil/include -I/usr/include/python2.6 -c util/cpumap.c -o temp.linux-i686-2.6/util/cpumap.o -fno-strict-aliasing -Wno-write-strings
gcc -pthread -fno-strict-aliasing -DNDEBUG -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i686 -mtune=atom -fasynchronous-unwind-tables -D_GNU_SOURCE -fPIC -fwrapv -fPIC -Iutil/include -I/usr/include/python2.6 -c util/thread_map.c -o temp.linux-i686-2.6/util/thread_map.o -fno-strict-aliasing -Wno-write-strings
gcc -pthread -fno-strict-aliasing -DNDEBUG -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i686 -mtune=atom -fasynchronous-unwind-tables -D_GNU_SOURCE -fPIC -fwrapv -fPIC -Iutil/include -I/usr/include/python2.6 -c util/util.c -o temp.linux-i686-2.6/util/util.o -fno-strict-aliasing -Wno-write-strings
gcc -pthread -fno-strict-aliasing -DNDEBUG -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i686 -mtune=atom -fasynchronous-unwind-tables -D_GNU_SOURCE -fPIC -fwrapv -fPIC -Iutil/include -I/usr/include/python2.6 -c util/xyarray.c -o temp.linux-i686-2.6/util/xyarray.o -fno-strict-aliasing -Wno-write-strings
gcc -pthread -shared temp.linux-i686-2.6/util/python.o temp.linux-i686-2.6/util/ctype.o temp.linux-i686-2.6/util/evlist.o temp.linux-i686-2.6/util/evsel.o temp.linux-i686-2.6/util/cpumap.o temp.linux-i686-2.6/util/thread_map.o temp.linux-i686-2.6/util/util.o temp.linux-i686-2.6/util/xyarray.o -L/usr/lib -lpython2.6 -o lib.linux-i686-2.6/perf.so
AR libperf.a
LINK perf
See all tose 'gcc -pthread ...' lines? Those should have the simplified 'CC '-alike
output as well i suspect.
Thanks,
Ingo
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [GIT PULL 0/5] perf/core improvements
2011-02-01 9:14 ` [GIT PULL 0/5] perf/core improvements Ingo Molnar
@ 2011-02-01 14:28 ` Arnaldo Carvalho de Melo
0 siblings, 0 replies; 8+ messages in thread
From: Arnaldo Carvalho de Melo @ 2011-02-01 14:28 UTC (permalink / raw)
To: Ingo Molnar
Cc: linux-kernel, Frederic Weisbecker, Mike Galbraith,
Paul Mackerras, Peter Zijlstra, Stephane Eranian, Tom Zanussi
Em Tue, Feb 01, 2011 at 10:14:47AM +0100, Ingo Molnar escreveu:
>
> * Arnaldo Carvalho de Melo <acme@infradead.org> wrote:
>
> > Hi Ingo,
> >
> > Please consider pulling from:
> >
> > git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux-2.6 perf/core
> >
>
> Pulled, thanks Arnaldo!
>
> Note, there's a beauty wart i noticed, while building perf on 32-bit Fedora: the build output
> includes new, unintended lines:
>
> CC util/scripting-engines/trace-event-perl.o
> CC scripts/perl/Perf-Trace-Util/Context.o
> CC util/scripting-engines/trace-event-python.o
> CC scripts/python/Perf-Trace-Util/Context.o
> CC perf.o
> CC builtin-help.o
> gcc -pthread -fno-strict-aliasing -DNDEBUG -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i686 -mtune=atom -fasynchronous-unwind-tables -D_GNU_SOURCE -fPIC -fwrapv -fPIC -Iutil/include -I/usr/include/python2.6 -c util/evlist.c -o temp.linux-i686-2.6/util/evlist.o -fno-strict-aliasing -Wno-write-strings
> gcc -pthread -fno-strict-aliasing -DNDEBUG -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m32 -march=i686 -mtune=atom -fasynchronous-unwind-tables -D_GNU_SOURCE -fPIC -fwrapv -fPIC -Iutil/include -I/usr/include/python2.6 -c util/evsel.c -o temp.linux-i686-2.6/util/evsel.o -fno-strict-aliasing -Wno-write-strings
Yeah, I noticed that too, its on my todo list, its the setup.py python
building part, I need to either make it silent and provide just one:
PYTHON perf.so
message or to ditch using setup.py and do everything directly, which I
may need to do anyway because 'python setup.py --build-base" (or
equivalent, that I haven't found) doesn't work with 'install', just with
'build'.
For now its just a nuisance, not something crucial, but I'll get that
nailed.
- Arnaldo
^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2011-02-01 14:28 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2011-01-31 21:29 [GIT PULL 0/5] perf/core improvements Arnaldo Carvalho de Melo
2011-01-31 21:29 ` [PATCH 1/5] perf evlist: Move evlist methods to evlist.c Arnaldo Carvalho de Melo
2011-01-31 21:29 ` [PATCH 2/5] perf evlist: Store pointer to the cpu and thread maps Arnaldo Carvalho de Melo
2011-01-31 21:29 ` [PATCH 3/5] perf top: Move display agnostic routines to util/top.[ch] Arnaldo Carvalho de Melo
2011-01-31 21:29 ` [PATCH 4/5] perf tools: Don't fallback to setup_pager unconditionally Arnaldo Carvalho de Melo
2011-01-31 21:29 ` [PATCH 5/5] perf top: Introduce slang based TUI Arnaldo Carvalho de Melo
2011-02-01 9:14 ` [GIT PULL 0/5] perf/core improvements Ingo Molnar
2011-02-01 14:28 ` Arnaldo Carvalho de Melo
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.