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

Groups > linux.kernel > #1443025 > unrolled thread

[GIT PULL 00/19] perf/core improvements and fixes

Started byArnaldo Carvalho de Melo <acme@kernel.org>
First post2016-07-14 04:30 +0200
Last post2016-07-14 09:00 +0200
Articles 6 — 2 participants

Back to article view | Back to linux.kernel

Contents

  [GIT PULL 00/19] perf/core improvements and fixes Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-07-14 04:30 +0200
    [PATCH 13/19] perf probe: Allow wildcard for cached events Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-07-14 04:30 +0200
    [PATCH 17/19] perf probe: Support a special SDT probe format Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-07-14 04:30 +0200
    [PATCH 07/19] perf bpf: Rename bpf__foreach_tev() to bpf__foreach_event() Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-07-14 04:30 +0200
    [PATCH 19/19] perf test: Add a test case for SDT event Arnaldo Carvalho de Melo <acme@kernel.org> - 2016-07-14 04:30 +0200
    Re: [GIT PULL 00/19] perf/core improvements and fixes Ingo Molnar <mingo@kernel.org> - 2016-07-14 09:00 +0200

#1443025 — [GIT PULL 00/19] perf/core improvements and fixes

FromArnaldo Carvalho de Melo <acme@kernel.org>
Date2016-07-14 04:30 +0200
Subject[GIT PULL 00/19] perf/core improvements and fixes
Message-ID<rUEzv-3o6-3@gated-at.bofh.it>
Hi Ingo,

	Please consider pulling,

I've added building objtool to most of the containers in my build test setup:

[root@jouet ~]# perf stat dm
alpine:3.4: Ok
centos:5: Ok
centos:6: Ok
centos:7: Ok
debian:7: Ok
debian:8: Ok
debian:experimental: Ok
fedora:21: Ok
fedora:22: Ok
fedora:23: Ok
fedora:24: Ok
fedora:rawhide: Ok
mageia:5: Ok
opensuse:13.2: Ok
opensuse:42.1: Ok
ubuntu:12.04.5: Ok
ubuntu:14.04.4: Ok
ubuntu:15.10: Ok
ubuntu:16.04: Ok

 Performance counter stats for 'dm':

       2601.121782      task-clock (msec)         #    0.002 CPUs utilized          
            86,368      context-switches          #    0.033 M/sec                  
             5,740      cpu-migrations            #    0.002 M/sec                  
            53,962      page-faults               #    0.021 M/sec                  
     7,217,605,183      cycles                    #    2.775 GHz                    
     6,534,540,119      instructions              #    0.91  insn per cycle         
     1,408,715,184      branches                  #  541.580 M/sec                  
        18,523,459      branch-misses             #    1.31% of all branches        

    1541.746171526 seconds time elapsed

[root@jouet ~]# 

- Arnaldo

The following changes since commit 7b39cafb7aa68ef8e32a9f51fbe737d96084ca74:

  tools: Work around BITS_PER_LONG related build failure in objtool (2016-07-13 09:37:43 +0200)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git tags/perf-core-for-mingo-20160713

for you to fetch changes up to 8e5dc848356ecf6ea8d27d641c4d7ad8d42fe92b:

  perf test: Add a test case for SDT event (2016-07-13 23:09:10 -0300)

----------------------------------------------------------------
perf/core improvements and fixes:

User visible:

- Finish merging initial SDT (Statically Defined Traces) support, see
  cset comments for details about how it all works (Masami Hiramatsu)

- Support attaching eBPF programs to tracepoints (Wang Nan)

Infrastructure:

- Fix up BITS_PER_LONG setting (Arnaldo Carvalho de Melo)

- Add fallback from ELF_C_READ_MMAP to ELF_C_READ in objtool, fixing
  the build in libelf implementations lacking that elf_begin() cmd,
  such as Alpine Linux's (Arnaldo Carvalho de Melo)

- Avoid checking code drift on busybox's diff in objtool (Arnaldo Carvalho de Melo)

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

----------------------------------------------------------------
Arnaldo Carvalho de Melo (3):
      tools: Fix up BITS_PER_LONG setting
      objtool: Add fallback from ELF_C_READ_MMAP to ELF_C_READ
      objtool: Avoid checking code drift on busybox's diff

Masami Hiramatsu (11):
      perf probe: Fix to show correct error message for $vars and $params
      perf probe: Accept %sdt and %cached event name
      perf probe: Make --list show only available cached events
      perf probe-cache: Add for_each_probe_cache_entry() wrapper
      perf probe: Allow wildcard for cached events
      perf probe: Search SDT/cached event from all probe caches
      perf list: Show SDT and pre-cached events
      perf probe: Support @BUILDID or @FILE suffix for SDT events
      perf probe: Support a special SDT probe format
      perf build: Add sdt feature detection
      perf test: Add a test case for SDT event

Wang Nan (5):
      tools lib bpf: New API to adjust type of a BPF program
      tools lib bpf: Report error when kernel doesn't support program type
      perf event parser: Add const qualifier to evt_name and sys_name
      perf bpf: Rename bpf__foreach_tev() to bpf__foreach_event()
      perf bpf: Support BPF program attach to tracepoints

 tools/build/Makefile.feature            |   3 +-
 tools/build/feature/Makefile            |   6 +-
 tools/build/feature/test-all.c          |   5 +
 tools/build/feature/test-sdt.c          |   7 +
 tools/include/asm-generic/bitsperlong.h |  24 ++-
 tools/lib/bpf/libbpf.c                  |  80 +++++++--
 tools/lib/bpf/libbpf.h                  |  10 ++
 tools/objtool/Makefile                  |   5 +-
 tools/objtool/elf.c                     |   7 +
 tools/perf/Documentation/perf-probe.txt |  11 +-
 tools/perf/Makefile.perf                |   3 +
 tools/perf/builtin-list.c               |   6 +-
 tools/perf/builtin-probe.c              |   2 +-
 tools/perf/config/Makefile              |  10 ++
 tools/perf/tests/Build                  |   1 +
 tools/perf/tests/builtin-test.c         |   4 +
 tools/perf/tests/make                   |   3 +-
 tools/perf/tests/sdt.c                  | 115 ++++++++++++
 tools/perf/tests/tests.h                |   1 +
 tools/perf/util/bpf-loader.c            |  73 +++++++-
 tools/perf/util/bpf-loader.h            |  12 +-
 tools/perf/util/build-id.c              |  76 +++++++-
 tools/perf/util/build-id.h              |   3 +-
 tools/perf/util/parse-events.c          | 110 ++++++++++--
 tools/perf/util/parse-events.h          |   4 +-
 tools/perf/util/probe-event.c           | 309 +++++++++++++++++++++++++++-----
 tools/perf/util/probe-event.h           |   1 +
 tools/perf/util/probe-file.c            |  57 ++++--
 tools/perf/util/probe-file.h            |  14 ++
 29 files changed, 850 insertions(+), 112 deletions(-)
 create mode 100644 tools/build/feature/test-sdt.c
 create mode 100644 tools/perf/tests/sdt.c

[toc] | [next] | [standalone]

#1443026 — [PATCH 13/19] perf probe: Allow wildcard for cached events

FromArnaldo Carvalho de Melo <acme@kernel.org>
Date2016-07-14 04:30 +0200
Subject[PATCH 13/19] perf probe: Allow wildcard for cached events
Message-ID<rUEzx-3o6-41@gated-at.bofh.it>
In reply to#1443025
From: Masami Hiramatsu <mhiramat@kernel.org>

Allo glob wildcard for reusing cached/SDT events. E.g.

  # perf probe -x /usr/lib64/libc-2.20.so -a %sdt_libc:\*

This example adds probes for all SDT in libc.
Note that the SDTs must have been scanned by perf buildid-cache.

Committer note:

Using it to check what of those SDT probes would take place when doing
a cargo run (rust):

  # trace --no-sys --event sdt_libc:* cargo run
     0.000 sdt_libc:setjmp:(7f326b69c4d1))
    28.423 sdt_libc:setjmp:(7f4b0a5364d1))
    29.000 sdt_libc:setjmp:(7f4b0a5364d1))
    88.597 sdt_libc:setjmp:(7fc01fd414d1))
    89.220 sdt_libc:setjmp:(7fc01fd414d1))
    95.501 sdt_libc:setjmp:(7f326b69c4d1))
     Running `target/debug/hello_world`
    97.110 sdt_libc:setjmp:(7f95e09234d1))
  Hello, world!
  #

Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@linux.vnet.ibm.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Hemant Kumar <hemant@linux.vnet.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/146831791813.17065.17846564230840594888.stgit@devbox
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/util/probe-event.c | 107 ++++++++++++++++++++++++++++++++++++++++--
 tools/perf/util/probe-file.c  |  38 ++++++++++++---
 tools/perf/util/probe-file.h  |   3 ++
 3 files changed, 138 insertions(+), 10 deletions(-)

diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c
index 85f25d41cf8d..7b96e687568e 100644
--- a/tools/perf/util/probe-event.c
+++ b/tools/perf/util/probe-event.c
@@ -1204,7 +1204,7 @@ static int parse_perf_probe_event_name(char **arg, struct perf_probe_event *pev)
 	ptr = strchr(*arg, ':');
 	if (ptr) {
 		*ptr = '\0';
-		if (!is_c_func_name(*arg))
+		if (!pev->sdt && !is_c_func_name(*arg))
 			goto ng_name;
 		pev->group = strdup(*arg);
 		if (!pev->group)
@@ -1212,7 +1212,7 @@ static int parse_perf_probe_event_name(char **arg, struct perf_probe_event *pev)
 		*arg = ptr + 1;
 	} else
 		pev->group = NULL;
-	if (!is_c_func_name(*arg)) {
+	if (!pev->sdt && !is_c_func_name(*arg)) {
 ng_name:
 		semantic_error("%s is bad for event name -it must "
 			       "follow C symbol-naming rule.\n", *arg);
@@ -1644,6 +1644,7 @@ int parse_probe_trace_command(const char *cmd, struct probe_trace_event *tev)
 			ret = -ENOMEM;
 			goto out;
 		}
+		tev->uprobes = (tp->module[0] == '/');
 		p++;
 	} else
 		p = argv[1];
@@ -2518,7 +2519,7 @@ static int probe_trace_event__set_name(struct probe_trace_event *tev,
 	int ret;
 
 	/* If probe_event or trace_event already have the name, reuse it */
-	if (pev->event)
+	if (pev->event && !pev->sdt)
 		event = pev->event;
 	else if (tev->event)
 		event = tev->event;
@@ -2531,7 +2532,7 @@ static int probe_trace_event__set_name(struct probe_trace_event *tev,
 		else
 			event = tev->point.realname;
 	}
-	if (pev->group)
+	if (pev->group && !pev->sdt)
 		group = pev->group;
 	else if (tev->group)
 		group = tev->group;
@@ -2894,6 +2895,100 @@ errout:
 
 bool __weak arch__prefers_symtab(void) { return false; }
 
+/* Concatinate two arrays */
+static void *memcat(void *a, size_t sz_a, void *b, size_t sz_b)
+{
+	void *ret;
+
+	ret = malloc(sz_a + sz_b);
+	if (ret) {
+		memcpy(ret, a, sz_a);
+		memcpy(ret + sz_a, b, sz_b);
+	}
+	return ret;
+}
+
+static int
+concat_probe_trace_events(struct probe_trace_event **tevs, int *ntevs,
+			  struct probe_trace_event **tevs2, int ntevs2)
+{
+	struct probe_trace_event *new_tevs;
+	int ret = 0;
+
+	if (ntevs == 0) {
+		*tevs = *tevs2;
+		*ntevs = ntevs2;
+		*tevs2 = NULL;
+		return 0;
+	}
+
+	if (*ntevs + ntevs2 > probe_conf.max_probes)
+		ret = -E2BIG;
+	else {
+		/* Concatinate the array of probe_trace_event */
+		new_tevs = memcat(*tevs, (*ntevs) * sizeof(**tevs),
+				  *tevs2, ntevs2 * sizeof(**tevs2));
+		if (!new_tevs)
+			ret = -ENOMEM;
+		else {
+			free(*tevs);
+			*tevs = new_tevs;
+			*ntevs += ntevs2;
+		}
+	}
+	if (ret < 0)
+		clear_probe_trace_events(*tevs2, ntevs2);
+	zfree(tevs2);
+
+	return ret;
+}
+
+/*
+ * Try to find probe_trace_event from given probe caches. Return the number
+ * of cached events found, if an error occurs return the error.
+ */
+static int find_cached_events(struct perf_probe_event *pev,
+			      struct probe_trace_event **tevs,
+			      const char *target)
+{
+	struct probe_cache *cache;
+	struct probe_cache_entry *entry;
+	struct probe_trace_event *tmp_tevs = NULL;
+	int ntevs = 0;
+	int ret = 0;
+
+	cache = probe_cache__new(target);
+	/* Return 0 ("not found") if the target has no probe cache. */
+	if (!cache)
+		return 0;
+
+	for_each_probe_cache_entry(entry, cache) {
+		/* Skip the cache entry which has no name */
+		if (!entry->pev.event || !entry->pev.group)
+			continue;
+		if ((!pev->group || strglobmatch(entry->pev.group, pev->group)) &&
+		    strglobmatch(entry->pev.event, pev->event)) {
+			ret = probe_cache_entry__get_event(entry, &tmp_tevs);
+			if (ret > 0)
+				ret = concat_probe_trace_events(tevs, &ntevs,
+								&tmp_tevs, ret);
+			if (ret < 0)
+				break;
+		}
+	}
+	probe_cache__delete(cache);
+	if (ret < 0) {
+		clear_probe_trace_events(*tevs, ntevs);
+		zfree(tevs);
+	} else {
+		ret = ntevs;
+		if (ntevs > 0 && target && target[0] == '/')
+			pev->uprobes = true;
+	}
+
+	return ret;
+}
+
 static int find_probe_trace_events_from_cache(struct perf_probe_event *pev,
 					      struct probe_trace_event **tevs)
 {
@@ -2903,6 +2998,10 @@ static int find_probe_trace_events_from_cache(struct perf_probe_event *pev,
 	struct str_node *node;
 	int ret, i;
 
+	if (pev->sdt)
+		/* For SDT/cached events, we use special search functions */
+		return find_cached_events(pev, tevs, pev->target);
+
 	cache = probe_cache__new(pev->target);
 	if (!cache)
 		return 0;
diff --git a/tools/perf/util/probe-file.c b/tools/perf/util/probe-file.c
index abfb05cf135b..9aed9c332da6 100644
--- a/tools/perf/util/probe-file.c
+++ b/tools/perf/util/probe-file.c
@@ -362,13 +362,38 @@ probe_cache_entry__new(struct perf_probe_event *pev)
 	return entry;
 }
 
-/* For the kernel probe caches, pass target = NULL */
+int probe_cache_entry__get_event(struct probe_cache_entry *entry,
+				 struct probe_trace_event **tevs)
+{
+	struct probe_trace_event *tev;
+	struct str_node *node;
+	int ret, i;
+
+	ret = strlist__nr_entries(entry->tevlist);
+	if (ret > probe_conf.max_probes)
+		return -E2BIG;
+
+	*tevs = zalloc(ret * sizeof(*tev));
+	if (!*tevs)
+		return -ENOMEM;
+
+	i = 0;
+	strlist__for_each_entry(node, entry->tevlist) {
+		tev = &(*tevs)[i++];
+		ret = parse_probe_trace_command(node->s, tev);
+		if (ret < 0)
+			break;
+	}
+	return i;
+}
+
+/* For the kernel probe caches, pass target = NULL or DSO__NAME_KALLSYMS */
 static int probe_cache__open(struct probe_cache *pcache, const char *target)
 {
 	char cpath[PATH_MAX];
 	char sbuildid[SBUILD_ID_SIZE];
 	char *dir_name = NULL;
-	bool is_kallsyms = !target;
+	bool is_kallsyms = false;
 	int ret, fd;
 
 	if (target && build_id_cache__cached(target)) {
@@ -378,12 +403,13 @@ static int probe_cache__open(struct probe_cache *pcache, const char *target)
 		goto found;
 	}
 
-	if (target)
-		ret = filename__sprintf_build_id(target, sbuildid);
-	else {
+	if (!target || !strcmp(target, DSO__NAME_KALLSYMS)) {
 		target = DSO__NAME_KALLSYMS;
+		is_kallsyms = true;
 		ret = sysfs__sprintf_build_id("/", sbuildid);
-	}
+	} else
+		ret = filename__sprintf_build_id(target, sbuildid);
+
 	if (ret < 0) {
 		pr_debug("Failed to get build-id from %s.\n", target);
 		return ret;
diff --git a/tools/perf/util/probe-file.h b/tools/perf/util/probe-file.h
index d513b346a70e..cafbe1d3f3bf 100644
--- a/tools/perf/util/probe-file.h
+++ b/tools/perf/util/probe-file.h
@@ -34,6 +34,9 @@ int probe_file__get_events(int fd, struct strfilter *filter,
 				  struct strlist *plist);
 int probe_file__del_strlist(int fd, struct strlist *namelist);
 
+int probe_cache_entry__get_event(struct probe_cache_entry *entry,
+				 struct probe_trace_event **tevs);
+
 struct probe_cache *probe_cache__new(const char *target);
 int probe_cache__add_entry(struct probe_cache *pcache,
 			   struct perf_probe_event *pev,
-- 
2.7.4

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

#1443027 — [PATCH 17/19] perf probe: Support a special SDT probe format

FromArnaldo Carvalho de Melo <acme@kernel.org>
Date2016-07-14 04:30 +0200
Subject[PATCH 17/19] perf probe: Support a special SDT probe format
Message-ID<rUEzx-3o6-43@gated-at.bofh.it>
In reply to#1443025
From: Masami Hiramatsu <mhiramat@kernel.org>

Support a special SDT probe format which can omit the '%' prefix only if
the SDT group name starts with "sdt_". So, for example both of
"%sdt_libc:setjump" and "sdt_libc:setjump" are acceptable for perf probe
--add.

E.g. without this:

  # perf probe -a sdt_libc:setjmp
  Semantic error :There is non-digit char in line number.
  ...

With this:

  # perf probe -a sdt_libc:setjmp
  Added new event:
    sdt_libc:setjmp      (on %setjmp in /usr/lib64/libc-2.20.so)

  You can now use it in all perf tools, such as:

  	perf record -e sdt_libc:setjmp -aR sleep 1

Suggested-by: Brendan Gregg <brendan.d.gregg@gmail.com>
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@linux.vnet.ibm.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Hemant Kumar <hemant@linux.vnet.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/146831794674.17065.13359473252168740430.stgit@devbox
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/Documentation/perf-probe.txt |  4 +++-
 tools/perf/util/probe-event.c           | 12 ++++++++++--
 2 files changed, 13 insertions(+), 3 deletions(-)

diff --git a/tools/perf/Documentation/perf-probe.txt b/tools/perf/Documentation/perf-probe.txt
index 39e387042098..736da44596e4 100644
--- a/tools/perf/Documentation/perf-probe.txt
+++ b/tools/perf/Documentation/perf-probe.txt
@@ -152,7 +152,9 @@ Probe points are defined by following syntax.
      [[GROUP:]EVENT=]SRC;PTN [ARG ...]
 
     4) Pre-defined SDT events or cached event with name
-     %[PROVIDER:]SDTEVENT
+     %[sdt_PROVIDER:]SDTEVENT
+     or,
+     sdt_PROVIDER:SDTEVENT
 
 'EVENT' specifies the name of new event, if omitted, it will be set the name of the probed function. You can also specify a group name by 'GROUP', if omitted, set 'probe' is used for kprobe and 'probe_<bin>' is used for uprobe.
 Note that using existing group name can conflict with other events. Especially, using the group name reserved for kernel modules can hide embedded events in the
diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c
index f12081e48a32..d4f8835c0a27 100644
--- a/tools/perf/util/probe-event.c
+++ b/tools/perf/util/probe-event.c
@@ -1243,9 +1243,17 @@ static int parse_perf_probe_point(char *arg, struct perf_probe_event *pev)
 	if (!arg)
 		return -EINVAL;
 
-	if (arg[0] == '%') {
+	/*
+	 * If the probe point starts with '%',
+	 * or starts with "sdt_" and has a ':' but no '=',
+	 * then it should be a SDT/cached probe point.
+	 */
+	if (arg[0] == '%' ||
+	    (!strncmp(arg, "sdt_", 4) &&
+	     !!strchr(arg, ':') && !strchr(arg, '='))) {
 		pev->sdt = true;
-		arg++;
+		if (arg[0] == '%')
+			arg++;
 	}
 
 	ptr = strpbrk(arg, ";=@+%");
-- 
2.7.4

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

#1443028 — [PATCH 07/19] perf bpf: Rename bpf__foreach_tev() to bpf__foreach_event()

FromArnaldo Carvalho de Melo <acme@kernel.org>
Date2016-07-14 04:30 +0200
Subject[PATCH 07/19] perf bpf: Rename bpf__foreach_tev() to bpf__foreach_event()
Message-ID<rUEzx-3o6-45@gated-at.bofh.it>
In reply to#1443025
From: Wang Nan <wangnan0@huawei.com>

Following commit will allow BPF script attach to tracepoints.
bpf__foreach_tev() will iterate over all events, not only kprobes.
Rename it to bpf__foreach_event().

Since only group and event are used by caller, there's no need to pass
full 'struct probe_trace_event' to bpf_prog_iter_callback_t. Pass only
these two strings. After this patch bpf_prog_iter_callback_t natually
support tracepoints.

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Zefan Li <lizefan@huawei.com>
Cc: pi3orama@163.com
Link: http://lkml.kernel.org/r/1468406646-21642-5-git-send-email-wangnan0@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/util/bpf-loader.c   |  8 ++++----
 tools/perf/util/bpf-loader.h   | 12 ++++++------
 tools/perf/util/parse-events.c | 16 ++++++++--------
 3 files changed, 18 insertions(+), 18 deletions(-)

diff --git a/tools/perf/util/bpf-loader.c b/tools/perf/util/bpf-loader.c
index 8445e89621fe..f22701497112 100644
--- a/tools/perf/util/bpf-loader.c
+++ b/tools/perf/util/bpf-loader.c
@@ -693,9 +693,9 @@ int bpf__load(struct bpf_object *obj)
 	return 0;
 }
 
-int bpf__foreach_tev(struct bpf_object *obj,
-		     bpf_prog_iter_callback_t func,
-		     void *arg)
+int bpf__foreach_event(struct bpf_object *obj,
+		       bpf_prog_iter_callback_t func,
+		       void *arg)
 {
 	struct bpf_program *prog;
 	int err;
@@ -728,7 +728,7 @@ int bpf__foreach_tev(struct bpf_object *obj,
 				return fd;
 			}
 
-			err = (*func)(tev, fd, arg);
+			err = (*func)(tev->group, tev->event, fd, arg);
 			if (err) {
 				pr_debug("bpf: call back failed, stop iterate\n");
 				return err;
diff --git a/tools/perf/util/bpf-loader.h b/tools/perf/util/bpf-loader.h
index 941e17275aa7..f2b737b225f2 100644
--- a/tools/perf/util/bpf-loader.h
+++ b/tools/perf/util/bpf-loader.h
@@ -46,7 +46,7 @@ struct bpf_object;
 struct parse_events_term;
 #define PERF_BPF_PROBE_GROUP "perf_bpf_probe"
 
-typedef int (*bpf_prog_iter_callback_t)(struct probe_trace_event *tev,
+typedef int (*bpf_prog_iter_callback_t)(const char *group, const char *event,
 					int fd, void *arg);
 
 #ifdef HAVE_LIBBPF_SUPPORT
@@ -67,8 +67,8 @@ int bpf__strerror_probe(struct bpf_object *obj, int err,
 int bpf__load(struct bpf_object *obj);
 int bpf__strerror_load(struct bpf_object *obj, int err,
 		       char *buf, size_t size);
-int bpf__foreach_tev(struct bpf_object *obj,
-		     bpf_prog_iter_callback_t func, void *arg);
+int bpf__foreach_event(struct bpf_object *obj,
+		       bpf_prog_iter_callback_t func, void *arg);
 
 int bpf__config_obj(struct bpf_object *obj, struct parse_events_term *term,
 		    struct perf_evlist *evlist, int *error_pos);
@@ -107,9 +107,9 @@ static inline int bpf__unprobe(struct bpf_object *obj __maybe_unused) { return 0
 static inline int bpf__load(struct bpf_object *obj __maybe_unused) { return 0; }
 
 static inline int
-bpf__foreach_tev(struct bpf_object *obj __maybe_unused,
-		 bpf_prog_iter_callback_t func __maybe_unused,
-		 void *arg __maybe_unused)
+bpf__foreach_event(struct bpf_object *obj __maybe_unused,
+		   bpf_prog_iter_callback_t func __maybe_unused,
+		   void *arg __maybe_unused)
 {
 	return 0;
 }
diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c
index d866824f2b42..6b4fff375d99 100644
--- a/tools/perf/util/parse-events.c
+++ b/tools/perf/util/parse-events.c
@@ -584,7 +584,7 @@ struct __add_bpf_event_param {
 	struct list_head *head_config;
 };
 
-static int add_bpf_event(struct probe_trace_event *tev, int fd,
+static int add_bpf_event(const char *group, const char *event, int fd,
 			 void *_param)
 {
 	LIST_HEAD(new_evsels);
@@ -595,27 +595,27 @@ static int add_bpf_event(struct probe_trace_event *tev, int fd,
 	int err;
 
 	pr_debug("add bpf event %s:%s and attach bpf program %d\n",
-		 tev->group, tev->event, fd);
+		 group, event, fd);
 
-	err = parse_events_add_tracepoint(&new_evsels, &evlist->idx, tev->group,
-					  tev->event, evlist->error,
+	err = parse_events_add_tracepoint(&new_evsels, &evlist->idx, group,
+					  event, evlist->error,
 					  param->head_config);
 	if (err) {
 		struct perf_evsel *evsel, *tmp;
 
 		pr_debug("Failed to add BPF event %s:%s\n",
-			 tev->group, tev->event);
+			 group, event);
 		list_for_each_entry_safe(evsel, tmp, &new_evsels, node) {
 			list_del(&evsel->node);
 			perf_evsel__delete(evsel);
 		}
 		return err;
 	}
-	pr_debug("adding %s:%s\n", tev->group, tev->event);
+	pr_debug("adding %s:%s\n", group, event);
 
 	list_for_each_entry(pos, &new_evsels, node) {
 		pr_debug("adding %s:%s to %p\n",
-			 tev->group, tev->event, pos);
+			 group, event, pos);
 		pos->bpf_fd = fd;
 	}
 	list_splice(&new_evsels, list);
@@ -661,7 +661,7 @@ int parse_events_load_bpf_obj(struct parse_events_evlist *data,
 		goto errout;
 	}
 
-	err = bpf__foreach_tev(obj, add_bpf_event, &param);
+	err = bpf__foreach_event(obj, add_bpf_event, &param);
 	if (err) {
 		snprintf(errbuf, sizeof(errbuf),
 			 "Attach events in BPF object failed");
-- 
2.7.4

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

#1443029 — [PATCH 19/19] perf test: Add a test case for SDT event

FromArnaldo Carvalho de Melo <acme@kernel.org>
Date2016-07-14 04:30 +0200
Subject[PATCH 19/19] perf test: Add a test case for SDT event
Message-ID<rUEzx-3o6-47@gated-at.bofh.it>
In reply to#1443025
From: Masami Hiramatsu <mhiramat@kernel.org>

Add a basic test case for SDT event support.  This test scans an SDT
event in perftools and check whether the SDT event is correctly stored
into the buildid cache.

Here is an example:

  ----
  $ perf test sdt -v
  47: Test SDT event probing                                   :
  --- start ---
  test child forked, pid 20732
  Found 72 SDTs in /home/mhiramat/ksrc/linux/tools/perf/perf
  Writing cache: %sdt_perf:test_target=test_target
  Cache committed: 0
  symbol:test_target file:(null) line:0 offset:0 return:0 lazy:(null)
  test child finished with 0
  ---- end ----
  Test SDT event probing: Ok
  ----

Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@linux.vnet.ibm.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Hemant Kumar <hemant@linux.vnet.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/146831796546.17065.1502584370844087537.stgit@devbox
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/tests/Build          |   1 +
 tools/perf/tests/builtin-test.c |   4 ++
 tools/perf/tests/sdt.c          | 115 ++++++++++++++++++++++++++++++++++++++++
 tools/perf/tests/tests.h        |   1 +
 4 files changed, 121 insertions(+)
 create mode 100644 tools/perf/tests/sdt.c

diff --git a/tools/perf/tests/Build b/tools/perf/tests/Build
index 66a28982547b..4158422cc2a6 100644
--- a/tools/perf/tests/Build
+++ b/tools/perf/tests/Build
@@ -39,6 +39,7 @@ perf-y += stat.o
 perf-y += event_update.o
 perf-y += event-times.o
 perf-y += backward-ring-buffer.o
+perf-y += sdt.o
 
 $(OUTPUT)tests/llvm-src-base.c: tests/bpf-script-example.c tests/Build
 	$(call rule_mkdir)
diff --git a/tools/perf/tests/builtin-test.c b/tools/perf/tests/builtin-test.c
index c23cbf733549..4dd2d050788a 100644
--- a/tools/perf/tests/builtin-test.c
+++ b/tools/perf/tests/builtin-test.c
@@ -218,6 +218,10 @@ static struct test generic_tests[] = {
 		.func = test__cpu_map_print,
 	},
 	{
+		.desc = "Test SDT event probing",
+		.func = test__sdt_event,
+	},
+	{
 		.func = NULL,
 	},
 };
diff --git a/tools/perf/tests/sdt.c b/tools/perf/tests/sdt.c
new file mode 100644
index 000000000000..f59d210e1baf
--- /dev/null
+++ b/tools/perf/tests/sdt.c
@@ -0,0 +1,115 @@
+#include <stdio.h>
+#include <sys/epoll.h>
+#include <util/util.h>
+#include <util/evlist.h>
+#include <linux/filter.h>
+#include "tests.h"
+#include "debug.h"
+#include "probe-file.h"
+#include "build-id.h"
+
+/* To test SDT event, we need libelf support to scan elf binary */
+#if defined(HAVE_SDT_EVENT) && defined(HAVE_LIBELF_SUPPORT)
+
+#include <sys/sdt.h>
+
+static int target_function(void)
+{
+	DTRACE_PROBE(perf, test_target);
+	return TEST_OK;
+}
+
+/* Copied from builtin-buildid-cache.c */
+static int build_id_cache__add_file(const char *filename)
+{
+	char sbuild_id[SBUILD_ID_SIZE];
+	u8 build_id[BUILD_ID_SIZE];
+	int err;
+
+	err = filename__read_build_id(filename, &build_id, sizeof(build_id));
+	if (err < 0) {
+		pr_debug("Failed to read build id of %s\n", filename);
+		return err;
+	}
+
+	build_id__sprintf(build_id, sizeof(build_id), sbuild_id);
+	err = build_id_cache__add_s(sbuild_id, filename, false, false);
+	if (err < 0)
+		pr_debug("Failed to add build id cache of %s\n", filename);
+	return err;
+}
+
+static char *get_self_path(void)
+{
+	char *buf = calloc(PATH_MAX, sizeof(char));
+
+	if (buf && readlink("/proc/self/exe", buf, PATH_MAX) < 0) {
+		pr_debug("Failed to get correct path of perf\n");
+		free(buf);
+		return NULL;
+	}
+	return buf;
+}
+
+static int search_cached_probe(const char *target,
+			       const char *group, const char *event)
+{
+	struct probe_cache *cache = probe_cache__new(target);
+	int ret = 0;
+
+	if (!cache) {
+		pr_debug("Failed to open probe cache of %s\n", target);
+		return -EINVAL;
+	}
+
+	if (!probe_cache__find_by_name(cache, group, event)) {
+		pr_debug("Failed to find %s:%s in the cache\n", group, event);
+		ret = -ENOENT;
+	}
+	probe_cache__delete(cache);
+
+	return ret;
+}
+
+int test__sdt_event(int subtests __maybe_unused)
+{
+	int ret = TEST_FAIL;
+	char __tempdir[] = "./test-buildid-XXXXXX";
+	char *tempdir = NULL, *myself = get_self_path();
+
+	if (myself == NULL || mkdtemp(__tempdir) == NULL) {
+		pr_debug("Failed to make a tempdir for build-id cache\n");
+		goto error;
+	}
+	/* Note that buildid_dir must be an absolute path */
+	tempdir = realpath(__tempdir, NULL);
+
+	/* At first, scan itself */
+	set_buildid_dir(tempdir);
+	if (build_id_cache__add_file(myself) < 0)
+		goto error_rmdir;
+
+	/* Open a cache and make sure the SDT is stored */
+	if (search_cached_probe(myself, "sdt_perf", "test_target") < 0)
+		goto error_rmdir;
+
+	/* TBD: probing on the SDT event and collect logs */
+
+	/* Call the target and get an event */
+	ret = target_function();
+
+error_rmdir:
+	/* Cleanup temporary buildid dir */
+	rm_rf(tempdir);
+error:
+	free(tempdir);
+	free(myself);
+	return ret;
+}
+#else
+int test__sdt_event(int subtests __maybe_unused)
+{
+	pr_debug("Skip SDT event test because SDT support is not compiled\n");
+	return TEST_SKIP;
+}
+#endif
diff --git a/tools/perf/tests/tests.h b/tools/perf/tests/tests.h
index 52f969570c97..a0288f8092b2 100644
--- a/tools/perf/tests/tests.h
+++ b/tools/perf/tests/tests.h
@@ -88,6 +88,7 @@ int test__event_update(int subtest);
 int test__event_times(int subtest);
 int test__backward_ring_buffer(int subtest);
 int test__cpu_map_print(int subtest);
+int test__sdt_event(int subtest);
 
 #if defined(__arm__) || defined(__aarch64__)
 #ifdef HAVE_DWARF_UNWIND_SUPPORT
-- 
2.7.4

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

#1443112

FromIngo Molnar <mingo@kernel.org>
Date2016-07-14 09:00 +0200
Message-ID<rUIMO-64Y-17@gated-at.bofh.it>
In reply to#1443025
* Arnaldo Carvalho de Melo <acme@kernel.org> wrote:

> Hi Ingo,
> 
> 	Please consider pulling,
> 
> I've added building objtool to most of the containers in my build test setup:
> 
> [root@jouet ~]# perf stat dm
> alpine:3.4: Ok
> centos:5: Ok
> centos:6: Ok
> centos:7: Ok
> debian:7: Ok
> debian:8: Ok
> debian:experimental: Ok
> fedora:21: Ok
> fedora:22: Ok
> fedora:23: Ok
> fedora:24: Ok
> fedora:rawhide: Ok
> mageia:5: Ok
> opensuse:13.2: Ok
> opensuse:42.1: Ok
> ubuntu:12.04.5: Ok
> ubuntu:14.04.4: Ok
> ubuntu:15.10: Ok
> ubuntu:16.04: Ok
> 
>  Performance counter stats for 'dm':
> 
>        2601.121782      task-clock (msec)         #    0.002 CPUs utilized          
>             86,368      context-switches          #    0.033 M/sec                  
>              5,740      cpu-migrations            #    0.002 M/sec                  
>             53,962      page-faults               #    0.021 M/sec                  
>      7,217,605,183      cycles                    #    2.775 GHz                    
>      6,534,540,119      instructions              #    0.91  insn per cycle         
>      1,408,715,184      branches                  #  541.580 M/sec                  
>         18,523,459      branch-misses             #    1.31% of all branches        
> 
>     1541.746171526 seconds time elapsed
> 
> [root@jouet ~]# 
> 
> - Arnaldo
> 
> The following changes since commit 7b39cafb7aa68ef8e32a9f51fbe737d96084ca74:
> 
>   tools: Work around BITS_PER_LONG related build failure in objtool (2016-07-13 09:37:43 +0200)
> 
> are available in the git repository at:
> 
>   git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git tags/perf-core-for-mingo-20160713
> 
> for you to fetch changes up to 8e5dc848356ecf6ea8d27d641c4d7ad8d42fe92b:
> 
>   perf test: Add a test case for SDT event (2016-07-13 23:09:10 -0300)
> 
> ----------------------------------------------------------------
> perf/core improvements and fixes:
> 
> User visible:
> 
> - Finish merging initial SDT (Statically Defined Traces) support, see
>   cset comments for details about how it all works (Masami Hiramatsu)
> 
> - Support attaching eBPF programs to tracepoints (Wang Nan)
> 
> Infrastructure:
> 
> - Fix up BITS_PER_LONG setting (Arnaldo Carvalho de Melo)
> 
> - Add fallback from ELF_C_READ_MMAP to ELF_C_READ in objtool, fixing
>   the build in libelf implementations lacking that elf_begin() cmd,
>   such as Alpine Linux's (Arnaldo Carvalho de Melo)
> 
> - Avoid checking code drift on busybox's diff in objtool (Arnaldo Carvalho de Melo)
> 
> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
> 
> ----------------------------------------------------------------
> Arnaldo Carvalho de Melo (3):
>       tools: Fix up BITS_PER_LONG setting
>       objtool: Add fallback from ELF_C_READ_MMAP to ELF_C_READ
>       objtool: Avoid checking code drift on busybox's diff
> 
> Masami Hiramatsu (11):
>       perf probe: Fix to show correct error message for $vars and $params
>       perf probe: Accept %sdt and %cached event name
>       perf probe: Make --list show only available cached events
>       perf probe-cache: Add for_each_probe_cache_entry() wrapper
>       perf probe: Allow wildcard for cached events
>       perf probe: Search SDT/cached event from all probe caches
>       perf list: Show SDT and pre-cached events
>       perf probe: Support @BUILDID or @FILE suffix for SDT events
>       perf probe: Support a special SDT probe format
>       perf build: Add sdt feature detection
>       perf test: Add a test case for SDT event
> 
> Wang Nan (5):
>       tools lib bpf: New API to adjust type of a BPF program
>       tools lib bpf: Report error when kernel doesn't support program type
>       perf event parser: Add const qualifier to evt_name and sys_name
>       perf bpf: Rename bpf__foreach_tev() to bpf__foreach_event()
>       perf bpf: Support BPF program attach to tracepoints
> 
>  tools/build/Makefile.feature            |   3 +-
>  tools/build/feature/Makefile            |   6 +-
>  tools/build/feature/test-all.c          |   5 +
>  tools/build/feature/test-sdt.c          |   7 +
>  tools/include/asm-generic/bitsperlong.h |  24 ++-
>  tools/lib/bpf/libbpf.c                  |  80 +++++++--
>  tools/lib/bpf/libbpf.h                  |  10 ++
>  tools/objtool/Makefile                  |   5 +-
>  tools/objtool/elf.c                     |   7 +
>  tools/perf/Documentation/perf-probe.txt |  11 +-
>  tools/perf/Makefile.perf                |   3 +
>  tools/perf/builtin-list.c               |   6 +-
>  tools/perf/builtin-probe.c              |   2 +-
>  tools/perf/config/Makefile              |  10 ++
>  tools/perf/tests/Build                  |   1 +
>  tools/perf/tests/builtin-test.c         |   4 +
>  tools/perf/tests/make                   |   3 +-
>  tools/perf/tests/sdt.c                  | 115 ++++++++++++
>  tools/perf/tests/tests.h                |   1 +
>  tools/perf/util/bpf-loader.c            |  73 +++++++-
>  tools/perf/util/bpf-loader.h            |  12 +-
>  tools/perf/util/build-id.c              |  76 +++++++-
>  tools/perf/util/build-id.h              |   3 +-
>  tools/perf/util/parse-events.c          | 110 ++++++++++--
>  tools/perf/util/parse-events.h          |   4 +-
>  tools/perf/util/probe-event.c           | 309 +++++++++++++++++++++++++++-----
>  tools/perf/util/probe-event.h           |   1 +
>  tools/perf/util/probe-file.c            |  57 ++++--
>  tools/perf/util/probe-file.h            |  14 ++
>  29 files changed, 850 insertions(+), 112 deletions(-)
>  create mode 100644 tools/build/feature/test-sdt.c
>  create mode 100644 tools/perf/tests/sdt.c

Pulled, thanks a lot Arnaldo!

	Ingo

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web