Merge branch 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/l…

…inux/kernel/git/tip/linux-2.6-tip

* 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  perf, amd: Use kmalloc_node(,__GFP_ZERO) for northbridge structure allocation
  perf_events: Fix time tracking in samples
  perf trace: update usage
  perf trace: update Documentation with new perf trace variants
  perf trace: live-mode command-line cleanup
  perf trace record: handle commands correctly
  perf record: make the record options available outside perf record
  perf trace scripting: remove system-wide param from shell scripts
  perf trace scripting: fix some small memory leaks and missing error checks
  perf: Fix usages of profile_cpu in builtin-top.c to use cpu_list
  perf, ui: Eliminate stack-smashing protection compiler complaint

arch/x86/kernel/cpu/perf_event_amd.c

@@ -280,11 +280,11 @@ static struct amd_nb *amd_alloc_nb(int cpu, int nb_id)
 	struct amd_nb *nb;
 	int i;
 
-	nb = kmalloc(sizeof(struct amd_nb), GFP_KERNEL);
+	nb = kmalloc_node(sizeof(struct amd_nb), GFP_KERNEL | __GFP_ZERO,
+			  cpu_to_node(cpu));
 	if (!nb)
 		return NULL;
 
-	memset(nb, 0, sizeof(*nb));
 	nb->nb_id = nb_id;
 
 	/*

include/linux/perf_event.h

@@ -747,6 +747,16 @@ struct perf_event {
 	u64				tstamp_running;
 	u64				tstamp_stopped;
 
+	/*
+	 * timestamp shadows the actual context timing but it can
+	 * be safely used in NMI interrupt context. It reflects the
+	 * context time as it was when the event was last scheduled in.
+	 *
+	 * ctx_time already accounts for ctx->timestamp. Therefore to
+	 * compute ctx_time for a sample, simply add perf_clock().
+	 */
+	u64				shadow_ctx_time;
+
 	struct perf_event_attr		attr;
 	struct hw_perf_event		hw;
 

kernel/perf_event.c

@@ -674,6 +674,8 @@ event_sched_in(struct perf_event *event,
 
 	event->tstamp_running += ctx->time - event->tstamp_stopped;
 
+	event->shadow_ctx_time = ctx->time - ctx->timestamp;
+
 	if (!is_software_event(event))
 		cpuctx->active_oncpu++;
 	ctx->nr_active++;
@@ -3396,19 +3398,20 @@ static u32 perf_event_tid(struct perf_event *event, struct task_struct *p)
 }
 
 static void perf_output_read_one(struct perf_output_handle *handle,
-				 struct perf_event *event)
+				 struct perf_event *event,
+				 u64 enabled, u64 running)
 {
 	u64 read_format = event->attr.read_format;
 	u64 values[4];
 	int n = 0;
 
 	values[n++] = perf_event_count(event);
 	if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED) {
-		values[n++] = event->total_time_enabled +
+		values[n++] = enabled +
 			atomic64_read(&event->child_total_time_enabled);
 	}
 	if (read_format & PERF_FORMAT_TOTAL_TIME_RUNNING) {
-		values[n++] = event->total_time_running +
+		values[n++] = running +
 			atomic64_read(&event->child_total_time_running);
 	}
 	if (read_format & PERF_FORMAT_ID)
@@ -3421,7 +3424,8 @@ static void perf_output_read_one(struct perf_output_handle *handle,
  * XXX PERF_FORMAT_GROUP vs inherited events seems difficult.
  */
 static void perf_output_read_group(struct perf_output_handle *handle,
-			    struct perf_event *event)
+			    struct perf_event *event,
+			    u64 enabled, u64 running)
 {
 	struct perf_event *leader = event->group_leader, *sub;
 	u64 read_format = event->attr.read_format;
@@ -3431,10 +3435,10 @@ static void perf_output_read_group(struct perf_output_handle *handle,
 	values[n++] = 1 + leader->nr_siblings;
 
 	if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED)
-		values[n++] = leader->total_time_enabled;
+		values[n++] = enabled;
 
 	if (read_format & PERF_FORMAT_TOTAL_TIME_RUNNING)
-		values[n++] = leader->total_time_running;
+		values[n++] = running;
 
 	if (leader != event)
 		leader->pmu->read(leader);
@@ -3459,13 +3463,35 @@ static void perf_output_read_group(struct perf_output_handle *handle,
 	}
 }
 
+#define PERF_FORMAT_TOTAL_TIMES (PERF_FORMAT_TOTAL_TIME_ENABLED|\
+				 PERF_FORMAT_TOTAL_TIME_RUNNING)
+
 static void perf_output_read(struct perf_output_handle *handle,
 			     struct perf_event *event)
 {
+	u64 enabled = 0, running = 0, now, ctx_time;
+	u64 read_format = event->attr.read_format;
+
+	/*
+	 * compute total_time_enabled, total_time_running
+	 * based on snapshot values taken when the event
+	 * was last scheduled in.
+	 *
+	 * we cannot simply called update_context_time()
+	 * because of locking issue as we are called in
+	 * NMI context
+	 */
+	if (read_format & PERF_FORMAT_TOTAL_TIMES) {
+		now = perf_clock();
+		ctx_time = event->shadow_ctx_time + now;
+		enabled = ctx_time - event->tstamp_enabled;
+		running = ctx_time - event->tstamp_running;
+	}
+
 	if (event->attr.read_format & PERF_FORMAT_GROUP)
-		perf_output_read_group(handle, event);
+		perf_output_read_group(handle, event, enabled, running);
 	else
-		perf_output_read_one(handle, event);
+		perf_output_read_one(handle, event, enabled, running);
 }
 
 void perf_output_sample(struct perf_output_handle *handle,

tools/perf/Documentation/perf-trace.txt

@@ -8,7 +8,11 @@ perf-trace - Read perf.data (created by perf record) and display trace output
 SYNOPSIS
 --------
 [verse]
-'perf trace' {record <script> | report <script> [args] }
+'perf trace' [<options>]
+'perf trace' [<options>] record <script> [<record-options>] <command>
+'perf trace' [<options>] report <script> [script-args]
+'perf trace' [<options>] <script> <required-script-args> [<record-options>] <command>
+'perf trace' [<options>] <top-script> [script-args]
 
 DESCRIPTION
 -----------
@@ -24,23 +28,53 @@ There are several variants of perf trace:
   available via 'perf trace -l').  The following variants allow you to
   record and run those scripts:
 
-  'perf trace record <script>' to record the events required for 'perf
-  trace report'.  <script> is the name displayed in the output of
-  'perf trace --list' i.e. the actual script name minus any language
-  extension.
+  'perf trace record <script> <command>' to record the events required
+  for 'perf trace report'.  <script> is the name displayed in the
+  output of 'perf trace --list' i.e. the actual script name minus any
+  language extension.  If <command> is not specified, the events are
+  recorded using the -a (system-wide) 'perf record' option.
 
-  'perf trace report <script>' to run and display the results of
-  <script>.  <script> is the name displayed in the output of 'perf
+  'perf trace report <script> [args]' to run and display the results
+  of <script>.  <script> is the name displayed in the output of 'perf
   trace --list' i.e. the actual script name minus any language
   extension.  The perf.data output from a previous run of 'perf trace
   record <script>' is used and should be present for this command to
-  succeed.
+  succeed.  [args] refers to the (mainly optional) args expected by
+  the script.
+
+  'perf trace <script> <required-script-args> <command>' to both
+  record the events required for <script> and to run the <script>
+  using 'live-mode' i.e. without writing anything to disk.  <script>
+  is the name displayed in the output of 'perf trace --list' i.e. the
+  actual script name minus any language extension.  If <command> is
+  not specified, the events are recorded using the -a (system-wide)
+  'perf record' option.  If <script> has any required args, they
+  should be specified before <command>.  This mode doesn't allow for
+  optional script args to be specified; if optional script args are
+  desired, they can be specified using separate 'perf trace record'
+  and 'perf trace report' commands, with the stdout of the record step
+  piped to the stdin of the report script, using the '-o -' and '-i -'
+  options of the corresponding commands.
+
+  'perf trace <top-script>' to both record the events required for
+  <top-script> and to run the <top-script> using 'live-mode'
+  i.e. without writing anything to disk.  <top-script> is the name
+  displayed in the output of 'perf trace --list' i.e. the actual
+  script name minus any language extension; a <top-script> is defined
+  as any script name ending with the string 'top'.
+
+  [<record-options>] can be passed to the record steps of 'perf trace
+  record' and 'live-mode' variants; this isn't possible however for
+  <top-script> 'live-mode' or 'perf trace report' variants.
 
   See the 'SEE ALSO' section for links to language-specific
   information on how to write and run your own trace scripts.
 
 OPTIONS
 -------
+<command>...::
+	Any command you can specify in a shell.
+
 -D::
 --dump-raw-trace=::
         Display verbose dump of the trace data.
@@ -64,6 +98,13 @@ OPTIONS
         Generate perf-trace.[ext] starter script for given language,
         using current perf.data.
 
+-a::
+        Force system-wide collection.  Scripts run without a <command>
+        normally use -a by default, while scripts run with a <command>
+        normally don't - this option allows the latter to be run in
+        system-wide mode.
+
+
 SEE ALSO
 --------
 linkperf:perf-record[1], linkperf:perf-trace-perl[1],

tools/perf/builtin-record.c

@@ -790,7 +790,7 @@ static const char * const record_usage[] = {
 
 static bool force, append_file;
 
-static const struct option options[] = {
+const struct option record_options[] = {
 	OPT_CALLBACK('e', "event", NULL, "event",
 		     "event selector. use 'perf list' to list available events",
 		     parse_events),
@@ -839,16 +839,16 @@ int cmd_record(int argc, const char **argv, const char *prefix __used)
 {
 	int i, j, err = -ENOMEM;
 
-	argc = parse_options(argc, argv, options, record_usage,
+	argc = parse_options(argc, argv, record_options, record_usage,
 			    PARSE_OPT_STOP_AT_NON_OPTION);
 	if (!argc && target_pid == -1 && target_tid == -1 &&
 		!system_wide && !cpu_list)
-		usage_with_options(record_usage, options);
+		usage_with_options(record_usage, record_options);
 
 	if (force && append_file) {
 		fprintf(stderr, "Can't overwrite and append at the same time."
 				" You need to choose between -f and -A");
-		usage_with_options(record_usage, options);
+		usage_with_options(record_usage, record_options);
 	} else if (append_file) {
 		write_mode = WRITE_APPEND;
 	} else {
@@ -871,7 +871,7 @@ int cmd_record(int argc, const char **argv, const char *prefix __used)
 		if (thread_num <= 0) {
 			fprintf(stderr, "Can't find all threads of pid %d\n",
 					target_pid);
-			usage_with_options(record_usage, options);
+			usage_with_options(record_usage, record_options);
 		}
 	} else {
 		all_tids=malloc(sizeof(pid_t));

tools/perf/builtin-top.c

@@ -69,7 +69,6 @@ static int			target_tid			=     -1;
 static pid_t			*all_tids			=      NULL;
 static int			thread_num			=      0;
 static bool			inherit				=  false;
-static int			profile_cpu			=     -1;
 static int			nr_cpus				=      0;
 static int			realtime_prio			=      0;
 static bool			group				=  false;
@@ -558,13 +557,13 @@ static void print_sym_table(void)
 	else
 		printf(" (all");
 
-	if (profile_cpu != -1)
-		printf(", cpu: %d)\n", profile_cpu);
+	if (cpu_list)
+		printf(", CPU%s: %s)\n", nr_cpus > 1 ? "s" : "", cpu_list);
 	else {
 		if (target_tid != -1)
 			printf(")\n");
 		else
-			printf(", %d CPUs)\n", nr_cpus);
+			printf(", %d CPU%s)\n", nr_cpus, nr_cpus > 1 ? "s" : "");
 	}
 
 	printf("%-*.*s\n", win_width, win_width, graph_dotted_line);
@@ -1187,11 +1186,10 @@ int group_fd;
 static void start_counter(int i, int counter)
 {
 	struct perf_event_attr *attr;
-	int cpu;
+	int cpu = -1;
 	int thread_index;
 
-	cpu = profile_cpu;
-	if (target_tid == -1 && profile_cpu == -1)
+	if (target_tid == -1)
 		cpu = cpumap[i];
 
 	attr = attrs + counter;