Skip to contentSkip to conversation

tracing: Fixes for 5.5

PATCH In review Steven Rostedt 4 participants 11 messages Jan 21, 2020 – Jan 22, 2020 1 message not in the archive

Subsystems (2)
lore
Share & tools
lore

Steven Rostedt

author maintainer
#0 lore raw

cover tracing: Fixes for 5.5

Changbin Du (1):
tracing: xen: Ordered comparison of function pointers

Masami Hiramatsu (2):
tracing/uprobe: Fix double perf_event linking on multiprobe uprobe
tracing: trigger: Replace unneeded RCU-list traversals

Masami Ichikawa (1):
tracing: Do not set trace clock if tracefs lockdown is in effect

Steven Rostedt (VMware) (1):
tracing: Fix histogram code when expression has same var as value

----

 include/trace/events/xen.h          |   6 +-
 kernel/trace/trace.c                |   5 ++
 kernel/trace/trace_events_hist.c    |  63 ++++++++++++++----
 kernel/trace/trace_events_trigger.c |  20 ++++--
 kernel/trace/trace_kprobe.c         |   2 +-
 kernel/trace/trace_probe.c          |   5 +-
 kernel/trace/trace_probe.h          |   3 +-
 kernel/trace/trace_uprobe.c         | 124 +++++++++++++++++++++++-------------
 8 files changed, 163 insertions(+), 65 deletions(-)
author maintainer
#1 lore raw

PATCH 1/5 tracing: xen: Ordered comparison of function pointers

From: Changbin Du <[email protected]>

Just as commit 0566e40ce7 ("tracing: initcall: Ordered comparison of function pointers"), this patch fixes another remaining one in xen.h found by clang-9.

In file included from arch/x86/xen/trace.c:21:
In file included from ./include/trace/events/xen.h:475:
In file included from ./include/trace/define_trace.h:102:
In file included from ./include/trace/trace_events.h:473:
./include/trace/events/xen.h:69:7: warning: ordered comparison of function \
pointers ('xen_mc_callback_fn_t' (aka 'void (*)(void *)') and 'xen_mc_callback_fn_t') [-Wordered-compare-function-pointers]

                    __field(xen_mc_callback_fn_t, fn)
                    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
./include/trace/trace_events.h:421:29: note: expanded from macro '__field'
                                ^
./include/trace/trace_events.h:407:6: note: expanded from macro '__field_ext'
                                 is_signed_type(type), filter_type);    \
                                 ^
./include/linux/trace_events.h:554:44: note: expanded from macro 'is_signed_type'
                                              ^
Fixes c796f213a6934 ("xen/trace: add multicall tracing")
---
 include/trace/events/xen.h | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
include/trace/events/xen.h +5 -1
index 9a0e8af21310..a5ccfa67bc5c 100644
@@ -66,7 +66,11 @@ TRACE_EVENT(xen_mc_callback,
 	    TP_PROTO(xen_mc_callback_fn_t fn, void *data),
 	    TP_ARGS(fn, data),
 	    TP_STRUCT__entry(
-		    __field(xen_mc_callback_fn_t, fn)
+		    /*
+		     * Use field_struct to avoid is_signed_type()
+		     * comparison of a function pointer.
+		     */
+		    __field_struct(xen_mc_callback_fn_t, fn)
 		    __field(void *, data)
 		    ),
 	    TP_fast_assign(
Signature · 1 line
2.24.1
author maintainer
#2 lore raw

PATCH 2/5 tracing/uprobe: Fix double perf_event linking on multiprobe uprobe

In reply to Steven Rostedt

From: Masami Hiramatsu <[email protected]>

Fix double perf_event linking to trace_uprobe_filter on multiple uprobe event by moving trace_uprobe_filter under trace_probe_event.

In uprobe perf event, trace_uprobe_filter data structure is managing target mm filters (in perf_event) related to each uprobe event.

Since commit 60d53e2c3b75 ("tracing/probe: Split trace_event related data from trace_probe") left the trace_uprobe_filter data structure in trace_uprobe, if a trace_probe_event has multiple trace_uprobe (multi-probe event), a perf_event is added to different trace_uprobe_filter on each trace_uprobe. This leads a linked list corruption.

To fix this issue, move trace_uprobe_filter to trace_probe_event and link it once on each event instead of each probe.

Cc Jiri Olsa <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc "Naveen N . Rao" <[email protected]>
Cc Anil S Keshavamurthy <[email protected]>
Cc "David S . Miller" <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc =?utf-8?q?Toke_H=C3=B8iland-J?= =?utf-8?b?w7hyZ2Vuc2Vu?= <[email protected]>
Cc Jean-Tsung Hsiao <[email protected]>
Cc Jesper Dangaard Brouer <[email protected]>
Fixes 60d53e2c3b75 ("tracing/probe: Split trace_event related data from trace_probe")
---
 kernel/trace/trace_kprobe.c |   2 +-
 kernel/trace/trace_probe.c  |   5 +-
 kernel/trace/trace_probe.h  |   3 +-
 kernel/trace/trace_uprobe.c | 124 +++++++++++++++++++++++-------------
 4 files changed, 86 insertions(+), 48 deletions(-)
kernel/trace/trace_kprobe.c +1 -1
index 7f890262c8a3..3e5f9c7d939c 100644
@@ -290,7 +290,7 @@ static struct trace_kprobe *alloc_trace_kprobe(const char *group,
 	INIT_HLIST_NODE(&tk->rp.kp.hlist);
 	INIT_LIST_HEAD(&tk->rp.kp.list);
 
-	ret = trace_probe_init(&tk->tp, event, group);
+	ret = trace_probe_init(&tk->tp, event, group, 0);
 	if (ret < 0)
 		goto error;
 
kernel/trace/trace_probe.c +3 -2
index 905b10af5d5c..bba18cf44a30 100644
@@ -984,7 +984,7 @@ void trace_probe_cleanup(struct trace_probe *tp)
 }
 
 int trace_probe_init(struct trace_probe *tp, const char *event,
-		     const char *group)
+		     const char *group, size_t event_data_size)
 {
 	struct trace_event_call *call;
 	int ret = 0;
@@ -992,7 +992,8 @@ int trace_probe_init(struct trace_probe *tp, const char *event,
 	if (!event || !group)
 		return -EINVAL;
 
-	tp->event = kzalloc(sizeof(struct trace_probe_event), GFP_KERNEL);
+	tp->event = kzalloc(sizeof(struct trace_probe_event) + event_data_size,
+			    GFP_KERNEL);
 	if (!tp->event)
 		return -ENOMEM;
 
kernel/trace/trace_probe.h +2 -1
index 4ee703728aec..03e4e180058d 100644
@@ -230,6 +230,7 @@ struct trace_probe_event {
 	struct trace_event_call		call;
 	struct list_head 		files;
 	struct list_head		probes;
+	char				data[0];
 };
 
 struct trace_probe {
@@ -322,7 +323,7 @@ static inline bool trace_probe_has_single_file(struct trace_probe *tp)
 }
 
 int trace_probe_init(struct trace_probe *tp, const char *event,
-		     const char *group);
+		     const char *group, size_t event_data_size);
 void trace_probe_cleanup(struct trace_probe *tp);
 int trace_probe_append(struct trace_probe *tp, struct trace_probe *to);
 void trace_probe_unlink(struct trace_probe *tp);
kernel/trace/trace_uprobe.c +80 -44

229 lines not on this page · raw message

Signature · 1 line
2.24.1
#3 lore raw

Re: [for-linus][PATCH 2/5] tracing/uprobe: Fix double perf_event linking on multiprobe uprobe

kernel/trace/trace_probe.h struct trace_probe_event { · line 234

Quoted 10 lines · Steven Rostedt
kernel/trace/trace_probe.h +1 -0
index 4ee703728aec..03e4e180058d 100644
@@ -230,6 +230,7 @@ struct trace_probe_event {
 	struct trace_event_call		call;
 	struct list_head 		files;
 	struct list_head		probes;
+	char				data[0];
 };

Note that this relies on pure 'luck'. If you stick anything <4 bytes in between the list_head and the data member it'll come unstuck real fast.

kernel/trace/trace_probe.h struct trace_probe_event {

Quoted 7 lines
kernel/trace/trace_probe.h +7 -0
+static struct trace_uprobe_filter *
+trace_uprobe_get_filter(struct trace_uprobe *tu)
+{
+	struct trace_probe_event *event = tu->tp.event;
+
+	return (struct trace_uprobe_filter *)&event->data[0];
+}

That is, the above does not consider the alignment requirements of struct trace_uprobe_filter.

#4 lore raw

RE: [for-linus][PATCH 2/5] tracing/uprobe: Fix double perf_event linking on multiprobe uprobe

From: Peter Zijlstra

Quoted 15 lines

Sent: 21 January 2020 14:50

kernel/trace/trace_probe.h struct trace_probe_event { · line 234

Quoted 10 lines · Steven Rostedt
kernel/trace/trace_probe.h +1 -0
index 4ee703728aec..03e4e180058d 100644
@@ -230,6 +230,7 @@ struct trace_probe_event {
 	struct trace_event_call		call;
 	struct list_head 		files;
 	struct list_head		probes;
+	char				data[0];
 };

Note that this relies on pure 'luck'. If you stick anything <4 bytes in between the list_head and the data member it'll come unstuck real fast.

Can you fix it by adding an unnamed struct as in:

struct trace_probe_event {
    	struct {
		struct trace_event_call		call;
		struct list_head 		files;
		struct list_head		probes;
	};
	char				data[0];
};

David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)

#5 lore raw

Re: [for-linus][PATCH 2/5] tracing/uprobe: Fix double perf_event linking on multiprobe uprobe

Quoted 17 lines · David Laight

From: Peter Zijlstra

Quoted 15 lines

Sent: 21 January 2020 14:50

kernel/trace/trace_probe.h struct trace_probe_event { · line 234

Quoted 10 lines · Steven Rostedt
kernel/trace/trace_probe.h +1 -0
index 4ee703728aec..03e4e180058d 100644
@@ -230,6 +230,7 @@ struct trace_probe_event {
 	struct trace_event_call		call;
 	struct list_head 		files;
 	struct list_head		probes;
+	char				data[0];
 };

Note that this relies on pure 'luck'. If you stick anything <4 bytes in between the list_head and the data member it'll come unstuck real fast.

Quoted 1 line

Can you fix it by adding an unnamed struct as in:

The trivial fix is like I suggested in the other thread:

	struct trace_uprobe_filter filters[0];

The alternative that Masami-San suggested should also work.

maintainer
#6 lore raw

Re: [for-linus][PATCH 2/5] tracing/uprobe: Fix double perf_event linking on multiprobe uprobe

Quoted 18 lines · On Tue, 21 Jan 2020 16:06:05 +0100 Peter Zijlstra <[email protected]> wrote:
Quoted 17 lines · David Laight

From: Peter Zijlstra

Quoted 15 lines

Sent: 21 January 2020 14:50

kernel/trace/trace_probe.h struct trace_probe_event { · line 234

Quoted 10 lines · Steven Rostedt
kernel/trace/trace_probe.h +1 -0
index 4ee703728aec..03e4e180058d 100644
@@ -230,6 +230,7 @@ struct trace_probe_event {
 	struct trace_event_call		call;
 	struct list_head 		files;
 	struct list_head		probes;
+	char				data[0];
 };

Note that this relies on pure 'luck'. If you stick anything <4 bytes in between the list_head and the data member it'll come unstuck real fast.

Quoted 7 lines
Quoted 1 line

Can you fix it by adding an unnamed struct as in:

The trivial fix is like I suggested in the other thread:

	struct trace_uprobe_filter filters[0];

The alternative that Masami-San suggested should also work.

I've sent a fix ( https://lkml.org/lkml/2020/1/21/1214 ) as Peter suggested, change the field to

	struct trace_uprobe_filter filters[0];

and change the code around that.

Thank you,

Signature · 1 line
Masami Hiramatsu <[email protected]>
author maintainer
#7 lore raw

PATCH 3/5 tracing: trigger: Replace unneeded RCU-list traversals

In reply to Steven Rostedt

From: Masami Hiramatsu <[email protected]>

With CONFIG_PROVE_RCU_LIST, I had many suspicious RCU warnings when I ran ftracetest trigger testcases.

-----

  # dmesg -c > /dev/null
  # ./ftracetest test.d/trigger
  ...
  # dmesg | grep "RCU-list traversed" | cut -f 2 -d ] | cut -f 2 -d " "
  kernel/trace/trace_events_hist.c:6070
  kernel/trace/trace_events_hist.c:1760
  kernel/trace/trace_events_hist.c:5911
  kernel/trace/trace_events_trigger.c:504
  kernel/trace/trace_events_hist.c:1810
  kernel/trace/trace_events_hist.c:3158
  kernel/trace/trace_events_hist.c:3105
  kernel/trace/trace_events_hist.c:5518
  kernel/trace/trace_events_hist.c:5998
  kernel/trace/trace_events_hist.c:6019
  kernel/trace/trace_events_hist.c:6044
  kernel/trace/trace_events_trigger.c:1500
  kernel/trace/trace_events_trigger.c:1540
  kernel/trace/trace_events_trigger.c:539
  kernel/trace/trace_events_trigger.c:584
-----

I investigated those warnings and found that the RCU-list traversals in event trigger and hist didn't need to use RCU version because those were called only under event_mutex.

I also checked other RCU-list traversals related to event trigger list, and found that most of them were called from event_hist_trigger_func() or hist_unregister_trigger() or register/unregister functions except for a few cases.

Replace these unneeded RCU-list traversals with normal list traversal macro and lockdep_assert_held() to check the event_mutex is held.

Fixes 30350d65ac567 ("tracing: Add variable support to hist triggers")
---
 kernel/trace/trace_events_hist.c    | 41 +++++++++++++++++++++--------
 kernel/trace/trace_events_trigger.c | 20 ++++++++++----
 2 files changed, 45 insertions(+), 16 deletions(-)
kernel/trace/trace_events_hist.c +30 -11
index f62de5f43e79..d33b046f985a 100644
@@ -1766,11 +1766,13 @@ static struct hist_field *find_var(struct hist_trigger_data *hist_data,
 	struct event_trigger_data *test;
 	struct hist_field *hist_field;
 
+	lockdep_assert_held(&event_mutex);
+
 	hist_field = find_var_field(hist_data, var_name);
 	if (hist_field)
 		return hist_field;
 
-	list_for_each_entry_rcu(test, &file->triggers, list) {
+	list_for_each_entry(test, &file->triggers, list) {
 		if (test->cmd_ops->trigger_type == ETT_EVENT_HIST) {
 			test_data = test->private_data;
 			hist_field = find_var_field(test_data, var_name);
@@ -1820,7 +1822,9 @@ static struct hist_field *find_file_var(struct trace_event_file *file,
 	struct event_trigger_data *test;
 	struct hist_field *hist_field;
 
-	list_for_each_entry_rcu(test, &file->triggers, list) {
+	lockdep_assert_held(&event_mutex);
+
+	list_for_each_entry(test, &file->triggers, list) {
 		if (test->cmd_ops->trigger_type == ETT_EVENT_HIST) {
 			test_data = test->private_data;
 			hist_field = find_var_field(test_data, var_name);
@@ -3115,7 +3119,9 @@ static char *find_trigger_filter(struct hist_trigger_data *hist_data,
 {
 	struct event_trigger_data *test;
 
-	list_for_each_entry_rcu(test, &file->triggers, list) {
+	lockdep_assert_held(&event_mutex);
+
+	list_for_each_entry(test, &file->triggers, list) {
 		if (test->cmd_ops->trigger_type == ETT_EVENT_HIST) {
 			if (test->private_data == hist_data)
 				return test->filter_str;
@@ -3166,9 +3172,11 @@ find_compatible_hist(struct hist_trigger_data *target_hist_data,
 	struct event_trigger_data *test;
 	unsigned int n_keys;
 
+	lockdep_assert_held(&event_mutex);
+
 	n_keys = target_hist_data->n_fields - target_hist_data->n_vals;
 
-	list_for_each_entry_rcu(test, &file->triggers, list) {
+	list_for_each_entry(test, &file->triggers, list) {
 		if (test->cmd_ops->trigger_type == ETT_EVENT_HIST) {
 			hist_data = test->private_data;
 
@@ -5528,7 +5536,7 @@ static int hist_show(struct seq_file *m, void *v)
 		goto out_unlock;
 	}
 
-	list_for_each_entry_rcu(data, &event_file->triggers, list) {
+	list_for_each_entry(data, &event_file->triggers, list) {
 		if (data->cmd_ops->trigger_type == ETT_EVENT_HIST)
 			hist_trigger_show(m, data, n++);
 	}
@@ -5921,7 +5929,9 @@ static int hist_register_trigger(char *glob, struct event_trigger_ops *ops,
 	if (hist_data->attrs->name && !named_data)
 		goto new;
 
-	list_for_each_entry_rcu(test, &file->triggers, list) {
+	lockdep_assert_held(&event_mutex);
+
+	list_for_each_entry(test, &file->triggers, list) {
 		if (test->cmd_ops->trigger_type == ETT_EVENT_HIST) {
 			if (!hist_trigger_match(data, test, named_data, false))
 				continue;
@@ -6005,10 +6015,12 @@ static bool have_hist_trigger_match(struct event_trigger_data *data,
 	struct event_trigger_data *test, *named_data = NULL;
 	bool match = false;
 
+	lockdep_assert_held(&event_mutex);
+
 	if (hist_data->attrs->name)
 		named_data = find_named_trigger(hist_data->attrs->name);
 
-	list_for_each_entry_rcu(test, &file->triggers, list) {
+	list_for_each_entry(test, &file->triggers, list) {
 		if (test->cmd_ops->trigger_type == ETT_EVENT_HIST) {
 			if (hist_trigger_match(data, test, named_data, false)) {
 				match = true;
@@ -6026,10 +6038,12 @@ static bool hist_trigger_check_refs(struct event_trigger_data *data,
 	struct hist_trigger_data *hist_data = data->private_data;
 	struct event_trigger_data *test, *named_data = NULL;
 
+	lockdep_assert_held(&event_mutex);
+
 	if (hist_data->attrs->name)
 		named_data = find_named_trigger(hist_data->attrs->name);
 
-	list_for_each_entry_rcu(test, &file->triggers, list) {
+	list_for_each_entry(test, &file->triggers, list) {
 		if (test->cmd_ops->trigger_type == ETT_EVENT_HIST) {
 			if (!hist_trigger_match(data, test, named_data, false))
 				continue;
@@ -6051,10 +6065,12 @@ static void hist_unregister_trigger(char *glob, struct event_trigger_ops *ops,
 	struct event_trigger_data *test, *named_data = NULL;
 	bool unregistered = false;
 
+	lockdep_assert_held(&event_mutex);
+
 	if (hist_data->attrs->name)
 		named_data = find_named_trigger(hist_data->attrs->name);
 
-	list_for_each_entry_rcu(test, &file->triggers, list) {
+	list_for_each_entry(test, &file->triggers, list) {
 		if (test->cmd_ops->trigger_type == ETT_EVENT_HIST) {
 			if (!hist_trigger_match(data, test, named_data, false))
 				continue;
@@ -6080,7 +6096,9 @@ static bool hist_file_check_refs(struct trace_event_file *file)
 	struct hist_trigger_data *hist_data;
 	struct event_trigger_data *test;
 
-	list_for_each_entry_rcu(test, &file->triggers, list) {
+	lockdep_assert_held(&event_mutex);
+
+	list_for_each_entry(test, &file->triggers, list) {
 		if (test->cmd_ops->trigger_type == ETT_EVENT_HIST) {
 			hist_data = test->private_data;
 			if (check_var_refs(hist_data))
@@ -6323,7 +6341,8 @@ hist_enable_trigger(struct event_trigger_data *data, void *rec,
 	struct enable_trigger_data *enable_data = data->private_data;
 	struct event_trigger_data *test;
 
-	list_for_each_entry_rcu(test, &enable_data->file->triggers, list) {
+	list_for_each_entry_rcu(test, &enable_data->file->triggers, list,
+				lockdep_is_held(&event_mutex)) {
 		if (test->cmd_ops->trigger_type == ETT_EVENT_HIST) {
 			if (enable_data->enable)
 				test->paused = false;
kernel/trace/trace_events_trigger.c +15 -5
index 2cd53ca21b51..40106fff06a4 100644
@@ -501,7 +501,9 @@ void update_cond_flag(struct trace_event_file *file)
 	struct event_trigger_data *data;
 	bool set_cond = false;
 
-	list_for_each_entry_rcu(data, &file->triggers, list) {
+	lockdep_assert_held(&event_mutex);
+
+	list_for_each_entry(data, &file->triggers, list) {
 		if (data->filter || event_command_post_trigger(data->cmd_ops) ||
 		    event_command_needs_rec(data->cmd_ops)) {
 			set_cond = true;
@@ -536,7 +538,9 @@ static int register_trigger(char *glob, struct event_trigger_ops *ops,
 	struct event_trigger_data *test;
 	int ret = 0;
 
-	list_for_each_entry_rcu(test, &file->triggers, list) {
+	lockdep_assert_held(&event_mutex);
+
+	list_for_each_entry(test, &file->triggers, list) {
 		if (test->cmd_ops->trigger_type == data->cmd_ops->trigger_type) {
 			ret = -EEXIST;
 			goto out;
@@ -581,7 +585,9 @@ static void unregister_trigger(char *glob, struct event_trigger_ops *ops,
 	struct event_trigger_data *data;
 	bool unregistered = false;
 
-	list_for_each_entry_rcu(data, &file->triggers, list) {
+	lockdep_assert_held(&event_mutex);
+
+	list_for_each_entry(data, &file->triggers, list) {
 		if (data->cmd_ops->trigger_type == test->cmd_ops->trigger_type) {
 			unregistered = true;
 			list_del_rcu(&data->list);
@@ -1497,7 +1503,9 @@ int event_enable_register_trigger(char *glob,
 	struct event_trigger_data *test;
 	int ret = 0;
 
-	list_for_each_entry_rcu(test, &file->triggers, list) {
+	lockdep_assert_held(&event_mutex);
+
+	list_for_each_entry(test, &file->triggers, list) {
 		test_enable_data = test->private_data;
 		if (test_enable_data &&
 		    (test->cmd_ops->trigger_type ==
@@ -1537,7 +1545,9 @@ void event_enable_unregister_trigger(char *glob,
 	struct event_trigger_data *data;
 	bool unregistered = false;
 
-	list_for_each_entry_rcu(data, &file->triggers, list) {
+	lockdep_assert_held(&event_mutex);
+
+	list_for_each_entry(data, &file->triggers, list) {
 		enable_data = data->private_data;
 		if (enable_data &&
 		    (data->cmd_ops->trigger_type ==
Signature · 1 line
2.24.1

A message that is not in the archive (sent off-list or not yet received): [email protected]

maintainer
#9 lore raw

Re: [for-linus][PATCH 3/5] tracing: trigger: Replace unneeded RCU-list traversals

Quoted 7 lines · On Wed, 22 Jan 2020 02:26:22 +0000 Sasha Levin <[email protected]> wrote:

Hi,

[This is an automated email]

This commit has been processed because it contains a "Fixes:" tag, fixing commit: 30350d65ac56 ("tracing: Add variable support to hist triggers").

Quoted 7 lines

The bot has tested the following trees: v5.4.13, v4.19.97.

v5.4.13: Build OK!
v4.19.97: Build failed! Errors:
kernel/trace/trace_events_hist.c:5725:34: error: macro "list_for_each_entry_rcu" passed 4 arguments, but takes just 3 kernel/trace/trace_events_hist.c:5724:2: error: ‘list_for_each_entry_rcu’ undeclared (first use in this function) kernel/trace/trace_events_hist.c:5724:25: error: expected ‘;’ before ‘{’ token

Ah, that depends on commit 28875945ba98 ("rcu: Add support for consolidated-RCU reader checking"). We need to remove lockdep_is_held() from that list_for_each_entry_rcu().
(I've attached it)

Thank you,

Signature · 1 line
Masami Hiramatsu <[email protected]>
author maintainer
#10 lore raw

PATCH 4/5 tracing: Fix histogram code when expression has same var as value

In reply to Steven Rostedt

From: "Steven Rostedt (VMware)" <[email protected]>

While working on a tool to convert SQL syntex into the histogram language of the kernel, I discovered the following bug:

 # echo 'first u64 start_time u64 end_time pid_t pid u64 delta' >> synthetic_events
 # echo 'hist:keys=pid:start=common_timestamp' > events/sched/sched_waking/trigger
 # echo 'hist:keys=next_pid:delta=common_timestamp-$start,start2=$start:onmatch(sched.sched_waking).trace(first,$start2,common_timestamp,next_pid,$delta)' > events/sched/sched_switch/trigger

Would not display any histograms in the sched_switch histogram side.

But if I were to swap the location of

"delta=common_timestamp-$start" with "start2=$start"

Such that the last line had:

 # echo 'hist:keys=next_pid:start2=$start,delta=common_timestamp-$start:onmatch(sched.sched_waking).trace(first,$start2,common_timestamp,next_pid,$delta)' > events/sched/sched_switch/trigger

The histogram works as expected.

What I found out is that the expressions clear out the value once it is resolved. As the variables are resolved in the order listed, when processing:

delta=common_timestamp-$start

The $start is cleared. When it gets to "start2=$start", it errors out with "unresolved symbol" (which is silent as this happens at the location of the trace), and the histogram is dropped.

When processing the histogram for variable references, instead of adding a new reference for a variable used twice, use the same reference. That way, not only is it more efficient, but the order will no longer matter in processing of the variables.

From Tom Zanussi:

"Just to clarify some more about what the problem was is that without
your patch, we would have two separate references to the same variable, and during resolve_var_refs(), they'd both want to be resolved separately, so in this case, since the first reference to start wasn't part of an expression, it wouldn't get the read-once flag set, so would be read normally, and then the second reference would do the read-once read and also be read but using read-once. So everything worked and you didn't see a problem:

from: start2=$start,delta=common_timestamp-$start

In the second case, when you switched them around, the first reference would be resolved by doing the read-once, and following that the second reference would try to resolve and see that the variable had already been read, so failed as unset, which caused it to short-circuit out and not do the trigger action to generate the synthetic event:

to: delta=common_timestamp-$start,start2=$start

With your patch, we only have the single resolution which happens correctly the one time it's resolved, so this can't happen."

Fixes 067fe038e70f6 ("tracing: Add variable reference handling to hist triggers")
---
 kernel/trace/trace_events_hist.c | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)
kernel/trace/trace_events_hist.c +22 -0
index d33b046f985a..6ac35b9e195d 100644
@@ -116,6 +116,7 @@ struct hist_field {
 	struct ftrace_event_field	*field;
 	unsigned long			flags;
 	hist_field_fn_t			fn;
+	unsigned int			ref;
 	unsigned int			size;
 	unsigned int			offset;
 	unsigned int                    is_signed;
@@ -2427,8 +2428,16 @@ static int contains_operator(char *str)
 	return field_op;
 }
 
+static void get_hist_field(struct hist_field *hist_field)
+{
+	hist_field->ref++;
+}
+
 static void __destroy_hist_field(struct hist_field *hist_field)
 {
+	if (--hist_field->ref > 1)
+		return;
+
 	kfree(hist_field->var.name);
 	kfree(hist_field->name);
 	kfree(hist_field->type);
@@ -2470,6 +2479,8 @@ static struct hist_field *create_hist_field(struct hist_trigger_data *hist_data,
 	if (!hist_field)
 		return NULL;
 
+	hist_field->ref = 1;
+
 	hist_field->hist_data = hist_data;
 
 	if (flags & HIST_FIELD_FL_EXPR || flags & HIST_FIELD_FL_ALIAS)
@@ -2665,6 +2676,17 @@ static struct hist_field *create_var_ref(struct hist_trigger_data *hist_data,
 {
 	unsigned long flags = HIST_FIELD_FL_VAR_REF;
 	struct hist_field *ref_field;
+	int i;
+
+	/* Check if the variable already exists */
+	for (i = 0; i < hist_data->n_var_refs; i++) {
+		ref_field = hist_data->var_refs[i];
+		if (ref_field->var.idx == var_field->var.idx &&
+		    ref_field->var.hist_data == var_field->hist_data) {
+			get_hist_field(ref_field);
+			return ref_field;
+		}
+	}
 
 	ref_field = create_hist_field(var_field->hist_data, NULL, flags, NULL);
 	if (ref_field) {
Signature · 1 line
2.24.1
author maintainer
#11 lore raw

PATCH 5/5 tracing: Do not set trace clock if tracefs lockdown is in effect

In reply to Steven Rostedt

From: Masami Ichikawa <[email protected]>

When trace_clock option is not set and unstable clcok detected, tracing_set_default_clock() sets trace_clock(ThinkPad A285 is one of case). In that case, if lockdown is in effect, null pointer dereference error happens in ring_buffer_set_clock().

Fixes 17911ff38aa58 ("tracing: Add locked_down checks to the open calls of files created for tracefs")

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1788488

---
 kernel/trace/trace.c | 5 +++++
 1 file changed, 5 insertions(+)
kernel/trace/trace.c +5 -0
index ddb7e7f5fe8d..5b6ee4aadc26 100644
@@ -9420,6 +9420,11 @@ __init static int tracing_set_default_clock(void)
 {
 	/* sched_clock_stable() is determined in late_initcall */
 	if (!trace_boot_clock && !sched_clock_stable()) {
+		if (security_locked_down(LOCKDOWN_TRACEFS)) {
+			pr_warn("Can not set tracing clock due to lockdown\n");
+			return -EPERM;
+		}
+
 		printk(KERN_WARNING
 		       "Unstable clock detected, switching default tracing clock to \"global\"\n"
 		       "If you want to keep using the local clock, then add:\n"
Signature · 1 line
2.24.1