Merge tag 'trace-v6.17' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace

Pull tracing updates from Steven Rostedt:

 - Deprecate auto-mounting tracefs to /sys/kernel/debug/tracing

   When tracefs was first introduced back in 2014, the directory
   /sys/kernel/tracing was added and is the designated location to mount
   tracefs. To keep backward compatibility, tracefs was auto-mounted in
   /sys/kernel/debug/tracing as well.

   All distros now mount tracefs on /sys/kernel/tracing. Having it seen
   in two different locations has lead to various issues and
   inconsistencies.

   The VFS folks have to also maintain debugfs_create_automount() for
   this single user.

   It's been over 10 years. Tooling and scripts should start replacing
   the debugfs location with the tracefs one. The reason tracefs was
   created in the first place was to allow access to the tracing
   facilities without the need to configure debugfs into the kernel.
   Using tracefs should now be more robust.

   A new config is created: CONFIG_TRACEFS_AUTOMOUNT_DEPRECATED which is
   default y, so that the kernel is still built with the automount. This
   config allows those that want to remove the automount from debugfs to
   do so.

   When tracefs is accessed from /sys/kernel/debug/tracing, the
   following printk is triggerd:

     pr_warn("NOTICE: Automounting of tracing to debugfs is deprecated and will be removed in 2030\n");

   This gives users another 5 years to fix their scripts.

 - Use queue_rcu_work() instead of call_rcu() for freeing event filters

   The number of filters to be free can be many depending on the number
   of events within an event system. Freeing them from softirq context
   can potentially cause undesired latency. Use the RCU workqueue to
   free them instead.

 - Remove pointless memory barriers in latency code

   Memory barriers were added to some of the latency code a long time
   ago with the idea of "making them visible", but that's not what
   memory barriers are for. They are to synchronize access between
   different variables. There was no synchronization here making them
   pointless.

 - Remove "__attribute__()" from the type field of event format

   When LLVM is used to compile the kernel with CONFIG_DEBUG_INFO_BTF=y
   and PAHOLE_HAS_BTF_TAG=y, some of the format fields get expanded with
   the following:

     field:const char * filename;      offset:24;      size:8; signed:0;

   Turns into:

     field:const char __attribute__((btf_type_tag("user"))) * filename;      offset:24;      size:8; signed:0;

   This confuses parsers. Add code to strip these tags from the strings.

 - Add eprobe config option CONFIG_EPROBE_EVENTS

   Eprobes were added back in 5.15 but were only enabled when another
   probe was enabled (kprobe, fprobe, uprobe, etc). The eprobes had no
   config option of their own. Add one as they should be a separate
   entity.

   It's default y to keep with the old kernels but still has
   dependencies on TRACING and HAVE_REGS_AND_STACK_ACCESS_API.

 - Add eprobe documentation

   When eprobes were added back in 5.15 no documentation was added to
   describe them. This needs to be rectified.

 - Replace open coded cpumask_next_wrap() in move_to_next_cpu()

 - Have preemptirq_delay_run() use off-stack CPU mask

 - Remove obsolete comment about pelt_cfs event

   DECLARE_TRACE() appends "_tp" to trace events now, but the comment
   above pelt_cfs still mentioned appending it manually.

 - Remove EVENT_FILE_FL_SOFT_MODE flag

   The SOFT_MODE flag was required when the soft enabling and disabling
   of trace events was first introduced. But there was a bug with this
   approach as it only worked for a single instance. When multiple users
   required soft disabling and disabling the code was changed to have a
   ref count. The SOFT_MODE flag is now set iff the ref count is non
   zero. This is redundant and just reading the ref count is good
   enough.

 - Fix typo in comment

* tag 'trace-v6.17' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  Documentation: tracing: Add documentation about eprobes
  tracing: Have eprobes have their own config option
  tracing: Remove "__attribute__()" from the type field of event format
  tracing: Deprecate auto-mounting tracefs in debugfs
  tracing: Fix comment in trace_module_remove_events()
  tracing: Remove EVENT_FILE_FL_SOFT_MODE flag
  tracing: Remove pointless memory barriers
  tracing/sched: Remove obsolete comment on suffixes
  kernel: trace: preemptirq_delay_test: use offstack cpu mask
  tracing: Use queue_rcu_work() to free filters
  tracing: Replace opencoded cpumask_next_wrap() in move_to_next_cpu()

Author: torvalds

kernel/trace/trace_events.c

@@ -768,6 +768,7 @@ static int __ftrace_event_enable_disable(struct trace_event_file *file,
 {
 	struct trace_event_call *call = file->event_call;
 	struct trace_array *tr = file->tr;
+	bool soft_mode = atomic_read(&file->sm_ref) != 0;
 	int ret = 0;
 	int disable;
 
@@ -782,19 +783,19 @@ static int __ftrace_event_enable_disable(struct trace_event_file *file,
 		 * is set we do not want the event to be enabled before we
 		 * clear the bit.
 		 *
-		 * When soft_disable is not set but the SOFT_MODE flag is,
+		 * When soft_disable is not set but the soft_mode is,
 		 * we do nothing. Do not disable the tracepoint, otherwise
 		 * "soft enable"s (clearing the SOFT_DISABLED bit) wont work.
 		 */
 		if (soft_disable) {
 			if (atomic_dec_return(&file->sm_ref) > 0)
 				break;
 			disable = file->flags & EVENT_FILE_FL_SOFT_DISABLED;
-			clear_bit(EVENT_FILE_FL_SOFT_MODE_BIT, &file->flags);
+			soft_mode = false;
 			/* Disable use of trace_buffered_event */
 			trace_buffered_event_disable();
 		} else
-			disable = !(file->flags & EVENT_FILE_FL_SOFT_MODE);
+			disable = !soft_mode;
 
 		if (disable && (file->flags & EVENT_FILE_FL_ENABLED)) {
 			clear_bit(EVENT_FILE_FL_ENABLED_BIT, &file->flags);
@@ -812,8 +813,8 @@ static int __ftrace_event_enable_disable(struct trace_event_file *file,
 
 			WARN_ON_ONCE(ret);
 		}
-		/* If in SOFT_MODE, just set the SOFT_DISABLE_BIT, else clear it */
-		if (file->flags & EVENT_FILE_FL_SOFT_MODE)
+		/* If in soft mode, just set the SOFT_DISABLE_BIT, else clear it */
+		if (soft_mode)
 			set_bit(EVENT_FILE_FL_SOFT_DISABLED_BIT, &file->flags);
 		else
 			clear_bit(EVENT_FILE_FL_SOFT_DISABLED_BIT, &file->flags);
@@ -823,7 +824,7 @@ static int __ftrace_event_enable_disable(struct trace_event_file *file,
 		 * When soft_disable is set and enable is set, we want to
 		 * register the tracepoint for the event, but leave the event
 		 * as is. That means, if the event was already enabled, we do
-		 * nothing (but set SOFT_MODE). If the event is disabled, we
+		 * nothing (but set soft_mode). If the event is disabled, we
 		 * set SOFT_DISABLED before enabling the event tracepoint, so
 		 * it still seems to be disabled.
 		 */
@@ -832,15 +833,15 @@ static int __ftrace_event_enable_disable(struct trace_event_file *file,
 		else {
 			if (atomic_inc_return(&file->sm_ref) > 1)
 				break;
-			set_bit(EVENT_FILE_FL_SOFT_MODE_BIT, &file->flags);
+			soft_mode = true;
 			/* Enable use of trace_buffered_event */
 			trace_buffered_event_enable();
 		}
 
 		if (!(file->flags & EVENT_FILE_FL_ENABLED)) {
 			bool cmd = false, tgid = false;
 
-			/* Keep the event disabled, when going to SOFT_MODE. */
+			/* Keep the event disabled, when going to soft mode. */
 			if (soft_disable)
 				set_bit(EVENT_FILE_FL_SOFT_DISABLED_BIT, &file->flags);
 
@@ -1792,8 +1793,7 @@ event_enable_read(struct file *filp, char __user *ubuf, size_t cnt,
 	    !(flags & EVENT_FILE_FL_SOFT_DISABLED))
 		strcpy(buf, "1");
 
-	if (flags & EVENT_FILE_FL_SOFT_DISABLED ||
-	    flags & EVENT_FILE_FL_SOFT_MODE)
+	if (atomic_read(&file->sm_ref) != 0)
 		strcat(buf, "*");
 
 	strcat(buf, "\n");
@@ -3267,43 +3267,120 @@ static void add_str_to_module(struct module *module, char *str)
 	list_add(&modstr->next, &module_strings);
 }
 
+#define ATTRIBUTE_STR "__attribute__("
+#define ATTRIBUTE_STR_LEN (sizeof(ATTRIBUTE_STR) - 1)
+
+/* Remove all __attribute__() from @type. Return allocated string or @type. */
+static char *sanitize_field_type(const char *type)
+{
+	char *attr, *tmp, *next, *ret = (char *)type;
+	int depth;
+
+	next = (char *)type;
+	while ((attr = strstr(next, ATTRIBUTE_STR))) {
+		/* Retry if "__attribute__(" is a part of another word. */
+		if (attr != next && !isspace(attr[-1])) {
+			next = attr + ATTRIBUTE_STR_LEN;
+			continue;
+		}
+
+		if (ret == type) {
+			ret = kstrdup(type, GFP_KERNEL);
+			if (WARN_ON_ONCE(!ret))
+				return NULL;
+			attr = ret + (attr - type);
+		}
+
+		/* the ATTRIBUTE_STR already has the first '(' */
+		depth = 1;
+		next = attr + ATTRIBUTE_STR_LEN;
+		do {
+			tmp = strpbrk(next, "()");
+			/* There is unbalanced parentheses */
+			if (WARN_ON_ONCE(!tmp)) {
+				kfree(ret);
+				return (char *)type;
+			}
+
+			if (*tmp == '(')
+				depth++;
+			else
+				depth--;
+			next = tmp + 1;
+		} while (depth > 0);
+		next = skip_spaces(next);
+		strcpy(attr, next);
+		next = attr;
+	}
+	return ret;
+}
+
+static char *find_replacable_eval(const char *type, const char *eval_string,
+				  int len)
+{
+	char *ptr;
+
+	if (!eval_string)
+		return NULL;
+
+	ptr = strchr(type, '[');
+	if (!ptr)
+		return NULL;
+	ptr++;
+
+	if (!isalpha(*ptr) && *ptr != '_')
+		return NULL;
+
+	if (strncmp(eval_string, ptr, len) != 0)
+		return NULL;
+
+	return ptr;
+}
+
 static void update_event_fields(struct trace_event_call *call,
 				struct trace_eval_map *map)
 {
 	struct ftrace_event_field *field;
+	const char *eval_string = NULL;
 	struct list_head *head;
+	int len = 0;
 	char *ptr;
 	char *str;
-	int len = strlen(map->eval_string);
 
 	/* Dynamic events should never have field maps */
-	if (WARN_ON_ONCE(call->flags & TRACE_EVENT_FL_DYNAMIC))
+	if (call->flags & TRACE_EVENT_FL_DYNAMIC)
 		return;
 
+	if (map) {
+		eval_string = map->eval_string;
+		len = strlen(map->eval_string);
+	}
+
 	head = trace_get_fields(call);
 	list_for_each_entry(field, head, link) {
-		ptr = strchr(field->type, '[');
-		if (!ptr)
-			continue;
-		ptr++;
-
-		if (!isalpha(*ptr) && *ptr != '_')
-			continue;
+		str = sanitize_field_type(field->type);
+		if (!str)
+			return;
 
-		if (strncmp(map->eval_string, ptr, len) != 0)
-			continue;
+		ptr = find_replacable_eval(str, eval_string, len);
+		if (ptr) {
+			if (str == field->type) {
+				str = kstrdup(field->type, GFP_KERNEL);
+				if (WARN_ON_ONCE(!str))
+					return;
+				ptr = str + (ptr - field->type);
+			}
 
-		str = kstrdup(field->type, GFP_KERNEL);
-		if (WARN_ON_ONCE(!str))
-			return;
-		ptr = str + (ptr - field->type);
-		ptr = eval_replace(ptr, map, len);
-		/* enum/sizeof string smaller than value */
-		if (WARN_ON_ONCE(!ptr)) {
-			kfree(str);
-			continue;
+			ptr = eval_replace(ptr, map, len);
+			/* enum/sizeof string smaller than value */
+			if (WARN_ON_ONCE(!ptr)) {
+				kfree(str);
+				continue;
+			}
 		}
 
+		if (str == field->type)
+			continue;
 		/*
 		 * If the event is part of a module, then we need to free the string
 		 * when the module is removed. Otherwise, it will stay allocated
@@ -3313,14 +3390,18 @@ static void update_event_fields(struct trace_event_call *call,
 			add_str_to_module(call->module, str);
 
 		field->type = str;
+		if (field->filter_type == FILTER_OTHER)
+			field->filter_type = filter_assign_type(field->type);
 	}
 }
 
-void trace_event_eval_update(struct trace_eval_map **map, int len)
+/* Update all events for replacing eval and sanitizing */
+void trace_event_update_all(struct trace_eval_map **map, int len)
 {
 	struct trace_event_call *call, *p;
 	const char *last_system = NULL;
 	bool first = false;
+	bool updated;
 	int last_i;
 	int i;
 
@@ -3333,6 +3414,7 @@ void trace_event_eval_update(struct trace_eval_map **map, int len)
 			last_system = call->class->system;
 		}
 
+		updated = false;
 		/*
 		 * Since calls are grouped by systems, the likelihood that the
 		 * next call in the iteration belongs to the same system as the
@@ -3352,8 +3434,12 @@ void trace_event_eval_update(struct trace_eval_map **map, int len)
 				}
 				update_event_printk(call, map[i]);
 				update_event_fields(call, map[i]);
+				updated = true;
 			}
 		}
+		/* If not updated yet, update field for sanitizing. */
+		if (!updated)
+			update_event_fields(call, NULL);
 		cond_resched();
 	}
 	up_write(&trace_event_sem);
@@ -3587,7 +3673,7 @@ static int probe_remove_event_call(struct trace_event_call *call)
 			continue;
 		/*
 		 * We can't rely on ftrace_event_enable_disable(enable => 0)
-		 * we are going to do, EVENT_FILE_FL_SOFT_MODE can suppress
+		 * we are going to do, soft mode can suppress
 		 * TRACE_REG_UNREGISTER.
 		 */
 		if (file->flags & EVENT_FILE_FL_ENABLED)
@@ -3698,7 +3784,7 @@ static void trace_module_remove_events(struct module *mod)
 		if (call->module == mod)
 			__trace_remove_event_call(call);
 	}
-	/* Check for any strings allocade for this module */
+	/* Check for any strings allocated for this module */
 	list_for_each_entry_safe(modstr, m, &module_strings, next) {
 		if (modstr->module != mod)
 			continue;
@@ -4002,7 +4088,7 @@ static int free_probe_data(void *data)
 
 	edata->ref--;
 	if (!edata->ref) {
-		/* Remove the SOFT_MODE flag */
+		/* Remove soft mode */
 		__ftrace_event_enable_disable(edata->file, 0, 1);
 		trace_event_put_ref(edata->file->event_call);
 		kfree(edata);