Merge tag 'trace-v6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace

Pull tracing updates from Steven Rostedt:

 - Add function names as a way to filter function addresses

 - Add sample module to test ftrace ops and dynamic trampolines

 - Allow stack traces to be passed from beginning event to end event for
   synthetic events. This will allow seeing the stack trace of when a
   task is scheduled out and recorded when it gets scheduled back in.

 - Add trace event helper __get_buf() to use as a temporary buffer when
   printing out trace event output.

 - Add kernel command line to create trace instances on boot up.

 - Add enabling of events to instances created at boot up.

 - Add trace_array_puts() to write into instances.

 - Allow boot instances to take a snapshot at the end of boot up.

 - Allow live patch modules to include trace events

 - Minor fixes and clean ups

* tag 'trace-v6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace: (31 commits)
  tracing: Remove unnecessary NULL assignment
  tracepoint: Allow livepatch module add trace event
  tracing: Always use canonical ftrace path
  tracing/histogram: Fix stacktrace histogram Documententation
  tracing/histogram: Fix stacktrace key
  tracing/histogram: Fix a few problems with stacktrace variable printing
  tracing: Add BUILD_BUG() to make sure stacktrace fits in strings
  tracing/histogram: Don't use strlen to find length of stacktrace variables
  tracing: Allow boot instances to have snapshot buffers
  tracing: Add trace_array_puts() to write into instance
  tracing: Add enabling of events to boot instances
  tracing: Add creation of instances at boot command line
  tracing: Fix trace_event_raw_event_synth() if else statement
  samples: ftrace: Make some global variables static
  ftrace: sample: avoid open-coded 64-bit division
  samples: ftrace: Include the nospec-branch.h only for x86
  tracing: Acquire buffer from temparary trace sequence
  tracing/histogram: Wrap remaining shell snippets in code blocks
  tracing/osnoise: No need for schedule_hrtimeout range
  bpf/tracing: Use stage6 of tracing to not duplicate macros
  ...

10 files changed, 268 insertions, 7 deletions

diff --git a/samples/Kconfig b/samples/Kconfig
index 0d81c00289ee36..daf14c35f071db 100644
--- a/samples/Kconfig
+++ b/samples/Kconfig
@@ -46,6 +46,13 @@ config SAMPLE_FTRACE_DIRECT_MULTI
 	  that hooks to wake_up_process and schedule, and prints
 	  the function addresses.
 
+config SAMPLE_FTRACE_OPS
+	tristate "Build custom ftrace ops example"
+	depends on FUNCTION_TRACER
+	help
+	  This builds an ftrace ops example that hooks two functions and
+	  measures the time taken to invoke one function a number of times.
+
 config SAMPLE_TRACE_ARRAY
         tristate "Build sample module for kernel access to Ftrace instancess"
 	depends on EVENT_TRACING && m
diff --git a/samples/Makefile b/samples/Makefile
index 9832ef3f8fcbaf..7cb632ef88eeb1 100644
--- a/samples/Makefile
+++ b/samples/Makefile
@@ -24,6 +24,7 @@ obj-$(CONFIG_SAMPLE_TRACE_CUSTOM_EVENTS) += trace_events/
 obj-$(CONFIG_SAMPLE_TRACE_PRINTK)	+= trace_printk/
 obj-$(CONFIG_SAMPLE_FTRACE_DIRECT)	+= ftrace/
 obj-$(CONFIG_SAMPLE_FTRACE_DIRECT_MULTI) += ftrace/
+obj-$(CONFIG_SAMPLE_FTRACE_OPS)		+= ftrace/
 obj-$(CONFIG_SAMPLE_TRACE_ARRAY)	+= ftrace/
 subdir-$(CONFIG_SAMPLE_UHID)		+= uhid
 obj-$(CONFIG_VIDEO_PCI_SKELETON)	+= v4l/
diff --git a/samples/ftrace/Makefile b/samples/ftrace/Makefile
index faf8cdb79c5f48..589baf2ec4e3da 100644
--- a/samples/ftrace/Makefile
+++ b/samples/ftrace/Makefile
@@ -5,6 +5,7 @@ obj-$(CONFIG_SAMPLE_FTRACE_DIRECT) += ftrace-direct-too.o
 obj-$(CONFIG_SAMPLE_FTRACE_DIRECT) += ftrace-direct-modify.o
 obj-$(CONFIG_SAMPLE_FTRACE_DIRECT_MULTI) += ftrace-direct-multi.o
 obj-$(CONFIG_SAMPLE_FTRACE_DIRECT_MULTI) += ftrace-direct-multi-modify.o
+obj-$(CONFIG_SAMPLE_FTRACE_OPS) += ftrace-ops.o
 
 CFLAGS_sample-trace-array.o := -I$(src)
 obj-$(CONFIG_SAMPLE_TRACE_ARRAY) += sample-trace-array.o
diff --git a/samples/ftrace/ftrace-direct-modify.c b/samples/ftrace/ftrace-direct-modify.c
index de5a0f67f320c3..d93abbcb1f4c54 100644
--- a/samples/ftrace/ftrace-direct-modify.c
+++ b/samples/ftrace/ftrace-direct-modify.c
@@ -3,7 +3,6 @@
 #include <linux/kthread.h>
 #include <linux/ftrace.h>
 #include <asm/asm-offsets.h>
-#include <asm/nospec-branch.h>
 
 extern void my_direct_func1(void);
 extern void my_direct_func2(void);
@@ -26,6 +25,7 @@ static unsigned long my_ip = (unsigned long)schedule;
 #ifdef CONFIG_X86_64
 
 #include <asm/ibt.h>
+#include <asm/nospec-branch.h>
 
 asm (
 "	.pushsection    .text, \"ax\", @progbits\n"
diff --git a/samples/ftrace/ftrace-direct-multi-modify.c b/samples/ftrace/ftrace-direct-multi-modify.c
index a825dbd2c9cfd2..b58c594efb51e8 100644
--- a/samples/ftrace/ftrace-direct-multi-modify.c
+++ b/samples/ftrace/ftrace-direct-multi-modify.c
@@ -3,7 +3,6 @@
 #include <linux/kthread.h>
 #include <linux/ftrace.h>
 #include <asm/asm-offsets.h>
-#include <asm/nospec-branch.h>
 
 extern void my_direct_func1(unsigned long ip);
 extern void my_direct_func2(unsigned long ip);
@@ -24,6 +23,7 @@ extern void my_tramp2(void *);
 #ifdef CONFIG_X86_64
 
 #include <asm/ibt.h>
+#include <asm/nospec-branch.h>
 
 asm (
 "	.pushsection    .text, \"ax\", @progbits\n"
diff --git a/samples/ftrace/ftrace-direct-multi.c b/samples/ftrace/ftrace-direct-multi.c
index d955a26506053f..c27cf130c3193b 100644
--- a/samples/ftrace/ftrace-direct-multi.c
+++ b/samples/ftrace/ftrace-direct-multi.c
@@ -5,7 +5,6 @@
 #include <linux/ftrace.h>
 #include <linux/sched/stat.h>
 #include <asm/asm-offsets.h>
-#include <asm/nospec-branch.h>
 
 extern void my_direct_func(unsigned long ip);
 
@@ -19,6 +18,7 @@ extern void my_tramp(void *);
 #ifdef CONFIG_X86_64
 
 #include <asm/ibt.h>
+#include <asm/nospec-branch.h>
 
 asm (
 "	.pushsection    .text, \"ax\", @progbits\n"
diff --git a/samples/ftrace/ftrace-direct-too.c b/samples/ftrace/ftrace-direct-too.c
index e13fb59a2b47a0..8139dce2a31cbd 100644
--- a/samples/ftrace/ftrace-direct-too.c
+++ b/samples/ftrace/ftrace-direct-too.c
@@ -4,7 +4,6 @@
 #include <linux/mm.h> /* for handle_mm_fault() */
 #include <linux/ftrace.h>
 #include <asm/asm-offsets.h>
-#include <asm/nospec-branch.h>
 
 extern void my_direct_func(struct vm_area_struct *vma,
 			   unsigned long address, unsigned int flags);
@@ -21,6 +20,7 @@ extern void my_tramp(void *);
 #ifdef CONFIG_X86_64
 
 #include <asm/ibt.h>
+#include <asm/nospec-branch.h>
 
 asm (
 "	.pushsection    .text, \"ax\", @progbits\n"
diff --git a/samples/ftrace/ftrace-direct.c b/samples/ftrace/ftrace-direct.c
index 1f769d0db20f39..1d3d307ca33d6e 100644
--- a/samples/ftrace/ftrace-direct.c
+++ b/samples/ftrace/ftrace-direct.c
@@ -4,7 +4,6 @@
 #include <linux/sched.h> /* for wake_up_process() */
 #include <linux/ftrace.h>
 #include <asm/asm-offsets.h>
-#include <asm/nospec-branch.h>
 
 extern void my_direct_func(struct task_struct *p);
 
@@ -18,6 +17,7 @@ extern void my_tramp(void *);
 #ifdef CONFIG_X86_64
 
 #include <asm/ibt.h>
+#include <asm/nospec-branch.h>
 
 asm (
 "	.pushsection    .text, \"ax\", @progbits\n"
diff --git a/samples/ftrace/ftrace-ops.c b/samples/ftrace/ftrace-ops.c
new file mode 100644
index 00000000000000..68d6685c80bdb1
--- /dev/null
+++ b/samples/ftrace/ftrace-ops.c
@@ -0,0 +1,252 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+#define pr_fmt(fmt)       KBUILD_MODNAME ": " fmt
+
+#include <linux/ftrace.h>
+#include <linux/ktime.h>
+#include <linux/module.h>
+
+#include <asm/barrier.h>
+
+/*
+ * Arbitrary large value chosen to be sufficiently large to minimize noise but
+ * sufficiently small to complete quickly.
+ */
+static unsigned int nr_function_calls = 100000;
+module_param(nr_function_calls, uint, 0);
+MODULE_PARM_DESC(nr_function_calls, "How many times to call the relevant tracee");
+
+/*
+ * The number of ops associated with a call site affects whether a tracer can
+ * be called directly or whether it's necessary to go via the list func, which
+ * can be significantly more expensive.
+ */
+static unsigned int nr_ops_relevant = 1;
+module_param(nr_ops_relevant, uint, 0);
+MODULE_PARM_DESC(nr_ops_relevant, "How many ftrace_ops to associate with the relevant tracee");
+
+/*
+ * On architectures where all call sites share the same trampoline, having
+ * tracers enabled for distinct functions can force the use of the list func
+ * and incur overhead for all call sites.
+ */
+static unsigned int nr_ops_irrelevant;
+module_param(nr_ops_irrelevant, uint, 0);
+MODULE_PARM_DESC(nr_ops_irrelevant, "How many ftrace_ops to associate with the irrelevant tracee");
+
+/*
+ * On architectures with DYNAMIC_FTRACE_WITH_REGS, saving the full pt_regs can
+ * be more expensive than only saving the minimal necessary regs.
+ */
+static bool save_regs;
+module_param(save_regs, bool, 0);
+MODULE_PARM_DESC(save_regs, "Register ops with FTRACE_OPS_FL_SAVE_REGS (save all registers in the trampoline)");
+
+static bool assist_recursion;
+module_param(assist_recursion, bool, 0);
+MODULE_PARM_DESC(assist_reursion, "Register ops with FTRACE_OPS_FL_RECURSION");
+
+static bool assist_rcu;
+module_param(assist_rcu, bool, 0);
+MODULE_PARM_DESC(assist_reursion, "Register ops with FTRACE_OPS_FL_RCU");
+
+/*
+ * By default, a trivial tracer is used which immediately returns to mimimize
+ * overhead. Sometimes a consistency check using a more expensive tracer is
+ * desireable.
+ */
+static bool check_count;
+module_param(check_count, bool, 0);
+MODULE_PARM_DESC(check_count, "Check that tracers are called the expected number of times\n");
+
+/*
+ * Usually it's not interesting to leave the ops registered after the test
+ * runs, but sometimes it can be useful to leave them registered so that they
+ * can be inspected through the tracefs 'enabled_functions' file.
+ */
+static bool persist;
+module_param(persist, bool, 0);
+MODULE_PARM_DESC(persist, "Successfully load module and leave ftrace ops registered after test completes\n");
+
+/*
+ * Marked as noinline to ensure that an out-of-line traceable copy is
+ * generated by the compiler.
+ *
+ * The barrier() ensures the compiler won't elide calls by determining there
+ * are no side-effects.
+ */
+static noinline void tracee_relevant(void)
+{
+	barrier();
+}
+
+/*
+ * Marked as noinline to ensure that an out-of-line traceable copy is
+ * generated by the compiler.
+ *
+ * The barrier() ensures the compiler won't elide calls by determining there
+ * are no side-effects.
+ */
+static noinline void tracee_irrelevant(void)
+{
+	barrier();
+}
+
+struct sample_ops {
+	struct ftrace_ops ops;
+	unsigned int count;
+};
+
+static void ops_func_nop(unsigned long ip, unsigned long parent_ip,
+			 struct ftrace_ops *op,
+			 struct ftrace_regs *fregs)
+{
+	/* do nothing */
+}
+
+static void ops_func_count(unsigned long ip, unsigned long parent_ip,
+			   struct ftrace_ops *op,
+			   struct ftrace_regs *fregs)
+{
+	struct sample_ops *self;
+
+	self = container_of(op, struct sample_ops, ops);
+	self->count++;
+}
+
+static struct sample_ops *ops_relevant;
+static struct sample_ops *ops_irrelevant;
+
+static struct sample_ops *ops_alloc_init(void *tracee, ftrace_func_t func,
+					 unsigned long flags, int nr)
+{
+	struct sample_ops *ops;
+
+	ops = kcalloc(nr, sizeof(*ops), GFP_KERNEL);
+	if (WARN_ON_ONCE(!ops))
+		return NULL;
+
+	for (unsigned int i = 0; i < nr; i++) {
+		ops[i].ops.func = func;
+		ops[i].ops.flags = flags;
+		WARN_ON_ONCE(ftrace_set_filter_ip(&ops[i].ops, (unsigned long)tracee, 0, 0));
+		WARN_ON_ONCE(register_ftrace_function(&ops[i].ops));
+	}
+
+	return ops;
+}
+
+static void ops_destroy(struct sample_ops *ops, int nr)
+{
+	if (!ops)
+		return;
+
+	for (unsigned int i = 0; i < nr; i++) {
+		WARN_ON_ONCE(unregister_ftrace_function(&ops[i].ops));
+		ftrace_free_filter(&ops[i].ops);
+	}
+
+	kfree(ops);
+}
+
+static void ops_check(struct sample_ops *ops, int nr,
+		      unsigned int expected_count)
+{
+	if (!ops || !check_count)
+		return;
+
+	for (unsigned int i = 0; i < nr; i++) {
+		if (ops->count == expected_count)
+			continue;
+		pr_warn("Counter called %u times (expected %u)\n",
+			ops->count, expected_count);
+	}
+}
+
+static ftrace_func_t tracer_relevant = ops_func_nop;
+static ftrace_func_t tracer_irrelevant = ops_func_nop;
+
+static int __init ftrace_ops_sample_init(void)
+{
+	unsigned long flags = 0;
+	ktime_t start, end;
+	u64 period;
+
+	if (!IS_ENABLED(CONFIG_DYNAMIC_FTRACE_WITH_REGS) && save_regs) {
+		pr_info("this kernel does not support saving registers\n");
+		save_regs = false;
+	} else if (save_regs) {
+		flags |= FTRACE_OPS_FL_SAVE_REGS;
+	}
+
+	if (assist_recursion)
+		flags |= FTRACE_OPS_FL_RECURSION;
+
+	if (assist_rcu)
+		flags |= FTRACE_OPS_FL_RCU;
+
+	if (check_count) {
+		tracer_relevant = ops_func_count;
+		tracer_irrelevant = ops_func_count;
+	}
+
+	pr_info("registering:\n"
+		"  relevant ops: %u\n"
+		"    tracee: %ps\n"
+		"    tracer: %ps\n"
+		"  irrelevant ops: %u\n"
+		"    tracee: %ps\n"
+		"    tracer: %ps\n"
+		"  saving registers: %s\n"
+		"  assist recursion: %s\n"
+		"  assist RCU: %s\n",
+		nr_ops_relevant, tracee_relevant, tracer_relevant,
+		nr_ops_irrelevant, tracee_irrelevant, tracer_irrelevant,
+		save_regs ? "YES" : "NO",
+		assist_recursion ? "YES" : "NO",
+		assist_rcu ? "YES" : "NO");
+
+	ops_relevant = ops_alloc_init(tracee_relevant, tracer_relevant,
+				      flags, nr_ops_relevant);
+	ops_irrelevant = ops_alloc_init(tracee_irrelevant, tracer_irrelevant,
+					flags, nr_ops_irrelevant);
+
+	start = ktime_get();
+	for (unsigned int i = 0; i < nr_function_calls; i++)
+		tracee_relevant();
+	end = ktime_get();
+
+	ops_check(ops_relevant, nr_ops_relevant, nr_function_calls);
+	ops_check(ops_irrelevant, nr_ops_irrelevant, 0);
+
+	period = ktime_to_ns(ktime_sub(end, start));
+
+	pr_info("Attempted %u calls to %ps in %lluns (%lluns / call)\n",
+		nr_function_calls, tracee_relevant,
+		period, div_u64(period, nr_function_calls));
+
+	if (persist)
+		return 0;
+
+	ops_destroy(ops_relevant, nr_ops_relevant);
+	ops_destroy(ops_irrelevant, nr_ops_irrelevant);
+
+	/*
+	 * The benchmark completed sucessfully, but there's no reason to keep
+	 * the module around. Return an error do the user doesn't have to
+	 * manually unload the module.
+	 */
+	return -EINVAL;
+}
+module_init(ftrace_ops_sample_init);
+
+static void __exit ftrace_ops_sample_exit(void)
+{
+	ops_destroy(ops_relevant, nr_ops_relevant);
+	ops_destroy(ops_irrelevant, nr_ops_irrelevant);
+}
+module_exit(ftrace_ops_sample_exit);
+
+MODULE_AUTHOR("Mark Rutland");
+MODULE_DESCRIPTION("Example of using custom ftrace_ops");
+MODULE_LICENSE("GPL");
diff --git a/samples/user_events/example.c b/samples/user_events/example.c
index d06dc24156ec03..18e34c9d708e28 100644
--- a/samples/user_events/example.c
+++ b/samples/user_events/example.c
@@ -23,8 +23,8 @@
 #endif
 
 /* Assumes debugfs is mounted */
-const char *data_file = "/sys/kernel/debug/tracing/user_events_data";
-const char *status_file = "/sys/kernel/debug/tracing/user_events_status";
+const char *data_file = "/sys/kernel/tracing/user_events_data";
+const char *status_file = "/sys/kernel/tracing/user_events_status";
 
 static int event_status(long **status)
 {