[ftrace-bpf 2/5] add bpf prog interface for ftrace

[yupeng0921]

create ftrace_bpf_filter in tracefs, to set a bpf filter, write a bpf
prog file descriptor to this file, to clean the bpf filter, write
empty string to this file.

Signed-off-by: yupeng0921@xxxxxxxxx
---
 include/linux/ftrace.h |  3 ++
 kernel/trace/ftrace.c  | 74 ++++++++++++++++++++++++++++++++++++++++++++++++++
 kernel/trace/trace.c   |  6 ++++
 kernel/trace/trace.h   |  4 +++
 4 files changed, 87 insertions(+)

diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h
index e54d257..9959435 100644
--- a/include/linux/ftrace.h
+++ b/include/linux/ftrace.h
@@ -769,6 +769,9 @@ static inline void ftrace_init(void) { }
 struct ftrace_graph_ent {
 	unsigned long func; /* Current function */
 	int depth;
+#ifdef FTRACE_BPF_FILTER
+	struct ftrace_regs *ctx;
+#endif
 } __packed;
 
 /*
diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index 8319e09..acb73f6 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -33,6 +33,7 @@
 #include <linux/list.h>
 #include <linux/hash.h>
 #include <linux/rcupdate.h>
+#include <linux/bpf.h>
 
 #include <trace/events/sched.h>
 
@@ -6415,10 +6416,83 @@ static const struct file_operations ftrace_pid_fops = {
 	.release	= ftrace_pid_release,
 };
 
+#ifdef FTRACE_BPF_FILTER
+static int
+ftrace_bpf_open(struct inode *inode, struct file *file)
+{
+	struct trace_array *tr = inode->i_private;
+
+	if (trace_array_get(tr) < 0)
+		return -ENODEV;
+
+	file->private_data = tr;
+
+	return 0;
+}
+
+static int
+ftrace_bpf_release(struct inode *inode, struct file *file)
+{
+	struct trace_array *tr = inode->i_private;
+
+	trace_array_put(tr);
+	return 0;
+}
+
+static ssize_t
+ftrace_bpf_write(struct file *filp, const char __user *ubuf,
+		   size_t cnt, loff_t *ppos)
+{
+	struct trace_array *tr = filp->private_data;
+	unsigned long prog_fd;
+	struct bpf_prog *prog, *old_prog;
+	ssize_t ret = 0;
+
+	old_prog = NULL;
+	mutex_lock(&ftrace_lock);
+	if (!cnt) {
+		old_prog = tr->prog;
+		goto out;
+	}
+	ret = kstrtoul_from_user(ubuf, cnt, 10, &prog_fd);
+	if (ret)
+		goto out;
+
+	prog = bpf_prog_get(prog_fd);
+	if (IS_ERR(prog)) {
+		ret = PTR_ERR(prog);
+		goto out;
+	}
+
+	old_prog = tr->prog;
+	rcu_assign_pointer(tr->prog, prog);
+
+ out:
+	mutex_unlock(&ftrace_lock);
+	if (old_prog) {
+		synchronize_rcu();
+		bpf_prog_put(old_prog);
+	}
+	if (ret)
+		return ret;
+
+	return cnt;
+}
+
+static const struct file_operations ftrace_bpf_fops = {
+	.open		= ftrace_bpf_open,
+	.write		= ftrace_bpf_write,
+	.release	= ftrace_bpf_release,
+};
+#endif	/* FTRACE_BPF_FILTER */
 void ftrace_init_tracefs(struct trace_array *tr, struct dentry *d_tracer)
 {
 	trace_create_file("set_ftrace_pid", 0644, d_tracer,
 			    tr, &ftrace_pid_fops);
+#ifdef FTRACE_BPF_FILTER
+	trace_create_file("set_ftrace_bpf", 0644, d_tracer,
+				tr, &ftrace_bpf_fops);
+#endif
 }
 
 void __init ftrace_init_tracefs_toplevel(struct trace_array *tr,
diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
index 752e5da..09c75c7 100644
--- a/kernel/trace/trace.c
+++ b/kernel/trace/trace.c
@@ -7714,6 +7714,9 @@ static int instance_mkdir(const char *name)
 	raw_spin_lock_init(&tr->start_lock);
 
 	tr->max_lock = (arch_spinlock_t)__ARCH_SPIN_LOCK_UNLOCKED;
+#ifdef FTRACE_BPF_FILTER
+	tr->prog = NULL;
+#endif
 
 	tr->current_trace = &nop_trace;
 
@@ -8354,6 +8357,9 @@ __init static int tracer_alloc_buffers(void)
 	global_trace.current_trace = &nop_trace;
 
 	global_trace.max_lock = (arch_spinlock_t)__ARCH_SPIN_LOCK_UNLOCKED;
+#ifdef FTRACE_BPF_FILTER
+	global_trace.prog = NULL;
+#endif
 
 	ftrace_init_global_array_ops(&global_trace);
 
diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h
index 401b063..dedecd6 100644
--- a/kernel/trace/trace.h
+++ b/kernel/trace/trace.h
@@ -17,6 +17,7 @@
 #include <linux/compiler.h>
 #include <linux/trace_seq.h>
 #include <linux/glob.h>
+#include <linux/filter.h>
 
 #ifdef CONFIG_FTRACE_SYSCALLS
 #include <asm/unistd.h>		/* For NR_SYSCALLS	     */
@@ -261,6 +262,9 @@ struct trace_array {
 	struct list_head	events;
 	cpumask_var_t		tracing_cpumask; /* only trace on set CPUs */
 	int			ref;
+#ifdef FTRACE_BPF_FILTER
+	struct bpf_prog __rcu *prog;
+#endif
 #ifdef CONFIG_FUNCTION_TRACER
 	struct ftrace_ops	*ops;
 	struct trace_pid_list	__rcu *function_pids;
-- 
2.7.4