bpf: Implement minimal BPF perf link

Introduce a new type of BPF link - BPF perf link. This brings perf_event-based
BPF program attachments (perf_event, tracepoints, kprobes, and uprobes) into
the common BPF link infrastructure, allowing to list all active perf_event
based attachments, auto-detaching BPF program from perf_event when link's FD
is closed, get generic BPF link fdinfo/get_info functionality.

BPF_LINK_CREATE command expects perf_event's FD as target_fd. No extra flags
are currently supported.

Force-detaching and atomic BPF program updates are not yet implemented, but
with perf_event-based BPF links we now have common framework for this without
the need to extend ioctl()-based perf_event interface.

One interesting consideration is a new value for bpf_attach_type, which
BPF_LINK_CREATE command expects. Generally, it's either 1-to-1 mapping from
bpf_attach_type to bpf_prog_type, or many-to-1 mapping from a subset of
bpf_attach_types to one bpf_prog_type (e.g., see BPF_PROG_TYPE_SK_SKB or
BPF_PROG_TYPE_CGROUP_SOCK). In this case, though, we have three different
program types (KPROBE, TRACEPOINT, PERF_EVENT) using the same perf_event-based
mechanism, so it's many bpf_prog_types to one bpf_attach_type. I chose to
define a single BPF_PERF_EVENT attach type for all of them and adjust
link_create()'s logic for checking correspondence between attach type and
program type.

The alternative would be to define three new attach types (e.g., BPF_KPROBE,
BPF_TRACEPOINT, and BPF_PERF_EVENT), but that seemed like unnecessary overkill
and BPF_KPROBE will cause naming conflicts with BPF_KPROBE() macro, defined by
libbpf. I chose to not do this to avoid unnecessary proliferation of
bpf_attach_type enum values and not have to deal with naming conflicts.

Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Yonghong Song <yhs@fb.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/bpf/20210815070609.987780-5-andrii@kernel.org

2 files changed, 102 insertions, 13 deletions

diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
index 9a2068e39d23..80c03bedd6e6 100644
--- a/kernel/bpf/syscall.c
+++ b/kernel/bpf/syscall.c
@@ -2906,6 +2906,79 @@ static const struct bpf_link_ops bpf_raw_tp_link_lops = {
 	.fill_link_info = bpf_raw_tp_link_fill_link_info,
 };
 
+#ifdef CONFIG_PERF_EVENTS
+struct bpf_perf_link {
+	struct bpf_link link;
+	struct file *perf_file;
+};
+
+static void bpf_perf_link_release(struct bpf_link *link)
+{
+	struct bpf_perf_link *perf_link = container_of(link, struct bpf_perf_link, link);
+	struct perf_event *event = perf_link->perf_file->private_data;
+
+	perf_event_free_bpf_prog(event);
+	fput(perf_link->perf_file);
+}
+
+static void bpf_perf_link_dealloc(struct bpf_link *link)
+{
+	struct bpf_perf_link *perf_link = container_of(link, struct bpf_perf_link, link);
+
+	kfree(perf_link);
+}
+
+static const struct bpf_link_ops bpf_perf_link_lops = {
+	.release = bpf_perf_link_release,
+	.dealloc = bpf_perf_link_dealloc,
+};
+
+static int bpf_perf_link_attach(const union bpf_attr *attr, struct bpf_prog *prog)
+{
+	struct bpf_link_primer link_primer;
+	struct bpf_perf_link *link;
+	struct perf_event *event;
+	struct file *perf_file;
+	int err;
+
+	if (attr->link_create.flags)
+		return -EINVAL;
+
+	perf_file = perf_event_get(attr->link_create.target_fd);
+	if (IS_ERR(perf_file))
+		return PTR_ERR(perf_file);
+
+	link = kzalloc(sizeof(*link), GFP_USER);
+	if (!link) {
+		err = -ENOMEM;
+		goto out_put_file;
+	}
+	bpf_link_init(&link->link, BPF_LINK_TYPE_PERF_EVENT, &bpf_perf_link_lops, prog);
+	link->perf_file = perf_file;
+
+	err = bpf_link_prime(&link->link, &link_primer);
+	if (err) {
+		kfree(link);
+		goto out_put_file;
+	}
+
+	event = perf_file->private_data;
+	err = perf_event_set_bpf_prog(event, prog);
+	if (err) {
+		bpf_link_cleanup(&link_primer);
+		goto out_put_file;
+	}
+	/* perf_event_set_bpf_prog() doesn't take its own refcnt on prog */
+	bpf_prog_inc(prog);
+
+	return bpf_link_settle(&link_primer);
+
+out_put_file:
+	fput(perf_file);
+	return err;
+}
+#endif /* CONFIG_PERF_EVENTS */
+
 #define BPF_RAW_TRACEPOINT_OPEN_LAST_FIELD raw_tracepoint.prog_fd
 
 static int bpf_raw_tracepoint_open(const union bpf_attr *attr)
@@ -4147,15 +4220,26 @@ static int link_create(union bpf_attr *attr, bpfptr_t uattr)
 	if (ret)
 		goto out;
 
-	if (prog->type == BPF_PROG_TYPE_EXT) {
+	switch (prog->type) {
+	case BPF_PROG_TYPE_EXT:
 		ret = tracing_bpf_link_attach(attr, uattr, prog);
 		goto out;
-	}
-
-	ptype = attach_type_to_prog_type(attr->link_create.attach_type);
-	if (ptype == BPF_PROG_TYPE_UNSPEC || ptype != prog->type) {
-		ret = -EINVAL;
-		goto out;
+	case BPF_PROG_TYPE_PERF_EVENT:
+	case BPF_PROG_TYPE_KPROBE:
+	case BPF_PROG_TYPE_TRACEPOINT:
+		if (attr->link_create.attach_type != BPF_PERF_EVENT) {
+			ret = -EINVAL;
+			goto out;
+		}
+		ptype = prog->type;
+		break;
+	default:
+		ptype = attach_type_to_prog_type(attr->link_create.attach_type);
+		if (ptype == BPF_PROG_TYPE_UNSPEC || ptype != prog->type) {
+			ret = -EINVAL;
+			goto out;
+		}
+		break;
 	}
 
 	switch (ptype) {
@@ -4180,6 +4264,13 @@ static int link_create(union bpf_attr *attr, bpfptr_t uattr)
 		ret = bpf_xdp_link_attach(attr, prog);
 		break;
 #endif
+#ifdef CONFIG_PERF_EVENTS
+	case BPF_PROG_TYPE_PERF_EVENT:
+	case BPF_PROG_TYPE_TRACEPOINT:
+	case BPF_PROG_TYPE_KPROBE:
+		ret = bpf_perf_link_attach(attr, prog);
+		break;
+#endif
 	default:
 		ret = -EINVAL;
 	}
diff --git a/kernel/events/core.c b/kernel/events/core.c
index 2f07718bd41c..9fd65667bcb2 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -4697,7 +4697,6 @@ errout:
 }
 
 static void perf_event_free_filter(struct perf_event *event);
-static void perf_event_free_bpf_prog(struct perf_event *event);
 
 static void free_event_rcu(struct rcu_head *head)
 {
@@ -5574,7 +5573,6 @@ static inline int perf_fget_light(int fd, struct fd *p)
 static int perf_event_set_output(struct perf_event *event,
 				 struct perf_event *output_event);
 static int perf_event_set_filter(struct perf_event *event, void __user *arg);
-static int perf_event_set_bpf_prog(struct perf_event *event, struct bpf_prog *prog);
 static int perf_copy_attr(struct perf_event_attr __user *uattr,
 			  struct perf_event_attr *attr);
 
@@ -10013,7 +10011,7 @@ static inline bool perf_event_is_tracing(struct perf_event *event)
 	return false;
 }
 
-static int perf_event_set_bpf_prog(struct perf_event *event, struct bpf_prog *prog)
+int perf_event_set_bpf_prog(struct perf_event *event, struct bpf_prog *prog)
 {
 	bool is_kprobe, is_tracepoint, is_syscall_tp;
 
@@ -10047,7 +10045,7 @@ static int perf_event_set_bpf_prog(struct perf_event *event, struct bpf_prog *pr
 	return perf_event_attach_bpf_prog(event, prog);
 }
 
-static void perf_event_free_bpf_prog(struct perf_event *event)
+void perf_event_free_bpf_prog(struct perf_event *event)
 {
 	if (!perf_event_is_tracing(event)) {
 		perf_event_free_bpf_handler(event);
@@ -10066,12 +10064,12 @@ static void perf_event_free_filter(struct perf_event *event)
 {
 }
 
-static int perf_event_set_bpf_prog(struct perf_event *event, struct bpf_prog *prog)
+int perf_event_set_bpf_prog(struct perf_event *event, struct bpf_prog *prog)
 {
 	return -ENOENT;
 }
 
-static void perf_event_free_bpf_prog(struct perf_event *event)
+void perf_event_free_bpf_prog(struct perf_event *event)
 {
 }
 #endif /* CONFIG_EVENT_TRACING */