BPF updates

index 70eda5a..b7b0fb1 100644
@@ -347,6 +347,10 @@ enum bpf_func_id {
 #define BPF_F_ZERO_CSUM_TX		(1ULL << 1)
 #define BPF_F_DONT_FRAGMENT		(1ULL << 2)
 
+/* BPF_FUNC_perf_event_output flags. */
+#define BPF_F_INDEX_MASK		0xffffffffULL
+#define BPF_F_CURRENT_CPU		BPF_F_INDEX_MASK
+
 /* user accessible mirror of in-kernel sk_buff.
  * new fields can only be added to the end of this structure
  */

index 6855878..6bfe55c 100644
@@ -225,11 +225,12 @@ static const struct bpf_func_proto bpf_perf_event_read_proto = {
 	.arg2_type	= ARG_ANYTHING,
 };
 
-static u64 bpf_perf_event_output(u64 r1, u64 r2, u64 index, u64 r4, u64 size)
+static u64 bpf_perf_event_output(u64 r1, u64 r2, u64 flags, u64 r4, u64 size)
 {
 	struct pt_regs *regs = (struct pt_regs *) (long) r1;
 	struct bpf_map *map = (struct bpf_map *) (long) r2;
 	struct bpf_array *array = container_of(map, struct bpf_array, map);
+	u64 index = flags & BPF_F_INDEX_MASK;
 	void *data = (void *) (long) r4;
 	struct perf_sample_data sample_data;
 	struct perf_event *event;

@@ -239,6 +240,10 @@ static u64 bpf_perf_event_output(u64 r1, u64 r2, u64 index, u64 r4, u64 size)
 		.data = data,
 	};
 
+	if (unlikely(flags & ~(BPF_F_INDEX_MASK)))
+		return -EINVAL;
+	if (index == BPF_F_CURRENT_CPU)
+		index = raw_smp_processor_id();
 	if (unlikely(index >= array->map.max_entries))
 		return -E2BIG;
 

index 5fb3c61..2116924 100644
@@ -167,6 +167,8 @@ struct bpf_array {
 
 u64 bpf_tail_call(u64 ctx, u64 r2, u64 index, u64 r4, u64 r5);
 u64 bpf_get_stackid(u64 r1, u64 r2, u64 r3, u64 r4, u64 r5);
+u64 bpf_perf_event_output(u64 r1, u64 r2, u64 flags, u64 r4, u64 size);
+
 void bpf_fd_array_map_clear(struct bpf_map *map);
 bool bpf_prog_array_compatible(struct bpf_array *array, const struct bpf_prog *fp);
 const struct bpf_func_proto *bpf_get_trace_printk_proto(void);

index 6bfe55c..7fe2f22 100644
@@ -225,7 +225,7 @@ static const struct bpf_func_proto bpf_perf_event_read_proto = {
 	.arg2_type	= ARG_ANYTHING,
 };
 
-static u64 bpf_perf_event_output(u64 r1, u64 r2, u64 flags, u64 r4, u64 size)
+u64 bpf_perf_event_output(u64 r1, u64 r2, u64 flags, u64 r4, u64 size)
 {
 	struct pt_regs *regs = (struct pt_regs *) (long) r1;
 	struct bpf_map *map = (struct bpf_map *) (long) r2;

index 5d2ac2b..3521252 100644
@@ -46,6 +46,7 @@
 #include <linux/seccomp.h>
 #include <linux/if_vlan.h>
 #include <linux/bpf.h>
+#include <linux/perf_event.h>
 #include <net/sch_generic.h>
 #include <net/cls_cgroup.h>
 #include <net/dst_metadata.h>

@@ -1584,6 +1585,33 @@ static const struct bpf_func_proto bpf_csum_diff_proto = {
 	.arg5_type	= ARG_ANYTHING,
 };
 
+u64 __weak bpf_perf_event_output(u64 r1, u64 r2, u64 flags, u64 r4, u64 size)
+{
+	return -EACCES;
+}
+
+static DEFINE_PER_CPU(struct pt_regs, bpf_pt_regs);
+
+static u64 bpf_skb_event_output(u64 r1, u64 r2, u64 flags, u64 r4, u64 size)
+{
+	struct pt_regs *regs = this_cpu_ptr(&bpf_pt_regs);
+
+	perf_fetch_caller_regs(regs);
+
+	return bpf_perf_event_output((long)regs, r2, flags, r4, size);
+}
+
+static const struct bpf_func_proto bpf_skb_event_output_proto = {
+	.func		= bpf_skb_event_output,
+	.gpl_only	= true,
+	.ret_type	= RET_INTEGER,
+	.arg1_type	= ARG_PTR_TO_CTX,
+	.arg2_type	= ARG_CONST_MAP_PTR,
+	.arg3_type	= ARG_ANYTHING,
+	.arg4_type	= ARG_PTR_TO_STACK,
+	.arg5_type	= ARG_CONST_STACK_SIZE,
+};
+
 static u64 bpf_clone_redirect(u64 r1, u64 ifindex, u64 flags, u64 r4, u64 r5)
 {
 	struct sk_buff *skb = (struct sk_buff *) (long) r1, *skb2;

@@ -2039,6 +2067,8 @@ tc_cls_act_func_proto(enum bpf_func_id func_id)
 		return &bpf_redirect_proto;
 	case BPF_FUNC_get_route_realm:
 		return &bpf_get_route_realm_proto;
+	case BPF_FUNC_perf_event_output:
+		return &bpf_skb_event_output_proto;
 	default:
 		return sk_filter_func_proto(func_id);
 	}