Merge branch 'bpf-map-in-map'

Martin KaFai Lau says:

====================
bpf: Add map-in-map support

This patchset adds map-in-map support (map->map).
One use case is the (vips -> webservers) in the L4 load balancer so
that different vips can be backed by different set of webservers.

Please refer to the individual commit log for details.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 67 insertions, 7 deletions

diff --git a/kernel/bpf/arraymap.c b/kernel/bpf/arraymap.c
index bcf9955fac95..bc9da93db403 100644
--- a/kernel/bpf/arraymap.c
+++ b/kernel/bpf/arraymap.c
@@ -17,6 +17,8 @@
 #include <linux/filter.h>
 #include <linux/perf_event.h>
 
+#include "map_in_map.h"
+
 static void bpf_array_free_percpu(struct bpf_array *array)
 {
 	int i;
@@ -117,20 +119,17 @@ static void *array_map_lookup_elem(struct bpf_map *map, void *key)
 /* emit BPF instructions equivalent to C code of array_map_lookup_elem() */
 static u32 array_map_gen_lookup(struct bpf_map *map, struct bpf_insn *insn_buf)
 {
-	struct bpf_array *array = container_of(map, struct bpf_array, map);
 	struct bpf_insn *insn = insn_buf;
-	u32 elem_size = array->elem_size;
+	u32 elem_size = round_up(map->value_size, 8);
 	const int ret = BPF_REG_0;
 	const int map_ptr = BPF_REG_1;
 	const int index = BPF_REG_2;
 
 	*insn++ = BPF_ALU64_IMM(BPF_ADD, map_ptr, offsetof(struct bpf_array, value));
 	*insn++ = BPF_LDX_MEM(BPF_W, ret, index, 0);
-	*insn++ = BPF_JMP_IMM(BPF_JGE, ret, array->map.max_entries,
-			      elem_size == 1 ? 2 : 3);
-	if (elem_size == 1) {
-		/* nop */
-	} else if (is_power_of_2(elem_size)) {
+	*insn++ = BPF_JMP_IMM(BPF_JGE, ret, map->max_entries, 3);
+
+	if (is_power_of_2(elem_size)) {
 		*insn++ = BPF_ALU64_IMM(BPF_LSH, ret, ilog2(elem_size));
 	} else {
 		*insn++ = BPF_ALU64_IMM(BPF_MUL, ret, elem_size);
@@ -605,3 +604,64 @@ static int __init register_cgroup_array_map(void)
 }
 late_initcall(register_cgroup_array_map);
 #endif
+
+static struct bpf_map *array_of_map_alloc(union bpf_attr *attr)
+{
+	struct bpf_map *map, *inner_map_meta;
+
+	inner_map_meta = bpf_map_meta_alloc(attr->inner_map_fd);
+	if (IS_ERR(inner_map_meta))
+		return inner_map_meta;
+
+	map = fd_array_map_alloc(attr);
+	if (IS_ERR(map)) {
+		bpf_map_meta_free(inner_map_meta);
+		return map;
+	}
+
+	map->inner_map_meta = inner_map_meta;
+
+	return map;
+}
+
+static void array_of_map_free(struct bpf_map *map)
+{
+	/* map->inner_map_meta is only accessed by syscall which
+	 * is protected by fdget/fdput.
+	 */
+	bpf_map_meta_free(map->inner_map_meta);
+	bpf_fd_array_map_clear(map);
+	fd_array_map_free(map);
+}
+
+static void *array_of_map_lookup_elem(struct bpf_map *map, void *key)
+{
+	struct bpf_map **inner_map = array_map_lookup_elem(map, key);
+
+	if (!inner_map)
+		return NULL;
+
+	return READ_ONCE(*inner_map);
+}
+
+static const struct bpf_map_ops array_of_map_ops = {
+	.map_alloc = array_of_map_alloc,
+	.map_free = array_of_map_free,
+	.map_get_next_key = array_map_get_next_key,
+	.map_lookup_elem = array_of_map_lookup_elem,
+	.map_delete_elem = fd_array_map_delete_elem,
+	.map_fd_get_ptr = bpf_map_fd_get_ptr,
+	.map_fd_put_ptr = bpf_map_fd_put_ptr,
+};
+
+static struct bpf_map_type_list array_of_map_type __ro_after_init = {
+	.ops = &array_of_map_ops,
+	.type = BPF_MAP_TYPE_ARRAY_OF_MAPS,
+};
+
+static int __init register_array_of_map(void)
+{
+	bpf_register_map_type(&array_of_map_type);
+	return 0;
+}
+late_initcall(register_array_of_map);