sock: Make sk_protocol a 16-bit value

Match the 16-bit width of skbuff->protocol. Fills an 8-bit hole so
sizeof(struct sock) does not change.

Also take care of BPF field access for sk_type/sk_protocol. Both of them
are now outside the bitfield, so we can use load instructions without
further shifting/masking.

v5 -> v6:
 - update eBPF accessors, too (Intel's kbuild test robot)
v2 -> v3:
 - keep 'sk_type' 2 bytes aligned (Eric)
v1 -> v2:
 - preserve sk_pacing_shift as bit field (Eric)

Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: bpf@vger.kernel.org
Co-developed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Co-developed-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

include/net/sock.h

@@ -436,30 +436,15 @@ struct sock {
 	 * Because of non atomicity rules, all
 	 * changes are protected by socket lock.
 	 */
-	unsigned int		__sk_flags_offset[0];
+	u8			sk_padding : 1,
-#ifdef __BIG_ENDIAN_BITFIELD
-#define SK_FL_PROTO_SHIFT  16
-#define SK_FL_PROTO_MASK   0x00ff0000
-
-#define SK_FL_TYPE_SHIFT   0
-#define SK_FL_TYPE_MASK    0x0000ffff
-#else
-#define SK_FL_PROTO_SHIFT  8
-#define SK_FL_PROTO_MASK   0x0000ff00
-
-#define SK_FL_TYPE_SHIFT   16
-#define SK_FL_TYPE_MASK    0xffff0000
-#endif
-
-	unsigned int		sk_padding : 1,
 				sk_kern_sock : 1,
 				sk_no_check_tx : 1,
 				sk_no_check_rx : 1,
-				sk_userlocks : 4,
+				sk_userlocks : 4;
-				sk_protocol  : 8,
-				sk_type      : 16;
-	u16			sk_gso_max_segs;
 	u8			sk_pacing_shift;
+	u16			sk_type;
+	u16			sk_protocol;
+	u16			sk_gso_max_segs;
 	unsigned long	        sk_lingertime;
 	struct proto		*sk_prot_creator;
 	rwlock_t		sk_callback_lock;

include/trace/events/sock.h

@@ -147,7 +147,7 @@ TRACE_EVENT(inet_sock_set_state,
 		__field(__u16, sport)
 		__field(__u16, dport)
 		__field(__u16, family)
-		__field(__u8, protocol)
+		__field(__u16, protocol)
 		__array(__u8, saddr, 4)
 		__array(__u8, daddr, 4)
 		__array(__u8, saddr_v6, 16)

net/core/filter.c

@@ -7607,21 +7607,21 @@ u32 bpf_sock_convert_ctx_access(enum bpf_access_type type,
 		break;
 
 	case offsetof(struct bpf_sock, type):
-		BUILD_BUG_ON(HWEIGHT32(SK_FL_TYPE_MASK) != BITS_PER_BYTE * 2);
+		*insn++ = BPF_LDX_MEM(
-		*insn++ = BPF_LDX_MEM(BPF_W, si->dst_reg, si->src_reg,
+			BPF_FIELD_SIZEOF(struct sock, sk_type),
-				      offsetof(struct sock, __sk_flags_offset));
+			si->dst_reg, si->src_reg,
-		*insn++ = BPF_ALU32_IMM(BPF_AND, si->dst_reg, SK_FL_TYPE_MASK);
+			bpf_target_off(struct sock, sk_type,
-		*insn++ = BPF_ALU32_IMM(BPF_RSH, si->dst_reg, SK_FL_TYPE_SHIFT);
+				       sizeof_field(struct sock, sk_type),
-		*target_size = 2;
+				       target_size));
 		break;
 
 	case offsetof(struct bpf_sock, protocol):
-		BUILD_BUG_ON(HWEIGHT32(SK_FL_PROTO_MASK) != BITS_PER_BYTE);
+		*insn++ = BPF_LDX_MEM(
-		*insn++ = BPF_LDX_MEM(BPF_W, si->dst_reg, si->src_reg,
+			BPF_FIELD_SIZEOF(struct sock, sk_protocol),
-				      offsetof(struct sock, __sk_flags_offset));
+			si->dst_reg, si->src_reg,
-		*insn++ = BPF_ALU32_IMM(BPF_AND, si->dst_reg, SK_FL_PROTO_MASK);
+			bpf_target_off(struct sock, sk_protocol,
-		*insn++ = BPF_ALU32_IMM(BPF_RSH, si->dst_reg, SK_FL_PROTO_SHIFT);
+				       sizeof_field(struct sock, sk_protocol),
-		*target_size = 1;
+				       target_size));
 		break;
 
 	case offsetof(struct bpf_sock, src_ip4):
@@ -7903,20 +7903,13 @@ static u32 sock_addr_convert_ctx_access(enum bpf_access_type type,
 		break;
 
 	case offsetof(struct bpf_sock_addr, type):
-		SOCK_ADDR_LOAD_NESTED_FIELD_SIZE_OFF(
+		SOCK_ADDR_LOAD_NESTED_FIELD(struct bpf_sock_addr_kern,
-			struct bpf_sock_addr_kern, struct sock, sk,
+					    struct sock, sk, sk_type);
-			__sk_flags_offset, BPF_W, 0);
-		*insn++ = BPF_ALU32_IMM(BPF_AND, si->dst_reg, SK_FL_TYPE_MASK);
-		*insn++ = BPF_ALU32_IMM(BPF_RSH, si->dst_reg, SK_FL_TYPE_SHIFT);
 		break;
 
 	case offsetof(struct bpf_sock_addr, protocol):
-		SOCK_ADDR_LOAD_NESTED_FIELD_SIZE_OFF(
+		SOCK_ADDR_LOAD_NESTED_FIELD(struct bpf_sock_addr_kern,
-			struct bpf_sock_addr_kern, struct sock, sk,
+					    struct sock, sk, sk_protocol);
-			__sk_flags_offset, BPF_W, 0);
-		*insn++ = BPF_ALU32_IMM(BPF_AND, si->dst_reg, SK_FL_PROTO_MASK);
-		*insn++ = BPF_ALU32_IMM(BPF_RSH, si->dst_reg,
-					SK_FL_PROTO_SHIFT);
 		break;
 
 	case offsetof(struct bpf_sock_addr, msg_src_ip4):
@@ -8835,11 +8828,11 @@ sk_reuseport_is_valid_access(int off, int size,
 				    skb,				\
 				    SKB_FIELD)
 
-#define SK_REUSEPORT_LOAD_SK_FIELD_SIZE_OFF(SK_FIELD, BPF_SIZE, EXTRA_OFF) \
+#define SK_REUSEPORT_LOAD_SK_FIELD(SK_FIELD)				\
-	SOCK_ADDR_LOAD_NESTED_FIELD_SIZE_OFF(struct sk_reuseport_kern,	\
+	SOCK_ADDR_LOAD_NESTED_FIELD(struct sk_reuseport_kern,		\
-					     struct sock,		\
+				    struct sock,			\
-					     sk,			\
+				    sk,					\
-					     SK_FIELD, BPF_SIZE, EXTRA_OFF)
+				    SK_FIELD)
 
 static u32 sk_reuseport_convert_ctx_access(enum bpf_access_type type,
 					   const struct bpf_insn *si,
@@ -8863,16 +8856,7 @@ static u32 sk_reuseport_convert_ctx_access(enum bpf_access_type type,
 		break;
 
 	case offsetof(struct sk_reuseport_md, ip_protocol):
-		BUILD_BUG_ON(HWEIGHT32(SK_FL_PROTO_MASK) != BITS_PER_BYTE);
+		SK_REUSEPORT_LOAD_SK_FIELD(sk_protocol);
-		SK_REUSEPORT_LOAD_SK_FIELD_SIZE_OFF(__sk_flags_offset,
-						    BPF_W, 0);
-		*insn++ = BPF_ALU32_IMM(BPF_AND, si->dst_reg, SK_FL_PROTO_MASK);
-		*insn++ = BPF_ALU32_IMM(BPF_RSH, si->dst_reg,
-					SK_FL_PROTO_SHIFT);
-		/* SK_FL_PROTO_MASK and SK_FL_PROTO_SHIFT are endian
-		 * aware.  No further narrowing or masking is needed.
-		 */
-		*target_size = 1;
 		break;
 
 	case offsetof(struct sk_reuseport_md, data_end):