bpf: allow programs to write to certain skb fields

allow programs read/write skb->mark, tc_index fields and
((struct qdisc_skb_cb *)cb)->data.

mark and tc_index are generically useful in TC.
cb[0]-cb[4] are primarily used to pass arguments from one
program to another called via bpf_tail_call() which can
be seen in sockex3_kern.c example.

All fields of 'struct __sk_buff' are readable to socket and tc_cls_act progs.
mark, tc_index are writeable from tc_cls_act only.
cb[0]-cb[4] are writeable by both sockets and tc_cls_act.

Add verifier tests and improve sample code.

Signed-off-by: Alexei Starovoitov <ast@plumgrid.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

2 files changed, 93 insertions, 26 deletions

diff --git a/samples/bpf/sockex3_kern.c b/samples/bpf/sockex3_kern.c
index 2625b98..41ae2fd 100644
--- a/samples/bpf/sockex3_kern.c
+++ b/samples/bpf/sockex3_kern.c
@@ -89,7 +89,6 @@ static inline __u32 ipv6_addr_hash(struct __sk_buff *ctx, __u64 off)
 
 struct globals {
 	struct flow_keys flow;
-	__u32 nhoff;
 };
 
 struct bpf_map_def SEC("maps") percpu_map = {
@@ -139,7 +138,7 @@ static void update_stats(struct __sk_buff *skb, struct globals *g)
 static __always_inline void parse_ip_proto(struct __sk_buff *skb,
 					   struct globals *g, __u32 ip_proto)
 {
-	__u32 nhoff = g->nhoff;
+	__u32 nhoff = skb->cb[0];
 	int poff;
 
 	switch (ip_proto) {
@@ -165,7 +164,7 @@ static __always_inline void parse_ip_proto(struct __sk_buff *skb,
 		if (gre_flags & GRE_SEQ)
 			nhoff += 4;
 
-		g->nhoff = nhoff;
+		skb->cb[0] = nhoff;
 		parse_eth_proto(skb, gre_proto);
 		break;
 	}
@@ -195,7 +194,7 @@ PROG(PARSE_IP)(struct __sk_buff *skb)
 	if (!g)
 		return 0;
 
-	nhoff = g->nhoff;
+	nhoff = skb->cb[0];
 
 	if (unlikely(ip_is_fragment(skb, nhoff)))
 		return 0;
@@ -210,7 +209,7 @@ PROG(PARSE_IP)(struct __sk_buff *skb)
 	verlen = load_byte(skb, nhoff + 0/*offsetof(struct iphdr, ihl)*/);
 	nhoff += (verlen & 0xF) << 2;
 
-	g->nhoff = nhoff;
+	skb->cb[0] = nhoff;
 	parse_ip_proto(skb, g, ip_proto);
 	return 0;
 }
@@ -223,7 +222,7 @@ PROG(PARSE_IPV6)(struct __sk_buff *skb)
 	if (!g)
 		return 0;
 
-	nhoff = g->nhoff;
+	nhoff = skb->cb[0];
 
 	ip_proto = load_byte(skb,
 			     nhoff + offsetof(struct ipv6hdr, nexthdr));
@@ -233,25 +232,21 @@ PROG(PARSE_IPV6)(struct __sk_buff *skb)
 				     nhoff + offsetof(struct ipv6hdr, daddr));
 	nhoff += sizeof(struct ipv6hdr);
 
-	g->nhoff = nhoff;
+	skb->cb[0] = nhoff;
 	parse_ip_proto(skb, g, ip_proto);
 	return 0;
 }
 
 PROG(PARSE_VLAN)(struct __sk_buff *skb)
 {
-	struct globals *g = this_cpu_globals();
 	__u32 nhoff, proto;
 
-	if (!g)
-		return 0;
-
-	nhoff = g->nhoff;
+	nhoff = skb->cb[0];
 
 	proto = load_half(skb, nhoff + offsetof(struct vlan_hdr,
 						h_vlan_encapsulated_proto));
 	nhoff += sizeof(struct vlan_hdr);
-	g->nhoff = nhoff;
+	skb->cb[0] = nhoff;
 
 	parse_eth_proto(skb, proto);
 
@@ -260,17 +255,13 @@ PROG(PARSE_VLAN)(struct __sk_buff *skb)
 
 PROG(PARSE_MPLS)(struct __sk_buff *skb)
 {
-	struct globals *g = this_cpu_globals();
 	__u32 nhoff, label;
 
-	if (!g)
-		return 0;
-
-	nhoff = g->nhoff;
+	nhoff = skb->cb[0];
 
 	label = load_word(skb, nhoff);
 	nhoff += sizeof(struct mpls_label);
-	g->nhoff = nhoff;
+	skb->cb[0] = nhoff;
 
 	if (label & MPLS_LS_S_MASK) {
 		__u8 verlen = load_byte(skb, nhoff);
@@ -288,14 +279,10 @@ PROG(PARSE_MPLS)(struct __sk_buff *skb)
 SEC("socket/0")
 int main_prog(struct __sk_buff *skb)
 {
-	struct globals *g = this_cpu_globals();
 	__u32 nhoff = ETH_HLEN;
 	__u32 proto = load_half(skb, 12);
 
-	if (!g)
-		return 0;
-
-	g->nhoff = nhoff;
+	skb->cb[0] = nhoff;
 	parse_eth_proto(skb, proto);
 	return 0;
 }
diff --git a/samples/bpf/test_verifier.c b/samples/bpf/test_verifier.c
index 12f3780..6936059 100644
--- a/samples/bpf/test_verifier.c
+++ b/samples/bpf/test_verifier.c
@@ -29,6 +29,7 @@ struct bpf_test {
 		ACCEPT,
 		REJECT
 	} result;
+	enum bpf_prog_type prog_type;
 };
 
 static struct bpf_test tests[] = {
@@ -743,6 +744,84 @@ static struct bpf_test tests[] = {
 		.errstr = "different pointers",
 		.result = REJECT,
 	},
+	{
+		"check skb->mark is not writeable by sockets",
+		.insns = {
+			BPF_STX_MEM(BPF_W, BPF_REG_1, BPF_REG_1,
+				    offsetof(struct __sk_buff, mark)),
+			BPF_EXIT_INSN(),
+		},
+		.errstr = "invalid bpf_context access",
+		.result = REJECT,
+	},
+	{
+		"check skb->tc_index is not writeable by sockets",
+		.insns = {
+			BPF_STX_MEM(BPF_W, BPF_REG_1, BPF_REG_1,
+				    offsetof(struct __sk_buff, tc_index)),
+			BPF_EXIT_INSN(),
+		},
+		.errstr = "invalid bpf_context access",
+		.result = REJECT,
+	},
+	{
+		"check non-u32 access to cb",
+		.insns = {
+			BPF_STX_MEM(BPF_H, BPF_REG_1, BPF_REG_1,
+				    offsetof(struct __sk_buff, cb[0])),
+			BPF_EXIT_INSN(),
+		},
+		.errstr = "invalid bpf_context access",
+		.result = REJECT,
+	},
+	{
+		"check out of range skb->cb access",
+		.insns = {
+			BPF_LDX_MEM(BPF_W, BPF_REG_0, BPF_REG_1,
+				    offsetof(struct __sk_buff, cb[60])),
+			BPF_EXIT_INSN(),
+		},
+		.errstr = "invalid bpf_context access",
+		.result = REJECT,
+		.prog_type = BPF_PROG_TYPE_SCHED_ACT,
+	},
+	{
+		"write skb fields from socket prog",
+		.insns = {
+			BPF_LDX_MEM(BPF_W, BPF_REG_0, BPF_REG_1,
+				    offsetof(struct __sk_buff, cb[4])),
+			BPF_JMP_IMM(BPF_JGE, BPF_REG_0, 0, 1),
+			BPF_LDX_MEM(BPF_W, BPF_REG_0, BPF_REG_1,
+				    offsetof(struct __sk_buff, mark)),
+			BPF_LDX_MEM(BPF_W, BPF_REG_0, BPF_REG_1,
+				    offsetof(struct __sk_buff, tc_index)),
+			BPF_JMP_IMM(BPF_JGE, BPF_REG_0, 0, 1),
+			BPF_STX_MEM(BPF_W, BPF_REG_1, BPF_REG_1,
+				    offsetof(struct __sk_buff, cb[0])),
+			BPF_STX_MEM(BPF_W, BPF_REG_1, BPF_REG_1,
+				    offsetof(struct __sk_buff, cb[2])),
+			BPF_EXIT_INSN(),
+		},
+		.result = ACCEPT,
+	},
+	{
+		"write skb fields from tc_cls_act prog",
+		.insns = {
+			BPF_LDX_MEM(BPF_W, BPF_REG_0, BPF_REG_1,
+				    offsetof(struct __sk_buff, cb[0])),
+			BPF_STX_MEM(BPF_W, BPF_REG_1, BPF_REG_0,
+				    offsetof(struct __sk_buff, mark)),
+			BPF_LDX_MEM(BPF_W, BPF_REG_0, BPF_REG_1,
+				    offsetof(struct __sk_buff, tc_index)),
+			BPF_STX_MEM(BPF_W, BPF_REG_1, BPF_REG_0,
+				    offsetof(struct __sk_buff, tc_index)),
+			BPF_STX_MEM(BPF_W, BPF_REG_1, BPF_REG_0,
+				    offsetof(struct __sk_buff, cb[3])),
+			BPF_EXIT_INSN(),
+		},
+		.result = ACCEPT,
+		.prog_type = BPF_PROG_TYPE_SCHED_CLS,
+	},
 };
 
 static int probe_filter_length(struct bpf_insn *fp)
@@ -775,6 +854,7 @@ static int test(void)
 
 	for (i = 0; i < ARRAY_SIZE(tests); i++) {
 		struct bpf_insn *prog = tests[i].insns;
+		int prog_type = tests[i].prog_type;
 		int prog_len = probe_filter_length(prog);
 		int *fixup = tests[i].fixup;
 		int map_fd = -1;
@@ -789,8 +869,8 @@ static int test(void)
 		}
 		printf("#%d %s ", i, tests[i].descr);
 
-		prog_fd = bpf_prog_load(BPF_PROG_TYPE_SOCKET_FILTER, prog,
-					prog_len * sizeof(struct bpf_insn),
+		prog_fd = bpf_prog_load(prog_type ?: BPF_PROG_TYPE_SOCKET_FILTER,
+					prog, prog_len * sizeof(struct bpf_insn),
 					"GPL", 0);
 
 		if (tests[i].result == ACCEPT) {