tc, bpf, examples: further bpf_api improvements

Add a couple of improvements to tc's BPF api, that facilitate program
development.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

examples/bpf/bpf_cyclic.c

@@ -11,9 +11,7 @@ BPF_PROG_ARRAY(jmp_tc, JMP_MAP_ID, PIN_OBJECT_NS, 1);
 __section_tail(JMP_MAP_ID, 0)
 int cls_loop(struct __sk_buff *skb)
 {
-	char fmt[] = "cb: %u\n";
-
-	trace_printk(fmt, sizeof(fmt), skb->cb[0]++);
+	printt("cb: %u\n", skb->cb[0]++);
 	tail_call(skb, &jmp_tc, 0);
 
 	skb->tc_classid = TC_H_MAKE(1, 42);

examples/bpf/bpf_graft.c

@@ -38,29 +38,22 @@ BPF_PROG_ARRAY(jmp_tc, 0, PIN_GLOBAL_NS, 1);
 __section("aaa")
 int cls_aaa(struct __sk_buff *skb)
 {
-	char fmt[] = "aaa\n";
-
-	trace_printk(fmt, sizeof(fmt));
+	printt("aaa\n");
 	return TC_H_MAKE(1, 42);
 }
 
 __section("bbb")
 int cls_bbb(struct __sk_buff *skb)
 {
-	char fmt[] = "bbb\n";
-
-	trace_printk(fmt, sizeof(fmt));
+	printt("bbb\n");
 	return TC_H_MAKE(1, 43);
 }
 
 __section_cls_entry
 int cls_entry(struct __sk_buff *skb)
 {
-	char fmt[] = "fallthrough\n";
-
 	tail_call(skb, &jmp_tc, 0);
-	trace_printk(fmt, sizeof(fmt));
-
+	printt("fallthrough\n");
 	return BPF_H_DEFAULT;
 }
 

examples/bpf/bpf_prog.c

@@ -233,7 +233,7 @@ struct flow_keys {
 	__u8 ip_proto;
 };
 
-static inline int flow_ports_offset(__u8 ip_proto)
+static __inline__ int flow_ports_offset(__u8 ip_proto)
 {
 	switch (ip_proto) {
 	case IPPROTO_TCP:
@@ -249,13 +249,13 @@ static inline int flow_ports_offset(__u8 ip_proto)
 	}
 }
 
-static inline bool flow_is_frag(struct __sk_buff *skb, int nh_off)
+static __inline__ bool flow_is_frag(struct __sk_buff *skb, int nh_off)
 {
 	return !!(load_half(skb, nh_off + offsetof(struct iphdr, frag_off)) &
 		  (IP_MF | IP_OFFSET));
 }
 
-static inline int flow_parse_ipv4(struct __sk_buff *skb, int nh_off,
+static __inline__ int flow_parse_ipv4(struct __sk_buff *skb, int nh_off,
 				      __u8 *ip_proto, struct flow_keys *flow)
 {
 	__u8 ip_ver_len;
@@ -279,7 +279,7 @@ static inline int flow_parse_ipv4(struct __sk_buff *skb, int nh_off,
 	return nh_off;
 }
 
-static inline __u32 flow_addr_hash_ipv6(struct __sk_buff *skb, int off)
+static __inline__ __u32 flow_addr_hash_ipv6(struct __sk_buff *skb, int off)
 {
 	__u32 w0 = load_word(skb, off);
 	__u32 w1 = load_word(skb, off + sizeof(w0));
@@ -289,7 +289,7 @@ static inline __u32 flow_addr_hash_ipv6(struct __sk_buff *skb, int off)
 	return w0 ^ w1 ^ w2 ^ w3;
 }
 
-static inline int flow_parse_ipv6(struct __sk_buff *skb, int nh_off,
+static __inline__ int flow_parse_ipv6(struct __sk_buff *skb, int nh_off,
 				      __u8 *ip_proto, struct flow_keys *flow)
 {
 	*ip_proto = load_byte(skb, nh_off + offsetof(struct ipv6hdr, nexthdr));
@@ -300,7 +300,7 @@ static inline int flow_parse_ipv6(struct __sk_buff *skb, int nh_off,
 	return nh_off + sizeof(struct ipv6hdr);
 }
 
-static inline bool flow_dissector(struct __sk_buff *skb,
+static __inline__ bool flow_dissector(struct __sk_buff *skb,
 				      struct flow_keys *flow)
 {
 	int poff, nh_off = BPF_LL_OFF + ETH_HLEN;
@@ -381,7 +381,7 @@ static inline bool flow_dissector(struct __sk_buff *skb,
 	return true;
 }
 
-static inline void cls_update_proto_map(const struct __sk_buff *skb,
+static __inline__ void cls_update_proto_map(const struct __sk_buff *skb,
 					    const struct flow_keys *flow)
 {
 	uint8_t proto = flow->ip_proto;
@@ -401,7 +401,7 @@ static inline void cls_update_proto_map(const struct __sk_buff *skb,
 	map_update_elem(&map_proto, &proto, &_ct, BPF_ANY);
 }
 
-static inline void cls_update_queue_map(const struct __sk_buff *skb)
+static __inline__ void cls_update_queue_map(const struct __sk_buff *skb)
 {
 	uint32_t queue = skb->queue_mapping;
 	struct count_queue *cq, _cq;
@@ -453,7 +453,7 @@ int cls_main(struct __sk_buff *skb)
 	return flow.ip_proto;
 }
 
-static inline void act_update_drop_map(void)
+static __inline__ void act_update_drop_map(void)
 {
 	uint32_t *count, cpu = get_smp_processor_id();
 

examples/bpf/bpf_shared.c

@@ -35,12 +35,11 @@ int emain(struct __sk_buff *skb)
 __section("ingress")
 int imain(struct __sk_buff *skb)
 {
-	char fmt[] = "map val: %d\n";
 	int key = 0, *val;
 
 	val = map_lookup_elem(&map_sh, &key);
 	if (val)
-		trace_printk(fmt, sizeof(fmt), *val);
+		printt("map val: %d\n", *val);
 
 	return BPF_H_DEFAULT;
 }

examples/bpf/bpf_tailcall.c

@@ -34,12 +34,11 @@ BPF_ARRAY4(map_sh, 0, PIN_OBJECT_NS, 1);
 __section_tail(FOO, ENTRY_0)
 int cls_case1(struct __sk_buff *skb)
 {
-	char fmt[] = "case1: map-val: %d from:%u\n";
 	int key = 0, *val;
 
 	val = map_lookup_elem(&map_sh, &key);
 	if (val)
-		trace_printk(fmt, sizeof(fmt), *val, skb->cb[0]);
+		printt("case1: map-val: %d from:%u\n", *val, skb->cb[0]);
 
 	skb->cb[0] = ENTRY_0;
 	tail_call(skb, &jmp_ex, ENTRY_0);
@@ -50,12 +49,11 @@ int cls_case1(struct __sk_buff *skb)
 __section_tail(FOO, ENTRY_1)
 int cls_case2(struct __sk_buff *skb)
 {
-	char fmt[] = "case2: map-val: %d from:%u\n";
 	int key = 0, *val;
 
 	val = map_lookup_elem(&map_sh, &key);
 	if (val)
-		trace_printk(fmt, sizeof(fmt), *val, skb->cb[0]);
+		printt("case2: map-val: %d from:%u\n", *val, skb->cb[0]);
 
 	skb->cb[0] = ENTRY_1;
 	tail_call(skb, &jmp_tc, ENTRY_0);
@@ -66,12 +64,11 @@ int cls_case2(struct __sk_buff *skb)
 __section_tail(BAR, ENTRY_0)
 int cls_exit(struct __sk_buff *skb)
 {
-	char fmt[] = "exit: map-val: %d from:%u\n";
 	int key = 0, *val;
 
 	val = map_lookup_elem(&map_sh, &key);
 	if (val)
-		trace_printk(fmt, sizeof(fmt), *val, skb->cb[0]);
+		printt("exit: map-val: %d from:%u\n", *val, skb->cb[0]);
 
 	/* Termination point. */
 	return BPF_H_DEFAULT;
@@ -80,7 +77,6 @@ int cls_exit(struct __sk_buff *skb)
 __section_cls_entry
 int cls_entry(struct __sk_buff *skb)
 {
-	char fmt[] = "fallthrough\n";
 	int key = 0, *val;
 
 	/* For transferring state, we can use skb->cb[0] ... skb->cb[4]. */
@@ -92,7 +88,7 @@ int cls_entry(struct __sk_buff *skb)
 		tail_call(skb, &jmp_tc, skb->hash & (MAX_JMP_SIZE - 1));
 	}
 
-	trace_printk(fmt, sizeof(fmt));
+	printt("fallthrough\n");
 	return BPF_H_DEFAULT;
 }
 

include/bpf_api.h

@@ -56,6 +56,10 @@
 # define ntohl(X)		__constant_ntohl((X))
 #endif
 
+#ifndef __inline__
+# define __inline__		__attribute__((always_inline))
+#endif
+
 /** Section helper macros. */
 
 #ifndef __section
@@ -146,7 +150,7 @@
 # define BPF_H_DEFAULT	-1
 #endif
 
-/** BPF helper functions for tc. */
+/** BPF helper functions for tc. Individual flags are in linux/bpf.h */
 
 #ifndef BPF_FUNC
 # define BPF_FUNC(NAME, ...)						\
@@ -163,8 +167,22 @@ static int BPF_FUNC(map_delete_elem, void *map, const void *key);
 static uint64_t BPF_FUNC(ktime_get_ns);
 
 /* Debugging */
+
+/* FIXME: __attribute__ ((format(printf, 1, 3))) not possible unless
+ * llvm bug https://llvm.org/bugs/show_bug.cgi?id=26243 gets resolved.
+ * It would require ____fmt to be made const, which generates a reloc
+ * entry (non-map).
+ */
 static void BPF_FUNC(trace_printk, const char *fmt, int fmt_size, ...);
 
+#ifndef printt
+# define printt(fmt, ...)						\
+	({								\
+		char ____fmt[] = fmt;					\
+		trace_printk(____fmt, sizeof(____fmt), ##__VA_ARGS__);	\
+	})
+#endif
+
 /* Random numbers */
 static uint32_t BPF_FUNC(get_prandom_u32);
 
@@ -185,12 +203,11 @@ static int BPF_FUNC(clone_redirect, struct __sk_buff *skb, int ifindex,
 		    uint32_t flags);
 
 /* Packet manipulation */
-#define BPF_PSEUDO_HDR			0x10
-#define BPF_HAS_PSEUDO_HDR(flags)	((flags) & BPF_PSEUDO_HDR)
-#define BPF_HDR_FIELD_SIZE(flags)	((flags) & 0x0f)
-
+static int BPF_FUNC(skb_load_bytes, struct __sk_buff *skb, uint32_t off,
+		    void *to, uint32_t len);
 static int BPF_FUNC(skb_store_bytes, struct __sk_buff *skb, uint32_t off,
-		    void *from, uint32_t len, uint32_t flags);
+		    const void *from, uint32_t len, uint32_t flags);
+
 static int BPF_FUNC(l3_csum_replace, struct __sk_buff *skb, uint32_t off,
 		    uint32_t from, uint32_t to, uint32_t flags);
 static int BPF_FUNC(l4_csum_replace, struct __sk_buff *skb, uint32_t off,
@@ -205,14 +222,37 @@ static int BPF_FUNC(skb_vlan_pop, struct __sk_buff *skb);
 static int BPF_FUNC(skb_get_tunnel_key, struct __sk_buff *skb,
 		    struct bpf_tunnel_key *to, uint32_t size, uint32_t flags);
 static int BPF_FUNC(skb_set_tunnel_key, struct __sk_buff *skb,
-		    struct bpf_tunnel_key *from, uint32_t size, uint32_t flags);
+		    const struct bpf_tunnel_key *from, uint32_t size,
+		    uint32_t flags);
 
-/** LLVM built-ins */
+/** LLVM built-ins, mem*() routines work for constant size */
 
 #ifndef lock_xadd
 # define lock_xadd(ptr, val)	((void) __sync_fetch_and_add(ptr, val))
 #endif
 
+#ifndef memset
+# define memset(s, c, n)	__builtin_memset((s), (c), (n))
+#endif
+
+#ifndef memcpy
+# define memcpy(d, s, n)	__builtin_memcpy((d), (s), (n))
+#endif
+
+#ifndef memmove
+# define memmove(d, s, n)	__builtin_memmove((d), (s), (n))
+#endif
+
+/* FIXME: __builtin_memcmp() is not yet fully useable unless llvm bug
+ * https://llvm.org/bugs/show_bug.cgi?id=26218 gets resolved. Also
+ * this one would generate a reloc entry (non-map), otherwise.
+ */
+#if 0
+#ifndef memcmp
+# define memcmp(a, b, n)	__builtin_memcmp((a), (b), (n))
+#endif
+#endif
+
 unsigned long long load_byte(void *skb, unsigned long long off)
 	asm ("llvm.bpf.load.byte");