UPSTREAM: bpf: Count the number of times recursion was prevented

Add per-program counter for number of times recursion prevention mechanism
was triggered and expose it via show_fdinfo and bpf_prog_info.
Teach bpftool to print it.

Change-Id: Ie47d3a1cf100d4c46be8b680be1a0bebf2044d15
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20210210033634.62081-7-alexei.starovoitov@gmail.com

include/linux/filter.h

@@ -587,6 +587,7 @@ struct bpf_binary_header {
 struct bpf_prog_stats {
 	u64 cnt;
 	u64 nsecs;
+	u64 misses;
 	struct u64_stats_sync syncp;
 } __aligned(2 * sizeof(u64));
 

include/uapi/linux/bpf.h

@@ -4510,6 +4510,7 @@ struct bpf_prog_info {
 	__aligned_u64 prog_tags;
 	__u64 run_time_ns;
 	__u64 run_cnt;
+	__u64 recursion_misses;
 } __attribute__((aligned(8)));
 
 struct bpf_map_info {

kernel/bpf/syscall.c

@@ -1755,25 +1755,28 @@ static int bpf_prog_release(struct inode *inode, struct file *filp)
 static void bpf_prog_get_stats(const struct bpf_prog *prog,
 			       struct bpf_prog_stats *stats)
 {
-	u64 nsecs = 0, cnt = 0;
+	u64 nsecs = 0, cnt = 0, misses = 0;
 	int cpu;
 
 	for_each_possible_cpu(cpu) {
 		const struct bpf_prog_stats *st;
 		unsigned int start;
-		u64 tnsecs, tcnt;
+		u64 tnsecs, tcnt, tmisses;
 
 		st = per_cpu_ptr(prog->stats, cpu);
 		do {
 			start = u64_stats_fetch_begin_irq(&st->syncp);
 			tnsecs = st->nsecs;
 			tcnt = st->cnt;
+			tmisses = st->misses;
 		} while (u64_stats_fetch_retry_irq(&st->syncp, start));
 		nsecs += tnsecs;
 		cnt += tcnt;
+		misses += tmisses;
 	}
 	stats->nsecs = nsecs;
 	stats->cnt = cnt;
+	stats->misses = misses;
 }
 
 #ifdef CONFIG_PROC_FS
@@ -1792,14 +1795,16 @@ static void bpf_prog_show_fdinfo(struct seq_file *m, struct file *filp)
 		   "memlock:\t%llu\n"
 		   "prog_id:\t%u\n"
 		   "run_time_ns:\t%llu\n"
-		   "run_cnt:\t%llu\n",
+		   "run_cnt:\t%llu\n"
+		   "recursion_misses:\t%llu\n",
 		   prog->type,
 		   prog->jited,
 		   prog_tag,
 		   prog->pages * 1ULL << PAGE_SHIFT,
 		   prog->aux->id,
 		   stats.nsecs,
-		   stats.cnt);
+		   stats.cnt,
+		   stats.misses);
 }
 #endif
 
@@ -3467,6 +3472,7 @@ static int bpf_prog_get_info_by_fd(struct file *file,
 	bpf_prog_get_stats(prog, &stats);
 	info.run_time_ns = stats.nsecs;
 	info.run_cnt = stats.cnt;
+	info.recursion_misses = stats.misses;
 
 	if (!bpf_capable()) {
 		info.jited_prog_len = 0;

kernel/bpf/trampoline.c

@@ -499,6 +499,16 @@ static u64 notrace bpf_prog_start_time(void)
 	return start;
 }
 
+static void notrace inc_misses_counter(struct bpf_prog *prog)
+{
+	struct bpf_prog_stats *stats;
+
+	stats = this_cpu_ptr(prog->stats);
+	u64_stats_update_begin(&stats->syncp);
+	stats->misses++;
+	u64_stats_update_end(&stats->syncp);
+}
+
 /* The logic is similar to BPF_PROG_RUN, but with an explicit
  * rcu_read_lock() and migrate_disable() which are required
  * for the trampoline. The macro is split into
@@ -517,8 +527,10 @@ u64 notrace __bpf_prog_enter(struct bpf_prog *prog)
 {
 	rcu_read_lock();
 	migrate_disable();
-	if (unlikely(__this_cpu_inc_return(*(prog->active)) != 1))
+	if (unlikely(__this_cpu_inc_return(*(prog->active)) != 1)) {
+		inc_misses_counter(prog);
 		return 0;
+	}
 	return bpf_prog_start_time();
 }
 
@@ -556,8 +568,10 @@ u64 notrace __bpf_prog_enter_sleepable(struct bpf_prog *prog)
 	rcu_read_lock_trace();
 	migrate_disable();
 	might_fault();
-	if (unlikely(__this_cpu_inc_return(*(prog->active)) != 1))
+	if (unlikely(__this_cpu_inc_return(*(prog->active)) != 1)) {
+		inc_misses_counter(prog);
 		return 0;
+	}
 	return bpf_prog_start_time();
 }
 

tools/bpf/bpftool/prog.c

@@ -245,6 +245,13 @@ static void print_prog_json(struct bpf_prog_info *info, int fd)
 		     info->tag[4], info->tag[5], info->tag[6], info->tag[7]);
 
 	jsonw_bool_field(json_wtr, "gpl_compatible", info->gpl_compatible);
+	if (info->run_time_ns) {
+		jsonw_uint_field(json_wtr, "run_time_ns", info->run_time_ns);
+		jsonw_uint_field(json_wtr, "run_cnt", info->run_cnt);
+	}
+	if (info->recursion_misses)
+		jsonw_uint_field(json_wtr, "recursion_misses", info->recursion_misses);
+}
 
 	print_dev_json(info->ifindex, info->netns_dev, info->netns_ino);
 
@@ -308,6 +315,11 @@ static void print_prog_plain(struct bpf_prog_info *info, int fd)
 	fprint_hex(stdout, info->tag, BPF_TAG_SIZE, "");
 	print_dev_plain(info->ifindex, info->netns_dev, info->netns_ino);
 	printf("%s", info->gpl_compatible ? "  gpl" : "");
+	if (info->run_time_ns)
+		printf(" run_time_ns %lld run_cnt %lld",
+		       info->run_time_ns, info->run_cnt);
+	if (info->recursion_misses)
+		printf(" recursion_misses %lld", info->recursion_misses);
 	printf("\n");
 
 	if (info->load_time) {

tools/include/uapi/linux/bpf.h

@@ -4435,6 +4435,7 @@ struct bpf_prog_info {
 	__aligned_u64 prog_tags;
 	__u64 run_time_ns;
 	__u64 run_cnt;
+	__u64 recursion_misses;
 } __attribute__((aligned(8)));
 
 struct bpf_map_info {