bpf: test_sockmap, add options for msg_pop_data() helper

Similar to msg_pull_data and msg_push_data add a set of options to
have msg_pop_data() exercised.
Signed-off-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

tools/testing/selftests/bpf/test_sockmap.c

@@ -79,6 +79,8 @@ int txmsg_start;
 int txmsg_end;
 int txmsg_start_push;
 int txmsg_end_push;
+int txmsg_start_pop;
+int txmsg_pop;
 int txmsg_ingress;
 int txmsg_skb;
 int ktls;
@@ -104,6 +106,8 @@ static const struct option long_options[] = {
 	{"txmsg_end",	required_argument,	NULL, 'e'},
 	{"txmsg_start_push", required_argument,	NULL, 'p'},
 	{"txmsg_end_push",   required_argument,	NULL, 'q'},
+	{"txmsg_start_pop",  required_argument,	NULL, 'w'},
+	{"txmsg_pop",	     required_argument,	NULL, 'x'},
 	{"txmsg_ingress", no_argument,		&txmsg_ingress, 1 },
 	{"txmsg_skb", no_argument,		&txmsg_skb, 1 },
 	{"ktls", no_argument,			&ktls, 1 },
@@ -473,13 +477,27 @@ static int msg_loop(int fd, int iov_count, int iov_length, int cnt,
 		clock_gettime(CLOCK_MONOTONIC, &s->end);
 	} else {
 		int slct, recvp = 0, recv, max_fd = fd;
+		float total_bytes, txmsg_pop_total;
 		int fd_flags = O_NONBLOCK;
 		struct timeval timeout;
-		float total_bytes;
 		fd_set w;
 
 		fcntl(fd, fd_flags);
+		/* Account for pop bytes noting each iteration of apply will
+		 * call msg_pop_data helper so we need to account for this
+		 * by calculating the number of apply iterations. Note user
+		 * of the tool can create cases where no data is sent by
+		 * manipulating pop/push/pull/etc. For example txmsg_apply 1
+		 * with txmsg_pop 1 will try to apply 1B at a time but each
+		 * iteration will then pop 1B so no data will ever be sent.
+		 * This is really only useful for testing edge cases in code
+		 * paths.
+		 */
 		total_bytes = (float)iov_count * (float)iov_length * (float)cnt;
+		txmsg_pop_total = txmsg_pop;
+		if (txmsg_apply)
+			txmsg_pop_total *= (total_bytes / txmsg_apply);
+		total_bytes -= txmsg_pop_total;
 		err = clock_gettime(CLOCK_MONOTONIC, &s->start);
 		if (err < 0)
 			perror("recv start time: ");
@@ -488,7 +506,7 @@ static int msg_loop(int fd, int iov_count, int iov_length, int cnt,
 				timeout.tv_sec = 0;
 				timeout.tv_usec = 300000;
 			} else {
-				timeout.tv_sec = 1;
+				timeout.tv_sec = 3;
 				timeout.tv_usec = 0;
 			}
 
@@ -503,7 +521,7 @@ static int msg_loop(int fd, int iov_count, int iov_length, int cnt,
 				goto out_errno;
 			} else if (!slct) {
 				if (opt->verbose)
-					fprintf(stderr, "unexpected timeout\n");
+					fprintf(stderr, "unexpected timeout: recved %zu/%f pop_total %f\n", s->bytes_recvd, total_bytes, txmsg_pop_total);
 				errno = -EIO;
 				clock_gettime(CLOCK_MONOTONIC, &s->end);
 				goto out_errno;
@@ -619,7 +637,7 @@ static int sendmsg_test(struct sockmap_options *opt)
 			iov_count = 1;
 		err = msg_loop(rx_fd, iov_count, iov_buf,
 			       cnt, &s, false, opt);
-		if (err && opt->verbose)
+		if (opt->verbose)
 			fprintf(stderr,
 				"msg_loop_rx: iov_count %i iov_buf %i cnt %i err %i\n",
 				iov_count, iov_buf, cnt, err);
@@ -931,6 +949,39 @@ static int run_options(struct sockmap_options *options, int cg_fd,  int test)
 			}
 		}
 
+		if (txmsg_start_pop) {
+			i = 4;
+			err = bpf_map_update_elem(map_fd[5],
+						  &i, &txmsg_start_pop, BPF_ANY);
+			if (err) {
+				fprintf(stderr,
+					"ERROR: bpf_map_update_elem %i@%i (txmsg_start_pop):  %d (%s)\n",
+					txmsg_start_pop, i, err, strerror(errno));
+				goto out;
+			}
+		} else {
+			i = 4;
+			bpf_map_update_elem(map_fd[5],
+						  &i, &txmsg_start_pop, BPF_ANY);
+		}
+
+		if (txmsg_pop) {
+			i = 5;
+			err = bpf_map_update_elem(map_fd[5],
+						  &i, &txmsg_pop, BPF_ANY);
+			if (err) {
+				fprintf(stderr,
+					"ERROR: bpf_map_update_elem %i@%i (txmsg_pop):  %d (%s)\n",
+					txmsg_pop, i, err, strerror(errno));
+				goto out;
+			}
+		} else {
+			i = 5;
+			bpf_map_update_elem(map_fd[5],
+					    &i, &txmsg_pop, BPF_ANY);
+
+		}
+
 		if (txmsg_ingress) {
 			int in = BPF_F_INGRESS;
 
@@ -1082,6 +1133,11 @@ static void test_options(char *options)
 		snprintf(tstr, OPTSTRING, "end %d,", txmsg_end);
 		strncat(options, tstr, OPTSTRING);
 	}
+	if (txmsg_start_pop) {
+		snprintf(tstr, OPTSTRING, "pop (%d,%d),",
+			 txmsg_start_pop, txmsg_start_pop + txmsg_pop);
+		strncat(options, tstr, OPTSTRING);
+	}
 	if (txmsg_ingress)
 		strncat(options, "ingress,", OPTSTRING);
 	if (txmsg_skb)
@@ -1264,6 +1320,7 @@ static int test_mixed(int cgrp)
 	txmsg_apply = txmsg_cork = 0;
 	txmsg_start = txmsg_end = 0;
 	txmsg_start_push = txmsg_end_push = 0;
+	txmsg_start_pop = txmsg_pop = 0;
 
 	/* Test small and large iov_count values with pass/redir/apply/cork */
 	txmsg_pass = 1;
@@ -1383,6 +1440,19 @@ static int test_start_end(int cgrp)
 	txmsg_end = 2;
 	txmsg_start_push = 1;
 	txmsg_end_push = 2;
+	txmsg_start_pop = 1;
+	txmsg_pop = 1;
+	err = test_txmsg(cgrp);
+	if (err)
+		goto out;
+
+	/* Cut a byte of pushed data but leave reamining in place */
+	txmsg_start = 1;
+	txmsg_end = 2;
+	txmsg_start_push = 1;
+	txmsg_end_push = 3;
+	txmsg_start_pop = 1;
+	txmsg_pop = 1;
 	err = test_txmsg(cgrp);
 	if (err)
 		goto out;
@@ -1393,6 +1463,9 @@ static int test_start_end(int cgrp)
 	opt.iov_length = 100;
 	txmsg_cork = 1600;
 
+	txmsg_start_pop = 0;
+	txmsg_pop = 0;
+
 	for (i = 99; i <= 1600; i += 500) {
 		txmsg_start = 0;
 		txmsg_end = i;
@@ -1403,6 +1476,17 @@ static int test_start_end(int cgrp)
 			goto out;
 	}
 
+	/* Test pop data in middle of cork */
+	for (i = 99; i <= 1600; i += 500) {
+		txmsg_start_pop = 10;
+		txmsg_pop = i;
+		err = test_exec(cgrp, &opt);
+		if (err)
+			goto out;
+	}
+	txmsg_start_pop = 0;
+	txmsg_pop = 0;
+
 	/* Test start/end with cork but pull data in middle */
 	for (i = 199; i <= 1600; i += 500) {
 		txmsg_start = 100;
@@ -1423,6 +1507,15 @@ static int test_start_end(int cgrp)
 	if (err)
 		goto out;
 
+	/* Test pop with cork pulling last sg entry */
+	txmsg_start_pop = 1500;
+	txmsg_pop = 1600;
+	err = test_exec(cgrp, &opt);
+	if (err)
+		goto out;
+	txmsg_start_pop = 0;
+	txmsg_pop = 0;
+
 	/* Test start/end pull of single byte in last page */
 	txmsg_start = 1111;
 	txmsg_end = 1112;
@@ -1432,6 +1525,13 @@ static int test_start_end(int cgrp)
 	if (err)
 		goto out;
 
+	/* Test pop of single byte in last page */
+	txmsg_start_pop = 1111;
+	txmsg_pop = 1112;
+	err = test_exec(cgrp, &opt);
+	if (err)
+		goto out;
+
 	/* Test start/end with end < start */
 	txmsg_start = 1111;
 	txmsg_end = 0;
@@ -1456,7 +1556,20 @@ static int test_start_end(int cgrp)
 	txmsg_start_push = 1601;
 	txmsg_end_push = 1600;
 	err = test_exec(cgrp, &opt);
+	if (err)
+		goto out;
+
+	/* Test pop with start > data */
+	txmsg_start_pop = 1601;
+	txmsg_pop = 1;
+	err = test_exec(cgrp, &opt);
+	if (err)
+		goto out;
 
+	/* Test pop with pop range > data */
+	txmsg_start_pop = 1599;
+	txmsg_pop = 10;
+	err = test_exec(cgrp, &opt);
 out:
 	txmsg_start = 0;
 	txmsg_end = 0;
@@ -1641,6 +1754,12 @@ int main(int argc, char **argv)
 		case 'q':
 			txmsg_end_push = atoi(optarg);
 			break;
+		case 'w':
+			txmsg_start_pop = atoi(optarg);
+			break;
+		case 'x':
+			txmsg_pop = atoi(optarg);
+			break;
 		case 'a':
 			txmsg_apply = atoi(optarg);
 			break;

tools/testing/selftests/bpf/test_sockmap_kern.h

@@ -74,7 +74,7 @@ struct bpf_map_def SEC("maps") sock_bytes = {
 	.type = BPF_MAP_TYPE_ARRAY,
 	.key_size = sizeof(int),
 	.value_size = sizeof(int),
-	.max_entries = 4
+	.max_entries = 6
 };
 
 struct bpf_map_def SEC("maps") sock_redir_flags = {
@@ -181,8 +181,8 @@ int bpf_sockmap(struct bpf_sock_ops *skops)
 SEC("sk_msg1")
 int bpf_prog4(struct sk_msg_md *msg)
 {
-	int *bytes, zero = 0, one = 1, two = 2, three = 3;
-	int *start, *end, *start_push, *end_push;
+	int *bytes, zero = 0, one = 1, two = 2, three = 3, four = 4, five = 5;
+	int *start, *end, *start_push, *end_push, *start_pop, *pop;
 
 	bytes = bpf_map_lookup_elem(&sock_apply_bytes, &zero);
 	if (bytes)
@@ -198,15 +198,19 @@ int bpf_prog4(struct sk_msg_md *msg)
 	end_push = bpf_map_lookup_elem(&sock_bytes, &three);
 	if (start_push && end_push)
 		bpf_msg_push_data(msg, *start_push, *end_push, 0);
+	start_pop = bpf_map_lookup_elem(&sock_bytes, &four);
+	pop = bpf_map_lookup_elem(&sock_bytes, &five);
+	if (start_pop && pop)
+		bpf_msg_pop_data(msg, *start_pop, *pop, 0);
 	return SK_PASS;
 }
 
 SEC("sk_msg2")
 int bpf_prog5(struct sk_msg_md *msg)
 {
-	int zero = 0, one = 1, two = 2, three = 3;
-	int *start, *end, *start_push, *end_push;
-	int *bytes, len1, len2 = 0, len3;
+	int zero = 0, one = 1, two = 2, three = 3, four = 4, five = 5;
+	int *start, *end, *start_push, *end_push, *start_pop, *pop;
+	int *bytes, len1, len2 = 0, len3, len4;
 	int err1 = -1, err2 = -1;
 
 	bytes = bpf_map_lookup_elem(&sock_apply_bytes, &zero);
@@ -247,6 +251,20 @@ int bpf_prog5(struct sk_msg_md *msg)
 		bpf_printk("sk_msg2: length push_update %i->%i\n",
 			   len2 ? len2 : len1, len3);
 	}
+	start_pop = bpf_map_lookup_elem(&sock_bytes, &four);
+	pop = bpf_map_lookup_elem(&sock_bytes, &five);
+	if (start_pop && pop) {
+		int err;
+
+		bpf_printk("sk_msg2: pop(%i@%i)\n",
+			   start_pop, pop);
+		err = bpf_msg_pop_data(msg, *start_pop, *pop, 0);
+		if (err)
+			bpf_printk("sk_msg2: pop_data err %i\n", err);
+		len4 = (__u64)msg->data_end - (__u64)msg->data;
+		bpf_printk("sk_msg2: length pop_data %i->%i\n",
+			   len1 ? len1 : 0,  len4);
+	}
 
 	bpf_printk("sk_msg2: data length %i err1 %i err2 %i\n",
 		   len1, err1, err2);
@@ -256,8 +274,8 @@ int bpf_prog5(struct sk_msg_md *msg)
 SEC("sk_msg3")
 int bpf_prog6(struct sk_msg_md *msg)
 {
-	int *bytes, *start, *end, *start_push, *end_push, *f;
-	int zero = 0, one = 1, two = 2, three = 3, key = 0;
+	int zero = 0, one = 1, two = 2, three = 3, four = 4, five = 5, key = 0;
+	int *bytes, *start, *end, *start_push, *end_push, *start_pop, *pop, *f;
 	__u64 flags = 0;
 
 	bytes = bpf_map_lookup_elem(&sock_apply_bytes, &zero);
@@ -277,6 +295,11 @@ int bpf_prog6(struct sk_msg_md *msg)
 	if (start_push && end_push)
 		bpf_msg_push_data(msg, *start_push, *end_push, 0);
 
+	start_pop = bpf_map_lookup_elem(&sock_bytes, &four);
+	pop = bpf_map_lookup_elem(&sock_bytes, &five);
+	if (start_pop && pop)
+		bpf_msg_pop_data(msg, *start_pop, *pop, 0);
+
 	f = bpf_map_lookup_elem(&sock_redir_flags, &zero);
 	if (f && *f) {
 		key = 2;
@@ -292,8 +315,9 @@ int bpf_prog6(struct sk_msg_md *msg)
 SEC("sk_msg4")
 int bpf_prog7(struct sk_msg_md *msg)
 {
-	int zero = 0, one = 1, two = 2, three = 3, len1, len2 = 0, len3;
-	int *bytes, *start, *end, *start_push, *end_push, *f;
+	int *bytes, *start, *end, *start_push, *end_push, *start_pop, *pop, *f;
+	int zero = 0, one = 1, two = 2, three = 3, four = 4, five = 5;
+	int len1, len2 = 0, len3, len4;
 	int err1 = 0, err2 = 0, key = 0;
 	__u64 flags = 0;
 
@@ -335,6 +359,22 @@ int bpf_prog7(struct sk_msg_md *msg)
 			   len2 ? len2 : len1, len3);
 	}
 
+	start_pop = bpf_map_lookup_elem(&sock_bytes, &four);
+	pop = bpf_map_lookup_elem(&sock_bytes, &five);
+	if (start_pop && pop) {
+		int err;
+
+		bpf_printk("sk_msg4: pop(%i@%i)\n",
+			   start_pop, pop);
+		err = bpf_msg_pop_data(msg, *start_pop, *pop, 0);
+		if (err)
+			bpf_printk("sk_msg4: pop_data err %i\n", err);
+		len4 = (__u64)msg->data_end - (__u64)msg->data;
+		bpf_printk("sk_msg4: length pop_data %i->%i\n",
+			   len1 ? len1 : 0,  len4);
+	}
+
+
 	f = bpf_map_lookup_elem(&sock_redir_flags, &zero);
 	if (f && *f) {
 		key = 2;
@@ -389,8 +429,8 @@ int bpf_prog9(struct sk_msg_md *msg)
 SEC("sk_msg7")
 int bpf_prog10(struct sk_msg_md *msg)
 {
-	int *bytes, *start, *end, *start_push, *end_push;
-	int zero = 0, one = 1, two = 2, three = 3;
+	int *bytes, *start, *end, *start_push, *end_push, *start_pop, *pop;
+	int zero = 0, one = 1, two = 2, three = 3, four = 4, five = 5;
 
 	bytes = bpf_map_lookup_elem(&sock_apply_bytes, &zero);
 	if (bytes)
@@ -406,7 +446,11 @@ int bpf_prog10(struct sk_msg_md *msg)
 	end_push = bpf_map_lookup_elem(&sock_bytes, &three);
 	if (start_push && end_push)
 		bpf_msg_push_data(msg, *start_push, *end_push, 0);
-
+	start_pop = bpf_map_lookup_elem(&sock_bytes, &four);
+	pop = bpf_map_lookup_elem(&sock_bytes, &five);
+	if (start_pop && pop)
+		bpf_msg_pop_data(msg, *start_pop, *pop, 0);
+	bpf_printk("return sk drop\n");
 	return SK_DROP;
 }