selftests: drv-net: Test tail-adjustment support

Add test to validate support for the two cases of tail adjustment: 1)
tail extension, and 2) tail shrinking across different frame sizes and
offset values. For each of the two cases, test both the single and
multi-buffer cases by choosing appropriate packet size.

The negative offset value result in growing of tailroom (shrinking of
payload) while the positive offset result in shrinking of tailroom
(growing of payload).

Since the support for tail adjustment varies across drivers, classify the
test as pass if at least one combination of packet size and offset from a
pre-selected list results in a successful run. In case of an unsuccessful
run, report the failure and highlight the packet size and offset values
that caused the test to fail, as well as the values that resulted in the
last successful run.

Note: The growing part of this test for netdevsim may appear flaky when
the offset value is larger than 1. This behavior occurs because tailroom
is not explicitly reserved for netdevsim, with 1 being the typical
tailroom value. However, in certain cases, such as payload being the last
in the page with additional available space, the truesize is expanded.
This also result increases the tailroom causing the test to pass
intermittently. In contrast, when tailrrom is explicitly reserved, such
as in the of fbnic, the test results are deterministic.

./drivers/net/xdp.py
TAP version 13
1..7
ok 1 xdp.test_xdp_native_pass_sb
ok 2 xdp.test_xdp_native_pass_mb
ok 3 xdp.test_xdp_native_drop_sb
ok 4 xdp.test_xdp_native_drop_mb
ok 5 xdp.test_xdp_native_tx_mb
\# Failed run: ... successful run: ... offset 1. Reason: Adjustment failed
ok 6 xdp.test_xdp_native_adjst_tail_grow_data
ok 7 xdp.test_xdp_native_adjst_tail_shrnk_data
\# Totals: pass:7 fail:0 xfail:0 xpass:0 skip:0 error:0

Signed-off-by: Mohsin Bashir <mohsin.bashr@gmail.com>
Link: https://patch.msgid.link/20250719083059.3209169-5-mohsin.bashr@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

1 files changed, 204 insertions, 2 deletions

diff --git a/tools/testing/selftests/net/lib/xdp_native.bpf.c b/tools/testing/selftests/net/lib/xdp_native.bpf.c
index d3c66c891589..7bb2dff86d8f 100644
--- a/tools/testing/selftests/net/lib/xdp_native.bpf.c
+++ b/tools/testing/selftests/net/lib/xdp_native.bpf.c
@@ -10,15 +10,22 @@
 #include <bpf/bpf_endian.h>
 #include <bpf/bpf_helpers.h>
 
+#define MAX_ADJST_OFFSET 256
+#define MAX_PAYLOAD_LEN 5000
+#define MAX_HDR_LEN 64
+
 enum {
 	XDP_MODE = 0,
 	XDP_PORT = 1,
+	XDP_ADJST_OFFSET = 2,
+	XDP_ADJST_TAG = 3,
 } xdp_map_setup_keys;
 
 enum {
 	XDP_MODE_PASS = 0,
 	XDP_MODE_DROP = 1,
 	XDP_MODE_TX = 2,
+	XDP_MODE_TAIL_ADJST = 3,
 } xdp_map_modes;
 
 enum {
@@ -26,22 +33,28 @@ enum {
 	STATS_PASS = 1,
 	STATS_DROP = 2,
 	STATS_TX = 3,
+	STATS_ABORT = 4,
 } xdp_stats;
 
 struct {
 	__uint(type, BPF_MAP_TYPE_ARRAY);
-	__uint(max_entries, 2);
+	__uint(max_entries, 5);
 	__type(key, __u32);
 	__type(value, __s32);
 } map_xdp_setup SEC(".maps");
 
 struct {
 	__uint(type, BPF_MAP_TYPE_ARRAY);
-	__uint(max_entries, 4);
+	__uint(max_entries, 5);
 	__type(key, __u32);
 	__type(value, __u64);
 } map_xdp_stats SEC(".maps");
 
+static __u32 min(__u32 a, __u32 b)
+{
+	return a < b ? a : b;
+}
+
 static void record_stats(struct xdp_md *ctx, __u32 stat_type)
 {
 	__u64 *count;
@@ -195,6 +208,193 @@ static int xdp_mode_tx_handler(struct xdp_md *ctx, __u16 port)
 	return XDP_PASS;
 }
 
+static void *update_pkt(struct xdp_md *ctx, __s16 offset, __u32 *udp_csum)
+{
+	void *data_end = (void *)(long)ctx->data_end;
+	void *data = (void *)(long)ctx->data;
+	struct udphdr *udph = NULL;
+	struct ethhdr *eth = data;
+	__u32 len, len_new;
+
+	if (data + sizeof(*eth) > data_end)
+		return NULL;
+
+	if (eth->h_proto == bpf_htons(ETH_P_IP)) {
+		struct iphdr *iph = data + sizeof(*eth);
+		__u16 total_len;
+
+		if (iph + 1 > (struct iphdr *)data_end)
+			return NULL;
+
+		iph->tot_len = bpf_htons(bpf_ntohs(iph->tot_len) + offset);
+
+		udph = (void *)eth + sizeof(*iph) + sizeof(*eth);
+		if (!udph || udph + 1 > (struct udphdr *)data_end)
+			return NULL;
+
+		len_new = bpf_htons(bpf_ntohs(udph->len) + offset);
+	} else if (eth->h_proto  == bpf_htons(ETH_P_IPV6)) {
+		struct ipv6hdr *ipv6h = data + sizeof(*eth);
+		__u16 payload_len;
+
+		if (ipv6h + 1 > (struct ipv6hdr *)data_end)
+			return NULL;
+
+		udph = (void *)eth + sizeof(*ipv6h) + sizeof(*eth);
+		if (!udph || udph + 1 > (struct udphdr *)data_end)
+			return NULL;
+
+		*udp_csum = ~((__u32)udph->check);
+
+		len = ipv6h->payload_len;
+		len_new = bpf_htons(bpf_ntohs(len) + offset);
+		ipv6h->payload_len = len_new;
+
+		*udp_csum = bpf_csum_diff(&len, sizeof(len), &len_new,
+					  sizeof(len_new), *udp_csum);
+
+		len = udph->len;
+		len_new = bpf_htons(bpf_ntohs(udph->len) + offset);
+		*udp_csum = bpf_csum_diff(&len, sizeof(len), &len_new,
+					  sizeof(len_new), *udp_csum);
+	} else {
+		return NULL;
+	}
+
+	udph->len = len_new;
+
+	return udph;
+}
+
+static __u16 csum_fold_helper(__u32 csum)
+{
+	return ~((csum & 0xffff) + (csum >> 16)) ? : 0xffff;
+}
+
+static int xdp_adjst_tail_shrnk_data(struct xdp_md *ctx, __u16 offset,
+				     __u32 hdr_len)
+{
+	char tmp_buff[MAX_ADJST_OFFSET];
+	__u32 buff_pos, udp_csum = 0;
+	struct udphdr *udph = NULL;
+	__u32 buff_len;
+
+	udph = update_pkt(ctx, 0 - offset, &udp_csum);
+	if (!udph)
+		return -1;
+
+	buff_len = bpf_xdp_get_buff_len(ctx);
+
+	offset = (offset & 0x1ff) >= MAX_ADJST_OFFSET ? MAX_ADJST_OFFSET :
+				     offset & 0xff;
+	if (offset == 0)
+		return -1;
+
+	/* Make sure we have enough data to avoid eating the header */
+	if (buff_len - offset < hdr_len)
+		return -1;
+
+	buff_pos = buff_len - offset;
+	if (bpf_xdp_load_bytes(ctx, buff_pos, tmp_buff, offset) < 0)
+		return -1;
+
+	udp_csum = bpf_csum_diff((__be32 *)tmp_buff, offset, 0, 0, udp_csum);
+	udph->check = (__u16)csum_fold_helper(udp_csum);
+
+	if (bpf_xdp_adjust_tail(ctx, 0 - offset) < 0)
+		return -1;
+
+	return 0;
+}
+
+static int xdp_adjst_tail_grow_data(struct xdp_md *ctx, __u16 offset)
+{
+	char tmp_buff[MAX_ADJST_OFFSET];
+	__u32 buff_pos, udp_csum = 0;
+	__u32 buff_len, hdr_len, key;
+	struct udphdr *udph;
+	__s32 *val;
+	__u8 tag;
+
+	/* Proceed to update the packet headers before attempting to adjuste
+	 * the tail. Once the tail is adjusted we lose access to the offset
+	 * amount of data at the end of the packet which is crucial to update
+	 * the checksum.
+	 * Since any failure beyond this would abort the packet, we should
+	 * not worry about passing a packet up the stack with wrong headers
+	 */
+	udph = update_pkt(ctx, offset, &udp_csum);
+	if (!udph)
+		return -1;
+
+	key = XDP_ADJST_TAG;
+	val = bpf_map_lookup_elem(&map_xdp_setup, &key);
+	if (!val)
+		return -1;
+
+	tag = (__u8)(*val);
+
+	for (int i = 0; i < MAX_ADJST_OFFSET; i++)
+		__builtin_memcpy(&tmp_buff[i], &tag, 1);
+
+	offset = (offset & 0x1ff) >= MAX_ADJST_OFFSET ? MAX_ADJST_OFFSET :
+				     offset & 0xff;
+	if (offset == 0)
+		return -1;
+
+	udp_csum = bpf_csum_diff(0, 0, (__be32 *)tmp_buff, offset, udp_csum);
+	udph->check = (__u16)csum_fold_helper(udp_csum);
+
+	buff_len = bpf_xdp_get_buff_len(ctx);
+
+	if (bpf_xdp_adjust_tail(ctx, offset) < 0) {
+		bpf_printk("Failed to adjust tail\n");
+		return -1;
+	}
+
+	if (bpf_xdp_store_bytes(ctx, buff_len, tmp_buff, offset) < 0)
+		return -1;
+
+	return 0;
+}
+
+static int xdp_adjst_tail(struct xdp_md *ctx, __u16 port)
+{
+	void *data = (void *)(long)ctx->data;
+	struct udphdr *udph = NULL;
+	__s32 *adjust_offset, *val;
+	__u32 key, hdr_len;
+	void *offset_ptr;
+	__u8 tag;
+	int ret;
+
+	udph = filter_udphdr(ctx, port);
+	if (!udph)
+		return XDP_PASS;
+
+	hdr_len = (void *)udph - data + sizeof(struct udphdr);
+	key = XDP_ADJST_OFFSET;
+	adjust_offset = bpf_map_lookup_elem(&map_xdp_setup, &key);
+	if (!adjust_offset)
+		return XDP_PASS;
+
+	if (*adjust_offset < 0)
+		ret = xdp_adjst_tail_shrnk_data(ctx,
+						(__u16)(0 - *adjust_offset),
+						hdr_len);
+	else
+		ret = xdp_adjst_tail_grow_data(ctx, (__u16)(*adjust_offset));
+	if (ret)
+		goto abort_pkt;
+
+	record_stats(ctx, STATS_PASS);
+	return XDP_PASS;
+
+abort_pkt:
+	record_stats(ctx, STATS_ABORT);
+	return XDP_ABORTED;
+}
+
 static int xdp_prog_common(struct xdp_md *ctx)
 {
 	__u32 key, *port;
@@ -217,6 +417,8 @@ static int xdp_prog_common(struct xdp_md *ctx)
 		return xdp_mode_drop_handler(ctx, (__u16)(*port));
 	case XDP_MODE_TX:
 		return xdp_mode_tx_handler(ctx, (__u16)(*port));
+	case XDP_MODE_TAIL_ADJST:
+		return xdp_adjst_tail(ctx, (__u16)(*port));
 	}
 
 	/* Default action is to simple pass */