[PATCH v2 0/2] Adding SO_PEEK_OFF for TCPv6

[PATCH v2 0/2] Adding SO_PEEK_OFF for TCPv6

[Jon Maloy]

Adding SO_PEEK_OFF for TCPv6 and selftest for it.

Jon Maloy (2):
  tcp: add SO_PEEK_OFF socket option tor TCPv6
  selftests: add selftest for tcp SO_PEEK_OFF support

 net/ipv6/af_inet6.c                           |   1 +
 tools/testing/selftests/net/Makefile          |   1 +
 tools/testing/selftests/net/tcp_so_peek_off.c | 181 ++++++++++++++++++
 3 files changed, 183 insertions(+)
 create mode 100644 tools/testing/selftests/net/tcp_so_peek_off.c

-- 
2.45.2

[PATCH v2 1/2] tcp: add SO_PEEK_OFF socket option tor TCPv6

[Jon Maloy]

When we added the SO_PEEK_OFF socket option to TCP we forgot
to add it even for TCP on IPv6.

We do that here.

Fixes: 05ea491641d3 ("tcp: add support for SO_PEEK_OFF socket option")
Signed-off-by: Jon Maloy <jmaloy@redhat.com>
---
 net/ipv6/af_inet6.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
index 90d2c7e3f5e9..ba69b86f1c7d 100644
--- a/net/ipv6/af_inet6.c
+++ b/net/ipv6/af_inet6.c
@@ -708,6 +708,7 @@ const struct proto_ops inet6_stream_ops = {
 	.splice_eof	   = inet_splice_eof,
 	.sendmsg_locked    = tcp_sendmsg_locked,
 	.splice_read	   = tcp_splice_read,
+	.set_peek_off      = sk_set_peek_off,
 	.read_sock	   = tcp_read_sock,
 	.read_skb	   = tcp_read_skb,
 	.peek_len	   = tcp_peek_len,
-- 
2.45.2

Re: [PATCH v2 1/2] tcp: add SO_PEEK_OFF socket option tor TCPv6

[David Gibson]

[-- Attachment #1: Type: text/plain, Size: 1142 bytes --]

On Thu, Aug 22, 2024 at 09:36:36PM -0400, Jon Maloy wrote:
> When we added the SO_PEEK_OFF socket option to TCP we forgot
> to add it even for TCP on IPv6.
> 
> We do that here.
> 
> Fixes: 05ea491641d3 ("tcp: add support for SO_PEEK_OFF socket option")
> Signed-off-by: Jon Maloy <jmaloy@redhat.com>

Reviewed-by: David Gibson <david@gibson.dropbear.id.au>

> ---
>  net/ipv6/af_inet6.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
> index 90d2c7e3f5e9..ba69b86f1c7d 100644
> --- a/net/ipv6/af_inet6.c
> +++ b/net/ipv6/af_inet6.c
> @@ -708,6 +708,7 @@ const struct proto_ops inet6_stream_ops = {
>  	.splice_eof	   = inet_splice_eof,
>  	.sendmsg_locked    = tcp_sendmsg_locked,
>  	.splice_read	   = tcp_splice_read,
> +	.set_peek_off      = sk_set_peek_off,
>  	.read_sock	   = tcp_read_sock,
>  	.read_skb	   = tcp_read_skb,
>  	.peek_len	   = tcp_peek_len,

-- 
David Gibson (he or they)	| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you, not the other way
				| around.
http://www.ozlabs.org/~dgibson

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [PATCH v2 1/2] tcp: add SO_PEEK_OFF socket option tor TCPv6

[Stefano Brivio]

On Thu, 22 Aug 2024 21:36:36 -0400
Jon Maloy <jmaloy@redhat.com> wrote:

> When we added the SO_PEEK_OFF socket option to TCP we forgot
> to add it even for TCP on IPv6.
> 
> We do that here.
> 
> Fixes: 05ea491641d3 ("tcp: add support for SO_PEEK_OFF socket option")
> Signed-off-by: Jon Maloy <jmaloy@redhat.com>

Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Tested-by: Stefano Brivio <sbrivio@redhat.com>

-- 
Stefano

[PATCH v2 2/2] selftests: add selftest for tcp SO_PEEK_OFF support

[Jon Maloy]

We add a selftest to check that the new feature added in
commit 05ea491641d3 ("tcp: add support for SO_PEEK_OFF socket option")
works correctly.

Signed-off-by: Jon Maloy <jmaloy@redhat.com>

---
v2: Fixed Makefile and improved code after suggestion from Stefano Brivio
---
 tools/testing/selftests/net/Makefile          |   1 +
 tools/testing/selftests/net/tcp_so_peek_off.c | 181 ++++++++++++++++++
 2 files changed, 182 insertions(+)
 create mode 100644 tools/testing/selftests/net/tcp_so_peek_off.c

diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile
index 8eaffd7a641c..1179e3261bef 100644
--- a/tools/testing/selftests/net/Makefile
+++ b/tools/testing/selftests/net/Makefile
@@ -80,6 +80,7 @@ TEST_PROGS += io_uring_zerocopy_tx.sh
 TEST_GEN_FILES += bind_bhash
 TEST_GEN_PROGS += sk_bind_sendto_listen
 TEST_GEN_PROGS += sk_connect_zero_addr
+TEST_GEN_PROGS += tcp_so_peek_off
 TEST_PROGS += test_ingress_egress_chaining.sh
 TEST_GEN_PROGS += so_incoming_cpu
 TEST_PROGS += sctp_vrf.sh
diff --git a/tools/testing/selftests/net/tcp_so_peek_off.c b/tools/testing/selftests/net/tcp_so_peek_off.c
new file mode 100644
index 000000000000..8379ea02e3d7
--- /dev/null
+++ b/tools/testing/selftests/net/tcp_so_peek_off.c
@@ -0,0 +1,181 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <errno.h>
+#include <sys/types.h>
+#include <netinet/in.h>
+#include <arpa/inet.h>
+#include "../kselftest.h"
+
+static char *afstr(int af)
+{
+	return af == AF_INET ? "TCP/IPv4" : "TCP/IPv6";
+}
+
+int tcp_peek_offset_probe(sa_family_t af)
+{
+	int optv = 0;
+	int ret = 0;
+	int s;
+
+	s = socket(af, SOCK_STREAM | SOCK_CLOEXEC, IPPROTO_TCP);
+	if (s < 0) {
+		ksft_perror("Temporary TCP socket creation failed");
+	} else {
+		if (!setsockopt(s, SOL_SOCKET, SO_PEEK_OFF, &optv, sizeof(int)))
+			ret = 1;
+		else
+			printf("%s does not support SO_PEEK_OFF\n", afstr(af));
+		close(s);
+	}
+	return ret;
+}
+
+static void tcp_peek_offset_set(int s, int offset)
+{
+	if (setsockopt(s, SOL_SOCKET, SO_PEEK_OFF, &offset, sizeof(offset)))
+		ksft_perror("Failed to set SO_PEEK_OFF value\n");
+}
+
+static int tcp_peek_offset_get(int s)
+{
+	int offset;
+	socklen_t len = sizeof(offset);
+
+	if (getsockopt(s, SOL_SOCKET, SO_PEEK_OFF, &offset, &len))
+		ksft_perror("Failed to get SO_PEEK_OFF value\n");
+	return offset;
+}
+
+static int tcp_peek_offset_test(sa_family_t af)
+{
+	union {
+		struct sockaddr sa;
+		struct sockaddr_in a4;
+		struct sockaddr_in6 a6;
+	} a;
+	int res = 0;
+	int s[2] = {0, 0};
+	int recv_sock = 0;
+	int offset = 0;
+	ssize_t len;
+	char buf;
+
+	memset(&a, 0, sizeof(a));
+	a.sa.sa_family = af;
+
+	s[0] = socket(af, SOCK_STREAM, IPPROTO_TCP);
+	s[1] = socket(af, SOCK_STREAM | SOCK_NONBLOCK, IPPROTO_TCP);
+
+	if (s[0] < 0 || s[1] < 0) {
+		ksft_perror("Temporary probe socket creation failed\n");
+		goto out;
+	}
+	if (bind(s[0], &a.sa, sizeof(a)) < 0) {
+		ksft_perror("Temporary probe socket bind() failed\n");
+		goto out;
+	}
+	if (getsockname(s[0], &a.sa, &((socklen_t) { sizeof(a) })) < 0) {
+		ksft_perror("Temporary probe socket getsockname() failed\n");
+		goto out;
+	}
+	if (listen(s[0], 0) < 0) {
+		ksft_perror("Temporary probe socket listen() failed\n");
+		goto out;
+	}
+	if (connect(s[1], &a.sa, sizeof(a)) >= 0 || errno != EINPROGRESS) {
+		ksft_perror("Temporary probe socket connect() failed\n");
+		goto out;
+	}
+	recv_sock = accept(s[0], NULL, NULL);
+	if (recv_sock <= 0) {
+		ksft_perror("Temporary probe socket accept() failed\n");
+		goto out;
+	}
+
+	/* Some basic tests of getting/setting offset */
+	offset = tcp_peek_offset_get(recv_sock);
+	if (offset != -1) {
+		ksft_perror("Initial value of socket offset not -1\n");
+		goto out;
+	}
+	tcp_peek_offset_set(recv_sock, 0);
+	offset = tcp_peek_offset_get(recv_sock);
+	if (offset != 0) {
+		ksft_perror("Failed to set socket offset to 0\n");
+		goto out;
+	}
+
+	/* Transfer a message */
+	if (send(s[1], (char *)("ab"), 2, 0) <= 0 || errno != EINPROGRESS) {
+		ksft_perror("Temporary probe socket send() failed\n");
+		goto out;
+	}
+	/* Read first byte */
+	len = recv(recv_sock, &buf, 1, MSG_PEEK);
+	if (len != 1 || buf != 'a') {
+		ksft_perror("Failed to read first byte of message\n");
+		goto out;
+	}
+	offset = tcp_peek_offset_get(recv_sock);
+	if (offset != 1) {
+		ksft_perror("Offset not forwarded correctly at first byte\n");
+		goto out;
+	}
+	/* Try to read beyond last byte */
+	len = recv(recv_sock, &buf, 2, MSG_PEEK);
+	if (len != 1 || buf != 'b') {
+		ksft_perror("Failed to read last byte of message\n");
+		goto out;
+	}
+	offset = tcp_peek_offset_get(recv_sock);
+	if (offset != 2) {
+		ksft_perror("Offset not forwarded correctly at last byte\n");
+		goto out;
+	}
+	/* Flush message */
+	len = recv(recv_sock, NULL, 2, MSG_TRUNC);
+	if (len != 2) {
+		ksft_perror("Failed to flush message\n");
+		goto out;
+	}
+	offset = tcp_peek_offset_get(recv_sock);
+	if (offset != 0) {
+		ksft_perror("Offset not reverted correctly after flush\n");
+		goto out;
+	}
+
+	printf("%s with MSG_PEEK_OFF works correctly\n", afstr(af));
+	res = 1;
+out:
+	close(recv_sock);
+	close(s[1]);
+	close(s[0]);
+	return res;
+}
+
+int main(void)
+{
+	int res4, res6;
+
+	res4 = tcp_peek_offset_probe(AF_INET);
+	res6 = tcp_peek_offset_probe(AF_INET6);
+
+	if (!res4 && !res6)
+		return KSFT_SKIP;
+
+	if (res4)
+		res4 = tcp_peek_offset_test(AF_INET);
+
+	if (res6)
+		res6 = tcp_peek_offset_test(AF_INET6);
+
+	if (!res4 || !res6)
+		return KSFT_FAIL;
+
+	return KSFT_PASS;
+}
+
-- 
2.45.2

Re: [PATCH v2 2/2] selftests: add selftest for tcp SO_PEEK_OFF support

[Stefano Brivio]

On Thu, 22 Aug 2024 21:36:37 -0400
Jon Maloy <jmaloy@redhat.com> wrote:

> We add a selftest to check that the new feature added in
> commit 05ea491641d3 ("tcp: add support for SO_PEEK_OFF socket option")
> works correctly.
> 
> Signed-off-by: Jon Maloy <jmaloy@redhat.com>

Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Tested-by: Stefano Brivio <sbrivio@redhat.com>

With or without fixing two nits below:

> ---
> v2: Fixed Makefile and improved code after suggestion from Stefano Brivio
> ---
>  tools/testing/selftests/net/Makefile          |   1 +
>  tools/testing/selftests/net/tcp_so_peek_off.c | 181 ++++++++++++++++++
>  2 files changed, 182 insertions(+)
>  create mode 100644 tools/testing/selftests/net/tcp_so_peek_off.c
> 
> diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile
> index 8eaffd7a641c..1179e3261bef 100644
> --- a/tools/testing/selftests/net/Makefile
> +++ b/tools/testing/selftests/net/Makefile
> @@ -80,6 +80,7 @@ TEST_PROGS += io_uring_zerocopy_tx.sh
>  TEST_GEN_FILES += bind_bhash
>  TEST_GEN_PROGS += sk_bind_sendto_listen
>  TEST_GEN_PROGS += sk_connect_zero_addr
> +TEST_GEN_PROGS += tcp_so_peek_off
>  TEST_PROGS += test_ingress_egress_chaining.sh
>  TEST_GEN_PROGS += so_incoming_cpu
>  TEST_PROGS += sctp_vrf.sh
> diff --git a/tools/testing/selftests/net/tcp_so_peek_off.c b/tools/testing/selftests/net/tcp_so_peek_off.c
> new file mode 100644
> index 000000000000..8379ea02e3d7
> --- /dev/null
> +++ b/tools/testing/selftests/net/tcp_so_peek_off.c
> @@ -0,0 +1,181 @@
> +// SPDX-License-Identifier: GPL-2.0
> +
> +#include <stdio.h>
> +#include <stdlib.h>
> +#include <string.h>
> +#include <unistd.h>
> +#include <errno.h>
> +#include <sys/types.h>
> +#include <netinet/in.h>
> +#include <arpa/inet.h>
> +#include "../kselftest.h"
> +
> +static char *afstr(int af)
> +{
> +	return af == AF_INET ? "TCP/IPv4" : "TCP/IPv6";
> +}
> +
> +int tcp_peek_offset_probe(sa_family_t af)
> +{
> +	int optv = 0;
> +	int ret = 0;
> +	int s;
> +
> +	s = socket(af, SOCK_STREAM | SOCK_CLOEXEC, IPPROTO_TCP);
> +	if (s < 0) {
> +		ksft_perror("Temporary TCP socket creation failed");
> +	} else {
> +		if (!setsockopt(s, SOL_SOCKET, SO_PEEK_OFF, &optv, sizeof(int)))
> +			ret = 1;
> +		else
> +			printf("%s does not support SO_PEEK_OFF\n", afstr(af));

You could use ksft_print_msg() for this one.

But the naming of those functions is pretty redundant (what else do you
want to do with a 'msg' other than printing it, I wonder), so it might
make the code unnecessarily unreadable.

> +		close(s);
> +	}
> +	return ret;
> +}
> +
> +static void tcp_peek_offset_set(int s, int offset)
> +{
> +	if (setsockopt(s, SOL_SOCKET, SO_PEEK_OFF, &offset, sizeof(offset)))
> +		ksft_perror("Failed to set SO_PEEK_OFF value\n");
> +}
> +
> +static int tcp_peek_offset_get(int s)
> +{
> +	int offset;
> +	socklen_t len = sizeof(offset);
> +
> +	if (getsockopt(s, SOL_SOCKET, SO_PEEK_OFF, &offset, &len))
> +		ksft_perror("Failed to get SO_PEEK_OFF value\n");
> +	return offset;
> +}
> +
> +static int tcp_peek_offset_test(sa_family_t af)
> +{
> +	union {
> +		struct sockaddr sa;
> +		struct sockaddr_in a4;
> +		struct sockaddr_in6 a6;
> +	} a;
> +	int res = 0;
> +	int s[2] = {0, 0};
> +	int recv_sock = 0;
> +	int offset = 0;
> +	ssize_t len;
> +	char buf;
> +
> +	memset(&a, 0, sizeof(a));
> +	a.sa.sa_family = af;
> +
> +	s[0] = socket(af, SOCK_STREAM, IPPROTO_TCP);
> +	s[1] = socket(af, SOCK_STREAM | SOCK_NONBLOCK, IPPROTO_TCP);
> +
> +	if (s[0] < 0 || s[1] < 0) {
> +		ksft_perror("Temporary probe socket creation failed\n");
> +		goto out;
> +	}
> +	if (bind(s[0], &a.sa, sizeof(a)) < 0) {
> +		ksft_perror("Temporary probe socket bind() failed\n");
> +		goto out;
> +	}
> +	if (getsockname(s[0], &a.sa, &((socklen_t) { sizeof(a) })) < 0) {
> +		ksft_perror("Temporary probe socket getsockname() failed\n");
> +		goto out;
> +	}
> +	if (listen(s[0], 0) < 0) {
> +		ksft_perror("Temporary probe socket listen() failed\n");
> +		goto out;
> +	}
> +	if (connect(s[1], &a.sa, sizeof(a)) >= 0 || errno != EINPROGRESS) {
> +		ksft_perror("Temporary probe socket connect() failed\n");
> +		goto out;
> +	}
> +	recv_sock = accept(s[0], NULL, NULL);
> +	if (recv_sock <= 0) {
> +		ksft_perror("Temporary probe socket accept() failed\n");
> +		goto out;
> +	}
> +
> +	/* Some basic tests of getting/setting offset */
> +	offset = tcp_peek_offset_get(recv_sock);
> +	if (offset != -1) {
> +		ksft_perror("Initial value of socket offset not -1\n");
> +		goto out;
> +	}
> +	tcp_peek_offset_set(recv_sock, 0);
> +	offset = tcp_peek_offset_get(recv_sock);
> +	if (offset != 0) {
> +		ksft_perror("Failed to set socket offset to 0\n");
> +		goto out;
> +	}
> +
> +	/* Transfer a message */
> +	if (send(s[1], (char *)("ab"), 2, 0) <= 0 || errno != EINPROGRESS) {
> +		ksft_perror("Temporary probe socket send() failed\n");
> +		goto out;
> +	}
> +	/* Read first byte */
> +	len = recv(recv_sock, &buf, 1, MSG_PEEK);
> +	if (len != 1 || buf != 'a') {
> +		ksft_perror("Failed to read first byte of message\n");
> +		goto out;
> +	}
> +	offset = tcp_peek_offset_get(recv_sock);
> +	if (offset != 1) {
> +		ksft_perror("Offset not forwarded correctly at first byte\n");
> +		goto out;
> +	}
> +	/* Try to read beyond last byte */
> +	len = recv(recv_sock, &buf, 2, MSG_PEEK);
> +	if (len != 1 || buf != 'b') {
> +		ksft_perror("Failed to read last byte of message\n");
> +		goto out;
> +	}
> +	offset = tcp_peek_offset_get(recv_sock);
> +	if (offset != 2) {
> +		ksft_perror("Offset not forwarded correctly at last byte\n");
> +		goto out;
> +	}
> +	/* Flush message */
> +	len = recv(recv_sock, NULL, 2, MSG_TRUNC);
> +	if (len != 2) {
> +		ksft_perror("Failed to flush message\n");
> +		goto out;
> +	}
> +	offset = tcp_peek_offset_get(recv_sock);
> +	if (offset != 0) {
> +		ksft_perror("Offset not reverted correctly after flush\n");
> +		goto out;
> +	}
> +
> +	printf("%s with MSG_PEEK_OFF works correctly\n", afstr(af));

Same here.

> +	res = 1;
> +out:
> +	close(recv_sock);
> +	close(s[1]);
> +	close(s[0]);
> +	return res;
> +}
> +
> +int main(void)
> +{
> +	int res4, res6;
> +
> +	res4 = tcp_peek_offset_probe(AF_INET);
> +	res6 = tcp_peek_offset_probe(AF_INET6);
> +
> +	if (!res4 && !res6)
> +		return KSFT_SKIP;
> +
> +	if (res4)
> +		res4 = tcp_peek_offset_test(AF_INET);
> +
> +	if (res6)
> +		res6 = tcp_peek_offset_test(AF_INET6);
> +
> +	if (!res4 || !res6)
> +		return KSFT_FAIL;
> +
> +	return KSFT_PASS;
> +}
> +

-- 
Stefano