Skip to contentSkip to conversation

net: ping: reset stored IP once the command finishes

PATCH New Marek Szyprowski 2 participants 3 messages Mar 25, 2020 – Jun 15, 2020 1 message not in the archive Networking

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): CGME20200325134211eucas1p25efcdbfaf6cd1a1088d6eb104e75db19@eucas1p2.samsung.com

#1 lore raw

PATCH 1/1 net: ping: reset stored IP once the command finishes

Reset stored ping IP address before leaving the netloop to ensure that the subsequent calls to the netloop, especially for the other protocols, won't be interrupted by the received ICMP_ECHO_REPLY packet.

---
 net/ping.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
net/ping.c +4 -1
index 633c942..d912e3d 100644
@@ -63,6 +63,7 @@ static int ping_send(void)
 static void ping_timeout_handler(void)
 {
 	eth_halt();
+	net_ping_ip.s_addr = 0;
 	net_set_state(NETLOOP_FAIL);	/* we did not get the reply */
 }
 
@@ -84,8 +85,10 @@ void ping_receive(struct ethernet_hdr *et, struct ip_udp_hdr *ip, int len)
 	switch (icmph->type) {
 	case ICMP_ECHO_REPLY:
 		src_ip = net_read_ip((void *)&ip->ip_src);
-		if (src_ip.s_addr == net_ping_ip.s_addr)
+		if (src_ip.s_addr == net_ping_ip.s_addr) {
+			net_ping_ip.s_addr = 0;
 			net_set_state(NETLOOP_SUCCESS);
+		}
 		return;
 	case ICMP_ECHO_REQUEST:
 		eth_hdr_size = net_update_ether(et, et->et_src, PROT_IP);
Signature · 1 line
1.9.1
#2 lore raw

PATCH 1/1 net: ping: reset stored IP once the command finishes

Quoted 23 lines · Marek Szyprowski

Reset stored ping IP address before leaving the netloop to ensure that the subsequent calls to the netloop, especially for the other protocols, won't be interrupted by the received ICMP_ECHO_REPLY packet.

---
 net/ping.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
net/ping.c +1 -0
index 633c942..d912e3d 100644
@@ -63,6 +63,7 @@ static int ping_send(void)
 static void ping_timeout_handler(void)
 {
 	eth_halt();
+	net_ping_ip.s_addr = 0;
 	net_set_state(NETLOOP_FAIL);	/* we did not get the reply */
 }
 
@@ -84,8 +85,10 @@ void ping_receive(struct ethernet_hdr *et, struct ip_udp_hdr *ip, int len)
 	switch (icmph->type) {

net/ping.c void ping_receive(struct ethernet_hdr *et, struct ip_udp_hdr *ip, int len) · line 94

Quoted 10 lines
net/ping.c +3 -1
 	case ICMP_ECHO_REPLY:
 		src_ip = net_read_ip((void *)&ip->ip_src);
-		if (src_ip.s_addr == net_ping_ip.s_addr)
+		if (src_ip.s_addr == net_ping_ip.s_addr) {
+			net_ping_ip.s_addr = 0;
 			net_set_state(NETLOOP_SUCCESS);
+		}
 		return;
 	case ICMP_ECHO_REQUEST:
 		eth_hdr_size = net_update_ether(et, et->et_src, PROT_IP);

This breaks a number of tests in test/dm/eth.c. Can you please adjust the tests as well to not rely on the behavior you're fixing? Thanks!

Signature · 8 lines
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20200612/a548967d/attachment.sig>
#3 lore raw

PATCH 1/1 net: ping: reset stored IP once the command finishes

Hi Tom,

Quoted 22 lines · Tom Rini
Quoted 21 lines · Marek Szyprowski

Reset stored ping IP address before leaving the netloop to ensure that the subsequent calls to the netloop, especially for the other protocols, won't be interrupted by the received ICMP_ECHO_REPLY packet.

---
  net/ping.c | 5 ++++-
  1 file changed, 4 insertions(+), 1 deletion(-)
net/ping.c +1 -0
index 633c942..d912e3d 100644
@@ -63,6 +63,7 @@ static int ping_send(void)
  static void ping_timeout_handler(void)
  {
  	eth_halt();
+	net_ping_ip.s_addr = 0;
  	net_set_state(NETLOOP_FAIL);	/* we did not get the reply */
  }
 
Quoted 14 lines

net/ping.c void ping_receive(struct ethernet_hdr *et, struct ip_udp_hdr *ip, int len) · line 94

Quoted 12 lines
net/ping.c +3 -1
@@ -84,8 +85,10 @@ void ping_receive(struct ethernet_hdr *et, struct ip_udp_hdr *ip, int len)
  	switch (icmph->type) {
  	case ICMP_ECHO_REPLY:
  		src_ip = net_read_ip((void *)&ip->ip_src);
-		if (src_ip.s_addr == net_ping_ip.s_addr)
+		if (src_ip.s_addr == net_ping_ip.s_addr) {
+			net_ping_ip.s_addr = 0;
  			net_set_state(NETLOOP_SUCCESS);
+		}
  		return;
  	case ICMP_ECHO_REQUEST:
  		eth_hdr_size = net_update_ether(et, et->et_src, PROT_IP);

This breaks a number of tests in test/dm/eth.c. Can you please adjust the tests as well to not rely on the behavior you're fixing? Thanks!

I've checked and it turned out that it is not possible to fix the net_retry test. It clearly shows that the net_retry feature relies on the fact that the state machine of the net related commands are preserved between the calls.

I wonder how to fix this without breaking the netretry feature. The only solution I see is to clear net_ping_ip at the beginning of the net_loop() if the protocol is not equal to PING. Quite ugly...

Best regards

Signature · 2 lines
Marek Szyprowski, PhD
Samsung R&D Institute Poland