Skip to contentSkip to conversation

busybox: Update udhcpc.script for stateful DHCPv6

PATCH Superseded Samuel Mendoza-Jonas 3 participants 6 messages Jul 26, 2018 – Apr 9, 2019

v4 of 5 Superseded Latest is v5 →

lore
Share & tools
lore
#0 lore raw

PATCH v4 1/1 busybox: Update udhcpc.script for stateful DHCPv6

udhcpc6 will call the default script with the address set in the "ipv6" variable. Set "ip" to this address if present.

udhcpc6 implements "stateful" DHCPv6 for explicitly requesting an address and other configuration information. A major difference between DHCPv4 and DHCPv6 is that DHCPv6 does *not* advertise a default route; this is determined by normal IPv6 autoconfiguration.
Included is a change from Rob, which if handling a DHCPv6 address waits a moment for the IPv6 route to be configured; as above this doesn't come from DHCPv6 but rather the IPv6 Router Advertisement (RA) which happens independently from udhcpc6. The intent here is to try and ensure that the interface is route-able upon the script's completion as it would be if called from udhcpc.

(wait for IPv6 route)
From: Robert Lippert <[email protected]>

---
v4: Expand description of stateful DHCPv6 in commit message
v3: Point out 'stateful' DHCPv6 in commit, send to the correct list...
v2: Avoid bashisms

 package/busybox/udhcpc.script | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
package/busybox/udhcpc.script +14 -0
index ad110d3a7f..65114b6cd0 100755
@@ -8,6 +8,16 @@ RESOLV_CONF="/etc/resolv.conf"
 [ -e $RESOLV_CONF ] || touch $RESOLV_CONF
 [ -n "$broadcast" ] && BROADCAST="broadcast $broadcast"
 [ -n "$subnet" ] && NETMASK="netmask $subnet"
+[ -n "$ipv6" ] && ip="$ipv6/128"
+
+wait_for_ipv6_default_route() {
+	attempts=10
+	while [ $attempts != 0 ] && [ -z "$(ip -6 route list | grep default)" ] ; do
+		sleep 1
+		attempts=$(($attempts-1))
+	done
+	[ $attempts = 0 ] && echo "failed to get default ipv6 route"
+}
 
 case "$1" in
 	deconfig)
@@ -37,6 +47,10 @@ case "$1" in
 			/usr/sbin/avahi-autoipd -k $interface
 		fi
 		/sbin/ifconfig $interface $ip $BROADCAST $NETMASK
+		if [ -n "$ipv6" ] ; then
+			echo "waiting for default ipv6 route set"
+			wait_for_ipv6_default_route
+		fi
 
 		if [ -n "$router" ] ; then
 			echo "deleting routers"
Signature · 1 line
2.18.0
#1 lore raw

PATCH v4 1/1 busybox: Update udhcpc.script for stateful DHCPv6

Quoted 14 lines · Samuel Mendoza-Jonas

udhcpc6 will call the default script with the address set in the "ipv6" variable. Set "ip" to this address if present.

udhcpc6 implements "stateful" DHCPv6 for explicitly requesting an address and other configuration information. A major difference between DHCPv4 and DHCPv6 is that DHCPv6 does *not* advertise a default route; this is determined by normal IPv6 autoconfiguration.
Included is a change from Rob, which if handling a DHCPv6 address waits a moment for the IPv6 route to be configured; as above this doesn't come from DHCPv6 but rather the IPv6 Router Advertisement (RA) which happens independently from udhcpc6. The intent here is to try and ensure that the interface is route-able upon the script's completion as it would be if called from udhcpc.

Quoted 7 lines

(wait for IPv6 route)
From: Robert Lippert <[email protected]>

---
v4: Expand description of stateful DHCPv6 in commit message
v3: Point out 'stateful' DHCPv6 in commit, send to the correct list...
v2: Avoid bashisms

Hi all, any more thoughts on this?

Quoted 36 lines
 package/busybox/udhcpc.script | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
package/busybox/udhcpc.script +14 -0
index ad110d3a7f..65114b6cd0 100755
@@ -8,6 +8,16 @@ RESOLV_CONF="/etc/resolv.conf"
 [ -e $RESOLV_CONF ] || touch $RESOLV_CONF
 [ -n "$broadcast" ] && BROADCAST="broadcast $broadcast"
 [ -n "$subnet" ] && NETMASK="netmask $subnet"
+[ -n "$ipv6" ] && ip="$ipv6/128"
+
+wait_for_ipv6_default_route() {
+	attempts=10
+	while [ $attempts != 0 ] && [ -z "$(ip -6 route list | grep default)" ] ; do
+		sleep 1
+		attempts=$(($attempts-1))
+	done
+	[ $attempts = 0 ] && echo "failed to get default ipv6 route"
+}
 
 case "$1" in
 	deconfig)
@@ -37,6 +47,10 @@ case "$1" in
 			/usr/sbin/avahi-autoipd -k $interface
 		fi
 		/sbin/ifconfig $interface $ip $BROADCAST $NETMASK
+		if [ -n "$ipv6" ] ; then
+			echo "waiting for default ipv6 route set"
+			wait_for_ipv6_default_route
+		fi
 
 		if [ -n "$router" ] ; then
 			echo "deleting routers"
#2 lore raw

PATCH v4 1/1 busybox: Update udhcpc.script for stateful DHCPv6

Quoted 23 lines · Samuel Mendoza-Jonas
Quoted 14 lines · Samuel Mendoza-Jonas

udhcpc6 will call the default script with the address set in the "ipv6" variable. Set "ip" to this address if present.

udhcpc6 implements "stateful" DHCPv6 for explicitly requesting an address and other configuration information. A major difference between DHCPv4 and DHCPv6 is that DHCPv6 does *not* advertise a default route; this is determined by normal IPv6 autoconfiguration.
Included is a change from Rob, which if handling a DHCPv6 address waits a moment for the IPv6 route to be configured; as above this doesn't come from DHCPv6 but rather the IPv6 Router Advertisement (RA) which happens independently from udhcpc6. The intent here is to try and ensure that the interface is route-able upon the script's completion as it would be if called from udhcpc.

Quoted 7 lines

(wait for IPv6 route)
From: Robert Lippert <[email protected]>

---
v4: Expand description of stateful DHCPv6 in commit message
v3: Point out 'stateful' DHCPv6 in commit, send to the correct list...
v2: Avoid bashisms

Quoted 1 line

Hi all, any more thoughts on this?

Even I nearly forgot about this one so bumping it again.

Any issues with the patch? Would it be worth splitting out the

Quoted 1 line

[ -n "$ipv6" ] && ip="$ipv6/128"

bit from the waiting section so at least IPv6 addresses will be properly added?

Cheers,
Sam

Quoted 41 lines
Quoted 26 lines
 package/busybox/udhcpc.script | 14 ++++++++++++++
 1 file changed, 14 insertions(+)
package/busybox/udhcpc.script +10 -0
index ad110d3a7f..65114b6cd0 100755
@@ -8,6 +8,16 @@ RESOLV_CONF="/etc/resolv.conf"
 [ -e $RESOLV_CONF ] || touch $RESOLV_CONF
 [ -n "$broadcast" ] && BROADCAST="broadcast $broadcast"
 [ -n "$subnet" ] && NETMASK="netmask $subnet"
+[ -n "$ipv6" ] && ip="$ipv6/128"
+
+wait_for_ipv6_default_route() {
+	attempts=10
+	while [ $attempts != 0 ] && [ -z "$(ip -6 route list | grep default)" ] ; do
+		sleep 1
+		attempts=$(($attempts-1))
+	done
+	[ $attempts = 0 ] && echo "failed to get default ipv6 route"
+}
 
 case "$1" in
 	deconfig)

package/busybox/udhcpc.script case "$1" in · line 56

Quoted 10 lines
package/busybox/udhcpc.script +4 -0
@@ -37,6 +47,10 @@ case "$1" in
 			/usr/sbin/avahi-autoipd -k $interface
 		fi
 		/sbin/ifconfig $interface $ip $BROADCAST $NETMASK
+		if [ -n "$ipv6" ] ; then
+			echo "waiting for default ipv6 route set"
+			wait_for_ipv6_default_route
+		fi
 
 		if [ -n "$router" ] ; then
 			echo "deleting routers"

_______________________________________________
buildroot mailing list
buildroot at busybox.net
http://lists.busybox.net/mailman/listinfo/buildroot

#3 lore raw

PATCH v4 1/1 busybox: Update udhcpc.script for stateful DHCPv6

Sam,

Quoted 6 lines · Samuel Mendoza-Jonas
Quoted 4 lines · Samuel Mendoza-Jonas
Quoted 3 lines · Samuel Mendoza-Jonas

udhcpc6 will call the default script with the address set in the "ipv6" variable. Set "ip" to this address if present.

Quoted 10 lines
Quoted 10 lines
Quoted 10 lines

udhcpc6 implements "stateful" DHCPv6 for explicitly requesting an address and other configuration information. A major difference between DHCPv4 and DHCPv6 is that DHCPv6 does *not* advertise a default route; this is determined by normal IPv6 autoconfiguration.
Included is a change from Rob, which if handling a DHCPv6 address waits a moment for the IPv6 route to be configured; as above this doesn't come from DHCPv6 but rather the IPv6 Router Advertisement (RA) which happens independently from udhcpc6. The intent here is to try and ensure that the interface is route-able upon the script's completion as it would be if called from udhcpc.

Based on what I understand of IPv6 this patch functionally looks correct.

I'd add a comment above the "[ -n "$ipv6" ] && ip="$ipv6/128"" line stating that udhcpc only supports stateful. Hopefully that would be enough for a user to then know they should install "odhcp6c" or another client for full RFC support.

In the "wait_for_ipv6_default_route()", do you know if the busybox "ip" tool supports the "$(ip -6 route list | grep default)" command/expected output or are you dependent on the full iproute2 package?

Also in that wait loop, I'd suggest mimic'n the loop and output behavior of this USB Nic wait script (package/ifupdown-scripts/network/if-pre-up.d/wait_iface).

Quoted 9 lines
Quoted 9 lines
Quoted 8 lines

(wait for IPv6 route)
From: Robert Lippert <[email protected]>

---
v4: Expand description of stateful DHCPv6 in commit message
v3: Point out 'stateful' DHCPv6 in commit, send to the correct list...
v2: Avoid bashisms

Quoted 8 lines
Quoted 1 line

Hi all, any more thoughts on this?

Even I nearly forgot about this one so bumping it again.

Any issues with the patch? Would it be worth splitting out the

Quoted 1 line

[ -n "$ipv6" ] && ip="$ipv6/128"

bit from the waiting section so at least IPv6 addresses will be properly added?

Sure, it would probably be good to have the changes broken into a initial stateful ip= fix and a second which adds the RA wait loop.

Thanks for the patch!

Matt

#4 lore raw

PATCH v4 1/1 busybox: Update udhcpc.script for stateful DHCPv6

In reply to Samuel Mendoza-Jonas

Hello Samuel,

Quoted 14 lines · On Thu, 26 Jul 2018 11:39:19 +1000 Samuel Mendoza-Jonas <[email protected]>…

udhcpc6 will call the default script with the address set in the "ipv6" variable. Set "ip" to this address if present.

udhcpc6 implements "stateful" DHCPv6 for explicitly requesting an address and other configuration information. A major difference between DHCPv4 and DHCPv6 is that DHCPv6 does *not* advertise a default route; this is determined by normal IPv6 autoconfiguration.
Included is a change from Rob, which if handling a DHCPv6 address waits a moment for the IPv6 route to be configured; as above this doesn't come from DHCPv6 but rather the IPv6 Router Advertisement (RA) which happens independently from udhcpc6. The intent here is to try and ensure that the interface is route-able upon the script's completion as it would be if called from udhcpc.

Quoted 7 lines

(wait for IPv6 route)
From: Robert Lippert <[email protected]>

---
v4: Expand description of stateful DHCPv6 in commit message
v3: Point out 'stateful' DHCPv6 in commit, send to the correct list...
v2: Avoid bashisms

On February 26, 2019, you received some comments on this patch (yes it took a lot of time for someone to review the patch, but it finally happened). Could you take into account those comments and post an updated version ?

Thanks a lot!

Thomas

Signature · 3 lines
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
#5 lore raw

PATCH v4 1/1 busybox: Update udhcpc.script for stateful DHCPv6

Quoted 27 lines · Thomas Petazzoni

Hello Samuel,

Quoted 14 lines · On Thu, 26 Jul 2018 11:39:19 +1000 Samuel Mendoza-Jonas <[email protected]>…

udhcpc6 will call the default script with the address set in the "ipv6" variable. Set "ip" to this address if present.

udhcpc6 implements "stateful" DHCPv6 for explicitly requesting an address and other configuration information. A major difference between DHCPv4 and DHCPv6 is that DHCPv6 does *not* advertise a default route; this is determined by normal IPv6 autoconfiguration.
Included is a change from Rob, which if handling a DHCPv6 address waits a moment for the IPv6 route to be configured; as above this doesn't come from DHCPv6 but rather the IPv6 Router Advertisement (RA) which happens independently from udhcpc6. The intent here is to try and ensure that the interface is route-able upon the script's completion as it would be if called from udhcpc.

Quoted 7 lines

(wait for IPv6 route)
From: Robert Lippert <[email protected]>

---
v4: Expand description of stateful DHCPv6 in commit message
v3: Point out 'stateful' DHCPv6 in commit, send to the correct list...
v2: Avoid bashisms

Quoted 8 lines

On February 26, 2019, you received some comments on this patch (yes it took a lot of time for someone to review the patch, but it finally happened). Could you take into account those comments and post an updated version ?

Thanks a lot!

Thomas

Hi Thomas,

So I did! This got lost in my inbox as well, I'll follow this up shortly.

Thanks,
Sam