[OpenWrt-Devel] [PATCH] [odhcp6c] ra: clear RA information and request new after link-up event or SIGUSR2
Pavel Merzlyakov
pavel.merzlyakov at gmail.com
Tue Mar 19 08:14:52 EDT 2019
>why is this not the case in your setup ?
Because sometimes odhcp6c miss restart, probably because of Bridge Mode
(it's attached to bridge interface).
Anyway, I use a hotplug script which sends SIGUSR2 signal to odhcp6c on
link-up event.
Of course I could send SIGTERM, but I think it's not the proper way to
solve problem.
I thought that it is an obvious bug because I see a lack of consistency.
Why odhcp6c on SIGUSR2
clear STATE_IA_ and resend DHCPV6_MSG_
and don't do same with RA - clear STATE_RA_ and resend Router Solicitation
message?
Br,
Pavel
On Tue, 19 Mar 2019 at 12:36, Hans Dedecker <dedeckeh at gmail.com> wrote:
> Hi,
>
> On Mon, Mar 18, 2019 at 8:51 PM Pavel Merzlyakov
> <pavel.merzlyakov at gmail.com> wrote:
> >
> > Hi,
> >
> > >Can you be elaborate more in detail what use case you want to cover
> > >with this patch; in other words what is not working now ?
> > Ok,
> > my setup:
> > Two peer routers A and B which both connected to gateway C.
> > Routers A and B have public IPv6 addresses on WAN interfaces from same
> subnet (because of RA on C) and private IPv6 addresses on LAN interfaces.
> > and issue:
> > If I unplug router B from C and then plug it in LAN port of router A:
> > 1) Router B does not send Router Solicitation message. So router B does
> not have a valid gateway for some time.
> Since we've linksensing support in netifd the wan/wan6 interface
> should be brought down when the cable is unplugged; the wan/wan6
> interface will be brought up again when the cable is plugged in. This
> means odhcp6c will be restarted and will start sending RS; why is this
> not the case in your setup ?
>
> Hans
> > 2) Router B does not delete old IPv6 address on the WAN. After first
> Router Advertisement message router B is capable to send a request via new
> gateway A, but it can't receive responses because of invalid/old source
> address in request.
> >
> >
> > >Why do you want to restart RA if an SIGUSR2 signal is received ?
> > Because I assume that SIGUSR2 signal is equivalent to link-up event
> (see odhcp6c:674 ->
> https://git.openwrt.org/?p=project/odhcp6c.git;a=blob;f=src/odhcp6c.c;h=19a86f2654bf3c59b0f47cf0aedd87235187bf89;hb=d2e247d8d87ecf8c60fcf0acdad05667bd379521#l674
> ).
> > In my case I use hotplug script witch on link-up sends SIGUSR2 signal to
> odhcp6c (it's connected to bridge interface).
> > ra_restart () could be called without condition in beginning of the main
> loop.
> >
> > Br,
> > Pavel
> >
> > On Mon, 18 Mar 2019 at 18:26, Hans Dedecker <dedeckeh at gmail.com> wrote:
> >>
> >> Hi,
> >>
> >> On Mon, Mar 18, 2019 at 2:43 PM <pavel.merzlyakov at gmail.com> wrote:
> >> >
> >> > From: Pavel Merzlyakov <pavel.merzlyakov at gmail.com>
> >> >
> >> > A subnet may be changed after link-up event
> >> Can you be elaborate more in detail what use case you want to cover
> >> with this patch; in other words what is not working now ?
> >> >
> >> > Signed-off-by: Pavel Merzlyakov <pavel.merzlyakov at gmail.com>
> >> > ---
> >> > src/odhcp6c.c | 3 +++
> >> > src/ra.c | 20 +++++++++++++++++---
> >> > src/ra.h | 1 +
> >> > 3 files changed, 21 insertions(+), 3 deletions(-)
> >> >
> >> > diff --git a/src/odhcp6c.c b/src/odhcp6c.c
> >> > index 19a86f2..dd20f39 100644
> >> > --- a/src/odhcp6c.c
> >> > +++ b/src/odhcp6c.c
> >> > @@ -455,6 +455,9 @@ int main(_unused int argc, char* const argv[])
> >> >
> >> > syslog(LOG_NOTICE, "(re)starting transaction on %s",
> ifname);
> >> >
> >> > + if (signal_usr2)
> >> > + ra_restart();
> >> Why do you want to restart RA if an SIGUSR2 signal is received ?
> >>
> >> Hans
> >> > +
> >> > signal_usr1 = signal_usr2 = false;
> >> > int mode = dhcpv6_set_ia_mode(ia_na_mode, ia_pd_mode);
> >> > if (mode != DHCPV6_STATELESS)
> >> > diff --git a/src/ra.c b/src/ra.c
> >> > index 898f449..917df11 100644
> >> > --- a/src/ra.c
> >> > +++ b/src/ra.c
> >> > @@ -55,6 +55,7 @@ static int sock = -1, rtnl = -1;
> >> > static int if_index = 0;
> >> > static char if_name[IF_NAMESIZE] = {0};
> >> > static volatile int rs_attempt = 0;
> >> > +static const int rs_attempt_limit = 4;
> >> > static struct in6_addr lladdr = IN6ADDR_ANY_INIT;
> >> > static unsigned int ra_options = 0;
> >> > static unsigned int ra_holdoff_interval = 0;
> >> > @@ -179,6 +180,20 @@ failure:
> >> > return -1;
> >> > }
> >> >
> >> > +void ra_restart(void)
> >> > +{
> >> > + const int rs_attempt_old = rs_attempt;
> >> > +
> >> > + odhcp6c_clear_state(STATE_RA_PREFIX);
> >> > + odhcp6c_clear_state(STATE_RA_ROUTE);
> >> > + odhcp6c_clear_state(STATE_RA_DNS);
> >> > + odhcp6c_clear_state(STATE_RA_SEARCH);
> >> > +
> >> > + rs_attempt = 0;
> >> > + if (rs_attempt_old == 0 || rs_attempt_old >= rs_attempt_limit)
> >> > + ra_send_rs(SIGALRM);
> >> > +}
> >> > +
> >> > static void ra_send_rs(int signal __attribute__((unused)))
> >> > {
> >> > const struct sockaddr_in6 dest = {AF_INET6, 0, 0,
> ALL_IPV6_ROUTERS, if_index};
> >> > @@ -193,7 +208,7 @@ static void ra_send_rs(int signal
> __attribute__((unused)))
> >> > if (sendto(sock, &rs, len, MSG_DONTWAIT, (struct
> sockaddr*)&dest, sizeof(dest)) < 0)
> >> > syslog(LOG_ERR, "Failed to send RS (%s)",
> strerror(errno));
> >> >
> >> > - if (++rs_attempt <= 3)
> >> > + if (++rs_attempt < rs_attempt_limit)
> >> > alarm(4);
> >> > }
> >> >
> >> > @@ -243,8 +258,7 @@ bool ra_link_up(void)
> >> > if (ret) {
> >> > syslog(LOG_NOTICE, "carrier => %i event on %s",
> (int)!nocarrier, if_name);
> >> >
> >> > - rs_attempt = 0;
> >> > - ra_send_rs(SIGALRM);
> >> > + ra_restart();
> >> > }
> >> >
> >> > return ret;
> >> > diff --git a/src/ra.h b/src/ra.h
> >> > index 9acc8cd..4ec208f 100644
> >> > --- a/src/ra.h
> >> > +++ b/src/ra.h
> >> > @@ -46,5 +46,6 @@ struct icmpv6_opt_route_info {
> >> >
> >> > int ra_init(const char *ifname, const struct in6_addr *ifid,
> >> > unsigned int options, unsigned int holdoff_interval);
> >> > +void ra_restart(void);
> >> > bool ra_link_up(void);
> >> > bool ra_process(void);
> >> > --
> >> > 2.21.0
> >> >
> >> >
> >> > _______________________________________________
> >> > openwrt-devel mailing list
> >> > openwrt-devel at lists.openwrt.org
> >> > https://lists.openwrt.org/mailman/listinfo/openwrt-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.infradead.org/pipermail/openwrt-devel/attachments/20190319/640bafef/attachment.htm>
-------------- next part --------------
_______________________________________________
openwrt-devel mailing list
openwrt-devel at lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel
More information about the openwrt-devel
mailing list