Re: kern/168742: detaching of ethernet adapter with configured vlans leads to panic
On Tuesday, August 07, 2012 12:48:31 am a...@freebsd.org wrote:
> Synopsis: detaching of ethernet adapter with configured vlans leads to panic
>
> Responsible-Changed-From-To: freebsd-bugs->freebsd-net
> Responsible-Changed-By: ae
> Responsible-Changed-When: Tue Aug 7 04:48:17 UTC 2012
> Responsible-Changed-Why:
> Reassign.
>
> http://www.freebsd.org/cgi/query-pr.cgi?pr=168742
Ugh, I thought I had fixed this in
http://svnweb.freebsd.org/base?view=revision&revision=208212
I think the problem is the assertion is wrong. We could add a new DETACHING
flag, but I think the simplest fix is to just remove it. I'm not sure if a
similar assertion in if_delmulti_ifma() should also be removed.
Index: if.c
===
--- if.c(revision 238992)
+++ if.c(working copy)
@@ -3058,19 +3058,7 @@ if_delmulti(struct ifnet *ifp, struct sockaddr *sa
{
struct ifmultiaddr *ifma;
int lastref;
-#ifdef INVARIANTS
- struct ifnet *oifp;
- IFNET_RLOCK_NOSLEEP();
- TAILQ_FOREACH(oifp, &V_ifnet, if_link)
- if (ifp == oifp)
- break;
- if (ifp != oifp)
- ifp = NULL;
- IFNET_RUNLOCK_NOSLEEP();
-
- KASSERT(ifp != NULL, ("%s: ifnet went away", __func__));
-#endif
if (ifp == NULL)
return (ENOENT);
--
John Baldwin
___
freebsd-net@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-net
To unsubscribe, send any mail to "freebsd-net-unsubscr...@freebsd.org"
Re: kern/120304: [netgraph] [patch] netgraph source assumes 32-bit timeval on AMD64
Hi! This is still the problem for 9.1-PRERELEASE: ng_source is unusable for 64bit systems. The patch really helps and fixes the problem for 64bit (and breaks for 32bit :-) Perhaps, it may be improved with some #ifdef's Eugene Grosbein ___ freebsd-net@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-net To unsubscribe, send any mail to "freebsd-net-unsubscr...@freebsd.org"
Re: kern/120304: [netgraph] [patch] netgraph source assumes 32-bit timeval on AMD64
The following reply was made to PR kern/120304; it has been noted by GNATS. From: Eugene Grosbein To: bug-follo...@freebsd.org, "n...@freebsd.org" Cc: Subject: Re: kern/120304: [netgraph] [patch] netgraph source assumes 32-bit timeval on AMD64 Date: Tue, 07 Aug 2012 19:38:08 +0700 Hi! This is still the problem for 9.1-PRERELEASE: ng_source is unusable for 64bit systems. The patch really helps and fixes the problem for 64bit (and breaks for 32bit :-) Perhaps, it may be improved with some #ifdef's Eugene Grosbein ___ freebsd-net@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-net To unsubscribe, send any mail to "freebsd-net-unsubscr...@freebsd.org"
Send traffic to itself using real NIC
Hi! I'd like to run some stability tests for some L2-transparent gigabit ethernet network hardware I've got. I have one spare FreeBSD 9.1-PRERELEASE with two-ports igb(4) card and connected both ports using my L2 hardward making a loop. Is it possible to configure FreeBSD in such way so it would send IP traffic for itself not via loopback interface but via igb0/igb1? I've tried to utilize multiple FIBs: sysctl net.add_addr_allfibs=0 ifconfig igb0 inet 10.0.0.1/24 mtu 9126 ifconfig igb1 inet 10.0.1.1/24 mtu 9126 setfib 1 route -q flush setfib 1 route add 10.0.1.1/32 -iface igb1 setfib 2 route -q flush setfib 2 route add 10.0.0.1/32 -iface igb0 But "setfib 1 ping -S 10.0.0.1 10.0.1.1" still runs via lo0 (and has no answer). The only way I've got working is unidirectional stream of packets generated with ng_source(4) (patched due to kern/120304) that shows me 10% packet loss (using netstat -I to see how many packets got through). I would really like using normal TCP/UDP/ICMP streams in addition of ng_source tests. Eugene Grosbein ___ freebsd-net@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-net To unsubscribe, send any mail to "freebsd-net-unsubscr...@freebsd.org"
Re: Send traffic to itself using real NIC
On Tue, Aug 7, 2012 at 3:53 PM, Eugene Grosbein wrote: > Hi! > > I'd like to run some stability tests for some L2-transparent > gigabit ethernet network hardware I've got. > > I have one spare FreeBSD 9.1-PRERELEASE with two-ports igb(4) card > and connected both ports using my L2 hardward making a loop. > > Is it possible to configure FreeBSD in such way so it would send > IP traffic for itself not via loopback interface but via igb0/igb1? > > I've tried to utilize multiple FIBs: > > sysctl net.add_addr_allfibs=0 > ifconfig igb0 inet 10.0.0.1/24 mtu 9126 > ifconfig igb1 inet 10.0.1.1/24 mtu 9126 > setfib 1 route -q flush > setfib 1 route add 10.0.1.1/32 -iface igb1 > setfib 2 route -q flush > setfib 2 route add 10.0.0.1/32 -iface igb0 > > But "setfib 1 ping -S 10.0.0.1 10.0.1.1" still runs via lo0 (and has no > answer). > > The only way I've got working is unidirectional stream of packets > generated with ng_source(4) (patched due to kern/120304) that shows me 10% > packet loss > (using netstat -I to see how many packets got through). > > I would really like using normal TCP/UDP/ICMP streams in addition of > ng_source tests. > > Eugene Grosbein > ___ > freebsd-net@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-net > To unsubscribe, send any mail to "freebsd-net-unsubscr...@freebsd.org" You can try using a VIMAGE enabled kernel and jails using epair(4) as the network interface. Bridge the epair interfaces on the host to igb0/igb1 and you should be able to communicate between the two jails via igb0 and igb1. ___ freebsd-net@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-net To unsubscribe, send any mail to "freebsd-net-unsubscr...@freebsd.org"
Re: Send traffic to itself using real NIC
On Tuesday 07 August 2012 16:04:17 Jacques Fourie wrote: > On Tue, Aug 7, 2012 at 3:53 PM, Eugene Grosbein wrote: > > Hi! > > > > I'd like to run some stability tests for some L2-transparent > > gigabit ethernet network hardware I've got. > > > > I have one spare FreeBSD 9.1-PRERELEASE with two-ports igb(4) card > > and connected both ports using my L2 hardward making a loop. > > > > Is it possible to configure FreeBSD in such way so it would send > > IP traffic for itself not via loopback interface but via igb0/igb1? > > > > I've tried to utilize multiple FIBs: > > > > sysctl net.add_addr_allfibs=0 > > ifconfig igb0 inet 10.0.0.1/24 mtu 9126 > > ifconfig igb1 inet 10.0.1.1/24 mtu 9126 > > setfib 1 route -q flush > > setfib 1 route add 10.0.1.1/32 -iface igb1 > > setfib 2 route -q flush > > setfib 2 route add 10.0.0.1/32 -iface igb0 > > > > But "setfib 1 ping -S 10.0.0.1 10.0.1.1" still runs via lo0 (and has no > > answer). > > > > The only way I've got working is unidirectional stream of packets > > generated with ng_source(4) (patched due to kern/120304) that shows me > > 10% packet loss (using netstat -I to see how many packets got through). > > > > I would really like using normal TCP/UDP/ICMP streams in addition of > > ng_source tests. > > > > Eugene Grosbein > > ___ > > freebsd-net@freebsd.org mailing list > > http://lists.freebsd.org/mailman/listinfo/freebsd-net > > To unsubscribe, send any mail to "freebsd-net-unsubscr...@freebsd.org" > > You can try using a VIMAGE enabled kernel and jails using epair(4) as > the network interface. Bridge the epair interfaces on the host to > igb0/igb1 and you should be able to communicate between the two jails > via igb0 and igb1. There's no need to use epair at all here. Just create a vnet jail, and assign one of the igb interfaces into it, done: # jail -c name=x vnet persist # ifconfig igb1 vnet x # jexec x ifconfig lo0: flags=8008 metric 0 mtu 16384 options=3 igb1: flags=8842<...> metric 0 mtu 1500 ... Marko ___ freebsd-net@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-net To unsubscribe, send any mail to "freebsd-net-unsubscr...@freebsd.org"
Re: Send traffic to itself using real NIC
On Tue, Aug 07, 2012 at 08:53:52PM +0700, Eugene Grosbein wrote: > Hi! > > I'd like to run some stability tests for some L2-transparent > gigabit ethernet network hardware I've got. > > I have one spare FreeBSD 9.1-PRERELEASE with two-ports igb(4) card > and connected both ports using my L2 hardward making a loop. > > Is it possible to configure FreeBSD in such way so it would send > IP traffic for itself not via loopback interface but via igb0/igb1? > > I've tried to utilize multiple FIBs: > > sysctl net.add_addr_allfibs=0 > ifconfig igb0 inet 10.0.0.1/24 mtu 9126 > ifconfig igb1 inet 10.0.1.1/24 mtu 9126 > setfib 1 route -q flush > setfib 1 route add 10.0.1.1/32 -iface igb1 > setfib 2 route -q flush > setfib 2 route add 10.0.0.1/32 -iface igb0 > > But "setfib 1 ping -S 10.0.0.1 10.0.1.1" still runs via lo0 (and has no > answer). > > The only way I've got working is unidirectional stream of packets > generated with ng_source(4) (patched due to kern/120304) that shows me 10% > packet loss > (using netstat -I to see how many packets got through). > > I would really like using normal TCP/UDP/ICMP streams in addition of > ng_source tests. > You could just throw a loopback plug in your adapter. All traffic going out is going to come right back in as fast as its sent out. Another alternative is a crossover cable from igb0 -> igb1. -- - (2^(N-1)) JJH48-ARIN pgpkiGAIimMC8.pgp Description: PGP signature
RE: networking perplexity
Thanks for the reply... Sorry for the delay.. 10gb0 - 192.168.33.0/24 10gb1 - 192.168.32.0/24 igb0 - 152.x.x.x/24 igb3 - 192.168.34.0/24 all routed out through 152.x.x.254 gateway rc.conf (attempted, hangs on boot): hostname="fqdn (resolvable)" ifconfig_igb3="inet 192.168.34.2 netmask 255.255.255.0" ifconfig_ix0="inet 192.168.33.2 netmask 255.255.255.0" ifconfig_ix1="inet 192.168.32.2 netmask 255.255.255.0" ifconfig_igb0="inet 152.x.x.x netmask 255.255.255.0" defaultrouter="152.x.x.254" sshd_enable="YES" firewall_enable="NO" #firewall_type="open" # Set dumpdev to "AUTO" to enable crash dumps, "NO" to disable dumpdev="NO" # Local services zfs_enable="YES" powerd_enable="YES" # Network services sshd_enable="YES" samba_enable="YES" # NFS rpcbind_enable="YES" nfs_server_enable="YES" mountd_enable="YES" mountd_flags="-r" rpc_lockd_enable="YES" rpc_statd_enable="YES" # Synchronise time ntpdate_enable="YES" ntpdate_hosts="timeserver (fqdn)" The workaround that works for a minute I comment out the above ugb3 config line and add the following in /usr/local/etc/rc.d !/bin/csh ifconfig igb3 inet 192.68.34.2 netmask 255.255.255.0 /etc/rc.d/routing restart Thanks again! Eddie From: Julian Elischer [jul...@freebsd.org] Sent: Tuesday, July 31, 2012 12:21 PM To: Dunn, George Jr Cc: freebsd-net@freebsd.org Subject: Re: networking perplexity On 7/31/12 7:21 AM, Dunn, George Jr wrote: > Hi, > > I am running Freebsd 9 with all current updates and am having some serious > issues with the networking/routing stack. > > I have a machine that has 2 - 10gb interfaces and 2 - 1gb interfaces that I > wish to configure. > The desired effect is to have all interfaces except one of the 1Gb ports to > be broadcast only private address spaces for running nfs. > I have one public fully routable address. > Trying to configure the interfaces with rc.conf I am able to get the 10gb > interfaces and the public working. > However when I add in the third private address on the other 1 gb interface > the whole stack seems to freak. It will completely hang the boot usually > requiring a hard power off and single user mode to fsck to reboot. > > I tried to work around by placing a script in the new rc.local place to bring > up the last interface and fix the routing table. It will boot then but > shortly after (15-45 mins) I lose all connectivity again. I have tried > different ports on the box as well as several different switches with the > same result so I don't think it's a hardware issue per se. doesn't sound particularly hairy.. can you draw a quick diag,with numbers and then show your config files? > > I really just need some input! Any takers? > > Thanks in advance! > > Eddie > > > ___ > freebsd-net@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-net > To unsubscribe, send any mail to "freebsd-net-unsubscr...@freebsd.org" > > ___ freebsd-net@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-net To unsubscribe, send any mail to "freebsd-net-unsubscr...@freebsd.org"
Re: Send traffic to itself using real NIC
On Tue, Aug 07, 2012 at 04:33:11PM +0200, Marko Zec wrote: > There's no need to use epair at all here. Just create a vnet jail, and > assign > one of the igb interfaces into it, done: > > # jail -c name=x vnet persist > # ifconfig igb1 vnet x > # jexec x ifconfig > lo0: flags=8008 metric 0 mtu 16384 > options=3 > igb1: flags=8842<...> metric 0 mtu 1500 ... That would be nice solution but jail(8) manual page says: "a fairly complete file system install of FreeBSD is required" and that seems to be pretty big overhead. Meantime, I'll try to think up some kind of NETGRAPH mesh. Eugene Grosbein ___ freebsd-net@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-net To unsubscribe, send any mail to "freebsd-net-unsubscr...@freebsd.org"
Re: Send traffic to itself using real NIC
On Tuesday 07 August 2012 18:20:28 Eugene Grosbein wrote: > On Tue, Aug 07, 2012 at 04:33:11PM +0200, Marko Zec wrote: > > There's no need to use epair at all here. Just create a vnet jail, and > > assign one of the igb interfaces into it, done: > > > > # jail -c name=x vnet persist > > # ifconfig igb1 vnet x > > # jexec x ifconfig > > lo0: flags=8008 metric 0 mtu 16384 > > options=3 > > igb1: flags=8842<...> metric 0 mtu 1500 ... > > That would be nice solution but jail(8) manual page says: > "a fairly complete file system install of FreeBSD is required" No, a separate filesystem is not required, that's a manpage bug. Just try the above 3 commands, they really work. cheers, Marko > and that seems to be pretty big overhead. Meantime, I'll try to think up > some kind of NETGRAPH mesh. > > Eugene Grosbein ___ freebsd-net@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-net To unsubscribe, send any mail to "freebsd-net-unsubscr...@freebsd.org"
