> On 27 Dec 2014, at 6:09 pm, Kapetanakis Giannis <[email protected]> > wrote: > > On 27/12/14 10:05, Kapetanakis Giannis wrote: >> On 26/12/14 12:23, Kapetanakis Giannis wrote: >>> Hi, >>> >>> Any ideas on this? I'm getting at least one panic every day. >>> >>> G >>> >>> On 24/12/14 06:13, Kapetanakis Giannis wrote: >>>> Today I've installed a 10Gb adapter and upgraded to latest snapshot. >>>> I've had a crash... >>>> >>>> Machine is a Fujitsu RX300 S6 and the adapter is an Intel X520 SR1 >>>> >>>> G >>>> >>>> ddb{0}> trace >>>> ixgbe_tx_ctx_setup(d4164980,d919df00,f55a0e5c,f55a0e60,4) at >>>> ixgbe_tx_ctx_setup >>>> +0x11a >>>> ixgbe_encap(d4164980,d919df00,0,1000,a) at ixgbe_encap+0x16a >>>> ixgbe_start(d4376030,d424f2c0,1,f55a0ed4,d43ac320) at ixgbe_start+0xa7 >>>> nettxintr(0,50,0,f55a0f08,d055e282) at nettxintr+0x47 >>>> softintr_dispatch(1) at softintr_dispatch+0x5a >>>> Xsoftnet() at Xsoftnet+0x17 >>>> --- interrupt --- >>>> cpu_idle_cycle(d0c54a00) at cpu_idle_cycle+0xf >>>> Bad frame pointer: 0xd0d1ee58 >> >> I'm still getting the crash at the same point. >> I've followed http://www.benzedrine.cx/crashreport.html >> and I have managed to trace it to line 2114 of revision 1.113 >> >> 2107 /* Set the ether header length */ >> 2108 vlan_macip_lens |= ehdrlen << IXGBE_ADVTXD_MACLEN_SHIFT; >> 2109 >> 2110 switch (etype) { >> 2111 case ETHERTYPE_IP: >> 2112 ip = (struct ip *)(mp->m_data + ehdrlen); >> 2113 ip_hlen = ip->ip_hl << 2; >>>> 2114 ipproto = ip->ip_p; >> 2115 type_tucmd_mlhl |= IXGBE_ADVTXD_TUCMD_IPV4; >> 2116 break; >> >> I've also tried to revert back to older revisions of if_ix.c (until 1.108) >> with no luck. >> I will continue a bit more to see if anything changes. >> >> regards, >> >> Giannis > > Just to add that I'm running vlan 802.1q trunk on this interface.
thats pretty cool. what kind of traffic are you sending on this interface? is it forwarded traffic or locally generated? cheers, dlg
