Bug 222273

Summary: igb(4): Kernel panic (fatal trap 12) due to network related processes
Product: Base System Reporter: aayla.secura.1138
Component: kernAssignee: Graham Perrin <grahamperrin>
Status: Closed Feedback Timeout    
Severity: Affects Only Me CC: eugen, net, shurd, stephane.rochoy, terry-freebsd
Priority: --- Keywords: crash, needs-qa
Version: 11.0-RELEASE   
Hardware: amd64   
OS: Any   
Attachments:
Description Flags
kgdb backtrace none

Description aayla.secura.1138 2017-09-13 01:06:31 UTC 
Created attachment 186312 [details]
kgdb backtrace

The steps to getting it to panic are a bit elaborate, but the bug is reproducible every time: If I stop the netif service, then run dhclient on the wired network interface and in the meantime (while dhclient is sending discover messages) I perform a network related operation, such as printing the routing table, the kernel panics with Fatal trap 12: page fault while in kernel mode.

FreeBSD version is 11.0-RELEASE-p1.

It is a server machine: The CPU is Xeon E3-1275v6. The motherboard is Intel DBS1200SPLR, it has two Ethernet ports, both using igb driver. I tried wiring each and the panic occurs either way, but only if dhclient is run on the active (plugged in) interface.

I attach the crash dump from one of the panics. The original forum thread I opened is here: https://forums.freebsd.org/threads/62338/

The bug may be the same, or related, to this: https://lists.freebsd.org/pipermail/freebsd-hackers/2017-June/051200.html
Comment 1 Terry Kennedy 2017-09-13 01:21:58 UTC 
This appears to be a NULL + 0x17 pointer reference at soclose+0x3c

uipc_socket.c:1046 error = (*so->so_proto->pr_usrreqs->pru_disconnect)(so);
Comment 2 Eugene Grosbein freebsd_committer freebsd_triage 2018-04-09 15:12:50 UTC 
Is it repeatable with 11.1-RELEASE, or better try 11.1-STABLE?