> >Synopsis:       ex(4) is broken and may cause kernel crash
> >Environment:
> NetBSD flam.gado 4.99.6 NetBSD 4.99.6 (FLAM) #0: Mon Dec 18 05:11:17 MSK 2006  dina@flam.gado:/volatile/worksrc/netbsd-curre
> nt/obj/sys/arch/i386/compile/FLAM i386
> Architecture: i386
> Machine: i386
> SMP machine
> >Description:
> I have this network card:
> ex0 at pci0 dev 11 function 0: 3Com 3c905B-TX 10/100 Ethernet (rev. 0x24)
> ex0: interrupting at ioapic0 pin 17 (irq 10)
> 
> # ifconfig ex0
> ex0: flags=8863<UP,BROADCAST,NOTRAILERS,RUNNING,SIMPLEX,MULTICAST> mtu 1500
>         capabilities=3f00<IP4CSUM_Rx,IP4CSUM_Tx,TCP4CSUM_Rx,TCP4CSUM_Tx,UDP4CSUM_Rx,UDP4CSUM_Tx>
>         enabled=3f00<IP4CSUM_Rx,IP4CSUM_Tx,TCP4CSUM_Rx,TCP4CSUM_Tx,UDP4CSUM_Rx,UDP4CSUM_Tx>
>         address: 00:10:4b:67:87:27
>         media: Ethernet autoselect (100baseTX full-duplex)
>         status: active
>         inet 10.239.17.213 netmask 0xffff0000 broadcast 10.239.255.255
>         inet6 fe80::210:4bff:fe67:8727%ex0 prefixlen 64 scopeid 0x1
> 
> it worked fine with -current 3 months old
> 
> Today either kernel crashes, or just network malfunctions
> (even with capabilities turned off and untouched)
> 
> This is example of crash (they are always in rtflushall):
> 
> uvm_fault(0xccc7eee8, 0, 1) -> oxe
> kernel: supervisor trap page fault, code=0
> Stopped in pid xxx.1 (ifconfig) at netbsd:in_rtflushall+0x4e: movl 0x38(%edx),%eax
> 
> 
> Second ex card seems to work fine:
> 
> ex1 at pci0 dev 12 function 0: 3Com 3c905-TX 10/100 Ethernet (rev. 0x0)
> ex1: interrupting at ioapic0 pin 16 (irq 11)
> 
> # ifconfig ex1
> ex1: flags=8a63<UP,BROADCAST,NOTRAILERS,RUNNING,ALLMULTI,SIMPLEX,MULTICAST> mtu 1500
>         address: 00:60:98:ef:e3:5d
>         media: Ethernet autoselect (100baseTX full-duplex)
>         status: active
>         inet 192.168.0.11 netmask 0xffffff00 broadcast 192.168.0.255
>         inet6 fe80::260:98ff:feef:e35d%ex1 prefixlen 64 scopeid 0x2
> 
> However it is just statically configured, and I don't know whether it will work in more complex configurations.
> 
> >How-To-Repeat:
> play with 3Com 3c905B-TX
> >Fix:
> N/A

it seems that crash is not related to ex(4)

following is backtrace on today kernel:

uvm_fault(0xccb69464, 0, 1) -> 0xe
kernel: supervisor trap page fault, code=0
Stopped in pid xxxx.1 (ifconfig) at netbsd:in_rtflushall+0x4e: movl  0x38(%edx),%eax

bt:
in_rtflushall(2,c1ef6ee4,cccad928,c1ef6ee4,cccad7d0)
rtrequest1(1,cccad928,cccad95c,1,ccbd1820)
rtinit(c2532100,1,1,246,0)
in_ifinit(c1d0b040,c2532100,cccadba8,0,8040691a)
in_control(c1eeb01c,8040691a,cccadb98,c1d0b040,ccbd1820)
compat_ifioctl(c1eeb01c,8040691,cccadb98,ccbd1820,80206910)
ifioctl(c1eeb01c,8040691a,cccadb98,ccbd1820,c1a5a488)
sys_ioctl(ccbd1820,cccadc48,cccadc68,c0561458,246)
syscall_plain()