Re: Double mutex destruction (from loss of gateway route?)

On Fri, Aug 17, 2007 at 01:45:26PM +0000, Alexander Kabaev wrote:
> Hi,
> 
> I left development box running 7.0 sitting idle at work overnight and
> was greeted with the following panic this morning:
> 
> panic: mtx_lock() of destroyed mutex _at_ ../../../net/route.c:1303
> cpuid = 1
> KDB: enter: panic
> [thread pid 4088 tid 100128 ]
> Stopped at      kdb_enter+0x32: leave
> db> bt
> Tracing pid 4088 tid 100128 td 0xc5539cc0
> kdb_enter(c07c56bd,1,c07c4619,f199f9bc,1,...) at kdb_enter+0x32
> panic(c07c4619,c07d2ce5,517,507,c5386a7c,...) at panic+0x124
> _mtx_lock_flags(c5aeb510,0,c07d2ce5,517,f199fa28,...) at _mtx_lock_flags+0x65
> rt_check(f199fa20,f199fa3c,c541ec30,c080a6c4,c07d2cee,...) at rt_check+0x111
> arpresolve(c51e4000,c5584bb8,c5587d00,c541ec30,f199fa56,...) at arpresolve+0xb0
> ether_output(c51e4000,c5587d00,c541ec30,c5584bb8,c5539cc0,...) at ether_output+0x7e
> ip_output(c5587d00,0,f199fac4,0,0,...) at ip_output+0x9f5
> udp_send(c5a72000,0,c5587d00,c5a17970,0,...) at udp_send+0x597
> sosend_dgram(c5a72000,c5a17970,f199fbe8,c5587d00,0,...) at sosend_dgram+0x351
> sosend(c5a72000,c5a17970,f199fbe8,0,0,...) at sosend+0x54
> kern_sendit(c5539cc0,4,f199fc64,0,0,...) at kern_sendit+0x104
> sendit(0,28228a34,0,c5a17970,10,...) at sendit+0x182
> sendto(c5539cc0,f199fcfc,18,c07c1d9e,c0804e18,...) at sendto+0x4f
> syscall(f199fd38) at syscall+0x29a
> Xint0x80_syscall() at Xint0x80_syscall+0x20
> --- syscall (133, FreeBSD ELF32, sendto), eip = 0x28152f23, esp = 0xbfbfe2cc, ebp = 0xbfbfe3e8 ---
> 
> 
> Is this one known? Unfortunately, vmcore from this panic got garbled somehow.

It looks like a panic I was having when the gateway route disappeared
from a sparc64 system.  The switch seemed to be dodgy and would
randomly drop the link, and the system would panic.

Kris