Subject: NetBSD NFS bug?
To: None <>
From: Neil A. Carson <>
List: tech-kern
Date: 03/31/1999 14:41:48

I'm using a 1.3-I NetBSD x86 system in a network of other NetBSD
Intel/PowerPC/StrongARM machines, Linux/Intel machines, and
FreeBSD/Intel machines. A common panic occurs on the NetBSD machine when
it's being served from the Linux server (doesn't happen with a NetBSD
server or FreeBSD server). An example panic is thus:

     > panic: getblk: block size invariant failed
     > Stopped in cpio at      _Debugger+0x4:  leave
     > db> trace
     > _Debugger(0,f102a74c,f80e4d70,f0186944,f0186838) at _Debugger+0x4
     > _panic(f0186838,f0668000,f7fe9888,2000,f062cc84) at _panic+0x55
     > _getblk(f7fe9888,0,2000,0,0) at _getblk+0xe8
     > _nfs_getcacheblk(f7fe9888,0,2000,f81277dc,1) at
     > _nfs_write(f80e4ea4,f80e4f88,200,f81277dc,f80e4ea4) at
     > _vn_write(f80f8658,f80f8670,f80e4ef0,f06f0380,1) at
     > _dofilewrite(f81277dc,4,f80f8658,12000,200) at _dofilewrite+0xd3
     > _sys_write(f81277dc,f80e4f88,f80e4f80,0,4) at _sys_write+0x43
     > _syscall() at _syscall+0x1c2
     > --- syscall (number 4) ---
     > 0x40093a37:
     > db>

Has anyone seen these before? They happen every couple of days (fairly
often) and I haven't had time to debug them, and wonder if someone
(Charles?) might have already fixed this one in -current...