Hi,

  on two of our systems, lfs has taken on killing the kernel within
seconds of boot; there is no way to use the systems sensibly anymore
other than mounting the lfs partitions read-only.

  What I get is:

| lfs_segwrite: possibly invalid checkpoint!
| lfs_segwrite: ifile still has dirty blocks?!
| bp=0xca38e3e8, lbn 21, flags 0x24080
| bp=0xca37c5f8, lbn 20, flags 0x24080
| panic: dirty blocks
| Stopped in pid 14.1 (ioflush) at   cpu_Debugger+0x4:    leave
| db> bt
| cpu_Debugger(0,e3399000,0,x02f5d6e,c059862e) at cpu_Debuuger+0x4
| panic(c059864c,e36e7000,e3399000,1,c11b5000) at panic+0xb8
| lfs_segwrite(c11b6e00,5,e3375f28,e3364694,0) at lfs_segwrite+0x4df
| lfs_sync(c11b6e00,3,c1016f00,e3364694,0) at lfs_sync+0x78
| sync_fsync(e3375f28,e09c5,e3375f4c,c033462b,e33968c8) at sync_fsync+0x05c
| VOP_FSYNC(e33968c8,c1016f00,8,0,0) at VOP_FSYNC+0x58
| sched_sync(e333e680,0,0,0,c010030c) at sched_sync+0xd5
| db>

The only thing I can see is that the filesystem filled to over 90%
before that happened; there are no other peculiar things that were
happening on that machine.

Unfortunately, I can't keep the machines in this state to do any tests
or trials on; I will be converting the partitions back to ffs today in
order to have the systems running again.

Konrad