geom_gate panic

[Daniel Eriksson]

Using geom_gate to mount a geom_stripe array from another machine (hooked up
through a crossover cable), the following happens when moving large amounts
of data:

g_vfs_done():ggate0[WRITE(offset=59507982336, length=131072)]error = 5
...
g_vfs_done():ggate0[WRITE(offset=1119618220032, length=131072)]error = 5
g_vfs_done():ggate0[WRITE(offset=1062713851904, length=16384)]error = 5
g_vfs_done():ggate0[WRITE(offset=1101811367936, length=49152)]error = 5
initiate_write_filepage: already started
panic: handle_written_filepage: attached
cpuid = 1
KDB: enter: panic
[thread pid 3 tid 100044 ]
Stopped at      kdb_enter+0x32: leave   
db> tr
Tracing pid 3 tid 100044 td 0xc1eb65c0
kdb_enter(c075cb66,1,c07692b8,e06c5c00,0) at kdb_enter+0x32
panic(c07692b8,c07a9d80,46,c1eb65c0,2) at panic+0x1e9
softdep_disk_write_complete(d1872668,246,c1eb65c0,0,e06c5ca0) at
softdep_disk_write_complete+0x839
bufdone(d1872668,c1eb65c0,0,0,0) at bufdone+0x2c2
g_vfs_done(c245c000,c07b2f48,24c,c0758cf2,c8) at g_vfs_done+0x69
g_io_schedule_up(c1eb65c0,c1eb91f8,e06c5d34,c0553252,0) at
g_io_schedule_up+0x70
g_up_procbody(0,e06c5d48,75003c80,74f73bfa,48a4e10) at g_up_procbody+0x1e
fork_exit(c052a2e0,0,e06c5d48) at fork_exit+0x62
fork_trampoline() at fork_trampoline+0x8
--- trap 0x1, eip = 0, esp = 0xe06c5d7c, ebp = 0 ---


Server: FBSD 6-CURRENT, UP, 2004.12.20.16.00.00
Client: FBSD 6-CURRENT, SMP, 2004.12.20.16.00.00

This happens with or without debug.mpsafenet=0.

I've seen the same error-messages when mounting a filesystem backed by a
normal disc (instead of the geom_stripe used above), but I aborted the
file-move before the machine had a chance to panic.

/Daniel Eriksson