[SSI-devel] Re: SSI-1.9 kernel panic - not syncing: __copy_to_user_inatomic:Cannot go remote

[Roger T. <rog...@gm...>]

I'm able to reproduce this panic by trying to dd a remote block
device.  Maybe we could make ssi_rcopy_to_user more graceful?

[user@node1]# dd if=3D/cluster/node3/dev/hda7 of=3Dhda7
dd: reading `/cluster/node3/dev/hda7': Object is remote
0+0 records in
0+0 records out

At node3...
Kernel panic - not syncing: __copy_to_user_inatomic:Cannot go remote


Entering kdb (current=3D0xf6ef9020, pid 197094) due to KDB_ENTER()
kdb> bt
Stack traceback for pid 197094
0xf6ef9020   197094        2  1    0   R  0xf6ef91e0 *icssvr_daemon
EBP        EIP        Function (args)
0xf6f4dc30 0xc028d727 kdb_panic+0x27 (0xc051ae50, 0x0, 0xc060c960,
0xc060c960, 0xc2678000)
0xf6f4dc50 0xc0128b65 notifier_call_chain+0x25 (0xc060c940, 0x0,
0xc060c960, 0xf6f4dc7c, 0x1)
0xf6f4dc6c 0xc011b04c panic+0x8c (0xc03db893, 0xc03c064c, 0x0,
0x40000, 0xf4dc94)
0xf6f4dc94 0xc0138a85 file_read_actor+0x1e5 (0xf6f4dd60, 0xc10602c0,
0x0, 0x1000, 0x40)
0xf6f4dd2c 0xc0138412 do_generic_mapping_read+0x182 (0xf6855a08,
0xf7d8c508, 0xf7d8c4c0, 0xf7d8c4e4, 0xf6f4dd60)
0xf6f4dd7c 0xc0138c62 __generic_file_aio_read+0x192 (0xf6f4dd9c,
0xf6f4dd94, 0x1, 0xf7d8c4e4, 0xb7da0000)
0xf6f4de20 0xc0138de7 generic_file_read+0xa7 (0xf7d8c4c0, 0xb7da0000,
0x40000, 0xf7d8c4e4, 0x3)
0xf6f4de84 0xc024c7d9 rssidev_read+0x79 (0x3, 0xf6f4deac, 0xf6f4df14,
0xf6f4debc, 0xf6f4df04)
0xf6f4df60 0xc024b463 svr_rssidev_read+0xa3
           0xc01009b5 kernel_thread_helper+0x5
kdb>

On 4/15/06, Roger Tsang <rog...@gm...> wrote:
> Hi,
>
> I managed to migrate `dd if=3D/dev/hda
> of=3D<some_filename_on_cfs_at_destination_node>` while dd is stopped.  I
> saw that dd has not migrated on the first attempt to `migrate` the
> process, so I tried `migrate` again which would then give me an error
> -11.  The node where dd was originally invoked crashes when I signal
> dd to continue.  Using OPENSSI-FC (1.9 CVS).
>
> Haven't tried reproducing this yet though and suppose dd actually
> migrated off the node before the crash.  Cluster remains up.
>
> Roger
>
>
> <0>Kernel panic - not syncing: __copy_to_user_inatomic:Cannot go remote
> <4>
> <4>
> kdb> bt
> Stack traceback for pid 199007
> 0xf1b42550   199007        2  1    0   R  0xf1b42710 *icssvr_daemon
> EBP        EIP        Function (args)
> 0xdb731c30 0xc028d727 kdb_panic+0x27 (0xc051ae50, 0x0, 0xc060c960,
> 0xc060c960, 0xd6cd6000)
> 0xdb731c50 0xc0128b65 notifier_call_chain+0x25 (0xc060c940, 0x0,
> 0xc060c960, 0xdb731c7c, 0x1)
> 0xdb731c6c 0xc011b04c panic+0x8c (0xc03db893, 0xc03c064c, 0x0,
> 0x40000, 0x731c94)
> 0xdb731c94 0xc0138a85 file_read_actor+0x1e5 (0xdb731d60, 0xc1390170,
> 0x0, 0x1000, 0x40)
> 0xdb731d2c 0xc0138412 do_generic_mapping_read+0x182 (0xf6853a08,
> 0xf66c6148, 0xf66c6100, 0xf66c6124, 0xdb731d60)
> 0xdb731d7c 0xc0138c62 __generic_file_aio_read+0x192 (0xdb731d9c,
> 0xdb731d94, 0x1, 0xf66c6124, 0xb7da0000)
> 0xdb731e20 0xc0138de7 generic_file_read+0xa7 (0xf66c6100, 0xb7da0000,
> 0x40000, 0xf66c6124, 0x3)
> 0xdb731e84 0xc024c7d9 rssidev_read+0x79 (0x3, 0xdb731eac, 0xdb731f14,
> 0xdb731ebc, 0xdb731f04)
> 0xdb731f60 0xc024b463 svr_rssidev_read+0xa3
>            0xc01009b5 kernel_thread_helper+0x5
> kdb>
>