On Tue, Nov 15, 2011 at 12:12 PM, Hannes Reinecke <hare@xxxxxxx> wrote: > On 11/15/2011 11:00 AM, Vasiliy Tolstov wrote: > > [ 1152.782672] Call Trace: > > [ 1152.782760] [<ffffffffa010dd13>] ? __scsi_queue_insert+0xbe/0xe5 [scsi_mod] > > [ 1152.782862] [<ffffffffa010e125>] ? scsi_io_completion+0x3eb/0x3fa [scsi_mod] > > [ 1152.782964] [<ffffffffa010abb0>] ? scsi_eh_flush_done_q+0xe3/0x104 > > [scsi_mod] > > [ 1152.783093] [<ffffffffa010bdc3>] ? scsi_error_handler+0x3cb/0x5b5 [scsi_mod] > > [ 1152.783195] [<ffffffffa010b9f8>] ? scsi_error_handler+0x0/0x5b5 [scsi_mod] > > [ 1152.783295] [<ffffffff81065c69>] ? kthread+0x79/0x81 > > [ 1152.783389] [<ffffffff81012baa>] ? child_rip+0xa/0x20 > > [ 1152.783482] [<ffffffff81011d61>] ? int_ret_from_sys_call+0x7/0x1b > > [ 1152.783577] [<ffffffff8101251d>] ? retint_restore_args+0x5/0x6 > > [ 1152.783673] [<ffffffff8100ef4f>] ? xen_restore_fl_direct_end+0x0/0x1 > > [ 1152.783769] [<ffffffff81012ba0>] ? child_rip+0x0/0x20Ouch. Requeue while the queue is dead. > > I'm pretty sure we need to fix the error handler for this case. Apparently when a SCSI host is removed the SCSI queue is killed from inside __scsi_remove_device() before the error handler thread is stopped from scsi_host_dev_release(). That order doesn't seem correct to me. Bart. -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html