iscsi target failure ...

Sean Bruno sbruno at miralink.com
Mon Jun 30 18:06:43 UTC 2008


Danny Braniss wrote:
>> Danny Braniss wrote:
>>     
>>> Hi,
>>> 	I'm trying to get the iscsi_initiator to handle errors in a better way,
>>> and so far if if I report to the cam via XPT_DONE, with status = 
>>> CAM_REQ_ABORTED
>>> the kernel panics after printing many
>>> 	g_vfs_done():da1s1d[WRITE(offset=6428672, length=2048)]error = 5
>>> (the offset= changes), 
>>> so I was wandering, is there a better (CAM_REQ_CMP_ERR?) status? the return 
>>> when
>>> a terminal error has occurred?
>>>
>>> danny
>>>       
>
>   
>> Are you talking about the macro XPT_DONE() or the CAM status XPT_DONE?
>>
>> Also, is there a spot in the code that you are specifically looking at?
>>
>>     
>
> Hi Sean,
> 	in iscsi_subr.c/iscsi_cleanup -> _scsi_done->[macro]XPT_DONE()
>
>
>   
How are you attempting to make the failure occur with the initiator?  
Just pulling the cable or some such operation while running an write?

Sean




More information about the freebsd-scsi mailing list