Re: [PATCH scsi-rc-fixes-2.6] mvsas: fix error handling.

[Ke Wei <kecore@gmail.com>]

James Bottomley wrote:
> On Fri, 2008-04-25 at 11:51 -0400, Jeff Garzik wrote:
>> Ke Wei wrote:
>>>  Changed error handling: Add a internal queue to send command in mvs_task_abort function.
>>>  Set max_id and max_lun to a valid value. It's good for 64xx chip.
>>>  BTW, DVD-ROM compatibility needs to test carefully.
>>>
>>>
>>> Signed-off-by: Ke Wei <kewei@marvell.com>
>> James, ping?
>>
>> Did this get missed?  Are we waiting on Marvell to update this patch?
> 
> Actually, the patch contains a red flag:
> 
>> -       mvi->shost->max_id = 21;
>> -       mvi->shost->max_lun = ~0;
>> +       mvi->shost->max_id = 128;
>> +       mvi->shost->max_lun = 1;
> 
> Setting max_lun to 1 can't be right.  My suspicion is that it's done to
> turn off the REPORT_LUN scan that I've already said twice is necessary.
> 
> The DVD that's showing the problems needs to be blacklisted.
> 
> So, it's on my list to check carefully to see what else might have been
> quietly slipped in there ... unfortunately, I haven't had much time
> recently (and probably won't until the merge window closes) to give this
> patch the thorough scrutiny it needs.
> 

First, Thank you for your help, Jeff and James. 
I am investigating whether other devices have the same issue as my testing DVD caused by REPORT_LUNS command.
So I think I doesn't need to add this device to blacklists before I check. My colleague is working on the libsas module so that we hope SAS transport is more powerful and stable.

Ke