MMC / Mt Fuji:: NormalACA and HiSup bits in INQUIRY data are still not consistent

Bill McFerrin billmc37 at ctesc.net
Tue Apr 7 14:35:21 PDT 2009


* From the T10 Reflector (t10 at t10.org), posted by:
* Bill McFerrin <billmc37 at ctesc.net>
*
Hi,
This is a mess that may be only esoteric. At least I hope it is.
The Host must be able to trust something as axiomatic - inviolable and 
unchanging. INQUIRY data - at least some of it - should be that way.
Byte 0: Bits 4 through 0 specifies the device type (5 for MM devices) 
must be in the minimum set. It has been defined in this way since SCSI-2 
(1993). One could argue that it goes back to SCSI-1, but SCSI-1 did not 
define a device type 5.
Byte 2: This has been "version" since SPC-2 (2001). Its original 
definition as 3 separate version fields was established in SCSI-1 (1985).
SPC-2 tells us that when version = 0, the drive is claiming no 
compliance with any (SPC) standard.
Theoretically, you don't need to trust anything else. As a practical 
matter, the Host needs more "trustable" information, so I'll stop the 
theoretical viewpoint.
The main Fuji, MMC-5 and SPC-3 disagreement is byte 3. I blew it in 
MMC-5 with response data format (bits 3-0). I had requested a unique 
version for MMC and I didn't get it, but I failed to fix MMC-5. This is 
corrected in MMC-6. In the case of Fuji, the Host is supposed to "know" 
that the device is either SCSI or ATAPI in order to determine the actual 
meaning of byte 3 (bits7-4). That may be possible, but typically, USB 
connected devices report as if they were ATAPI. As the editor of MMC-6, 
I cannot change the field definitions of the INQUIRY data specified in 
SPC-3. T10 will not permit it. I would hope for a change to Fuji, but I 
that will also be difficult.
There are additional problems associated with version = 0. Luckily, 
these are resolved by specifying MM device behavior.  You probably don't 
care anyway. For example: the REPORT LUNS command is now mandatory 
according to SPC-3. We are permitted to not implement such a useless 
command when version = 0. I simply need to specify it.
Yes, this does not help much.
Cheers,
Bill
David Burg wrote:
>
> Hello,
>
>  
>
> Last year I reported an inconsistency on 4 fields between MMC and Mt 
> Fuji for the inquiry command response:
>
>  
>
> http://www.t10.org/ftp/t10/t10r/2008/r0804024.htm
>
>  
>
> The following meeting discussed and resolved the conflict on version 
> and response data format fields:
>
>  
>
> http://www.t10.org/ftp/t10/document.08/08-236r0.pdf
>
>  
>
> However it looks like the NormalACA and HiSup bits were overlooked. 
> Should these values be changed to 1b?
>
>  
>
> With regards,
>
>  
>
> David.
>
*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org



More information about the T10 mailing list