[laptop-discuss] [driver-discuss] draft blk2scsa functional spec posted

[Garrett D'Amore]

Artem Kachitchkine wrote:
>
>> Yes, but it also is true that the particular behavior in SCSI-2 
>> reservations prevents excludes FORMAT.  Its this particular need in 
>> SCSI-2, that causes RESERVE and RELEASE to be exposed to blk2scsa 
>> dependent drivers.
>>
>> These commands are are issued by blk2scsa in response to the 
>> corresponding reserve and release commands in the SCSI-2 protocol.
>
> What does blk2scsa do with SCMD's that don't have a corresponding 
> B2S_CMD? Reservation commands could be handled in the same way (faked 
> success). Even if you tried to write a blk2scsa-based driver for 
> devices that support multiple initiators, you got bigger problems and 
> B2S_CMD_RESERVE is not going to help.

I was thinking the same thing earlier this morning.  I originally pushed 
it down to the drivers so that I could preserve queueing order, but I'm 
thinking now that since this only interacts with format, I can just fake 
it in the layer above.  I'll remove it from the spec.

    -- Garrett
>
> -Artem