Re: [smartmontools-support] [smartmontools-devel] [PATCH] atacb support (fwd)

[matthieu c. <cas...@fr...>]

Hi Douglas,

Douglas Gilbert wrote:
> Bruce,
> It certainly looks worth including. The pass-through looks
> remarkably like SAT (i.e. SCSI ATA PASS_THROUGH(16) command)
> but uses opcode 0x24. That is a bit strange (but most likely
> not fatal) as that range of SCSI commands are reserved for
> 10 byte long, t10 defined SCSI commands.
Is more like ATA PASS_THROUGH(12) as it doesn't support LBA48.
Also do you know if there is some usb device supporting SAT.
After reading some specs SAT and mass storage seems incompatible :
SAT are based on spc3 but mass storage on spc2 and there is some 
incompatibility with sense format.


> 
> The pdf seems to indicate (table 6) that the command opcode
> byte (i.e. 0x24) is taken from eeprom. That implies that
> products using the same chip could have that "magic" ATACB
> command at other locations in the SCSI opcode range (which
> is 0 to 0xff).
Yes, I suppose we could make 0x24 the default and override it with "-d 
atacb,opcode" option.

> 
> 
> A question to Matthieu: can the fact that a cy7c68300c_8
> chip is present in a USB (external) device be detected
> from the OS?
It suppose it can be detected by usb id.
On mine they also put "Cypress AT2LP" in the usb product string.
But may be we could try the atacb command and check if there is an error 
reported in the sense (my previous patch doesn't do that ATM) ?

Matthieu

PS : I also start a SAT emulation in Linux mass storage driver : 
http://thread.gmane.org/gmane.linux.usb.general/2050