This message is from the T13 list server.
> > Linux has ... bidirectional ATA pass thru defined, > > also Microsoft ... > > > > to add ATA into SCSI, we could emulate ... > > with a command out/ data out command > > followed by a data in/ status in command. > .... > some other solution (such as an ioctl directly > submitting the taskfile) > if SCSI CDB size limitations prove too annoying. Also I've seen proposed a sequence of four: 1) a command out command, promising data in or data out next 2) a data out command 3) a data in command 4) a status in command. That's how SCSI over USB works. > The biggest task is attempting to fit all information I need to properly > set up an ATA taskfile into a 16-byte CDB. Have to fit in 12 bytes to make much ATAPI happy. The theory at t13.org that an ATAPI device can ask for always 16 bytes of Command Out rather than 12 maybe hasn't been widely adopted. Have to fit in 12 bytes - or was that 16 - to make SCSI over FireWire traces look normal. > I may use SCSI's variable CDB standard, or Have to fit in 16 bytes to make SCSI over USB happy. Have to fit in 12 bytes to make legacy look-up-cdb-length-by-op transport happy, popular especially in SPI legacy. Have to fit in 10 bytes to make SCSI-1 transport happy, I mis/remember. Pat LaVarre
