This message is from the T13 list server.
Andy Warner wrote:
Curtis Stevens wrote:
Jeff
It turns out that during the last T10 meeting I discovered that I have to pass the transfer length in the CDB. I have a was to do that, but it is going to cost all the transfer bits. I will have a new command doc out next week.
Two comments on T10/04-262r3:
1. My linux implementation also finds itself needing to be told the direction of DMA operations. I pick it up from SCSI housekeeping structures (which is where I also get the transfer length from.) If the direction of each transfer was explicitly stated in the CDB, I think it would be more self contained.
Well, TBH I think it's redundant information since libata (the Linux implementation, for our fair readers) already carries the data direction implicitly. I cannot see a case where the data direction is not available at all.
2. I am struggling to know how to use the Bus Idle protocol, but I can believe that it may be useful to a hardware implementation that was not built on top of an already complete SATA controller. I just thought I'd point that out.
Bus Idle, UDMA-In and UDMA-Out should be removed, IMHO.
Jeff
