This message is from the T13 list server.
I wrote a possibility. >In case of native SATA ODD, it can report Check Condition when DMA >direction is incorrect. Sense data may be "invalid field in CDB"? >Please no. >By the t10 English, SK ASC x 5 24 means CDB[0] known but CDB contains >unsupported values in fields not ignored as reserved. I have seen a >vendor-specific choice of SK ASC x 2 FF used because t10 seemingly gave >no clear guidance on what SK ASC to use to say host & device disagreed >over direction & length. We might prefer falsely reporting a UDMA CRC >error? Phase failure can happen parallel SCSI. In case of SCSI Message Phase, it should be transport layer failure that application software cannot know. It is similar with CRC failure. Data In / Data Out Phase of a SCSI transaction is assigned by application software not by port driver. And the mismatch is checked by drive comparing with SCSI ope-code. So it could be handled as an application error. If ordinary PATA port driver has such error code/message for application, DMADIR mismatch should use same error code. CRC failure is physical failure, so it may not be the best solution. User may want to change the SATA cable instead of application update to fix the error code. cam-r12b.pdf has following section. 11.3.8.6 Command reception errors and data phase errors handling. gpp-r09.pdf has following section. C.2.1 INVALID BUS PHASE DETECTED So SCSI port driver may have some error handling way. Let me confirm. PATA interface drive cannot say the DMADIR error. There is no such mechanism in PATA. SATA-PATA bridge can check the error using PIO mode at PATA side. SATA drive can check the error. Creating something new for SATA only means incompatibility with PATA/SCSI. Best regards, Keiji Katata PIONEER CORP.
