This message is from the T13 list server.
Steve Livaccari (IBM) said: > It's back because I need this function to ship > product and stay in business. But the proposal makes no sense. The proposal says (and note my "comments" inclosed in [] within this quoted text): 5 DRQ=0 when ERR=1 Function During the processing of a command devices set either BSY=1 or DRQ=1. DRQ=1 indicates the device is requesting data [transfer]. DRQ may be asserted [signals are asserted, DRQ is not a signal] when an error has occurred (ERR=1). This behavior is a legacy of buffer management where parallel ATA devices would transfer data to their buffer and set DRQ before detecting the error. The recommended Serial ATA protocol is to disallow the DRQ=1 and ERR=1 condition (Design Guide 21). If this feature is implemented on a Serial ATA drive the drive is guaranteed to do the recommended behavior. This feature allows the host to require a device to clear DRQ to zero when ERR is set to one, so the error data does not have to be transferred to end the command. This legacy buffer management you talk about is a violation of the ATA PIO command protocols going back to ATA-3. So you want to add a feature that must be enabled in order for a device to conform to the ATA/ATAPI-x standards? That makes so sense at all to me. There must be something about this that you are not telling us? It sure would be nice to know what problem this proposal is really trying to fix. I could understand a feature like this if it was documented as "when enabled the feature causes the device to be ATA-1 or ATA-2 compatible and allows BSY=0 DRQ=1 ERR=1" status. But who would want that these days? Anyone reading the the ATA/ATAPI-x PIO command state diagrams would know that BSY=0 DRQ=1 ERR=1 status is not valid; and it looks like the SATA folks wanted to make sure people understand this too. [I am extremely disappointed that T13 would spend time on things like this while refusing to addressing the very serious problems that people are seeing every day using SATA (especially random and undetected errors that frequently result in data corruption). Maybe it is time for T13 to give up and go away and let some other organization take over?] Hale *** Hale Landis *** www.ata-atapi.com ***
