This message is from the T13 list server.
The history... Alexander Krebs wrote:
This message is from the T13 list server. Hi all, Could anybody please explain why the Features registers hold the sector count whereas the Sector Count registers hold the tag for the READ/WRITE FPDMA QUEUED commands? This looks weird to me.
Many years ago, the ATAPI folks, working outside of X3T9.2 and X3T10, developed the idea of "tagged command queuing" for ATAPI devices. The PACKET command uses the Sector Count register as the Interrupt Reason register and the tag value is need along with the interrupt reason data, so they put the tag value into this register. I've never seen an ATAPI device that implements "tagged command queuing".
Also many years ago, some ATA hard disk folks thought that ATA devices should also have "tagged command queuing". These ATA folks decided that since the tag data is needed by host controller hardware at interrupt time, the tag data should also be in the same bits of the Sector Count. Of course this means that the ATA sector count must be moved somewhere else and the only place was to the Features register. There have been a few ATA devices and host controllers that implement "tagged command queuing".
Most recently we can only assume that the SATA Secret Society also decide to maintain the tag data in the same ATA registers for backward compatibility. But of course we will never know because we don't see the minutes of their meetings, or even who attended their meetings, and perhaps worse, T13 is not allowed to discuss SATA items without the prior approval of the Secret Society. I've have not seen any indication that T13 has been given permission to discuss the First Party DMA QUEUED commands. A question for the T13 officials would be: Has such approval been given and where is the letter that says this (what is the T13 doc number for that letter)?
Anyone notice how the ANSI standards processes has been converted from a mostly "open" process to a very "closed/private/secret" process? Does anyone care?
Hale -- ++ Hale Landis ++ www.ata-atapi.com ++
