This message is from the T13 list server.
Sergei, You are correct, the device ID should not be forced by a spec/standard. Many companies control the device ID, for one reason or another, and cannot have this dictated to them. My own company identifies similar (ATA) devices with different device ID's mainly because we potentially step old devices to fix production issues while new devices are in development or also being manufactured. This makes identifying different devices through the rev ID very difficult to "guess" how many steppings (and rev ID's) one must reserve for the future. Other company-specific issues exist that dictate how the device ID is used, which each company will definitely have their own quirks. The registers you enumerate (below) are decent for device capabilities detection, but putting an identification field as part of the ADMA memory mapped registers would be a more flexible "standard" method for determining if a controller meets this proposed standard. The id field is used in other controller types, and would work fine here too. Another method is to require a new capabilities pointer, but this would only work for ADMA as a brand new device type unencumbered by the legacy ATA controller features. Sincerely, Michael K. Eschmann Intel Storage Solutions -----Original Message----- From: Sergei Shtylyov [mailto:[EMAIL PROTECTED]] Sent: Saturday, February 23, 2002 1:38 AM To: [EMAIL PROTECTED] Subject: [t13] D1510: PCI device ID 1841h? This message is from the T13 list server. Hello, T13. My question is: why it's requered for ADMA compatible device to have some pre-assigned PCI device ID? Isn't it discriminated enough by the base class & subclass & programming interface registers? I've expected this error to be gone after the 1st revision of draft but it's still here in the latest one... -- Sergei Shtylyov E-mail: [EMAIL PROTECTED] FidoNet: 2:5020/118.24 Subscribe/Unsubscribe instructions can be found at www.t13.org. Subscribe/Unsubscribe instructions can be found at www.t13.org.
