Prelim HSC50 HostIntfSpec 0.9 Jul9

This document is a "Preliminary HSC50 Host Interface Specification, Version 0.9," dated July 18, 1979, with a print date of April 9, 1982. It outlines the interface between a host computer and the Digital Equipment Corporation's HSC50 Mass Storage Controller.

Key aspects of the specification include:

1. Core Updates: This version incorporates a new packet format from the Port and Protocol Task Force, expands the host port ID to 32 bits, defines the controller's device class as DV.MSC for mass storage, includes minor revisions to characteristics and status block formats, and requires all reserved fields to be zero.
2. Terminology & Introduction: It provides an introduction to the interface and defines key architectural terms like Subsystem, Controller, Device Classes (Disk and Tape), Class Driver, Port Driver, and Unit. It also defines disk and control protocol terminology, including block, logical block number (LBN), replacement block number (RBN), and packet.
3. Host & Controller Responsibilities: Detailed sections delineate the functional responsibilities of both the controller (e.g., diagnostics, status change notification, error reporting, command integrity, data integrity, formatting, bad block management, shadowing) and the host (e.g., interface diagnostics, resynchronization, command validity, managing controller operation, unique command reference numbers, command timeouts, error recovery, bad block revectoring, multi-host coordination, error logging).
4. Physical Interconnect & Protocol: It describes the assumptions and guarantees required of the underlying physical interconnect and control packet mechanism, including message ordering, error detection/recovery, and bus-level initialization functions. The control protocol is described as having four logical levels, operating in a master/slave relationship where the host dominates, and supporting asynchronous command execution.
5. Logical States and Transitions: The document extensively defines the logical states for the subsystem (Controller Offline/Online) and individual units (Unit Offline/Available/Online) relative to a host, and details the rules governing transitions between these states, including diagrams for clarity.
6. Command Flow & Error Recovery: It illustrates the general flow for host-to-controller commands like READ and WRITE, explaining how commands are processed and data transferred. It specifies host error recovery procedures for various command failures, involving resynchronization and reinitialization, and includes a detailed recovery sequence diagram.
7. Connection Management: It outlines the automatic establishment and orderly termination of logical connections between the host and controller, detailing a configuration sequence for initial setup.
8. Messages & Shadowing: It describes the various controller-to-host messages: END (command completion), ATTENTION (status changes), LOG (error information), and PSUEDO-TERM OUT (terminal output). It also details the functionality and rules for "shadowing" disk units for data duplication, including command restrictions and error handling.
9. Packet Formats & Commands: General command packet formats are provided for both host-to-controller commands and controller-to-host messages, specifying byte offsets for fields like opcode, device class, unit, reference number, and arguments. It then specifies numerous host-to-controller commands (e.g., ABORT, COMPARE, DOWNLINE LOAD, EXECUTE, FLUSH, GET STATUS, INIT, INVALIDATE, NOP, POSITION, READ, SET STATUS, WRITE) and their "gatekeeper classes" which define their processing behavior.
10. Definition Tables: The document concludes with comprehensive tables defining opcodes, modifiers, condition codes, status formats, log information formats, characteristics formats, device/unit numbers, and priority levels.