Skip to content

SCSI Optical Disk Driver (IP-P2-SCSI-OPDI)

Module: IP-P2-SCSI-OPDI.NPL Module ID: 8SCOD (SCSI Optical Disk) Type: Level 11 Driver Purpose: SCSI optical disk operations with write recovery and device management

1. Overview

The SCSI Optical Disk Driver (IP-P2-SCSI-OPDI.NPL) is a specialized driver for optical storage devices in the Sintran L operating system. Unlike magnetic disk drivers, this driver implements sophisticated write recovery mechanisms, strict device initialization, and optical-specific optimizations.

Key Differences from Magnetic Disks

Optical disks have fundamentally different characteristics than magnetic disks:

Characteristic Optical Disk Magnetic Disk
Write Speed 10-100x slower Fast
Seek Time 100-500ms 5-15ms
Media Type WORM/Rewritable Rewritable
Write Recovery Block-level verification Simple retry
Initialization Mandatory MODE SELECT Optional
Queue Sorting Not beneficial (SEEK) Highly beneficial
Cache Usually disabled Usually enabled

Optical-Specific Features

  1. Write Recovery (CCRWO): Block-level COMPARE and selective rewrite after power failures
  2. WORM Support: Write-Once-Read-Many media handling
  3. Device Type Validation: Only types 3 (optical) and 4 (WORM) supported
  4. Mandatory MODE SELECT: Required device configuration before operations
  5. No Queue Sorting for SEEK: Long seek times make sorting ineffective
  6. BLANK CHECK Handling: Expected condition during write recovery

2. Complete API Reference

Supported Functions

Function Name Description Double Address
0 READ Read data blocks from disk 60
1 WRITE Write data blocks to disk 61
2 READ PARITY Read with parity checking 62
3 COMPARE Compare disk data with memory 63
4 SEEK Position read/write head -
34 RESERVE DEVICE Reserve device for exclusive use -
35 RELEASE DEVICE Release reserved device -
37 READ EXTENDED STATUS Retrieve SCSI sense information -
42 READ FORMAT Get disk format and capacity -
73 TEST UNIT READY Check if device is ready -
74 USER COMMAND Execute custom SCSI command -
75 INQUIRY Read device type information -

Function Entry Point: SCOPTICAL

Entry Parameters: - X = DAQ (Disk Activity Queue) pointer - X.NFUNC = Function code (0-75) - X.ABFUN = Low 7 bits contain function code - X.MEMAD = Memory address for data transfer - X.ABPA2 = Disk address (single or double word) - X.ABP32 = Transfer amount (blocks) - L = Return address

Exit Parameters: - T = Status code (0 = success, non-zero = error) - A = Sense data (if T=0) - X = Updated DAQ pointer - SCTRG = Operation status (function code + error code)

SCOPTICAL Routine

Purpose: Main dispatcher for all optical disk operations

Key Operations: 1. Extract and validate function code 2. Convert single to double addresses if needed 3. Validate unit data field 4. Queue management and sorting (if applicable) 5. Device initialization (INQUIRY + MODE SELECT) 6. Write recovery check (4SRWO flag) 7. Execute SCSI command via SCSID 8. Sense data validation 9. Post-processing (function 42)

RETOP - Operation Termination

Purpose: Terminate operation and update status

Entry Parameters: - B = Disk data field - X = DAQ pointer - T = Driver status - A = Sense data (if T=0)

Key Operations: 1. Translate sense data to error codes 2. Handle COMPARE special case (miscompare = success) 3. Log errors if 3SERR flag set 4. Return to level 11 scheduler

Status Translation (NEWST Array):

Sense Value Meaning Error Code
0 NO SENSE No error 000000
1 RECOVERED ERROR Device corrected error 000000
2 NOT READY Device not ready 100020
3 MEDIUM ERROR Defect in media 100020
4 HARDWARE ERROR Device hardware failure 100020
5 ILLEGAL REQUEST Invalid command 100020
6 UNIT ATTENTION Device reset/media changed 100020
7 DATA PROTECT Media write-protected 000020
8 BLANK CHECK Unwritten blocks 000020
9 VENDOR SPECIFIC Vendor-specific error 100020
10 COPY ABORTED Copy command failed 100020
11 ABORTED COMMAND Command interrupted 100020
12 EQUAL COMPARE matched 000000
13 VOLUME OVERFLOW End of partition 100020
14 MISCOMPARE COMPARE failed 000020
15 RESERVED Reserved sense code 100020

MDSEL - MODE SELECT

Purpose: Configure optical disk parameters

Entry Parameters: - X = DAQ pointer - SUTYP = Device subtype (bits 10-13 = device type) - X.CMAD1:CMAD2 = Physical buffer address

Key Operations: 1. Check device type (3 = optical, others = default) 2. Build MODE SELECT command block 3. Set parameter count (4 bytes) 4. Execute via SCSID

MODE SELECT Parameters: - Block size configuration - Write verification mode - Error recovery parameters - Cache control (typically disabled)

TER42 - READ FORMAT Post-Processing

Purpose: Return disk format and capacity information

Entry Parameters: - X = DAQ pointer - SURSZ = Sector/record size - X.MEMA1:MEMA2 = User buffer address

Exit Data Structure:

Offset 00: Disk Layout Index
           41 (51 decimal) = 2KB sectors (SURSZ=4000 octal)
           40 (50 decimal) = 1KB sectors (SURSZ=2000 octal)
           -1 = Non-standard size

Offset 10: Available Blocks (Double word)
           Total capacity - 1MB test reserve
           Calculated as: total - (SURSZ × 18432 / 15)

Test Area Reservation: The driver reserves approximately 1 megabyte for testing and calibration. This is calculated as: 

Reserved blocks = (SURSZ × 18432 / 15) + 1
Available blocks = Total capacity - Reserved blocks

CHSEN - Sense Data Validation

Purpose: Determine if sense data indicates recoverable error

Entry Parameters: - A = Sense data from SCSI command - X = DAQ pointer - B = Disk data field

Actions:

  1. Sense 0-1 (Acceptable): Return to caller, continue operation
  2. Sense 6 (Unit Attention): Decrement retry counter, RETRY
  3. Sense 13 (Aborted Command): Decrement retry counter, RETRY
  4. Other Sense: Jump to RETEX with error status

Error Handling Routines

ERRET - Hardware Error Recovery

Handles hardware failures with retry logic: - PFAIL: Power failure - SBRST: SCSI bus reset - LIRST: Link interface reset

For WRITE operations (functions 1, 61): 1. Set 4SRWO flag (write recovery needed) 2. Decrement retry counter 3. Jump to RETRY

ERREX - Error Exit Preparation

Prepares error exit with status code 20 (octal 16 decimal).

ERR1 - Early Error Exit

Handles configuration errors: - ILAOP: Illegal operation (invalid function code) - NOLUN: No logical unit (unit not configured)

RETEX - Return with Status

Returns from operation: 1. Clean up queue pointers 2. Call RETOP to terminate 3. Check for next queued operation 4. Start next operation or return to scheduler

3. Write Recovery Mechanism (CCRWO)

The CCRWO (Check/Compare/Rewrite Operation) routine is unique to the optical disk driver and implements sophisticated write recovery after hardware failures.

Why Write Recovery?

Optical disk characteristics make write recovery essential:

  1. Slow Write Times: Optical writes take 10-100x longer than magnetic
  2. WORM Media: Some optical media is write-once, cannot simply rewrite
  3. Power Failure Resilience: Long writes increase probability of interruption
  4. Partial Write Success: Some blocks may write successfully before failure
  5. Data Integrity: Must verify what was actually written

Recovery Strategy

The recovery process uses a two-phase approach:

Phase 1: COMPARE - Verify Written Blocks 1. Build SCSI COMPARE command 2. Set original disk address 3. Set original block count 4. Set original memory address 5. Execute COMPARE via function 74 (user command)

Phase 2: REWRITE - Selective Block Rewrite 1. If BLANK CHECK returned: some blocks failed 2. Get failing block address from sense data 3. Calculate remaining blocks to write 4. Adjust memory address to failed block offset 5. Build SCSI WRITE command for remaining blocks 6. Execute WRITE via function 74

CCRWO Entry Point

Trigger: 4SRWO flag set in OPSTA field

When Set: - Write operation (function 1 or 61) in progress - Hardware error occurred (PFAIL, SBRST, or LIRST) - ERRET routine sets flag before retry

Entry Check (Line 226): 

IF X.OPSTA BIT 4SRWO GO FAR CCRWO

COMPARE Command Construction

Command Block Format: 

Byte 0-1: 27402 (octal) = SCSI COMPARE command
Byte 2-5: Device block address (from original ABPA2)
Byte 6-8: Number of blocks (from original ABPA3)
Byte 9-11: Byte count (blocks × block size)

Memory Address: Original memory address (data being written)

Function: 74 (user-specified SCSI command)

COMPARE Results

Case 1: All Blocks Match (Sense = 0) - All blocks written successfully before failure - Recovery complete - Jump to RETEX with success - Next operation clears 4SRWO flag

Case 2: Some Blocks Failed (Sense = 8, BLANK CHECK) - Hardware returns failing block address - Indicates first unwritten or mismatched block - Proceed to rewrite phase

REWRITE Command Construction

Calculate Remaining Work: 

Original: Write N blocks starting at LBA X from memory M
COMPARE failed at: LBA Y

Remaining blocks = N - (Y - X)
New start LBA = Y
New memory address = M + ((Y - X) × block_size)

Command Block Format: 

Byte 0-1: 25000 (octal) = SCSI WRITE command
Byte 2-5: Failing block address (from COMPARE sense)
Byte 6-8: Remaining block count
Byte 9-11: New byte count

Memory Address Calculation: 

AD SHZ -1                    % Byte count → word count
A:=:D+X.MEMA2=:X.MEMA2      % Add to low word
A:=D+C+X.MEMA1=:X.MEMA1     % Add carry to high word

Recovery Examples

Example 1: Successful Recovery 

Initial Write:
- Function: 1 (WRITE)
- Address: LBA 5000
- Count: 20 blocks
- Memory: 0x20000

Event: Power failure after block 12

Recovery:
1. Set 4SRWO flag
2. Next operation checks 4SRWO → CCRWO
3. COMPARE 20 blocks at LBA 5000
4. BLANK CHECK at LBA 5012
5. Calculate: 20 - 12 = 8 blocks remain
6. Memory: 0x20000 + (12 × 512) = 0x23000
7. WRITE 8 blocks at LBA 5012 from 0x23000
8. Success - clear 4SRWO

Result: All 20 blocks written, no data loss

Example 2: Multiple Retry Attempts 

Initial Write:
- Function: 61 (WRITE double address)
- Address: LBA 100000
- Count: 50 blocks
- Memory: 0x40000

Event: SCSI bus reset during write

Attempt 1:
1. Set 4SRWO, retry = 5
2. COMPARE → BLANK CHECK at LBA 100025
3. WRITE remaining 25 blocks
4. Hardware error → retry = 4

Attempt 2:
1. COMPARE → BLANK CHECK at LBA 100025
2. WRITE remaining 25 blocks
3. Success
4. Clear 4SRWO

Result: Recovery required 2 attempts, all blocks written

Example 3: Unrecoverable Error 

Initial Write:
- Address: LBA 2000
- Count: 10 blocks

Event: Link reset during write

Attempts 1-5:
- Each attempt fails at LBA 2005
- Hardware error persists
- Retry counter exhausted

Result:
- Blocks 2000-2004: Written successfully
- Blocks 2005-2009: Failed
- Error returned to caller
- 4SRWO remains set

Design Rationale

Why COMPARE Instead of READ?

  1. Efficiency: COMPARE is faster (no data transfer)
  2. Accuracy: Atomic comparison on device (no race conditions)
  3. BLANK CHECK: SCSI provides specific sense with failing address
  4. Precision: Identifies exact block that failed

Why Block-Level Recovery?

  1. WORM Media: Cannot rewrite successfully written blocks
  2. Performance: Rewriting only failures saves time on slow optical media
  3. Reliability: Preserves partial success from long writes
  4. Wear Reduction: Minimizes rewrites on optical media

Why User Command (Function 74)?

  1. Flexibility: COMPARE not a standard disk function
  2. Control: Direct SCSI command block construction
  3. Compatibility: Standard SCSI COMPARE command
  4. Recovery: Full control over recovery process

Recovery Limitations

Cannot Recover: 1. Media Defects: Physical damage to optical disk 2. Device Failure: Drive hardware malfunction 3. Bad Blocks: Permanently unwritable blocks

Retry Count: - Initialized from TACNS (typically 3-5) - Decremented on each recovery attempt - Error returned when exhausted - Prevents infinite loops

Flag Persistence: - 4SRWO persists until successful recovery - Ensures recovery attempted on next operation - Manual intervention may be needed if all retries fail

4. Mermaid Diagrams

Optical Disk Operation Flow

flowchart TD
    A[SCOPTICAL Entry] --> B{Valid Function?}
    B -->|No| C[ERR1: ILAOP]
    B -->|Yes| D[Convert Addresses if Needed]
    D --> E{Unit Valid?}
    E -->|No| F[ERR1: NOLUN]
    E -->|Yes| G{Queue Sorting?}
    G -->|Yes, SLINK valid| H[DSORT]
    H --> I[SWT11 Return]
    G -->|No / New Op| J[NEWOP: Initialize Operation]
    J --> K{Device Initialized?<br/>4SINI flag}
    K -->|No| L[Execute INQUIRY<br/>Function 36]
    L --> M{Device Type 3 or 4?}
    M -->|No| N[ERREX: TYPER]
    M -->|Yes| O[Set 4SINI Flag]
    O --> P{MODE SELECT Done?<br/>4SMSL flag}
    K -->|Yes| P
    P -->|No| Q[Execute MDSEL]
    Q --> R[Set 4SMSL Flag]
    P -->|Yes| S{Write Recovery?<br/>4SRWO flag}
    R --> S
    S -->|Yes| T[CCRWO<br/>Write Recovery]
    S -->|No| U[Execute SCSID]
    T --> V[RETEX Success]
    U --> W{Error?}
    W -->|Yes| X[ERRET/ERREX]
    W -->|No| Y[Check Sense: CHSEN]
    Y --> Z{Function 42?}
    Z -->|Yes| AA[TER42 Post-Process]
    Z -->|No| V
    AA --> V
    X --> V
    C --> V
    F --> V
    N --> V
    V --> AB[RETOP]
    AB --> AC{Queue Empty?}
    AC -->|No| J
    AC -->|Yes| I

Write Recovery Mechanism (CCRWO)

flowchart TD
    A[Write Operation in Progress] --> B[Hardware Failure<br/>PFAIL/SBRST/LIRST]
    B --> C{Is WRITE Operation?<br/>Function 1 or 61?}
    C -->|No| D[Simple Retry]
    C -->|Yes| E[Set 4SRWO Flag]
    E --> F[Decrement Retry Counter]
    F --> G[Jump to RETRY]
    G --> H{4SRWO Set?}
    H -->|Yes| I[Enter CCRWO]
    I --> J[Phase 1: Build COMPARE Command]
    J --> K[Set Original Disk Address]
    K --> L[Set Original Block Count]
    L --> M[Set Original Memory Address]
    M --> N[Execute COMPARE<br/>via Function 74]
    N --> O{COMPARE Result?}
    O -->|All Match<br/>Sense = 0| P[Success Path]
    O -->|BLANK CHECK<br/>Sense = 8| Q[Phase 2: Get Failing Address]
    Q --> R[Calculate Remaining Blocks:<br/>Total - Written]
    R --> S[Adjust Memory Address:<br/>Original + Offset]
    S --> T[Build WRITE Command]
    T --> U[Execute WRITE<br/>Remaining Blocks Only]
    U --> V{Write Success?}
    V -->|Yes| P
    V -->|No| W{Retries Remain?}
    W -->|Yes| F
    W -->|No| X[Error Return<br/>Partial Write Success]
    P --> Y[Check Sense: CHSEN]
    Y --> Z[RETEX Success]
    Z --> AA[Clear 4SRWO on Next Op]
    AA --> AB[Operation Complete]

MODE SELECT Sequence

sequenceDiagram
    participant App as Application
    participant SCOPT as SCOPTICAL
    participant MDSEL as MDSEL Routine
    participant SCSID as SCSI Driver
    participant Device as Optical Device

    App->>SCOPT: First Operation (READ/WRITE)
    SCOPT->>SCOPT: Check 4SINI Flag
    Note over SCOPT: Not set - initialization needed
    SCOPT->>SCSID: Execute INQUIRY (Function 36)
    SCSID->>Device: INQUIRY Command
    Device-->>SCSID: Device Type = 3 (Optical)
    SCSID-->>SCOPT: Success
    SCOPT->>SCOPT: Validate Type 3 or 4
    SCOPT->>SCOPT: Set 4SINI Flag
    SCOPT->>SCOPT: Check 4SMSL Flag
    Note over SCOPT: Not set - MODE SELECT needed
    SCOPT->>MDSEL: Call MDSEL
    MDSEL->>MDSEL: Check Device Type
    Note over MDSEL: Type 3 = Optical parameters
    MDSEL->>MDSEL: Build MODE SELECT Command
    Note over MDSEL: 4 bytes: header + parameters
    MDSEL->>SCSID: Execute MODE SELECT
    SCSID->>Device: MODE SELECT Command
    Note over Device: Configure:<br/>- Block size<br/>- Write verification<br/>- Error recovery<br/>- Cache control
    Device-->>SCSID: Success
    SCSID-->>MDSEL: Complete
    MDSEL-->>SCOPT: Success
    SCOPT->>SCOPT: Set 4SMSL Flag
    Note over SCOPT: Device now initialized
    SCOPT->>SCSID: Execute Original Operation
    SCSID->>Device: READ/WRITE Command
    Device-->>SCSID: Data Transfer
    SCSID-->>SCOPT: Success
    SCOPT-->>App: Operation Complete

Initialization Flow

stateDiagram-v2
    [*] --> Uninitialized: Power On / Device Reset

    Uninitialized --> CheckingDevice: First Operation<br/>(Not 3SNTR)

    CheckingDevice --> InquiryExecute: 4SINI not set
    InquiryExecute --> ValidateType: INQUIRY Success
    ValidateType --> TypeError: Type not 3 or 4
    TypeError --> [*]: Error Return

    ValidateType --> InitComplete: Type Valid
    InitComplete --> CheckModeSelect: Set 4SINI

    CheckModeSelect --> ModeSelectExecute: 4SMSL not set
    ModeSelectExecute --> DeviceReady: MODE SELECT Success
    DeviceReady --> Ready: Set 4SMSL

    CheckModeSelect --> Ready: 4SMSL already set

    Ready --> CheckWriteRecovery: Device Initialized

    CheckWriteRecovery --> WriteRecovery: 4SRWO set
    WriteRecovery --> OperationExecute: Recovery Complete

    CheckWriteRecovery --> OperationExecute: 4SRWO not set

    OperationExecute --> Success: Operation OK
    OperationExecute --> HardwareError: PFAIL/SBRST/LIRST

    HardwareError --> SetRecoveryFlag: WRITE operation?
    SetRecoveryFlag --> Retry: Set 4SRWO

    HardwareError --> Retry: Other operation

    Retry --> CheckingDevice: Retry Count OK
    Retry --> [*]: Retries Exhausted

    Success --> Ready: Next Operation

    Ready --> Uninitialized: Unit Attention<br/>(Media Change/Reset)

    note right of Uninitialized
        OPSTA = 000000
        All flags clear
    end note

    note right of InitComplete
        OPSTA = 000001
        4SINI set
    end note

    note right of Ready
        OPSTA = 000003
        4SINI + 4SMSL set
    end note

    note right of SetRecoveryFlag
        OPSTA = 000013
        4SRWO set
    end note

Error Handling with BLANK CHECK

flowchart TD
    A[SCSI Command Execution] --> B{Command Type}
    B -->|WRITE| C[Execute WRITE Command]
    B -->|READ| D[Execute READ Command]
    B -->|COMPARE| E[Execute COMPARE Command]

    C --> F{Result?}
    D --> F
    E --> F

    F -->|Success<br/>Sense = 0| G[RETOP: Success]
    F -->|Recovered Error<br/>Sense = 1| G
    F -->|BLANK CHECK<br/>Sense = 8| H{Context?}

    H -->|WRITE Recovery<br/>CCRWO Active| I[Expected Result]
    I --> J[Get Failing Block Address]
    J --> K[Calculate Remaining Blocks]
    K --> L[Rewrite Failed Blocks]
    L --> M{Rewrite Success?}
    M -->|Yes| G
    M -->|No| N{Retries Remain?}
    N -->|Yes| L
    N -->|No| O[RETOP: Error 000020]

    H -->|Normal Operation| P[Unexpected BLANK CHECK]
    P --> O

    F -->|Other Errors<br/>Sense 2-7, 9-15| Q{Recoverable?}
    Q -->|Unit Attention<br/>Sense = 6| R[CHSEN: Retry]
    Q -->|Aborted Command<br/>Sense = 13| R
    R --> S{Retries Remain?}
    S -->|Yes| T[Decrement Counter]
    T --> U[RETRY: Reinitialize]
    U --> A
    S -->|No| O

    Q -->|Non-Recoverable| V[CHSEN: Error]
    V --> W[Translate Sense to Error Code]
    W --> X{Error Type}
    X -->|Serious<br/>100020| O
    X -->|Minor<br/>000020| O

    G --> Y[Update Status in SCTRG]
    O --> Z{3SERR Flag Set?}
    Z -->|Yes| AA[Log Error to System]
    Z -->|No| AB[Silent Error]
    AA --> AC[Return to Caller]
    AB --> AC
    Y --> AC

    style I fill:#90EE90
    style P fill:#FFB6C1
    style G fill:#90EE90
    style O fill:#FFB6C1

5. Optical-Specific Features

WORM Support (Write-Once-Read-Many)

Device Type 4: WORM optical disk

Characteristics: - Write operations are permanent - Cannot overwrite or erase - Sequential or random write access - Read operations unlimited

Driver Support: 1. Device type validation (types 3 and 4 accepted) 2. Write recovery preserves written blocks (cannot rewrite) 3. COMPARE verifies without reading back 4. Error handling for "media full" conditions

WORM Implications: - Failed write blocks cannot be reused - Write recovery must identify exact failed blocks - CCRWO only rewrites unwritten blocks - Bad blocks permanently lost

No Queue Sorting for Long Seeks

Problem: Optical disk seeks are very slow (100-500ms)

Solution: Mark SEEK as special operation (3SPES flag)

OPTYP(4) for SEEK: 

110004 (octal) = 3SERR + 3SPES
- 3SERR: Log errors
- 3SPES: Special operation, no sorting

Effect: - SEEK operations execute immediately - No insertion into sorted queue - Elevator algorithm bypassed for SEEK - Read/write operations may still be sorted

Comparison:

Operation Optical Magnetic Reason
READ Sorted Sorted Benefits from optimization
WRITE Sorted Sorted Benefits from optimization
SEEK Not sorted (3SPES) Sorted Too slow to benefit

Code (Lines 173-178): 

IF L NBIT 3SPES THEN
   X.TYPCO BONE SSEEK=:X.TYPCO    % Mark sortable
FI

If 3SPES is set (SEEK), SSEEK flag is NOT set, so operation bypasses DSORT.

Block Verification

Automatic Verification: During write recovery (CCRWO)

COMPARE Command: - Verifies all written blocks - No data transfer to memory - Atomic operation on device - Returns exact failing block address

Verification Process: 1. Write operation completes (or fails) 2. If hardware error: set 4SRWO 3. Next operation: CCRWO entry 4. COMPARE written blocks with source memory 5. If BLANK CHECK: identify failed blocks 6. Rewrite only failed blocks

Benefits: - Ensures data integrity - Minimizes rewrites on slow media - Preserves WORM media (no unnecessary writes) - Detects media defects early

Manual Verification: Function 3 (COMPARE)

Applications can explicitly verify data: 

1. WRITE blocks to disk
2. Check status (success)
3. COMPARE blocks (function 3)
4. Check result:
   - Sense 0: Match
   - Sense 14: Miscompare

Device Type Restrictions

Supported Types:

Type SCSI Name Support Reason
3 Optical Disk YES Primary target (rewritable optical)
4 WORM YES Write-once optical media
0 Direct Access NO Magnetic disk (use IP-P2-SCSI-DISK)
1 Sequential Access NO Tape (different driver)
2 Printer NO Not a disk
5+ Other NO Unsupported types

Validation Code (Lines 214-216): 

IF SUTYP SHZ -10><3 AND ><4 THEN
   T:=TYPER; GO FAR ERREX
FI

SUTYP Field: - Bits 10-13: Device type - Shift right 10 positions to extract - Compare with 3 and 4 - Reject if not match

Why Restrict?

  1. Optical-Specific Features: CCRWO only makes sense for optical
  2. MODE SELECT Parameters: Type 3 requires special configuration
  3. Performance Characteristics: Optimizations tuned for optical
  4. Media Properties: Write-once, slow seeks, etc.
  5. Code Clarity: Separate drivers for different media types

MODE SELECT Dependency (Lines 114-118): 

MDSEL: IF SUTYP SHZ -10=3 THEN
          1                        % Type 3: Optical parameters
       ELSE
          "0"                      % Default parameters
       FI

Type 3 devices receive different MODE SELECT data than type 4 or others.

6. Comparison Table: Optical vs Magnetic Disk

Feature Comparison

Feature Optical Disk (OPDI) Magnetic Disk (DISK) Significance
Write Recovery CCRWO (COMPARE + rewrite) Simple retry Optical: Preserves partial writes
4SRWO Flag YES (bit 3 of OPSTA) NO Optical: Write recovery state
MODE SELECT Mandatory (4SMSL flag) Optional Optical: Requires configuration
Device Types 3, 4 only 0, 1, 3, 4 Optical: Specialized
SEEK Sorting No (3SPES set) Yes (sortable) Optical: Long seeks don't benefit
BLANK CHECK Expected (000020 minor) Error (100020 serious) Optical: Used in recovery
MISCOMPARE Expected (000020 minor) Error (100020 serious) Optical: Used in verification
Function 74 Heavy use (CCRWO) Limited use Optical: Custom commands for recovery
Retry Strategy Block-level recovery Full retry Optical: More sophisticated
Cache Usually disabled Usually enabled Optical: Less benefit

Performance Characteristics

Metric Optical Disk Magnetic Disk Ratio
Seek Time 100-500 ms 5-15 ms 10-50x slower
Write Speed 150-600 KB/s 50-150 MB/s 100x slower
Read Speed 1-6 MB/s 50-150 MB/s 10-25x slower
Rotation Speed 1500-3000 RPM 5400-15000 RPM 2-10x slower
Capacity 128 MB - 9.1 GB (1990s) 40 MB - 2 GB (1990s) Similar
Access Time 150-600 ms 8-20 ms 10-30x slower

Error Handling Differences

NEWST Array Translation:

Sense Optical Error Code Magnetic Error Code Difference
0 (NO SENSE) 000000 000000 Same
1 (RECOVERED ERROR) 000000 000000 Same
2 (NOT READY) 100020 100020 Same
3 (MEDIUM ERROR) 100020 100020 Same
4 (HARDWARE ERROR) 100020 100020 Same
5 (ILLEGAL REQUEST) 100020 100020 Same
6 (UNIT ATTENTION) 100020 100020 Same
7 (DATA PROTECT) 000020 000020 Same
8 (BLANK CHECK) 000020 (minor) 100020 (serious) Optical: expected in recovery
9 (VENDOR SPECIFIC) 100020 100020 Same
10 (COPY ABORTED) 100020 100020 Same
11 (ABORTED COMMAND) 100020 100020 Same
12 (EQUAL) 000000 000000 Same
13 (VOLUME OVERFLOW) 100020 100020 Same
14 (MISCOMPARE) 000020 (minor) 100020 (serious) Optical: expected in verification
15 (RESERVED) 100020 100020 Same

Key Differences: - BLANK CHECK (8): Optical treats as minor (used in CCRWO recovery), magnetic treats as serious - MISCOMPARE (14): Optical treats as minor (verification result), magnetic treats as serious

Operation Type Control Words (OPTYP)

Significant Differences:

Function Name Optical OPTYP Magnetic OPTYP Difference
0 READ 100004 100004 Same
1 WRITE 100004 100004 Same
2 READ PARITY 100004 100004 Same
3 COMPARE 100004 100004 Same
4 SEEK 110004 000004 Optical: 3SPES set (no sort)
34 RESERVE 106004 106004 Same
35 RELEASE 000000 000000 Same
42 READ FORMAT 000000 (special) 000000 (special) Different processing
60 READ (double) 074004 074004 Same
61 WRITE (double) 134002 134002 Same
62 READ PARITY (double) 000000 074004 Optical: not implemented
63 COMPARE (double) 000000 074004 Optical: not implemented

OPTYP Flag Breakdown:

Function 4 (SEEK): - Optical: 110004 = 3SERR (log errors) + 3SPES (special, no sort) - Magnetic: 000004 = Sortable operation

Why Different? - Optical seeks are 10-50x slower than magnetic - Sorting provides minimal benefit for optical - Better to execute seek immediately than wait in queue - Magnetic seeks benefit significantly from elevator algorithm

Write Recovery Comparison

Scenario: 100-block write with power failure after 75 blocks

Optical Recovery (CCRWO): 

1. WRITE 100 blocks (LBA 1000-1099)
2. Power failure after block 75
3. Set 4SRWO flag
4. Retry: Enter CCRWO
5. COMPARE 100 blocks
6. BLANK CHECK at LBA 1075
7. Calculate: 100 - 75 = 25 blocks remain
8. WRITE 25 blocks (LBA 1075-1099)
9. Success - clear 4SRWO

Time: - Compare: ~1 second - Write 25 blocks: ~5 seconds - Total: ~6 seconds

Data: - 75 blocks preserved - 25 blocks rewritten - 75% efficiency

Magnetic Recovery (Simple Retry): 

1. WRITE 100 blocks (LBA 1000-1099)
2. Power failure
3. Retry: WRITE 100 blocks again
4. Success

Time: - Write 100 blocks: ~0.5 seconds - Total: ~0.5 seconds

Data: - All 100 blocks rewritten - 0% efficiency (but fast enough not to matter)

Conclusion: - Optical: Sophisticated recovery essential due to slow writes - Magnetic: Simple retry sufficient due to fast writes - Optical: Data preservation important for WORM media - Magnetic: Full rewrite is faster than verification

Initialization Comparison

Optical Initialization: 

First Operation:
1. Check 4SINI flag (not set)
2. Execute INQUIRY (function 36)
3. Validate device type = 3 or 4
4. Set 4SINI flag
5. Check 4SMSL flag (not set)
6. Execute MODE SELECT (mandatory)
7. Set 4SMSL flag
8. Execute operation

Required: INQUIRY + MODE SELECT

Time: ~100-200ms

Magnetic Initialization: 

First Operation:
1. Check device ready
2. Execute operation
(MODE SELECT only if needed)

Required: None (optional MODE SELECT)

Time: ~10-20ms

Why Different? - Optical requires block size configuration - Optical needs write parameter setup - Optical benefits from specific error recovery settings - Magnetic works with defaults

Function 42 (READ FORMAT) Differences

Optical TER42: - Returns layout index 40 or 41 (1KB or 2KB sectors) - Reserves ~1MB for test area - Calculation: (SURSZ × 18432 / 15) + 1 blocks reserved - Returns: Total capacity - Reserved blocks

Magnetic Function 42: - Returns standard layout indices for magnetic - No test area reservation - Simpler capacity calculation - Returns: Total capacity

Example (2KB sectors, 1GB disk):

Optical: 

Total: 524288 blocks (1GB / 2KB)
Reserved: (2048 × 18432 / 15) + 1 = 2516 blocks (~5MB)
Available: 521772 blocks
Layout: 41 (2KB)

Magnetic: 

Total: 524288 blocks
Reserved: 0
Available: 524288 blocks
Layout: Different index

Cross-Reference

  • IP-P2-SCSI-DISK.NPL: Magnetic SCSI disk driver (types 0, 1)
  • IP-P2-SCSI-TAPE.NPL: SCSI tape driver (type 1, sequential)
  • IP-P2-SCSI.NPL: Common SCSI routines (SCSID, PUNDF, etc.)

Common SCSI Features

Both optical and magnetic drivers share: - PUNDF: Unit data field validation - SCSID: SCSI command execution - DSORT: Disk queue sorting (when applicable) - TO11Q: Return to level 11 scheduler - Error logging: PHLOG, 9FLEX

Function 74 Usage

Optical: Heavy use in CCRWO - COMPARE command construction - WRITE command for failed blocks - Custom SCSI command blocks

Magnetic: Limited use - User-initiated custom commands - Special diagnostic operations

Status Flags

Optical-Specific (OPSTA): - Bit 0 (4SINI): Initialization performed - Bit 1 (4SMSL): MODE SELECT performed - Bit 3 (4SRWO): Write recovery in progress

Common: - SLINK: Queue link pointer - TACOU: Retry counter - STPRW: Access owner

Operation Control Flags (OPTYP)

Common Flags: - Bit 17 (3SERR): Error logging enabled - Bit 16 (3SNTR): Neutral operation (skip init) - Bit 15 (3DPA3): Parameter 3 double precision - Bit 14 (3DPA2): Parameter 2 double precision - Bit 13 (3SPES): Special operation (no sort) - Bit 12 (3SF42): Function 42 post-processing

Optical-Specific Usage: - 3SPES: Set for SEEK (too slow to sort) - 3SNTR: Set for device management operations - 3SF42: Not used (special function 42 handling)

Implementation Notes

Performance Considerations

  1. Seek Optimization: Don't sort seeks (too slow)
  2. Write Recovery: Essential for slow optical writes
  3. Initialization: MODE SELECT required, cache appropriately
  4. Compare vs Read: Use COMPARE for verification (faster)
  5. Block-Level Recovery: Minimizes rewrites on slow media

Reliability Features

  1. 4SRWO Flag: Ensures write recovery after hardware failures
  2. BLANK CHECK: Expected result during recovery
  3. Retry Counter: Prevents infinite loops
  4. Device Validation: Only optical types accepted
  5. Error Logging: Comprehensive diagnostics

WORM Media Handling

  1. Cannot Rewrite: Recovery only writes unwritten blocks
  2. COMPARE Essential: Verifies without destructive read
  3. Partial Success: Preserves successfully written blocks
  4. Bad Blocks: Permanently lost on WORM

Porting Considerations

When adapting this driver:

  1. Hardware Timing: Optical seeks and writes are much slower
  2. MODE SELECT: Required for optical devices
  3. Write Recovery: Critical for data integrity
  4. Device Types: Validate types 3 and 4 only
  5. BLANK CHECK: Handle as expected result in recovery
  6. Function 74: Must support user-specified commands

Summary

The SCSI Optical Disk Driver is a sophisticated, specialized driver optimized for the unique characteristics of optical storage media. Its key innovations are:

  1. CCRWO Write Recovery: Block-level verification and selective rewrite after hardware failures
  2. WORM Support: Preserves successfully written blocks on write-once media
  3. Optimized for Slow Media: No seek sorting, block-level recovery
  4. Mandatory Initialization: INQUIRY and MODE SELECT required
  5. Sophisticated Error Handling: BLANK CHECK and MISCOMPARE as expected results

The driver demonstrates careful engineering for the challenges of optical storage: slow performance, write-once media, and the need for data integrity in long-running operations.

For magnetic disk operations, use IP-P2-SCSI-DISK.NPL instead. For common SCSI functions, see IP-P2-SCSI.NPL.

Document Path: Source Code\Sintran L\NPL\IP-P2-SCSI-OPDI.md

Source Analysis: Source Code\Sintran L\NPL\IP-P2-SCSI-OPDI-Analysis.md

Source Code: Source Code\Sintran L\NPL\IP-P2-SCSI-OPDI.NPL