VIKING ROST ADAPTERS USER GUIDE P/N 900-00036 Revision B, December 1989 Copyright (C) 1989 by TD Systems, Incorporated All Rights Reserved. Printed in the U.S.A. The material in this manual is intended for informational purposes. TD Systems assumes no responsibility for errors in this manual, and reserves the right to make changes in this manual and the described U product without prior notice. U TD Systems welcomes questions and comments from the reader concerning its products and the material in this manual. TD Systems Corporation 100 Bearfoot Road Northborough, MA 01532 Tel # 508-393-0861 Fax # 508-393-0858 U U U DEC, Q-Bus, Unibus, VAX, MicroVAX, MicroPDP, RSTS, RSX, VMS are trademarks of the U Digital Equipment Corporation. U U WARRANTY TD Systems products are warranted against defects in materials and workmanship. This warranty applies for a period of ninety (90) days from the date of shipment. This warranty shall not cover damage from misuse, or damage caused by interfacing with equipment not supplied by TD Systems. Under this warranty TD Systems will, at its option, repair or replace products which prove to be defective within the warranty period. This warranty includes labor, parts, and surface transportation costs. Equipment returned to TD Systems for repair, whether under warranty or not, must be shipped freight prepaid and be accompanied by a Material Return Authorization number issued by TD Systems. This warranty does not constitute a statement of merchantibility or fitness for a particular use. No other warranty is expressed or im- plied. TD Systems shall not be liable for consequential damages. TABLE OF CONTENTS SECTION 1: INTRODUCTION 1 SECTION 2: INSTALLATION 2 2.1 Determine CSR Address 2.2 Set CSR Address Jumpers 2.3 Q-Bus Installation 2.4 Unibus Installation 2.5 SCSI Installation SECTION 3: OPERATION 8 3.1 Power-up 3.2 Auto-configuration 3.3 Model QDA Bootstrap 3.4 Running the Onboard Utilities 3.5 Configuration Options SECTION 4: PROBLEM SOLVING 13 4.1 General 4.2 Debugging Tools 4.3 Configuration Problems 4.4 Device Errors or Erratic Operation 4.5 Viking Affects Other Devices in System APPENDIX A: VMS NOTES 17 Al VMS Sysgen Procedure A2 VMS Backup Command Switches A3 Persistent Tape Initialization Errors A4 VMS Console Error Messages AS Setting Disk Cluster Size APPENDIX B: ALTERNATE ADDRESS OPTION CHIPS 21 APPENDIX C: ERROR LOG MESSAGE FORMAT 24 TABLE 1: VIKING MODELS AND SUPPORTED SCSI DEVICES I J LOGICAL PHYSICAL MODEL INTERFACE EMULATION UNITS DEVICES COMMENTS 0-BUS: Auto-configures with SCSI Com QDO DUAL-WIDE mon Command Set (CCS) disks. 022 SLOT ONE MSCP 8 'DU' UNITS SCSI: 7 MAX Diagnostic utilities allow manual DISK PORT MAXIMUM DISKS: 8 MAX configuration, disk formatting, UNIBUS: and verification through the on UDO QUAD-WIDE board serial port (requires SPC SLOT optional cable adapter panel). 0-BUS: ONE MSCP The QDA combines the features QDA DUAL-WIDE DISK PORT& 8'DU' UNITS SCSI: 7MAX of the ODO with a bootstrap for SLOT BOOTSTRAP MAXIMUM DISKS: 8 MAX and MicroPDP systems. 0-BUS: This model works with almost all QTO DUAL-WIDE SCSI tape drives. It auto-con- 022 SLOT ONEMSCP 8'MU'UNITS SCSI: 7MAX figures, so drive types may be TAPE PORT MAXIMUM TAPES: 8 MAX mixed. Specify the 'A' version UNIBUS: for half-inch and other variable UTO QUAD-WIDE block drives, or the 'F' version for SPC SLOT fixed block cartridge drives. 0-BUS: A two-port board that combines QDT DUAL-WIDE ONEMSCP 8'DU'UNITS disk MSCP emulation on port A 022 SLOT DISK PORT MAXIMUM SCSI: 7MAX with tape MSCP emulation on AND AND DISKS: 8 MAX port B. The disk port has the UNIBUS: ONE MSCP 4'MU' UNITS TAPES:4 MAX same features as the ODOIUDO UDT QUAD-WIDE TAPE PORT MAXIMUM while the tape port has the same SPC SLOT features as the OTO/UTO. 0-BUS: A two-port board with the same QDD DUAL-WIDE 8'DU'UNITS features as the ODT/UDT, but 022 SLOT TWO MSCP MAX PORTA SCSI: 7MAX the second port is a disk port. DISK PORTS AND DISKS: 12 MAX This model is appropriate for UNIBUS: 'A'AND'B' 4'DU'UNITS applications that combine disk UDD QUAD-WIDE MAXPORTB MSCP emulation on port A with SPC SLOT direct SCSI on port B. DISK DRIVES SUPPORTED, TAPE DRIVES SUPPORTED, DISK CONTROLLERS: EMBEDDED SCSI: 'A', 'B' FIRMWARE: ADAPTEC DPT . SMS FIXED-DISK DRIVES FROM ANY DAT DRIVES FROM MOST OMTI WESTERN DIGITAL VENDOR. MOST ERASABLE MANUFACTURERS. 8MM FROM TAPE CONTROLLERS: OPTICALS. OTHER CARTRIDGE EXABYTE. MOST 9-TRACKS. EMULEX . NCR . WESTERN DRIVES INCLUDING IOMEGA CARTRIDGE DRIVES FROM AND SYQUEST. HIGH-CAPACITY STC, FUJITSU, DEC, OTHERS. DIGITAL FLOPPY DRIVES. CDROMS. F' FIRMWARE: MOST QUARTER-INCH DRIVES. TABLE 2: ViKING SPECIFICATIONS AND CONNECTOR PINOUTS SPECIFICATIONS: 0-BUS MODELS Identification: Models ODA, ODO, OTO, QDT, ODD: Assembly no.210-00031. Board Size: Standard DEC dual: 5.2 wide x 8.8 high including handles. Bus Slot Required: Standard DEC 022 Power Required: +5VDC +/- 5% at 2.8A maximum. Bus Loads: 2 AC, 1 DC. Interrupt Vector: Programmable, set by Host. Interrupt Priority: Level 4 (single-level). DMA Data Rate: 2.5 MByte/second peak, block mode. SPECIFICATIONS: UNIBUS MODELS Identification: Models UDO, UTO, UDT, UDD: Assembly no.210-00036. Board Size: Standard DEC quad: 10.4 wide x 8.8 high including handles. Bus Slot Required: Standard DEC SPC Power Required: +5VDC +/- 5% at 2.8A maximum. Bus Loads: 2 AC, 1 DC. Interrupt Vector: Programmable, set by Host. Interrupt Priority: Level 5 fixed. DMA Data Rate: 2.0 MByte/second peak, burst mode. SPECIFICATIONS: SCSI, ALL MODELS SCSI Devices: 8 maximum (Viiking is one). Drives / SCSI Controller: 2 maximum. Data Rate: Greater than 1.25 MByte/second asynchronous. Interface: Single-ended. Connector: 50-pin flat ribbon cable header, shrouded and center-keyed. Termination: 220/330 ohm permanent. Terminator Power: Fused +5v available on pin 26. Cable Length: 6 meters (18 feet) maximum. PINOUTS: VIKING scsi CONNECTOR (JI) PINOUTS: ADAPTER BOARD TO SCSI DEVICE PIN SIGNAL PIN SIGNAL PIN SIGNAL PIN SIGNAL 2 .DB(0) 1 GROUND 2 -DB(0) 1 GROUND 4 -DB(1) 3 GROUND 4 -DB(1) 3 GROUND 6 -DB(2) 5 GROUND 6 -DB(2) 5 GROUND 8 -DB(3) 7 GROUND 8 -DB(3) 7 GROUND 10 .DB(4) 9 GROUND 10 -DB(4) 9 GROUND 12 -DB(5) 11 GROUND 12 -DB(5) 11 GROUND 14 -DB(6) 13 GROUND 14 -DB(6) 13 GROUND 16 -DB(") 15 GROUND 16 -DB(7) 15 GROUND 18 -DB(P) 17 GROUND 18 -DB(P) 17 GROUND 20 CONTX(RS232) 19 GROUND 20 GROUND 19 GROUND 22 CON Rx (RS232) 21 GROUND 22 GROUND 21 GROUND 24 GROUND 23 GROUND 24 GROUND 23 GROUND 26 +5V FUSED 25 (OPEN) 26 TERMPWR 25 (OPEN) 28 FR TX (RS232) 27 GROUND 28 GROUND 27 GROUND 30 FR RX (RS232) 29 GROUND 30 GROUND 29 GROUND 32 -ATN 31 GROUND 32 -ATN 31 GROUND 34 GROUND 33 GROUND 34 GROUND 33 GROUND 36 -BSY 35 GROUND 36 -BSY 35 GROUND 38 -ACK 37 GROUND 38 -ACK 37 GROUND 40 -RST 39 GROUND 40 .RST 39 GROUND 42 -MSG 41 GROUND 42 -MSG 41 GROUND 44 -SEL 43 GROUND 44 SEL 43 GROUND 46 -CID 45 GROUND 46 C/D 45 GROUND 48 -REQ 47 GROUND 48 -REQ 47 GROUND 50 I/O 49 GROUND 50 -I/O 49 GROUND SECTION 1: INTRODUCTION The Viking series is a family of Host Adapters that allow SCSI (Small Computer System Interface) devices to be attached to DEC Q-Bus and Unibus computers. The Vikings are intelligent Host Adapters that emu- late DEC's MSCP (Mass Storage Control Protocol) disk and tape control- lers. This allows SCSI mass storage devices to be accessed via stan- dard DEC operating system drivers, eliminating the need for special software in the Host computer. Viking Host Adapters are available in disk only, tape only, and disk- tape combination models for both Q-Bus and Unibus computers. The var- ious Viking models are summarized in Table 1 below. Table 1 also lists those manufacturers whose SCSI products are supported by the Viking. Table 2 lists Viking specifications and connector pinouts. All Viking models implement a unique 'Auto-configuration' capability. Auto-configuration provides a very flexible method of connecting a va- riety of devices to the Viking. Disks and tapes may be mixed on the same SCSI bus (for models QDT and UDT). Units can be added or removed without having to change the Viking's configuration. Parity and dis- connect are handled transparently. For most applications, then, Viking installation or reconfiguration is simply a matter of 'plug and play'. This manual covers Viking installation procedures, operation, and trouble-shooting, with additional notes on VMS in Appendix A. Empha- sis is on 'standard' installations that use the Viking default auto- configuration capability. For more information on non-standard in- stallations, manual configuration, and the Viking's on-board utili- ties, refer to the Viking Technical Manual (TD document 900-00031). 900-00036 Viking User Guide Page 1 SECTION 2: INSTALLATION The Viking installation procedure consists of five steps, which are explained in detail in the following paragraphs: Step 1. Determine the correct Viking CSR base address for your system. Step 2. Set the Viking address jumpers. Step 3. Select and prepare a backplane slot for the Viking. Step 4. Install the Viking in your system. Step 5. Connect the SCSI devices to the Viking 2.1 DETERMINE CSR ADDRESS The correct CSR address for a Viking port is the one that the operat- ing system will recognize when it configures the target system. In all its operating systems DEC assigns standard addresses in system I/O space to the first port of each mass-storage device type: 772150 octal for the first disk MSCP controller and 774500 octal for the first tape MSCP controller. When the Viking is the first device of its type in the system, use the standard (default) address option. When the Viking is installed in a system that already has one or more of the same device type, then the Viking's CSR address will have to be set to a value in 'floating' I/O space. If the operating system is VMS, use the SYSGEN procedure described in Appendix A. For PDP-ll systems, refer to the PDP-ll Architecture Handbook (DEC EB-23657-18) or use the FLOAT utility run from the XXDP+ Diagnostic Supervisor to determine the correct address. Be aware, however, that some PDP-ll Operating systems have no built-in provisions for additional devices, and will have to be re-SYSGENed to accommodate the Viking. For details, refer to the system generation manual for the particular 0/S used on your system. 2.2 SET CSR ADDRESS JUMPERS The Viking's MSCP port CSR addresses are set by a group of four option jumper plugs in the arrangement shown below. Refer to either Figure 1 (Q-Bus) or Figure 2 (Unibus) for the location of the option jumpers on your Viking. The range of available port addresses and the exact function of each option jumper depend on which Viking model you have. The models and their jumper options are listed below. 900-00036 Viking User Guide Page 2 2.2.3 Models QDT and UDT Viking models QDT and UDT appear to the Host computer as TWO mass- storage ports, one a disk port, and the other a tape port. Both ports implement DEC's MSCP protocol. The disk port emulates DEC's 'DU' devices, while the tape port emulates one or more TK5O's with Tape MSCP. The following port address options are available: PLUGS: 1 2 DISK PORT ADDRESS IN IN 772150 (Standard) IN OUT 760334 (Floating device) OUT IN 760354 (Floating device) OUT OUT 760374 (Floating device) PLUGS: 3 4 TAPE PORT ADDRESS IN IN 774500 (Standard) IN OUT 760404 (Floating device) OUT IN 760444 (Floating device) OUT OUT 760504 (Floating device) 2.2.4 Models QDD and UDD Viking models QDD and UDD appear to the Host computer as TWO mass- storage ports, both implementing DEC's MSCP protocol. Port 'A' works with SCSI magnetic disks, while Port 'B' works with SCSI optical disks. Both types of disks appear to the Host system as 'DU' devices. The following port address options are available: PLUGS: 1 2 DISK PORT 'A' ADDRESS IN IN 772150 (Standard) IN OUT 760334 (Floating device) OUT IN 760354 (Floating device) OUT OUT 760374 (Floating device) PLUGS: 3 4 DISK PORT 'B' ADDRESS IN IN 772154 (Standard) IN OUT 763000 (Floating device) OUT IN 763100 (Floating device) OUT OUT 763200 (Floating device) 2.2.5 Model QDA Viking model QDA appears to the Host computer as one mass-storage port, implementing DEC's Disk MSCP protocol. The port works with SCSI magnetic disks, which appear to the Host system as 'DU' devices. Port B on the QDA, when enabled, provides a Bootstrap routine for MicroPDP Q-Bus systems ONLY. The Bootstrap address may be set to either 773000 (DEC standard) or 771000 (alternate). NOTE: If the on-board bootstrap is to be enabled, make sure that all other bootstraps are disabled or do not conflict with the selected bootstrap address. 900-00036 Viking User Guide Page 4 1 <----| |~DiskPortAddressSelect 2 <-| 3 < Bootstrap Addesss: IN = 773000 OUT = 771000 4 < IN = Bootstrap Enable IPLUGS 1 2 3 4 DISK PORT ADDRESS IN IN - IN 772150 (Standard) IN OUT - IN 763000 (Floating device) CUT IN - IN 763100 (Floating device) IOUT OUT - IN 763200 (Floating device) IN IN - OUT 772150 (Standard) IN OUT - OUT 760334 (Floating device) OUT IN - OUT 760354 (Floating device) OUT OUT - OUT 760374 (Floating device) 2.3 Q-BUS INSTALLATION Viking Q-Bus models require a single dual-wide Q22 slot in the Q-Bus backplane. Because the Viking system is fully buffered, DMA priority is not critical. The Viking supports only single-level interrupt priority, so it should be placed AFTER any multi-level interrupting devices on the Q-Bus. 2.3.1 Installation in a BA23/123 Cabinet The Viking may be installed in any Q22 slot. Figure 3 shows a typical BA23/123 backplane. Note that the C-D connectors in the first 3 slots are not Q22 compatible. Also, the DMA Grant daisy chain must be con- tinuous from the processor in slot 1 to the last board in the back- plane; there can be no empty Q22 slots between between boards. NOTE: The old DEC RQDXl (RX5O/RD5O/RD5l) controller does NOT pass DMA grants. The Viking must be installed AHEAD of the RQDXl in the backplane. 2.3.2 Installation in a MiCroVAX 3xxx The MicroVAX 3000 series have a new cabinet (the BA2OO series) and backplane layout. The backplane has 13 quad-width slots, but only the A-B connectors have Q22 signals. The C-D connectors are unwired. The Viking can only be installed in the A-B connectors. DMA Grant routing is straightforward, as shown in Figure 6. Due to the new design, however, additional hardware is required to maintain the cabinet's EMI shielding and airflow characteristics. DEC sells the hardware pieces either individually or in a kit. The kit is DEC model number H3651. It includes a handle with dual I/O cutouts and the necessary filler panels. 900-00036 Viking User Guide Page 5 2.4 UNIBUB INSTALLATION 2e4e1 Backplane Requirements The Unibus Viking models require a single quad-wide SPC slot in a Small Peripheral Controller (SPC) backplane. The backplane must be wired to 'SPC PRIME' specifications: that is, it must accommodate DMA peripherals. All new backplanes, VAX Unibus adapters, and most other Unibus backplanes currently in service meet this requirement. If in doubt, verify that other DMA peripherals are installed, or contact your DEC Field Engineer. The Viking is fully buffered, so its DMA (NPR) priority is not criti- cal, and it may be placed anywhere in the backplane. Its BR level is fixed at Level 5, so its interrupt priority may only be altered by repositioning it in the backplane. Figure 5 shows a typical Unibus backplane with the Viking installed. •4e2 Slot Preparation cS2elect an unused slot in the backplane and remove the grant continuity ard from it. If the grant card is a full-height card mounted in con- nectors C and D, no further preparation is necessary. If the grant card is the small G727A model installed in connector D, the NPG jumper also must be removed from between pins CAl and CBl of the selected slot. This jumper is found on the wiring side of the backplane. Figure 6 shows the NPG wiring as it would appear from the top of the backplane if you had x-ray vision. 2.4.3 Installation The Viking is inserted into connectors C-D-E-F of the prepared slot, with its component side facing slot 1 as shown in Figure 5. NOTE: Under NO circumstances should the Viking be inserted into connector A or B, as there are voltages present on these connectors that will damage the Viking. 2.5 SCSI INSTALLATION 2e5e1 General Rules The following is a brief summary of the rules for connecting SCSI devices to the Viking. See Figure 7. Refer also to the manuals for the devices to be installed for additional information. Maximum Number of Devices. No more than eight SCSI devices may be connected together on the SCSI Bus. The Viking represents one device. Each device must have a unique SCSI ID (address) in the range 0-7. Most SCSI devices have jumpers or DIP switches to set their SCSI ID. The Viking is shipped with its SCSI ID=7. This may be changed if necessary by using the on-board utilities accessed via the Viking's serial port. Refer to the Viking Technical Manual (900-00031) for more details. 900-00036 Viking User Guide Page 6 Cable Type. SCSI devices are connected together by a 50-conductor 'daisy-chain' cable. Most SCSI devices have connectors that accept mass-terminated flat cable. This cable may be shielded or unshielded. Shielded cable is preferred when the cable length is long, or when op- erating in an electrically noisy environment. Cable Length. The maximum length for a SCSI single-ended cable is 6 meters (18 feet). In general 'the shorter the better' is a good rule to follow, as this minimizes cable delays and noise pickup. I Stub Length. SCSI requires that the 'stub' length (see Figure 7) of any connection be less than 0.1M (3.6 in). Excessive stub length will adversely affect the electrical characteristics of the SCSI bus, often causing erratic operation and random data errors. Termination. Each end of the SCSI cable must be terminated with a resistor network. This means that the devices connected to the middle I of the cable must have their terminators (if any) removed. It is recommended that the Viking be at one end of the cable because its I terminators are non-removable. Terminator Power. The terminating resistor networks require +5V. The I Viking's terminators are powered directly by the Host system's +5V supply. Fused +5V is available on pin 26 of the Viking SCSI connector for powering external terminations. The fuse may be removed for ap- plications that don't require the +5V. I 2.5.2 Cabling with an Optional Adapter Panel optional adapter panel (Part no. 210-00032) is available. This panel accomplishes two things: 1) allows for easier cabling to ex- I ternal mass storage subsystems, and 2) provides a serial RS232 con- nector for access to the Viking's diagnostic and utility programs. The panel may be permanently mounted in one of the small I/O cutouts I of a DEC BA23 or BA123 cabinet. To connect the Viking to a controller with the adapter panel, simply run one cable from the Viking 50-pin connector to the 50-pin connector I on the rear of the adapter panel. This is the connector on the side OPPOSITE the R5232 connector. The controller may now be cabled to the 50-pin connector on the front of the adapter. Refer to Figure 8. Fused terminator power with a series diode is provided on pin 26 of the controller connector. If terminator power is not desired, the diode may be removed. It is located on the rear of the adapter panel. The serial R5232 port will work with any terminal that is set up to support asynchronous serial communications with the following charac- teristics: 9600 baud, 8 data bits, one stop bit, no parity. The R5232 pinouts are given in Figure 8. Generally, a 'null modem' cable I is required to connect to the terminal. This is simply a cable that that connects the transmit data signal on one end to the receive data signal on the other, and vice versa. Check your terminal's manual for I specifics. I 900-00036 Viking User Guide Page 7 SECTION 3: OPERATION 3.1 POWER-UP On power-up the Viking executes a series of self-tests and displays the results via its LED. The LED is beside the 50 pin SCSI connector. 3.1.1 Self-test Sequence ROM test: A ROM checksum test is performed to verify ROM contents and processor functions. RAM test: The on-board RAM is thoroughly tested. Buffer and I/O: The high-speed buffer RAM is tested, the SCSI interface chip is initialized and tested. 3.1.2 Normal LED Power-on Sequence ON 1 second: ROM test. OFF 1/2 second: RAM test. ON 1/4 second: Buffer and I/O test. ON DIM: Tests passed, waiting for Host initialization. 3.1.3 LED Failure Indications Always OFF: Board dead. 1 FLASH/sec: ROM failure. 2 FLASH/sec: RAM failure. 4 FLASH/sec: Buffer or SCSI failure. 3.1.4 Normal LED Indications after Self-Test ON DIM: Offline, waiting for Host initialization. OFF: Online, inactive. ON BRIGHT: SCSI activity. 3.2 AUTO-CONFIGURATION In its default auto-configure mode, each Viking port assigns MSCP unit numbers to SCSI devices based on the following auto-configure method, starting with unit 0 or the user-selected beginning unit number (see Paragraph 3.5) on each port: 1) Find the first SCSI controller that responds to device selection. Start with SCSI address 0 and work up to 7. Skip the Viking's host adapter address (normally=7). 2) Determine if the first drive on a responding controller is an appropriate device type: that is it exists, is a disk (or tape) and is a full SCSI implementation. 3) If the device is configurable, assign it the next unit number. Test the device to determine if it supports parity and disconnect. Enable/disable accordingly. 4) Advance the unit number, repeat steps 2 and 3 for the next device on the controller. If the next device doesn't exist, advance to the next SCSI address and repeat step 1. 900-00036 Viking User Guide Page 8 There are several constraints on this auto-configure method: 1) The SCSI devices must support the following commands: Request sense (extended) Inquiry Read capacity (disks) Read block limits (tape) Mode sense/select (tape) 2) A maximum of two drives per controller is allowed. 3) A disk port can support a maximum of eight (8) units. 4) A tape port can support a maximum of eight (8) units unless the Viking is a disk/tape model, in which case the maximum is four (4) units. 5) No disk partitioning is supported. 3.3 MODEL QDA BOOTSTRAP An optional onboard bootstrap is implemented on the Viking QDA. It-may be enabled by installing option plug 4 (described in Paragraph 2.2.5 above). Once enabled, one of two boot addresses may be selected with option plug 3. The following limitations apply: 1) If the host system has an LSI-11, 11/23, or 11/73 processor, either Viking boot address may be selected. Make sure that the selected address does not conflict with existing bootstraps on other boards. 2) If the Host system has a processor with an internal bootstrap (an 11/23+, 11/53, or 11/83), the Viking boot address must be set to 771000. 3) The Viking disk port address MUST be set to DEC's standard 'DU' address (772150). Other bus addresses will not boot. 4) Bootstrap routines are provided for 'DU' and 'DL' devices only. The DL controller must be set to its standard address (774400). 3.3.1 Running the Onboard Bootstrap The onboard bootstrap may be started automatically on power-up (reter to the DEC processor manual for power-up options), with a front panel 'BOOT' switch, or manually by halting the processor and restarting it at the selected bootstrap address. Once started, the bootstrap pro- gram displays a countdown message and then attempts to boot from DUO. 3 2..... 1 Booting from DUO If any key is struck during the countdown, the boot prompt will be displayed. The user may now enter the desired DU or DL unit to boot from by typing 'DUn' or 'DLn'. 3.3.2 Running DMON from the Bootstrap The Diagnostic Monitor (DMON) may also be started from bootstrap. Simply type 'V' in response to the bootstrap prompt, then another to get the DMON prompt. 900-00036 Viking User Guide Page 9 3.4 RUNNING THE ONBOARD UTILITIES The Diagnostic Monitor (DMON) is a set of programs that are resident in the Viking and provide the user with the tools necessary to con- figure the Viking, set up SCSI devices that are attached to it, and format/verify disk drives. DMON may be run either off line with any R5232 serial terminal, or with the QDA on MicroPDP/LSI-11 systems, directly through the system console. 3g4e1 Offline with an RB232 Terminal Connect the terminal to the R5232 connector of the breakout panel. Refer to Paragraph 2.5.2 and Figure 8 for connector pinouts and ca- bling information. Power the system up and verify that the Viking has passed its self-tests. Now press any key on the terminal. The screen will display the DMON sign-on message with Viking model and serial number followed by the DMON prompt. 3.4.2 Using the System Console (Viking QDA Only) The Diagnostic Monitor (DMON) may also be started from bootstrap. Simply type 'V' in response to the bootstrap prompt, then another to get the DMON prompt. 3.4.3 DMON Commands Typing H in response to the DMON prompt will display a brief menu of commands: Command list: A Abort MSCP C Display configuration D Display SCSI devices F Format unit M Mode sense/select R Read block S SCSI command T Test menu V Verify unit W Write block X Cold start exit All DMON commands are initiated by typing a single letter followed by a Carriage Return . Letters may be either upper or lower case. When typing a command or any other input to DMON, the following spe- cial characters are recognized: Control C Aborts the command and returns to the DMON prompt. Delete Deletes the last character typed. When DMON requests a numerical value during a command, it may need ei- ther a decimal (D) value or a hexadecimal (X) value and will so indi- cate: Enter cylinder (D)= 12345 Enter command byte (X)= 900-00036 Viking User Guide Page 10 In the decimal case above, responding with just a will preserve the current value '12345'. In the hexadecimal case where no current value is displayed, just a will enter the value '0'. This is true for both decimal and hex values. When the command is one that references a SCSI controller and drive, DMON will first ask for the target ID (SCSI address) of the control- ler, and then for the LUN (SCSI logical unit number, usually the same as the drive number) of the dri:ve. For devices with embedded SCSI controllers, this value will be 0. Most of the commands are self-explanatory. Note that the Read/Write Block, Verify, and Format commands apply only to disk drives. Refer to the Viking Technical Manual (900-00031) for detailed information on commands and Viking configuration options. 3.5 CONFIGURATION OPTIONS Two configuration options are available for all Viking models. The options are accessed via the DMON utility using the 'C' command. 3.5.1 First Unit Offset This option allows the user to specify the unit number of the first unit on the selected port. The offset may be set to any value from 0 to 7. A different offset may be specified for each port of a two-port board (QDT or UDT). Normally this feature is used in the Autocon- figure mode to accomodate installations in which the operating system requires sequential (or unique) unit numbers over several controllers. When used with Autoconf igure, any additional units would be assigned unit numbers in ascending sequential order. Example: if the first unit is selected to be Unit 6, the second unit would be unit 7, etc. Once set, the starting unit number is saved in non-volatile memory and retrieved on subsequent power-ups. Example: The following is a description of how you would enter a unit offset of 3, but leave the port in Autoconfigure mode. The first drive on the SCSI bus will appear as Unit 3 to the Host system, and any additional drives will be numbered in sequence (4, 5, etc.). Start by getting the onboard diagnostics running. Then do a 'C' com- mand. After the current configuration is displayed, enter the follow- ing answers in response to the questions shown: Change host adapter parameters? (Y/N) N Change port configuration? (YIN) Y Enter unit number (0-7) of first unit= 3 Autoconf igure? (YIN) x Save new Configuration? (YIN) Y The new configuration will take affect after the next power- up, or you can type 'X' to restart the Viking. 900-00036 Viking User Guide Page 11 3.5e2 Manual Configuration of Ports Both tape and disk ports may also be manually configured. This option disables Autocont igure mode and lets the user force the Viking to con- figure units at specific SCSI addresses and LUN's. Furthermore, a manually configured port will always report the units as existing, even if they are not physically attached to the SCSI bus. Up to eight units may be manually configured on single-port boards (Viking models QTO, UTO, QDO, QDA- and UDO). On disk/tape models (QDT and UDT) the limit is four units on either port. Once entered, the manual configuration is saved in non-volatile memory and will always appear to the Host system, even if the target drives are not attached. Manual configuration may be combined with a first unit offset to con- figure unit numbers from 0-10. Example: The following is a description of how you would enter a fixed tape port configuration if you have one tape drive set to SCSI ID 0 and you wish to have it appear as unit 10. Start by getting the onboard diagnostics running. Then do a 'C' command. After the cur- rent configuration is displayed, enter the following answers in response to the questions shown: Change host adapter parameters? (Y/N) N Change port Configuration? (Y/N) x Enter unit number (0-7) of first unit= 7 AutoConf igure? (Y/N) N MU7 exist? (YIN) N MU8 exist? (YIN) N MU9 exist? (Y/N) N MUlO exist? (Y/N) Y Controller ID (D) = 0 (The SCSI address of the tape controller) LUN (D) = 0 (The SCSI drive number of the tape drive) Enable disconnect? (YIN) Y (Normally enabled) Enable parity? (YIN) x (Normally enabled) Enable unbuffered mode? (YIN) N (Normally buffered mode is enabled) Save new Configuration? (YIN) x 900-00036 Viking User Guide Page 12 SECTION 4: PROBLEM SOLVING 4.1 GENERAL The intent of this section is to aid you, the user, in dealing with problems that may occur with the Viking or associated equipment. It is not intended as a complete technical reference, but rather as a guide to approaching problems in a logical manner with the resources at hand. The section is written with VMS as a model, but the same procedure is applicable to any operating system. If, after following the procedures suggested in this section, you still haven't solved the problem, contact your subsystem supplier or TD Systems for technical assistance. It will be of great help if you have the following information at hand: 1. Viking model, serial number and firmware version. 2. SCSI devices installed: manufacturers and models. 3. Operating system and version. 4. List of symptoms including console error messages. 5. Applicable error log entries. 4.2 DEBUGGING TOOLS The following is a brief list of VMS commands that are useful in debugging problems with the Viking. DCL commands are shown prefaced with a '$', while SYSGEN commands start with the SYSGEN> prompt. $ SHOW DEVICE MSCP devices show up in two places in a SHOW DEVICE command. The 'port' devices PU (disk) and PT (tape) are the low-level MSCP ports accessed by the PUDRIVER. DU (disk) and MU (tape) devices are the ac- tual mass-storage units controlled by the DUDRIVER and TUDRIVER respectively. SHOW DEVICE lists each device, its online/offline state, and its current error count. Port devices normally show an er- ror count of one, corresponding to the first initialization. The online/offline status indicates whether the unit/port is available or not. It does not indicate the 'ready' state of the device. $ ANALYZE IERROR The error log utility allows the operator to obtain additional in- formation on most types of port and unit errors. Adding the /INCLUDE= modifier restricts the display to certain classes of errors. Besides the obvious /INCLUDE=(DU,MU,PU,PT) options, two other error classes are of interest: /INCLUDE=UNSOLICITED and /INCLUDE=UNKNOWN. SYSGEN> SHOW IUNIBUS 900-00036 Viking User Guide Page 13 Lists all the I/O locations that respond to the processor and displays their contents. This is particularly useful in determining whether device addresses are correctly set. 4.3 CONFIGURATION PROBLEMS This subsection outlines a debugging procedure to be used when devices attached to the Viking fail to appear in your system configuration. The procedure concentrates on new installations where there is a prob- ability that the problem is a result of incorrect setup or installa- tion. Sudden problems with existing installations are usually the result of a Viking or SCSI device failure, but the possibility of a bad cable or connector should not be overlooked, particularly if the problems cropped up after devices were moved or rearranged. Power-up the system: Viking passes self-tests? (see Paragraph 3.1) NO: Viking failed. SCSI devices pass self-tests (if applicable)? NO: SCSI device failed. Boot system, log into system manager account. Run SYSGEN Do SHOW IUNIBUS Viking CSR at selected address? NO: Recheck Viking address jumpers. If jumpers OK, Viking failed. CSR contents = Viking firmware version? NO: Viking failed. Exit SYSGEN Do SHOW DEVICE P Viking port in configuration? NO: Problem is either that the CSR address is not correct for the system configuration or the system startup file does not configure the Viking devices. Redo the SYSGEN procedure described in Appendix Al to verify the CSR address, and then examine the system startup file (usually SYS$SYSTEM:STARTUP.COM) to make sure that the AUTOCONFIGURE ALL statement is being executed. If AUTOCONFIGURE ALL includes a /SELECT= modifier, add PU (disk) or PT (tape) as appropriate to the select list. Viking port online? NO: Error count incrementing every 30 seconds? NO: Viking initialize error. Check for DMA grant continuity. On Q-Bus systems, make sure that Viking is positioned after any multi- level interrupt devices. If OK, Viking failed. YES: Viking can find no devices to configure or is detecting a fatal SCSI bus error. Check SCSI cabling, SCSI device ID's, power. Do SHOW DEVICE DU (disk) or MU (tape) Expected unit(s) appear in configuration? NO: Possible SCSI device failure or incompatible 900-00036 Viking User Guide Page 14 device. Check SCSI cabling, SCSI device ID's, power. 4e4 DEVICE ERRORS OR ERRATIC OPERATION Excessive device errors or erratic operation may be caused by a marginal component (Viking or SCSI device) , or by 'environmental' problems such as heat, electrical noise, excessive cable length, im- proper termination, or poor power supplies. It is best to eliminate any environmental problems before proceeding further: Step 1. Check the power to your SCSI devices to make sure that it is within specification. Measure it while the device is actually operating, if possible. Step 2. Verify that your SCSI cable is properly terminated at each end, and that no other terminators are installed if there are more than two SCSI devices (including the Viking). Check the Viking's terminators to make sure that they are not cracked. Check socketed terminators to make sure that the resistor packs are securely inserted with no bent leads. Make sure that all terminators have terminator power applied. Step 3. Check that your drives are operating within their temperature rating. Remember that magnetic media are particularly susceptible to temperature extremes: most drives are rated for operation at 50ºC (122ºF) maximum. Step 4. Make sure that your SCSI cable is as short as possible, and dressed away from any electrically noisy components. This is particularly important with the older VAXs such as 11/780's and 785's which are fierce generators of EMI. Shielded cable is strongly recommended in all cases where an external SCSI connection must be made, Having eliminated environmental problems as a cause, the easiest way to isolate faulty components is to swap them with known good ones. If this is impossible, gather as much information as you can (see the list in subsection 4.1) and contact your subsystem supplier or TD Sys- tems for technical assistance. 4.5 VIKING AFFECTS OTHER DEVICES IN SYSTEM ere we have two basic possibilities: one is that the Viking's CSR Haddress or vector is conflicting with the other device's. The second is that the Viking's DMA rate results in 'Data Late' type errors from other devices. 4.5.1 Address or Vector Conflict Redo the SYSGEN procedure described in Appendix Al again. Do it first without including the Viking port(s), then again with the Viking port(s). Make careful notes of the changes in addresses AND vectors equired to add the Viking. Often other devices will retain their required CSR addresses, but require that their vectors be changed. 900-00036 Viking User Guide Page 15 For example, when adding a Viking QTO as the second 'TU8l' in a Q-Bus system with DHV-ll's, the DHV-1l CSR addresses will remain the same, bfut their vectors will change, requiring that the boards be removed rom the system and their DIP switches reset. If this is not done, the system will configure the DHV's (because their CSR's are correct), but users will quickly complain that their terminals don't respond to keystrokes, because the DHV's vectors are assigned to the Viking. 4.5.2 Data Late Errors froin Other Devices Some I/O devices are quite sensitive to DMA latency (the amount of time from the point at which the device requests the bus to the point when it gains control of the bus), and will report errors if the latency is too long. For example, Unibus Ethernet controllers (DUENA's) are sensitive to Unibus NPR latency, and the Unibus Adapter tends to be somewhat slow in this area. Add to this a Viking doing a lot of data transfers, and occasionally the DEUNA may log errors. The same holds true of some older disk and tape controllers that have small data buffers. The first step in solving this problem is to make sure that the Viking is installed AFTER these devices in the backplane, that is in a higher numbered slot. If this doesn't work, then the Viking's DMA burst count will have to be reduced. Refer to the Viking Technical Manual (900-00031) for details. 900-00036 Viking User Guide Page 16 APPENDIX A: VMS NOTES Al. VMS SYSGEN PROCEDURE Each time a VMS system is booted, the VMS 'SYSGEN' utility is invoked to determine what physical I/O devices are present and to bring them online and attach their drivers. - If the Viking is optioned with the correct port address(es), VMS will automatically detect and configure it. To determine the proper addresses for a Viking board in a VMS system, the user should use the SYSGEN 'CONFIGURE' utility: 1. On a running VAX/VMS system, log onto the system manager queue and run the SYSGEN utility: $ MCR SYSGEN SYSGEN> 2. Enter the CONFIGURE facility of SYSGEN as follows: SYSGEN> CONFIGURE DEVICE> 3. At the DEVICE prompt, enter the list of Q-Bus or Unibus options present in the VAX system. Make sure that each device count is cor- rect and that MSCP disk and tape device counts include both existing devices and the new Viking MSCP ports. Refer to the table below for a list of common options and their DEVICE names. For example, suppose you are installing a Viking QDT in a MicroVAX system with a TK5O, an RD54, a DZV1l, and a DEQNA Ethernet controller. The Viking QDT represents one 'UDA' disk port and one 'TU8l' tape port, no matter how many drives are attached to it. Adding these to the existing devices, enter: DEVICE> QNA (one DEQNA) DEVICE> DZll (one DZV1l) DEVICE> UDA 2 (two MSCP disk ports including Viking) DEVICE> TU81 2 (two MSCP tape ports including Viking) 3. When all devices have been entered, press CTRL Z to end the input session and display the results. Each device will be listed along with its port name, CSR address, interrupt vector, and driver support status. The new Viking ports will be the last device of each type listed. For the Viking QDT in the above example, SYSGEN will display: Device: UDA Name: PUA CSR:772150 etc. Device: TU8l Name: PTA CSR:774500 etc. Device: QNA Name: XQA CSR:774440 etc. Device: DZll Name: TTA CSR:760100 etc. Device: UDA Name: PUB CSR:760354 etc. Device: TU8l Name: PTB CSR:760444 etc. 900-00036 Viking User Guide Page 17 4. The Viking may now be optioned for the correct address and in- stalled in the system. Again using the example of a Viking QDT, the correct addresses are shown below. Note that the Viking's interrupt vectors are programmable and not selected by option jumpers. Viking disk port: Address= 760354 Viking tape port: Address= 760444 SYSGEN DEVICE TABLE: Physical VMS Sysgen Device Device 'DEVICE' Name Name Name Disk: Viking DU 'UDA' RQDXn DU 'UDA' UDASO DU 'UDA' KDA5O DU 'UDA' RRD5O DU 'UDA' RLl1,RLV1l/12 DL 'RLl1' RXll,RXV1l DX,DY 'RX21l' RK06 DM 'RK611' Tape: Viking MU 'TU81' TU8l MU 'TU8l' TK5O MU 'TU81' TS11,TSV11 MS 'TS1l' TUS8 DD 'TUS8' Comm: DEUNA XE 'UNA' DEQNA QE 'QNA' DZ11,DZV11 TT 'DZ1l' DH11 YH 'DH11' DHV11 TX 'DHVl1' DMF32 TX 'DMF32' LPl1 LP 'LP11' CR11 CR 'CR11' DR11W XA 'DR11W' 900-00036 Viking User Guide Page 18 A2. VMS BACKUP COMMMND SWITCHES BACKUP QUALIFIERS: /IMAGE Produces a functionally equivalent copy of the input vequilume on the output volume. Copies all the files on the input volume... no file selection qualifiers are permitted. Output volume must be mounted /FOREIGN. BACKUP/IMAGE to a disk will initialize the destination disk with the characteristics of the source disk unless the /NOINIT qualifier is used. IPHYSICAL Copies the entire input volume onto the output volume on a block-by- block basis, ignoring any file structure. Output volume must be mounted /FOREIGN. Note that this command copies the entire disk ca- pacity regardless of how much of the disk is used. /BUFFER COUNT=n Specifies the number of I/O buffers to be used in the backup opera- tion. The default is 3. A maximum of 5 may be specified. Use of more buffers than the default may improve operation with streaming tape drives. OUTPUT QUALIFIERS: /NOCRC Inhibits the calculation and storing of Cyclic Redundancy Check records on the output medium. The creation of these records requires considerable CPU overhead as well as tape overhead, and uneccessarily duplicates the error detection mechanisms of most tape controllers. /GROUP_SIZE=n BACKUP calculates and appends one error correction record to every group of 'n' data records, allowing it to correct one 'uncorrectable' read error in each group. The default is 10. A value of n=0 inhibits the entire process. IBLOCK_SIZE=n Specifies the output record size in bytes. The default for tape devices is 8192 (8 KBytes) . The allowable range is 2048 to 65024 bytes. Larger block sizes may improve the performance of streaming tape drives, but may also cause problems at End-of-Tape if the con- troller's logical EOT occurs less than the specified block size before physical EOT. EXAMPLE: $ BACKUP /IMAGE /BUFFER COUNT=5 DUAO: MUBO:name.bck - /NOCRC /GROUP_SIZE=0 /BLOCK_SIZE=32768 This example produces an image backup with no CRC records, no error correction records, and a block size of 32 KB. This set of command switches will produce the most efficient image backup in terms of time required, but at the expense of BACKUP's normal error correction capa- bilities. 900-00036 Viking User Guide Page 19 A3. PERSISTENT TAPE INITIALIZATION ERRORS 3efore the VMS INITIALIZE utility writes to a tape, it attempts to read the tape's header records to test the accessibility, expiration jate, and ownership fields. If it decides on the basis of the tests that the tape should not be initialized, it displays the error message 'MEDIUM IS OFFLINE'. Under some circumstances a blank tape may yield this error. In either case, the user may override the tests and force tnitialization by using the following command switches: $ INITIALIZE /OVERRIDE=(ACCESS, EXPIR,OWNER) MUBO: name A4. VMS CONSOLE ERROR MESSAGES the following is a list of common VMS console error messages along with the corresponding error log status (if any) and the likely cause of the error. console message: No Such Device Available error Log Status: (No error log) source and reason: Viking reports non-existent device console message: Volume Is Not Software Enabled e:rror Log Status: (No error log) source and reason: SCSI device reports Not Ready console message: Medium is Offline error Log Status: (No error log) ource and reason: See Paragraph A3 above. onsole message: Parity Error rror Log Status: Data Error, Invalid Header (disk) Long Gap Encountered (tape) ource and reason: SCSI device reports Media Error or Blank Check onsole message: (Fatal) Controller Error rror Log Status: Controller Error ource and reason: SCSI device reports Illegal Command, or Viking detects SCSI bus error. onsole message: Position Lost (tape) rror Log Status: Controller Error ource and reason: SCSI device reports Illegal Command or Viking receives unexpected response from device. This message usually occurs with Fatal Controller Error. onsole message: Invalid Buffer Length rror Log Status: Host Buffer Error ource and reason: Viking detects odd byte count (disk) or odd buffer address. 00-00036 Viking User Guide Page 20 A5. SETTING DISK CLUSTER SIZE WHAT IS A DISK 'CLUSTER'? For a Files-li structured disk the disk's 'cluster size' is the basic space-allocation unit. The cluster size is set to some number of vir- tual disk blocks (virtual blocks are always 512 bytes long for MSCP disks) when the disk is initialized. Thereafter, VMS allocates space on the disk in 'cluster-size' chunks and keeps track of allocated space using a bit map in which each bit corresponds to one cluster. The bit map is stored on the disk and cannot be larger than 255 disk blocks. If the disk is large and the cluster size is small ('1' for example) there could be too many clusters to fit in the available bit map space, and the allocation mechanism would fail. WHAT CLUSTER SIZES CAN BE USED? The range of possible cluster sizes depends on the disk capacity. For a disk with a capacity of 'N' 512-byte blocks (or the equivalent in other physical block sizes) the smallest cluster size that can be used is N/255*4096 (N divided by 1,044,480). If this is a fractional value, it must be rounded up to the next greater integer. Example: the disk capacity (N) is 1,680,234 512-blocks (860MB). The minimum cluster size is calculated to be 1.61 which must be rounded up to 2. The largest cluster size is N/100. WRAT CLUSTER SIZE IS APPROPRIATE? In selecting the cluster size you trade of f wasted space at the ends of files (if the cluster size is set large) versus increased bit map size and file fragmentation (if the cluster size is set small). The appropriate size will depend on the specific system in which the disk is installed and the uses to which the disk will be put. For disks with physical block sizes LARGER than 512 bytes, the cluster size should be set to an even multiple of the physical size in order to maximize Viking performance. Example: a disk has 1024-byte blocks. set the cluster size to 2 (2 x 512 = 1024) in order to force VMS to a1locate disk space on even 1024-byte block boundaries. U SING THE INITIALIZE COMMAND TO SET THE CLUSTER SIZE ro initialize a disk and set the cluster size, use the VMS INITIALIZE ommand with the /CLUSTER_SIZE qualifier: $ INITIALIZE /CLUSTER_SIZE=2 DUxx: volume_label (MAGE BACKUPS AND CLUSTER SIZE A cautionary note concerning disk-to-disk image backups to newly- Lnitialized disks: the BACKUP /IMAGE command will by default re- Lnitialize the destination disk, using the cluster size of the source disk. If the resulting cluster size on the destination disk is in- correct, VMS will so indicate with an error message that the CLUSTER SIZE IS INAPPROPRIATE. In order to retain the selected cluster size )n the destination disk, use BACKUP with the /NOINITIALIZE qualifier: $ BACKUP /IMAGE /NOINITIALIZE DUxx: DUyy: 000-00036 Viking User Guide Page 21 q q APPENDIX B: ALTERNATE ADDRESS OPTION CHIPS GENERAL Several alternate address option chips (PAL's) are available for Viking disk-only and tape-only models. These PAL's provide new VMS- compatible CSR base addresses to supplement the standard ones. They 'nay be used in installations where VMS SYSGEN requires a CSR address that is not supported by the standard address PAL. The alternates are I listed below with the address options implemented with each. ORDERING ALTERNATE ADDRESS CHIPS ISpecify the designation of the desired alternate address chip: 'DO-2' for disk-only Vikings; 'TO-2' or 'TO-3' for tape-only Vikings. Q-BUS INSTALLATION Remove the standard address PAL from its socket at location E4. Care- q fully insert the new PAL into the socket with pin 1 in the same orientation. Pin 1 is at the corner with the embossed dot. Do not go q by the PAL label orientation. UNIBUS INSTALLATION IRemove the standard address PAL from its socket at location E24. Carefully insert the new PAL into the socket with pin 1 in the same orientation. Pin 1 is at the corner with the embossed dot. Do not go by the PAL label orientation. DO-2 PAL FOR VIKING QDO AND UDO DISK-ONLY MODELS PLUGS 1 2 3 DISK PORT ADDRESS IN IN IN 772154 (System I/O page) IN IN OUT 760364 (Floating device) IN OUT IN 760370 (Floating device) IN OUT OUT 760374 (Floating device) OUT IN IN 772154 (System I/O page) OUT IN OUT 760454 (Floating device) OUT OUT IN 760554 (Floating device) OUT OUT OUT 760654 (Floating device) (PLUG 4 is not used and may be IN or OUT) I I900-00036 Viking User Guide Page 22 TO-2 PAL FOR VIKING QTO AND UTO TAPE-ONLY MODELS PLUGS 1 2 3 TAPE PORT ADDRESS IN IN IN 760704 (Floating device) IN IN OUT 760410 (Floating device) IN OUT IN 760510 (Floating device) IN OUT OUT 760610 (Floating device) OUT IN IN 760704 (Floating device) OUT IN OUT 760424 (Floating device) OUT OUT IN 760524 (Floating device) OUT OUT OUT 760624 (Floating device) (PLUG 4 is not used and may be IN or OUT) TO-3 PAL FOR VIKING QTO AND UTO TAPE-ONLY MODELS PLUGS 1 2 3 TAPE PORT ADDRESS IN IN IN 774500 (Standard) IN IN OUT 760404 (Floating device) IN OUT IN 760444 (Floating device) IN OUT OUT 760504 (Floating device) OUT IN IN 774500 (Standard) OUT IN OUT 760410 (Floating device) OUT OUT IN 760450 (Floating device) OUT OUT OUT 760510 (Floating device) (PLUG 4 is not used and may be IN or OUT) TO-4 PAL FOR VIKING QTO AND UTO TAPE-ONLY MODELS PLUGS 1 2 3 TAPE PORT ADDRESS IN IN IN 774500 (Standard) IN IN OUT 760404 (Floating device) IN OUT IN 760444 (Floating device) IN OUT OUT 760450 (Floating device) OUT IN IN 774500 (Standard) OUT IN OUT 760454 (Floating device) OUT OUT IN 760460 (Floating device) OUT OUT OUT 760464 (Floating device) (PLUG 4 is not used and may be IN or OUT) )00-00036 Viking User Guide Page 23 APPENDIX C: ERROR LOG MESSAGE FORMAT All error log messages sent to the Host by the Viking are formatted as 'Controller Errors'. This free-form format allows the Viking to in- clude internal SCSI information in the error message. Controller er- ror log messages under VMS result in the following display when exam- ined using an 'analyze/error' command. Other operating systems will produce similar displays. Each of the displayed hex bytes is de- scribed below: CONTROLLER DEPENDENT INFORMATION (byte) 3 2 1 0 LONGWORD 1. STH STL FLG COD LONGWORD 2. ADL TRY ERR TSK LONGWORD 3. BCM BCL ADH ADM LONGWORD 4. XCH XCM XCL BCH LONGWORD 5. 1N3 1N2 IN1 KEY LONGWORD 6. CDB CDB CDB CTL LONGWORD 7. CDB CDB CDB CDB LONGWORD 8. CDB* CDB* CDB CDB LONGWORD 9. SMB SSB LUN CDB LONGWORD 10. REVISION COD: The MSCP end code returned for the command. FLG: The MSCP end flags returned for the command: 7 6 5 4 3 2 1 0 BB REP BB UNR ERRLOG SER_EX EOT PLOST DLOST badblk badblk erriog serious end of positio data report unrep sent except tape lost lost STL,STH: MSCP status word returned for the command. 900-00036 Viking User Guide Page 24 TSK: Internal Viking task byte: 5 4 3 2 1 0 PEN DEN DIR L76 _____ parity disconn data enable enable direct ERR: Internal Viking error byte: 7 6 5 4 3 2 1 0 EDMA ECMP EFAT EPHA ESEL EABRT EPAR ECHK dma compare scsi scsi scsi task scsi scsi timeout error fatal phase select aborted parity status TRY: Internal Viking retry count. ADL,ADM,ADH: Internal Viking Host bus address: Low, Mid, High. BCL,BCM,BCK: Host requested byte count: Low, Mid, High. XCL,XCM,XCH: Number bytes actually transferred: Low, Mid, High. KEY: Sense key byte from last SCSI 'Request Sense' command. (Bit 4= 'info valid' bit from first byte of data). IN1,1N2,1N3: Information bytes from last SCSI Request Sense. CTL: SCSI target device ID. CDB: SCSI Command Descriptor Block (12 bytes). * Exabyte: Longword 8, byte 2 = Sense byte 19 * Exabyte: Longword 8, byte 3 = Sense byte 20 LUN: SCSI target LUN. SSB: SCSI end status byte. 6MB: SCSI message byte. REVISION: Firmware revision level. 900-00036 Viking User Guide Page 25