179 PROG: H3000 Crash Dump Proceedures Crash Dump Procedures Procedures for capturing HYDRA 3000 crash information The following procedures apply to HYDRA 3000 units populated by boards using either a Motorola 68010 or 68030 processor.  Procedures differ slightly between the two processors.   Contact JDS Systems Support for assistance whenever possible; they can arrange for Engineering support if necessary.  Please keep in mind that failures in back-leveled code can not be easily analyzed by the Engineering staff, and that the usual solution is to first upgrade to current code. 1.  Write down the error code displayed on the board.  Error codes begin with an equal sign.  They have the format:  =tt.aaaaaa, where tt is a two-digit error type and aaaaaa is a six-digit code address.  Both values are hexadecimal.  Be sure to distinguish between a 6 and a b; the 6 will have the top horizontal segment turned on, the b will not.  If this is all you have time to do before resetting, it will be of some help.  If you have time and opportunity to do more, go on to the next step. 2.  If you have PC terminal emulation software that will allow you to capture to disk everything transmitted to the PC (JDS' HNet/PC will do this), attach a PC to Control Port 0 (CTL0) in the Group 0 connectors (68010 Master Processor) or the separate 2-port panel (68030 MP) on the back of the HYDRA 3000, using a HYDRA3000-to-PC cable.  Fire up the PC and the emulation software.  Make sure the software is set for the correct communications port, using 8 data bits, 1 stop bit, no parity, and XON/XOFF flow control.  Top speed of the CTL-port is 38,400 bps.  When all is set correctly, hit Enter.  You should see a  FAT:  prompt. (Note that this stands for Fault Analysis Tool; it is not a personal comment.)  Now open a disk file to capture all subsequent output from the CTL-port.  (With HNet\PC, hold the control key down and hit F1, then type a file name in standard DOS format at the prompt in the status line and hit Enter.)  In the following steps, beware of mistyped commands.  The Fault Analysis Tool is a debugger.  As such it has great power over the unit.  It assumes you know what you are doing, so very little error checking is performed.  A mistyped command can cause a more serious crash, or even code or configuration loss.  Double-check each command before hitting Enter.  Use the Tab key to clear input to the debugger if necessary.  Use only the commands listed below or those given to you by JDS Systems Support or Engineering personnel. If a PC is not available and convenient, you can attach a modem to the control port using a HYDRA3000-to-modem cable. The modem must be configured with a maximum serial rate of 38,400 bps, XON/XOFF flow control, error correction, suppressed command echo and results, and normal Carrier Detect and Data Terminal Ready handling.  Then you (or JDS Support) can dial up to the control port from a remote PC and perform the following steps. 3.  Type MAP and hit Enter.  This will display a list of all installed, enabled boards and their current status.  Note the type and position of the broken board, and the module and offset of the failure.  You will need this information in what follows. 4.  Type ID and hit Enter.  This displays the version and date of the microcode currently running in this board (or in all boards starting with Release 2.0). 5.  As of Release 2.1, the control port is password protected regarding most debugger commands (the above are exceptions).  For good measure, to make sure you can execute all of the following commands, type PW password using the high-level password configured on your HYDRA3000 (it must have the same capabilities as the default password, HYDRASNA) 6. For code Release 2.0 or later, type SB  bn (there must be a space between B and b, but none between b and n), where b is the board type (M, C, A, or N), and n is its position number (starting with 0).  (For code releases prior to 2.0, type SB  b  n; there must be a space between B and b, and between b and n.)  Hit Enter.  This Select Board command allows you to interrogate the crashed board.  Example:  SB  A2 selects Asynchronous Interface board 2.  When you issue this command, the Fault Analysis Tool will automatically display a status report for the failed board.  You will notice that the PC (Program Counter) field of the status display will contain the same address as noted in Step 1.  Note that for 68030 boards the code addresses are 8-digit numbers, where only the last 6 digits are diplayed on the 7-segment LED; here the first 2 digits are always 20, so the full address would be 20aaaaaa.  The Module and Offset corresponding to this address are given above the display of register contents.  You may redisplay the status report at any time by typing S and hitting Enter. 7.  Type D  (SSP) and hit Enter.  This will display the contents of the System Stack. 8.  Scan the stack contents for values in the code range.  This will be 180000-19FFFF for 68010 CI, AI, or NI boards and 220000-23FFFF for the 68010 MP board, and 20000000-2001FFFF for most of the code on all 68030 boards.  Some code addresses will be split between two long words in the memory display, i.e., you may see something like xxxx0018 2345yyyy.  For each of these addresses, type FM aaaaaa (or FM aaaaaaaa), where aaaaaa (or aaaaaaaa) represents one of the addresses you found in the stack display. 9.  Look at the contents of the A0 register (you may need to issue another S command to redisplay the status report); if that value is less than 00200000, then type D  (A0) and hit Enter.  This will display the area of memory pointed to by address register A0.  Caution:  If you attempt to use the D (An) command (or the equivalent D aaaaaa command) with an address greater then 001FFFFF, you will cause a Bus Error on the Master Processor, which will stop all remaining active processes on the unit, and replace the original crash status report with that of the Bus Error.  A reset will clear the Bus Error.  It is still possible, prior to resetting, to complete Step 9, but you will have to look back in the captured display data for the actual address register contents, and use the D aaaaaa form of the Display command. 10.  Repeat the above step for address registers A1 through A6.  If the failed board is an AI board, A6 points to the beginning of a large section of memory related to the active port at the time of failure.  After typing D (A6), hit Enter 30 times to display that entire block of memory.   11.  Type GS (A6) to display the name of the port that was active at the time of failure. 12.  Type XB PORT PORT0n, where n is the 11.  Type XB  TASKS and hit Enter.  This displays a list of the active tasks for each port at the time of the failure. 12.  Type XB  MMV and hit Enter.  This displays memory allocations at the time of failure. 13.  Close the capture file (hit another Ctrl-F1 if running HNet/PC); make sure you make a written notation about the name of the capture file. Reset the unit from the control port by typing RS or by hitting the reset toggle on the Master Processor as soon convenient.  Disconnect the cable at the control port (or force the modem to hang up, if dialed up to the control port). Do not attempt to reset single boards.