Memory Maps

This appendix presents memory maps for selected product platforms, processors, and interface cards. Memory map information is useful for technically qualified users who understand concepts of low-level operating systems and have a basic understanding of bus structures and address mapping in computer systems.

When using this appendix, be aware of the distinct difference between program counter values and operand addresses. The addresses that appear in this appendix are operand values and should not be confused with program counter values.

Memory Maps and Troubleshooting

Memory map information can be useful when you are determining whether a problem exists in the software or in the hardware. The system software can provide information on the reasons for a system crash. This information appears in the form of error messages issued by the read-only memory (ROM) monitor when an exception is encountered.

Failure Types

When a system crashes, the ROM monitor reports a failure type. The failure type is important both in its own right and as a guide to interpreting the other information the system provides. Failure types are usually one of the following:

• Bus errors

• Address errors

• Watchdog timeouts

• Parity errors

• Emulator traps

Bus Errors

The system encounters a bus error when the processor tries to use a device or a memory location that either does not exist or does not respond properly. Bus errors indicate either a software bug or a hardware problem. The address the processor was trying to access when the system crashed provides a key as to whether the failure is due to software or hardware.

If the operand address is valid, the problem is probably in the hardware. The address maps list addresses for selected hardware platforms.

Bus errors on an address not in the map usually indicate a software bug.

Address Errors

Address errors occur when the software tries to access data on incorrectly aligned boundaries. For example, 2- and 4-byte accesses are allowed only on even addresses. An address error usually indicates a software bug.

Watchdog Timeouts

Cisco processors have timers that guard against certain types of system hangs. The central processing unit (CPU) periodically resets a watchdog timer. If the timer is not reset, a trap will occur. Failure to service the watchdog timer indicates either a hardware or a software bug.

Parity Errors

Parity errors indicate that internal hardware error checks have failed. A parity failure is almost certainly a hardware problem. Use the address map to locate the affected hardware.

Emulator Traps

Emulator traps indicate that the processor has executed an illegal instruction. Emulator traps can be caused either by software taking illegal branches or by hardware failures, notably ROM failures.

Error Addresses

By observing the operand address, you can locate the general area of the router where the error occurred. Hardware problems can be inferred only from a bus error on a legal address, not from an emulator trap or illegal instruction trap. When looking at the bus error, the operand address--not the program counter address--provides the memory map location of the error.

show stacks Command

By using the show stacks EXEC command, you can display data saved by the ROM monitor, which includes a failure type, an operand address, and a failure program counter. This data is overwritten when the system is reloaded, so you might want to check your configuration register settings and decide how you want to recover from system crashes. Stack traces can be used by qualified technical support representatives who have access to symbol tables, object files, and source code.

Figure D-1 shows an example of the show stacks output from a software failure. The message "Software forced crash" indicates that the software detected a condition it did not expect and from which it could not recover. When investigated by a technical support representative, the listed program counter provides a trace to the code responsible for the failure.

Figure D-2 shows output from a hardware error and includes an example of a hardware operand address that can be used with the memory maps in this appendix. The operand address points to the register space for MCI unit 0 and indicates a hardware or microcode problem with that unit.

Memory Maps

The following tables summarize memory map information for the various Cisco platforms:

• Table D-1 describes the Cisco 2000 memory map.

• Table D-2 describes the Cisco 2500 memory map.

• Table D-3 describes the Cisco 3000 memory map.

• Table D-4 describes the memory map for the Cisco 3104 and Cisco 3204; Table D-5 describes the Cisco 3104 and Cisco 3204 memory map of onboard registers and chips.

• Table D-6 describes the Cisco 4000 memory map; Table D-7 describes the Cisco 4000 memory map of onboard resources.

• Table D-8 describes the Cisco 4500 memory map; Table D-9 describes the Cisco 4500 memory map of onboard resources.

• Table D-10 describes the Cisco 7000 memory map.

• Table D-11 describes the Cisco 500-CS memory map.

• Table D-12 describes Multibus memory space assignment; Table D-13 describes Multibus I/O space assignment.

• Table D-14 describes the Cisco Cx-RP memory map.

• Table D-15 describes the Cisco CSC/3 memory map.

• Table D-16 describes the Cisco CSC/4 memory map.

• Table D-17 describes the processor memory map for the Cisco CSC/2, CSC/3, CSC/4 cards, including the IGS and Cisco 3000.