Apparatus, method and system for using cache memory as fail-over memory

• US 6,467,048 B1
• Filed: 10/07/1999
• Issued: 10/15/2002
• Est. Priority Date: 10/07/1999
• Status: Expired due to Fees

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Compaq Computer Corporation (HP Inc.)
• Inventors: Carbajal, Christopher M., Jenne, John E., Olarig, Sompong P.
• Primary Examiner(s): LE, DIEU MINH T
• Application Number:
  US09/414,206
• Time in Patent Office: 1,104 Days
• Field of Search: 714/7, 714/8, 714/5, 714/6, 714/54, 714/718, 711/104, 711/108, 711/151, 711/152
• US Class Current: 714/6.32
• CPC Class Codes: G06F 12/0802 Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
• CPC Class Codes: G11C 29/70 Masking faults in memories by using spares or by reconfiguring

Abstract

A computer system having a main memory and a cache memory, the computer system uses portions of the cache memory to store information from defective main memory locations until the main memory can be repaired. The address space of the main memory is always maintained by substituting cache-lines of cache memory for the defective main memory locations. A fail-over memory status bit in the cache memory controller indicates when a cache line of the cache memory contains fail-over information from the defective or failing main memory so that that cache-line will not be written over by a cache replacement algorithm. When the fail-over status bit is set, the contents of the fail-over memory location(s) remains in the cache-line and all memory reads and writes are directed to only that cache-line of the cache memory and not the main memory for the fail-over memory location(s). Also no write-back of the fail-over memory location(s) from cache memory to the main memory is required nor desired until the main memory location is repaired or replaced. Indicator lights may be used to represent the different activities of the main and cache memories at detection of the defective memory location, during and after fail-over from the main memory location to the cache-line, and transfer back to the main memory once repaired. A plurality of cache memories may migrate fail-over information therebetween.

Claims

1. A computer system providing automatic fail-over of information stored in memory locations having an undesired number of errors to cache-lines of a cache memory during operation of the computer system, the computer system comprising:

a central processing unit;

a memory controller connected to the central processing unit;

a main memory comprising a plurality of memory locations having information stored therein, wherein the main memory is connected to the memory controller;

memory error detection and correction logic for determining and correcting errors in the information stored in the plurality of memory locations during access of the main memory;

a first cache memory comprising a plurality of first cache-lines and a plurality of first cache tag registers, each of the first cache tag registers being associated with a respective one of the first cache-lines, each of the first cache tag registers comprising an address portion, a status portion and a fail-over memory portion, wherein information is stored in at least one of the first cache-lines of the first cache memory during access of the main memory; and

fail-over logic connected to the memory error detection and correction logic and first cache memory, wherein the fail-over logic sets the fail-over memory portion of the first cache tag register associated with the first cache-line storing information from at least one of the plurality of memory locations that has an undesired number of errors detected by the memory error detection and correction logic, and so long as the fail-over memory portion is set, the respective first cache-line cannot be written over during subsequent accesses of the main memory.

2. The computer system of claim 1, wherein said main memory is dynamic random access memory (DRAM).

3. The computer system of claim 1, wherein said first cache memory is static random access memory (SRAM).

4. The computer system of claim 1, further comprising a plurality of central processing units.

5. The computer system of claim 1, wherein the fail-over logic resets the fail-over memory portion of the cache tag register when the at least one of the plurality of memory locations having the undesired number of errors is replaced.

6. The computer system of claim 1, wherein the main memory comprises at least one random access memory module (RAM).

7. The computer system of claim 6, wherein the random access memory module is a single in-line memory module (SIMM).

8. The computer system of claim 6, wherein the random access memory module is a dual in-line memory module (DIMM).

9. The computer system of claim 6, wherein the random access memory module is a RAMBUS dual in-line memory module (RIMM).

10. The computer system of claim 1, further comprising an indicator to indicate when the undesired number of errors is detected, the indicator being controlled by logic connected to the memory error detection and correction logic.

11. The computer system of claim 10, wherein the indicator is a light emitting diode (LED).

12. The computer system of claim 10, wherein the indicator is a message on a computer administrator display.

13. The computer system of claim 1, wherein the memory error detection and correction logic and fail-over logic are part of a core logic chipset.

14. The computer system of claim 13, wherein the first cache memory is part of the core logic chipset.

15. The computer system of claim 1, wherein the memory error detection and correction logic and fail-over logic are part of the central processing unit.

16. The computer system of claim 15, wherein the first cache memory is part of the central processing unit.

17. The computer system of claim 13, further comprising a second cache memory.

18. The computer system of claim 17, wherein the second cache memory is part of the core logic chipset.

19. The computer system of claim 17, wherein the second cache memory is part of the central processing unit.

20. The computer system of claim 17, wherein the second cache memory comprises:

a plurality of second cache-lines and a plurality of first cache tag registers, each of the second cache tag registers being associated with a respective one of the second cache-lines, each of the second cache tag registers comprising an address portion, a status portion and a fail-over memory portion, wherein information is stored in at least one of the second cache-lines of the first cache memory during access of the first cache memory; and

the fail-over logic connected to the memory error detection and correction logic and second cache memory, wherein the fail-over logic sets the fail-over memory portion of the second cache tag register associated with the second cache-line storing information from at least one of the first cache-lines that has an undesired number of errors detected by the memory error detection and correction logic, and so long as the fail-over memory portion is set, the respective second cache-line cannot be written over during subsequent accesses of the first cache memory.

21. The computer system of claim 1, further comprising a third cache memory associated with the central processing unit.

22. The computer system of claim 1, further comprising a hard disk drive, a floppy disk drive, a CD ROM, a modem, a keyboard, a video controller and monitor, a mouse, a real time clock, a serial port, a parallel port, and a network interface card.

32. The computer system of claim 1, further comprising a plurality of cache memories, each of said plurality of cache memories comprising a plurality of cache-lines and a plurality of cache tag registers, each of the plurality of cache tag registers being associated with a respective one of the plurality of cache-lines, each of the plurality of cache tag registers comprising an address portion, a status portion and a fail-over memory portion, wherein information is stored in at least one of the plurality of cache-lines of a one of the plurality of cache memories during access of the main memory, wherein said fail-over logic sets the fail-over memory portion of the cache tag register associated with the at least one of the plurality of cache lines of the one of the plurality of cache memories.

33. The computer system of claim 32, further comprising logic for migrating information stored in at least one cache-line of the one cache memory to a cache line of another one of the plurality of cache memories and setting the fail-over memory portion of the cache tag register associated with the another one of the plurality of cache memories.

34. The computer system of claim 33, wherein the fail-over memory portion of the cache tag register of the at least one cache-line is reset when the fail-over memory portion of the cache tag register of the another one of the plurality of cache memories is set.

23. A core logic chipset, comprising:

a host bus interface adapted for connection to a central processing unit;

a memory controller adapted for connection to a main memory comprising a plurality of memory locations having information stored therein;

memory error detection and correction logic for determining and correcting errors in the information stored in the plurality of memory locations during access of the main memory;

a cache memory comprising a plurality of cache-lines and a plurality of cache tag registers, each of the cache tag registers being associated with a respective one of the cache-lines, each of the cache tag registers comprising an address portion, a status portion and a fail-over memory portion, wherein information is stored in at least one of the cache-lines of the cache memory during access of the main memory; and

fail-over logic connected to the memory error detection and correction logic and cache memory, wherein the fail-over logic sets the fail-over memory portion of the cache tag register associated with the cache-line storing information from at least one of the plurality of memory locations that has an undesired number of errors detected by the memory error detection and correction logic, and so long as the fail-over memory portion is set, the respective cache-line cannot be written over during subsequent accesses of the main memory.

24. A method in a computer system for providing automatic fail-over of information stored in memory locations having an undesired number of errors to cache-lines of a cache memory during operation of the computer system, the method comprising the steps of:

detecting an error in information read from a memory location in a main memory;

determining if the memory location has an undesired number of errors, wherein

if the memory location has the undesired number of errors, then

writing the information to a first cache-line of a first cache memory, and

setting a first fail-over status bit associated with the first cache-line so as to prevent the first cache-line from being written over during subsequent cache-line replacements, and

if the memory location does not have the undesired number of errors, then writing the information to the first cache-line of the first cache memory without setting the first fail-over status bit.

25. The method of claim 24, further comprising the step of activating a first indicator representing an occurrence of the information error.

26. The method of claim 24, further comprising the step of activating a second indicator representing an occurrence of the undesired number of errors.

27. The method of claim 24, further comprising the step of activating a third indicator representing a transfer of the information in the memory location having the undesired number of errors to the cache-line.

28. The method of claim 24, further comprising the step of replacing a portion of the main memory containing the memory location of the information having the undesired number of errors with a new portion of memory.

29. The method of claim 28, further comprising the step of writing back the information stored in the cache-line to the new portion of main memory and resetting the fail-over status bit associated with the cache-line.

30. The method of claim 29, further comprising the step of de-activating the second indicator.

31. The method of claim 24, further comprising the step of alerting an operator when the memory location has an undesired number of errors.

35. The method of claim 24, further comprising the step of logging the memory location of the information having the undesired number of errors.

36. The method of claim 24, further comprising the step of migrating the information stored in the first cache-line of the first cache memory having the associated first fail-over status bit set to a second cache-line of a second cache memory and setting a second fail-over status bit associated with the second cache-line.

37. The method of claim 36, wherein the first fail-over status bit is reset when the second fail-over status bit is set.

38. The method of claim 36, further comprising the step of migrating the information stored in the second cache-line of the second cache memory having the associated second fail-over status bit set to a third cache-line of a third cache memory and setting a third fail-over status bit associated with the third cache-line.

39. The method of claim 38, wherein the second fail-over status bit is reset when the third fail-over status bit is set.