Managing memory faults

• US 8,201,024 B2
• Filed: 05/17/2010
• Issued: 06/12/2012
• Est. Priority Date: 05/17/2010
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Burger, Doug, Larus, James, Strauss, Karin, Condit, Jeremy
• Primary Examiner(s): Baderman, Scott
• Assistant Examiner(s): TRUONG, LOAN
• Application Number:
  US12/780,931
• Publication Number:
  US 20110283135A1
• Time in Patent Office: 757 Days
• Field of Search: 714/42, 714/6.1, 714/6.13, 711/206
• US Class Current: 714/42
• CPC Class Codes: G06F 11/073 in a memory management context, e.g. virtual memory or cache management memory management G06F12/00; testing of static memory units G11C29/00
• CPC Class Codes: G06F 11/0766 Error or fault reporting or storing
• CPC Class Codes: G06F 11/0793 Remedial or corrective actions recovery from an exception in an instruction pipeline G06F9/3861; by retry G06F11/1402; for recovering from a failure of a protocol instance or entity H04L69/40

Abstract

Embodiments are described for managing memory faults. An example system can include a memory controller module to manage memory cells and report memory faults. An error buffer module can store memory fault information received from the memory controller. A notification module can be in communication with the error buffer module. The notification module may generate a notification of a memory fault in a memory access operation. A system software module can provide services and manage executing programs on a processor. In addition, the system software module can receive the notifications of the memory fault for the memory access operation. A notification handler may be activated by an interrupt when the notification of the memory fault in the memory access operation is received.

Claims

1. A system for managing a hardware memory module, the system comprising:

a memory controller configured to manage memory cells of the hardware memory module, the memory controller comprising;

an error buffer module configured to store memory fault information received from the hardware memory module by the memory controller, the memory fault information reflecting a memory fault by an individual cell or memory region within a physical page of memory of the hardware memory module; and

a notification module in communication with the error buffer module configured to generate a notification of the memory fault;

a system software module configured to provide services and manage executing programs on a processor, wherein the system software module is configured to receive the notification of the memory fault and re-use other memory cells or memory regions in the physical page of memory having the memory fault;

a notification handler configured to be activated by an interrupt when the notification of the memory fault is received; and

a memory allocator configured to map a whole virtual memory page to partial physical memory pages having memory faults.

2. The system as in claim 1, wherein the error buffer module stores information including:

whether a write operation failed, data used in a write attempt, and an address of a physical location where a write failure occurred.

3. The system as in claim 1, wherein the error buffer module is configured to mark in a storage buffer to indicate that individual memory regions with faults have been permanently decommissioned by the memory controller.

4. The system as in claim 1, wherein the notification module is configured to generate fault interrupts.

5. The system as in claim 4, wherein an individual fault interrupt is a hardware fault interrupt indicating that an individual memory region has been permanently decommissioned.

6. The system as in claim 1, wherein the notification module is configured to generate a fault interrupt indicating that an amount of temporary storage used has reached a certain threshold and a plurality of memory faults, including the memory fault by the individual cell or memory region, are available to be batch processed by the system software.

7. The system as in claim 1, wherein the memory cells are a non-volatile type of memory.

8. A method for managing physical memory faults using system software, the method comprising:

receiving, by the system software, a notification that a memory cell fault has occurred for a failed physical memory location of a physical memory, the memory cell fault occurring during a write operation to a physical memory page that includes the failed physical memory location;

finding a working physical memory location in the physical memory to move failed write data of the write operation;

moving the failed write data of the write operation to the working physical memory location of the physical memory for storage;

remapping a virtual address from the failed physical memory location having the memory cell fault to an address of the working physical memory location, the remapping comprising finding a plurality of virtual pages that map to the physical memory page that includes the failed physical memory location; and

reclaiming working memory cells in the physical memory page having the failed physical memory location.

9. The method as in claim 8, wherein finding the plurality of virtual pages further comprises scanning system page tables.

10. The method as in claim 8, wherein finding the plurality of virtual pages further comprises maintaining a reverse translation table of physical to virtual page addresses.

11. The method as in claim 8, wherein reclaiming the working memory cells further comprises marking the working memory cells as being available in a memory fault map to allow the system software to find the working memory cells at a defined level of memory granularity.

12. The method as in claim 8, wherein the system software module uses a memory allocator and a memory fault map to avoid allocating the failed physical memory location.

13. A system comprising:

a memory fault map configured to;

track faulty physical memory regions or faulty physical memory cells of a physical memory; and

identify working physical memory cells in a physical memory page that includes one or more of the faulty physical memory regions or one or more of the faulty physical memory cells;

a system software module configured to reclaim the working physical memory cells from the physical memory page that includes the one or more faulty physical memory regions or the one or more faulty physical memory cells;

a fractional mapping module configured to map a whole virtual page into partial physical memory pages, the partial physical memory pages including the physical memory page that includes the one or more faulty physical memory regions or the one or more faulty physical memory cells; and

one or more processors configured to execute the system software module.

14. The system according to claim 13, the whole virtual page being partially mapped to the working physical memory cells of the physical memory page and partially mapped to other working physical memory cells of another physical memory page.

15. The system according to claim 13, further comprising:

a notification module configured to provide a hardware interrupt fault to the system software module when a memory access operation attempts to access the one or more of the faulty physical memory regions or the one or more of the faulty physical memory cells in the physical page.

16. The system according to claim 15, the memory access operation comprising a write operation.