Satisfying memory ordering requirements between partial reads and non-snoop accesses

• US 8,250,311 B2
• Filed: 07/07/2008
• Issued: 08/21/2012
• Est. Priority Date: 07/07/2008
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Beers, Robert H., Chou, Ching-Tsun, Safranek, Robert J., Vash, James
• Primary Examiner(s): YU, JAE UN
• Application Number:
  US12/168,613
• Publication Number:
  US 20100005246A1
• Time in Patent Office: 1,506 Days
• Field of Search: 711/141, 711/143, 711/144, 711/146, 711/145
• US Class Current: 711/146
• CPC Class Codes: G06F 12/0806 Multiuser, multiprocessor or multiprocessing cache systems
• CPC Class Codes: G06F 12/0813 with a network or matrix configuration
• CPC Class Codes: G06F 12/0824 Distributed directories, e.g. linked lists of caches
• CPC Class Codes: G06F 12/0831 using a bus scheme, e.g. with bus monitoring or watching means
• CPC Class Codes: G06F 12/0833 in combination with broadcast means (e.g. for invalidation or updating)
• CPC Class Codes: G06F 12/12 Replacement control
• CPC Class Codes: G06F 12/128 adapted to multidimensional cache systems, e.g. set-associative, multicache, multiset or multilevel
• CPC Class Codes: G06F 2212/2542 Non-uniform memory access [NUMA] architecture
• CPC Class Codes: G06F 2212/62 Details of cache specific to multiprocessor cache arrangements
• CPC Class Codes: G06F 2212/621 Coherency control relating to peripheral accessing, e.g. from DMA or I/O device
• CPC Class Codes: G06F 2212/70 Details relating to dynamic memory management
• CPC Class Codes: H04L 67/56 Provisioning of proxy services store-and-forward switching systems in data switching networks H04L12/54
• CPC Class Codes: H04L 67/568 Storing data temporarily at an intermediate stage, e.g. caching

Abstract

A method and apparatus for preserving memory ordering in a cache coherent link based interconnect in light of partial and non-coherent memory accesses is herein described. In one embodiment, partial memory accesses, such as a partial read, is implemented utilizing a Read Invalidate and/or Snoop Invalidate message. When a peer node receives a Snoop Invalidate message referencing data from a requesting node, the peer node is to invalidate a cache line associated with the data and is not to directly forward the data to the requesting node. In one embodiment, when the peer node holds the referenced cache line in a Modified coherency state, in response to receiving the Snoop Invalidate message, the peer node is to writeback the data to a home node associated with the data.

Claims

1. An apparatus comprising:

a requesting agent including,

protocol layer logic to generate a first message associated with a read of a data element, wherein the first message is to indicate to a peer agent that a cache line associated with the data element is to be invalidated in the peer agent and the peer agent is not to forward the cache line directly to the requesting agent; and

physical layer logic coupled to the protocol layer logic to transmit the first message directed for the peer agent, wherein the first message is to comprise a Snoop Invalidate No Forward message and the Snoop Invalidate No Forward message is to cause the peer node to write back the data element to a home node associated with the data element in response to the data element being held in a modified cache coherency state in the peer node, invalidate the cache line, and not provide the data element directly to the requesting node.

2. The apparatus of claim 1, wherein the protocol layer logic is also to generate a Read Invalidate No Forward message to be routed to a home agent associated with the data element.

3. The apparatus of claim 1, wherein the first message includes a Snoop Invalidate message.

4. The apparatus of claim 1, wherein the read of the data element includes a partial read of the data element from the cache line.

5. The apparatus of claim 1, wherein the protocol layer logic is coupled in an interconnect layer stack to the physical layer logic, such that the interconnect layer stack includes the physical layer logic, link layer logic, and the protocol layer logic.

6. The apparatus of claim 5, wherein the physical layer logic is to couple to a serial point-to-point link to the peer agent, and wherein the protocol layer logic is to implement a MESIF cache coherence protocol.

7. A method comprising:

generating a first message associated with a read of a data element in a requesting node;

transmitting the first message to a peer agent; and

invalidating a cache line in the peer agent associated with the data element and not forwarding the cache line directly from the peer agent to the requesting node in response to the peer agent receiving the first message and the cache line being held in a modified, exclusive, shared, or invalid coherency state, wherein the first message is to comprise a Snoop Invalidate No Forward message and the Snoop Invalidate No Forward message is to cause the peer node to write back the data element to a home node associated with the data element in response to the data element being held in a modified cache coherency state in the peer node, invalidate the cache line, and not provide the data element directly to the requesting node.

8. The method of claim 7, further comprising generating in a requesting node a Read Invalidate No Forward message to be routed to a home agent associated with the data element.

9. The method of claim 7, wherein the first message includes a Snoop Invalidate message.

10. The method of claim 7, wherein the read of the data element includes a partial read of the data element from the cache line.

11. The method of claim 7, further comprising writing the cache line back to the home agent before invalidating the cache line in response to the peer agent receiving the Snoop Invalidate No Forward Message and the peer agent holding the cache line in a modified or exclusive coherency state.

12. A system comprising:

a requesting node adapted to generate a snoop message associated with a partial read of a cache line and to transmit the snoop message to be routed to a peer node; and

a peer node coupled to the requesting node, the peer node being adapted to;

receive the snoop message,

invalidate the cache line in the peer agent, and

not forward the cache line directly from the peer agent to the requesting node responsive to receiving the snoop message and the cache line being held in a modified, exclusive, shared, or invalid coherency state, wherein the first message is to comprise a Snoop Invalidate No Forward message and the Snoop Invalidate No Forward message is to cause the peer node to write back the data element to a home node associated with the data element in response to the data element being held in a modified cache coherency state in the peer node, invalidate the cache line, and not provide the data element directly to the requesting node.

13. The system of claim 12, further comprising generating in a requesting node a Read Invalidate No Forward message to be routed to a home agent associated with the data element.

14. The system of claim 12, wherein the snoop message includes a Snoop Invalidate message.

15. The system of claim 12, further comprising writing the cache line back to the home agent before invalidating the cache line in response to the peer agent receiving the Snoop Invalidate No Forward Message and the peer agent holding the cache line in a modified or exclusive coherency state.