DYNAMIC PROCESSOR CACHE TO AVOID SPECULATIVE VULNERABILITY

• US 20190332379A1
• Filed: 04/30/2018
• Published: 10/31/2019
• Est. Priority Date: 04/30/2018
• Status: Active Grant

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Inventors: Calhoun, Michael Bozich, Chitturi, Divakar
• Granted Patent: US 11,169,805 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 12/0848 Partitioned cache, e.g. separate instruction and operand caches
• CPC Class Codes: G06F 12/123 with age lists, e.g. queue, most recently used [MRU] list or least recently used [LRU] list
• CPC Class Codes: G06F 12/126 with special data handling, e.g. priority of data or instructions, handling errors or pinning
• CPC Class Codes: G06F 2212/1016 Performance improvement
• CPC Class Codes: G06F 2212/1052 Security improvement
• CPC Class Codes: G06F 2212/507 using speculative control
• CPC Class Codes: G06F 9/30047 Prefetch instructions; cache control instructions
• CPC Class Codes: G06F 9/3802 Instruction prefetching
• CPC Class Codes: G06F 9/3814 Implementation provisions of instruction buffers, e.g. prefetch buffer; banks
• CPC Class Codes: G06F 9/3842 Speculative instruction execution

Abstract

A processor including a logic unit configured to execute multiple instructions being one of a speculative instruction or an architectural instruction is provided. The processor also includes a split cache comprising multiple lines, each line including a data accessed by an instruction and copied into the split cache from a memory address, wherein a line is identified as a speculative line for the speculative instruction, and as an architectural line for the architectural instruction. The processor includes a cache manager configured to select a number of speculative lines allocated in the split cache. The cache manager prevents an architectural line from being replaced by a speculative line based on a number of speculative lines allotted in the split cache, and manages the number of speculative lines to be allocated in the split cache based on the number of speculative lines relative to a number of architectural lines.

Claims

1. A processor, comprising:

a logic unit configured to execute multiple instructions, each of the multiple instructions being one of a speculative instruction or an architectural instruction;

a split cache comprising multiple lines, each line including a data accessed by an instruction and copied into the split cache from a memory address, wherein a line is identified as a speculative line for the speculative instruction, and as an architectural line for the architectural instruction; and

a cache manager configured to;

select a number of speculative lines to be allocated in the split cache, based on a current state of the processor,

prevent an architectural line from being replaced by a speculative line based on a number of speculative lines allotted in the split cache, and

manage the number of speculative lines to be allocated in the split cache when the speculative instruction requests a data access based on the number of speculative lines relative to a number of architectural lines.

2. The processor of claim 1, wherein the cache manager is configured to modify the number of speculative lines in the split cache based on one of a least recently used state, a modified state, an exclusive state, a shared state, or an invalid state, of one or more architectural lines in the split cache.

3. The processor of claim 1, wherein the cache manager is configured to increase the number of speculative lines when an architectural line in the split cache is free or invalid, and to wait for architecture lines to be empty for allocation to a speculative line when the number of speculative lines are fully occupied,

4. The processor of claim 1, wherein the cache manager is configured to wait for a pre-selected number of accesses to the number of speculative lines before adjusting the number of speculative lines.

5. The processor of claim 1, wherein the cache manager is configured to adjust the number of speculative lines based on a value of a miss counter, a performance data, or a branch prediction.

6. The processor of claim 1, wherein the cache manager is configured to pre-select the number of speculative lines based on a pre-analysis of a processor workload.

7. The processor of claim 1, wherein the cache manager is configured to increase a number of architectural lines allocated to the split cache when a speculative line is free or invalid, the number of architectural lines are occupied, and a data access is issued from an architectural instruction.

8. The processor of claim 1, wherein the cache manager is configured to maintain the number of speculative lines within one or two of the number of architectural lines.

9. The processor of claim 1, wherein the cache manager further comprises a tunable parameter to select the number of speculative lines to be allocated in the split cache.

10. The processor of claim 1, wherein the cache manager is further configured to convert a speculative line into an architectural line by adjusting a status bit when the speculative line is accessed by an architectural instruction.

11. The processor of claim 1, wherein the cache manager is further configured to prevent an architectural line in the split cache from being evicted by a speculative line.

12. The processor of claim 1, wherein the cache manager is further configured to track and clear speculative lines when speculative execution is discarded.

13. The processor of claim 1, wherein the cache manager is further configured to prevent a cache probe instruction to access a speculative line in the split cache.

14. The processor of claim 1, wherein the cache manager is further configured to prevent an external source to access data in a speculative line, wherein the external source comprises at least one of a read line presence, a cache flush request, and a read from a second processor.

15. A cache, comprising:

a management unit;

a first plurality of architectural lines storing data fetched by an architectural instruction to be executed by a processor coupled with the cache, the architectural instruction, when executed, configured to modify an architecture of the cache; and

a second plurality of speculative lines storing data fetched by a speculative instruction, wherein the speculative instruction is invalidated by the management unit, wherein a number of the first plurality of architectural lines and the second plurality of speculative lines is dynamically determined based on a state of the processor.

16. The cache of claim 15, wherein the management unit is configured to convert a speculative line into an architectural line when the speculative line is accessed by an architectural instruction.

17. A method, comprising:

retrieving, with a fetch engine, a data associated with an instruction to be executed by a logic unit of a processor, from an external memory, wherein the instruction is one of a speculative instruction or an architectural instruction;

copying the data from a memory address to a data in a speculative line in a split cache, when the data is fetched by a speculative instruction, and to an architectural line in the split cache when the data is fetched by an architectural instruction;

selecting a number of speculative lines to be allocated in the split cache based on a current state of the processor;

preventing an architectural line in the split cache from being replaced by a speculative line; and

managing the number of speculative lines to be allocated in the split cache when a speculative instruction requests a data access based on the number of speculative lines relative to a number of architectural lines in the split cache.

18. The method of claim 17, further comprising modifying the number of speculative lines based on one of a least recently used state, a modified state, an exclusive state, a shared state, or an invalid state, of one or more architectural lines in the split cache.

19. The method of claim 17, further comprising increasing the number of speculative lines when an architectural line in the split cache is free or invalid.

20. The method of claim 17, further comprising selecting a number of accesses to the number of speculative lines before adjusting the number of speculative lines.