Method and apparatus for loading and storing misaligned data on an out-of-order execution computer system

Claim 2

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing a data load operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: (a) first misalignment detection circuitry disposed in either said EU or said at least one MOI component for detecting data chunk split misalignment for said load operation when said load operation is dispatched from said EU to said MEM; and (b) buffering circuitry disposed in said at least one MOI component and coupled to said first misalignment detection circuitry for buffering said load operation, and dispatching said load operation to said MEM non-speculatively if data chunk split misalignment is detected for said load operation.

Claim 3

The apparatus as set forth in claim 1, wherein, said apparatus further comprises (c) second misalignment detection circuitry disposed in either said EU or said at least one MOI component for detecting cache line split misalignment for said load operation when said load operation is dispatched from said EU to said MEM; said buffering circuitry being also coupled to said second misalignment detection circuitry further dispatches said load operation as successive aligned subset load operations, each destined for a subset of source memory locations that are data boundary aligned if cache line split misalignment is detected for said load operation.

Claim 4

The apparatus as set forth in claim 2, wherein, said apparatus further comprises (d) address and data size calculation circuitry coupled to said buffering circuitry and disposed in the same MOI component as said buffering circuitry for incrementing addresses and decrementing data sizes for each subsequent one of said successive aligned subset load operations to be dispatched for said load operation.

Claim 5

The apparatus as set forth in claim 2, wherein said apparatus further comprises: (d) third misalignment detection circuitry disposed in said at least one MOI component coupled to said buffering circuitry for detecting memory page split misalignment for said load operation when a first of said successive aligned subset load operations is dispatched by said buffering circuitry; (e) first fault generation circuitry disposed in said at least one MOI component and coupled to said third misalignment detection circuitry for faulting said first of said successive aligned subset load operations if memory page split misalignment is detected for said load operation; and (f) second fault generation circuitry disposed in said at least one MOI component for also faulting said first of said successive aligned subset load operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset load operations would result in fault.

Claim 6

The apparatus as set forth in claim 4, wherein said buffering circuitry further redispatches said first of said successive aligned subset load operations for said load operation if said first of said successive aligned subset load operation was only faulted for said load operation being also memory page split misaligned, and then dispatches each subsequent one of said successive aligned subset load operations.

Claim 7

The apparatus as set forth in claim 5, wherein, said EU comprises an address generation unit for generating a linear address for said load operation, said address generation unit comprising said first misalignment detection circuitry; said at least one MOI component comprises a memory order buffer (MOB) having a load buffer (LB) for buffering load operations, said LB comprising said buffering circuitry and said second misalignment detection circuitry; said at least one MOI component further comprises a data translation lookaside buffer (DTLB), said DTLB comprising said third alignment detection circuitry, and said first and second fault generation circuitry.

Claim 8

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing a data load operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: (a) first misalignment detection circuitry disposed in either said EU or said at least one MOI component for detecting cache line split misalignment for said load operation when said load operation is dispatched from said EU to said MEM; and (b) buffering circuitry disposed in said at least one MOI component and coupled to said first misalignment detection circuitry for buffering said load operation, and dispatching said load operation to said MEM non-speculatively if data chunk split misalignment is detected for said load operation, said buffering circuitry dispatching said cache line split misaligned load operation as successive aligned subset load operations, each destined for a subset of source memory locations that are data boundary aligned.

Claim 9

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing a data load operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: (a) first misalignment detection circuitry disposed in either said EU or said at least one MOI component for detecting cache line split misalignment for said load operation when said load operation is dispatched from said EU to said MEM; and (b) buffering circuitry disposed in said at least one MOI component and coupled to said first misalignment detection circuitry for buffering said load operation, and dispatching said load operation to said MEM non-speculatively if data chunk split misalignment is detected for said load operation, said buffering circuitry dispatching said cache line split misaligned load operation as successive aligned subset load operations, each destined for a subset of source memory locations that are data boundary aligned.

Claim 10

The apparatus as set forth in claim 7, wherein, said apparatus further comprises (c) address and data size calculation circuitry coupled to said buffering circuitry and disposed in the same MOI component as said buffering circuitry for incrementing addresses and decrementing data sizes for each subsequent one of said successive aligned subset load operations to be dispatched for said load operation.

Claim 11

The apparatus as set forth in claim 7, wherein said apparatus further comprises: (c) second misalignment detection circuitry disposed in either said EU or said at least one MOI component coupled to said buffering circuitry for detecting memory page split misalignment for said load operation when a first of said successive aligned subset load operations is dispatched by said buffering circuitry; (d) first fault generation circuitry disposed in said at least one MOI component and coupled to said second misalignment detection circuitry for faulting said first of said successive aligned subset load operations if memory page split misalignment is detected for said load operation; and (e) second fault generation circuitry disposed in said at least one MOI component for also faulting said first of said successive aligned subset load operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset load operations would result in fault.

Claim 12

The apparatus as set forth in claim 9, wherein said buffering circuitry further redispatches said first of said successive aligned subset load operations for said load operation if said first of said successive aligned subset load operation was only faulted for said load operation being also memory page split misaligned, and then dispatches each subsequent one of said successive aligned subset load operations.

Claim 13

The apparatus as set forth in claim 10, wherein, said at least one MOI component comprises a memory order buffer (MOB) having a load buffer (LB) for buffering load operations, said LB comprising said buffering circuitry and said first misalignment detection circuitry; said at least one MOI component further comprises a data translation lookaside buffer (DTLB), said DTLB comprising said second alignment detection circuitry, and said first and second fault generation circuitry.

Claim 14

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO s execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing a data load operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: (a) first misalignment detection circuitry disposed in either said EU or said at least one MOI component coupled to said buffering circuitry for detecting memory page split misalignment for said load operation when a first of successive aligned subset load operations is dispatched for said load operation; (b) first fault generation circuitry disposed in said at least one MOI component and coupled to said first misalignment detection circuitry for faulting said first of said successive aligned subset load operations if memory page split misalignment is detected for said load operation; and (c) second fault generation circuitry disposed in said at least one MOI component for also faulting said first of said successive aligned subset load operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset load operations would result in fault.

Claim 15

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO s execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing a data load operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: (a) first misalignment detection circuitry disposed in either said EU or said at least one MOI component coupled to said buffering circuitry for detecting memory page split misalignment for said load operation when a first of successive aligned subset load operations is dispatched for said load operation; (b) first fault generation circuitry disposed in said at least one MOI component and coupled to said first misalignment detection circuitry for faulting said first of said successive aligned subset load operations if memory page split misalignment is detected for said load operation; and (c) second fault generation circuitry disposed in said at least one MOI component for also faulting said first of said successive aligned subset load operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset load operations would result in fault.

Claim 16

The apparatus as set forth in claim 12, wherein, said at least one MOI component comprises a data translation lookaside buffer (DTLB), said DTLB comprising said first alignment detection circuitry, and said first and second fault generation circuitry.

Claim 17

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, a method for performing a data load operation of misaligned data from said memory subsystem into said execution unit, said method comprising the steps of: (a) detecting data chunk split misalignment for said load operation when said load operation is dispatched from said EU to said MEM; and (b) buffering said load operation, and dispatching said load operation to said MEM non-speculatively if data chunk split misalignment is detected for said load operation; (b.1) dispatching a first of said successive aligned subset load operations; (b.2) detecting memory page split misalignment for said load operation when said first of said successive aligned subset load operations is dispatched; (b.3) faulting said first of said successive aligned subset load operations if memory page split misalignment is also detected for said load operation; (b.4) faulting also said first of said successive aligned subset load operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset load operations would result in fault; (b.5) redispatching said first of said successive aligned subset load operations if said first of said successive aligned subset load operations is faulted for only memory page split misalignment being detected for said load operation; and (b.6) dispatching each subsequent one of said successive aligned subset load operations if either memory page split misalignment is not detected for said load operation, or memory page split misalignment is detected for said load operation but said first of said successive aligned subset load operation is redispatched.

Claim 18

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, a method for performing a data load operation of misaligned data from said memory subsystem into said execution unit, said method comprising the steps of: (a) detecting data chunk split misalignment for said load operation when said load operation is dispatched from said EU to said MEM; and (b) buffering said load operation, and dispatching said load operation to said MEM non-speculatively if data chunk split misalignment is detected for said load operation; (b.1) dispatching a first of said successive aligned subset load operations; (b.2) detecting memory page split misalignment for said load operation when said first of said successive aligned subset load operations is dispatched; (b.3) faulting said first of said successive aligned subset load operations if memory page split misalignment is also detected for said load operation; (b.4) faulting also said first of said successive aligned subset load operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset load operations would result in fault; (b.5) redispatching said first of said successive aligned subset load operations if said first of said successive aligned subset load operations is faulted for only memory page split misalignment being detected for said load operation; and (b.6) dispatching each subsequent one of said successive aligned subset load operations if either memory page split misalignment is not detected for said load operation, or memory page split misalignment is detected for said load operation but said first of said successive aligned subset load operation is redispatched.

Claim 19

The method as set forth in claim 14, wherein, said step (a) further comprises detecting cache line split misalignment for said load operation when said load operation is dispatched from said EU to said MEM if said load operation is detected to be data chunk split misaligned; said step (b) comprises dispatching said load operation as successive aligned subset load operations, each destined for a subset of source memory locations that are data boundary aligned if cache line split misalignment is also detected for said load operation.

Claim 20

The method as set forth in claim 15, wherein said step (b) comprises: (b.1) dispatching a first of said successive aligned subset load operations; (b.2) detecting memory page split misalignment for said load operation when said first of said successive aligned subset load operations is dispatched; (b.3) faulting said first of said successive aligned subset load operations if memory page split misalignment is also detected for said load operation; (b.4) faulting also said first of said successive aligned subset load operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset load operations would result in fault; (b.5) redispatching said first of said successive aligned subset load operations if said first of said successive aligned subset load operations is faulted for only memory page split misalignment being detected for said load operation; and (b.6) dispatching each subsequent one of said successive aligned subset load operations if either memory page split misalignment is not detected for said load operation, or memory page split misalignment is detected for said load operation but said first of said successive aligned subset load operation is redispatched.

Claim 21

An out-of-order (OOO) execution processor comprising: (a) a data cache (DC) for caching data; (b) an instruction fetch and issue unit for fetching and issuing instructions; (c) an execution unit (EU) coupled to the instruction fetch and issue unit comprising an address generation unit (AGU) for generating a linear address for a load operation dispatched from said EU, said AGU having a first misalignment detection circuitry for detecting data chunk split misalignment for said load operation when said load operation is dispatched from said EU to said MEM; (d) a memory order buffer (MOB) coupled to said EU and said DC comprising a load buffer (LB) having buffering circuitry for buffering said load operation dispatched from said EU, and dispatching said load operation to said DC non-speculatively if data chunk split misalignment is detected for said load operation.

Claim 22

An out-of-order (OOO) execution processor comprising: (a) a data cache (DC) for caching data; (b) an instruction fetch and issue unit for fetching and issuing instructions; (c) an execution unit (EU) coupled to the instruction fetch and issue unit comprising an address generation unit (AGU) for generating a linear address for a load operation dispatched from said EU, said AGU having a first misalignment detection circuitry for detecting data chunk split misalignment for said load operation when said load operation is dispatched from said EU to said MEM; (d) a memory order buffer (MOB) coupled to said EU and said DC comprising a load buffer (LB) having buffering circuitry for buffering said load operation dispatched from said EU, and dispatching said load operation to said DC non-speculatively if data chunk split misalignment is detected for said load operation.

Claim 23

The processor as set forth in claim 17, wherein, said LB further having second misalignment detection circuitry for detecting cache line split misalignment for said load operation when said load operation is dispatched from said EU to said MEM; said buffering circuitry being also coupled to said second misalignment detection circuitry further dispatches said load operation as successive aligned subset load operations, each destined for a subset of source memory locations that are data boundary aligned if cache line split misalignment is detected for said load operation.

Claim 24

The processor as set forth in claim 18, wherein said processor further comprises (e) a data translation lookaside buffer (DTLB) coupled to said MOB and said DC for translating a linear address to a physical address, and performing a number of fault detections, said DTLB having fault detection circuitry comprising: (e.1) third misalignment detection circuitry for detecting memory page split misalignment for said load operation when a first of said successive aligned subset load operations is dispatched by said buffering circuitry; (e.2) first fault generation circuitry coupled to said third misalignment detection circuitry for faulting said first of said successive aligned subset load operations if memory page split misalignment is detected for said load operation; and (e.3) second fault generation circuitry for also faulting said first of said successive aligned subset load operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset load operations would result in fault.

Claim 25

The processor as set forth in claim 19, wherein said buffering circuitry further redispatches said first of said successive aligned subset load operations for said load operation if said first of said successive aligned subset load operation was only faulted for said load operation being also memory page split misaligned, and then dispatched each subsequent one of said successive aligned subset load operations.

Claim 26

The processor as forth in claim 20, wherein said processor is disposed in a single silicon based integrated circuit.

Claim 27

An out-of-order (OOO) computer system comprising: (a) a main memory for storing data; (b) a data cache (DC) coupled to said main memory for caching a subset of said data stored in said main memory; (c) an instruction fetch and issue unit for fetching and issuing instructions; (d) an execution unit (EU) coupled to the instruction fetch and issue unit comprising a first misalignment detection circuitry for detecting data chunk split misalignment for said load operation when said load operation is dispatched from said EU to said MEM; (e) a memory order buffer (MOB) coupled to said EU and said DC comprising buffering circuitry for buffering said load operation dispatched from said EU, and dispatching said load operation to said DC non-speculatively if data chunk split misalignment is detected for said load operation.

Claim 28

An out-of-order (OOO) computer system comprising: (a) a main memory for storing data; (b) a data cache (DC) coupled to said main memory for caching a subset of said data stored in said main memory; (c) an instruction fetch and issue unit for fetching and issuing instructions; (d) an execution unit (EU) coupled to the instruction fetch and issue unit comprising a first misalignment detection circuitry for detecting data chunk split misalignment for said load operation when said load operation is dispatched from said EU to said MEM; (e) a memory order buffer (MOB) coupled to said EU and said DC comprising buffering circuitry for buffering said load operation dispatched from said EU, and dispatching said load operation to said DC non-speculatively if data chunk split misalignment is detected for said load operation.

Claim 29

The computer system as set forth in claim 22, wherein, said MOB further having second misalignment detection circuitry for detecting cache line split misalignment for said load operation when said load operation is dispatched from said EU to said MEM; said buffering circuitry being also coupled to said second misalignment detection circuitry further dispatches said load operation as successive aligned subset load operations, each destined for a subset of source memory locations that are data boundary aligned if cache line split misalignment is detected for said load operation.

Claim 30

The computer system as set forth in claim 23, wherein said computer system further comprises (f) a data translation lookaside buffer (DTLB) coupled to said MOB and said DC for translating a linear address to a physical address, and performing a number of fault detections, said DTLB having fault detection circuitry comprising: (f.1) third misalignment detection circuitry for detecting memory page split misalignment for said load operation when a first of said successive aligned subset load operations is dispatched by said buffering circuitry; (f.2) first fault generation circuitry coupled to said third misalignment detection circuitry for faulting said first of said successive aligned subset load operations if memory page split misalignment is detected for said load operation; and (d.3) second fault generation circuitry for also faulting said first of said successive aligned subset load operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset load operations would result in fault.

Claim 31

The computer system as set forth in claim 24, wherein said buffering circuitry further redispatches said first of said successive aligned subset load operations for said load operation if said first of said successive aligned subset load operation was only faulted for said load operation being also memory page split misaligned, and then dispatched each subsequent one of said successive aligned subset load operations.

Claim 32

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing a data store operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: a) first misalignment detection circuitry disposed in either said EU or said at least one MOI component for detecting data chunk split misalignment for said store operation when said store operation is dispatched from said EU to said MEM; and b) buffering circuitry disposed in said at least one MOI component and coupled to said first misalignment detection circuitry for buffering said store operation, allowing said misaligned data of said store operation to be promoted to a system commitment ready state, and dispatching said store operation to said MEM with an appropriate byte mask, after said misaligned data of said store operation has been promoted to said system commitment ready state, if data chunk split misalignment is detected for said store operation.

Claim 33

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing a data store operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: a) first misalignment detection circuitry disposed in either said EU or said at least one MOI component for detecting data chunk split misalignment for said store operation when said store operation is dispatched from said EU to said MEM; and b) buffering circuitry disposed in said at least one MOI component and coupled to said first misalignment detection circuitry for buffering said store operation, allowing said misaligned data of said store operation to be promoted to a system commitment ready state, and dispatching said store operation to said MEM with an appropriate byte mask, after said misaligned data of said store operation has been promoted to said system commitment ready state, if data chunk split misalignment is detected for said store operation.

Claim 34

The apparatus as set forth in claim 26, wherein, said apparatus further comprises (c) second misalignment detection circuitry disposed in either said EU or said at least one MOI component for detecting cache line split misalignment for said store operation when said store operation is dispatched from said EU to said MEM; said buffering circuitry being also coupled to said second misalignment detection circuitry prevents said misaligned data of said store operation from being promoted to said system commitment ready state if cache line split misalignment is detected for said operation until said store operation is ready to be dispatched to said MEM, and dispatches said store operation as successive aligned subset store operations, each destined for a subset of source memory locations that are data boundary aligned, concurrently with committing said misaligned data of said store operation to system state.

Claim 35

The apparatus as set forth in claim 27, wherein, said apparatus further comprises (d) address and data size calculation circuitry coupled to said buffering circuitry and disposed in the same MOI component as said buffering circuitry for incrementing addresses and decrementing data sizes for each subsequent one of said successive aligned subset store operations to be dispatched for said store operation.

Claim 36

The apparatus as set forth in claim 27, wherein said apparatus further comprises: (d) third misalignment detection circuitry disposed in either said EU or said at least one MOI component coupled to said buffering circuitry for detecting memory page split misalignment for said store operation when a first of said successive aligned subset store operations is dispatched by said buffering circuitry; (e) first fault generation circuitry disposed in said at least one MOI component and coupled to said third misalignment detection circuitry for faulting said first of said successive aligned subset store operations if memory page split misalignment is detected for said store operation; and (f) second fault generation circuitry disposed in said at least one MOI component for also faulting said first of said successive aligned subset store operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset store operations would result in fault.

Claim 37

The apparatus as set forth in claim 29, wherein said buffering circuitry further redispatches said first of said successive aligned subset store operations for said store operation if said first of said successive aligned subset store operation was only faulted for said store operation being also memory page split misaligned, and then dispatches each subsequent one of said successive aligned subset store operations.

Claim 38

The apparatus as set forth in claim 30, wherein, said EU comprises an address generation unit for generating a linear address for said store operation, said address generation unit comprising said first misalignment detection circuitry; said at least one MOI component comprises a memory order buffer (MOB) having a store buffer (SB) for buffering store operations, said SB comprising said buffering circuitry and said second misalignment detection circuitry; said at least one MOI component further comprises a data translation lookaside buffer (DTLB), said DTLB comprising said third alignment detection circuitry, and said first and second fault generation circuitry.

Claim 39

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing a data store operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: (a) first misalignment detection circuitry disposed in either said EU or said at least one MOI component for detecting cache line split misalignment for said store operation when said store operation is dispatched from said EU to said MEM; and (b) buffering circuitry disposed in said at least one MOI component and coupled to said first misalignment detection circuitry for buffering said store operation, preventing said misaligned data of said store operation from being promoted to a system commitment ready state if cache line split misalignment is detected for said store operation until said store operation is ready to be dispatched to said MEM, and dispatching said cache line split misaligned store operation as successive aligned subset store operations, each destined for a subset of source memory locations that are data boundary aligned, concurrent with committing said misaligned data of said store operation to system state.

Claim 40

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing a data store operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: (a) first misalignment detection circuitry disposed in either said EU or said at least one MOI component for detecting cache line split misalignment for said store operation when said store operation is dispatched from said EU to said MEM; and (b) buffering circuitry disposed in said at least one MOI component and coupled to said first misalignment detection circuitry for buffering said store operation, preventing said misaligned data of said store operation from being promoted to a system commitment ready state if cache line split misalignment is detected for said store operation until said store operation is ready to be dispatched to said MEM, and dispatching said cache line split misaligned store operation as successive aligned subset store operations, each destined for a subset of source memory locations that are data boundary aligned, concurrent with committing said misaligned data of said store operation to system state.

Claim 41

The apparatus as set forth in claim 32, wherein, said apparatus further comprises (c) address and data size calculation circuitry coupled to said buffering circuitry and disposed in the same MOI component as said buffering circuitry for incrementing addresses and decrementing data sizes for each subsequent one of said successive aligned subset store operations to be dispatched for said store operation.

Claim 42

The apparatus as set forth in claim 32, wherein said apparatus further comprises: (c) second misalignment detection circuitry disposed in either said EU or said at least one MOI component coupled to said buffering circuitry for detecting memory page split misalignment for said store operation when a first of said successive aligned subset store operations is dispatched by said buffering circuitry; (d) first fault generation circuitry disposed in said at least one MOI component and coupled to said second misalignment detection circuitry for faulting said first of said successive aligned subset store operations if memory page split misalignment is detected for said store operation; and (e) second fault generation circuitry disposed in said at least one MOI component for also faulting said first of said successive aligned subset store operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset store operations would result in fault.

Claim 43

The apparatus as set forth in claim 34, wherein said buffering circuitry further redispatches said first of said successive aligned subset store operations for said store operation if said first of said successive aligned subset store operation was only faulted for said store operation being also memory page split misaligned, and then dispatches each subsequent one of said successive aligned subset store operations.

Claim 44

The apparatus as set forth in claim 35, wherein, said at least one MOI component comprises a memory order buffer (MOB) having a store buffer (SB) for buffering store operations, said SB comprising said buffering circuitry and said first misalignment detection circuitry; said at least one MOI component further comprises a data translation lookaside buffer (DTLB), said DTLB comprising said second alignment detection circuitry, and said first and second fault generation circuitry.

Claim 45

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing a data store operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: (a) first misalignment detection circuitry disposed in either said EU or said at least one MOI component coupled to said buffering circuitry for detecting memory page split misalignment for said store operation when a first of successive aligned subset store operations is dispatched for said store operation; (b) first fault generation circuitry disposed in said at least one MOI component and coupled to said first misalignment detection circuitry for faulting said first of said successive aligned subset store operations if memory page split misalignment is detected for said store operation; and (c) second fault generation circuitry disposed in said at least one MOI component for also faulting said first of said successive aligned subset store operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset store operations would result in fault.

Claim 46

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing a data store operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: (a) first misalignment detection circuitry disposed in either said EU or said at least one MOI component coupled to said buffering circuitry for detecting memory page split misalignment for said store operation when a first of successive aligned subset store operations is dispatched for said store operation; (b) first fault generation circuitry disposed in said at least one MOI component and coupled to said first misalignment detection circuitry for faulting said first of said successive aligned subset store operations if memory page split misalignment is detected for said store operation; and (c) second fault generation circuitry disposed in said at least one MOI component for also faulting said first of said successive aligned subset store operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset store operations would result in fault.

Claim 47

The apparatus as set forth in claim 37, wherein, said at least one MOI component comprises a data translation lookaside buffer (DTLB), said DTLB comprising said first alignment detection circuitry, and said first and second fault generation circuitry.

Claim 48

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, a method for performing a data store operation of misaligned data from said memory subsystem into said execution unit, said method comprising the steps of: (a) detecting data chunk split misalignment for said store operation when said store operation is dispatched from said EU to said MEM; and (b) buffering said store operation, allowing said misaligned data of said store operation to be promoted to a system commitment ready state, and dispatching said store operation to said MEM with an appropriate byte mask, after said misaligned data of said store operation has been promoted to said system commitment ready state, if data chunk split misalignment is detected for said store operation.

Claim 49

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, a method for performing a data store operation of misaligned data from said memory subsystem into said execution unit, said method comprising the steps of: (a) detecting data chunk split misalignment for said store operation when said store operation is dispatched from said EU to said MEM; and (b) buffering said store operation, allowing said misaligned data of said store operation to be promoted to a system commitment ready state, and dispatching said store operation to said MEM with an appropriate byte mask, after said misaligned data of said store operation has been promoted to said system commitment ready state, if data chunk split misalignment is detected for said store operation.

Claim 50

The method as set forth in claim 39, wherein, said step (a) further comprises detecting cache line split misalignment for said store operation when said store operation is dispatched from said EU to said MEM if said store operation is detected to be data chunk split misaligned; said step (b) comprises preventing said misaligned data of said store operation from being promoted to said system commitment ready state until said store operation is ready to be dispatched to said MEM instead if cache line split misalignment is also detected for said store operation, and dispatching said store operation as successive aligned subset store operations, each destined for a subset of source memory locations that are data boundary aligned, concurrent with committing said misaligned data of said store operation to system state.

Claim 51

The method as set forth in claim 40, wherein said dispatching of said cache line split misaligned store operation as successive aligned subset store operations in said step (b) comprises: (b.1) dispatching a first of said successive aligned subset store operations; (b.2) detecting memory page split misalignment for said store operation when said first of said successive aligned subset store operations is dispatched; (b.3) faulting said first of said successive aligned subset store operations if memory page split misalignment is also detected for said store operation; (b.4) faulting also said first of said successive aligned subset store operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset store operations would result in fault; (b.5) redispatching said first of said successive aligned subset store operations if said first of said successive aligned subset store operations is faulted for only memory page split misalignment being detected for said store operation; and (b.6) dispatching each subsequent one of said successive aligned subset store operations if either memory page split misalignment is not detected for said store operation, or memory page split misalignment is detected for said store operation but said first of said successive aligned subset store operation is redispatched.

Claim 52

An out-of-order (OOO) execution processor comprising: (a) a data cache (DC) for caching data; (b) an instruction fetch and issue unit for fetching and issuing instructions; (c) an execution unit (EU) coupled to said instruction fetch and issue unit comprising an address generation unit (AGU) for generating a linear address for a store operation dispatched from said EU, said AGU having a first misalignment detection circuitry for detecting data chunk split misalignment for said store operation when said store operation is dispatched from said EU to said MEM; (d) a memory order buffer (MOB) coupled to said EU and said DC comprising a store buffer (SB) having buffering circuitry for buffering said store operation dispatched from said EU, allowing said misaligned data of said store operation to be promoted to a system commitment ready state, and dispatching said store operation to said DC with an appropriate byte mask, after said misaligned data of said store operation has been promoted to said system commitment ready state, if data chunk split misalignment is detected for said store operation.

Claim 53

An out-of-order (OOO) execution processor comprising: (a) a data cache (DC) for caching data; (b) an instruction fetch and issue unit for fetching and issuing instructions; (c) an execution unit (EU) coupled to said instruction fetch and issue unit comprising an address generation unit (AGU) for generating a linear address for a store operation dispatched from said EU, said AGU having a first misalignment detection circuitry for detecting data chunk split misalignment for said store operation when said store operation is dispatched from said EU to said MEM; (d) a memory order buffer (MOB) coupled to said EU and said DC comprising a store buffer (SB) having buffering circuitry for buffering said store operation dispatched from said EU, allowing said misaligned data of said store operation to be promoted to a system commitment ready state, and dispatching said store operation to said DC with an appropriate byte mask, after said misaligned data of said store operation has been promoted to said system commitment ready state, if data chunk split misalignment is detected for said store operation.

Claim 54

The processor as set forth in claim 42, wherein, said SB further having second misalignment detection circuitry for detecting cache line split misalignment for said store operation when said store operation is dispatched from said EU to said MEM; said buffering circuitry being also coupled to said second misalignment detection circuitry prevents said misaligned data of said store operation from being promoted to said system commitment ready state until said store operation is ready to be dispatched to said MEM if cache line split misalignment is detected for said store operation, and dispatches said store operation as successive aligned subset store operations, each destined for a subset of source memory locations that are data boundary aligned, concurrent with committing said misaligned data of said store operation to system state.

Claim 55

The processor as set forth in claim 43, wherein said processor further comprises (e) a data translation lookaside buffer (DTLB) coupled to said MOB and said DC for translating a linear address to a physical address, and performing a number of fault detections, said DTLB having fault detection circuitry comprising: (e.1) third misalignment detection circuitry for detecting memory page split misalignment for said store operation when a first of said successive aligned subset store operations is dispatched by said buffering circuitry; (e.2) first fault generation circuitry coupled to said third misalignment detection circuitry for faulting said first of said successive aligned subset store operations if memory page split misalignment is detected for said store operation; and (e.3) second fault generation circuitry for also faulting said first of said successive aligned subset store operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset store operations would result in fault.

Claim 56

The processor as set forth in claim 44, wherein said buffering circuitry further redispatches said first of said successive aligned subset store operations for said store operation if said first of said successive aligned subset store operation was only faulted for said store operation being also memory page split misaligned, and then dispatched each subsequent one of said successive aligned subset store operations.

Claim 57

The processor as forth in claim 45, wherein said processor is disposed in a single silicon based integrated circuit.

Claim 58

An out-of-order (OOO) computer system comprising: (a) a main memory for storing data; (b) a data cache (DC) coupled to said main memory for caching a subset of said data stored in said main memory; (c) an instruction fetch and issue unit for fetching and issuing instructions; (d) an execution unit (EU) coupled to said instruction fetch and issue unit comprising a first misalignment detection circuitry for detecting data chunk split misalignment for said store operation when said store operation is dispatched from said EU to said MEM; (e) a memory order buffer (MOB) coupled to said EU and said DC comprising buffering circuitry for buffering said store operation dispatched from said EU, allowing said misaligned data of said store operation to be promoted to a system commitment ready state, and dispatching said store operation to said DC with an appropriate byte mask, after said misaligned data of said stoe operation has been promoted to said system commitment ready state, if data chunk split misalignment is detected for said store operation.

Claim 59

An out-of-order (OOO) computer system comprising: (a) a main memory for storing data; (b) a data cache (DC) coupled to said main memory for caching a subset of said data stored in said main memory; (c) an instruction fetch and issue unit for fetching and issuing instructions; (d) an execution unit (EU) coupled to said instruction fetch and issue unit comprising a first misalignment detection circuitry for detecting data chunk split misalignment for said store operation when said store operation is dispatched from said EU to said MEM; (e) a memory order buffer (MOB) coupled to said EU and said DC comprising buffering circuitry for buffering said store operation dispatched from said EU, allowing said misaligned data of said store operation to be promoted to a system commitment ready state, and dispatching said store operation to said DC with an appropriate byte mask, after said misaligned data of said stoe operation has been promoted to said system commitment ready state, if data chunk split misalignment is detected for said store operation.

Claim 60

The computer system as set forth in claim 47, wherein, said MOB further having second misalignment detection circuitry for detecting cache line split misalignment for said store operation when said store operation is dispatched from said EU to said MEM; said buffering circuitry being also coupled to said second misalignment detection circuitry prevents said misaligned data of said store operation from being promoted to said system commitment ready state until said store operation is ready to be dispatched to said MEM if cache line split misalignment is detected for said store operation, and dispatches said store operation as successive aligned subset store operations, each destined for a subset of source memory locations that are data boundary aligned, concurrent with committing said misaligned data of said store operation to system state.

Claim 61

The computer system as set forth in claim 48, wherein said computer system further comprises (e) a data translation lookaside buffer (DTLB) coupled to said MOB and said DC for translating a linear address to a physical address, and performing a number of fault detections, said DTLB having fault detection circuitry comprising: (e.1) third misalignment detection circuitry for detecting memory page split misalignment for said store operation when a first of said successive aligned subset store operations is dispatched by said buffering circuitry; (e.2) first fault generation circuitry coupled to said third misalignment detection circuitry for faulting said first of said successive aligned subset store operations if memory page split misalignment is detected for said store operation; and (e.3) second fault generation circuitry for also faulting said first of said successive aligned subset store operations if at least one reference to a memory page to be referenced by a subsequent one of said successive aligned subset store operations would result in fault.

Claim 62

The computer system as set forth in claim 49, wherein said buffering circuitry further redispatches said first of said successive aligned subset store operations for said store operation if said first of said successive aligned subset store operation was only faulted for said store operation being also memory page split misaligned, and then dispatched each subsequent one of said successive aligned subset store operations.

Claim 63

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing data load operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: a) at least one misalignment detection circuitry disposed in either said EU or said at least one MOI component for detecting at least one type of data misalignment for said load operation when said load operation is dispatched from said EU to said MEM; and b) buffering circuitry disposed in said at least one MOI component and coupled to said at least one misalignment detection circuitry for buffering said load operation, and dispatching said load operation to said MEM in a manner depending on whether data misalignment is detected for said load operation, and the type of data misalignment detected.

Claim 64

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing data load operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: a) at least one misalignment detection circuitry disposed in either said EU or said at least one MOI component for detecting at least one type of data misalignment for said load operation when said load operation is dispatched from said EU to said MEM; and b) buffering circuitry disposed in said at least one MOI component and coupled to said at least one misalignment detection circuitry for buffering said load operation, and dispatching said load operation to said MEM in a manner depending on whether data misalignment is detected for said load operation, and the type of data misalignment detected.

Claim 65

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing a data store operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: a) at least one misalignment detection circuitry disposed in either said EU or said at least one MOI component for detecting at least one type of data misalignment for said store operation when said store operation is dispatched from said EU to said MEM; and b) buffering circuitry disposed in said at least one MOI component and coupled to said at least one misalignment detection circuitry for buffering said store operation, and dispatching said store operation to said MEM in a manner depending on whether data misalignment is detected for said store operation, and the type of data misalignment detected.

Claim 66

In an out-of-order (OOO) computer system comprising an instruction fetch and issue unit for fetching and issuing instructions, an OOO execution unit (EU) for executing instructions, a memory subsystem (MEM) for storing data, and at least one memory ordering interface (MOI) component coupled to said EU and said MEM for interfacing between said EU and MEM, an apparatus for performing a data store operation of misaligned data from said memory subsystem into said execution unit, said apparatus comprising: a) at least one misalignment detection circuitry disposed in either said EU or said at least one MOI component for detecting at least one type of data misalignment for said store operation when said store operation is dispatched from said EU to said MEM; and b) buffering circuitry disposed in said at least one MOI component and coupled to said at least one misalignment detection circuitry for buffering said store operation, and dispatching said store operation to said MEM in a manner depending on whether data misalignment is detected for said store operation, and the type of data misalignment detected.