Apple Inc.

Jesse Pan

Ramesh B Gunna

James B Keller

Tse-Yu Yeh

Wei-Han Lien

Po-Yung Chang

A system and method for efficiently handling translation look-aside buffer (TLB) misses. A memory management unit (MMU) detects when a given virtual address misses in each available translation-lookaside-buffer (TLB). The MMU determines whether a memory access operation associated with the given virtual address is the oldest, uncompleted memory access operation in a scheduler. If this is the case, a demand table walk (TW) request may be stored in an available entry in a TW queue. During this time, the utilization of the memory subsystem resources may be low. While a demand TW request is stored in the TW queue, subsequent speculative TW requests may be stored in the TW queue. When the TW queue does not store a demand TW request, no more entries of the TW queue may be allocated to store TW requests.

Memory management unit speculative hardware table walk scheme

In one embodiment, a processor comprises a memory management unit (MMU) and an interface unit coupled to the MMU and to an interface unit of the processor. The MMU comprises a queue configured to store pending hardware-generated page table entry (PTE) updates. The interface unit is configured to receive a synchronization operation on the interface that is defined to cause the pending hardware-generated PTE updates, if any, to be written to memory. The MMU is configured to accept a subsequent hardware-generated PTE update generated subsequent to receiving the synchronization operation even if the synchronization operation has not completed on the interface. In some embodiments, the MMU may accept the subsequent PTE update responsive to transmitting the pending PTE updates from the queue. In other embodiments, the pending PTE updates may be identified in the queue and subsequent updates may be received.

R and C bit update handling

In one embodiment, a processor comprises a scheduler configured to issue a first instruction operation to be executed and an execution core coupled to the scheduler. Configured to execute the first instruction operation, the execution core comprises a plurality of replay sources configured to cause a replay of the first instruction operation responsive to detecting at least one of a plurality of replay cases. The scheduler is configured to inhibit issuance of the first instruction operation subsequent to the replay for a subset of the plurality of replay cases. The scheduler is coupled to receive an acknowledgement indication corresponding to each of the plurality of replay cases in the subset, and is configured to inhibit issuance of the first instruction operation until the acknowledgement indication is asserted that corresponds to an identified replay case of the subset.

Replay reduction for power saving

In one embodiment, a processor comprises a scheduler configured to issue a first instruction operation to be executed and an execution core coupled to the scheduler. Configured to execute the first instruction operation, the execution core comprises a plurality of replay sources configured to cause a replay of the first instruction operation responsive to detecting at least one of a plurality of replay cases. The scheduler is configured to inhibit issuance of the first instruction operation subsequent to the replay for a subset of the plurality of replay cases. The scheduler is coupled to receive an acknowledgement indication corresponding to each of the plurality of replay cases in the subset, and is configured to inhibit issuance of the first instruction operation until the acknowledge indication is asserted that corresponds to an identified replay case of the subset.

Growing community of inventors

San Jose, CA, United States of America

Jesse Pan