International Business Machines Corporation

Thomas G Koehler

Frank Siegfried Lehnert

Markus Helms

Uwe Brandt

Christian Jacobi

Markus Kaltenbach

Ute Gaertner

Lisa Cranton Heller

Jennifer A Navarro

Chung-Lung K Shum

Michael F Fee

Martin Recktenwald

Ulrich Mayer

Brian David Barrick

Erwin F Pfeffer

Aaron Tsai

Bernd Nerz

Scott Barnett Swaney

Klaus J Getzlaff

Willm Hinrichs

Gilles Gervais

Norbert Schumacher

Joerg Deutschle

Thomas Fuchs

Ingemar Holm

Simon H Friedmann

Siegmund Schlechter

Helmut Kohler

Gerhard Zilles

Driving address translations in a microprocessor system by sending a rejected Lx+1 cache request from a first set of caches of a first level Lx to a central request unit, transferring an Lx+1 cache request having a translation of a virtual address into a physical address stored in a first buffer, from the central request unit to the at least one Lx+1 cache, and keeping an Lx+1 cache request lacking a translation of a virtual address into a physical address stored in the first buffer, pending in the central request unit.

Fetch request arbiter

An address translation facility is provided for multiple virtualization levels, where a guest virtual address may be translated to a guest non-virtual address, the guest non-virtual address corresponding without translation to a host virtual address, and the host virtual address may be translated to a host non-virtual address, where translation within a virtualization level may be specified as a sequence of accesses to address translation tables. The address translation facility may include a first translation engine and a second translation engine, where the first and second translation engines each have capacity to perform address translation within a single virtualization level of the multiple virtualization levels. In operation, based on the first translation engine performing a guest level translation, the second translation engine may perform a host level translation of a resulting guest non-virtual address to a host non-virtual address based on the guest level translation by the first translation engine.

Multi-engine address translation facility

Embodiments include techniques for using a zone-SDID mapping for translation lookaside buffer (TLB) purges, the techniques include receiving a zone purge request, including zone attribute information, and searching for matching zone attribute information in a zone register using the zone purge request. The techniques also include computing, based at least in part on the search, a state descriptor identifier (SDID) vector for each matching zone of the zone register, and reading TLB entries referenced in the zone purge request. The techniques include comparing an SDID of the TLB entry against an SDID specified in the SDID vector, and purging the TLB entries based on the comparison.

Zone-SDID mapping scheme for TLB purges

Embodiments include techniques for using a zone-SDID mapping for translation lookaside buffer (TLB) purges, the embodiments include receiving a zone purge request, including zone attribute information, and searching for matching zone attribute information in a zone register using the zone purge request. Embodiments also include computing, based at least in part on the search, a state descriptor identifier (SDID) vector for each matching zone of the zone register, and reading TLB entries referenced in the zone purge request. Embodiments include comparing an SDID of the TLB entry against an SDID specified in the SDID vector, and purging the TLB entries based on the comparison.

Methods and apparatuses for performing a quiesce operation in a multithread environment is provided. A processor receives a first thread quiesce request from a first thread executing on the processor. A processor sends a first processor quiesce request to a system controller to initiate a quiesce operation. A processor performs one or more operations of the first thread based, at least in part, on receiving a response from the system controller.

Quiesce handling in multithreaded environments

The present disclosure relates to a method of operating a hierarchical translation lookaside buffer (TLB). The TLB comprises at least two TLB levels, wherein a given entry of the upper level TLB comprises a portion of bits for indicating related entries in the lower level TLB. The method comprises the following when a TLB miss is encountered for a requested first virtual address. A first table walk is performed to obtain the absolute memory address for the first virtual address. A logical tag is stored. The logical tag comprises the portion of bits that has been identified in association with the first table walk. In response to determining that a concurrent second table walk, of the ongoing first table walk, that has a second virtual address that addresses the same entry in the upper level TLB as the first virtual address is writing in the TLB, the stored logical tag may be incremented. And, the incremented logical tag and the obtained absolute memory address may be stored in the TLB.

Link consistency in a hierarchical TLB with concurrent table walks

A translation engine for a processor system to translate virtual memory addresses to physical addresses of a main memory of a computer system is provided, where a sequence of accesses to multiple address translation tables is performed to support a computer system virtualization level. The translation engine includes: a first pipeline having at least, a first pipeline stage to receive a value for an original address or an address translation table entry requested in a previous pass through the first pipeline; a second pipeline stage using the value as an operand in a translation operation eventually yielding the address translation result or yielding a table index to an entry in a next address translation table; and a third pipeline stage issuing a read request for the entry in the next address translation table.

Translating virtual memory addresses to physical addresses

A computer system has physical processors supporting virtual addressing. Virtual processors represent multiple execution threads, and logical state of all threads of a virtual processor is stored in a state descriptor field in main memory when the virtual processor is removed from one of the physical processors. Each thread has assigned a thread identifier, which is unique in the respective virtual processor only, and each virtual processor has assigned a unique state descriptor identifier. Address translations for the threads of the multiple virtual processors under their respective thread identifier and state descriptor identifier are stored, and a sequence number is generated when an entry in the translation lookaside buffer is created. The sequence number is stored together with a respective thread identifier, state descriptor identifier, and a valid bit in a respective translation lookaside buffer entry. A determination is made as to whether an address translation is stored in the translation lookaside buffer for a current thread identifier and a current state descriptor identifier by comparing the translation lookaside buffer entries with the entries in the state descriptor/thread array.

Multiprocessor computer system

Growing community of inventors

Holzgerlingen, Germany

Thomas G Koehler