Intel Corporation

John G Holm

Subhankar Panda

Theodros Yigzaw

Gaurav Porwal

Glenn J Hinton

Avinash Sodani

Ashok Raj

Deborah T Marr

Sarathy Jayakumar

Ronak Singhal

Sebastien Hily

Steven E Raasch

Kuljit Singh Bains

Bret L Toll

Mohan J Kumar

Jason W Brandt

Wei P Chen

Rajat Agarwal

Kjersten E Criss

Hsing-Min Chen

Omar Avelar Suarez

Gilbert A Neiger

Rajesh M Sankaran

Martin Guy Dixon

Shubhendu Sekhar Mukherjee

Matthew C Merten

James B Crossland

Wei Wu

Narayan Ranganathan

Sergiu D Ghetie

Jing Ling

Leena K Puthiyedath

Adrian C Moga

Deep K Buch

Arijit Biswas

Andrew M Rudoff

Sreenivas Mandava

Morris Marden

Rebecca Z Loop

Santhosh Srinath

Anthony Luck

Qiuxu Zhuo

Raicsh Parthasarathy

Eswar Konduru

Hugo Enrique Gonzalez Chavero

Satyaprakash Nanda

Guarav Porwal

Vaibhav Singh

Karthik Ananthanarayanan

In one embodiment, an apparatus comprises: a first circuit to compact a plurality of data blocks to a compacted data block and to compact a plurality of error correction codes (ECCs) associated with the plurality of data blocks to a compacted ECC; and a second circuit to generate a generated ECC for the compacted data block. The apparatus may directly send the plurality of data blocks to a destination circuit without error detection on the plurality of data blocks based at least in part on the compacted ECC and the generated ECC. Other embodiments are described and claimed.

System, method and apparatus for reducing power consumption of error correction coding using compacted data blocks

Techniques and mechanisms for supporting machine check functionality with a handler which is implemented in firmware. In an embodiment, a processor executes first firmware code to implement a machine check event (MCE) detector. The MCE detector detects a hardware error of a platform which includes the processor, and generates a call to invoke an MCE handler which the processor implements by executing second firmware code. The MCE handler is called, outside of a software context, to attempt a recovery from the hardware error. The call is performed independent of any system management interrupt being based on the detected hardware error. In another embodiment, another MCE handler of an operating system is conditionally invoked where it is determined that the attempted recovery by the first MCE handler was unsuccessful.

Firmware first handling of a machine check event

A memory subsystem with error checking and scrubbing (ECS) logic on-device on the memory can adapt the rate of ECS operations in response to detection of errors in the memory when the memory device is in automatic ECS mode. The ECS logic can include an indication of rows of memory that have been offlined by the host. The ECS logic can skip the offlined rows in ECS operation counts. The ECS logic can include requests or hints by the host to have ECS operations performed. An internal address generator of the ECS logic can select between generated addresses and the hints. The system can allow a memory controller to detect multibit errors (MBEs) related to a specific address of the associated memory. When the detected MBEs indicate a pattern of errors, the memory controller triggers a row hammer response for the specific address.

Adaptive internal error scrubbing and error handling

A memory subsystem includes memory devices with space dynamically allocated for improvement of reliability, availability, and serviceability (RAS) in the system. Error checking and correction (ECC) logic detects an error in all or a portion of a memory device. In response to error detection, the system can dynamically perform one or more of: allocate active memory device space for sparing to spare a failed memory segment; write a poison pattern into a failed cacheline to mark it as failed; perform permanent fault detection (PFD) and adjust application of ECC based on PFD detection; or, spare only a portion of a device and leave another portion active, including adjusting ECC based on the spared portion. The error detection can be based on bits of an ECC device, and error correction based on those bits and additional bits stored on the data devices.

Adaptive error correction to improve system memory reliability, availability, and serviceability (RAS)

An embodiment of an electronic apparatus may comprise one or more substrates, and a controller coupled to the one or more substrates, the controller including circuitry to provide management of a connected hardware subsystem with respect to one or more of reliability, availability and serviceability, and coordinate the management of the connected hardware subsystem with respect to one or more of reliability, availability and serviceability between the connected hardware subsystem and a host. Other embodiments are disclosed and claimed.

Cloud scale server reliability management

Techniques and mechanisms for providing performance monitoring information. In an embodiment, a performance monitor circuit receives a communication which indicates a format comprising multiple fields which are each to store a respective count of monitored events. A programming of the performance monitor circuit, based on the communication, designates first bits and second bits of the register to provide, respectively, a first first field and a second field according to the format. Performance monitoring subsequent to the programming successively tallies a first count of first events which occur during a first period of time, and a second count of second events which occur during a second period of time. In another embodiment, performance monitoring results in the register concurrently storing both the first count and the second count.

Device, system, and method to concurrently store multiple PMON counts in a single register

Upon occurrence of multiple errors in a central processing unit (CPU) package, data indicating the errors is stored in machine check (MC) banks. A timestamp corresponding to each error is stored, the timestamp indicating a time of occurrence for each error. A machine check exception (MCE) handler is generated to address the errors based on the timestamps. The timestamps can be stored in the MC banks or in a utility box (U-box). The MCE handler can then address the errors based on order of occurrence, for example by determining that the first error in time causes the remaining error. The MCE can isolate hardware/software associated with the first error to recover from a failure. The MCE can report only the first error to the operating system (OS) or other error management software/hardware. The U-Box may also convert the timestamps into real time to support user debugging.

Serializing machine check exceptions for predictive failure analysis

In an embodiment, a processor for reverse translation includes a plurality of processing engines (PEs) to execute threads and a reverse translation circuit. The reverse translation circuit is to: determine a target module address of a corrupt portion of a memory module; determine a plurality of system physical address (SPA) addresses associated with the memory module; and for each SPA address in the plurality of SPA addresses, translate the SPA address into a translated module address, and in response to a determination that the translated module address matches the target module address, log the SPA address as a result of a reverse translation of the target module address. Other embodiments are described and claimed.

Hardware unit for reverse translation in a processor

An apparatus and method for machine check bank reporting in a processor. For example, one embodiment includes a processor comprising: one or more cores to execute instructions and process data; a plurality of machine check architecture banks to store errors detected during execution of the instructions; error monitoring circuitry to detect the errors and responsively update the MCA banks; and a first error register (FERR) into which a first error vector is to be stored to identify an MCA bank containing a first error in an error sequence, the error monitoring circuitry to update the first error vector responsive to detecting the first error; and one or more next error registers (NERRs) to store one or more error vectors to one or more other MCA banks containing subsequent errors occurring after the first error.

Apparatus and method for vectored machine check bank reporting

In one embodiment, a processor includes a plurality of cores including a first core to be reserved for execution in a protected domain, the first core to be hidden from an operating system. The processor may further include a filter coupled to the plurality of cores, where the filter includes a plurality of fields each associated with one of the plurality of cores to indicate whether an interrupt of the protected domain is to be directed to the corresponding core. Other embodiments are described and claimed.

Providing dedicated resources for a system management mode of a processor

Methods and apparatuses relating to a common architectural state presentation for a processor having cores of different types are described. In one embodiment, a processor includes a first core, a second core, wherein the first core comprises a unique architectural state and a common architectural state with the second core, and circuitry to migrate a thread from said first core to said second core, said circuitry to migrate the common architectural state from the first core to the second core, and migrate the unique architectural state to a storage external from the second core.

Common architectural state presentation for processor having processing cores of different types

In one embodiment, the present invention includes a method for controlling redundant execution such that if an exceptional event occurs, the redundant execution is stopped, non-redundant execution is performed in one of the threads until the exceptional event has been-resolved, after which a state of the threads is synchronized, and redundant execution is continued. Other embodiments are described and claimed.

Controlling non-redundant execution in a redundant multithreading (RMT) processor

A method is described that includes deciding to migrate a thread from a first processing core to a second processing core. The method also includes automatically in hardware migrating first context of the thread of the first processing core whose register definition is also found on the second processing core to the second processing core. The method also includes automatically in hardware migrating second context of the thread of the first processing core whose register definition is not found on the second processing core to a first storage location external to the second processing core. The message also includes automatically in hardware migrating third context of the thread from a second storage location external to the second processing core to register definition found on the second processing core but not found on the first processing core.

Common architecture state presentation for processor having processing cores of different types

Growing community of inventors

Beaverton, OR, United States of America

John G Holm