International Business Machines Corporation

Intel Corporation

Tahoe Research, Ltd.

Intel-ne, Inc.

Vadim Makhervaks

Giora Biran

Renato John Recio

Leah Shalev

Richard Louis Arndt

Jaya Srikrishnan

Zorik Machulsky

Harvey Gene Kiel

Patrick Allen Buckland

Eliezer Tamir

Ben-Zion Friedman

Julian Satran

Theodore L Willke

Kenneth G Keels

Phil C Cayton

Ilan Shimony

Brian S Hausauer

Robert O Sharp

John Lewis Hufferd

Donald Wood

Ali S Oztaskin

Shaul Yohai Yifrach

Kalman Zvi Meth

Steen K Larsen

Alexander Mesh

Tal Sostheim

Mark Epshtein

Eric Jon Schneider

Leah Shaley

Examples are disclosed for access to a storage device maintained at a server. In some examples, a network input/output device coupled to the server may allocate, in a memory of the server, a buffer, a doorbell, and a queue pair accessible to a client remote to the server. For these examples, the network input/output device may assign an Non-Volatile Memory Express (NVMe) namespace context to the client. For these examples, indications of the allocated buffer, doorbell, queue pair, and namespace context may be transmitted to the client. Other examples are described and claimed.

Techniques for command validation for access to a storage device by a remote client

Examples are disclosed for command validation for access to a storage device maintained at a server. In some examples, a network input/output device coupled to the server may receive the command from a client remote to the server. For these examples, elements or modules of the network input/output device may be capable of validating the command and reporting the status of the received command to the client. Other examples are described and claimed.

Generally, this disclosure relates to a shared send queue in a networked system. A method, apparatus and system are configured to support a plurality of reliable communication channels using a shared send queue. The reliable communication channels are configured to carry messages from a host to a plurality of destinations and to ensure completed order of messages is related to a transmission order.

Shared send queue

Methods, apparatus and systems for reducing usage of Doorbell Rings in connection with RDMA operations. A portion of system memory is employed as a Memory-Mapped Input/Output (MMIO) address space configured to be accessed via a hardware networking device. A Send Queue (SQ) is stored in MMIO and is used to facilitate processing of Work Requests (WRs) that are written to SQ entries by software and read from the SQ via the hardware networking device. The software and logic in the hardware networking device employ pointers identifying locations in the SQ corresponding to a next write WR entry slot and last read WR entry slot that are implemented to enable WRs to be written to and read from the SQ during ongoing operations under which the SQ is not emptied such that doorbell rings to notify the hardware networking device that new WRs have been written to the SQ are not required.

Methods, apparatus and systems for facilitating RDMA operations with reduced doorbell rings

An embodiment may include circuitry that may write a message from a system memory in a host to a memory space in an input/output (I/O) controller in the host. A host operating system may reside, at least in part, in the system memory. The message may include both data and at least one descriptor associated with the data. The data may be included in the at least one descriptor. The circuitry also may signal the I/O controller that the writing has occurred.

Writing message to controller memory space

A method of offloading, from a host data processing unit (), the generation of data corruption-detection digests for iSCSI PDUs to be transmitted as TCP segments over respective TCP connections. An iSCSI layer processing software () executed by the host data processing unit provides a command descriptor list () containing command descriptors adapted to identify portions of at least one iSCSI PDU to be transmitted, and data corruption-detection digest descriptors (CRC DESC(PDUa); CRC DESC(PDUb)), each one associated with a respective PDU data corruption-detection digest. An iSCSI processing offload engine () transmits the iSCSI PDU over the respective TCP connection, based on the descriptors in the command descriptor list; during the transmission, the iSCSI PDU data corruption-detection digest are calculated, and the calculated data corruption-detection digest is saved in the corresponding data corruption-detection digest descriptor in the command descriptor list.

Method of offloading iSCSI PDU corruption-detection digest generation from a host processing unit, and related iSCSI offload engine

A method and system for completion coalescing by a Transmission Control Protocol (TCP) receiver (e.g., in a TCP/IP offload engine (TOE)). The method comprises: processing inbound TCP segments; and performing completion processing of received TCP ACKS and/or RDMA Read Requests independently of the processing of the inbound TCP segments.

Completion coalescing by TCP receiver

A method for receiving data in a network acceleration architecture for use with TCP (transport control protocol), iSCSI (Internet Small Computer System Interface) and RDMA (Remote Direct Memory Access) over TCP, including providing a hardware acceleration engine, called a streamer, adapted for communication with and processing data from a consumer application in a system that supports TCP, iSCSI and RDMA over TCP, providing a software protocol processor adapted for carrying out TCP implementation, the software control processor being called a TCE (TCP Control Engine), and providing an asynchronous dual-queue interface for exchanging information between the streamer and TCE, wherein the streamer and TCE are adapted to operate asynchronously and independently of one another.

Asynchronous dual-queue interface for use in network acceleration architecture

A method, computer program product, and distributed data processing system that enables host software or firmware to allocate virtual resources to one or more system images from a single physical I/O adapter, such as a PCI, PCI-X, or PCI-E adapter, is provided. Adapter resource groups are assigned to respective system images. An adapter resource group is exclusively available to the system image to which the adapter resource group assignment was made. Assignment of adapter resource groups may be made per a relative resource assignment or an absolute resource assignment. In another embodiment, adapter resource groups are assigned to system images on a first come, first served basis.

System and method for virtual adapter resource allocation matrix that defines the amount of resources of a physical I/O adapter

An RNIC implementation that performs direct data placement to memory where all segments of a particular connection are aligned, or moves data through reassembly buffers where all segments of a particular connection are non-aligned. The type of connection that cuts-through without accessing the reassembly buffers is referred to as a 'Fast' connection because it is highly likely to be aligned, while the other type is referred to as a 'Slow' connection. When a consumer establishes a connection, it specifies a connection type. The connection type can change from Fast to Slow and back. The invention reduces memory bandwidth, latency, error recovery using TCP retransmit and provides for a &#x201c;graceful recovery&#x201d; from an empty receive queue. The implementation also may conduct CRC validation for a majority of inbound DDP segments in the Fast connection before sending a TCP acknowledgement (Ack) confirming segment reception.

Data transfer error checking

A method, system, and computer program product for sharing adapter resources among multiple operating system instances. The present invention provides a mechanism for dynamically allocating virtualized I/O adapter resources. The present invention separates the operation of adapter resource allocation from adapter resource management. Protection attributes within the adapter resource context are used to allow the adapter to enforce access control over the adapter resources. The hypervisor allocates an available adapter resource to a given partition. The adapter is notified of the allocation, and the adapter updates its internal structure to reflect the allocation. The hypervisor may revoke ownership of and reassign adapter resources to another OS instance. In this manner, the allocation described above allows for the simple reassignment of resources from one partition to another.

Association of memory access through protection attributes that are associated to an access control level on a PCI adapter that supports virtualization

A method for receiving data in a network acceleration architecture for use with TCP (transport control protocol), iSCSI (Internet Small Computer System Interface) and RDMA (Remote Direct Memory Access) over TCP, including providing a hardware acceleration engine, called a streamer, adapted for communication with and processing data from a consumer application in a system that supports TCP, iSCSI and RDMA over TCP, providing a software protocol processor adapted for carrying out TCP implementation, the software control processor being called a TCE (TCP Control Engine), wherein the streamer and the TCE are adapted to operate asynchronously and independently of one another, and receiving an inbound TCP segment with the streamer.

Growing community of inventors

Austin, TX, United States of America

Vadim Makhervaks