International Business Machines Corporation

Paul Frank Lecocq

James Stephen Fields, Jr

Kevin Franklin Reick

Michael Stephen Floyd

Thomas Pflueger

Scott Barnett Swaney

Timothy Michael Skergan

John Andrew Schumann

Brian Chan Monwai

Lakshminarayana Baba Arimilli

Benjiman L Goodman

Bernard Charles Drerup

Hanhong Xue

Praveen S Reddy

Shakti Kapoor

Manoj Dusanapudi

Debapriya Chatterjee

Larry Scott Leitner

Karen E Yokum

Bryant Cockcroft

Ra'ed Mohammad Al-Omari

Tharunachalam Pindicura

Paul Umbarger

Jeffrey W Kellington

Lance G Hehenberger

Kevin Barnett

Jamory Hawkins

Lawrence Leitner

Omesh Bajaj

Jody B Joyner

Michael Stephen Floys

A method of checking for a stall condition in a processor is disclosed, the method including inserting an inline instruction sequence into a thread, the inline instruction sequence configured to read the result from a timing register during processing of a first instruction and store the result in a first general purpose register, wherein the timing register functions as a timer for the processor; and read the results from the timing register during processing of a second instruction and store the results in a second general purpose register, wherein the second instruction is the next consecutive instruction after the first instruction. The inline thread sequence may be inserted in sequence with the thread and further configured to compare the difference between the result in the first and second general purpose register to a programmable threshold.

Method and system for detection of thread stall

Embodiments herein provide a testing apparatus (whether physical or simulated) for testing a non-core MMU in a processor chip. Unlike core MMUs, non-core MMUs may be located in a part of the processor chip outside of the processing cores in the chip. Instead of being used to perform address translation requests sent by the processing core, the non-core MMUs may be used by other hardware modules in the processor chip such as compression engines, crypto engines, accelerators, etc. In one embodiment, the testing apparatus includes a MMU testor that transmits the translation requests to the non-core MMU which tests its functionality. Using the data provided in the translation requests, the non-core MMU performs virtual to physical address translations. The non-core MMU transmits the results of these translations to the MMU testor which compares these results to expected results to identify any design flaws in the non-core MMU.

Testing a non-core MMU

A mechanism is provided in a collective acceleration unit for performing a collective operation to distribute or collect data among a plurality of participant nodes. The mechanism receives an input collective packet for a collective operation from a neighbor node within a collective tree. The input collective packet comprises a tree identifier and an input data field and wherein the collective tree comprises a plurality of sub trees. The mechanism maps the tree identifier to an index within the collective acceleration unit. The index identifies a portion of resources within the collective acceleration unit and is associated with a set of neighbor nodes in a given sub tree within the collective tree. For each neighbor node the collective acceleration unit stores destination information. The collective acceleration unit performs an operation on the input data field using the portion of resources to effect the collective operation.

Collective acceleration unit tree structure

A mechanism is provided for collective acceleration unit tree flow control forms a logical tree (sub-network) among those processors and transfers 'collective' packets on this tree. The system supports many collective trees, and each collective acceleration unit (CAU) includes resources to support a subset of the trees. Each CAU has limited buffer space, and the connection between two CAUs is not completely reliable. Therefore, to address the challenge of collective packets traversing on the tree without colliding with each other for buffer space and guaranteeing the end-to-end packet delivery, each CAU in the system effectively flow controls the packets, detects packet loss, and retransmits lost packets.

Collective acceleration unit tree flow control and retransmit

In a multiprocessor environment, by executing cache-inhibited reads or writes to registers, a scan communication is used to rapidly access registers inside and outside a chip originating the command. Cumbersome locking of the memory location may be thus avoided. Setting of busy latches at the outset virtually eliminates the chance of collisions, and status bits are set to inform the requesting core processor that a command is done and free of error, if that is the case.

Apparatus for operating cache-inhibited memory mapped commands to access registers

Illustrative embodiments disclose a data processing system providing low-level hardware access to in-band and out-of-band firmware. The data processing system includes a plurality of chips that includes at least one processor chip and a plurality of support chips. At least one processor chip includes a field replaceable unit support interface master that uses a field replaceable unit support interface serial transmission protocol to communicate with the plurality of support chips. Each one of the plurality of support chips includes a field replaceable unit support interface slave in, with ones of the plurality of chips that include a processor also include the field replaceable unit support interface master, and ones of the plurality of chips that do not include the processor include only the field replaceable unit support interface slave. Only the field replaceable unit support interface master possesses conversion logic.

Providing low-level hardware access to in-band and out-of-band firmware

A method, apparatus, and computer program product are disclosed in a data processing system for synchronizing the triggering of multiple hardware trace facilities using an existing bus. The multiple hardware trace facilities include a first hardware trace facility and a second hardware trace facility. The data processing system includes a first processor that includes the first hardware trace facility and first processing units that are coupled together utilizing the system bus, and a second processor that includes the second hardware trace facility and second processing units that are coupled together utilizing the system bus. Information is transmitted among the first and second processing units utilizing the system bus when the processors are in a normal, non-tracing mode, where the information is formatted according to a standard system bus protocol. Trigger events are transmitted to the hardware trace facilities utilizing the same standard system bus, where the trigger events are also formatted according to the standard system bus protocol.

Synchronizing triggering of multiple hardware trace facilities using an existing system bus

A method and apparatus are provided for a support interface for memory-mapped resources. A support processor sends a sequence of commands over and FSI interface to a memory-mapped support interface on a processor chip. The memory-mapped support interface updates memory, memory-mapped registers or memory-mapped resources. The interface uses fabric packet generation logic to generate a single command packet in a protocol for the coherency fabric which consists of an address, command and/or data. Fabric commands are converted to FSI protocol and forwarded to attached support chips to access the memory-mapped resource, and responses from the support chips are converted back to fabric response packets. Fabric snoop logic monitors the coherency fabric and decodes responses for packets previously sent by fabric packet generation logic. The fabric snoop logic updates status register and/or writes response data to a read data register. The system also reports any errors that are encountered.

Method for indirect access to a support interface for memory-mapped resources to reduce system connectivity from out-of-band support processor

A method, apparatus, and program for systematically testing the functionality of all connections in a multi-tiered bus system that connects a large number of processors. Each bus controller is instructed to send a test version of a snoop request to all of the other processors and to wait for the replies. If a connection is bad, the port associated with that connection will time out. Detection of a time-out will cause the initialization process to be halted until the problem can be isolated and resolved.

Apparatus to use fabric initialization to test functionality of all inter-chip paths between processors in system

In-band firmware executes instructions which cause commands to be sent on a coherency fabric. Fabric snoop logic monitors the coherency fabric for command packets that target a resource in one of the support chips attached via an FSI link. Conversion logic converts the information from the fabric packet into an FSI protocol. An FSI command is transmitted via the FSI transmit link to an FSI slave of the intended support chip. An FSI receive link receives response data from the FSI slave of the intended support chip. Conversion logic converts the information from the support chip received via the FSI receive link into the fabric protocol. Response packet generation logic generates the fabric response packet and returns it on the coherency fabric. An identical FSI link between a support processor and support chips allows direct access to the same resources on the support chips by out-of-band firmware.

Method for providing low-level hardware access to in-band and out-of-band firmware

A method, apparatus, and computer instructions are provided by the present invention to automatically recover from a failed node concurrent maintenance operation. A control logic is provided to send a first test command to processors of a new node. If the first test command is successful, a second test command is sent to all processors or to the remaining nodes if nodes are removed. If the second command is successful, system operation is resumed with the newly configured topology with either nodes added or removed. If the response is incorrect or a timeout has occurred, the control logic restores values to the current mode register and sends a third test command to check for an error. A fatal system attention is sent to a service processor or system software if an error is encountered. If no error, system operation is resumed with previously configured topology.

Method and apparatus for automatic recovery from a failed node concurrent maintenance operation

Method to use fabric initialization to test functionality of all inter-chip paths between processors in system

Growing community of inventors

Cedar Park, TX, United States of America

Paul Frank Lecocq