摘要:
Hot plug modules comprising processors, memory, and/or I/O hubs may be added to and removed from a running computing device without rebooting the running computing device. The hot plug modules and computing device comprise hot plug interfaces that support hot plug addition and hot plug removal of the hot plug modules.
摘要:
An apparatus and method for enumeration of processors during hot-plug of a compute node are described. The method includes the enumeration, in response to a hot-plug reset, of one or more processors. The enumeration is provided to a system architecture operating system in which a compute node is hot-plugged. Once enumeration is complete, the compute node is started in response to an operating system activation request. Accordingly, once device enumeration, as well as resource enumeration are complete, the one or more processors of the processor memory node are activated, such that the operating system may begin utilizing the processors of the hot-plugged compute node.
摘要:
An apparatus and method for enumeration of processors during hot-plug of a compute node are described. The method includes the enumeration, in response to a hot-plug reset, of one or more processors. The enumeration is provided to a system architecture operating system in which a compute node is hot-plugged. Once enumeration is complete, the compute node is started in response to an operating system activation request. Accordingly, once device enumeration, as well as resource enumeration are complete, the one or more processors of the processor memory node are activated, such that the operating system may begin utilizing the processors of the hot-plugged compute node.
摘要:
One aspect of the invention relates to a method for supporting hibernation despite the presence of hot-plugged nodes and non-deterministic boot operations. The method comprises invoking a management interrupt in response to a Hibernate request. The management interrupt is used to obtain and store platform configuration information into a non-volatile storage location. The platform configuration information includes data to indicate whether a next boot sequence for a platform occurs as a deterministic boot sequence or a non-deterministic boot sequence as well as a boot node identifier and a listing of an order in which processors of the platform are initialized.
摘要:
One aspect of the invention relates to creation of a container object being part of software that is stored in platform readable medium and executed by a processor within a platform. The container comprises (i) a hardware identification object to identify to an operating system of the platform that a type of device represented by the container object is a node and (ii) a plurality of component objects to identify constituent components of the node. Another aspect of the invention is the distribution of BIOS to handle initiation of components of a substrate in response to hot-plug addition of that substrate.
摘要:
An apparatus and method for enumeration of processors during hot-plug of a compute node are described. The method includes the enumeration, in response to a hot-plug reset, of one or more processors. The enumeration is provided to a system architecture operating system in which a compute node is hot-plugged. Once enumeration is complete, the compute node is started in response to an operating system activation request. Accordingly, once device enumeration, as well as resource enumeration are complete, the one or more processors of the processor memory node are activated, such that the operating system may begin utilizing the processors of the hot-plugged compute node.
摘要:
Error detection methods, systems and medium are provided. The error detection method may comprise processing error conditions associated with transactions in a manner that may enable error source identification. The system may comprise a plurality of nodes of components. The nodes may include storage elements to record an error condition indicative of whether a component provided an indication of detecting an error in response to processing the transaction.
摘要:
A system that may optionally be partitioned into multiple domains is disclosed. Each domain is capable of independently powering on, executing a firmware program, and loading an operating system, including a legacy operating system, as well as running an application program that is distinct from programs running on another domain. Interrupts, including boot interrupts, reset handlers, and inter-chassis communications are initialized differently, depending on whether the system is to be partitioned or not. The cost of redundant hardware and/or firmware is substantially avoided, yet the system fully supports multiple domains.
摘要:
A small computer system interface (SCSI) controller circuit, connected between a host computer bus and a SCSI bus, includes a storage buffer shared between command and data signals provided by a host computer to a target SCSI peripheral device. The shared buffer includes a bidirectional FIFO which is used for intermediate storage of command blocks between transferred the host and the target device as well as data blocks. In the case of data transfers from the host to the target device, a command block is written to the bidirectional FIFO followed by as much of the data block as will fit in the remaining FIFO space. After the target device has read the command block, it accesses the SCSI bus and begins the transfer of data from the FIFO. In the case of data transfers from the target device to the host, a data transfer command block is first written to the FIFO by the host. The command block is read by the target device which then provides the requested data to the FIFO over the SCSI bus. The host system receives the requested data from the same shared storage buffer to which the command block was written.