Patent Application
20130159572
The present disclosure relates generally to computer systems, and more particularly, to managing virtual functions that are hosted by a virtualized input/output (I/O) adapter.
Single Root I/O Virtualization (SR-IOV) is a specification that allows a Peripheral Component Interconnect Express (PCIe) device to appear to be multiple separate physical PCIe devices. SR-IOV enables a virtualization intermediary (VI), such as a hypervisor or virtual input/output (I/O) server operating system, to configure an I/O adapter into a number of virtual functions (VFs). The virtual functions may be assigned to different operating system images (OSIs), or logical partitions (LPARs).
The virtual functions belong to a PCI hierarchy and are of a device type that may be undefined in operating system and system firmware. Configuration of the virtual functions may require significant administrator man-hours and system downtime. Association and management of the virtual functions with a PCI adapter or slot location that is subject to PCI adapter maintenance and administrative operations, such as adapter hot plug and dynamic assignment to or from logical partitions, may be undefined in operating systems and system firmware.
In a particular embodiment, a computer implemented method of managing an adapter includes determining an adapter is single root input/output virtualization (SR-IOV) capable; determining that an operating system is capable of using the adapter in SR-IOV mode, and configuring the adapter in SR-IOV mode by generating an SR-IOV function associated with the adapter. The SR-IOV function may be used to emulate an aspect of a Peripheral Component Interconnect (PCI) multi-function device at the operating system.
In another particular embodiment, an apparatus includes an adapter, a processor, and a memory storing program code, the program code executable by the processor to determine the adapter is SR-IOV capable, to determine that an operating system is capable of using the adapter in SR-IOV mode, to configure the adapter in SR-IOV mode by generating an SR-IOV function associated with the adapter, and to present the SR-IOV function to the operating system in such a manner as to emulate an aspect of a PCI multi-function device.
In another particular embodiment, a computer program product includes a computer usable medium having computer usable program code embodied therewith. The computer usable program code may be executable by a processor to determine an adapter is SR-IOV capable; to determine that an operating system is capable of using the adapter in SR-IOV mode; to configure the adapter in SR-IOV mode by generating an SR-IOV function associated with the adapter; and to present the SR-IOV function to the operating system in such a manner as to emulate an aspect of a PCI multi-function device.
These and other advantages and features that characterize embodiments of the disclosure are set forth in the claims listed below. However, for a better understanding of the disclosure, and of the advantages and objectives attained through its use, reference should be made to the drawings and to the accompanying descriptive matter in which there are described exemplary embodiments of the disclosure.
In a virtualized computer system, a hardware input/output (I/O) adapter may be capable of providing virtual functions to multiple logical partitions. For example, the hardware I/O adapter may be a single root input/output virtualized (SR-IOV) adapter or a multiple root input/output virtualized (MR-IOV) adapter. A virtualization intermediary (VI), such as a hypervisor, a hosting operating system, or other firmware or software entity within a virtualized computer system acting as a virtualization management agent, may manage the execution of the multiple logical partitions and assign one or more of the virtual functions to particular logical partitions to enable the logical partitions to perform I/O operations.
Each virtual function may have an associated configuration space that is located at a memory of the hardware I/O adapter. The configuration space may include a read-only portion and a read-write portion. For example, the read-only portion may provide information associated with the virtual function, such as a device identifier and a vendor identifier, and information associated with the hardware I/O adapter, such as a number of ports of the hardware I/O adapter and an arrangement of the ports. The read-write portion may include parameters that can be configured (e.g., by a logical partition or by an application executing in the logical partition), such as enabling/disabling memory-mapped I/O (MMIO), enabling/disabling direct memory access (DMA), setting a maximum link speed, enabling/disabling advanced error handling, setting another virtual function parameter or any combination thereof. In a particular embodiment, the configuration space may include one or more registers, such as read-only registers and read-write registers.
The virtualization intermediary may provide an access mechanism to enable a logical partition to access the configuration space that is associated with the virtual function that is assigned to the logical partition. The access mechanism provided by the virtualization intermediary may be a high-level access mechanism that uses lower-level access mechanisms to access the configuration space of each virtual function. For example, the access mechanism provided by the virtualization intermediary may call a configuration space access mechanism of a root complex, an adapter provided configuration mechanism, another access mechanism, or any combination thereof.
A particular embodiment facilitates the implementation and application of Peripheral Component Interconnect Express (PCIe) Single Root I/O Virtualization (SR-IOV) adapter by presenting the SR-IOV adapter and associated virtual functions to system components in a manner that avoids change to the system components. Illustrative such system components may be outside of the virtualization intermediary, such as system or platform management systems, operating systems, system firmware, and I/O device drivers. The virtualization intermediary may detect and initialize physical functions and virtual functions correctly and appropriately.
An embodiment enables a virtualization intermediary to present and operate SR-IOV adapters and virtual functions within system management, operating system, and system firmware components in a manner that substantially conforms to that of non-SR-IOV PCI adapters. SR-IOV technology may be adapted to operating systems and firmware that already support PCI-e adapters in an automatic and inexpensive manner.
An SR-IOV adapter may be virtualized to be shared by multiple OSIs/LPARs within a logically partitioned environment, or may be assigned to one OSI/LPAR as a dedicated adapter. In the shared case, a virtualization intermediary may configure the adapter in SR-IOV-enabled mode and make individual virtual functions available for assignment to an individual operating system or logical partition.
In the case of a dedicated (e.g., non-shared) operating system, the operating system may desire to use the adapter in legacy mode. In legacy mode, the SR-IOV capabilities may not be enabled or used. Another legacy mode scenario may include an adapter enabled for SR-IOV and an operating system that implements a single device driver for the virtual function (or for each virtual function of a plurality of multiple functions). The device driver arrangement may avoid development of a more complex device driver that encompasses both virtual function and adapter physical and management functions.
Where a platform management administers logical partitions and shares SR-IOV adapters as individual virtual functions, an SR-IOV-enabled adapter may be dedicated to a single operating system or logical partition by assigning all of the adapter virtual functions to the operating system or logical partition. This dedicated assignment may allow the operating system or logical partition to provide a virtual function device driver and may delegate the larger adapter configuration and management or service functions to the platform management and virtualization intermediary.
A computing system that is not under such a partition management agent (i.e., a non-managed system) may inherit ownership of all of the PCI devices. The operating system and system firmware may perform all adapter configuration and management operations. The operating system may provide device driver resources to manage the adapter, whether virtualized or not. Further, an operating system may desire to use a non-shared adapter in a legacy mode, i.e., without SR-IOV being enabled. Other operating system instances running on the same logically partitioned system may desire to use the adapter in a non-shared, virtualized mode (e.g., SR-IOV-enabled) when ownership of the adapter is transferred to the operating system or logical partition. An embodiment may enable an SR-IOV adapter to be assigned to, or on a non-managed system to default to be owned by, an operating system or logical partition as wholly owned by that operating system or logical partition in either a virtualized or non-virtualized mode. According to an embodiment, the virtualization intermediary automatically and selectively translates between an SR-IOV function to an emulated PCI-standard function to enable control by the operating system.
Referring to
More particularly, the system 100 may include a hardware server 102 that is managed by the virtualization intermediary 110, such as a hypervisor. The hardware server 102 may include hardware resources, such as a first board 104, a second board 105, and a third board 106. While three boards are illustrated in
The virtualization intermediary 110 may create and manage logical partitions, such as virtual servers 112, 113, 143. A logical partition may be a subset of the resources of the hardware server 102 that is virtualized as a separate virtual server. Each of the virtual servers 112, 113, 143 may have its own set of virtual resources, similar to a physical server. For example, the first virtual server 112 may include virtual processors 120, virtual memory 122, and virtual I/O adapters 124. The second virtual server 113 may include virtual processors 121, virtual memory 123, and virtual I/O adapters 125. The second virtual server 143 may include virtual processors 143, virtual memory 145, and virtual I/O adapters 146. The virtualization intermediary 110 may map the hardware of the hardware server 102 to the virtual servers 112, 113, 143. For example, the processors 130-132 may be mapped to the virtual processors 120, 121; the memory 133-135 may be mapped to the virtual memory 122, 123, and the I/O adapters 136-138 may be mapped to the virtual I/O adapters 124-125. Each of the virtual servers 112, 113, 143 may include a physical I/O adapter 147-149. The physical I/O adapters 147-149 may correspond to I/O adapters 136-138. The virtualization intermediary 110 may manage the selection of portions of the hardware server 102 and their temporary assignment to portions of the virtual servers 112, 113, including assignment of one or a plurality of physical adapters 136-138 to one virtual server.
The virtualization intermediary 110 may provide a configuration mechanism 180 to configure and manage a PCI hierarchy that includes a PCI host bridge and virtual functions. SR-IOV virtual functions may be presented to an operating system 114, 115 as non-IOV functions of a PCI multi-function device. According to another embodiment, the configuration mechanism 180 may not configure the adapters 136-138 in SR-IOV mode, and may instead allow the operating system 114, 115 to operate the adapters 136-138 as legacy PCI adapters.
Referring to
The physical functions 230, 231 may include PCI functions that support single root I/O virtualization capabilities. Each of the virtual functions 232-235 may be associated with one of the physical functions 230, 231 and may share one or more physical resources of the hardware I/O adapter 202.
Software modules, such as a physical function (PF) manager 220, may assist the virtualization intermediary in managing the physical functions 230, 231 and the virtual functions 232-235. For example, a user may specify a particular configuration and the PF manager 220 may configure the virtual functions 232-235 from the physical functions 230, 231 accordingly.
In operation, the PF manager 220 may enable the first virtual function instances 232-234 from the first physical function 230. The PF manager 220 may enable the second virtual function 235 from the second physical function 231. The virtual functions 232-235 may be enabled based on a user provided configuration. Each of the logical partitions 206-208 may execute an operating system (not shown) and client applications (not shown). The client applications that execute at the logical partitions 206-208 may perform virtual input/output operations. For example, a first client application executing at the first logical partition 206 may include first client virtual I/O 226, and a second client application executing at the first logical partition 206 may include a second client virtual I/O 227. The first client virtual I/O 226 may access the first instance of the first virtual function 232. The second client virtual I/O 227 may access the second virtual function 235. A third client virtual I/O 228 executing at the second logical partition 207 may access the second instance of the first virtual function 233. An Nth client virtual I/O 229 executing at the Nth logical partition 208 may access the Nth instance of the first virtual function 233.
The virtualization intermediary 204 may assign the first instance of the first virtual function 232 and the first instance of the second virtual function 235 to the first logical partition 206. The virtualization intermediary 204 may provide the first logical partition 206 with two tokens (not shown), such as a first token and a second token, to enable the first logical partition 206 to access the virtual functions 232 and 235. The token may include a group identifier that identifies a physical slot location of the hardware I/O adapter 202 that hosts the virtual functions 232 and 235. The hardware I/O adapter 202 that hosts the virtual functions 232 and 235 may be moved from a first physical slot location to a second physical slot location. After the move, the virtualization intermediary 202 may associate the group identifier with the second physical slot location to enable the virtual functions 232 and 235 to be provided to the first logical partition 206.
It will be appreciated by one skilled in the art that the present invention is equally suited to embodiments that do not utilize a virtual function (VF) manager and client virtual I/O to enable a logical partition to access a virtual function, and instead enable a device driver within a logical partition to directly manage the virtual function. The virtualization intermediary 204 may provide a configuration mechanism 280 to selectively and automatically associate SR-IOV virtual functions with non-SR-IOV functions, such as PCI-standard functions virtual functions.
Referring to
The service processor 314 may be a microcontroller that is embedded in a hardware server (e.g., the hardware server 102 of
The virtualization intermediary 304 may enable hardware devices, such as the hardware I/O adapter 302, to be logically divided into virtual resources and accessed by one or more logical partitions (e.g., the N logical partitions 206-208 of
The I/O hub 306 may be a hardware device (e.g., a microchip on a computer motherboard) that is under the control of the virtualization intermediary 304. The I/O hub 306 may enable the virtualization intermediary 304 to control I/O devices, such as the hardware I/O adapter 302.
The processors 308 may include one more processors, such as central processing units (CPUs), digital signal processors (DSPs), other types of processors, or any combination thereof. One or more of the processors 308 may be configured in a symmetric multiprocessor (SMP) configuration.
The memory 310 may include various types of memory storage devices, such as random access memory (RAM) and disk storage devices. The memory 310 may be used to store and retrieve various types of data. For example, the memory 310 may be used to store and to retrieve operational instructions that are executable by one or more of the processors 308.
The operating system 312 may execute within the logical partition 311. The virtual I/O of client applications (e.g., the client virtual I/Os 226-229 of
The PF manager 318 may include an adapter abstraction layer 320 and an adapter driver 322. The adapter abstraction layer 320 may include a generic abstraction to enable configuration of physical functions and virtual functions of the hardware I/O adapter 302. The adapter driver 322 may be specific to each particular model of hardware adapter. The adapter driver 322 may be provided by a manufacturer of the hardware I/O adapter 302.
The hardware I/O adapter 302 may include physical functions and ports, such as a first physical function 324, a second physical function 325, a first port 326, and a second port 327. The PF manager 318 may configure virtual functions based on the physical functions 324, 325 and associate the virtual functions with one or more of the ports 326, 327 of the hardware I/O adapter 302. For example, the PF manager 318 may configure the first physical function 324 to host multiple instances of a first virtual function, such as the first instance of the first virtual function 330 and the Mth instance of the first virtual function 331, where M is greater than 1. The instances of the first virtual function 330, 331 may be associated with the second port 327. The PF manager 318 may configure the second physical function 325 to host multiple instances of a second virtual function, such as the first instance of the second virtual function 332 and the Pth instance of the second virtual function 333, where P is greater than 1. The instances of the second virtual function 332, 333 may be associated with the first port 326. The PF manager 318 may configure multiple instances of an Nth virtual function, such as the first instance of the Nth virtual function 334 and the Qth instance of the Nth virtual function 335, where N is greater than 2, and Q is greater than 1. The instances of the Nth virtual function 334, 335 may be associated with the second port 327. The instances of the Nth virtual function 334, 335 may be hosted by a physical function, such as one of the first physical function 324, the second physical function 325, and another physical function (not shown).
Each virtual function (e.g., each of the virtual functions 330-335) may have an associated virtual function identifier (ID). For example, in the system 300, the first instance of the first virtual function 330 may have an associated identifier 340, the Mth instance of the first virtual function 331 may have an associated identifier 341, the first instance of the second virtual function 332 may have an associated identifier 342, the Pth instance of the second virtual function 333 may have an associated identifier 343, the first instance of the Nth virtual function 334 may have an associated identifier 344, and the Qth instance of the Nth virtual function 335 may have an associated identifier 345.
Each virtual function identifier may uniquely identify a particular virtual function that is hosted by the hardware I/O adapter 302. For example, when a message (not shown) is routed to a particular virtual function, the message may include the identifier associated with the particular virtual function. As another example, a token 313 may be provided to the operating system 312 to enable the operating system 312 to access one of the virtual functions 330-335 at the hardware I/O adapter 302. The token 313 may include a configuration mechanism 380 that is associated with the accessed virtual function. For example, the first instance of the first virtual function 330 may be assigned to the operating system 312. The token 313 may be provided to the operating system 312 to access the first instance of the first virtual function 330. The token 313 may include the virtual function identifier 380. The virtual function identifier 380 may comprise the identifier 340 that is associated with the first instance of the first virtual function 330.
The virtualization intermediary 304 may assign one or more of the virtual functions 330-335 to the logical partition 311. For each virtual function that is assigned to the logical partition 311, the virtualization intermediary 304 may provide the logical partition 206 with a token (not shown) to enable the logical partition 311 to access the virtual function. The token may include a group identifier that identifies a physical slot location of the hardware I/O adapter 302 that hosts the assigned virtual functions.
The virtualization intermediary 304 may provide an access mechanism 380 to enable logical partitions (e.g., the logical partition 311) to access configuration space associated with one or more of the virtual functions 330-335. The virtualization intermediary 304 may include an access mechanism 279 to enable logical partitions to access the PCI memory space, PCI DMA space, and interrupt ranges associate with virtual functions. In a legacy or SR-IOV model, the operating system device driver may access to the PCI memory that maps BARs, as well as access to a DMA window that the virtual function can use to DMA to memory, and a range of PCI interrupts the device driver can use to enable the virtual function to signal interrupts. This feature may provide for virtual functions in the same or a similar manner to that of legacy mode adapter function.
The PCI hardware 404 may include a PCI host bridge (PHB) 406, associated with a PCI-express root port (not shown). The PCI host bridge 406 may be coupled to the PCI adapter 408 via a PCI bus 410 representing a PCIe physical link connection (not shown) between the PCIe root port and a PCI adapter 408. The PCI adapter 408 may include a plurality of functions 412 associated, respectively, with a plurality of ports 414, as is common among a class of PCI adapters termed “PCI multi-function adapters”.
The operating system 402 may include a PCI device tree 416 and PCI device drivers 436 to operate on the functions 412 of the PCI adapter 608. The PCI device tree 416 may include a PCI host bridge node 418 and device nodes 420. The PCI host bridge node 418 may include a hot plug identifier (ID) 422, a dynamic logical partitioning (DLPAR) ID 424, and PCI bus properties 426. The device nodes 420 may include a configuration space 428, memory-mapped I/O (MIMIO) and direct memory access (DMA) space 430, a PCI read only memory base address register/read-only memory (ROMBAR/ROM) space 432, and an interrupt 434.
The PCI host bridge 406 may create an instance of the PCIe bus 410 connected to the PCI adapter 408. The function(s) 412 may be individually addressable in PCI configuration address space. For example, the function(s) 412 may have the same PCI device number and differing PCI function numbers (e.g., ranging from 0 to 7). Alternatively, the PCI adapter 408 may use PCI alternate routing ID (ARI) configuration addressing. Each function 412 may have a unique configuration function number ranging from 0 to 255 at an implied device number of 0. Each function 412 may be associated with a unique physical port 414 within the PCI adapter 408. The physical port 414 may create a connection to an external peripheral I/O interconnect, such as Ethernet, Fiber Channel, or another peripheral device interconnect.
The function(s) 412 may form a device driver programming interface by which the operating system 402 may utilize the PCI device drivers 436. The PCI host bridge nodes 418 may represent the PCI host bridge(s) 406, and the PCI device nodes 420 may represent each instance of the function(s) within the PCI adapter 408.
The PCI host bridge node 418 may include properties, or functions, descriptive of the PCI host bridge 406. Such properties may include characteristics of the PCIe bus 410 created by that PCI host bridge 406. The characteristics may be used by the operating system 402 to manage the PCI host bridge 406 and by the PCI device drivers 436 to perform PCI bus transactions. For example, the PCI host bridge node properties may include an identifier used for a hot plug domain 422 and an identifier for a DLPAR domain 424.
The operating system 402 may utilize the configuration firmware 438 to detect the presence of PCI devices, such as the function(s) 412. For each detected function 412, the configuration firmware 438 may generate a device node 420 associated with the PCI host bridge node 418 of the PCI device tree 416. The device node 420 may include functions, or properties, associated uniquely with the function 412 and used by the operating system 402 to identify the type and programming interface of the function 412. Illustrative such functions may relate to the configuration space 428 and the ROMBAR/ROM space 432. The properties may further be used by the device driver 436 to perform PCI bus transactions specific to that function 412, as well as to properties relating to the MIMIO and DMA space 430, the ROMBAR/ROM space 432, and the interrupts 434.
For each device node 420 within the PCI device tree 416, the operating system 402 may activate an instance of the device driver 436 to control the characteristics of the associated function 412. Data transfer operations may be performed between the operating system 402, the external interconnect, and devices accessed through the corresponding physical port 414.
The hot plug ID 422 of the PCI host bridge node 418 may be used to identify the PCI bus 410 physical connection point, or slot. The slot may be located between the PCI host bridge 406 and the PCI adapter 408. The operating system 402 may use the hot plug ID 422 when adapter a power-off or power-on operation is performed. The operating system 402 may be running and may be in control of the PCI host bridge 406 and the PCI bus 410.
To power-off the adapter 105, the operating system 402 may correlate a hot plug ID of a hot plug power-off/on operation with the hot plug ID 422 of the PCI host bridge node 418. As part of performing the power-off operation, the operating system 402 may first deactivate the device driver(s) 436. As discussed herein, the device driver(s) 436 may be associated with each device node 420, and each device node 420 may be associated with the PCI host bridge node 418.
When powering-on the PCI adapter 105, the configuration firmware 438 associated with the operating system 402 may interrogate each possible PCI configuration address of the PCI bus 410 to detect each function 412. The configuration firmware 438 may construct a device node 420 that is associated with the PCI host bridge node 418. The operating system 402 may create instances of the device driver(s) 436 that are associated with each device node 420. The device driver(s) 436 may control each of the associated functions 412.
The PCI host bridge(s) 406 may be connected individually to PCI slots. A slot may be a connection point at which the PCI adapters 408 may be added at a future time. The configuration firmware 438 may generate the PCI host bridge node(s) 418 of the PCI device tree 416 for each PCI host bridge 406. This generation may occur at an instance where the PCI host bridge 406 is connected to a PCI slot that is empty (e.g., does not have a PCI adapter 408 present).
The PCI adapter 408 may be transferable to different logical partitions using DLPAR. The PCI host bridge node 418 of the PCI device tree 416 may represent the domain of the functions 412 that are transferred, collectively, between logical partitions of the operating system 402. The virtualization intermediary 440 may act as a management agent of a system administrator to automatically associate elements of the PCI hardware 404 with an operating system(s) 402 comprising logical partitions.
The virtualization intermediary 440 may function as a system administrator for DLPAR by removing the PCI adapter 408 from the operating system 402. More specifically, the virtualization intermediary 440 may signal to the operating system 402 to initiate removal of a particular PCI adapter 408 having a DLPAR ID that references a matching DLPAR ID 424 of the operating system 402. As part of removing the PCI adapter 408 from the operating system PCI configuration, the operating system 402 may deactivate the PCI device driver(s) 436 associated with each device node 420 that is associated with that PCI host bridge node 418. The operating system 402 may release control of the PCI host bridge 406 and the PCI adapter 408 to the virtualization intermediary 440.
When adding a PCI adapter 408 to the PCI configuration of an executing operating system 402, the virtualization intermediary 440 may signal the operating system 402 to add the PCI host bridge node 418 to the PCI device tree 416. The virtual PCI host bridge node 418 may correspond to the physical PCI host bridge 406 and to the associated PCIe bus 410. The operating system 402 may invoke the configuration firmware 438 to detect the functions 412 of the PCI adapter 408. The configuration firmware 438 may update the PCI device tree 416 with a device node 420 corresponding to each detected function 412 that is associated with the PCI host bridge node 418 and/or PCIe bus 410. The operating system 402 may create an instance of the PCI device driver 436. The PCI device driver 436 may be associated with each device node 420 in order to control each of the associated functions 412.
The PCI hardware 504 may include a PCI host bridge (PHB) 506, associated with a PCIe root port (not shown). The PCI host bridge 506 may be coupled to the SR-IOV adapter 508 via a PCI bus 510, representing a PCIe physical link connection (not shown) between the PCI-express root port and the SR-IOV adapter 508. The SR-IOV adapter 508 may include physical functions (PFs) 516, 518 respectively coupled to ports 520 and 522. The SR-IOV adapter 508 may further include virtual functions (VFs) 524, 526 associated with the physical function 516, and virtual functions 528, 530 associated with the physical function 518. The operating system 502 may include multiple PCI virtual function device drivers 532, 534.
The SR-IOV adapter 508 may present one or more of the physical functions 516, 518 at the PCI bus device 510 across a PCI link. The physical functions 516, 518 may respond to configuration read and write cycles (e.g., at physical functions 516, 518 numbering 0 through 7). Alternatively, the SR-IOV adapter 508 may be designed according to PCI alternate routing ID (ARI) configuration addressing. Each physical function 516, 518 may have a unique configuration function number (e.g., ranging from 0 to 255 at an implied device number of 0). The ports 520, 522 may create a connection to an external peripheral I/O interconnect, such as Ethernet, Fiber Channel, or other peripheral device interconnects.
Each physical function 516, 518 may be further configured by the virtualization intermediary 512 into one or more of the virtual functions 524, 526, 528, 530. An embodiment of the virtualization intermediary 512 may include program code residing within firmware of the computer system 500. An embodiment of the virtualization intermediary 512 may include a hypervisor. The hypervisor may be a component of the computer system firmware or a type of operating system, or program within an operating system, that is a host to the operating systems 502. Another embodiment of the hypervisor may be a PCI manager program within the computer system having access to the SR-IOV adapter 508 by some physical interconnect that may be a PCI link or other physical connection. The PCI manager of an embodiment may be located locally or remotely, e.g., in a separate processor or memory.
Each virtual function 524, 526, 528, 530 may provide a PCI device programming interface that may be controlled by a PCI virtual function device driver 532, 534. The PCI virtual function device drivers 532, 534 may control the virtual functions 524, 526, 528, 530 to perform I/O transactions through the ports 520, 522 on behalf of the operating system 502.
As discussed herein, the virtual functions 524, 526, 528, 530 may be created under physical functions 516, 518, which may be associated with the ports 520, 522. The virtual functions 524, 526, 528, 530 may thus share the physical facilities of the ports 520, 522. The virtual functions 524, 526, 528, 530 may have a limited ability to perform I/O transactions through the ports 520, 522 and may affect the physical states of the ports 520, 522. The virtual functions 524, 526, 528, 530 may reconfigure the number and capabilities of the individual physical function 516, 518 within the SR-IOV adapter 508.
The computer system hardware 604 may include a PCI host bridge (PHB) 614 coupled to the SR-IOV adapter 606 via a PCIe link 616. A PCIe bus (not shown) may be logically superimposed on the PCIe link 616 to facilitate PCI bus transactions between the PCI host bridge 614 and the SR-IOV adapter 606.
The SR-IOV adapter 606 may include physical functions (PFs) 618, 620 that are respectively coupled to ports 622 and 624. The SR-IOV adapter 606 may further include a virtual function (VF) 626 associated with the physical function 618, and a virtual function 628 associated with the physical function 620. As shown in
The operating system 610 may include a PCI device tree 634 and multiple PCI virtual function device drivers 636, 638.
A virtual PCI host bridge node 648 of the PCI device tree 634 may be associated with the virtual PCI host bridge domain 630. The virtual PCI host bridge node 648 may include a hot plug ID 650, a DLPAR ID 652, and PCI bus properties 654. The virtual PCI host bridge node 648 may be associated with device nodes 656 and 674. The device node 656 may also be associated with the virtual function 628 and the PCI virtual function device driver 636. The device node 656 may include a configuration space 658, MIMIO and DMA space 660, PCI ROMBAR/ROM space 662, and interrupts 664.
The device node 674 may also be associated with the virtual function 626 and the PCI virtual function device driver 638. The device node 674 may include a configuration space 676, MIMIO and DMA space 678, PCI ROMBAR/ROM space 680, and interrupts 682 that are an exclusive subset of the MMIO, DMA, and ROMBAR spaces and interrupts provided by the physical PCI host bridge 614.
The SR-IOV adapter 606 may present one or a plurality of the physical functions 618, 620 at the PCIe link 616 across a PCIe bus. The physical functions 618, 620 may respond to configuration read and write cycles. Alternatively, the SR-IOV adapter 606 may be designed according to PCI ARI configuration addressing. Each physical function 618, 620 may have a unique configuration function number. The ports 622, 624 may create a connection to an external peripheral I/O interconnect, such as Ethernet, Fiber Channel, or other peripheral device interconnects.
Each physical function 618, 620 may be further configured by the virtualization intermediary 608 into one or more of the virtual functions 626, 628. An embodiment of the virtualization intermediary 608 may include a program code within firmware of the computer system 600. Another embodiment of the virtualization intermediary 608 may be a hypervisor. The virtualization intermediary 608 may be a component of the computer system firmware or a type of operating system that is a host to the operating systems 610. Another embodiment of the virtualization intermediary 608 may be a PCI manager.
Each virtual function 626, 628 may provide a PCI device programming interface that may be controlled by PCI virtual function device drivers 636, 638. The PCI virtual function device drivers 636, 638 may control the virtual functions 626, 628 to perform I/O transactions through the ports 622, 624 on behalf of the operating system 610.
As discussed herein, the virtual functions 626, 628 may be created under the physical functions 618, 620, which may be associated with the ports 622, 624. The virtual functions 626, 628 may thus share the physical facilities of the ports 622, 624. The virtual functions 626, 628 may have a limited ability to perform I/O transactions through the ports 622, 624 and may affect the physical state of the port 622, 624. The virtual functions 626, 628 may reconfigure the number and capabilities of individual physical function 618, 620 within the SR-IOV adapter 606. The SR-IOV adapter 606 may be assigned to a single logical partition (e.g., and may not be shared by other logical partitions). The computer system 600 may be configured with the single logical partition 602 and the associated operating system 610 so as to appear as a non-partitioned computer system. The PCI virtual function device driver 636 may be configured for a particular type of virtual function owned by a single operating system 610 regardless of whether the SRIOV adapters are located in a logically partitioned computing system.
The configuration firmware 612 may determine the PCI hierarchy containing the SR-IOV adapter 606. Prior to that determination, the virtualization intermediary 608 may detect and configure the SR-IOV adapter to establish a virtual function 626, 628 for each of the physical ports 622, 624. For an illustrative SR-IOV adapter 606, the virtualization intermediary 608 may configure virtual functions 626, 628 to be in a one-to-one correspondence with each physical port 618, 620.
The SR-IOV adapter 606 may support different peripheral device protocols to concurrently access a physical port 618, 620. For example, the SR-IOV adapter 606 may be a converged network adapters configured to enable Ethernet and Fibre-Channel-Over-Ethernet (FCoE) protocols to simultaneously operate over a single physical port 618, 620.
The virtualization intermediary 608 may create a unique instance of a virtual function 626, 628 for each protocol and on each physical port 618, 620 configured to operate multiple protocols. For example, for an illustrative SR-IOV adapter having four physical ports and enabling two protocols (e.g., Ethernet and FCoE), the virtualization intermediary 608 may configure two virtual functions on each physical port, for a total of eight virtual functions.
The virtualization intermediary 608 may provide the configuration firmware 612 with information to construct the PCI device tree 634 having the virtual PCI host bridge node 648. The virtual PCI host bridge node 648 may be associated with the virtual functions 626, 628 of the SR-IOV adapter 606 assigned to the logical partition 602 The virtual PCI host bridge node 648 may be representative of the combined PCI bus and DLPAR domain properties of the PCI host bridge 614, the SR-IOV adapter 606, and the physical functions 618, 620 indicated as the virtual PCI host bridge domain 630.
The PCI bus properties 654 may be used by the virtualization intermediary 608 to address the virtual PCI host bridge domain 630. For instance, the virtualization intermediary 608 may translate PCI bus operations targeting the virtual PCI host bridge node 648. As such, the presence of the physical function 620 may be transparent to the operating system 610, as well as to the configuration firmware 612 of the logical partition 602.
The configuration firmware 612 may perform PCI hierarchy detection using PCI configuration read operations across the PCIe link 616. The configuration firmware 612 may thus detect the presence of a PCI function at various possible device addresses. For example, a function may be detected at function numbers 0 through 7, or alternatively at ARI function numbers 0 through 255 of an implied ARI device number.
The virtualization intermediary 608 may intercept PCI configuration read or write transactions to the PCIe link 616. The virtualization intermediary 608 may respond to a PCI bus configuration read operation such that the configuration firmware 612 first detects the virtual function 626 at an emulated function number 0 of the virtual PCI host bridge bus and device 0. The virtualization intermediary 608 may respond to the configuration firmware reads that are directed to PCI device 0 and function 0 below the virtual PCI host bridge. In the case that the SRIOV adapter is configured for just a single virtual function 626 associated with a single physical function 618 further associated with a single port 622, the configuration firmware 612 may detect only a single PCI function, at function 0, in the PCI hierarchy below the virtual PCI host bridge. In another case, the SRIOV adapter is configured for a plurality of virtual functions 626, 628 and the configuration firmware 612 may detect a multi-function PCI function at function 0 corresponding for example to virtual function 626, in the PCI hierarchy below the virtual PCI host bridge, and detect another multi-function PCI function at function 1 corresponding for example to virtual function 628. The virtualization intermediary 608 may thereby make the presence of the physical functions 620 transparent to the configuration firmware 612. The virtual functions 626, 628 may thus be represented to the operating system 610 in a manner analogous to that of functions of a PCI multi-function legacy adapter, such as the PCI adapter 408 of
The virtualization intermediary 608 may pass configuration read operations directly to an actual virtual function configuration register within the SR-IOV adapter 606. The logical virtual PCI host bridge bus number and device function number may be translated to the actual PCI configuration bus/device/function number utilized on the physical PCI bus, or PCIe link 616.
In another embodiment, the virtualization intermediary 608 may respond directly to the configuration firmware read operations with an emulated register value. The virtualization intermediary 608 may have derived the emulated register value as part of configuring the SR-IOV adapter 606 in SR-IOV mode. This action may maintain the appearance of the virtual functions 626, 628 as functions of a PCI multi-function device. The transparency of the physical functions 618, 620 on the virtual PCI host bridge bus may further be maintained with respect to the configuration firmware 612.
The configuration firmware 612 may also be modified from a legacy PCI function configuration to account for limitations of the PCI SR-IOV Architecture. The limitations may relate to the assignment of memory mapped address spaces associated with the virtual functions 626, 628. The configuration firmware 612 may write to the PCI base address registers of a PCI function to determine the size of the PCI memory space used by that base address register of that function. The configuration firmware 612 may select a location within PCI memory at which to bind the base address register and associated PCI memory space. However, the virtualization intermediary 608 may establish a location of the PCI memory regions to map virtual function PCI memory spaces using base address registers in the physical functions 618, 620.
According to the SR-IOV architecture, the virtual functions 626, 628 may not actually implement the PCI base address registers of a PCI function. As such, the PCI bus properties 654 of the virtual PCI host bridge node 648 may specify that the PCI base address registers are read only and cannot be changed in relation to their PCI memory location. As discussed herein, the PCI base address registers may belong to the device(s) on the PCI bus associated with the virtual PCI host bridge node 648.
In order for the configuration firmware 612 to determine the size of each PCI base address space within the virtual functions 626, 628, the configuration firmware 612 may perform the configuration write of all-ones data to each base address register. The virtualization intermediary 608 may emulate the action by storing temporary all-ones values. Where the configuration firmware 612 reads from the base address register, the virtualization intermediary 608 may return an emulated value of all-one bits. The emulated value may indicate the power of two size of the PCI memory space associated with the virtual function base address register. The virtual functions 626, 628 may then return the actual PCI address associated with that virtual function base address register for subsequent configuration reads from that virtual function base address register.
A legacy PCI function may be connected to a ROM device containing adapter vital product data or boot drivers used with that PCI function or adapter. The PCI function may include a ROMBAR that is subject to location within PCI memory by the configuration firmware 612. The virtualization intermediary 608 and configuration firmware 612 may perform the same sequence regarding the ROM base address register within the virtual function configuration space.
The operating system 610 may provide hot plug support. A hot plug module may enable a user to use an application interface within the operating system 610 to select a particular physical slot. The physical slot may include a PCI adapter for powering off or on. The hot plug module may enable the user to remove or add a PCI adapter without disrupting other functions of the computer system 600.
The PCI device tree 634 may be generated by the virtualization intermediary 608. The PCI host bridge node 640 of the PCI device tree 634 may represent the physical PCI host bridge 614 of the computer system hardware 604. The PCI host bridge node 640 may not include a PCI device within its hierarchy, but may include a hot plug ID 642. The operating system 610 may associate the hot plug ID 642 with a physical location of a PCIe slot. The PCIe slot may accommodate an adapter, such as the SR-IOV adapter 606, or a legacy, non-SRI-OV PCIE adapter, in the same location connected to the PCIe link 616.
The hot plug ID 642 may be a logical ID that corresponds to a physical slot location or a power domain associated with the physical slot. The hot plug ID 642 may, itself, be the physical location ID, such as a system physical location code. Hot plug power operations may utilize the hot plug ID 642 to instruct the operating system 610 with the physical location of a power domain within the computer system hardware 604. The power domain may be the object of a power off or power on operation. The operating system 610 may use the hot plug ID 642 to determine PCI host bridges and PCI devices within the PCI device tree 634 that are affected by a power off or power on to the hot plug location.
An empty PCI slot may be assigned to a logical partition 602, and a PCI adapter may later be added to the PCI slot. The virtualization intermediary 608 may present the operating system 610 with the PCI host bridge node 640. The operating system 610 may use the PCI host bridge node 640 to identify the location of a hot plug power on operation. Such a hot plug power operation may add a PCIe adapter to a physical PCI host bridge 614.
While the adapter shown in
According to a particular embodiment, the virtualization intermediary 608 determines that the operating system 610 does not use virtual function device drivers for the adapter 606. In such a scenario, the virtualization intermediary 608 may take no further action. As shown in
The virtualization intermediary 608 may generate the PCI device tree 634 for the operating system 610. The operating system 610 may include the virtual PCI host bridge node 648 for the virtual functions 626, 628. The virtualization intermediary 608 may intercept PCI configuration cycles of the configuration firmware 612 to the PCI bus. The PCI bus may be associated with the PCI host bridge 614. The virtualization intermediary 608 may return that there are no devices present. For example, the PCI host bridge node 640 may have no associated device nodes 656. The configuration firmware 612 may detect PCI functions of a PCI multi-function device at the PCI host bridge node 614 to generate a device node 656 for that associated virtual function. The configuration firmware 612 may further create an instance of a virtual function device driver 636, 638 in association with the device nodes 656.
An embodiment may enable the powering-off an adapter that is configured within a running logical partition 602. The power-off operation may allow repair or replacement of an adapter with an alternative adapter. The new adapter may be a different type than the original adapter.
The hot plug power off operation may use the hot plug ID 650 to identify a power domain containing a PCIE adapter. Accordingly, the hot plug ID 650 may enable the virtualization intermediary 608 to identify all PCI devices within the shared hot plug domain represented by the physical slot location of the adapter 606.
Prior to performing the physical power off operation, the operating system 610 may determine all affected PCI devices by correlating the hot plug ID specified in the operation with the hot plug ID 650 in the virtual PCI host bridge node 648. The operating system(s) 610 may then terminate the operations of the device drivers 636, 638 associated with the device nodes 656, 674 under the virtual PCI host bridge node 648 having that same hot plug ID 650. Once the device drivers 636, 638 have terminated operations, the virtualization intermediary 608 and hot plug module may continue with the physical power off operation of the hot plug domain associated with that hot plug ID 650.
Where a PCIe slot containing an adapter has been powered off, it may be possible for the system user or a service representative to repair or replace the adapter. The replacement adapter may be a different type of adapter (e.g., replacing a PCIe adapter with an SR-IOV capable adapter or vice versa). In either case, a subsequent power-on of the PCIe slot may result in the virtualization intermediary 608 presenting the operating system 610 with an updated PCI device tree 634. The operating system 610 may use the SR-IOV virtual function device drivers 636, 638, along with virtual PCI host bridge node 648 for each of the SR-IOV virtual functions 626, 628 that have been configured by the virtualization intermediary 608.
A PCI slot may be removed from or added to the control of a particular running logical partition 602. A PCI adapter may be removed from a logical partition to transfer that adapter to another logical partition during a dynamic logical partitioning (DLPAR) operation. DLPAR operations may reference a PCIE adapter. For example, the PCIE slot location within the computer system 600 may be referenced using a DLPAR ID 652.
According to an embodiment, a PCIe slot associated with the PCI host bridge 614 may not be assigned initially to the logical partition 602 at the time that the logical partition 602 is booted. Adding the PCIe slot to the logical partition 602 may result in the virtualization intermediary 608 determining whether the adapter is an SRIOV type. Where the adapter is a non-SR-IOV type, the virtualization intermediary 608 may add a PCI host bridge node 418 to the PCIE device tree 416 as shown in
Where the adapter 606 is SR-IOV capable, the virtualization intermediary 608 may determine whether the operating system 610 uses the SR-IOV virtual function device drivers 636, 638 as non-SR-IOV mode device drivers. Where the SR-IOV virtual function device drivers 636, 638 are not used, the virtualization intermediary 608 may add a PCI host bridge node 418 to the PCIE device tree 416, as shown in
The virtualization intermediary 608 may intercept the PCI configuration cycles of the configuration firmware 612 to the PCI bus associated with the vPHB 648. The configuration firmware 612 may then detect PCI functions of a PCI multi-function device at the PCI host bridge node 648 to generate a device node 656, 674 for the associated virtual function. The configuration firmware 612 may further create an instance of a virtual function device driver 636, 638 associated with the device node 656, 674.
According to an embodiment, no adapter may be physically plugged into a PCI slot that has been transferred to the running logical partition 602. A later hot plug power-on operation may add an adapter to the running logical partition 602. For example, the virtualization intermediary 608 may enable the operating system 610 to selectively use the adapter 606 according to a legacy or an SR-IOV configuration. Conversely, a user may initiate the automatic removal of the SR-IOV adapter 606 from the running logical partition 608. The DLPAR ID 652 may be used by the operating system 610 of that logical partition 608 to identify the PCI devices and functions that will be removed during the DLPAR operation.
As represented by the PCI device tree 634, the operating system 610, may determine the affected PCI devices prior to relinquishing control of the affected PCI devices. The operating system 610 may correlate the DLPR ID specified in the operation with the DLPAR ID 652 in the virtual PCI host bridge node 648. The operating system(s) 610 may then terminate the operations of the virtual PCI host bridge (vPHB) 648 having DLPAR ID 652 and the device nodes 656, 674 under the virtual PCI host bridge 648 (e.g., having that same DLPAR ID 652). Once all the device drivers 636, 638 have terminated operations, the slot may be assigned to another, different logical partition 602. The slot may alternatively be added back to the original logical partition 602.
According to a particular embodiment, an SR-IOV adapter may be plugged below a PCI bridge, such as a PCI bridge of a PCIe switch. The PCIe switch may form a PCIe link below a bridge that is analogous to the PCIe link 616. The virtual PCI host bridge 648 may be presented to the logical partition configuration firmware 612. The PCI bus properties 654 of the virtual PCI host bridge 648 may account for combined properties of the physical PCI host bridge 614 and the PCIe switch. Illustrative such properties may include PCI bus memory and DMA address ranges, as well as interrupt assignments.
According to a particular embodiment, the virtualization intermediary 608 may not configure the SR-IOV adapter 606 for SR-IOV mode. Alternatively, the virtualization intermediary 608 may enable the configuration firmware 612 to fully detect and control configuration functions of the SR-IOV adapter 606. As such, the configuration firmware 612 may, itself, configure the SR-IOV adapter 606 for SR-IOV mode. The configuration firmware 612 may function as a virtualization intermediary local to the logical partition 602. The local virtualization intermediary may make the SR-IOV aspects of the SR-IOV adapter 606 visible to elements of the logical partition 602, the operating system 610, the device tree 634, or the device drivers 636, 638.
Operations of an embodiment are not limited by whether or not the configuration firmware 612 enables SR-IOV mode within an SR-IOV adapter that has not been virtualized by a virtualization intermediary 608 external to the logical partition 602. Operability may further be independent of what by method the logical partition 602 represents the SR-IOV adapter 606 within its device tree 634 or enables device driver translations to the functions of the SR-IOV adapter 606.
The SR-IOV adapter 606 may be assigned to a single operating system within a logical partition that is non-shared. The system 600 of a particular embodiment may determine whether to configure the SR-IOV adapter 606 for SR-IOV mode based on a configuration file accessible to the virtualization intermediary 608 upon detecting that the adapter is SR-IOV-capable.
The PCIM may, in turn, initialize the SR-IOV adapter at 704. A request from the virtualization intermediary may be received by the PCIM at 706. The request may prompt the PCIM to configure the adapter in non-shared mode with virtual PCI host bridges.
At 708, the PCIM may locate a first physical port on the SR-IOV adapter. The first protocol supported by the port may be determined at 710. Accordingly, the PCIM may configure at 712 a virtual function on the located physical port with the determined protocol.
The PCIM may at 714 retrieve identification and I/O resource information for the virtual function. Illustrative virtual function information may include: identification of the owning logical partition, a routing ID (e.g., a configuration space address), an allocated system interrupt source, a PCI MMIO BAR address and size, a PCI DMA address and size, vendor and device IDs, a virtual function port ID, and a virtual function ID.
At 716, the virtualization intermediary may store the virtual function information for later use. For example, the virtual function information may be used during an operating system boot. The PCIM may at 720 communicate the virtual function information to the virtualization intermediary.
The PCIM may determine at 720 whether another protocol is supported at the located physical port. Where another protocol is supported at the located physical port, the PCIM may determine the next protocol supported by the physical port at 722. The PCIM may configure at 712 a virtual function on the located physical port with the next, determined protocol.
Where no other protocol is supported at the located physical port at 720, the PCIM may determine whether there is another physical port at 724 for which to configure virtual functions. At 726, the PCIM may locate the next physical port on the adapter. A protocol supported by the port may be determined at 710. Accordingly, the PCIM may configure at 712 a virtual function on the next located physical port with the determined protocol.
A hot plug power-on operation may be initiated at 804. For instance, the user may use an operating system utility to initiate the hot plug power-on operation. At 806, the operating system may request that the virtualization intermediary initiate a physical power-on of the adapter slot. The virtualization intermediary may perform at 808 the hardware operations to apply power to the SR-IOV adapter. The virtualization intermediary may further prepare the bridge, adapter slot, and adapter to be used by a logical partition.
The virtualization intermediary may determine whether the SR-IOV is supported on the adapter slot at 810. The virtualization intermediary may further determine whether a PCIM driver is available for the adapter. The virtualization intermediary may read identification information, such as the PCI configuration ID, from the SR-IOV adapter. The virtualization intermediary may use the identification information to determine whether the adapter is SR-IOV-capable and has a compatible PCIM driver.
Where the adapter is either not SR-IOV-capable or does not have a compatible PCIM driver, legacy support may be provided to power-on the adapter at 812. An indication that the power-on operation is complete may be sent to the operating system.
Where the adapter is alternatively SR-IOV-capable and has a compatible PCIM driver, the virtualization intermediary may determine at 814 whether the logical partition that owns the adapter slot needs non-shared mode SR-IOV enablement. Where non-shared mode SR-IOV enablement is unneeded, legacy support may be provided to power-on the adapter at 812, and an indication that the power-on operation is complete may be sent to the operating system.
Where non-shared mode SR-IOV enablement is alternatively desired at 814, the non-shared mode configuration may be initiated at 816. Processes illustrative of the non-shared mode configuration are described in connection with the method 700 of
The operating system at 820 may query the virtualization intermediary for a list of new PCI host bridges. At 822, the virtualization intermediary may generate a unique PCI host bridge ID for the virtual function(s) that has been configured by the PCIM. The virtualization intermediary may further return the host bridge ID to the operating system and/or logical partition.
The operating system at 824 may request information for the PCI host bridge ID from the virtualization intermediary. The virtualization intermediary may at 826 present the virtual PCI host bridge and the virtual function device to the configuration firmware (i.e., partition firmware), as described in the method 1,000 of
The configuration firmware may at 828 present bus and virtual functions to the operating system as functions of a multi-function device adapter. Illustrative such presentation processes are described in connection with the method 1,100 described of the flowchart of
At 904, an operating system may locate a first device that matches location information of an adapter slot. The operating system may terminate applications and/or device drivers for the device at 906. At 908, the operating system may relinquish control of the device back to the virtualization intermediary.
The operating system may determine at 910 whether there is a device that matches the location information of the adapter slot. Where a device matches the location information of the adapter slot, the operating system at 912 may locate the next device matching the location information of the adapter slot. Alternatively, the operating system may request at 914 that the virtualization intermediary initiate the physical power-off of the slot in response to there being no devices that match the location information of the adapter slot. The virtualization intermediary may perform hardware operations at 916 to power-off the adapter and may return to the operating system.
The virtualization intermediary may create and store a PCI host bridge unique ID at 1,004. The PCI host bridge unique ID may map directly to the unique ID of the virtual PHB in the PCI host bridge data structure PCI host bridge ID fields. At 1,006, the virtualization intermediary may store configuration space address information with a predetermined primary bus number. The information may be stored in the PCI host bridge data structure PCI configuration space fields. The virtualization intermediary may locate the virtual function's adapter's parent physical PCI host bridge MMIO extents. The virtualization intermediary may store the information in the virtual function's virtual PCI host bridge data structure MMIO range fields.
At 1,010, the virtualization intermediary may locate the virtual function's PCI DMA window information. The virtualization intermediary may store the information in the virtual function's virtual PCI host bridge data structure DMA range fields.
The virtualization intermediary may create at 1,012 a data structure to identify a virtual function to the logical partition. The virtualization intermediary may identify the virtual function data structure at 1,014 as the child to the virtual PCI host bridge data structure and may store the virtual function's unique ID to the virtual function data structure. The virtualization intermediary may use at 1,016 the physical link information from the virtual function's adapter's parent physical PCI host bridge for the virtual function's PCI host bridge data structure physical link information fields. The virtualization intermediary at 1,018 may store the PCI configuration space information to the virtual function data structure, including a bus number range that includes the predetermined bus number, device zero, and function zero to address the virtual function.
At 1,020, the virtualization intermediary may store the virtual function PCI MMIO BAR information to the virtual function data structure MMIO range fields. The virtualization intermediary may store at 1,022 the virtual function PCI DMA window information to the virtual function data structure DMA range fields. The virtualization intermediary may locate at 1,024 the virtual function's parent adapter's slot physical, mechanical, and location data and may store the information to the virtual function data structure. At 1,026, the virtualization intermediary may copy the PCI host bridge and virtual function data structures to logical partition memory.
At 1,106, the configuration firmware may probe the device as a physical PCI adapter. The virtualization intermediary may emulate at 1,108 the configuration space. At 1,110, the configuration firmware determines that a device is present at the function number provided in the probe. At 1,110, the configuration firmware may detect that no device is present, as result of the probe. The configuration firmware at 1,122 may increment the function number to use in the probe. The configuration firmware at 1,124 may determine that the incremented device number is beyond the range of function numbers to probe and discontinue probing at 1,126. In conventional configuration addressing mode a function number greater than 7 may be determined to be beyond the range of functions to probe. In ARI mode a function number greater than 255 may be determined to be beyond the range of functions to probe.
Alternatively, the configuration firmware may determine at 1,124 that the function number is within the range of function numbers to probe. At 1,110 the configuration firmware may locate a PCI multi-function device as a result of the probe, corresponding to a virtual function of the SRIOV adapter.
At 1,112, the configuration firmware may use the PCI host bridge child virtual function data structure for the device that was found during the probe to determine PCI MMIO BAR sizes. The configuration firmware may read at 1,114 the PCI BAR registers or us the virtual function data structure to determine address for the PCI BAR regions. The configuration firmware may use the PCI host bridge child virtual function data structure at 1,116 to determine other I/O resource allocation parameters as any other PCI device.
The configuration firmware may at 1,118 insert data the virtual function into an operating system-level data structure(s) that describes a function of a PCI multi-function device. Information in the operating system-level data structure(s) for a function (e.g., PCI physical device or virtual function) may include: a PCI configuration space address (e.g., a routing ID), PCI IDs, DMA window addresses and sizes, MMIO BAR region addresses and sizes, MSI interrupt assignments, and vendor product data. At 1,120, the configuration firmware may store data for the bus and the device to an appropriate location(s) for the operating system. At 1,122, the configuration firmware may increment the function number. At 1,124, the configuration firmware may determine that this function number is beyond the maximum function number to probe and discontinue probing. Alternatively, at 1,124 the configuration firmware may determine that the incremented function number is within the range of function numbers to probe and at 1,106, the configuration firmware may probe for a device at that function number.
A PCI adapter slot that includes an SR-IOV adapter may be controlled by a single operating system instance or a logical partition of a computing system. The PCI adapter slot may be controlled either by default on a non-partitioned system or where the PCI slot and the SR-IOV adapter are assigned to a logical partition of a logically partitioned computer. More particularly, the PCI slot and the SR-IOV adapter may be assigned to a virtualization intermediary, such as a hypervisor, a PCIM, a physical function manager, or system firmware.
A determination may be made as to whether the adapter is SR-IOV-capable and as to whether the operating system desires or is capable of using the adapter as SR-IOV enabled. The adapter may be configured and enabled for SR-IOV with a subset of possible SR-IOV virtual functions. Control of virtual function PCI hierarchies comprising virtual PCI host bridges each connecting a single function PCI device may be transferred to the single operating system instance or logical partition. The adapter may be controlled via the SR-IOV function, and the adapter may not be configurable directly within the operating system.
The virtualization intermediary may present the PCI adapter slot and PCI host bridge/root port hierarchy to the operating system as a PCI slot alike to other PCI slots. The virtualization intermediary may emulate the device in the slot as a PCI multi-function device in either conventional or ARI configuration addressing modes. The virtual functions of the SR-IOV adapter may be controlled by the operating system instance or the logical partition. There may be one virtual function per adapter protocol, per physical adapter port, or physical function.
The virtualization intermediary may present the PCI host bridge and the adapter slot to the operating system instance or the logical partition the same as for any PCI adapter slot including a non-SR-IOV adapter. The virtualization intermediary may present the virtual functions to the operating system as non-SR-IOV functions of a conventional PCI multi-function device. The virtualization intermediary may identify that virtual function as device 0 and a function number within the range of function numbers appropriate for conventional or ARI mode configuration addressing. The virtual functions may be connected to a common PCI bus of a virtual PCI host bridge. The virtualization intermediary may not expose the SR-IOV physical functions or the presence of some of the virtual functions sharing the same physical PCI bus. The virtualization intermediary/firmware may emulate virtual function BAR registers in the base PCI configuration space of each virtual function. The virtual functions may include values that the virtualization intermediary/firmware assigned in setting the virtual function's parent physical function SR-IOV capability virtual function BARs. The virtualization intermediary may present a ROMBAR that is mapped by a virtual function to enable the extraction of the adapter VPD and a (UEFI or FCODE) virtual function boot driver, and additional resources the virtualization intermediary/firmware. The virtual function may be operated within a logical partition and/or operating system as an individual function endpoint of a PCIe single function device.
The operating system may configure and manage the PCI hierarchy that includes PCI host bridges and PCI endpoint functions no differently than it would for a non-SR-IOV PCI multi-function device that is connected to a PCI host bridge. This may be facilitated by the virtualization intermediary presenting each of the virtual functions as non-IOV multi-function devices connected to a common virtualized PCI host bridge.
According to an embodiment, the virtualization intermediary or firmware may not configure the adapter in SR-IOV mode. Instead, the operating system may operate the adapter as a legacy PCI-Express adapter. Whether to configure the adapter for SR-IOV may be determined by an adapter profile registering the type, version, or capability of an operating system.
A particular embodiment may enable PCI hot plug operations on a PCI adapter slot that includes an SR-IOV adapter that is under the control of a single operating system instance or a logical partition. The operating system may associate the virtual functions with the PCI adapter hot plug domain in a manner identical to that for functions of a non-SR-IOV PCI multi-function adapter. For instance, the operating system may deactivate the virtual function device drivers and de-configure the PCI devices or functions associated with the PCI slot prior to performing a power off operation of the adapter.
During a hot plug power operation, the virtualization intermediary may determine that the adapter is SR-IOV-capable and that the operating system desires or is capable of using the adapter as SR-IOV-enabled. The virtualization intermediary may enable and configure the adapter for SR-IOV with a sufficient subset of possible SR-IOV virtual functions. Control of the collection of virtual PCI host bridge/root port and virtual function PCI devices may be transferred to the logical partition or operating system. The virtualization intermediary may present the PCI adapter slot and PCI host bridge/root port hierarchy to the operating system as a PCI slot similar to other PCI slots, except that the operating system may be aware that the slot containing the PCI adapter is not configurable directly within the operating system, but has associated virtual resources. The subset of virtual functions may include one virtual function per adapter protocol, per physical adapter port, or physical function.
The virtualization intermediary may present the PCI host bridge/root port and adapter slot to the operating system and logical partition in the same manner as for any PCI adapter slot containing a non-SR-IOV adapter. The virtualization intermediary may further present the virtual functions to the operating system as non-SR-IOV functions of a conventional PCI multi-function device, identified as device 0 and a number within the range of function numbers appropriate to conventional or ARI mode configuration addressing. The virtualization intermediary may not expose the SR-IOV physical functions or expose some of the virtual functions sharing a common virtual PCI host bridge. The virtualization intermediary may emulate virtual function BAR registers as RO BARs 0-5 in the base PCI configuration space of each virtual function having the values the virtualization intermediary/firmware assigned in setting the virtual function's parent physical function SR-IOV capability virtual function BARs. The virtualization intermediary may present a ROMBAR that is mapped RO by a virtual function to enable extraction of adapter VPD and a (UEFI or FCODE) virtual function boot driver to emulate and operate the virtual functions within a logical partition/operating system as an individual function endpoint of a PCIe multi-function device.
The operating system may configure and manage the PCI hierarchy including the PCI host bridge/root port and virtual functions the same as for a non-SR-IOV PCI multi-function device connected to that PCI host bridge/root port by virtue of the virtualization intermediary presenting the virtual functions as non-IOV multi-function devices connected to a common virtualized PCI host bridge. According to an embodiment, the virtualization intermediary does not configure the adapter in SR-IOV mode, and instead allows the operating system to operate the adapter as a legacy PCI-Express adapter. The configuration of the adapter may be determined by an adapter profile registering the type, and possibly an operating system version or capability.
During a hot plug power operation, the virtualization intermediary may not configure the adapter in SR-IOV mode and may alternatively allow the operating system to operate the adapter as a legacy PCI-Express adapter. The adapter slot that is the target of an operating system hot plug power-on operation may have been empty prior to the power-on operation. According to another embodiment, the targeted adapter slot of an operating system hot plug power-on may have been a slot occupied by a non-SR-IOV capable adapter that was powered off and replaced with an SR-IOV capable adapter prior to the power on operation.
According to another embodiment, the targeted adapter slot of an operating system hot plug power-on may have been a slot occupied by an SR-IOV capable adapter that was powered off and replaced with a different SR-IOV capable adapter prior to the power on operation. According to another embodiment, the targeted adapter slot of an operating system hot plug power-on may have been a slot occupied by an SR-IOV capable adapter that was powered off and replaced with another of the same type, or the very same, SR-IOV capable adapter prior to the power-on operation.
Another operation may add or remove assignments of logical partitions to a PCI adapter slot and an associated SR-IOV-capable adapter. An “add” assignment to a logical partition may result in the virtualization intermediary determining that the adapter is SR-10V-capable and that the operating system desires or is capable of using the adapter in SR-IOV-enabled mode. The virtualization intermediary may enable and configure the adapter for SR-IOV with a sufficient subset of possible SR-IOV virtual functions.
The adapter slot containing the SR-IOV-enabled adapter may be an object of logical partition remove operation. A logical partition remove operation of an SR-IOV-enabled adapter from an operating system or a logical partition may result in the operating system associating the virtual functions with the PCI adapter DLPAR domain in a manner identical to that for functions of non-SR-IOV PCI adapters sharing a common DLPAR domain. The virtual function device drivers may be deactivated, and the PCI devices associated with the PCI DLPAR domain may be de-configured prior to completing the removal of the adapter from the logical partition. A logical partition remove assignment of the SR-IOV-enabled adapter may result in the removal of the virtualization intermediary and related virtualization intermediary storage.
Particular embodiments described herein may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment containing both hardware and software elements. In a particular embodiment, the disclosed methods are implemented in software that is embedded in processor readable storage medium and executed by a processor, which includes but is not limited to firmware, resident software, microcode, etc.
Further, embodiments of the present disclosure, such as the one or more embodiments may take the form of a computer program product accessible from a computer-usable or computer-readable storage medium providing program code for use by or in connection with a computer or any instruction execution system. For the purposes of this description, a computer-usable or computer-readable storage medium may be any apparatus that may tangibly embody a computer program and that may contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
In various embodiments, the medium may include an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system (or apparatus or device) or a propagation medium. Examples of a computer-readable storage medium include a semiconductor or solid state memory, magnetic tape, a removable computer diskette, a random access memory (RAM), a read-only memory (ROM), a rigid magnetic disk and an optical disk. Current examples of optical disks include compact disk-read-only memory (CD-ROM), compact disk-read/write (CD-R/W) and digital versatile disk (DVD).
A data processing system suitable for storing and/or executing program code may include at least one processor coupled directly or indirectly to memory elements through a system bus. The memory elements may include local memory employed during actual execution of the program code, bulk storage, and cache memories which provide temporary storage of at least some program code in order to reduce the number of times code must be retrieved from bulk storage during execution.
Input/output or I/O devices (including but not limited to keyboards, displays, pointing devices, etc.) may be coupled to the data processing system either directly or through intervening I/O controllers. Network adapters may also be coupled to the data processing system to enable the data processing system to become coupled to other data processing systems or remote printers or storage devices through intervening private or public networks. Modems, cable modems, and Ethernet cards are just a few of the currently available types of network adapters.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the disclosed embodiments. Use of the terms Peripheral Component Interconnect Express (PCIe) and Peripheral Component Interconnect (PCI) may be used interchangeably in some instances. Moreover, the terms operating system and logical partition may be used interchangeably in certain of the embodiments described herein. Various modifications to these embodiments, including embodiments of I/O adapters virtualized in multi-root input/output virtualization (MR-IOV) embodiments, or virtualized using software virtualization intermediaries, will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the scope of the disclosure. Thus, the present disclosure is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope possible consistent with the principles and features as defined by the following claims.
