The present invention relates generally to the data processing field, and more particularly, relates to a method, apparatus and computer program product for detecting and powering off unused input/output (I/O) slots in a computer system including a logically partitioned computer system.
Low power consumption by computer systems and components is a selling point for computers because it saves customers money through lower utility bills and less machine room cooling equipment.
On some known logically partitioned computer systems, I/O slots with adapters in them or occupied I/O slots are powered on automatically by the hardware when the enclosure, or power domain, in which the slots reside goes through a Power-On-Reset (POR), that is when the enclosure or power domain powers on.
Some known computer systems that support logical partitioning do not attempt to power off unused I/O slots. Rather, occupied I/O slots remain powered on as long as the platform is powered on, regardless of whether the I/O slots are being used or not. Powering these I/O slots takes more power and costs the customer more money.
On a logically partitioned system there are well defined and platform controlled situations where I/O adapters are not being used for extended periods of time. One example is when an I/O slot is not assigned to a logical partition. Another example is when the logical partition to which an I/O slot is assigned is powered off.
A need exists for a mechanism for detecting and powering off unused I/O slots in a logically partitioned computer system.
A principal object of the present invention is to provide a method, apparatus and computer program product for detecting and powering off unused I/O slots in a computer system. Another important object of the present invention is to provide such method, apparatus and computer program product for detecting and powering off unused I/O slots substantially without negative effect and that overcomes some of the disadvantages of prior art arrangements.
In brief, a method, apparatus and computer program product are provided for detecting and powering off unused I/O slots in a computer system including a logically partitioned computer system. For each of the I/O slots, the current slot is checked to determine if the slot is powered on. Responsive to the current slot being powered on, the current slot is checked to determine if the slot is unlocked. Responsive to the current slot being unlocked, the current slot is checked to determine if the current slot is a candidate for being powered off. Responsive to the current slot being unlocked and being a candidate for being powered off, the current slot is powered off and checking continues with a next one of the I/O slots.
In accordance with features of the invention, when the current slot is unlocked and is not already a candidate for being powered off, then the current slot is recorded as being a candidate for being powered off. Then checking continues with a next one of the I/O slots. Recording the slot as a candidate for being powered off and then repeating an iteration of checking the I/O slots before the candidate slot is powered off ensures a proper delay occurs before the slot is actually powered off and avoids powering off a slot that is about to be locked, which could cause slot power cycles in quick succession.
The present invention together with the above and other objects and advantages may best be understood from the following detailed description of the preferred embodiments of the invention illustrated in the drawings, wherein:
In accordance with features of the preferred embodiment, unused I/O slots are powered off automatically. A method is provided to continuously monitor for and detect I/O slots that are not needed on a logically partitioned computer system so that unneeded I/O slots can be powered off. First an unused slot is recorded as a candidate for being powered off, then a delay occurs before the candidate slot is actually powered off to avoid otherwise possible slot power cycles in quick succession.
Referring now to the drawings, in
Computer system 100 is shown in simplified form sufficient for an understanding of the invention. Computer system 100 can be implemented with various computers, for example, with one of the IBM eServer line of computers manufactured by International Business Machines Corporation. It should be understood that the present invention applies to any computer system that includes one or more I/O slots, where I/O adapters are not being used for extended periods of time and the I/O slots advantageously are powered off in accordance with features of the preferred embodiment.
Referring now to
Referring now to
Hypervisor 300 includes a hypervisor hardware manager 308 that encapsulates hypervisor functions to access and control hardware states of the PCI adapter slots 1–8, 118A–118H including power on/power off slot mechanism 112. The I/O slot lock mechanism 108 encapsulate the functions to set the ownership of the slot lock 110 and to serialize transfer of a slot lock between hypervisor 300 and logical partitions 114A–114N. Logical partitions 114A–114N communicate with the hypervisor 300 to invoke hypervisor functions. Hypervisor 300 is an agent of a system administrator interface 310, and performs partition configuration and platform service operations requested by the system administrator 312 through interface 310. System administrator 312 includes an administration console 314 and a hardware console 316.
Hypervisor operating system kernel 304 includes a task for implementing methods for detecting and powering off unused I/O slots in accordance with the preferred embodiment as illustrated and described in
It should be understood that the present invention is not limited for use with the illustrated computer system 100, hardware arrangement 200 or hypervisor 300. The illustrated computer system 100, hardware arrangement 200 and hypervisor 300 are not intended to imply architectural or operating system limitations. The present invention can be used with various computer systems having a plurality of I/O slots and various other internal hardware or peripheral devices. For example, computer system 100 may contain multiple buses, rather than the single system bus 103 shown in
In accordance with features of the preferred embodiment, the method involves continuously iterating through all I/O slots 118 in the system 100 looking for slots that are powered on but not being used. For example, determining whether a slot is being used or not is done based on the status of the slot lock associated with the slot. If the slot lock for a slot is not held, the slot is considered unused. There are periods of time when I/O adapters or slots 118 in the logically partitioned system 100 are not being used and are powered off by the method of the preferred embodiment. Then when I/O slots 118 are needed the slots are powered on again.
In accordance with features of the preferred embodiment, to ensure unused slots 118 are powered off, a hypervisor task in hypervisor operating system kernel 304 is started at platform initial program load (IPL) and runs continuously while the hypervisor 300 is active; that is, the hypervisor task exists until the hypervisor 300 is terminated typically at platform power off or reboot.
For example, consider the logically partitioned system 200 with eight I/O slots, 1–8, 118A–118H, that are occupied and may be contained in a single I/O expansion unit or enclosure 204. I/O slots 1–4, 118A–118D are assigned to partition 1, 114, which is powered on. I/O slots 5–7118E–118G are assigned to partition 2, 114, which is powered off. I/O slot 8, 118H is not assigned to a partition. At some point while the platform is up and running, the enclosure 204 in which I/O slots 1–8118A–118H reside goes through a power-on-reset (POR). The POR could be the result of one of multiple events. Temporary loss and restoration of A/C power would cause a POR, as would happen if the power cord accidentally got pulled out and re-inserted, or electricity temporarily is interrupted and restored. A concurrent maintenance procedure on the enclosure 204 would also cause a POR, such as a concurrent repair action where the hardware management console is used to power off the enclosure, a part is replaced, and the hardware management console is used to power on the enclosure again. When the enclosure 204 goes through the POR, all contained I/O slots 1–8, 118A–118H are powered on automatically by the hardware.
Hypervisor task of hypervisor operating system kernel 304 of the preferred embodiment that started at hypervisor IPL and runs continuously while the hypervisor 300 is powered on, is continuously looking for any of the I/O slots 1–8, 118A–118H that are not needed and can be powered off, such as slots 5–8118E–118H in the above example. The task uses a hypervisor interface, such as power on/power off slot mechanism 112 to determine whether a slot is powered on and to power off an unused slot. The task determines whether a slot is needed based on whether the slot lock 110 for a slot is held or not. The slot lock 110 is a mechanism used to ensure an entity accessing a particular one of slots 1–8, 118A–118H is authorized to do so, as well as to serve as a synchronization point for resetting the slot and disabling memory and other mappings associated with the slot during a dynamic transfer of the slot between partitions 1-N, 114A–114N. In other words, before an entity, such as a partition 1-N, 114A–114N or an external service agent such as hardware management console 316, is allowed to access a particular I/O slot 1–8, 118A–118H through hypervisor controls associated with the slot, the entity must hold the slot look 110 for the slot. If the slot lock 110 associated with a slot 1–8, 118A–118H is not held or the slot is not locked, the slot is not being used. Thus, the hypervisor task of the preferred embodiment considers an I/O slot 1–8, 118A–118H that is powered on, but for which the lock is not held, a candidate for power off.
In accordance with features of the preferred embodiment, to prevent repeated power cycles of I/O slots in quick succession during enclosure concurrent maintenance procedures and procedures involving the dynamic transfer of a slot between partitions (LPAR), which can be detrimental to some I/O adapters, the hypervisor task of the preferred embodiment does not immediately power off a slot upon detecting the slot lock not held. Rather, the hypervisor task records the fact that the slot was found to be a candidate for power off and continues iterating through slots in the system. When the hypervisor task of the preferred embodiment has finished iteration through all I/O slots 1–8, 118A–118H in the system 100, recording those that are candidates for power off, it delays for a short time after which it begins another iteration through the slots. Each slot that is found to be a candidate for power off, that was also recorded as a candidate the previous iteration through the slots, is powered off and the candidate status cleared. When a slot is encountered that was recorded as a candidate for power off during the previous iteration through the slots, the candidate status is cleared regardless of whether the slot is still a candidate or not, so that the next time the slot is found to be a candidate for power off the slot does not appear as though the slot was found to be a candidate during the previous iteration. This ensures the proper delay occurs before the slot is actually powered off so as to not power off a slot that is about to be locked, which could cause slot power cycles in quick succession.
Referring now to
It should be understood that variations of the exemplary steps of
Referring now to
A sequence of program instructions or a logical assembly of one or more interrelated modules defined by the recorded program means 504, 506, 508, 510, direct the computer system 100 for detecting and powering off unused I/O slots of the preferred embodiment.
While the present invention has been described with reference to the details of the embodiments of the invention shown in the drawing, these details are not intended to limit the scope of the invention as claimed in the appended claims.
Number | Name | Date | Kind |
---|---|---|---|
4747041 | Engel et al. | May 1988 | A |
5371867 | George et al. | Dec 1994 | A |
5423046 | Nunnelley et al. | Jun 1995 | A |
5655146 | Baum et al. | Aug 1997 | A |
5708790 | White et al. | Jan 1998 | A |
5721858 | White et al. | Feb 1998 | A |
5875310 | Buckland et al. | Feb 1999 | A |
6243817 | Melo et al. | Jun 2001 | B1 |
6247109 | Kleinsorge et al. | Jun 2001 | B1 |
6279046 | Armstrong et al. | Aug 2001 | B1 |
6332180 | Kauffman et al. | Dec 2001 | B1 |
6381682 | Noel et al. | Apr 2002 | B1 |
6542926 | Zalewski et al. | Apr 2003 | B1 |
6633916 | Kauffman | Oct 2003 | B1 |
6678831 | Mustafa et al. | Jan 2004 | B1 |
6748546 | Mirov et al. | Jun 2004 | B1 |
6760852 | Gulick | Jul 2004 | B1 |
6795930 | Laurenti et al. | Sep 2004 | B1 |
6910142 | Cross et al. | Jun 2005 | B1 |
6959399 | King et al. | Oct 2005 | B1 |
20020161869 | Griffin et al. | Oct 2002 | A1 |
20020161891 | Higuchi et al. | Oct 2002 | A1 |
20030041090 | Armstrong et al. | Feb 2003 | A1 |
20040111596 | Rawson, III | Jun 2004 | A1 |
20040181625 | Armstrong et al. | Sep 2004 | A1 |
20040181657 | Armstrong et al. | Sep 2004 | A1 |
Number | Date | Country |
---|---|---|
0 917 056 | May 1999 | EP |
362217314 | Sep 1987 | JP |
Number | Date | Country | |
---|---|---|---|
20040215985 A1 | Oct 2004 | US |