Technique to automatically notify an operating system level...

Details Technique to automatically notify an operating system level... Technique to automatically notify an operating system level...

1999-04-08

2002-10-08

Lefkowitz, Sumati (Department: 2181)

Electrical computers and digital data processing systems: input/

Interrupt processing

Reexamination Certificate

active

06463492

ABSTRACT:

BACKGROUND
The invention relates generally to computer system management and, more particularly, to the exchange of information between an operating system and an interrupt processing routine.
In many computer systems, management activities such as those associated with temperature, voltage, and front panel button monitoring are performed by routines that operate transparently to the operating system. Referring to
FIG. 1
, for example, when a system management event occurs a system management interrupt (SMI) is generated that causes the operating system to transfer control to a SMI handler (block
100
). Once invoked, the SMI handler determines what event occurred and takes the appropriate action (block
102
). As part of the handler's operation, it may modify one or more hardware registers and/or system memory locations to indicate the occurrence of the event. Finally, the SMI handler transfers control back to the operating system which restarts (i.e., begins executing) that program/application that was executing at the time the SMI occurred (block
104
).
During execution of a SMI handler, neither the operating system or any application executing under it has any knowledge of what system management event occurred or what actions the responding handler has taken. Thus, to obtain information about the occurrence of system management events applications such as LANDesk® (from Intel Corporation of California) and Unicenter TNG® (from Computer Associates International of New York) may poll the operating system as shown in FIG.
2
. On expiration of an application specific timer (block
200
) the system management application initiates a query that, eventually, examines the hardware registers and/or system memory locations that the SMI handler modified (block
202
). If the SMI handler modifications indicate a change relevant to the initiating application (the ‘yes’ prong of diamond
204
), this information is returned to the querying application (block
206
) and processing continues (block
208
). If the SMI handler modifications do not indicate a change relevant to the initiating application (the ‘no’ prong of diamond
204
), normal processing continues (block
208
).
Disadvantages of polling techniques include wasted processor time and a lack of event to application-action synchronization. Processor time is wasted every time a polling operation (i.e., acts in accordance with
FIG. 2
) is performed when a system management event has not occurred. Synchronization between the occurrence of a system management event and an application taking action in response to that event is lost because of the granularity of the associated event timer. In prior art systems, optimization of one of these problems tends to exacerbate the other. For example, to improve synchronization an application's timer must be very short. This, however, generates more application queries or polling cycles which, in turn, use more processor time. Alternatively, to reduce processor use, an application's timer may be set to a larger value. A larger timer value, however, adds to the aforementioned synchronization problem. Thus, it would be beneficial to provide techniques to improve system management event processing.
SUMMARY
The invention provides a mechanism by which an operating system level application/program is notified of a system management event. The invention conserves computer system processor time by avoiding the use of polling techniques and provides synchronization between the occurrence of a system management event and the detection of that event by the operating system level application.
In one embodiment, the invention provides a method that includes setting a value (while executing in the system management mode) that is visible to the operating system before returning control to the operating system. In one embodiment, the operating system visible value may cause an interrupt service routine to execute. The interrupt service routine may determine various characteristics of the system management event and pass this information to the operating system level application.


REFERENCES:
patent: 5357628 (1994-10-01), Yuen
patent: 5708819 (1998-01-01), Dunnihoo
patent: 5974573 (1999-10-01), Martin
patent: 6055643 (2000-04-01), Chaiken
patent: 6070253 (2000-05-01), Tavallaei et al.
patent: 6219742 (2001-04-01), Stanley
patent: 2001/0007117 (2001-07-01), Cooper et al.

Affiliated with

Also associated with

Rating

Technique to automatically notify an operating system level... does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Technique to automatically notify an operating system level..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Technique to automatically notify an operating system level... will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2966171