From mboxrd@z Thu Jan 1 00:00:00 1970 From: mds@paradyne.com (Mark Studebaker) Date: Thu, 19 May 2005 06:23:34 +0000 Subject: Question ... Message-Id: <3D124A33.83C92B8B@paradyne.com> List-Id: References: In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: lm-sensors@vger.kernel.org Almost every sensor chip has working alarm bits that _latch_ the alarm condition until the register is read. This is much better than polling because an intermittent condition will not be missed. So I think the way we do it now is the best way. It may be that you have found a bug, or perhaps something has already been fixed (2.6.0 is one year old) - you can look at our CHANGES document (linked on our download page) to see. If you think you have an alarm bit that is not working then it is probably a bug, or caused by missing documentation. In that case, of course, tell us what bit, what driver, what sensor, what chip. mds "Dr. Ing. Dieter Jurzitza" wrote: > > Dear lm_sensors people, > first of all: many many thanks for the time you spent writing lm_sensors and > making it public available. I am very happy with that tool because it is highly > effective in device control (watching overheat issues with my mb for example). > > One question to you: you write in the docs, that alarm triggers only occur by > the sensor chip. This is nice - if the chip honors the limits you give > (currently I am using a Tyan 2460 MB (amd756 / wXXX)) and want to shut it down > in case of an overtemperature event, kernel 2.4.18 (SuSE), lm_sensors version > 2.6.0). This is difficult if there is no alarm trigger even though the condition > for a trigger is met (i.e. temperature above the limiting value). > > Two possibilities: > 1.) tell me upgrade here and there and things (may) get better, > 2.) do some awk magic (this is what I did) and check for the actual reading to > be larger than some value XY. > > However, I have to run (and have) awk every other minute to ensure apropriate > operation. Would it be bad to have an /etc/sensors.conf option that > allows lm_sensors itself to interpret the limits given and say "alarm" if > an "alarm" condition is met? According to my understanding there must be > a program internal structure containing those values. Or even allow a user > command to be executed as soon as some limiting values cross a certain border > (shutdown -h now for example)? > > Please give me your opinion on that (and again, don't take me wrong, I am > *very* happy with this software and to have a chance to build a thermal > supervision on my own, as long as ACPI is still in its early stages. > Take care, > > Dieter Jurzitza > > -- > ----------------------------------------------------------- > E-Mail: Dr. Ing. Dieter Jurzitza > Date: 18-Jun-2002 > Time: 21:02:55 | > \ > /\_/\ | > | ~x~ |/-----\ / > \ /- \_/ > ^^__ _ / _ ____ / > || || _| _| _| _| > > if you really want to see the pictures above - use some font > with constant spacing like courier! :-) > -----------------------------------------------------------