Hi Jean, > -----Original Message----- > From: Jean Delvare [mailto:khali at linux-fr.org] > Sent: 23 April 2006 19:40 > To: Roger Lucas > Cc: LM Sensors > Subject: Re: Kernel fault from ACPI when using W83627EHG on > 2.6.15 > > Hi Roger, > > When starting a new thread, please do NOT use the "reply" function of > your e-mail client on a random post. This confuses users of threading > e-mail clients a lot. I thought that I had started the thread with a new subject line. Is there something else that I am missing that happens behind-the-scenes? > > > Although booting with "acpi=off" is a workaround, I would _really_ like > to > > have ACPI running as it means that the power button on the front of the > box > > does a controlled shutdown (rather than a hard power-off) if pressed > briefly > > as well as a few other nice features. > > The various ACPI features can be built modular. You could try this, and > only load the parts you need (e.g. ac, button) and not the ones which > are probably responsible for the crash (e.g. thermal). This might help > you survive for now, and isolate the problem within acpi. > > And this definitely needs to be reported to the acpi people (bugzilla?) I've raised the problem on the mailing list: http://permalink.gmane.org/gmane.linux.acpi.devel/17523 I still want to know what is really going on though. This ACPI area is not one I am familiar with, and although I have tried reading the documentation, much of it is old and much more seems to indicate that a lot is manufacturer-specific. I have a few work-arounds for this problem, but I'm not comfortable using them until I know what is actually going on. Disabling ACPI completely resolves the problem (but I lose ACPI). Removing the "thermal" module in ACPI stops the kernel dump, but the machine still runs very slowly until the over-temperature situation clears. This may be correct behaviour (e.g. the processor is run at a much lower clock speed to help cool it down) but I don't know for sure. If I just set a very high cpu temperature limit (e.g. 95C) then this also "fixes" the problem, but it is really just a hack rather than a real solution. Hopefully, the guys on the ACPI mailing list can help. This doesn't seem to be a problem with lm-sensors (as far as I can see). > > -- > Jean Delvare
