I'm about to transfer all my data to a new unRAID Plus system based upon Intel D945GCLF2 with 2GB RAM and 4 SATA drives (2 via onboard SATA controller and 2 via PCI Promise SATA300 TX4). However I am a bit worried about my syslog being flooded by the following error messages.

 

Dec  2 10:11:57 babylon kernel: ACPI Error (psparse-0537): Method parse/execution failed [\_SB_.PCI0.LPC_.SMBR] (Node f7411c48), AE_AML_INFINITE_LOOP
Dec  2 10:11:57 babylon kernel: ACPI Error (psparse-0537): Method parse/execution failed [\_SB_.PCI0.LPC_.INIT] (Node f7411c30), AE_AML_INFINITE_LOOP
Dec  2 10:11:57 babylon kernel: ACPI Error (psparse-0537): Method parse/execution failed [\_GPE._L00] (Node f740e1b0), AE_AML_INFINITE_LOOP
Dec  2 10:11:57 babylon kernel: ACPI Exception: AE_AML_INFINITE_LOOP, while evaluating GPE method [_L00] 20090521 evgpe-568

 

Is this a critical error ? What does it relate to ?

 

You can fetch a full copy of the latest syslog here :

[EDIT] Here is a working direct link: http://natzo.com/files/syslog-2009-12-02.txt

 

Thank you in advance for your suggestions.

Alphazo

 

As far as I know, ACPI errors are corrected by an update to the BIOS, or an update to the kernel, or possibly a change in the hardware components.  Sorry, no good choices here.  You can look for a BIOS update, and hopefully that is all you need.  You already look up-to-date with the kernel release, but perhaps a future one will have either a relevant bug fix or workaround.  I noticed a bit of confusion over your CPU, so it is slightly possible that this BIOS is not fully compatible with your CPU.  If you happen to have another around, of the same generation as this board, that *might* make a difference, but I don't think your chances are great, probably not worth much effort here.

 

Otherwise, you can either turn ACPI off completely (acpi=off), or try the various ACPI-related boot codes, to find one that selectively disables a part of ACPI, but not the whole ACPI subsystem.  See the Boot Codes wiki page, and check the ACPI links at the bottom.  One that might be helpful is pci=noacpi.

 

It could still be some other issue, such as a problem with the motherboard.  I can't think of anything else.  ACPI errors, as reported in the syslog, have usually been harmless, but yours threatens to blow up the syslog, filling all of memory.  The syslog growth looks slow enough that you could probably get by, by rebooting once a week.

My BIOS seems to be the latest one (LF94510J.86A.0229.2009.0729.0209) according to Intel:

http://downloadcenter.intel.com/SearchResult.aspx?lang=eng&ProductFamily=Desktop+Boards&ProductLine=Intel%C2%AE+945+Express+Chipset+Family+Boards&ProductProduct=Intel%C2%AE+Desktop+Board+D945GCLF2

 

I don't recall seeing any ACPI switch in the BIOS but I will look again carefully.

 

I would prefer to have some kind of ACPI feature active like power button and also shutdown via apcupsd. Thank you for the boot options, now I need to figure out which one to use

 

According to Tom there is a workaround but it requires a kernel rebuild. Maybe I can wait until a new beta release.

 

Alphazo

 

[EDIT] This morning I did a cold boot and warm boot (reboot) and the problem didn't show up at all (nothing as changed in the configuration since yesterday). It looks like a pretty random issue.