Red Hat Bugzilla – Bug 42030
SPM needs APM completely absent in compiled kernel, on some motherboards.
Last modified: 2007-04-18 12:33:23 EDT
From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux 2.4.4-xfs i686; en-US; 0.7) Gecko/20010316 Description of problem: On some SMP motherboards (ABit BP6 is the one I have), when using an SMP kernel, Power Management apparently needs to be disabled in the kernel config file. I am still doing some other testing to verify that this is the only solution, but it is the first thing that has worked so far. Symptoms include: -- Console doesn't work after finishing boot -- Random failure of network or disk devices w/large FTP transfers. How reproducible: Always Steps to Reproduce: 1.Boot standard RH 7.1 SMP kernel on ABit BP6 w/2 processors. 2.FTP very large file over network to this machine (my test file was ~2GB, 8.5GB w/XFS kernel which had identical problem to stock RH 7.1 SMP kernel). 3.notice network or disk hang at some point in transfer. For the 8.5GB file (which is too big for ext2fs), I would always get a failure before the end of the file. For ext2fs file ~2GB, I might need 2-3 transfers to get it to hang the network or disk. Occasionally it would lock up the whole machine. Also, I noticed that the text console, after the boot sequence finishes and it's switched to multiple console mode, would not respond to keyboard to mouse input. This would go away by starting X in some fashion. Then keyboard and mouse input, even on the text consoles, works fine. I thought this was a 2.4.X kernel problem til I got rid of APM and it worked fine afterward. Actual Results: Network or disk hung. Console unresponsive to keyboard or mouse once multi-console mode after boot sequence finishes turns on (i.e. the gettys are active). Again, this could be worked around by starting X in some way, but completely removing APM from the kernel config fixed it completely. Additional info: I just think this board has a buggy APM BIOS anyway (as the console problem was present on the UP kernel w/APM enabled), but in general APM and SMP don't mix too well. It would be better to completely remove APM from the default SPM config file.
APM is disabled at runtime by the kernel, but enabling it allows you to poweroff the machine at shutdown. Re the BP6. I have one at home, and it works well with our default kernel. However, there has been a batch of BP6 boards that are "different" (some call it "less stable") and require you to pass "noapic" to the kernel on the lilo prompt. Could you try this ?