Tejun Heo wrote: > On 02/09/2010 08:16 AM, Rafael J. Wysocki wrote: > >> On Tuesday 09 February 2010, Bill Davidsen wrote: >> >>> Pretty simple to reproduce, boot, suspend, press shift >>> >>> Acer Aspire 1681, Celeron, 2.6.33-rc6. Trace and config attached. >>> >> This looks suspicious: >> >> Feb 8 17:32:47 roadwarrior3 kernel: ata2.00: ACPI cmd 03/42:00:00:00:a0:ef (CFA REQUEST EXTENDED ERROR) rejected by device (Stat=0x51 Err=0x04) >> Feb 8 17:32:47 roadwarrior3 kernel: sd 0:0:0:0: [sda] Starting disk >> [snip] > These aren't harmful in themselves. It just means that the device > failed commands BIOS requested via ACPI. Just in case, does > "libata.noacpi=1" make any difference to the oops? > I tried that, most interesting result. The kernel still crashes but then starts removing modules at random, re-crashing and declaring itself tainted, and finally gets up (for some definition of up) enough to issue shutdown. It then crashes its way to a halt. I have attached the first BUG here, but the whole log, boot to shutdown is pretty huge to inflict on people, so I put it up for download on the theory that it is unlikely to be helpful or interesting, just big. At this point I'm going to build 2.6.33-rc{latest} and try again, I hoped this could be useful quickly, so I'll use the latest in the hope that it might show the problem fixed. The whole log is at http://www.tmr.com/~davidsen/Private/noacpi_crash.trace.bz2 -- Bill Davidsen "We can't solve today's problems by using the same thinking we used in creating them." - Einstein