On Tuesday, November 20, 2012 08:01:40 AM Ben Hutchings wrote: > RCX: f7ffc9000052c1d8 > > Current 64-bit x86 processors really only use 48-bit virtual addresses, > which must begin with 0000 or ffff. So this address seems to have been > corrupted: a single bit has changed. > > This could be caused by a software bug but it looks much more like a > hardware fault. You may be able to confirm a RAM fault with > memtest86+. Unfortunately it doesn't reliably find faults, and it > doesn't particularly stress the CPU or other parts of the system that > could also be at fault.
Ran memtst+ for 6 hours on the two 4GiB sticks. It found two problems. Tst Pass Fail Addr Location Good Bad Err Count 5 3 0002A453310 676.3MB ffffffe7 fffffff7 00000010 1 5 3 0002E4532f0 740.3MB ffffffe7 fffffff7 00000010 2 At least it's GSkill memory with lifetime warranty. Could this single bit error somehow correlate with the address in RCX? (If it's there, I don't recognize it.) -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/201211202240.29662.neal.p.mur...@alum.wpi.edu
