On Wed, 2003-09-24 at 10:23, Luc Van Oostenryck wrote: Like what, "kernel: hey, your expensive new memory is defective"? I wish, then we'd know what the problem was. :P Certainly it shows nothing userspace-segfault-related. Here's a diff between the kernel boot messages for a regular boot and a boot with the new memory. The differences are mostly a) more memory and b) slight differences in timing that you'll see on every boot. 11,12c11,12 < pliny kernel: BIOS-e820: 0000000000100000 - 000000007ff80000 (usable) < pliny kernel: BIOS-e820: 000000007ff80000 - 0000000080000000 (reserved) --- 16c16 < pliny kernel: 1151MB HIGHMEM available. --- 23c23 < pliny kernel: On node 0 totalpages: 524160 --- 26c26 < pliny kernel: zone(2): 294784 pages. --- 36c36 < pliny kernel: Detected 1533.402 MHz processor. --- 39c39 < pliny kernel: Memory: 2068800k/2096640k available (1559k kernel code, 27452k reserved, 646k data, 112k init, 1179136k highmem) --- 41,43c41,43 < pliny kernel: Inode cache hash table entries: 131072 (order: 8, 1048576 bytes) < pliny kernel: Mount-cache hash table entries: 32768 (order: 6, 262144 bytes) < pliny kernel: Buffer-cache hash table entries: 131072 (order: 7, 524288 bytes) --- 68c68 < pliny kernel: Calibrating delay loop... 3060.53 BogoMIPS --- 73c73 < pliny kernel: Total of 2 processors activated (6121.06 BogoMIPS). --- 83,88c83,88 < pliny kernel: ..... CPU clock speed is 1533.3984 MHz. < pliny kernel: ..... host bus clock speed is 266.6779 MHz. < pliny kernel: cpu: 0, clocks: 2666779, slice: 888926 < pliny kernel: CPU0<T0:2666768,T1:1777840,D:2,S:888926,C:2666779> < pliny kernel: cpu: 1, clocks: 2666779, slice: 888926 < pliny kernel: CPU1<T0:2666768,T1:888912,D:4,S:888926,C:2666779> --- 115a116,117 133c135 < pliny kernel: blk: queue f7b96018, I/O limit 4095Mb (mask 0xffffffff) --- 164c166 < pliny kernel: blk: queue f7b59018, I/O limit 4095Mb (mask 0xffffffff) --- 167c169 < pliny kernel: blk: queue f7b58018, I/O limit 4095Mb (mask 0xffffffff) --- 183c185 < pliny kernel: IP: routing cache hash table of 16384 buckets, 128Kbytes --- 207a210 -- brion vibber (brion @ pobox.com)

On Wed, 2003-09-24 at 12:15, Luc Van Oostenryck wrote: That's on... Well, I'm nowhere near the place so all I can do is log in if the machine is up or call Jason and complain while he's trying to get it back online. ;) We don't have any other machines that can take those chips right now, so until the other parts come the only way to test them would be to take pliny offline again, put the chips in, and run memtest86 until either it dies or we're satisfied that there's not a problem. When the new parts are in, we can instead test them in larousse, which can be taken offline without interrupting wiki access too badly, since pliny can take over larousse's job but not vice-versa. -- brion vibber (brion @ pobox.com)