Doubling of 3.1's Running Time for i686-pc-linux-gnu

[Jeffrey Oldham <oldham at codesourcery dot com>]

My nightly gcc 3.1 builds and tests now require about double the
amount of time compared with those on 19Jul or before.

start times are all 1:01am
ending times for bootstrap and testing:
    Jul  1 03:59
    Jul  2 04:08
    Jul  3 03:46
    Jul  4 03:49
    Jul  5 03:41
    Jul  6 03:43
    Jul  7 03:30
    Jul  8 03:42
    Jul  9 02:47
    Jul 10 01:03			failed to build
    Jul 11 01:42			failed to build
    Jul 12 01:26			failed to build
    Jul 13 03:55
    Jul 14 03:57
    Jul 15 04:01
    Jul 16 04:04
    Jul 17 02:58
    Jul 18 04:23
    Jul 19 04:22
    Jul 20 09:36	double the normal
    Jul 21 05:43	higher than normal
    Jul 22 01:29			failed to build
    Jul 23 06:03	higher than normal
    Jul 24 09:28	double the normal
    Jul 25 09:34	double the normal

Is this expected?

Thanks,
Jeffrey D. Oldham
oldham@codesourcery.com