2.5.56 strange behaviour

[Arador <diegocg _at_ xxxxxxxxxxx>]

Hi, i've observed an "strange" thing in 2.5.56 (and 2.5.55 too)

procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
 r  b   swpd   free  inact active   si   so    bi    bo   in    cs us sy id wa
 0  0      0   6228  92524 134716    0    0     0     0 1698  2257  1  0 98  0
<two "dd if=/dev/zero of=test bs=4M" here>
 3  0      0  26384  79468 131012    0    0    12  8192 1784  2163  1 33 60  5
 1  1      0   4404 115528 119768    0    0     4 43616 2013  3825  2 63 14 22 
 0  4      0   2076 118480 118896    0    0     8 40456 2027  2251  1 44  6 49
 0  3      0   4100 116444 118792    0    0     0 41184 2044  2614  2  5  0 94
 3  2      0   4028 116368 119120    0    0     0 37768 2041  2196  1 53  6 40
 3  2      0  12040 117608 109764    0    0     8 41456 2048  2403  1 55 10 33
 1  3      0   3852 127036 108344    0    0     4 36140 2012  2473  1 53  6 40
 0  5    284   2168 130556 106468    0  284     0 47564 2061  2665  2 44  2 51
 0  3   1884   4064 129808 105288    0 1504     0 42028 2062  2360  1 44 12 44
<one "dd if=/dev/zero of=test bs=4M" here>
 1  4   2424   2228 132204 104732    0  540    16 39768 2091  2508  1 41  2 57
 1  3   3232   4072 130772 104332    0  904    12 43156 2050  2569  1 43  6 49
 1  2   4524   4584 130892 103656    0 1292     4 37680 2031  2608  2 43  6 49
 1  3   4524   5136 130340 103724    0    0     0 47024 2060  2491  1 49  4 46
 4  4   4524   2240 132948 103972    0    0     0 40588 2063  5791  1 48  8 44
<two "dd if=/dev/zero of=test bs=4M" here>
 4  4   5188   4052 130768 104264    0  664     4 41896 2063 199652  1 80  0 18
 3  1   5188   4232 131112 103752    0    0    44 40172 2156 237791  3 94  0  3
 4  3   5188   3572 132208 103320    0    0    16 44332 2068 252186  1 95  1  4
 2  1   5188   4292 131488 103256    0    0     0 40852 2056 302306  1 81  7 10
 0  3   5188   4192 131528 103248    0    0   340 29672 1974 182712  1 69 14 17
							     ^^^^^^

 1  0   5188  90728  47484 103436   28    0    68 21224 1931  2178  2 64 25  8
 1  2   5188  32432 104000 103492    0    0     0 44924 2047  2200  2 57  9 33
 0  3   5188   7068 128576 103512    0    0     0 35064 2028  2200  1 27 15 58
 1  3   6108   3800 131884 103176    0  920     0 40184 2041  2135  1 32 11 56
 1  3   6108   2188 133672 103028    0    0     0 42048 2037  2351  2 32  5 61
diego _at_ estel:~$

As you see, the cs rate goes from 2??? to 200000-300000.
I wanted to know if it's the expected behaviour.

The test file is the same in all the dd's.


If you need info:

The machine is a P3 2x800 with 256 ram; UDMA 100 ide disk;
FS is ext3 (no htree or the orlov thing).
everything is in the same disk. While it does that;
it's doing nothing except X+gnome (in the background)
playing ogg (which increases the cs)
I reproduced it without X/ogg playing.

I've managed to save slabinfo attached (if it helps...)

I observed it in 2.5.55 for the first time; it was not easy
to reproduce but it happened. Strangely it happened
the first time this afternoon after a mail/web/irc idling session.

Attachment: slabinfo
Description: Binary data