Re: 2.6.6-rc3-mm1

From: Pasi Savolainen
Date: Sat May 01 2004 - 20:13:22 EST


* Andrew Morton <akpm@xxxxxxxx>:
> ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.6-rc3/2.6.6-rc3-mm1/

I'm having severe interactivity problems with 2.6 tree on a dual Athlon system.
I mostly get 'X screen freezes/mouse pointer immovable for a several
seconds' and rather often audio skips.

I can't really pinpoint it on some activity, but I can trigger this easily
just by switching working desktops. Applications I use are not really
anything special (sorted by virtual memory usage):

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
1352 root 15 0 144m 43m 65m S 1.3 8.6 109:24.39 XFree86
31502 pvsavola 15 0 142m 32m 66m S 0.3 6.4 1:58.98 opera
31707 pvsavola 15 0 101m 29m 30m S 0.0 5.9 3:05.41 firefox-bin
31040 pvsavola 15 0 63216 8572 20m S 0.7 1.7 0:28.09 xmms
31805 pvsavola 15 0 42012 14m 21m S 0.0 2.9 0:16.30 straw
4466 pvsavola 15 0 39800 34m 13m S 0.0 6.9 0:04.01 gimp-2.0
31508 privoxy 16 0 36180 896 1904 S 0.0 0.2 0:00.09 privoxy

Most degenerate case comes however when _closing_ some application,
especially if it has been dropped into swap. Below is 'vmstat 1' of my
shutting down (swapped-out) GIMP2.0 after I load some other memory-hungry
applications. 'freeze' and 'unfreeze' mark points at which vmstats
output really comes onto terminal.

-I really pushed 'close' on GIMP half-a minute earlier-
procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
timestamp r b swpd free buff cache si so bi bo in cs us sy id wa
03:27:39 0 2 381196 4416 320 21836 1744 0 1768 0 1615 2798 3 3 37 58
03:27:40 0 2 381196 3120 328 21896 1224 0 1240 44 1654 2812 3 3 50 43
03:27:41 0 2 381192 4276 296 22844 116 0 2312 4 1778 2907 4 4 32 60
03:27:42 0 1 381024 3584 304 23288 372 0 912 16 1668 2707 10 3 32 56
03:27:43 0 1 381024 3904 292 23288 896 0 920 4 1651 1244 3 3 49 45
03:27:44 0 2 381024 4112 296 23492 724 0 848 0 1628 1069 3 2 32 64
03:27:45 0 2 381024 3248 308 23476 548 0 632 12 1639 1391 4 2 34 61
03:27:46 0 2 381020 3952 312 24552 56 0 2312 28 1737 2993 5 5 25 66
03:27:47 1 3 381020 2976 316 25076 356 0 1004 12 1749 2756 5 4 14 77
03:27:48 1 2 380308 3864 324 25164 1244 0 1276 4 1684 2821 5 4 38 55
03:27:49 0 3 380116 3224 324 25160 1312 0 1356 0 1659 2561 10 3 38 50
03:27:50 0 3 380116 3408 332 24408 880 0 904 40 1656 1106 4 1 28 66
03:27:51 1 2 380116 3104 388 24420 448 0 968 4 1639 2078 4 3 34 60
03:27:52 0 2 378492 3400 424 24888 640 0 1376 16 1719 2773 8 5 27 60
03:27:53 0 3 378484 4112 400 24296 1304 0 1328 0 1686 1626 5 2 34 60
03:27:54 0 4 378484 3224 404 24232 1536 0 1624 0 1693 1006 4 2 40 54
03:27:55 1 0 378484 3680 428 23624 1048 0 1116 0 1688 2772 9 3 48 39
03:27:56 procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
03:27:56 r b swpd free buff cache si so bi bo in cs us sy id wa
03:27:56 1 3 378484 3896 404 23680 200 0 1364 436 1731 2914 11 5 24 60
03:27:57 0 2 378484 3272 452 23816 88 0 604 0 1640 2810 6 3 45 46
-freeze-
03:27:58 0 2 258376 106600 456 24128 216 0 1188 4 1637 2311 4 5 40 51
03:27:59 0 2 258376 105400 464 24156 1120 0 1120 48 1702 1018 1 2 50 47
03:28:00 0 1 258360 104776 464 23384 1188 0 1188 0 1675 880 0 2 50 49
03:28:01 0 2 258360 103432 472 23360 1240 0 1240 16 1707 940 0 1 25 73
03:28:02 0 1 258360 102216 472 23392 988 0 988 0 1675 748 0 1 50 49
03:28:03 0 1 258360 101132 472 23368 908 0 908 0 1667 706 1 1 50 49
03:28:04 0 1 258360 99920 472 23320 1136 0 1136 0 1637 744 1 1 50 49
03:28:05 0 1 258360 98752 472 23312 1028 0 1028 0 1635 743 0 2 50 49
03:28:06 0 1 258360 97280 488 23360 1228 0 1228 28 1703 822 0 1 26 73
03:28:07 0 1 258360 95936 488 23324 1124 0 1124 0 1669 741 1 2 50 49
03:28:08 0 1 258360 94784 488 23336 1008 0 1008 0 1672 726 0 1 50 49
03:28:09 0 1 258360 93616 488 23340 1220 0 1220 0 1677 726 0 0 50 49
03:28:10 0 1 258360 92464 488 23332 1096 0 1096 0 1673 736 0 2 50 48
03:28:11 0 2 258360 91376 492 23320 960 0 960 12 1649 691 0 1 33 65
03:28:12 0 2 258360 90544 504 23340 716 0 716 16 1635 741 0 1 16 83
03:28:13 0 1 258360 89432 504 23312 1048 0 1048 0 1631 585 0 2 50 47
03:28:14 0 1 258360 88408 504 23276 988 0 988 0 1609 616 0 1 50 49
03:28:15 0 4 258360 88860 524 23872 472 0 856 0 1608 800 8 2 33 57
03:28:16 0 4 258360 84428 552 28124 208 0 4700 12 1606 593 1 2 49 48
03:28:17 procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
03:28:17 r b swpd free buff cache si so bi bo in cs us sy id wa
03:28:17 1 1 258360 82764 560 28812 1004 0 1692 4 1735 1736 5 2 34 59
03:28:18 0 0 258360 82508 568 28872 0 0 0 52 1560 2844 5 2 90 2
-unfreeze-
03:28:19 1 0 258360 82516 568 28872 0 0 0 0 1553 2770 5 3 93 0
03:28:20 0 0 258360 82516 568 28872 0 0 0 0 1550 2793 5 2 93 0
03:28:21 0 0 258360 82604 576 28864 0 0 8 0 1458 2635 4 2 94 1
03:28:22 0 0 258360 82604 584 28856 0 0 0 16 1551 2674 4 2 93 0


I've also wondered a bit about slabtop, is this really OK for half-a-day
uptime and not really changing since?:

OBJS ACTIVE USE OBJ SIZE SLABS OBJ/SLAB CACHE SIZE NAME
348220 348135 99% 0.19K 17411 20 69644K dentry_cache
313138 313049 99% 0.50K 44734 7 178936K ext3_inode_cache
10530 7354 69% 0.05K 135 78 540K buffer_head


This machine is rather standard 2xAthlon/512MB, 1G swap/UDMA2 disk.

Is there anything I can do to help solve this?


Thanks.
--
Psi -- <http://www.iki.fi/pasi.savolainen>

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/