* 2.5.70-mm1 Strangeness
@ 2003-05-29 20:16 Bongani Hlope
[not found] ` <20030529135541.7c926896.akpm@digeo.com>
0 siblings, 1 reply; 8+ messages in thread
From: Bongani Hlope @ 2003-05-29 20:16 UTC (permalink / raw)
To: akpm; +Cc: linux-kernel
[-- Attachment #1: Type: text/plain, Size: 23689 bytes --]
Hi Andrew
I'm experiencing a strange behaviour with 2.5.70-mm1 on my work PC,
Pentium IV 1.6GHZ with 512MB memory. Let me know if you need any more input.
After login in to KDE /proc/meminfo
MemTotal: 514864 kB
MemFree: 16308 kB
Buffers: 10780 kB
Cached: 86268 kB
SwapCached: 0 kB
Active: 122096 kB
Inactive: 31680 kB
HighTotal: 0 kB
HighFree: 0 kB
LowTotal: 514864 kB
LowFree: 16308 kB
SwapTotal: 409208 kB
SwapFree: 409208 kB
Dirty: 4604 kB
Writeback: 0 kB
Mapped: 93668 kB
Slab: 340540 kB
Committed_AS: 83492 kB
PageTables: 1148 kB
VmallocTotal: 516024 kB
VmallocUsed: 33644 kB
VmallocChunk: 482272 kB
Too much memory has gone to Slab. cat /proc/slabinfo follows bellow.
After a while the kernel hangs, and after rebooting my /var/log/messages
has this as the last output:
May 29 15:10:06 bongani1 last message repeated 689 times
May 29 15:10:06 bongani1 kernel: e:0x20
May 29 15:10:06 bongani1 kernel: lisa: page allocation failure. order:0, mode:0x20
I was running a couple of applications Mozilla-mail, KDE, Some Java Application Server ...
When I load the a application, some memory is freed from the Slabs. The PC was just
sitting there both times running those applications and both times I returned and found it
hanging.
cat /proc/slabinfo looks like this
slabinfo - version: 2.0 (statistics)
# name active_objs num_objs objsize objperslab pagesperslab
#! tunables batchcount limit sharedfactor
#! slabdata active_slabs num_slabs sharedavail
#! globalstat listallocs maxobjs grown reaped error maxfreeable freelimit
#! cpustat N allochit allocmiss freehit freemiss
unix_sock 125 128 4096 1 1
! tunables 24 12 0
! slabdata 125 128 0
! globalstat 193 133 165 37 0 0 25
! cpustat 0 2699 186 2760 0
ip_mrt_cache 0 0 88 43 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 75
tcp_tw_bucket 2 29 132 29 1
! tunables 32 16 0
! slabdata 1 1 0
! globalstat 48 18 2 1 0 0 61
! cpustat 0 3 3 4 0
tcp_bind_bucket 11 123 28 123 1
! tunables 32 16 0
! slabdata 1 1 0
! globalstat 80 27 1 0 0 0 155
! cpustat 0 10 5 4 0
tcp_open_request 0 0 96 40 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 48 16 3 3 0 0 72
! cpustat 0 3 3 6 0
inet_peer_cache 6 71 52 71 1
! tunables 32 16 0
! slabdata 1 1 0
! globalstat 80 21 1 0 0 0 103
! cpustat 0 1 5 0 0
secpath4_cache 0 0 136 28 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 60
xfrm_dst_cache 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
ip_fib_hash 10 123 28 123 1
! tunables 32 16 0
! slabdata 1 1 0
! globalstat 32 25 1 0 0 0 155
! cpustat 0 17 2 9 0
ip_dst_cache 19470 19470 4096 1 1
! tunables 24 12 0
! slabdata 19470 19470 0
! globalstat 69869 32578 69655 220 0 25 25
! cpustat 0 288 69681 46319 4181
arp_cache 211 459 140 27 1
! tunables 32 16 0
! slabdata 17 17 0
! globalstat 544 513 19 0 0 0 59
! cpustat 0 475 41 295 10
raw4_sock 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
udp_sock 10 10 4096 1 1
! tunables 24 12 0
! slabdata 10 10 0
! globalstat 15 12 13 3 0 0 25
! cpustat 0 18 15 23 0
tcp_sock 15 15 4096 1 1
! tunables 24 12 0
! slabdata 15 15 0
! globalstat 50 18 41 26 0 0 25
! cpustat 0 1899 50 1934 0
flow_cache 0 0 104 36 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 68
uhci_urb_priv 0 0 72 53 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 85
udf_inode_cache 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
smb_request 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
smb_inode_cache 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
devfsd_event 0 0 32 112 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 270 128 4 3 0 0 144
! cpustat 0 608 17 615 10
isofs_inode_cache 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
fat_inode_cache 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
ext2_inode_cache 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 7 7 7 7 0 0 25
! cpustat 0 0 7 7 0
ext2_xattr 0 0 56 66 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 98
journal_handle 9 90 40 90 1
! tunables 32 16 0
! slabdata 1 1 0
! globalstat 144 16 5 4 0 0 122
! cpustat 0 33806 9 33815 0
journal_head 1240 1647 60 61 1
! tunables 32 16 0
! slabdata 27 27 0
! globalstat 10764 1756 31 0 0 1 93
! cpustat 0 11463 679 10348 578
revoke_table 2 144 24 144 1
! tunables 32 16 0
! slabdata 1 1 0
! globalstat 16 16 1 0 0 0 176
! cpustat 0 1 1 0 0
revoke_record 0 0 28 123 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 224 89 5 5 0 0 155
! cpustat 0 177 14 184 7
ext3_inode_cache 26178 26179 4096 1 1
! tunables 24 12 0
! slabdata 26178 26179 0
! globalstat 29765 27105 29416 21 0 25 25
! cpustat 0 1218 29457 4199 298
ext3_xattr 0 0 56 66 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 98
dquot 0 0 116 33 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 65
eventpoll_pwq 0 0 48 77 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 109
eventpoll_epi 0 0 84 45 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 77
async_poll 0 0 44 84 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 116
kioctx 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
kiocb 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
dnotify_cache 44 224 32 112 1
! tunables 32 16 0
! slabdata 2 2 0
! globalstat 352 218 2 0 0 0 144
! cpustat 0 330 22 298 16
file_lock_cache 12 36 104 36 1
! tunables 32 16 0
! slabdata 1 1 0
! globalstat 208 31 1 0 0 0 68
! cpustat 0 2066 13 2067 0
fasync_cache 1 123 28 123 1
! tunables 32 16 0
! slabdata 1 1 0
! globalstat 64 16 4 3 0 0 155
! cpustat 0 0 4 3 0
shmem_inode_cache 2 2 4096 1 1
! tunables 24 12 0
! slabdata 2 2 0
! globalstat 10 4 8 6 0 0 25
! cpustat 0 12 9 19 0
idr_layer_cache 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
posix_timers_cache 0 0 88 43 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 75
uid_cache 5 99 36 99 1
! tunables 32 16 0
! slabdata 1 1 0
! globalstat 32 20 1 0 0 0 131
! cpustat 0 3 2 0 0
cfq_pool 64 84 44 84 1
! tunables 32 16 0
! slabdata 1 1 0
! globalstat 64 64 1 0 0 0 116
! cpustat 0 60 4 0 0
crq_pool 0 0 48 77 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 109
as_arq 39 53 72 53 1
! tunables 32 16 0
! slabdata 1 1 0
! globalstat 984 159 6 3 0 1 85
! cpustat 0 17326 65 17327 44
deadline_drq 0 0 60 61 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 93
blkdev_requests 26 26 4096 1 1
! tunables 24 12 0
! slabdata 26 26 0
! globalstat 881 157 547 131 0 25 25
! cpustat 0 16804 587 17307 64
biovec-BIO_MAX_PAGES 256 256 4096 1 1
! tunables 24 12 0
! slabdata 256 256 0
! globalstat 438 306 370 91 0 25 25
! cpustat 0 116 380 236 4
biovec-128 256 256 4096 1 1
! tunables 24 12 0
! slabdata 256 256 0
! globalstat 365 276 325 69 0 0 25
! cpustat 0 200 333 277 0
biovec-64 256 256 4096 1 1
! tunables 24 12 0
! slabdata 256 256 0
! globalstat 397 285 365 109 0 12 25
! cpustat 0 437 371 549 3
biovec-16 258 260 4096 1 1
! tunables 24 12 0
! slabdata 258 260 0
! globalstat 336 267 311 51 0 0 25
! cpustat 0 1118 317 1179 0
biovec-4 256 305 60 61 1
! tunables 32 16 0
! slabdata 5 5 0
! globalstat 420 275 5 0 0 0 93
! cpustat 0 1339 27 1110 0
biovec-1 274 432 24 144 1
! tunables 32 16 0
! slabdata 3 3 0
! globalstat 9310 864 31 3 0 1 176
! cpustat 0 23648 585 23431 546
bio 274 392 68 56 1
! tunables 32 16 0
! slabdata 7 7 0
! globalstat 9998 856 105 4 0 1 88
! cpustat 0 29127 672 28954 589
sock_inode_cache 160 160 4096 1 1
! tunables 24 12 0
! slabdata 160 160 0
! globalstat 168 160 165 5 0 0 25
! cpustat 0 3803 168 3817 0
skbuff_head_cache 626 651 4096 1 1
! tunables 24 12 0
! slabdata 626 651 0
! globalstat 8141 874 5173 97 0 25 25
! cpustat 0 223010 5476 227246 626
sock 3 3 4096 1 1
! tunables 24 12 0
! slabdata 3 3 0
! globalstat 4 4 4 1 0 0 25
! cpustat 0 10 4 11 0
proc_inode_cache 230 230 4096 1 1
! tunables 24 12 0
! slabdata 230 230 0
! globalstat 370 230 369 61 0 25 25
! cpustat 0 10687 370 10821 9
sigqueue 17 42 4096 1 1
! tunables 24 12 0
! slabdata 17 42 0
! globalstat 213 209 213 4 0 25 25
! cpustat 0 11150 213 11347 16
radix_tree_node 2271 2271 4096 1 1
! tunables 24 12 0
! slabdata 2271 2271 0
! globalstat 2757 2271 2510 20 0 25 25
! cpustat 0 626 2537 854 38
bdev_cache 6 40 96 40 1
! tunables 32 16 0
! slabdata 1 1 0
! globalstat 224 192 5 0 0 0 72
! cpustat 0 198 16 198 10
mnt_cache 18 53 72 53 1
! tunables 32 16 0
! slabdata 1 1 0
! globalstat 32 29 1 0 0 0 85
! cpustat 0 24 2 8 0
inode_cache 2953 2953 4096 1 1
! tunables 24 12 0
! slabdata 2953 2953 0
! globalstat 4237 4137 4197 34 0 25 25
! cpustat 0 1158 4203 2303 105
dentry_cache 25891 25891 4096 1 1
! tunables 24 12 0
! slabdata 25891 25891 0
! globalstat 95071 33059 82069 10 0 26 25
! cpustat 0 38623 83540 90513 5765
filp 1341 1392 132 29 1
! tunables 32 16 0
! slabdata 48 48 0
! globalstat 3821 1437 53 1 0 1 61
! cpustat 0 50750 249 49504 155
names_cache 8 8 4096 1 1
! tunables 24 12 0
! slabdata 8 8 0
! globalstat 9 8 8 0 0 0 25
! cpustat 0 293086 9 293095 0
buffer_head 6327 6588 60 61 1
! tunables 32 16 0
! slabdata 108 108 0
! globalstat 14121 6620 111 1 0 1 93
! cpustat 0 13983 904 8112 467
mm_struct 59 59 4096 1 1
! tunables 24 12 0
! slabdata 59 59 0
! globalstat 95 64 84 25 0 0 25
! cpustat 0 3451 90 3483 0
vm_area_struct 3454 3600 76 50 1
! tunables 32 16 0
! slabdata 72 72 0
! globalstat 35887 3716 92 0 0 1 82
! cpustat 0 108028 2312 104871 2027
fs_cache 72 84 44 84 1
! tunables 32 16 0
! slabdata 1 1 0
! globalstat 224 72 1 0 0 0 116
! cpustat 0 1985 14 1942 0
files_cache 57 57 4096 1 1
! tunables 24 12 0
! slabdata 57 57 0
! globalstat 80 62 75 18 0 0 25
! cpustat 0 1920 79 1942 0
signal_cache 84 132 56 66 1
! tunables 32 16 0
! slabdata 2 2 0
! globalstat 239 84 2 0 0 0 98
! cpustat 0 2004 16 1951 0
sighand_cache 63 63 4096 1 1
! tunables 24 12 0
! slabdata 63 63 0
! globalstat 89 69 81 18 0 0 25
! cpustat 0 1919 85 1941 0
task_struct 71 71 4084 1 1
! tunables 24 12 0
! slabdata 71 71 0
! globalstat 98 77 88 17 0 0 25
! cpustat 0 1927 93 1949 0
pte_chain 14194 14280 128 30 1
! tunables 32 16 0
! slabdata 476 476 0
! globalstat 45675 14446 815 2 0 2 62
! cpustat 0 51527 2975 38358 1957
pgd 59 59 4096 1 1
! tunables 24 12 0
! slabdata 59 59 0
! globalstat 100 64 83 24 0 0 25
! cpustat 0 3452 89 3483 0
size-131072(DMA) 0 0 131072 1 32
! tunables 8 4 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 9
size-131072 0 0 131072 1 32
! tunables 8 4 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 9
size-65536(DMA) 0 0 65536 1 16
! tunables 8 4 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 9
size-65536 1 1 65536 1 16
! tunables 8 4 0
! slabdata 1 1 0
! globalstat 1 1 1 0 0 0 9
! cpustat 0 0 1 0 0
size-32768(DMA) 0 0 32768 1 8
! tunables 8 4 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 9
size-32768 1 1 32768 1 8
! tunables 8 4 0
! slabdata 1 1 0
! globalstat 2 2 2 1 0 0 9
! cpustat 0 1 2 2 0
size-16384(DMA) 0 0 16384 1 4
! tunables 8 4 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 9
size-16384 0 0 16384 1 4
! tunables 8 4 0
! slabdata 0 0 0
! globalstat 19 6 17 17 0 0 9
! cpustat 0 725 19 744 0
size-8192(DMA) 0 0 8192 1 2
! tunables 8 4 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 9
size-8192 11 13 8192 1 2
! tunables 8 4 0
! slabdata 11 13 0
! globalstat 36 13 29 16 0 0 9
! cpustat 0 268 35 292 0
size-4096(DMA) 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
size-4096 60 60 4096 1 1
! tunables 24 12 0
! slabdata 60 60 0
! globalstat 297 132 248 94 0 25 25
! cpustat 0 6288 255 6474 17
size-2048(DMA) 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
size-2048 81 81 4096 1 1
! tunables 24 12 0
! slabdata 81 81 0
! globalstat 103 82 96 15 0 0 25
! cpustat 0 1112 103 1138 0
size-1024(DMA) 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
size-1024 76 76 4096 1 1
! tunables 24 12 0
! slabdata 76 76 0
! globalstat 93 88 88 12 0 12 25
! cpustat 0 11184 91 11217 1
size-512(DMA) 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
size-512 641 654 4096 1 1
! tunables 24 12 0
! slabdata 641 654 0
! globalstat 7999 846 5007 92 0 25 25
! cpustat 0 136047 5307 140123 613
size-256(DMA) 0 0 4096 1 1
! tunables 24 12 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 25
size-256 252 252 4096 1 1
! tunables 24 12 0
! slabdata 252 252 0
! globalstat 255 252 254 2 0 0 25
! cpustat 0 40315 255 40320 0
size-128(DMA) 0 0 140 27 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 59
size-128 82705 94716 140 27 1
! tunables 32 16 0
! slabdata 3508 3508 0
! globalstat 195853 94797 3511 0 0 0 59
! cpustat 0 193572 13350 117172 7061
size-64(DMA) 0 0 76 50 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 82
size-64 248 250 76 50 1
! tunables 32 16 0
! slabdata 5 5 0
! globalstat 279 248 5 0 0 0 82
! cpustat 0 4089 30 3893 0
size-32(DMA) 0 0 44 84 1
! tunables 32 16 0
! slabdata 0 0 0
! globalstat 0 0 0 0 0 0 116
size-32 1324 1344 44 84 1
! tunables 32 16 0
! slabdata 16 16 0
! globalstat 1516 1324 16 0 0 0 116
! cpustat 0 53594 131 52405 12
kmem_cache 112 120 160 24 1
! tunables 32 16 0
! slabdata 5 5 0
! globalstat 112 112 5 0 0 0 56
! cpustat 0 70 33 0 0
[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: 2.5.70-mm1 Strangeness
[not found] ` <20030529135541.7c926896.akpm@digeo.com>
@ 2003-05-30 0:11 ` David S. Miller
2003-05-30 0:51 ` Andrew Morton
0 siblings, 1 reply; 8+ messages in thread
From: David S. Miller @ 2003-05-30 0:11 UTC (permalink / raw)
To: akpm; +Cc: bonganilinux, linux-kernel
From: Andrew Morton <akpm@digeo.com>
Date: Thu, 29 May 2003 13:55:41 -0700
The ip_dst_cache seems unreasonably large. Unless your desktop is a
backbone router or something.
Lots of DST entries can result on any machine actually. We create one
per source address, not just per destination address. So if you talk
to a lot of sites, or lots of sites talk to you, you'll get a lot of
DST entries.
Regardless, 80MB _IS_ excessive. That's nearly 400,000 entries.
It definitely indicates there is a leak somewhere.
Although it say:
ip_dst_cache 19470 19470 4096 1 1
Which is 19470 active objects right?
There is a known aparent issue with IPV6, there is some DST
leak there, but that is irrelevant here since we're clearly
talking about the ipv4 dst cache.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: 2.5.70-mm1 Strangeness
2003-05-30 0:11 ` David S. Miller
@ 2003-05-30 0:51 ` Andrew Morton
2003-05-30 0:58 ` David S. Miller
` (3 more replies)
0 siblings, 4 replies; 8+ messages in thread
From: Andrew Morton @ 2003-05-30 0:51 UTC (permalink / raw)
To: David S. Miller; +Cc: bonganilinux, linux-kernel, Manfred Spraul
"David S. Miller" <davem@redhat.com> wrote:
>
> From: Andrew Morton <akpm@digeo.com>
> Date: Thu, 29 May 2003 13:55:41 -0700
>
> The ip_dst_cache seems unreasonably large. Unless your desktop is a
> backbone router or something.
>
> Lots of DST entries can result on any machine actually. We create one
> per source address, not just per destination address. So if you talk
> to a lot of sites, or lots of sites talk to you, you'll get a lot of
> DST entries.
>
> Regardless, 80MB _IS_ excessive. That's nearly 400,000 entries.
> It definitely indicates there is a leak somewhere.
>
> Although it say:
>
> ip_dst_cache 19470 19470 4096 1 1
>
> Which is 19470 active objects right?
Yes, 19470 entries. But note that each entry is 4096 bytes.
Something seems to have gone and bumped the object size from 240 bytes up
to 4096. This is actually what I want CONFIG_DEBUG_PAGEALLOC to do, but I
don't think it does it yet.
Bongani, if you have CONFIG_DEBUG_PAGEALLOC enabled then please try turning
it off. And maybe Manfred can throw some light on what slab has done
there.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: 2.5.70-mm1 Strangeness
2003-05-30 0:51 ` Andrew Morton
@ 2003-05-30 0:58 ` David S. Miller
2003-05-30 4:12 ` Bongani Hlope
` (2 subsequent siblings)
3 siblings, 0 replies; 8+ messages in thread
From: David S. Miller @ 2003-05-30 0:58 UTC (permalink / raw)
To: akpm; +Cc: bonganilinux, linux-kernel, manfred
From: Andrew Morton <akpm@digeo.com>
Date: Thu, 29 May 2003 17:51:35 -0700
Yes, 19470 entries. But note that each entry is 4096 bytes.
Something seems to have gone and bumped the object size from 240
bytes up to 4096.
I see.
Meanwhile, you can control how large the routing cache allows itself
to grow by changing /proc/sys/net/ipv4/route/max_size. It is the
maximum number of entries the routing cache will allow itself to
expand to.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: 2.5.70-mm1 Strangeness
2003-05-30 0:51 ` Andrew Morton
2003-05-30 0:58 ` David S. Miller
@ 2003-05-30 4:12 ` Bongani Hlope
2003-05-30 5:02 ` Manfred Spraul
2003-05-30 19:25 ` Bongani Hlope
3 siblings, 0 replies; 8+ messages in thread
From: Bongani Hlope @ 2003-05-30 4:12 UTC (permalink / raw)
To: Andrew Morton; +Cc: davem, linux-kernel, manfred, bonganih
[-- Attachment #1: Type: text/plain, Size: 1362 bytes --]
On Thu, 29 May 2003 17:51:35 -0700
Andrew Morton <akpm@digeo.com> wrote:
> "David S. Miller" <davem@redhat.com> wrote:
> >
> > From: Andrew Morton <akpm@digeo.com>
> > Date: Thu, 29 May 2003 13:55:41 -0700
> >
> > The ip_dst_cache seems unreasonably large. Unless your desktop is a
> > backbone router or something.
> >
> > Lots of DST entries can result on any machine actually. We create one
> > per source address, not just per destination address. So if you talk
> > to a lot of sites, or lots of sites talk to you, you'll get a lot of
> > DST entries.
> >
> > Regardless, 80MB _IS_ excessive. That's nearly 400,000 entries.
> > It definitely indicates there is a leak somewhere.
> >
> > Although it say:
> >
> > ip_dst_cache 19470 19470 4096 1 1
> >
> > Which is 19470 active objects right?
>
> Yes, 19470 entries. But note that each entry is 4096 bytes.
>
> Something seems to have gone and bumped the object size from 240 bytes up
> to 4096. This is actually what I want CONFIG_DEBUG_PAGEALLOC to do, but I
> don't think it does it yet.
>
> Bongani, if you have CONFIG_DEBUG_PAGEALLOC enabled then please try turning
> it off. And maybe Manfred can throw some light on what slab has done
> there.
Cool, I'll let you know how it goes. I'll also try davem,s suggestion about
/proc/sys/net/ipv4/route/max_size
[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: 2.5.70-mm1 Strangeness
2003-05-30 0:51 ` Andrew Morton
2003-05-30 0:58 ` David S. Miller
2003-05-30 4:12 ` Bongani Hlope
@ 2003-05-30 5:02 ` Manfred Spraul
2003-05-30 19:25 ` Bongani Hlope
3 siblings, 0 replies; 8+ messages in thread
From: Manfred Spraul @ 2003-05-30 5:02 UTC (permalink / raw)
To: Andrew Morton; +Cc: David S. Miller, bonganilinux, linux-kernel
Andrew Morton wrote:
>Something seems to have gone and bumped the object size from 240 bytes up
>to 4096. This is actually what I want CONFIG_DEBUG_PAGEALLOC to do, but I
>don't think it does it yet.
>
>
Yes, it does it:
> #ifdef CONFIG_DEBUG_PAGEALLOC
> if (size < PAGE_SIZE-3*BYTES_PER_WORD && size > 128)
> size = PAGE_SIZE-3*BYTES_PER_WORD;
> #endif
--
Manfred
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: 2.5.70-mm1 Strangeness
2003-05-30 0:51 ` Andrew Morton
` (2 preceding siblings ...)
2003-05-30 5:02 ` Manfred Spraul
@ 2003-05-30 19:25 ` Bongani Hlope
2003-05-30 19:58 ` Andrew Morton
3 siblings, 1 reply; 8+ messages in thread
From: Bongani Hlope @ 2003-05-30 19:25 UTC (permalink / raw)
To: Andrew Morton; +Cc: davem, linux-kernel, manfred
[-- Attachment #1: Type: text/plain, Size: 2319 bytes --]
On Thu, 29 May 2003 17:51:35 -0700
Andrew Morton <akpm@digeo.com> wrote:
> "David S. Miller" <davem@redhat.com> wrote:
> >
> > From: Andrew Morton <akpm@digeo.com>
> > Date: Thu, 29 May 2003 13:55:41 -0700
> >
> > The ip_dst_cache seems unreasonably large. Unless your desktop is a
> > backbone router or something.
> >
> > Lots of DST entries can result on any machine actually. We create one
> > per source address, not just per destination address. So if you talk
> > to a lot of sites, or lots of sites talk to you, you'll get a lot of
> > DST entries.
> >
> > Regardless, 80MB _IS_ excessive. That's nearly 400,000 entries.
> > It definitely indicates there is a leak somewhere.
> >
> > Although it say:
> >
> > ip_dst_cache 19470 19470 4096 1 1
> >
> > Which is 19470 active objects right?
>
> Yes, 19470 entries. But note that each entry is 4096 bytes.
>
> Something seems to have gone and bumped the object size from 240 bytes up
> to 4096. This is actually what I want CONFIG_DEBUG_PAGEALLOC to do, but I
> don't think it does it yet.
>
> Bongani, if you have CONFIG_DEBUG_PAGEALLOC enabled then please try turning
> it off. And maybe Manfred can throw some light on what slab has done
> there.
>
Turning off CONFIG_DEBUG_PAGEALLOC helped, everything is running smoothly
cat /proc/sys/net/ipv4/route/max_size gave me 65536
The slabs were reclaimed before the PC died. I had a script that checked /proc/slabinfo
every 10 seconds and dup it to a file. The last output was: (for the slabs you asked me to watch)
ip_dst_cache 61 61 4096 1 1
! tunables 24 12 0
! slabdata 61 61 0
! globalstat 342536 53783 340902 2351 0 25 25
! cpustat 0 2231 406871 314930 28356
ext3_inode_cache 4331 4331 4096 1 1
! tunables 24 12 0
! slabdata 4331 4331 0
! globalstat 34388 26451 33205 580 0 25 25
! cpustat 0 2459 33332 28998 2462
dentry_cache 11016 11016 4096 1 1
! tunables 24 12 0
! slabdata 11016 11016 0
! globalstat 109937 29529 94505 418 0 25 25
! cpustat 0 568369 96256 645381 8243
So it seems like CONFIG_DEBUG_PAGEALLOC was causing the problem.
[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: 2.5.70-mm1 Strangeness
2003-05-30 19:25 ` Bongani Hlope
@ 2003-05-30 19:58 ` Andrew Morton
0 siblings, 0 replies; 8+ messages in thread
From: Andrew Morton @ 2003-05-30 19:58 UTC (permalink / raw)
To: Bongani Hlope; +Cc: davem, linux-kernel, manfred
Bongani Hlope <bonganilinux@mweb.co.za> wrote:
>
> So it seems like CONFIG_DEBUG_PAGEALLOC was causing the problem.
hm, okay. Thanks for your help in working all that out.
Everything should have just balanced itself out - nothing in the memory
reclaim area cares or knows about the actual size of the objects.
You could run out of memory simply because the things are 20x bigger, but
only if they were pinned - say, a larger number of files open, or a large
number of files in ramfs or sysfs.
But from your last-gasp slabinfo the three caches seem to be of a
reasonable size. It's a bit of a mystery.
^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2003-05-30 19:45 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-05-29 20:16 2.5.70-mm1 Strangeness Bongani Hlope
[not found] ` <20030529135541.7c926896.akpm@digeo.com>
2003-05-30 0:11 ` David S. Miller
2003-05-30 0:51 ` Andrew Morton
2003-05-30 0:58 ` David S. Miller
2003-05-30 4:12 ` Bongani Hlope
2003-05-30 5:02 ` Manfred Spraul
2003-05-30 19:25 ` Bongani Hlope
2003-05-30 19:58 ` Andrew Morton
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox