public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* [2.6.7, ia64] continual memory leak at ~102kB/s...
@ 2004-08-06 14:44 Daniel Blueman
  2004-08-06 14:54 ` William Lee Irwin III
  0 siblings, 1 reply; 3+ messages in thread
From: Daniel Blueman @ 2004-08-06 14:44 UTC (permalink / raw)
  To: linux-ia64, davidm, linux-kernel

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="us-ascii", Size: 3854 bytes --]

When running 2.6.7 on a generic ia64 system, I see memory being leaked in
the kernel. Most of the fancy (preempt, hot-plug procs, ...) features are
disabled, and the system in a quiescent state [1].

/proc/meminfo shows the memory as unaccounted for [2], so it seems likely it
has been kmalloc()d somehere. A small script shows memory disappearing at
102kB/s [3].

Anyone else seen this on ia64?

--- [1]

# ps -ef
UID        PID  PPID  C STIME TTY          TIME CMD
root         1     0  1 10:29 ?        00:00:04 init [S]              
root         2     1  0 10:29 ?        00:00:00 [migration/0]
root         3     1  0 10:29 ?        00:00:00 [ksoftirqd/0]
root         4     1  0 10:29 ?        00:00:00 [migration/1]
root         5     1  0 10:29 ?        00:00:00 [ksoftirqd/1]
root         6     1  0 10:29 ?        00:00:00 [migration/2]
root         7     1  0 10:29 ?        00:00:00 [ksoftirqd/2]
root         8     1  0 10:29 ?        00:00:00 [migration/3]
root         9     1  0 10:29 ?        00:00:00 [ksoftirqd/3]
root        10     1  0 10:29 ?        00:00:00 [events/0]
root        11     1  0 10:29 ?        00:00:00 [events/1]
root        12     1  0 10:29 ?        00:00:00 [events/2]
root        13     1  0 10:29 ?        00:00:00 [events/3]
root        14    10  0 10:29 ?        00:00:00 [khelper]
root        15    10  0 10:29 ?        00:00:00 [kacpid]
root        70    10  0 10:29 ?        00:00:00 [kblockd/0]
root        71    10  0 10:29 ?        00:00:00 [kblockd/1]
root        72    10  0 10:29 ?        00:00:00 [kblockd/2]
root        73    10  0 10:29 ?        00:00:00 [kblockd/3]
root        83    10  0 10:29 ?        00:00:00 [pdflush]
root        84    10  0 10:29 ?        00:00:00 [pdflush]
root        85     1  0 10:29 ?        00:00:00 [kswapd0]
root        86    10  0 10:29 ?        00:00:00 [aio/0]
root        87    10  0 10:29 ?        00:00:00 [aio/1]
root        88    10  0 10:29 ?        00:00:00 [aio/2]
root        89    10  0 10:29 ?        00:00:00 [aio/3]
root       194     1  0 10:29 ?        00:00:00 [scsi_eh_0]
root       206     1  0 10:29 ?        00:00:00 [scsi_eh_1]
root       210     1  0 10:29 ?        00:00:00 [kjournald]
root       836     1  0 10:29 ttyS1    00:00:00 init [S]              
root       837   836  0 10:29 ttyS1    00:00:00 /bin/sh
root       846   837  0 10:33 ttyS1    00:00:00 ps -ef

--- [2]

# cat /proc/meminfo; sleep 10; cat /proc/meminfo
MemTotal:      1006272 kB
MemFree:        671104 kB
Buffers:         14720 kB
Cached:          18304 kB
SwapCached:          0 kB
Active:          35584 kB
Inactive:         9216 kB
HighTotal:           0 kB
HighFree:            0 kB
LowTotal:      1006272 kB
LowFree:        671104 kB
SwapTotal:           0 kB
SwapFree:            0 kB
Dirty:               0 kB
Writeback:           0 kB
Mapped:           6592 kB
Slab:           212160 kB
Committed_AS:     8576 kB
PageTables:       1344 kB
VmallocTotal: 35184363699328 kB
VmallocUsed:       384 kB
VmallocChunk: 35184363698944 kB

MemTotal:      1006272 kB
MemFree:        670016 kB
Buffers:         14720 kB
Cached:          18304 kB
SwapCached:          0 kB
Active:          35648 kB
Inactive:         9152 kB
HighTotal:           0 kB
HighFree:            0 kB
LowTotal:      1006272 kB
LowFree:        670016 kB
SwapTotal:           0 kB
SwapFree:            0 kB
Dirty:             256 kB
Writeback:           0 kB
Mapped:           6592 kB
Slab:           212224 kB
Committed_AS:     8576 kB
PageTables:       1344 kB
VmallocTotal: 35184363699328 kB
VmallocUsed:       384 kB
VmallocChunk: 35184363698944 kB

--- [3]

# ./vm.pl; sleep 10; ./vm.pl 
leaked=104640KB
leaked=105664KB
# 102.4kB/s leak rate here

-- 
Daniel J Blueman

NEU: WLAN-Router für 0,- EUR* - auch für DSL-Wechsler!
GMX DSL = supergünstig & kabellos http://www.gmx.net/de/go/dsl


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [2.6.7, ia64] continual memory leak at ~102kB/s...
  2004-08-06 14:44 [2.6.7, ia64] continual memory leak at ~102kB/s Daniel Blueman
@ 2004-08-06 14:54 ` William Lee Irwin III
  2004-08-09  9:22   ` Daniel Blueman
  0 siblings, 1 reply; 3+ messages in thread
From: William Lee Irwin III @ 2004-08-06 14:54 UTC (permalink / raw)
  To: Daniel Blueman; +Cc: linux-ia64, davidm, linux-kernel

On Fri, Aug 06, 2004 at 04:44:18PM +0200, Daniel Blueman wrote:
> When running 2.6.7 on a generic ia64 system, I see memory being leaked in
> the kernel. Most of the fancy (preempt, hot-plug procs, ...) features are
> disabled, and the system in a quiescent state [1].
> /proc/meminfo shows the memory as unaccounted for [2], so it seems likely it
> has been kmalloc()d somehere. A small script shows memory disappearing at
> 102kB/s [3].
> Anyone else seen this on ia64?

Could you dump /proc/slabinfo?


-- wli

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [2.6.7, ia64] continual memory leak at ~102kB/s...
  2004-08-06 14:54 ` William Lee Irwin III
@ 2004-08-09  9:22   ` Daniel Blueman
  0 siblings, 0 replies; 3+ messages in thread
From: Daniel Blueman @ 2004-08-09  9:22 UTC (permalink / raw)
  To: William Lee Irwin III; +Cc: linux-ia64, davidm, linux-kernel

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="us-ascii", Size: 3953 bytes --]

> On Fri, Aug 06, 2004 at 04:44:18PM +0200, Daniel Blueman wrote:
> > When running 2.6.7 on a generic ia64 system, I see memory being leaked
> in
> > the kernel. Most of the fancy (preempt, hot-plug procs, ...) features
> are
> > disabled, and the system in a quiescent state [1].
> > /proc/meminfo shows the memory as unaccounted for [2], so it seems
> likely it
> > has been kmalloc()d somehere. A small script shows memory disappearing
> at
> > 102kB/s [3].
> > Anyone else seen this on ia64?
> 
> Could you dump /proc/slabinfo?

Sampling /proc/slabinfo and again after 30 minutes while the system is
quiescent, nothing looks suspicious:

 tcp_tw_bucket  0       0
 tcp_bind_bucket        12      1129
 tcp_open_request       0       0
-inet_peer_cache        0       0
+inet_peer_cache        1       668
 ip_fib_hash    11      1309
-ip_dst_cache   11      177
+ip_dst_cache   14      177
 arp_cache      2       253
 raw4_sock      0       0
 udp_sock       6       81
 tcp_sock       18      44
 flow_cache     0       0
-scsi_cmd_cache 11      129
+scsi_cmd_cache 2       129
 nfs_write_data 36      85
 nfs_read_data  32      88
 nfs_inode_cache        16      72
@@ -23,8 +23,8 @@
 isofs_inode_cache      0       0
 fat_inode_cache        1       104
 ext2_inode_cache       0       0
-journal_handle 18      885
-journal_head   26      574
+journal_handle 16      885
+journal_head   18      574
 revoke_table   2       1559
 revoke_record  0       0
 ext3_inode_cache       981     1080
@@ -45,42 +45,42 @@
 sgpool-64      32      62
 sgpool-32      32      62
 sgpool-16      32      121
-sgpool-8       41      232
+sgpool-8       32      232
 cfq_pool       64      798
 crq_pool       0       0
 deadline_drq   0       0
-as_arq 25      474
+as_arq 16      474
 blkdev_ioc     18      1129
 blkdev_queue   28      75
-blkdev_requests        25      225
+blkdev_requests        16      225
 biovec-(256)   256     270
 biovec-128     256     279
 biovec-64      256     310
 biovec-16      256     464
 biovec-4       256     727
-biovec-1       265     1559
-bio    265     536
-sock_inode_cache       40      100
-skbuff_head_cache      268     506
+biovec-1       256     1559
+bio    256     536
+sock_inode_cache       39      100
+skbuff_head_cache      267     506
 sock   2       110
-proc_inode_cache       171     216
-sigqueue       0       404
+proc_inode_cache       177     216
+sigqueue       0       0
 radix_tree_node        644     696
 bdev_cache     5       85
 mnt_cache      23      474
 inode_cache    1369    1482
-dentry_cache   3031    3150
-filp   223     478
-names_cache    10      15
+dentry_cache   3038    3150
+filp   230     478
+names_cache    15      15
 idr_layer_cache        31      118
-buffer_head    13792   13936
-mm_struct      35      130
-vm_area_struct 508     648
+buffer_head    13888   13936
+mm_struct      34      130
+vm_area_struct 500     648
 fs_cache       36      727
 files_cache    37      154
 signal_cache   74      503
 sighand_cache  65      123
-anon_vma       292     1309
+anon_vma       291     1309
 size-131072(DMA)       0       0
 size-131072    0       0
 size-65536(DMA)        0       0
@@ -92,17 +92,17 @@
 size-8192(DMA) 0       0
 size-8192      25      32
 size-4096(DMA) 0       0
-size-4096      315     315
+size-4096      314     315
 size-2048(DMA) 0       0
-size-2048      176     186
+size-2048      175     186
 size-1024(DMA) 0       0
 size-1024      219     682
 size-512(DMA)  0       0
-size-512       496     38720
+size-512       495     38720
 size-256(DMA)  0       0
-size-256       1094    25984
+size-256       1092    25984
 size-128(DMA)  0       0
 size-128       2201    6800
 size-64(DMA)   0       0
-size-64        2527    103961
+size-64        2521    103961
 kmem_cache     120     169

-- 
Daniel J Blueman

NEU: WLAN-Router für 0,- EUR* - auch für DSL-Wechsler!
GMX DSL = supergünstig & kabellos http://www.gmx.net/de/go/dsl


^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2004-08-09  9:22 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-08-06 14:44 [2.6.7, ia64] continual memory leak at ~102kB/s Daniel Blueman
2004-08-06 14:54 ` William Lee Irwin III
2004-08-09  9:22   ` Daniel Blueman

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox