[Openswan Users] kernel memory leak 2.4.7, 2.4.33 - misconfiguration or bug? (fwd)

[Brad Langhorst]

On Fri, 2007-05-11 at 11:55 -0400, Michael Richardson wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> 
> >>>>> "Brad" == Brad Langhorst <brad at coopmetrics.coop> writes:
>     Brad> thanks for your response!
>     Brad> please find answers to your questions interspersed below..
> 
>     Brad> one new piece of information - memory only seems to leak when i'm
>     Brad> transferring from the remote side to the problem side, not
>     Brad> vice versa.
> 
>   okay.
> 
>   So, slabwatch is definitely what you want to run.
>   You only need to run the collection program on the router.
> 
>   => "slabwatch"
>   Shall I write it as sh+awk for you?... HOW about this?
> 
>   while true
>   do
>      awk '{ printf "echo >>%s %s %s\n", $1, $2, $3;}' /proc/slabinfo | tr -d'()'
>      sleep 1
>   done | sh
>   
i captured slabinfo before an after a transfer that had a 100K decrease
in free memory

here's the diff
--- /tmp/slabinfo_before        Fri May 11 12:35:06 2007
+++ /tmp/slabinfo_after Fri May 11 12:36:25 2007
@@ -1,32 +1,32 @@
               total         used         free       shared      buffers
-  Mem:       119664        65232        54432            0           56
+  Mem:       119664        65332        54332            0           56
  Swap:            0            0            0
-Total:       119664        65232        54432
+Total:       119664        65332        54332
 echo >>slabinfo - version:
 echo >>kmem_cache 57 72
 echo >>ip_fib_hash 18 203
 echo >>urb_priv 0 0
-echo >>ip_conntrack 65 300
+echo >>ip_conntrack 84 300
 echo >>tcp_tw_bucket 0 0
 echo >>tcp_bind_bucket 2 203
 echo >>tcp_open_request 0 40
 echo >>inet_peer_cache 6 78
-echo >>ip_dst_cache 180 462
+echo >>ip_dst_cache 179 462
 echo >>arp_cache 19 40
 echo >>blkdev_requests 2048 2080
 echo >>dnotify_cache 0 0
 echo >>file_lock_cache 1 42
 echo >>fasync_cache 0 0
 echo >>uid_cache 2 113
-echo >>skbuff_head_cache 29855 29920
+echo >>skbuff_head_cache 29934 29942
 echo >>sock 35 42
 echo >>sigqueue 0 29
 echo >>kiobuf 0 0
 echo >>cdev_cache 284 312
 echo >>bdev_cache 1 59
 echo >>mnt_cache 10 59
-echo >>inode_cache 1664 1768
-echo >>dentry_cache 1662 1785
+echo >>inode_cache 1665 1768
+echo >>dentry_cache 1663 1785
 echo >>filp 223 245
 echo >>names_cache 0 2
 echo >>buffer_head 100 8920
@@ -50,7 +50,7 @@
 echo >>size-2048DMA 0 0
 echo >>size-2048 38 118
 echo >>size-1024DMA 0 0
-echo >>size-1024 29574 29588
+echo >>size-1024 29668 29672
 echo >>size-512DMA 0 0
 echo >>size-512 177 216
 echo >>size-256DMA 0 0
@@ -58,6 +58,6 @@
 echo >>size-128DMA 2 30
 echo >>size-128 642 690
 echo >>size-64DMA 0 0
-echo >>size-64 7501 7552
+echo >>size-64 7522 7552
 echo >>size-32DMA 34 113
 echo >>size-32 423 452


i think these values are in bytes...
21+ 96 +2 +79 -1 +19  ~ 215 bytes , but I'm losing 100K

do I have the units wrong?

Brad

ps: i added the openswan users list back in, hope that's okay

-- 
Brad Langhorst
CTO - CoopMetrics