2.0.31-pre9 leaking memory badly

Michael K. Johnson (johnsonm@redhat.com)
Thu, 18 Sep 1997 10:34:26 -0400

I'm amazed that I've only seen one reference to this so far on linux-kernel,
but I can certainly substantiate that 2.0.31-pre9 is leaking memory all over
the floor in at least some configurations. I think that pre-2 from davem
also leaked, but less badly; we had a machine run out of memory after a bit
over a month of uptime (could not get a free page/could not load interpreter).

We rebooted into pre-9 yesterday afternoon, and this is what we have so far:

Directly after a fresh boot:
[root@porkchop log]# free
total used free shared buffers cached
Mem: 63192 52392 10800 12812 16300 22968
-/+ buffers: 13124 50068
Swap: 72288 8 72280

A few hours, and some NFS traffic later:
[root@porkchop log]# free
total used free shared buffers cached
Mem: 63192 62188 1004 13456 16264 30228
-/+ buffers: 15696 47496
Swap: 72288 8 72280

Perhaps a gig of NFS traffic and a few hours after that:
[root@porkchop log]# free
total used free shared buffers cached
Mem: 63192 62224 968 15264 10996 23424
-/+ buffers: 27804 35388
Swap: 72288 8 72280

The next morning:
[root@porkchop log]# free
total used free shared buffers cached
Mem: 63192 62580 612 6376 1552 3804
-/+ buffers: 57224 5968
Swap: 72288 18880 53408

Either it is leaking badly, or stats are not being kept up.

Machine configuration:
[root@porkchop log]# cat /proc/cpuinfo
processor : 0
cpu : 586
model : Pentium 75+
vendor_id : GenuineIntel
stepping : 12
fdiv_bug : no
hlt_bug : no
fpu : yes
fpu_exception : yes
cpuid : yes
wp : yes
flags : fpu vme de pse tsc msr mce cx8
bogomips : 66.36
[root@porkchop log]# cat /proc/modules
nfs 12 6 (autoclean)
de4x5 9 1 (autoclean)
BusLogic 20 6

If you want more configuration information, just ask. We'll be rebooting
now; this is a production server.

michaelkjohnson

"Magazines all too frequently lead to books and should be regarded by the
prudent as the heavy petting of literature." -- Fran Lebowitz