Re: 2.6.23-rc6-mm1: BUG kmalloc-16: Object padding overwritten(sysfs?)

From: Christoph Lameter
Date: Wed Sep 19 2007 - 17:36:22 EST


On Wed, 19 Sep 2007, Christoph Lameter wrote:

> Hmmm... This works fine here. Doing the cat may trigger slab validation
> and show objects corruptions that were heretofore unnoticed.

Not exactly true since reading a file cannot trigger slab validation. The
catting of all files will touch the following:

clameter@schroedinger:/sys/slab/kmalloc-64$ ls -l
total 0
-r--r--r-- 1 root root 4096 Sep 19 14:28 aliases
-r--r--r-- 1 root root 4096 Sep 19 14:28 align
-r--r--r-- 1 root root 4096 Sep 19 14:28 alloc_calls
-r--r--r-- 1 root root 4096 Sep 19 14:28 cache_dma
-r--r--r-- 1 root root 4096 Sep 19 14:28 cpu_slabs
-r--r--r-- 1 root root 4096 Sep 19 14:28 ctor
-r--r--r-- 1 root root 4096 Sep 19 14:28 destroy_by_rcu
-r--r--r-- 1 root root 4096 Sep 19 14:28 free_calls
-r--r--r-- 1 root root 4096 Sep 19 14:28 hwcache_align
-r--r--r-- 1 root root 4096 Sep 19 14:28 object_size
-r--r--r-- 1 root root 4096 Sep 19 14:28 objects
-r--r--r-- 1 root root 4096 Sep 19 14:28 objs_per_slab
-r--r--r-- 1 root root 4096 Sep 19 14:28 order
-r--r--r-- 1 root root 4096 Sep 19 14:28 partial
-rw-r--r-- 1 root root 4096 Sep 19 14:28 poison
-rw-r--r-- 1 root root 4096 Sep 19 14:28 reclaim_account
-rw-r--r-- 1 root root 4096 Sep 19 14:28 red_zone
-rw-r--r-- 1 root root 4096 Sep 19 14:28 sanity_checks
-rw-r--r-- 1 root root 4096 Sep 19 14:28 shrink
-r--r--r-- 1 root root 4096 Sep 19 14:28 slab_size
-r--r--r-- 1 root root 4096 Sep 19 14:28 slabs
-rw-r--r-- 1 root root 4096 Sep 19 14:28 store_user
-rw-r--r-- 1 root root 4096 Sep 19 14:28 trace
-rw-r--r-- 1 root root 0 Sep 19 14:28 validate

catting

alloc_calls

and
free_calls

will perform a global scan over all objects to determine the callers.
Before we do that we have to flush the per cpu slabs. The trace shows that
one of the errors was detected in there and the other one probably later.

So I would think that this is an actual memory corruption issue on your
system and the /sys/slab stuff works fine. slabinfo -v will perform a more
details scan to help you located the problem.

-
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/