Re: dentry_cache using up all my zone normal memory -- also seen on 2.6.14-rc2

[Dipankar Sarma <dipankar@in.ibm.com>]

On Thu, Sep 22, 2005 at 09:16:58AM -0600, Christopher Friesen wrote:
> Al Viro wrote:
> 
> >Umm...   How many RCU callbacks are pending?
> 
> I added an atomic counter that is incremented just before call_rcu() in 
> d_free(), and decremented just after kmem_cache_free() in d_callback().
> 
> According to this we had 4127306 pending rcu callbacks.  A few seconds 
> later it was down to 0.
> 
> 
> /proc/sys/fs/dentry-state:
> 1611    838     45      0       0       0
> 

Hmm.. This clearly indicates that there are very few allocated dentries
and they are just not returned to slab by RCU.

Since then, I have done some testing myself, but I can't reproduce
this problem in two of my systems - x86 and x86_64. I ran rename14
in a loop too, but after exhausting a lot of free memory, dcache
does get shrunk and I don't see dentries stuck in RCU queues at all.
I tried UP kernel too.

So, there must be something else in your system that I
am missing in my setup. Could you please mail me your .config ?

Thanks
Dipankar
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/