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

[Al Viro]

On Wed, Sep 21, 2005 at 09:54:14PM -0600, Christopher Friesen wrote:
> Al Viro wrote:
> 
> >>Hmm... could there be a race in shmem_rename()??
> 
> >Not likely - in that setup all calls of ->unlink() and ->rename()
> >are completely serialized by ->i_sem on parent.  One question:
> >is it dcache or icache that ends up leaking?
> 
> dcache.  Here's some information I sent to dipankar earlier, with his 
> debug patch applied.  This is within half a second of the oom killer 
> kicking in.

Umm...   How many RCU callbacks are pending?  Since past the OOM you get
the sucker back to normal...  Sounds like you've got a bunch of dentries
on their way to be freed, but the thing that should've been doing final
kmem_cache_free() is getting postponed too much.
-
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/