xfs
[Top] [All Lists]

Re: [PATCH] (repost) kmem_cache_zalloc

To: Steve Lord <lord@xxxxxxx>
Subject: Re: [PATCH] (repost) kmem_cache_zalloc
From: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Date: Thu, 25 Apr 2002 14:56:46 +0100
Cc: Eric Sandeen <sandeen@xxxxxxx>, linux-xfs@xxxxxxxxxxx
In-reply-to: <1019741880.8353.8.camel@jen.americas.sgi.com>; from lord@sgi.com on Thu, Apr 25, 2002 at 08:38:00AM -0500
References: <1019682472.15455.33.camel@stout.americas.sgi.com> <20020425101333.A18583@infradead.org> <1019741880.8353.8.camel@jen.americas.sgi.com>
Sender: owner-linux-xfs@xxxxxxxxxxx
User-agent: Mutt/1.2.5.1i
On Thu, Apr 25, 2002 at 08:38:00AM -0500, Steve Lord wrote:
> > There are only two callers, one is in the pagebuf code and would benefit
> > from letting slab cache initialized objects and the other in the emulation
> > of IRIX's kmen_zone_zalloc..
> 
> It is a little more complex than those two places - they do not know
> how big the memory is, that is hidden in the zone itself - which is
> not accessible outside slab.c. You need to go back to the callers of
> kmem_zone_zalloc and do the work there. Actually there are only 19
> of them, so it is not too bad.

<mode="provoke Al">
parse /proc/slabinfo in kernelspace to find it
</mode>

When looking over all uses of kmen_zone_* in the XFS code I see some
place that look like opportunities for a Linux SLAB cache - and that one
should be easily emulatable on IRIX by always calling the constructure
after getting memory from the lowlevel allocator (sorry for lack of IRIX
lowlevel knowledge :))

        Christoph


<Prev in Thread] Current Thread [Next in Thread>