From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from gate.crashing.org (gate.crashing.org [63.228.1.57]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client did not present a certificate) by ozlabs.org (Postfix) with ESMTP id 5D094DDDE4 for ; Wed, 9 May 2007 07:17:40 +1000 (EST) Subject: Re: [PATCH 5/5] powerpc: Don't use SLAB/SLUB for PTE pages From: Benjamin Herrenschmidt To: Hugh Dickins In-Reply-To: References: <20070508062750.A5CBBDDE42@ozlabs.org> Content-Type: text/plain Date: Wed, 09 May 2007 07:17:30 +1000 Message-Id: <1178659050.14928.78.camel@localhost.localdomain> Mime-Version: 1.0 Cc: linuxppc-dev@ozlabs.org, Paul Mackerras List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Tue, 2007-05-08 at 20:50 +0100, Hugh Dickins wrote: > On Tue, 8 May 2007, Benjamin Herrenschmidt wrote: > > From: Hugh Dickins > > > > The SLUB allocator relies on struct page fields first_page and slab, > > overwritten by ptl when SPLIT_PTLOCK: so the SLUB allocator cannot then > > be used for the lowest level of pagetable pages. This was obstructing > > SLUB on PowerPC, which uses kmem_caches for its pagetables. So convert > > its pte level to use normal gfp pages (whereas pmd, pud and 64k-page pgd > > want partpages, so continue to use kmem_caches for pmd, pud and pgd). > > > > Signed-off-by: Hugh Dickins > > Signed-off-by: Benjamin Herrenschmidt > > --- > > Looks and runs fine for me (though would have seemed nicer to use > __GFP_ZERO than clear_page - or does that subvert your fast zeroing?). Nah, you are right, I'll respin. Forgot about __GFP_ZERO. It just called clear_highpage on the returned pages right ? This is the same as calling clear_page from my code. > But you also need to undo the ARCH_USES_SLAB_PAGE_STRUCT from > arch/powerpc/Kconfig now that's gone into Linus' tree: patch > I added in testing below (either append to this one or add a 6/5). I was thinking about adding a new patch. > For some people that interim disablement was working, for other > people it wasn't: mysterious, and the sooner it's got rid of the > better. I'm sorry to find that oldconfig doesn't actually offer > the SLAB or SLUB menu (to me anyway) when moving from SLUB disabled > to SLUB enabled: so again, the sooner we get rid of that interim > disablement the better for SLUB exposure. Ok. I'll post a new patch along with the final serie after I've done some testing later today. Cheers, Ben.