From mboxrd@z Thu Jan 1 00:00:00 1970 From: Matthew Wilcox Date: Wed, 16 Jan 2019 14:20:14 +0000 Subject: Re: [PATCH V2] mm: Introduce GFP_PGTABLE Message-Id: <20190116142014.GJ6310@bombadil.infradead.org> List-Id: References: <1547619692-7946-1-git-send-email-anshuman.khandual@arm.com> <20190116065703.GE24149@dhcp22.suse.cz> <20190116123018.GF6310@bombadil.infradead.org> <07d6a264-dccd-78ab-e8a9-2410bbef7b97@arm.com> <20190116131827.GH6310@bombadil.infradead.org> <521d8511-4c87-49c6-de03-67a71d5bacca@c-s.fr> In-Reply-To: <521d8511-4c87-49c6-de03-67a71d5bacca@c-s.fr> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable To: Christophe Leroy Cc: mark.rutland@arm.com, christoffer.dall@arm.com, linux-sh@vger.kernel.org, peterz@infradead.org, catalin.marinas@arm.com, palmer@sifive.com, will.deacon@arm.com, Michal Hocko , linux-mm@kvack.org, linux-riscv@lists.infradead.org, kvmarm@lists.cs.columbia.edu, aneesh.kumar@linux.ibm.com, greentime@andestech.com, linux@armlinux.org.uk, mingo@redhat.com, linux-arm-kernel@lists.infradead.org, rientjes@google.com, dave.hansen@linux.intel.com, Anshuman Khandual , marc.zyngier@arm.com, rppt@linux.vnet.ibm.com, shakeelb@google.com, kirill@shutemov.name, tglx@linutronix.de, vbabka@suse.cz, ard.biesheuvel@linaro.org, linuxppc-dev@lists.ozlabs.org, steve.capper@arm.com, linux-kernel@vger.kernel.org, james.morse@arm.com, akpm@linux-foundation.org, robin.murphy@arm.com On Wed, Jan 16, 2019 at 02:47:16PM +0100, Christophe Leroy wrote: > Le 16/01/2019 =E0 14:18, Matthew Wilcox a =E9crit=A0: > > I disagree with your objective. Making more code common is a great ide= a, > > but this patch is too unambitious. We should be heading towards one or > > two page table allocation functions instead of having every architectur= e do > > its own thing. > >=20 > > So start there. Move the x86 function into common code and convert one > > other architecture to use it too. >=20 > Are we talking about pte_alloc_one_kernel() and pte_alloc_one() ? >=20 > I'm not sure x86 function is the best common one, as it seems to allocate= a > multiple of PAGE_SIZE only. And that's the common case. Most architectures use a single page for at least one level of the pte/pmd/pud/p4d/pgd hierarchy. Some use multiple pages and some use a fraction of a page. > Some arches like powerpc use pagetables which are smaller than a page, for > instance powerpc 8xx uses 4k pagetables even with 16k pages, which means a > single page can be used by 4 pagetables. Those can be added later. Note I said "one or two", and that's what I had in mind; I think we want one function that allocates just a page and another that allocates a page fragment. Then we can have a good discussion about what method we use; s390 and ppc use different techniques today and there's really no good reason for that.