Received-SPF: pass (google.com: domain of linux-kernel+bounces-189715-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Date: Sun, 26 May 2024 13:04:51 +0200
From: Oscar Salvador <osalvador@suse.de>
To: Christophe Leroy <christophe.leroy@csgroup.eu>
Cc: Andrew Morton <akpm@linux-foundation.org>,
	Jason Gunthorpe <jgg@nvidia.com>, Peter Xu <peterx@redhat.com>,
	Michael Ellerman <mpe@ellerman.id.au>,
	Nicholas Piggin <npiggin@gmail.com>, linux-kernel@vger.kernel.org,
	linux-mm@kvack.org, linuxppc-dev@lists.ozlabs.org
Subject: Re: [RFC PATCH v3 00/16] Reimplement huge pages without hugepd on
 powerpc (8xx, e500, book3s/64)
Message-ID: <ZlMXU_4sbm02jEwe@localhost.localdomain>
References: <cover.1716714720.git.christophe.leroy@csgroup.eu>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <cover.1716714720.git.christophe.leroy@csgroup.eu>

On Sun, May 26, 2024 at 11:22:20AM +0200, Christophe Leroy wrote:
> This is the continuation of the RFC v1 series "Reimplement huge pages
> without hugepd on powerpc 8xx". It now get rid of hugepd completely
> after handling also e500 and book3s/64
> 
> Also see https://github.com/linuxppc/issues/issues/483
> 
> Unlike most architectures, powerpc 8xx HW requires a two-level
> pagetable topology for all page sizes. So a leaf PMD-contig approach
> is not feasible as such.
> 
> Possible sizes on 8xx are 4k, 16k, 512k and 8M.
> 
> First level (PGD/PMD) covers 4M per entry. For 8M pages, two PMD entries
> must point to a single entry level-2 page table. Until now that was
> done using hugepd. This series changes it to use standard page tables
> where the entry is replicated 1024 times on each of the two pagetables
> refered by the two associated PMD entries for that 8M page.
> 
> For e500 and book3s/64 there are less constraints because it is not
> tied to the HW assisted tablewalk like on 8xx, so it is easier to use
> leaf PMDs (and PUDs).
> 
> On e500 the supported page sizes are 4M, 16M, 64M, 256M and 1G. All at
> PMD level on e500/32 (mpc85xx) and mix of PMD and PUD for e500/64. We
> encode page size with 4 available bits in PTE entries. On e300/32 PGD
> entries size is increases to 64 bits in order to allow leaf-PMD entries
> because PTE are 64 bits on e500.
> 
> On book3s/64 only the hash-4k mode is concerned. It supports 16M pages
> as cont-PMD and 16G pages as cont-PUD. In other modes (radix-4k, radix-6k
> and hash-64k) the sizes match with PMD and PUD sizes so that's just leaf
> entries. The hash processing make things a bit more complex. To ease
> things, __hash_page_huge() is modified to bail out when DIRTY or ACCESSED
> bits are missing, leaving it to mm core to fix it.
> 
> Global changes in v3:
> - Removed patches 1 and 2
> - Squashed patch 11 into patch 5
> - Replaced patches 12 and 13 with a series from Michael
> - Reordered patches a bit to have more general patches up front
> 
> For more details on changes, see in each patch.
> 
> Christophe Leroy (15):
>   mm: Define __pte_leaf_size() to also take a PMD entry
>   mm: Provide mm_struct and address to huge_ptep_get()
>   powerpc/mm: Remove _PAGE_PSIZE
>   powerpc/mm: Fix __find_linux_pte() on 32 bits with PMD leaf entries
>   powerpc/mm: Allow hugepages without hugepd
>   powerpc/8xx: Fix size given to set_huge_pte_at()
>   powerpc/8xx: Rework support for 8M pages using contiguous PTE entries
>   powerpc/8xx: Simplify struct mmu_psize_def
>   powerpc/e500: Remove enc and ind fields from struct mmu_psize_def
>   powerpc/e500: Switch to 64 bits PGD on 85xx (32 bits)
>   powerpc/e500: Encode hugepage size in PTE bits
>   powerpc/e500: Use contiguous PMD instead of hugepd
>   powerpc/64s: Use contiguous PMD/PUD instead of HUGEPD
>   powerpc/mm: Remove hugepd leftovers
>   mm: Remove CONFIG_ARCH_HAS_HUGEPD

I glanced over it and it looks much better, not having to fiddle with other arch
code and generic declarations is a big plus.
I plan to do a proper review tomorrow.

Thanks for working on this Christophe!


-- 
Oscar Salvador
SUSE Labs