Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Wed, 14 Jun 2023 11:35:18 -0400
From:   Peter Xu <peterx@redhat.com>
To:     Matthew Wilcox <willy@infradead.org>
Cc:     linux-kernel@vger.kernel.org, linux-mm@kvack.org,
        Andrea Arcangeli <aarcange@redhat.com>,
        John Hubbard <jhubbard@nvidia.com>,
        Mike Rapoport <rppt@kernel.org>,
        David Hildenbrand <david@redhat.com>,
        Vlastimil Babka <vbabka@suse.cz>,
        "Kirill A . Shutemov" <kirill@shutemov.name>,
        Andrew Morton <akpm@linux-foundation.org>,
        Mike Kravetz <mike.kravetz@oracle.com>,
        James Houghton <jthoughton@google.com>,
        Hugh Dickins <hughd@google.com>
Subject: Re: [PATCH 6/7] mm/gup: Accelerate thp gup even for "pages != NULL"
Message-ID: <ZIneNhvxGGb6zLyq@x1n>
References: <20230613215346.1022773-1-peterx@redhat.com>
 <20230613215346.1022773-7-peterx@redhat.com>
 <ZInVmrJdLWxOEkeD@casper.infradead.org>
 <ZInalCeSNmAiG2K4@x1n>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <ZInalCeSNmAiG2K4@x1n>
Precedence: bulk

On Wed, Jun 14, 2023 at 11:19:48AM -0400, Peter Xu wrote:
> > > +			for (j = 0; j < page_increm; j++) {
> > > +				subpage = nth_page(page, j);
> > > +				pages[i+j] = subpage;
> > > +				flush_anon_page(vma, subpage, start + j * PAGE_SIZE);
> > > +				flush_dcache_page(subpage);
> > 
> > You're better off calling flush_dcache_folio() right at the end.
> 
> Will do.

Ah when I start to modify it I noticed it's a two-sided sword: we'll then
also do flush dcache over the whole folio even if we gup one page.

We'll start to get benefit only if some arch at least starts to impl
flush_dcache_folio() (which seems to be none, right now..), and we'll
already start to lose on amplifying the flush when gup on partial folio.

Perhaps I still keep it as-is which will still be accurate, always faster
than old code, and definitely not regress in any form?

-- 
Peter Xu