Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Fri, 28 Jan 2022 14:15:56 +0800
From:   Peter Xu <peterx@redhat.com>
To:     John Hubbard <jhubbard@nvidia.com>
Cc:     Jason Gunthorpe <jgg@ziepe.ca>, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org,
        Andrew Morton <akpm@linux-foundation.org>,
        Andrea Arcangeli <aarcange@redhat.com>,
        Jan Kara <jack@suse.cz>,
        =?utf-8?B?SsOpcsO0bWU=?= Glisse <jglisse@redhat.com>,
        "Kirill A . Shutemov" <kirill.shutemov@linux.intel.com>,
        Alex Williamson <alex.williamson@redhat.com>
Subject: Re: [PATCH] mm: Fix invalid page pointer returned with FOLL_PIN gups
Message-ID: <YfOKHMbk2caH4d5g@xz-m1.local>
References: <20220125033700.69705-1-peterx@redhat.com>
 <f5400544-f602-0bb4-5cb1-5ac682e41974@nvidia.com>
 <20220127004206.GP8034@ziepe.ca>
 <YfJjhop3senAUjue@xz-m1.local>
 <20220127152538.GQ8034@ziepe.ca>
 <YfNIjqPpty0YkLJP@xz-m1.local>
 <20220128023127.GR8034@ziepe.ca>
 <YfNiWHDYH0dtj9rK@xz-m1.local>
 <b773a0ab-e226-6acb-eab5-24036b0cf6f4@nvidia.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <b773a0ab-e226-6acb-eab5-24036b0cf6f4@nvidia.com>
Precedence: bulk

On Thu, Jan 27, 2022 at 09:57:34PM -0800, John Hubbard wrote:
> On 1/27/22 19:26, Peter Xu wrote:
> ...
> > I see that both you and John has a strong preference on at least the
> > WARN_ON_ONCE() in the patch.
> > 
> > Do you think it's okay I repost with only the one-liner fix, which will keep
> > the Fixes but drop the WARN_ON_ONCE?  Then we can leave the rest as follow up.
> > 
> 
> I think that's OK with me, anyway. You'll recall that I initially requested
> that you split this into two patches, after all.
> 
> Would you like me to post a follow-up that does the refactoring that Jason
> and I are requesting? I see that we have some fundamental differences in
> opinion about how this should be done, so rather than drive you crazy with
> debating, maybe that would be smoother?  :)

Sure thing. :-)

Please just double check that the pud devmap will still always work on mlock().
I believe both of you are much more familiar than me on that; it just still
seems a little bit tricky.

I'll repost this one, thanks.

-- 
Peter Xu