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;
Message-ID: <8420a781-6ac2-e976-cc43-a04d59d2d043@deltatee.com>
Date:   Tue, 6 Dec 2022 11:40:32 -0700
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.5.0
Content-Language: en-CA
To:     David Howells <dhowells@redhat.com>
Cc:     Al Viro <viro@zeniv.linux.org.uk>, Christoph Hellwig <hch@lst.de>,
        John Hubbard <jhubbard@nvidia.com>,
        Matthew Wilcox <willy@infradead.org>,
        linux-fsdevel@vger.kernel.org, linux-mm@kvack.org,
        Christoph Hellwig <hch@infradead.org>,
        Jeff Layton <jlayton@kernel.org>, linux-kernel@vger.kernel.org
References: <0d62615c-57d1-922b-5ebc-32faabf33327@deltatee.com>
 <166997419665.9475.15014699817597102032.stgit@warthog.procyon.org.uk>
 <166997421646.9475.14837976344157464997.stgit@warthog.procyon.org.uk>
 <1251303.1670351737@warthog.procyon.org.uk>
From:   Logan Gunthorpe <logang@deltatee.com>
In-Reply-To: <1251303.1670351737@warthog.procyon.org.uk>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Subject: Re: [PATCH v3 2/4] iov_iter: Add a function to extract a page list
 from an iterator
Precedence: bulk


On 2022-12-06 11:35, David Howells wrote:
> Logan Gunthorpe <logang@deltatee.com> wrote:
> 
>> If this is going to be a general replacement for iov_iter_get_pages()
>> it's going to need to pass through gup_flags. My recent patchset added
>> versions with these and I think it should be in during the next merge
>> cycle. [1]
> 
> Cool.  Note that the current iov_iter_get_pages2() is broken, which is why Al
> wanted a replacement.  It should not be taking a ref on the pages in an
> XARRAY, BVEC or PIPE - and it should be pinning rather than getting a ref on
> pages in IOVEC or UBUF if the buffer is being read into.  I'm guessing that
> your changes move the latter decision to the caller?

My changes maintained the status quo in terms of brokenness. They simply
added the gup_flags so I could pass a P2PDMA flag in a couple specific
cases. I have no objections to the other changes and vaguely look ok to
me, but having not seen patches to convert the users I care about, I
thought I'd point out that the P2PDMA use case will need to be supported
somehow.

Logan