Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Wed, 2 Oct 2019 11:27:31 -0300
From:   Jason Gunthorpe <jgg@ziepe.ca>
To:     Alan Mikhak <alan.mikhak@sifive.com>
Cc:     linux-kernel@vger.kernel.org, martin.petersen@oracle.com,
        alexios.zavras@intel.com, ming.lei@redhat.com,
        gregkh@linuxfoundation.org, tglx@linutronix.de,
        christophe.leroy@c-s.fr, thellstrom@vmware.com,
        galpress@amazon.com, palmer@sifive.com, paul.walmsley@sifive.com
Subject: Re: [PATCH] scatterlist: Comment on pages for sg_set_page()
Message-ID: <20191002142731.GC17152@ziepe.ca>
References: <1569954770-11477-1-git-send-email-alan.mikhak@sifive.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <1569954770-11477-1-git-send-email-alan.mikhak@sifive.com>
User-Agent: Mutt/1.9.4 (2018-02-28)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Tue, Oct 01, 2019 at 11:32:50AM -0700, Alan Mikhak wrote:
> From: Alan Mikhak <alan.mikhak@sifive.com>
> 
> Update the description of sg_set_page() to communicate current
> requirements for the page pointer parameter.
> 
> Signed-off-by: Alan Mikhak <alan.mikhak@sifive.com>
>  include/linux/scatterlist.h | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/include/linux/scatterlist.h b/include/linux/scatterlist.h
> index 6eec50fb36c8..6dda865893aa 100644
> +++ b/include/linux/scatterlist.h
> @@ -112,6 +112,12 @@ static inline void sg_assign_page(struct scatterlist *sg, struct page *page)
>   *   of the page pointer. See sg_page() for looking up the page belonging
>   *   to an sg entry.
>   *
> + *   Scatterlist currently expects the page parameter to be a pointer to
> + *   a page that is backed by a page struct.
> + *
> + *   Page pointers derived from addresses obtained from ioremap() are
> + *   currently not supported since they require use of iomem safe memcpy.
> + *
>   **/
>  static inline void sg_set_page(struct scatterlist *sg, struct page *page,
>  			       unsigned int len, unsigned int offset)

It seems a bit weird to have a comment explaining that 'struct page
*page' must actually be a valid pointer. Of course it must.

Computing a 'struct page *' to something that doesn't actually have a
struct page is simply a bug in whoever did that.

Code should never be interchanging ioremap results with the struct
page* world.

Jason