[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAL9ddJciz2MD8CYqdbFLhYCKFk=ouHzzEndQwmcfQ-UqNNgJxQ@mail.gmail.com>
Date: Wed, 2 Sep 2020 11:42:37 -0700
From: David Awogbemila <awogbemila@...gle.com>
To: Jakub Kicinski <kuba@...nel.org>
Cc: netdev@...r.kernel.org, Catherine Sullivan <csully@...gle.com>,
Yangchun Fu <yangchun@...gle.com>
Subject: Re: [PATCH net-next v2 4/9] gve: Add support for dma_mask register
On Tue, Sep 1, 2020 at 5:34 PM Jakub Kicinski <kuba@...nel.org> wrote:
>
> On Tue, 1 Sep 2020 14:51:44 -0700 David Awogbemila wrote:
> > From: Catherine Sullivan <csully@...gle.com>
> >
> > Add the dma_mask register and read it to set the dma_masks.
> > gve_alloc_page will alloc_page with:
> > GFP_DMA if priv->dma_mask is 24,
> > GFP_DMA32 if priv->dma_mask is 32.
>
> What about Andrew's request to CC someone who can review this for
> correctness?
I didn't realize I needed to CC someone. How may I find suitable reviewers?
>
> What's your use case here? Do you really have devices with 24bit
> addressing?
I don't think there is a specific 24-bit device in mind here, only
that we have seen 32-bit addressing use cases where the guest ran out
of SWIOTLB space and restricting to GFP_DMA32 helped.. so we thought
it would be natural for the driver to handle the 24 bit case in case
it ever came along.
Powered by blists - more mailing lists