From mboxrd@z Thu Jan  1 00:00:00 1970
From: Christoph Hellwig <hch@infradead.org>
Subject: Re: [PATCH] net: mvpp2: avoid bouncing buffers
Date: Sun, 19 Aug 2018 23:23:26 -0700
Message-ID: <20180820062326.GA22222@infradead.org>
References: <20180820024730.9147-1-brian.brooks@linaro.org>
 <20180819.195505.1988137313680465320.davem@davemloft.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: brian.brooks@linaro.org, antoine.tenart@bootlin.com,
        maxime.chevallier@bootlin.com, ymarkman@marvell.com,
        stefanc@marvell.com, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org, bjorn.topel@intel.com,
        brian.brooks@arm.com
To: David Miller <davem@davemloft.net>
Return-path: <netdev-owner@vger.kernel.org>
Received: from bombadil.infradead.org ([198.137.202.133]:48082 "EHLO
        bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
        with ESMTP id S1726026AbeHTJhu (ORCPT
        <rfc822;netdev@vger.kernel.org>); Mon, 20 Aug 2018 05:37:50 -0400
Content-Disposition: inline
In-Reply-To: <20180819.195505.1988137313680465320.davem@davemloft.net>
Sender: netdev-owner@vger.kernel.org
List-ID: <netdev.vger.kernel.org>

On Sun, Aug 19, 2018 at 07:55:05PM -0700, David Miller wrote:
> From: Brian Brooks <brian.brooks@linaro.org>
> Date: Sun, 19 Aug 2018 21:47:30 -0500
> 
> > @@ -5126,6 +5126,12 @@ static int mvpp2_probe(struct platform_device *pdev)
> >  	}
> >  
> >  	if (priv->hw_version == MVPP22) {
> > +		/* Platform code may have set dev->dma_mask to point
> > +		 * to dev->coherent_dma_mask, but we want to ensure
> > +		 * they take different values due to comment below.
> > +		 */
> > +		pdev->dev.dma_mask = &priv->dma_mask;
> 
> The platform code might be doing this exactly because it cannot support
> different coherent and streaming DMA masks.
> 
> Well, in any case, the platform code is doing it for a reason and
> overriding this in a "driver" of all places seems totally
> inappropriate and at best a layering violation.
> 
> I would rather you fix this in a way that involves well defined APIs
> that set the DMA masks or whatever to the values that you need, rather
> than going behind the platform code's back and changing the DMA mask
> pointer like this.

Agreed.  What platform do you see this issue on?