From mboxrd@z Thu Jan 1 00:00:00 1970 From: Fredrik Noring Subject: Re: [PATCH] dma-mapping: Relax warnings for per-device areas Date: Tue, 17 Jul 2018 18:33:30 +0200 Message-ID: <20180717163330.GE2342@localhost.localdomain> References: <1f8262d206c6886072d04cc93454f6e3f812bd20.1530623284.git.robin.murphy@arm.com> <20180705193613.GA28905@lst.de> <20180715122826.GB2342@localhost.localdomain> <20180717145725.GA22109@lst.de> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Return-path: Content-Disposition: inline In-Reply-To: <20180717145725.GA22109@lst.de> Sender: linux-kernel-owner@vger.kernel.org To: Christoph Hellwig Cc: Robin Murphy , m.szyprowski@samsung.com, Geoff Levand , iommu@lists.linux-foundation.org, linux-kernel@vger.kernel.org, JuergenUrban@gmx.de List-Id: iommu@lists.linux-foundation.org Hi Christoph, Geoff, [ CC-ing Geoff to give him an opportunity to chime in about the PS3 part. ] > > Here are three other regressions related to the coherent mask WARN_ON_ONCE: > > They are a pretty strong indication that yes, you should really set > the coherent mask if you ever do coherent allocations.. I'm unfortunately unfamiliar with the USB drivers for the PS3. Geoff, what do you think about setting a coherent mask? Christoph, what mask value would you suggest for the PS2 driver? Typical DMA addresses 0-0x200000 are mapped to 0x1c000000-0x1c200000 and is memory managed exclusively by the IOP. Robin indicated that DMA_BIT_MASK(20) or a nonzero but useless value such as 1 are possibly candidates. It seems quite unclear what the coherent mask actually means in this case, doesn't it? Fredrik