From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Mon, 14 Jan 2019 01:53:15 -0800 From: Christoph Hellwig Subject: Re: [rpmsg PATCH v2 1/1] rpmsg: virtio_rpmsg_bus: fix unexpected huge vmap mappings Message-ID: <20190114095315.GA24495@infradead.org> References: <1545812449-32455-1-git-send-email-fugang.duan@nxp.com> <20181226145048.GA24307@infradead.org> <20181227121901.GA20892@infradead.org> <20190110140726.GA6223@infradead.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: To: Andy Duan Cc: Christoph Hellwig , "bjorn.andersson@linaro.org" , "ohad@wizery.com" , Ard Biesheuvel , Andrew Morton , Linux-MM , Robin Murphy , "linux-remoteproc@vger.kernel.org" , "anup@brainfault.org" , "loic.pallardy@st.com" , dl-linux-imx , Richard Zhu , Jason Liu , Peng Fan List-ID: On Fri, Jan 11, 2019 at 01:28:46AM +0000, Andy Duan wrote: > As NXP i.MX8 platform requirement that M4 only access the fixed memory region, so do > You have any suggestion to fix the issue and satisfy the requirement ? Or do you have plan > To fix the root cause ? I think the answer is to use RESERVEDMEM_OF_DECLARE without the DMA coherent boilerplate code. For the initial prototype just do it inside the driver, although I'd like to eventually factor this out into common code, especially if my proposal for more general availability of DMA_ATTR_NON_CONSISTENT goes ahead.