From mboxrd@z Thu Jan 1 00:00:00 1970 From: will.deacon@arm.com (Will Deacon) Date: Fri, 2 Mar 2018 18:02:20 +0000 Subject: RFC on Kdump and PCIe on ARM64 In-Reply-To: <84121a8d-5294-0567-6c6b-f57e85dc4961@codeaurora.org> References: <20180301190552.GK13722@bhelgaas-glaptop.roam.corp.google.com> <2b2de17c-8527-e49b-2ef2-2a3d1801e4f9@codeaurora.org> <20180302000303.GD74737@bhelgaas-glaptop.roam.corp.google.com> <20180302103032.GB19323@arm.com> <84121a8d-5294-0567-6c6b-f57e85dc4961@codeaurora.org> Message-ID: <20180302180219.GB20579@arm.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Fri, Mar 02, 2018 at 09:12:29AM -0500, Sinan Kaya wrote: > On 3/2/2018 5:30 AM, Will Deacon wrote: > >> Do you really have to reset the IOMMU? Can you just give it new page > >> tables that start out with all IOVAs from all devices being invalid, > >> then add valid mappings as drivers need them (presumably after the > >> driver has done whatever it needs to so the device stops using the old > >> DMA addresses)? > > We already have the option to do that via the command line using the > > disable_bypass option, so it just sounds like we need to take this into > > account when resetting the SMMU to take care that GBPA is configured so > > that transactions are terminated when SMMUEN=0. > > How about the points that Baoquan highlighted in his email regarding the > solution from AMD and X86? Which specific points do you think this proposal doesn't address? > I have not read the entire thread but, is this just a matter of following > what Bjorn recommended or there is more to it? I'm trying to say how I think Bjorn's idea can be implemented for SMMUv3. I basically want to avoid a situation where the SMMU driver tries to walk the in-memory data structures left by a previous kernel and infer the setup from that. Will