Hi Christoph , On Wed, Nov 28, 2018 at 1:10 PM Christoph Hellwig <hch@xxxxxxxxxxxxx> wrote: > > > + /* > > + * dma_sync_sg_*() flush the physical pages, so point > > + * sg->dma_address to the physical ones for the right behavior. > > + */ > > + for_each_sg(msm_obj->sgt->sgl, s, msm_obj->sgt->nents, i) > > + sg_dma_address(s) = sg_phys(s); > > + > > I'm sorry, but this is completely bogus and not acceptable. > > The only place that is allowed to initialize sg_dma_address is > dma_map_sg. If the default dma ops don't work for your setup we have > major a problem and need to fix the dma api / iommu integration instead > of hacking around it. Thanks for reviewing this change. >From what I understand the things in drm, we don't use the default iommu-dma domain for drm devices. Rather we allocate a new iommu domain, and therefore we can't use the default dma ops. Hence we need separate dma_sync_sg*() to flush/invalidate the cache. For this we need to initialize the sg's addresses. Please correct me if my understanding is incomplete. Best regards Vivek -- QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, hosted by The Linux Foundation
