Messages in this thread | | | From | Peng Fan <> | Subject | RE: [RFC] virtio_ring: check dma_mem for xen_domain | Date | Tue, 22 Jan 2019 02:32:19 +0000 |
| |
Hi
> -----Original Message----- > From: hch@infradead.org [mailto:hch@infradead.org] > Sent: 2019年1月21日 16:29 > To: Peng Fan <peng.fan@nxp.com> > Cc: mst@redhat.com; jasowang@redhat.com; sstabellini@kernel.org; > hch@infradead.org; virtualization@lists.linux-foundation.org; > xen-devel@lists.xenproject.org; linux-kernel@vger.kernel.org; > linux-remoteproc@vger.kernel.org > Subject: Re: [RFC] virtio_ring: check dma_mem for xen_domain > > On Mon, Jan 21, 2019 at 04:51:57AM +0000, Peng Fan wrote: > > on i.MX8QM, M4_1 is communicating with DomU using rpmsg with a fixed > > address as the dma mem buffer which is predefined. > > > > Without this patch, the flow is: > > vring_map_one_sg -> vring_use_dma_api > > -> dma_map_page > > -> __swiotlb_map_page > > ->swiotlb_map_page > > ->__dma_map_area(phys_to_virt(dma_to_phys(dev, > dev_addr)), size, > > dir); However we are using per device dma area for rpmsg, phys_to_virt > > could not return a correct virtual address for virtual address in > > vmalloc area. Then kernel panic. > > And that is the right thing to do. You must not call dma_map_* on memory > that was allocated from dma_alloc_*.
Understand. But the current code is that vring_use_dma_api will always return true, if the current OS is a xen VM.
Actually it needs to return false for my case, then we could use sg_phys(sg) to get the correct physical address.
> > We actually have another thread which appears to be for this same issue.
You mean https://patchwork.kernel.org/patch/10742923/ ?
You suggest use cma there, but vring_use_dma_api will still return true if the OS is running on xen. Then vring_map_one_sg will still runs into __dma_map_area.
In my case, just need vring_use_dma_api to return false and use sg_phys(sg) to get the correct physical address, whether per dma reserved area or per device cma.
Thanks, Peng.
| |