On Tue, Jul 23, 2019 at 5:22 PM Jungo Lin <jungo.lin@xxxxxxxxxxxx> wrote:
>
> Hi, Tomasz:
>
> On Tue, 2019-07-23 at 16:20 +0900, Tomasz Figa wrote:
> > Hi Jungo,
> >
> > On Fri, Jul 5, 2019 at 4:59 PM Jungo Lin <jungo.lin@xxxxxxxxxxxx> wrote:
> > >
> > > Hi Tomasz:
> > >
> > > On Fri, 2019-07-05 at 13:22 +0900, Tomasz Figa wrote:
> > > > Hi Jungo,
> > > >
> > > > On Fri, Jul 5, 2019 at 12:33 PM Jungo Lin <jungo.lin@xxxxxxxxxxxx> wrote:
> > > > >
> > > > > Hi Tomasz,
> > >
> > > [snip]
> > >
> > > > > After applying your suggestion in SCP device driver, we could remove
> > > > > mtk_cam-smem.h/c. Currently, we use dma_alloc_coherent with SCP device
> > > > > to get SCP address. We could touch the buffer with this SCP address in
> > > > > SCP processor.
> > > > >
> > > > > After that, we use dma_map_page_attrs with P1 device which supports
> > > > > IOMMU domain to get IOVA address. For this address, we will assign
> > > > > it to our ISP HW device to proceed.
> > > > >
> > > > > Below is the snippet for ISP P1 compose buffer initialization.
> > > > >
> > > > > ptr = dma_alloc_coherent(p1_dev->cam_dev.smem_dev,
> > > > > MAX_COMPOSER_SIZE, &addr, GFP_KERNEL);
> > > > > if (!ptr) {
> > > > > dev_err(dev, "failed to allocate compose memory\n");
> > > > > return -ENOMEM;
> > > > > }
> > > > > isp_ctx->scp_mem_pa = addr;
> > > >
> > > > addr contains a DMA address, not a physical address. Could we call it
> > > > scp_mem_dma instead?
> > > >
> > > > > dev_dbg(dev, "scp addr:%pad\n", &addr);
> > > > >
> > > > > /* get iova address */
> > > > > addr = dma_map_page_attrs(dev, phys_to_page(addr), 0,
> > > >
> > > > addr is a DMA address, so phys_to_page() can't be called on it. The
> > > > simplest thing here would be to use dma_map_single() with ptr as the
> > > > CPU address expected.
> > > >
> > >
> > > We have changed to use ma_map_single() with ptr, but encounter IOMMU
> > > error. From the debug log of iommu_dma_map_page[3], we got
> > > 0x0000000054800000 instead of expected address: 0x0000000050800000[2].
> > > There is a address offset(0x4000000). If we change to use
> > > dma_map_page_attrs with phys_to_page(addr), the address is correct as we
> > > expected[2]. Do you have any suggestion on this issue? Do we miss
> > > something?
> >
> > Sorry for the late reply. Could you show me the code changes you made
> > to use dma_map_single()? It would sound like the virtual address
> > passed to dma_map_single() isn't correct.
> >
> > Best regards,
> > Tomasz
> >
>
>
> Please check the below code snippet in today's testing.
>
> p1_dev->cam_dev.smem_dev = &p1_dev->scp_pdev->dev;
> ptr = dma_alloc_coherent(p1_dev->cam_dev.smem_dev,
> MTK_ISP_COMPOSER_MEM_SIZE, &addr, GFP_KERNEL);
> if (!ptr) {
> dev_err(dev, "failed to allocate compose memory\n");
> return -ENOMEM;
> }
> p1_dev->composer_scp_addr = addr;
> p1_dev->composer_virt_addr = ptr;
> dev_info(dev, "scp addr:%pad va:%pK\n", &addr, ptr);
>
> /* get iova address */
> addr = dma_map_single(dev, ptr, MTK_ISP_COMPOSER_MEM_SIZE,
> DMA_BIDIRECTIONAL);
> if (dma_mapping_error(dev, addr)) {
> dma_free_coherent(p1_dev->cam_dev.smem_dev,
> MTK_ISP_COMPOSER_MEM_SIZE,
> ptr, p1_dev->composer_scp_addr);
> dev_err(dev, "Failed to map scp iova\n");
> ret = -ENOMEM;
> goto fail_free_mem;
> }
> p1_dev->composer_iova = addr;
> dev_info(dev, "scp iova addr:%pad\n", &addr);
>
> Moreover, below is extracted log[2].
>
> We guess the virtual address which is returned by dma_alloc_coherent
> function is not valid kernel logical address. It is actually returned by
> memremap() in dma_init_coherent_memory(). Moreover, dma_map_single()
> will call virt_to_page() function. For virt_to_page function, it
> requires a logical address[1].
>
> [1]https://www.oreilly.com/library/view/linux-device-drivers/0596005903/ch15.html
>
Indeed virt_to_page() works only with kernel LOWMEM addresses. Whether
virt_to_page() is the right thing to do in dma_map_single() is a good
question, but let's assume it was implemented like this for a reason.
However, you also can't call phys_to_page() on the DMA addresses
returned by dma_alloc_*() either. It works just by luck, because SCP
DMA addresses and CPU physical addresses are numerically the same.
Could you try dma_get_sgtable() with the SCP struct device and then
dma_map_sg() with the P1 struct device?
Best regards,
Tomasz
> [2]
> 322 [ 1.238269] mtk-cam-p1 1a006000.camisp: scp
> addr:0x0000000052000000 va:00000000a3adc471
> 323 [ 1.239582] mtk-cam-p1 1a006000.camisp: scp iova
> addr:0x00000000fde00000
> 7716 [ 1.238963] mtk-cam-p1 1a006000.camisp: scp
> addr:0x0000000052000000 va:0000000042ec580f
> 7717 [ 1.240276] mtk-cam-p1 1a006000.camisp: scp iova
> addr:0x00000000fde00000
> 15088 [ 1.239309] mtk-cam-p1 1a006000.camisp: scp
> addr:0x0000000052000000 va:000000005e5b3462
> 15089 [ 1.240626] mtk-cam-p1 1a006000.camisp: scp iova
> addr:0x00000000fde00000
>
> Best regards,
>
> Jungo
>
> > >
> > > [1]
> > > [ 1.344786] __dma_alloc_from_coherent: 0x800000 PAGE_SHIFT:12
> > > device_base:0x0000000050000000 dma:0x0000000050800000
> > > virt_base:ffffff8014000000 va:ffffff8014800000
> > >
> > > [ 1.346890] mtk-cam 1a000000.camisp: scp addr:0x0000000050800000
> > > va:ffffff8014800000
> > >
> > > [ 1.347864] iommu_dma_map_page:0x0000000054800000 offset:0
> > > [ 1.348562] mtk-cam 1a000000.camisp: iova addr:0x00000000fde00000
> > >
> > > [2]
> > > [ 1.346738] __dma_alloc_from_coherent: 0x800000 PAGE_SHIFT:12
> > > device_base:0x0000000050000000 dma:0x0000000050800000
> > > virt_base:ffffff8014000000 va:ffffff8014800000
> > > [ 1.348841] mtk-cam 1a000000.camisp: scp addr:0x0000000050800000
> > > va:ffffff8014800000
> > > [ 1.349816] iommu_dma_map_page:0x0000000050800000 offset:0
> > > [ 1.350514] mtk-cam 1a000000.camisp: iova addr:0x00000000fde00000
> > >
> > >
> > > [3]
> > > dma_addr_t iommu_dma_map_page(struct device *dev, struct page *page,
> > > unsigned long offset, size_t size, int prot)
> > > {
> > > phys_addr_t phys = page_to_phys(page);
> > > pr_err("iommu_dma_map_page:%pa offset:%lu\n", &phys, offset);
> > >
> > > return __iommu_dma_map(dev, page_to_phys(page) + offset, size, prot,
> > > iommu_get_dma_domain(dev));
> > > }
> > >
> > > [snip]
> > >
> > > Best regards,
> > >
> > > Jungo
> > >
> >
> > _______________________________________________
> > Linux-mediatek mailing list
> > Linux-mediatek@xxxxxxxxxxxxxxxxxxx
> > http://lists.infradead.org/mailman/listinfo/linux-mediatek
>
>
