On 11-05-18, 19:58, Baolin Wang wrote:

On 11 May 2018 at 19:53, Vinod Koul [off-list ref] wrote:

quoted

quoted

quoted

quoted

+/*
+ * struct sprd_dma_config - DMA configuration structure
+ * @cfg: dma slave channel runtime config
+ * @src_addr: the source physical address
+ * @dst_addr: the destination physical address
+ * @block_len: specify one block transfer length
+ * @transcation_len: specify one transcation transfer length
+ * @src_step: source transfer step
+ * @dst_step: destination transfer step
+ * @wrap_ptr: wrap pointer address, once the transfer address reaches the
+ * 'wrap_ptr', the next transfer address will jump to the 'wrap_to' address.
+ * @wrap_to: wrap jump to address
+ * @req_mode: specify the DMA request mode
+ * @int_mode: specify the DMA interrupt type
+ */
+struct sprd_dma_config {
+     struct dma_slave_config cfg;
+     phys_addr_t src_addr;
+     phys_addr_t dst_addr;

these are already in cfg so why duplicate, same for few more here.

We save them in 'struct sprd_dma_config' as one parameter for
sprd_dma_config(), otherwise we need add 2 more parameters (src and
dst) for sprd_dma_config().

I am not sure I follow...

I meant you can use sprd_dma_config.cfg->src_addr/dst_addr and remove src_addr &
dst_addr in the sprd_dma_config. It duplicate..

But we can not set values if dir == DMA_MEM_TO_DEV, since they
represent the device address, right?
sprd_dma_config.cfg->src_addr = sg_dma_address(sg);

Ah right and that is another issue to fix then :) we should not store addresses
you get for prep_ calls. They are supposed to be stored in the descriptor.
Current approach is not correct.

You can get different prepare calls  (i think fixing this will remove 1 sg
limitation).. and use the argument values and store in respective descriptors.
Pls do see the other driver examples..