> --------- Original Message --------- > Sender : Christoph Hellwig <hch@xxxxxxxxxxxxx> > Date : 2022-03-31 00:45 (GMT+9) > Title : Re: [PATCH] PCI: dwc: Modify the check about MSI DMA mask 32-bit > > On Wed, Mar 30, 2022 at 11:35:26AM +0200, Alexander Lobakin wrote: >> I'm not super familiar with the DMA internals, so adding Chris here, >> maybe he'd like to comment, but anyway, the lower/arch layer must >> not give the DMA addresses wider than the number of bits passed to >> dma_set_mask() if that call returned 0. > > So, the basic assumption in the kernel is that 32-bit DMA is always > supported, and dma_set_maks for that should not fail. If the > system (or root port, internal interconnect) supports less than that > we'll bounce buffer. But how and why would you hand out addresses > larger than that? It really is not valid, but I can't even see how > it could happen. Hello, thank you for your review. Yes, the dma address should not be used in excess of the mask value set through dma_set_mask(). If I want to use 33+bit dma address on 64bit system, I have to call dma_set_mask(33+) again after below code in dw_pcie_host_init() is performed. This is because dw_pcie_host_init(32) is always called in dw_pcie_host_init() without any conditions. Is this right? Also, if I assign 33+bit dma address before dw_pcie_host_init() to use 33+bit dma address on 64bit system, dma_set_mask(32) is called and dma_mask is changed to 32 but dma address is maintained 33+, so error occurs. it is not a issue with the dma_set_mask() function, but the called condition must need to be modified. Thank you.
