On 3/22/19 9:34 AM, Wolfram Sang wrote: > >> - rcar_pci_write_reg(pcie, 0, PCIEMSIAUR); >> + rcar_pci_write_reg(pcie, base >> 32, PCIEMSIAUR); > > Seeing the in-depth discussion between you and Geert, I'll leave this to > you guys :) I think this fix is clear, the discussion is now rather about whether we should allocate the reserved page with GFP_DMA32 and keep it in 32bit address range or not. I'll reword the commit message though. -- Best regards, Marek Vasut
