Hello Sören: > > Get rid of cdns_uart_readl() and cdns_uart_writel() and just call > > readl() and writel() directly. > > I don't know whether there are any ramifications from migrating from > io(read|write) to (read|write)l. For most use-cases I assume it is fine. > But there are some folks plugging Zynq platforms into their x86(_64) PCIe > slots and use Zynq peripherals from the x86 host. Could such a case be > affected in any way by this migration? The driver only supports memory-mapped I/O [by ioremap()], so readl/writel is actually the right thing to do, IMO. On x86 and ia64, ioread32/iowrite for MMIO addresses also maps to readl/writel. Are there any plans to support port-mapped I/O? My guess is that ioread32/iowrite32 was simply copied from some other driver long ago (uartlite?), and forgotten. I have looked at various Zynq peripheral drivers, and they all use some variant of readl/writel. The only exception is the CAN driver, drivers/net/can/xilinx_can.c, which uses ioread32/iowrite32 or ioread32be/iowrite32be, depending on endianness (similar to uartlite). I assume that this is because there is no big-endian version of readl/writel, so ioread32 vs. ioread32be looks better than, say, readl vs. be32_to_cpu(__raw_readl); or is there some other reason? Best regards, Thomas Betker -- To unsubscribe from this list: send the line "unsubscribe linux-serial" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
