On 06/21/2013 06:23 PM, Arnd Bergmann wrote: >> static void ace_datain_8(struct ace_device *ace) >> @@ -248,7 +248,7 @@ static void ace_datain_8(struct ace_device *ace) >> u8 *dst = ace->data_ptr; >> int i = ACE_FIFO_SIZE; >> while (i--) >> - *dst++ = in_8(r++); >> + *dst++ = ioread8(r++); >> ace->data_ptr = dst; >> } > > Why not just use ioread8_rep() to replace the loop? My intention with this first patch was to do a simple iteration that is supposed to not break anything. As it it said in patch title - just replace accessors. And for now (for this patch) leave improper implemented functionality as it is. And in the next patch I will introduce changes that you've suggested with "_rep" accessors. Maybe I should have sent both patches at once so my intention would be clear. > > I guess the last two modifications are actually needed for correct > operation independent of CPU endianess. I don't what led to the > combination of big- and little-endian accessors, but my guess is > that it was trial-and-error together with cut-and-paste. The point > is that the single register access should match the endianess of > the device while the streaming access should match the endianess > of the CPU. Completely agree. -Alexey -- To unsubscribe from this list: send the line "unsubscribe linux-arch" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
