On 30/11/2017 13:51, Thomas Huth wrote: > The two micrablaze data arrays are completely different, since one is > big endian, the other is little, so I'd need to byte-swap the whole > array on the fly, too. Not sure whether it makes sense to add such > complex code just to safe 16 bytes of blob data...?
*facepalm* :) >> Likewise, you could have just two copies of the code for all ARM boards >> that have a pl011 (or any other UART with a simple byte-long transmit >> register), one 32-bit and one 64-bit. > > OK, having a patching mechanism in place likely makes sense as soon as > we want to include multiple ARM boards here. I can do that, but I'd > rather like to do it as a second step. It was already quite hard work to > come up with all these assembler snippets (from CPUs where I don't have > a clue of) and to determine which machines could be tested this way at > all, so I'd first like to wait and see whether this patch series is > acceptable at all or not, since Peter has objections. Given your remark on endianness then no, having one snippet per binary is enough. Paolo
