On Fri, 2014-07-25 at 07:26 +0300, Siarhei Siamashka wrote:
- number_of_lanes = mctl_get_number_of_lanes();
There is a subtle functional change here since number_of_lanes can be 1 whereas n could never have been 2. Is that intended/ok? Please mention in the commit message.
I tried to experiment with setting the 8-bit bus width and it is semi-workable (single byte access is OK, but accessing more than one byte is broken). This part of the patch looks like a forgotten leftover of these experiments. But it clearly has no practical value and we only normally deal with the 16-bit or 32-bit bus width.
The most correct way of handling this unexpected code branch would be to panic. But that's an unnecessarily increase of the code size. So I think that the best solution is just to keep the old code logic (expect only 16-bit or 32-bit bus width and 2 or 4 lanes).
Not sure how much a panic actually adds to the code size or if it is worth worrying about, but removing the 8-bit stuff is fine by me anyway.
Ian.