13 Apr
2011
13 Apr
'11
1:04 p.m.
On Wed, Mar 2, 2011 at 11:22 PM, matt.waddel@linaro.org wrote:
+static int convert_from_bytes_to_power_of_two(unsigned int x) +{
- int y = 0;
- y = (x & 0xAAAA) ? 1 : 0;
- y |= ((x & 0xCCCC) ? 1 : 0) << 1;
- y |= ((x & 0xF0F0) ? 1 : 0) << 2;
- y |= ((x & 0xFF00) ? 1 : 0) << 3;
- return y;
+}
Let's not try to be this clever. U-Boot already has a function that tells you which power of two a number corresponds to. Unless the caller of this will be highly sensitive to several cycles that you may gain from knowing that the input is already a power-of-two, there's just no cause to creating a separate implementation of ffs().
Other than that, it looks fine, and as soon as you respin, I'll add it to the mmc tree.
Andy