I know that this is an Edison thread, but thought since I have a Galileo Gen 2 and the code looks very similar, except it does not appear to have the global tristate object, that I would test the ping out on it as well and also make similar changes.... So far with the MRAA code base.
I could put up Logic Analyzer pictures showing before and after, but would look more or less the same as the Edison ones.
Timing showed for the Change from Output to Input went from about 1.9ms to now about .62.
I uploaded the changes to the branch gpio-dir of my fork of MRAA.
The thing that took me the longest time was being able to buld MRAA. I tried downloading the linux version of cmake, but it did not want to run, so I downloaded the sources and built cmake at which point I could then build MRAA...
Kurt