Logical cores iteration

[Jimmy-653]

Basically copied from you comment.....
Seems kinda hacky to get a PR that way.....

[coveralls]

Coverage remained the same at 87.975% when pulling d39bbfb on J-Jimmy:upstream/1360_bugfix_64_cores into f131fa1 on oshi:master.

[dbwiddis]

It's not hacky to do an important bug fix! But you need a changelog entry and probably a better title for this PR.

Also I'm going to hold off on merging this while resolving #1361 which is probably in that same iteration. I suspect using i from the bitmask should be replaced by an incrementing (from 0) counter for each core instead, and it'd be nice to fix both problems in one PR.

[Jimmy-653]

I'm really not following all this code about CPU cores....
And I have no environment to test/reproduce and analyze the behaviour in order to fix it.

[dbwiddis]

I don't have a test environment either but I understand the problem well enough to describe the fix, and the user can test it.

I'll go ahead and merge this.

[dbwiddis]

Or not. Sorry, as I dig into this I keep finding more. The fix you did so far applies for the pre-Win7 branch of the code.

Same fix needs to be applied in getMatchingPackage(), getMatchingNumaNode(), and getMatchingCore(). Everywhere you see ... mask ... & (1L << lp)) > 0 should be ... mask ... & (1L << lp)) != 0

[dbwiddis]

And also in getLogicalProcessorInformationEx()