On Mon, Apr 30, 2018 at 12:45:07PM -0400, Theodore Y. Ts'o wrote:
On Mon, Apr 30, 2018 at 10:24:53AM +0200, Lukas Czerner wrote:
Well, so much for a discussion :-/ The patch is upstream now...
Sorry, it's been crazy busy and there were other fixes I really needed to get upstream.
Yeah, I understand.
If it really is superfluous, the compiler will optimize it away. I have been bitten more times than I can count with fuzz testers and C's _fine_ signed vs. unsigned integer semantics. So I prefer to be careful here.
If we can really be sure it's safe, we can always remove the extra comparison in a separate patch, since that's unrelated to the real bug we're trying to fix.
That's fine, I understand that it was better to fix it sooner rather than later and deal with the optimization at a different time. Short reply stating that would have helped though, there is my sign-off below the patch as well after all.
-Lukas
- Ted