Coverity complained that asBmpDigit() may return a negative number, so assigning that to an unsigned variable may thus underflow. The apparent intention of the code is to use the unsigned variable to map negative values to values ≥ 10, to save on one comparison. This is clever™ code, and if Coverity doesn't understand it, the next programmer reading the code likely won't, either. Fix by receiving the result of asBmpDigit() unchanged and explicitly checking for ≥ 0 and < 10. Optimizing these two into a single unsigned comparison is something that the compiler will do automatically, not something the programmer needs to bother about. Even GCC 8.1, even at -O1, turns this into a single cmp <reg>, 9 ja <target> Amends 4931fe9b02f68d59a3a68f4eaf5cf74695686add. Coverity-Id: 474086 Pick-to: 6.10 6.9 6.8 Change-Id: I3df36eb3f545ca0689f2008d45693176f96bfb73 Reviewed-by: Edward Welbourne <edward.welbourne@qt.io> Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>