On Sun, 2015-12-06 at 01:21 +0100, Rasmus Villemoes wrote: > On Sun, Dec 06 2015, Joe Perches <joe@xxxxxxxxxxx> wrote: > > > On Sun, 2015-12-06 at 01:05 +0100, Rasmus Villemoes wrote: > > > If char is signed and ch happens to be negative, printing ch with > > > "%02x" will not do as intended (when ch is -19, one will get > > > "ffffffed"). Fix that by masking with 0xff. > > > > I presume there are a lot of these in the kernel. > > Did you use a tool to find this or just inspection? > > Initially I just used coccinelle, for the most obvious candidates > (with > --include-headers-for-types): > > @r depends on !patch@ > char c; > @@ > * \( sprintf \| snprintf \| scnprintf \) (..., c, ...) > > That gives lots of false positives (arguments to %c), but it's not > too > bad piping to less, searching for "%[0.]2[xX]", and then checking > manually. > > I'm now doing a much wider range of printf functions, but it's really > past my bedtime, so feel free to pick up the ball :-) > Thanks, but no thanks. It's your ball. Enjoy... You might consider looking for s8 and s16 emitted as %u too. _______________________________________________ devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxx http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
