Re: [ft-devel] Curb the scaling

[Werner LEMBERG]

> As per our discussion I propose the following patch to limit the
> outline scaling so that ppem values do not exceed 65535.  As far as
> I can see, no such limits are currently in place, ppem values are
> simply truncated while the scaling factors are not limited.

Looks good.

> My secret goal is to limit those integer addition/subtraction
> overflows in the future and, perhaps, revisit the old ones.  To that
> effect, if this is accepted, I propose turning the ADD_ and SUB_
> macros back to normal arithmetic and see how many overflows truly
> survive.  I hope to see some of the macros gone if they are
> triggered by crazy scaling.

Kostya, is there a possibility to make the fuzzer temporarily watch a
branch of FreeType also, using the already created set of malformed
files of master?  Then Alexei's suggestion could be tested quite
easily.

Or to ask differently: Where can we download the complete set of
malformed files that are currently in use by the fuzzer?  Then Alexei
could easily run a clang binary by himself.


   Werner