[cfe-dev] Adding taint sources to GenericTaintChecker

[Artem Dergachev via cfe-dev]

On 12.04.2016 08:35, Ashwin Ganesh wrote:
 > when I use float instead of int, the taint doesn't propagate at all

Umm yeah, and then we run into a bit of an issue - the analyzer does not 
use symbols for floats yet. The problem was that constraints on floats 
are not yet supported, otherwise there shouldn't be a problem to replace 
UnknownVal's with float-type symbols. This is item #2 in 
http://clang-analyzer.llvm.org/open_projects.html . Maybe we should go 
ahead and produce symbols anyway, and ignore them in the constraint 
manager, solely for the purposes of taint analysis, before we have 
anything better? Not sure right now what kinds of false positives we may 
unleash this way.