I suspect, this code below explains why you are getting the false positives.
The issue you highlight in the example is that sometimes the analyzer doesn't
know what the value of a variable is. The existing checkers minimize false
positives by issuing a warning only when it's known that a value is "bad". For
example, we would only warn if StateOverflow && !StateNotOverflow. This will
flag much less issues, but should not produce a lot of false positives.
Are the false positives you are getting being flagged by the first if clause?
if (StateOverflow && StateNotOverflow) {
if (Pack.LValueIsTainted) {
Msg.assign("Possible integer overflow while " + Pack.Operation +
". Left operand is tainted: " + Pack.LValue + " AND " +
Pack.RValue);
reportBug(Msg, C, SL);
} else if (Pack.RValueIsTainted) {
Msg.assign("Possible integer overflow while " + Pack.Operation +
". Right operand is tainted: " + Pack.LValue + " AND " +
Pack.RValue);
reportBug(Msg, C, SL);
}
return;
}
if (StateOverflow) {
Msg.assign("Integer overflow while " + Pack.Operation + ". " + Pack.LValue +
" AND " + Pack.RValue);
reportBug(Msg, C, SL);
}
================
Comment at: lib/StaticAnalyzer/Checkers/IntegerOverflowChecker.cpp:35
@@ +34,3 @@
+ mutable std::unique_ptr<BuiltinBug> BT;
+
+ mutable std::set<SourceLocation> OverflowLoc;
----------------
j.trofimovich wrote:
> zaks.anna wrote:
> > Are you getting multiple reports on the same location? I don't think that
> > should be happening - the bug reporting infrastructure should unique
> > reports.
> In what way should bug reporting infrastructure unique reports? scan-build
> prevents existence of fully identical reports by computing digest
> (Digest::MD5->new->addfile(*FILE)->hexdigest; scan-build, line 247) but cases
> when alerts differs by message only aren't caught.
Identical issues should have the same message. Do you have identical issues
with different messages?
http://reviews.llvm.org/D4066
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
