[PATCH] D13126: New static analyzer checker for loss of sign/precision

Daniel Marjamäki via cfe-commits cfe-commits at lists.llvm.org
Tue Mar 22 05:57:07 PDT 2016


danielmarjamaki added a comment.

In http://reviews.llvm.org/D13126#379306, @zaks.anna wrote:

> Why is there such a large jump in the number of warnings reported in the last patch iteration?
>  It went from "1678 projects where scanned. In total I got 124 warnings" to "In 2215 projects it found 875 warnings." Did the number of warnings in the initial 1678 projects stay the same?


No I improved the check.

My previous patch was limited: I have limited this patch hoping that it will be easier to triage and review. Right now I only warn if there is assignment and RHS is a DeclRefExpr and only for loss of precision.

> Is it possible to take a look at the nature of the false positives, as per NoQ's request above?


I'll try when I get some time.

> This checker would benefit greatly from explaining why the errors occur. For example, where the values of variables are being constrained. Other checkers use BugReporterVisitor to generate the rich diagnostic information. Dow you have plans on how to accomplish that for this checker?


Yes that sounds like a good idea.. I''ll try to look at that.


http://reviews.llvm.org/D13126





More information about the cfe-commits mailing list