================ @@ -0,0 +1,21 @@ +// RUN: %clang_analyze_cc1 -std=c++20 \ +// RUN: -analyzer-checker=core,alpha.deadcode.UnreachableCode \ +// RUN: -analyzer-config experimental-enable-naive-ctu-analysis=true \ +// RUN: -analyzer-config max-nodes=10 \ +// RUN: -verify=ctu-on %s +// ctu-on-no-diagnostics ---------------- NagyDonat wrote:
> I fear that CTU was enabled here, but was ineffective. Could you demonstrate > that this in fact enables CTU that does work. Demonstrate it by a CTU report. According to the `CoreEngine` constructor code, enabling CTU always creates a `CTUWorkList` even if there are actually no other relevant translation units. I'm convinced that this test does correctly validate the intended behavior – but you are right that perhaps it would be nice to showcase a CTU report (e.g. by copying a trivial test from one of the CTU test files) as a visible demonstration. > Could you also demonstrate that the `alpha.deadcode.UnreachableCode` works > otherwise to establish a baseline that its effective. This is tested in other test files (or at least I very much hope so :sweat_smile:), but I'm not opposed to having a trivial test that produces an `UnreachableCode` report. https://github.com/llvm/llvm-project/pull/219225 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
