================ @@ -1,8 +1,23 @@ -// RUN: %clang_analyze_cc1 -analyzer-checker=core,deadcode.DeadStores,debug.Stats -verify -Wno-unreachable-code -analyzer-max-loop 4 %s +// RUN: %clang_analyze_cc1 \ +// RUN: -analyzer-checker=core,deadcode.DeadStores,debug.Stats \ +// RUN: -Wno-unreachable-code \ +// RUN: -verify=default %s + +// NOTE: analyzer-max-loop option is only meaningful if unroll-loops is false, +// that's why we do not pass it in the first case, as unroll-loops is +// true by default. ---------------- NagyDonat wrote:
This is not true -- `unroll-loops` only unrolls a subset of loops that are simple and predictable in some sense (e.g. do not perform any state splits where both branches are feasible). The option `analyzer-max-loop` is still important as it controls the behavior of the rest of the loops. Note that I didn't read this test file yet, I'm just reacting to this comment. https://github.com/llvm/llvm-project/pull/178911 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
