On Tue, 17 Nov 2026 19:17:12 +0000 Bruce Richardson <[email protected]> wrote:
> On Tue, Nov 17, 2026 at 09:40:59AM -0800, Stephen Hemminger wrote: > > Clang evalutes strlen of constant string at compile time, then > > complains about unreachable code. Use extra parenthesis to > > mark code as explcitly dead. > > > > Bugzilla ID: 2135 > > > > Signed-off-by: Stephen Hemminger <[email protected]> > > --- > > lib/eal/common/rte_version.c | 4 +++- > > 1 file changed, 3 insertions(+), 1 deletion(-) > > > > diff --git a/lib/eal/common/rte_version.c b/lib/eal/common/rte_version.c > > index 627b89d4a8..62572428fc 100644 > > --- a/lib/eal/common/rte_version.c > > +++ b/lib/eal/common/rte_version.c > > @@ -36,7 +36,9 @@ rte_version(void) > > static char version[32]; > > if (version[0] != 0) > > return version; > > - if (strlen(RTE_VER_SUFFIX) == 0) > > + > > + /* extra paren to silence unreachable warnings */ > > + if ((strlen(RTE_VER_SUFFIX) == 0)) > > snprintf(version, sizeof(version), "%s %d.%02d.%d", > > RTE_VER_PREFIX, > > RTE_VER_YEAR, > > -- > > Is it guaranteed that the checker won't look inside the double brackets? An > alternative suggestion is to change RTE_VER_SUFFIX to the function call to > the (public) API rte_version_suffix() and use strlen on that. We can even > mark the function as __rte_noinline to avoid optimizer getting rid of the > call in future and reintroducing the warning. > > Also, when fixing this, I think it would be nice to also fix the indent of > the else leg. > > /Bruce Clang suggests it as a fix so I think that should always work. Gcc unreachable flag is a noop, the compiler structure makes it too hard for them to do the analysis. Ok, can fix the indent as well.

