On Wed, Mar 18, 2026 at 01:03:03PM +0100, Krzysztof Kozlowski wrote: > On 18/03/2026 12:58, Segher Boessenkool wrote: > > On Wed, Mar 18, 2026 at 12:54:30PM +0100, Krzysztof Kozlowski wrote: > >> On 18/03/2026 12:51, Segher Boessenkool wrote: > >>> On Tue, Mar 17, 2026 at 02:08:24PM +0100, Krzysztof Kozlowski wrote: > >>>> Return value of ps3_start_probe_thread() is not used, so code can be > >>>> simplified to fix W=1 clang warnings: > >>>> > >>>> arch/powerpc/platforms/ps3/device-init.c:953:6: error: variable > >>>> 'result' set but not used [-Werror,-Wunused-but-set-variable] > >>> > >>> And this is exactly why -Werror is harmful. A boring harmless warning > >>> turned into a build break :-( > >> > >> -Werror does not matter here. We do not want warnings either, assuming > >> of course they are correct. > >> > >> So if this warning is correct, then patch should be applied and WERROR > >> is irrelevant. > > > > The build is broken. -Werror is positively harmful. > > > > Yes, you want the warnings fixed, but maybe something else has priority > > right now? > > What do you mean? This is waiting on the list for long time - note that > it is a resend - so if warning is correct, for how long it should wait > to get to the "priority" box? > > And why anything needs even the priority here? Why priority has to be > discussed, especially for some legacy (I think?) code? Why correct > patches cannot be simply applied after some time/review?
It is totally simple: without -Werror, you can build a kernel. With it, you cannot. The ***warning*** is correct, there is something in the code that can be improved. That is why such things are warnings: the compiler can produce useful output, there is no reason to scream bloody murder. The main reason some people like -Werror is because they are too lazy to look at the warning messages they get from normal builds. Such developers need potty training, this is not a reason to punish all more reasonable people! Segher
