"Christophe Leroy (CS GROUP)" <[email protected]> writes: > Le 23/03/2026 à 08:32, Venkat Rao Bagalkote a écrit : >> v2: >> - Added missing Suggested-by tag from Ritesh Harjani (IBM) >> > > Patch history must go _after_ the --- below, otherwise it will appear in > the commit message when applied, which is pointless. > >> drivers/char/nvram.c defines a static mutex 'nvram_mutex' which is never >> used. This results in a compiler warning on linux-next builds: > > It is probably not only linux-next builds, I think the problem exists > since 20e07af71f34 ("powerpc: Adopt nvram module for PPC64") > >> >> warning: 'nvram_mutex' defined but not used [-Wunused-variable] >> >> Remove the unused definition to avoid the warning. > > It is not what you are doing. > > You are just hiding the probleme by saying 'maybe it is used, maybe it > is not used, I don't know I don't care".
Venkat, do cares about this warning, and hence he sent the patch in trying to fix it ;) I think, I missed seeing the upper #ifdef block of PPC, and hence suggested him to use __maybe_unused, instead of complicating it further with... #if defined(CONFIG_PPC32) || defined(CONFIG_X86) || defined(CONFIG_M68K), > Please properly fix the problem instead. > I agree, make sense. > I think the fix is probably to remove the #ifdef CONFIG_PPC32 around > IOC_NVRAM_SYNC. > If you think it is important to return -ENOTTY on CONFIG_PPC64, just add: > That make sense and I should have thought of that. However, I looked at the suggestions from Arnd, and I too agree that all underneath function operations already do their own locking, so I agree that we could just kill this nvram_mutex lock itself. -ritesh
