On Sat, Oct 17, 2020 at 02:48:56PM +0200, Arnd Bergmann wrote: > On Fri, Oct 16, 2020 at 4:12 PM Andrew Lunn <andrew@xxxxxxx> wrote: > > > > > One drawback of your approach is that > > > you cannot set KBUILD_CFLAGS_W1_20200930 > > > until you eliminate all the warnings in the > > > sub-directory in interest. > > > (i.e. all or nothing approach) > > > > Hi Mashiro > > > > That actual works well for my use case. drivers/net/ethernet is W=1 > > clean. So is drivers/net/phy, drivers/net/mdio. Developers generally > > clean up a subsystem by adding W=1 to the command line because that is > > the simple tool they have. > > > > And my aim here is to keep those subsystem W=1 clean. I don't care > > about individual warnings within W=1, because those subsystems are > > passed that stage already. > > I tried to get a better grasp of what kind of warnings we are actually talking > about and looked at the x86 allmodconfig W=1 output on today's linux-next. Hi Arnd The work done to cleanup drivers/net/ethernet was mostly done by an Intel team. When built for ARM there are few warnings left, mostly due to missing COMPILE_TEST. I have fixes for that. But this raises the question, can we be a bit more tolerant of warnings for not x86 to start with? 0-day should help us weed out the remaining warnings on other architectures. As for the plan, it looks O.K. to me. I can definitely help with driver/net and net. Andrew
