On Fri, Jul 26, 2019 at 12:27 AM Jan Kiszka <jan.kiszka@xxxxxxxxxxx> wrote: > > On 25.07.19 17:18, Masahiro Yamada wrote: > > Hi Jan, > > > > On Thu, Jul 25, 2019 at 5:39 PM Jan Kiszka <jan.kiszka@xxxxxxxxxxx> wrote: > >>> > >> > >> This affects also external modules builds: I have patterns here that do > >> > >> [Makefile] > >> subdir-y := some-module > >> > >> [some-module/Makefile] > >> obj-m := some-module.o > >> > >> and since this patch, the final some-module.ko is no longer built. Am I missing > >> something in the kbuild makefiles, or is this a regression? > > > > Thanks for the report. > > Interesting. I have never imagined that Makefiles were written like that. > > > > I just wrote a fix-up, but I have not determined to apply it. > > https://patchwork.kernel.org/patch/11059033/ > > > > It is easy to fixup your Makefile, though. > > Thanks for addressing this quickly! I'm happy to adjust our code [1]. Is the > suggested pattern usable with recent stable kernels as well, say down to 4.4 at > least? Yes, should work for any kernel version. > Jan > > [1] https://github.com/siemens/jailhouse/blob/master/Kbuild#L54 Maybe, obj-m := driver/ hypervisor/ configs/ inmates/ tools/ and $(patsubst %/,$(obj)/%,$(obj-m)): $(GEN_CONFIG_MK) Not tested at all. -- Best Regards Masahiro Yamada
