Hi Mashiro, On Tue, 2016-11-08 at 10:50 +0900, Masahiro Yamada wrote: > 2016-11-07 21:52 GMT+09:00 Paul Bolle <pebolle@xxxxxxxxxx>: > > So it seems the odd $(LINUXINCLUDE) variable in that Makefile could be > > replaced with something like: > > -include $(srctree)/include/generated/autoconf.h > > This would break O= build because autoconf.h is a generated file. > > Rather, it should be > -include $(objtree)/include/generated/autoconf.h Three cheers for weasel words like "something like"! > I thought of this at first, but I was not quite sure > if the file path include/generated/autoconf.h is a guaranteed interface. > > Basically, now we are supposed to include autoconf.h via kconfig.h. Yes, that seems to go back to commit 2a11c8ea20bf ("kconfig: Introduce IS_ENABLED(), IS_BUILTIN() and IS_MODULE()"). And when the current approach to the IS_*() macros was introduced - with that breathtaking hack that introduced __is_defined() - this was no longer needed but was not changed again. > So, I thought $(LINUXINCLUDE) is a more stable interface > than specifying the exact path to autoconf.h > > I doubt that nobody would try to change it, but it is just two my cents. A bit of cruft accumulated around LINUXINCLUDE: a few dubious uses of it (and I think this is one of those); typos (ie, LINUX_INCLUDE); the pointless USERINCLUDE; things like that. It would be nice to remove that cruft. But it needs to be done carefully. > Anyway, arch/x86/boot/Makefile already > referenced the path to autoconf.h > > So, if you want to change it, I will not oppose to it. Paul Bolle -- To unsubscribe from this list: send the line "unsubscribe linux-s390" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html