Re: [tip:locking/core] locking/refcount: Create unchecked atomic_t implementation

From: Alexey Dobriyan
Date: Tue Sep 05 2017 - 14:15:50 EST


On Mon, Sep 04, 2017 at 09:36:43PM +0200, Peter Zijlstra wrote:
> On Mon, Sep 04, 2017 at 08:34:44PM +0300, Alexey Dobriyan wrote:
> > On Mon, Sep 04, 2017 at 02:37:24PM +0200, Peter Zijlstra wrote:
> > > - once.o refcount.o usercopy.o errseq.o
> > > -obj-y += string_helpers.o
> > > + once.o usercopy.o errseq.o string_helpers.o hexdump.o kstrtox.o
> > > +
> > > +obj-$(CONFIG_REFCOUNT_FULL) += refcount.o
> > > obj-$(CONFIG_TEST_STRING_HELPERS) += test-string_helpers.o
> > > -obj-y += hexdump.o
> > > obj-$(CONFIG_TEST_HEXDUMP) += test_hexdump.o
> > > -obj-y += kstrtox.o
> >
> > File per line so everything is not lumped together even if it is obj-y.
>
> If that is policy mass convert everything and be done with it. Otherwise
> I'll continue to use the predominant pattern, which in this case is a
> giant obj-y +=. But mixed stuff is terribly annoying.

It is not a policy, just a "grassroot" movement to give people good
example and wait until someone is annoyed enough to mass convert
everything. -)

Worked for fs/proc/Makefile .