Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230321074249.2221674-1-leobras@redhat.com> <20230321074249.2221674-2-leobras@redhat.com>
 <ZBpPms0PGv4sWR1Y@andrea>
In-Reply-To: <ZBpPms0PGv4sWR1Y@andrea>
From:   Leonardo Bras Soares Passos <leobras@redhat.com>
Date:   Wed, 22 Mar 2023 00:50:45 -0300
Message-ID: <CAJ6HWG46-an3r3WH+n8mHHge8SjqgMbohjTmmk=_EXVmwKZSeA@mail.gmail.com>
Subject: Re: [RESEND RFC PATCH v2 1/6] riscv/cmpxchg: Deduplicate cmpxchg()
 asm functions
To:     Andrea Parri <parri.andrea@gmail.com>
Cc:     Paul Walmsley <paul.walmsley@sifive.com>,
        Palmer Dabbelt <palmer@dabbelt.com>,
        Albert Ou <aou@eecs.berkeley.edu>, Guo Ren <guoren@kernel.org>,
        linux-riscv@lists.infradead.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Tue, Mar 21, 2023 at 9:45=E2=80=AFPM Andrea Parri <parri.andrea@gmail.co=
m> wrote:
>
> On Tue, Mar 21, 2023 at 04:42:44AM -0300, Leonardo Bras wrote:
> > In this header every cmpxchg define (_relaxed, _acquire, _release,
> > vanilla) contain it's own asm file, both for 4-byte variables an 8-byte
> > variables, on a total of 8 versions of mostly the same asm.
> >
> > This is usually bad, as it means any change may be done in up to 8
> > different places.
> >
> > Unify those versions by creating a new define with enough parameters to
> > generate any version of the previous 8.
> >
> > (This did not cause any change in generated asm)
> >
> > Signed-off-by: Leonardo Bras <leobras@redhat.com>
> > ---
> >  arch/riscv/include/asm/cmpxchg.h | 102 ++++++++-----------------------
> >  1 file changed, 24 insertions(+), 78 deletions(-)
>
> Diff looks nice,

Hello Andrea, thanks for reviewing!


> > +#define ___cmpxchg(lr_sfx, sc_sfx, prepend, append)                  \
> > +{                                                                    \
> > +     __asm__ __volatile__ (                                          \
> > +             prepend                                                 \
> > +             "0:     lr" lr_sfx " %0, %2\n"                          \
> > +             "       bne  %0, %z3, 1f\n"                             \
> > +             "       sc" sc_sfx " %1, %z4, %2\n"                     \
> > +             "       bnez %1, 0b\n"                                  \
> > +             append                                                  \
> > +             "1:\n"                                                  \
> > +             : "=3D&r" (__ret), "=3D&r" (__rc), "+A" (*__ptr)         =
   \
> > +             : "rJ" ((long)__old), "rJ" (__new)                      \
> > +             : "memory");                                            \
> > +}
>
> Though I'm not really a fan of macros depending on local variables with
> "magic" names, can this be avoided?

You mean __ret, __rc, __ptr, __old and __new ?
If so, yes, we could add them to the macro signature, but it would
become quite big with not much to gain. To reduce impact it could be
something like:

+#define ___cmpxchg(lr_sfx, sc_sfx, prepend, append, ret, r, p, o, n)

Is this a possible fix?
>
> Thanks,
>   Andrea
>

Thank you!
Leo