From: heiko@sntech.de (Heiko =?utf-8?q?St=C3=BCbner?=)
Date: Mon, 3 Jun 2013 23:39:38 +0200
Subject: [PATCH RFC 2/3] clk: dt: binding for basic multiplexor clock
In-Reply-To: <201306032215.45983.heiko@sntech.de>
References: <1370281990-15090-1-git-send-email-mturquette@linaro.org>
 <20130603200722.6077.85426@quantum> <201306032215.45983.heiko@sntech.de>
Message-ID: <201306032339.39176.heiko@sntech.de>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

Am Montag, 3. Juni 2013, 22:15:45 schrieb Heiko St?bner:
> Am Montag, 3. Juni 2013, 22:07:22 schrieb Mike Turquette:
> > Quoting Heiko St?bner (2013-06-03 12:33:19)
> > 
> > > Hi Mike,
> > > 
> > > I think it's a multiplexEr clock in the patch title, and see below
> > 
> > Doh, you are right.  But "xor" is so much cooler looking than "xer"...
> > 
> > > Am Montag, 3. Juni 2013, 19:53:09 schrieb Mike Turquette:
> > > > Device Tree binding for the basic clock multiplexor, plus the setup
> > > > function to register the clock.  Based on the existing fixed-clock
> > > > binding.
> > > > 
> > > > Also relocate declaration of of_fixed_factor_clk_setup to keep things
> > > > tidy.
> > > > 
> > > > Signed-off-by: Mike Turquette <mturquette@linaro.org>
> > > 
> > > [...]
> > > 
> > > > +
> > > > +     reg = of_iomap(node, 0);
> > > > +     pr_err("%s: reg is 0x%p\n", __func__, reg);
> > > > +
> > > > +     if (of_property_read_u32(node, "mask", &mask)) {
> > > > +             pr_err("%s: missing mask property for %s\n", __func__,
> > > > node->name); +             return;
> > > > +     }
> > > > +
> > > > +     if (of_property_read_u32(node, "shift", &shift))
> > > > +             pr_debug("%s: missing shift property defaults to zero
> > > > for %s\n", +                             __func__, node->name);
> > > > +
> > > > +     if (of_property_read_bool(node, "index_one"))
> > > > +             clk_mux_flags |= CLK_MUX_INDEX_ONE;
> > > > +
> > > > +     clk = clk_register_mux_table(NULL, clk_name, parent_names,
> > > > num_parents, +                     0, reg, 0, mask, clk_mux_flags,
> > > > 
> > >                       ^- should probably be shift
> > > 
> > > Otherwise looks cool and I'm currently trying it with my Rockchip code.
> > 
> > Right again.  My test platform seems to not shift the mask at all so
> > this did not cause a visible bug for me.
> 
> I'm currently converting my rockchip stuff to it, as it would solve a lot
> of problems with the numerous clock types I'm defining just to put the
> shift and mask values somewhere.
> 
> So lets see if anything more turns up ;-)

yep ... of_mux_clk_setup is quite talkative ... there are quite some (debug-?) 
pr_errs in it ;-)


Heiko