RE: [PATCH E 08/14] OMAP clock: move rate recalc, propagation code up to plat-omap/clock.c

From: "Woodruff, Richard" <r-woodruff2@ti.com>
To: Russell King - ARM Linux <linux@arm.linux.org.uk>,
	Paul Walmsley <paul@pwsan.com>
Cc: "linux-arm-kernel@lists.arm.linux.org.uk" 
	<linux-arm-kernel@lists.arm.linux.org.uk>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"linux-omap@vger.kernel.org" <linux-omap@vger.kernel.org>,
	Tony Lindgren <tony@atomide.com>
Subject: RE: [PATCH E 08/14] OMAP clock: move rate recalc, propagation code up to plat-omap/clock.c
Date: Fri, 30 Jan 2009 08:23:49 -0600	[thread overview]
Message-ID: <13B9B4C6EF24D648824FF11BE89671620376DEC116@dlee02.ent.ti.com> (raw)
In-Reply-To: <20090130085249.GB6864@n2100.arm.linux.org.uk>

> There's one bug that your version highlights in mine - the virtual mpu
> clock in omap1 touches the DPLL and repropagates that rate.  I've
> removed that repropagation, so that needs fixing.
>
> However, this raises a question: why is the virtual mpu clock touching
> some other part of the clock tree.  I wonder whether this should be
> handled a different way, though the first thing that needs answering is
> why we have this alias for 'arm_ck' ?

At one point in time the virtual clock allowed control for a set of clocks with some dependencies.  The mpu alias provided a convenient control point.

Tukka did original omap1 in this format. I forget its exact meaning.

The original omap2 code had behavior:
        - mpu-dpll had a clock node which allowed direct mpu speed control
        - mpu-virt-clock allowed switching between sets that have fixed ratio dependency which were pegged to current mpu-dll speed.
                * This way round_rate would allow you to discover all valid OPPs.

The first pass OMAP3 code TI did followed omap2 but it wasn't strongly bound by ratio-sets given the multiple async-dplls. Paul has taken code in a bit different direction.  He can comment on current behavior here.

Regards,
Richard W.