Re: [PATCH] clk: mediatek: Allow changing PLL rate when it is off

[James Liao]

Hi Sascha,

On Fri, 2016-01-08 at 12:21 +0100, Sascha Hauer wrote:
> On Fri, Jan 08, 2016 at 05:48:53PM +0800, James Liao wrote:
> > On Fri, 2016-01-08 at 10:15 +0100, Sascha Hauer wrote:
> > > On Fri, Jan 08, 2016 at 04:16:37PM +0800, James Liao wrote:
> > > > Some modules may need to change its clock rate before turn on it.
> > > > So changing PLL's rate when it is off should be allowed.
> > > > This patch removes PLL enabled check before set rate, so that
> > > > PLLs can set new frequency even if they are off.
> > > 
> > > This sounds like the software refused to change the rate on disabled
> > > PLLs, but this is not the case.
> > 
> > In fact the major change of this patch is trigger (set) CON0_PCW_CHG no
> > matter PLL is on or not.
> > 
> > > > 
> > > > Signed-off-by: James Liao <jamesjj.liao@xxxxxxxxxxxx>
> > > > ---
> > > >  drivers/clk/mediatek/clk-pll.c | 9 ++-------
> > > >  1 file changed, 2 insertions(+), 7 deletions(-)
> > > > 
> > > > diff --git a/drivers/clk/mediatek/clk-pll.c b/drivers/clk/mediatek/clk-pll.c
> > > > index 966cab1..8e31fae 100644
> > > > --- a/drivers/clk/mediatek/clk-pll.c
> > > > +++ b/drivers/clk/mediatek/clk-pll.c
> > > > @@ -91,9 +91,6 @@ static void mtk_pll_set_rate_regs(struct mtk_clk_pll *pll, u32 pcw,
> > > >  		int postdiv)
> > > >  {
> > > >  	u32 con1, val;
> > > > -	int pll_en;
> > > > -
> > > > -	pll_en = readl(pll->base_addr + REG_CON0) & CON0_BASE_EN;
> > > >  
> > > >  	/* set postdiv */
> > > >  	val = readl(pll->pd_addr);
> > > > @@ -114,15 +111,13 @@ static void mtk_pll_set_rate_regs(struct mtk_clk_pll *pll, u32 pcw,
> > > >  
> > > >  	con1 = readl(pll->base_addr + REG_CON1);
> > > >  
> > > > -	if (pll_en)
> > > > -		con1 |= CON0_PCW_CHG;
> > > > +	con1 |= CON0_PCW_CHG;
> > > 
> > > This bit is described as "Feedback divide ratio update". To me this
> > > sounds like we have to inform the hardware that the PLL registers have
> > > been updated. The current code only sets this bit when the PLL is
> > > enabled which sounds sane to me.
> > > 
> > > >  
> > > >  	writel(con1, pll->base_addr + REG_CON1);
> > > >  	if (pll->tuner_addr)
> > > >  		writel(con1 + 1, pll->tuner_addr);
> > > >  
> > > > -	if (pll_en)
> > > > -		udelay(20);
> > > > +	udelay(20);
> > > 
> > > We seem to have to wait here until the PLL is really running at the new
> > > frequency. Normally we don't have to do this when the PLL is disabled.
> > > 
> > > I'm sure this patch solves a real problem, from looking at it it's just
> > > not clear to me what the problem is. Could you clarify this a bit?
> > 
> > On MT8173 for example, ARMPLL's enable bit can be controlled by other
> > HW. That means ARMPLL may be turned on even if we (CPU / SW) set
> > ARMPLL's enable bit as 0. In this case, SW may want and can still change
> > ARMPLL's rate by changing its pcw and postdiv settings. But without this
> > patch, new pcw setting will not be applied because its enable bit is 0.
> 
> Ok, thanks for explaining. Could you add that to the commit message?

OK. I'll send a new patch with these comments.


Best regards,

James