[PATCH] Fix iProc PLL output clock frequency calculation

[PATCH] Fix iProc PLL output clock frequency calculation

[Ray Jui]

This patch fixes incorrect iProc PLL clock frequency calculation. The issue
is exposed when Cygnus audio PLL was being added.

This patch is based on v4.3-rc5 and has been tested on Cygnus bcm958305k
wirelss audio board

This full tree is available here:
repo: https://github.com/Broadcom/cygnus-linux.git
branch: iproc-clk-pll-fix-v1

Simran Rai (1):
  clk: iproc: Fix PLL output frequency calculation

 drivers/clk/bcm/clk-iproc-pll.c | 13 +++++--------
 1 file changed, 5 insertions(+), 8 deletions(-)

-- 
1.9.1

[PATCH] clk: iproc: Fix PLL output frequency calculation

[Ray Jui]

From: Simran Rai <ssimran@broadcom.com>

This patch affects the clocks that use fractional ndivider in their
PLL output frequency calculation. Instead of 2^20 divide factor, the
clock's ndiv integer shift was used. Fixed the bug by replacing ndiv
integer shift with 2^20 factor.

Signed-off-by: Simran Rai <ssimran@broadcom.com>
Signed-off-by: Ray Jui <rjui@broadcom.com>
Reviewed-by: Scott Branden <sbranden@broadcom.com>
---
 drivers/clk/bcm/clk-iproc-pll.c | 13 +++++--------
 1 file changed, 5 insertions(+), 8 deletions(-)

diff --git a/drivers/clk/bcm/clk-iproc-pll.c b/drivers/clk/bcm/clk-iproc-pll.c
index 2dda4e8..d679ab8 100644
--- a/drivers/clk/bcm/clk-iproc-pll.c
+++ b/drivers/clk/bcm/clk-iproc-pll.c
@@ -345,8 +345,8 @@ static unsigned long iproc_pll_recalc_rate(struct clk_hw *hw,
 	struct iproc_pll *pll = clk->pll;
 	const struct iproc_pll_ctrl *ctrl = pll->ctrl;
 	u32 val;
-	u64 ndiv;
-	unsigned int ndiv_int, ndiv_frac, pdiv;
+	u64 ndiv, ndiv_int, ndiv_frac;
+	unsigned int pdiv;
 
 	if (parent_rate == 0)
 		return 0;
@@ -366,22 +366,19 @@ static unsigned long iproc_pll_recalc_rate(struct clk_hw *hw,
 	val = readl(pll->pll_base + ctrl->ndiv_int.offset);
 	ndiv_int = (val >> ctrl->ndiv_int.shift) &
 		bit_mask(ctrl->ndiv_int.width);
-	ndiv = (u64)ndiv_int << ctrl->ndiv_int.shift;
+	ndiv = ndiv_int << 20;
 
 	if (ctrl->flags & IPROC_CLK_PLL_HAS_NDIV_FRAC) {
 		val = readl(pll->pll_base + ctrl->ndiv_frac.offset);
 		ndiv_frac = (val >> ctrl->ndiv_frac.shift) &
 			bit_mask(ctrl->ndiv_frac.width);
-
-		if (ndiv_frac != 0)
-			ndiv = ((u64)ndiv_int << ctrl->ndiv_int.shift) |
-				ndiv_frac;
+		ndiv += ndiv_frac;
 	}
 
 	val = readl(pll->pll_base + ctrl->pdiv.offset);
 	pdiv = (val >> ctrl->pdiv.shift) & bit_mask(ctrl->pdiv.width);
 
-	clk->rate = (ndiv * parent_rate) >> ctrl->ndiv_int.shift;
+	clk->rate = (ndiv * parent_rate) >> 20;
 
 	if (pdiv == 0)
 		clk->rate *= 2;
-- 
1.9.1

Re: [PATCH] clk: iproc: Fix PLL output frequency calculation

[Stephen Boyd]

On 10/16, Ray Jui wrote:
> From: Simran Rai <ssimran@broadcom.com>
> 
> This patch affects the clocks that use fractional ndivider in their
> PLL output frequency calculation. Instead of 2^20 divide factor, the
> clock's ndiv integer shift was used. Fixed the bug by replacing ndiv
> integer shift with 2^20 factor.
> 
> Signed-off-by: Simran Rai <ssimran@broadcom.com>
> Signed-off-by: Ray Jui <rjui@broadcom.com>
> Reviewed-by: Scott Branden <sbranden@broadcom.com>
> ---

Did you want any Fixes: tag so this goes back to stable?

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project

Re: [PATCH] clk: iproc: Fix PLL output frequency calculation

[Ray Jui]

On 15-10-19 02:49 PM, Stephen Boyd wrote:
> On 10/16, Ray Jui wrote:
>> From: Simran Rai <ssimran@broadcom.com>
>>
>> This patch affects the clocks that use fractional ndivider in their
>> PLL output frequency calculation. Instead of 2^20 divide factor, the
>> clock's ndiv integer shift was used. Fixed the bug by replacing ndiv
>> integer shift with 2^20 factor.
>>
>> Signed-off-by: Simran Rai <ssimran@broadcom.com>
>> Signed-off-by: Ray Jui <rjui@broadcom.com>
>> Reviewed-by: Scott Branden <sbranden@broadcom.com>
>> ---
>
> Did you want any Fixes: tag so this goes back to stable?
>

Yes, we want this fix to go back to the stable kernel.

I will re-submit the patch with the Fixes: tag and at the same time cc 
stable@vger.kernel.org .

Thanks,

Ray

Re: [PATCH] clk: iproc: Fix PLL output frequency calculation

[Stephen Boyd]

On 10/19/2015 02:55 PM, Ray Jui wrote:
>
>
> On 15-10-19 02:49 PM, Stephen Boyd wrote:
>> On 10/16, Ray Jui wrote:
>>> From: Simran Rai <ssimran@broadcom.com>
>>>
>>> This patch affects the clocks that use fractional ndivider in their
>>> PLL output frequency calculation. Instead of 2^20 divide factor, the
>>> clock's ndiv integer shift was used. Fixed the bug by replacing ndiv
>>> integer shift with 2^20 factor.
>>>
>>> Signed-off-by: Simran Rai <ssimran@broadcom.com>
>>> Signed-off-by: Ray Jui <rjui@broadcom.com>
>>> Reviewed-by: Scott Branden <sbranden@broadcom.com>
>>> ---
>>
>> Did you want any Fixes: tag so this goes back to stable?
>>
>
> Yes, we want this fix to go back to the stable kernel.
>
> I will re-submit the patch with the Fixes: tag and at the same time cc
> stable@vger.kernel.org .

Please just tell me the commit that this fixes, I can add the tag and
apply to the right tree then.

-- 
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project

Re: [PATCH] clk: iproc: Fix PLL output frequency calculation

[Michael Turquette]

Quoting Stephen Boyd (2015-10-19 15:16:05)
> On 10/19/2015 02:55 PM, Ray Jui wrote:
> >
> >
> > On 15-10-19 02:49 PM, Stephen Boyd wrote:
> >> On 10/16, Ray Jui wrote:
> >>> From: Simran Rai <ssimran@broadcom.com>
> >>>
> >>> This patch affects the clocks that use fractional ndivider in their
> >>> PLL output frequency calculation. Instead of 2^20 divide factor, the
> >>> clock's ndiv integer shift was used. Fixed the bug by replacing ndiv
> >>> integer shift with 2^20 factor.
> >>>
> >>> Signed-off-by: Simran Rai <ssimran@broadcom.com>
> >>> Signed-off-by: Ray Jui <rjui@broadcom.com>
> >>> Reviewed-by: Scott Branden <sbranden@broadcom.com>
> >>> ---
> >>
> >> Did you want any Fixes: tag so this goes back to stable?
> >>
> >
> > Yes, we want this fix to go back to the stable kernel.
> >
> > I will re-submit the patch with the Fixes: tag and at the same time cc
> > stable@vger.kernel.org .
> 
> Please just tell me the commit that this fixes, I can add the tag and
> apply to the right tree then.

I applied v2 of this patch and amended the commit log with:

Cc: <stable@vger.kernel.org> # v4.1+

Regards,
Mike

> 
> -- 
> Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
> a Linux Foundation Collaborative Project
>