[RESEND,v1,4/4] clk: rockchip: rk3036: fix and add node id for emac clock
diff mbox

Message ID 1562268.fb58QJ4jV7@diego
State New
Headers show

Commit Message

Heiko Stuebner Jan. 1, 2016, 10:10 p.m. UTC
Hi Xing,

Am Dienstag, 29. Dezember 2015, 10:34:09 schrieb Xing Zheng:
> On 2015?12?29? 09:59, Yakir Yang wrote:
> > On 12/28/2015 08:41 PM, Heiko Stübner wrote:
> >> Am Montag, 28. Dezember 2015, 17:03:53 schrieb Xing Zheng:
> >>> Due to referred old version TRM, there is incorrect emac clock node,
> >>> we should fix it. The SEL_21_9 is the parent of SEL_21_4.
> >>> 
> >>> In the emac driver, we need to refer HCLK_MAC, and because There are
> >>> only 3PLLs (APLL/GPLL/DPLL) on the rk3036, most clock are under the
> >>> GPLL, and it is unable to provide the accurate rate for mac_ref which
> >>> need to 50MHz probability, we should let it under the APLL and are
> >>> able to set the freq which integer multiples of 50MHz, so we add these
> >>> emac node for reference.
> >> 
> >> I don't really follow here. While I do understand that the emac needs
> >> 50MHz, I
> >> don't think using the APLL as source is helpful.
> >> 
> >> The APLL is the main clocksource for the cpu-cores, including frequency
> >> scaling, and while it currently only lists 816MHz as sole frequency,
> >> you're
> >> pretty much guaranteed to not get your correct multiple of 50MHz from
> >> there
> >> either. And limiting the cpu to just do 600MHz to get the mac working
> >> sounds
> >> pretty bad ;-) .
> >> 
> >> 
> >> In the rk3036 cru-node the gpll gets set to 594MHz. Is there a
> >> special reason
> >> why it needs to be 594MHz and cannot be a round 600MHz? Because that
> >> would
> >> also provide your 50MHz-multiple nicely.
> > 
> > Yes, this magic 594MHz would help to support the standard HDMI
> > resolutions, here are the math:
> > 
> > 1920x1080-60Hz DCLK = 148.5MHz = 594MHz / 4
> > 1280x720-60Hz DCLK = 74.25MHz = 594MHz / 8
> > 720x480-60Hz DCLK = 27MHz = 594MHz / 22
> > 
> > Thanks,
> > - Yakir
> 
> Thanks Yakir.
> 
> Hi Heiko,
>  From the above, do you have better idea for the RK3036's emac without
> ext-oscillator?

During the last days I did play a bit with the clock framework. As I don't 
have a Kylin (or any rk3036) board, I did build a test-case with pclk_cpu on 
the rk3188 (which can be affected by the armclk if not reparented to the 
gpll), which got sucessfully adapted to get back to (or near) the originally 
requested frequency.

So ideally you could roll back your mux/div split here and try the attached 
diff. In theory it should help :-) .
As can be seen by the FIXMEs, not fully finished, but I'd like to determine if 
it fixes the issue at least.


Heiko

Comments

zhengxing Jan. 2, 2016, 2:34 a.m. UTC | #1
Hi Heiko,
    Thank you for your patch, I will apply and test it later.

Thanks.

> ? 2016?1?2??06:10?Heiko Stübner <heiko@sntech.de> ???
> 
> Hi Xing,
> 
> Am Dienstag, 29. Dezember 2015, 10:34:09 schrieb Xing Zheng:
>> On 2015?12?29? 09:59, Yakir Yang wrote:
>>> On 12/28/2015 08:41 PM, Heiko Stübner wrote:
>>>> Am Montag, 28. Dezember 2015, 17:03:53 schrieb Xing Zheng:
>>>>> Due to referred old version TRM, there is incorrect emac clock node,
>>>>> we should fix it. The SEL_21_9 is the parent of SEL_21_4.
>>>>> 
>>>>> In the emac driver, we need to refer HCLK_MAC, and because There are
>>>>> only 3PLLs (APLL/GPLL/DPLL) on the rk3036, most clock are under the
>>>>> GPLL, and it is unable to provide the accurate rate for mac_ref which
>>>>> need to 50MHz probability, we should let it under the APLL and are
>>>>> able to set the freq which integer multiples of 50MHz, so we add these
>>>>> emac node for reference.
>>>> 
>>>> I don't really follow here. While I do understand that the emac needs
>>>> 50MHz, I
>>>> don't think using the APLL as source is helpful.
>>>> 
>>>> The APLL is the main clocksource for the cpu-cores, including frequency
>>>> scaling, and while it currently only lists 816MHz as sole frequency,
>>>> you're
>>>> pretty much guaranteed to not get your correct multiple of 50MHz from
>>>> there
>>>> either. And limiting the cpu to just do 600MHz to get the mac working
>>>> sounds
>>>> pretty bad ;-) .
>>>> 
>>>> 
>>>> In the rk3036 cru-node the gpll gets set to 594MHz. Is there a
>>>> special reason
>>>> why it needs to be 594MHz and cannot be a round 600MHz? Because that
>>>> would
>>>> also provide your 50MHz-multiple nicely.
>>> 
>>> Yes, this magic 594MHz would help to support the standard HDMI
>>> resolutions, here are the math:
>>> 
>>> 1920x1080-60Hz DCLK = 148.5MHz = 594MHz / 4
>>> 1280x720-60Hz DCLK = 74.25MHz = 594MHz / 8
>>> 720x480-60Hz DCLK = 27MHz = 594MHz / 22
>>> 
>>> Thanks,
>>> - Yakir
>> 
>> Thanks Yakir.
>> 
>> Hi Heiko,
>> From the above, do you have better idea for the RK3036's emac without
>> ext-oscillator?
> 
> During the last days I did play a bit with the clock framework. As I don't 
> have a Kylin (or any rk3036) board, I did build a test-case with pclk_cpu on 
> the rk3188 (which can be affected by the armclk if not reparented to the 
> gpll), which got sucessfully adapted to get back to (or near) the originally 
> requested frequency.
> 
> So ideally you could roll back your mux/div split here and try the attached 
> diff. In theory it should help :-) .
> As can be seen by the FIXMEs, not fully finished, but I'd like to determine if 
> it fixes the issue at least.
> 
> 
> Heiko
> <clk-keep-req-rate.diff>
zhengxing Jan. 6, 2016, 1:05 p.m. UTC | #2
Hi Heiko,

On 2016?01?02? 10:34, Xing Zheng wrote:
> Hi Heiko,
>     Thank you for your patch, I will apply and test it later.
>
> Thanks.
>
>> ? 2016?1?2??06:10?Heiko Stübner <heiko@sntech.de> ???
>>
>> Hi Xing,
>>
>> Am Dienstag, 29. Dezember 2015, 10:34:09 schrieb Xing Zheng:
>>> On 2015?12?29? 09:59, Yakir Yang wrote:
>>>> On 12/28/2015 08:41 PM, Heiko Stübner wrote:
>>>>> Am Montag, 28. Dezember 2015, 17:03:53 schrieb Xing Zheng:
>>>>>> Due to referred old version TRM, there is incorrect emac clock node,
>>>>>> we should fix it. The SEL_21_9 is the parent of SEL_21_4.
>>>>>>
>>>>>> In the emac driver, we need to refer HCLK_MAC, and because There are
>>>>>> only 3PLLs (APLL/GPLL/DPLL) on the rk3036, most clock are under the
>>>>>> GPLL, and it is unable to provide the accurate rate for mac_ref which
>>>>>> need to 50MHz probability, we should let it under the APLL and are
>>>>>> able to set the freq which integer multiples of 50MHz, so we add these
>>>>>> emac node for reference.
>>>>> I don't really follow here. While I do understand that the emac needs
>>>>> 50MHz, I
>>>>> don't think using the APLL as source is helpful.
>>>>>
>>>>> The APLL is the main clocksource for the cpu-cores, including frequency
>>>>> scaling, and while it currently only lists 816MHz as sole frequency,
>>>>> you're
>>>>> pretty much guaranteed to not get your correct multiple of 50MHz from
>>>>> there
>>>>> either. And limiting the cpu to just do 600MHz to get the mac working
>>>>> sounds
>>>>> pretty bad ;-) .
>>>>>
>>>>>
>>>>> In the rk3036 cru-node the gpll gets set to 594MHz. Is there a
>>>>> special reason
>>>>> why it needs to be 594MHz and cannot be a round 600MHz? Because that
>>>>> would
>>>>> also provide your 50MHz-multiple nicely.
>>>> Yes, this magic 594MHz would help to support the standard HDMI
>>>> resolutions, here are the math:
>>>>
>>>> 1920x1080-60Hz DCLK = 148.5MHz = 594MHz / 4
>>>> 1280x720-60Hz DCLK = 74.25MHz = 594MHz / 8
>>>> 720x480-60Hz DCLK = 27MHz = 594MHz / 22
>>>>
>>>> Thanks,
>>>> - Yakir
>>> Thanks Yakir.
>>>
>>> Hi Heiko,
>>> From the above, do you have better idea for the RK3036's emac without
>>> ext-oscillator?
>> During the last days I did play a bit with the clock framework. As I don't 
>> have a Kylin (or any rk3036) board, I did build a test-case with pclk_cpu on 
>> the rk3188 (which can be affected by the armclk if not reparented to the 
>> gpll), which got sucessfully adapted to get back to (or near) the originally 
>> requested frequency.
>>
>> So ideally you could roll back your mux/div split here and try the attached 
>> diff. In theory it should help :-) .
>> As can be seen by the FIXMEs, not fully finished, but I'd like to determine if 
>> it fixes the issue at least.
>>
>>
>> Heiko
>> <clk-keep-req-rate.diff>

I tested your patch just now, it seems to work well. I will do more tests.

Thank you. :-)

Patch
diff mbox

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index 7bbb0fd..83a7234 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -1423,6 +1423,9 @@  static struct clk_core *clk_propagate_rate_change(struct clk_core *core,
 	return fail_clk;
 }
 
+static int clk_core_set_rate_nolock(struct clk_core *core,
+				    unsigned long req_rate);
+
 /*
  * walk down a subtree and set the new rates notifying the rate
  * change on the way
@@ -1438,6 +1441,7 @@  static void clk_change_rate(struct clk_core *core)
 
 	old_rate = core->rate;
 
+printk("%s: %s requested %lu, new %lu\n", __func__, core->name, core->req_rate, core->new_rate);
 	if (core->new_parent)
 		best_parent_rate = core->new_parent->rate;
 	else if (core->parent)
@@ -1507,6 +1511,12 @@  static void clk_change_rate(struct clk_core *core)
 	/* handle the new child who might not be in core->children yet */
 	if (core->new_child)
 		clk_change_rate(core->new_child);
+
+	/* FIXME: add flag to limit to specific clocks? */
+	if (core->req_rate && core->new_rate != old_rate) {
+		printk("\t\ttrying to adapt %s\n", core->name);
+		clk_core_set_rate_nolock(core, core->req_rate);
+	}
 }
 
 static int clk_core_set_rate_nolock(struct clk_core *core,
@@ -1520,8 +1530,10 @@  static int clk_core_set_rate_nolock(struct clk_core *core,
 		return 0;
 
 	/* bail early if nothing to do */
-	if (rate == clk_core_get_rate_nolock(core))
+	if (rate == clk_core_get_rate_nolock(core)) {
+		core->req_rate = req_rate;
 		return 0;
+	}
 
 	if ((core->flags & CLK_SET_RATE_GATE) && core->prepare_count)
 		return -EBUSY;
@@ -1612,9 +1624,14 @@  int clk_set_rate_range(struct clk *clk, unsigned long min, unsigned long max)
 	clk_prepare_lock();
 
 	if (min != clk->min_rate || max != clk->max_rate) {
+		unsigned long rate = clk->core->req_rate;
+
+		if (!rate)
+			rate = clk->core->rate;
+
 		clk->min_rate = min;
 		clk->max_rate = max;
-		ret = clk_core_set_rate_nolock(clk->core, clk->core->req_rate);
+		ret = clk_core_set_rate_nolock(clk->core, rate);
 	}
 
 	clk_prepare_unlock();
@@ -2451,7 +2468,7 @@  static int __clk_init(struct device *dev, struct clk *clk_user)
 		rate = core->parent->rate;
 	else
 		rate = 0;
-	core->rate = core->req_rate = rate;
+	core->rate = rate;
 
 	/*
 	 * walk the list of orphan clocks and reparent any that are children of
@@ -2798,6 +2815,7 @@  int __clk_get(struct clk *clk)
 
 void __clk_put(struct clk *clk)
 {
+	unsigned long rate;
 	struct module *owner;
 
 	if (!clk || WARN_ON_ONCE(IS_ERR(clk)))
@@ -2806,9 +2824,13 @@  void __clk_put(struct clk *clk)
 	clk_prepare_lock();
 
 	hlist_del(&clk->clks_node);
-	if (clk->min_rate > clk->core->req_rate ||
-	    clk->max_rate < clk->core->req_rate)
-		clk_core_set_rate_nolock(clk->core, clk->core->req_rate);
+
+	rate = clk->core->req_rate;
+	if (!rate)
+		rate = clk->core->rate;
+
+	if (clk->min_rate > rate || clk->max_rate < rate)
+		clk_core_set_rate_nolock(clk->core, rate);
 
 	owner = clk->core->owner;
 	kref_put(&clk->core->ref, __clk_release);