linux-omap.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Tero Kristo <t-kristo@ti.com>
To: Nishanth Menon <nm@ti.com>,
	linux-omap@vger.kernel.org, paul@pwsan.com, tony@atomide.com,
	bcousson@baylibre.com, rnayak@ti.com, mturquette@linaro.org
Cc: linux-arm-kernel@lists.infradead.org, devicetree@vger.kernel.org
Subject: Re: [PATCHv9 06/43] CLK: ti: add init support for clock IP blocks
Date: Fri, 1 Nov 2013 11:12:53 +0200	[thread overview]
Message-ID: <52737095.10900@ti.com> (raw)
In-Reply-To: <52727A58.9010903@ti.com>

On 10/31/2013 05:42 PM, Nishanth Menon wrote:
> On 10/25/2013 10:57 AM, Tero Kristo wrote:
>> ti_dt_clk_init_provider() can now be used to initialize the contents of
>> a single clock IP block. This parses all the clocks under the IP block
>> and calls the corresponding init function for them.
>>
>> Signed-off-by: Tero Kristo <t-kristo@ti.com>
>> ---
>>   drivers/clk/ti/clk.c   |   59 ++++++++++++++++++++++++++++++++++++++++++++++++
>>   include/linux/clk/ti.h |    1 +
>>   2 files changed, 60 insertions(+)
>>
>> diff --git a/drivers/clk/ti/clk.c b/drivers/clk/ti/clk.c
>> index ad58b01..7f030d7 100644
>> --- a/drivers/clk/ti/clk.c
>> +++ b/drivers/clk/ti/clk.c
>> @@ -19,6 +19,9 @@
>>   #include <linux/clkdev.h>
>>   #include <linux/clk/ti.h>
>>   #include <linux/of.h>
>> +#include <linux/list.h>
>> +
>> +extern struct of_device_id __clk_of_table[];
>>
>>   /**
>>    * ti_dt_clocks_register - register DT duplicate clocks during boot
>> @@ -50,3 +53,59 @@ void __init ti_dt_clocks_register(struct ti_dt_clk oclks[])
>>   		}
>>   	}
>>   }
>> +
>> +typedef int (*ti_of_clk_init_cb_t)(struct device_node *, struct regmap *);
>> +
>> +struct clk_init_item {
>> +	struct regmap *regmap;
>> +	struct device_node *np;
>> +	ti_of_clk_init_cb_t init_cb;
>> +	struct list_head node;
>> +};
>> +
>> +static LIST_HEAD(retry_list);
>> +
>> +void __init ti_dt_clk_init_provider(struct device_node *parent,
>> +				    struct regmap *regmap)
>> +{
>> +	const struct of_device_id *match;
>> +	struct device_node *np;
>> +	ti_of_clk_init_cb_t clk_init_cb;
>> +	struct clk_init_item *retry;
>> +	struct clk_init_item *tmp;
>> +	int ret;
>> +
>> +	for_each_child_of_node(parent, np) {
>> +		match = of_match_node(__clk_of_table, np);
>> +		if (!match)
>> +			continue;
>> +		clk_init_cb = match->data;
>
> I must admit I am confused here.

Yea this patch is something I am not quite comfortable myself yet and 
would like improvement ideas....

> a) of_clk_init (in the generic clk.c) uses of_clk_init_cb_t as match
> data. The prototype of the generic of_clk_init_cb_t is typedef void
> (*of_clk_init_cb_t)(struct device_node *);
> b) both of_clk_init and ti_dt_clk_init_provider looks up clock nodes
> from __clk_of_table

__clk_of_table contains the function pointers for the clock init 
functions, not clock nodes.

>
> I assume we need ti_dt_clk_init_provider to be always called with
> clock list, as a result of_clk_init(NULL); will never be invoked. This
> is counter productive as you have have generic non SoC clock providers
> as well who would have been invoked with of_clk_init(NULL);

Can't call of_clk_init(NULL) anymore, as Paul wants to map the clock 
nodes under respective IP blocks. Two reasons here:

a) This requires me to pass some info from the IP block (CM1/CM2/PRM) to 
the clock init functions, basically the pointer to the memory map region 
(regmap.)
b) of_clk_init(NULL) will initialize all the clock nodes in the system, 
irrespective of the hierarchy considerations.

Only thing that can be done, is to make the API introduced in this patch 
a generic API and call it something like of_clk_init_children().

>
> I do strongly encourage using of_clk_init(NULL) and not having to
> switch the clk_init call back with SoC specific one as it forces
> un-intended dependency.

Can't do this as mentioned above.

>
>
> Further such as SoC specific callback should have been in a header.
>> +		pr_debug("%s: initializing: %s\n", __func__, np->name);
>
>
> one further comment - using pr_fmt will save us from using __func__
> for every pr_* message :(

Hmm good comment, will fix that for next rev.

>
>> +		ret = clk_init_cb(np, regmap);
>> +		if (ret == -EAGAIN) {
>> +			pr_debug("%s: adding to again list...\n", np->name);
>> +			retry = kzalloc(sizeof(*retry), GFP_KERNEL);
>> +			retry->np = np;
>> +			retry->regmap = regmap;
>> +			retry->init_cb = clk_init_cb;
>> +			list_add(&retry->node, &retry_list);
>> +		} else if (ret) {
>> +			pr_err("%s: clock init failed for %s (%d)!\n", __func__,
>> +			       np->name, ret);
>> +		}
>> +	}
>> +
>> +	list_for_each_entry_safe(retry, tmp, &retry_list, node) {
>> +		pr_debug("%s: retry-init: %s\n", __func__, retry->np->name);
>> +		ret = retry->init_cb(retry->np, retry->regmap);
>> +		if (ret == -EAGAIN) {
>> +			pr_debug("%s failed again?\n", retry->np->name);
>> +		} else {
>> +			if (ret)
>> +				pr_err("%s: clock init failed for %s (%d)!\n",
>> +				       __func__, retry->np->name, ret);
>> +			list_del(&retry->node);
>> +			kfree(retry);
>> +		}
>> +	}
>> +}
>> diff --git a/include/linux/clk/ti.h b/include/linux/clk/ti.h
>> index 9786752..7ab02fa 100644
>> --- a/include/linux/clk/ti.h
>> +++ b/include/linux/clk/ti.h
>> @@ -177,6 +177,7 @@ int omap3_dpll4_set_rate(struct clk_hw *clk, unsigned long rate,
>>   			 unsigned long parent_rate);
>>
>>   void ti_dt_clocks_register(struct ti_dt_clk *oclks);
>> +void ti_dt_clk_init_provider(struct device_node *np, struct regmap *regmap);
>>
>>   extern const struct clk_hw_omap_ops clkhwops_omap3_dpll;
>>   extern const struct clk_hw_omap_ops clkhwops_omap4_dpllmx;
>>
>
>


  reply	other threads:[~2013-11-01  9:13 UTC|newest]

Thread overview: 106+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-10-25 15:56 [PATCHv9 00/43] ARM: TI SoC clock DT conversion Tero Kristo
2013-10-25 15:56 ` [PATCHv9 01/43] clk: Add support for regmap register read/write Tero Kristo
2013-10-31 14:03   ` Nishanth Menon
2013-10-31 14:40     ` Tero Kristo
2013-10-31 15:46       ` Nishanth Menon
2013-11-01  8:57         ` Tero Kristo
2013-11-05 21:43       ` Gerhard Sittig
2013-11-06 10:54         ` Tero Kristo
2013-11-02 13:26   ` Tomasz Figa
2013-10-25 15:56 ` [PATCHv9 02/43] clk: divider: add init call which supports regmap Tero Kristo
2013-10-25 15:56 ` [PATCHv9 03/43] clk: mux: " Tero Kristo
2013-10-25 15:56 ` [PATCHv9 04/43] CLK: TI: Add DPLL clock support Tero Kristo
2013-10-31 14:19   ` Nishanth Menon
2013-10-31 14:56     ` Tero Kristo
2013-10-31 15:25       ` Nishanth Menon
2013-10-25 15:56 ` [PATCHv9 05/43] CLK: TI: add DT alias clock registration mechanism Tero Kristo
2013-10-29 17:50   ` Matt Sealey
2013-10-30  8:29     ` Tero Kristo
2013-10-30 17:38       ` Matt Sealey
2013-10-31  9:18         ` Tero Kristo
2013-11-02 13:49     ` Tomasz Figa
2013-11-04 22:45       ` Matt Sealey
2013-10-25 15:57 ` [PATCHv9 06/43] CLK: ti: add init support for clock IP blocks Tero Kristo
2013-10-31 15:42   ` Nishanth Menon
2013-11-01  9:12     ` Tero Kristo [this message]
2013-11-01 19:13       ` Nishanth Menon
2013-11-04  7:23         ` Tero Kristo
2013-10-25 15:57 ` [PATCHv9 07/43] CLK: TI: add autoidle support Tero Kristo
2013-10-31 16:05   ` Nishanth Menon
2013-11-01  9:18     ` Tero Kristo
2013-11-01 19:16       ` Nishanth Menon
2013-11-04 10:00         ` Tero Kristo
2013-11-04 14:59           ` Nishanth Menon
2013-11-05  8:10             ` Tero Kristo
2013-10-25 15:57 ` [PATCHv9 08/43] clk: ti: add composite clock support Tero Kristo
2013-10-31 16:27   ` Nishanth Menon
2013-10-31 16:32     ` Nishanth Menon
2013-11-01  9:40       ` Tero Kristo
2013-11-01  9:35     ` Tero Kristo
2013-11-01 19:24       ` Nishanth Menon
2013-10-25 15:57 ` [PATCHv9 09/43] CLK: ti: add support for ti divider-clock Tero Kristo
2013-10-31 18:02   ` Nishanth Menon
2013-11-01  9:48     ` Tero Kristo
2013-11-01  9:54       ` Tero Kristo
2013-11-01 19:35         ` Nishanth Menon
2013-11-04 10:54           ` Tero Kristo
2013-10-25 15:57 ` [PATCHv9 10/43] clk: ti: add support for TI fixed factor clock Tero Kristo
2013-10-31 18:12   ` Nishanth Menon
2013-11-01  9:52     ` Tero Kristo
2013-10-25 15:57 ` [PATCHv9 11/43] CLK: TI: add support for gate clock Tero Kristo
2013-11-01 20:11   ` Nishanth Menon
2013-11-04 12:23     ` Tero Kristo
2013-10-25 15:57 ` [PATCHv9 12/43] CLK: TI: add support for clockdomain binding Tero Kristo
2013-11-01 20:22   ` Nishanth Menon
     [not found]     ` <52740D79.3090107-l0cyMroinI0@public.gmane.org>
2013-11-04 14:30       ` Tero Kristo
2013-10-25 15:57 ` [PATCHv9 13/43] clk: ti: add support for basic mux clock Tero Kristo
2013-11-01 21:01   ` Nishanth Menon
2013-11-05  8:09     ` Tero Kristo
2013-10-25 15:57 ` [PATCHv9 14/43] CLK: TI: add omap4 clock init file Tero Kristo
2013-10-25 15:57 ` [PATCHv9 15/43] CLK: TI: add omap5 " Tero Kristo
2013-10-25 15:57 ` [PATCHv9 16/43] CLK: TI: omap5: Initialize USB_DPLL at boot Tero Kristo
2013-10-25 15:57 ` [PATCHv9 17/43] CLK: TI: DRA7: Add APLL support Tero Kristo
2013-10-25 15:57 ` [PATCHv9 18/43] CLK: TI: add dra7 clock init file Tero Kristo
2013-10-25 15:57 ` [PATCHv9 19/43] CLK: TI: add am33xx " Tero Kristo
2013-10-25 15:57 ` [PATCHv9 21/43] CLK: TI: add omap3 " Tero Kristo
2013-10-25 15:57 ` [PATCHv9 22/43] CLK: TI: add am43xx " Tero Kristo
2013-10-25 15:57 ` [PATCHv9 23/43] ARM: dts: omap4 clock data Tero Kristo
2013-10-25 15:57 ` [PATCHv9 24/43] ARM: dts: omap5 " Tero Kristo
     [not found] ` <1382716658-6964-1-git-send-email-t-kristo-l0cyMroinI0@public.gmane.org>
2013-10-25 15:57   ` [PATCHv9 20/43] CLK: TI: add interface clock support for OMAP3 Tero Kristo
2013-10-25 15:57   ` [PATCHv9 25/43] ARM: dts: dra7 clock data Tero Kristo
2013-10-25 15:57   ` [PATCHv9 36/43] ARM: OMAP: hwmod: fix an incorrect clk type cast with _get_clkdm Tero Kristo
2013-10-25 15:57   ` [PATCHv9 40/43] ARM: OMAP4: remove old clock data and link in new clock init code Tero Kristo
2013-10-25 15:57 ` [PATCHv9 26/43] ARM: dts: clk: Add apll related clocks Tero Kristo
2013-10-25 15:57 ` [PATCHv9 27/43] ARM: dts: DRA7: Change apll_pcie_m2_ck to fixed factor clock Tero Kristo
2013-10-25 15:57 ` [PATCHv9 28/43] ARM: dts: DRA7: Add PCIe related clock nodes Tero Kristo
2013-10-25 15:57 ` [PATCHv9 29/43] ARM: dts: DRA7: link in clock DT data Tero Kristo
2013-10-25 15:57 ` [PATCHv9 30/43] ARM: dts: am33xx clock data Tero Kristo
2013-10-26  0:36   ` Jan Lübbe
2013-10-26 12:46     ` Tero Kristo
2013-10-28  9:59       ` Jan Lübbe
2013-10-28 10:12         ` Tero Kristo
2013-10-25 15:57 ` [PATCHv9 31/43] ARM: dts: omap3 " Tero Kristo
2013-10-25 15:57 ` [PATCHv9 32/43] ARM: dts: AM35xx: use DT " Tero Kristo
2013-11-01 21:18   ` Nishanth Menon
2013-11-05  8:12     ` Tero Kristo
2013-10-25 15:57 ` [PATCHv9 33/43] ARM: dts: am43xx " Tero Kristo
2013-11-01 21:16   ` Nishanth Menon
2013-11-04 14:15     ` Tero Kristo
2013-10-25 15:57 ` [PATCHv9 34/43] ARM: OMAP2+: clock: add support for regmap Tero Kristo
2013-10-25 15:57 ` [PATCHv9 35/43] ARM: OMAP2+: clock: use driver API instead of direct memory read/write Tero Kristo
2013-10-25 15:57 ` [PATCHv9 37/43] ARM: OMAP3: hwmod: initialize clkdm from clkdm_name Tero Kristo
2013-10-25 15:57 ` [PATCHv9 38/43] ARM: OMAP2+: PRM: add support for initializing PRCM clock modules from DT Tero Kristo
2013-11-01 21:07   ` Nishanth Menon
2013-11-05  8:22     ` Tero Kristo
2013-10-25 15:57 ` [PATCHv9 39/43] ARM: OMAP2+: io: use new clock init API Tero Kristo
2013-10-25 15:57 ` [PATCHv9 41/43] ARM: OMAP: DRA7: Enable clock init Tero Kristo
2013-10-25 15:57 ` [PATCHv9 42/43] ARM: AM33xx: remove old clock data and link in new clock init code Tero Kristo
2013-10-25 15:57 ` [PATCHv9 43/43] ARM: OMAP3: use DT clock init if DT data is available Tero Kristo
2013-10-29 16:19 ` [PATCHv9 00/43] ARM: TI SoC clock DT conversion Nishanth Menon
2013-10-30  8:23   ` Tero Kristo
2013-10-30 15:00     ` Nishanth Menon
2013-10-30 20:10       ` Nishanth Menon
2013-10-31  9:10         ` Tero Kristo
2013-10-31 13:55           ` Nishanth Menon
2013-11-01 21:25             ` Nishanth Menon
2013-11-04  7:15               ` Tero Kristo

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=52737095.10900@ti.com \
    --to=t-kristo@ti.com \
    --cc=bcousson@baylibre.com \
    --cc=devicetree@vger.kernel.org \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-omap@vger.kernel.org \
    --cc=mturquette@linaro.org \
    --cc=nm@ti.com \
    --cc=paul@pwsan.com \
    --cc=rnayak@ti.com \
    --cc=tony@atomide.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).