linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Chanwoo Choi <cw00.choi@samsung.com>
To: Tomasz Figa <t.figa@samsung.com>
Cc: myungjoo.ham@samsung.com, kyungmin.park@samsung.com,
	rafael.j.wysocki@intel.com, nm@ti.com, b.zolnierkie@samsaung.com,
	pawel.moll@arm.com, mark.rutland@arm.com, swarren@wwwdotorg.org,
	ijc+devicetree@hellion.org.uk, linux-pm@vger.kernel.org,
	linux-kernel@vger.kernel.org, linux-samsung-soc@vger.kernel.org,
	devicetree@vger.kernel.org, linux-doc@vger.kernel.org
Subject: Re: [PATCHv2 3/8] devfreq: exynos4: Add ppmu's clock control and code clean about regulator control
Date: Mon, 17 Mar 2014 11:51:22 +0900	[thread overview]
Message-ID: <5326632A.8030801@samsung.com> (raw)
In-Reply-To: <53233F78.8020500@samsung.com>

Hi Tomasz,

On 03/15/2014 02:42 AM, Tomasz Figa wrote:
> Hi Chanwoo,
> 
> On 13.03.2014 09:17, Chanwoo Choi wrote:
>> There are not the clock controller of ppmudmc0/1. This patch control the clock
>> of ppmudmc0/1 which is used for monitoring memory bus utilization.
>>
>> Also, this patch code clean about regulator control and free resource
>> when calling exit/remove function.
>>
>> For example,
>> busfreq@106A0000 {
>>     compatible = "samsung,exynos4x12-busfreq";
>>
>>     /* Clock for PPMUDMC0/1 */
>>     clocks = <&clock CLK_PPMUDMC0>, <&clock CLK_PPMUDMC1>;
>>     clock-names = "ppmudmc0", "ppmudmc1";
>>
>>     /* Regulator for MIF/INT block */
>>     vdd_mif-supply = <&buck1_reg>;
>>     vdd_int-supply = <&buck3_reg>;
>> };
>>
>> Signed-off-by: Chanwoo Choi <cw00.choi@samsung.com>
>> ---
>>   drivers/devfreq/exynos/exynos4_bus.c | 114 ++++++++++++++++++++++++++++++-----
>>   1 file changed, 100 insertions(+), 14 deletions(-)
>>
>> diff --git a/drivers/devfreq/exynos/exynos4_bus.c b/drivers/devfreq/exynos/exynos4_bus.c
>> index 1a0effa..a2a3a47 100644
>> --- a/drivers/devfreq/exynos/exynos4_bus.c
>> +++ b/drivers/devfreq/exynos/exynos4_bus.c
>> @@ -62,6 +62,11 @@ enum exynos_ppmu_idx {
>>       PPMU_END,
>>   };
>>
>> +static const char *exynos_ppmu_clk_name[] = {
>> +    [PPMU_DMC0]    = "ppmudmc0",
>> +    [PPMU_DMC1]    = "ppmudmc1",
>> +};
>> +
>>   #define EX4210_LV_MAX    LV_2
>>   #define EX4x12_LV_MAX    LV_4
>>   #define EX4210_LV_NUM    (LV_2 + 1)
>> @@ -86,6 +91,7 @@ struct busfreq_data {
>>       struct regulator *vdd_mif; /* Exynos4412/4212 only */
>>       struct busfreq_opp_info curr_oppinfo;
>>       struct exynos_ppmu ppmu[PPMU_END];
>> +    struct clk *clk_ppmu[PPMU_END];
>>
>>       struct notifier_block pm_notifier;
>>       struct mutex lock;
>> @@ -722,8 +728,26 @@ static int exynos4_bus_get_dev_status(struct device *dev,
>>   static void exynos4_bus_exit(struct device *dev)
>>   {
>>       struct busfreq_data *data = dev_get_drvdata(dev);
>> +    int i;
>> +
>> +    /*
>> +     * Un-map memory map and disable regulator/clocks
>> +     * to prevent power leakage.
>> +     */
>> +    regulator_disable(data->vdd_int);
>> +    if (data->type == TYPE_BUSF_EXYNOS4x12)
>> +        regulator_disable(data->vdd_mif);
>> +
>> +    for (i = 0; i < PPMU_END; i++) {
>> +        if (data->clk_ppmu[i])
> 
> This check is invalid. Clock pointers must be checked for validity using the IS_ERR() macro, because NULL is a valid clock pointer value indicating a dummy clock.

OK, I'll check it by using the IS_ERR() macro as following:

	if (IS_ERR(data->clk_ppmu[i]) {


> 
>> +            clk_disable_unprepare(data->clk_ppmu[i]);
>> +    }
>>
>> -    devfreq_unregister_opp_notifier(dev, data->devfreq);
>> +    for (i = 0; i < PPMU_END; i++) {
>> +        if (data->ppmu[i].hw_base)
> 
> Can this even happen? Is there a PPMU without registers?
> 
>> +            iounmap(data->ppmu[i].hw_base);
>> +
>> +    }
>>   }
>>
>>   static struct devfreq_dev_profile exynos4_devfreq_profile = {
>> @@ -987,6 +1011,7 @@ static int exynos4_busfreq_parse_dt(struct busfreq_data *data)
>>   {
>>       struct device *dev = data->dev;
>>       struct device_node *np = dev->of_node;
>> +    const char **clk_name = exynos_ppmu_clk_name;
>>       int i, ret;
>>
>>       if (!np) {
>> @@ -1005,8 +1030,70 @@ static int exynos4_busfreq_parse_dt(struct busfreq_data *data)
>>           }
>>       }
>>
>> +    /*
>> +     * Get PPMU's clocks to control them. But, if PPMU's clocks
>> +     * is default 'pass' state, this driver don't need control
>> +     * PPMU's clock.
>> +     */
>> +    for (i = 0; i < PPMU_END; i++) {
>> +        data->clk_ppmu[i] = devm_clk_get(dev, clk_name[i]);
>> +        if (IS_ERR_OR_NULL(data->clk_ppmu[i])) {
> 
> Again, this check is invalid. Only IS_ERR() is the correct way to check whether returned clock pointer is valid.

ditto.
	if (IS_ERR(data->clk_ppmu[i]) {

> 
>> +            dev_warn(dev, "Cannot get %s clock\n", clk_name[i]);
>> +            data->clk_ppmu[i] = NULL;
> 
> This assignment is wrong. To allow further checking whether the clock was found the value returned from devm_clk_get() must be retained and then IS_ERR() used in further code.
> 
> However, I believe it should be an error if a clock is not provided. The driver must make sure that PPMU clocks are ungated before trying to access them, otherwise the system might hang.

OK, I'll use IS_ERR() macro when checking / handling clock instance of 'data->clk_ppmu[i]'.
And If this driver can't get the clock of ppmu, handel error exception.

> 
>> +        }
>> +
>> +        ret = clk_prepare_enable(data->clk_ppmu[i]);
> 
> The code above allows the clock to be skipped, but this line doesn't check whether it is valid. Still, I think the clock should be always required.

OK, I'll require clock of ppmu without exception.

> 
>> +        if (ret < 0) {
>> +            dev_warn(dev, "Cannot enable %s clock\n", clk_name[i]);
>> +            data->clk_ppmu[i] = NULL;
>> +            goto err_clocks;
>> +        }
>> +    }
>> +
>> +    /* Get regulator to control voltage of int block */
>> +    data->vdd_int = devm_regulator_get(dev, "vdd_int");
>> +    if (IS_ERR(data->vdd_int)) {
>> +        dev_err(dev, "Failed to get the regulator of vdd_int\n");
>> +        ret = PTR_ERR(data->vdd_int);
>> +        goto err_clocks;
>> +    }
>> +    ret = regulator_enable(data->vdd_int);
>> +    if (ret < 0) {
>> +        dev_err(dev, "Failed to enable regulator of vdd_int\n");
>> +        goto err_clocks;
>> +    }
>> +
>> +    switch (data->type) {
>> +    case TYPE_BUSF_EXYNOS4210:
>> +        break;
>> +    case TYPE_BUSF_EXYNOS4x12:
>> +        /* Get regulator to control voltage of mif blk if Exynos4x12 */
>> +        data->vdd_mif = devm_regulator_get(dev, "vdd_mif");
>> +        if (IS_ERR(data->vdd_mif)) {
>> +            dev_err(dev, "Failed to get the regulator vdd_mif\n");
>> +            ret = PTR_ERR(data->vdd_mif);
>> +            goto err_regulator;
>> +        }
>> +        ret = regulator_enable(data->vdd_mif);
>> +        if (ret < 0) {
>> +            dev_err(dev, "Failed to enable regulator of vdd_mif\n");
>> +            goto err_regulator;
>> +        }
>> +        break;
>> +    default:
>> +        dev_err(dev, "Unknown device type : %d\n", data->type);
>> +        return -EINVAL;
>> +    };
>> +
>>       return 0;
>>
>> +err_regulator:
>> +    regulator_disable(data->vdd_int);
>> +err_clocks:
>> +    for (i = 0; i < PPMU_END; i++) {
>> +        if (data->clk_ppmu[i])
> 
> Invalid check.

Modify it as following:

	if (!IS_ERR(data->clk_ppmu[i]) {


Best Regards,
Chanwoo Choi


  reply	other threads:[~2014-03-17  2:51 UTC|newest]

Thread overview: 41+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-03-13  8:17 [PATCHv2 0/8] devfreq: exynos4: Support dt and use common ppmu driver Chanwoo Choi
2014-03-13  8:17 ` [PATCHv2 1/8] devfreq: exynos4: Support devicetree to get device id of Exynos4 SoC Chanwoo Choi
2014-03-13  8:17 ` [PATCHv2 2/8] devfreq: exynos4: Use common ppmu driver and get ppmu address from dt data Chanwoo Choi
2014-03-13  8:17 ` [PATCHv2 3/8] devfreq: exynos4: Add ppmu's clock control and code clean about regulator control Chanwoo Choi
2014-03-14 17:42   ` Tomasz Figa
2014-03-17  2:51     ` Chanwoo Choi [this message]
2014-03-17  5:35       ` Chanwoo Choi
2014-03-17  5:59       ` Chanwoo Choi
2014-03-18 11:13         ` Tomasz Figa
2014-03-19  2:44           ` Chanwoo Choi
2014-03-13  8:17 ` [PATCHv2 4/8] devfreq: exynos4: Fix bug of resource leak and code clean on probe() Chanwoo Choi
2014-03-14 17:49   ` Tomasz Figa
2014-03-17  5:05     ` Chanwoo Choi
2014-03-18 12:18       ` Tomasz Figa
2014-03-19  2:46         ` Chanwoo Choi
2014-03-13  8:17 ` [PATCHv2 5/8] devfreq: exynos4: Use SET_SYSTEM_SLEEP_PM_OPS macro Chanwoo Choi
2014-03-13  8:17 ` [PATCHv2 6/8] devfreq: exynos4: Fix power-leakage of clock on suspend state Chanwoo Choi
2014-03-14 17:52   ` Tomasz Figa
2014-03-17  2:58     ` Chanwoo Choi
2014-03-13  8:17 ` [PATCHv2 7/8] devfreq: exynos4: Add CONFIG_PM_OPP dependency to fix probe fail Chanwoo Choi
2014-03-13  8:17 ` [PATCHv2 8/8] devfreq: exynos4: Add busfreq driver for exynos4210/exynos4x12 Chanwoo Choi
2014-03-13 16:50   ` Bartlomiej Zolnierkiewicz
2014-03-13 17:53   ` Mark Rutland
2014-03-14  7:14     ` Chanwoo Choi
2014-03-14 10:35       ` Mark Rutland
2014-03-14 10:56         ` Chanwoo Choi
2014-03-14 17:35           ` Tomasz Figa
2014-03-15 11:36             ` Kyungmin Park
2014-03-15 12:41               ` Tomasz Figa
2014-03-17  5:19             ` Chanwoo Choi
2014-03-18 15:46               ` Tomasz Figa
2014-03-19  9:47                 ` Chanwoo Choi
2014-03-19 10:23                   ` Tomasz Figa
2014-03-13 16:43 ` [PATCHv2 0/8] devfreq: exynos4: Support dt and use common ppmu driver Bartlomiej Zolnierkiewicz
2014-03-14  3:14   ` Chanwoo Choi
2014-03-14 10:47     ` Bartlomiej Zolnierkiewicz
2014-03-17  1:56       ` Chanwoo Choi
2014-03-14 17:58 ` Tomasz Figa
2014-03-17  1:58   ` Chanwoo Choi
2014-03-18 15:47     ` Tomasz Figa
2014-07-09 13:06     ` Tomeu Vizoso

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=5326632A.8030801@samsung.com \
    --to=cw00.choi@samsung.com \
    --cc=b.zolnierkie@samsaung.com \
    --cc=devicetree@vger.kernel.org \
    --cc=ijc+devicetree@hellion.org.uk \
    --cc=kyungmin.park@samsung.com \
    --cc=linux-doc@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-pm@vger.kernel.org \
    --cc=linux-samsung-soc@vger.kernel.org \
    --cc=mark.rutland@arm.com \
    --cc=myungjoo.ham@samsung.com \
    --cc=nm@ti.com \
    --cc=pawel.moll@arm.com \
    --cc=rafael.j.wysocki@intel.com \
    --cc=swarren@wwwdotorg.org \
    --cc=t.figa@samsung.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).