Linux Power Management development
 help / color / mirror / Atom feed
From: Jiada Wang <jiada_wang@mentor.com>
To: Daniel Lezcano <daniel.lezcano@linaro.org>,
	rui.zhang@intel.com, edubezval@gmail.com
Cc: linux-pm@vger.kernel.org, linux-kernel@vger.kernel.org,
	erosca@de.adit-jv.com, "Frkuska,
	Joshua" <Joshua_Frkuska@mentor.com>
Subject: Re: [PATCH v1 1/1] thermal: rcar_gen3_thermal: request IRQ after device initialization
Date: Thu, 18 Apr 2019 20:36:54 +0900	[thread overview]
Message-ID: <804954a6-9f5b-28ca-ead1-891be93735ee@mentor.com> (raw)
In-Reply-To: <5e685e1d-dca7-7292-1342-dbe609c517d2@linaro.org>

Hi Daniel

Thanks for your comments

On 2019/04/17 17:05, Daniel Lezcano wrote:
> On 17/04/2019 05:01, Jiada Wang wrote:
>> Hi Daniel
>>
>> On 2019/04/17 4:22, Daniel Lezcano wrote:
>>> On 11/04/2019 12:03, Jiada Wang wrote:
>>>> Currently IRQ is remain enabled after .remove, later if device is
>>>> probed,
>>>> IRQ is requested before .thermal_init, this may cause IRQ function be
>>>> triggered but not able to clear IRQ status, thus cause system to hang.
>>>>
>>>> this patch by moving request of IRQ after device initialization to
>>>> avoid this issue.
>>>
>>> Why not disable the interrupt and clear the irq status in the .remove
>>> callback, so the place is clean when probing again?
>>>
>>>
>>>           struct rcar_gen3_thermal_priv *priv = dev_get_drvdata(dev);
>>>
>>>           rcar_thermal_irq_set(priv, false);
>>>
>>> should do the trick no ?
>>>
>> yes, this issue also can be addressed by disable the interrupt in .remove.
>>
>> But there is race condition between .remove and irq thread function,
>> (which enables IRQ)
>> so driver need to ensure threaded irq has been disabled before call
>> rcar_thermal_irq_set(priv, false) in .remove.
>> this adds additional complexity.
>>
>> I am fine with both solutions,
>> what is your opinion?
> 
> My opinion is it is the tree hiding the forest.
> 
> After a quick look at the irq setup and handling, it appears the
> implementation is cumbersome. This part should be fixed before the rest:
> 
>   - check IRQF_ONESHOT flag
>   - remove the lock in the interrupt handlers
>   - remove rcar_gen3_thermal_irq() which is pointless
>   - check the IRQF_SHARED flag is correct (I doubt)
> 

yes, I think rather than IRQF_SHARED, IRQF_ONESHOT flag need to be used.
these suggestions make sense, I will add these changes in v2 patch-set

> As the function devm_request_threaded_irq() is called 3 times, you
> should add the priv->tscs[i]->zone in the private data and the irq
> thread handler should look like:
> 
> static irqreturn_t rcar_gen3_thermal_irq_thread(int irq, void *data)
> {
> 	struct thermal_zone_device *tz = data;
> 
> 	thermal_zone_device_update(tz, THERMAL_EVENT_UNSPECIFIED);
> 
> 	[ ... ]
> 
>          return IRQ_HANDLED;
> }
> 
hmmm, IRQ is requested 2 times to monitor low and high temperature of 
all tzs.

Thanks,
Jiada

> When the implementation is fixed, then we can take care of the .remove
> 
>>>> Signed-off-by: Jiada Wang <jiada_wang@mentor.com>
>>>> ---
>>>>    drivers/thermal/rcar_gen3_thermal.c | 48 ++++++++++++++++-------------
>>>>    1 file changed, 26 insertions(+), 22 deletions(-)
>>>>
>>>> diff --git a/drivers/thermal/rcar_gen3_thermal.c
>>>> b/drivers/thermal/rcar_gen3_thermal.c
>>>> index 88fa41cf16e8..4d095d7f9763 100644
>>>> --- a/drivers/thermal/rcar_gen3_thermal.c
>>>> +++ b/drivers/thermal/rcar_gen3_thermal.c
>>>> @@ -375,28 +375,6 @@ static int rcar_gen3_thermal_probe(struct
>>>> platform_device *pdev)
>>>>          platform_set_drvdata(pdev, priv);
>>>>    -    /*
>>>> -     * Request 2 (of the 3 possible) IRQs, the driver only needs to
>>>> -     * to trigger on the low and high trip points of the current
>>>> -     * temp window at this point.
>>>> -     */
>>>> -    for (i = 0; i < 2; i++) {
>>>> -        irq = platform_get_irq(pdev, i);
>>>> -        if (irq < 0)
>>>> -            return irq;
>>>> -
>>>> -        irqname = devm_kasprintf(dev, GFP_KERNEL, "%s:ch%d",
>>>> -                     dev_name(dev), i);
>>>> -        if (!irqname)
>>>> -            return -ENOMEM;
>>>> -
>>>> -        ret = devm_request_threaded_irq(dev, irq,
>>>> rcar_gen3_thermal_irq,
>>>> -                        rcar_gen3_thermal_irq_thread,
>>>> -                        IRQF_SHARED, irqname, priv);
>>>> -        if (ret)
>>>> -            return ret;
>>>> -    }
>>>> -
>>>>        pm_runtime_enable(dev);
>>>>        pm_runtime_get_sync(dev);
>>>>    @@ -458,6 +436,32 @@ static int rcar_gen3_thermal_probe(struct
>>>> platform_device *pdev)
>>>>            goto error_unregister;
>>>>        }
>>>>    +    /*
>>>> +     * Request 2 (of the 3 possible) IRQs, the driver only needs to
>>>> +     * to trigger on the low and high trip points of the current
>>>> +     * temp window at this point.
>>>> +     */
>>>> +    for (i = 0; i < 2; i++) {
>>>> +        irq = platform_get_irq(pdev, i);
>>>> +        if (irq < 0) {
>>>> +            ret = irq;
>>>> +            goto error_unregister;
>>>> +        }
>>>> +
>>>> +        irqname = devm_kasprintf(dev, GFP_KERNEL, "%s:ch%d",
>>>> +                     dev_name(dev), i);
>>>> +        if (!irqname) {
>>>> +            ret = -ENOMEM;
>>>> +            goto error_unregister;
>>>> +        }
>>>> +
>>>> +        ret = devm_request_threaded_irq(dev, irq,
>>>> rcar_gen3_thermal_irq,
>>>> +                        rcar_gen3_thermal_irq_thread,
>>>> +                        IRQF_SHARED, irqname, priv);
>>>> +        if (ret)
>>>> +            goto error_unregister;
>>>> +    }
>>>> +
>>>>        rcar_thermal_irq_set(priv, true);
>>>>          return 0;
>>>>
>>>
>>>
> 
> 

WARNING: multiple messages have this Message-ID (diff)
From: Jiada Wang <jiada_wang@mentor.com>
To: Daniel Lezcano <daniel.lezcano@linaro.org>, <rui.zhang@intel.com>,
	<edubezval@gmail.com>
Cc: <linux-pm@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
	<erosca@de.adit-jv.com>,
	"Frkuska, Joshua" <Joshua_Frkuska@mentor.com>
Subject: Re: [PATCH v1 1/1] thermal: rcar_gen3_thermal: request IRQ after device initialization
Date: Thu, 18 Apr 2019 20:36:54 +0900	[thread overview]
Message-ID: <804954a6-9f5b-28ca-ead1-891be93735ee@mentor.com> (raw)
Message-ID: <20190418113654.MUOk1PXdBewjovlGjsI0M5gy7pxk8BpipCDJyiZo9rI@z> (raw)
In-Reply-To: <5e685e1d-dca7-7292-1342-dbe609c517d2@linaro.org>

Hi Daniel

Thanks for your comments

On 2019/04/17 17:05, Daniel Lezcano wrote:
> On 17/04/2019 05:01, Jiada Wang wrote:
>> Hi Daniel
>>
>> On 2019/04/17 4:22, Daniel Lezcano wrote:
>>> On 11/04/2019 12:03, Jiada Wang wrote:
>>>> Currently IRQ is remain enabled after .remove, later if device is
>>>> probed,
>>>> IRQ is requested before .thermal_init, this may cause IRQ function be
>>>> triggered but not able to clear IRQ status, thus cause system to hang.
>>>>
>>>> this patch by moving request of IRQ after device initialization to
>>>> avoid this issue.
>>>
>>> Why not disable the interrupt and clear the irq status in the .remove
>>> callback, so the place is clean when probing again?
>>>
>>>
>>>           struct rcar_gen3_thermal_priv *priv = dev_get_drvdata(dev);
>>>
>>>           rcar_thermal_irq_set(priv, false);
>>>
>>> should do the trick no ?
>>>
>> yes, this issue also can be addressed by disable the interrupt in .remove.
>>
>> But there is race condition between .remove and irq thread function,
>> (which enables IRQ)
>> so driver need to ensure threaded irq has been disabled before call
>> rcar_thermal_irq_set(priv, false) in .remove.
>> this adds additional complexity.
>>
>> I am fine with both solutions,
>> what is your opinion?
> 
> My opinion is it is the tree hiding the forest.
> 
> After a quick look at the irq setup and handling, it appears the
> implementation is cumbersome. This part should be fixed before the rest:
> 
>   - check IRQF_ONESHOT flag
>   - remove the lock in the interrupt handlers
>   - remove rcar_gen3_thermal_irq() which is pointless
>   - check the IRQF_SHARED flag is correct (I doubt)
> 

yes, I think rather than IRQF_SHARED, IRQF_ONESHOT flag need to be used.
these suggestions make sense, I will add these changes in v2 patch-set

> As the function devm_request_threaded_irq() is called 3 times, you
> should add the priv->tscs[i]->zone in the private data and the irq
> thread handler should look like:
> 
> static irqreturn_t rcar_gen3_thermal_irq_thread(int irq, void *data)
> {
> 	struct thermal_zone_device *tz = data;
> 
> 	thermal_zone_device_update(tz, THERMAL_EVENT_UNSPECIFIED);
> 
> 	[ ... ]
> 
>          return IRQ_HANDLED;
> }
> 
hmmm, IRQ is requested 2 times to monitor low and high temperature of 
all tzs.

Thanks,
Jiada

> When the implementation is fixed, then we can take care of the .remove
> 
>>>> Signed-off-by: Jiada Wang <jiada_wang@mentor.com>
>>>> ---
>>>>    drivers/thermal/rcar_gen3_thermal.c | 48 ++++++++++++++++-------------
>>>>    1 file changed, 26 insertions(+), 22 deletions(-)
>>>>
>>>> diff --git a/drivers/thermal/rcar_gen3_thermal.c
>>>> b/drivers/thermal/rcar_gen3_thermal.c
>>>> index 88fa41cf16e8..4d095d7f9763 100644
>>>> --- a/drivers/thermal/rcar_gen3_thermal.c
>>>> +++ b/drivers/thermal/rcar_gen3_thermal.c
>>>> @@ -375,28 +375,6 @@ static int rcar_gen3_thermal_probe(struct
>>>> platform_device *pdev)
>>>>          platform_set_drvdata(pdev, priv);
>>>>    -    /*
>>>> -     * Request 2 (of the 3 possible) IRQs, the driver only needs to
>>>> -     * to trigger on the low and high trip points of the current
>>>> -     * temp window at this point.
>>>> -     */
>>>> -    for (i = 0; i < 2; i++) {
>>>> -        irq = platform_get_irq(pdev, i);
>>>> -        if (irq < 0)
>>>> -            return irq;
>>>> -
>>>> -        irqname = devm_kasprintf(dev, GFP_KERNEL, "%s:ch%d",
>>>> -                     dev_name(dev), i);
>>>> -        if (!irqname)
>>>> -            return -ENOMEM;
>>>> -
>>>> -        ret = devm_request_threaded_irq(dev, irq,
>>>> rcar_gen3_thermal_irq,
>>>> -                        rcar_gen3_thermal_irq_thread,
>>>> -                        IRQF_SHARED, irqname, priv);
>>>> -        if (ret)
>>>> -            return ret;
>>>> -    }
>>>> -
>>>>        pm_runtime_enable(dev);
>>>>        pm_runtime_get_sync(dev);
>>>>    @@ -458,6 +436,32 @@ static int rcar_gen3_thermal_probe(struct
>>>> platform_device *pdev)
>>>>            goto error_unregister;
>>>>        }
>>>>    +    /*
>>>> +     * Request 2 (of the 3 possible) IRQs, the driver only needs to
>>>> +     * to trigger on the low and high trip points of the current
>>>> +     * temp window at this point.
>>>> +     */
>>>> +    for (i = 0; i < 2; i++) {
>>>> +        irq = platform_get_irq(pdev, i);
>>>> +        if (irq < 0) {
>>>> +            ret = irq;
>>>> +            goto error_unregister;
>>>> +        }
>>>> +
>>>> +        irqname = devm_kasprintf(dev, GFP_KERNEL, "%s:ch%d",
>>>> +                     dev_name(dev), i);
>>>> +        if (!irqname) {
>>>> +            ret = -ENOMEM;
>>>> +            goto error_unregister;
>>>> +        }
>>>> +
>>>> +        ret = devm_request_threaded_irq(dev, irq,
>>>> rcar_gen3_thermal_irq,
>>>> +                        rcar_gen3_thermal_irq_thread,
>>>> +                        IRQF_SHARED, irqname, priv);
>>>> +        if (ret)
>>>> +            goto error_unregister;
>>>> +    }
>>>> +
>>>>        rcar_thermal_irq_set(priv, true);
>>>>          return 0;
>>>>
>>>
>>>
> 
> 

  parent reply	other threads:[~2019-04-18 11:36 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-04-11 10:03 [PATCH v1 1/1] thermal: rcar_gen3_thermal: request IRQ after device initialization Jiada Wang
2019-04-11 10:03 ` Jiada Wang
2019-04-16 17:48 ` Eugeniu Rosca
2019-04-16 17:48   ` Eugeniu Rosca
2019-04-17  4:40   ` Jiada Wang
2019-04-17  4:40     ` Jiada Wang
2019-04-23 10:01   ` Simon Horman
2019-04-23 10:01     ` Simon Horman
2019-04-23 11:24     ` Eugeniu Rosca
2019-04-23 11:24       ` Eugeniu Rosca
2019-04-24  8:41       ` Simon Horman
2019-04-24  8:41         ` Simon Horman
2019-04-16 19:22 ` Daniel Lezcano
2019-04-16 19:22   ` Daniel Lezcano
2019-04-17  3:01   ` Jiada Wang
2019-04-17  3:01     ` Jiada Wang
2019-04-17  8:05     ` Daniel Lezcano
2019-04-17  8:05       ` Daniel Lezcano
2019-04-18 11:36       ` Jiada Wang [this message]
2019-04-18 11:36         ` Jiada Wang

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=804954a6-9f5b-28ca-ead1-891be93735ee@mentor.com \
    --to=jiada_wang@mentor.com \
    --cc=Joshua_Frkuska@mentor.com \
    --cc=daniel.lezcano@linaro.org \
    --cc=edubezval@gmail.com \
    --cc=erosca@de.adit-jv.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-pm@vger.kernel.org \
    --cc=rui.zhang@intel.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