Patchwork mtd: atmel_nand: fix bug driver will in a dead lock if no nand detected

login
register
mail settings
Submitter Wu, Josh
Date Nov. 5, 2013, 9:59 a.m.
Message ID <1383645547-21813-1-git-send-email-josh.wu@atmel.com>
Download mbox | patch
Permalink /patch/288453/
State Accepted
Commit a749d13acd8e079ed4c77a9456d842dc94af8f17
Headers show

Comments

Wu, Josh - Nov. 5, 2013, 9:59 a.m.
In the atmel driver probe function, the code shows like following:
  atmel_nand_probe(...) {
        ...

  err_nand_ioremap:
        platform_driver_unregister(&atmel_nand_nfc_driver);
        return res;
  }

If no nand flash detected, the driver probe function will goto
err_nand_ioremap label.
Then platform_driver_unregister() will be called. It will get the
lock of atmel_nand device since it is parent of nfc_device. The
problem is the lock is already hold by atmel_nand_probe itself.
So system will be in a dead lock.

This patch just simply removed to platform_driver_unregister() call.
When atmel_nand driver is quit the platform_driver_unregister() will
be called in atmel_nand_remove().

Signed-off-by: Josh Wu <josh.wu@atmel.com>
---
 drivers/mtd/nand/atmel_nand.c |    1 -
 1 file changed, 1 deletion(-)
Brian Norris - Nov. 7, 2013, 8:39 a.m.
Hi Josh,

On Tue, Nov 05, 2013 at 05:59:07PM +0800, Josh Wu wrote:
> In the atmel driver probe function, the code shows like following:
>   atmel_nand_probe(...) {
>         ...
> 
>   err_nand_ioremap:
>         platform_driver_unregister(&atmel_nand_nfc_driver);
>         return res;
>   }
> 
> If no nand flash detected, the driver probe function will goto
> err_nand_ioremap label.
> Then platform_driver_unregister() will be called. It will get the
> lock of atmel_nand device since it is parent of nfc_device. The
> problem is the lock is already hold by atmel_nand_probe itself.
> So system will be in a dead lock.
> 
> This patch just simply removed to platform_driver_unregister() call.
> When atmel_nand driver is quit the platform_driver_unregister() will
> be called in atmel_nand_remove().

The real key, here, is that the platform-driver probe() has no business
un-registering another driver, right? Shouldn't both drivers just be
registered/unregistered in the module init/exit, and not in probe()?

Also, should this be marked for -stable?

> Signed-off-by: Josh Wu <josh.wu@atmel.com>
> ---
>  drivers/mtd/nand/atmel_nand.c |    1 -
>  1 file changed, 1 deletion(-)
> 
> diff --git a/drivers/mtd/nand/atmel_nand.c b/drivers/mtd/nand/atmel_nand.c
> index ef9c9f5..469d4e2 100644
> --- a/drivers/mtd/nand/atmel_nand.c
> +++ b/drivers/mtd/nand/atmel_nand.c
> @@ -2142,7 +2142,6 @@ err_no_card:
>  	if (host->dma_chan)
>  		dma_release_channel(host->dma_chan);
>  err_nand_ioremap:
> -	platform_driver_unregister(&atmel_nand_nfc_driver);
>  	return res;
>  }
>  

Brian
Wu, Josh - Nov. 7, 2013, 10:26 a.m.
Hi, Brian

On 11/7/2013 4:39 PM, Brian Norris wrote:
> Hi Josh,
>
> On Tue, Nov 05, 2013 at 05:59:07PM +0800, Josh Wu wrote:
>> In the atmel driver probe function, the code shows like following:
>>    atmel_nand_probe(...) {
>>          ...
>>
>>    err_nand_ioremap:
>>          platform_driver_unregister(&atmel_nand_nfc_driver);
>>          return res;
>>    }
>>
>> If no nand flash detected, the driver probe function will goto
>> err_nand_ioremap label.
>> Then platform_driver_unregister() will be called. It will get the
>> lock of atmel_nand device since it is parent of nfc_device. The
>> problem is the lock is already hold by atmel_nand_probe itself.
>> So system will be in a dead lock.
>>
>> This patch just simply removed to platform_driver_unregister() call.
>> When atmel_nand driver is quit the platform_driver_unregister() will
>> be called in atmel_nand_remove().
> The real key, here, is that the platform-driver probe() has no business
> un-registering another driver, right?

right.

> Shouldn't both drivers just be
> registered/unregistered in the module init/exit, and not in probe()?

currently the NFC driver is registered in the beginning of nand probe 
function. After NFC driver is initialized, the rest of the nand probe 
function
will check the NFC driver status.
I am not sure it is proper to registered another driver in a probe().

>
> Also, should this be marked for -stable?

Shouldn't cc to stable after the patch is merged in mtd mail list?

Best Regards,
Josh Wu

>
>> Signed-off-by: Josh Wu <josh.wu@atmel.com>
>> ---
>>   drivers/mtd/nand/atmel_nand.c |    1 -
>>   1 file changed, 1 deletion(-)
>>
>> diff --git a/drivers/mtd/nand/atmel_nand.c b/drivers/mtd/nand/atmel_nand.c
>> index ef9c9f5..469d4e2 100644
>> --- a/drivers/mtd/nand/atmel_nand.c
>> +++ b/drivers/mtd/nand/atmel_nand.c
>> @@ -2142,7 +2142,6 @@ err_no_card:
>>   	if (host->dma_chan)
>>   		dma_release_channel(host->dma_chan);
>>   err_nand_ioremap:
>> -	platform_driver_unregister(&atmel_nand_nfc_driver);
>>   	return res;
>>   }
>>   
> Brian
Nicolas Ferre - Nov. 7, 2013, 10:55 a.m.
On 07/11/2013 11:26, Josh Wu :
> Hi, Brian
>
> On 11/7/2013 4:39 PM, Brian Norris wrote:
>> Hi Josh,
>>
>> On Tue, Nov 05, 2013 at 05:59:07PM +0800, Josh Wu wrote:
>>> In the atmel driver probe function, the code shows like following:
>>>     atmel_nand_probe(...) {
>>>           ...
>>>
>>>     err_nand_ioremap:
>>>           platform_driver_unregister(&atmel_nand_nfc_driver);
>>>           return res;
>>>     }
>>>
>>> If no nand flash detected, the driver probe function will goto
>>> err_nand_ioremap label.
>>> Then platform_driver_unregister() will be called. It will get the
>>> lock of atmel_nand device since it is parent of nfc_device. The
>>> problem is the lock is already hold by atmel_nand_probe itself.
>>> So system will be in a dead lock.
>>>
>>> This patch just simply removed to platform_driver_unregister() call.
>>> When atmel_nand driver is quit the platform_driver_unregister() will
>>> be called in atmel_nand_remove().
>> The real key, here, is that the platform-driver probe() has no business
>> un-registering another driver, right?
>
> right.
>
>> Shouldn't both drivers just be
>> registered/unregistered in the module init/exit, and not in probe()?
>
> currently the NFC driver is registered in the beginning of nand probe
> function. After NFC driver is initialized, the rest of the nand probe
> function
> will check the NFC driver status.
> I am not sure it is proper to registered another driver in a probe().
>
>>
>> Also, should this be marked for -stable?
>
> Shouldn't cc to stable after the patch is merged in mtd mail list?

Whether you add a Cc: tag in your signature location or you send another 
patch to the stable mailing list *but* only when the patch is accepted 
in Linus' git tree.

You can find the information in
Documentation/stable_kernel_rules.txt

Bye,

>>> Signed-off-by: Josh Wu <josh.wu@atmel.com>
>>> ---
>>>    drivers/mtd/nand/atmel_nand.c |    1 -
>>>    1 file changed, 1 deletion(-)
>>>
>>> diff --git a/drivers/mtd/nand/atmel_nand.c b/drivers/mtd/nand/atmel_nand.c
>>> index ef9c9f5..469d4e2 100644
>>> --- a/drivers/mtd/nand/atmel_nand.c
>>> +++ b/drivers/mtd/nand/atmel_nand.c
>>> @@ -2142,7 +2142,6 @@ err_no_card:
>>>    	if (host->dma_chan)
>>>    		dma_release_channel(host->dma_chan);
>>>    err_nand_ioremap:
>>> -	platform_driver_unregister(&atmel_nand_nfc_driver);
>>>    	return res;
>>>    }
>>>
>> Brian
>
>
>
Brian Norris - Nov. 7, 2013, 6:09 p.m.
On Thu, Nov 07, 2013 at 06:26:39PM +0800, Josh Wu wrote:
> Hi, Brian
> 
> On 11/7/2013 4:39 PM, Brian Norris wrote:
> >Hi Josh,
> >
> >On Tue, Nov 05, 2013 at 05:59:07PM +0800, Josh Wu wrote:
> >>In the atmel driver probe function, the code shows like following:
> >>   atmel_nand_probe(...) {
> >>         ...
> >>
> >>   err_nand_ioremap:
> >>         platform_driver_unregister(&atmel_nand_nfc_driver);
> >>         return res;
> >>   }
> >>
> >>If no nand flash detected, the driver probe function will goto
> >>err_nand_ioremap label.
> >>Then platform_driver_unregister() will be called. It will get the
> >>lock of atmel_nand device since it is parent of nfc_device. The
> >>problem is the lock is already hold by atmel_nand_probe itself.
> >>So system will be in a dead lock.
> >>
> >>This patch just simply removed to platform_driver_unregister() call.
> >>When atmel_nand driver is quit the platform_driver_unregister() will
> >>be called in atmel_nand_remove().
> >The real key, here, is that the platform-driver probe() has no business
> >un-registering another driver, right?
> 
> right.
> 
> >Shouldn't both drivers just be
> >registered/unregistered in the module init/exit, and not in probe()?
> 
> currently the NFC driver is registered in the beginning of nand
> probe function. After NFC driver is initialized, the rest of the
> nand probe function
> will check the NFC driver status.
> I am not sure it is proper to registered another driver in a probe().

This whole 2-driver registration process seems like a totally racy mess.
Unless I'm reading this wrong, the code is entirely wrong. You are not
properly expressing the dependency between the NFC device and the
atmel_nand device, so I think you're relying purely on happenstance that
when the NAND probe registers the NFC driver, it will complete the NFC
probe before the NAND probe reaches:

  if (nand_nfc.is_initialized) {
    ...

This is a race, no?

Anyway, I think the current patch (fixing a deadlock) is still ready for
stable, while I would expect you can fix up this racy situation in a
subsequent patch. Or perhaps I'm misreading something?

Brian
Brian Norris - Nov. 7, 2013, 6:43 p.m.
On Tue, Nov 05, 2013 at 05:59:07PM +0800, Josh Wu wrote:
> In the atmel driver probe function, the code shows like following:
>   atmel_nand_probe(...) {
>         ...
> 
>   err_nand_ioremap:
>         platform_driver_unregister(&atmel_nand_nfc_driver);
>         return res;
>   }
> 
> If no nand flash detected, the driver probe function will goto
> err_nand_ioremap label.
> Then platform_driver_unregister() will be called. It will get the
> lock of atmel_nand device since it is parent of nfc_device. The
> problem is the lock is already hold by atmel_nand_probe itself.
> So system will be in a dead lock.
> 
> This patch just simply removed to platform_driver_unregister() call.
> When atmel_nand driver is quit the platform_driver_unregister() will
> be called in atmel_nand_remove().
> 
> Signed-off-by: Josh Wu <josh.wu@atmel.com>

Pushed to l2-mtd.git, with a little extra note and a
Cc: <stable@vger.kernel.org>

Thanks,
Brian
Wu, Josh - Nov. 8, 2013, 3:46 a.m.
Hi, Brain

I resend with non-html format to make it acceptable in mail list.

On 11/8/2013 2:09 AM, Brian Norris wrote:
> On Thu, Nov 07, 2013 at 06:26:39PM +0800, Josh Wu wrote:
>> Hi, Brian
>>
>> On 11/7/2013 4:39 PM, Brian Norris wrote:
>>> Hi Josh,
>>>
>>> On Tue, Nov 05, 2013 at 05:59:07PM +0800, Josh Wu wrote:
>>>> In the atmel driver probe function, the code shows like following:
>>>>    atmel_nand_probe(...) {
>>>>          ...
>>>>
>>>>    err_nand_ioremap:
>>>>          platform_driver_unregister(&atmel_nand_nfc_driver);
>>>>          return res;
>>>>    }
>>>>
>>>> If no nand flash detected, the driver probe function will goto
>>>> err_nand_ioremap label.
>>>> Then platform_driver_unregister() will be called. It will get the
>>>> lock of atmel_nand device since it is parent of nfc_device. The
>>>> problem is the lock is already hold by atmel_nand_probe itself.
>>>> So system will be in a dead lock.
>>>>
>>>> This patch just simply removed to platform_driver_unregister() call.
>>>> When atmel_nand driver is quit the platform_driver_unregister() will
>>>> be called in atmel_nand_remove().
>>> The real key, here, is that the platform-driver probe() has no business
>>> un-registering another driver, right?
>> right.
>>
>>> Shouldn't both drivers just be
>>> registered/unregistered in the module init/exit, and not in probe()?
>> currently the NFC driver is registered in the beginning of nand
>> probe function. After NFC driver is initialized, the rest of the
>> nand probe function
>> will check the NFC driver status.
>> I am not sure it is proper to registered another driver in a probe().
> This whole 2-driver registration process seems like a totally racy mess.
> Unless I'm reading this wrong, the code is entirely wrong. You are not
> properly expressing the dependency between the NFC device and the
> atmel_nand device, so I think you're relying purely on happenstance that
> when the NAND probe registers the NFC driver, it will complete the NFC
> probe before the NAND probe reaches:
>
>    if (nand_nfc.is_initialized) {
>      ...

Yes, exactly.
And the NAND probe will also load the NFC device before it reach above 
check code.
That is what I want: make *nfc probe function* is running before reach 
above check code.

>
> This is a race, no?

So I don't think there is a race, as NFC probe will always run before 
the check code:
   if (nand_nfc.is_initialized) {
     ...

>
> Anyway, I think the current patch (fixing a deadlock) is still ready for
> stable, while I would expect you can fix up this racy situation in a
> subsequent patch. Or perhaps I'm misreading something?

I think my subsequent patch is to move the register/unregister() 
function to the module_init/exit().
And in the module_init(), make sure register NFC driver before NAND driver.
That means I will revert the change that convert module_init/exit() to 
module_platform_driver_probe().

So the whole flow will like that:

module_init()
   register nfc driver
   register nand driver

nand driver probe() {
     # register nfc driver --> move to module_init()
     populate nfc device
       since nfc driver is registered, so nfc_probe() is called (init nfc)
     check whether nfc is initialized
     ...
   }

module_exit()
   unregister nand driver
   unregister nfc driver

>
> Brian
Best Regards,
Josh Wu
Brian Norris - Nov. 13, 2013, 12:10 a.m.
+ Greg KH

Hi Josh,

On Fri, Nov 08, 2013 at 11:46:30AM +0800, Josh Wu wrote:
> On 11/8/2013 2:09 AM, Brian Norris wrote:
> >On Thu, Nov 07, 2013 at 06:26:39PM +0800, Josh Wu wrote:
> >>On 11/7/2013 4:39 PM, Brian Norris wrote:
> >>>On Tue, Nov 05, 2013 at 05:59:07PM +0800, Josh Wu wrote:
> >>>>In the atmel driver probe function, the code shows like following:
> >>>>   atmel_nand_probe(...) {
> >>>>         ...
> >>>>
> >>>>   err_nand_ioremap:
> >>>>         platform_driver_unregister(&atmel_nand_nfc_driver);
> >>>>         return res;
> >>>>   }
> >>>>
> >>>>If no nand flash detected, the driver probe function will goto
> >>>>err_nand_ioremap label.
> >>>>Then platform_driver_unregister() will be called. It will get the
> >>>>lock of atmel_nand device since it is parent of nfc_device. The
> >>>>problem is the lock is already hold by atmel_nand_probe itself.
> >>>>So system will be in a dead lock.
> >>>>
> >>>>This patch just simply removed to platform_driver_unregister() call.
> >>>>When atmel_nand driver is quit the platform_driver_unregister() will
> >>>>be called in atmel_nand_remove().
> >>>The real key, here, is that the platform-driver probe() has no business
> >>>un-registering another driver, right?
> >>right.
> >>
> >>>Shouldn't both drivers just be
> >>>registered/unregistered in the module init/exit, and not in probe()?
> >>currently the NFC driver is registered in the beginning of nand
> >>probe function. After NFC driver is initialized, the rest of the
> >>nand probe function
> >>will check the NFC driver status.
> >>I am not sure it is proper to registered another driver in a probe().
> >This whole 2-driver registration process seems like a totally racy mess.
> >
> >Unless I'm reading this wrong, the code is entirely wrong. You are not
> >properly expressing the dependency between the NFC device and the
> >atmel_nand device, so I think you're relying purely on happenstance that
> >when the NAND probe registers the NFC driver, it will complete the NFC
> >probe before the NAND probe reaches:
> >
> >   if (nand_nfc.is_initialized) {
> >     ...
> 
> Yes, exactly.
> And the NAND probe will also load the NFC device before it reach
> above check code.

No, it loads the *driver*, not the *device*. I'm not familiar with the
driver core guarantees, but I don't think you can guarantee that just
because a driver was registered before another that the corresponding
devices will be probed in that order. Or are you relying on the
dependencies captured by device tree? (I don't think even the
parent/child dependency between NAND/NFC gives you enough.)

> That is what I want: make *nfc probe function* is running before
> reach above check code.
> 
> >
> >This is a race, no?
> 
> So I don't think there is a race, as NFC probe will always run
> before the check code:
>   if (nand_nfc.is_initialized) {
>     ...

It is not clear to me how you guarantee this.

> >
> >Anyway, I think the current patch (fixing a deadlock) is still ready for
> >stable, while I would expect you can fix up this racy situation in a
> >subsequent patch. Or perhaps I'm misreading something?
> 
> I think my subsequent patch is to move the register/unregister()
> function to the module_init/exit().
> And in the module_init(), make sure register NFC driver before NAND driver.
> That means I will revert the change that convert module_init/exit()
> to module_platform_driver_probe().
> 
> So the whole flow will like that:
> 
> module_init()
>   register nfc driver
>   register nand driver
> 
> nand driver probe() {
>     # register nfc driver --> move to module_init()
>     populate nfc device

How do you guarantee this "populate" happens here?

>       since nfc driver is registered, so nfc_probe() is called (init nfc)

I'm not sure where you get this. Can you point to a line in the code?

>     check whether nfc is initialized
>     ...
>   }
> 
> module_exit()
>   unregister nand driver
>   unregister nfc driver

Brian
Wu, Josh - Nov. 13, 2013, 3:26 a.m.
On 11/13/2013 8:10 AM, Brian Norris wrote:
> + Greg KH
>
> Hi Josh,
>
> On Fri, Nov 08, 2013 at 11:46:30AM +0800, Josh Wu wrote:
>> On 11/8/2013 2:09 AM, Brian Norris wrote:
>>> On Thu, Nov 07, 2013 at 06:26:39PM +0800, Josh Wu wrote:
>>>> On 11/7/2013 4:39 PM, Brian Norris wrote:
>>>>> On Tue, Nov 05, 2013 at 05:59:07PM +0800, Josh Wu wrote:
>>>>>> In the atmel driver probe function, the code shows like following:
>>>>>>    atmel_nand_probe(...) {
>>>>>>          ...
>>>>>>
>>>>>>    err_nand_ioremap:
>>>>>>          platform_driver_unregister(&atmel_nand_nfc_driver);
>>>>>>          return res;
>>>>>>    }
>>>>>>
>>>>>> If no nand flash detected, the driver probe function will goto
>>>>>> err_nand_ioremap label.
>>>>>> Then platform_driver_unregister() will be called. It will get the
>>>>>> lock of atmel_nand device since it is parent of nfc_device. The
>>>>>> problem is the lock is already hold by atmel_nand_probe itself.
>>>>>> So system will be in a dead lock.
>>>>>>
>>>>>> This patch just simply removed to platform_driver_unregister() call.
>>>>>> When atmel_nand driver is quit the platform_driver_unregister() will
>>>>>> be called in atmel_nand_remove().
>>>>> The real key, here, is that the platform-driver probe() has no business
>>>>> un-registering another driver, right?
>>>> right.
>>>>
>>>>> Shouldn't both drivers just be
>>>>> registered/unregistered in the module init/exit, and not in probe()?
>>>> currently the NFC driver is registered in the beginning of nand
>>>> probe function. After NFC driver is initialized, the rest of the
>>>> nand probe function
>>>> will check the NFC driver status.
>>>> I am not sure it is proper to registered another driver in a probe().
>>> This whole 2-driver registration process seems like a totally racy mess.
>>>
>>> Unless I'm reading this wrong, the code is entirely wrong. You are not
>>> properly expressing the dependency between the NFC device and the
>>> atmel_nand device, so I think you're relying purely on happenstance that
>>> when the NAND probe registers the NFC driver, it will complete the NFC
>>> probe before the NAND probe reaches:
>>>
>>>    if (nand_nfc.is_initialized) {
>>>      ...
>> Yes, exactly.
>> And the NAND probe will also load the NFC device before it reach
>> above check code.
> No, it loads the *driver*, not the *device*. I'm not familiar with the
> driver core guarantees, but I don't think you can guarantee that just
> because a driver was registered before another that the corresponding
> devices will be probed in that order. Or are you relying on the
> dependencies captured by device tree? (I don't think even the
> parent/child dependency between NAND/NFC gives you enough.)
Hi, Brain

I put  more code here to make it clearer:

arch/arm/boot/dts/sama5d3.dtsi:
nand0: nand@60000000 {
             compatible = "atmel,at91rm9200-nand";
             ...
             ranges;
             ...
             nfc@70000000 {
                 compatible = "atmel,sama5d3-nfc";
                 ...
             };
         };

drivers/mtd/nand/atmel_nand.c
int atmel_of_init_port() {
    ...
    /* load the nfc driver if there is */
   of_platform_populate(np, NULL, NULL, host->dev);   # <--- Manually 
populate the sub node (NFC node) to load NFC driver
   ...
}

static int atmel_nand_nfc_probe(struct platform_device *pdev)
{
     struct atmel_nfc *nfc = &nand_nfc;
     ...
     nfc->is_initialized = true;
     dev_info(&pdev->dev, "NFC is probed.\n");
     return 0;
}

atmel_nand_probe() {
    ...
    res = platform_driver_register(&atmel_nand_nfc_driver);  # Register 
NFC driver.
    ...
    if (IS_ENABLED(CONFIG_OF) && pdev->dev.of_node) {
       res = atmel_of_init_port(host, pdev->dev.of_node);
       #       ^^^^^^^^^^ ^^^^^^  In this function, it will populate the 
NFC node shows in above code block.
       #   It triggers the NFC driver's probe() function. Which will set 
the 'nand_nfc.is_initialized' as true.

    } else {
       ...
    }
    ...

    if (nand_nfc.is_initialized) {               # now we know the NFC 
driver is probed.
       ...
    }
    ...
}

in summary, the following flow to make sure NFC is initialized in 
atmel_nand_probe():
    atmel_nand_probe() {
       Register NFC.
       dt property check. That populates the NFC dt node.
           Via the NFC compatible string, system find the registered NFC 
driver, so call the NFC probe() function.
           NFC probe will initialized the NFC.
       Check NFC is initialized or not.
    }

>
>> That is what I want: make *nfc probe function* is running before
>> reach above check code.
>>
>>> This is a race, no?
>> So I don't think there is a race, as NFC probe will always run
>> before the check code:
>>    if (nand_nfc.is_initialized) {
>>      ...
> It is not clear to me how you guarantee this.
>
>>> Anyway, I think the current patch (fixing a deadlock) is still ready for
>>> stable, while I would expect you can fix up this racy situation in a
>>> subsequent patch. Or perhaps I'm misreading something?
>> I think my subsequent patch is to move the register/unregister()
>> function to the module_init/exit().
>> And in the module_init(), make sure register NFC driver before NAND driver.
>> That means I will revert the change that convert module_init/exit()
>> to module_platform_driver_probe().
>>
>> So the whole flow will like that:
>>
>> module_init()
>>    register nfc driver
>>    register nand driver
>>
>> nand driver probe() {
>>      # register nfc driver --> move to module_init()
>>      populate nfc device
> How do you guarantee this "populate" happens here?

>
>>        since nfc driver is registered, so nfc_probe() is called (init nfc)
> I'm not sure where you get this. Can you point to a line in the code?
atmel_nand_probe() {
    ...
    atmel_of_init_port(host, pdev->dev.of_node);
    ...
}

int atmel_of_init_port() {
    ...
    /* load the nfc driver if there is */
   of_platform_populate(np, NULL, NULL, host->dev);
   ...
}

>
>>      check whether nfc is initialized
>>      ...
>>    }
>>
>> module_exit()
>>    unregister nand driver
>>    unregister nfc driver
> Brian

Best Regards,
Josh Wu
Brian Norris - Nov. 13, 2013, 8:44 a.m.
Hi Josh,

On Wed, Nov 13, 2013 at 11:26:28AM +0800, Josh Wu wrote:
> On 11/13/2013 8:10 AM, Brian Norris wrote:
> >On Fri, Nov 08, 2013 at 11:46:30AM +0800, Josh Wu wrote:
> >>On 11/8/2013 2:09 AM, Brian Norris wrote:
> >>>   if (nand_nfc.is_initialized) {
> >>>     ...
> >>Yes, exactly.
> >>And the NAND probe will also load the NFC device before it reach
> >>above check code.
> >No, it loads the *driver*, not the *device*. I'm not familiar with the
> >driver core guarantees, but I don't think you can guarantee that just
> >because a driver was registered before another that the corresponding
> >devices will be probed in that order. Or are you relying on the
> >dependencies captured by device tree? (I don't think even the
> >parent/child dependency between NAND/NFC gives you enough.)
> 
> I put  more code here to make it clearer:
> 
> arch/arm/boot/dts/sama5d3.dtsi:
> nand0: nand@60000000 {
>             compatible = "atmel,at91rm9200-nand";
>             ...
>             ranges;
>             ...
>             nfc@70000000 {
>                 compatible = "atmel,sama5d3-nfc";
>                 ...
>             };
>         };

Ah, so you're focusing purely on the device tree support? It seems this
driver also supports traditional platform devices, but nobody uses NFC
without DT? That explains some things.

> drivers/mtd/nand/atmel_nand.c
> int atmel_of_init_port() {
>    ...
>    /* load the nfc driver if there is */
>   of_platform_populate(np, NULL, NULL, host->dev);   # <--- Manually
> populate the sub node (NFC node) to load NFC driver
>   ...
> }

Right, that makes sense. So the ordering of the driver registration
stuff is kind of a distraction in this case, then. It's this portion of
the NAND probe that triggers bus_probe_device() to initialize the NFC
device before continuing on.

[snip rest of explanation; thanks!]

So I don't think you have a problem for device tree platforms. But if
someone were to try to instantiate these devices via
platform_device_register(), they will have a problem.

Please, do follow up with the cleanup to the driver registration that
you mentioned.

Brian
Nicolas Ferre - Nov. 13, 2013, 10:33 a.m.
On 13/11/2013 09:44, Brian Norris :
> Hi Josh,
>
> On Wed, Nov 13, 2013 at 11:26:28AM +0800, Josh Wu wrote:
>> On 11/13/2013 8:10 AM, Brian Norris wrote:
>>> On Fri, Nov 08, 2013 at 11:46:30AM +0800, Josh Wu wrote:
>>>> On 11/8/2013 2:09 AM, Brian Norris wrote:
>>>>>    if (nand_nfc.is_initialized) {
>>>>>      ...
>>>> Yes, exactly.
>>>> And the NAND probe will also load the NFC device before it reach
>>>> above check code.
>>> No, it loads the *driver*, not the *device*. I'm not familiar with the
>>> driver core guarantees, but I don't think you can guarantee that just
>>> because a driver was registered before another that the corresponding
>>> devices will be probed in that order. Or are you relying on the
>>> dependencies captured by device tree? (I don't think even the
>>> parent/child dependency between NAND/NFC gives you enough.)
>>
>> I put  more code here to make it clearer:
>>
>> arch/arm/boot/dts/sama5d3.dtsi:
>> nand0: nand@60000000 {
>>              compatible = "atmel,at91rm9200-nand";
>>              ...
>>              ranges;
>>              ...
>>              nfc@70000000 {
>>                  compatible = "atmel,sama5d3-nfc";
>>                  ...
>>              };
>>          };
>
> Ah, so you're focusing purely on the device tree support? It seems this
> driver also supports traditional platform devices, but nobody uses NFC
> without DT? That explains some things.

Absolutely: the NFC is only available on SoC that are pure-DT.

The traditional platform device support is for older SoC that can use 
this driver but do not have the NFC.

I let Josh continue the discussion on the registration topic.

[..]

Bye,

Patch

diff --git a/drivers/mtd/nand/atmel_nand.c b/drivers/mtd/nand/atmel_nand.c
index ef9c9f5..469d4e2 100644
--- a/drivers/mtd/nand/atmel_nand.c
+++ b/drivers/mtd/nand/atmel_nand.c
@@ -2142,7 +2142,6 @@  err_no_card:
 	if (host->dma_chan)
 		dma_release_channel(host->dma_chan);
 err_nand_ioremap:
-	platform_driver_unregister(&atmel_nand_nfc_driver);
 	return res;
 }