[PATCH 0/2] mmc: meson-gx: fix module unloading

[PATCH 0/2] mmc: meson-gx: fix module unloading

[Michał Zegan]

This patch contains fixes for the meson-gx-mmc driver, that allow the module to be unloaded correctly.
Before trying to unload the module caused kernel warnings.

Michał Zegan (2):
  mmc: meson-gx: prevent cfg_div_clk from being disabled on init
  mmc: meson-gx: remove mmc host on device removal

 drivers/mmc/host/meson-gx-mmc.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

-- 
2.11.0

[PATCH 1/2] mmc: meson-gx: prevent cfg_div_clk from being disabled on init

[Michał Zegan]

At the end of function meson_mmc_clk_init, the cfg_div clock was
prepared, enabled and configured, but then immediately disabled due to bogus if statements.
That made later calls to clk_disable_unprepare executed during module removal to fail with a kernel warning.
Fix that by changing the code to disable clock only when it failed to be configured.

Signed-off-by: Michał Zegan <webczat@webczatnet.pl>
---
 drivers/mmc/host/meson-gx-mmc.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/drivers/mmc/host/meson-gx-mmc.c b/drivers/mmc/host/meson-gx-mmc.c
index 09739352834c..d444b6bfa02b 100644
--- a/drivers/mmc/host/meson-gx-mmc.c
+++ b/drivers/mmc/host/meson-gx-mmc.c
@@ -324,11 +324,12 @@ static int meson_mmc_clk_init(struct meson_host *host)
 	writel(cfg, host->regs + SD_EMMC_CFG);
 
 	ret = clk_prepare_enable(host->cfg_div_clk);
-	if (!ret)
+	if (!ret) {
 		ret = meson_mmc_clk_set(host, f_min);
 
-	if (!ret)
-		clk_disable_unprepare(host->cfg_div_clk);
+		if (ret)
+			clk_disable_unprepare(host->cfg_div_clk);
+	}
 
 	return ret;
 }
-- 
2.11.0

[PATCH 2/2] mmc: meson-gx: remove mmc host on device removal

[Michał Zegan]

The mmc host was added in meson_mmc_probe, but never removed in meson_mmc_remove.
Fix that by removing the host before deallocating other resources.

Signed-off-by: Michał Zegan <webczat@webczatnet.pl>
---
 drivers/mmc/host/meson-gx-mmc.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/mmc/host/meson-gx-mmc.c b/drivers/mmc/host/meson-gx-mmc.c
index d444b6bfa02b..bb83446118a3 100644
--- a/drivers/mmc/host/meson-gx-mmc.c
+++ b/drivers/mmc/host/meson-gx-mmc.c
@@ -818,6 +818,8 @@ static int meson_mmc_remove(struct platform_device *pdev)
 	if (WARN_ON(!host))
 		return 0;
 
+	mmc_remove_host(host->mmc);
+
 	if (host->bounce_buf)
 		dma_free_coherent(host->dev, host->bounce_buf_size,
 				  host->bounce_buf, host->bounce_dma_addr);
-- 
2.11.0

Re: [PATCH 1/2] mmc: meson-gx: prevent cfg_div_clk from being disabled on init

[Kevin Hilman]

Michał Zegan <webczat@webczatnet.pl> writes:

> At the end of function meson_mmc_clk_init, the cfg_div clock was
> prepared, enabled and configured, but then immediately disabled due to bogus if statements.
> That made later calls to clk_disable_unprepare executed during module removal to fail with a kernel warning.
> Fix that by changing the code to disable clock only when it failed to be configured.
>
> Signed-off-by: Michał Zegan <webczat@webczatnet.pl>

I believe this one is not needed any more after Heiner's cleanups?

Kevin

Re: [PATCH 2/2] mmc: meson-gx: remove mmc host on device removal

[Kevin Hilman]

Michał Zegan <webczat@webczatnet.pl> writes:

> The mmc host was added in meson_mmc_probe, but never removed in meson_mmc_remove.
> Fix that by removing the host before deallocating other resources.
>
> Signed-off-by: Michał Zegan <webczat@webczatnet.pl>

Reviewed-by: Kevin Hilman <khilman@baylibre.com>

> ---
>  drivers/mmc/host/meson-gx-mmc.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/drivers/mmc/host/meson-gx-mmc.c b/drivers/mmc/host/meson-gx-mmc.c
> index d444b6bfa02b..bb83446118a3 100644
> --- a/drivers/mmc/host/meson-gx-mmc.c
> +++ b/drivers/mmc/host/meson-gx-mmc.c
> @@ -818,6 +818,8 @@ static int meson_mmc_remove(struct platform_device *pdev)
>  	if (WARN_ON(!host))
>  		return 0;
>  
> +	mmc_remove_host(host->mmc);
> +
>  	if (host->bounce_buf)
>  		dma_free_coherent(host->dev, host->bounce_buf_size,
>  				  host->bounce_buf, host->bounce_dma_addr);

Re: [PATCH 1/2] mmc: meson-gx: prevent cfg_div_clk from being disabled on init

[Michał Zegan]

W dniu 17.02.2017 o 20:44, Kevin Hilman pisze:
> Michał Zegan <webczat@webczatnet.pl> writes:
>
>> At the end of function meson_mmc_clk_init, the cfg_div clock was
>> prepared, enabled and configured, but then immediately disabled due to bogus if statements.
>> That made later calls to clk_disable_unprepare executed during module removal to fail with a kernel warning.
>> Fix that by changing the code to disable clock only when it failed to be configured.
>>
>> Signed-off-by: Michał Zegan <webczat@webczatnet.pl>
> I believe this one is not needed any more after Heiner's cleanups?
I think it is not needed, yes.
>
> Kevin

Re: [PATCH 2/2] mmc: meson-gx: remove mmc host on device removal

[Michał Zegan]

W dniu 17.02.2017 o 20:47, Kevin Hilman pisze:
> Michał Zegan <webczat@webczatnet.pl> writes:
>
>> The mmc host was added in meson_mmc_probe, but never removed in meson_mmc_remove.
>> Fix that by removing the host before deallocating other resources.
>>
>> Signed-off-by: Michał Zegan <webczat@webczatnet.pl>
> Reviewed-by: Kevin Hilman <khilman@baylibre.com>
I do not know how the mmc driver looks like after Heiner's cleanup. Does
this patch get obsoleted by the prior cleanup, or conflict with it in
some way?
>
>> ---
>>  drivers/mmc/host/meson-gx-mmc.c | 2 ++
>>  1 file changed, 2 insertions(+)
>>
>> diff --git a/drivers/mmc/host/meson-gx-mmc.c b/drivers/mmc/host/meson-gx-mmc.c
>> index d444b6bfa02b..bb83446118a3 100644
>> --- a/drivers/mmc/host/meson-gx-mmc.c
>> +++ b/drivers/mmc/host/meson-gx-mmc.c
>> @@ -818,6 +818,8 @@ static int meson_mmc_remove(struct platform_device *pdev)
>>  	if (WARN_ON(!host))
>>  		return 0;
>>  
>> +	mmc_remove_host(host->mmc);
>> +
>>  	if (host->bounce_buf)
>>  		dma_free_coherent(host->dev, host->bounce_buf_size,
>>  				  host->bounce_buf, host->bounce_dma_addr);

Re: [PATCH 2/2] mmc: meson-gx: remove mmc host on device removal

[Kevin Hilman]

Michał Zegan <webczat@webczatnet.pl> writes:

> W dniu 17.02.2017 o 20:47, Kevin Hilman pisze:
>> Michał Zegan <webczat@webczatnet.pl> writes:
>>
>>> The mmc host was added in meson_mmc_probe, but never removed in meson_mmc_remove.
>>> Fix that by removing the host before deallocating other resources.
>>>
>>> Signed-off-by: Michał Zegan <webczat@webczatnet.pl>
>> Reviewed-by: Kevin Hilman <khilman@baylibre.com>
>
> I do not know how the mmc driver looks like after Heiner's cleanup. Does
> this patch get obsoleted by the prior cleanup, or conflict with it in
> some way?

It's not obsoleted, as I don't think Heiner's series fixes this problem,
but it may have minor conflicts.

I suggest you rebase this patch on top of Heiner's v2 to double check,
and resend if necessary, noting in the patch (after the '---') what it
applies on top of so that the MMC maintainers don't have to figure it
out.

Thanks,

Kevin

>>
>>> ---
>>>  drivers/mmc/host/meson-gx-mmc.c | 2 ++
>>>  1 file changed, 2 insertions(+)
>>>
>>> diff --git a/drivers/mmc/host/meson-gx-mmc.c b/drivers/mmc/host/meson-gx-mmc.c
>>> index d444b6bfa02b..bb83446118a3 100644
>>> --- a/drivers/mmc/host/meson-gx-mmc.c
>>> +++ b/drivers/mmc/host/meson-gx-mmc.c
>>> @@ -818,6 +818,8 @@ static int meson_mmc_remove(struct platform_device *pdev)
>>>  	if (WARN_ON(!host))
>>>  		return 0;
>>>  
>>> +	mmc_remove_host(host->mmc);
>>> +
>>>  	if (host->bounce_buf)
>>>  		dma_free_coherent(host->dev, host->bounce_buf_size,
>>>  				  host->bounce_buf, host->bounce_dma_addr);