media: synopsys: hdmirx: Fix signedness bug in hdmirx_parse_dt()

[PATCH next] media: synopsys: hdmirx: Fix signedness bug in hdmirx_parse_dt()

Posted by Dan Carpenter 11 months, 1 week ago

The num_clks is set this way:

	hdmirx_dev->num_clks = devm_clk_bulk_get_all(dev, &hdmirx_dev->clks);
        if (hdmirx_dev->num_clks < 1)
                return -ENODEV;

The devm_clk_bulk_get_all() function returns negative error codes so the
hdmirx_dev->num_cks variable needs to be signed for the error handling to
work.

Fixes: 7b59b132ad43 ("media: platform: synopsys: Add support for HDMI input driver")
Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
---
 drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c b/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
index 4ffc86ad6c35..e0d3fed87a92 100644
--- a/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
+++ b/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
@@ -154,7 +154,7 @@ struct snps_hdmirx_dev {
 	bool hpd_trigger_level_high;
 	bool tmds_clk_ratio;
 	bool plugged;
-	u32 num_clks;
+	int num_clks;
 	u32 edid_blocks_written;
 	u32 cur_fmt_fourcc;
 	u32 color_depth;
-- 
2.47.2

Re: [PATCH next] media: synopsys: hdmirx: Fix signedness bug in hdmirx_parse_dt()

Posted by Dmitry Osipenko 11 months, 1 week ago

On 3/7/25 12:30, Dan Carpenter wrote:
> The num_clks is set this way:
> 
> 	hdmirx_dev->num_clks = devm_clk_bulk_get_all(dev, &hdmirx_dev->clks);
>         if (hdmirx_dev->num_clks < 1)
>                 return -ENODEV;
> 
> The devm_clk_bulk_get_all() function returns negative error codes so the
> hdmirx_dev->num_cks variable needs to be signed for the error handling to
> work.
> 
> Fixes: 7b59b132ad43 ("media: platform: synopsys: Add support for HDMI input driver")
> Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
> ---
>  drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c b/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
> index 4ffc86ad6c35..e0d3fed87a92 100644
> --- a/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
> +++ b/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
> @@ -154,7 +154,7 @@ struct snps_hdmirx_dev {
>  	bool hpd_trigger_level_high;
>  	bool tmds_clk_ratio;
>  	bool plugged;
> -	u32 num_clks;
> +	int num_clks;
>  	u32 edid_blocks_written;
>  	u32 cur_fmt_fourcc;
>  	u32 color_depth;

Reviewed-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>

Would be also good to return the original error code. There is no need
to check for the < 1 clock, it should be the < 0 check. Can be done in a
separate patch later. Thanks for the fix!

-- 
Best regards,
Dmitry

Re: [PATCH next] media: synopsys: hdmirx: Fix signedness bug in hdmirx_parse_dt()

Posted by Dan Carpenter 11 months, 1 week ago

On Fri, Mar 07, 2025 at 12:36:47PM +0300, Dmitry Osipenko wrote:
> On 3/7/25 12:30, Dan Carpenter wrote:
> > diff --git a/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c b/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
> > index 4ffc86ad6c35..e0d3fed87a92 100644
> > --- a/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
> > +++ b/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
> > @@ -154,7 +154,7 @@ struct snps_hdmirx_dev {
> >  	bool hpd_trigger_level_high;
> >  	bool tmds_clk_ratio;
> >  	bool plugged;
> > -	u32 num_clks;
> > +	int num_clks;
> >  	u32 edid_blocks_written;
> >  	u32 cur_fmt_fourcc;
> >  	u32 color_depth;
> 
> Reviewed-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
> 
> Would be also good to return the original error code. There is no need
> to check for the < 1 clock, it should be the < 0 check. Can be done in a
> separate patch later. Thanks for the fix!

I'm not very familiar with th edevm_clk_bulk_get_all() function and it's
not documented.  But clk_bulk_get_all() does return zero, so I can see why
people would be confused.

regards,
dan carpenter

Re: [PATCH next] media: synopsys: hdmirx: Fix signedness bug in hdmirx_parse_dt()

Posted by Dmitry Osipenko 11 months, 1 week ago

On 3/7/25 12:45, Dan Carpenter wrote:
> On Fri, Mar 07, 2025 at 12:36:47PM +0300, Dmitry Osipenko wrote:
>> On 3/7/25 12:30, Dan Carpenter wrote:
>>> diff --git a/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c b/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
>>> index 4ffc86ad6c35..e0d3fed87a92 100644
>>> --- a/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
>>> +++ b/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
>>> @@ -154,7 +154,7 @@ struct snps_hdmirx_dev {
>>>  	bool hpd_trigger_level_high;
>>>  	bool tmds_clk_ratio;
>>>  	bool plugged;
>>> -	u32 num_clks;
>>> +	int num_clks;
>>>  	u32 edid_blocks_written;
>>>  	u32 cur_fmt_fourcc;
>>>  	u32 color_depth;
>>
>> Reviewed-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
>>
>> Would be also good to return the original error code. There is no need
>> to check for the < 1 clock, it should be the < 0 check. Can be done in a
>> separate patch later. Thanks for the fix!
> 
> I'm not very familiar with th edevm_clk_bulk_get_all() function and it's
> not documented.  But clk_bulk_get_all() does return zero, so I can see why
> people would be confused.

We will take care of it, thanks again.

-- 
Best regards,
Dmitry

Re: [PATCH next] media: synopsys: hdmirx: Fix signedness bug in hdmirx_parse_dt()

Posted by Hans Verkuil 11 months, 1 week ago

On 07/03/2025 10:47, Dmitry Osipenko wrote:
> On 3/7/25 12:45, Dan Carpenter wrote:
>> On Fri, Mar 07, 2025 at 12:36:47PM +0300, Dmitry Osipenko wrote:
>>> On 3/7/25 12:30, Dan Carpenter wrote:
>>>> diff --git a/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c b/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
>>>> index 4ffc86ad6c35..e0d3fed87a92 100644
>>>> --- a/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
>>>> +++ b/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
>>>> @@ -154,7 +154,7 @@ struct snps_hdmirx_dev {
>>>>  	bool hpd_trigger_level_high;
>>>>  	bool tmds_clk_ratio;
>>>>  	bool plugged;
>>>> -	u32 num_clks;
>>>> +	int num_clks;
>>>>  	u32 edid_blocks_written;
>>>>  	u32 cur_fmt_fourcc;
>>>>  	u32 color_depth;
>>>
>>> Reviewed-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
>>>
>>> Would be also good to return the original error code. There is no need
>>> to check for the < 1 clock, it should be the < 0 check. Can be done in a
>>> separate patch later. Thanks for the fix!
>>
>> I'm not very familiar with th edevm_clk_bulk_get_all() function and it's
>> not documented.  But clk_bulk_get_all() does return zero, so I can see why
>> people would be confused.
> 
> We will take care of it, thanks again.
> 

I'm confused. Is Dan's patch correct or is more work needed? If more work is
needed, then I prefer to have a single patch correcting the devm_clk_bulk_get_all()
handling.

Regards,

	Hans

Re: [PATCH next] media: synopsys: hdmirx: Fix signedness bug in hdmirx_parse_dt()

Posted by Dmitry Osipenko 11 months, 1 week ago

On 3/7/25 12:58, Hans Verkuil wrote:
> On 07/03/2025 10:47, Dmitry Osipenko wrote:
>> On 3/7/25 12:45, Dan Carpenter wrote:
>>> On Fri, Mar 07, 2025 at 12:36:47PM +0300, Dmitry Osipenko wrote:
>>>> On 3/7/25 12:30, Dan Carpenter wrote:
>>>>> diff --git a/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c b/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
>>>>> index 4ffc86ad6c35..e0d3fed87a92 100644
>>>>> --- a/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
>>>>> +++ b/drivers/media/platform/synopsys/hdmirx/snps_hdmirx.c
>>>>> @@ -154,7 +154,7 @@ struct snps_hdmirx_dev {
>>>>>  	bool hpd_trigger_level_high;
>>>>>  	bool tmds_clk_ratio;
>>>>>  	bool plugged;
>>>>> -	u32 num_clks;
>>>>> +	int num_clks;
>>>>>  	u32 edid_blocks_written;
>>>>>  	u32 cur_fmt_fourcc;
>>>>>  	u32 color_depth;
>>>>
>>>> Reviewed-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
>>>>
>>>> Would be also good to return the original error code. There is no need
>>>> to check for the < 1 clock, it should be the < 0 check. Can be done in a
>>>> separate patch later. Thanks for the fix!
>>>
>>> I'm not very familiar with th edevm_clk_bulk_get_all() function and it's
>>> not documented.  But clk_bulk_get_all() does return zero, so I can see why
>>> people would be confused.
>>
>> We will take care of it, thanks again.
>>
> 
> I'm confused. Is Dan's patch correct or is more work needed? If more work is
> needed, then I prefer to have a single patch correcting the devm_clk_bulk_get_all()
> handling.

Dan's patch is good, please merge it.

Propagation of original error code that I said about will be additional
improvement.

-- 
Best regards,
Dmitry