Re: [PATCH] fbdev: omapfb: panel-sony-acx565akm: Simplify show_cabc_available_modes()

[Christophe JAILLET <christophe.jaillet@wanadoo.fr>]

Le 09/08/2024 à 16:42, Dan Carpenter a écrit :
> On Thu, Aug 08, 2024 at 11:46:11AM +0200, Christophe JAILLET wrote:
>> Use sysfs_emit_at() instead of snprintf() + custom logic.
>> Using sysfs_emit_at() is much more simple.
>>
>> Also, sysfs_emit() is already used in this function, so using
>> sysfs_emit_at() is more consistent.
>>
>> Also simplify the logic:
>>    - always add a space after an entry
>>    - change the last space into a '\n'
>>
>> Finally it is easy to see that, given the size of cabc_modes, PAGE_SIZE
>> can not be reached.
>> So better keep everything simple (and correct).
>>
>> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
>> ---
>>   .../omap2/omapfb/displays/panel-sony-acx565akm.c  | 15 ++++++++-------
>>   1 file changed, 8 insertions(+), 7 deletions(-)
>>
>> diff --git a/drivers/video/fbdev/omap2/omapfb/displays/panel-sony-acx565akm.c b/drivers/video/fbdev/omap2/omapfb/displays/panel-sony-acx565akm.c
>> index 71d2e015960c..fc975615d5c9 100644
>> --- a/drivers/video/fbdev/omap2/omapfb/displays/panel-sony-acx565akm.c
>> +++ b/drivers/video/fbdev/omap2/omapfb/displays/panel-sony-acx565akm.c
>> @@ -466,19 +466,20 @@ static ssize_t show_cabc_available_modes(struct device *dev,
>>   		char *buf)
>>   {
>>   	struct panel_drv_data *ddata = dev_get_drvdata(dev);
>> -	int len;
>> +	int len = 0;
>>   	int i;
>>   
>>   	if (!ddata->has_cabc)
>>   		return sysfs_emit(buf, "%s\n", cabc_modes[0]);
>>   
>> -	for (i = 0, len = 0;
>> -	     len < PAGE_SIZE && i < ARRAY_SIZE(cabc_modes); i++)
>> -		len += snprintf(&buf[len], PAGE_SIZE - len, "%s%s%s",
>> -			i ? " " : "", cabc_modes[i],
>> -			i == ARRAY_SIZE(cabc_modes) - 1 ? "\n" : "");
>> +	for (i = 0; i < ARRAY_SIZE(cabc_modes); i++)
>> +		len += sysfs_emit_at(buf, len, "%s ", cabc_modes[i]);
>> +
>> +	/* Remove the trailing space */
>> +	if (len)
>> +		buf[len - 1] = '\n';
> 
> I'm uncomfortable with this line.  It assumes we don't overflow PAGE_SIZE where
> the original code was careful about checking.  Probably easiest to do what the
> original code did and say:
> 

Hi Dan,

I don't follow you. AFAIK, it does not assume anything.

Thanks to sysfs_emit_at(), len can only be in [0..PAGE_SIZE-1] because 
the trailing \0 is not counted.

So, as len != 0, len-1 is in [0..PAGE_SIZE-2].

How do you think an overflow could happen?


Also, all this code does is buildind:
	"off, ui, still-image, moving-image\n"

So in any case, an overflow is impossible, and really unlikely in the 
future as well.


 From my PoV, my proposed patch is both much more readable and still 
correct in all cases.

CJ

> 	for (i = 0; i < ARRAY_SIZE(cabc_modes); i++)
> 		len += sysfs_emit_at(buf, len, "%s%s", cabc_modes[i],
> 				     i == ARRAY_SIZE(cabc_modes) - 1 ? "\n" : "");
> 
> regards,
> dan carpenter
> 
> 
> 
>