Re: [PATCH] Input: atmel_mxt_ts - Use snprintf for sysfs attributeshow method

[Daniel Kurtz]

Hi Dmitry,

I fear the following patch, and its 'Acked-by' from Nick Dyer, may
have gotten lost:

On Fri, Sep 16, 2011 at 3:40 AM, Daniel Kurtz <djkurtz@xxxxxxxxxxxx> wrote:
>
> Sysfs attribute show methods are always passed a buffer of length
> PAGE_SIZE.  To keep from overwriting this buffer and causing havoc, use
> snprintf() to guarantee we never write more than the buffer can hold.
>
> In addition, at least for my touchscreen, the number and size of objects
> was far too big to fit in a single 4K page.  Therefore, this patch also
> trims some redundant framing text to leave more room for actual data.
>
> Signed-off-by: Daniel Kurtz <djkurtz@xxxxxxxxxxxx>
> ---
>  drivers/input/touchscreen/atmel_mxt_ts.c |   21 +++++++++++++++------
>  1 files changed, 15 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/input/touchscreen/atmel_mxt_ts.c b/drivers/input/touchscreen/atmel_mxt_ts.c
> index f5d6685..a596c27 100644
> --- a/drivers/input/touchscreen/atmel_mxt_ts.c
> +++ b/drivers/input/touchscreen/atmel_mxt_ts.c
> @@ -910,12 +910,17 @@ static ssize_t mxt_object_show(struct device *dev,
>        for (i = 0; i < data->info.object_num; i++) {
>                object = data->object_table + i;
>
> -               count += sprintf(buf + count,
> -                               "Object Table Element %d(Type %d)\n",
> +               count += snprintf(buf + count, PAGE_SIZE - count,
> +                               "Object[%d] (Type %d)\n",
>                                i + 1, object->type);
> +               if (count >= PAGE_SIZE)
> +                       return PAGE_SIZE - 1;
>
>                if (!mxt_object_readable(object->type)) {
> -                       count += sprintf(buf + count, "\n");
> +                       count += snprintf(buf + count, PAGE_SIZE - count,
> +                                       "\n");
> +                       if (count >= PAGE_SIZE)
> +                               return PAGE_SIZE - 1;
>                        continue;
>                }
>
> @@ -925,11 +930,15 @@ static ssize_t mxt_object_show(struct device *dev,
>                        if (error)
>                                return error;
>
> -                       count += sprintf(buf + count,
> -                                       "  Byte %d: 0x%x (%d)\n", j, val, val);
> +                       count += snprintf(buf + count, PAGE_SIZE - count,
> +                                       "\t[%2d]: %02x (%d)\n", j, val, val);
> +                       if (count >= PAGE_SIZE)
> +                               return PAGE_SIZE - 1;
>                }
>
> -               count += sprintf(buf + count, "\n");
> +               count += snprintf(buf + count, PAGE_SIZE - count, "\n");
> +               if (count >= PAGE_SIZE)
> +                       return PAGE_SIZE - 1;
>        }
>
>        return count;
> --
> 1.7.3.1
>

On Thu, Oct 6, 2011 at 5:59 PM, Nick Dyer <nick.dyer@xxxxxxxxxxx> wrote:
> Hi Daniel-
>
> Just to introduce myself, I'm working for Atmel in UK on the atmel_mxt_ts
> driver.
>
> Daniel Kurtz wrote:
>> Sysfs attribute show methods are always passed a buffer of length
>> PAGE_SIZE.  To keep from overwriting this buffer and causing havoc,
>> use snprintf() to guarantee we never write more than the buffer can
>> hold.
>>
>> In addition, at least for my touchscreen, the number and size of
>> objects was far too big to fit in a single 4K page.  Therefore, this
>> patch also trims some redundant framing text to leave more room for
>> actual data.
>
> Yes, I'd noticed this problem as well. I was planning to address it by
> splitting up the file, but your patch looks like a good solution for now.
>
> I've tested it successfully on several chips.
>
> Acked-by: Nick Dyer <nick.dyer@xxxxxxxxxxx>
>
> --
> Nick Dyer
> Software Engineer, ITDev  Ltd
>
> Hardware and Software Development Consultancy
>
> Email:   nick.dyer@xxxxxxxxxxx
> Website: http://www.itdev.co.uk
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/