Re: [PATCH 4/4] UBI: replace MTD_UBI_BEB_LIMIT with user-space parameter

[Richard Genoud <richard.genoud@gmail.com>]

2012/8/20 Artem Bityutskiy <dedekind1@gmail.com>:
> On Mon, 2012-08-20 at 08:55 +0200, Richard Genoud wrote:
>> Hi Artem,
>> 2012/8/19 Artem Bityutskiy <dedekind1@gmail.com>:
>> > Yeah, I wanted to make it 1..256 but forgot, will do now. 0..256 would
>> > need some more work to avoid division by 0.
>> Division by 0 is handled in the get_bad_peb_limit() function, I don't
>> see another dangerous place.
>
>         if (mult_frac(limit, 1024, max_beb_per1024) < device_pebs)
>
> will divide by 0 if max_beb_per1024 is 0.
>
just a few lines before, you've got:
        /*
         * We don't want a division by 0, and having max_beb_per1024 == 0 is ok
         */
        if (!max_beb_per1024)
                return 0;
from commit abae1f1
(or I'm not looking at the right line ?)

>> (and if we want to be coherent with user-space, it should be 0..255,
>> as the range is coded with an u8)
>
> I think it should be uint16_t instead, because we are defining ABI here
> and we should not assume no one will ever nee values higher than 255.
I agree with you, even if 25% of reserved space for bad block seems
insane, we never know...
I'll update that.



-- 
for me, ck means con kolivas and not calvin klein... does it mean I'm a geek ?