Hello Oleg,

(I wrote that patch.)

You are right, it works only if the property does not already exist.
In that case we'd probably also need a free() in the else-branch,
where of_new_property is called (otherwise we have a leak), because
that actually copies what is contained in data.

Sorry for that regression.


Best Regards
Florian

Am Montag, den 27.08.2018, 13:57 +0000 schrieb [email protected]:
> With commit 748d1b8 the function do_of_property_set_now is introduced. With 
> this
> commit an unwanted free(data) is added wich leads to that an free on the value
> is done that is previously set to the property. As a consequence the command
> of_dump reads freed data an printis the value as hex. Fix this by moving
> free(data) to the right place like before the above commit.
> 
> Signed-off-by: Oleg Karfich <[email protected]>
> ---
>  commands/of_property.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/commands/of_property.c b/commands/of_property.c
> index 8ce12e5..ea32e51 100644
> --- a/commands/of_property.c
> +++ b/commands/of_property.c
> @@ -275,10 +275,12 @@ static int do_of_property_set_now(struct device_node 
> *root, const char *path,
>               pp->value_const = NULL;
>  
>               /* limit property data to the actual size */
> -             if (len)
> +             if (len) {
>                       pp->value = xrealloc(data, len);
> -             else
> +             } else {
>                       pp->value = NULL;
> +                     free(data);
> +             }
>  
>               pp->length = len;
>       } else {
> @@ -374,7 +376,6 @@ static int do_of_property(int argc, char *argv[])
>                               free(data);
>               } else {
>                       ret = do_of_property_set_now(NULL, path, propname, 
> data, len);
> -                     free(data);
>               }
>  
>               return ret;
_______________________________________________
barebox mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/barebox

Reply via email to