Re: [PATCH] block: fix part_pack_uuid() build error

[Andrew Morton <akpm@linux-foundation.org>]

On Sun, 24 Feb 2013 10:15:44 -0500
Mimi Zohar <zohar@linux.vnet.ibm.com> wrote:

> Fix a build error when CONFIG_BLOCK is not enabled, by defining
> a wrapper called blk_part_pack_uuid().  The wrapper returns
> -EINVAL, when CONFIG_BLOCK is not defined.
> 
> security/integrity/ima/ima_policy.c:538:4: error: implicit declaration
> of function 'part_pack_uuid' [-Werror=implicit-function-declaration]
> 
> ...
>
> diff --git a/security/integrity/ima/ima_policy.c b/security/integrity/ima/ima_policy.c
> index b27535a..399433a 100644
> --- a/security/integrity/ima/ima_policy.c
> +++ b/security/integrity/ima/ima_policy.c
>  			ima_log_string(ab, "fsuuid", args[0].from);
>  
>  			if (memchr_inv(entry->fsuuid, 0x00,
> -			    sizeof(entry->fsuuid))) {
> +				       sizeof(entry->fsuuid))) {
>  				result = -EINVAL;
>  				break;
>  			}
>  
> -			part_pack_uuid(args[0].from, entry->fsuuid);
> -			entry->flags |= IMA_FSUUID;
> -			result = 0;
> +			result = blk_part_pack_uuid(args[0].from,
> +						    entry->fsuuid);
> +			if (!result)
> +				entry->flags |= IMA_FSUUID;

This will cause ima_parse_rule() to newly return -EINVAL if the fsuuid=
option is used when CONFIG_BLOCK=n.

This functional change was not changelogged, forcing me to ask: was it
deliberate or was it accidental?

And it is a non-back-compatible change, introducing some potential to
break existing userspace code.  Is the risk considered acceptable?  If
so, why?