Re: [PATCH 21/46] selinux: Two function calls less in range_read() after error detection

[Paul Moore]

On Tue, Jan 17, 2017 at 11:35 AM, Casey Schaufler
<casey@xxxxxxxxxxxxxxxx> wrote:
> On 1/15/2017 7:21 AM, SF Markus Elfring wrote:
>> From: Markus Elfring <elfring@xxxxxxxxxxxxxxxxxxxxx>
>> Date: Sat, 14 Jan 2017 20:20:15 +0100
>>
>> Adjust a jump target to avoid two calls of the function "kfree" at the end
>> after a memory allocation failed for the local variable "rt".
>>
>> Signed-off-by: Markus Elfring <elfring@xxxxxxxxxxxxxxxxxxxxx>
>> ---
>>  security/selinux/ss/policydb.c | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/security/selinux/ss/policydb.c b/security/selinux/ss/policydb.c
>> index 4cd96ce51322..0d2f64558c0a 100644
>> --- a/security/selinux/ss/policydb.c
>> +++ b/security/selinux/ss/policydb.c
>> @@ -1857,7 +1857,7 @@ static int range_read(struct policydb *p, void *fp)
>>               rt = kzalloc(sizeof(*rt), GFP_KERNEL);
>>               if (!rt) {
>>                       rc = -ENOMEM;
>> -                     goto out;
>> +                     goto exit;
>
> Why not "return rc;"?
> goto to a return is wrong.

Agree with Casey, but also see my previous comments about the
convenience of using a single error handling goto for loops like this.

>>               }
>>
>>               rc = next_entry(buf, fp, (sizeof(u32) * 2));
>> @@ -1909,6 +1909,7 @@ static int range_read(struct policydb *p, void *fp)
>>  out:
>>       kfree(rt);
>>       kfree(r);
>> +exit:
>>       return rc;
>>  }
>>
>



-- 
paul moore
www.paul-moore.com