Re: [RFC] [PATCH] merge *_vm_enough_memory()s into a common helper

[Chris Wright <chrisw@osdl.org>]

* Serge E. Hallyn (serue@us.ibm.com) wrote:

I'm fine with this with a few nits.  Although I don't think it will apply
to current bk which has merge error in this area right now.  Stephen,
are you ok with the way this one generates audit messages?

> +	return __vm_enough_memory(pages,
> +			(cap_capable(current, CAP_SYS_ADMIN) == 0));

A temp variable isn't going to be costly and makes it easier to read.

> +	return __vm_enough_memory(pages,
> +			(dummy_capable(current, CAP_SYS_ADMIN) == 0));

same here

> + * Note that secondary_ops->capable and task_has_perm return 0 if
> + * the capability is granted, but __vm_enough_memory requires 1 if
> + * the capability is granted.
>   */
>  static int selinux_vm_enough_memory(long pages)
>  {
> -	unsigned long free, allowed;
> -	int rc;
> +	int rc, cap_sys_admin = 0;
<snip>
> +	rc = secondary_ops->capable(current, CAP_SYS_ADMIN);
> +	if (rc == 0)
> +		cap_sys_admin = avc_has_perm_noaudit(tsec->sid, tsec->sid,
> +					SECCLASS_CAPABILITY,
> +					CAP_TO_MASK(CAP_SYS_ADMIN),
> +					NULL);
>  
> -	vm_unacct_memory(pages);
> +	if (rc == 0)
> +		cap_sys_admin = 1;

This sure looks wrong.  Did you mean rc = avc_has_perm_noaudit()?

thanks,
-chris
-- 
Linux Security Modules     http://lsm.immunix.org     http://lsm.bkbits.net