Re: [PATCH v2 3/8] kernel/params.c: generalize bool_enable_only

["Luis R. Rodriguez" <mcgrof@suse.com>]

On Thu, Apr 23, 2015 at 11:22:49AM -0400, Tejun Heo wrote:
> Hello,
> 
> On Wed, Apr 22, 2015 at 02:55:06PM -0700, Luis R. Rodriguez wrote:
> > +int param_set_bool_enable_only(const char *val, const struct kernel_param *kp)
> > +{
> > +	int err = 0;
> > +	bool new_value;
> > +	bool orig_value = *(bool *)kp->arg;
> > +	struct kernel_param dummy_kp = *kp;
> > +
> > +	dummy_kp.arg = &new_value;
> > +
> > +	err = param_set_bool(val, &dummy_kp);
> > +	if (err)
> > +		return err;
> > +
> > +	/* Don't let them unset it once it's set! */
> > +	if (!new_value && orig_value)
> > +		return -EROFS;
> 
> I know that this was moved from another place but as we're making it
> generic now I'm a bit curious about -EROFS.  Wouldn't -EINVAL be a
> more conventional choice here?

Let's see, I tested to see what errors we get:

Userspace:

-EROFS:

mcgrof@ergon ~/devel/test-sig-force-general $ sudo insmod ./hello.ko test_sig_enforce=1
insmod: ERROR: could not insert module ./hello.ko: Read-only file system

-EINVAL:

mcgrof@ergon ~/devel/test-sig-force-general $ sudo insmod ./hello.ko test_sig_enforce=1
insmod: ERROR: could not insert module ./hello.ko: Invalid parameters

Kernel space:

Both of these returns yield this on the kernel ring buffer:

mcgrof@ergon ~/devel/test-sig-force-general $ sudo dmesg -c
[124677.202875] hello: `1' invalid for parameter `test_sig_enforce'

Alternative candidates:

#define EBADRQC         56      /* Invalid request code */ 
#define EOPNOTSUPP      95      /* Operation not supported on transport endpoint */

Perhaps EOPNOTSUPP is best?

  Luis