Re: [PATCH 04/14] crypto: jitterentropy - Add library API

[Eric Biggers <ebiggers@kernel.org>]

On Fri, Nov 13, 2026 at 12:28:12PM +0100, Thomas Huth wrote:
> Am 18.10.26 um 22:46 schrieb Eric Biggers:
> > Make the jitterentropy RNG accessible using a regular set of functions.
> > 
> > This will allow drbg.c to just access it directly, instead of going
> > through the overly-general crypto_rng abstraction.
> > 
> > Later commits will remove crypto_rng.
> > 
> > Note that the library API should also continue to meet all the same
> > certification requirements, including the health check which occurs on
> > the first call to crypto_jent_alloc().
> > 
> > Signed-off-by: Eric Biggers <ebiggers@kernel.org>
> > ---
> ...
> > +int crypto_jent_get_bytes(struct jitterentropy *rng, u8 *out,
> > +			  unsigned int out_len)
> > +{
> > +	int ret;
> > +
> > +	guard(mutex)(&rng->jent_lock);
> > +
> > +	ret = jent_read_entropy(rng->entropy_collector, out, out_len);
> > +
> > +	if (ret == -3) {
> > +		/* Handle permanent health test error */
> > +		/*
> > +		 * If the kernel was booted with fips=1, it implies that
> > +		 * the entire kernel acts as a FIPS 140 module. In this case
> > +		 * an SP800-90B permanent health test error is treated as
> > +		 * a FIPS module error.
> > +		 */
> > +		if (fips_enabled)
> > +			panic("Jitter RNG permanent health test failure\n");
> > +
> > +		pr_err("Jitter RNG permanent health test failure\n");
> > +		return -EFAULT;
> EFAULT means "bad address" in the sense of a page fault ... not sure whether
> this is the appropriate error code here? Maybe -EIO would be a better fit?

I intentionally left the EFAULT unchanged from the old code in
jent_kcapi_random(), to limit the scope of this conversion.  I agree
that EFAULT is not an appropriate error code here though, and feel free
to send a separate patch that changes it to something else.  Thanks!

- Eric