Re: [PATCH v7 3/5] crypto: tpm2_key: Introduce a TPM2 key type

[Stefan Berger]

On 6/4/24 13:23, Jarkko Sakkinen wrote:
> On Fri May 31, 2024 at 3:35 AM EEST, Stefan Berger wrote:
> >

> > >    
> > > -	rc = tpm2_key_decode(payload, options, &blob);
> > > -	if (rc) {
> > > -		/* old form */
> > > +	key = tpm2_key_decode(payload->blob, payload->blob_len);
> > > +	if (IS_ERR(key)) {
> > > +		/* Get the error code and reset the pointer to the key: */
> > > +		rc = PTR_ERR(key);
> > > +		key = NULL;
> > > +
> > > +		if (rc == -ENOMEM)
> > > +			return -ENOMEM;
> > > +
> > > +		/* A sanity check, as only -EBADMSG or -ENOMEM are expected: */
> > > +		if (rc != -EBADMSG)
> > > +			pr_err("tpm2_key_decode(): spurious error code %d\n", rc);
> >
> > tpm2_key_decode seems simple enough that it only returns key, -ENOMEM or
> > EBADMSG.
>
> So what is your suggestion here?

You can remove the check resuling in pr_err().

> The reasoning here is that asymmetric keys use -EBADMSG not only as
> error but also iterator, when probing which can load a specific key.