[PATCH v2 4/6] KEYS: trusted: Move tpm2_key_decode() to the TPM driver

[Jarkko Sakkinen]

On Tue May 21, 2024 at 9:18 PM EEST, James Bottomley wrote:
> On Tue, 2024-05-21 at 06:16 +0300, Jarkko Sakkinen wrote:
> [...]
> > diff --git a/include/crypto/tpm2_key.h b/include/crypto/tpm2_key.h
> > new file mode 100644
> > index 000000000000..acf41b2e0c92
> > --- /dev/null
> > +++ b/include/crypto/tpm2_key.h
> > @@ -0,0 +1,33 @@
> > +/* SPDX-License-Identifier: GPL-2.0-only */
> > +#ifndef __LINUX_TPM2_KEY_H__
> > +#define __LINUX_TPM2_KEY_H__
> > +
> > +#include <linux/slab.h>
> > +
> > +/*
> > + * TPM2 ASN.1 key
> > + */
> > +struct tpm2_key {
> > +       u32 parent;
> > +       const u8 *blob;
> > +       u32 blob_len;
> > +       const u8 *pub;
> > +       u32 pub_len;
> > +       const u8 *priv;
> > +       u32 priv_len;
> > +};
> > +
> > +int tpm2_key_decode(const u8 *src, u32 src_len, struct tpm2_key
> > *key,
> > +                   u32 max_key_len);
>
> I don't think this is a good idea.  Trusted keys already have a pre-
> defined max payload size (MAX_BLOB_SIZE in include/keys/trusted-type.h)
> and I've already had to increase this several times because once you
> get policy attached to a key, it can get pretty big (over a page). 
> Exactly the same thing will happen to asymmetric keys as well, so it
> does make sense that they share the same maximum (probably in a more
> generic header, though).

ECDSA and RSA have different space requirements. With that solution you
actually max out space requirements given same cap for everything.

Even tpm2_key_ecdsa should use a different value than tpm2_key_rsa to
save memory.

> Since the code already right sizes the allocation and all we check with
> this is whether it's over a pre-defined maximum, it's way easier if
> that maximum is defined in a header rather than passed in in several
> places making increasing the maximum really hard because you have to
> chase all the threading.

You don't save a single byte of memory with any constant that dictates
the size requirements for multiple modules in two disjoint subsystems.

You are maximizing the use of memory.

> James

BR, Jarkko