Re: [PATCH v2] KEYS: trusted: Fix memory leak in tpm2_key_encode()

[Jarkko Sakkinen <jarkko@kernel.org>]

On Thu, Jun 09, 2022 at 08:21:20AM +0300, Jarkko Sakkinen wrote:
> On Wed, Jun 08, 2022 at 09:17:32PM +0800, Jianglei Nie wrote:
> > tpm2_key_encode() allocates a memory chunk from scratch with kmalloc(),
> > but it is never freed, which leads to a memory leak. Free the memory
> > chunk with kfree() in the return path.
> > 
> > Signed-off-by: Jianglei Nie <niejianglei2021@163.com>
> > ---
> 
> You should write down the changelog ere. No idea what changed
> from the previous version.
> 
> >  security/keys/trusted-keys/trusted_tpm2.c | 23 +++++++++++++++++------
> >  1 file changed, 17 insertions(+), 6 deletions(-)
> > 
> > diff --git a/security/keys/trusted-keys/trusted_tpm2.c b/security/keys/trusted-keys/trusted_tpm2.c
> > index 0165da386289..8b7ab22950d1 100644
> > --- a/security/keys/trusted-keys/trusted_tpm2.c
> > +++ b/security/keys/trusted-keys/trusted_tpm2.c
> > @@ -32,6 +32,7 @@ static int tpm2_key_encode(struct trusted_key_payload *payload,
> >  			   struct trusted_key_options *options,
> >  			   u8 *src, u32 len)
> >  {
> > +	int err;
> 
> Declare as the last local variable (reverse christmas tree order).
> 
> Also, I'd use "int ret" since in other functions that is used.
> 
> >  	const int SCRATCH_SIZE = PAGE_SIZE;
> >  	u8 *scratch = kmalloc(SCRATCH_SIZE, GFP_KERNEL);
> >  	u8 *work = scratch, *work1;
> > @@ -57,8 +58,10 @@ static int tpm2_key_encode(struct trusted_key_payload *payload,
> >  		unsigned char bool[3], *w = bool;
> >  		/* tag 0 is emptyAuth */
> >  		w = asn1_encode_boolean(w, w + sizeof(bool), true);
> > -		if (WARN(IS_ERR(w), "BUG: Boolean failed to encode"))
> > -			return PTR_ERR(w);
> > +		if (WARN(IS_ERR(w), "BUG: Boolean failed to encode")) {
> > +			err = PTR_ERR(w);
> > +			goto out;
> > +		}
> >  		work = asn1_encode_tag(work, end_work, 0, bool, w - bool);
> >  	}
> >  
> > @@ -69,8 +72,10 @@ static int tpm2_key_encode(struct trusted_key_payload *payload,
> >  	 * trigger, so if it does there's something nefarious going on
> >  	 */
> >  	if (WARN(work - scratch + pub_len + priv_len + 14 > SCRATCH_SIZE,
> > -		 "BUG: scratch buffer is too small"))
> > -		return -EINVAL;
> > +		 "BUG: scratch buffer is too small")) {
> > +		err = -EINVAL;
> > +		goto out;
> > +	}
> >  
> >  	work = asn1_encode_integer(work, end_work, options->keyhandle);
> >  	work = asn1_encode_octet_string(work, end_work, pub, pub_len);
> > @@ -79,10 +84,16 @@ static int tpm2_key_encode(struct trusted_key_payload *payload,
> >  	work1 = payload->blob;
> >  	work1 = asn1_encode_sequence(work1, work1 + sizeof(payload->blob),
> >  				     scratch, work - scratch);
> > -	if (WARN(IS_ERR(work1), "BUG: ASN.1 encoder failed"))
> > -		return PTR_ERR(work1);
> > +	if (WARN(IS_ERR(work1), "BUG: ASN.1 encoder failed")) {
> > +		err = -EINVAL;
> > +		goto out;
> 
> Why you are changing the return value from PTR_ERR(work1
> to -EINVAL?
> 
> > +	}
> > +	kfree(scratch);
> >  
> >  	return work1 - payload->blob;
> > +
> > +out:
> 
> Nit:
> 
> err:
> 
> It's only used for the error path.
> 
> > +	return err;
> >  }
> >  
> >  struct tpm2_key_context {
> > -- 
> > 2.25.1
> > 
> 
> BR, Jarkko

And the patch is also missing fixes tag:

Fixes: f2219745250f ("security: keys: trusted: use ASN.1 TPM2 key format for the blobs")

BR, Jarkko