Re: [PATCH] tpm: fix a race condition tpm2_unseal_trusted()

[Jarkko Sakkinen]

On Wed, Jul 20, 2016 at 10:48:18AM -0600, Jason Gunthorpe wrote:
> On Wed, Jul 20, 2016 at 03:16:32AM +0300, Jarkko Sakkinen wrote:
> > Unseal and load operations should be done as an atomic unit. This
> > commit fixes the issue by moving TPM mutex handling to tpm_try_get_ops()
> > and tpm_put_ops(), which is probably more logical place for it anyway.
> 
> No..
> 
> 'get_ops' is to be used to hold a persisent kref to a single tpm. It
> cannot block other tpm access.
> 
> Eg a upper protocol might get_ops to for a long period to ensure it
> consistently talks to the same TPM in a multi-tpm system.
> 
> We need something else to solve whatever you are concerned with
> here..

The only use cases I see at the moment for it work this way:

1. Call tpm_try_get_ops.
2. Send a TPM command.
3. Call tpm_put_ops.

I did not find any other form of use. The only use is to make sure that
there are no transactions running before the ops are cleared. Or did I
overlook something perhaps?

Trusted key unseal operation with TPM2 is broken into two operations:

1. Load the given key blob.
2. Unseal the data.

Without locking and unlocking mutex only once there is a race condition.

/Jarkko