Subject: Re: [tpmdd-devel] tpm: read burstcount from TPM_STS in one 32-bit
 transaction
From: James Bottomley <jejb@linux.vnet.ibm.com>
To: Michal =?ISO-8859-1?Q?Such=E1nek?= <msuchanek@suse.de>,
        Christophe Ricard <christophe.ricard@gmail.com>,
        linux-kernel@vger.kernel.org, tpmdd-devel@lists.sourceforge.net,
        Jarkko Sakkinen <jarkko.sakkinen@linux.intel.com>,
        apronin@chromium.org
Date: Tue, 25 Jul 2017 10:36:11 -0700
In-Reply-To: <20170725150443.7cf8fc91@kitsune.suse.cz>
References: <20170725150443.7cf8fc91@kitsune.suse.cz>
Content-Type: text/plain; charset="UTF-8"
Mime-Version: 1.0
Content-Transfer-Encoding: 8bit
Message-Id: <1501004171.3689.25.camel@linux.vnet.ibm.com>
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 887
Lines: 20

On Tue, 2017-07-25 at 15:04 +0200, Michal Suchánek wrote:
> Hello,
> 
> in commit 9754d45e9970 ("tpm: read burstcount from TPM_STS in one
> 32-bit transaction") you change reading of two 8-bit values to one
> 32bit read. This is obviously wrong wrt endianess unless the
> underlying tpm_tis_read32 does endian conversion. 

Some of the bus read primitives do do endianness conversions.  The
problem is with the SPI attachment, which has unclear endianness.  A
standard PCI bus attachment uses ioread32() which automatically
transforms from a little endian bus to the cpu endianness, however SPI
is forced to transfer the bytes one at a time over the serial bus and
then transform.  The assumption seems to be that the TIS TPM is
replying in little endian format when SPI connected.

We can probably get the PPC people to confirm this, I believe they have
a SPI attached TPM.

James