On Jan 26, 2017, at 5:58 PM, Ashley Lai [off-list ref] wrote:
=20
Adding Vicky from IBM.
=20
=20
On 01/26/2017 04:05 PM, Jason Gunthorpe wrote:

quoted

On Thu, Jan 26, 2017 at 09:22:48PM +0100, Michal Such??nek wrote:
=20

quoted

This is repeated a few times in the driver so I added memset to =

quiet

quoted

quoted

gcc and make behavior deterministic in case the unused fields get =

some

quoted

quoted

meaning in the future.

Yep, reserved certainly needs to be zeroed.. Can you send a patch?
memset is overkill...
=20

quoted

However, in tpm_ibmvtpm_send the structure is initialized as
=20
	struct ibmvtpm_crq crq;
        __be64 *word =3D (__be64 *)&crq;
...
        crq.valid =3D (u8)IBMVTPM_VALID_CMD;
        crq.msg =3D (u8)VTPM_TPM_COMMAND;
        crq.len =3D cpu_to_be16(count);
        crq.data =3D cpu_to_be32(ibmvtpm->rtce_dma_handle);
=20
and submitted with
=20
	rc =3D ibmvtpm_send_crq(ibmvtpm->vdev, be64_to_cpu(word[0]),
                              be64_to_cpu(word[1]));
meaning it is swapped twice.

No idea, Nayna may know.
=20
My guess is that '__be64 *word' should be 'u64 *word'...
=20
Jason

=20

I don=92t think we want =91word' to be changed back to be of type =91u64=92=
.   Please see commit 62dfd912ab3b5405b6fe72d0135c37e9648071f1


Vicky=