Re: [Qemu-devel] [PATCH V9 5/5] Add a TPM Passthrough backend driver implementation

[Stefan Berger <stefanb@linux.vnet.ibm.com>]

On 09/26/2011 04:24 PM, Michael S. Tsirkin wrote:
> On Mon, Sep 26, 2011 at 04:12:19PM -0400, Stefan Berger wrote:
>
>>>> +    if (tpm_passthrough_test_tpmdev(tb->s.tpm_pt->tpm_fd)) {
>>>> +        fprintf(stderr,
>>>> +                "'%s' is not a TPM device.\n",
>>>> +                tb->s.tpm_pt->tpm_dev);
>>>> +        goto err_close_tpmdev;
>>> Is this a must? Is it common to have more than one
>>> tpm device available on a computer? Maybe there's
>>> a good default in case only one tpm exists there ...
>>>
>> Well, passing /dev/tty48 in the place of /dev/tpm0 ends up in a
>> disappointment. So I'd rather check what that device is and refuse
>> to start if it is found not to be a TPM.
> Sorry, I mean can path= be made optional in case there's a single
> /dev/tmpXXX? Is it even common to have any tpms except
> /dev/tpm0?
No, typically there is only one TPM in the machine and it's accessible 
using /dev/tpm0. SR-IOV TPMs afaik don't exist. So, yes, I am adding 
'/dev/tpm0' as the default now.
>
> All good answers shall likely go into the documentation :)
I'll write about it... Will likely post v10 tomorrow.

    Stefan