[PATCH v3 5/9] domain_conf.c: XML parsing for VIR_DOMAIN_TPM_MODEL_SPAPR_PROXY

Daniel Henrique Barboza danielhb413 at gmail.com
Fri May 15 22:35:51 UTC 2020



On 5/14/20 10:07 PM, Daniel Henrique Barboza wrote:
> 
> 
> On 5/14/20 11:32 AM, Daniel Henrique Barboza wrote:
>>
>>
>> On 5/14/20 11:09 AM, Ján Tomko wrote:
>>> On a Wednesday in 2020, Daniel Henrique Barboza wrote:
>>>> Aside from trivial XML parsing/format changes, this patch adds
>>>> additional rules for TPM device support to better accomodate
> 
> [...]
> 
>>>
>>> Any reason why you store them separately?
>>>
>>> It seems they are treated the same in every place except when building
>>> QEMU command line. Switching to a def->tpms array would better reflect
>>> the XML. The Validate function would then check wheteher there's just
>>> one copy of each device type.


Just sent a v4 with this approach. I attempted to do the trick I mentioned previously
of having def->tpms[0] to always hold a non-proxy device, but in the end I decided it
wasn't worth it - I would make everyone reading qemu_extdevices.c to wonder "why is this
code passing def->tpms[0] instead of interacting the whole array".  It was too
messy for someone who doesn't know about the TPM Proxy history.


Thanks,

DHB




More information about the libvir-list mailing list