[pve-devel] [PATCH pve-manager] cpumodelselector : add EPYC

[Thomas Lamprecht]

On 9/3/18 1:58 PM, Dominik Csapak wrote:
> On 09/03/2018 01:42 PM, Thomas Lamprecht wrote:
>> On 9/3/18 12:35 PM, Dominik Csapak wrote:
>>> On 09/02/2018 07:52 AM, Alexandre Derumier wrote:
>>>> ---
>>>>    www/manager6/form/CPUModelSelector.js | 1 +
>>>>    1 file changed, 1 insertion(+)
>>>>
>>>> diff --git a/www/manager6/form/CPUModelSelector.js b/www/manager6/form/CPUModelSelector.js
>>>> index d3315513..9eb5b0e9 100644
>>>> --- a/www/manager6/form/CPUModelSelector.js
>>>> +++ b/www/manager6/form/CPUModelSelector.js
>>>> @@ -31,6 +31,7 @@ Ext.define('PVE.form.CPUModelSelector', {
>>>>        ['Opteron_G3', 'Opteron_G3'],
>>>>        ['Opteron_G4', 'Opteron_G4'],
>>>>        ['Opteron_G5', 'Opteron_G5'],
>>>> +    ['EPYC', 'EPYC'],
>>>>        ['host', 'host']
>>>>          ]
>>>>
>>>
>>> looks good, but i wanted to generate that component from qemu help output (qemu-system-x86_64 -cpu ?) for some time now
>>>
>>> i will look at how hard this would be and send that, otherwise
>>> we can include your patch
>>>
>>
>> As the list gets not often updated (new major QEMU release, or special
>> occasions like a shiny big new security hole in a CPU) I'm not sure if
>> that's worth it. If, I'd only generate one at build time, that way we also
>> still have control about what exactly gets shown, e.g., we do not show 'max'
>> or 'base' and values not included in $cpu_vendor_list from PVE::QemuServer
>> cannot be used anyway.
>> Those, approx., two CPU list updates a year could be managed manually just
>> fine, IMO.
>>
> 
> yeah, my idea was to include a script which generates the file on-demand, so no automatic build step or something like that
> 

that approach would be OK for me.
Let's just apply Alexandre's patch (as it's already there), and the next time
this is needed we go straight to your approach? If you already implemented it
we can naturally add it now :)

> but i'm also ok with manually updating it :)
>