On 5/25/2018 1:44 AM, Zhang, Qi Z wrote:
>
>
>> -----Original Message-----
>> From: Thomas Monjalon [mailto:tho...@monjalon.net]
>> Sent: Friday, May 25, 2018 1:06 AM
>> To: Varghese, Vipin <vipin.vargh...@intel.com>
>> Cc: dev@dpdk.org; Zhang, Qi Z <qi.z.zh...@intel.com>; Kovacevic, Marko
>> <marko.kovace...@intel.com>; Jain, Deepak K <deepak.k.j...@intel.com>;
>> Yigit, Ferruh <ferruh.yi...@intel.com>
>> Subject: Re: [dpdk-dev] [PATCH v1] doc/eth: update document for functional
>> limitation
>>
>> 23/05/2018 12:05, Vipin Varghese:
>>> --- a/lib/librte_ethdev/rte_ethdev.h
>>> +++ b/lib/librte_ethdev/rte_ethdev.h
>>> @@ -1479,6 +1479,9 @@ int rte_eth_dev_attach(const char *devargs,
>> uint16_t *port_id);
>>> * This function must be called when the device is in the
>>> * closed state.
>>> *
>>> + * Note:
>>> + * - Currently supported for primary process only.
>>
>> I think it should be possible to attach a device in secondary process.
>> If it is a recent limitation, it should be in "known issues" of the release
>> notes.
>
> For PCI device, we can only attached a device that is already resource mapped
> in primary process (for example, a device not in secondary process' white
> list but in primary process's)
> And we should not detached a device in secondary process, that will mess
> primary process and cause it can't be attached again.
>
> For vdev, I think we still can attached/detach a new device which does not
> exist in primary process.
For vdev it was possible to attach a new device in secondary, but it seems
primary process checks has been added to virtual PMDs probe() function by [1]
which seems breaking this capability, can you please check it?
[1]
Fixes: ee27edbe0c10 ("drivers/net: share vdev data to secondary process")
>
>>
>>
>
