On 12/12/2019 12:04 PM, Alfredo Cardigliano wrote:
> 
> 
>> On 10 Dec 2019, at 19:50, Ferruh Yigit <ferruh.yi...@intel.com> wrote:
>>
>> On 12/9/2019 9:46 PM, Alfredo Cardigliano wrote:
>>> Register the Pensando ionic PMD (net_ionic) and define initial probe
>>> and remove callbacks with adapter initialization.
>>>
>>> Signed-off-by: Alfredo Cardigliano <cardigli...@ntop.org>
>>> Reviewed-by: Shannon Nelson <snel...@pensando.io>
>>
>> <...>
>>
>>> +/*
>>> + * There is no room in struct rte_pci_driver to keep a reference
>>> + * to the adapter, using a static list for the time being.
>>> + */
>>> +static LIST_HEAD(ionic_pci_adapters_list, ionic_adapter) 
>>> ionic_pci_adapters =
>>> +           LIST_HEAD_INITIALIZER(ionic_pci_adapters);
>>> +static rte_spinlock_t ionic_pci_adapters_lock = RTE_SPINLOCK_INITIALIZER;
>>
>> As described in previous version of the patchset, it is possible to remove 
>> the
>> PMD maintained list of adapters, is there any benefit to keep it?
> 
> I will try to implement the approach you recommended. If that looks messy, 
> is it a problem to keep this static list?
> 

No problem, that looks reasonable. I believe it will work without being messy,
please check octeontx PMD [1], it does something similar, but if it turns out to
be messy please feel free to keep the static list.

[1]
http://lxr.dpdk.org/dpdk/v19.11/source/drivers/net/octeontx/octeontx_ethdev.c#L1146

Reply via email to