> On Nov 4, 2014, at 3:00 PM, Thomas Monjalon <thomas.monjalon at 6wind.com>
> wrote:
>
> 2014-11-03 13:02, Aaron Campbell:
>>> On Jul 8, 2014, at 5:28 AM, Simon Kuenzer <simon.kuenzer at neclab.eu>
>>> wrote:
>>>
>>> + else if (!strcmp(lgopts[option_index].name,
>>> OPT_MASTER_LCORE)) {
>>> + if (!coremask_ok) {
>>> + RTE_LOG(ERR, EAL, "please specify the
>>> master "
>>> + "lcore id after
>>> specifying "
>>> + "the coremask\n");
>>> + eal_usage(prgname);
>>> + return -1;
>>> + }
>>
>>
>> Hi Simon,
>>
>> I think that forcing a particular command line order is not that clean.
>> It might be better to remove the cfg->master_lcore setting from
>> eal_parse_coremask(), and defer the selection of the master lcore until
>> all of the command-line options have been parsed. If ?master-lcore was
>> specified, save the value and use that, otherwise
>> rte_get_next_lcore(-1, 0, 0) can return the first bit set in the coremask.
>
> It's not sufficient: eal_parse_master_lcore() requires cfg->lcore_role
> to be set. There is a real dependency between these 2 options.
> I'm going to submit a v2. Feel free to improve it with another patch.
I was nit-picking; although it might be nice if the new option is given, to
verify the specified lcore is in the coremask. I will ack v2 though and this
can be improved some other time.
-Aaron
