nfp: bring back support for different ring counts
authorJakub Kicinski <jakub.kicinski@netronome.com>
Mon, 31 Oct 2016 20:43:22 +0000 (20:43 +0000)
committerDavid S. Miller <davem@davemloft.net>
Tue, 1 Nov 2016 15:05:01 +0000 (11:05 -0400)
commitcbeaf7aa733a134721cdcda44688e53a6380cd31
tree79e2fbf8d604d7213029666cd7f84d46f7ab7eaa
parentb33ae997a362d64a1178be17804c2c3899770a4b
nfp: bring back support for different ring counts

We used to always allocate the same number of TX and RX rings
so the support for having r_vectors without one of the rings
was dropped.  That makes us, however, unnecessarily limited
to 8 TX rings (8 is the Linux RSS default) most of the time.
Also we are about to add channel count configuration via
ethtool, so bring that support back.  TX rings can now default
to num_online_cpus() and RX rings to netif_get_num_default_rss_queues().

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/ethernet/netronome/nfp/nfp_net_common.c