On Wed, Jun 17, 2020 at 09:38:47PM +0000, Alexander Lobakin wrote:
> Hi Michal,
>
> On Thursday, 18 June 2020, 0:18, Michal Kubecek <mkube...@suse.cz> wrote:
>
> > On Wed, Jun 17, 2020 at 08:42:47PM +0000, Alexander Lobakin wrote:
> >
> > > Commit 3b33583265ed ("net: Add fraglist GRO/GSO feature flags") missed
> > > an entry for NETIF_F_GSO_FRAGLIST in netdev_features_strings array. As
> > > a result, fraglist GSO feature is not shown in 'ethtool -k' output and
> > > can't be toggled on/off.
> > > The fix is trivial.
> > >
> > > Fixes: 3b33583265ed ("net: Add fraglist GRO/GSO feature flags")
> > > Signed-off-by: Alexander Lobakin aloba...@pm.me
> > >
> > > ----------------------------------------------------------------------------------------------------------------
> > >
> > > net/ethtool/common.c | 1 +
> > > 1 file changed, 1 insertion(+)
> > > diff --git a/net/ethtool/common.c b/net/ethtool/common.c
> > > index 423e640e3876..47f63526818e 100644
> > > --- a/net/ethtool/common.c
> > > +++ b/net/ethtool/common.c
> > > @@ -43,6 +43,7 @@ const char
> > > netdev_features_strings[NETDEV_FEATURE_COUNT][ETH_GSTRING_LEN] = {
> > > [NETIF_F_GSO_SCTP_BIT] = "tx-sctp-segmentation",
> > > [NETIF_F_GSO_ESP_BIT] = "tx-esp-segmentation",
> > > [NETIF_F_GSO_UDP_L4_BIT] = "tx-udp-segmentation",
> > >
> > > - [NETIF_F_GSO_FRAGLIST_BIT] = "tx-gso-list",
> > > [NETIF_F_FCOE_CRC_BIT] = "tx-checksum-fcoe-crc",
> > > [NETIF_F_SCTP_CRC_BIT] = "tx-checksum-sctp",
> > >
> >
> > Reviewed-by: Michal kubecekmkube...@suse.cz
>
> Thanks!
>
> > AFAICS the name for NETIF_F_GSO_TUNNEL_REMCSUM_BIT is also missing but
> > IMHO it will be better to fix that by a separate patch with its own
> > Fixes tag.
>
> Oh, nice catch! I'll make a separate for this one.
> I also wanted to add any sort of static_assert() / BUILD_BUG_ON() to
> prevent such misses, but don't see any easy pattern to check for now,
> as netdev_features_strings[] is always NETDEV_FEATURE_COUNT-sized.
The key problem is that unlike e.g. link modes, new netdev features are
not always added at the end. So even if we changed
netdev_features_strings[] array to have size determined automatically
from initializers, adding new netdev feature somewhere in the middle
(e.g. GSO one) without updating netdev_features_strings[] would not be
caught as simple array size check would not notice the hole (actually,
there are also two intended holes in the array).
Michal
