On 10/18/2019 10:23 AM, Olivier Matz wrote: > On Fri, Oct 18, 2019 at 10:47:37AM +0200, David Marchand wrote: >> On Fri, Oct 18, 2019 at 10:43 AM Olivier Matz <olivier.m...@6wind.com> wrote: >>> >>> Hi David, >>> >>> 2lOn Fri, Sep 27, 2019 at 01:58:02PM +0200, David Marchand wrote: >>>> OVS currently maintains a copy of those headers with the right endianness >>>> annotations so that sparse checks can pass. >>>> >>>> We introduced rte_beXX_t for better readibility in v17.08. >>>> Let's make use of them, OVS then only needs to override those rte_beXX_t >>>> types by exposing a tweaked rte_byteorder.h header. >>>> >>>> Other existing dpdk users won't be affected since rte_beXX_t types are >>>> mapped to uintXX_t types. >>>> >>>> Signed-off-by: David Marchand <david.march...@redhat.com> >>>> --- >>>> lib/librte_net/rte_icmp.h | 12 +++++++----- >>>> lib/librte_net/rte_ip.h | 28 ++++++++++++++-------------- >>>> lib/librte_net/rte_sctp.h | 10 ++++++---- >>>> lib/librte_net/rte_tcp.h | 20 +++++++++++--------- >>>> lib/librte_net/rte_udp.h | 10 ++++++---- >>>> 5 files changed, 44 insertions(+), 36 deletions(-) >>> >>> Is there a reason why you didn't change arp, gre, ... ? >> >> Let me look at the other headers, I had focused on what OVS used. >> >> Is this current patch ok? > > Yes, it looks good to me. >
Converting to an explicit ack J Acked-by: Olivier Matz <olivier.m...@6wind.com>