Hi, xiaoyun
> -----Original Message-----
> From: Li, Xiaoyun <xiaoyun...@intel.com>
> Sent: Friday, March 19, 2021 3:57 PM
> To: Guo, Jia <jia....@intel.com>; or...@nvidia.com; Zhang, Qi Z
> <qi.z.zh...@intel.com>; Xing, Beilei <beilei.x...@intel.com>; Wu, Jingjing
> <jingjing...@intel.com>
> Cc: Zhang, Yuying <yuying.zh...@intel.com>; Xu, Ting <ting...@intel.com>;
> dev@dpdk.org
> Subject: RE: [PATCH v1 4/4] net/iavf: support FDIR for IP fragment packet
>
> Hi
>
> > -----Original Message-----
> > From: Guo, Jia <jia....@intel.com>
> > Sent: Wednesday, March 17, 2021 11:12
> > To: or...@nvidia.com; Zhang, Qi Z <qi.z.zh...@intel.com>; Xing, Beilei
> > <beilei.x...@intel.com>; Li, Xiaoyun <xiaoyun...@intel.com>; Wu,
> > Jingjing <jingjing...@intel.com>
> > Cc: Zhang, Yuying <yuying.zh...@intel.com>; Xu, Ting
> > <ting...@intel.com>; dev@dpdk.org; Guo, Jia <jia....@intel.com>
> > Subject: [PATCH v1 4/4] net/iavf: support FDIR for IP fragment packet
> >
> > New FDIR parsing are added to handle the fragmented IPv4/IPv6 packet.
> >
> > Signed-off-by: Ting Xu <ting...@intel.com>
> > Signed-off-by: Jeff Guo <jia....@intel.com>
> > ---
> > drivers/net/iavf/iavf_fdir.c | 278 ++++++++++++++++++---------
> > drivers/net/iavf/iavf_generic_flow.h | 5 +
> > 2 files changed, 190 insertions(+), 93 deletions(-)
> >
> > diff --git a/drivers/net/iavf/iavf_fdir.c
> > b/drivers/net/iavf/iavf_fdir.c index
> > e3f3b5f22a..348d423081 100644
> > --- a/drivers/net/iavf/iavf_fdir.c
> > +++ b/drivers/net/iavf/iavf_fdir.c
> > @@ -34,7 +34,7 @@
> <snip>
> >
> > uint8_t ipv6_addr_mask[16] = {
> > 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, @@ -528,12
> > +534,6 @@ iavf_fdir_parse_pattern(__rte_unused struct iavf_adapter
> > +*ad,
> > };
> >
> > for (item = pattern; item->type != RTE_FLOW_ITEM_TYPE_END; item++) {
> > -if (item->last) {
>
> You can't directly remove the check. If users use "last" for items like
> RTE_FLOW_ITEM_TYPE_ETH. The driver doesn't support that.
>
Oh, that is like as you said. Specific enabling should be addressed.
> > -rte_flow_error_set(error, EINVAL,
> > -RTE_FLOW_ERROR_TYPE_ITEM, item,
> > -"Not support range");
> > -}
> > -
> > item_type = item->type;
> >
> <snip>
> > +if (ipv4_mask->hdr.version_ihl ||
> > + ipv4_mask->hdr.total_length ||
> > + ipv4_mask->hdr.hdr_checksum) {
> > +rte_flow_error_set(error, EINVAL,
> > +
> > RTE_FLOW_ERROR_TYPE_ITEM,
> > + item, "Invalid IPv4 mask.");
> > +return -rte_errno;
> > +}
> >
> > -rte_memcpy(hdr->buffer,
> > -&ipv4_spec->hdr,
> > -sizeof(ipv4_spec->hdr));
>
> The ipv4_mask is checked. But what about ipv4_last?
> What if users set a rule which includes "last" for hdr.version_ihl? The code
> doesn't process that and not err return.
> You should block all situations in ipv4_last except for ipv4_last-
> >hdr.fragment_offset.
>
Ok.
> <snip>
> > +
> > +if (ipv4_mask->hdr.packet_id == UINT16_MAX ||
> > + ipv4_mask->hdr.fragment_offset == UINT16_MAX) {
>
> And I don't understand your logic here.
> Only if ipv4_mask->hdr.fragment_offset and ipv4_mask->hdr.packet_id Are
> UINT16_MAX, you process this case. But what about other cases? Shouldn't
> those cases return err like not supported?
> And the mask for fragment_offset shouldn't be 0xffff (UINT16_MAX), it
> should be 0x3fff (RTE_IPV4_HDR_OFFSET_MASK |
> RTE_IPV4_HDR_MF_FLAG). Because only the last 14 bits matters. The other 2
> bits are reserved bit and DF bit, it doesn't matter if it's fragment packets
> or
> not.
>
Em, there are definitely something not cover. The negotiate should be specific
at device but not at avf.
> > +if (ipv4_last &&
> > + ipv4_spec->hdr.packet_id == 0 &&
> > + ipv4_last->hdr.packet_id == 0xffff)
>
> And I don't understand this part too. I thought it's a fragment rule and non-
> fragment rule. Why is this related to packet_id? And what about
> fragment_offset spec and last?
> In ovs usercase, the rule for fragment packets is like flow create 0 ingress
> pattern eth / ipv4 src is xxx dst is xxx fragment_offset spec 0x8
> fragment_offset last 0x2000 fragment_offset mask 0x3fff / end actions
> queue index 1 / end
>
> And the rule for non-fragment rule is like:
> flow create 0 ingress pattern eth / ipv4 src is xxx dst is xxx fragment_offset
> spec 0 fragment_offset mask 0x3fff / end actions queue index 1 / end or
> flow create 0 ingress pattern eth / ipv4 src is xxx dst is xxx fragment_offset
> mask 0x3fff / end actions queue index 1 / end
>
> How can your codes here make sure the rules behave correctly?
>
What i want is that use a dummy input set for the fragment packet. At lease, we
have handle 5tuple/specific id/any packet id for Fragment packet, and for
one-fragment packet, seems that no need to
use fragment_offset to identify it seem default is for ip other.
> > +spec_all_pid = true;
> > +
> > +/* All IPv4 fragment packet has the same
> > + * ethertype, if the spec is for all invalid
> > + * packet id, set the ethertype into input set.
> > + */
> > +input_set |= spec_all_pid ?
> > +IAVF_INSET_ETHERTYPE :
> > +IAVF_INSET_IPV4_ID;
> > +
> > +if (spec_all_pid)
> > +
> > VIRTCHNL_ADD_PROTO_HDR_FIELD_BIT(hdr1,
> > +ETH, ETHERTYPE);
> > +else
> > +
> > VIRTCHNL_ADD_PROTO_HDR_FIELD_BIT(hdr,
> > +IPV4, PKID);
> > }
> >
> > +rte_memcpy(hdr->buffer, &ipv4_spec->hdr,
> > + sizeof(ipv4_spec->hdr));
> > +
> > hdrs.count = ++layer;
> > break;
> >
>
> I'm not familiar with IPv6 fragment rule. So not review that part.
>
> <snip>
> > --
> > 2.20.1
>
