DPDK patches and discussions
 help / color / mirror / Atom feed
From: Jack Min <jackmin@nvidia.com>
To: Ferruh Yigit <ferruh.yigit@intel.com>,
	Xiaoyu Min <jackmin@mellanox.com>,
	 Jingjing Wu <jingjing.wu@intel.com>,
	Beilei Xing <beilei.xing@intel.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>,
	NBU-Contact-Thomas Monjalon <thomas@monjalon.net>,
	Andrew Rybchenko <arybchenko@solarflare.com>,
	Ori Kam <orika@nvidia.com>, Dekel Peled <dekelp@nvidia.com>
Subject: Re: [dpdk-dev] [PATCH 4/5] net/iavf: fix protocol size for virtchnl copy
Date: Sun, 22 Nov 2020 13:28:23 +0000
Message-ID: <MN2PR12MB33441EF29DD89174A29F7BC4C1FD0@MN2PR12MB3344.namprd12.prod.outlook.com> (raw)
In-Reply-To: <8990f612-b3b1-bdbc-8da3-9a0e45d5c3bc@intel.com>

> -----Original Message-----
> From: Ferruh Yigit <ferruh.yigit@intel.com>
> Sent: Tuesday, November 17, 2020 00:23
> To: Xiaoyu Min <jackmin@mellanox.com>; Jingjing Wu <jingjing.wu@intel.com>;
> Beilei Xing <beilei.xing@intel.com>
> Cc: dev@dpdk.org; Jack Min <jackmin@nvidia.com>; NBU-Contact-Thomas
> Monjalon <thomas@monjalon.net>; Andrew Rybchenko
> <arybchenko@solarflare.com>; Ori Kam <orika@nvidia.com>; Dekel Peled
> <dekelp@nvidia.com>
> Subject: Re: [dpdk-dev] [PATCH 4/5] net/iavf: fix protocol size for virtchnl copy
> 
> On 11/16/2020 7:55 AM, Xiaoyu Min wrote:
> > From: Xiaoyu Min <jackmin@nvidia.com>
> >
> > The rte_flow_item_vlan items are refined.
> > The structs do not exactly represent the packet bits captured on the
> > wire anymore so should only copy real header instead of the whole struct.
> >
> > Replace the rte_flow_item_* with the existing corresponding rte_*_hdr.
> >
> > Fixes: 09315fc83861 ("ethdev: add VLAN attributes to ethernet and VLAN
> items")
> >
> > Signed-off-by: Xiaoyu Min <jackmin@nvidia.com>
> > ---
> >   drivers/net/iavf/iavf_fdir.c | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/net/iavf/iavf_fdir.c b/drivers/net/iavf/iavf_fdir.c
> > index d683a468c1..7054bde0b9 100644
> > --- a/drivers/net/iavf/iavf_fdir.c
> > +++ b/drivers/net/iavf/iavf_fdir.c
> > @@ -541,7 +541,7 @@ iavf_fdir_parse_pattern(__rte_unused struct
> iavf_adapter *ad,
> >   				VIRTCHNL_ADD_PROTO_HDR_FIELD_BIT(hdr,
> ETH, ETHERTYPE);
> >
> >   				rte_memcpy(hdr->buffer,
> > -					eth_spec, sizeof(*eth_spec));
> > +					eth_spec, sizeof(struct rte_ether_hdr));
> 
> This requires 'struct rte_flow_item_eth' should have 'struct rte_ether_hdr' as
> first element, and I suspect this usage exists in a few more locations, but I
> wonder if this assumption is real and documented in somewhere?
> I am not just talking about 'struct rte_flow_item_eth', but for all
> 'rte_flow_item_*'...
> 
I think this is not documented and this assumption is not real.
I've created one ticket on Bugzilla (https://bugs.dpdk.org/show_bug.cgi?id=581) to track.

> 
> 
> btw, while checking for the 'struct rte_flow_item_eth', pahole shows it is using
> 20 bytes, and I suspect this is not the intention with the reserved field:
> 
> struct rte_flow_item_eth {
> 	struct rte_ether_addr      dst;                  /*     0     6 */
> 	struct rte_ether_addr      src;                  /*     6     6 */
> 	uint16_t                   type;                 /*    12     2 */
> 
> 	/* Bitfield combined with previous fields */
> 
> 	uint32_t                   has_vlan:1;           /*    12:15  4 */
> 
> 	/* XXX 31 bits hole, try to pack */
> 
> 	uint32_t                   reserved:31;          /*    16: 1  4 */
> 
> 	/* size: 20, cachelines: 1, members: 5 */
> 	/* bit holes: 1, sum bit holes: 31 bits */
> 	/* bit_padding: 1 bits */
> 	/* last cacheline: 20 bytes */
> };
> 
> 'has_vlan' seems combined with previous field to make together 32 bits. So the
> 'reserved' field is occupying a new 32 bits all by itself.
> 
> What about changing the struct as following, while we can change the ABI:
> struct rte_flow_item_eth {
> 	struct rte_ether_addr      dst;                  /*     0     6 */
> 	struct rte_ether_addr      src;                  /*     6     6 */
> 	uint16_t                   type;                 /*    12     2 */
> 	uint16_t                   has_vlan:1;           /*    14:15  2 */
> 	uint16_t                   reserved:15;          /*    14: 0  2 */
> 
> 	/* size: 16, cachelines: 1, members: 5 */
> 	/* last cacheline: 16 bytes */
> };
> 

Well we probably need to discuss this in next release. 
It's too late to change this API at this moment.

-Jack


  reply	other threads:[~2020-11-22 13:28 UTC|newest]

Thread overview: 30+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-11-16  7:55 [dpdk-dev] [PATCH 0/5] fix protocol size calculation Xiaoyu Min
2020-11-16  7:55 ` [dpdk-dev] [PATCH 1/5] net/mlx5: fix protocol size for raw encap judgement Xiaoyu Min
2020-11-17 13:25   ` Matan Azrad
2020-11-22 14:21     ` Thomas Monjalon
2020-11-22 15:32   ` Thomas Monjalon
2020-11-22 16:04     ` Matan Azrad
2020-11-23  7:54       ` Ori Kam
2020-11-23  8:12         ` Thomas Monjalon
2020-11-16  7:55 ` [dpdk-dev] [PATCH 2/5] app/flow-perf: fix protocol size for raw encap Xiaoyu Min
2020-11-16  7:55 ` [dpdk-dev] [PATCH 3/5] net/bnxt: fix protocol size for VXLAN encap copy Xiaoyu Min
2020-11-16 16:12   ` Ferruh Yigit
2020-11-18  0:34     ` Ajit Khaparde
2020-11-18  6:37       ` Andrew Rybchenko
2020-11-18 17:44         ` Ajit Khaparde
2020-11-16  7:55 ` [dpdk-dev] [PATCH 4/5] net/iavf: fix protocol size for virtchnl copy Xiaoyu Min
2020-11-16 16:23   ` Ferruh Yigit
2020-11-22 13:28     ` Jack Min [this message]
2020-11-22 14:15       ` Thomas Monjalon
2020-11-23 10:02         ` Ferruh Yigit
2020-11-23 10:49   ` Ferruh Yigit
2020-11-24 21:58     ` Thomas Monjalon
2020-11-27  1:17   ` Xing, Beilei
2020-11-16  7:55 ` [dpdk-dev] [PATCH 5/5] net/softnic: update headers size calculation Xiaoyu Min
2020-11-16 16:27   ` Ferruh Yigit
2020-11-16 19:09     ` Dekel Peled
2020-11-17 10:30       ` Ferruh Yigit
2020-11-17 12:41         ` Dekel Peled
2020-11-17 15:43           ` Singh, Jasvinder
2020-11-18  2:23   ` Zhou, JunX W
2020-11-22 16:11 ` [dpdk-dev] [PATCH 0/5] fix protocol " Thomas Monjalon

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=MN2PR12MB33441EF29DD89174A29F7BC4C1FD0@MN2PR12MB3344.namprd12.prod.outlook.com \
    --to=jackmin@nvidia.com \
    --cc=arybchenko@solarflare.com \
    --cc=beilei.xing@intel.com \
    --cc=dekelp@nvidia.com \
    --cc=dev@dpdk.org \
    --cc=ferruh.yigit@intel.com \
    --cc=jackmin@mellanox.com \
    --cc=jingjing.wu@intel.com \
    --cc=orika@nvidia.com \
    --cc=thomas@monjalon.net \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

DPDK patches and discussions

This inbox may be cloned and mirrored by anyone:

	git clone --mirror https://inbox.dpdk.org/dev/0 dev/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 dev dev/ https://inbox.dpdk.org/dev \
		dev@dpdk.org
	public-inbox-index dev

Example config snippet for mirrors.
Newsgroup available over NNTP:
	nntp://inbox.dpdk.org/inbox.dpdk.dev


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git