From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mga03.intel.com (mga03.intel.com [134.134.136.65]) by dpdk.org (Postfix) with ESMTP id 87C00DE0 for ; Mon, 8 Dec 2014 11:55:28 +0100 (CET) Received: from orsmga003.jf.intel.com ([10.7.209.27]) by orsmga103.jf.intel.com with ESMTP; 08 Dec 2014 02:53:36 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.04,691,1406617200"; d="scan'208";a="495364821" Received: from irsmsx104.ger.corp.intel.com ([163.33.3.159]) by orsmga003.jf.intel.com with ESMTP; 08 Dec 2014 02:51:48 -0800 Received: from irsmsx105.ger.corp.intel.com ([169.254.7.144]) by IRSMSX104.ger.corp.intel.com ([169.254.5.209]) with mapi id 14.03.0195.001; Mon, 8 Dec 2014 10:55:25 +0000 From: "Ananyev, Konstantin" To: "Zhang, Helin" , "dev@dpdk.org" Thread-Topic: [PATCH v2 2/2] mbuf: assign valid bit values for some RX and TX flags Thread-Index: AQHQEC1U9lXpd1X4/U+Iw0lE/Pu4KJyAy3bQgADuHlCAAAdHMIADxstA Date: Mon, 8 Dec 2014 10:55:25 +0000 Message-ID: <2601191342CEEE43887BDE71AB977258213BDAC4@IRSMSX105.ger.corp.intel.com> References: <1416982032-28519-1-git-send-email-helin.zhang@intel.com> <1417743988-15604-1-git-send-email-helin.zhang@intel.com> <1417743988-15604-3-git-send-email-helin.zhang@intel.com> <2601191342CEEE43887BDE71AB977258213BCE0C@IRSMSX105.ger.corp.intel.com> In-Reply-To: Accept-Language: en-IE, en-US Content-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [163.33.239.180] Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: quoted-printable MIME-Version: 1.0 Subject: Re: [dpdk-dev] [PATCH v2 2/2] mbuf: assign valid bit values for some RX and TX flags X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 08 Dec 2014 10:55:29 -0000 > -----Original Message----- > From: Zhang, Helin > Sent: Saturday, December 06, 2014 1:08 AM > To: Ananyev, Konstantin; dev@dpdk.org > Cc: Cao, Waterman; Cao, Min; Wu, Jingjing; Liu, Jijiang; olivier.matz@6wi= nd.com > Subject: RE: [PATCH v2 2/2] mbuf: assign valid bit values for some RX and= TX flags >=20 >=20 >=20 > > -----Original Message----- > > From: Zhang, Helin > > Sent: Saturday, December 6, 2014 8:40 AM > > To: Ananyev, Konstantin; dev@dpdk.org > > Cc: Cao, Waterman; Cao, Min; Wu, Jingjing; Liu, Jijiang; > > olivier.matz@6wind.com > > Subject: RE: [PATCH v2 2/2] mbuf: assign valid bit values for some RX a= nd TX > > flags > > > > OK. I will send out another patch according to your comments. Thanks a = lot! > > > > Regards, > > Helin > > > > > -----Original Message----- > > > From: Ananyev, Konstantin > > > Sent: Friday, December 5, 2014 6:50 PM > > > To: Zhang, Helin; dev@dpdk.org > > > Cc: Cao, Waterman; Cao, Min; Wu, Jingjing; Liu, Jijiang; > > > olivier.matz@6wind.com > > > Subject: RE: [PATCH v2 2/2] mbuf: assign valid bit values for some RX > > > and TX flags > > > > > > Hi Helin, > > > > > > > -----Original Message----- > > > > From: Zhang, Helin > > > > Sent: Friday, December 05, 2014 1:46 AM > > > > To: dev@dpdk.org > > > > Cc: Cao, Waterman; Cao, Min; Wu, Jingjing; Liu, Jijiang; Ananyev, > > > > Konstantin; olivier.matz@6wind.com; Zhang, Helin > > > > Subject: [PATCH v2 2/2] mbuf: assign valid bit values for some RX > > > > and TX flags > > > > > > > > Before redefining mbuf structure, there was lack of free bits in > > > > 'ol_flags' (32 bits in total) for new RX or TX flags. So it tried t= o > > > > reuse existant bits as most as possible, or even assigning 0 to som= e > > > > of bit flags. After new mbuf structure defined, there are quite a > > > > lot of free bits. So those newly added bit flags should be assigned > > > > with correct and valid bit values, and getting their names should b= e > > > > enabled as well. Note that 'RECIP' should be removed, as nowhere > > > > will use it. > > > > > > > > Signed-off-by: Helin Zhang > > > > --- > > > > lib/librte_mbuf/rte_mbuf.c | 9 ++++----- > > > > lib/librte_mbuf/rte_mbuf.h > > > > | 19 +++++++++---------- > > > > 2 files changed, 13 insertions(+), 15 deletions(-) > > > > > > > > v2 changes: > > > > * Removed error flag of 'ECIPE' processing only in mbuf. All other = error flags > > > > were added back. > > > > * Assigned error flags with correct and valid values, as their prev= ious values > > > > were invalid. > > > > * Enabled getting all error flag names. > > > > > > > > diff --git a/lib/librte_mbuf/rte_mbuf.c b/lib/librte_mbuf/rte_mbuf.= c > > > > index 87c2963..3ce7c8d 100644 > > > > --- a/lib/librte_mbuf/rte_mbuf.c > > > > +++ b/lib/librte_mbuf/rte_mbuf.c > > > > @@ -210,11 +210,10 @@ const char *rte_get_rx_ol_flag_name(uint64_t > > > mask) > > > > case PKT_RX_FDIR: return "PKT_RX_FDIR"; > > > > case PKT_RX_L4_CKSUM_BAD: return "PKT_RX_L4_CKSUM_BAD"; > > > > case PKT_RX_IP_CKSUM_BAD: return "PKT_RX_IP_CKSUM_BAD"; > > > > - /* case PKT_RX_EIP_CKSUM_BAD: return "PKT_RX_EIP_CKSUM_BAD"; */ > > > > - /* case PKT_RX_OVERSIZE: return "PKT_RX_OVERSIZE"; */ > > > > - /* case PKT_RX_HBUF_OVERFLOW: return "PKT_RX_HBUF_OVERFLOW"; > > > */ > > > > - /* case PKT_RX_RECIP_ERR: return "PKT_RX_RECIP_ERR"; */ > > > > - /* case PKT_RX_MAC_ERR: return "PKT_RX_MAC_ERR"; */ > > > > + case PKT_RX_EIP_CKSUM_BAD: return "PKT_RX_EIP_CKSUM_BAD"; > > > > + case PKT_RX_OVERSIZE: return "PKT_RX_OVERSIZE"; > > > > + case PKT_RX_HBUF_OVERFLOW: return > > "PKT_RX_HBUF_OVERFLOW"; > > > > + case PKT_RX_MAC_ERR: return "PKT_RX_MAC_ERR"; > > > > case PKT_RX_IPV4_HDR: return "PKT_RX_IPV4_HDR"; > > > > case PKT_RX_IPV4_HDR_EXT: return "PKT_RX_IPV4_HDR_EXT"; > > > > case PKT_RX_IPV6_HDR: return "PKT_RX_IPV6_HDR"; diff --git > > > > a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h index > > > > 2e5fce5..c9591c0 100644 > > > > --- a/lib/librte_mbuf/rte_mbuf.h > > > > +++ b/lib/librte_mbuf/rte_mbuf.h > > > > @@ -84,11 +84,6 @@ extern "C" { > > > > #define PKT_RX_FDIR (1ULL << 2) /**< RX packet with FDIR > > > match indicate. */ > > > > #define PKT_RX_L4_CKSUM_BAD (1ULL << 3) /**< L4 cksum of RX pkt. > > > is > > > > not OK. */ #define PKT_RX_IP_CKSUM_BAD (1ULL << 4) /**< IP cksum > > > of > > > > RX pkt. is not OK. */ -#define PKT_RX_EIP_CKSUM_BAD (0ULL << 0) > > > > /**< > > > External IP header checksum error. */ > > > > -#define PKT_RX_OVERSIZE (0ULL << 0) /**< Num of desc of an R= X > > > pkt oversize. */ > > > > -#define PKT_RX_HBUF_OVERFLOW (0ULL << 0) /**< Header buffer > > > overflow. */ > > > > -#define PKT_RX_RECIP_ERR (0ULL << 0) /**< Hardware processing > > > error. */ > > > > -#define PKT_RX_MAC_ERR (0ULL << 0) /**< MAC error. */ > > > > #define PKT_RX_IPV4_HDR (1ULL << 5) /**< RX packet with IPv4 > > > header. */ > > > > #define PKT_RX_IPV4_HDR_EXT (1ULL << 6) /**< RX packet with > > > extended IPv4 header. */ > > > > #define PKT_RX_IPV6_HDR (1ULL << 7) /**< RX packet with IPv6 > > > header. */ > > > > @@ -99,6 +94,10 @@ extern "C" { > > > > #define PKT_RX_TUNNEL_IPV6_HDR (1ULL << 12) /**< RX tunnel packet > > > with IPv6 header. */ > > > > #define PKT_RX_FDIR_ID (1ULL << 13) /**< FD id reported if F= DIR > > > match. */ > > > > #define PKT_RX_FDIR_FLX (1ULL << 14) /**< Flexible bytes > > reported > > > if FDIR match. */ > > > > +#define PKT_RX_EIP_CKSUM_BAD (1ULL << 15) /**< External IP header > > > checksum error. */ > > > > +#define PKT_RX_OVERSIZE (1ULL << 16) /**< Num of desc of an > > RX > > > pkt oversize. */ > > > > +#define PKT_RX_HBUF_OVERFLOW (1ULL << 17) /**< Header buffer > > > overflow. */ > > > > +#define PKT_RX_MAC_ERR (1ULL << 18) /**< MAC error. */ > > > > /* add new RX flags here */ > > > > > > I still think there is no point to have several flags to indicate HW > > > error for the packet. > > > As I suggested before we can collapse 3 of them (OVERSIZE, > > > HBUF_OVERFLOW, > > > MAC_ERR) into one. > > > As I remember, Oliver even suggested to drop such packets. > > > As was said above - if is not a whole packet SW can't do much with it= anyway. > > > The only thing such bad packets can probably be used for - some sort > > > of debugging. > > > So we probably can combine both things: > > > - in normal operation just drop such packet > > > - if PMD_DEBUG_RX is enabled, then write a log record, set RX_HW_ERR > > > and deliver a packet to the upper layer. > I still do not want to drop the bad packet here, as it may affect vector = processing. > At least it should be investigated how much impact on vector RX. I prefer= to let > up-layer software do that. But right now, i40e doesn't have any vector RX support. You probably meant implications with scatter RX, no? Another thing - if you HW can't receive packets normally, then something re= ally wrong is going on. In such situation, you probably wouldn't worry about your RX performance an= yway :)=20 But I suppose it is a good first step, let's just collapse 3 error flags in= to one for now. Then we can decide should we drop such packets or not.=20 Konstantin >=20 > > > > > > > > > > > /* add new TX flags here */ > > > > @@ -141,13 +140,13 @@ extern "C" { > > > > #define PKT_TX_IP_CKSUM (1ULL << 54) /**< IP cksum of TX pkt. > > > computed by NIC. */ > > > > #define PKT_TX_IPV4_CSUM PKT_TX_IP_CKSUM /**< Alias of > > > PKT_TX_IP_CKSUM. */ > > > > > > > > -/** Tell the NIC it's an IPv4 packet. Required for L4 checksum off= load or > > TSO. > > > */ > > > > -#define PKT_TX_IPV4 PKT_RX_IPV4_HDR > > > > +/** Tell the NIC it's an IPv4 packet. */ > > > > +#define PKT_TX_IPV4 (1ULL << 55) /**< TX packet is a IPV4 > > > packet. */ > > > > > > > > -/** Tell the NIC it's an IPv6 packet. Required for L4 checksum off= load or > > TSO. > > > */ > > > > -#define PKT_TX_IPV6 PKT_RX_IPV6_HDR > > > > +/** Tell the NIC it's an IPv6 packet. */ > > > > +#define PKT_TX_IPV6 (1ULL << 56) /**< TX packet is a IPV6 > > > packet. */ > > > > > > > > -#define PKT_TX_VLAN_PKT (1ULL << 55) /**< TX packet is a 802.= 1q > > > VLAN packet. */ > > > > +#define PKT_TX_VLAN_PKT (1ULL << 57) /**< TX packet is a VLAN > > > packet. */ > > > > > > I don't think these changes should be part of that patch. > > > They violate another patch that Frank sent before. > > > > > > Konstantin > > > > > > > > > > > /* Use final bit of flags to indicate a control mbuf */ > > > > #define CTRL_MBUF_FLAG (1ULL << 63) /**< Mbuf contains > > control > > > data */ > > > > -- > > > > 1.9.3