Message ID | 20191017120245.984-1-pbhagavatula@marvell.com (mailing list archive) |
---|---|
Headers |
Return-Path: <dev-bounces@dpdk.org> X-Original-To: patchwork@dpdk.org Delivered-To: patchwork@dpdk.org Received: from [92.243.14.124] (localhost [127.0.0.1]) by dpdk.org (Postfix) with ESMTP id C1D091E566; Thu, 17 Oct 2019 14:02:52 +0200 (CEST) Received: from mx0b-0016f401.pphosted.com (mx0a-0016f401.pphosted.com [67.231.148.174]) by dpdk.org (Postfix) with ESMTP id D09B61DFE7 for <dev@dpdk.org>; Thu, 17 Oct 2019 14:02:51 +0200 (CEST) Received: from pps.filterd (m0045849.ppops.net [127.0.0.1]) by mx0a-0016f401.pphosted.com (8.16.0.42/8.16.0.42) with SMTP id x9HBuAKT008222; Thu, 17 Oct 2019 05:02:51 -0700 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=marvell.com; h=from : to : cc : subject : date : message-id : in-reply-to : references : mime-version : content-type : content-transfer-encoding; s=pfpt0818; bh=z4FC1gIw4rahHIzTAiiOxWI8BtuvXZxZHzgHat1B88M=; b=HHIl3a7jMWioVKeUNO2lMfaz7b6mli3XLRVKZipWSKa6jBiBruOHsdfP81VohZ/9gQ/h ZJZdc44RQHzs2VW1DAGRijyaPBRsY08b9WeNsZPs+gYwlJM5mhbBJQ5IrBlI5Z6FO7vU Hy7AuLhBDrkYJK5Jab6C/xb6S5NJS54zFcHAjNjGu1Jg9com4hFQZnBTK0nnx7Gmmhe7 R8t621IxGW017xrh68vIjK4NwgttvEN4ZIsCwLLPew+mJaBdC4QCBfZGmPBjQfaTIVMJ kLJsEJ6eXjePs31YE1erlzGV+U2VO2GGhim1AAS1+m2Vr7KdqKB6HRGuiIzIsSRvUZhI sw== Received: from sc-exch01.marvell.com ([199.233.58.181]) by mx0a-0016f401.pphosted.com with ESMTP id 2vnpmbqgtm-1 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-SHA384 bits=256 verify=NOT); Thu, 17 Oct 2019 05:02:49 -0700 Received: from SC-EXCH01.marvell.com (10.93.176.81) by SC-EXCH01.marvell.com (10.93.176.81) with Microsoft SMTP Server (TLS) id 15.0.1367.3; Thu, 17 Oct 2019 05:02:48 -0700 Received: from maili.marvell.com (10.93.176.43) by SC-EXCH01.marvell.com (10.93.176.81) with Microsoft SMTP Server id 15.0.1367.3 via Frontend Transport; Thu, 17 Oct 2019 05:02:48 -0700 Received: from BG-LT7430.marvell.com (bg-lt7430.marvell.com [10.28.10.255]) by maili.marvell.com (Postfix) with ESMTP id AC6B03F7040; Thu, 17 Oct 2019 05:02:46 -0700 (PDT) From: <pbhagavatula@marvell.com> To: <ferruh.yigit@intel.com>, <arybchenko@solarflare.com>, <jerinj@marvell.com> CC: <dev@dpdk.org>, Pavan Nikhilesh <pbhagavatula@marvell.com> Date: Thu, 17 Oct 2019 17:32:37 +0530 Message-ID: <20191017120245.984-1-pbhagavatula@marvell.com> X-Mailer: git-send-email 2.17.1 In-Reply-To: <20191010105130.3382-1-pbhagavatula@marvell.com> References: <20191010105130.3382-1-pbhagavatula@marvell.com> MIME-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 8bit X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10434:6.0.95,1.0.8 definitions=2019-10-17_04:2019-10-17,2019-10-17 signatures=0 Subject: [dpdk-dev] [PATCH v12 0/7] ethdev: add new Rx offload flags X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions <dev.dpdk.org> List-Unsubscribe: <https://mails.dpdk.org/options/dev>, <mailto:dev-request@dpdk.org?subject=unsubscribe> List-Archive: <http://mails.dpdk.org/archives/dev/> List-Post: <mailto:dev@dpdk.org> List-Help: <mailto:dev-request@dpdk.org?subject=help> List-Subscribe: <https://mails.dpdk.org/listinfo/dev>, <mailto:dev-request@dpdk.org?subject=subscribe> Errors-To: dev-bounces@dpdk.org Sender: "dev" <dev-bounces@dpdk.org> |
Series |
ethdev: add new Rx offload flags
|
|
Message
Pavan Nikhilesh Bhagavatula
Oct. 17, 2019, 12:02 p.m. UTC
From: Pavan Nikhilesh <pbhagavatula@marvell.com>
Add new Rx offload flags `DEV_RX_OFFLOAD_RSS_HASH` and
`DEV_RX_OFFLOAD_FLOW_MARK`. These flags can be used to
enable/disable PMD writes to rte_mbuf fields `hash.rss` and `hash.fdir.hi`
and also `ol_flags:PKT_RX_RSS` and `ol_flags:PKT_RX_FDIR`.
Add new packet type set function `rte_eth_dev_set_supported_ptypes`,
allows application to inform PMDs about the packet types it is interested
in. Based on ptypes requested by application PMDs can optimize the Rx path.
For example, if a given PMD doesn't support any packet types that the
application is interested in then the application can disable[1] writes to
`mbuf.packet_type` done by the PMD and use a software ptype parser.
[1] rte_eth_dev_set_supported_ptypes(*port_id*, RTE_PTYPE_UNKNOWN,
NULL, 0);
v12 Changes:
-----------
- Rebase onto next-net.
v11 Changes:
-----------
- Use RTE_DIM to get array size.
- Since we are using a list of MASKs to validate ptype_mask return -EINVAL
if any unknown mask is set.
- Rebase to TOT.
v10 Changes:
-----------
- Modify ptype_mask validation in set_supported_ptypes.(Andrew)
v9 Changes:
----------
- Add ptype_mask validation in set_supported_ptypes.(Andrew)
- Make description more verbose.
v8 Changes:
----------
- Make description more verbose.
- Set RTE_PTYPE_UNKNOWN in set_ptypes array when either get ot set ptypes
is not supported by ethernet device.
v7 Changes:
----------
- Fix unused variable in net/octeontx2
v6 Changes:
----------
- Add additional checks for set supported ptypes.(Andrew)
- Clarify `rte_eth_dev_set_supported_ptypes` documentation.
- Remove DEV_RX_OFFLOAD_FLOW_MARK emulation from net/octeontx2.
v5 Changes:
----------
- Fix typos.
v4 Changes:
----------
- Set the last element in set_ptype array as RTE_PTYPE_UNKNOWN to mark the end
of array.
- Fix invalid set ptype function call in examples.
- Remove setting rte_eth_dev_set_supported_ptypes to UNKNOWN in l3fwd-power.
v3 Changes:
----------
- Add missing release notes. (Andrew)
- Re-word various descriptions.
- Fix ptype set logic.
v2 Changes:
----------
- Update release notes. (Andrew)
- Redo commit logs. (Andrew)
- Disable ptype parsing for unsupported examples. (Jerin)
- Disable RSS write only in generic mode eventdev_pipeline. (Jerin)
- Modify set_supported_ptypes function to return successfuly set mask
instead of failure.
- Dropped set_supported_ptypes to drivers by handling in library
layer, interested PMD can add it in.
Pavan Nikhilesh (7):
ethdev: add set ptype function
ethdev: add mbuf RSS update as an offload
ethdev: add flow action type update as an offload
drivers/net: update Rx RSS hash offload capabilities
drivers/net: update Rx flow flag and mark capabilities
examples/eventdev_pipeline: add new Rx RSS hash offload
examples: disable Rx packet type parsing
doc/guides/nics/features.rst | 22 ++-
doc/guides/rel_notes/release_19_11.rst | 24 ++++
drivers/net/bnxt/bnxt_ethdev.c | 4 +-
drivers/net/cxgbe/cxgbe.h | 3 +-
drivers/net/dpaa/dpaa_ethdev.c | 3 +-
drivers/net/dpaa2/dpaa2_ethdev.c | 3 +-
drivers/net/e1000/igb_rxtx.c | 3 +-
drivers/net/enic/enic_res.c | 4 +-
drivers/net/fm10k/fm10k_ethdev.c | 3 +-
drivers/net/hinic/hinic_pmd_ethdev.c | 3 +-
drivers/net/i40e/i40e_ethdev.c | 4 +-
drivers/net/iavf/iavf_ethdev.c | 4 +-
drivers/net/ice/ice_ethdev.c | 4 +-
drivers/net/ixgbe/ixgbe_rxtx.c | 4 +-
drivers/net/liquidio/lio_ethdev.c | 3 +-
drivers/net/mlx4/mlx4_rxq.c | 3 +-
drivers/net/mlx5/mlx5_rxq.c | 4 +-
drivers/net/netvsc/hn_rndis.c | 3 +-
drivers/net/nfp/nfp_net.c | 3 +-
drivers/net/octeontx2/otx2_ethdev.c | 3 +-
drivers/net/octeontx2/otx2_ethdev.h | 16 ++-
drivers/net/octeontx2/otx2_flow.c | 9 +-
drivers/net/octeontx2/otx2_flow.h | 1 -
drivers/net/octeontx2/otx2_flow_parse.c | 5 +-
drivers/net/qede/qede_ethdev.c | 3 +-
drivers/net/sfc/sfc_ef10_essb_rx.c | 3 +-
drivers/net/sfc/sfc_ef10_rx.c | 3 +-
drivers/net/sfc/sfc_rx.c | 3 +-
drivers/net/thunderx/nicvf_ethdev.h | 3 +-
drivers/net/vmxnet3/vmxnet3_ethdev.c | 3 +-
examples/bbdev_app/main.c | 2 +
examples/bond/main.c | 2 +
examples/distributor/Makefile | 1 +
examples/distributor/main.c | 1 +
examples/distributor/meson.build | 1 +
examples/eventdev_pipeline/main.c | 130 +----------------
examples/eventdev_pipeline/meson.build | 1 +
.../pipeline_worker_generic.c | 132 ++++++++++++++++++
.../eventdev_pipeline/pipeline_worker_tx.c | 128 +++++++++++++++++
examples/exception_path/Makefile | 1 +
examples/exception_path/main.c | 1 +
examples/exception_path/meson.build | 1 +
examples/flow_classify/flow_classify.c | 1 +
examples/flow_filtering/Makefile | 1 +
examples/flow_filtering/main.c | 2 +
examples/flow_filtering/meson.build | 1 +
examples/ip_pipeline/link.c | 1 +
examples/ip_reassembly/Makefile | 1 +
examples/ip_reassembly/main.c | 2 +
examples/ip_reassembly/meson.build | 1 +
examples/ipsec-secgw/ipsec-secgw.c | 2 +
examples/ipv4_multicast/Makefile | 1 +
examples/ipv4_multicast/main.c | 3 +
examples/ipv4_multicast/meson.build | 1 +
examples/kni/main.c | 1 +
examples/l2fwd-cat/Makefile | 1 +
examples/l2fwd-cat/l2fwd-cat.c | 1 +
examples/l2fwd-cat/meson.build | 1 +
examples/l2fwd-crypto/main.c | 2 +
examples/l2fwd-jobstats/Makefile | 1 +
examples/l2fwd-jobstats/main.c | 2 +
examples/l2fwd-jobstats/meson.build | 1 +
examples/l2fwd-keepalive/Makefile | 1 +
examples/l2fwd-keepalive/main.c | 2 +
examples/l2fwd-keepalive/meson.build | 1 +
examples/l2fwd/Makefile | 1 +
examples/l2fwd/main.c | 2 +
examples/l2fwd/meson.build | 1 +
examples/l3fwd-acl/Makefile | 1 +
examples/l3fwd-acl/main.c | 2 +
examples/l3fwd-acl/meson.build | 1 +
examples/l3fwd-vf/Makefile | 1 +
examples/l3fwd-vf/main.c | 2 +
examples/l3fwd-vf/meson.build | 1 +
examples/link_status_interrupt/Makefile | 1 +
examples/link_status_interrupt/main.c | 2 +
examples/link_status_interrupt/meson.build | 1 +
examples/load_balancer/Makefile | 1 +
examples/load_balancer/init.c | 2 +
examples/load_balancer/meson.build | 1 +
examples/packet_ordering/Makefile | 1 +
examples/packet_ordering/main.c | 1 +
examples/packet_ordering/meson.build | 1 +
examples/ptpclient/Makefile | 1 +
examples/ptpclient/meson.build | 1 +
examples/ptpclient/ptpclient.c | 1 +
examples/qos_meter/Makefile | 1 +
examples/qos_meter/main.c | 2 +
examples/qos_meter/meson.build | 1 +
examples/qos_sched/Makefile | 1 +
examples/qos_sched/init.c | 1 +
examples/qos_sched/meson.build | 1 +
examples/quota_watermark/qw/Makefile | 1 +
examples/quota_watermark/qw/init.c | 1 +
examples/rxtx_callbacks/main.c | 1 +
examples/server_node_efd/server/Makefile | 1 +
examples/server_node_efd/server/init.c | 1 +
examples/skeleton/Makefile | 1 +
examples/skeleton/basicfwd.c | 1 +
examples/skeleton/meson.build | 1 +
examples/tep_termination/Makefile | 1 +
examples/tep_termination/meson.build | 1 +
examples/tep_termination/vxlan_setup.c | 1 +
examples/vhost/Makefile | 1 +
examples/vhost/main.c | 1 +
examples/vm_power_manager/Makefile | 1 +
examples/vm_power_manager/main.c | 1 +
examples/vm_power_manager/meson.build | 1 +
examples/vmdq/Makefile | 1 +
examples/vmdq/main.c | 1 +
examples/vmdq/meson.build | 1 +
examples/vmdq_dcb/Makefile | 1 +
examples/vmdq_dcb/main.c | 1 +
examples/vmdq_dcb/meson.build | 1 +
lib/librte_ethdev/rte_ethdev.c | 89 +++++++++++-
lib/librte_ethdev/rte_ethdev.h | 39 ++++++
lib/librte_ethdev/rte_ethdev_core.h | 19 +++
lib/librte_ethdev/rte_ethdev_version.map | 1 +
lib/librte_ethdev/rte_flow.h | 6 +-
119 files changed, 621 insertions(+), 176 deletions(-)
--
2.17.1
Comments
On 10/17/2019 1:02 PM, pbhagavatula@marvell.com wrote: > From: Pavan Nikhilesh <pbhagavatula@marvell.com> > > Add new Rx offload flags `DEV_RX_OFFLOAD_RSS_HASH` and > `DEV_RX_OFFLOAD_FLOW_MARK`. These flags can be used to > enable/disable PMD writes to rte_mbuf fields `hash.rss` and `hash.fdir.hi` > and also `ol_flags:PKT_RX_RSS` and `ol_flags:PKT_RX_FDIR`. Hi Pavan, Initially sorry for involving late, When we expose an interface to the applications, they will expect those will be respected by underlying PMDs. As far as I can see drivers are updated to report new added Rx offload flags as supported capabilities but drivers are not using those flags at all, so application providing that flag won't really enable/disable anything, I think this is a problem and it is wrong to lie even for the PMDs J Specific to `DEV_RX_OFFLOAD_RSS_HASH`, we have already some RSS config structures and it is part of the 'rte_eth_dev_configure()' API, won't it create multiple way to do same thing? And for the `ol_flags:PKT_RX_RSS` flag, it was already used to mark that 'mbuf::hash::rss' is valid, right? Is there anything new related that in the set? Specific to the `DEV_RX_OFFLOAD_FLOW_MARK` and `RTE_FLOW_ACTION_FLAG`, they are rte_flow actions, application can verify and later request these actions via rte_flow APIs. Why we are adding an additional RX_OFFLOAD flag for them? > > Add new packet type set function `rte_eth_dev_set_supported_ptypes`, > allows application to inform PMDs about the packet types it is interested > in. Based on ptypes requested by application PMDs can optimize the Rx path. OK to the API, but why "Packet type parsing" feature updated to say it should implement this API? Is this API really required to say "Packet type parsing" supported by PMD? > > For example, if a given PMD doesn't support any packet types that the > application is interested in then the application can disable[1] writes to > `mbuf.packet_type` done by the PMD and use a software ptype parser. > [1] rte_eth_dev_set_supported_ptypes(*port_id*, RTE_PTYPE_UNKNOWN, > NULL, 0); And for the 7/7 patch, why we are updating all examples, is the API something do we really need to call for any DPDK application? I am for leaving the default behavior unless there is a very specific case for set or disable packet typing. Instead implement a command in testpmd to test this feature. > > v12 Changes: > ----------- > - Rebase onto next-net. > > v11 Changes: > ----------- > - Use RTE_DIM to get array size. > - Since we are using a list of MASKs to validate ptype_mask return -EINVAL > if any unknown mask is set. > - Rebase to TOT. > > v10 Changes: > ----------- > - Modify ptype_mask validation in set_supported_ptypes.(Andrew) > > v9 Changes: > ---------- > - Add ptype_mask validation in set_supported_ptypes.(Andrew) > - Make description more verbose. > > v8 Changes: > ---------- > - Make description more verbose. > - Set RTE_PTYPE_UNKNOWN in set_ptypes array when either get ot set ptypes > is not supported by ethernet device. > > v7 Changes: > ---------- > - Fix unused variable in net/octeontx2 > > v6 Changes: > ---------- > - Add additional checks for set supported ptypes.(Andrew) > - Clarify `rte_eth_dev_set_supported_ptypes` documentation. > - Remove DEV_RX_OFFLOAD_FLOW_MARK emulation from net/octeontx2. > > v5 Changes: > ---------- > - Fix typos. > > v4 Changes: > ---------- > - Set the last element in set_ptype array as RTE_PTYPE_UNKNOWN to mark the end > of array. > - Fix invalid set ptype function call in examples. > - Remove setting rte_eth_dev_set_supported_ptypes to UNKNOWN in l3fwd-power. > > v3 Changes: > ---------- > - Add missing release notes. (Andrew) > - Re-word various descriptions. > - Fix ptype set logic. > > v2 Changes: > ---------- > - Update release notes. (Andrew) > - Redo commit logs. (Andrew) > - Disable ptype parsing for unsupported examples. (Jerin) > - Disable RSS write only in generic mode eventdev_pipeline. (Jerin) > - Modify set_supported_ptypes function to return successfuly set mask > instead of failure. > - Dropped set_supported_ptypes to drivers by handling in library > layer, interested PMD can add it in. > > Pavan Nikhilesh (7): > ethdev: add set ptype function > ethdev: add mbuf RSS update as an offload > ethdev: add flow action type update as an offload > drivers/net: update Rx RSS hash offload capabilities > drivers/net: update Rx flow flag and mark capabilities > examples/eventdev_pipeline: add new Rx RSS hash offload > examples: disable Rx packet type parsing <...>
Hi Ferruh, since I've reviewed I'll reply as I understand it. On 10/17/19 8:43 PM, Ferruh Yigit wrote: > On 10/17/2019 1:02 PM, pbhagavatula@marvell.com wrote: >> From: Pavan Nikhilesh <pbhagavatula@marvell.com> >> >> Add new Rx offload flags `DEV_RX_OFFLOAD_RSS_HASH` and >> `DEV_RX_OFFLOAD_FLOW_MARK`. These flags can be used to >> enable/disable PMD writes to rte_mbuf fields `hash.rss` and `hash.fdir.hi` >> and also `ol_flags:PKT_RX_RSS` and `ol_flags:PKT_RX_FDIR`. > Hi Pavan, > > Initially sorry for involving late, > > When we expose an interface to the applications, they will expect those will be > respected by underlying PMDs. > As far as I can see drivers are updated to report new added Rx offload flags as > supported capabilities but drivers are not using those flags at all, so > application providing that flag won't really enable/disable anything, I think > this is a problem and it is wrong to lie even for the PMDs J It is required to let applications know that the offload is supported. There are a number of cases when an offload cannot be disabled, but it does not mean that the offload must not be advertised. If driver see benefits from disabling the offload (e.g. avoid delivery of RSS hash from NIC to host), it can do it after the patchset. > Specific to `DEV_RX_OFFLOAD_RSS_HASH`, we have already some RSS config > structures and it is part of the 'rte_eth_dev_configure()' API, won't it create > multiple way to do same thing? No, a new offload is responsible for RSS hash delivery from NIC to host and fill in in mbuf returned to application on Rx. > And for the `ol_flags:PKT_RX_RSS` flag, it was already used to mark that > 'mbuf::hash::rss' is valid, right? Is there anything new related that in the set? As I understand you mean, ol_flags::PKT_RX_RSS_HASH. Yes, the new offload allows say if application needs it or now. Basically it decouples RSS distribution and hash delivery. > Specific to the `DEV_RX_OFFLOAD_FLOW_MARK` and `RTE_FLOW_ACTION_FLAG`, they are > rte_flow actions, application can verify and later request these actions via > rte_flow APIs. Why we are adding an additional RX_OFFLOAD flag for them? The reason is basically the same as above. HW needs to know in advance, if application is going to use flow marks and configure Rx queue to enable the information delivery. >> Add new packet type set function `rte_eth_dev_set_supported_ptypes`, >> allows application to inform PMDs about the packet types it is interested >> in. Based on ptypes requested by application PMDs can optimize the Rx path. > OK to the API, but why "Packet type parsing" feature updated to say it should > implement this API? > Is this API really required to say "Packet type parsing" supported by PMD? As I understand it is not strictly required, but related to the feature. >> For example, if a given PMD doesn't support any packet types that the >> application is interested in then the application can disable[1] writes to >> `mbuf.packet_type` done by the PMD and use a software ptype parser. >> [1] rte_eth_dev_set_supported_ptypes(*port_id*, RTE_PTYPE_UNKNOWN, >> NULL, 0); > And for the 7/7 patch, why we are updating all examples, is the API something do > we really need to call for any DPDK application? I am for leaving the default > behavior unless there is a very specific case for set or disable packet typing. > Instead implement a command in testpmd to test this feature. If an application does not use packet types provided in mbuf, it is better to inform PMD that the information is not required to allow PMD to do optimizations. Yes, may be it would be better to have it as the default behaviour, but it would be behaviour change in comparison to previous DPDK releases and it is better to avoid it. Thanks, Andrew. >> v12 Changes: >> ----------- >> - Rebase onto next-net. >> >> v11 Changes: >> ----------- >> - Use RTE_DIM to get array size. >> - Since we are using a list of MASKs to validate ptype_mask return -EINVAL >> if any unknown mask is set. >> - Rebase to TOT. >> >> v10 Changes: >> ----------- >> - Modify ptype_mask validation in set_supported_ptypes.(Andrew) >> >> v9 Changes: >> ---------- >> - Add ptype_mask validation in set_supported_ptypes.(Andrew) >> - Make description more verbose. >> >> v8 Changes: >> ---------- >> - Make description more verbose. >> - Set RTE_PTYPE_UNKNOWN in set_ptypes array when either get ot set ptypes >> is not supported by ethernet device. >> >> v7 Changes: >> ---------- >> - Fix unused variable in net/octeontx2 >> >> v6 Changes: >> ---------- >> - Add additional checks for set supported ptypes.(Andrew) >> - Clarify `rte_eth_dev_set_supported_ptypes` documentation. >> - Remove DEV_RX_OFFLOAD_FLOW_MARK emulation from net/octeontx2. >> >> v5 Changes: >> ---------- >> - Fix typos. >> >> v4 Changes: >> ---------- >> - Set the last element in set_ptype array as RTE_PTYPE_UNKNOWN to mark the end >> of array. >> - Fix invalid set ptype function call in examples. >> - Remove setting rte_eth_dev_set_supported_ptypes to UNKNOWN in l3fwd-power. >> >> v3 Changes: >> ---------- >> - Add missing release notes. (Andrew) >> - Re-word various descriptions. >> - Fix ptype set logic. >> >> v2 Changes: >> ---------- >> - Update release notes. (Andrew) >> - Redo commit logs. (Andrew) >> - Disable ptype parsing for unsupported examples. (Jerin) >> - Disable RSS write only in generic mode eventdev_pipeline. (Jerin) >> - Modify set_supported_ptypes function to return successfuly set mask >> instead of failure. >> - Dropped set_supported_ptypes to drivers by handling in library >> layer, interested PMD can add it in. >> >> Pavan Nikhilesh (7): >> ethdev: add set ptype function >> ethdev: add mbuf RSS update as an offload >> ethdev: add flow action type update as an offload >> drivers/net: update Rx RSS hash offload capabilities >> drivers/net: update Rx flow flag and mark capabilities >> examples/eventdev_pipeline: add new Rx RSS hash offload >> examples: disable Rx packet type parsing > <...>
On 10/18/2019 8:32 AM, Andrew Rybchenko wrote: > Hi Ferruh, > > since I've reviewed I'll reply as I understand it. > > On 10/17/19 8:43 PM, Ferruh Yigit wrote: >> On 10/17/2019 1:02 PM, pbhagavatula@marvell.com wrote: >>> From: Pavan Nikhilesh <pbhagavatula@marvell.com> >>> >>> Add new Rx offload flags `DEV_RX_OFFLOAD_RSS_HASH` and >>> `DEV_RX_OFFLOAD_FLOW_MARK`. These flags can be used to >>> enable/disable PMD writes to rte_mbuf fields `hash.rss` and `hash.fdir.hi` >>> and also `ol_flags:PKT_RX_RSS` and `ol_flags:PKT_RX_FDIR`. >> Hi Pavan, >> >> Initially sorry for involving late, >> >> When we expose an interface to the applications, they will expect those will be >> respected by underlying PMDs. >> As far as I can see drivers are updated to report new added Rx offload flags as >> supported capabilities but drivers are not using those flags at all, so >> application providing that flag won't really enable/disable anything, I think >> this is a problem and it is wrong to lie even for the PMDs J > > It is required to let applications know that the offload is supported. > There are a number of cases when an offload cannot be disabled, > but it does not mean that the offload must not be advertised. Can't disable is something else, although I believe that is rare case, in this case driver can enable/disable the RSS and representing this as an offload capability. But when user want to configure this offload by setting or unsetting in offload config, driver just ignores it. > If driver see benefits from disabling the offload (e.g. avoid delivery > of RSS hash from NIC to host), it can do it after the patchset. Yes but I think this patchset shouldn't ignore that disabling the feature is not implemented yet. If those PMDs that has been updated to report the HASH capability has RSS enabled by default, I suggest adding a check for this offload in PMD, if it is requested to disable (which means not requested for enable), print a log saying disabling HASH is not supported and set this flag in the offload configuration to say PMD is configured to calculate the HASH. Later PMD maintainers may prefer to replace that error log with actual disable code. > >> Specific to `DEV_RX_OFFLOAD_RSS_HASH`, we have already some RSS config >> structures and it is part of the 'rte_eth_dev_configure()' API, won't it create >> multiple way to do same thing? > > No, a new offload is responsible for RSS hash delivery from NIC to host > and fill in in mbuf returned to application on Rx. What you have described is already happening without the new offload flag and this is my concern that we are duplicating it. There is a 'struct rte_eth_rxmode' (under 'struct rte_eth_conf') which has 'enum rte_eth_rx_mq_mode mq_mode;' If "mq_mode == ETH_MQ_RX_NONE" hash calculation is disabled, and 'mbuf::hash::rss' is not updated. (Thanks Bruce to helping finding it out) > >> And for the `ol_flags:PKT_RX_RSS` flag, it was already used to mark that >> 'mbuf::hash::rss' is valid, right? Is there anything new related that in the set? > > As I understand you mean, ol_flags::PKT_RX_RSS_HASH. > Yes, the new offload allows say if application needs it or now. > Basically it decouples RSS distribution and hash delivery. Setting 'ol_flags::PKT_RX_RSS_HASH' and 'mbuf::hash::rss' already there and not changing. I just want to clarify since this is not clear in the commit log. Only addition is to add a new flag to control PMD to enable/disable hash calculation (which PMDs ignore in the patch ???) > >> Specific to the `DEV_RX_OFFLOAD_FLOW_MARK` and `RTE_FLOW_ACTION_FLAG`, they are >> rte_flow actions, application can verify and later request these actions via >> rte_flow APIs. Why we are adding an additional RX_OFFLOAD flag for them? > > The reason is basically the same as above. HW needs to know in advance, > if application is going to use flow marks and configure Rx queue to enable > the information delivery. What you described is done via 'rte_flow_create()' API, application will request those actions via API and Rx queue will be configured accordingly, this is more dynamic approach. Why application need to set this additional configuration flag? And as above the new RX offload flags ignored by PMDs, hard to understand what is the intention here. Above usage of flags feels like the intention is adding some capability information for the PMDs more that adding new offload configuration. If so this is bigger/older problem, and instead of abusing the offload flags we can think of an API to present device capabilities, and move features.ini content to the API in long term. > >>> Add new packet type set function `rte_eth_dev_set_supported_ptypes`, >>> allows application to inform PMDs about the packet types it is interested >>> in. Based on ptypes requested by application PMDs can optimize the Rx path. >> OK to the API, but why "Packet type parsing" feature updated to say it should >> implement this API? >> Is this API really required to say "Packet type parsing" supported by PMD? > > As I understand it is not strictly required, but related to the feature. I am OK with "related", but it is documented as "implements", so doc says it is required. > >>> For example, if a given PMD doesn't support any packet types that the >>> application is interested in then the application can disable[1] writes to >>> `mbuf.packet_type` done by the PMD and use a software ptype parser. >>> [1] rte_eth_dev_set_supported_ptypes(*port_id*, RTE_PTYPE_UNKNOWN, >>> NULL, 0); >> And for the 7/7 patch, why we are updating all examples, is the API something do >> we really need to call for any DPDK application? I am for leaving the default >> behavior unless there is a very specific case for set or disable packet typing. >> Instead implement a command in testpmd to test this feature. > > If an application does not use packet types provided in mbuf, it is > better to inform PMD that the information is not required to allow PMD > to do optimizations. > I can see disabling packet type detection may increase the performance but sample applications are to demonstrate a specific feature, adding these kind of APIs will pollute them. 'skeleton' app that shows the most basic code for forwarding sample, why it is now having "experimental" 'set_supported_ptypes()' API? Same for other. As said before I think a testpmd command suits better here. > Yes, may be it would be better to have it as the > default behaviour, but it would be behaviour change in comparison > to previous DPDK releases and it is better to avoid it. Sorry I missed why not calling this function cause behavior change? I think it is other way around, no? > Thanks, > Andrew. > >>> v12 Changes: >>> ----------- >>> - Rebase onto next-net. >>> >>> v11 Changes: >>> ----------- >>> - Use RTE_DIM to get array size. >>> - Since we are using a list of MASKs to validate ptype_mask return -EINVAL >>> if any unknown mask is set. >>> - Rebase to TOT. >>> >>> v10 Changes: >>> ----------- >>> - Modify ptype_mask validation in set_supported_ptypes.(Andrew) >>> >>> v9 Changes: >>> ---------- >>> - Add ptype_mask validation in set_supported_ptypes.(Andrew) >>> - Make description more verbose. >>> >>> v8 Changes: >>> ---------- >>> - Make description more verbose. >>> - Set RTE_PTYPE_UNKNOWN in set_ptypes array when either get ot set ptypes >>> is not supported by ethernet device. >>> >>> v7 Changes: >>> ---------- >>> - Fix unused variable in net/octeontx2 >>> >>> v6 Changes: >>> ---------- >>> - Add additional checks for set supported ptypes.(Andrew) >>> - Clarify `rte_eth_dev_set_supported_ptypes` documentation. >>> - Remove DEV_RX_OFFLOAD_FLOW_MARK emulation from net/octeontx2. >>> >>> v5 Changes: >>> ---------- >>> - Fix typos. >>> >>> v4 Changes: >>> ---------- >>> - Set the last element in set_ptype array as RTE_PTYPE_UNKNOWN to mark the end >>> of array. >>> - Fix invalid set ptype function call in examples. >>> - Remove setting rte_eth_dev_set_supported_ptypes to UNKNOWN in l3fwd-power. >>> >>> v3 Changes: >>> ---------- >>> - Add missing release notes. (Andrew) >>> - Re-word various descriptions. >>> - Fix ptype set logic. >>> >>> v2 Changes: >>> ---------- >>> - Update release notes. (Andrew) >>> - Redo commit logs. (Andrew) >>> - Disable ptype parsing for unsupported examples. (Jerin) >>> - Disable RSS write only in generic mode eventdev_pipeline. (Jerin) >>> - Modify set_supported_ptypes function to return successfuly set mask >>> instead of failure. >>> - Dropped set_supported_ptypes to drivers by handling in library >>> layer, interested PMD can add it in. >>> >>> Pavan Nikhilesh (7): >>> ethdev: add set ptype function >>> ethdev: add mbuf RSS update as an offload >>> ethdev: add flow action type update as an offload >>> drivers/net: update Rx RSS hash offload capabilities >>> drivers/net: update Rx flow flag and mark capabilities >>> examples/eventdev_pipeline: add new Rx RSS hash offload >>> examples: disable Rx packet type parsing >> <...> >
On 10/18/19 12:42 PM, Ferruh Yigit wrote: > On 10/18/2019 8:32 AM, Andrew Rybchenko wrote: >> Hi Ferruh, >> >> since I've reviewed I'll reply as I understand it. >> >> On 10/17/19 8:43 PM, Ferruh Yigit wrote: >>> On 10/17/2019 1:02 PM, pbhagavatula@marvell.com wrote: >>>> From: Pavan Nikhilesh <pbhagavatula@marvell.com> >>>> >>>> Add new Rx offload flags `DEV_RX_OFFLOAD_RSS_HASH` and >>>> `DEV_RX_OFFLOAD_FLOW_MARK`. These flags can be used to >>>> enable/disable PMD writes to rte_mbuf fields `hash.rss` and `hash.fdir.hi` >>>> and also `ol_flags:PKT_RX_RSS` and `ol_flags:PKT_RX_FDIR`. >>> Hi Pavan, >>> >>> Initially sorry for involving late, >>> >>> When we expose an interface to the applications, they will expect those will be >>> respected by underlying PMDs. >>> As far as I can see drivers are updated to report new added Rx offload flags as >>> supported capabilities but drivers are not using those flags at all, so >>> application providing that flag won't really enable/disable anything, I think >>> this is a problem and it is wrong to lie even for the PMDs J >> It is required to let applications know that the offload is supported. >> There are a number of cases when an offload cannot be disabled, >> but it does not mean that the offload must not be advertised. > Can't disable is something else, although I believe that is rare case, in this > case driver can enable/disable the RSS and representing this as an offload > capability. It is not enabling/disabling the RSS. It is enabling/disabling RSS hash delivery together with an mbuf. > But when user want to configure this offload by setting or unsetting in offload > config, driver just ignores it. When application enables offload, it says that it needs it and going to use (required). When the offload is not enabled, application simply don't care. So, if the information is still provided it does not harm. >> If driver see benefits from disabling the offload (e.g. avoid delivery >> of RSS hash from NIC to host), it can do it after the patchset. > Yes but I think this patchset shouldn't ignore that disabling the feature is not > implemented yet. If those PMDs that has been updated to report the HASH > capability has RSS enabled by default, I suggest adding a check for this offload > in PMD, > if it is requested to disable (which means not requested for enable), print a > log saying disabling HASH is not supported and set this flag in the offload > configuration to say PMD is configured to calculate the HASH. > Later PMD maintainers may prefer to replace that error log with actual disable code. It is possible to do. Of course, it is better to provide real offload values on get, but eth_conf is const in rte_eth_dev_configure(), so, we can't change it and it is good. So, the only way is rte_eth_rx_queue_info_get(). I guess there is a lot of space for the same improvement for other Rx offloads in various PMDs. Also I worry that it could be not that trivial to do in all effected PMDs. >>> Specific to `DEV_RX_OFFLOAD_RSS_HASH`, we have already some RSS config >>> structures and it is part of the 'rte_eth_dev_configure()' API, won't it create >>> multiple way to do same thing? >> No, a new offload is responsible for RSS hash delivery from NIC to host >> and fill in in mbuf returned to application on Rx. > What you have described is already happening without the new offload flag and > this is my concern that we are duplicating it. > > > There is a 'struct rte_eth_rxmode' (under 'struct rte_eth_conf') > which has 'enum rte_eth_rx_mq_mode mq_mode;' > > If "mq_mode == ETH_MQ_RX_NONE" hash calculation is disabled, and > 'mbuf::hash::rss' is not updated. No-no. It binds RSS distribution and hash delivery. What the new offload allows to achieve: I want Rx to spread traffic over many Rx queues, but I don't need RSS hash. > (Thanks Bruce to helping finding it out) > > >>> And for the `ol_flags:PKT_RX_RSS` flag, it was already used to mark that >>> 'mbuf::hash::rss' is valid, right? Is there anything new related that in the set? >> As I understand you mean, ol_flags::PKT_RX_RSS_HASH. >> Yes, the new offload allows say if application needs it or now. >> Basically it decouples RSS distribution and hash delivery. > Setting 'ol_flags::PKT_RX_RSS_HASH' and 'mbuf::hash::rss' already there and not > changing. I just want to clarify since this is not clear in the commit log. > > Only addition is to add a new flag to control PMD to enable/disable hash > calculation (which PMDs ignore in the patch ???) It is not calculation, but delivery of the value from HW to applications. Yes, commit log may/should be improved. >>> Specific to the `DEV_RX_OFFLOAD_FLOW_MARK` and `RTE_FLOW_ACTION_FLAG`, they are >>> rte_flow actions, application can verify and later request these actions via >>> rte_flow APIs. Why we are adding an additional RX_OFFLOAD flag for them? >> The reason is basically the same as above. HW needs to know in advance, >> if application is going to use flow marks and configure Rx queue to enable >> the information delivery. > What you described is done via 'rte_flow_create()' API, application will request > those actions via API and Rx queue will be configured accordingly, this is more > dynamic approach. Why application need to set this additional configuration flag? More dynamic approach is definitely better, but it is not always possible. Some PMDs can't even change MTU dynamically or MTU changing requires restart which is hardly really a dynamic change. Of course, it is unlikely that MTU is changed when traffic is running etc, but still possible. The information about necessity to support flow marks delivery may be required on Rx queue setup and cannot be changed dynamically when Rx queue is running and application would like to add flow rule with mark action. > And as above the new RX offload flags ignored by PMDs, hard to understand what > is the intention here. > > > Above usage of flags feels like the intention is adding some capability > information for the PMDs more that adding new offload configuration. > If so this is bigger/older problem, and instead of abusing the offload flags we > can think of an API to present device capabilities, and move features.ini > content to the API in long term. What I really like with these new offload flags for Rx hash and flow mark is that it makes features which provide information in mbuf on Rx consistent: - want timestamp? => DEV_RX_OFFLOAD_TIMESTAMP - want Rx checksum flags => DEV_RX_OFFLOAD_CHECKSUM - want to strip VLAN? => DEV_RX_OFFLOAD_VLAN_STRIP - want RSS hash? => DEV_RX_OFFLOAD_RSS_HASH - want flow mark support? => DEV_RX_OFFLOAD_FLOW_MARK Also it perfectly fits dynamic mbuf fields and allows to make RSS hash and flow mark fields dynamic with the new offloads as controls. >>>> Add new packet type set function `rte_eth_dev_set_supported_ptypes`, >>>> allows application to inform PMDs about the packet types it is interested >>>> in. Based on ptypes requested by application PMDs can optimize the Rx path. >>> OK to the API, but why "Packet type parsing" feature updated to say it should >>> implement this API? >>> Is this API really required to say "Packet type parsing" supported by PMD? >> As I understand it is not strictly required, but related to the feature. > I am OK with "related", but it is documented as "implements", so doc says it is > required. Agreed. >>>> For example, if a given PMD doesn't support any packet types that the >>>> application is interested in then the application can disable[1] writes to >>>> `mbuf.packet_type` done by the PMD and use a software ptype parser. >>>> [1] rte_eth_dev_set_supported_ptypes(*port_id*, RTE_PTYPE_UNKNOWN, >>>> NULL, 0); >>> And for the 7/7 patch, why we are updating all examples, is the API something do >>> we really need to call for any DPDK application? I am for leaving the default >>> behavior unless there is a very specific case for set or disable packet typing. >>> Instead implement a command in testpmd to test this feature. >> If an application does not use packet types provided in mbuf, it is >> better to inform PMD that the information is not required to allow PMD >> to do optimizations. >> > I can see disabling packet type detection may increase the performance but > sample applications are to demonstrate a specific feature, adding these kind of > APIs will pollute them. > 'skeleton' app that shows the most basic code for forwarding sample, why it is > now having "experimental" 'set_supported_ptypes()' API? Same for other. As said > before I think a testpmd command suits better here. May be you're right and we should reconsider which applications are updated and which are ignored. I guess before the criteria was simple: don't use packet type information, say so to take benefits from all possible optimizations. >> Yes, may be it would be better to have it as the >> default behaviour, but it would be behaviour change in comparison >> to previous DPDK releases and it is better to avoid it. > Sorry I missed why not calling this function cause behavior change? I think it > is other way around, no? Just misunderstanding. What I was trying to say is that it could be more logical to have packet type parsing and delivery disabled by default (as we have for all other offloads), but it would be behaviour change from application point of view. That's why it is necessary to disable explicitly. Thanks, Andrew. >> Thanks, >> Andrew. >> >>>> v12 Changes: >>>> ----------- >>>> - Rebase onto next-net. >>>> >>>> v11 Changes: >>>> ----------- >>>> - Use RTE_DIM to get array size. >>>> - Since we are using a list of MASKs to validate ptype_mask return -EINVAL >>>> if any unknown mask is set. >>>> - Rebase to TOT. >>>> >>>> v10 Changes: >>>> ----------- >>>> - Modify ptype_mask validation in set_supported_ptypes.(Andrew) >>>> >>>> v9 Changes: >>>> ---------- >>>> - Add ptype_mask validation in set_supported_ptypes.(Andrew) >>>> - Make description more verbose. >>>> >>>> v8 Changes: >>>> ---------- >>>> - Make description more verbose. >>>> - Set RTE_PTYPE_UNKNOWN in set_ptypes array when either get ot set ptypes >>>> is not supported by ethernet device. >>>> >>>> v7 Changes: >>>> ---------- >>>> - Fix unused variable in net/octeontx2 >>>> >>>> v6 Changes: >>>> ---------- >>>> - Add additional checks for set supported ptypes.(Andrew) >>>> - Clarify `rte_eth_dev_set_supported_ptypes` documentation. >>>> - Remove DEV_RX_OFFLOAD_FLOW_MARK emulation from net/octeontx2. >>>> >>>> v5 Changes: >>>> ---------- >>>> - Fix typos. >>>> >>>> v4 Changes: >>>> ---------- >>>> - Set the last element in set_ptype array as RTE_PTYPE_UNKNOWN to mark the end >>>> of array. >>>> - Fix invalid set ptype function call in examples. >>>> - Remove setting rte_eth_dev_set_supported_ptypes to UNKNOWN in l3fwd-power. >>>> >>>> v3 Changes: >>>> ---------- >>>> - Add missing release notes. (Andrew) >>>> - Re-word various descriptions. >>>> - Fix ptype set logic. >>>> >>>> v2 Changes: >>>> ---------- >>>> - Update release notes. (Andrew) >>>> - Redo commit logs. (Andrew) >>>> - Disable ptype parsing for unsupported examples. (Jerin) >>>> - Disable RSS write only in generic mode eventdev_pipeline. (Jerin) >>>> - Modify set_supported_ptypes function to return successfuly set mask >>>> instead of failure. >>>> - Dropped set_supported_ptypes to drivers by handling in library >>>> layer, interested PMD can add it in. >>>> >>>> Pavan Nikhilesh (7): >>>> ethdev: add set ptype function >>>> ethdev: add mbuf RSS update as an offload >>>> ethdev: add flow action type update as an offload >>>> drivers/net: update Rx RSS hash offload capabilities >>>> drivers/net: update Rx flow flag and mark capabilities >>>> examples/eventdev_pipeline: add new Rx RSS hash offload >>>> examples: disable Rx packet type parsing >>> <...>
On 10/18/2019 11:31 AM, Andrew Rybchenko wrote: > On 10/18/19 12:42 PM, Ferruh Yigit wrote: >> On 10/18/2019 8:32 AM, Andrew Rybchenko wrote: >>> Hi Ferruh, >>> >>> since I've reviewed I'll reply as I understand it. >>> >>> On 10/17/19 8:43 PM, Ferruh Yigit wrote: >>>> On 10/17/2019 1:02 PM, pbhagavatula@marvell.com wrote: >>>>> From: Pavan Nikhilesh <pbhagavatula@marvell.com> >>>>> >>>>> Add new Rx offload flags `DEV_RX_OFFLOAD_RSS_HASH` and >>>>> `DEV_RX_OFFLOAD_FLOW_MARK`. These flags can be used to >>>>> enable/disable PMD writes to rte_mbuf fields `hash.rss` and `hash.fdir.hi` >>>>> and also `ol_flags:PKT_RX_RSS` and `ol_flags:PKT_RX_FDIR`. >>>> Hi Pavan, >>>> >>>> Initially sorry for involving late, >>>> >>>> When we expose an interface to the applications, they will expect those will be >>>> respected by underlying PMDs. >>>> As far as I can see drivers are updated to report new added Rx offload flags as >>>> supported capabilities but drivers are not using those flags at all, so >>>> application providing that flag won't really enable/disable anything, I think >>>> this is a problem and it is wrong to lie even for the PMDs J >>> It is required to let applications know that the offload is supported. >>> There are a number of cases when an offload cannot be disabled, >>> but it does not mean that the offload must not be advertised. >> Can't disable is something else, although I believe that is rare case, in this >> case driver can enable/disable the RSS and representing this as an offload >> capability. > > It is not enabling/disabling the RSS. It is enabling/disabling RSS hash > delivery > together with an mbuf. Got it, it is related to the RSS hash delivery. > >> But when user want to configure this offload by setting or unsetting in offload >> config, driver just ignores it. > > When application enables offload, it says that it needs it and going to use > (required). When the offload is not enabled, application simply don't care. > So, if the information is still provided it does not harm. Not sure if there is no harm, a config option not respected by underlying PMDs silently is a problem I think. > >>> If driver see benefits from disabling the offload (e.g. avoid delivery >>> of RSS hash from NIC to host), it can do it after the patchset. >> Yes but I think this patchset shouldn't ignore that disabling the feature is not >> implemented yet. If those PMDs that has been updated to report the HASH >> capability has RSS enabled by default, I suggest adding a check for this offload >> in PMD, >> if it is requested to disable (which means not requested for enable), print a >> log saying disabling HASH is not supported and set this flag in the offload >> configuration to say PMD is configured to calculate the HASH. >> Later PMD maintainers may prefer to replace that error log with actual disable code. > > It is possible to do. Of course, it is better to provide real offload > values on get, but > eth_conf is const in rte_eth_dev_configure(), so, we can't change it and > it is good. > So, the only way is rte_eth_rx_queue_info_get(). > I guess there is a lot of space for the same improvement for other Rx > offloads > in various PMDs. We don't need the update 'eth_conf' parameter of the 'rte_eth_dev_configure()', that is what user requested, but config stored in 'dev->data->dev_conf' which can be updated. > Also I worry that it could be not that trivial to do in all effected PMDs. Yes it can be some work, and if this patchset doesn't do it, who will do the work? > >>>> Specific to `DEV_RX_OFFLOAD_RSS_HASH`, we have already some RSS config >>>> structures and it is part of the 'rte_eth_dev_configure()' API, won't it create >>>> multiple way to do same thing? >>> No, a new offload is responsible for RSS hash delivery from NIC to host >>> and fill in in mbuf returned to application on Rx. >> What you have described is already happening without the new offload flag and >> this is my concern that we are duplicating it. >> >> >> There is a 'struct rte_eth_rxmode' (under 'struct rte_eth_conf') >> which has 'enum rte_eth_rx_mq_mode mq_mode;' >> >> If "mq_mode == ETH_MQ_RX_NONE" hash calculation is disabled, and >> 'mbuf::hash::rss' is not updated. > > No-no. It binds RSS distribution and hash delivery. What the new > offload allows to achieve: I want Rx to spread traffic over many Rx > queues, but I don't need RSS hash. I see, so RSS configuration will stay same, but driver needs to take care the new flags to decide to update or not the mbuf::rss::hash field. I don't know if disabling RSS but calculating hash is supported, if not supported that case also should be checked by driver. > >> (Thanks Bruce to helping finding it out) >> >> >>>> And for the `ol_flags:PKT_RX_RSS` flag, it was already used to mark that >>>> 'mbuf::hash::rss' is valid, right? Is there anything new related that in the set? >>> As I understand you mean, ol_flags::PKT_RX_RSS_HASH. >>> Yes, the new offload allows say if application needs it or now. >>> Basically it decouples RSS distribution and hash delivery. >> Setting 'ol_flags::PKT_RX_RSS_HASH' and 'mbuf::hash::rss' already there and not >> changing. I just want to clarify since this is not clear in the commit log. >> >> Only addition is to add a new flag to control PMD to enable/disable hash >> calculation (which PMDs ignore in the patch ???) > > It is not calculation, but delivery of the value from HW to applications. OK > Yes, commit log may/should be improved.> >>>> Specific to the `DEV_RX_OFFLOAD_FLOW_MARK` and `RTE_FLOW_ACTION_FLAG`, they are >>>> rte_flow actions, application can verify and later request these actions via >>>> rte_flow APIs. Why we are adding an additional RX_OFFLOAD flag for them? >>> The reason is basically the same as above. HW needs to know in advance, >>> if application is going to use flow marks and configure Rx queue to enable >>> the information delivery. >> What you described is done via 'rte_flow_create()' API, application will request >> those actions via API and Rx queue will be configured accordingly, this is more >> dynamic approach. Why application need to set this additional configuration flag? > > More dynamic approach is definitely better, but it is not always possible. > Some PMDs can't even change MTU dynamically or MTU changing requires > restart which is hardly really a dynamic change. Of course, it is > unlikely that > MTU is changed when traffic is running etc, but still possible. > The information about necessity to support flow marks delivery may > be required on Rx queue setup and cannot be changed dynamically when > Rx queue is running and application would like to add flow rule with mark > action. It doesn't need to be changed dynamically, application can call 'rte_flow_validate()' and learn if it can set this action or not. Perhaps I am missing something, when it is required to have this as configuration option? > >> And as above the new RX offload flags ignored by PMDs, hard to understand what >> is the intention here. >> >> >> Above usage of flags feels like the intention is adding some capability >> information for the PMDs more that adding new offload configuration. >> If so this is bigger/older problem, and instead of abusing the offload flags we >> can think of an API to present device capabilities, and move features.ini >> content to the API in long term. > > What I really like with these new offload flags for Rx hash and flow mark is > that it makes features which provide information in mbuf on Rx consistent: > - want timestamp? => DEV_RX_OFFLOAD_TIMESTAMP > - want Rx checksum flags => DEV_RX_OFFLOAD_CHECKSUM > - want to strip VLAN? => DEV_RX_OFFLOAD_VLAN_STRIP > - want RSS hash? => DEV_RX_OFFLOAD_RSS_HASH > - want flow mark support? => DEV_RX_OFFLOAD_FLOW_MARK > > Also it perfectly fits dynamic mbuf fields and allows to make RSS hash > and flow mark fields dynamic with the new offloads as controls Agree RSS_HASH fits well, my main concern with the patchset is driver implementations are missing and just ignored. I am not so sure about FLOW_MARK one, it looks like it is duplicating the rte_flow API. > >>>>> Add new packet type set function `rte_eth_dev_set_supported_ptypes`, >>>>> allows application to inform PMDs about the packet types it is interested >>>>> in. Based on ptypes requested by application PMDs can optimize the Rx path. >>>> OK to the API, but why "Packet type parsing" feature updated to say it should >>>> implement this API? >>>> Is this API really required to say "Packet type parsing" supported by PMD? >>> As I understand it is not strictly required, but related to the feature. >> I am OK with "related", but it is documented as "implements", so doc says it is >> required. > > Agreed. > >>>>> For example, if a given PMD doesn't support any packet types that the >>>>> application is interested in then the application can disable[1] writes to >>>>> `mbuf.packet_type` done by the PMD and use a software ptype parser. >>>>> [1] rte_eth_dev_set_supported_ptypes(*port_id*, RTE_PTYPE_UNKNOWN, >>>>> NULL, 0); >>>> And for the 7/7 patch, why we are updating all examples, is the API something do >>>> we really need to call for any DPDK application? I am for leaving the default >>>> behavior unless there is a very specific case for set or disable packet typing. >>>> Instead implement a command in testpmd to test this feature. >>> If an application does not use packet types provided in mbuf, it is >>> better to inform PMD that the information is not required to allow PMD >>> to do optimizations. >>> >> I can see disabling packet type detection may increase the performance but >> sample applications are to demonstrate a specific feature, adding these kind of >> APIs will pollute them. >> 'skeleton' app that shows the most basic code for forwarding sample, why it is >> now having "experimental" 'set_supported_ptypes()' API? Same for other. As said >> before I think a testpmd command suits better here. > > May be you're right and we should reconsider which applications > are updated and which are ignored. I guess before the criteria > was simple: don't use packet type information, say so to take > benefits from all possible optimizations. > >>> Yes, may be it would be better to have it as the >>> default behaviour, but it would be behaviour change in comparison >>> to previous DPDK releases and it is better to avoid it. >> Sorry I missed why not calling this function cause behavior change? I think it >> is other way around, no? > > Just misunderstanding. What I was trying to say is that it could > be more logical to have packet type parsing and delivery > disabled by default (as we have for all other offloads), but > it would be behaviour change from application point of view. > That's why it is necessary to disable explicitly. > > Thanks, > Andrew. > >>> Thanks, >>> Andrew. >>> >>>>> v12 Changes: >>>>> ----------- >>>>> - Rebase onto next-net. >>>>> >>>>> v11 Changes: >>>>> ----------- >>>>> - Use RTE_DIM to get array size. >>>>> - Since we are using a list of MASKs to validate ptype_mask return -EINVAL >>>>> if any unknown mask is set. >>>>> - Rebase to TOT. >>>>> >>>>> v10 Changes: >>>>> ----------- >>>>> - Modify ptype_mask validation in set_supported_ptypes.(Andrew) >>>>> >>>>> v9 Changes: >>>>> ---------- >>>>> - Add ptype_mask validation in set_supported_ptypes.(Andrew) >>>>> - Make description more verbose. >>>>> >>>>> v8 Changes: >>>>> ---------- >>>>> - Make description more verbose. >>>>> - Set RTE_PTYPE_UNKNOWN in set_ptypes array when either get ot set ptypes >>>>> is not supported by ethernet device. >>>>> >>>>> v7 Changes: >>>>> ---------- >>>>> - Fix unused variable in net/octeontx2 >>>>> >>>>> v6 Changes: >>>>> ---------- >>>>> - Add additional checks for set supported ptypes.(Andrew) >>>>> - Clarify `rte_eth_dev_set_supported_ptypes` documentation. >>>>> - Remove DEV_RX_OFFLOAD_FLOW_MARK emulation from net/octeontx2. >>>>> >>>>> v5 Changes: >>>>> ---------- >>>>> - Fix typos. >>>>> >>>>> v4 Changes: >>>>> ---------- >>>>> - Set the last element in set_ptype array as RTE_PTYPE_UNKNOWN to mark the end >>>>> of array. >>>>> - Fix invalid set ptype function call in examples. >>>>> - Remove setting rte_eth_dev_set_supported_ptypes to UNKNOWN in l3fwd-power. >>>>> >>>>> v3 Changes: >>>>> ---------- >>>>> - Add missing release notes. (Andrew) >>>>> - Re-word various descriptions. >>>>> - Fix ptype set logic. >>>>> >>>>> v2 Changes: >>>>> ---------- >>>>> - Update release notes. (Andrew) >>>>> - Redo commit logs. (Andrew) >>>>> - Disable ptype parsing for unsupported examples. (Jerin) >>>>> - Disable RSS write only in generic mode eventdev_pipeline. (Jerin) >>>>> - Modify set_supported_ptypes function to return successfuly set mask >>>>> instead of failure. >>>>> - Dropped set_supported_ptypes to drivers by handling in library >>>>> layer, interested PMD can add it in. >>>>> >>>>> Pavan Nikhilesh (7): >>>>> ethdev: add set ptype function >>>>> ethdev: add mbuf RSS update as an offload >>>>> ethdev: add flow action type update as an offload >>>>> drivers/net: update Rx RSS hash offload capabilities >>>>> drivers/net: update Rx flow flag and mark capabilities >>>>> examples/eventdev_pipeline: add new Rx RSS hash offload >>>>> examples: disable Rx packet type parsing >>>> <...> >