[dpdk-dev] [PATCH v12 0/7] ethdev: add new Rx offload flags

Ferruh Yigit ferruh.yigit at intel.com
Thu Oct 17 19:43:39 CEST 2019


On 10/17/2019 1:02 PM, pbhagavatula at marvell.com wrote:
> From: Pavan Nikhilesh <pbhagavatula at 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

<...>


More information about the dev mailing list