[PATCH v2 1/1] ethdev: support congestion management

[Bruce Richardson]

On Tue, Sep 27, 2022 at 04:36:18PM +0200, Thomas Monjalon wrote:
> 19/09/2022 14:15, skori at marvell.com:
> > From: Jerin Jacob <jerinj at marvell.com>
> > 
> > NIC HW controllers often come with congestion management support on
> > various HW objects such as Rx queue depth or mempool queue depth.
> > 
> > Also, it can support various modes of operation such as RED (Random
> > early discard), WRED etc on those HW objects.
> > 
> > This patch adds a framework to express such modes(enum rte_cman_mode)
> > and introduce (enum rte_eth_cman_obj) to enumerate the different
> > objects where the modes can operate on.
> [...]
> >  doc/guides/nics/features.rst         |  12 +++
> >  doc/guides/nics/features/default.ini |   1 +
> >  lib/eal/include/meson.build          |   1 +
> >  lib/eal/include/rte_cman.h           |  55 ++++++++++
> >  lib/ethdev/ethdev_driver.h           |  25 +++++
> >  lib/ethdev/meson.build               |   1 + lib/ethdev/rte_cman.c |
> >  101 ++++++++++++++++++ lib/ethdev/rte_ethdev.h              | 151
> >  +++++++++++++++++++++++++++ lib/ethdev/version.map               |   6
> >  ++
> 
> I feel EAL is not the right place for CMAN definitions.
> 
> After a discussion with Jerin, I understand we could use the same
> definitions in other API classes, not only ethdev.  However I think this
> .h file should be better hosted in lib/ethdev/ with its own namespace
> rte_cman.  Then other libs could include this rte_cman.h without having a
> strong dependency on ethdev.
> 
> Deal?
> 
Would rte_net also be an option? Alternatively, is this related to the work
and structures defined in the meter library header, since it seems also
related to congestion management on ingress?

/Bruce