Re: [dpdk-dev] [PATCH] config: disable iova phys for dpaa and 1588 for dpaa2

[Nipun Gupta <nipun.gupta@nxp.com>]

Hi David,

> >
> > > -----Original Message-----
> > > From: David Marchand <david.marchand@redhat.com>
> > > Subject: Re: [dpdk-dev] [PATCH] config: disable iova phys for dpaa and
> 1588
> > > for dpaa2
> > >
> > > On Fri, Oct 4, 2019 at 1:26 PM Nipun Gupta <nipun.gupta@nxp.com>
> wrote:
> > > >
> > > > IOVA_PHYS flag is not required in the DPAA config, thus disable it.
> > > > Also, disable the 1588 timer support by default on DPAA2 platform
> > > > due to the performance impact
> > >
> > > With this patch, we don't have a single target using the 1588 timer
> support.
> > > This means that it will get broken in the future.
> > >
> > > Is this used?
> >
> > Yes, this is enabled when PTP client application is used. We have by default
> disabled it
> > because it has performance implications. We enable it in our system testing
> from time to
> > time basis.
> 
> Ok, but beware that, starting from now, we (maintainers) won't see
> that the compilation gets broken with this feature enabled.
> Hope you can catch issues quickly.

Yes.. we will take care of it.

> 
> 
> > > With this patch, the meson configuration is unaligned with the make
> > > configuration.
> >
> > I am not sure if I get this. In meson build as you mentioned:
> > -
> config/meson.build:dpdk_conf.set('RTE_LIBRTE_DPAA2_USE_PHYS_IOVA',
> true)
> > - config/arm/meson.build: ['RTE_LIBRTE_DPAA2_USE_PHYS_IOVA', false]]
> >
> > And in Makefile:
> > config/common_base: CONFIG_RTE_LIBRTE_DPAA2_USE_PHYS_IOVA=y
> > config/defconfig_arm64-dpaa2-linuxapp-
> gcc:CONFIG_RTE_LIBRTE_DPAA2_USE_PHYS_IOVA=n
> >
> > And now we have:
> > config/defconfig_arm64-dpaa-linuxapp-
> gcc:CONFIG_RTE_LIBRTE_DPAA2_USE_PHYS_IOVA=n
> >
> > Can you please let me know if I need to make any change for meson build.
> 
> You updated the configuration for the dpaa machine in make.
> config/arm/meson.build: ['RTE_LIBRTE_DPAA2_USE_PHYS_IOVA', false]]
> only applies to dpaa2.
> So in meson, the dpaa machine still has
> RTE_LIBRTE_DPAA2_USE_PHYS_IOVA=y
> 
> >
> > >
> > > If I look before the patch:
> > >
> > >
> origin/master:config/meson.build:dpdk_conf.set('RTE_LIBRTE_DPAA2_USE_
> > > PHYS_IOVA',
> > > true)
> > > origin/master:config/arm/meson.build:
> > > ['RTE_LIBRTE_DPAA2_USE_PHYS_IOVA', false]]
> > >
> > >
> origin/master:config/common_base:CONFIG_RTE_LIBRTE_DPAA2_USE_PHY
> > > S_IOVA=y
> > > origin/master:config/defconfig_arm64-dpaa2-linuxapp-
> > > gcc:CONFIG_RTE_LIBRTE_DPAA2_USE_PHYS_IOVA=n
> > >
> > > And in the history, I understand:
> > > - option (with dpaa2 in the name) introduced and enabled for all
> > > - later, disabled for dpaa2
> > >
> > > Does it mean the option can be removed?
> >
> > Here again, we use this option when running DPDK in Virtual Machine
> (Direct Assignment).
> > By default this option is disabled, but for VM scenarios we use this option
> extensively.
> 
> I don't understand when or on which platform you want this
> enabled/disabled.
> 
> 
> If this option is supposed to be disabled by default, then the value
> in those files (below) should be changed to n/false.
> config/common_base:CONFIG_RTE_LIBRTE_DPAA2_USE_PHYS_IOVA=y
> config/meson.build:dpdk_conf.set('RTE_LIBRTE_DPAA2_USE_PHYS_IOVA',
> true)
> doc/guides/nics/dpaa2.rst:-
> ``CONFIG_RTE_LIBRTE_DPAA2_USE_PHYS_IOVA``
> (default ``y``)

Reflecting and analyzing your comments, yes, there are discrepancies.
I will send patch to clean it up.

Thanks,
Nipun

> 
> And then, we end up in the same situation than described above with no
> configuration referencing this option enabled.
> 
> Please clarify.
> 
> 
> --
> David Marchand