[dpdk-dev] [PATCH v14 00/12] Arm build options rework

[Honnappa Nagarahalli]

<snip>

> > >
> > >> >
> > >> > > >
> > >> > > >v14:
> > >> > > >Resent for retesting (because of suspected false negative).
> > >> > > >
> > >> > > >Series Acked-by: Jerin Jacob <jerinj at marvell.com> Series
> > >> > > >Tested-
> > >by:
> > >> > > >Jerin Jacob <jerinj at marvell.com> Series Tested-by: Dharmik
> > >Thakkar
> > >> > > ><dharmik.thakkar at arm.com> Series Tested-by: Vimal Chungath
> > >> > > ><vcchunga at amazon.com>
> > >> > > >
> > >> > >
> > >> > > Series Tested-by: Pavan Nikhilesh <pbhagavatula at marvell.com>
> > >> > >
> > >> > > Also, I believe we need to take a look at ARCH_SOC_ID support
> > >coming
> > >> > > up in
> > >> > > 5.9 kernel.
> > >> > >
> > >> > > https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F
> > >> > > %2
> > >> > > Furldefense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-
> > &data=04%7C
> > >> > >
> >
> 01%7Chemant.agrawal%40nxp.com%7Cf05a2623ec9347bd254708d8c2d4896
> > c%
> > >> > >
> > 7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C637473565383043785
> > %7
> > >> > >
> >
> CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJ
> B
> > Ti
> > >> > >
> >
> I6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=o7%2BXNileFgI2By08IHv
> > EY
> > >> > > PWrDjRikXlfFtal1MCdyCM%3D&reserved=0
> > >3A__lore.kernel.org_linux-2Darm-2Dkernel_20200506164411.3284-
> > >2D1-
> > >2D&d=DwIGaQ&c=nKjWec2b6R0mOyPaz7xtfQ&r=1cjuAHrGh745jHNmj
> > >2fD85sUMIJ2IPIDsIJzo6FN6Z0&m=A7kOn5DYMe_WZiykt8BVATgv7-
> > >qTVhfo4iSgHBg-
> > >BTo&s=dslFI0W1qZdDxkbXDMTzPSUzlKMFfMUsE6tllPaF9Oc&e=
> > >> > > sudeep.holla at arm.com/
> > >> >
> > >> > Agree, it can come after this series is merged.
> > >> >
> > >> > Note that it requires firmware with SMCCC v1.2 support in the
> > >> > underlying platform.
> > >> >
> > >>
> > >> I have a few questions to everyone:
> > >> Is there any comprehensive documentation about how to use this?
> > >Things
> > >> like where in Linux can I get the ARCH_SOC_ID and how are the ID's
> > >mapped
> > >> to socs?
> > >Kernel patch is at [1] - merged in Kernel 5.9 Makes use of SMCCC
> > >v1.2, SMCCC_ARCH_SOC_ID call.
> > >SMCCC v1.2 spec is at [2].
> > >
> > >The SoC ID becomes available at:
> > >/sys/devices/socX/family
> > >/sys/devices/socX/soc_id
> > >/sys/devices/socX/revision
> > >
> > >SOC_ID makes use of JEP-106 code [3] for the SiP which makes it unique.
> > >
> > >[1]
> > >https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Furl
> > >d
> > e
> > >fense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-
> > &data=04%7C01%7Chemant.
> >
> >agrawal%40nxp.com%7Cf05a2623ec9347bd254708d8c2d4896c%7C686ea1d
> > 3bc2b4c6f
> > >a92cd99c5c301635%7C0%7C0%7C637473565383043785%7CUnknown%7CT
> > WFpbGZsb3d8e
> >
> >yJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3
> > D%7C100
> >
> >0&sdata=o7%2BXNileFgI2By08IHvEYPWrDjRikXlfFtal1MCdyCM%3D&a
> > mp;reserv
> > >ed=0
> 3A__lkml.org_lkml_2020_6_25_208&d=DwIGaQ&c=nKjWec2b6R0mOy
> > >Paz7xtfQ&r=1cjuAHrGh745jHNmj2fD85sUMIJ2IPIDsIJzo6FN6Z0&m=A7k
> > >On5DYMe_WZiykt8BVATgv7-qTVhfo4iSgHBg-
> > >BTo&s=5xXdzD7DpbcDeG81JVGZf396EFNbV0rSl00hRLuZJBc&e=
> > >[2]
> > >https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Furl
> > >d
> > e
> > >fense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-
> > &data=04%7C01%7Chemant.
> >
> >agrawal%40nxp.com%7Cf05a2623ec9347bd254708d8c2d4896c%7C686ea1d
> > 3bc2b4c6f
> > >a92cd99c5c301635%7C0%7C0%7C637473565383043785%7CUnknown%7CT
> > WFpbGZsb3d8e
> >
> >yJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3
> > D%7C100
> >
> >0&sdata=o7%2BXNileFgI2By08IHvEYPWrDjRikXlfFtal1MCdyCM%3D&a
> > mp;reserv
> > >ed=0
> > 3A__developer.arm.com_documentation_den0028_c&d=DwIGaQ&c=n
> > >KjWec2b6R0mOyPaz7xtfQ&r=1cjuAHrGh745jHNmj2fD85sUMIJ2IPIDsIJz
> > >o6FN6Z0&m=A7kOn5DYMe_WZiykt8BVATgv7-qTVhfo4iSgHBg-
> > >BTo&s=DFG7a9oFDFczCcODfR0wVGjkLQXwMU19q_ogaFVw90I&e=
> > >[3]
> > >https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Furl
> > >d
> > e
> > >fense.proofpoint.com%2Fv2%2Furl%3Fu%3Dhttps-
> > &data=04%7C01%7Chemant.
> >
> >agrawal%40nxp.com%7Cf05a2623ec9347bd254708d8c2d4896c%7C686ea1d
> > 3bc2b4c6f
> > >a92cd99c5c301635%7C0%7C0%7C637473565383043785%7CUnknown%7CT
> > WFpbGZsb3d8e
> >
> >yJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3
> > D%7C100
> >
> >0&sdata=o7%2BXNileFgI2By08IHvEYPWrDjRikXlfFtal1MCdyCM%3D&a
> > mp;reserv
> > >ed=0
> > >3A__developer.arm.com_documentation_ka001301_1-
> > >2D0&d=DwIGaQ&c=nKjWec2b6R0mOyPaz7xtfQ&r=1cjuAHrGh745jHNm
> > >j2fD85sUMIJ2IPIDsIJzo6FN6Z0&m=A7kOn5DYMe_WZiykt8BVATgv7-
> > >qTVhfo4iSgHBg-
> > >BTo&s=GQsJJuft9k1iooO8VNNIM8ZUUgGjCEPB6eRxN3qm62U&e=
> > >
> > >> If the ID mappings are not public then the plaform owners would
> > >> need
> > >to
> > >> add the support, right?
> > >I agree here. There are firmware changes required and we do not know
> > >the IDs as well. It will be good for Marvell/NXP/NVIDIA to do this.
> > >Pavan, is this ok for you?
> >
> > Yeah, I believe we need fallback to the current mechanisms if SOC_ID
> > support is not present or platform owners have not updated them.
> 
> [Hemant]  +1
> 
Agree.
Just to be clear, my ask is, can any of you take this work as Arm will not have all the parts to test this?