Date: Mon, 8 Sep 2025 13:45:13 -0600 From: Warner Losh <imp@bsdimp.com> To: Mark Johnston <markj@freebsd.org> Cc: Michael Tuexen <tuexen@freebsd.org>, "Bjoern A. Zeeb" <bzeeb-lists@lists.zabbadoz.net>, srcmgr@freebsd.org, "src-committers@freebsd.org" <src-committers@freebsd.org>, "dev-commits-src-all@freebsd.org" <dev-commits-src-all@freebsd.org>, "dev-commits-src-main@freebsd.org" <dev-commits-src-main@freebsd.org> Subject: Re: NOIP/NOINET/... for arm64 too? [was: Re: git: 1c23d8f9f398 - main - vtnet: improve checksum offloading] Message-ID: <CANCZdfr9=WW6eP1H2iJ-BzTntZKLLzpbxSK8FdyrLeRWnQZGrg@mail.gmail.com> In-Reply-To: <aL7Re2OBn4eTbMpO@nuc> References: <202509041126.584BQBBt067763@gitrepo.freebsd.org> <23rq9r12-75nn-sqs8-06o7-p785s2q915o3@yvfgf.mnoonqbm.arg> <8A4EFC13-ABA3-471B-9726-5940E2B88DED@FreeBSD.org> <0p6ns5r7-ss8q-n81s-s83n-6n16o6pr37p2@yvfgf.mnoonqbm.arg> <C03E215B-121D-428E-87FB-E2B55CDB700B@FreeBSD.org> <aL7Re2OBn4eTbMpO@nuc>
next in thread | previous in thread | raw e-mail | index | archive | help
--000000000000cbeb3c063e4f6d35 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable On Mon, Sep 8, 2025 at 6:52=E2=80=AFAM Mark Johnston <markj@freebsd.org> wr= ote: > On Sat, Sep 06, 2025 at 02:59:25PM +0200, Michael Tuexen wrote: > > > On 6. Sep 2025, at 10:31, Bjoern A. Zeeb < > bzeeb-lists@lists.zabbadoz.net> wrote: > > > > > > On Sat, 6 Sep 2025, Michael Tuexen wrote: > > > > > >> OK, that is fixed in > > >> > https://cgit.FreeBSD.org/src/commit/?id=3D7c448f5aab8be6977d0860e608e7d49= 7b495d28c > < > https://cgit.freebsd.org/src/commit/?id=3D7c448f5aab8be6977d0860e608e7d49= 7b495d28c > > > > > > > > Thanks! > > > > > >> Thanks again for reporting and I am sorry for breaking the NOIP > (again). > > >> BTW: Is there a reason we don't have the NOINET, NOINET6, and NOIP > kernel > > >> configs for arm64? > > > > > > I think initially we only added all the extra kernel configs 10-15 > years > > > ago to amd64 as that was the fastest way to build a universe (for a > > > target). I am likely guilty for NOINET and VIMAGE and NOIP... > > > initially generating them if I do not misremember. > > > > > > Adding them to a 2nd arch would increase full make universe build > times but > > > given a lot of people have started to do native builds on higher core > arm64 > > > machines I wonder if it would be time to add them there? > > Does adding a kernel config also requires it to activate it in make > universe? > > arm64 is now tier one and I use arm64 platforms more often then amd64 f= or > > development. > > I believe adding a new config to sys/arm64/conf will automatically cause > it to be included in universe builds. > > > > I'll Cc: source manager and see how they feel about this... > > I don't see any particular problem with building the extra kernels on > arm64 as well. It might be useful to build-test NOIP etc. > configurations with "options FDT". > > But even if one is using arm64 for development, it's trivial to build > the corresponding amd64 kernels when build-testing, just set TARGET. > I agree. We should just add them to arm64. It's part of being Tier 1. And while we can test them today, I'd like to see them explicitly tested since we're still breaking these configurations too often. Warner --000000000000cbeb3c063e4f6d35 Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable <div dir=3D"ltr"><div dir=3D"ltr"><br></div><br><div class=3D"gmail_quote g= mail_quote_container"><div dir=3D"ltr" class=3D"gmail_attr">On Mon, Sep 8, = 2025 at 6:52=E2=80=AFAM Mark Johnston <<a href=3D"mailto:markj@freebsd.o= rg">markj@freebsd.org</a>> wrote:<br></div><blockquote class=3D"gmail_qu= ote" style=3D"margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,20= 4);padding-left:1ex">On Sat, Sep 06, 2025 at 02:59:25PM +0200, Michael Tuex= en wrote:<br> > > On 6. Sep 2025, at 10:31, Bjoern A. Zeeb <<a href=3D"mailto:bz= eeb-lists@lists.zabbadoz.net" target=3D"_blank">bzeeb-lists@lists.zabbadoz.= net</a>> wrote:<br> > > <br> > > On Sat, 6 Sep 2025, Michael Tuexen wrote:<br> > > <br> > >> OK, that is fixed in<br> > >> <a href=3D"https://cgit.FreeBSD.org/src/commit/?id=3D7c448f5a= ab8be6977d0860e608e7d497b495d28c" rel=3D"noreferrer" target=3D"_blank">http= s://cgit.FreeBSD.org/src/commit/?id=3D7c448f5aab8be6977d0860e608e7d497b495d= 28c</a> <<a href=3D"https://cgit.freebsd.org/src/commit/?id=3D7c448f5aab= 8be6977d0860e608e7d497b495d28c" rel=3D"noreferrer" target=3D"_blank">https:= //cgit.freebsd.org/src/commit/?id=3D7c448f5aab8be6977d0860e608e7d497b495d28= c</a>><br> > > <br> > > Thanks!<br> > > <br> > >> Thanks again for reporting and I am sorry for breaking the NO= IP (again).<br> > >> BTW: Is there a reason we don't have the NOINET, NOINET6,= and NOIP kernel<br> > >> configs for arm64?<br> > > <br> > > I think initially we only added all the extra kernel configs 10-1= 5 years<br> > > ago to amd64 as that was the fastest way to build a universe (for= a<br> > > target).=C2=A0 =C2=A0I am likely guilty for NOINET and VIMAGE and= NOIP...<br> > > initially generating them if I do not misremember.<br> > > <br> > > Adding them to a 2nd arch would increase full make universe build= times but<br> > > given a lot of people have started to do native builds on higher = core arm64<br> > > machines I wonder if it would be time to add them there?<br> > Does adding a kernel config also requires it to activate it in make un= iverse?<br> > arm64 is now tier one and I use arm64 platforms more often then amd64 = for<br> > development.<br> <br> I believe adding a new config to sys/arm64/conf will automatically cause<br= > it to be included in universe builds.<br> <br> > > I'll Cc: source manager and see how they feel about this...<b= r> <br> I don't see any particular problem with building the extra kernels on<b= r> arm64 as well.=C2=A0 It might be useful to build-test NOIP etc.<br> configurations with "options FDT".<br> <br> But even if one is using arm64 for development, it's trivial to build<b= r> the corresponding amd64 kernels when build-testing, just set TARGET.<br></b= lockquote><div><br></div><div>I agree. We should just add them to arm64. It= 's part of being Tier 1. And while we can test them today, I'd like= to see them explicitly tested since we're still breaking these configu= rations too often.</div><div><br></div><div>Warner</div></div></div> --000000000000cbeb3c063e4f6d35--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CANCZdfr9=WW6eP1H2iJ-BzTntZKLLzpbxSK8FdyrLeRWnQZGrg>