From: Henning Schild <henning.schild@siemens.com>
To: "Hombourger, Cedric" <Cedric_Hombourger@mentor.com>,
"Kiszka, Jan (CT RDA IOT SES-DE)" <jan.kiszka@siemens.com>
Cc: "isar-users@googlegroups.com" <isar-users@googlegroups.com>
Subject: Re: [PATCH] linux-custom: repack linux-libc-dev
Date: Thu, 13 Jun 2019 18:09:28 +0200 [thread overview]
Message-ID: <20190613180928.358b01fd@md1za8fc.ad001.siemens.net> (raw)
In-Reply-To: <20190613173733.483cdb98@md1za8fc.ad001.siemens.net>
Am Thu, 13 Jun 2019 17:37:33 +0200
schrieb "[ext] Henning Schild" <henning.schild@siemens.com>:
> Am Thu, 13 Jun 2019 08:32:26 +0000
> schrieb "Hombourger, Cedric" <Cedric_Hombourger@mentor.com>:
>
> > Hi Henning
> >
> > Agreed but I would think we would need a work-around in Isar until
> > the proposed change gets accepted by k.org folks (on a different
> > note, I also need to submit a patch to arch/mips/Makefile for
> > builddeb to succeed - my linux-custom recipe would not build without
> > a patch as dtbs don't get built by default while they do on e.g.
> > aarch64)
> >
> > I would therefore suggest to have build-kernel.sh do something like:
> >
> > if [ -d <...>/usr/include/asm ]; then
> > echo "warning: your kernel has incorrectly packaged
> > linux-libc-dev headers" >&2 # apply work-around and repack
> > linux-libc-dev ...
> > mv <...>/usr/include/asm <...>/usr/include/${LIBC_HEADERS_DIR}/
> > ...
> > fi
>
> I guess a backport or such a hack would not be a big deal, after
> upstream accepted your patch. k.org is probably faster than ilbers
>
> Lets at least see what the discussion brings and how fast it starts.
> Maybe upstream will aknowledge the bug right away. Or they might give
> valuable feedback on what we do not understand. ... and should not be
> putting into Isar on the fast-lane
Thinking about it again. The whole repacking should maybe be replaced
with patching the original packing code. That would make clearer what
the changes to the kernel actually are.
If patches do not apply anymore on changing kernels, we would have a
clear warning to review that again.
so we would carry meta/recipes-kernel/linux/files/
linux-image-depends.patch
linux-image-aarch64-uncompressed_4.19.patch
linux-image-aarch64-uncompressed_4.4.patch
...
All these would be SRC_URI += ;apply=no from the inc, which would also
get a new magic patch function deciding which ones to apply depending
on PV.
The result would also be a kernel tree that you could actually
"builddeb" outside of Isar, and skipping the repack might save time.
Henning
> Henning
>
> > I can submit a v2 if the approach is agreeable
> > will roll-up my sleeves a prepare a patch for k.org
> >
> > Cedric
> > ________________________________________
> > From: Henning Schild <henning.schild@siemens.com>
> > Sent: Thursday, June 13, 2019 10:10 AM
> > To: Hombourger, Cedric
> > Cc: isar-users@googlegroups.com
> > Subject: Re: [PATCH] linux-custom: repack linux-libc-dev
> >
> > Am Wed, 12 Jun 2019 22:10:02 +0200
> > schrieb Cedric Hombourger <Cedric_Hombourger@mentor.com>:
> >
> > > Debian's linux-libc-dev package has "asm" headers in the following
> > > directory: /usr/include/<libc-arch>/asm while the builddeb script
> > > from the Linux kernel places them in: /usr/include/asm. Repack the
> > > linux-libc-dev package (fixing a build error for packages built
> > > against Isar-created distributions shipping a custom kernel in
> > > lieu of Debian's).
> >
> > To me that sounds like an upstream bug that should not just get
> > covered up by repacking. I would propose looking into a kernel patch
> > for the "debpkg" step. That patch could be carried in your layer(s)
> > and Isar while it is pending upstream.
> >
> > Looking at both ubuntu and debian, old and new .. it should never
> > be /usr/include/asm/.
> >
> > > Signed-off-by: Cedric Hombourger <Cedric_Hombourger@mentor.com>
> > > ---
> > > meta/recipes-kernel/linux/files/build-kernel.sh | 38
> > > +++++++++++++++++++++++-- 1 file changed, 35 insertions(+), 3
> > > deletions(-)
> > >
> > > diff --git a/meta/recipes-kernel/linux/files/build-kernel.sh
> > > b/meta/recipes-kernel/linux/files/build-kernel.sh index
> > > 8b7b23b..2c97932 100644 ---
> > > a/meta/recipes-kernel/linux/files/build-kernel.sh +++
> > > b/meta/recipes-kernel/linux/files/build-kernel.sh @@ -9,8 +9,20 @@
> > >
> > > source /isar/common.sh
> > >
> > > -host_arch=$(dpkg --print-architecture)
> > > +unsupported_arch() {
> > > + echo "error: unsupported architecture ($1)" >&2
> > > + exit 1
> > > +}
> > > +
> > > +case $target_arch in
> > > + amd64) LIBC_HEADERS_DIR="x86_64-linux-gnu" ;;
> > > + armhf) LIBC_HEADERS_DIR="arm-linux-gnueabihf" ;;
> > > + arm64) LIBC_HEADERS_DIR="aarch64-linux-gnu" ;;
> > > + mipsel) LIBC_HEADERS_DIR="mipsel-linux-gnu" ;;
> > > + *) unsupported_arch "$target_arch" ;;
> > > +esac
> >
> > Looks suspiciously like "${CROSS-COMPILE}gcc -dumpmachine". Using
> > that would relieve us from carrying those magic strings around.
> >
> >
> > > +host_arch=$(dpkg --print-architecture)
> > > if [ "$host_arch" != "$target_arch" ]; then
> > > case $target_arch in
> > > armhf)
> > > @@ -26,8 +38,7 @@ if [ "$host_arch" != "$target_arch" ]; then
> > > export CROSS_COMPILE="mipsel-linux-gnu-"
> > > ;;
> > > *)
> > > - echo "error: unsupported architecture ($target_arch)"
> > > - exit 1
> > > + unsupported_arch "$target_arch"
> > > ;;
> > > esac
> > > fi
> > > @@ -35,6 +46,7 @@ fi
> > > REPACK_DIR="$1/../repack"
> > > REPACK_LINUX_IMAGE_DIR="${REPACK_DIR}/linux-image"
> > > REPACK_LINUX_HEADERS_DIR="${REPACK_DIR}/linux-headers"
> > > +REPACK_LIBC_HEADERS_DIR="${REPACK_DIR}/libc-headers"
> > >
> > > if [ -e .config ]; then
> > > make olddefconfig
> > > @@ -56,6 +68,7 @@ rm -rf "${REPACK_DIR}"
> > > mkdir -p "${REPACK_DIR}"
> > > mkdir -p "${REPACK_LINUX_IMAGE_DIR}"
> > > mkdir -p "${REPACK_LINUX_HEADERS_DIR}"
> > > +mkdir -p "${REPACK_LIBC_HEADERS_DIR}"
> > >
> > > cp -a debian "${REPACK_DIR}"
> > >
> > > @@ -67,6 +80,7 @@ cd ..
> > >
> > > dpkg-deb -R linux-image-${PV}_${PV}-1_*.deb
> > > "${REPACK_LINUX_IMAGE_DIR}" dpkg-deb -R
> > > linux-headers-${PV}_${PV}-1_*.deb "${REPACK_LINUX_HEADERS_DIR}"
> > > +dpkg-deb -R linux-libc-dev_${PV}-1_*.deb
> > > "${REPACK_LIBC_HEADERS_DIR}" dpkg-gencontrol
> > > -crepack/debian/control \ -lrepack/debian/changelog \
> > > @@ -110,6 +124,10 @@ if [ "$target_arch" = "arm64" ]; then
> > > gunzip "$vmlinuz.gz"
> > > fi
> > >
> > > +# Move libc's asm headers where they are expected
> > > +mkdir ${REPACK_LIBC_HEADERS_DIR}/usr/include/${LIBC_HEADERS_DIR}
> > > +mv ${REPACK_LIBC_HEADERS_DIR}/usr/include/asm
> > > ${REPACK_LIBC_HEADERS_DIR}/usr/include/${LIBC_HEADERS_DIR} +
> > > dpkg-gencontrol -crepack/debian/control \
> > > -lrepack/debian/changelog \
> > > -frepack/debian/files \
> > > @@ -121,12 +139,26 @@ dpkg-gencontrol -crepack/debian/control \
> > > -DDepends="${KERNEL_HEADERS_DEBIAN_DEPENDS}" \
> > > -DArchitecture=$target_arch
> > >
> > > +dpkg-gencontrol -crepack/debian/control \
> > > + -lrepack/debian/changelog \
> > > + -frepack/debian/files \
> > > + -plinux-libc-dev \
> > > + -P"${REPACK_LIBC_HEADERS_DIR}" \
> > > + -Vkernel:debarch="${KERNEL_NAME}" \
> > > + -DPackage="linux-libc-dev" \
> > > + -DSection=kernel \
> > > + -DDepends="${LIBC_HEADERS_DEBIAN_DEPENDS}" \
> > > + -DArchitecture=$target_arch
> > > +
> > > fakeroot dpkg-deb -b "${REPACK_LINUX_IMAGE_DIR}" \
> > > linux-image-${KERNEL_NAME}_${PV}-1_${KERNEL_NAME}.deb
> > > rm -f linux-image-${PV}_${PV}-1_*.deb
> > > fakeroot dpkg-deb -b "${REPACK_LINUX_HEADERS_DIR}" \
> > > linux-headers-${KERNEL_NAME}_${PV}-1_${KERNEL_NAME}.deb
> > > rm -f linux-headers-${PV}_${PV}-1_*.deb
> > > +rm -f linux-libc-dev_${PV}-1_*.deb
> > > +fakeroot dpkg-deb -b "${REPACK_LIBC_HEADERS_DIR}" \
> > > + linux-libc-dev_${PV}-1_${KERNEL_NAME}.deb
> > >
> > > # linux-libc-dev causes dependency problems if we downgrade
> > > # remove it after the build so the downgraded version does not
> > > get deployed
> >
> > Ahh this guy is still somehow in the patch. You might want to have a
> > look at 7a1f14ca. My guess is that skipping the downgrade is still
> > OK because the problem just shows once you install the libc-dev
> > package.
>
next prev parent reply other threads:[~2019-06-13 16:09 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-06-12 20:10 Cedric Hombourger
2019-06-13 8:10 ` Henning Schild
2019-06-13 8:32 ` Hombourger, Cedric
2019-06-13 15:37 ` Henning Schild
2019-06-13 16:09 ` Henning Schild [this message]
2019-06-13 16:05 ` Baurzhan Ismagulov
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20190613180928.358b01fd@md1za8fc.ad001.siemens.net \
--to=henning.schild@siemens.com \
--cc=Cedric_Hombourger@mentor.com \
--cc=isar-users@googlegroups.com \
--cc=jan.kiszka@siemens.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox