| Message ID | 363ef9b2-335b-44ed-b2e9-6df662bc0452@siemens.com |
|---|---|
| State | Accepted, archived |
| Headers | show |
| Series | sdk: Fix build with custom DISTRO when host==target | expand |
On Thu, 2024-03-14 at 08:54 +0100, 'Jan Kiszka' via isar-users wrote: > From: Jan Kiszka <jan.kiszka@siemens.com> > > HOST_DISTRO is not always the right choice for building an SDK. If the > host arch is identical to the target and DISTRO was customized, there is > no HOST_DISTRO to get the bootstrap from. Account for that. > > This issue could have been reproduced e.g by xenomai-images when > selecting an x86 target and requesting to build the SDK for it. > > Reported-by: Fabian Scheler <fabian.scheler@siemens.com> > Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com> > --- [snip] Is there no test in place that covers this case? The SDK was broken a couple of times now. Maybe it's time to increase the test coverage. Florian
On Fri, 2024-03-15 at 09:42 +0100, 'Florian Bezdeka' via isar-users wrote: > On Thu, 2024-03-14 at 08:54 +0100, 'Jan Kiszka' via isar-users wrote: > > From: Jan Kiszka <jan.kiszka@siemens.com> > > > > HOST_DISTRO is not always the right choice for building an SDK. If > > the > > host arch is identical to the target and DISTRO was customized, > > there is > > no HOST_DISTRO to get the bootstrap from. Account for that. > > > > This issue could have been reproduced e.g by xenomai-images when > > selecting an x86 target and requesting to build the SDK for it. > > > > Reported-by: Fabian Scheler <fabian.scheler@siemens.com> > > Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com> > > --- > [snip] > > Is there no test in place that covers this case? The SDK was broken a > couple of times now. Maybe it's time to increase the test coverage. > > Florian > Hello everyone. In order not to delay merging, we are ready to apply this to next. Possible CI tests for sdk could be added later. Best regards, Uladzimir.
15/03/2024 10:42, 'Florian Bezdeka' via isar-users wrote: > [snip] > > Is there no test in place that covers this case? The SDK was broken a > couple of times now. Maybe it's time to increase the test coverage. > > Florian > Is anyone interested in helping create such a test case?
On 02.04.24 16:58, Anton Mikanovich wrote: > 15/03/2024 10:42, 'Florian Bezdeka' via isar-users wrote: >> [snip] >> >> Is there no test in place that covers this case? The SDK was broken a >> couple of times now. Maybe it's time to increase the test coverage. >> >> Florian >> > Is anyone interested in helping create such a test case? > This is not a trivial thing, given that we have no pattern yet to model downstream layers (with custom ) in the testsuite. In any case, this should not delay merging this fix. Jan
On Thu, 2024-03-14 at 08:54 +0100, 'Jan Kiszka' via isar-users wrote: > From: Jan Kiszka <jan.kiszka@siemens.com> > > HOST_DISTRO is not always the right choice for building an SDK. If > the > host arch is identical to the target and DISTRO was customized, there > is > no HOST_DISTRO to get the bootstrap from. Account for that. > > This issue could have been reproduced e.g by xenomai-images when > selecting an x86 target and requesting to build the SDK for it. > > Reported-by: Fabian Scheler <fabian.scheler@siemens.com> > Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com> > --- > meta/classes/sdk.bbclass | 10 +++++++++- > 1 file changed, 9 insertions(+), 1 deletion(-) > > diff --git a/meta/classes/sdk.bbclass b/meta/classes/sdk.bbclass > index 74c0acb9..71db6f3a 100644 > --- a/meta/classes/sdk.bbclass > +++ b/meta/classes/sdk.bbclass > @@ -44,9 +44,17 @@ SDK_PREINSTALL += " \ > devscripts \ > equivs" > > +def get_rootfs_distro(d): > + host_arch = d.getVar('HOST_ARCH') > + distro_arch = d.getVar('DISTRO_ARCH') > + if host_arch == distro_arch: > + return d.getVar('DISTRO') > + else: > + return d.getVar('HOST_DISTRO') > + > # rootfs/image overrides for the SDK > ROOTFS_ARCH:class-sdk = "${HOST_ARCH}" > -ROOTFS_DISTRO:class-sdk = "${HOST_DISTRO}" > +ROOTFS_DISTRO:class-sdk = "${@get_rootfs_distro(d)}" > ROOTFS_PACKAGES:class-sdk = "sdk-files ${SDK_TOOLCHAIN} > ${SDK_PREINSTALL} ${@isar_multiarch_packages('SDK_INSTALL', d)}" > ROOTFS_FEATURES:append:class-sdk = " clean-package-cache generate- > manifest export-dpkg-status" > ROOTFS_MANIFEST_DEPLOY_DIR:class-sdk = "${DEPLOY_DIR_SDKCHROOT}" > -- > 2.35.3 > Applied to next, thanks.
diff --git a/meta/classes/sdk.bbclass b/meta/classes/sdk.bbclass index 74c0acb9..71db6f3a 100644 --- a/meta/classes/sdk.bbclass +++ b/meta/classes/sdk.bbclass @@ -44,9 +44,17 @@ SDK_PREINSTALL += " \ devscripts \ equivs" +def get_rootfs_distro(d): + host_arch = d.getVar('HOST_ARCH') + distro_arch = d.getVar('DISTRO_ARCH') + if host_arch == distro_arch: + return d.getVar('DISTRO') + else: + return d.getVar('HOST_DISTRO') + # rootfs/image overrides for the SDK ROOTFS_ARCH:class-sdk = "${HOST_ARCH}" -ROOTFS_DISTRO:class-sdk = "${HOST_DISTRO}" +ROOTFS_DISTRO:class-sdk = "${@get_rootfs_distro(d)}" ROOTFS_PACKAGES:class-sdk = "sdk-files ${SDK_TOOLCHAIN} ${SDK_PREINSTALL} ${@isar_multiarch_packages('SDK_INSTALL', d)}" ROOTFS_FEATURES:append:class-sdk = " clean-package-cache generate-manifest export-dpkg-status" ROOTFS_MANIFEST_DEPLOY_DIR:class-sdk = "${DEPLOY_DIR_SDKCHROOT}"