But something like this inside of the external kernel module should work, right? If yes, then it could be moved to module.bbclass or something.
At least as long as the RPATH is set correctly, that worked for me with some old 4.9 kernels in: https://github.com/shr-distribution/linux/commit/409f7adf4a862e439b5bec04739b67713e08b8ba On Thu, Oct 26, 2023 at 10:35 AM Jose Quaresma <quaresma.j...@gmail.com> wrote: > > > Khem Raj <raj.k...@gmail.com> escreveu no dia quarta, 25/10/2023 à(s) > 18:18: > >> I wonder if something like below would help >> >> diff --git a/meta/recipes-kernel/make-mod-scripts/make-mod-scripts_1.0.bb >> b/meta/recipes-kernel/make-mod-scripts/make-mod-scripts_1.0.bb >> index 2cff48c39c3..6133721334c 100644 >> --- a/meta/recipes-kernel/make-mod-scripts/make-mod-scripts_1.0.bb >> +++ b/meta/recipes-kernel/make-mod-scripts/make-mod-scripts_1.0.bb >> @@ -12,7 +12,7 @@ S = "${WORKDIR}" >> >> do_configure[depends] += "virtual/kernel:do_shared_workdir >> openssl-native:do_populate_sysroot" >> do_compile[depends] += "virtual/kernel:do_compile_kernelmodules" >> - >> +do_populate_sysroot[depends] += "openssl-native:do_populate_sysroot" >> > > This doesn't solve the problem because at compile time the openssl is > already there on native sysroot > and this is the reason for the sign-file to be linked with openssl. > The make-mod-script doesn't install anything and only calls make on the > shared staging kernel dir > and the consequence is creating some binaries on the shared staging kernel > builddir. > One of the tools it builds is the sign-file linked dynamically with the > openssl in the native sysroot of > make-mod-script. At the end of make-mod-script the rm_work deletes the > native sysroot > and this will remove the openssl libcrypto.so.3. > Next time the sign-file in the shared staging kernel builddir will run the > libcrypto.so.3 will not be found. > Without using the rm_work class everything works as expected. > > Jose > > >> DEPENDS += "bc-native bison-native" >> DEPENDS += "gmp-native" >> >> On Wed, Oct 25, 2023 at 7:45 AM Bruce Ashfield <bruce.ashfi...@gmail.com> >> wrote: >> > >> > On Wed, Oct 25, 2023 at 7:40 AM Jose Quaresma <quaresma.j...@gmail.com> >> wrote: >> > > >> > > Hi Joerg, >> > > >> > > This topic has already been discussed previously in >> > > https://lists.openembedded.org/g/openembedded-core/topic/98296212 >> > > >> > > Bruce even sent a patch but it was never integrated, probably because >> it was a very invasive change. >> > > https://lists.openembedded.org/g/openembedded-core/message/182024 >> > >> > Heh. No, that wasn't the reason. The change wasn't invasive, it is >> > isolated to just the single recipe. >> > >> > I just didn't have the cycles to chase down host specific issues and >> > uninative updates. >> > >> > Clobbering rm_work isn't the right solution IMHO, but it can do the job. >> > >> > Bruce >> > >> > > >> > > Anyway you can solve this problem with: >> > > RM_WORK_EXCLUDE += "make-mod-scripts" >> > > >> > > Jose >> > > >> > > Jörg Sommer via lists.openembedded.org <joerg.sommer= >> navimatix...@lists.openembedded.org> escreveu no dia quarta, 25/10/2023 >> à(s) 09:57: >> > >> >> > >> Hi, >> > >> >> > >> I'm having an odd problem building the external kernel module >> > >> kernel-module-imx-gpu-viv [1]. I'm using kirkstone and have rm_work >> > >> enabled. >> > >> >> > >> [1] https://layers.openembedded.org/layerindex/recipe/40267/ >> > >> >> > >> The module builds fine, but the install step contains the module >> signing >> > >> which **silently** fails with *libcrypto.so.3* not found: >> > >> >> > >> ``` >> > >> | make -C /build/tmp/work-shared/ziborium/kernel-source >> > >> >> M=/build/tmp/work/machine-pyxis-linux/kernel-module-imx-gpu-viv/6.4.3.p4.2-r0/git/src >> > >> modules_install >> > >> | make[1]: Entering directory >> '/build/tmp/work-shared/machine/kernel-source' >> > >> | make -C /build/tmp/work-shared/machine/kernel-build-artifacts -f >> > >> /build/tmp/work-shared/machine/kernel-source/Makefile modules_install >> > >> | make[2]: Entering directory >> > >> '/build/tmp/work-shared/machine/kernel-build-artifacts' >> > >> | test -e include/generated/autoconf.h -a -e >> include/config/auto.conf || >> > >> ( \ >> > >> | echo >&2; \ >> > >> | echo >&2 " ERROR: Kernel configuration is invalid."; \ >> > >> | echo >&2 " include/generated/autoconf.h or >> > >> include/config/auto.conf are missing.";\ >> > >> | echo >&2 " Run 'make oldconfig && make prepare' on kernel >> src >> > >> to fix it."; \ >> > >> | echo >&2 ; \ >> > >> | /bin/false) >> > >> | make -f >> > >> /build/tmp/work-shared/machine/kernel-source/scripts/Makefile.modinst >> > >> | mkdir -p >> > >> >> /build/tmp/work/machine-dist-linux/kernel-module-imx-gpu-viv/6.4.3.p4.2-r0/image/usr/lib/modules/5.15.60-imx8mm+g343e81c0e39e/extra/; >> > >> cp >> > >> >> /build/tmp/work/machine-dist-linux/kernel-module-imx-gpu-viv/6.4.3.p4.2-r0/git/src/galcore.ko >> > >> >> /build/tmp/work/machine-dist-linux/kernel-module-imx-gpu-viv/6.4.3.p4.2-r0/image/usr/lib/modules/5.15.60-imx8mm+g343e81c0e39e/extra/galcore.ko >> > >> | : >> > >> | scripts/sign-file "sha1" "certs/signing_key.pem" >> > >> certs/signing_key.x509 >> > >> >> /build/tmp/work/machine-dist-linux/kernel-module-imx-gpu-viv/6.4.3.p4.2-r0/image/usr/lib/modules/5.15.60-imx8mm+g343e81c0e39e/extra/galcore.ko >> > >> || true >> > >> | scripts/sign-file: error while loading shared libraries: >> > >> libcrypto.so.3: cannot open shared object file: No such file or >> directory >> > >> | sh /build/tmp/work-shared/machine/kernel-source/scripts/depmod.sh >> > >> echo 5.15.60-imx8mm+g343e81c0e39e >> > >> | Warning: modules_install: missing 'System.map' file. Skipping >> depmod. >> > >> | make[2]: Leaving directory >> > >> '/build/tmp/work-shared/machine/kernel-build-artifacts' >> > >> | make[1]: Leaving directory >> '/build/tmp/work-shared/machine/kernel-source' >> > >> ``` >> > >> >> > >> I noticed that there are special dependency definitions in >> make-mod-scripts >> > >> >> > >> ``` >> > >> do_configure[depends] += "virtual/kernel:do_shared_workdir >> > >> openssl-native:do_populate_sysroot" >> > >> do_compile[depends] += "virtual/kernel:do_compile_kernelmodules" >> > >> ``` >> > >> >> > >> and module-base.bbclass (inherited by module.bbclass) >> > >> >> > >> ``` >> > >> # We do the dependency this way because the output is not preserved >> > >> # in sstate, so we must force do_compile to run (once). >> > >> do_configure[depends] += "make-mod-scripts:do_compile" >> > >> ``` >> > >> >> > >> Is there an additional dependency needed to address rm_work? >> > >> >> > >> Kind regards >> > >> >> > >> Jörg Sommer >> > >> -- >> > >> Navimatix GmbH >> > >> Tatzendpromenade 2 >> > >> D-07745 Jena >> > >> Geschäftsführer: Steffen Späthe, Jan Rommeley >> > >> Registergericht: Amtsgericht Jena, HRB 501480 >> > >> >> > >> >> > >> >> > > >> > > >> > > -- >> > > Best regards, >> > > >> > > José Quaresma >> > >> > >> > >> > -- >> > - Thou shalt not follow the NULL pointer, for chaos and madness await >> > thee at its end >> > - "Use the force Harry" - Gandalf, Star Trek II >> > >> > >> > >> > > > -- > Best regards, > > José Quaresma > > > >
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#105732): https://lists.openembedded.org/g/openembedded-devel/message/105732 Mute This Topic: https://lists.openembedded.org/mt/102174355/21656 Group Owner: openembedded-devel+ow...@lists.openembedded.org Unsubscribe: https://lists.openembedded.org/g/openembedded-devel/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-
