On Fri, 29 Apr 2022 at 06:51, leimao...@fujitsu.com <leimao...@fujitsu.com> wrote: > Sorry, It seems that I should add an explain in commit message. > I found that if enable fips, a binary build for target will be executed in > do_compile task. The log(MACHINE ?= "qemuarm64") is as the following: > ------------------------------------------------------ > | ./fipshmac .libs/libgnutls.so.30 > .libs/.libgnutls.so.30.hmac-t && mv > .libs/.libgnutls.so.30.hmac-t .libs/.libgnutls.so.30.hmac > | libtool: link: (cd ".libs" && rm -f "libgnutlsxx.so.30" && ln -s > "libgnutlsxx.so.30.0.0" "libgnutlsxx.so.30") > | libtool: link: (cd ".libs" && rm -f "libgnutlsxx.so" && ln -s > "libgnutlsxx.so.30.0.0" "libgnutlsxx.so") > | /lib/ld-linux-aarch64.so.1: No such file or directory > > So, I tweaked do_compile/do_install to use fipshmac from native instead of > target. If there is better way, please tell me. > And if there are no questions, I'll send a v2 patch with an issue fixed for > multilib.
Can you please show the fragment in the Makefile where the target binary is executed? It is better to make a real patch that changes it to run a native binary, instead of changing it with sed. (sed patching is problematic because it's not clear from the recipe what is being done exactly, and how the original and the changed code looks like). Alex
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#164996): https://lists.openembedded.org/g/openembedded-core/message/164996 Mute This Topic: https://lists.openembedded.org/mt/90748530/21656 Group Owner: openembedded-core+ow...@lists.openembedded.org Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-