When same kernel tree is built with gcc 5.3.1 from Xenial, the generated modlues
can't be loaded any more and '-mcmodel=large' is still passed to gcc
during kernel
building, so it looks like a compiler bug:

ubuntu@ubuntu:~/git$ gcc -v
Using built-in specs.
COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=/usr/lib/gcc/aarch64-linux-gnu/5/lto-wrapper
Target: aarch64-linux-gnu
Configured with: ../src/configure -v --with-pkgversion='Ubuntu/Linaro
5.3.1-5ubuntu2' --with-bugurl=file:///usr/share/doc/gcc-5/README.Bugs
--enable-languages=c,ada,c++,java,go,d,fortran,objc,obj-c++
--prefix=/usr --program-suffix=-5 --enable-shared
--enable-linker-build-id --libexecdir=/usr/lib
--without-included-gettext --enable-threads=posix --libdir=/usr/lib
--enable-nls --with-sysroot=/ --enable-clocale=gnu
--enable-libstdcxx-debug --enable-libstdcxx-time=yes
--with-default-libstdcxx-abi=new --enable-gnu-unique-object
--disable-libquadmath --enable-plugin --with-system-zlib
--disable-browser-plugin --enable-java-awt=gtk --enable-gtk-cairo
--with-java-home=/usr/lib/jvm/java-1.5.0-gcj-5-arm64/jre
--enable-java-home
--with-jvm-root-dir=/usr/lib/jvm/java-1.5.0-gcj-5-arm64
--with-jvm-jar-dir=/usr/lib/jvm-exports/java-1.5.0-gcj-5-arm64
--with-arch-directory=aarch64
--with-ecj-jar=/usr/share/java/eclipse-ecj.jar --enable-multiarch
--disable-werror --enable-checking=release --build=aarch64-linux-gnu
--host=aarch64-linux-gnu --target=aarch64-linux-gnu
Thread model: posix
gcc version 5.3.1 20160108 (Ubuntu/Linaro 5.3.1-5ubuntu2)



On Wed, Jan 13, 2016 at 9:11 AM, Ming Lei <email address hidden> wrote:
> When I built 4.3.0-7-generic on arm64(mustang) Wily with the following steps,
>
>          fakeroot debian/rules clean
>          fakeroot debian/rules binary-generic
>
> by this compiler:
>
> ubuntu@ubuntu:~$ gcc -v
> Using built-in specs.
> COLLECT_GCC=gcc
> COLLECT_LTO_WRAPPER=/usr/lib/gcc/aarch64-linux-gnu/5/lto-wrapper
> Target: aarch64-linux-gnu
> Configured with: ../src/configure -v --with-pkgversion='Ubuntu
> 5.2.1-22ubuntu2' --with-bugurl=file:///usr/share/doc/gcc-5/README.Bugs
> --enable-languages=c,ada,c++,java,go,d,fortran,objc,obj-c++
> --prefix=/usr --program-suffix=-5 --enable-shared
> --enable-linker-build-id --libexecdir=/usr/lib
> --without-included-gettext --enable-threads=posix --libdir=/usr/lib
> --enable-nls --with-sysroot=/ --enable-clocale=gnu
> --enable-libstdcxx-debug --enable-libstdcxx-time=yes
> --with-default-libstdcxx-abi=new --enable-gnu-unique-object
> --disable-libquadmath --enable-plugin --with-system-zlib
> --disable-browser-plugin --enable-java-awt=gtk --enable-gtk-cairo
> --with-java-home=/usr/lib/jvm/java-1.5.0-gcj-5-arm64/jre
> --enable-java-home
> --with-jvm-root-dir=/usr/lib/jvm/java-1.5.0-gcj-5-arm64
> --with-jvm-jar-dir=/usr/lib/jvm-exports/java-1.5.0-gcj-5-arm64
> --with-arch-directory=aarch64
> --with-ecj-jar=/usr/share/java/eclipse-ecj.jar --enable-multiarch
> --disable-werror --enable-checking=release --build=aarch64-linux-gnu
> --host=aarch64-linux-gnu --target=aarch64-linux-gnu
>
>
> the built image just works well:
>
>       ubuntu@ubuntu:~$ uname -a
>       Linux ubuntu 4.3.0-7-generic #18 SMP Tue Jan 12 10:19:24 EST 2016
>       aarch64 aarch64 aarch64 GNU/Linux
>
>
> But I can see the issue when booting a kernel from the following package[1]:
>
>          http://launchpadlibrarian.net/230287220/linux-image-4.3.0-5-generic_4.3.0-5.16_arm64.deb
>
> It is a build environment issue instead of kernel issue, please see
> comments from Ard Biesheuvel <email address hidden> [2]:
>
>         RELA #275 is the relocation against ADRP instructions, which GCC
>         should not emit anymore when -mcmodel=large is in effect.
>
>         Can you confirm that the modules have been rebuilt with this config as
>         well? Can you double check the GCC command line (with V=1) when doing
>         'make modules' to ensure that '-mcmodel=large' is being passed? Can
>         you check with 'readelf -r' which objects still contain
>         R_AARCH64_ADR_PREL_PG_HI21 relocations?
>
> I have checked the gcc flag in my building environment, and '-mcmodel=large' is
> passed, and CONFIG_ARM64_ERRATUM_843419 is enabled too.
>
> I don't know how the image in [1] is built, so could anyone check the build
> environment for this building?
>
>
> [1] http://launchpadlibrarian.net/230287220/linux-image-4.3.0-5-generic_4.3.0-5.16_arm64.deb
> [2] http://www.spinics.net/lists/arm-kernel/msg449991.html
>
>
> On Tue, Jan 12, 2016 at 7:47 PM, Raghuram Kota
> <email address hidden> wrote:
>> ** Tags added: hs-arm64
>>
>> --
>> You received this bug notification because you are subscribed to linux
>> in Ubuntu.
>> https://bugs.launchpad.net/bugs/1533009
>>
>> Title:
>>   arm64: "unsupported RELA relocation"
>>
>> Status in linux package in Ubuntu:
>>   Confirmed
>>
>> Bug description:
>>   linux-image-4.3.0-5-generic 4.3.0-5.16 arm64 fails to load modules
>>   (and therefore boot). It emits messages like the following for each
>>   attempted module load:
>>
>>   [    2.156817] module libahci: unsupported RELA relocation: 275
>>
>>   This is reminiscent of LP: #1502946 - except that fix appears to still
>>   be in-tact. What has changed, however, is the build environment. If I
>>   rebuild the same kernel source in a wily chroot, it boots fine.
>>
>>   Marking "Confirmed" because Paulo Pisatti reported this to me, and I
>>   reproduced.
>>
>> To manage notifications about this bug go to:
>> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1533009/+subscriptions