[colm] colm-0.13.0.5 fails to build on IA32

Jan Engelhardt jengelh at inai.de
Tue Jun 13 21:24:57 UTC 2017


On Tuesday 2017-06-13 00:58, Adrian Thurston wrote:

>Hmm I tried to reproduce this the other day on a fresh Ubuntu VM but couldn't.  Can you give more
>platform information?

It seems to only happen within KVM. I have a image files ready if you
want to try. These are image extracts from the OBS (which means they
don't have all the software). These may be found at
http://inai.de/files/colm/ .


$ ./COLM-start 
[    1.206078] dracut-pre-udev[175]: modprobe: FATAL: Module kqemu not found in directory /lib/modules/4.11.3-1-default
[    1.213907] dracut-pre-udev[175]: modprobe: FATAL: Module ext2 not found in directory /lib/modules/4.11.3-1-default
[    1.217805] dracut-pre-udev[175]: modprobe: FATAL: Module ext3 not found in directory /lib/modules/4.11.3-1-default
[    2.051420] dracut-pre-udev[175]: modprobe: FATAL: Module virtio_mmio not found in directory /lib/modules/4.11.3-1-default
[    2.065650] dracut-pre-udev[175]: modprobe: FATAL: Module ibmvscsi not found in directory /lib/modules/4.11.3-1-default
[    2.421544] dracut-pre-udev[175]: modprobe: FATAL: Module ibmveth not found in directory /lib/modules/4.11.3-1-default
sh-4.4# su - abuild
abuild at localhost:~> cd rpmbuild/BUILD/colm-0.13.0.5/src/
abuild at localhost:~/rpmbuild/BUILD/colm-0.13.0.5/src> make
make  all-am
make[1]: Entering directory '/home/abuild/rpmbuild/BUILD/colm-0.13.0.5/src'
mkdir -p gen
./bootstrap1 -c -o gen/parse2.c -e gen/if2.h -x gen/if2.cc colm.lm
bootstrap1: tree.c:953: colm_tree_downref: Assertion `tree->refs > 0' failed.
make[1]: *** [Makefile:1568: gen/parse2.c] Aborted (core dumped)


Running the same bootstrap1 command from the host system completes
fine.

# mount COLM-img /mnt
# cd /mnt/home/abuild/rpmbuild/BUILD/colm-0.13.0.5/src
# LD_LIBRARY_PATH=/mnt/lib:/mnt/usr/lib:$PWD/.libs /mnt/lib/ld-linux.so.2 .libs/bootstrap1 -c -o gen/parse2.c -e gen/if2.h -x gen/if2.cc colm.lm

It really starts to smell like a KVM issue, but I hope that that is
not off-putting to debugging efforts. It does come a bit strange that
it "only" affects colm out of all the other things people do in KVM.




More information about the colm-users mailing list