On Fri, Sep 29, 2017 at 01:11:13PM +0300, Alexey Perevalov wrote: > This small commit fixes build of tests/migration/stress on Ubuntu16.04 > (on RHEL7.4 the same situation, after typical configure with following options > --target-list=x86_64-linux-user,i386-linux-user,x86_64-softmmu,i386-softmmu \ > --enable-numa --enable-kvm --enable-vhost-net --enable-tpm --enable-debug \ > --disable-gtk ) > PTHREAD_LIB is empty after successfully compiled pthread sample in configure. > tests/migration/stress relay on it, but -pthread is not there > and linker fails trying to resolve pthread symbols. > > I still have a questions: > 1. Is there a better way to obtain -thread option from QEMU_CFLAGS > 2. Is linker options really necessary for stress application, because only -O3 > is used, so w/o linker options in Makefile there is no need in PTHREAD_LIB > at all, I mean in whole project. > 3. LINKPROG will be expanded into c++ firstly, and -static in > tests/Makefile.include will require static libstdc++, but configure doesn't > check it and in case of missing libstdc++-static.x86_64 (RHEL package name), > build will fail with the following message: "ld: cannot find -lstdc++", > I think here reasonably to use just $(CC) instead of LINKPROG, stress.c is on > C.
Yeah, changing to $(CC) is fine - and adding -lpthread to the end of the args instead of $PTHREAD_LIB > > Also ./tests/migration/guestperf.py can not be started w/o > super user privileges, and reports in this case something useless like this: > Error: [Errno 104] Connection reset by peer IIRC, that's an Ubuntu problem - they pointlessly restrict their /boot/vmlinuz files so you can't read them as non-root. Use the --kernel arg to point to a readable image. There's no need for the test to run privileged Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
