On Tue, Feb 05, 2019 at 11:06:42AM -0500, Michael S. Tsirkin wrote: > On Tue, Feb 05, 2019 at 12:41:55PM +0000, Peter Maydell wrote: > > On Tue, 5 Feb 2019 at 01:52, Michael S. Tsirkin <m...@redhat.com> wrote: > > > So I think I fixed the build errors. And who knows maybe > > > that will fix the runtime ones somehow .. > > > Could you try pls? Re-pushed to same tag. > > > > these failures still remain: > > > > unused variable for OSX and the BSDs: > > > > tests/vhost-user-test.c:942:17: warning: unused variable 'hugefs' > > [-Wunused-variable] > > const char *hugefs; > > ^ > > > > Assertion failure on sparc, s390x and PPC64 hosts: > > > > MALLOC_PERTURB_=${MALLOC_PERTURB_:-$(( ${RANDOM:-0} % 255 + 1))} > > QTEST_QEMU_BINARY=i386-softmmu/qemu-system-i386 > > QTEST_QEMU_IMG=qemu-img tests/vhost-user-test -m=quick -k --tap < > > /dev/null | ./scripts/tap-driver.pl --test-name="vhost-user-test" > > qemu-system-i386: backend does not support LE vnet headers; falling > > back on userspace virtio > > qemu-system-i386: backend does not support LE vnet headers; falling > > back on userspace virtio > > ** > > ERROR:/home/pm215/qemu/tests/vhost-user-test.c:243:wait_for_fds: > > assertion failed: (s->fds_num) > > ERROR - too few tests run (expected 4, got 0) > > > > Given the set of hosts affected this is almost certainly > > a big-endian-host specific bug. > > > > thanks > > -- PMM > > You were right. > > OK I reverted the whole part dealing with vhost-user and reposted.
I noticed that the merged pull request could possibly have squashed the below two patches (in previous pull) into one super patch (a56de056c91f87e1e): i386/kvm: ignore masked irqs when update msi routes contrib/vhost-user-blk: fix the compilation issue Here, the first patch lost its commit message, and the last patch lost its real author. ;) Regards, -- Peter Xu