Re: [Qemu-devel] [PATCH v13 00/10] Block replication for continuous checkpoints

["Dr. David Alan Gilbert" <dgilbert@redhat.com>]

Hi,
  I can trigger a segfault if I wire in the block replication together with
a quorum instance; it only triggers with both of them present but,
it looks like the problem is a disagreement about the number of quorum
members;  I'm triggering this on the 'colo-v2.4-periodic-mode' branch
that is posted in the colo-framework set that I think includes this set
(from https://github.com/coloft/qemu.git).

To trigger:
./git/colo/jan-16/try/x86_64-softmmu/qemu-system-x86_64 -nographic -S

(qemu) drive_add 0 if=none,id=colo-disk0,file.filename=/home/localvms/bugzilla.raw,driver=raw,node-name=node0
(qemu) drive_add 1 if=none,id=active-disk0,throttling.bps-total=70000000,driver=replication,mode=secondary,file.driver=qcow2,file.file.filename=/run/colo-active-disk.qcow2,file.backing.driver=qcow2,file.backing.file.filename=/run/colo-hidden-disk.qcow2,file.backing.backing=colo-disk0
(qemu) drive_add 2 if=none,id=top-quorum,driver=quorum,read-pattern=fifo,vote-threshold=1,children.0=active-disk0
(qemu) device_add virtio-blk-pci,drive=top-quorum,addr=9

*** Error in `/root/colo/jan-2016/./try/x86_64-softmmu/qemu-system-x86_64': free(): invalid pointer: 0x0000555555a8fdf0 ***
======= Backtrace: =========
/lib64/libc.so.6(+0x7cfe1)[0x7ffff110ffe1]
/lib64/libglib-2.0.so.0(g_free+0xf)[0x7ffff1ecc36f]
/root/colo/jan-2016/./try/x86_64-softmmu/qemu-system-x86_64
Program received signal SIGABRT, Aborted.
0x00007ffff10c85f7 in raise () from /lib64/libc.so.6
(gdb) where
#0  0x00007ffff10c85f7 in raise () from /lib64/libc.so.6
#1  0x00007ffff10c9ce8 in abort () from /lib64/libc.so.6
#2  0x00007ffff1108317 in __libc_message () from /lib64/libc.so.6
#3  0x00007ffff110ffe1 in _int_free () from /lib64/libc.so.6
#4  0x00007ffff1ecc36f in g_free () from /lib64/libglib-2.0.so.0
#5  0x00005555559dfdd7 in qemu_iovec_destroy (qiov=0x555557815410) at /root/colo/jan-2016/qemu/util/iov.c:378
#6  0x0000555555989cce in quorum_aio_finalize (acb=0x555557815350) at /root/colo/jan-2016/qemu/block/quorum.c:171
171	            qemu_iovec_destroy(&acb->qcrs[i].qiov);
(gdb) list
166	
167	    if (acb->is_read) {
168	        /* on the quorum case acb->child_iter == s->num_children - 1 */
169	        for (i = 0; i <= acb->child_iter; i++) {
170	            qemu_vfree(acb->qcrs[i].buf);
171	            qemu_iovec_destroy(&acb->qcrs[i].qiov);
172	        }
173	    }
174	
175	    g_free(acb->qcrs);
(gdb) p acb->child_iter
$1 = 1
(gdb) p i
$3 = 1

#7  0x000055555598afca in quorum_aio_cb (opaque=<optimized out>, ret=-5)
    at /root/colo/jan-2016/qemu/block/quorum.c:302
#8  0x00005555559990ee in bdrv_co_complete (acb=0x555557815410) at /root/colo/jan-2016/qemu/block/io.c:2122
.....

So I guess acb->child_iter is wrong, since we only have one child on that quorum?
and we're trying to do a destroy on the second child.

Dave
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK