dnode_evict_dbufs leads to panic

[Shiva Bhanujan]

Hello,

We're running FreeBSD 10.2-RELEASE-p20, as a VM (HVM) on XenServer 6.5.  It serves as a backend storage system, and exports ZFS via NFS.  After running for a day and a half, we notice the FreeBSD VM getting hung, or has crashed.

Looking at the following traceback, I have found this issue being reported multiple times.

#1  0xffffffff8093ce62 in kern_reboot (howto=260)
    at /usr/src/sys/kern/kern_shutdown.c:451
#2  0xffffffff8093d245 in vpanic (fmt=<value optimized out>,
    ap=<value optimized out>) at /usr/src/sys/kern/kern_shutdown.c:758
#3  0xffffffff8093d0d3 in panic (fmt=0x0)
    at /usr/src/sys/kern/kern_shutdown.c:687
#4  0xffffffff80d3f22b in trap_fatal (frame=<value optimized out>,
    eva=<value optimized out>) at /usr/src/sys/amd64/amd64/trap.c:851
#5  0xffffffff80d3f52d in trap_pfault (frame=0xfffffe032937f660,
    usermode=<value optimized out>) at /usr/src/sys/amd64/amd64/trap.c:674
#6  0xffffffff80d3ebca in trap (frame=0xfffffe032937f660)
    at /usr/src/sys/amd64/amd64/trap.c:440
#7  0xffffffff80d24f42 in calltrap ()
    at /usr/src/sys/amd64/amd64/exception.S:236
#8  0xffffffff8094550b in _sx_xlock_hard (sx=0xfffff800b3e35918,
    tid=18446735282677829632, opts=<value optimized out>,
    file=0xfffff8012dad5000 "", line=49749) at /usr/src/sys/kern/kern_sx.c:568
#9  0xffffffff8094501d in _sx_xlock (sx=0xfffff8012dad5000, opts=766332928,
    file=<value optimized out>, line=766332928) at sx.h:152
#10 0xffffffff819ca67d in dnode_evict_dbufs (dn=0xfffff8008c4728d0)
    at /usr/src/sys/modules/zfs/../../cddl/contrib/opensolaris/uts/common/fs/zfs/dnode_sync.c:416
#11 0xffffffff819c151b in dmu_objset_evict_dbufs (os=<value optimized out>)
    at /usr/src/sys/modules/zfs/../../cddl/contrib/opensolaris/uts/common/fs/zfs/dmu_objset.c:620
#12 0xffffffff81a49ff1 in zfsvfs_teardown (zfsvfs=0xfffff800adb6e000,
    unmounting=<value optimized out>)
    at /usr/src/sys/modules/zfs/../../cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vfsops.c:1915
#13 0xffffffff81a4b965 in zfs_umount (vfsp=0xfffff8007035e330,
    fflag=<value optimized out>)
    at /usr/src/sys/modules/zfs/../../cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vfsops.c:2009
#14 0xffffffff809d951f in dounmount (mp=0xfffff8007035e330, flags=524288,
    td=0xfffff8012dad5000) at /usr/src/sys/kern/vfs_mount.c:1316
#15 0xffffffff809d9026 in sys_unmount (td=0xfffff8012dad5000,
    uap=0xfffffe032937fb80) at /usr/src/sys/kern/vfs_mount.c:1208
#16 0xffffffff80d3fb47 in amd64_syscall (td=0xfffff8012dad5000, traced=0)
    at subr_syscall.c:134

However, I haven't been able to find a definite answer as to how this is to be fixed.  Can anybody provide some guidance if this is something that has a fix in a subsequent FreeBSD release?  If there is more information that is needed, please do let me know?  I can send the core file if needed, haven't attached it here so as not to spam.  Please let me know if I should provide it?

Regards,
Shiva