xen server crashes when used with network bonding modes 5 or 6

Issue

• Almost immediately after starting Xend, the servers' network connections stopped working, and a number of bonding-related call traces started filling the console.

BUG: soft lockup - CPU#2 stuck for 10s! [ip:3646]

Pid: 3646, comm:                   ip
EIP: 0061:[<c0617021>] CPU: 2
EIP is at __write_lock_failed+0x9/0x20
 EFLAGS: 00000206    Not tainted  (2.6.18-164.el5xen #1)
EAX: ebf9b41c EBX: ebf9b41c ECX: 00000000 EDX: eaff3000
ESI: 00000001 EDI: ebf9b11e EBP: ebf9b400 DS: 007b ES: 007b
CR0: 8005003b CR2: 00d5a558 CR3: 2b93a000 CR4: 00002660
 [<c061909c>] _write_lock_bh+0x18/0x1a
 [<ee28ef8a>] bond_alb_set_mac_address+0x231/0x24f [bonding]
 [<c05bb32b>] dev_set_mac_address+0x2e/0x4a
 [<c05bda6b>] dev_ioctl+0x399/0x462
 [<c04c1464>] inode_has_perm+0x54/0x5c
 [<c05b25d2>] sock_ioctl+0x191/0x1b3
 [<c05b2441>] sock_ioctl+0x0/0x1b3
 [<c0480afd>] do_ioctl+0x1c/0x5d
 [<c0481091>] vfs_ioctl+0x47b/0x4d3
 [<c0481131>] sys_ioctl+0x48/0x5f
 [<c0405413>] syscall_call+0x7/0xb

This continues until the server reboots.

This problem only occurs if network-bridge is used with a non-tagged mode 5 or 6 bond0 interface.