[Devel] Re-entrance of kernel do_exit() while VE is stopping

[Bojie Li]

Hello,

I have met several uninterruptible (D state) processes in containers
while VE is shutting down (vzctl stop). The re-entrance of do_exit()
caused the kernel to print an alert message and scheduled the process
away permanently. I wonder whether or not this bug has been fixed in
the mainline OpenVZ kernel. Thanks in advance.

Kernel: 2.6.32-5-openvz-amd64 (in Debian package
linux-image-2.6.32-5-openvz-amd64, 2.6.32-48squeeze4)

PID 5942 kernel thread [logger], VE 377
PID 11524 kernel thread [sshd], VE 720 (same stack trace)

[<ffffffff81051b55>] ? do_exit+0xdf/0x75b
[<ffffffff8104e7ee>] ? release_console_sem+0x192/0x1c4
[<ffffffff812ec5bc>] ? oops_end+0xaf/0xb4
[<ffffffff81032353>] ? no_context+0x1e9/0x1f8
[<ffffffff810d24dc>] ? free_pgtables+0x55/0xbd
[<ffffffff81032506>] ? __bad_area_nosemaphore+0x1a4/0x1c8
[<ffffffff812eba85>] ? page_fault+0x25/0x30
[<ffffffff810d24dc>] ? free_pgtables+0x55/0xbd
[<ffffffff812eb619>] ? _spin_lock_irqsave+0x1a/0x34
[<ffffffff8107365f>] ? uncharge_beancounter+0x1f/0x50
[<ffffffff810741d4>] ? ub_slab_uncharge+0x29/0x42
[<ffffffff810e9314>] ? kmem_cache_free+0xc3/0xd1
[<ffffffff810d24dc>] ? free_pgtables+0x55/0xbd
[<ffffffff810d3df2>] ? exit_mmap+0xf4/0x14d
[<ffffffff8104bbca>] ? mmput+0x2b/0xf4
[<ffffffff810502dc>] ? exit_mm+0x115/0x120
[<ffffffff81051c7c>] ? do_exit+0x206/0x75b
[<ffffffff81052247>] ? do_group_exit+0x76/0x9d
[<ffffffff81052280>] ? sys_exit_group+0x12/0x16
[<ffffffff81010c12>] ? system_call_fastpath+0x16/0x1b

====================================

PID 30803 kernel thread [mysqld], VE 678

[<ffffffff81051b55>] ? do_exit+0xdf/0x75b
[<ffffffff8104e7ee>] ? release_console_sem+0x192/0x1c4
[<ffffffff812ec5bc>] ? oops_end+0xaf/0xb4
[<ffffffff81032353>] ? no_context+0x1e9/0x1f8
[<ffffffff810d24dc>] ? free_pgtables+0x55/0xbd
[<ffffffff81032506>] ? __bad_area_nosemaphore+0x1a4/0x1c8
[<ffffffff810bf313>] ? release_pages+0x17b/0x18d
[<ffffffff810e6b86>] ? add_partial+0x11/0x58
[<ffffffff810e8f4d>] ? __slab_free+0x7f/0x278
[<ffffffff812eba85>] ? page_fault+0x25/0x30
[<ffffffff810d24dc>] ? free_pgtables+0x55/0xbd
[<ffffffff812eb619>] ? _spin_lock_irqsave+0x1a/0x34
[<ffffffff8107365f>] ? uncharge_beancounter+0x1f/0x50
[<ffffffff810741d4>] ? ub_slab_uncharge+0x29/0x42
[<ffffffff810e9314>] ? kmem_cache_free+0xc3/0xd1
[<ffffffff8117d290>] ? prio_tree_remove+0xbd/0xc5
[<ffffffff810d24dc>] ? free_pgtables+0x55/0xbd
[<ffffffff810d3df2>] ? exit_mmap+0xf4/0x14d
[<ffffffff8104bbca>] ? mmput+0x2b/0xf4
[<ffffffff810502dc>] ? exit_mm+0x115/0x120
[<ffffffff81051c7c>] ? do_exit+0x206/0x75b
[<ffffffff81052247>] ? do_group_exit+0x76/0x9d
[<ffffffff81052280>] ? sys_exit_group+0x12/0x16
[<ffffffff81010c12>] ? system_call_fastpath+0x16/0x1b

=================================

PID 21299, VE 378
User-mode process: /usr/sbin/apache2

[<ffffffff81051b55>] ? do_exit+0xdf/0x75b
[<ffffffff8104e7ee>] ? release_console_sem+0x192/0x1c4
[<ffffffff812ec5bc>] ? oops_end+0xaf/0xb4
[<ffffffff81032353>] ? no_context+0x1e9/0x1f8
[<ffffffff810d24dc>] ? free_pgtables+0x55/0xbd
[<ffffffff81032506>] ? __bad_area_nosemaphore+0x1a4/0x1c8
[<ffffffff810bf313>] ? release_pages+0x17b/0x18d
[<ffffffff8101172e>] ? apic_timer_interrupt+0xe/0x20
[<ffffffff812eba85>] ? page_fault+0x25/0x30
[<ffffffff810d24dc>] ? free_pgtables+0x55/0xbd
[<ffffffff812eb619>] ? _spin_lock_irqsave+0x1a/0x34
[<ffffffff8107365f>] ? uncharge_beancounter+0x1f/0x50
[<ffffffff810741d4>] ? ub_slab_uncharge+0x29/0x42
[<ffffffff810e9314>] ? kmem_cache_free+0xc3/0xd1
[<ffffffff8117d290>] ? prio_tree_remove+0xbd/0xc5
[<ffffffff810d24dc>] ? free_pgtables+0x55/0xbd
[<ffffffff810d3df2>] ? exit_mmap+0xf4/0x14d
[<ffffffff8104bbca>] ? mmput+0x2b/0xf4
[<ffffffff810502dc>] ? exit_mm+0x115/0x120
[<ffffffff81051c7c>] ? do_exit+0x206/0x75b
[<ffffffff81052247>] ? do_group_exit+0x76/0x9d
[<ffffffff81052280>] ? sys_exit_group+0x12/0x16
[<ffffffff81010c12>] ? system_call_fastpath+0x16/0x1b