[PATCH v3] KVM: Move VM's worker kthreads back to the original cgroup before exiting.

[PATCH v3] KVM: Move VM's worker kthreads back to the original cgroup before exiting.

[Vipin Sharma]

VM worker kthreads can linger in the VM process's cgroup for sometime
after KVM terminates the VM process.

KVM terminates the worker kthreads by calling kthread_stop() which waits
on the 'exited' completion, triggered by exit_mm(), via mm_release(), in
do_exit() during the kthread's exit.  However, these kthreads are
removed from the cgroup using the cgroup_exit() which happens after the
exit_mm(). Therefore, a VM process can terminate in between the
exit_mm() and cgroup_exit() calls, leaving only worker kthreads in the
cgroup.

Moving worker kthreads back to the original cgroup (kthreadd_task's
cgroup) makes sure that the cgroup is empty as soon as the main VM
process is terminated.

Signed-off-by: Vipin Sharma <vipinsh@google.com>
---

v3:
- Use 'current->real_parent' (kthreadd_task) in the
  cgroup_attach_task_all() call.
- Revert cgroup APIs changes in v2. Now, patch does not touch cgroup
  APIs.
- Update commit and comment message

v2: https://lore.kernel.org/lkml/20211222225350.1912249-1-vipinsh@google.com/
- Use kthreadd_task in the cgroup API to avoid build issue.

v1: https://lore.kernel.org/lkml/20211214050708.4040200-1-vipinsh@google.com/

 virt/kvm/kvm_main.c | 21 +++++++++++++++++++--
 1 file changed, 19 insertions(+), 2 deletions(-)

diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 83c57bcc6eb6..2c9dcfffb606 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -5813,7 +5813,7 @@ static int kvm_vm_worker_thread(void *context)
 	struct kvm *kvm = init_context->kvm;
 	kvm_vm_thread_fn_t thread_fn = init_context->thread_fn;
 	uintptr_t data = init_context->data;
-	int err;
+	int err, reattach_err;
 
 	err = kthread_park(current);
 	/* kthread_park(current) is never supposed to return an error */
@@ -5836,7 +5836,7 @@ static int kvm_vm_worker_thread(void *context)
 	init_context = NULL;
 
 	if (err)
-		return err;
+		goto out;
 
 	/* Wait to be woken up by the spawner before proceeding. */
 	kthread_parkme();
@@ -5844,6 +5844,23 @@ static int kvm_vm_worker_thread(void *context)
 	if (!kthread_should_stop())
 		err = thread_fn(kvm, data);
 
+out:
+	/*
+	 * Move kthread back to its original cgroup to prevent it lingering in
+	 * the cgroup of the VM process, after the latter finishes its
+	 * execution.
+	 *
+	 * kthread_stop() waits on the 'exited' completion condition which is
+	 * set in exit_mm(), via mm_release(), in do_exit(). However, the
+	 * kthread is removed from the cgroup in the cgroup_exit() which is
+	 * called after the exit_mm(). This causes the kthread_stop() to return
+	 * before the kthread actually quits the cgroup.
+	 */
+	reattach_err = cgroup_attach_task_all(current->real_parent, current);
+	if (reattach_err) {
+		kvm_err("%s: cgroup_attach_task_all failed on reattach with err %d\n",
+			__func__, reattach_err);
+	}
 	return err;
 }
 

base-commit: db6e7adf8de9b3b99a9856acb73870cc3a70e3ca
-- 
2.35.1.265.g69c8d7142f-goog

Re: [PATCH v3] KVM: Move VM's worker kthreads back to the original cgroup before exiting.

[kernel test robot]

Hi Vipin,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on db6e7adf8de9b3b99a9856acb73870cc3a70e3ca]

url:    https://github.com/0day-ci/linux/commits/Vipin-Sharma/KVM-Move-VM-s-worker-kthreads-back-to-the-original-cgroup-before-exiting/20220217-141723
base:   db6e7adf8de9b3b99a9856acb73870cc3a70e3ca
config: s390-randconfig-s032-20220217 (https://download.01.org/0day-ci/archive/20220217/202202172046.GuW8pHQc-lkp@intel.com/config)
compiler: s390-linux-gcc (GCC) 11.2.0
reproduce:
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # apt-get install sparse
        # sparse version: v0.6.4-dirty
        # https://github.com/0day-ci/linux/commit/1abffef71ef85b6fb8f1296e6ef38febc4f2b007
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Vipin-Sharma/KVM-Move-VM-s-worker-kthreads-back-to-the-original-cgroup-before-exiting/20220217-141723
        git checkout 1abffef71ef85b6fb8f1296e6ef38febc4f2b007
        # save the config file to linux build tree
        mkdir build_dir
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-11.2.0 make.cross C=1 CF='-fdiagnostic-prefix -D__CHECK_ENDIAN__' O=build_dir ARCH=s390 SHELL=/bin/bash

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>


sparse warnings: (new ones prefixed by >>)
   arch/s390/kvm/../../../virt/kvm/kvm_main.c: note: in included file:
   include/linux/kvm_host.h:1877:54: sparse: sparse: array of flexible structures
   include/linux/kvm_host.h:1879:56: sparse: sparse: array of flexible structures
>> arch/s390/kvm/../../../virt/kvm/kvm_main.c:5859:54: sparse: sparse: incorrect type in argument 1 (different address spaces) @@     expected struct task_struct *from @@     got struct task_struct [noderef] __rcu *real_parent @@
   arch/s390/kvm/../../../virt/kvm/kvm_main.c:5859:54: sparse:     expected struct task_struct *from
   arch/s390/kvm/../../../virt/kvm/kvm_main.c:5859:54: sparse:     got struct task_struct [noderef] __rcu *real_parent
   arch/s390/kvm/../../../virt/kvm/kvm_main.c:2522:9: sparse: sparse: context imbalance in 'hva_to_pfn_remapped' - unexpected unlock

vim +5859 arch/s390/kvm/../../../virt/kvm/kvm_main.c

  5805	
  5806	static int kvm_vm_worker_thread(void *context)
  5807	{
  5808		/*
  5809		 * The init_context is allocated on the stack of the parent thread, so
  5810		 * we have to locally copy anything that is needed beyond initialization
  5811		 */
  5812		struct kvm_vm_worker_thread_context *init_context = context;
  5813		struct kvm *kvm = init_context->kvm;
  5814		kvm_vm_thread_fn_t thread_fn = init_context->thread_fn;
  5815		uintptr_t data = init_context->data;
  5816		int err, reattach_err;
  5817	
  5818		err = kthread_park(current);
  5819		/* kthread_park(current) is never supposed to return an error */
  5820		WARN_ON(err != 0);
  5821		if (err)
  5822			goto init_complete;
  5823	
  5824		err = cgroup_attach_task_all(init_context->parent, current);
  5825		if (err) {
  5826			kvm_err("%s: cgroup_attach_task_all failed with err %d\n",
  5827				__func__, err);
  5828			goto init_complete;
  5829		}
  5830	
  5831		set_user_nice(current, task_nice(init_context->parent));
  5832	
  5833	init_complete:
  5834		init_context->err = err;
  5835		complete(&init_context->init_done);
  5836		init_context = NULL;
  5837	
  5838		if (err)
  5839			goto out;
  5840	
  5841		/* Wait to be woken up by the spawner before proceeding. */
  5842		kthread_parkme();
  5843	
  5844		if (!kthread_should_stop())
  5845			err = thread_fn(kvm, data);
  5846	
  5847	out:
  5848		/*
  5849		 * Move kthread back to its original cgroup to prevent it lingering in
  5850		 * the cgroup of the VM process, after the latter finishes its
  5851		 * execution.
  5852		 *
  5853		 * kthread_stop() waits on the 'exited' completion condition which is
  5854		 * set in exit_mm(), via mm_release(), in do_exit(). However, the
  5855		 * kthread is removed from the cgroup in the cgroup_exit() which is
  5856		 * called after the exit_mm(). This causes the kthread_stop() to return
  5857		 * before the kthread actually quits the cgroup.
  5858		 */
> 5859		reattach_err = cgroup_attach_task_all(current->real_parent, current);
  5860		if (reattach_err) {
  5861			kvm_err("%s: cgroup_attach_task_all failed on reattach with err %d\n",
  5862				__func__, reattach_err);
  5863		}
  5864		return err;
  5865	}
  5866	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

Re: [PATCH v3] KVM: Move VM's worker kthreads back to the original cgroup before exiting.

[Paolo Bonzini]

On 2/17/22 13:34, kernel test robot wrote:
>> 5859		reattach_err = cgroup_attach_task_all(current->real_parent, current);

This needs to be within rcu_dereference().

Paolo

>    5860		if (reattach_err) {
>    5861			kvm_err("%s: cgroup_attach_task_all failed on reattach with err %d\n",
>    5862				__func__, reattach_err);
>    5863		}
>    5864		return err;
>    5865	}
>    5866	
> 
> ---
> 0-DAY CI Kernel Test Service, Intel Corporation
> https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org
> 

Re: [PATCH v3] KVM: Move VM's worker kthreads back to the original cgroup before exiting.

[kernel test robot]

Hi Vipin,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on db6e7adf8de9b3b99a9856acb73870cc3a70e3ca]

url:    https://github.com/0day-ci/linux/commits/Vipin-Sharma/KVM-Move-VM-s-worker-kthreads-back-to-the-original-cgroup-before-exiting/20220217-141723
base:   db6e7adf8de9b3b99a9856acb73870cc3a70e3ca
config: arm64-randconfig-s032-20220217 (https://download.01.org/0day-ci/archive/20220218/202202180218.msk1UR5R-lkp@intel.com/config)
compiler: aarch64-linux-gcc (GCC) 11.2.0
reproduce:
        wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
        chmod +x ~/bin/make.cross
        # apt-get install sparse
        # sparse version: v0.6.4-dirty
        # https://github.com/0day-ci/linux/commit/1abffef71ef85b6fb8f1296e6ef38febc4f2b007
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Vipin-Sharma/KVM-Move-VM-s-worker-kthreads-back-to-the-original-cgroup-before-exiting/20220217-141723
        git checkout 1abffef71ef85b6fb8f1296e6ef38febc4f2b007
        # save the config file to linux build tree
        mkdir build_dir
        COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-11.2.0 make.cross C=1 CF='-fdiagnostic-prefix -D__CHECK_ENDIAN__' O=build_dir ARCH=arm64 SHELL=/bin/bash

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>


sparse warnings: (new ones prefixed by >>)
   arch/arm64/kvm/../../../virt/kvm/kvm_main.c: note: in included file:
   include/linux/kvm_host.h:1877:54: sparse: sparse: array of flexible structures
   include/linux/kvm_host.h:1879:56: sparse: sparse: array of flexible structures
>> arch/arm64/kvm/../../../virt/kvm/kvm_main.c:5859:54: sparse: sparse: incorrect type in argument 1 (different address spaces) @@     expected struct task_struct *from @@     got struct task_struct [noderef] __rcu *real_parent @@
   arch/arm64/kvm/../../../virt/kvm/kvm_main.c:5859:54: sparse:     expected struct task_struct *from
   arch/arm64/kvm/../../../virt/kvm/kvm_main.c:5859:54: sparse:     got struct task_struct [noderef] __rcu *real_parent
   arch/arm64/kvm/../../../virt/kvm/kvm_main.c:538:9: sparse: sparse: context imbalance in 'kvm_mmu_notifier_change_pte' - different lock contexts for basic block
   arch/arm64/kvm/../../../virt/kvm/kvm_main.c:538:9: sparse: sparse: context imbalance in 'kvm_mmu_notifier_invalidate_range_start' - different lock contexts for basic block
   arch/arm64/kvm/../../../virt/kvm/kvm_main.c:538:9: sparse: sparse: context imbalance in 'kvm_mmu_notifier_invalidate_range_end' - different lock contexts for basic block
   arch/arm64/kvm/../../../virt/kvm/kvm_main.c:538:9: sparse: sparse: context imbalance in 'kvm_mmu_notifier_clear_flush_young' - different lock contexts for basic block
   arch/arm64/kvm/../../../virt/kvm/kvm_main.c:538:9: sparse: sparse: context imbalance in 'kvm_mmu_notifier_clear_young' - different lock contexts for basic block
   arch/arm64/kvm/../../../virt/kvm/kvm_main.c:538:9: sparse: sparse: context imbalance in 'kvm_mmu_notifier_test_young' - different lock contexts for basic block
   arch/arm64/kvm/../../../virt/kvm/kvm_main.c:2522:9: sparse: sparse: context imbalance in 'hva_to_pfn_remapped' - unexpected unlock

vim +5859 arch/arm64/kvm/../../../virt/kvm/kvm_main.c

  5805	
  5806	static int kvm_vm_worker_thread(void *context)
  5807	{
  5808		/*
  5809		 * The init_context is allocated on the stack of the parent thread, so
  5810		 * we have to locally copy anything that is needed beyond initialization
  5811		 */
  5812		struct kvm_vm_worker_thread_context *init_context = context;
  5813		struct kvm *kvm = init_context->kvm;
  5814		kvm_vm_thread_fn_t thread_fn = init_context->thread_fn;
  5815		uintptr_t data = init_context->data;
  5816		int err, reattach_err;
  5817	
  5818		err = kthread_park(current);
  5819		/* kthread_park(current) is never supposed to return an error */
  5820		WARN_ON(err != 0);
  5821		if (err)
  5822			goto init_complete;
  5823	
  5824		err = cgroup_attach_task_all(init_context->parent, current);
  5825		if (err) {
  5826			kvm_err("%s: cgroup_attach_task_all failed with err %d\n",
  5827				__func__, err);
  5828			goto init_complete;
  5829		}
  5830	
  5831		set_user_nice(current, task_nice(init_context->parent));
  5832	
  5833	init_complete:
  5834		init_context->err = err;
  5835		complete(&init_context->init_done);
  5836		init_context = NULL;
  5837	
  5838		if (err)
  5839			goto out;
  5840	
  5841		/* Wait to be woken up by the spawner before proceeding. */
  5842		kthread_parkme();
  5843	
  5844		if (!kthread_should_stop())
  5845			err = thread_fn(kvm, data);
  5846	
  5847	out:
  5848		/*
  5849		 * Move kthread back to its original cgroup to prevent it lingering in
  5850		 * the cgroup of the VM process, after the latter finishes its
  5851		 * execution.
  5852		 *
  5853		 * kthread_stop() waits on the 'exited' completion condition which is
  5854		 * set in exit_mm(), via mm_release(), in do_exit(). However, the
  5855		 * kthread is removed from the cgroup in the cgroup_exit() which is
  5856		 * called after the exit_mm(). This causes the kthread_stop() to return
  5857		 * before the kthread actually quits the cgroup.
  5858		 */
> 5859		reattach_err = cgroup_attach_task_all(current->real_parent, current);
  5860		if (reattach_err) {
  5861			kvm_err("%s: cgroup_attach_task_all failed on reattach with err %d\n",
  5862				__func__, reattach_err);
  5863		}
  5864		return err;
  5865	}
  5866	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

Re: [PATCH v3] KVM: Move VM's worker kthreads back to the original cgroup before exiting.

[kernel test robot]

Hi Vipin,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on db6e7adf8de9b3b99a9856acb73870cc3a70e3ca]

url:    https://github.com/0day-ci/linux/commits/Vipin-Sharma/KVM-Move-VM-s-worker-kthreads-back-to-the-original-cgroup-before-exiting/20220217-141723
base:   db6e7adf8de9b3b99a9856acb73870cc3a70e3ca
config: x86_64-rhel-8.3-kselftests (https://download.01.org/0day-ci/archive/20220218/202202180730.AAgeOZkF-lkp@intel.com/config)
compiler: gcc-9 (Debian 9.3.0-22) 9.3.0
reproduce:
        # apt-get install sparse
        # sparse version: v0.6.4-dirty
        # https://github.com/0day-ci/linux/commit/1abffef71ef85b6fb8f1296e6ef38febc4f2b007
        git remote add linux-review https://github.com/0day-ci/linux
        git fetch --no-tags linux-review Vipin-Sharma/KVM-Move-VM-s-worker-kthreads-back-to-the-original-cgroup-before-exiting/20220217-141723
        git checkout 1abffef71ef85b6fb8f1296e6ef38febc4f2b007
        # save the config file to linux build tree
        mkdir build_dir
        make W=1 C=1 CF='-fdiagnostic-prefix -D__CHECK_ENDIAN__' O=build_dir ARCH=x86_64 SHELL=/bin/bash

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@intel.com>


sparse warnings: (new ones prefixed by >>)
   arch/x86/kvm/../../../virt/kvm/kvm_main.c: note: in included file:
   include/linux/kvm_host.h:1877:54: sparse: sparse: array of flexible structures
   include/linux/kvm_host.h:1879:56: sparse: sparse: array of flexible structures
>> arch/x86/kvm/../../../virt/kvm/kvm_main.c:5859:54: sparse: sparse: incorrect type in argument 1 (different address spaces) @@     expected struct task_struct *from @@     got struct task_struct [noderef] __rcu *real_parent @@
   arch/x86/kvm/../../../virt/kvm/kvm_main.c:5859:54: sparse:     expected struct task_struct *from
   arch/x86/kvm/../../../virt/kvm/kvm_main.c:5859:54: sparse:     got struct task_struct [noderef] __rcu *real_parent
   arch/x86/kvm/../../../virt/kvm/kvm_main.c:538:9: sparse: sparse: context imbalance in 'kvm_mmu_notifier_change_pte' - different lock contexts for basic block
   arch/x86/kvm/../../../virt/kvm/kvm_main.c:538:9: sparse: sparse: context imbalance in 'kvm_mmu_notifier_invalidate_range_start' - different lock contexts for basic block
   arch/x86/kvm/../../../virt/kvm/kvm_main.c:538:9: sparse: sparse: context imbalance in 'kvm_mmu_notifier_invalidate_range_end' - different lock contexts for basic block
   arch/x86/kvm/../../../virt/kvm/kvm_main.c:538:9: sparse: sparse: context imbalance in 'kvm_mmu_notifier_clear_flush_young' - different lock contexts for basic block
   arch/x86/kvm/../../../virt/kvm/kvm_main.c:538:9: sparse: sparse: context imbalance in 'kvm_mmu_notifier_clear_young' - different lock contexts for basic block
   arch/x86/kvm/../../../virt/kvm/kvm_main.c:538:9: sparse: sparse: context imbalance in 'kvm_mmu_notifier_test_young' - different lock contexts for basic block
   arch/x86/kvm/../../../virt/kvm/kvm_main.c:2522:9: sparse: sparse: context imbalance in 'hva_to_pfn_remapped' - unexpected unlock

vim +5859 arch/x86/kvm/../../../virt/kvm/kvm_main.c

  5805	
  5806	static int kvm_vm_worker_thread(void *context)
  5807	{
  5808		/*
  5809		 * The init_context is allocated on the stack of the parent thread, so
  5810		 * we have to locally copy anything that is needed beyond initialization
  5811		 */
  5812		struct kvm_vm_worker_thread_context *init_context = context;
  5813		struct kvm *kvm = init_context->kvm;
  5814		kvm_vm_thread_fn_t thread_fn = init_context->thread_fn;
  5815		uintptr_t data = init_context->data;
  5816		int err, reattach_err;
  5817	
  5818		err = kthread_park(current);
  5819		/* kthread_park(current) is never supposed to return an error */
  5820		WARN_ON(err != 0);
  5821		if (err)
  5822			goto init_complete;
  5823	
  5824		err = cgroup_attach_task_all(init_context->parent, current);
  5825		if (err) {
  5826			kvm_err("%s: cgroup_attach_task_all failed with err %d\n",
  5827				__func__, err);
  5828			goto init_complete;
  5829		}
  5830	
  5831		set_user_nice(current, task_nice(init_context->parent));
  5832	
  5833	init_complete:
  5834		init_context->err = err;
  5835		complete(&init_context->init_done);
  5836		init_context = NULL;
  5837	
  5838		if (err)
  5839			goto out;
  5840	
  5841		/* Wait to be woken up by the spawner before proceeding. */
  5842		kthread_parkme();
  5843	
  5844		if (!kthread_should_stop())
  5845			err = thread_fn(kvm, data);
  5846	
  5847	out:
  5848		/*
  5849		 * Move kthread back to its original cgroup to prevent it lingering in
  5850		 * the cgroup of the VM process, after the latter finishes its
  5851		 * execution.
  5852		 *
  5853		 * kthread_stop() waits on the 'exited' completion condition which is
  5854		 * set in exit_mm(), via mm_release(), in do_exit(). However, the
  5855		 * kthread is removed from the cgroup in the cgroup_exit() which is
  5856		 * called after the exit_mm(). This causes the kthread_stop() to return
  5857		 * before the kthread actually quits the cgroup.
  5858		 */
> 5859		reattach_err = cgroup_attach_task_all(current->real_parent, current);
  5860		if (reattach_err) {
  5861			kvm_err("%s: cgroup_attach_task_all failed on reattach with err %d\n",
  5862				__func__, reattach_err);
  5863		}
  5864		return err;
  5865	}
  5866	

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org

Re: [PATCH v3] KVM: Move VM's worker kthreads back to the original cgroup before exiting.

[Sean Christopherson]

On Thu, Feb 17, 2022, Paolo Bonzini wrote:
> On 2/17/22 13:34, kernel test robot wrote:
> > > 5859		reattach_err = cgroup_attach_task_all(current->real_parent, current);
> 
> This needs to be within rcu_dereference().

As Vipin found out the hard way, cgroup_attach_task_all() can't be inside an RCU
critical section as it might sleep due to mutex_lock().

My sched knowledge is sketchy at best, but I believe KVM just needs to guarantee
the liveliness of real_parent when passing it to cgroup_attach_task_all().
Presumably kthreadd_task can (in theory) be killed/exiting while this is going on,
but at that point I doubt anyone cares much about cgroup accuracy.

So I think this can be:

	rcu_read_lock();
	parent = rcu_dereference(current->real_parent);
	get_task_struct(parent);
	rcu_read_unlock();

	(void)cgroup_attach_task_all(parent, current);
        put_task_struct(parent);


> >    5860		if (reattach_err) {
> >    5861			kvm_err("%s: cgroup_attach_task_all failed on reattach with err %d\n",
> >    5862				__func__, reattach_err);

Eh, I wouldn't bother logging the error, userspace can't take action on it, and
if the system is OOM it's just more noise in the log to dig through.

Re: [PATCH v3] KVM: Move VM's worker kthreads back to the original cgroup before exiting.

[Paolo Bonzini]

On 2/19/22 01:30, Sean Christopherson wrote:
> 
> So I think this can be:
> 
> 	rcu_read_lock();
> 	parent = rcu_dereference(current->real_parent);
> 	get_task_struct(parent);
> 	rcu_read_unlock();
> 
> 	(void)cgroup_attach_task_all(parent, current);
>          put_task_struct(parent);

Yes, I agree.

Paolo