lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:   Mon, 17 Feb 2020 10:23:18 +0100
From:   Giuseppe Scrivano <gscrivan@...hat.com>
To:     "Paul E. McKenney" <paulmck@...nel.org>
Cc:     linux-kernel@...r.kernel.org, rcu@...r.kernel.org,
        ebiederm@...ssion.com
Subject: Re: [PATCH] ipc: use a work queue to free_ipc

Hi Paul,

"Paul E. McKenney" <paulmck@...nel.org> writes:

> Nice speedup!
>
> However, I am not convinced that the code shown below is safe.
>
> I believe that you need either a synchronize_rcu() in your free_ipc()
> function or that you need to pass free_ipc() to queue_rcu_work() instead
> of directly schedule_work().  As things are, I would expect you to see
> free_ipc_ns() being invoke too soon on heavily loaded CONFIG_PREEMPT=y
> kernels.  Which can be quite a pain to debug!
>
> Or am I missing something?

thanks for the review!

Would being called too soon be a problem?  The current version calls
free_ipc_ns() as part of the put_ipc_ns cleanup.

free_ipc() calls immediately synchronize_rcu() through free_ipc_ns():

free_ipc_ns() -> mq_put_mnt() -> kern_unmount() -> synchronize_rcu()

Do you think we should make it more explicit and add a synchronize_rcu()
as part of the free_ipc_ns() function itself?

Regards,
Giuseppe


>
> 							Thanx, Paul
>
>> Signed-off-by: Giuseppe Scrivano <gscrivan@...hat.com>
>> ---
>>  include/linux/ipc_namespace.h |  2 ++
>>  ipc/namespace.c               | 17 +++++++++++++++--
>>  2 files changed, 17 insertions(+), 2 deletions(-)
>> 
>> diff --git a/include/linux/ipc_namespace.h b/include/linux/ipc_namespace.h
>> index c309f43bde45..a06a78c67f19 100644
>> --- a/include/linux/ipc_namespace.h
>> +++ b/include/linux/ipc_namespace.h
>> @@ -68,6 +68,8 @@ struct ipc_namespace {
>>  	struct user_namespace *user_ns;
>>  	struct ucounts *ucounts;
>>  
>> +	struct llist_node mnt_llist;
>> +
>>  	struct ns_common ns;
>>  } __randomize_layout;
>>  
>> diff --git a/ipc/namespace.c b/ipc/namespace.c
>> index b3ca1476ca51..37d27e1b807a 100644
>> --- a/ipc/namespace.c
>> +++ b/ipc/namespace.c
>> @@ -117,6 +117,7 @@ void free_ipcs(struct ipc_namespace *ns, struct ipc_ids *ids,
>>  
>>  static void free_ipc_ns(struct ipc_namespace *ns)
>>  {
>> +	mq_put_mnt(ns);
>>  	sem_exit_ns(ns);
>>  	msg_exit_ns(ns);
>>  	shm_exit_ns(ns);
>> @@ -127,6 +128,17 @@ static void free_ipc_ns(struct ipc_namespace *ns)
>>  	kfree(ns);
>>  }
>>  
>> +static LLIST_HEAD(free_ipc_list);
>> +static void free_ipc(struct work_struct *unused)
>> +{
>> +	struct llist_node *node = llist_del_all(&free_ipc_list);
>> +	struct ipc_namespace *n, *t;
>> +
>> +	llist_for_each_entry_safe(n, t, node, mnt_llist)
>> +		free_ipc_ns(n);
>> +}
>> +static DECLARE_WORK(free_ipc_work, free_ipc);
>> +
>>  /*
>>   * put_ipc_ns - drop a reference to an ipc namespace.
>>   * @ns: the namespace to put
>> @@ -148,8 +160,9 @@ void put_ipc_ns(struct ipc_namespace *ns)
>>  	if (refcount_dec_and_lock(&ns->count, &mq_lock)) {
>>  		mq_clear_sbinfo(ns);
>>  		spin_unlock(&mq_lock);
>> -		mq_put_mnt(ns);
>> -		free_ipc_ns(ns);
>> +
>> +		if (llist_add(&ns->mnt_llist, &free_ipc_list))
>> +			schedule_work(&free_ipc_work);
>>  	}
>>  }
>>  
>> -- 
>> 2.24.1
>> 

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ