[Devel] [PATCH v3 2/4] SUNRPC: fix races on PipeFS UMOUNT notifications

Stanislav Kinsbursky skinsbursky at parallels.com
Tue Jun 25 22:32:09 PDT 2013


25.06.2013 20:13, Myklebust, Trond пишет:
> On Mon, 2013-06-24 at 11:52 +0400, Stanislav Kinsbursky wrote:
>> CPU#0                                   CPU#1
>> -----------------------------           -----------------------------
>> rpc_kill_sb
>> sn->pipefs_sb = NULL                    rpc_release_client
>> (UMOUNT_EVENT)                          rpc_free_auth
>> rpc_pipefs_event
>> rpc_get_client_for_event
>> !atomic_inc_not_zero(cl_count)
>> <skip the client>
>>                                          atomic_inc(cl_count)
>>                                          rpc_free_client
>>                                          rpc_clnt_remove_pipedir
>>                                          <skip client dir removing>
>>
>> To fix this, this patch does the following:
>>
>> 1) Calls RPC_PIPEFS_UMOUNT notification with sn->pipefs_sb_lock being held.
>> 2) Removes SUNRPC client from the list AFTER pipes destroying.
>> 3) Doesn't hold RPC client on notification: if client in the list, then it
>> can't be destroyed while sn->pipefs_sb_lock in hold by notification caller.
>>
>> Signed-off-by: Stanislav Kinsbursky <skinsbursky at parallels.com>
>> Cc: stable at vger.kernel.org
>> ---
>>   net/sunrpc/clnt.c     |    5 +----
>>   net/sunrpc/rpc_pipe.c |    2 +-
>>   2 files changed, 2 insertions(+), 5 deletions(-)
>
> <snip>
>
>> diff --git a/net/sunrpc/rpc_pipe.c b/net/sunrpc/rpc_pipe.c
>> index c512448..efca2f7 100644
>> --- a/net/sunrpc/rpc_pipe.c
>> +++ b/net/sunrpc/rpc_pipe.c
>> @@ -1165,7 +1165,6 @@ static void rpc_kill_sb(struct super_block *sb)
>>   		goto out;
>>   	}
>>   	sn->pipefs_sb = NULL;
>> -	mutex_unlock(&sn->pipefs_sb_lock);
>>   	dprintk("RPC:       sending pipefs UMOUNT notification for net %p%s\n",
>>   		net, NET_NAME(net));
>>   	blocking_notifier_call_chain(&rpc_pipefs_notifier_list,
>> @@ -1173,6 +1172,7 @@ static void rpc_kill_sb(struct super_block *sb)
>>   					   sb);
>>   	put_net(net);
>>   out:
>> +	mutex_unlock(&sn->pipefs_sb_lock);
>
> Is this safe to do after the put_net()?
>

Sure it's not. Sorry.
Will send the patch once more.

>>   	kill_litter_super(sb);
>>   }
>>
>>
>
>


-- 
Best regards,
Stanislav Kinsbursky



More information about the Devel mailing list