Re: [PATCH v5 4/4] ocfs2: fix sparse warnings

From: Heming Zhao <heming.zhao@suse.com>
To: Joseph Qi <joseph.qi@linux.alibaba.com>
Cc: ocfs2-devel@lists.linux.dev, ailiop@suse.com
Subject: Re: [PATCH v5 4/4] ocfs2: fix sparse warnings
Date: Thu, 28 Mar 2024 18:50:24 +0800	[thread overview]
Message-ID: <bb3fdfc2-ad16-4907-9d20-d3696c6f1a86@suse.com> (raw)
In-Reply-To: <b1982b17-0674-4080-8ed4-de2c0107c7f3@linux.alibaba.com>

On 3/28/24 17:01, Joseph Qi wrote:
> 
> 
> On 3/28/24 4:29 PM, Heming Zhao wrote:
>> 1.
>> fs/ocfs2/localalloc.c:1224:41: warning: incorrect type in argument 1 (different base types)
>> fs/ocfs2/localalloc.c:1224:41:    expected unsigned long long val1
>> fs/ocfs2/localalloc.c:1224:41:    got restricted __le32 [usertype] la_bm_off
>>
>> 2.
>> fs/ocfs2/export.c:258:32: warning: cast to restricted __le32
>> fs/ocfs2/export.c:259:33: warning: cast to restricted __le32
>> fs/ocfs2/export.c:260:32: warning: cast to restricted __le32
>> fs/ocfs2/export.c:272:32: warning: cast to restricted __le32
>> fs/ocfs2/export.c:273:33: warning: cast to restricted __le32
>> fs/ocfs2/export.c:274:32: warning: cast to restricted __le32
>>
>> 3.
>> fs/ocfs2/inode.c:1623:13: warning: context imbalance in 'ocfs2_inode_cache_lock' - wrong count at exit
>> fs/ocfs2/inode.c:1630:13: warning: context imbalance in 'ocfs2_inode_cache_unlock' - unexpected unlock
>>
>> 4.
>> fs/ocfs2/refcounttree.c:633:27: warning: incorrect type in assignment (different base types)
>> fs/ocfs2/refcounttree.c:633:27:    expected restricted __le32 [usertype] rf_generation
>> fs/ocfs2/refcounttree.c:633:27:    got unsigned int
>>
>> 5.
>> fs/ocfs2/dlm/dlmdomain.c:1316:20: warning: context imbalance in 'dlm_query_nodeinfo_handler' - different lock contexts for basic block
>>
>> 6.
>> fs/ocfs2/dlm/dlmrecovery.c:2950:9: warning: context imbalance in 'dlm_finalize_reco_handler' - different lock contexts for basic block
>>
>> Signed-off-by: Heming Zhao <heming.zhao@suse.com>
>> ---
>> v5: (first version)
>> - localalloc.c fix (No: 1) is split from v4 patch 3/3.
>> - new sparse warnings fixes (No: 2~6) for other files.
>>
>> ---
>>   fs/ocfs2/dlm/dlmdomain.c   | 11 +++++------
>>   fs/ocfs2/dlm/dlmrecovery.c |  4 ++++
>>   fs/ocfs2/export.c          | 12 ++++++------
>>   fs/ocfs2/inode.c           |  2 ++
>>   fs/ocfs2/localalloc.c      |  4 ++--
>>   fs/ocfs2/refcounttree.c    |  2 +-
>>   6 files changed, 20 insertions(+), 15 deletions(-)
>>
>> diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c
>> index 5c04dde99981..2e0a2f338282 100644
>> --- a/fs/ocfs2/dlm/dlmdomain.c
>> +++ b/fs/ocfs2/dlm/dlmdomain.c
>> @@ -1274,7 +1274,7 @@ static int dlm_query_nodeinfo_handler(struct o2net_msg *msg, u32 len,
>>   {
>>   	struct dlm_query_nodeinfo *qn;
>>   	struct dlm_ctxt *dlm = NULL;
>> -	int locked = 0, status = -EINVAL;
>> +	int status = -EINVAL;
>>   
>>   	qn = (struct dlm_query_nodeinfo *) msg->buf;
>>   
>> @@ -1290,12 +1290,11 @@ static int dlm_query_nodeinfo_handler(struct o2net_msg *msg, u32 len,
>>   	}
>>   
>>   	spin_lock(&dlm->spinlock);
>> -	locked = 1;
>>   	if (dlm->joining_node != qn->qn_nodenum) {
>>   		mlog(ML_ERROR, "Node %d queried nodes on domain %s but "
>>   		     "joining node is %d\n", qn->qn_nodenum, qn->qn_domain,
>>   		     dlm->joining_node);
>> -		goto bail;
>> +		goto unlock;
>>   	}
>>   
>>   	/* Support for node query was added in 1.1 */
>> @@ -1305,14 +1304,14 @@ static int dlm_query_nodeinfo_handler(struct o2net_msg *msg, u32 len,
>>   		     "but active dlm protocol is %d.%d\n", qn->qn_nodenum,
>>   		     qn->qn_domain, dlm->dlm_locking_proto.pv_major,
>>   		     dlm->dlm_locking_proto.pv_minor);
>> -		goto bail;
>> +		goto unlock;
>>   	}
>>   
>>   	status = dlm_match_nodes(dlm, qn);
>>   
>> +unlock:
>> +	spin_unlock(&dlm->spinlock);
>>   bail:
>> -	if (locked)
>> -		spin_unlock(&dlm->spinlock);
>>   	spin_unlock(&dlm_domain_lock);
>>   
>>   	return status;
>> diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c
>> index 50da8af988c1..40a0e5df3de0 100644
>> --- a/fs/ocfs2/dlm/dlmrecovery.c
>> +++ b/fs/ocfs2/dlm/dlmrecovery.c
>> @@ -2917,8 +2917,11 @@ int dlm_finalize_reco_handler(struct o2net_msg *msg, u32 len, void *data,
>>   		BUG();
>>   	}
>>   
>> +	spin_unlock(&dlm->spinlock);
>> +
> 
> I'm not sure unlock first and then lock is a right way.
> I'd rather leave it as of now.
> 
> Joseph

Me too, I will restore this function code in v6.
Thanks for reviewing.

-Heming