On Mon, Nov 05 2018, Jeff Layton wrote:

> On Mon, 2018-11-05 at 12:30 +1100, NeilBrown wrote:
>> Rather than assuming all-zeros is sufficient, use the available API to
>> initialize the file_lock structure use for unlock.
>> VFS-level changes will soon make it important that the
>> list_heads in file_lock are always properly initialized.
>> 
>> Signed-off-by: NeilBrown <neilb@suse.com>
>> ---
>>  fs/gfs2/file.c |   10 +++++-----
>>  1 file changed, 5 insertions(+), 5 deletions(-)
>> 
>> diff --git a/fs/gfs2/file.c b/fs/gfs2/file.c
>> index 45a17b770d97..271f847705e3 100644
>> --- a/fs/gfs2/file.c
>> +++ b/fs/gfs2/file.c
>> @@ -1199,13 +1199,13 @@ static int do_flock(struct file *file, int cmd, struct file_lock *fl)
>>  	mutex_lock(&fp->f_fl_mutex);
>>  
>>  	if (gfs2_holder_initialized(fl_gh)) {
>> +		struct file_lock request;
>>  		if (fl_gh->gh_state == state)
>>  			goto out;
>> -		locks_lock_file_wait(file,
>> -				     &(struct file_lock) {
>> -					     .fl_type = F_UNLCK,
>> -					     .fl_flags = FL_FLOCK
>> -				     });
>> +		locks_init_lock(&request);
>> +		request.fl_type = F_UNLOCK;
>
> F_UNLCK ?
>
> The ocfs2 patch has the same bug.

Anyone would think that I hadn't even compile tested.....

This is true for OCFS2 :-( but I had actually compile-tested with GFS2
enabled.
But CONFIG_DLM *wasn't* enabled, so GFS2 was compiled without locking
support.
I guess there is a good reason that GFS2 doesn't require DLM.

Do you want me to resend the series, to will you just update those
patches.

Sorry about that,
NeilBrown


>
>> +		request.fl_flags = FL_FLOCK;
>> +		locks_lock_file_wait(file, &request);
>>  		gfs2_glock_dq(fl_gh);
>>  		gfs2_holder_reinit(state, flags, fl_gh);
>>  	} else {
>> 
>> 
>
> -- 
> Jeff Layton <jlayton@kernel.org>