[PATCH] Changes to inline comments in posix locking

[Jeremy Allison]

On Fri, Aug 03, 2018 at 12:11:20PM +0530, Anoop C S via samba-technical wrote:
> Hi,
> 
> Attached patch attempts to make the intention clear inside source3/locking/posix.c by replacing and
> modifying some inline comments around management of entries inside in-memory posix_pending_close_db
> database.
> 
> Reviews are appreciated.

Reviewed-by: Jeremy Allison <jra at samba.org>

Can I get a second Team reviewer ?

Thanks,

	Jeremy.

> From 09eb282d798963591d03cc989eb0eb32ffb26a0c Mon Sep 17 00:00:00 2001
> From: Anoop C S <anoopcs at redhat.com>
> Date: Fri, 3 Aug 2018 11:51:25 +0530
> Subject: [PATCH] s3/locking: Corrections and improvements to inline comments
> 
> Signed-off-by: Anoop C S <anoopcs at redhat.com>
> ---
>  source3/locking/posix.c | 35 ++++++++++++++++++++++-------------
>  1 file changed, 22 insertions(+), 13 deletions(-)
> 
> diff --git a/source3/locking/posix.c b/source3/locking/posix.c
> index 13992663280..79c33cfb0fa 100644
> --- a/source3/locking/posix.c
> +++ b/source3/locking/posix.c
> @@ -410,20 +410,12 @@ bool posix_locking_end(void)
>  }
>  
>  /****************************************************************************
> - Next - the functions that deal with storing fd's that have outstanding
> - POSIX locks when closed.
> -****************************************************************************/
> -
> -/****************************************************************************
> - The records in posix_pending_close_db are composed of an array of
> - ints keyed by dev/ino pair. Those ints are the fd's that were open on
> - this dev/ino pair that should have been closed, but can't as the lock
> - ref count is non zero.
> + Next - the functions that deal with reference count of number of locks open
> + on a dev/ino pair.
>  ****************************************************************************/
>  
>  /****************************************************************************
> - Keep a reference count of the number of locks open on this dev/ino
> - pair. Creates entry if it doesn't exist.
> + Increase the lock ref count. Creates lock_ref_count entry if it doesn't exist.
>  ****************************************************************************/
>  
>  static void increment_lock_ref_count(const files_struct *fsp)
> @@ -443,6 +435,10 @@ static void increment_lock_ref_count(const files_struct *fsp)
>  		  fsp_str_dbg(fsp), (int)lock_ref_count));
>  }
>  
> +/****************************************************************************
> + Reduce the lock ref count.
> +****************************************************************************/
> +
>  static void decrement_lock_ref_count(const files_struct *fsp)
>  {
>  	struct lock_ref_count_key tmp;
> @@ -500,6 +496,18 @@ static void delete_lock_ref_count(const files_struct *fsp)
>  		  fsp_str_dbg(fsp)));
>  }
>  
> +/****************************************************************************
> + Next - the functions that deal with storing fd's that have outstanding
> + POSIX locks when closed.
> +****************************************************************************/
> +
> +/****************************************************************************
> + The records in posix_pending_close_db are composed of an array of
> + ints keyed by dev/ino pair. Those ints are the fd's that were open on
> + this dev/ino pair that should have been closed, but can't as the lock
> + ref count is non zero.
> +****************************************************************************/
> +
>  /****************************************************************************
>   Add an fd to the pending close tdb.
>  ****************************************************************************/
> @@ -618,8 +626,9 @@ int fd_close_posix(const struct files_struct *fsp)
>  
>  		/*
>  		 * There are outstanding locks on this dev/inode pair on
> -		 * other fds. Add our fd to the pending close tdb and set
> -		 * fsp->fh->fd to -1.
> +		 * other fds. Add our fd to the pending close db. We also
> +		 * set fsp->fh->fd to -1 inside fd_close() after returning
> +		 * from VFS layer.
>  		 */
>  
>  		add_fd_to_close_entry(fsp);
> -- 
> 2.17.1
>