[PATCH 0/2] xfsprog: fix xfs_inode lifetime issue

[PATCH 0/2] xfsprog: fix xfs_inode lifetime issue

[Mark Tinguely]

These patches clean up the xfs(_trans)_iput() and fix the
xfs_inode life time so that inodes with a extended attribute
fork can add an attributes.

Patch 1	removes the unused argument in _iput routines and
	introduces the IRELSE define for xfs_iput().
Patch 2 fixes the xfs_inode lifetime problem.

Mark.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 1/2] xfsprogs: remove unused argument in trans_iput

[Mark Tinguely]

[-- Attachment #1: xfsprogs-remove-extra-var-xfs_iput.patch --]
[-- Type: text/plain, Size: 4684 bytes --]

Remove the unused second argument to xfs_iput() and
xfs_trans_iput().

Introduce the define "IRELE()" and use in place of xfs_iput().
 
Signed-off-by: Mark Tinguely <tinguely@sgi.com>
---
 db/attrset.c     |    4 ++--
 include/libxfs.h |    6 ++++--
 libxfs/init.c    |    4 ++--
 libxfs/rdwr.c    |    2 +-
 libxfs/trans.c   |   11 +++++------
 mkfs/proto.c     |    2 +-
 repair/phase6.c  |    2 +-
 repair/phase7.c  |    2 +-
 8 files changed, 17 insertions(+), 16 deletions(-)

Index: b/db/attrset.c
===================================================================
--- a/db/attrset.c
+++ b/db/attrset.c
@@ -170,7 +170,7 @@ attr_set_f(
 out:
 	mp->m_flags &= ~LIBXFS_MOUNT_COMPAT_ATTR;
 	if (ip)
-		libxfs_iput(ip, 0);
+		IRELE(ip);
 	if (value)
 		free(value);
 	return 0;
@@ -244,6 +244,6 @@ attr_remove_f(
 out:
 	mp->m_flags &= ~LIBXFS_MOUNT_COMPAT_ATTR;
 	if (ip)
-		libxfs_iput(ip, 0);
+		IRELE(ip);
 	return 0;
 }
Index: b/include/libxfs.h
===================================================================
--- a/include/libxfs.h
+++ b/include/libxfs.h
@@ -532,7 +532,7 @@ extern xfs_buf_t	*libxfs_trans_getsb (xf
 
 extern int	libxfs_trans_iget (xfs_mount_t *, xfs_trans_t *, xfs_ino_t,
 				uint, uint, struct xfs_inode **);
-extern void	libxfs_trans_iput(xfs_trans_t *, struct xfs_inode *, uint);
+extern void	libxfs_trans_iput(xfs_trans_t *, struct xfs_inode *);
 extern void	libxfs_trans_ijoin (xfs_trans_t *, struct xfs_inode *, uint);
 extern void	libxfs_trans_ihold (xfs_trans_t *, struct xfs_inode *);
 extern void	libxfs_trans_ijoin_ref(xfs_trans_t *, struct xfs_inode *, int);
@@ -653,7 +653,9 @@ extern int	libxfs_iflush_int (xfs_inode_
 /* Inode Cache Interfaces */
 extern int	libxfs_iget (xfs_mount_t *, xfs_trans_t *, xfs_ino_t,
 				uint, xfs_inode_t **, xfs_daddr_t);
-extern void	libxfs_iput (xfs_inode_t *, uint);
+extern void	libxfs_iput (xfs_inode_t *);
+
+#define IRELE(ip) libxfs_iput(ip)
 
 /* Shared utility routines */
 extern unsigned int	libxfs_log2_roundup(unsigned int i);
Index: b/libxfs/init.c
===================================================================
--- a/libxfs/init.c
+++ b/libxfs/init.c
@@ -778,9 +778,9 @@ void
 libxfs_rtmount_destroy(xfs_mount_t *mp)
 {
 	if (mp->m_rsumip)
-		libxfs_iput(mp->m_rsumip, 0);
+		IRELE(mp->m_rsumip);
 	if (mp->m_rbmip)
-		libxfs_iput(mp->m_rbmip, 0);
+		IRELE(mp->m_rbmip);
 	mp->m_rsumip = mp->m_rbmip = NULL;
 }
 
Index: b/libxfs/rdwr.c
===================================================================
--- a/libxfs/rdwr.c
+++ b/libxfs/rdwr.c
@@ -1076,7 +1076,7 @@ libxfs_idestroy(xfs_inode_t *ip)
 }
 
 void
-libxfs_iput(xfs_inode_t *ip, uint lock_flags)
+libxfs_iput(xfs_inode_t *ip)
 {
 	if (ip->i_itemp)
 		kmem_zone_free(xfs_ili_zone, ip->i_itemp);
Index: b/libxfs/trans.c
===================================================================
--- a/libxfs/trans.c
+++ b/libxfs/trans.c
@@ -250,13 +250,12 @@ libxfs_trans_iget(
 void
 libxfs_trans_iput(
 	xfs_trans_t		*tp,
-	xfs_inode_t		*ip,
-	uint			lock_flags)
+	xfs_inode_t		*ip)
 {
 	xfs_inode_log_item_t	*iip;
 
 	if (tp == NULL) {
-		libxfs_iput(ip, lock_flags);
+		IRELE(ip);
 		return;
 	}
 
@@ -265,7 +264,7 @@ libxfs_trans_iput(
 	ASSERT(iip != NULL);
 	xfs_trans_del_item(&iip->ili_item);
 
-	libxfs_iput(ip, lock_flags);
+	IRELE(ip);
 }
 
 void
@@ -737,7 +736,7 @@ ili_done:
 		return;
 	}
 	/* free the inode */
-	libxfs_iput(ip, 0);
+	IRELE(ip);
 }
 
 static void
@@ -819,7 +818,7 @@ inode_item_unlock(
 
 	iip->ili_flags = 0;
 	if (!iip->ili_lock_flags)
-		libxfs_iput(ip, 0);
+		IRELE(ip);
 	else
 		iip->ili_lock_flags = 0;
 }
Index: b/mkfs/proto.c
===================================================================
--- a/mkfs/proto.c
+++ b/mkfs/proto.c
@@ -589,7 +589,7 @@ parseproto(
 				break;
 			parseproto(mp, ip, fsxp, pp, name);
 		}
-		libxfs_iput(ip, 0);
+		IRELE(ip);
 		return;
 	default:
 		ASSERT(0);
Index: b/repair/phase6.c
===================================================================
--- a/repair/phase6.c
+++ b/repair/phase6.c
@@ -2873,7 +2873,7 @@ process_dir_inode(
 					|XFS_TRANS_SYNC);
 		}
 	}
-	libxfs_iput(ip, 0);
+	IRELE(ip);
 }
 
 /*
Index: b/repair/phase7.c
===================================================================
--- a/repair/phase7.c
+++ b/repair/phase7.c
@@ -99,7 +99,7 @@ update_inode_nlinks(
 	set_nlinks(&ip->i_d, ino, nlinks, &dirty);
 
 	if (!dirty)  {
-		libxfs_trans_iput(tp, ip, 0);
+		libxfs_trans_iput(tp, ip);
 		libxfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES);
 	} else  {
 		libxfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);


_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 1/2] xfsprogs: remove unused argument in trans_iput

[Eric Sandeen]

On 4/30/14, 8:48 AM, Mark Tinguely wrote:
> Remove the unused second argument to xfs_iput() and
> xfs_trans_iput().
> 
> Introduce the define "IRELE()" and use in place of xfs_iput().

Why do this?  We had been moving away from the upper-case-macro-
redefined-to-a-function meme... what does this #define gain?

libxfs_iget/libxfs_iput pairs seem more obvious than
libxfs_iget/IRELE()...

Thanks,
-Eric

> Signed-off-by: Mark Tinguely <tinguely@sgi.com>
> ---
>  db/attrset.c     |    4 ++--
>  include/libxfs.h |    6 ++++--
>  libxfs/init.c    |    4 ++--
>  libxfs/rdwr.c    |    2 +-
>  libxfs/trans.c   |   11 +++++------
>  mkfs/proto.c     |    2 +-
>  repair/phase6.c  |    2 +-
>  repair/phase7.c  |    2 +-
>  8 files changed, 17 insertions(+), 16 deletions(-)
> 
> Index: b/db/attrset.c
> ===================================================================
> --- a/db/attrset.c
> +++ b/db/attrset.c
> @@ -170,7 +170,7 @@ attr_set_f(
>  out:
>  	mp->m_flags &= ~LIBXFS_MOUNT_COMPAT_ATTR;
>  	if (ip)
> -		libxfs_iput(ip, 0);
> +		IRELE(ip);
>  	if (value)
>  		free(value);
>  	return 0;
> @@ -244,6 +244,6 @@ attr_remove_f(
>  out:
>  	mp->m_flags &= ~LIBXFS_MOUNT_COMPAT_ATTR;
>  	if (ip)
> -		libxfs_iput(ip, 0);
> +		IRELE(ip);
>  	return 0;
>  }
> Index: b/include/libxfs.h
> ===================================================================
> --- a/include/libxfs.h
> +++ b/include/libxfs.h
> @@ -532,7 +532,7 @@ extern xfs_buf_t	*libxfs_trans_getsb (xf
>  
>  extern int	libxfs_trans_iget (xfs_mount_t *, xfs_trans_t *, xfs_ino_t,
>  				uint, uint, struct xfs_inode **);
> -extern void	libxfs_trans_iput(xfs_trans_t *, struct xfs_inode *, uint);
> +extern void	libxfs_trans_iput(xfs_trans_t *, struct xfs_inode *);
>  extern void	libxfs_trans_ijoin (xfs_trans_t *, struct xfs_inode *, uint);
>  extern void	libxfs_trans_ihold (xfs_trans_t *, struct xfs_inode *);
>  extern void	libxfs_trans_ijoin_ref(xfs_trans_t *, struct xfs_inode *, int);
> @@ -653,7 +653,9 @@ extern int	libxfs_iflush_int (xfs_inode_
>  /* Inode Cache Interfaces */
>  extern int	libxfs_iget (xfs_mount_t *, xfs_trans_t *, xfs_ino_t,
>  				uint, xfs_inode_t **, xfs_daddr_t);
> -extern void	libxfs_iput (xfs_inode_t *, uint);
> +extern void	libxfs_iput (xfs_inode_t *);
> +
> +#define IRELE(ip) libxfs_iput(ip)
>  
>  /* Shared utility routines */
>  extern unsigned int	libxfs_log2_roundup(unsigned int i);
> Index: b/libxfs/init.c
> ===================================================================
> --- a/libxfs/init.c
> +++ b/libxfs/init.c
> @@ -778,9 +778,9 @@ void
>  libxfs_rtmount_destroy(xfs_mount_t *mp)
>  {
>  	if (mp->m_rsumip)
> -		libxfs_iput(mp->m_rsumip, 0);
> +		IRELE(mp->m_rsumip);
>  	if (mp->m_rbmip)
> -		libxfs_iput(mp->m_rbmip, 0);
> +		IRELE(mp->m_rbmip);
>  	mp->m_rsumip = mp->m_rbmip = NULL;
>  }
>  
> Index: b/libxfs/rdwr.c
> ===================================================================
> --- a/libxfs/rdwr.c
> +++ b/libxfs/rdwr.c
> @@ -1076,7 +1076,7 @@ libxfs_idestroy(xfs_inode_t *ip)
>  }
>  
>  void
> -libxfs_iput(xfs_inode_t *ip, uint lock_flags)
> +libxfs_iput(xfs_inode_t *ip)
>  {
>  	if (ip->i_itemp)
>  		kmem_zone_free(xfs_ili_zone, ip->i_itemp);
> Index: b/libxfs/trans.c
> ===================================================================
> --- a/libxfs/trans.c
> +++ b/libxfs/trans.c
> @@ -250,13 +250,12 @@ libxfs_trans_iget(
>  void
>  libxfs_trans_iput(
>  	xfs_trans_t		*tp,
> -	xfs_inode_t		*ip,
> -	uint			lock_flags)
> +	xfs_inode_t		*ip)
>  {
>  	xfs_inode_log_item_t	*iip;
>  
>  	if (tp == NULL) {
> -		libxfs_iput(ip, lock_flags);
> +		IRELE(ip);
>  		return;
>  	}
>  
> @@ -265,7 +264,7 @@ libxfs_trans_iput(
>  	ASSERT(iip != NULL);
>  	xfs_trans_del_item(&iip->ili_item);
>  
> -	libxfs_iput(ip, lock_flags);
> +	IRELE(ip);
>  }
>  
>  void
> @@ -737,7 +736,7 @@ ili_done:
>  		return;
>  	}
>  	/* free the inode */
> -	libxfs_iput(ip, 0);
> +	IRELE(ip);
>  }
>  
>  static void
> @@ -819,7 +818,7 @@ inode_item_unlock(
>  
>  	iip->ili_flags = 0;
>  	if (!iip->ili_lock_flags)
> -		libxfs_iput(ip, 0);
> +		IRELE(ip);
>  	else
>  		iip->ili_lock_flags = 0;
>  }
> Index: b/mkfs/proto.c
> ===================================================================
> --- a/mkfs/proto.c
> +++ b/mkfs/proto.c
> @@ -589,7 +589,7 @@ parseproto(
>  				break;
>  			parseproto(mp, ip, fsxp, pp, name);
>  		}
> -		libxfs_iput(ip, 0);
> +		IRELE(ip);
>  		return;
>  	default:
>  		ASSERT(0);
> Index: b/repair/phase6.c
> ===================================================================
> --- a/repair/phase6.c
> +++ b/repair/phase6.c
> @@ -2873,7 +2873,7 @@ process_dir_inode(
>  					|XFS_TRANS_SYNC);
>  		}
>  	}
> -	libxfs_iput(ip, 0);
> +	IRELE(ip);
>  }
>  
>  /*
> Index: b/repair/phase7.c
> ===================================================================
> --- a/repair/phase7.c
> +++ b/repair/phase7.c
> @@ -99,7 +99,7 @@ update_inode_nlinks(
>  	set_nlinks(&ip->i_d, ino, nlinks, &dirty);
>  
>  	if (!dirty)  {
> -		libxfs_trans_iput(tp, ip, 0);
> +		libxfs_trans_iput(tp, ip);
>  		libxfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES);
>  	} else  {
>  		libxfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
> 
> 
> _______________________________________________
> xfs mailing list
> xfs@oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs
> 

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 1/2] xfsprogs: remove unused argument in trans_iput

[Mark Tinguely]

On 04/30/14 09:47, Eric Sandeen wrote:
> On 4/30/14, 8:48 AM, Mark Tinguely wrote:
>> >  Remove the unused second argument to xfs_iput() and
>> >  xfs_trans_iput().
>> >
>> >  Introduce the define "IRELE()" and use in place of xfs_iput().
> Why do this?  We had been moving away from the upper-case-macro-
> redefined-to-a-function meme... what does this #define gain?
>
> libxfs_iget/libxfs_iput pairs seem more obvious than
> libxfs_iget/IRELE()...
>
> Thanks,
> -Eric
>

To be consistent with the kernel code.

--Mark.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 1/2] xfsprogs: remove unused argument in trans_iput

[Eric Sandeen]

On 4/30/14, 9:50 AM, Mark Tinguely wrote:
> On 04/30/14 09:47, Eric Sandeen wrote:
>> On 4/30/14, 8:48 AM, Mark Tinguely wrote:
>>> >  Remove the unused second argument to xfs_iput() and
>>> >  xfs_trans_iput().
>>> >
>>> >  Introduce the define "IRELE()" and use in place of xfs_iput().
>> Why do this?  We had been moving away from the upper-case-macro-
>> redefined-to-a-function meme... what does this #define gain?
>>
>> libxfs_iget/libxfs_iput pairs seem more obvious than
>> libxfs_iget/IRELE()...
>>
>> Thanks,
>> -Eric
>>
> 
> To be consistent with the kernel code.

Gah, hohum, I'm not even going to try to explain why I thought
we didn't have it in kernel code.

sorry for the noise,
-Eric

> --Mark.
> 

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 1/2] xfsprogs: remove unused argument in trans_iput

[Christoph Hellwig]

On Wed, Apr 30, 2014 at 09:47:50AM -0500, Eric Sandeen wrote:
> On 4/30/14, 8:48 AM, Mark Tinguely wrote:
> > Remove the unused second argument to xfs_iput() and
> > xfs_trans_iput().
> > 
> > Introduce the define "IRELE()" and use in place of xfs_iput().
> 
> Why do this?  We had been moving away from the upper-case-macro-
> redefined-to-a-function meme... what does this #define gain?
> 
> libxfs_iget/libxfs_iput pairs seem more obvious than
> libxfs_iget/IRELE()...

Mostly to match the kernel that makes it do an iput underneath..

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 1/2] xfsprogs: remove unused argument in trans_iput

[Christoph Hellwig]

On Wed, Apr 30, 2014 at 08:48:45AM -0500, Mark Tinguely wrote:
> Remove the unused second argument to xfs_iput() and
> xfs_trans_iput().
> 
> Introduce the define "IRELE()" and use in place of xfs_iput().
>  
> Signed-off-by: Mark Tinguely <tinguely@sgi.com>

Looks good,

Reviewed-by: Christoph Hellwig <hch@lst.de>

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 2/2] xfsprogs: dont free xfs_inode until complete

[Mark Tinguely]

[-- Attachment #1: xfsprogs-fix-release-xfs_inode.patch --]
[-- Type: text/plain, Size: 13288 bytes --]

Originally, the xfs_inode are released upon the first
call to xfs_trans_cancel, xfs_trans_commit, or
inode_item_done. This code used the log item lock field
to prevent the release of the inode on the next call to
one of the above functions. This is a unusual use of the
log item lock field which is suppose to specify which lock
is to be release on transaction commit or cancel. User
space does not perform locking in transactions..

Unfortunately, this breaks any code that relies on multiple
transaction operations. For example, adding an extended
attribute to an inode that does not have an attribute fork
will fail:

 # xfs_db -x XFS_DEVICE
 xfs_db> inode INO_NUM
 xfs_db> attr_set newattribute

This patch does the following:
 1) Removes the iput from the transaction completion and
    requires that the xfs_inode allocators call IRELE()
    when they are done with the pointer. The real time
    inodes are pointed to by the xfs_mount and have a longer
    lifetime.
 2) Removes libxfs_trans_iput() because transaction entries
    are removed in transaction commit and cancel.
 3) Removes libxfs_trans_ihold() which is an obsolete interface.
 4) Removes the now unneeded ili_flags from the xfs_inode_log_item
    structure.
 
Signed-off-by: Mark Tinguely <tinguely@sgi.com>
---
 include/libxfs.h |    3 --
 libxfs/trans.c   |   59 +++++--------------------------------------------------
 libxfs/util.c    |    1 
 libxfs/xfs.h     |    1 
 mkfs/proto.c     |   16 +-------------
 repair/phase6.c  |   13 +++---------
 repair/phase7.c  |    2 -
 7 files changed, 13 insertions(+), 82 deletions(-)

Index: b/include/libxfs.h
===================================================================
--- a/include/libxfs.h
+++ b/include/libxfs.h
@@ -477,7 +477,6 @@ typedef struct xfs_inode_log_item {
 	unsigned int		ili_fields;		/* fields to be logged */
 	unsigned int		ili_last_fields;	/* fields when flushed*/
 	xfs_inode_log_format_t	ili_format;		/* logged structure */
-	int			ili_lock_flags;
 } xfs_inode_log_item_t;
 
 typedef struct xfs_buf_log_item {
@@ -532,9 +531,7 @@ extern xfs_buf_t	*libxfs_trans_getsb (xf
 
 extern int	libxfs_trans_iget (xfs_mount_t *, xfs_trans_t *, xfs_ino_t,
 				uint, uint, struct xfs_inode **);
-extern void	libxfs_trans_iput(xfs_trans_t *, struct xfs_inode *);
 extern void	libxfs_trans_ijoin (xfs_trans_t *, struct xfs_inode *, uint);
-extern void	libxfs_trans_ihold (xfs_trans_t *, struct xfs_inode *);
 extern void	libxfs_trans_ijoin_ref(xfs_trans_t *, struct xfs_inode *, int);
 extern void	libxfs_trans_log_inode (xfs_trans_t *, struct xfs_inode *,
 				uint);
Index: b/libxfs/trans.c
===================================================================
--- a/libxfs/trans.c
+++ b/libxfs/trans.c
@@ -110,7 +110,7 @@ libxfs_trans_roll(
 	/*
 	 * Commit the current transaction.
 	 * If this commit failed, then it'd just unlock those items that
-	 * are not marked ihold. That also means that a filesystem shutdown
+	 * are marked to be released. That also means that a filesystem shutdown
 	 * is in progress. The caller takes the responsibility to cancel
 	 * the duplicate transaction that gets returned.
 	 */
@@ -248,26 +248,6 @@ libxfs_trans_iget(
 }
 
 void
-libxfs_trans_iput(
-	xfs_trans_t		*tp,
-	xfs_inode_t		*ip)
-{
-	xfs_inode_log_item_t	*iip;
-
-	if (tp == NULL) {
-		IRELE(ip);
-		return;
-	}
-
-	ASSERT(ip->i_transp == tp);
-	iip = ip->i_itemp;
-	ASSERT(iip != NULL);
-	xfs_trans_del_item(&iip->ili_item);
-
-	IRELE(ip);
-}
-
-void
 libxfs_trans_ijoin(
 	xfs_trans_t		*tp,
 	xfs_inode_t		*ip,
@@ -300,7 +280,6 @@ libxfs_trans_ijoin_ref(
 	ASSERT(ip->i_itemp != NULL);
 
 	xfs_trans_ijoin(tp, ip, lock_flags);
-	ip->i_itemp->ili_lock_flags = lock_flags;
 
 #ifdef XACT_DEBUG
 	fprintf(stderr, "ijoin_ref'd inode %llu, transaction %p\n", ip->i_ino, tp);
@@ -308,21 +287,6 @@ libxfs_trans_ijoin_ref(
 }
 
 void
-libxfs_trans_ihold(
-	xfs_trans_t		*tp,
-	xfs_inode_t		*ip)
-{
-	ASSERT(ip->i_transp == tp);
-	ASSERT(ip->i_itemp != NULL);
-
-	ip->i_itemp->ili_lock_flags = 1;
-
-#ifdef XACT_DEBUG
-	fprintf(stderr, "ihold'd inode %llu, transaction %p\n", ip->i_ino, tp);
-#endif
-}
-
-void
 libxfs_trans_inode_alloc_buf(
 	xfs_trans_t		*tp,
 	xfs_buf_t		*bp)
@@ -701,7 +665,7 @@ inode_item_done(
 	if (!(iip->ili_fields & XFS_ILOG_ALL)) {
 		ip->i_transp = NULL;	/* disassociate from transaction */
 		iip->ili_flags = 0;	/* reset all flags */
-		goto ili_done;
+		return;
 	}
 
 	/*
@@ -711,7 +675,7 @@ inode_item_done(
 	if (error) {
 		fprintf(stderr, _("%s: warning - imap_to_bp failed (%d)\n"),
 			progname, error);
-		goto ili_done;
+		return;
 	}
 
 	XFS_BUF_SET_FSPRIVATE(bp, iip);
@@ -719,7 +683,7 @@ inode_item_done(
 	if (error) {
 		fprintf(stderr, _("%s: warning - iflush_int failed (%d)\n"),
 			progname, error);
-		goto ili_done;
+		return;
 	}
 
 	ip->i_transp = NULL;	/* disassociate from transaction */
@@ -727,16 +691,9 @@ inode_item_done(
 	XFS_BUF_SET_FSPRIVATE2(bp, NULL);	/* remove xact ptr */
 	libxfs_writebuf(bp, 0);
 #ifdef XACT_DEBUG
-	fprintf(stderr, "flushing dirty inode %llu, buffer %p (hold=%u)\n",
-			ip->i_ino, bp, iip->ili_lock_flags);
+	fprintf(stderr, "flushing dirty inode %llu, buffer %p\n",
+			ip->i_ino, bp);
 #endif
-ili_done:
-	if (iip->ili_lock_flags) {
-		iip->ili_lock_flags = 0;
-		return;
-	}
-	/* free the inode */
-	IRELE(ip);
 }
 
 static void
@@ -817,10 +774,6 @@ inode_item_unlock(
 	ip->i_transp = NULL;
 
 	iip->ili_flags = 0;
-	if (!iip->ili_lock_flags)
-		IRELE(ip);
-	else
-		iip->ili_lock_flags = 0;
 }
 
 /*
Index: b/libxfs/util.c
===================================================================
--- a/libxfs/util.c
+++ b/libxfs/util.c
@@ -595,7 +595,6 @@ libxfs_alloc_file_space(
 			break;
 		}
 		xfs_trans_ijoin(tp, ip, 0);
-		xfs_trans_ihold(tp, ip);
 
 		xfs_bmap_init(&free_list, &firstfsb);
 		error = xfs_bmapi_write(tp, ip, startoffset_fsb, allocatesize_fsb,
Index: b/libxfs/xfs.h
===================================================================
--- a/libxfs/xfs.h
+++ b/libxfs/xfs.h
@@ -292,7 +292,6 @@ roundup_64(__uint64_t x, __uint32_t y)
 #define xfs_trans_get_buf		libxfs_trans_get_buf
 #define xfs_trans_getsb			libxfs_trans_getsb
 #define xfs_trans_iget			libxfs_trans_iget
-#define xfs_trans_ihold			libxfs_trans_ihold
 #define xfs_trans_ijoin			libxfs_trans_ijoin
 #define xfs_trans_ijoin_ref		libxfs_trans_ijoin_ref
 #define xfs_trans_init			libxfs_trans_init
Index: b/mkfs/proto.c
===================================================================
--- a/mkfs/proto.c
+++ b/mkfs/proto.c
@@ -196,7 +196,6 @@ rsvfile(
 	tp = libxfs_trans_alloc(mp, 0);
 
 	libxfs_trans_ijoin(tp, ip, 0);
-	libxfs_trans_ihold(tp, ip);
 
 	ip->i_d.di_mode &= ~S_ISUID;
 
@@ -463,7 +462,6 @@ parseproto(
 		libxfs_trans_ijoin(tp, pip, 0);
 		xname.type = XFS_DIR3_FT_REG_FILE;
 		newdirent(mp, tp, pip, &xname, ip->i_ino, &first, &flist);
-		libxfs_trans_ihold(tp, pip);
 		break;
 
 	case IF_RESERVED:			/* pre-allocated space only */
@@ -480,7 +478,6 @@ parseproto(
 
 		xname.type = XFS_DIR3_FT_REG_FILE;
 		newdirent(mp, tp, pip, &xname, ip->i_ino, &first, &flist);
-		libxfs_trans_ihold(tp, pip);
 		libxfs_trans_log_inode(tp, ip, flags);
 
 		error = libxfs_bmap_finish(&tp, &flist, &committed);
@@ -488,6 +485,7 @@ parseproto(
 			fail(_("Pre-allocated file creation failed"), error);
 		libxfs_trans_commit(tp, 0);
 		rsvfile(mp, ip, llen);
+		IRELE(ip);
 		return;
 
 	case IF_BLOCK:
@@ -502,7 +500,6 @@ parseproto(
 		libxfs_trans_ijoin(tp, pip, 0);
 		xname.type = XFS_DIR3_FT_BLKDEV;
 		newdirent(mp, tp, pip, &xname, ip->i_ino, &first, &flist);
-		libxfs_trans_ihold(tp, pip);
 		flags |= XFS_ILOG_DEV;
 		break;
 
@@ -517,7 +514,6 @@ parseproto(
 		libxfs_trans_ijoin(tp, pip, 0);
 		xname.type = XFS_DIR3_FT_CHRDEV;
 		newdirent(mp, tp, pip, &xname, ip->i_ino, &first, &flist);
-		libxfs_trans_ihold(tp, pip);
 		flags |= XFS_ILOG_DEV;
 		break;
 
@@ -530,7 +526,6 @@ parseproto(
 		libxfs_trans_ijoin(tp, pip, 0);
 		xname.type = XFS_DIR3_FT_FIFO;
 		newdirent(mp, tp, pip, &xname, ip->i_ino, &first, &flist);
-		libxfs_trans_ihold(tp, pip);
 		break;
 	case IF_SYMLINK:
 		buf = getstr(pp);
@@ -544,7 +539,6 @@ parseproto(
 		libxfs_trans_ijoin(tp, pip, 0);
 		xname.type = XFS_DIR3_FT_SYMLINK;
 		newdirent(mp, tp, pip, &xname, ip->i_ino, &first, &flist);
-		libxfs_trans_ihold(tp, pip);
 		break;
 	case IF_DIRECTORY:
 		getres(tp, 0);
@@ -564,7 +558,6 @@ parseproto(
 			newdirent(mp, tp, pip, &xname, ip->i_ino,
 				  &first, &flist);
 			pip->i_d.di_nlink++;
-			libxfs_trans_ihold(tp, pip);
 			libxfs_trans_log_inode(tp, pip, XFS_ILOG_CORE);
 		}
 		newdirectory(mp, tp, ip, pip);
@@ -572,7 +565,6 @@ parseproto(
 		error = libxfs_bmap_finish(&tp, &flist, &committed);
 		if (error)
 			fail(_("Directory creation failed"), error);
-		libxfs_trans_ihold(tp, ip);
 		libxfs_trans_commit(tp, 0);
 		/*
 		 * RT initialization.  Do this here to ensure that
@@ -602,6 +594,7 @@ parseproto(
 			error);
 	}
 	libxfs_trans_commit(tp, 0);
+	IRELE(ip);
 }
 
 void
@@ -664,7 +657,6 @@ rtinit(
 	*(__uint64_t *)&rbmip->i_d.di_atime = 0;
 	libxfs_trans_log_inode(tp, rbmip, XFS_ILOG_CORE);
 	libxfs_mod_sb(tp, XFS_SB_RBMINO);
-	libxfs_trans_ihold(tp, rbmip);
 	mp->m_rbmip = rbmip;
 	error = libxfs_inode_alloc(&tp, NULL, S_IFREG, 1, 0,
 					&creds, &fsxattrs, &rsumip);
@@ -675,7 +667,6 @@ rtinit(
 	rsumip->i_d.di_size = mp->m_rsumsize;
 	libxfs_trans_log_inode(tp, rsumip, XFS_ILOG_CORE);
 	libxfs_mod_sb(tp, XFS_SB_RSUMINO);
-	libxfs_trans_ihold(tp, rsumip);
 	libxfs_trans_commit(tp, 0);
 	mp->m_rsumip = rsumip;
 	/*
@@ -688,7 +679,6 @@ rtinit(
 		res_failed(i);
 
 	libxfs_trans_ijoin(tp, rbmip, 0);
-	libxfs_trans_ihold(tp, rbmip);
 	bno = 0;
 	xfs_bmap_init(&flist, &first);
 	while (bno < mp->m_sb.sb_rbmblocks) {
@@ -725,7 +715,6 @@ rtinit(
 	if (i)
 		res_failed(i);
 	libxfs_trans_ijoin(tp, rsumip, 0);
-	libxfs_trans_ihold(tp, rsumip);
 	bno = 0;
 	xfs_bmap_init(&flist, &first);
 	while (bno < nsumblocks) {
@@ -761,7 +750,6 @@ rtinit(
 		if (i)
 			res_failed(i);
 		libxfs_trans_ijoin(tp, rbmip, 0);
-		libxfs_trans_ihold(tp, rbmip);
 		xfs_bmap_init(&flist, &first);
 		ebno = XFS_RTMIN(mp->m_sb.sb_rextents,
 			bno + NBBY * mp->m_sb.sb_blocksize);
Index: b/repair/phase6.c
===================================================================
--- a/repair/phase6.c
+++ b/repair/phase6.c
@@ -505,7 +505,6 @@ mk_rbmino(xfs_mount_t *mp)
 	 * commit changes
 	 */
 	libxfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
-	libxfs_trans_ihold(tp, ip);
 	libxfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES|XFS_TRANS_SYNC);
 
 	/*
@@ -762,7 +761,6 @@ mk_rsumino(xfs_mount_t *mp)
 	 * commit changes
 	 */
 	libxfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
-	libxfs_trans_ihold(tp, ip);
 	libxfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES|XFS_TRANS_SYNC);
 
 	/*
@@ -1024,6 +1022,8 @@ mk_orphanage(xfs_mount_t *mp)
 
 
 	libxfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES|XFS_TRANS_SYNC);
+	IRELE(ip);
+	IRELE(pip);
 	add_inode_reached(irec,ino_offset);
 
 	return(ino);
@@ -1221,6 +1221,8 @@ mv_orphanage(
 
 		libxfs_trans_commit(tp, XFS_TRANS_RELEASE_LOG_RES|XFS_TRANS_SYNC);
 	}
+	IRELE(ino_p);
+	IRELE(orphanage_ip);
 }
 
 static int
@@ -1291,7 +1293,6 @@ longform_dir2_rebuild(
 	if (error)
 		res_failed(error);
 	libxfs_trans_ijoin(tp, ip, 0);
-	libxfs_trans_ihold(tp, ip);
 
 	if ((error = libxfs_bmap_last_offset(tp, ip, &lastblock,
 						XFS_DATA_FORK)))
@@ -1331,7 +1332,6 @@ longform_dir2_rebuild(
 			res_failed(error);
 
 		libxfs_trans_ijoin(tp, ip, 0);
-		libxfs_trans_ihold(tp, ip);
 
 		xfs_bmap_init(&flist, &firstblock);
 		error = libxfs_dir_createname(tp, ip, &p->name, p->inum,
@@ -1389,7 +1389,6 @@ dir2_kill_block(
 	if (error)
 		res_failed(error);
 	libxfs_trans_ijoin(tp, ip, 0);
-	libxfs_trans_ihold(tp, ip);
 	libxfs_trans_bjoin(tp, bp);
 	memset(&args, 0, sizeof(args));
 	xfs_bmap_init(&flist, &firstblock);
@@ -1577,7 +1576,6 @@ longform_dir2_entry_check_data(
 	if (error)
 		res_failed(error);
 	libxfs_trans_ijoin(tp, ip, 0);
-	libxfs_trans_ihold(tp, ip);
 	libxfs_trans_bjoin(tp, bp);
 	libxfs_trans_bhold(tp, bp);
 	xfs_bmap_init(&flist, &firstblock);
@@ -2744,7 +2742,6 @@ process_dir_inode(
 				res_failed(error);
 
 			libxfs_trans_ijoin(tp, ip, 0);
-			libxfs_trans_ihold(tp, ip);
 
 			shortform_dir2_entry_check(mp, ino, ip, &dirty,
 						irec, ino_offset,
@@ -2792,7 +2789,6 @@ process_dir_inode(
 			res_failed(error);
 
 		libxfs_trans_ijoin(tp, ip, 0);
-		libxfs_trans_ihold(tp, ip);
 
 		xfs_bmap_init(&flist, &first);
 
@@ -2854,7 +2850,6 @@ process_dir_inode(
 				res_failed(error);
 
 			libxfs_trans_ijoin(tp, ip, 0);
-			libxfs_trans_ihold(tp, ip);
 
 			xfs_bmap_init(&flist, &first);
 
Index: b/repair/phase7.c
===================================================================
--- a/repair/phase7.c
+++ b/repair/phase7.c
@@ -99,7 +99,6 @@ update_inode_nlinks(
 	set_nlinks(&ip->i_d, ino, nlinks, &dirty);
 
 	if (!dirty)  {
-		libxfs_trans_iput(tp, ip);
 		libxfs_trans_cancel(tp, XFS_TRANS_RELEASE_LOG_RES);
 	} else  {
 		libxfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
@@ -113,6 +112,7 @@ update_inode_nlinks(
 
 		ASSERT(error == 0);
 	}
+	IRELE(ip);
 }
 
 void


_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 2/2] xfsprogs: dont free xfs_inode until complete

[Christoph Hellwig]

On Wed, Apr 30, 2014 at 08:48:46AM -0500, Mark Tinguely wrote:
> Originally, the xfs_inode are released upon the first
> call to xfs_trans_cancel, xfs_trans_commit, or
> inode_item_done. This code used the log item lock field
> to prevent the release of the inode on the next call to
> one of the above functions. This is a unusual use of the
> log item lock field which is suppose to specify which lock
> is to be release on transaction commit or cancel. User
> space does not perform locking in transactions..
> 
> Unfortunately, this breaks any code that relies on multiple
> transaction operations. For example, adding an extended
> attribute to an inode that does not have an attribute fork
> will fail:
> 
>  # xfs_db -x XFS_DEVICE
>  xfs_db> inode INO_NUM
>  xfs_db> attr_set newattribute
> 
> This patch does the following:
>  1) Removes the iput from the transaction completion and
>     requires that the xfs_inode allocators call IRELE()
>     when they are done with the pointer. The real time
>     inodes are pointed to by the xfs_mount and have a longer
>     lifetime.

Makes sense, the kernel hasn't done an iput during transaction
completion for a long time.

>  2) Removes libxfs_trans_iput() because transaction entries
>     are removed in transaction commit and cancel.

Also matches the kernel and makes sense.

>  3) Removes libxfs_trans_ihold() which is an obsolete interface.

Ditto.

>  4) Removes the now unneeded ili_flags from the xfs_inode_log_item
>     structure.

ili_lock_flags, not ili_flags.  Given that we never lock inodes in
userspace there is no need to unlock them during transaction commit,
and no need to track this.  There's also no shared code that references
this field, so it can safely go.

The patch looks good to me, and having consistent behavior with the
kernel code is very useful to have!

Reviewed-by: Christoph Hellwig <hch@lst.de>

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 0/2] xfsprog: fix xfs_inode lifetime issue

[Dave Chinner]

On Wed, Apr 30, 2014 at 08:48:44AM -0500, Mark Tinguely wrote:
> These patches clean up the xfs(_trans)_iput() and fix the
> xfs_inode life time so that inodes with a extended attribute
> fork can add an attributes.
> 
> Patch 1	removes the unused argument in _iput routines and
> 	introduces the IRELSE define for xfs_iput().
> Patch 2 fixes the xfs_inode lifetime problem.

These look OK, but I'm going to leave them until after the 3.2.0
release. Thanks for doing this, Mark!

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs