xfs_growfs fix backport for 2.6.16.y

PATCH Ed Cashin 2 participants 5 messages Aug 25, 2008 – Sep 5, 2008 XFS filesystem

Share & tools

Ed Cashin

author

2008-08-25 15:38 UTC #0 lore raw

Dave Chinner, hello.

Your fix for the counters that had been preventing an XFS from growing by more than two terabytes went into 2.6.21, after the introduction of the per-cpu in-core superblock counters. The per-cpu in-core superblock counters are not in the latest version of the long-lived stable kernel 2.6.16.y, maintained by Adrian Bunk.

The problem with growing by more than two terabytes appears to be there, though. The 2.6.16.62 kernel will not allow me to grow an XFS by, e.g., 10 terabytes, so xfs_growfs does not report any change in the data blocks.

I backported your fix,

commit 20f4ebf2bf2f57c1a9abb3655391336cc90314b3
Author: David Chinner <[email protected]>
Date: Sat Feb 10 18:36:10 2007 +1100

[XFS] Make growfs work for amounts greater than 2TB

The free block modification code has a 32bit interface, limiting the size the filesystem can be grown even on 64 bit machines. On 32 bit machines, there are other 32bit variables in transaction structures and interfaces that need to be expanded to allow this to work.

SGI-PV: 959978
SGI-Modid: xfs-linux-melb:xfs-kern:27894a

Signed-off-by David Chinner <[email protected]>

Signed-off-by Christoph Hellwig <[email protected]>

Signed-off-by Tim Shimmin <[email protected]>

... to the 2.6.16.y git tree, and the result is included below. When I apply this backported fix to 2.6.16.62, I can grow an online XFS by 10 terabytes without any trouble.

Do you see any problems with this backport? If not, I will submit it for inclusion in the next 2.6.16.y release.

-------------

merge 20f4ebf2: [XFS] Make growfs work for amounts greater than 2TB

This is a fix from David Chinner that went into 2.6.21. It followed the introduction of the per-cpu superblock counters, but 2.6.16.y does not have those.

SGI-PV: 959978
SGI-Modid: xfs-linux-melb:xfs-kern:27894a

fs/xfs/xfs_bmap.c +13 ~~-13~~

index 70625e5..18e57b6 100644
@@ -687,7 +687,7 @@ xfs_bmap_add_extent(
 		ASSERT(nblks <= da_old);
 		if (nblks < da_old)
 			xfs_mod_incore_sb(ip->i_mount, XFS_SBS_FDBLOCKS,
-				(int)(da_old - nblks), rsvd);
+				(int64_t)(da_old - nblks), rsvd);
 	}
 	/*
 	 * Clear out the allocated field, done with it now in any case.

@@ -1180,7 +1180,7 @@ xfs_bmap_add_extent_delay_real(
 		diff = (int)(temp + temp2 - STARTBLOCKVAL(PREV.br_startblock) -
 			(cur ? cur->bc_private.b.allocated : 0));
 		if (diff > 0 &&
-		    xfs_mod_incore_sb(ip->i_mount, XFS_SBS_FDBLOCKS, -diff, rsvd)) {
+		    xfs_mod_incore_sb(ip->i_mount, XFS_SBS_FDBLOCKS, -((int64_t)diff), rsvd)) {
 			/*
 			 * Ick gross gag me with a spoon.
 			 */

@@ -1191,7 +1191,7 @@ xfs_bmap_add_extent_delay_real(
 					diff--;
 					if (!diff ||
 					    !xfs_mod_incore_sb(ip->i_mount,
-						    XFS_SBS_FDBLOCKS, -diff, rsvd))
+						    XFS_SBS_FDBLOCKS, -((int64_t)diff), rsvd))
 						break;
 				}
 				if (temp2) {

@@ -1199,7 +1199,7 @@ xfs_bmap_add_extent_delay_real(
 					diff--;
 					if (!diff ||
 					    !xfs_mod_incore_sb(ip->i_mount,
-						    XFS_SBS_FDBLOCKS, -diff, rsvd))
+						    XFS_SBS_FDBLOCKS, -((int64_t)diff), rsvd))
 						break;
 				}
 			}

@@ -1916,7 +1916,7 @@ xfs_bmap_add_extent_hole_delay(
 	if (oldlen != newlen) {
 		ASSERT(oldlen > newlen);
 		xfs_mod_incore_sb(ip->i_mount, XFS_SBS_FDBLOCKS,
-			(int)(oldlen - newlen), rsvd);
+			(int64_t)(oldlen - newlen), rsvd);
 		/*
 		 * Nothing to do for disk quota accounting here.
 		 */

@@ -3187,7 +3187,7 @@ xfs_bmap_del_extent(
 	 */
 	ASSERT(da_old >= da_new);
 	if (da_old > da_new)
-		xfs_mod_incore_sb(mp, XFS_SBS_FDBLOCKS, (int)(da_old - da_new),
+		xfs_mod_incore_sb(mp, XFS_SBS_FDBLOCKS, (int64_t)(da_old - da_new),
 			rsvd);
 done:
 	*logflagsp = flags;

@@ -4875,24 +4875,24 @@ xfs_bmapi(
 				if (rt) {
 					error = xfs_mod_incore_sb(mp,
 							XFS_SBS_FREXTENTS,
-							-(extsz), rsvd);
+							-((int64_t)extsz), rsvd);
 				} else {
 					error = xfs_mod_incore_sb(mp,
 							XFS_SBS_FDBLOCKS,
-							-(alen), rsvd);
+							-((int64_t)alen), rsvd);
 				}
 				if (!error) {
 					error = xfs_mod_incore_sb(mp,
 							XFS_SBS_FDBLOCKS,
-							-(indlen), rsvd);
+							-((int64_t)indlen), rsvd);
 					if (error && rt)
 						xfs_mod_incore_sb(mp,
 							XFS_SBS_FREXTENTS,
-							extsz, rsvd);
+							(int64_t)extsz, rsvd);
 					else if (error)
 						xfs_mod_incore_sb(mp,
 							XFS_SBS_FDBLOCKS,
-							alen, rsvd);
+							(int64_t)alen, rsvd);
 				}
 
 				if (error) {

@@ -5538,13 +5538,13 @@ xfs_bunmapi(
 				rtexts = XFS_FSB_TO_B(mp, del.br_blockcount);
 				do_div(rtexts, mp->m_sb.sb_rextsize);
 				xfs_mod_incore_sb(mp, XFS_SBS_FREXTENTS,
-						(int)rtexts, rsvd);
+						(int64_t)rtexts, rsvd);
 				(void)XFS_TRANS_RESERVE_QUOTA_NBLKS(mp,
 					NULL, ip, -((long)del.br_blockcount), 0,
 					XFS_QMOPT_RES_RTBLKS);
 			} else {
 				xfs_mod_incore_sb(mp, XFS_SBS_FDBLOCKS,
-						(int)del.br_blockcount, rsvd);
+						(int64_t)del.br_blockcount, rsvd);
 				(void)XFS_TRANS_RESERVE_QUOTA_NBLKS(mp,
 					NULL, ip, -((long)del.br_blockcount), 0,
 					XFS_QMOPT_RES_REGBLKS);

fs/xfs/xfs_mount.c +10 -3

index 62188ea..ab3d014 100644
@@ -1238,8 +1238,11 @@ xfs_mod_sb(xfs_trans_t *tp, __int64_t fields)
  * The SB_LOCK must be held when this routine is called.
  */
 STATIC int
-xfs_mod_incore_sb_unlocked(xfs_mount_t *mp, xfs_sb_field_t field,
-			int delta, int rsvd)
+xfs_mod_incore_sb_unlocked(
+	xfs_mount_t	*mp,
+	xfs_sb_field_t	field,
+	int64_t		delta,
+	int		rsvd)
 {
 	int		scounter;	/* short counter for 32 bit fields */
 	long long	lcounter;	/* long counter for 64 bit fields */

@@ -1401,7 +1404,11 @@ xfs_mod_incore_sb_unlocked(xfs_mount_t *mp, xfs_sb_field_t field,
  * routine to do the work.
  */
 int
-xfs_mod_incore_sb(xfs_mount_t *mp, xfs_sb_field_t field, int delta, int rsvd)
+xfs_mod_incore_sb(
+	xfs_mount_t	*mp,
+	xfs_sb_field_t	field,
+	int64_t		delta,
+	int		rsvd)
 {
 	unsigned long	s;
 	int	status;

fs/xfs/xfs_mount.h +3 -2

index cd3cf96..100d5b9 100644
@@ -526,10 +526,11 @@ xfs_daddr_to_agbno(struct xfs_mount *mp, xfs_daddr_t d)
 
 /*
  * This structure is for use by the xfs_mod_incore_sb_batch() routine.
+ * xfs_growfs can specify a few fields which are more than int limit
  */
 typedef struct xfs_mod_sb {
 	xfs_sb_field_t	msb_field;	/* Field to modify, see below */
-	int		msb_delta;	/* Change to make to specified field */
+	int64_t		msb_delta;	/* Change to make to specified field */
 } xfs_mod_sb_t;
 
 #define	XFS_MOUNT_ILOCK(mp)	mutex_lock(&((mp)->m_ilock))

@@ -547,7 +548,7 @@ extern int	xfs_unmountfs(xfs_mount_t *, struct cred *);
 extern void	xfs_unmountfs_close(xfs_mount_t *, struct cred *);
 extern int	xfs_unmountfs_writesb(xfs_mount_t *);
 extern int	xfs_unmount_flush(xfs_mount_t *, int);
-extern int	xfs_mod_incore_sb(xfs_mount_t *, xfs_sb_field_t, int, int);
+extern int	xfs_mod_incore_sb(xfs_mount_t *, xfs_sb_field_t, int64_t, int);
 extern int	xfs_mod_incore_sb_batch(xfs_mount_t *, xfs_mod_sb_t *,
 			uint, int);
 extern struct xfs_buf *xfs_getsb(xfs_mount_t *, int);

fs/xfs/xfs_trans.c +16 ~~-16~~

index d3d714e..68b5895 100644
@@ -232,7 +232,7 @@ xfs_trans_reserve(
 	 */
 	if (blocks > 0) {
 		error = xfs_mod_incore_sb(tp->t_mountp, XFS_SBS_FDBLOCKS,
-					  -blocks, rsvd);
+					  -((int64_t)blocks), rsvd);
 		if (error != 0) {
                         PFLAGS_RESTORE_FSTRANS(&tp->t_pflags);
 			return (XFS_ERROR(ENOSPC));

@@ -273,7 +273,7 @@ xfs_trans_reserve(
 	 */
 	if (rtextents > 0) {
 		error = xfs_mod_incore_sb(tp->t_mountp, XFS_SBS_FREXTENTS,
-					  -rtextents, rsvd);
+					  -((int64_t)rtextents), rsvd);
 		if (error) {
 			error = XFS_ERROR(ENOSPC);
 			goto undo_log;

@@ -303,7 +303,7 @@ undo_log:
 undo_blocks:
 	if (blocks > 0) {
 		(void) xfs_mod_incore_sb(tp->t_mountp, XFS_SBS_FDBLOCKS,
-					 blocks, rsvd);
+					 (int64_t)blocks, rsvd);
 		tp->t_blk_res = 0;
 	}

@@ -325,7 +325,7 @@ void
 xfs_trans_mod_sb(
 	xfs_trans_t	*tp,
 	uint		field,
-	long		delta)
+	int64_t		delta)
 {
 
 	switch (field) {

@@ -556,62 +556,62 @@ xfs_trans_unreserve_and_mod_sb(
 	if (tp->t_flags & XFS_TRANS_SB_DIRTY) {
 		if (tp->t_icount_delta != 0) {
 			msbp->msb_field = XFS_SBS_ICOUNT;
-			msbp->msb_delta = (int)tp->t_icount_delta;
+			msbp->msb_delta = tp->t_icount_delta;
 			msbp++;
 		}
 		if (tp->t_ifree_delta != 0) {
 			msbp->msb_field = XFS_SBS_IFREE;
-			msbp->msb_delta = (int)tp->t_ifree_delta;
+			msbp->msb_delta = tp->t_ifree_delta;
 			msbp++;
 		}
 		if (tp->t_fdblocks_delta != 0) {
 			msbp->msb_field = XFS_SBS_FDBLOCKS;
-			msbp->msb_delta = (int)tp->t_fdblocks_delta;
+			msbp->msb_delta = tp->t_fdblocks_delta;
 			msbp++;
 		}
 		if (tp->t_frextents_delta != 0) {
 			msbp->msb_field = XFS_SBS_FREXTENTS;
-			msbp->msb_delta = (int)tp->t_frextents_delta;
+			msbp->msb_delta = tp->t_frextents_delta;
 			msbp++;
 		}
 		if (tp->t_dblocks_delta != 0) {
 			msbp->msb_field = XFS_SBS_DBLOCKS;
-			msbp->msb_delta = (int)tp->t_dblocks_delta;
+			msbp->msb_delta = tp->t_dblocks_delta;
 			msbp++;
 		}
 		if (tp->t_agcount_delta != 0) {
 			msbp->msb_field = XFS_SBS_AGCOUNT;
-			msbp->msb_delta = (int)tp->t_agcount_delta;
+			msbp->msb_delta = tp->t_agcount_delta;
 			msbp++;
 		}
 		if (tp->t_imaxpct_delta != 0) {
 			msbp->msb_field = XFS_SBS_IMAX_PCT;
-			msbp->msb_delta = (int)tp->t_imaxpct_delta;
+			msbp->msb_delta = tp->t_imaxpct_delta;
 			msbp++;
 		}
 		if (tp->t_rextsize_delta != 0) {
 			msbp->msb_field = XFS_SBS_REXTSIZE;
-			msbp->msb_delta = (int)tp->t_rextsize_delta;
+			msbp->msb_delta = tp->t_rextsize_delta;
 			msbp++;
 		}
 		if (tp->t_rbmblocks_delta != 0) {
 			msbp->msb_field = XFS_SBS_RBMBLOCKS;
-			msbp->msb_delta = (int)tp->t_rbmblocks_delta;
+			msbp->msb_delta = tp->t_rbmblocks_delta;
 			msbp++;
 		}
 		if (tp->t_rblocks_delta != 0) {
 			msbp->msb_field = XFS_SBS_RBLOCKS;
-			msbp->msb_delta = (int)tp->t_rblocks_delta;
+			msbp->msb_delta = tp->t_rblocks_delta;
 			msbp++;
 		}
 		if (tp->t_rextents_delta != 0) {
 			msbp->msb_field = XFS_SBS_REXTENTS;
-			msbp->msb_delta = (int)tp->t_rextents_delta;
+			msbp->msb_delta = tp->t_rextents_delta;
 			msbp++;
 		}
 		if (tp->t_rextslog_delta != 0) {
 			msbp->msb_field = XFS_SBS_REXTSLOG;
-			msbp->msb_delta = (int)tp->t_rextslog_delta;
+			msbp->msb_delta = tp->t_rextslog_delta;
 			msbp++;
 		}
 	}

fs/xfs/xfs_trans.h +21 ~~-21~~

60 lines not on this page · raw message

Signature · 1 line

  Ed Cashin <[email protected]>

commit b4558635478e0c4d0f0761fc6287c7616f8f271f

Author: Ed L. Cashin <[email protected]>
Date:   Fri Aug 22 16:46:49 2008 -0400

merge 20f4ebf2: [XFS] Make growfs work for amounts greater than 2TB

This is a fix from David Chinner that went into 2.6.21. It followed the introduction of the per-cpu superblock counters, but 2.6.16.y does not have those.

SGI-PV: 959978
SGI-Modid: xfs-linux-melb:xfs-kern:27894a

fs/xfs/xfs_bmap.c +13 ~~-13~~

index 70625e5..18e57b6 100644
@@ -687,7 +687,7 @@ xfs_bmap_add_extent(
 		ASSERT(nblks <= da_old);
 		if (nblks < da_old)
 			xfs_mod_incore_sb(ip->i_mount, XFS_SBS_FDBLOCKS,
-				(int)(da_old - nblks), rsvd);
+				(int64_t)(da_old - nblks), rsvd);
 	}
 	/*
 	 * Clear out the allocated field, done with it now in any case.

@@ -1180,7 +1180,7 @@ xfs_bmap_add_extent_delay_real(
 		diff = (int)(temp + temp2 - STARTBLOCKVAL(PREV.br_startblock) -
 			(cur ? cur->bc_private.b.allocated : 0));
 		if (diff > 0 &&
-		    xfs_mod_incore_sb(ip->i_mount, XFS_SBS_FDBLOCKS, -diff, rsvd)) {
+		    xfs_mod_incore_sb(ip->i_mount, XFS_SBS_FDBLOCKS, -((int64_t)diff), rsvd)) {
 			/*
 			 * Ick gross gag me with a spoon.
 			 */

@@ -1191,7 +1191,7 @@ xfs_bmap_add_extent_delay_real(
 					diff--;
 					if (!diff ||
 					    !xfs_mod_incore_sb(ip->i_mount,
-						    XFS_SBS_FDBLOCKS, -diff, rsvd))
+						    XFS_SBS_FDBLOCKS, -((int64_t)diff), rsvd))
 						break;
 				}
 				if (temp2) {

@@ -1199,7 +1199,7 @@ xfs_bmap_add_extent_delay_real(
 					diff--;
 					if (!diff ||
 					    !xfs_mod_incore_sb(ip->i_mount,
-						    XFS_SBS_FDBLOCKS, -diff, rsvd))
+						    XFS_SBS_FDBLOCKS, -((int64_t)diff), rsvd))
 						break;
 				}
 			}

@@ -1916,7 +1916,7 @@ xfs_bmap_add_extent_hole_delay(
 	if (oldlen != newlen) {
 		ASSERT(oldlen > newlen);
 		xfs_mod_incore_sb(ip->i_mount, XFS_SBS_FDBLOCKS,
-			(int)(oldlen - newlen), rsvd);
+			(int64_t)(oldlen - newlen), rsvd);
 		/*
 		 * Nothing to do for disk quota accounting here.
 		 */

@@ -3187,7 +3187,7 @@ xfs_bmap_del_extent(
 	 */
 	ASSERT(da_old >= da_new);
 	if (da_old > da_new)
-		xfs_mod_incore_sb(mp, XFS_SBS_FDBLOCKS, (int)(da_old - da_new),
+		xfs_mod_incore_sb(mp, XFS_SBS_FDBLOCKS, (int64_t)(da_old - da_new),
 			rsvd);
 done:
 	*logflagsp = flags;

@@ -4875,24 +4875,24 @@ xfs_bmapi(
 				if (rt) {
 					error = xfs_mod_incore_sb(mp,
 							XFS_SBS_FREXTENTS,
-							-(extsz), rsvd);
+							-((int64_t)extsz), rsvd);
 				} else {
 					error = xfs_mod_incore_sb(mp,
 							XFS_SBS_FDBLOCKS,
-							-(alen), rsvd);
+							-((int64_t)alen), rsvd);
 				}
 				if (!error) {
 					error = xfs_mod_incore_sb(mp,
 							XFS_SBS_FDBLOCKS,
-							-(indlen), rsvd);
+							-((int64_t)indlen), rsvd);
 					if (error && rt)
 						xfs_mod_incore_sb(mp,
 							XFS_SBS_FREXTENTS,
-							extsz, rsvd);
+							(int64_t)extsz, rsvd);
 					else if (error)
 						xfs_mod_incore_sb(mp,
 							XFS_SBS_FDBLOCKS,
-							alen, rsvd);
+							(int64_t)alen, rsvd);
 				}
 
 				if (error) {

@@ -5538,13 +5538,13 @@ xfs_bunmapi(
 				rtexts = XFS_FSB_TO_B(mp, del.br_blockcount);
 				do_div(rtexts, mp->m_sb.sb_rextsize);
 				xfs_mod_incore_sb(mp, XFS_SBS_FREXTENTS,
-						(int)rtexts, rsvd);
+						(int64_t)rtexts, rsvd);
 				(void)XFS_TRANS_RESERVE_QUOTA_NBLKS(mp,
 					NULL, ip, -((long)del.br_blockcount), 0,
 					XFS_QMOPT_RES_RTBLKS);
 			} else {
 				xfs_mod_incore_sb(mp, XFS_SBS_FDBLOCKS,
-						(int)del.br_blockcount, rsvd);
+						(int64_t)del.br_blockcount, rsvd);
 				(void)XFS_TRANS_RESERVE_QUOTA_NBLKS(mp,
 					NULL, ip, -((long)del.br_blockcount), 0,
 					XFS_QMOPT_RES_REGBLKS);

fs/xfs/xfs_mount.c +10 -3

index 62188ea..ab3d014 100644
@@ -1238,8 +1238,11 @@ xfs_mod_sb(xfs_trans_t *tp, __int64_t fields)
  * The SB_LOCK must be held when this routine is called.
  */
 STATIC int
-xfs_mod_incore_sb_unlocked(xfs_mount_t *mp, xfs_sb_field_t field,
-			int delta, int rsvd)
+xfs_mod_incore_sb_unlocked(
+	xfs_mount_t	*mp,
+	xfs_sb_field_t	field,
+	int64_t		delta,
+	int		rsvd)
 {
 	int		scounter;	/* short counter for 32 bit fields */
 	long long	lcounter;	/* long counter for 64 bit fields */

@@ -1401,7 +1404,11 @@ xfs_mod_incore_sb_unlocked(xfs_mount_t *mp, xfs_sb_field_t field,
  * routine to do the work.
  */
 int
-xfs_mod_incore_sb(xfs_mount_t *mp, xfs_sb_field_t field, int delta, int rsvd)
+xfs_mod_incore_sb(
+	xfs_mount_t	*mp,
+	xfs_sb_field_t	field,
+	int64_t		delta,
+	int		rsvd)
 {
 	unsigned long	s;
 	int	status;

fs/xfs/xfs_mount.h +3 -2

index cd3cf96..100d5b9 100644
@@ -526,10 +526,11 @@ xfs_daddr_to_agbno(struct xfs_mount *mp, xfs_daddr_t d)
 
 /*
  * This structure is for use by the xfs_mod_incore_sb_batch() routine.
+ * xfs_growfs can specify a few fields which are more than int limit
  */
 typedef struct xfs_mod_sb {
 	xfs_sb_field_t	msb_field;	/* Field to modify, see below */
-	int		msb_delta;	/* Change to make to specified field */
+	int64_t		msb_delta;	/* Change to make to specified field */
 } xfs_mod_sb_t;
 
 #define	XFS_MOUNT_ILOCK(mp)	mutex_lock(&((mp)->m_ilock))

@@ -547,7 +548,7 @@ extern int	xfs_unmountfs(xfs_mount_t *, struct cred *);
 extern void	xfs_unmountfs_close(xfs_mount_t *, struct cred *);
 extern int	xfs_unmountfs_writesb(xfs_mount_t *);
 extern int	xfs_unmount_flush(xfs_mount_t *, int);
-extern int	xfs_mod_incore_sb(xfs_mount_t *, xfs_sb_field_t, int, int);
+extern int	xfs_mod_incore_sb(xfs_mount_t *, xfs_sb_field_t, int64_t, int);
 extern int	xfs_mod_incore_sb_batch(xfs_mount_t *, xfs_mod_sb_t *,
 			uint, int);
 extern struct xfs_buf *xfs_getsb(xfs_mount_t *, int);

fs/xfs/xfs_trans.c +16 ~~-16~~

index d3d714e..68b5895 100644
@@ -232,7 +232,7 @@ xfs_trans_reserve(
 	 */
 	if (blocks > 0) {
 		error = xfs_mod_incore_sb(tp->t_mountp, XFS_SBS_FDBLOCKS,
-					  -blocks, rsvd);
+					  -((int64_t)blocks), rsvd);
 		if (error != 0) {
                         PFLAGS_RESTORE_FSTRANS(&tp->t_pflags);
 			return (XFS_ERROR(ENOSPC));

@@ -273,7 +273,7 @@ xfs_trans_reserve(
 	 */
 	if (rtextents > 0) {
 		error = xfs_mod_incore_sb(tp->t_mountp, XFS_SBS_FREXTENTS,
-					  -rtextents, rsvd);
+					  -((int64_t)rtextents), rsvd);
 		if (error) {
 			error = XFS_ERROR(ENOSPC);
 			goto undo_log;

@@ -303,7 +303,7 @@ undo_log:
 undo_blocks:
 	if (blocks > 0) {
 		(void) xfs_mod_incore_sb(tp->t_mountp, XFS_SBS_FDBLOCKS,
-					 blocks, rsvd);
+					 (int64_t)blocks, rsvd);
 		tp->t_blk_res = 0;
 	}

@@ -325,7 +325,7 @@ void
 xfs_trans_mod_sb(
 	xfs_trans_t	*tp,
 	uint		field,
-	long		delta)
+	int64_t		delta)
 {
 
 	switch (field) {

@@ -556,62 +556,62 @@ xfs_trans_unreserve_and_mod_sb(
 	if (tp->t_flags & XFS_TRANS_SB_DIRTY) {
 		if (tp->t_icount_delta != 0) {
 			msbp->msb_field = XFS_SBS_ICOUNT;
-			msbp->msb_delta = (int)tp->t_icount_delta;
+			msbp->msb_delta = tp->t_icount_delta;
 			msbp++;
 		}
 		if (tp->t_ifree_delta != 0) {
 			msbp->msb_field = XFS_SBS_IFREE;
-			msbp->msb_delta = (int)tp->t_ifree_delta;
+			msbp->msb_delta = tp->t_ifree_delta;
 			msbp++;
 		}
 		if (tp->t_fdblocks_delta != 0) {
 			msbp->msb_field = XFS_SBS_FDBLOCKS;
-			msbp->msb_delta = (int)tp->t_fdblocks_delta;
+			msbp->msb_delta = tp->t_fdblocks_delta;
 			msbp++;
 		}
 		if (tp->t_frextents_delta != 0) {
 			msbp->msb_field = XFS_SBS_FREXTENTS;
-			msbp->msb_delta = (int)tp->t_frextents_delta;
+			msbp->msb_delta = tp->t_frextents_delta;
 			msbp++;
 		}
 		if (tp->t_dblocks_delta != 0) {
 			msbp->msb_field = XFS_SBS_DBLOCKS;
-			msbp->msb_delta = (int)tp->t_dblocks_delta;
+			msbp->msb_delta = tp->t_dblocks_delta;
 			msbp++;
 		}
 		if (tp->t_agcount_delta != 0) {
 			msbp->msb_field = XFS_SBS_AGCOUNT;
-			msbp->msb_delta = (int)tp->t_agcount_delta;
+			msbp->msb_delta = tp->t_agcount_delta;
 			msbp++;
 		}
 		if (tp->t_imaxpct_delta != 0) {
 			msbp->msb_field = XFS_SBS_IMAX_PCT;
-			msbp->msb_delta = (int)tp->t_imaxpct_delta;
+			msbp->msb_delta = tp->t_imaxpct_delta;
 			msbp++;
 		}
 		if (tp->t_rextsize_delta != 0) {
 			msbp->msb_field = XFS_SBS_REXTSIZE;
-			msbp->msb_delta = (int)tp->t_rextsize_delta;
+			msbp->msb_delta = tp->t_rextsize_delta;
 			msbp++;
 		}
 		if (tp->t_rbmblocks_delta != 0) {
 			msbp->msb_field = XFS_SBS_RBMBLOCKS;
-			msbp->msb_delta = (int)tp->t_rbmblocks_delta;
+			msbp->msb_delta = tp->t_rbmblocks_delta;
 			msbp++;
 		}
 		if (tp->t_rblocks_delta != 0) {
 			msbp->msb_field = XFS_SBS_RBLOCKS;
-			msbp->msb_delta = (int)tp->t_rblocks_delta;
+			msbp->msb_delta = tp->t_rblocks_delta;
 			msbp++;
 		}
 		if (tp->t_rextents_delta != 0) {
 			msbp->msb_field = XFS_SBS_REXTENTS;
-			msbp->msb_delta = (int)tp->t_rextents_delta;
+			msbp->msb_delta = tp->t_rextents_delta;
 			msbp++;
 		}
 		if (tp->t_rextslog_delta != 0) {
 			msbp->msb_field = XFS_SBS_REXTSLOG;
-			msbp->msb_delta = (int)tp->t_rextslog_delta;
+			msbp->msb_delta = tp->t_rextslog_delta;
 			msbp++;
 		}
 	}

fs/xfs/xfs_trans.h +21 ~~-21~~

60 lines not on this page · raw message

Dave Chinner

2008-08-26 01:59 UTC #1 lore raw

Quoted 33 lines · Ed Cashin

Dave Chinner, hello.
Your fix for the counters that had been preventing an XFS from growing by more than two terabytes went into 2.6.21, after the introduction of the per-cpu in-core superblock counters. The per-cpu in-core superblock counters are not in the latest version of the long-lived stable kernel 2.6.16.y, maintained by Adrian Bunk.
The problem with growing by more than two terabytes appears to be there, though. The 2.6.16.62 kernel will not allow me to grow an XFS by, e.g., 10 terabytes, so xfs_growfs does not report any change in the data blocks.
I backported your fix,
commit 20f4ebf2bf2f57c1a9abb3655391336cc90314b3
Author: David Chinner <[email protected]>
Date: Sat Feb 10 18:36:10 2007 +1100
[XFS] Make growfs work for amounts greater than 2TB
The free block modification code has a 32bit interface, limiting the size the filesystem can be grown even on 64 bit machines. On 32 bit machines, there are other 32bit variables in transaction structures and interfaces that need to be expanded to allow this to work.
SGI-PV: 959978
SGI-Modid: xfs-linux-melb:xfs-kern:27894a
Signed-off-by David Chinner <[email protected]>
Signed-off-by Christoph Hellwig <[email protected]>
Signed-off-by Tim Shimmin <[email protected]>

Quoted 6 lines

... to the 2.6.16.y git tree, and the result is included below. When I apply this backported fix to 2.6.16.62, I can grow an online XFS by 10 terabytes without any trouble.
Do you see any problems with this backport? If not, I will submit it for inclusion in the next 2.6.16.y release.

I suggest you make sure it passes test 078 in the xfsqa suite (part of the xfs-cmds tree) as that tests all the nasty growfs corner cases. You'll need to test it on 32 bit and 64 bit machines....

If it passes that then I don't see any problems - SGI backported this for sles10 which is based on 2.6.16 a long time ago.

Cheers,

Dave.

Signature · 2 lines

Dave Chinner
[email protected]

Ed Cashin

author

2008-09-03 18:24 UTC #2 lore raw

Quoted 1 line · Dave Chinner

On Mon, Aug 25, 2008 at 11:39:31AM -0400, Ed Cashin wrote:

...

Quoted 3 lines

Quoted 3 lines
I backported your fix,
    commit 20f4ebf2bf2f57c1a9abb3655391336cc90314b3

...

Quoted 3 lines

Quoted 3 lines
... to the 2.6.16.y git tree, and the result is included below. When I apply this backported fix to 2.6.16.62, I can grow an online XFS by 10 terabytes without any trouble.

...

Quoted 6 lines

I suggest you make sure it passes test 078 in the xfsqa suite (part of the xfs-cmds tree) as that tests all the nasty growfs corner cases. You'll need to test it on 32 bit and 64 bit machines....
If it passes that then I don't see any problems - SGI backported this for sles10 which is based on 2.6.16 a long time ago.

On a 64-bit system running 2.6.16.62 with this patch, test 078 does not succeed because of one difference in the output file, the line in the diff below. Instead of a new size of 4194304001 blocks, the new size is 4194304000 blocks.

Maybe this looks familiar to you and suggests that I missed something. If so, please let me know.

ellijay:/mnt/makki/ecashin/opt-amd64/xfstests# diff -U15 078.out 078.out.bad
--- 078.out     2008-03-05 00:25:08.000000000 -0500
+++ 078.out.bad 2008-08-29 16:41:34.537888000 -0400
@@ -194,18 +194,18 @@

=== GROWFS (from 1t to 16000g, 4096 blocksize)

*** mkfs loop file (size=1t)
meta-data=DDEV isize=XXX agcount=N, agsize=XXX blks
data = bsize=XXX blocks=XXX, imaxpct=PCT
= sunit=XXX swidth=XXX, unwritten=X
naming =VERN bsize=XXX
log =LDEV bsize=XXX blocks=XXX
realtime =RDEV extsz=XXX blocks=XXX, rtextents=XXX *** extend loop file
wrote 4096/4096 bytes at offset 17179869184000 *** mount loop filesystem
*** grow loop filesystem
xfs_growfs --BlockSize=4096 --Blocks=268435456
-data blocks changed from 268435456 to 4194304001 +data blocks changed from 268435456 to 4194304000
*** unmount
*** all done
ellijay:/mnt/makki/ecashin/opt-amd64/xfstests#

Signature · 1 line

  Ed Cashin <[email protected]>

Dave Chinner

2008-09-04 23:13 UTC #3 lore raw

Quoted 16 lines · Ed Cashin

Quoted 1 line · Dave Chinner
On Mon, Aug 25, 2008 at 11:39:31AM -0400, Ed Cashin wrote:
...
Quoted 3 lines
Quoted 3 lines
I backported your fix,
    commit 20f4ebf2bf2f57c1a9abb3655391336cc90314b3
...
Quoted 3 lines
Quoted 3 lines
... to the 2.6.16.y git tree, and the result is included below. When I apply this backported fix to 2.6.16.62, I can grow an online XFS by 10 terabytes without any trouble.
...
Quoted 4 lines
I suggest you make sure it passes test 078 in the xfsqa suite (part of the xfs-cmds tree) as that tests all the nasty growfs corner cases. You'll need to test it on 32 bit and 64 bit machines....

Quoted 7 lines

Quoted 2 lines
If it passes that then I don't see any problems - SGI backported this for sles10 which is based on 2.6.16 a long time ago.
On a 64-bit system running 2.6.16.62 with this patch, test 078 does not succeed because of one difference in the output file, the line in the diff below. Instead of a new size of 4194304001 blocks, the new size is 4194304000 blocks.

Ah, yes. That. I think Barry can try to explain that one because:

http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-cmds/xfstests/078.out.diff?r1=1.3;r2=1.4

The test golden output was changed instead of someone understanding why the fixes to growfs changed the size that the filesystem was grown to. ISTR being opposed to changing the golden output because it was the wrong thing to do and would break QA on older kernels, not to mention that it indicated some possible off-by-one bug in a change that had been made at some point...

Other than that, the backport should be fine given it passed all the other parts of the test....

Cheers,

Dave.

Signature · 2 lines

Dave Chinner
[email protected]

Ed Cashin

author

2008-09-05 16:27 UTC #4 lore raw

On Fri, Sep 05, 2008 at 09:14:18AM +1000, Dave Chinner wrote:
...

Quoted 11 lines

Ah, yes. That. I think Barry can try to explain that one because:
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-cmds/xfstests/078.out.diff?r1=1.3;r2=1.4
The test golden output was changed instead of someone understanding why the fixes to growfs changed the size that the filesystem was grown to. ISTR being opposed to changing the golden output because it was the wrong thing to do and would break QA on older kernels, not to mention that it indicated some possible off-by-one bug in a change that had been made at some point...

Quoted 2 lines

Other than that, the backport should be fine given it passed all the other parts of the test....

Oh, that's good news, thanks.

I just need to test the patch on a 32-bit O.S. now, but I am having a difficult time building xfs-cmds/xfstests on a CentOS 5 VM.

Does anyone have a recommendation for an O.S. I can use to build and run the xfsqa test 078? It would be running the 2.6.16.62 kernel with the patch in question, not the stock distro kernel.

Signature · 1 line

  Ed Cashin <[email protected]>