[PATCH 00/25] xfs_bmapi split and cleanups

[PATCH 00/25] xfs_bmapi split and cleanups

[Christoph Hellwig]

This series, which is largely based on one that Dave published a couple month
ago, splits up the many armed hyrda called xfs_bmapi into multiple smaller
functions, and reduced stack usage in those pathes.

The main visible change to the rest of xfs is that there are three new
interfaces that replace xfs_bmapi:  xfs_bmapi_read to read extent maps,
xfs_bmapi_delay to add new unwritten extents, and xfs_bmapi to allocate
or convert real and unwrittent extents.

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

[PATCH 01/25] xfs: remove the first extent special case in xfs_bmap_add_extent

[Christoph Hellwig]

[-- Attachment #1: xfs-bmap_add_extent-remove-special-cases --]
[-- Type: text/plain, Size: 1414 bytes --]

Both xfs_bmap_add_extent_hole_delay and xfs_bmap_add_extent_hole_real
already contain code to handle the case where there is no extent to
merge with, which is effectively the same as the code duplicated here.

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

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-07-13 09:14:14.375340664 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-07-13 09:16:07.041396967 +0200
@@ -466,26 +466,9 @@ xfs_bmap_add_extent(
 	ASSERT(*idx <= nextents);
 
 	/*
-	 * This is the first extent added to a new/empty file.
-	 * Special case this one, so other routines get to assume there are
-	 * already extents in the list.
-	 */
-	if (nextents == 0) {
-		xfs_iext_insert(ip, *idx, 1, new,
-				whichfork == XFS_ATTR_FORK ? BMAP_ATTRFORK : 0);
-
-		ASSERT(cur == NULL);
-
-		if (!isnullstartblock(new->br_startblock)) {
-			XFS_IFORK_NEXT_SET(ip, whichfork, 1);
-			logflags = XFS_ILOG_CORE | xfs_ilog_fext(whichfork);
-		} else
-			logflags = 0;
-	}
-	/*
 	 * Any kind of new delayed allocation goes here.
 	 */
-	else if (isnullstartblock(new->br_startblock)) {
+	if (isnullstartblock(new->br_startblock)) {
 		if (cur)
 			ASSERT((cur->bc_private.b.flags &
 				XFS_BTCUR_BPRV_WASDEL) == 0);

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

[PATCH 02/25] xfs: remove impossible to read code in xfs_bmap_add_extent_delay_real

[Christoph Hellwig]

[-- Attachment #1: xfs-cleanup-xfs_bmap_add_extent_delay_real-eh --]
[-- Type: text/plain, Size: 1920 bytes --]

We already have the worst case blocks reserved, so xfs_icsb_modify_counters
won't fail in xfs_bmap_add_extent_delay_real.  In fact we've had an assert
to catch this case since day and it never triggered.  So remove the code
to try smaller reservations, and just return the error for that case in
addition to keeping the assert.

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

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-07-15 11:29:03.408815322 -0700
+++ xfs/fs/xfs/xfs_bmap.c	2011-07-15 11:29:16.985408438 -0700
@@ -1045,34 +1045,15 @@ xfs_bmap_add_extent_delay_real(
 		temp2 = xfs_bmap_worst_indlen(ip, temp2);
 		diff = (int)(temp + temp2 - startblockval(PREV.br_startblock) -
 			(cur ? cur->bc_private.b.allocated : 0));
-		if (diff > 0 &&
-		    xfs_icsb_modify_counters(ip->i_mount, XFS_SBS_FDBLOCKS,
-					     -((int64_t)diff), 0)) {
-			/*
-			 * Ick gross gag me with a spoon.
-			 */
-			ASSERT(0);	/* want to see if this ever happens! */
-			while (diff > 0) {
-				if (temp) {
-					temp--;
-					diff--;
-					if (!diff ||
-					    !xfs_icsb_modify_counters(ip->i_mount,
-						    XFS_SBS_FDBLOCKS,
-						    -((int64_t)diff), 0))
-						break;
-				}
-				if (temp2) {
-					temp2--;
-					diff--;
-					if (!diff ||
-					    !xfs_icsb_modify_counters(ip->i_mount,
-						    XFS_SBS_FDBLOCKS,
-						    -((int64_t)diff), 0))
-						break;
-				}
-			}
+		if (diff > 0) {
+			error = xfs_icsb_modify_counters(ip->i_mount,
+					XFS_SBS_FDBLOCKS,
+					-((int64_t)diff), 0);
+			ASSERT(!error);
+			if (error)
+				goto done;
 		}
+
 		ep = xfs_iext_get_ext(ifp, *idx);
 		xfs_bmbt_set_startblock(ep, nullstartblock((int)temp));
 		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);

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

[PATCH 03/25] xfs: remove the nextents variable in xfs_bmapi

[Christoph Hellwig]

[-- Attachment #1: xfs-bmapi-cleanup-nextents --]
[-- Type: text/plain, Size: 1948 bytes --]

Instead of using a local variable that needs to updated when we modify
the extent map just check ifp->if_bytes directly where we use it.

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

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:07:50.707607838 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:07:51.327604480 +0200
@@ -4294,7 +4294,6 @@ xfs_bmapi(
 	xfs_mount_t	*mp;		/* xfs mount structure */
 	int		n;		/* current extent index */
 	int		nallocs;	/* number of extents alloc'd */
-	xfs_extnum_t	nextents;	/* number of extents in file */
 	xfs_fileoff_t	obno;		/* old block number (offset) */
 	xfs_bmbt_irec_t	prev;		/* previous file extent record */
 	int		tmp_logflags;	/* temp flags holder */
@@ -4380,7 +4379,6 @@ xfs_bmapi(
 		goto error0;
 	ep = xfs_bmap_search_extents(ip, bno, whichfork, &eof, &lastx, &got,
 		&prev);
-	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
 	n = 0;
 	end = bno + len;
 	obno = bno;
@@ -4622,7 +4620,6 @@ xfs_bmapi(
 			if (error)
 				goto error0;
 			ep = xfs_iext_get_ext(ifp, lastx);
-			nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
 			xfs_bmbt_get_all(ep, &got);
 			ASSERT(got.br_startoff <= aoff);
 			ASSERT(got.br_startoff + got.br_blockcount >=
@@ -4723,7 +4720,6 @@ xfs_bmapi(
 			if (error)
 				goto error0;
 			ep = xfs_iext_get_ext(ifp, lastx);
-			nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
 			xfs_bmbt_get_all(ep, &got);
 			/*
 			 * We may have combined previously unwritten
@@ -4781,7 +4777,7 @@ xfs_bmapi(
 		 * Else go on to the next record.
 		 */
 		prev = got;
-		if (++lastx < nextents) {
+		if (++lastx < ifp->if_bytes / sizeof(xfs_bmbt_rec_t)) {
 			ep = xfs_iext_get_ext(ifp, lastx);
 			xfs_bmbt_get_all(ep, &got);
 		} else {

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

[PATCH 04/25] xfs: factor extent map manipulations out of xfs_bmapi

[Christoph Hellwig]

[-- Attachment #1: xfs-bmapi-factor-map-manipulation --]
[-- Type: text/plain, Size: 6920 bytes --]

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:07:51.327604480 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:08:45.150646228 +0200
@@ -4249,6 +4249,107 @@ xfs_bmap_validate_ret(
 
 
 /*
+ * Trim the returned map to the required bounds
+ */
+STATIC void
+xfs_bmapi_trim_map(
+	struct xfs_bmbt_irec	*mval,
+	struct xfs_bmbt_irec	*got,
+	xfs_fileoff_t		*bno,
+	xfs_filblks_t		len,
+	xfs_fileoff_t		obno,
+	xfs_fileoff_t		end,
+	int			n,
+	int			flags)
+{
+	if ((flags & XFS_BMAPI_ENTIRE) ||
+	    got->br_startoff + got->br_blockcount <= obno) {
+		*mval = *got;
+		if (isnullstartblock(got->br_startblock))
+			mval->br_startblock = DELAYSTARTBLOCK;
+		return;
+	}
+
+	if (obno > *bno)
+		*bno = obno;
+	ASSERT((*bno >= obno) || (n == 0));
+	ASSERT(*bno < end);
+	mval->br_startoff = *bno;
+	if (isnullstartblock(got->br_startblock))
+		mval->br_startblock = DELAYSTARTBLOCK;
+	else
+		mval->br_startblock = got->br_startblock +
+					(*bno - got->br_startoff);
+	/*
+	 * Return the minimum of what we got and what we asked for for
+	 * the length.  We can use the len variable here because it is
+	 * modified below and we could have been there before coming
+	 * here if the first part of the allocation didn't overlap what
+	 * was asked for.
+	 */
+	mval->br_blockcount = XFS_FILBLKS_MIN(end - *bno,
+			got->br_blockcount - (*bno - got->br_startoff));
+	mval->br_state = got->br_state;
+	ASSERT(mval->br_blockcount <= len);
+	return;
+}
+
+/*
+ * Update and validate the extent map to return
+ */
+STATIC void
+xfs_bmapi_update_map(
+	struct xfs_bmbt_irec	**map,
+	xfs_fileoff_t		*bno,
+	xfs_filblks_t		*len,
+	xfs_fileoff_t		obno,
+	xfs_fileoff_t		end,
+	int			*n,
+	int			flags)
+{
+	xfs_bmbt_irec_t	*mval = *map;
+
+	ASSERT((flags & XFS_BMAPI_ENTIRE) ||
+	       ((mval->br_startoff + mval->br_blockcount) <= end));
+	ASSERT((flags & XFS_BMAPI_ENTIRE) || (mval->br_blockcount <= *len) ||
+	       (mval->br_startoff < obno));
+
+	*bno = mval->br_startoff + mval->br_blockcount;
+	*len = end - *bno;
+	if (*n > 0 && mval->br_startoff == mval[-1].br_startoff) {
+		/* update previous map with new information */
+		ASSERT(mval->br_startblock == mval[-1].br_startblock);
+		ASSERT(mval->br_blockcount > mval[-1].br_blockcount);
+		ASSERT(mval->br_state == mval[-1].br_state);
+		mval[-1].br_blockcount = mval->br_blockcount;
+		mval[-1].br_state = mval->br_state;
+	} else if (*n > 0 && mval->br_startblock != DELAYSTARTBLOCK &&
+		   mval[-1].br_startblock != DELAYSTARTBLOCK &&
+		   mval[-1].br_startblock != HOLESTARTBLOCK &&
+		   mval->br_startblock == mval[-1].br_startblock +
+					  mval[-1].br_blockcount &&
+		   ((flags & XFS_BMAPI_IGSTATE) ||
+			mval[-1].br_state == mval->br_state)) {
+		ASSERT(mval->br_startoff ==
+		       mval[-1].br_startoff + mval[-1].br_blockcount);
+		mval[-1].br_blockcount += mval->br_blockcount;
+	} else if (*n > 0 &&
+		   mval->br_startblock == DELAYSTARTBLOCK &&
+		   mval[-1].br_startblock == DELAYSTARTBLOCK &&
+		   mval->br_startoff ==
+		   mval[-1].br_startoff + mval[-1].br_blockcount) {
+		mval[-1].br_blockcount += mval->br_blockcount;
+		mval[-1].br_state = mval->br_state;
+	} else if (!((*n == 0) &&
+		     ((mval->br_startoff + mval->br_blockcount) <=
+		      obno))) {
+		mval++;
+		(*n)++;
+	}
+	*map = mval;
+}
+
+/*
  * Map file blocks to filesystem blocks.
  * File range is given by the bno/len pair.
  * Adds blocks to file if a write ("flags & XFS_BMAPI_WRITE" set)
@@ -4650,44 +4751,9 @@ xfs_bmapi(
 			n++;
 			continue;
 		}
-		/*
-		 * Then deal with the allocated space we found.
-		 */
-		ASSERT(ep != NULL);
-		if (!(flags & XFS_BMAPI_ENTIRE) &&
-		    (got.br_startoff + got.br_blockcount > obno)) {
-			if (obno > bno)
-				bno = obno;
-			ASSERT((bno >= obno) || (n == 0));
-			ASSERT(bno < end);
-			mval->br_startoff = bno;
-			if (isnullstartblock(got.br_startblock)) {
-				ASSERT(!wr || (flags & XFS_BMAPI_DELAY));
-				mval->br_startblock = DELAYSTARTBLOCK;
-			} else
-				mval->br_startblock =
-					got.br_startblock +
-					(bno - got.br_startoff);
-			/*
-			 * Return the minimum of what we got and what we
-			 * asked for for the length.  We can use the len
-			 * variable here because it is modified below
-			 * and we could have been there before coming
-			 * here if the first part of the allocation
-			 * didn't overlap what was asked for.
-			 */
-			mval->br_blockcount =
-				XFS_FILBLKS_MIN(end - bno, got.br_blockcount -
-					(bno - got.br_startoff));
-			mval->br_state = got.br_state;
-			ASSERT(mval->br_blockcount <= len);
-		} else {
-			*mval = got;
-			if (isnullstartblock(mval->br_startblock)) {
-				ASSERT(!wr || (flags & XFS_BMAPI_DELAY));
-				mval->br_startblock = DELAYSTARTBLOCK;
-			}
-		}
+
+		/* Deal with the allocated space we found.  */
+		xfs_bmapi_trim_map(mval, &got, &bno, len, obno, end, n, flags);
 
 		/*
 		 * Check if writing previously allocated but
@@ -4730,42 +4796,9 @@ xfs_bmapi(
 				continue;
 		}
 
-		ASSERT((flags & XFS_BMAPI_ENTIRE) ||
-		       ((mval->br_startoff + mval->br_blockcount) <= end));
-		ASSERT((flags & XFS_BMAPI_ENTIRE) ||
-		       (mval->br_blockcount <= len) ||
-		       (mval->br_startoff < obno));
-		bno = mval->br_startoff + mval->br_blockcount;
-		len = end - bno;
-		if (n > 0 && mval->br_startoff == mval[-1].br_startoff) {
-			ASSERT(mval->br_startblock == mval[-1].br_startblock);
-			ASSERT(mval->br_blockcount > mval[-1].br_blockcount);
-			ASSERT(mval->br_state == mval[-1].br_state);
-			mval[-1].br_blockcount = mval->br_blockcount;
-			mval[-1].br_state = mval->br_state;
-		} else if (n > 0 && mval->br_startblock != DELAYSTARTBLOCK &&
-			   mval[-1].br_startblock != DELAYSTARTBLOCK &&
-			   mval[-1].br_startblock != HOLESTARTBLOCK &&
-			   mval->br_startblock ==
-			   mval[-1].br_startblock + mval[-1].br_blockcount &&
-			   ((flags & XFS_BMAPI_IGSTATE) ||
-				mval[-1].br_state == mval->br_state)) {
-			ASSERT(mval->br_startoff ==
-			       mval[-1].br_startoff + mval[-1].br_blockcount);
-			mval[-1].br_blockcount += mval->br_blockcount;
-		} else if (n > 0 &&
-			   mval->br_startblock == DELAYSTARTBLOCK &&
-			   mval[-1].br_startblock == DELAYSTARTBLOCK &&
-			   mval->br_startoff ==
-			   mval[-1].br_startoff + mval[-1].br_blockcount) {
-			mval[-1].br_blockcount += mval->br_blockcount;
-			mval[-1].br_state = mval->br_state;
-		} else if (!((n == 0) &&
-			     ((mval->br_startoff + mval->br_blockcount) <=
-			      obno))) {
-			mval++;
-			n++;
-		}
+		/* update the extent map to return */
+		xfs_bmapi_update_map(&mval, &bno, &len, obno, end, &n, flags);
+
 		/*
 		 * If we're done, stop now.  Stop when we've allocated
 		 * XFS_BMAP_MAX_NMAP extents no matter what.  Otherwise

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

[PATCH 05/25] xfs: introduce xfs_bmapi_read()

[Christoph Hellwig]

[-- Attachment #1: xfs-bmapi-add-xfs_bmapi_read --]
[-- Type: text/plain, Size: 16787 bytes --]

xfs_bmapi() currently handles both extent map reading and
allocation. As a result, the code is littered with "if (wr)"
branches to conditionally do allocation operations if required.
This makes the code much harder to follow and causes significant
indent issues with the code.

Given that read mapping is much simpler than allocation, we can
split out read mapping from xfs_bmapi() and reuse the logic that
we have already factored out do do all the hard work of handling the
extent map manipulations. The results in a much simpler function for
the common extent read operations, and will allow the allocation
code to be simplified in another commit.

Once xfs_bmapi_read() is implemented, convert all the callers of
xfs_bmapi() that are only reading extents to use the new function.

Signed-off-by: Dave Chinner <dchinner@redhat.com>

Index: xfs/fs/xfs/xfs_aops.c
===================================================================
--- xfs.orig/fs/xfs/xfs_aops.c	2011-08-23 20:43:28.488862691 +0200
+++ xfs/fs/xfs/xfs_aops.c	2011-08-23 21:09:32.897054232 +0200
@@ -287,8 +287,8 @@ xfs_map_blocks(
 		count = mp->m_maxioffset - offset;
 	end_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + count);
 	offset_fsb = XFS_B_TO_FSBT(mp, offset);
-	error = xfs_bmapi(NULL, ip, offset_fsb, end_fsb - offset_fsb,
-			  bmapi_flags,  NULL, 0, imap, &nimaps, NULL);
+	error = xfs_bmapi_read(ip, offset_fsb, end_fsb - offset_fsb,
+				imap, &nimaps, bmapi_flags);
 	xfs_iunlock(ip, XFS_ILOCK_SHARED);
 
 	if (error)
@@ -1102,8 +1102,8 @@ __xfs_get_blocks(
 	end_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + size);
 	offset_fsb = XFS_B_TO_FSBT(mp, offset);
 
-	error = xfs_bmapi(NULL, ip, offset_fsb, end_fsb - offset_fsb,
-			  XFS_BMAPI_ENTIRE,  NULL, 0, &imap, &nimaps, NULL);
+	error = xfs_bmapi_read(ip, offset_fsb, end_fsb - offset_fsb,
+				&imap, &nimaps, XFS_BMAPI_ENTIRE);
 	if (error)
 		goto out_unlock;
 
Index: xfs/fs/xfs/xfs_file.c
===================================================================
--- xfs.orig/fs/xfs/xfs_file.c	2011-08-23 20:43:28.488862691 +0200
+++ xfs/fs/xfs/xfs_file.c	2011-08-23 21:09:32.900387547 +0200
@@ -496,11 +496,9 @@ xfs_zero_last_block(
 
 	last_fsb = XFS_B_TO_FSBT(mp, isize);
 	nimaps = 1;
-	error = xfs_bmapi(NULL, ip, last_fsb, 1, 0, NULL, 0, &imap,
-			  &nimaps, NULL);
-	if (error) {
+	error = xfs_bmapi_read(ip, last_fsb, 1, &imap, &nimaps, 0);
+	if (error)
 		return error;
-	}
 	ASSERT(nimaps > 0);
 	/*
 	 * If the block underlying isize is just a hole, then there
@@ -591,8 +589,8 @@ xfs_zero_eof(
 	while (start_zero_fsb <= end_zero_fsb) {
 		nimaps = 1;
 		zero_count_fsb = end_zero_fsb - start_zero_fsb + 1;
-		error = xfs_bmapi(NULL, ip, start_zero_fsb, zero_count_fsb,
-				  0, NULL, 0, &imap, &nimaps, NULL);
+		error = xfs_bmapi_read(ip, start_zero_fsb, zero_count_fsb,
+					  &imap, &nimaps, 0);
 		if (error) {
 			ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL));
 			return error;
Index: xfs/fs/xfs/xfs_dquot.c
===================================================================
--- xfs.orig/fs/xfs/xfs_dquot.c	2011-08-23 03:56:39.791586092 +0200
+++ xfs/fs/xfs/xfs_dquot.c	2011-08-23 21:09:32.900387547 +0200
@@ -485,9 +485,8 @@ xfs_qm_dqtobp(
 	/*
 	 * Find the block map; no allocations yet
 	 */
-	error = xfs_bmapi(NULL, quotip, dqp->q_fileoffset,
-			  XFS_DQUOT_CLUSTER_SIZE_FSB, XFS_BMAPI_METADATA,
-			  NULL, 0, &map, &nmaps, NULL);
+	error = xfs_bmapi_read(quotip, dqp->q_fileoffset,
+			       XFS_DQUOT_CLUSTER_SIZE_FSB, &map, &nmaps, 0);
 
 	xfs_iunlock(quotip, XFS_ILOCK_SHARED);
 	if (error)
Index: xfs/fs/xfs/xfs_qm.c
===================================================================
--- xfs.orig/fs/xfs/xfs_qm.c	2011-08-23 03:56:17.321707822 +0200
+++ xfs/fs/xfs/xfs_qm.c	2011-08-23 21:09:32.903720862 +0200
@@ -1347,11 +1347,8 @@ xfs_qm_dqiterate(
 		 * the inode is never added to the transaction.
 		 */
 		xfs_ilock(qip, XFS_ILOCK_SHARED);
-		error = xfs_bmapi(NULL, qip, lblkno,
-				  maxlblkcnt - lblkno,
-				  XFS_BMAPI_METADATA,
-				  NULL,
-				  0, map, &nmaps, NULL);
+		error = xfs_bmapi_read(qip, lblkno, maxlblkcnt - lblkno,
+				       map, &nmaps, 0);
 		xfs_iunlock(qip, XFS_ILOCK_SHARED);
 		if (error)
 			break;
Index: xfs/fs/xfs/xfs_attr.c
===================================================================
--- xfs.orig/fs/xfs/xfs_attr.c	2011-08-23 21:07:49.000000000 +0200
+++ xfs/fs/xfs/xfs_attr.c	2011-08-23 21:09:32.907054177 +0200
@@ -1963,10 +1963,9 @@ xfs_attr_rmtval_get(xfs_da_args_t *args)
 	lblkno = args->rmtblkno;
 	while (valuelen > 0) {
 		nmap = ATTR_RMTVALUE_MAPSIZE;
-		error = xfs_bmapi(args->trans, args->dp, (xfs_fileoff_t)lblkno,
-				  args->rmtblkcnt,
-				  XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
-				  NULL, 0, map, &nmap, NULL);
+		error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
+				       args->rmtblkcnt, map, &nmap,
+				       XFS_BMAPI_ATTRFORK);
 		if (error)
 			return(error);
 		ASSERT(nmap >= 1);
@@ -2092,14 +2091,11 @@ xfs_attr_rmtval_set(xfs_da_args_t *args)
 		 */
 		xfs_bmap_init(args->flist, args->firstblock);
 		nmap = 1;
-		error = xfs_bmapi(NULL, dp, (xfs_fileoff_t)lblkno,
-				  args->rmtblkcnt,
-				  XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
-				  args->firstblock, 0, &map, &nmap,
-				  NULL);
-		if (error) {
+		error = xfs_bmapi_read(dp, (xfs_fileoff_t)lblkno,
+				       args->rmtblkcnt, &map, &nmap,
+				       XFS_BMAPI_ATTRFORK);
+		if (error)
 			return(error);
-		}
 		ASSERT(nmap == 1);
 		ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
 		       (map.br_startblock != HOLESTARTBLOCK));
@@ -2155,16 +2151,12 @@ xfs_attr_rmtval_remove(xfs_da_args_t *ar
 		/*
 		 * Try to remember where we decided to put the value.
 		 */
-		xfs_bmap_init(args->flist, args->firstblock);
 		nmap = 1;
-		error = xfs_bmapi(NULL, args->dp, (xfs_fileoff_t)lblkno,
-					args->rmtblkcnt,
-					XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
-					args->firstblock, 0, &map, &nmap,
-					args->flist);
-		if (error) {
+		error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
+				       args->rmtblkcnt, &map, &nmap,
+				       XFS_BMAPI_ATTRFORK);
+		if (error)
 			return(error);
-		}
 		ASSERT(nmap == 1);
 		ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
 		       (map.br_startblock != HOLESTARTBLOCK));
Index: xfs/fs/xfs/xfs_attr_leaf.c
===================================================================
--- xfs.orig/fs/xfs/xfs_attr_leaf.c	2011-08-23 21:07:10.000000000 +0200
+++ xfs/fs/xfs/xfs_attr_leaf.c	2011-08-23 21:09:32.910387492 +0200
@@ -2926,9 +2926,8 @@ xfs_attr_leaf_freextent(xfs_trans_t **tr
 		 * Try to remember where we decided to put the value.
 		 */
 		nmap = 1;
-		error = xfs_bmapi(*trans, dp, (xfs_fileoff_t)tblkno, tblkcnt,
-					XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
-					NULL, 0, &map, &nmap, NULL);
+		error = xfs_bmapi_read(dp, (xfs_fileoff_t)tblkno, tblkcnt,
+				       &map, &nmap, XFS_BMAPI_ATTRFORK);
 		if (error) {
 			return(error);
 		}
Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:08:45.000000000 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:09:32.913720807 +0200
@@ -4350,6 +4350,95 @@ xfs_bmapi_update_map(
 }
 
 /*
+ * Map file blocks to filesystem blocks without allocation.
+ */
+int
+xfs_bmapi_read(
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		bno,
+	xfs_filblks_t		len,
+	struct xfs_bmbt_irec	*mval,
+	int			*nmap,
+	int			flags)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_ifork	*ifp;
+	struct xfs_bmbt_irec	got;
+	struct xfs_bmbt_irec	prev;
+	xfs_fileoff_t		obno;
+	xfs_fileoff_t		end;
+	xfs_extnum_t		lastx;
+	int			error;
+	int			eof;
+	int			n = 0;
+	int			whichfork = (flags & XFS_BMAPI_ATTRFORK) ?
+						XFS_ATTR_FORK : XFS_DATA_FORK;
+	ASSERT(*nmap >= 1);
+	ASSERT(!(flags & ~(XFS_BMAPI_ATTRFORK|XFS_BMAPI_ENTIRE|
+			   XFS_BMAPI_IGSTATE)));
+
+	error = XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS &&
+		XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE;
+	if (unlikely(XFS_TEST_ERROR(error, mp, XFS_ERRTAG_BMAPIFORMAT,
+				    XFS_RANDOM_BMAPIFORMAT))) {
+		XFS_ERROR_REPORT("xfs_bmapi", XFS_ERRLEVEL_LOW, mp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	XFS_STATS_INC(xs_blk_mapr);
+
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	ASSERT(ifp->if_ext_max ==
+	       XFS_IFORK_SIZE(ip, whichfork) / (uint)sizeof(xfs_bmbt_rec_t));
+	if (!(ifp->if_flags & XFS_IFEXTENTS)) {
+		error = xfs_iread_extents(NULL, ip, whichfork);
+		if (error)
+			return error;
+	}
+
+	xfs_bmap_search_extents(ip, bno, whichfork, &eof, &lastx, &got, &prev);
+	end = bno + len;
+	obno = bno;
+
+	while (bno < end && n < *nmap) {
+		/* Reading past eof, act as though there's a hole up to end. */
+		if (eof)
+			got.br_startoff = end;
+		if (got.br_startoff > bno) {
+			/* Reading in a hole.  */
+			mval->br_startoff = bno;
+			mval->br_startblock = HOLESTARTBLOCK;
+			mval->br_blockcount =
+				XFS_FILBLKS_MIN(len, got.br_startoff - bno);
+			mval->br_state = XFS_EXT_NORM;
+			bno += mval->br_blockcount;
+			len -= mval->br_blockcount;
+			mval++;
+			n++;
+			continue;
+		}
+
+		/* set up the extent map to return. */
+		xfs_bmapi_trim_map(mval, &got, &bno, len, obno, end, n, flags);
+		xfs_bmapi_update_map(&mval, &bno, &len, obno, end, &n, flags);
+
+		/* If we're done, stop now. */
+		if (bno >= end || n >= *nmap)
+			break;
+
+		/* Else go on to the next record. */
+		if (++lastx < ifp->if_bytes / sizeof(xfs_bmbt_rec_t))
+			xfs_bmbt_get_all(xfs_iext_get_ext(ifp, lastx), &got);
+		else
+			eof = 1;
+	}
+	*nmap = n;
+	return 0;
+}
+
+/*
  * Map file blocks to filesystem blocks.
  * File range is given by the bno/len pair.
  * Adds blocks to file if a write ("flags & XFS_BMAPI_WRITE" set)
@@ -5490,10 +5579,9 @@ xfs_getbmap(
 
 	do {
 		nmap = (nexleft > subnex) ? subnex : nexleft;
-		error = xfs_bmapi(NULL, ip, XFS_BB_TO_FSBT(mp, bmv->bmv_offset),
-				  XFS_BB_TO_FSB(mp, bmv->bmv_length),
-				  bmapi_flags, NULL, 0, map, &nmap,
-				  NULL);
+		error = xfs_bmapi_read(ip, XFS_BB_TO_FSBT(mp, bmv->bmv_offset),
+				       XFS_BB_TO_FSB(mp, bmv->bmv_length),
+				       map, &nmap, bmapi_flags);
 		if (error)
 			goto out_free_map;
 		ASSERT(nmap <= subnex);
@@ -6084,9 +6172,8 @@ xfs_bmap_punch_delalloc_range(
 		 * trying to remove a real extent (which requires a
 		 * transaction) or a hole, which is probably a bad idea...
 		 */
-		error = xfs_bmapi(NULL, ip, start_fsb, 1,
-				XFS_BMAPI_ENTIRE,  NULL, 0, &imap,
-				&nimaps, NULL);
+		error = xfs_bmapi_read(ip, start_fsb, 1, &imap, &nimaps,
+				       XFS_BMAPI_ENTIRE);
 
 		if (error) {
 			/* something screwed, just bail */
Index: xfs/fs/xfs/xfs_bmap.h
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.h	2011-07-26 17:43:02.000000000 +0200
+++ xfs/fs/xfs/xfs_bmap.h	2011-08-23 21:09:32.917054122 +0200
@@ -294,6 +294,10 @@ xfs_bmapi(
 	int			*nmap,		/* i/o: mval size/count */
 	xfs_bmap_free_t		*flist);	/* i/o: list extents to free */
 
+int	xfs_bmapi_read(struct xfs_inode *ip, xfs_fileoff_t bno,
+		xfs_filblks_t len, struct xfs_bmbt_irec *mval,
+		int *nmap, int flags);
+
 /*
  * Map file blocks to filesystem blocks, simple version.
  * One block only, read-only.
Index: xfs/fs/xfs/xfs_dir2_leaf.c
===================================================================
--- xfs.orig/fs/xfs/xfs_dir2_leaf.c	2011-07-28 15:35:56.000000000 +0200
+++ xfs/fs/xfs/xfs_dir2_leaf.c	2011-08-23 21:09:32.920387437 +0200
@@ -888,12 +888,10 @@ xfs_dir2_leaf_getdents(
 				 * we already have in the table.
 				 */
 				nmap = map_size - map_valid;
-				error = xfs_bmapi(NULL, dp,
-					map_off,
+				error = xfs_bmapi_read(dp, map_off,
 					xfs_dir2_byte_to_da(mp,
 						XFS_DIR2_LEAF_OFFSET) - map_off,
-					XFS_BMAPI_METADATA, NULL, 0,
-					&map[map_valid], &nmap, NULL);
+					&map[map_valid], &nmap, 0);
 				/*
 				 * Don't know if we should ignore this or
 				 * try to return an error.
Index: xfs/fs/xfs/xfs_inode.c
===================================================================
--- xfs.orig/fs/xfs/xfs_inode.c	2011-08-23 21:07:09.000000000 +0200
+++ xfs/fs/xfs/xfs_inode.c	2011-08-23 21:09:32.923720752 +0200
@@ -1187,6 +1187,7 @@ xfs_isize_check(
 	xfs_fileoff_t		map_first;
 	int			nimaps;
 	xfs_bmbt_irec_t		imaps[2];
+	int			error;
 
 	if (!S_ISREG(ip->i_d.di_mode))
 		return;
@@ -1203,13 +1204,12 @@ xfs_isize_check(
 	 * The filesystem could be shutting down, so bmapi may return
 	 * an error.
 	 */
-	if (xfs_bmapi(NULL, ip, map_first,
+	error = xfs_bmapi_read(ip, map_first,
 			 (XFS_B_TO_FSB(mp,
-				       (xfs_ufsize_t)XFS_MAXIOFFSET(mp)) -
-			  map_first),
-			 XFS_BMAPI_ENTIRE, NULL, 0, imaps, &nimaps,
-			 NULL))
-	    return;
+			       (xfs_ufsize_t)XFS_MAXIOFFSET(mp)) - map_first),
+			 imaps, &nimaps, XFS_BMAPI_ENTIRE);
+	if (error)
+		return;
 	ASSERT(nimaps == 1);
 	ASSERT(imaps[0].br_startblock == HOLESTARTBLOCK);
 }
Index: xfs/fs/xfs/xfs_iomap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_iomap.c	2011-07-27 18:48:07.000000000 +0200
+++ xfs/fs/xfs/xfs_iomap.c	2011-08-23 21:09:32.923720752 +0200
@@ -300,8 +300,8 @@ xfs_iomap_eof_want_preallocate(
 	while (count_fsb > 0) {
 		imaps = nimaps;
 		firstblock = NULLFSBLOCK;
-		error = xfs_bmapi(NULL, ip, start_fsb, count_fsb, 0,
-				  &firstblock, 0, imap, &imaps, NULL);
+		error = xfs_bmapi_read(ip, start_fsb, count_fsb, imap, &imaps,
+				       0);
 		if (error)
 			return error;
 		for (n = 0; n < imaps; n++) {
Index: xfs/fs/xfs/xfs_vnodeops.c
===================================================================
--- xfs.orig/fs/xfs/xfs_vnodeops.c	2011-08-23 04:40:04.000000000 +0200
+++ xfs/fs/xfs/xfs_vnodeops.c	2011-08-23 21:09:32.927054068 +0200
@@ -72,8 +72,8 @@ xfs_readlink_bmap(
 	xfs_buf_t	*bp;
 	int		error = 0;
 
-	error = xfs_bmapi(NULL, ip, 0, XFS_B_TO_FSB(mp, pathlen), 0, NULL, 0,
-			mval, &nmaps, NULL);
+	error = xfs_bmapi_read(ip, 0, XFS_B_TO_FSB(mp, pathlen), mval, &nmaps,
+			       0);
 	if (error)
 		goto out;
 
@@ -178,8 +178,7 @@ xfs_free_eofblocks(
 
 	nimaps = 1;
 	xfs_ilock(ip, XFS_ILOCK_SHARED);
-	error = xfs_bmapi(NULL, ip, end_fsb, map_len, 0,
-			  NULL, 0, &imap, &nimaps, NULL);
+	error = xfs_bmapi_read(ip, end_fsb, map_len, &imap, &nimaps, 0);
 	xfs_iunlock(ip, XFS_ILOCK_SHARED);
 
 	if (!error && (nimaps != 0) &&
@@ -297,9 +296,9 @@ xfs_inactive_symlink_rmt(
 	done = 0;
 	xfs_bmap_init(&free_list, &first_block);
 	nmaps = ARRAY_SIZE(mval);
-	if ((error = xfs_bmapi(tp, ip, 0, XFS_B_TO_FSB(mp, size),
-			XFS_BMAPI_METADATA, &first_block, 0, mval, &nmaps,
-			&free_list)))
+	error = xfs_bmapi_read(ip, 0, XFS_B_TO_FSB(mp, size),
+				mval, &nmaps, 0);
+	if (error)
 		goto error0;
 	/*
 	 * Invalidate the block(s).
@@ -1974,8 +1973,7 @@ xfs_zero_remaining_bytes(
 	for (offset = startoff; offset <= endoff; offset = lastoffset + 1) {
 		offset_fsb = XFS_B_TO_FSBT(mp, offset);
 		nimap = 1;
-		error = xfs_bmapi(NULL, ip, offset_fsb, 1, 0,
-			NULL, 0, &imap, &nimap, NULL);
+		error = xfs_bmapi_read(ip, offset_fsb, 1, &imap, &nimap, 0);
 		if (error || nimap < 1)
 			break;
 		ASSERT(imap.br_blockcount >= 1);
@@ -2094,8 +2092,8 @@ xfs_free_file_space(
 	 */
 	if (rt && !xfs_sb_version_hasextflgbit(&mp->m_sb)) {
 		nimap = 1;
-		error = xfs_bmapi(NULL, ip, startoffset_fsb,
-			1, 0, NULL, 0, &imap, &nimap, NULL);
+		error = xfs_bmapi_read(ip, startoffset_fsb, 1,
+					&imap, &nimap, 0);
 		if (error)
 			goto out_unlock_iolock;
 		ASSERT(nimap == 0 || nimap == 1);
@@ -2109,8 +2107,8 @@ xfs_free_file_space(
 				startoffset_fsb += mp->m_sb.sb_rextsize - mod;
 		}
 		nimap = 1;
-		error = xfs_bmapi(NULL, ip, endoffset_fsb - 1,
-			1, 0, NULL, 0, &imap, &nimap, NULL);
+		error = xfs_bmapi_read(ip, endoffset_fsb - 1, 1,
+					&imap, &nimap, 0);
 		if (error)
 			goto out_unlock_iolock;
 		ASSERT(nimap == 0 || nimap == 1);
Index: xfs/fs/xfs/xfs_da_btree.c
===================================================================
--- xfs.orig/fs/xfs/xfs_da_btree.c	2011-08-12 04:36:18.000000000 +0200
+++ xfs/fs/xfs/xfs_da_btree.c	2011-08-23 21:09:32.930387384 +0200
@@ -1995,11 +1995,10 @@ xfs_da_do_buf(
 		} else {
 			mapp = kmem_alloc(sizeof(*mapp) * nfsb, KM_SLEEP);
 			nmap = nfsb;
-			if ((error = xfs_bmapi(trans, dp, (xfs_fileoff_t)bno,
-					nfsb,
-					XFS_BMAPI_METADATA |
-						xfs_bmapi_aflag(whichfork),
-					NULL, 0, mapp, &nmap, NULL)))
+			error = xfs_bmapi_read(dp, (xfs_fileoff_t)bno, nfsb,
+					       mapp, &nmap,
+					       xfs_bmapi_aflag(whichfork));
+			if (error)
 				goto exit0;
 		}
 	} else {

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

[PATCH 06/25] xfs: remove xfs_bmapi_single()

[Christoph Hellwig]

[-- Attachment #1: xfs-bmapi-remove-xfs_bmapi_single --]
[-- Type: text/plain, Size: 6299 bytes --]

Now we have xfs_bmapi_read, there is no need for xfs_bmapi_single().
Change the remaining caller over and kill the function.

Signed-off-by: Dave Chinner <dchinner@redhat.com>

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:09:32.913720807 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:10:14.383496146 +0200
@@ -4966,58 +4966,6 @@ error0:
 }
 
 /*
- * Map file blocks to filesystem blocks, simple version.
- * One block (extent) only, read-only.
- * For flags, only the XFS_BMAPI_ATTRFORK flag is examined.
- * For the other flag values, the effect is as if XFS_BMAPI_METADATA
- * was set and all the others were clear.
- */
-int						/* error */
-xfs_bmapi_single(
-	xfs_trans_t	*tp,		/* transaction pointer */
-	xfs_inode_t	*ip,		/* incore inode */
-	int		whichfork,	/* data or attr fork */
-	xfs_fsblock_t	*fsb,		/* output: mapped block */
-	xfs_fileoff_t	bno)		/* starting file offs. mapped */
-{
-	int		eof;		/* we've hit the end of extents */
-	int		error;		/* error return */
-	xfs_bmbt_irec_t	got;		/* current file extent record */
-	xfs_ifork_t	*ifp;		/* inode fork pointer */
-	xfs_extnum_t	lastx;		/* last useful extent number */
-	xfs_bmbt_irec_t	prev;		/* previous file extent record */
-
-	ifp = XFS_IFORK_PTR(ip, whichfork);
-	if (unlikely(
-	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE &&
-	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS)) {
-	       XFS_ERROR_REPORT("xfs_bmapi_single", XFS_ERRLEVEL_LOW,
-				ip->i_mount);
-	       return XFS_ERROR(EFSCORRUPTED);
-	}
-	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
-		return XFS_ERROR(EIO);
-	XFS_STATS_INC(xs_blk_mapr);
-	if (!(ifp->if_flags & XFS_IFEXTENTS) &&
-	    (error = xfs_iread_extents(tp, ip, whichfork)))
-		return error;
-	(void)xfs_bmap_search_extents(ip, bno, whichfork, &eof, &lastx, &got,
-		&prev);
-	/*
-	 * Reading past eof, act as though there's a hole
-	 * up to end.
-	 */
-	if (eof || got.br_startoff > bno) {
-		*fsb = NULLFSBLOCK;
-		return 0;
-	}
-	ASSERT(!isnullstartblock(got.br_startblock));
-	ASSERT(bno < got.br_startoff + got.br_blockcount);
-	*fsb = got.br_startblock + (bno - got.br_startoff);
-	return 0;
-}
-
-/*
  * Unmap (remove) blocks from a file.
  * If nexts is nonzero then the number of extents to remove is limited to
  * that value.  If not all extents in the block range can be removed then
Index: xfs/fs/xfs/xfs_bmap.h
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.h	2011-08-23 21:09:32.917054122 +0200
+++ xfs/fs/xfs/xfs_bmap.h	2011-08-23 21:10:14.383496146 +0200
@@ -299,21 +299,6 @@ int	xfs_bmapi_read(struct xfs_inode *ip,
 		int *nmap, int flags);
 
 /*
- * Map file blocks to filesystem blocks, simple version.
- * One block only, read-only.
- * For flags, only the XFS_BMAPI_ATTRFORK flag is examined.
- * For the other flag values, the effect is as if XFS_BMAPI_METADATA
- * was set and all the others were clear.
- */
-int						/* error */
-xfs_bmapi_single(
-	struct xfs_trans	*tp,		/* transaction pointer */
-	struct xfs_inode	*ip,		/* incore inode */
-	int			whichfork,	/* data or attr fork */
-	xfs_fsblock_t		*fsb,		/* output: mapped block */
-	xfs_fileoff_t		bno);		/* starting file offs. mapped */
-
-/*
  * Unmap (remove) blocks from a file.
  * If nexts is nonzero then the number of extents to remove is limited to
  * that value.  If not all extents in the block range can be removed then
Index: xfs/fs/xfs/xfs_rtalloc.c
===================================================================
--- xfs.orig/fs/xfs/xfs_rtalloc.c	2011-08-12 04:36:18.000000000 +0200
+++ xfs/fs/xfs/xfs_rtalloc.c	2011-08-23 21:12:01.472915992 +0200
@@ -856,33 +856,23 @@ xfs_rtbuf_get(
 	xfs_buf_t	**bpp)		/* output: buffer for the block */
 {
 	xfs_buf_t	*bp;		/* block buffer, result */
-	xfs_daddr_t	d;		/* disk addr of block */
-	int		error;		/* error value */
-	xfs_fsblock_t	fsb;		/* fs block number for block */
 	xfs_inode_t	*ip;		/* bitmap or summary inode */
+	xfs_bmbt_irec_t	map;
+	int		nmap;
+	int		error;		/* error value */
 
 	ip = issum ? mp->m_rsumip : mp->m_rbmip;
-	/*
-	 * Map from the file offset (block) and inode number to the
-	 * file system block.
-	 */
-	error = xfs_bmapi_single(tp, ip, XFS_DATA_FORK, &fsb, block);
-	if (error) {
+
+	error = xfs_bmapi_read(ip, block, 1, &map, &nmap, XFS_DATA_FORK);
+	if (error)
 		return error;
-	}
-	ASSERT(fsb != NULLFSBLOCK);
-	/*
-	 * Convert to disk address for buffer cache.
-	 */
-	d = XFS_FSB_TO_DADDR(mp, fsb);
-	/*
-	 * Read the buffer.
-	 */
-	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp, d,
+
+	ASSERT(map.br_startblock != NULLFSBLOCK);
+	error = xfs_trans_read_buf(mp, tp, mp->m_ddev_targp,
+				   XFS_FSB_TO_DADDR(mp, map.br_startblock),
 				   mp->m_bsize, 0, &bp);
-	if (error) {
+	if (error)
 		return error;
-	}
 	ASSERT(!xfs_buf_geterror(bp));
 	*bpp = bp;
 	return 0;
Index: xfs/fs/xfs/xfs_da_btree.c
===================================================================
--- xfs.orig/fs/xfs/xfs_da_btree.c	2011-08-23 21:09:32.000000000 +0200
+++ xfs/fs/xfs/xfs_da_btree.c	2011-08-23 21:10:14.390162776 +0200
@@ -1975,32 +1975,16 @@ xfs_da_do_buf(
 		/*
 		 * Optimize the one-block case.
 		 */
-		if (nfsb == 1) {
-			xfs_fsblock_t	fsb;
-
-			if ((error =
-			    xfs_bmapi_single(trans, dp, whichfork, &fsb,
-				    (xfs_fileoff_t)bno))) {
-				return error;
-			}
+		if (nfsb == 1)
 			mapp = &map;
-			if (fsb == NULLFSBLOCK) {
-				nmap = 0;
-			} else {
-				map.br_startblock = fsb;
-				map.br_startoff = (xfs_fileoff_t)bno;
-				map.br_blockcount = 1;
-				nmap = 1;
-			}
-		} else {
+		else
 			mapp = kmem_alloc(sizeof(*mapp) * nfsb, KM_SLEEP);
-			nmap = nfsb;
-			error = xfs_bmapi_read(dp, (xfs_fileoff_t)bno, nfsb,
-					       mapp, &nmap,
-					       xfs_bmapi_aflag(whichfork));
-			if (error)
-				goto exit0;
-		}
+
+		nmap = nfsb;
+		error = xfs_bmapi_read(dp, (xfs_fileoff_t)bno, nfsb, mapp,
+				       &nmap, xfs_bmapi_aflag(whichfork));
+		if (error)
+			goto exit0;
 	} else {
 		map.br_startblock = XFS_DADDR_TO_FSB(mp, mappedbno);
 		map.br_startoff = (xfs_fileoff_t)bno;

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

[PATCH 07/25] xfs: factor delalloc reservations out of xfs_bmapi

[Christoph Hellwig]

[-- Attachment #1: xfs-bmapi-factor-delalloc-reservation --]
[-- Type: text/plain, Size: 7211 bytes --]

Move the reservation of delayed allocations, and addition of delalloc
regions to the extent trees into a new helper function.  For now
this adds some twisted goto logic to xfs_bmapi, but that will be
cleaned up in the following patches.

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

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:10:14.383496146 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:12:56.272619118 +0200
@@ -4438,6 +4438,120 @@ xfs_bmapi_read(
 	return 0;
 }
 
+STATIC int
+xfs_bmapi_reserve_delalloc(
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		aoff,
+	xfs_filblks_t		len,
+	struct xfs_bmbt_irec	*got,
+	struct xfs_bmbt_irec	*prev,
+	xfs_extnum_t		*lastx,
+	int			eof)
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
+	xfs_extlen_t		alen;
+	xfs_extlen_t		indlen;
+	xfs_fsblock_t		firstblock = NULLFSBLOCK;
+	struct xfs_btree_cur	*cur = NULL;
+	int			tmp_logflags = 0;
+	char			rt = XFS_IS_REALTIME_INODE(ip);
+	xfs_extlen_t		extsz;
+	int			error;
+
+	alen = XFS_FILBLKS_MIN(len, MAXEXTLEN);
+	if (!eof)
+		alen = XFS_FILBLKS_MIN(alen, got->br_startoff - aoff);
+
+	/* Figure out the extent size, adjust alen */
+	extsz = xfs_get_extsz_hint(ip);
+	if (extsz) {
+		/*
+		 * Make sure we don't exceed a single extent length when we
+		 * align the extent by reducing length we are going to
+		 * allocate by the maximum amount extent size aligment may
+		 * require.
+		 */
+		alen = XFS_FILBLKS_MIN(len, MAXEXTLEN - (2 * extsz - 1));
+		error = xfs_bmap_extsize_align(mp, got, prev, extsz, rt, eof,
+					       1, 0, &aoff, &alen);
+		ASSERT(!error);
+	}
+
+	if (rt)
+		extsz = alen / mp->m_sb.sb_rextsize;
+
+	/*
+	 * Make a transaction-less quota reservation for delayed allocation
+	 * blocks.  This number gets adjusted later.  We return if we haven't
+	 * allocated blocks already inside this loop.
+	 */
+	error = xfs_trans_reserve_quota_nblks(NULL, ip, (long)alen, 0,
+			rt ? XFS_QMOPT_RES_RTBLKS : XFS_QMOPT_RES_REGBLKS);
+	if (error)
+		return error;
+
+	/*
+	 * Split changing sb for alen and indlen since they could be coming
+	 * from different places.
+	 */
+	indlen = (xfs_extlen_t)xfs_bmap_worst_indlen(ip, alen);
+	ASSERT(indlen > 0);
+
+	if (rt) {
+		error = xfs_mod_incore_sb(mp, XFS_SBS_FREXTENTS,
+					  -((int64_t)extsz), 0);
+	} else {
+		error = xfs_icsb_modify_counters(mp, XFS_SBS_FDBLOCKS,
+						 -((int64_t)alen), 0);
+	}
+
+	if (error)
+		goto out_unreserve_quota;
+
+	error = xfs_icsb_modify_counters(mp, XFS_SBS_FDBLOCKS,
+					 -((int64_t)indlen), 0);
+	if (error)
+		goto out_unreserve_blocks;
+
+
+	ip->i_delayed_blks += alen;
+
+	got->br_startoff = aoff;
+	got->br_startblock = nullstartblock(indlen);
+	got->br_blockcount = alen;
+	got->br_state = XFS_EXT_NORM;
+
+	error = xfs_bmap_add_extent(NULL, ip, lastx, &cur, got, &firstblock,
+				    NULL, &tmp_logflags, XFS_DATA_FORK);
+	ASSERT(!error);
+	ASSERT(!tmp_logflags);
+	ASSERT(!cur);
+
+	/*
+	 * Update our extent pointer, given that xfs_bmap_add_extent might
+	 * have merged it into one of the neighbouring ones.
+	 */
+	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx), got);
+
+	ASSERT(got->br_startoff <= aoff);
+	ASSERT(got->br_startoff + got->br_blockcount >= aoff + alen);
+	ASSERT(isnullstartblock(got->br_startblock));
+	ASSERT(got->br_state == XFS_EXT_NORM);
+	return 0;
+
+out_unreserve_blocks:
+	if (rt)
+		xfs_mod_incore_sb(mp, XFS_SBS_FREXTENTS, extsz, 0);
+	else
+		xfs_icsb_modify_counters(mp, XFS_SBS_FDBLOCKS, alen, 0);
+out_unreserve_quota:
+	if (XFS_IS_QUOTA_ON(mp))
+		xfs_trans_unreserve_quota_nblks(NULL, ip, alen, 0, rt ?
+				XFS_QMOPT_RES_RTBLKS : XFS_QMOPT_RES_REGBLKS);
+	return error;
+}
+
 /*
  * Map file blocks to filesystem blocks.
  * File range is given by the bno/len pair.
@@ -4476,7 +4590,6 @@ xfs_bmapi(
 	int		error;		/* error return */
 	xfs_bmbt_irec_t	got;		/* current file extent record */
 	xfs_ifork_t	*ifp;		/* inode fork pointer */
-	xfs_extlen_t	indlen;		/* indirect blocks length */
 	xfs_extnum_t	lastx;		/* last useful extent number */
 	int		logflags;	/* flags for transaction logging */
 	xfs_extlen_t	minleft;	/* min blocks left after allocation */
@@ -4612,43 +4725,8 @@ xfs_bmapi(
 			}
 			minlen = (flags & XFS_BMAPI_CONTIG) ? alen : 1;
 			if (flags & XFS_BMAPI_DELAY) {
-				xfs_extlen_t	extsz;
-
-				/* Figure out the extent size, adjust alen */
-				extsz = xfs_get_extsz_hint(ip);
-				if (extsz) {
-					/*
-					 * make sure we don't exceed a single
-					 * extent length when we align the
-					 * extent by reducing length we are
-					 * going to allocate by the maximum
-					 * amount extent size aligment may
-					 * require.
-					 */
-					alen = XFS_FILBLKS_MIN(len,
-						   MAXEXTLEN - (2 * extsz - 1));
-					error = xfs_bmap_extsize_align(mp,
-							&got, &prev, extsz,
-							rt, eof,
-							flags&XFS_BMAPI_DELAY,
-							flags&XFS_BMAPI_CONVERT,
-							&aoff, &alen);
-					ASSERT(!error);
-				}
-
-				if (rt)
-					extsz = alen / mp->m_sb.sb_rextsize;
-
-				/*
-				 * Make a transaction-less quota reservation for
-				 * delayed allocation blocks. This number gets
-				 * adjusted later.  We return if we haven't
-				 * allocated blocks already inside this loop.
-				 */
-				error = xfs_trans_reserve_quota_nblks(
-						NULL, ip, (long)alen, 0,
-						rt ? XFS_QMOPT_RES_RTBLKS :
-						     XFS_QMOPT_RES_REGBLKS);
+				error = xfs_bmapi_reserve_delalloc(ip, bno, len, &got,
+								   &prev, &lastx, eof);
 				if (error) {
 					if (n == 0) {
 						*nmap = 0;
@@ -4658,51 +4736,7 @@ xfs_bmapi(
 					break;
 				}
 
-				/*
-				 * Split changing sb for alen and indlen since
-				 * they could be coming from different places.
-				 */
-				indlen = (xfs_extlen_t)
-					xfs_bmap_worst_indlen(ip, alen);
-				ASSERT(indlen > 0);
-
-				if (rt) {
-					error = xfs_mod_incore_sb(mp,
-							XFS_SBS_FREXTENTS,
-							-((int64_t)extsz), 0);
-				} else {
-					error = xfs_icsb_modify_counters(mp,
-							XFS_SBS_FDBLOCKS,
-							-((int64_t)alen), 0);
-				}
-				if (!error) {
-					error = xfs_icsb_modify_counters(mp,
-							XFS_SBS_FDBLOCKS,
-							-((int64_t)indlen), 0);
-					if (error && rt)
-						xfs_mod_incore_sb(mp,
-							XFS_SBS_FREXTENTS,
-							(int64_t)extsz, 0);
-					else if (error)
-						xfs_icsb_modify_counters(mp,
-							XFS_SBS_FDBLOCKS,
-							(int64_t)alen, 0);
-				}
-
-				if (error) {
-					if (XFS_IS_QUOTA_ON(mp))
-						/* unreserve the blocks now */
-						(void)
-						xfs_trans_unreserve_quota_nblks(
-							NULL, ip,
-							(long)alen, 0, rt ?
-							XFS_QMOPT_RES_RTBLKS :
-							XFS_QMOPT_RES_REGBLKS);
-					break;
-				}
-
-				ip->i_delayed_blks += alen;
-				abno = nullstartblock(indlen);
+				goto trim_extent;
 			} else {
 				/*
 				 * If first time, allocate and fill in
@@ -4840,7 +4874,7 @@ xfs_bmapi(
 			n++;
 			continue;
 		}
-
+trim_extent:
 		/* Deal with the allocated space we found.  */
 		xfs_bmapi_trim_map(mval, &got, &bno, len, obno, end, n, flags);
 

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

[PATCH 08/25] xfs: introduce xfs_bmapi_delay()

[Christoph Hellwig]

[-- Attachment #1: xfs-bmapi-add-xfs_bmapi_delay --]
[-- Type: text/plain, Size: 8132 bytes --]

Delalloc reservations are much simpler than allocations, so give them a
separate bmapi-level interface.  Using the previously added
xfs_bmapi_reserve_delalloc we get a function that is only minimally
more complicated than xfs_bmapi_read, which is far from the complexity
in xfs_bmapi.  Also remove the XFS_BMAPI_DELAY code after switching
over the only user to xfs_bmapi_delay.

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

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:12:56.272619118 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:12:59.355935746 +0200
@@ -4238,7 +4238,7 @@ xfs_bmap_validate_ret(
 		ASSERT(i == 0 ||
 		       mval[i - 1].br_startoff + mval[i - 1].br_blockcount ==
 		       mval[i].br_startoff);
-		if ((flags & XFS_BMAPI_WRITE) && !(flags & XFS_BMAPI_DELAY))
+		if (flags & XFS_BMAPI_WRITE)
 			ASSERT(mval[i].br_startblock != DELAYSTARTBLOCK &&
 			       mval[i].br_startblock != HOLESTARTBLOCK);
 		ASSERT(mval[i].br_state == XFS_EXT_NORM ||
@@ -4553,6 +4553,90 @@ out_unreserve_quota:
 }
 
 /*
+ * Map file blocks to filesystem blocks, adding delayed allocations as needed.
+ */
+int
+xfs_bmapi_delay(
+	struct xfs_inode	*ip,	/* incore inode */
+	xfs_fileoff_t		bno,	/* starting file offs. mapped */
+	xfs_filblks_t		len,	/* length to map in file */
+	struct xfs_bmbt_irec	*mval,	/* output: map values */
+	int			*nmap,	/* i/o: mval size/count */
+	int			flags)	/* XFS_BMAPI_... */
+{
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
+	struct xfs_bmbt_irec	got;	/* current file extent record */
+	struct xfs_bmbt_irec	prev;	/* previous file extent record */
+	xfs_fileoff_t		obno;	/* old block number (offset) */
+	xfs_fileoff_t		end;	/* end of mapped file region */
+	xfs_extnum_t		lastx;	/* last useful extent number */
+	int			eof;	/* we've hit the end of extents */
+	int			n = 0;	/* current extent index */
+	int			error = 0;
+
+	ASSERT(*nmap >= 1);
+	ASSERT(*nmap <= XFS_BMAP_MAX_NMAP);
+	ASSERT(!(flags & ~XFS_BMAPI_ENTIRE));
+
+	if (unlikely(XFS_TEST_ERROR(
+	    (XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) != XFS_DINODE_FMT_EXTENTS &&
+	     XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) != XFS_DINODE_FMT_BTREE),
+	     mp, XFS_ERRTAG_BMAPIFORMAT, XFS_RANDOM_BMAPIFORMAT))) {
+		XFS_ERROR_REPORT("xfs_bmapi_delay", XFS_ERRLEVEL_LOW, mp);
+		return XFS_ERROR(EFSCORRUPTED);
+	}
+
+	if (XFS_FORCED_SHUTDOWN(mp))
+		return XFS_ERROR(EIO);
+
+	XFS_STATS_INC(xs_blk_mapw);
+
+	if (!(ifp->if_flags & XFS_IFEXTENTS)) {
+		error = xfs_iread_extents(NULL, ip, XFS_DATA_FORK);
+		if (error)
+			return error;
+	}
+
+	xfs_bmap_search_extents(ip, bno, XFS_DATA_FORK, &eof, &lastx, &got, &prev);
+	end = bno + len;
+	obno = bno;
+
+	while (bno < end && n < *nmap) {
+		if (eof || got.br_startoff > bno) {
+			error = xfs_bmapi_reserve_delalloc(ip, bno, len, &got,
+							   &prev, &lastx, eof);
+			if (error) {
+				if (n == 0) {
+					*nmap = 0;
+					return error;
+				}
+				break;
+			}
+		}
+
+		/* set up the extent map to return. */
+		xfs_bmapi_trim_map(mval, &got, &bno, len, obno, end, n, flags);
+		xfs_bmapi_update_map(&mval, &bno, &len, obno, end, &n, flags);
+
+		/* If we're done, stop now. */
+		if (bno >= end || n >= *nmap)
+			break;
+
+		/* Else go on to the next record. */
+		prev = got;
+		if (++lastx < ifp->if_bytes / sizeof(xfs_bmbt_rec_t))
+			xfs_bmbt_get_all(xfs_iext_get_ext(ifp, lastx), &got);
+		else
+			eof = 1;
+	}
+
+	*nmap = n;
+	return 0;
+}
+
+
+/*
  * Map file blocks to filesystem blocks.
  * File range is given by the bno/len pair.
  * Adds blocks to file if a write ("flags & XFS_BMAPI_WRITE" set)
@@ -4660,7 +4744,6 @@ xfs_bmapi(
 	 */
 	if ((flags & XFS_BMAPI_IGSTATE) && wr)	/* if writing unwritten space */
 		wr = 0;				/* no allocations are allowed */
-	ASSERT(wr || !(flags & XFS_BMAPI_DELAY));
 	logflags = 0;
 	nallocs = 0;
 	cur = NULL;
@@ -4695,8 +4778,7 @@ xfs_bmapi(
 		if (eof && !wr)
 			got.br_startoff = end;
 		inhole = eof || got.br_startoff > bno;
-		wasdelay = wr && !inhole && !(flags & XFS_BMAPI_DELAY) &&
-			isnullstartblock(got.br_startblock);
+		wasdelay = wr && !inhole && isnullstartblock(got.br_startblock);
 		/*
 		 * First, deal with the hole before the allocated space
 		 * that we found, if any.
@@ -4724,20 +4806,7 @@ xfs_bmapi(
 				aoff = bno;
 			}
 			minlen = (flags & XFS_BMAPI_CONTIG) ? alen : 1;
-			if (flags & XFS_BMAPI_DELAY) {
-				error = xfs_bmapi_reserve_delalloc(ip, bno, len, &got,
-								   &prev, &lastx, eof);
-				if (error) {
-					if (n == 0) {
-						*nmap = 0;
-						ASSERT(cur == NULL);
-						return error;
-					}
-					break;
-				}
-
-				goto trim_extent;
-			} else {
+			{
 				/*
 				 * If first time, allocate and fill in
 				 * once-only bma fields.
@@ -4848,14 +4917,8 @@ xfs_bmapi(
 			ASSERT(got.br_startoff <= aoff);
 			ASSERT(got.br_startoff + got.br_blockcount >=
 				aoff + alen);
-#ifdef DEBUG
-			if (flags & XFS_BMAPI_DELAY) {
-				ASSERT(isnullstartblock(got.br_startblock));
-				ASSERT(startblockval(got.br_startblock) > 0);
-			}
 			ASSERT(got.br_state == XFS_EXT_NORM ||
 			       got.br_state == XFS_EXT_UNWRITTEN);
-#endif
 			/*
 			 * Fall down into the found allocated space case.
 			 */
@@ -4874,7 +4937,7 @@ xfs_bmapi(
 			n++;
 			continue;
 		}
-trim_extent:
+
 		/* Deal with the allocated space we found.  */
 		xfs_bmapi_trim_map(mval, &got, &bno, len, obno, end, n, flags);
 
@@ -4884,7 +4947,7 @@ trim_extent:
 		 */
 		if (wr &&
 		    ((mval->br_state == XFS_EXT_UNWRITTEN &&
-		      ((flags & (XFS_BMAPI_PREALLOC|XFS_BMAPI_DELAY)) == 0)) ||
+		      ((flags & XFS_BMAPI_PREALLOC) == 0)) ||
 		     (mval->br_state == XFS_EXT_NORM &&
 		      ((flags & (XFS_BMAPI_PREALLOC|XFS_BMAPI_CONVERT)) ==
 				(XFS_BMAPI_PREALLOC|XFS_BMAPI_CONVERT))))) {
Index: xfs/fs/xfs/xfs_bmap.h
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.h	2011-08-23 21:10:14.383496146 +0200
+++ xfs/fs/xfs/xfs_bmap.h	2011-08-23 21:12:59.355935746 +0200
@@ -65,7 +65,6 @@ typedef	struct xfs_bmap_free
  * Flags for xfs_bmapi
  */
 #define	XFS_BMAPI_WRITE		0x001	/* write operation: allocate space */
-#define XFS_BMAPI_DELAY		0x002	/* delayed write operation */
 #define XFS_BMAPI_ENTIRE	0x004	/* return entire extent, not trimmed */
 #define XFS_BMAPI_METADATA	0x008	/* mapping metadata not user data */
 #define XFS_BMAPI_ATTRFORK	0x010	/* use attribute fork not data */
@@ -82,7 +81,6 @@ typedef	struct xfs_bmap_free
 
 #define XFS_BMAPI_FLAGS \
 	{ XFS_BMAPI_WRITE,	"WRITE" }, \
-	{ XFS_BMAPI_DELAY,	"DELAY" }, \
 	{ XFS_BMAPI_ENTIRE,	"ENTIRE" }, \
 	{ XFS_BMAPI_METADATA,	"METADATA" }, \
 	{ XFS_BMAPI_ATTRFORK,	"ATTRFORK" }, \
@@ -297,6 +295,9 @@ xfs_bmapi(
 int	xfs_bmapi_read(struct xfs_inode *ip, xfs_fileoff_t bno,
 		xfs_filblks_t len, struct xfs_bmbt_irec *mval,
 		int *nmap, int flags);
+int	xfs_bmapi_delay(struct xfs_inode *ip, xfs_fileoff_t bno,
+		xfs_filblks_t len, struct xfs_bmbt_irec *mval,
+		int *nmap, int flags);
 
 /*
  * Unmap (remove) blocks from a file.
Index: xfs/fs/xfs/xfs_iomap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_iomap.c	2011-08-23 21:09:32.923720752 +0200
+++ xfs/fs/xfs/xfs_iomap.c	2011-08-23 21:12:59.355935746 +0200
@@ -381,7 +381,6 @@ xfs_iomap_write_delay(
 	xfs_fileoff_t	last_fsb;
 	xfs_off_t	aligned_offset;
 	xfs_fileoff_t	ioalign;
-	xfs_fsblock_t	firstblock;
 	xfs_extlen_t	extsz;
 	int		nimaps;
 	xfs_bmbt_irec_t imap[XFS_WRITE_IMAPS];
@@ -425,12 +424,8 @@ retry:
 	}
 
 	nimaps = XFS_WRITE_IMAPS;
-	firstblock = NULLFSBLOCK;
-	error = xfs_bmapi(NULL, ip, offset_fsb,
-			  (xfs_filblks_t)(last_fsb - offset_fsb),
-			  XFS_BMAPI_DELAY | XFS_BMAPI_WRITE |
-			  XFS_BMAPI_ENTIRE, &firstblock, 1, imap,
-			  &nimaps, NULL);
+	error = xfs_bmapi_delay(ip, offset_fsb, last_fsb - offset_fsb,
+				imap, &nimaps, XFS_BMAPI_ENTIRE);
 	switch (error) {
 	case 0:
 	case ENOSPC:

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

[PATCH 09/25] xfs: do not use xfs_bmap_add_extent for adding delalloc extents

[Christoph Hellwig]

[-- Attachment #1: xfs-bmapi-delay-simplify --]
[-- Type: text/plain, Size: 3990 bytes --]

We can just call xfs_bmap_add_extent_hole_delay directly to add a delayed
allocated regions to the extent tree, instead of going through all the
complexities of xfs_bmap_add_extent that aren't needed for this simple
case.

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

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:12:59.355935746 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:13:31.625760927 +0200
@@ -102,17 +102,6 @@ xfs_bmap_add_extent_delay_real(
 
 /*
  * Called by xfs_bmap_add_extent to handle cases converting a hole
- * to a delayed allocation.
- */
-STATIC int				/* error */
-xfs_bmap_add_extent_hole_delay(
-	xfs_inode_t		*ip,	/* incore inode pointer */
-	xfs_extnum_t		*idx,	/* extent number to update/insert */
-	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
-	int			*logflagsp); /* inode logging flags */
-
-/*
- * Called by xfs_bmap_add_extent to handle cases converting a hole
  * to a real allocation.
  */
 STATIC int				/* error */
@@ -431,8 +420,7 @@ xfs_bmap_add_attrfork_local(
 }
 
 /*
- * Called by xfs_bmapi to update file extent records and the btree
- * after allocating space (or doing a delayed allocation).
+ * Update file extent records and the btree after allocating space.
  */
 STATIC int				/* error */
 xfs_bmap_add_extent(
@@ -464,21 +452,12 @@ xfs_bmap_add_extent(
 
 	ASSERT(*idx >= 0);
 	ASSERT(*idx <= nextents);
+	ASSERT(!isnullstartblock(new->br_startblock));
 
 	/*
-	 * Any kind of new delayed allocation goes here.
-	 */
-	if (isnullstartblock(new->br_startblock)) {
-		if (cur)
-			ASSERT((cur->bc_private.b.flags &
-				XFS_BTCUR_BPRV_WASDEL) == 0);
-		error = xfs_bmap_add_extent_hole_delay(ip, idx, new,
-						       &logflags);
-	}
-	/*
 	 * Real allocation off the end of the file.
 	 */
-	else if (*idx == nextents) {
+	if (*idx == nextents) {
 		if (cur)
 			ASSERT((cur->bc_private.b.flags &
 				XFS_BTCUR_BPRV_WASDEL) == 0);
@@ -1581,16 +1560,13 @@ done:
 }
 
 /*
- * Called by xfs_bmap_add_extent to handle cases converting a hole
- * to a delayed allocation.
+ * Convert a hole to a delayed allocation.
  */
-/*ARGSUSED*/
-STATIC int				/* error */
+STATIC void
 xfs_bmap_add_extent_hole_delay(
 	xfs_inode_t		*ip,	/* incore inode pointer */
 	xfs_extnum_t		*idx,	/* extent number to update/insert */
-	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
-	int			*logflagsp) /* inode logging flags */
+	xfs_bmbt_irec_t		*new)	/* new data to add to file extents */
 {
 	xfs_ifork_t		*ifp;	/* inode fork pointer */
 	xfs_bmbt_irec_t		left;	/* left neighbor extent entry */
@@ -1725,8 +1701,6 @@ xfs_bmap_add_extent_hole_delay(
 		 * Nothing to do for disk quota accounting here.
 		 */
 	}
-	*logflagsp = 0;
-	return 0;
 }
 
 /*
@@ -4452,9 +4426,6 @@ xfs_bmapi_reserve_delalloc(
 	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
 	xfs_extlen_t		alen;
 	xfs_extlen_t		indlen;
-	xfs_fsblock_t		firstblock = NULLFSBLOCK;
-	struct xfs_btree_cur	*cur = NULL;
-	int			tmp_logflags = 0;
 	char			rt = XFS_IS_REALTIME_INODE(ip);
 	xfs_extlen_t		extsz;
 	int			error;
@@ -4521,16 +4492,11 @@ xfs_bmapi_reserve_delalloc(
 	got->br_startblock = nullstartblock(indlen);
 	got->br_blockcount = alen;
 	got->br_state = XFS_EXT_NORM;
-
-	error = xfs_bmap_add_extent(NULL, ip, lastx, &cur, got, &firstblock,
-				    NULL, &tmp_logflags, XFS_DATA_FORK);
-	ASSERT(!error);
-	ASSERT(!tmp_logflags);
-	ASSERT(!cur);
+	xfs_bmap_add_extent_hole_delay(ip, lastx, got);
 
 	/*
-	 * Update our extent pointer, given that xfs_bmap_add_extent might
-	 * have merged it into one of the neighbouring ones.
+	 * Update our extent pointer, given that xfs_bmap_add_extent_hole_delay
+	 * might have merged it into one of the neighbouring ones.
 	 */
 	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx), got);
 

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

[PATCH 10/25] xfs: factor extent allocation out of xfs_bmapi

[Christoph Hellwig]

[-- Attachment #1: xfs-bmapi-factor-allocate --]
[-- Type: text/plain, Size: 10403 bytes --]

To further improve the readability of xfs_bmapi(), factor the extent
allocation out into a separate function. This removes a large block
of logic from the xfs_bmapi() code loop and makes it easier to see
the operational logic flow for xfs_bmapi().

Signed-off-by: Dave Chinner <dchinner@redhat.com>

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:13:31.625760927 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:13:33.855748845 +0200
@@ -4602,6 +4602,147 @@ xfs_bmapi_delay(
 }
 
 
+STATIC int
+xfs_bmapi_allocate(
+	struct xfs_bmalloca	*bma,
+	xfs_extnum_t		*lastx,
+	struct xfs_btree_cur	**cur,
+	xfs_fsblock_t		*firstblock,
+	struct xfs_bmap_free	*flist,
+	int			flags,
+	int			*nallocs,
+	int			*logflags)
+{
+	struct xfs_mount	*mp = bma->ip->i_mount;
+	int			whichfork = (flags & XFS_BMAPI_ATTRFORK) ?
+						XFS_ATTR_FORK : XFS_DATA_FORK;
+	struct xfs_ifork	*ifp = XFS_IFORK_PTR(bma->ip, whichfork);
+	xfs_fsblock_t		abno;
+	xfs_extlen_t		alen;
+	xfs_fileoff_t		aoff;
+	int			error;
+	int			rt;
+
+	rt = (whichfork == XFS_DATA_FORK) && XFS_IS_REALTIME_INODE(bma->ip);
+
+	/*
+	 * For the wasdelay case, we could also just allocate the stuff asked
+	 * for in this bmap call but that wouldn't be as good.
+	 */
+	if (bma->wasdel) {
+		alen = (xfs_extlen_t)bma->gotp->br_blockcount;
+		aoff = bma->gotp->br_startoff;
+		if (*lastx != NULLEXTNUM && *lastx) {
+			xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx - 1),
+					 bma->prevp);
+		}
+	} else {
+		alen = (xfs_extlen_t)XFS_FILBLKS_MIN(bma->alen, MAXEXTLEN);
+		if (!bma->eof)
+			alen = (xfs_extlen_t)XFS_FILBLKS_MIN(alen,
+					bma->gotp->br_startoff - bma->off);
+		aoff = bma->off;
+	}
+
+	/*
+	 * Indicate if this is the first user data in the file, or just any
+	 * user data.
+	 */
+	if (!(flags & XFS_BMAPI_METADATA)) {
+		bma->userdata = (aoff == 0) ?
+			XFS_ALLOC_INITIAL_USER_DATA : XFS_ALLOC_USERDATA;
+	}
+
+	/*
+	 * Fill in changeable bma fields.
+	 */
+	bma->alen = alen;
+	bma->off = aoff;
+	bma->firstblock = *firstblock;
+	bma->minlen = (flags & XFS_BMAPI_CONTIG) ? alen : 1;
+	bma->low = flist->xbf_low;
+	bma->aeof = 0;
+
+	/*
+	 * Only want to do the alignment at the eof if it is userdata and
+	 * allocation length is larger than a stripe unit.
+	 */
+	if (mp->m_dalign && alen >= mp->m_dalign &&
+	    !(flags & XFS_BMAPI_METADATA) && whichfork == XFS_DATA_FORK) {
+		error = xfs_bmap_isaeof(bma->ip, aoff, whichfork, &bma->aeof);
+		if (error)
+			return error;
+	}
+
+	error = xfs_bmap_alloc(bma);
+	if (error)
+		return error;
+
+	/*
+	 * Copy out result fields.
+	 */
+	abno = bma->rval;
+	flist->xbf_low = bma->low;
+	alen = bma->alen;
+	aoff = bma->off;
+	ASSERT(*firstblock == NULLFSBLOCK ||
+	       XFS_FSB_TO_AGNO(mp, *firstblock) ==
+	       XFS_FSB_TO_AGNO(mp, bma->firstblock) ||
+	       (flist->xbf_low &&
+		XFS_FSB_TO_AGNO(mp, *firstblock) <
+			XFS_FSB_TO_AGNO(mp, bma->firstblock)));
+	*firstblock = bma->firstblock;
+	if (*cur)
+		(*cur)->bc_private.b.firstblock = *firstblock;
+	if (abno == NULLFSBLOCK)
+		return 0;
+	if ((ifp->if_flags & XFS_IFBROOT) && !*cur) {
+		(*cur) = xfs_bmbt_init_cursor(mp, bma->tp, bma->ip, whichfork);
+		(*cur)->bc_private.b.firstblock = *firstblock;
+		(*cur)->bc_private.b.flist = flist;
+	}
+	/*
+	 * Bump the number of extents we've allocated
+	 * in this call.
+	 */
+	(*nallocs)++;
+
+	if (*cur)
+		(*cur)->bc_private.b.flags =
+			bma->wasdel ? XFS_BTCUR_BPRV_WASDEL : 0;
+
+	bma->gotp->br_startoff = aoff;
+	bma->gotp->br_startblock = abno;
+	bma->gotp->br_blockcount = alen;
+	bma->gotp->br_state = XFS_EXT_NORM;
+
+	/*
+	 * A wasdelay extent has been initialized, so shouldn't be flagged
+	 * as unwritten.
+	 */
+	if (!bma->wasdel && (flags & XFS_BMAPI_PREALLOC) &&
+	    xfs_sb_version_hasextflgbit(&mp->m_sb))
+		bma->gotp->br_state = XFS_EXT_UNWRITTEN;
+
+	error = xfs_bmap_add_extent(bma->tp, bma->ip, lastx, cur, bma->gotp,
+				    firstblock, flist, logflags, whichfork);
+	if (error)
+		return error;
+
+	/*
+	 * Update our extent pointer, given that xfs_bmap_add_extent  might
+	 * have merged it into one of the neighbouring ones.
+	 */
+	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx), bma->gotp);
+
+	ASSERT(bma->gotp->br_startoff <= aoff);
+	ASSERT(bma->gotp->br_startoff + bma->gotp->br_blockcount >=
+		aoff + alen);
+	ASSERT(bma->gotp->br_state == XFS_EXT_NORM ||
+	       bma->gotp->br_state == XFS_EXT_UNWRITTEN);
+	return 0;
+}
+
 /*
  * Map file blocks to filesystem blocks.
  * File range is given by the bno/len pair.
@@ -4629,9 +4770,6 @@ xfs_bmapi(
 	int		*nmap,		/* i/o: mval size/count */
 	xfs_bmap_free_t	*flist)		/* i/o: list extents to free */
 {
-	xfs_fsblock_t	abno;		/* allocated block number */
-	xfs_extlen_t	alen;		/* allocated extent length */
-	xfs_fileoff_t	aoff;		/* allocated file offset */
 	xfs_bmalloca_t	bma = { 0 };	/* args for xfs_bmap_alloc */
 	xfs_btree_cur_t	*cur;		/* bmap btree cursor */
 	xfs_fileoff_t	end;		/* end of mapped file region */
@@ -4643,7 +4781,6 @@ xfs_bmapi(
 	xfs_extnum_t	lastx;		/* last useful extent number */
 	int		logflags;	/* flags for transaction logging */
 	xfs_extlen_t	minleft;	/* min blocks left after allocation */
-	xfs_extlen_t	minlen;		/* min allocation size */
 	xfs_mount_t	*mp;		/* xfs mount structure */
 	int		n;		/* current extent index */
 	int		nallocs;	/* number of extents alloc'd */
@@ -4734,7 +4871,13 @@ xfs_bmapi(
 	n = 0;
 	end = bno + len;
 	obno = bno;
-	bma.ip = NULL;
+
+	bma.tp = tp;
+	bma.ip = ip;
+	bma.prevp = &prev;
+	bma.gotp = &got;
+	bma.total = total;
+	bma.userdata = 0;
 
 	while (bno < end && n < *nmap) {
 		/*
@@ -4750,144 +4893,31 @@ xfs_bmapi(
 		 * that we found, if any.
 		 */
 		if (wr && (inhole || wasdelay)) {
-			/*
-			 * For the wasdelay case, we could also just
-			 * allocate the stuff asked for in this bmap call
-			 * but that wouldn't be as good.
-			 */
-			if (wasdelay) {
-				alen = (xfs_extlen_t)got.br_blockcount;
-				aoff = got.br_startoff;
-				if (lastx != NULLEXTNUM && lastx) {
-					ep = xfs_iext_get_ext(ifp, lastx - 1);
-					xfs_bmbt_get_all(ep, &prev);
-				}
-			} else {
-				alen = (xfs_extlen_t)
-					XFS_FILBLKS_MIN(len, MAXEXTLEN);
-				if (!eof)
-					alen = (xfs_extlen_t)
-						XFS_FILBLKS_MIN(alen,
-							got.br_startoff - bno);
-				aoff = bno;
-			}
-			minlen = (flags & XFS_BMAPI_CONTIG) ? alen : 1;
-			{
-				/*
-				 * If first time, allocate and fill in
-				 * once-only bma fields.
-				 */
-				if (bma.ip == NULL) {
-					bma.tp = tp;
-					bma.ip = ip;
-					bma.prevp = &prev;
-					bma.gotp = &got;
-					bma.total = total;
-					bma.userdata = 0;
-				}
-				/* Indicate if this is the first user data
-				 * in the file, or just any user data.
-				 */
-				if (!(flags & XFS_BMAPI_METADATA)) {
-					bma.userdata = (aoff == 0) ?
-						XFS_ALLOC_INITIAL_USER_DATA :
-						XFS_ALLOC_USERDATA;
-				}
-				/*
-				 * Fill in changeable bma fields.
-				 */
-				bma.eof = eof;
-				bma.firstblock = *firstblock;
-				bma.alen = alen;
-				bma.off = aoff;
-				bma.conv = !!(flags & XFS_BMAPI_CONVERT);
-				bma.wasdel = wasdelay;
-				bma.minlen = minlen;
-				bma.low = flist->xbf_low;
-				bma.minleft = minleft;
-				/*
-				 * Only want to do the alignment at the
-				 * eof if it is userdata and allocation length
-				 * is larger than a stripe unit.
-				 */
-				if (mp->m_dalign && alen >= mp->m_dalign &&
-				    (!(flags & XFS_BMAPI_METADATA)) &&
-				    (whichfork == XFS_DATA_FORK)) {
-					if ((error = xfs_bmap_isaeof(ip, aoff,
-							whichfork, &bma.aeof)))
-						goto error0;
-				} else
-					bma.aeof = 0;
-				/*
-				 * Call allocator.
-				 */
-				if ((error = xfs_bmap_alloc(&bma)))
-					goto error0;
-				/*
-				 * Copy out result fields.
-				 */
-				abno = bma.rval;
-				if ((flist->xbf_low = bma.low))
-					minleft = 0;
-				alen = bma.alen;
-				aoff = bma.off;
-				ASSERT(*firstblock == NULLFSBLOCK ||
-				       XFS_FSB_TO_AGNO(mp, *firstblock) ==
-				       XFS_FSB_TO_AGNO(mp, bma.firstblock) ||
-				       (flist->xbf_low &&
-					XFS_FSB_TO_AGNO(mp, *firstblock) <
-					XFS_FSB_TO_AGNO(mp, bma.firstblock)));
-				*firstblock = bma.firstblock;
-				if (cur)
-					cur->bc_private.b.firstblock =
-						*firstblock;
-				if (abno == NULLFSBLOCK)
-					break;
-				if ((ifp->if_flags & XFS_IFBROOT) && !cur) {
-					cur = xfs_bmbt_init_cursor(mp, tp,
-						ip, whichfork);
-					cur->bc_private.b.firstblock =
-						*firstblock;
-					cur->bc_private.b.flist = flist;
+			bma.eof = eof;
+			bma.conv = !!(flags & XFS_BMAPI_CONVERT);
+			bma.wasdel = wasdelay;
+			bma.alen = len;
+			bma.off = bno;
+			bma.minleft = minleft;
+
+			error = xfs_bmapi_allocate(&bma, &lastx, &cur,
+					firstblock, flist, flags, &nallocs,
+					&tmp_logflags);
+			if (error == ENOSPC || error == EDQUOT) {
+				if (n == 0) {
+					*nmap = 0;
+					ASSERT(cur == NULL);
+					return error;
 				}
-				/*
-				 * Bump the number of extents we've allocated
-				 * in this call.
-				 */
-				nallocs++;
-			}
-			if (cur)
-				cur->bc_private.b.flags =
-					wasdelay ? XFS_BTCUR_BPRV_WASDEL : 0;
-			got.br_startoff = aoff;
-			got.br_startblock = abno;
-			got.br_blockcount = alen;
-			got.br_state = XFS_EXT_NORM;	/* assume normal */
-			/*
-			 * Determine state of extent, and the filesystem.
-			 * A wasdelay extent has been initialized, so
-			 * shouldn't be flagged as unwritten.
-			 */
-			if (wr && xfs_sb_version_hasextflgbit(&mp->m_sb)) {
-				if (!wasdelay && (flags & XFS_BMAPI_PREALLOC))
-					got.br_state = XFS_EXT_UNWRITTEN;
 			}
-			error = xfs_bmap_add_extent(tp, ip, &lastx, &cur, &got,
-				firstblock, flist, &tmp_logflags,
-				whichfork);
+
 			logflags |= tmp_logflags;
 			if (error)
 				goto error0;
-			ep = xfs_iext_get_ext(ifp, lastx);
-			xfs_bmbt_get_all(ep, &got);
-			ASSERT(got.br_startoff <= aoff);
-			ASSERT(got.br_startoff + got.br_blockcount >=
-				aoff + alen);
-			ASSERT(got.br_state == XFS_EXT_NORM ||
-			       got.br_state == XFS_EXT_UNWRITTEN);
-			/*
-			 * Fall down into the found allocated space case.
-			 */
+			if (flist && flist->xbf_low)
+				minleft = 0;
+			if (bma.rval == NULLFSBLOCK)
+				break;
 		} else if (inhole) {
 			/*
 			 * Reading in a hole.

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

[PATCH 11/25] xfs: factor unwritten extent map manipulations out of xfs_bmapi

[Christoph Hellwig]

[-- Attachment #1: xfs-bmapi-factor-unwrittent-extent-conversion --]
[-- Type: text/plain, Size: 4465 bytes --]

To further improve the readability of xfs_bmapi(), factor the unwritten
extent conversion out into a separate function. This removes
large block of logic from the xfs_bmapi() code loop and makes it
easier to see the operational logic flow for xfs_bmapi().

Signed-off-by: Dave Chinner <dchinner@redhat.com>

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:13:33.855748845 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:13:35.465740123 +0200
@@ -4089,7 +4089,6 @@ xfs_bmap_read_extents(
 		xfs_extnum_t	num_recs;
 		xfs_extnum_t	start;
 
-
 		num_recs = xfs_btree_get_numrecs(block);
 		if (unlikely(i + num_recs > room)) {
 			ASSERT(i + num_recs <= room);
@@ -4743,6 +4742,69 @@ xfs_bmapi_allocate(
 	return 0;
 }
 
+STATIC int
+xfs_bmapi_convert_unwritten(
+	struct xfs_bmalloca	*bma,
+	struct xfs_bmbt_irec	*mval,
+	xfs_filblks_t		len,
+	xfs_extnum_t		*lastx,
+	struct xfs_btree_cur	**cur,
+	xfs_fsblock_t		*firstblock,
+	struct xfs_bmap_free	*flist,
+	int			flags,
+	int			*logflags)
+{
+	int			whichfork = (flags & XFS_BMAPI_ATTRFORK) ?
+						XFS_ATTR_FORK : XFS_DATA_FORK;
+	struct xfs_ifork	*ifp = XFS_IFORK_PTR(bma->ip, whichfork);
+	int			error;
+
+	*logflags = 0;
+
+	/* check if we need to do unwritten->real conversion */
+	if (mval->br_state == XFS_EXT_UNWRITTEN &&
+	    (flags & XFS_BMAPI_PREALLOC))
+		return 0;
+
+	/* check if we need to do real->unwritten conversion */
+	if (mval->br_state == XFS_EXT_NORM &&
+	    (flags & (XFS_BMAPI_PREALLOC | XFS_BMAPI_CONVERT)) !=
+			(XFS_BMAPI_PREALLOC | XFS_BMAPI_CONVERT))
+		return 0;
+
+	/*
+	 * Modify (by adding) the state flag, if writing.
+	 */
+	ASSERT(mval->br_blockcount <= len);
+	if ((ifp->if_flags & XFS_IFBROOT) && !*cur) {
+		*cur = xfs_bmbt_init_cursor(bma->ip->i_mount, bma->tp,
+					bma->ip, whichfork);
+		(*cur)->bc_private.b.firstblock = *firstblock;
+		(*cur)->bc_private.b.flist = flist;
+	}
+	mval->br_state = (mval->br_state == XFS_EXT_UNWRITTEN)
+				? XFS_EXT_NORM : XFS_EXT_UNWRITTEN;
+
+	error = xfs_bmap_add_extent(bma->tp, bma->ip, lastx, cur, mval,
+				firstblock, flist, logflags, whichfork);
+	if (error)
+		return error;
+
+	/*
+	 * Update our extent pointer, given that xfs_bmap_add_extent  might
+	 * have merged it into one of the neighbouring ones.
+	 */
+	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx), bma->gotp);
+
+	/*
+	 * We may have combined previously unwritten space with written space,
+	 * so generate another request.
+	 */
+	if (mval->br_blockcount < len)
+		return EAGAIN;
+	return 0;
+}
+
 /*
  * Map file blocks to filesystem blocks.
  * File range is given by the bno/len pair.
@@ -4937,45 +4999,16 @@ xfs_bmapi(
 		/* Deal with the allocated space we found.  */
 		xfs_bmapi_trim_map(mval, &got, &bno, len, obno, end, n, flags);
 
-		/*
-		 * Check if writing previously allocated but
-		 * unwritten extents.
-		 */
-		if (wr &&
-		    ((mval->br_state == XFS_EXT_UNWRITTEN &&
-		      ((flags & XFS_BMAPI_PREALLOC) == 0)) ||
-		     (mval->br_state == XFS_EXT_NORM &&
-		      ((flags & (XFS_BMAPI_PREALLOC|XFS_BMAPI_CONVERT)) ==
-				(XFS_BMAPI_PREALLOC|XFS_BMAPI_CONVERT))))) {
-			/*
-			 * Modify (by adding) the state flag, if writing.
-			 */
-			ASSERT(mval->br_blockcount <= len);
-			if ((ifp->if_flags & XFS_IFBROOT) && !cur) {
-				cur = xfs_bmbt_init_cursor(mp,
-					tp, ip, whichfork);
-				cur->bc_private.b.firstblock =
-					*firstblock;
-				cur->bc_private.b.flist = flist;
-			}
-			mval->br_state = (mval->br_state == XFS_EXT_UNWRITTEN)
-						? XFS_EXT_NORM
-						: XFS_EXT_UNWRITTEN;
-			error = xfs_bmap_add_extent(tp, ip, &lastx, &cur, mval,
-				firstblock, flist, &tmp_logflags,
-				whichfork);
+		/* Execute unwritten extent conversion if necessary */
+		if (wr) {
+			error = xfs_bmapi_convert_unwritten(&bma, mval, len,
+						&lastx, &cur, firstblock, flist, flags,
+						&tmp_logflags);
 			logflags |= tmp_logflags;
+			if (error == EAGAIN)
+				continue;
 			if (error)
 				goto error0;
-			ep = xfs_iext_get_ext(ifp, lastx);
-			xfs_bmbt_get_all(ep, &got);
-			/*
-			 * We may have combined previously unwritten
-			 * space with written space, so generate
-			 * another request.
-			 */
-			if (mval->br_blockcount < len)
-				continue;
 		}
 
 		/* update the extent map to return */

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

[PATCH 12/25] xfs: rename xfs_bmapi to xfs_bmapi_write

[Christoph Hellwig]

[-- Attachment #1: xfs-bmapi-rename-xfs_bmapi --]
[-- Type: text/plain, Size: 22698 bytes --]

Once this is done, rename xfs_bmapi to xfs_bmapi_write to reflect the fact it
is for allocation only. This enables us to kill the XFS_BMAPI_WRITE flag as
well.

Also clean up xfs_bmapi_write to the style used in the newly added
xfs_bmapi_read/delay functions.

Signed-off-by: Dave Chinner <dchinner@redhat.com>

Index: xfs/fs/xfs/xfs_dquot.c
===================================================================
--- xfs.orig/fs/xfs/xfs_dquot.c	2011-08-23 21:09:32.900387547 +0200
+++ xfs/fs/xfs/xfs_dquot.c	2011-08-23 21:13:59.315610918 +0200
@@ -379,14 +379,12 @@ xfs_qm_dqalloc(
 
 	xfs_trans_ijoin_ref(tp, quotip, XFS_ILOCK_EXCL);
 	nmaps = 1;
-	if ((error = xfs_bmapi(tp, quotip,
-			      offset_fsb, XFS_DQUOT_CLUSTER_SIZE_FSB,
-			      XFS_BMAPI_METADATA | XFS_BMAPI_WRITE,
-			      &firstblock,
-			      XFS_QM_DQALLOC_SPACE_RES(mp),
-			      &map, &nmaps, &flist))) {
+	error = xfs_bmapi_write(tp, quotip, offset_fsb,
+				XFS_DQUOT_CLUSTER_SIZE_FSB, XFS_BMAPI_METADATA,
+				&firstblock, XFS_QM_DQALLOC_SPACE_RES(mp),
+				&map, &nmaps, &flist);
+	if (error)
 		goto error0;
-	}
 	ASSERT(map.br_blockcount == XFS_DQUOT_CLUSTER_SIZE_FSB);
 	ASSERT(nmaps == 1);
 	ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
Index: xfs/fs/xfs/xfs_attr.c
===================================================================
--- xfs.orig/fs/xfs/xfs_attr.c	2011-08-23 21:09:32.000000000 +0200
+++ xfs/fs/xfs/xfs_attr.c	2011-08-23 21:13:59.315610918 +0200
@@ -2039,10 +2039,9 @@ xfs_attr_rmtval_set(xfs_da_args_t *args)
 		 */
 		xfs_bmap_init(args->flist, args->firstblock);
 		nmap = 1;
-		error = xfs_bmapi(args->trans, dp, (xfs_fileoff_t)lblkno,
+		error = xfs_bmapi_write(args->trans, dp, (xfs_fileoff_t)lblkno,
 				  blkcnt,
-				  XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA |
-							XFS_BMAPI_WRITE,
+				  XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
 				  args->firstblock, args->total, &map, &nmap,
 				  args->flist);
 		if (!error) {
Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:13:35.000000000 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:13:59.318944234 +0200
@@ -4211,9 +4211,8 @@ xfs_bmap_validate_ret(
 		ASSERT(i == 0 ||
 		       mval[i - 1].br_startoff + mval[i - 1].br_blockcount ==
 		       mval[i].br_startoff);
-		if (flags & XFS_BMAPI_WRITE)
-			ASSERT(mval[i].br_startblock != DELAYSTARTBLOCK &&
-			       mval[i].br_startblock != HOLESTARTBLOCK);
+		ASSERT(mval[i].br_startblock != DELAYSTARTBLOCK &&
+		       mval[i].br_startblock != HOLESTARTBLOCK);
 		ASSERT(mval[i].br_state == XFS_EXT_NORM ||
 		       mval[i].br_state == XFS_EXT_UNWRITTEN);
 	}
@@ -4806,60 +4805,57 @@ xfs_bmapi_convert_unwritten(
 }
 
 /*
- * Map file blocks to filesystem blocks.
- * File range is given by the bno/len pair.
- * Adds blocks to file if a write ("flags & XFS_BMAPI_WRITE" set)
- * into a hole or past eof.
- * Only allocates blocks from a single allocation group,
- * to avoid locking problems.
+ * Map file blocks to filesystem blocks, and allocate blocks or convert the
+ * extent state if nessecary.  Details behaviour is controlled by the flags
+ * paramter.  Only allocates blocks from a single allocation group, to avoid
+ * locking problems.
+ *
  * The returned value in "firstblock" from the first call in a transaction
  * must be remembered and presented to subsequent calls in "firstblock".
  * An upper bound for the number of blocks to be allocated is supplied to
  * the first call in "total"; if no allocation group has that many free
  * blocks then the call will fail (return NULLFSBLOCK in "firstblock").
  */
-int					/* error */
-xfs_bmapi(
-	xfs_trans_t	*tp,		/* transaction pointer */
-	xfs_inode_t	*ip,		/* incore inode */
-	xfs_fileoff_t	bno,		/* starting file offs. mapped */
-	xfs_filblks_t	len,		/* length to map in file */
-	int		flags,		/* XFS_BMAPI_... */
-	xfs_fsblock_t	*firstblock,	/* first allocated block
-					   controls a.g. for allocs */
-	xfs_extlen_t	total,		/* total blocks needed */
-	xfs_bmbt_irec_t	*mval,		/* output: map values */
-	int		*nmap,		/* i/o: mval size/count */
-	xfs_bmap_free_t	*flist)		/* i/o: list extents to free */
+int
+xfs_bmapi_write(
+	struct xfs_trans	*tp,		/* transaction pointer */
+	struct xfs_inode	*ip,		/* incore inode */
+	xfs_fileoff_t		bno,		/* starting file offs. mapped */
+	xfs_filblks_t		len,		/* length to map in file */
+	int			flags,		/* XFS_BMAPI_... */
+	xfs_fsblock_t		*firstblock,	/* first allocated block
+						   controls a.g. for allocs */
+	xfs_extlen_t		total,		/* total blocks needed */
+	struct xfs_bmbt_irec	*mval,		/* output: map values */
+	int			*nmap,		/* i/o: mval size/count */
+	struct xfs_bmap_free	*flist)		/* i/o: list extents to free */
 {
-	xfs_bmalloca_t	bma = { 0 };	/* args for xfs_bmap_alloc */
-	xfs_btree_cur_t	*cur;		/* bmap btree cursor */
-	xfs_fileoff_t	end;		/* end of mapped file region */
-	int		eof;		/* we've hit the end of extents */
-	xfs_bmbt_rec_host_t *ep;	/* extent record pointer */
-	int		error;		/* error return */
-	xfs_bmbt_irec_t	got;		/* current file extent record */
-	xfs_ifork_t	*ifp;		/* inode fork pointer */
-	xfs_extnum_t	lastx;		/* last useful extent number */
-	int		logflags;	/* flags for transaction logging */
-	xfs_extlen_t	minleft;	/* min blocks left after allocation */
-	xfs_mount_t	*mp;		/* xfs mount structure */
-	int		n;		/* current extent index */
-	int		nallocs;	/* number of extents alloc'd */
-	xfs_fileoff_t	obno;		/* old block number (offset) */
-	xfs_bmbt_irec_t	prev;		/* previous file extent record */
-	int		tmp_logflags;	/* temp flags holder */
-	int		whichfork;	/* data or attr fork */
-	char		inhole;		/* current location is hole in file */
-	char		wasdelay;	/* old extent was delayed */
-	char		wr;		/* this is a write request */
-	char		rt;		/* this is a realtime file */
+	struct xfs_mount	*mp = ip->i_mount;
+	struct xfs_bmalloca	bma = { 0 };	/* args for xfs_bmap_alloc */
+	struct xfs_btree_cur	*cur;		/* bmap btree cursor */
+	xfs_fileoff_t		end;		/* end of mapped file region */
+	int			eof;		/* after the end of extents */
+	int			error;		/* error return */
+	struct xfs_bmbt_irec	got;		/* current file extent record */
+	xfs_ifork_t		*ifp;		/* inode fork pointer */
+	xfs_extnum_t		lastx;		/* last useful extent number */
+	int			logflags;	/* flags for transaction logging */
+	xfs_extlen_t		minleft;	/* min blocks left after allocation */
+	int			n;		/* current extent index */
+	int			nallocs;	/* number of extents alloc'd */
+	xfs_fileoff_t		obno;		/* old block number (offset) */
+	struct xfs_bmbt_irec	prev;		/* previous file extent record */
+	int			tmp_logflags;	/* temp flags holder */
+	int			whichfork;	/* data or attr fork */
+	char			inhole;		/* current location is hole in file */
+	char			wasdelay;	/* old extent was delayed */
+
 #ifdef DEBUG
-	xfs_fileoff_t	orig_bno;	/* original block number value */
-	int		orig_flags;	/* original flags arg value */
-	xfs_filblks_t	orig_len;	/* original value of len arg */
-	xfs_bmbt_irec_t	*orig_mval;	/* original value of mval */
-	int		orig_nmap;	/* original value of *nmap */
+	xfs_fileoff_t		orig_bno;	/* original block number value */
+	int			orig_flags;	/* original flags arg value */
+	xfs_filblks_t		orig_len;	/* original value of len arg */
+	struct xfs_bmbt_irec	*orig_mval;	/* original value of mval */
+	int			orig_nmap;	/* original value of *nmap */
 
 	orig_bno = bno;
 	orig_len = len;
@@ -4867,69 +4863,60 @@ xfs_bmapi(
 	orig_mval = mval;
 	orig_nmap = *nmap;
 #endif
+
 	ASSERT(*nmap >= 1);
-	ASSERT(*nmap <= XFS_BMAP_MAX_NMAP || !(flags & XFS_BMAPI_WRITE));
+	ASSERT(*nmap <= XFS_BMAP_MAX_NMAP);
+	ASSERT(!(flags & XFS_BMAPI_IGSTATE));
+	ASSERT(tp != NULL);
+
 	whichfork = (flags & XFS_BMAPI_ATTRFORK) ?
 		XFS_ATTR_FORK : XFS_DATA_FORK;
-	mp = ip->i_mount;
+
 	if (unlikely(XFS_TEST_ERROR(
 	    (XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS &&
 	     XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE &&
 	     XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_LOCAL),
 	     mp, XFS_ERRTAG_BMAPIFORMAT, XFS_RANDOM_BMAPIFORMAT))) {
-		XFS_ERROR_REPORT("xfs_bmapi", XFS_ERRLEVEL_LOW, mp);
+		XFS_ERROR_REPORT("xfs_bmapi_write", XFS_ERRLEVEL_LOW, mp);
 		return XFS_ERROR(EFSCORRUPTED);
 	}
+
 	if (XFS_FORCED_SHUTDOWN(mp))
 		return XFS_ERROR(EIO);
-	rt = (whichfork == XFS_DATA_FORK) && XFS_IS_REALTIME_INODE(ip);
+
 	ifp = XFS_IFORK_PTR(ip, whichfork);
 	ASSERT(ifp->if_ext_max ==
 	       XFS_IFORK_SIZE(ip, whichfork) / (uint)sizeof(xfs_bmbt_rec_t));
-	if ((wr = (flags & XFS_BMAPI_WRITE)) != 0)
-		XFS_STATS_INC(xs_blk_mapw);
-	else
-		XFS_STATS_INC(xs_blk_mapr);
-	/*
-	 * IGSTATE flag is used to combine extents which
-	 * differ only due to the state of the extents.
-	 * This technique is used from xfs_getbmap()
-	 * when the caller does not wish to see the
-	 * separation (which is the default).
-	 *
-	 * This technique is also used when writing a
-	 * buffer which has been partially written,
-	 * (usually by being flushed during a chunkread),
-	 * to ensure one write takes place. This also
-	 * prevents a change in the xfs inode extents at
-	 * this time, intentionally. This change occurs
-	 * on completion of the write operation, in
-	 * xfs_strat_comp(), where the xfs_bmapi() call
-	 * is transactioned, and the extents combined.
-	 */
-	if ((flags & XFS_BMAPI_IGSTATE) && wr)	/* if writing unwritten space */
-		wr = 0;				/* no allocations are allowed */
+
+	XFS_STATS_INC(xs_blk_mapw);
+
 	logflags = 0;
 	nallocs = 0;
 	cur = NULL;
+
 	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_LOCAL) {
-		ASSERT(wr && tp);
-		if ((error = xfs_bmap_local_to_extents(tp, ip,
-				firstblock, total, &logflags, whichfork)))
+		error = xfs_bmap_local_to_extents(tp, ip, firstblock, total,
+						  &logflags, whichfork);
+		if (error)
 			goto error0;
 	}
-	if (wr && *firstblock == NULLFSBLOCK) {
+
+	if (*firstblock == NULLFSBLOCK) {
 		if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_BTREE)
 			minleft = be16_to_cpu(ifp->if_broot->bb_level) + 1;
 		else
 			minleft = 1;
-	} else
+	} else {
 		minleft = 0;
-	if (!(ifp->if_flags & XFS_IFEXTENTS) &&
-	    (error = xfs_iread_extents(tp, ip, whichfork)))
-		goto error0;
-	ep = xfs_bmap_search_extents(ip, bno, whichfork, &eof, &lastx, &got,
-		&prev);
+	}
+
+	if (!(ifp->if_flags & XFS_IFEXTENTS)) {
+		error = xfs_iread_extents(tp, ip, whichfork);
+		if (error)
+			goto error0;
+	}
+
+	xfs_bmap_search_extents(ip, bno, whichfork, &eof, &lastx, &got, &prev);
 	n = 0;
 	end = bno + len;
 	obno = bno;
@@ -4942,19 +4929,14 @@ xfs_bmapi(
 	bma.userdata = 0;
 
 	while (bno < end && n < *nmap) {
-		/*
-		 * Reading past eof, act as though there's a hole
-		 * up to end.
-		 */
-		if (eof && !wr)
-			got.br_startoff = end;
 		inhole = eof || got.br_startoff > bno;
-		wasdelay = wr && !inhole && isnullstartblock(got.br_startblock);
+		wasdelay = !inhole && isnullstartblock(got.br_startblock);
+
 		/*
 		 * First, deal with the hole before the allocated space
 		 * that we found, if any.
 		 */
-		if (wr && (inhole || wasdelay)) {
+		if (inhole || wasdelay) {
 			bma.eof = eof;
 			bma.conv = !!(flags & XFS_BMAPI_CONVERT);
 			bma.wasdel = wasdelay;
@@ -4980,36 +4962,20 @@ xfs_bmapi(
 				minleft = 0;
 			if (bma.rval == NULLFSBLOCK)
 				break;
-		} else if (inhole) {
-			/*
-			 * Reading in a hole.
-			 */
-			mval->br_startoff = bno;
-			mval->br_startblock = HOLESTARTBLOCK;
-			mval->br_blockcount =
-				XFS_FILBLKS_MIN(len, got.br_startoff - bno);
-			mval->br_state = XFS_EXT_NORM;
-			bno += mval->br_blockcount;
-			len -= mval->br_blockcount;
-			mval++;
-			n++;
-			continue;
 		}
 
 		/* Deal with the allocated space we found.  */
 		xfs_bmapi_trim_map(mval, &got, &bno, len, obno, end, n, flags);
 
 		/* Execute unwritten extent conversion if necessary */
-		if (wr) {
-			error = xfs_bmapi_convert_unwritten(&bma, mval, len,
-						&lastx, &cur, firstblock, flist, flags,
-						&tmp_logflags);
-			logflags |= tmp_logflags;
-			if (error == EAGAIN)
-				continue;
-			if (error)
-				goto error0;
-		}
+		error = xfs_bmapi_convert_unwritten(&bma, mval, len, &lastx,
+						    &cur, firstblock, flist,
+						    flags, &tmp_logflags);
+		logflags |= tmp_logflags;
+		if (error == EAGAIN)
+			continue;
+		if (error)
+			goto error0;
 
 		/* update the extent map to return */
 		xfs_bmapi_update_map(&mval, &bno, &len, obno, end, &n, flags);
@@ -5021,24 +4987,22 @@ xfs_bmapi(
 		 */
 		if (bno >= end || n >= *nmap || nallocs >= *nmap)
 			break;
-		/*
-		 * Else go on to the next record.
-		 */
+
+		/* Else go on to the next record. */
 		prev = got;
-		if (++lastx < ifp->if_bytes / sizeof(xfs_bmbt_rec_t)) {
-			ep = xfs_iext_get_ext(ifp, lastx);
-			xfs_bmbt_get_all(ep, &got);
-		} else {
+		if (++lastx < ifp->if_bytes / sizeof(xfs_bmbt_rec_t))
+			xfs_bmbt_get_all(xfs_iext_get_ext(ifp, lastx), &got);
+		else
 			eof = 1;
-		}
 	}
 	*nmap = n;
+
 	/*
 	 * Transform from btree to extents, give it cur.
 	 */
-	if (tp && XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_BTREE &&
+	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_BTREE &&
 	    XFS_IFORK_NEXTENTS(ip, whichfork) <= ifp->if_ext_max) {
-		ASSERT(wr && cur);
+		ASSERT(cur);
 		error = xfs_bmap_btree_to_extents(tp, ip, cur,
 			&tmp_logflags, whichfork);
 		logflags |= tmp_logflags;
@@ -5066,10 +5030,9 @@ error0:
 	 * detecting a case where the data is changed, there's an error,
 	 * and it's not logged so we don't shutdown when we should.
 	 */
-	if (logflags) {
-		ASSERT(tp && wr);
+	if (logflags)
 		xfs_trans_log_inode(tp, ip, logflags);
-	}
+
 	if (cur) {
 		if (!error) {
 			ASSERT(*firstblock == NULLFSBLOCK ||
Index: xfs/fs/xfs/xfs_bmap.h
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.h	2011-08-23 21:12:59.000000000 +0200
+++ xfs/fs/xfs/xfs_bmap.h	2011-08-23 21:13:59.318944234 +0200
@@ -62,25 +62,23 @@ typedef	struct xfs_bmap_free
 #define	XFS_BMAP_MAX_NMAP	4
 
 /*
- * Flags for xfs_bmapi
+ * Flags for xfs_bmapi_*
  */
-#define	XFS_BMAPI_WRITE		0x001	/* write operation: allocate space */
-#define XFS_BMAPI_ENTIRE	0x004	/* return entire extent, not trimmed */
-#define XFS_BMAPI_METADATA	0x008	/* mapping metadata not user data */
-#define XFS_BMAPI_ATTRFORK	0x010	/* use attribute fork not data */
-#define	XFS_BMAPI_PREALLOC	0x040	/* preallocation op: unwritten space */
-#define	XFS_BMAPI_IGSTATE	0x080	/* Ignore state - */
+#define XFS_BMAPI_ENTIRE	0x001	/* return entire extent, not trimmed */
+#define XFS_BMAPI_METADATA	0x002	/* mapping metadata not user data */
+#define XFS_BMAPI_ATTRFORK	0x004	/* use attribute fork not data */
+#define XFS_BMAPI_PREALLOC	0x008	/* preallocation op: unwritten space */
+#define XFS_BMAPI_IGSTATE	0x010	/* Ignore state - */
 					/* combine contig. space */
-#define	XFS_BMAPI_CONTIG	0x100	/* must allocate only one extent */
+#define XFS_BMAPI_CONTIG	0x020	/* must allocate only one extent */
 /*
  * unwritten extent conversion - this needs write cache flushing and no additional
  * allocation alignments. When specified with XFS_BMAPI_PREALLOC it converts
  * from written to unwritten, otherwise convert from unwritten to written.
  */
-#define XFS_BMAPI_CONVERT	0x200
+#define XFS_BMAPI_CONVERT	0x040
 
 #define XFS_BMAPI_FLAGS \
-	{ XFS_BMAPI_WRITE,	"WRITE" }, \
 	{ XFS_BMAPI_ENTIRE,	"ENTIRE" }, \
 	{ XFS_BMAPI_METADATA,	"METADATA" }, \
 	{ XFS_BMAPI_ATTRFORK,	"ATTRFORK" }, \
@@ -265,39 +263,17 @@ xfs_bmap_read_extents(
 	struct xfs_inode	*ip,		/* incore inode */
 	int			whichfork);	/* data or attr fork */
 
-/*
- * Map file blocks to filesystem blocks.
- * File range is given by the bno/len pair.
- * Adds blocks to file if a write ("flags & XFS_BMAPI_WRITE" set)
- * into a hole or past eof.
- * Only allocates blocks from a single allocation group,
- * to avoid locking problems.
- * The returned value in "firstblock" from the first call in a transaction
- * must be remembered and presented to subsequent calls in "firstblock".
- * An upper bound for the number of blocks to be allocated is supplied to
- * the first call in "total"; if no allocation group has that many free
- * blocks then the call will fail (return NULLFSBLOCK in "firstblock").
- */
-int						/* error */
-xfs_bmapi(
-	struct xfs_trans	*tp,		/* transaction pointer */
-	struct xfs_inode	*ip,		/* incore inode */
-	xfs_fileoff_t		bno,		/* starting file offs. mapped */
-	xfs_filblks_t		len,		/* length to map in file */
-	int			flags,		/* XFS_BMAPI_... */
-	xfs_fsblock_t		*firstblock,	/* first allocated block
-						   controls a.g. for allocs */
-	xfs_extlen_t		total,		/* total blocks needed */
-	struct xfs_bmbt_irec	*mval,		/* output: map values */
-	int			*nmap,		/* i/o: mval size/count */
-	xfs_bmap_free_t		*flist);	/* i/o: list extents to free */
-
 int	xfs_bmapi_read(struct xfs_inode *ip, xfs_fileoff_t bno,
 		xfs_filblks_t len, struct xfs_bmbt_irec *mval,
 		int *nmap, int flags);
 int	xfs_bmapi_delay(struct xfs_inode *ip, xfs_fileoff_t bno,
 		xfs_filblks_t len, struct xfs_bmbt_irec *mval,
 		int *nmap, int flags);
+int	xfs_bmapi_write(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_fileoff_t bno, xfs_filblks_t len, int flags,
+		xfs_fsblock_t *firstblock, xfs_extlen_t total,
+		struct xfs_bmbt_irec *mval, int *nmap,
+		struct xfs_bmap_free *flist);
 
 /*
  * Unmap (remove) blocks from a file.
Index: xfs/fs/xfs/xfs_da_btree.c
===================================================================
--- xfs.orig/fs/xfs/xfs_da_btree.c	2011-08-23 21:10:14.000000000 +0200
+++ xfs/fs/xfs/xfs_da_btree.c	2011-08-23 21:13:59.322277549 +0200
@@ -1578,9 +1578,8 @@ xfs_da_grow_inode_int(
 	 */
 	nmap = 1;
 	ASSERT(args->firstblock != NULL);
-	error = xfs_bmapi(tp, dp, *bno, count,
-			xfs_bmapi_aflag(w)|XFS_BMAPI_WRITE|XFS_BMAPI_METADATA|
-			XFS_BMAPI_CONTIG,
+	error = xfs_bmapi_write(tp, dp, *bno, count,
+			xfs_bmapi_aflag(w)|XFS_BMAPI_METADATA|XFS_BMAPI_CONTIG,
 			args->firstblock, args->total, &map, &nmap,
 			args->flist);
 	if (error)
@@ -1602,9 +1601,8 @@ xfs_da_grow_inode_int(
 		for (b = *bno, mapi = 0; b < *bno + count; ) {
 			nmap = MIN(XFS_BMAP_MAX_NMAP, count);
 			c = (int)(*bno + count - b);
-			error = xfs_bmapi(tp, dp, b, c,
-					xfs_bmapi_aflag(w)|XFS_BMAPI_WRITE|
-					XFS_BMAPI_METADATA,
+			error = xfs_bmapi_write(tp, dp, b, c,
+					xfs_bmapi_aflag(w)|XFS_BMAPI_METADATA,
 					args->firstblock, args->total,
 					&mapp[mapi], &nmap, args->flist);
 			if (error)
Index: xfs/fs/xfs/xfs_iomap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_iomap.c	2011-08-23 21:12:59.000000000 +0200
+++ xfs/fs/xfs/xfs_iomap.c	2011-08-23 21:13:59.325610864 +0200
@@ -210,20 +210,18 @@ xfs_iomap_write_direct(
 
 	xfs_trans_ijoin(tp, ip);
 
-	bmapi_flag = XFS_BMAPI_WRITE;
+	bmapi_flag = 0;
 	if (offset < ip->i_size || extsz)
 		bmapi_flag |= XFS_BMAPI_PREALLOC;
 
 	/*
-	 * Issue the xfs_bmapi() call to allocate the blocks.
-	 *
 	 * From this point onwards we overwrite the imap pointer that the
 	 * caller gave to us.
 	 */
 	xfs_bmap_init(&free_list, &firstfsb);
 	nimaps = 1;
-	error = xfs_bmapi(tp, ip, offset_fsb, count_fsb, bmapi_flag,
-		&firstfsb, 0, imap, &nimaps, &free_list);
+	error = xfs_bmapi_write(tp, ip, offset_fsb, count_fsb, bmapi_flag,
+				&firstfsb, 0, imap, &nimaps, &free_list);
 	if (error)
 		goto error0;
 
@@ -582,14 +580,12 @@ xfs_iomap_write_allocate(
 			}
 
 			/*
-			 * Go get the actual blocks.
-	 	 	 *
 			 * From this point onwards we overwrite the imap
 			 * pointer that the caller gave to us.
 			 */
-			error = xfs_bmapi(tp, ip, map_start_fsb, count_fsb,
-					XFS_BMAPI_WRITE, &first_block, 1,
-					imap, &nimaps, &free_list);
+			error = xfs_bmapi_write(tp, ip, map_start_fsb,
+						count_fsb, 0, &first_block, 1,
+						imap, &nimaps, &free_list);
 			if (error)
 				goto trans_cancel;
 
@@ -703,8 +699,8 @@ xfs_iomap_write_unwritten(
 		 */
 		xfs_bmap_init(&free_list, &firstfsb);
 		nimaps = 1;
-		error = xfs_bmapi(tp, ip, offset_fsb, count_fsb,
-				  XFS_BMAPI_WRITE|XFS_BMAPI_CONVERT, &firstfsb,
+		error = xfs_bmapi_write(tp, ip, offset_fsb, count_fsb,
+				  XFS_BMAPI_CONVERT, &firstfsb,
 				  1, &imap, &nimaps, &free_list);
 		if (error)
 			goto error_on_bmapi_transaction;
Index: xfs/fs/xfs/xfs_rtalloc.c
===================================================================
--- xfs.orig/fs/xfs/xfs_rtalloc.c	2011-08-23 21:12:01.000000000 +0200
+++ xfs/fs/xfs/xfs_rtalloc.c	2011-08-23 21:13:59.328944179 +0200
@@ -120,9 +120,9 @@ xfs_growfs_rt_alloc(
 		 */
 		nmap = 1;
 		cancelflags |= XFS_TRANS_ABORT;
-		error = xfs_bmapi(tp, ip, oblocks, nblocks - oblocks,
-			XFS_BMAPI_WRITE | XFS_BMAPI_METADATA, &firstblock,
-			resblks, &map, &nmap, &flist);
+		error = xfs_bmapi_write(tp, ip, oblocks, nblocks - oblocks,
+					XFS_BMAPI_METADATA, &firstblock,
+					resblks, &map, &nmap, &flist);
 		if (!error && nmap < 1)
 			error = XFS_ERROR(ENOSPC);
 		if (error)
Index: xfs/fs/xfs/xfs_vnodeops.c
===================================================================
--- xfs.orig/fs/xfs/xfs_vnodeops.c	2011-08-23 21:09:32.000000000 +0200
+++ xfs/fs/xfs/xfs_vnodeops.c	2011-08-23 21:13:59.332277494 +0200
@@ -1629,10 +1629,9 @@ xfs_symlink(
 		first_fsb = 0;
 		nmaps = SYMLINK_MAPS;
 
-		error = xfs_bmapi(tp, ip, first_fsb, fs_blocks,
-				  XFS_BMAPI_WRITE | XFS_BMAPI_METADATA,
-				  &first_block, resblks, mval, &nmaps,
-				  &free_list);
+		error = xfs_bmapi_write(tp, ip, first_fsb, fs_blocks,
+				  XFS_BMAPI_METADATA, &first_block, resblks,
+				  mval, &nmaps, &free_list);
 		if (error)
 			goto error2;
 
@@ -1775,7 +1774,6 @@ xfs_alloc_file_space(
 	xfs_fileoff_t		startoffset_fsb;
 	xfs_fsblock_t		firstfsb;
 	int			nimaps;
-	int			bmapi_flag;
 	int			quota_flag;
 	int			rt;
 	xfs_trans_t		*tp;
@@ -1803,7 +1801,6 @@ xfs_alloc_file_space(
 	count = len;
 	imapp = &imaps[0];
 	nimaps = 1;
-	bmapi_flag = XFS_BMAPI_WRITE | alloc_type;
 	startoffset_fsb	= XFS_B_TO_FSBT(mp, offset);
 	allocatesize_fsb = XFS_B_TO_FSB(mp, count);
 
@@ -1876,14 +1873,10 @@ xfs_alloc_file_space(
 
 		xfs_trans_ijoin(tp, ip);
 
-		/*
-		 * Issue the xfs_bmapi() call to allocate the blocks
-		 */
 		xfs_bmap_init(&free_list, &firstfsb);
-		error = xfs_bmapi(tp, ip, startoffset_fsb,
-				  allocatesize_fsb, bmapi_flag,
-				  &firstfsb, 0, imapp, &nimaps,
-				  &free_list);
+		error = xfs_bmapi_write(tp, ip, startoffset_fsb,
+					allocatesize_fsb, alloc_type, &firstfsb,
+					0, imapp, &nimaps, &free_list);
 		if (error) {
 			goto error0;
 		}

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

[PATCH 13/25] xfs: introduce xfs_bmap_last_extent

[Christoph Hellwig]

[-- Attachment #1: xfs-simplify-xfs_bmap_isaeof --]
[-- Type: text/plain, Size: 8120 bytes --]

Add a common helper for finding the last extent in a file.

Largely based on a patch from Dave Chinner.

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

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:13:59.318944234 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:14:31.535436367 +0200
@@ -204,19 +204,6 @@ xfs_bmap_search_extents(
 	xfs_bmbt_irec_t	*prevp);	/* out: previous extent entry found */
 
 /*
- * Check the last inode extent to determine whether this allocation will result
- * in blocks being allocated at the end of the file. When we allocate new data
- * blocks at the end of the file which do not start at the previous data block,
- * we will try to align the new blocks at stripe unit boundaries.
- */
-STATIC int				/* error */
-xfs_bmap_isaeof(
-	xfs_inode_t	*ip,		/* incore inode pointer */
-	xfs_fileoff_t   off,		/* file offset in fsblocks */
-	int             whichfork,	/* data or attribute fork */
-	char		*aeof);		/* return value */
-
-/*
  * Compute the worst-case number of indirect blocks that will be used
  * for ip's delayed extent of length "len".
  */
@@ -3924,42 +3911,120 @@ xfs_bmap_last_before(
 	return 0;
 }
 
+STATIC int
+xfs_bmap_last_extent(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	int			whichfork,
+	struct xfs_bmbt_irec	*rec,
+	int			*is_empty)
+{
+	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, whichfork);
+	int			error;
+	int			nextents;
+
+	if (!(ifp->if_flags & XFS_IFEXTENTS)) {
+		error = xfs_iread_extents(tp, ip, whichfork);
+		if (error)
+			return error;
+	}
+
+	nextents = ifp->if_bytes / sizeof(xfs_bmbt_rec_t);
+	if (nextents == 0) {
+		*is_empty = 1;
+		return 0;
+	}
+
+	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, nextents - 1), rec);
+	*is_empty = 0;
+	return 0;
+}
+
+/*
+ * Check the last inode extent to determine whether this allocation will result
+ * in blocks being allocated at the end of the file. When we allocate new data
+ * blocks at the end of the file which do not start at the previous data block,
+ * we will try to align the new blocks at stripe unit boundaries.
+ */
+STATIC int
+xfs_bmap_isaeof(
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		off,
+	int			whichfork,
+	char			*aeof)
+{
+	struct xfs_bmbt_irec	rec;
+	int			is_empty;
+	int			error;
+
+	*aeof = 0;
+	error = xfs_bmap_last_extent(NULL, ip, whichfork, &rec, &is_empty);
+	if (error || is_empty)
+		return error;
+
+	/*
+	 * Check we are allocating in the last extent (for delayed allocations)
+	 * or past the last extent for non-delayed allocations.
+	 */
+	*aeof = (off >= rec.br_startoff &&
+		 off < rec.br_startoff + rec.br_blockcount &&
+		 isnullstartblock(rec.br_startblock)) ||
+		off >= rec.br_startoff + rec.br_blockcount;
+	return 0;
+}
+
+/*
+ * Check if the endoff is outside the last extent. If so the caller will grow
+ * the allocation to a stripe unit boundary.
+ */
+int
+xfs_bmap_eof(
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		endoff,
+	int			whichfork,
+	int			*eof)
+{
+	struct xfs_bmbt_irec	rec;
+	int			error;
+
+	error = xfs_bmap_last_extent(NULL, ip, whichfork, &rec, eof);
+	if (error || *eof)
+		return error;
+
+	*eof = endoff >= rec.br_startoff + rec.br_blockcount;
+	return 0;
+}
+
 /*
  * Returns the file-relative block number of the first block past eof in
  * the file.  This is not based on i_size, it is based on the extent records.
  * Returns 0 for local files, as they do not have extent records.
  */
-int						/* error */
+int
 xfs_bmap_last_offset(
-	xfs_trans_t	*tp,			/* transaction pointer */
-	xfs_inode_t	*ip,			/* incore inode */
-	xfs_fileoff_t	*last_block,		/* last block */
-	int		whichfork)		/* data or attr fork */
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	xfs_fileoff_t		*last_block,
+	int			whichfork)
 {
-	xfs_bmbt_rec_host_t *ep;		/* pointer to last extent */
-	int		error;			/* error return value */
-	xfs_ifork_t	*ifp;			/* inode fork pointer */
-	xfs_extnum_t	nextents;		/* number of extent entries */
+	struct xfs_bmbt_irec	rec;
+	int			is_empty;
+	int			error;
+
+	*last_block = 0;
+
+	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_LOCAL)
+		return 0;
 
 	if (XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE &&
-	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS &&
-	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_LOCAL)
+	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS)
 	       return XFS_ERROR(EIO);
-	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_LOCAL) {
-		*last_block = 0;
-		return 0;
-	}
-	ifp = XFS_IFORK_PTR(ip, whichfork);
-	if (!(ifp->if_flags & XFS_IFEXTENTS) &&
-	    (error = xfs_iread_extents(tp, ip, whichfork)))
+
+	error = xfs_bmap_last_extent(NULL, ip, whichfork, &rec, &is_empty);
+	if (error || is_empty)
 		return error;
-	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
-	if (!nextents) {
-		*last_block = 0;
-		return 0;
-	}
-	ep = xfs_iext_get_ext(ifp, nextents - 1);
-	*last_block = xfs_bmbt_get_startoff(ep) + xfs_bmbt_get_blockcount(ep);
+
+	*last_block = rec.br_startoff + rec.br_blockcount;
 	return 0;
 }
 
@@ -5692,89 +5757,6 @@ xfs_getbmap(
 	return error;
 }
 
-/*
- * Check the last inode extent to determine whether this allocation will result
- * in blocks being allocated at the end of the file. When we allocate new data
- * blocks at the end of the file which do not start at the previous data block,
- * we will try to align the new blocks at stripe unit boundaries.
- */
-STATIC int				/* error */
-xfs_bmap_isaeof(
-	xfs_inode_t	*ip,		/* incore inode pointer */
-	xfs_fileoff_t   off,		/* file offset in fsblocks */
-	int             whichfork,	/* data or attribute fork */
-	char		*aeof)		/* return value */
-{
-	int		error;		/* error return value */
-	xfs_ifork_t	*ifp;		/* inode fork pointer */
-	xfs_bmbt_rec_host_t *lastrec;	/* extent record pointer */
-	xfs_extnum_t	nextents;	/* number of file extents */
-	xfs_bmbt_irec_t	s;		/* expanded extent record */
-
-	ASSERT(whichfork == XFS_DATA_FORK);
-	ifp = XFS_IFORK_PTR(ip, whichfork);
-	if (!(ifp->if_flags & XFS_IFEXTENTS) &&
-	    (error = xfs_iread_extents(NULL, ip, whichfork)))
-		return error;
-	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
-	if (nextents == 0) {
-		*aeof = 1;
-		return 0;
-	}
-	/*
-	 * Go to the last extent
-	 */
-	lastrec = xfs_iext_get_ext(ifp, nextents - 1);
-	xfs_bmbt_get_all(lastrec, &s);
-	/*
-	 * Check we are allocating in the last extent (for delayed allocations)
-	 * or past the last extent for non-delayed allocations.
-	 */
-	*aeof = (off >= s.br_startoff &&
-		 off < s.br_startoff + s.br_blockcount &&
-		 isnullstartblock(s.br_startblock)) ||
-		off >= s.br_startoff + s.br_blockcount;
-	return 0;
-}
-
-/*
- * Check if the endoff is outside the last extent. If so the caller will grow
- * the allocation to a stripe unit boundary.
- */
-int					/* error */
-xfs_bmap_eof(
-	xfs_inode_t	*ip,		/* incore inode pointer */
-	xfs_fileoff_t	endoff,		/* file offset in fsblocks */
-	int		whichfork,	/* data or attribute fork */
-	int		*eof)		/* result value */
-{
-	xfs_fsblock_t	blockcount;	/* extent block count */
-	int		error;		/* error return value */
-	xfs_ifork_t	*ifp;		/* inode fork pointer */
-	xfs_bmbt_rec_host_t *lastrec;	/* extent record pointer */
-	xfs_extnum_t	nextents;	/* number of file extents */
-	xfs_fileoff_t	startoff;	/* extent starting file offset */
-
-	ASSERT(whichfork == XFS_DATA_FORK);
-	ifp = XFS_IFORK_PTR(ip, whichfork);
-	if (!(ifp->if_flags & XFS_IFEXTENTS) &&
-	    (error = xfs_iread_extents(NULL, ip, whichfork)))
-		return error;
-	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
-	if (nextents == 0) {
-		*eof = 1;
-		return 0;
-	}
-	/*
-	 * Go to the last extent
-	 */
-	lastrec = xfs_iext_get_ext(ifp, nextents - 1);
-	startoff = xfs_bmbt_get_startoff(lastrec);
-	blockcount = xfs_bmbt_get_blockcount(lastrec);
-	*eof = endoff >= startoff + blockcount;
-	return 0;
-}
-
 #ifdef DEBUG
 STATIC struct xfs_buf *
 xfs_bmap_get_bp(

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

[PATCH 14/25] xfs: remove xfs_bmap_add_extent

[Christoph Hellwig]

[-- Attachment #1: xfs-bmapi-split-xfs_bmap_extent-2 --]
[-- Type: text/plain, Size: 21547 bytes --]

There is no real need to the xfs_bmap_add_extent, as the callers know what
kind of extents they need to it.  Removing it means duplicating the
extents to btree conversion logic in three places, but overall it's still
a lot less, and much simpler code.

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

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:14:31.535436367 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:14:33.735424450 +0200
@@ -50,17 +50,22 @@
 #include "xfs_trace.h"
 
 
-#ifdef DEBUG
-STATIC void
-xfs_bmap_check_leaf_extents(xfs_btree_cur_t *cur, xfs_inode_t *ip, int whichfork);
-#endif
-
 kmem_zone_t		*xfs_bmap_free_item_zone;
 
 /*
  * Prototypes for internal bmap routines.
  */
 
+#ifdef DEBUG
+STATIC void
+xfs_bmap_check_leaf_extents(
+	struct xfs_btree_cur	*cur,
+	struct xfs_inode	*ip,
+	int			whichfork);
+#else
+#define xfs_bmap_check_leaf_extents(cur, ip, whichfork)		do { } while (0)
+#endif
+
 
 /*
  * Called from xfs_bmap_add_attrfork to handle extents format files.
@@ -85,47 +90,6 @@ xfs_bmap_add_attrfork_local(
 	int			*flags);	/* inode logging flags */
 
 /*
- * Called by xfs_bmap_add_extent to handle cases converting a delayed
- * allocation to a real allocation.
- */
-STATIC int				/* error */
-xfs_bmap_add_extent_delay_real(
-	struct xfs_trans	*tp,	/* transaction pointer */
-	xfs_inode_t		*ip,	/* incore inode pointer */
-	xfs_extnum_t		*idx,	/* extent number to update/insert */
-	xfs_btree_cur_t		**curp,	/* if *curp is null, not a btree */
-	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
-	xfs_filblks_t		*dnew,	/* new delayed-alloc indirect blocks */
-	xfs_fsblock_t		*first,	/* pointer to firstblock variable */
-	xfs_bmap_free_t		*flist,	/* list of extents to be freed */
-	int			*logflagsp); /* inode logging flags */
-
-/*
- * Called by xfs_bmap_add_extent to handle cases converting a hole
- * to a real allocation.
- */
-STATIC int				/* error */
-xfs_bmap_add_extent_hole_real(
-	xfs_inode_t		*ip,	/* incore inode pointer */
-	xfs_extnum_t		*idx,	/* extent number to update/insert */
-	xfs_btree_cur_t		*cur,	/* if null, not a btree */
-	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
-	int			*logflagsp, /* inode logging flags */
-	int			whichfork); /* data or attr fork */
-
-/*
- * Called by xfs_bmap_add_extent to handle cases converting an unwritten
- * allocation to a real allocation or vice versa.
- */
-STATIC int				/* error */
-xfs_bmap_add_extent_unwritten_real(
-	xfs_inode_t		*ip,	/* incore inode pointer */
-	xfs_extnum_t		*idx,	/* extent number to update/insert */
-	xfs_btree_cur_t		**curp,	/* if *curp is null, not a btree */
-	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
-	int			*logflagsp); /* inode logging flags */
-
-/*
  * xfs_bmap_alloc is called by xfs_bmapi to allocate an extent for a file.
  * It figures out where to ask the underlying allocator to put the new extent.
  */
@@ -407,147 +371,7 @@ xfs_bmap_add_attrfork_local(
 }
 
 /*
- * Update file extent records and the btree after allocating space.
- */
-STATIC int				/* error */
-xfs_bmap_add_extent(
-	struct xfs_trans	*tp,	/* transaction pointer */
-	xfs_inode_t		*ip,	/* incore inode pointer */
-	xfs_extnum_t		*idx,	/* extent number to update/insert */
-	xfs_btree_cur_t		**curp,	/* if *curp is null, not a btree */
-	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
-	xfs_fsblock_t		*first,	/* pointer to firstblock variable */
-	xfs_bmap_free_t		*flist,	/* list of extents to be freed */
-	int			*logflagsp, /* inode logging flags */
-	int			whichfork) /* data or attr fork */
-{
-	xfs_btree_cur_t		*cur;	/* btree cursor or null */
-	xfs_filblks_t		da_new; /* new count del alloc blocks used */
-	xfs_filblks_t		da_old; /* old count del alloc blocks used */
-	int			error;	/* error return value */
-	xfs_ifork_t		*ifp;	/* inode fork ptr */
-	int			logflags; /* returned value */
-	xfs_extnum_t		nextents; /* number of extents in file now */
-
-	XFS_STATS_INC(xs_add_exlist);
-
-	cur = *curp;
-	ifp = XFS_IFORK_PTR(ip, whichfork);
-	nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
-	da_old = da_new = 0;
-	error = 0;
-
-	ASSERT(*idx >= 0);
-	ASSERT(*idx <= nextents);
-	ASSERT(!isnullstartblock(new->br_startblock));
-
-	/*
-	 * Real allocation off the end of the file.
-	 */
-	if (*idx == nextents) {
-		if (cur)
-			ASSERT((cur->bc_private.b.flags &
-				XFS_BTCUR_BPRV_WASDEL) == 0);
-		error = xfs_bmap_add_extent_hole_real(ip, idx, cur, new,
-				&logflags, whichfork);
-	} else {
-		xfs_bmbt_irec_t	prev;	/* old extent at offset idx */
-
-		/*
-		 * Get the record referred to by idx.
-		 */
-		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *idx), &prev);
-		/*
-		 * If it's a real allocation record, and the new allocation ends
-		 * after the start of the referred to record, then we're filling
-		 * in a delayed or unwritten allocation with a real one, or
-		 * converting real back to unwritten.
-		 */
-		if (!isnullstartblock(new->br_startblock) &&
-		    new->br_startoff + new->br_blockcount > prev.br_startoff) {
-			if (prev.br_state != XFS_EXT_UNWRITTEN &&
-			    isnullstartblock(prev.br_startblock)) {
-				da_old = startblockval(prev.br_startblock);
-				if (cur)
-					ASSERT(cur->bc_private.b.flags &
-						XFS_BTCUR_BPRV_WASDEL);
-				error = xfs_bmap_add_extent_delay_real(tp, ip,
-						idx, &cur, new, &da_new,
-						first, flist, &logflags);
-			} else {
-				ASSERT(new->br_state == XFS_EXT_NORM ||
-				       new->br_state == XFS_EXT_UNWRITTEN);
-
-				error = xfs_bmap_add_extent_unwritten_real(ip,
-						idx, &cur, new, &logflags);
-				if (error)
-					goto done;
-			}
-		}
-		/*
-		 * Otherwise we're filling in a hole with an allocation.
-		 */
-		else {
-			if (cur)
-				ASSERT((cur->bc_private.b.flags &
-					XFS_BTCUR_BPRV_WASDEL) == 0);
-			error = xfs_bmap_add_extent_hole_real(ip, idx, cur,
-					new, &logflags, whichfork);
-		}
-	}
-
-	if (error)
-		goto done;
-	ASSERT(*curp == cur || *curp == NULL);
-
-	/*
-	 * Convert to a btree if necessary.
-	 */
-	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_EXTENTS &&
-	    XFS_IFORK_NEXTENTS(ip, whichfork) > ifp->if_ext_max) {
-		int	tmp_logflags;	/* partial log flag return val */
-
-		ASSERT(cur == NULL);
-		error = xfs_bmap_extents_to_btree(tp, ip, first,
-			flist, &cur, da_old > 0, &tmp_logflags, whichfork);
-		logflags |= tmp_logflags;
-		if (error)
-			goto done;
-	}
-	/*
-	 * Adjust for changes in reserved delayed indirect blocks.
-	 * Nothing to do for disk quotas here.
-	 */
-	if (da_old || da_new) {
-		xfs_filblks_t	nblks;
-
-		nblks = da_new;
-		if (cur)
-			nblks += cur->bc_private.b.allocated;
-		ASSERT(nblks <= da_old);
-		if (nblks < da_old)
-			xfs_icsb_modify_counters(ip->i_mount, XFS_SBS_FDBLOCKS,
-				(int64_t)(da_old - nblks), 0);
-	}
-	/*
-	 * Clear out the allocated field, done with it now in any case.
-	 */
-	if (cur) {
-		cur->bc_private.b.allocated = 0;
-		*curp = cur;
-	}
-done:
-#ifdef DEBUG
-	if (!error)
-		xfs_bmap_check_leaf_extents(*curp, ip, whichfork);
-#endif
-	*logflagsp = logflags;
-	return error;
-}
-
-/*
- * Called by xfs_bmap_add_extent to handle cases converting a delayed
- * allocation to a real allocation.
+ * Convert a delayed allocation to a real allocation.
  */
 STATIC int				/* error */
 xfs_bmap_add_extent_delay_real(
@@ -556,7 +380,6 @@ xfs_bmap_add_extent_delay_real(
 	xfs_extnum_t		*idx,	/* extent number to update/insert */
 	xfs_btree_cur_t		**curp,	/* if *curp is null, not a btree */
 	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
-	xfs_filblks_t		*dnew,	/* new delayed-alloc indirect blocks */
 	xfs_fsblock_t		*first,	/* pointer to firstblock variable */
 	xfs_bmap_free_t		*flist,	/* list of extents to be freed */
 	int			*logflagsp) /* inode logging flags */
@@ -572,10 +395,24 @@ xfs_bmap_add_extent_delay_real(
 					/* left is 0, right is 1, prev is 2 */
 	int			rval=0;	/* return value (logging flags) */
 	int			state = 0;/* state bits, accessed thru macros */
-	xfs_filblks_t		temp=0;	/* value for dnew calculations */
-	xfs_filblks_t		temp2=0;/* value for dnew calculations */
+	xfs_filblks_t		da_new; /* new count del alloc blocks used */
+	xfs_filblks_t		da_old; /* old count del alloc blocks used */
+	xfs_filblks_t		temp=0;	/* value for da_new calculations */
+	xfs_filblks_t		temp2=0;/* value for da_new calculations */
 	int			tmp_rval;	/* partial logging flags */
 
+	ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
+	cur = *curp;
+
+	ASSERT(*idx >= 0);
+	ASSERT(*idx <= ifp->if_bytes / sizeof(struct xfs_bmbt_rec));
+	ASSERT(!isnullstartblock(new->br_startblock));
+	ASSERT(!cur || (cur->bc_private.b.flags & XFS_BTCUR_BPRV_WASDEL));
+
+	XFS_STATS_INC(xs_add_exlist);
+
+	*logflagsp = 0;
+
 #define	LEFT		r[0]
 #define	RIGHT		r[1]
 #define	PREV		r[2]
@@ -583,14 +420,15 @@ xfs_bmap_add_extent_delay_real(
 	/*
 	 * Set up a bunch of variables to make the tests simpler.
 	 */
-	cur = *curp;
-	ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
 	ep = xfs_iext_get_ext(ifp, *idx);
 	xfs_bmbt_get_all(ep, &PREV);
 	new_endoff = new->br_startoff + new->br_blockcount;
 	ASSERT(PREV.br_startoff <= new->br_startoff);
 	ASSERT(PREV.br_startoff + PREV.br_blockcount >= new_endoff);
 
+	da_old = startblockval(PREV.br_startblock);
+	da_new = 0;
+
 	/*
 	 * Set flags determining what part of the previous delayed allocation
 	 * extent is being replaced by a real allocation.
@@ -688,7 +526,6 @@ xfs_bmap_add_extent_delay_real(
 					RIGHT.br_blockcount, LEFT.br_state)))
 				goto done;
 		}
-		*dnew = 0;
 		break;
 
 	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING | BMAP_LEFT_CONTIG:
@@ -719,7 +556,6 @@ xfs_bmap_add_extent_delay_real(
 					PREV.br_blockcount, LEFT.br_state)))
 				goto done;
 		}
-		*dnew = 0;
 		break;
 
 	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
@@ -749,8 +585,6 @@ xfs_bmap_add_extent_delay_real(
 					RIGHT.br_blockcount, PREV.br_state)))
 				goto done;
 		}
-
-		*dnew = 0;
 		break;
 
 	case BMAP_LEFT_FILLING | BMAP_RIGHT_FILLING:
@@ -778,8 +612,6 @@ xfs_bmap_add_extent_delay_real(
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		}
-
-		*dnew = 0;
 		break;
 
 	case BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG:
@@ -813,13 +645,12 @@ xfs_bmap_add_extent_delay_real(
 					LEFT.br_state)))
 				goto done;
 		}
-		temp = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
 			startblockval(PREV.br_startblock));
-		xfs_bmbt_set_startblock(ep, nullstartblock((int)temp));
+		xfs_bmbt_set_startblock(ep, nullstartblock(da_new));
 		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
 
 		--*idx;
-		*dnew = temp;
 		break;
 
 	case BMAP_LEFT_FILLING:
@@ -856,14 +687,12 @@ xfs_bmap_add_extent_delay_real(
 			if (error)
 				goto done;
 		}
-		temp = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
 			startblockval(PREV.br_startblock) -
 			(cur ? cur->bc_private.b.allocated : 0));
 		ep = xfs_iext_get_ext(ifp, *idx + 1);
-		xfs_bmbt_set_startblock(ep, nullstartblock((int)temp));
+		xfs_bmbt_set_startblock(ep, nullstartblock(da_new));
 		trace_xfs_bmap_post_update(ip, *idx + 1, state, _THIS_IP_);
-
-		*dnew = temp;
 		break;
 
 	case BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
@@ -896,14 +725,13 @@ xfs_bmap_add_extent_delay_real(
 				goto done;
 		}
 
-		temp = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
 			startblockval(PREV.br_startblock));
 		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
-		xfs_bmbt_set_startblock(ep, nullstartblock((int)temp));
+		xfs_bmbt_set_startblock(ep, nullstartblock(da_new));
 		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
 
 		++*idx;
-		*dnew = temp;
 		break;
 
 	case BMAP_RIGHT_FILLING:
@@ -939,15 +767,14 @@ xfs_bmap_add_extent_delay_real(
 			if (error)
 				goto done;
 		}
-		temp = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
 			startblockval(PREV.br_startblock) -
 			(cur ? cur->bc_private.b.allocated : 0));
 		ep = xfs_iext_get_ext(ifp, *idx);
-		xfs_bmbt_set_startblock(ep, nullstartblock((int)temp));
+		xfs_bmbt_set_startblock(ep, nullstartblock(da_new));
 		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
 
 		++*idx;
-		*dnew = temp;
 		break;
 
 	case 0:
@@ -1029,7 +856,7 @@ xfs_bmap_add_extent_delay_real(
 		trace_xfs_bmap_post_update(ip, *idx + 2, state, _THIS_IP_);
 
 		++*idx;
-		*dnew = temp + temp2;
+		da_new = temp + temp2;
 		break;
 
 	case BMAP_LEFT_FILLING | BMAP_LEFT_CONTIG | BMAP_RIGHT_CONTIG:
@@ -1044,9 +871,39 @@ xfs_bmap_add_extent_delay_real(
 		 */
 		ASSERT(0);
 	}
-	*curp = cur;
+
+	/* convert to a btree if necessary */
+	if (XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) == XFS_DINODE_FMT_EXTENTS &&
+	    XFS_IFORK_NEXTENTS(ip, XFS_DATA_FORK) > ifp->if_ext_max) {
+		int	tmp_logflags;	/* partial log flag return val */
+
+		ASSERT(cur == NULL);
+		error = xfs_bmap_extents_to_btree(tp, ip, first, flist, &cur,
+				da_old > 0, &tmp_logflags, XFS_DATA_FORK);
+		*logflagsp |= tmp_logflags;
+		if (error)
+			goto done;
+	}
+
+	/* adjust for changes in reserved delayed indirect blocks */
+	if (da_old || da_new) {
+		temp = da_new;
+		if (cur)
+			temp += cur->bc_private.b.allocated;
+		ASSERT(temp <= da_old);
+		if (temp < da_old)
+			xfs_icsb_modify_counters(ip->i_mount, XFS_SBS_FDBLOCKS,
+				(int64_t)(da_old - temp), 0);
+	}
+
+	/* clear out the allocated field, done with it now in any case. */
+	if (cur) {
+		cur->bc_private.b.allocated = 0;
+		*curp = cur;
+	}
+	xfs_bmap_check_leaf_extents(cur, ip, XFS_DATA_FORK);
 done:
-	*logflagsp = rval;
+	*logflagsp |= rval;
 	return error;
 #undef	LEFT
 #undef	RIGHT
@@ -1054,15 +911,17 @@ done:
 }
 
 /*
- * Called by xfs_bmap_add_extent to handle cases converting an unwritten
- * allocation to a real allocation or vice versa.
+ * Convert an unwritten allocation to a real allocation or vice versa.
  */
 STATIC int				/* error */
 xfs_bmap_add_extent_unwritten_real(
+	struct xfs_trans	*tp,
 	xfs_inode_t		*ip,	/* incore inode pointer */
 	xfs_extnum_t		*idx,	/* extent number to update/insert */
 	xfs_btree_cur_t		**curp,	/* if *curp is null, not a btree */
 	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
+	xfs_fsblock_t		*first,	/* pointer to firstblock variable */
+	xfs_bmap_free_t		*flist,	/* list of extents to be freed */
 	int			*logflagsp) /* inode logging flags */
 {
 	xfs_btree_cur_t		*cur;	/* btree cursor */
@@ -1078,15 +937,25 @@ xfs_bmap_add_extent_unwritten_real(
 	int			rval=0;	/* return value (logging flags) */
 	int			state = 0;/* state bits, accessed thru macros */
 
+	*logflagsp = 0;
+
+	cur = *curp;
+	ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
+
+	ASSERT(*idx >= 0);
+	ASSERT(*idx <= ifp->if_bytes / sizeof(struct xfs_bmbt_rec));
+	ASSERT(!isnullstartblock(new->br_startblock));
+
+	XFS_STATS_INC(xs_add_exlist);
+
 #define	LEFT		r[0]
 #define	RIGHT		r[1]
 #define	PREV		r[2]
+
 	/*
 	 * Set up a bunch of variables to make the tests simpler.
 	 */
 	error = 0;
-	cur = *curp;
-	ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
 	ep = xfs_iext_get_ext(ifp, *idx);
 	xfs_bmbt_get_all(ep, &PREV);
 	newext = new->br_state;
@@ -1537,9 +1406,29 @@ xfs_bmap_add_extent_unwritten_real(
 		 */
 		ASSERT(0);
 	}
-	*curp = cur;
+
+	/* convert to a btree if necessary */
+	if (XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) == XFS_DINODE_FMT_EXTENTS &&
+	    XFS_IFORK_NEXTENTS(ip, XFS_DATA_FORK) > ifp->if_ext_max) {
+		int	tmp_logflags;	/* partial log flag return val */
+
+		ASSERT(cur == NULL);
+		error = xfs_bmap_extents_to_btree(tp, ip, first, flist, &cur,
+				0, &tmp_logflags, XFS_DATA_FORK);
+		*logflagsp |= tmp_logflags;
+		if (error)
+			goto done;
+	}
+
+	/* clear out the allocated field, done with it now in any case. */
+	if (cur) {
+		cur->bc_private.b.allocated = 0;
+		*curp = cur;
+	}
+
+	xfs_bmap_check_leaf_extents(*curp, ip, XFS_DATA_FORK);
 done:
-	*logflagsp = rval;
+	*logflagsp |= rval;
 	return error;
 #undef	LEFT
 #undef	RIGHT
@@ -1691,30 +1580,42 @@ xfs_bmap_add_extent_hole_delay(
 }
 
 /*
- * Called by xfs_bmap_add_extent to handle cases converting a hole
- * to a real allocation.
+ * Convert a hole to a real allocation.
  */
 STATIC int				/* error */
 xfs_bmap_add_extent_hole_real(
+	struct xfs_trans	*tp,
 	xfs_inode_t		*ip,	/* incore inode pointer */
 	xfs_extnum_t		*idx,	/* extent number to update/insert */
-	xfs_btree_cur_t		*cur,	/* if null, not a btree */
+	xfs_btree_cur_t		**curp,	/* if null, not a btree */
 	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
+	xfs_fsblock_t		*first,	/* pointer to firstblock variable */
+	xfs_bmap_free_t		*flist,	/* list of extents to be freed */
 	int			*logflagsp, /* inode logging flags */
 	int			whichfork) /* data or attr fork */
 {
 	int			error;	/* error return value */
 	int			i;	/* temp state */
+	xfs_btree_cur_t		*cur;	/* if null, not a btree */
 	xfs_ifork_t		*ifp;	/* inode fork pointer */
 	xfs_bmbt_irec_t		left;	/* left neighbor extent entry */
 	xfs_bmbt_irec_t		right;	/* right neighbor extent entry */
 	int			rval=0;	/* return value (logging flags) */
 	int			state;	/* state bits, accessed thru macros */
 
+	*logflagsp = 0;
+
 	ifp = XFS_IFORK_PTR(ip, whichfork);
-	ASSERT(*idx <= ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t));
-	state = 0;
+	cur = *curp;
+
+	ASSERT(*idx >= 0);
+	ASSERT(*idx <= ifp->if_bytes / sizeof(struct xfs_bmbt_rec));
+	ASSERT(!isnullstartblock(new->br_startblock));
+	ASSERT(!cur || !(cur->bc_private.b.flags & XFS_BTCUR_BPRV_WASDEL));
+
+	XFS_STATS_INC(xs_add_exlist);
 
+	state = 0;
 	if (whichfork == XFS_ATTR_FORK)
 		state |= BMAP_ATTRFORK;
 
@@ -1897,8 +1798,28 @@ xfs_bmap_add_extent_hole_real(
 		}
 		break;
 	}
+
+	/* convert to a btree if necessary */
+	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_EXTENTS &&
+	    XFS_IFORK_NEXTENTS(ip, whichfork) > ifp->if_ext_max) {
+		int	tmp_logflags;	/* partial log flag return val */
+
+		ASSERT(cur == NULL);
+		error = xfs_bmap_extents_to_btree(tp, ip, first,
+			flist, &cur, 0, &tmp_logflags, whichfork);
+		*logflagsp |= tmp_logflags;
+		if (error)
+			goto done;
+	}
+
+	/* clear out the allocated field, done with it now in any case. */
+	if (cur) {
+		cur->bc_private.b.allocated = 0;
+		*curp = cur;
+	}
+	xfs_bmap_check_leaf_extents(cur, ip, whichfork);
 done:
-	*logflagsp = rval;
+	*logflagsp |= rval;
 	return error;
 }
 
@@ -4787,14 +4708,22 @@ xfs_bmapi_allocate(
 	    xfs_sb_version_hasextflgbit(&mp->m_sb))
 		bma->gotp->br_state = XFS_EXT_UNWRITTEN;
 
-	error = xfs_bmap_add_extent(bma->tp, bma->ip, lastx, cur, bma->gotp,
-				    firstblock, flist, logflags, whichfork);
+	if (bma->wasdel) {
+		error = xfs_bmap_add_extent_delay_real(bma->tp, bma->ip, lastx,
+				cur, bma->gotp, firstblock, flist, logflags);
+	} else {
+		error = xfs_bmap_add_extent_hole_real(bma->tp, bma->ip, lastx,
+				cur, bma->gotp, firstblock, flist, logflags,
+				whichfork);
+	}
+
 	if (error)
 		return error;
 
 	/*
-	 * Update our extent pointer, given that xfs_bmap_add_extent  might
-	 * have merged it into one of the neighbouring ones.
+	 * Update our extent pointer, given that xfs_bmap_add_extent_delay_real
+	 * or xfs_bmap_add_extent_hole_real might have merged it into one of
+	 * the neighbouring ones.
 	 */
 	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx), bma->gotp);
 
@@ -4849,14 +4778,15 @@ xfs_bmapi_convert_unwritten(
 	mval->br_state = (mval->br_state == XFS_EXT_UNWRITTEN)
 				? XFS_EXT_NORM : XFS_EXT_UNWRITTEN;
 
-	error = xfs_bmap_add_extent(bma->tp, bma->ip, lastx, cur, mval,
-				firstblock, flist, logflags, whichfork);
+	error = xfs_bmap_add_extent_unwritten_real(bma->tp, bma->ip, lastx,
+			cur, mval, firstblock, flist, logflags);
 	if (error)
 		return error;
 
 	/*
-	 * Update our extent pointer, given that xfs_bmap_add_extent  might
-	 * have merged it into one of the neighbouring ones.
+	 * Update our extent pointer, given that
+	 * xfs_bmap_add_extent_unwritten_real might have merged it into one
+	 * of the neighbouring ones.
 	 */
 	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx), bma->gotp);
 
@@ -5290,9 +5220,9 @@ xfs_bunmapi(
 				del.br_blockcount = mod;
 			}
 			del.br_state = XFS_EXT_UNWRITTEN;
-			error = xfs_bmap_add_extent(tp, ip, &lastx, &cur, &del,
-				firstblock, flist, &logflags,
-				XFS_DATA_FORK);
+			error = xfs_bmap_add_extent_unwritten_real(tp, ip,
+					&lastx, &cur, &del, firstblock, flist,
+					&logflags);
 			if (error)
 				goto error0;
 			goto nodelete;
@@ -5348,18 +5278,18 @@ xfs_bunmapi(
 				}
 				prev.br_state = XFS_EXT_UNWRITTEN;
 				lastx--;
-				error = xfs_bmap_add_extent(tp, ip, &lastx,
-						&cur, &prev, firstblock, flist,
-						&logflags, XFS_DATA_FORK);
+				error = xfs_bmap_add_extent_unwritten_real(tp,
+						ip, &lastx, &cur, &prev,
+						firstblock, flist, &logflags);
 				if (error)
 					goto error0;
 				goto nodelete;
 			} else {
 				ASSERT(del.br_state == XFS_EXT_NORM);
 				del.br_state = XFS_EXT_UNWRITTEN;
-				error = xfs_bmap_add_extent(tp, ip, &lastx,
-						&cur, &del, firstblock, flist,
-						&logflags, XFS_DATA_FORK);
+				error = xfs_bmap_add_extent_unwritten_real(tp,
+						ip, &lastx, &cur, &del,
+						firstblock, flist, &logflags);
 				if (error)
 					goto error0;
 				goto nodelete;

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

[PATCH 15/25] xfs: pass bmalloca structure to xfs_bmap_isaeof

[Christoph Hellwig]

[-- Attachment #1: xfs-bmalloca-isaeof --]
[-- Type: text/plain, Size: 1911 bytes --]

All the variables xfs_bmap_isaeof() is passed are contained within
the xfs_bmalloca structure. Pass that instead.

Signed-off-by: Dave Chinner <dchinner@redhat.com>

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:14:33.735424450 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:14:35.358748989 +0200
@@ -3869,17 +3869,16 @@ xfs_bmap_last_extent(
  */
 STATIC int
 xfs_bmap_isaeof(
-	struct xfs_inode	*ip,
-	xfs_fileoff_t		off,
-	int			whichfork,
-	char			*aeof)
+	struct xfs_bmalloca	*bma,
+	int			whichfork)
 {
 	struct xfs_bmbt_irec	rec;
 	int			is_empty;
 	int			error;
 
-	*aeof = 0;
-	error = xfs_bmap_last_extent(NULL, ip, whichfork, &rec, &is_empty);
+	bma->aeof = 0;
+	error = xfs_bmap_last_extent(bma->tp, bma->ip, whichfork, &rec,
+				     &is_empty);
 	if (error || is_empty)
 		return error;
 
@@ -3887,10 +3886,10 @@ xfs_bmap_isaeof(
 	 * Check we are allocating in the last extent (for delayed allocations)
 	 * or past the last extent for non-delayed allocations.
 	 */
-	*aeof = (off >= rec.br_startoff &&
-		 off < rec.br_startoff + rec.br_blockcount &&
-		 isnullstartblock(rec.br_startblock)) ||
-		off >= rec.br_startoff + rec.br_blockcount;
+	bma->aeof = (bma->off >= rec.br_startoff &&
+		     bma->off < rec.br_startoff + rec.br_blockcount &&
+		     isnullstartblock(rec.br_startblock)) ||
+		    bma->off >= rec.br_startoff + rec.br_blockcount;
 	return 0;
 }
 
@@ -4653,7 +4652,7 @@ xfs_bmapi_allocate(
 	 */
 	if (mp->m_dalign && alen >= mp->m_dalign &&
 	    !(flags & XFS_BMAPI_METADATA) && whichfork == XFS_DATA_FORK) {
-		error = xfs_bmap_isaeof(bma->ip, aoff, whichfork, &bma->aeof);
+		error = xfs_bmap_isaeof(bma, whichfork);
 		if (error)
 			return error;
 	}

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

[PATCH 16/25] xfs: move extent records into bmalloca structure

[Christoph Hellwig]

[-- Attachment #1: xfs-bmalloca-extents --]
[-- Type: text/plain, Size: 10188 bytes --]

Rather that putting extent records on the stack and then pointing to
them in the bmalloca structure which is in the same stack frame, put
the extent records directly in the bmalloca structure. This reduces
the number of args that need to be passed around.

Signed-off-by: Dave Chinner <dchinner@redhat.com>

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:14:35.358748989 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:14:37.402071253 +0200
@@ -2013,18 +2013,18 @@ xfs_bmap_adjacent(
 	 * If allocating at eof, and there's a previous real block,
 	 * try to use its last block as our starting point.
 	 */
-	if (ap->eof && ap->prevp->br_startoff != NULLFILEOFF &&
-	    !isnullstartblock(ap->prevp->br_startblock) &&
-	    ISVALID(ap->prevp->br_startblock + ap->prevp->br_blockcount,
-		    ap->prevp->br_startblock)) {
-		ap->rval = ap->prevp->br_startblock + ap->prevp->br_blockcount;
+	if (ap->eof && ap->prev.br_startoff != NULLFILEOFF &&
+	    !isnullstartblock(ap->prev.br_startblock) &&
+	    ISVALID(ap->prev.br_startblock + ap->prev.br_blockcount,
+		    ap->prev.br_startblock)) {
+		ap->rval = ap->prev.br_startblock + ap->prev.br_blockcount;
 		/*
 		 * Adjust for the gap between prevp and us.
 		 */
 		adjust = ap->off -
-			(ap->prevp->br_startoff + ap->prevp->br_blockcount);
+			(ap->prev.br_startoff + ap->prev.br_blockcount);
 		if (adjust &&
-		    ISVALID(ap->rval + adjust, ap->prevp->br_startblock))
+		    ISVALID(ap->rval + adjust, ap->prev.br_startblock))
 			ap->rval += adjust;
 	}
 	/*
@@ -2042,17 +2042,17 @@ xfs_bmap_adjacent(
 		 * If there's a previous (left) block, select a requested
 		 * start block based on it.
 		 */
-		if (ap->prevp->br_startoff != NULLFILEOFF &&
-		    !isnullstartblock(ap->prevp->br_startblock) &&
-		    (prevbno = ap->prevp->br_startblock +
-			       ap->prevp->br_blockcount) &&
-		    ISVALID(prevbno, ap->prevp->br_startblock)) {
+		if (ap->prev.br_startoff != NULLFILEOFF &&
+		    !isnullstartblock(ap->prev.br_startblock) &&
+		    (prevbno = ap->prev.br_startblock +
+			       ap->prev.br_blockcount) &&
+		    ISVALID(prevbno, ap->prev.br_startblock)) {
 			/*
 			 * Calculate gap to end of previous block.
 			 */
 			adjust = prevdiff = ap->off -
-				(ap->prevp->br_startoff +
-				 ap->prevp->br_blockcount);
+				(ap->prev.br_startoff +
+				 ap->prev.br_blockcount);
 			/*
 			 * Figure the startblock based on the previous block's
 			 * end and the gap size.
@@ -2063,7 +2063,7 @@ xfs_bmap_adjacent(
 			 */
 			if (prevdiff <= XFS_ALLOC_GAP_UNITS * ap->alen &&
 			    ISVALID(prevbno + prevdiff,
-				    ap->prevp->br_startblock))
+				    ap->prev.br_startblock))
 				prevbno += adjust;
 			else
 				prevdiff += adjust;
@@ -2084,16 +2084,16 @@ xfs_bmap_adjacent(
 		 * If there's a following (right) block, select a requested
 		 * start block based on it.
 		 */
-		if (!isnullstartblock(ap->gotp->br_startblock)) {
+		if (!isnullstartblock(ap->got.br_startblock)) {
 			/*
 			 * Calculate gap to start of next block.
 			 */
-			adjust = gotdiff = ap->gotp->br_startoff - ap->off;
+			adjust = gotdiff = ap->got.br_startoff - ap->off;
 			/*
 			 * Figure the startblock based on the next block's
 			 * start and the gap size.
 			 */
-			gotbno = ap->gotp->br_startblock;
+			gotbno = ap->got.br_startblock;
 			/*
 			 * Heuristic!
 			 * If the gap is large relative to the piece we're
@@ -2151,7 +2151,7 @@ xfs_bmap_rtalloc(
 	mp = ap->ip->i_mount;
 	align = xfs_get_extsz_hint(ap->ip);
 	prod = align / mp->m_sb.sb_rextsize;
-	error = xfs_bmap_extsize_align(mp, ap->gotp, ap->prevp,
+	error = xfs_bmap_extsize_align(mp, &ap->got, &ap->prev,
 					align, 1, ap->eof, 0,
 					ap->conv, &ap->off, &ap->alen);
 	if (error)
@@ -2374,7 +2374,7 @@ xfs_bmap_btalloc(
 	mp = ap->ip->i_mount;
 	align = ap->userdata ? xfs_get_extsz_hint(ap->ip) : 0;
 	if (unlikely(align)) {
-		error = xfs_bmap_extsize_align(mp, ap->gotp, ap->prevp,
+		error = xfs_bmap_extsize_align(mp, &ap->got, &ap->prev,
 						align, 0, ap->eof, 0, ap->conv,
 						&ap->off, &ap->alen);
 		ASSERT(!error);
@@ -4613,17 +4613,17 @@ xfs_bmapi_allocate(
 	 * for in this bmap call but that wouldn't be as good.
 	 */
 	if (bma->wasdel) {
-		alen = (xfs_extlen_t)bma->gotp->br_blockcount;
-		aoff = bma->gotp->br_startoff;
+		alen = (xfs_extlen_t)bma->got.br_blockcount;
+		aoff = bma->got.br_startoff;
 		if (*lastx != NULLEXTNUM && *lastx) {
 			xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx - 1),
-					 bma->prevp);
+					 &bma->prev);
 		}
 	} else {
 		alen = (xfs_extlen_t)XFS_FILBLKS_MIN(bma->alen, MAXEXTLEN);
 		if (!bma->eof)
 			alen = (xfs_extlen_t)XFS_FILBLKS_MIN(alen,
-					bma->gotp->br_startoff - bma->off);
+					bma->got.br_startoff - bma->off);
 		aoff = bma->off;
 	}
 
@@ -4694,10 +4694,10 @@ xfs_bmapi_allocate(
 		(*cur)->bc_private.b.flags =
 			bma->wasdel ? XFS_BTCUR_BPRV_WASDEL : 0;
 
-	bma->gotp->br_startoff = aoff;
-	bma->gotp->br_startblock = abno;
-	bma->gotp->br_blockcount = alen;
-	bma->gotp->br_state = XFS_EXT_NORM;
+	bma->got.br_startoff = aoff;
+	bma->got.br_startblock = abno;
+	bma->got.br_blockcount = alen;
+	bma->got.br_state = XFS_EXT_NORM;
 
 	/*
 	 * A wasdelay extent has been initialized, so shouldn't be flagged
@@ -4705,14 +4705,14 @@ xfs_bmapi_allocate(
 	 */
 	if (!bma->wasdel && (flags & XFS_BMAPI_PREALLOC) &&
 	    xfs_sb_version_hasextflgbit(&mp->m_sb))
-		bma->gotp->br_state = XFS_EXT_UNWRITTEN;
+		bma->got.br_state = XFS_EXT_UNWRITTEN;
 
 	if (bma->wasdel) {
 		error = xfs_bmap_add_extent_delay_real(bma->tp, bma->ip, lastx,
-				cur, bma->gotp, firstblock, flist, logflags);
+				cur, &bma->got, firstblock, flist, logflags);
 	} else {
 		error = xfs_bmap_add_extent_hole_real(bma->tp, bma->ip, lastx,
-				cur, bma->gotp, firstblock, flist, logflags,
+				cur, &bma->got, firstblock, flist, logflags,
 				whichfork);
 	}
 
@@ -4724,13 +4724,12 @@ xfs_bmapi_allocate(
 	 * or xfs_bmap_add_extent_hole_real might have merged it into one of
 	 * the neighbouring ones.
 	 */
-	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx), bma->gotp);
+	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx), &bma->got);
 
-	ASSERT(bma->gotp->br_startoff <= aoff);
-	ASSERT(bma->gotp->br_startoff + bma->gotp->br_blockcount >=
-		aoff + alen);
-	ASSERT(bma->gotp->br_state == XFS_EXT_NORM ||
-	       bma->gotp->br_state == XFS_EXT_UNWRITTEN);
+	ASSERT(bma->got.br_startoff <= aoff);
+	ASSERT(bma->got.br_startoff + bma->got.br_blockcount >= aoff + alen);
+	ASSERT(bma->got.br_state == XFS_EXT_NORM ||
+	       bma->got.br_state == XFS_EXT_UNWRITTEN);
 	return 0;
 }
 
@@ -4787,7 +4786,7 @@ xfs_bmapi_convert_unwritten(
 	 * xfs_bmap_add_extent_unwritten_real might have merged it into one
 	 * of the neighbouring ones.
 	 */
-	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx), bma->gotp);
+	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx), &bma->got);
 
 	/*
 	 * We may have combined previously unwritten space with written space,
@@ -4830,7 +4829,6 @@ xfs_bmapi_write(
 	xfs_fileoff_t		end;		/* end of mapped file region */
 	int			eof;		/* after the end of extents */
 	int			error;		/* error return */
-	struct xfs_bmbt_irec	got;		/* current file extent record */
 	xfs_ifork_t		*ifp;		/* inode fork pointer */
 	xfs_extnum_t		lastx;		/* last useful extent number */
 	int			logflags;	/* flags for transaction logging */
@@ -4838,7 +4836,6 @@ xfs_bmapi_write(
 	int			n;		/* current extent index */
 	int			nallocs;	/* number of extents alloc'd */
 	xfs_fileoff_t		obno;		/* old block number (offset) */
-	struct xfs_bmbt_irec	prev;		/* previous file extent record */
 	int			tmp_logflags;	/* temp flags holder */
 	int			whichfork;	/* data or attr fork */
 	char			inhole;		/* current location is hole in file */
@@ -4910,21 +4907,20 @@ xfs_bmapi_write(
 			goto error0;
 	}
 
-	xfs_bmap_search_extents(ip, bno, whichfork, &eof, &lastx, &got, &prev);
+	xfs_bmap_search_extents(ip, bno, whichfork, &eof, &lastx, &bma.got,
+				&bma.prev);
 	n = 0;
 	end = bno + len;
 	obno = bno;
 
 	bma.tp = tp;
 	bma.ip = ip;
-	bma.prevp = &prev;
-	bma.gotp = &got;
 	bma.total = total;
 	bma.userdata = 0;
 
 	while (bno < end && n < *nmap) {
-		inhole = eof || got.br_startoff > bno;
-		wasdelay = !inhole && isnullstartblock(got.br_startblock);
+		inhole = eof || bma.got.br_startoff > bno;
+		wasdelay = !inhole && isnullstartblock(bma.got.br_startblock);
 
 		/*
 		 * First, deal with the hole before the allocated space
@@ -4959,7 +4955,8 @@ xfs_bmapi_write(
 		}
 
 		/* Deal with the allocated space we found.  */
-		xfs_bmapi_trim_map(mval, &got, &bno, len, obno, end, n, flags);
+		xfs_bmapi_trim_map(mval, &bma.got, &bno, len, obno,
+							end, n, flags);
 
 		/* Execute unwritten extent conversion if necessary */
 		error = xfs_bmapi_convert_unwritten(&bma, mval, len, &lastx,
@@ -4983,9 +4980,9 @@ xfs_bmapi_write(
 			break;
 
 		/* Else go on to the next record. */
-		prev = got;
+		bma.prev = bma.got;
 		if (++lastx < ifp->if_bytes / sizeof(xfs_bmbt_rec_t))
-			xfs_bmbt_get_all(xfs_iext_get_ext(ifp, lastx), &got);
+			xfs_bmbt_get_all(xfs_iext_get_ext(ifp, lastx), &bma.got);
 		else
 			eof = 1;
 	}
Index: xfs/fs/xfs/xfs_bmap.h
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.h	2011-08-23 21:13:59.318944234 +0200
+++ xfs/fs/xfs/xfs_bmap.h	2011-08-23 21:14:37.402071253 +0200
@@ -114,8 +114,8 @@ typedef struct xfs_bmalloca {
 	xfs_fileoff_t		off;	/* offset in file filling in */
 	struct xfs_trans	*tp;	/* transaction pointer */
 	struct xfs_inode	*ip;	/* incore inode pointer */
-	struct xfs_bmbt_irec	*prevp;	/* extent before the new one */
-	struct xfs_bmbt_irec	*gotp;	/* extent after, or delayed */
+	struct xfs_bmbt_irec	prev;	/* extent before the new one */
+	struct xfs_bmbt_irec	got;	/* extent after, or delayed */
 	xfs_extlen_t		alen;	/* i/o length asked/allocated */
 	xfs_extlen_t		total;	/* total blocks needed for xaction */
 	xfs_extlen_t		minlen;	/* minimum allocation size (blocks) */

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

[PATCH 17/25] xfs: move firstblock and bmap freelist cursor into bmalloca structure

[Christoph Hellwig]

[-- Attachment #1: xfs-bmalloca-freelist --]
[-- Type: text/plain, Size: 10091 bytes --]

Rather than passing the freeblock and freelist structure around, embed it
into the bmalloca structure and remove it from the function parameters.

This also enables the minleft parameter to be set only once in
xfs_bmapi_write(), and the freelist cursor directly queried in
xfs_bmapi_allocate to clear it when the lowspace algorithm is activated.
    
    Signed-off-by: Dave Chinner <dchinner@redhat.com>

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:14:37.402071253 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:14:39.985390591 +0200
@@ -2006,9 +2006,9 @@ xfs_bmap_adjacent(
 		XFS_FSB_TO_AGBNO(mp, x) < mp->m_sb.sb_agblocks)
 
 	mp = ap->ip->i_mount;
-	nullfb = ap->firstblock == NULLFSBLOCK;
+	nullfb = *ap->firstblock == NULLFSBLOCK;
 	rt = XFS_IS_REALTIME_INODE(ap->ip) && ap->userdata;
-	fb_agno = nullfb ? NULLAGNUMBER : XFS_FSB_TO_AGNO(mp, ap->firstblock);
+	fb_agno = nullfb ? NULLAGNUMBER : XFS_FSB_TO_AGNO(mp, *ap->firstblock);
 	/*
 	 * If allocating at eof, and there's a previous real block,
 	 * try to use its last block as our starting point.
@@ -2380,8 +2380,8 @@ xfs_bmap_btalloc(
 		ASSERT(!error);
 		ASSERT(ap->alen);
 	}
-	nullfb = ap->firstblock == NULLFSBLOCK;
-	fb_agno = nullfb ? NULLAGNUMBER : XFS_FSB_TO_AGNO(mp, ap->firstblock);
+	nullfb = *ap->firstblock == NULLFSBLOCK;
+	fb_agno = nullfb ? NULLAGNUMBER : XFS_FSB_TO_AGNO(mp, *ap->firstblock);
 	if (nullfb) {
 		if (ap->userdata && xfs_inode_is_filestream(ap->ip)) {
 			ag = xfs_filestream_lookup_ag(ap->ip);
@@ -2391,7 +2391,7 @@ xfs_bmap_btalloc(
 			ap->rval = XFS_INO_TO_FSB(mp, ap->ip->i_ino);
 		}
 	} else
-		ap->rval = ap->firstblock;
+		ap->rval = *ap->firstblock;
 
 	xfs_bmap_adjacent(ap);
 
@@ -2402,7 +2402,7 @@ xfs_bmap_btalloc(
 	if (nullfb || XFS_FSB_TO_AGNO(mp, ap->rval) == fb_agno)
 		;
 	else
-		ap->rval = ap->firstblock;
+		ap->rval = *ap->firstblock;
 	/*
 	 * Normal allocation, done through xfs_alloc_vextent.
 	 */
@@ -2413,13 +2413,13 @@ xfs_bmap_btalloc(
 
 	/* Trim the allocation back to the maximum an AG can fit. */
 	args.maxlen = MIN(ap->alen, XFS_ALLOC_AG_MAX_USABLE(mp));
-	args.firstblock = ap->firstblock;
+	args.firstblock = *ap->firstblock;
 	blen = 0;
 	if (nullfb) {
 		error = xfs_bmap_btalloc_nullfb(ap, &args, &blen);
 		if (error)
 			return error;
-	} else if (ap->low) {
+	} else if (ap->flist->xbf_low) {
 		if (xfs_inode_is_filestream(ap->ip))
 			args.type = XFS_ALLOCTYPE_FIRST_AG;
 		else
@@ -2452,7 +2452,7 @@ xfs_bmap_btalloc(
 	 * is >= the stripe unit and the allocation offset is
 	 * at the end of file.
 	 */
-	if (!ap->low && ap->aeof) {
+	if (!ap->flist->xbf_low && ap->aeof) {
 		if (!ap->off) {
 			args.alignment = mp->m_dalign;
 			atype = args.type;
@@ -2540,12 +2540,25 @@ xfs_bmap_btalloc(
 		args.minleft = 0;
 		if ((error = xfs_alloc_vextent(&args)))
 			return error;
-		ap->low = 1;
+		ap->flist->xbf_low = 1;
 	}
 	if (args.fsbno != NULLFSBLOCK) {
-		ap->firstblock = ap->rval = args.fsbno;
+		/*
+		 * check the allocation happened at the same or higher AG than
+		 * the first block that was allocated.
+		 */
+		ASSERT(*ap->firstblock == NULLFSBLOCK ||
+		       XFS_FSB_TO_AGNO(mp, *ap->firstblock) ==
+		       XFS_FSB_TO_AGNO(mp, args.fsbno) ||
+		       (ap->flist->xbf_low &&
+			XFS_FSB_TO_AGNO(mp, *ap->firstblock) <
+			XFS_FSB_TO_AGNO(mp, args.fsbno)));
+
+		ap->rval = args.fsbno;
+		if (*ap->firstblock == NULLFSBLOCK)
+			*ap->firstblock = args.fsbno;
 		ASSERT(nullfb || fb_agno == args.agno ||
-		       (ap->low && fb_agno < args.agno));
+		       (ap->flist->xbf_low && fb_agno < args.agno));
 		ap->alen = args.len;
 		ap->ip->i_d.di_nblocks += args.len;
 		xfs_trans_log_inode(ap->tp, ap->ip, XFS_ILOG_CORE);
@@ -4590,8 +4603,6 @@ xfs_bmapi_allocate(
 	struct xfs_bmalloca	*bma,
 	xfs_extnum_t		*lastx,
 	struct xfs_btree_cur	**cur,
-	xfs_fsblock_t		*firstblock,
-	struct xfs_bmap_free	*flist,
 	int			flags,
 	int			*nallocs,
 	int			*logflags)
@@ -4641,9 +4652,7 @@ xfs_bmapi_allocate(
 	 */
 	bma->alen = alen;
 	bma->off = aoff;
-	bma->firstblock = *firstblock;
 	bma->minlen = (flags & XFS_BMAPI_CONTIG) ? alen : 1;
-	bma->low = flist->xbf_low;
 	bma->aeof = 0;
 
 	/*
@@ -4665,24 +4674,18 @@ xfs_bmapi_allocate(
 	 * Copy out result fields.
 	 */
 	abno = bma->rval;
-	flist->xbf_low = bma->low;
 	alen = bma->alen;
 	aoff = bma->off;
-	ASSERT(*firstblock == NULLFSBLOCK ||
-	       XFS_FSB_TO_AGNO(mp, *firstblock) ==
-	       XFS_FSB_TO_AGNO(mp, bma->firstblock) ||
-	       (flist->xbf_low &&
-		XFS_FSB_TO_AGNO(mp, *firstblock) <
-			XFS_FSB_TO_AGNO(mp, bma->firstblock)));
-	*firstblock = bma->firstblock;
+	if (bma->flist->xbf_low)
+		bma->minleft = 0;
 	if (*cur)
-		(*cur)->bc_private.b.firstblock = *firstblock;
+		(*cur)->bc_private.b.firstblock = *bma->firstblock;
 	if (abno == NULLFSBLOCK)
 		return 0;
 	if ((ifp->if_flags & XFS_IFBROOT) && !*cur) {
 		(*cur) = xfs_bmbt_init_cursor(mp, bma->tp, bma->ip, whichfork);
-		(*cur)->bc_private.b.firstblock = *firstblock;
-		(*cur)->bc_private.b.flist = flist;
+		(*cur)->bc_private.b.firstblock = *bma->firstblock;
+		(*cur)->bc_private.b.flist = bma->flist;
 	}
 	/*
 	 * Bump the number of extents we've allocated
@@ -4709,11 +4712,12 @@ xfs_bmapi_allocate(
 
 	if (bma->wasdel) {
 		error = xfs_bmap_add_extent_delay_real(bma->tp, bma->ip, lastx,
-				cur, &bma->got, firstblock, flist, logflags);
+				cur, &bma->got, bma->firstblock, bma->flist,
+				logflags);
 	} else {
 		error = xfs_bmap_add_extent_hole_real(bma->tp, bma->ip, lastx,
-				cur, &bma->got, firstblock, flist, logflags,
-				whichfork);
+				cur, &bma->got, bma->firstblock, bma->flist,
+				logflags, whichfork);
 	}
 
 	if (error)
@@ -4740,8 +4744,6 @@ xfs_bmapi_convert_unwritten(
 	xfs_filblks_t		len,
 	xfs_extnum_t		*lastx,
 	struct xfs_btree_cur	**cur,
-	xfs_fsblock_t		*firstblock,
-	struct xfs_bmap_free	*flist,
 	int			flags,
 	int			*logflags)
 {
@@ -4770,14 +4772,14 @@ xfs_bmapi_convert_unwritten(
 	if ((ifp->if_flags & XFS_IFBROOT) && !*cur) {
 		*cur = xfs_bmbt_init_cursor(bma->ip->i_mount, bma->tp,
 					bma->ip, whichfork);
-		(*cur)->bc_private.b.firstblock = *firstblock;
-		(*cur)->bc_private.b.flist = flist;
+		(*cur)->bc_private.b.firstblock = *bma->firstblock;
+		(*cur)->bc_private.b.flist = bma->flist;
 	}
 	mval->br_state = (mval->br_state == XFS_EXT_UNWRITTEN)
 				? XFS_EXT_NORM : XFS_EXT_UNWRITTEN;
 
 	error = xfs_bmap_add_extent_unwritten_real(bma->tp, bma->ip, lastx,
-			cur, mval, firstblock, flist, logflags);
+			cur, mval, bma->firstblock, bma->flist, logflags);
 	if (error)
 		return error;
 
@@ -4832,7 +4834,6 @@ xfs_bmapi_write(
 	xfs_ifork_t		*ifp;		/* inode fork pointer */
 	xfs_extnum_t		lastx;		/* last useful extent number */
 	int			logflags;	/* flags for transaction logging */
-	xfs_extlen_t		minleft;	/* min blocks left after allocation */
 	int			n;		/* current extent index */
 	int			nallocs;	/* number of extents alloc'd */
 	xfs_fileoff_t		obno;		/* old block number (offset) */
@@ -4894,11 +4895,11 @@ xfs_bmapi_write(
 
 	if (*firstblock == NULLFSBLOCK) {
 		if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_BTREE)
-			minleft = be16_to_cpu(ifp->if_broot->bb_level) + 1;
+			bma.minleft = be16_to_cpu(ifp->if_broot->bb_level) + 1;
 		else
-			minleft = 1;
+			bma.minleft = 1;
 	} else {
-		minleft = 0;
+		bma.minleft = 0;
 	}
 
 	if (!(ifp->if_flags & XFS_IFEXTENTS)) {
@@ -4917,6 +4918,8 @@ xfs_bmapi_write(
 	bma.ip = ip;
 	bma.total = total;
 	bma.userdata = 0;
+	bma.flist = flist;
+	bma.firstblock = firstblock;
 
 	while (bno < end && n < *nmap) {
 		inhole = eof || bma.got.br_startoff > bno;
@@ -4932,11 +4935,9 @@ xfs_bmapi_write(
 			bma.wasdel = wasdelay;
 			bma.alen = len;
 			bma.off = bno;
-			bma.minleft = minleft;
 
-			error = xfs_bmapi_allocate(&bma, &lastx, &cur,
-					firstblock, flist, flags, &nallocs,
-					&tmp_logflags);
+			error = xfs_bmapi_allocate(&bma, &lastx, &cur, flags,
+						   &nallocs, &tmp_logflags);
 			if (error == ENOSPC || error == EDQUOT) {
 				if (n == 0) {
 					*nmap = 0;
@@ -4948,8 +4949,6 @@ xfs_bmapi_write(
 			logflags |= tmp_logflags;
 			if (error)
 				goto error0;
-			if (flist && flist->xbf_low)
-				minleft = 0;
 			if (bma.rval == NULLFSBLOCK)
 				break;
 		}
@@ -4960,8 +4959,7 @@ xfs_bmapi_write(
 
 		/* Execute unwritten extent conversion if necessary */
 		error = xfs_bmapi_convert_unwritten(&bma, mval, len, &lastx,
-						    &cur, firstblock, flist,
-						    flags, &tmp_logflags);
+						    &cur, flags, &tmp_logflags);
 		logflags |= tmp_logflags;
 		if (error == EAGAIN)
 			continue;
Index: xfs/fs/xfs/xfs_bmap.h
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.h	2011-08-23 21:14:37.402071253 +0200
+++ xfs/fs/xfs/xfs_bmap.h	2011-08-23 21:14:39.988723907 +0200
@@ -109,7 +109,8 @@ static inline void xfs_bmap_init(xfs_bma
  * Argument structure for xfs_bmap_alloc.
  */
 typedef struct xfs_bmalloca {
-	xfs_fsblock_t		firstblock; /* i/o first block allocated */
+	xfs_fsblock_t		*firstblock; /* i/o first block allocated */
+	struct xfs_bmap_free	*flist;	/* bmap freelist */
 	xfs_fsblock_t		rval;	/* starting block of new extent */
 	xfs_fileoff_t		off;	/* offset in file filling in */
 	struct xfs_trans	*tp;	/* transaction pointer */
Index: xfs/fs/xfs/xfs_filestream.c
===================================================================
--- xfs.orig/fs/xfs/xfs_filestream.c	2011-08-09 09:55:48.579356195 +0200
+++ xfs/fs/xfs/xfs_filestream.c	2011-08-23 21:14:39.992057222 +0200
@@ -761,7 +761,7 @@ xfs_filestream_new_ag(
 	 */
 	ag = (ag == NULLAGNUMBER) ? 0 : (ag + 1) % mp->m_sb.sb_agcount;
 	flags = (ap->userdata ? XFS_PICK_USERDATA : 0) |
-	        (ap->low ? XFS_PICK_LOWSPACE : 0);
+	        (ap->flist->xbf_low ? XFS_PICK_LOWSPACE : 0);
 
 	err = _xfs_filestream_pick_ag(mp, ag, agp, flags, minlen);
 	if (err || *agp == NULLAGNUMBER)

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

[PATCH 18/25] xfs: move allocation ranges inode bmalloca structure

[Christoph Hellwig]

[-- Attachment #1: xfs-bmalloca-ranges --]
[-- Type: text/plain, Size: 16675 bytes --]

Most of the allocation functions take the offset length and block
number as parameters for allocation. often they are modified so are
passed by reference.  Make the version in the bmalloca structure the
primary version and stop passing them around individually.

Signed-off-by: Dave Chinner <dchinner@redhat.com>

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:14:39.985390591 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:14:41.285383547 +0200
@@ -2017,15 +2017,15 @@ xfs_bmap_adjacent(
 	    !isnullstartblock(ap->prev.br_startblock) &&
 	    ISVALID(ap->prev.br_startblock + ap->prev.br_blockcount,
 		    ap->prev.br_startblock)) {
-		ap->rval = ap->prev.br_startblock + ap->prev.br_blockcount;
+		ap->blkno = ap->prev.br_startblock + ap->prev.br_blockcount;
 		/*
 		 * Adjust for the gap between prevp and us.
 		 */
-		adjust = ap->off -
+		adjust = ap->offset -
 			(ap->prev.br_startoff + ap->prev.br_blockcount);
 		if (adjust &&
-		    ISVALID(ap->rval + adjust, ap->prev.br_startblock))
-			ap->rval += adjust;
+		    ISVALID(ap->blkno + adjust, ap->prev.br_startblock))
+			ap->blkno += adjust;
 	}
 	/*
 	 * If not at eof, then compare the two neighbor blocks.
@@ -2050,7 +2050,7 @@ xfs_bmap_adjacent(
 			/*
 			 * Calculate gap to end of previous block.
 			 */
-			adjust = prevdiff = ap->off -
+			adjust = prevdiff = ap->offset -
 				(ap->prev.br_startoff +
 				 ap->prev.br_blockcount);
 			/*
@@ -2061,7 +2061,7 @@ xfs_bmap_adjacent(
 			 * allocating, or using it gives us an invalid block
 			 * number, then just use the end of the previous block.
 			 */
-			if (prevdiff <= XFS_ALLOC_GAP_UNITS * ap->alen &&
+			if (prevdiff <= XFS_ALLOC_GAP_UNITS * ap->length &&
 			    ISVALID(prevbno + prevdiff,
 				    ap->prev.br_startblock))
 				prevbno += adjust;
@@ -2088,7 +2088,7 @@ xfs_bmap_adjacent(
 			/*
 			 * Calculate gap to start of next block.
 			 */
-			adjust = gotdiff = ap->got.br_startoff - ap->off;
+			adjust = gotdiff = ap->got.br_startoff - ap->offset;
 			/*
 			 * Figure the startblock based on the next block's
 			 * start and the gap size.
@@ -2101,12 +2101,12 @@ xfs_bmap_adjacent(
 			 * number, then just use the start of the next block
 			 * offset by our length.
 			 */
-			if (gotdiff <= XFS_ALLOC_GAP_UNITS * ap->alen &&
+			if (gotdiff <= XFS_ALLOC_GAP_UNITS * ap->length &&
 			    ISVALID(gotbno - gotdiff, gotbno))
 				gotbno -= adjust;
-			else if (ISVALID(gotbno - ap->alen, gotbno)) {
-				gotbno -= ap->alen;
-				gotdiff += adjust - ap->alen;
+			else if (ISVALID(gotbno - ap->length, gotbno)) {
+				gotbno -= ap->length;
+				gotdiff += adjust - ap->length;
 			} else
 				gotdiff += adjust;
 			/*
@@ -2124,14 +2124,14 @@ xfs_bmap_adjacent(
 			gotbno = NULLFSBLOCK;
 		/*
 		 * If both valid, pick the better one, else the only good
-		 * one, else ap->rval is already set (to 0 or the inode block).
+		 * one, else ap->blkno is already set (to 0 or the inode block).
 		 */
 		if (prevbno != NULLFSBLOCK && gotbno != NULLFSBLOCK)
-			ap->rval = prevdiff <= gotdiff ? prevbno : gotbno;
+			ap->blkno = prevdiff <= gotdiff ? prevbno : gotbno;
 		else if (prevbno != NULLFSBLOCK)
-			ap->rval = prevbno;
+			ap->blkno = prevbno;
 		else if (gotbno != NULLFSBLOCK)
-			ap->rval = gotbno;
+			ap->blkno = gotbno;
 	}
 #undef ISVALID
 }
@@ -2153,22 +2153,22 @@ xfs_bmap_rtalloc(
 	prod = align / mp->m_sb.sb_rextsize;
 	error = xfs_bmap_extsize_align(mp, &ap->got, &ap->prev,
 					align, 1, ap->eof, 0,
-					ap->conv, &ap->off, &ap->alen);
+					ap->conv, &ap->offset, &ap->length);
 	if (error)
 		return error;
-	ASSERT(ap->alen);
-	ASSERT(ap->alen % mp->m_sb.sb_rextsize == 0);
+	ASSERT(ap->length);
+	ASSERT(ap->length % mp->m_sb.sb_rextsize == 0);
 
 	/*
 	 * If the offset & length are not perfectly aligned
 	 * then kill prod, it will just get us in trouble.
 	 */
-	if (do_mod(ap->off, align) || ap->alen % align)
+	if (do_mod(ap->offset, align) || ap->length % align)
 		prod = 1;
 	/*
 	 * Set ralen to be the actual requested length in rtextents.
 	 */
-	ralen = ap->alen / mp->m_sb.sb_rextsize;
+	ralen = ap->length / mp->m_sb.sb_rextsize;
 	/*
 	 * If the old value was close enough to MAXEXTLEN that
 	 * we rounded up to it, cut it back so it's valid again.
@@ -2189,15 +2189,15 @@ xfs_bmap_rtalloc(
 	 * If it's an allocation to an empty file at offset 0,
 	 * pick an extent that will space things out in the rt area.
 	 */
-	if (ap->eof && ap->off == 0) {
+	if (ap->eof && ap->offset == 0) {
 		xfs_rtblock_t uninitialized_var(rtx); /* realtime extent no */
 
 		error = xfs_rtpick_extent(mp, ap->tp, ralen, &rtx);
 		if (error)
 			return error;
-		ap->rval = rtx * mp->m_sb.sb_rextsize;
+		ap->blkno = rtx * mp->m_sb.sb_rextsize;
 	} else {
-		ap->rval = 0;
+		ap->blkno = 0;
 	}
 
 	xfs_bmap_adjacent(ap);
@@ -2205,23 +2205,23 @@ xfs_bmap_rtalloc(
 	/*
 	 * Realtime allocation, done through xfs_rtallocate_extent.
 	 */
-	atype = ap->rval == 0 ?  XFS_ALLOCTYPE_ANY_AG : XFS_ALLOCTYPE_NEAR_BNO;
-	do_div(ap->rval, mp->m_sb.sb_rextsize);
-	rtb = ap->rval;
-	ap->alen = ralen;
-	if ((error = xfs_rtallocate_extent(ap->tp, ap->rval, 1, ap->alen,
+	atype = ap->blkno == 0 ?  XFS_ALLOCTYPE_ANY_AG : XFS_ALLOCTYPE_NEAR_BNO;
+	do_div(ap->blkno, mp->m_sb.sb_rextsize);
+	rtb = ap->blkno;
+	ap->length = ralen;
+	if ((error = xfs_rtallocate_extent(ap->tp, ap->blkno, 1, ap->length,
 				&ralen, atype, ap->wasdel, prod, &rtb)))
 		return error;
 	if (rtb == NULLFSBLOCK && prod > 1 &&
-	    (error = xfs_rtallocate_extent(ap->tp, ap->rval, 1,
-					   ap->alen, &ralen, atype,
+	    (error = xfs_rtallocate_extent(ap->tp, ap->blkno, 1,
+					   ap->length, &ralen, atype,
 					   ap->wasdel, 1, &rtb)))
 		return error;
-	ap->rval = rtb;
-	if (ap->rval != NULLFSBLOCK) {
-		ap->rval *= mp->m_sb.sb_rextsize;
+	ap->blkno = rtb;
+	if (ap->blkno != NULLFSBLOCK) {
+		ap->blkno *= mp->m_sb.sb_rextsize;
 		ralen *= mp->m_sb.sb_rextsize;
-		ap->alen = ralen;
+		ap->length = ralen;
 		ap->ip->i_d.di_nblocks += ralen;
 		xfs_trans_log_inode(ap->tp, ap->ip, XFS_ILOG_CORE);
 		if (ap->wasdel)
@@ -2234,7 +2234,7 @@ xfs_bmap_rtalloc(
 			ap->wasdel ? XFS_TRANS_DQ_DELRTBCOUNT :
 					XFS_TRANS_DQ_RTBCOUNT, (long) ralen);
 	} else {
-		ap->alen = 0;
+		ap->length = 0;
 	}
 	return 0;
 }
@@ -2349,7 +2349,7 @@ xfs_bmap_btalloc_nullfb(
 	 * AG as the stream may have moved.
 	 */
 	if (xfs_inode_is_filestream(ap->ip))
-		ap->rval = args->fsbno = XFS_AGB_TO_FSB(mp, ag, 0);
+		ap->blkno = args->fsbno = XFS_AGB_TO_FSB(mp, ag, 0);
 
 	return 0;
 }
@@ -2376,9 +2376,9 @@ xfs_bmap_btalloc(
 	if (unlikely(align)) {
 		error = xfs_bmap_extsize_align(mp, &ap->got, &ap->prev,
 						align, 0, ap->eof, 0, ap->conv,
-						&ap->off, &ap->alen);
+						&ap->offset, &ap->length);
 		ASSERT(!error);
-		ASSERT(ap->alen);
+		ASSERT(ap->length);
 	}
 	nullfb = *ap->firstblock == NULLFSBLOCK;
 	fb_agno = nullfb ? NULLAGNUMBER : XFS_FSB_TO_AGNO(mp, *ap->firstblock);
@@ -2386,33 +2386,33 @@ xfs_bmap_btalloc(
 		if (ap->userdata && xfs_inode_is_filestream(ap->ip)) {
 			ag = xfs_filestream_lookup_ag(ap->ip);
 			ag = (ag != NULLAGNUMBER) ? ag : 0;
-			ap->rval = XFS_AGB_TO_FSB(mp, ag, 0);
+			ap->blkno = XFS_AGB_TO_FSB(mp, ag, 0);
 		} else {
-			ap->rval = XFS_INO_TO_FSB(mp, ap->ip->i_ino);
+			ap->blkno = XFS_INO_TO_FSB(mp, ap->ip->i_ino);
 		}
 	} else
-		ap->rval = *ap->firstblock;
+		ap->blkno = *ap->firstblock;
 
 	xfs_bmap_adjacent(ap);
 
 	/*
-	 * If allowed, use ap->rval; otherwise must use firstblock since
+	 * If allowed, use ap->blkno; otherwise must use firstblock since
 	 * it's in the right allocation group.
 	 */
-	if (nullfb || XFS_FSB_TO_AGNO(mp, ap->rval) == fb_agno)
+	if (nullfb || XFS_FSB_TO_AGNO(mp, ap->blkno) == fb_agno)
 		;
 	else
-		ap->rval = *ap->firstblock;
+		ap->blkno = *ap->firstblock;
 	/*
 	 * Normal allocation, done through xfs_alloc_vextent.
 	 */
 	tryagain = isaligned = 0;
 	args.tp = ap->tp;
 	args.mp = mp;
-	args.fsbno = ap->rval;
+	args.fsbno = ap->blkno;
 
 	/* Trim the allocation back to the maximum an AG can fit. */
-	args.maxlen = MIN(ap->alen, XFS_ALLOC_AG_MAX_USABLE(mp));
+	args.maxlen = MIN(ap->length, XFS_ALLOC_AG_MAX_USABLE(mp));
 	args.firstblock = *ap->firstblock;
 	blen = 0;
 	if (nullfb) {
@@ -2433,14 +2433,14 @@ xfs_bmap_btalloc(
 	/* apply extent size hints if obtained earlier */
 	if (unlikely(align)) {
 		args.prod = align;
-		if ((args.mod = (xfs_extlen_t)do_mod(ap->off, args.prod)))
+		if ((args.mod = (xfs_extlen_t)do_mod(ap->offset, args.prod)))
 			args.mod = (xfs_extlen_t)(args.prod - args.mod);
 	} else if (mp->m_sb.sb_blocksize >= PAGE_CACHE_SIZE) {
 		args.prod = 1;
 		args.mod = 0;
 	} else {
 		args.prod = PAGE_CACHE_SIZE >> mp->m_sb.sb_blocklog;
-		if ((args.mod = (xfs_extlen_t)(do_mod(ap->off, args.prod))))
+		if ((args.mod = (xfs_extlen_t)(do_mod(ap->offset, args.prod))))
 			args.mod = (xfs_extlen_t)(args.prod - args.mod);
 	}
 	/*
@@ -2453,7 +2453,7 @@ xfs_bmap_btalloc(
 	 * at the end of file.
 	 */
 	if (!ap->flist->xbf_low && ap->aeof) {
-		if (!ap->off) {
+		if (!ap->offset) {
 			args.alignment = mp->m_dalign;
 			atype = args.type;
 			isaligned = 1;
@@ -2506,7 +2506,7 @@ xfs_bmap_btalloc(
 		 * turned on.
 		 */
 		args.type = atype;
-		args.fsbno = ap->rval;
+		args.fsbno = ap->blkno;
 		args.alignment = mp->m_dalign;
 		args.minlen = nextminlen;
 		args.minalignslop = 0;
@@ -2520,7 +2520,7 @@ xfs_bmap_btalloc(
 		 * try again.
 		 */
 		args.type = atype;
-		args.fsbno = ap->rval;
+		args.fsbno = ap->blkno;
 		args.alignment = 0;
 		if ((error = xfs_alloc_vextent(&args)))
 			return error;
@@ -2529,7 +2529,7 @@ xfs_bmap_btalloc(
 	    args.minlen > ap->minlen) {
 		args.minlen = ap->minlen;
 		args.type = XFS_ALLOCTYPE_START_BNO;
-		args.fsbno = ap->rval;
+		args.fsbno = ap->blkno;
 		if ((error = xfs_alloc_vextent(&args)))
 			return error;
 	}
@@ -2554,12 +2554,12 @@ xfs_bmap_btalloc(
 			XFS_FSB_TO_AGNO(mp, *ap->firstblock) <
 			XFS_FSB_TO_AGNO(mp, args.fsbno)));
 
-		ap->rval = args.fsbno;
+		ap->blkno = args.fsbno;
 		if (*ap->firstblock == NULLFSBLOCK)
 			*ap->firstblock = args.fsbno;
 		ASSERT(nullfb || fb_agno == args.agno ||
 		       (ap->flist->xbf_low && fb_agno < args.agno));
-		ap->alen = args.len;
+		ap->length = args.len;
 		ap->ip->i_d.di_nblocks += args.len;
 		xfs_trans_log_inode(ap->tp, ap->ip, XFS_ILOG_CORE);
 		if (ap->wasdel)
@@ -2573,8 +2573,8 @@ xfs_bmap_btalloc(
 					XFS_TRANS_DQ_BCOUNT,
 			(long) args.len);
 	} else {
-		ap->rval = NULLFSBLOCK;
-		ap->alen = 0;
+		ap->blkno = NULLFSBLOCK;
+		ap->length = 0;
 	}
 	return 0;
 }
@@ -3899,10 +3899,10 @@ xfs_bmap_isaeof(
 	 * Check we are allocating in the last extent (for delayed allocations)
 	 * or past the last extent for non-delayed allocations.
 	 */
-	bma->aeof = (bma->off >= rec.br_startoff &&
-		     bma->off < rec.br_startoff + rec.br_blockcount &&
+	bma->aeof = (bma->offset >= rec.br_startoff &&
+		     bma->offset < rec.br_startoff + rec.br_blockcount &&
 		     isnullstartblock(rec.br_startblock)) ||
-		    bma->off >= rec.br_startoff + rec.br_blockcount;
+		    bma->offset >= rec.br_startoff + rec.br_blockcount;
 	return 0;
 }
 
@@ -4611,9 +4611,6 @@ xfs_bmapi_allocate(
 	int			whichfork = (flags & XFS_BMAPI_ATTRFORK) ?
 						XFS_ATTR_FORK : XFS_DATA_FORK;
 	struct xfs_ifork	*ifp = XFS_IFORK_PTR(bma->ip, whichfork);
-	xfs_fsblock_t		abno;
-	xfs_extlen_t		alen;
-	xfs_fileoff_t		aoff;
 	int			error;
 	int			rt;
 
@@ -4624,18 +4621,17 @@ xfs_bmapi_allocate(
 	 * for in this bmap call but that wouldn't be as good.
 	 */
 	if (bma->wasdel) {
-		alen = (xfs_extlen_t)bma->got.br_blockcount;
-		aoff = bma->got.br_startoff;
+		bma->length = bma->got.br_blockcount;
+		bma->offset = bma->got.br_startoff;
 		if (*lastx != NULLEXTNUM && *lastx) {
 			xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx - 1),
 					 &bma->prev);
 		}
 	} else {
-		alen = (xfs_extlen_t)XFS_FILBLKS_MIN(bma->alen, MAXEXTLEN);
+		bma->length = XFS_EXTLEN_MIN(bma->length, MAXEXTLEN);
 		if (!bma->eof)
-			alen = (xfs_extlen_t)XFS_FILBLKS_MIN(alen,
-					bma->got.br_startoff - bma->off);
-		aoff = bma->off;
+			bma->length = XFS_EXTLEN_MIN(bma->length,
+					bma->got.br_startoff - bma->offset);
 	}
 
 	/*
@@ -4643,23 +4639,20 @@ xfs_bmapi_allocate(
 	 * user data.
 	 */
 	if (!(flags & XFS_BMAPI_METADATA)) {
-		bma->userdata = (aoff == 0) ?
+		bma->userdata = (bma->offset == 0) ?
 			XFS_ALLOC_INITIAL_USER_DATA : XFS_ALLOC_USERDATA;
 	}
 
 	/*
 	 * Fill in changeable bma fields.
 	 */
-	bma->alen = alen;
-	bma->off = aoff;
-	bma->minlen = (flags & XFS_BMAPI_CONTIG) ? alen : 1;
-	bma->aeof = 0;
+	bma->minlen = (flags & XFS_BMAPI_CONTIG) ? bma->length : 1;
 
 	/*
 	 * Only want to do the alignment at the eof if it is userdata and
 	 * allocation length is larger than a stripe unit.
 	 */
-	if (mp->m_dalign && alen >= mp->m_dalign &&
+	if (mp->m_dalign && bma->length >= mp->m_dalign &&
 	    !(flags & XFS_BMAPI_METADATA) && whichfork == XFS_DATA_FORK) {
 		error = xfs_bmap_isaeof(bma, whichfork);
 		if (error)
@@ -4670,17 +4663,11 @@ xfs_bmapi_allocate(
 	if (error)
 		return error;
 
-	/*
-	 * Copy out result fields.
-	 */
-	abno = bma->rval;
-	alen = bma->alen;
-	aoff = bma->off;
 	if (bma->flist->xbf_low)
 		bma->minleft = 0;
 	if (*cur)
 		(*cur)->bc_private.b.firstblock = *bma->firstblock;
-	if (abno == NULLFSBLOCK)
+	if (bma->blkno == NULLFSBLOCK)
 		return 0;
 	if ((ifp->if_flags & XFS_IFBROOT) && !*cur) {
 		(*cur) = xfs_bmbt_init_cursor(mp, bma->tp, bma->ip, whichfork);
@@ -4697,9 +4684,9 @@ xfs_bmapi_allocate(
 		(*cur)->bc_private.b.flags =
 			bma->wasdel ? XFS_BTCUR_BPRV_WASDEL : 0;
 
-	bma->got.br_startoff = aoff;
-	bma->got.br_startblock = abno;
-	bma->got.br_blockcount = alen;
+	bma->got.br_startoff = bma->offset;
+	bma->got.br_startblock = bma->blkno;
+	bma->got.br_blockcount = bma->length;
 	bma->got.br_state = XFS_EXT_NORM;
 
 	/*
@@ -4730,8 +4717,9 @@ xfs_bmapi_allocate(
 	 */
 	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx), &bma->got);
 
-	ASSERT(bma->got.br_startoff <= aoff);
-	ASSERT(bma->got.br_startoff + bma->got.br_blockcount >= aoff + alen);
+	ASSERT(bma->got.br_startoff <= bma->offset);
+	ASSERT(bma->got.br_startoff + bma->got.br_blockcount >=
+	       bma->offset + bma->length);
 	ASSERT(bma->got.br_state == XFS_EXT_NORM ||
 	       bma->got.br_state == XFS_EXT_UNWRITTEN);
 	return 0;
@@ -4933,8 +4921,8 @@ xfs_bmapi_write(
 			bma.eof = eof;
 			bma.conv = !!(flags & XFS_BMAPI_CONVERT);
 			bma.wasdel = wasdelay;
-			bma.alen = len;
-			bma.off = bno;
+			bma.length = len;
+			bma.offset = bno;
 
 			error = xfs_bmapi_allocate(&bma, &lastx, &cur, flags,
 						   &nallocs, &tmp_logflags);
@@ -4949,7 +4937,7 @@ xfs_bmapi_write(
 			logflags |= tmp_logflags;
 			if (error)
 				goto error0;
-			if (bma.rval == NULLFSBLOCK)
+			if (bma.blkno == NULLFSBLOCK)
 				break;
 		}
 
Index: xfs/fs/xfs/xfs_bmap.h
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.h	2011-08-23 21:14:39.988723907 +0200
+++ xfs/fs/xfs/xfs_bmap.h	2011-08-23 21:14:41.288716862 +0200
@@ -111,13 +111,15 @@ static inline void xfs_bmap_init(xfs_bma
 typedef struct xfs_bmalloca {
 	xfs_fsblock_t		*firstblock; /* i/o first block allocated */
 	struct xfs_bmap_free	*flist;	/* bmap freelist */
-	xfs_fsblock_t		rval;	/* starting block of new extent */
-	xfs_fileoff_t		off;	/* offset in file filling in */
 	struct xfs_trans	*tp;	/* transaction pointer */
 	struct xfs_inode	*ip;	/* incore inode pointer */
 	struct xfs_bmbt_irec	prev;	/* extent before the new one */
 	struct xfs_bmbt_irec	got;	/* extent after, or delayed */
-	xfs_extlen_t		alen;	/* i/o length asked/allocated */
+
+	xfs_fileoff_t		offset;	/* offset in file filling in */
+	xfs_extlen_t		length;	/* i/o length asked/allocated */
+	xfs_fsblock_t		blkno;	/* starting block of new extent */
+
 	xfs_extlen_t		total;	/* total blocks needed for xaction */
 	xfs_extlen_t		minlen;	/* minimum allocation size (blocks) */
 	xfs_extlen_t		minleft; /* amount must be left after alloc */
Index: xfs/fs/xfs/xfs_filestream.c
===================================================================
--- xfs.orig/fs/xfs/xfs_filestream.c	2011-08-23 21:14:39.992057222 +0200
+++ xfs/fs/xfs/xfs_filestream.c	2011-08-23 21:14:41.288716862 +0200
@@ -682,7 +682,7 @@ xfs_filestream_new_ag(
 	ip = ap->ip;
 	mp = ip->i_mount;
 	cache = mp->m_filestream;
-	minlen = ap->alen;
+	minlen = ap->length;
 	*agp = NULLAGNUMBER;
 
 	/*

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

[PATCH 19/25] xfs: move btree cursor into bmalloca

[Christoph Hellwig]

[-- Attachment #1: xfs-bmalloca-cursor --]
[-- Type: text/plain, Size: 6291 bytes --]

Signed-off-by: Dave Chinner <dchinner@redhat.com>

Index: xfs/fs/xfs/xfs_bmap.h
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.h	2011-08-23 21:14:41.288716862 +0200
+++ xfs/fs/xfs/xfs_bmap.h	2011-08-23 21:14:42.535376777 +0200
@@ -120,6 +120,8 @@ typedef struct xfs_bmalloca {
 	xfs_extlen_t		length;	/* i/o length asked/allocated */
 	xfs_fsblock_t		blkno;	/* starting block of new extent */
 
+	struct xfs_btree_cur	*cur;	/* btree cursor */
+
 	xfs_extlen_t		total;	/* total blocks needed for xaction */
 	xfs_extlen_t		minlen;	/* minimum allocation size (blocks) */
 	xfs_extlen_t		minleft; /* amount must be left after alloc */
Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:14:41.285383547 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:14:42.542043407 +0200
@@ -4602,7 +4602,6 @@ STATIC int
 xfs_bmapi_allocate(
 	struct xfs_bmalloca	*bma,
 	xfs_extnum_t		*lastx,
-	struct xfs_btree_cur	**cur,
 	int			flags,
 	int			*nallocs,
 	int			*logflags)
@@ -4665,14 +4664,14 @@ xfs_bmapi_allocate(
 
 	if (bma->flist->xbf_low)
 		bma->minleft = 0;
-	if (*cur)
-		(*cur)->bc_private.b.firstblock = *bma->firstblock;
+	if (bma->cur)
+		bma->cur->bc_private.b.firstblock = *bma->firstblock;
 	if (bma->blkno == NULLFSBLOCK)
 		return 0;
-	if ((ifp->if_flags & XFS_IFBROOT) && !*cur) {
-		(*cur) = xfs_bmbt_init_cursor(mp, bma->tp, bma->ip, whichfork);
-		(*cur)->bc_private.b.firstblock = *bma->firstblock;
-		(*cur)->bc_private.b.flist = bma->flist;
+	if ((ifp->if_flags & XFS_IFBROOT) && !bma->cur) {
+		bma->cur = xfs_bmbt_init_cursor(mp, bma->tp, bma->ip, whichfork);
+		bma->cur->bc_private.b.firstblock = *bma->firstblock;
+		bma->cur->bc_private.b.flist = bma->flist;
 	}
 	/*
 	 * Bump the number of extents we've allocated
@@ -4680,8 +4679,8 @@ xfs_bmapi_allocate(
 	 */
 	(*nallocs)++;
 
-	if (*cur)
-		(*cur)->bc_private.b.flags =
+	if (bma->cur)
+		bma->cur->bc_private.b.flags =
 			bma->wasdel ? XFS_BTCUR_BPRV_WASDEL : 0;
 
 	bma->got.br_startoff = bma->offset;
@@ -4699,12 +4698,12 @@ xfs_bmapi_allocate(
 
 	if (bma->wasdel) {
 		error = xfs_bmap_add_extent_delay_real(bma->tp, bma->ip, lastx,
-				cur, &bma->got, bma->firstblock, bma->flist,
-				logflags);
+				&bma->cur, &bma->got, bma->firstblock,
+				bma->flist, logflags);
 	} else {
 		error = xfs_bmap_add_extent_hole_real(bma->tp, bma->ip, lastx,
-				cur, &bma->got, bma->firstblock, bma->flist,
-				logflags, whichfork);
+				&bma->cur, &bma->got, bma->firstblock,
+				bma->flist, logflags, whichfork);
 	}
 
 	if (error)
@@ -4731,7 +4730,6 @@ xfs_bmapi_convert_unwritten(
 	struct xfs_bmbt_irec	*mval,
 	xfs_filblks_t		len,
 	xfs_extnum_t		*lastx,
-	struct xfs_btree_cur	**cur,
 	int			flags,
 	int			*logflags)
 {
@@ -4757,17 +4755,17 @@ xfs_bmapi_convert_unwritten(
 	 * Modify (by adding) the state flag, if writing.
 	 */
 	ASSERT(mval->br_blockcount <= len);
-	if ((ifp->if_flags & XFS_IFBROOT) && !*cur) {
-		*cur = xfs_bmbt_init_cursor(bma->ip->i_mount, bma->tp,
+	if ((ifp->if_flags & XFS_IFBROOT) && !bma->cur) {
+		bma->cur = xfs_bmbt_init_cursor(bma->ip->i_mount, bma->tp,
 					bma->ip, whichfork);
-		(*cur)->bc_private.b.firstblock = *bma->firstblock;
-		(*cur)->bc_private.b.flist = bma->flist;
+		bma->cur->bc_private.b.firstblock = *bma->firstblock;
+		bma->cur->bc_private.b.flist = bma->flist;
 	}
 	mval->br_state = (mval->br_state == XFS_EXT_UNWRITTEN)
 				? XFS_EXT_NORM : XFS_EXT_UNWRITTEN;
 
 	error = xfs_bmap_add_extent_unwritten_real(bma->tp, bma->ip, lastx,
-			cur, mval, bma->firstblock, bma->flist, logflags);
+			&bma->cur, mval, bma->firstblock, bma->flist, logflags);
 	if (error)
 		return error;
 
@@ -4815,7 +4813,6 @@ xfs_bmapi_write(
 {
 	struct xfs_mount	*mp = ip->i_mount;
 	struct xfs_bmalloca	bma = { 0 };	/* args for xfs_bmap_alloc */
-	struct xfs_btree_cur	*cur;		/* bmap btree cursor */
 	xfs_fileoff_t		end;		/* end of mapped file region */
 	int			eof;		/* after the end of extents */
 	int			error;		/* error return */
@@ -4872,7 +4869,6 @@ xfs_bmapi_write(
 
 	logflags = 0;
 	nallocs = 0;
-	cur = NULL;
 
 	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_LOCAL) {
 		error = xfs_bmap_local_to_extents(tp, ip, firstblock, total,
@@ -4924,12 +4920,12 @@ xfs_bmapi_write(
 			bma.length = len;
 			bma.offset = bno;
 
-			error = xfs_bmapi_allocate(&bma, &lastx, &cur, flags,
+			error = xfs_bmapi_allocate(&bma, &lastx, flags,
 						   &nallocs, &tmp_logflags);
 			if (error == ENOSPC || error == EDQUOT) {
 				if (n == 0) {
 					*nmap = 0;
-					ASSERT(cur == NULL);
+					ASSERT(bma.cur == NULL);
 					return error;
 				}
 			}
@@ -4947,7 +4943,7 @@ xfs_bmapi_write(
 
 		/* Execute unwritten extent conversion if necessary */
 		error = xfs_bmapi_convert_unwritten(&bma, mval, len, &lastx,
-						    &cur, flags, &tmp_logflags);
+						    flags, &tmp_logflags);
 		logflags |= tmp_logflags;
 		if (error == EAGAIN)
 			continue;
@@ -4979,8 +4975,8 @@ xfs_bmapi_write(
 	 */
 	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_BTREE &&
 	    XFS_IFORK_NEXTENTS(ip, whichfork) <= ifp->if_ext_max) {
-		ASSERT(cur);
-		error = xfs_bmap_btree_to_extents(tp, ip, cur,
+		ASSERT(bma.cur);
+		error = xfs_bmap_btree_to_extents(tp, ip, bma.cur,
 			&tmp_logflags, whichfork);
 		logflags |= tmp_logflags;
 		if (error)
@@ -5010,19 +5006,19 @@ error0:
 	if (logflags)
 		xfs_trans_log_inode(tp, ip, logflags);
 
-	if (cur) {
+	if (bma.cur) {
 		if (!error) {
 			ASSERT(*firstblock == NULLFSBLOCK ||
 			       XFS_FSB_TO_AGNO(mp, *firstblock) ==
 			       XFS_FSB_TO_AGNO(mp,
-				       cur->bc_private.b.firstblock) ||
+				       bma.cur->bc_private.b.firstblock) ||
 			       (flist->xbf_low &&
 				XFS_FSB_TO_AGNO(mp, *firstblock) <
 				XFS_FSB_TO_AGNO(mp,
-					cur->bc_private.b.firstblock)));
-			*firstblock = cur->bc_private.b.firstblock;
+					bma.cur->bc_private.b.firstblock)));
+			*firstblock = bma.cur->bc_private.b.firstblock;
 		}
-		xfs_btree_del_cursor(cur,
+		xfs_btree_del_cursor(bma.cur,
 			error ? XFS_BTREE_ERROR : XFS_BTREE_NOERROR);
 	}
 	if (!error)

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

[PATCH 20/25] xfs: move lastx and nallocs into bmalloca

[Christoph Hellwig]

[-- Attachment #1: xfs-bmalloca-lastx --]
[-- Type: text/plain, Size: 6126 bytes --]

Signed-off-by: Dave Chinner <dchinner@redhat.com>

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:14:42.542043407 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:14:43.972035659 +0200
@@ -4601,9 +4601,7 @@ xfs_bmapi_delay(
 STATIC int
 xfs_bmapi_allocate(
 	struct xfs_bmalloca	*bma,
-	xfs_extnum_t		*lastx,
 	int			flags,
-	int			*nallocs,
 	int			*logflags)
 {
 	struct xfs_mount	*mp = bma->ip->i_mount;
@@ -4622,8 +4620,8 @@ xfs_bmapi_allocate(
 	if (bma->wasdel) {
 		bma->length = bma->got.br_blockcount;
 		bma->offset = bma->got.br_startoff;
-		if (*lastx != NULLEXTNUM && *lastx) {
-			xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx - 1),
+		if (bma->idx != NULLEXTNUM && bma->idx) {
+			xfs_bmbt_get_all(xfs_iext_get_ext(ifp, bma->idx - 1),
 					 &bma->prev);
 		}
 	} else {
@@ -4677,7 +4675,7 @@ xfs_bmapi_allocate(
 	 * Bump the number of extents we've allocated
 	 * in this call.
 	 */
-	(*nallocs)++;
+	bma->nallocs++;
 
 	if (bma->cur)
 		bma->cur->bc_private.b.flags =
@@ -4697,13 +4695,14 @@ xfs_bmapi_allocate(
 		bma->got.br_state = XFS_EXT_UNWRITTEN;
 
 	if (bma->wasdel) {
-		error = xfs_bmap_add_extent_delay_real(bma->tp, bma->ip, lastx,
-				&bma->cur, &bma->got, bma->firstblock,
-				bma->flist, logflags);
+		error = xfs_bmap_add_extent_delay_real(bma->tp, bma->ip,
+				&bma->idx, &bma->cur, &bma->got,
+				bma->firstblock, bma->flist, logflags);
 	} else {
-		error = xfs_bmap_add_extent_hole_real(bma->tp, bma->ip, lastx,
-				&bma->cur, &bma->got, bma->firstblock,
-				bma->flist, logflags, whichfork);
+		error = xfs_bmap_add_extent_hole_real(bma->tp, bma->ip,
+				&bma->idx, &bma->cur, &bma->got,
+				bma->firstblock, bma->flist, logflags,
+				whichfork);
 	}
 
 	if (error)
@@ -4714,7 +4713,7 @@ xfs_bmapi_allocate(
 	 * or xfs_bmap_add_extent_hole_real might have merged it into one of
 	 * the neighbouring ones.
 	 */
-	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx), &bma->got);
+	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, bma->idx), &bma->got);
 
 	ASSERT(bma->got.br_startoff <= bma->offset);
 	ASSERT(bma->got.br_startoff + bma->got.br_blockcount >=
@@ -4729,7 +4728,6 @@ xfs_bmapi_convert_unwritten(
 	struct xfs_bmalloca	*bma,
 	struct xfs_bmbt_irec	*mval,
 	xfs_filblks_t		len,
-	xfs_extnum_t		*lastx,
 	int			flags,
 	int			*logflags)
 {
@@ -4764,7 +4762,7 @@ xfs_bmapi_convert_unwritten(
 	mval->br_state = (mval->br_state == XFS_EXT_UNWRITTEN)
 				? XFS_EXT_NORM : XFS_EXT_UNWRITTEN;
 
-	error = xfs_bmap_add_extent_unwritten_real(bma->tp, bma->ip, lastx,
+	error = xfs_bmap_add_extent_unwritten_real(bma->tp, bma->ip, &bma->idx,
 			&bma->cur, mval, bma->firstblock, bma->flist, logflags);
 	if (error)
 		return error;
@@ -4774,7 +4772,7 @@ xfs_bmapi_convert_unwritten(
 	 * xfs_bmap_add_extent_unwritten_real might have merged it into one
 	 * of the neighbouring ones.
 	 */
-	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *lastx), &bma->got);
+	xfs_bmbt_get_all(xfs_iext_get_ext(ifp, bma->idx), &bma->got);
 
 	/*
 	 * We may have combined previously unwritten space with written space,
@@ -4817,10 +4815,8 @@ xfs_bmapi_write(
 	int			eof;		/* after the end of extents */
 	int			error;		/* error return */
 	xfs_ifork_t		*ifp;		/* inode fork pointer */
-	xfs_extnum_t		lastx;		/* last useful extent number */
 	int			logflags;	/* flags for transaction logging */
 	int			n;		/* current extent index */
-	int			nallocs;	/* number of extents alloc'd */
 	xfs_fileoff_t		obno;		/* old block number (offset) */
 	int			tmp_logflags;	/* temp flags holder */
 	int			whichfork;	/* data or attr fork */
@@ -4868,7 +4864,6 @@ xfs_bmapi_write(
 	XFS_STATS_INC(xs_blk_mapw);
 
 	logflags = 0;
-	nallocs = 0;
 
 	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_LOCAL) {
 		error = xfs_bmap_local_to_extents(tp, ip, firstblock, total,
@@ -4892,7 +4887,7 @@ xfs_bmapi_write(
 			goto error0;
 	}
 
-	xfs_bmap_search_extents(ip, bno, whichfork, &eof, &lastx, &bma.got,
+	xfs_bmap_search_extents(ip, bno, whichfork, &eof, &bma.idx, &bma.got,
 				&bma.prev);
 	n = 0;
 	end = bno + len;
@@ -4920,8 +4915,7 @@ xfs_bmapi_write(
 			bma.length = len;
 			bma.offset = bno;
 
-			error = xfs_bmapi_allocate(&bma, &lastx, flags,
-						   &nallocs, &tmp_logflags);
+			error = xfs_bmapi_allocate(&bma, flags, &tmp_logflags);
 			if (error == ENOSPC || error == EDQUOT) {
 				if (n == 0) {
 					*nmap = 0;
@@ -4942,7 +4936,7 @@ xfs_bmapi_write(
 							end, n, flags);
 
 		/* Execute unwritten extent conversion if necessary */
-		error = xfs_bmapi_convert_unwritten(&bma, mval, len, &lastx,
+		error = xfs_bmapi_convert_unwritten(&bma, mval, len,
 						    flags, &tmp_logflags);
 		logflags |= tmp_logflags;
 		if (error == EAGAIN)
@@ -4958,14 +4952,15 @@ xfs_bmapi_write(
 		 * XFS_BMAP_MAX_NMAP extents no matter what.  Otherwise
 		 * the transaction may get too big.
 		 */
-		if (bno >= end || n >= *nmap || nallocs >= *nmap)
+		if (bno >= end || n >= *nmap || bma.nallocs >= *nmap)
 			break;
 
 		/* Else go on to the next record. */
 		bma.prev = bma.got;
-		if (++lastx < ifp->if_bytes / sizeof(xfs_bmbt_rec_t))
-			xfs_bmbt_get_all(xfs_iext_get_ext(ifp, lastx), &bma.got);
-		else
+		if (++bma.idx < ifp->if_bytes / sizeof(xfs_bmbt_rec_t)) {
+			xfs_bmbt_get_all(xfs_iext_get_ext(ifp, bma.idx),
+					 &bma.got);
+		} else
 			eof = 1;
 	}
 	*nmap = n;
Index: xfs/fs/xfs/xfs_bmap.h
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.h	2011-08-23 21:14:42.535376777 +0200
+++ xfs/fs/xfs/xfs_bmap.h	2011-08-23 21:14:43.972035659 +0200
@@ -121,6 +121,8 @@ typedef struct xfs_bmalloca {
 	xfs_fsblock_t		blkno;	/* starting block of new extent */
 
 	struct xfs_btree_cur	*cur;	/* btree cursor */
+	xfs_extnum_t		idx;	/* current extent index */
+	int			nallocs;/* number of extents alloc'd */
 
 	xfs_extlen_t		total;	/* total blocks needed for xaction */
 	xfs_extlen_t		minlen;	/* minimum allocation size (blocks) */

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

[PATCH 21/25] xfs: move logflags into bmalloca

[Christoph Hellwig]

[-- Attachment #1: xfs-bmalloca-logflags --]
[-- Type: text/plain, Size: 5925 bytes --]

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

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:14:43.972035659 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:14:45.412027859 +0200
@@ -4601,13 +4601,13 @@ xfs_bmapi_delay(
 STATIC int
 xfs_bmapi_allocate(
 	struct xfs_bmalloca	*bma,
-	int			flags,
-	int			*logflags)
+	int			flags)
 {
 	struct xfs_mount	*mp = bma->ip->i_mount;
 	int			whichfork = (flags & XFS_BMAPI_ATTRFORK) ?
 						XFS_ATTR_FORK : XFS_DATA_FORK;
 	struct xfs_ifork	*ifp = XFS_IFORK_PTR(bma->ip, whichfork);
+	int			tmp_logflags = 0;
 	int			error;
 	int			rt;
 
@@ -4697,14 +4697,15 @@ xfs_bmapi_allocate(
 	if (bma->wasdel) {
 		error = xfs_bmap_add_extent_delay_real(bma->tp, bma->ip,
 				&bma->idx, &bma->cur, &bma->got,
-				bma->firstblock, bma->flist, logflags);
+				bma->firstblock, bma->flist, &tmp_logflags);
 	} else {
 		error = xfs_bmap_add_extent_hole_real(bma->tp, bma->ip,
 				&bma->idx, &bma->cur, &bma->got,
-				bma->firstblock, bma->flist, logflags,
+				bma->firstblock, bma->flist, &tmp_logflags,
 				whichfork);
 	}
 
+	bma->logflags |= tmp_logflags;
 	if (error)
 		return error;
 
@@ -4728,16 +4729,14 @@ xfs_bmapi_convert_unwritten(
 	struct xfs_bmalloca	*bma,
 	struct xfs_bmbt_irec	*mval,
 	xfs_filblks_t		len,
-	int			flags,
-	int			*logflags)
+	int			flags)
 {
 	int			whichfork = (flags & XFS_BMAPI_ATTRFORK) ?
 						XFS_ATTR_FORK : XFS_DATA_FORK;
 	struct xfs_ifork	*ifp = XFS_IFORK_PTR(bma->ip, whichfork);
+	int			tmp_logflags = 0;
 	int			error;
 
-	*logflags = 0;
-
 	/* check if we need to do unwritten->real conversion */
 	if (mval->br_state == XFS_EXT_UNWRITTEN &&
 	    (flags & XFS_BMAPI_PREALLOC))
@@ -4763,7 +4762,9 @@ xfs_bmapi_convert_unwritten(
 				? XFS_EXT_NORM : XFS_EXT_UNWRITTEN;
 
 	error = xfs_bmap_add_extent_unwritten_real(bma->tp, bma->ip, &bma->idx,
-			&bma->cur, mval, bma->firstblock, bma->flist, logflags);
+			&bma->cur, mval, bma->firstblock, bma->flist,
+			&tmp_logflags);
+	bma->logflags |= tmp_logflags;
 	if (error)
 		return error;
 
@@ -4815,10 +4816,8 @@ xfs_bmapi_write(
 	int			eof;		/* after the end of extents */
 	int			error;		/* error return */
 	xfs_ifork_t		*ifp;		/* inode fork pointer */
-	int			logflags;	/* flags for transaction logging */
 	int			n;		/* current extent index */
 	xfs_fileoff_t		obno;		/* old block number (offset) */
-	int			tmp_logflags;	/* temp flags holder */
 	int			whichfork;	/* data or attr fork */
 	char			inhole;		/* current location is hole in file */
 	char			wasdelay;	/* old extent was delayed */
@@ -4863,11 +4862,9 @@ xfs_bmapi_write(
 
 	XFS_STATS_INC(xs_blk_mapw);
 
-	logflags = 0;
-
 	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_LOCAL) {
 		error = xfs_bmap_local_to_extents(tp, ip, firstblock, total,
-						  &logflags, whichfork);
+						  &bma.logflags, whichfork);
 		if (error)
 			goto error0;
 	}
@@ -4915,7 +4912,7 @@ xfs_bmapi_write(
 			bma.length = len;
 			bma.offset = bno;
 
-			error = xfs_bmapi_allocate(&bma, flags, &tmp_logflags);
+			error = xfs_bmapi_allocate(&bma, flags);
 			if (error == ENOSPC || error == EDQUOT) {
 				if (n == 0) {
 					*nmap = 0;
@@ -4924,7 +4921,6 @@ xfs_bmapi_write(
 				}
 			}
 
-			logflags |= tmp_logflags;
 			if (error)
 				goto error0;
 			if (bma.blkno == NULLFSBLOCK)
@@ -4936,9 +4932,7 @@ xfs_bmapi_write(
 							end, n, flags);
 
 		/* Execute unwritten extent conversion if necessary */
-		error = xfs_bmapi_convert_unwritten(&bma, mval, len,
-						    flags, &tmp_logflags);
-		logflags |= tmp_logflags;
+		error = xfs_bmapi_convert_unwritten(&bma, mval, len, flags);
 		if (error == EAGAIN)
 			continue;
 		if (error)
@@ -4970,10 +4964,12 @@ xfs_bmapi_write(
 	 */
 	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_BTREE &&
 	    XFS_IFORK_NEXTENTS(ip, whichfork) <= ifp->if_ext_max) {
+		int		tmp_logflags = 0;
+
 		ASSERT(bma.cur);
 		error = xfs_bmap_btree_to_extents(tp, ip, bma.cur,
 			&tmp_logflags, whichfork);
-		logflags |= tmp_logflags;
+		bma.logflags |= tmp_logflags;
 		if (error)
 			goto error0;
 	}
@@ -4987,19 +4983,19 @@ error0:
 	 * Log everything.  Do this after conversion, there's no point in
 	 * logging the extent records if we've converted to btree format.
 	 */
-	if ((logflags & xfs_ilog_fext(whichfork)) &&
+	if ((bma.logflags & xfs_ilog_fext(whichfork)) &&
 	    XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS)
-		logflags &= ~xfs_ilog_fext(whichfork);
-	else if ((logflags & xfs_ilog_fbroot(whichfork)) &&
+		bma.logflags &= ~xfs_ilog_fext(whichfork);
+	else if ((bma.logflags & xfs_ilog_fbroot(whichfork)) &&
 		 XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE)
-		logflags &= ~xfs_ilog_fbroot(whichfork);
+		bma.logflags &= ~xfs_ilog_fbroot(whichfork);
 	/*
 	 * Log whatever the flags say, even if error.  Otherwise we might miss
 	 * detecting a case where the data is changed, there's an error,
 	 * and it's not logged so we don't shutdown when we should.
 	 */
-	if (logflags)
-		xfs_trans_log_inode(tp, ip, logflags);
+	if (bma.logflags)
+		xfs_trans_log_inode(tp, ip, bma.logflags);
 
 	if (bma.cur) {
 		if (!error) {
Index: xfs/fs/xfs/xfs_bmap.h
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.h	2011-08-23 21:14:43.972035659 +0200
+++ xfs/fs/xfs/xfs_bmap.h	2011-08-23 21:14:45.412027859 +0200
@@ -123,6 +123,7 @@ typedef struct xfs_bmalloca {
 	struct xfs_btree_cur	*cur;	/* btree cursor */
 	xfs_extnum_t		idx;	/* current extent index */
 	int			nallocs;/* number of extents alloc'd */
+	int			logflags;/* flags for transaction logging */
 
 	xfs_extlen_t		total;	/* total blocks needed for xaction */
 	xfs_extlen_t		minlen;	/* minimum allocation size (blocks) */

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

[PATCH 22/25] xfs: pass bmalloca to xfs_bmap_add_extent_delay_real

[Christoph Hellwig]

[-- Attachment #1: xfs-bmalloca-xfs_bmap_add_extent_delay_real --]
[-- Type: text/plain, Size: 23250 bytes --]

All the parameters passed to xfs_bmap_add_extent_delay_real() are in the
xfs_bmalloca structure now. Just pass the bmalloca parameter to the
function instead of 8 separate parameters.

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

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:14:45.412027859 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:14:46.848686741 +0200
@@ -375,16 +375,9 @@ xfs_bmap_add_attrfork_local(
  */
 STATIC int				/* error */
 xfs_bmap_add_extent_delay_real(
-	struct xfs_trans	*tp,	/* transaction pointer */
-	xfs_inode_t		*ip,	/* incore inode pointer */
-	xfs_extnum_t		*idx,	/* extent number to update/insert */
-	xfs_btree_cur_t		**curp,	/* if *curp is null, not a btree */
-	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
-	xfs_fsblock_t		*first,	/* pointer to firstblock variable */
-	xfs_bmap_free_t		*flist,	/* list of extents to be freed */
-	int			*logflagsp) /* inode logging flags */
+	struct xfs_bmalloca	*bma)
 {
-	xfs_btree_cur_t		*cur;	/* btree cursor */
+	struct xfs_bmbt_irec	*new = &bma->got;
 	int			diff;	/* temp value */
 	xfs_bmbt_rec_host_t	*ep;	/* extent entry for idx */
 	int			error;	/* error return value */
@@ -401,18 +394,16 @@ xfs_bmap_add_extent_delay_real(
 	xfs_filblks_t		temp2=0;/* value for da_new calculations */
 	int			tmp_rval;	/* partial logging flags */
 
-	ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
-	cur = *curp;
+	ifp = XFS_IFORK_PTR(bma->ip, XFS_DATA_FORK);
 
-	ASSERT(*idx >= 0);
-	ASSERT(*idx <= ifp->if_bytes / sizeof(struct xfs_bmbt_rec));
+	ASSERT(bma->idx >= 0);
+	ASSERT(bma->idx <= ifp->if_bytes / sizeof(struct xfs_bmbt_rec));
 	ASSERT(!isnullstartblock(new->br_startblock));
-	ASSERT(!cur || (cur->bc_private.b.flags & XFS_BTCUR_BPRV_WASDEL));
+	ASSERT(!bma->cur ||
+	       (bma->cur->bc_private.b.flags & XFS_BTCUR_BPRV_WASDEL));
 
 	XFS_STATS_INC(xs_add_exlist);
 
-	*logflagsp = 0;
-
 #define	LEFT		r[0]
 #define	RIGHT		r[1]
 #define	PREV		r[2]
@@ -420,7 +411,7 @@ xfs_bmap_add_extent_delay_real(
 	/*
 	 * Set up a bunch of variables to make the tests simpler.
 	 */
-	ep = xfs_iext_get_ext(ifp, *idx);
+	ep = xfs_iext_get_ext(ifp, bma->idx);
 	xfs_bmbt_get_all(ep, &PREV);
 	new_endoff = new->br_startoff + new->br_blockcount;
 	ASSERT(PREV.br_startoff <= new->br_startoff);
@@ -442,9 +433,9 @@ xfs_bmap_add_extent_delay_real(
 	 * Check and set flags if this segment has a left neighbor.
 	 * Don't set contiguous if the combined extent would be too large.
 	 */
-	if (*idx > 0) {
+	if (bma->idx > 0) {
 		state |= BMAP_LEFT_VALID;
-		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *idx - 1), &LEFT);
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, bma->idx - 1), &LEFT);
 
 		if (isnullstartblock(LEFT.br_startblock))
 			state |= BMAP_LEFT_DELAY;
@@ -462,9 +453,9 @@ xfs_bmap_add_extent_delay_real(
 	 * Don't set contiguous if the combined extent would be too large.
 	 * Also check for all-three-contiguous being too large.
 	 */
-	if (*idx < ip->i_df.if_bytes / (uint)sizeof(xfs_bmbt_rec_t) - 1) {
+	if (bma->idx < bma->ip->i_df.if_bytes / (uint)sizeof(xfs_bmbt_rec_t) - 1) {
 		state |= BMAP_RIGHT_VALID;
-		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *idx + 1), &RIGHT);
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, bma->idx + 1), &RIGHT);
 
 		if (isnullstartblock(RIGHT.br_startblock))
 			state |= BMAP_RIGHT_DELAY;
@@ -495,35 +486,39 @@ xfs_bmap_add_extent_delay_real(
 		 * Filling in all of a previously delayed allocation extent.
 		 * The left and right neighbors are both contiguous with new.
 		 */
-		--*idx;
-		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
-		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx),
+		bma->idx--;
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, bma->idx),
 			LEFT.br_blockcount + PREV.br_blockcount +
 			RIGHT.br_blockcount);
-		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_post_update(bma->ip, bma->idx, state, _THIS_IP_);
 
-		xfs_iext_remove(ip, *idx + 1, 2, state);
-		ip->i_d.di_nextents--;
-		if (cur == NULL)
+		xfs_iext_remove(bma->ip, bma->idx + 1, 2, state);
+		bma->ip->i_d.di_nextents--;
+		if (bma->cur == NULL)
 			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
 		else {
 			rval = XFS_ILOG_CORE;
-			if ((error = xfs_bmbt_lookup_eq(cur, RIGHT.br_startoff,
+			error = xfs_bmbt_lookup_eq(bma->cur, RIGHT.br_startoff,
 					RIGHT.br_startblock,
-					RIGHT.br_blockcount, &i)))
+					RIGHT.br_blockcount, &i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_btree_delete(cur, &i)))
+			error = xfs_btree_delete(bma->cur, &i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_btree_decrement(cur, 0, &i)))
+			error = xfs_btree_decrement(bma->cur, 0, &i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_update(cur, LEFT.br_startoff,
+			error = xfs_bmbt_update(bma->cur, LEFT.br_startoff,
 					LEFT.br_startblock,
 					LEFT.br_blockcount +
 					PREV.br_blockcount +
-					RIGHT.br_blockcount, LEFT.br_state)))
+					RIGHT.br_blockcount, LEFT.br_state);
+			if (error)
 				goto done;
 		}
 		break;
@@ -533,27 +528,29 @@ xfs_bmap_add_extent_delay_real(
 		 * Filling in all of a previously delayed allocation extent.
 		 * The left neighbor is contiguous, the right is not.
 		 */
-		--*idx;
+		bma->idx--;
 
-		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
-		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx),
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, bma->idx),
 			LEFT.br_blockcount + PREV.br_blockcount);
-		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_post_update(bma->ip, bma->idx, state, _THIS_IP_);
 
-		xfs_iext_remove(ip, *idx + 1, 1, state);
-		if (cur == NULL)
+		xfs_iext_remove(bma->ip, bma->idx + 1, 1, state);
+		if (bma->cur == NULL)
 			rval = XFS_ILOG_DEXT;
 		else {
 			rval = 0;
-			if ((error = xfs_bmbt_lookup_eq(cur, LEFT.br_startoff,
+			error = xfs_bmbt_lookup_eq(bma->cur, LEFT.br_startoff,
 					LEFT.br_startblock, LEFT.br_blockcount,
-					&i)))
+					&i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_update(cur, LEFT.br_startoff,
+			error = xfs_bmbt_update(bma->cur, LEFT.br_startoff,
 					LEFT.br_startblock,
 					LEFT.br_blockcount +
-					PREV.br_blockcount, LEFT.br_state)))
+					PREV.br_blockcount, LEFT.br_state);
+			if (error)
 				goto done;
 		}
 		break;
@@ -563,26 +560,28 @@ xfs_bmap_add_extent_delay_real(
 		 * Filling in all of a previously delayed allocation extent.
 		 * The right neighbor is contiguous, the left is not.
 		 */
-		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx, state, _THIS_IP_);
 		xfs_bmbt_set_startblock(ep, new->br_startblock);
 		xfs_bmbt_set_blockcount(ep,
 			PREV.br_blockcount + RIGHT.br_blockcount);
-		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_post_update(bma->ip, bma->idx, state, _THIS_IP_);
 
-		xfs_iext_remove(ip, *idx + 1, 1, state);
-		if (cur == NULL)
+		xfs_iext_remove(bma->ip, bma->idx + 1, 1, state);
+		if (bma->cur == NULL)
 			rval = XFS_ILOG_DEXT;
 		else {
 			rval = 0;
-			if ((error = xfs_bmbt_lookup_eq(cur, RIGHT.br_startoff,
+			error = xfs_bmbt_lookup_eq(bma->cur, RIGHT.br_startoff,
 					RIGHT.br_startblock,
-					RIGHT.br_blockcount, &i)))
+					RIGHT.br_blockcount, &i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_update(cur, PREV.br_startoff,
+			error = xfs_bmbt_update(bma->cur, PREV.br_startoff,
 					new->br_startblock,
 					PREV.br_blockcount +
-					RIGHT.br_blockcount, PREV.br_state)))
+					RIGHT.br_blockcount, PREV.br_state);
+			if (error)
 				goto done;
 		}
 		break;
@@ -593,22 +592,24 @@ xfs_bmap_add_extent_delay_real(
 		 * Neither the left nor right neighbors are contiguous with
 		 * the new one.
 		 */
-		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx, state, _THIS_IP_);
 		xfs_bmbt_set_startblock(ep, new->br_startblock);
-		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_post_update(bma->ip, bma->idx, state, _THIS_IP_);
 
-		ip->i_d.di_nextents++;
-		if (cur == NULL)
+		bma->ip->i_d.di_nextents++;
+		if (bma->cur == NULL)
 			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
 		else {
 			rval = XFS_ILOG_CORE;
-			if ((error = xfs_bmbt_lookup_eq(cur, new->br_startoff,
+			error = xfs_bmbt_lookup_eq(bma->cur, new->br_startoff,
 					new->br_startblock, new->br_blockcount,
-					&i)))
+					&i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
-			cur->bc_rec.b.br_state = XFS_EXT_NORM;
-			if ((error = xfs_btree_insert(cur, &i)))
+			bma->cur->bc_rec.b.br_state = XFS_EXT_NORM;
+			error = xfs_btree_insert(bma->cur, &i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		}
@@ -619,38 +620,40 @@ xfs_bmap_add_extent_delay_real(
 		 * Filling in the first part of a previous delayed allocation.
 		 * The left neighbor is contiguous.
 		 */
-		trace_xfs_bmap_pre_update(ip, *idx - 1, state, _THIS_IP_);
-		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx - 1),
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx - 1, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, bma->idx - 1),
 			LEFT.br_blockcount + new->br_blockcount);
 		xfs_bmbt_set_startoff(ep,
 			PREV.br_startoff + new->br_blockcount);
-		trace_xfs_bmap_post_update(ip, *idx - 1, state, _THIS_IP_);
+		trace_xfs_bmap_post_update(bma->ip, bma->idx - 1, state, _THIS_IP_);
 
 		temp = PREV.br_blockcount - new->br_blockcount;
-		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx, state, _THIS_IP_);
 		xfs_bmbt_set_blockcount(ep, temp);
-		if (cur == NULL)
+		if (bma->cur == NULL)
 			rval = XFS_ILOG_DEXT;
 		else {
 			rval = 0;
-			if ((error = xfs_bmbt_lookup_eq(cur, LEFT.br_startoff,
+			error = xfs_bmbt_lookup_eq(bma->cur, LEFT.br_startoff,
 					LEFT.br_startblock, LEFT.br_blockcount,
-					&i)))
+					&i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_update(cur, LEFT.br_startoff,
+			error = xfs_bmbt_update(bma->cur, LEFT.br_startoff,
 					LEFT.br_startblock,
 					LEFT.br_blockcount +
 					new->br_blockcount,
-					LEFT.br_state)))
+					LEFT.br_state);
+			if (error)
 				goto done;
 		}
-		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(bma->ip, temp),
 			startblockval(PREV.br_startblock));
 		xfs_bmbt_set_startblock(ep, nullstartblock(da_new));
-		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_post_update(bma->ip, bma->idx, state, _THIS_IP_);
 
-		--*idx;
+		bma->idx--;
 		break;
 
 	case BMAP_LEFT_FILLING:
@@ -658,41 +661,43 @@ xfs_bmap_add_extent_delay_real(
 		 * Filling in the first part of a previous delayed allocation.
 		 * The left neighbor is not contiguous.
 		 */
-		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx, state, _THIS_IP_);
 		xfs_bmbt_set_startoff(ep, new_endoff);
 		temp = PREV.br_blockcount - new->br_blockcount;
 		xfs_bmbt_set_blockcount(ep, temp);
-		xfs_iext_insert(ip, *idx, 1, new, state);
-		ip->i_d.di_nextents++;
-		if (cur == NULL)
+		xfs_iext_insert(bma->ip, bma->idx, 1, new, state);
+		bma->ip->i_d.di_nextents++;
+		if (bma->cur == NULL)
 			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
 		else {
 			rval = XFS_ILOG_CORE;
-			if ((error = xfs_bmbt_lookup_eq(cur, new->br_startoff,
+			error = xfs_bmbt_lookup_eq(bma->cur, new->br_startoff,
 					new->br_startblock, new->br_blockcount,
-					&i)))
+					&i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
-			cur->bc_rec.b.br_state = XFS_EXT_NORM;
-			if ((error = xfs_btree_insert(cur, &i)))
+			bma->cur->bc_rec.b.br_state = XFS_EXT_NORM;
+			error = xfs_btree_insert(bma->cur, &i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		}
-		if (ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
-		    ip->i_d.di_nextents > ip->i_df.if_ext_max) {
-			error = xfs_bmap_extents_to_btree(tp, ip,
-					first, flist, &cur, 1, &tmp_rval,
-					XFS_DATA_FORK);
+		if (bma->ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
+		    bma->ip->i_d.di_nextents > bma->ip->i_df.if_ext_max) {
+			error = xfs_bmap_extents_to_btree(bma->tp, bma->ip,
+					bma->firstblock, bma->flist,
+					&bma->cur, 1, &tmp_rval, XFS_DATA_FORK);
 			rval |= tmp_rval;
 			if (error)
 				goto done;
 		}
-		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(bma->ip, temp),
 			startblockval(PREV.br_startblock) -
-			(cur ? cur->bc_private.b.allocated : 0));
-		ep = xfs_iext_get_ext(ifp, *idx + 1);
+			(bma->cur ? bma->cur->bc_private.b.allocated : 0));
+		ep = xfs_iext_get_ext(ifp, bma->idx + 1);
 		xfs_bmbt_set_startblock(ep, nullstartblock(da_new));
-		trace_xfs_bmap_post_update(ip, *idx + 1, state, _THIS_IP_);
+		trace_xfs_bmap_post_update(bma->ip, bma->idx + 1, state, _THIS_IP_);
 		break;
 
 	case BMAP_RIGHT_FILLING | BMAP_RIGHT_CONTIG:
@@ -701,37 +706,39 @@ xfs_bmap_add_extent_delay_real(
 		 * The right neighbor is contiguous with the new allocation.
 		 */
 		temp = PREV.br_blockcount - new->br_blockcount;
-		trace_xfs_bmap_pre_update(ip, *idx + 1, state, _THIS_IP_);
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx + 1, state, _THIS_IP_);
 		xfs_bmbt_set_blockcount(ep, temp);
-		xfs_bmbt_set_allf(xfs_iext_get_ext(ifp, *idx + 1),
+		xfs_bmbt_set_allf(xfs_iext_get_ext(ifp, bma->idx + 1),
 			new->br_startoff, new->br_startblock,
 			new->br_blockcount + RIGHT.br_blockcount,
 			RIGHT.br_state);
-		trace_xfs_bmap_post_update(ip, *idx + 1, state, _THIS_IP_);
-		if (cur == NULL)
+		trace_xfs_bmap_post_update(bma->ip, bma->idx + 1, state, _THIS_IP_);
+		if (bma->cur == NULL)
 			rval = XFS_ILOG_DEXT;
 		else {
 			rval = 0;
-			if ((error = xfs_bmbt_lookup_eq(cur, RIGHT.br_startoff,
+			error = xfs_bmbt_lookup_eq(bma->cur, RIGHT.br_startoff,
 					RIGHT.br_startblock,
-					RIGHT.br_blockcount, &i)))
+					RIGHT.br_blockcount, &i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_update(cur, new->br_startoff,
+			error = xfs_bmbt_update(bma->cur, new->br_startoff,
 					new->br_startblock,
 					new->br_blockcount +
 					RIGHT.br_blockcount,
-					RIGHT.br_state)))
+					RIGHT.br_state);
+			if (error)
 				goto done;
 		}
 
-		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(bma->ip, temp),
 			startblockval(PREV.br_startblock));
-		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx, state, _THIS_IP_);
 		xfs_bmbt_set_startblock(ep, nullstartblock(da_new));
-		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_post_update(bma->ip, bma->idx, state, _THIS_IP_);
 
-		++*idx;
+		bma->idx++;
 		break;
 
 	case BMAP_RIGHT_FILLING:
@@ -740,41 +747,43 @@ xfs_bmap_add_extent_delay_real(
 		 * The right neighbor is not contiguous.
 		 */
 		temp = PREV.br_blockcount - new->br_blockcount;
-		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx, state, _THIS_IP_);
 		xfs_bmbt_set_blockcount(ep, temp);
-		xfs_iext_insert(ip, *idx + 1, 1, new, state);
-		ip->i_d.di_nextents++;
-		if (cur == NULL)
+		xfs_iext_insert(bma->ip, bma->idx + 1, 1, new, state);
+		bma->ip->i_d.di_nextents++;
+		if (bma->cur == NULL)
 			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
 		else {
 			rval = XFS_ILOG_CORE;
-			if ((error = xfs_bmbt_lookup_eq(cur, new->br_startoff,
+			error = xfs_bmbt_lookup_eq(bma->cur, new->br_startoff,
 					new->br_startblock, new->br_blockcount,
-					&i)))
+					&i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
-			cur->bc_rec.b.br_state = XFS_EXT_NORM;
-			if ((error = xfs_btree_insert(cur, &i)))
+			bma->cur->bc_rec.b.br_state = XFS_EXT_NORM;
+			error = xfs_btree_insert(bma->cur, &i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		}
-		if (ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
-		    ip->i_d.di_nextents > ip->i_df.if_ext_max) {
-			error = xfs_bmap_extents_to_btree(tp, ip,
-				first, flist, &cur, 1, &tmp_rval,
-				XFS_DATA_FORK);
+		if (bma->ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
+		    bma->ip->i_d.di_nextents > bma->ip->i_df.if_ext_max) {
+			error = xfs_bmap_extents_to_btree(bma->tp, bma->ip,
+				bma->firstblock, bma->flist, &bma->cur, 1,
+				&tmp_rval, XFS_DATA_FORK);
 			rval |= tmp_rval;
 			if (error)
 				goto done;
 		}
-		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(ip, temp),
+		da_new = XFS_FILBLKS_MIN(xfs_bmap_worst_indlen(bma->ip, temp),
 			startblockval(PREV.br_startblock) -
-			(cur ? cur->bc_private.b.allocated : 0));
-		ep = xfs_iext_get_ext(ifp, *idx);
+			(bma->cur ? bma->cur->bc_private.b.allocated : 0));
+		ep = xfs_iext_get_ext(ifp, bma->idx);
 		xfs_bmbt_set_startblock(ep, nullstartblock(da_new));
-		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_post_update(bma->ip, bma->idx, state, _THIS_IP_);
 
-		++*idx;
+		bma->idx++;
 		break;
 
 	case 0:
@@ -800,46 +809,48 @@ xfs_bmap_add_extent_delay_real(
 		 */
 		temp = new->br_startoff - PREV.br_startoff;
 		temp2 = PREV.br_startoff + PREV.br_blockcount - new_endoff;
-		trace_xfs_bmap_pre_update(ip, *idx, 0, _THIS_IP_);
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx, 0, _THIS_IP_);
 		xfs_bmbt_set_blockcount(ep, temp);	/* truncate PREV */
 		LEFT = *new;
 		RIGHT.br_state = PREV.br_state;
 		RIGHT.br_startblock = nullstartblock(
-				(int)xfs_bmap_worst_indlen(ip, temp2));
+				(int)xfs_bmap_worst_indlen(bma->ip, temp2));
 		RIGHT.br_startoff = new_endoff;
 		RIGHT.br_blockcount = temp2;
 		/* insert LEFT (r[0]) and RIGHT (r[1]) at the same time */
-		xfs_iext_insert(ip, *idx + 1, 2, &LEFT, state);
-		ip->i_d.di_nextents++;
-		if (cur == NULL)
+		xfs_iext_insert(bma->ip, bma->idx + 1, 2, &LEFT, state);
+		bma->ip->i_d.di_nextents++;
+		if (bma->cur == NULL)
 			rval = XFS_ILOG_CORE | XFS_ILOG_DEXT;
 		else {
 			rval = XFS_ILOG_CORE;
-			if ((error = xfs_bmbt_lookup_eq(cur, new->br_startoff,
+			error = xfs_bmbt_lookup_eq(bma->cur, new->br_startoff,
 					new->br_startblock, new->br_blockcount,
-					&i)))
+					&i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
-			cur->bc_rec.b.br_state = XFS_EXT_NORM;
-			if ((error = xfs_btree_insert(cur, &i)))
+			bma->cur->bc_rec.b.br_state = XFS_EXT_NORM;
+			error = xfs_btree_insert(bma->cur, &i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		}
-		if (ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
-		    ip->i_d.di_nextents > ip->i_df.if_ext_max) {
-			error = xfs_bmap_extents_to_btree(tp, ip,
-					first, flist, &cur, 1, &tmp_rval,
-					XFS_DATA_FORK);
+		if (bma->ip->i_d.di_format == XFS_DINODE_FMT_EXTENTS &&
+		    bma->ip->i_d.di_nextents > bma->ip->i_df.if_ext_max) {
+			error = xfs_bmap_extents_to_btree(bma->tp, bma->ip,
+					bma->firstblock, bma->flist, &bma->cur,
+					1, &tmp_rval, XFS_DATA_FORK);
 			rval |= tmp_rval;
 			if (error)
 				goto done;
 		}
-		temp = xfs_bmap_worst_indlen(ip, temp);
-		temp2 = xfs_bmap_worst_indlen(ip, temp2);
+		temp = xfs_bmap_worst_indlen(bma->ip, temp);
+		temp2 = xfs_bmap_worst_indlen(bma->ip, temp2);
 		diff = (int)(temp + temp2 - startblockval(PREV.br_startblock) -
-			(cur ? cur->bc_private.b.allocated : 0));
+			(bma->cur ? bma->cur->bc_private.b.allocated : 0));
 		if (diff > 0) {
-			error = xfs_icsb_modify_counters(ip->i_mount,
+			error = xfs_icsb_modify_counters(bma->ip->i_mount,
 					XFS_SBS_FDBLOCKS,
 					-((int64_t)diff), 0);
 			ASSERT(!error);
@@ -847,15 +858,15 @@ xfs_bmap_add_extent_delay_real(
 				goto done;
 		}
 
-		ep = xfs_iext_get_ext(ifp, *idx);
+		ep = xfs_iext_get_ext(ifp, bma->idx);
 		xfs_bmbt_set_startblock(ep, nullstartblock((int)temp));
-		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
-		trace_xfs_bmap_pre_update(ip, *idx + 2, state, _THIS_IP_);
-		xfs_bmbt_set_startblock(xfs_iext_get_ext(ifp, *idx + 2),
+		trace_xfs_bmap_post_update(bma->ip, bma->idx, state, _THIS_IP_);
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx + 2, state, _THIS_IP_);
+		xfs_bmbt_set_startblock(xfs_iext_get_ext(ifp, bma->idx + 2),
 			nullstartblock((int)temp2));
-		trace_xfs_bmap_post_update(ip, *idx + 2, state, _THIS_IP_);
+		trace_xfs_bmap_post_update(bma->ip, bma->idx + 2, state, _THIS_IP_);
 
-		++*idx;
+		bma->idx++;
 		da_new = temp + temp2;
 		break;
 
@@ -873,14 +884,15 @@ xfs_bmap_add_extent_delay_real(
 	}
 
 	/* convert to a btree if necessary */
-	if (XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) == XFS_DINODE_FMT_EXTENTS &&
-	    XFS_IFORK_NEXTENTS(ip, XFS_DATA_FORK) > ifp->if_ext_max) {
+	if (XFS_IFORK_FORMAT(bma->ip, XFS_DATA_FORK) == XFS_DINODE_FMT_EXTENTS &&
+	    XFS_IFORK_NEXTENTS(bma->ip, XFS_DATA_FORK) > ifp->if_ext_max) {
 		int	tmp_logflags;	/* partial log flag return val */
 
-		ASSERT(cur == NULL);
-		error = xfs_bmap_extents_to_btree(tp, ip, first, flist, &cur,
+		ASSERT(bma->cur == NULL);
+		error = xfs_bmap_extents_to_btree(bma->tp, bma->ip,
+				bma->firstblock, bma->flist, &bma->cur,
 				da_old > 0, &tmp_logflags, XFS_DATA_FORK);
-		*logflagsp |= tmp_logflags;
+		bma->logflags |= tmp_logflags;
 		if (error)
 			goto done;
 	}
@@ -888,22 +900,22 @@ xfs_bmap_add_extent_delay_real(
 	/* adjust for changes in reserved delayed indirect blocks */
 	if (da_old || da_new) {
 		temp = da_new;
-		if (cur)
-			temp += cur->bc_private.b.allocated;
+		if (bma->cur)
+			temp += bma->cur->bc_private.b.allocated;
 		ASSERT(temp <= da_old);
 		if (temp < da_old)
-			xfs_icsb_modify_counters(ip->i_mount, XFS_SBS_FDBLOCKS,
-				(int64_t)(da_old - temp), 0);
+			xfs_icsb_modify_counters(bma->ip->i_mount,
+					XFS_SBS_FDBLOCKS,
+					(int64_t)(da_old - temp), 0);
 	}
 
 	/* clear out the allocated field, done with it now in any case. */
-	if (cur) {
-		cur->bc_private.b.allocated = 0;
-		*curp = cur;
-	}
-	xfs_bmap_check_leaf_extents(cur, ip, XFS_DATA_FORK);
+	if (bma->cur)
+		bma->cur->bc_private.b.allocated = 0;
+
+	xfs_bmap_check_leaf_extents(bma->cur, bma->ip, XFS_DATA_FORK);
 done:
-	*logflagsp |= rval;
+	bma->logflags |= rval;
 	return error;
 #undef	LEFT
 #undef	RIGHT
@@ -4695,9 +4707,7 @@ xfs_bmapi_allocate(
 		bma->got.br_state = XFS_EXT_UNWRITTEN;
 
 	if (bma->wasdel) {
-		error = xfs_bmap_add_extent_delay_real(bma->tp, bma->ip,
-				&bma->idx, &bma->cur, &bma->got,
-				bma->firstblock, bma->flist, &tmp_logflags);
+		error = xfs_bmap_add_extent_delay_real(bma);
 	} else {
 		error = xfs_bmap_add_extent_hole_real(bma->tp, bma->ip,
 				&bma->idx, &bma->cur, &bma->got,

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

[PATCH 23/25] xfs: pass bmalloca to xfs_bmap_add_extent_hole_real

[Christoph Hellwig]

[-- Attachment #1: xfs-bmalloca-xfs_bmap_add_extent_hole_real --]
[-- Type: text/plain, Size: 10425 bytes --]

All the parameters passed to xfs_bmap_add_extent_hole_real() are in the
xfs_bmalloca structure now. Just pass the bmalloca parameter to the
function instead of 8 separate parameters.

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

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:14:46.848686741 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:14:48.488677858 +0200
@@ -1596,34 +1596,25 @@ xfs_bmap_add_extent_hole_delay(
  */
 STATIC int				/* error */
 xfs_bmap_add_extent_hole_real(
-	struct xfs_trans	*tp,
-	xfs_inode_t		*ip,	/* incore inode pointer */
-	xfs_extnum_t		*idx,	/* extent number to update/insert */
-	xfs_btree_cur_t		**curp,	/* if null, not a btree */
-	xfs_bmbt_irec_t		*new,	/* new data to add to file extents */
-	xfs_fsblock_t		*first,	/* pointer to firstblock variable */
-	xfs_bmap_free_t		*flist,	/* list of extents to be freed */
-	int			*logflagsp, /* inode logging flags */
-	int			whichfork) /* data or attr fork */
+	struct xfs_bmalloca	*bma,
+	int			whichfork)
 {
+	struct xfs_bmbt_irec	*new = &bma->got;
 	int			error;	/* error return value */
 	int			i;	/* temp state */
-	xfs_btree_cur_t		*cur;	/* if null, not a btree */
 	xfs_ifork_t		*ifp;	/* inode fork pointer */
 	xfs_bmbt_irec_t		left;	/* left neighbor extent entry */
 	xfs_bmbt_irec_t		right;	/* right neighbor extent entry */
 	int			rval=0;	/* return value (logging flags) */
 	int			state;	/* state bits, accessed thru macros */
 
-	*logflagsp = 0;
+	ifp = XFS_IFORK_PTR(bma->ip, whichfork);
 
-	ifp = XFS_IFORK_PTR(ip, whichfork);
-	cur = *curp;
-
-	ASSERT(*idx >= 0);
-	ASSERT(*idx <= ifp->if_bytes / sizeof(struct xfs_bmbt_rec));
+	ASSERT(bma->idx >= 0);
+	ASSERT(bma->idx <= ifp->if_bytes / sizeof(struct xfs_bmbt_rec));
 	ASSERT(!isnullstartblock(new->br_startblock));
-	ASSERT(!cur || !(cur->bc_private.b.flags & XFS_BTCUR_BPRV_WASDEL));
+	ASSERT(!bma->cur ||
+	       !(bma->cur->bc_private.b.flags & XFS_BTCUR_BPRV_WASDEL));
 
 	XFS_STATS_INC(xs_add_exlist);
 
@@ -1634,9 +1625,9 @@ xfs_bmap_add_extent_hole_real(
 	/*
 	 * Check and set flags if this segment has a left neighbor.
 	 */
-	if (*idx > 0) {
+	if (bma->idx > 0) {
 		state |= BMAP_LEFT_VALID;
-		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *idx - 1), &left);
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, bma->idx - 1), &left);
 		if (isnullstartblock(left.br_startblock))
 			state |= BMAP_LEFT_DELAY;
 	}
@@ -1645,9 +1636,9 @@ xfs_bmap_add_extent_hole_real(
 	 * Check and set flags if this segment has a current value.
 	 * Not true if we're inserting into the "hole" at eof.
 	 */
-	if (*idx < ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t)) {
+	if (bma->idx < ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t)) {
 		state |= BMAP_RIGHT_VALID;
-		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, *idx), &right);
+		xfs_bmbt_get_all(xfs_iext_get_ext(ifp, bma->idx), &right);
 		if (isnullstartblock(right.br_startblock))
 			state |= BMAP_RIGHT_DELAY;
 	}
@@ -1684,39 +1675,42 @@ xfs_bmap_add_extent_hole_real(
 		 * left and on the right.
 		 * Merge all three into a single extent record.
 		 */
-		--*idx;
-		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
-		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx),
+		--bma->idx;
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, bma->idx),
 			left.br_blockcount + new->br_blockcount +
 			right.br_blockcount);
-		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_post_update(bma->ip, bma->idx, state, _THIS_IP_);
 
-		xfs_iext_remove(ip, *idx + 1, 1, state);
+		xfs_iext_remove(bma->ip, bma->idx + 1, 1, state);
 
-		XFS_IFORK_NEXT_SET(ip, whichfork,
-			XFS_IFORK_NEXTENTS(ip, whichfork) - 1);
-		if (cur == NULL) {
+		XFS_IFORK_NEXT_SET(bma->ip, whichfork,
+			XFS_IFORK_NEXTENTS(bma->ip, whichfork) - 1);
+		if (bma->cur == NULL) {
 			rval = XFS_ILOG_CORE | xfs_ilog_fext(whichfork);
 		} else {
 			rval = XFS_ILOG_CORE;
-			if ((error = xfs_bmbt_lookup_eq(cur,
-					right.br_startoff,
-					right.br_startblock,
-					right.br_blockcount, &i)))
+			error = xfs_bmbt_lookup_eq(bma->cur, right.br_startoff,
+					right.br_startblock, right.br_blockcount,
+					&i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_btree_delete(cur, &i)))
+			error = xfs_btree_delete(bma->cur, &i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_btree_decrement(cur, 0, &i)))
+			error = xfs_btree_decrement(bma->cur, 0, &i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_update(cur, left.br_startoff,
+			error = xfs_bmbt_update(bma->cur, left.br_startoff,
 					left.br_startblock,
 					left.br_blockcount +
 						new->br_blockcount +
 						right.br_blockcount,
-					left.br_state)))
+					left.br_state);
+			if (error)
 				goto done;
 		}
 		break;
@@ -1727,27 +1721,28 @@ xfs_bmap_add_extent_hole_real(
 		 * on the left.
 		 * Merge the new allocation with the left neighbor.
 		 */
-		--*idx;
-		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
-		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, *idx),
+		--bma->idx;
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx, state, _THIS_IP_);
+		xfs_bmbt_set_blockcount(xfs_iext_get_ext(ifp, bma->idx),
 			left.br_blockcount + new->br_blockcount);
-		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_post_update(bma->ip, bma->idx, state, _THIS_IP_);
 
-		if (cur == NULL) {
+		if (bma->cur == NULL) {
 			rval = xfs_ilog_fext(whichfork);
 		} else {
 			rval = 0;
-			if ((error = xfs_bmbt_lookup_eq(cur,
-					left.br_startoff,
-					left.br_startblock,
-					left.br_blockcount, &i)))
+			error = xfs_bmbt_lookup_eq(bma->cur, left.br_startoff,
+					left.br_startblock, left.br_blockcount,
+					&i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_update(cur, left.br_startoff,
+			error = xfs_bmbt_update(bma->cur, left.br_startoff,
 					left.br_startblock,
 					left.br_blockcount +
 						new->br_blockcount,
-					left.br_state)))
+					left.br_state);
+			if (error)
 				goto done;
 		}
 		break;
@@ -1758,28 +1753,30 @@ xfs_bmap_add_extent_hole_real(
 		 * on the right.
 		 * Merge the new allocation with the right neighbor.
 		 */
-		trace_xfs_bmap_pre_update(ip, *idx, state, _THIS_IP_);
-		xfs_bmbt_set_allf(xfs_iext_get_ext(ifp, *idx),
+		trace_xfs_bmap_pre_update(bma->ip, bma->idx, state, _THIS_IP_);
+		xfs_bmbt_set_allf(xfs_iext_get_ext(ifp, bma->idx),
 			new->br_startoff, new->br_startblock,
 			new->br_blockcount + right.br_blockcount,
 			right.br_state);
-		trace_xfs_bmap_post_update(ip, *idx, state, _THIS_IP_);
+		trace_xfs_bmap_post_update(bma->ip, bma->idx, state, _THIS_IP_);
 
-		if (cur == NULL) {
+		if (bma->cur == NULL) {
 			rval = xfs_ilog_fext(whichfork);
 		} else {
 			rval = 0;
-			if ((error = xfs_bmbt_lookup_eq(cur,
+			error = xfs_bmbt_lookup_eq(bma->cur,
 					right.br_startoff,
 					right.br_startblock,
-					right.br_blockcount, &i)))
+					right.br_blockcount, &i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
-			if ((error = xfs_bmbt_update(cur, new->br_startoff,
+			error = xfs_bmbt_update(bma->cur, new->br_startoff,
 					new->br_startblock,
 					new->br_blockcount +
 						right.br_blockcount,
-					right.br_state)))
+					right.br_state);
+			if (error)
 				goto done;
 		}
 		break;
@@ -1790,21 +1787,23 @@ xfs_bmap_add_extent_hole_real(
 		 * real allocation.
 		 * Insert a new entry.
 		 */
-		xfs_iext_insert(ip, *idx, 1, new, state);
-		XFS_IFORK_NEXT_SET(ip, whichfork,
-			XFS_IFORK_NEXTENTS(ip, whichfork) + 1);
-		if (cur == NULL) {
+		xfs_iext_insert(bma->ip, bma->idx, 1, new, state);
+		XFS_IFORK_NEXT_SET(bma->ip, whichfork,
+			XFS_IFORK_NEXTENTS(bma->ip, whichfork) + 1);
+		if (bma->cur == NULL) {
 			rval = XFS_ILOG_CORE | xfs_ilog_fext(whichfork);
 		} else {
 			rval = XFS_ILOG_CORE;
-			if ((error = xfs_bmbt_lookup_eq(cur,
+			error = xfs_bmbt_lookup_eq(bma->cur,
 					new->br_startoff,
 					new->br_startblock,
-					new->br_blockcount, &i)))
+					new->br_blockcount, &i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 0, done);
-			cur->bc_rec.b.br_state = new->br_state;
-			if ((error = xfs_btree_insert(cur, &i)))
+			bma->cur->bc_rec.b.br_state = new->br_state;
+			error = xfs_btree_insert(bma->cur, &i);
+			if (error)
 				goto done;
 			XFS_WANT_CORRUPTED_GOTO(i == 1, done);
 		}
@@ -1812,26 +1811,26 @@ xfs_bmap_add_extent_hole_real(
 	}
 
 	/* convert to a btree if necessary */
-	if (XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_EXTENTS &&
-	    XFS_IFORK_NEXTENTS(ip, whichfork) > ifp->if_ext_max) {
+	if (XFS_IFORK_FORMAT(bma->ip, whichfork) == XFS_DINODE_FMT_EXTENTS &&
+	    XFS_IFORK_NEXTENTS(bma->ip, whichfork) > ifp->if_ext_max) {
 		int	tmp_logflags;	/* partial log flag return val */
 
-		ASSERT(cur == NULL);
-		error = xfs_bmap_extents_to_btree(tp, ip, first,
-			flist, &cur, 0, &tmp_logflags, whichfork);
-		*logflagsp |= tmp_logflags;
+		ASSERT(bma->cur == NULL);
+		error = xfs_bmap_extents_to_btree(bma->tp, bma->ip,
+				bma->firstblock, bma->flist, &bma->cur,
+				0, &tmp_logflags, whichfork);
+		bma->logflags |= tmp_logflags;
 		if (error)
 			goto done;
 	}
 
 	/* clear out the allocated field, done with it now in any case. */
-	if (cur) {
-		cur->bc_private.b.allocated = 0;
-		*curp = cur;
-	}
-	xfs_bmap_check_leaf_extents(cur, ip, whichfork);
+	if (bma->cur)
+		bma->cur->bc_private.b.allocated = 0;
+
+	xfs_bmap_check_leaf_extents(bma->cur, bma->ip, whichfork);
 done:
-	*logflagsp |= rval;
+	bma->logflags |= rval;
 	return error;
 }
 
@@ -4706,14 +4705,10 @@ xfs_bmapi_allocate(
 	    xfs_sb_version_hasextflgbit(&mp->m_sb))
 		bma->got.br_state = XFS_EXT_UNWRITTEN;
 
-	if (bma->wasdel) {
+	if (bma->wasdel)
 		error = xfs_bmap_add_extent_delay_real(bma);
-	} else {
-		error = xfs_bmap_add_extent_hole_real(bma->tp, bma->ip,
-				&bma->idx, &bma->cur, &bma->got,
-				bma->firstblock, bma->flist, &tmp_logflags,
-				whichfork);
-	}
+	else
+		error = xfs_bmap_add_extent_hole_real(bma, whichfork);
 
 	bma->logflags |= tmp_logflags;
 	if (error)

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

[PATCH 24/25] xfs: dont ignore error code from xfs_bmbt_update

[Christoph Hellwig]

[-- Attachment #1: xfs-xfs_bmap_add_extent_unwritten_real-fix-error-handling --]
[-- Type: text/plain, Size: 892 bytes --]

Fix a case in xfs_bmap_add_extent_unwritten_real where we aren't passing
the returned error on.

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

Index: xfs/fs/xfs/xfs_bmap.c
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:14:48.488677858 +0200
+++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:14:50.002002992 +0200
@@ -1217,10 +1217,11 @@ xfs_bmap_add_extent_unwritten_real(
 				goto done;
 			if ((error = xfs_btree_decrement(cur, 0, &i)))
 				goto done;
-			if (xfs_bmbt_update(cur, LEFT.br_startoff,
+			error = xfs_bmbt_update(cur, LEFT.br_startoff,
 				LEFT.br_startblock,
 				LEFT.br_blockcount + new->br_blockcount,
-				LEFT.br_state))
+				LEFT.br_state);
+			if (error)
 				goto done;
 		}
 		break;

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

[PATCH 25/25] xfs: cleanup xfs_bmap.h

[Christoph Hellwig]

[-- Attachment #1: xfs-bmap-header-cleanup --]
[-- Type: text/plain, Size: 9093 bytes --]

Convert all function prototypes to the short form used elsewhere, and
remove duplicates of comments already placed at the function body.

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

Index: xfs/fs/xfs/xfs_bmap.h
===================================================================
--- xfs.orig/fs/xfs/xfs_bmap.h	2011-08-23 21:14:45.412027859 +0200
+++ xfs/fs/xfs/xfs_bmap.h	2011-08-23 21:17:13.324559881 +0200
@@ -156,121 +156,29 @@ typedef struct xfs_bmalloca {
 	{ BMAP_RIGHT_FILLING,	"RF" }, \
 	{ BMAP_ATTRFORK,	"ATTR" }
 
-/*
- * Add bmap trace insert entries for all the contents of the extent list.
- *
- * Quite excessive tracing.  Only do this for debug builds.
- */
 #if defined(__KERNEL) && defined(DEBUG)
-void
-xfs_bmap_trace_exlist(
-	struct xfs_inode	*ip,		/* incore inode pointer */
-	xfs_extnum_t		cnt,		/* count of entries in list */
-	int			whichfork,
-	unsigned long		caller_ip);	/* data or attr fork */
+void	xfs_bmap_trace_exlist(struct xfs_inode *ip, xfs_extnum_t cnt,
+		int whichfork, unsigned long caller_ip);
 #define	XFS_BMAP_TRACE_EXLIST(ip,c,w)	\
 	xfs_bmap_trace_exlist(ip,c,w, _THIS_IP_)
 #else
 #define	XFS_BMAP_TRACE_EXLIST(ip,c,w)
 #endif
 
-/*
- * Convert inode from non-attributed to attributed.
- * Must not be in a transaction, ip must not be locked.
- */
-int					/* error code */
-xfs_bmap_add_attrfork(
-	struct xfs_inode	*ip,	/* incore inode pointer */
-	int			size,	/* space needed for new attribute */
-	int			rsvd);	/* flag for reserved block allocation */
-
-/*
- * Add the extent to the list of extents to be free at transaction end.
- * The list is maintained sorted (by block number).
- */
-void
-xfs_bmap_add_free(
-	xfs_fsblock_t		bno,		/* fs block number of extent */
-	xfs_filblks_t		len,		/* length of extent */
-	xfs_bmap_free_t		*flist,		/* list of extents */
-	struct xfs_mount	*mp);		/* mount point structure */
-
-/*
- * Routine to clean up the free list data structure when
- * an error occurs during a transaction.
- */
-void
-xfs_bmap_cancel(
-	xfs_bmap_free_t		*flist);	/* free list to clean up */
-
-/*
- * Compute and fill in the value of the maximum depth of a bmap btree
- * in this filesystem.  Done once, during mount.
- */
-void
-xfs_bmap_compute_maxlevels(
-	struct xfs_mount	*mp,	/* file system mount structure */
-	int			whichfork);	/* data or attr fork */
-
-/*
- * Returns the file-relative block number of the first unused block in the file.
- * This is the lowest-address hole if the file has holes, else the first block
- * past the end of file.
- */
-int						/* error */
-xfs_bmap_first_unused(
-	struct xfs_trans	*tp,		/* transaction pointer */
-	struct xfs_inode	*ip,		/* incore inode */
-	xfs_extlen_t		len,		/* size of hole to find */
-	xfs_fileoff_t		*unused,	/* unused block num */
-	int			whichfork);	/* data or attr fork */
-
-/*
- * Returns the file-relative block number of the last block + 1 before
- * last_block (input value) in the file.
- * This is not based on i_size, it is based on the extent list.
- * Returns 0 for local files, as they do not have an extent list.
- */
-int						/* error */
-xfs_bmap_last_before(
-	struct xfs_trans	*tp,		/* transaction pointer */
-	struct xfs_inode	*ip,		/* incore inode */
-	xfs_fileoff_t		*last_block,	/* last block */
-	int			whichfork);	/* data or attr fork */
-
-/*
- * Returns the file-relative block number of the first block past eof in
- * the file.  This is not based on i_size, it is based on the extent list.
- * Returns 0 for local files, as they do not have an extent list.
- */
-int						/* error */
-xfs_bmap_last_offset(
-	struct xfs_trans	*tp,		/* transaction pointer */
-	struct xfs_inode	*ip,		/* incore inode */
-	xfs_fileoff_t		*unused,	/* last block num */
-	int			whichfork);	/* data or attr fork */
-
-/*
- * Returns whether the selected fork of the inode has exactly one
- * block or not.  For the data fork we check this matches di_size,
- * implying the file's range is 0..bsize-1.
- */
-int
-xfs_bmap_one_block(
-	struct xfs_inode	*ip,		/* incore inode */
-	int			whichfork);	/* data or attr fork */
-
-/*
- * Read in the extents to iu_extents.
- * All inode fields are set up by caller, we just traverse the btree
- * and copy the records in.
- */
-int						/* error */
-xfs_bmap_read_extents(
-	struct xfs_trans	*tp,		/* transaction pointer */
-	struct xfs_inode	*ip,		/* incore inode */
-	int			whichfork);	/* data or attr fork */
-
+int	xfs_bmap_add_attrfork(struct xfs_inode *ip, int size, int rsvd);
+void	xfs_bmap_add_free(xfs_fsblock_t bno, xfs_filblks_t len,
+		struct xfs_bmap_free *flist, struct xfs_mount *mp);
+void	xfs_bmap_cancel(struct xfs_bmap_free *flist);
+void	xfs_bmap_compute_maxlevels(struct xfs_mount *mp, int whichfork);
+int	xfs_bmap_first_unused(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_extlen_t len, xfs_fileoff_t *unused, int whichfork);
+int	xfs_bmap_last_before(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_fileoff_t *last_block, int whichfork);
+int	xfs_bmap_last_offset(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_fileoff_t *unused, int whichfork);
+int	xfs_bmap_one_block(struct xfs_inode *ip, int whichfork);
+int	xfs_bmap_read_extents(struct xfs_trans *tp, struct xfs_inode *ip,
+		int whichfork);
 int	xfs_bmapi_read(struct xfs_inode *ip, xfs_fileoff_t bno,
 		xfs_filblks_t len, struct xfs_bmbt_irec *mval,
 		int *nmap, int flags);
@@ -282,95 +190,28 @@ int	xfs_bmapi_write(struct xfs_trans *tp
 		xfs_fsblock_t *firstblock, xfs_extlen_t total,
 		struct xfs_bmbt_irec *mval, int *nmap,
 		struct xfs_bmap_free *flist);
-
-/*
- * Unmap (remove) blocks from a file.
- * If nexts is nonzero then the number of extents to remove is limited to
- * that value.  If not all extents in the block range can be removed then
- * *done is set.
- */
-int						/* error */
-xfs_bunmapi(
-	struct xfs_trans	*tp,		/* transaction pointer */
-	struct xfs_inode	*ip,		/* incore inode */
-	xfs_fileoff_t		bno,		/* starting offset to unmap */
-	xfs_filblks_t		len,		/* length to unmap in file */
-	int			flags,		/* XFS_BMAPI_... */
-	xfs_extnum_t		nexts,		/* number of extents max */
-	xfs_fsblock_t		*firstblock,	/* first allocated block
-						   controls a.g. for allocs */
-	xfs_bmap_free_t		*flist,		/* i/o: list extents to free */
-	int			*done);		/* set if not done yet */
-
-/*
- * Check an extent list, which has just been read, for
- * any bit in the extent flag field.
- */
-int
-xfs_check_nostate_extents(
-	struct xfs_ifork	*ifp,
-	xfs_extnum_t		idx,
-	xfs_extnum_t		num);
-
-uint
-xfs_default_attroffset(
-	struct xfs_inode	*ip);
+int	xfs_bunmapi(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_fileoff_t bno, xfs_filblks_t len, int flags,
+		xfs_extnum_t nexts, xfs_fsblock_t *firstblock,
+		struct xfs_bmap_free *flist, int *done);
+int	xfs_check_nostate_extents(struct xfs_ifork *ifp, xfs_extnum_t idx,
+		xfs_extnum_t num);
+uint	xfs_default_attroffset(struct xfs_inode *ip);
 
 #ifdef __KERNEL__
-
-/*
- * Routine to be called at transaction's end by xfs_bmapi, xfs_bunmapi
- * caller.  Frees all the extents that need freeing, which must be done
- * last due to locking considerations.
- *
- * Return 1 if the given transaction was committed and a new one allocated,
- * and 0 otherwise.
- */
-int						/* error */
-xfs_bmap_finish(
-	struct xfs_trans	**tp,		/* transaction pointer addr */
-	xfs_bmap_free_t		*flist,		/* i/o: list extents to free */
-	int			*committed);	/* xact committed or not */
-
 /* bmap to userspace formatter - copy to user & advance pointer */
 typedef int (*xfs_bmap_format_t)(void **, struct getbmapx *, int *);
 
-/*
- * Get inode's extents as described in bmv, and format for output.
- */
-int						/* error code */
-xfs_getbmap(
-	xfs_inode_t		*ip,
-	struct getbmapx		*bmv,		/* user bmap structure */
-	xfs_bmap_format_t	formatter,	/* format to user */
-	void			*arg);		/* formatter arg */
-
-/*
- * Check if the endoff is outside the last extent. If so the caller will grow
- * the allocation to a stripe unit boundary
- */
-int
-xfs_bmap_eof(
-	struct xfs_inode        *ip,
-	xfs_fileoff_t           endoff,
-	int                     whichfork,
-	int                     *eof);
-
-/*
- * Count fsblocks of the given fork.
- */
-int
-xfs_bmap_count_blocks(
-	xfs_trans_t		*tp,
-	struct xfs_inode	*ip,
-	int			whichfork,
-	int			*count);
-
-int
-xfs_bmap_punch_delalloc_range(
-	struct xfs_inode	*ip,
-	xfs_fileoff_t		start_fsb,
-	xfs_fileoff_t		length);
+int	xfs_bmap_finish(struct xfs_trans **tp, struct xfs_bmap_free *flist,
+		int *committed);
+int	xfs_getbmap(struct xfs_inode *ip, struct getbmapx *bmv,
+		xfs_bmap_format_t formatter, void *arg);
+int	xfs_bmap_eof(struct xfs_inode *ip, xfs_fileoff_t endoff,
+		int whichfork, int *eof);
+int	xfs_bmap_count_blocks(struct xfs_trans *tp, struct xfs_inode *ip,
+		int whichfork, int *count);
+int	xfs_bmap_punch_delalloc_range(struct xfs_inode *ip,
+		xfs_fileoff_t start_fsb, xfs_fileoff_t length);
 #endif	/* __KERNEL__ */
 
 #endif	/* __XFS_BMAP_H__ */

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

Re: [PATCH 00/25] xfs_bmapi split and cleanups

[Christoph Hellwig]

Looks like quilt mail lost some From: attributions.  A lot of these
patches are directly from Dave or had only minimal changes so they
should be attributed to him.  It's still visible in the Signed-off-by
lines.

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

Re: [PATCH 01/25] xfs: remove the first extent special case in xfs_bmap_add_extent

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> Both xfs_bmap_add_extent_hole_delay and xfs_bmap_add_extent_hole_real
> already contain code to handle the case where there is no extent to
> merge with, which is effectively the same as the code duplicated here.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

It looks like an attribute fork will never get a delayed
allocation extent assigned to it.  At least I assume so,
because xfs_bmap_add_extent_hole_delay() only ever works
on the data fork.  (I didn't know that.)

Anyway, it took a bit to get myself into this--no surprise,
this is the bmapi code--but assuming the above is true this
does produce the same result as before.

Reviewed-by: Alex Elder <aelder@sgi.com>


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

Re: [PATCH 02/25] xfs: remove impossible to read code in xfs_bmap_add_extent_delay_real

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> We already have the worst case blocks reserved, so xfs_icsb_modify_counters
> won't fail in xfs_bmap_add_extent_delay_real.  In fact we've had an assert
> to catch this case since day and it never triggered.  So remove the code
> to try smaller reservations, and just return the error for that case in
> addition to keeping the assert.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

It was a very weird block of code that I never did
really understand.  It appears to be trying to recover
from running out of space by reserving less and
retrying.  But that means you're not reserving enough
to cover worst case, which wouldn't be safe.

Anyway, given this function already has an error
return path that could handle it if we actually
had not pre-reserved enough blocks, this looks good.

Reviewed-by: Alex Elder <aelder@sgi.com>


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

Re: [PATCH 03/25] xfs: remove the nextents variable in xfs_bmapi

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> Instead of using a local variable that needs to updated when we modify
> the extent map just check ifp->if_bytes directly where we use it.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

Looks good.

Reviewed-by: Alex Elder <aelder@sgi.com>


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

Re: [PATCH 04/25] xfs: factor extent map manipulations out of xfs_bmapi

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:

Unfortunately the content of the new xfs_bmapi_update_map()
is as hideous as it was before, but at least it's isolated.
The -1 index values are just awful, even if we do know from
the logic that they're indexing non-negative offsets into the
underlying array.  I do think it's good that you encapsulated
this without changing the existing logic, but someday maybe
this function can be improved.

Reviewed-by: Alex Elder <aelder@sgi.com>

You included no signoff.  I am going to assume it included
this like the rest of the series:

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


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

Re: [PATCH 05/25] xfs: introduce xfs_bmapi_read()

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> xfs_bmapi() currently handles both extent map reading and
> allocation. As a result, the code is littered with "if (wr)"
> branches to conditionally do allocation operations if required.
> This makes the code much harder to follow and causes significant
> indent issues with the code.
> 
> Given that read mapping is much simpler than allocation, we can
> split out read mapping from xfs_bmapi() and reuse the logic that
> we have already factored out do do all the hard work of handling the
> extent map manipulations. The results in a much simpler function for
> the common extent read operations, and will allow the allocation
> code to be simplified in another commit.
> 
> Once xfs_bmapi_read() is implemented, convert all the callers of
> xfs_bmapi() that are only reading extents to use the new function.
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

Very nice.  About a third of the uses of XFS_BMAPI_METADATA
were unnecessary because that flag only matters on writes.
That's a good demonstration of how refactoring like this
improves clarity (if the simplicity of the xfs_bmapi_read()
function isn't evidence enough).

Reviewed-by: Alex Elder <aelder@sgi.com>


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

Re: [PATCH 06/25] xfs: remove xfs_bmapi_single()

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> Now we have xfs_bmapi_read, there is no need for xfs_bmapi_single().
> Change the remaining caller over and kill the function.
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

Nice.

Reviewed-by: Alex Elder <aelder@sgi.com>


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

Re: [PATCH 07/25] xfs: factor delalloc reservations out of xfs_bmapi

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> plain text document attachment (xfs-bmapi-factor-delalloc-reservation)
> Move the reservation of delayed allocations, and addition of delalloc
> regions to the extent trees into a new helper function.  For now
> this adds some twisted goto logic to xfs_bmapi, but that will be
> cleaned up in the following patches.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

Looks good.

Reviewed-by: Alex Elder <aelder@sgi.com>


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

Re: [PATCH 08/25] xfs: introduce xfs_bmapi_delay()

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> plain text document attachment (xfs-bmapi-add-xfs_bmapi_delay)
> Delalloc reservations are much simpler than allocations, so give them a
> separate bmapi-level interface.  Using the previously added
> xfs_bmapi_reserve_delalloc we get a function that is only minimally
> more complicated than xfs_bmapi_read, which is far from the complexity
> in xfs_bmapi.  Also remove the XFS_BMAPI_DELAY code after switching
> over the only user to xfs_bmapi_delay.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

I have a minor suggestion but it's not worth
rejecting this over.  Also a question.  But
regardless, this looks good:

Reviewed-by: Alex Elder <aelder@sgi.com>

I'm out of time today.  I will have to continue
later.

> Index: xfs/fs/xfs/xfs_bmap.c
> ===================================================================
> --- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:12:56.272619118 +0200
> +++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:12:59.355935746 +0200
> @@ -4238,7 +4238,7 @@ xfs_bmap_validate_ret(
>  		ASSERT(i == 0 ||
>  		       mval[i - 1].br_startoff + mval[i - 1].br_blockcount ==
>  		       mval[i].br_startoff);
> -		if ((flags & XFS_BMAPI_WRITE) && !(flags & XFS_BMAPI_DELAY))
> +		if (flags & XFS_BMAPI_WRITE)
>  			ASSERT(mval[i].br_startblock != DELAYSTARTBLOCK &&
>  			       mval[i].br_startblock != HOLESTARTBLOCK);
>  		ASSERT(mval[i].br_state == XFS_EXT_NORM ||
> @@ -4553,6 +4553,90 @@ out_unreserve_quota:
>  }
>  
>  /*
> + * Map file blocks to filesystem blocks, adding delayed allocations as needed.
> + */
> +int
> +xfs_bmapi_delay(
> +	struct xfs_inode	*ip,	/* incore inode */
> +	xfs_fileoff_t		bno,	/* starting file offs. mapped */
> +	xfs_filblks_t		len,	/* length to map in file */
> +	struct xfs_bmbt_irec	*mval,	/* output: map values */
> +	int			*nmap,	/* i/o: mval size/count */
> +	int			flags)	/* XFS_BMAPI_... */
> +{
> +	struct xfs_mount	*mp = ip->i_mount;
> +	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
> +	struct xfs_bmbt_irec	got;	/* current file extent record */
> +	struct xfs_bmbt_irec	prev;	/* previous file extent record */
> +	xfs_fileoff_t		obno;	/* old block number (offset) */
> +	xfs_fileoff_t		end;	/* end of mapped file region */
> +	xfs_extnum_t		lastx;	/* last useful extent number */
> +	int			eof;	/* we've hit the end of extents */
> +	int			n = 0;	/* current extent index */
> +	int			error = 0;
> +
> +	ASSERT(*nmap >= 1);
> +	ASSERT(*nmap <= XFS_BMAP_MAX_NMAP);
> +	ASSERT(!(flags & ~XFS_BMAPI_ENTIRE));
> +

Rearrange the following test to use the pattern (assigning error)
used in xfs_bmapi_read().

> +	if (unlikely(XFS_TEST_ERROR(
> +	    (XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) != XFS_DINODE_FMT_EXTENTS &&
> +	     XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) != XFS_DINODE_FMT_BTREE),

Are you certain that XFS_DINODE_FMT_LOCAL is not possible here?
I tried to trace it back but I'm still not sure.  The transaction
pointer passed is null, so it would have tripped an assertion
in the previous code.  (A simple explanation would reassure me.)

> +	     mp, XFS_ERRTAG_BMAPIFORMAT, XFS_RANDOM_BMAPIFORMAT))) {
> +		XFS_ERROR_REPORT("xfs_bmapi_delay", XFS_ERRLEVEL_LOW, mp);
> +		return XFS_ERROR(EFSCORRUPTED);
> +	}
> +
> +	if (XFS_FORCED_SHUTDOWN(mp))
> +		return XFS_ERROR(EIO);
> +
> +	XFS_STATS_INC(xs_blk_mapw);
> +
> +
. . .

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

Re: [PATCH 01/25] xfs: remove the first extent special case in xfs_bmap_add_extent

[Christoph Hellwig]

On Fri, Sep 02, 2011 at 05:22:56PM -0500, Alex Elder wrote:
> On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> > Both xfs_bmap_add_extent_hole_delay and xfs_bmap_add_extent_hole_real
> > already contain code to handle the case where there is no extent to
> > merge with, which is effectively the same as the code duplicated here.
> > 
> > Signed-off-by: Christoph Hellwig <hch@lst.de>
> 
> It looks like an attribute fork will never get a delayed
> allocation extent assigned to it.  At least I assume so,
> because xfs_bmap_add_extent_hole_delay() only ever works
> on the data fork.  (I didn't know that.)

Yes, we only use delayed allocation for file data.  The same is
also true for unwritten extents.

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

Re: [PATCH 04/25] xfs: factor extent map manipulations out of xfs_bmapi

[Christoph Hellwig]

On Fri, Sep 02, 2011 at 05:23:11PM -0500, Alex Elder wrote:
> You included no signoff.  I am going to assume it included
> this like the rest of the series:
> 
>     Signed-off-by: Christoph Hellwig <hch@lst.de>

It's actually a patch from Dave that I didn't modify in any significant
way, so it should have his signoff.

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

Re: [PATCH 08/25] xfs: introduce xfs_bmapi_delay()

[Christoph Hellwig]

> > +	xfs_extnum_t		lastx;	/* last useful extent number */
> > +	int			eof;	/* we've hit the end of extents */
> > +	int			n = 0;	/* current extent index */
> > +	int			error = 0;
> > +
> > +	ASSERT(*nmap >= 1);
> > +	ASSERT(*nmap <= XFS_BMAP_MAX_NMAP);
> > +	ASSERT(!(flags & ~XFS_BMAPI_ENTIRE));
> > +
> 
> Rearrange the following test to use the pattern (assigning error)
> used in xfs_bmapi_read().

Hmm - given that error is used as a boolean there I don't actually like
that pattern very much as error is generally used to hold an errno
value.

> 
> > +	if (unlikely(XFS_TEST_ERROR(
> > +	    (XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) != XFS_DINODE_FMT_EXTENTS &&
> > +	     XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) != XFS_DINODE_FMT_BTREE),
> 
> Are you certain that XFS_DINODE_FMT_LOCAL is not possible here?
> I tried to trace it back but I'm still not sure.  The transaction
> pointer passed is null, so it would have tripped an assertion
> in the previous code.  (A simple explanation would reassure me.)

We can't hit it because we do not support the local format for regular
files at all, and we do not support delayed allocations for anything
but regular files.

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

Re: [PATCH 08/25] xfs: introduce xfs_bmapi_delay()

[Alex Elder]

On Tue, 2011-09-06 at 11:27 -0400, Christoph Hellwig wrote:
> > > +	xfs_extnum_t		lastx;	/* last useful extent number */
> > > +	int			eof;	/* we've hit the end of extents */
> > > +	int			n = 0;	/* current extent index */
> > > +	int			error = 0;
> > > +
> > > +	ASSERT(*nmap >= 1);
> > > +	ASSERT(*nmap <= XFS_BMAP_MAX_NMAP);
> > > +	ASSERT(!(flags & ~XFS_BMAPI_ENTIRE));
> > > +
> > 
> > Rearrange the following test to use the pattern (assigning error)
> > used in xfs_bmapi_read().
> 
> Hmm - given that error is used as a boolean there I don't actually like
> that pattern very much as error is generally used to hold an errno
> value.

My main thought was "make them consistent."  But
looking at it again I agree that "error" is not
the right name.  It doesn't really need changing
but could be beautified in some later patch.

> > 
> > > +	if (unlikely(XFS_TEST_ERROR(
> > > +	    (XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) != XFS_DINODE_FMT_EXTENTS &&
> > > +	     XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) != XFS_DINODE_FMT_BTREE),
> > 
> > Are you certain that XFS_DINODE_FMT_LOCAL is not possible here?
> > I tried to trace it back but I'm still not sure.  The transaction
> > pointer passed is null, so it would have tripped an assertion
> > in the previous code.  (A simple explanation would reassure me.)
> 
> We can't hit it because we do not support the local format for regular
> files at all, and we do not support delayed allocations for anything
> but regular files.

OK.  That's another thing I didn't realize about XFS...

					-Alex


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

Re: [PATCH 04/25] xfs: factor extent map manipulations out of xfs_bmapi

[Alex Elder]

On Tue, 2011-09-06 at 07:55 -0400, Christoph Hellwig wrote:
> On Fri, Sep 02, 2011 at 05:23:11PM -0500, Alex Elder wrote:
> > You included no signoff.  I am going to assume it included
> > this like the rest of the series:
> > 
> >     Signed-off-by: Christoph Hellwig <hch@lst.de>
> 
> It's actually a patch from Dave that I didn't modify in any significant
> way, so it should have his signoff.
> 

OK.  I won't commit it without recording that.
Do you want your own signoff included also?

					-Alex

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

Re: [PATCH 04/25] xfs: factor extent map manipulations out of xfs_bmapi

[Christoph Hellwig]

On Wed, Sep 07, 2011 at 12:21:44PM -0500, Alex Elder wrote:
> OK.  I won't commit it without recording that.
> Do you want your own signoff included also?

I have fixed all the attributions in my local tree.  I'll wait
for the reviews to finish and will resend it after that.  I'll
also pick anything else that needs uptdating.

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

Re: [PATCH 09/25] xfs: do not use xfs_bmap_add_extent for adding delalloc extents

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote: 
> We can just call xfs_bmap_add_extent_hole_delay directly to add a delayed
> allocated regions to the extent tree, instead of going through all the
> complexities of xfs_bmap_add_extent that aren't needed for this simple
> case.

Looks good.

By the way, while looking at this I noticed the code in xfs_bunmapi()
seemed to be doing arithmetic with a possibly null br_startblock,
which can't be right...  (In assigning sum just after wasdel gets
assigned.)

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

Reviewed-by: Alex Elder <aelder@sgi.com>



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

Re: [PATCH 10/25] xfs: factor extent allocation out of xfs_bmapi

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> To further improve the readability of xfs_bmapi(), factor the extent
> allocation out into a separate function. This removes a large block
> of logic from the xfs_bmapi() code loop and makes it easier to see
> the operational logic flow for xfs_bmapi().
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

OK, this looks very good.  I have a spot that I chased
for a while to verify it produced the same functionality
as before, but I just gave up because it was just taking
much too much time.  I'll point it out below, just for
the record, but I'm not too concerned about it.  Everything
else looks good.

Reviewed-by: Alex Elder <aelder@sgi.com>

> Index: xfs/fs/xfs/xfs_bmap.c
> ===================================================================

. . .

> @@ -4750,144 +4893,31 @@ xfs_bmapi(
>  		 * that we found, if any.
>  		 */
>  		if (wr && (inhole || wasdelay)) {
> -			/*
> -			 * For the wasdelay case, we could also just
> -			 * allocate the stuff asked for in this bmap call

. . .

> -						ip, whichfork);
> -					cur->bc_private.b.firstblock =
> -						*firstblock;
> -					cur->bc_private.b.flist = flist;
> +			bma.eof = eof;
> +			bma.conv = !!(flags & XFS_BMAPI_CONVERT);
> +			bma.wasdel = wasdelay;
> +			bma.alen = len;
> +			bma.off = bno;
> +			bma.minleft = minleft;
> +
> +			error = xfs_bmapi_allocate(&bma, &lastx, &cur,
> +					firstblock, flist, flags, &nallocs,
> +					&tmp_logflags);

> +			if (error == ENOSPC || error == EDQUOT) {
> +				if (n == 0) {
> +					*nmap = 0;
> +					ASSERT(cur == NULL);
> +					return error;
>  				}

Here is the spot I mentioned above.  I was trying to find out
the circumstances under which ENOSPC or EDQUOT could get returned
by xfs_bmapi_allocate() in order to confirm that this in fact
produces the same effect as before.

I also had a little trouble because there were spots--such
as calling xfs_bmap_isaeof()--that are now encapsulated within
xfs_bmapi_allocate() that previously jumped to error0, but now
will produce an error return from that function.  So now this
doesn't execute the code at error0 in this case.  I didn't
work through it but I trust that the code there would end
up being a series of no-ops anyway.



> -				/*
> -				 * Bump the number of extents we've allocated
> -				 * in this call.
> -				 */
> -				nallocs++;
> -			}

. . . .

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

Re: [PATCH 11/25] xfs: factor unwritten extent map manipulations out of xfs_bmapi

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> To further improve the readability of xfs_bmapi(), factor the unwritten
> extent conversion out into a separate function. This removes
> large block of logic from the xfs_bmapi() code loop and makes it
> easier to see the operational logic flow for xfs_bmapi().
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

Looks good.

Reviewed-by: Alex Elder <aelder@sgi.com>


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

Re: [PATCH 12/25] xfs: rename xfs_bmapi to xfs_bmapi_write

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> plain text document attachment (xfs-bmapi-rename-xfs_bmapi)
> Once this is done, rename xfs_bmapi to xfs_bmapi_write to reflect the fact it
> is for allocation only. This enables us to kill the XFS_BMAPI_WRITE flag as
> well.
> 
> Also clean up xfs_bmapi_write to the style used in the newly added
> xfs_bmapi_read/delay functions.


I have a few spelling corrections, otherwise this
is just beautiful.

Reviewed-by: Alex Elder <aelder@sgi.com>

. . .

> @@ -4806,60 +4805,57 @@ xfs_bmapi_convert_unwritten(
>  }
>  
>  /*
> - * Map file blocks to filesystem blocks.
> - * File range is given by the bno/len pair.
> - * Adds blocks to file if a write ("flags & XFS_BMAPI_WRITE" set)
> - * into a hole or past eof.
> - * Only allocates blocks from a single allocation group,
> - * to avoid locking problems.
> + * Map file blocks to filesystem blocks, and allocate blocks or convert the
> + * extent state if nessecary.  Details behaviour is controlled by the flags
                     necessary.  Detailed
> + * paramter.  Only allocates blocks from a single allocation group, to avoid
     parameter

> + * locking problems.
> + *


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

Re: [PATCH 13/25] xfs: introduce xfs_bmap_last_extent

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> plain text document attachment (xfs-simplify-xfs_bmap_isaeof)
> Add a common helper for finding the last extent in a file.
> 
> Largely based on a patch from Dave Chinner.


The new version of xfs_bmap_isaeof() no longer asserts
that the data fork is the one being operated on.  Why?

I have a couple of suggested comment clarifications,
and a suggested logic simplification below.

Reviewed-by: Alex Elder <aelder@sgi.com>

(Still reviewing the rest, but thought I'd fire off
a batch of the ones that I'm done with.)


> Signed-off-by: Christoph Hellwig <hch@lst.de>
> 
> Index: xfs/fs/xfs/xfs_bmap.c

. . .

> +
> +/*
> + * Check the last inode extent to determine whether this allocation will result
> + * in blocks being allocated at the end of the file. When we allocate new data
> + * blocks at the end of the file which do not start at the previous data block,
> + * we will try to align the new blocks at stripe unit boundaries.

It is not obvious what value will be returned when the
fork is empty, so mention that case in the comment header:
 
    Returns 0 in *aeof if the file (fork) is empty.

(Maybe you could explain why this is true at a more
abstract level though.)

> + */
> +STATIC int
> +xfs_bmap_isaeof(
> +	struct xfs_inode	*ip,
> +	xfs_fileoff_t		off,
> +	int			whichfork,
> +	char			*aeof)
> +{
> +	struct xfs_bmbt_irec	rec;
> +	int			is_empty;
> +	int			error;
> +
> +	*aeof = 0;
> +	error = xfs_bmap_last_extent(NULL, ip, whichfork, &rec, &is_empty);
> +	if (error || is_empty)
> +		return error;
> +
> +	/*
> +	 * Check we are allocating in the last extent (for delayed allocations)
> +	 * or past the last extent for non-delayed allocations.
> +	 */
> +	*aeof = (off >= rec.br_startoff &&
> +		 off < rec.br_startoff + rec.br_blockcount &&
> +		 isnullstartblock(rec.br_startblock)) ||
> +		off >= rec.br_startoff + rec.br_blockcount;

This logic could be shortened:

		*aeof = off >= rec.br_startoff + rec.br_blockcount ||
			(off >= rec.br_startoff &&
				 isnullstartblock(rec.br_startblock);

> +	return 0;
> +}
> +
> +/*
> + * Check if the endoff is outside the last extent. If so the caller will grow
> + * the allocation to a stripe unit boundary.

    All offsets are considered outside the end of file for an
    empty file (fork), so 1 is returned in *eof in that case.

> + */
> +int
> +xfs_bmap_eof(
> +	struct xfs_inode	*ip,
> +	xfs_fileoff_t		endoff,
> +	int			whichfork,
> +	int			*eof)
> +{
> +	struct xfs_bmbt_irec	rec;
> +	int			error;
> +
> +	error = xfs_bmap_last_extent(NULL, ip, whichfork, &rec, eof);
> +	if (error || *eof)
> +		return error;
> +
> +	*eof = endoff >= rec.br_startoff + rec.br_blockcount;
> +	return 0;
> +}
> +
. . .

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

Re: [PATCH 14/25] xfs: remove xfs_bmap_add_extent

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> plain text document attachment (xfs-bmapi-split-xfs_bmap_extent-2)
> There is no real need to the xfs_bmap_add_extent, as the callers know what
> kind of extents they need to it.  Removing it means duplicating the
> extents to btree conversion logic in three places, but overall it's still
> a lot less, and much simpler code.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

Not really a *lot* less code, but it is an improvement.

Simple question below (coding style) but this looks good.

Reviewed-by: Alex Elder <aelder@sgi.com>


. . .

>  
> +#ifdef DEBUG
> +STATIC void
> +xfs_bmap_check_leaf_extents(
> +	struct xfs_btree_cur	*cur,
> +	struct xfs_inode	*ip,
> +	int			whichfork);
> +#else
> +#define xfs_bmap_check_leaf_extents(cur, ip, whichfork)		do { } while (0)
> +#endif

Why do you use "do {} while (0)" rather than just an empty right
hand side? 

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

Re: [PATCH 15/25] xfs: pass bmalloca structure to xfs_bmap_isaeof

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> All the variables xfs_bmap_isaeof() is passed are contained within
> the xfs_bmalloca structure. Pass that instead.
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

This looks good.

Now that the transaction pointer is available in
xfs_bmap_isaeof(), it gets used in the call to
xfs_bmap_last_extent().  It looks to me like
this means btree block buffers will be added to
and removed from the transaction's item list
in xfs_bmap_read_extents(), and that list will
be scanned for these buffers in xfs_trans_read_buf()
(unlike before).

I don't question whether that's correct, but
is that desirable?  Would we be just as well
off *not* providing the transaction pointer?

Anyway:

Reviewed-by: Alex Elder <aelder@sgi.com>
 
> Index: xfs/fs/xfs/xfs_bmap.c
> ===================================================================
> --- xfs.orig/fs/xfs/xfs_bmap.c	2011-08-23 21:14:33.735424450 +0200
> +++ xfs/fs/xfs/xfs_bmap.c	2011-08-23 21:14:35.358748989 +0200
> @@ -3869,17 +3869,16 @@ xfs_bmap_last_extent(
>   */
>  STATIC int
>  xfs_bmap_isaeof(
> -	struct xfs_inode	*ip,
> -	xfs_fileoff_t		off,
> -	int			whichfork,
> -	char			*aeof)
> +	struct xfs_bmalloca	*bma,
> +	int			whichfork)
>  {
>  	struct xfs_bmbt_irec	rec;
>  	int			is_empty;
>  	int			error;
>  
> -	*aeof = 0;
> -	error = xfs_bmap_last_extent(NULL, ip, whichfork, &rec, &is_empty);
> +	bma->aeof = 0;
> +	error = xfs_bmap_last_extent(bma->tp, bma->ip, whichfork, &rec,
> +				     &is_empty);

Here is the spot I'm referring to.

>  	if (error || is_empty)
>  		return error;
>  


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

Re: [PATCH 16/25] xfs: move extent records into bmalloca structure

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> Rather that putting extent records on the stack and then pointing to
> them in the bmalloca structure which is in the same stack frame, put
> the extent records directly in the bmalloca structure. This reduces
> the number of args that need to be passed around.
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

Interesting that the only place that uses that
structure (defines an object of that type anyway)
is xfs_bmapi_write().

Looks good.

Reviewed-by: Alex Elder <aelder@sgi.com>

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

Re: [PATCH 17/25] xfs: move firstblock and bmap freelist cursor into bmalloca structure

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> Rather than passing the freeblock and freelist structure around, embed it
                          firstblock
> into the bmalloca structure and remove it from the function parameters.
> 
> This also enables the minleft parameter to be set only once in
> xfs_bmapi_write(), and the freelist cursor directly queried in
> xfs_bmapi_allocate to clear it when the lowspace algorithm is activated.
>     
>     Signed-off-by: Dave Chinner <dchinner@redhat.com>

A couple of comments:
- The struct xfs_bmalloca field "low" ought to be eliminated
  (or renamed "pad" or something), since it's now unused
  (->flist->xbf_low is used in its place).
- It might have been a bit simpler to review if using
  ->flist->xbf_low in place of ->low and ->minleft
  in place of minleft were done as separate, min-steps.

In any case, this looks good.

Reviewed-by: Alex Elder <aelder@sgi.com>



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

Re: [PATCH 18/25] xfs: move allocation ranges inode bmalloca structure

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> Most of the allocation functions take the offset length and block
> number as parameters for allocation. often they are modified so are
> passed by reference.  Make the version in the bmalloca structure the
> primary version and stop passing them around individually.
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

Lots of this patch is simply renaming xfs_bmalloca fields
"off" -> "offset", "rval" -> "blkno", and "alen" -> "length".
Those are very much appreciated changes.  I know there are
already a lot of patches in this series but it would have
been easier to see the "real" change had that part been
pulled out as a trivial extra one.

But it looks good, anyway.

Reviewed-by: Alex Elder <aelder@sgi.com>


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

Re: [PATCH 19/25] xfs: move btree cursor into bmalloca

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

Looks good.

Reviewed-by: Alex Elder <aelder@sgi.com>


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

Re: [PATCH 20/25] xfs: move lastx and nallocs into bmalloca

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

Looks good.

Reviewed-by: Alex Elder <aelder@sgi.com>


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

Re: [PATCH 21/25] xfs: move logflags into bmalloca

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> Signed-off-by: Christoph Hellwig <hch@lst.de>

Looks good.

Reviewed-by: Alex Elder <aelder@sgi.com>

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

Re: [PATCH 22/25] xfs: pass bmalloca to xfs_bmap_add_extent_delay_real

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> All the parameters passed to xfs_bmap_add_extent_delay_real() are in the
> xfs_bmalloca structure now. Just pass the bmalloca parameter to the
> function instead of 8 separate parameters.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

Looks good.  Maybe xfs_bmalloca field "got" could be renamed
but I can't think of a good one.

Reviewed-by: Alex Elder <aelder@sgi.com>

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

Re: [PATCH 23/25] xfs: pass bmalloca to xfs_bmap_add_extent_hole_real

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> All the parameters passed to xfs_bmap_add_extent_hole_real() are in the
> xfs_bmalloca structure now. Just pass the bmalloca parameter to the
> function instead of 8 separate parameters.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

Looks good.

Reviewed-by: Alex Elder <aelder@sgi.com>


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

Re: [PATCH 24/25] xfs: dont ignore error code from xfs_bmbt_update

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> Fix a case in xfs_bmap_add_extent_unwritten_real where we aren't passing
> the returned error on.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

Looks good.

Reviewed-by: Alex Elder <aelder@sgi.com>


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

Re: [PATCH 25/25] xfs: cleanup xfs_bmap.h

[Alex Elder]

On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> Convert all function prototypes to the short form used elsewhere, and
> remove duplicates of comments already placed at the function body.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

Sounds good to me.

Reviewed-by: Alex Elder <aelder@sgi.com>


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

Re: [PATCH 13/25] xfs: introduce xfs_bmap_last_extent

[Christoph Hellwig]

On Fri, Sep 09, 2011 at 03:23:46PM -0500, Alex Elder wrote:
> On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> > plain text document attachment (xfs-simplify-xfs_bmap_isaeof)
> > Add a common helper for finding the last extent in a file.
> > 
> > Largely based on a patch from Dave Chinner.
> 
> 
> The new version of xfs_bmap_isaeof() no longer asserts
> that the data fork is the one being operated on.  Why?

Because it would work perfectly fine for the attr fork, too - even
if it's fairly useless for it.

> fork is empty, so mention that case in the comment header:
>  
>     Returns 0 in *aeof if the file (fork) is empty.
> 
> (Maybe you could explain why this is true at a more
> abstract level though.)

I've done both.

> This logic could be shortened:
> 
> 		*aeof = off >= rec.br_startoff + rec.br_blockcount ||
> 			(off >= rec.br_startoff &&
> 				 isnullstartblock(rec.br_startblock);

Done (including a comment update)

> > +	return 0;
> > +}
> > +
> > +/*
> > + * Check if the endoff is outside the last extent. If so the caller will grow
> > + * the allocation to a stripe unit boundary.
> 
>     All offsets are considered outside the end of file for an
>     empty file (fork), so 1 is returned in *eof in that case.

Ok.

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

Re: [PATCH 14/25] xfs: remove xfs_bmap_add_extent

[Christoph Hellwig]

On Fri, Sep 09, 2011 at 06:55:49PM -0500, Alex Elder wrote:
> On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> > plain text document attachment (xfs-bmapi-split-xfs_bmap_extent-2)
> > There is no real need to the xfs_bmap_add_extent, as the callers know what
> > kind of extents they need to it.  Removing it means duplicating the
> > extents to btree conversion logic in three places, but overall it's still
> > a lot less, and much simpler code.
> > 
> > Signed-off-by: Christoph Hellwig <hch@lst.de>
> 
> Not really a *lot* less code, but it is an improvement.
> 
> Simple question below (coding style) but this looks good.

It's about 70 lines.  I've toned the description down a bit for the next
version.

> > +STATIC void
> > +xfs_bmap_check_leaf_extents(
> > +	struct xfs_btree_cur	*cur,
> > +	struct xfs_inode	*ip,
> > +	int			whichfork);
> > +#else
> > +#define xfs_bmap_check_leaf_extents(cur, ip, whichfork)		do { } while (0)
> > +#endif
> 
> Why do you use "do {} while (0)" rather than just an empty right
> hand side? 

That way it can safely be used e.g. inside single-line conditionals
without braces. It's an idiom used quite a lot in the kernel.

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

Re: [PATCH 15/25] xfs: pass bmalloca structure to xfs_bmap_isaeof

[Christoph Hellwig]

On Fri, Sep 09, 2011 at 06:55:56PM -0500, Alex Elder wrote:
> On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> > All the variables xfs_bmap_isaeof() is passed are contained within
> > the xfs_bmalloca structure. Pass that instead.
> > 
> > Signed-off-by: Dave Chinner <dchinner@redhat.com>
> 
> This looks good.
> 
> Now that the transaction pointer is available in
> xfs_bmap_isaeof(), it gets used in the call to
> xfs_bmap_last_extent().  It looks to me like
> this means btree block buffers will be added to
> and removed from the transaction's item list
> in xfs_bmap_read_extents(), and that list will
> be scanned for these buffers in xfs_trans_read_buf()
> (unlike before).
> 
> I don't question whether that's correct, but
> is that desirable?  Would we be just as well
> off *not* providing the transaction pointer?

We shouldn't do it, if just to avoid random changes in this patch.

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

Re: [PATCH 18/25] xfs: move allocation ranges inode bmalloca structure

[Christoph Hellwig]

On Fri, Sep 09, 2011 at 06:56:16PM -0500, Alex Elder wrote:
> Lots of this patch is simply renaming xfs_bmalloca fields
> "off" -> "offset", "rval" -> "blkno", and "alen" -> "length".
> Those are very much appreciated changes.  I know there are
> already a lot of patches in this series but it would have
> been easier to see the "real" change had that part been
> pulled out as a trivial extra one.

I'll look into it.

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

Re: [PATCH 10/25] xfs: factor extent allocation out of xfs_bmapi

[Christoph Hellwig]

On Fri, Sep 09, 2011 at 03:23:24PM -0500, Alex Elder wrote:
> On Wed, 2011-08-24 at 02:04 -0400, Christoph Hellwig wrote:
> > To further improve the readability of xfs_bmapi(), factor the extent
> > allocation out into a separate function. This removes a large block
> > of logic from the xfs_bmapi() code loop and makes it easier to see
> > the operational logic flow for xfs_bmapi().
> > 
> > Signed-off-by: Dave Chinner <dchinner@redhat.com>
> 
> OK, this looks very good.  I have a spot that I chased
> for a while to verify it produced the same functionality
> as before, but I just gave up because it was just taking
> much too much time.  I'll point it out below, just for
> the record, but I'm not too concerned about it.  Everything
> else looks good.
> 
> Reviewed-by: Alex Elder <aelder@sgi.com>
> 

> > +			bma.minleft = minleft;
> > +
> > +			error = xfs_bmapi_allocate(&bma, &lastx, &cur,
> > +					firstblock, flist, flags, &nallocs,
> > +					&tmp_logflags);
> 
> > +			if (error == ENOSPC || error == EDQUOT) {
> > +				if (n == 0) {
> > +					*nmap = 0;
> > +					ASSERT(cur == NULL);
> > +					return error;
> >  				}
> 
> Here is the spot I mentioned above.  I was trying to find out
> the circumstances under which ENOSPC or EDQUOT could get returned
> by xfs_bmapi_allocate() in order to confirm that this in fact
> produces the same effect as before.

It shouldn't be there, and never hit.  This is a leftover from Dave's
earlier version where we did the delalloc reservation from
xfs_bmapi_allocate.  I have removed it.

> I also had a little trouble because there were spots--such
> as calling xfs_bmap_isaeof()--that are now encapsulated within
> xfs_bmapi_allocate() that previously jumped to error0, but now
> will produce an error return from that function.  So now this
> doesn't execute the code at error0 in this case.  I didn't
> work through it but I trust that the code there would end
> up being a series of no-ops anyway.

We still got to error0 in that case, it's just below the code you
quoted:

			logflags |= tmp_logflags;
			if (error)
				goto error0;

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