[PATCH v2 0/3] discontiguous buffer patches

[PATCH v2 0/3] discontiguous buffer patches

[Mark Tinguely]

Eric Sundeen's "userspace vs. fragmented multiblock dir2", xfstest 291
commit 2a4ed, causes a xfs_buf lock hang:

[<ffffffff81065d87>] down+0x47/0x50
[<ffffffffa03a25c6>] xfs_buf_lock+0x66/0xe0 [xfs]
[<ffffffffa03a33c9>] _xfs_buf_find+0x1f9/0x320 [xfs]
[<ffffffffa03a393f>] xfs_buf_get_map+0x2f/0x170 [xfs]
[<ffffffffa03a3f7a>] xfs_buf_read_map+0x2a/0x100 [xfs]
[<ffffffffa0411630>] xfs_trans_read_buf_map+0x3b0/0x590 [xfs]
[<ffffffffa03e1c5e>] xfs_da_read_buf+0xbe/0x230 [xfs]
[<ffffffffa03e78dc>] xfs_dir2_block_addname+0x7c/0x980 [xfs]
[<ffffffffa03f1468>] xfs_dir2_sf_addname+0x3e8/0x450 [xfs]
[<ffffffffa03e634c>] xfs_dir_createname+0x17c/0x1e0 [xfs]
[<ffffffffa03b9fe2>] xfs_create+0x4c2/0x5f0 [xfs]
[<ffffffffa03b0c8a>] xfs_vn_mknod+0x8a/0x1a0 [xfs]
[<ffffffffa03b0dce>] xfs_vn_create+0xe/0x10 [xfs]
[<ffffffff8115695c>] vfs_create+0xac/0xd0
[<ffffffff811587de>] do_last+0x8be/0x960
[<ffffffff8115940c>] path_openat+0xdc/0x410
[<ffffffff81159853>] do_filp_open+0x43/0xa0
[<ffffffff8114a502>] do_sys_open+0x152/0x1e0
[<ffffffff8114a5cc>] sys_open+0x1c/0x20
[<ffffffff81423df9>] system_call_fastpath+0x16/0x1b
[<ffffffffffffffff>] 0xffffffffffffffff

That bisect a problem to the discontiguous xfs_buf support series. 
xfs_trans_buf_item_match() is looking for the block number of the
buffer in the single segment map area. After fixing the b_map issue,
there were couple asserts associated with the buffer log format
not correctly using the bli_formats[] structure.

Changes to the series from version 1:

Patch 1: Moved b_map structure to __b_map to avoid future confusion
	  per Christoph's suggestion.

Patch 2: Moved bli_format structure to __bli_format to avoid future
	  confusion per Christoph's suggestion.

	 Removed the buffer XFS_TRANS_DEBUG routines per Dave's suggestion.

Patch 3: Moved the detection of an empty format to a new file and
	  add parenthesis per Dave's suggestion.

	 Cleared all buffer log formats per Dave's suggestion.

--Mark.

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

[PATCH v2 1/3] xfs: use b_maps[] for discontiguous buffers

[Mark Tinguely]

[-- Attachment #1: 1-3-xfs-use-xfs_buf-b_maps.patch --]
[-- Type: text/plain, Size: 3436 bytes --]

Commits starting at 77c1a08 introduced a multiple segment support
to xfs_buf. xfs_trans_buf_item_match() could not find a multi-segment
buffer in the transaction because it was looking at the single segment
block number rather than the multi-segment b_maps[0].bm.bn. This
results on a recursive buffer lock that can never be satisfied.

This patch:
 1) Changed the remaining b_map accesses to be b_maps[0] accesses.
 2) Renames the single segment b_map structure to __b_map to avoid
    future confusion.

Signed-off-by: Mark Tinguely <tinguely@sgi.com>

---
 fs/xfs/xfs_buf.c |   12 ++++++------
 fs/xfs/xfs_buf.h |    6 +++---
 2 files changed, 9 insertions(+), 9 deletions(-)

Index: b/fs/xfs/xfs_buf.c
===================================================================
--- a/fs/xfs/xfs_buf.c
+++ b/fs/xfs/xfs_buf.c
@@ -175,7 +175,7 @@ xfs_buf_get_maps(
 	bp->b_map_count = map_count;
 
 	if (map_count == 1) {
-		bp->b_maps = &bp->b_map;
+		bp->b_maps = &bp->__b_map;
 		return 0;
 	}
 
@@ -193,7 +193,7 @@ static void
 xfs_buf_free_maps(
 	struct xfs_buf	*bp)
 {
-	if (bp->b_maps != &bp->b_map) {
+	if (bp->b_maps != &bp->__b_map) {
 		kmem_free(bp->b_maps);
 		bp->b_maps = NULL;
 	}
@@ -377,8 +377,8 @@ xfs_buf_allocate_memory(
 	}
 
 use_alloc_page:
-	start = BBTOB(bp->b_map.bm_bn) >> PAGE_SHIFT;
-	end = (BBTOB(bp->b_map.bm_bn + bp->b_length) + PAGE_SIZE - 1)
+	start = BBTOB(bp->b_maps[0].bm_bn) >> PAGE_SHIFT;
+	end = (BBTOB(bp->b_maps[0].bm_bn + bp->b_length) + PAGE_SIZE - 1)
 								>> PAGE_SHIFT;
 	page_count = end - start;
 	error = _xfs_buf_get_pages(bp, page_count, flags);
@@ -640,7 +640,7 @@ _xfs_buf_read(
 	xfs_buf_flags_t		flags)
 {
 	ASSERT(!(flags & XBF_WRITE));
-	ASSERT(bp->b_map.bm_bn != XFS_BUF_DADDR_NULL);
+	ASSERT(bp->b_maps[0].bm_bn != XFS_BUF_DADDR_NULL);
 
 	bp->b_flags &= ~(XBF_WRITE | XBF_ASYNC | XBF_READ_AHEAD);
 	bp->b_flags |= flags & (XBF_READ | XBF_ASYNC | XBF_READ_AHEAD);
@@ -1709,7 +1709,7 @@ xfs_buf_cmp(
 	struct xfs_buf	*bp = container_of(b, struct xfs_buf, b_list);
 	xfs_daddr_t		diff;
 
-	diff = ap->b_map.bm_bn - bp->b_map.bm_bn;
+	diff = ap->b_maps[0].bm_bn - bp->b_maps[0].bm_bn;
 	if (diff < 0)
 		return -1;
 	if (diff > 0)
Index: b/fs/xfs/xfs_buf.h
===================================================================
--- a/fs/xfs/xfs_buf.h
+++ b/fs/xfs/xfs_buf.h
@@ -151,7 +151,7 @@ typedef struct xfs_buf {
 	struct page		**b_pages;	/* array of page pointers */
 	struct page		*b_page_array[XB_PAGES]; /* inline pages */
 	struct xfs_buf_map	*b_maps;	/* compound buffer map */
-	struct xfs_buf_map	b_map;		/* inline compound buffer map */
+	struct xfs_buf_map	__b_map;	/* inline compound buffer map */
 	int			b_map_count;
 	int			b_io_length;	/* IO size in BBs */
 	atomic_t		b_pin_count;	/* pin count */
@@ -330,8 +330,8 @@ void xfs_buf_stale(struct xfs_buf *bp);
  * In future, uncached buffers will pass the block number directly to the io
  * request function and hence these macros will go away at that point.
  */
-#define XFS_BUF_ADDR(bp)		((bp)->b_map.bm_bn)
-#define XFS_BUF_SET_ADDR(bp, bno)	((bp)->b_map.bm_bn = (xfs_daddr_t)(bno))
+#define XFS_BUF_ADDR(bp)		((bp)->b_maps[0].bm_bn)
+#define XFS_BUF_SET_ADDR(bp, bno)	((bp)->b_maps[0].bm_bn = (xfs_daddr_t)(bno))
 
 static inline void xfs_buf_set_ref(struct xfs_buf *bp, int lru_ref)
 {


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

[PATCH v2 2/3] xfs: fix the buffer log format for contiguous buffers

[Mark Tinguely]

[-- Attachment #1: 2-3-xfs-fix-blf-for-multiple-segment.patch --]
[-- Type: text/plain, Size: 11833 bytes --]

Not every segment in a multi-segment buffer is dirty in a
transaction and they will not be outputted. The assert in
xfs_buf_item_format_segment() that checks for the at least
one chunk of data in the segment to be used is not necessary
true for multi-segmented buffers.

This patch:

 1) Change xfs_buf_item_format_segment() to skip over non-dirty
    segments.
 2) Change the bli_format structure to __bli_format so it is not
    accidently confused with the bli_formats pointer.
 3) Remove the buffer XFS_TRANS_DEBUG routines xfs_buf_item_log_debug()
    and xfs_buf_item_log_check(). They are not used and are not
    multi-segment aware.

Signed-off-by: Mark Tinguely <tinguely@sgi.com>

---
 fs/xfs/xfs_buf_item.c  |  151 +++++++++----------------------------------------
 fs/xfs/xfs_buf_item.h  |    2 
 fs/xfs/xfs_trans_buf.c |   24 +++----
 3 files changed, 41 insertions(+), 136 deletions(-)

Index: b/fs/xfs/xfs_buf_item.c
===================================================================
--- a/fs/xfs/xfs_buf_item.c
+++ b/fs/xfs/xfs_buf_item.c
@@ -37,109 +37,6 @@ static inline struct xfs_buf_log_item *B
 	return container_of(lip, struct xfs_buf_log_item, bli_item);
 }
 
-
-#ifdef XFS_TRANS_DEBUG
-/*
- * This function uses an alternate strategy for tracking the bytes
- * that the user requests to be logged.  This can then be used
- * in conjunction with the bli_orig array in the buf log item to
- * catch bugs in our callers' code.
- *
- * We also double check the bits set in xfs_buf_item_log using a
- * simple algorithm to check that every byte is accounted for.
- */
-STATIC void
-xfs_buf_item_log_debug(
-	xfs_buf_log_item_t	*bip,
-	uint			first,
-	uint			last)
-{
-	uint	x;
-	uint	byte;
-	uint	nbytes;
-	uint	chunk_num;
-	uint	word_num;
-	uint	bit_num;
-	uint	bit_set;
-	uint	*wordp;
-
-	ASSERT(bip->bli_logged != NULL);
-	byte = first;
-	nbytes = last - first + 1;
-	bfset(bip->bli_logged, first, nbytes);
-	for (x = 0; x < nbytes; x++) {
-		chunk_num = byte >> XFS_BLF_SHIFT;
-		word_num = chunk_num >> BIT_TO_WORD_SHIFT;
-		bit_num = chunk_num & (NBWORD - 1);
-		wordp = &(bip->bli_format.blf_data_map[word_num]);
-		bit_set = *wordp & (1 << bit_num);
-		ASSERT(bit_set);
-		byte++;
-	}
-}
-
-/*
- * This function is called when we flush something into a buffer without
- * logging it.  This happens for things like inodes which are logged
- * separately from the buffer.
- */
-void
-xfs_buf_item_flush_log_debug(
-	xfs_buf_t	*bp,
-	uint		first,
-	uint		last)
-{
-	xfs_buf_log_item_t	*bip = bp->b_fspriv;
-	uint			nbytes;
-
-	if (bip == NULL || (bip->bli_item.li_type != XFS_LI_BUF))
-		return;
-
-	ASSERT(bip->bli_logged != NULL);
-	nbytes = last - first + 1;
-	bfset(bip->bli_logged, first, nbytes);
-}
-
-/*
- * This function is called to verify that our callers have logged
- * all the bytes that they changed.
- *
- * It does this by comparing the original copy of the buffer stored in
- * the buf log item's bli_orig array to the current copy of the buffer
- * and ensuring that all bytes which mismatch are set in the bli_logged
- * array of the buf log item.
- */
-STATIC void
-xfs_buf_item_log_check(
-	xfs_buf_log_item_t	*bip)
-{
-	char		*orig;
-	char		*buffer;
-	int		x;
-	xfs_buf_t	*bp;
-
-	ASSERT(bip->bli_orig != NULL);
-	ASSERT(bip->bli_logged != NULL);
-
-	bp = bip->bli_buf;
-	ASSERT(bp->b_length > 0);
-	ASSERT(bp->b_addr != NULL);
-	orig = bip->bli_orig;
-	buffer = bp->b_addr;
-	for (x = 0; x < BBTOB(bp->b_length); x++) {
-		if (orig[x] != buffer[x] && !btst(bip->bli_logged, x)) {
-			xfs_emerg(bp->b_mount,
-				"%s: bip %x buffer %x orig %x index %d",
-				__func__, bip, bp, orig, x);
-			ASSERT(0);
-		}
-	}
-}
-#else
-#define		xfs_buf_item_log_debug(x,y,z)
-#define		xfs_buf_item_log_check(x)
-#endif
-
 STATIC void	xfs_buf_do_callbacks(struct xfs_buf *bp);
 
 /*
@@ -237,7 +134,7 @@ xfs_buf_item_size(
 		 * cancel flag in it.
 		 */
 		trace_xfs_buf_item_size_stale(bip);
-		ASSERT(bip->bli_format.blf_flags & XFS_BLF_CANCEL);
+		ASSERT(bip->__bli_format.blf_flags & XFS_BLF_CANCEL);
 		return bip->bli_format_count;
 	}
 
@@ -278,7 +175,7 @@ xfs_buf_item_format_segment(
 	uint		buffer_offset;
 
 	/* copy the flags across from the base format item */
-	blfp->blf_flags = bip->bli_format.blf_flags;
+	blfp->blf_flags = bip->__bli_format.blf_flags;
 
 	/*
 	 * Base size is the actual size of the ondisk structure - it reflects
@@ -287,11 +184,6 @@ xfs_buf_item_format_segment(
 	 */
 	base_size = offsetof(struct xfs_buf_log_format, blf_data_map) +
 			(blfp->blf_map_size * sizeof(blfp->blf_data_map[0]));
-	vecp->i_addr = blfp;
-	vecp->i_len = base_size;
-	vecp->i_type = XLOG_REG_TYPE_BFORMAT;
-	vecp++;
-	nvecs = 1;
 
 	if (bip->bli_flags & XFS_BLI_STALE) {
 		/*
@@ -301,7 +193,11 @@ xfs_buf_item_format_segment(
 		 */
 		trace_xfs_buf_item_format_stale(bip);
 		ASSERT(blfp->blf_flags & XFS_BLF_CANCEL);
-		blfp->blf_size = nvecs;
+		vecp->i_addr = blfp;
+		vecp->i_len = base_size;
+		vecp->i_type = XLOG_REG_TYPE_BFORMAT;
+		vecp++;
+		blfp->blf_size = 1;
 		return vecp;
 	}
 
@@ -309,7 +205,19 @@ xfs_buf_item_format_segment(
 	 * Fill in an iovec for each set of contiguous chunks.
 	 */
 	first_bit = xfs_next_bit(blfp->blf_data_map, blfp->blf_map_size, 0);
-	ASSERT(first_bit != -1);
+	if (first_bit == -1) {
+		/* If the map is not be dirty in the transaction, mark
+		 * the size as zero and do not advance the vector pointer.
+		 */
+		blfp->blf_size = 0;
+		return(vecp);
+	}
+
+	vecp->i_addr = blfp;
+	vecp->i_len = base_size;
+	vecp->i_type = XLOG_REG_TYPE_BFORMAT;
+	vecp++;
+	nvecs = 1;
 	last_bit = first_bit;
 	nbits = 1;
 	for (;;) {
@@ -371,7 +279,7 @@ xfs_buf_item_format_segment(
 			nbits++;
 		}
 	}
-	bip->bli_format.blf_size = nvecs;
+	blfp->blf_size = nvecs;
 	return vecp;
 }
 
@@ -405,7 +313,7 @@ xfs_buf_item_format(
 	if (bip->bli_flags & XFS_BLI_INODE_BUF) {
 		if (!((bip->bli_flags & XFS_BLI_INODE_ALLOC_BUF) &&
 		      xfs_log_item_in_current_chkpt(lip)))
-			bip->bli_format.blf_flags |= XFS_BLF_INODE_BUF;
+			bip->__bli_format.blf_flags |= XFS_BLF_INODE_BUF;
 		bip->bli_flags &= ~XFS_BLI_INODE_BUF;
 	}
 
@@ -419,7 +327,6 @@ xfs_buf_item_format(
 	 * Check to make sure everything is consistent.
 	 */
 	trace_xfs_buf_item_format(bip);
-	xfs_buf_item_log_check(bip);
 }
 
 /*
@@ -485,7 +392,7 @@ xfs_buf_item_unpin(
 		ASSERT(bip->bli_flags & XFS_BLI_STALE);
 		ASSERT(xfs_buf_islocked(bp));
 		ASSERT(XFS_BUF_ISSTALE(bp));
-		ASSERT(bip->bli_format.blf_flags & XFS_BLF_CANCEL);
+		ASSERT(bip->__bli_format.blf_flags & XFS_BLF_CANCEL);
 
 		trace_xfs_buf_item_unpin_stale(bip);
 
@@ -631,7 +538,7 @@ xfs_buf_item_unlock(
 	 */
 	if (bip->bli_flags & XFS_BLI_STALE) {
 		trace_xfs_buf_item_unlock_stale(bip);
-		ASSERT(bip->bli_format.blf_flags & XFS_BLF_CANCEL);
+		ASSERT(bip->__bli_format.blf_flags & XFS_BLF_CANCEL);
 		if (!aborted) {
 			atomic_dec(&bip->bli_refcount);
 			return;
@@ -644,8 +551,8 @@ xfs_buf_item_unlock(
 	 * If the buf item isn't tracking any data, free it, otherwise drop the
 	 * reference we hold to it.
 	 */
-	if (xfs_bitmap_empty(bip->bli_format.blf_data_map,
-			     bip->bli_format.blf_map_size))
+	if (xfs_bitmap_empty(bip->__bli_format.blf_data_map,
+			     bip->__bli_format.blf_map_size))
 		xfs_buf_item_relse(bp);
 	else
 		atomic_dec(&bip->bli_refcount);
@@ -716,7 +623,7 @@ xfs_buf_item_get_format(
 	bip->bli_format_count = count;
 
 	if (count == 1) {
-		bip->bli_formats = &bip->bli_format;
+		bip->bli_formats = &bip->__bli_format;
 		return 0;
 	}
 
@@ -731,7 +638,7 @@ STATIC void
 xfs_buf_item_free_format(
 	struct xfs_buf_log_item	*bip)
 {
-	if (bip->bli_formats != &bip->bli_format) {
+	if (bip->bli_formats != &bip->__bli_format) {
 		kmem_free(bip->bli_formats);
 		bip->bli_formats = NULL;
 	}
@@ -898,8 +805,6 @@ xfs_buf_item_log_segment(
 		mask = (1 << end_bit) - 1;
 		*wordp |= mask;
 	}
-
-	xfs_buf_item_log_debug(bip, first, last);
 }
 
 /*
Index: b/fs/xfs/xfs_buf_item.h
===================================================================
--- a/fs/xfs/xfs_buf_item.h
+++ b/fs/xfs/xfs_buf_item.h
@@ -104,7 +104,7 @@ typedef struct xfs_buf_log_item {
 #endif
 	int			bli_format_count;	/* count of headers */
 	struct xfs_buf_log_format *bli_formats;	/* array of in-log header ptrs */
-	struct xfs_buf_log_format bli_format;	/* embedded in-log header */
+	struct xfs_buf_log_format __bli_format;	/* embedded in-log header */
 } xfs_buf_log_item_t;
 
 void	xfs_buf_item_init(struct xfs_buf *, struct xfs_mount *);
Index: b/fs/xfs/xfs_trans_buf.c
===================================================================
--- a/fs/xfs/xfs_trans_buf.c
+++ b/fs/xfs/xfs_trans_buf.c
@@ -93,7 +93,7 @@ _xfs_trans_bjoin(
 	xfs_buf_item_init(bp, tp->t_mountp);
 	bip = bp->b_fspriv;
 	ASSERT(!(bip->bli_flags & XFS_BLI_STALE));
-	ASSERT(!(bip->bli_format.blf_flags & XFS_BLF_CANCEL));
+	ASSERT(!(bip->__bli_format.blf_flags & XFS_BLF_CANCEL));
 	ASSERT(!(bip->bli_flags & XFS_BLI_LOGGED));
 	if (reset_recur)
 		bip->bli_recur = 0;
@@ -432,7 +432,7 @@ xfs_trans_brelse(xfs_trans_t	*tp,
 	bip = bp->b_fspriv;
 	ASSERT(bip->bli_item.li_type == XFS_LI_BUF);
 	ASSERT(!(bip->bli_flags & XFS_BLI_STALE));
-	ASSERT(!(bip->bli_format.blf_flags & XFS_BLF_CANCEL));
+	ASSERT(!(bip->__bli_format.blf_flags & XFS_BLF_CANCEL));
 	ASSERT(atomic_read(&bip->bli_refcount) > 0);
 
 	trace_xfs_trans_brelse(bip);
@@ -519,7 +519,7 @@ xfs_trans_bhold(xfs_trans_t	*tp,
 	ASSERT(bp->b_transp == tp);
 	ASSERT(bip != NULL);
 	ASSERT(!(bip->bli_flags & XFS_BLI_STALE));
-	ASSERT(!(bip->bli_format.blf_flags & XFS_BLF_CANCEL));
+	ASSERT(!(bip->__bli_format.blf_flags & XFS_BLF_CANCEL));
 	ASSERT(atomic_read(&bip->bli_refcount) > 0);
 
 	bip->bli_flags |= XFS_BLI_HOLD;
@@ -539,7 +539,7 @@ xfs_trans_bhold_release(xfs_trans_t	*tp,
 	ASSERT(bp->b_transp == tp);
 	ASSERT(bip != NULL);
 	ASSERT(!(bip->bli_flags & XFS_BLI_STALE));
-	ASSERT(!(bip->bli_format.blf_flags & XFS_BLF_CANCEL));
+	ASSERT(!(bip->__bli_format.blf_flags & XFS_BLF_CANCEL));
 	ASSERT(atomic_read(&bip->bli_refcount) > 0);
 	ASSERT(bip->bli_flags & XFS_BLI_HOLD);
 
@@ -598,7 +598,7 @@ xfs_trans_log_buf(xfs_trans_t	*tp,
 		bip->bli_flags &= ~XFS_BLI_STALE;
 		ASSERT(XFS_BUF_ISSTALE(bp));
 		XFS_BUF_UNSTALE(bp);
-		bip->bli_format.blf_flags &= ~XFS_BLF_CANCEL;
+		bip->__bli_format.blf_flags &= ~XFS_BLF_CANCEL;
 	}
 
 	tp->t_flags |= XFS_TRANS_DIRTY;
@@ -657,8 +657,8 @@ xfs_trans_binval(
 		 */
 		ASSERT(XFS_BUF_ISSTALE(bp));
 		ASSERT(!(bip->bli_flags & (XFS_BLI_LOGGED | XFS_BLI_DIRTY)));
-		ASSERT(!(bip->bli_format.blf_flags & XFS_BLF_INODE_BUF));
-		ASSERT(bip->bli_format.blf_flags & XFS_BLF_CANCEL);
+		ASSERT(!(bip->__bli_format.blf_flags & XFS_BLF_INODE_BUF));
+		ASSERT(bip->__bli_format.blf_flags & XFS_BLF_CANCEL);
 		ASSERT(bip->bli_item.li_desc->lid_flags & XFS_LID_DIRTY);
 		ASSERT(tp->t_flags & XFS_TRANS_DIRTY);
 		return;
@@ -668,10 +668,10 @@ xfs_trans_binval(
 
 	bip->bli_flags |= XFS_BLI_STALE;
 	bip->bli_flags &= ~(XFS_BLI_INODE_BUF | XFS_BLI_LOGGED | XFS_BLI_DIRTY);
-	bip->bli_format.blf_flags &= ~XFS_BLF_INODE_BUF;
-	bip->bli_format.blf_flags |= XFS_BLF_CANCEL;
-	memset((char *)(bip->bli_format.blf_data_map), 0,
-	      (bip->bli_format.blf_map_size * sizeof(uint)));
+	bip->__bli_format.blf_flags &= ~XFS_BLF_INODE_BUF;
+	bip->__bli_format.blf_flags |= XFS_BLF_CANCEL;
+	memset((char *)(bip->__bli_format.blf_data_map), 0,
+	      (bip->__bli_format.blf_map_size * sizeof(uint)));
 	bip->bli_item.li_desc->lid_flags |= XFS_LID_DIRTY;
 	tp->t_flags |= XFS_TRANS_DIRTY;
 }
@@ -775,5 +775,5 @@ xfs_trans_dquot_buf(
 	       type == XFS_BLF_GDQUOT_BUF);
 	ASSERT(atomic_read(&bip->bli_refcount) > 0);
 
-	bip->bli_format.blf_flags |= type;
+	bip->__bli_format.blf_flags |= type;
 }


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

[PATCH v2 3/3] xfs: fix the multi-segment log buffer format

[Mark Tinguely]

[-- Attachment #1: 3-3-xfs-extend-blf-multiple-segment.patch --]
[-- Type: text/plain, Size: 2362 bytes --]

Per Dave Chinner suggestion, this patch:
 1) Corrects the detection of whether a multi-segment buffer is
    still tracking data.
 2) Clears all the buffer log formats for a multi-segment buffer.

Signed-off-by: Mark Tinguely <tinguely@sgi.com>

---
 fs/xfs/xfs_buf_item.c  |   13 ++++++++++---
 fs/xfs/xfs_trans_buf.c |    7 +++++--
 2 files changed, 15 insertions(+), 5 deletions(-)

Index: b/fs/xfs/xfs_buf_item.c
===================================================================
--- a/fs/xfs/xfs_buf_item.c
+++ b/fs/xfs/xfs_buf_item.c
@@ -508,7 +508,7 @@ xfs_buf_item_unlock(
 {
 	struct xfs_buf_log_item	*bip = BUF_ITEM(lip);
 	struct xfs_buf		*bp = bip->bli_buf;
-	int			aborted;
+	int			aborted, empty, i;
 	uint			hold;
 
 	/* Clear the buffer's association with this transaction. */
@@ -551,8 +551,15 @@ xfs_buf_item_unlock(
 	 * If the buf item isn't tracking any data, free it, otherwise drop the
 	 * reference we hold to it.
 	 */
-	if (xfs_bitmap_empty(bip->__bli_format.blf_data_map,
-			     bip->__bli_format.blf_map_size))
+	empty = 1;
+	for (i = 0; i < bip->bli_format_count; i++) {
+		if (!xfs_bitmap_empty(bip->bli_formats[i].blf_data_map,
+			     bip->bli_formats[i].blf_map_size)) {
+			empty = 0;
+			break;
+		}
+	}
+	if (empty)
 		xfs_buf_item_relse(bp);
 	else
 		atomic_dec(&bip->bli_refcount);
Index: b/fs/xfs/xfs_trans_buf.c
===================================================================
--- a/fs/xfs/xfs_trans_buf.c
+++ b/fs/xfs/xfs_trans_buf.c
@@ -643,6 +643,7 @@ xfs_trans_binval(
 	xfs_buf_t	*bp)
 {
 	xfs_buf_log_item_t	*bip = bp->b_fspriv;
+	int			i;
 
 	ASSERT(bp->b_transp == tp);
 	ASSERT(bip != NULL);
@@ -670,8 +671,10 @@ xfs_trans_binval(
 	bip->bli_flags &= ~(XFS_BLI_INODE_BUF | XFS_BLI_LOGGED | XFS_BLI_DIRTY);
 	bip->__bli_format.blf_flags &= ~XFS_BLF_INODE_BUF;
 	bip->__bli_format.blf_flags |= XFS_BLF_CANCEL;
-	memset((char *)(bip->__bli_format.blf_data_map), 0,
-	      (bip->__bli_format.blf_map_size * sizeof(uint)));
+	for (i = 0; i < bip->bli_format_count; i++) {
+		memset((char *)(bip->bli_formats[i].blf_data_map), 0,
+		       (bip->bli_formats[i].blf_map_size * sizeof(uint)));
+	}
 	bip->bli_item.li_desc->lid_flags |= XFS_LID_DIRTY;
 	tp->t_flags |= XFS_TRANS_DIRTY;
 }


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

Re: [PATCH v2 1/3] xfs: use b_maps[] for discontiguous buffers

[Christoph Hellwig]

>  use_alloc_page:
> -	start = BBTOB(bp->b_map.bm_bn) >> PAGE_SHIFT;
> -	end = (BBTOB(bp->b_map.bm_bn + bp->b_length) + PAGE_SIZE - 1)
> +	start = BBTOB(bp->b_maps[0].bm_bn) >> PAGE_SHIFT;
> +	end = (BBTOB(bp->b_maps[0].bm_bn + bp->b_length) + PAGE_SIZE - 1)

Btw, how is this logic supposed to work for discontiguous buffers?

Each of them might straddle boundaries individually, so doing the
start/end calculation for the number of pages isn't going to be correct.

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

Re: [PATCH v2 1/3] xfs: use b_maps[] for discontiguous buffers

[Mark Tinguely]

On 11/30/12 10:09, Christoph Hellwig wrote:
>>   use_alloc_page:
>> -	start = BBTOB(bp->b_map.bm_bn)>>  PAGE_SHIFT;
>> -	end = (BBTOB(bp->b_map.bm_bn + bp->b_length) + PAGE_SIZE - 1)
>> +	start = BBTOB(bp->b_maps[0].bm_bn)>>  PAGE_SHIFT;
>> +	end = (BBTOB(bp->b_maps[0].bm_bn + bp->b_length) + PAGE_SIZE - 1)
>
> Btw, how is this logic supposed to work for discontiguous buffers?
>
> Each of them might straddle boundaries individually, so doing the
> start/end calculation for the number of pages isn't going to be correct.
>

eeek. yep, I will loop through and count the pages needed in each
segment.

--Mark.

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

Re: [PATCH v2 1/3] xfs: use b_maps[] for discontiguous buffers

[Dave Chinner]

On Fri, Nov 30, 2012 at 11:09:20AM -0500, Christoph Hellwig wrote:
> >  use_alloc_page:
> > -	start = BBTOB(bp->b_map.bm_bn) >> PAGE_SHIFT;
> > -	end = (BBTOB(bp->b_map.bm_bn + bp->b_length) + PAGE_SIZE - 1)
> > +	start = BBTOB(bp->b_maps[0].bm_bn) >> PAGE_SHIFT;
> > +	end = (BBTOB(bp->b_maps[0].bm_bn + bp->b_length) + PAGE_SIZE - 1)
> 
> Btw, how is this logic supposed to work for discontiguous buffers?
> 
> Each of them might straddle boundaries individually, so doing the
> start/end calculation for the number of pages isn't going to be correct.

AFAICT it is correct - the first page is always aligned to the start
of the buffer (i.e. b_offset = 0), and the end is extended to ensure
that it covers any partial tail page. e.g:

Page	0		1		2
	+---------------+---------------+---------------+
map[0]	|----------|
map[1]		   |------------|
map[2]				|--------------|
        |---------------bp->b_length-----------|
	|						|
	start						end

Maps are always multiples of basic blocks, so every segment boundary
aligns to something that IO can be done on, and the array of pages
always spans the entire range of the maps.

Hence I'm not sure what the problem is here, Christoph. Can you
expand on the issue?

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

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

Re: [PATCH v2 1/3] xfs: use b_maps[] for discontiguous buffers

[Dave Chinner]

On Fri, Nov 30, 2012 at 11:55:08AM -0600, Mark Tinguely wrote:
> On 11/30/12 10:09, Christoph Hellwig wrote:
> >>  use_alloc_page:
> >>-	start = BBTOB(bp->b_map.bm_bn)>>  PAGE_SHIFT;
> >>-	end = (BBTOB(bp->b_map.bm_bn + bp->b_length) + PAGE_SIZE - 1)
> >>+	start = BBTOB(bp->b_maps[0].bm_bn)>>  PAGE_SHIFT;
> >>+	end = (BBTOB(bp->b_maps[0].bm_bn + bp->b_length) + PAGE_SIZE - 1)
> >
> >Btw, how is this logic supposed to work for discontiguous buffers?
> >
> >Each of them might straddle boundaries individually, so doing the
> >start/end calculation for the number of pages isn't going to be correct.
> >
> 
> eeek. yep, I will loop through and count the pages needed in each
> segment.

That's wrong - the pages must be mappable as a contiguous memory
range. That's how this code avoids copying the data from
discontiguous page ranges into a contiguous mapped memory range.
If you separate each IO into to it's own set of pages, then you have
to completely rewrite _xfs_buf_ioapply(), which AFAICT works just
fine with the page allocation that is done right now....

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

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

Re: [PATCH v2 1/3] xfs: use b_maps[] for discontiguous buffers

[Christoph Hellwig]

On Sat, Dec 01, 2012 at 09:21:10AM +1100, Dave Chinner wrote:
> > eeek. yep, I will loop through and count the pages needed in each
> > segment.
> 
> That's wrong - the pages must be mappable as a contiguous memory
> range. That's how this code avoids copying the data from
> discontiguous page ranges into a contiguous mapped memory range.
> If you separate each IO into to it's own set of pages, then you have
> to completely rewrite _xfs_buf_ioapply(), which AFAICT works just
> fine with the page allocation that is done right now....

Yeah, you're right.

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

Re: [PATCH v2 3/3] xfs: fix the multi-segment log buffer format

[Christoph Hellwig]

Looks good,

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

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

Re: [PATCH v2 1/3] xfs: use b_maps[] for discontiguous buffers

[Dave Chinner]

On Wed, Nov 28, 2012 at 04:23:10PM -0600, Mark Tinguely wrote:
> Commits starting at 77c1a08 introduced a multiple segment support
> to xfs_buf. xfs_trans_buf_item_match() could not find a multi-segment
> buffer in the transaction because it was looking at the single segment
> block number rather than the multi-segment b_maps[0].bm.bn. This
> results on a recursive buffer lock that can never be satisfied.
> 
> This patch:
>  1) Changed the remaining b_map accesses to be b_maps[0] accesses.
>  2) Renames the single segment b_map structure to __b_map to avoid
>     future confusion.
> 
> Signed-off-by: Mark Tinguely <tinguely@sgi.com>

Looks good.

Reviewed-by: Dave Chinner <dchinner@redhat.com>
-- 
Dave Chinner
david@fromorbit.com

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

Re: [PATCH v2 2/3] xfs: fix the buffer log format for contiguous buffers

[Dave Chinner]

On Wed, Nov 28, 2012 at 04:23:11PM -0600, Mark Tinguely wrote:
> Not every segment in a multi-segment buffer is dirty in a
> transaction and they will not be outputted. The assert in
> xfs_buf_item_format_segment() that checks for the at least
> one chunk of data in the segment to be used is not necessary
> true for multi-segmented buffers.
> 
> This patch:

Has three separate, unrelated changes. That usually means 3
separate patches....

>  1) Change xfs_buf_item_format_segment() to skip over non-dirty
>     segments.

Comments about this below.

>  2) Change the bli_format structure to __bli_format so it is not
>     accidently confused with the bli_formats pointer.

This change looks ok.

>  3) Remove the buffer XFS_TRANS_DEBUG routines xfs_buf_item_log_debug()
>     and xfs_buf_item_log_check(). They are not used and are not
>     multi-segment aware.

Removing XFS_TRANS_DEBUG should definitely be done as a separate
patch and it should remove it from everywhere (there's several stale
uses) rather than just this one place. If we don't use it here, we
don't use it anywhere....

....

> @@ -287,11 +184,6 @@ xfs_buf_item_format_segment(
>  	 */
>  	base_size = offsetof(struct xfs_buf_log_format, blf_data_map) +
>  			(blfp->blf_map_size * sizeof(blfp->blf_data_map[0]));
> -	vecp->i_addr = blfp;
> -	vecp->i_len = base_size;
> -	vecp->i_type = XLOG_REG_TYPE_BFORMAT;
> -	vecp++;
> -	nvecs = 1;
>  
>  	if (bip->bli_flags & XFS_BLI_STALE) {
>  		/*
> @@ -301,7 +193,11 @@ xfs_buf_item_format_segment(
>  		 */
>  		trace_xfs_buf_item_format_stale(bip);
>  		ASSERT(blfp->blf_flags & XFS_BLF_CANCEL);
> -		blfp->blf_size = nvecs;
> +		vecp->i_addr = blfp;
> +		vecp->i_len = base_size;
> +		vecp->i_type = XLOG_REG_TYPE_BFORMAT;
> +		vecp++;
> +		blfp->blf_size = 1;
>  		return vecp;
>  	}
>  
> @@ -309,7 +205,19 @@ xfs_buf_item_format_segment(
>  	 * Fill in an iovec for each set of contiguous chunks.
>  	 */
>  	first_bit = xfs_next_bit(blfp->blf_data_map, blfp->blf_map_size, 0);
> -	ASSERT(first_bit != -1);
> +	if (first_bit == -1) {
> +		/* If the map is not be dirty in the transaction, mark
> +		 * the size as zero and do not advance the vector pointer.
> +		 */

Comment format.

> +		blfp->blf_size = 0;
> +		return(vecp);

		return vecp;
> +	}
> +
> +	vecp->i_addr = blfp;
> +	vecp->i_len = base_size;
> +	vecp->i_type = XLOG_REG_TYPE_BFORMAT;
> +	vecp++;
> +	nvecs = 1;

This results in duplicating the code. Perhaps it woul dbe better to
rearrange the code slightly to avoid needing to do that. i.e:

{
	nvecs = 0;
	first_bit = xfs_next_bit(blfp->blf_data_map, blfp->blf_map_size, 0);

	if (!(bip->bli_flags & XFS_BLI_STALE) && first_bit == -1) {
		/*
		 * If the map is not be dirty in the transaction, mark
		 * the size as zero and do not advance the vector pointer.
		 */
		goto out;
	}

	vecp->i_addr = blfp;
	vecp->i_len = base_size;
	vecp->i_type = XLOG_REG_TYPE_BFORMAT;
	vecp++;
	nvecs = 1;

	if (bip->bli_flags & XFS_BLI_STALE) {
		.....
		goto out;
	}
	.....

out:
	blfp->blf_size = nvecs;
	return vecp;
}

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

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

Re: [PATCH v2 3/3] xfs: fix the multi-segment log buffer format

[Dave Chinner]

On Wed, Nov 28, 2012 at 04:23:12PM -0600, Mark Tinguely wrote:
> Per Dave Chinner suggestion, this patch:
>  1) Corrects the detection of whether a multi-segment buffer is
>     still tracking data.
>  2) Clears all the buffer log formats for a multi-segment buffer.
> 
> Signed-off-by: Mark Tinguely <tinguely@sgi.com>
> 
> ---
>  fs/xfs/xfs_buf_item.c  |   13 ++++++++++---
>  fs/xfs/xfs_trans_buf.c |    7 +++++--
>  2 files changed, 15 insertions(+), 5 deletions(-)
> 
> Index: b/fs/xfs/xfs_buf_item.c
> ===================================================================
> --- a/fs/xfs/xfs_buf_item.c
> +++ b/fs/xfs/xfs_buf_item.c
> @@ -508,7 +508,7 @@ xfs_buf_item_unlock(
>  {
>  	struct xfs_buf_log_item	*bip = BUF_ITEM(lip);
>  	struct xfs_buf		*bp = bip->bli_buf;
> -	int			aborted;
> +	int			aborted, empty, i;
>  	uint			hold;
>  
>  	/* Clear the buffer's association with this transaction. */
> @@ -551,8 +551,15 @@ xfs_buf_item_unlock(
>  	 * If the buf item isn't tracking any data, free it, otherwise drop the
>  	 * reference we hold to it.
>  	 */
> -	if (xfs_bitmap_empty(bip->__bli_format.blf_data_map,
> -			     bip->__bli_format.blf_map_size))
> +	empty = 1;
> +	for (i = 0; i < bip->bli_format_count; i++) {
> +		if (!xfs_bitmap_empty(bip->bli_formats[i].blf_data_map,
> +			     bip->bli_formats[i].blf_map_size)) {
> +			empty = 0;
> +			break;
> +		}
> +	}
> +	if (empty)
>  		xfs_buf_item_relse(bp);
>  	else
>  		atomic_dec(&bip->bli_refcount);

I think "clean" is a better name for the temporary variable, as what
we are actually checking is whether the buffer has been modified or
not (i.e. whether it is clean or dirty)....

> Index: b/fs/xfs/xfs_trans_buf.c
> ===================================================================
> --- a/fs/xfs/xfs_trans_buf.c
> +++ b/fs/xfs/xfs_trans_buf.c
> @@ -643,6 +643,7 @@ xfs_trans_binval(
>  	xfs_buf_t	*bp)
>  {
>  	xfs_buf_log_item_t	*bip = bp->b_fspriv;
> +	int			i;
>  
>  	ASSERT(bp->b_transp == tp);
>  	ASSERT(bip != NULL);
> @@ -670,8 +671,10 @@ xfs_trans_binval(
>  	bip->bli_flags &= ~(XFS_BLI_INODE_BUF | XFS_BLI_LOGGED | XFS_BLI_DIRTY);
>  	bip->__bli_format.blf_flags &= ~XFS_BLF_INODE_BUF;
>  	bip->__bli_format.blf_flags |= XFS_BLF_CANCEL;
> -	memset((char *)(bip->__bli_format.blf_data_map), 0,
> -	      (bip->__bli_format.blf_map_size * sizeof(uint)));
> +	for (i = 0; i < bip->bli_format_count; i++) {
> +		memset((char *)(bip->bli_formats[i].blf_data_map), 0,
> +		       (bip->bli_formats[i].blf_map_size * sizeof(uint)));

No cast needed - memset() takes a void pointer....

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

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