Merge branch 'guilt/xlog-intent-whiteouts' into xfs-5.19-for-next

fs/xfs/libxfs/xfs_defer.c

@@ -186,7 +186,7 @@ static const struct xfs_defer_op_type *defer_op_types[] = {
 	[XFS_DEFER_OPS_TYPE_AGFL_FREE]	= &xfs_agfl_free_defer_type,
 };
 
-static void
+static bool
 xfs_defer_create_intent(
 	struct xfs_trans		*tp,
 	struct xfs_defer_pending	*dfp,
@@ -197,6 +197,7 @@ xfs_defer_create_intent(
 	if (!dfp->dfp_intent)
 		dfp->dfp_intent = ops->create_intent(tp, &dfp->dfp_work,
 						     dfp->dfp_count, sort);
+	return dfp->dfp_intent != NULL;
 }
 
 /*
@@ -204,16 +205,18 @@ xfs_defer_create_intent(
  * associated extents, then add the entire intake list to the end of
  * the pending list.
  */
-STATIC void
+static bool
 xfs_defer_create_intents(
 	struct xfs_trans		*tp)
 {
 	struct xfs_defer_pending	*dfp;
+	bool				ret = false;
 
 	list_for_each_entry(dfp, &tp->t_dfops, dfp_list) {
 		trace_xfs_defer_create_intent(tp->t_mountp, dfp);
-		xfs_defer_create_intent(tp, dfp, true);
+		ret |= xfs_defer_create_intent(tp, dfp, true);
 	}
+	return ret;
 }
 
 /* Abort all the intents that were committed. */
@@ -487,7 +490,7 @@ int
 xfs_defer_finish_noroll(
 	struct xfs_trans		**tp)
 {
-	struct xfs_defer_pending	*dfp;
+	struct xfs_defer_pending	*dfp = NULL;
 	int				error = 0;
 	LIST_HEAD(dop_pending);
 
@@ -506,17 +509,20 @@ xfs_defer_finish_noroll(
 		 * of time that any one intent item can stick around in memory,
 		 * pinning the log tail.
 		 */
-		xfs_defer_create_intents(*tp);
+		bool has_intents = xfs_defer_create_intents(*tp);
+
 		list_splice_init(&(*tp)->t_dfops, &dop_pending);
 
-		error = xfs_defer_trans_roll(tp);
-		if (error)
-			goto out_shutdown;
+		if (has_intents || dfp) {
+			error = xfs_defer_trans_roll(tp);
+			if (error)
+				goto out_shutdown;
 
-		/* Possibly relog intent items to keep the log moving. */
-		error = xfs_defer_relog(tp, &dop_pending);
-		if (error)
-			goto out_shutdown;
+			/* Relog intent items to keep the log moving. */
+			error = xfs_defer_relog(tp, &dop_pending);
+			if (error)
+				goto out_shutdown;
+		}
 
 		dfp = list_first_entry(&dop_pending, struct xfs_defer_pending,
 				       dfp_list);

fs/xfs/libxfs/xfs_inode_fork.c

@@ -36,7 +36,7 @@ xfs_init_local_fork(
 	int64_t			size)
 {
 	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, whichfork);
-	int			mem_size = size, real_size = 0;
+	int			mem_size = size;
 	bool			zero_terminate;
 
 	/*
@@ -50,8 +50,7 @@ xfs_init_local_fork(
 		mem_size++;
 
 	if (size) {
-		real_size = roundup(mem_size, 4);
-		ifp->if_u1.if_data = kmem_alloc(real_size, KM_NOFS);
+		ifp->if_u1.if_data = kmem_alloc(mem_size, KM_NOFS);
 		memcpy(ifp->if_u1.if_data, data, size);
 		if (zero_terminate)
 			ifp->if_u1.if_data[size] = '\0';
@@ -497,12 +496,7 @@ xfs_idata_realloc(
 		return;
 	}
 
-	/*
-	 * For inline data, the underlying buffer must be a multiple of 4 bytes
-	 * in size so that it can be logged and stay on word boundaries.
-	 * We enforce that here.
-	 */
-	ifp->if_u1.if_data = krealloc(ifp->if_u1.if_data, roundup(new_size, 4),
+	ifp->if_u1.if_data = krealloc(ifp->if_u1.if_data, new_size,
 				      GFP_NOFS | __GFP_NOFAIL);
 	ifp->if_bytes = new_size;
 }

fs/xfs/libxfs/xfs_shared.h

@@ -54,13 +54,23 @@ void	xfs_log_get_max_trans_res(struct xfs_mount *mp,
 /*
  * Values for t_flags.
  */
-#define	XFS_TRANS_DIRTY		0x01	/* something needs to be logged */
-#define	XFS_TRANS_SB_DIRTY	0x02	/* superblock is modified */
-#define	XFS_TRANS_PERM_LOG_RES	0x04	/* xact took a permanent log res */
-#define	XFS_TRANS_SYNC		0x08	/* make commit synchronous */
-#define XFS_TRANS_RESERVE	0x20    /* OK to use reserved data blocks */
-#define XFS_TRANS_NO_WRITECOUNT 0x40	/* do not elevate SB writecount */
-#define XFS_TRANS_RES_FDBLKS	0x80	/* reserve newly freed blocks */
+/* Transaction needs to be logged */
+#define XFS_TRANS_DIRTY			(1u << 0)
+/* Superblock is dirty and needs to be logged */
+#define XFS_TRANS_SB_DIRTY		(1u << 1)
+/* Transaction took a permanent log reservation */
+#define XFS_TRANS_PERM_LOG_RES		(1u << 2)
+/* Synchronous transaction commit needed */
+#define XFS_TRANS_SYNC			(1u << 3)
+/* Transaction can use reserve block pool */
+#define XFS_TRANS_RESERVE		(1u << 4)
+/* Transaction should avoid VFS level superblock write accounting */
+#define XFS_TRANS_NO_WRITECOUNT		(1u << 5)
+/* Transaction has freed blocks returned to it's reservation */
+#define XFS_TRANS_RES_FDBLKS		(1u << 6)
+/* Transaction contains an intent done log item */
+#define XFS_TRANS_HAS_INTENT_DONE	(1u << 7)
+
 /*
  * LOWMODE is used by the allocator to activate the lowspace algorithm - when
  * free space is running low the extent allocator may choose to allocate an

fs/xfs/xfs_bmap_item.c

@@ -39,6 +39,7 @@ STATIC void
 xfs_bui_item_free(
 	struct xfs_bui_log_item	*buip)
 {
+	kmem_free(buip->bui_item.li_lv_shadow);
 	kmem_cache_free(xfs_bui_cache, buip);
 }
 
@@ -54,10 +55,11 @@ xfs_bui_release(
 	struct xfs_bui_log_item	*buip)
 {
 	ASSERT(atomic_read(&buip->bui_refcount) > 0);
-	if (atomic_dec_and_test(&buip->bui_refcount)) {
-		xfs_trans_ail_delete(&buip->bui_item, SHUTDOWN_LOG_IO_ERROR);
-		xfs_bui_item_free(buip);
-	}
+	if (!atomic_dec_and_test(&buip->bui_refcount))
+		return;
+
+	xfs_trans_ail_delete(&buip->bui_item, 0);
+	xfs_bui_item_free(buip);
 }
 
 
@@ -198,14 +200,24 @@ xfs_bud_item_release(
 	struct xfs_bud_log_item	*budp = BUD_ITEM(lip);
 
 	xfs_bui_release(budp->bud_buip);
+	kmem_free(budp->bud_item.li_lv_shadow);
 	kmem_cache_free(xfs_bud_cache, budp);
 }
 
+static struct xfs_log_item *
+xfs_bud_item_intent(
+	struct xfs_log_item	*lip)
+{
+	return &BUD_ITEM(lip)->bud_buip->bui_item;
+}
+
 static const struct xfs_item_ops xfs_bud_item_ops = {
-	.flags		= XFS_ITEM_RELEASE_WHEN_COMMITTED,
+	.flags		= XFS_ITEM_RELEASE_WHEN_COMMITTED |
+			  XFS_ITEM_INTENT_DONE,
 	.iop_size	= xfs_bud_item_size,
 	.iop_format	= xfs_bud_item_format,
 	.iop_release	= xfs_bud_item_release,
+	.iop_intent	= xfs_bud_item_intent,
 };
 
 static struct xfs_bud_log_item *
@@ -254,7 +266,7 @@ xfs_trans_log_finish_bmap_update(
 	 * 1.) releases the BUI and frees the BUD
 	 * 2.) shuts down the filesystem
 	 */
-	tp->t_flags |= XFS_TRANS_DIRTY;
+	tp->t_flags |= XFS_TRANS_DIRTY | XFS_TRANS_HAS_INTENT_DONE;
 	set_bit(XFS_LI_DIRTY, &budp->bud_item.li_flags);
 
 	return error;
@@ -586,6 +598,7 @@ xfs_bui_item_relog(
 }
 
 static const struct xfs_item_ops xfs_bui_item_ops = {
+	.flags		= XFS_ITEM_INTENT,
 	.iop_size	= xfs_bui_item_size,
 	.iop_format	= xfs_bui_item_format,
 	.iop_unpin	= xfs_bui_item_unpin,

fs/xfs/xfs_extfree_item.c

@@ -58,10 +58,11 @@ xfs_efi_release(
 	struct xfs_efi_log_item	*efip)
 {
 	ASSERT(atomic_read(&efip->efi_refcount) > 0);
-	if (atomic_dec_and_test(&efip->efi_refcount)) {
-		xfs_trans_ail_delete(&efip->efi_item, SHUTDOWN_LOG_IO_ERROR);
-		xfs_efi_item_free(efip);
-	}
+	if (!atomic_dec_and_test(&efip->efi_refcount))
+		return;
+
+	xfs_trans_ail_delete(&efip->efi_item, 0);
+	xfs_efi_item_free(efip);
 }
 
 /*
@@ -306,11 +307,20 @@ xfs_efd_item_release(
 	xfs_efd_item_free(efdp);
 }
 
+static struct xfs_log_item *
+xfs_efd_item_intent(
+	struct xfs_log_item	*lip)
+{
+	return &EFD_ITEM(lip)->efd_efip->efi_item;
+}
+
 static const struct xfs_item_ops xfs_efd_item_ops = {
-	.flags		= XFS_ITEM_RELEASE_WHEN_COMMITTED,
+	.flags		= XFS_ITEM_RELEASE_WHEN_COMMITTED |
+			  XFS_ITEM_INTENT_DONE,
 	.iop_size	= xfs_efd_item_size,
 	.iop_format	= xfs_efd_item_format,
 	.iop_release	= xfs_efd_item_release,
+	.iop_intent	= xfs_efd_item_intent,
 };
 
 /*
@@ -380,7 +390,7 @@ xfs_trans_free_extent(
 	 * 1.) releases the EFI and frees the EFD
 	 * 2.) shuts down the filesystem
 	 */
-	tp->t_flags |= XFS_TRANS_DIRTY;
+	tp->t_flags |= XFS_TRANS_DIRTY | XFS_TRANS_HAS_INTENT_DONE;
 	set_bit(XFS_LI_DIRTY, &efdp->efd_item.li_flags);
 
 	next_extent = efdp->efd_next_extent;
@@ -688,6 +698,7 @@ xfs_efi_item_relog(
 }
 
 static const struct xfs_item_ops xfs_efi_item_ops = {
+	.flags		= XFS_ITEM_INTENT,
 	.iop_size	= xfs_efi_item_size,
 	.iop_format	= xfs_efi_item_format,
 	.iop_unpin	= xfs_efi_item_unpin,

fs/xfs/xfs_icreate_item.c

@@ -63,6 +63,7 @@ STATIC void
 xfs_icreate_item_release(
 	struct xfs_log_item	*lip)
 {
+	kmem_free(ICR_ITEM(lip)->ic_item.li_lv_shadow);
 	kmem_cache_free(xfs_icreate_cache, ICR_ITEM(lip));
 }
 

fs/xfs/xfs_inode_item.c

@@ -71,7 +71,7 @@ xfs_inode_item_data_fork_size(
 	case XFS_DINODE_FMT_LOCAL:
 		if ((iip->ili_fields & XFS_ILOG_DDATA) &&
 		    ip->i_df.if_bytes > 0) {
-			*nbytes += roundup(ip->i_df.if_bytes, 4);
+			*nbytes += xlog_calc_iovec_len(ip->i_df.if_bytes);
 			*nvecs += 1;
 		}
 		break;
@@ -112,7 +112,7 @@ xfs_inode_item_attr_fork_size(
 	case XFS_DINODE_FMT_LOCAL:
 		if ((iip->ili_fields & XFS_ILOG_ADATA) &&
 		    ip->i_afp->if_bytes > 0) {
-			*nbytes += roundup(ip->i_afp->if_bytes, 4);
+			*nbytes += xlog_calc_iovec_len(ip->i_afp->if_bytes);
 			*nvecs += 1;
 		}
 		break;
@@ -204,17 +204,12 @@ xfs_inode_item_format_data_fork(
 			~(XFS_ILOG_DEXT | XFS_ILOG_DBROOT | XFS_ILOG_DEV);
 		if ((iip->ili_fields & XFS_ILOG_DDATA) &&
 		    ip->i_df.if_bytes > 0) {
-			/*
-			 * Round i_bytes up to a word boundary.
-			 * The underlying memory is guaranteed
-			 * to be there by xfs_idata_realloc().
-			 */
-			data_bytes = roundup(ip->i_df.if_bytes, 4);
 			ASSERT(ip->i_df.if_u1.if_data != NULL);
 			ASSERT(ip->i_disk_size > 0);
 			xlog_copy_iovec(lv, vecp, XLOG_REG_TYPE_ILOCAL,
-					ip->i_df.if_u1.if_data, data_bytes);
-			ilf->ilf_dsize = (unsigned)data_bytes;
+					ip->i_df.if_u1.if_data,
+					ip->i_df.if_bytes);
+			ilf->ilf_dsize = (unsigned)ip->i_df.if_bytes;
 			ilf->ilf_size++;
 		} else {
 			iip->ili_fields &= ~XFS_ILOG_DDATA;
@@ -288,17 +283,11 @@ xfs_inode_item_format_attr_fork(
 
 		if ((iip->ili_fields & XFS_ILOG_ADATA) &&
 		    ip->i_afp->if_bytes > 0) {
-			/*
-			 * Round i_bytes up to a word boundary.
-			 * The underlying memory is guaranteed
-			 * to be there by xfs_idata_realloc().
-			 */
-			data_bytes = roundup(ip->i_afp->if_bytes, 4);
 			ASSERT(ip->i_afp->if_u1.if_data != NULL);
 			xlog_copy_iovec(lv, vecp, XLOG_REG_TYPE_IATTR_LOCAL,
 					ip->i_afp->if_u1.if_data,
-					data_bytes);
-			ilf->ilf_asize = (unsigned)data_bytes;
+					ip->i_afp->if_bytes);
+			ilf->ilf_asize = (unsigned)ip->i_afp->if_bytes;
 			ilf->ilf_size++;
 		} else {
 			iip->ili_fields &= ~XFS_ILOG_ADATA;

fs/xfs/xfs_inode_item_recover.c

@@ -462,7 +462,7 @@ xlog_recover_inode_commit_pass2(
 	ASSERT(in_f->ilf_size <= 4);
 	ASSERT((in_f->ilf_size == 3) || (fields & XFS_ILOG_AFORK));
 	ASSERT(!(fields & XFS_ILOG_DFORK) ||
-	       (len == in_f->ilf_dsize));
+	       (len == xlog_calc_iovec_len(in_f->ilf_dsize)));
 
 	switch (fields & XFS_ILOG_DFORK) {
 	case XFS_ILOG_DDATA:
@@ -497,7 +497,7 @@ xlog_recover_inode_commit_pass2(
 		}
 		len = item->ri_buf[attr_index].i_len;
 		src = item->ri_buf[attr_index].i_addr;
-		ASSERT(len == in_f->ilf_asize);
+		ASSERT(len == xlog_calc_iovec_len(in_f->ilf_asize));
 
 		switch (in_f->ilf_fields & XFS_ILOG_AFORK) {
 		case XFS_ILOG_ADATA:

fs/xfs/xfs_log.h

@@ -21,23 +21,59 @@ struct xfs_log_vec {
 
 #define XFS_LOG_VEC_ORDERED	(-1)
 
+/*
+ * Calculate the log iovec length for a given user buffer length. Intended to be
+ * used by ->iop_size implementations when sizing buffers of arbitrary
+ * alignments.
+ */
+static inline int
+xlog_calc_iovec_len(int len)
+{
+	return roundup(len, sizeof(uint32_t));
+}
+
 void *xlog_prepare_iovec(struct xfs_log_vec *lv, struct xfs_log_iovec **vecp,
 		uint type);
 
 static inline void
-xlog_finish_iovec(struct xfs_log_vec *lv, struct xfs_log_iovec *vec, int len)
+xlog_finish_iovec(struct xfs_log_vec *lv, struct xfs_log_iovec *vec,
+		int data_len)
 {
 	struct xlog_op_header	*oph = vec->i_addr;
-
-	/* opheader tracks payload length, logvec tracks region length */
+	int			len;
+
+	/*
+	 * Always round up the length to the correct alignment so callers don't
+	 * need to know anything about this log vec layout requirement. This
+	 * means we have to zero the area the data to be written does not cover.
+	 * This is complicated by fact the payload region is offset into the
+	 * logvec region by the opheader that tracks the payload.
+	 */
+	len = xlog_calc_iovec_len(data_len);
+	if (len - data_len != 0) {
+		char	*buf = vec->i_addr + sizeof(struct xlog_op_header);
+
+		memset(buf + data_len, 0, len - data_len);
+	}
+
+	/*
+	 * The opheader tracks aligned payload length, whilst the logvec tracks
+	 * the overall region length.
+	 */
 	oph->oh_len = cpu_to_be32(len);
 
 	len += sizeof(struct xlog_op_header);
 	lv->lv_buf_len += len;
 	lv->lv_bytes += len;
 	vec->i_len = len;
+
+	/* Catch buffer overruns */
+	ASSERT((void *)lv->lv_buf + lv->lv_bytes <= (void *)lv + lv->lv_size);
 }
 
+/*
+ * Copy the amount of data requested by the caller into a new log iovec.
+ */
 static inline void *
 xlog_copy_iovec(struct xfs_log_vec *lv, struct xfs_log_iovec **vecp,
 		uint type, void *data, int len)

fs/xfs/xfs_log_cil.c

@@ -47,6 +47,38 @@ xlog_cil_ticket_alloc(
 	return tic;
 }
 
+/*
+ * Check if the current log item was first committed in this sequence.
+ * We can't rely on just the log item being in the CIL, we have to check
+ * the recorded commit sequence number.
+ *
+ * Note: for this to be used in a non-racy manner, it has to be called with
+ * CIL flushing locked out. As a result, it should only be used during the
+ * transaction commit process when deciding what to format into the item.
+ */
+static bool
+xlog_item_in_current_chkpt(
+	struct xfs_cil		*cil,
+	struct xfs_log_item	*lip)
+{
+	if (list_empty(&lip->li_cil))
+		return false;
+
+	/*
+	 * li_seq is written on the first commit of a log item to record the
+	 * first checkpoint it is written to. Hence if it is different to the
+	 * current sequence, we're in a new checkpoint.
+	 */
+	return lip->li_seq == READ_ONCE(cil->xc_current_sequence);
+}
+
+bool
+xfs_log_item_in_current_chkpt(
+	struct xfs_log_item *lip)
+{
+	return xlog_item_in_current_chkpt(lip->li_log->l_cilp, lip);
+}
+
 /*
  * Unavoidable forward declaration - xlog_cil_push_work() calls
  * xlog_cil_ctx_alloc() itself.
@@ -444,7 +476,8 @@ xlog_cil_insert_format_items(
 static void
 xlog_cil_insert_items(
 	struct xlog		*log,
-	struct xfs_trans	*tp)
+	struct xfs_trans	*tp,
+	uint32_t		released_space)
 {
 	struct xfs_cil		*cil = log->l_cilp;
 	struct xfs_cil_ctx	*ctx = cil->xc_ctx;
@@ -493,7 +526,9 @@ xlog_cil_insert_items(
 		ASSERT(tp->t_ticket->t_curr_res >= len);
 	}
 	tp->t_ticket->t_curr_res -= len;
+	tp->t_ticket->t_curr_res += released_space;
 	ctx->space_used += len;
+	ctx->space_used -= released_space;
 
 	/*
 	 * If we've overrun the reservation, dump the tx details before we move
@@ -934,6 +969,65 @@ xlog_cil_build_trans_hdr(
 	tic->t_curr_res -= lvhdr->lv_bytes;
 }
 
+/*
+ * Pull all the log vectors off the items in the CIL, and remove the items from
+ * the CIL. We don't need the CIL lock here because it's only needed on the
+ * transaction commit side which is currently locked out by the flush lock.
+ *
+ * If a log item is marked with a whiteout, we do not need to write it to the
+ * journal and so we just move them to the whiteout list for the caller to
+ * dispose of appropriately.
+ */
+static void
+xlog_cil_build_lv_chain(
+	struct xfs_cil		*cil,
+	struct xfs_cil_ctx	*ctx,
+	struct list_head	*whiteouts,
+	uint32_t		*num_iovecs,
+	uint32_t		*num_bytes)
+{
+	struct xfs_log_vec	*lv = NULL;
+
+	while (!list_empty(&cil->xc_cil)) {
+		struct xfs_log_item	*item;
+
+		item = list_first_entry(&cil->xc_cil,
+					struct xfs_log_item, li_cil);
+
+		if (test_bit(XFS_LI_WHITEOUT, &item->li_flags)) {
+			list_move(&item->li_cil, whiteouts);
+			trace_xfs_cil_whiteout_skip(item);
+			continue;
+		}
+
+		list_del_init(&item->li_cil);
+		if (!ctx->lv_chain)
+			ctx->lv_chain = item->li_lv;
+		else
+			lv->lv_next = item->li_lv;
+		lv = item->li_lv;
+		item->li_lv = NULL;
+		*num_iovecs += lv->lv_niovecs;
+
+		/* we don't write ordered log vectors */
+		if (lv->lv_buf_len != XFS_LOG_VEC_ORDERED)
+			*num_bytes += lv->lv_bytes;
+	}
+}
+
+static void
+xlog_cil_cleanup_whiteouts(
+	struct list_head	*whiteouts)
+{
+	while (!list_empty(whiteouts)) {
+		struct xfs_log_item *item = list_first_entry(whiteouts,
+						struct xfs_log_item, li_cil);
+		list_del_init(&item->li_cil);
+		trace_xfs_cil_whiteout_unpin(item);
+		item->li_ops->iop_unpin(item, 1);
+	}
+}
+
 /*
  * Push the Committed Item List to the log.
  *
@@ -956,7 +1050,6 @@ xlog_cil_push_work(
 		container_of(work, struct xfs_cil_ctx, push_work);
 	struct xfs_cil		*cil = ctx->cil;
 	struct xlog		*log = cil->xc_log;
-	struct xfs_log_vec	*lv;
 	struct xfs_cil_ctx	*new_ctx;
 	int			num_iovecs = 0;
 	int			num_bytes = 0;
@@ -965,6 +1058,7 @@ xlog_cil_push_work(
 	struct xfs_log_vec	lvhdr = { NULL };
 	xfs_csn_t		push_seq;
 	bool			push_commit_stable;
+	LIST_HEAD		(whiteouts);
 
 	new_ctx = xlog_cil_ctx_alloc();
 	new_ctx->ticket = xlog_cil_ticket_alloc(log);
@@ -1033,31 +1127,7 @@ xlog_cil_push_work(
 	list_add(&ctx->committing, &cil->xc_committing);
 	spin_unlock(&cil->xc_push_lock);
 
-	/*
-	 * Pull all the log vectors off the items in the CIL, and remove the
-	 * items from the CIL. We don't need the CIL lock here because it's only
-	 * needed on the transaction commit side which is currently locked out
-	 * by the flush lock.
-	 */
-	lv = NULL;
-	while (!list_empty(&cil->xc_cil)) {
-		struct xfs_log_item	*item;
-
-		item = list_first_entry(&cil->xc_cil,
-					struct xfs_log_item, li_cil);
-		list_del_init(&item->li_cil);
-		if (!ctx->lv_chain)
-			ctx->lv_chain = item->li_lv;
-		else
-			lv->lv_next = item->li_lv;
-		lv = item->li_lv;
-		item->li_lv = NULL;
-		num_iovecs += lv->lv_niovecs;
-
-		/* we don't write ordered log vectors */
-		if (lv->lv_buf_len != XFS_LOG_VEC_ORDERED)
-			num_bytes += lv->lv_bytes;
-	}
+	xlog_cil_build_lv_chain(cil, ctx, &whiteouts, &num_iovecs, &num_bytes);
 
 	/*
 	 * Switch the contexts so we can drop the context lock and move out
@@ -1160,6 +1230,7 @@ xlog_cil_push_work(
 	/* Not safe to reference ctx now! */
 
 	spin_unlock(&log->l_icloglock);
+	xlog_cil_cleanup_whiteouts(&whiteouts);
 	return;
 
 out_skip:
@@ -1171,6 +1242,7 @@ xlog_cil_push_work(
 out_abort_free_ticket:
 	xfs_log_ticket_ungrant(log, ctx->ticket);
 	ASSERT(xlog_is_shutdown(log));
+	xlog_cil_cleanup_whiteouts(&whiteouts);
 	if (!ctx->commit_iclog) {
 		xlog_cil_committed(ctx);
 		return;
@@ -1319,6 +1391,43 @@ xlog_cil_empty(
 	return empty;
 }
 
+/*
+ * If there are intent done items in this transaction and the related intent was
+ * committed in the current (same) CIL checkpoint, we don't need to write either
+ * the intent or intent done item to the journal as the change will be
+ * journalled atomically within this checkpoint. As we cannot remove items from
+ * the CIL here, mark the related intent with a whiteout so that the CIL push
+ * can remove it rather than writing it to the journal. Then remove the intent
+ * done item from the current transaction and release it so it doesn't get put
+ * into the CIL at all.
+ */
+static uint32_t
+xlog_cil_process_intents(
+	struct xfs_cil		*cil,
+	struct xfs_trans	*tp)
+{
+	struct xfs_log_item	*lip, *ilip, *next;
+	uint32_t		len = 0;
+
+	list_for_each_entry_safe(lip, next, &tp->t_items, li_trans) {
+		if (!(lip->li_ops->flags & XFS_ITEM_INTENT_DONE))
+			continue;
+
+		ilip = lip->li_ops->iop_intent(lip);
+		if (!ilip || !xlog_item_in_current_chkpt(cil, ilip))
+			continue;
+		set_bit(XFS_LI_WHITEOUT, &ilip->li_flags);
+		trace_xfs_cil_whiteout_mark(ilip);
+		len += ilip->li_lv->lv_bytes;
+		kmem_free(ilip->li_lv);
+		ilip->li_lv = NULL;
+
+		xfs_trans_del_item(lip);
+		lip->li_ops->iop_release(lip);
+	}
+	return len;
+}
+
 /*
  * Commit a transaction with the given vector to the Committed Item List.
  *
@@ -1341,6 +1450,7 @@ xlog_cil_commit(
 {
 	struct xfs_cil		*cil = log->l_cilp;
 	struct xfs_log_item	*lip, *next;
+	uint32_t		released_space = 0;
 
 	/*
 	 * Do all necessary memory allocation before we lock the CIL.
@@ -1352,7 +1462,10 @@ xlog_cil_commit(
 	/* lock out background commit */
 	down_read(&cil->xc_ctx_lock);
 
-	xlog_cil_insert_items(log, tp);
+	if (tp->t_flags & XFS_TRANS_HAS_INTENT_DONE)
+		released_space = xlog_cil_process_intents(cil, tp);
+
+	xlog_cil_insert_items(log, tp, released_space);
 
 	if (regrant && !xlog_is_shutdown(log))
 		xfs_log_ticket_regrant(log, tp->t_ticket);
@@ -1508,32 +1621,6 @@ xlog_cil_force_seq(
 	return 0;
 }
 
-/*
- * Check if the current log item was first committed in this sequence.
- * We can't rely on just the log item being in the CIL, we have to check
- * the recorded commit sequence number.
- *
- * Note: for this to be used in a non-racy manner, it has to be called with
- * CIL flushing locked out. As a result, it should only be used during the
- * transaction commit process when deciding what to format into the item.
- */
-bool
-xfs_log_item_in_current_chkpt(
-	struct xfs_log_item	*lip)
-{
-	struct xfs_cil		*cil = lip->li_log->l_cilp;
-
-	if (list_empty(&lip->li_cil))
-		return false;
-
-	/*
-	 * li_seq is written on the first commit of a log item to record the
-	 * first checkpoint it is written to. Hence if it is different to the
-	 * current sequence, we're in a new checkpoint.
-	 */
-	return lip->li_seq == READ_ONCE(cil->xc_current_sequence);
-}
-
 /*
  * Perform initial CIL structure initialisation.
  */

fs/xfs/xfs_refcount_item.c

@@ -35,6 +35,7 @@ STATIC void
 xfs_cui_item_free(
 	struct xfs_cui_log_item	*cuip)
 {
+	kmem_free(cuip->cui_item.li_lv_shadow);
 	if (cuip->cui_format.cui_nextents > XFS_CUI_MAX_FAST_EXTENTS)
 		kmem_free(cuip);
 	else
@@ -53,10 +54,11 @@ xfs_cui_release(
 	struct xfs_cui_log_item	*cuip)
 {
 	ASSERT(atomic_read(&cuip->cui_refcount) > 0);
-	if (atomic_dec_and_test(&cuip->cui_refcount)) {
-		xfs_trans_ail_delete(&cuip->cui_item, SHUTDOWN_LOG_IO_ERROR);
-		xfs_cui_item_free(cuip);
-	}
+	if (!atomic_dec_and_test(&cuip->cui_refcount))
+		return;
+
+	xfs_trans_ail_delete(&cuip->cui_item, 0);
+	xfs_cui_item_free(cuip);
 }
 
 
@@ -204,14 +206,24 @@ xfs_cud_item_release(
 	struct xfs_cud_log_item	*cudp = CUD_ITEM(lip);
 
 	xfs_cui_release(cudp->cud_cuip);
+	kmem_free(cudp->cud_item.li_lv_shadow);
 	kmem_cache_free(xfs_cud_cache, cudp);
 }
 
+static struct xfs_log_item *
+xfs_cud_item_intent(
+	struct xfs_log_item	*lip)
+{
+	return &CUD_ITEM(lip)->cud_cuip->cui_item;
+}
+
 static const struct xfs_item_ops xfs_cud_item_ops = {
-	.flags		= XFS_ITEM_RELEASE_WHEN_COMMITTED,
+	.flags		= XFS_ITEM_RELEASE_WHEN_COMMITTED |
+			  XFS_ITEM_INTENT_DONE,
 	.iop_size	= xfs_cud_item_size,
 	.iop_format	= xfs_cud_item_format,
 	.iop_release	= xfs_cud_item_release,
+	.iop_intent	= xfs_cud_item_intent,
 };
 
 static struct xfs_cud_log_item *
@@ -259,7 +271,7 @@ xfs_trans_log_finish_refcount_update(
 	 * 1.) releases the CUI and frees the CUD
 	 * 2.) shuts down the filesystem
 	 */
-	tp->t_flags |= XFS_TRANS_DIRTY;
+	tp->t_flags |= XFS_TRANS_DIRTY | XFS_TRANS_HAS_INTENT_DONE;
 	set_bit(XFS_LI_DIRTY, &cudp->cud_item.li_flags);
 
 	return error;
@@ -600,6 +612,7 @@ xfs_cui_item_relog(
 }
 
 static const struct xfs_item_ops xfs_cui_item_ops = {
+	.flags		= XFS_ITEM_INTENT,
 	.iop_size	= xfs_cui_item_size,
 	.iop_format	= xfs_cui_item_format,
 	.iop_unpin	= xfs_cui_item_unpin,

fs/xfs/xfs_rmap_item.c

@@ -35,6 +35,7 @@ STATIC void
 xfs_rui_item_free(
 	struct xfs_rui_log_item	*ruip)
 {
+	kmem_free(ruip->rui_item.li_lv_shadow);
 	if (ruip->rui_format.rui_nextents > XFS_RUI_MAX_FAST_EXTENTS)
 		kmem_free(ruip);
 	else
@@ -53,10 +54,11 @@ xfs_rui_release(
 	struct xfs_rui_log_item	*ruip)
 {
 	ASSERT(atomic_read(&ruip->rui_refcount) > 0);
-	if (atomic_dec_and_test(&ruip->rui_refcount)) {
-		xfs_trans_ail_delete(&ruip->rui_item, SHUTDOWN_LOG_IO_ERROR);
-		xfs_rui_item_free(ruip);
-	}
+	if (!atomic_dec_and_test(&ruip->rui_refcount))
+		return;
+
+	xfs_trans_ail_delete(&ruip->rui_item, 0);
+	xfs_rui_item_free(ruip);
 }
 
 STATIC void
@@ -227,14 +229,24 @@ xfs_rud_item_release(
 	struct xfs_rud_log_item	*rudp = RUD_ITEM(lip);
 
 	xfs_rui_release(rudp->rud_ruip);
+	kmem_free(rudp->rud_item.li_lv_shadow);
 	kmem_cache_free(xfs_rud_cache, rudp);
 }
 
+static struct xfs_log_item *
+xfs_rud_item_intent(
+	struct xfs_log_item	*lip)
+{
+	return &RUD_ITEM(lip)->rud_ruip->rui_item;
+}
+
 static const struct xfs_item_ops xfs_rud_item_ops = {
-	.flags		= XFS_ITEM_RELEASE_WHEN_COMMITTED,
+	.flags		= XFS_ITEM_RELEASE_WHEN_COMMITTED |
+			  XFS_ITEM_INTENT_DONE,
 	.iop_size	= xfs_rud_item_size,
 	.iop_format	= xfs_rud_item_format,
 	.iop_release	= xfs_rud_item_release,
+	.iop_intent	= xfs_rud_item_intent,
 };
 
 static struct xfs_rud_log_item *
@@ -327,7 +339,7 @@ xfs_trans_log_finish_rmap_update(
 	 * 1.) releases the RUI and frees the RUD
 	 * 2.) shuts down the filesystem
 	 */
-	tp->t_flags |= XFS_TRANS_DIRTY;
+	tp->t_flags |= XFS_TRANS_DIRTY | XFS_TRANS_HAS_INTENT_DONE;
 	set_bit(XFS_LI_DIRTY, &rudp->rud_item.li_flags);
 
 	return error;
@@ -630,6 +642,7 @@ xfs_rui_item_relog(
 }
 
 static const struct xfs_item_ops xfs_rui_item_ops = {
+	.flags		= XFS_ITEM_INTENT,
 	.iop_size	= xfs_rui_item_size,
 	.iop_format	= xfs_rui_item_format,
 	.iop_unpin	= xfs_rui_item_unpin,

fs/xfs/xfs_trace.h

@@ -1332,6 +1332,9 @@ DEFINE_LOG_ITEM_EVENT(xfs_ail_push);
 DEFINE_LOG_ITEM_EVENT(xfs_ail_pinned);
 DEFINE_LOG_ITEM_EVENT(xfs_ail_locked);
 DEFINE_LOG_ITEM_EVENT(xfs_ail_flushing);
+DEFINE_LOG_ITEM_EVENT(xfs_cil_whiteout_mark);
+DEFINE_LOG_ITEM_EVENT(xfs_cil_whiteout_skip);
+DEFINE_LOG_ITEM_EVENT(xfs_cil_whiteout_unpin);
 
 DECLARE_EVENT_CLASS(xfs_ail_class,
 	TP_PROTO(struct xfs_log_item *lip, xfs_lsn_t old_lsn, xfs_lsn_t new_lsn),

fs/xfs/xfs_trans.h

@@ -55,13 +55,15 @@ struct xfs_log_item {
 #define	XFS_LI_IN_AIL	0
 #define	XFS_LI_ABORTED	1
 #define	XFS_LI_FAILED	2
-#define	XFS_LI_DIRTY	3	/* log item dirty in transaction */
+#define	XFS_LI_DIRTY	3
+#define	XFS_LI_WHITEOUT	4
 
 #define XFS_LI_FLAGS \
 	{ (1u << XFS_LI_IN_AIL),	"IN_AIL" }, \
 	{ (1u << XFS_LI_ABORTED),	"ABORTED" }, \
 	{ (1u << XFS_LI_FAILED),	"FAILED" }, \
-	{ (1u << XFS_LI_DIRTY),		"DIRTY" }
+	{ (1u << XFS_LI_DIRTY),		"DIRTY" }, \
+	{ (1u << XFS_LI_WHITEOUT),	"WHITEOUT" }
 
 struct xfs_item_ops {
 	unsigned flags;
@@ -78,30 +80,32 @@ struct xfs_item_ops {
 	bool (*iop_match)(struct xfs_log_item *item, uint64_t id);
 	struct xfs_log_item *(*iop_relog)(struct xfs_log_item *intent,
 			struct xfs_trans *tp);
+	struct xfs_log_item *(*iop_intent)(struct xfs_log_item *intent_done);
 };
 
-/* Is this log item a deferred action intent? */
+/*
+ * Log item ops flags
+ */
+/*
+ * Release the log item when the journal commits instead of inserting into the
+ * AIL for writeback tracking and/or log tail pinning.
+ */
+#define XFS_ITEM_RELEASE_WHEN_COMMITTED	(1 << 0)
+#define XFS_ITEM_INTENT			(1 << 1)
+#define XFS_ITEM_INTENT_DONE		(1 << 2)
+
 static inline bool
 xlog_item_is_intent(struct xfs_log_item *lip)
 {
-	return lip->li_ops->iop_recover != NULL &&
-	       lip->li_ops->iop_match != NULL;
+	return lip->li_ops->flags & XFS_ITEM_INTENT;
 }
 
-/* Is this a log intent-done item? */
 static inline bool
 xlog_item_is_intent_done(struct xfs_log_item *lip)
 {
-	return lip->li_ops->iop_unpin == NULL &&
-	       lip->li_ops->iop_push == NULL;
+	return lip->li_ops->flags & XFS_ITEM_INTENT_DONE;
 }
 
-/*
- * Release the log item as soon as committed.  This is for items just logging
- * intents that never need to be written back in place.
- */
-#define XFS_ITEM_RELEASE_WHEN_COMMITTED	(1 << 0)
-
 void	xfs_log_item_init(struct xfs_mount *mp, struct xfs_log_item *item,
 			  int type, const struct xfs_item_ops *ops);