[Ocfs2-devel] [PATCH 0/3] Ocfs2: Complete rest of o2info patches(v5).

[Ocfs2-devel] [PATCH 0/3] Ocfs2: Complete rest of o2info patches(v5).

[Tristan Ye]

Joel,
	Please take the following three patches as the final version for the rest
of o2info patches, it's applicable atop latest 2.6.39 kernel, and has been well-
tested.
	Btw, move_exts patches series will be coming soon, which should be applied
right after o2info patches, so apply o2info first, then move_exts into your merge_
window.

Tristan.

[Ocfs2-devel] [PATCH 1/3] Ocfs2: Using inline funcs to set/clear *FILLED* flags in info handler.

[Tristan Ye]

It just removes some macros for the sake of typechecking gains.

Signed-off-by: Tristan Ye <tristan.ye@oracle.com>
---
 fs/ocfs2/ioctl.c |   47 +++++++++++++++++++----------------------------
 1 files changed, 19 insertions(+), 28 deletions(-)

diff --git a/fs/ocfs2/ioctl.c b/fs/ocfs2/ioctl.c
index 8f13c59..373c1d1 100644
--- a/fs/ocfs2/ioctl.c
+++ b/fs/ocfs2/ioctl.c
@@ -35,32 +35,23 @@
  * be -EFAULT.  The error will be returned from the ioctl(2) call.  It's
  * just a best-effort to tell userspace that this request caused the error.
  */
-static inline void __o2info_set_request_error(struct ocfs2_info_request *kreq,
+static inline void o2info_set_request_error(struct ocfs2_info_request *kreq,
 					struct ocfs2_info_request __user *req)
 {
 	kreq->ir_flags |= OCFS2_INFO_FL_ERROR;
 	(void)put_user(kreq->ir_flags, (__u32 __user *)&(req->ir_flags));
 }
 
-#define o2info_set_request_error(a, b) \
-		__o2info_set_request_error((struct ocfs2_info_request *)&(a), b)
-
-static inline void __o2info_set_request_filled(struct ocfs2_info_request *req)
+static inline void o2info_set_request_filled(struct ocfs2_info_request *req)
 {
 	req->ir_flags |= OCFS2_INFO_FL_FILLED;
 }
 
-#define o2info_set_request_filled(a) \
-		__o2info_set_request_filled((struct ocfs2_info_request *)&(a))
-
-static inline void __o2info_clear_request_filled(struct ocfs2_info_request *req)
+static inline void o2info_clear_request_filled(struct ocfs2_info_request *req)
 {
 	req->ir_flags &= ~OCFS2_INFO_FL_FILLED;
 }
 
-#define o2info_clear_request_filled(a) \
-		__o2info_clear_request_filled((struct ocfs2_info_request *)&(a))
-
 static int ocfs2_get_inode_attr(struct inode *inode, unsigned *flags)
 {
 	int status;
@@ -153,7 +144,7 @@ int ocfs2_info_handle_blocksize(struct inode *inode,
 
 	oib.ib_blocksize = inode->i_sb->s_blocksize;
 
-	o2info_set_request_filled(oib);
+	o2info_set_request_filled(&oib.ib_req);
 
 	if (o2info_to_user(oib, req))
 		goto bail;
@@ -161,7 +152,7 @@ int ocfs2_info_handle_blocksize(struct inode *inode,
 	status = 0;
 bail:
 	if (status)
-		o2info_set_request_error(oib, req);
+		o2info_set_request_error(&oib.ib_req, req);
 
 	return status;
 }
@@ -178,7 +169,7 @@ int ocfs2_info_handle_clustersize(struct inode *inode,
 
 	oic.ic_clustersize = osb->s_clustersize;
 
-	o2info_set_request_filled(oic);
+	o2info_set_request_filled(&oic.ic_req);
 
 	if (o2info_to_user(oic, req))
 		goto bail;
@@ -186,7 +177,7 @@ int ocfs2_info_handle_clustersize(struct inode *inode,
 	status = 0;
 bail:
 	if (status)
-		o2info_set_request_error(oic, req);
+		o2info_set_request_error(&oic.ic_req, req);
 
 	return status;
 }
@@ -203,7 +194,7 @@ int ocfs2_info_handle_maxslots(struct inode *inode,
 
 	oim.im_max_slots = osb->max_slots;
 
-	o2info_set_request_filled(oim);
+	o2info_set_request_filled(&oim.im_req);
 
 	if (o2info_to_user(oim, req))
 		goto bail;
@@ -211,7 +202,7 @@ int ocfs2_info_handle_maxslots(struct inode *inode,
 	status = 0;
 bail:
 	if (status)
-		o2info_set_request_error(oim, req);
+		o2info_set_request_error(&oim.im_req, req);
 
 	return status;
 }
@@ -228,7 +219,7 @@ int ocfs2_info_handle_label(struct inode *inode,
 
 	memcpy(oil.il_label, osb->vol_label, OCFS2_MAX_VOL_LABEL_LEN);
 
-	o2info_set_request_filled(oil);
+	o2info_set_request_filled(&oil.il_req);
 
 	if (o2info_to_user(oil, req))
 		goto bail;
@@ -236,7 +227,7 @@ int ocfs2_info_handle_label(struct inode *inode,
 	status = 0;
 bail:
 	if (status)
-		o2info_set_request_error(oil, req);
+		o2info_set_request_error(&oil.il_req, req);
 
 	return status;
 }
@@ -253,7 +244,7 @@ int ocfs2_info_handle_uuid(struct inode *inode,
 
 	memcpy(oiu.iu_uuid_str, osb->uuid_str, OCFS2_TEXT_UUID_LEN + 1);
 
-	o2info_set_request_filled(oiu);
+	o2info_set_request_filled(&oiu.iu_req);
 
 	if (o2info_to_user(oiu, req))
 		goto bail;
@@ -261,7 +252,7 @@ int ocfs2_info_handle_uuid(struct inode *inode,
 	status = 0;
 bail:
 	if (status)
-		o2info_set_request_error(oiu, req);
+		o2info_set_request_error(&oiu.iu_req, req);
 
 	return status;
 }
@@ -280,7 +271,7 @@ int ocfs2_info_handle_fs_features(struct inode *inode,
 	oif.if_incompat_features = osb->s_feature_incompat;
 	oif.if_ro_compat_features = osb->s_feature_ro_compat;
 
-	o2info_set_request_filled(oif);
+	o2info_set_request_filled(&oif.if_req);
 
 	if (o2info_to_user(oif, req))
 		goto bail;
@@ -288,7 +279,7 @@ int ocfs2_info_handle_fs_features(struct inode *inode,
 	status = 0;
 bail:
 	if (status)
-		o2info_set_request_error(oif, req);
+		o2info_set_request_error(&oif.if_req, req);
 
 	return status;
 }
@@ -305,7 +296,7 @@ int ocfs2_info_handle_journal_size(struct inode *inode,
 
 	oij.ij_journal_size = osb->journal->j_inode->i_size;
 
-	o2info_set_request_filled(oij);
+	o2info_set_request_filled(&oij.ij_req);
 
 	if (o2info_to_user(oij, req))
 		goto bail;
@@ -313,7 +304,7 @@ int ocfs2_info_handle_journal_size(struct inode *inode,
 	status = 0;
 bail:
 	if (status)
-		o2info_set_request_error(oij, req);
+		o2info_set_request_error(&oij.ij_req, req);
 
 	return status;
 }
@@ -327,7 +318,7 @@ int ocfs2_info_handle_unknown(struct inode *inode,
 	if (o2info_from_user(oir, req))
 		goto bail;
 
-	o2info_clear_request_filled(oir);
+	o2info_clear_request_filled(&oir);
 
 	if (o2info_to_user(oir, req))
 		goto bail;
@@ -335,7 +326,7 @@ int ocfs2_info_handle_unknown(struct inode *inode,
 	status = 0;
 bail:
 	if (status)
-		o2info_set_request_error(oir, req);
+		o2info_set_request_error(&oir, req);
 
 	return status;
 }
-- 
1.5.5

[Ocfs2-devel] [PATCH 2/3] Ocfs2: Add a new code 'OCFS2_INFO_FREEINODE' for o2info ioctl.

[Tristan Ye]

The new code is dedicated to calculate free inodes number of all inode_allocs,
then return the info to userpace in terms of an array.

Specially, flag 'OCFS2_INFO_FL_NON_COHERENT', manipulated by '--cluster-coherent'
from userspace, is now going to be involved. setting the flag on means no cluster
coherency considered, usually, userspace tools choose none-coherency strategy by
default for the sake of performace.

Signed-off-by: Tristan Ye <tristan.ye@oracle.com>
---
 fs/ocfs2/ioctl.c       |  128 ++++++++++++++++++++++++++++++++++++++++++++++++
 fs/ocfs2/ocfs2_ioctl.h |   11 ++++
 2 files changed, 139 insertions(+), 0 deletions(-)

diff --git a/fs/ocfs2/ioctl.c b/fs/ocfs2/ioctl.c
index 373c1d1..4216739 100644
--- a/fs/ocfs2/ioctl.c
+++ b/fs/ocfs2/ioctl.c
@@ -22,6 +22,9 @@
 #include "ioctl.h"
 #include "resize.h"
 #include "refcounttree.h"
+#include "sysfile.h"
+#include "dir.h"
+#include "buffer_head_io.h"
 
 #include <linux/ext2_fs.h>
 
@@ -52,6 +55,11 @@ static inline void o2info_clear_request_filled(struct ocfs2_info_request *req)
 	req->ir_flags &= ~OCFS2_INFO_FL_FILLED;
 }
 
+static inline int o2info_coherent(struct ocfs2_info_request *req)
+{
+	return (!(req->ir_flags & OCFS2_INFO_FL_NON_COHERENT));
+}
+
 static int ocfs2_get_inode_attr(struct inode *inode, unsigned *flags)
 {
 	int status;
@@ -309,6 +317,122 @@ bail:
 	return status;
 }
 
+int ocfs2_info_scan_inode_alloc(struct ocfs2_super *osb,
+				struct inode *inode_alloc, u64 blkno,
+				struct ocfs2_info_freeinode *fi, u32 slot)
+{
+	int status = 0, unlock = 0;
+
+	struct buffer_head *bh = NULL;
+	struct ocfs2_dinode *dinode_alloc = NULL;
+
+	if (inode_alloc)
+		mutex_lock(&inode_alloc->i_mutex);
+
+	if (o2info_coherent(&fi->ifi_req)) {
+		status = ocfs2_inode_lock(inode_alloc, &bh, 0);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+		unlock = 1;
+	} else {
+		status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+	}
+
+	dinode_alloc = (struct ocfs2_dinode *)bh->b_data;
+
+	fi->ifi_stat[slot].lfi_total =
+		le32_to_cpu(dinode_alloc->id1.bitmap1.i_total);
+	fi->ifi_stat[slot].lfi_free =
+		le32_to_cpu(dinode_alloc->id1.bitmap1.i_total) -
+		le32_to_cpu(dinode_alloc->id1.bitmap1.i_used);
+
+bail:
+	if (unlock)
+		ocfs2_inode_unlock(inode_alloc, 0);
+
+	if (inode_alloc)
+		mutex_unlock(&inode_alloc->i_mutex);
+
+	brelse(bh);
+
+	return status;
+}
+
+int ocfs2_info_handle_freeinode(struct inode *inode,
+				struct ocfs2_info_request __user *req)
+{
+	u32 i;
+	u64 blkno = -1;
+	char namebuf[40];
+	int status = -EFAULT, type = INODE_ALLOC_SYSTEM_INODE;
+	struct ocfs2_info_freeinode *oifi = NULL;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct inode *inode_alloc = NULL;
+
+	oifi = kzalloc(sizeof(struct ocfs2_info_freeinode), GFP_KERNEL);
+	if (!oifi) {
+		status = -ENOMEM;
+		mlog_errno(status);
+		goto bail;
+	}
+
+	if (o2info_from_user(*oifi, req))
+		goto bail;
+
+	oifi->ifi_slotnum = osb->max_slots;
+
+	for (i = 0; i < oifi->ifi_slotnum; i++) {
+		if (o2info_coherent(&oifi->ifi_req)) {
+			inode_alloc = ocfs2_get_system_file_inode(osb, type, i);
+			if (!inode_alloc) {
+				mlog(ML_ERROR, "unable to get alloc inode in "
+				    "slot %u\n", i);
+				status = -EIO;
+				goto bail;
+			}
+		} else {
+			ocfs2_sprintf_system_inode_name(namebuf,
+							sizeof(namebuf),
+							type, i);
+			status = ocfs2_lookup_ino_from_name(osb->sys_root_inode,
+							    namebuf,
+							    strlen(namebuf),
+							    &blkno);
+			if (status < 0) {
+				status = -ENOENT;
+				goto bail;
+			}
+		}
+
+		status = ocfs2_info_scan_inode_alloc(osb, inode_alloc, blkno, oifi, i);
+		if (status < 0)
+			goto bail;
+
+		iput(inode_alloc);
+		inode_alloc = NULL;
+	}
+
+	o2info_set_request_filled(&oifi->ifi_req);
+
+	if (o2info_to_user(*oifi, req))
+		goto bail;
+
+	status = 0;
+bail:
+	if (status)
+		o2info_set_request_error(&oifi->ifi_req, req);
+
+	kfree(oifi);
+
+	return status;
+}
+
 int ocfs2_info_handle_unknown(struct inode *inode,
 			      struct ocfs2_info_request __user *req)
 {
@@ -380,6 +504,10 @@ int ocfs2_info_handle_request(struct inode *inode,
 		if (oir.ir_size == sizeof(struct ocfs2_info_journal_size))
 			status = ocfs2_info_handle_journal_size(inode, req);
 		break;
+	case OCFS2_INFO_FREEINODE:
+		if (oir.ir_size == sizeof(struct ocfs2_info_freeinode))
+			status = ocfs2_info_handle_freeinode(inode, req);
+		break;
 	default:
 		status = ocfs2_info_handle_unknown(inode, req);
 		break;
diff --git a/fs/ocfs2/ocfs2_ioctl.h b/fs/ocfs2/ocfs2_ioctl.h
index b46f39b..6b4b39a 100644
--- a/fs/ocfs2/ocfs2_ioctl.h
+++ b/fs/ocfs2/ocfs2_ioctl.h
@@ -142,6 +142,16 @@ struct ocfs2_info_journal_size {
 	__u64 ij_journal_size;
 };
 
+struct ocfs2_info_freeinode {
+	struct ocfs2_info_request ifi_req;
+	struct ocfs2_info_local_freeinode {
+		__u64 lfi_total;
+		__u64 lfi_free;
+	} ifi_stat[OCFS2_MAX_SLOTS];
+	__u32 ifi_slotnum; /* out */
+	__u32 ifi_pad;
+};
+
 /* Codes for ocfs2_info_request */
 enum ocfs2_info_type {
 	OCFS2_INFO_CLUSTERSIZE = 1,
@@ -151,6 +161,7 @@ enum ocfs2_info_type {
 	OCFS2_INFO_UUID,
 	OCFS2_INFO_FS_FEATURES,
 	OCFS2_INFO_JOURNAL_SIZE,
+	OCFS2_INFO_FREEINODE,
 	OCFS2_INFO_NUM_TYPES
 };
 
-- 
1.5.5

[Ocfs2-devel] [PATCH 3/3] Ocfs2: Add a new code 'OCFS2_INFO_FREEFRAG' for o2info ioctl.

[Tristan Ye]

This new code is a bit more complicated than former ones, the goal is to
show user all statistics required to take a deep insight into filesystem
on how the disk is being fragmentaed.

The goal is achieved by scaning global bitmap from (cluster)group to group
to figure out following factors in the filesystem:

        - How many free chunks in a fixed size as user requested.
        - How many real free chunks in all size.
        - Min/Max/Avg size(in) clusters of free chunks.
        - How do free chunks distribute(in size) in terms of a histogram,
          just like following:
          ---------------------------------------------------------
          Extent Size Range :  Free extents  Free Clusters  Percent
             32K...   64K-  :             1             1    0.00%
              1M...    2M-  :             9           288    0.03%
              8M...   16M-  :             2           831    0.09%
             32M...   64M-  :             1          2047    0.23%
            128M...  256M-  :             1          8191    0.92%
            256M...  512M-  :             2         21706    2.43%
            512M... 1024M-  :            27        858623   96.29%
          ---------------------------------------------------------

Userspace ioctl() call eventually gets the above info returned by passing
a 'struct ocfs2_info_freefrag' with the chunk_size being specified first.

Signed-off-by: Tristan Ye <tristan.ye@oracle.com>
---
 fs/ocfs2/ioctl.c       |  290 ++++++++++++++++++++++++++++++++++++++++++++++++
 fs/ocfs2/ocfs2_ioctl.h |   23 ++++
 2 files changed, 313 insertions(+), 0 deletions(-)

diff --git a/fs/ocfs2/ioctl.c b/fs/ocfs2/ioctl.c
index 4216739..fd248ed 100644
--- a/fs/ocfs2/ioctl.c
+++ b/fs/ocfs2/ioctl.c
@@ -25,6 +25,7 @@
 #include "sysfile.h"
 #include "dir.h"
 #include "buffer_head_io.h"
+#include "suballoc.h"
 
 #include <linux/ext2_fs.h>
 
@@ -433,6 +434,291 @@ bail:
 	return status;
 }
 
+static void o2ffg_update_histogram(struct ocfs2_info_free_chunk_list *hist,
+				   unsigned int chunksize)
+{
+	int index;
+
+	index = __ilog2_u32(chunksize);
+	if (index >= OCFS2_INFO_MAX_HIST)
+		index = OCFS2_INFO_MAX_HIST - 1;
+
+	hist->fc_chunks[index]++;
+	hist->fc_clusters[index] += chunksize;
+}
+
+static void o2ffg_update_stats(struct ocfs2_info_freefrag_stats *stats,
+			       unsigned int chunksize)
+{
+	if (chunksize > stats->ffs_max)
+		stats->ffs_max = chunksize;
+
+	if (chunksize < stats->ffs_min)
+		stats->ffs_min = chunksize;
+
+	stats->ffs_avg += chunksize;
+	stats->ffs_free_chunks_real++;
+}
+
+void ocfs2_info_update_ffg(struct ocfs2_info_freefrag *ffg,
+			   unsigned int chunksize)
+{
+	o2ffg_update_histogram(&(ffg->iff_ffs.ffs_fc_hist), chunksize);
+	o2ffg_update_stats(&(ffg->iff_ffs), chunksize);
+}
+
+int ocfs2_info_freefrag_scan_chain(struct ocfs2_super *osb,
+				   struct inode *gb_inode,
+				   struct ocfs2_dinode *gb_dinode,
+				   struct ocfs2_chain_rec *rec,
+				   struct ocfs2_info_freefrag *ffg,
+				   u32 chunks_in_group)
+{
+	int status = 0, used;
+	u64 blkno;
+
+	struct buffer_head *bh = NULL;
+	struct ocfs2_group_desc *bg = NULL;
+
+	unsigned int max_bits, num_clusters;
+	unsigned int offset = 0, cluster, chunk;
+	unsigned int chunk_free, last_chunksize = 0;
+
+	if (!le32_to_cpu(rec->c_free))
+		goto bail;
+
+	do {
+		if (!bg)
+			blkno = le64_to_cpu(rec->c_blkno);
+		else
+			blkno = le64_to_cpu(bg->bg_next_group);
+
+		if (bh) {
+			brelse(bh);
+			bh = NULL;
+		}
+
+		if (o2info_coherent(&ffg->iff_req))
+			status = ocfs2_read_group_descriptor(gb_inode,
+							     gb_dinode,
+							     blkno, &bh);
+		else
+			status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
+
+		if (status < 0) {
+			mlog(ML_ERROR, "Can't read the group descriptor # "
+			     "%llu from device.", (unsigned long long)blkno);
+			status = -EIO;
+			goto bail;
+		}
+
+		bg = (struct ocfs2_group_desc *)bh->b_data;
+
+		if (!le16_to_cpu(bg->bg_free_bits_count))
+			continue;
+
+		max_bits = le16_to_cpu(bg->bg_bits);
+		offset = 0;
+
+		for (chunk = 0; chunk < chunks_in_group; chunk++) {
+			/*
+			 * last chunk may be not an entire one.
+			 */
+			if ((offset + ffg->iff_chunksize) > max_bits)
+				num_clusters = max_bits - offset;
+			else
+				num_clusters = ffg->iff_chunksize;
+
+			chunk_free = 0;
+			for (cluster = 0; cluster < num_clusters; cluster++) {
+				used = ocfs2_test_bit(offset,
+						(unsigned long *)bg->bg_bitmap);
+				/*
+				 * - chunk_free counts free clusters in #N chunk.
+				 * - last_chunksize records the size(in) clusters
+				 *   for the last real free chunk being counted.
+				 */
+				if (!used) {
+					last_chunksize++;
+					chunk_free++;
+				}
+
+				if (used && last_chunksize) {
+					ocfs2_info_update_ffg(ffg,
+							      last_chunksize);
+					last_chunksize = 0;
+				}
+
+				offset++;
+			}
+
+			if (chunk_free == ffg->iff_chunksize)
+				ffg->iff_ffs.ffs_free_chunks++;
+		}
+
+		/*
+		 * need to update the info for last free chunk.
+		 */
+		if (last_chunksize)
+			ocfs2_info_update_ffg(ffg, last_chunksize);
+
+	} while (le64_to_cpu(bg->bg_next_group));
+
+bail:
+	brelse(bh);
+
+	return status;
+}
+
+int ocfs2_info_freefrag_scan_bitmap(struct ocfs2_super *osb,
+				    struct inode *gb_inode, u64 blkno,
+				    struct ocfs2_info_freefrag *ffg)
+{
+	u32 chunks_in_group;
+	int status = 0, unlock = 0, i;
+
+	struct buffer_head *bh = NULL;
+	struct ocfs2_chain_list *cl = NULL;
+	struct ocfs2_chain_rec *rec = NULL;
+	struct ocfs2_dinode *gb_dinode = NULL;
+
+	if (gb_inode)
+		mutex_lock(&gb_inode->i_mutex);
+
+	if (o2info_coherent(&ffg->iff_req)) {
+		status = ocfs2_inode_lock(gb_inode, &bh, 0);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+		unlock = 1;
+	} else {
+		status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+	}
+
+	gb_dinode = (struct ocfs2_dinode *)bh->b_data;
+	cl = &(gb_dinode->id2.i_chain);
+
+	/*
+	 * Chunksize(in) clusters from userspace should be
+	 * less than clusters in a group.
+	 */
+	if (ffg->iff_chunksize > le16_to_cpu(cl->cl_cpg)) {
+		status = -EINVAL;
+		goto bail;
+	}
+
+	memset(&ffg->iff_ffs, 0, sizeof(struct ocfs2_info_freefrag_stats));
+
+	ffg->iff_ffs.ffs_min = ~0U;
+	ffg->iff_ffs.ffs_clusters =
+			le32_to_cpu(gb_dinode->id1.bitmap1.i_total);
+	ffg->iff_ffs.ffs_free_clusters = ffg->iff_ffs.ffs_clusters -
+			le32_to_cpu(gb_dinode->id1.bitmap1.i_used);
+
+	chunks_in_group = le16_to_cpu(cl->cl_cpg) / ffg->iff_chunksize + 1;
+
+	for (i = 0; i < le16_to_cpu(cl->cl_next_free_rec); i++) {
+		rec = &(cl->cl_recs[i]);
+		status = ocfs2_info_freefrag_scan_chain(osb, gb_inode,
+							gb_dinode,
+							rec, ffg,
+							chunks_in_group);
+		if (status)
+			goto bail;
+	}
+
+	if (ffg->iff_ffs.ffs_free_chunks_real)
+		ffg->iff_ffs.ffs_avg = (ffg->iff_ffs.ffs_avg /
+					ffg->iff_ffs.ffs_free_chunks_real);
+bail:
+	if (unlock)
+		ocfs2_inode_unlock(gb_inode, 0);
+
+	if (gb_inode)
+		mutex_unlock(&gb_inode->i_mutex);
+
+	if (gb_inode)
+		iput(gb_inode);
+
+	brelse(bh);
+
+	return status;
+}
+
+int ocfs2_info_handle_freefrag(struct inode *inode,
+			       struct ocfs2_info_request __user *req)
+{
+	u64 blkno = -1;
+	char namebuf[40];
+	int status = -EFAULT, type = GLOBAL_BITMAP_SYSTEM_INODE;
+
+	struct ocfs2_info_freefrag *oiff;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct inode *gb_inode = NULL;
+
+	oiff = kzalloc(sizeof(struct ocfs2_info_freefrag), GFP_KERNEL);
+	if (!oiff) {
+		status = -ENOMEM;
+		mlog_errno(status);
+		goto bail;
+	}
+
+	if (o2info_from_user(*oiff, req))
+		goto bail;
+	/*
+	 * chunksize from userspace should be power of 2.
+	 */
+	if ((oiff->iff_chunksize & (oiff->iff_chunksize - 1)) ||
+	    (!oiff->iff_chunksize)) {
+		status = -EINVAL;
+		goto bail;
+	}
+
+	if (o2info_coherent(&oiff->iff_req)) {
+		gb_inode = ocfs2_get_system_file_inode(osb, type,
+						       OCFS2_INVALID_SLOT);
+		if (!gb_inode) {
+			mlog(ML_ERROR, "unable to get global_bitmap inode\n");
+			status = -EIO;
+			goto bail;
+		}
+	} else {
+		ocfs2_sprintf_system_inode_name(namebuf, sizeof(namebuf), type,
+						OCFS2_INVALID_SLOT);
+		status = ocfs2_lookup_ino_from_name(osb->sys_root_inode,
+						    namebuf,
+						    strlen(namebuf),
+						    &blkno);
+		if (status < 0) {
+			status = -ENOENT;
+			goto bail;
+		}
+	}
+
+	status = ocfs2_info_freefrag_scan_bitmap(osb, gb_inode, blkno, oiff);
+	if (status < 0)
+		goto bail;
+
+	o2info_set_request_filled(&oiff->iff_req);
+
+	if (o2info_to_user(*oiff, req))
+		goto bail;
+
+	status = 0;
+bail:
+	if (status)
+		o2info_set_request_error(&oiff->iff_req, req);
+
+	kfree(oiff);
+
+	return status;
+}
+
 int ocfs2_info_handle_unknown(struct inode *inode,
 			      struct ocfs2_info_request __user *req)
 {
@@ -508,6 +794,10 @@ int ocfs2_info_handle_request(struct inode *inode,
 		if (oir.ir_size == sizeof(struct ocfs2_info_freeinode))
 			status = ocfs2_info_handle_freeinode(inode, req);
 		break;
+	case OCFS2_INFO_FREEFRAG:
+		if (oir.ir_size == sizeof(struct ocfs2_info_freefrag))
+			status = ocfs2_info_handle_freefrag(inode, req);
+		break;
 	default:
 		status = ocfs2_info_handle_unknown(inode, req);
 		break;
diff --git a/fs/ocfs2/ocfs2_ioctl.h b/fs/ocfs2/ocfs2_ioctl.h
index 6b4b39a..18b6770 100644
--- a/fs/ocfs2/ocfs2_ioctl.h
+++ b/fs/ocfs2/ocfs2_ioctl.h
@@ -152,6 +152,28 @@ struct ocfs2_info_freeinode {
 	__u32 ifi_pad;
 };
 
+#define OCFS2_INFO_MAX_HIST     (32)
+
+struct ocfs2_info_freefrag {
+	struct ocfs2_info_request iff_req;
+	struct ocfs2_info_freefrag_stats { /* (out) */
+		struct ocfs2_info_free_chunk_list {
+			__u32 fc_chunks[OCFS2_INFO_MAX_HIST];
+			__u32 fc_clusters[OCFS2_INFO_MAX_HIST];
+		} ffs_fc_hist;
+		__u32 ffs_clusters;
+		__u32 ffs_free_clusters;
+		__u32 ffs_free_chunks;
+		__u32 ffs_free_chunks_real;
+		__u32 ffs_min; /* Minimum free chunksize in clusters */
+		__u32 ffs_max;
+		__u32 ffs_avg;
+		__u32 ffs_pad;
+	} iff_ffs;
+	__u32 iff_chunksize; /* chunksize in clusters(in) */
+	__u32 iff_pad;
+};
+
 /* Codes for ocfs2_info_request */
 enum ocfs2_info_type {
 	OCFS2_INFO_CLUSTERSIZE = 1,
@@ -162,6 +184,7 @@ enum ocfs2_info_type {
 	OCFS2_INFO_FS_FEATURES,
 	OCFS2_INFO_JOURNAL_SIZE,
 	OCFS2_INFO_FREEINODE,
+	OCFS2_INFO_FREEFRAG,
 	OCFS2_INFO_NUM_TYPES
 };
 
-- 
1.5.5

[Ocfs2-devel] [PATCH 3/3] Ocfs2: Add a new code 'OCFS2_INFO_FREEFRAG' for o2info ioctl.

[Tristan Ye]

This new code is a bit more complicated than former ones, the goal is to
show user all statistics required to take a deep insight into filesystem
on how the disk is being fragmentaed.

The goal is achieved by scaning global bitmap from (cluster)group to group
to figure out following factors in the filesystem:

        - How many free chunks in a fixed size as user requested.
        - How many real free chunks in all size.
        - Min/Max/Avg size(in) clusters of free chunks.
        - How do free chunks distribute(in size) in terms of a histogram,
          just like following:
          ---------------------------------------------------------
          Extent Size Range :  Free extents  Free Clusters  Percent
             32K...   64K-  :             1             1    0.00%
              1M...    2M-  :             9           288    0.03%
              8M...   16M-  :             2           831    0.09%
             32M...   64M-  :             1          2047    0.23%
            128M...  256M-  :             1          8191    0.92%
            256M...  512M-  :             2         21706    2.43%
            512M... 1024M-  :            27        858623   96.29%
          ---------------------------------------------------------

Userspace ioctl() call eventually gets the above info returned by passing
a 'struct ocfs2_info_freefrag' with the chunk_size being specified first.

Signed-off-by: Tristan Ye <tristan.ye@oracle.com>
---
 fs/ocfs2/ioctl.c       |  292 ++++++++++++++++++++++++++++++++++++++++++++++++
 fs/ocfs2/ocfs2_ioctl.h |   23 ++++
 2 files changed, 315 insertions(+), 0 deletions(-)

diff --git a/fs/ocfs2/ioctl.c b/fs/ocfs2/ioctl.c
index ae7f89d..a2b562b 100644
--- a/fs/ocfs2/ioctl.c
+++ b/fs/ocfs2/ioctl.c
@@ -26,6 +26,7 @@
 #include "sysfile.h"
 #include "dir.h"
 #include "buffer_head_io.h"
+#include "suballoc.h"
 
 #include <linux/ext2_fs.h>
 
@@ -437,6 +438,293 @@ bail:
 	return status;
 }
 
+static void o2ffg_update_histogram(struct ocfs2_info_free_chunk_list *hist,
+				   unsigned int chunksize)
+{
+	int index;
+
+	index = __ilog2_u32(chunksize);
+	if (index >= OCFS2_INFO_MAX_HIST)
+		index = OCFS2_INFO_MAX_HIST - 1;
+
+	hist->fc_chunks[index]++;
+	hist->fc_clusters[index] += chunksize;
+}
+
+static void o2ffg_update_stats(struct ocfs2_info_freefrag_stats *stats,
+			       unsigned int chunksize)
+{
+	if (chunksize > stats->ffs_max)
+		stats->ffs_max = chunksize;
+
+	if (chunksize < stats->ffs_min)
+		stats->ffs_min = chunksize;
+
+	stats->ffs_avg += chunksize;
+	stats->ffs_free_chunks_real++;
+}
+
+void ocfs2_info_update_ffg(struct ocfs2_info_freefrag *ffg,
+			   unsigned int chunksize)
+{
+	o2ffg_update_histogram(&(ffg->iff_ffs.ffs_fc_hist), chunksize);
+	o2ffg_update_stats(&(ffg->iff_ffs), chunksize);
+}
+
+int ocfs2_info_freefrag_scan_chain(struct ocfs2_super *osb,
+				   struct inode *gb_inode,
+				   struct ocfs2_dinode *gb_dinode,
+				   struct ocfs2_chain_rec *rec,
+				   struct ocfs2_info_freefrag *ffg,
+				   u32 chunks_in_group)
+{
+	int status = 0, used;
+	u64 blkno;
+
+	struct buffer_head *bh = NULL;
+	struct ocfs2_group_desc *bg = NULL;
+
+	unsigned int max_bits, num_clusters;
+	unsigned int offset = 0, cluster, chunk;
+	unsigned int chunk_free, last_chunksize = 0;
+
+	if (!le32_to_cpu(rec->c_free))
+		goto bail;
+
+	do {
+		if (!bg)
+			blkno = le64_to_cpu(rec->c_blkno);
+		else
+			blkno = le64_to_cpu(bg->bg_next_group);
+
+		if (bh) {
+			brelse(bh);
+			bh = NULL;
+		}
+
+		if (o2info_coherent(&ffg->iff_req))
+			status = ocfs2_read_group_descriptor(gb_inode,
+							     gb_dinode,
+							     blkno, &bh);
+		else
+			status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
+
+		if (status < 0) {
+			mlog(ML_ERROR, "Can't read the group descriptor # "
+			     "%llu from device.", (unsigned long long)blkno);
+			status = -EIO;
+			goto bail;
+		}
+
+		bg = (struct ocfs2_group_desc *)bh->b_data;
+
+		if (!le16_to_cpu(bg->bg_free_bits_count))
+			continue;
+
+		max_bits = le16_to_cpu(bg->bg_bits);
+		offset = 0;
+
+		for (chunk = 0; chunk < chunks_in_group; chunk++) {
+			/*
+			 * last chunk may be not an entire one.
+			 */
+			if ((offset + ffg->iff_chunksize) > max_bits)
+				num_clusters = max_bits - offset;
+			else
+				num_clusters = ffg->iff_chunksize;
+
+			chunk_free = 0;
+			for (cluster = 0; cluster < num_clusters; cluster++) {
+				used = ocfs2_test_bit(offset,
+						(unsigned long *)bg->bg_bitmap);
+				/*
+				 * - chunk_free counts free clusters in #N chunk.
+				 * - last_chunksize records the size(in) clusters
+				 *   for the last real free chunk being counted.
+				 */
+				if (!used) {
+					last_chunksize++;
+					chunk_free++;
+				}
+
+				if (used && last_chunksize) {
+					ocfs2_info_update_ffg(ffg,
+							      last_chunksize);
+					last_chunksize = 0;
+				}
+
+				offset++;
+			}
+
+			if (chunk_free == ffg->iff_chunksize)
+				ffg->iff_ffs.ffs_free_chunks++;
+		}
+
+		/*
+		 * need to update the info for last free chunk.
+		 */
+		if (last_chunksize)
+			ocfs2_info_update_ffg(ffg, last_chunksize);
+
+	} while (le64_to_cpu(bg->bg_next_group));
+
+bail:
+	brelse(bh);
+
+	mlog_exit(status);
+	return status;
+}
+
+int ocfs2_info_freefrag_scan_bitmap(struct ocfs2_super *osb,
+				    struct inode *gb_inode, u64 blkno,
+				    struct ocfs2_info_freefrag *ffg)
+{
+	u32 chunks_in_group;
+	int status = 0, unlock = 0, i;
+
+	struct buffer_head *bh = NULL;
+	struct ocfs2_chain_list *cl = NULL;
+	struct ocfs2_chain_rec *rec = NULL;
+	struct ocfs2_dinode *gb_dinode = NULL;
+
+	if (gb_inode)
+		mutex_lock(&gb_inode->i_mutex);
+
+	if (o2info_coherent(&ffg->iff_req)) {
+		status = ocfs2_inode_lock(gb_inode, &bh, 0);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+		unlock = 1;
+	} else {
+		status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+	}
+
+	gb_dinode = (struct ocfs2_dinode *)bh->b_data;
+	cl = &(gb_dinode->id2.i_chain);
+
+	/*
+	 * Chunksize(in) clusters from userspace should be
+	 * less than clusters in a group.
+	 */
+	if (ffg->iff_chunksize > le16_to_cpu(cl->cl_cpg)) {
+		status = -EINVAL;
+		goto bail;
+	}
+
+	memset(&ffg->iff_ffs, 0, sizeof(struct ocfs2_info_freefrag_stats));
+
+	ffg->iff_ffs.ffs_min = ~0U;
+	ffg->iff_ffs.ffs_clusters =
+			le32_to_cpu(gb_dinode->id1.bitmap1.i_total);
+	ffg->iff_ffs.ffs_free_clusters = ffg->iff_ffs.ffs_clusters -
+			le32_to_cpu(gb_dinode->id1.bitmap1.i_used);
+
+	chunks_in_group = le16_to_cpu(cl->cl_cpg) / ffg->iff_chunksize + 1;
+
+	for (i = 0; i < le16_to_cpu(cl->cl_next_free_rec); i++) {
+		rec = &(cl->cl_recs[i]);
+		status = ocfs2_info_freefrag_scan_chain(osb, gb_inode,
+							gb_dinode,
+							rec, ffg,
+							chunks_in_group);
+		if (status)
+			goto bail;
+	}
+
+	if (ffg->iff_ffs.ffs_free_chunks_real)
+		ffg->iff_ffs.ffs_avg = (ffg->iff_ffs.ffs_avg /
+					ffg->iff_ffs.ffs_free_chunks_real);
+bail:
+	if (unlock)
+		ocfs2_inode_unlock(gb_inode, 0);
+
+	if (gb_inode)
+		mutex_unlock(&gb_inode->i_mutex);
+
+	if (gb_inode)
+		iput(gb_inode);
+
+	brelse(bh);
+
+	mlog_exit(status);
+	return status;
+}
+
+int ocfs2_info_handle_freefrag(struct inode *inode,
+			       struct ocfs2_info_request __user *req)
+{
+	u64 blkno = -1;
+	char namebuf[40];
+	int status = -EFAULT, type = GLOBAL_BITMAP_SYSTEM_INODE;
+
+	struct ocfs2_info_freefrag *oiff;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct inode *gb_inode = NULL;
+
+	oiff = kzalloc(sizeof(struct ocfs2_info_freefrag), GFP_KERNEL);
+	if (!oiff) {
+		status = -ENOMEM;
+		mlog_errno(status);
+		goto bail;
+	}
+
+	if (o2info_from_user(*oiff, req))
+		goto bail;
+	/*
+	 * chunksize from userspace should be power of 2.
+	 */
+	if ((oiff->iff_chunksize & (oiff->iff_chunksize - 1)) ||
+	    (!oiff->iff_chunksize)) {
+		status = -EINVAL;
+		goto bail;
+	}
+
+	if (o2info_coherent(&oiff->iff_req)) {
+		gb_inode = ocfs2_get_system_file_inode(osb, type,
+						       OCFS2_INVALID_SLOT);
+		if (!gb_inode) {
+			mlog(ML_ERROR, "unable to get global_bitmap inode\n");
+			status = -EIO;
+			goto bail;
+		}
+	} else {
+		ocfs2_sprintf_system_inode_name(namebuf, sizeof(namebuf), type,
+						OCFS2_INVALID_SLOT);
+		status = ocfs2_lookup_ino_from_name(osb->sys_root_inode,
+						    namebuf,
+						    strlen(namebuf),
+						    &blkno);
+		if (status < 0) {
+			status = -ENOENT;
+			goto bail;
+		}
+	}
+
+	status = ocfs2_info_freefrag_scan_bitmap(osb, gb_inode, blkno, oiff);
+	if (status < 0)
+		goto bail;
+
+	o2info_set_request_filled(&oiff->iff_req);
+
+	if (o2info_to_user(*oiff, req))
+		goto bail;
+
+	status = 0;
+bail:
+	if (status)
+		o2info_set_request_error(&oiff->iff_req, req);
+
+	kfree(oiff);
+
+	return status;
+}
+
 int ocfs2_info_handle_unknown(struct inode *inode,
 			      struct ocfs2_info_request __user *req)
 {
@@ -512,6 +800,10 @@ int ocfs2_info_handle_request(struct inode *inode,
 		if (oir.ir_size == sizeof(struct ocfs2_info_freeinode))
 			status = ocfs2_info_handle_freeinode(inode, req);
 		break;
+	case OCFS2_INFO_FREEFRAG:
+		if (oir.ir_size == sizeof(struct ocfs2_info_freefrag))
+			status = ocfs2_info_handle_freefrag(inode, req);
+		break;
 	default:
 		status = ocfs2_info_handle_unknown(inode, req);
 		break;
diff --git a/fs/ocfs2/ocfs2_ioctl.h b/fs/ocfs2/ocfs2_ioctl.h
index 6b4b39a..18b6770 100644
--- a/fs/ocfs2/ocfs2_ioctl.h
+++ b/fs/ocfs2/ocfs2_ioctl.h
@@ -152,6 +152,28 @@ struct ocfs2_info_freeinode {
 	__u32 ifi_pad;
 };
 
+#define OCFS2_INFO_MAX_HIST     (32)
+
+struct ocfs2_info_freefrag {
+	struct ocfs2_info_request iff_req;
+	struct ocfs2_info_freefrag_stats { /* (out) */
+		struct ocfs2_info_free_chunk_list {
+			__u32 fc_chunks[OCFS2_INFO_MAX_HIST];
+			__u32 fc_clusters[OCFS2_INFO_MAX_HIST];
+		} ffs_fc_hist;
+		__u32 ffs_clusters;
+		__u32 ffs_free_clusters;
+		__u32 ffs_free_chunks;
+		__u32 ffs_free_chunks_real;
+		__u32 ffs_min; /* Minimum free chunksize in clusters */
+		__u32 ffs_max;
+		__u32 ffs_avg;
+		__u32 ffs_pad;
+	} iff_ffs;
+	__u32 iff_chunksize; /* chunksize in clusters(in) */
+	__u32 iff_pad;
+};
+
 /* Codes for ocfs2_info_request */
 enum ocfs2_info_type {
 	OCFS2_INFO_CLUSTERSIZE = 1,
@@ -162,6 +184,7 @@ enum ocfs2_info_type {
 	OCFS2_INFO_FS_FEATURES,
 	OCFS2_INFO_JOURNAL_SIZE,
 	OCFS2_INFO_FREEINODE,
+	OCFS2_INFO_FREEFRAG,
 	OCFS2_INFO_NUM_TYPES
 };
 
-- 
1.5.5

[Ocfs2-devel] [PATCH 3/3] Ocfs2: Add a new code 'OCFS2_INFO_FREEFRAG' for o2info ioctl.

[Tristan Ye]

This new code is a bit more complicated than former ones, the goal is to
show user all statistics required to take a deep insight into filesystem
on how the disk is being fragmentaed.

The goal is achieved by scaning global bitmap from (cluster)group to group
to figure out following factors in the filesystem:

        - How many free chunks in a fixed size as user requested.
        - How many real free chunks in all size.
        - Min/Max/Avg size(in) clusters of free chunks.
        - How do free chunks distribute(in size) in terms of a histogram,
          just like following:
          ---------------------------------------------------------
          Extent Size Range :  Free extents  Free Clusters  Percent
             32K...   64K-  :             1             1    0.00%
              1M...    2M-  :             9           288    0.03%
              8M...   16M-  :             2           831    0.09%
             32M...   64M-  :             1          2047    0.23%
            128M...  256M-  :             1          8191    0.92%
            256M...  512M-  :             2         21706    2.43%
            512M... 1024M-  :            27        858623   96.29%
          ---------------------------------------------------------

Userspace ioctl() call eventually gets the above info returned by passing
a 'struct ocfs2_info_freefrag' with the chunk_size being specified first.

Signed-off-by: Tristan Ye <tristan.ye@oracle.com>
---
 fs/ocfs2/ioctl.c       |  292 ++++++++++++++++++++++++++++++++++++++++++++++++
 fs/ocfs2/ocfs2_ioctl.h |   23 ++++
 2 files changed, 315 insertions(+), 0 deletions(-)

diff --git a/fs/ocfs2/ioctl.c b/fs/ocfs2/ioctl.c
index ae7f89d..a2b562b 100644
--- a/fs/ocfs2/ioctl.c
+++ b/fs/ocfs2/ioctl.c
@@ -26,6 +26,7 @@
 #include "sysfile.h"
 #include "dir.h"
 #include "buffer_head_io.h"
+#include "suballoc.h"
 
 #include <linux/ext2_fs.h>
 
@@ -437,6 +438,293 @@ bail:
 	return status;
 }
 
+static void o2ffg_update_histogram(struct ocfs2_info_free_chunk_list *hist,
+				   unsigned int chunksize)
+{
+	int index;
+
+	index = __ilog2_u32(chunksize);
+	if (index >= OCFS2_INFO_MAX_HIST)
+		index = OCFS2_INFO_MAX_HIST - 1;
+
+	hist->fc_chunks[index]++;
+	hist->fc_clusters[index] += chunksize;
+}
+
+static void o2ffg_update_stats(struct ocfs2_info_freefrag_stats *stats,
+			       unsigned int chunksize)
+{
+	if (chunksize > stats->ffs_max)
+		stats->ffs_max = chunksize;
+
+	if (chunksize < stats->ffs_min)
+		stats->ffs_min = chunksize;
+
+	stats->ffs_avg += chunksize;
+	stats->ffs_free_chunks_real++;
+}
+
+void ocfs2_info_update_ffg(struct ocfs2_info_freefrag *ffg,
+			   unsigned int chunksize)
+{
+	o2ffg_update_histogram(&(ffg->iff_ffs.ffs_fc_hist), chunksize);
+	o2ffg_update_stats(&(ffg->iff_ffs), chunksize);
+}
+
+int ocfs2_info_freefrag_scan_chain(struct ocfs2_super *osb,
+				   struct inode *gb_inode,
+				   struct ocfs2_dinode *gb_dinode,
+				   struct ocfs2_chain_rec *rec,
+				   struct ocfs2_info_freefrag *ffg,
+				   u32 chunks_in_group)
+{
+	int status = 0, used;
+	u64 blkno;
+
+	struct buffer_head *bh = NULL;
+	struct ocfs2_group_desc *bg = NULL;
+
+	unsigned int max_bits, num_clusters;
+	unsigned int offset = 0, cluster, chunk;
+	unsigned int chunk_free, last_chunksize = 0;
+
+	if (!le32_to_cpu(rec->c_free))
+		goto bail;
+
+	do {
+		if (!bg)
+			blkno = le64_to_cpu(rec->c_blkno);
+		else
+			blkno = le64_to_cpu(bg->bg_next_group);
+
+		if (bh) {
+			brelse(bh);
+			bh = NULL;
+		}
+
+		if (o2info_coherent(&ffg->iff_req))
+			status = ocfs2_read_group_descriptor(gb_inode,
+							     gb_dinode,
+							     blkno, &bh);
+		else
+			status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
+
+		if (status < 0) {
+			mlog(ML_ERROR, "Can't read the group descriptor # "
+			     "%llu from device.", (unsigned long long)blkno);
+			status = -EIO;
+			goto bail;
+		}
+
+		bg = (struct ocfs2_group_desc *)bh->b_data;
+
+		if (!le16_to_cpu(bg->bg_free_bits_count))
+			continue;
+
+		max_bits = le16_to_cpu(bg->bg_bits);
+		offset = 0;
+
+		for (chunk = 0; chunk < chunks_in_group; chunk++) {
+			/*
+			 * last chunk may be not an entire one.
+			 */
+			if ((offset + ffg->iff_chunksize) > max_bits)
+				num_clusters = max_bits - offset;
+			else
+				num_clusters = ffg->iff_chunksize;
+
+			chunk_free = 0;
+			for (cluster = 0; cluster < num_clusters; cluster++) {
+				used = ocfs2_test_bit(offset,
+						(unsigned long *)bg->bg_bitmap);
+				/*
+				 * - chunk_free counts free clusters in #N chunk.
+				 * - last_chunksize records the size(in) clusters
+				 *   for the last real free chunk being counted.
+				 */
+				if (!used) {
+					last_chunksize++;
+					chunk_free++;
+				}
+
+				if (used && last_chunksize) {
+					ocfs2_info_update_ffg(ffg,
+							      last_chunksize);
+					last_chunksize = 0;
+				}
+
+				offset++;
+			}
+
+			if (chunk_free == ffg->iff_chunksize)
+				ffg->iff_ffs.ffs_free_chunks++;
+		}
+
+		/*
+		 * need to update the info for last free chunk.
+		 */
+		if (last_chunksize)
+			ocfs2_info_update_ffg(ffg, last_chunksize);
+
+	} while (le64_to_cpu(bg->bg_next_group));
+
+bail:
+	brelse(bh);
+
+	mlog_exit(status);
+	return status;
+}
+
+int ocfs2_info_freefrag_scan_bitmap(struct ocfs2_super *osb,
+				    struct inode *gb_inode, u64 blkno,
+				    struct ocfs2_info_freefrag *ffg)
+{
+	u32 chunks_in_group;
+	int status = 0, unlock = 0, i;
+
+	struct buffer_head *bh = NULL;
+	struct ocfs2_chain_list *cl = NULL;
+	struct ocfs2_chain_rec *rec = NULL;
+	struct ocfs2_dinode *gb_dinode = NULL;
+
+	if (gb_inode)
+		mutex_lock(&gb_inode->i_mutex);
+
+	if (o2info_coherent(&ffg->iff_req)) {
+		status = ocfs2_inode_lock(gb_inode, &bh, 0);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+		unlock = 1;
+	} else {
+		status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+	}
+
+	gb_dinode = (struct ocfs2_dinode *)bh->b_data;
+	cl = &(gb_dinode->id2.i_chain);
+
+	/*
+	 * Chunksize(in) clusters from userspace should be
+	 * less than clusters in a group.
+	 */
+	if (ffg->iff_chunksize > le16_to_cpu(cl->cl_cpg)) {
+		status = -EINVAL;
+		goto bail;
+	}
+
+	memset(&ffg->iff_ffs, 0, sizeof(struct ocfs2_info_freefrag_stats));
+
+	ffg->iff_ffs.ffs_min = ~0U;
+	ffg->iff_ffs.ffs_clusters =
+			le32_to_cpu(gb_dinode->id1.bitmap1.i_total);
+	ffg->iff_ffs.ffs_free_clusters = ffg->iff_ffs.ffs_clusters -
+			le32_to_cpu(gb_dinode->id1.bitmap1.i_used);
+
+	chunks_in_group = le16_to_cpu(cl->cl_cpg) / ffg->iff_chunksize + 1;
+
+	for (i = 0; i < le16_to_cpu(cl->cl_next_free_rec); i++) {
+		rec = &(cl->cl_recs[i]);
+		status = ocfs2_info_freefrag_scan_chain(osb, gb_inode,
+							gb_dinode,
+							rec, ffg,
+							chunks_in_group);
+		if (status)
+			goto bail;
+	}
+
+	if (ffg->iff_ffs.ffs_free_chunks_real)
+		ffg->iff_ffs.ffs_avg = (ffg->iff_ffs.ffs_avg /
+					ffg->iff_ffs.ffs_free_chunks_real);
+bail:
+	if (unlock)
+		ocfs2_inode_unlock(gb_inode, 0);
+
+	if (gb_inode)
+		mutex_unlock(&gb_inode->i_mutex);
+
+	if (gb_inode)
+		iput(gb_inode);
+
+	brelse(bh);
+
+	mlog_exit(status);
+	return status;
+}
+
+int ocfs2_info_handle_freefrag(struct inode *inode,
+			       struct ocfs2_info_request __user *req)
+{
+	u64 blkno = -1;
+	char namebuf[40];
+	int status = -EFAULT, type = GLOBAL_BITMAP_SYSTEM_INODE;
+
+	struct ocfs2_info_freefrag *oiff;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct inode *gb_inode = NULL;
+
+	oiff = kzalloc(sizeof(struct ocfs2_info_freefrag), GFP_NOFS);
+	if (!oiff) {
+		status = -ENOMEM;
+		mlog_errno(status);
+		goto bail;
+	}
+
+	if (o2info_from_user(*oiff, req))
+		goto bail;
+	/*
+	 * chunksize from userspace should be power of 2.
+	 */
+	if ((oiff->iff_chunksize & (oiff->iff_chunksize - 1)) ||
+	    (!oiff->iff_chunksize)) {
+		status = -EINVAL;
+		goto bail;
+	}
+
+	if (o2info_coherent(&oiff->iff_req)) {
+		gb_inode = ocfs2_get_system_file_inode(osb, type,
+						       OCFS2_INVALID_SLOT);
+		if (!gb_inode) {
+			mlog(ML_ERROR, "unable to get global_bitmap inode\n");
+			status = -EIO;
+			goto bail;
+		}
+	} else {
+		ocfs2_sprintf_system_inode_name(namebuf, sizeof(namebuf), type,
+						OCFS2_INVALID_SLOT);
+		status = ocfs2_lookup_ino_from_name(osb->sys_root_inode,
+						    namebuf,
+						    strlen(namebuf),
+						    &blkno);
+		if (status < 0) {
+			status = -ENOENT;
+			goto bail;
+		}
+	}
+
+	status = ocfs2_info_freefrag_scan_bitmap(osb, gb_inode, blkno, oiff);
+	if (status < 0)
+		goto bail;
+
+	o2info_set_request_filled(&oiff->iff_req);
+
+	if (o2info_to_user(*oiff, req))
+		goto bail;
+
+	status = 0;
+bail:
+	if (status)
+		o2info_set_request_error(&oiff->iff_req, req);
+
+	kfree(oiff);
+
+	return status;
+}
+
 int ocfs2_info_handle_unknown(struct inode *inode,
 			      struct ocfs2_info_request __user *req)
 {
@@ -512,6 +800,10 @@ int ocfs2_info_handle_request(struct inode *inode,
 		if (oir.ir_size == sizeof(struct ocfs2_info_freeinode))
 			status = ocfs2_info_handle_freeinode(inode, req);
 		break;
+	case OCFS2_INFO_FREEFRAG:
+		if (oir.ir_size == sizeof(struct ocfs2_info_freefrag))
+			status = ocfs2_info_handle_freefrag(inode, req);
+		break;
 	default:
 		status = ocfs2_info_handle_unknown(inode, req);
 		break;
diff --git a/fs/ocfs2/ocfs2_ioctl.h b/fs/ocfs2/ocfs2_ioctl.h
index 6b4b39a..18b6770 100644
--- a/fs/ocfs2/ocfs2_ioctl.h
+++ b/fs/ocfs2/ocfs2_ioctl.h
@@ -152,6 +152,28 @@ struct ocfs2_info_freeinode {
 	__u32 ifi_pad;
 };
 
+#define OCFS2_INFO_MAX_HIST     (32)
+
+struct ocfs2_info_freefrag {
+	struct ocfs2_info_request iff_req;
+	struct ocfs2_info_freefrag_stats { /* (out) */
+		struct ocfs2_info_free_chunk_list {
+			__u32 fc_chunks[OCFS2_INFO_MAX_HIST];
+			__u32 fc_clusters[OCFS2_INFO_MAX_HIST];
+		} ffs_fc_hist;
+		__u32 ffs_clusters;
+		__u32 ffs_free_clusters;
+		__u32 ffs_free_chunks;
+		__u32 ffs_free_chunks_real;
+		__u32 ffs_min; /* Minimum free chunksize in clusters */
+		__u32 ffs_max;
+		__u32 ffs_avg;
+		__u32 ffs_pad;
+	} iff_ffs;
+	__u32 iff_chunksize; /* chunksize in clusters(in) */
+	__u32 iff_pad;
+};
+
 /* Codes for ocfs2_info_request */
 enum ocfs2_info_type {
 	OCFS2_INFO_CLUSTERSIZE = 1,
@@ -162,6 +184,7 @@ enum ocfs2_info_type {
 	OCFS2_INFO_FS_FEATURES,
 	OCFS2_INFO_JOURNAL_SIZE,
 	OCFS2_INFO_FREEINODE,
+	OCFS2_INFO_FREEFRAG,
 	OCFS2_INFO_NUM_TYPES
 };
 
-- 
1.5.5

[Ocfs2-devel] [PATCH 3/3] Ocfs2: Add a new code 'OCFS2_INFO_FREEFRAG' for o2info ioctl.

[Tristan Ye]

This new code is a bit more complicated than former ones, the goal is to
show user all statistics required to take a deep insight into filesystem
on how the disk is being fragmentaed.

The goal is achieved by scaning global bitmap from (cluster)group to group
to figure out following factors in the filesystem:

        - How many free chunks in a fixed size as user requested.
        - How many real free chunks in all size.
        - Min/Max/Avg size(in) clusters of free chunks.
        - How do free chunks distribute(in size) in terms of a histogram,
          just like following:
          ---------------------------------------------------------
          Extent Size Range :  Free extents  Free Clusters  Percent
             32K...   64K-  :             1             1    0.00%
              1M...    2M-  :             9           288    0.03%
              8M...   16M-  :             2           831    0.09%
             32M...   64M-  :             1          2047    0.23%
            128M...  256M-  :             1          8191    0.92%
            256M...  512M-  :             2         21706    2.43%
            512M... 1024M-  :            27        858623   96.29%
          ---------------------------------------------------------

Userspace ioctl() call eventually gets the above info returned by passing
a 'struct ocfs2_info_freefrag' with the chunk_size being specified first.

Signed-off-by: Tristan Ye <tristan.ye@oracle.com>
---
 fs/ocfs2/ioctl.c       |  283 ++++++++++++++++++++++++++++++++++++++++++++++++
 fs/ocfs2/ocfs2_ioctl.h |   23 ++++
 2 files changed, 306 insertions(+), 0 deletions(-)

diff --git a/fs/ocfs2/ioctl.c b/fs/ocfs2/ioctl.c
index 8aa2100..396798c 100644
--- a/fs/ocfs2/ioctl.c
+++ b/fs/ocfs2/ioctl.c
@@ -26,6 +26,7 @@
 #include "sysfile.h"
 #include "dir.h"
 #include "buffer_head_io.h"
+#include "suballoc.h"
 
 #include <linux/ext2_fs.h>
 
@@ -428,6 +429,284 @@ bail:
 	return status;
 }
 
+static void o2ffg_update_histogram(struct ocfs2_info_free_chunk_list *hist,
+				   unsigned int chunksize)
+{
+	int index;
+
+	index = __ilog2_u32(chunksize);
+	if (index >= OCFS2_INFO_MAX_HIST)
+		index = OCFS2_INFO_MAX_HIST - 1;
+
+	hist->fc_chunks[index]++;
+	hist->fc_clusters[index] += chunksize;
+}
+
+static void o2ffg_update_stats(struct ocfs2_info_freefrag_stats *stats,
+			       unsigned int chunksize)
+{
+	if (chunksize > stats->ffs_max)
+		stats->ffs_max = chunksize;
+
+	if (chunksize < stats->ffs_min)
+		stats->ffs_min = chunksize;
+
+	stats->ffs_avg += chunksize;
+	stats->ffs_free_chunks_real++;
+}
+
+void ocfs2_info_update_ffg(struct ocfs2_info_freefrag *ffg,
+			   unsigned int chunksize)
+{
+	o2ffg_update_histogram(&(ffg->iff_ffs.ffs_fc_hist), chunksize);
+	o2ffg_update_stats(&(ffg->iff_ffs), chunksize);
+}
+
+int ocfs2_info_freefrag_scan_chain(struct ocfs2_super *osb,
+				   struct inode *gb_inode,
+				   struct ocfs2_dinode *gb_dinode,
+				   struct ocfs2_chain_rec *rec,
+				   struct ocfs2_info_freefrag *ffg,
+				   u32 chunks_in_group)
+{
+	int status = 0, used;
+	u64 blkno;
+
+	struct buffer_head *bh = NULL;
+	struct ocfs2_group_desc *bg = NULL;
+
+	unsigned int max_bits, num_clusters;
+	unsigned int offset = 0, cluster, chunk;
+	unsigned int chunk_free, last_chunksize = 0;
+
+	if (!le32_to_cpu(rec->c_free))
+		goto bail;
+
+	do {
+		if (!bg)
+			blkno = le64_to_cpu(rec->c_blkno);
+		else
+			blkno = le64_to_cpu(bg->bg_next_group);
+
+		if (bh) {
+			brelse(bh);
+			bh = NULL;
+		}
+
+		if (o2info_coherent(&ffg->iff_req))
+			status = ocfs2_read_group_descriptor(gb_inode,
+							     gb_dinode,
+							     blkno, &bh);
+		else
+			status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
+
+		if (status < 0) {
+			mlog(ML_ERROR, "Can't read the group descriptor # "
+			     "%llu from device.", (unsigned long long)blkno);
+			status = -EIO;
+			goto bail;
+		}
+
+		bg = (struct ocfs2_group_desc *)bh->b_data;
+
+		if (!le16_to_cpu(bg->bg_free_bits_count))
+			continue;
+
+		max_bits = le16_to_cpu(bg->bg_bits);
+		offset = 0;
+
+		for (chunk = 0; chunk < chunks_in_group; chunk++) {
+			/*
+			 * last chunk may be not an entire one.
+			 */
+			if ((offset + ffg->iff_chunksize) > max_bits)
+				num_clusters = max_bits - offset;
+			else
+				num_clusters = ffg->iff_chunksize;
+
+			chunk_free = 0;
+			for (cluster = 0; cluster < num_clusters; cluster++) {
+				used = ocfs2_test_bit(offset,
+						(unsigned long *)bg->bg_bitmap);
+				/*
+				 * - chunk_free counts free clusters in #N chunk.
+				 * - last_chunksize records the size(in) clusters
+				 *   for the last real free chunk being counted.
+				 */
+				if (!used) {
+					last_chunksize++;
+					chunk_free++;
+				}
+
+				if (used && last_chunksize) {
+					ocfs2_info_update_ffg(ffg,
+							      last_chunksize);
+					last_chunksize = 0;
+				}
+
+				offset++;
+			}
+
+			if (chunk_free == ffg->iff_chunksize)
+				ffg->iff_ffs.ffs_free_chunks++;
+		}
+
+		/*
+		 * need to update the info for last free chunk.
+		 */
+		if (last_chunksize)
+			ocfs2_info_update_ffg(ffg, last_chunksize);
+
+	} while (le64_to_cpu(bg->bg_next_group));
+
+bail:
+	brelse(bh);
+
+	mlog_exit(status);
+	return status;
+}
+
+int ocfs2_info_freefrag_scan_bitmap(struct ocfs2_super *osb,
+				    struct inode *gb_inode, u64 blkno,
+				    struct ocfs2_info_freefrag *ffg)
+{
+	u32 chunks_in_group;
+	int status = 0, unlock = 0, i;
+
+	struct buffer_head *bh = NULL;
+	struct ocfs2_chain_list *cl = NULL;
+	struct ocfs2_chain_rec *rec = NULL;
+	struct ocfs2_dinode *gb_dinode = NULL;
+
+	if (gb_inode)
+		mutex_lock(&gb_inode->i_mutex);
+
+	if (o2info_coherent(&ffg->iff_req)) {
+		status = ocfs2_inode_lock(gb_inode, &bh, 0);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+		unlock = 1;
+	} else {
+		status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+	}
+
+	gb_dinode = (struct ocfs2_dinode *)bh->b_data;
+	cl = &(gb_dinode->id2.i_chain);
+
+	/*
+	 * Chunksize(in) clusters from userspace should be
+	 * less than clusters in a group.
+	 */
+	if (ffg->iff_chunksize > le16_to_cpu(cl->cl_cpg)) {
+		status = -EINVAL;
+		goto bail;
+	}
+
+	memset(&ffg->iff_ffs, 0, sizeof(struct ocfs2_info_freefrag_stats));
+
+	ffg->iff_ffs.ffs_min = ~0U;
+	ffg->iff_ffs.ffs_clusters =
+			le32_to_cpu(gb_dinode->id1.bitmap1.i_total);
+	ffg->iff_ffs.ffs_free_clusters = ffg->iff_ffs.ffs_clusters -
+			le32_to_cpu(gb_dinode->id1.bitmap1.i_used);
+
+	chunks_in_group = le16_to_cpu(cl->cl_cpg) / ffg->iff_chunksize + 1;
+
+	for (i = 0; i < le16_to_cpu(cl->cl_next_free_rec); i++) {
+		rec = &(cl->cl_recs[i]);
+		status = ocfs2_info_freefrag_scan_chain(osb, gb_inode,
+							gb_dinode,
+							rec, ffg,
+							chunks_in_group);
+		if (status)
+			goto bail;
+	}
+
+	if (ffg->iff_ffs.ffs_free_chunks_real)
+		ffg->iff_ffs.ffs_avg = (ffg->iff_ffs.ffs_avg /
+					ffg->iff_ffs.ffs_free_chunks_real);
+bail:
+	if (unlock)
+		ocfs2_inode_unlock(gb_inode, 0);
+
+	if (gb_inode)
+		mutex_unlock(&gb_inode->i_mutex);
+
+	if (gb_inode)
+		iput(gb_inode);
+
+	brelse(bh);
+
+	mlog_exit(status);
+	return status;
+}
+
+int ocfs2_info_handle_freefrag(struct inode *inode,
+			       struct ocfs2_info_request __user *req)
+{
+	u64 blkno = -1;
+	char namebuf[40];
+	int status = -EFAULT, type = GLOBAL_BITMAP_SYSTEM_INODE;
+
+	struct ocfs2_info_freefrag oiff;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct inode *gb_inode = NULL;
+
+	if (o2info_from_user(oiff, req))
+		goto bail;
+	/*
+	 * chunksize from userspace should be power of 2.
+	 */
+	if ((oiff.iff_chunksize & (oiff.iff_chunksize - 1)) ||
+	    (!oiff.iff_chunksize)) {
+		status = -EINVAL;
+		goto bail;
+	}
+
+	if (o2info_coherent(&oiff.iff_req)) {
+		gb_inode = ocfs2_get_system_file_inode(osb, type,
+						       OCFS2_INVALID_SLOT);
+		if (!gb_inode) {
+			mlog(ML_ERROR, "unable to get global_bitmap inode\n");
+			status = -EIO;
+			goto bail;
+		}
+	} else {
+		ocfs2_sprintf_system_inode_name(namebuf, sizeof(namebuf), type,
+						OCFS2_INVALID_SLOT);
+		status = ocfs2_lookup_ino_from_name(osb->sys_root_inode,
+						    namebuf,
+						    strlen(namebuf),
+						    &blkno);
+		if (status < 0) {
+			status = -ENOENT;
+			goto bail;
+		}
+	}
+
+	status = ocfs2_info_freefrag_scan_bitmap(osb, gb_inode, blkno, &oiff);
+	if (status < 0)
+		goto bail;
+
+	o2info_set_request_filled(&oiff.iff_req);
+
+	if (o2info_to_user(oiff, req))
+		goto bail;
+
+	status = 0;
+bail:
+	if (status)
+		o2info_set_request_error(&oiff.iff_req, req);
+
+	return status;
+}
+
 int ocfs2_info_handle_unknown(struct inode *inode,
 			      struct ocfs2_info_request __user *req)
 {
@@ -503,6 +782,10 @@ int ocfs2_info_handle_request(struct inode *inode,
 		if (oir.ir_size == sizeof(struct ocfs2_info_freeinode))
 			status = ocfs2_info_handle_freeinode(inode, req);
 		break;
+	case OCFS2_INFO_FREEFRAG:
+		if (oir.ir_size == sizeof(struct ocfs2_info_freefrag))
+			status = ocfs2_info_handle_freefrag(inode, req);
+		break;
 	default:
 		status = ocfs2_info_handle_unknown(inode, req);
 		break;
diff --git a/fs/ocfs2/ocfs2_ioctl.h b/fs/ocfs2/ocfs2_ioctl.h
index b74bd33..b7afff6 100644
--- a/fs/ocfs2/ocfs2_ioctl.h
+++ b/fs/ocfs2/ocfs2_ioctl.h
@@ -152,6 +152,28 @@ struct ocfs2_info_freeinode {
 	} ifi_stat[OCFS2_MAX_SLOTS];
 };
 
+#define OCFS2_INFO_MAX_HIST     (32)
+
+struct ocfs2_info_freefrag {
+	struct ocfs2_info_request iff_req;
+	struct ocfs2_info_freefrag_stats { /* (out) */
+		struct ocfs2_info_free_chunk_list {
+			__u32 fc_chunks[OCFS2_INFO_MAX_HIST];
+			__u32 fc_clusters[OCFS2_INFO_MAX_HIST];
+		} ffs_fc_hist;
+		__u32 ffs_clusters;
+		__u32 ffs_free_clusters;
+		__u32 ffs_free_chunks;
+		__u32 ffs_free_chunks_real;
+		__u32 ffs_min; /* Minimum free chunksize in clusters */
+		__u32 ffs_max;
+		__u32 ffs_avg;
+		__u32 ffs_pad;
+	} iff_ffs;
+	__u32 iff_chunksize; /* chunksize in clusters(in) */
+	__u32 iff_pad;
+};
+
 /* Codes for ocfs2_info_request */
 enum ocfs2_info_type {
 	OCFS2_INFO_CLUSTERSIZE = 1,
@@ -162,6 +184,7 @@ enum ocfs2_info_type {
 	OCFS2_INFO_FS_FEATURES,
 	OCFS2_INFO_JOURNAL_SIZE,
 	OCFS2_INFO_FREEINODE,
+	OCFS2_INFO_FREEFRAG,
 	OCFS2_INFO_NUM_TYPES
 };
 
-- 
1.5.5

[Ocfs2-devel] [PATCH 3/3] Ocfs2: Add a new code 'OCFS2_INFO_FREEFRAG' for o2info ioctl.

[Tristan Ye]

This new code is a bit more complicated than former ones, the goal is to
show user all statistics required to take a deep insight into filesystem
on how the disk is being fragmentaed.

The goal is achieved by scaning global bitmap from (cluster)group to group
to figure out following factors in the filesystem:

        - How many free chunks in a fixed size as user requested.
        - How many real free chunks in all size.
        - Min/Max/Avg size(in) clusters of free chunks.
        - How do free chunks distribute(in size) in terms of a histogram,
          just like following:
          ---------------------------------------------------------
          Extent Size Range :  Free extents  Free Clusters  Percent
             32K...   64K-  :             1             1    0.00%
              1M...    2M-  :             9           288    0.03%
              8M...   16M-  :             2           831    0.09%
             32M...   64M-  :             1          2047    0.23%
            128M...  256M-  :             1          8191    0.92%
            256M...  512M-  :             2         21706    2.43%
            512M... 1024M-  :            27        858623   96.29%
          ---------------------------------------------------------

Userspace ioctl() call eventually gets the above info returned by passing
a 'struct ocfs2_info_freefrag' with the chunk_size being specified first.

Signed-off-by: Tristan Ye <tristan.ye@oracle.com>
---
 fs/ocfs2/ioctl.c       |  283 ++++++++++++++++++++++++++++++++++++++++++++++++
 fs/ocfs2/ocfs2_ioctl.h |   23 ++++
 2 files changed, 306 insertions(+), 0 deletions(-)

diff --git a/fs/ocfs2/ioctl.c b/fs/ocfs2/ioctl.c
index 18812be..9c5ee27 100644
--- a/fs/ocfs2/ioctl.c
+++ b/fs/ocfs2/ioctl.c
@@ -26,6 +26,7 @@
 #include "sysfile.h"
 #include "dir.h"
 #include "buffer_head_io.h"
+#include "suballoc.h"
 
 #include <linux/ext2_fs.h>
 
@@ -439,6 +440,284 @@ bail:
 	return status;
 }
 
+static void o2ffg_update_histogram(struct ocfs2_info_free_chunk_list *hist,
+				   unsigned int chunksize)
+{
+	int index;
+
+	index = __ilog2_u32(chunksize);
+	if (index >= OCFS2_INFO_MAX_HIST)
+		index = OCFS2_INFO_MAX_HIST - 1;
+
+	hist->fc_chunks[index]++;
+	hist->fc_clusters[index] += chunksize;
+}
+
+static void o2ffg_update_stats(struct ocfs2_info_freefrag_stats *stats,
+			       unsigned int chunksize)
+{
+	if (chunksize > stats->ffs_max)
+		stats->ffs_max = chunksize;
+
+	if (chunksize < stats->ffs_min)
+		stats->ffs_min = chunksize;
+
+	stats->ffs_avg += chunksize;
+	stats->ffs_free_chunks_real++;
+}
+
+void ocfs2_info_update_ffg(struct ocfs2_info_freefrag *ffg,
+			   unsigned int chunksize)
+{
+	o2ffg_update_histogram(&(ffg->iff_ffs.ffs_fc_hist), chunksize);
+	o2ffg_update_stats(&(ffg->iff_ffs), chunksize);
+}
+
+int ocfs2_info_freefrag_scan_chain(struct ocfs2_super *osb,
+				   struct inode *gb_inode,
+				   struct ocfs2_dinode *gb_dinode,
+				   struct ocfs2_chain_rec *rec,
+				   struct ocfs2_info_freefrag *ffg,
+				   u32 chunks_in_group)
+{
+	int status = 0, used;
+	u64 blkno;
+
+	struct buffer_head *bh = NULL;
+	struct ocfs2_group_desc *bg = NULL;
+
+	unsigned int max_bits, num_clusters;
+	unsigned int offset = 0, cluster, chunk;
+	unsigned int chunk_free, last_chunksize = 0;
+
+	if (!le32_to_cpu(rec->c_free))
+		goto bail;
+
+	do {
+		if (!bg)
+			blkno = le64_to_cpu(rec->c_blkno);
+		else
+			blkno = le64_to_cpu(bg->bg_next_group);
+
+		if (bh) {
+			brelse(bh);
+			bh = NULL;
+		}
+
+		if (o2info_coherent(*ffg))
+			status = ocfs2_read_group_descriptor(gb_inode,
+							     gb_dinode,
+							     blkno, &bh);
+		else
+			status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
+
+		if (status < 0) {
+			mlog(ML_ERROR, "Can't read the group descriptor # "
+			     "%llu from device.", (unsigned long long)blkno);
+			status = -EIO;
+			goto bail;
+		}
+
+		bg = (struct ocfs2_group_desc *)bh->b_data;
+
+		if (!le16_to_cpu(bg->bg_free_bits_count))
+			continue;
+
+		max_bits = le16_to_cpu(bg->bg_bits);
+		offset = 0;
+
+		for (chunk = 0; chunk < chunks_in_group; chunk++) {
+			/*
+			 * last chunk may be not an entire one.
+			 */
+			if ((offset + ffg->iff_chunksize) > max_bits)
+				num_clusters = max_bits - offset;
+			else
+				num_clusters = ffg->iff_chunksize;
+
+			chunk_free = 0;
+			for (cluster = 0; cluster < num_clusters; cluster++) {
+				used = ocfs2_test_bit(offset,
+						(unsigned long *)bg->bg_bitmap);
+				/*
+				 * - chunk_free counts free clusters in #N chunk.
+				 * - last_chunksize records the size(in) clusters
+				 *   for the last real free chunk being counted.
+				 */
+				if (!used) {
+					last_chunksize++;
+					chunk_free++;
+				}
+
+				if (used && last_chunksize) {
+					ocfs2_info_update_ffg(ffg,
+							      last_chunksize);
+					last_chunksize = 0;
+				}
+
+				offset++;
+			}
+
+			if (chunk_free == ffg->iff_chunksize)
+				ffg->iff_ffs.ffs_free_chunks++;
+		}
+
+		/*
+		 * need to update the info for last free chunk.
+		 */
+		if (last_chunksize)
+			ocfs2_info_update_ffg(ffg, last_chunksize);
+
+	} while (le64_to_cpu(bg->bg_next_group));
+
+bail:
+	brelse(bh);
+
+	mlog_exit(status);
+	return status;
+}
+
+int ocfs2_info_freefrag_scan_bitmap(struct ocfs2_super *osb,
+				    struct inode *gb_inode, u64 blkno,
+				    struct ocfs2_info_freefrag *ffg)
+{
+	u32 chunks_in_group;
+	int status = 0, unlock = 0, i;
+
+	struct buffer_head *bh = NULL;
+	struct ocfs2_chain_list *cl = NULL;
+	struct ocfs2_chain_rec *rec = NULL;
+	struct ocfs2_dinode *gb_dinode = NULL;
+
+	if (gb_inode)
+		mutex_lock(&gb_inode->i_mutex);
+
+	if (o2info_coherent(*ffg)) {
+		status = ocfs2_inode_lock(gb_inode, &bh, 0);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+		unlock = 1;
+	} else {
+		status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+	}
+
+	gb_dinode = (struct ocfs2_dinode *)bh->b_data;
+	cl = &(gb_dinode->id2.i_chain);
+
+	/*
+	 * Chunksize(in) clusters from userspace should be
+	 * less than clusters in a group.
+	 */
+	if (ffg->iff_chunksize > le16_to_cpu(cl->cl_cpg)) {
+		status = -EINVAL;
+		goto bail;
+	}
+
+	memset(&ffg->iff_ffs, 0, sizeof(struct ocfs2_info_freefrag_stats));
+
+	ffg->iff_ffs.ffs_min = ~0U;
+	ffg->iff_ffs.ffs_clusters =
+			le32_to_cpu(gb_dinode->id1.bitmap1.i_total);
+	ffg->iff_ffs.ffs_free_clusters = ffg->iff_ffs.ffs_clusters -
+			le32_to_cpu(gb_dinode->id1.bitmap1.i_used);
+
+	chunks_in_group = le16_to_cpu(cl->cl_cpg) / ffg->iff_chunksize + 1;
+
+	for (i = 0; i < le16_to_cpu(cl->cl_next_free_rec); i++) {
+		rec = &(cl->cl_recs[i]);
+		status = ocfs2_info_freefrag_scan_chain(osb, gb_inode,
+							gb_dinode,
+							rec, ffg,
+							chunks_in_group);
+		if (status)
+			goto bail;
+	}
+
+	if (ffg->iff_ffs.ffs_free_chunks_real)
+		ffg->iff_ffs.ffs_avg = (ffg->iff_ffs.ffs_avg /
+					ffg->iff_ffs.ffs_free_chunks_real);
+bail:
+	if (unlock)
+		ocfs2_inode_unlock(gb_inode, 0);
+
+	if (gb_inode)
+		mutex_unlock(&gb_inode->i_mutex);
+
+	if (gb_inode)
+		iput(gb_inode);
+
+	brelse(bh);
+
+	mlog_exit(status);
+	return status;
+}
+
+int ocfs2_info_handle_freefrag(struct inode *inode,
+			       struct ocfs2_info_request __user *req)
+{
+	u64 blkno = -1;
+	char namebuf[40];
+	int status = -EFAULT, type = GLOBAL_BITMAP_SYSTEM_INODE;
+
+	struct ocfs2_info_freefrag oiff;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct inode *gb_inode = NULL;
+
+	if (o2info_from_user(oiff, req))
+		goto bail;
+	/*
+	 * chunksize from userspace should be power of 2.
+	 */
+	if ((oiff.iff_chunksize & (oiff.iff_chunksize - 1)) ||
+	    (!oiff.iff_chunksize)) {
+		status = -EINVAL;
+		goto bail;
+	}
+
+	if (o2info_coherent(oiff)) {
+		gb_inode = ocfs2_get_system_file_inode(osb, type,
+						       OCFS2_INVALID_SLOT);
+		if (!gb_inode) {
+			mlog(ML_ERROR, "unable to get global_bitmap inode\n");
+			status = -EIO;
+			goto bail;
+		}
+	} else {
+		ocfs2_sprintf_system_inode_name(namebuf, sizeof(namebuf), type,
+						OCFS2_INVALID_SLOT);
+		status = ocfs2_lookup_ino_from_name(osb->sys_root_inode,
+						    namebuf,
+						    strlen(namebuf),
+						    &blkno);
+		if (status < 0) {
+			status = -ENOENT;
+			goto bail;
+		}
+	}
+
+	status = ocfs2_info_freefrag_scan_bitmap(osb, gb_inode, blkno, &oiff);
+	if (status < 0)
+		goto bail;
+
+	o2info_set_request_filled(oiff);
+
+	if (o2info_to_user(oiff, req))
+		goto bail;
+
+	status = 0;
+bail:
+	if (status)
+		o2info_set_request_error(oiff, req);
+
+	return status;
+}
+
 int ocfs2_info_handle_unknown(struct inode *inode,
 			      struct ocfs2_info_request __user *req)
 {
@@ -514,6 +793,10 @@ int ocfs2_info_handle_request(struct inode *inode,
 		if (oir.ir_size == sizeof(struct ocfs2_info_freeinode))
 			status = ocfs2_info_handle_freeinode(inode, req);
 		break;
+	case OCFS2_INFO_FREEFRAG:
+		if (oir.ir_size == sizeof(struct ocfs2_info_freefrag))
+			status = ocfs2_info_handle_freefrag(inode, req);
+		break;
 	default:
 		status = ocfs2_info_handle_unknown(inode, req);
 		break;
diff --git a/fs/ocfs2/ocfs2_ioctl.h b/fs/ocfs2/ocfs2_ioctl.h
index 6b4b39a..18b6770 100644
--- a/fs/ocfs2/ocfs2_ioctl.h
+++ b/fs/ocfs2/ocfs2_ioctl.h
@@ -152,6 +152,28 @@ struct ocfs2_info_freeinode {
 	__u32 ifi_pad;
 };
 
+#define OCFS2_INFO_MAX_HIST     (32)
+
+struct ocfs2_info_freefrag {
+	struct ocfs2_info_request iff_req;
+	struct ocfs2_info_freefrag_stats { /* (out) */
+		struct ocfs2_info_free_chunk_list {
+			__u32 fc_chunks[OCFS2_INFO_MAX_HIST];
+			__u32 fc_clusters[OCFS2_INFO_MAX_HIST];
+		} ffs_fc_hist;
+		__u32 ffs_clusters;
+		__u32 ffs_free_clusters;
+		__u32 ffs_free_chunks;
+		__u32 ffs_free_chunks_real;
+		__u32 ffs_min; /* Minimum free chunksize in clusters */
+		__u32 ffs_max;
+		__u32 ffs_avg;
+		__u32 ffs_pad;
+	} iff_ffs;
+	__u32 iff_chunksize; /* chunksize in clusters(in) */
+	__u32 iff_pad;
+};
+
 /* Codes for ocfs2_info_request */
 enum ocfs2_info_type {
 	OCFS2_INFO_CLUSTERSIZE = 1,
@@ -162,6 +184,7 @@ enum ocfs2_info_type {
 	OCFS2_INFO_FS_FEATURES,
 	OCFS2_INFO_JOURNAL_SIZE,
 	OCFS2_INFO_FREEINODE,
+	OCFS2_INFO_FREEFRAG,
 	OCFS2_INFO_NUM_TYPES
 };
 
-- 
1.5.5

[Ocfs2-devel] [PATCH 3/3] Ocfs2: Add a new code 'OCFS2_INFO_FREEFRAG' for o2info ioctl.

[Joel Becker]

On Tue, Dec 07, 2010 at 09:45:38AM +0800, Tristan Ye wrote:
> >>+	gb_dinode = (struct ocfs2_dinode *)bh->b_data;
> >>+	cl = &(gb_dinode->id2.i_chain);
> >
> >	This is safe because we never remove a chain entry from an
> >inode.  However, if we ever shrink disks, we'll need to coordinate with
> >this code.  Perhaps we should comment that.
> 
>    Alright, and the recoordination only would be needed for
> none-coherency case, right?
> 
> Fully-coherent case will be holding the inode lock, and never allow
> disk to be shrinked during the inquiry.

	You are correct.

>    Oh, Wait, an arbitrary 'fdisk' can shrink the disk regardless of
> cluster-coherency.

	We're not worried about that.  That's a mistake made by the
sysadmin, and there is no way to prevent it.  I'm only talking about a
tunefs operation that validly shrinks the ocfs2 filesystem.  We don't do
it today, but if we ever do this dirty read has to know about it.  Not
for accuracy, just not to crash.

Joel

-- 

"What no boss of a programmer can ever understand is that a programmer
 is working when he's staring out of the window"
	- With apologies to Burton Rascoe

Joel Becker
Senior Development Manager
Oracle
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127

[Ocfs2-devel] [PATCH 3/3] Ocfs2: Add a new code 'OCFS2_INFO_FREEFRAG' for o2info ioctl.

[Tristan Ye]

Joel Becker wrote:
> On Tue, Nov 16, 2010 at 06:13:42PM +0800, Tristan Ye wrote:
>> +static void o2ffg_update_histogram(struct ocfs2_info_free_chunk_list *hist,
>> +				   unsigned int chunksize)
>> +{
>> +	int index;
>> +
>> +	index = __ilog2_u32(chunksize);
>> +	if (index >= OCFS2_INFO_MAX_HIST)
>> +		index = OCFS2_INFO_MAX_HIST - 1;
>> +
>> +	hist->fc_chunks[index]++;
>> +	hist->fc_clusters[index] += chunksize;
>> +}
>> +
>> +static void o2ffg_update_stats(struct ocfs2_info_freefrag_stats *stats,
>> +			       unsigned int chunksize)
>> +{
>> +	if (chunksize > stats->ffs_max)
>> +		stats->ffs_max = chunksize;
>> +
>> +	if (chunksize < stats->ffs_min)
>> +		stats->ffs_min = chunksize;
>> +
>> +	stats->ffs_avg += chunksize;
>> +	stats->ffs_free_chunks_real++;
>> +}
>
> 	MUCH more readable.
>
>> +int ocfs2_info_freefrag_scan_bitmap(struct ocfs2_super *osb,
>> +				    struct inode *gb_inode, u64 blkno,
>> +				    struct ocfs2_info_freefrag *ffg)
>> +{
>> +	u32 chunks_in_group;
>> +	int status = 0, unlock = 0, i;
>> +
>> +	struct buffer_head *bh = NULL;
>> +	struct ocfs2_chain_list *cl = NULL;
>> +	struct ocfs2_chain_rec *rec = NULL;
>> +	struct ocfs2_dinode *gb_dinode = NULL;
>> +
>> +	if (gb_inode)
>> +		mutex_lock(&gb_inode->i_mutex);
>> +
>> +	if (o2info_coherent(*ffg)) {
>> +		status = ocfs2_inode_lock(gb_inode, &bh, 0);
>> +		if (status < 0) {
>> +			mlog_errno(status);
>> +			goto bail;
>> +		}
>> +		unlock = 1;
>> +	} else {
>> +		status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
>> +		if (status < 0) {
>> +			mlog_errno(status);
>> +			goto bail;
>> +		}
>> +	}
>> +
>> +	gb_dinode = (struct ocfs2_dinode *)bh->b_data;
>> +	cl = &(gb_dinode->id2.i_chain);
>
> 	This is safe because we never remove a chain entry from an
> inode.  However, if we ever shrink disks, we'll need to coordinate with
> this code.  Perhaps we should comment that.

    Alright, and the recoordination only would be needed for 
none-coherency case, right?

Fully-coherent case will be holding the inode lock, and never allow disk 
to be shrinked during the inquiry.

    Oh, Wait, an arbitrary 'fdisk' can shrink the disk regardless of 
cluster-coherency.

>
> Joel
>

[Ocfs2-devel] [PATCH 3/3] Ocfs2: Add a new code 'OCFS2_INFO_FREEFRAG' for o2info ioctl.

[Joel Becker]

On Tue, Nov 16, 2010 at 06:13:42PM +0800, Tristan Ye wrote:
> +static void o2ffg_update_histogram(struct ocfs2_info_free_chunk_list *hist,
> +				   unsigned int chunksize)
> +{
> +	int index;
> +
> +	index = __ilog2_u32(chunksize);
> +	if (index >= OCFS2_INFO_MAX_HIST)
> +		index = OCFS2_INFO_MAX_HIST - 1;
> +
> +	hist->fc_chunks[index]++;
> +	hist->fc_clusters[index] += chunksize;
> +}
> +
> +static void o2ffg_update_stats(struct ocfs2_info_freefrag_stats *stats,
> +			       unsigned int chunksize)
> +{
> +	if (chunksize > stats->ffs_max)
> +		stats->ffs_max = chunksize;
> +
> +	if (chunksize < stats->ffs_min)
> +		stats->ffs_min = chunksize;
> +
> +	stats->ffs_avg += chunksize;
> +	stats->ffs_free_chunks_real++;
> +}

	MUCH more readable.

> +int ocfs2_info_freefrag_scan_bitmap(struct ocfs2_super *osb,
> +				    struct inode *gb_inode, u64 blkno,
> +				    struct ocfs2_info_freefrag *ffg)
> +{
> +	u32 chunks_in_group;
> +	int status = 0, unlock = 0, i;
> +
> +	struct buffer_head *bh = NULL;
> +	struct ocfs2_chain_list *cl = NULL;
> +	struct ocfs2_chain_rec *rec = NULL;
> +	struct ocfs2_dinode *gb_dinode = NULL;
> +
> +	if (gb_inode)
> +		mutex_lock(&gb_inode->i_mutex);
> +
> +	if (o2info_coherent(*ffg)) {
> +		status = ocfs2_inode_lock(gb_inode, &bh, 0);
> +		if (status < 0) {
> +			mlog_errno(status);
> +			goto bail;
> +		}
> +		unlock = 1;
> +	} else {
> +		status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
> +		if (status < 0) {
> +			mlog_errno(status);
> +			goto bail;
> +		}
> +	}
> +
> +	gb_dinode = (struct ocfs2_dinode *)bh->b_data;
> +	cl = &(gb_dinode->id2.i_chain);

	This is safe because we never remove a chain entry from an
inode.  However, if we ever shrink disks, we'll need to coordinate with
this code.  Perhaps we should comment that.

Joel

-- 

"Also, all of life's big problems include the words 'indictment' or
	'inoperable.' Everything else is small stuff."
	- Alton Brown

Joel Becker
Senior Development Manager
Oracle
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127

[Ocfs2-devel] [PATCH 3/3] Ocfs2: Add a new code 'OCFS2_INFO_FREEFRAG' for o2info ioctl.

[Tristan Ye]

This new code is a bit more complicated than former ones, the goal is to
show user all statistics required to take a deep insight into filesystem
on how the disk is being fragmentaed.

The goal is achieved by scaning global bitmap from (cluster)group to group
to figure out following factors in the filesystem:

        - How many free chunks in a fixed size as user requested.
        - How many real free chunks in all size.
        - Min/Max/Avg size(in) clusters of free chunks.
        - How do free chunks distribute(in size) in terms of a histogram,
          just like following:
          ---------------------------------------------------------
          Extent Size Range :  Free extents  Free Clusters  Percent
             32K...   64K-  :             1             1    0.00%
              1M...    2M-  :             9           288    0.03%
              8M...   16M-  :             2           831    0.09%
             32M...   64M-  :             1          2047    0.23%
            128M...  256M-  :             1          8191    0.92%
            256M...  512M-  :             2         21706    2.43%
            512M... 1024M-  :            27        858623   96.29%
          ---------------------------------------------------------

Userspace ioctl() call eventually gets the above info returned by passing
a 'struct ocfs2_info_freefrag' with the chunk_size being specified first.

Signed-off-by: Tristan Ye <tristan.ye@oracle.com>
---
 fs/ocfs2/ioctl.c       |  283 ++++++++++++++++++++++++++++++++++++++++++++++++
 fs/ocfs2/ocfs2_ioctl.h |   23 ++++
 2 files changed, 306 insertions(+), 0 deletions(-)

diff --git a/fs/ocfs2/ioctl.c b/fs/ocfs2/ioctl.c
index 18812be..c796860 100644
--- a/fs/ocfs2/ioctl.c
+++ b/fs/ocfs2/ioctl.c
@@ -26,6 +26,7 @@
 #include "sysfile.h"
 #include "dir.h"
 #include "buffer_head_io.h"
+#include "suballoc.h"
 
 #include <linux/ext2_fs.h>
 
@@ -439,6 +440,284 @@ bail:
 	return status;
 }
 
+static void o2ffg_update_histogram(struct ocfs2_info_free_chunk_list *hist,
+				   unsigned int chunksize)
+{
+	int index;
+
+	index = __ilog2_u32(chunksize);
+	if (index >= OCFS2_INFO_MAX_HIST)
+		index = OCFS2_INFO_MAX_HIST - 1;
+
+	hist->fc_chunks[index]++;
+	hist->fc_clusters[index] += chunksize;
+}
+
+static void o2ffg_update_stats(struct ocfs2_info_freefrag_stats *stats,
+			       unsigned int chunksize)
+{
+	if (chunksize > stats->ffs_max)
+		stats->ffs_max = chunksize;
+
+	if (chunksize < stats->ffs_min)
+		stats->ffs_min = chunksize;
+
+	stats->ffs_avg += chunksize;
+	stats->ffs_free_chunks_real++;
+}
+
+void ocfs2_info_update_ffg(struct ocfs2_info_freefrag *ffg,
+			   unsigned int chunksize)
+{
+	o2ffg_update_histogram(&(ffg->iff_ffs.ffs_fc_hist), chunksize);
+	o2ffg_update_stats(&(ffg->iff_ffs), chunksize);
+}
+
+int ocfs2_info_freefrag_scan_chain(struct ocfs2_super *osb,
+				   struct inode *gb_inode,
+				   struct ocfs2_dinode *gb_dinode,
+				   struct ocfs2_chain_rec *rec,
+				   struct ocfs2_info_freefrag *ffg,
+				   u32 chunks_in_group)
+{
+	int status = 0, used;
+	u64 blkno;
+
+	struct buffer_head *bh = NULL;
+	struct ocfs2_group_desc *bg = NULL;
+
+	unsigned int max_bits, num_clusters;
+	unsigned int offset = 0, cluster, chunk;
+	unsigned int chunk_free, last_chunksize = 0;
+
+	if (!le32_to_cpu(rec->c_free))
+		goto bail;
+
+	do {
+		if (!bg)
+			blkno = le64_to_cpu(rec->c_blkno);
+		else
+			blkno = le64_to_cpu(bg->bg_next_group);
+
+		if (bh) {
+			brelse(bh);
+			bh = NULL;
+		}
+
+		if (o2info_coherent(*ffg))
+			status = ocfs2_read_group_descriptor(gb_inode,
+							     gb_dinode,
+							     blkno, &bh);
+		else
+			status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
+
+		if (status < 0) {
+			mlog(ML_ERROR, "Can't read the group descriptor # "
+			     "%llu from device.", (unsigned long long)blkno);
+			status = -EIO;
+			goto bail;
+		}
+
+		bg = (struct ocfs2_group_desc *)bh->b_data;
+
+		if (!le16_to_cpu(bg->bg_free_bits_count))
+			continue;
+
+		max_bits = le16_to_cpu(bg->bg_bits);
+		offset = 0;
+
+		for (chunk = 0; chunk < chunks_in_group; chunk++) {
+			/*
+			 * last chunk may be not an entire one.
+			 */
+			if ((offset + ffg->iff_chunksize) > max_bits)
+				num_clusters = max_bits - offset;
+			else
+				num_clusters = ffg->iff_chunksize;
+
+			chunk_free = 0;
+			for (cluster = 0; cluster < num_clusters; cluster++) {
+				used = ocfs2_test_bit(offset,
+						(unsigned long *)bg->bg_bitmap);
+				/*
+				 * - chunk_free counts free clusters in #N chunk.
+				 * - last_chunksize records the size(in) clusters
+				 *   for the last real free chunk being counted.
+				 */
+				if (!used) {
+					last_chunksize++;
+					chunk_free++;
+				}
+
+				if (used && last_chunksize) {
+					ocfs2_info_update_ffg(ffg,
+							      last_chunksize);
+					last_chunksize = 0;
+				}
+
+				offset++;
+			}
+
+			if (chunk_free == ffg->iff_chunksize)
+				ffg->iff_ffs.ffs_free_chunks++;
+		}
+
+		/*
+		 * need to update the info for last free chunk.
+		 */
+		if (last_chunksize)
+			ocfs2_info_update_ffg(ffg, last_chunksize);
+
+	} while (le64_to_cpu(bg->bg_next_group));
+
+bail:
+	brelse(bh);
+
+	mlog_exit(status);
+	return status;
+}
+
+int ocfs2_info_freefrag_scan_bitmap(struct ocfs2_super *osb,
+				    struct inode *gb_inode, u64 blkno,
+				    struct ocfs2_info_freefrag *ffg)
+{
+	u32 chunks_in_group;
+	int status = 0, unlock = 0, i;
+
+	struct buffer_head *bh = NULL;
+	struct ocfs2_chain_list *cl = NULL;
+	struct ocfs2_chain_rec *rec = NULL;
+	struct ocfs2_dinode *gb_dinode = NULL;
+
+	if (gb_inode)
+		mutex_lock(&gb_inode->i_mutex);
+
+	if (o2info_coherent(*ffg)) {
+		status = ocfs2_inode_lock(gb_inode, &bh, 0);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+		unlock = 1;
+	} else {
+		status = ocfs2_read_blocks_sync(osb, blkno, 1, &bh);
+		if (status < 0) {
+			mlog_errno(status);
+			goto bail;
+		}
+	}
+
+	gb_dinode = (struct ocfs2_dinode *)bh->b_data;
+	cl = &(gb_dinode->id2.i_chain);
+
+	/* 
+	 * Chunksize(in) clusters from userspace should be
+	 * less than clusters in a group.
+	 */
+	if (ffg->iff_chunksize > le16_to_cpu(cl->cl_cpg)) {
+		status = -EINVAL;
+		goto bail;
+	}
+
+	memset(&ffg->iff_ffs, 0, sizeof(struct ocfs2_info_freefrag_stats));
+
+	ffg->iff_ffs.ffs_min = ~0U;
+	ffg->iff_ffs.ffs_clusters =
+			le32_to_cpu(gb_dinode->id1.bitmap1.i_total);
+	ffg->iff_ffs.ffs_free_clusters = ffg->iff_ffs.ffs_clusters -
+			le32_to_cpu(gb_dinode->id1.bitmap1.i_used);
+
+	chunks_in_group = le16_to_cpu(cl->cl_cpg) / ffg->iff_chunksize + 1;
+
+	for (i = 0; i < le16_to_cpu(cl->cl_next_free_rec); i++) {
+		rec = &(cl->cl_recs[i]);
+		status = ocfs2_info_freefrag_scan_chain(osb, gb_inode,
+							gb_dinode,
+							rec, ffg,
+							chunks_in_group);
+		if (status)
+			goto bail;
+	}
+
+	if (ffg->iff_ffs.ffs_free_chunks_real)
+		ffg->iff_ffs.ffs_avg = (ffg->iff_ffs.ffs_avg /
+					ffg->iff_ffs.ffs_free_chunks_real);
+bail:
+	if (unlock)
+		ocfs2_inode_unlock(gb_inode, 0);
+
+	if (gb_inode)
+		mutex_unlock(&gb_inode->i_mutex);
+
+	if (gb_inode)
+		iput(gb_inode);
+
+	brelse(bh);
+
+	mlog_exit(status);
+	return status;
+}
+
+int ocfs2_info_handle_freefrag(struct inode *inode,
+			       struct ocfs2_info_request __user *req)
+{
+	u64 blkno = -1;
+	char namebuf[40];
+	int status = -EFAULT, type = GLOBAL_BITMAP_SYSTEM_INODE;
+
+	struct ocfs2_info_freefrag oiff;
+	struct ocfs2_super *osb = OCFS2_SB(inode->i_sb);
+	struct inode *gb_inode = NULL;
+
+	if (o2info_from_user(oiff, req))
+		goto bail;
+	/*
+	 * chunksize from userspace should be power of 2.
+	 */
+	if ((oiff.iff_chunksize & (oiff.iff_chunksize - 1)) ||
+	    (!oiff.iff_chunksize)) {
+		status = -EINVAL;
+		goto bail;
+	}
+
+	if (o2info_coherent(oiff)) {
+		gb_inode = ocfs2_get_system_file_inode(osb, type,
+						       OCFS2_INVALID_SLOT);
+		if (!gb_inode) {
+			mlog(ML_ERROR, "unable to get global_bitmap inode\n");
+			status = -EIO;
+			goto bail;
+		}
+	} else {
+		ocfs2_sprintf_system_inode_name(namebuf, sizeof(namebuf), type,
+						OCFS2_INVALID_SLOT);
+		status = ocfs2_lookup_ino_from_name(osb->sys_root_inode,
+						    namebuf,
+						    strlen(namebuf),
+						    &blkno);
+		if (status < 0) {
+			status = -ENOENT;
+			goto bail;
+		}
+	}
+
+	status = ocfs2_info_freefrag_scan_bitmap(osb, gb_inode, blkno, &oiff);
+	if (status < 0)
+		goto bail;
+
+	o2info_set_request_filled(oiff);
+
+	if (o2info_to_user(oiff, req))
+		goto bail;
+
+	status = 0;
+bail:
+	if (status)
+		o2info_set_request_error(oiff, req);
+
+	return status;
+}
+
 int ocfs2_info_handle_unknown(struct inode *inode,
 			      struct ocfs2_info_request __user *req)
 {
@@ -514,6 +793,10 @@ int ocfs2_info_handle_request(struct inode *inode,
 		if (oir.ir_size == sizeof(struct ocfs2_info_freeinode))
 			status = ocfs2_info_handle_freeinode(inode, req);
 		break;
+	case OCFS2_INFO_FREEFRAG:
+		if (oir.ir_size == sizeof(struct ocfs2_info_freefrag))
+			status = ocfs2_info_handle_freefrag(inode, req);
+		break;
 	default:
 		status = ocfs2_info_handle_unknown(inode, req);
 		break;
diff --git a/fs/ocfs2/ocfs2_ioctl.h b/fs/ocfs2/ocfs2_ioctl.h
index 6b4b39a..18b6770 100644
--- a/fs/ocfs2/ocfs2_ioctl.h
+++ b/fs/ocfs2/ocfs2_ioctl.h
@@ -152,6 +152,28 @@ struct ocfs2_info_freeinode {
 	__u32 ifi_pad;
 };
 
+#define OCFS2_INFO_MAX_HIST     (32)
+
+struct ocfs2_info_freefrag {
+	struct ocfs2_info_request iff_req;
+	struct ocfs2_info_freefrag_stats { /* (out) */
+		struct ocfs2_info_free_chunk_list {
+			__u32 fc_chunks[OCFS2_INFO_MAX_HIST];
+			__u32 fc_clusters[OCFS2_INFO_MAX_HIST];
+		} ffs_fc_hist;
+		__u32 ffs_clusters;
+		__u32 ffs_free_clusters;
+		__u32 ffs_free_chunks;
+		__u32 ffs_free_chunks_real;
+		__u32 ffs_min; /* Minimum free chunksize in clusters */
+		__u32 ffs_max;
+		__u32 ffs_avg;
+		__u32 ffs_pad;
+	} iff_ffs;
+	__u32 iff_chunksize; /* chunksize in clusters(in) */
+	__u32 iff_pad;
+};
+
 /* Codes for ocfs2_info_request */
 enum ocfs2_info_type {
 	OCFS2_INFO_CLUSTERSIZE = 1,
@@ -162,6 +184,7 @@ enum ocfs2_info_type {
 	OCFS2_INFO_FS_FEATURES,
 	OCFS2_INFO_JOURNAL_SIZE,
 	OCFS2_INFO_FREEINODE,
+	OCFS2_INFO_FREEFRAG,
 	OCFS2_INFO_NUM_TYPES
 };
 
-- 
1.5.5