Skip to contentSkip to conversation

exofs patches for after 2.6.32

PATCH New Boaz Harrosh 1 participant 7 messages Nov 8, 2009

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 exofs patches for after 2.6.32

Assorted bunch of patches to the exofs file system. Most of these are a resend, they did not make it to 2.6.32 because the dependent bits from drivers/scsi/osd did not make it into 2.6.32. I hope they finally will for 2.6.33.

These patches are just preliminary cleanups to some new fixtures that will be sent next.

Thank you for reviewing

list of patches:
[PATCH 1/6] trivial: some small fixes in exofs documentation [PATCH 2/6] exofs: More sane debug print
[PATCH 3/6] exofs: dbg-print less
[PATCH 4/6] exofs: refactor exofs_i_info initialization into common helper [PATCH 5/6] exofs: Prints on mount and unmout
[PATCH 6/6] exofs: statfs blocks is sectors not FS blocks

Boaz

#1 lore raw

PATCH 1/6 trivial: some small fixes in exofs documentation

From: Thadeu Lima de Souza Cascardo <[email protected]>

Add exofs.txt to filesystems Documentation index and fix some typos, identation and grammar.

---
 Documentation/filesystems/00-INDEX  |    2 ++
 Documentation/filesystems/exofs.txt |   23 ++++++++++++-----------
 2 files changed, 14 insertions(+), 11 deletions(-)
Documentation/filesystems/00-INDEX +2 -0
index f15621e..7001782 100644
@@ -36,6 +36,8 @@ dnotify.txt
 	- info about directory notification in Linux.
 ecryptfs.txt
 	- docs on eCryptfs: stacked cryptographic filesystem for Linux.
+exofs.txt
+	- info, usage, mount options, design about EXOFS.
 ext2.txt
 	- info, mount options and specifications for the Ext2 filesystem.
 ext3.txt
Documentation/filesystems/exofs.txt +12 -11
index 0ced74c..abd2a9b 100644
@@ -60,13 +60,13 @@ USAGE
 
    mkfs.exofs --pid=65536 --format /dev/osd0
 
-   The --format is optional if not specified no OSD_FORMAT will be
-   preformed and a clean file system will be created in the specified pid,
+   The --format is optional. If not specified, no OSD_FORMAT will be
+   performed and a clean file system will be created in the specified pid,
    in the available space of the target. (Use --format=size_in_meg to limit
    the total LUN space available)
 
-   If pid already exist it will be deleted and a new one will be created in it's
-   place. Be careful.
+   If pid already exists, it will be deleted and a new one will be created in
+   its place. Be careful.
 
    An exofs lives inside a single OSD partition. You can create multiple exofs
    filesystems on the same device using multiple pids.
@@ -81,7 +81,7 @@ USAGE
 
 7. For reference (See do-exofs example script):
 	do-exofs start - an example of how to perform the above steps.
-	do-exofs stop -  an example of how to unmount the file system.
+	do-exofs stop - an example of how to unmount the file system.
 	do-exofs format - an example of how to format and mkfs a new exofs.
 
 8. Extra compilation flags (uncomment in fs/exofs/Kbuild):
@@ -104,8 +104,8 @@ Where:
     exofs specific options: Options are separated by commas (,)
 		pid=<integer> - The partition number to mount/create as
                                 container of the filesystem.
-                                This option is mandatory
-                to=<integer>  - Timeout in ticks for a single command
+                                This option is mandatory.
+                to=<integer>  - Timeout in ticks for a single command.
                                 default is (60 * HZ) [for debugging only]
 
 ===============================================================================
@@ -116,7 +116,7 @@ DESIGN
   with a special ID (defined in common.h).
   Information included in the file system control block is used to fill the
   in-memory superblock structure at mount time. This object is created before
-  the file system is used by mkexofs.c It contains information such as:
+  the file system is used by mkexofs.c. It contains information such as:
 	- The file system's magic number
 	- The next inode number to be allocated
 
@@ -134,8 +134,8 @@ DESIGN
   attributes. This applies to both regular files and other types (directories,
   device files, symlinks, etc.).
 
-* Credentials are generated per object (inode and superblock) when they is
-  created in memory (read off disk or created). The credential works for all
+* Credentials are generated per object (inode and superblock) when they are
+  created in memory (read from disk or created). The credential works for all
   operations and is used as long as the object remains in memory.
 
 * Async OSD operations are used whenever possible, but the target may execute
@@ -145,7 +145,8 @@ DESIGN
   from executing in reverse order:
 	- The following are handled with the OBJ_CREATED and OBJ_2BCREATED
 	  flags. OBJ_CREATED is set when we know the object exists on the OSD -
-	  in create's callback function, and when we successfully do a read_inode.
+	  in create's callback function, and when we successfully do a
+	  read_inode.
 	  OBJ_2BCREATED is set in the beginning of the create function, so we
 	  know that we should wait.
 		- create/delete: delete should wait until the object is created
Signature · 1 line
1.6.5.1
#2 lore raw

PATCH 2/6 exofs: More sane debug print

In reply to Boaz Harrosh

debug prints should be somewhat useful without actually reading the source code

---
 fs/exofs/inode.c |    3 +--
 1 files changed, 1 insertions(+), 2 deletions(-)
fs/exofs/inode.c +1 -2
index 6c10f74..4474861 100644
@@ -950,8 +950,7 @@ struct inode *exofs_iget(struct super_block *sb, unsigned long ino)
 #ifdef EXOFS_DEBUG_OBJ_ISIZE
 	if ((inode->i_size != sanity) &&
 		(!exofs_inode_is_fast_symlink(inode))) {
-		EXOFS_ERR("WARNING: Size of object from inode and "
-			  "attributes differ (%lld != %llu)\n",
+		EXOFS_ERR("WARNING: Size of inode=%llu != object=%llu\n",
 			  inode->i_size, _LLU(sanity));
 	}
 #endif
Signature · 1 line
1.6.5.1
#3 lore raw

PATCH 3/6 exofs: dbg-print less

In reply to Boaz Harrosh

Iner-loops printing is converted to EXOFS_DBG2 which is #defined to nothing.

It is now almost bareable to just leave debug-on. Every operation is printed once, with most relevant info (I hope).

---
 fs/exofs/inode.c |   10 ++++++----
 1 files changed, 6 insertions(+), 4 deletions(-)
fs/exofs/inode.c +6 -4
index 4474861..dc90f7c 100644
@@ -41,6 +41,8 @@
 #  define EXOFS_DEBUG_OBJ_ISIZE 1
 #endif
 
+#define EXOFS_DBGMSG2(M...) do{}while(0)
+
 struct page_collect {
 	struct exofs_sb_info *sbi;
 	struct request_queue *req_q;
@@ -198,7 +200,7 @@ static int __readpages_done(struct osd_request *or, struct page_collect *pcol,
 		else
 			page_stat = ret;
 
-		EXOFS_DBGMSG("    readpages_done(0x%lx, 0x%lx) %s\n",
+		EXOFS_DBGMSG2("    readpages_done(0x%lx, 0x%lx) %s\n",
 			  inode->i_ino, page->index,
 			  page_stat ? "bad_bytes" : "good_bytes");
 
@@ -370,7 +372,7 @@ try_again:
 	if (len != PAGE_CACHE_SIZE)
 		zero_user(page, len, PAGE_CACHE_SIZE - len);
 
-	EXOFS_DBGMSG("    readpage_strip(0x%lx, 0x%lx) len=0x%zx\n",
+	EXOFS_DBGMSG2("    readpage_strip(0x%lx, 0x%lx) len=0x%zx\n",
 		     inode->i_ino, page->index, len);
 
 	ret = pcol_add_page(pcol, page, len);
@@ -482,7 +484,7 @@ static void writepages_done(struct osd_request *or, void *p)
 
 		update_write_page(page, page_stat);
 		unlock_page(page);
-		EXOFS_DBGMSG("    writepages_done(0x%lx, 0x%lx) status=%d\n",
+		EXOFS_DBGMSG2("    writepages_done(0x%lx, 0x%lx) status=%d\n",
 			     inode->i_ino, page->index, page_stat);
 
 		length += bvec->bv_len;
@@ -609,7 +611,7 @@ try_again:
 			goto fail;
 	}
 
-	EXOFS_DBGMSG("    writepage_strip(0x%lx, 0x%lx) len=0x%zx\n",
+	EXOFS_DBGMSG2("    writepage_strip(0x%lx, 0x%lx) len=0x%zx\n",
 		     inode->i_ino, page->index, len);
 
 	ret = pcol_add_page(pcol, page, len);
Signature · 1 line
1.6.5.1
#4 lore raw

PATCH 4/6 exofs: refactor exofs_i_info initialization into common helper

In reply to Boaz Harrosh

There are two places that initialize inodes: exofs_iget() and exofs_new_inode()

As more members of exofs_i_info that need initialization are added this code will grow. (soon)

---
 fs/exofs/inode.c |   10 ++++++++--
 1 files changed, 8 insertions(+), 2 deletions(-)
fs/exofs/inode.c +8 -2
index dc90f7c..243bc46 100644
@@ -908,6 +908,12 @@ out:
 	return ret;
 }
 
+
+static void __oi_init(struct exofs_i_info *oi)
+{
+	init_waitqueue_head(&oi->i_wq);
+	oi->i_flags = 0;
+}
 /*
  * Fill in an inode read from the OSD and set it up for use
  */
@@ -925,13 +931,13 @@ struct inode *exofs_iget(struct super_block *sb, unsigned long ino)
 	if (!(inode->i_state & I_NEW))
 		return inode;
 	oi = exofs_i(inode);
+	__oi_init(oi);
 
 	/* read the inode from the osd */
 	ret = exofs_get_inode(sb, oi, &fcb, &sanity);
 	if (ret)
 		goto bad_inode;
 
-	init_waitqueue_head(&oi->i_wq);
 	set_obj_created(oi);
 
 	/* copy stuff from on-disk struct to in-memory struct */
@@ -1062,8 +1068,8 @@ struct inode *exofs_new_inode(struct inode *dir, int mode)
 		return ERR_PTR(-ENOMEM);
 
 	oi = exofs_i(inode);
+	__oi_init(oi);
 
-	init_waitqueue_head(&oi->i_wq);
 	set_obj_2bcreated(oi);
 
 	sbi = sb->s_fs_info;
Signature · 1 line
1.6.5.1
#5 lore raw

PATCH 5/6 exofs: Prints on mount and unmout

In reply to Boaz Harrosh

It is important to print in the logs when a filesystem was mounted and eventually unmounted.

Print the osd-device's osd_name and pid the FS was mounted/unmounted on.

TODO: How to also print the namespace path the filesystem was
mounted on?

---
 fs/exofs/super.c |   11 +++++++++++
 1 files changed, 11 insertions(+), 0 deletions(-)
fs/exofs/super.c +11 -0
index 9f500de..920f016 100644
@@ -257,6 +257,15 @@ static void exofs_write_super(struct super_block *sb)
 		sb->s_dirt = 0;
 }
 
+static void _exofs_print_device(const char *msg, const char *dev_path,
+				struct osd_dev *od, u64 pid)
+{
+	const struct osd_dev_info *odi = osduld_device_info(od);
+
+	printk(KERN_NOTICE "exofs: %s %s osd_name-%s pid-0x%llx\n",
+		msg, dev_path ?: "", odi->osdname, _LLU(pid));
+}
+
 /*
  * This function is called when the vfs is freeing the superblock.  We just
  * need to free our own part.
@@ -279,6 +288,7 @@ static void exofs_put_super(struct super_block *sb)
 				  msecs_to_jiffies(100));
 	}
 
+	_exofs_print_device("Unmounting", NULL, sbi->s_dev, sbi->s_pid);
 	osduld_put_device(sbi->s_dev);
 	kfree(sb->s_fs_info);
 	sb->s_fs_info = NULL;
@@ -395,6 +405,7 @@ static int exofs_fill_super(struct super_block *sb, void *data, int silent)
 		goto free_sbi;
 	}
 
+	_exofs_print_device("Mounting", opts->dev_name, sbi->s_dev, sbi->s_pid);
 	ret = 0;
 out:
 	if (or)
Signature · 1 line
1.6.5.1
#6 lore raw

PATCH 6/6 exofs: statfs blocks is sectors not FS blocks

In reply to Boaz Harrosh

Even though exofs has a 4k block size, statfs blocks is in sectors (512 bytes).

Also if target returns 0 for capacity then make it ULLONG_MAX. df does not like zero-size filesystems

---
 fs/exofs/super.c |   10 ++++++----
 1 files changed, 6 insertions(+), 4 deletions(-)
fs/exofs/super.c +6 -4
index 920f016..28add3e 100644
@@ -473,9 +473,11 @@ static int exofs_statfs(struct dentry *dentry, struct kstatfs *buf)
 		goto out;
 
 	ret = extract_attr_from_req(or, &attrs[0]);
-	if (likely(!ret))
+	if (likely(!ret)) {
 		capacity = get_unaligned_be64(attrs[0].val_ptr);
-	else
+		if (unlikely(!capacity))
+			capacity = ULLONG_MAX;
+	} else
 		EXOFS_DBGMSG("exofs_statfs: get capacity failed.\n");
 
 	ret = extract_attr_from_req(or, &attrs[1]);
@@ -487,8 +489,8 @@ static int exofs_statfs(struct dentry *dentry, struct kstatfs *buf)
 	/* fill in the stats buffer */
 	buf->f_type = EXOFS_SUPER_MAGIC;
 	buf->f_bsize = EXOFS_BLKSIZE;
-	buf->f_blocks = (capacity >> EXOFS_BLKSHIFT);
-	buf->f_bfree = ((capacity - used) >> EXOFS_BLKSHIFT);
+	buf->f_blocks = capacity >> 9;
+	buf->f_bfree = (capacity - used) >> 9;
 	buf->f_bavail = buf->f_bfree;
 	buf->f_files = sbi->s_numfiles;
 	buf->f_ffree = EXOFS_MAX_ID - sbi->s_numfiles;
Signature · 1 line
1.6.5.1