Add basic superblock and inode handling routines for OMFS
    Signed-off-by: Bob Copeland <me@bobcopeland.com>

---

 fs/omfs/inode.c |  439 +++++++++++++++++++++++++++++++++++++++++++++++++++++++

 1 files changed, 439 insertions(+), 0 deletions(-)

 create mode 100644 fs/omfs/inode.c
diff --git a/fs/omfs/inode.c b/fs/omfs/inode.c

new file mode 100644

index 0000000..92d794f

--- /dev/null

+++ b/fs/omfs/inode.c

@@ -0,0 +1,439 @@

+/*

+ * Optimized MPEG FS - inode and super operations.

+ * Copyright (C) 2006 Bob Copeland <me@bobcopeland.com>

+ * Released under GPL v2.

+ */

+#include <linux/version.h>

+#include <linux/module.h>

+#include <linux/sched.h>

+#include <linux/fs.h>

+#include <linux/vfs.h>

+#include <linux/buffer_head.h>

+#include <linux/vmalloc.h>

+#include "omfs.h"

+

+MODULE_AUTHOR("Bob Copeland <me@bobcopeland.com>");

+MODULE_DESCRIPTION("OMFS (ReplayTV/Karma) Filesystem for Linux");

+MODULE_LICENSE("GPL");

+

+struct inode *omfs_new_inode(struct inode *dir, int mode)

+{

+	struct inode *inode;

+	u64 new_block;

+	int res;

+	int len;

+	struct omfs_sb_info *sbi = OMFS_SB(dir->i_sb);

+

+	inode = new_inode(dir->i_sb);

+	if (!inode)

+		return ERR_PTR(-ENOMEM);

+

+	res = omfs_allocate_range(dir->i_sb, sbi->s_mirrors, sbi->s_mirrors,

+			&new_block, &len);

+	if (res)

+		return ERR_PTR(res);

+

+	inode->i_ino = new_block;

+	inode->i_mode = mode;

+	inode->i_uid = current->fsuid;

+	inode->i_gid = current->fsgid;

+	inode->i_blocks = 0;

+	inode->i_mapping->a_ops = &omfs_aops;

+

+	inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;

+	switch (mode & S_IFMT) {

+	case S_IFDIR:

+		inode->i_op = &omfs_dir_inops;

+		inode->i_fop = &omfs_dir_operations;

+		inode->i_size = sbi->s_sys_blocksize;

+		inode->i_nlink++;

+		break;

+	case S_IFREG:

+		inode->i_op = &omfs_file_inops;

+		inode->i_fop ...
[ message continues ]

can be omfs_sb->s_magic != cpu_to_be32(OMFS_MAGIC)
this code leaks omfs_sbi_info (sbi)

--

Yup, I saw that too.  I dropped these and added the labels per Miklos'
This one I'm not seeing.  FS core calls put_super if ret is nonzero so

this should be ok?
Thanks!

--

Bob Copeland %% www.bobcopeland.com 
--

Yep, you are right. Sorry for the noise.

I was mislead by other filesystems code.
Marcin

--

Hmm, looks like error handling needs a makeover if this is really to

become example code.  See comments inline.
Somebody said this is fun?  Please do a proper review of this patchset

then, thank you.  ("Political" eh?  Now I think *that* is really

insulting to Andrew)
This form is preferred:
	ret = -EIO;

	if (!bh)
Ditto.  If err is the same as assigned previously it doesn't need to
Should be:
	if (!bh)

		goto iget_failed;

...
 iget_failed:

	iget_failed(inode);
This is weird.  This should be done by jumping to the proper label

--

It was - if sync fails, it should still try writing the mirrors.  Plus
Hrm, brelse(NULL) is allowed so the check is suspect anyway.  I did

this in a couple of other places, so I'll fix those up too.
Thanks!
--

Bob Copeland %% www.bobcopeland.com

--

Several of these look like they should be checked before use elsewhere in

the code (eg blocksize is fed into shifts later)
--

This if check is no needed, as your fill_super method never returns

success without setting it up.
--