Tomasz Kvarsin wrote:
> This I got during boot with 2.6.20-rc4:
> =============================================
> [ INFO: possible recursive locking detected ]
> 2.6.20-rc4 #3
> ---------------------------------------------
> mount/5819 is trying to acquire lock:
> (&jfs_ip->commit_mutex){--..}, at: [<c03395e1>] mutex_lock+0x21/0x30
>
> but task is already holding lock:
> (&jfs_ip->commit_mutex){--..}, at: [<c03395e1>] mutex_lock+0x21/0x30
>
> other info that might help us debug this:
> 2 locks held by mount/5819:
> #0:  (&inode->i_mutex){--..}, at: [<c03395e1>] mutex_lock+0x21/0x30
> #1:  (&jfs_ip->commit_mutex){--..}, at: [<c03395e1>] mutex_lock+0x21/0x30
>
Problem was in jfs_create(), it was calling 2 mutex_calls.
=============================================
static int jfs_create(struct inode *dip, struct dentry *dentry, int mode,
                struct nameidata *nd)
{
        int rc = 0;
        tid_t tid;              /* transaction id */
        struct inode *ip = NULL;        /* child directory inode */
        ino_t ino;
        struct component_name dname;    /* child directory name */
        struct btstack btstack;
        struct inode *iplist[2];
        struct tblock *tblk;

        jfs_info("jfs_create: dip:0x%p name:%s", dip, dentry->d_name.name);
      .....................................
       tid = txBegin(dip->i_sb, 0);

        mutex_lock(&JFS_IP(dip)->commit_mutex);
        mutex_lock(&JFS_IP(ip)->commit_mutex);

        rc = jfs_init_acl(tid, ip, dip);
        if (rc)
                goto out3;
=======================================

So below patch should fix this problem,please test this. Let me know 
your comments on this.

Signed-off-by: Srinivasa DS <srinivasa@in.ibm.com>