Alexander Viro wrote:
>
> In all fairness, ->read_inode() should not be a method. Correct way to
> deal with that would be along the lines
> [...]
>
> + if (inode->i_state & I_NEW) {
> + ext2_read_inode(inode);
> + unlock_new_inode(inode);
> + }
Why wouldn't I_CLEAR work for this? Then we'd be left with a generic
inode unlock that's also a common factor of get_new_inode and sync_one:
+void unlock_inode (struct inode *inode)
+{
+ inode->i_state &= ~I_LOCK;
+ wake_up(&inode->i_wait);
}
While you're in there slicing and dicing, the 'This is special!' comment
no longer means anything in its new context and, and even in the
original context it just set me off hunting for something genuinely
special.
--
Daniel
-
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to [EMAIL PROTECTED]
