Hello.

Christoph Hellwig wrote:
quoted text
> > NACK to this. Passing function parameters through the task_struct is
> > definitely not an acceptable hack
> Exactly.  Having a vfsmount other than the current processes root or
> current working directory in task_struct doesn't make any sense.
The process's root and cwd vfsmounts are used permanently, but
this last_vfsmount passed via task_struct behaves like temporary variable
pushed on stack memory.
In other words, last_vfsmount becomes NULL when it becomes invalid.

| static inline int vfs_create2(struct inode *dir, struct dentry *dentry,
| 			      int mode, struct nameidata *nd)
| {
| 	int ret;
| 	struct vfsmount *mnt = nd ? nd->path.mnt : NULL;
| 	struct task_struct *task = current;
| 	struct vfsmount *prev_mnt = task->last_vfsmount;
| 	task->last_vfsmount = mntget(mnt);
| 	ret = vfs_create(dir, dentry, mode, nd);
| 	task->last_vfsmount = prev_mnt;
| 	mntput(mnt);
| 	return ret;
| }

I agree that keeping last_vfsmount after it lost it's reference count is bad, but
I don't understand why keeping last_vfsmount while it has it's reference count is bad too.

Regards.

-
unsubscribe notice
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/