Re: Change in NFS client behavior

[Andrew Morton <akpm@osdl.org>]

Trond Myklebust <trond.myklebust@fys.uio.no> wrote:
>

Looks good from a quick scan.

> +static int do_posix_truncate(struct dentry *dentry, loff_t length)
>  +{
>  +	int err = 0;
>  +	struct iattr newattrs;
>  +
>  +	newattrs.ia_size = length;
>  +	newattrs.ia_valid = ATTR_SIZE;
>  +
>  +	down(&dentry->d_inode->i_sem);
>  +	/* In SuS/Posix lore, truncate to the current file size is a no-op */
>  +	if (length != i_size_read(dentry->d_inode))
>  +		err = notify_change(dentry, &newattrs);
>  +	up(&dentry->d_inode->i_sem);
>  +	return err;
>  +}

I'm not sure that we really need to bother putting the i_size test inside
i_sem.  If somebody is changing the file size in parallel with truncate
then they'll get unpredictable results anyway.  Needs deep thought.