Re: [PATCH 1/3] Vectorize aio_read/aio_write methods

[Andrew Morton <akpm@osdl.org>]

Badari Pulavarty <pbadari@us.ibm.com> wrote:
>
>  static ssize_t ep_aio_read_retry(struct kiocb *iocb)
>  {
>  	struct kiocb_priv	*priv = iocb->private;
> -	ssize_t			status = priv->actual;
> +	ssize_t			len, total;
>  
>  	/* we "retry" to get the right mm context for this: */
> -	status = copy_to_user(priv->ubuf, priv->buf, priv->actual);
> -	if (unlikely(0 != status))
> -		status = -EFAULT;
> -	else
> -		status = priv->actual;
> +
> +	/* copy stuff into user buffers */
> +	total = priv->actual;
> +	len = 0;
> +	for (i=0; i < priv->count; i++) {

	for (i = 0

> +		ssize_t this = min(priv->iv[i].iov_len, (size_t)total);

min_t().

Strange mixture of size_t and ssize_t there.

> +		if (copy_to_user(priv->iv[i].iov_buf, priv->buf, this))
> +			break;
> +
> +		total -= this;
> +		len += this;
> +		if (total <= 0)
> +			break;
> +	}
> +
> +	if (unlikely(len != 0))
> +		len = -EFAULT;

This looks wrong.  I think you meant (total != 0).




Together these three patches shrink the kernel by 113 lines.  I don't know
what the effect is on text size, but that's a pretty modest saving, at a
pretty high risk level.

What else do we get in return for this risk?