Re: [PATCH take2 1/5] dio: centralize completion in dio_complete()

[Andrew Morton <akpm@osdl.org>]

On Mon,  2 Oct 2006 16:21:25 -0700 (PDT)
Zach Brown <zach.brown@oracle.com> wrote:

> dio: centralize completion in dio_complete()
> 
> The mechanics which decide the result of a direct IO operation were duplicated
> in the sync and async paths.
> 
> The async path didn't check page_errors which can manifest as silently
> returning success when the final pointer in an operation faults and its
> matching file region is filled with zeros.
> 
> The sync path and async path differed in whether they passed errors to the
> caller's dio->end_io operation.  The async path was passing errors to it which
> trips an assertion in XFS, though it is apparently harmless.
> 
> This centralizes the completion phase of dio ops in one place.  AIO will now
> return EFAULT consistently and all paths fall back to the previously sync
> behaviour of passing the number of bytes 'transferred' to the dio->end_io
> callback, regardless of errors.
> 
> dio_await_completion() doesn't have to propogate EIO from non-uptodate
> bios now that it's being propogated through dio_complete() via dio->io_error.
> This lets it return void which simplifies its sole caller.
> 
> ...
>
> -static void dio_complete(struct dio *dio, loff_t offset, ssize_t bytes)
> +static int dio_complete(struct dio *dio, loff_t offset, int ret)
>  {
> +	ssize_t transferred = 0;
> +
> +	if (dio->result) {
> +		transferred = dio->result;
> +
> +		/* Check for short read case */
> +		if ((dio->rw == READ) && ((offset + transferred) > dio->i_size))
> +			transferred = dio->i_size - offset;

On 32-bit machines ssize_t is `int' and loff_t is `long long'.  I guess
`transferred' cannot overflow because you can't write >4G.  And I guess
`transferred' cannot go negative because you cannot write >=2G.  Can you
confirm that thinking?