Re: [Qemu-devel] [PATCH 01/12] aio: Introduce "type" in aio_set_fd_handler and aio_set_event_notifier

[Kevin Wolf]

Am 09.10.2015 um 07:45 hat Fam Zheng geschrieben:
> The parameter is added but not used.
> 
> Signed-off-by: Fam Zheng <address@hidden>
> ---
>  aio-posix.c                     |  4 ++-
>  aio-win32.c                     |  2 ++
>  async.c                         |  3 ++-
>  block/curl.c                    | 14 +++++-----
>  block/iscsi.c                   |  9 +++----
>  block/linux-aio.c               |  5 ++--
>  block/nbd-client.c              | 10 ++++---
>  block/nfs.c                     | 17 +++++-------
>  block/sheepdog.c                | 38 ++++++++++++++++++---------
>  block/ssh.c                     |  5 ++--
>  block/win32-aio.c               |  5 ++--
>  hw/block/dataplane/virtio-blk.c |  6 +++--
>  hw/scsi/virtio-scsi-dataplane.c | 24 +++++++++++------
>  include/block/aio.h             |  5 ++++
>  iohandler.c                     |  3 ++-
>  nbd.c                           |  4 ++-
>  roms/ipxe                       |  2 +-

This one looks unrelated.

>  tests/test-aio.c                | 58 
> +++++++++++++++++++++++------------------
>  18 files changed, 129 insertions(+), 85 deletions(-)

> diff --git a/include/block/aio.h b/include/block/aio.h
> index 400b1b0..830a0e8 100644
> --- a/include/block/aio.h
> +++ b/include/block/aio.h
> @@ -275,6 +275,9 @@ bool aio_pending(AioContext *ctx);
>   */
>  bool aio_dispatch(AioContext *ctx);
>  
> +#define AIO_CLIENT_UNSPECIFIED    (1 << 0)
> +#define AIO_CLIENT_MASK_ALL       -1
> +
>  /* Progress in completing AIO work to occur.  This can issue new pending
>   * aio as a result of executing I/O completion or bh callbacks.
>   *
> @@ -299,6 +302,7 @@ bool aio_poll(AioContext *ctx, bool blocking);
>   */
>  void aio_set_fd_handler(AioContext *ctx,
>                          int fd,
> +                        int type,
>                          IOHandler *io_read,
>                          IOHandler *io_write,
>                          void *opaque);
> @@ -312,6 +316,7 @@ void aio_set_fd_handler(AioContext *ctx,
>   */
>  void aio_set_event_notifier(AioContext *ctx,
>                              EventNotifier *notifier,
> +                            int type,
>                              EventNotifierHandler *io_read);

I'll leave the details like that AIO_CLIENT_MASK_ALL depends on two's
complement representation to László, but as a matter of personal taste,
I'd still prefer unsigned types for bitmasks.

I would still give my R-b if it weren't for the submodule change.

Kevin