RE: [PATCH v1] aio-posix: fix build failure io_uring 2.2

[Wang, Haiyue]

> -----Original Message-----
> From: Stefan Hajnoczi <stefanha@redhat.com>
> Sent: Monday, February 21, 2022 22:54
> To: Wang, Haiyue <haiyue.wang@intel.com>
> Cc: qemu-devel@nongnu.org; Fam Zheng <fam@euphon.net>; open list:Block I/O 
> path <qemu-block@nongnu.org>
> Subject: Re: [PATCH v1] aio-posix: fix build failure io_uring 2.2
> 
> On Fri, Feb 18, 2022 at 12:16:27AM +0800, Haiyue Wang wrote:
> > The io_uring fixed "Don't truncate addr fields to 32-bit on 32-bit":
> > https://git.kernel.dk/cgit/liburing/commit/?id=d84c29b19ed0b130000619cff40141bb1fc3615b
> >
> > This leads to build failure:
> > ../util/fdmon-io_uring.c: In function ‘add_poll_remove_sqe’:
> > ../util/fdmon-io_uring.c:182:36: error: passing argument 2 of 
> > ‘io_uring_prep_poll_remove’ makes
> integer from pointer without a cast [-Werror=int-conversion]
> >   182 |     io_uring_prep_poll_remove(sqe, node);
> >       |                                    ^~~~
> >       |                                    |
> >       |                                    AioHandler *
> > In file included from /root/io/qemu/include/block/aio.h:18,
> >                  from ../util/aio-posix.h:20,
> >                  from ../util/fdmon-io_uring.c:49:
> > /usr/include/liburing.h:415:17: note: expected ‘__u64’ {aka ‘long long 
> > unsigned int’} but argument
> is of type ‘AioHandler *’
> >   415 |           __u64 user_data)
> >       |           ~~~~~~^~~~~~~~~
> > cc1: all warnings being treated as errors
> >
> > So convert the paramter to right type according to the io_uring define.
> >
> > Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
> > ---
> >  util/fdmon-io_uring.c | 4 ++++
> >  1 file changed, 4 insertions(+)
> >
> > diff --git a/util/fdmon-io_uring.c b/util/fdmon-io_uring.c
> > index 1461dfa407..e7febbf11f 100644
> > --- a/util/fdmon-io_uring.c
> > +++ b/util/fdmon-io_uring.c
> > @@ -179,7 +179,11 @@ static void add_poll_remove_sqe(AioContext *ctx, 
> > AioHandler *node)
> >  {
> >      struct io_uring_sqe *sqe = get_sqe(ctx);
> >
> > +#ifdef LIBURING_HAVE_DATA64
> > +    io_uring_prep_poll_remove(sqe, (__u64)node);
> > +#else
> >      io_uring_prep_poll_remove(sqe, node);
> > +#endif
> >  }
> 
> Thanks for the patch. I suggest using the same approach as the liburing
> commit to avoid the #ifdef:

Old version is "void *user_data", so need to use ' LIBURING_HAVE_DATA64',
also, as the comment said:

/*
 * Tell the app the have the 64-bit variants of the get/set userdata
 */
#define LIBURING_HAVE_DATA64

And yes, two casts seems be more safe: (__u64)(uintptr_t)node

> 
> diff --git a/test/poll-cancel-ton.c b/test/poll-cancel-ton.c
> index f0875cd..05bf565 100644
> --- a/test/poll-cancel-ton.c
> +++ b/test/poll-cancel-ton.c
> @@ -55,7 +55,7 @@ static int del_polls(struct io_uring *ring, int fd, int nr)
> 
>                         sqe = io_uring_get_sqe(ring);
>                         data = sqe_index[lrand48() % nr];
> -                       io_uring_prep_poll_remove(sqe, data);
> +                       io_uring_prep_poll_remove(sqe, 
> (__u64)(uintptr_t)data);
> 
> So the QEMU add_poll_remove_sqe() function would do:
> 
>   io_uring_prep_poll_remove(sqe, (__u64)(uintptr_t)node);
> 
> Was there a reason why you chose an #ifdef instead?
> 
> Thanks,
> Stefan