Re: [Qemu-devel] [PATCH] virtio-scsi: Fix build with gcc 9

[Peter Maydell]

On Thu, 28 Feb 2019 at 17:59, Greg Kurz <gr...@kaod.org> wrote:
>
> Build fails with gcc 9:
>
>   CC      ppc64-softmmu/hw/scsi/virtio-scsi.o
> hw/scsi/virtio-scsi.c: In function ‘virtio_scsi_do_tmf’:
> hw/scsi/virtio-scsi.c:265:39: error: taking address of packed member of 
> ‘struct virtio_scsi_ctrl_tmf_req’ may result in an unaligned pointer value 
> [-Werror=address-of-packed-member]
>   265 |     virtio_tswap32s(VIRTIO_DEVICE(s), &req->req.tmf.subtype);
>       |                                       ^~~~~~~~~~~~~~~~~~~~~
> cc1: all warnings being treated as errors
>
> All the fields in struct virtio_scsi_ctrl_tmf_req are naturally aligned,
> so we could in theory drop QEMU_PACKED. Unfortunately, the header file
> is imported from linux which already has the packed attribute. Trying to
> fix that in the update-linux-headers.sh script is likely to produce
> ugliness. Turn the call to virtio_tswap32s() into an assignment instead.
>
> Signed-off-by: Greg Kurz <gr...@kaod.org>
> ---
>  hw/scsi/virtio-scsi.c |    8 +++++++-
>  1 file changed, 7 insertions(+), 1 deletion(-)
>
> diff --git a/hw/scsi/virtio-scsi.c b/hw/scsi/virtio-scsi.c
> index ce99d288b035..839f1202567d 100644
> --- a/hw/scsi/virtio-scsi.c
> +++ b/hw/scsi/virtio-scsi.c
> @@ -262,7 +262,13 @@ static int virtio_scsi_do_tmf(VirtIOSCSI *s, 
> VirtIOSCSIReq *req)
>      /* Here VIRTIO_SCSI_S_OK means "FUNCTION COMPLETE".  */
>      req->resp.tmf.response = VIRTIO_SCSI_S_OK;
>
> -    virtio_tswap32s(VIRTIO_DEVICE(s), &req->req.tmf.subtype);
> +    /*
> +     * req->req.tmf has the QEMU_PACKED attribute. Don't use 
> virtio_tswap32s()
> +     * to avoid compiler errors.
> +     */
> +    req->req.tmf.subtype =
> +        virtio_tswap32(VIRTIO_DEVICE(s), req->req.tmf.subtype);
> +
>      switch (req->req.tmf.subtype) {
>      case VIRTIO_SCSI_T_TMF_ABORT_TASK:
>      case VIRTIO_SCSI_T_TMF_QUERY_TASK:

Reviewed-by: Peter Maydell <peter.mayd...@linaro.org>

I haven't generally bothered to add comments about packed
structs when I've switched away from the *s versions of byteswap
functions, but it doesn't hurt.

We should consider just replacing the other dozen uses
of virtio_tswap*s() with the non-s versions and dropping the
s functions entirely...

thanks
-- PMM