Re: [Qemu-devel] [RFC/PATCH] virtio-9p: Add Read only support for 9p export.

[Stefan Hajnoczi <stefanha@gmail.com>]

On Mon, May 23, 2011 at 8:58 AM, M. Mohan Kumar <mohan@in.ibm.com> wrote:
> A new fsdev parameter "access" is introduced to control accessing 9p export.
> access=ro|rw can be used to specify the access type. By default rw access
> is given to 9p export.

It would be consistent with -drive to use readonly=on|off (default:
off).  Do you have any future access modes in mind that require more
than readonly on|off?

It seems the client does not know ahead of time that the mount is
read-only.  I wonder if this exposes any weirdness compared to
actually mounting ro on the client side (i.e. operations that appear
to start okay but then fail with -EROFS at a different point than if
you had mounted ro)?  It could cause apps to error in new ways.

> +static inline int is_ro_export(FsContext *fs_ctx)
> +{
> +        if (fs_ctx->flags & FS_RDONLY) {
> +                return 1;
> +        }
> +        return 0;
> +}

Please use bool and fix indentation:

static inline bool is_ro_export(FsContext *fs_ctx)
{
    return fs_ctx->flags & FS_RDONLY;
}

> +
>  static void v9fs_version(V9fsState *s, V9fsPDU *pdu)
>  {
>     V9fsString version;
> @@ -1734,6 +1742,13 @@ static void v9fs_open_post_lstat(V9fsState *s, V9fsOpenState *vs, int err)
>         vs->fidp->fs.dir = v9fs_do_opendir(s, &vs->fidp->path);
>         v9fs_open_post_opendir(s, vs, err);
>     } else {
> +        if (is_ro_export(&s->ctx)) {
> +            if (vs->mode & O_WRONLY || vs->mode & O_RDWR ||
> +                            vs->mode & O_APPEND) {
> +                err = -EROFS;
> +                goto out;
> +            }
> +        }
>         if (s->proto_version == V9FS_PROTO_2000L) {
>             flags = vs->mode;
>             flags &= ~(O_NOCTTY | O_ASYNC | O_CREAT);
> @@ -3606,6 +3621,33 @@ static pdu_handler_t *pdu_handlers[] = {
>     [P9_TREMOVE] = v9fs_remove,
>  };
>
> +static inline int is_read_only_op(int id)

bool

Stefan