Re: [Xen-devel] libxl_device.c, stat() and remote disks

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Mon, 2013-04-22 at 17:54 +0100, Dave Scott wrote:
> On 22/04/13 15:36, Ian Campbell wrote:
> > On Mon, 2013-04-22 at 15:07 +0100, David Scott wrote:
> >> Is it safe to remove the stat() from libxl_device.c?
> >
> > I expect it only makes sense when backendtype=phy and perhaps the
> > disk->script check has served as an imperfect surrogate for that until
> > now?
> >
> > Does changing
> >       } else if (!disk->script) {
> > into
> >       } else if (disk->backendtype == ...PHY && !disk->script) {
> >
> > Work for you also?
> 
> Yes, it works nicely against 4.2.1! The exact patch I applied was:
> 
> --- tools/libxl/libxl_device.c.orig     2013-04-22 14:52:54.745001092 +0000
> +++ tools/libxl/libxl_device.c  2013-04-22 14:54:11.566001097 +0000
> @@ -236,7 +236,7 @@
>               return ERROR_INVAL;
>           }
>           memset(&a.stab, 0, sizeof(a.stab));
> -    } else if (!disk->script) {
> +    } else if (disk->backend == LIBXL_DISK_BACKEND_PHY && !disk->script) {
>           if (stat(disk->pdev_path, &a.stab)) {
>               LIBXL__LOG_ERRNO(ctx, LIBXL__LOG_ERROR, "Disk vdev=%s "
>                                "failed to stat: %s",
> 
> Would you like me to retest this against -unstable? Also would you like 
> me to 'git format-patch/send-email' or is it too trivial to bother?

A changelog and a S-o-b should be sufficient, thanks.

However, I think this code is probably broken for driver domains as
well. I think it probably needs a disk->backend_domid check in addition
to what's there -- if you wouldn't mind folding that into the patch that
would be awesome.

There's so many exclusions now I'm wondering if the stat is actually
useful, oh well. I suppose it is still active in some of the more common
cases.

Ian,


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel