On Fri, 2011-01-14 at 14:55 +0000, Kamala Narasimhan wrote:
> On Fri, Jan 14, 2011 at 4:05 AM, Ian Campbell <Ian.Campbell@xxxxxxxxxx> wrote:
> > On Thu, 2011-01-13 at 15:35 +0000, Kamala Narasimhan wrote:
> >> This patch performs some very basic validation on the virtual disk
> >> file passed through the config file. This validation ensures that we
> >> don't go too far with the initialization like spawn qemu and more
> >> while there could be some potentially fundamental issues. Obviously,
> >> there is a lot of room for improvement in the kind of validations we
> >> could do but the below is a minimal first stab at it. Please consider
> >> this for inclusion or feel free to tweak it as necessary.
> >> Signed-off-by: Kamala Narasimhan <kamala.narasimhan@xxxxxxxxxx>
> > I wonder if the validation function should be part of libxl?
> We are better off performing these checks early on as they are very
> basic. Getting far enough to spawn qemu and getting to its block
> device initialization code and failing there is a bit of a chase when
> it comes to troubleshooting these issues, the cause of which are
> rather trivial. That said, in the long run we might want to move
> these validations to upstream qemu as qemu also must perform these
> checks especially when run without an accelerator (as there wouldn't
> be a toolstack to perform these checks for it in that case). But,
> till that is accomplished these checks need to be somewhere and libxl
> seem like a reasonable place in my opinion.
I think Ians point is that your change affects the 'xl' binary and not
the libxl.so library.
Perhaps libxl_device_disk_add() and libxl_cdrom_insert() would be the
reasonable places to add this, replacing fprintf() calls with libxl
Xen-devel mailing list