libxl__device_nextid and holes

libxl__device_nextid and holes

[Olaf Hering]

I wonder if libxl__device_nextid is robust enough to handle holes. If
there are 3 devices (0,1,2) of type X, and device #0 or #1 is removed
with "X-detach", wouldn't the code return nextid==2? The result would be
that the caller will write to the wrong path.

I have not veified this theory in practice.

Olaf

Re: libxl__device_nextid and holes

[Wei Liu]

On Mon, Mar 09, 2015 at 08:58:38AM +0100, Olaf Hering wrote:
> I wonder if libxl__device_nextid is robust enough to handle holes. If
> there are 3 devices (0,1,2) of type X, and device #0 or #1 is removed
> with "X-detach", wouldn't the code return nextid==2? The result would be
> that the caller will write to the wrong path.
> 
> I have not veified this theory in practice.
> 

I once had the same question. IIRC I did some experiments and that
function seemed to work fine.

You can verify this by attaching / detaching devices to a domain.

And I think the following line is the meat.

1956         nextid = strtoul(l[nb - 1], NULL, 10) + 1;


Wei.

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