Re: Possible bug in tools/libxl/libxl.c -- Variable passed by reference not set in one possible case

[Ian Campbell <Ian.Campbell@citrix.com>]

On Fri, 2014-10-10 at 10:06 +0100, Wei Liu wrote:
> Please send plain text email in the future. Some (if not all) developers
> only have very shabby text editor to read / reply to your email. ;-)
> 
> On Thu, Oct 09, 2014 at 08:28:53PM -0500, ayush ruia wrote:
> > What I am trying to say is that shouldn't the code block highlighted in
> > yellow should come before the block marked in green. Then it would update
> > the value of *target_memkb and *max_memkb in all possible situations.
> > 
> 
> I don't think so. This function means "if there's no such values in
> xenstore then retrieve them from hypervisor and fill them in xenstore,
> optionally return those values to the caller".
> 
> So if those values already are present in xenstore this function doesn't
> need to do anything.

If I call this function and receive a return code of zero how can I tell
if the target_memkb pointer I passed has been initialised or not?

If all of target, staticmax and freememslack are already set the
function returns 0 without updating those pointers, so I don't think we
can tell.

It does seem like the 
 if (target && staticmax && freememslack) {
        rc = 0;
        goto out;
    }
belongs after the code which writes back the two variables.

Ian.