Possible bug in xenbus __xenbus_switch_state

Possible bug in xenbus __xenbus_switch_state

[Andrew Reimers]

HI,

Looking at the function __xenbus_switch_state in file source/drivers/xen/xenbus/xenbus_client.c,

The comment says:

"We check whether the state is currently set to the given value, and
	   if not, then the state is set."

The code loads the old state value, but never does anything with it.

I would expect it to look something like:

err = xenbus_scanf(xbt, dev->nodename, "state", "%d", &current_state);
if (err != 1 || current_state != state) {
	err = xenbus_printf(xbt, dev->nodename, "state", "%d", state);
		if (err) {
		xenbus_switch_fatal(dev, depth, err, "writing new state");
		goto abort;
	}
}

--
Andrew Reimers

Re: Possible bug in xenbus __xenbus_switch_state

[David Vrabel]

On 04/12/14 09:40, Andrew Reimers wrote:
> HI,
> 
> Looking at the function __xenbus_switch_state in file source/drivers/xen/xenbus/xenbus_client.c,
> 
> The comment says:
> 
> "We check whether the state is currently set to the given value, and
> 	   if not, then the state is set."
> 
> The code loads the old state value, but never does anything with it.

It does:

        if (state == dev->state)
                return 0;

The read of the state key is to check it still exists.

   "Furthermore, if the node has gone, we don't write
    to it, as the device will be tearing down, and we don't want to
    resurrect that directory."

David