On Thu, Jan 09, 2014 at 09:05:05PM +0000, KY Srinivasan wrote:
> >
> > We've had this discussion before where you urge me to trust the host...
>
> I am just implementing the protocol specification given by the host. If I cannot trust the
> specified protocol, I am not sure what else can be done here.
I'm sorry I got too upset and we're not communicating correctly.
Anyway, looking at the code, there is actually locking in
process_chn_event() which protect ->fcopy_context. It's still sucky to
only store the last message in the queue but that's ok.
+static int fcopy_handle_handshake(int op)
+{
+ int ret = 1;
+
+ pr_info("FCP: user-mode registering done.\n");
+ fcopy_transaction.active = false;
+ set_channel_read_state((struct vmbus_channel *)
+ fcopy_transaction.fcopy_context,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Dereferenced inside the call to set_channel_read_state()
+ true);
+
+ if (fcopy_transaction.fcopy_context)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Checked for NULL.
+ hv_fcopy_onchannelcallback(fcopy_transaction.fcopy_context);
+
+ in_hand_shake = false;
+ return ret;
+}
regards,
dan carpenter
_______________________________________________
devel mailing list
devel@xxxxxxxxxxxxxxxxxxxxxx
http://driverdev.linuxdriverproject.org/mailman/listinfo/driverdev-devel
