[etherlab-users] SoE configuration failes (during startup).

[Florian Pose]

On Thu, Dec 02, 2010 at 04:09:01PM +0100, Baur Andreas wrote:
> Sometimes SoE configuration failes (during startup).
> 
> I use DC, RTAI, the newest devel-version of the master and a Rexroth servo drive.
> 
> Is it a synchronization problem among threads? See the error message:

Maybe. What is really strange, is that the datagram state seems to be different
in subsequent calls. Please try the below patch:

diff -r b5391b329b5d master/fsm_soe.c
--- a/master/fsm_soe.c  Tue Nov 30 15:24:21 2010 +0100
+++ b/master/fsm_soe.c  Tue Dec 07 14:49:04 2010 +0100
@@ -646,11 +646,18 @@
     uint8_t *data, mbox_prot, opcode, error_flag;
     uint16_t idn;
     size_t rec_size;
+    uint8_t state = datagram->state;
 
-    if (datagram->state == EC_DATAGRAM_TIMED_OUT && fsm->retries--)
+    if (state == EC_DATAGRAM_SENT
+        || state == EC_DATAGRAM_QUEUED) {
+        // datagram was not sent or received yet.
+        return;
+    }
+
+    if (state == EC_DATAGRAM_TIMED_OUT && fsm->retries--)
         return; // FIXME: request again?
 
-    if (datagram->state != EC_DATAGRAM_RECEIVED) {
+    if (state != EC_DATAGRAM_RECEIVED) {
         fsm->state = ec_fsm_soe_error;
         EC_SLAVE_ERR(slave, "Failed to receive SoE write"
                 " response datagram: ");


-- 
Best regards,
Florian Pose

http://etherlab.org