Fix a race condition in the trap handler

status.priv may be read after returning from the trap handler, which
causes sq_interrupt_word_wave.priv to be 0 even though the s_sendmsg
instruction was initiated when status.priv was 1.

To work around this, added a s_waitcnt lgkmcnt(0) after s_sendmsg
to make sure the message is sent before continuing.

Signed-off-by: Jay Cornwall <Jay.Cornwall@amd.com>
Signed-off-by: Laurent Morichetti <Laurent.Morichetti@amd.com>
Change-Id: Ieb75005ca1559ef03d0efac80e966f521e41fcb7

runtime/hsa-runtime/core/runtime/trap_handler/trap_handler.s

@@ -212,6 +212,7 @@ trap_entry:
   s_mov_b32            m0, ttmp3
   s_nop                0x0 // Manually inserted wait states
   s_sendmsg            sendmsg(MSG_INTERRUPT)
+  s_waitcnt            lgkmcnt(0) // Wait for the message to go out.
   s_mov_b32            m0, ttmp2
 
   // Parking the wave requires saving the original pc in the preserved ttmps.