pldmd: do not exit on recv fail with af-mctp transport

Under the Linux kernel AF_MCTP transport, transient receive failures
(e.g., PLDM_REQUESTER_RECV_FAIL returned when a downstream endpoint is
resetting, disconnected, or sending unmapped packets) do not indicate
that the underlying transport socket has been closed by a daemon.

Exiting the event loop on PLDM_REQUESTER_RECV_FAIL was originally
introduced for mctp-demux UNIX domain sockets, where recv failure
indicated the demux daemon process had terminated. In the AF_MCTP
architecture, this causes pldmd to terminate and restart unnecessarily
whenever a downstream device resets.

Restrict the socket-close exit logic to PLDM_TRANSPORT_WITH_MCTP_DEMUX
so that AF_MCTP logs a warning and continues event processing.

Tested: Verified error handling for AF_MCTP transport.
Google-Bug-Id: 543041824
Change-Id: I5c1b0c0b57dd6e27d173bdceaf0cd80010b0afee
Signed-off-by: Guangzong Chen <guangzong@google.com>
TAG=agy
CONV=f6e4e885-dc2d-47ba-99c7-be407d766de9

pldmd/pldmd.cpp

diff --git a/pldmd/pldmd.cpp b/pldmd/pldmd.cpp
index a32f6dc..25a976e 100644
--- a/pldmd/pldmd.cpp
+++ b/pldmd/pldmd.cpp
@@ -434,6 +434,7 @@
         // TODO check that we get here if mctp-demux dies?
         else if (returnCode == PLDM_REQUESTER_RECV_FAIL)
         {
+#if defined(PLDM_TRANSPORT_WITH_MCTP_DEMUX)
             // MCTP daemon has closed the socket this daemon is connected to.
             // This may or may not be an error scenario, in either case the
             // recovery mechanism for this daemon is to restart, and hence exit
@@ -443,6 +444,10 @@
                 "MCTP daemon closed the socket, IO exiting with response code '{RC}'",
                 "RC", returnCode);
             io.get_event().exit(0);
+#else
+            warning("Failed to receive PLDM message, response code '{RC}'",
+                    "RC", returnCode);
+#endif
         }
         else
         {