hw/fsp: Do not queue SP and SPCN class messages during reset/reload

During FSP R/R, the FSP is inaccessible and will lose state. Messages to the
FSP are generally queued for sending later.

It does seem like the FSP fails to process any subseuqent messages of certain
classes (SP info -- ipmi) if it receives queued mbox messages it isn't expecting.

In certain other cases (sensors), the FSP driver returns a default code (async
completion) even though there is no known bound from the time of this error
return to the actual data being available. The kernel driver keeps waiting
leading to soft-lockup on the host side.

Mitigate both these (known) cases by returning OPAL_BUSY so the host driver
knows to retry later.

With this change, the sensors command works fine when the FSP comes back.

This version also resolves the remaining IPMI issues

Signed-off-by: Ananth N Mavinakayanahalli <ananth@linux.vnet.ibm.com>
Tested-by: Pridhiviraj Paidipeddi <ppaidipe@linux.vnet.ibm.com>
Signed-off-by: Stewart Smith <stewart@linux.vnet.ibm.com>

1 files changed, 6 insertions, 0 deletions

diff --git a/hw/fsp/fsp-sensor.c b/hw/fsp/fsp-sensor.c
index 0fa31156..63635303 100644
--- a/hw/fsp/fsp-sensor.c
+++ b/hw/fsp/fsp-sensor.c
@@ -376,6 +376,9 @@ static int64_t fsp_sensor_send_read_request(struct opal_sensor_data *attr)
 	uint32_t align;
 	uint32_t cmd_header;
 
+	if (fsp_in_rr())
+		return OPAL_BUSY;
+
 	prlog(PR_INSANE, "Get the data for modifier [%x]\n",
 	      spcn_mod_data[attr->mod_index].mod);
 
@@ -516,6 +519,9 @@ int64_t fsp_opal_read_sensor(uint32_t sensor_hndl, int token,
 
 	prlog(PR_INSANE, "fsp_opal_read_sensor [%08x]\n", sensor_hndl);
 
+	if (fsp_in_rr())
+		return OPAL_BUSY;
+
 	if (sensor_state == SENSOR_PERMANENT_ERROR) {
 		rc = OPAL_HARDWARE;
 		goto out;