samus: add automatic retries for host commands from EC to PD

Add three retries for EC to PD host commands. With this change,
removed retry mechanism in host_command_pd.c which was a retry
only for the specific EC_CMD_PD_EXCHANGE_STATUS host command.

BUG=chrome-os-partner:32006
BRANCH=none
TEST=Loaded EC code onto samus. Added the following code
for testing failed host commands to samus_pd common/host_command.c
host_command_task():
if ((evt & TASK_EVENT_CMD_PENDING) && pending_args) {
	if (i++ != 4)
		pending_args->result =
			host_command_process(pending_args);
	else {
		pending_args->result = -7;
		i = 0;
	}
	host_send_response(pending_args);
}
This test code on samus_pd drops one in every five host commands.

With this code, from the EC, I send "pdcmd 0 0", and verified that
1 out of 5 times the EC prints a host command failed code, but then
retries successfully.

Change-Id: Ibf43feefbfc7d791c45c6689b82c66f5d71046ab
Signed-off-by: Alec Berg <alecaberg@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/217461
Reviewed-by: Todd Broch <tbroch@chromium.org>

3 files changed, 27 insertions, 22 deletions

diff --git a/common/host_command_master.c b/common/host_command_master.c
index 38acd9a560..11d0dbca06 100644
--- a/common/host_command_master.c
+++ b/common/host_command_master.c
@@ -20,6 +20,9 @@
 /* Host command timeout */
 #define HOST_COMMAND_TIMEOUT_US SECOND
 
+/* Number of attempts for each PD host command */
+#define PD_HOST_COMMAND_ATTEMPTS 3
+
 static struct mutex pd_mutex;
 
 /**
@@ -86,7 +89,7 @@ static int pd_host_command_internal(int command, int version,
 	if (ret) {
 		i2c_lock(I2C_PORT_PD_MCU, 0);
 		CPRINTF("[%T i2c transaction 1 failed: %d]\n", ret);
-		return -ret;
+		return -EC_RES_BUS_ERROR;
 	}
 
 	resp_len = resp_buf[1];
@@ -107,7 +110,7 @@ static int pd_host_command_internal(int command, int version,
 	i2c_lock(I2C_PORT_PD_MCU, 0);
 	if (ret) {
 		CPRINTF("[%T i2c transaction 2 failed: %d]\n", ret);
-		return -ret;
+		return -EC_RES_BUS_ERROR;
 	}
 
 	/* Check for host command error code */
@@ -151,7 +154,7 @@ static int pd_host_command_internal(int command, int version,
 	if ((uint8_t)sum) {
 		CPRINTF("[%T command 0x%02x bad checksum returned: "
 			"%d]\n", command, sum);
-		return -EC_RES_ERROR;
+		return -EC_RES_INVALID_CHECKSUM;
 	}
 
 	/* Return output buffer size */
@@ -163,16 +166,23 @@ int pd_host_command(int command, int version,
 		    void *indata, int insize)
 {
 	int rv;
-
-	/* Acquire mutex */
-	mutex_lock(&pd_mutex);
-
-	/* Call internal version of host command */
-	rv = pd_host_command_internal(command, version, outdata, outsize,
-				      indata, insize);
-
-	/* Release mutex */
-	mutex_unlock(&pd_mutex);
+	int tries = 0;
+
+	/* Try multiple times to send host command. */
+	for (tries = 0; tries < PD_HOST_COMMAND_ATTEMPTS; tries++) {
+		/* Acquire mutex */
+		mutex_lock(&pd_mutex);
+		/* Call internal version of host command */
+		rv = pd_host_command_internal(command, version, outdata,
+					      outsize, indata, insize);
+		/* Release mutex */
+		mutex_unlock(&pd_mutex);
+
+		/* If host command error due to i2c bus error, try again. */
+		if (rv != -EC_RES_BUS_ERROR)
+			break;
+		task_wait_event(50*MSEC);
+	}
 
 	return rv;
 }
diff --git a/common/host_command_pd.c b/common/host_command_pd.c
index ecd5e07cf5..444e9e3288 100644
--- a/common/host_command_pd.c
+++ b/common/host_command_pd.c
@@ -26,7 +26,7 @@ static void pd_exchange_status(void)
 {
 	struct ec_params_pd_status ec_status;
 	struct ec_response_pd_status pd_status;
-	int rv = 0, tries = 0;
+	int rv = 0;
 
 	/* Send battery state of charge */
 	if (charge_get_flags() & CHARGE_FLAG_BATT_RESPONSIVE)
@@ -34,15 +34,9 @@ static void pd_exchange_status(void)
 	else
 		ec_status.batt_soc = -1;
 
-	/* Try 3 times to get the PD MCU status. */
-	while (tries++ < 3) {
-		rv = pd_host_command(EC_CMD_PD_EXCHANGE_STATUS, 0, &ec_status,
+	rv = pd_host_command(EC_CMD_PD_EXCHANGE_STATUS, 0, &ec_status,
 			     sizeof(struct ec_params_pd_status), &pd_status,
 			     sizeof(struct ec_response_pd_status));
-		if (rv >= 0)
-			break;
-		task_wait_event(500*MSEC);
-	}
 
 	if (rv < 0) {
 		CPRINTS("Host command to PD MCU failed");
diff --git a/include/ec_commands.h b/include/ec_commands.h
index 4428648511..951551769b 100644
--- a/include/ec_commands.h
+++ b/include/ec_commands.h
@@ -224,7 +224,8 @@ enum ec_status {
 	EC_RES_OVERFLOW = 11,		/* Table / data overflow */
 	EC_RES_INVALID_HEADER = 12,     /* Header contains invalid data */
 	EC_RES_REQUEST_TRUNCATED = 13,  /* Didn't get the entire request */
-	EC_RES_RESPONSE_TOO_BIG = 14    /* Response was too big to handle */
+	EC_RES_RESPONSE_TOO_BIG = 14,   /* Response was too big to handle */
+	EC_RES_BUS_ERROR = 15           /* Communications bus error */
 };
 
 /*