edac: Allow panic on correctable errors (CE)

Add an EDAC device flag and associated sysfs entries to
allow an EDAC driver to be configured to panic the kernel
if a correctable error is detected. Though correctable
errors (by definition) have no adverse system effects,
a panic may still be useful, since a correctable error may
be indicative of a marginal system state.

Change-Id: I98921469254aa7b999979c1c7d9186286f982a0c
Signed-off-by: Stepan Moskovchenko <stepanm@codeaurora.org>

drivers/edac/edac_core.h

@@ -223,6 +223,7 @@ struct edac_device_ctl_info {
 	/* Per instance controls for this edac_device */
 	int log_ue;		/* boolean for logging UEs */
 	int log_ce;		/* boolean for logging CEs */
+	int panic_on_ce;	/* boolean for panic'ing on an CE */
 	int panic_on_ue;	/* boolean for panic'ing on an UE */
 	unsigned poll_msec;	/* number of milliseconds to poll interval */
 	unsigned long delay;	/* number of jiffies for poll_msec */

drivers/edac/edac_device.c

@@ -611,6 +611,12 @@ static inline int edac_device_get_log_ue(struct edac_device_ctl_info *edac_dev)
 	return edac_dev->log_ue;
 }
 
+static inline int edac_device_get_panic_on_ce(struct edac_device_ctl_info
+					*edac_dev)
+{
+	return edac_dev->panic_on_ce;
+}
+
 static inline int edac_device_get_panic_on_ue(struct edac_device_ctl_info
 					*edac_dev)
 {
@@ -660,6 +666,11 @@ void edac_device_handle_ce(struct edac_device_ctl_info *edac_dev,
 				"CE: %s instance: %s block: %s '%s'\n",
 				edac_dev->ctl_name, instance->name,
 				block ? block->name : "N/A", msg);
+
+	if (edac_device_get_panic_on_ce(edac_dev))
+		panic("EDAC %s: CE instance: %s block %s '%s'\n",
+			edac_dev->ctl_name, instance->name,
+			block ? block->name : "N/A", msg);
 }
 EXPORT_SYMBOL_GPL(edac_device_handle_ce);
 

drivers/edac/edac_device_sysfs.c

@@ -62,6 +62,13 @@ static ssize_t edac_device_ctl_log_ce_store(struct edac_device_ctl_info
 	return count;
 }
 
+/* 'panic_on_ce' */
+static ssize_t edac_device_ctl_panic_on_ce_show(struct edac_device_ctl_info
+						*ctl_info, char *data)
+{
+	return snprintf(data, PAGE_SIZE, "%u\n", ctl_info->panic_on_ce);
+}
+
 /* 'panic_on_ue' */
 static ssize_t edac_device_ctl_panic_on_ue_show(struct edac_device_ctl_info
 						*ctl_info, char *data)
@@ -69,6 +76,21 @@ static ssize_t edac_device_ctl_panic_on_ue_show(struct edac_device_ctl_info
 	return sprintf(data, "%u\n", ctl_info->panic_on_ue);
 }
 
+static ssize_t edac_device_ctl_panic_on_ce_store(struct edac_device_ctl_info
+						 *ctl_info, const char *data,
+						 size_t count)
+{
+	unsigned long val;
+
+	/* if parameter is zero, turn off flag, if non-zero turn on flag */
+	if (kstrtoul(data, 0, &val) < 0)
+		return -EINVAL;
+
+	ctl_info->panic_on_ce = !!val;
+
+	return count;
+}
+
 static ssize_t edac_device_ctl_panic_on_ue_store(struct edac_device_ctl_info
 						 *ctl_info, const char *data,
 						 size_t count)
@@ -156,6 +178,9 @@ CTL_INFO_ATTR(log_ue, S_IRUGO | S_IWUSR,
 	edac_device_ctl_log_ue_show, edac_device_ctl_log_ue_store);
 CTL_INFO_ATTR(log_ce, S_IRUGO | S_IWUSR,
 	edac_device_ctl_log_ce_show, edac_device_ctl_log_ce_store);
+CTL_INFO_ATTR(panic_on_ce, S_IRUGO | S_IWUSR,
+	edac_device_ctl_panic_on_ce_show,
+	edac_device_ctl_panic_on_ce_store);
 CTL_INFO_ATTR(panic_on_ue, S_IRUGO | S_IWUSR,
 	edac_device_ctl_panic_on_ue_show,
 	edac_device_ctl_panic_on_ue_store);
@@ -164,6 +189,7 @@ CTL_INFO_ATTR(poll_msec, S_IRUGO | S_IWUSR,
 
 /* Base Attributes of the EDAC_DEVICE ECC object */
 static struct ctl_info_attribute *device_ctrl_attr[] = {
+	&attr_ctl_info_panic_on_ce,
 	&attr_ctl_info_panic_on_ue,
 	&attr_ctl_info_log_ue,
 	&attr_ctl_info_log_ce,