devlink: Fix reporter's recovery condition

Devlink health core conditions the reporter's recovery with the
expiration of the grace period. This is not relevant for the first
recovery. Explicitly demand that the grace period will only apply to
recoveries other than the first.

Fixes: c8e1da0bf923 ("devlink: Add health report functionality")
Signed-off-by: Aya Levin <ayal@mellanox.com>
Reviewed-by: Moshe Shemesh <moshe@mellanox.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/net/core/devlink.c b/net/core/devlink.c
index 1ec2e9f..899edce 100644 (file)
--- a/net/core/devlink.c
+++ b/net/core/devlink.c
@@ -5368,6 +5368,7 @@ int devlink_health_report(struct devlink_health_reporter *reporter,
 {
        enum devlink_health_reporter_state prev_health_state;
        struct devlink *devlink = reporter->devlink;
+       unsigned long recover_ts_threshold;
 
        /* write a log message of the current error */
        WARN_ON(!msg);
@@ -5378,10 +5379,12 @@ int devlink_health_report(struct devlink_health_reporter *reporter,
        devlink_recover_notify(reporter, DEVLINK_CMD_HEALTH_REPORTER_RECOVER);
 
        /* abort if the previous error wasn't recovered */
+       recover_ts_threshold = reporter->last_recovery_ts +
+                              msecs_to_jiffies(reporter->graceful_period);
        if (reporter->auto_recover &&
            (prev_health_state != DEVLINK_HEALTH_REPORTER_STATE_HEALTHY ||
-            jiffies - reporter->last_recovery_ts <
-            msecs_to_jiffies(reporter->graceful_period))) {
+            (reporter->last_recovery_ts && reporter->recovery_count &&
+             time_is_after_jiffies(recover_ts_threshold)))) {
                trace_devlink_health_recover_aborted(devlink,
                                                     reporter->ops->name,
                                                     reporter->health_state,