SM doesn't quit endless loop when S3 service returns non-retryable error

The recent incident unveils that if S3 service continuously returns non-retryable error both GET and PUT stuck in an endless loop w/o any reaction(no logs produced). From one perspective this protects the persistent storage from future corruption but doesn't facility root-cause troubleshooting.
The solution would be to add extra log messages with non-retryable error type received from S3 service.