ena driver TX timeouts leading to softirq hangs

Issue

• AWS EC2 instances running are intermittently going into a hung state. Network transmit timeouts start occurring and the NIC is hung:

ena 0000:00:05.0 eth0: Found a Tx that wasn't completed on time, qid 1, index 492. 13094000 usecs have passed since last napi execution. Missing Tx timeout value 5000 msecs
ena 0000:00:05.0 eth0: Found a Tx that wasn't completed on time, qid 1, index 493. 13097000 usecs have passed since last napi execution. Missing Tx timeout value 5000 msecs
ena 0000:00:05.0 eth0: Found a Tx that wasn't completed on time, qid 1, index 494. 13101000 usecs have passed since last napi execution. Missing Tx timeout value 5000 msecs
ena 0000:00:05.0 eth0: Found a Tx that wasn't completed on time, qid 1, index 495. 13104000 usecs have passed since last napi execution. Missing Tx timeout value 5000 msecs
ena 0000:00:05.0 eth0: The number of lost tx completions is above the threshold (132 > 128). Reset the device
ena 0000:00:05.0 eth0: Trigger reset is on
ena 0000:00:05.0 eth0: tx_timeout: 9
ena 0000:00:05.0 eth0: suspend: 0
ena 0000:00:05.0 eth0: resume: 0
ena 0000:00:05.0 eth0: wd_expired: 0

• The kernel log (dmesg) shows multiple watchdog: BUG: soft lockup - CPU#1 stuck for 23s! errors, which then causes ena to restart and connection loss.