block_cache checksum frequently triggerred

[wangyuyue]

As mentioned in #226,
I run the ssd_perf/graph_cache_leader workload for 64 navy writer threads and 64 navy reader threads. After 50 minutes, the terminal output is flooded with the item header checksum mismatch error.
Is this a software error or my SSD's hardware error?

Thanks in advance for any suggestions.