Cache Robustness

[turbo]

How is the caching logic implemented? The docs don't really disclose that. Specifically, is the implementation vulnerable against a cache stampede attack?

Vulnerable:

if (!inCache(key)) 
  expensive_calc()
  setCache(key)

Safe:

if (!inCache(key)) 
  aquire_global_lock(key)
  expensive_calc()
  setCache(key)
  release_global_lock(key)