作者: Thomas Shrimpton , R. Seth Terashima
DOI: 10.1007/978-3-662-46800-5_4
关键词:
摘要: We provide the first provable-security analysis of Intel Secure Key hardware RNG (ISK-RNG), versions which have appeared in processors since late 2011. To model ISK-RNG, we generalize PRNG-with-inputs primitive, introduced by Dodis et al. at CCS’13 for their /dev/[u]random analysis. The concrete security bounds uncover tell a mixed story. find that ISK-RNG lacks backward-security altogether, and forward-security bound “truly random” bits fetched \(\mathtt {RDSEED}\) instruction is potentially worrisome. On other hand, are able to prove stronger pseudorandom {RDRAND}\) instruction. En route these results, our main technical efforts focus on way employs CBCMAC as an entropy extractor.