可信计算4动可信度量.ppt

* More on stop sign: Used to differentiate outputs from secure code. Also helps with seal (talk about later). * * When S terminates, untrusted OS will regain control, so we need a mechanism to protect S’s state until it is run again. TPM provides Seal and Unseal options… * Be a bit more explicit about what the “Flicker Platform” is. * Run detector over OS image in memory (not FS) * Emphasize properties: Even if server is compromised: Plaintext password protected. Potentially prevent offline dictionary attacks too. * * Unseal on today’s TPMs is an RSA decrypt. This is for our complete implementation. With other microbenchmarks, we observe variance across different TPM implementations. Point out that these are purely Flicker-induced overheads; essentially 0 w/o Flicker * Explain what the Quote is. * Our recommendations explicitly designed as minor tweaks to existing functionality. Be careful about discussion of multicore. * * * * * * * * Emphasize that VMM launch not really designed to return Define “VMM”. Mention SKINIT will be used (but SENTER very similar). SKINIT also performs other ops too (like DMA protection) * * * * * * * * * * * * * * * * * * * * * * * * I’m going to present work we’ve done exploring how far you can minimize the amount of code that an application must trust for its security. * * Smart application developer working on a program for signing certificates or checking passwords. * * Mention that it eliminates 50,000,000 LoC (Windows), 5,000,000 (Linux) or 80,000 (Xen) Mention that shim should be amenable to manual audit and formal analysis * * * * Application: SSH Passwords nonce Start Gen {K, K-1} K EncryptK(passwd) EncryptK(passwd) OK! Shim S K Shim S K-1 Shim S K-1 Shim S EncryptK(passwd) passwd * Other Applications Implemented Enhanced Certificate Authority (CA) Private signing key isolated from entire system Verifiable distributed computing Verifiably perform a computational task on a remote computer Ex: SETI@Home, Foldin