2 ms·
I'd start with the requirements for secure SCM's in general. Wheeler has those: http://www.dwheeler.com/essays/scm-security.html http://www.dwheeler.com/essays
by nickpsecurity 11y ago
I'd start with the requirements for secure SCM's in general. Wheeler has those:
http://www.dwheeler.com/essays/scm-security.html http://www.dwheeler.com/essays/scm-security.html
Then add endpoint security. Then add certified compilation to know compiler passes don't add vulnerabilities. Then add reproducible builds, cryptographic signing, and so on. Now one sees the scope of the problem. At this point, most people will just ignore most of the problem in favor of implementing a subset and saying they "secured the build and deployment process." ;)
- CiPHPerCoder 11y agoOn that note, definitely check out boringcc: https://groups.google.com/forum/#!topic/boring-crypto/48qa1kWignU https://groups.google.com/forum/#!topic/boring-crypto/48qa1k...
- nickpsecurity 11y agoI was definitely glad to see that when he posted it. Getting some mainstream attention. Meanwhile, CompSci has been doing things like CompCert C compiler, strong analysis showing absence of common issues, and so on. We have stuff to use right now at varying stages of completion. People just have to heed the call and build on it.