3 ms·

This is a really glib view of the vast and complicated reasons people are encouraged to update their OS. It's not mere whimsy that in a rapidly developing malwa

This is a really glib view of the vast and complicated reasons people are encouraged to update their OS. It's not mere whimsy that in a rapidly developing malware landscape, updating your OS is a necessity.

grishka 5y ago
Security patches are one thing. The constant moving things around and putting "fresh coats of paint" onto everything without solving any real problems is another. But then, at some point, all mainstream OSes will have been rewritten in memory-safe languages and there won't be any more exploitable vulnerabilities remaining. What then?
- nippoo 5y ago
  Not all vulnerabilities are due to memory safety!
  - grishka 5y ago
    But most are, and they usually are most devastating. Mistakes in logic can sometimes be vulnerabilities that could, for example, disclose some information to an unintended party, but I can't imagine how one could lead to remote code execution.
- jchw 5y ago
  If that last bit truly isn’t a joke, let me be the first to point out that moving to memory-safe languages will not eliminate vulnerabilities. It won’t even really be close. Formal proofs of correctness would probably get us closer to software that is bug-free, but literally nothing is a panacea, and at the scale of attack surface we’re dealing with most of the extreme options (even including “rewrite it in a memory safe language”) are too big to foresee happening in the short or mid-term in any meaningful way.
- krylon 5y ago
  Memory-safe languages do not guarantee absence of bugs or security issues. I remember when Heartbleed was discovered, and people were groaning how that would never have happened had OpenSSL been written in Rust, one of the OpenBSD developers wrote a blog post to explain how Rust's memory safety guarantees would not have prevented Heartbleed or a comparable bug. It would no doubt eliminate a lot of bugs, but security researchers and black hats would almost certainly continue to find exploitable bugs. It would just get somewhat harder. But AFAIU, finding exploitable bugs at the OS-level has been getting progressively harder for a long time, to the point that these days the main attack surface are web browsers and users who unquestioningly open file attachments in email. And people who eagerly add new features to operating systems will no doubt provide a generous supply of security problems to discover and "fix in the next release". ;-)

grishka 5y ago

Security patches are one thing. The constant moving things around and putting "fresh coats of paint" onto everything without solving any real problems is another. But then, at some point, all mainstream OSes will have been rewritten in memory-safe languages and there won't be any more exploitable vulnerabilities remaining. What then?

nippoo 5y ago
Not all vulnerabilities are due to memory safety!
- grishka 5y ago
  But most are, and they usually are most devastating. Mistakes in logic can sometimes be vulnerabilities that could, for example, disclose some information to an unintended party, but I can't imagine how one could lead to remote code execution.
jchw 5y ago
If that last bit truly isn’t a joke, let me be the first to point out that moving to memory-safe languages will not eliminate vulnerabilities. It won’t even really be close. Formal proofs of correctness would probably get us closer to software that is bug-free, but literally nothing is a panacea, and at the scale of attack surface we’re dealing with most of the extreme options (even including “rewrite it in a memory safe language”) are too big to foresee happening in the short or mid-term in any meaningful way.
krylon 5y ago
Memory-safe languages do not guarantee absence of bugs or security issues. I remember when Heartbleed was discovered, and people were groaning how that would never have happened had OpenSSL been written in Rust, one of the OpenBSD developers wrote a blog post to explain how Rust's memory safety guarantees would not have prevented Heartbleed or a comparable bug. It would no doubt eliminate a lot of bugs, but security researchers and black hats would almost certainly continue to find exploitable bugs. It would just get somewhat harder. But AFAIU, finding exploitable bugs at the OS-level has been getting progressively harder for a long time, to the point that these days the main attack surface are web browsers and users who unquestioningly open file attachments in email. And people who eagerly add new features to operating systems will no doubt provide a generous supply of security problems to discover and "fix in the next release". ;-)