Linus Torvalds, creator of Linux, has never suffered fools gladly. In particular, he unequivocally dislikes people who make improving confidence in Linux some-more difficulty than it needs to be. Most recently, in his possess incalculable style, he called some confidence developers “f*cking morons”. But, Torvalds, while mostly colorful, also gave instruction to confidence programmers.
It all started when Torvalds took Google Pixel developer Kees Cook, who had submitted a lift request that could have caused Linux heart panics, to task. Torvalds snarled, “Honestly, this is a kind of completely unsuitable ‘security person’ behavior that we had with a strange user entrance hardening too, and finished that most some-more unpleasant than it ever should have been. IT IS NOT ACCEPTABLE when confidence people set enchanting new rules, and afterwards make a heart panic when those new manners are violated.”
He reminded confidence programmers that “security problems are usually bugs.” And, that confidence hardening rags should never outcome “in murdering processes. The usually routine I’m meddlesome in is a _development_ process, where we find bugs and repair them.”
His position isn’t new. In 2008, Torvalds wrote, “To me, confidence is important. But it’s no reduction critical than all *else* that is also important!”
Torvalds isn’t a usually one who sees it that way. An eccentric confidence researcher, pronounced on a Linux Kernel Mailing List (LKML), “Some confidence people sneer during other confidence people’s mania with ‘security bugs’. The confidence attention is mostly spooky by anticipating (and selling/using/patching/reporting/showcasing/stockpiling/detecting/stealing) these ‘dangerous/useful’ accumulation of bugs. And this mania is ceaselessly over since bugs keep function — that is usually a inlet of program growth — and so this ‘security bug’ ardour continues.”
This is a vital reason we have unconstrained stories about a latest Android, Windows, whatever, confidence hole. Torvalds, for one, has prolonged been ill of this.
It’s not that confidence bugs aren’t genuine nor that they’re important. But, a sky is not descending scarcely as mostly as chicken-little confidence companies and developers would have we believe.
Torvalds explained to Donenfeld that from where he stands, “Security-first people tend to see a large win is when a [insecure] entrance is _stopped_. That’s the finish of a story from a confidence standpoint.“
“But,” Torvalds continued, “from a developer standpoint, things _really_ usually are not done. Not even close. From a developer standpoint, a bad entrance was usually a symptom, and it needs to be reported, and debugged, and fixed, so that a bug indeed gets corrected. So, from a developer standpoint, a finish indicate of hardening is usually a starting point, and when _you_ consider you’re done, we’re unequivocally usually removing started.”
So how should confidence developers proceed Torvalds when they contention a patch?
Well-known confidence and Linux developer Matthew Garrett acted that doubt to Torvalds on a LKML. True, Garrett observed, “Kees schooled from that experience and combined a default fallback in response to it. Let’s prerogative people for training from past problems rather than screaming during them.” After all, Garrett continued, “The series of people peaceful to work on confidence things is singular adequate for several reasons, let’s try to keep reason of a ones we have!”
Torvalds certified he had been busy and he apologized. “Sorry for a clever words.“
So what do confidence developers do when operative on Linux heart code? Here are Torvalds’ manners for confidence developers:
- When adding hardening features, we as a confidence chairman should always see that hardening to be a _endpoint_, though not a evident goal.
- When adding hardening features, a initial step should *ALWAYS* be “just news it”. Not murdering things, not even interlude a access. Report it. Nothing else.
- “Do no harm” should be your mantra for any new hardening work.
If programmers hang to these rules, Torvalds will swear reduction and fewer bugs will make it into a Linux kernel.
- Linus Torvalds: ‘I don’t trust confidence people to do lucid things’
- Linux creator Linus Torvalds: This is what drives me nuts about IT security
- Linus Torvalds says targeted fuzzing is improving Linux security