Skip to main content

in reply to daniel:// stenberg://

"Typical server latency is likely "slow" enough to trigger this bug"

Nitpicking - but if it's the typical case it sounds a bit strange to refer to it as slow? :)

in reply to daniel:// stenberg://

You are only human. But you're a pretty good human.

Thanks for the clear and detailed write up.

in reply to daniel:// stenberg://

refactoring is always risky. Kinda wonder what an exploit scenario would look like though; it has to be a malicious server, the user has to be intentionally trying to reach it via SOCKS, and they must have some kind of ongoing service relationship for the server to have fingerprinted the client well enough to push an effective payload.
in reply to Howard Chu @ Symas

@hyc yeah. Perhaps most realistically, a Tor user (which normally uses SOCKS5) going to a HTTPS site that has been breached or similar
in reply to daniel:// stenberg://

I wonder if the upcoming bounds checked c rfc features from Apple in clang would be helpful at the very least as a defensive measure. but they also seem to be moving fairly slowly :(
in reply to daniel:// stenberg://

Great write up!

There is still something fundamentally wrong that a single person should be (read: feel) accountable for something like this. People make mistakes and there is nothing to feel bad about.

in reply to daniel:// stenberg://

speaking from personal experience, it's okay to make mistakes: github.com/fwupd/fwupd/commit/… :)

Well done on the handling -- and please don't take any of the criticism personally; you're doing a great job.

in reply to daniel:// stenberg://

The hackerone report behind it: hackerone.com/reports/2187833
in reply to daniel:// stenberg://

“In hindsight, shipping a heap overflow in code installed in over twenty billion instances is not an experience I would recommend.”

Wise words indeed