Brace yourselves: USB bugs are coming!
Quote Tweet
Slides for my "Coverage-guided USB Fuzzing with Syzkaller" talk @offensive_con docs.google.com/presentation/d
1
8
46
Conversation
Replying to
I applied 's deny_new_usb patch around the time it appeared. Syzkaller keeps vindicating that choice. 👍
1
1
3
I didn't make the deny_new_usb feature. I developed the Android integration for it including integration into Keyguard for automatically enabling it when the screen is locked. It existed as a kernel feature for many years prior but hadn't been properly integrated as intended.
3
1
2
I picked up the patch from the Linux Hardened tree you were maintaining a while back (thanks!), so the patch I use has your name as the git author. What was the original provenance? PaX/grsec?
1
The commit states the origin (grsecurity). Being the author of a commit doesn't imply being the original author of the code. The origin of the patch wasn't published as a Git repository and even if it was there wouldn't necessarily be a usable commit to directly extract that way.
1
It also wouldn't be right to claim someone else as the author for commits which have been changed from the original code. You need to look at Git commit messages for authorship information including an edit history of the commit leading up to the final version that was applied.
1
I was going to work on splitting up / porting more PaX and grsecurity features but didn't want to deal with the hostility and fighting anymore. I also had an in-progress hardened slab allocator similar to github.com/AndroidHardeni in userspace which is stalled for similar reasons.
2
I stopped working on that as a whole due to a mix of the terrible attitude of the upstream maintainers / developers / leadership combined with people trying to prevent their derivative works from being landed upstream. Some people like are willing to deal with that.
1
1
Yes, both upstream and PaX / grsecurity along with other people with out-of-tree security work. The worst case with an out-of-tree security project doesn't even involve PaX and grsecurity. I wasn't willing to deal with all that drama, dishonesty and endless fighting anymore.
1
1
I prefer working on projects where things are much more cooperative, civil and constructive. It's also nice working on projects where security isn't such a hopeless cause like the Linux kernel and where both the security engineers and other developers have less tunnel vision.