Twitter | Search | |
Bryan Steele
AKA brynet@. I like tinkering with , occasionally other Unix-like systems. In other words, not a MCP. Infosec? DMs open. He/Him. Tutu.🍕🇨🇦
9,492
Tweets
1,050
Following
1,467
Followers
Tweets
Bryan Steele 1h
🤔 🍕💻🥺🤞
Reply Retweet Like
Bryan Steele retweeted
NaN 10h
A lot of malloc implementations hold on to memory pages received from the kernel. 's is reasonably aggressive with unmapping pages that are unused, catching use-after-free bugs with segmentation violations.
Reply Retweet Like
Bryan Steele retweeted
OpenBSD Journal 9h
Microsoft goes Gold for 2018!
Reply Retweet Like
Bryan Steele retweeted
OpenBSD src Changes 19h
kn@ modified sbin/ifconfig/ifconfig.c: Warn on deprecated 'vlan' and 'vlandev' option usage These were superseeded by 'vnetid' and 'parent' in june 2017 and will be removed in the future. "Looks right" deraadt, OK benno
Reply Retweet Like
Bryan Steele 17h
Replying to @LuigiThirty
Was that one not well liked? I don't change the channel when it's on.. ☺️
Reply Retweet Like
Bryan Steele 19h
No, sorry. It doesn't appear to be out yet? might be interested though, given the eMMC storage. 😁
Reply Retweet Like
Bryan Steele 20h
Thanks for the b-day treats, ! ☺️🐱
Reply Retweet Like
Bryan Steele retweeted
Daniel Gruss Nov 13
We systematically analyzed , and and came across interesting new transient execution attacks: Great collaboration with , Frank Piessens :)
Reply Retweet Like
Bryan Steele retweeted
NaN 21h
's stores meta-data separated from the memory returned to the caller, making it harder to use heap overflows to achieve code injection and execution. See e.g. and references. Having malloc meta-data near program data is dangerous!
Reply Retweet Like
Bryan Steele 23h
I came back from lunch subscribed to Tadpole computer, and NT-on-dinos facts. 😉
Reply Retweet Like
Bryan Steele Nov 14
Reminds me of 's very modest request for ☺️
Reply Retweet Like
Bryan Steele Nov 14
Replying to @HWadrr
🥳
Reply Retweet Like
Bryan Steele retweeted
NaN Nov 13
's randomizes allocations all over the address space. It makes attacks harder because each run has a different memory layout, catches bugs (allocations are surrounded by unmapped memory) and allows realloc to grow an allocation without copying in most cases.
Reply Retweet Like
Bryan Steele Nov 14
Reply Retweet Like
Bryan Steele Nov 13
Amazing hardware/driver debug work by to fix a laptop speaker. ☺️
Reply Retweet Like
Bryan Steele retweeted
joshua stein Nov 13
I recently used VFIO on Linux to get stereo sound working on the Dolby Atmos sound chip on my OpenBSD laptop
Reply Retweet Like
Bryan Steele retweeted
NaN Nov 13
's does not touch the pages of an allocation by default. Allocating 100kB is roughly as fast as 1000kB. But *using* the newly allocated memory takes time because of page faults.This is also the reason why option J has a big performance impact.
Reply Retweet Like
Bryan Steele Nov 13
Reply Retweet Like
Bryan Steele Nov 12
Replying to @__dotblake
That doesn't look particularly relevant to what I'm working on, not doing anything with VoIP. My personal opinion is that pcap parsers should be privsep and pledged. There's a lot of buzzwords on that page that concern me. ☺️
Reply Retweet Like
Bryan Steele Nov 12
Replying to @shirkdog
Yep! The packet parser process already ran as an unpriv user, and tightly pledged. The change here is the privsep monitor process, stay tuned! 😉 tcpdump(8) is already unveiled, thanks to mestre@! That made it into 6.4.
Reply Retweet Like