The DMZ

« first day (4156 days earlier) ← previous day next day → last day (1022 days later) »

forest

00:45

gruss.cc

This guy is great.

He's one of the authors of gruss.cc/files/rambleed.pdf, which showed that even ECC memory correcting rowhammer bit flips couldn't protect from a confidentiality-only-violating variant of rowhammer.

> Furthermore, in contrast to Rowhammer write side channels,
which require persistent bit flips, our read channel succeeds even
when ECC memory detects and corrects every bit flip. Thus,
we demonstrate the first security implication of successfully-
corrected bit flips, which were previously considered benign.

Although, thinking about it, I imagine using SECDED ECC with the system set to panic on MCE (or optionally to also panic on a corrected error) would largely mitigate such risks. Assuming aggressive patrol scrub intervals...

forest

01:48

-3

Q: Does anyone in the world care about people

I been trying to get help understanding this whole sex games, data hacking bullshit, that has allowed my life to be destroyed. They knew things about me from my childhood, they used everything against me to mess me up and I asked them for help almost every day for a year, I’ve asked everyone I co...

haha what

belkarx

02:43

@forest It's just facial recognition, and appears to be locally stored. There's a disclosure that it shouldn't be a primary password. What's am I missing that makes it "terrifying"?

forest

@belkarx The fact that no one will heed that advice and will use it as the sole authentication anyway.

belkarx

Fair. Though I do feel Linux users tend to be a bit more security conscious as a whole

forest

Eh, you'd be surprised.

But maybe I'm just too pessimistic.

belkarx

And maybe I'm not cynical or knowledgeable enough

Pessimism is realism, honestly

forest

It might be an interesting base on which to design a secure "panic mode", though. Have the system check your face every 5 seconds and if it changes (e.g. because someone broke into your house and you got up), the system shuts down.

belkarx

02:48

Might be too many false positives due to environmental changes there for that to be viable. Depends on how good the model is ig

forest

It could check every one second and if it fails 5 times in a row, it could trigger the event.

Although detection of user inattention could be done in other ways as well, I suppose.

Very niche threat model, though.

belkarx

I agree. Interesting concept though

forest

I designed something similar, albeit based on distance instead of facial recognition.

And much more primitive.

belkarx

With IR?

forest

03:07

@belkarx Actually ultrasonic, which kinda sucked. Was mostly just a project for fun.

So there was a lot of room for optimization that I missed.

forest

03:18

krebsonsecurity.com/2022/03/…

Very interesting. Linked to me in another chat room.

belkarx

@forest Posted this here a while ago :) chat.stackexchange.com/transcript/151?m=60818450#60818450

forest

Hah, must have missed it.

belkarx

It's social engineering at its finest. Brilliant idea

forest

Also see wyden.senate.gov/news/press-releases/… (Wyden's response, linked in that article)

Unrelated, but Clive Robinson, as usual, understands what's really going on re. Huawei: schneier.com/blog/archives/2022/04/…

belkarx

@forest Nice, requiring signing was also what I first thought of when I saw the article; I hope the bill gets passed at some point

forest

03:29

Hope so too. Although it will be a loss of a very interesting... social engineering exploit primitive?

As I said elsewhere, Wyden is probably the only politician ever to live who knows what a digital signature is, much less use it in a cogent sentence.

belkarx

Alas. I suppose anyone else smart/involved enough to know what they are knows being a politician is not worthwhile

forest

Or smart and involved, but corrupt.

belkarx

Certainly a possibility

forest

We can't rely on the government to make good decisions regarding tech, of course.

Just see that repulsive EARN IT act... Basically bans end-to-end encryption, and will probably pass.

belkarx

Seriously?

forest

03:34

Yes.

All in the name of "protect the children".

belkarx

Riii-ight. Because the bad guys won't find a way around this

I love these names ngl. Wonder who comes up with them
\>Eliminating Abusive and Rampant Neglect of Interactive Technologies Act

forest

Tl;dr if social media can't find a magic way to break encryption and let someone post kiddie porn, they can be sued for a million bucks each. Since they can't break encryption, it means they can't use encryption.

I mean ffs, anyone who uses Facebook for kiddie porn is going to get caught anyway.

No matter whether or not the chats are e2e.

EU is also trying to ban end-to-end encryption: eff.org/deeplinks/2022/03/…

belkarx

@forest Yup, can't the govt just request the data from the companies ... how does e2ee play into this

forest

@belkarx The idea is that people can use chat which is e2e.

belkarx

Whyyyy would you chat about your illegal doings on a social media site

If you do, you deserve to be caught anyway haha

forest

03:39

People are idiots. Pedos are no exception.

This is just a cheap shot by the feds to try to avoid having to do actual *gasp* investigation!

belkarx

Understandable

forest

It's not like idiot pedos using social media for their porn have stellar opsec.

belkarx

Sigh. People.

forest

As much as I hate feds, I'm all for them doing traditional investigations into criminals.

belkarx

Depends on the criminals. Opinions on Snowden?

forest

03:41

Oh I'm usually on the side of the criminals, don't get me wrong.

But at least I can understand the feds doing what they do, as long as they do it legitimately.

But when they try to screw everyone over, that makes my blood boil.

belkarx

I mean ... it's not just them that benefit from lack of e2ee I presume. There have got to be others pushing this

forest

The IC in general.

So not just American feds but the international intelligence community.

Considering they hate everyone who likes privacy. Fun fact: The IETF is one of the biggest thorns in the side of the IC!

belkarx

Noted. Sounds like a great org

forest

Well, they're a big org. I wouldn't say they're great, but they're the only ones with serious influence and they aren't big fans of surveillance. They're the ones who manage RFCs and basically control internet standards.

belkarx

What's the difference between them and the IEEE

I was under the impression that the latter dealt with RFCs

forest

03:48

IEEE is electrical engineering.

IEEE is basically OSI layer 1 and 2 vs IETF which is the higher OSI layers.

belkarx

Oh I see

forest

IETF does things like HTTP/2 and TLS. IEEE does stuff like 802.11 (Wi-Fi). But unlike IETF, IEEE also deals with stuff that isn't networking-related (like standards for data buses and integrated circuits).

9 hours later…

RoraΖ

12:54

Joining IEEE can be a big benefit. A lot of research papers get published there that are really interesting. Might even get your employer to pay for the membership if you work in a related field.

I think it also provides discounts for certifications, it's been a while since I looked into it.

ThoriumBR

13:47

@forest I used BlueProximity for years when working in the office, so if I had to walk away from my computer, it would lock before any coworker had any time to do anything funny with my computer...

7 hours later…

forest

20:56

A: How can I protect my home/office computer against row hammer?

The only reliable hardware mitigation that does not involve downgrading to a motherboard with DDR2 is to use memory that supports TRR (Target Row Refresh), which is optional for LPDDR4 (not the same as DDR4). Unfortunately, many DDR4 modules do not support TRR, and there is often no easy way to t...

Regarding this answer that I wrote a few years ago, I realized yesterday that you can't panic on a correctable ECC error without a kernel patch. I haven't tested it, but I think this should work for kernel 5.17.5:

diff --git a/drivers/edac/edac_device.c b/drivers/edac/edac_device.c
index 09d19b0..ac33b51 100644
--- a/drivers/edac/edac_device.c
+++ b/drivers/edac/edac_device.c
@@ -554,6 +554,12 @@ static inline int edac_device_get_panic_on_ue(struct edac_device_ctl_info
 	return edac_dev->panic_on_ue;
 }

+static inline int edac_device_get_panic_on_ce(struct edac_device_ctl_info
+					*edac_dev)
+{
+	return edac_dev->panic_on_ce;
+}
+
 void edac_device_handle_ce_count(struct edac_device_ctl_info *edac_dev,

Posting here just in case anyone wants to further harden their ECC system against rowhammer.

ThoriumBR

21:09

interesting

forest

The patch would be made most effective in combination with a reduced patrol scrubbing interval and a low poll_msec value.

Assuming the system's BIOS allows performing patrol scrubbing irrespective of current system or memory bus load.

It should also help against RAMBleed (lol those names) which succeeds even if the bitflips are corrected.

« first day (4156 days earlier) ← previous day next day → last day (1022 days later) »

all rooms