The DMZ

Terry Chia

01:19

Nice, a 64-bit processor in the new iPhone?

Xander

@tylerl Ugh. What a hassle.

Gilles

@TerryChia how much RAM?

Terry Chia

@Gilles No mention of that like usual. I'd be surprised if it's more than 1GB.

I see it as more of a transition thing.

Gilles

01:57

@TerryChia I see it as “F!RST”

Terry Chia

@Gilles Maybe... but considering how Apple doesn't usually focus on hardware specs, that will surprise me.

Perhaps the iPad in a generation or two will have > 4GB of RAM but I don't see the point of that in a phone.

Simon

@TerryChia I feel like programming the generator he wants.

Terry Chia

@Simon Do it! Shouldn't be that difficult...

Simon

It shouldn't be too hard, you keep in a DB which character should be replaced with which one.

Then you go through the word and generate the possible entries. However, the algorithm to generate every single possibilities might be harder to figure out.

Terry Chia

I'm actually surprised he needs it. AFAIK password crackers like john already mangles the dictionaries you pass in.

Simon

02:14

I'll think about it tomorrow when I'm not half asleep and I might start something later.

Simon

02:25

Free flag/downvote: security.stackexchange.com/questions/42202/…

Terry Chia

02:54

@Simon If you are bored, I got a few projects you could contribute to. :P

Iszi

03:57

Aside from just being weird, does anyone else feel like this is fairly off-topic?

0

Q: Using only private non-government resources, how can something be sent such that the NSA will see it?

Supposing you want the NSA to see something, what does one have to do to guarantee it. This means in their normal data collection it will be swept up, not in a targeted probe of an individual or with (or without) a subpoena. The restrictions are as follows: No use of government resources (can...

ManishEarth

04:18

I posted a question on the main site too, but this is more discussiony: What do you guys think? Should browsers block passive content loaded over HTTP on HTTPS pages? (Mixed passive content)

The context is bugzilla.mozilla.org/show_bug.cgi?id=906069

Stack Exchange

'At least humans are better at quietly amusing ourselves, oblivious to our pending obsolescence' thought the human, as a nearby Dell Inspiron contentedly displayed the same bouncing geometric shape screensaver it had been running for years.

Terry Chia

04:59

@ManishEarth Interesting...

Never really thought about passive content before.

TildalWave

@ManishEarth why is that link you include linking to Optics Assignments.rar file? :O

Anyway, I've replaced it... but do note I've seen the link you posted here in the DMZ only now, AFTER I posted my answer. :P (I'll prolly read your discussion a bit later, maybe edit a few things in my answer, if that's necessary)

And if that wrong link is sensitive, you might wanna flag your question and ask mods to purge history, or whatever needs to be done to delete revisions.

I never had to do it myself, but I imagine mods need to request it in TL?

Terry Chia

And when I asked about this on Google+ (didn't get an answer), it automatically tagged my post with #Cat and #Caturday — Matt 2 hours ago

Now I understand why G+ is awesome.

AviD

05:54

@TerryChia no, this is why:

On one hand, you'll never be able to convince your parents to switch. On the other hand, you'll never be able to convince your parents to switch!

xkcd has the answer to everything.

Lucas Kauffman

morning

Terry Chia

06:33

@LucasKauffman Morning

@AviD That's very true.

Rory McCune

07:48

mornin' all

Lucas Kauffman

sup @RoryMcCune

Rory McCune

not much just report writing (finishing big mobile assignment this week :) )

rest of the month is already mapped out as a web app review for a bank and a big external for a public body

life of the jobbing pen tester :)

Lucas Kauffman

08:14

@RoryMcCune good that you got a lot of engagements then ^^

ManishEarth

08:25

@TildalWave Oopsies. Nope, not sensitive

thanks

TildalWave

@ManishEarth No problem ;) BTW I've borrowed a bit of wording from one of the Firefox articles on this same stuff, but added my own points to it too. I'm not too sure I chose the best approach tho, but I'm too tired to go through it again now. Maybe later, after I sleep a bit too LOL

ManishEarth

@TildalWave Feel free to chime in on the bugzilla thread too :)

The guy is a bit aggressive and keeps screaming that Mozilla doesn't understand security

Well, nobody does, but he's just being imperative without giving reasons why things should be that way

And I'm no security expert, so I can't refute things clearly :s

Terry Chia

@ManishEarth You should make an offering to the gods and sic @ThomasPornin on him.

ManishEarth

@TerryChia What does TP like?

Do we feed him rainbow tables?

Terry Chia

@ManishEarth Salmon and math.

Lucas Kauffman

08:34

@ManishEarth salmon

ManishEarth

ah

George Salmon

George Salmon (25 September 1819 – 22 January 1904) was, firstly, a mathematician whose publications in algebraic geometry were widely read in the second half of the 19th century. He was also an Anglican theologian who devoted himself mostly to theology for the last forty years of his life. His publications in theology were widely read, too. He spent his entire career at Trinity College Dublin. Personal life Salmon was almost certainly born in Dublin, to Michael Salmon and Helen Weekes (the daughter of the Reverend Edward Weekes), but he spent his boyhood in Cork City, Ireland, where hi...

What about a mathematician named Salmon?

@TerryChia There's also the chance that that guy's right and I'm wrong , and then Thomas will maul me :S

Terry Chia

@ManishEarth It's a risk you have to take when invoking the power of The Bear.

TildalWave

@ManishEarth This is a bit funny attitude actually, considering all this is better addressed in Firefox than currently in any other browser that I know of. :| And with some latest updates, these settings were even exposed more for users to notice them easier. What does he expect? That they'll decide what's best for users?

ManishEarth

@TildalWave Yeah

Though I personally disagree with the color of the mixed content symbol and the other HTTPS things

Because IE and Chrome use bright colors. Noticeable. FF uses symbols with I think less color

But that's just me (I'm more used to chrome)

TildalWave

@ManishEarth If that was blocked by default, then Mozilla would be hit by all the rage their users are capable of, with most of the pages displaying improperly. Including SE

where HTTPS is enabled, Q&A still doesn't have it

ManishEarth

08:38

@TildalWave Yep.

TildalWave

@ManishEarth Well me too, but I miss HTTPS Everywhere working as good as it does in Firefox tho

ManishEarth

He said that browsers should always take the more secure route. I said that by that logic browsers should just block all HTTP

TildalWave

@ManishEarth Hehe true. And accept only httponly and secure tagged cookies, and block third-party contents, and make sure you're not accessing through or to a blacklisted IP, and ad nauseum until you break the bleepin' internet.

ManishEarth

@TildalWave At the bottom of my comment I linked him to today's Abstruse Goose. Very pertinent in this case

TildalWave

They should also include a plugin that disconnects your NIC

ManishEarth

08:45

Alt text: If you simply drop any packets that have a destination address, you should be alright.

@TildalWave ha

TildalWave

@ManishEarth just tell the guy that what he's after is called "offline browsers"

ManishEarth

lol

Terry Chia

09:00

Huh, I got another answer on the supercollider.

It's not really a reptrain though, probably just a slow day.

Rory McCune

11:47

this looks interesting indiegogo.com/projects/…

linux/android dual-boot 10" tablet at $249..

Terry Chia

@RoryMcCune Those specs strike me as "too good to be true".

That's like 2nd gen Nexus 7 specs.

Rory McCune

@TerryChia mm possibly apparently they've shipped stuff before, so you'd hope it's not complete vapourware...

that said I've never heard of 'em up until now.

Terry Chia

@RoryMcCune I'm still calling bullshit. That screen is iPad level.

The processor and RAM as good as a Nexus 10.

And they are selling it cheaper?

Terry Chia

12:06

Salting has nothing to do with the cracking time of a single password, it only prevents running attacks in parallel. — Brendan Long 16 hours ago

lolwut?

Rory McCune

well if you don't buy one you'll never know :) I'm moving house next month, so aff gadgets till after then...

Terry Chia

@RoryMcCune I don't have that kind of disposable income. We need to find someone who does. :P

Simon

12:31

@TerryChia Well, I have plenty of stuff that I can do but I'm so not motivated to do any >.<

Terry Chia

@Simon Hehe.

I'm doing some stuff with the Dropbox API now. Pretty fun stuff.

Lucas Kauffman

@TerryChia TWSS

Simon

Automatically uploading all the porn pictures to your dropbox whenever you browse those websites?

Terry Chia

@LucasKauffman That's bad.

Simon

Clever, clever.

Terry Chia

12:33

@Simon No, I'm afraid you will sue me.

I'm steering clear of all porn-based ideas now.

Scott Pack

In Singapore? How is that possible?

Lucas Kauffman

@ScottPack They will kill TerryChia

Scott Pack

I'm sure they would start with caning.

Lucas Kauffman

@ScottPack or worse, they make him cut up Durian for a day :p

Scott Pack

@LucasKauffman They can't, it's banned in Singapore.

Terry Chia

12:39

@LucasKauffman How is that worse?

@ScottPack Wut?

Lucas Kauffman

@TerryChia You will smell bad for the rest of your life

Scott Pack

Oh, only on mass transit. My bad.

Lucas Kauffman

@ScottPack it's allowed, just not on public transport

Terry Chia

@ScottPack That's like the unofficial national fruit of Singapore...

Scott Pack

Nice

> The edible flesh emits a distinctive odour that is strong and penetrating even when the husk is intact. Some people regard the durian as pleasantly fragrant; others find the aroma overpowering and revolting. The smell evokes reactions from deep appreciation to intense disgust, and has been described variously as almonds, rotten onions, turpentine, and raw sewage. The persistence of its odour has led to the fruit's banishment from certain hotels and public transportation in southeast Asia.

Here we go

No fine listed for durian. Must be immediate execution.

Terry Chia

12:42

@LucasKauffman or anyone that knows python.

1

Q: How to delete a file from the filesystem based on the returned case insensitive delta?

I am attempting to write a code to fulfill this particular case documented in the Dropbox Core API python SDK. [path, nil]: Indicates that there is no file/folder at the path on Dropbox. To update your local state to match, delete whatever is at path, including any children (you will sometime...

python dropbox dropbox-api

Any ideas for this one?

Repcapped for the second day in a row. Huzzah!

Simon

Brag about it.

Terry Chia

Hmm, I can actually get 10 more rep.

Simon

Bragging and lying.

Lucas Kauffman

@Simon you need to answer more questions if you want more rep

Terry Chia

@Simon Well, I'm at 205 but 15 rep came from an accept.

Simon

12:45

Or I can simply not do it and keep complaining about my low rep.

Terry Chia

@Simon You are such a whiny kid. :P

Simon

Well duh, I'm 13.

Thomas Pornin

@TerryChia Well, it is true... for the right notion of "parallel", of course. Rainbow tables are a parallel attack, if you apply the adequate change of coordinates on space-time.

Terry Chia

@ThomasPornin heh

Damn, this is pretty sweet deal. arstechnica.com/apple/2013/09/…

Thomas Pornin

I have exactly 61000 rep difference between my two accounts.

Simon

12:56

gz but low

Terry Chia

@ThomasPornin This month will see you hitting two major milestones. 100k for Big Bear and 3rd place for Small Bear.

Thomas Pornin

@TerryChia Yeah, but @Gilles told me that there was a free T-shirt when reaching 100k. Now that this milestone is not completely meaningless, I find that it lacks elegance.

Simon

Is there a "Hall of fame" of rep?

Terry Chia

@Simon You wouldn't know.

Simon

WELL DUH, that's why I'm asking.

Silly Terry.

Terry Chia

12:59

@ThomasPornin Man, that's pretty cheap of SE. They should at least ship some salmon to you.

Thomas Pornin

@Simon There are the leagues: stackexchange.com/leagues

@TerryChia When they distributed stuff upon site graduation from beta, Bear #2 suggested that they send socks, not T-shirts, but they did not get the joke.

Terry Chia

@ThomasPornin Hahahaha!

Simon

@ThomasPornin Yep, that's what I was looking for, thanks.

Terry Chia

I'm still waiting for a redesign of the site logo.

@Simon Meh, you can get the same information on the /users page.

The leagues page lags behind by about a day I think.

Simon

Unacceptable.

Scott Pack

13:10

@Simon So's your face.

Thomas Pornin

Well, bummer. Google considers that my "network" may be sending automated queries and refuses to talk to me.

Simon

@ScottPack Rude.

@ThomasPornin Google is also rude.

Thomas Pornin

Their loss, I say. I'll do all my googling from memory, then.

Simon

Alright guys, who wants to pay me for the information?

TildalWave

Woot I just saw like 40 questions with reopen votes... every second a few less... what's going on?

Simon

13:12

I made it too obvious. There's 32 reopen votes.

Terry Chia

Blah, I only grabbed 3.

@ThomasPornin You can always Bing that shit.

TildalWave

Where did they come from? Who's trigger happy?

Scott Pack

I only got through 3 before the queue emptied

Terry Chia

@TildalWave I think someone mentioned it was a change by SE that applied to questions retroactively.

Scott Pack

And i only got credit for 1. :(

Simon

13:15

Again, nobody's thanking me.

Terry Chia

@ScottPack Stat page shows 3 for you.

Simon

I really gotta figure out a way to make you guys pay.

Terry Chia

@Simon Well, @TildalWave mentioned the key detail before you did. :P

Simon

@TerryChia NO.

Scott Pack

@TerryChia Still only showing 1 to me. I blame the devs.

Terry Chia

13:17

I think they applied a hard cap of 3 reviews.

TildalWave

@TerryChia Interesting. You have any link? Because I don't see that in effect on SEx.SE.... yet

Terry Chia

@TildalWave Nope, someone mentioned it in chat yesterday I think.

Can't remember who.

Simon

@TerryChia Your mom.

TildalWave

@TerryChia That's what she said

Simon

@TildalWave We teamed up on him.

Terry Chia

13:21

@Simon @TildalWave WEAK!

Scott Pack

@Simon You tag teamed Terry's mom?

2

TildalWave

LOL

Simon

@ScottPack I've been single for 3 days now, I'm not that type of man.

Scott Pack

►

Oh yeah. NSFW

Lucas Kauffman

@ScottPack I had someone make me open the bubblebut video by Major Lazer today.

AT WORK

Scott Pack

13:31

Fine. So NSFsW

My favorite features are the undocumented ones where, when I ask my tech product manager for advice he says, "Yeah, the documentation blows. We get a lot of complaints about it, it's right here."

TildalWave

13:50

made 160 points on space already today... time to hit the pillows. t/c all!

Scott Pack

Ew.

ManishEarth

Incoming

1

Q: Should I know my users passwords so I can check they can logon?

I have an issue with support for a system and I don't know what approach to take. It boils down to the question: Should I know my users passwords so that I can check with certainty that a particular user can logon and there is not a problem with their account? If I can view my users pa...

security passwords support

Any objections?

Iszi

@ManishEarth Nope

And the answer is: "No. Your application should be smart enough to be able to tell the difference between an account lockout and bad password."

ManishEarth

Yep

Terry Chia

@ManishEarth Thanks for the warning. :P

ManishEarth

13:59

Ha :)

Rory Alsop

@ThomasPornin there is a t-shirt - why was I not told???

Terry Chia

Annnnnnnd rep capped. (For real this time).

Thomas Pornin

@RoryAlsop I don't have confirmation (yet). Right now this is, from my point of view, a rumour.

Terry Chia

High chance I'll be passing @AJHenderson in rep tomorrow. :)

Thomas Pornin

But if the T-shirt is real, I'll be sure to tell you.

I mean, I'll be sure to gloat in the DMZ.

Rory Alsop

14:04

@ThomasPornin well, sadly it isn't total site-wide rep, otherwise I'd already have one :-(

Simon

Can we expect a picture of you wearing it?

Rory Alsop

@ThomasPornin oh, of course

ManishEarth

@ThomasPornin They don't have t shirts in Bear size

Rory Alsop

@Simon and a salmon

Terry Chia

We should have a Bear Day to celebrate this momentous occasion.

Simon

14:05

@RoryAlsop That'd be awesome.

Rory Alsop

@Terry - best be careful... your avatar is the only thing in the way of a Bear's full house on stackexchange.com/leagues

Terry Chia

@RoryAlsop Oh dear....

Thomas Pornin

@RoryAlsop This happens occasionally. Rep trains tend to wreck weekly stats. I just have to wait for next week to see the situation go back to normal.

Rory Alsop

:-)

Thomas Pornin

Unless @TerryChia decides to make more lasting efforts, an event which I could only applaud to.

Active users who write good answers: we cannot have too many of them.

4

ManishEarth

14:07

@ThomasPornin Normal as in last week normal or normal as in before-Terry-existed normal? :p

Terry Chia

@ThomasPornin Nah, I usually hit 2-3 reptrains in a row then go on a long hiatus.

CodesInChaos

You're falling behind on crypto. Soon poncho will overtake you in the all time stats

AJ Henderson

@TerryChia Noooooooo!!!

grats

yeah, Photography and AVP have been taking too much of my time to get answers in quick here and I don't like posting follow up answers if there is already a good answer

Terry Chia

@ThomasPornin Plus my reptraining answers certainly aren't good. Hell, most of them does not break the 100 word mark.

ManishEarth

Oh, now that there are more people here, what do you think of bugzilla.mozilla.org/show_bug.cgi?id=906069 ? Should FireFox block mixed passive content on HTTPS by default?

CodesInChaos

14:09

IMO being able to embed http images in a https site without too many issues is essential

else you can't practically migrate forums etc. to https

Rory Alsop

@AJHenderson it's okay - if you leave AVP alone for a wee bit I might get some answers in and get my rep up :-)

Terry Chia

@ManishEarth What @Code said.

CodesInChaos

I'll take the risk of goatse

AJ Henderson

@RoryAlsop yeah, but I just got the kitty full house

Rory Alsop

@AJHenderson lol

AJ Henderson

14:11

though I suppose 5x anyone else's rep gain is probably unnecessarily overkill

ManishEarth

@CodesInChaos Amended question, by default?

Rory Alsop

my 5k is not far off - that's my net target

@AJHenderson tell that to the bears!

Terry Chia

@ManishEarth Well, that's the most optimal user experience.

CodesInChaos

at last in a way that the website can opt-in, not the user

Terry Chia

Think about it, what's the worse that could happen? Some images get MITM-ed and replaced?

ManishEarth

14:12

@CodesInChaos The most optimal UX is no warnings whatsoever :p

Terry Chia

No lasting harm, important content should be served over HTTPS anyway.

ManishEarth

@TerryChia Well, the HTTP images have full URLs and there's a chance that they can be tracked back to the HTTPS site visited

I find this over paranoid though

CodesInChaos

mostly replacing images. There are some minor risks regarding referer, IP leaks and XSRF. But those can be mitigated.

Terry Chia

@ManishEarth Sorry, a bit sleepy. I can't seem to parse this sentence.

CodesInChaos

@ManishEarth one can do similar attacks by using a unique domain, which will leak through SSL

ManishEarth

14:14

@TerryChia I visit abc.com/a.htm, which contains b.com/b.jpg. On abc.com, b.jpg is only used on a.htm. Normally the attacker would just know the domain, nothing more. Now he can guess the full path

CodesInChaos

but if somebody embeds unique.something.com tracking works even with SSL

or use an attacker controlled server

ManishEarth

@CodesInChaos oh yeah

Terry Chia

@ManishEarth So yeah, I really think it's not a big deal if it's not blocking. That bug report is rather stupid.

ManishEarth

@TerryChia I find it stupid. But I find myself stupid, so I thought I was missing something security-related

Rod MacPherson

Does anyone know what might port scan it's local network for UDP 2054 ? I currently have 2 separate Windows 7 machines on our guest WiFi that are doing this, I haven't seen that before and googling 2054 UDP doesn't help.

Terry Chia

14:23

@ManishEarth Your point about Vista is appropriate. If you throw too many warnings at the user he is gonna ignore it anyway. There was a UX question about it on the supercollider yesterday.

ManishEarth

@TerryChia yeah, saw that

Terry Chia

@RodMacPherson You want to find out what process is listening on UDP 2054 on your machines?

ManishEarth

doesn't netstat do that?

Scott Pack

netstat -panu | grep 2054

Terry Chia

@ManishEarth Yeah... that's why I wanted him to clarify it. Seems a bit basic. :P

0

Q: How could `iptables -F` lock me out of the server?

I was just configuring IPtables and whilst doing so, had the default chain policies set to ACCEPT, but when happy with the generated rules I changed that to DROP for all chains. I did not run iptables-save at any time. After do this, I still had good access, but noticed that one of the rules wa...

linux webserver iptables

U&L/SU?

ManishEarth

14:26

yes

Simon

Pretty sure.

Terry Chia

@ManishEarth Well, which? :P

Rod MacPherson

@TerryChia If only they were my machines. That would make it easy. They are not my machines, they are on a public access point at a marina that I manage the WiFi for. I don't know who's they are, but if it is an infection I'd like to notify the marina staff who do know who the users are.

ManishEarth

@TerryChia all of them :P

Terry Chia

@RodMacPherson Ahh, so you want to identify the remote service. That's a little tricky...

Are the machines just listening on the port or actively communicating on it?

Rod MacPherson

14:29

I suppose if it continues I'll packet capture and see what that can reveal, but I was wondering if anyone here might have seen this behavior before.

Terry Chia

You could sniff the traffic and try to deduce the process, but it can pretty much be anything.

Rod MacPherson

They are sending out packets in a scan fashion, same port one IP after another

Like they are looking for a service

Terry Chia

@RodMacPherson Ahhh, so they are scanning for UDP 2054?

Rod MacPherson

yes

Terry Chia

Huh, that's a weird one.

What sort of packets? SYN?

ManishEarth

14:36

Google says it's for "weblogin"

Iszi

15:16

Is this really us, or should it go to User Experience or somewhere else? Or is it just not an SE question at all?

0

Q: Data Cleaning in User Profiles

I have no idea if this is the right stack exchange group for this question, but at work we recently adopted a new name, and that led to a new website. This means that users went from having [email protected] to [email protected]. During the rebranding of our various applications the topic of data cle...

databases

Polynomial

15:48

UX or SF sounds right to me.

probably more an SF question than a UX one, since it's about back-end admin

Iszi

@Polynomial No, I think it's more about user respect and trust than the actual admin work.

Polynomial

hmm, maybe

I didn't really give it a thorough read

I have a headache and I'm tired

plus it's the last day of work for me today. at 44con tomorrow/friday.

and delving through reams of IDA output is hurting my head

Rory McCune

16:14

@Polynomial so between that and all the booze at 44con it's not really going to be a good week for your head is it!

Polynomial

@RoryMcCune hardly :P

also house party on Saturday

so yeah, hardcore head-fuck

Rory McCune

so a busy week all in! let me know if there's any cool stuff at 44con :)

CodesInChaos

16:29

Ported XSalsa20Poly1305 (what a great name) to C#. 25cpb, so about 5x as slow as fast implementations.

I blame lack of SIMD and lack of rotation support

symmetric crypto tends to have a much bigger gap that asymmetric (ECC etc.)

Rory McCune

could be of interest to crypto types blog.howareyou.com/post/60928192861/how-are-you-cryptography

Thomas Pornin

@CodesInChaos A 3x gap is typical, but the algorithms from Bernstein tend to be very optimized for usage of some specific opcodes (e.g. SIMD), and suffer quite a bit when such opcodes are not used.

My main point in all the SHA-3 process was to explain, again and again, that high-end platforms with nifty opcodes are not an important target for performance, since they don't have such performance issues to begin with.

CodesInChaos

The difference for Curve25519/Ed25519 on the other hand is pretty small. .net is barely slower than equivalent c code (32x32->64 based), and even 64x64->128 based c code doesn't gain that much on top of that.

Thomas Pornin

16:45

@CodesInChaos For some algorithms I had Java code which was only 30% slower than optimized C code. But usually a 2x to 4x factor is obtained. Array accesses, in particular, kill performance because the array bounds are checked (both in .NET and Java). Maybe some "unsafe pointers" could speed up things ? I have yet to lookup exact usage conditions for those.

CodesInChaos

I used mostly structs and local variables

I generally avoid arrays in my crypto code

Thomas Pornin

@CodesInChaos Sometimes you cannot (e.g. RC4).

Or AES, for that matter.

CodesInChaos

I mostly implement/port those fixed memory access bernstein style functions (Blake, Salsa, Ed25519, Poly1305)

Thomas Pornin

strangely enough, one algorithm where I got the "only 30%" figure was Sosemanuk, a stream cipher where the implementation does use some arrays.

CodesInChaos

array access is not that expensive, probably because branch prediction works really well

and sometimes the check gets eliminated. But the JITter is rather dumb in that area

Lack of SIMD and lack of specific instructions (big multiplication, integer rotation, etc.) are the biggest issues in my experience

Rory McCune

16:52

I was with you two up to "Ported" :op

CodesInChaos

in first order it's copy&paste of the c code, and then fixing it until it compiles

Thomas Pornin

@CodesInChaos Well, sure -- but it is not totally a C# issue. When dealing with a "small" platform (embedded ARM or Mips... the kind which actually has performance issues and benefits from the fastest algorithms) you do not have these nifty SIMD instruction even with native code.

Simon

Can you launch JS directly from the button e.g. onclick="function() { alert('test'); }"?

If so, what's the damn syntax?

CodesInChaos

modern ARM has quite a bit of SIMD

Thomas Pornin

@RoryMcCune Well, they sure did a bit of research, but not enough. They speak about indistinguishability of AES-CBC, but that holds only for unpredictable IV, and they don't speak about it. They also don't speak about active attacks (malicious alterations).

Last but not least, they begin the post with an interesting goal: ability to compute "aggregate answers" without revealing the individual identities. But that goal disappears quite quickly.

Rory McCune

16:57

@ThomasPornin see I thought it was an interesting approach but not being crypto types they may not have thought out all the angles..

Thomas Pornin

@CodesInChaos Modern big ARM has quite a bit of SIMD. You'll have that in smartphones. But not in smart cards. Or in RFID tokens. Or even payment terminals.

Rory McCune

@ThomasPornin yep there'd need to be more details there, I'd guess that some of the data isn't encrypted in the same way but that with only the unencrypted data it shouldn't be possible to tie it back to a unique person...

Thomas Pornin

@RoryMcCune Actually it is an hard and essentially unsolved problem, which is why I deem that goal "interesting".

For a practical situation, they would have to make a "trusted aggregator": some shielded machine which is granted access to the user data, but will refuse to reveal it except as "aggregate answers"

(and, even so, some dichotomic requests can reveal quite a lot)

Rory McCune

@ThomasPornin yep that bit is tricky. I've not looked at it at all in healthcare, but I thought in a similar case the panoptikon stuff about browser fingerprinting made an excellent point about how data which doesn't appear to be uniquely identifiable may well be..

with health data I wouldn't think you'd need too many factors to tie one set of data to a real person..

CodesInChaos

Personally I care most about PC and smartphones, so I live in a world where SIMD is available

Thomas Pornin

17:04

@CodesInChaos That's fine. However, that's less fine when researchers trying to define an all-encompassing standard only care about PC and smartphones as well.

That was the point which I relentlessly make (to a large extent, it made me quite "unpopular" with DJB).

CodesInChaos

I think Salsa doesn't that badly even on smaller devices

One thing I'm interested in is how well schemes can be sped up with special purpose instructions that aren't too expensive to implement

Thomas Pornin

@CodesInChaos By "too expensive", you mean "in hardware" ?

The SHA-3 competition is a good example: if you have opcodes which compute one round of AES (and that's rather cheap in terms of silicon area, much cheaper than, say, a multiplication), then you can go real fast.

Simon

Does anyone else feel like he doesn't belong here anymore?

Nope, just me, alright.

Rory McCune

@Simon don't worry I think you'll find that a fair amount of crypto stuff goes right over the heads of most security people (myself included) :)

Simon

@RoryMcCune Good because that's exactly what I'm doing, haha.

Totally ignoring it.

CodesInChaos

17:23

where the additional hardware for the instruction is small, but the instruction gives a nice speedup for the algo

Thomas Pornin

17:41

@CodesInChaos Then there will be few bargains better than including an AES round, at least for symmetric crypto. Also add a "carryless multiplication": it will help about everything which does operations in GF(2^m), in particular GCM, but also binary elliptic curves.

I should try to write an optimized implementation of Koblitz curves with the carryless multiplication opcode added by Intel with the AES instructions.

CodesInChaos

to my surprise AES based SHA-3 submissions didn't seem to have too much success

Thomas Pornin

@CodesInChaos Well, that's because you don't always have the choice to add an AES hardware implementation to your CPU.

And without it, they kind of plain sucked.

CodesInChaos

groestl didn't seem to be too great even with AES-NI

Thomas Pornin

@CodesInChaos Yes, that was the other surprise: given the 3 AES-based candidates, why did NIST choose the slow one ? It was as if they wanted to kill the idea of an AES-based SHA-3...

CodesInChaos

I only looked into SHA-3 in the final phase, so I don't know too much about round 2 candidates.

Thomas Pornin

17:47

@CodesInChaos Basically none of them was broken. The second half of the SHA-3 competition was boring.

The bright side is that we now have many backup algorithms.

Thomas Pornin

18:25

Does anybody knows a way to tell to Android (2.2) that I do not want to be bothered with updates for some app (namely, the Facebook app, which I don't use but cannot remove because it is part of the vendor-provided firmware) ?

If I install the updates, they cumulatively saturate the internal Flash (and the app being part of the OS, the OS refuses to move it to the SD card).

Simon

I don't think it's possible. You could root the device to remove it if you're motivated.

CodesInChaos

it seems like pre-installed applications take twice the space one you update them. Once for the original version, and once for the updated version. Really annoying.

Thomas Pornin

@CodesInChaos I concur. I am really annoyed.

Rory McCune

@ThomasPornin android 2.2 <shudder> that's .... old

CodesInChaos

I think my nexus 1 had 2.3 before it died. And all newer nexus devices have less memory WTF

Thomas Pornin

18:37

@RoryMcCune <ScottMode> Not so much as your mom </ScottMode>

Rory McCune

also less vulns (well apart from the master key one which is still present on fully patched samsung Galaxy S3s amongst other devices)

ManishEarth

@ThomasPornin me too, please

Same's happening for me

Internal memory filling up

I'll probably hard reset one of these days

Thomas Pornin

I don't want to change my phone because it works well for what I want it to do (mostly, to be a phone; also to serve as Internet access for my laptop, and occasionally to use Google Maps) and it has good autonomy (it can keep on running for five days on its battery).

CodesInChaos

why can't google sell some 32 GB nexus (don't care if SD or internal) grrr

Oleg Ostroumov

@ThomasPornin 5 days android huh? And why 2 bears

Thomas Pornin

18:44

@OlegOstroumov Yes, five days being on continuously, on a full charge. I cannot abide smartphones which must be recharged every night.

As for the two bears, it is for showing off. I show that I have so much rep that I can afford to maintain two high-rep accounts.

CodesInChaos

18:54

I think there is some correlation between question quality and the choice of bear. But that might just be the usual human fault of seeing patterns in randomness.

Tom Leek seems to be a bit harsher in his answers

Thomas Pornin

That's voluntary. Bear #2 is "less polite".

Main drive for bear selection is whatever account has not reach rep cap yet that day, though.

AviD

@CodesInChaos "patterns in randomness"... "codes in chaos"... hmm, I think I see a trend.

Thomas Pornin

I tend to use Bear #1 for the hairy crypto questions where I complement my explanations with the authority of my name.

Transcript for