The DMZ

A stupid question... it has probably been asked on the site, but I can't seem to find it. When hashing passwords, does it make sense to repeat the hashing several times? Like SHA(SHA(SHA(pass))) ?

Thomas Pornin

1:10 PM

@bilbo_pingouin Ah ! "probably"

@bilbo_pingouin Read that:

602

A: How to securely hash passwords?

The Theory We need to hash passwords as a second line of defence. A server which can authenticate users necessarily contains, somewhere in its entrails, some data which can be used to validate a password. A very simple system would just store the passwords themselves, and validation would be a s...

Xander

@ThomasPornin So what was the deal with Wifi blocking SSL2 yesterday? And was it actually Wifi, or something on the other side of the AP?

Thomas Pornin

Don't ever think about touching your keyboard again until you have read it thoroughly.

Xander

@ThomasPornin Or did you not care enough to pursue it further?

Thomas Pornin

@Xander When sending a SSL 2.0 ClientHello, instead of receiving a ServerHello, I got a plain HTTP answer that identified the system that was denying the access.

Xander

@ThomasPornin Interesting.

Thomas Pornin

1:12 PM

I don't exactly know what is expected from that; a Web browser that issues a SSL "hello" won't expect a non-SSL answer with some HTML.

Anyway, I could compile an old OpenSSL that supports SSL 2.0, and run tests on localhost.

Xander

@bilbo_pingouin That specific form of the question has been asked a few times on the site, but the answer is always to link back to the answer @ThomasPornin just posted.

Mark Buffalo

@ThomasPornin If I recall correctly, didn't you tell us to avoid SSL 2 like the debil?

Thomas Pornin

@MarkBuffalo Cops tell that shooting other people is bad, yet they train hard to do it.

Mark Buffalo

@ThomasPornin haha, I figured that was why you're using it.

Thomas Pornin

I am not making a full SSL 2.0 handshake -- I am sending a ClientHello to see if the server at the other end would accept to do a SSL 2.0 handshake (and I extract the certificate and cipher suites).

Mark Buffalo

1:18 PM

Just wondering if that's the version you said to stay away from

Xander

@MarkBuffalo SSLv2 and SSLv3 should both be disabled.

Thomas Pornin

If you use SSL 3.0 then you are doing it bad. If you use SSL 2.0 then you are doing it worse.

TLS 1.0 is kind of OK but it requires implementations to exercise great care (constant-time unpadding and MAC checks, 1/n-1 splits to get IV unpredictability...).

TLS 1.1 is better. TLS 1.2 allows for AEAD cipher suites (GCM) which are even better.

Xander

@MarkBuffalo SSL v2 has been explicitly prohibited by RFC for 5 years. tools.ietf.org/html/rfc6176

Mark Buffalo

Thanks for the info

bilbo_pingouin

@ThomasPornin and @Xander I did look more specifically to that form of asking that question. Not for a related questions. Nice posts, thanks for sharing. :)

Xander

1:22 PM

8

Q: What is the problem with chain hashing?

Let's say that my password is a single character: "a". Couldn't I chain hash it 1000 (or more) times and make it nearly invulnerable to rainbow table attacks and brute force? Why isn't this preferred to salting and what are the problems with this technique, if any? EDIT: Clarification: By chai...

hash

Mark Buffalo

horray for my free meal

bilbo_pingouin

@Xander From that I gather that one does some hash(hash(hash(pass+salt)))...right? In that post 1000 iteration is mentioned. Is that current standards?

Mark Buffalo

I should get knifed more often

bilbo_pingouin

@MarkBuffalo it's probably lethal in the long run

Mark Buffalo

@bilbo_pingouin No, see... I've been cut so much that I built up a resistance

SmokeDispenser

1:27 PM

@MarkBuffalo Did the subway pay or the old lady?

Mark Buffalo

@SmokeDispenser Neither. The food booth next to them did... lol

"Sorry she threw a knife at you. Here's a free apology card meal"

SmokeDispenser

O_o

Mark Buffalo

"Awesome!"

SmokeDispenser

nice.

lucky you :)

TildalWave

wow new faces

Thomas Pornin

1:29 PM

@bilbo_pingouin It is not standard. It is widespread. Many people hash passwords most horribly.

TildalWave

runs new faces through new faces identification algo

Xander

@bilbo_pingouin No, the current standard is in the post that @ThomasPornin linked to. Use bcrypt, failing that scrypt, failing that PBKDF2 with work factor as high is as reasonable for your application.

Mark Buffalo

@Xander I used PBKDF2 for a bit

Switching to bcrypt

Xander

@MarkBuffalo I use PBKFD2, because it is implemented in the .NET framework. Which reminds, me, I need to post issues in GitHub, asking for bcrypt and/or scrypt.

Terry Chia

@bilbo_pingouin More like 100,000 at least.

Mark Buffalo

1:32 PM

@Xander Wait. This is already implemented in .NET? But I wrote an implementation for it....

ffs

The more you know.

Xander

@MarkBuffalo Yup. System.Security.Cryptography.Rfc2898DeriveBytes

Mark Buffalo

That's why I didn't find it

That, and not caring

I don't work with crypto / hashing all that much

bilbo_pingouin

I read about the basic theory, but never had to implement any password hashing myself. So I never thought about iterations. Good to know, thanks a lot. And 100,000 sounds like a lot... I'm wondering if someone bothered to evaluate the resources wasted on password hashing (for both side)... ^^"

Thomas Pornin

@bilbo_pingouin Evaluating, that's your job. The whole idea of the iteration count is that it is part of the configuration which is, by nature, specific to each deployment site.

The good count is exactly the one that is the highest and still tolerable on your hardware with your expected peak work load.

Xander

@bilbo_pingouin For offline applications where security is vital, such as volume encryption, or password manager master passwords, the iteration count can reach into the millions.

TildalWave

1:37 PM

@bilbo_pingouin you'd have to include all the workload used to compute cryptcurrencies

Shadow Wizard

@Telkitty so how can I know it's really you, and not someone who hacked your account? ;)

Xander

@TildalWave Which, incidentally, is insanely inefficient. I was reading awhile back that each bitcoin transaction verification requires roughly the same amount of electricity as would power 1.5 American houses for an entire day. Which is a lot.

Shadow Wizard

@RoryAlsop awesome! Well, can't beat the source... :D

kalina

two days in a row :o

Shadow Wizard

Who? Where?

bilbo_pingouin

1:41 PM

@ThomasPornin that's not what I meant by evaluating. Hashing needs processors. Processors need power. Power need resources. Oil, wind, uranium, you name it. And as @TildalWave mention there are more to add to that.

Xander

@bilbo_pingouin Passwords also need security. Which is a consideration.

silverpenguin

@kalina ???

bilbo_pingouin

@Xander That's clear that we do it because we need to do it. I'm just curious about how much that represents.

kalina

@ShadowWizard it's yoooooooooou points

TildalWave

@ShadowWizard you can't, for all you know some of us might be refrigerators :P

kalina

1:45 PM

or dogs

diagprov

Or bears.

kalina

or ponies

Mark Buffalo

2:02 PM

Where is @Simon?

I feel incomplete without @Simon.

@kalina top of the morning to you, siress.

SmokeDispenser

Oh gawd, that was a lot of link searching. Hope, OP can reasonably use it.

Ohnana

@SmokeDispenser stickers for you

SmokeDispenser

haha, that feeling when done and closing 20 tabs;)

Mark Buffalo

@SmokeDispenser Close all but this?

Close other tabs!

It feels like I had a shower after closing all of those tabs

SmokeDispenser

@MarkBuffalo, 3 left. this, the front page of sec.SE and an article I'm reading

Mark Buffalo

2:12 PM

We're less than half-way to next month, then we can finally receive more words of wisdom from @BarryCarlyon

Rory Alsop

@TildalWave or pyromaniacs

kalina

siress?

Rory Alsop

I'm guessing instead of Ma'am. But it makes me think of some kind of Siren

Lighty

@RoryAlsop At the amount of voice @kalina could produce with those trained vocal cords, Siren is prolly a good name

Rоry McCune

lol, some london start-ups are getting desperate, one just e-mailed me about a job as a backend engineer based on my github profile... I can't imagine anyone who'd looked at my code thinking that was a good idea...

kalina

2:15 PM

@RoryAlsop I would have probably accepted siren

@RоryMcCune I heard you like the backend

Rоry McCune

@kalina Who the heck have you been talking to!

Mark Buffalo

@RоryMcCune Yeah, this happens to me all the time.

"Oh, I see you've created a few PHP functions... PLEASE WORK WITH OUR MAGENTO STACK"

kalina

Oh i see bla bla bla bla words letters numbers bla would you bla bla bla bla bla?

Mark Buffalo

"Oh, I see you've done some sysadmin work several years ago. We're looking for a SENIOR-PRINCPAL SYSTEMS ADMINISTRATOR"

@kalina precisely

SmokeDispenser

@RоryMcCune :D

@MarkBuffalo, like your tin foil answer on checking links. Didn't think of the VPN, good call :)

Mark Buffalo

2:24 PM

I'm adding more info

SmokeDispenser

@MarkBuffalo, commented :)

@MarkBuffalo, for additional being-curious;)

... aaaand about to hit the rep cap again. It this a usual thing here? I'm not even answering half as much as I used to on SO. This is so much nicer and more rewarding in terms of imaginary unicorn internet points than SO.

Mark Buffalo

@SmokeDispenser Good catch... that's usually detected in the javascript though

SmokeDispenser

@MarkBuffalo, being curious. the server could try and hide the attack from testing.

Thomas Pornin

@SmokeDispenser There is a lot less competition on answering here, so answers with lots of information don't get outrepped by dozens of one-line answers that were posted first.

Xander

@SmokeDispenser Since we have only a tiny a fraction of the traffic and activity of SO, questions and answers don't tend to get lost and slip under the radar like they do at SO.

@ThomasPornin And this too.

Thomas Pornin

2:31 PM

Generally speaking, Sec.SE has an audience that has higher "quality" than SO.

SmokeDispenser

also, probably, good answers do not get downvoted because 'no copypasta', I guess

silverpenguin

@ThomasPornin if you dont say so your self hey bear ? :P

Thomas Pornin

It is nevertheless usually felt that quality of questions has declined over the last year or so, maybe as an aftereffect of increasing exposure.

SmokeDispenser

I do feel like I should've turned by back on SO a while earlier;)

silverpenguin

@SmokeDispenser SO users are the biggest keyboard warriors. where their personal preference is the right answer...

IMO of course

Thomas Pornin

2:34 PM

@SmokeDispenser The journey matters more than the destination. By doing all the link searching, you learned more yourself. (Although you did miss a few links, which I have added as my own answer.)

SmokeDispenser

@ThomasPornin, yeah, that is probably a reason. But that must not be a bad thing, in 'our' case. cf. the SO-question I answered and whichs deletion (now undeleted) sent me this way.

Mark Buffalo

@ThomasPornin Agreed

SmokeDispenser

@ThomasPornin, my link searching was more about demonstrating that there is no management-friendly translation available. I think I didn't learn more than OP will learn;)

@ThomasPornin, but I like your answer, will get a vote:)

Mark Buffalo

I need to release one of my monitor programs

that monitors changes to the full file system, and who dun it

@ThomasPornin I love this about secse. It's true to an extent on SO as well... but most of it is a bunch of one-liners. Sickens me to see all the suggested SQL injection attacks, though

SmokeDispenser

@ThomasPornin A different approach though, yours. One I like better personally (because more thoroughly researched, confined and put together) but will probably not help OP as much. I think I'm SO-damaged, trying to help OP regardless of style and/or scientific quality. anyone repair me?;)

@MarkBuffalo, did you read my linked meta on security related problems on SO? their community puts them side to side with performance issues, like 'doesn't matter, had sex'. One of the reasons I'm here now.

Mark Buffalo

2:43 PM

let me see

Link it

Thomas Pornin

@SmokeDispenser Give a man a fish and you feed him for a day; teach a man to fish and he will hate your condescending butt for a lifetime.

4

Rоry McCune

@ThomasPornin speaking of SSL stuff have you seen testssl.sh

Mark Buffalo

lmfao

SmokeDispenser

@MarkBuffalo, linked in this gist, with some context

Mark Buffalo

...and github is blocked

SmokeDispenser

2:44 PM

@ThomasPornin :D

Mark Buffalo

wtf

SmokeDispenser

so no gist?

then... wait.

Mark Buffalo

no gist. no github

probably because some idiot would accidentally upload stuff to github

SmokeDispenser

@MarkBuffalo the meta has linked the question w/ my answer.

Mark Buffalo

that's probably why it's blocked now

Ohnana

2:46 PM

@SmokeDispenser we are not afraid to call something a stupid question

SmokeDispenser

@Ohnana, I'm neither.

Rоry McCune

wow a 5KLoc shell script, that must be ... fun... to maintain

Mark Buffalo

@SmokeDispenser I just downvoted CodeCaster

Thomas Pornin

@RоryMcCune I had not seen that. Interesting. Uses OpenSSL, though, so it is delicate to use as a tool on generic Windows systems.

SmokeDispenser

@Ohnana with 'regardless of style or quality' I refered to my answer, not a question:)

Rоry McCune

2:47 PM

@ThomasPornin yeah I've done a little docker container for using it which uses the statically linked OpenSSL that's on that site which increases the number of ciphers checked for a lot from base and obv. re-enables things like SSLv2

good for lone or small numbers of servers but I use github.com/nabla-c0d3/sslyze for larger numbers as it has XML out so I can parse/extract as needed more easily.

SmokeDispenser

@MarkBuffalo, the interesting part is that the Q linked in the meta got deleted by SO's meta-burning-torches rantmob:)

like 'who cares if the answer is good, there is no copypasta'

Thomas Pornin

@RоryMcCune I should add a parseable report format to TestSSLServer, but I am a bit unsure about XML.

Any suggestion ?

SmokeDispenser

@ThomasPornin JSON?

Ohnana

NGL, i was reading your goodbye letter and i was like "wow, does he realize no one cares on a big site like SO" nad then i saw the chat transcript of everyone flipping out when they'd realized you'd gone

:(

Mark Buffalo

@ThomasPornin Do you mean you haven't learned to read the XML format, or?

SmokeDispenser

2:50 PM

Hmm. I thought the rep cap was 200? how do I get 224?

Mark Buffalo

@SmokeDispenser Two accepted answers?

and a few downvotes

accepted answer = +15. Two of them means +30. So if you got 224, then you were downvoted 3 times

SmokeDispenser

@MarkBuffalo, only votes count to the rep cap?

@MarkBuffalo, the 4 come from 2 edits then

Mark Buffalo

accepted answers and bounties help you exceed the rep cap

SmokeDispenser

as I'm still being reviewed;)

Thomas Pornin

@MarkBuffalo I have written my own XML parser. I thus came to the conclusion that it is an invention of the Devil to cause the perversion of Mankind.

SmokeDispenser

2:52 PM

Oh, no, one edit, and the first accept

Mark Buffalo

@ThomasPornin Ha. I have done the same... and have come to the same conclusion

SmokeDispenser

@ThomasPornin XML is just wrong. altogether.

Mark Buffalo

One of my previous tasks involved parsing terabytes of XML files, and dumping the records into a database. ETL, essentially.

Xander

@SmokeDispenser Yes. Repcap is for upvotes only.

Rоry McCune

@ThomasPornin JSON and XML are the two ones I see most often, they both have very good parser support in most languages so as long as the format's reasonably explicable it's easy to work with

@ThomasPornin there's nmap which has a "greppable" format as well but that's not something I see very often

Mark Buffalo

2:54 PM

@RоryMcCune would probably work with XML instead of JSON

Thomas Pornin

@RоryMcCune JSON seems reasonable enough.

Rоry McCune

@ThomasPornin whilst XML can be a bit of a pain, after you get XPath it's not too horrible to extract stuff from

SmokeDispenser

@Xander, thanks (for your comment, too. Will edit the answer :))

Mark Buffalo

there are a lot of funky deserialization exploits with many JSON parsers

Xander

@SmokeDispenser Cheers.

Rоry McCune

2:55 PM

@MarkBuffalo well working with untrusted input is always dangerous, but as this should be tool output it shouldn't be too nasty

@ThomasPornin yeah JSON is very popular these days and pretty good for simple output...

Thomas Pornin

I suppose there are subtleties with character encoding and what constitutes whitespace. Also, string literal definition uses "\u" with four hex digits, not 6, so it is insufficient for actual Unicode support (unless using UTF-16 surrogate pairs...).

But for technical ASCII output, that should work.

diagprov

@ThomasPornin Java has a number of XML serializers you could use. It's been a very long while since I've done it, though.

Mark Buffalo

I refuse to use serialization

I will manually parse according to my expected format

Thomas Pornin

@diagprov I have switched to C# -- mainly so that the resulting executable may be used on a stock Windows system without installing any extra framework.

SmokeDispenser

@Xander, edited. Feel free to ping me here for such things if I'm around anyhow.

diagprov

2:59 PM

@ThomasPornin Ahhh. Well, C# has lots of XML serializers too :)

Both in .net and third party.

Thomas Pornin

In fact I was toying with the idea of using a CSV file for people who want to read it with Excel.

Because people do weird things like that.

SmokeDispenser

@ThomasPornin, yeah, unfortunately.

Mark Buffalo

techcrunch.com/2016/03/16/…

Finally. It will make my life so much easier... having my shoes tied for me

diagprov

@ThomasPornin A good serialization library should be able to do both JSON and XML. I.e. you should get both for free.

Mark Buffalo

It also should be able to let you execute arbitrary code

mutters under his breath about serialization

HamZa

3:17 PM

8

Q: How to create a SQL injection attack with Shift-JIS and CP932?

I'm writing some unit tests to ensure my code isn't vulnerable to SQL injection under various charsets. According to this answer, you can create a vulnerability by injecting \xbf\x27 using one of the following charsets: big5, cp932, gb2312, gbk and sjis This is because if your escaper is not co...

php mysql security encoding sql-injection

^ Internet pointz 4 ownage!

silverpenguin

@HamZa man, CBA :< can you put a bounty on a better question e_e hahaha

HamZa

@silverpenguin you think that question is of poor quality?

Ohnana

@HamZa i think in this case "better" is defined as "easier" :P

HamZa

I've spent hours debugging this. So it is hard, trust me.

Ohnana

i know that's my eternal struggle with bounties lol

HamZa

3:24 PM

I've given around 3K in form of bounties, as long as I stay at 10K, I'm fine with it :P

Ohnana

wow!

silverpenguin

@HamZa Well its just the qay i understand questions... if he gave me his code or a variation I could test out some theories.... answering something like this from the top of my head cant be a thing requires more research and use of other peoples work rather than solidly my own

HamZa

@silverpenguin well, that's the thing: SJIS SQLi doesn't work like gbk.

Remember we used the Yen symbol

So he started on the wrong foot.
Now I tried with the right injection character and still failed

The only thing I didn't try is to use an OLD version of PHP...
Maybe <5.3...

Gotta go to work

SmokeDispenser

@HamZa hf :)

diagprov

@MarkBuffalo so do you have a recommended serializer that can let me marshall/unmarshall data without getting exploited? Or a guide I can use so I know how to avoid making my product vulnerable?

Given that my default choice will be "whatever is in .net or Java" unless there is some particularly justifiable need not to.

Mark Buffalo

3:35 PM

@diagprov I am just very suspicious in general of arbitrary data to begin with, especially stuff that can be serialized into arbitrary class/types, and then do all manner of fonkey things.

and therefore do not use serialization... at all...

I create my own parsers with expected formats, and go from there

and I check every frigging input value to ensure it makes the cut

SmokeDispenser

So, still looking for some guideance regarding the tone and professionalism of answers here. I noticed over here, fun is perfectly acceptable. Is this answer of mine too much fun and/or too little 'further reading'?

(waiting for the formal logic jokes answering 'yes') ;)

Rоry McCune

@SmokeDispenser given the level of comedy/snark that goes into answers here, that seems fine to me but then IANAM (you need @AviD @RoryAlsop @schroeder or @JeffFerland for that)

heck if we disapproved of potential snark in answers, little bear wouldn't have as much rep as he does

SmokeDispenser

@RоryMcCune, just playing safe. went 'to school' at a place where sarcasm, let alone fun, is frowned upon;)

Rоry McCune

e.g. security.stackexchange.com/a/33471/37 has 962 upvotes...

@SmokeDispenser or security.stackexchange.com/questions/9487/… with 308 upvotes

SmokeDispenser

@RоryMcCune :DD

Rоry McCune

3:50 PM

as you can see fun/sarcasm is rarely a problem here

god if security people weren't allowed to be sarcastic, I think we'd all explode!

SmokeDispenser

:D

@RоryMcCune I think OP might didn't get it

Rоry McCune

that is the risk with humour...

silverpenguin

@RоryMcCune I actually got an upvote just for being funny apprently

3

A: How can an attacker eavesdrop on users in public WIFI?

When accessing a public wifi an attacker might use applications such as wire shark to capture packet information (the data of users), if encrypted the attacker will most likely attempt some kind of MitM MitM ARP poison, this will essentially tell the victims machine "Yo, Im the next hop I swear"...

wasnt that funny but it gave me a single up vote atleast

point farming

Thomas Pornin

@SmokeDispenser Fun is always acceptable as long as it is actually funny.

It has been recently discussed: meta.security.stackexchange.com/q/2261/655

Iszi

@RоryMcCune I find it mildly amusing that UANAM but you're the one with the pinned comment about mods not being good chat role models.

Rory Alsop

3:59 PM

@RоryMcCune I detected no snark of any discernible level in that answer

Transcript for