Hacknbeers sqli and cryptography
- 1. ì Information Security SQL Injec*on and Cryptography
- 2. About me ì Miguel Ibarra ì PHP developer since 2000 ì Actually, coding since 1986 ì Projects from simple web pages to GRP’s ì SoLware Engineer @ Tiempo Development ì Webservices, security and cryptography fan
- 3. ì I can has ur data?
- 4. I can has ur data? ì Today, informa*on assets can be more valuable than physical assets… ì Lost your USB s*ck lately? ì Data theL is becoming an every-‐day issue and concern…
- 5. ì Data Dump I can has ur data?
- 6. I can has ur data? ì linkedin.com ì 6th June 2012 ì More than 6 million passwords ì eharmony.com ì 6th June 2012 ì More than 1.5 million passwords ì last.fm ì 7th June 2012 ì ? Million passwords ì yahoo.com ì 12th June 2012 ì 443K passwords ì Sql injec*on ì Passwords in plain text…
- 7. ì I can has ur data? Hot news!
- 8. ì 2010 -‐ 2013 World’s biggest data breaches
- 9. ì 2010 – 2013 (Web only) World’s biggest data breaches
- 10. I can has ur data? ì Public web applica*ons expose an authen,cated and authorized connec,on to DBMS servers ì DBMS have their own authen*ca*on and authoriza*on systems ì Applica*ons that use such DBMS need creden*als to connect to DB servers ì This type of apps can have their own auth procedures ì They can be vulnerable
- 11. I can has ur data? ì It does not maeer if your DBMS is behind a firewall and/or private network User Public network Web server Private network DBMS Server
- 12. ì SQL Injection How I got your data
- 13. SQL Injection ì Golden rule: ì Filter/Validate inputs, escape outputs ì …many ‘developers’ do not follow this rule…
- 14. SQL Injection ì Our database schema ì Sample resultset WTF?
- 15. SQL Injection ì Live demonstra*on
- 16. ì Protecting your data – Password Hashing
- 17. Protecting your data – Password Hashing ì Hash ì Algorithm that maps data of variable length to data of fixed length ì One way func*on ì Output cannot be reversed using an efficient algorithm ì Also called ‘pseudo-‐random func*on’ ì Output indis*nguishable from true random data ì Popular hashing algorithms ì md5 ì sha1
- 18. Protecting your data – Password Hashing ì Hash proper*es ì Output yields a fixed length result ì md5(‘1’) = ‘c4ca4238a0b923820dcc509a6f75849’ ì md5(‘Hello world’) = ‘3e25960a79dbc69b674cd4ec67a72c62’ ì The slightest change, totally different results ì md5(‘Hello World’) = ‘b10a8db164e0754105b7a99be72e3fe5’
- 19. Protecting your data – Password Hashing Message space Result space
- 20. Protecting your data – Password Hashing Hashing encryp*ng
- 21. Protecting your data – Password Hashing ì Our new schema ì Sample resultset
- 22. Protecting your data – Password Hashing ì Again, cannot revert a hash using an efficient algorithm ì … but can be cracked… ì Crack <> Hack Hash Algorithm Data
- 23. Protecting your data – Password Hashing ì Brute force aeack ì Generate hashes from a dic*onary ì Common words ì Generate character combina*ons ì Exhaus*ve search ì Generated hash = target hash? ì Bingo ì Inneficient, but… ì CPU power is growing (mul*ple cores) ì GPU can be used too (thousands of cores) ì Cloud systems
- 24. ì ATI HD 5970 String consis*ng of a-‐z|A-‐Z|0-‐9 Algorithm Speed 8 chars 9 chars 10 chars md5 5600 million h/s 10 hours 27 days 4.5 years sha1 2300 million h/s 26 hours 68 days 11.5 years
- 25. Protecting your data – Password Hashing ì Hash database aeack ì Query a database with pregenerated hashes ì Several sites offer this service, free ì Google.com ì hep://bit.ly/15O4SLN
- 26. Protecting your data – Password Hashing ì Gesng a hash through sql injec*on ì Live demonstra*on
- 28. Password salting ì Salt ì Random character string ì f(password, salt) = hash(password + salt)
- 29. Password salting Result Salt Password
- 30. Password salting ì Our new schema ì Sample resultset
- 31. Password salting ì Salt has to be stored in clear text as to authen*cate a user ì if( hash(<provided password> + <salt field>) == <password field> ) then ì User creden*als are valid
- 32. Password salting ì Hash database aeack becomes improbable ì If hash remains unknown, brute force aeack becomes improbable ì Total characters: 42 ì Calcula*ons per second: 4 billion ì Possible combina*ons: 522 duovigin*llion ì Total *me to crack: 4 septendecillion years* ì According to heps://howsecureismypassword.net/
- 33. Password salting ì But the aeacker could modify the aeack to obtain the salt field… ì DEMO ì The aeacker would only need to launch a brute force aeack ì Generate some character combina*on string ì Concatenate salt and hash ì Compare hashes
- 35. Password stretching ì To mi,gate men*oned aeack, use password stretching technique ì Create a recursive / itera*ve algorithm that calculates a hash value over itself thousand (or more) *mes
- 36. Salt Password Hash Salt Password Hash Salt Password Hash Algorithm Hash Algorithm Thousands of *mes
- 37. Password stretching ì This algorithm should iterate enough to delay each calcula*on by 1 second ì In order to crack with a brute force aeack, the aeacker… ì Should know the exact itera*on count ì +/-‐ 1 itera*on will result in totally different hash value ì Should wait 1 second between each aeemp ì This makes the aeack highly improbable
- 38. Password stretching ì Several standard algorithms for password stretching ì PBKDF2 ì Bcrypt ì Scrypt ì …
- 39. ì I can still has ur data…
- 40. I can still has ur data… ì We give it away freely… ì Facebook ì Twieer ì Foursquare ì If it is free, you are the product
- 41. I can still has ur data… ì We uninten*onally give it away… ì Phishing scams ì Social engineering ì Adware / Spyware / Browser bars / Apps ì Weak passwords ì Names ì Birthdays ì Phone numbers ì Common passwords
- 42. I can still has ur data… ì It is forcibly/unlawfully taken from us… ì Extor*ons / black mail ì Unethical prac*ces ì banks ì Government spy programs ì NSA’s PRISM ì Communica*ons Intercept System Mexico ì Requested by USDoS to Mexican Federal Government in 2007 ì Request cancelled in 2012… ì Unknown sponsored spy programs ì Rumored FinFisher program probably running in networks belonging to Uninet, Iusacell and Televisa
- 44. Cryptography 101 ì Krypto ì Hidden ì Graphos ì Script ì Tecnique to modify a linguis*cal or caligrafical presenta*on of a message ì Ruled by an algorithm ì Must allow forward and backward process
- 45. Cryptography 101 ì 700BC -‐ Scytale
- 46. Cryptography 101 ì <40BC – Caesar Cipher
- 47. Cryptography 101 ì 1467 – Alber* Cipher
- 48. Cryptography 101 ì 1797 – Jefferson Wheel
- 49. Cryptography 101 ì 1943 – Enigma machine
- 50. Cryptography 101 ì Today, cryptography is performed by an automated algorithm: Cipher ì Short name for pseudo-‐random permuta*on ì Takes an input ì Applies a reversible algorithm ì Outputs data indis*nguishable from a truly random data stream ì Result space is equal to message space ì No collisions
- 51. Message space Result space
- 52. Cryptography 101 ì Modern algorithms use a ‘key’ ì They key is used to transform a message into a pseudo-‐random string ì This is called ‘cipher’ ì This pseudo-‐random string can be transformed back to the original message only with this key ì ‘decipher’
- 53. Cryptography 101 ì 2 key types ì Symetric ì Asymetric
- 55. Symetric ciphers ì The same key is used to cipher and decipher ì The 2 endpoints must agree on this key ì Security relies mainly in this key ì Key must be improbably guessed ì Key space has to be large…
- 56. Symetric ciphers ì Simple and safest symetric key chipher algorithm ì XOR ì 1 ⊕ 0 = 1 ì 1 ⊕ 1 = 0 ì 0 ⊕ 0 = 0 ì ‘a’ = ascii(97) ì 9710 = 110000012 ì Random key = 01010000
- 57. 1 1 0 0 0 0 0 1 0 1 0 1 0 0 0 0 1 0 0 1 0 0 0 1 1 0 0 1 0 0 0 1 0 1 0 1 0 0 0 0 Data to cipher Random key Ciphered data Data to decipher Random key Original data
- 58. Symetric ciphers ì Ciphered data is impossible to decipher without the key by an efficient algorithm ì This is, no exhaus*ve search for the key ì Is very simple ì Key length must be the same as message length ì Security measures applied while securily sharing the key might as well we applied to the unciphered message
- 59. Symetric ciphers ì In fact, every symetrical cipher weakest link is the key ì An aeacker, instead of brute forcing the key, might as well focus on incercep*ng the key ì Popular cipher algorithms ì DES ì 3DES ì AES
- 61. Asymetric ciphers ì Has a key pair ì Private key: only the owner can know it ì Public key: owner can share it freely ì Message ciphered with the public key can only be deciphered with the private key ì Message ciphered with the private key can be deciphered with the public key ì This adds a message authen*ca*on mechanism
- 62. Asymetric ciphers ì Algorithms are based on prime number and one way func*ons ì Way too easy to mul*ply to prime numbers ì Factorizing a number into its prime factors is very difficult ì Usually involves very large prime numbers ì Hundreds of digits
- 63. Asymetric ciphers ì Asymetric ciphers require more processing *me ì Keys are required to be large ì As today’s standards, 2048 bits ì Ciphered message is bigger than the original message ì Popular algorithms ì RSA ì Hybrid symetric/asymetric algorithms ì HTTPS/TLS
- 64. ì Just encrypt it and you are safe …are you?
- 65. Just encrypt it and you are safe… are you? ì Weakest links in cryptography ì Again, the key… ì … and how it is implemented ì Aircrack anyone? ì WEP algor*hm: example of bad crypto implementa*on
- 66. Just encrypt it and you are safe… are you? ì Common cryptography implementa*on misconcep*on C Message Key Cipher Message
- 67. Just encrypt it and you are safe… are you? ì Do not… ì Use a short key ì Use weak random data to generate a key ì Use directly the generated key ì Try to derive the key first ì PBKDF2 ì Pseudo-‐random func*on ì Use the same key to cipher iden*cal messages ì Informa*on leak ì Use the same key to cipher mul*ple messages ì WEP’s Achilles’ heel ì If you need to, use nonces ì Nonce ≃ salt ì Nonce is included with the message ì Cipher and send… ì Always add signature verifica*on mechanism ì Hash-‐mac ì Hash with a key ì Hash-‐mac signature included in message
- 68. ì Do not use weak random data Random data, PHP 4’s rand() func*on output on Windows converted to bitmap
- 69. ì Encryption in databases
- 70. Encryption in databases ì We have the following schema
- 71. Encryption in databases ì The aeacker, unable to get login details, could modify the aeack to get personal informa*on:
- 72. Encryption in databases ì MySQL supports AES cipher algorithm, then we could modify our schema
- 73. Encryption in databases ì If the aeacker executes the injec*on
- 74. Encryption in databases ì S*ll informa*on can be leaked ì Suppose the 2 users share the same phone number, the aeacker could no*ce this since data was ciphered with the same key
- 76. Tweakable encryption ì Remember, do not use the same key to cipher mul*ple messages… ì Do we need to generate a new key for each record? ì Say, we have thousands of records, do we need thousand keys? ì NO ì Use a tweak
- 77. Tweakable encryption ì Every table should have a primary key ì So the values inside the primary key must be unique to every record ì Use the master key + primary key value, then hash ì f(key, pk value) = hash(key + pk value) ì You’ll have an unique cipher key for each record ì Now, fields with the same plain text value will appear to be completely different when ciphered
- 78. Last Words ì Last words… ì Informa*on privacy is YOUR RIGHT ì Do you consider it to be a paranoid idea… ì …or a daily life concern? ì First informa*on privacy law from 1890 US ì Laws cannot keep up with technology
- 79. QA ì QA ì Thank y’all!