Implementing TALT

1
Implementing TALT

• William Lovas
• with Karl Crary

2
Grid Computing

• Low-cost supercomputing using idle cycles of
  Internet-connected PCs
• Cray X1 supercomputer 8.4 million
• Pentium 4, 2.2 GHz 699 (with rebate!)
• Getting everyone elses PCs to do your work for
  you priceless

3
Grid Computing
PCs on The Internet
Developer
4
Grid Computing

• SETI_at_Home project at Berkeley
• scanning radio signals for signs of intelligence
• Folding_at_Home project at Stanford
• studying diseases related to protein folding
• distributed.net
• cracking encryption algorithms since 1997

5
Grid Computing

• Very powerful distributed.net, 22 hours to brute
  force 1 DES key out of 256 ( over 72
  quadrillion!)
• distributed.net equivalent to 160,000 Pentium II
  266 MHz PCs

6
Grid Computing

• Few examples, because grid computing depends on
  users trust in the code they run
• ConCert Project to develop a framework for
  trustless grid computing

7
Roadmap

• Motivation check!
• Methodology of TALT the how and why
• A taste of TALT to whet your appetite
• My work on TALT what I accomplished, what I
  learned

8
Establishing Safety

• How can we verify that code is safe?
• High-level languages use type systems
• Why cant low-level languages, too?

9
Typed Assembly Language

• Formalizes a subset of x86 instruction set
• Adds a type system that demonstrates safety
• Idea employ powerful tools from type theory to
  tame the assembly language beast!
• subtyping, polymorphism

10
TALT -- TAL Two

• Refines and extends the ideas from TAL
• Establishes a framework for future type systems
• One type system wont meet everyones needs

Two type systems good gt Four type systems
better! -- Benjamin C. Pierce
11
How TALT works

• Trusted
• Twelf proof checker
• and meta-theorem prover
• Safety policy (SP)
• safe machine states, safe transitions
• Runtime

• Untrusted
• Code
• Type system (TS)
• Safety proof (TS(p) gt SP(p))
• Machine-checkable, via Twelf

12
How TALT works
code
type system
Twelf
type system
safety proof
safety policy
code
runtime
13
A taste of TALT

• Many interesting ideas, lets talk about 3
• Register file subtyping
• Universal polymorphism
• Sized kinds

white lies ahead!
14
A taste of TALT
Register file subtyping
15
A taste of TALT

• Register file type looks like
• r0 t0, r1 t1, rn tn
• Suppose a code-block expects
• eax int, ebx int
• Okay if register file actually has type
• eax int, ebx int, ecx int
• Because of register file subtyping!

16
A taste of TALT
Universal polymorphism
17
A taste of TALT

• Suppose a function f leaves the stack unchanged
• Stack pointer stored in register esp
• How does a calling function know?
• f has a type polymorphic in esp!
• Just like SML

- fun id x x val id fn 'a -gt 'a
18
A taste of TALT
Sized kinds
19
A taste of TALT

• Kinds the types of types
• In TALT, the kind of a type specifies the size of
  that types values
• A 4-byte word has type B4, which has kind T4
• Pair of words has type B4B4, which has kind T8
• Typing rules know the sizes of operands
• Essential for complex addressing modes

20
My work on TALT

• Implementation of infrastructure
• Runtime
• Certificate generator
• Assembler
• Creation of new code as well as
  simplification/extension of existing code

21
Runtime implementation

• Runtime provides primitive operations, like
  malloc, exit, return
• Works directly with machine code, so written in
  C, with some x86 assembly
• Great care taken to adhere to specification
  runtime is trusted!

22
Certificate Generation

• TALT binary looks like this

magic number
machine code
code
certificate
23
Certificate Generation

• Certificate generator generates

magic number
machine code
code
certificate
24
Certificate Generation

• When I arrived
• Certificate was a typing derivation in the TALT
  type system
• Partially implemented and buggy derivation
  generator
• Discovery typing derivation too big to use as a
  practical certificate

25
Certificate Generation

• And now
• Certificate is an XTALT program, in a notation
  that Twelf understands
• Certificate generator is 97 complete (and
  hopefully not buggy!)
• Code is much simpler
• Type system is a little simpler, too

26
Assembler

• Next step translate XTALT programs into straight
  x86 assembly
• Assembly can be run through a standard assembler
  to produce machine code

27
Assembler

• Now we also have

magic number
machine code
code
certificate
28
Assembler

• Then, only have to put the pieces together to
  produce TALT binaries!

29
Summary

• Implemented TALT runtime in C, assembly
• Implemented certificate generator in SML
• Simplified typing rules
• Simplified code
• Hope to implement assembler by next week
• Very close to a working system!

30
Summary

• Didnt discover anything new, but
• Learned a lot about type theory, and
• Hacked on some interesting code!

Questions?