Don t let data Go astray

Size: px

Start display at page:

Download "Don t let data Go astray"

Charity Ferguson
5 years ago
Views:

1 Don t let data Go astray A Context-Sensitive Taint Analysis for Concurrent Programs in Go Volker Stolz Bergen University College, Norway & University of Oslo, Norway 28 th Nordic Workshop on Programming Theory (NWPT 16) 1 st November 2016 Supported by the bilateral Norwegian/German project GoRETech Go Runtime Enforcement Techniques & EU COST Action IC1402 ARVI Applied Runtime Verification

2 Don t let data Go astray Violet Ka I Pun Martin Steffen Volker Stolz Anna-Katharina Wickert Eric Bodden Michael Eichberg

3 Motivation Taint analysis: data flow analysis Secure information flow Prevent untrusted/sensitive data from reaching sensitive locations Examples (the usual suspects): SQL injection (user input flows unfiltered into SQL query) leaks (clear-text password ending up in log/debugging output) Volker Stolz Don t let data Go astray NWPT 16 1 / 20

The Go language I Backed by Google I imperative

typed I garbage collected; dynamic race checker I

4 The Go language I Backed by Google I imperative (C-programmers should be able to read it) I object-oriented (maybe... ) I concurrent (goroutines) I structurally typed I garbage collected; dynamic race checker I higher-order functions and closures Volker Stolz Don t let data Go astray NWPT 16 2 / 20

5 What are methods? procedures functions methods methods are specific functions Example: add1 / add2 not that much different from each other type Number struct { n int } func add1 (x Number, y Number) func (x Number) add2 (y Number) int { return x.n+y.n } method function with special first argument f (o, v) vs. o.f (v) elsewhere often: special keyword for first argument: this (or self) Volker Stolz Don t let data Go astray NWPT 16 3 / 20

6 Higher-order functions known from functional languages languages with higher-order functions functions as first-class data func add (x int) (func (int) int) { return func ( y int ) ( int ) { return y + x } } add : int (int int) = λx :int.λy :int.x + y Volker Stolz Don t let data Go astray NWPT 16 4 / 20

7 Deferred functions Each function/method can be called: 1 conventionally 2 deferred 3 asynchronously (see later) Also in Apple s Swift language func main () { defer fmt. Println ( 1 ) fmt. Println ( 2 ) } Deferred call (guaranteed to be) executed when the surrounding function body returns Eval d for side-effect only, returned value irrelevant Deferred calls can be nested, too Volker Stolz Don t let data Go astray NWPT 16 5 / 20

8 Concurrency in Go Go s concurrency mantra Don t communicate by sharing memory, share memory by communicating! Go concurrency goroutines + channels claimed to be easy first-class, typed channels Volker Stolz Don t let data Go astray NWPT 16 6 / 20

9 Channels named pipes FIFO, bounded, non-lossy communication crucial data type with synchronization power taking a back-seat: locks mutexes monitors semaphores... channels: first-class data channels can send (references to) channels inspired by CSP (and CCS, and, actually π) directed channels Volker Stolz Don t let data Go astray NWPT 16 7 / 20

10 Channels Simple example: synchronized channel for strings package main import fmt func main () { messages := make ( chan string, 0 ) } go func () { messages ping } () msg := messages fmt. Println (msg) Volker Stolz Don t let data Go astray NWPT 16 8 / 20

11 Back to Taint Analysis... Identifies flows of private information to untrusted places Terminology: Source (produce tainted data) Sink (consumer of data) Flow (from source to sink) Both given as user-defined sets of methods (Q: How to untaint data? Sanitizers, future work) Volker Stolz Don t let data Go astray NWPT 16 9 / 20

12 Abstract syntax Abstract syntax captures essence of Go Gory details in SSA representation in implementation s ::= x := e x.f := e if v then s else s defer((λx.s) v) go s x y x y return v s ; s e ::= v v v makechan v ::= x x.f () true false λx.s Volker Stolz Don t let data Go astray NWPT / 20

13 Lattice for Taint Analysis Simple lattice: uninitialized / untainted / tainted / both Volker Stolz Don t let data Go astray NWPT / 20

14 Standing on the Shoulders... Existing Go compiler infrastructure: SSA representation Basic blocks / call-graph Points-to analysis à la Andersen Interprocedural, context-sensitive analysis: Padhye / Khedker, SOAP@PLDI 2013 standard worklist algorithm w/ calling context & value Volker Stolz Don t let data Go astray NWPT / 20

15 Standing on the Shoulders... Existing Go compiler infrastructure: SSA representation Basic blocks / call-graph Points-to analysis à la Andersen Interprocedural, context-sensitive analysis: Padhye / Khedker, SOAP@PLDI 2013 standard worklist algorithm w/ calling context & value... or their toes? Volker Stolz Don t let data Go astray NWPT / 20

16 Our analysis example main() h(f *os.file) (c string, r int) n 1 c 1 n 3 a := Hello World b := g(a) fmt.print(b) // sink f, := os.openfile(./pw.txt ) n 4 n 5 b := make([]byte, 8) r, = f.read(b) c = string(b[:]) return c 2 s, n := h(f) n 6 fmt.print(s) for n > 0 // sink c 3 s, n = h(f) c 4 t := g(s) n 8 fmt.print(t) // sink g(a string) (b string) n 9 exit n 2 n 7 b = a return Volker Stolz Don t let data Go astray NWPT / 20

17 Our analysis example main() h(f *os.file) (c string, r int) n 1 c 1 n 3 a := Hello World b := g(a) fmt.print(b) // sink f, := os.openfile(./pw.txt ) n 4 n 5 b := make([]byte, 8) r, = f.read(b) c = string(b[:]) return c 2 s, n := h(f) n 6 c 3 fmt.print(s) for n > 0 s, n = h(f) // sink Untainted c 4 t := g(s) n 8 fmt.print(t) // sink g(a string) (b string) n 9 exit n 2 n 7 b = a return Volker Stolz Don t let data Go astray NWPT / 20

18 Our analysis example main() h(f *os.file) (c string, r int) n 1 c 1 n 3 a := Hello World b := g(a) fmt.print(b) // sink f, := os.openfile(./pw.txt ) n 4 n 5 b := make([]byte, 8) r, = f.read(b) c = string(b[:]) return c 2 n 6 c 3 s, n := h(f) fmt.print(s) for n > 0 s, n = h(f) // sink Tainted Untainted c 4 t := g(s) n 8 fmt.print(t) // sink g(a string) (b string) n 9 exit n 2 n 7 b = a return Volker Stolz Don t let data Go astray NWPT / 20

19 Channel Handling Q: Can we handle flow through channels in the same framework? n 1 x := Hello World n 2 ch := make(chan, string) c 1 go f(ch) fn 1 func f(ch chan string) n 3 sink(x) fn 2 y := ch n 4 x = tainted fn 3 sink(y) n 5 ch x n 6... Volker Stolz Don t let data Go astray NWPT / 20

20 Channel Handling Q: Can we handle flow through channels in the same framework? n 1 x := Hello World n 2 ch := make(chan, string) c 1 go f(ch) fn 1 func f(ch chan string) TA(n 5 ) ch = [ch S(x)] n 3 sink(x) fn 2 y := ch n 4 x = tainted fn 3 sink(y) n 5 ch x n 6... A: Add feedback from writes to reads Overapproximation; only feed back taint value of channel along this edge Volker Stolz Don t let data Go astray NWPT / 20

21 Our analysis Nielson-style data-flow analysis (monotone framework) Analysis info of a node = transfer function of node applied to (union over all incoming flows) TA(l) = Φ(S, N l ) where S = {TA(l ) (l, l) flow (P)} and N l nodes(p). Volker Stolz Don t let data Go astray NWPT / 20

22 Our analysis Assignments, from/to struct members, calls: Φ(S, [x := e] l ) = φ(s, x, e, l) Volker Stolz Don t let data Go astray NWPT / 20

23 Our analysis Assignments, from/to struct members, calls: Φ(S, [x := e] l ) = φ(s, x, e, l) φ(s, x, y.f, l) = S[x {TA(l ) y [y := e] l [y ch] l f.a. (l : y ) aliases(l : y)}] φ(s, x.f, y.f, l) = S[x.f {TA(l ) [y := e] l [y ch] l y f.a. (l : y ) aliases(l : y)}] φ(s, x, v 1 v 2, l) = { S[x 1] if v1 is a source S[x Φ v 1 exit ] otherwise aliases(l : u) produced by existing context-insensitive pta. tainting a struct-member taints the entire struct same for slices & built-in key/value map data structure Volker Stolz Don t let data Go astray NWPT / 20

24 Our analysis Assignments, from/to struct members, calls: Φ(S, [x := e] l ) = φ(s, x, e, l) φ(s, x, y.f, l) = S[x {TA(l ) y [y := e] l [y ch] l f.a. (l : y ) aliases(l : y)}] φ(s, x, v 1 v 2, l) = { S[x 1] if v1 is a source S[x Φ v 1 exit ] otherwise aliases(l : u) produced by existing context-insensitive pta. tainting a struct-member taints the entire struct same for slices & built-in key/value map data structure Volker Stolz Don t let data Go astray NWPT / 20

25 Channel Handling Φ(S, [x ch] l ) = S[ch S(x)] Φ(S, [x ch] l ) = S[x A] where A = {TA(l ) ch [x ch ] l } f.a. (l : ch ) aliases(l : ch) existing pta knows about channels we lose information about ordering of messages in channels (even in obvious cases) Volker Stolz Don t let data Go astray NWPT / 20

26 Sanitizers & Monitoring Taint analysis with only sources and sinks too restrictive Certain operations untaint data: SQL injections: filter out dangerous characters Password example: data flowing through hash-function sanitized Natural extension through third set of operations: sanitizers Interesting questions: Where to report tainted flow at runtime (early/late) Minimize taint-tagging at runtime Automated placement of sanitizers to repair programs (for C : Livshits, POPL 13) Volker Stolz Don t let data Go astray NWPT / 20

27 Practical Evaluation Extensive list of sources & sinks from Eric s security projects Test suite with hand-crafted examples: Runtime bounded by lattice height, calling contexts Real-world case study: well... fine-grained analysis descends into imported packages (fmt.print) case study requires domain specific properties (and by now everyone is avoiding SQL injections... ) SSA-based infrastructure requires setup that actually compiles (git clone github.com/random/goproject not enough) High-profile targets: Docker, Dropbox, Android development Volker Stolz Don t let data Go astray NWPT / 20

28 Summary: Challenge in Go Traditional data-flow analysis, for a newer language Information flow: sources to sinks (via sanitizers) mixes problems from C (structs, pointers) with problems from π-calculus (channels) combination of off-the-shelf components (SSA representation & pta from Go compiler infrastructure, vasco interprocedural analysis) 1 More research is needed on... : fine tune white-list/black-list for imports or: precompute & cache standard libraries? collect (performance) data on real-world Go code 1 E. Bodden, K. I Pun, M. Steffen, V. Stolz, A.-K. Wickert Information Flow Analysis for Go, ISoLA Volker Stolz Don t let data Go astray NWPT / 20

Summary: Challenge in Go Traditional data-flow analysis, for a newer language Information flow: sources to sinks (via sanitizers) mixes problems from C (structs, pointers) with problems from

29 Summary: Challenge in Go Traditional data-flow analysis, for a newer language Information flow: sources to sinks (via sanitizers) mixes problems from C (structs, pointers) with problems from π-calculus (channels) combination of off-the-shelf components (SSA representation & pta from Go compiler infrastructure, vasco interprocedural analysis) 1 More research is needed on... : fine tune white-list/black-list for imports or: precompute & cache standard libraries? collect (performance) data on real-world Go code 1 E. Bodden, K. I Pun, M. Steffen, V. Stolz, A.-K. Wickert Information Flow Analysis for Go, ISoLA Volker Stolz Don t let data Go astray NWPT / 20

The Go Programming Language. Frank Roberts

The Go Programming Language. Frank Roberts The Go Programming Language Frank Roberts frank.roberts@uky.edu - C++ (1983), Java (1995), Python (1991): not modern - Java is 18 years old; how has computing changed in 10? - multi/many core - web programming