A concrete memory model for CompCert

Transcription

1 A concrete memory model for CompCert Frédéric Besson Sandrine Blazy Pierre Wilke Rennes, France P. Wilke A concrete memory model for CompCert 1 / 28

2 CompCert real-world C to ASM compiler used in industry (commercialised by AbsInt) proven correct in Coq: it does not introduce bugs! C Clight Cminor RTL ASM P. Wilke A concrete memory model for CompCert 2 / 28

3 CompCert real-world C to ASM compiler used in industry (commercialised by AbsInt) proven correct in Coq: it does not introduce bugs! C Clight Cminor RTL ASM P. Wilke A concrete memory model for CompCert 2 / 28

4 CompCert real-world C to ASM compiler used in industry (commercialised by AbsInt) proven correct in Coq: it does not introduce bugs! C Clight Cminor RTL ASM Each language has a Formal Semantics i.e. a mathematical meaning for programs P. Wilke A concrete memory model for CompCert 2 / 28

5 CompCert real-world C to ASM compiler used in industry (commercialised by AbsInt) proven correct in Coq: it does not introduce bugs! C Clight Cminor RTL ASM Each language has a Formal Semantics i.e. a mathematical meaning for programs Proof of semantic preservation For every source program S that has a defined semantics, If the compiler succeeds to generate a target program T, Then T has the same behavior as S. P. Wilke A concrete memory model for CompCert 2 / 28

6 CompCert real-world C to ASM compiler used in industry (commercialised by AbsInt) proven correct in Coq: it does not introduce bugs! Memory model C Clight Cminor RTL ASM Each language has a Formal Semantics i.e. a mathematical meaning for programs Proof of semantic preservation For every source program S that has a defined semantics, If the compiler succeeds to generate a target program T, Then T has the same behavior as S. P. Wilke A concrete memory model for CompCert 2 / 28

7 Goal: Make the semantics of C more defined Why did C leave some behaviors undefined? Portability Performance Why do we want to make it more defined? real-life programs use features that are undefined, according to C the compilation theorem will be more useful What kind of undefined behaviors do we aim at? undefined pointer arithmetic, i.e. bitwise operators use of uninitialised memory Our starting point: CompCert P. Wilke A concrete memory model for CompCert 3 / 28

8 An example of low-level C program in CompCert b p int main(){ int * p = (int *) malloc (sizeof (int)); *p = 42; int * q = p 5; int * r = (q» 3) «3; return *r; } b q b r P. Wilke A concrete memory model for CompCert 4 / 28

9 An example of low-level C program in CompCert b p b q int main(){ int * p = (int *) malloc (sizeof (int)); *p = 42; int * q = p 5; int * r = (q» 3) «3; return *r; } (b,0) b b r P. Wilke A concrete memory model for CompCert 4 / 28

10 An example of low-level C program in CompCert b p b q int main(){ int * p = (int *) malloc (sizeof (int)); *p = 42; int * q = p 5; int * r = (q» 3) «3; return *r; } (b,0) b 42 b r P. Wilke A concrete memory model for CompCert 4 / 28

11 An example of low-level C program in CompCert b p b q int main(){ int * p = (int *) malloc (sizeof (int)); *p = 42; int * q = p 5; int * r = (q» 3) «3; return *r; } (b,0) b 42 b r Bitwise operators on pointers are undefined behavior! CompCert [JAR 09], KCC [POPL 12], Krebbers [POPL 14], Norrish [PhD 98]: undefined behavior Kang et al. [PLDI 15]: don t model bitwise operators P. Wilke A concrete memory model for CompCert 4 / 28

12 Contributions Previous work [APLAS 14]: A memory model for low-level programs This work: integration of the memory model inside CompCert correctness proofs of the memory model correctness proofs of the transformations of the frontend (up to Cminor) P. Wilke A concrete memory model for CompCert 5 / 28

13 Outline 1 CompCert s memory model 2 New features of the memory model 3 Consistency of the memory models 4 CompCert proof: Overview 5 Conclusion P. Wilke A concrete memory model for CompCert 6 / 28

14 Outline 1 CompCert s memory model 2 New features of the memory model 3 Consistency of the memory models 4 CompCert proof: Overview 5 Conclusion P. Wilke A concrete memory model for CompCert 7 / 28

15 New features of the memory model Symbolic expressions val ::= i (b, o) not expressive enough We change the semantic domain to: expr ::= val op 1 expr expr op 2 expr P. Wilke A concrete memory model for CompCert 8 / 28

16 New features of the memory model Symbolic expressions val ::= i (b, o) not expressive enough We change the semantic domain to: Alignment constraints expr ::= val op 1 expr expr op 2 expr We need information about some bits of the concrete address of a pointer The alloc primitive takes an extra parameter mask, such that: A(b) & mask = A(b) P. Wilke A concrete memory model for CompCert 8 / 28

17 Interaction with the memory model What is the semantics of reading from memory: *p? In CompCert, p is evaluated into a pointer (b,i), then we can use load(m,b,i) In our model, p is a symbolic expression. It needs to be transformed into a pointer so that we can use load. normalise : mem expr val We need to modify the semantics to include calls to normalise memory accesses (load and store) conditionnal branches P. Wilke A concrete memory model for CompCert 9 / 28

18 Back to the example b p int main(){ int * p = (int *) malloc (sizeof (int)); *p = 42; int * q = p 5; int * r = (q» 3) «3; return *r; } (b,0) b 8 42 b q b r P. Wilke A concrete memory model for CompCert 10 / 28

19 Back to the example b p b q int main(){ int * p = (int *) malloc (sizeof (int)); *p = 42; int * q = p 5; int * r = (q» 3) «3; return *r; } (b,0) (b,0) 5 b 8 42 b r P. Wilke A concrete memory model for CompCert 10 / 28

20 Back to the example b p b q b r int main(){ int * p = (int *) malloc (sizeof (int)); *p = 42; int * q = p 5; int * r = (q» 3) «3; return *r; } (b,0) (b,0) 5 ( ((b,0) 5 ) 3 ) 3 b 8 42 P. Wilke A concrete memory model for CompCert 10 / 28

21 Back to the example b p b q int main(){ int * p = (int *) malloc (sizeof (int)); *p = 42; int * q = p 5; int * r = (q» 3) «3; return *r; } (b,0) (b,0) 5 b 8 42 b r ( ((b,0) 5 ) 3 ) 3 normalise (b,0) P. Wilke A concrete memory model for CompCert 10 / 28

22 Back to the example b p b q int main(){ int * p = (int *) malloc (sizeof (int)); *p = 42; int * q = p 5; int * r = (q» 3) «3; return *r; } (b,0) (b,0) 5 b 8 42 b r ( ((b,0) 5 ) 3 ) 3 normalise (b,0) P. Wilke A concrete memory model for CompCert 10 / 28

23 Normalisation specification: concrete memories (b 2,2) 5 Abstract memory m cm 1 cm 2 Concrete memories of m cm i m cm 3 cm 4 cm 5 cm 6 range : ]0;55[ no overlap alignment P. Wilke A concrete memory model for CompCert 11 / 28

24 Normalisation: example 1 e = ((( b,0) 5) 3) 3 cm 1 8 = (b,o) cm1 cm 2 8 = (b,o) cm2 cm 3 16 = (b,o) cm3 cm 4 24 = (b,o) cm4 cm 5 32 = (b,o) cm5 cm 6 32 = (b,o) cm e cm1 = (((cm 1 (b) + 0) 5) 3) = ((8 5) 3) = ((0b1000 5) 3) 3 = (0b1101 3) 3 = 0b = 0b1000 = 8 = cm 1 (b) i, e cmi = cm i (b), hence e normalises into (b,0) P. Wilke A concrete memory model for CompCert 12 / 28

25 Normalisation: example 2 e = ( b,0) > ( b,0) cm 1 cm 2 cm 3 cm 4 cm 5 cm 6 true true true false false false There is no v such that i, e cmi = v cmi, hence e doesn t normalise P. Wilke A concrete memory model for CompCert 13 / 28

26 CompCert with symbolic expressions expr ::= val op 1 expr expr op 2 expr b 1 (b 2,2) b 3 5 b (b,o) 5 Memory model C Clight Cminor RTL ASM S S S S S P. Wilke A concrete memory model for CompCert 14 / 28

27 Outline 1 CompCert s memory model 2 New features of the memory model 3 Consistency of the memory models 4 CompCert proof: Overview 5 Conclusion P. Wilke A concrete memory model for CompCert 15 / 28

28 How does our model compare to CompCert? x(t) x(t) Behaviors in CompCert t Behaviors with symbolic expressions t We are an extension of CompCert P. Wilke A concrete memory model for CompCert 16 / 28

29 How does our model compare to CompCert? Formally, Lemma expr_add_ok: v 1 v 2 m v, sem_add v 1 v 2 m = v e, sem_add_expr v 1 v 2 m = e normalise m e = v. If the addition of v 1 and v 2 succeeds in CompCert, Then it should succeed in our model as well, And the expression we compute should normalise into the same value. P. Wilke A concrete memory model for CompCert 17 / 28

30 Discovery of bugs 2 cases where our model disagrees with CompCert Bug in CompCert 2.4: Pointer comparison to NULL (fixed in CompCert 2.5) Bug in our model: incorrect handling of pointers one past the end P. Wilke A concrete memory model for CompCert 18 / 28

31 Incorrect pointer comparison to NULL In CompCert: pointers are pairs (b, o) the NULL pointer is represented as the integer 0 p == 0 was incorrectly defined to always evaluate to false when p is a pointer. b But we need to check that o is a valid offset of b (b,o) cm = cm(b) + o is not zero only in that case otherwise (b, 8) evaluates to zero P. Wilke A concrete memory model for CompCert 19 / 28

32 Outline 1 CompCert s memory model 2 New features of the memory model 3 Consistency of the memory models 4 CompCert proof: Overview 5 Conclusion P. Wilke A concrete memory model for CompCert 20 / 28

33 Overview of CompCert architecture C Clight C minor Cminor frontend backend Linear LTL RTL CminorSel Mach ASM : conserves the memory layout : modifies the memory layout P. Wilke A concrete memory model for CompCert 21 / 28

34 Memory injections: a generic memory transformation In CompCert C, each local variable has its own block. During the compilation these variables are merged into a stack frame. mem_inject f m m m m f b b 1 b 2 1 (b 3,o) f f 1 (b,o + δ 2 ) 37 δ 1 δ 2 b 3 37 Adapting to symbolic expressions: generalization of the injection over values lots of proofs to adapt (relation with normalisation) P. Wilke A concrete memory model for CompCert 22 / 28

35 Memory injections - Central theorem Theorem norm_inject: f m m e e (Minj: inject f m m ) (Einj: expr_inject f e e ), val_inject f (normalise m e) (normalise m e ). We can show that: v,val_inject f (normalise m e) v Let s now prove that: normalise m e = v cm m, e cm = v cm From the specification of the normalisation of e in m we know: cm m, e cm = normalise m e cm We need a theorem relating evaluations in cm and cm! P. Wilke A concrete memory model for CompCert 23 / 28

36 Memory injections - Evaluation mem_inject f m m Concrete memories of m Concrete memories of m pre_cm(f,cm ): recovers a concrete memory as it was before injection Definition pre_cm f cm := fun (b: block) let (b, delta) := f b in cm b + delta. Theorem expr_inject_eval: f cm e e (Einj: expr_inject f e e ), e cm = e pre_cm(f,cm ). P. Wilke A concrete memory model for CompCert 24 / 28

37 Memory injections - Central theorem Theorem norm_inject: f m m e e (Minj: inject f m m ) (Einj: expr_inject f e e ), val_inject f (normalise m e) (normalise m e ). Concrete memories of m Concrete memories of m We are left to prove: cm m, e cm = v cm expr_inject_eval : expr_inject f e e e cm = e pre_cm(f,cm ) We rewrite both sides using expr_inject_eval, the goal becomes: cm m, e pre_cm(f,cm ) = normalise m e pre_cm(f,cm ) From the specification of the normalisation of e in m we know: which solves our goal. cm m, e cm = normalise m e cm P. Wilke A concrete memory model for CompCert 25 / 28

38 Outline 1 CompCert s memory model 2 New features of the memory model 3 Consistency of the memory models 4 CompCert proof: Overview 5 Conclusion P. Wilke A concrete memory model for CompCert 26 / 28

39 Conclusion A semantics for C more precise than CompCert s compatible with CompCert nearly as proven correct as CompCert Future directions finish the proof by adapting the last remaining unproven pass add a more concrete assembly language to the certified compilation chain plug back in optimizations at RTL level (precision improvement?, still sound?) P. Wilke A concrete memory model for CompCert 27 / 28

40 Questions? P. Wilke A concrete memory model for CompCert 28 / 28