source: Papers/fopara2013/fopara13.tex @ 3430

Last change on this file since 3430 was 3430, checked in by campbell, 6 years ago

Revisions up to the end of section 3.

File size: 64.0 KB
12            bookmarks,bookmarksopen,bookmarksdepth=2]{hyperref}
31        {\setlength{\fboxsep}{5pt}
32                \setlength{\mylength}{\linewidth}%
33                \addtolength{\mylength}{-2\fboxsep}%
34                \addtolength{\mylength}{-2\fboxrule}%
35                \Sbox
36                \minipage{\mylength}%
37                        \setlength{\abovedisplayskip}{0pt}%
38                        \setlength{\belowdisplayskip}{0pt}%
39                }%
40                {\endminipage\endSbox
41                        \[\fbox{\TheSbox}\]}
43\title{Certified Complexity (CerCo)\thanks{The project CerCo acknowledges the
44financial support of the Future and Emerging Technologies (FET) programme within
45the Seventh Framework Programme for Research of the European Commission, under
46FET-Open grant number: 243881}}
49R.~M. Amadio$^{4}$ \and
51N.~Ayache$^{3,4}$ \and
53F.~Bobot$^{3,4}$ \and
55J.~P. Boender$^1$ \and
57B.~Campbell$^2$ \and
59I.~Garnier$^2$ \and
61A.~Madet$^4$ \and
63J.~McKinna$^2$ \and
65D.~P. Mulligan$^1$ \and
67M.~Piccolo$^1$ \and
69R.~Pollack$^2$ \and
71Y.~R\'egis-Gianas$^{3,4}$ \and
73C.~Sacerdoti Coen$^1$ \and
75I.~Stark$^2$ \and
78\institute{Dipartimento di Informatica - Scienza e Ingegneria, Universit\'a di
79Bologna \and
80LFCS, School of Informatics, University of Edinburgh
81\and INRIA (Team $\pi$r2 )
83Universit\`e Paris Diderot
92% Brian: I've changed "without loss of accuracy" because we discuss
93% important precision/complexity tradeoffs for advanced architectures
94% in the dependent labelling section.
96This paper provides an overview of the FET-Open Project CerCo
97(`Certified Complexity'). The project's main achievement is
98the development of a technique for analysing non-functional
99properties of programs (time, space) at the source level, with little or no loss of accuracy
100and with a small trusted code base. The main software component
101developed is a certified compiler producing cost annotations. The compiler
102translates source code to object code and produces an instrumented copy of the
103source code. This instrumentation exposes at
104the source level---and tracks precisely---the actual (non-asymptotic)
105computational cost of the input program. Untrusted invariant generators and trusted theorem provers
106can then be used to compute and certify the parametric execution time of the
110% ---------------------------------------------------------------------------- %
111% SECTION                                                                      %
112% ---------------------------------------------------------------------------- %
115%\paragraph{Problem statement.}
116Computer programs can be specified with both
117functional constraints (what a program must do) and non-functional constraints
118(what resources---time, space, etc.---the program may use).  In the current
119state of the art, functional properties are verified for high-level source code
120by combining user annotations (e.g. preconditions and invariants) with a
121multitude of automated analyses (invariant generators, type systems, abstract
122interpretation, theorem proving, and so on). By contrast, non-functional properties
123are generally checked on low-level object code, but also demand information
124about high-level functional behaviour that must somehow be reconstructed.
126This situation presents several problems: 1) it can be hard to infer this
127high-level structure in the presence of compiler optimisations; 2) techniques
128working on object code are not useful in early development, yet problems
129detected later are more expensive to tackle; 3) parametric cost analysis is very
130hard: how can we reflect a cost that depends on the execution state (e.g. the
131value of a register or a carry bit) to a cost that the user can understand
132looking at source code? 4) performing functional analysis on the object code
133makes it hard for the programmer to provide information, leading to less precision in
134the estimates.
136\paragraph{Vision and approach.} We want to reconcile functional and
137non-functional analyses: to share information and perform both at the same time
138on source code.
140What has previously prevented this approach is the lack of a uniform and precise
141cost model for high-level code: 1) each statement occurrence is compiled
142differently and optimisations may change control flow; 2) the cost of an object
143code instruction may depend on the runtime state of hardware components like
144pipelines and caches, all of which are not visible in the source code.
146To solve the issue, we envision a new generation of compilers able to keep track
147of program structure through compilation and optimisation, and to exploit that
148information to define a cost model for source code that is precise, non-uniform,
149and which accounts for runtime state. With such a source-level cost model we can
150reduce non-functional verification to the functional case and exploit the state
151of the art in automated high-level verification~\cite{survey}. The techniques
152currently used by the Worst Case Execution Time (WCET) community, which perform the analysis on object code,
153are still available but can now be coupled with an additional source-level
154analysis. Where the approach produces precise cost models too complex to reason
155about, safe approximations can be used to trade complexity with precision.
156Finally, source code analysis can be used during the early stages of development, when
157components have been specified but not implemented: source code modularity means
158that it is enough to specify the non-functional behaviour of unimplemented
161\paragraph{Contributions.} We have developed what we refer to as \emph{the labeling approach} \cite{labeling}, a
162technique to implement compilers that induce cost models on source programs by
163very lightweight tracking of code changes through compilation. We have studied
164how to formally prove the correctness of compilers implementing this technique.
165We have implemented such a compiler from C to object binaries for the 8051
166micro-controller, and verified it in an interactive theorem prover.
168To demonstrate source-level verification of costs we have implemented
169a Frama-C plugin \cite{framac} that invokes the compiler on a source
170program and uses this to generate invariants on the high-level source
171that correctly model low-level costs. The plugin certifies that the
172program respects these costs by calling automated theorem provers, a
173new and innovative technique in the field of cost analysis. Finally,
174we have conduct several case studies, including showing that the
175plugin can automatically compute and certify the exact reaction time
176of Lustre~\cite{lustre} data flow programs compiled into C.
178\section{Project context and approach}
179Formal methods for verification of functional properties of programs have
180now reached a level of maturity and automation that is facilitating a slow but
181increasing adoption in production environments. For safety critical code, it is
182becoming commonplace to combine rigorous software engineering methodologies and testing
183with static analysis, taking the strong points of each approach and mitigating
184their weaknesses. Of particular interest are open frameworks
185for the combination of different formal methods, where the programs can be
186progressively specified and are continuously enriched with new safety
187guarantees: every method contributes knowledge (e.g. new invariants) that
188becomes an assumption for later analysis.
190The scenario for the verification of non-functional properties (time spent,
191memory used, energy consumed) is bleaker.
192% and the future seems to be getting even worse.
193Most industries verify that real time systems meet their deadlines
194by simply performing many runs of the system and timing their execution,  computing the
195maximum time and adding an empirical safety margin, claiming the result to be a
196bound for the WCET of the program. Formal methods and software to statically
197analyse the WCET of programs exist, but they often produce bounds that are too
198pessimistic to be useful. Recent advancements in hardware architecture
199have been
200focused on the improvement of the average case performance, not the
201predictability of the worst case. Execution time is becoming increasingly
202dependent on execution history and the internal state of
203hardware components like pipelines and caches. Multi-core processors and non-uniform
204memory models are drastically reducing the possibility of performing
205static analysis in isolation, because programs are less and less time
206composable. Clock-precise hardware models are necessary for static analysis, and
207obtaining them is becoming harder due to the increased sophistication
208of hardware design.
210Despite the aforementioned problems, the need for reliable real time
211systems and programs is increasing, and there is increasing pressure
212from the research community towards the introduction of alternative
213hardware with more predictable behaviour, which would be more suitable
214for static analysis.  One example being investigated by the Proartis
215project~\cite{proartis} is to decouple execution time from execution
216history by introducing randomisation.
218In the CerCo project \cite{cerco} we do not try to address this problem, optimistically
219assuming that improvements in low-level timing analysis or architecture will make
220verification feasible in the longer term. The main objective of our work is
221instead to bring together static analysis of functional and non-functional
222properties, which, according to the current state of the art, are completely
223independent activities with limited exchange of information: while the
224functional properties are verified on the source code, the analysis of
225non-functional properties is entirely performed on the object code to exploit
226clock-precise hardware models.
228\subsection{Current object-code methods}
230Analysis currently takes place on object code for two main reasons.
231First, there cannot be a uniform, precise cost model for source
232code instructions (or even basic blocks). During compilation, high level
233instructions are torn apart and reassembled in context-specific ways so that
234identifying a fragment of object code and a single high level instruction is
235infeasible. Even the control flow of the object and source code can be very
236different as a result of optimisations, for example aggressive loop
237optimisations may completely transform source level loops. Despite the lack of a uniform, compilation- and
238program-independent cost model on the source language, the literature on the
239analysis of non-asymptotic execution time on high level languages that assumes
240such a model is growing and gaining momentum. However, unless we provide a
241replacement for such cost models, this literature's future practical impact looks
242to be minimal. Some hope has been provided by the EmBounded project \cite{embounded}, which
243compositionally compiles high-level code to a byte code that is executed by an
244interpreter with guarantees on the maximal execution time spent for each byte code
245instruction. This provides a uniform model at the expense of the model's
246precision (each cost is a pessimistic upper bound) and the performance of the
247executed code (because the byte code is interpreted compositionally instead of
248performing a fully non-compositional compilation).
250The second reason to perform the analysis on the object code is that bounding
251the worst case execution time of small code fragments in isolation (e.g. loop
252bodies) and then adding up the bounds yields very poor estimates because no
253knowledge of the hardware state on executing the fragment can be assumed. By
254analysing longer runs the bound obtained becomes more precise because the lack
255of information about the initial state has a relatively small impact.
257To calculate the cost of an execution, value and control flow analyses
258are required to bound the number of times each basic block is
259executed.  Current state
260of the art WCET technology such as AbsInt performs these analyses on the
261object code, where the logic of the program is harder to reconstruct
262and most information available at the source code level has been lost;
263see~\cite{stateart} for a survey. Imprecision in the analysis can lead to useless bounds. To
264augment precision, the tools ask the user to provide constraints on
265the object code control flow, usually in the form of bounds on the
266number of iterations of loops or linear inequalities on them. This
267requires the user to manually link the source and object code,
268translating his assumptions on the source code (which may be wrong) to
269object code constraints. The task is error prone and hard, especially
270in the presence of complex compiler optimisations.
272Traditional techniques for WCET that work on object code are also affected by
273another problem: they cannot be applied before the generation of the object
274code. Functional properties can be analysed in early development stages, while
275analysis of non-functional properties may come too late to avoid expensive
276changes to the program architecture.
280In CerCo we propose a radically new approach to the problem: we reject the idea
281of a uniform cost model and we propose that the compiler, which knows how the
282code is translated, must return the cost model for basic blocks of high level
283instructions. It must do so by keeping track of the control flow modifications
284to reverse them and by interfacing with processor timing analysis.
286By embracing compilation, instead of avoiding it like EmBounded did, a CerCo
287compiler can both produce efficient code and return costs that are
288as precise as the processor timing analysis can be. Moreover, our costs can be
289parametric: the cost of a block can depend on actual program data, on a
290summary of the execution history, or on an approximated representation of the
291hardware state. For example, loop optimizations may assign to a loop body a cost
292that is a function of the number of iterations performed. As another example,
293the cost of a block may be a function of the vector of stalled pipeline states,
294which can be exposed in the source code and updated at each basic block exit. It
295is parametricity that allows one to analyse small code fragments without losing
296precision: in the analysis of the code fragment we do not have to ignore the
297initial hardware state. On the contrary, we can assume that we know exactly which
298state (or mode, as the WCET literature calls it) we are in.
300The CerCo approach has the potential to dramatically improve the state of the
301art. By performing control and data flow analyses on the source code, the error
302prone translation of invariants is completely avoided. It is in fact performed
303by the compiler itself when it induces the cost model on the source language.
304Moreover, any available technique for the verification of functional properties
305can be immediately reused and multiple techniques can collaborate together to
306infer and certify cost invariants for the program. Parametric cost analysis
307becomes the default one, with non-parametric bounds used as a last resort when
308trading the complexity of the analysis with its precision. \emph{A priori}, no
309technique previously used in traditional WCET is lost: processor
310timing analyses can be used by the compiler on the object code, and the rest can be applied
311at the source code level.
312 Our approach can also work in the early
313stages of development by axiomatically attaching costs to unimplemented components.
316All software used to verify properties of programs must be as bug free as
317possible. The trusted code base for verification consists of the code that needs
318to be trusted to believe that the property holds. The trusted code base of
319state-of-the-art WCET tools is very large: one needs to trust the control flow
320analyser and the linear programming libraries it uses and also the formal models
321of the hardware. In CerCo we are moving the control flow analysis to the source
322code and we are introducing a non-standard compiler too. To reduce the trusted
323code base, we implemented a prototype and a static analyser in an interactive
324theorem prover, which was used to certify that the costs added to the source
325code are indeed those incurred by the hardware. Formal models of the
326hardware and of the high level source languages were also implemented in the
327interactive theorem prover. Control flow analysis on the source code has been
328obtained using invariant generators, tools to produce proof obligations from
329generated invariants and automatic theorem provers to verify the obligations. If
330the automatic provers are able to generate proof traces that can be
331independently checked, the only remaining component that enters the trusted code
332base is an off-the-shelf invariant generator which, in turn, can be proved
333correct using an interactive theorem prover. Therefore we achieve the double
334objective of allowing the use of more off-the-shelf components (e.g. provers and
335invariant generators) whilst reducing the trusted code base at the same time.
337%\paragraph{Summary of the CerCo objectives.} To summarize, the goal of CerCo is
338% to reconcile functional with non-functional analysis by performing them together
339% on the source code, sharing common knowledge about execution invariants. We want
340% to achieve the goal implementing a new generation of compilers that induce a
341% parametric, precise cost model for basic blocks on the source code. The compiler
342% should be certified using an interactive theorem prover to minimize the trusted
343% code base of the analysis. Once the cost model is induced, off-the-shelf tools
344% and techniques can be combined together to infer and prove parametric cost
345% bounds.
346%The long term benefits of the CerCo vision are expected to be:
347%1. the possibility to perform static analysis during early development stages
348%2.  parametric bounds made easier
349%3.  the application of off-the-shelf techniques currently unused for the
350% analysis of non-functional properties, like automated proving and type systems
351%4. simpler and safer interaction with the user, that is still asked for
352% knowledge, but on the source code, with the additional possibility of actually
353% verifying the provided knowledge
354%5. a reduced trusted code base
355%6. the increased accuracy of the bounds themselves.
357%The long term success of the project is hindered by the increased complexity of
358% the static prediction of the non-functional behaviour of modern hardware. In the
359% time frame of the European contribution we have focused on the general
360% methodology and on the difficulties related to the development and certification
361% of a cost model inducing compiler.
363\section{The typical CerCo workflow}
367char a[] = {3, 2, 7, 14};
368char threshold = 4;
370int count(char *p, int len) {
371  char j;
372  int found = 0;
373  for (j=0; j < len; j++) {
374    if (*p <= threshold)
375      found++;
376    p++;
377    }
378  return found;
381int main() {
382  return count(a,4);
386%  $\vcenter{\includegraphics[width=7.5cm]{interaction_diagram.pdf}}$
388    baseline={([yshift=-.5ex]current bounding box)},
389    element/.style={draw, text width=1.6cm, on chain, text badly centered},
390    >=stealth
391    ]
392{[start chain=going below, node distance=.5cm]
393\node [element] (cerco) {CerCo\\compiler};
394\node [element] (cost)  {CerCo\\cost plugin};
395{[densely dashed]
396\node [element] (ded)   {Deductive\\platform};
397\node [element] (check) {Proof\\checker};
400\coordinate [left=3.5cm of cerco] (left);
401{[every node/.style={above, text width=3.5cm, text badly centered,
402                     font=\scriptsize}]
403\draw [<-] ([yshift=-1ex]cerco.north west) coordinate (t) --
404    node {C source}
405    (t-|left);
406\draw [->] (cerco) -- (cost);
407\draw [->] ([yshift=1ex]cerco.south west) coordinate (t) --
408    node {C source+\color{red}{cost annotations}}
409    (t-|left) coordinate (cerco out);
410\draw [->] ([yshift=1ex]cost.south west) coordinate (t) --
411    node {C source+\color{red}{cost annotations}\\+\color{blue}{synthesized assertions}}
412    (t-|left) coordinate (out);
413{[densely dashed]
414\draw [<-] ([yshift=-1ex]ded.north west) coordinate (t) --
415    node {C source+\color{red}{cost annotations}\\+\color{blue}{complexity assertions}}
416    (t-|left) coordinate (ded in) .. controls +(-.5, 0) and +(-.5, 0) .. (out);
417\draw [->] ([yshift=1ex]ded.south west) coordinate (t) --
418    node {complexity obligations}
419    (t-|left) coordinate (out);
420\draw [<-] ([yshift=-1ex]check.north west) coordinate (t) --
421    node {complexity proof}
422    (t-|left) .. controls +(-.5, 0) and +(-.5, 0) .. (out);
423\draw [dash phase=2.5pt] (cerco out) .. controls +(-1, 0) and +(-1, 0) ..
424    (ded in);
426%% user:
427% head
428\draw (current bounding box.west) ++(-.2,.5)
429    circle (.2) ++(0,-.2) -- ++(0,-.1) coordinate (t);
430% arms
431\draw (t) -- +(-.3,-.2);
432\draw (t) -- +(.3,-.2);
433% body
434\draw (t) -- ++(0,-.4) coordinate (t);
435% legs
436\draw (t) -- +(-.2,-.6);
437\draw (t) -- +(.2,-.6);
440\caption{On the left: code to count the array's elements
441 that are less than or equal to the threshold. On the right: CerCo's interaction
442 diagram. CerCo's components are drawn solid.}
445We illustrate the workflow we envisage (on the right
446of~\autoref{test1}) on an example program (on the left
447of~\autoref{test1}).  The user writes the program and feeds it to the
448CerCo compiler, which outputs an instrumented version of the same
449program that updates global variables that record the elapsed
450execution time and the stack space usage.  The red lines in
451\autoref{itest1} introducing variables, functions and function calls
452starting with \lstinline'__cost' and \lstinline'__stack' are the instrumentation introduced by the
453compiler.  For example, the two calls at the start of
454\lstinline'count' say that 4 bytes of stack are required, and that it
455takes 111 cycles to reach the next cost annotation (in the loop body).
456The compiler measures these on the labeled object code that it generates.
458 The annotated program can then be enriched with complexity
459assertions in the style of Hoare logic, that are passed to a deductive
460platform (in our case Frama-C). We provide as a Frama-C cost plugin a
461simple automatic synthesiser for complexity assertions which can
462be overridden by the user to increase or decrease accuracy.  These are the blue
463comments starting with \lstinline'/*@' in \autoref{itest1}.  From the
464assertions, a general purpose deductive platform produces proof
465obligations which in turn can be closed by automatic or interactive
466provers, ending in a proof certificate.
468% NB: if you try this example with the live CD you should increase the timeout
470Twelve proof obligations are generated from~\autoref{itest1} (to prove
471that the loop invariant holds after one execution if it holds before,
472to prove that the whole program execution takes at most 1358 cycles,
473etc.).  Note that the synthesised time bound for \lstinline'count',
474$178+214*(1+\text{\lstinline'len'})$ cycles, is parametric in the length of
475the array. The CVC3 prover
476closes all obligations within half a minute on routine commodity
477hardware.  A simpler non-parametric version can be solved in a few
482        moredelim=[is][\color{blue}]{$}{$},
483        moredelim=[is][\color{red}]{|}{|},
484        lineskip=-2pt}
487|int __cost = 33, __stack = 5, __stack_max = 5;|
488|void __cost_incr(int incr) { __cost += incr; }|
489|void __stack_incr(int incr) {
490  __stack += incr;
491  __stack_max = __stack_max < __stack ? __stack : __stack_max;
494char a[4] = {3, 2, 7, 14};  char threshold = 4;
496/*@ behavior stack_cost:
497      ensures __stack_max <= __max(\old(__stack_max), 4+\old(__stack));
498      ensures __stack == \old(__stack);
499    behavior time_cost:
500      ensures __cost <= \old(__cost)+(178+214*__max(1+\at(len,Pre), 0));
502int count(char *p, int len) {
503  char j;  int found = 0;
504  |__stack_incr(4);  __cost_incr(111);|
505  $__l: /* internal */$
506  /*@ for time_cost: loop invariant
507        __cost <= \at(__cost,__l)+
508                  214*(__max(\at((len-j)+1,__l), 0)-__max(1+(len-j), 0));
509      for stack_cost: loop invariant
510        __stack_max == \at(__stack_max,__l);
511      for stack_cost: loop invariant
512        __stack == \at(__stack,__l);
513      loop variant len-j;
514  */
515  for (j = 0; j < len; j++) {
516    |__cost_incr(78);|
517    if (*p <= threshold) { |__cost_incr(136);| found ++; }
518    else { |__cost_incr(114);| }
519    p ++;
520  }
521  |__cost_incr(67);  __stack_incr(-4);|
522  return found;
525/*@ behavior stack_cost:
526      ensures __stack_max <= __max(\old(__stack_max), 6+\old(__stack));
527      ensures __stack == \old(__stack);
528    behavior time_cost:
529      ensures __cost <= \old(__cost)+1358;
531int main(void) {
532  int t;
533  |__stack_incr(2);  __cost_incr(110);|
534  t = count(a,4);
535  |__stack_incr(-2);|
536  return t;
539\caption{The instrumented version of the program in \autoref{test1},
540 with instrumentation added by the CerCo compiler in red and cost invariants
541 added by the CerCo Frama-C plugin in blue. The \texttt{\_\_cost},
542 \texttt{\_\_stack} and \texttt{\_\_stack\_max} variables hold the elapsed time
543in clock cycles and the current and maximum stack usage. Their initial values
544hold the clock cycles spent in initialising the global data before calling
545\texttt{main} and the space required by global data (and thus unavailable for
546the stack).
550\section{Main scientific and technical results}
551First we describe the basic labeling approach and our compiler
552implementations that use it.  This is suitable for basic architectures
553with simple cost models.  Then we will discuss the dependent labeling
554extension which is suitable for more advanced processor architectures
555and compiler optimisations.  At the end of this section we will
556demonstrate automated high level reasoning about the source level
557costs provided by the compilers.
559% \emph{Dependent labeling~\cite{?}.} The basic labeling approach assumes a
560% bijective mapping between object code and source code O(1) blocks (called basic
561% blocks). This assumption is violated by many program optimizations (e.g. loop
562% peeling and loop unrolling). It also assumes the cost model computed on the
563% object code to be non parametric: every block must be assigned a cost that does
564% not depend on the state. This assumption is violated by stateful hardware like
565% pipelines, caches, branch prediction units. The dependent labeling approach is
566% an extension of the basic labeling approach that allows to handle parametric
567% cost models. We showed how the method allows to deal with loop optimizations and
568% pipelines, and we speculated about its applications to caches.
570% \emph{Techniques to exploit the induced cost model.} Every technique used for
571% the analysis of functional properties of programs can be adapted to analyse the
572% non-functional properties of the source code instrumented by compilers that
573% implement the labeling approach. In order to gain confidence in this claim, we
574% showed how to implement a cost invariant generator combining abstract
575% interpretation with separation logic ideas \cite{separation}. We integrated
576% everything in the Frama-C modular architecture, in order to compute proof
577% obligations from functional and cost invariants and to use automatic theorem
578% provers on them. This is an example of a new technique that is not currently
579% exploited in WCET analysis. It also shows how precise functional invariants
580% benefits the non-functional analysis too. Finally, we show how to fully
581% automatically analyse the reaction time of Lustre nodes that are first compiled
582% to C using a standard Lustre compiler and then processed by a C compiler that
583% implements the labeling approach.
585% \emph{The CerCo compiler.} This is a compiler from a large subset of the C
586% program to 8051/8052 object code,
587% integrating the labeling approach and a static analyser for 8051 executables.
588% The latter can be implemented easily and does not require dependent costs
589% because the 8051 microprocessor is a very simple one, with constant-cost
590% instruction. It was chosen to separate the issue of exact propagation of the
591% cost model from the orthogonal problem of the static analysis of object code
592% that may require approximations or dependent costs. The compiler comes in
593% several versions: some are prototypes implemented directly in OCaml, and they
594% implement both the basic and dependent labeling approaches; the final version
595% is extracted from a Matita certification and at the moment implements only the
596% basic approach.
598\subsection{The (basic) labeling approach}
599The labeling approach is the foundational insight that underlies all the developments in CerCo.
600It allows the tracking of the evolution of
601basic blocks during the compilation process in order to propagate the cost model from the
602object code to the source code without losing precision in the process.
603\paragraph{Problem statement.} Given a source program $P$, we want to obtain an
604instrumented source program $P'$,  written in the same programming language, and
605the object code $O$ such that: 1) $P'$ is obtained by inserting into $P$ some
606additional instructions to update global cost information like the amount of
607time spent during execution or the maximal stack space required; 2) $P$ and $P'$ 
608must have the same functional behaviour, i.e.\ they must produce that same output
609and intermediate observables; 3) $P$ and $O$ must have the same functional
610behaviour; 4) after execution and in interesting points during execution, the
611cost information computed by $P'$ must be an upper bound of the one spent by $O$ 
612to perform the corresponding operations (soundness property); 5) the difference
613between the costs computed by $P'$ and the execution costs of $O$ must be
614bounded by a program-dependent constant (precision property).
616\paragraph{The labeling software components.} We solve the problem in four
617stages \cite{labeling}, implemented by four software components that are used
618in sequence.
620The first component labels the source program $P$ by injecting label emission
621statements in appropriate positions to mark the beginning of basic blocks.
622% The
623% set of labels with their positions is called labeling.
624The syntax and semantics
625of the source programming language is augmented with label emission statements.
626The statement ``EMIT $\ell$'' behaves like a NOP instruction that does not affect the
627program state or control flow, but its execution is observable.
628% Therefore the observables of a run of a program becomes a stream
629% of label emissions: $\ell_1,\ldots,\ell_n$, called the program trace. We clarify the conditions
630% that the labeling must respect later.
632The second component is a labeling preserving compiler. It can be obtained from
633an existing compiler by adding label emission statements to every intermediate
634language and by propagating label emission statements during compilation. The
635compiler is correct if it preserves both the functional behaviour of the program
636and the traces of observables.
637% We may also ask that the function that erases the cost
638% emission statements commute with compilation. This optional property grants that
639% the labeling does not interfere with the original compiler behaviour. A further
640% set of requirements will be added later.
642The third component is a static object code analyser. It takes as input a labeled
643object code and it computes the scope of each of its label emission statements,
644i.e.\ the tree of instructions that may be executed after the statement and before
645a new label emission is encountered.
646It is a tree and not a sequence because the scope
647may contain a branching statement. In order to grant that such a finite tree
648exists, the object code must not contain any loop that is not broken by a label
649emission statement. This is the first requirement of a sound labeling. The
650analyser fails if the labeling is unsound. For each scope, the analyser
651computes an upper bound of the execution time required by the scope, using the
652maximum of the costs of the two branches in case of a conditional statement.
653Finally, the analyser computes the cost of a label by taking the maximum of the
654costs of the scopes of every statement that emits that label.
656The fourth and last component takes in input the cost model computed at step 3
657and the labelled code computed at step 1. It outputs a source program obtained
658by replacing each label emission statement with a statement that increments the
659global cost variable with the cost associated to the label by the cost model. 
660The obtained source code is the instrumented source code.
662\paragraph{Correctness.} Requirements 1, 2 and 3 of the problem statement
663obviously hold, with 2 and 3 being a consequence of the definition of a correct
664labeling preserving compiler. It is also obvious that the value of the global
665cost variable of an instrumented source code is at any time equal to the sum of
666the costs of the labels emitted by the corresponding labelled code. Moreover,
667because the compiler preserves all traces, the sum of the costs of the labels
668emitted in the source and target labelled code are the same. Therefore, to
669satisfy the fourth requirement, we need to grant that the time taken to execute
670the object code is equal to the sum of the costs of the labels emitted by the
671object code. We collect all the necessary conditions for this to happen in the
672definition of a sound labeling: a) all loops must be broken by a cost emission
673statement;  b) all program instructions must be in the scope of some cost
674emission statement. To satisfy also the fifth requirement, additional
675requirements must be imposed on the object code labeling to avoid all uses of
676the maximum in the cost computation: the labeling is precise if every label is
677emitted at most once and both branches of a conditional jump start with a label
678emission statement.
680The correctness and precision of the labeling approach only rely on the
681correctness and precision of the object code labeling. The simplest
682way to achieve them is to impose correctness and precision
683requirements also on the initial labeling produced by the first software
684component, and to ask the compiler to preserve these
685properties too. The latter requirement imposes serious limitations on the
686compilation strategy and optimizations: the compiler may not duplicate any code
687that contains label emission statements, like loop bodies. Therefore several
688loop optimisations like peeling or unrolling are prevented. Moreover, precision
689of the object code labeling is not sufficient \emph{per se} to obtain global
690precision: we implicitly assumed that a precise constant cost can be assigned
691to every instruction. This is not possible in
692the presence of stateful hardware whose state influences the cost of operations,
693like pipelines and caches. In Section~\ref{lab:deplabel} we will see an extension of the
694basic labeling approach to cover this situation.
696In CerCo we have developed several cost preserving compilers based on
697the labeling approach. Excluding an initial certified compiler for a
698`while' language, all remaining compilers target realistic source
699languages---a pure higher order functional language and a large subset
700of C with pointers, gotos and all data structures---and real world
701target processors---MIPS and the Intel 8051/8052 processor
702family. Moreover, they achieve a level of optimisation that ranges
703from moderate (comparable to GCC level 1) to intermediate (including
704loop peeling and unrolling, hoisting and late constant propagation).
705We describe the C compilers in detail in the following section.
707Two compilation chains were implemented for a purely functional
708higher-order language~\cite{labeling2}. The two main changes required
709to deal with functional languages are: 1) because global variables and
710updates are not available, the instrumentation phase produces monadic
711code to `update' the global costs; 2) the requirements for a sound and
712precise labeling of the source code must be changed when the
713compilation is based on CPS translations.  In particular, we need to
714introduce labels emitted before a statement is executed and also
715labels emitted after a statement is executed. The latter capture code
716that is inserted by the CPS translation and that would escape all
717label scopes.
719% Brian: one of the reviewers pointed out that standard Prolog implementations
720% do have some global state that apparently survives backtracking and might be
721% used.  As we haven't experimented with this, I think it's best to elide it
722% entirely.
724% Phases 1, 2 and 3 can be applied as well to logic languages (e.g. Prolog).
725% However, the instrumentation phase cannot: in standard Prolog there is no notion
726% of (global) variable whose state is not retracted during backtracking.
727% Therefore, the cost of executing computations that are later backtracked would
728% not be correctly counted in. Any extension of logic languages with
729% non-backtrackable state could support our labeling approach.
731\subsection{The CerCo C compilers}
732We implemented two C compilers, one implemented directly in OCaml and
733the other implemented in the interactive theorem prover
734Matita~\cite{matita}.  The first acted as a prototype for the second,
735but also supported MIPS and acted as a testbed for more advanced
736features such as the dependent labeling approach
737in Section~\ref{lab:deplabel}.
739The second C compiler is the \emph{Trusted CerCo Compiler}, whose cost
740predictions are formally verified. The code distributed
741is extracted OCaml code from the Matita implementation. In the rest of
742this section we will only focus on the Trusted CerCo Compiler, that only targets
743the C language and the 8051/8052 family, and that does not implement any
744advanced optimisations yet. Its user interface, however, is the same as the one
745of the other version, in order to trade safety with additional performances. In
746particular, the Frama-C CerCo plugin can work without recompilation
747with both of our C compilers.
749The 8051/8052 microprocessor is a very simple one, with constant-cost
750instructions. It was chosen to separate the issue of exact propagation of the
751cost model from the orthogonal problem of the static analysis of object code
752that may require approximations or dependent costs.
754The (trusted) CerCo compiler implements the following optimisations: cast
755simplification, constant propagation in expressions, liveness analysis driven
756spilling of registers, dead code elimination, branch displacement, and tunneling.
757The two latter optimisations are performed by our optimising assembler
758\cite{correctness}. The back-end of the compiler works on three address
759instructions, preferred to static single assignment code for the simplicity of
760the formal certification.
762The CerCo compiler is loosely based on the CompCert compiler \cite{compcert}, a
763recently developed certified compiler from C to the PowerPC, ARM and x86
764microprocessors. In contrast to CompCert, both the CerCo code and its
765certification are fully open source. Some data structures and language definitions for
766the front-end are directly taken from CompCert, while the back-end is a redesign
767of a compiler from Pascal to MIPS used by Fran\c{c}ois Pottier for a course at the
768Ecole Polytechnique.
770The main differences in the CerCo compiler are:
772\item All the intermediate languages include label emitting instructions to
773implement the labeling approach, and the compiler preserves execution traces.
774\item Traditionally compilers target an assembly language with additional
775macro-instructions to be expanded before assembly; for CerCo we need to go all
776the way down to object code in order to perform the static analysis. Therefore
777we integrated also an optimising assembler and a static analyser.
778\item In order to avoid the additional work of implementing a linker
779 and a loader, we do not support separate
780compilation and external calls. Adding them is orthogonal
781to the labeling approach and should not introduce extra problems.
782\item We target an 8-bit processor, in contrast to CompCert's 32-bit targets. This requires
783many changes and more compiler code, but it is not fundamentally more
784complex. The proof of correctness, however, becomes much harder.
785\item We target a microprocessor that has a non-uniform memory model, which is
786still often the case for microprocessors used in embedded systems and that is
787becoming common again in multi-core processors. Therefore the compiler has to
788keep track of data and it must move data between memory regions in the proper
789way. Moreover the size of pointers to different regions is not uniform. An
790additional difficulty was that the space available for the stack in internal
791memory in the 8051 is tiny, allowing only a minor number of nested calls. To
792support full recursion in order to test the CerCo tools also on recursive
793programs, the compiler implements a stack in external memory.
796\subsection{Formal certification of the CerCo compiler}
797We have formally
798certified in the Matita interactive proof assistant that the cost models induced on the source code by the
799Trusted CerCo Compiler correctly and precisely
800predict the object code behaviour. There are two cost models, one for execution
801time and one for stack space consumption. We show the correctness of the prediction
802only for those programs that do not exhaust the available machine stack space, a
803property that---thanks to the stack cost model---we can statically analyse on the
804source code using our Frama-C tool. The preservation of functional properties we
805take as an assumption, not itself formally proved in CerCo.  Other projects have
806already certified the preservation of functional semantics in similar compilers,
807and we have not attempted to directly repeat that work. In order to complete the
808proof for non-functional properties, we have introduced a new semantics for
809programming languages based on a new kind of structured observables with the
810relative notions of forward similarity and the study of the intensional
811consequences of forward similarity. We have also introduced a unified
812representation for back-end intermediate languages that was exploited to provide
813a uniform proof of forward similarity.
815The details on the proof techniques employed
817the proof sketch can be collected from the CerCo deliverables and papers.
818In this section we will only hint at the correctness statement, which turned
819out to be more complex than what we expected at the beginning.
821\paragraph{The statement.}
822Real time programs are often reactive programs that loop forever responding to
823events (inputs) by performing some computation followed by some action (output)
824and the return to the initial state. For looping programs the overall execution
825time does not make sense. The same happens for reactive programs that spend an
826unpredictable amount of time in I/O. What is interesting is the reaction time
827that measure the time spent between I/O events. Moreover, we are interested in
828predicting and ruling out crashes due to running out of space on certain
830Therefore we need to look for a statement that talks about sub-runs of a
831program. The most natural statement is that the time predicted on the source
832code and spent on the object code by two corresponding sub-runs are the same.
833The problem to solve to make this statement formal is how to identify the
834corresponding sub-runs and how to single out those that are meaningful.
835The solution we found is based on the notion of measurability. We say that a run
836has a \emph{measurable sub-run} when both the prefix of the sub-run and the
837sub-run do not exhaust the stack space, the number of function calls and returns
838in the sub-run is the same, the sub-run does not perform any I/O and the sub-run
839starts with a label emission statement and ends with a return or another label
840emission statements. The stack usage is estimated using the stack usage model
841that is computed by the compiler.
843The statement that we formally proved is: for each C run with a measurable
844sub-run, there exists an object code run with a sub-run, such that the observables
845of the pairs of prefixes and sub-runs are the same and the time spent by the
846object code in the sub-run is the same as the one predicted on the source code
847using the time cost model generated by the compiler.
848We briefly discuss the constraints for measurability. Not exhausting the stack
849space is a clear requirement of meaningfulness of a run, because source programs
850do not crash for lack of space while object code ones do. The balancing of
851function calls and returns is a requirement for precision: the labeling approach
852allows the scope of label emission statements to extend after function calls to
853minimize the number of labels. Therefore a label pays for all the instructions
854in a block, excluding those executed in nested function calls. If the number of
855calls/returns is unbalanced, it means that there is a call we have not returned
856to that could be followed by additional instructions whose cost has already been
857taken in account. To make the statement true (but less precise) in this
858situation, we could only say that the cost predicted on the source code is a
859safe bound, not that it is exact. The last condition on the entry/exit points of
860a run is used to identify sub-runs whose code correspond to a sequence of blocks
861that we can measure precisely. Any other choice would start or end the run in the
862middle of a block and we would be forced again to weaken the statement taking as
863a bound the cost obtained counting in all the instructions that precede the
864starting one in the block, or follow the final one in the block.
865I/O operations can be performed in the prefix of the run, but not in the
866measurable sub-run. Therefore we prove that we can predict reaction times, but
867not I/O times, as desired.
869\subsection{Dependent labeling}
871The core idea of the basic labeling approach is to establish a tight connection
872between basic blocks executed in the source and target languages. Once the
873connection is established, any cost model computed on the object code can be
874transferred to the source code, without affecting the code of the compiler or
875its proof. In particular, it is immediate that we can also transport cost models
876that associate to each label functions from hardware state to natural numbers.
877However, a problem arises during the instrumentation phase that replaces cost
878emission statements with increments of global cost variables. The global cost
879variable must be incremented with the result of applying the function associated
880to the label to the hardware state at the time of execution of the block.
881The hardware state comprises both the functional state that affects the
882computation (the value of the registers and memory) and the non-functional
883state that does not (the pipeline and cache contents, for example). The former is
884in correspondence with the source code state, but reconstructing the
885correspondence may be hard and lifting the cost model to work on the source code
886state is likely to produce cost expressions that are too complex to understand and reason about.
887Fortunately, in all modern architectures the cost of executing single
888instructions is either independent of the functional state or the jitter---the
889difference between the worst and best case execution times---is small enough
890to be bounded without losing too much precision. Therefore we only consider
891dependencies on the `non-functional' parts of the state.
893The non-functional state is not directly related to the high level
894state and does not influence the functional properties. What can be
895done is to expose key aspects of the non-functional state in the
896source code. We present here the basic intuition in a simplified form:
897the technical details that allow us to handle the general case are
898more complex and can be found in~\cite{paolo}. We add to the source
899code an additional global variable that represents the non-functional
900state and another one that remembers the last few labels emitted. The
901state variable must be updated at every label emission statement,
902using an update function which is computed during static analysis
903too. The update function associates to each label a function from the
904recently emitted labels and old state to the new state. It is computed
905by composing the semantics of every instruction in a basic block and
906restricting it to the non-functional part of the state.
908Not all the details of the non-functional state needs to be exposed, and the
909technique works better when the part of state that is required can be summarized
910in a simple data structure. For example, to handle simple but realistic
911pipelines it is sufficient to remember a short integer that encodes the position
912of bubbles (stuck instructions) in the pipeline. In any case, it is not necessary
913for the user to understand the meaning of the state to reason over the properties
914of the
915program. Moreover, at any moment the user, or the invariant generator tools that
916analyse the instrumented source code produced by the compiler, can decide to
917trade precision of the analysis for simplicity by approximating the
918cost with safe bounds that do not depend on the processor state. Interestingly, the functional analysis of
919the code could determine which blocks are executed more frequently in order to
920use more aggressive approximations for the ones that are executed least.
922Dependent labeling can also be applied to allow the compiler to duplicate
923blocks that contain labels (e.g. in loop optimisations)~\cite{paolo}. The effect is to assign
924a different cost to the different occurrences of a duplicated label. For
925example, loop peeling turns a loop into the concatenation of a copy of the loop
926body for the first iteration and the conditional execution of the
927loop for successive iterations. Further optimisations will compile the two
928copies of the loop differently, with the first body usually
929taking more time.
931By introducing a variable that keeps track of the iteration number, we can
932associate to the label a cost that is a function of the iteration number. The
933same technique works for loop unrolling without modifications: the function will
934assign a cost to the even iterations and another cost to the odd ones. The
935actual work to be done consists of introducing within the source code, and for each
936loop, a variable that counts the number of iterations. The loop optimisation code
937that duplicate the loop bodies must also modify the code to correctly propagate
938the update of the iteration numbers. The technical details are more complex and
939can be found in the CerCo reports and publications. The implementation, however,
940is quite simple (and forms part of our OCaml version of the compiler)
941and the changes to a loop optimising compiler are minimal.
943\subsection{Techniques to exploit the induced cost model}
944We now turn our attention to synthesising high-level costs, such as
945the execution time of a whole program.  We consider as our starting point source level costs
946provided by basic labeling, in other words annotations
947on the source code which are constants that provide a sound and sufficiently
948precise upper bound on the cost of executing the blocks after compilation to
949object code.
951The principle that we have followed in designing the cost synthesis tools is
952that the synthetic bounds should be expressed and proved within a general
953purpose tool built to reason on the source code. In particular, we rely on the
954Frama-C tool to reason on C code and on the Coq proof-assistant to reason on
955higher-order functional programs.
957This principle entails that: 1)
958The inferred synthetic bounds are indeed correct as long as the general purpose
959tool is; 2) there is no limitation on the class of programs that can be handled
960as long as the user is willing to carry on an interactive proof.
962Of course, automation is desirable whenever possible. Within this framework,
963automation means writing programs that give hints to the general purpose tool.
964These hints may take the form, say, of loop invariants/variants, of predicates
965describing the structure of the heap, or of types in a light logic. If these
966hints are correct and sufficiently precise the general purpose tool will produce
967a proof automatically, otherwise, user interaction is required.
969\paragraph{The Cost plugin and its application to the Lustre compiler.}
970Frama-C \cite{framac} is a set of analysers for C programs with a
971specification language called ACSL. New analyses can be dynamically added
972via a plugin system. For instance, the Jessie plugin allows deductive
973verification of C programs with respect to their specification in ACSL, with
974various provers as back-end tools.
975We developed the CerCo Cost plugin for the Frama-C platform as a proof of
976concept of an automatic environment exploiting the cost annotations produced by
977the CerCo compiler. It consists of an OCaml program which essentially
978takes the following actions: 1) it receives as input a C program, 2) it
979applies the CerCo compiler to produce a related C program with cost annotations,
9803) it applies some heuristics to produce a tentative bound on the cost of
981executing the C functions of the program as a function of the value of their
982parameters, 4) the user can then call the Jessie plugin to discharge the
983related proof obligations.
984In the following we elaborate on the soundness of the framework and the
985experiments we performed with the Cost tool on the C programs produced by a
986Lustre compiler.
988\paragraph{Soundness.} The soundness of the whole framework depends on the cost
989annotations added by the CerCo compiler, the synthetic costs produced by the
990cost plugin, the verification conditions (VCs) generated by Jessie, and the
991external provers discharging the VCs. The synthetic costs being in ACSL format,
992Jessie can be used to verify them. Thus, even if the added synthetic costs are
993incorrect (relatively to the cost annotations), the process as a whole is still
994correct: indeed, Jessie will not validate incorrect costs and no conclusion can
995be made about the WCET of the program in this case. In other terms, the
996soundness does not really depend on the action of the cost plugin, which can in
997principle produce any synthetic cost. However, in order to be able to actually
998prove a WCET of a C function, we need to add correct annotations in a way that
999Jessie and subsequent automatic provers have enough information to deduce their
1000validity. In practice this is not straightforward even for very simple programs
1001composed of branching and assignments (no loops and no recursion) because a fine
1002analysis of the VCs associated with branching may lead to a complexity blow up.
1003\paragraph{Experience with Lustre.} Lustre is a data-flow language for programming
1004synchronous systems, with a compiler which targets C. We designed a
1005wrapper for supporting Lustre files.
1006The C function produced by the compiler implements the step function of the
1007synchronous system and computing the WCET of the function amounts to obtain a
1008bound on the reaction time of the system. We tested the Cost plugin and the
1009Lustre wrapper on the C programs generated by the Lustre compiler. For programs
1010consisting of a few hundred lines of code, the cost plugin computes a
1011WCET and Alt-Ergo is able to discharge all VCs automatically.
1013\paragraph{Handling C programs with simple loops.}
1014The cost annotations added by the CerCo compiler take the form of C instructions
1015that update a fresh global variable called the cost variable by a constant.
1016Synthesizing a WCET of a C function thus consists of statically resolving an
1017upper bound of the difference between the value of the cost variable before and
1018after the execution of the function, i.e. finding the instructions
1019that update the cost variable and establish the number of times they are passed
1020through during the flow of execution. To perform the analysis the plugin
1021makes the following assumptions on the programs:
10221) there are no recursive functions;
10232) every loop must be annotated with a variant. The variants of `for' loops are
1024automatically inferred.
1026The plugin proceeds as follows.
1027First the call graph of the program is computed.
1028Then the computation of the cost of the function is performed by traversing its
1029control flow graph. If the function $f$ calls the function $g$ 
1030then the function $g$ 
1031is processed before the function $f$. The cost at a node is the maximum of the
1032costs of the successors.
1033In the case of a loop with a body that has a constant cost for every step of the
1034loop, the cost is the product of the cost of the body and of the variant taken
1035at the start of the loop.
1036In the case of a loop with a body whose cost depends on the values of some free
1037variables, a fresh logic function $f$ is introduced to represent the cost of the
1038loop in the logic assertions. This logic function takes the variant as a first
1039parameter. The other parameters of $f$ are the free variables of the body of the
1040loop. An axiom is added to account the fact that the cost is accumulated at each
1041step of the loop.
1042The cost of the function is directly added as post-condition of the function.
1044The user can influence the annotation by two different means:
10451) by using more precise variants;
10462) by annotating functions with cost specifications. The plugin will use this cost
1047for the function instead of computing it.
1048\paragraph{C programs with pointers.}
1049When it comes to verifying programs involving pointer-based data structures,
1050such as linked lists, trees, or graphs, the use of traditional first-order logic
1051to specify, and of SMT solvers to verify, shows some limitations. Separation
1052logic~\cite{separation} is an elegant alternative. It is a program logic
1053with a new notion of conjunction to express spatial heap separation. Bobot has
1054recently introduced automatically generated separation
1055predicates to simulate separation logic reasoning in the Jessie plugin where the specification language, the verification condition
1056generator, and the theorem provers were not designed with separation logic in
1057mind. CerCo's plugin can exploit the separation predicates to automatically
1058reason on the cost of execution of simple heap manipulation programs such as an
1059in-place list reversal.
1061\section{Conclusions and future work}
1063All the CerCo software and deliverables can be found on the CerCo homepage at~\url{}.
1065The results obtained so far are encouraging and provide evidence that
1066it is possible to perform static time and space analysis at the source level
1067without losing accuracy, reducing the trusted code base and reconciling the
1068study of functional and non-functional properties of programs. The
1069techniques introduced seem to be scalable, cover both imperative and
1070functional languages and are compatible with every compiler optimisation
1071considered by us so far.
1073To prove that compilers can keep track of optimisations
1074and induce a precise cost model on the source code, we targeted a simple
1075architecture that admits a cost model that is execution history independent.
1076The most important future work is dealing with hardware architectures
1077characterized by history dependent stateful components, like caches and
1078pipelines. The main issue is to assign a parametric, dependent cost
1079to basic blocks that can be later transferred by the labeling approach to
1080the source code and represented in a meaningful way to the user. The dependent
1081labeling approach that we have studied seems a promising tool to achieve
1082this goal, but more work is required to provide good source level
1083approximations of the relevant processor state.
1085Other examples of future work are to improve the cost invariant
1086generator algorithms and the coverage of compiler optimizations, to combining
1087the labeling approach with the type and effect discipline of~\cite{typeffects}
1088to handle languages with implicit memory management, and to experiment with
1089our tools in the early phases of development. Larger case studies are also necessary
1090to evaluate the CerCo's prototype on realistic, industrial-scale programs.
1092% \bibliographystyle{splncs}
1094% \begin{thebibliography}{19}
1096% \bibitem{survey} \textbf{A Survey of Static Program Analysis Techniques}
1097% W.~W\"ogerer. Technical report. Technische Universit\"at Wien 2005
1099% \bibitem{cerco} \textbf{Certified Complexity}. R.M. Amadio, A. Asperti, N. Ayache,
1100% B. Campbell, D. P. Mulligan, R. Pollack, Y. Regis-Gianas, C. Sacerdoti Coen, I.
1101% Stark, in Procedia Computer Science, Volume 7, 2011, Proceedings of the 2 nd
1102% European Future Technologies Conference and Exhibition 2011 (FET 11), 175-177.
1104% \bibitem{labeling} \textbf{Certifying and Reasoning on Cost Annotations in C
1105% Programs}, N.  Ayache, R.M. Amadio, Y.R\'{e}gis-Gianas, in Proc. FMICS, Springer
1106% LNCS
1107% 7437: 32-46, 2012.
1108% %, DOI:10.1007/978-3-642-32469-7\_3.
1110% \bibitem{labeling2} \textbf{Certifying and reasoning on cost annotations of
1111% functional programs}.
1112% R.M. Amadio, Y. R\'{e}gis-Gianas. Proceedings of the Second international conference
1113% on Foundational and Practical Aspects of Resource Analysis FOPARA 2011 Springer
1114% LNCS 7177:72-89, 2012.
1116% \bibitem{compcert} \textbf{Formal verification of a realistic compiler}. X. Leroy,  In Commun. ACM 52(7), 107–115, 2009.
1118% \bibitem{framac} \textbf{Frama-C user manual}. L. Correnson, P. Cuoq, F. Kirchner, V. Prevosto, A. Puccetti, J. Signoles,
1119% B. Yakobowski. in CEA-LIST, Software Safety Laboratory, Saclay, F-91191,
1120% \url{}.
1122% \bibitem{paolo} \textbf{Indexed Labels for Loop Iteration Dependent Costs}. P.
1123% Tranquilli, in Proceedings of the 11th International Workshop on Quantitative
1124% Aspects of Programming Languages and Systems (QAPL 2013), Rome, 23rd-24th March
1125% 2013, Electronic Proceedings in Theoretical Computer Science, to appear in 2013.
1127% \bibitem{separation} \textbf{Intuitionistic reasoning about shared mutable data
1128% structure} J.C. Reynolds. In Millennial Perspectives in Computer Science,
1129% pages 303–321, Houndsmill, Hampshire, 2000. Palgrave.
1131% \bibitem{lustre} \textbf{LUSTRE: a declarative language for real-time
1132% programming}
1133% P. Caspi, D. Pilaud, N. Halbwachs, J.A. Plaice. In Proceedings of
1134% the 14th ACM SIGACT-SIGPLAN symposium on Principles of programming languages ACM
1135% 1987.
1137% \bibitem{correctness} \textbf{On the correctness of an optimising assembler for
1138% the intel MCS-51 microprocessor}.
1139%   D. P. Mulligan, C. Sacerdoti Coen. In Proceedings of the Second
1140% international conference on Certified Programs and Proofs, Springer-Verlag 2012.
1142% \bibitem{proartis} \textbf{PROARTIS: Probabilistically Analysable Real-Time
1143% Systems}, F.J. Cazorla, E. Qui\~{n}ones, T. Vardanega, L. Cucu, B. Triquet, G.
1144% Bernat, E. Berger, J. Abella, F. Wartel, M. Houston, et al., in ACM Transactions
1145% on Embedded Computing Systems, 2012.
1147% \bibitem{embounded} \textbf{The EmBounded project (project paper)}. K. Hammond,
1148% R. Dyckhoff, C. Ferdinand, R. Heckmann, M. Hofmann, H. Loidl, G. Michaelson, J.
1149% Serot, A. Wallace, in Trends in Functional Programming, Volume 6, Intellect
1150% Press, 2006.
1152% \bibitem{matita}
1153% \textbf{The Matita Interactive Theorem Prover}.
1154% A. Asperti, C. Sacerdoti Coen, W. Ricciotti, E. Tassi.
1155% 23rd International Conference on Automated Deduction, CADE 2011.
1157% \bibitem{typeffects} \textbf{The Type and Effect Discipline}. J.-P. Talpin,
1158%  P. Jouvelot.
1159%   In Proceedings of the Seventh Annual Symposium on Logic in Computer Science
1160% (LICS '92), Santa Cruz, California, USA, June 22-25, 1992.
1161% IEEE Computer Society 1992.
1163% \bibitem{stateart} \textbf{The worst-case execution-time problem overview of
1164% methods
1165% and survey of tools.} R. Wilhelm et al., in  ACM Transactions on Embedded
1166% Computing Systems, 7:1–53, May 2008.
1168% %\bibitem{proartis2} \textbf{A Cache Design for Probabilistic Real-Time
1169% % Systems}, L. Kosmidis, J. Abella, E. Quinones, and F. Cazorla, in Design,
1170% % Automation, and Test in Europe (DATE), Grenoble, France, 03/2013.
1172% \end{thebibliography}
Note: See TracBrowser for help on using the repository browser.