Make Frama-c show dependencies even of "dead branches" - frama-c

I am using frama-c Aluminium-20160502 version and I want to find out the dependencies in a large program. When using the option -deps in the command line I found some dependencies are missing. In particular when several conditions are joined in one if, the dependency analysis stops whenever one condition is false. In this example here:
#include<stdio.h>
#include<stdbool.h>
/*Global variable definitions*/
bool A = true;
bool B = false;
bool C = true;
bool X;
bool Y;
bool res;
void main(){
if (A && B && C) {
res = X;
}else res = Y;
}
when I try: frama-c -deps program.c
frama shows the following dependencies:
[from] ====== DEPENDENCIES COMPUTED ======
These dependencies hold at termination for the executions that terminate:
[from] Function main:
res FROM A; B; Y
[from] ====== END OF DEPENDENCIES ======
so it does not reach the condition C because already B is false.
I wonder if there is a way to tell frama to compute all dependencies even if the condition is not fulfilled. I tried with the option -slevel but with no results. I know there is a way to use an interval Frama_C_interval(0,1) but when I use it the variable using this function is not shown in the dependencies. I would like to get X and Y dependent on A,B,C and res dependent on A,B,C,X,Y
Any ideas?

The From plugin uses the results of the Value analysis plugin. In your example, the values of A and B are sufficiently precise that Value is able to infer that the condition is entirely determined (since the && operator is lazily evaluated, from left to right) before reaching C, therefore C never affects the outcome and is thus not a dependency from the point of view of From.
Unfortunately, Frama_C_interval cannot be used directly at global initializers:
user error: Call to Frama_C_interval in constant.
You can, however, use a "hack" (not always the best solution, but works here):
volatile bool nondet;
bool A = nondet;
bool B = nondet;
bool C = nondet;
...
Note that because nondet is volatile, each variable A, B and C is assigned a different non-deterministic value.
In this case, Value has to consider both branches of the conditionals, and therefore C becomes a dependency in your example, since it is possible that C will be read during the execution of main. You'll then have:
These dependencies hold at termination for the executions that terminate:
[from] Function main:
res FROM A; B; C; X; Y
Note that some plugins require special treatment when dealing with volatile variables, so this is not always the best solution.
This however only deals with some kinds of dependencies. As mentioned in Chapter 6 of the Value Analysis user manual, the From plugin computes functional, imperative and operational dependencies. These do not include indirect control dependencies, which are those such as X from A, B, C, as you mention. For those, you need the PDG (Program Dependence Graph) plugin, but it does not currently have a textual output of the dependencies. You can use -pdg to compute it, and then -pdg-dot <file> to export the dependency graph in dot (graphviz) format. Here's what I get for your main function (using the volatile hack as mentioned previously):
Finally, as a side note: -slevel is mostly used to improve precision, but in your example you already have too much precision (that is, Value is already able to infer that C is never read inside main).

Related

Assigning Rcpp::XPtr on the R side

I'm learning about external pointers, XPtr, in Rcpp. I made the following test functions:
// [[Rcpp::export]]
Rcpp::XPtr<arma::Mat<int>> create_xptr(int i, int j) {
arma::Mat<int>* ptr(new arma::Mat<int>(i, j));
Rcpp::XPtr<arma::Mat<int>> p(ptr, true);
return p;
}
// [[Rcpp::export]]
void fill_xptr(Rcpp::XPtr<arma::Mat<int>>& xptr, int k) {
(*xptr).fill(k);
}
// [[Rcpp::export]]
arma::Mat<int> return_val(Rcpp::XPtr<arma::Mat<int>>& xptr) {
return *xptr;
}
Now on the R side I can of-course create an instance and work with it:
x <- create_xptr(1000, 1000) # (1)
Say, for some reason I accidentally called create_xptr again and assigned the result to x, i.e
x <- create_xptr(1000, 1000) # (2)
Then, I have no longer access to the pointer i created in (1) which makes sense and hence, I cannot free the memory. What I would like is, that the second time (2) it just overwrite the first one (1). And secondly, if I create an external pointer in some local scope (say a simple for loop), should the memory used then be freed automatically when it goes out of scope? I've tried the following:
for (i in 1:3) {
a <- create_xptr(1000, 100000)
fill_xptr(a, 1)
}
But it just adds to the memory usage for each i.
I have tried reading some code on different git-repos, old posts here on stack read a little about finalizers and garbage collection in R. I can't seem to put together the puzzle.
We use external pointers for things that do not have already existing interfaces such as database connections or objects from other new libraries. You may be better off looking at some existing uses of XPtr (or the other external pointer variants in some other packages, there are two small ones on CRAN).
And I don't think I can think of an example directly referencing this in R. It is mostly for "wrapping" external objects to hold on to them and to pass them around for further use elsewhere. And you are correct that you need to read up a little on finalizers. I find reading Writing R Extensions, as dense as it is, to be the best source because you need to get the initial "basics" in C right first.

Ada: Re-exporting enum type values

I have an ada package in the form of
package C is
type Test is (Test_1, Test_2, Test_3);
end C;
and another package in the form of
with C;
package B is
subtype Test is C.Test;
end B;
and yet another in the form of
with B;
package A is
subtype Test is B.Test;
Item: Test := Test_1;
end A;
Naively, I would expect that a subtype defined in B, which is later subtyped by A, would be able to access members of the original type. However, upon inspection, the members of C are not even visible in the scope of B. This can be fixed by adding use c;, this seems like a find solution to a degree, however to use it in A, you would have to add with c; use c; to every package that transitively depends on C. This could lead to confusion as it wouldn't be implicitly obvious that you should be using C.
I would like to be able to "rexport" these types, so that I can abstract layers of my program better.
If you change package A to
with B;
package A is
subtype Test is B.Test;
use all type Test;
Item: Test := Test_1;
end A;
the code compiles. Not sure whether this will help.
This is an Ada 2012 feature; see ARM 8.4(8) and (8.1). use all type makes the primitive operations of the type visible (which includes enumeration literals).
If you really want to re-export the values, you can do this:
with C;
package B is
subtype Test is C.Test;
function Test_1 return Test is (C.Test_1);
function Test_2 return Test is (C.Test_2);
function Test_3 return Test is (C.Test_3);
end B;
Unfortunately, you can't use named numbers since enumerations are not numeric types. You can make these functions normal constants instead, but conceptually, this would execute code at elaboration time (the compiler will probably optimize it away, but you can't use pragma Preelaborate anymore).
This allows you to access the literals in A using B.Test_1 etc. This is also a proper abstraction as A will not depend on the literals defined in C anymore (you can rename the literals in C without affecting A, you do need to update B though to map the new names to the original ones).
Simply importing the literal names into the namespace of A is not an abstraction.
with B;
package A is
type Test is new B.Test;
Item: Test := Test_1;
end A;
...and all is well!
Ada has the concept of visibility, with a lot of rules that define when, where, and how things are visible. Understanding visibility is key to understanding Ada. One of the best discussions of visibility is in Ada Distilled. What you're encountering is a consequence of the visibility rules. This goes even further than enumeration literals:
package Pack is
type Int is range -(2 ** 15) + 1 .. 2 ** 15 - 1;
end Pack;
with Pack;
procedure Proc is
subtype Int is Pack.Int;
-- None of the operators of Int ("+", "*", ...) are directly visible here
begin
null;
end Proc;
Declarations in a package spec are only visible outside the package by using dotted notation, unless you explicitly make them directly visible elsewhere. use clauses are one way to make them visible; others are renames and, for enumeration literals, declaring constants of the type initialized using dotted notation.

Frama-c : Trouble understanding WP memory models

I'm looking for WP options/model that could allow me to prove basic C memory manipulations like :
memcpy : I've tried to prove this simple code :
struct header_src{
char t1;
char t2;
char t3;
char t4;
};
struct header_dest{
short t1;
short t2;
};
/*# requires 0<=n<=UINT_MAX;
# requires \valid(dest);
# requires \valid_read(src);
# assigns (dest)[0..n-1] \from (src)[0..n-1];
# assigns \result \from dest;
# ensures dest[0..n] == src[0..n];
# ensures \result == dest;
*/
void* Frama_C_memcpy(char *dest, const char *src, uint32_t n);
int main(void)
{
struct header_src p_header_src;
struct header_dest p_header_dest;
p_header_src.t1 = 'e';
p_header_src.t2 = 'b';
p_header_src.t3 = 'c';
p_header_src.t4 = 'd';
p_header_dest.t1 = 0x0000;
p_header_dest.t2 = 0x0000;
//# assert \valid(&p_header_dest);
Frama_C_memcpy((char*)&p_header_dest, (char*)&p_header_src, sizeof(struct header_src));
//# assert p_header_dest.t1 == 0x6265;
//# assert p_header_dest.t2 == 0x6463;
}
but the two last assert weren't verified by WP (with default prover Alt-Ergo). It can be proved thanks to Value analysis, but I mostly want to be able to prove the code not using abstract interpretation.
Cast pointer to int : Since I'm programming embedded code, I want to be able to specify something like:
#define MEMORY_ADDR 0x08000000
#define SOME_SIZE 10
struct some_struct {
uint8_t field1[SOME_SIZE];
uint32_t field2[SOME_SIZE];
}
// [...]
// some function body {
struct some_struct *p = (some_struct*)MEMORY_ADDR;
if(p == NULL) {
// Handle error
} else {
// Do something
}
// } body end
I've looked a little bit at WP's documentation and it seems that the version of frama-c that I use (Magnesium-20151002) has several memory model (Hoare, Typed , +cast, +ref, ...) but none of the given example were proved with any of the model above. It is explicitly said in the documentation that Typed model does not handle pointer-to-int casts. I've a lot of trouble to understand what's really going on under the hood with each wp-model. It would really help me if I was able to verify at least post-conditions of the memcpy function. Plus, I have seen this issue about void pointer that apparently are not very well handled by WP at least in the Magnesium version. I didn't tried another version of frama-c yet, but I think that newer version handle void pointer in a better way.
Thank you very much in advance for your suggestions !
memcpy
Reasoning about the result of memcpy (or Frama_C_memcpy) is out of range of the current WP plugin. The only memory model that would work in your case is Bytes (page 13 of the manual for Chlorine), but it is not implemented.
Independently, please note that your postcondition from Frama_C_memcpy is not what you want here. You are asserting the equality of the sets dest[0..n] and src[0..n]. First, you should stop at n-1. Second, and more importantly, this is far too weak, and is in fact not sufficient to prove the two assertions in the caller. What you want is a quantification on all bytes. See e.g. the predicate memcmp in Frama-C's stdlib, or the variant \forall int i; 0 <= i < n -> dest[i] == src[i];
By the way, this postcondition holds only if dest and src are properly separated, which your function does not require. Otherwise, you should write dest[i] == \at (src[i], Pre). And your requires are also too weak for another reason, as you only require the first character to be valid, not the n first ones.
Cast pointer to int
Basically, all current models implemented in WP are unable to reason on codes in which the memory is accessed with multiple incompatible types (through the use of unions or pointer casts). In some cases, such as Typed, the cast is detected syntactically, and a warning is issued to warn that the code cannot be analyzed. The Typed+Cast model is a variant of Typed in which some casts are accepted. The analysis is correct only if the pointer is re-cast into its original type before being used. The idea is to allow using some libc functions that require a cast to void*, but not much more.
Your example is again a bit different, because it is likely that MEMORY_ADDR is always addressed with type some_stuct. Would it be acceptable to change the code slightly, and change your function as taking a pointer to this type? This way, you would "hide" the cast to MEMORY_ADDR inside a function that would remain unproven.
I tried this example in the latest version of Frama-C (of course the format is modified a little bit).
for the memcpy case
Assertion 2 fails but assertion 3 is successfully proved (basically because the failure of assertion 2 leads to a False assumption, which proves everything).
So in fact both assertion cannot be proved, same as your problem.
This conclusion is sound because the memory models used in the wp plugin (as far as I know) has no assumption on the relation between fields in a struct, i.e. in header_src the first two fields are 8 bit chars, but they may not be nestedly organized in the physical memory like char[2]. Instead, there may be paddings between them (refer to wiki for detailed description). So when you try to copy bits in such a struct to another struct, Frama-C becomes completely confused and has no idea what you are doing.
As far as I am concerned, Frama-C does not support any approach to precisely control the memory layout, e.g. gcc's PACKED which forces the compiler to remove paddings.
I am facing the same problem, and the (not elegant at all) solution is, use arrays instead. Arrays are always nested, so if you try to copy a char[4] to a short[2], I think the assertion can be proved.
for the Cast pointer to int case
With memory model Typed+cast, the current version I am using (Chlorine-20180501) supports casting between pointers and uint64_t. You may want to try this version.
Moreover, it is strongly suggested to call Z3 and CVC4 through why3, whose performance is certainly better than Alt-Ergo.

Frama-C generates confusing assertions about pointer comparison

I'm getting assertions that don't make sense to me.
Code:
struct s {
int pred;
};
/*# assigns \result \from \nothing;
ensures \valid(\result);
*/
struct s *get(void);
int main(void)
{
return !get()->pred;
}
Frama-C output:
$ frama-c -val frama-ptr.c
[...]
frama-ptr.c:12:[kernel] warning: pointer comparison.
assert \pointer_comparable((void *)0, (void *)tmp->pred);
(tmp from get())
Am I doing something wrong or is it a bug in Frama-C?
This is not really a bug, although this behavior is not really satisfactory either. I recommend that you observe the analysis in the GUI (frama-c-gui -val frama-ptr.c), in particular the value of tmp->pred at line 12 in the Values tab.
Value before:
∈
{{ garbled mix of &{alloced_return_get}
(origin: Library function {c/ptrcmp.c:12}) }}
This is a very imprecise value, that has been generated because the return value of function get is a pointer. The Eva plugin generates a dummy variable (alloced_return_get), and fills it with dummy contents (garbled mix of &{alloced_return_get}). Even though the field pred is an integer, the analyzer assumes that (imprecise) pointers can also be present. This is why an alarm for pointer comparison is emitted.
There are at least two ways to avoid this:
use option -val-warn-undefined-pointer-comparison pointer, so that pointer comparison alarms are emitted only on comparisons involving lvalues with pointer type. The contents of the field pred will remain imprecise, though.
add a proper body to function get, possibly written using malloc, so that the field pred receives a more accurate value.
On a more general note, the Eva/Value plugin requires precise specifications for functions without a body; see section 7.2 of the manual. For functions that return pointers, you won't be able to write satisfactory specifications: you need to write allocates clauses, and those are currently not handled by Eva/Value. Thus, I suggest that you write a body for get.

Frama-C: No splitting of if-statement

I have the following problem when analyzing if-conditions with my plugin. When I analyze code like
if ((a && b) || c)
Frama-C creates code like this:
if (a) {
if (b){
goto _LOR;
}
else{
goto _LAND;
}
}
else{
_LAND: ;
if (c) {
_LOR: //....
For my analysis, I want to get the control-flow without the conditions split up, so that one statement remains. I want this, because by splitting up the conditional, the reaching of _LOR is dependent on the or-part only.
Example: creating the CFG as in the viewcfg plugin of the developer guide leads to this:
a=(a+b)+c can be reached in the then and the else-path of if a, which I cancel out in my plugin in the block-statement (so that after a "simple" if, the statement is not dependent on the condition anymore).
Is there a possibility to block the splitting of the if, by a command-line argument or something like that?
An undocumented and unsupported solution exists. Before compiling Frama-C, in function initCIL of cil.ml, change
theMachine.useLogicalOperators <- false (* do not use lazy LAND and LOR *);
into
theMachine.useLogicalOperators <- true;
The normalization will use logical || and && operators instead of gotos.
Please note that this is unsupported for a good reason. The Frama-C plugins packaged with the kernel expect an AST in which those operators are not used, so they will probably crash or do something unsound on your program. Use at your own risk!

Resources