[cfe-dev] Finding function declaration?

Anna Zaks ganna at apple.com
Sat May 16 15:26:25 PDT 2015



Sent from my iPhone

> On May 16, 2015, at 8:15 AM, Manuel Klimek <klimek at google.com> wrote:
> 
>> On Thu, May 14, 2015 at 3:46 PM Ben Laurie <benl at google.com> wrote:
>>> On 14 May 2015 at 14:43, Manuel Klimek <klimek at google.com> wrote:
>>> Ah, wrong parent map :) the ast context has a function to retrieve parents that supports crossing into decls. I can look up the details when I'm back at a real computer ...
>>> 
>> 
>> Ah! Thanks.
> 
> You can get the ASTContext from the CheckerContext, and then call getParents() on it.
> 
CheckerContext is the way to go.
> Also cc'ing Anna and Jordan, who might have better ideas how to capture context info from a static analyzer check.
>  
>>  
>>> BTW do you need the symbolic execution engine for what you want to check? If the ast is enough consider using a clang-tidy check, those are a bit simpler to write with the ast matchers.
>>> 
>> 
>> I do need symbolic execution.
>>  
>>> 
>>>> On Thu, May 14, 2015, 1:20 PM Ben Laurie <benl at google.com> wrote:
>>>> Manuel asked me for more info.
>>>> 
>>>> Given a checker like this:
>>>> 
>>>> class MyChecker : 
>>>>     public Checker< check::PreStmt<ReturnStmt> > {
>>>> public:
>>>>   void checkPreStmt(const ReturnStmt *RS, CheckerContext &C) const;
>>>> };
>>>> 
>>>> void MyChecker::checkPreStmt(const ReturnStmt *RS,
>>>> 				     CheckerContext &C) const {
>>>>   RS->dump();
>>>> 
>>>>   const Stmt *s = RS;
>>>> 
>>>>   ParentMap &PM = C.getLocationContext()->getParentMap();
>>>> 
>>>>   while ((s = PM.getParent(s)) != NULL) {
>>>>     std::cerr << "------\n";
>>>>     s->dump();
>>>>   }
>>>> }
>>>> 
>>>> And this as input:
>>>> 
>>>> int f(void)
>>>>     {
>>>>     return 1;
>>>>     }
>>>> 
>>>> 
>>>> This is the output:
>>>> 
>>>> ReturnStmt 0x807cbdda8
>>>> `-IntegerLiteral 0x807cbdd88 'int' 1
>>>> ------
>>>> CompoundStmt 0x807cbddc0
>>>> `-ReturnStmt 0x807cbdda8
>>>>   `-IntegerLiteral 0x807cbdd88 'int' 1
>>>> 
>>>> As you can see, it terminates at the function body, not the declaration.
>>>> 
>>>> 
>>>>> On 14 May 2015 at 07:03, Ben Laurie <benl at google.com> wrote:
>>>>> 
>>>>> 
>>>>>> On 11 May 2015 at 11:20, Manuel Klimek <klimek at google.com> wrote:
>>>>>>> On Sun, May 10, 2015 at 11:54 PM Nikola Smiljanic <popizdeh at gmail.com> wrote:
>>>>>>> I think you'll need to use ParentMap from ASTContext.
>>>>>> 
>>>>>> Note that the parent-map is lazy-built (that is, the first time you get a parent, you build the parent-map for the whole TU), and thus has a pretty high cost. Usually we first try to get away with not using it.
>>>>>> 
>>>>>> Can you give a bit more context on what you're trying to do?
>>>>> 
>>>>> Sorry for delay ... I am trying to label functions to have certain checks done on their return values if an attribute is set on the function.
>>>>> 
>>>>> ParentMap appears not to work for this case (it stops finding parents once you have the function body).
>>>>> 
>>>>> Currently, the only way I could make this work was to construct a map of function bodies that have the attribute set in an AST observer, and then using ParentMap to get back to the body, which is looked up in the map.
>>>>> 
>>>>> Doesn't seem optimal!
>>>>> 
>>>>>>  
>>>>>>> 
>>>>>>>> On Mon, May 11, 2015 at 4:18 AM, Ben Laurie <benl at google.com> wrote:
>>>>>>>> If I have a checker for check::PreStmt<ReturnStmt>, how do I get from
>>>>>>>> the ReturnStmt to the function it is part of (I want to check
>>>>>>>> attributes on that function)?
>>>>>>>> _______________________________________________
>>>>>>>> cfe-dev mailing list
>>>>>>>> cfe-dev at cs.uiuc.edu
>>>>>>>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
>>>>>>> 
>>>>>>> _______________________________________________
>>>>>>> cfe-dev mailing list
>>>>>>> cfe-dev at cs.uiuc.edu
>>>>>>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20150516/b4a4ee9b/attachment.html>


More information about the cfe-dev mailing list