[cfe-commits] [PATCH] Expressions have lvalues and rvalues

[Ted Kremenek]

Hi Zhongxing,

This is an interesting patch.  I'm going to scrutinize it so that I  
understand its full ramifications (both in terms of design and  
functionality), so I might not get back to you immediately.

Ted

On Oct 9, 2008, at 6:45 AM, Zhongxing Xu wrote:

> This patch introduces VisitLValue() to evaluate the lvalue of an  
> expression. VisitLValue()
> is different from VisitLVal(). The motivation of this patch is as  
> follows.
>
> We have two kinds of RVal: LVal and NonLVal. LVal represents  
> location value,
> NonLVal represents non-location value.
>
> Expressions have two kinds of value: rvalue and lvalue. The rvalue  
> of an
> expression is is the normal value it evaluates to when placed at the  
> RHS of an
> assignment. The lvalue of an expression is the address of the memory  
> object it
> represents.
>
> We should distinguish rvalue/lvalue from NonLVal/LVal. An  
> expression's lvalue
> must be a LVal. An expression's rvalue can be either NonLVal or  
> LVal. Every
> expressoin has its rvalue. But not all of the expressions have  
> lvalues. For
> example:
>
> int x = 3;
>
> "3" has rvalue 3, but no lvalue
> "x" has rvalue 3, lvalue is its associated memregion MemRegionVal
> "&x" has no lvalue, but has rvalue, which is the lvalue of 'x', i.e.  
> MemRegionVal of x
>
> So I suggest when we evaluate expressions, we pass on an argument  
> indicating
> whether we want to get its lvalue/rvalue, instead of specifying we  
> are getting a
> LVal or NonLVal.
>
> In this patch, I use LValue/RValue to distinguish from LVal/NonLVal.
>
> This patch cleans up a bunch of code. It lets store manager to  
> evaluate the lvalue
> of an expression.
> <lvalue.patch>_______________________________________________
> cfe-commits mailing list
> cfe-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits