variable size array declarations

variable size array declarations

[Dan Carpenter]

[-- Attachment #1: Type: text/plain, Size: 618 bytes --]

I am trying to check something in Smatch with variable size arrays
and I'm running into problems.  Here is the function I'm trying to
parse.

void func(int x)
{
	char buf[x];
}

Array symbols have an ->array_size.  It gets set correctly in parse.c
to expression of type EXPR_SYMBOL.  But by the time I parse it in
Smatch the ->array_size is now an EXPR_PREOP (a dereference).  In
fact all the variables seem to get change to dereferences of the
variables.

So if I have:
	char buf[a + b + *c + 4];
Smatch gets:
	char buf[*a + *b + **c + 4];

I'm not sure where this is getting changed or why.

regards,
dan carpenter


[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: variable size array declarations

[Dan Carpenter]

[-- Attachment #1: Type: text/plain, Size: 1161 bytes --]

On Mon, Oct 31, 2011 at 05:08:29PM +0300, Dan Carpenter wrote:
> I am trying to check something in Smatch with variable size arrays
> and I'm running into problems.  Here is the function I'm trying to
> parse.
> 
> void func(int x)
> {
> 	char buf[x];
> }
> 
> Array symbols have an ->array_size.  It gets set correctly in parse.c
> to expression of type EXPR_SYMBOL.  But by the time I parse it in
> Smatch the ->array_size is now an EXPR_PREOP (a dereference).  In
> fact all the variables seem to get change to dereferences of the
> variables.
>

I tracked this down.  In smatch, how I use sparse is that I call
sparse_keep_tokens() and then I parse the resulting symbol list
myself.

In examine_array_type() we call get_expression_value() which changes
the symbols from normal symbols to dereferences.  The call tree is:
examine_array_type()
-> get_expression_value()
   -> __get_expression_value()
      -> evaluate_expression()
         -> evaluate_symbol_expression() <- change happens here.

The problem for me is that not all expressions have been evualated
like this so it's not consistent.

regards,
dan carpenter


[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]