misleading warning [pointerSize]

[Yannick HOUDOT]

Hi,

I don't understand why the warning is about the size of pointer 'p' ?!?

Best regards.

=== test.c ===
void * func(size_t n)
{
char *p;

if (!(p = malloc( n * sizeof(void*) )))
{
}
return p;

}

void main() { free( func(16 ) ) ; }
=== test.c == end

$ /opt/cppcheck-2.10/cppcheck --enable=all test.c
Checking test.c ...
test.c:5:11: warning: Size of pointer 'p' used instead of size of its data. [pointerSize]
if (!(p = malloc( n +sizeof(void*) + sizeof(size_t))))
^
$

[Andrew C Aitchison]

Do you want to allocate space for n pointers or n chars ?
You probably want either
void *p;
or
sizeof(char)

[Yannick HOUDOT]

ok, so if I understand correctly, this test checks for a match between the pointer type and the types in the memory size calculation...

ok, but I stand by the message being misleading when it talks about the size of 'p' instead of a none match between types ...

Thanks for the hint !
Best.

[Daniel Marjamäki]

Yes.. hmm the message is confusing. Thanks.

I have created this ticket:
https://trac.cppcheck.net/ticket/11754

[Daniel Marjamäki]

I wonder if you think this is a false positive. I mean was do you think your own code was better or was it better to tweak the code.

If I understand it correctly you wanted to allocate this amount of bytes:

p = malloc( n + sizeof(void*) + sizeof(size_t))

Maybe we should implement some exception. If sizeof(void*) is used it's likely the user wanted to get "size of pointer".. there is no big danger there is an extra "*" at least.

[Yannick HOUDOT]

This code looks like an allocation for a "command bloc", it's like a packed structure.
there's no alignement involved, so it use a char* pointer to fill it up.
I used them all day long.

I corrected the code by using a temporary "void*" pointer instead, no more warning.

[Daniel Marjamäki]

I corrected the code by using a temporary "void*" pointer instead, no more warning.

I don't like that such correction is used to silence cppcheck. So I think this is a false positive.

[Yannick HOUDOT]

late reply, but yes, I do think myself that it's a false positive.
the main issue is really the warning sentence... not clear enough, or misleading.
Thanks for the ticket opening.