variable length array (was: question about dynamic memory allocation)

[Fernando Barsoba]

Hi,

I decided to start a new topic about my memory allocation question. One
of the answers was that variable length arrays are possible using C99. I
found this topic that also mentions it.

http://groups.google.com/group/comp...variable+length+array&rnum=2#f5335d22d6e6fb34

I did some test, but I get strange results:

int size_array = sizeof(struct ip) + cnf->msg_length;
unsigned char packet[size_array];

And then, when asking the size of the 'packet' variable I get 4. How is
that? It should be another number (31).. but I guess i'm getting the
size of type unsigned char. Right?


Thanks,

FBM

 

[Skarmander]

I decided to start a new topic about my memory allocation question. One
of the answers was that variable length arrays are possible using C99. I
found this topic that also mentions it.

http://groups.google.com/group/comp...variable+length+array&rnum=2#f5335d22d6e6fb34


I did some test, but I get strange results:

int size_array = sizeof(struct ip) + cnf->msg_length;
unsigned char packet[size_array];

And then, when asking the size of the 'packet' variable I get 4. How is
that? It should be another number (31).. but I guess i'm getting the
size of type unsigned char. Right?

Technically there's no way to tell, because we don't know what system
you're compiling this on.

That said, you're probably getting the size of an unsigned char*, to
which `packet' decays.

If sizeof packet != size_array after the declarations above, your
compiler does not conform to the C99 standard. gcc in particular does
not support VLAs; it supports an earlier gcc extension that is similar
but not identical to C99 VLAs.

S.

 

[Artie Gold]

Hi,

I decided to start a new topic about my memory allocation question. One
of the answers was that variable length arrays are possible using C99. I
found this topic that also mentions it.

http://groups.google.com/group/comp...variable+length+array&rnum=2#f5335d22d6e6fb34


I did some test, but I get strange results:

int size_array = sizeof(struct ip) + cnf->msg_length;
unsigned char packet[size_array];

And then, when asking the size of the 'packet' variable I get 4. How is
that? It should be another number (31).. but I guess i'm getting the
size of type unsigned char. Right?

Show us the code you used to test.
[You passed `packet' to a function, didn't you...]

HTH,
--ag

 

[Fernando Barsoba]

Hi,

I decided to start a new topic about my memory allocation question. One
of the answers was that variable length arrays are possible using C99. I
found this topic that also mentions it.

http://groups.google.com/group/comp...variable+length+array&rnum=2#f5335d22d6e6fb34


I did some test, but I get strange results:

int size_array = sizeof(struct ip) + cnf->msg_length;
unsigned char packet[size_array];

And then, when asking the size of the 'packet' variable I get 4. How is
that? It should be another number (31).. but I guess i'm getting the
size of type unsigned char. Right?

Show us the code you used to test.
[You passed `packet' to a function, didn't you...]

I pass 'packet' to a function later on.. but not for this test. I
checked the size of 'packet' after

>>unsigned char packet[size_array];

and there's where I get the value 4.

Regarding to the other question, I'm using Eclipse + CDT (and the
compiler is the one that comes with Cygwin which I am using too).

FBM

 

[Keith Thompson]

I did some test, but I get strange results:

int size_array = sizeof(struct ip) + cnf->msg_length;
unsigned char packet[size_array];

And then, when asking the size of the 'packet' variable I get 4. How is
that? It should be another number (31).. but I guess i'm getting the
size of type unsigned char. Right?

No, sizeof(unsigned char) is 1 by definition. It looks like it could
be sizeof(unsigned char*), but that's not

How exactly did you ask the size of the 'packet' variable? Show us
some code.

Here's an example that works properly with gcc 2.95.2, 3.4.4, and
4.0.2:

#include <stdio.h>
int main(void)
{
int len = 42;
char vla[len];
printf("sizeof vla = %d\n", (int)sizeof vla);
return 0;
}

The output is:

sizeof vla = 42

If you passed it as an argument to a function and applied sizeof to
the parameter, you'd get the size of the pointer, but that's because
array expressions decay to pointers in most contexts; if you apply
sizeof directly to the array object, it doesn't decay and you get the
size of the array.

 

[Skarmander]

I did some test, but I get strange results:

int size_array = sizeof(struct ip) + cnf->msg_length;
unsigned char packet[size_array];

And then, when asking the size of the 'packet' variable I get 4. How is
that? It should be another number (31).. but I guess i'm getting the
size of type unsigned char. Right?

No, sizeof(unsigned char) is 1 by definition. It looks like it could
be sizeof(unsigned char*), but that's not

How exactly did you ask the size of the 'packet' variable? Show us
some code.

Here's an example that works properly with gcc 2.95.2, 3.4.4, and
4.0.2:

<snip>
And there's me badmouthing gcc about not supporting VLAs yet. At least
this part works.

S.

 

[Barry Schwarz]

Hi,

I decided to start a new topic about my memory allocation question. One
of the answers was that variable length arrays are possible using C99. I
found this topic that also mentions it.

http://groups.google.com/group/comp...variable+length+array&rnum=2#f5335d22d6e6fb34

I did some test, but I get strange results:

int size_array = sizeof(struct ip) + cnf->msg_length;
unsigned char packet[size_array];

And then, when asking the size of the 'packet' variable I get 4. How is
that? It should be another number (31).. but I guess i'm getting the
size of type unsigned char. Right?

Not likely. An unsigned char is guaranteed to have a size of one.


<<Remove the del for email>>

 

[Keith Thompson]

Fernando Barsoba wrote: [snip]

I did some test, but I get strange results:

int size_array = sizeof(struct ip) + cnf->msg_length;
unsigned char packet[size_array];

And then, when asking the size of the 'packet' variable I get 4. How is
that? It should be another number (31).. but I guess i'm getting
the size of type unsigned char. Right?

Show us the code you used to test.
[You passed `packet' to a function, didn't you...]

I pass 'packet' to a function later on.. but not for this test. I
checked the size of 'packet' after

unsigned char packet[size_array];

and there's where I get the value 4.

That shouldn't be happening.

Show us the code you used to test. That means posting a complete
self-contained program that we can compile and run ourselves, along
with the exact output that it gave you. Without that, we really can't
help.

Regarding to the other question, I'm using Eclipse + CDT (and the
compiler is the one that comes with Cygwin which I am using too).

Cygwin comes with gcc (version 3.4.4 if you've updated it recently).

 

[Fernando Barsoba]

Fernando Barsoba wrote: [snip]


Show us the code you used to test.
[You passed `packet' to a function, didn't you...]

I pass 'packet' to a function later on.. but not for this test. I
checked the size of 'packet' after

unsigned char packet[size_array];

and there's where I get the value 4.

That shouldn't be happening.

Show us the code you used to test. That means posting a complete
self-contained program that we can compile and run ourselves, along
with the exact output that it gave you. Without that, we really can't
help.

Ok, I tested the following code:

#include <stdio.h>
#include <stdlib.h>
#include <netinet/ip.h>


int main(int argc, char **argv) {

int size_array = sizeof(struct ip) + 11;
unsigned char packet[size_array];
// unsigned char ipheader_cpy[sizeof(struct ip)];

printf("Result for sizeof(packet) is: %d", sizeof(packet));
exit(0);

}

And I got as a result:
$ ./Test.exe
Result for sizeof(packet) is: 31

So, I think I have to assume that it's working correctly. The problem is
that I'm using Eclipse/CDT and the debugger seems to have problems with
this kind of code (with VL Arrays). It showed me the '4' that I was
referring to, in the debugging perspective/window from Eclipse... I
thought the problem was with the code, but all you made me focus on the
test.. I guess I relied too much on the debugger...sigh..

Thanks,

FBM