Data Buffer

Hello,
I am writing an outgoing data buffer for a program that will be sending it across a socket to a server. When I form this packet, the first 4 bytes are important identifiers to the server. I am assigning the first and second byte ok, but the 3rd and 4th is where I am having a problem.
The 3rd and 4th contain the size of the packet in bytes. Obviously, I am storing this as an unsigned short int. When I go to write to this portion, it has errors when the value is higher than 255.

Here is the basic snippet. I am getting an assertion error on the line where it assigns the unsigned short int. Could anyway take a crack at this?

Code:

---

char buffer[1024] = {0};
unsigned short int bufferlen = 4;

void sendPacket(unsigned char id)
{
        int i, j, k;
        int line = 0;

        *(unsigned char *)(buffer) = 0xFF;
        *(unsigned char *)(buffer + 1) = id;
        *(unsigned short int *)(buffer + 2) = bufferlen;
 
        ...
}

---

Edit: I just noticed this. To check to see if my packet structure is correct, I print it to the console. If I choose not to print it to the console, there is no error. Also, I checked the content of the 3rd and 4th bytes and they are correct. Why would this error when I print the characters out? If you want to see the way I output it, just ask, but it's kind of long.

Edit: Wow, just never mind. I've been recompiling this for 20 mins and this last compile decided to work. I am sorry. Does anyone else get this with visual studio 2008? I changed no code, but just recompiled about 20 times.

Edit: And here we go again, now its erroring. Could this be a compiler issue?

Exactly what error do you get? To me, there's nothing wrong with that line [from looking at it].

--
Mats

Heres a screenshot
http://i278.photobucket.com/albums/k...1029/error.jpg

And line 56 is here

Code:

---

0052    void insertINT32(unsigned long data)
0053    {
0054            *(unsigned long *)(buffer + bufferlen) = data;
0055            bufferlen += 4;
0056    }

---

Now that I see that, that looks like a library error, here is where i use a function from ctype.h

Code:

---

                for (j = 0; j < 16; j++)
                {
                        if (isgraph(buffer[i+j]))
                                printf("&#37;c", buffer[i+j]);
                        else
                                printf(".");
                }

---

It's not a library error. You are passing a value that is greater than 255. A char, a character, can only assume numbers through 0 to 255, but you are passing a greater value, so the error is yours.
What type is buffer? Long? Int?

It should work. buffer is char[]

Bahh, I just changed the line:

Code:

---

if (isgraph(*(unsigned char *)(buffer + i + j)))

---

Thanks

Then the error may lie elsewhere?
Try this: pick retry on the assert, then use the call stack (it's in the bottom right window) [it's a tab], and walk backwards in the list until you find your function which called the library function. Then you should be able to find the line which caused the error.
Check data types, check variable values.

You can then set a breakpoint in the function and step using the debugger with F11 and F10. Keep a look out for any invalid value range or such.

You can post the code if you have troubles.

Code:

---

char buffer[1024] = {0};
...
*(unsigned short int *)(buffer + 2) = bufferlen;

---

What happens if (buffer+2) is not aligned on 16bits? Perhaps that's why your program is sometimes working, this is not related to the recompilation (which should always produce the same code) but to the location in memory @runtime which may vary.

Quote:

---

Originally Posted by root4

Code:

---

char buffer[1024] = {0};
...
*(unsigned short int *)(buffer + 2) = bufferlen;

---

What happens if (buffer+2) is not aligned on 16bits? Perhaps that's why your program is sometimes working, this is not related to the recompilation (which should always produce the same code) but to the location in memory @runtime which may vary.

---

Well, it's working now, but what you pointed out hasn't been my error. I do understand what you are saying though. If a situation like you described happened in memory, how could I prepare my code for it?

Quote:

---

Originally Posted by carrotcake1029

Well, it's working now, but what you pointed out hasn't been my error. I do understand what you are saying though. If a situation like you described happened in memory, how could I prepare my code for it?

---

You could simply VERIFY that your buffer is evenly aligned - since it's a global, it likely is. Something like this:

Code:

---

unsigned int a;
a = (unsigned int) buffer;
if (a & 3 != 0) printf("Not aligned to a longword address");

---

Since you are writing a longword in your function at line 56, this is the minimum required alignment.

Various compilers (e.g. gcc) allow you to specify the alignment of a variable or type, so you could use that to ensure the alignment if you are using a compiler that can do it.

--
Mats

Are you using htons() & ntohs() somewhere in your code? If not, you should be.

Better make that into a macro or a function ;)
That looks nasty and not at all obvious that it's checking for alignment.

Ok guys, thanks for your help. Glad to learn any flaws with my code.

Quote:

---

Originally Posted by Elysia

Better make that into a macro or a function ;)
That looks nasty and not at all obvious that it's checking for alignment.

---

Well, it is obvious if you know about alignment checks [it is very often used in drivers, so I've seen it many times] - but I didn't actually mean to imply that it should be done inline - a function or macro is an excellent idea.

--
Mats

Quote:

---

...but what you pointed out hasn't been my error.

---

Indeed, it does come from the isgraph() call:
"When used with a debug CRT library, isgraph will display a CRT assert if passed a parameter that is not EOF or in the range of 0 through 0xFF"[from MSDN]

... so it's using an "int" as input type but only works with bytes, how clever is that. I guess this is using an int only for the special case with EOF(=-1?).

Quote:

---

how could I prepare my code for it?

---

As far as I know you used the only case which could raise such a problem (accessing an array of bytes by using another (longer) type - or something equivalent) . The compiler takes care of aligning everything that need to be otherwise.

I decided to see what msdn said about alignment, and here is what page I came to:
http://msdn.microsoft.com/en-us/library/83ythb65.aspx

Is this syntax correct and does it do what I am hoping to achieve?

Code:

---

__declspec(align(1024)) char buffer[1024] = {0};

---

Doing this, my code still compiles fine and runs ok.