[ bzflag-Bugs-520395 ] ID patching not right for MsgGrab/Drop

[<no...@so...>]

Bugs item #520395, was opened at 2002-02-20 01:37
You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=103248&aid=520395&group_id=3248

Category: BZFS
Group: None
Status: Open
Resolution: None
Priority: 5
Submitted By: Dave Brosius (dbrosius)
Assigned to: Tim Riker (timriker)
Summary: ID patching not right for MsgGrab/Drop

Initial Comment:
Would someone please do a sanity check here?

In 

static void patchMessage(PlayerId fromId, PlayerId 
toId, const void *msg)

in bzfs.cpp is this code


    case MsgDropFlag:
    case MsgGrabFlag:
      if (len > 28)
	// server version
	patchPlayerId(fromId, toId, msg, 4);
	patchPlayerId(fromId, toId, msg, 20);
      break;
      ;;

When these messages go from client to server, besides 
sending the PlayerID at pos 4, they send
MsgDropFlag -> position
MsgGrabFlag -> flagIndex

(in other words small amount of data)

When these messages go from server to client, besides 
sending the PlayerID at pos 4, they send

Both -> flagIndex, flagInfo

So when the buffer is big, there's a flagInfo there, 
and inside that is an owner with a playerID, that 
needs patching.

So, I think the code should be

    case MsgDropFlag:
    case MsgGrabFlag:
      if (len > 28) {
	// server version
        // patch owner in flagInfo
	patchPlayerId(fromId, toId, msg, 16);
      }
      patchPlayerId(fromId, toId, msg, 4);
      break;
      ;

>> Secondly, the offset I believe is 16

ByteOffset     Info
-------------------------
0              Msgcode/length
4              PlayerID
8              flagIndex
10             FlagID
12             flagstatus
14             flagtype
16             OWNER


Right?


----------------------------------------------------------------------

Comment By: Dave Brosius (dbrosius)
Date: 2002-02-20 06:34

Message:
Logged In: YES 
user_id=66596

Hey Frank, Thanks...

Of course,, dang your right. PlayeriD length is 8.... was 
looking at Address.unpack. I'm an idiot.

And *blush* yup client messages don't have playerIDs at 
offset 4. Slap me again. <Protocol.h is wrong then>

But, Where is the PlayerID at offset 20 for client 
messages?

    case MsgDropFlag:
    case MsgGrabFlag:
      if (len > 28)
	// server version
	patchPlayerId(fromId, toId, msg, 4);
>>>>	patchPlayerId(fromId, toId, msg, 20);
      break;

The client does:

void			ServerLink::sendDropFlag(const 
float* position)
{
  char msg[12];
  void* buf = msg;
  buf = nboPackVector(buf, position);
  send(MsgDropFlag, sizeof(msg), msg);
}


Then doesn't that if need braces?

    case MsgDropFlag:
    case MsgGrabFlag:
      if (len > 28) {
	// server version
	patchPlayerId(fromId, toId, msg, 4);
	patchPlayerId(fromId, toId, msg, 20);
      }
      break;





----------------------------------------------------------------------

Comment By: Frank Thilo (chestal)
Date: 2002-02-20 02:52

Message:
Logged In: YES 
user_id=48940

I think the current code is correct. I am not using E5
server right now, but when I used it I remember that at
least that part of it worked flawless. Other bugs have been
fixed in the meantime, except for the UDP incompatibility
with old clients (which of course is actualyl a bug in the
old clients).

The offset of +20 is correct, PlayerID has a size of 8
bytes, not 4 as in your table. The client messages do not
need patching because they do not contain a PlayerId at all.
For MsgGrabFlag the flag number is at offset 4, for DropFlag
it's the x coordinate.

----------------------------------------------------------------------

You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=103248&aid=520395&group_id=3248