SBL2E XA Error in ReadDataCb

[nsclfrib]

My application is using the buffered TCP Network interface to handle a Telnet connection.

After about 10-12 hours of operation, I get a message on the debug console
"
"****** Error in ReadDataCb, nPut=81, nGet=81
"
Have no idea what/where that is coming from, except once it comes in, only way to get things restarted is by a reset.

This won't do for a device that's going to be buried inside equipment.

Any idea's what is causing this?

[rnixon]

What buffer size did you specify when you created the object?

[nsclfrib]

I allocated 256 bytes. Here's the code:
"
// create tcp buffer object
#define TCP_BUFF_SIZE 256
BYTE buff_obj_buffer[TCP_BUFF_SIZE];
BufferedTcpObject tcp(buff_obj_buffer, TCP_BUFF_SIZE, NULL);
"

Just got another error, last night. Message on the debug console is:
"
***** Error in ReadDataCb, nPut=242, nGet=242
"

Later-
David

[nsclfrib]

A little extra information related to this latest crash (since it happened before, I trapped when it happened this time):

My client sends a continuous stream of 4 byte commands to the NetBurner. It received somewhere around 8000-10,000 of them before this error, and logged a continuous stream of commands / responses on the terminal. When the error occurred, the NetBurner had been sent a 4 byte command, but it only received the first 2 bytes of this command, and then this error occurred. The client sent the 4 bytes out (according to it's log), but the NetBurner only received the first 2.

[pbreed]

Strange.....
Can you modify the file nburn\system_sc\tcp_buffer.cpp

In the function

void BufferedTcpObject::ReadDataCb(PBYTE Data, int len)

Add the following code before the for loop:

if(len > RxWindowCb())
{
iprintf("****** ERROR More data than the allowed window! len=%d win=%d nPut=%d, nGet=%d\r\n",len,RxWindowCb(), nPut, nGet);
}


Recompile everything and see if you get the "new" error.....

Paul

[rnixon]

Not sure how difficult it would be for you, but a wireshark trace would really help to determine what is going on, since it appears to be related to window size and flow control.

[nsclfrib]

I added the code indicated into tcp_buffer.cpp,(and changed the working of the other error message as well, to insure that the new code got loaded), however when I get the error, I just get my new error message.

Later-

[nsclfrib]

I meant, I get my reworded error message. The new code I added doesn't return an error.

[pbreed]

That is strange....
The combination you describe should not happen unless some kind of memory corruption is happening....
Do you have other tasks that might be overflowing their stack?

[nsclfrib]

This application only has a Telnet socket enabled now.

I've disabled most everything else.

When I look at the uc/OS Tasks dump I get (real disappointed that mtty doesn't let me grab text)

uc/OS Stacks
Prio StackPtr Stack Bottom Free Now Minimum Free
31 | 0x20005ac8 | 0x20005920 | 424 | 288
5 | 0x20006878 | 0x20005cd0 | 2984 | 2460
10 | 0x20002ab0 | 0x20001f20 | 2960 | 2740

uc/OS Tasks
Prio State Ticks Call Stack
31??|Ready |forever |at: ffc09b02
5??|Running
NET |Ready |forever |ffc09f0c->ffc06e0c-><END>
: