Assembly language bug in source code

[+9640News]

I've been going over this XMD source file in the attachment for quite some time. The source file is for the Geneve, but for any assembly language programmers, the fact it is a Geneve file does not matter if you are familiar with assembly language.

 

The attached file (XMD) in the DSK image is an XMODEM transfer routine.

 

I recall years ago, Mike Maksimik informed me there was a bug and I found it once before. I thought it was on the Receive portion of the file transfer, but it could be on Send side. I just do not see it in either section. It's possible I had to correct it on both sides, just not 100% sure.

What I do remember was the bug was specifically related to an issue of a bad block during one of the 128/1024 block transfers. The error was with one of the following registers for the send and/or receive routine:

 

R6 Buffer Pointer

R7 Record Number

 

I believe the code was incrementing the Buffer Pointer too soon, and with an error, it offset the buffer pointer by either the 128/1024 block transfer. The solution was in the error responding code to decrement either the Buffer Pointer or the Record Number (or both).

 

In 2003, I left a comment in the source code after I setup the Geneve emulator pointing to where I thought the error was. Then, and now, I can not see the issue.

 

If someone else can take a look at it, I would appreciate it. I have attached a DSK image, as well as TXT file so anyone can view it under whatever means they desire.

 

Thanks for any help.

 

Beery

 

 

Disk0005.dsk

XMD.txt

[+9640News]

I've been going over this XMD source file in the attachment for quite some time. The source file is for the Geneve, but for any assembly language programmers, the fact it is a Geneve file does not matter if you are familiar with assembly language.

The attached file (XMD) in the DSK image is an XMODEM transfer routine.

I recall years ago, Mike Maksimik informed me there was a bug and I found it once before. I thought it was on the Receive portion of the file transfer, but it could be on Send side. I just do not see it in either section. It's possible I had to correct it on both sides, just not 100% sure.

What I do remember was the bug was specifically related to an issue of a bad block during one of the 128/1024 block transfers. The error was with one of the following registers for the send and/or receive routine:

R6 Buffer Pointer

R7 Record Number

I believe the code was incrementing the Buffer Pointer too soon, and with an error, it offset the buffer pointer by either the 128/1024 block transfer. The solution was in the error responding code to decrement either the Buffer Pointer or the Record Number (or both).

In 2003, I left a comment in the source code after I setup the Geneve emulator pointing to where I thought the error was. Then, and now, I can not see the issue.

If someone else can take a look at it, I would appreciate it. I have attached a DSK image, as well as TXT file so anyone can view it under whatever means they desire.

Thanks for any help.

Beery

XMD.txt

Disk0005.dsk

[+mizapf]

If you are interested for a MDOS-based XMODEM, I can offer you my tool, natively writting in assembly language for MDOS.

 

http://atariage.com/forums/topic/255547-xmodem-utility-for-geneve/

[+9640News]

If you are interested for a MDOS-based XMODEM, I can offer you my tool, natively writting in assembly language for MDOS.

 

http://atariage.com/forums/topic/255547-xmodem-utility-for-geneve/

 

I grabbed the DSK image, however, I actually have more grander plans in mind and I was wanting to start with the source I had from My-Term that ran under MDOS.

 

I knew there was that one bug in the original My-Term source code I received from Riccio. I spent more time today going over both routines and it is not obvious and I am fairly certain this was the original source code Riccio sent me years ago and not source I went back through and fixed in the 90's. If nobody else is able to see the issue, I may need to go back and compare the distributed program with the image of late to see where those lines of code were.