[Ti-Asm(83+)] Link port problems

Deathsting · Post by **Deathsting** » Sat 30 Jun, 2007 9:03 am

I cant figure out whats wrong with this routine.
well i have two calc one runs CheckServer while the other runs CheckHost
When executed Check host routine executes and then exits while the CheckServer routine gets stuck at CheckServer1.
Any idea to why this is happening??? Am i checking the right bits???

Code: Select all

CheckServer:
	ld a,3     ; Set tip low, ring low
	out (0),a
CheckServer1:
       	in a,(0)                             
	bit 0,a         ; Check tip
	jr z,CheckServer1		;tip is not yet high wait till high
            jp exit

CheckHost:
	in a,(0)            
	bit 0,a         ; Check tip
	jr nz,CheckHost		;wait till low
CheckHost1:
	ld a,2     ; Set tip high, ring low
	out (0),a
            jp exit

Also to my knowledge both the ti lcd and keypad require a specific delay before each read and write, do the link ports require any such delays??

Jim e · Post by **Jim e** » Sat 30 Jun, 2007 12:13 pm

Sorry some one forgot to write something extremely important on the wiki.

The values you write to the port are inverted of the values you read. So writing 0 you would read 3 which means both lines are high. Writhing 3 you would read 0 so both lines are low.

The link port doesn't really need a delay I believe.

Deathsting · Post by **Deathsting** » Sat 30 Jun, 2007 5:52 pm

thank you :bow
Finally...this thing has been driving me crazy since yesterday...i ll try it out soon

Timendus · Post by **Timendus** » Mon 02 Jul, 2007 3:41 pm

@Jim: He is using the Bell defines, so the read and write values should be just fine.

As I wrote in my last e-mail he's got the logic mixed up. If the "server" routine pulls the tip low, there's no way the "client" routine can get it back up. That's just not how it works.

I'm sorry I'm not around more often, but I have many many other things to work on besides z80, so it's a good thing you e-mailed me, Deathsting.

Jim e · Post by **Jim e** » Mon 02 Jul, 2007 6:38 pm

Bell defines?

Deathsting · Post by **Deathsting** » Mon 02 Jul, 2007 7:02 pm

umm after realizing that the bits were wrong i used some code from timendus's "BELL" routines and emailed him...
Anyway i sort of get it (although not completly)

SOMEBODY should fix Wiki TI

One thing to keep in mind when writing link port related software is that the Ti-OS checks for silent transfers in the background of input routines. When two calculators are connected and one pulls a line low, the other calculator will respond by pulling the other line low to acknowledge that it has received a bit. Even when the other calculator is running an assembly program that uses the getkey romcall. This phenomenon is known to cause severe headaches for programmers who attempt to write synchronization routines .

That doesnt talk much about any or gate now does it

KermMartian · Post by **KermMartian** » Mon 02 Jul, 2007 7:31 pm

Deathsting wrote:umm after realizing that the bits were wrong i used some code from timendus's "BELL" routines and emailed him...
Anyway i sort of get it (although not completly)

SOMEBODY should fix Wiki TI
One thing to keep in mind when writing link port related software is that the Ti-OS checks for silent transfers in the background of input routines. When two calculators are connected and one pulls a line low, the other calculator will respond by pulling the other line low to acknowledge that it has received a bit. Even when the other calculator is running an assembly program that uses the getkey romcall. This phenomenon is known to cause severe headaches for programmers who attempt to write synchronization routines .
That doesnt talk much about any or gate now does it

Why don't you fix it? That's the point of a wiki.

Yeah, this always confused the hell out of me until some combination of Timendus and Ben Ryves helped me wrap my head around it.

Timendus · Post by **Timendus** » Mon 02 Jul, 2007 7:54 pm

Deathsting wrote:That doesnt talk much about any or gate now does it

It doesn't. Slightly higher on the page, under "read values" though it reads:

When idle (no transfer in progress, no cable plugged in, etc), both lines are usually high (1). When a cable is connected on both ends, a line reads high if and only if both ends have set the line high. The line will read low if either calculator sets it low.

In other words; there's nothing wrong with the WikiTi entry, you just need to read more carefully

But perhaps it could benefit from a nice schematic showing how it works.

Deathsting · Post by **Deathsting** » Tue 03 Jul, 2007 10:10 am

why do i have the weird feeling that someone just added that in...i dont remeber seeing

Bit 0 is the tip and bit 1 is the ring."

...and that just drove be crazy

@Timenedus - u were right rgp and bell screw each other up...the interupts in rgp seem to make bell routines go out of sync and for some odd reason bell_connect just causes both calculators to hang

Timendus · Post by **Timendus** » Tue 03 Jul, 2007 10:19 am

Yup, that part has been added on the 30th of june by Jim e. It's a wiki, you can look back it's history

Edit: You could try to increase the timeout delay, but I doubt it's going to help much. Maybe check if RGP does any weird magic with the link port. But as I said; you're the first to try to combine the two, so I'm not surprised that things don't work smoothly.

Jim e · Post by **Jim e** » Tue 03 Jul, 2007 12:04 pm

I'm not surprised. It takes about 60% of the cpu time. If you want linking you'll probably need to use the link assist. 83+ and the 83+SE Both have a receiving link assist with a 2 second timeout.

But honestly I don't suggest linking with grey. it will only slow things down more.