Again
Registers are global ! That means if r6 was written in main thread, it stores same data when cpu jumps to your subroutine, do rhings in it and rets back. So just avoid using registers which first instruction after your custom call are reading from register and feel free if its write to register
Its just a simple logic idk how to say

Is this okay?

So this is how original look like



I set here call function



In my function I check 386000( my flag) to set zwgru to -30 or to not touch it all (leave at it is)



After that I put this thing two things like was original (still don't know what this [r0+] mean )


After some rest I will check if I forget something in code or if something is wrong .

Only question I want to know did I made jump and return from and to original function?

Don't have car here to test it

I am still exploring how all this works. I know I can do that way, but I want to find more ways so I can understand it better.


RL4 or R4?  I know that RL4 is low byte of R4, so touching RL4 will change value of R4.  But general speaking zwgru is uint_8 value?
If after:  movb  zwgru, rl4
...
XXXX
...
I have DB 00 (like in this case)   does that mean, that in most case rl4 value is no longer needed(it will be most likely overwriten in some next function) so I am free to use it XXXX area



Tnx for shortcout about Z flag and all other help

; conditions
exts   #38h, #1
movb   rl5, LC_flag
jmpr    cc_z, loc_end
;eof

;params conditions

loc_cut:
exts   #8Ah, #1
movb   rl4, LC_zwgru

loc_end:
movb    zwgru, rl4
rets

This is code of my LC, called from  loc_8B89B2 (in your case).
new fucntion called in place where rl4 moved to zwgru. so rl4 contain zwgru value and you want to touch it in your routine, rl5 used for conditional checks
8bytes signed int8_t


absolutely no. check how map calculations are done. in most cases they writes calculated value into r4 and rets to main code
general rule here - look at the main code flow after your routine (even after rets) if you see some register reading first, then you cannot use it. just imagine this situation

mov r5, rl_w
mov r4, plsol_w
shr r4, #8
movb plsol, rl4
mov rlkh_w, r5
mov r7, rkat_w

and you change it to

mov r5, rl_w
mov r4, plsol_w
shr r4, #8
calls #8Ah, myRoutine ; < r4 contain plsol value!
mov rlkh_w, r5 ;  r5 goes through your routine, do not use it!
mov r7, rkat_w ; r7 only written here, so in your routine its not contain any useful data, could be used

<myRoutine:>

mov r4, nmot_w ; < and now you changed it to nmot_w
cmp r4, myNmotLimit
jmpr cc_ule, loc_end
....
loc_end:
movb plsol, rl4 ;< feeding plsol with part of nmot (bad idea)
rets

as of the last screen - this wont work
you overwrite zwgru with stock value anyway

I assume this is not possible but can I make call function after I already make call


Working example:
at 8A000   I make Call 8E540
At 8E540 my function
....
...
Rets (this will return to 8A004)

But If I make something like this

at 8A000   I make Call 8E540
At 8E540 my function
....
Call 8E640
Rets

At 8E640 my function#2
....
Rets  

this will return to next array where I called 8E640, but after that I will again have rets. Will after that rets send to 8A004?

Reading the user manual helps.
Every time a call is made the PC is pushed onto the stack and popped off with RET.
In case of CALLS also the segment is pushed onto the stack and then both the segment and the address get popped off.
As long as you don't exceed the stack size you can chain as many calls as you like, obviously the amount of calls and returns has to be balanced.

If you are patching an already existing call, you can also just jump to the original call location after you're done with your code.
This is becomes a lot more important on TriCore where calls automagically save and restore a bunch of registers.

You can safely use few calls
Personally ive never faced with stack overflow with 1-2 incapsulated calls and map calculation (vars pushed in stack)