PCM Hammer P12 development

[Gampy]

Antus, I understand the expense for nothing ... Not a big deal to me, gives me something to do.

2004 Trailblazer uses the P10 PCM.

Thank ya sir!

Well, we now know the dog lives for between 20,000 and 40,000 iterations ... 20,000 iterations returns ack, 40,000 does not.

So, now it's time to start hitting Dog scratching techniques with a 40,000 (or greater) iteration loop ...

I've been using the technique,

Code: Select all

    move.b  #0x55, (0xFFFA55).l         | Reset COP1
    move.b  #0xAA, (0xFFFA55).l
    bclr    #7,   (0xFFFA21).l         | Reset COP2
    bset    #7,   (0xFFFA21).l      

Obviously it fails to keep the dog happy.

So, the next test is,

Code: Select all

    move.b  #0x55, (0xFFFA55).l         | Reset COP1
    move.b  #0xAA, (0xFFFA55).l
    eori.b  #0x80, (0xFFFA21).l

with a 40,000 iteration loop.

And has been sent ...

Any thoughts, input on dog scratching techniques are welcome!

[antus]

It changes nothing but food for thought - the way the code is written it looks like the 55/AA dog needs constant scratching, and the bit flip dog, might be a start timer, then disabled at the end of a critical section. For our purposes though, scratching them both together constantly is probably fine.


edit: unless the second one is much shorter than the 55/AA one, and maybe the order of set then clear or clear then set matters.

[Gampy]

So other then food for thought, am I wasting my time ...

[antus]

of course you are! hahaha no just jokes. your time is up to you. its gotta be possible, but if I were you I would get a pcm, so you can just tweak things one step at a time and try try again. also note that there is a comment about keeping the slave quiet here, maybe your hitting in to lack of that. viewtopic.php?f=42&t=7742&start=10#p114215

[Gampy]

My goal ATM is to get a loop that we can keep alive ... And when we exit our code, we get an ack from the Os.

I do not believe the first dog is happy (FA55 / FA21) ... 40,000 iterations is nothing time wise, I believe that is over it's lifespan, I'm guessing the timeout is closer to 30,000 (like 25,000) iterations.
Really no way to tell what's causing it to die ...

For my next test (after the one that's out), I'm going to try the 20,000 and 40,000 loops without scratching the dog, and see if we get the same results as with it scratched.

If I'm on the wrong track, by all means kick me off!
Kick hard, I am extremely bull headed ... I won't take offense.

[Gampy]

Holy cow ... Just got the results from a second 40,000 iteration test, we got ack!

This was the test that flipped (eori) the COP2 bit versus bclr/bset.

I have just sent a longer loop ...

[darkman5001]

Where in the hardware is the watchdogs and timers located? Are they a separate chip from the CPU and flash chip? And what is the COP? Trying to learn this terminology.

[Gampy]

And just got the results from that, it survived 0x13880 (80,000) iterations ...
I stepped it up BIG ... Just sent one with 0xFFFFFF (16,777,215) iterations!

I think that should say the dog is happy if successful ... What do you all think!

[Gampy]

Where in the hardware is the watchdogs and timers located? Are they a separate chip from the CPU and flash chip? And what is the COP? Trying to learn this terminology.

The WatchDog and Timers are in the CPU ...

Actually technically speaking the Watchdog is called COP (Computer Operating Properly) ... They are one and the same!

And I apologize for using COP, it's what I know ... Mostly they use Watchdog around here.

[darkman5001]

Okay, starting to make more sense now.