IceCore myStorm firmware

Folknology 2019-08-03 16:58:29 UTC #281

The flash code now looks sound, however I'm still not getting it to boot from flash, if anyone cares to take a look please do so as I pushed it..

lawrie 2019-08-03 17:18:02 UTC #282

Your latest mystorm.raw works for me and boots from flash for me. I did not rebuild it. That suggests you haven't got a valid bitstream in flash.

I then tried using your firmware to do the flashing, which I assume I do by press the mode button to switch the yellow mode led on and then cat the bitstream to /dev/ttyACM0. That hung. And after unplugging and plugging back in, it was no longer booting from flash, presumably as something had been written to flash.

Folknology 2019-08-03 17:34:49 UTC #283

Ok thanks @lawrie looks like I have some more work todo, must be corrupting the image during write somehow..

lawrie 2019-08-03 17:41:36 UTC #284

I am currently modifying one of my SoC programs to look at the image you write, and compare it with one that I write.

One difference is that you don't write the 0xFF0000FF, but I do.

But for me the cat hangs and the red status LED stays on when I try to use your firmware to do the flash, so the full bitsream doesn't get written.

Folknology 2019-08-03 18:18:39 UTC #285

Do I need that?

Cat a regular Ice40 program first before the pressing the mode button and cat'ing it again for flash, how does that go for you?

AndyCap 2019-08-03 18:23:45 UTC #286

No you don't.

@lawrie You definitely need to do this.

@Folknology Do you have any ideas about getting the Ice40 to not mess around with the SPI when there is no valid flash image?

I will do a merge and test tomorrow, I have code that can read the flash image and compare it so we can see the problem, to many beers drunk to do this now.

lawrie 2019-08-03 18:35:50 UTC #287

I don't believe so, but I always write it as that was easiest for me, so I know if works with it. But @AndyCap confirms it works without it.

That seems to be better, although the the green CDONE led seems to come on briefly and then go off.

The problem with the flash image seems to be that the end of it is not written. It is all FFs from address 0x20700.

Folknology 2019-08-03 18:43:20 UTC #288

That's odd, but might be a clue..

Folknology 2019-08-03 18:46:01 UTC #289

That just means it resets ok, but then CDone fails (bad image in flash)

lawrie 2019-08-03 18:53:38 UTC #290

This is the start of flash after doing a flash with your firmware:

000000 7E99AA7E 05010051 62200092 01726703 00008210 01010011 00000000 00000000
000020 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000

And this is where it ends:

0206E0 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
020700 FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF

It should end at 0x20FB8.

Folknology 2019-08-03 18:58:00 UTC #291

Yeah I'm trying to work out where I'm dropping 2k+ bytes!!

lawrie 2019-08-03 19:49:44 UTC #292

Is it this line, at line 770 in Flash:Stream:

while (++nbytes < end){

Should it be:

while (nbytes++ < end){

I think that might lose 1 byte per 64-byte USB packet and (135100 / 64) is just over 2k.

There is also some confusion over whether the parameter to erase is the command code or the number of kb (32 or 64).

lawrie 2019-08-04 07:44:41 UTC #293

No that still increments nbytes too much. Its should be:

                while (nbytes < end){
                        nbytes++;

I tested that, and it works.

And I no longer have to send an image in non-flash mode before flash mode works.

I removed the call to mode_led_toggle in erase_write as it was confusing having the mode led turned off while still in flash mode.

Folknology 2019-08-04 09:12:30 UTC #294

I think you need to show the whole chunk of code, I changed it last night to:

while (nbytes < end){
pagebuf[hd++] = *img++;
if( (nbytes >= NBYTES) || (hd == 256) ) { // Lets write flash page/remainder
  if(erase_write(ERASE64)) 
    err = 1;
  hd = 0;
  if(nbytes >= NBYTES) 
    break;
  } 
  nbytes++;
}

However this still doesn't boot from flash..

lawrie 2019-08-04 09:14:20 UTC #295

                while (nbytes < end){
                        nbytes++;
                        pagebuf[hd++] = *img++;
                        if( (nbytes >= NBYTES) || (hd == 256) ) { // Lets write flash page/remainder
                                if(erase_write(pagebuf, hd, ERASE64)) // We don't need to pass pagebuf or hd!
                                        err = 1;
                                hd = 0;
                                if(nbytes >= NBYTES)
                                        break;
                        }
                }

Folknology 2019-08-04 09:14:35 UTC #296

I use this to count how many 64K erases occur as initially I had an issue with this, I will use the led for progress later..

lawrie 2019-08-04 09:18:51 UTC #297

As we currently have two modes, I like the led to show which mode we are in.

My code, posted above, is definitely working, as I can flash images and they are immediately started and still there when I remove and reinsert the cable.

Folknology 2019-08-04 09:19:26 UTC #298

lawrie:

while (nbytes < end){
                        nbytes++;
                        pagebuf[hd++] = *img++;
                        if( (nbytes <= NBYTES) || (hd == 256) ) { // Lets write flash page/remainder
                                if(erase_write(pagebuf, hd, ERASE64)) // We don't need to pass pagebuf or hd!
                                        err = 1;
                                hd = 0;
                                if(nbytes <= NBYTES)
                                        break;
                        }
                }

I don't understand this, how it would work, also I tried it just in case, it didn't work for me

Folknology 2019-08-04 09:19:48 UTC #299

Yes I return the led to the mode at the end of programming

Folknology 2019-08-04 09:22:18 UTC #300

Why is it working for you but not me?