How to start

ManAtWork

Maybe writing to flash fails after the first page or something like that. As I said earlier I'd feel a lot more confident if the flash programmer would do a verify.

ManAtWork

Hmm, the program seems to be a lot shorter than one page.

TAQOZ# $10000 $180 sf dump ---
0001_0000: 60 00 00 00  00 24 60 FD  E8 01 80 FF  1F 20 65 FD     '`....$`......
 e.'
0001_0010: 03 24 44 F5  00 24 60 FD  3E F8 0C FC  80 08 80 FF     '.$D..$`.>....
...'
0001_0020: 3E 0E 1C FC  41 7C 64 FD  09 90 5C FB  08 82 5C FB     '>...A|d...\..
.\.'
0001_0030: 07 A0 5C FB  06 A0 5C FB  05 B2 5C FB  04 40 5C FB     '..\...\...\..
@\.'
0001_0040: A1 07 80 FF  1F 80 64 FD  DC FF 9F FD  48 03 04 01     '......d.....H
...'
0001_0050: 3E FE 97 FA  40 7C 6C FD  3E EE 27 BC  2D 00 7C BD     '>...@|l.>.'.-
.|.'
0001_0060: EC FF 9F FD  00 00 00 00  00 00 00 00  00 00 00 00     '.............
...'
0001_0070: 00 00 00 00  00 00 00 00  00 00 00 00  00 00 00 00     '.............
...'
(00 repeating until end)

evanh

That's the second stage blinky program at that address. Currently unused.

The testing program is in the first stage located at the start of the flash, from address 0. Here's the start of it as placed in the binary:

00400 000             | stage1	
00400 000 59 7A 64 FD | 		drvh	#spi_cs			'SPI chip select - start not selected!
00404 001 50 74 64 FD | 		fltl	#spi_do
00408 002 50 76 64 FD | 		fltl	#spi_di
0040c 003             | 
0040c 003             | 'config one smartpin for SPI clock
0040c 003 50 78 64 FD | 		fltl	#spi_clk
00410 004 3C D0 04 FC | 		wrpin	.mode_clk, #spi_clk			'unregistered SPI clock pin, helps with read timing
00414 005 3C CC 14 FC | 		wxpin	.spi_clk_ca, #spi_clk			'sysclock/8 bitrate (min of /8 for tx smartpin!)
00418 006 41 78 64 FD | 		dirh	#spi_clk
0041c 007             | 
0041c 007 21 F8 64 FD | 		setse2	#E_PIN_RISE | spi_clk	'setup event on completed SPI clock pulses from smartpin
00420 008             | 
00420 008             | 'faster sys-clock for faster loading
00420 008 00 BE 60 FD | 		hubset	 .clk_mode		'config crystal and PLL - still running RCFAST
00424 009 1F C0 60 FD | 		waitx	 .pause			'wait for crystal/PLL to ramp up
00428 00a 03 BE 44 F5 | 		or	 .clk_mode, #XSEL	'select clock mode
0042c 00b 00 BE 60 FD | 		hubset	 .clk_mode		'engage
00430 00c             | 
00430 00c             | 'set up serial output for progress messages
00430 00c 40 7C 64 FD | 		dirl	#ser_tx_pin
00434 00d 3E F8 0C FC | 		wrpin	#SPM_ASER_TX, #ser_tx_pin
00438 00e 00 F3 81 FF 
0043c 00f 3E 0E 1C FC | 		wxpin	##(CLOCKFREQ / SERIAL_BAUD)<<16 | 7, #ser_tx_pin
00440 010 41 7C 64 FD | 		dirh	#ser_tx_pin
00444 011             | 
00444 011             | '------------------------------------
00444 011 38 DA 4C FB | 		callpa	#.title_msg, #.puts
00448 012 64 00 B0 FD | 		call	#.test_modes		'clk=REGD, data=UNREGD
...

evanh

The code for programming the flash is virtually untouched from Brian's original. I had tweaked the clock ratio a small amount to make it 50:50 is all. The fact that there is valid code at the right location seems it's working fine.

I might have a look at the ROM loader, it'll be different again ...

ManAtWork

Ok, the code at $000 is there and matches. So it must be the ROM bootloader. If that's true I have to get another flash chip, I fear. As long as I don't know the cause of the incompatibility I can only try...

Issi IS25LP080
Microchip SST25VF080
Amic A25L080
Adesto AT25SF041
ON Semi LE25U40
Winbond W25Q80

BTW, the latter is also quite cheap ($0,28)

ManAtWork

I've repeated your test with the Winbond W25Q80 and the Adesto AT25SF041 flash chips. Both give the same results

Loading G:/Projekte/Speedcube3/Source/Propeller/P2ES_flashloader-2MHz.binary - 4
224 bytes
chksum: 97 OK
( Entering terminal mode.  Press Ctrl-] to exit. )
P2-ES Flash Programmer
Erasing flash...
Programming...
Done
  DPI   Dual   Fast   SPI
--------------------------
 Pass   Pass   Pass   Pass
 Pass   Pass   Pass   Pass
 Fail   Fail   Fail   Fail
 Pass   Pass   Pass   Pass
 Pass   Pass   Pass   Pass
 Pass   Pass   Pass   Pass
 Pass   Pass   Pass   Pass
 Fail   Fail   Fail   Fail

So they should be compatible with the Dual SPI modes you use. I looked up in the data sheet that the Adesto chip supports single pin read up to 50MHz and dual read mode up to 104MHz.

evanh

Yep, so far, every part tested seems to support all modes. Only very old flash chips don't.

Aside, I was pondering just adopting Chip's much simpler code since it goes plenty fast enough. But then I realised that I'd still like to be able to run tests on the revA boards with the revA IC's.