QSPI read fail.

RE: QSPI read fail.

Florian Zierer — Wed, 31 Oct 2018 14:12:02 GMT

@haakonsh would it be possible to add an ASSERT in the SDK to check if the address is word aligned to e.g. nrfx_qspi_write and nrfx_qspi_read?

I had right now the same problem and it was really hard to find the cause of the problem.

RE: QSPI read fail.

haakonsh — Wed, 19 Sep 2018 11:59:01 GMT

From QSPI — Quad serial peripheral interface:

Listed here are the main features for the QSPI peripheral:

Single/dual/quad SPI input/output
2–32 MHz configurable clock frequency
Single-word read/write access from/to external flash
EasyDMA for block read and write transfers
Execute in place (XIP) for executing program directly from external flash
XIP execution speed: Up to 8 million instruction fetches per second for 16 bit instructions, up to 4 million instruction fetches per second for 32 bit instructions.

I think word-aligned address access is the standard for external flash chips.

RE: QSPI read fail.

Justin_Yeh — Wed, 19 Sep 2018 09:34:41 GMT

Yes, that's the root cause, Why data buffer must be aligned ? Could you help to explain this QSPI mechanism ?

Is this platform bug ? Thanks.

RE: QSPI read fail.

haakonsh — Tue, 18 Sep 2018 13:23:21 GMT

I think this is an issue where the addresses of your data are not word-aligned.

I need access to your source code if I am to review it.

RE: QSPI read fail.

Justin_Yeh — Fri, 14 Sep 2018 02:16:07 GMT

We found the root cause, if transmission data uses uint8_t, not the uint32_t, system will insert 24 bit zero at first, Could you help to explain why ? Thanks.

:020000041200E8
:10000000000000A5A5A5A5A5A5A5A5A5A5A5A5A58F
:10001000A5A5A5A5A5A5A5A5A5A5A5A5A5A5A5A590

......

RE: QSPI read fail.

haakonsh — Thu, 13 Sep 2018 11:10:41 GMT

Hey Justin,

Sorry for the late reply. I've not been able to reproduce the issue. I'm going to ask the developers/IC designers what they think.

Cheers,

Håkon.

RE: QSPI read fail.

Justin_Yeh — Mon, 10 Sep 2018 02:16:33 GMT

Hi, Do you have any update ? Thanks.

RE: QSPI read fail.

Justin_Yeh — Thu, 06 Sep 2018 11:00:07 GMT

Case 2

uart_cli:~$ extflash earse 0
done
uart_cli:~$ extflash write 0 4
done
[00:00:22.271,534] <info> app: test pattern is

0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 uart_cli 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5
0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 uart_cli 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5
done
uart_cli:~$ extflash read 0 32
0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 uart_cli 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5
0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 0xA5 uart_cli 0xA5 0xA5 0xA5 0x0 0x0 0x0
done

---------------------------------------

the nrfjprog log.

:020000041200E8
:10000000000000A5A5A5A5A5A5A5A5A5A5A5A5A58F
:10001000A5A5A5A5A5A5A5A5A5A5A5A5A5A5A5A590
:10002000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFE0
:10003000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFD0

......

RE: QSPI read fail.

Justin_Yeh — Thu, 06 Sep 2018 10:54:26 GMT

my test steps...

uart_cli:~$ extflash init
done
[00:00:20.095,334] <info> app: QSPI init started.
uart_cli:~$ extflash earse 0
done
uart_cli:~$ extflash read 0 32
0xFF 0xFF 0xFF 0xFF 0xFF 0xFF 0xFF 0xFF 0xFF uart_cli0xFF 0xFF 0xFF 0xFF 0xFF 0xFF 0xFF 0xFF
0xFF 0xFF 0xFF 0xFF 0xFF 0xFF 0xFF 0xFF 0xFF uart_cli0xFF 0xFF 0xFF 0x0 0x0 0x0
done

And the nrfjprog log.

:020000041200E8
:10000000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF00
:10001000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF0
:10002000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFE0
:10003000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFD0

.....

RE: QSPI read fail.

haakonsh — Thu, 06 Sep 2018 08:53:38 GMT

I want to replicate your issue on my end. I need a detailed description of your process and the file generated by nrfjprog when reading the ext flash.

RE: QSPI read fail.

Justin_Yeh — Thu, 06 Sep 2018 01:26:30 GMT

Hi haakonsh

Do you find any issue on my file ? Thanks.

RE: QSPI read fail.

Justin_Yeh — Wed, 05 Sep 2018 10:11:48 GMT

please give me your mail box, I send file to you.

RE: QSPI read fail.

haakonsh — Wed, 05 Sep 2018 10:06:10 GMT

I want to replicate your issue on my end. I need a detailed description of your process and the file generated by nrfjprog when reading the ext flash.

RE: QSPI read fail.

Justin_Yeh — Wed, 05 Sep 2018 09:33:53 GMT

I write 0x00, 0x01 ~ 0x1F to address 0 to 31, but it not match to nrfjprog.exe result.

:020000041200E8
:10000000000000000102030405060708090A0B0CA2
:100010000D0E0F101112131415161718191A1B1C98

it should be..

:020000041200E8
:10000000 00~0F
:10001000 10~1F

Do you know why ?

RE: QSPI read fail.

Justin_Yeh — Wed, 05 Sep 2018 09:20:44 GMT

I checked by nrfjprog.exe, read/write always lost the last 3~5 bytes.

RE: QSPI read fail.

haakonsh — Wed, 05 Sep 2018 08:08:35 GMT

This is wierd, do you mind sending me the file you generated?

Also have you tried to erase the ext flash with --qspieraseall ?

RE: QSPI read fail.

Justin_Yeh — Wed, 05 Sep 2018 07:52:25 GMT

C:\Program Files (x86)\Nordic Semiconductor\nrf5x\bin>nrfjprog.exe --readqspi ./
log
WARNING: An operation that can take up to several minutes is being executed.
WARNING: Please remain patient.
Initializing the QSPI peripheral.
Reading from the QSPI-connected memory.
Uninitializing the QSPI peripheral.
Storing data in './log'.

:020000041200E8
:10000000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF00
:10001000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF0
:10002000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFE0
:10003000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFD0
:10004000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFC0
:10005000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFB0
:10006000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFA0
:10007000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF90
:10008000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF80
:10009000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF70
:1000A000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF60
:1000B000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF50
:1000C000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF40
:1000D000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF30
:1000E000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF20
:1000F000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF10
:10010000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
:10011000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFEF
:10012000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFDF
:10013000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFCF
:10014000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFBF
:10015000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFAF
:10016000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF9F
:10017000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF8F
:10018000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF7F
:10019000FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF6F

....

RE: QSPI read fail.

haakonsh — Wed, 05 Sep 2018 07:42:33 GMT

See nrfjprog commands.

You can use --memrd to read out the content of the XiP memory region of the external flash, but you will have to use --readqspi if you want to read data stored outside of the XiP memory region.

RE: QSPI read fail.

Justin_Yeh — Wed, 05 Sep 2018 04:10:19 GMT

I don't sure how to check external flash content by nrfjprog, the read back data look are nrf52840 internal flash content. ??

C:\Program Files (x86)\Nordic Semiconductor\nrf5x\bin>nrfjprog.exe --memrd 0 --n
128
0x00000000: 20000400 00000999 0000062D 00000979 |... ....-...y...|
0x00000010: 00000637 00000641 0000064B 00000000 |7...A...K.......|
0x00000020: 00000000 00000000 00000000 000009BD |................|
0x00000030: 00000655 00000000 0000065F 00000669 |U......._...i...|
0x00000040: 00000673 0000067D 00000687 00000691 |s...}...........|
0x00000050: 0000069B 000006A5 000006AF 000006B9 |................|
0x00000060: 000006C3 000006CD 000006D7 000006E1 |................|
0x00000070: 000006EB 000006F5 000006FF 00000709 |................|

RE: QSPI read fail.

Justin_Yeh — Wed, 05 Sep 2018 04:01:10 GMT

like this ?

nrfjprog.exe --memrd 0 --w 32

0x00000000: 20000400 |... |

RE: QSPI read fail.

haakonsh — Tue, 04 Sep 2018 13:42:08 GMT

By verifying I mean comparing what you read out with nrfjprog with what the uart log prints.

RE: QSPI read fail.

Justin_Yeh — Tue, 04 Sep 2018 13:01:04 GMT

Could you teach me how to verify content by nrfprog ? Thanks.

RE: QSPI read fail.

haakonsh — Tue, 04 Sep 2018 12:26:13 GMT

We need to verify that the content of the external flash has changed. Can you read out and verify the the content of the external memory with nrfjprog after you have erased it?