drivers: flash: andes_qspi: Fix bug when RX length exceeds 512 bytes and optimize the check for being used as a flash fetch device.

[kevinwang821020]

This PR addresses two main issues.
The first one is fixing a bug that occurs when the flash receives more than 512 bytes of data. The root cause is that ATCSPI hardware only supports a maximum of 512 bytes per transfer. Therefore, if the received data exceeds this limit, the transfer needs to be split into multiple transactions.
The second issue is related to determining whether the SPI is used as a flash fetch device in XIP mode. Being in XIP mode does not necessarily mean that the program is executed from flash. The SPI node is only used as a flash fetch device when the program is actually running from flash. In other cases, the SPI node can still exist.

[de-nordic]

@kevinwang821020 Can you log Zephyr bug for what you fix here and link the issue to this fix?

[kevinwang821020]

Hi @de-nordic ,
This bug does not appear in the tests or samples in the Zephyr upstream package (because single read operations do not exceed 512 bytes).
We discovered this issue based on the atcspi200 hardware specifications and verified its existence by local testing.
Therefore, no issue has been filed.

[de-nordic]

Hi @de-nordic , This bug does not appear in the tests or samples in the Zephyr upstream package (because single read operations do not exceed 512 bytes). We discovered this issue based on the atcspi200 hardware specifications and verified its existence by local testing. Therefore, no issue has been filed.

I am not sure whether I am allowed to insist on reporting the bug, as it is in your platform.

Nevertheless, I think that a role of a bug report, aside from reporting a problem for somebody to fix, is a way to inform users on a problem so that people with similar problems could find information that the issue may not be specific for their setup, and also provide information that the issue is tracked and being addressed.

@kartben What do you think?

Also, In case you would like to backport the issue fix, into previous release, then bug report is needed.

[kevinwang821020]

I think this is a great suggestion. If others feel the same way, I will follow this suggestion and raise an issue (bug).
By the way, I think this issue may not be necessary to backport to previous release. After all, it seems more like an improvement to this flash driver's tolerance for read length limits (even though the limit was rarely encountered).

[de-nordic]

Looks OK to me.
I will re-assign the bug to representative of Andes to make sure vendor agrees on the change.

[wtlee1995]

Looks good to me.