feat(spi_nand_flash): Add Kconfig option to verify write operations

[RathiSonika]

Change description

1. Added an Kconfig option to verify write operations which will be useful when debugging hardware issues or issues with the driver itself.
2. Fixed an issue in the spi_nand_erase_chip() API where the mutex was not released, leading to a deadlock if another API attempted to acquire the mutex after calling spi_nand_erase_chip(). Updated test case for the same.
3. Added spi_nand_flash_copy_sector() API in nand.c and test case for the same.

[igrr]

Thanks for the PR @RathiSonika.

I left a few minor comments. The major one is whether we should indeed verify writes above Dhara layer, like you did, or below it, in nand.c dhara_glue.c.

To me it seems that verifying writes at lower layer achieves the goal of making sure our Flash operations work correctly, while verifying at upper layer doesn't.

We do writes to Flash in dhara_nand_prog, dhara_nand_copy, dhara_nand_mark_bad. If we fail to perform the write in dhara_nand_mark_bad correctly, we might not get the verification error in spi_nand_flash_write_sector. Dhara will consider the page as "bad" until next reset, but after reset we will mistakenly detect the page as not "bad" because the bad block marker wasn't written correctly.

[RathiSonika]

The major one is whether we should indeed verify writes above Dhara layer, like you did, or below it, in nand.c.

I had the same thought to add it in below Dhara layer. Currently added it to nand.c. I wasn’t entirely sure if it should go in dhara_glue.c, given its tight coupling with the Dhara library. However, you’re right; we should verify the write operation below Dhara layer. So, for the current structure, it makes sense to add this in dhara_glue.c. We could consider refactoring this in the future.

[igrr]

Technically we also have to verify the used marker here

[RathiSonika]

Oh, I see. So you mean it should return -1/failure if the write verification fails, correct?

[igrr]

I mean that a few lines above, we write two parts:

    ESP_GOTO_ON_ERROR(spi_nand_program_load(dev->config.device_handle, data, 0, dev->page_size),
                      fail, TAG, "");
    ESP_GOTO_ON_ERROR(spi_nand_program_load(dev->config.device_handle, (uint8_t *)&used_marker,
                                            dev->page_size + 2, 2),

This s_verify_write call checks that the first part (from 0 to page_size) is written correctly. We should call s_verify_write second time to check that the used page marker (page_size ... page_size + 2) is also written correctly.

[RathiSonika]

Oh, my mistake—I misunderstood. Yes, it should verify the markers. I’ll update it. Thank you!