Spi flash fix

[EmergReanimator]

Tested with STM32 F4VE board

https://stm32-base.org/boards/STM32F407VET6-STM32-F4VE-V2.0.html

Re-configuration the SPI on STM32F407 causes SCK signal to toggle.
This leads to Fast ReaD (0x0B) command failure in check_fs: 0x0B is treated as 0x85

  /* Find an FAT partition on the drive. Supports only generic partitioning rules, FDISK (MBR) and SFD (w/o partition). */
    bsect = 0;
    fmt = check_fs(fs, bsect);          /* Load sector 0 and check if it is an FAT-VBR as SFD */
    if (fmt == 2 || (fmt < 2 && LD2PT(fs) != 0)) { /* Not an FAT-VBR or forced partition number */
        for (i = 0; i < 4; i++) {       /* Get partition offset */
            pt = fs->win + (MBR_Table + i * SZ_PTE);
            br[i] = pt[PTE_System] ? ld_dword(pt + PTE_StLba) : 0;
        }
        i = LD2PT(fs);                  /* Partition number: 0:auto, 1-4:forced */
        if (i != 0) i--;
        do {                            /* Find an FAT volume */
            bsect = br[i];
            fmt = bsect ? check_fs(fs, bsect) : 3;  /* Check the partition */
        } while (LD2PT(fs) == 0 && fmt >= 2 && ++i < 4);
    }
    if (fmt == 4) return FR_DISK_ERR;       /* An error occured in the disk I/O layer */
    if (fmt >= 2) return FR_NO_FILESYSTEM;  /* No FAT volume is found */

[tannewt]

I don't think you want to do this. Removing this configures will break cases where the flash is on a shared SPI bus with a device that has different settings. This is rare but possible.

Instead, the port should guard against repeated calls and exit early. https://github.com/adafruit/circuitpython/blob/main/ports/atmel-samd/common-hal/busio/SPI.c#L216-L222 This protects the SPI from similar issues when used by SPIDevice which also calls configure before each transaction.

[EmergReanimator]

Removing this configures will break cases where the flash is on a shared SPI bus with a device that has different settings. This is rare but possible.

I see. I also was not sure about this change. But I was not aware that shared SPI bus use cases are possible with circuitpython.

Instead, the port should guard against repeated calls and exit early. https://github.com/adafruit/circuitpython/blob/main/ports/atmel-samd/common-hal/busio/SPI.c#L216-L222 This protects the SPI from similar issues when used by SPIDevice which also calls configure before each transaction.

Actually STM port does check the SPI configuration change before altering SPI state.
https://github.com/adafruit/circuitpython/blob/main/ports/stm/common-hal/busio/SPI.c#L279
The thing is that common_hal_busio_spi_construct sets default SPI frequency (baudrate) that differs from the SPI flash settings. And the first call of spi_flash_read_data made out from check_fs causes the issue.

I have added common_hal_busio_spi_configure back to spi_flash_read_data and spi_flash_write_data. But they are called before chip select signal is activated now. Moreover I have added common_hal_busio_spi_configure to spi_flash_init_device.

[tannewt]

The thing is that common_hal_busio_spi_construct sets default SPI frequency (baudrate) that differs from the SPI flash settings.

We're careful with the starting baudrate due to picky SD cards. That's where the 250k (IIRC) comes from.

I have added common_hal_busio_spi_configure back to spi_flash_read_data and spi_flash_write_data. But they are called before chip select signal is activated now. Moreover I have added common_hal_busio_spi_configure to spi_flash_init_device.

That sounds perfect!

[tannewt]

Looks good to me! Thank you!