stm32h5/qspi, drivers/mtd/gd25: add QSPIMEM_QUADDATA flag and GD25 QSPI support

[sammytranGeo]

Summary

include/nuttx/spi/qspi.h currently has QSPIMEM_QUADIO for 1-4-4 transfers (address and data both on quad lines) but no way to express 1-1-4 (single-line address, quad-width data only). Several NOR flash devices use this mode for page program commands.

This adds QSPIMEM_QUADDATA (bit 1) and QSPIMEM_ISQUADDATA() to fill that gap, and extends the STM32H5 QSPI driver to set CCR_DMODE_QUAD when either QUADIO or QUADDATA is set. The address-mode selection block is unchanged, so QUADDATA keeps addrmode at single-line.

The GD25 MTD driver is updated as the first user, gaining a CONFIG_GD25_QSPI build option that switches it from SPI to QuadSPI. When enabled, gd25_initialize() accepts a qspi_dev_s pointer and a bool unprotect. Both 3- and 4-byte address modes are supported, and the existing SPI path is fully preserved when CONFIG_GD25_QSPI=n.

Impact

• qspi.h: additive change only — new flag and macro, no existing behaviour changed.
• stm32h5/stm32_qspi.c: the else if (QSPIMEM_ISQUADIO(...)) branch gains an || QSPIMEM_ISQUADDATA(...) condition. No impact on existing QUADIO users.
• drivers/mtd/gd25.c: new CONFIG_GD25_QSPI Kconfig option, off by default. Existing SPI users are unaffected.

Testing

Host: Linux x86_64, arm-none-eabi-gcc 13.2
Board: Custom STM32H563ZI board with GD25Q256 NOR flash on OCTOSPI1
Config: Custom board NSH config with CONFIG_STM32_QSPI1=y, CONFIG_MTD_GD25=y, CONFIG_GD25_QSPI=y, CONFIG_GD25_QSPIFREQUENCY=40000000, CONFIG_FS_LITTLEFS=y

LittleFS was mounted on /dev/mtd0 and a file written and read back to exercise both transfer modes end-to-end:

nsh> mount -t littlefs -o forceformat /dev/mtd0 /mnt
nsh> echo "hello qspi" > /mnt/test.txt
nsh> cat /mnt/test.txt
hello qspi

[github-actions]

MemBrowse Memory Report

No memory changes detected for:

• arduino-mega2560
• esp32-devkitc
• hifive1-revb
• mirtoo
• qemu-armv8a
• qemu-intel64
• rx65n-rsk2mb
• s698pm-dkit
• stm32-nucleo-f103rb

[acassis]

@sammytranGeo please fix the CI error (long line and wrong block comment found)

You can use the tools/checkpatch.sh to find coding style issues in the source code before creating the commit