On Tue, 8 Jun 2010 15:20:57 -0400 Mike Frysinger vapier@gentoo.org wrote:
speaking historically, the dataflash code absolutely has its place. it existed long before the SPI flash framework. but i'm looking forward only now.
Yes, of course, understood.
However, the Dataflash aren't normal SPI flash, they don't have the same opcodes. For example, drivers/mtd/spi/spi_flash.c assumes that it can probe the "ID code" of the SPI flash by sending the CMD_READ_ID (0x9F) command (in spi_flash_probe()). This works for SPI flashes, but not for Dataflashes. The identification of Dataflashes takes place with command GET_STATUS (0xD7) in drivers/mtd/at45.c, which has a different return value than the 0x9F command of SPI flashes. Am I missing something ?
In terms of code infrastructure/organization, how do you suggest to handle this ?
i think the first step would be to convert the boards we can and leave a #warning for the rest that the dataflash code is being killed off. then after some time, if no one has fixed the remainders, we do our best to convert them. -mike
Ok.
Thomas