Re: [PATCH 1/2] Revert "mtd: atmel_nand: Support variable RB_EDGE interrupts"

[Romain Izard]

Hi Boris,

2016-05-10 10:55 GMT+02:00 Boris Brezillon <boris.brezillon@xxxxxxxxxxxxxxxxxx>:
>
> Romain, I thought you had a real use case on sama5d4 where this patch
> was needed to make the whole thing work. Not sure why you submitted
> this patch if you couldn't test it on a real board.

My target CPU is SAMA5D2, but I chose the SAMA5D4 for the compatible
string as it was the earliest design whose datasheet mentioned the
RB_EDGE3 bit.

I wrote the SAMA5D2 support in advance, as I was waiting for my board to
be ready, basing myself on what was found in the datasheet.  The changes
included both the variable RB_EDGE support, and the PMECC register
layout changes, which were a prerequisite to use the SAMA5D2 NAND
controller.

I tested the code against regressions on sama5d3xek, and Wenyou reported
that he tested it on Atmel's SAMA5D2 PTC, which was the only existing
board at that time with both a SAMA5D2 SoC and a NAND chip.
Unfortunately, as the bug is only seen when writing on the flash, and it
only affects the speed of the device, he did not notice it.

I sent the patches early because I expected the submission process to
be long, and I wanted to be able to freeze the kernel version to be used
on my board as soon as possible.

Best regards,
-- 
Romain Izard