[U-Boot] [PATCH v3 1/3] cmd_sf: add 'release' command

[Valentin Longchamp <valentin.longchamp@keymile.com>]

Hi Simon,

On 20/11/2015 18:19, Simon Glass wrote:
> Hi,
> 
> On 20 November 2015 at 03:13, Valentin Longchamp
> <valentin.longchamp@keymile.com> wrote:
>> On 19/11/2015 17:57, Jagan Teki wrote:
>>> On 13 November 2015 at 18:55, Valentin Longchamp
>>> <valentin.longchamp@keymile.com> wrote:
>>>> The release command is the pendant of the probe command. This command
>>>> allows to call spi_flash_free from the command line. This may be
>>>> necessary for some boards where sf probe does change the state of the
>>>> hardware (like with some pin multiplexing changes for instance).
>>>
>>> So you want to change the state of pin multiplexing on your board with
>>> connected slave devices example: spi nor flash is it? what exactly the
>>> need of releasing? why can't we use pin multiplexing changes like
>>> selecting or deselecting particular lines through driver or from board
>>> files itself.
>>
>> That's our use case yes. Let me explain you it again in detail. Some of the
>> signals used to access the NAND Flash and the SPI NOR are shared. At reset, they
>> are available for the SPI NOR, since u-boot is in there and the CPU then
>> accesses it.
>>
>> In an usual boot sequence, the SPI NOR is accessed first (copying u-boot to the
>> RAM, reading out the environment) and so the pins are configured in hardware at
>> boot time for accessing the SPI NOR. After that, they are configured to access
>> the NAND where the kernel and filesystem are stored to boot Linux thanks to
>> env_relocate_spec() calling spi_flash_free() on exit in conjunction with [1]
>>
>> Now in the case where the boot sequence is interrupted and some accesses are
>> done to the SPI NOR, the pins are changed again to SPI NOR to perform these
>> accesses. But we have to make sure that the pins are configured back to NAND by
>> calling spi_flash_free() after these accesses and that's why I introduced the
>> release() function.
>>
>> In our case, there are 2 types of such accesses:
>> - environment variables write: this is the first patch of the series. It simply
>> adds calls to spi_flash_free() at function exit no only in env_relocate_spec()
>> but also in saveenv() so that the behavior here is coherent for the whole env_sf
>> file (spi_flash_probe() at function start, spi_flash_free() at function exit).
>> - updating u-boot: this is solved for us with the last 2 patches of the series.
>> The first one just adds a sf release command that does the opposite/cleanup to
>> sf probe and the second patch just calls this command in our scripts where
>> u-boot is updated/the SPI NOR is written.
>>
>> We are *indeed* using pin multiplexing changes, in our case, they are
>> implemented in the spi controller driver: drivers/spi/kirkwood_spi.c. To be very
>> specific, in our case this sf release command allows to explicitely call
>> spi_flash_free() which calls spi_free_slave(), which in our case
>> (kirkwood_spi.c) sets the pins back to their previous configuration.
> 
> Does your board use driver model from SPI and SPI flash? If not I
> think that should be the first step.
> 

No we don't. Could you please elaborate on how this would cover this use case
and should be the first step ?

I am open to other ways to cover this use case of ours, especially since this
was done more than 2 years ago and u-boot has changed since then. However I
don't see the direct link between the driver model and how it would allow to
make sure spi_flash_free() is called in our u-boot env scripts.

Valentin