Re: [PATCH v14 8/8] soc: amd: Add support for AMD Pensando SoC Controller

[Brad Larson <blarson@amd.com>]

Hi Arnd,

> On Mon, May 15, 2023, at 20:16, Brad Larson wrote:
>> The Pensando SoC controller is a SPI connected companion device
>> that is present in all Pensando SoC board designs.  The essential
>> board management registers are accessed on chip select 0 with
>> board mgmt IO support accessed using additional chip selects.
>>
>> Signed-off-by: Brad Larson <blarson@amd.com>
>
> Hi Brad,
>
> I'm sorry I wasn't paying enough attention to this driver as the
> past 13 revisions went by.
>
No worries, bit of a saga.  See explanation below.

>> v10 changes:
>> - Different driver implementation specific to this Pensando controller device.
>> - Moved to soc/amd directory under new name based on guidance.  This driver is
>>   of no use to any design other than all Pensando SoC based cards.
>> - Removed use of builtin_driver, can be built as a module.
>
> it looks like this was a fundamental change that I failed to see.

See explanation below.

>> +# SPDX-License-Identifier: GPL-2.0-only
>> +menu "AMD Pensando SoC drivers"
>> +
>> +config AMD_PENSANDO_CTRL
>> +	tristate "AMD Pensando SoC Controller"
>> +	depends on SPI_MASTER=y
>> +	depends on (ARCH_PENSANDO && OF) || COMPILE_TEST
>> +	default ARCH_PENSANDO
>> +	select REGMAP_SPI
>> +	select MFD_SYSCON
>> +	help
>> +	  Enables AMD Pensando SoC controller device support.  This is a SPI
>> +	  attached companion device in all Pensando SoC board designs which
>> +	  provides essential board control/status registers and management IO
>> +	  support.
>
> So generally speaking, I don't want custom user interfaces in
> drivers/soc. It looks like this one has internal interfaces for
> a reset controller and the regmap, so those parts are fine, but
> I'm confused about the purpose of the ioctl interface:
>
>> +static long
>> +penctrl_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
>> +{
>
>> +	if (num_msgs > 1) {
>> +		msg++;
>> +		if (msg->len > PENCTRL_MAX_MSG_LEN) {
>> +			ret = -EINVAL;
>> +			goto out_unlock;
>> +		}
>> +		t[1].rx_buf = rx_buf;
>> +		t[1].len = msg->len;
>> +	}
>> +	spi_message_init_with_transfers(&m, t, num_msgs);
>
> This seems to be just a passthrough of user space messages, which
> is what the spidev driver already provides, but using a different
> ioctl interface. I don't really want any user-level interfaces
> in drivers/soc as a rule, but having one that duplicates existing
> functionality seems particularly wrong.
>
> Can you explain what the purpose is? Is this about serializing
> access between the in-kernel reset control and the user-side
> access?
>
> Also, can you explain why this needs a low-lever user interface
> in the first place, rather than something that can be expressed
> using high-level abstractions as you already do with the reset
> control?
>
> All of the above should be part of the changelog text to get a
> driver like this merged. I don't think we can get a quick
> solution here though, so maybe you can start by removing the
> ioctl side and having the rest of the driver in drivers/reset?

In the original patchset I added a pensando compatible to spidev and that
was nacked in review and reusing some random compatible that made it into 
spidev was just wrong.  Further it was recommended this should be a system 
specific driver and don't rely on a debug driver like spidev.  I changed 
over to a platform specific driver and at that time I also needed to include 
a reset controller (emmc reset only).  I put these in drivers/mfd and 
drivers/reset.  Review of the device tree for this approach went back and 
forth to _not_ have four child nodes on the spi device each with the same 
compatible. Decision was to squash the child nodes into the parent and put 
the reset-controller there also.  One driver and since its pensando
specific its currently in drivers/soc/amd.

There are five different user processes and some utilities that access the 
functionality in the cpld/fpga.  You're correct, its passing messages that 
are specific to the IP accessed via chip-select.  No Elba system will boot 
without this driver providing ioctl access.

Regards,
Brad