linux-kernel - Re: [RFC PATCH 2/2] mfd: rohm-bd71828: Add power off functionality

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Mon, 25 Mar 2024 14:44:43 +0200
From: Matti Vaittinen <mazziesaccount@...il.com>
To: Andreas Kemnade <andreas@...nade.info>
Cc: lee@...nel.org, robh+dt@...nel.org, krzysztof.kozlowski+dt@...aro.org,
 conor+dt@...nel.org, devicetree@...r.kernel.org,
 linux-kernel@...r.kernel.org, imx@...ts.linux.dev
Subject: Re: [RFC PATCH 2/2] mfd: rohm-bd71828: Add power off functionality

Hi Andreas,

On 3/25/24 14:16, Andreas Kemnade wrote:
> On Mon, 25 Mar 2024 13:31:15 +0200
> Matti Vaittinen <mazziesaccount@...il.com> wrote:
> 
>> On 3/24/24 22:12, Andreas Kemnade wrote:
>>> Since the chip can power off the system, add the corresponding
>>> functionality.
>>> Based on https://github.com/kobolabs/Kobo-Reader/raw/master/hw/imx6sll-clara2e/kernel.tar.bz2
>>> No information source about the magic numbers found.
>>
>> Oh, interesting repository :) Thanks for linking to it! I didn't know
>> someone had reworked this driver...
>>
> which btw: contains this interesting snippet (output from fdtdump)
>    bd71828-i2c@4b {
>                      reg = <0x0000004b>;
>                      compatible = "rohm,bd71828";
>                      gpio_int = <0x00000008 0x00000013 0x00000001>;
>                      gpio_wdogb = <0x00000039 0x00000018 0x00000001>;
>                      #address-cells = <0x00000001>;
>                      #size-cells = <0x00000000>;
>                      pmic@4b {
>                          compatible = "rohm,bd71828";
>               	        regulators {
>                          	BUCK1 {
>                              		regulator-name = "buck1";
> 
> 
> and to make it work since basically no regulators are registered
> instead just some regmap_write()s are done to configure something
> in probe(). It is a pitfall to think that the information below pmic@4b
> is used, especially since it is not that obvious in the source.

Just to ensure there will be no misunderstanding - I have not authored 
the modifications seen in "Kobo-Reader" repository. Upstream driver does 
register the regulators - and it does not use the oddly named gpio_int 
or the gpio_wdogb:

https://elixir.bootlin.com/linux/v6.9-rc1/source/drivers/regulator/bd71828-regulator.c#L750

>> I have access to the data-sheets so I also have some pieces of
>> information. I hope I can clarify part of the puzzle. Unfortunately I
>> have no information about the magic delays. I guess I could try asking
>> though.
>>
>> Oh, it seems to me this handler is only working on BD71828, not on
>> BD71815. So, it should be tied to the ROHM_CHIP_TYPE_BD71828.
>>
>>> Signed-off-by: Andreas Kemnade <andreas@...nade.info>
>>> ---
>>>    drivers/mfd/rohm-bd71828.c | 31 ++++++++++++++++++++++++++++++-
>>>    1 file changed, 30 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/mfd/rohm-bd71828.c b/drivers/mfd/rohm-bd71828.c
>>> index 594718f7e8e1..5a55aa3620d0 100644
>>> --- a/drivers/mfd/rohm-bd71828.c
>>> +++ b/drivers/mfd/rohm-bd71828.c
>>> @@ -464,6 +464,24 @@ static int set_clk_mode(struct device *dev, struct regmap *regmap,
>>>    				  OUT32K_MODE_CMOS);
>>>    }
>>>    
>>> +static struct i2c_client *bd71828_dev;
>>
>> I'm not sure why to store pointer to the device and not a pointer to the
>> regmap?
>>
>>> +static void bd71828_power_off(void)
>>> +{
>>> +	i2c_smbus_write_byte_data(bd71828_dev, 0x03, 0xff);
>>
>> 0x03 is a "reset reason" - register. Spec I have states that the
>> register should clear when a reset occurs - but it also says the bits
>> are "write '1' to clear". So, for some reason(?), this clears the
>> previous reset reason.
> 
> well, so just check in bootloader what the reset reason is and check if
> there is anything odd.

I understand why bootloader would check this register - but I don't 
understand why the driver would clear it as it should be cleared by 
reset (unless I missread the spec).

>> I am unsure why i2c_smbus_write_byte_data() and
>> not regmap()?
>>
> regmap involves mutex_lock() and we are not allowed to sleep here.

Right. It makes perfect sense now. Thanks! (Do you think it's worth a 
comment?)

>>> +	mdelay(500);
>>> +	i2c_smbus_write_byte_data(bd71828_dev, BD71828_REG_INT_DCIN2, 0x02);
>>
>> This clears the DCIN monitoring status bit from the IRQ status register.
>> I don't understand the purpose though.
>>
> so maybe something to prevent power on by just plugging a usb cable? Will
> experiment a bit with it.

I still think this is odd because, if I read it right, this is an IRQ 
status and not a mask register. Well, wouldn't be a first undocumented 
"feature" I see...

>>> +	mdelay(500);
>>> +	while (true) {
>>> +		i2c_smbus_write_byte_data(bd71828_dev, BD71828_REG_PS_CTRL_1, 0x02);
>>
>> This write to PS_CTRL_1 initiates a state transition. 0x2 equals to HBNT
>> state. Eg, in usual cases this should be a start of the power-off sequence.
>>
>>> +		mdelay(500);
>>> +	}
>>> +}
>>
>> If you have the hardware to test this on, then it'd be great to see if
>> clearing the reset reason and IRQ status could be dropped. I can't
>> immediately think of a reason for those.
>>
> I will to so. That will also remove the need for all those delays.

Thanks :) Very much appreciated!

Yours,
	-- Matti

-- 
Matti Vaittinen
Linux kernel developer at ROHM Semiconductors
Oulu Finland

~~ When things go utterly wrong vim users can always type :help! ~~