[PATCH v17 03/12] power: reset: reboot-mode: Add support for 64 bit magic

[Shivendra Pratap]

On 11/11/2025 9:55 PM, Bjorn Andersson wrote:
> On Tue, Nov 11, 2025 at 08:20:43PM +0530, Shivendra Pratap wrote:
>>
>>
>> On 11/11/2025 12:03 AM, Bjorn Andersson wrote:
>>> On Mon, Nov 10, 2025 at 11:22:40PM +0530, Shivendra Pratap wrote:
>>>>
>>>>
>>>> On 11/10/2025 10:00 PM, Bjorn Andersson wrote:
>>>>> On Mon, Nov 10, 2025 at 07:15:29PM +0530, Mukesh Ojha wrote:
>>>>>> On Sun, Nov 09, 2025 at 08:07:16PM +0530, Shivendra Pratap wrote:
>>>>>>> Current reboot-mode supports a single 32-bit argument for any
>>>>>>> supported mode. Some reboot-mode based drivers may require
>>>>>>> passing two independent 32-bit arguments during a reboot
>>>>>>> sequence, for uses-cases, where a mode requires an additional
>>>>>>> argument. Such drivers may not be able to use the reboot-mode
>>>>>>> driver. For example, ARM PSCI vendor-specific resets, need two
>>>>>>> arguments for its operation – reset_type and cookie, to complete
>>>>>>> the reset operation. If a driver wants to implement this
>>>>>>> firmware-based reset, it cannot use reboot-mode framework.
>>>>>>>
>>>>>>> Introduce 64-bit magic values in reboot-mode driver to
>>>>>>> accommodate dual 32-bit arguments when specified via device tree.
>>>>>>> In cases, where no second argument is passed from device tree,
>>>>>>> keep the upper 32-bit of magic un-changed(0) to maintain backward
>>>>>>> compatibility.
>>>>>>>
>>>>>>> Update the current drivers using reboot-mode for a 64-bit magic
>>>>>>> value.
>>>>
>>>> [SNIP..]
>>>>
>>>>>>> +	if (magic > U32_MAX)
>>>>>>> +		return -EINVAL;
>>>>>>> +
>>>>>>> +	magic_32 = magic;
>>>>>>> +
>>>>>>>  	syscon_rbm = container_of(reboot, struct syscon_reboot_mode, reboot);
>>>>>>>  
>>>>>>>  	ret = regmap_update_bits(syscon_rbm->map, syscon_rbm->offset,
>>>>>>> -				 syscon_rbm->mask, magic);
>>>>>>> +				 syscon_rbm->mask, magic_32);
>>>>>
>>>>> As above, if we're no longer silently discarding bits, I think we should
>>>>> compare the magic against syscon_rbm->mask.
>>>>>
>>>>> No need for a local variable, just type cast after checking the validity.
>>>>
>>>> Trying to summarize below why we added these check-
>>>>
>>>> the patch in v11 used typecasting and did not have any of these checks(link below):
>>>> https://lore.kernel.org/all/20250717-arm-psci-system_reset2-vendor-reboots-v11-2-df3e2b2183c3@oss.qualcomm.com/
>>>>
>>>> As per below upstream review, type cast was removed and bound checks were added all-over patchset:
>>>> "As a general rule of thumb, code with casts is poor quality code. Try
>>>> to write the code without casts." - 
>>>> https://lore.kernel.org/all/8d4a42b6-657f-4c30-8e25-4213d8d53a89@lunn.ch/
>>>>
>>>> We can revert to the typecast way. Please suggest.
>>>>
>>>
>>> Okay, I'm okay with Andrew's original request, stick to that for the
>>> nvmem case. Although I don't fancy the name "magic_32", and would prefer
>>> that you just call it "value" or something.
>>
>> sure will make it proper wherever applicable. 
>>
>>>
>>>
>>> For pon and syscon however, I'm wondering why you have ignored Andrew's
>>> other request from that same email:
>>>
>>> """
>>> You might be able to go further, and validate that magic actually fits
>>> into the field when you consider the << pon->reason_shift.
>>> """
>>>
>>> Writing "if (magic > U32_MAX)" in a snippet of code where magic isn't
>>> allowed to be more than either 32 or 64 is misleading.
>>>
>>> For syscon, it's true that the parameter is an unsigned long, but the
>>> actual limit better be based on syscon_rbm->mask.
>>
>> May be i was not able to interpret it correctly. Basically tried to
>> make sure that magic that now coming in a "u64 magic" should be passed
>> ahead only it its a 32 bit value.
>>
> 
> That is the correct interpretation of the original ask. But what I'm
> saying is that if you write:
> 
> if (magic > U32_MAX)
> 
> then that means "check that magic isn't larger than 32 bits". So the
> reader will see that and understand that magic can only be 32 bits.
> 
> But the actual PON magic value is 5 or 6 bits, not 32 - so you should
> check that the value fits in 5 or 6 bits.

sure. thanks.

> 
>> So should i get rid of the checks done here for syscon and pon?
>>
> 
> Continuing to silently ignoring other bits would be one option, but
> you've been asked to sanity check the values, so please do that. You
> have the code, just compare with the correct value.

ok. got it. thanks.

thanks,
Shivendra