linux-kernel - Re: [PATCH 02/10] hwmon: (amc6821) Make reading and writing fan speed limits consistent

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, 1 Jul 2024 10:21:35 -0700
From: Guenter Roeck <linux@...ck-us.net>
To: Quentin Schulz <quentin.schulz@...rry.de>, linux-hwmon@...r.kernel.org
Cc: linux-kernel@...r.kernel.org, Farouk Bouabid <farouk.bouabid@...rry.de>
Subject: Re: [PATCH 02/10] hwmon: (amc6821) Make reading and writing fan speed
 limits consistent

On 7/1/24 09:13, Quentin Schulz wrote:
> Hi Guenter,
> 
> On 7/1/24 4:37 PM, Guenter Roeck wrote:
>> On 7/1/24 07:11, Guenter Roeck wrote:
>>> On 7/1/24 04:05, Quentin Schulz wrote:
>>>> Hi Guenter,
>>>>
>>>> On 6/28/24 5:13 PM, Guenter Roeck wrote:
>>>>> The default value of the maximum fan speed limit register is 0,
>>>>> essentially translating to an unlimited fan speed. When reading
>>>>> the limit, a value of 0 is reported in this case. However, writing
>>>>> a value of 0 results in writing a value of 0xffff into the register,
>>>>> which is inconsistent.
>>>>>  > Signed-off-by: Guenter Roeck <linux@...ck-us.net>
>>>>> ---
>>>>>   drivers/hwmon/amc6821.c | 6 +++---
>>>>>   1 file changed, 3 insertions(+), 3 deletions(-)
>>>>>
>>>>> diff --git a/drivers/hwmon/amc6821.c b/drivers/hwmon/amc6821.c
>>>>> index 3c614a0bd192..e37257ae1a6b 100644
>>>>> --- a/drivers/hwmon/amc6821.c
>>>>> +++ b/drivers/hwmon/amc6821.c
>>>>> @@ -601,7 +601,7 @@ static ssize_t fan_show(struct device *dev, struct device_attribute *devattr,
>>>>>       struct amc6821_data *data = amc6821_update_device(dev);
>>>>>       int ix = to_sensor_dev_attr(devattr)->index;
>>>>>       if (0 == data->fan[ix])
>>>>> -        return sprintf(buf, "0");
>>>>> +        return sprintf(buf, "6000000");
>>>>>       return sprintf(buf, "%d\n", (int)(6000000 / data->fan[ix]));
>>>>>   }
>>>>> @@ -625,10 +625,10 @@ static ssize_t fan_store(struct device *dev, struct device_attribute *attr,
>>>>>       int ret = kstrtol(buf, 10, &val);
>>>>>       if (ret)
>>>>>           return ret;
>>>>> -    val = 1 > val ? 0xFFFF : 6000000/val;
>>>>> +    val = val < 1 ? 0xFFFF : 6000000 / val;
>>>>>       mutex_lock(&data->update_lock);
>>>>> -    data->fan[ix] = (u16) clamp_val(val, 1, 0xFFFF);
>>>>> +    data->fan[ix] = (u16)clamp_val(val, 0, 0xFFFF);
>>>>
>>>> This is an unrelated change I believe and I would therefore have this in its own commit with proper documentation in the commit log. Indeed:
>>>>
>>>> 1- Change in fan_show handles the default 0x0 register value (which can only currently be achieved via the default value of the registers)
>>>> 2- Allow (re-)setting unlimited fan speed by allowing the user to pass 6000001+ instead of clamping it to 6000000 RPM.
>>>>
>>>
>>> Both changes are related.
>>>
>>> The whole point of this commit is to report and permit consistent values when
>>> the register value is 0. But you do have a point - reading it after my changes
>>> returns 6000000, but writing the same value sets the register to 1. So I think
>>> the proper change would be to display 6000001 as speed if the register value is
>>> 0, and provide a more detailed explanation. Would that address your concerns ?
>>>
>>
>> Ah, never  mind, I'll do it differently:
>>
>> - If the register value is 0, keep reporting 0.
> 
> Or...... maybe UINT_MAX?
> 

Problem with that is that disconnected fans would report that value as fan speed.
Traditionally drivers report a fan speed of 0 in that case.

On the other side I agree that reporting "0" as "maximum fan speed" doesn't
make much sense either because the real limit _is_ unlimited. But reporting
4294967295 in that case isn't really any better.

>> - If the value written is 0, write 0, otherwise limit the range to 1..6000000
>>    and write clamp_val(6000000 / val, 1, 0xffff)
>>
> 
> Mmmm... I'm a bit worried about the implication of writing 0 in TACH-Low-Limit, what is actually going to happen in that scenario? I assume **every** possible RPM returned by TACH-DATA will be deemed invalid/below the limit then? Reading `Fan Spin-Up` section, if FSPD bit from register 0x20 (which we don't write to yet I think?) is set to 0, a spin-up is started whenever the fan is detected to be running at too low speed. And we would also be getting an interrupt for that too-low event.
> 
> Basically, wondering if we shouldn't gate the writing of 0 to only the MAX setting?
> 

Hmm, good point, and make sense.

>> This minimizes user visibility of the changes, and also ensures that
>> the reported fan speed is 0 if the register value is 0 when reading the fan
>> speed.
>>
> 
> But didn't you say this means the fan is running at unknown 60 000 000+ RPMs? Do we really want to return 0 even if the fan is actually running? In which case max < current (possibly) but with no event happening (which I would expect, reading the datasheet).
> 

Did I say that ? If so, I must have meant something different. The register counts the
pulse period, so, yes, it would be 0 if rpm is above 6,000,000. But that is really not
realistic. In practice I don't know what the controller reports in the register if no
fan is connected - that would require real hardware which obviously I don't have.

Overall I think I'll stick with the minimum, at least for now: Permit writing 0
into the high limit register only, and otherwise keep the currently permitted ranges.

Thanks,
Guenter