Re: [PATCH v2] ACPI: Replace msleep() with usleep_range() in acpi_os_sleep().

[Pierre Gondois <pierre.gondois@arm.com>]

On 11/20/24 13:06, Dietmar Eggemann wrote:
> On 20/11/2024 10:01, Pierre Gondois wrote:
>>
>>
>> On 11/18/24 13:02, Rafael J. Wysocki wrote:
>>> Hi Hans,
>>>
>>> On Mon, Nov 18, 2024 at 12:38 PM Hans de Goede <hdegoede@redhat.com>
>>> wrote:
>>>>
>>>> Hi Rafael, Len,
>>>>
>>>> On 18-Nov-24 12:03 PM, Rafael J. Wysocki wrote:
>>>>> On Sat, Nov 16, 2024 at 12:11 AM Len Brown <lenb@kernel.org> wrote:
> 
> [...]
> 
>> FWIW, testing the above version on an Arm Juno platform by executing
>> the following method:
>>
>> Method (SLEE, 1, Serialized)  {
>>    Sleep(Arg0)
>> }
>>
>> _wo: without patch
>> _w: with patch
>> - Values in ns.
>> - Requesting to sleep X ms
>> - Tested over 10 iterations
>> - HZ=250
>> +------+------------+----------+------------+---------+-----------+
>> |   ms |    mean_wo |   std_wo |    mean_w  |  std_w  |     ratio |
>> +------+------------+----------+------------+---------+-----------+
>> |    1 |    8087797 |  2079703 |    1313920 |   55066 | -83.75429 |
>> |    2 |    7942471 |  2201985 |    2416064 |  111604 | -69.58044 |
>> |    3 |    8373704 |   144274 |    3632537 |  111037 | -56.61970 |
>> |    4 |    7946013 |  2214330 |    4606028 |  255838 | -42.03346 |
>> |    5 |   11418920 |  1673914 |    5955548 |  131862 | -47.84490 |
>> |    6 |   11427042 |  1677519 |    7045713 |  211439 | -38.34176 |
>> |    7 |   12301242 |   221580 |    8174633 |  330050 | -33.54628 |
>> |    8 |   11411606 |  1672182 |    9191048 |  431767 | -19.45877 |
>> |    9 |   16722304 |  1288625 |   10517284 |  103274 | -37.10625 |
>> |   10 |   16746542 |  1280385 |   11564426 |  417218 | -30.94439 |
>> |   20 |   24294957 |    70703 |   22756497 |  673936 |  -6.33243 |
>> |   30 |   36284782 |    74340 |   34131455 |  391473 |  -5.93452 |
>> |   40 |   44703162 |  1199709 |   45407108 |  289715 |   1.57471 |
>> |   50 |   56311282 |   281418 |   56098040 |  607739 |  -0.37868 |
>> |   60 |   64225811 |   247587 |   64302246 |  132059 |   0.11901 |
>> |   70 |   76299457 |    99853 |   76282497 |   83910 |  -0.02223 |
>> |  100 |  104214393 |    38642 |  104212524 |  244424 |  -0.00179 |
>> | 1000 | 1016131215 |   245725 | 1017051744 | 2748280 |   0.09059 |
>> | 2000 | 2007711297 |  1325094 | 2007628922 | 1421807 |  -0.00410 |
>> +------+------------+----------+------------+---------+-----------+
>> - With the patch, the min sleep duration is never below the requested
>>    sleep duration
>>
>> So indeed the penalty of using msleep is big for small sleep durations.
> 
> Just to make sure, this is with Rafael's proposal, using msleep() for
> values >= 48ms = (12 * 1000/250)ms and usleep_range() otherwise?
> 

Yes exact