lkml.org 
[lkml]   [2023]   [Oct]   [16]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: [PATCH v4 2/2] mm: Init page count in reserve_bootmem_region when MEMINIT_EARLY
From

On 2023/10/16 16:16, David Hildenbrand wrote:
> On 16.10.23 10:10, Yajun Deng wrote:
>>
>> On 2023/10/16 14:33, Mike Rapoport wrote:
>>> On Fri, Oct 13, 2023 at 05:29:19PM +0800, Yajun Deng wrote:
>>>> On 2023/10/13 16:48, Mike Rapoport wrote:
>>>>> On Thu, Oct 12, 2023 at 05:53:22PM +0800, Yajun Deng wrote:
>>>>>> On 2023/10/12 17:23, David Hildenbrand wrote:
>>>>>>> On 10.10.23 04:31, Yajun Deng wrote:
>>>>>>>> On 2023/10/8 16:57, Yajun Deng wrote:
>>>>>>>>>> That looks wrong. if the page count would by pure luck be 0
>>>>>>>>>> already for hotplugged memory, you wouldn't clear the reserved
>>>>>>>>>> flag.
>>>>>>>>>>
>>>>>>>>>> These changes make me a bit nervous.
>>>>>>>>> Is 'if (page_count(page) || PageReserved(page))' be safer? Or
>>>>>>>>> do I
>>>>>>>>> need to do something else?
>>>>>>>>>
>>>>>>>> How about the following if statement? But it needs to add more
>>>>>>>> patch
>>>>>>>> like v1 ([PATCH 2/4] mm: Introduce MEMINIT_LATE context).
>>>>>>>>
>>>>>>>> It'll be safer, but more complex. Please comment...
>>>>>>>>
>>>>>>>>        if (context != MEMINIT_EARLY || (page_count(page) ||
>>>>>>>> PageReserved(page)) {
>>>>>>>>
>>>>>>> Ideally we could make initialization only depend on the context,
>>>>>>> and not
>>>>>>> check for count or the reserved flag.
>>>>>>>
>>>>>> This link is v1,
>>>>>> https://lore.kernel.org/all/20230922070923.355656-1-yajun.deng@linux.dev/
>>>>>>
>>>>>>
>>>>>> If we could make initialization only depend on the context, I'll
>>>>>> modify it
>>>>>> based on v1.
>>>>> Although ~20% improvement looks impressive, this is only
>>>>> optimization of a
>>>>> fraction of the boot time, and realistically, how much 56 msec
>>>>> saves from
>>>>> the total boot time when you boot a machine with 190G of RAM?
>>>> There are a lot of factors that can affect the total boot time. 56
>>>> msec
>>>> saves may be insignificant.
>>>>
>>>> But if we look at the boot log, we'll see there's a significant
>>>> time jump.
>>>>
>>>> before:
>>>>
>>>> [    0.250334] ACPI: PM-Timer IO Port: 0x508
>>>> [    0.618994] Memory: 173413056K/199884452K available (18440K
>>>> kernel code,
>>>>
>>>> after:
>>>>
>>>> [    0.260229] software IO TLB: area num 32.
>>>> [    0.563497] Memory: 173413056K/199884452K available (18440K
>>>> kernel code,
>>>> Memory:
>>>> Memory initialization is time consuming in the boot log.
>>> You just confirmed that 56 msec is insignificant and then you send
>>> again
>>> the improvement of ~60 msec in memory initialization.
>>>
>>> What does this improvement gain in percentage of total boot time?
>>
>>
>> before:
>>
>> [   10.692708] Run /init as init process
>>
>>
>> after:
>>
>> [   10.666290] Run /init as init process
>>
>>
>> About 0.25%. The total boot time is variable, depending on how many
>> drivers need to be initialized.
>>
>>
>>>>> I still think the improvement does not justify the churn, added
>>>>> complexity
>>>>> and special casing of different code paths of initialization of
>>>>> struct pages.
>>>>
>>>> Because there is a loop, if the order is MAX_ORDER, the loop will
>>>> run 1024
>>>> times. The following 'if' would be safer:
>>>>
>>>> 'if (context != MEMINIT_EARLY || (page_count(page) || >>
>>>> PageReserved(page))
>>>> {'
>>> No, it will not.
>>>
>>> As the matter of fact any condition here won't be 'safer' because it
>>> makes
>>> the code more complex and less maintainable.
>>> Any future change in __free_pages_core() or one of it's callers will
>>> have
>>> to reason what will happen with that condition after the change.
>>
>>
>> To avoid introducing MEMINIT_LATE context and make code simpler. This
>> might be a better option.
>>
>> if (page_count(page) || PageReserved(page))
>
> I'll have to side with Mike here; this change might not be worth it.
>

Okay, I got it. Thanks!

\
 
 \ /
  Last update: 2023-10-16 10:34    [W:0.130 / U:0.472 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site