It's not completely obvious why we have to shuffle the complete zone, as
some sort of shuffling is already performed when onlining pages via
__free_one_page(), placing MAX_ORDER-1 pages either to the head or the tail
of the freelist. Let's document why we have to shuffle the complete zone
when exposing larger, contiguous physical memory areas to the buddy.
Cc: Andrew Morton <[email protected]>
Cc: Alexander Duyck <[email protected]>
Cc: Dan Williams <[email protected]>
Cc: Michal Hocko <[email protected]>
Signed-off-by: David Hildenbrand <[email protected]>
---
mm/memory_hotplug.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index 9b34e03e730a4..a0d81d404823d 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -822,6 +822,14 @@ int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
zone->zone_pgdat->node_present_pages += onlined_pages;
pgdat_resize_unlock(zone->zone_pgdat, &flags);
+ /*
+ * When exposing larger, physically contiguous memory areas to the
+ * buddy, shuffling in the buddy (when freeing onlined pages, putting
+ * them either to the head or the tail of the freelist) is only helpful
+ * for mainining the shuffle, but not for creating the initial shuffle.
+ * Shuffle the whole zone to make sure the just onlined pages are
+ * properly distributed across the whole freelist.
+ */
shuffle_zone(zone);
node_states_set_node(nid, &arg);
--
2.26.2
On Fri, Jun 19, 2020 at 6:00 AM David Hildenbrand <[email protected]> wrote:
>
> It's not completely obvious why we have to shuffle the complete zone, as
> some sort of shuffling is already performed when onlining pages via
> __free_one_page(), placing MAX_ORDER-1 pages either to the head or the tail
> of the freelist. Let's document why we have to shuffle the complete zone
> when exposing larger, contiguous physical memory areas to the buddy.
>
How about?
Fixes: e900a918b098 ("mm: shuffle initial free memory to improve
memory-side-cache utilization")
...just like Patch1 since that original commit was missing the proper
commentary in the code?
> Cc: Andrew Morton <[email protected]>
> Cc: Alexander Duyck <[email protected]>
> Cc: Dan Williams <[email protected]>
> Cc: Michal Hocko <[email protected]>
> Signed-off-by: David Hildenbrand <[email protected]>
> ---
> mm/memory_hotplug.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
> index 9b34e03e730a4..a0d81d404823d 100644
> --- a/mm/memory_hotplug.c
> +++ b/mm/memory_hotplug.c
> @@ -822,6 +822,14 @@ int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
> zone->zone_pgdat->node_present_pages += onlined_pages;
> pgdat_resize_unlock(zone->zone_pgdat, &flags);
>
> + /*
> + * When exposing larger, physically contiguous memory areas to the
> + * buddy, shuffling in the buddy (when freeing onlined pages, putting
> + * them either to the head or the tail of the freelist) is only helpful
> + * for mainining the shuffle, but not for creating the initial shuffle.
s/mainining/maintaining/
> + * Shuffle the whole zone to make sure the just onlined pages are
> + * properly distributed across the whole freelist.
> + */
> shuffle_zone(zone);
>
> node_states_set_node(nid, &arg);
Other than the above minor fixups you can add:
Acked-by: Dan Williams <[email protected]>
On 20.06.20 03:41, Dan Williams wrote:
> On Fri, Jun 19, 2020 at 6:00 AM David Hildenbrand <[email protected]> wrote:
>>
>> It's not completely obvious why we have to shuffle the complete zone, as
>> some sort of shuffling is already performed when onlining pages via
>> __free_one_page(), placing MAX_ORDER-1 pages either to the head or the tail
>> of the freelist. Let's document why we have to shuffle the complete zone
>> when exposing larger, contiguous physical memory areas to the buddy.
>>
>
> How about?
>
> Fixes: e900a918b098 ("mm: shuffle initial free memory to improve
> memory-side-cache utilization")
>
> ...just like Patch1 since that original commit was missing the proper
> commentary in the code?
Hmm, mixed feelings. I (working for a distributor :) ) prefer fixes tags
for actual BUGs, as described in
Documentation/process/submitting-patches.rst: "If your patch fixes a bug
in a specific commit, e.g. you found an issue using ``git bisect``,
please use the 'Fixes:' tag with the first 12 characters" ...
So unless there are strong feelings, I'll not add a fixes tag (although
I agree, that it should have been contained in the original commit).
>> Cc: Andrew Morton <[email protected]>
>> Cc: Alexander Duyck <[email protected]>
>> Cc: Dan Williams <[email protected]>
>> Cc: Michal Hocko <[email protected]>
>> Signed-off-by: David Hildenbrand <[email protected]>
>> ---
>> mm/memory_hotplug.c | 8 ++++++++
>> 1 file changed, 8 insertions(+)
>>
>> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
>> index 9b34e03e730a4..a0d81d404823d 100644
>> --- a/mm/memory_hotplug.c
>> +++ b/mm/memory_hotplug.c
>> @@ -822,6 +822,14 @@ int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
>> zone->zone_pgdat->node_present_pages += onlined_pages;
>> pgdat_resize_unlock(zone->zone_pgdat, &flags);
>>
>> + /*
>> + * When exposing larger, physically contiguous memory areas to the
>> + * buddy, shuffling in the buddy (when freeing onlined pages, putting
>> + * them either to the head or the tail of the freelist) is only helpful
>> + * for mainining the shuffle, but not for creating the initial shuffle.
>
> s/mainining/maintaining/
Huh, what went wrong there :) Thanks!
--
Thanks,
David / dhildenb
On Fri 19-06-20 14:59:21, David Hildenbrand wrote:
> It's not completely obvious why we have to shuffle the complete zone, as
> some sort of shuffling is already performed when onlining pages via
> __free_one_page(), placing MAX_ORDER-1 pages either to the head or the tail
> of the freelist. Let's document why we have to shuffle the complete zone
> when exposing larger, contiguous physical memory areas to the buddy.
>
> Cc: Andrew Morton <[email protected]>
> Cc: Alexander Duyck <[email protected]>
> Cc: Dan Williams <[email protected]>
> Cc: Michal Hocko <[email protected]>
> Signed-off-by: David Hildenbrand <[email protected]>
OK, this is an improvement. I would still prefer to have this claim
backed by some numbers but it seems we are not going to get any so we
can at least pretend to try as hard as possible especially when this is
not a hot path.
Acked-by: Michal Hocko <[email protected]>
> ---
> mm/memory_hotplug.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
> index 9b34e03e730a4..a0d81d404823d 100644
> --- a/mm/memory_hotplug.c
> +++ b/mm/memory_hotplug.c
> @@ -822,6 +822,14 @@ int __ref online_pages(unsigned long pfn, unsigned long nr_pages,
> zone->zone_pgdat->node_present_pages += onlined_pages;
> pgdat_resize_unlock(zone->zone_pgdat, &flags);
>
> + /*
> + * When exposing larger, physically contiguous memory areas to the
> + * buddy, shuffling in the buddy (when freeing onlined pages, putting
> + * them either to the head or the tail of the freelist) is only helpful
> + * for mainining the shuffle, but not for creating the initial shuffle.
> + * Shuffle the whole zone to make sure the just onlined pages are
> + * properly distributed across the whole freelist.
> + */
> shuffle_zone(zone);
>
> node_states_set_node(nid, &arg);
> --
> 2.26.2
--
Michal Hocko
SUSE Labs
On Mon, Jun 22, 2020 at 12:28 AM David Hildenbrand <[email protected]> wrote:
>
> On 20.06.20 03:41, Dan Williams wrote:
> > On Fri, Jun 19, 2020 at 6:00 AM David Hildenbrand <[email protected]> wrote:
> >>
> >> It's not completely obvious why we have to shuffle the complete zone, as
> >> some sort of shuffling is already performed when onlining pages via
> >> __free_one_page(), placing MAX_ORDER-1 pages either to the head or the tail
> >> of the freelist. Let's document why we have to shuffle the complete zone
> >> when exposing larger, contiguous physical memory areas to the buddy.
> >>
> >
> > How about?
> >
> > Fixes: e900a918b098 ("mm: shuffle initial free memory to improve
> > memory-side-cache utilization")
> >
> > ...just like Patch1 since that original commit was missing the proper
> > commentary in the code?
>
> Hmm, mixed feelings. I (working for a distributor :) ) prefer fixes tags
> for actual BUGs, as described in
>
> Documentation/process/submitting-patches.rst: "If your patch fixes a bug
> in a specific commit, e.g. you found an issue using ``git bisect``,
> please use the 'Fixes:' tag with the first 12 characters" ...
>
> So unless there are strong feelings, I'll not add a fixes tag (although
> I agree, that it should have been contained in the original commit).
It doesn't need to be "Fixes", but how about at least mentioning the
original commit as a breadcrumb so that some future "git blame"
archaeology effort is streamlined.
On 23.06.20 23:15, Dan Williams wrote:
> On Mon, Jun 22, 2020 at 12:28 AM David Hildenbrand <[email protected]> wrote:
>>
>> On 20.06.20 03:41, Dan Williams wrote:
>>> On Fri, Jun 19, 2020 at 6:00 AM David Hildenbrand <[email protected]> wrote:
>>>>
>>>> It's not completely obvious why we have to shuffle the complete zone, as
>>>> some sort of shuffling is already performed when onlining pages via
>>>> __free_one_page(), placing MAX_ORDER-1 pages either to the head or the tail
>>>> of the freelist. Let's document why we have to shuffle the complete zone
>>>> when exposing larger, contiguous physical memory areas to the buddy.
>>>>
>>>
>>> How about?
>>>
>>> Fixes: e900a918b098 ("mm: shuffle initial free memory to improve
>>> memory-side-cache utilization")
>>>
>>> ...just like Patch1 since that original commit was missing the proper
>>> commentary in the code?
>>
>> Hmm, mixed feelings. I (working for a distributor :) ) prefer fixes tags
>> for actual BUGs, as described in
>>
>> Documentation/process/submitting-patches.rst: "If your patch fixes a bug
>> in a specific commit, e.g. you found an issue using ``git bisect``,
>> please use the 'Fixes:' tag with the first 12 characters" ...
>>
>> So unless there are strong feelings, I'll not add a fixes tag (although
>> I agree, that it should have been contained in the original commit).
>
> It doesn't need to be "Fixes", but how about at least mentioning the
> original commit as a breadcrumb so that some future "git blame"
> archaeology effort is streamlined.
>
Makes sense, I'll mention it as
It's not completely obvious why we have to shuffle the complete zone (
introduced in commit e900a918b098 ("mm: shuffle initial free memory to
...
thanks!
--
Thanks,
David / dhildenb