This was the behavior prior to making the function arch-specific with
commit ecd096506922 ("mm: make deferred init's max threads
arch-specific")
Architectures can override the generic implementation that uses only one
CPU. Setting DEFERRED_STRUCT_PAGE_INIT and testing on a few arm64
platforms shows faster deferred_init_memmap completions:
| | x13s | SA8775p-ride | Ampere R137-P31 | Ampere HR330 |
| | Metal, 32GB | VM, 36GB | VM, 58GB | Metal, 128GB |
| | 8cpus | 8cpus | 8cpus | 32cpus |
|---------|-------------|--------------|-----------------|--------------|
| threads | ms (%) | ms (%) | ms (%) | ms (%) |
|---------|-------------|--------------|-----------------|--------------|
| 1 | 108 (0%) | 72 (0%) | 224 (0%) | 324 (0%) |
| cpus | 24 (-77%) | 36 (-50%) | 40 (-82%) | 56 (-82%) |
Signed-off-by: Eric Chanudet <[email protected]>
---
arch/arm64/mm/init.c | 7 +++++++
1 file changed, 7 insertions(+)
diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
index 9b5ab6818f7f..71f5188fe63d 100644
--- a/arch/arm64/mm/init.c
+++ b/arch/arm64/mm/init.c
@@ -158,6 +158,13 @@ static void __init zone_sizes_init(void)
free_area_init(max_zone_pfns);
}
+#ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
+int __init deferred_page_init_max_threads(const struct cpumask *node_cpumask)
+{
+ return max_t(int, cpumask_weight(node_cpumask), 1);
+}
+#endif
+
int pfn_is_map_memory(unsigned long pfn)
{
phys_addr_t addr = PFN_PHYS(pfn);
--
2.44.0
On 05/20/24 at 07:15pm, Eric Chanudet wrote:
> This was the behavior prior to making the function arch-specific with
> commit ecd096506922 ("mm: make deferred init's max threads
> arch-specific")
>
> Architectures can override the generic implementation that uses only one
> CPU. Setting DEFERRED_STRUCT_PAGE_INIT and testing on a few arm64
> platforms shows faster deferred_init_memmap completions:
>
> | | x13s | SA8775p-ride | Ampere R137-P31 | Ampere HR330 |
> | | Metal, 32GB | VM, 36GB | VM, 58GB | Metal, 128GB |
> | | 8cpus | 8cpus | 8cpus | 32cpus |
> |---------|-------------|--------------|-----------------|--------------|
> | threads | ms (%) | ms (%) | ms (%) | ms (%) |
> |---------|-------------|--------------|-----------------|--------------|
> | 1 | 108 (0%) | 72 (0%) | 224 (0%) | 324 (0%) |
> | cpus | 24 (-77%) | 36 (-50%) | 40 (-82%) | 56 (-82%) |
>
> Signed-off-by: Eric Chanudet <[email protected]>
> ---
> arch/arm64/mm/init.c | 7 +++++++
> 1 file changed, 7 insertions(+)
>
> diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
> index 9b5ab6818f7f..71f5188fe63d 100644
> --- a/arch/arm64/mm/init.c
> +++ b/arch/arm64/mm/init.c
> @@ -158,6 +158,13 @@ static void __init zone_sizes_init(void)
> free_area_init(max_zone_pfns);
> }
>
> +#ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
> +int __init deferred_page_init_max_threads(const struct cpumask *node_cpumask)
> +{
> + return max_t(int, cpumask_weight(node_cpumask), 1);
> +}
> +#endif
LGTM,
Reviewed-by: Baoquan He <[email protected]>
> +
> int pfn_is_map_memory(unsigned long pfn)
> {
> phys_addr_t addr = PFN_PHYS(pfn);
> --
> 2.44.0
>
(added powerpc folks)
On Mon, May 20, 2024 at 07:15:59PM -0400, Eric Chanudet wrote:
> This was the behavior prior to making the function arch-specific with
> commit ecd096506922 ("mm: make deferred init's max threads
> arch-specific")
>
> Architectures can override the generic implementation that uses only one
> CPU. Setting DEFERRED_STRUCT_PAGE_INIT and testing on a few arm64
> platforms shows faster deferred_init_memmap completions:
>
> | | x13s | SA8775p-ride | Ampere R137-P31 | Ampere HR330 |
> | | Metal, 32GB | VM, 36GB | VM, 58GB | Metal, 128GB |
> | | 8cpus | 8cpus | 8cpus | 32cpus |
> |---------|-------------|--------------|-----------------|--------------|
> | threads | ms (%) | ms (%) | ms (%) | ms (%) |
> |---------|-------------|--------------|-----------------|--------------|
> | 1 | 108 (0%) | 72 (0%) | 224 (0%) | 324 (0%) |
> | cpus | 24 (-77%) | 36 (-50%) | 40 (-82%) | 56 (-82%) |
>
> Signed-off-by: Eric Chanudet <[email protected]>
> ---
> arch/arm64/mm/init.c | 7 +++++++
> 1 file changed, 7 insertions(+)
>
> diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
> index 9b5ab6818f7f..71f5188fe63d 100644
> --- a/arch/arm64/mm/init.c
> +++ b/arch/arm64/mm/init.c
> @@ -158,6 +158,13 @@ static void __init zone_sizes_init(void)
> free_area_init(max_zone_pfns);
> }
>
> +#ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
> +int __init deferred_page_init_max_threads(const struct cpumask *node_cpumask)
> +{
> + return max_t(int, cpumask_weight(node_cpumask), 1);
> +}
> +#endif
> +
Maybe we should make this default and let architectures that want a single
thread override deferred_page_init_max_threads() to return 1?
> int pfn_is_map_memory(unsigned long pfn)
> {
> phys_addr_t addr = PFN_PHYS(pfn);
> --
> 2.44.0
>
--
Sincerely yours,
Mike.
On Tue, May 21, 2024 at 07:10:07PM +0300, Mike Rapoport wrote:
> (added powerpc folks)
>
> On Mon, May 20, 2024 at 07:15:59PM -0400, Eric Chanudet wrote:
> > This was the behavior prior to making the function arch-specific with
> > commit ecd096506922 ("mm: make deferred init's max threads
> > arch-specific")
> >
> > Architectures can override the generic implementation that uses only one
> > CPU. Setting DEFERRED_STRUCT_PAGE_INIT and testing on a few arm64
> > platforms shows faster deferred_init_memmap completions:
> >
> > | | x13s | SA8775p-ride | Ampere R137-P31 | Ampere HR330 |
> > | | Metal, 32GB | VM, 36GB | VM, 58GB | Metal, 128GB |
> > | | 8cpus | 8cpus | 8cpus | 32cpus |
> > |---------|-------------|--------------|-----------------|--------------|
> > | threads | ms (%) | ms (%) | ms (%) | ms (%) |
> > |---------|-------------|--------------|-----------------|--------------|
> > | 1 | 108 (0%) | 72 (0%) | 224 (0%) | 324 (0%) |
> > | cpus | 24 (-77%) | 36 (-50%) | 40 (-82%) | 56 (-82%) |
> >
> > Signed-off-by: Eric Chanudet <[email protected]>
> > ---
> > arch/arm64/mm/init.c | 7 +++++++
> > 1 file changed, 7 insertions(+)
> >
> > diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
> > index 9b5ab6818f7f..71f5188fe63d 100644
> > --- a/arch/arm64/mm/init.c
> > +++ b/arch/arm64/mm/init.c
> > @@ -158,6 +158,13 @@ static void __init zone_sizes_init(void)
> > free_area_init(max_zone_pfns);
> > }
> >
> > +#ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
> > +int __init deferred_page_init_max_threads(const struct cpumask *node_cpumask)
> > +{
> > + return max_t(int, cpumask_weight(node_cpumask), 1);
> > +}
> > +#endif
> > +
>
> Maybe we should make this default and let architectures that want a single
> thread override deferred_page_init_max_threads() to return 1?
It would affect more archs than I can try this on. Currently, only x86
(with this change, arm64) return more than one thread.
I'm happy to send a v2 inverting the logic if you find it preferable.
Best,
> > int pfn_is_map_memory(unsigned long pfn)
> > {
> > phys_addr_t addr = PFN_PHYS(pfn);
> > --
> > 2.44.0
> >
>
> --
> Sincerely yours,
> Mike.
>
--
Eric Chanudet
Eric Chanudet <[email protected]> writes:
> On Tue, May 21, 2024 at 07:10:07PM +0300, Mike Rapoport wrote:
>> (added powerpc folks)
Thanks Mike.
>> On Mon, May 20, 2024 at 07:15:59PM -0400, Eric Chanudet wrote:
>> > This was the behavior prior to making the function arch-specific with
>> > commit ecd096506922 ("mm: make deferred init's max threads
>> > arch-specific")
>> >
>> > Architectures can override the generic implementation that uses only one
>> > CPU. Setting DEFERRED_STRUCT_PAGE_INIT and testing on a few arm64
>> > platforms shows faster deferred_init_memmap completions:
>> >
>> > | | x13s | SA8775p-ride | Ampere R137-P31 | Ampere HR330 |
>> > | | Metal, 32GB | VM, 36GB | VM, 58GB | Metal, 128GB |
>> > | | 8cpus | 8cpus | 8cpus | 32cpus |
>> > |---------|-------------|--------------|-----------------|--------------|
>> > | threads | ms (%) | ms (%) | ms (%) | ms (%) |
>> > |---------|-------------|--------------|-----------------|--------------|
>> > | 1 | 108 (0%) | 72 (0%) | 224 (0%) | 324 (0%) |
>> > | cpus | 24 (-77%) | 36 (-50%) | 40 (-82%) | 56 (-82%) |
How did you measure this, just some printks in page_alloc_init_late() or
something more sophisticated? Just so I can do some comparable measurements.
>> > diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
>> > index 9b5ab6818f7f..71f5188fe63d 100644
>> > --- a/arch/arm64/mm/init.c
>> > +++ b/arch/arm64/mm/init.c
>> > @@ -158,6 +158,13 @@ static void __init zone_sizes_init(void)
>> > free_area_init(max_zone_pfns);
>> > }
>> >
>> > +#ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
>> > +int __init deferred_page_init_max_threads(const struct cpumask *node_cpumask)
>> > +{
>> > + return max_t(int, cpumask_weight(node_cpumask), 1);
>> > +}
>> > +#endif
>> > +
>>
>> Maybe we should make this default and let architectures that want a single
>> thread override deferred_page_init_max_threads() to return 1?
>
> It would affect more archs than I can try this on. Currently, only x86
> (with this change, arm64) return more than one thread.
I can test powerpc and we can find someone to test s390. No other arches
have it enabled in their defconfig.
> I'm happy to send a v2 inverting the logic if you find it preferable.
That seems preferable. It's a scalability feature, it makes no sense for
the default to be a single thread AFAICS.
cheers
On Wed, May 22, 2024 at 11:41:07PM +1000, Michael Ellerman wrote:
> Eric Chanudet <[email protected]> writes:
> > On Tue, May 21, 2024 at 07:10:07PM +0300, Mike Rapoport wrote:
> >> (added powerpc folks)
>
> Thanks Mike.
>
> >> On Mon, May 20, 2024 at 07:15:59PM -0400, Eric Chanudet wrote:
> >> > This was the behavior prior to making the function arch-specific with
> >> > commit ecd096506922 ("mm: make deferred init's max threads
> >> > arch-specific")
> >> >
> >> > Architectures can override the generic implementation that uses only one
> >> > CPU. Setting DEFERRED_STRUCT_PAGE_INIT and testing on a few arm64
> >> > platforms shows faster deferred_init_memmap completions:
> >> >
> >> > | | x13s | SA8775p-ride | Ampere R137-P31 | Ampere HR330 |
> >> > | | Metal, 32GB | VM, 36GB | VM, 58GB | Metal, 128GB |
> >> > | | 8cpus | 8cpus | 8cpus | 32cpus |
> >> > |---------|-------------|--------------|-----------------|--------------|
> >> > | threads | ms (%) | ms (%) | ms (%) | ms (%) |
> >> > |---------|-------------|--------------|-----------------|--------------|
> >> > | 1 | 108 (0%) | 72 (0%) | 224 (0%) | 324 (0%) |
> >> > | cpus | 24 (-77%) | 36 (-50%) | 40 (-82%) | 56 (-82%) |
>
> How did you measure this, just some printks in page_alloc_init_late() or
> something more sophisticated? Just so I can do some comparable measurements.
I used the existing pr_info in deferred_init_memmap().
> >> > diff --git a/arch/arm64/mm/init.c b/arch/arm64/mm/init.c
> >> > index 9b5ab6818f7f..71f5188fe63d 100644
> >> > --- a/arch/arm64/mm/init.c
> >> > +++ b/arch/arm64/mm/init.c
> >> > @@ -158,6 +158,13 @@ static void __init zone_sizes_init(void)
> >> > free_area_init(max_zone_pfns);
> >> > }
> >> >
> >> > +#ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
> >> > +int __init deferred_page_init_max_threads(const struct cpumask *node_cpumask)
> >> > +{
> >> > + return max_t(int, cpumask_weight(node_cpumask), 1);
> >> > +}
> >> > +#endif
> >> > +
> >>
> >> Maybe we should make this default and let architectures that want a single
> >> thread override deferred_page_init_max_threads() to return 1?
> >
> > It would affect more archs than I can try this on. Currently, only x86
> > (with this change, arm64) return more than one thread.
>
> I can test powerpc and we can find someone to test s390. No other
> arches have it enabled in their defconfig.
Many thanks!
> > I'm happy to send a v2 inverting the logic if you find it preferable.
>
> That seems preferable. It's a scalability feature, it makes no sense for
> the default to be a single thread AFAICS.
Understood, I'll respin.
--
Eric Chanudet