* [PATCH v3] mm: add per-order mTHP swap-in fallback/fallback_charge counters
@ 2024-12-02 12:47 Wenchao Hao
2024-12-02 22:44 ` Barry Song
2024-12-06 11:21 ` Lance Yang
0 siblings, 2 replies; 3+ messages in thread
From: Wenchao Hao @ 2024-12-02 12:47 UTC (permalink / raw)
To: Jonathan Corbet, Andrew Morton, David Hildenbrand, Barry Song,
Ryan Roberts, Baolin Wang, Usama Arif, Lance Yang,
Matthew Wilcox, Peter Xu, linux-doc, linux-kernel, linux-mm
Cc: Wenchao Hao
Currently, large folio swap-in is supported, but we lack a method to
analyze their success ratio. Similar to anon_fault_fallback, we introduce
per-order mTHP swpin_fallback and swpin_fallback_charge counters for
calculating their success ratio. The new counters are located at:
/sys/kernel/mm/transparent_hugepage/hugepages-<size>/stats/
swpin_fallback
swpin_fallback_charge
Signed-off-by: Wenchao Hao <haowenchao22@gmail.com>
---
V3:
Update description about swpin_fallback and swpin_fallback_charge
V2:
Introduce swapin_fallback_charge, which increments if it fails to
charge a huge page to memory despite successful allocation.
Documentation/admin-guide/mm/transhuge.rst | 10 ++++++++++
include/linux/huge_mm.h | 2 ++
mm/huge_memory.c | 6 ++++++
mm/memory.c | 2 ++
4 files changed, 20 insertions(+)
diff --git a/Documentation/admin-guide/mm/transhuge.rst b/Documentation/admin-guide/mm/transhuge.rst
index 333958ef0d5f..156a03af0a88 100644
--- a/Documentation/admin-guide/mm/transhuge.rst
+++ b/Documentation/admin-guide/mm/transhuge.rst
@@ -591,6 +591,16 @@ swpin
is incremented every time a huge page is swapped in from a non-zswap
swap device in one piece.
+swpin_fallback
+ is incremented if swapin fails to allocate or charge a huge page
+ and instead falls back to using huge pages with lower orders or
+ small pages.
+
+swpin_fallback_charge
+ is incremented if swapin fails to charge a huge page and instead
+ falls back to using huge pages with lower orders or small pages
+ even though the allocation was successful.
+
swpout
is incremented every time a huge page is swapped out to a non-zswap
swap device in one piece without splitting.
diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
index b94c2e8ee918..93e509b6c00e 100644
--- a/include/linux/huge_mm.h
+++ b/include/linux/huge_mm.h
@@ -121,6 +121,8 @@ enum mthp_stat_item {
MTHP_STAT_ANON_FAULT_FALLBACK_CHARGE,
MTHP_STAT_ZSWPOUT,
MTHP_STAT_SWPIN,
+ MTHP_STAT_SWPIN_FALLBACK,
+ MTHP_STAT_SWPIN_FALLBACK_CHARGE,
MTHP_STAT_SWPOUT,
MTHP_STAT_SWPOUT_FALLBACK,
MTHP_STAT_SHMEM_ALLOC,
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index ab46ef718b44..d062b257376d 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -617,6 +617,8 @@ DEFINE_MTHP_STAT_ATTR(anon_fault_fallback, MTHP_STAT_ANON_FAULT_FALLBACK);
DEFINE_MTHP_STAT_ATTR(anon_fault_fallback_charge, MTHP_STAT_ANON_FAULT_FALLBACK_CHARGE);
DEFINE_MTHP_STAT_ATTR(zswpout, MTHP_STAT_ZSWPOUT);
DEFINE_MTHP_STAT_ATTR(swpin, MTHP_STAT_SWPIN);
+DEFINE_MTHP_STAT_ATTR(swpin_fallback, MTHP_STAT_SWPIN_FALLBACK);
+DEFINE_MTHP_STAT_ATTR(swpin_fallback_charge, MTHP_STAT_SWPIN_FALLBACK_CHARGE);
DEFINE_MTHP_STAT_ATTR(swpout, MTHP_STAT_SWPOUT);
DEFINE_MTHP_STAT_ATTR(swpout_fallback, MTHP_STAT_SWPOUT_FALLBACK);
#ifdef CONFIG_SHMEM
@@ -637,6 +639,8 @@ static struct attribute *anon_stats_attrs[] = {
#ifndef CONFIG_SHMEM
&zswpout_attr.attr,
&swpin_attr.attr,
+ &swpin_fallback_attr.attr,
+ &swpin_fallback_charge_attr.attr,
&swpout_attr.attr,
&swpout_fallback_attr.attr,
#endif
@@ -669,6 +673,8 @@ static struct attribute *any_stats_attrs[] = {
#ifdef CONFIG_SHMEM
&zswpout_attr.attr,
&swpin_attr.attr,
+ &swpin_fallback_attr.attr,
+ &swpin_fallback_charge_attr.attr,
&swpout_attr.attr,
&swpout_fallback_attr.attr,
#endif
diff --git a/mm/memory.c b/mm/memory.c
index d5a1b0a6bf1f..a44547600c02 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -4189,8 +4189,10 @@ static struct folio *alloc_swap_folio(struct vm_fault *vmf)
if (!mem_cgroup_swapin_charge_folio(folio, vma->vm_mm,
gfp, entry))
return folio;
+ count_mthp_stat(order, MTHP_STAT_SWPIN_FALLBACK_CHARGE);
folio_put(folio);
}
+ count_mthp_stat(order, MTHP_STAT_SWPIN_FALLBACK);
order = next_order(&orders, order);
}
--
2.45.0
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH v3] mm: add per-order mTHP swap-in fallback/fallback_charge counters
2024-12-02 12:47 [PATCH v3] mm: add per-order mTHP swap-in fallback/fallback_charge counters Wenchao Hao
@ 2024-12-02 22:44 ` Barry Song
2024-12-06 11:21 ` Lance Yang
1 sibling, 0 replies; 3+ messages in thread
From: Barry Song @ 2024-12-02 22:44 UTC (permalink / raw)
To: Wenchao Hao
Cc: Jonathan Corbet, Andrew Morton, David Hildenbrand, Ryan Roberts,
Baolin Wang, Usama Arif, Lance Yang, Matthew Wilcox, Peter Xu,
linux-doc, linux-kernel, linux-mm
On Tue, Dec 3, 2024 at 1:47 AM Wenchao Hao <haowenchao22@gmail.com> wrote:
>
> Currently, large folio swap-in is supported, but we lack a method to
> analyze their success ratio. Similar to anon_fault_fallback, we introduce
> per-order mTHP swpin_fallback and swpin_fallback_charge counters for
> calculating their success ratio. The new counters are located at:
>
> /sys/kernel/mm/transparent_hugepage/hugepages-<size>/stats/
> swpin_fallback
> swpin_fallback_charge
>
> Signed-off-by: Wenchao Hao <haowenchao22@gmail.com>
LGTM, thanks!
Reviewed-by: Barry Song <baohua@kernel.org>
> ---
> V3:
> Update description about swpin_fallback and swpin_fallback_charge
>
> V2:
> Introduce swapin_fallback_charge, which increments if it fails to
> charge a huge page to memory despite successful allocation.
>
> Documentation/admin-guide/mm/transhuge.rst | 10 ++++++++++
> include/linux/huge_mm.h | 2 ++
> mm/huge_memory.c | 6 ++++++
> mm/memory.c | 2 ++
> 4 files changed, 20 insertions(+)
>
> diff --git a/Documentation/admin-guide/mm/transhuge.rst b/Documentation/admin-guide/mm/transhuge.rst
> index 333958ef0d5f..156a03af0a88 100644
> --- a/Documentation/admin-guide/mm/transhuge.rst
> +++ b/Documentation/admin-guide/mm/transhuge.rst
> @@ -591,6 +591,16 @@ swpin
> is incremented every time a huge page is swapped in from a non-zswap
> swap device in one piece.
>
> +swpin_fallback
> + is incremented if swapin fails to allocate or charge a huge page
> + and instead falls back to using huge pages with lower orders or
> + small pages.
> +
> +swpin_fallback_charge
> + is incremented if swapin fails to charge a huge page and instead
> + falls back to using huge pages with lower orders or small pages
> + even though the allocation was successful.
> +
> swpout
> is incremented every time a huge page is swapped out to a non-zswap
> swap device in one piece without splitting.
> diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
> index b94c2e8ee918..93e509b6c00e 100644
> --- a/include/linux/huge_mm.h
> +++ b/include/linux/huge_mm.h
> @@ -121,6 +121,8 @@ enum mthp_stat_item {
> MTHP_STAT_ANON_FAULT_FALLBACK_CHARGE,
> MTHP_STAT_ZSWPOUT,
> MTHP_STAT_SWPIN,
> + MTHP_STAT_SWPIN_FALLBACK,
> + MTHP_STAT_SWPIN_FALLBACK_CHARGE,
> MTHP_STAT_SWPOUT,
> MTHP_STAT_SWPOUT_FALLBACK,
> MTHP_STAT_SHMEM_ALLOC,
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index ab46ef718b44..d062b257376d 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -617,6 +617,8 @@ DEFINE_MTHP_STAT_ATTR(anon_fault_fallback, MTHP_STAT_ANON_FAULT_FALLBACK);
> DEFINE_MTHP_STAT_ATTR(anon_fault_fallback_charge, MTHP_STAT_ANON_FAULT_FALLBACK_CHARGE);
> DEFINE_MTHP_STAT_ATTR(zswpout, MTHP_STAT_ZSWPOUT);
> DEFINE_MTHP_STAT_ATTR(swpin, MTHP_STAT_SWPIN);
> +DEFINE_MTHP_STAT_ATTR(swpin_fallback, MTHP_STAT_SWPIN_FALLBACK);
> +DEFINE_MTHP_STAT_ATTR(swpin_fallback_charge, MTHP_STAT_SWPIN_FALLBACK_CHARGE);
> DEFINE_MTHP_STAT_ATTR(swpout, MTHP_STAT_SWPOUT);
> DEFINE_MTHP_STAT_ATTR(swpout_fallback, MTHP_STAT_SWPOUT_FALLBACK);
> #ifdef CONFIG_SHMEM
> @@ -637,6 +639,8 @@ static struct attribute *anon_stats_attrs[] = {
> #ifndef CONFIG_SHMEM
> &zswpout_attr.attr,
> &swpin_attr.attr,
> + &swpin_fallback_attr.attr,
> + &swpin_fallback_charge_attr.attr,
> &swpout_attr.attr,
> &swpout_fallback_attr.attr,
> #endif
> @@ -669,6 +673,8 @@ static struct attribute *any_stats_attrs[] = {
> #ifdef CONFIG_SHMEM
> &zswpout_attr.attr,
> &swpin_attr.attr,
> + &swpin_fallback_attr.attr,
> + &swpin_fallback_charge_attr.attr,
> &swpout_attr.attr,
> &swpout_fallback_attr.attr,
> #endif
> diff --git a/mm/memory.c b/mm/memory.c
> index d5a1b0a6bf1f..a44547600c02 100644
> --- a/mm/memory.c
> +++ b/mm/memory.c
> @@ -4189,8 +4189,10 @@ static struct folio *alloc_swap_folio(struct vm_fault *vmf)
> if (!mem_cgroup_swapin_charge_folio(folio, vma->vm_mm,
> gfp, entry))
> return folio;
> + count_mthp_stat(order, MTHP_STAT_SWPIN_FALLBACK_CHARGE);
> folio_put(folio);
> }
> + count_mthp_stat(order, MTHP_STAT_SWPIN_FALLBACK);
> order = next_order(&orders, order);
> }
>
> --
> 2.45.0
>
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH v3] mm: add per-order mTHP swap-in fallback/fallback_charge counters
2024-12-02 12:47 [PATCH v3] mm: add per-order mTHP swap-in fallback/fallback_charge counters Wenchao Hao
2024-12-02 22:44 ` Barry Song
@ 2024-12-06 11:21 ` Lance Yang
1 sibling, 0 replies; 3+ messages in thread
From: Lance Yang @ 2024-12-06 11:21 UTC (permalink / raw)
To: Wenchao Hao
Cc: Jonathan Corbet, Andrew Morton, David Hildenbrand, Barry Song,
Ryan Roberts, Baolin Wang, Usama Arif, Matthew Wilcox, Peter Xu,
linux-doc, linux-kernel, linux-mm
On Mon, Dec 2, 2024 at 8:47 PM Wenchao Hao <haowenchao22@gmail.com> wrote:
>
> Currently, large folio swap-in is supported, but we lack a method to
> analyze their success ratio. Similar to anon_fault_fallback, we introduce
> per-order mTHP swpin_fallback and swpin_fallback_charge counters for
> calculating their success ratio. The new counters are located at:
>
> /sys/kernel/mm/transparent_hugepage/hugepages-<size>/stats/
> swpin_fallback
> swpin_fallback_charge
>
> Signed-off-by: Wenchao Hao <haowenchao22@gmail.com>
LGTM. Feel free to add:
Reviewed-by: Lance Yang <ioworker0@gmail.com>
Thanks,
Lance
> ---
> V3:
> Update description about swpin_fallback and swpin_fallback_charge
>
> V2:
> Introduce swapin_fallback_charge, which increments if it fails to
> charge a huge page to memory despite successful allocation.
>
> Documentation/admin-guide/mm/transhuge.rst | 10 ++++++++++
> include/linux/huge_mm.h | 2 ++
> mm/huge_memory.c | 6 ++++++
> mm/memory.c | 2 ++
> 4 files changed, 20 insertions(+)
>
> diff --git a/Documentation/admin-guide/mm/transhuge.rst b/Documentation/admin-guide/mm/transhuge.rst
> index 333958ef0d5f..156a03af0a88 100644
> --- a/Documentation/admin-guide/mm/transhuge.rst
> +++ b/Documentation/admin-guide/mm/transhuge.rst
> @@ -591,6 +591,16 @@ swpin
> is incremented every time a huge page is swapped in from a non-zswap
> swap device in one piece.
>
> +swpin_fallback
> + is incremented if swapin fails to allocate or charge a huge page
> + and instead falls back to using huge pages with lower orders or
> + small pages.
> +
> +swpin_fallback_charge
> + is incremented if swapin fails to charge a huge page and instead
> + falls back to using huge pages with lower orders or small pages
> + even though the allocation was successful.
> +
> swpout
> is incremented every time a huge page is swapped out to a non-zswap
> swap device in one piece without splitting.
> diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
> index b94c2e8ee918..93e509b6c00e 100644
> --- a/include/linux/huge_mm.h
> +++ b/include/linux/huge_mm.h
> @@ -121,6 +121,8 @@ enum mthp_stat_item {
> MTHP_STAT_ANON_FAULT_FALLBACK_CHARGE,
> MTHP_STAT_ZSWPOUT,
> MTHP_STAT_SWPIN,
> + MTHP_STAT_SWPIN_FALLBACK,
> + MTHP_STAT_SWPIN_FALLBACK_CHARGE,
> MTHP_STAT_SWPOUT,
> MTHP_STAT_SWPOUT_FALLBACK,
> MTHP_STAT_SHMEM_ALLOC,
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index ab46ef718b44..d062b257376d 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -617,6 +617,8 @@ DEFINE_MTHP_STAT_ATTR(anon_fault_fallback, MTHP_STAT_ANON_FAULT_FALLBACK);
> DEFINE_MTHP_STAT_ATTR(anon_fault_fallback_charge, MTHP_STAT_ANON_FAULT_FALLBACK_CHARGE);
> DEFINE_MTHP_STAT_ATTR(zswpout, MTHP_STAT_ZSWPOUT);
> DEFINE_MTHP_STAT_ATTR(swpin, MTHP_STAT_SWPIN);
> +DEFINE_MTHP_STAT_ATTR(swpin_fallback, MTHP_STAT_SWPIN_FALLBACK);
> +DEFINE_MTHP_STAT_ATTR(swpin_fallback_charge, MTHP_STAT_SWPIN_FALLBACK_CHARGE);
> DEFINE_MTHP_STAT_ATTR(swpout, MTHP_STAT_SWPOUT);
> DEFINE_MTHP_STAT_ATTR(swpout_fallback, MTHP_STAT_SWPOUT_FALLBACK);
> #ifdef CONFIG_SHMEM
> @@ -637,6 +639,8 @@ static struct attribute *anon_stats_attrs[] = {
> #ifndef CONFIG_SHMEM
> &zswpout_attr.attr,
> &swpin_attr.attr,
> + &swpin_fallback_attr.attr,
> + &swpin_fallback_charge_attr.attr,
> &swpout_attr.attr,
> &swpout_fallback_attr.attr,
> #endif
> @@ -669,6 +673,8 @@ static struct attribute *any_stats_attrs[] = {
> #ifdef CONFIG_SHMEM
> &zswpout_attr.attr,
> &swpin_attr.attr,
> + &swpin_fallback_attr.attr,
> + &swpin_fallback_charge_attr.attr,
> &swpout_attr.attr,
> &swpout_fallback_attr.attr,
> #endif
> diff --git a/mm/memory.c b/mm/memory.c
> index d5a1b0a6bf1f..a44547600c02 100644
> --- a/mm/memory.c
> +++ b/mm/memory.c
> @@ -4189,8 +4189,10 @@ static struct folio *alloc_swap_folio(struct vm_fault *vmf)
> if (!mem_cgroup_swapin_charge_folio(folio, vma->vm_mm,
> gfp, entry))
> return folio;
> + count_mthp_stat(order, MTHP_STAT_SWPIN_FALLBACK_CHARGE);
> folio_put(folio);
> }
> + count_mthp_stat(order, MTHP_STAT_SWPIN_FALLBACK);
> order = next_order(&orders, order);
> }
>
> --
> 2.45.0
>
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2024-12-06 11:21 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-12-02 12:47 [PATCH v3] mm: add per-order mTHP swap-in fallback/fallback_charge counters Wenchao Hao
2024-12-02 22:44 ` Barry Song
2024-12-06 11:21 ` Lance Yang
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox