diff mbox series

[-v4,8/9] migrate_pages: batch flushing TLB

Message ID 20230206063313.635011-9-ying.huang@intel.com
State New
Headers
Series migrate_pages(): batch TLB flushing |

Commit Message

Huang, Ying Feb. 6, 2023, 6:33 a.m. UTC 
  The TLB flushing will cost quite some CPU cycles during the folio
migration in some situations.  For example, when migrate a folio of a
process with multiple active threads that run on multiple CPUs.  After
batching the _unmap and _move in migrate_pages(), the TLB flushing can
be batched easily with the existing TLB flush batching mechanism.
This patch implements that.

We use the following test case to test the patch.

On a 2-socket Intel server,

- Run pmbench memory accessing benchmark

- Run `migratepages` to migrate pages of pmbench between node 0 and
  node 1 back and forth.

With the patch, the TLB flushing IPI reduces 99.1% during the test and
the number of pages migrated successfully per second increases 291.7%.

NOTE: TLB flushing is batched only for normal folios, not for THP
folios.  Because the overhead of TLB flushing for THP folios is much
lower than that for normal folios (about 1/512 on x86 platform).

Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
Cc: Zi Yan <ziy@nvidia.com>
Cc: Yang Shi <shy828301@gmail.com>
Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Bharata B Rao <bharata@amd.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: haoxin <xhao@linux.alibaba.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com>
---
 mm/migrate.c |  4 +++-
 mm/rmap.c    | 20 +++++++++++++++++---
 2 files changed, 20 insertions(+), 4 deletions(-)

Comments

Zi Yan Feb. 7, 2023, 2:52 p.m. UTC | #1 
On 6 Feb 2023, at 1:33, Huang Ying wrote:

> The TLB flushing will cost quite some CPU cycles during the folio
> migration in some situations.  For example, when migrate a folio of a
> process with multiple active threads that run on multiple CPUs.  After
> batching the _unmap and _move in migrate_pages(), the TLB flushing can
> be batched easily with the existing TLB flush batching mechanism.
> This patch implements that.
>
> We use the following test case to test the patch.
>
> On a 2-socket Intel server,
>
> - Run pmbench memory accessing benchmark
>
> - Run `migratepages` to migrate pages of pmbench between node 0 and
>   node 1 back and forth.
>
> With the patch, the TLB flushing IPI reduces 99.1% during the test and
> the number of pages migrated successfully per second increases 291.7%.
>
> NOTE: TLB flushing is batched only for normal folios, not for THP
> folios.  Because the overhead of TLB flushing for THP folios is much
> lower than that for normal folios (about 1/512 on x86 platform).
>
> Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
> Cc: Zi Yan <ziy@nvidia.com>
> Cc: Yang Shi <shy828301@gmail.com>
> Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
> Cc: Oscar Salvador <osalvador@suse.de>
> Cc: Matthew Wilcox <willy@infradead.org>
> Cc: Bharata B Rao <bharata@amd.com>
> Cc: Alistair Popple <apopple@nvidia.com>
> Cc: haoxin <xhao@linux.alibaba.com>
> Cc: Minchan Kim <minchan@kernel.org>
> Cc: Mike Kravetz <mike.kravetz@oracle.com>
> Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com>
> ---
>  mm/migrate.c |  4 +++-
>  mm/rmap.c    | 20 +++++++++++++++++---
>  2 files changed, 20 insertions(+), 4 deletions(-)
>
> diff --git a/mm/migrate.c b/mm/migrate.c
> index 9378fa2ad4a5..ca6e2ff02a09 100644
> --- a/mm/migrate.c
> +++ b/mm/migrate.c
> @@ -1230,7 +1230,7 @@ static int migrate_folio_unmap(new_page_t get_new_page, free_page_t put_new_page
>  		/* Establish migration ptes */
>  		VM_BUG_ON_FOLIO(folio_test_anon(src) &&
>  			       !folio_test_ksm(src) && !anon_vma, src);
> -		try_to_migrate(src, 0);
> +		try_to_migrate(src, TTU_BATCH_FLUSH);
>  		page_was_mapped = 1;
>  	}
>
> @@ -1781,6 +1781,8 @@ static int migrate_pages_batch(struct list_head *from, new_page_t get_new_page,
>  	stats->nr_thp_failed += thp_retry;
>  	stats->nr_failed_pages += nr_retry_pages;
>  move:

Maybe a comment:
/* Flush TLBs for all the unmapped pages */

> +	try_to_unmap_flush();
> +
>  	retry = 1;
>  	for (pass = 0;
>  	     pass < NR_MAX_MIGRATE_PAGES_RETRY && (retry || large_retry);
> diff --git a/mm/rmap.c b/mm/rmap.c
> index b616870a09be..2e125f3e462e 100644
> --- a/mm/rmap.c
> +++ b/mm/rmap.c
> @@ -1976,7 +1976,21 @@ static bool try_to_migrate_one(struct folio *folio, struct vm_area_struct *vma,
>  		} else {
>  			flush_cache_page(vma, address, pte_pfn(*pvmw.pte));
>  			/* Nuke the page table entry. */
> -			pteval = ptep_clear_flush(vma, address, pvmw.pte);
> +			if (should_defer_flush(mm, flags)) {
> +				/*
> +				 * We clear the PTE but do not flush so potentially
> +				 * a remote CPU could still be writing to the folio.
> +				 * If the entry was previously clean then the
> +				 * architecture must guarantee that a clear->dirty
> +				 * transition on a cached TLB entry is written through
> +				 * and traps if the PTE is unmapped.
> +				 */
> +				pteval = ptep_get_and_clear(mm, address, pvmw.pte);
> +
> +				set_tlb_ubc_flush_pending(mm, pte_dirty(pteval));
> +			} else {
> +				pteval = ptep_clear_flush(vma, address, pvmw.pte);
> +			}
>  		}
>
>  		/* Set the dirty flag on the folio now the pte is gone. */
> @@ -2148,10 +2162,10 @@ void try_to_migrate(struct folio *folio, enum ttu_flags flags)
>
>  	/*
>  	 * Migration always ignores mlock and only supports TTU_RMAP_LOCKED and
> -	 * TTU_SPLIT_HUGE_PMD and TTU_SYNC flags.
> +	 * TTU_SPLIT_HUGE_PMD, TTU_SYNC, and TTU_BATCH_FLUSH flags.
>  	 */
>  	if (WARN_ON_ONCE(flags & ~(TTU_RMAP_LOCKED | TTU_SPLIT_HUGE_PMD |
> -					TTU_SYNC)))
> +					TTU_SYNC | TTU_BATCH_FLUSH)))
>  		return;
>
>  	if (folio_is_zone_device(folio) &&
> -- 
> 2.35.1

Everything else looks good to me. Reviewed-by: Zi Yan <ziy@nvidia.com>

--
Best Regards,
Yan, Zi
haoxin Feb. 7, 2023, 5:44 p.m. UTC | #2 
在 2023/2/6 下午2:33, Huang Ying 写道:
> The TLB flushing will cost quite some CPU cycles during the folio
> migration in some situations.  For example, when migrate a folio of a
> process with multiple active threads that run on multiple CPUs.  After
> batching the _unmap and _move in migrate_pages(), the TLB flushing can
> be batched easily with the existing TLB flush batching mechanism.
> This patch implements that.
>
> We use the following test case to test the patch.
>
> On a 2-socket Intel server,
>
> - Run pmbench memory accessing benchmark
>
> - Run `migratepages` to migrate pages of pmbench between node 0 and
>    node 1 back and forth.
>
> With the patch, the TLB flushing IPI reduces 99.1% during the test and
> the number of pages migrated successfully per second increases 291.7%.
>
> NOTE: TLB flushing is batched only for normal folios, not for THP
> folios.  Because the overhead of TLB flushing for THP folios is much
> lower than that for normal folios (about 1/512 on x86 platform).
>
> Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
> Cc: Zi Yan <ziy@nvidia.com>
> Cc: Yang Shi <shy828301@gmail.com>
> Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
> Cc: Oscar Salvador <osalvador@suse.de>
> Cc: Matthew Wilcox <willy@infradead.org>
> Cc: Bharata B Rao <bharata@amd.com>
> Cc: Alistair Popple <apopple@nvidia.com>
> Cc: haoxin <xhao@linux.alibaba.com>
> Cc: Minchan Kim <minchan@kernel.org>
> Cc: Mike Kravetz <mike.kravetz@oracle.com>
> Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com>
> ---
>   mm/migrate.c |  4 +++-
>   mm/rmap.c    | 20 +++++++++++++++++---
>   2 files changed, 20 insertions(+), 4 deletions(-)
>
> diff --git a/mm/migrate.c b/mm/migrate.c
> index 9378fa2ad4a5..ca6e2ff02a09 100644
> --- a/mm/migrate.c
> +++ b/mm/migrate.c
> @@ -1230,7 +1230,7 @@ static int migrate_folio_unmap(new_page_t get_new_page, free_page_t put_new_page
>   		/* Establish migration ptes */
>   		VM_BUG_ON_FOLIO(folio_test_anon(src) &&
>   			       !folio_test_ksm(src) && !anon_vma, src);
> -		try_to_migrate(src, 0);
> +		try_to_migrate(src, TTU_BATCH_FLUSH);
>   		page_was_mapped = 1;
>   	}
>   
> @@ -1781,6 +1781,8 @@ static int migrate_pages_batch(struct list_head *from, new_page_t get_new_page,
>   	stats->nr_thp_failed += thp_retry;
>   	stats->nr_failed_pages += nr_retry_pages;
>   move:
> +	try_to_unmap_flush();
> +
>   	retry = 1;
>   	for (pass = 0;
>   	     pass < NR_MAX_MIGRATE_PAGES_RETRY && (retry || large_retry);
> diff --git a/mm/rmap.c b/mm/rmap.c
> index b616870a09be..2e125f3e462e 100644
> --- a/mm/rmap.c
> +++ b/mm/rmap.c
> @@ -1976,7 +1976,21 @@ static bool try_to_migrate_one(struct folio *folio, struct vm_area_struct *vma,
>   		} else {
>   			flush_cache_page(vma, address, pte_pfn(*pvmw.pte));
>   			/* Nuke the page table entry. */
> -			pteval = ptep_clear_flush(vma, address, pvmw.pte);
> +			if (should_defer_flush(mm, flags)) {
> +				/*
> +				 * We clear the PTE but do not flush so potentially
> +				 * a remote CPU could still be writing to the folio.
> +				 * If the entry was previously clean then the
> +				 * architecture must guarantee that a clear->dirty
> +				 * transition on a cached TLB entry is written through
> +				 * and traps if the PTE is unmapped.
> +				 */
> +				pteval = ptep_get_and_clear(mm, address, pvmw.pte);
Nice work， Reviewed-by: Xin Hao <xhao@linux.alibaba.com>
> +
> +				set_tlb_ubc_flush_pending(mm, pte_dirty(pteval));
> +			} else {
> +				pteval = ptep_clear_flush(vma, address, pvmw.pte);
> +			}
>   		}
>   
>   		/* Set the dirty flag on the folio now the pte is gone. */
> @@ -2148,10 +2162,10 @@ void try_to_migrate(struct folio *folio, enum ttu_flags flags)
>   
>   	/*
>   	 * Migration always ignores mlock and only supports TTU_RMAP_LOCKED and
> -	 * TTU_SPLIT_HUGE_PMD and TTU_SYNC flags.
> +	 * TTU_SPLIT_HUGE_PMD, TTU_SYNC, and TTU_BATCH_FLUSH flags.
>   	 */
>   	if (WARN_ON_ONCE(flags & ~(TTU_RMAP_LOCKED | TTU_SPLIT_HUGE_PMD |
> -					TTU_SYNC)))
> +					TTU_SYNC | TTU_BATCH_FLUSH)))
>   		return;
>   
>   	if (folio_is_zone_device(folio) &&
Huang, Ying Feb. 8, 2023, 11:27 a.m. UTC | #3 
Zi Yan <ziy@nvidia.com> writes:

> On 6 Feb 2023, at 1:33, Huang Ying wrote:
>
>> The TLB flushing will cost quite some CPU cycles during the folio
>> migration in some situations.  For example, when migrate a folio of a
>> process with multiple active threads that run on multiple CPUs.  After
>> batching the _unmap and _move in migrate_pages(), the TLB flushing can
>> be batched easily with the existing TLB flush batching mechanism.
>> This patch implements that.
>>
>> We use the following test case to test the patch.
>>
>> On a 2-socket Intel server,
>>
>> - Run pmbench memory accessing benchmark
>>
>> - Run `migratepages` to migrate pages of pmbench between node 0 and
>>   node 1 back and forth.
>>
>> With the patch, the TLB flushing IPI reduces 99.1% during the test and
>> the number of pages migrated successfully per second increases 291.7%.
>>
>> NOTE: TLB flushing is batched only for normal folios, not for THP
>> folios.  Because the overhead of TLB flushing for THP folios is much
>> lower than that for normal folios (about 1/512 on x86 platform).
>>
>> Signed-off-by: "Huang, Ying" <ying.huang@intel.com>
>> Cc: Zi Yan <ziy@nvidia.com>
>> Cc: Yang Shi <shy828301@gmail.com>
>> Cc: Baolin Wang <baolin.wang@linux.alibaba.com>
>> Cc: Oscar Salvador <osalvador@suse.de>
>> Cc: Matthew Wilcox <willy@infradead.org>
>> Cc: Bharata B Rao <bharata@amd.com>
>> Cc: Alistair Popple <apopple@nvidia.com>
>> Cc: haoxin <xhao@linux.alibaba.com>
>> Cc: Minchan Kim <minchan@kernel.org>
>> Cc: Mike Kravetz <mike.kravetz@oracle.com>
>> Cc: Hyeonggon Yoo <42.hyeyoo@gmail.com>
>> ---
>>  mm/migrate.c |  4 +++-
>>  mm/rmap.c    | 20 +++++++++++++++++---
>>  2 files changed, 20 insertions(+), 4 deletions(-)
>>
>> diff --git a/mm/migrate.c b/mm/migrate.c
>> index 9378fa2ad4a5..ca6e2ff02a09 100644
>> --- a/mm/migrate.c
>> +++ b/mm/migrate.c
>> @@ -1230,7 +1230,7 @@ static int migrate_folio_unmap(new_page_t get_new_page, free_page_t put_new_page
>>  		/* Establish migration ptes */
>>  		VM_BUG_ON_FOLIO(folio_test_anon(src) &&
>>  			       !folio_test_ksm(src) && !anon_vma, src);
>> -		try_to_migrate(src, 0);
>> +		try_to_migrate(src, TTU_BATCH_FLUSH);
>>  		page_was_mapped = 1;
>>  	}
>>
>> @@ -1781,6 +1781,8 @@ static int migrate_pages_batch(struct list_head *from, new_page_t get_new_page,
>>  	stats->nr_thp_failed += thp_retry;
>>  	stats->nr_failed_pages += nr_retry_pages;
>>  move:
>
> Maybe a comment:
> /* Flush TLBs for all the unmapped pages */

OK.  Will do that in the next version.

Best Regards,
Huang, Ying

>> +	try_to_unmap_flush();
>> +
>>  	retry = 1;
>>  	for (pass = 0;
>>  	     pass < NR_MAX_MIGRATE_PAGES_RETRY && (retry || large_retry);
>> diff --git a/mm/rmap.c b/mm/rmap.c
>> index b616870a09be..2e125f3e462e 100644
>> --- a/mm/rmap.c
>> +++ b/mm/rmap.c
>> @@ -1976,7 +1976,21 @@ static bool try_to_migrate_one(struct folio *folio, struct vm_area_struct *vma,
>>  		} else {
>>  			flush_cache_page(vma, address, pte_pfn(*pvmw.pte));
>>  			/* Nuke the page table entry. */
>> -			pteval = ptep_clear_flush(vma, address, pvmw.pte);
>> +			if (should_defer_flush(mm, flags)) {
>> +				/*
>> +				 * We clear the PTE but do not flush so potentially
>> +				 * a remote CPU could still be writing to the folio.
>> +				 * If the entry was previously clean then the
>> +				 * architecture must guarantee that a clear->dirty
>> +				 * transition on a cached TLB entry is written through
>> +				 * and traps if the PTE is unmapped.
>> +				 */
>> +				pteval = ptep_get_and_clear(mm, address, pvmw.pte);
>> +
>> +				set_tlb_ubc_flush_pending(mm, pte_dirty(pteval));
>> +			} else {
>> +				pteval = ptep_clear_flush(vma, address, pvmw.pte);
>> +			}
>>  		}
>>
>>  		/* Set the dirty flag on the folio now the pte is gone. */
>> @@ -2148,10 +2162,10 @@ void try_to_migrate(struct folio *folio, enum ttu_flags flags)
>>
>>  	/*
>>  	 * Migration always ignores mlock and only supports TTU_RMAP_LOCKED and
>> -	 * TTU_SPLIT_HUGE_PMD and TTU_SYNC flags.
>> +	 * TTU_SPLIT_HUGE_PMD, TTU_SYNC, and TTU_BATCH_FLUSH flags.
>>  	 */
>>  	if (WARN_ON_ONCE(flags & ~(TTU_RMAP_LOCKED | TTU_SPLIT_HUGE_PMD |
>> -					TTU_SYNC)))
>> +					TTU_SYNC | TTU_BATCH_FLUSH)))
>>  		return;
>>
>>  	if (folio_is_zone_device(folio) &&
>> -- 
>> 2.35.1
>
> Everything else looks good to me. Reviewed-by: Zi Yan <ziy@nvidia.com>
>
> --
> Best Regards,
> Yan, Zi
diff mbox series

Patch

diff --git a/mm/migrate.c b/mm/migrate.c
index 9378fa2ad4a5..ca6e2ff02a09 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -1230,7 +1230,7 @@  static int migrate_folio_unmap(new_page_t get_new_page, free_page_t put_new_page
 		/* Establish migration ptes */
 		VM_BUG_ON_FOLIO(folio_test_anon(src) &&
 			       !folio_test_ksm(src) && !anon_vma, src);
-		try_to_migrate(src, 0);
+		try_to_migrate(src, TTU_BATCH_FLUSH);
 		page_was_mapped = 1;
 	}
 
@@ -1781,6 +1781,8 @@  static int migrate_pages_batch(struct list_head *from, new_page_t get_new_page,
 	stats->nr_thp_failed += thp_retry;
 	stats->nr_failed_pages += nr_retry_pages;
 move:
+	try_to_unmap_flush();
+
 	retry = 1;
 	for (pass = 0;
 	     pass < NR_MAX_MIGRATE_PAGES_RETRY && (retry || large_retry);
diff --git a/mm/rmap.c b/mm/rmap.c
index b616870a09be..2e125f3e462e 100644
--- a/mm/rmap.c
+++ b/mm/rmap.c
@@ -1976,7 +1976,21 @@  static bool try_to_migrate_one(struct folio *folio, struct vm_area_struct *vma,
 		} else {
 			flush_cache_page(vma, address, pte_pfn(*pvmw.pte));
 			/* Nuke the page table entry. */
-			pteval = ptep_clear_flush(vma, address, pvmw.pte);
+			if (should_defer_flush(mm, flags)) {
+				/*
+				 * We clear the PTE but do not flush so potentially
+				 * a remote CPU could still be writing to the folio.
+				 * If the entry was previously clean then the
+				 * architecture must guarantee that a clear->dirty
+				 * transition on a cached TLB entry is written through
+				 * and traps if the PTE is unmapped.
+				 */
+				pteval = ptep_get_and_clear(mm, address, pvmw.pte);
+
+				set_tlb_ubc_flush_pending(mm, pte_dirty(pteval));
+			} else {
+				pteval = ptep_clear_flush(vma, address, pvmw.pte);
+			}
 		}
 
 		/* Set the dirty flag on the folio now the pte is gone. */
@@ -2148,10 +2162,10 @@  void try_to_migrate(struct folio *folio, enum ttu_flags flags)
 
 	/*
 	 * Migration always ignores mlock and only supports TTU_RMAP_LOCKED and
-	 * TTU_SPLIT_HUGE_PMD and TTU_SYNC flags.
+	 * TTU_SPLIT_HUGE_PMD, TTU_SYNC, and TTU_BATCH_FLUSH flags.
 	 */
 	if (WARN_ON_ONCE(flags & ~(TTU_RMAP_LOCKED | TTU_SPLIT_HUGE_PMD |
-					TTU_SYNC)))
+					TTU_SYNC | TTU_BATCH_FLUSH)))
 		return;
 
 	if (folio_is_zone_device(folio) &&