[PATCH V2] mm/thp/migration: switch from flush_tlb_range to flush_pmd_tlb_range

[PATCH V2] mm/thp/migration: switch from flush_tlb_range to flush_pmd_tlb_range

[Aneesh Kumar K.V]

We remove one instace of flush_tlb_range here. That was added by
f714f4f20e59ea6eea264a86b9a51fd51b88fc54 ("mm: numa: call MMU notifiers
on THP migration"). But the pmdp_huge_clear_flush_notify should have
done the require flush for us. Hence remove the extra flush.

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
---
Changes from V1:
* fix build error

 include/asm-generic/pgtable.h | 17 +++++++++++++++++
 mm/migrate.c                  |  8 +++++---
 mm/pgtable-generic.c          | 14 --------------
 3 files changed, 22 insertions(+), 17 deletions(-)

diff --git a/include/asm-generic/pgtable.h b/include/asm-generic/pgtable.h
index c370b261c720..9401f4819891 100644
--- a/include/asm-generic/pgtable.h
+++ b/include/asm-generic/pgtable.h
@@ -783,6 +783,23 @@ static inline int pmd_clear_huge(pmd_t *pmd)
 }
 #endif	/* CONFIG_HAVE_ARCH_HUGE_VMAP */
 
+#ifndef __HAVE_ARCH_FLUSH_PMD_TLB_RANGE
+#ifdef CONFIG_TRANSPARENT_HUGEPAGE
+/*
+ * ARCHes with special requirements for evicting THP backing TLB entries can
+ * implement this. Otherwise also, it can help optimize normal TLB flush in
+ * THP regime. stock flush_tlb_range() typically has optimization to nuke the
+ * entire TLB TLB if flush span is greater than a threshold, which will
+ * likely be true for a single huge page. Thus a single thp flush will
+ * invalidate the entire TLB which is not desitable.
+ * e.g. see arch/arc: flush_pmd_tlb_range
+ */
+#define flush_pmd_tlb_range(vma, addr, end)	flush_tlb_range(vma, addr, end)
+#else
+#define flush_pmd_tlb_range(vma, addr, end)	BUILD_BUG()
+#endif
+#endif
+
 #endif /* !__ASSEMBLY__ */
 
 #ifndef io_remap_pfn_range
diff --git a/mm/migrate.c b/mm/migrate.c
index b1034f9c77e7..c079c115d038 100644
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -1767,7 +1767,10 @@ int migrate_misplaced_transhuge_page(struct mm_struct *mm,
 		put_page(new_page);
 		goto out_fail;
 	}
-
+	/*
+	 * We are not sure a pending tlb flush here is for a huge page
+	 * mapping or not. Hence use the tlb range variant
+	 */
 	if (mm_tlb_flush_pending(mm))
 		flush_tlb_range(vma, mmun_start, mmun_end);
 
@@ -1823,12 +1826,11 @@ fail_putback:
 	page_add_anon_rmap(new_page, vma, mmun_start, true);
 	pmdp_huge_clear_flush_notify(vma, mmun_start, pmd);
 	set_pmd_at(mm, mmun_start, pmd, entry);
-	flush_tlb_range(vma, mmun_start, mmun_end);
 	update_mmu_cache_pmd(vma, address, &entry);
 
 	if (page_count(page) != 2) {
 		set_pmd_at(mm, mmun_start, pmd, orig_entry);
-		flush_tlb_range(vma, mmun_start, mmun_end);
+		flush_pmd_tlb_range(vma, mmun_start, mmun_end);
 		mmu_notifier_invalidate_range(mm, mmun_start, mmun_end);
 		update_mmu_cache_pmd(vma, address, &entry);
 		page_remove_rmap(new_page, true);
diff --git a/mm/pgtable-generic.c b/mm/pgtable-generic.c
index 9d4767698a1c..3c9c78400300 100644
--- a/mm/pgtable-generic.c
+++ b/mm/pgtable-generic.c
@@ -84,20 +84,6 @@ pte_t ptep_clear_flush(struct vm_area_struct *vma, unsigned long address,
 
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
 
-#ifndef __HAVE_ARCH_FLUSH_PMD_TLB_RANGE
-
-/*
- * ARCHes with special requirements for evicting THP backing TLB entries can
- * implement this. Otherwise also, it can help optimize normal TLB flush in
- * THP regime. stock flush_tlb_range() typically has optimization to nuke the
- * entire TLB TLB if flush span is greater than a threshhold, which will
- * likely be true for a single huge page. Thus a single thp flush will
- * invalidate the entire TLB which is not desitable.
- * e.g. see arch/arc: flush_pmd_tlb_range
- */
-#define flush_pmd_tlb_range(vma, addr, end)	flush_tlb_range(vma, addr, end)
-#endif
-
 #ifndef __HAVE_ARCH_PMDP_SET_ACCESS_FLAGS
 int pmdp_set_access_flags(struct vm_area_struct *vma,
 			  unsigned long address, pmd_t *pmdp,
-- 
2.5.0

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH V2] mm/thp/migration: switch from flush_tlb_range to flush_pmd_tlb_range

[Aneesh Kumar K.V]

"Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com> writes:

> We remove one instace of flush_tlb_range here. That was added by
> f714f4f20e59ea6eea264a86b9a51fd51b88fc54 ("mm: numa: call MMU notifiers
> on THP migration"). But the pmdp_huge_clear_flush_notify should have
> done the require flush for us. Hence remove the extra flush.
>
> Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
> ---
> Changes from V1:
> * fix build error
>
>  include/asm-generic/pgtable.h | 17 +++++++++++++++++
>  mm/migrate.c                  |  8 +++++---
>  mm/pgtable-generic.c          | 14 --------------
>  3 files changed, 22 insertions(+), 17 deletions(-)
>
> diff --git a/include/asm-generic/pgtable.h b/include/asm-generic/pgtable.h
> index c370b261c720..9401f4819891 100644
> --- a/include/asm-generic/pgtable.h
> +++ b/include/asm-generic/pgtable.h
> @@ -783,6 +783,23 @@ static inline int pmd_clear_huge(pmd_t *pmd)
>  }
>  #endif	/* CONFIG_HAVE_ARCH_HUGE_VMAP */
>
> +#ifndef __HAVE_ARCH_FLUSH_PMD_TLB_RANGE
> +#ifdef CONFIG_TRANSPARENT_HUGEPAGE
> +/*
> + * ARCHes with special requirements for evicting THP backing TLB entries can
> + * implement this. Otherwise also, it can help optimize normal TLB flush in
> + * THP regime. stock flush_tlb_range() typically has optimization to nuke the
> + * entire TLB TLB if flush span is greater than a threshold, which will
> + * likely be true for a single huge page. Thus a single thp flush will
> + * invalidate the entire TLB which is not desitable.
> + * e.g. see arch/arc: flush_pmd_tlb_range
> + */
> +#define flush_pmd_tlb_range(vma, addr, end)	flush_tlb_range(vma, addr, end)
> +#else
> +#define flush_pmd_tlb_range(vma, addr, end)	BUILD_BUG()
> +#endif
> +#endif
> +
>  #endif /* !__ASSEMBLY__ */
>
>  #ifndef io_remap_pfn_range
> diff --git a/mm/migrate.c b/mm/migrate.c
> index b1034f9c77e7..c079c115d038 100644
> --- a/mm/migrate.c
> +++ b/mm/migrate.c
> @@ -1767,7 +1767,10 @@ int migrate_misplaced_transhuge_page(struct mm_struct *mm,
>  		put_page(new_page);
>  		goto out_fail;
>  	}
> -
> +	/*
> +	 * We are not sure a pending tlb flush here is for a huge page
> +	 * mapping or not. Hence use the tlb range variant
> +	 */
>  	if (mm_tlb_flush_pending(mm))
>  		flush_tlb_range(vma, mmun_start, mmun_end);
>

I was thinking we should be able to switch this flush_pmd_tlb_range. But
Kirill was not sure when we discussed this last time. Can we have a
pending tlb flush with PAGE_SIZE page translation, when we are parallely
trying to handle a autonuma fault on that ?


> @@ -1823,12 +1826,11 @@ fail_putback:
>  	page_add_anon_rmap(new_page, vma, mmun_start, true);
>  	pmdp_huge_clear_flush_notify(vma, mmun_start, pmd);
>  	set_pmd_at(mm, mmun_start, pmd, entry);
> -	flush_tlb_range(vma, mmun_start, mmun_end);
>  	update_mmu_cache_pmd(vma, address, &entry);
>
>  	if (page_count(page) != 2) {
>  		set_pmd_at(mm, mmun_start, pmd, orig_entry);
> -		flush_tlb_range(vma, mmun_start, mmun_end);
> +		flush_pmd_tlb_range(vma, mmun_start, mmun_end);
>  		mmu_notifier_invalidate_range(mm, mmun_start, mmun_end);
>  		update_mmu_cache_pmd(vma, address, &entry);
>  		page_remove_rmap(new_page, true);
> diff --git a/mm/pgtable-generic.c b/mm/pgtable-generic.c
> index 9d4767698a1c..3c9c78400300 100644
> --- a/mm/pgtable-generic.c
> +++ b/mm/pgtable-generic.c
> @@ -84,20 +84,6 @@ pte_t ptep_clear_flush(struct vm_area_struct *vma, unsigned long address,
>
>  #ifdef CONFIG_TRANSPARENT_HUGEPAGE
>
> -#ifndef __HAVE_ARCH_FLUSH_PMD_TLB_RANGE
> -
> -/*
> - * ARCHes with special requirements for evicting THP backing TLB entries can
> - * implement this. Otherwise also, it can help optimize normal TLB flush in
> - * THP regime. stock flush_tlb_range() typically has optimization to nuke the
> - * entire TLB TLB if flush span is greater than a threshhold, which will
> - * likely be true for a single huge page. Thus a single thp flush will
> - * invalidate the entire TLB which is not desitable.
> - * e.g. see arch/arc: flush_pmd_tlb_range
> - */
> -#define flush_pmd_tlb_range(vma, addr, end)	flush_tlb_range(vma, addr, end)
> -#endif
> -
>  #ifndef __HAVE_ARCH_PMDP_SET_ACCESS_FLAGS
>  int pmdp_set_access_flags(struct vm_area_struct *vma,
>  			  unsigned long address, pmd_t *pmdp,
> -- 
> 2.5.0

-aneesh

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

ARC !THP broken in linux-next (was Re: [PATCH V2] mm/thp/migration: switch from flush_tlb_range to flush_pmd_tlb_range)

[Vineet Gupta]

Hi Aneesh,

On Wednesday 10 February 2016 09:05 PM, Aneesh Kumar K.V wrote:
> We remove one instace of flush_tlb_range here. That was added by
> f714f4f20e59ea6eea264a86b9a51fd51b88fc54 ("mm: numa: call MMU notifiers
> on THP migration"). But the pmdp_huge_clear_flush_notify should have
> done the require flush for us. Hence remove the extra flush.
> 
> Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>

linux-next fails for ARC with this - although this is not your fault.
Back then, per your request I did verify that ARC builds fine with this for
CONFIG_TRANSPARENT_HUGEPAGE worked. However I failed to check the !THP case which
is broken.

@Andrew could you please add the patch below to mm tree !

Thx,
Vineet

------------>

Re: ARC !THP broken in linux-next (was Re: [PATCH V2] mm/thp/migration: switch from flush_tlb_range to flush_pmd_tlb_range)

[Vineet Gupta]

On Thursday 17 March 2016 03:06 PM, Vineet Gupta wrote:
> @Andrew could you please add the patch below to mm tree !

Never mind - I pushed it to my tree as this fix has no real dependency on original
patch and seems to be the right things with or w/o it.

Thx,
-Vineet

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>