Re: [PATCH v2] mm/mempolicy,hugetlb: Checking hstate for hugetlbfs page in vma_migratable

["Li Xinhai" <lixinhai.lxh@gmail.com>]

On 2020-01-15 at 14:43 Li Xinhai wrote:
>Checking hstate at early phase when isolating page, instead of during
>unmap and move phase, to avoid useless isolation.
>
>Signed-off-by: Li Xinhai <lixinhai.lxh@gmail.com>
>Cc: Michal Hocko <mhocko@suse.com>
>Cc: Mike Kravetz <mike.kravetz@oracle.com> 

It is bad to declare the function in two files, would it be better to declare
the interface in hugetlb_inline.h as this file is created in purpose for break
circular reference between hugetlb and other components.

>---
>v1->v2:
>New function
>bool vm_hugepage_migration_supported(struct vm_area_struct *vma)
>is introduced to simplify the inter dependency of
>include/linux/mempolicy.h and include/linux/hugetlb.h, and could be
>useful for other caller.
>
> include/linux/hugetlb.h   |  2 ++
> include/linux/mempolicy.h |  6 +++---
> mm/hugetlb.c              | 10 ++++++++++
> 3 files changed, 15 insertions(+), 3 deletions(-)
>
>diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
>index 31d4920..52fc034 100644
>--- a/include/linux/hugetlb.h
>+++ b/include/linux/hugetlb.h
>@@ -834,6 +834,8 @@ static inline void set_huge_swap_pte_at(struct mm_struct *mm, unsigned long addr
> }
> #endif	/* CONFIG_HUGETLB_PAGE */
>
>+extern bool vm_hugepage_migration_supported(struct vm_area_struct *vma);
>+
> static inline spinlock_t *huge_pte_lock(struct hstate *h,
> struct mm_struct *mm, pte_t *pte)
> {
>diff --git a/include/linux/mempolicy.h b/include/linux/mempolicy.h
>index 5228c62..6637166 100644
>--- a/include/linux/mempolicy.h
>+++ b/include/linux/mempolicy.h
>@@ -172,6 +172,7 @@ int do_migrate_pages(struct mm_struct *mm, const nodemask_t *from,
>
> extern void mpol_to_str(char *buffer, int maxlen, struct mempolicy *pol);
>
>+extern bool vm_hugepage_migration_supported(struct vm_area_struct *vma);
> /* Check if a vma is migratable */
> static inline bool vma_migratable(struct vm_area_struct *vma)
> {
>@@ -185,10 +186,9 @@ static inline bool vma_migratable(struct vm_area_struct *vma)
> if (vma_is_dax(vma))
> return false;
>
>-#ifndef CONFIG_ARCH_ENABLE_HUGEPAGE_MIGRATION
>-	if (vma->vm_flags & VM_HUGETLB)
>+	if (is_vm_hugetlb_page(vma) &&
>+	!vm_hugepage_migration_supported(vma))
> return false;
>-#endif
>
> /*
> * Migration allocates pages in the highest zone. If we cannot
>diff --git a/mm/hugetlb.c b/mm/hugetlb.c
>index dd8737a..fce149c 100644
>--- a/mm/hugetlb.c
>+++ b/mm/hugetlb.c
>@@ -1316,6 +1316,16 @@ int PageHeadHuge(struct page *page_head)
> return get_compound_page_dtor(page_head) == free_huge_page;
> }
>
>+bool vm_hugepage_migration_supported(struct vm_area_struct *vma)
>+{
>+#ifdef CONFIG_HUGETLB_PAGE
>+	VM_BUG_ON(!is_vm_hugetlb_page(vma));
>+	if (hugepage_migration_supported(hstate_vma(vma)))
>+	return true;
>+#endif
>+	return false;
>+}
>+
> pgoff_t __basepage_index(struct page *page)
> {
> struct page *page_head = compound_head(page);
>--
>1.8.3.1
>