Commit f55e1014 authored by Linus Torvalds's avatar Linus Torvalds

Revert "mm, thp: Do not make pmd/pud dirty without a reason"

This reverts commit 152e93af.

It was a nice cleanup in theory, but as Nicolai Stange points out, we do
need to make the page dirty for the copy-on-write case even when we
didn't end up making it writable, since the dirty bit is what we use to
check that we've gone through a COW cycle.
Reported-by: default avatarMichal Hocko <mhocko@kernel.org>
Acked-by: default avatarKirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: default avatarLinus Torvalds <torvalds@linux-foundation.org>
parent 43570f03
...@@ -474,13 +474,10 @@ static int __init setup_transparent_hugepage(char *str) ...@@ -474,13 +474,10 @@ static int __init setup_transparent_hugepage(char *str)
} }
__setup("transparent_hugepage=", setup_transparent_hugepage); __setup("transparent_hugepage=", setup_transparent_hugepage);
pmd_t maybe_pmd_mkwrite(pmd_t pmd, struct vm_area_struct *vma, bool dirty) pmd_t maybe_pmd_mkwrite(pmd_t pmd, struct vm_area_struct *vma)
{ {
if (likely(vma->vm_flags & VM_WRITE)) { if (likely(vma->vm_flags & VM_WRITE))
pmd = pmd_mkwrite(pmd); pmd = pmd_mkwrite(pmd);
if (dirty)
pmd = pmd_mkdirty(pmd);
}
return pmd; return pmd;
} }
...@@ -602,7 +599,7 @@ static int __do_huge_pmd_anonymous_page(struct vm_fault *vmf, struct page *page, ...@@ -602,7 +599,7 @@ static int __do_huge_pmd_anonymous_page(struct vm_fault *vmf, struct page *page,
} }
entry = mk_huge_pmd(page, vma->vm_page_prot); entry = mk_huge_pmd(page, vma->vm_page_prot);
entry = maybe_pmd_mkwrite(entry, vma, true); entry = maybe_pmd_mkwrite(pmd_mkdirty(entry), vma);
page_add_new_anon_rmap(page, vma, haddr, true); page_add_new_anon_rmap(page, vma, haddr, true);
mem_cgroup_commit_charge(page, memcg, false, true); mem_cgroup_commit_charge(page, memcg, false, true);
lru_cache_add_active_or_unevictable(page, vma); lru_cache_add_active_or_unevictable(page, vma);
...@@ -744,8 +741,8 @@ static void insert_pfn_pmd(struct vm_area_struct *vma, unsigned long addr, ...@@ -744,8 +741,8 @@ static void insert_pfn_pmd(struct vm_area_struct *vma, unsigned long addr,
if (pfn_t_devmap(pfn)) if (pfn_t_devmap(pfn))
entry = pmd_mkdevmap(entry); entry = pmd_mkdevmap(entry);
if (write) { if (write) {
entry = pmd_mkyoung(entry); entry = pmd_mkyoung(pmd_mkdirty(entry));
entry = maybe_pmd_mkwrite(entry, vma, true); entry = maybe_pmd_mkwrite(entry, vma);
} }
if (pgtable) { if (pgtable) {
...@@ -791,14 +788,10 @@ int vmf_insert_pfn_pmd(struct vm_area_struct *vma, unsigned long addr, ...@@ -791,14 +788,10 @@ int vmf_insert_pfn_pmd(struct vm_area_struct *vma, unsigned long addr,
EXPORT_SYMBOL_GPL(vmf_insert_pfn_pmd); EXPORT_SYMBOL_GPL(vmf_insert_pfn_pmd);
#ifdef CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD #ifdef CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
static pud_t maybe_pud_mkwrite(pud_t pud, struct vm_area_struct *vma, static pud_t maybe_pud_mkwrite(pud_t pud, struct vm_area_struct *vma)
bool dirty)
{ {
if (likely(vma->vm_flags & VM_WRITE)) { if (likely(vma->vm_flags & VM_WRITE))
pud = pud_mkwrite(pud); pud = pud_mkwrite(pud);
if (dirty)
pud = pud_mkdirty(pud);
}
return pud; return pud;
} }
...@@ -814,8 +807,8 @@ static void insert_pfn_pud(struct vm_area_struct *vma, unsigned long addr, ...@@ -814,8 +807,8 @@ static void insert_pfn_pud(struct vm_area_struct *vma, unsigned long addr,
if (pfn_t_devmap(pfn)) if (pfn_t_devmap(pfn))
entry = pud_mkdevmap(entry); entry = pud_mkdevmap(entry);
if (write) { if (write) {
entry = pud_mkyoung(entry); entry = pud_mkyoung(pud_mkdirty(entry));
entry = maybe_pud_mkwrite(entry, vma, true); entry = maybe_pud_mkwrite(entry, vma);
} }
set_pud_at(mm, addr, pud, entry); set_pud_at(mm, addr, pud, entry);
update_mmu_cache_pud(vma, addr, pud); update_mmu_cache_pud(vma, addr, pud);
...@@ -1286,7 +1279,7 @@ int do_huge_pmd_wp_page(struct vm_fault *vmf, pmd_t orig_pmd) ...@@ -1286,7 +1279,7 @@ int do_huge_pmd_wp_page(struct vm_fault *vmf, pmd_t orig_pmd)
if (reuse_swap_page(page, NULL)) { if (reuse_swap_page(page, NULL)) {
pmd_t entry; pmd_t entry;
entry = pmd_mkyoung(orig_pmd); entry = pmd_mkyoung(orig_pmd);
entry = maybe_pmd_mkwrite(entry, vma, true); entry = maybe_pmd_mkwrite(pmd_mkdirty(entry), vma);
if (pmdp_set_access_flags(vma, haddr, vmf->pmd, entry, 1)) if (pmdp_set_access_flags(vma, haddr, vmf->pmd, entry, 1))
update_mmu_cache_pmd(vma, vmf->address, vmf->pmd); update_mmu_cache_pmd(vma, vmf->address, vmf->pmd);
ret |= VM_FAULT_WRITE; ret |= VM_FAULT_WRITE;
...@@ -1356,7 +1349,7 @@ int do_huge_pmd_wp_page(struct vm_fault *vmf, pmd_t orig_pmd) ...@@ -1356,7 +1349,7 @@ int do_huge_pmd_wp_page(struct vm_fault *vmf, pmd_t orig_pmd)
} else { } else {
pmd_t entry; pmd_t entry;
entry = mk_huge_pmd(new_page, vma->vm_page_prot); entry = mk_huge_pmd(new_page, vma->vm_page_prot);
entry = maybe_pmd_mkwrite(entry, vma, true); entry = maybe_pmd_mkwrite(pmd_mkdirty(entry), vma);
pmdp_huge_clear_flush_notify(vma, haddr, vmf->pmd); pmdp_huge_clear_flush_notify(vma, haddr, vmf->pmd);
page_add_new_anon_rmap(new_page, vma, haddr, true); page_add_new_anon_rmap(new_page, vma, haddr, true);
mem_cgroup_commit_charge(new_page, memcg, false, true); mem_cgroup_commit_charge(new_page, memcg, false, true);
...@@ -2935,7 +2928,7 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new) ...@@ -2935,7 +2928,7 @@ void remove_migration_pmd(struct page_vma_mapped_walk *pvmw, struct page *new)
if (pmd_swp_soft_dirty(*pvmw->pmd)) if (pmd_swp_soft_dirty(*pvmw->pmd))
pmde = pmd_mksoft_dirty(pmde); pmde = pmd_mksoft_dirty(pmde);
if (is_write_migration_entry(entry)) if (is_write_migration_entry(entry))
pmde = maybe_pmd_mkwrite(pmde, vma, false); pmde = maybe_pmd_mkwrite(pmde, vma);
flush_cache_range(vma, mmun_start, mmun_start + HPAGE_PMD_SIZE); flush_cache_range(vma, mmun_start, mmun_start + HPAGE_PMD_SIZE);
page_add_anon_rmap(new, vma, mmun_start, true); page_add_anon_rmap(new, vma, mmun_start, true);
......
...@@ -328,8 +328,7 @@ static inline void mlock_migrate_page(struct page *newpage, struct page *page) ...@@ -328,8 +328,7 @@ static inline void mlock_migrate_page(struct page *newpage, struct page *page)
} }
} }
extern pmd_t maybe_pmd_mkwrite(pmd_t pmd, struct vm_area_struct *vma, extern pmd_t maybe_pmd_mkwrite(pmd_t pmd, struct vm_area_struct *vma);
bool dirty);
/* /*
* At what user virtual address is page expected in @vma? * At what user virtual address is page expected in @vma?
......
...@@ -1057,7 +1057,7 @@ static void collapse_huge_page(struct mm_struct *mm, ...@@ -1057,7 +1057,7 @@ static void collapse_huge_page(struct mm_struct *mm,
pgtable = pmd_pgtable(_pmd); pgtable = pmd_pgtable(_pmd);
_pmd = mk_huge_pmd(new_page, vma->vm_page_prot); _pmd = mk_huge_pmd(new_page, vma->vm_page_prot);
_pmd = maybe_pmd_mkwrite(_pmd, vma, false); _pmd = maybe_pmd_mkwrite(pmd_mkdirty(_pmd), vma);
/* /*
* spin_lock() below is not the equivalent of smp_wmb(), so * spin_lock() below is not the equivalent of smp_wmb(), so
......
...@@ -3335,7 +3335,7 @@ static int do_set_pmd(struct vm_fault *vmf, struct page *page) ...@@ -3335,7 +3335,7 @@ static int do_set_pmd(struct vm_fault *vmf, struct page *page)
entry = mk_huge_pmd(page, vma->vm_page_prot); entry = mk_huge_pmd(page, vma->vm_page_prot);
if (write) if (write)
entry = maybe_pmd_mkwrite(entry, vma, true); entry = maybe_pmd_mkwrite(pmd_mkdirty(entry), vma);
add_mm_counter(vma->vm_mm, MM_FILEPAGES, HPAGE_PMD_NR); add_mm_counter(vma->vm_mm, MM_FILEPAGES, HPAGE_PMD_NR);
page_add_file_rmap(page, true); page_add_file_rmap(page, true);
......
...@@ -2068,7 +2068,7 @@ int migrate_misplaced_transhuge_page(struct mm_struct *mm, ...@@ -2068,7 +2068,7 @@ int migrate_misplaced_transhuge_page(struct mm_struct *mm,
} }
entry = mk_huge_pmd(new_page, vma->vm_page_prot); entry = mk_huge_pmd(new_page, vma->vm_page_prot);
entry = maybe_pmd_mkwrite(entry, vma, false); entry = maybe_pmd_mkwrite(pmd_mkdirty(entry), vma);
/* /*
* Clear the old entry under pagetable lock and establish the new PTE. * Clear the old entry under pagetable lock and establish the new PTE.
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment