mm: account pud page tables

On a machine with 5-level paging support a process can allocate
significant amount of memory and stay unnoticed by oom-killer and memory
cgroup.  The trick is to allocate a lot of PUD page tables.  We don't
account PUD page tables, only PMD and PTE.

We already addressed the same issue for PMD page tables, see commit
dc6c9a35 ("mm: account pmd page tables to the process").
Introduction of 5-level paging brings the same issue for PUD page
tables.

The patch expands accounting to PUD level.

[kirill.shutemov@linux.intel.com: s/pmd_t/pud_t/]
  Link: http://lkml.kernel.org/r/20171004074305.x35eh5u7ybbt5kar@black.fi.intel.com
[heiko.carstens@de.ibm.com: s390/mm: fix pud table accounting]
  Link: http://lkml.kernel.org/r/20171103090551.18231-1-heiko.carstens@de.ibm.com
Link: http://lkml.kernel.org/r/20171002080427.3320-1-kirill.shutemov@linux.intel.comSigned-off-by: NKirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Acked-by: NRik van Riel <riel@redhat.com>
Acked-by: NMichal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Documentation/sysctl/vm.txt

@@ -629,10 +629,10 @@ oom_dump_tasks
 
 Enables a system-wide task dump (excluding kernel threads) to be produced
 when the kernel performs an OOM-killing and includes such information as
-pid, uid, tgid, vm size, rss, nr_ptes, nr_pmds, swapents, oom_score_adj
-score, and name.  This is helpful to determine why the OOM killer was
-invoked, to identify the rogue task that caused it, and to determine why
-the OOM killer chose the task it did to kill.
+pid, uid, tgid, vm size, rss, nr_ptes, nr_pmds, nr_puds, swapents,
+oom_score_adj score, and name.  This is helpful to determine why the OOM
+killer was invoked, to identify the rogue task that caused it, and to
+determine why the OOM killer chose the task it did to kill.
 
 If this is set to zero, this information is suppressed.  On very
 large systems with thousands of tasks it may not be feasible to dump

arch/powerpc/mm/hugetlbpage.c

@@ -433,6 +433,7 @@ static void hugetlb_free_pud_range(struct mmu_gather *tlb, pgd_t *pgd,
 	pud = pud_offset(pgd, start);
 	pgd_clear(pgd);
 	pud_free_tlb(tlb, pud, start);
+	mm_dec_nr_puds(tlb->mm);
 }
 
 /*

arch/s390/include/asm/mmu_context.h

@@ -44,6 +44,8 @@ static inline int init_new_context(struct task_struct *tsk,
 		mm->context.asce_limit = STACK_TOP_MAX;
 		mm->context.asce = __pa(mm->pgd) | _ASCE_TABLE_LENGTH |
 				   _ASCE_USER_BITS | _ASCE_TYPE_REGION3;
+		/* pgd_alloc() did not account this pud */
+		mm_inc_nr_puds(mm);
 		break;
 	case -PAGE_SIZE:
 		/* forked 5-level task, set new asce with new_mm->pgd */
@@ -59,7 +61,7 @@ static inline int init_new_context(struct task_struct *tsk,
 		/* forked 2-level compat task, set new asce with new mm->pgd */
 		mm->context.asce = __pa(mm->pgd) | _ASCE_TABLE_LENGTH |
 				   _ASCE_USER_BITS | _ASCE_TYPE_SEGMENT;
-		/* pgd_alloc() did not increase mm->nr_pmds */
+		/* pgd_alloc() did not account this pmd */
 		mm_inc_nr_pmds(mm);
 	}
 	crst_table_init((unsigned long *) mm->pgd, pgd_entry_type(mm));

arch/sparc/mm/hugetlbpage.c

@@ -472,6 +472,7 @@ static void hugetlb_free_pud_range(struct mmu_gather *tlb, pgd_t *pgd,
 	pud = pud_offset(pgd, start);
 	pgd_clear(pgd);
 	pud_free_tlb(tlb, pud, start);
+	mm_dec_nr_puds(tlb->mm);
 }
 
 void hugetlb_free_pgd_range(struct mmu_gather *tlb,

fs/proc/task_mmu.c

@@ -26,7 +26,7 @@
 
 void task_mem(struct seq_file *m, struct mm_struct *mm)
 {
-	unsigned long text, lib, swap, ptes, pmds, anon, file, shmem;
+	unsigned long text, lib, swap, ptes, pmds, puds, anon, file, shmem;
 	unsigned long hiwater_vm, total_vm, hiwater_rss, total_rss;
 
 	anon = get_mm_counter(mm, MM_ANONPAGES);
@@ -52,6 +52,7 @@ void task_mem(struct seq_file *m, struct mm_struct *mm)
 	swap = get_mm_counter(mm, MM_SWAPENTS);
 	ptes = PTRS_PER_PTE * sizeof(pte_t) * atomic_long_read(&mm->nr_ptes);
 	pmds = PTRS_PER_PMD * sizeof(pmd_t) * mm_nr_pmds(mm);
+	puds = PTRS_PER_PUD * sizeof(pud_t) * mm_nr_puds(mm);
 	seq_printf(m,
 		"VmPeak:\t%8lu kB\n"
 		"VmSize:\t%8lu kB\n"
@@ -68,6 +69,7 @@ void task_mem(struct seq_file *m, struct mm_struct *mm)
 		"VmLib:\t%8lu kB\n"
 		"VmPTE:\t%8lu kB\n"
 		"VmPMD:\t%8lu kB\n"
+		"VmPUD:\t%8lu kB\n"
 		"VmSwap:\t%8lu kB\n",
 		hiwater_vm << (PAGE_SHIFT-10),
 		total_vm << (PAGE_SHIFT-10),
@@ -82,6 +84,7 @@ void task_mem(struct seq_file *m, struct mm_struct *mm)
 		mm->stack_vm << (PAGE_SHIFT-10), text, lib,
 		ptes >> 10,
 		pmds >> 10,
+		puds >> 10,
 		swap << (PAGE_SHIFT-10));
 	hugetlb_report_usage(m, mm);
 }

include/linux/mm.h

@@ -1599,14 +1599,44 @@ static inline int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd,
 int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
 #endif
 
-#ifdef __PAGETABLE_PUD_FOLDED
+#if defined(__PAGETABLE_PUD_FOLDED) || !defined(CONFIG_MMU)
 static inline int __pud_alloc(struct mm_struct *mm, p4d_t *p4d,
 						unsigned long address)
 {
 	return 0;
 }
+
+static inline unsigned long mm_nr_puds(const struct mm_struct *mm)
+{
+	return 0;
+}
+
+static inline void mm_nr_puds_init(struct mm_struct *mm) {}
+static inline void mm_inc_nr_puds(struct mm_struct *mm) {}
+static inline void mm_dec_nr_puds(struct mm_struct *mm) {}
+
 #else
 int __pud_alloc(struct mm_struct *mm, p4d_t *p4d, unsigned long address);
+
+static inline void mm_nr_puds_init(struct mm_struct *mm)
+{
+	atomic_long_set(&mm->nr_puds, 0);
+}
+
+static inline unsigned long mm_nr_puds(const struct mm_struct *mm)
+{
+	return atomic_long_read(&mm->nr_puds);
+}
+
+static inline void mm_inc_nr_puds(struct mm_struct *mm)
+{
+	atomic_long_inc(&mm->nr_puds);
+}
+
+static inline void mm_dec_nr_puds(struct mm_struct *mm)
+{
+	atomic_long_dec(&mm->nr_puds);
+}
 #endif
 
 #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU)
@@ -1618,7 +1648,7 @@ static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
 
 static inline void mm_nr_pmds_init(struct mm_struct *mm) {}
 
-static inline unsigned long mm_nr_pmds(struct mm_struct *mm)
+static inline unsigned long mm_nr_pmds(const struct mm_struct *mm)
 {
 	return 0;
 }
@@ -1634,7 +1664,7 @@ static inline void mm_nr_pmds_init(struct mm_struct *mm)
 	atomic_long_set(&mm->nr_pmds, 0);
 }
 
-static inline unsigned long mm_nr_pmds(struct mm_struct *mm)
+static inline unsigned long mm_nr_pmds(const struct mm_struct *mm)
 {
 	return atomic_long_read(&mm->nr_pmds);
 }

include/linux/mm_types.h

@@ -404,6 +404,9 @@ struct mm_struct {
 	atomic_long_t nr_ptes;			/* PTE page table pages */
 #if CONFIG_PGTABLE_LEVELS > 2
 	atomic_long_t nr_pmds;			/* PMD page table pages */
+#endif
+#if CONFIG_PGTABLE_LEVELS > 3
+	atomic_long_t nr_puds;			/* PUD page table pages */
 #endif
 	int map_count;				/* number of VMAs */
 

kernel/fork.c

@@ -819,6 +819,7 @@ static struct mm_struct *mm_init(struct mm_struct *mm, struct task_struct *p,
 	mm->core_state = NULL;
 	atomic_long_set(&mm->nr_ptes, 0);
 	mm_nr_pmds_init(mm);
+	mm_nr_puds_init(mm);
 	mm->map_count = 0;
 	mm->locked_vm = 0;
 	mm->pinned_vm = 0;
@@ -878,6 +879,9 @@ static void check_mm(struct mm_struct *mm)
 	if (mm_nr_pmds(mm))
 		pr_alert("BUG: non-zero nr_pmds on freeing mm: %ld\n",
 				mm_nr_pmds(mm));
+	if (mm_nr_puds(mm))
+		pr_alert("BUG: non-zero nr_puds on freeing mm: %ld\n",
+				mm_nr_puds(mm));
 
 #if defined(CONFIG_TRANSPARENT_HUGEPAGE) && !USE_SPLIT_PMD_PTLOCKS
 	VM_BUG_ON_MM(mm->pmd_huge_pte, mm);

mm/debug.c

@@ -105,7 +105,8 @@ void dump_mm(const struct mm_struct *mm)
 		"get_unmapped_area %p\n"
 #endif
 		"mmap_base %lu mmap_legacy_base %lu highest_vm_end %lu\n"
-		"pgd %p mm_users %d mm_count %d nr_ptes %lu nr_pmds %lu map_count %d\n"
+		"pgd %p mm_users %d mm_count %d\n"
+		"nr_ptes %lu nr_pmds %lu nr_puds %lu map_count %d\n"
 		"hiwater_rss %lx hiwater_vm %lx total_vm %lx locked_vm %lx\n"
 		"pinned_vm %lx data_vm %lx exec_vm %lx stack_vm %lx\n"
 		"start_code %lx end_code %lx start_data %lx end_data %lx\n"
@@ -136,7 +137,8 @@ void dump_mm(const struct mm_struct *mm)
 		mm->pgd, atomic_read(&mm->mm_users),
 		atomic_read(&mm->mm_count),
 		atomic_long_read((atomic_long_t *)&mm->nr_ptes),
-		mm_nr_pmds((struct mm_struct *)mm),
+		mm_nr_pmds(mm),
+		mm_nr_puds(mm),
 		mm->map_count,
 		mm->hiwater_rss, mm->hiwater_vm, mm->total_vm, mm->locked_vm,
 		mm->pinned_vm, mm->data_vm, mm->exec_vm, mm->stack_vm,

mm/memory.c

@@ -506,6 +506,7 @@ static inline void free_pud_range(struct mmu_gather *tlb, p4d_t *p4d,
 	pud = pud_offset(p4d, start);
 	p4d_clear(p4d);
 	pud_free_tlb(tlb, pud, start);
+	mm_dec_nr_puds(tlb->mm);
 }
 
 static inline void free_p4d_range(struct mmu_gather *tlb, pgd_t *pgd,
@@ -4149,15 +4150,17 @@ int __pud_alloc(struct mm_struct *mm, p4d_t *p4d, unsigned long address)
 
 	spin_lock(&mm->page_table_lock);
 #ifndef __ARCH_HAS_5LEVEL_HACK
-	if (p4d_present(*p4d))		/* Another has populated it */
-		pud_free(mm, new);
-	else
+	if (!p4d_present(*p4d)) {
+		mm_inc_nr_puds(mm);
 		p4d_populate(mm, p4d, new);
-#else
-	if (pgd_present(*p4d))		/* Another has populated it */
+	} else	/* Another has populated it */
 		pud_free(mm, new);
-	else
+#else
+	if (!pgd_present(*p4d)) {
+		mm_inc_nr_puds(mm);
 		pgd_populate(mm, p4d, new);
+	} else	/* Another has populated it */
+		pud_free(mm, new);
 #endif /* __ARCH_HAS_5LEVEL_HACK */
 	spin_unlock(&mm->page_table_lock);
 	return 0;

mm/oom_kill.c

@@ -221,7 +221,8 @@ unsigned long oom_badness(struct task_struct *p, struct mem_cgroup *memcg,
 	 * task's rss, pagetable and swap space use.
 	 */
 	points = get_mm_rss(p->mm) + get_mm_counter(p->mm, MM_SWAPENTS) +
-		atomic_long_read(&p->mm->nr_ptes) + mm_nr_pmds(p->mm);
+		atomic_long_read(&p->mm->nr_ptes) + mm_nr_pmds(p->mm) +
+		mm_nr_puds(p->mm);
 	task_unlock(p);
 
 	/*
@@ -397,7 +398,7 @@ static void dump_tasks(struct mem_cgroup *memcg, const nodemask_t *nodemask)
 	struct task_struct *p;
 	struct task_struct *task;
 
-	pr_info("[ pid ]   uid  tgid total_vm      rss nr_ptes nr_pmds swapents oom_score_adj name\n");
+	pr_info("[ pid ]   uid  tgid total_vm      rss nr_ptes nr_pmds nr_puds swapents oom_score_adj name\n");
 	rcu_read_lock();
 	for_each_process(p) {
 		if (oom_unkillable_task(p, memcg, nodemask))
@@ -413,11 +414,12 @@ static void dump_tasks(struct mem_cgroup *memcg, const nodemask_t *nodemask)
 			continue;
 		}
 
-		pr_info("[%5d] %5d %5d %8lu %8lu %7ld %7ld %8lu         %5hd %s\n",
+		pr_info("[%5d] %5d %5d %8lu %8lu %7ld %7ld %7ld %8lu         %5hd %s\n",
 			task->pid, from_kuid(&init_user_ns, task_uid(task)),
 			task->tgid, task->mm->total_vm, get_mm_rss(task->mm),
 			atomic_long_read(&task->mm->nr_ptes),
 			mm_nr_pmds(task->mm),
+			mm_nr_puds(task->mm),
 			get_mm_counter(task->mm, MM_SWAPENTS),
 			task->signal->oom_score_adj, task->comm);
 		task_unlock(task);