[PATCH v3 1/2] mm: protect free_pgtables with mmap_lock write lock in exit_mmap

* [PATCH v3 1/2] mm: protect free_pgtables with mmap_lock write lock in exit_mmap
@ 2021-12-07 21:50 Suren Baghdasaryan
  2021-12-07 21:50 ` [PATCH v3 2/2] mm/oom_kill: allow process_mrelease to run under mmap_lock protection Suren Baghdasaryan
  2021-12-07 22:07 ` [PATCH v3 1/2] mm: protect free_pgtables with mmap_lock write lock in exit_mmap Matthew Wilcox
  0 siblings, 2 replies; 13+ messages in thread
From: Suren Baghdasaryan @ 2021-12-07 21:50 UTC (permalink / raw)
  To: akpm
  Cc: mhocko, mhocko, rientjes, willy, hannes, guro, riel, minchan,
	kirill, aarcange, christian, hch, oleg, david, jannh, shakeelb,
	luto, christian.brauner, fweimer, jengelh, timmurray, linux-mm,
	linux-kernel, kernel-team, surenb

oom-reaper and process_mrelease system call should protect against
races with exit_mmap which can destroy page tables while they
walk the VMA tree. oom-reaper protects from that race by setting
MMF_OOM_VICTIM and by relying on exit_mmap to set MMF_OOM_SKIP
before taking and releasing mmap_write_lock. process_mrelease has
to elevate mm->mm_users to prevent such race. Both oom-reaper and
process_mrelease hold mmap_read_lock when walking the VMA tree.
The locking rules and mechanisms could be simpler if exit_mmap takes
mmap_write_lock while executing destructive operations such as
free_pgtables.
Change exit_mmap to hold the mmap_write_lock when calling
free_pgtables and remove_vma. Operations like unmap_vmas and
unlock_range are not destructive and could run under mmap_read_lock
but for simplicity we take one mmap_write_lock during almost the entire
operation. Note also that because oom-reaper checks VM_LOCKED flag,
unlock_range() should not be allowed to race with it.
Before this patch, remove_vma used to be called with no locks held,
however with fput being executed asynchronously and vm_ops->close
not being allowed to hold mmap_lock (it is called from __split_vma
with mmap_sem held for write), changing that should be fine.
In most cases this lock should be uncontended. Previously, Kirill
reported ~4% regression caused by a similar change [1]. We reran the
same test and although the individual results are quite noisy, the
percentiles show lower regression with 1.6% being the worst case [2].
The change allows oom-reaper and process_mrelease to execute safely
under mmap_read_lock without worries that exit_mmap might destroy page
tables from under them.

[1] https://lore.kernel.org/all/20170725141723.ivukwhddk2voyhuc@node.shutemov.name/
[2] https://lore.kernel.org/all/CAJuCfpGC9-c9P40x7oy=jy5SphMcd0o0G_6U1-+JAziGKG6dGA@mail.gmail.com/

Signed-off-by: Suren Baghdasaryan <surenb@google.com>
---
changes in v3
- Amended patch description to explain why remove_vma can be called while
holding mmap_write_lock, per Michal Hocko
- Added a comment for vm_operations_struct::close, documenting restriction for
taking mmap_lock in the callback, per Michal Hocko and Matthew Wilcox

 include/linux/mm.h |  4 ++++
 mm/mmap.c          | 16 ++++++++--------
 2 files changed, 12 insertions(+), 8 deletions(-)

diff --git a/include/linux/mm.h b/include/linux/mm.h
index a7e4a9e7d807..97e1a05c3b2c 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -577,6 +577,10 @@ enum page_entry_size {
  */
 struct vm_operations_struct {
 	void (*open)(struct vm_area_struct * area);
+    /*
+     * Called with mmap_lock lock held for write from __split_vma and
+     * remove_vma, therefore should never take that lock.
+     */
 	void (*close)(struct vm_area_struct * area);
 	/* Called any time before splitting to check if it's allowed */
 	int (*may_split)(struct vm_area_struct *area, unsigned long addr);
diff --git a/mm/mmap.c b/mm/mmap.c
index bfb0ea164a90..f4e09d390a07 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -3142,25 +3142,27 @@ void exit_mmap(struct mm_struct *mm)
 		 * to mmu_notifier_release(mm) ensures mmu notifier callbacks in
 		 * __oom_reap_task_mm() will not block.
 		 *
-		 * This needs to be done before calling munlock_vma_pages_all(),
+		 * This needs to be done before calling unlock_range(),
 		 * which clears VM_LOCKED, otherwise the oom reaper cannot
 		 * reliably test it.
 		 */
 		(void)__oom_reap_task_mm(mm);
 
 		set_bit(MMF_OOM_SKIP, &mm->flags);
-		mmap_write_lock(mm);
-		mmap_write_unlock(mm);
 	}
 
+	mmap_write_lock(mm);
 	if (mm->locked_vm)
 		unlock_range(mm->mmap, ULONG_MAX);
 
 	arch_exit_mmap(mm);
 
 	vma = mm->mmap;
-	if (!vma)	/* Can happen if dup_mmap() received an OOM */
+	if (!vma) {
+		/* Can happen if dup_mmap() received an OOM */
+		mmap_write_unlock(mm);
 		return;
+	}
 
 	lru_add_drain();
 	flush_cache_mm(mm);
@@ -3171,16 +3173,14 @@ void exit_mmap(struct mm_struct *mm)
 	free_pgtables(&tlb, vma, FIRST_USER_ADDRESS, USER_PGTABLES_CEILING);
 	tlb_finish_mmu(&tlb);
 
-	/*
-	 * Walk the list again, actually closing and freeing it,
-	 * with preemption enabled, without holding any MM locks.
-	 */
+	/* Walk the list again, actually closing and freeing it. */
 	while (vma) {
 		if (vma->vm_flags & VM_ACCOUNT)
 			nr_accounted += vma_pages(vma);
 		vma = remove_vma(vma);
 		cond_resched();
 	}
+	mmap_write_unlock(mm);
 	vm_unacct_memory(nr_accounted);
 }
 
-- 
2.34.1.400.ga245620fadb-goog


^ permalink raw reply related	[flat|nested] 13+ messages in thread