mm: handle swap page faults under per-VMA lock

When page fault is handled under per-VMA lock protection, all swap page
faults are retried with mmap_lock because folio_lock_or_retry has to drop
and reacquire mmap_lock if folio could not be immediately locked.  Follow
the same pattern as mmap_lock to drop per-VMA lock when waiting for folio
and retrying once folio is available.

With this obstacle removed, enable do_swap_page to operate under per-VMA
lock protection.  Drivers implementing ops->migrate_to_ram might still
rely on mmap_lock, therefore we have to fall back to mmap_lock in that
particular case.

Note that the only time do_swap_page calls synchronous swap_readpage is
when SWP_SYNCHRONOUS_IO is set, which is only set for
QUEUE_FLAG_SYNCHRONOUS devices: brd, zram and nvdimms (both btt and pmem).
Therefore we don't sleep in this path, and there's no need to drop the
mmap or per-VMA lock.

Link: https://lkml.kernel.org/r/20230630211957.1341547-6-surenb@google.com
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Tested-by: Alistair Popple <apopple@nvidia.com>
Reviewed-by: Alistair Popple <apopple@nvidia.com>
Acked-by: Peter Xu <peterx@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Hillf Danton <hdanton@sina.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Josef Bacik <josef@toxicpanda.com>
Cc: Laurent Dufour <ldufour@linux.ibm.com>
Cc: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michel Lespinasse <michel@lespinasse.org>
Cc: Minchan Kim <minchan@google.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Punit Agrawal <punit.agrawal@bytedance.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

2 files changed, 18 insertions, 15 deletions

diff --git a/mm/filemap.c b/mm/filemap.c
index 40514493014a..8040545954bc 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -1671,27 +1671,26 @@ static int __folio_lock_async(struct folio *folio, struct wait_page_queue *wait)
  * Return values:
  * 0 - folio is locked.
  * non-zero - folio is not locked.
- *     mmap_lock has been released (mmap_read_unlock(), unless flags had both
- *     FAULT_FLAG_ALLOW_RETRY and FAULT_FLAG_RETRY_NOWAIT set, in
- *     which case mmap_lock is still held.
+ *     mmap_lock or per-VMA lock has been released (mmap_read_unlock() or
+ *     vma_end_read()), unless flags had both FAULT_FLAG_ALLOW_RETRY and
+ *     FAULT_FLAG_RETRY_NOWAIT set, in which case the lock is still held.
  *
  * If neither ALLOW_RETRY nor KILLABLE are set, will always return 0
- * with the folio locked and the mmap_lock unperturbed.
+ * with the folio locked and the mmap_lock/per-VMA lock is left unperturbed.
  */
 vm_fault_t __folio_lock_or_retry(struct folio *folio, struct vm_fault *vmf)
 {
-	struct mm_struct *mm = vmf->vma->vm_mm;
 	unsigned int flags = vmf->flags;
 
 	if (fault_flag_allow_retry_first(flags)) {
 		/*
-		 * CAUTION! In this case, mmap_lock is not released
-		 * even though return VM_FAULT_RETRY.
+		 * CAUTION! In this case, mmap_lock/per-VMA lock is not
+		 * released even though returning VM_FAULT_RETRY.
 		 */
 		if (flags & FAULT_FLAG_RETRY_NOWAIT)
 			return VM_FAULT_RETRY;
 
-		mmap_read_unlock(mm);
+		release_fault_lock(vmf);
 		if (flags & FAULT_FLAG_KILLABLE)
 			folio_wait_locked_killable(folio);
 		else
@@ -1703,7 +1702,7 @@ vm_fault_t __folio_lock_or_retry(struct folio *folio, struct vm_fault *vmf)
 
 		ret = __folio_lock_killable(folio);
 		if (ret) {
-			mmap_read_unlock(mm);
+			release_fault_lock(vmf);
 			return VM_FAULT_RETRY;
 		}
 	} else {
diff --git a/mm/memory.c b/mm/memory.c
index 080e1d59d752..5748a41c164c 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -3746,12 +3746,6 @@ vm_fault_t do_swap_page(struct vm_fault *vmf)
 	if (!pte_unmap_same(vmf))
 		goto out;
 
-	if (vmf->flags & FAULT_FLAG_VMA_LOCK) {
-		ret = VM_FAULT_RETRY;
-		vma_end_read(vma);
-		goto out;
-	}
-
 	entry = pte_to_swp_entry(vmf->orig_pte);
 	if (unlikely(non_swap_entry(entry))) {
 		if (is_migration_entry(entry)) {
@@ -3761,6 +3755,16 @@ vm_fault_t do_swap_page(struct vm_fault *vmf)
 			vmf->page = pfn_swap_entry_to_page(entry);
 			ret = remove_device_exclusive_entry(vmf);
 		} else if (is_device_private_entry(entry)) {
+			if (vmf->flags & FAULT_FLAG_VMA_LOCK) {
+				/*
+				 * migrate_to_ram is not yet ready to operate
+				 * under VMA lock.
+				 */
+				vma_end_read(vma);
+				ret = VM_FAULT_RETRY;
+				goto out;
+			}
+
 			vmf->page = pfn_swap_entry_to_page(entry);
 			vmf->pte = pte_offset_map_lock(vma->vm_mm, vmf->pmd,
 					vmf->address, &vmf->ptl);