1 files changed, 87 insertions, 59 deletions

diff --git a/fs/buffer.c b/fs/buffer.c
index c7abb4a029dc..ead4dc85debd 100644
--- a/fs/buffer.c
+++ b/fs/buffer.c
@@ -176,18 +176,8 @@ void end_buffer_write_sync(struct buffer_head *bh, int uptodate)
 }
 EXPORT_SYMBOL(end_buffer_write_sync);
 
-/*
- * Various filesystems appear to want __find_get_block to be non-blocking.
- * But it's the page lock which protects the buffers.  To get around this,
- * we get exclusion from try_to_free_buffers with the blockdev mapping's
- * i_private_lock.
- *
- * Hack idea: for the blockdev mapping, i_private_lock contention
- * may be quite high.  This code could TryLock the page, and if that
- * succeeds, there is no need to take i_private_lock.
- */
 static struct buffer_head *
-__find_get_block_slow(struct block_device *bdev, sector_t block)
+__find_get_block_slow(struct block_device *bdev, sector_t block, bool atomic)
 {
 	struct address_space *bd_mapping = bdev->bd_mapping;
 	const int blkbits = bd_mapping->host->i_blkbits;
@@ -204,10 +194,28 @@ __find_get_block_slow(struct block_device *bdev, sector_t block)
 	if (IS_ERR(folio))
 		goto out;
 
-	spin_lock(&bd_mapping->i_private_lock);
+	/*
+	 * Folio lock protects the buffers. Callers that cannot block
+	 * will fallback to serializing vs try_to_free_buffers() via
+	 * the i_private_lock.
+	 */
+	if (atomic)
+		spin_lock(&bd_mapping->i_private_lock);
+	else
+		folio_lock(folio);
+
 	head = folio_buffers(folio);
 	if (!head)
 		goto out_unlock;
+	/*
+	 * Upon a noref migration, the folio lock serializes here;
+	 * otherwise bail.
+	 */
+	if (test_bit_acquire(BH_Migrate, &head->b_state)) {
+		WARN_ON(!atomic);
+		goto out_unlock;
+	}
+
 	bh = head;
 	do {
 		if (!buffer_mapped(bh))
@@ -236,7 +244,10 @@ __find_get_block_slow(struct block_device *bdev, sector_t block)
 		       1 << blkbits);
 	}
 out_unlock:
-	spin_unlock(&bd_mapping->i_private_lock);
+	if (atomic)
+		spin_unlock(&bd_mapping->i_private_lock);
+	else
+		folio_unlock(folio);
 	folio_put(folio);
 out:
 	return ret;
@@ -286,7 +297,6 @@ static void end_buffer_async_read(struct buffer_head *bh, int uptodate)
 
 still_busy:
 	spin_unlock_irqrestore(&first->b_uptodate_lock, flags);
-	return;
 }
 
 struct postprocess_bh_ctx {
@@ -411,7 +421,6 @@ static void end_buffer_async_write(struct buffer_head *bh, int uptodate)
 
 still_busy:
 	spin_unlock_irqrestore(&first->b_uptodate_lock, flags);
-	return;
 }
 
 /*
@@ -656,7 +665,9 @@ EXPORT_SYMBOL(generic_buffers_fsync);
 void write_boundary_block(struct block_device *bdev,
 			sector_t bblock, unsigned blocksize)
 {
-	struct buffer_head *bh = __find_get_block(bdev, bblock + 1, blocksize);
+	struct buffer_head *bh;
+
+	bh = __find_get_block_nonatomic(bdev, bblock + 1, blocksize);
 	if (bh) {
 		if (buffer_dirty(bh))
 			write_dirty_buffer(bh, 0);
@@ -1109,27 +1120,26 @@ static struct buffer_head *
 __getblk_slow(struct block_device *bdev, sector_t block,
 	     unsigned size, gfp_t gfp)
 {
-	/* Size must be multiple of hard sectorsize */
-	if (unlikely(size & (bdev_logical_block_size(bdev)-1) ||
-			(size < 512 || size > PAGE_SIZE))) {
-		printk(KERN_ERR "getblk(): invalid block size %d requested\n",
-					size);
-		printk(KERN_ERR "logical block size: %d\n",
-					bdev_logical_block_size(bdev));
+	bool blocking = gfpflags_allow_blocking(gfp);
 
-		dump_stack();
+	if (WARN_ON_ONCE(!IS_ALIGNED(size, bdev_logical_block_size(bdev)))) {
+		printk(KERN_ERR "getblk(): block size %d not aligned to logical block size %d\n",
+		       size, bdev_logical_block_size(bdev));
 		return NULL;
 	}
 
 	for (;;) {
 		struct buffer_head *bh;
 
-		bh = __find_get_block(bdev, block, size);
-		if (bh)
-			return bh;
-
 		if (!grow_buffers(bdev, block, size, gfp))
 			return NULL;
+
+		if (blocking)
+			bh = __find_get_block_nonatomic(bdev, block, size);
+		else
+			bh = __find_get_block(bdev, block, size);
+		if (bh)
+			return bh;
 	}
 }
 
@@ -1207,10 +1217,8 @@ void mark_buffer_write_io_error(struct buffer_head *bh)
 	/* FIXME: do we need to set this in both places? */
 	if (bh->b_folio && bh->b_folio->mapping)
 		mapping_set_error(bh->b_folio->mapping, -EIO);
-	if (bh->b_assoc_map) {
+	if (bh->b_assoc_map)
 		mapping_set_error(bh->b_assoc_map, -EIO);
-		errseq_set(&bh->b_assoc_map->host->i_sb->s_wb_err, -EIO);
-	}
 }
 EXPORT_SYMBOL(mark_buffer_write_io_error);
 
@@ -1386,16 +1394,18 @@ lookup_bh_lru(struct block_device *bdev, sector_t block, unsigned size)
 /*
  * Perform a pagecache lookup for the matching buffer.  If it's there, refresh
  * it in the LRU and mark it as accessed.  If it is not present then return
- * NULL
+ * NULL. Atomic context callers may also return NULL if the buffer is being
+ * migrated; similarly the page is not marked accessed either.
  */
-struct buffer_head *
-__find_get_block(struct block_device *bdev, sector_t block, unsigned size)
+static struct buffer_head *
+find_get_block_common(struct block_device *bdev, sector_t block,
+			unsigned size, bool atomic)
 {
 	struct buffer_head *bh = lookup_bh_lru(bdev, block, size);
 
 	if (bh == NULL) {
 		/* __find_get_block_slow will mark the page accessed */
-		bh = __find_get_block_slow(bdev, block);
+		bh = __find_get_block_slow(bdev, block, atomic);
 		if (bh)
 			bh_lru_install(bh);
 	} else
@@ -1403,8 +1413,23 @@ __find_get_block(struct block_device *bdev, sector_t block, unsigned size)
 
 	return bh;
 }
+
+struct buffer_head *
+__find_get_block(struct block_device *bdev, sector_t block, unsigned size)
+{
+	return find_get_block_common(bdev, block, size, true);
+}
 EXPORT_SYMBOL(__find_get_block);
 
+/* same as __find_get_block() but allows sleeping contexts */
+struct buffer_head *
+__find_get_block_nonatomic(struct block_device *bdev, sector_t block,
+			   unsigned size)
+{
+	return find_get_block_common(bdev, block, size, false);
+}
+EXPORT_SYMBOL(__find_get_block_nonatomic);
+
 /**
  * bdev_getblk - Get a buffer_head in a block device's buffer cache.
  * @bdev: The block device.
@@ -1422,7 +1447,12 @@ EXPORT_SYMBOL(__find_get_block);
 struct buffer_head *bdev_getblk(struct block_device *bdev, sector_t block,
 		unsigned size, gfp_t gfp)
 {
-	struct buffer_head *bh = __find_get_block(bdev, block, size);
+	struct buffer_head *bh;
+
+	if (gfpflags_allow_blocking(gfp))
+		bh = __find_get_block_nonatomic(bdev, block, size);
+	else
+		bh = __find_get_block(bdev, block, size);
 
 	might_alloc(gfp);
 	if (bh)
@@ -1578,8 +1608,8 @@ static void discard_buffer(struct buffer_head * bh)
 	bh->b_bdev = NULL;
 	b_state = READ_ONCE(bh->b_state);
 	do {
-	} while (!try_cmpxchg(&bh->b_state, &b_state,
-			      b_state & ~BUFFER_FLAGS_DISCARD));
+	} while (!try_cmpxchg_relaxed(&bh->b_state, &b_state,
+				      b_state & ~BUFFER_FLAGS_DISCARD));
 	unlock_buffer(bh);
 }
 
@@ -1644,7 +1674,6 @@ void block_invalidate_folio(struct folio *folio, size_t offset, size_t length)
 		filemap_release_folio(folio, 0);
 out:
 	folio_clear_mappedtodisk(folio);
-	return;
 }
 EXPORT_SYMBOL(block_invalidate_folio);
 
@@ -2236,9 +2265,8 @@ int block_write_begin(struct address_space *mapping, loff_t pos, unsigned len,
 }
 EXPORT_SYMBOL(block_write_begin);
 
-int block_write_end(struct file *file, struct address_space *mapping,
-			loff_t pos, unsigned len, unsigned copied,
-			struct folio *folio, void *fsdata)
+int block_write_end(loff_t pos, unsigned len, unsigned copied,
+		struct folio *folio)
 {
 	size_t start = pos - folio_pos(folio);
 
@@ -2269,15 +2297,15 @@ int block_write_end(struct file *file, struct address_space *mapping,
 }
 EXPORT_SYMBOL(block_write_end);
 
-int generic_write_end(struct file *file, struct address_space *mapping,
-			loff_t pos, unsigned len, unsigned copied,
-			struct folio *folio, void *fsdata)
+int generic_write_end(const struct kiocb *iocb, struct address_space *mapping,
+		      loff_t pos, unsigned len, unsigned copied,
+		      struct folio *folio, void *fsdata)
 {
 	struct inode *inode = mapping->host;
 	loff_t old_size = inode->i_size;
 	bool i_size_changed = false;
 
-	copied = block_write_end(file, mapping, pos, len, copied, folio, fsdata);
+	copied = block_write_end(pos, len, copied, folio);
 
 	/*
 	 * No need to use i_size_read() here, the i_size cannot change under us
@@ -2466,7 +2494,8 @@ out:
 }
 EXPORT_SYMBOL(generic_cont_expand_simple);
 
-static int cont_expand_zero(struct file *file, struct address_space *mapping,
+static int cont_expand_zero(const struct kiocb *iocb,
+			    struct address_space *mapping,
 			    loff_t pos, loff_t *bytes)
 {
 	struct inode *inode = mapping->host;
@@ -2490,12 +2519,12 @@ static int cont_expand_zero(struct file *file, struct address_space *mapping,
 		}
 		len = PAGE_SIZE - zerofrom;
 
-		err = aops->write_begin(file, mapping, curpos, len,
+		err = aops->write_begin(iocb, mapping, curpos, len,
 					    &folio, &fsdata);
 		if (err)
 			goto out;
 		folio_zero_range(folio, offset_in_folio(folio, curpos), len);
-		err = aops->write_end(file, mapping, curpos, len, len,
+		err = aops->write_end(iocb, mapping, curpos, len, len,
 						folio, fsdata);
 		if (err < 0)
 			goto out;
@@ -2523,12 +2552,12 @@ static int cont_expand_zero(struct file *file, struct address_space *mapping,
 		}
 		len = offset - zerofrom;
 
-		err = aops->write_begin(file, mapping, curpos, len,
+		err = aops->write_begin(iocb, mapping, curpos, len,
 					    &folio, &fsdata);
 		if (err)
 			goto out;
 		folio_zero_range(folio, offset_in_folio(folio, curpos), len);
-		err = aops->write_end(file, mapping, curpos, len, len,
+		err = aops->write_end(iocb, mapping, curpos, len, len,
 						folio, fsdata);
 		if (err < 0)
 			goto out;
@@ -2543,17 +2572,16 @@ out:
  * For moronic filesystems that do not allow holes in file.
  * We may have to extend the file.
  */
-int cont_write_begin(struct file *file, struct address_space *mapping,
-			loff_t pos, unsigned len,
-			struct folio **foliop, void **fsdata,
-			get_block_t *get_block, loff_t *bytes)
+int cont_write_begin(const struct kiocb *iocb, struct address_space *mapping,
+		     loff_t pos, unsigned len, struct folio **foliop,
+		     void **fsdata, get_block_t *get_block, loff_t *bytes)
 {
 	struct inode *inode = mapping->host;
 	unsigned int blocksize = i_blocksize(inode);
 	unsigned int zerofrom;
 	int err;
 
-	err = cont_expand_zero(file, mapping, pos, bytes);
+	err = cont_expand_zero(iocb, mapping, pos, bytes);
 	if (err)
 		return err;
 
@@ -2575,7 +2603,7 @@ EXPORT_SYMBOL(cont_write_begin);
  * holes and correct delalloc and unwritten extent mapping on filesystems that
  * support these features.
  *
- * We are not allowed to take the i_mutex here so we have to play games to
+ * We are not allowed to take the i_rwsem here so we have to play games to
  * protect against truncate races as the page could now be beyond EOF.  Because
  * truncate writes the inode size before removing pages, once we have the
  * page lock we can determine safely if the page is beyond EOF. If it is not
@@ -2695,7 +2723,7 @@ unlock:
 EXPORT_SYMBOL(block_truncate_page);
 
 /*
- * The generic ->writepage function for buffer-backed address_spaces
+ * The generic write folio function for buffer-backed address_spaces
  */
 int block_write_full_folio(struct folio *folio, struct writeback_control *wbc,
 		void *get_block)
@@ -2715,7 +2743,7 @@ int block_write_full_folio(struct folio *folio, struct writeback_control *wbc,
 
 	/*
 	 * The folio straddles i_size.  It must be zeroed out on each and every
-	 * writepage invocation because it may be mmapped.  "A file is mapped
+	 * writeback invocation because it may be mmapped.  "A file is mapped
 	 * in multiples of the page size.  For a file that is not a multiple of
 	 * the page size, the remaining memory is zeroed when mapped, and
 	 * writes to that region are not written out to the file."