Merge patch series "fs: refactor write_begin/write_end and add ext4 IOCB_DONTCACHE support"

陈涛涛 Taotao Chen <chentaotao@didiglobal.com> says:

This patch series refactors the address_space_operations write_begin()
and write_end() callbacks to take const struct kiocb * as their first
argument, allowing IOCB flags such as IOCB_DONTCACHE to propagate to the
filesystem's buffered I/O path.

Ext4 is updated to implement handling of the IOCB_DONTCACHE flag and
advertises support via the FOP_DONTCACHE file operation flag.

Additionally, the i915 driver's shmem write paths are updated to bypass
the legacy write_begin/write_end interface in favor of directly
calling write_iter() with a constructed synchronous kiocb. Another i915
change replaces a manual write loop with kernel_write() during GEM shmem
object creation.

Tested with ext4 and i915 GEM workloads.

* patches from https://lore.kernel.org/20250716093559.217344-1-chentaotao@didiglobal.com:
  ext4: support uncached buffered I/O
  mm/pagemap: add write_begin_get_folio() helper function
  fs: change write_begin/write_end interface to take struct kiocb *
  drm/i915: Refactor shmem_pwrite() to use kiocb and write_iter
  drm/i915: Use kernel_write() in shmem object create

Link: https://lore.kernel.org/20250716093559.217344-1-chentaotao@didiglobal.com
Signed-off-by: Christian Brauner <brauner@kernel.org>

Author: brauner

Documentation/filesystems/locking.rst

@@ -253,10 +253,10 @@ prototypes::
 	int (*writepages)(struct address_space *, struct writeback_control *);
 	bool (*dirty_folio)(struct address_space *, struct folio *folio);
 	void (*readahead)(struct readahead_control *);
-	int (*write_begin)(struct file *, struct address_space *mapping,
+	int (*write_begin)(const struct kiocb *, struct address_space *mapping,
 				loff_t pos, unsigned len,
 				struct folio **foliop, void **fsdata);
-	int (*write_end)(struct file *, struct address_space *mapping,
+	int (*write_end)(const struct kiocb *, struct address_space *mapping,
 				loff_t pos, unsigned len, unsigned copied,
 				struct folio *folio, void *fsdata);
 	sector_t (*bmap)(struct address_space *, sector_t);

Documentation/filesystems/vfs.rst

@@ -823,10 +823,10 @@ cache in your filesystem.  The following members are defined:
 		int (*writepages)(struct address_space *, struct writeback_control *);
 		bool (*dirty_folio)(struct address_space *, struct folio *);
 		void (*readahead)(struct readahead_control *);
-		int (*write_begin)(struct file *, struct address_space *mapping,
+		int (*write_begin)(const struct kiocb *, struct address_space *mapping,
 				   loff_t pos, unsigned len,
-				struct page **pagep, void **fsdata);
-		int (*write_end)(struct file *, struct address_space *mapping,
+				   struct page **pagep, void **fsdata);
+		int (*write_end)(const struct kiocb *, struct address_space *mapping,
 				 loff_t pos, unsigned len, unsigned copied,
 				 struct folio *folio, void *fsdata);
 		sector_t (*bmap)(struct address_space *, sector_t);

block/fops.c

@@ -496,15 +496,18 @@ static void blkdev_readahead(struct readahead_control *rac)
 	mpage_readahead(rac, blkdev_get_block);
 }
 
-static int blkdev_write_begin(struct file *file, struct address_space *mapping,
-		loff_t pos, unsigned len, struct folio **foliop, void **fsdata)
+static int blkdev_write_begin(const struct kiocb *iocb,
+			      struct address_space *mapping, loff_t pos,
+			      unsigned len, struct folio **foliop,
+			      void **fsdata)
 {
 	return block_write_begin(mapping, pos, len, foliop, blkdev_get_block);
 }
 
-static int blkdev_write_end(struct file *file, struct address_space *mapping,
-		loff_t pos, unsigned len, unsigned copied, struct folio *folio,
-		void *fsdata)
+static int blkdev_write_end(const struct kiocb *iocb,
+			    struct address_space *mapping,
+			    loff_t pos, unsigned len, unsigned copied,
+			    struct folio *folio, void *fsdata)
 {
 	int ret;
 	ret = block_write_end(pos, len, copied, folio);

drivers/gpu/drm/i915/gem/i915_gem_shmem.c

@@ -6,6 +6,7 @@
 #include <linux/pagevec.h>
 #include <linux/shmem_fs.h>
 #include <linux/swap.h>
+#include <linux/uio.h>
 
 #include <drm/drm_cache.h>
 
@@ -400,12 +401,12 @@ static int
 shmem_pwrite(struct drm_i915_gem_object *obj,
 	     const struct drm_i915_gem_pwrite *arg)
 {
-	struct address_space *mapping = obj->base.filp->f_mapping;
-	const struct address_space_operations *aops = mapping->a_ops;
 	char __user *user_data = u64_to_user_ptr(arg->data_ptr);
-	u64 remain;
-	loff_t pos;
-	unsigned int pg;
+	struct file *file = obj->base.filp;
+	struct kiocb kiocb;
+	struct iov_iter iter;
+	ssize_t written;
+	u64 size = arg->size;
 
 	/* Caller already validated user args */
 	GEM_BUG_ON(!access_ok(user_data, arg->size));
@@ -428,63 +429,24 @@ shmem_pwrite(struct drm_i915_gem_object *obj,
 	if (obj->mm.madv != I915_MADV_WILLNEED)
 		return -EFAULT;
 
-	/*
-	 * Before the pages are instantiated the object is treated as being
-	 * in the CPU domain. The pages will be clflushed as required before
-	 * use, and we can freely write into the pages directly. If userspace
-	 * races pwrite with any other operation; corruption will ensue -
-	 * that is userspace's prerogative!
-	 */
+	if (size > MAX_RW_COUNT)
+		return -EFBIG;
 
-	remain = arg->size;
-	pos = arg->offset;
-	pg = offset_in_page(pos);
+	if (!file->f_op->write_iter)
+		return -EINVAL;
 
-	do {
-		unsigned int len, unwritten;
-		struct folio *folio;
-		void *data, *vaddr;
-		int err;
-		char __maybe_unused c;
-
-		len = PAGE_SIZE - pg;
-		if (len > remain)
-			len = remain;
-
-		/* Prefault the user page to reduce potential recursion */
-		err = __get_user(c, user_data);
-		if (err)
-			return err;
-
-		err = __get_user(c, user_data + len - 1);
-		if (err)
-			return err;
-
-		err = aops->write_begin(obj->base.filp, mapping, pos, len,
-					&folio, &data);
-		if (err < 0)
-			return err;
-
-		vaddr = kmap_local_folio(folio, offset_in_folio(folio, pos));
-		pagefault_disable();
-		unwritten = __copy_from_user_inatomic(vaddr, user_data, len);
-		pagefault_enable();
-		kunmap_local(vaddr);
-
-		err = aops->write_end(obj->base.filp, mapping, pos, len,
-				      len - unwritten, folio, data);
-		if (err < 0)
-			return err;
-
-		/* We don't handle -EFAULT, leave it to the caller to check */
-		if (unwritten)
-			return -ENODEV;
-
-		remain -= len;
-		user_data += len;
-		pos += len;
-		pg = 0;
-	} while (remain);
+	init_sync_kiocb(&kiocb, file);
+	kiocb.ki_pos = arg->offset;
+	iov_iter_ubuf(&iter, ITER_SOURCE, (void __user *)user_data, size);
+
+	written = file->f_op->write_iter(&kiocb, &iter);
+	BUG_ON(written == -EIOCBQUEUED);
+
+	if (written != size)
+		return -EIO;
+
+	if (written < 0)
+		return written;
 
 	return 0;
 }
@@ -637,9 +599,8 @@ i915_gem_object_create_shmem_from_data(struct drm_i915_private *i915,
 {
 	struct drm_i915_gem_object *obj;
 	struct file *file;
-	const struct address_space_operations *aops;
-	loff_t pos;
-	int err;
+	loff_t pos = 0;
+	ssize_t err;
 
 	GEM_WARN_ON(IS_DGFX(i915));
 	obj = i915_gem_object_create_shmem(i915, round_up(size, PAGE_SIZE));
@@ -649,29 +610,15 @@ i915_gem_object_create_shmem_from_data(struct drm_i915_private *i915,
 	GEM_BUG_ON(obj->write_domain != I915_GEM_DOMAIN_CPU);
 
 	file = obj->base.filp;
-	aops = file->f_mapping->a_ops;
-	pos = 0;
-	do {
-		unsigned int len = min_t(typeof(size), size, PAGE_SIZE);
-		struct folio *folio;
-		void *fsdata;
-
-		err = aops->write_begin(file, file->f_mapping, pos, len,
-					&folio, &fsdata);
-		if (err < 0)
-			goto fail;
+	err = kernel_write(file, data, size, &pos);
 
-		memcpy_to_folio(folio, offset_in_folio(folio, pos), data, len);
+	if (err < 0)
+		goto fail;
 
-		err = aops->write_end(file, file->f_mapping, pos, len, len,
-				      folio, fsdata);
-		if (err < 0)
-			goto fail;
-
-		size -= len;
-		data += len;
-		pos += len;
-	} while (size);
+	if (err != size) {
+		err = -EIO;
+		goto fail;
+	}
 
 	return obj;
 

fs/adfs/inode.c

@@ -53,13 +53,14 @@ static void adfs_write_failed(struct address_space *mapping, loff_t to)
 		truncate_pagecache(inode, inode->i_size);
 }
 
-static int adfs_write_begin(struct file *file, struct address_space *mapping,
-			loff_t pos, unsigned len,
-			struct folio **foliop, void **fsdata)
+static int adfs_write_begin(const struct kiocb *iocb,
+			    struct address_space *mapping,
+			    loff_t pos, unsigned len,
+			    struct folio **foliop, void **fsdata)
 {
 	int ret;
 
-	ret = cont_write_begin(file, mapping, pos, len, foliop, fsdata,
+	ret = cont_write_begin(iocb, mapping, pos, len, foliop, fsdata,
 				adfs_get_block,
 				&ADFS_I(mapping->host)->mmu_private);
 	if (unlikely(ret))

fs/affs/file.c

@@ -415,13 +415,14 @@ affs_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
 	return ret;
 }
 
-static int affs_write_begin(struct file *file, struct address_space *mapping,
-			loff_t pos, unsigned len,
-			struct folio **foliop, void **fsdata)
+static int affs_write_begin(const struct kiocb *iocb,
+			    struct address_space *mapping,
+			    loff_t pos, unsigned len,
+			    struct folio **foliop, void **fsdata)
 {
 	int ret;
 
-	ret = cont_write_begin(file, mapping, pos, len, foliop, fsdata,
+	ret = cont_write_begin(iocb, mapping, pos, len, foliop, fsdata,
 				affs_get_block,
 				&AFFS_I(mapping->host)->mmu_private);
 	if (unlikely(ret))
@@ -430,14 +431,15 @@ static int affs_write_begin(struct file *file, struct address_space *mapping,
 	return ret;
 }
 
-static int affs_write_end(struct file *file, struct address_space *mapping,
-			  loff_t pos, unsigned int len, unsigned int copied,
+static int affs_write_end(const struct kiocb *iocb,
+			  struct address_space *mapping, loff_t pos,
+			  unsigned int len, unsigned int copied,
 			  struct folio *folio, void *fsdata)
 {
 	struct inode *inode = mapping->host;
 	int ret;
 
-	ret = generic_write_end(file, mapping, pos, len, copied, folio, fsdata);
+	ret = generic_write_end(iocb, mapping, pos, len, copied, folio, fsdata);
 
 	/* Clear Archived bit on file writes, as AmigaOS would do */
 	if (AFFS_I(inode)->i_protect & FIBF_ARCHIVED) {
@@ -645,7 +647,8 @@ static int affs_read_folio_ofs(struct file *file, struct folio *folio)
 	return err;
 }
 
-static int affs_write_begin_ofs(struct file *file, struct address_space *mapping,
+static int affs_write_begin_ofs(const struct kiocb *iocb,
+				struct address_space *mapping,
 				loff_t pos, unsigned len,
 				struct folio **foliop, void **fsdata)
 {
@@ -684,9 +687,10 @@ static int affs_write_begin_ofs(struct file *file, struct address_space *mapping
 	return err;
 }
 
-static int affs_write_end_ofs(struct file *file, struct address_space *mapping,
-				loff_t pos, unsigned len, unsigned copied,
-				struct folio *folio, void *fsdata)
+static int affs_write_end_ofs(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;
 	struct super_block *sb = inode->i_sb;

fs/bcachefs/fs-io-buffered.c

@@ -674,7 +674,7 @@ int bch2_writepages(struct address_space *mapping, struct writeback_control *wbc
 
 /* buffered writes: */
 
-int bch2_write_begin(struct file *file, struct address_space *mapping,
+int bch2_write_begin(const struct kiocb *iocb, struct address_space *mapping,
 		     loff_t pos, unsigned len,
 		     struct folio **foliop, void **fsdata)
 {
@@ -757,7 +757,7 @@ int bch2_write_begin(struct file *file, struct address_space *mapping,
 	return bch2_err_class(ret);
 }
 
-int bch2_write_end(struct file *file, struct address_space *mapping,
+int bch2_write_end(const struct kiocb *iocb, struct address_space *mapping,
 		   loff_t pos, unsigned len, unsigned copied,
 		   struct folio *folio, void *fsdata)
 {

fs/bcachefs/fs-io-buffered.h

@@ -10,9 +10,9 @@ int bch2_read_folio(struct file *, struct folio *);
 int bch2_writepages(struct address_space *, struct writeback_control *);
 void bch2_readahead(struct readahead_control *);
 
-int bch2_write_begin(struct file *, struct address_space *, loff_t pos,
+int bch2_write_begin(const struct kiocb *, struct address_space *, loff_t pos,
 		     unsigned len, struct folio **, void **);
-int bch2_write_end(struct file *, struct address_space *, loff_t,
+int bch2_write_end(const struct kiocb *, struct address_space *, loff_t,
 		   unsigned len, unsigned copied, struct folio *, void *);
 
 ssize_t bch2_write_iter(struct kiocb *, struct iov_iter *);

fs/bfs/file.c

@@ -170,9 +170,10 @@ static void bfs_write_failed(struct address_space *mapping, loff_t to)
 		truncate_pagecache(inode, inode->i_size);
 }
 
-static int bfs_write_begin(struct file *file, struct address_space *mapping,
-			loff_t pos, unsigned len,
-			struct folio **foliop, void **fsdata)
+static int bfs_write_begin(const struct kiocb *iocb,
+			   struct address_space *mapping,
+			   loff_t pos, unsigned len,
+			   struct folio **foliop, void **fsdata)
 {
 	int ret;