[PATCH v2] exfat: drain dirty pages after forced shutdown

[PATCH v2] exfat: drain dirty pages after forced shutdown

[Chi Zhiling]

From: Chi Zhiling <chizhiling@kylinos.cn>

A forced shutdown sets EXFAT_FLAGS_SHUTDOWN, after which
exfat_writepages() returns -EIO.  Pages dirtied before the shutdown can
then never be written back, so a writer throttled in
balance_dirty_pages() while extending valid_size sleeps forever.

Stuck stack:
  balance_dirty_pages+0x4f0/0xb40
  exfat_prepare_valid_range+0x43/0x140
  exfat_file_write_iter+0x413/0x520
  iter_file_splice_write+0x32d/0x5a0
  direct_splice_actor+0x51/0x270
  splice_direct_to_actor+0xf5/0x270
  do_splice_direct+0x76/0xc0
  vfs_copy_file_range+0x244/0x710
  __do_sys_copy_file_range+0xe6/0x240
  do_syscall_64+0x106/0x5f0
  entry_SYSCALL_64_after_hwframe+0x77/0x7f

Do not reject writeback up front.  Reject it per folio from
->writeback_range() instead: iomap then ends writeback and clears the
dirty state without issuing I/O, which lets the dirty limits drain and
unblocks the writer.

Write syscalls and page faults are still rejected in
exfat_file_write_iter() and exfat_page_mkwrite().  A write that is
already in progress may continue to dirty folios; writeback retires
them without I/O as described above.

Fixes: f761fcdd289d0 ("exfat: Implement sops->shutdown and ioctl")
Signed-off-by: Chi Zhiling <chizhiling@kylinos.cn>
---
 fs/exfat/file.c  | 3 +++
 fs/exfat/inode.c | 3 ---
 fs/exfat/iomap.c | 7 ++++++-
 3 files changed, 9 insertions(+), 4 deletions(-)

diff --git a/fs/exfat/file.c b/fs/exfat/file.c
index 1386149916bb..c358899e77f3 100644
--- a/fs/exfat/file.c
+++ b/fs/exfat/file.c
@@ -975,6 +975,9 @@ static vm_fault_t exfat_page_mkwrite(struct vm_fault *vmf)
 	vm_fault_t ret;
 	loff_t new_valid_size, mmap_valid_size, fault_page_start;
 
+	if (exfat_forced_shutdown(inode->i_sb))
+		return vmf_fs_error(-EIO);
+
 	if (!inode_trylock(inode))
 		return VM_FAULT_RETRY;
 
diff --git a/fs/exfat/inode.c b/fs/exfat/inode.c
index 5129df1e79b4..4bea9e9deb94 100644
--- a/fs/exfat/inode.c
+++ b/fs/exfat/inode.c
@@ -283,9 +283,6 @@ static int exfat_writepages(struct address_space *mapping,
 		.ops		= &exfat_writeback_ops,
 	};
 
-	if (unlikely(exfat_forced_shutdown(mapping->host->i_sb)))
-		return -EIO;
-
 	return iomap_writepages(&wpc);
 }
 
diff --git a/fs/exfat/iomap.c b/fs/exfat/iomap.c
index e50ecbfbfbe0..2089da8439e4 100644
--- a/fs/exfat/iomap.c
+++ b/fs/exfat/iomap.c
@@ -261,11 +261,16 @@ const struct iomap_ops exfat_write_iomap_ops = {
 static ssize_t exfat_writeback_range(struct iomap_writepage_ctx *wpc,
 		struct folio *folio, u64 offset, unsigned int len, u64 end_pos)
 {
+	struct inode *inode = wpc->inode;
+
+	if (exfat_forced_shutdown(inode->i_sb))
+		return -EIO;
+
 	if (offset < wpc->iomap.offset ||
 	    offset >= wpc->iomap.offset + wpc->iomap.length) {
 		int error;
 
-		error = __exfat_iomap_begin(wpc->inode, offset, len,
+		error = __exfat_iomap_begin(inode, offset, len,
 				0, &wpc->iomap, false);
 		if (error)
 			return error;
-- 
2.53.0

Re: [PATCH v2] exfat: drain dirty pages after forced shutdown

[Namjae Jeon]

On Tue, Oct 27, 2026 at 2:47 PM Chi Zhiling <chizhiling@163.com> wrote:
>
> From: Chi Zhiling <chizhiling@kylinos.cn>
>
> A forced shutdown sets EXFAT_FLAGS_SHUTDOWN, after which
> exfat_writepages() returns -EIO.  Pages dirtied before the shutdown can
> then never be written back, so a writer throttled in
> balance_dirty_pages() while extending valid_size sleeps forever.
>
> Stuck stack:
>   balance_dirty_pages+0x4f0/0xb40
>   exfat_prepare_valid_range+0x43/0x140
>   exfat_file_write_iter+0x413/0x520
>   iter_file_splice_write+0x32d/0x5a0
>   direct_splice_actor+0x51/0x270
>   splice_direct_to_actor+0xf5/0x270
>   do_splice_direct+0x76/0xc0
>   vfs_copy_file_range+0x244/0x710
>   __do_sys_copy_file_range+0xe6/0x240
>   do_syscall_64+0x106/0x5f0
>   entry_SYSCALL_64_after_hwframe+0x77/0x7f
>
> Do not reject writeback up front.  Reject it per folio from
> ->writeback_range() instead: iomap then ends writeback and clears the
> dirty state without issuing I/O, which lets the dirty limits drain and
> unblocks the writer.
>
> Write syscalls and page faults are still rejected in
> exfat_file_write_iter() and exfat_page_mkwrite().  A write that is
> already in progress may continue to dirty folios; writeback retires
> them without I/O as described above.
>
> Fixes: f761fcdd289d0 ("exfat: Implement sops->shutdown and ioctl")
> Signed-off-by: Chi Zhiling <chizhiling@kylinos.cn>
Applied it to #dev.
Thanks!