Thread (49 messages) 49 messages, 9 authors, 2017-10-15
STALE3196d
Revisions (2)
  1. v7 current
  2. v8 [diff vs current]

[PATCH v7 03/12] fs: introduce i_mapdcount

From: Dan Williams <hidden>
Date: 2017-10-06 22:35:32
Also in: linux-fsdevel, linux-mm, linux-rdma, linux-xfs, nvdimm
Subsystem: filesystems (vfs and infrastructure), the rest, xfs filesystem · Maintainers: Alexander Viro, Christian Brauner, Linus Torvalds, Carlos Maiolino

When ->iomap_begin() sees this count being non-zero and determines that
the block map of the file needs to be modified to satisfy the I/O
request it will instead return an error. This is needed for MAP_DIRECT
where, due to locking constraints, we can't rely on xfs_break_layouts()
to protect against allocating write-faults either from the process that
setup the MAP_DIRECT mapping nor other processes that have the file
mapped.  xfs_break_layouts() requires XFS_IOLOCK which is problematic to
mix with the XFS_MMAPLOCK in the fault path.

Cc: Jan Kara <jack@suse.cz>
Cc: Jeff Moyer <redacted>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Dave Chinner <david@fromorbit.com>
Cc: "Darrick J. Wong" <redacted>
Cc: Ross Zwisler <redacted>
Cc: Jeff Layton <redacted>
Cc: "J. Bruce Fields" <redacted>
Signed-off-by: Dan Williams <redacted>
---
 fs/xfs/xfs_iomap.c |    9 +++++++++
 include/linux/fs.h |   31 +++++++++++++++++++++++++++++++
 2 files changed, 40 insertions(+)
diff --git a/fs/xfs/xfs_iomap.c b/fs/xfs/xfs_iomap.c
index a1909bc064e9..6816f8ebbdcf 100644
--- a/fs/xfs/xfs_iomap.c
+++ b/fs/xfs/xfs_iomap.c
@@ -1053,6 +1053,15 @@ xfs_file_iomap_begin(
 			goto out_unlock;
 		}
 		/*
+		 * If a file has MAP_DIRECT mappings disable block map
+		 * updates. This should only effect mmap write faults as
+		 * other paths are protected by an FL_LAYOUT lease.
+		 */
+		if (i_mapdcount_read(inode)) {
+			error = -ETXTBSY;
+			goto out_unlock;
+		}
+		/*
 		 * We cap the maximum length we map here to MAX_WRITEBACK_PAGES
 		 * pages to keep the chunks of work done where somewhat symmetric
 		 * with the work writeback does. This is a completely arbitrary
diff --git a/include/linux/fs.h b/include/linux/fs.h
index c2b9bf3dc4e9..f83871b188ff 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -642,6 +642,9 @@ struct inode {
 	atomic_t		i_count;
 	atomic_t		i_dio_count;
 	atomic_t		i_writecount;
+#ifdef CONFIG_FS_DAX
+	atomic_t		i_mapdcount;	/* count of MAP_DIRECT vmas */
+#endif
 #ifdef CONFIG_IMA
 	atomic_t		i_readcount; /* struct files open RO */
 #endif
@@ -2784,6 +2787,34 @@ static inline void i_readcount_inc(struct inode *inode)
 	return;
 }
 #endif
+
+#ifdef CONFIG_FS_DAX
+static inline void i_mapdcount_dec(struct inode *inode)
+{
+	BUG_ON(!atomic_read(&inode->i_mapdcount));
+	atomic_dec(&inode->i_mapdcount);
+}
+static inline void i_mapdcount_inc(struct inode *inode)
+{
+	atomic_inc(&inode->i_mapdcount);
+}
+static inline int i_mapdcount_read(struct inode *inode)
+{
+	return atomic_read(&inode->i_mapdcount);
+}
+#else
+static inline void i_mapdcount_dec(struct inode *inode)
+{
+}
+static inline void i_mapdcount_inc(struct inode *inode)
+{
+}
+static inline int i_mapdcount_read(struct inode *inode)
+{
+	return 0;
+}
+#endif
+
 extern int do_pipe_flags(int *, int);
 
 #define __kernel_read_file_id(id) \

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help