From: Darrick J. Wong <djwong@xxxxxxxxxx>
If we have shared realtime files and the rt extent size is larger than a
single fs block, we need to extend writeback requests to be aligned to
rt extent size granularity because we cannot share partial rt extents.
The front end should have set us up for this by dirtying the relevant
ranges.
Signed-off-by: "Darrick J. Wong" <djwong@xxxxxxxxxx>
---
fs/xfs/xfs_aops.c | 38 ++++++++++++++++++++++++++++++++++++++
fs/xfs/xfs_trace.h | 1 +
2 files changed, 39 insertions(+)
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index f51f2f5f76d0f6..9bc2d7d92e4c46 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -466,6 +466,38 @@ static const struct iomap_writeback_ops xfs_writeback_ops = {
.discard_folio = xfs_discard_folio,
};
+/*
+ * Extend the writeback range to allocation unit granularity and alignment.
+ * This is a requirement for blocksize > pagesize scenarios such as realtime
+ * copy on write, since we can only share full rt extents.
+ */
+static inline void
+xfs_vm_writepages_extend(
+ struct xfs_inode *ip,
+ struct writeback_control *wbc)
+{
+ unsigned int bsize = xfs_inode_alloc_unitsize(ip);
+ long long int pages_to_write;
+ loff_t next = wbc->range_end + 1;
+
+ wbc->range_start = rounddown_64(wbc->range_start, bsize);
+ if (wbc->range_end != LLONG_MAX)
+ wbc->range_end = roundup_64(next, bsize) - 1;
+
+ if (wbc->nr_to_write != LONG_MAX) {
+ pgoff_t pg_start = wbc->range_start >> PAGE_SHIFT;
+ pgoff_t pg_next = (wbc->range_end + 1) >> PAGE_SHIFT;
+
+ pages_to_write = pg_next - pg_start;
+ if (pages_to_write >= LONG_MAX)
+ pages_to_write = LONG_MAX;
+ if (wbc->nr_to_write < pages_to_write)
+ wbc->nr_to_write = pages_to_write;
+ }
+
+ trace_xfs_vm_writepages_extend(ip, wbc);
+}
+
STATIC int
xfs_vm_writepages(
struct address_space *mapping,
@@ -476,6 +508,9 @@ xfs_vm_writepages(
trace_xfs_vm_writepages(ip, wbc);
+ if (xfs_inode_needs_cow_around(ip))
+ xfs_vm_writepages_extend(ip, wbc);
+
xfs_iflags_clear(ip, XFS_ITRUNCATED);
return iomap_writepages(mapping, wbc, &wpc.ctx, &xfs_writeback_ops);
}
@@ -489,6 +524,9 @@ xfs_dax_writepages(
trace_xfs_dax_writepages(ip, wbc);
+ if (xfs_inode_needs_cow_around(ip))
+ xfs_vm_writepages_extend(ip, wbc);
+
xfs_iflags_clear(ip, XFS_ITRUNCATED);
return dax_writeback_mapping_range(mapping,
xfs_inode_buftarg(ip)->bt_daxdev, wbc);
diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h
index 0234af78cea9a1..021ea65909c915 100644
--- a/fs/xfs/xfs_trace.h
+++ b/fs/xfs/xfs_trace.h
@@ -1586,6 +1586,7 @@ DEFINE_EVENT(xfs_writeback_class, name, \
TP_PROTO(struct xfs_inode *ip, const struct writeback_control *wbc), \
TP_ARGS(ip, wbc))
DEFINE_WRITEBACK_EVENT(xfs_vm_writepages);
+DEFINE_WRITEBACK_EVENT(xfs_vm_writepages_extend);
DEFINE_WRITEBACK_EVENT(xfs_dax_writepages);
DECLARE_EVENT_CLASS(xfs_simple_io_class,
