Fix lseek(SEEK_DATA/SEEK_HOLE) mmap consistency

When using lseek(2) to report data/holes memory mapped regions of
the file were ignored.  This could result in incorrect results.
To handle this zfs_holey_common() was updated to asynchronously
writeback any dirty mmap(2) regions prior to reporting holes.

Additionally, while not strictly required, the dn_struct_rwlock is
now held over the dirty check to prevent the dnode structure from
changing.  This ensures that a clean dnode can't be dirtied before
the data/hole is located.  The range lock is now also taken to
ensure the call cannot race with zfs_write().

Furthermore, the code was refactored to provide a dnode_is_dirty()
helper function which checks the dnode for any dirty records to
determine its dirtiness.

Reviewed-by: Matthew Ahrens <[email protected]>
Reviewed-by: Tony Hutter <[email protected]>
Reviewed-by: Rich Ercolani <[email protected]>
Signed-off-by: Brian Behlendorf <[email protected]>
Issue #11900
Closes #12724

4 files changed, 22 insertions, 1 deletions

diff --git a/include/os/freebsd/spl/sys/vnode.h b/include/os/freebsd/spl/sys/vnode.h
index 3670712a0..3bc8a18ee 100644
--- a/include/os/freebsd/spl/sys/vnode.h
+++ b/include/os/freebsd/spl/sys/vnode.h
@@ -59,6 +59,8 @@ enum symfollow { NO_FOLLOW = NOFOLLOW };
 #include <sys/file.h>
 #include <sys/filedesc.h>
 #include <sys/syscallsubr.h>
+#include <sys/vm.h>
+#include <vm/vm_object.h>
 
 typedef	struct vop_vector	vnodeops_t;
 #define	VOP_FID		VOP_VPTOFH
@@ -83,6 +85,22 @@ vn_is_readonly(vnode_t *vp)
 #define	vn_has_cached_data(vp)	\
 	((vp)->v_object != NULL && \
 	(vp)->v_object->resident_page_count > 0)
+
+static __inline void
+vn_flush_cached_data(vnode_t *vp, boolean_t sync)
+{
+#if __FreeBSD_version > 1300054
+	if (vm_object_mightbedirty(vp->v_object)) {
+#else
+	if (vp->v_object->flags & OBJ_MIGHTBEDIRTY) {
+#endif
+		int flags = sync ? OBJPC_SYNC : 0;
+		zfs_vmobject_wlock(vp->v_object);
+		vm_object_page_clean(vp->v_object, 0, 0, flags);
+		zfs_vmobject_wunlock(vp->v_object);
+	}
+}
+
 #define	vn_exists(vp)		do { } while (0)
 #define	vn_invalid(vp)		do { } while (0)
 #define	vn_renamepath(tdvp, svp, tnm, lentnm)	do { } while (0)
diff --git a/include/os/freebsd/zfs/sys/zfs_znode_impl.h b/include/os/freebsd/zfs/sys/zfs_znode_impl.h
index e90008c70..edb28d041 100644
--- a/include/os/freebsd/zfs/sys/zfs_znode_impl.h
+++ b/include/os/freebsd/zfs/sys/zfs_znode_impl.h
@@ -118,7 +118,8 @@ extern minor_t zfsdev_minor_alloc(void);
 #define	Z_ISLNK(type) ((type) == VLNK)
 #define	Z_ISDIR(type) ((type) == VDIR)
 
-#define	zn_has_cached_data(zp)	vn_has_cached_data(ZTOV(zp))
+#define	zn_has_cached_data(zp)		vn_has_cached_data(ZTOV(zp))
+#define	zn_flush_cached_data(zp, sync)	vn_flush_cached_data(ZTOV(zp), sync)
 #define	zn_rlimit_fsize(zp, uio) \
     vn_rlimit_fsize(ZTOV(zp), GET_UIO_STRUCT(uio), zfs_uio_td(uio))
 
diff --git a/include/os/linux/zfs/sys/zfs_znode_impl.h b/include/os/linux/zfs/sys/zfs_znode_impl.h
index 0a6273442..de46fc8f2 100644
--- a/include/os/linux/zfs/sys/zfs_znode_impl.h
+++ b/include/os/linux/zfs/sys/zfs_znode_impl.h
@@ -71,6 +71,7 @@ extern "C" {
 #define	Z_ISDIR(type)	S_ISDIR(type)
 
 #define	zn_has_cached_data(zp)		((zp)->z_is_mapped)
+#define	zn_flush_cached_data(zp, sync)	write_inode_now(ZTOI(zp), sync)
 #define	zn_rlimit_fsize(zp, uio)	(0)
 
 /*
diff --git a/include/sys/dnode.h b/include/sys/dnode.h
index 2cdc5b879..af8775b9e 100644
--- a/include/sys/dnode.h
+++ b/include/sys/dnode.h
@@ -425,6 +425,7 @@ boolean_t dnode_add_ref(dnode_t *dn, void *ref);
 void dnode_rele(dnode_t *dn, void *ref);
 void dnode_rele_and_unlock(dnode_t *dn, void *tag, boolean_t evicting);
 int dnode_try_claim(objset_t *os, uint64_t object, int slots);
+boolean_t dnode_is_dirty(dnode_t *dn);
 void dnode_setdirty(dnode_t *dn, dmu_tx_t *tx);
 void dnode_set_dirtyctx(dnode_t *dn, dmu_tx_t *tx, void *tag);
 void dnode_sync(dnode_t *dn, dmu_tx_t *tx);