MFC r320352: zfs: port vdev_file part of illumos change 3306

3306 zdb should be able to issue reads in parallel
illumos/illumos-gate/31d7e8fa33fae995f558673adb22641b5aa8b6e1
https://www.illumos.org/issues/3306

The upstream change was made before we started to import upstream commits
individually.  It was imported into the illumos vendor area as r242733.
That commit was MFV-ed in r260138, but as the commit message says
vdev_file.c was left intact.

This commit actually implements the parallel I/O for vdev_file using a
taskqueue with multiple thread.  This implementation does not depend on
the illumos or FreeBSD bio interface at all, but uses zio_t to pass
around all the relevent data.  So, the code looks a bit different from
the upstream.

This commit also incorporates ZoL commit
zfsonlinux/zfs/bc25c9325b0e5ced897b9820dad239539d561ec9 that fixed
https://github.com/zfsonlinux/zfs/issues/2270
We need to use a dedicated taskqueue for exactly the same reason as ZoL
as we do not implement TASKQ_DYNAMIC.

3 files changed, 61 insertions, 20 deletions

diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa_misc.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa_misc.c
index cafa083..34b1b6e 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa_misc.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa_misc.c
@@ -39,6 +39,7 @@
 #include <sys/zap.h>
 #include <sys/zil.h>
 #include <sys/vdev_impl.h>
+#include <sys/vdev_file.h>
 #include <sys/metaslab.h>
 #include <sys/uberblock_impl.h>
 #include <sys/txg.h>
@@ -2013,6 +2014,7 @@ spa_init(int mode)
 	dmu_init();
 	zil_init();
 	vdev_cache_stat_init();
+	vdev_file_init();
 	zfs_prop_init();
 	zpool_prop_init();
 	zpool_feature_init();
@@ -2032,6 +2034,7 @@ spa_fini(void)
 
 	spa_evict_all();
 
+	vdev_file_fini();
 	vdev_cache_stat_fini();
 	zil_fini();
 	dmu_fini();
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/vdev_file.h b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/vdev_file.h
index cd49673..0260b4a 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/vdev_file.h
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/vdev_file.h
@@ -39,6 +39,9 @@ typedef struct vdev_file {
 	vnode_t		*vf_vnode;
 } vdev_file_t;
 
+extern void vdev_file_init(void);
+extern void vdev_file_fini(void);
+
 #ifdef	__cplusplus
 }
 #endif
diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_file.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_file.c
index 8228af1..e083000 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_file.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/vdev_file.c
@@ -35,6 +35,21 @@
  * Virtual device vector for files.
  */
 
+static taskq_t *vdev_file_taskq;
+
+void
+vdev_file_init(void)
+{
+	vdev_file_taskq = taskq_create("z_vdev_file", MAX(max_ncpus, 16),
+	    minclsyspri, max_ncpus, INT_MAX, 0);
+}
+
+void
+vdev_file_fini(void)
+{
+	taskq_destroy(vdev_file_taskq);
+}
+
 static void
 vdev_file_hold(vdev_t *vd)
 {
@@ -156,27 +171,58 @@ vdev_file_close(vdev_t *vd)
 	vd->vdev_tsd = NULL;
 }
 
+/*
+ * Implements the interrupt side for file vdev types. This routine will be
+ * called when the I/O completes allowing us to transfer the I/O to the
+ * interrupt taskqs. For consistency, the code structure mimics disk vdev
+ * types.
+ */
 static void
-vdev_file_io_start(zio_t *zio)
+vdev_file_io_intr(zio_t *zio)
 {
+	zio_delay_interrupt(zio);
+}
+
+static void
+vdev_file_io_strategy(void *arg)
+{
+	zio_t *zio = arg;
 	vdev_t *vd = zio->io_vd;
 	vdev_file_t *vf;
 	vnode_t *vp;
 	ssize_t resid;
 
-	if (!vdev_readable(vd)) {
-		zio->io_error = SET_ERROR(ENXIO);
-		zio_interrupt(zio);
-		return;
-	}
-
 	vf = vd->vdev_tsd;
 	vp = vf->vf_vnode;
 
+	ASSERT(zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE);
+	zio->io_error = vn_rdwr(zio->io_type == ZIO_TYPE_READ ?
+	    UIO_READ : UIO_WRITE, vp, zio->io_data, zio->io_size,
+	    zio->io_offset, UIO_SYSSPACE, 0, RLIM64_INFINITY, kcred, &resid);
+
+	if (resid != 0 && zio->io_error == 0)
+		zio->io_error = ENOSPC;
+
+	vdev_file_io_intr(zio);
+}
+
+static void
+vdev_file_io_start(zio_t *zio)
+{
+	vdev_t *vd = zio->io_vd;
+	vdev_file_t *vf = vd->vdev_tsd;
+
 	if (zio->io_type == ZIO_TYPE_IOCTL) {
+		/* XXPOLICY */
+		if (!vdev_readable(vd)) {
+			zio->io_error = SET_ERROR(ENXIO);
+			zio_interrupt(zio);
+			return;
+		}
+
 		switch (zio->io_cmd) {
 		case DKIOCFLUSHWRITECACHE:
-			zio->io_error = VOP_FSYNC(vp, FSYNC | FDSYNC,
+			zio->io_error = VOP_FSYNC(vf->vf_vnode, FSYNC | FDSYNC,
 			    kcred, NULL);
 			break;
 		default:
@@ -190,19 +236,8 @@ vdev_file_io_start(zio_t *zio)
 	ASSERT(zio->io_type == ZIO_TYPE_READ || zio->io_type == ZIO_TYPE_WRITE);
 	zio->io_target_timestamp = zio_handle_io_delay(zio);
 
-	zio->io_error = vn_rdwr(zio->io_type == ZIO_TYPE_READ ?
-	    UIO_READ : UIO_WRITE, vp, zio->io_data, zio->io_size,
-	    zio->io_offset, UIO_SYSSPACE, 0, RLIM64_INFINITY, kcred, &resid);
-
-	if (resid != 0 && zio->io_error == 0)
-		zio->io_error = ENOSPC;
-
-	zio_delay_interrupt(zio);
-
-#ifdef illumos
-	VERIFY3U(taskq_dispatch(system_taskq, vdev_file_io_strategy, bp,
+	VERIFY3U(taskq_dispatch(vdev_file_taskq, vdev_file_io_strategy, zio,
 	    TQ_SLEEP), !=, 0);
-#endif
 }
 
 /* ARGSUSED */