Use taskq for dump_bytes()

The vn_rdwr() function performs I/O by calling the vfs_write() or
vfs_read() functions.  These functions reside just below the system
call layer and the expectation is they have almost the entire 8k of
stack space to work with.  In fact, certain layered configurations
such as ext+lvm+md+multipath require the majority of this stack to
avoid stack overflows.

To avoid this posibility the vn_rdwr() call in dump_bytes() has been
moved to the ZIO_TYPE_FREE, taskq.  This ensures that all I/O will be
performed with the majority of the stack space available.  This ends
up being very similiar to as if the I/O were issued via sys_write()
or sys_read().

Signed-off-by: Brian Behlendorf <[email protected]>
Closes #1399
Closes #1423

2 files changed, 59 insertions, 6 deletions

diff --git a/module/zfs/dmu_send.c b/module/zfs/dmu_send.c
index 921c3d76f..b2c6bfe2b 100644
--- a/module/zfs/dmu_send.c
+++ b/module/zfs/dmu_send.c
@@ -39,6 +39,7 @@
 #include <sys/dsl_prop.h>
 #include <sys/dsl_pool.h>
 #include <sys/dsl_synctask.h>
+#include <sys/spa_impl.h>
 #include <sys/zfs_ioctl.h>
 #include <sys/zap.h>
 #include <sys/zio_checksum.h>
@@ -53,21 +54,48 @@ int zfs_send_corrupt_data = B_FALSE;
 
 static char *dmu_recv_tag = "dmu_recv_tag";
 
-static int
-dump_bytes(dmu_sendarg_t *dsp, void *buf, int len)
+typedef struct dump_bytes_io {
+	dmu_sendarg_t	*dbi_dsp;
+	void		*dbi_buf;
+	int		dbi_len;
+} dump_bytes_io_t;
+
+static void
+dump_bytes_strategy(void *arg)
 {
+	dump_bytes_io_t *dbi = (dump_bytes_io_t *)arg;
+	dmu_sendarg_t *dsp = dbi->dbi_dsp;
 	dsl_dataset_t *ds = dsp->dsa_os->os_dsl_dataset;
 	ssize_t resid; /* have to get resid to get detailed errno */
-	ASSERT3U(len % 8, ==, 0);
+	ASSERT3U(dbi->dbi_len % 8, ==, 0);
 
-	fletcher_4_incremental_native(buf, len, &dsp->dsa_zc);
+	fletcher_4_incremental_native(dbi->dbi_buf, dbi->dbi_len, &dsp->dsa_zc);
 	dsp->dsa_err = vn_rdwr(UIO_WRITE, dsp->dsa_vp,
-	    (caddr_t)buf, len,
+	    (caddr_t)dbi->dbi_buf, dbi->dbi_len,
 	    0, UIO_SYSSPACE, FAPPEND, RLIM64_INFINITY, CRED(), &resid);
 
 	mutex_enter(&ds->ds_sendstream_lock);
-	*dsp->dsa_off += len;
+	*dsp->dsa_off += dbi->dbi_len;
 	mutex_exit(&ds->ds_sendstream_lock);
+}
+
+static int
+dump_bytes(dmu_sendarg_t *dsp, void *buf, int len)
+{
+	dump_bytes_io_t dbi;
+
+	dbi.dbi_dsp = dsp;
+	dbi.dbi_buf = buf;
+	dbi.dbi_len = len;
+
+	/*
+	 * The vn_rdwr() call is performed in a taskq to ensure that there is
+	 * always enough stack space to write safely to the target filesystem.
+	 * The ZIO_TYPE_FREE threads are used because there can be a lot of
+	 * them and they are used in vdev_file.c for a similar purpose.
+	 */
+	spa_taskq_dispatch_sync(dmu_objset_spa(dsp->dsa_os), ZIO_TYPE_FREE,
+	    ZIO_TASKQ_ISSUE, dump_bytes_strategy, &dbi, TQ_SLEEP);
 
 	return (dsp->dsa_err);
 }
diff --git a/module/zfs/spa.c b/module/zfs/spa.c
index e022c3258..82ee445ab 100644
--- a/module/zfs/spa.c
+++ b/module/zfs/spa.c
@@ -926,6 +926,31 @@ spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q,
 	taskq_dispatch_ent(tq, func, arg, flags, ent);
 }
 
+/*
+ * Same as spa_taskq_dispatch_ent() but block on the task until completion.
+ */
+void
+spa_taskq_dispatch_sync(spa_t *spa, zio_type_t t, zio_taskq_type_t q,
+    task_func_t *func, void *arg, uint_t flags)
+{
+	spa_taskqs_t *tqs = &spa->spa_zio_taskq[t][q];
+	taskq_t *tq;
+	taskqid_t id;
+
+	ASSERT3P(tqs->stqs_taskq, !=, NULL);
+	ASSERT3U(tqs->stqs_count, !=, 0);
+
+	if (tqs->stqs_count == 1) {
+		tq = tqs->stqs_taskq[0];
+	} else {
+		tq = tqs->stqs_taskq[gethrtime() % tqs->stqs_count];
+	}
+
+	id = taskq_dispatch(tq, func, arg, flags);
+	if (id)
+		taskq_wait_id(tq, id);
+}
+
 static void
 spa_create_zio_taskqs(spa_t *spa)
 {