From f66aea7fc31aa8a0da17ff9b63657af0aa38f14f Mon Sep 17 00:00:00 2001
From: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Date: Tue, 10 Sep 2019 11:44:11 +0900
Subject: [PATCH] bdevperf: Use zcopy APIs for simple write I/O case

Add a bdev_io pointer to struct bdevperf_task.

In bdevperf_zcopy_get_buf_complete() which is the callback to
spdk_bdev_zcopy_start(), bdev_io is saved into the current task.

Then bdevperf_submit_task() will call spdk_bdev_zcopy_end()
by using the saved bdev_io.

Besides, when spdk_bdev_zcopy_start() is called with populate=false,
increment target->current_queue_depth when it completes successfully,
and do not increment target->current_queue_depth when the corresponding
spdk_bdev_zcopy_end() is called with commit=true.

The reason is that IO processing is already started when
spdk_bdev_zcopy_start() is called.

The next patch will use zcopy APIs for write I/O with verify or
reset cases.

Signed-off-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Change-Id: I12f3b1ccac726abe345a64f06e33d65d2a3538fd
Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/467900
Tested-by: SPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: Changpeng Liu <changpeng.liu@intel.com>
Reviewed-by: Paul Luse <paul.e.luse@intel.com>
---
 test/bdev/bdevperf/bdevperf.c | 70 +++++++++++++++++++++++++++++------
 1 file changed, 59 insertions(+), 11 deletions(-)

diff --git a/test/bdev/bdevperf/bdevperf.c b/test/bdev/bdevperf/bdevperf.c
index 1593f3b22..30049b3ae 100644
--- a/test/bdev/bdevperf/bdevperf.c
+++ b/test/bdev/bdevperf/bdevperf.c
@@ -47,6 +47,7 @@
 struct bdevperf_task {
 	struct iovec			iov;
 	struct io_target		*target;
+	struct spdk_bdev_io		*bdev_io;
 	void				*buf;
 	void				*md_buf;
 	uint64_t			offset_blocks;
@@ -616,17 +617,22 @@ bdevperf_submit_task(void *arg)
 		if (rc == 0) {
 			cb_fn = (g_verify || g_reset) ? bdevperf_verify_write_complete : bdevperf_complete;
 
-			if (spdk_bdev_is_md_separate(target->bdev)) {
-				rc = spdk_bdev_writev_blocks_with_md(desc, ch, &task->iov, 1,
-								     task->md_buf,
+			if (g_zcopy && !g_verify && !g_reset) {
+				spdk_bdev_zcopy_end(task->bdev_io, true, cb_fn, task);
+				return;
+			} else {
+				if (spdk_bdev_is_md_separate(target->bdev)) {
+					rc = spdk_bdev_writev_blocks_with_md(desc, ch, &task->iov, 1,
+									     task->md_buf,
+									     task->offset_blocks,
+									     target->io_size_blocks,
+									     cb_fn, task);
+				} else {
+					rc = spdk_bdev_writev_blocks(desc, ch, &task->iov, 1,
 								     task->offset_blocks,
 								     target->io_size_blocks,
 								     cb_fn, task);
-			} else {
-				rc = spdk_bdev_writev_blocks(desc, ch, &task->iov, 1,
-							     task->offset_blocks,
-							     target->io_size_blocks,
-							     cb_fn, task);
+				}
 			}
 		}
 		break;
@@ -677,6 +683,43 @@ bdevperf_submit_task(void *arg)
 	target->current_queue_depth++;
 }
 
+static void
+bdevperf_zcopy_get_buf_complete(struct spdk_bdev_io *bdev_io, bool success, void *cb_arg)
+{
+	struct bdevperf_task	*task = cb_arg;
+	struct io_target	*target = task->target;
+
+	if (!success) {
+		target->is_draining = true;
+		g_run_failed = true;
+		return;
+	}
+
+	task->bdev_io = bdev_io;
+	task->io_type = SPDK_BDEV_IO_TYPE_WRITE;
+
+	bdevperf_submit_task(task);
+}
+
+static void
+bdevperf_prep_zcopy_write_task(void *arg)
+{
+	struct bdevperf_task	*task = arg;
+	struct io_target	*target = task->target;
+	int			rc;
+
+	rc = spdk_bdev_zcopy_start(target->bdev_desc, target->ch,
+				   task->offset_blocks, target->io_size_blocks,
+				   false, bdevperf_zcopy_get_buf_complete, task);
+	if (rc != 0) {
+		assert(rc == -ENOMEM);
+		bdevperf_queue_io_wait_with_cb(task, bdevperf_prep_zcopy_write_task);
+		return;
+	}
+
+	target->current_queue_depth++;
+}
+
 static __thread unsigned int seed = 0;
 
 static void
@@ -713,9 +756,14 @@ bdevperf_prep_task(struct bdevperf_task *task)
 		   (g_rw_percentage != 0 && ((rand_r(&seed) % 100) < g_rw_percentage))) {
 		task->io_type = SPDK_BDEV_IO_TYPE_READ;
 	} else {
-		task->iov.iov_base = task->buf;
-		task->iov.iov_len = g_buf_size;
-		task->io_type = SPDK_BDEV_IO_TYPE_WRITE;
+		if (g_zcopy) {
+			bdevperf_prep_zcopy_write_task(task);
+			return;
+		} else {
+			task->iov.iov_base = task->buf;
+			task->iov.iov_len = g_buf_size;
+			task->io_type = SPDK_BDEV_IO_TYPE_WRITE;
+		}
 	}
 
 	bdevperf_submit_task(task);