[PATCH 0/9] target: Add support for EXTENDED_COPY (VAAI) offload emulation

[PATCH 0/9] target: Add support for EXTENDED_COPY (VAAI) offload emulation

[Nicholas A. Bellinger]

From: Nicholas Bellinger <nab@daterainc.com>

Hi folks!

This series adds support to target-core for generic EXTENDED_COPY offload
emulation as defined by SPC-4 using virtual (IBLOCK, FILEIO, RAMDISK)
backends.

EXTENDED_COPY is a VMWare ESX VAAI primative that is used to perform copy
offload, that allows a target to perform local READ + WRITE I/O requests
for bulk data transfers (cloning a virtual machine for example), instead
of requiring these I/Os to actually be sent to/from the requesting SCSI
initiator port.

This implemenation fully supports copy offload between the same device
backend, and across multiple device backends.  It supports copy offload
transparently across multiple target ports of different fabrics, eg:
iSCSI -> FC, FC -> iSER, iSER -> FCoE and so on.

It also supports both PUSH and PULL models of operation, so the actual
EXTENDED_COPY CDB may be received on either source or destination logical
unit.

For Target Descriptors, it currently supports the NAA IEEE Registered
Extended designator (type 0xe4), which allows the reference of target
ports to occur independent of fabric type using EVPD 0x83 WWNs.  For
Segment Descriptors, it currently supports copy from block to block
(0x02) mode.

Here's a quick snippet of the code in action with sg_xcopy performing
copy offload between two IBLOCK and FILEIO backends:

[  644.638215] Processing XCOPY with list_id: 0x00 list_id_usage: 0x10 tdll: 64 sdll: 28 inline_dl: 0
[  644.648227] XCOPY 0xe4: RELATIVE INITIATOR PORT IDENTIFIER: 0
[  644.654639] XCOPY 0xe4: desig_len: 16
[  644.658722] XCOPY 0xe4: Set xop->src_dev ffff88045d770000 from source received xop
[  644.667179] XCOPY 0xe4: RELATIVE INITIATOR PORT IDENTIFIER: 0
[  644.673597] XCOPY 0xe4: desig_len: 16
[  644.677699] XCOPY 0xe4: Setting xop->dst_dev: ffff88045d771048 from located se_dev
[  644.686297] Called configfs_depend_item for subsys: ffffffffa00f2570 se_dev: ffff88045d771048 se_dev->se_dev_group: ffff88045d7714f8
[  644.699607] XCOPY TGT desc: Source dev: ffff88045d770000 NAA IEEE WWN: 0x6001405d2e0745b08564acea3ca401e5
[  644.710296] XCOPY TGT desc: Dest dev: ffff88045d771048 NAA IEEE WWN: 0x60014056da9d8672d4b437596ab764b3
[  644.720782] XCOPY: Processed 2 target descriptors, length: 64
[  644.727203] XCOPY seg desc 0x02: desc_len: 24 stdi: 0 dtdi: 1, DC: 2
[  644.734304] XCOPY seg desc 0x02: nolb: 1 src_lba: 0 dst_lba: 0
[  644.740819] XCOPY seg desc 0x02: DC=1 w/ dbl: 0
[  644.745881] XCOPY: Processed 1 segment descriptors, length: 28
[  644.752402] target_xcopy_do_work: nolb: 1, max_nolb: 1024 end_lba: 1
[  644.759504] target_xcopy_do_work: Starting src_lba: 0, dst_lba: 0
[  644.766303] target_xcopy_do_work: Calling read src_dev: ffff88045d770000 src_lba: 0, cur_nolb: 1
[  644.776115] XCOPY: Built READ_16: LBA: 0 Sectors: 1 Length: 512
[  644.782751] Honoring local SRC port from ec_cmd->se_dev: ffff88045d770000
[  644.790335] Honoring local SRC port from ec_cmd->se_lun: ffff88085a1977e0
[  644.797921] XCOPY-READ: Saved xop->xop_data_sg: ffff880459d3e3a8, num: 1 for READ memory
[  644.807203] target_xcopy_issue_pt_cmd(): SCSI status: 0x00
[  644.813333] target_xcopy_do_work: Incremented READ src_lba to 1
[  644.819947] target_xcopy_do_work: Calling write dst_dev: ffff88045d771048 dst_lba: 0, cur_nolb: 1
[  644.829854] XCOPY: Built WRITE_16: LBA: 0 Sectors: 1 Length: 512
[  644.836568] Setup emulated se_dev: ffff88045d771048 from se_dev
[  644.843185] Setup emulated se_dev: ffff88045d771048 to pt_cmd->se_lun->lun_se_dev
[  644.851545] Setup emulated remote DEST xcopy_pt_port: ffffffffa00f7610 to cmd->se_lun->lun_sep for X-COPY data PUSH
[  644.863198] Setup PASSTHROUGH_NOALLOC t_data_sg: ffff880459d3e3a8 t_data_nents: 1
[  644.895203] target_xcopy_issue_pt_cmd(): SCSI status: 0x00
[  644.901332] target_xcopy_do_work: Incremented WRITE dst_lba to 1
[  644.908044] Calling configfs_undepend_item for subsys: ffffffffa00f2570 remote_dev: ffff88045d771048 remote_dev->dev_group: ffff88045d7714f8
[  644.922129] target_xcopy_do_work: Final src_lba: 1, dst_lba: 1
[  644.928646] target_xcopy_do_work: Blocks copied: 1, Bytes Copied: 512
[  644.935840] target_xcopy_do_work: Setting X-COPY GOOD status -> sending response

For all intensive purposes this code is completely standalone, and the amount
of changes required to enable it's function within target-core code is small.

Please review as v3.12 material.

Thank you,

--nab

Nicholas Bellinger (9):
  target: Make target_core_subsystem defined as non static
  target: Make spc_parse_naa_6h_vendor_specific non static
  target: Make helpers non static for EXTENDED_COPY command setup
  target: Add global device list for EXTENDED_COPY
  target: Avoid non-existent tg_pt_gp_mem in target_alua_state_check
  target: Add support for EXTENDED_COPY copy offload emulation
  target: Enable EXTENDED_COPY setup in spc_parse_cdb
  target: Add Third Party Copy (3PC) bit in INQUIRY response
  target: Enable global EXTENDED_COPY setup/release

 drivers/target/Makefile                |    3 +-
 drivers/target/target_core_alua.c      |    3 +
 drivers/target/target_core_configfs.c  |   12 +-
 drivers/target/target_core_device.c    |   27 +
 drivers/target/target_core_internal.h  |    1 +
 drivers/target/target_core_spc.c       |   20 +-
 drivers/target/target_core_transport.c |    5 +-
 drivers/target/target_core_xcopy.c     | 1122 ++++++++++++++++++++++++++++++++
 drivers/target/target_core_xcopy.h     |   62 ++
 include/target/target_core_backend.h   |    4 +
 include/target/target_core_base.h      |    5 +
 11 files changed, 1255 insertions(+), 9 deletions(-)
 create mode 100644 drivers/target/target_core_xcopy.c
 create mode 100644 drivers/target/target_core_xcopy.h

-- 
1.7.2.5

[PATCH 1/9] target: Make target_core_subsystem defined as non static

[Nicholas A. Bellinger]

From: Nicholas Bellinger <nab@daterainc.com>

This patch makes the top-level target_core_subsystem array available
to other target code, which is required by EXTENDED_COPY to pin the
backend se_device using configfs_depend_item(), in order to ensure
it can't be removed for the duration of a EXTENDED_COPY operation.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Martin Petersen <martin.petersen@oracle.com>
Cc: Chris Mason <chris.mason@fusionio.com>
Cc: Roland Dreier <roland@purestorage.com>
Cc: Zach Brown <zab@redhat.com>
Cc: James Bottomley <JBottomley@Parallels.com>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Nicholas Bellinger <nab@daterainc.com>
---
 drivers/target/target_core_configfs.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/target/target_core_configfs.c b/drivers/target/target_core_configfs.c
index 24517d4..939ecc5 100644
--- a/drivers/target/target_core_configfs.c
+++ b/drivers/target/target_core_configfs.c
@@ -268,7 +268,7 @@ static struct configfs_subsystem target_core_fabrics = {
 	},
 };
 
-static struct configfs_subsystem *target_core_subsystem[] = {
+struct configfs_subsystem *target_core_subsystem[] = {
 	&target_core_fabrics,
 	NULL,
 };
-- 
1.7.2.5

[PATCH 2/9] target: Make spc_parse_naa_6h_vendor_specific non static

[Nicholas A. Bellinger]

From: Nicholas Bellinger <nab@daterainc.com>

This patch makes spc_parse_naa_6h_vendor_specific() available to
other target code, which is required by EXTENDED_COPY when comparing
the received NAA WWN device identifer for locating the associated
se_device backend.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Martin Petersen <martin.petersen@oracle.com>
Cc: Chris Mason <chris.mason@fusionio.com>
Cc: Roland Dreier <roland@purestorage.com>
Cc: Zach Brown <zab@redhat.com>
Cc: James Bottomley <JBottomley@Parallels.com>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Nicholas Bellinger <nab@daterainc.com>
---
 drivers/target/target_core_spc.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/target/target_core_spc.c b/drivers/target/target_core_spc.c
index ed7077a..bd16a93 100644
--- a/drivers/target/target_core_spc.c
+++ b/drivers/target/target_core_spc.c
@@ -126,8 +126,8 @@ spc_emulate_evpd_80(struct se_cmd *cmd, unsigned char *buf)
 	return 0;
 }
 
-static void spc_parse_naa_6h_vendor_specific(struct se_device *dev,
-		unsigned char *buf)
+void spc_parse_naa_6h_vendor_specific(struct se_device *dev,
+				      unsigned char *buf)
 {
 	unsigned char *p = &dev->t10_wwn.unit_serial[0];
 	int cnt;
-- 
1.7.2.5

[PATCH 3/9] target: Make helpers non static for EXTENDED_COPY command setup

[Nicholas A. Bellinger]

From: Nicholas Bellinger <nab@daterainc.com>

Both transport_generic_get_mem() and transport_generic_map_mem_to_cmd()
are required by EXTENDED_COPY logic when setting up internally
dispatched command descriptors, so go ahead and make both of these
non static.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Martin Petersen <martin.petersen@oracle.com>
Cc: Chris Mason <chris.mason@fusionio.com>
Cc: Roland Dreier <roland@purestorage.com>
Cc: Zach Brown <zab@redhat.com>
Cc: James Bottomley <JBottomley@Parallels.com>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Nicholas Bellinger <nab@daterainc.com>
---
 drivers/target/target_core_transport.c |    5 ++---
 include/target/target_core_backend.h   |    4 ++++
 2 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/drivers/target/target_core_transport.c b/drivers/target/target_core_transport.c
index 3009cda..2f9c402 100644
--- a/drivers/target/target_core_transport.c
+++ b/drivers/target/target_core_transport.c
@@ -67,7 +67,6 @@ struct kmem_cache *t10_alua_tg_pt_gp_mem_cache;
 static void transport_complete_task_attr(struct se_cmd *cmd);
 static void transport_handle_queue_full(struct se_cmd *cmd,
 		struct se_device *dev);
-static int transport_generic_get_mem(struct se_cmd *cmd);
 static int transport_put_cmd(struct se_cmd *cmd);
 static void target_complete_ok_work(struct work_struct *work);
 
@@ -1254,7 +1253,7 @@ int transport_handle_cdb_direct(
 }
 EXPORT_SYMBOL(transport_handle_cdb_direct);
 
-static sense_reason_t
+sense_reason_t
 transport_generic_map_mem_to_cmd(struct se_cmd *cmd, struct scatterlist *sgl,
 		u32 sgl_count, struct scatterlist *sgl_bidi, u32 sgl_bidi_count)
 {
@@ -2164,7 +2163,7 @@ out:
 	return -ENOMEM;
 }
 
-static int
+int
 transport_generic_get_mem(struct se_cmd *cmd)
 {
 	u32 length = cmd->data_length;
diff --git a/include/target/target_core_backend.h b/include/target/target_core_backend.h
index 77f25e0..9f07231 100644
--- a/include/target/target_core_backend.h
+++ b/include/target/target_core_backend.h
@@ -74,6 +74,10 @@ int	transport_set_vpd_ident(struct t10_vpd *, unsigned char *);
 /* core helpers also used by command snooping in pscsi */
 void	*transport_kmap_data_sg(struct se_cmd *);
 void	transport_kunmap_data_sg(struct se_cmd *);
+/* core helpers also used by xcopy during internal command setup */
+int	transport_generic_get_mem(struct se_cmd *);
+sense_reason_t	transport_generic_map_mem_to_cmd(struct se_cmd *,
+		struct scatterlist *, u32, struct scatterlist *, u32);
 
 void	array_free(void *array, int n);
 
-- 
1.7.2.5

[PATCH 4/9] target: Add global device list for EXTENDED_COPY

[Nicholas A. Bellinger]

From: Nicholas Bellinger <nab@daterainc.com>

EXTENDED_COPY needs to be able to search a global list of devices
based on NAA WWN device identifiers, so add a simple g_device_list
protected by g_device_mutex.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Martin Petersen <martin.petersen@oracle.com>
Cc: Chris Mason <chris.mason@fusionio.com>
Cc: Roland Dreier <roland@purestorage.com>
Cc: Zach Brown <zab@redhat.com>
Cc: James Bottomley <JBottomley@Parallels.com>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Nicholas Bellinger <nab@daterainc.com>
---
 drivers/target/target_core_device.c |   13 +++++++++++++
 include/target/target_core_base.h   |    1 +
 2 files changed, 14 insertions(+), 0 deletions(-)

diff --git a/drivers/target/target_core_device.c b/drivers/target/target_core_device.c
index de89046..458944e 100644
--- a/drivers/target/target_core_device.c
+++ b/drivers/target/target_core_device.c
@@ -47,6 +47,9 @@
 #include "target_core_pr.h"
 #include "target_core_ua.h"
 
+DEFINE_MUTEX(g_device_mutex);
+LIST_HEAD(g_device_list);
+
 static struct se_hba *lun0_hba;
 /* not static, needed by tpg.c */
 struct se_device *g_lun0_dev;
@@ -1406,6 +1409,7 @@ struct se_device *target_alloc_device(struct se_hba *hba, const char *name)
 	INIT_LIST_HEAD(&dev->delayed_cmd_list);
 	INIT_LIST_HEAD(&dev->state_list);
 	INIT_LIST_HEAD(&dev->qf_cmd_list);
+	INIT_LIST_HEAD(&dev->g_dev_node);
 	spin_lock_init(&dev->stats_lock);
 	spin_lock_init(&dev->execute_task_lock);
 	spin_lock_init(&dev->delayed_cmd_lock);
@@ -1525,6 +1529,11 @@ int target_configure_device(struct se_device *dev)
 	spin_lock(&hba->device_lock);
 	hba->dev_count++;
 	spin_unlock(&hba->device_lock);
+
+	mutex_lock(&g_device_mutex);
+	list_add_tail(&dev->g_dev_node, &g_device_list);
+	mutex_unlock(&g_device_mutex);
+
 	return 0;
 
 out_free_alua:
@@ -1543,6 +1552,10 @@ void target_free_device(struct se_device *dev)
 	if (dev->dev_flags & DF_CONFIGURED) {
 		destroy_workqueue(dev->tmr_wq);
 
+		mutex_lock(&g_device_mutex);
+		list_del(&dev->g_dev_node);
+		mutex_unlock(&g_device_mutex);
+
 		spin_lock(&hba->device_lock);
 		hba->dev_count--;
 		spin_unlock(&hba->device_lock);
diff --git a/include/target/target_core_base.h b/include/target/target_core_base.h
index 0783b2c..6b14f3c 100644
--- a/include/target/target_core_base.h
+++ b/include/target/target_core_base.h
@@ -686,6 +686,7 @@ struct se_device {
 	struct list_head	delayed_cmd_list;
 	struct list_head	state_list;
 	struct list_head	qf_cmd_list;
+	struct list_head	g_dev_node;
 	/* Pointer to associated SE HBA */
 	struct se_hba		*se_hba;
 	/* T10 Inquiry and VPD WWN Information */
-- 
1.7.2.5

[PATCH 5/9] target: Avoid non-existent tg_pt_gp_mem in target_alua_state_check

[Nicholas A. Bellinger]

From: Nicholas Bellinger <nab@daterainc.com>

This patch adds an check for a non-existent port->sep_alua_tg_pt_gp_mem
within target_alua_state_check(), which is not present for internally
dispatched EXTENDED_COPY WRITE I/O to the destination target port.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Martin Petersen <martin.petersen@oracle.com>
Cc: Chris Mason <chris.mason@fusionio.com>
Cc: Roland Dreier <roland@purestorage.com>
Cc: Zach Brown <zab@redhat.com>
Cc: James Bottomley <JBottomley@Parallels.com>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Nicholas Bellinger <nab@daterainc.com>
---
 drivers/target/target_core_alua.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)

diff --git a/drivers/target/target_core_alua.c b/drivers/target/target_core_alua.c
index 5403186..ea928c4 100644
--- a/drivers/target/target_core_alua.c
+++ b/drivers/target/target_core_alua.c
@@ -557,6 +557,9 @@ target_alua_state_check(struct se_cmd *cmd)
 	 * a ALUA logical unit group.
 	 */
 	tg_pt_gp_mem = port->sep_alua_tg_pt_gp_mem;
+	if (!tg_pt_gp_mem)
+		return 0;
+
 	spin_lock(&tg_pt_gp_mem->tg_pt_gp_mem_lock);
 	tg_pt_gp = tg_pt_gp_mem->tg_pt_gp;
 	out_alua_state = atomic_read(&tg_pt_gp->tg_pt_gp_alua_access_state);
-- 
1.7.2.5

[PATCH 6/9] target: Add support for EXTENDED_COPY copy offload emulation

[Nicholas A. Bellinger]

From: Nicholas Bellinger <nab@daterainc.com>

This patch adds support for EXTENDED_COPY emulation from SPC-3, that
enables full copy offload target support within both a single virtual
backend device, and across multiple virtual backend devices.  It also
functions independent of target fabric, and supports copy offload
across multiple target fabric ports.

This implemenation supports both EXTENDED_COPY PUSH and PULL models
of operation, so the actual CDB may be received on either source or
desination logical unit.

For Target Descriptors, it currently supports the NAA IEEE Registered
Extended designator (type 0xe4), which allows the reference of target
ports to occur independent of fabric type using EVPD 0x83 WWNs.

For Segment Descriptors, it currently supports copy from block to
block (0x02) mode.

It also honors any present SCSI reservations of the destination target
port.  Note that only Supports No List Identifier (SNLID=1) mode is
supported.

Also included is basic RECEIVE_COPY_RESULTS with service action type
OPERATING PARAMETERS (0x03) required for SNLID=1 operation.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Martin Petersen <martin.petersen@oracle.com>
Cc: Chris Mason <chris.mason@fusionio.com>
Cc: Roland Dreier <roland@purestorage.com>
Cc: Zach Brown <zab@redhat.com>
Cc: James Bottomley <JBottomley@Parallels.com>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Nicholas Bellinger <nab@daterainc.com>
---
 drivers/target/Makefile            |    3 +-
 drivers/target/target_core_xcopy.c | 1122 ++++++++++++++++++++++++++++++++++++
 drivers/target/target_core_xcopy.h |   62 ++
 include/target/target_core_base.h  |    1 +
 4 files changed, 1187 insertions(+), 1 deletions(-)
 create mode 100644 drivers/target/target_core_xcopy.c
 create mode 100644 drivers/target/target_core_xcopy.h

diff --git a/drivers/target/Makefile b/drivers/target/Makefile
index 9fdcb56..85b012d 100644
--- a/drivers/target/Makefile
+++ b/drivers/target/Makefile
@@ -13,7 +13,8 @@ target_core_mod-y		:= target_core_configfs.o \
 				   target_core_spc.o \
 				   target_core_ua.o \
 				   target_core_rd.o \
-				   target_core_stat.o
+				   target_core_stat.o \
+				   target_core_xcopy.o
 
 obj-$(CONFIG_TARGET_CORE)	+= target_core_mod.o
 
diff --git a/drivers/target/target_core_xcopy.c b/drivers/target/target_core_xcopy.c
new file mode 100644
index 0000000..e0fabea
--- /dev/null
+++ b/drivers/target/target_core_xcopy.c
@@ -0,0 +1,1122 @@
+/*******************************************************************************
+ * Filename: target_core_xcopy.c
+ *
+ * This file contains support for SPC-4 Extended-Copy offload with generic
+ * TCM backends.
+ *
+ * Copyright (c) 2011-2013 Datera, Inc. All rights reserved.
+ *
+ * Author:
+ * Nicholas A. Bellinger <nab@daterainc.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ ******************************************************************************/
+
+#include <linux/version.h>
+#include <linux/slab.h>
+#include <linux/spinlock.h>
+#include <linux/list.h>
+#include <linux/configfs.h>
+#include <scsi/scsi.h>
+#include <scsi/scsi_cmnd.h>
+#include <asm/unaligned.h>
+
+#include <target/target_core_base.h>
+#include <target/target_core_backend.h>
+#include <target/target_core_fabric.h>
+#include <target/target_core_configfs.h>
+
+#include "target_core_pr.h"
+#include "target_core_ua.h"
+#include "target_core_xcopy.h"
+
+/* #define XCOPY_DBG_CTL */
+#ifdef XCOPY_DBG_CTL
+#define XCOPY_CTL(x...) printk(KERN_INFO x)
+#else
+#define XCOPY_CTL(x...)
+#endif
+
+/* #define XCOPY_DBG_IO */
+#ifdef XCOPY_DBG_IO
+#define XCOPY_IO(x...) printk(KERN_INFO x)
+#else
+#define XCOPY_IO(x...)
+#endif
+
+static struct workqueue_struct *xcopy_wq = NULL;
+/*
+ * From target_core_spc.c
+ */
+extern void spc_parse_naa_6h_vendor_specific(struct se_device *, unsigned char *);
+/*
+ * From target_core_device.c
+ */
+extern struct mutex g_device_mutex;
+extern struct list_head g_device_list;
+/*
+ * From target_core_configfs.c
+ */
+extern struct configfs_subsystem *target_core_subsystem[];
+
+static int target_xcopy_gen_naa_ieee(struct se_device *dev, unsigned char *buf)
+{
+	int off = 0;
+
+	buf[off++] = (0x6 << 4);
+	buf[off++] = 0x01;
+	buf[off++] = 0x40;
+	buf[off] = (0x5 << 4);
+
+	spc_parse_naa_6h_vendor_specific(dev, &buf[off]);
+	return 0;
+}
+
+static int target_xcopy_locate_se_dev_e4(struct se_cmd *se_cmd, struct xcopy_op *xop,
+					bool src)
+{
+	struct se_device *se_dev;
+	struct configfs_subsystem *subsys = target_core_subsystem[0];
+	unsigned char tmp_dev_wwn[XCOPY_NAA_IEEE_REGEX_LEN], *dev_wwn;
+	int rc;
+
+	if (src == true)
+		dev_wwn = &xop->dst_tid_wwn[0];
+	else
+		dev_wwn = &xop->src_tid_wwn[0];
+
+	mutex_lock(&g_device_mutex);
+	list_for_each_entry(se_dev, &g_device_list, g_dev_node) {
+
+		memset(&tmp_dev_wwn[0], 0, XCOPY_NAA_IEEE_REGEX_LEN);
+		target_xcopy_gen_naa_ieee(se_dev, &tmp_dev_wwn[0]);
+
+		rc = memcmp(&tmp_dev_wwn[0], dev_wwn, XCOPY_NAA_IEEE_REGEX_LEN);
+		if (rc != 0)
+			continue;
+
+		if (src == true) {
+			xop->dst_dev = se_dev;
+			XCOPY_CTL("XCOPY 0xe4: Setting xop->dst_dev: %p from located"
+				" se_dev\n", xop->dst_dev);
+		} else {
+			xop->src_dev = se_dev;
+			XCOPY_CTL("XCOPY 0xe4: Setting xop->src_dev: %p from located"
+				" se_dev\n", xop->src_dev);
+		}
+
+		rc = configfs_depend_item(subsys,
+				&se_dev->dev_group.cg_item);
+		if (rc != 0) {
+			pr_err("configfs_depend_item attempt failed:"
+				" %d for se_dev: %p\n", rc, se_dev);
+			mutex_unlock(&g_device_mutex);
+			return rc;
+		}
+
+		XCOPY_CTL("Called configfs_depend_item for subsys: %p se_dev: %p"
+			" se_dev->se_dev_group: %p\n", subsys, se_dev,
+			&se_dev->dev_group);
+
+		mutex_unlock(&g_device_mutex);
+		return 0;
+	}
+	mutex_unlock(&g_device_mutex);
+
+	pr_err("Unable to locate 0xe4 descriptor for EXTENDED_COPY\n");
+	return -EINVAL;
+}
+
+static int target_xcopy_parse_tiddesc_e4(struct se_cmd *se_cmd, struct xcopy_op *xop,
+				unsigned char *p, bool src)
+{
+	unsigned char *desc = p;
+	unsigned short ript;
+	u8 desig_len;
+	/*
+	 * Extract RELATIVE INITIATOR PORT IDENTIFIER
+	 */
+	ript = get_unaligned_be16(&desc[2]);
+	XCOPY_CTL("XCOPY 0xe4: RELATIVE INITIATOR PORT IDENTIFIER: %hu\n", ript);
+	/*
+	 * Check for supported code set, association, and designator type
+	 */
+	if ((desc[4] & 0x0f) != 0x1) {
+		pr_err("XCOPY 0xe4: code set of non binary type not supported\n");
+		return -EINVAL;
+	}
+	if ((desc[5] & 0x30) != 0x00) {
+		pr_err("XCOPY 0xe4: association other than LUN not supported\n");
+		return -EINVAL;
+	}
+	if ((desc[5] & 0x0f) != 0x3) {
+		pr_err("XCOPY 0xe4: designator type unsupported: 0x%02x\n",
+				(desc[5] & 0x0f));
+		return -EINVAL;
+	}
+	/*
+	 * Check for matching 16 byte length for NAA IEEE Registered Extended
+	 * Assigned designator
+	 */
+	desig_len = desc[7];
+	if (desig_len != 16) {
+		pr_err("XCOPY 0xe4: invalid desig_len: %d\n", (int)desig_len);
+		return -EINVAL;
+	}
+	XCOPY_CTL("XCOPY 0xe4: desig_len: %d\n", (int)desig_len);
+	/*
+	 * Check for NAA IEEE Registered Extended Assigned header..
+	 */
+	if ((desc[8] & 0xf0) != 0x60) {
+		pr_err("XCOPY 0xe4: Unsupported DESIGNATOR TYPE: 0x%02x\n",
+					(desc[8] & 0xf0));
+		return -EINVAL;
+	}
+
+	if (src == true) {
+		memcpy(&xop->src_tid_wwn[0], &desc[8], XCOPY_NAA_IEEE_REGEX_LEN);
+		/*
+		 * Determine if the source designator matches the local device
+		 */
+		if (!memcmp(&xop->local_dev_wwn[0], &xop->src_tid_wwn[0],
+				XCOPY_NAA_IEEE_REGEX_LEN)) {
+			xop->op_origin = XCOL_SOURCE_RECV_OP;
+			xop->src_dev = se_cmd->se_dev;
+			XCOPY_CTL("XCOPY 0xe4: Set xop->src_dev %p from source"
+					" received xop\n", xop->src_dev);
+		}
+	} else {
+		memcpy(&xop->dst_tid_wwn[0], &desc[8], XCOPY_NAA_IEEE_REGEX_LEN);
+		/*
+		 * Determine if the destination designator matches the local device
+		 */
+		if (!memcmp(&xop->local_dev_wwn[0], &xop->dst_tid_wwn[0],
+				XCOPY_NAA_IEEE_REGEX_LEN)) {
+			xop->op_origin = XCOL_DEST_RECV_OP;
+			xop->dst_dev = se_cmd->se_dev;
+			XCOPY_CTL("XCOPY 0xe4: Set xop->dst_dev: %p from destination"
+				" received xop\n", xop->dst_dev);
+		}
+	}
+
+	return 0;
+}
+
+static int target_xcopy_parse_target_descriptors(struct se_cmd *se_cmd,
+				struct xcopy_op *xop, unsigned char *p,
+				unsigned short tdll)
+{
+	struct se_device *local_dev = se_cmd->se_dev;
+	unsigned char *desc = p;
+	int offset = tdll % XCOPY_TARGET_DESC_LEN, rc, ret = 0;
+	unsigned short start = 0;
+	bool src = true;
+
+	if (offset != 0) {
+		pr_err("XCOPY target descriptor list length is not"
+			" multiple of %d\n", XCOPY_TARGET_DESC_LEN);
+		return -EINVAL;
+	}
+	if (tdll > 64) {
+		pr_err("XCOPY target descriptor supports a maximum"
+			" two src/dest descriptors, tdll: %hu too large..\n", tdll);
+		return -EINVAL;
+	}
+	/*
+	 * Generate an IEEE Registered Extended designator based upon the
+	 * se_device the XCOPY was received upon..
+	 */
+	memset(&xop->local_dev_wwn[0], 0, XCOPY_NAA_IEEE_REGEX_LEN);
+	target_xcopy_gen_naa_ieee(local_dev, &xop->local_dev_wwn[0]);
+
+	while (start < tdll) {
+		/*
+		 * Check target descriptor identification with 0xE4 type with
+		 * use VPD 0x83 WWPN matching ..
+		 */
+		switch (desc[0]) {
+		case 0xe4:
+			rc = target_xcopy_parse_tiddesc_e4(se_cmd, xop,
+							&desc[0], src);
+			if (rc != 0)
+				goto out;
+			/*
+			 * Assume target descriptors are in source -> destination order..
+			 */
+			if (src == true)
+				src = false;
+			else
+				src = true;
+			start += XCOPY_TARGET_DESC_LEN;
+			desc += XCOPY_TARGET_DESC_LEN;
+			ret++;
+			break;
+		default:
+			pr_err("XCOPY unsupported descriptor type code:"
+					" 0x%02x\n", desc[0]);
+			goto out;
+		}
+	}
+
+	if (xop->op_origin == XCOL_SOURCE_RECV_OP)
+		rc = target_xcopy_locate_se_dev_e4(se_cmd, xop, true);
+	else
+		rc = target_xcopy_locate_se_dev_e4(se_cmd, xop, false);
+
+	if (rc < 0)
+		goto out;
+
+	XCOPY_CTL("XCOPY TGT desc: Source dev: %p NAA IEEE WWN: "
+		"0x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x"
+		"%02x%02x%02x%02x%02x%02x\n", xop->src_dev, xop->src_tid_wwn[0],
+		xop->src_tid_wwn[1], xop->src_tid_wwn[2], xop->src_tid_wwn[3],
+		xop->src_tid_wwn[4], xop->src_tid_wwn[5], xop->src_tid_wwn[6],
+		xop->src_tid_wwn[7], xop->src_tid_wwn[8], xop->src_tid_wwn[9],
+		xop->src_tid_wwn[10], xop->src_tid_wwn[11], xop->src_tid_wwn[12],
+		xop->src_tid_wwn[13], xop->src_tid_wwn[14], xop->src_tid_wwn[15]);
+	XCOPY_CTL("XCOPY TGT desc: Dest dev: %p NAA IEEE WWN: "
+		"0x%02x%02x%02x%02x%02x%02x%02x%02x%02x%02x"
+		"%02x%02x%02x%02x%02x%02x\n",  xop->dst_dev, xop->dst_tid_wwn[0],
+		xop->dst_tid_wwn[1], xop->dst_tid_wwn[2], xop->dst_tid_wwn[3],
+		xop->dst_tid_wwn[4], xop->dst_tid_wwn[5], xop->dst_tid_wwn[6],
+		xop->dst_tid_wwn[7], xop->dst_tid_wwn[8], xop->dst_tid_wwn[9],
+		xop->dst_tid_wwn[10], xop->dst_tid_wwn[11], xop->dst_tid_wwn[12],
+		xop->dst_tid_wwn[13], xop->dst_tid_wwn[14], xop->dst_tid_wwn[15]);
+
+	return ret;
+
+out:
+	return -EINVAL;
+}
+
+static int target_xcopy_parse_segdesc_02(struct se_cmd *se_cmd, struct xcopy_op *xop,
+					unsigned char *p)
+{
+	unsigned char *desc = p;
+	int dc = (desc[1] & 0x02);
+	unsigned short desc_len;
+
+	desc_len = get_unaligned_be16(&desc[2]);
+	if (desc_len != 0x18) {
+		pr_err("XCOPY segment desc 0x02: Illegal desc_len:"
+				" %hu\n", desc_len);
+		return -EINVAL;
+	}
+
+	xop->stdi = get_unaligned_be16(&desc[4]);
+	xop->dtdi = get_unaligned_be16(&desc[6]);
+	XCOPY_CTL("XCOPY seg desc 0x02: desc_len: %hu stdi: %hu dtdi: %hu, DC: %d\n",
+		desc_len, xop->stdi, xop->dtdi, dc);
+
+	xop->nolb = get_unaligned_be16(&desc[10]);
+	xop->src_lba = get_unaligned_be64(&desc[12]);
+	xop->dst_lba = get_unaligned_be64(&desc[20]);
+	XCOPY_CTL("XCOPY seg desc 0x02: nolb: %hu src_lba: %llu dst_lba: %llu\n",
+		xop->nolb, (unsigned long long)xop->src_lba,
+		(unsigned long long)xop->dst_lba);
+
+	if (dc != 0) {
+		xop->dbl = (desc[29] << 16) & 0xff;
+		xop->dbl |= (desc[30] << 8) & 0xff;
+		xop->dbl |= desc[31] & 0xff;
+
+		XCOPY_CTL("XCOPY seg desc 0x02: DC=1 w/ dbl: %u\n", xop->dbl);
+	}
+	return 0;
+}
+
+static int target_xcopy_parse_segment_descriptors(struct se_cmd *se_cmd,
+				struct xcopy_op *xop, unsigned char *p,
+				unsigned int sdll)
+{
+	unsigned char *desc = p;
+	unsigned int start = 0;
+	int offset = sdll % XCOPY_SEGMENT_DESC_LEN, rc, ret = 0;
+
+	if (offset != 0) {
+		pr_err("XCOPY segment descriptor list length is not"
+			" multiple of %d\n", XCOPY_SEGMENT_DESC_LEN);
+		return -EINVAL;
+	}
+
+	while (start < sdll) {
+		/*
+		 * Check segment descriptor type code for block -> block
+		 */
+		switch (desc[0]) {
+		case 0x02:
+			rc = target_xcopy_parse_segdesc_02(se_cmd, xop, desc);
+			if (rc < 0)
+				goto out;
+
+			ret++;
+			start += XCOPY_SEGMENT_DESC_LEN;
+			desc += XCOPY_SEGMENT_DESC_LEN;
+			break;
+		default:
+			pr_err("XCOPY unspported segment descriptor"
+				"type: 0x%02x\n", desc[0]);
+			goto out;
+		}
+	}
+
+	return ret;
+
+out:
+	return -EINVAL;
+}
+
+/*
+ * Start xcopy_pt ops
+ */
+
+struct xcopy_pt_cmd {
+	bool remote_port;
+	struct se_cmd se_cmd;
+	struct xcopy_op *xcopy_op;
+	struct completion xpt_passthrough_wsem;
+	struct completion xpt_passthrough_sem;
+};
+
+static struct se_port xcopy_pt_port;
+static struct se_portal_group xcopy_pt_tpg;
+static struct se_session xcopy_pt_sess;
+static struct se_node_acl xcopy_pt_nacl;
+
+static char *xcopy_pt_get_fabric_name(void)
+{
+        return "xcopy-pt";
+}
+
+static u32 xcopy_pt_get_tag(struct se_cmd *se_cmd)
+{
+        return 0;
+}
+
+static int xcopy_pt_get_cmd_state(struct se_cmd *se_cmd)
+{
+        return 0;
+}
+
+static void xcopy_pt_undepend_remotedev(struct xcopy_op *xop)
+{
+	struct configfs_subsystem *subsys = target_core_subsystem[0];
+	struct se_device *remote_dev;
+
+	if (xop->op_origin == XCOL_SOURCE_RECV_OP)
+		remote_dev = xop->dst_dev;
+	else
+		remote_dev = xop->src_dev;
+
+	XCOPY_CTL("Calling configfs_undepend_item for subsys: %p"
+		  " remote_dev: %p remote_dev->dev_group: %p\n",
+		  subsys, remote_dev, &remote_dev->dev_group.cg_item);
+
+	configfs_undepend_item(subsys, &remote_dev->dev_group.cg_item);
+}
+
+static void xcopy_pt_release_cmd(struct se_cmd *se_cmd)
+{
+	struct xcopy_pt_cmd *xpt_cmd = container_of(se_cmd,
+				struct xcopy_pt_cmd, se_cmd);
+
+	if (xpt_cmd->remote_port)
+		kfree(se_cmd->se_lun);
+
+	kfree(xpt_cmd);
+}
+
+static int xcopy_pt_check_stop_free(struct se_cmd *se_cmd)
+{
+	struct xcopy_pt_cmd *xpt_cmd = container_of(se_cmd,
+				struct xcopy_pt_cmd, se_cmd);
+
+	complete(&xpt_cmd->xpt_passthrough_sem);
+	return 0;
+}
+
+static int xcopy_pt_write_pending(struct se_cmd *se_cmd)
+{
+	struct xcopy_pt_cmd *xpt_cmd = container_of(se_cmd,
+				struct xcopy_pt_cmd, se_cmd);
+
+	complete(&xpt_cmd->xpt_passthrough_wsem);
+	return 0;
+}
+
+static int xcopy_pt_write_pending_status(struct se_cmd *se_cmd)
+{
+	return 0;
+}
+
+static int xcopy_pt_queue_data_in(struct se_cmd *se_cmd)
+{
+	return 0;
+}
+
+static int xcopy_pt_queue_status(struct se_cmd *se_cmd)
+{
+	return 0;
+}
+
+static struct target_core_fabric_ops xcopy_pt_tfo = {
+	.get_fabric_name	= xcopy_pt_get_fabric_name,
+	.get_task_tag		= xcopy_pt_get_tag,
+	.get_cmd_state		= xcopy_pt_get_cmd_state,
+	.release_cmd		= xcopy_pt_release_cmd,
+	.check_stop_free	= xcopy_pt_check_stop_free,
+	.write_pending		= xcopy_pt_write_pending,
+	.write_pending_status	= xcopy_pt_write_pending_status,
+	.queue_data_in		= xcopy_pt_queue_data_in,
+	.queue_status		= xcopy_pt_queue_status,
+};
+
+/*
+ * End xcopy_pt_ops
+ */
+
+int target_xcopy_setup_pt(void)
+{
+	xcopy_wq = alloc_workqueue("xcopy_wq", WQ_MEM_RECLAIM, 0);
+	if (!xcopy_wq) {
+		pr_err("Unable to allocate xcopy_wq\n");
+		return -ENOMEM;
+	}
+
+	memset(&xcopy_pt_port, 0, sizeof(struct se_port));
+	INIT_LIST_HEAD(&xcopy_pt_port.sep_alua_list);
+	INIT_LIST_HEAD(&xcopy_pt_port.sep_list);
+	mutex_init(&xcopy_pt_port.sep_tg_pt_md_mutex);
+
+	memset(&xcopy_pt_tpg, 0, sizeof(struct se_portal_group));
+	INIT_LIST_HEAD(&xcopy_pt_tpg.se_tpg_node);
+	INIT_LIST_HEAD(&xcopy_pt_tpg.acl_node_list);
+	INIT_LIST_HEAD(&xcopy_pt_tpg.tpg_sess_list);
+
+	xcopy_pt_port.sep_tpg = &xcopy_pt_tpg;
+	xcopy_pt_tpg.se_tpg_tfo = &xcopy_pt_tfo;
+
+	memset(&xcopy_pt_nacl, 0, sizeof(struct se_node_acl));
+	INIT_LIST_HEAD(&xcopy_pt_nacl.acl_list);
+	INIT_LIST_HEAD(&xcopy_pt_nacl.acl_sess_list);
+	memset(&xcopy_pt_sess, 0, sizeof(struct se_session));
+	INIT_LIST_HEAD(&xcopy_pt_sess.sess_list);
+	INIT_LIST_HEAD(&xcopy_pt_sess.sess_acl_list);
+
+	xcopy_pt_nacl.se_tpg = &xcopy_pt_tpg;
+	xcopy_pt_nacl.nacl_sess = &xcopy_pt_sess;
+
+	xcopy_pt_sess.se_tpg = &xcopy_pt_tpg;
+	xcopy_pt_sess.se_node_acl = &xcopy_pt_nacl;
+
+	return 0;
+}
+
+void target_xcopy_release_pt(void)
+{
+	if (xcopy_wq)
+		destroy_workqueue(xcopy_wq);
+}
+
+static void target_xcopy_setup_pt_port(
+	struct xcopy_pt_cmd *xpt_cmd,
+	struct xcopy_op *xop,
+	bool remote_port)
+{
+	struct se_cmd *ec_cmd = xop->xop_se_cmd;
+	struct se_cmd *pt_cmd = &xpt_cmd->se_cmd;
+
+	if (xop->op_origin == XCOL_SOURCE_RECV_OP) {
+		/*
+		 * Honor destination port reservations for X-COPY PUSH emulation
+		 * when CDB is received on local source port, and READs blocks to
+		 * WRITE on remote destination port.
+		 */
+		if (remote_port) {
+			xpt_cmd->remote_port = remote_port;
+			pt_cmd->se_lun->lun_sep = &xcopy_pt_port;
+			XCOPY_IO("Setup emulated remote DEST xcopy_pt_port: %p to"
+				" cmd->se_lun->lun_sep for X-COPY data PUSH\n",
+				pt_cmd->se_lun->lun_sep);
+		} else {
+			pt_cmd->se_lun = ec_cmd->se_lun;
+			pt_cmd->se_dev = ec_cmd->se_dev;
+
+			XCOPY_IO("Honoring local SRC port from ec_cmd->se_dev:"
+				" %p\n", pt_cmd->se_dev);
+			pt_cmd->se_lun = ec_cmd->se_lun;
+			XCOPY_IO("Honoring local SRC port from ec_cmd->se_lun: %p\n",
+				pt_cmd->se_lun);
+		}
+	} else {
+		/*
+		 * Honor source port reservation for X-COPY PULL emulation
+		 * when CDB is received on local desintation port, and READs
+		 * blocks from the remote source port to WRITE on local
+		 * destination port.
+		 */
+		if (remote_port) {
+			xpt_cmd->remote_port = remote_port;
+			pt_cmd->se_lun->lun_sep = &xcopy_pt_port;
+			XCOPY_IO("Setup emulated remote SRC xcopy_pt_port: %p to"
+				" cmd->se_lun->lun_sep for X-COPY data PULL\n",
+				pt_cmd->se_lun->lun_sep);
+		} else {
+			pt_cmd->se_lun = ec_cmd->se_lun;
+			pt_cmd->se_dev = ec_cmd->se_dev;
+
+			XCOPY_IO("Honoring local DST port from ec_cmd->se_dev:"
+				" %p\n", pt_cmd->se_dev);
+			pt_cmd->se_lun = ec_cmd->se_lun;
+			XCOPY_IO("Honoring local DST port from ec_cmd->se_lun: %p\n",
+				pt_cmd->se_lun);
+		}
+	}
+}
+
+static int target_xcopy_init_pt_lun(
+	struct xcopy_pt_cmd *xpt_cmd,
+	struct xcopy_op *xop,
+	struct se_device *se_dev,
+	struct se_cmd *pt_cmd,
+	bool remote_port)
+{
+	/*
+	 * Don't allocate + init an pt_cmd->se_lun if honoring local port for
+	 * reservations.  The pt_cmd->se_lun pointer will be setup from within
+	 * target_xcopy_setup_pt_port()
+	 */
+	if (remote_port == false) {
+		pt_cmd->se_cmd_flags |= SCF_SE_LUN_CMD | SCF_CMD_XCOPY_PASSTHROUGH;
+		return 0;
+	}
+
+	pt_cmd->se_lun = kzalloc(sizeof(struct se_lun), GFP_KERNEL);
+	if (!pt_cmd->se_lun) {
+		pr_err("Unable to allocate pt_cmd->se_lun\n");
+		return -ENOMEM;
+	}
+	init_completion(&pt_cmd->se_lun->lun_shutdown_comp);
+	INIT_LIST_HEAD(&pt_cmd->se_lun->lun_cmd_list);
+	INIT_LIST_HEAD(&pt_cmd->se_lun->lun_acl_list);
+	spin_lock_init(&pt_cmd->se_lun->lun_acl_lock);
+	spin_lock_init(&pt_cmd->se_lun->lun_cmd_lock);
+	spin_lock_init(&pt_cmd->se_lun->lun_sep_lock);
+
+	pt_cmd->se_dev = se_dev;
+
+	XCOPY_IO("Setup emulated se_dev: %p from se_dev\n", pt_cmd->se_dev);
+	pt_cmd->se_lun->lun_se_dev = se_dev;
+	pt_cmd->se_cmd_flags |= SCF_SE_LUN_CMD | SCF_CMD_XCOPY_PASSTHROUGH;
+
+	XCOPY_IO("Setup emulated se_dev: %p to pt_cmd->se_lun->lun_se_dev\n",
+		pt_cmd->se_lun->lun_se_dev);
+
+	return 0;
+}
+
+static int target_xcopy_setup_pt_cmd(
+	struct xcopy_pt_cmd *xpt_cmd,
+	struct xcopy_op *xop,
+	struct se_device *se_dev,
+	unsigned char *cdb,
+	bool remote_port,
+	bool alloc_mem)
+{
+	struct se_cmd *cmd = &xpt_cmd->se_cmd;
+	sense_reason_t sense_rc;
+	int ret = 0, rc;
+	/*
+	 * Setup LUN+port to honor reservations based upon xop->op_origin for
+	 * X-COPY PUSH or X-COPY PULL based upon where the CDB was received.
+	 */
+	rc = target_xcopy_init_pt_lun(xpt_cmd, xop, se_dev, cmd, remote_port);
+	if (rc < 0) {
+		ret = rc;
+		goto out;
+	}
+	xpt_cmd->xcopy_op = xop;
+	target_xcopy_setup_pt_port(xpt_cmd, xop, remote_port);
+
+	sense_rc = target_setup_cmd_from_cdb(cmd, cdb);
+	if (sense_rc) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	if (alloc_mem) {
+		rc = transport_generic_get_mem(cmd);
+		if (rc < 0) {
+			ret = rc;
+			goto out;
+		}
+		/*
+		 * Set this bit so that transport_free_pages() allows the
+		 * caller to release SGLs + physical memory allocated by
+		 * transport_generic_get_mem()..
+		 */
+		cmd->se_cmd_flags |= SCF_PASSTHROUGH_SG_TO_MEM_NOALLOC;
+	} else {
+		/*
+		 * Here the previously allocated SGLs for the internal READ
+		 * are mapped zero-copy to the internal WRITE.
+		 */
+		sense_rc = transport_generic_map_mem_to_cmd(cmd,
+					xop->xop_data_sg, xop->xop_data_nents,
+					NULL, 0);
+		if (sense_rc) {
+			ret = -EINVAL;
+			goto out;
+		}
+
+		XCOPY_IO("Setup PASSTHROUGH_NOALLOC t_data_sg: %p t_data_nents:"
+			 " %u\n", cmd->t_data_sg, cmd->t_data_nents);
+	}
+
+	return 0;
+
+out:
+	if (remote_port == true)
+		kfree(cmd->se_lun);
+	return ret;
+}
+
+static int target_xcopy_issue_pt_cmd(struct xcopy_pt_cmd *xpt_cmd)
+{
+	struct se_cmd *se_cmd = &xpt_cmd->se_cmd;
+	sense_reason_t sense_rc;
+
+	sense_rc = transport_generic_new_cmd(se_cmd);
+	if (sense_rc)
+		return -EINVAL;
+
+	if (se_cmd->data_direction == DMA_TO_DEVICE) {
+		if (wait_for_completion_interruptible(&xpt_cmd->xpt_passthrough_wsem) != 0)
+			return -EINVAL;
+
+		target_execute_cmd(se_cmd);
+	}
+
+	wait_for_completion_interruptible(&xpt_cmd->xpt_passthrough_sem);
+
+	XCOPY_IO("target_xcopy_issue_pt_cmd(): SCSI status: 0x%02x\n",
+			se_cmd->scsi_status);
+	return 0;
+}
+
+static int target_xcopy_read_source(
+	struct se_cmd *ec_cmd,
+	struct xcopy_op *xop,
+	struct se_device *src_dev,
+	sector_t src_lba,
+	u32 src_sectors)
+{
+	struct xcopy_pt_cmd *xpt_cmd;
+	struct se_cmd *se_cmd;
+	u32 length = (src_sectors * src_dev->dev_attrib.block_size);
+	int rc;
+	unsigned char cdb[16];
+	bool remote_port = (xop->op_origin == XCOL_DEST_RECV_OP);
+
+	xpt_cmd = kzalloc(sizeof(struct xcopy_pt_cmd), GFP_KERNEL);
+	if (!xpt_cmd) {
+		pr_err("Unable to allocate xcopy_pt_cmd\n");
+		return -ENOMEM;
+	}
+	init_completion(&xpt_cmd->xpt_passthrough_wsem);
+	init_completion(&xpt_cmd->xpt_passthrough_sem);
+	se_cmd = &xpt_cmd->se_cmd;
+
+	memset(&cdb[0], 0, 16);
+	cdb[0] = READ_16;
+	put_unaligned_be64(src_lba, &cdb[2]);
+	put_unaligned_be32(src_sectors, &cdb[10]);
+	XCOPY_IO("XCOPY: Built READ_16: LBA: %llu Sectors: %u Length: %u\n",
+		(unsigned long long)src_lba, src_sectors, length);
+
+	transport_init_se_cmd(se_cmd, &xcopy_pt_tfo, NULL, length,
+				DMA_FROM_DEVICE, 0, NULL);
+	xop->src_pt_cmd = xpt_cmd;
+
+	rc = target_xcopy_setup_pt_cmd(xpt_cmd, xop, src_dev, &cdb[0],
+				remote_port, true);
+	if (rc < 0) {
+		transport_generic_free_cmd(se_cmd, 0);
+		return rc;
+	}
+
+	xop->xop_data_sg = se_cmd->t_data_sg;
+	xop->xop_data_nents = se_cmd->t_data_nents;
+	XCOPY_IO("XCOPY-READ: Saved xop->xop_data_sg: %p, num: %u for READ"
+		" memory\n", xop->xop_data_sg, xop->xop_data_nents);
+
+	rc = target_xcopy_issue_pt_cmd(xpt_cmd);
+	if (rc < 0) {
+		transport_generic_free_cmd(se_cmd, 0);
+		return rc;
+	}
+	/*
+	 * Clear off the allocated t_data_sg, that has been saved for
+	 * zero-copy WRITE submission reuse in struct xcopy_op..
+	 */
+	se_cmd->t_data_sg = NULL;
+	se_cmd->t_data_nents = 0;
+
+	return 0;
+}
+
+static int target_xcopy_write_destination(
+	struct se_cmd *ec_cmd,
+	struct xcopy_op *xop,
+	struct se_device *dst_dev,
+	sector_t dst_lba,
+	u32 dst_sectors)
+{
+	struct xcopy_pt_cmd *xpt_cmd;
+	struct se_cmd *se_cmd;
+	u32 length = (dst_sectors * dst_dev->dev_attrib.block_size);
+	int rc;
+	unsigned char cdb[16];
+	bool remote_port = (xop->op_origin == XCOL_SOURCE_RECV_OP);
+
+	xpt_cmd = kzalloc(sizeof(struct xcopy_pt_cmd), GFP_KERNEL);
+	if (!xpt_cmd) {
+		pr_err("Unable to allocate xcopy_pt_cmd\n");
+		return -ENOMEM;
+	}
+	init_completion(&xpt_cmd->xpt_passthrough_wsem);
+	init_completion(&xpt_cmd->xpt_passthrough_sem);
+	se_cmd = &xpt_cmd->se_cmd;
+
+	memset(&cdb[0], 0, 16);
+	cdb[0] = WRITE_16;
+	put_unaligned_be64(dst_lba, &cdb[2]);
+	put_unaligned_be32(dst_sectors, &cdb[10]);
+	XCOPY_IO("XCOPY: Built WRITE_16: LBA: %llu Sectors: %u Length: %u\n",
+		(unsigned long long)dst_lba, dst_sectors, length);
+
+	transport_init_se_cmd(se_cmd, &xcopy_pt_tfo, NULL, length,
+				DMA_TO_DEVICE, 0, NULL);
+	xop->dst_pt_cmd = xpt_cmd;
+
+	rc = target_xcopy_setup_pt_cmd(xpt_cmd, xop, dst_dev, &cdb[0],
+				remote_port, false);
+	if (rc < 0) {
+		struct se_cmd *src_cmd = &xop->src_pt_cmd->se_cmd;
+		/*
+		 * If the failure happened before the t_mem_list hand-off in
+		 * target_xcopy_setup_pt_cmd(), Reset memory + clear flag so that
+		 * core releases this memory on error during X-COPY WRITE I/O.
+		 */
+		src_cmd->se_cmd_flags &= ~SCF_PASSTHROUGH_SG_TO_MEM_NOALLOC;
+		src_cmd->t_data_sg = xop->xop_data_sg;
+		src_cmd->t_data_nents = xop->xop_data_nents;
+
+		transport_generic_free_cmd(se_cmd, 0);
+		return rc;
+	}
+
+	rc = target_xcopy_issue_pt_cmd(xpt_cmd);
+	if (rc < 0) {
+		se_cmd->se_cmd_flags &= ~SCF_PASSTHROUGH_SG_TO_MEM_NOALLOC;
+		transport_generic_free_cmd(se_cmd, 0);
+		return rc;
+	}
+
+	return 0;
+}
+
+static void target_xcopy_do_work(struct work_struct *work)
+{
+	struct xcopy_op *xop = container_of(work, struct xcopy_op, xop_work);
+	struct se_device *src_dev = xop->src_dev, *dst_dev = xop->dst_dev;
+	struct se_cmd *ec_cmd = xop->xop_se_cmd;
+	sector_t src_lba = xop->src_lba, dst_lba = xop->dst_lba, end_lba;
+	unsigned int max_sectors;
+	int rc;
+	unsigned short nolb = xop->nolb, cur_nolb, max_nolb, copied_nolb = 0;
+
+	end_lba = src_lba + nolb;
+	/*
+	 * Break up XCOPY I/O into hw_max_sectors sized I/O based on the
+	 * smallest max_sectors between src_dev + dev_dev, or
+	 */
+	max_sectors = min(src_dev->dev_attrib.hw_max_sectors,
+			  dst_dev->dev_attrib.hw_max_sectors);
+	max_sectors = min_t(u32, max_sectors, XCOPY_MAX_SECTORS);
+
+	max_nolb = min_t(u16, max_sectors, ((u16)(~0U)));
+
+	XCOPY_CTL("target_xcopy_do_work: nolb: %hu, max_nolb: %hu end_lba: %llu\n",
+			nolb, max_nolb, (unsigned long long)end_lba);
+	XCOPY_CTL("target_xcopy_do_work: Starting src_lba: %llu, dst_lba: %llu\n",
+			(unsigned long long)src_lba, (unsigned long long)dst_lba);
+
+	while (src_lba < end_lba) {
+		cur_nolb = min(nolb, max_nolb);
+
+		XCOPY_IO("target_xcopy_do_work: Calling read src_dev: %p src_lba: %llu,"
+			" cur_nolb: %hu\n", src_dev, (unsigned long long)src_lba, cur_nolb);
+
+		rc = target_xcopy_read_source(ec_cmd, xop, src_dev, src_lba, cur_nolb);
+		if (rc < 0)
+			goto out;
+
+		src_lba += cur_nolb;
+		XCOPY_IO("target_xcopy_do_work: Incremented READ src_lba to %llu\n",
+				(unsigned long long)src_lba);
+
+		XCOPY_IO("target_xcopy_do_work: Calling write dst_dev: %p dst_lba: %llu,"
+			" cur_nolb: %hu\n", dst_dev, (unsigned long long)dst_lba, cur_nolb);
+
+		rc = target_xcopy_write_destination(ec_cmd, xop, dst_dev,
+						dst_lba, cur_nolb);
+		if (rc < 0) {
+			transport_generic_free_cmd(&xop->src_pt_cmd->se_cmd, 0);
+			goto out;
+		}
+
+		dst_lba += cur_nolb;
+		XCOPY_IO("target_xcopy_do_work: Incremented WRITE dst_lba to %llu\n",
+				(unsigned long long)dst_lba);
+
+		copied_nolb += cur_nolb;
+		nolb -= cur_nolb;
+
+		transport_generic_free_cmd(&xop->src_pt_cmd->se_cmd, 0);
+		xop->dst_pt_cmd->se_cmd.se_cmd_flags &= ~SCF_PASSTHROUGH_SG_TO_MEM_NOALLOC;
+
+		transport_generic_free_cmd(&xop->dst_pt_cmd->se_cmd, 0);
+	}
+
+	xcopy_pt_undepend_remotedev(xop);
+	kfree(xop);
+
+	XCOPY_CTL("target_xcopy_do_work: Final src_lba: %llu, dst_lba: %llu\n",
+		(unsigned long long)src_lba, (unsigned long long)dst_lba);
+	XCOPY_CTL("target_xcopy_do_work: Blocks copied: %hu, Bytes Copied: %u\n",
+		copied_nolb, copied_nolb * dst_dev->dev_attrib.block_size);
+
+	XCOPY_CTL("target_xcopy_do_work: Setting X-COPY GOOD status -> sending response\n");
+	target_complete_cmd(ec_cmd, SAM_STAT_GOOD);
+	return;
+
+out:
+	xcopy_pt_undepend_remotedev(xop);
+	kfree(xop);
+
+	printk("target_xcopy_do_work: Setting X-COPY CHECK_CONDITION -> sending response\n");
+	ec_cmd->scsi_status = SAM_STAT_CHECK_CONDITION;
+	target_complete_cmd(ec_cmd, SAM_STAT_CHECK_CONDITION);
+}
+
+sense_reason_t target_do_xcopy(struct se_cmd *se_cmd)
+{
+	struct xcopy_op *xop = NULL;
+	unsigned char *p = NULL, *seg_desc;
+	unsigned int list_id, list_id_usage, sdll, inline_dl;
+	int rc;
+	unsigned short tdll;
+
+	p = transport_kmap_data_sg(se_cmd);
+	if (!p) {
+		pr_err("transport_kmap_data_sg() failed in target_do_xcopy\n");
+		return TCM_OUT_OF_RESOURCES;
+	}
+
+	list_id = p[0];
+	if (list_id != 0x00) {
+		pr_err("XCOPY with non zero list_id: 0x%02x\n", list_id);
+		goto out;
+	}
+	/*
+	 * Check list id usage value:
+	 */
+	list_id_usage = (p[1] & 0x18);
+#if 0
+	if (list_id_usage != 0x0) {
+		pr_err("XCOPY with non zero list identifier not"
+			" supported: 0x%02x\n", list_id_usage);
+		goto out;
+	}
+#endif
+	/*
+	 * Determine TARGET DESCRIPTOR LIST LENGTH + SEGMENT DESCRIPTOR LIST LENGTH
+	 */
+	tdll = get_unaligned_be16(&p[2]);
+	sdll = get_unaligned_be32(&p[8]);
+
+	inline_dl = get_unaligned_be32(&p[12]);
+	if (inline_dl != 0) {
+		pr_err("XCOPY with non zero inline data length\n");
+		goto out;
+	}
+
+	xop = kzalloc(sizeof(struct xcopy_op), GFP_KERNEL);
+	if (!xop) {
+		pr_err("Unable to allocate xcopy_op\n");
+		goto out;
+	}
+	xop->xop_se_cmd = se_cmd;
+
+	XCOPY_CTL("Processing XCOPY with list_id: 0x%02x list_id_usage: 0x%02x"
+		" tdll: %hu sdll: %u inline_dl: %u\n", list_id, list_id_usage,
+		tdll, sdll, inline_dl);
+
+	rc = target_xcopy_parse_target_descriptors(se_cmd, xop, &p[16], tdll);
+	if (rc <= 0)
+		goto out;
+
+	XCOPY_CTL("XCOPY: Processed %d target descriptors, length: %u\n", rc,
+				rc * XCOPY_TARGET_DESC_LEN);
+	seg_desc = &p[16];
+	seg_desc += (rc * XCOPY_TARGET_DESC_LEN);
+
+	rc = target_xcopy_parse_segment_descriptors(se_cmd, xop, seg_desc, sdll);
+	if (rc <= 0) {
+		xcopy_pt_undepend_remotedev(xop);
+		goto out;
+	}
+	transport_kunmap_data_sg(se_cmd);
+
+	XCOPY_CTL("XCOPY: Processed %d segment descriptors, length: %u\n", rc,
+				rc * XCOPY_SEGMENT_DESC_LEN);
+	INIT_WORK(&xop->xop_work, target_xcopy_do_work);
+	queue_work(xcopy_wq, &xop->xop_work);
+	return TCM_NO_SENSE;
+
+out:
+	if (p)
+		transport_kunmap_data_sg(se_cmd);
+	kfree(xop);
+	return TCM_INVALID_CDB_FIELD;
+}
+
+static sense_reason_t target_rcr_operating_parameters(struct se_cmd *se_cmd)
+{
+	unsigned char *p;
+
+	p = transport_kmap_data_sg(se_cmd);
+	if (!p) {
+		pr_err("transport_kmap_data_sg failed in"
+		       " target_rcr_operating_parameters\n");
+		return TCM_OUT_OF_RESOURCES;
+	}
+
+	if (se_cmd->data_length < 54) {
+		pr_err("Receive Copy Results Op Parameters length"
+		       " too small: %u\n", se_cmd->data_length);
+		transport_kunmap_data_sg(se_cmd);
+		return TCM_INVALID_CDB_FIELD;
+	}
+	/*
+	 * Set SNLID=1 (Supports no List ID)
+	 */
+	p[4] = 0x1;
+	/*
+	 * MAXIMUM TARGET DESCRIPTOR COUNT
+	 */
+	put_unaligned_be16(RCR_OP_MAX_TARGET_DESC_COUNT, &p[8]);
+	/*
+	 * MAXIMUM SEGMENT DESCRIPTOR COUNT
+	 */
+	put_unaligned_be16(RCR_OP_MAX_SG_DESC_COUNT, &p[10]);
+	/*
+	 * MAXIMUM DESCRIPTOR LIST LENGTH
+	 */
+	put_unaligned_be32(RCR_OP_MAX_DESC_LIST_LEN, &p[12]);
+	/*
+	 * MAXIMUM SEGMENT LENGTH
+	 */
+	put_unaligned_be32(RCR_OP_MAX_SEGMENT_LEN, &p[16]);
+	/*
+	 * MAXIMUM INLINE DATA LENGTH for SA 0x04 (NOT SUPPORTED)
+	 */
+	put_unaligned_be32(0x0, &p[20]);
+	/*
+	 * HELD DATA LIMIT
+	 */
+	put_unaligned_be32(0x0, &p[24]);
+	/*
+	 * MAXIMUM STREAM DEVICE TRANSFER SIZE
+	 */
+	put_unaligned_be32(0x0, &p[28]);
+	/*
+	 * TOTAL CONCURRENT COPIES
+	 */
+	put_unaligned_be16(RCR_OP_TOTAL_CONCURR_COPIES, &p[34]);
+	/*
+	 * MAXIMUM CONCURRENT COPIES
+	 */
+	p[36] = RCR_OP_MAX_CONCURR_COPIES;
+	/*
+	 * DATA SEGMENT GRANULARITY (log 2)
+	 */
+	p[37] = RCR_OP_DATA_SEG_GRAN_LOG2;
+	/*
+	 * INLINE DATA GRANULARITY log 2)
+	 */
+	p[38] = RCR_OP_INLINE_DATA_GRAN_LOG2;
+	/*
+	 * HELD DATA GRANULARITY
+	 */
+	p[39] = RCR_OP_HELD_DATA_GRAN_LOG2;
+	/*
+	 * IMPLEMENTED DESCRIPTOR LIST LENGTH
+	 */
+	p[43] = 0x2;
+	/*
+	 * List of implemented descriptor type codes (ordered)
+	 */
+	p[44] = 0x02; /* Copy Block to Block device */
+	p[45] = 0xe4; /* Identification descriptor target descriptor */
+
+	/*
+	 * AVAILABLE DATA (n-3)
+	 */
+	put_unaligned_be32(42, &p[0]);
+
+	transport_kunmap_data_sg(se_cmd);
+	target_complete_cmd(se_cmd, GOOD);
+
+	return TCM_NO_SENSE;
+}
+
+sense_reason_t target_do_receive_copy_results(struct se_cmd *se_cmd)
+{
+	unsigned char *cdb = &se_cmd->t_task_cdb[0];
+	int sa = (cdb[1] & 0x1f), list_id = cdb[2];
+	int rc = TCM_NO_SENSE;
+
+	XCOPY_CTL("Entering target_do_receive_copy_results: SA: 0x%02x, List ID:"
+		" 0x%02x, AL: %u\n", sa, list_id, se_cmd->data_length);
+
+	if (list_id != 0) {
+		pr_err("Receive Copy Results with non zero list identifier"
+		       " not supported\n");
+		return TCM_INVALID_CDB_FIELD;
+	}
+
+	switch (sa) {
+	case RCR_SA_OPERATING_PARAMETERS:
+		rc = target_rcr_operating_parameters(se_cmd);
+		break;
+	case RCR_SA_COPY_STATUS:
+	case RCR_SA_RECEIVE_DATA:
+	case RCR_SA_FAILED_SEGMENT_DETAILS:
+	default:
+		pr_err("Unsupported SA for receive copy results: 0x%02x\n", sa);
+		return TCM_INVALID_CDB_FIELD;
+	}
+
+	return rc;
+}
diff --git a/drivers/target/target_core_xcopy.h b/drivers/target/target_core_xcopy.h
new file mode 100644
index 0000000..700a981
--- /dev/null
+++ b/drivers/target/target_core_xcopy.h
@@ -0,0 +1,62 @@
+#define XCOPY_TARGET_DESC_LEN		32
+#define XCOPY_SEGMENT_DESC_LEN		28
+#define XCOPY_NAA_IEEE_REGEX_LEN	16
+#define XCOPY_MAX_SECTORS		1024
+
+enum xcopy_origin_list {
+	XCOL_SOURCE_RECV_OP = 0x01,
+	XCOL_DEST_RECV_OP = 0x02,
+};
+
+struct xcopy_pt_cmd;
+
+struct xcopy_op {
+	int op_origin;
+
+	struct se_cmd *xop_se_cmd;
+	struct se_device *src_dev;
+	unsigned char src_tid_wwn[XCOPY_NAA_IEEE_REGEX_LEN];
+	struct se_device *dst_dev;
+	unsigned char dst_tid_wwn[XCOPY_NAA_IEEE_REGEX_LEN];
+	unsigned char local_dev_wwn[XCOPY_NAA_IEEE_REGEX_LEN];
+
+	sector_t src_lba;
+	sector_t dst_lba;
+	unsigned short stdi;
+	unsigned short dtdi;
+	unsigned short nolb;
+	unsigned int dbl;
+
+	struct xcopy_pt_cmd *src_pt_cmd;
+	struct xcopy_pt_cmd *dst_pt_cmd;
+
+	u32 xop_data_nents;
+	struct scatterlist *xop_data_sg;
+	struct work_struct xop_work;
+};
+
+/*
+ * Receive Copy Results Sevice Actions
+ */
+#define RCR_SA_COPY_STATUS		0x00
+#define RCR_SA_RECEIVE_DATA		0x01
+#define RCR_SA_OPERATING_PARAMETERS	0x03
+#define RCR_SA_FAILED_SEGMENT_DETAILS	0x04
+
+/*
+ * Receive Copy Results defs for Operating Parameters
+ */
+#define RCR_OP_MAX_TARGET_DESC_COUNT	0x2
+#define RCR_OP_MAX_SG_DESC_COUNT	0x1
+#define RCR_OP_MAX_DESC_LIST_LEN	1024
+#define RCR_OP_MAX_SEGMENT_LEN		268435456 /* 256 MB */
+#define RCR_OP_TOTAL_CONCURR_COPIES	0x1 /* Must be <= 16384 */
+#define RCR_OP_MAX_CONCURR_COPIES	0x1 /* Must be <= 255 */
+#define RCR_OP_DATA_SEG_GRAN_LOG2	9 /* 512 bytes in log 2 */
+#define RCR_OP_INLINE_DATA_GRAN_LOG2	9 /* 512 bytes in log 2 */
+#define RCR_OP_HELD_DATA_GRAN_LOG2	9 /* 512 bytes in log 2 */
+
+extern int target_xcopy_setup_pt(void);
+extern void target_xcopy_release_pt(void);
+extern sense_reason_t target_do_xcopy(struct se_cmd *);
+extern sense_reason_t target_do_receive_copy_results(struct se_cmd *);
diff --git a/include/target/target_core_base.h b/include/target/target_core_base.h
index 6b14f3c..f54a015 100644
--- a/include/target/target_core_base.h
+++ b/include/target/target_core_base.h
@@ -163,6 +163,7 @@ enum se_cmd_flags_table {
 	SCF_ACK_KREF			= 0x00040000,
 	SCF_COMPARE_AND_WRITE		= 0x00080000,
 	SCF_COMPARE_AND_WRITE_POST	= 0x00100000,
+	SCF_CMD_XCOPY_PASSTHROUGH	= 0x00200000,
 };
 
 /* struct se_dev_entry->lun_flags and struct se_lun->lun_access */
-- 
1.7.2.5

[PATCH 7/9] target: Enable EXTENDED_COPY setup in spc_parse_cdb

[Nicholas A. Bellinger]

From: Nicholas Bellinger <nab@daterainc.com>

Setup up the se_cmd->execute_cmd() pointers for EXTENDED_COPY and
RECEIVE_COPY_RESULTS handling within spc_parse_cdb()

Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Martin Petersen <martin.petersen@oracle.com>
Cc: Chris Mason <chris.mason@fusionio.com>
Cc: Roland Dreier <roland@purestorage.com>
Cc: Zach Brown <zab@redhat.com>
Cc: James Bottomley <JBottomley@Parallels.com>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Nicholas Bellinger <nab@daterainc.com>
---
 drivers/target/target_core_spc.c |   10 ++++++++--
 1 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/drivers/target/target_core_spc.c b/drivers/target/target_core_spc.c
index bd16a93..894e83b 100644
--- a/drivers/target/target_core_spc.c
+++ b/drivers/target/target_core_spc.c
@@ -35,7 +35,7 @@
 #include "target_core_alua.h"
 #include "target_core_pr.h"
 #include "target_core_ua.h"
-
+#include "target_core_xcopy.h"
 
 static void spc_fill_alua_data(struct se_port *port, unsigned char *buf)
 {
@@ -1252,8 +1252,14 @@ spc_parse_cdb(struct se_cmd *cmd, unsigned int *size)
 		*size = (cdb[6] << 24) | (cdb[7] << 16) | (cdb[8] << 8) | cdb[9];
 		break;
 	case EXTENDED_COPY:
-	case READ_ATTRIBUTE:
+		*size = get_unaligned_be32(&cdb[10]);
+		cmd->execute_cmd = target_do_xcopy;
+		break;
 	case RECEIVE_COPY_RESULTS:
+		*size = get_unaligned_be32(&cdb[10]);
+		cmd->execute_cmd = target_do_receive_copy_results;
+		break;
+	case READ_ATTRIBUTE:
 	case WRITE_ATTRIBUTE:
 		*size = (cdb[10] << 24) | (cdb[11] << 16) |
 		       (cdb[12] << 8) | cdb[13];
-- 
1.7.2.5

[PATCH 8/9] target: Add Third Party Copy (3PC) bit in INQUIRY response

[Nicholas A. Bellinger]

From: Nicholas Bellinger <nab@daterainc.com>

This patch adds the Third Party Copy (3PC) bit to signal support
for EXTENDED_COPY within standard inquiry response data.

Also add emulate_3pc device attribute in configfs (enabled by default)
to allow the exposure of this bit to be disabled, if necessary.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Martin Petersen <martin.petersen@oracle.com>
Cc: Chris Mason <chris.mason@fusionio.com>
Cc: Roland Dreier <roland@purestorage.com>
Cc: Zach Brown <zab@redhat.com>
Cc: James Bottomley <JBottomley@Parallels.com>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Nicholas Bellinger <nab@daterainc.com>
---
 drivers/target/target_core_configfs.c |    4 ++++
 drivers/target/target_core_device.c   |   14 ++++++++++++++
 drivers/target/target_core_internal.h |    1 +
 drivers/target/target_core_spc.c      |    6 ++++++
 include/target/target_core_base.h     |    3 +++
 5 files changed, 28 insertions(+), 0 deletions(-)

diff --git a/drivers/target/target_core_configfs.c b/drivers/target/target_core_configfs.c
index 939ecc5..026e42b 100644
--- a/drivers/target/target_core_configfs.c
+++ b/drivers/target/target_core_configfs.c
@@ -639,6 +639,9 @@ SE_DEV_ATTR(emulate_tpws, S_IRUGO | S_IWUSR);
 DEF_DEV_ATTRIB(emulate_caw);
 SE_DEV_ATTR(emulate_caw, S_IRUGO | S_IWUSR);
 
+DEF_DEV_ATTRIB(emulate_3pc);
+SE_DEV_ATTR(emulate_3pc, S_IRUGO | S_IWUSR);
+
 DEF_DEV_ATTRIB(enforce_pr_isids);
 SE_DEV_ATTR(enforce_pr_isids, S_IRUGO | S_IWUSR);
 
@@ -697,6 +700,7 @@ static struct configfs_attribute *target_core_dev_attrib_attrs[] = {
 	&target_core_dev_attrib_emulate_tpu.attr,
 	&target_core_dev_attrib_emulate_tpws.attr,
 	&target_core_dev_attrib_emulate_caw.attr,
+	&target_core_dev_attrib_emulate_3pc.attr,
 	&target_core_dev_attrib_enforce_pr_isids.attr,
 	&target_core_dev_attrib_is_nonrot.attr,
 	&target_core_dev_attrib_emulate_rest_reord.attr,
diff --git a/drivers/target/target_core_device.c b/drivers/target/target_core_device.c
index 458944e..6f492c7 100644
--- a/drivers/target/target_core_device.c
+++ b/drivers/target/target_core_device.c
@@ -906,6 +906,19 @@ int se_dev_set_emulate_caw(struct se_device *dev, int flag)
 	return 0;
 }
 
+int se_dev_set_emulate_3pc(struct se_device *dev, int flag)
+{
+	if (flag != 0 && flag != 1) {
+		pr_err("Illegal value %d\n", flag);
+		return -EINVAL;
+	}
+	dev->dev_attrib.emulate_3pc = flag;
+	pr_debug("dev[%p]: SE Device 3rd Party Copy (EXTENDED_COPY): %d\n",
+		dev, flag);
+
+	return 0;
+}
+
 int se_dev_set_enforce_pr_isids(struct se_device *dev, int flag)
 {
 	if ((flag != 0) && (flag != 1)) {
@@ -1442,6 +1455,7 @@ struct se_device *target_alloc_device(struct se_hba *hba, const char *name)
 	dev->dev_attrib.emulate_tpu = DA_EMULATE_TPU;
 	dev->dev_attrib.emulate_tpws = DA_EMULATE_TPWS;
 	dev->dev_attrib.emulate_caw = DA_EMULATE_CAW;
+	dev->dev_attrib.emulate_3pc = DA_EMULATE_3PC;
 	dev->dev_attrib.enforce_pr_isids = DA_ENFORCE_PR_ISIDS;
 	dev->dev_attrib.is_nonrot = DA_IS_NONROT;
 	dev->dev_attrib.emulate_rest_reord = DA_EMULATE_REST_REORD;
diff --git a/drivers/target/target_core_internal.h b/drivers/target/target_core_internal.h
index 805ceb4..579128a 100644
--- a/drivers/target/target_core_internal.h
+++ b/drivers/target/target_core_internal.h
@@ -34,6 +34,7 @@ int	se_dev_set_emulate_tas(struct se_device *, int);
 int	se_dev_set_emulate_tpu(struct se_device *, int);
 int	se_dev_set_emulate_tpws(struct se_device *, int);
 int	se_dev_set_emulate_caw(struct se_device *, int);
+int	se_dev_set_emulate_3pc(struct se_device *, int);
 int	se_dev_set_enforce_pr_isids(struct se_device *, int);
 int	se_dev_set_is_nonrot(struct se_device *, int);
 int	se_dev_set_emulate_rest_reord(struct se_device *dev, int);
diff --git a/drivers/target/target_core_spc.c b/drivers/target/target_core_spc.c
index 894e83b..566dd27 100644
--- a/drivers/target/target_core_spc.c
+++ b/drivers/target/target_core_spc.c
@@ -95,6 +95,12 @@ spc_emulate_inquiry_std(struct se_cmd *cmd, unsigned char *buf)
 	 */
 	spc_fill_alua_data(lun->lun_sep, buf);
 
+	/*
+	 * Set Third-Party Copy (3PC) bit to indicate support for EXTENDED_COPY
+	 */
+	if (dev->dev_attrib.emulate_3pc)
+		buf[5] |= 0x8;
+
 	buf[7] = 0x2; /* CmdQue=1 */
 
 	snprintf(&buf[8], 8, "LIO-ORG");
diff --git a/include/target/target_core_base.h b/include/target/target_core_base.h
index f54a015..ba9ca79 100644
--- a/include/target/target_core_base.h
+++ b/include/target/target_core_base.h
@@ -99,6 +99,8 @@
 #define DA_EMULATE_TPWS				0
 /* Emulation for CompareAndWrite (AtomicTestandSet) by default */
 #define DA_EMULATE_CAW				1
+/* Emulation for 3rd Party Copy (ExtendedCopy) by default */
+#define DA_EMULATE_3PC				1
 /* No Emulation for PSCSI by default */
 #define DA_EMULATE_ALUA				0
 /* Enforce SCSI Initiator Port TransportID with 'ISID' for PR */
@@ -606,6 +608,7 @@ struct se_dev_attrib {
 	int		emulate_tpu;
 	int		emulate_tpws;
 	int		emulate_caw;
+	int		emulate_3pc;
 	int		enforce_pr_isids;
 	int		is_nonrot;
 	int		emulate_rest_reord;
-- 
1.7.2.5

[PATCH 9/9] target: Enable global EXTENDED_COPY setup/release

[Nicholas A. Bellinger]

From: Nicholas Bellinger <nab@daterainc.com>

Add calls to target_xcopy_setup_pt() + target_xcopy_release_pt() to
target_core_init_configfs() and target_core_exit_configfs()
respectively.

Cc: Christoph Hellwig <hch@lst.de>
Cc: Hannes Reinecke <hare@suse.de>
Cc: Martin Petersen <martin.petersen@oracle.com>
Cc: Chris Mason <chris.mason@fusionio.com>
Cc: Roland Dreier <roland@purestorage.com>
Cc: Zach Brown <zab@redhat.com>
Cc: James Bottomley <JBottomley@Parallels.com>
Cc: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Nicholas Bellinger <nab@daterainc.com>
---
 drivers/target/target_core_configfs.c |    6 ++++++
 1 files changed, 6 insertions(+), 0 deletions(-)

diff --git a/drivers/target/target_core_configfs.c b/drivers/target/target_core_configfs.c
index 026e42b..328f425 100644
--- a/drivers/target/target_core_configfs.c
+++ b/drivers/target/target_core_configfs.c
@@ -48,6 +48,7 @@
 #include "target_core_alua.h"
 #include "target_core_pr.h"
 #include "target_core_rd.h"
+#include "target_core_xcopy.h"
 
 extern struct t10_alua_lu_gp *default_lu_gp;
 
@@ -2935,6 +2936,10 @@ static int __init target_core_init_configfs(void)
 	if (ret < 0)
 		goto out;
 
+	ret = target_xcopy_setup_pt();
+	if (ret < 0)
+		goto out;
+
 	return 0;
 
 out:
@@ -3007,6 +3012,7 @@ static void __exit target_core_exit_configfs(void)
 
 	core_dev_release_virtual_lun0();
 	rd_module_exit();
+	target_xcopy_release_pt();
 	release_se_kmem_caches();
 }
 
-- 
1.7.2.5

Re: [PATCH 0/9] target: Add support for EXTENDED_COPY (VAAI) offload emulation

[Douglas Gilbert]

On 13-08-23 04:26 AM, Nicholas A. Bellinger wrote:
> From: Nicholas Bellinger <nab@daterainc.com>
>
> Hi folks!
>
> This series adds support to target-core for generic EXTENDED_COPY offload
> emulation as defined by SPC-4 using virtual (IBLOCK, FILEIO, RAMDISK)
> backends.
>
> EXTENDED_COPY is a VMWare ESX VAAI primative that is used to perform copy
> offload, that allows a target to perform local READ + WRITE I/O requests
> for bulk data transfers (cloning a virtual machine for example), instead
> of requiring these I/Os to actually be sent to/from the requesting SCSI
> initiator port.

Recently I have been looking at EXTENDED COPY since
T10 has been working on it. The SCSI opcodes associated
with it (0x83 and 0x84) have been renamed THIRD PARTY
COPY OUT and IN, and each have several service actions.
The Extended copy found in SPC-2 and SPC-3 is now
termed as "LID1" (for List Identifier length of 1 byte)
while the new stuff is termed "LID4" in SPC-4. The
"LID4" variants are "ROD" token based. A ROD
(Representation Of Data) token is 512 bytes long.

sg_xcopy (written by Hannes Reinecke and found in the
sg3_utils package) only supports LID1 but that covers
most of the existing hardware including kit that
supports VMWare ESX VAAI.

As defined by T10 both EXTENDED COPY (LID1) and (LID4)
do copies between disks, tapes and memory (all
combinations (e.g. disk->tape) (but maybe not memory
to memory).

There is a stripped down version of EXTENDED COPY(LID4)
that was called XCOPY LITE at one stage that only does
disk to disk copies (based on a ROD token). That is
the basis of Microsoft's Offload Data Transfer (ODX).
It uses commands defined in SBC-3 (e.g. POPULATE TOKEN
and WRITE USING TOKEN).


Confused? I certainly was. Feel free to correct and
clarify the above.

Doug Gilbert

Re: [PATCH 0/9] target: Add support for EXTENDED_COPY (VAAI) offload emulation

[Martin K. Petersen]

>>>>> "Doug" == Douglas Gilbert <dgilbert@interlog.com> writes:

Doug> The SCSI opcodes associated with it (0x83 and 0x84) have been
Doug> renamed THIRD PARTY COPY OUT and IN, and

Where did you see that? My SPC still has EXTENDED COPY.


Doug> Confused? I certainly was.

Yeah, this is UNMAP all over again, just 100 times worse :(

Anyway. Excited to see nab posting the patches! My copy offload code
from the spring has been getting stale both in the T10 and the kernel
sense. But at least we know what I'll be working on next week :)

-- 
Martin K. Petersen	Oracle Linux Engineering

Re: [PATCH 0/9] target: Add support for EXTENDED_COPY (VAAI) offload emulation

[Douglas Gilbert]

On 13-08-23 02:33 PM, Martin K. Petersen wrote:
>>>>>> "Doug" == Douglas Gilbert <dgilbert@interlog.com> writes:
>
> Doug> The SCSI opcodes associated with it (0x83 and 0x84) have been
> Doug> renamed THIRD PARTY COPY OUT and IN, and
>
> Where did you see that? My SPC still has EXTENDED COPY.

SCSI _opcodes_ == SCSI operation codes. In other words the
name associated with all service actions (commands) that have
as their first byte 0x83 and 0x84 .

spc4r36h.pdf Annex F.3.1 and changed in spc4r34. Yes, well
hidden but IMO useful. So now we can use the term "Third party
copy" to cover:
   - EXTENDED COPY(LID1) and associated commands
     also found in SPC-2 and SPC-3 (with the "LID1" suffix)
   - EXTENDED COPY(LID4) and associated commands
   - the XCOPY LITE commands: POPULATE TOKEN and WRITE USING TOKEN

> Doug> Confused? I certainly was.
>
> Yeah, this is UNMAP all over again, just 100 times worse :(

Well what EXTENDED COPY (offload copy) is trying to do
ain't simple but obviously there could be a substantial
performance pay-off.

There is a "Third-party copy implementation and usage" Annex
(D) in spc4r36h.pdf . It could do with some more explanatory
text.

> Anyway. Excited to see nab posting the patches! My copy offload code
> from the spring has been getting stale both in the T10 and the kernel
> sense. But at least we know what I'll be working on next week :)

BTW I have ported the sg_xcopy "LID1" xcopy logic into
my ddpt utility. That gives two advantages:
   - can cope with ibs!=obs
   - runs on OSes other than Linux

Doug Gilbert