d/numam-dpdk
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

examples/vhost_scsi: introduce a new sample app

Browse Source 
vhost-user protocol is common to many virtio devices, such as
virtio_net/virtio_scsi/virtio_blk. Since DPDK vhost library
removed the NET specific data structures, the vhost library
is common to other virtio devices, such as virtio-scsi.

Here we introduce a simple memory based block device that
can be presented to Guest VM through vhost-user-scsi-pci
controller. Similar with vhost-net, the sample application
will process the I/Os sent via virt rings.

Signed-off-by: Changpeng Liu <changpeng.liu@intel.com>
This commit is contained in:
Changpeng Liu 2017-07-20 17:16:29 +08:00 committed by Yuanhan Liu
parent 78b2e3bae1
commit db75c7af19
9 changed files with 1792 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
MAINTAINERS
View File

@ -447,6 +447,8 @@ F: lib/librte_vhost/
F: doc/guides/prog_guide/vhost_lib.rst
F: examples/vhost/
F: doc/guides/sample_app_ug/vhost.rst
F: examples/vhost_scsi/
F: doc/guides/sample_app_ug/vhost_scsi.rst
Vhost PMD
M: Tetsuya Mukawa <mtetsuyah@gmail.com>

1
doc/guides/sample_app_ug/index.rst
View File

@ -66,6 +66,7 @@ Sample Applications User Guides
packet_ordering
vmdq_dcb_forwarding
vhost
vhost_scsi
netmap_compatibility
ip_pipeline
test_pipeline

115
doc/guides/sample_app_ug/vhost_scsi.rst Normal file
View File

@ -0,0 +1,115 @@
.. BSD LICENSE
Copyright(c) 2010-2017 Intel Corporation. All rights reserved.
All rights reserved.
Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions
are met:
* Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in
the documentation and/or other materials provided with the
distribution.
* Neither the name of Intel Corporation nor the names of its
contributors may be used to endorse or promote products derived
from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Vhost_scsi Sample Application
=============================
The vhost_scsi sample application implemented a simple SCSI block device,
which used as the backend of Qemu vhost-user-scsi device. Users can extend
the exist example to use other type of block device(e.g. AIO) besides
memory based block device. Similar with vhost-user-net device, the sample
application used domain socket to communicate with Qemu, and the virtio
ring was processed by vhost_scsi sample application.
The sample application reuse lots codes from SPDK(Storage Performance
Development Kit, https://github.com/spdk/spdk) vhost-user-scsi target,
for DPDK vhost library used in storage area, user can take SPDK as
reference as well.
Testing steps
-------------
This section shows the steps how to start a VM with the block device as
fast data path for critical application.
Build
~~~~~
Follow the *Getting Started Guide for Linux* on generic info about
environment setup and building DPDK from source.
In this example, you need build DPDK both on the host and inside guest.
Also, you need build this example.
.. code-block:: console
export RTE_SDK=/path/to/dpdk_source
export RTE_TARGET=x86_64-native-linuxapp-gcc
cd ${RTE_SDK}/examples/vhost_scsi
make
Start the vhost_scsi example
~~~~~~~~~~~~~~~~~~~~~~~~~~~~
.. code-block:: console
./vhost_scsi -m 1024
.. _vhost_scsi_app_run_vm:
Start the VM
~~~~~~~~~~~~
.. code-block:: console
qemu-system-x86_64 -machine accel=kvm \
-m $mem -object memory-backend-file,id=mem,size=$mem,\
mem-path=/dev/hugepages,share=on -numa node,memdev=mem \
-drive file=os.img,if=none,id=disk \
-device ide-hd,drive=disk,bootindex=0 \
-chardev socket,id=char0,path=/tmp/vhost.socket \
-device vhost-user-scsi-pci,chardev=char0,bootindex=2 \
...
.. note::
You must check whether your Qemu can support "vhost-user-scsi" or not,
Qemu v2.10 or newer version is required.
Vhost_scsi Common Issues
------------------------
* vhost_scsi can not start with block size 512 Bytes:
Currently DPDK vhost library was designed for NET device(althrough the APIs
are generic now), for 512 Bytes block device, Qemu BIOS(x86 BIOS Enhanced
Disk Device) will enumerate all block device and do some IOs to those block
devices with 512 Bytes sector size. DPDK vhost library can not process such
scenarios(both BIOS and OS will enumerate the block device), so as a
workaround, the vhost_scsi example application hardcoded the block size
with 4096 Bytes.
* vhost_scsi can only support the block device as fast data disk(non OS image):
Make sure ``bootindex=2`` Qemu option is given to vhost-user-scsi-pci device.

2
examples/Makefile
View File

@ -88,7 +88,7 @@ ifeq ($(CONFIG_RTE_LIBRTE_HASH),y)
DIRS-$(CONFIG_RTE_LIBRTE_VHOST) += tep_termination
endif
DIRS-$(CONFIG_RTE_LIBRTE_TIMER) += timer
DIRS-$(CONFIG_RTE_LIBRTE_VHOST) += vhost
DIRS-$(CONFIG_RTE_LIBRTE_VHOST) += vhost vhost_scsi
DIRS-$(CONFIG_RTE_LIBRTE_XEN_DOM0) += vhost_xen
DIRS-y += vmdq
DIRS-y += vmdq_dcb

59
examples/vhost_scsi/Makefile Normal file
View File

@ -0,0 +1,59 @@
# BSD LICENSE
#
# Copyright(c) 2010-2017 Intel Corporation. All rights reserved.
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions
# are met:
#
# * Redistributions of source code must retain the above copyright
# notice, this list of conditions and the following disclaimer.
# * Redistributions in binary form must reproduce the above copyright
# notice, this list of conditions and the following disclaimer in
# the documentation and/or other materials provided with the
# distribution.
# * Neither the name of Intel Corporation nor the names of its
# contributors may be used to endorse or promote products derived
# from this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
ifeq ($(RTE_SDK),)
$(error "Please define RTE_SDK environment variable")
endif
# Default target, can be overridden by command line or environment
RTE_TARGET ?= x86_64-native-linuxapp-gcc
include $(RTE_SDK)/mk/rte.vars.mk
ifneq ($(CONFIG_RTE_EXEC_ENV),"linuxapp")
$(info This application can only operate in a linuxapp environment, \
please change the definition of the RTE_TARGET environment variable)
all:
else
# binary name
APP = vhost-scsi
# all source are stored in SRCS-y
SRCS-y := scsi.c vhost_scsi.c
CFLAGS += -O2 -D_FILE_OFFSET_BITS=64
CFLAGS += $(WERROR_FLAGS)
CFLAGS += -D_GNU_SOURCE
include $(RTE_SDK)/mk/rte.extapp.mk
endif

539
examples/vhost_scsi/scsi.c Normal file
View File

@ -0,0 +1,539 @@
/*-
* BSD LICENSE
*
* Copyright(c) 2010-2017 Intel Corporation. All rights reserved.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
* * Neither the name of Intel Corporation nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/**
* This work is largely based on the "vhost-user-scsi" implementation by
* SPDK(https://github.com/spdk/spdk).
*/
#include <stdio.h>
#include <stdint.h>
#include <unistd.h>
#include <assert.h>
#include <ctype.h>
#include <string.h>
#include <stddef.h>
#include <rte_atomic.h>
#include <rte_cycles.h>
#include <rte_log.h>
#include <rte_malloc.h>
#include <rte_byteorder.h>
#include "vhost_scsi.h"
#include "scsi_spec.h"
#define INQ_OFFSET(field) (offsetof(struct scsi_cdb_inquiry_data, field) + \
sizeof(((struct scsi_cdb_inquiry_data *)0x0)->field))
static void
vhost_strcpy_pad(void *dst, const char *src, size_t size, int pad)
{
size_t len;
len = strlen(src);
if (len < size) {
memcpy(dst, src, len);
memset((char *)dst + len, pad, size - len);
} else {
memcpy(dst, src, size);
}
}
static int
vhost_hex2bin(char ch)
{
if ((ch >= '0') && (ch <= '9'))
return ch - '0';
ch = tolower(ch);
if ((ch >= 'a') && (ch <= 'f'))
return ch - 'a' + 10;
return (int)ch;
}
static void
vhost_bdev_scsi_set_naa_ieee_extended(const char *name, uint8_t *buf)
{
int i, value, count = 0;
uint64_t *temp64, local_value;
for (i = 0; (i < 16) && (name[i] != '\0'); i++) {
value = vhost_hex2bin(name[i]);
if (i % 2)
buf[count++] |= value << 4;
else
buf[count] = value;
}
local_value = *(uint64_t *)buf;
/*
* see spc3r23 7.6.3.6.2,
* NAA IEEE Extended identifer format
*/
local_value &= 0x0fff000000ffffffull;
/* NAA 02, and 00 03 47 for IEEE Intel */
local_value |= 0x2000000347000000ull;
temp64 = (uint64_t *)buf;
*temp64 = rte_cpu_to_be_64(local_value);
}
static void
scsi_task_build_sense_data(struct vhost_scsi_task *task, int sk,
int asc, int ascq)
{
uint8_t *cp;
int resp_code;
resp_code = 0x70; /* Current + Fixed format */
/* Sense Data */
cp = (uint8_t *)task->resp->sense;
/* VALID(7) RESPONSE CODE(6-0) */
cp[0] = 0x80 | resp_code;
/* Obsolete */
cp[1] = 0;
/* FILEMARK(7) EOM(6) ILI(5) SENSE KEY(3-0) */
cp[2] = sk & 0xf;
/* INFORMATION */
memset(&cp[3], 0, 4);
/* ADDITIONAL SENSE LENGTH */
cp[7] = 10;
/* COMMAND-SPECIFIC INFORMATION */
memset(&cp[8], 0, 4);
/* ADDITIONAL SENSE CODE */
cp[12] = asc;
/* ADDITIONAL SENSE CODE QUALIFIER */
cp[13] = ascq;
/* FIELD REPLACEABLE UNIT CODE */
cp[14] = 0;
/* SKSV(7) SENSE KEY SPECIFIC(6-0,7-0,7-0) */
cp[15] = 0;
cp[16] = 0;
cp[17] = 0;
/* SenseLength */
task->resp->sense_len = 18;
}
static void
scsi_task_set_status(struct vhost_scsi_task *task, int sc, int sk,
int asc, int ascq)
{
if (sc == SCSI_STATUS_CHECK_CONDITION)
scsi_task_build_sense_data(task, sk, asc, ascq);
task->resp->status = sc;
}
static int
vhost_bdev_scsi_inquiry_command(struct vhost_block_dev *bdev,
struct vhost_scsi_task *task)
{
int hlen = 0;
uint32_t alloc_len = 0;
uint16_t len = 0;
uint16_t *temp16;
int pc;
int pd;
int evpd;
int i;
uint8_t *buf;
struct scsi_cdb_inquiry *inq;
inq = (struct scsi_cdb_inquiry *)task->req->cdb;
assert(task->iovs_cnt == 1);
/* At least 36Bytes for inquiry command */
if (task->data_len < 0x24)
goto inq_error;
pd = SPC_PERIPHERAL_DEVICE_TYPE_DISK;
pc = inq->page_code;
evpd = inq->evpd & 0x1;
if (!evpd && pc)
goto inq_error;
if (evpd) {
struct scsi_vpd_page *vpage = (struct scsi_vpd_page *)
task->iovs[0].iov_base;
/* PERIPHERAL QUALIFIER(7-5) PERIPHERAL DEVICE TYPE(4-0) */
vpage->peripheral = pd;
/* PAGE CODE */
vpage->page_code = pc;
switch (pc) {
case SPC_VPD_SUPPORTED_VPD_PAGES:
hlen = 4;
vpage->params[0] = SPC_VPD_SUPPORTED_VPD_PAGES;
vpage->params[1] = SPC_VPD_UNIT_SERIAL_NUMBER;
vpage->params[2] = SPC_VPD_DEVICE_IDENTIFICATION;
len = 3;
/* PAGE LENGTH */
vpage->alloc_len = rte_cpu_to_be_16(len);
break;
case SPC_VPD_UNIT_SERIAL_NUMBER:
hlen = 4;
strncpy((char *)vpage->params, bdev->name, 32);
vpage->alloc_len = rte_cpu_to_be_16(32);
break;
case SPC_VPD_DEVICE_IDENTIFICATION:
buf = vpage->params;
struct scsi_desig_desc *desig;
hlen = 4;
/* NAA designator */
desig = (struct scsi_desig_desc *)buf;
desig->code_set = SPC_VPD_CODE_SET_BINARY;
desig->protocol_id = SPC_PROTOCOL_IDENTIFIER_ISCSI;
desig->type = SPC_VPD_IDENTIFIER_TYPE_NAA;
desig->association = SPC_VPD_ASSOCIATION_LOGICAL_UNIT;
desig->reserved0 = 0;
desig->piv = 1;
desig->reserved1 = 0;
desig->len = 8;
vhost_bdev_scsi_set_naa_ieee_extended(bdev->name,
desig->desig);
len = sizeof(struct scsi_desig_desc) + 8;
buf += sizeof(struct scsi_desig_desc) + desig->len;
/* T10 Vendor ID designator */
desig = (struct scsi_desig_desc *)buf;
desig->code_set = SPC_VPD_CODE_SET_ASCII;
desig->protocol_id = SPC_PROTOCOL_IDENTIFIER_ISCSI;
desig->type = SPC_VPD_IDENTIFIER_TYPE_T10_VENDOR_ID;
desig->association = SPC_VPD_ASSOCIATION_LOGICAL_UNIT;
desig->reserved0 = 0;
desig->piv = 1;
desig->reserved1 = 0;
desig->len = 8 + 16 + 32;
strncpy((char *)desig->desig, "INTEL", 8);
vhost_strcpy_pad((char *)&desig->desig[8],
bdev->product_name, 16, ' ');
strncpy((char *)&desig->desig[24], bdev->name, 32);
len += sizeof(struct scsi_desig_desc) + 8 + 16 + 32;
buf += sizeof(struct scsi_desig_desc) + desig->len;
/* SCSI Device Name designator */
desig = (struct scsi_desig_desc *)buf;
desig->code_set = SPC_VPD_CODE_SET_UTF8;
desig->protocol_id = SPC_PROTOCOL_IDENTIFIER_ISCSI;
desig->type = SPC_VPD_IDENTIFIER_TYPE_SCSI_NAME;
desig->association = SPC_VPD_ASSOCIATION_TARGET_DEVICE;
desig->reserved0 = 0;
desig->piv = 1;
desig->reserved1 = 0;
desig->len = snprintf((char *)desig->desig,
255, "%s", bdev->name);
len += sizeof(struct scsi_desig_desc) + desig->len;
buf += sizeof(struct scsi_desig_desc) + desig->len;
vpage->alloc_len = rte_cpu_to_be_16(len);
break;
default:
goto inq_error;
}
} else {
struct scsi_cdb_inquiry_data *inqdata =
(struct scsi_cdb_inquiry_data *)task->iovs[0].iov_base;
/* Standard INQUIRY data */
/* PERIPHERAL QUALIFIER(7-5) PERIPHERAL DEVICE TYPE(4-0) */
inqdata->peripheral = pd;
/* RMB(7) */
inqdata->rmb = 0;
/* VERSION */
/* See SPC3/SBC2/MMC4/SAM2 for more details */
inqdata->version = SPC_VERSION_SPC3;
/* NORMACA(5) HISUP(4) RESPONSE DATA FORMAT(3-0) */
/* format 2 */ /* hierarchical support */
inqdata->response = 2 | 1 << 4;
hlen = 5;
/* SCCS(7) ACC(6) TPGS(5-4) 3PC(3) PROTECT(0) */
/* Not support TPGS */
inqdata->flags = 0;
/* MULTIP */
inqdata->flags2 = 0x10;
/* WBUS16(5) SYNC(4) LINKED(3) CMDQUE(1) VS(0) */
/* CMDQUE */
inqdata->flags3 = 0x2;
/* T10 VENDOR IDENTIFICATION */
strncpy((char *)inqdata->t10_vendor_id, "INTEL", 8);
/* PRODUCT IDENTIFICATION */
strncpy((char *)inqdata->product_id, bdev->product_name, 16);
/* PRODUCT REVISION LEVEL */
strncpy((char *)inqdata->product_rev, "0001", 4);
/* Standard inquiry data ends here. Only populate
* remaining fields if alloc_len indicates enough
* space to hold it.
*/
len = INQ_OFFSET(product_rev) - 5;
if (alloc_len >= INQ_OFFSET(vendor)) {
/* Vendor specific */
memset(inqdata->vendor, 0x20, 20);
len += sizeof(inqdata->vendor);
}
if (alloc_len >= INQ_OFFSET(ius)) {
/* CLOCKING(3-2) QAS(1) IUS(0) */
inqdata->ius = 0;
len += sizeof(inqdata->ius);
}
if (alloc_len >= INQ_OFFSET(reserved)) {
/* Reserved */
inqdata->reserved = 0;
len += sizeof(inqdata->reserved);
}
/* VERSION DESCRIPTOR 1-8 */
if (alloc_len >= INQ_OFFSET(reserved) + 2) {
temp16 = (uint16_t *)&inqdata->desc[0];
*temp16 = rte_cpu_to_be_16(0x0960);
len += 2;
}
if (alloc_len >= INQ_OFFSET(reserved) + 4) {
/* SPC-3 (no version claimed) */
temp16 = (uint16_t *)&inqdata->desc[2];
*temp16 = rte_cpu_to_be_16(0x0300);
len += 2;
}
if (alloc_len >= INQ_OFFSET(reserved) + 6) {
/* SBC-2 (no version claimed) */
temp16 = (uint16_t *)&inqdata->desc[4];
*temp16 = rte_cpu_to_be_16(0x0320);
len += 2;
}
if (alloc_len >= INQ_OFFSET(reserved) + 8) {
/* SAM-2 (no version claimed) */
temp16 = (uint16_t *)&inqdata->desc[6];
*temp16 = rte_cpu_to_be_16(0x0040);
len += 2;
}
if (alloc_len > INQ_OFFSET(reserved) + 8) {
i = alloc_len - (INQ_OFFSET(reserved) + 8);
if (i > 30)
i = 30;
memset(&inqdata->desc[8], 0, i);
len += i;
}
/* ADDITIONAL LENGTH */
inqdata->add_len = len;
}
/* STATUS GOOD */
scsi_task_set_status(task, SCSI_STATUS_GOOD, 0, 0, 0);
return hlen + len;
inq_error:
scsi_task_set_status(task, SCSI_STATUS_CHECK_CONDITION,
SCSI_SENSE_ILLEGAL_REQUEST,
SCSI_ASC_INVALID_FIELD_IN_CDB,
SCSI_ASCQ_CAUSE_NOT_REPORTABLE);
return 0;
}
static int
vhost_bdev_scsi_readwrite(struct vhost_block_dev *bdev,
struct vhost_scsi_task *task,
uint64_t lba, __rte_unused uint32_t xfer_len)
{
uint32_t i;
uint64_t offset;
uint32_t nbytes = 0;
offset = lba * bdev->blocklen;
for (i = 0; i < task->iovs_cnt; i++) {
if (task->dxfer_dir == SCSI_DIR_TO_DEV)
memcpy(bdev->data + offset, task->iovs[i].iov_base,
task->iovs[i].iov_len);
else
memcpy(task->iovs[i].iov_base, bdev->data + offset,
task->iovs[i].iov_len);
offset += task->iovs[i].iov_len;
nbytes += task->iovs[i].iov_len;
}
return nbytes;
}
static int
vhost_bdev_scsi_process_block(struct vhost_block_dev *bdev,
struct vhost_scsi_task *task)
{
uint64_t lba, *temp64;
uint32_t xfer_len, *temp32;
uint16_t *temp16;
uint8_t *cdb = (uint8_t *)task->req->cdb;
switch (cdb[0]) {
case SBC_READ_6:
case SBC_WRITE_6:
lba = (uint64_t)cdb[1] << 16;
lba |= (uint64_t)cdb[2] << 8;
lba |= (uint64_t)cdb[3];
xfer_len = cdb[4];
if (xfer_len == 0)
xfer_len = 256;
return vhost_bdev_scsi_readwrite(bdev, task, lba, xfer_len);
case SBC_READ_10:
case SBC_WRITE_10:
temp32 = (uint32_t *)&cdb[2];
lba = rte_be_to_cpu_32(*temp32);
temp16 = (uint16_t *)&cdb[7];
xfer_len = rte_be_to_cpu_16(*temp16);
return vhost_bdev_scsi_readwrite(bdev, task, lba, xfer_len);
case SBC_READ_12:
case SBC_WRITE_12:
temp32 = (uint32_t *)&cdb[2];
lba = rte_be_to_cpu_32(*temp32);
temp32 = (uint32_t *)&cdb[6];
xfer_len = rte_be_to_cpu_32(*temp32);
return vhost_bdev_scsi_readwrite(bdev, task, lba, xfer_len);
case SBC_READ_16:
case SBC_WRITE_16:
temp64 = (uint64_t *)&cdb[2];
lba = rte_be_to_cpu_64(*temp64);
temp32 = (uint32_t *)&cdb[10];
xfer_len = rte_be_to_cpu_32(*temp32);
return vhost_bdev_scsi_readwrite(bdev, task, lba, xfer_len);
case SBC_READ_CAPACITY_10: {
uint8_t buffer[8];
if (bdev->blockcnt - 1 > 0xffffffffULL)
memset(buffer, 0xff, 4);
else {
temp32 = (uint32_t *)buffer;
*temp32 = rte_cpu_to_be_32(bdev->blockcnt - 1);
}
temp32 = (uint32_t *)&buffer[4];
*temp32 = rte_cpu_to_be_32(bdev->blocklen);
memcpy(task->iovs[0].iov_base, buffer, sizeof(buffer));
task->resp->status = SCSI_STATUS_GOOD;
return sizeof(buffer);
}
case SBC_SYNCHRONIZE_CACHE_10:
case SBC_SYNCHRONIZE_CACHE_16:
task->resp->status = SCSI_STATUS_GOOD;
return 0;
}
scsi_task_set_status(task, SCSI_STATUS_CHECK_CONDITION,
SCSI_SENSE_ILLEGAL_REQUEST,
SCSI_ASC_INVALID_FIELD_IN_CDB,
SCSI_ASCQ_CAUSE_NOT_REPORTABLE);
return 0;
}
int
vhost_bdev_process_scsi_commands(struct vhost_block_dev *bdev,
struct vhost_scsi_task *task)
{
int len;
uint8_t *data;
uint64_t *temp64, fmt_lun = 0;
uint32_t *temp32;
const uint8_t *lun;
uint8_t *cdb = (uint8_t *)task->req->cdb;
lun = (const uint8_t *)task->req->lun;
/* only 1 LUN supported */
if (lun[0] != 1 || lun[1] >= 1)
return -1;
switch (cdb[0]) {
case SPC_INQUIRY:
len = vhost_bdev_scsi_inquiry_command(bdev, task);
task->data_len = len;
break;
case SPC_REPORT_LUNS:
data = (uint8_t *)task->iovs[0].iov_base;
fmt_lun |= (0x0ULL & 0x00ffULL) << 48;
temp64 = (uint64_t *)&data[8];
*temp64 = rte_cpu_to_be_64(fmt_lun);
temp32 = (uint32_t *)data;
*temp32 = rte_cpu_to_be_32(8);
task->data_len = 16;
scsi_task_set_status(task, SCSI_STATUS_GOOD, 0, 0, 0);
break;
case SPC_MODE_SELECT_6:
case SPC_MODE_SELECT_10:
/* don't support it now */
scsi_task_set_status(task, SCSI_STATUS_GOOD, 0, 0, 0);
break;
case SPC_MODE_SENSE_6:
case SPC_MODE_SENSE_10:
/* don't support it now */
scsi_task_set_status(task, SCSI_STATUS_GOOD, 0, 0, 0);
break;
case SPC_TEST_UNIT_READY:
scsi_task_set_status(task, SCSI_STATUS_GOOD, 0, 0, 0);
break;
default:
len = vhost_bdev_scsi_process_block(bdev, task);
task->data_len = len;
}
return 0;
}

493
examples/vhost_scsi/scsi_spec.h Normal file
View File

@ -0,0 +1,493 @@
/*-
* BSD LICENSE
*
* Copyright (c) Intel Corporation.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
* * Neither the name of Intel Corporation nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
/**
* SCSI specification definition
* refer http://www.t10.org/drafts.htm#SPC_Family for SPC-3 and SBC-3
*/
#ifndef _SCSI_SPEC_H
#define _SCSI_SPEC_H
#include <stdint.h>
enum scsi_group_code {
SCSI_6BYTE_CMD = 0x00,
SCSI_10BYTE_CMD = 0x20,
SCSI_10BYTE_CMD2 = 0x40,
SCSI_16BYTE_CMD = 0x80,
SCSI_12BYTE_CMD = 0xa0,
};
#define SCSI_GROUP_MASK 0xe0
#define SCSI_OPCODE_MASK 0x1f
enum scsi_status {
SCSI_STATUS_GOOD = 0x00,
SCSI_STATUS_CHECK_CONDITION = 0x02,
SCSI_STATUS_CONDITION_MET = 0x04,
SCSI_STATUS_BUSY = 0x08,
SCSI_STATUS_INTERMEDIATE = 0x10,
SCSI_STATUS_INTERMEDIATE_CONDITION_MET = 0x14,
SCSI_STATUS_RESERVATION_CONFLICT = 0x18,
SCSI_STATUS_Obsolete = 0x22,
SCSI_STATUS_TASK_SET_FULL = 0x28,
SCSI_STATUS_ACA_ACTIVE = 0x30,
SCSI_STATUS_TASK_ABORTED = 0x40,
};
enum scsi_sense {
SCSI_SENSE_NO_SENSE = 0x00,
SCSI_SENSE_RECOVERED_ERROR = 0x01,
SCSI_SENSE_NOT_READY = 0x02,
SCSI_SENSE_MEDIUM_ERROR = 0x03,
SCSI_SENSE_HARDWARE_ERROR = 0x04,
SCSI_SENSE_ILLEGAL_REQUEST = 0x05,
SCSI_SENSE_UNIT_ATTENTION = 0x06,
SCSI_SENSE_DATA_PROTECT = 0x07,
SCSI_SENSE_BLANK_CHECK = 0x08,
SCSI_SENSE_VENDOR_SPECIFIC = 0x09,
SCSI_SENSE_COPY_ABORTED = 0x0a,
SCSI_SENSE_ABORTED_COMMAND = 0x0b,
SCSI_SENSE_VOLUME_OVERFLOW = 0x0d,
SCSI_SENSE_MISCOMPARE = 0x0e,
};
enum scsi_asc {
SCSI_ASC_NO_ADDITIONAL_SENSE = 0x00,
SCSI_ASC_PERIPHERAL_DEVICE_WRITE_FAULT = 0x03,
SCSI_ASC_LOGICAL_UNIT_NOT_READY = 0x04,
SCSI_ASC_WARNING = 0x0b,
SCSI_ASC_LOGICAL_BLOCK_GUARD_CHECK_FAILED = 0x10,
SCSI_ASC_LOGICAL_BLOCK_APP_TAG_CHECK_FAILED = 0x10,
SCSI_ASC_LOGICAL_BLOCK_REF_TAG_CHECK_FAILED = 0x10,
SCSI_ASC_UNRECOVERED_READ_ERROR = 0x11,
SCSI_ASC_MISCOMPARE_DURING_VERIFY_OPERATION = 0x1d,
SCSI_ASC_INVALID_COMMAND_OPERATION_CODE = 0x20,
SCSI_ASC_ACCESS_DENIED = 0x20,
SCSI_ASC_LOGICAL_BLOCK_ADDRESS_OUT_OF_RANGE = 0x21,
SCSI_ASC_INVALID_FIELD_IN_CDB = 0x24,
SCSI_ASC_LOGICAL_UNIT_NOT_SUPPORTED = 0x25,
SCSI_ASC_WRITE_PROTECTED = 0x27,
SCSI_ASC_FORMAT_COMMAND_FAILED = 0x31,
SCSI_ASC_INTERNAL_TARGET_FAILURE = 0x44,
};
enum scsi_ascq {
SCSI_ASCQ_CAUSE_NOT_REPORTABLE = 0x00,
SCSI_ASCQ_BECOMING_READY = 0x01,
SCSI_ASCQ_FORMAT_COMMAND_FAILED = 0x01,
SCSI_ASCQ_LOGICAL_BLOCK_GUARD_CHECK_FAILED = 0x01,
SCSI_ASCQ_LOGICAL_BLOCK_APP_TAG_CHECK_FAILED = 0x02,
SCSI_ASCQ_NO_ACCESS_RIGHTS = 0x02,
SCSI_ASCQ_LOGICAL_BLOCK_REF_TAG_CHECK_FAILED = 0x03,
SCSI_ASCQ_POWER_LOSS_EXPECTED = 0x08,
SCSI_ASCQ_INVALID_LU_IDENTIFIER = 0x09,
};
enum spc_opcode {
/* SPC3 related */
SPC_ACCESS_CONTROL_IN = 0x86,
SPC_ACCESS_CONTROL_OUT = 0x87,
SPC_EXTENDED_COPY = 0x83,
SPC_INQUIRY = 0x12,
SPC_LOG_SELECT = 0x4c,
SPC_LOG_SENSE = 0x4d,
SPC_MODE_SELECT_6 = 0x15,
SPC_MODE_SELECT_10 = 0x55,
SPC_MODE_SENSE_6 = 0x1a,
SPC_MODE_SENSE_10 = 0x5a,
SPC_PERSISTENT_RESERVE_IN = 0x5e,
SPC_PERSISTENT_RESERVE_OUT = 0x5f,
SPC_PREVENT_ALLOW_MEDIUM_REMOVAL = 0x1e,
SPC_READ_ATTRIBUTE = 0x8c,
SPC_READ_BUFFER = 0x3c,
SPC_RECEIVE_COPY_RESULTS = 0x84,
SPC_RECEIVE_DIAGNOSTIC_RESULTS = 0x1c,
SPC_REPORT_LUNS = 0xa0,
SPC_REQUEST_SENSE = 0x03,
SPC_SEND_DIAGNOSTIC = 0x1d,
SPC_TEST_UNIT_READY = 0x00,
SPC_WRITE_ATTRIBUTE = 0x8d,
SPC_WRITE_BUFFER = 0x3b,
SPC_SERVICE_ACTION_IN_12 = 0xab,
SPC_SERVICE_ACTION_OUT_12 = 0xa9,
SPC_SERVICE_ACTION_IN_16 = 0x9e,
SPC_SERVICE_ACTION_OUT_16 = 0x9f,
SPC_VARIABLE_LENGTH = 0x7f,
SPC_MO_CHANGE_ALIASES = 0x0b,
SPC_MO_SET_DEVICE_IDENTIFIER = 0x06,
SPC_MO_SET_PRIORITY = 0x0e,
SPC_MO_SET_TARGET_PORT_GROUPS = 0x0a,
SPC_MO_SET_TIMESTAMP = 0x0f,
SPC_MI_REPORT_ALIASES = 0x0b,
SPC_MI_REPORT_DEVICE_IDENTIFIER = 0x05,
SPC_MI_REPORT_PRIORITY = 0x0e,
SPC_MI_REPORT_SUPPORTED_OPERATION_CODES = 0x0c,
SPC_MI_REPORT_SUPPORTED_TASK_MANAGEMENT_FUNCTIONS = 0x0d,
SPC_MI_REPORT_TARGET_PORT_GROUPS = 0x0a,
SPC_MI_REPORT_TIMESTAMP = 0x0f,
/* SPC2 related (Obsolete) */
SPC2_RELEASE_6 = 0x17,
SPC2_RELEASE_10 = 0x57,
SPC2_RESERVE_6 = 0x16,
SPC2_RESERVE_10 = 0x56,
};
enum scc_opcode {
SCC_MAINTENANCE_IN = 0xa3,
SCC_MAINTENANCE_OUT = 0xa4,
};
enum sbc_opcode {
SBC_COMPARE_AND_WRITE = 0x89,
SBC_FORMAT_UNIT = 0x04,
SBC_GET_LBA_STATUS = 0x0012009e,
SBC_ORWRITE_16 = 0x8b,
SBC_PRE_FETCH_10 = 0x34,
SBC_PRE_FETCH_16 = 0x90,
SBC_READ_6 = 0x08,
SBC_READ_10 = 0x28,
SBC_READ_12 = 0xa8,
SBC_READ_16 = 0x88,
SBC_READ_ATTRIBUTE = 0x8c,
SBC_READ_BUFFER = 0x3c,
SBC_READ_CAPACITY_10 = 0x25,
SBC_READ_DEFECT_DATA_10 = 0x37,
SBC_READ_DEFECT_DATA_12 = 0xb7,
SBC_READ_LONG_10 = 0x3e,
SBC_REASSIGN_BLOCKS = 0x07,
SBC_SANITIZE = 0x48,
SBC_START_STOP_UNIT = 0x1b,
SBC_SYNCHRONIZE_CACHE_10 = 0x35,
SBC_SYNCHRONIZE_CACHE_16 = 0x91,
SBC_UNMAP = 0x42,
SBC_VERIFY_10 = 0x2f,
SBC_VERIFY_12 = 0xaf,
SBC_VERIFY_16 = 0x8f,
SBC_WRITE_6 = 0x0a,
SBC_WRITE_10 = 0x2a,
SBC_WRITE_12 = 0xaa,
SBC_WRITE_16 = 0x8a,
SBC_WRITE_AND_VERIFY_10 = 0x2e,
SBC_WRITE_AND_VERIFY_12 = 0xae,
SBC_WRITE_AND_VERIFY_16 = 0x8e,
SBC_WRITE_LONG_10 = 0x3f,
SBC_WRITE_SAME_10 = 0x41,
SBC_WRITE_SAME_16 = 0x93,
SBC_XDREAD_10 = 0x52,
SBC_XDWRITE_10 = 0x50,
SBC_XDWRITEREAD_10 = 0x53,
SBC_XPWRITE_10 = 0x51,
SBC_SAI_READ_CAPACITY_16 = 0x10,
SBC_SAI_READ_LONG_16 = 0x11,
SBC_SAO_WRITE_LONG_16 = 0x11,
SBC_VL_READ_32 = 0x0009,
SBC_VL_VERIFY_32 = 0x000a,
SBC_VL_WRITE_32 = 0x000b,
SBC_VL_WRITE_AND_VERIFY_32 = 0x000c,
SBC_VL_WRITE_SAME_32 = 0x000d,
SBC_VL_XDREAD_32 = 0x0003,
SBC_VL_XDWRITE_32 = 0x0004,
SBC_VL_XDWRITEREAD_32 = 0x0007,
SBC_VL_XPWRITE_32 = 0x0006,
};
enum mmc_opcode {
/* MMC6 */
MMC_READ_DISC_STRUCTURE = 0xad,
/* MMC4 */
MMC_BLANK = 0xa1,
MMC_CLOSE_TRACK_SESSION = 0x5b,
MMC_ERASE_10 = 0x2c,
MMC_FORMAT_UNIT = 0x04,
MMC_GET_CONFIGURATION = 0x46,
MMC_GET_EVENT_STATUS_NOTIFICATION = 0x4a,
MMC_GET_PERFORMANCE = 0xac,
MMC_INQUIRY = 0x12,
MMC_LOAD_UNLOAD_MEDIUM = 0xa6,
MMC_MECHANISM_STATUS = 0xbd,
MMC_MODE_SELECT_10 = 0x55,
MMC_MODE_SENSE_10 = 0x5a,
MMC_PAUSE_RESUME = 0x4b,
MMC_PLAY_AUDIO_10 = 0x45,
MMC_PLAY_AUDIO_12 = 0xa5,
MMC_PLAY_AUDIO_MSF = 0x47,
MMC_PREVENT_ALLOW_MEDIUM_REMOVAL = 0x1e,
MMC_READ_10 = 0x28,
MMC_READ_12 = 0xa8,
MMC_READ_BUFFER = 0x3c,
MMC_READ_BUFFER_CAPACITY = 0x5c,
MMC_READ_CAPACITY = 0x25,
MMC_READ_CD = 0xbe,
MMC_READ_CD_MSF = 0xb9,
MMC_READ_DISC_INFORMATION = 0x51,
MMC_READ_DVD_STRUCTURE = 0xad,
MMC_READ_FORMAT_CAPACITIES = 0x23,
MMC_READ_SUB_CHANNEL = 0x42,
MMC_READ_TOC_PMA_ATIP = 0x43,
MMC_READ_TRACK_INFORMATION = 0x52,
MMC_REPAIR_TRACK = 0x58,
MMC_REPORT_KEY = 0xa4,
MMC_REQUEST_SENSE = 0x03,
MMC_RESERVE_TRACK = 0x53,
MMC_SCAN = 0xba,
MMC_SEEK_10 = 0x2b,
MMC_SEND_CUE_SHEET = 0x5d,
MMC_SEND_DVD_STRUCTURE = 0xbf,
MMC_SEND_KEY = 0xa3,
MMC_SEND_OPC_INFORMATION = 0x54,
MMC_SET_CD_SPEED = 0xbb,
MMC_SET_READ_AHEAD = 0xa7,
MMC_SET_STREAMING = 0xb6,
MMC_START_STOP_UNIT = 0x1b,
MMC_STOP_PLAY_SCAN = 0x4e,
MMC_SYNCHRONIZE_CACHE = 0x35,
MMC_TEST_UNIT_READY = 0x00,
MMC_VERIFY_10 = 0x2f,
MMC_WRITE_10 = 0xa2,
MMC_WRITE_12 = 0xaa,
MMC_WRITE_AND_VERIFY_10 = 0x2e,
MMC_WRITE_BUFFER = 0x3b,
};
enum ssc_opcode {
SSC_ERASE_6 = 0x19,
SSC_FORMAT_MEDIUM = 0x04,
SSC_LOAD_UNLOAD = 0x1b,
SSC_LOCATE_10 = 0x2b,
SSC_LOCATE_16 = 0x92,
SSC_MOVE_MEDIUM_ATTACHED = 0xa7,
SSC_READ_6 = 0x08,
SSC_READ_BLOCK_LIMITS = 0x05,
SSC_READ_ELEMENT_STATUS_ATTACHED = 0xb4,
SSC_READ_POSITION = 0x34,
SSC_READ_REVERSE_6 = 0x0f,
SSC_RECOVER_BUFFERED_DATA = 0x14,
SSC_REPORT_DENSITY_SUPPORT = 0x44,
SSC_REWIND = 0x01,
SSC_SET_CAPACITY = 0x0b,
SSC_SPACE_6 = 0x11,
SSC_SPACE_16 = 0x91,
SSC_VERIFY_6 = 0x13,
SSC_WRITE_6 = 0x0a,
SSC_WRITE_FILEMARKS_6 = 0x10,
};
enum spc_vpd {
SPC_VPD_DEVICE_IDENTIFICATION = 0x83,
SPC_VPD_EXTENDED_INQUIRY_DATA = 0x86,
SPC_VPD_MANAGEMENT_NETWORK_ADDRESSES = 0x85,
SPC_VPD_MODE_PAGE_POLICY = 0x87,
SPC_VPD_SCSI_PORTS = 0x88,
SPC_VPD_SOFTWARE_INTERFACE_IDENTIFICATION = 0x84,
SPC_VPD_SUPPORTED_VPD_PAGES = 0x00,
SPC_VPD_UNIT_SERIAL_NUMBER = 0x80,
SPC_VPD_BLOCK_LIMITS = 0xb0,
SPC_VPD_BLOCK_DEV_CHARS = 0xb1,
SPC_VPD_BLOCK_THIN_PROVISION = 0xb2,
};
enum {
SPC_PERIPHERAL_DEVICE_TYPE_DISK = 0x00,
SPC_PERIPHERAL_DEVICE_TYPE_TAPE = 0x01,
SPC_PERIPHERAL_DEVICE_TYPE_DVD = 0x05,
SPC_PERIPHERAL_DEVICE_TYPE_CHANGER = 0x08,
SPC_VERSION_NONE = 0x00,
SPC_VERSION_SPC = 0x03,
SPC_VERSION_SPC2 = 0x04,
SPC_VERSION_SPC3 = 0x05,
SPC_VERSION_SPC4 = 0x06,
SPC_PROTOCOL_IDENTIFIER_FC = 0x00,
SPC_PROTOCOL_IDENTIFIER_PSCSI = 0x01,
SPC_PROTOCOL_IDENTIFIER_SSA = 0x02,
SPC_PROTOCOL_IDENTIFIER_IEEE1394 = 0x03,
SPC_PROTOCOL_IDENTIFIER_RDMA = 0x04,
SPC_PROTOCOL_IDENTIFIER_ISCSI = 0x05,
SPC_PROTOCOL_IDENTIFIER_SAS = 0x06,
SPC_PROTOCOL_IDENTIFIER_ADT = 0x07,
SPC_PROTOCOL_IDENTIFIER_ATA = 0x08,
SPC_VPD_CODE_SET_BINARY = 0x01,
SPC_VPD_CODE_SET_ASCII = 0x02,
SPC_VPD_CODE_SET_UTF8 = 0x03,
SPC_VPD_ASSOCIATION_LOGICAL_UNIT = 0x00,
SPC_VPD_ASSOCIATION_TARGET_PORT = 0x01,
SPC_VPD_ASSOCIATION_TARGET_DEVICE = 0x02,
SPC_VPD_IDENTIFIER_TYPE_VENDOR_SPECIFIC = 0x00,
SPC_VPD_IDENTIFIER_TYPE_T10_VENDOR_ID = 0x01,
SPC_VPD_IDENTIFIER_TYPE_EUI64 = 0x02,
SPC_VPD_IDENTIFIER_TYPE_NAA = 0x03,
SPC_VPD_IDENTIFIER_TYPE_RELATIVE_TARGET_PORT = 0x04,
SPC_VPD_IDENTIFIER_TYPE_TARGET_PORT_GROUP = 0x05,
SPC_VPD_IDENTIFIER_TYPE_LOGICAL_UNIT_GROUP = 0x06,
SPC_VPD_IDENTIFIER_TYPE_MD5_LOGICAL_UNIT = 0x07,
SPC_VPD_IDENTIFIER_TYPE_SCSI_NAME = 0x08,
};
struct scsi_cdb_inquiry {
uint8_t opcode;
uint8_t evpd;
uint8_t page_code;
uint16_t alloc_len;
uint8_t control;
};
struct scsi_cdb_inquiry_data {
uint8_t peripheral;
uint8_t rmb;
uint8_t version;
uint8_t response;
uint8_t add_len;
uint8_t flags;
uint8_t flags2;
uint8_t flags3;
uint8_t t10_vendor_id[8];
uint8_t product_id[16];
uint8_t product_rev[4];
uint8_t vendor[20];
uint8_t ius;
uint8_t reserved;
uint8_t desc[];
};
struct scsi_vpd_page {
uint8_t peripheral;
uint8_t page_code;
uint16_t alloc_len;
uint8_t params[];
};
#define SCSI_VEXT_REF_CHK 0x01
#define SCSI_VEXT_APP_CHK 0x02
#define SCSI_VEXT_GRD_CHK 0x04
#define SCSI_VEXT_SIMPSUP 0x01
#define SCSI_VEXT_ORDSUP 0x02
#define SCSI_VEXT_HEADSUP 0x04
#define SCSI_VEXT_PRIOR_SUP 0x08
#define SCSI_VEXT_GROUP_SUP 0x10
#define SCSI_VEXT_UASK_SUP 0x20
#define SCSI_VEXT_V_SUP 0x01
#define SCSI_VEXT_NV_SUP 0x02
#define SCSI_VEXT_CRD_SUP 0x04
#define SCSI_VEXT_WU_SUP 0x08
struct scsi_vpd_ext_inquiry {
uint8_t peripheral;
uint8_t page_code;
uint16_t alloc_len;
uint8_t check;
uint8_t sup;
uint8_t sup2;
uint8_t luiclr;
uint8_t cbcs;
uint8_t micro_dl;
uint8_t reserved[54];
};
#define SPC_VPD_DESIG_PIV 0x80
/* designation descriptor */
struct scsi_desig_desc {
uint8_t code_set : 4;
uint8_t protocol_id : 4;
uint8_t type : 4;
uint8_t association : 2;
uint8_t reserved0 : 1;
uint8_t piv : 1;
uint8_t reserved1;
uint8_t len;
uint8_t desig[];
};
/* mode page policy descriptor */
struct scsi_mpage_policy_desc {
uint8_t page_code;
uint8_t sub_page_code;
uint8_t policy;
uint8_t reserved;
};
/* target port descriptor */
struct scsi_tgt_port_desc {
uint8_t code_set;
uint8_t desig_type;
uint8_t reserved;
uint8_t len;
uint8_t designator[];
};
/* SCSI port designation descriptor */
struct scsi_port_desc {
uint16_t reserved;
uint16_t rel_port_id;
uint16_t reserved2;
uint16_t init_port_len;
uint16_t init_port_id;
uint16_t reserved3;
uint16_t tgt_desc_len;
uint8_t tgt_desc[];
};
/* SCSI UNMAP block descriptor */
struct scsi_unmap_bdesc {
/* UNMAP LOGICAL BLOCK ADDRESS */
uint64_t lba;
/* NUMBER OF LOGICAL BLOCKS */
uint32_t block_count;
/* RESERVED */
uint32_t reserved;
};
#define SCSI_UNMAP_LBPU (1 << 7)
#define SCSI_UNMAP_LBPWS (1 << 6)
#define SCSI_UNMAP_LBPWS10 (1 << 5)
#define SCSI_UNMAP_FULL_PROVISIONING 0x00
#define SCSI_UNMAP_RESOURCE_PROVISIONING 0x01
#define SCSI_UNMAP_THIN_PROVISIONING 0x02
#endif /* _SCSI_SPEC_H */

474
examples/vhost_scsi/vhost_scsi.c Normal file
View File

@ -0,0 +1,474 @@
/*-
* BSD LICENSE
*
* Copyright(c) 2010-2017 Intel Corporation. All rights reserved.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
* * Neither the name of Intel Corporation nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#include <stdint.h>
#include <unistd.h>
#include <stdbool.h>
#include <signal.h>
#include <assert.h>
#include <semaphore.h>
#include <linux/virtio_scsi.h>
#include <linux/virtio_ring.h>
#include <rte_atomic.h>
#include <rte_cycles.h>
#include <rte_log.h>
#include <rte_malloc.h>
#include <rte_vhost.h>
#include "vhost_scsi.h"
#include "scsi_spec.h"
#define VIRTIO_SCSI_FEATURES ((1 << VIRTIO_F_NOTIFY_ON_EMPTY) |\
(1 << VIRTIO_RING_F_EVENT_IDX) |\
(1 << VIRTIO_SCSI_F_INOUT) |\
(1 << VIRTIO_SCSI_F_CHANGE))
/* Path to folder where character device will be created. Can be set by user. */
static char dev_pathname[PATH_MAX] = "";
static struct vhost_scsi_ctrlr *g_vhost_ctrlr;
static int g_should_stop;
static sem_t exit_sem;
static struct vhost_scsi_ctrlr *
vhost_scsi_ctrlr_find(__rte_unused const char *ctrlr_name)
{
/* currently we only support 1 socket file fd */
return g_vhost_ctrlr;
}
static uint64_t gpa_to_vva(int vid, uint64_t gpa)
{
char path[PATH_MAX];
struct vhost_scsi_ctrlr *ctrlr;
int ret = 0;
ret = rte_vhost_get_ifname(vid, path, PATH_MAX);
if (ret) {
fprintf(stderr, "Cannot get socket name\n");
assert(ret != 0);
}
ctrlr = vhost_scsi_ctrlr_find(path);
if (!ctrlr) {
fprintf(stderr, "Controller is not ready\n");
assert(ctrlr != NULL);
}
assert(ctrlr->mem != NULL);
return rte_vhost_gpa_to_vva(ctrlr->mem, gpa);
}
static struct vring_desc *
descriptor_get_next(struct vring_desc *vq_desc, struct vring_desc *cur_desc)
{
return &vq_desc[cur_desc->next];
}
static bool
descriptor_has_next(struct vring_desc *cur_desc)
{
return !!(cur_desc->flags & VRING_DESC_F_NEXT);
}
static bool
descriptor_is_wr(struct vring_desc *cur_desc)
{
return !!(cur_desc->flags & VRING_DESC_F_WRITE);
}
static void
submit_completion(struct vhost_scsi_task *task)
{
struct rte_vhost_vring *vq;
struct vring_used *used;
vq = task->vq;
used = vq->used;
/* Fill out the next entry in the "used" ring. id = the
* index of the descriptor that contained the SCSI request.
* len = the total amount of data transferred for the SCSI
* request. We must report the correct len, for variable
* length SCSI CDBs, where we may return less data than
* allocated by the guest VM.
*/
used->ring[used->idx & (vq->size - 1)].id = task->req_idx;
used->ring[used->idx & (vq->size - 1)].len = task->data_len;
used->idx++;
/* Send an interrupt back to the guest VM so that it knows
* a completion is ready to be processed.
*/
eventfd_write(vq->callfd, (eventfd_t)1);
}
static void
vhost_process_read_payload_chain(struct vhost_scsi_task *task)
{
void *data;
task->iovs_cnt = 0;
task->resp = (void *)(uintptr_t)gpa_to_vva(task->bdev->vid,
task->desc->addr);
while (descriptor_has_next(task->desc)) {
task->desc = descriptor_get_next(task->vq->desc, task->desc);
data = (void *)(uintptr_t)gpa_to_vva(task->bdev->vid,
task->desc->addr);
task->iovs[task->iovs_cnt].iov_base = data;
task->iovs[task->iovs_cnt].iov_len = task->desc->len;
task->data_len += task->desc->len;
task->iovs_cnt++;
}
}
static void
vhost_process_write_payload_chain(struct vhost_scsi_task *task)
{
void *data;
task->iovs_cnt = 0;
do {
data = (void *)(uintptr_t)gpa_to_vva(task->bdev->vid,
task->desc->addr);
task->iovs[task->iovs_cnt].iov_base = data;
task->iovs[task->iovs_cnt].iov_len = task->desc->len;
task->data_len += task->desc->len;
task->iovs_cnt++;
task->desc = descriptor_get_next(task->vq->desc, task->desc);
} while (descriptor_has_next(task->desc));
task->resp = (void *)(uintptr_t)gpa_to_vva(task->bdev->vid,
task->desc->addr);
}
static struct vhost_block_dev *
vhost_scsi_bdev_construct(const char *bdev_name, const char *bdev_serial,
uint32_t blk_size, uint64_t blk_cnt,
bool wce_enable)
{
struct vhost_block_dev *bdev;
bdev = rte_zmalloc(NULL, sizeof(*bdev), RTE_CACHE_LINE_SIZE);
if (!bdev)
return NULL;
strncpy(bdev->name, bdev_name, sizeof(bdev->name));
strncpy(bdev->product_name, bdev_serial, sizeof(bdev->product_name));
bdev->blocklen = blk_size;
bdev->blockcnt = blk_cnt;
bdev->write_cache = wce_enable;
/* use memory as disk storage space */
bdev->data = rte_zmalloc(NULL, blk_cnt * blk_size, 0);
if (!bdev->data) {
fprintf(stderr, "no enough reseverd huge memory for disk\n");
return NULL;
}
return bdev;
}
static void
process_requestq(struct vhost_scsi_ctrlr *ctrlr, uint32_t q_idx)
{
int ret;
struct vhost_scsi_queue *scsi_vq;
struct rte_vhost_vring *vq;
scsi_vq = &ctrlr->bdev->queues[q_idx];
vq = &scsi_vq->vq;
ret = rte_vhost_get_vhost_vring(ctrlr->bdev->vid, q_idx, vq);
assert(ret == 0);
while (vq->avail->idx != scsi_vq->last_used_idx) {
int req_idx;
uint16_t last_idx;
struct vhost_scsi_task *task;
last_idx = scsi_vq->last_used_idx & (vq->size - 1);
req_idx = vq->avail->ring[last_idx];
task = rte_zmalloc(NULL, sizeof(*task), 0);
assert(task != NULL);
task->ctrlr = ctrlr;
task->bdev = ctrlr->bdev;
task->vq = vq;
task->req_idx = req_idx;
task->desc = &task->vq->desc[task->req_idx];
/* does not support indirect descriptors */
assert((task->desc->flags & VRING_DESC_F_INDIRECT) == 0);
scsi_vq->last_used_idx++;
task->req = (void *)(uintptr_t)gpa_to_vva(task->bdev->vid,
task->desc->addr);
task->desc = descriptor_get_next(task->vq->desc, task->desc);
if (!descriptor_has_next(task->desc)) {
task->dxfer_dir = SCSI_DIR_NONE;
task->resp = (void *)(uintptr_t)
gpa_to_vva(task->bdev->vid,
task->desc->addr);
} else if (!descriptor_is_wr(task->desc)) {
task->dxfer_dir = SCSI_DIR_TO_DEV;
vhost_process_write_payload_chain(task);
} else {
task->dxfer_dir = SCSI_DIR_FROM_DEV;
vhost_process_read_payload_chain(task);
}
ret = vhost_bdev_process_scsi_commands(ctrlr->bdev, task);
if (ret) {
/* invalid response */
task->resp->response = VIRTIO_SCSI_S_BAD_TARGET;
} else {
/* successfully */
task->resp->response = VIRTIO_SCSI_S_OK;
task->resp->status = 0;
task->resp->resid = 0;
}
submit_completion(task);
rte_free(task);
}
}
/* Main framework for processing IOs */
static void *
ctrlr_worker(void *arg)
{
uint32_t idx, num;
struct vhost_scsi_ctrlr *ctrlr = (struct vhost_scsi_ctrlr *)arg;
cpu_set_t cpuset;
pthread_t thread;
thread = pthread_self();
CPU_ZERO(&cpuset);
CPU_SET(0, &cpuset);
pthread_setaffinity_np(thread, sizeof(cpu_set_t), &cpuset);
num = rte_vhost_get_vring_num(ctrlr->bdev->vid);
fprintf(stdout, "Ctrlr Worker Thread Started with %u Vring\n", num);
if (num != NUM_OF_SCSI_QUEUES) {
fprintf(stderr, "Only 1 IO queue are supported\n");
exit(0);
}
while (!g_should_stop && ctrlr->bdev != NULL) {
/* At least 3 vrings, currently only can support 1 IO queue
* Queue 2 for IO queue, does not support TMF and hotplug
* for the example application now
*/
for (idx = 2; idx < num; idx++)
process_requestq(ctrlr, idx);
}
fprintf(stdout, "Ctrlr Worker Thread Exiting\n");
sem_post(&exit_sem);
return NULL;
}
static int
new_device(int vid)
{
char path[PATH_MAX];
struct vhost_scsi_ctrlr *ctrlr;
struct vhost_scsi_queue *scsi_vq;
struct rte_vhost_vring *vq;
pthread_t tid;
int i, ret;
ret = rte_vhost_get_ifname(vid, path, PATH_MAX);
if (ret) {
fprintf(stderr, "Cannot get socket name\n");
return -1;
}
ctrlr = vhost_scsi_ctrlr_find(path);
if (!ctrlr) {
fprintf(stderr, "Controller is not ready\n");
return -1;
}
ret = rte_vhost_get_mem_table(vid, &ctrlr->mem);
if (ret) {
fprintf(stderr, "Get Controller memory region failed\n");
return -1;
}
assert(ctrlr->mem != NULL);
/* hardcoded block device information with 128MiB */
ctrlr->bdev = vhost_scsi_bdev_construct("malloc0", "vhost_scsi_malloc0",
4096, 32768, 0);
if (!ctrlr->bdev)
return -1;
ctrlr->bdev->vid = vid;
/* Disable Notifications */
for (i = 0; i < NUM_OF_SCSI_QUEUES; i++) {
rte_vhost_enable_guest_notification(vid, i, 0);
/* restore used index */
scsi_vq = &ctrlr->bdev->queues[i];
vq = &scsi_vq->vq;
ret = rte_vhost_get_vhost_vring(ctrlr->bdev->vid, i, vq);
assert(ret == 0);
scsi_vq->last_used_idx = vq->used->idx;
scsi_vq->last_avail_idx = vq->used->idx;
}
g_should_stop = 0;
fprintf(stdout, "New Device %s, Device ID %d\n", path, vid);
if (pthread_create(&tid, NULL, &ctrlr_worker, ctrlr) < 0) {
fprintf(stderr, "Worker Thread Started Failed\n");
return -1;
}
pthread_detach(tid);
return 0;
}
static void
destroy_device(int vid)
{
char path[PATH_MAX];
struct vhost_scsi_ctrlr *ctrlr;
rte_vhost_get_ifname(vid, path, PATH_MAX);
fprintf(stdout, "Destroy %s Device ID %d\n", path, vid);
ctrlr = vhost_scsi_ctrlr_find(path);
if (!ctrlr) {
fprintf(stderr, "Destroy Ctrlr Failed\n");
return;
}
ctrlr->bdev = NULL;
g_should_stop = 1;
sem_wait(&exit_sem);
}
static const struct vhost_device_ops vhost_scsi_device_ops = {
.new_device = new_device,
.destroy_device = destroy_device,
};
static struct vhost_scsi_ctrlr *
vhost_scsi_ctrlr_construct(const char *ctrlr_name)
{
int ret;
struct vhost_scsi_ctrlr *ctrlr;
char *path;
char cwd[PATH_MAX];
/* always use current directory */
path = getcwd(cwd, PATH_MAX);
if (!path) {
fprintf(stderr, "Cannot get current working directory\n");
return NULL;
}
snprintf(dev_pathname, sizeof(dev_pathname), "%s/%s", path, ctrlr_name);
if (access(dev_pathname, F_OK) != -1) {
if (unlink(dev_pathname) != 0)
rte_exit(EXIT_FAILURE, "Cannot remove %s.\n",
dev_pathname);
}
if (rte_vhost_driver_register(dev_pathname, 0) != 0) {
fprintf(stderr, "socket %s already exists\n", dev_pathname);
return NULL;
}
fprintf(stdout, "socket file: %s created\n", dev_pathname);
ret = rte_vhost_driver_set_features(dev_pathname, VIRTIO_SCSI_FEATURES);
if (ret != 0) {
fprintf(stderr, "Set vhost driver features failed\n");
return NULL;
}
ctrlr = rte_zmalloc(NULL, sizeof(*ctrlr), RTE_CACHE_LINE_SIZE);
if (!ctrlr)
return NULL;
rte_vhost_driver_callback_register(dev_pathname,
&vhost_scsi_device_ops);
return ctrlr;
}
static void
signal_handler(__rte_unused int signum)
{
if (access(dev_pathname, F_OK) == 0)
unlink(dev_pathname);
exit(0);
}
int main(int argc, char *argv[])
{
int ret;
signal(SIGINT, signal_handler);
/* init EAL */
ret = rte_eal_init(argc, argv);
if (ret < 0)
rte_exit(EXIT_FAILURE, "Error with EAL initialization\n");
g_vhost_ctrlr = vhost_scsi_ctrlr_construct("vhost.socket");
if (g_vhost_ctrlr == NULL) {
fprintf(stderr, "Construct vhost scsi controller failed\n");
return 0;
}
if (sem_init(&exit_sem, 0, 0) < 0) {
fprintf(stderr, "Error init exit_sem\n");
return -1;
}
rte_vhost_driver_start(dev_pathname);
/* loop for exit the application */
while (1)
sleep(1);
return 0;
}

108
examples/vhost_scsi/vhost_scsi.h Normal file
View File

@ -0,0 +1,108 @@
/*-
* BSD LICENSE
*
* Copyright(c) 2010-2017 Intel Corporation. All rights reserved.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in
* the documentation and/or other materials provided with the
* distribution.
* * Neither the name of Intel Corporation nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef _VHOST_SCSI_H_
#define _VHOST_SCSI_H_
#include <sys/uio.h>
#include <stdint.h>
#include <linux/virtio_scsi.h>
#include <linux/virtio_ring.h>
#include <rte_vhost.h>
struct vhost_scsi_queue {
struct rte_vhost_vring vq;
uint16_t last_avail_idx;
uint16_t last_used_idx;
};
#define NUM_OF_SCSI_QUEUES 3
struct vhost_block_dev {
/** ID for vhost library. */
int vid;
/** Queues for the block device */
struct vhost_scsi_queue queues[NUM_OF_SCSI_QUEUES];
/** Unique name for this block device. */
char name[64];
/** Unique product name for this kind of block device. */
char product_name[256];
/** Size in bytes of a logical block for the backend */
uint32_t blocklen;
/** Number of blocks */
uint64_t blockcnt;
/** write cache enabled, not used at the moment */
int write_cache;
/** use memory as disk storage space */
uint8_t *data;
};
struct vhost_scsi_ctrlr {
/** Only support 1 LUN for the example */
struct vhost_block_dev *bdev;
/** VM memory region */
struct rte_vhost_memory *mem;
} __rte_cache_aligned;
#define VHOST_SCSI_MAX_IOVS 128
enum scsi_data_dir {
SCSI_DIR_NONE = 0,
SCSI_DIR_TO_DEV = 1,
SCSI_DIR_FROM_DEV = 2,
};
struct vhost_scsi_task {
int req_idx;
uint32_t dxfer_dir;
uint32_t data_len;
struct virtio_scsi_cmd_req *req;
struct virtio_scsi_cmd_resp *resp;
struct iovec iovs[VHOST_SCSI_MAX_IOVS];
uint32_t iovs_cnt;
struct vring_desc *desc;
struct rte_vhost_vring *vq;
struct vhost_block_dev *bdev;
struct vhost_scsi_ctrlr *ctrlr;
};
int vhost_bdev_process_scsi_commands(struct vhost_block_dev *bdev,
struct vhost_scsi_task *task);
#endif /* _VHOST_SCSI_H_ */