* [RFC PATCH v01] Xen PVSCSI drivers for pvops xen/stable-2.6.32.x kernel
@ 2011-01-06 22:31 Pasi Kärkkäinen
2011-01-07 10:10 ` Ian Campbell
2011-11-30 16:23 ` Konrad Rzeszutek Wilk
0 siblings, 2 replies; 10+ messages in thread
From: Pasi Kärkkäinen @ 2011-01-06 22:31 UTC (permalink / raw)
To: xen-devel; +Cc: Jeremy Fitzhardinge, Ian Campbell
[-- Attachment #1: Type: text/plain, Size: 708 bytes --]
Hello,
http://pasik.reaktio.net/xen/patches/xen-pvscsi-drivers-linux-2.6.32.27-pvops-v01.diff
This is the first version of Xen PVSCSI drivers, both the scsiback backend and
scsifront frontend, ported from Novell SLES11SP1 2.6.32 Xenlinux kernel to
pvops xen/stable-2.6.32.x branch.
At the moment it's *only* compile-tested with the latest xen/stable-2.6.32.x
git kernel as of today (2.6.32.27), on Fedora 14 x86_64.
Comments welcome.
I'm sure there are still things to fix in it.
Hopefully I managed to properly fix all the differences between Xenlinux <-> pvops..
Let me know how it goes, if you feel adventurous enough to try it :)
More info: http://wiki.xensource.com/xenwiki/XenPVSCSI
-- Pasi
[-- Attachment #2: xen-pvscsi-drivers-linux-2.6.32.27-pvops-v01.diff --]
[-- Type: text/x-diff, Size: 97110 bytes --]
diff -N -u -r linux-2.6-xen-orig/drivers/scsi/Kconfig linux-2.6-xen-pvscsi/drivers/scsi/Kconfig
--- linux-2.6-xen-orig/drivers/scsi/Kconfig 2011-01-06 17:23:22.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/scsi/Kconfig 2011-01-06 21:41:30.000000000 +0200
@@ -1838,6 +1838,14 @@
To compile this driver as a module, choose M here. The module will
be called bfa.
+config XEN_SCSI_FRONTEND
+ tristate "Xen PVSCSI frontend driver"
+ depends on XEN && SCSI
+ default m
+ help
+ The SCSI frontend driver allows the kernel to access SCSI Devices
+ within another guest OS.
+
endif # SCSI_LOWLEVEL
source "drivers/scsi/pcmcia/Kconfig"
diff -N -u -r linux-2.6-xen-orig/drivers/scsi/Makefile linux-2.6-xen-pvscsi/drivers/scsi/Makefile
--- linux-2.6-xen-orig/drivers/scsi/Makefile 2011-01-06 17:23:22.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/scsi/Makefile 2011-01-06 21:39:50.000000000 +0200
@@ -133,6 +133,7 @@
obj-$(CONFIG_SCSI_BNX2_ISCSI) += libiscsi.o bnx2i/
obj-$(CONFIG_BE2ISCSI) += libiscsi.o be2iscsi/
obj-$(CONFIG_SCSI_PMCRAID) += pmcraid.o
+obj-$(CONFIG_XEN_SCSI_FRONTEND) += xen-scsifront/
obj-$(CONFIG_ARM) += arm/
diff -N -u -r linux-2.6-xen-orig/drivers/scsi/xen-scsifront/common.h linux-2.6-xen-pvscsi/drivers/scsi/xen-scsifront/common.h
--- linux-2.6-xen-orig/drivers/scsi/xen-scsifront/common.h 1970-01-01 02:00:00.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/scsi/xen-scsifront/common.h 2011-01-06 21:22:56.000000000 +0200
@@ -0,0 +1,137 @@
+/*
+ * Xen SCSI frontend driver
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef __XEN_DRIVERS_SCSIFRONT_H__
+#define __XEN_DRIVERS_SCSIFRONT_H__
+
+#include <linux/version.h>
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/device.h>
+#include <linux/kthread.h>
+#include <linux/wait.h>
+#include <linux/interrupt.h>
+#include <linux/spinlock.h>
+#include <linux/sched.h>
+#include <linux/blkdev.h>
+#include <scsi/scsi_cmnd.h>
+#include <scsi/scsi_device.h>
+#include <scsi/scsi.h>
+#include <scsi/scsi_host.h>
+#include <asm/xen/page.h>
+#include <xen/xenbus.h>
+#include <xen/grant_table.h>
+#include <xen/events.h>
+#include <xen/evtchn.h>
+#include <xen/interface/xen.h>
+#include <xen/interface/io/ring.h>
+#include <xen/interface/io/vscsiif.h>
+#include <xen/interface/grant_table.h>
+#include <xen/interface/io/protocols.h>
+#include <asm/delay.h>
+#include <asm/hypervisor.h>
+/*#include <asm/maddr.h>*/
+
+#ifdef HAVE_XEN_PLATFORM_COMPAT_H
+#include <xen/platform-compat.h>
+#endif
+
+#define GRANT_INVALID_REF 0
+#define VSCSI_IN_ABORT 1
+#define VSCSI_IN_RESET 2
+
+/* tuning point*/
+#define VSCSIIF_DEFAULT_CMD_PER_LUN 10
+#define VSCSIIF_MAX_TARGET 64
+#define VSCSIIF_MAX_LUN 255
+
+#define VSCSIIF_RING_SIZE __CONST_RING_SIZE(vscsiif, PAGE_SIZE)
+#define VSCSIIF_MAX_REQS VSCSIIF_RING_SIZE
+
+struct vscsifrnt_shadow {
+ uint16_t next_free;
+
+ /* command between backend and frontend
+ * VSCSIIF_ACT_SCSI_CDB or VSCSIIF_ACT_SCSI_RESET */
+ unsigned char act;
+
+ /* do reset function */
+ wait_queue_head_t wq_reset; /* reset work queue */
+ int wait_reset; /* reset work queue condition */
+ int32_t rslt_reset; /* reset response status */
+ /* (SUCESS or FAILED) */
+
+ /* for DMA_TO_DEVICE(1), DMA_FROM_DEVICE(2), DMA_NONE(3)
+ requests */
+ unsigned int sc_data_direction;
+
+ /* Number of pieces of scatter-gather */
+ unsigned int nr_segments;
+
+ /* requested struct scsi_cmnd is stored from kernel */
+ unsigned long req_scsi_cmnd;
+ int gref[VSCSIIF_SG_TABLESIZE];
+};
+
+struct vscsifrnt_info {
+ struct xenbus_device *dev;
+
+ struct Scsi_Host *host;
+
+ spinlock_t io_lock;
+ spinlock_t shadow_lock;
+ unsigned int evtchn;
+ unsigned int irq;
+
+ grant_ref_t ring_ref;
+ struct vscsiif_front_ring ring;
+ struct vscsiif_response ring_res;
+
+ struct vscsifrnt_shadow shadow[VSCSIIF_MAX_REQS];
+ uint32_t shadow_free;
+
+ struct task_struct *kthread;
+ wait_queue_head_t wq;
+ unsigned int waiting_resp;
+
+};
+
+#define DPRINTK(_f, _a...) \
+ pr_debug("(file=%s, line=%d) " _f, \
+ __FILE__ , __LINE__ , ## _a )
+
+int scsifront_xenbus_init(void);
+void scsifront_xenbus_unregister(void);
+int scsifront_schedule(void *data);
+irqreturn_t scsifront_intr(int irq, void *dev_id);
+int scsifront_cmd_done(struct vscsifrnt_info *info);
+
+
+#endif /* __XEN_DRIVERS_SCSIFRONT_H__ */
diff -N -u -r linux-2.6-xen-orig/drivers/scsi/xen-scsifront/Makefile linux-2.6-xen-pvscsi/drivers/scsi/xen-scsifront/Makefile
--- linux-2.6-xen-orig/drivers/scsi/xen-scsifront/Makefile 1970-01-01 02:00:00.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/scsi/xen-scsifront/Makefile 2011-01-06 20:20:58.000000000 +0200
@@ -0,0 +1,4 @@
+
+obj-$(CONFIG_XEN_SCSI_FRONTEND) := xen-scsifront.o
+xen-scsifront-objs := scsifront.o xenbus.o
+
diff -N -u -r linux-2.6-xen-orig/drivers/scsi/xen-scsifront/scsifront.c linux-2.6-xen-pvscsi/drivers/scsi/xen-scsifront/scsifront.c
--- linux-2.6-xen-orig/drivers/scsi/xen-scsifront/scsifront.c 1970-01-01 02:00:00.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/scsi/xen-scsifront/scsifront.c 2011-01-06 20:47:56.000000000 +0200
@@ -0,0 +1,478 @@
+/*
+ * Xen SCSI frontend driver
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+
+#include <linux/version.h>
+#include "common.h"
+
+static int get_id_from_freelist(struct vscsifrnt_info *info)
+{
+ unsigned long flags;
+ uint32_t free;
+
+ spin_lock_irqsave(&info->shadow_lock, flags);
+
+ free = info->shadow_free;
+ BUG_ON(free > VSCSIIF_MAX_REQS);
+ info->shadow_free = info->shadow[free].next_free;
+ info->shadow[free].next_free = 0x0fff;
+
+ info->shadow[free].wait_reset = 0;
+
+ spin_unlock_irqrestore(&info->shadow_lock, flags);
+
+ return free;
+}
+
+static void add_id_to_freelist(struct vscsifrnt_info *info, uint32_t id)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->shadow_lock, flags);
+
+ info->shadow[id].next_free = info->shadow_free;
+ info->shadow[id].req_scsi_cmnd = 0;
+ info->shadow_free = id;
+
+ spin_unlock_irqrestore(&info->shadow_lock, flags);
+}
+
+
+struct vscsiif_request * scsifront_pre_request(struct vscsifrnt_info *info)
+{
+ struct vscsiif_front_ring *ring = &(info->ring);
+ vscsiif_request_t *ring_req;
+ uint32_t id;
+
+ ring_req = RING_GET_REQUEST(&(info->ring), ring->req_prod_pvt);
+
+ ring->req_prod_pvt++;
+
+ id = get_id_from_freelist(info); /* use id by response */
+ ring_req->rqid = (uint16_t)id;
+
+ return ring_req;
+}
+
+
+static void scsifront_notify_work(struct vscsifrnt_info *info)
+{
+ info->waiting_resp = 1;
+ wake_up(&info->wq);
+}
+
+
+static void scsifront_do_request(struct vscsifrnt_info *info)
+{
+ struct vscsiif_front_ring *ring = &(info->ring);
+ unsigned int irq = info->irq;
+ int notify;
+
+ RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(ring, notify);
+ if (notify)
+ notify_remote_via_irq(irq);
+}
+
+irqreturn_t scsifront_intr(int irq, void *dev_id)
+{
+ scsifront_notify_work((struct vscsifrnt_info *)dev_id);
+ return IRQ_HANDLED;
+}
+
+
+static void scsifront_gnttab_done(struct vscsifrnt_shadow *s, uint32_t id)
+{
+ int i;
+
+ if (s->sc_data_direction == DMA_NONE)
+ return;
+
+ if (s->nr_segments) {
+ for (i = 0; i < s->nr_segments; i++) {
+ if (unlikely(gnttab_query_foreign_access(
+ s->gref[i]) != 0)) {
+ printk(KERN_ALERT "scsifront: "
+ "grant still in use by backend.\n");
+ BUG();
+ }
+ gnttab_end_foreign_access(s->gref[i], 0, 0UL);
+ }
+ }
+
+ return;
+}
+
+
+static void scsifront_cdb_cmd_done(struct vscsifrnt_info *info,
+ vscsiif_response_t *ring_res)
+{
+ struct scsi_cmnd *sc;
+ uint32_t id;
+ uint8_t sense_len;
+
+ id = ring_res->rqid;
+ sc = (struct scsi_cmnd *)info->shadow[id].req_scsi_cmnd;
+
+ if (sc == NULL)
+ BUG();
+
+ scsifront_gnttab_done(&info->shadow[id], id);
+ add_id_to_freelist(info, id);
+
+ sc->result = ring_res->rslt;
+ scsi_set_resid(sc, ring_res->residual_len);
+
+ if (ring_res->sense_len > VSCSIIF_SENSE_BUFFERSIZE)
+ sense_len = VSCSIIF_SENSE_BUFFERSIZE;
+ else
+ sense_len = ring_res->sense_len;
+
+ if (sense_len)
+ memcpy(sc->sense_buffer, ring_res->sense_buffer, sense_len);
+
+ sc->scsi_done(sc);
+
+ return;
+}
+
+
+static void scsifront_sync_cmd_done(struct vscsifrnt_info *info,
+ vscsiif_response_t *ring_res)
+{
+ uint16_t id = ring_res->rqid;
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->shadow_lock, flags);
+ info->shadow[id].wait_reset = 1;
+ info->shadow[id].rslt_reset = ring_res->rslt;
+ spin_unlock_irqrestore(&info->shadow_lock, flags);
+
+ wake_up(&(info->shadow[id].wq_reset));
+}
+
+
+int scsifront_cmd_done(struct vscsifrnt_info *info)
+{
+ vscsiif_response_t *ring_res;
+
+ RING_IDX i, rp;
+ int more_to_do = 0;
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->io_lock, flags);
+
+ rp = info->ring.sring->rsp_prod;
+ rmb();
+ for (i = info->ring.rsp_cons; i != rp; i++) {
+
+ ring_res = RING_GET_RESPONSE(&info->ring, i);
+
+ if (info->shadow[ring_res->rqid].act == VSCSIIF_ACT_SCSI_CDB)
+ scsifront_cdb_cmd_done(info, ring_res);
+ else
+ scsifront_sync_cmd_done(info, ring_res);
+ }
+
+ info->ring.rsp_cons = i;
+
+ if (i != info->ring.req_prod_pvt) {
+ RING_FINAL_CHECK_FOR_RESPONSES(&info->ring, more_to_do);
+ } else {
+ info->ring.sring->rsp_event = i + 1;
+ }
+
+ spin_unlock_irqrestore(&info->io_lock, flags);
+
+
+ /* Yield point for this unbounded loop. */
+ cond_resched();
+
+ return more_to_do;
+}
+
+
+
+
+int scsifront_schedule(void *data)
+{
+ struct vscsifrnt_info *info = (struct vscsifrnt_info *)data;
+
+ while (!kthread_should_stop()) {
+ wait_event_interruptible(
+ info->wq,
+ info->waiting_resp || kthread_should_stop());
+
+ info->waiting_resp = 0;
+ smp_mb();
+
+ if (scsifront_cmd_done(info))
+ info->waiting_resp = 1;
+ }
+
+ return 0;
+}
+
+
+
+static int map_data_for_request(struct vscsifrnt_info *info,
+ struct scsi_cmnd *sc, vscsiif_request_t *ring_req, uint32_t id)
+{
+ grant_ref_t gref_head;
+ struct page *page;
+ int err, ref, ref_cnt = 0;
+ int write = (sc->sc_data_direction == DMA_TO_DEVICE);
+ unsigned int i, nr_pages, off, len, bytes;
+ unsigned long buffer_pfn;
+
+ if (sc->sc_data_direction == DMA_NONE)
+ return 0;
+
+ err = gnttab_alloc_grant_references(VSCSIIF_SG_TABLESIZE, &gref_head);
+ if (err) {
+ printk(KERN_ERR "scsifront: gnttab_alloc_grant_references() error\n");
+ return -ENOMEM;
+ }
+
+ if (scsi_bufflen(sc)) {
+ /* quoted scsi_lib.c/scsi_req_map_sg . */
+ struct scatterlist *sg, *sgl = scsi_sglist(sc);
+ unsigned int data_len = scsi_bufflen(sc);
+
+ nr_pages = (data_len + sgl->offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
+ if (nr_pages > VSCSIIF_SG_TABLESIZE) {
+ printk(KERN_ERR "scsifront: Unable to map request_buffer for command!\n");
+ ref_cnt = (-E2BIG);
+ goto big_to_sg;
+ }
+
+ for_each_sg (sgl, sg, scsi_sg_count(sc), i) {
+ page = sg_page(sg);
+ off = sg->offset;
+ len = sg->length;
+
+ buffer_pfn = page_to_phys(page) >> PAGE_SHIFT;
+
+ while (len > 0 && data_len > 0) {
+ /*
+ * sg sends a scatterlist that is larger than
+ * the data_len it wants transferred for certain
+ * IO sizes
+ */
+ bytes = min_t(unsigned int, len, PAGE_SIZE - off);
+ bytes = min(bytes, data_len);
+
+ ref = gnttab_claim_grant_reference(&gref_head);
+ BUG_ON(ref == -ENOSPC);
+
+ gnttab_grant_foreign_access_ref(ref, info->dev->otherend_id,
+ buffer_pfn, write);
+
+ info->shadow[id].gref[ref_cnt] = ref;
+ ring_req->seg[ref_cnt].gref = ref;
+ ring_req->seg[ref_cnt].offset = (uint16_t)off;
+ ring_req->seg[ref_cnt].length = (uint16_t)bytes;
+
+ buffer_pfn++;
+ len -= bytes;
+ data_len -= bytes;
+ off = 0;
+ ref_cnt++;
+ }
+ }
+ }
+
+big_to_sg:
+
+ gnttab_free_grant_references(gref_head);
+
+ return ref_cnt;
+}
+
+static int scsifront_queuecommand(struct scsi_cmnd *sc,
+ void (*done)(struct scsi_cmnd *))
+{
+ struct vscsifrnt_info *info =
+ (struct vscsifrnt_info *) sc->device->host->hostdata;
+ vscsiif_request_t *ring_req;
+ int ref_cnt;
+ uint16_t rqid;
+
+ if (RING_FULL(&info->ring)) {
+ goto out_host_busy;
+ }
+
+ sc->scsi_done = done;
+ sc->result = 0;
+
+ ring_req = scsifront_pre_request(info);
+ rqid = ring_req->rqid;
+ ring_req->act = VSCSIIF_ACT_SCSI_CDB;
+
+ ring_req->id = sc->device->id;
+ ring_req->lun = sc->device->lun;
+ ring_req->channel = sc->device->channel;
+ ring_req->cmd_len = sc->cmd_len;
+
+ BUG_ON(sc->cmd_len > VSCSIIF_MAX_COMMAND_SIZE);
+
+ if ( sc->cmd_len )
+ memcpy(ring_req->cmnd, sc->cmnd, sc->cmd_len);
+ else
+ memset(ring_req->cmnd, 0, VSCSIIF_MAX_COMMAND_SIZE);
+
+ ring_req->sc_data_direction = (uint8_t)sc->sc_data_direction;
+ ring_req->timeout_per_command = (sc->request->timeout / HZ);
+
+ info->shadow[rqid].req_scsi_cmnd = (unsigned long)sc;
+ info->shadow[rqid].sc_data_direction = sc->sc_data_direction;
+ info->shadow[rqid].act = ring_req->act;
+
+ ref_cnt = map_data_for_request(info, sc, ring_req, rqid);
+ if (ref_cnt < 0) {
+ add_id_to_freelist(info, rqid);
+ if (ref_cnt == (-ENOMEM))
+ goto out_host_busy;
+ else {
+ sc->result = (DID_ERROR << 16);
+ goto out_fail_command;
+ }
+ }
+
+ ring_req->nr_segments = (uint8_t)ref_cnt;
+ info->shadow[rqid].nr_segments = ref_cnt;
+
+ scsifront_do_request(info);
+
+ return 0;
+
+out_host_busy:
+ return SCSI_MLQUEUE_HOST_BUSY;
+
+out_fail_command:
+ done(sc);
+ return 0;
+}
+
+
+static int scsifront_eh_abort_handler(struct scsi_cmnd *sc)
+{
+ return (FAILED);
+}
+
+/* vscsi supports only device_reset, because it is each of LUNs */
+static int scsifront_dev_reset_handler(struct scsi_cmnd *sc)
+{
+ struct Scsi_Host *host = sc->device->host;
+ struct vscsifrnt_info *info =
+ (struct vscsifrnt_info *) sc->device->host->hostdata;
+
+ vscsiif_request_t *ring_req;
+ uint16_t rqid;
+ int err;
+
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,12)
+ spin_lock_irq(host->host_lock);
+#endif
+
+ ring_req = scsifront_pre_request(info);
+ ring_req->act = VSCSIIF_ACT_SCSI_RESET;
+
+ rqid = ring_req->rqid;
+ info->shadow[rqid].act = VSCSIIF_ACT_SCSI_RESET;
+
+ ring_req->channel = sc->device->channel;
+ ring_req->id = sc->device->id;
+ ring_req->lun = sc->device->lun;
+ ring_req->cmd_len = sc->cmd_len;
+
+ if ( sc->cmd_len )
+ memcpy(ring_req->cmnd, sc->cmnd, sc->cmd_len);
+ else
+ memset(ring_req->cmnd, 0, VSCSIIF_MAX_COMMAND_SIZE);
+
+ ring_req->sc_data_direction = (uint8_t)sc->sc_data_direction;
+ ring_req->timeout_per_command = (sc->request->timeout / HZ);
+ ring_req->nr_segments = 0;
+
+ scsifront_do_request(info);
+
+ spin_unlock_irq(host->host_lock);
+ wait_event_interruptible(info->shadow[rqid].wq_reset,
+ info->shadow[rqid].wait_reset);
+ spin_lock_irq(host->host_lock);
+
+ err = info->shadow[rqid].rslt_reset;
+
+ add_id_to_freelist(info, rqid);
+
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,12)
+ spin_unlock_irq(host->host_lock);
+#endif
+ return (err);
+}
+
+
+struct scsi_host_template scsifront_sht = {
+ .module = THIS_MODULE,
+ .name = "Xen SCSI frontend driver",
+ .queuecommand = scsifront_queuecommand,
+ .eh_abort_handler = scsifront_eh_abort_handler,
+ .eh_device_reset_handler= scsifront_dev_reset_handler,
+ .cmd_per_lun = VSCSIIF_DEFAULT_CMD_PER_LUN,
+ .can_queue = VSCSIIF_MAX_REQS,
+ .this_id = -1,
+ .sg_tablesize = VSCSIIF_SG_TABLESIZE,
+ .use_clustering = DISABLE_CLUSTERING,
+ .proc_name = "scsifront",
+};
+
+
+static int __init scsifront_init(void)
+{
+ int err;
+
+ if (!xen_domain())
+ return -ENODEV;
+
+ err = scsifront_xenbus_init();
+
+ return err;
+}
+
+static void __exit scsifront_exit(void)
+{
+ scsifront_xenbus_unregister();
+}
+
+module_init(scsifront_init);
+module_exit(scsifront_exit);
+
+MODULE_DESCRIPTION("Xen SCSI frontend driver");
+MODULE_LICENSE("GPL");
diff -N -u -r linux-2.6-xen-orig/drivers/scsi/xen-scsifront/xenbus.c linux-2.6-xen-pvscsi/drivers/scsi/xen-scsifront/xenbus.c
--- linux-2.6-xen-orig/drivers/scsi/xen-scsifront/xenbus.c 1970-01-01 02:00:00.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/scsi/xen-scsifront/xenbus.c 2011-01-06 21:27:02.000000000 +0200
@@ -0,0 +1,420 @@
+/*
+ * Xen SCSI frontend driver
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+
+#include <linux/version.h>
+#include "common.h"
+
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,11)
+ #define DEFAULT_TASK_COMM_LEN 16
+#else
+ #define DEFAULT_TASK_COMM_LEN TASK_COMM_LEN
+#endif
+
+extern struct scsi_host_template scsifront_sht;
+
+static void scsifront_free(struct vscsifrnt_info *info)
+{
+ struct Scsi_Host *host = info->host;
+
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,14)
+ if (host->shost_state != SHOST_DEL) {
+#else
+ if (!test_bit(SHOST_DEL, &host->shost_state)) {
+#endif
+ scsi_remove_host(info->host);
+ }
+
+ if (info->ring_ref != GRANT_INVALID_REF) {
+ gnttab_end_foreign_access(info->ring_ref,
+ 0, (unsigned long)info->ring.sring);
+ info->ring_ref = GRANT_INVALID_REF;
+ info->ring.sring = NULL;
+ }
+
+ if (info->irq)
+ unbind_from_irqhandler(info->irq, info);
+ info->irq = 0;
+
+ scsi_host_put(info->host);
+}
+
+
+static int scsifront_alloc_ring(struct vscsifrnt_info *info)
+{
+ struct xenbus_device *dev = info->dev;
+ struct vscsiif_sring *sring;
+ int err = -ENOMEM;
+
+
+ info->ring_ref = GRANT_INVALID_REF;
+
+ /***** Frontend to Backend ring start *****/
+ sring = (struct vscsiif_sring *) __get_free_page(GFP_KERNEL);
+ if (!sring) {
+ xenbus_dev_fatal(dev, err, "fail to allocate shared ring (Front to Back)");
+ return err;
+ }
+ SHARED_RING_INIT(sring);
+ FRONT_RING_INIT(&info->ring, sring, PAGE_SIZE);
+
+ err = xenbus_grant_ring(dev, virt_to_mfn(sring));
+ if (err < 0) {
+ free_page((unsigned long) sring);
+ info->ring.sring = NULL;
+ xenbus_dev_fatal(dev, err, "fail to grant shared ring (Front to Back)");
+ goto free_sring;
+ }
+ info->ring_ref = err;
+
+ err = bind_evtchn_to_irqhandler(
+ dev->otherend_id, scsifront_intr,
+ IRQF_SAMPLE_RANDOM, "scsifront", info);
+
+ if (err <= 0) {
+ xenbus_dev_fatal(dev, err, "bind_evtchn_to_irqhandler");
+ goto free_sring;
+ }
+ info->irq = err;
+
+ return 0;
+
+/* free resource */
+free_sring:
+ scsifront_free(info);
+
+ return err;
+}
+
+
+static int scsifront_init_ring(struct vscsifrnt_info *info)
+{
+ struct xenbus_device *dev = info->dev;
+ struct xenbus_transaction xbt;
+ int err;
+
+ DPRINTK("%s\n",__FUNCTION__);
+
+ err = scsifront_alloc_ring(info);
+ if (err)
+ return err;
+ DPRINTK("%u %u\n", info->ring_ref, info->evtchn);
+
+again:
+ err = xenbus_transaction_start(&xbt);
+ if (err) {
+ xenbus_dev_fatal(dev, err, "starting transaction");
+ }
+
+ err = xenbus_printf(xbt, dev->nodename, "ring-ref", "%u",
+ info->ring_ref);
+ if (err) {
+ xenbus_dev_fatal(dev, err, "%s", "writing ring-ref");
+ goto fail;
+ }
+
+ err = xenbus_printf(xbt, dev->nodename, "event-channel", "%u",
+ info->evtchn);
+
+ if (err) {
+ xenbus_dev_fatal(dev, err, "%s", "writing event-channel");
+ goto fail;
+ }
+
+ err = xenbus_transaction_end(xbt, 0);
+ if (err) {
+ if (err == -EAGAIN)
+ goto again;
+ xenbus_dev_fatal(dev, err, "completing transaction");
+ goto free_sring;
+ }
+
+ return 0;
+
+fail:
+ xenbus_transaction_end(xbt, 1);
+free_sring:
+ /* free resource */
+ scsifront_free(info);
+
+ return err;
+}
+
+
+static int scsifront_probe(struct xenbus_device *dev,
+ const struct xenbus_device_id *id)
+{
+ struct vscsifrnt_info *info;
+ struct Scsi_Host *host;
+ int i, err = -ENOMEM;
+ char name[DEFAULT_TASK_COMM_LEN];
+
+ host = scsi_host_alloc(&scsifront_sht, sizeof(*info));
+ if (!host) {
+ xenbus_dev_fatal(dev, err, "fail to allocate scsi host");
+ return err;
+ }
+ info = (struct vscsifrnt_info *) host->hostdata;
+ info->host = host;
+
+
+ dev_set_drvdata(&dev->dev, info);
+ info->dev = dev;
+
+ for (i = 0; i < VSCSIIF_MAX_REQS; i++) {
+ info->shadow[i].next_free = i + 1;
+ init_waitqueue_head(&(info->shadow[i].wq_reset));
+ info->shadow[i].wait_reset = 0;
+ }
+ info->shadow[VSCSIIF_MAX_REQS - 1].next_free = 0x0fff;
+
+ err = scsifront_init_ring(info);
+ if (err) {
+ scsi_host_put(host);
+ return err;
+ }
+
+ init_waitqueue_head(&info->wq);
+ spin_lock_init(&info->io_lock);
+ spin_lock_init(&info->shadow_lock);
+
+ snprintf(name, DEFAULT_TASK_COMM_LEN, "vscsiif.%d", info->host->host_no);
+
+ info->kthread = kthread_run(scsifront_schedule, info, name);
+ if (IS_ERR(info->kthread)) {
+ err = PTR_ERR(info->kthread);
+ info->kthread = NULL;
+ printk(KERN_ERR "scsifront: kthread start err %d\n", err);
+ goto free_sring;
+ }
+
+ host->max_id = VSCSIIF_MAX_TARGET;
+ host->max_channel = 0;
+ host->max_lun = VSCSIIF_MAX_LUN;
+ host->max_sectors = (VSCSIIF_SG_TABLESIZE - 1) * PAGE_SIZE / 512;
+
+ err = scsi_add_host(host, &dev->dev);
+ if (err) {
+ printk(KERN_ERR "scsifront: fail to add scsi host %d\n", err);
+ goto free_sring;
+ }
+
+ xenbus_switch_state(dev, XenbusStateInitialised);
+
+ return 0;
+
+free_sring:
+ /* free resource */
+ scsifront_free(info);
+ return err;
+}
+
+static int scsifront_remove(struct xenbus_device *dev)
+{
+ struct vscsifrnt_info *info = dev_get_drvdata(&dev->dev);
+
+ DPRINTK("%s: %s removed\n",__FUNCTION__ ,dev->nodename);
+
+ if (info->kthread) {
+ kthread_stop(info->kthread);
+ info->kthread = NULL;
+ }
+
+ scsifront_free(info);
+
+ return 0;
+}
+
+
+static int scsifront_disconnect(struct vscsifrnt_info *info)
+{
+ struct xenbus_device *dev = info->dev;
+ struct Scsi_Host *host = info->host;
+
+ DPRINTK("%s: %s disconnect\n",__FUNCTION__ ,dev->nodename);
+
+ /*
+ When this function is executed, all devices of
+ Frontend have been deleted.
+ Therefore, it need not block I/O before remove_host.
+ */
+
+ scsi_remove_host(host);
+ xenbus_frontend_closed(dev);
+
+ return 0;
+}
+
+#define VSCSIFRONT_OP_ADD_LUN 1
+#define VSCSIFRONT_OP_DEL_LUN 2
+
+static void scsifront_do_lun_hotplug(struct vscsifrnt_info *info, int op)
+{
+ struct xenbus_device *dev = info->dev;
+ int i, err = 0;
+ char str[64], state_str[64];
+ char **dir;
+ unsigned int dir_n = 0;
+ unsigned int device_state;
+ unsigned int hst, chn, tgt, lun;
+ struct scsi_device *sdev;
+
+ dir = xenbus_directory(XBT_NIL, dev->otherend, "vscsi-devs", &dir_n);
+ if (IS_ERR(dir))
+ return;
+
+ for (i = 0; i < dir_n; i++) {
+ /* read status */
+ snprintf(str, sizeof(str), "vscsi-devs/%s/state", dir[i]);
+ err = xenbus_scanf(XBT_NIL, dev->otherend, str, "%u",
+ &device_state);
+ if (XENBUS_EXIST_ERR(err))
+ continue;
+
+ /* virtual SCSI device */
+ snprintf(str, sizeof(str), "vscsi-devs/%s/v-dev", dir[i]);
+ err = xenbus_scanf(XBT_NIL, dev->otherend, str,
+ "%u:%u:%u:%u", &hst, &chn, &tgt, &lun);
+ if (XENBUS_EXIST_ERR(err))
+ continue;
+
+ /* front device state path */
+ snprintf(state_str, sizeof(state_str), "vscsi-devs/%s/state", dir[i]);
+
+ switch (op) {
+ case VSCSIFRONT_OP_ADD_LUN:
+ if (device_state == XenbusStateInitialised) {
+ sdev = scsi_device_lookup(info->host, chn, tgt, lun);
+ if (sdev) {
+ printk(KERN_ERR "scsifront: Device already in use.\n");
+ scsi_device_put(sdev);
+ xenbus_printf(XBT_NIL, dev->nodename,
+ state_str, "%d", XenbusStateClosed);
+ } else {
+ scsi_add_device(info->host, chn, tgt, lun);
+ xenbus_printf(XBT_NIL, dev->nodename,
+ state_str, "%d", XenbusStateConnected);
+ }
+ }
+ break;
+ case VSCSIFRONT_OP_DEL_LUN:
+ if (device_state == XenbusStateClosing) {
+ sdev = scsi_device_lookup(info->host, chn, tgt, lun);
+ if (sdev) {
+ scsi_remove_device(sdev);
+ scsi_device_put(sdev);
+ xenbus_printf(XBT_NIL, dev->nodename,
+ state_str, "%d", XenbusStateClosed);
+ }
+ }
+ break;
+ default:
+ break;
+ }
+ }
+
+ kfree(dir);
+ return;
+}
+
+
+
+
+static void scsifront_backend_changed(struct xenbus_device *dev,
+ enum xenbus_state backend_state)
+{
+ struct vscsifrnt_info *info = dev_get_drvdata(&dev->dev);
+
+ DPRINTK("%p %u %u\n", dev, dev->state, backend_state);
+
+ switch (backend_state) {
+ case XenbusStateUnknown:
+ case XenbusStateInitialising:
+ case XenbusStateInitWait:
+ case XenbusStateClosed:
+ break;
+
+ case XenbusStateInitialised:
+ break;
+
+ case XenbusStateConnected:
+ if (xenbus_read_driver_state(dev->nodename) ==
+ XenbusStateInitialised) {
+ scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_ADD_LUN);
+ }
+
+ if (dev->state == XenbusStateConnected)
+ break;
+
+ xenbus_switch_state(dev, XenbusStateConnected);
+ break;
+
+ case XenbusStateClosing:
+ scsifront_disconnect(info);
+ break;
+
+ case XenbusStateReconfiguring:
+ scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_DEL_LUN);
+ xenbus_switch_state(dev, XenbusStateReconfiguring);
+ break;
+
+ case XenbusStateReconfigured:
+ scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_ADD_LUN);
+ xenbus_switch_state(dev, XenbusStateConnected);
+ break;
+ }
+}
+
+
+static struct xenbus_device_id scsifront_ids[] = {
+ { "vscsi" },
+ { "" }
+};
+MODULE_ALIAS("xen:vscsi");
+
+static struct xenbus_driver scsifront_driver = {
+ .name = "vscsi",
+ .ids = scsifront_ids,
+ .probe = scsifront_probe,
+ .remove = scsifront_remove,
+/* .resume = scsifront_resume, */
+ .otherend_changed = scsifront_backend_changed,
+};
+
+int scsifront_xenbus_init(void)
+{
+ return xenbus_register_frontend(&scsifront_driver);
+}
+
+void scsifront_xenbus_unregister(void)
+{
+ xenbus_unregister_driver(&scsifront_driver);
+}
+
diff -N -u -r linux-2.6-xen-orig/drivers/xen/Kconfig linux-2.6-xen-pvscsi/drivers/xen/Kconfig
--- linux-2.6-xen-orig/drivers/xen/Kconfig 2011-01-06 17:23:23.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/xen/Kconfig 2011-01-06 21:36:39.000000000 +0200
@@ -43,6 +43,14 @@
Implement the network backend driver, which passes packets
from the guest domain's frontend drivers to the network.
+config XEN_SCSI_BACKEND
+ tristate "Xen PVSCSI backend driver"
+ depends on XEN_BACKEND && SCSI
+ default m
+ help
+ The PVSCSI backend driver allows the kernel to export its SCSI Devices
+ to other Xen guests via a high-performance shared-memory interface.
+
config XEN_BLKDEV_BACKEND
tristate "Block-device backend driver"
depends on XEN_BACKEND && BLOCK
diff -N -u -r linux-2.6-xen-orig/drivers/xen/Makefile linux-2.6-xen-pvscsi/drivers/xen/Makefile
--- linux-2.6-xen-orig/drivers/xen/Makefile 2011-01-06 17:23:23.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/xen/Makefile 2011-01-06 21:37:12.000000000 +0200
@@ -14,6 +14,7 @@
obj-$(CONFIG_XEN_BLKDEV_BACKEND) += blkback/
obj-$(CONFIG_XEN_BLKDEV_TAP) += blktap/
obj-$(CONFIG_XEN_NETDEV_BACKEND) += netback/
+obj-$(CONFIG_XEN_SCSI_BACKEND) += scsiback/
obj-$(CONFIG_XENFS) += xenfs/
obj-$(CONFIG_XEN_SYS_HYPERVISOR) += sys-hypervisor.o
obj-$(CONFIG_XEN_MCE) += mce.o
diff -N -u -r linux-2.6-xen-orig/drivers/xen/scsiback/common.h linux-2.6-xen-pvscsi/drivers/xen/scsiback/common.h
--- linux-2.6-xen-orig/drivers/xen/scsiback/common.h 1970-01-01 02:00:00.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/xen/scsiback/common.h 2011-01-06 19:46:31.000000000 +0200
@@ -0,0 +1,189 @@
+/*
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * Based on the blkback driver code.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef __SCSIIF__BACKEND__COMMON_H__
+#define __SCSIIF__BACKEND__COMMON_H__
+
+#include <linux/version.h>
+#include <linux/module.h>
+#include <linux/interrupt.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <linux/wait.h>
+#include <linux/sched.h>
+#include <linux/kthread.h>
+#include <linux/blkdev.h>
+#include <linux/list.h>
+#include <linux/kthread.h>
+#include <scsi/scsi.h>
+#include <scsi/scsi_cmnd.h>
+#include <scsi/scsi_host.h>
+#include <scsi/scsi_device.h>
+#include <scsi/scsi_dbg.h>
+#include <scsi/scsi_eh.h>
+#include <asm/io.h>
+#include <asm/setup.h>
+#include <asm/pgalloc.h>
+#include <asm/delay.h>
+#include <xen/evtchn.h>
+#include <asm/hypervisor.h>
+#include <xen/xen.h>
+#include <xen/events.h>
+#include <xen/grant_table.h>
+#include <xen/xenbus.h>
+#include <xen/page.h>
+#include <xen/interface/xen.h>
+#include <xen/interface/io/ring.h>
+#include <xen/interface/grant_table.h>
+#include <xen/interface/io/vscsiif.h>
+
+
+#define DPRINTK(_f, _a...) \
+ pr_debug("(file=%s, line=%d) " _f, \
+ __FILE__ , __LINE__ , ## _a )
+
+struct ids_tuple {
+ unsigned int hst; /* host */
+ unsigned int chn; /* channel */
+ unsigned int tgt; /* target */
+ unsigned int lun; /* LUN */
+};
+
+struct v2p_entry {
+ struct ids_tuple v; /* translate from */
+ struct scsi_device *sdev; /* translate to */
+ struct list_head l;
+};
+
+struct vscsibk_info {
+ struct xenbus_device *dev;
+
+ domid_t domid;
+ unsigned int evtchn;
+ unsigned int irq;
+
+ int feature;
+
+ struct vscsiif_back_ring ring;
+ struct vm_struct *ring_area;
+ grant_handle_t shmem_handle;
+ grant_ref_t shmem_ref;
+
+ spinlock_t ring_lock;
+ atomic_t nr_unreplied_reqs;
+
+ spinlock_t v2p_lock;
+ struct list_head v2p_entry_lists;
+
+ struct task_struct *kthread;
+ wait_queue_head_t waiting_to_free;
+ wait_queue_head_t wq;
+ unsigned int waiting_reqs;
+ struct page **mmap_pages;
+
+};
+
+typedef struct {
+ unsigned char act;
+ struct vscsibk_info *info;
+ struct scsi_device *sdev;
+
+ uint16_t rqid;
+
+ uint16_t v_chn, v_tgt;
+
+ uint8_t nr_segments;
+ uint8_t cmnd[VSCSIIF_MAX_COMMAND_SIZE];
+ uint8_t cmd_len;
+
+ uint8_t sc_data_direction;
+ uint16_t timeout_per_command;
+
+ uint32_t request_bufflen;
+ struct scatterlist *sgl;
+ grant_ref_t gref[VSCSIIF_SG_TABLESIZE];
+
+ int32_t rslt;
+ uint32_t resid;
+ uint8_t sense_buffer[VSCSIIF_SENSE_BUFFERSIZE];
+
+ struct list_head free_list;
+} pending_req_t;
+
+
+
+#define scsiback_get(_b) (atomic_inc(&(_b)->nr_unreplied_reqs))
+#define scsiback_put(_b) \
+ do { \
+ if (atomic_dec_and_test(&(_b)->nr_unreplied_reqs)) \
+ wake_up(&(_b)->waiting_to_free);\
+ } while (0)
+
+#define VSCSIIF_TIMEOUT (900*HZ)
+
+#define VSCSI_TYPE_HOST 1
+
+irqreturn_t scsiback_intr(int, void *);
+int scsiback_init_sring(struct vscsibk_info *info,
+ unsigned long ring_ref, unsigned int evtchn);
+int scsiback_schedule(void *data);
+
+
+struct vscsibk_info *vscsibk_info_alloc(domid_t domid);
+void scsiback_free(struct vscsibk_info *info);
+void scsiback_disconnect(struct vscsibk_info *info);
+int __init scsiback_interface_init(void);
+void scsiback_interface_exit(void);
+int scsiback_xenbus_init(void);
+void scsiback_xenbus_unregister(void);
+
+void scsiback_init_translation_table(struct vscsibk_info *info);
+
+int scsiback_add_translation_entry(struct vscsibk_info *info,
+ struct scsi_device *sdev, struct ids_tuple *v);
+
+int scsiback_del_translation_entry(struct vscsibk_info *info,
+ struct ids_tuple *v);
+struct scsi_device *scsiback_do_translation(struct vscsibk_info *info,
+ struct ids_tuple *v);
+void scsiback_release_translation_entry(struct vscsibk_info *info);
+
+
+void scsiback_cmd_exec(pending_req_t *pending_req);
+void scsiback_do_resp_with_sense(char *sense_buffer, int32_t result,
+ uint32_t resid, pending_req_t *pending_req);
+void scsiback_fast_flush_area(pending_req_t *req);
+
+void scsiback_rsp_emulation(pending_req_t *pending_req);
+void scsiback_req_emulation_or_cmdexec(pending_req_t *pending_req);
+void scsiback_emulation_init(void);
+
+
+#endif /* __SCSIIF__BACKEND__COMMON_H__ */
diff -N -u -r linux-2.6-xen-orig/drivers/xen/scsiback/emulate.c linux-2.6-xen-pvscsi/drivers/xen/scsiback/emulate.c
--- linux-2.6-xen-orig/drivers/xen/scsiback/emulate.c 1970-01-01 02:00:00.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/xen/scsiback/emulate.c 2011-01-06 18:10:02.000000000 +0200
@@ -0,0 +1,471 @@
+/*
+ * Xen SCSI backend driver
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <scsi/scsi.h>
+#include <scsi/scsi_cmnd.h>
+#include <scsi/scsi_device.h>
+#include "common.h"
+
+/* Following SCSI commands are not defined in scsi/scsi.h */
+#define EXTENDED_COPY 0x83 /* EXTENDED COPY command */
+#define REPORT_ALIASES 0xa3 /* REPORT ALIASES command */
+#define CHANGE_ALIASES 0xa4 /* CHANGE ALIASES command */
+#define SET_PRIORITY 0xa4 /* SET PRIORITY command */
+
+
+/*
+ The bitmap in order to control emulation.
+ (Bit 3 to 7 are reserved for future use.)
+*/
+#define VSCSIIF_NEED_CMD_EXEC 0x01 /* If this bit is set, cmd exec */
+ /* is required. */
+#define VSCSIIF_NEED_EMULATE_REQBUF 0x02 /* If this bit is set, need */
+ /* emulation reqest buff before */
+ /* cmd exec. */
+#define VSCSIIF_NEED_EMULATE_RSPBUF 0x04 /* If this bit is set, need */
+ /* emulation resp buff after */
+ /* cmd exec. */
+
+/* Additional Sense Code (ASC) used */
+#define NO_ADDITIONAL_SENSE 0x0
+#define LOGICAL_UNIT_NOT_READY 0x4
+#define UNRECOVERED_READ_ERR 0x11
+#define PARAMETER_LIST_LENGTH_ERR 0x1a
+#define INVALID_OPCODE 0x20
+#define ADDR_OUT_OF_RANGE 0x21
+#define INVALID_FIELD_IN_CDB 0x24
+#define INVALID_FIELD_IN_PARAM_LIST 0x26
+#define POWERON_RESET 0x29
+#define SAVING_PARAMS_UNSUP 0x39
+#define THRESHOLD_EXCEEDED 0x5d
+#define LOW_POWER_COND_ON 0x5e
+
+
+
+/* Number os SCSI op_code */
+#define VSCSI_MAX_SCSI_OP_CODE 256
+static unsigned char bitmap[VSCSI_MAX_SCSI_OP_CODE];
+
+#define NO_EMULATE(cmd) \
+ bitmap[cmd] = VSCSIIF_NEED_CMD_EXEC; \
+ pre_function[cmd] = NULL; \
+ post_function[cmd] = NULL
+
+
+
+/*
+ Emulation routines for each SCSI op_code.
+*/
+static void (*pre_function[VSCSI_MAX_SCSI_OP_CODE])(pending_req_t *, void *);
+static void (*post_function[VSCSI_MAX_SCSI_OP_CODE])(pending_req_t *, void *);
+
+
+static const int check_condition_result =
+ (DRIVER_SENSE << 24) | SAM_STAT_CHECK_CONDITION;
+
+static void scsiback_mk_sense_buffer(uint8_t *data, uint8_t key,
+ uint8_t asc, uint8_t asq)
+{
+ data[0] = 0x70; /* fixed, current */
+ data[2] = key;
+ data[7] = 0xa; /* implies 18 byte sense buffer */
+ data[12] = asc;
+ data[13] = asq;
+}
+
+static void resp_not_supported_cmd(pending_req_t *pending_req, void *data)
+{
+ scsiback_mk_sense_buffer(pending_req->sense_buffer, ILLEGAL_REQUEST,
+ INVALID_OPCODE, 0);
+ pending_req->resid = 0;
+ pending_req->rslt = check_condition_result;
+}
+
+
+static int __copy_to_sg(struct scatterlist *sgl, unsigned int nr_sg,
+ void *buf, unsigned int buflen)
+{
+ struct scatterlist *sg;
+ void *from = buf;
+ void *to;
+ unsigned int from_rest = buflen;
+ unsigned int to_capa;
+ unsigned int copy_size = 0;
+ unsigned int i;
+ unsigned long pfn;
+
+ for_each_sg (sgl, sg, nr_sg, i) {
+ if (sg_page(sg) == NULL) {
+ printk(KERN_WARNING "%s: inconsistent length field in "
+ "scatterlist\n", __FUNCTION__);
+ return -ENOMEM;
+ }
+
+ to_capa = sg->length;
+ copy_size = min_t(unsigned int, to_capa, from_rest);
+
+ pfn = page_to_pfn(sg_page(sg));
+ to = pfn_to_kaddr(pfn) + (sg->offset);
+ memcpy(to, from, copy_size);
+
+ from_rest -= copy_size;
+ if (from_rest == 0) {
+ return 0;
+ }
+
+ from += copy_size;
+ }
+
+ printk(KERN_WARNING "%s: no space in scatterlist\n",
+ __FUNCTION__);
+ return -ENOMEM;
+}
+
+static int __copy_from_sg(struct scatterlist *sgl, unsigned int nr_sg,
+ void *buf, unsigned int buflen)
+{
+ struct scatterlist *sg;
+ void *from;
+ void *to = buf;
+ unsigned int from_rest;
+ unsigned int to_capa = buflen;
+ unsigned int copy_size;
+ unsigned int i;
+ unsigned long pfn;
+
+ for_each_sg (sgl, sg, nr_sg, i) {
+ if (sg_page(sg) == NULL) {
+ printk(KERN_WARNING "%s: inconsistent length field in "
+ "scatterlist\n", __FUNCTION__);
+ return -ENOMEM;
+ }
+
+ from_rest = sg->length;
+ if ((from_rest > 0) && (to_capa < from_rest)) {
+ printk(KERN_WARNING
+ "%s: no space in destination buffer\n",
+ __FUNCTION__);
+ return -ENOMEM;
+ }
+ copy_size = from_rest;
+
+ pfn = page_to_pfn(sg_page(sg));
+ from = pfn_to_kaddr(pfn) + (sg->offset);
+ memcpy(to, from, copy_size);
+
+ to_capa -= copy_size;
+ to += copy_size;
+ }
+
+ return 0;
+}
+
+static int __nr_luns_under_host(struct vscsibk_info *info)
+{
+ struct v2p_entry *entry;
+ struct list_head *head = &(info->v2p_entry_lists);
+ unsigned long flags;
+ int lun_cnt = 0;
+
+ spin_lock_irqsave(&info->v2p_lock, flags);
+ list_for_each_entry(entry, head, l) {
+ lun_cnt++;
+ }
+ spin_unlock_irqrestore(&info->v2p_lock, flags);
+
+ return (lun_cnt);
+}
+
+
+/* REPORT LUNS Define*/
+#define VSCSI_REPORT_LUNS_HEADER 8
+#define VSCSI_REPORT_LUNS_RETRY 3
+
+/* quoted scsi_debug.c/resp_report_luns() */
+static void __report_luns(pending_req_t *pending_req, void *data)
+{
+ struct vscsibk_info *info = pending_req->info;
+ unsigned int channel = pending_req->v_chn;
+ unsigned int target = pending_req->v_tgt;
+ unsigned int nr_seg = pending_req->nr_segments;
+ unsigned char *cmd = (unsigned char *)pending_req->cmnd;
+
+ unsigned char *buff = NULL;
+ unsigned char alloc_len;
+ unsigned int alloc_luns = 0;
+ unsigned int req_bufflen = 0;
+ unsigned int actual_len = 0;
+ unsigned int retry_cnt = 0;
+ int select_report = (int)cmd[2];
+ int i, lun_cnt = 0, lun, upper, err = 0;
+
+ struct v2p_entry *entry;
+ struct list_head *head = &(info->v2p_entry_lists);
+ unsigned long flags;
+
+ struct scsi_lun *one_lun;
+
+ req_bufflen = cmd[9] + (cmd[8] << 8) + (cmd[7] << 16) + (cmd[6] << 24);
+ if ((req_bufflen < 4) || (select_report != 0))
+ goto fail;
+
+ alloc_luns = __nr_luns_under_host(info);
+ alloc_len = sizeof(struct scsi_lun) * alloc_luns
+ + VSCSI_REPORT_LUNS_HEADER;
+retry:
+ if ((buff = kzalloc(alloc_len, GFP_KERNEL)) == NULL) {
+ printk(KERN_ERR "scsiback:%s kmalloc err\n", __FUNCTION__);
+ goto fail;
+ }
+
+ one_lun = (struct scsi_lun *) &buff[8];
+ spin_lock_irqsave(&info->v2p_lock, flags);
+ list_for_each_entry(entry, head, l) {
+ if ((entry->v.chn == channel) &&
+ (entry->v.tgt == target)) {
+
+ /* check overflow */
+ if (lun_cnt >= alloc_luns) {
+ spin_unlock_irqrestore(&info->v2p_lock,
+ flags);
+
+ if (retry_cnt < VSCSI_REPORT_LUNS_RETRY) {
+ retry_cnt++;
+ if (buff)
+ kfree(buff);
+ goto retry;
+ }
+
+ goto fail;
+ }
+
+ lun = entry->v.lun;
+ upper = (lun >> 8) & 0x3f;
+ if (upper)
+ one_lun[lun_cnt].scsi_lun[0] = upper;
+ one_lun[lun_cnt].scsi_lun[1] = lun & 0xff;
+ lun_cnt++;
+ }
+ }
+
+ spin_unlock_irqrestore(&info->v2p_lock, flags);
+
+ buff[2] = ((sizeof(struct scsi_lun) * lun_cnt) >> 8) & 0xff;
+ buff[3] = (sizeof(struct scsi_lun) * lun_cnt) & 0xff;
+
+ actual_len = lun_cnt * sizeof(struct scsi_lun)
+ + VSCSI_REPORT_LUNS_HEADER;
+ req_bufflen = 0;
+ for (i = 0; i < nr_seg; i++)
+ req_bufflen += pending_req->sgl[i].length;
+
+ err = __copy_to_sg(pending_req->sgl, nr_seg, buff,
+ min(req_bufflen, actual_len));
+ if (err)
+ goto fail;
+
+ memset(pending_req->sense_buffer, 0, VSCSIIF_SENSE_BUFFERSIZE);
+ pending_req->rslt = 0x00;
+ pending_req->resid = req_bufflen - min(req_bufflen, actual_len);
+
+ kfree(buff);
+ return;
+
+fail:
+ scsiback_mk_sense_buffer(pending_req->sense_buffer, ILLEGAL_REQUEST,
+ INVALID_FIELD_IN_CDB, 0);
+ pending_req->rslt = check_condition_result;
+ pending_req->resid = 0;
+ if (buff)
+ kfree(buff);
+ return;
+}
+
+
+
+int __pre_do_emulation(pending_req_t *pending_req, void *data)
+{
+ uint8_t op_code = pending_req->cmnd[0];
+
+ if ((bitmap[op_code] & VSCSIIF_NEED_EMULATE_REQBUF) &&
+ pre_function[op_code] != NULL) {
+ pre_function[op_code](pending_req, data);
+ }
+
+ /*
+ 0: no need for native driver call, so should return immediately.
+ 1: non emulation or should call native driver
+ after modifing the request buffer.
+ */
+ return !!(bitmap[op_code] & VSCSIIF_NEED_CMD_EXEC);
+}
+
+void scsiback_rsp_emulation(pending_req_t *pending_req)
+{
+ uint8_t op_code = pending_req->cmnd[0];
+
+ if ((bitmap[op_code] & VSCSIIF_NEED_EMULATE_RSPBUF) &&
+ post_function[op_code] != NULL) {
+ post_function[op_code](pending_req, NULL);
+ }
+
+ return;
+}
+
+
+void scsiback_req_emulation_or_cmdexec(pending_req_t *pending_req)
+{
+ if (__pre_do_emulation(pending_req, NULL)) {
+ scsiback_cmd_exec(pending_req);
+ }
+ else {
+ scsiback_fast_flush_area(pending_req);
+ scsiback_do_resp_with_sense(pending_req->sense_buffer,
+ pending_req->rslt, pending_req->resid, pending_req);
+ }
+}
+
+
+/*
+ Following are not customizable functions.
+*/
+void scsiback_emulation_init(void)
+{
+ int i;
+
+ /* Initialize to default state */
+ for (i = 0; i < VSCSI_MAX_SCSI_OP_CODE; i++) {
+ bitmap[i] = (VSCSIIF_NEED_EMULATE_REQBUF |
+ VSCSIIF_NEED_EMULATE_RSPBUF);
+ pre_function[i] = resp_not_supported_cmd;
+ post_function[i] = NULL;
+ /* means,
+ - no need for pre-emulation
+ - no need for post-emulation
+ - call native driver
+ */
+ }
+
+ /*
+ Register appropriate functions below as you need.
+ (See scsi/scsi.h for definition of SCSI op_code.)
+ */
+
+ /*
+ Following commands do not require emulation.
+ */
+ NO_EMULATE(TEST_UNIT_READY); /*0x00*/
+ NO_EMULATE(REZERO_UNIT); /*0x01*/
+ NO_EMULATE(REQUEST_SENSE); /*0x03*/
+ NO_EMULATE(FORMAT_UNIT); /*0x04*/
+ NO_EMULATE(READ_BLOCK_LIMITS); /*0x05*/
+ /*NO_EMULATE(REASSIGN_BLOCKS); *//*0x07*/
+ /*NO_EMULATE(INITIALIZE_ELEMENT_STATUS); *//*0x07*/
+ NO_EMULATE(READ_6); /*0x08*/
+ NO_EMULATE(WRITE_6); /*0x0a*/
+ /*NO_EMULATE(SEEK_6); *//*0x0b*/
+ /*NO_EMULATE(READ_REVERSE); *//*0x0f*/
+ NO_EMULATE(WRITE_FILEMARKS); /*0x10*/
+ NO_EMULATE(SPACE); /*0x11*/
+ NO_EMULATE(INQUIRY); /*0x12*/
+ /*NO_EMULATE(RECOVER_BUFFERED_DATA); *//*0x14*/
+ /*NO_EMULATE(MODE_SELECT); *//*0x15*/
+ /*NO_EMULATE(RESERVE); *//*0x16*/
+ /*NO_EMULATE(RELEASE); *//*0x17*/
+ /*NO_EMULATE(COPY); *//*0x18*/
+ NO_EMULATE(ERASE); /*0x19*/
+ NO_EMULATE(MODE_SENSE); /*0x1a*/
+ /*NO_EMULATE(START_STOP); *//*0x1b*/
+ /*NO_EMULATE(RECEIVE_DIAGNOSTIC); *//*0x1c*/
+ NO_EMULATE(SEND_DIAGNOSTIC); /*0x1d*/
+ /*NO_EMULATE(ALLOW_MEDIUM_REMOVAL); *//*0x1e*/
+
+ /*NO_EMULATE(SET_WINDOW); *//*0x24*/
+ NO_EMULATE(READ_CAPACITY); /*0x25*/
+ NO_EMULATE(READ_10); /*0x28*/
+ NO_EMULATE(WRITE_10); /*0x2a*/
+ /*NO_EMULATE(SEEK_10); *//*0x2b*/
+ /*NO_EMULATE(POSITION_TO_ELEMENT); *//*0x2b*/
+ /*NO_EMULATE(WRITE_VERIFY); *//*0x2e*/
+ /*NO_EMULATE(VERIFY); *//*0x2f*/
+ /*NO_EMULATE(SEARCH_HIGH); *//*0x30*/
+ /*NO_EMULATE(SEARCH_EQUAL); *//*0x31*/
+ /*NO_EMULATE(SEARCH_LOW); *//*0x32*/
+ /*NO_EMULATE(SET_LIMITS); *//*0x33*/
+ /*NO_EMULATE(PRE_FETCH); *//*0x34*/
+ /*NO_EMULATE(READ_POSITION); *//*0x34*/
+ /*NO_EMULATE(SYNCHRONIZE_CACHE); *//*0x35*/
+ /*NO_EMULATE(LOCK_UNLOCK_CACHE); *//*0x36*/
+ /*NO_EMULATE(READ_DEFECT_DATA); *//*0x37*/
+ /*NO_EMULATE(MEDIUM_SCAN); *//*0x38*/
+ /*NO_EMULATE(COMPARE); *//*0x39*/
+ /*NO_EMULATE(COPY_VERIFY); *//*0x3a*/
+ /*NO_EMULATE(WRITE_BUFFER); *//*0x3b*/
+ /*NO_EMULATE(READ_BUFFER); *//*0x3c*/
+ /*NO_EMULATE(UPDATE_BLOCK); *//*0x3d*/
+ /*NO_EMULATE(READ_LONG); *//*0x3e*/
+ /*NO_EMULATE(WRITE_LONG); *//*0x3f*/
+ /*NO_EMULATE(CHANGE_DEFINITION); *//*0x40*/
+ /*NO_EMULATE(WRITE_SAME); *//*0x41*/
+ /*NO_EMULATE(READ_TOC); *//*0x43*/
+ /*NO_EMULATE(LOG_SELECT); *//*0x4c*/
+ /*NO_EMULATE(LOG_SENSE); *//*0x4d*/
+ /*NO_EMULATE(MODE_SELECT_10); *//*0x55*/
+ /*NO_EMULATE(RESERVE_10); *//*0x56*/
+ /*NO_EMULATE(RELEASE_10); *//*0x57*/
+ /*NO_EMULATE(MODE_SENSE_10); *//*0x5a*/
+ /*NO_EMULATE(PERSISTENT_RESERVE_IN); *//*0x5e*/
+ /*NO_EMULATE(PERSISTENT_RESERVE_OUT); *//*0x5f*/
+ /* REPORT_LUNS *//*0xa0*//*Full emulaiton*/
+ /*NO_EMULATE(MOVE_MEDIUM); *//*0xa5*/
+ /*NO_EMULATE(EXCHANGE_MEDIUM); *//*0xa6*/
+ /*NO_EMULATE(READ_12); *//*0xa8*/
+ /*NO_EMULATE(WRITE_12); *//*0xaa*/
+ /*NO_EMULATE(WRITE_VERIFY_12); *//*0xae*/
+ /*NO_EMULATE(SEARCH_HIGH_12); *//*0xb0*/
+ /*NO_EMULATE(SEARCH_EQUAL_12); *//*0xb1*/
+ /*NO_EMULATE(SEARCH_LOW_12); *//*0xb2*/
+ /*NO_EMULATE(READ_ELEMENT_STATUS); *//*0xb8*/
+ /*NO_EMULATE(SEND_VOLUME_TAG); *//*0xb6*/
+ /*NO_EMULATE(WRITE_LONG_2); *//*0xea*/
+ /*NO_EMULATE(READ_16); *//*0x88*/
+ /*NO_EMULATE(WRITE_16); *//*0x8a*/
+ /*NO_EMULATE(VERIFY_16); *//*0x8f*/
+ /*NO_EMULATE(SERVICE_ACTION_IN); *//*0x9e*/
+
+ /*
+ Following commands require emulation.
+ */
+ pre_function[REPORT_LUNS] = __report_luns;
+ bitmap[REPORT_LUNS] = (VSCSIIF_NEED_EMULATE_REQBUF |
+ VSCSIIF_NEED_EMULATE_RSPBUF);
+
+ return;
+}
diff -N -u -r linux-2.6-xen-orig/drivers/xen/scsiback/interface.c linux-2.6-xen-pvscsi/drivers/xen/scsiback/interface.c
--- linux-2.6-xen-orig/drivers/xen/scsiback/interface.c 1970-01-01 02:00:00.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/xen/scsiback/interface.c 2011-01-06 18:10:02.000000000 +0200
@@ -0,0 +1,185 @@
+/*
+ * interface management.
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * Based on the blkback driver code.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <scsi/scsi.h>
+#include <scsi/scsi_host.h>
+#include <scsi/scsi_device.h>
+#include "common.h"
+
+#include <xen/evtchn.h>
+#include <linux/kthread.h>
+#include <linux/delay.h>
+
+
+static struct kmem_cache *scsiback_cachep;
+
+struct vscsibk_info *vscsibk_info_alloc(domid_t domid)
+{
+ struct vscsibk_info *info;
+
+ info = kmem_cache_zalloc(scsiback_cachep, GFP_KERNEL);
+ if (!info)
+ return ERR_PTR(-ENOMEM);
+
+ info->domid = domid;
+ spin_lock_init(&info->ring_lock);
+ atomic_set(&info->nr_unreplied_reqs, 0);
+ init_waitqueue_head(&info->wq);
+ init_waitqueue_head(&info->waiting_to_free);
+
+ return info;
+}
+
+static int map_frontend_page( struct vscsibk_info *info,
+ unsigned long ring_ref)
+{
+ struct gnttab_map_grant_ref op;
+ int err;
+
+ gnttab_set_map_op(&op, (unsigned long)info->ring_area->addr,
+ GNTMAP_host_map, ring_ref,
+ info->domid);
+
+ do {
+ err = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, &op, 1);
+ BUG_ON(err);
+ msleep(10);
+ } while(op.status == GNTST_eagain);
+
+ if (op.status) {
+ printk(KERN_ERR "scsiback: Grant table operation failure !\n");
+ return op.status;
+ }
+
+ info->shmem_ref = ring_ref;
+ info->shmem_handle = op.handle;
+
+ return (GNTST_okay);
+}
+
+static void unmap_frontend_page(struct vscsibk_info *info)
+{
+ struct gnttab_unmap_grant_ref op;
+ int err;
+
+ gnttab_set_unmap_op(&op, (unsigned long)info->ring_area->addr,
+ GNTMAP_host_map, info->shmem_handle);
+
+ err = HYPERVISOR_grant_table_op(GNTTABOP_unmap_grant_ref, &op, 1);
+ BUG_ON(err);
+
+}
+
+int scsiback_init_sring(struct vscsibk_info *info,
+ unsigned long ring_ref, unsigned int evtchn)
+{
+ struct vscsiif_sring *sring;
+ int err;
+
+ if (info->irq) {
+ printk(KERN_ERR "scsiback: Already connected through?\n");
+ return -1;
+ }
+
+ info->ring_area = alloc_vm_area(PAGE_SIZE);
+ if (!info)
+ return -ENOMEM;
+
+ err = map_frontend_page(info, ring_ref);
+ if (err)
+ goto free_vm;
+
+ sring = (struct vscsiif_sring *) info->ring_area->addr;
+ BACK_RING_INIT(&info->ring, sring, PAGE_SIZE);
+
+ err = bind_interdomain_evtchn_to_irqhandler(
+ info->domid, evtchn,
+ scsiback_intr, 0, "vscsiif-backend", info);
+
+ if (err < 0)
+ goto unmap_page;
+
+ info->irq = err;
+
+ return 0;
+
+unmap_page:
+ unmap_frontend_page(info);
+free_vm:
+ free_vm_area(info->ring_area);
+
+ return err;
+}
+
+void scsiback_disconnect(struct vscsibk_info *info)
+{
+ if (info->kthread) {
+ kthread_stop(info->kthread);
+ info->kthread = NULL;
+ }
+
+ wait_event(info->waiting_to_free,
+ atomic_read(&info->nr_unreplied_reqs) == 0);
+
+ if (info->irq) {
+ unbind_from_irqhandler(info->irq, info);
+ info->irq = 0;
+ }
+
+ if (info->ring.sring) {
+ unmap_frontend_page(info);
+ free_vm_area(info->ring_area);
+ info->ring.sring = NULL;
+ }
+}
+
+void scsiback_free(struct vscsibk_info *info)
+{
+ kmem_cache_free(scsiback_cachep, info);
+}
+
+int __init scsiback_interface_init(void)
+{
+ scsiback_cachep = kmem_cache_create("vscsiif_cache",
+ sizeof(struct vscsibk_info), 0, 0, NULL);
+ if (!scsiback_cachep) {
+ printk(KERN_ERR "scsiback: can't init scsi cache\n");
+ return -ENOMEM;
+ }
+
+ return 0;
+}
+
+void scsiback_interface_exit(void)
+{
+ kmem_cache_destroy(scsiback_cachep);
+}
diff -N -u -r linux-2.6-xen-orig/drivers/xen/scsiback/Makefile linux-2.6-xen-pvscsi/drivers/xen/scsiback/Makefile
--- linux-2.6-xen-orig/drivers/xen/scsiback/Makefile 1970-01-01 02:00:00.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/xen/scsiback/Makefile 2011-01-06 20:07:53.000000000 +0200
@@ -0,0 +1,4 @@
+obj-$(CONFIG_XEN_SCSI_BACKEND) := xen-scsiback.o
+
+xen-scsiback-y := interface.o scsiback.o xenbus.o translate.o emulate.o
+
diff -N -u -r linux-2.6-xen-orig/drivers/xen/scsiback/scsiback.c linux-2.6-xen-pvscsi/drivers/xen/scsiback/scsiback.c
--- linux-2.6-xen-orig/drivers/xen/scsiback/scsiback.c 1970-01-01 02:00:00.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/xen/scsiback/scsiback.c 2011-01-06 19:54:19.000000000 +0200
@@ -0,0 +1,737 @@
+/*
+ * Xen SCSI backend driver
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * Based on the blkback driver code.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <linux/spinlock.h>
+#include <linux/kthread.h>
+#include <linux/list.h>
+#include <linux/delay.h>
+#include <xen/balloon.h>
+#include <asm/hypervisor.h>
+#include <scsi/scsi.h>
+#include <scsi/scsi_cmnd.h>
+#include <scsi/scsi_host.h>
+#include <scsi/scsi_device.h>
+#include <scsi/scsi_dbg.h>
+#include <scsi/scsi_eh.h>
+
+#include "common.h"
+
+
+struct list_head pending_free;
+DEFINE_SPINLOCK(pending_free_lock);
+DECLARE_WAIT_QUEUE_HEAD(pending_free_wq);
+
+int vscsiif_reqs = VSCSIIF_BACK_MAX_PENDING_REQS;
+module_param_named(reqs, vscsiif_reqs, int, 0);
+MODULE_PARM_DESC(reqs, "Number of scsiback requests to allocate");
+
+static unsigned int log_print_stat = 0;
+module_param(log_print_stat, int, 0644);
+
+#define SCSIBACK_INVALID_HANDLE (~0)
+
+static pending_req_t *pending_reqs;
+static struct page **pending_pages;
+static grant_handle_t *pending_grant_handles;
+
+static int vaddr_pagenr(pending_req_t *req, int seg)
+{
+ return (req - pending_reqs) * VSCSIIF_SG_TABLESIZE + seg;
+}
+
+static unsigned long vaddr(pending_req_t *req, int seg)
+{
+ unsigned long pfn = page_to_pfn(pending_pages[vaddr_pagenr(req, seg)]);
+ return (unsigned long)pfn_to_kaddr(pfn);
+}
+
+#define pending_handle(_req, _seg) \
+ (pending_grant_handles[vaddr_pagenr(_req, _seg)])
+
+
+void scsiback_fast_flush_area(pending_req_t *req)
+{
+ struct gnttab_unmap_grant_ref unmap[VSCSIIF_SG_TABLESIZE];
+ unsigned int i, invcount = 0;
+ grant_handle_t handle;
+ int err;
+
+ if (req->nr_segments) {
+ for (i = 0; i < req->nr_segments; i++) {
+ handle = pending_handle(req, i);
+ if (handle == SCSIBACK_INVALID_HANDLE)
+ continue;
+ gnttab_set_unmap_op(&unmap[i], vaddr(req, i),
+ GNTMAP_host_map, handle);
+ pending_handle(req, i) = SCSIBACK_INVALID_HANDLE;
+ invcount++;
+ }
+
+ err = HYPERVISOR_grant_table_op(
+ GNTTABOP_unmap_grant_ref, unmap, invcount);
+ BUG_ON(err);
+ kfree(req->sgl);
+ }
+
+ return;
+}
+
+
+static pending_req_t * alloc_req(struct vscsibk_info *info)
+{
+ pending_req_t *req = NULL;
+ unsigned long flags;
+
+ spin_lock_irqsave(&pending_free_lock, flags);
+ if (!list_empty(&pending_free)) {
+ req = list_entry(pending_free.next, pending_req_t, free_list);
+ list_del(&req->free_list);
+ }
+ spin_unlock_irqrestore(&pending_free_lock, flags);
+ return req;
+}
+
+
+static void free_req(pending_req_t *req)
+{
+ unsigned long flags;
+ int was_empty;
+
+ spin_lock_irqsave(&pending_free_lock, flags);
+ was_empty = list_empty(&pending_free);
+ list_add(&req->free_list, &pending_free);
+ spin_unlock_irqrestore(&pending_free_lock, flags);
+ if (was_empty)
+ wake_up(&pending_free_wq);
+}
+
+
+static void scsiback_notify_work(struct vscsibk_info *info)
+{
+ info->waiting_reqs = 1;
+ wake_up(&info->wq);
+}
+
+void scsiback_do_resp_with_sense(char *sense_buffer, int32_t result,
+ uint32_t resid, pending_req_t *pending_req)
+{
+ vscsiif_response_t *ring_res;
+ struct vscsibk_info *info = pending_req->info;
+ int notify;
+ int more_to_do = 1;
+ struct scsi_sense_hdr sshdr;
+ unsigned long flags;
+
+ DPRINTK("%s\n",__FUNCTION__);
+
+ spin_lock_irqsave(&info->ring_lock, flags);
+
+ ring_res = RING_GET_RESPONSE(&info->ring, info->ring.rsp_prod_pvt);
+ info->ring.rsp_prod_pvt++;
+
+ ring_res->rslt = result;
+ ring_res->rqid = pending_req->rqid;
+
+ if (sense_buffer != NULL) {
+ if (scsi_normalize_sense(sense_buffer,
+ sizeof(sense_buffer), &sshdr)) {
+
+ int len = 8 + sense_buffer[7];
+
+ if (len > VSCSIIF_SENSE_BUFFERSIZE)
+ len = VSCSIIF_SENSE_BUFFERSIZE;
+
+ memcpy(ring_res->sense_buffer, sense_buffer, len);
+ ring_res->sense_len = len;
+ }
+ } else {
+ ring_res->sense_len = 0;
+ }
+
+ ring_res->residual_len = resid;
+
+ RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&info->ring, notify);
+ if (info->ring.rsp_prod_pvt == info->ring.req_cons) {
+ RING_FINAL_CHECK_FOR_REQUESTS(&info->ring, more_to_do);
+ } else if (RING_HAS_UNCONSUMED_REQUESTS(&info->ring)) {
+ more_to_do = 1;
+ }
+
+ spin_unlock_irqrestore(&info->ring_lock, flags);
+
+ if (more_to_do)
+ scsiback_notify_work(info);
+
+ if (notify)
+ notify_remote_via_irq(info->irq);
+
+ free_req(pending_req);
+}
+
+static void scsiback_print_status(char *sense_buffer, int errors,
+ pending_req_t *pending_req)
+{
+ struct scsi_device *sdev = pending_req->sdev;
+
+ printk(KERN_ERR "scsiback: %d:%d:%d:%d ",sdev->host->host_no,
+ sdev->channel, sdev->id, sdev->lun);
+ printk(KERN_ERR "status = 0x%02x, message = 0x%02x, host = 0x%02x, driver = 0x%02x\n",
+ status_byte(errors), msg_byte(errors),
+ host_byte(errors), driver_byte(errors));
+
+ printk(KERN_ERR "scsiback: cmnd[0]=0x%02X\n",
+ pending_req->cmnd[0]);
+
+ if (CHECK_CONDITION & status_byte(errors))
+ __scsi_print_sense("scsiback", sense_buffer, SCSI_SENSE_BUFFERSIZE);
+}
+
+
+static void scsiback_cmd_done(struct request *req, int uptodate)
+{
+ pending_req_t *pending_req = req->end_io_data;
+ unsigned char *sense_buffer;
+ unsigned int resid;
+ int errors;
+
+ sense_buffer = req->sense;
+ resid = blk_rq_bytes(req);
+ errors = req->errors;
+
+ if (errors != 0) {
+ if (log_print_stat)
+ scsiback_print_status(sense_buffer, errors, pending_req);
+ }
+
+ /* The Host mode is through as for Emulation. */
+ if (pending_req->info->feature != VSCSI_TYPE_HOST)
+ scsiback_rsp_emulation(pending_req);
+
+ scsiback_fast_flush_area(pending_req);
+ scsiback_do_resp_with_sense(sense_buffer, errors, resid, pending_req);
+ scsiback_put(pending_req->info);
+
+ __blk_put_request(req->q, req);
+}
+
+
+static int scsiback_gnttab_data_map(vscsiif_request_t *ring_req,
+ pending_req_t *pending_req)
+{
+ u32 flags;
+ int write;
+ int i, err = 0;
+ unsigned int data_len = 0;
+ struct gnttab_map_grant_ref map[VSCSIIF_SG_TABLESIZE];
+ struct vscsibk_info *info = pending_req->info;
+
+ int data_dir = (int)pending_req->sc_data_direction;
+ unsigned int nr_segments = (unsigned int)pending_req->nr_segments;
+
+ write = (data_dir == DMA_TO_DEVICE);
+
+ if (nr_segments) {
+ struct scatterlist *sg;
+
+ /* free of (sgl) in fast_flush_area()*/
+ pending_req->sgl = kmalloc(sizeof(struct scatterlist) * nr_segments,
+ GFP_KERNEL);
+ if (!pending_req->sgl) {
+ printk(KERN_ERR "scsiback: %s: kmalloc() error.\n", __FUNCTION__);
+ return -ENOMEM;
+ }
+
+ sg_init_table(pending_req->sgl, nr_segments);
+
+ for (i = 0; i < nr_segments; i++) {
+ flags = GNTMAP_host_map;
+ if (write)
+ flags |= GNTMAP_readonly;
+ gnttab_set_map_op(&map[i], vaddr(pending_req, i), flags,
+ ring_req->seg[i].gref,
+ info->domid);
+ }
+
+ err = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, map, nr_segments);
+ BUG_ON(err);
+ /* Retry maps with GNTST_eagain */
+ for(i=0; i < nr_segments; i++) {
+ while(unlikely(map[i].status == GNTST_eagain))
+ {
+ msleep(10);
+ err = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref,
+ &map[i],
+ 1);
+ BUG_ON(err);
+ }
+ }
+
+ for_each_sg (pending_req->sgl, sg, nr_segments, i) {
+ struct page *pg;
+
+ if (unlikely(map[i].status != 0)) {
+ printk(KERN_ERR "scsiback: invalid buffer -- could not remap it\n");
+ map[i].handle = SCSIBACK_INVALID_HANDLE;
+ err |= 1;
+ }
+
+ pending_handle(pending_req, i) = map[i].handle;
+
+ if (err)
+ continue;
+
+ pg = pending_pages[vaddr_pagenr(pending_req, i)];
+
+ set_phys_to_machine(page_to_pfn(pg),
+ FOREIGN_FRAME(map[i].dev_bus_addr >> PAGE_SHIFT));
+
+ sg_set_page(sg, pg, ring_req->seg[i].length,
+ ring_req->seg[i].offset);
+ data_len += sg->length;
+
+ barrier();
+ if (sg->offset >= PAGE_SIZE ||
+ sg->length > PAGE_SIZE ||
+ sg->offset + sg->length > PAGE_SIZE)
+ err |= 1;
+
+ }
+
+ if (err)
+ goto fail_flush;
+ }
+
+ pending_req->request_bufflen = data_len;
+
+ return 0;
+
+fail_flush:
+ scsiback_fast_flush_area(pending_req);
+ return -ENOMEM;
+}
+
+/* quoted scsi_lib.c/scsi_bi_endio */
+static void scsiback_bi_endio(struct bio *bio, int error)
+{
+ bio_put(bio);
+}
+
+
+
+/* quoted scsi_lib.c/scsi_req_map_sg . */
+static struct bio *request_map_sg(pending_req_t *pending_req)
+{
+ struct request_queue *q = pending_req->sdev->request_queue;
+ unsigned int nsegs = (unsigned int)pending_req->nr_segments;
+ unsigned int i, len, bytes, off, nr_pages, nr_vecs = 0;
+ struct scatterlist *sg;
+ struct page *page;
+ struct bio *bio = NULL, *bio_first = NULL, *bio_last = NULL;
+ int err;
+
+ for_each_sg (pending_req->sgl, sg, nsegs, i) {
+ page = sg_page(sg);
+ off = sg->offset;
+ len = sg->length;
+
+ nr_pages = (len + off + PAGE_SIZE - 1) >> PAGE_SHIFT;
+ while (len > 0) {
+ bytes = min_t(unsigned int, len, PAGE_SIZE - off);
+
+ if (!bio) {
+ nr_vecs = min_t(unsigned int, BIO_MAX_PAGES,
+ nr_pages);
+ nr_pages -= nr_vecs;
+ bio = bio_alloc(GFP_KERNEL, nr_vecs);
+ if (!bio) {
+ err = -ENOMEM;
+ goto free_bios;
+ }
+ bio->bi_end_io = scsiback_bi_endio;
+ if (bio_last)
+ bio_last->bi_next = bio;
+ else
+ bio_first = bio;
+ bio_last = bio;
+ }
+
+ if (bio_add_pc_page(q, bio, page, bytes, off) !=
+ bytes) {
+ bio_put(bio);
+ err = -EINVAL;
+ goto free_bios;
+ }
+
+ if (bio->bi_vcnt >= nr_vecs) {
+ bio->bi_flags &= ~(1 << BIO_SEG_VALID);
+ if (pending_req->sc_data_direction == WRITE)
+ bio->bi_rw |= (1 << BIO_RW);
+ bio = NULL;
+ }
+
+ page++;
+ len -= bytes;
+ off = 0;
+ }
+ }
+
+ return bio_first;
+
+free_bios:
+ while ((bio = bio_first) != NULL) {
+ bio_first = bio->bi_next;
+ bio_put(bio);
+ }
+
+ return ERR_PTR(err);
+}
+
+
+void scsiback_cmd_exec(pending_req_t *pending_req)
+{
+ int cmd_len = (int)pending_req->cmd_len;
+ int data_dir = (int)pending_req->sc_data_direction;
+ unsigned int timeout;
+ struct request *rq;
+ int write;
+
+ DPRINTK("%s\n",__FUNCTION__);
+
+ /* because it doesn't timeout backend earlier than frontend.*/
+ if (pending_req->timeout_per_command)
+ timeout = pending_req->timeout_per_command * HZ;
+ else
+ timeout = VSCSIIF_TIMEOUT;
+
+ write = (data_dir == DMA_TO_DEVICE);
+ if (pending_req->nr_segments) {
+ struct bio *bio = request_map_sg(pending_req);
+
+ if (IS_ERR(bio)) {
+ printk(KERN_ERR "scsiback: SG Request Map Error\n");
+ return;
+ }
+
+ rq = blk_make_request(pending_req->sdev->request_queue, bio,
+ GFP_KERNEL);
+ if (IS_ERR(rq)) {
+ printk(KERN_ERR "scsiback: Make Request Error\n");
+ return;
+ }
+
+ rq->buffer = NULL;
+ } else {
+ rq = blk_get_request(pending_req->sdev->request_queue, write,
+ GFP_KERNEL);
+ if (unlikely(!rq)) {
+ printk(KERN_ERR "scsiback: Get Request Error\n");
+ return;
+ }
+ }
+
+ rq->cmd_type = REQ_TYPE_BLOCK_PC;
+ rq->cmd_len = cmd_len;
+ memcpy(rq->cmd, pending_req->cmnd, cmd_len);
+
+ memset(pending_req->sense_buffer, 0, VSCSIIF_SENSE_BUFFERSIZE);
+ rq->sense = pending_req->sense_buffer;
+ rq->sense_len = 0;
+
+ /* not allowed to retry in backend. */
+ rq->retries = 0;
+ rq->timeout = timeout;
+ rq->end_io_data = pending_req;
+
+ scsiback_get(pending_req->info);
+ blk_execute_rq_nowait(rq->q, NULL, rq, 1, scsiback_cmd_done);
+
+ return ;
+}
+
+
+static void scsiback_device_reset_exec(pending_req_t *pending_req)
+{
+ struct vscsibk_info *info = pending_req->info;
+ int err;
+ struct scsi_device *sdev = pending_req->sdev;
+
+ scsiback_get(info);
+ err = scsi_reset_provider(sdev, SCSI_TRY_RESET_DEVICE);
+
+ scsiback_do_resp_with_sense(NULL, err, 0, pending_req);
+ scsiback_put(info);
+
+ return;
+}
+
+
+irqreturn_t scsiback_intr(int irq, void *dev_id)
+{
+ scsiback_notify_work((struct vscsibk_info *)dev_id);
+ return IRQ_HANDLED;
+}
+
+static int prepare_pending_reqs(struct vscsibk_info *info,
+ vscsiif_request_t *ring_req, pending_req_t *pending_req)
+{
+ struct scsi_device *sdev;
+ struct ids_tuple vir;
+ int err = -EINVAL;
+
+ DPRINTK("%s\n",__FUNCTION__);
+
+ pending_req->rqid = ring_req->rqid;
+ pending_req->act = ring_req->act;
+
+ pending_req->info = info;
+
+ pending_req->v_chn = vir.chn = ring_req->channel;
+ pending_req->v_tgt = vir.tgt = ring_req->id;
+ vir.lun = ring_req->lun;
+
+ rmb();
+ sdev = scsiback_do_translation(info, &vir);
+ if (!sdev) {
+ pending_req->sdev = NULL;
+ DPRINTK("scsiback: doesn't exist.\n");
+ err = -ENODEV;
+ goto invalid_value;
+ }
+ pending_req->sdev = sdev;
+
+ /* request range check from frontend */
+ pending_req->sc_data_direction = ring_req->sc_data_direction;
+ barrier();
+ if ((pending_req->sc_data_direction != DMA_BIDIRECTIONAL) &&
+ (pending_req->sc_data_direction != DMA_TO_DEVICE) &&
+ (pending_req->sc_data_direction != DMA_FROM_DEVICE) &&
+ (pending_req->sc_data_direction != DMA_NONE)) {
+ DPRINTK("scsiback: invalid parameter data_dir = %d\n",
+ pending_req->sc_data_direction);
+ err = -EINVAL;
+ goto invalid_value;
+ }
+
+ pending_req->nr_segments = ring_req->nr_segments;
+ barrier();
+ if (pending_req->nr_segments > VSCSIIF_SG_TABLESIZE) {
+ DPRINTK("scsiback: invalid parameter nr_seg = %d\n",
+ pending_req->nr_segments);
+ err = -EINVAL;
+ goto invalid_value;
+ }
+
+ pending_req->cmd_len = ring_req->cmd_len;
+ barrier();
+ if (pending_req->cmd_len > VSCSIIF_MAX_COMMAND_SIZE) {
+ DPRINTK("scsiback: invalid parameter cmd_len = %d\n",
+ pending_req->cmd_len);
+ err = -EINVAL;
+ goto invalid_value;
+ }
+ memcpy(pending_req->cmnd, ring_req->cmnd, pending_req->cmd_len);
+
+ pending_req->timeout_per_command = ring_req->timeout_per_command;
+
+ if(scsiback_gnttab_data_map(ring_req, pending_req)) {
+ DPRINTK("scsiback: invalid buffer\n");
+ err = -EINVAL;
+ goto invalid_value;
+ }
+
+ return 0;
+
+invalid_value:
+ return err;
+}
+
+
+static int scsiback_do_cmd_fn(struct vscsibk_info *info)
+{
+ struct vscsiif_back_ring *ring = &info->ring;
+ vscsiif_request_t *ring_req;
+
+ pending_req_t *pending_req;
+ RING_IDX rc, rp;
+ int err, more_to_do = 0;
+
+ DPRINTK("%s\n",__FUNCTION__);
+
+ rc = ring->req_cons;
+ rp = ring->sring->req_prod;
+ rmb();
+
+ while ((rc != rp)) {
+ if (RING_REQUEST_CONS_OVERFLOW(ring, rc))
+ break;
+ pending_req = alloc_req(info);
+ if (NULL == pending_req) {
+ more_to_do = 1;
+ break;
+ }
+
+ ring_req = RING_GET_REQUEST(ring, rc);
+ ring->req_cons = ++rc;
+
+ err = prepare_pending_reqs(info, ring_req,
+ pending_req);
+ if (err == -EINVAL) {
+ scsiback_do_resp_with_sense(NULL, (DRIVER_ERROR << 24),
+ 0, pending_req);
+ continue;
+ } else if (err == -ENODEV) {
+ scsiback_do_resp_with_sense(NULL, (DID_NO_CONNECT << 16),
+ 0, pending_req);
+ continue;
+ }
+
+ if (pending_req->act == VSCSIIF_ACT_SCSI_CDB) {
+
+ /* The Host mode is through as for Emulation. */
+ if (info->feature == VSCSI_TYPE_HOST)
+ scsiback_cmd_exec(pending_req);
+ else
+ scsiback_req_emulation_or_cmdexec(pending_req);
+
+ } else if (pending_req->act == VSCSIIF_ACT_SCSI_RESET) {
+ scsiback_device_reset_exec(pending_req);
+ } else {
+ printk(KERN_ERR "scsiback: invalid parameter for request\n");
+ scsiback_do_resp_with_sense(NULL, (DRIVER_ERROR << 24),
+ 0, pending_req);
+ continue;
+ }
+ }
+
+ if (RING_HAS_UNCONSUMED_REQUESTS(ring))
+ more_to_do = 1;
+
+ /* Yield point for this unbounded loop. */
+ cond_resched();
+
+ return more_to_do;
+}
+
+
+int scsiback_schedule(void *data)
+{
+ struct vscsibk_info *info = (struct vscsibk_info *)data;
+
+ DPRINTK("%s\n",__FUNCTION__);
+
+ while (!kthread_should_stop()) {
+ wait_event_interruptible(
+ info->wq,
+ info->waiting_reqs || kthread_should_stop());
+ wait_event_interruptible(
+ pending_free_wq,
+ !list_empty(&pending_free) || kthread_should_stop());
+
+ info->waiting_reqs = 0;
+ smp_mb();
+
+ if (scsiback_do_cmd_fn(info))
+ info->waiting_reqs = 1;
+ }
+
+ return 0;
+}
+
+
+static int __init scsiback_init(void)
+{
+ int i, mmap_pages;
+
+ if (!xen_domain())
+ return -ENODEV;
+
+ mmap_pages = vscsiif_reqs * VSCSIIF_SG_TABLESIZE;
+
+ pending_reqs = kzalloc(sizeof(pending_reqs[0]) *
+ vscsiif_reqs, GFP_KERNEL);
+ pending_grant_handles = kmalloc(sizeof(pending_grant_handles[0]) *
+ mmap_pages, GFP_KERNEL);
+ pending_pages = alloc_empty_pages_and_pagevec(mmap_pages);
+
+ if (!pending_reqs || !pending_grant_handles || !pending_pages)
+ goto out_of_memory;
+
+ for (i = 0; i < mmap_pages; i++)
+ pending_grant_handles[i] = SCSIBACK_INVALID_HANDLE;
+
+ if (scsiback_interface_init() < 0)
+ goto out_of_kmem;
+
+ INIT_LIST_HEAD(&pending_free);
+
+ for (i = 0; i < vscsiif_reqs; i++)
+ list_add_tail(&pending_reqs[i].free_list, &pending_free);
+
+ if (scsiback_xenbus_init())
+ goto out_of_xenbus;
+
+ scsiback_emulation_init();
+
+ return 0;
+
+out_of_xenbus:
+ scsiback_xenbus_unregister();
+out_of_kmem:
+ scsiback_interface_exit();
+out_of_memory:
+ kfree(pending_reqs);
+ kfree(pending_grant_handles);
+ free_empty_pages_and_pagevec(pending_pages, mmap_pages);
+ printk(KERN_ERR "scsiback: %s: out of memory\n", __FUNCTION__);
+ return -ENOMEM;
+}
+
+#if 0
+static void __exit scsiback_exit(void)
+{
+ scsiback_xenbus_unregister();
+ scsiback_interface_exit();
+ kfree(pending_reqs);
+ kfree(pending_grant_handles);
+ free_empty_pages_and_pagevec(pending_pages, (vscsiif_reqs * VSCSIIF_SG_TABLESIZE));
+
+}
+#endif
+
+module_init(scsiback_init);
+
+#if 0
+module_exit(scsiback_exit);
+#endif
+
+MODULE_DESCRIPTION("Xen SCSI backend driver");
+MODULE_LICENSE("Dual BSD/GPL");
diff -N -u -r linux-2.6-xen-orig/drivers/xen/scsiback/translate.c linux-2.6-xen-pvscsi/drivers/xen/scsiback/translate.c
--- linux-2.6-xen-orig/drivers/xen/scsiback/translate.c 1970-01-01 02:00:00.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/xen/scsiback/translate.c 2011-01-06 18:10:02.000000000 +0200
@@ -0,0 +1,168 @@
+/*
+ * Xen SCSI backend driver
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <linux/list.h>
+#include <linux/gfp.h>
+
+#include "common.h"
+
+/*
+ Initialize the translation entry list
+*/
+void scsiback_init_translation_table(struct vscsibk_info *info)
+{
+ INIT_LIST_HEAD(&info->v2p_entry_lists);
+ spin_lock_init(&info->v2p_lock);
+}
+
+
+/*
+ Add a new translation entry
+*/
+int scsiback_add_translation_entry(struct vscsibk_info *info,
+ struct scsi_device *sdev, struct ids_tuple *v)
+{
+ int err = 0;
+ struct v2p_entry *entry;
+ struct v2p_entry *new;
+ struct list_head *head = &(info->v2p_entry_lists);
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->v2p_lock, flags);
+
+ /* Check double assignment to identical virtual ID */
+ list_for_each_entry(entry, head, l) {
+ if ((entry->v.chn == v->chn) &&
+ (entry->v.tgt == v->tgt) &&
+ (entry->v.lun == v->lun)) {
+ printk(KERN_WARNING "scsiback: Virtual ID is already used. "
+ "Assignment was not performed.\n");
+ err = -EEXIST;
+ goto out;
+ }
+
+ }
+
+ /* Create a new translation entry and add to the list */
+ if ((new = kmalloc(sizeof(struct v2p_entry), GFP_ATOMIC)) == NULL) {
+ printk(KERN_ERR "scsiback: %s: kmalloc() error.\n", __FUNCTION__);
+ err = -ENOMEM;
+ goto out;
+ }
+ new->v = *v;
+ new->sdev = sdev;
+ list_add_tail(&new->l, head);
+
+out:
+ spin_unlock_irqrestore(&info->v2p_lock, flags);
+ return err;
+}
+
+
+/*
+ Delete the translation entry specfied
+*/
+int scsiback_del_translation_entry(struct vscsibk_info *info,
+ struct ids_tuple *v)
+{
+ struct v2p_entry *entry;
+ struct list_head *head = &(info->v2p_entry_lists);
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->v2p_lock, flags);
+ /* Find out the translation entry specified */
+ list_for_each_entry(entry, head, l) {
+ if ((entry->v.chn == v->chn) &&
+ (entry->v.tgt == v->tgt) &&
+ (entry->v.lun == v->lun)) {
+ goto found;
+ }
+ }
+
+ spin_unlock_irqrestore(&info->v2p_lock, flags);
+ return 1;
+
+found:
+ /* Delete the translation entry specfied */
+ scsi_device_put(entry->sdev);
+ list_del(&entry->l);
+ kfree(entry);
+
+ spin_unlock_irqrestore(&info->v2p_lock, flags);
+ return 0;
+}
+
+
+/*
+ Perform virtual to physical translation
+*/
+struct scsi_device *scsiback_do_translation(struct vscsibk_info *info,
+ struct ids_tuple *v)
+{
+ struct v2p_entry *entry;
+ struct list_head *head = &(info->v2p_entry_lists);
+ struct scsi_device *sdev = NULL;
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->v2p_lock, flags);
+ list_for_each_entry(entry, head, l) {
+ if ((entry->v.chn == v->chn) &&
+ (entry->v.tgt == v->tgt) &&
+ (entry->v.lun == v->lun)) {
+ sdev = entry->sdev;
+ goto out;
+ }
+ }
+out:
+ spin_unlock_irqrestore(&info->v2p_lock, flags);
+ return sdev;
+}
+
+
+/*
+ Release the translation entry specfied
+*/
+void scsiback_release_translation_entry(struct vscsibk_info *info)
+{
+ struct v2p_entry *entry, *tmp;
+ struct list_head *head = &(info->v2p_entry_lists);
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->v2p_lock, flags);
+ list_for_each_entry_safe(entry, tmp, head, l) {
+ scsi_device_put(entry->sdev);
+ list_del(&entry->l);
+ kfree(entry);
+ }
+
+ spin_unlock_irqrestore(&info->v2p_lock, flags);
+ return;
+
+}
diff -N -u -r linux-2.6-xen-orig/drivers/xen/scsiback/xenbus.c linux-2.6-xen-pvscsi/drivers/xen/scsiback/xenbus.c
--- linux-2.6-xen-orig/drivers/xen/scsiback/xenbus.c 1970-01-01 02:00:00.000000000 +0200
+++ linux-2.6-xen-pvscsi/drivers/xen/scsiback/xenbus.c 2011-01-06 18:10:02.000000000 +0200
@@ -0,0 +1,377 @@
+/*
+ * Xen SCSI backend driver
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * Based on the blkback driver code.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <stdarg.h>
+#include <linux/module.h>
+#include <linux/kthread.h>
+#include <scsi/scsi.h>
+#include <scsi/scsi_host.h>
+#include <scsi/scsi_device.h>
+
+#include "common.h"
+
+struct backend_info
+{
+ struct xenbus_device *dev;
+ struct vscsibk_info *info;
+};
+
+
+static int __vscsiif_name(struct backend_info *be, char *buf)
+{
+ struct xenbus_device *dev = be->dev;
+ unsigned int domid, id;
+
+ sscanf(dev->nodename, "backend/vscsi/%u/%u", &domid, &id);
+ snprintf(buf, TASK_COMM_LEN, "vscsi.%u.%u", be->info->domid, id);
+
+ return 0;
+}
+
+static int scsiback_map(struct backend_info *be)
+{
+ struct xenbus_device *dev = be->dev;
+ unsigned long ring_ref;
+ unsigned int evtchn;
+ int err;
+ char name[TASK_COMM_LEN];
+
+ err = xenbus_gather(XBT_NIL, dev->otherend,
+ "ring-ref", "%lu", &ring_ref,
+ "event-channel", "%u", &evtchn, NULL);
+ if (err) {
+ xenbus_dev_fatal(dev, err, "reading %s ring", dev->otherend);
+ return err;
+ }
+
+ err = scsiback_init_sring(be->info, ring_ref, evtchn);
+ if (err)
+ return err;
+
+ err = __vscsiif_name(be, name);
+ if (err) {
+ xenbus_dev_error(dev, err, "get scsiback dev name");
+ return err;
+ }
+
+ be->info->kthread = kthread_run(scsiback_schedule, be->info, name);
+ if (IS_ERR(be->info->kthread)) {
+ err = PTR_ERR(be->info->kthread);
+ be->info->kthread = NULL;
+ xenbus_dev_error(be->dev, err, "start vscsiif");
+ return err;
+ }
+
+ return 0;
+}
+
+
+struct scsi_device *scsiback_get_scsi_device(struct ids_tuple *phy)
+{
+ struct Scsi_Host *shost;
+ struct scsi_device *sdev = NULL;
+
+ shost = scsi_host_lookup(phy->hst);
+ if (IS_ERR(shost)) {
+ printk(KERN_ERR "scsiback: host%d doesn't exist.\n",
+ phy->hst);
+ return NULL;
+ }
+ sdev = scsi_device_lookup(shost, phy->chn, phy->tgt, phy->lun);
+ if (!sdev) {
+ printk(KERN_ERR "scsiback: %d:%d:%d:%d doesn't exist.\n",
+ phy->hst, phy->chn, phy->tgt, phy->lun);
+ scsi_host_put(shost);
+ return NULL;
+ }
+
+ scsi_host_put(shost);
+ return (sdev);
+}
+
+#define VSCSIBACK_OP_ADD_OR_DEL_LUN 1
+#define VSCSIBACK_OP_UPDATEDEV_STATE 2
+
+
+static void scsiback_do_lun_hotplug(struct backend_info *be, int op)
+{
+ int i, err = 0;
+ struct ids_tuple phy, vir;
+ int device_state;
+ char str[64], state_str[64];
+ char **dir;
+ unsigned int dir_n = 0;
+ struct xenbus_device *dev = be->dev;
+ struct scsi_device *sdev;
+
+ dir = xenbus_directory(XBT_NIL, dev->nodename, "vscsi-devs", &dir_n);
+ if (IS_ERR(dir))
+ return;
+
+ for (i = 0; i < dir_n; i++) {
+
+ /* read status */
+ snprintf(state_str, sizeof(state_str), "vscsi-devs/%s/state", dir[i]);
+ err = xenbus_scanf(XBT_NIL, dev->nodename, state_str, "%u",
+ &device_state);
+ if (XENBUS_EXIST_ERR(err))
+ continue;
+
+ /* physical SCSI device */
+ snprintf(str, sizeof(str), "vscsi-devs/%s/p-dev", dir[i]);
+ err = xenbus_scanf(XBT_NIL, dev->nodename, str,
+ "%u:%u:%u:%u", &phy.hst, &phy.chn, &phy.tgt, &phy.lun);
+ if (XENBUS_EXIST_ERR(err)) {
+ xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateClosed);
+ continue;
+ }
+
+ /* virtual SCSI device */
+ snprintf(str, sizeof(str), "vscsi-devs/%s/v-dev", dir[i]);
+ err = xenbus_scanf(XBT_NIL, dev->nodename, str,
+ "%u:%u:%u:%u", &vir.hst, &vir.chn, &vir.tgt, &vir.lun);
+ if (XENBUS_EXIST_ERR(err)) {
+ xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateClosed);
+ continue;
+ }
+
+ switch (op) {
+ case VSCSIBACK_OP_ADD_OR_DEL_LUN:
+ if (device_state == XenbusStateInitialising) {
+ sdev = scsiback_get_scsi_device(&phy);
+ if (!sdev)
+ xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateClosed);
+ else {
+ err = scsiback_add_translation_entry(be->info, sdev, &vir);
+ if (!err) {
+ if (xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateInitialised)) {
+ printk(KERN_ERR "scsiback: xenbus_printf error %s\n", state_str);
+ scsiback_del_translation_entry(be->info, &vir);
+ }
+ } else {
+ scsi_device_put(sdev);
+ xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateClosed);
+ }
+ }
+ }
+
+ if (device_state == XenbusStateClosing) {
+ if (!scsiback_del_translation_entry(be->info, &vir)) {
+ if (xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateClosed))
+ printk(KERN_ERR "scsiback: xenbus_printf error %s\n", state_str);
+ }
+ }
+ break;
+
+ case VSCSIBACK_OP_UPDATEDEV_STATE:
+ if (device_state == XenbusStateInitialised) {
+ /* modify vscsi-devs/dev-x/state */
+ if (xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateConnected)) {
+ printk(KERN_ERR "scsiback: xenbus_printf error %s\n", state_str);
+ scsiback_del_translation_entry(be->info, &vir);
+ xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateClosed);
+ }
+ }
+ break;
+ /*When it is necessary, processing is added here.*/
+ default:
+ break;
+ }
+ }
+
+ kfree(dir);
+ return ;
+}
+
+
+static void scsiback_frontend_changed(struct xenbus_device *dev,
+ enum xenbus_state frontend_state)
+{
+ struct backend_info *be = dev_get_drvdata(&dev->dev);
+ int err;
+
+ switch (frontend_state) {
+ case XenbusStateInitialising:
+ break;
+ case XenbusStateInitialised:
+ err = scsiback_map(be);
+ if (err)
+ break;
+
+ scsiback_do_lun_hotplug(be, VSCSIBACK_OP_ADD_OR_DEL_LUN);
+ xenbus_switch_state(dev, XenbusStateConnected);
+
+ break;
+ case XenbusStateConnected:
+
+ scsiback_do_lun_hotplug(be, VSCSIBACK_OP_UPDATEDEV_STATE);
+
+ if (dev->state == XenbusStateConnected)
+ break;
+
+ xenbus_switch_state(dev, XenbusStateConnected);
+
+ break;
+
+ case XenbusStateClosing:
+ scsiback_disconnect(be->info);
+ xenbus_switch_state(dev, XenbusStateClosing);
+ break;
+
+ case XenbusStateClosed:
+ xenbus_switch_state(dev, XenbusStateClosed);
+ if (xenbus_dev_is_online(dev))
+ break;
+ /* fall through if not online */
+ case XenbusStateUnknown:
+ device_unregister(&dev->dev);
+ break;
+
+ case XenbusStateReconfiguring:
+ scsiback_do_lun_hotplug(be, VSCSIBACK_OP_ADD_OR_DEL_LUN);
+
+ xenbus_switch_state(dev, XenbusStateReconfigured);
+
+ break;
+
+ default:
+ xenbus_dev_fatal(dev, -EINVAL, "saw state %d at frontend",
+ frontend_state);
+ break;
+ }
+}
+
+
+static int scsiback_remove(struct xenbus_device *dev)
+{
+ struct backend_info *be = dev_get_drvdata(&dev->dev);
+
+ if (be->info) {
+ scsiback_disconnect(be->info);
+ scsiback_release_translation_entry(be->info);
+ scsiback_free(be->info);
+ be->info = NULL;
+ }
+
+ kfree(be);
+ dev_set_drvdata(&dev->dev, NULL);
+
+ return 0;
+}
+
+
+static int scsiback_probe(struct xenbus_device *dev,
+ const struct xenbus_device_id *id)
+{
+ int err;
+ unsigned val = 0;
+
+ struct backend_info *be = kzalloc(sizeof(struct backend_info),
+ GFP_KERNEL);
+
+ DPRINTK("%p %d\n", dev, dev->otherend_id);
+
+ if (!be) {
+ xenbus_dev_fatal(dev, -ENOMEM,
+ "allocating backend structure");
+ return -ENOMEM;
+ }
+ be->dev = dev;
+ dev_set_drvdata(&dev->dev, be);
+
+ be->info = vscsibk_info_alloc(dev->otherend_id);
+ if (IS_ERR(be->info)) {
+ err = PTR_ERR(be->info);
+ be->info = NULL;
+ xenbus_dev_fatal(dev, err, "creating scsihost interface");
+ goto fail;
+ }
+
+ be->info->dev = dev;
+ be->info->irq = 0;
+ be->info->feature = 0; /*default not HOSTMODE.*/
+
+ scsiback_init_translation_table(be->info);
+
+ err = xenbus_scanf(XBT_NIL, dev->nodename,
+ "feature-host", "%d", &val);
+ if (XENBUS_EXIST_ERR(err))
+ val = 0;
+
+ if (val)
+ be->info->feature = VSCSI_TYPE_HOST;
+
+ err = xenbus_switch_state(dev, XenbusStateInitWait);
+ if (err)
+ goto fail;
+
+ return 0;
+
+
+fail:
+ printk(KERN_WARNING "scsiback: %s failed\n",__FUNCTION__);
+ scsiback_remove(dev);
+
+ return err;
+}
+
+
+static struct xenbus_device_id scsiback_ids[] = {
+ { "vscsi" },
+ { "" }
+};
+
+static struct xenbus_driver scsiback = {
+ .name = "vscsi",
+ .ids = scsiback_ids,
+ .probe = scsiback_probe,
+ .remove = scsiback_remove,
+ .otherend_changed = scsiback_frontend_changed
+};
+
+int scsiback_xenbus_init(void)
+{
+ return xenbus_register_backend(&scsiback);
+}
+
+void scsiback_xenbus_unregister(void)
+{
+ xenbus_unregister_driver(&scsiback);
+}
diff -N -u -r linux-2.6-xen-orig/include/xen/interface/grant_table.h linux-2.6-xen-pvscsi/include/xen/interface/grant_table.h
--- linux-2.6-xen-orig/include/xen/interface/grant_table.h 2011-01-06 17:23:24.000000000 +0200
+++ linux-2.6-xen-pvscsi/include/xen/interface/grant_table.h 2011-01-06 21:34:55.000000000 +0200
@@ -385,6 +385,8 @@
#define GNTST_permission_denied (-8) /* Not enough privilege for operation. */
#define GNTST_bad_page (-9) /* Specified page was invalid for op. */
#define GNTST_bad_copy_arg (-10) /* copy arguments cross page boundary */
+#define GNTST_address_too_big (-11) /* transfer page address too large. */
+#define GNTST_eagain (-12) /* Could not map at the moment. Retry. */
#define GNTTABOP_error_msgs { \
"okay", \
diff -N -u -r linux-2.6-xen-orig/include/xen/interface/io/vscsiif.h linux-2.6-xen-pvscsi/include/xen/interface/io/vscsiif.h
--- linux-2.6-xen-orig/include/xen/interface/io/vscsiif.h 1970-01-01 02:00:00.000000000 +0200
+++ linux-2.6-xen-pvscsi/include/xen/interface/io/vscsiif.h 2011-01-06 21:38:09.000000000 +0200
@@ -0,0 +1,105 @@
+/******************************************************************************
+ * vscsiif.h
+ *
+ * Based on the blkif.h code.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *
+ * Copyright(c) FUJITSU Limited 2008.
+ */
+
+#ifndef __XEN__PUBLIC_IO_SCSI_H__
+#define __XEN__PUBLIC_IO_SCSI_H__
+
+#include "ring.h"
+#include "../grant_table.h"
+
+/* command between backend and frontend */
+#define VSCSIIF_ACT_SCSI_CDB 1 /* SCSI CDB command */
+#define VSCSIIF_ACT_SCSI_ABORT 2 /* SCSI Device(Lun) Abort*/
+#define VSCSIIF_ACT_SCSI_RESET 3 /* SCSI Device(Lun) Reset*/
+
+
+#define VSCSIIF_BACK_MAX_PENDING_REQS 128
+
+/*
+ * Maximum scatter/gather segments per request.
+ *
+ * Considering balance between allocating al least 16 "vscsiif_request"
+ * structures on one page (4096bytes) and number of scatter gather
+ * needed, we decided to use 26 as a magic number.
+ */
+#define VSCSIIF_SG_TABLESIZE 26
+
+/*
+ * base on linux kernel 2.6.18
+ */
+#define VSCSIIF_MAX_COMMAND_SIZE 16
+#define VSCSIIF_SENSE_BUFFERSIZE 96
+
+
+struct vscsiif_request {
+ uint16_t rqid; /* private guest value, echoed in resp */
+ uint8_t act; /* command between backend and frontend */
+ uint8_t cmd_len;
+
+ uint8_t cmnd[VSCSIIF_MAX_COMMAND_SIZE];
+ uint16_t timeout_per_command; /* The command is issued by twice
+ the value in Backend. */
+ uint16_t channel, id, lun;
+ uint16_t padding;
+ uint8_t sc_data_direction; /* for DMA_TO_DEVICE(1)
+ DMA_FROM_DEVICE(2)
+ DMA_NONE(3) requests */
+ uint8_t nr_segments; /* Number of pieces of scatter-gather */
+
+ struct scsiif_request_segment {
+ grant_ref_t gref;
+ uint16_t offset;
+ uint16_t length;
+ } seg[VSCSIIF_SG_TABLESIZE];
+ uint32_t reserved[3];
+};
+typedef struct vscsiif_request vscsiif_request_t;
+
+struct vscsiif_response {
+ uint16_t rqid;
+ uint8_t padding;
+ uint8_t sense_len;
+ uint8_t sense_buffer[VSCSIIF_SENSE_BUFFERSIZE];
+ int32_t rslt;
+ uint32_t residual_len; /* request bufflen -
+ return the value from physical device */
+ uint32_t reserved[36];
+};
+typedef struct vscsiif_response vscsiif_response_t;
+
+DEFINE_RING_TYPES(vscsiif, struct vscsiif_request, struct vscsiif_response);
+
+
+#endif /*__XEN__PUBLIC_IO_SCSI_H__*/
+/*
+ * Local variables:
+ * mode: C
+ * c-set-style: "BSD"
+ * c-basic-offset: 4
+ * tab-width: 4
+ * indent-tabs-mode: nil
+ * End:
+ */
[-- Attachment #3: Type: text/plain, Size: 138 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
^ permalink raw reply [flat|nested] 10+ messages in thread* Re: [RFC PATCH v01] Xen PVSCSI drivers for pvops xen/stable-2.6.32.x kernel
2011-01-06 22:31 [RFC PATCH v01] Xen PVSCSI drivers for pvops xen/stable-2.6.32.x kernel Pasi Kärkkäinen
@ 2011-01-07 10:10 ` Ian Campbell
2011-01-07 10:31 ` Pasi Kärkkäinen
2011-11-30 16:23 ` Konrad Rzeszutek Wilk
1 sibling, 1 reply; 10+ messages in thread
From: Ian Campbell @ 2011-01-07 10:10 UTC (permalink / raw)
To: Pasi Kärkkäinen
Cc: Fitzhardinge, xen-devel@lists.xensource.com, Jeremy
On Thu, 2011-01-06 at 22:31 +0000, Pasi Kärkkäinen wrote:
> Hello,
>
> http://pasik.reaktio.net/xen/patches/xen-pvscsi-drivers-linux-2.6.32.27-pvops-v01.diff
>
> This is the first version of Xen PVSCSI drivers, both the scsiback backend and
> scsifront frontend, ported from Novell SLES11SP1 2.6.32 Xenlinux kernel to
> pvops xen/stable-2.6.32.x branch.
>
> At the moment it's *only* compile-tested with the latest xen/stable-2.6.32.x
> git kernel as of today (2.6.32.27), on Fedora 14 x86_64.
>
> Comments welcome.
Without wanting to sound like a broken record:
<insert same comments regarding maintainership and upstreaming as the
recent pvusb thread ;-)>
>
> I'm sure there are still things to fix in it.
> Hopefully I managed to properly fix all the differences between Xenlinux <-> pvops..
> Let me know how it goes, if you feel adventurous enough to try it :)
>
> More info: http://wiki.xensource.com/xenwiki/XenPVSCSI
>
> -- Pasi
>
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [RFC PATCH v01] Xen PVSCSI drivers for pvops xen/stable-2.6.32.x kernel
2011-01-07 10:10 ` Ian Campbell
@ 2011-01-07 10:31 ` Pasi Kärkkäinen
0 siblings, 0 replies; 10+ messages in thread
From: Pasi Kärkkäinen @ 2011-01-07 10:31 UTC (permalink / raw)
To: Ian Campbell; +Cc: Jeremy Fitzhardinge, xen-devel@lists.xensource.com
On Fri, Jan 07, 2011 at 10:10:20AM +0000, Ian Campbell wrote:
> On Thu, 2011-01-06 at 22:31 +0000, Pasi Kärkkäinen wrote:
> > Hello,
> >
> > http://pasik.reaktio.net/xen/patches/xen-pvscsi-drivers-linux-2.6.32.27-pvops-v01.diff
> >
> > This is the first version of Xen PVSCSI drivers, both the scsiback backend and
> > scsifront frontend, ported from Novell SLES11SP1 2.6.32 Xenlinux kernel to
> > pvops xen/stable-2.6.32.x branch.
> >
> > At the moment it's *only* compile-tested with the latest xen/stable-2.6.32.x
> > git kernel as of today (2.6.32.27), on Fedora 14 x86_64.
> >
> > Comments welcome.
>
> Without wanting to sound like a broken record:
>
> <insert same comments regarding maintainership and upstreaming as the
> recent pvusb thread ;-)>
>
Yeah I knew this comment was coming up, so maybe I should have
mentioned already earlier that I'm aware :)
This patch needs testing and cleaning up, and obviously upstreaming..
We'll get there :)
-- Pasi
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [RFC PATCH v01] Xen PVSCSI drivers for pvops xen/stable-2.6.32.x kernel
2011-01-06 22:31 [RFC PATCH v01] Xen PVSCSI drivers for pvops xen/stable-2.6.32.x kernel Pasi Kärkkäinen
2011-01-07 10:10 ` Ian Campbell
@ 2011-11-30 16:23 ` Konrad Rzeszutek Wilk
1 sibling, 0 replies; 10+ messages in thread
From: Konrad Rzeszutek Wilk @ 2011-11-30 16:23 UTC (permalink / raw)
To: Pasi Kärkkäinen; +Cc: Jeremy Fitzhardinge, xen-devel, Ian Campbell
[-- Attachment #1: Type: text/plain, Size: 2244 bytes --]
On Fri, Jan 07, 2011 at 12:31:21AM +0200, Pasi Kärkkäinen wrote:
> Hello,
>
> http://pasik.reaktio.net/xen/patches/xen-pvscsi-drivers-linux-2.6.32.27-pvops-v01.diff
>
> This is the first version of Xen PVSCSI drivers, both the scsiback backend and
> scsifront frontend, ported from Novell SLES11SP1 2.6.32 Xenlinux kernel to
> pvops xen/stable-2.6.32.x branch.
>
> At the moment it's *only* compile-tested with the latest xen/stable-2.6.32.x
> git kernel as of today (2.6.32.27), on Fedora 14 x86_64.
>
> Comments welcome.
>
> I'm sure there are still things to fix in it.
> Hopefully I managed to properly fix all the differences between Xenlinux <-> pvops..
> Let me know how it goes, if you feel adventurous enough to try it :)
I took a look at the patches and rebased them on top of v3.0 some time ago.
Had to fix up some things, but not much (most of the P2M API calls, and some of the
grant table modifications) and stuck the patches in:
git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen.git devel/xen-scsi.v1.0
Attached is also the full patch against v3.0 kernel. I found that it works
with my SCSI disks, but you need to use Xen 4.1 (and xm). Earlier versions
do something weird. I am not really sure who is using it
Ah, I also put the 2Tb fix in it.
drivers/scsi/Kconfig | 16 +
drivers/scsi/Makefile | 2 +
drivers/scsi/xen-scsiback/Makefile | 4 +
drivers/scsi/xen-scsiback/common.h | 187 ++++++++
drivers/scsi/xen-scsiback/emulate.c | 478 ++++++++++++++++++++
drivers/scsi/xen-scsiback/interface.c | 141 ++++++
drivers/scsi/xen-scsiback/scsiback.c | 757 ++++++++++++++++++++++++++++++++
drivers/scsi/xen-scsiback/translate.c | 168 +++++++
drivers/scsi/xen-scsiback/xenbus.c | 374 ++++++++++++++++
drivers/scsi/xen-scsifront/Makefile | 4 +
drivers/scsi/xen-scsifront/common.h | 137 ++++++
drivers/scsi/xen-scsifront/scsifront.c | 469 ++++++++++++++++++++
drivers/scsi/xen-scsifront/xenbus.c | 414 +++++++++++++++++
include/xen/interface/grant_table.h | 4 +
include/xen/interface/io/vscsiif.h | 105 +++++
15 files changed, 3260 insertions(+), 0 deletions(-)
[-- Attachment #2: pv-scsi-against.v3.0.patch --]
[-- Type: text/plain, Size: 94347 bytes --]
diff --git a/drivers/scsi/Kconfig b/drivers/scsi/Kconfig
index 8d9dae8..380e4f05 100644
--- a/drivers/scsi/Kconfig
+++ b/drivers/scsi/Kconfig
@@ -1909,6 +1909,22 @@ config SCSI_BFA_FC
To compile this driver as a module, choose M here. The module will
be called bfa.
+config XEN_SCSI_FRONTEND
+ tristate "Xen PVSCSI frontend driver"
+ depends on XEN && SCSI
+ default m
+ help
+ The SCSI frontend driver allows the kernel to access SCSI Devices
+ within another guest OS.
+
+config XEN_SCSI_BACKEND
+ tristate "Xen PVSCSI backend driver"
+ depends on XEN_BACKEND && SCSI
+ default m
+ help
+ The PVSCSI backend driver allows the kernel to export its SCSI Devices
+ to other Xen guests via a high-performance shared-memory interface.
+
endif # SCSI_LOWLEVEL
source "drivers/scsi/pcmcia/Kconfig"
diff --git a/drivers/scsi/Makefile b/drivers/scsi/Makefile
index 3c08f53..7b6f4d2 100644
--- a/drivers/scsi/Makefile
+++ b/drivers/scsi/Makefile
@@ -140,6 +140,8 @@ obj-$(CONFIG_SCSI_CXGB4_ISCSI) += libiscsi.o libiscsi_tcp.o cxgbi/
obj-$(CONFIG_SCSI_BNX2_ISCSI) += libiscsi.o bnx2i/
obj-$(CONFIG_BE2ISCSI) += libiscsi.o be2iscsi/
obj-$(CONFIG_SCSI_PMCRAID) += pmcraid.o
+obj-$(CONFIG_XEN_SCSI_FRONTEND) += xen-scsifront/
+obj-$(CONFIG_XEN_SCSI_BACKEND) += xen-scsiback/
obj-$(CONFIG_VMWARE_PVSCSI) += vmw_pvscsi.o
obj-$(CONFIG_ARM) += arm/
diff --git a/drivers/scsi/xen-scsiback/Makefile b/drivers/scsi/xen-scsiback/Makefile
new file mode 100644
index 0000000..94926dc
--- /dev/null
+++ b/drivers/scsi/xen-scsiback/Makefile
@@ -0,0 +1,4 @@
+obj-$(CONFIG_XEN_SCSI_BACKEND) := xen-scsiback.o
+
+xen-scsiback-y := interface.o scsiback.o xenbus.o translate.o emulate.o
+
diff --git a/drivers/scsi/xen-scsiback/common.h b/drivers/scsi/xen-scsiback/common.h
new file mode 100644
index 0000000..dafa79e
--- /dev/null
+++ b/drivers/scsi/xen-scsiback/common.h
@@ -0,0 +1,187 @@
+/*
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * Based on the blkback driver code.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef __SCSIIF__BACKEND__COMMON_H__
+#define __SCSIIF__BACKEND__COMMON_H__
+
+#include <linux/version.h>
+#include <linux/module.h>
+#include <linux/interrupt.h>
+#include <linux/slab.h>
+#include <linux/vmalloc.h>
+#include <linux/wait.h>
+#include <linux/sched.h>
+#include <linux/kthread.h>
+#include <linux/blkdev.h>
+#include <linux/list.h>
+#include <linux/kthread.h>
+#include <scsi/scsi.h>
+#include <scsi/scsi_cmnd.h>
+#include <scsi/scsi_host.h>
+#include <scsi/scsi_device.h>
+#include <scsi/scsi_dbg.h>
+#include <scsi/scsi_eh.h>
+#include <asm/io.h>
+#include <asm/setup.h>
+#include <asm/pgalloc.h>
+#include <asm/delay.h>
+#include <xen/evtchn.h>
+#include <asm/hypervisor.h>
+#include <xen/xen.h>
+#include <xen/events.h>
+#include <xen/grant_table.h>
+#include <xen/xenbus.h>
+#include <xen/page.h>
+#include <xen/interface/xen.h>
+#include <xen/interface/io/ring.h>
+#include <xen/interface/grant_table.h>
+#include <xen/interface/io/vscsiif.h>
+
+
+#define DPRINTK(_f, _a...) \
+ pr_debug("(file=%s, line=%d) " _f, \
+ __FILE__ , __LINE__ , ## _a )
+
+struct ids_tuple {
+ unsigned int hst; /* host */
+ unsigned int chn; /* channel */
+ unsigned int tgt; /* target */
+ unsigned int lun; /* LUN */
+};
+
+struct v2p_entry {
+ struct ids_tuple v; /* translate from */
+ struct scsi_device *sdev; /* translate to */
+ struct list_head l;
+};
+
+struct vscsibk_info {
+ struct xenbus_device *dev;
+
+ domid_t domid;
+ unsigned int evtchn;
+ unsigned int irq;
+
+ int feature;
+
+ struct vscsiif_back_ring ring;
+ void *ring_area;
+
+ spinlock_t ring_lock;
+ atomic_t nr_unreplied_reqs;
+
+ spinlock_t v2p_lock;
+ struct list_head v2p_entry_lists;
+
+ struct task_struct *kthread;
+ wait_queue_head_t waiting_to_free;
+ wait_queue_head_t wq;
+ unsigned int waiting_reqs;
+ struct page **mmap_pages;
+
+};
+
+typedef struct {
+ unsigned char act;
+ struct vscsibk_info *info;
+ struct scsi_device *sdev;
+
+ uint16_t rqid;
+
+ uint16_t v_chn, v_tgt;
+
+ uint8_t nr_segments;
+ uint8_t cmnd[VSCSIIF_MAX_COMMAND_SIZE];
+ uint8_t cmd_len;
+
+ uint8_t sc_data_direction;
+ uint16_t timeout_per_command;
+
+ uint32_t request_bufflen;
+ struct scatterlist *sgl;
+ grant_ref_t gref[VSCSIIF_SG_TABLESIZE];
+
+ int32_t rslt;
+ uint32_t resid;
+ uint8_t sense_buffer[VSCSIIF_SENSE_BUFFERSIZE];
+
+ struct list_head free_list;
+} pending_req_t;
+
+
+
+#define scsiback_get(_b) (atomic_inc(&(_b)->nr_unreplied_reqs))
+#define scsiback_put(_b) \
+ do { \
+ if (atomic_dec_and_test(&(_b)->nr_unreplied_reqs)) \
+ wake_up(&(_b)->waiting_to_free);\
+ } while (0)
+
+#define VSCSIIF_TIMEOUT (900*HZ)
+
+#define VSCSI_TYPE_HOST 1
+
+irqreturn_t scsiback_intr(int, void *);
+int scsiback_init_sring(struct vscsibk_info *info,
+ unsigned long ring_ref, unsigned int evtchn);
+int scsiback_schedule(void *data);
+
+
+struct vscsibk_info *vscsibk_info_alloc(domid_t domid);
+void scsiback_free(struct vscsibk_info *info);
+void scsiback_disconnect(struct vscsibk_info *info);
+int __init scsiback_interface_init(void);
+void scsiback_interface_exit(void);
+int scsiback_xenbus_init(void);
+void scsiback_xenbus_unregister(void);
+
+void scsiback_init_translation_table(struct vscsibk_info *info);
+
+int scsiback_add_translation_entry(struct vscsibk_info *info,
+ struct scsi_device *sdev, struct ids_tuple *v);
+
+int scsiback_del_translation_entry(struct vscsibk_info *info,
+ struct ids_tuple *v);
+struct scsi_device *scsiback_do_translation(struct vscsibk_info *info,
+ struct ids_tuple *v);
+void scsiback_release_translation_entry(struct vscsibk_info *info);
+
+
+void scsiback_cmd_exec(pending_req_t *pending_req);
+void scsiback_do_resp_with_sense(char *sense_buffer, int32_t result,
+ uint32_t resid, pending_req_t *pending_req);
+void scsiback_fast_flush_area(pending_req_t *req);
+
+void scsiback_rsp_emulation(pending_req_t *pending_req);
+void scsiback_req_emulation_or_cmdexec(pending_req_t *pending_req);
+void scsiback_emulation_init(void);
+
+
+#endif /* __SCSIIF__BACKEND__COMMON_H__ */
diff --git a/drivers/scsi/xen-scsiback/emulate.c b/drivers/scsi/xen-scsiback/emulate.c
new file mode 100644
index 0000000..c5b0999
--- /dev/null
+++ b/drivers/scsi/xen-scsiback/emulate.c
@@ -0,0 +1,478 @@
+/*
+ * Xen SCSI backend driver
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+/*
+* Patched to support >2TB drives + allow tape & autoloader operations
+* 2010, Samuel Kvasnica, IMS Nanofabrication AG
+*/
+
+#include <scsi/scsi.h>
+#include <scsi/scsi_cmnd.h>
+#include <scsi/scsi_device.h>
+#include "common.h"
+
+/* Following SCSI commands are not defined in scsi/scsi.h */
+#define EXTENDED_COPY 0x83 /* EXTENDED COPY command */
+#define REPORT_ALIASES 0xa3 /* REPORT ALIASES command */
+#define CHANGE_ALIASES 0xa4 /* CHANGE ALIASES command */
+#define SET_PRIORITY 0xa4 /* SET PRIORITY command */
+
+
+/*
+ The bitmap in order to control emulation.
+ (Bit 3 to 7 are reserved for future use.)
+*/
+#define VSCSIIF_NEED_CMD_EXEC 0x01 /* If this bit is set, cmd exec */
+ /* is required. */
+#define VSCSIIF_NEED_EMULATE_REQBUF 0x02 /* If this bit is set, need */
+ /* emulation reqest buff before */
+ /* cmd exec. */
+#define VSCSIIF_NEED_EMULATE_RSPBUF 0x04 /* If this bit is set, need */
+ /* emulation resp buff after */
+ /* cmd exec. */
+
+/* Additional Sense Code (ASC) used */
+#define NO_ADDITIONAL_SENSE 0x0
+#define LOGICAL_UNIT_NOT_READY 0x4
+#define UNRECOVERED_READ_ERR 0x11
+#define PARAMETER_LIST_LENGTH_ERR 0x1a
+#define INVALID_OPCODE 0x20
+#define ADDR_OUT_OF_RANGE 0x21
+#define INVALID_FIELD_IN_CDB 0x24
+#define INVALID_FIELD_IN_PARAM_LIST 0x26
+#define POWERON_RESET 0x29
+#define SAVING_PARAMS_UNSUP 0x39
+#define THRESHOLD_EXCEEDED 0x5d
+#define LOW_POWER_COND_ON 0x5e
+
+
+
+/* Number os SCSI op_code */
+#define VSCSI_MAX_SCSI_OP_CODE 256
+static unsigned char bitmap[VSCSI_MAX_SCSI_OP_CODE];
+
+#define NO_EMULATE(cmd) \
+ bitmap[cmd] = VSCSIIF_NEED_CMD_EXEC; \
+ pre_function[cmd] = NULL; \
+ post_function[cmd] = NULL
+
+
+
+/*
+ Emulation routines for each SCSI op_code.
+*/
+static void (*pre_function[VSCSI_MAX_SCSI_OP_CODE])(pending_req_t *, void *);
+static void (*post_function[VSCSI_MAX_SCSI_OP_CODE])(pending_req_t *, void *);
+
+
+static const int check_condition_result =
+ (DRIVER_SENSE << 24) | SAM_STAT_CHECK_CONDITION;
+
+static void scsiback_mk_sense_buffer(uint8_t *data, uint8_t key,
+ uint8_t asc, uint8_t asq)
+{
+ data[0] = 0x70; /* fixed, current */
+ data[2] = key;
+ data[7] = 0xa; /* implies 18 byte sense buffer */
+ data[12] = asc;
+ data[13] = asq;
+}
+
+static void resp_not_supported_cmd(pending_req_t *pending_req, void *data)
+{
+ scsiback_mk_sense_buffer(pending_req->sense_buffer, ILLEGAL_REQUEST,
+ INVALID_OPCODE, 0);
+ pending_req->resid = 0;
+ pending_req->rslt = check_condition_result;
+}
+
+
+static int __copy_to_sg(struct scatterlist *sgl, unsigned int nr_sg,
+ void *buf, unsigned int buflen)
+{
+ struct scatterlist *sg;
+ void *from = buf;
+ void *to;
+ unsigned int from_rest = buflen;
+ unsigned int to_capa;
+ unsigned int copy_size = 0;
+ unsigned int i;
+ unsigned long pfn;
+
+ for_each_sg (sgl, sg, nr_sg, i) {
+ if (sg_page(sg) == NULL) {
+ printk(KERN_WARNING "%s: inconsistent length field in "
+ "scatterlist\n", __FUNCTION__);
+ return -ENOMEM;
+ }
+
+ to_capa = sg->length;
+ copy_size = min_t(unsigned int, to_capa, from_rest);
+
+ pfn = page_to_pfn(sg_page(sg));
+ to = pfn_to_kaddr(pfn) + (sg->offset);
+ memcpy(to, from, copy_size);
+
+ from_rest -= copy_size;
+ if (from_rest == 0) {
+ return 0;
+ }
+
+ from += copy_size;
+ }
+
+ printk(KERN_WARNING "%s: no space in scatterlist\n",
+ __FUNCTION__);
+ return -ENOMEM;
+}
+#if 0
+static int __copy_from_sg(struct scatterlist *sgl, unsigned int nr_sg,
+ void *buf, unsigned int buflen)
+{
+ struct scatterlist *sg;
+ void *from;
+ void *to = buf;
+ unsigned int from_rest;
+ unsigned int to_capa = buflen;
+ unsigned int copy_size;
+ unsigned int i;
+ unsigned long pfn;
+
+ for_each_sg (sgl, sg, nr_sg, i) {
+ if (sg_page(sg) == NULL) {
+ printk(KERN_WARNING "%s: inconsistent length field in "
+ "scatterlist\n", __FUNCTION__);
+ return -ENOMEM;
+ }
+
+ from_rest = sg->length;
+ if ((from_rest > 0) && (to_capa < from_rest)) {
+ printk(KERN_WARNING
+ "%s: no space in destination buffer\n",
+ __FUNCTION__);
+ return -ENOMEM;
+ }
+ copy_size = from_rest;
+
+ pfn = page_to_pfn(sg_page(sg));
+ from = pfn_to_kaddr(pfn) + (sg->offset);
+ memcpy(to, from, copy_size);
+
+ to_capa -= copy_size;
+ to += copy_size;
+ }
+
+ return 0;
+}
+#endif
+static int __nr_luns_under_host(struct vscsibk_info *info)
+{
+ struct v2p_entry *entry;
+ struct list_head *head = &(info->v2p_entry_lists);
+ unsigned long flags;
+ int lun_cnt = 0;
+
+ spin_lock_irqsave(&info->v2p_lock, flags);
+ list_for_each_entry(entry, head, l) {
+ lun_cnt++;
+ }
+ spin_unlock_irqrestore(&info->v2p_lock, flags);
+
+ return (lun_cnt);
+}
+
+
+/* REPORT LUNS Define*/
+#define VSCSI_REPORT_LUNS_HEADER 8
+#define VSCSI_REPORT_LUNS_RETRY 3
+
+/* quoted scsi_debug.c/resp_report_luns() */
+static void __report_luns(pending_req_t *pending_req, void *data)
+{
+ struct vscsibk_info *info = pending_req->info;
+ unsigned int channel = pending_req->v_chn;
+ unsigned int target = pending_req->v_tgt;
+ unsigned int nr_seg = pending_req->nr_segments;
+ unsigned char *cmd = (unsigned char *)pending_req->cmnd;
+
+ unsigned char *buff = NULL;
+ unsigned char alloc_len;
+ unsigned int alloc_luns = 0;
+ unsigned int req_bufflen = 0;
+ unsigned int actual_len = 0;
+ unsigned int retry_cnt = 0;
+ int select_report = (int)cmd[2];
+ int i, lun_cnt = 0, lun, upper, err = 0;
+
+ struct v2p_entry *entry;
+ struct list_head *head = &(info->v2p_entry_lists);
+ unsigned long flags;
+
+ struct scsi_lun *one_lun;
+
+ req_bufflen = cmd[9] + (cmd[8] << 8) + (cmd[7] << 16) + (cmd[6] << 24);
+ if ((req_bufflen < 4) || (select_report != 0))
+ goto fail;
+
+ alloc_luns = __nr_luns_under_host(info);
+ alloc_len = sizeof(struct scsi_lun) * alloc_luns
+ + VSCSI_REPORT_LUNS_HEADER;
+retry:
+ if ((buff = kzalloc(alloc_len, GFP_KERNEL)) == NULL) {
+ printk(KERN_ERR "scsiback:%s kmalloc err\n", __FUNCTION__);
+ goto fail;
+ }
+
+ one_lun = (struct scsi_lun *) &buff[8];
+ spin_lock_irqsave(&info->v2p_lock, flags);
+ list_for_each_entry(entry, head, l) {
+ if ((entry->v.chn == channel) &&
+ (entry->v.tgt == target)) {
+ /* check overflow */
+ if (lun_cnt >= alloc_luns) {
+ spin_unlock_irqrestore(&info->v2p_lock,
+ flags);
+
+ if (retry_cnt < VSCSI_REPORT_LUNS_RETRY) {
+ retry_cnt++;
+ if (buff)
+ kfree(buff);
+ goto retry;
+ }
+
+ goto fail;
+ }
+
+ lun = entry->v.lun;
+ upper = (lun >> 8) & 0x3f;
+ if (upper)
+ one_lun[lun_cnt].scsi_lun[0] = upper;
+ one_lun[lun_cnt].scsi_lun[1] = lun & 0xff;
+ lun_cnt++;
+ }
+ }
+
+ spin_unlock_irqrestore(&info->v2p_lock, flags);
+
+ buff[2] = ((sizeof(struct scsi_lun) * lun_cnt) >> 8) & 0xff;
+ buff[3] = (sizeof(struct scsi_lun) * lun_cnt) & 0xff;
+
+ actual_len = lun_cnt * sizeof(struct scsi_lun)
+ + VSCSI_REPORT_LUNS_HEADER;
+ req_bufflen = 0;
+ for (i = 0; i < nr_seg; i++)
+ req_bufflen += pending_req->sgl[i].length;
+
+ err = __copy_to_sg(pending_req->sgl, nr_seg, buff,
+ min(req_bufflen, actual_len));
+ if (err)
+ goto fail;
+
+ memset(pending_req->sense_buffer, 0, VSCSIIF_SENSE_BUFFERSIZE);
+ pending_req->rslt = 0x00;
+ pending_req->resid = req_bufflen - min(req_bufflen, actual_len);
+
+ kfree(buff);
+ return;
+
+fail:
+ scsiback_mk_sense_buffer(pending_req->sense_buffer, ILLEGAL_REQUEST,
+ INVALID_FIELD_IN_CDB, 0);
+ pending_req->rslt = check_condition_result;
+ pending_req->resid = 0;
+ if (buff)
+ kfree(buff);
+ return;
+}
+
+
+
+int __pre_do_emulation(pending_req_t *pending_req, void *data)
+{
+ uint8_t op_code = pending_req->cmnd[0];
+
+ if ((bitmap[op_code] & VSCSIIF_NEED_EMULATE_REQBUF) &&
+ pre_function[op_code] != NULL) {
+ pre_function[op_code](pending_req, data);
+ }
+
+ /*
+ 0: no need for native driver call, so should return immediately.
+ 1: non emulation or should call native driver
+ after modifing the request buffer.
+ */
+ return !!(bitmap[op_code] & VSCSIIF_NEED_CMD_EXEC);
+}
+
+void scsiback_rsp_emulation(pending_req_t *pending_req)
+{
+ uint8_t op_code = pending_req->cmnd[0];
+
+ if ((bitmap[op_code] & VSCSIIF_NEED_EMULATE_RSPBUF) &&
+ post_function[op_code] != NULL) {
+ post_function[op_code](pending_req, NULL);
+ }
+
+ return;
+}
+
+
+void scsiback_req_emulation_or_cmdexec(pending_req_t *pending_req)
+{
+ if (__pre_do_emulation(pending_req, NULL)) {
+ scsiback_cmd_exec(pending_req);
+ }
+ else {
+ scsiback_fast_flush_area(pending_req);
+ scsiback_do_resp_with_sense(pending_req->sense_buffer,
+ pending_req->rslt, pending_req->resid, pending_req);
+ }
+}
+
+
+/*
+ Following are not customizable functions.
+*/
+void scsiback_emulation_init(void)
+{
+ int i;
+
+ /* Initialize to default state */
+ for (i = 0; i < VSCSI_MAX_SCSI_OP_CODE; i++) {
+ bitmap[i] = (VSCSIIF_NEED_EMULATE_REQBUF |
+ VSCSIIF_NEED_EMULATE_RSPBUF);
+ pre_function[i] = resp_not_supported_cmd;
+ post_function[i] = NULL;
+ /* means,
+ - no need for pre-emulation
+ - no need for post-emulation
+ - call native driver
+ */
+ }
+
+ /*
+ Register appropriate functions below as you need.
+ (See scsi/scsi.h for definition of SCSI op_code.)
+ */
+
+ /*
+ Following commands do not require emulation.
+ */
+ NO_EMULATE(TEST_UNIT_READY); /*0x00*/ /* sd,st */
+ NO_EMULATE(REZERO_UNIT); /*0x01*/ /* st */
+ NO_EMULATE(REQUEST_SENSE); /*0x03*/
+ NO_EMULATE(FORMAT_UNIT); /*0x04*/
+ NO_EMULATE(READ_BLOCK_LIMITS); /*0x05*/ /* st */
+ /*NO_EMULATE(REASSIGN_BLOCKS); *//*0x07*/
+ NO_EMULATE(INITIALIZE_ELEMENT_STATUS); /*0x07*/ /* ch */
+ NO_EMULATE(READ_6); /*0x08*/ /* sd,st */
+ NO_EMULATE(WRITE_6); /*0x0a*/ /* sd,st */
+ NO_EMULATE(SEEK_6); /*0x0b*/
+ /*NO_EMULATE(READ_REVERSE); *//*0x0f*/
+ NO_EMULATE(WRITE_FILEMARKS); /*0x10*/ /* st */
+ NO_EMULATE(SPACE); /*0x11*/ /* st */
+ NO_EMULATE(INQUIRY); /*0x12*/
+ /*NO_EMULATE(RECOVER_BUFFERED_DATA); *//*0x14*/
+ NO_EMULATE(MODE_SELECT); /*0x15*/ /* st */
+ /*NO_EMULATE(RESERVE); *//*0x16*/
+ /*NO_EMULATE(RELEASE); *//*0x17*/
+ /*NO_EMULATE(COPY); *//*0x18*/
+ NO_EMULATE(ERASE); /*0x19*/ /* st */
+ NO_EMULATE(MODE_SENSE); /*0x1a*/ /* st */
+ NO_EMULATE(START_STOP); /*0x1b*/ /* sd,st */
+ NO_EMULATE(RECEIVE_DIAGNOSTIC); /*0x1c*/
+ NO_EMULATE(SEND_DIAGNOSTIC); /*0x1d*/
+ NO_EMULATE(ALLOW_MEDIUM_REMOVAL); /*0x1e*/
+
+ /*NO_EMULATE(SET_WINDOW); *//*0x24*/
+ NO_EMULATE(READ_CAPACITY); /*0x25*/ /* sd */
+ NO_EMULATE(READ_10); /*0x28*/ /* sd */
+ NO_EMULATE(WRITE_10); /*0x2a*/ /* sd */
+ NO_EMULATE(SEEK_10); /*0x2b*/ /* st */
+ NO_EMULATE(POSITION_TO_ELEMENT); /*0x2b*/ /* ch */
+ /*NO_EMULATE(WRITE_VERIFY); *//*0x2e*/
+ /*NO_EMULATE(VERIFY); *//*0x2f*/
+ /*NO_EMULATE(SEARCH_HIGH); *//*0x30*/
+ /*NO_EMULATE(SEARCH_EQUAL); *//*0x31*/
+ /*NO_EMULATE(SEARCH_LOW); *//*0x32*/
+ NO_EMULATE(SET_LIMITS); /*0x33*/
+ NO_EMULATE(PRE_FETCH); /*0x34*/ /* st! */
+ NO_EMULATE(READ_POSITION); /*0x34*/ /* st */
+ NO_EMULATE(SYNCHRONIZE_CACHE); /*0x35*/ /* sd */
+ NO_EMULATE(LOCK_UNLOCK_CACHE); /*0x36*/
+ NO_EMULATE(READ_DEFECT_DATA); /*0x37*/
+ NO_EMULATE(MEDIUM_SCAN); /*0x38*/
+ /*NO_EMULATE(COMPARE); *//*0x39*/
+ /*NO_EMULATE(COPY_VERIFY); *//*0x3a*/
+ NO_EMULATE(WRITE_BUFFER); /*0x3b*/
+ NO_EMULATE(READ_BUFFER); /*0x3c*/ /* osst */
+ /*NO_EMULATE(UPDATE_BLOCK); *//*0x3d*/
+ /*NO_EMULATE(READ_LONG); *//*0x3e*/
+ /*NO_EMULATE(WRITE_LONG); *//*0x3f*/
+ /*NO_EMULATE(CHANGE_DEFINITION); *//*0x40*/
+ /*NO_EMULATE(WRITE_SAME); *//*0x41*/
+ NO_EMULATE(READ_TOC); /*0x43*/ /* sr */
+ NO_EMULATE(LOG_SELECT); /*0x4c*/
+ NO_EMULATE(LOG_SENSE); /*0x4d*/ /* st! */
+ /*NO_EMULATE(MODE_SELECT_10); *//*0x55*/
+ /*NO_EMULATE(RESERVE_10); *//*0x56*/
+ /*NO_EMULATE(RELEASE_10); *//*0x57*/
+ NO_EMULATE(MODE_SENSE_10); /*0x5a*/ /* scsi_lib */
+ /*NO_EMULATE(PERSISTENT_RESERVE_IN); *//*0x5e*/
+ /*NO_EMULATE(PERSISTENT_RESERVE_OUT); *//*0x5f*/
+ /* REPORT_LUNS *//*0xa0*//*Full emulaiton*/
+ NO_EMULATE(MAINTENANCE_IN); /*0xa3*/ /* IFT alua */
+ NO_EMULATE(MAINTENANCE_OUT); /*0xa4*/ /* IFT alua */
+ NO_EMULATE(MOVE_MEDIUM); /*0xa5*/ /* ch */
+ NO_EMULATE(EXCHANGE_MEDIUM); /*0xa6*/ /* ch */
+ /*NO_EMULATE(READ_12); *//*0xa8*/
+ /*NO_EMULATE(WRITE_12); *//*0xaa*/
+ /*NO_EMULATE(WRITE_VERIFY_12); *//*0xae*/
+ /*NO_EMULATE(SEARCH_HIGH_12); *//*0xb0*/
+ /*NO_EMULATE(SEARCH_EQUAL_12); *//*0xb1*/
+ /*NO_EMULATE(SEARCH_LOW_12); *//*0xb2*/
+ NO_EMULATE(READ_ELEMENT_STATUS); /*0xb8*/ /* ch */
+ NO_EMULATE(SEND_VOLUME_TAG); /*0xb6*/ /* ch */
+ /*NO_EMULATE(WRITE_LONG_2); *//*0xea*/
+ NO_EMULATE(READ_16); /*0x88*/ /* sd >2TB */
+ NO_EMULATE(WRITE_16); /*0x8a*/ /* sd >2TB */
+ NO_EMULATE(VERIFY_16); /*0x8f*/
+ NO_EMULATE(SERVICE_ACTION_IN); /*0x9e*/ /* sd >2TB */
+
+/* st: QFA_REQUEST_BLOCK, QFA_SEEK_BLOCK might be needed ? */
+ /*
+ Following commands require emulation.
+ */
+ pre_function[REPORT_LUNS] = __report_luns;
+ bitmap[REPORT_LUNS] = (VSCSIIF_NEED_EMULATE_REQBUF |
+ VSCSIIF_NEED_EMULATE_RSPBUF);
+
+ return;
+}
diff --git a/drivers/scsi/xen-scsiback/interface.c b/drivers/scsi/xen-scsiback/interface.c
new file mode 100644
index 0000000..663568e
--- /dev/null
+++ b/drivers/scsi/xen-scsiback/interface.c
@@ -0,0 +1,141 @@
+/*
+ * interface management.
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * Based on the blkback driver code.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <scsi/scsi.h>
+#include <scsi/scsi_host.h>
+#include <scsi/scsi_device.h>
+#include "common.h"
+
+#include <xen/evtchn.h>
+#include <linux/kthread.h>
+#include <linux/delay.h>
+
+
+static struct kmem_cache *scsiback_cachep;
+
+struct vscsibk_info *vscsibk_info_alloc(domid_t domid)
+{
+ struct vscsibk_info *info;
+
+ info = kmem_cache_zalloc(scsiback_cachep, GFP_KERNEL);
+ if (!info)
+ return ERR_PTR(-ENOMEM);
+
+ info->domid = domid;
+ spin_lock_init(&info->ring_lock);
+ atomic_set(&info->nr_unreplied_reqs, 0);
+ init_waitqueue_head(&info->wq);
+ init_waitqueue_head(&info->waiting_to_free);
+
+ return info;
+}
+
+int scsiback_init_sring(struct vscsibk_info *info,
+ unsigned long ring_ref, unsigned int evtchn)
+{
+ struct vscsiif_sring *sring;
+ int err;
+
+ if (!info)
+ return -ENODEV;
+
+ if (info->irq) {
+ printk(KERN_ERR "scsiback: Already connected through?\n");
+ return -1;
+ }
+
+ err = xenbus_map_ring_valloc(info->dev, ring_ref, &info->ring_area);
+ if (err < 0)
+ return -ENOMEM;
+
+ sring = (struct vscsiif_sring *) info->ring_area;
+ BACK_RING_INIT(&info->ring, sring, PAGE_SIZE);
+
+ err = bind_interdomain_evtchn_to_irqhandler(
+ info->domid, evtchn,
+ scsiback_intr, 0, "vscsiif-backend", info);
+ if (err < 0)
+ goto unmap_page;
+
+ info->irq = err;
+
+ return 0;
+
+unmap_page:
+ xenbus_unmap_ring_vfree(info->dev, info->ring_area);
+
+ return err;
+}
+
+void scsiback_disconnect(struct vscsibk_info *info)
+{
+ if (info->kthread) {
+ kthread_stop(info->kthread);
+ info->kthread = NULL;
+ }
+
+ wait_event(info->waiting_to_free,
+ atomic_read(&info->nr_unreplied_reqs) == 0);
+
+ if (info->irq) {
+ unbind_from_irqhandler(info->irq, info);
+ info->irq = 0;
+ }
+
+ if (info->ring.sring || info->ring_area) {
+ xenbus_unmap_ring_vfree(info->dev, info->ring_area);
+ info->ring.sring = NULL;
+ info->ring_area = NULL;
+ }
+}
+
+void scsiback_free(struct vscsibk_info *info)
+{
+ kmem_cache_free(scsiback_cachep, info);
+}
+
+int __init scsiback_interface_init(void)
+{
+ scsiback_cachep = kmem_cache_create("vscsiif_cache",
+ sizeof(struct vscsibk_info), 0, 0, NULL);
+ if (!scsiback_cachep) {
+ printk(KERN_ERR "scsiback: can't init scsi cache\n");
+ return -ENOMEM;
+ }
+
+ return 0;
+}
+
+void scsiback_interface_exit(void)
+{
+ kmem_cache_destroy(scsiback_cachep);
+}
diff --git a/drivers/scsi/xen-scsiback/scsiback.c b/drivers/scsi/xen-scsiback/scsiback.c
new file mode 100644
index 0000000..a209f87
--- /dev/null
+++ b/drivers/scsi/xen-scsiback/scsiback.c
@@ -0,0 +1,757 @@
+/*
+ * Xen SCSI backend driver
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * Based on the blkback driver code.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <linux/spinlock.h>
+#include <linux/kthread.h>
+#include <linux/list.h>
+#include <linux/delay.h>
+#include <xen/balloon.h>
+#include <asm/hypervisor.h>
+#include <scsi/scsi.h>
+#include <scsi/scsi_cmnd.h>
+#include <scsi/scsi_host.h>
+#include <scsi/scsi_device.h>
+#include <scsi/scsi_dbg.h>
+#include <scsi/scsi_eh.h>
+
+#include "common.h"
+
+
+struct list_head pending_free;
+DEFINE_SPINLOCK(pending_free_lock);
+DECLARE_WAIT_QUEUE_HEAD(pending_free_wq);
+
+int vscsiif_reqs = VSCSIIF_BACK_MAX_PENDING_REQS;
+module_param_named(reqs, vscsiif_reqs, int, 0);
+MODULE_PARM_DESC(reqs, "Number of scsiback requests to allocate");
+
+static unsigned int log_print_stat;
+module_param(log_print_stat, int, 0644);
+
+#define SCSIBACK_INVALID_HANDLE (~0)
+
+static pending_req_t *pending_reqs;
+static struct page **pending_pages;
+static grant_handle_t *pending_grant_handles;
+
+static int vaddr_pagenr(pending_req_t *req, int seg)
+{
+ return (req - pending_reqs) * VSCSIIF_SG_TABLESIZE + seg;
+}
+
+static unsigned long vaddr(pending_req_t *req, int seg)
+{
+ unsigned long pfn = page_to_pfn(pending_pages[vaddr_pagenr(req, seg)]);
+ return (unsigned long)pfn_to_kaddr(pfn);
+}
+
+#define pending_handle(_req, _seg) \
+ (pending_grant_handles[vaddr_pagenr(_req, _seg)])
+
+
+void scsiback_fast_flush_area(pending_req_t *req)
+{
+ struct gnttab_unmap_grant_ref unmap[VSCSIIF_SG_TABLESIZE];
+ unsigned int i, invcount = 0;
+ grant_handle_t handle;
+ int err;
+
+ if (req->nr_segments) {
+ for (i = 0; i < req->nr_segments; i++) {
+ handle = pending_handle(req, i);
+ if (handle == SCSIBACK_INVALID_HANDLE)
+ continue;
+ gnttab_set_unmap_op(&unmap[i], vaddr(req, i),
+ GNTMAP_host_map, handle);
+ pending_handle(req, i) = SCSIBACK_INVALID_HANDLE;
+ invcount++;
+ }
+
+ err = HYPERVISOR_grant_table_op(
+ GNTTABOP_unmap_grant_ref, unmap, invcount);
+ BUG_ON(err);
+ for (i = 0; i <invcount; i++) {
+ err = m2p_remove_override(
+ virt_to_page(unmap[i].host_addr), false);
+ WARN_ON(err);
+ }
+ kfree(req->sgl);
+ }
+
+ return;
+}
+
+
+static pending_req_t * alloc_req(struct vscsibk_info *info)
+{
+ pending_req_t *req = NULL;
+ unsigned long flags;
+
+ spin_lock_irqsave(&pending_free_lock, flags);
+ if (!list_empty(&pending_free)) {
+ req = list_entry(pending_free.next, pending_req_t, free_list);
+ list_del(&req->free_list);
+ }
+ spin_unlock_irqrestore(&pending_free_lock, flags);
+ return req;
+}
+
+
+static void free_req(pending_req_t *req)
+{
+ unsigned long flags;
+ int was_empty;
+
+ spin_lock_irqsave(&pending_free_lock, flags);
+ was_empty = list_empty(&pending_free);
+ list_add(&req->free_list, &pending_free);
+ spin_unlock_irqrestore(&pending_free_lock, flags);
+ if (was_empty)
+ wake_up(&pending_free_wq);
+}
+
+
+static void scsiback_notify_work(struct vscsibk_info *info)
+{
+ info->waiting_reqs = 1;
+ wake_up(&info->wq);
+}
+
+void scsiback_do_resp_with_sense(char *sense_buffer, int32_t result,
+ uint32_t resid, pending_req_t *pending_req)
+{
+ vscsiif_response_t *ring_res;
+ struct vscsibk_info *info = pending_req->info;
+ int notify;
+ int more_to_do = 1;
+ struct scsi_sense_hdr sshdr;
+ unsigned long flags;
+
+ DPRINTK("%s\n",__FUNCTION__);
+
+ spin_lock_irqsave(&info->ring_lock, flags);
+
+ ring_res = RING_GET_RESPONSE(&info->ring, info->ring.rsp_prod_pvt);
+ info->ring.rsp_prod_pvt++;
+
+ ring_res->rslt = result;
+ ring_res->rqid = pending_req->rqid;
+
+ if (sense_buffer != NULL) {
+ if (scsi_normalize_sense(sense_buffer,
+ sizeof(sense_buffer), &sshdr)) {
+
+ int len = 8 + sense_buffer[7];
+
+ if (len > VSCSIIF_SENSE_BUFFERSIZE)
+ len = VSCSIIF_SENSE_BUFFERSIZE;
+
+ memcpy(ring_res->sense_buffer, sense_buffer, len);
+ ring_res->sense_len = len;
+ }
+ } else {
+ ring_res->sense_len = 0;
+ }
+
+ ring_res->residual_len = resid;
+
+ RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&info->ring, notify);
+ if (info->ring.rsp_prod_pvt == info->ring.req_cons) {
+ RING_FINAL_CHECK_FOR_REQUESTS(&info->ring, more_to_do);
+ } else if (RING_HAS_UNCONSUMED_REQUESTS(&info->ring)) {
+ more_to_do = 1;
+ }
+
+ spin_unlock_irqrestore(&info->ring_lock, flags);
+
+ if (more_to_do)
+ scsiback_notify_work(info);
+
+ if (notify)
+ notify_remote_via_irq(info->irq);
+
+ free_req(pending_req);
+}
+
+static void scsiback_print_status(char *sense_buffer, int errors,
+ pending_req_t *pending_req)
+{
+ struct scsi_device *sdev = pending_req->sdev;
+
+ printk(KERN_ERR "scsiback: %d:%d:%d:%d ",sdev->host->host_no,
+ sdev->channel, sdev->id, sdev->lun);
+ printk(KERN_ERR "status = 0x%02x, message = 0x%02x, host = 0x%02x, driver = 0x%02x\n",
+ status_byte(errors), msg_byte(errors),
+ host_byte(errors), driver_byte(errors));
+
+ printk(KERN_ERR "scsiback: cmnd[0]=0x%02X\n",
+ pending_req->cmnd[0]);
+
+ if (CHECK_CONDITION & status_byte(errors))
+ __scsi_print_sense("scsiback", sense_buffer, SCSI_SENSE_BUFFERSIZE);
+}
+
+
+static void scsiback_cmd_done(struct request *req, int uptodate)
+{
+ pending_req_t *pending_req = req->end_io_data;
+ unsigned char *sense_buffer;
+ unsigned int resid;
+ int errors;
+
+ sense_buffer = req->sense;
+ resid = blk_rq_bytes(req);
+ errors = req->errors;
+
+ if (errors != 0) {
+ if (log_print_stat)
+ scsiback_print_status(sense_buffer, errors, pending_req);
+ }
+
+ /* The Host mode is through as for Emulation. */
+ if (pending_req->info->feature != VSCSI_TYPE_HOST)
+ scsiback_rsp_emulation(pending_req);
+
+ scsiback_fast_flush_area(pending_req);
+ scsiback_do_resp_with_sense(sense_buffer, errors, resid, pending_req);
+ scsiback_put(pending_req->info);
+
+ __blk_put_request(req->q, req);
+}
+
+
+static int scsiback_gnttab_data_map(vscsiif_request_t *ring_req,
+ pending_req_t *pending_req)
+{
+ u32 flags;
+ int write;
+ int i, err = 0;
+ unsigned int data_len = 0;
+ struct gnttab_map_grant_ref map[VSCSIIF_SG_TABLESIZE];
+ struct vscsibk_info *info = pending_req->info;
+
+ int data_dir = pending_req->sc_data_direction;
+ unsigned int nr_segments = (unsigned int)pending_req->nr_segments;
+
+ write = (data_dir == DMA_TO_DEVICE);
+
+ if (nr_segments) {
+ struct scatterlist *sg;
+
+ /* free of (sgl) in fast_flush_area()*/
+ pending_req->sgl = kmalloc(sizeof(struct scatterlist) * nr_segments,
+ GFP_KERNEL);
+ if (!pending_req->sgl) {
+ printk(KERN_ERR "scsiback: %s: kmalloc() error.\n", __FUNCTION__);
+ return -ENOMEM;
+ }
+
+ sg_init_table(pending_req->sgl, nr_segments);
+
+ for (i = 0; i < nr_segments; i++) {
+ flags = GNTMAP_host_map;
+ if (write)
+ flags |= GNTMAP_readonly;
+
+ gnttab_set_map_op(&map[i], vaddr(pending_req, i), flags,
+ ring_req->seg[i].gref,
+ info->domid);
+ }
+
+ err = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, map, nr_segments);
+ BUG_ON(err);
+
+ /* Retry maps with GNTST_eagain */
+ for(i=0; i < nr_segments; i++) {
+ while(unlikely(map[i].status == GNTST_eagain))
+ {
+ msleep(10);
+ err = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref,
+ &map[i],
+ 1);
+ BUG_ON(err);
+ }
+ }
+
+ for_each_sg (pending_req->sgl, sg, nr_segments, i) {
+ struct page *pg;
+
+ if (unlikely(map[i].status != 0)) {
+ printk(KERN_ERR "scsiback: invalid buffer -- could not remap it: " \
+ "%d/%d, err:%d\n", i, nr_segments, map[i].status);
+ map[i].handle = SCSIBACK_INVALID_HANDLE;
+ err |= 1;
+ }
+
+ pending_handle(pending_req, i) = map[i].handle;
+
+ if (err)
+ continue;
+
+ pg = pending_pages[vaddr_pagenr(pending_req, i)];
+
+ m2p_add_override(PFN_DOWN(map[i].dev_bus_addr), pg, false);
+ sg_set_page(sg, pg, ring_req->seg[i].length,
+ ring_req->seg[i].offset);
+ data_len += sg->length;
+
+ barrier();
+ if (sg->offset >= PAGE_SIZE ||
+ sg->length > PAGE_SIZE ||
+ sg->offset + sg->length > PAGE_SIZE)
+ err |= 1;
+
+ }
+
+ if (err)
+ goto fail_flush;
+ }
+
+ pending_req->request_bufflen = data_len;
+
+ return 0;
+
+fail_flush:
+ scsiback_fast_flush_area(pending_req);
+ return -ENOMEM;
+}
+
+/* quoted scsi_lib.c/scsi_bi_endio */
+static void scsiback_bi_endio(struct bio *bio, int error)
+{
+ bio_put(bio);
+}
+
+
+
+/* quoted scsi_lib.c/scsi_req_map_sg . */
+static struct bio *request_map_sg(pending_req_t *pending_req)
+{
+ struct request_queue *q = pending_req->sdev->request_queue;
+ unsigned int nsegs = (unsigned int)pending_req->nr_segments;
+ unsigned int i, len, bytes, off, nr_pages, nr_vecs = 0;
+ struct scatterlist *sg;
+ struct page *page;
+ struct bio *bio = NULL, *bio_first = NULL, *bio_last = NULL;
+ int err;
+
+ for_each_sg (pending_req->sgl, sg, nsegs, i) {
+ page = sg_page(sg);
+ off = sg->offset;
+ len = sg->length;
+
+ nr_pages = (len + off + PAGE_SIZE - 1) >> PAGE_SHIFT;
+ while (len > 0) {
+ bytes = min_t(unsigned int, len, PAGE_SIZE - off);
+
+ if (!bio) {
+ nr_vecs = min_t(unsigned int, BIO_MAX_PAGES,
+ nr_pages);
+ nr_pages -= nr_vecs;
+ bio = bio_alloc(GFP_KERNEL, nr_vecs);
+ if (!bio) {
+ err = -ENOMEM;
+ goto free_bios;
+ }
+ bio->bi_end_io = scsiback_bi_endio;
+ if (bio_last)
+ bio_last->bi_next = bio;
+ else
+ bio_first = bio;
+ bio_last = bio;
+ }
+
+ if (bio_add_pc_page(q, bio, page, bytes, off) !=
+ bytes) {
+ bio_put(bio);
+ err = -EINVAL;
+ goto free_bios;
+ }
+
+ if (bio->bi_vcnt >= nr_vecs) {
+ bio->bi_flags &= ~(1 << BIO_SEG_VALID);
+ if (pending_req->sc_data_direction == WRITE)
+ bio->bi_rw |= REQ_WRITE;
+ bio = NULL;
+ }
+
+ page++;
+ len -= bytes;
+ off = 0;
+ }
+ }
+
+ return bio_first;
+
+free_bios:
+ while ((bio = bio_first) != NULL) {
+ bio_first = bio->bi_next;
+ bio_put(bio);
+ }
+
+ return ERR_PTR(err);
+}
+
+
+void scsiback_cmd_exec(pending_req_t *pending_req)
+{
+ int cmd_len = (int)pending_req->cmd_len;
+ int data_dir = (int)pending_req->sc_data_direction;
+ unsigned int timeout;
+ struct request *rq;
+ int write;
+
+ DPRINTK("%s\n",__FUNCTION__);
+
+ /* because it doesn't timeout backend earlier than frontend.*/
+ if (pending_req->timeout_per_command)
+ timeout = pending_req->timeout_per_command * HZ;
+ else
+ timeout = VSCSIIF_TIMEOUT;
+
+ write = (data_dir == DMA_TO_DEVICE);
+ if (pending_req->nr_segments) {
+ struct bio *bio = request_map_sg(pending_req);
+
+ if (IS_ERR(bio)) {
+ printk(KERN_ERR "scsiback: SG Request Map Error\n");
+ return;
+ }
+
+ rq = blk_make_request(pending_req->sdev->request_queue, bio,
+ GFP_KERNEL);
+ if (IS_ERR(rq)) {
+ printk(KERN_ERR "scsiback: Make Request Error\n");
+ return;
+ }
+
+ rq->buffer = NULL;
+ } else {
+ rq = blk_get_request(pending_req->sdev->request_queue, write,
+ GFP_KERNEL);
+ if (unlikely(!rq)) {
+ printk(KERN_ERR "scsiback: Get Request Error\n");
+ return;
+ }
+ }
+
+ rq->cmd_type = REQ_TYPE_BLOCK_PC;
+ rq->cmd_len = cmd_len;
+ memcpy(rq->cmd, pending_req->cmnd, cmd_len);
+
+ memset(pending_req->sense_buffer, 0, VSCSIIF_SENSE_BUFFERSIZE);
+ rq->sense = pending_req->sense_buffer;
+ rq->sense_len = 0;
+
+ /* not allowed to retry in backend. */
+ rq->retries = 0;
+ rq->timeout = timeout;
+ rq->end_io_data = pending_req;
+
+ scsiback_get(pending_req->info);
+ blk_execute_rq_nowait(rq->q, NULL, rq, 1, scsiback_cmd_done);
+
+ return ;
+}
+
+
+static void scsiback_device_reset_exec(pending_req_t *pending_req)
+{
+ struct vscsibk_info *info = pending_req->info;
+ int err;
+ struct scsi_device *sdev = pending_req->sdev;
+
+ scsiback_get(info);
+ err = scsi_reset_provider(sdev, SCSI_TRY_RESET_DEVICE);
+
+ scsiback_do_resp_with_sense(NULL, err, 0, pending_req);
+ scsiback_put(info);
+
+ return;
+}
+
+
+irqreturn_t scsiback_intr(int irq, void *dev_id)
+{
+ scsiback_notify_work((struct vscsibk_info *)dev_id);
+ return IRQ_HANDLED;
+}
+
+static int prepare_pending_reqs(struct vscsibk_info *info,
+ vscsiif_request_t *ring_req, pending_req_t *pending_req)
+{
+ struct scsi_device *sdev;
+ struct ids_tuple vir;
+ int err = -EINVAL;
+
+ DPRINTK("%s\n",__FUNCTION__);
+
+ pending_req->rqid = ring_req->rqid;
+ pending_req->act = ring_req->act;
+
+ pending_req->info = info;
+
+ pending_req->v_chn = vir.chn = ring_req->channel;
+ pending_req->v_tgt = vir.tgt = ring_req->id;
+ vir.lun = ring_req->lun;
+
+ rmb();
+ sdev = scsiback_do_translation(info, &vir);
+ if (!sdev) {
+ pending_req->sdev = NULL;
+ DPRINTK("scsiback: doesn't exist.\n");
+ err = -ENODEV;
+ goto invalid_value;
+ }
+ pending_req->sdev = sdev;
+
+ /* request range check from frontend */
+ pending_req->sc_data_direction = ring_req->sc_data_direction;
+ barrier();
+ if ((pending_req->sc_data_direction != DMA_BIDIRECTIONAL) &&
+ (pending_req->sc_data_direction != DMA_TO_DEVICE) &&
+ (pending_req->sc_data_direction != DMA_FROM_DEVICE) &&
+ (pending_req->sc_data_direction != DMA_NONE)) {
+ DPRINTK("scsiback: invalid parameter data_dir = %d\n",
+ pending_req->sc_data_direction);
+ err = -EINVAL;
+ goto invalid_value;
+ }
+
+ pending_req->nr_segments = ring_req->nr_segments;
+ barrier();
+ if (pending_req->nr_segments > VSCSIIF_SG_TABLESIZE) {
+ DPRINTK("scsiback: invalid parameter nr_seg = %d\n",
+ pending_req->nr_segments);
+ err = -EINVAL;
+ goto invalid_value;
+ }
+
+ pending_req->cmd_len = ring_req->cmd_len;
+ barrier();
+ if (pending_req->cmd_len > VSCSIIF_MAX_COMMAND_SIZE) {
+ DPRINTK("scsiback: invalid parameter cmd_len = %d\n",
+ pending_req->cmd_len);
+ err = -EINVAL;
+ goto invalid_value;
+ }
+ memcpy(pending_req->cmnd, ring_req->cmnd, pending_req->cmd_len);
+
+ pending_req->timeout_per_command = ring_req->timeout_per_command;
+
+ if(scsiback_gnttab_data_map(ring_req, pending_req)) {
+ DPRINTK("scsiback: invalid buffer\n");
+ err = -EINVAL;
+ goto invalid_value;
+ }
+
+ return 0;
+
+invalid_value:
+ return err;
+}
+
+
+static int scsiback_do_cmd_fn(struct vscsibk_info *info)
+{
+ struct vscsiif_back_ring *ring = &info->ring;
+ vscsiif_request_t *ring_req;
+
+ pending_req_t *pending_req;
+ RING_IDX rc, rp;
+ int err, more_to_do = 0;
+
+ DPRINTK("%s\n",__FUNCTION__);
+
+ rc = ring->req_cons;
+ rp = ring->sring->req_prod;
+ rmb();
+
+ while ((rc != rp)) {
+ if (RING_REQUEST_CONS_OVERFLOW(ring, rc))
+ break;
+ pending_req = alloc_req(info);
+ if (NULL == pending_req) {
+ more_to_do = 1;
+ break;
+ }
+
+ ring_req = RING_GET_REQUEST(ring, rc);
+ ring->req_cons = ++rc;
+
+ err = prepare_pending_reqs(info, ring_req,
+ pending_req);
+ if (err == -EINVAL) {
+ scsiback_do_resp_with_sense(NULL, (DRIVER_ERROR << 24),
+ 0, pending_req);
+ continue;
+ } else if (err == -ENODEV) {
+ scsiback_do_resp_with_sense(NULL, (DID_NO_CONNECT << 16),
+ 0, pending_req);
+ continue;
+ }
+
+ if (pending_req->act == VSCSIIF_ACT_SCSI_CDB) {
+
+ /* The Host mode is through as for Emulation. */
+ if (info->feature == VSCSI_TYPE_HOST)
+ scsiback_cmd_exec(pending_req);
+ else
+ scsiback_req_emulation_or_cmdexec(pending_req);
+
+ } else if (pending_req->act == VSCSIIF_ACT_SCSI_RESET) {
+ scsiback_device_reset_exec(pending_req);
+ } else {
+ printk(KERN_ERR "scsiback: invalid parameter for request\n");
+ scsiback_do_resp_with_sense(NULL, (DRIVER_ERROR << 24),
+ 0, pending_req);
+ continue;
+ }
+ }
+
+ if (RING_HAS_UNCONSUMED_REQUESTS(ring))
+ more_to_do = 1;
+
+ /* Yield point for this unbounded loop. */
+ cond_resched();
+
+ return more_to_do;
+}
+
+
+int scsiback_schedule(void *data)
+{
+ struct vscsibk_info *info = (struct vscsibk_info *)data;
+
+ DPRINTK("%s\n",__FUNCTION__);
+
+ while (!kthread_should_stop()) {
+ wait_event_interruptible(
+ info->wq,
+ info->waiting_reqs || kthread_should_stop());
+ wait_event_interruptible(
+ pending_free_wq,
+ !list_empty(&pending_free) || kthread_should_stop());
+
+ info->waiting_reqs = 0;
+ smp_mb();
+
+ if (scsiback_do_cmd_fn(info))
+ info->waiting_reqs = 1;
+ }
+
+ return 0;
+}
+
+
+static int __init scsiback_init(void)
+{
+ int i, mmap_pages;
+
+ if (!xen_domain())
+ return -ENODEV;
+
+ mmap_pages = vscsiif_reqs * VSCSIIF_SG_TABLESIZE;
+
+ pending_reqs = kzalloc(sizeof(pending_reqs[0]) *
+ vscsiif_reqs, GFP_KERNEL);
+ pending_grant_handles = kmalloc(sizeof(pending_grant_handles[0]) *
+ mmap_pages, GFP_KERNEL);
+ pending_pages = kzalloc(sizeof(pending_pages[0]) *
+ mmap_pages, GFP_KERNEL);
+
+ if (!pending_reqs || !pending_grant_handles || !pending_pages)
+ goto out_of_memory;
+
+ for (i = 0; i < mmap_pages; i++) {
+ pending_grant_handles[i] = SCSIBACK_INVALID_HANDLE;
+ pending_pages[i] = alloc_page(GFP_KERNEL);
+ if (pending_pages[i] == NULL)
+ goto out_of_memory;
+ }
+ if (scsiback_interface_init() < 0)
+ goto out_of_kmem;
+
+ memset(pending_reqs, 0, sizeof(pending_reqs));
+ INIT_LIST_HEAD(&pending_free);
+
+ for (i = 0; i < vscsiif_reqs; i++)
+ list_add_tail(&pending_reqs[i].free_list, &pending_free);
+
+ if (scsiback_xenbus_init())
+ goto out_of_xenbus;
+
+ scsiback_emulation_init();
+
+ return 0;
+
+out_of_xenbus:
+ scsiback_xenbus_unregister();
+out_of_kmem:
+ scsiback_interface_exit();
+out_of_memory:
+ if (pending_pages) {
+ for (i = 0; i < mmap_pages; i++) {
+ if (pending_pages[i])
+ __free_page(pending_pages[i]);
+ }
+ kfree(pending_pages);
+ }
+ kfree(pending_reqs);
+ kfree(pending_grant_handles);
+ printk(KERN_ERR "scsiback: %s: out of memory\n", __FUNCTION__);
+ return -ENOMEM;
+}
+
+
+static void __exit scsiback_exit(void)
+{
+ scsiback_xenbus_unregister();
+ scsiback_interface_exit();
+ kfree(pending_reqs);
+ kfree(pending_grant_handles);
+ if (pending_pages) {
+ unsigned int i;
+ unsigned int mmap_pages = vscsiif_reqs * VSCSIIF_SG_TABLESIZE;
+ for (i = 0; i < mmap_pages; i++) {
+ if (pending_pages[i])
+ __free_page(pending_pages[i]);
+ }
+ kfree(pending_pages);
+ }
+}
+
+module_init(scsiback_init);
+module_exit(scsiback_exit);
+
+MODULE_DESCRIPTION("Xen SCSI backend driver");
+MODULE_LICENSE("Dual BSD/GPL");
diff --git a/drivers/scsi/xen-scsiback/translate.c b/drivers/scsi/xen-scsiback/translate.c
new file mode 100644
index 0000000..36873cc
--- /dev/null
+++ b/drivers/scsi/xen-scsiback/translate.c
@@ -0,0 +1,168 @@
+/*
+ * Xen SCSI backend driver
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <linux/list.h>
+#include <linux/gfp.h>
+
+#include "common.h"
+
+/*
+ Initialize the translation entry list
+*/
+void scsiback_init_translation_table(struct vscsibk_info *info)
+{
+ INIT_LIST_HEAD(&info->v2p_entry_lists);
+ spin_lock_init(&info->v2p_lock);
+}
+
+
+/*
+ Add a new translation entry
+*/
+int scsiback_add_translation_entry(struct vscsibk_info *info,
+ struct scsi_device *sdev, struct ids_tuple *v)
+{
+ int err = 0;
+ struct v2p_entry *entry;
+ struct v2p_entry *new;
+ struct list_head *head = &(info->v2p_entry_lists);
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->v2p_lock, flags);
+
+ /* Check double assignment to identical virtual ID */
+ list_for_each_entry(entry, head, l) {
+ if ((entry->v.chn == v->chn) &&
+ (entry->v.tgt == v->tgt) &&
+ (entry->v.lun == v->lun)) {
+ printk(KERN_WARNING "scsiback: Virtual ID is already used. "
+ "Assignment was not performed.\n");
+ err = -EEXIST;
+ goto out;
+ }
+
+ }
+
+ /* Create a new translation entry and add to the list */
+ if ((new = kmalloc(sizeof(struct v2p_entry), GFP_ATOMIC)) == NULL) {
+ printk(KERN_ERR "scsiback: %s: kmalloc() error.\n", __FUNCTION__);
+ err = -ENOMEM;
+ goto out;
+ }
+ new->v = *v;
+ new->sdev = sdev;
+ list_add_tail(&new->l, head);
+
+out:
+ spin_unlock_irqrestore(&info->v2p_lock, flags);
+ return err;
+}
+
+
+/*
+ Delete the translation entry specfied
+*/
+int scsiback_del_translation_entry(struct vscsibk_info *info,
+ struct ids_tuple *v)
+{
+ struct v2p_entry *entry;
+ struct list_head *head = &(info->v2p_entry_lists);
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->v2p_lock, flags);
+ /* Find out the translation entry specified */
+ list_for_each_entry(entry, head, l) {
+ if ((entry->v.chn == v->chn) &&
+ (entry->v.tgt == v->tgt) &&
+ (entry->v.lun == v->lun)) {
+ goto found;
+ }
+ }
+
+ spin_unlock_irqrestore(&info->v2p_lock, flags);
+ return 1;
+
+found:
+ /* Delete the translation entry specfied */
+ scsi_device_put(entry->sdev);
+ list_del(&entry->l);
+ kfree(entry);
+
+ spin_unlock_irqrestore(&info->v2p_lock, flags);
+ return 0;
+}
+
+
+/*
+ Perform virtual to physical translation
+*/
+struct scsi_device *scsiback_do_translation(struct vscsibk_info *info,
+ struct ids_tuple *v)
+{
+ struct v2p_entry *entry;
+ struct list_head *head = &(info->v2p_entry_lists);
+ struct scsi_device *sdev = NULL;
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->v2p_lock, flags);
+ list_for_each_entry(entry, head, l) {
+ if ((entry->v.chn == v->chn) &&
+ (entry->v.tgt == v->tgt) &&
+ (entry->v.lun == v->lun)) {
+ sdev = entry->sdev;
+ goto out;
+ }
+ }
+out:
+ spin_unlock_irqrestore(&info->v2p_lock, flags);
+ return sdev;
+}
+
+
+/*
+ Release the translation entry specfied
+*/
+void scsiback_release_translation_entry(struct vscsibk_info *info)
+{
+ struct v2p_entry *entry, *tmp;
+ struct list_head *head = &(info->v2p_entry_lists);
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->v2p_lock, flags);
+ list_for_each_entry_safe(entry, tmp, head, l) {
+ scsi_device_put(entry->sdev);
+ list_del(&entry->l);
+ kfree(entry);
+ }
+
+ spin_unlock_irqrestore(&info->v2p_lock, flags);
+ return;
+
+}
diff --git a/drivers/scsi/xen-scsiback/xenbus.c b/drivers/scsi/xen-scsiback/xenbus.c
new file mode 100644
index 0000000..0816c0e
--- /dev/null
+++ b/drivers/scsi/xen-scsiback/xenbus.c
@@ -0,0 +1,374 @@
+/*
+ * Xen SCSI backend driver
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * Based on the blkback driver code.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <stdarg.h>
+#include <linux/module.h>
+#include <linux/kthread.h>
+#include <scsi/scsi.h>
+#include <scsi/scsi_host.h>
+#include <scsi/scsi_device.h>
+
+#include "common.h"
+
+struct backend_info
+{
+ struct xenbus_device *dev;
+ struct vscsibk_info *info;
+};
+
+
+static int __vscsiif_name(struct backend_info *be, char *buf)
+{
+ struct xenbus_device *dev = be->dev;
+ unsigned int domid, id;
+
+ sscanf(dev->nodename, "backend/vscsi/%u/%u", &domid, &id);
+ snprintf(buf, TASK_COMM_LEN, "vscsi.%u.%u", be->info->domid, id);
+
+ return 0;
+}
+
+static int scsiback_map(struct backend_info *be)
+{
+ struct xenbus_device *dev = be->dev;
+ unsigned long ring_ref = 0;
+ unsigned int evtchn = 0;
+ int err;
+ char name[TASK_COMM_LEN];
+
+ err = xenbus_gather(XBT_NIL, dev->otherend,
+ "ring-ref", "%lu", &ring_ref,
+ "event-channel", "%u", &evtchn, NULL);
+ if (err) {
+ xenbus_dev_fatal(dev, err, "reading %s ring", dev->otherend);
+ return err;
+ }
+ err = scsiback_init_sring(be->info, ring_ref, evtchn);
+ if (err)
+ return err;
+
+ err = __vscsiif_name(be, name);
+ if (err) {
+ xenbus_dev_error(dev, err, "get scsiback dev name");
+ return err;
+ }
+
+ be->info->kthread = kthread_run(scsiback_schedule, be->info, name);
+ if (IS_ERR(be->info->kthread)) {
+ err = PTR_ERR(be->info->kthread);
+ be->info->kthread = NULL;
+ xenbus_dev_error(be->dev, err, "start vscsiif");
+ return err;
+ }
+
+ return 0;
+}
+
+
+struct scsi_device *scsiback_get_scsi_device(struct ids_tuple *phy)
+{
+ struct Scsi_Host *shost;
+ struct scsi_device *sdev = NULL;
+
+ shost = scsi_host_lookup(phy->hst);
+ if (IS_ERR(shost)) {
+ printk(KERN_ERR "scsiback: host%d doesn't exist.\n",
+ phy->hst);
+ return NULL;
+ }
+ sdev = scsi_device_lookup(shost, phy->chn, phy->tgt, phy->lun);
+ if (!sdev) {
+ printk(KERN_ERR "scsiback: %d:%d:%d:%d doesn't exist.\n",
+ phy->hst, phy->chn, phy->tgt, phy->lun);
+ scsi_host_put(shost);
+ return NULL;
+ }
+
+ scsi_host_put(shost);
+ return (sdev);
+}
+
+#define VSCSIBACK_OP_ADD_OR_DEL_LUN 1
+#define VSCSIBACK_OP_UPDATEDEV_STATE 2
+
+
+static void scsiback_do_lun_hotplug(struct backend_info *be, int op)
+{
+ int i, err = 0;
+ struct ids_tuple phy, vir;
+ int device_state;
+ char str[64], state_str[64];
+ char **dir;
+ unsigned int dir_n = 0;
+ struct xenbus_device *dev = be->dev;
+ struct scsi_device *sdev;
+
+ dir = xenbus_directory(XBT_NIL, dev->nodename, "vscsi-devs", &dir_n);
+ if (IS_ERR(dir))
+ return;
+
+ for (i = 0; i < dir_n; i++) {
+ /* read status */
+ snprintf(state_str, sizeof(state_str), "vscsi-devs/%s/state", dir[i]);
+ err = xenbus_scanf(XBT_NIL, dev->nodename, state_str, "%u",
+ &device_state);
+ if (XENBUS_EXIST_ERR(err))
+ continue;
+
+ /* physical SCSI device */
+ snprintf(str, sizeof(str), "vscsi-devs/%s/p-dev", dir[i]);
+ err = xenbus_scanf(XBT_NIL, dev->nodename, str,
+ "%u:%u:%u:%u", &phy.hst, &phy.chn, &phy.tgt, &phy.lun);
+ if (XENBUS_EXIST_ERR(err)) {
+ xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateClosed);
+ continue;
+ }
+
+ /* virtual SCSI device */
+ snprintf(str, sizeof(str), "vscsi-devs/%s/v-dev", dir[i]);
+ err = xenbus_scanf(XBT_NIL, dev->nodename, str,
+ "%u:%u:%u:%u", &vir.hst, &vir.chn, &vir.tgt, &vir.lun);
+ if (XENBUS_EXIST_ERR(err)) {
+ xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateClosed);
+ continue;
+ }
+
+ switch (op) {
+ case VSCSIBACK_OP_ADD_OR_DEL_LUN:
+ if (device_state == XenbusStateInitialising) {
+ sdev = scsiback_get_scsi_device(&phy);
+ if (!sdev)
+ xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateClosed);
+ else {
+ err = scsiback_add_translation_entry(be->info, sdev, &vir);
+ if (!err) {
+ if (xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateInitialised)) {
+ printk(KERN_ERR "scsiback: xenbus_printf error %s\n", state_str);
+ scsiback_del_translation_entry(be->info, &vir);
+ }
+ } else {
+ scsi_device_put(sdev);
+ xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateClosed);
+ }
+ }
+ }
+
+ if (device_state == XenbusStateClosing) {
+ if (!scsiback_del_translation_entry(be->info, &vir)) {
+ if (xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateClosed))
+ printk(KERN_ERR "scsiback: xenbus_printf error %s\n", state_str);
+ }
+ }
+ break;
+
+ case VSCSIBACK_OP_UPDATEDEV_STATE:
+ if (device_state == XenbusStateInitialised) {
+ /* modify vscsi-devs/dev-x/state */
+ if (xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateConnected)) {
+ printk(KERN_ERR "scsiback: xenbus_printf error %s\n", state_str);
+ scsiback_del_translation_entry(be->info, &vir);
+ xenbus_printf(XBT_NIL, dev->nodename, state_str,
+ "%d", XenbusStateClosed);
+ }
+ }
+ break;
+ /*When it is necessary, processing is added here.*/
+ default:
+ break;
+ }
+ }
+
+ kfree(dir);
+ return ;
+}
+
+
+static void scsiback_frontend_changed(struct xenbus_device *dev,
+ enum xenbus_state frontend_state)
+{
+ struct backend_info *be = dev_get_drvdata(&dev->dev);
+ int err;
+
+ switch (frontend_state) {
+ case XenbusStateInitialising:
+ break;
+ case XenbusStateInitialised:
+ err = scsiback_map(be);
+ if (err)
+ break;
+
+ scsiback_do_lun_hotplug(be, VSCSIBACK_OP_ADD_OR_DEL_LUN);
+ xenbus_switch_state(dev, XenbusStateConnected);
+
+ break;
+ case XenbusStateConnected:
+
+ scsiback_do_lun_hotplug(be, VSCSIBACK_OP_UPDATEDEV_STATE);
+
+ if (dev->state == XenbusStateConnected)
+ break;
+
+ xenbus_switch_state(dev, XenbusStateConnected);
+
+ break;
+
+ case XenbusStateClosing:
+ scsiback_disconnect(be->info);
+ xenbus_switch_state(dev, XenbusStateClosing);
+ break;
+
+ case XenbusStateClosed:
+ xenbus_switch_state(dev, XenbusStateClosed);
+ if (xenbus_dev_is_online(dev))
+ break;
+ /* fall through if not online */
+ case XenbusStateUnknown:
+ device_unregister(&dev->dev);
+ break;
+
+ case XenbusStateReconfiguring:
+ scsiback_do_lun_hotplug(be, VSCSIBACK_OP_ADD_OR_DEL_LUN);
+
+ xenbus_switch_state(dev, XenbusStateReconfigured);
+
+ break;
+
+ default:
+ xenbus_dev_fatal(dev, -EINVAL, "saw state %d at frontend",
+ frontend_state);
+ break;
+ }
+}
+
+
+static int scsiback_remove(struct xenbus_device *dev)
+{
+ struct backend_info *be = dev_get_drvdata(&dev->dev);
+
+ if (be->info) {
+ scsiback_disconnect(be->info);
+ scsiback_release_translation_entry(be->info);
+ scsiback_free(be->info);
+ be->info = NULL;
+ }
+
+ kfree(be);
+ dev_set_drvdata(&dev->dev, NULL);
+
+ return 0;
+}
+
+
+static int scsiback_probe(struct xenbus_device *dev,
+ const struct xenbus_device_id *id)
+{
+ int err;
+ unsigned val = 0;
+
+ struct backend_info *be = kzalloc(sizeof(struct backend_info),
+ GFP_KERNEL);
+
+ if (!be) {
+ xenbus_dev_fatal(dev, -ENOMEM,
+ "allocating backend structure");
+ return -ENOMEM;
+ }
+ be->dev = dev;
+ dev_set_drvdata(&dev->dev, be);
+
+ be->info = vscsibk_info_alloc(dev->otherend_id);
+ if (IS_ERR(be->info)) {
+ err = PTR_ERR(be->info);
+ be->info = NULL;
+ xenbus_dev_fatal(dev, err, "creating scsihost interface");
+ goto fail;
+ }
+
+ be->info->dev = dev;
+ be->info->irq = 0;
+ be->info->feature = 0; /*default not HOSTMODE.*/
+
+ scsiback_init_translation_table(be->info);
+
+ err = xenbus_scanf(XBT_NIL, dev->nodename,
+ "feature-host", "%d", &val);
+ if (XENBUS_EXIST_ERR(err))
+ val = 0;
+
+ if (val)
+ be->info->feature = VSCSI_TYPE_HOST;
+
+ err = xenbus_switch_state(dev, XenbusStateInitWait);
+ if (err)
+ goto fail;
+
+ return 0;
+
+
+fail:
+ printk(KERN_WARNING "scsiback: %s failed\n",__func__);
+ scsiback_remove(dev);
+
+ return err;
+}
+
+
+static struct xenbus_device_id scsiback_ids[] = {
+ { "vscsi" },
+ { "" }
+};
+
+static struct xenbus_driver scsiback = {
+ .name = "vscsi",
+ .owner = THIS_MODULE,
+ .ids = scsiback_ids,
+ .probe = scsiback_probe,
+ .remove = scsiback_remove,
+ .otherend_changed = scsiback_frontend_changed
+};
+
+int scsiback_xenbus_init(void)
+{
+ return xenbus_register_backend(&scsiback);
+}
+
+void scsiback_xenbus_unregister(void)
+{
+ xenbus_unregister_driver(&scsiback);
+}
diff --git a/drivers/scsi/xen-scsifront/Makefile b/drivers/scsi/xen-scsifront/Makefile
new file mode 100644
index 0000000..18a5329
--- /dev/null
+++ b/drivers/scsi/xen-scsifront/Makefile
@@ -0,0 +1,4 @@
+
+obj-$(CONFIG_XEN_SCSI_FRONTEND) := xen-scsifront.o
+xen-scsifront-objs := scsifront.o xenbus.o
+
diff --git a/drivers/scsi/xen-scsifront/common.h b/drivers/scsi/xen-scsifront/common.h
new file mode 100644
index 0000000..cfa1c32
--- /dev/null
+++ b/drivers/scsi/xen-scsifront/common.h
@@ -0,0 +1,137 @@
+/*
+ * Xen SCSI frontend driver
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef __XEN_DRIVERS_SCSIFRONT_H__
+#define __XEN_DRIVERS_SCSIFRONT_H__
+
+#include <linux/version.h>
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/device.h>
+#include <linux/kthread.h>
+#include <linux/wait.h>
+#include <linux/interrupt.h>
+#include <linux/spinlock.h>
+#include <linux/sched.h>
+#include <linux/blkdev.h>
+#include <scsi/scsi_cmnd.h>
+#include <scsi/scsi_device.h>
+#include <scsi/scsi.h>
+#include <scsi/scsi_host.h>
+#include <asm/xen/page.h>
+#include <xen/xenbus.h>
+#include <xen/grant_table.h>
+#include <xen/events.h>
+#include <xen/evtchn.h>
+#include <xen/interface/xen.h>
+#include <xen/interface/io/ring.h>
+#include <xen/interface/io/vscsiif.h>
+#include <xen/interface/grant_table.h>
+#include <xen/interface/io/protocols.h>
+#include <asm/delay.h>
+#include <asm/hypervisor.h>
+/*#include <asm/maddr.h>*/
+
+#ifdef HAVE_XEN_PLATFORM_COMPAT_H
+#include <xen/platform-compat.h>
+#endif
+
+#define GRANT_INVALID_REF 0
+#define VSCSI_IN_ABORT 1
+#define VSCSI_IN_RESET 2
+
+/* tuning point*/
+#define VSCSIIF_DEFAULT_CMD_PER_LUN 10
+#define VSCSIIF_MAX_TARGET 64
+#define VSCSIIF_MAX_LUN 255
+
+#define VSCSIIF_RING_SIZE __CONST_RING_SIZE(vscsiif, PAGE_SIZE)
+#define VSCSIIF_MAX_REQS VSCSIIF_RING_SIZE
+
+struct vscsifrnt_shadow {
+ uint16_t next_free;
+
+ /* command between backend and frontend
+ * VSCSIIF_ACT_SCSI_CDB or VSCSIIF_ACT_SCSI_RESET */
+ unsigned char act;
+
+ /* do reset function */
+ wait_queue_head_t wq_reset; /* reset work queue */
+ int wait_reset; /* reset work queue condition */
+ int32_t rslt_reset; /* reset response status */
+ /* (SUCESS or FAILED) */
+
+ /* for DMA_TO_DEVICE(1), DMA_FROM_DEVICE(2), DMA_NONE(3)
+ requests */
+ unsigned int sc_data_direction;
+
+ /* Number of pieces of scatter-gather */
+ unsigned int nr_segments;
+
+ /* requested struct scsi_cmnd is stored from kernel */
+ unsigned long req_scsi_cmnd;
+ int gref[VSCSIIF_SG_TABLESIZE];
+};
+
+struct vscsifrnt_info {
+ struct xenbus_device *dev;
+
+ struct Scsi_Host *host;
+
+ spinlock_t io_lock;
+ spinlock_t shadow_lock;
+ unsigned int evtchn;
+ unsigned int irq;
+
+ grant_ref_t ring_ref;
+ struct vscsiif_front_ring ring;
+ struct vscsiif_response ring_res;
+
+ struct vscsifrnt_shadow shadow[VSCSIIF_MAX_REQS];
+ uint32_t shadow_free;
+
+ struct task_struct *kthread;
+ wait_queue_head_t wq;
+ unsigned int waiting_resp;
+
+};
+
+#define DPRINTK(_f, _a...) \
+ pr_debug("(file=%s, line=%d) " _f, \
+ __FILE__ , __LINE__ , ## _a )
+
+int scsifront_xenbus_init(void);
+void scsifront_xenbus_unregister(void);
+int scsifront_schedule(void *data);
+irqreturn_t scsifront_intr(int irq, void *dev_id);
+int scsifront_cmd_done(struct vscsifrnt_info *info);
+
+
+#endif /* __XEN_DRIVERS_SCSIFRONT_H__ */
diff --git a/drivers/scsi/xen-scsifront/scsifront.c b/drivers/scsi/xen-scsifront/scsifront.c
new file mode 100644
index 0000000..2d5e25f
--- /dev/null
+++ b/drivers/scsi/xen-scsifront/scsifront.c
@@ -0,0 +1,469 @@
+/*
+ * Xen SCSI frontend driver
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <linux/version.h>
+#include "common.h"
+
+static int get_id_from_freelist(struct vscsifrnt_info *info)
+{
+ unsigned long flags;
+ uint32_t free;
+
+ spin_lock_irqsave(&info->shadow_lock, flags);
+
+ free = info->shadow_free;
+ BUG_ON(free > VSCSIIF_MAX_REQS);
+ info->shadow_free = info->shadow[free].next_free;
+ info->shadow[free].next_free = 0x0fff;
+
+ info->shadow[free].wait_reset = 0;
+
+ spin_unlock_irqrestore(&info->shadow_lock, flags);
+
+ return free;
+}
+
+static void add_id_to_freelist(struct vscsifrnt_info *info, uint32_t id)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->shadow_lock, flags);
+
+ info->shadow[id].next_free = info->shadow_free;
+ info->shadow[id].req_scsi_cmnd = 0;
+ info->shadow_free = id;
+
+ spin_unlock_irqrestore(&info->shadow_lock, flags);
+}
+
+
+struct vscsiif_request * scsifront_pre_request(struct vscsifrnt_info *info)
+{
+ struct vscsiif_front_ring *ring = &(info->ring);
+ vscsiif_request_t *ring_req;
+ uint32_t id;
+
+ ring_req = RING_GET_REQUEST(&(info->ring), ring->req_prod_pvt);
+
+ ring->req_prod_pvt++;
+
+ id = get_id_from_freelist(info); /* use id by response */
+ ring_req->rqid = (uint16_t)id;
+
+ return ring_req;
+}
+
+
+static void scsifront_notify_work(struct vscsifrnt_info *info)
+{
+ info->waiting_resp = 1;
+ wake_up(&info->wq);
+}
+
+
+static void scsifront_do_request(struct vscsifrnt_info *info)
+{
+ struct vscsiif_front_ring *ring = &(info->ring);
+ unsigned int irq = info->irq;
+ int notify;
+
+ RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(ring, notify);
+ if (notify)
+ notify_remote_via_irq(irq);
+}
+
+irqreturn_t scsifront_intr(int irq, void *dev_id)
+{
+ scsifront_notify_work((struct vscsifrnt_info *)dev_id);
+ return IRQ_HANDLED;
+}
+
+
+static void scsifront_gnttab_done(struct vscsifrnt_shadow *s, uint32_t id)
+{
+ int i;
+
+ if (s->sc_data_direction == DMA_NONE)
+ return;
+
+ if (s->nr_segments) {
+ for (i = 0; i < s->nr_segments; i++) {
+ if (unlikely(gnttab_query_foreign_access(
+ s->gref[i]) != 0)) {
+ printk(KERN_ALERT "scsifront: "
+ "grant still in use by backend.\n");
+ BUG();
+ }
+ gnttab_end_foreign_access(s->gref[i], 0, 0UL);
+ }
+ }
+
+ return;
+}
+
+
+static void scsifront_cdb_cmd_done(struct vscsifrnt_info *info,
+ vscsiif_response_t *ring_res)
+{
+ struct scsi_cmnd *sc;
+ uint32_t id;
+ uint8_t sense_len;
+
+ id = ring_res->rqid;
+ sc = (struct scsi_cmnd *)info->shadow[id].req_scsi_cmnd;
+
+ if (sc == NULL)
+ BUG();
+
+ scsifront_gnttab_done(&info->shadow[id], id);
+ add_id_to_freelist(info, id);
+
+ sc->result = ring_res->rslt;
+ scsi_set_resid(sc, ring_res->residual_len);
+
+ if (ring_res->sense_len > VSCSIIF_SENSE_BUFFERSIZE)
+ sense_len = VSCSIIF_SENSE_BUFFERSIZE;
+ else
+ sense_len = ring_res->sense_len;
+
+ if (sense_len)
+ memcpy(sc->sense_buffer, ring_res->sense_buffer, sense_len);
+
+ sc->scsi_done(sc);
+
+ return;
+}
+
+
+static void scsifront_sync_cmd_done(struct vscsifrnt_info *info,
+ vscsiif_response_t *ring_res)
+{
+ uint16_t id = ring_res->rqid;
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->shadow_lock, flags);
+ info->shadow[id].wait_reset = 1;
+ info->shadow[id].rslt_reset = ring_res->rslt;
+ spin_unlock_irqrestore(&info->shadow_lock, flags);
+
+ wake_up(&(info->shadow[id].wq_reset));
+}
+
+
+int scsifront_cmd_done(struct vscsifrnt_info *info)
+{
+ vscsiif_response_t *ring_res;
+
+ RING_IDX i, rp;
+ int more_to_do = 0;
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->io_lock, flags);
+
+ rp = info->ring.sring->rsp_prod;
+ rmb();
+ for (i = info->ring.rsp_cons; i != rp; i++) {
+
+ ring_res = RING_GET_RESPONSE(&info->ring, i);
+
+ if (info->shadow[ring_res->rqid].act == VSCSIIF_ACT_SCSI_CDB)
+ scsifront_cdb_cmd_done(info, ring_res);
+ else
+ scsifront_sync_cmd_done(info, ring_res);
+ }
+
+ info->ring.rsp_cons = i;
+
+ if (i != info->ring.req_prod_pvt) {
+ RING_FINAL_CHECK_FOR_RESPONSES(&info->ring, more_to_do);
+ } else {
+ info->ring.sring->rsp_event = i + 1;
+ }
+
+ spin_unlock_irqrestore(&info->io_lock, flags);
+
+
+ /* Yield point for this unbounded loop. */
+ cond_resched();
+
+ return more_to_do;
+}
+
+
+
+
+int scsifront_schedule(void *data)
+{
+ struct vscsifrnt_info *info = (struct vscsifrnt_info *)data;
+
+ while (!kthread_should_stop()) {
+ wait_event_interruptible(
+ info->wq,
+ info->waiting_resp || kthread_should_stop());
+
+ info->waiting_resp = 0;
+ smp_mb();
+
+ if (scsifront_cmd_done(info))
+ info->waiting_resp = 1;
+ }
+
+ return 0;
+}
+
+
+
+static int map_data_for_request(struct vscsifrnt_info *info,
+ struct scsi_cmnd *sc, vscsiif_request_t *ring_req, uint32_t id)
+{
+ grant_ref_t gref_head;
+ int err, ref, ref_cnt = 0;
+ int write = (sc->sc_data_direction == DMA_TO_DEVICE);
+ unsigned int i, nr_pages, off, len, bytes;
+ unsigned long buffer_mfn, buffer_pfn;
+
+ if (sc->sc_data_direction == DMA_NONE)
+ return 0;
+
+ err = gnttab_alloc_grant_references(VSCSIIF_SG_TABLESIZE, &gref_head);
+ if (err) {
+ printk(KERN_ERR "scsifront: gnttab_alloc_grant_references() error\n");
+ return -ENOMEM;
+ }
+
+ if (scsi_bufflen(sc)) {
+ /* quoted scsi_lib.c/scsi_req_map_sg . */
+ struct scatterlist *sg, *sgl = scsi_sglist(sc);
+ unsigned int data_len = scsi_bufflen(sc);
+
+ nr_pages = (data_len + sgl->offset + PAGE_SIZE - 1) >> PAGE_SHIFT;
+ if (nr_pages > VSCSIIF_SG_TABLESIZE) {
+ printk(KERN_ERR "scsifront: Unable to map request_buffer for command!\n");
+ ref_cnt = (-E2BIG);
+ goto big_to_sg;
+ }
+
+ for_each_sg (sgl, sg, scsi_sg_count(sc), i) {
+ off = sg->offset;
+ len = sg->length;
+
+ buffer_pfn = page_to_pfn(sg_page(sg));
+ while (len > 0 && data_len > 0) {
+ /*
+ * sg sends a scatterlist that is larger than
+ * the data_len it wants transferred for certain
+ * IO sizes
+ */
+ bytes = min_t(unsigned int, len, PAGE_SIZE - off);
+ bytes = min(bytes, data_len);
+
+ ref = gnttab_claim_grant_reference(&gref_head);
+ BUG_ON(ref == -ENOSPC);
+
+ buffer_mfn = pfn_to_mfn(buffer_pfn);
+ gnttab_grant_foreign_access_ref(ref, info->dev->otherend_id,
+ buffer_mfn, write);
+
+ info->shadow[id].gref[ref_cnt] = ref;
+ ring_req->seg[ref_cnt].gref = ref;
+ ring_req->seg[ref_cnt].offset = (uint16_t)off;
+ ring_req->seg[ref_cnt].length = (uint16_t)bytes;
+
+ buffer_pfn ++;
+ len -= bytes;
+ data_len -= bytes;
+ off = 0;
+ ref_cnt++;
+ }
+ }
+ }
+
+big_to_sg:
+
+ gnttab_free_grant_references(gref_head);
+
+ return ref_cnt;
+}
+
+static int scsifront_queuecommand(struct Scsi_Host *h, struct scsi_cmnd *sc)
+{
+ struct vscsifrnt_info *info =
+ (struct vscsifrnt_info *) sc->device->host->hostdata;
+ vscsiif_request_t *ring_req;
+ int ref_cnt;
+ uint16_t rqid;
+
+ if (RING_FULL(&info->ring)) {
+ goto out_host_busy;
+ }
+
+ sc->result = 0;
+
+ ring_req = scsifront_pre_request(info);
+ rqid = ring_req->rqid;
+ ring_req->act = VSCSIIF_ACT_SCSI_CDB;
+
+ ring_req->id = sc->device->id;
+ ring_req->lun = sc->device->lun;
+ ring_req->channel = sc->device->channel;
+ ring_req->cmd_len = sc->cmd_len;
+
+ BUG_ON(sc->cmd_len > VSCSIIF_MAX_COMMAND_SIZE);
+
+ if (sc->cmd_len)
+ memcpy(ring_req->cmnd, sc->cmnd, sc->cmd_len);
+ else
+ memset(ring_req->cmnd, 0, VSCSIIF_MAX_COMMAND_SIZE);
+
+ ring_req->sc_data_direction = (uint8_t)sc->sc_data_direction;
+ ring_req->timeout_per_command = (sc->request->timeout / HZ);
+
+ info->shadow[rqid].req_scsi_cmnd = (unsigned long)sc;
+ info->shadow[rqid].sc_data_direction = sc->sc_data_direction;
+ info->shadow[rqid].act = ring_req->act;
+
+ ref_cnt = map_data_for_request(info, sc, ring_req, rqid);
+ if (ref_cnt < 0) {
+ add_id_to_freelist(info, rqid);
+ if (ref_cnt == (-ENOMEM))
+ goto out_host_busy;
+ else {
+ sc->result = (DID_ERROR << 16);
+ goto out_fail_command;
+ }
+ }
+
+ ring_req->nr_segments = (uint8_t)ref_cnt;
+ info->shadow[rqid].nr_segments = ref_cnt;
+
+ scsifront_do_request(info);
+
+ return 0;
+
+out_host_busy:
+ return SCSI_MLQUEUE_HOST_BUSY;
+
+out_fail_command:
+ sc->scsi_done(sc);
+ return 0;
+}
+
+
+static int scsifront_eh_abort_handler(struct scsi_cmnd *sc)
+{
+ return (FAILED);
+}
+
+/* vscsi supports only device_reset, because it is each of LUNs */
+static int scsifront_dev_reset_handler(struct scsi_cmnd *sc)
+{
+ struct Scsi_Host *host = sc->device->host;
+ struct vscsifrnt_info *info =
+ (struct vscsifrnt_info *) sc->device->host->hostdata;
+
+ vscsiif_request_t *ring_req;
+ uint16_t rqid;
+ int err;
+
+ spin_lock_irq(host->host_lock);
+
+ ring_req = scsifront_pre_request(info);
+ ring_req->act = VSCSIIF_ACT_SCSI_RESET;
+
+ rqid = ring_req->rqid;
+ info->shadow[rqid].act = VSCSIIF_ACT_SCSI_RESET;
+
+ ring_req->channel = sc->device->channel;
+ ring_req->id = sc->device->id;
+ ring_req->lun = sc->device->lun;
+ ring_req->cmd_len = sc->cmd_len;
+
+ if ( sc->cmd_len )
+ memcpy(ring_req->cmnd, sc->cmnd, sc->cmd_len);
+ else
+ memset(ring_req->cmnd, 0, VSCSIIF_MAX_COMMAND_SIZE);
+
+ ring_req->sc_data_direction = (uint8_t)sc->sc_data_direction;
+ ring_req->timeout_per_command = (sc->request->timeout / HZ);
+ ring_req->nr_segments = 0;
+
+ scsifront_do_request(info);
+
+ spin_unlock_irq(host->host_lock);
+ wait_event_interruptible(info->shadow[rqid].wq_reset,
+ info->shadow[rqid].wait_reset);
+ spin_lock_irq(host->host_lock);
+
+ err = info->shadow[rqid].rslt_reset;
+
+ add_id_to_freelist(info, rqid);
+
+ spin_unlock_irq(host->host_lock);
+ return (err);
+}
+
+
+struct scsi_host_template scsifront_sht = {
+ .module = THIS_MODULE,
+ .name = "Xen SCSI frontend driver",
+ .queuecommand = scsifront_queuecommand,
+ .eh_abort_handler = scsifront_eh_abort_handler,
+ .eh_device_reset_handler= scsifront_dev_reset_handler,
+ .cmd_per_lun = VSCSIIF_DEFAULT_CMD_PER_LUN,
+ .can_queue = VSCSIIF_MAX_REQS,
+ .this_id = -1,
+ .sg_tablesize = VSCSIIF_SG_TABLESIZE,
+ .use_clustering = DISABLE_CLUSTERING,
+ .proc_name = "scsifront",
+};
+
+
+static int __init scsifront_init(void)
+{
+ int err;
+
+ if (!xen_domain())
+ return -ENODEV;
+
+ err = scsifront_xenbus_init();
+
+ return err;
+}
+
+static void __exit scsifront_exit(void)
+{
+ scsifront_xenbus_unregister();
+}
+
+module_init(scsifront_init);
+module_exit(scsifront_exit);
+
+MODULE_DESCRIPTION("Xen SCSI frontend driver");
+MODULE_LICENSE("GPL");
diff --git a/drivers/scsi/xen-scsifront/xenbus.c b/drivers/scsi/xen-scsifront/xenbus.c
new file mode 100644
index 0000000..3b9f04a
--- /dev/null
+++ b/drivers/scsi/xen-scsifront/xenbus.c
@@ -0,0 +1,414 @@
+/*
+ * Xen SCSI frontend driver
+ *
+ * Copyright (c) 2008, FUJITSU Limited
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License version 2
+ * as published by the Free Software Foundation; or, when distributed
+ * separately from the Linux kernel or incorporated into other
+ * software packages, subject to the following license:
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this source file (the "Software"), to deal in the Software without
+ * restriction, including without limitation the rights to use, copy, modify,
+ * merge, publish, distribute, sublicense, and/or sell copies of the Software,
+ * and to permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <linux/version.h>
+#include "common.h"
+
+extern struct scsi_host_template scsifront_sht;
+
+static void scsifront_free(struct vscsifrnt_info *info)
+{
+ struct Scsi_Host *host = info->host;
+
+ if (host->shost_state != SHOST_DEL)
+ scsi_remove_host(info->host);
+
+ if (info->ring_ref != GRANT_INVALID_REF) {
+ gnttab_end_foreign_access(info->ring_ref,
+ 0, (unsigned long)info->ring.sring);
+ info->ring_ref = GRANT_INVALID_REF;
+ info->ring.sring = NULL;
+ }
+
+ if (info->irq)
+ unbind_from_irqhandler(info->irq, info);
+ info->irq = 0;
+
+ scsi_host_put(info->host);
+}
+
+
+static int scsifront_alloc_ring(struct vscsifrnt_info *info)
+{
+ struct xenbus_device *dev = info->dev;
+ struct vscsiif_sring *sring;
+ int err = -ENOMEM;
+
+
+ info->ring_ref = GRANT_INVALID_REF;
+
+ /***** Frontend to Backend ring start *****/
+ sring = (struct vscsiif_sring *) __get_free_page(GFP_KERNEL);
+ if (!sring) {
+ xenbus_dev_fatal(dev, err, "fail to allocate shared ring (Front to Back)");
+ return err;
+ }
+ SHARED_RING_INIT(sring);
+ FRONT_RING_INIT(&info->ring, sring, PAGE_SIZE);
+
+ err = xenbus_grant_ring(dev, virt_to_mfn(sring));
+ if (err < 0) {
+ free_page((unsigned long) sring);
+ info->ring.sring = NULL;
+ xenbus_dev_fatal(dev, err, "fail to grant shared ring (Front to Back)");
+ goto free_sring;
+ }
+ info->ring_ref = err;
+
+ err = xenbus_alloc_evtchn(dev, &info->evtchn);
+ if (err)
+ goto free_sring;
+
+ err = bind_evtchn_to_irqhandler(
+ info->evtchn, scsifront_intr,
+ IRQF_SAMPLE_RANDOM, "scsifront", info);
+
+ if (err <= 0) {
+ xenbus_dev_fatal(dev, err, "bind_evtchn_to_irqhandler");
+ goto free_sring;
+ }
+ info->irq = err;
+
+ return 0;
+
+/* free resource */
+free_sring:
+ scsifront_free(info);
+
+ return err;
+}
+
+
+static int scsifront_init_ring(struct vscsifrnt_info *info)
+{
+ struct xenbus_device *dev = info->dev;
+ struct xenbus_transaction xbt;
+ int err;
+
+ DPRINTK("%s\n",__FUNCTION__);
+
+ err = scsifront_alloc_ring(info);
+ if (err)
+ return err;
+ DPRINTK("%u %u\n", info->ring_ref, info->evtchn);
+
+again:
+ err = xenbus_transaction_start(&xbt);
+ if (err) {
+ xenbus_dev_fatal(dev, err, "starting transaction");
+ }
+
+ err = xenbus_printf(xbt, dev->nodename, "ring-ref", "%u",
+ info->ring_ref);
+ if (err) {
+ xenbus_dev_fatal(dev, err, "%s", "writing ring-ref");
+ goto fail;
+ }
+
+ err = xenbus_printf(xbt, dev->nodename, "event-channel", "%u",
+ info->evtchn);
+
+ if (err) {
+ xenbus_dev_fatal(dev, err, "%s", "writing event-channel");
+ goto fail;
+ }
+
+ err = xenbus_transaction_end(xbt, 0);
+ if (err) {
+ if (err == -EAGAIN)
+ goto again;
+ xenbus_dev_fatal(dev, err, "completing transaction");
+ goto free_sring;
+ }
+
+ return 0;
+
+fail:
+ xenbus_transaction_end(xbt, 1);
+free_sring:
+ /* free resource */
+ scsifront_free(info);
+
+ return err;
+}
+
+
+static int scsifront_probe(struct xenbus_device *dev,
+ const struct xenbus_device_id *id)
+{
+ struct vscsifrnt_info *info;
+ struct Scsi_Host *host;
+ int i, err = -ENOMEM;
+ char name[TASK_COMM_LEN];
+
+ host = scsi_host_alloc(&scsifront_sht, sizeof(*info));
+ if (!host) {
+ xenbus_dev_fatal(dev, err, "fail to allocate scsi host");
+ return err;
+ }
+ info = (struct vscsifrnt_info *) host->hostdata;
+ info->host = host;
+
+
+ dev_set_drvdata(&dev->dev, info);
+ info->dev = dev;
+
+ for (i = 0; i < VSCSIIF_MAX_REQS; i++) {
+ info->shadow[i].next_free = i + 1;
+ init_waitqueue_head(&(info->shadow[i].wq_reset));
+ info->shadow[i].wait_reset = 0;
+ }
+ info->shadow[VSCSIIF_MAX_REQS - 1].next_free = 0x0fff;
+
+ err = scsifront_init_ring(info);
+ if (err) {
+ scsi_host_put(host);
+ return err;
+ }
+
+ init_waitqueue_head(&info->wq);
+ spin_lock_init(&info->io_lock);
+ spin_lock_init(&info->shadow_lock);
+
+ snprintf(name, TASK_COMM_LEN, "vscsiif.%d", info->host->host_no);
+
+ info->kthread = kthread_run(scsifront_schedule, info, name);
+ if (IS_ERR(info->kthread)) {
+ err = PTR_ERR(info->kthread);
+ info->kthread = NULL;
+ printk(KERN_ERR "scsifront: kthread start err %d\n", err);
+ goto free_sring;
+ }
+
+ host->max_id = VSCSIIF_MAX_TARGET;
+ host->max_channel = 0;
+ host->max_lun = VSCSIIF_MAX_LUN;
+ host->max_sectors = (VSCSIIF_SG_TABLESIZE - 1) * PAGE_SIZE / 512;
+ host->max_cmd_len = VSCSIIF_MAX_COMMAND_SIZE;
+
+ err = scsi_add_host(host, &dev->dev);
+ if (err) {
+ printk(KERN_ERR "scsifront: fail to add scsi host %d\n", err);
+ goto free_sring;
+ }
+
+ xenbus_switch_state(dev, XenbusStateInitialised);
+
+ return 0;
+
+free_sring:
+ /* free resource */
+ scsifront_free(info);
+ return err;
+}
+
+static int scsifront_remove(struct xenbus_device *dev)
+{
+ struct vscsifrnt_info *info = dev_get_drvdata(&dev->dev);
+
+ DPRINTK("%s: %s removed\n",__FUNCTION__ ,dev->nodename);
+
+ if (info->kthread) {
+ kthread_stop(info->kthread);
+ info->kthread = NULL;
+ }
+
+ scsifront_free(info);
+
+ return 0;
+}
+
+
+static int scsifront_disconnect(struct vscsifrnt_info *info)
+{
+ struct xenbus_device *dev = info->dev;
+ struct Scsi_Host *host = info->host;
+
+ DPRINTK("%s: %s disconnect\n",__FUNCTION__ ,dev->nodename);
+
+ /*
+ When this function is executed, all devices of
+ Frontend have been deleted.
+ Therefore, it need not block I/O before remove_host.
+ */
+
+ scsi_remove_host(host);
+ xenbus_frontend_closed(dev);
+
+ return 0;
+}
+
+#define VSCSIFRONT_OP_ADD_LUN 1
+#define VSCSIFRONT_OP_DEL_LUN 2
+
+static void scsifront_do_lun_hotplug(struct vscsifrnt_info *info, int op)
+{
+ struct xenbus_device *dev = info->dev;
+ int i, err = 0;
+ char str[64], state_str[64];
+ char **dir;
+ unsigned int dir_n = 0;
+ unsigned int device_state;
+ unsigned int hst, chn, tgt, lun;
+ struct scsi_device *sdev;
+
+ dir = xenbus_directory(XBT_NIL, dev->otherend, "vscsi-devs", &dir_n);
+ if (IS_ERR(dir))
+ return;
+
+ for (i = 0; i < dir_n; i++) {
+ /* read status */
+ snprintf(str, sizeof(str), "vscsi-devs/%s/state", dir[i]);
+ err = xenbus_scanf(XBT_NIL, dev->otherend, str, "%u",
+ &device_state);
+ if (XENBUS_EXIST_ERR(err))
+ continue;
+
+ /* virtual SCSI device */
+ snprintf(str, sizeof(str), "vscsi-devs/%s/v-dev", dir[i]);
+ err = xenbus_scanf(XBT_NIL, dev->otherend, str,
+ "%u:%u:%u:%u", &hst, &chn, &tgt, &lun);
+ if (XENBUS_EXIST_ERR(err))
+ continue;
+
+ /* front device state path */
+ snprintf(state_str, sizeof(state_str), "vscsi-devs/%s/state", dir[i]);
+
+ switch (op) {
+ case VSCSIFRONT_OP_ADD_LUN:
+ if (device_state == XenbusStateInitialised) {
+ sdev = scsi_device_lookup(info->host, chn, tgt, lun);
+ if (sdev) {
+ printk(KERN_ERR "scsifront: Device already in use.\n");
+ scsi_device_put(sdev);
+ xenbus_printf(XBT_NIL, dev->nodename,
+ state_str, "%d", XenbusStateClosed);
+ } else {
+ scsi_add_device(info->host, chn, tgt, lun);
+ xenbus_printf(XBT_NIL, dev->nodename,
+ state_str, "%d", XenbusStateConnected);
+ }
+ }
+ break;
+ case VSCSIFRONT_OP_DEL_LUN:
+ if (device_state == XenbusStateClosing) {
+ sdev = scsi_device_lookup(info->host, chn, tgt, lun);
+ if (sdev) {
+ scsi_remove_device(sdev);
+ scsi_device_put(sdev);
+ xenbus_printf(XBT_NIL, dev->nodename,
+ state_str, "%d", XenbusStateClosed);
+ }
+ }
+ break;
+ default:
+ break;
+ }
+ }
+
+ kfree(dir);
+ return;
+}
+
+
+
+
+static void scsifront_backend_changed(struct xenbus_device *dev,
+ enum xenbus_state backend_state)
+{
+ struct vscsifrnt_info *info = dev_get_drvdata(&dev->dev);
+
+ DPRINTK("%p %u %u\n", dev, dev->state, backend_state);
+
+ switch (backend_state) {
+ case XenbusStateUnknown:
+ case XenbusStateInitialising:
+ case XenbusStateInitWait:
+ case XenbusStateClosed:
+ break;
+
+ case XenbusStateInitialised:
+ break;
+
+ case XenbusStateConnected:
+ if (xenbus_read_driver_state(dev->nodename) ==
+ XenbusStateInitialised) {
+ scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_ADD_LUN);
+ }
+
+ if (dev->state == XenbusStateConnected)
+ break;
+
+ xenbus_switch_state(dev, XenbusStateConnected);
+ break;
+
+ case XenbusStateClosing:
+ scsifront_disconnect(info);
+ break;
+
+ case XenbusStateReconfiguring:
+ scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_DEL_LUN);
+ xenbus_switch_state(dev, XenbusStateReconfiguring);
+ break;
+
+ case XenbusStateReconfigured:
+ scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_ADD_LUN);
+ xenbus_switch_state(dev, XenbusStateConnected);
+ break;
+ }
+}
+
+
+static struct xenbus_device_id scsifront_ids[] = {
+ { "vscsi" },
+ { "" }
+};
+MODULE_ALIAS("xen:vscsi");
+
+static struct xenbus_driver scsifront_driver = {
+ .name = "vscsi",
+ .owner = THIS_MODULE,
+ .ids = scsifront_ids,
+ .probe = scsifront_probe,
+ .remove = scsifront_remove,
+/* .resume = scsifront_resume, */
+ .otherend_changed = scsifront_backend_changed,
+};
+
+int scsifront_xenbus_init(void)
+{
+ return xenbus_register_frontend(&scsifront_driver);
+}
+
+void scsifront_xenbus_unregister(void)
+{
+ xenbus_unregister_driver(&scsifront_driver);
+}
+
diff --git a/include/xen/interface/grant_table.h b/include/xen/interface/grant_table.h
index 39e5717..ef2b377 100644
--- a/include/xen/interface/grant_table.h
+++ b/include/xen/interface/grant_table.h
@@ -363,6 +363,8 @@ DEFINE_GUEST_HANDLE_STRUCT(gnttab_query_size);
#define GNTST_permission_denied (-8) /* Not enough privilege for operation. */
#define GNTST_bad_page (-9) /* Specified page was invalid for op. */
#define GNTST_bad_copy_arg (-10) /* copy arguments cross page boundary */
+#define GNTST_address_too_big (-11) /* transfer page address too large. */
+#define GNTST_eagain (-12) /* Could not map at the moment. Retry. */
#define GNTTABOP_error_msgs { \
"okay", \
@@ -376,6 +378,8 @@ DEFINE_GUEST_HANDLE_STRUCT(gnttab_query_size);
"permission denied", \
"bad page", \
"copy arguments cross page boundary" \
+ "page address size too large", \
+ "could not map at the moment, retry" \
}
#endif /* __XEN_PUBLIC_GRANT_TABLE_H__ */
diff --git a/include/xen/interface/io/vscsiif.h b/include/xen/interface/io/vscsiif.h
new file mode 100644
index 0000000..7fbe688
--- /dev/null
+++ b/include/xen/interface/io/vscsiif.h
@@ -0,0 +1,105 @@
+/******************************************************************************
+ * vscsiif.h
+ *
+ * Based on the blkif.h code.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ *
+ * Copyright(c) FUJITSU Limited 2008.
+ */
+
+#ifndef __XEN__PUBLIC_IO_SCSI_H__
+#define __XEN__PUBLIC_IO_SCSI_H__
+
+#include "ring.h"
+#include "../grant_table.h"
+
+/* command between backend and frontend */
+#define VSCSIIF_ACT_SCSI_CDB 1 /* SCSI CDB command */
+#define VSCSIIF_ACT_SCSI_ABORT 2 /* SCSI Device(Lun) Abort*/
+#define VSCSIIF_ACT_SCSI_RESET 3 /* SCSI Device(Lun) Reset*/
+
+
+#define VSCSIIF_BACK_MAX_PENDING_REQS 128
+
+/*
+ * Maximum scatter/gather segments per request.
+ *
+ * Considering balance between allocating al least 16 "vscsiif_request"
+ * structures on one page (4096bytes) and number of scatter gather
+ * needed, we decided to use 26 as a magic number.
+ */
+#define VSCSIIF_SG_TABLESIZE 26
+
+/*
+ * base on linux kernel 2.6.18
+ */
+#define VSCSIIF_MAX_COMMAND_SIZE 16
+#define VSCSIIF_SENSE_BUFFERSIZE 96
+
+
+struct vscsiif_request {
+ uint16_t rqid; /* private guest value, echoed in resp */
+ uint8_t act; /* command between backend and frontend */
+ uint8_t cmd_len;
+
+ uint8_t cmnd[VSCSIIF_MAX_COMMAND_SIZE];
+ uint16_t timeout_per_command; /* The command is issued by twice
+ the value in Backend. */
+ uint16_t channel, id, lun;
+ uint16_t padding;
+ uint8_t sc_data_direction; /* for DMA_TO_DEVICE(1)
+ DMA_FROM_DEVICE(2)
+ DMA_NONE(3) requests */
+ uint8_t nr_segments; /* Number of pieces of scatter-gather */
+
+ struct scsiif_request_segment {
+ grant_ref_t gref;
+ uint16_t offset;
+ uint16_t length;
+ } seg[VSCSIIF_SG_TABLESIZE];
+ uint32_t reserved[3];
+};
+typedef struct vscsiif_request vscsiif_request_t;
+
+struct vscsiif_response {
+ uint16_t rqid;
+ uint8_t padding;
+ uint8_t sense_len;
+ uint8_t sense_buffer[VSCSIIF_SENSE_BUFFERSIZE];
+ int32_t rslt;
+ uint32_t residual_len; /* request bufflen -
+ return the value from physical device */
+ uint32_t reserved[36];
+};
+typedef struct vscsiif_response vscsiif_response_t;
+
+DEFINE_RING_TYPES(vscsiif, struct vscsiif_request, struct vscsiif_response);
+
+
+#endif /*__XEN__PUBLIC_IO_SCSI_H__*/
+/*
+ * Local variables:
+ * mode: C
+ * c-set-style: "BSD"
+ * c-basic-offset: 4
+ * tab-width: 4
+ * indent-tabs-mode: nil
+ * End:
+ */
[-- Attachment #3: Type: text/plain, Size: 138 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
^ permalink raw reply related [flat|nested] 10+ messages in thread
* Re: [RFC PATCH v01] Xen PVSCSI drivers for pvops xen/stable-2.6.32.x kernel
@ 2011-01-07 10:03 Nathanael Rensen
2011-01-07 10:26 ` Ian Campbell
2011-01-07 10:35 ` Pasi Kärkkäinen
0 siblings, 2 replies; 10+ messages in thread
From: Nathanael Rensen @ 2011-01-07 10:03 UTC (permalink / raw)
To: Pasi Kärkkäinen; +Cc: Jeremy Fitzhardinge, xen-devel, Ian Campbell
On 7 January 2011 06:35, Pasi Kärkkäinen <pasik@iki.fi> wrote:
> Hello,
>
> http://pasik.reaktio.net/xen/patches/xen-pvscsi-drivers-linux-2.6.32.27-pvops-v01.diff
>
> This is the first version of Xen PVSCSI drivers, both the scsiback backend and
> scsifront frontend, ported from Novell SLES11SP1 2.6.32 Xenlinux kernel to
> pvops xen/stable-2.6.32.x branch.
>
> -- Pasi
That's great. Some comments based on my experience with pvusb:
1) You've added GNTST_eagain into include/xen/interface/grant_table.h.
When porting the pvusb drivers I removed the "check and retry" loops
based on GNTST_eagain because I noticed that netback and blkback don't
do that. Looking at the grant_table code in xen-4.0-testing I don't
think GNTST_eagain is used now. Perhaps someone with knowledge of the
history of GNTST_eagain could help to clarify.
2) When porting pvusb I found I had to take care with the
gnttab_grant_foreign_access_ref calls. Following the example of
blkback I replaced
buffer_pfn = page_to_phys(page) >> PAGE_SHIFT;
with
buffer_mfn = pfn_to_mfn(page_to_pfn(page));
The situation for pvscsi looks a bit more involved (e.g. buffer_pfn++).
3) The existing xen-netfront, xen-blkfront and xen-pcifront have been
converted to a single .c module when ported to pvops. I followed that
lead with pvusb. You might like to do the same for consistency.
Nathanael
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [RFC PATCH v01] Xen PVSCSI drivers for pvops xen/stable-2.6.32.x kernel
2011-01-07 10:03 Nathanael Rensen
@ 2011-01-07 10:26 ` Ian Campbell
2011-01-07 10:32 ` Pasi Kärkkäinen
2011-01-10 10:15 ` Olaf Hering
2011-01-07 10:35 ` Pasi Kärkkäinen
1 sibling, 2 replies; 10+ messages in thread
From: Ian Campbell @ 2011-01-07 10:26 UTC (permalink / raw)
To: Nathanael Rensen
Cc: Patrick Colp, Jeremy Fitzhardinge, xen-devel@lists.xensource.com,
Olaf Hering
On Fri, 2011-01-07 at 10:03 +0000, Nathanael Rensen wrote:
> On 7 January 2011 06:35, Pasi Kärkkäinen <pasik@iki.fi> wrote:
> > Hello,
> >
> > http://pasik.reaktio.net/xen/patches/xen-pvscsi-drivers-linux-2.6.32.27-pvops-v01.diff
> >
> > This is the first version of Xen PVSCSI drivers, both the scsiback backend and
> > scsifront frontend, ported from Novell SLES11SP1 2.6.32 Xenlinux kernel to
> > pvops xen/stable-2.6.32.x branch.
> >
> > -- Pasi
>
> That's great. Some comments based on my experience with pvusb:
>
> 1) You've added GNTST_eagain into include/xen/interface/grant_table.h.
> When porting the pvusb drivers I removed the "check and retry" loops
> based on GNTST_eagain because I noticed that netback and blkback don't
> do that. Looking at the grant_table code in xen-4.0-testing I don't
> think GNTST_eagain is used now. Perhaps someone with knowledge of the
> history of GNTST_eagain could help to clarify.
IIRC GNTST_eagain was added to the classic Xen trees but never forward
ported to pvops.
It's part of the page sharing paging work (so dom0 can deal with a
paged/shared guest rather than for paging of dom0 itself). I guess those
don't work with pvops kernels? CC'ing the guys who are working on that
stuff.
(The presence of the use of msleep in the eagain case is a bit
terrifying. Stripping it out for now and redoing it might not be a bad
plan)
Ian.
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [RFC PATCH v01] Xen PVSCSI drivers for pvops xen/stable-2.6.32.x kernel
2011-01-07 10:26 ` Ian Campbell
@ 2011-01-07 10:32 ` Pasi Kärkkäinen
2011-01-10 10:15 ` Olaf Hering
1 sibling, 0 replies; 10+ messages in thread
From: Pasi Kärkkäinen @ 2011-01-07 10:32 UTC (permalink / raw)
To: Ian Campbell
Cc: Patrick Colp, Jeremy Fitzhardinge, xen-devel@lists.xensource.com,
Olaf Hering, Nathanael Rensen
On Fri, Jan 07, 2011 at 10:26:56AM +0000, Ian Campbell wrote:
> On Fri, 2011-01-07 at 10:03 +0000, Nathanael Rensen wrote:
> > On 7 January 2011 06:35, Pasi Kärkkäinen <pasik@iki.fi> wrote:
> > > Hello,
> > >
> > > http://pasik.reaktio.net/xen/patches/xen-pvscsi-drivers-linux-2.6.32.27-pvops-v01.diff
> > >
> > > This is the first version of Xen PVSCSI drivers, both the scsiback backend and
> > > scsifront frontend, ported from Novell SLES11SP1 2.6.32 Xenlinux kernel to
> > > pvops xen/stable-2.6.32.x branch.
> > >
> > > -- Pasi
> >
> > That's great. Some comments based on my experience with pvusb:
> >
> > 1) You've added GNTST_eagain into include/xen/interface/grant_table.h.
> > When porting the pvusb drivers I removed the "check and retry" loops
> > based on GNTST_eagain because I noticed that netback and blkback don't
> > do that. Looking at the grant_table code in xen-4.0-testing I don't
> > think GNTST_eagain is used now. Perhaps someone with knowledge of the
> > history of GNTST_eagain could help to clarify.
>
> IIRC GNTST_eagain was added to the classic Xen trees but never forward
> ported to pvops.
>
Yeah, I noticed GNTST_eagain was missing from pvops, so I just blindly
copied it from Xenlinux to pvops..
With the first version of PVSCSI for pvops I tried to do minimal
changes, just enough to make it compile..
So that stuff needs some more attention.
> It's part of the page sharing paging work (so dom0 can deal with a
> paged/shared guest rather than for paging of dom0 itself). I guess those
> don't work with pvops kernels? CC'ing the guys who are working on that
> stuff.
>
> (The presence of the use of msleep in the eagain case is a bit
> terrifying. Stripping it out for now and redoing it might not be a bad
> plan)
>
-- Pasi
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [RFC PATCH v01] Xen PVSCSI drivers for pvops xen/stable-2.6.32.x kernel
2011-01-07 10:26 ` Ian Campbell
2011-01-07 10:32 ` Pasi Kärkkäinen
@ 2011-01-10 10:15 ` Olaf Hering
2011-01-10 13:40 ` Ian Campbell
1 sibling, 1 reply; 10+ messages in thread
From: Olaf Hering @ 2011-01-10 10:15 UTC (permalink / raw)
To: Ian Campbell
Cc: Patrick Colp, Jeremy Fitzhardinge, xen-devel@lists.xensource.com,
Nathanael Rensen
On Fri, Jan 07, Ian Campbell wrote:
> On Fri, 2011-01-07 at 10:03 +0000, Nathanael Rensen wrote:
> > On 7 January 2011 06:35, Pasi Kärkkäinen <pasik@iki.fi> wrote:
> > > Hello,
> > >
> > > http://pasik.reaktio.net/xen/patches/xen-pvscsi-drivers-linux-2.6.32.27-pvops-v01.diff
> > >
> > > This is the first version of Xen PVSCSI drivers, both the scsiback backend and
> > > scsifront frontend, ported from Novell SLES11SP1 2.6.32 Xenlinux kernel to
> > > pvops xen/stable-2.6.32.x branch.
> > >
> > > -- Pasi
> >
> > That's great. Some comments based on my experience with pvusb:
> >
> > 1) You've added GNTST_eagain into include/xen/interface/grant_table.h.
> > When porting the pvusb drivers I removed the "check and retry" loops
> > based on GNTST_eagain because I noticed that netback and blkback don't
> > do that. Looking at the grant_table code in xen-4.0-testing I don't
> > think GNTST_eagain is used now. Perhaps someone with knowledge of the
> > history of GNTST_eagain could help to clarify.
>
> IIRC GNTST_eagain was added to the classic Xen trees but never forward
> ported to pvops.
>
> It's part of the page sharing paging work (so dom0 can deal with a
> paged/shared guest rather than for paging of dom0 itself). I guess those
> don't work with pvops kernels? CC'ing the guys who are working on that
> stuff.
I have not tried the pvops kernel with xenpaging.
> (The presence of the use of msleep in the eagain case is a bit
> terrifying. Stripping it out for now and redoing it might not be a bad
> plan)
Right now the retry is required for xenpaging, otherwise the guests
filesystem will be corrupted.
In the long run the newly added waitqueue feature may be used to make
the page access transparent to the guest and dom0.
Olaf
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [RFC PATCH v01] Xen PVSCSI drivers for pvops xen/stable-2.6.32.x kernel
2011-01-10 10:15 ` Olaf Hering
@ 2011-01-10 13:40 ` Ian Campbell
0 siblings, 0 replies; 10+ messages in thread
From: Ian Campbell @ 2011-01-10 13:40 UTC (permalink / raw)
To: Olaf Hering
Cc: Patrick Colp, Jeremy Fitzhardinge, xen-devel@lists.xensource.com,
Nathanael Rensen
On Mon, 2011-01-10 at 10:15 +0000, Olaf Hering wrote:
>
> > (The presence of the use of msleep in the eagain case is a bit
> > terrifying. Stripping it out for now and redoing it might not be a
> bad
> > plan)
>
> Right now the retry is required for xenpaging, otherwise the guests
> filesystem will be corrupted.
> In the long run the newly added waitqueue feature may be used to make
> the page access transparent to the guest and dom0.
I mistakenly thought msleep was a spinning type wait, but it does
actually sleep properly so it's probably not as bad as I thought.
Ian.
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [RFC PATCH v01] Xen PVSCSI drivers for pvops xen/stable-2.6.32.x kernel
2011-01-07 10:03 Nathanael Rensen
2011-01-07 10:26 ` Ian Campbell
@ 2011-01-07 10:35 ` Pasi Kärkkäinen
1 sibling, 0 replies; 10+ messages in thread
From: Pasi Kärkkäinen @ 2011-01-07 10:35 UTC (permalink / raw)
To: Nathanael Rensen; +Cc: Jeremy Fitzhardinge, xen-devel, Ian Campbell
On Fri, Jan 07, 2011 at 06:03:59PM +0800, Nathanael Rensen wrote:
> On 7 January 2011 06:35, Pasi Kärkkäinen <pasik@iki.fi> wrote:
> > Hello,
> >
> > http://pasik.reaktio.net/xen/patches/xen-pvscsi-drivers-linux-2.6.32.27-pvops-v01.diff
> >
> > This is the first version of Xen PVSCSI drivers, both the scsiback backend and
> > scsifront frontend, ported from Novell SLES11SP1 2.6.32 Xenlinux kernel to
> > pvops xen/stable-2.6.32.x branch.
> >
> > -- Pasi
>
> That's great. Some comments based on my experience with pvusb:
>
> 1) You've added GNTST_eagain into include/xen/interface/grant_table.h.
> When porting the pvusb drivers I removed the "check and retry" loops
> based on GNTST_eagain because I noticed that netback and blkback don't
> do that. Looking at the grant_table code in xen-4.0-testing I don't
> think GNTST_eagain is used now. Perhaps someone with knowledge of the
> history of GNTST_eagain could help to clarify.
>
Yep, I commented about this on the other email.
> 2) When porting pvusb I found I had to take care with the
> gnttab_grant_foreign_access_ref calls. Following the example of
> blkback I replaced
> buffer_pfn = page_to_phys(page) >> PAGE_SHIFT;
> with
> buffer_mfn = pfn_to_mfn(page_to_pfn(page));
> The situation for pvscsi looks a bit more involved (e.g. buffer_pfn++).
>
Yeah, I had to hack that stuff aswell.. I was pretty tired so not sure
at all I did it correctly :)
I'll get back to it some day now and take another look at it.
And actually do some *testing* with the patch :)
> 3) The existing xen-netfront, xen-blkfront and xen-pcifront have been
> converted to a single .c module when ported to pvops. I followed that
> lead with pvusb. You might like to do the same for consistency.
>
Yep, I noticed this aswell. The first version was meant to be
"minimal changes just to make it compile". I need to do this change aswell.
Thanks for the comments!
-- Pasi
^ permalink raw reply [flat|nested] 10+ messages in thread
end of thread, other threads:[~2011-11-30 16:23 UTC | newest]
Thread overview: 10+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2011-01-06 22:31 [RFC PATCH v01] Xen PVSCSI drivers for pvops xen/stable-2.6.32.x kernel Pasi Kärkkäinen
2011-01-07 10:10 ` Ian Campbell
2011-01-07 10:31 ` Pasi Kärkkäinen
2011-11-30 16:23 ` Konrad Rzeszutek Wilk
-- strict thread matches above, loose matches on Subject: below --
2011-01-07 10:03 Nathanael Rensen
2011-01-07 10:26 ` Ian Campbell
2011-01-07 10:32 ` Pasi Kärkkäinen
2011-01-10 10:15 ` Olaf Hering
2011-01-10 13:40 ` Ian Campbell
2011-01-07 10:35 ` Pasi Kärkkäinen
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.