From: Juergen Gross <jgross@xxxxxxxx>
Introduces the XEN pvSCSI frontend. With pvSCSI it is possible for a XEN domU
to issue SCSI commands to a SCSI LUN assigned to that domU. The SCSI commands
are passed to the pvSCSI backend in a driver domain (usually Dom0) which is
owner of the physical device. This allows e.g. to use SCSI tape drives in a
XEN domU.
The code is taken from the pvSCSI implementation in XEN done by Fujitsu based
on Linux kernel 2.6.18.
Changes from the original version are:
- port to upstream kernel
- put all code in just one source file
- move module to appropriate location in kernel tree
- adapt to Linux style guide
- some minor code simplifications
- replace constants with defines
- remove not used defines
- add support for larger SG lists by putting them in a granted page
Signed-off-by: Juergen Gross <jgross@xxxxxxxx>
---
drivers/scsi/Kconfig | 9 +
drivers/scsi/Makefile | 1 +
drivers/scsi/xen-scsifront.c | 1006 ++++++++++++++++++++++++++++++++++++++++++
3 files changed, 1016 insertions(+)
create mode 100644 drivers/scsi/xen-scsifront.c
diff --git a/drivers/scsi/Kconfig b/drivers/scsi/Kconfig
index baca589..e860c16 100644
--- a/drivers/scsi/Kconfig
+++ b/drivers/scsi/Kconfig
@@ -611,6 +611,15 @@ config VMWARE_PVSCSI
To compile this driver as a module, choose M here: the
module will be called vmw_pvscsi.
+config XEN_SCSI_FRONTEND
+ tristate "XEN SCSI frontend driver"
+ depends on SCSI && XEN
+ help
+ The XEN SCSI frontend driver allows the kernel to access SCSI Devices
+ within another guest OS (usually Dom0).
+ Only needed if the kernel is running in a XEN guest and generic
+ SCSI access to a device is needed.
+
config HYPERV_STORAGE
tristate "Microsoft Hyper-V virtual storage driver"
depends on SCSI && HYPERV
diff --git a/drivers/scsi/Makefile b/drivers/scsi/Makefile
index e172d4f..a4ee9c5 100644
--- a/drivers/scsi/Makefile
+++ b/drivers/scsi/Makefile
@@ -144,6 +144,7 @@ obj-$(CONFIG_SCSI_ESAS2R) += esas2r/
obj-$(CONFIG_SCSI_PMCRAID) += pmcraid.o
obj-$(CONFIG_SCSI_VIRTIO) += virtio_scsi.o
obj-$(CONFIG_VMWARE_PVSCSI) += vmw_pvscsi.o
+obj-$(CONFIG_XEN_SCSI_FRONTEND) += xen-scsifront.o
obj-$(CONFIG_HYPERV_STORAGE) += hv_storvsc.o
obj-$(CONFIG_ARM) += arm/
diff --git a/drivers/scsi/xen-scsifront.c b/drivers/scsi/xen-scsifront.c
new file mode 100644
index 0000000..469fc54
--- /dev/null
+++ b/drivers/scsi/xen-scsifront.c
@@ -0,0 +1,1006 @@
+/*
+ * 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.
+ */
+
+/*
+ * Patched to support >2TB drives
+ * 2010, Samuel Kvasnica, IMS Nanofabrication AG
+ */
+
+#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 <linux/pfn.h>
+#include <linux/slab.h>
+
+#include <scsi/scsi_cmnd.h>
+#include <scsi/scsi_device.h>
+#include <scsi/scsi.h>
+#include <scsi/scsi_host.h>
+
+#include <xen/xen.h>
+#include <xen/xenbus.h>
+#include <xen/grant_table.h>
+#include <xen/events.h>
+#include <xen/page.h>
+
+#include <xen/interface/grant_table.h>
+#include <xen/interface/io/vscsiif.h>
+#include <xen/interface/io/protocols.h>
+
+#include <asm/xen/hypervisor.h>
+
+
+#define GRANT_INVALID_REF 0
+
+#define VSCSIFRONT_OP_ADD_LUN 1
+#define VSCSIFRONT_OP_DEL_LUN 2
+
+#define DEFAULT_TASK_COMM_LEN TASK_COMM_LEN
+
+/* 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
+
+#define vscsiif_grants_sg(_sg) (PFN_UP((_sg) * \
+ sizeof(struct scsiif_request_segment)))
+
+struct vscsifrnt_shadow {
+ uint16_t next_free;
+
+ /* command between backend and frontend */
+ unsigned char act;
+
+ /* Number of pieces of scatter-gather */
+ unsigned int nr_grants;
+ struct scsiif_request_segment *sg;
+
+ /* do reset or abort 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) */
+
+ /* requested struct scsi_cmnd is stored from kernel */
+ struct scsi_cmnd *sc;
+ int gref[vscsiif_grants_sg(SG_ALL) + SG_ALL];
+};
+
+struct vscsifrnt_info {
+ struct xenbus_device *dev;
+
+ struct Scsi_Host *host;
+
+ 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;
+ wait_queue_head_t wq_sync;
+ unsigned int waiting_resp:1;
+ unsigned int waiting_sync:1;
+};
+
+#define DPRINTK(_f, _a...) \
+ pr_debug("(file=%s, line=%d) " _f, __FILE__ , __LINE__ , ## _a)
+
+#define PREFIX(lvl) KERN_##lvl "scsifront: "
+
+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 = VSCSIIF_MAX_REQS;
+ 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)
+{
+ info->shadow[id].next_free = info->shadow_free;
+ info->shadow[id].sc = NULL;
+ info->shadow_free = id;
+}
+
+static void add_id_to_freelist(struct vscsifrnt_info *info, uint32_t id)
+{
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->shadow_lock, flags);
+ _add_id_to_freelist(info, id);
+ spin_unlock_irqrestore(&info->shadow_lock, flags);
+}
+
+static struct vscsiif_request *scsifront_pre_req(struct vscsifrnt_info *info)
+{
+ struct vscsiif_front_ring *ring = &(info->ring);
+ struct vscsiif_request *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);
+ int notify;
+
+ RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(ring, notify);
+ if (notify)
+ notify_remote_via_irq(info->irq);
+}
+
+static 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_info *info, uint32_t id)
+{
+ struct vscsifrnt_shadow *s = &info->shadow[id];
+ int i;
+
+ if (s->sc->sc_data_direction == DMA_NONE)
+ return;
+
+ for (i = 0; i < s->nr_grants; i++) {
+ if (unlikely(gnttab_query_foreign_access(s->gref[i]) != 0)) {
+ shost_printk(PREFIX(ALERT), info->host,
+ "grant still in use by backend\n");
+ BUG();
+ }
+ gnttab_end_foreign_access(s->gref[i], 0, 0UL);
+ }
+
+ kfree(s->sg);
+ s->sg = NULL;
+}
+
+static void scsifront_cdb_cmd_done(struct vscsifrnt_info *info,
+ struct vscsiif_response *ring_res)
+{
+ struct scsi_cmnd *sc;
+ uint32_t id;
+ uint8_t sense_len;
+
+ id = ring_res->rqid;
+ sc = info->shadow[id].sc;
+
+ BUG_ON(sc == NULL);
+
+ scsifront_gnttab_done(info, id);
+ add_id_to_freelist(info, id);
+
+ sc->result = ring_res->rslt;
+ scsi_set_resid(sc, ring_res->residual_len);
+
+ sense_len = min_t(uint8_t, VSCSIIF_SENSE_BUFFERSIZE,
+ ring_res->sense_len);
+
+ if (sense_len)
+ memcpy(sc->sense_buffer, ring_res->sense_buffer, sense_len);
+
+ sc->scsi_done(sc);
+}
+
+static void scsifront_sync_cmd_done(struct vscsifrnt_info *info,
+ struct vscsiif_response *ring_res)
+{
+ uint16_t id = ring_res->rqid;
+ unsigned long flags;
+
+ spin_lock_irqsave(&info->shadow_lock, flags);
+ info->shadow[id].wait_reset = 1;
+ switch (info->shadow[id].rslt_reset) {
+ case 0:
+ info->shadow[id].rslt_reset = ring_res->rslt;
+ break;
+ case -1:
+ _add_id_to_freelist(info, id);
+ break;
+ default:
+ shost_printk(PREFIX(ERR), info->host,
+ "bad reset state %d, possibly leaking %u\n",
+ info->shadow[id].rslt_reset, id);
+ break;
+ }
+ spin_unlock_irqrestore(&info->shadow_lock, flags);
+
+ wake_up(&(info->shadow[id].wq_reset));
+}
+
+static int scsifront_cmd_done(struct vscsifrnt_info *info)
+{
+ struct vscsiif_response *ring_res;
+ RING_IDX i, rp;
+ int more_to_do = 0;
+ unsigned long flags;
+
+ spin_lock_irqsave(info->host->host_lock, flags);
+
+ rp = info->ring.sring->rsp_prod;
+ rmb(); /* ordering required respective to dom0 */
+ 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;
+
+ info->waiting_sync = 0;
+
+ spin_unlock_irqrestore(info->host->host_lock, flags);
+
+ wake_up(&info->wq_sync);
+
+ /* Yield point for this unbounded loop. */
+ cond_resched();
+
+ return more_to_do;
+}
+
+static 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(); /* other thread uses waiting_resp */
+
+ 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,
+ struct vscsiif_request *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, off, len, bytes;
+ unsigned int data_len = scsi_bufflen(sc);
+ unsigned int data_grants = 0, seg_grants = 0;
+ struct scatterlist *sg;
+ unsigned long mfn;
+ struct scsiif_request_segment *seg;
+
+ ring_req->nr_segments = 0;
+ if (sc->sc_data_direction == DMA_NONE || !data_len)
+ return 0;
+
+ scsi_for_each_sg(sc, sg, scsi_sg_count(sc), i)
+ data_grants += PFN_UP(sg->offset + sg->length);
+
+ if (data_grants > VSCSIIF_SG_TABLESIZE) {
+ if (data_grants > info->host->sg_tablesize) {
+ shost_printk(PREFIX(ERR), info->host,
+ "Unable to map request_buffer for command!\n");
+ return -E2BIG;
+ }
+ seg_grants = vscsiif_grants_sg(data_grants);
+ info->shadow[id].sg = kcalloc(data_grants,
+ sizeof(struct scsiif_request_segment), GFP_NOIO);
+ if (!info->shadow[id].sg)
+ return -ENOMEM;
+ }
+ seg = info->shadow[id].sg ? : ring_req->seg;
+
+ err = gnttab_alloc_grant_references(seg_grants + data_grants,
+ &gref_head);
+ if (err) {
+ kfree(info->shadow[id].sg);
+ info->shadow[id].sg = NULL;
+ shost_printk(PREFIX(ERR), info->host,
+ "gnttab_alloc_grant_references() error\n");
+ return -ENOMEM;
+ }
+
+ if (seg_grants) {
+ page = virt_to_page(seg);
+ off = (unsigned long)seg & ~PAGE_MASK;
+ len = sizeof(struct scsiif_request_segment) * data_grants;
+ while (len > 0) {
+ bytes = min_t(unsigned int, len, PAGE_SIZE - off);
+
+ ref = gnttab_claim_grant_reference(&gref_head);
+ BUG_ON(ref == -ENOSPC);
+
+ mfn = pfn_to_mfn(page_to_pfn(page));
+ gnttab_grant_foreign_access_ref(ref,
+ info->dev->otherend_id, mfn, 1);
+ 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;
+
+ page++;
+ len -= bytes;
+ off = 0;
+ ref_cnt++;
+ }
+ BUG_ON(seg_grants < ref_cnt);
+ seg_grants = ref_cnt;
+ }
+
+ scsi_for_each_sg(sc, sg, scsi_sg_count(sc), i) {
+ page = sg_page(sg);
+ off = sg->offset;
+ len = sg->length;
+
+ 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);
+
+ mfn = pfn_to_mfn(page_to_pfn(page));
+ gnttab_grant_foreign_access_ref(ref,
+ info->dev->otherend_id, mfn, write);
+
+ info->shadow[id].gref[ref_cnt] = ref;
+ seg->gref = ref;
+ seg->offset = (uint16_t)off;
+ seg->length = (uint16_t)bytes;
+
+ page++;
+ seg++;
+ len -= bytes;
+ data_len -= bytes;
+ off = 0;
+ ref_cnt++;
+ }
+ }
+
+ if (seg_grants)
+ ring_req->nr_segments = VSCSIIF_SG_GRANT | seg_grants;
+ else
+ ring_req->nr_segments = (uint8_t)ref_cnt;
+ info->shadow[id].nr_grants = ref_cnt;
+
+ return 0;
+}
+
+static struct vscsiif_request *scsifront_command2ring(
+ struct vscsifrnt_info *info, struct scsi_cmnd *sc)
+{
+ struct vscsiif_request *ring_req;
+
+ ring_req = scsifront_pre_req(info);
+
+ 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;
+
+ return ring_req;
+}
+
+static int scsifront_queuecommand(struct Scsi_Host *shost,
+ struct scsi_cmnd *sc)
+{
+ struct vscsifrnt_info *info = shost_priv(shost);
+ struct vscsiif_request *ring_req;
+ unsigned long flags;
+ int err;
+ uint16_t rqid;
+
+/* debug printk to identify more missing scsi commands
+ shost_printk(KERN_INFO "scsicmd: ", sc->device->host,
+ "len=%u %#x,%#x,%#x,%#x,%#x,%#x,%#x,%#x,%#x,%#x\n",
+ sc->cmd_len, sc->cmnd[0], sc->cmnd[1],
+ sc->cmnd[2], sc->cmnd[3], sc->cmnd[4], sc->cmnd[5],
+ sc->cmnd[6], sc->cmnd[7], sc->cmnd[8], sc->cmnd[9]);
+*/
+ spin_lock_irqsave(shost->host_lock, flags);
+ scsi_cmd_get_serial(shost, sc);
+ if (RING_FULL(&info->ring)) {
+ spin_unlock_irqrestore(shost->host_lock, flags);
+ return SCSI_MLQUEUE_HOST_BUSY;
+ }
+
+ sc->result = 0;
+
+ ring_req = scsifront_command2ring(info, sc);
+ rqid = ring_req->rqid;
+ ring_req->act = VSCSIIF_ACT_SCSI_CDB;
+
+ info->shadow[rqid].sc = sc;
+ info->shadow[rqid].act = VSCSIIF_ACT_SCSI_CDB;
+
+ err = map_data_for_request(info, sc, ring_req, rqid);
+ if (err < 0) {
+ add_id_to_freelist(info, rqid);
+ spin_unlock_irqrestore(shost->host_lock, flags);
+ if (err == -ENOMEM)
+ return SCSI_MLQUEUE_HOST_BUSY;
+ sc->result = DID_ERROR << 16;
+ sc->scsi_done(sc);
+ return 0;
+ }
+
+ scsifront_do_request(info);
+ spin_unlock_irqrestore(shost->host_lock, flags);
+
+ return 0;
+}
+
+static int scsifront_action_handler(struct scsi_cmnd *sc, uint8_t act)
+{
+ struct Scsi_Host *host = sc->device->host;
+ struct vscsifrnt_info *info = shost_priv(host);
+ struct vscsiif_request *ring_req;
+ uint16_t rqid;
+ int err = 0;
+
+ for (;;) {
+ spin_lock_irq(host->host_lock);
+ if (!RING_FULL(&info->ring))
+ break;
+ if (err) {
+ spin_unlock_irq(host->host_lock);
+ return FAILED;
+ }
+ info->waiting_sync = 1;
+ spin_unlock_irq(host->host_lock);
+ err = wait_event_interruptible(info->wq_sync,
+ !info->waiting_sync);
+ spin_lock_irq(host->host_lock);
+ }
+
+ ring_req = scsifront_command2ring(info, sc);
+ rqid = ring_req->rqid;
+ ring_req->act = act;
+
+ info->shadow[rqid].act = act;
+ info->shadow[rqid].rslt_reset = 0;
+
+ ring_req->nr_segments = 0;
+
+ scsifront_do_request(info);
+
+ spin_unlock_irq(host->host_lock);
+ err = wait_event_interruptible(info->shadow[rqid].wq_reset,
+ info->shadow[rqid].wait_reset);
+ spin_lock_irq(host->host_lock);
+
+ if (!err) {
+ err = info->shadow[rqid].rslt_reset;
+ add_id_to_freelist(info, rqid);
+ } else {
+ spin_lock(&info->shadow_lock);
+ info->shadow[rqid].rslt_reset = -1;
+ spin_unlock(&info->shadow_lock);
+ err = FAILED;
+ }
+
+ spin_unlock_irq(host->host_lock);
+ return err;
+}
+
+static int scsifront_eh_abort_handler(struct scsi_cmnd *sc)
+{
+ return scsifront_action_handler(sc, VSCSIIF_ACT_SCSI_ABORT);
+}
+
+static int scsifront_dev_reset_handler(struct scsi_cmnd *sc)
+{
+ return scsifront_action_handler(sc, VSCSIIF_ACT_SCSI_RESET);
+}
+
+static 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 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;
+ info->evtchn = 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,
+ 0, "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", __func__);
+
+ 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_free = 0;
+
+ err = scsifront_init_ring(info);
+ if (err) {
+ scsi_host_put(host);
+ return err;
+ }
+
+ init_waitqueue_head(&info->wq);
+ init_waitqueue_head(&info->wq_sync);
+ 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;
+ dev_err(&dev->dev, "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 = (host->sg_tablesize - 1) * PAGE_SIZE / 512;
+ host->max_cmd_len = VSCSIIF_MAX_COMMAND_SIZE;
+
+ err = scsi_add_host(host, &dev->dev);
+ if (err) {
+ dev_err(&dev->dev, "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", __func__, 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", __func__, 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;
+}
+
+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) {
+ dev_err(&dev->dev,
+ "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);
+}
+
+static void scsifront_read_backend_params(struct xenbus_device *dev,
+ struct vscsifrnt_info *info)
+{
+ unsigned int sg_grant;
+ int ret;
+ struct Scsi_Host *host = info->host;
+
+ ret = xenbus_scanf(XBT_NIL, dev->otherend, "feature-sg-grant", "%u",
+ &sg_grant);
+ if (ret == 1 && sg_grant) {
+ sg_grant = min_t(unsigned int, sg_grant, SG_ALL);
+ host->sg_tablesize = min_t(unsigned int, sg_grant,
+ VSCSIIF_SG_TABLESIZE * PAGE_SIZE /
+ sizeof(struct scsiif_request_segment));
+ dev_info(&dev->dev, "using up to %d SG entries\n",
+ host->sg_tablesize);
+ host->max_sectors = (host->sg_tablesize - 1) * PAGE_SIZE / 512;
+ }
+}
+
+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 XenbusStateInitialised:
+ break;
+
+ case XenbusStateConnected:
+ scsifront_read_backend_params(dev, info);
+ if (xenbus_read_driver_state(dev->nodename) ==
+ XenbusStateInitialised) {
+ scsifront_do_lun_hotplug(info, VSCSIFRONT_OP_ADD_LUN);
+ }
+
+ if (dev->state != XenbusStateConnected)
+ xenbus_switch_state(dev, XenbusStateConnected);
+ break;
+
+ case XenbusStateClosed:
+ if (dev->state == XenbusStateClosed)
+ break;
+ /* Missed the backend's Closing state -- fallthrough */
+ 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 const struct xenbus_device_id scsifront_ids[] = {
+ { "vscsi" },
+ { "" }
+};
+
+static DEFINE_XENBUS_DRIVER(scsifront, ,
+ .probe = scsifront_probe,
+ .remove = scsifront_remove,
+/* .resume = scsifront_resume, */
+ .otherend_changed = scsifront_backend_changed,
+);
+
+static int __init scsifront_init(void)
+{
+ if (!xen_domain())
+ return -ENODEV;
+
+ return xenbus_register_frontend(&scsifront_driver);
+}
+module_init(scsifront_init);
+
+static void __exit scsifront_exit(void)
+{
+ xenbus_unregister_driver(&scsifront_driver);
+}
+module_exit(scsifront_exit);
+
+MODULE_DESCRIPTION("Xen SCSI frontend driver");
+MODULE_LICENSE("GPL");
+MODULE_ALIAS("xen:vscsi");
--
1.8.4.5
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
