/******************************************************************************
- * arch/xen/drivers/blkif/backend/main.c
*
* Back-end of the driver for virtual block devices. This portion of the
* driver exports a 'unified' block-device interface that can be accessed
* by any operating system that implements a compatible front end. A
* reference front-end implementation can be found in:
- * arch/xen/drivers/blkif/frontend
+ * drivers/block/xen-blkfront.c
*
* Copyright (c) 2003-2004, Keir Fraser & Steve Hand
* Copyright (c) 2005, Christopher Clark
#include <linux/list.h>
#include <linux/delay.h>
#include <linux/freezer.h>
-#include <xen/balloon.h>
+
#include <xen/events.h>
#include <xen/page.h>
#include <asm/xen/hypervisor.h>
#include <asm/xen/hypercall.h>
#include "common.h"
+#define WRITE_BARRIER (REQ_WRITE | REQ_FLUSH | REQ_FUA)
+
/*
* These are rather arbitrary. They are fairly large because adjacent requests
* pulled from a communication ring are quite likely to end up being part of
MODULE_PARM_DESC(reqs, "Number of blkback requests to allocate");
/* Run-time switchable: /sys/module/blkback/parameters/ */
-static unsigned int log_stats = 0;
-static unsigned int debug_lvl = 0;
+static unsigned int log_stats;
+static unsigned int debug_lvl;
module_param(log_stats, int, 0644);
module_param(debug_lvl, int, 0644);
* the pendcnt towards zero. When it hits zero, the specified domain has a
* response queued for it, with the saved 'id' passed back.
*/
-typedef struct {
- blkif_t *blkif;
+struct pending_req {
+ struct blkif_st *blkif;
u64 id;
int nr_pages;
atomic_t pendcnt;
unsigned short operation;
int status;
struct list_head free_list;
-} pending_req_t;
-
-static pending_req_t *pending_reqs;
-static struct list_head pending_free;
-static DEFINE_SPINLOCK(pending_free_lock);
-static DECLARE_WAIT_QUEUE_HEAD(pending_free_wq);
+};
#define BLKBACK_INVALID_HANDLE (~0)
-static struct page **pending_pages;
-static grant_handle_t *pending_grant_handles;
+struct xen_blkbk {
+ struct pending_req *pending_reqs;
+ /* List of all 'pending_req' available */
+ struct list_head pending_free;
+ /* And its spinlock. */
+ spinlock_t pending_free_lock;
+ wait_queue_head_t pending_free_wq;
+ /* The list of all pages that are available. */
+ struct page **pending_pages;
+ /* And the grant handles that are available. */
+ grant_handle_t *pending_grant_handles;
+};
+
+static struct xen_blkbk *blkbk;
-static inline int vaddr_pagenr(pending_req_t *req, int seg)
+/*
+ * Little helpful macro to figure out the index and virtual address of the
+ * pending_pages[..]. For each 'pending_req' we have have up to
+ * BLKIF_MAX_SEGMENTS_PER_REQUEST (11) pages. The seg would be from 0 through
+ * 10 and would index in the pending_pages[..]. */
+static inline int vaddr_pagenr(struct pending_req *req, int seg)
{
- return (req - pending_reqs) * BLKIF_MAX_SEGMENTS_PER_REQUEST + seg;
+ return (req - blkbk->pending_reqs) *
+ BLKIF_MAX_SEGMENTS_PER_REQUEST + seg;
}
-static inline unsigned long vaddr(pending_req_t *req, int seg)
+#define pending_page(req, seg) pending_pages[vaddr_pagenr(req, seg)]
+
+static inline unsigned long vaddr(struct pending_req *req, int seg)
{
- unsigned long pfn = page_to_pfn(pending_pages[vaddr_pagenr(req, seg)]);
+ unsigned long pfn = page_to_pfn(blkbk->pending_page(req, seg));
return (unsigned long)pfn_to_kaddr(pfn);
}
#define pending_handle(_req, _seg) \
- (pending_grant_handles[vaddr_pagenr(_req, _seg)])
+ (blkbk->pending_grant_handles[vaddr_pagenr(_req, _seg)])
-static int do_block_io_op(blkif_t *blkif);
-static void dispatch_rw_block_io(blkif_t *blkif,
+static int do_block_io_op(struct blkif_st *blkif);
+static void dispatch_rw_block_io(struct blkif_st *blkif,
struct blkif_request *req,
- pending_req_t *pending_req);
-static void make_response(blkif_t *blkif, u64 id,
+ struct pending_req *pending_req);
+static void make_response(struct blkif_st *blkif, u64 id,
unsigned short op, int st);
-/******************************************************************
- * misc small helpers
+/*
+ * Retrieve from the 'pending_reqs' a free pending_req structure to be used.
*/
-static pending_req_t* alloc_req(void)
+static struct pending_req *alloc_req(void)
{
- pending_req_t *req = NULL;
+ struct pending_req *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);
+ spin_lock_irqsave(&blkbk->pending_free_lock, flags);
+ if (!list_empty(&blkbk->pending_free)) {
+ req = list_entry(blkbk->pending_free.next, struct pending_req,
+ free_list);
list_del(&req->free_list);
}
- spin_unlock_irqrestore(&pending_free_lock, flags);
+ spin_unlock_irqrestore(&blkbk->pending_free_lock, flags);
return req;
}
-static void free_req(pending_req_t *req)
+/*
+ * Return the 'pending_req' structure back to the freepool. We also
+ * wake up the thread if it was waiting for a free page.
+ */
+static void free_req(struct pending_req *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);
+ spin_lock_irqsave(&blkbk->pending_free_lock, flags);
+ was_empty = list_empty(&blkbk->pending_free);
+ list_add(&req->free_list, &blkbk->pending_free);
+ spin_unlock_irqrestore(&blkbk->pending_free_lock, flags);
if (was_empty)
- wake_up(&pending_free_wq);
-}
-
-static void unplug_queue(blkif_t *blkif)
-{
- if (blkif->plug == NULL)
- return;
- if (blkif->plug->unplug_fn)
- blkif->plug->unplug_fn(blkif->plug);
- blk_put_queue(blkif->plug);
- blkif->plug = NULL;
+ wake_up(&blkbk->pending_free_wq);
}
-static void plug_queue(blkif_t *blkif, struct block_device *bdev)
+/*
+ * Notification from the guest OS.
+ */
+static void blkif_notify_work(struct blkif_st *blkif)
{
- struct request_queue *q = bdev_get_queue(bdev);
-
- if (q == blkif->plug)
- return;
- unplug_queue(blkif);
- blk_get_queue(q);
- blkif->plug = q;
+ blkif->waiting_reqs = 1;
+ wake_up(&blkif->wq);
}
-static void fast_flush_area(pending_req_t *req)
+irqreturn_t blkif_be_int(int irq, void *dev_id)
{
- struct gnttab_unmap_grant_ref unmap[BLKIF_MAX_SEGMENTS_PER_REQUEST];
- unsigned int i, invcount = 0;
- grant_handle_t handle;
- int ret;
-
- for (i = 0; i < req->nr_pages; i++) {
- handle = pending_handle(req, i);
- if (handle == BLKBACK_INVALID_HANDLE)
- continue;
- gnttab_set_unmap_op(&unmap[invcount], vaddr(req, i),
- GNTMAP_host_map, handle);
- pending_handle(req, i) = BLKBACK_INVALID_HANDLE;
- invcount++;
- }
-
- ret = HYPERVISOR_grant_table_op(
- GNTTABOP_unmap_grant_ref, unmap, invcount);
- BUG_ON(ret);
+ blkif_notify_work(dev_id);
+ return IRQ_HANDLED;
}
-/******************************************************************
+/*
* SCHEDULER FUNCTIONS
*/
-static void print_stats(blkif_t *blkif)
+static void print_stats(struct blkif_st *blkif)
{
printk(KERN_DEBUG "%s: oo %3d | rd %4d | wr %4d | br %4d\n",
current->comm, blkif->st_oo_req,
int blkif_schedule(void *arg)
{
- blkif_t *blkif = arg;
+ struct blkif_st *blkif = arg;
+ struct vbd *vbd = &blkif->vbd;
blkif_get(blkif);
while (!kthread_should_stop()) {
if (try_to_freeze())
continue;
+ if (unlikely(vbd->size != vbd_size(vbd)))
+ vbd_resize(blkif);
wait_event_interruptible(
blkif->wq,
blkif->waiting_reqs || kthread_should_stop());
wait_event_interruptible(
- pending_free_wq,
- !list_empty(&pending_free) || kthread_should_stop());
+ blkbk->pending_free_wq,
+ !list_empty(&blkbk->pending_free) ||
+ kthread_should_stop());
blkif->waiting_reqs = 0;
smp_mb(); /* clear flag *before* checking for work */
if (do_block_io_op(blkif))
blkif->waiting_reqs = 1;
- unplug_queue(blkif);
if (log_stats && time_after(jiffies, blkif->st_print))
print_stats(blkif);
return 0;
}
-/******************************************************************
- * COMPLETION CALLBACK -- Called as bh->b_end_io()
+/*
+ * Unmap the grant references, and also remove the M2P over-rides
+ * used in the 'pending_req'.
+*/
+static void fast_flush_area(struct pending_req *req)
+{
+ struct gnttab_unmap_grant_ref unmap[BLKIF_MAX_SEGMENTS_PER_REQUEST];
+ unsigned int i, invcount = 0;
+ grant_handle_t handle;
+ int ret;
+
+ for (i = 0; i < req->nr_pages; i++) {
+ handle = pending_handle(req, i);
+ if (handle == BLKBACK_INVALID_HANDLE)
+ continue;
+ gnttab_set_unmap_op(&unmap[invcount], vaddr(req, i),
+ GNTMAP_host_map, handle);
+ pending_handle(req, i) = BLKBACK_INVALID_HANDLE;
+ invcount++;
+ }
+
+ ret = HYPERVISOR_grant_table_op(
+ GNTTABOP_unmap_grant_ref, unmap, invcount);
+ BUG_ON(ret);
+ /* Note, we use invcount, so nr->pages, so we can't index
+ * using vaddr(req, i).
+ */
+ for (i = 0; i < invcount; i++) {
+ ret = m2p_remove_override(
+ virt_to_page(unmap[i].host_addr), false);
+ if (ret) {
+ printk(KERN_ALERT "Failed to remove M2P override for " \
+ "%lx\n", (unsigned long)unmap[i].host_addr);
+ continue;
+ }
+ }
+}
+/*
+ * Completion callback on the bio's. Called as bh->b_end_io()
*/
-static void __end_block_io_op(pending_req_t *pending_req, int error)
+static void __end_block_io_op(struct pending_req *pending_req, int error)
{
/* An error fails the entire request. */
if ((pending_req->operation == BLKIF_OP_WRITE_BARRIER) &&
pending_req->status = BLKIF_RSP_ERROR;
}
+ /* If all of the bio's have completed it is time to unmap
+ * the grant references associated with 'request' and provide
+ * the proper response on the ring.
+ */
if (atomic_dec_and_test(&pending_req->pendcnt)) {
fast_flush_area(pending_req);
make_response(pending_req->blkif, pending_req->id,
}
}
+/*
+ * bio callback.
+ */
static void end_block_io_op(struct bio *bio, int error)
{
__end_block_io_op(bio->bi_private, error);
}
-/******************************************************************************
- * NOTIFICATION FROM GUEST OS.
- */
-
-static void blkif_notify_work(blkif_t *blkif)
-{
- blkif->waiting_reqs = 1;
- wake_up(&blkif->wq);
-}
-
-irqreturn_t blkif_be_int(int irq, void *dev_id)
-{
- blkif_notify_work(dev_id);
- return IRQ_HANDLED;
-}
-
-
-/******************************************************************
- * DOWNWARD CALLS -- These interface with the block-device layer proper.
+/*
+ * Function to copy the from the ring buffer the 'struct blkif_request'
+ * (which has the sectors we want, number of them, grant references, etc),
+ * and transmute it to the block API to hand it over to the proper block disk.
*/
-
-static int do_block_io_op(blkif_t *blkif)
+static int do_block_io_op(struct blkif_st *blkif)
{
union blkif_back_rings *blk_rings = &blkif->blk_rings;
struct blkif_request req;
- pending_req_t *pending_req;
+ struct pending_req *pending_req;
RING_IDX rc, rp;
int more_to_do = 0;
return more_to_do;
}
-static void dispatch_rw_block_io(blkif_t *blkif,
+/*
+ * Transumation of the 'struct blkif_request' to a proper 'struct bio'
+ * and call the 'submit_bio' to pass it to the underlaying storage.
+ */
+static void dispatch_rw_block_io(struct blkif_st *blkif,
struct blkif_request *req,
- pending_req_t *pending_req)
+ struct pending_req *pending_req)
{
- extern void ll_rw_block(int rw, int nr, struct buffer_head * bhs[]);
struct gnttab_map_grant_ref map[BLKIF_MAX_SEGMENTS_PER_REQUEST];
struct phys_req preq;
struct {
} seg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
unsigned int nseg;
struct bio *bio = NULL;
- int ret, i;
+ struct bio *biolist[BLKIF_MAX_SEGMENTS_PER_REQUEST];
+ int ret, i, nbio = 0;
int operation;
+ struct blk_plug plug;
+ struct request_queue *q;
switch (req->operation) {
case BLKIF_OP_READ:
BUG();
}
- /* Check that number of segments is sane. */
+ /* Check that the number of segments is sane. */
nseg = req->nr_segments;
if (unlikely(nseg == 0 && operation != WRITE_BARRIER) ||
unlikely(nseg > BLKIF_MAX_SEGMENTS_PER_REQUEST)) {
}
preq.dev = req->handle;
- preq.sector_number = req->sector_number;
+ preq.sector_number = req->u.rw.sector_number;
preq.nr_sects = 0;
pending_req->blkif = blkif;
pending_req->status = BLKIF_RSP_OKAY;
pending_req->nr_pages = nseg;
+ /* Fill out preq.nr_sects with proper amount of sectors, and setup
+ * assign map[..] with the PFN of the page in our domain with the
+ * corresponding grant reference for each page.
+ */
for (i = 0; i < nseg; i++) {
uint32_t flags;
- seg[i].nsec = req->seg[i].last_sect -
- req->seg[i].first_sect + 1;
-
- if ((req->seg[i].last_sect >= (PAGE_SIZE >> 9)) ||
- (req->seg[i].last_sect < req->seg[i].first_sect))
+ seg[i].nsec = req->u.rw.seg[i].last_sect -
+ req->u.rw.seg[i].first_sect + 1;
+ if ((req->u.rw.seg[i].last_sect >= (PAGE_SIZE >> 9)) ||
+ (req->u.rw.seg[i].last_sect < req->u.rw.seg[i].first_sect))
goto fail_response;
preq.nr_sects += seg[i].nsec;
if (operation != READ)
flags |= GNTMAP_readonly;
gnttab_set_map_op(&map[i], vaddr(pending_req, i), flags,
- req->seg[i].gref, blkif->domid);
+ req->u.rw.seg[i].gref, blkif->domid);
}
ret = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, map, nseg);
BUG_ON(ret);
+ /* Now swizzel the MFN in our domain with the MFN from the other domain
+ * so that when we access vaddr(pending_req,i) it has the contents of
+ * the page from the other domain.
+ */
for (i = 0; i < nseg; i++) {
if (unlikely(map[i].status != 0)) {
DPRINTK("invalid buffer -- could not remap it\n");
if (ret)
continue;
- set_phys_to_machine(__pa(vaddr(
- pending_req, i)) >> PAGE_SHIFT,
- FOREIGN_FRAME(map[i].dev_bus_addr >> PAGE_SHIFT));
+ ret = m2p_add_override(PFN_DOWN(map[i].dev_bus_addr),
+ blkbk->pending_page(pending_req, i), false);
+ if (ret) {
+ printk(KERN_ALERT "Failed to install M2P override for"\
+ " %lx (ret: %d)\n", (unsigned long)
+ map[i].dev_bus_addr, ret);
+ /* We could switch over to GNTTABOP_copy */
+ continue;
+ }
+
seg[i].buf = map[i].dev_bus_addr |
- (req->seg[i].first_sect << 9);
+ (req->u.rw.seg[i].first_sect << 9);
}
+ /* If we have failed at this point, we need to undo the M2P override,
+ * set gnttab_set_unmap_op on all of the grant references and perform
+ * the hypercall to unmap the grants - that is all done in
+ * fast_flush_area.
+ */
if (ret)
goto fail_flush;
goto fail_flush;
}
- plug_queue(blkif, preq.bdev);
- atomic_set(&pending_req->pendcnt, 1);
+ /* This corresponding blkif_put is done in __end_block_io_op */
blkif_get(blkif);
for (i = 0; i < nseg; i++) {
if (((int)preq.sector_number|(int)seg[i].nsec) &
- ((bdev_hardsect_size(preq.bdev) >> 9) - 1)) {
+ ((bdev_logical_block_size(preq.bdev) >> 9) - 1)) {
DPRINTK("Misaligned I/O request from domain %d",
blkif->domid);
goto fail_put_bio;
while ((bio == NULL) ||
(bio_add_page(bio,
- virt_to_page(vaddr(pending_req, i)),
+ blkbk->pending_page(pending_req, i),
seg[i].nsec << 9,
seg[i].buf & ~PAGE_MASK) == 0)) {
- if (bio) {
- atomic_inc(&pending_req->pendcnt);
- submit_bio(operation, bio);
- }
- bio = bio_alloc(GFP_KERNEL, nseg-i);
+ bio = biolist[nbio++] = bio_alloc(GFP_KERNEL, nseg-i);
if (unlikely(bio == NULL))
goto fail_put_bio;
preq.sector_number += seg[i].nsec;
}
+ /* This will be hit if the operation was a barrier. */
if (!bio) {
BUG_ON(operation != WRITE_BARRIER);
- bio = bio_alloc(GFP_KERNEL, 0);
+ bio = biolist[nbio++] = bio_alloc(GFP_KERNEL, 0);
if (unlikely(bio == NULL))
goto fail_put_bio;
bio->bi_sector = -1;
}
- submit_bio(operation, bio);
+
+ /* We set it one so that the last submit_bio does not have to call
+ * atomic_inc.
+ */
+ atomic_set(&pending_req->pendcnt, nbio);
+
+ /* Get a reference count for the disk queue and start sending I/O */
+ blk_get_queue(q);
+ blk_start_plug(&plug);
+
+ for (i = 0; i < nbio; i++)
+ submit_bio(operation, biolist[i]);
+
+ blk_finish_plug(&plug);
+ /* Let the I/Os go.. */
+ blk_put_queue(q);
if (operation == READ)
blkif->st_rd_sect += preq.nr_sects;
fail_flush:
fast_flush_area(pending_req);
fail_response:
+ /* Haven't submitted any bio's yet. */
make_response(blkif, req->id, req->operation, BLKIF_RSP_ERROR);
free_req(pending_req);
msleep(1); /* back off a bit */
return;
fail_put_bio:
+ for (i = 0; i < (nbio-1); i++)
+ bio_put(biolist[i]);
__end_block_io_op(pending_req, -EINVAL);
- if (bio)
- bio_put(bio);
- unplug_queue(blkif);
msleep(1); /* back off a bit */
return;
}
-/******************************************************************
- * MISCELLANEOUS SETUP / TEARDOWN / DEBUGGING
+/*
+ * Put a response on the ring on how the operation fared.
*/
-
-
-static void make_response(blkif_t *blkif, u64 id,
+static void make_response(struct blkif_st *blkif, u64 id,
unsigned short op, int st)
{
struct blkif_response resp;
static int __init blkif_init(void)
{
int i, mmap_pages;
+ int rc = 0;
- printk(KERN_CRIT "***blkif_init\n");
if (!xen_pv_domain())
return -ENODEV;
+ blkbk = kzalloc(sizeof(struct xen_blkbk), GFP_KERNEL);
+ if (!blkbk) {
+ printk(KERN_ALERT "%s: out of memory!\n", __func__);
+ return -ENOMEM;
+ }
+
mmap_pages = blkif_reqs * BLKIF_MAX_SEGMENTS_PER_REQUEST;
- pending_reqs = kmalloc(sizeof(pending_reqs[0]) *
+ blkbk->pending_reqs = kmalloc(sizeof(blkbk->pending_reqs[0]) *
blkif_reqs, GFP_KERNEL);
- pending_grant_handles = kmalloc(sizeof(pending_grant_handles[0]) *
+ blkbk->pending_grant_handles = kzalloc(sizeof(blkbk->pending_grant_handles[0]) *
+ mmap_pages, GFP_KERNEL);
+ blkbk->pending_pages = kzalloc(sizeof(blkbk->pending_pages[0]) *
mmap_pages, GFP_KERNEL);
- pending_pages = alloc_empty_pages_and_pagevec(mmap_pages);
- if (!pending_reqs || !pending_grant_handles || !pending_pages)
+ if (!blkbk->pending_reqs || !blkbk->pending_grant_handles ||
+ !blkbk->pending_pages) {
+ rc = -ENOMEM;
goto out_of_memory;
+ }
- for (i = 0; i < mmap_pages; i++)
- pending_grant_handles[i] = BLKBACK_INVALID_HANDLE;
+ for (i = 0; i < mmap_pages; i++) {
+ blkbk->pending_grant_handles[i] = BLKBACK_INVALID_HANDLE;
+ blkbk->pending_pages[i] = alloc_page(GFP_KERNEL);
+ if (blkbk->pending_pages[i] == NULL) {
+ rc = -ENOMEM;
+ goto out_of_memory;
+ }
+ }
+ rc = blkif_interface_init();
+ if (rc)
+ goto failed_init;
- blkif_interface_init();
+ memset(blkbk->pending_reqs, 0, sizeof(blkbk->pending_reqs));
- memset(pending_reqs, 0, sizeof(pending_reqs));
- INIT_LIST_HEAD(&pending_free);
+ INIT_LIST_HEAD(&blkbk->pending_free);
+ spin_lock_init(&blkbk->pending_free_lock);
+ init_waitqueue_head(&blkbk->pending_free_wq);
for (i = 0; i < blkif_reqs; i++)
- list_add_tail(&pending_reqs[i].free_list, &pending_free);
+ list_add_tail(&blkbk->pending_reqs[i].free_list,
+ &blkbk->pending_free);
- blkif_xenbus_init();
+ rc = blkif_xenbus_init();
+ if (rc)
+ goto failed_init;
return 0;
out_of_memory:
- kfree(pending_reqs);
- kfree(pending_grant_handles);
- free_empty_pages_and_pagevec(pending_pages, mmap_pages);
- printk("%s: out of memory\n", __FUNCTION__);
- return -ENOMEM;
+ printk(KERN_ERR "%s: out of memory\n", __func__);
+ failed_init:
+ kfree(blkbk->pending_reqs);
+ kfree(blkbk->pending_grant_handles);
+ for (i = 0; i < mmap_pages; i++) {
+ if (blkbk->pending_pages[i])
+ __free_page(blkbk->pending_pages[i]);
+ }
+ kfree(blkbk->pending_pages);
+ kfree(blkbk);
+ blkbk = NULL;
+ return rc;
}
module_init(blkif_init);