2011-04-14 22:25:47 +00:00
|
|
|
/******************************************************************************
|
|
|
|
*
|
|
|
|
* 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:
|
2011-04-14 21:05:23 +00:00
|
|
|
* drivers/block/xen-blkfront.c
|
2011-04-14 22:25:47 +00:00
|
|
|
*
|
|
|
|
* Copyright (c) 2003-2004, Keir Fraser & Steve Hand
|
|
|
|
* Copyright (c) 2005, Christopher Clark
|
|
|
|
*
|
|
|
|
* 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>
|
2009-02-09 20:05:51 +00:00
|
|
|
#include <linux/freezer.h>
|
2009-09-15 21:12:37 +00:00
|
|
|
|
2009-02-09 20:05:51 +00:00
|
|
|
#include <xen/events.h>
|
|
|
|
#include <xen/page.h>
|
|
|
|
#include <asm/xen/hypervisor.h>
|
|
|
|
#include <asm/xen/hypercall.h>
|
2011-04-14 22:25:47 +00:00
|
|
|
#include "common.h"
|
|
|
|
|
|
|
|
/*
|
|
|
|
* 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
|
|
|
|
* the same scatter/gather request at the disc.
|
|
|
|
*
|
2011-04-20 15:50:43 +00:00
|
|
|
* ** TRY INCREASING 'xen_blkif_reqs' IF WRITE SPEEDS SEEM TOO LOW **
|
2011-04-14 22:25:47 +00:00
|
|
|
*
|
|
|
|
* This will increase the chances of being able to write whole tracks.
|
|
|
|
* 64 should be enough to keep us competitive with Linux.
|
|
|
|
*/
|
2011-04-20 15:50:43 +00:00
|
|
|
static int xen_blkif_reqs = 64;
|
|
|
|
module_param_named(reqs, xen_blkif_reqs, int, 0);
|
2011-04-14 22:25:47 +00:00
|
|
|
MODULE_PARM_DESC(reqs, "Number of blkback requests to allocate");
|
|
|
|
|
|
|
|
/* Run-time switchable: /sys/module/blkback/parameters/ */
|
2011-04-14 21:42:07 +00:00
|
|
|
static unsigned int log_stats;
|
2011-04-14 22:25:47 +00:00
|
|
|
module_param(log_stats, int, 0644);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Each outstanding request that we've passed to the lower device layers has a
|
|
|
|
* 'pending_req' allocated to it. Each buffer_head that completes decrements
|
|
|
|
* the pendcnt towards zero. When it hits zero, the specified domain has a
|
|
|
|
* response queued for it, with the saved 'id' passed back.
|
|
|
|
*/
|
2011-04-14 21:42:07 +00:00
|
|
|
struct pending_req {
|
2011-05-12 20:47:48 +00:00
|
|
|
struct xen_blkif *blkif;
|
2011-05-11 19:57:09 +00:00
|
|
|
u64 id;
|
|
|
|
int nr_pages;
|
|
|
|
atomic_t pendcnt;
|
|
|
|
unsigned short operation;
|
|
|
|
int status;
|
|
|
|
struct list_head free_list;
|
2011-04-14 21:42:07 +00:00
|
|
|
};
|
2011-04-14 22:25:47 +00:00
|
|
|
|
|
|
|
#define BLKBACK_INVALID_HANDLE (~0)
|
|
|
|
|
2011-02-25 15:51:29 +00:00
|
|
|
struct xen_blkbk {
|
2011-04-14 21:42:07 +00:00
|
|
|
struct pending_req *pending_reqs;
|
2011-04-14 21:05:23 +00:00
|
|
|
/* List of all 'pending_req' available */
|
2011-02-25 15:51:29 +00:00
|
|
|
struct list_head pending_free;
|
2011-04-14 21:05:23 +00:00
|
|
|
/* And its spinlock. */
|
2011-02-25 15:51:29 +00:00
|
|
|
spinlock_t pending_free_lock;
|
|
|
|
wait_queue_head_t pending_free_wq;
|
2011-04-14 21:05:23 +00:00
|
|
|
/* The list of all pages that are available. */
|
2011-02-25 15:51:29 +00:00
|
|
|
struct page **pending_pages;
|
2011-04-14 21:05:23 +00:00
|
|
|
/* And the grant handles that are available. */
|
2011-02-25 15:51:29 +00:00
|
|
|
grant_handle_t *pending_grant_handles;
|
|
|
|
};
|
|
|
|
|
|
|
|
static struct xen_blkbk *blkbk;
|
2011-04-14 22:25:47 +00:00
|
|
|
|
2011-04-14 21:05:23 +00:00
|
|
|
/*
|
|
|
|
* 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
|
2011-05-11 19:57:09 +00:00
|
|
|
* 10 and would index in the pending_pages[..].
|
|
|
|
*/
|
2011-04-14 21:42:07 +00:00
|
|
|
static inline int vaddr_pagenr(struct pending_req *req, int seg)
|
2011-04-14 22:25:47 +00:00
|
|
|
{
|
2011-04-14 21:42:07 +00:00
|
|
|
return (req - blkbk->pending_reqs) *
|
|
|
|
BLKIF_MAX_SEGMENTS_PER_REQUEST + seg;
|
2011-04-14 22:25:47 +00:00
|
|
|
}
|
|
|
|
|
2010-02-05 19:19:33 +00:00
|
|
|
#define pending_page(req, seg) pending_pages[vaddr_pagenr(req, seg)]
|
|
|
|
|
2011-04-14 21:42:07 +00:00
|
|
|
static inline unsigned long vaddr(struct pending_req *req, int seg)
|
2011-04-14 22:25:47 +00:00
|
|
|
{
|
2011-02-25 15:51:29 +00:00
|
|
|
unsigned long pfn = page_to_pfn(blkbk->pending_page(req, seg));
|
2011-04-14 22:25:47 +00:00
|
|
|
return (unsigned long)pfn_to_kaddr(pfn);
|
|
|
|
}
|
|
|
|
|
|
|
|
#define pending_handle(_req, _seg) \
|
2011-02-25 15:51:29 +00:00
|
|
|
(blkbk->pending_grant_handles[vaddr_pagenr(_req, _seg)])
|
2011-04-14 22:25:47 +00:00
|
|
|
|
|
|
|
|
2011-05-12 20:47:48 +00:00
|
|
|
static int do_block_io_op(struct xen_blkif *blkif);
|
|
|
|
static int dispatch_rw_block_io(struct xen_blkif *blkif,
|
2011-05-05 17:37:23 +00:00
|
|
|
struct blkif_request *req,
|
|
|
|
struct pending_req *pending_req);
|
2011-05-12 20:47:48 +00:00
|
|
|
static void make_response(struct xen_blkif *blkif, u64 id,
|
2011-04-14 22:25:47 +00:00
|
|
|
unsigned short op, int st);
|
|
|
|
|
2011-04-14 21:05:23 +00:00
|
|
|
/*
|
|
|
|
* Retrieve from the 'pending_reqs' a free pending_req structure to be used.
|
2011-04-14 22:25:47 +00:00
|
|
|
*/
|
2011-04-14 21:42:07 +00:00
|
|
|
static struct pending_req *alloc_req(void)
|
2011-04-14 22:25:47 +00:00
|
|
|
{
|
2011-04-14 21:42:07 +00:00
|
|
|
struct pending_req *req = NULL;
|
2011-04-14 22:25:47 +00:00
|
|
|
unsigned long flags;
|
|
|
|
|
2011-02-25 15:51:29 +00:00
|
|
|
spin_lock_irqsave(&blkbk->pending_free_lock, flags);
|
|
|
|
if (!list_empty(&blkbk->pending_free)) {
|
2011-04-14 21:42:07 +00:00
|
|
|
req = list_entry(blkbk->pending_free.next, struct pending_req,
|
|
|
|
free_list);
|
2011-04-14 22:25:47 +00:00
|
|
|
list_del(&req->free_list);
|
|
|
|
}
|
2011-02-25 15:51:29 +00:00
|
|
|
spin_unlock_irqrestore(&blkbk->pending_free_lock, flags);
|
2011-04-14 22:25:47 +00:00
|
|
|
return req;
|
|
|
|
}
|
|
|
|
|
2011-04-14 21:05:23 +00:00
|
|
|
/*
|
|
|
|
* Return the 'pending_req' structure back to the freepool. We also
|
|
|
|
* wake up the thread if it was waiting for a free page.
|
|
|
|
*/
|
2011-04-14 21:42:07 +00:00
|
|
|
static void free_req(struct pending_req *req)
|
2011-04-14 22:25:47 +00:00
|
|
|
{
|
|
|
|
unsigned long flags;
|
|
|
|
int was_empty;
|
|
|
|
|
2011-02-25 15:51:29 +00:00
|
|
|
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);
|
2011-04-14 22:25:47 +00:00
|
|
|
if (was_empty)
|
2011-02-25 15:51:29 +00:00
|
|
|
wake_up(&blkbk->pending_free_wq);
|
2011-04-14 22:25:47 +00:00
|
|
|
}
|
|
|
|
|
2011-04-20 14:57:29 +00:00
|
|
|
/*
|
|
|
|
* Routines for managing virtual block devices (vbds).
|
|
|
|
*/
|
2011-05-12 20:53:56 +00:00
|
|
|
static int xen_vbd_translate(struct phys_req *req, struct xen_blkif *blkif,
|
|
|
|
int operation)
|
2011-04-20 14:57:29 +00:00
|
|
|
{
|
2011-05-12 20:53:56 +00:00
|
|
|
struct xen_vbd *vbd = &blkif->vbd;
|
2011-04-20 14:57:29 +00:00
|
|
|
int rc = -EACCES;
|
|
|
|
|
|
|
|
if ((operation != READ) && vbd->readonly)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
if (unlikely((req->sector_number + req->nr_sects) > vbd_sz(vbd)))
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
req->dev = vbd->pdevice;
|
|
|
|
req->bdev = vbd->bdev;
|
|
|
|
rc = 0;
|
|
|
|
|
|
|
|
out:
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
|
2011-05-12 20:53:56 +00:00
|
|
|
static void xen_vbd_resize(struct xen_blkif *blkif)
|
2011-04-20 14:57:29 +00:00
|
|
|
{
|
2011-05-12 20:53:56 +00:00
|
|
|
struct xen_vbd *vbd = &blkif->vbd;
|
2011-04-20 14:57:29 +00:00
|
|
|
struct xenbus_transaction xbt;
|
|
|
|
int err;
|
2011-04-20 15:50:43 +00:00
|
|
|
struct xenbus_device *dev = xen_blkbk_xenbus(blkif->be);
|
2011-04-20 15:21:43 +00:00
|
|
|
unsigned long long new_size = vbd_sz(vbd);
|
2011-04-20 14:57:29 +00:00
|
|
|
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_info(DRV_PFX "VBD Resize: Domid: %d, Device: (%d, %d)\n",
|
2011-04-20 14:57:29 +00:00
|
|
|
blkif->domid, MAJOR(vbd->pdevice), MINOR(vbd->pdevice));
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_info(DRV_PFX "VBD Resize: new size %llu\n", new_size);
|
2011-04-20 14:57:29 +00:00
|
|
|
vbd->size = new_size;
|
|
|
|
again:
|
|
|
|
err = xenbus_transaction_start(&xbt);
|
|
|
|
if (err) {
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_warn(DRV_PFX "Error starting transaction");
|
2011-04-20 14:57:29 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
err = xenbus_printf(xbt, dev->nodename, "sectors", "%llu",
|
2011-04-20 15:21:43 +00:00
|
|
|
(unsigned long long)vbd_sz(vbd));
|
2011-04-20 14:57:29 +00:00
|
|
|
if (err) {
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_warn(DRV_PFX "Error writing new size");
|
2011-04-20 14:57:29 +00:00
|
|
|
goto abort;
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* Write the current state; we will use this to synchronize
|
|
|
|
* the front-end. If the current state is "connected" the
|
|
|
|
* front-end will get the new size information online.
|
|
|
|
*/
|
|
|
|
err = xenbus_printf(xbt, dev->nodename, "state", "%d", dev->state);
|
|
|
|
if (err) {
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_warn(DRV_PFX "Error writing the state");
|
2011-04-20 14:57:29 +00:00
|
|
|
goto abort;
|
|
|
|
}
|
|
|
|
|
|
|
|
err = xenbus_transaction_end(xbt, 0);
|
|
|
|
if (err == -EAGAIN)
|
|
|
|
goto again;
|
|
|
|
if (err)
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_warn(DRV_PFX "Error ending transaction");
|
2011-04-20 14:57:29 +00:00
|
|
|
abort:
|
|
|
|
xenbus_transaction_end(xbt, 1);
|
|
|
|
}
|
|
|
|
|
2011-04-14 21:05:23 +00:00
|
|
|
/*
|
2011-04-15 14:58:05 +00:00
|
|
|
* Notification from the guest OS.
|
|
|
|
*/
|
2011-05-12 20:47:48 +00:00
|
|
|
static void blkif_notify_work(struct xen_blkif *blkif)
|
2011-04-14 22:25:47 +00:00
|
|
|
{
|
2011-04-15 14:58:05 +00:00
|
|
|
blkif->waiting_reqs = 1;
|
|
|
|
wake_up(&blkif->wq);
|
|
|
|
}
|
2011-04-14 22:25:47 +00:00
|
|
|
|
2011-04-20 15:50:43 +00:00
|
|
|
irqreturn_t xen_blkif_be_int(int irq, void *dev_id)
|
2011-04-15 14:58:05 +00:00
|
|
|
{
|
|
|
|
blkif_notify_work(dev_id);
|
|
|
|
return IRQ_HANDLED;
|
2011-04-14 22:25:47 +00:00
|
|
|
}
|
|
|
|
|
2011-04-14 21:42:07 +00:00
|
|
|
/*
|
2011-04-14 22:25:47 +00:00
|
|
|
* SCHEDULER FUNCTIONS
|
|
|
|
*/
|
|
|
|
|
2011-05-12 20:47:48 +00:00
|
|
|
static void print_stats(struct xen_blkif *blkif)
|
2011-04-14 22:25:47 +00:00
|
|
|
{
|
2011-05-12 21:23:30 +00:00
|
|
|
pr_info("xen-blkback (%s): oo %3d | rd %4d | wr %4d | f %4d\n",
|
2011-05-12 20:42:31 +00:00
|
|
|
current->comm, blkif->st_oo_req,
|
|
|
|
blkif->st_rd_req, blkif->st_wr_req, blkif->st_f_req);
|
2011-04-14 22:25:47 +00:00
|
|
|
blkif->st_print = jiffies + msecs_to_jiffies(10 * 1000);
|
|
|
|
blkif->st_rd_req = 0;
|
|
|
|
blkif->st_wr_req = 0;
|
|
|
|
blkif->st_oo_req = 0;
|
|
|
|
}
|
|
|
|
|
2011-04-20 15:50:43 +00:00
|
|
|
int xen_blkif_schedule(void *arg)
|
2011-04-14 22:25:47 +00:00
|
|
|
{
|
2011-05-12 20:47:48 +00:00
|
|
|
struct xen_blkif *blkif = arg;
|
2011-05-12 20:53:56 +00:00
|
|
|
struct xen_vbd *vbd = &blkif->vbd;
|
2011-04-14 22:25:47 +00:00
|
|
|
|
2011-04-20 15:50:43 +00:00
|
|
|
xen_blkif_get(blkif);
|
2011-04-14 22:25:47 +00:00
|
|
|
|
|
|
|
while (!kthread_should_stop()) {
|
|
|
|
if (try_to_freeze())
|
|
|
|
continue;
|
2011-04-20 15:21:43 +00:00
|
|
|
if (unlikely(vbd->size != vbd_sz(vbd)))
|
2011-05-12 20:53:56 +00:00
|
|
|
xen_vbd_resize(blkif);
|
2011-04-14 22:25:47 +00:00
|
|
|
|
|
|
|
wait_event_interruptible(
|
|
|
|
blkif->wq,
|
|
|
|
blkif->waiting_reqs || kthread_should_stop());
|
|
|
|
wait_event_interruptible(
|
2011-02-25 15:51:29 +00:00
|
|
|
blkbk->pending_free_wq,
|
2011-04-14 21:42:07 +00:00
|
|
|
!list_empty(&blkbk->pending_free) ||
|
|
|
|
kthread_should_stop());
|
2011-04-14 22:25:47 +00:00
|
|
|
|
|
|
|
blkif->waiting_reqs = 0;
|
|
|
|
smp_mb(); /* clear flag *before* checking for work */
|
|
|
|
|
|
|
|
if (do_block_io_op(blkif))
|
|
|
|
blkif->waiting_reqs = 1;
|
|
|
|
|
|
|
|
if (log_stats && time_after(jiffies, blkif->st_print))
|
|
|
|
print_stats(blkif);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (log_stats)
|
|
|
|
print_stats(blkif);
|
|
|
|
|
|
|
|
blkif->xenblkd = NULL;
|
2011-04-20 15:50:43 +00:00
|
|
|
xen_blkif_put(blkif);
|
2011-04-14 22:25:47 +00:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2011-04-15 15:35:13 +00:00
|
|
|
struct seg_buf {
|
|
|
|
unsigned long buf;
|
|
|
|
unsigned int nsec;
|
|
|
|
};
|
2011-04-15 14:58:05 +00:00
|
|
|
/*
|
|
|
|
* Unmap the grant references, and also remove the M2P over-rides
|
|
|
|
* used in the 'pending_req'.
|
2011-05-11 19:57:09 +00:00
|
|
|
*/
|
2011-04-15 15:50:34 +00:00
|
|
|
static void xen_blkbk_unmap(struct pending_req *req)
|
2011-04-15 14:58:05 +00:00
|
|
|
{
|
|
|
|
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);
|
2011-05-11 19:57:09 +00:00
|
|
|
/*
|
|
|
|
* Note, we use invcount, so nr->pages, so we can't index
|
2011-04-15 14:58:05 +00:00
|
|
|
* using vaddr(req, i).
|
|
|
|
*/
|
|
|
|
for (i = 0; i < invcount; i++) {
|
|
|
|
ret = m2p_remove_override(
|
|
|
|
virt_to_page(unmap[i].host_addr), false);
|
|
|
|
if (ret) {
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_alert(DRV_PFX "Failed to remove M2P override for %lx\n",
|
2011-05-12 20:42:31 +00:00
|
|
|
(unsigned long)unmap[i].host_addr);
|
2011-04-15 14:58:05 +00:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2011-05-11 19:57:09 +00:00
|
|
|
|
|
|
|
static int xen_blkbk_map(struct blkif_request *req,
|
|
|
|
struct pending_req *pending_req,
|
2011-04-15 15:50:34 +00:00
|
|
|
struct seg_buf seg[])
|
2011-04-15 15:35:13 +00:00
|
|
|
{
|
|
|
|
struct gnttab_map_grant_ref map[BLKIF_MAX_SEGMENTS_PER_REQUEST];
|
|
|
|
int i;
|
|
|
|
int nseg = req->nr_segments;
|
|
|
|
int ret = 0;
|
2011-05-11 19:57:09 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Fill out preq.nr_sects with proper amount of sectors, and setup
|
2011-04-15 15:35:13 +00:00
|
|
|
* 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;
|
|
|
|
|
|
|
|
flags = GNTMAP_host_map;
|
|
|
|
if (pending_req->operation != BLKIF_OP_READ)
|
|
|
|
flags |= GNTMAP_readonly;
|
|
|
|
gnttab_set_map_op(&map[i], vaddr(pending_req, i), flags,
|
2011-05-11 19:57:09 +00:00
|
|
|
req->u.rw.seg[i].gref,
|
|
|
|
pending_req->blkif->domid);
|
2011-04-15 15:35:13 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
ret = HYPERVISOR_grant_table_op(GNTTABOP_map_grant_ref, map, nseg);
|
|
|
|
BUG_ON(ret);
|
|
|
|
|
2011-05-11 19:57:09 +00:00
|
|
|
/*
|
|
|
|
* Now swizzle the MFN in our domain with the MFN from the other domain
|
2011-04-15 15:35:13 +00:00
|
|
|
* 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)) {
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_debug(DRV_PFX "invalid buffer -- could not remap it\n");
|
2011-04-15 15:35:13 +00:00
|
|
|
map[i].handle = BLKBACK_INVALID_HANDLE;
|
|
|
|
ret |= 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
pending_handle(pending_req, i) = map[i].handle;
|
|
|
|
|
|
|
|
if (ret)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
ret = m2p_add_override(PFN_DOWN(map[i].dev_bus_addr),
|
|
|
|
blkbk->pending_page(pending_req, i), false);
|
|
|
|
if (ret) {
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_alert(DRV_PFX "Failed to install M2P override for %lx (ret: %d)\n",
|
2011-05-12 20:42:31 +00:00
|
|
|
(unsigned long)map[i].dev_bus_addr, ret);
|
2011-04-15 15:35:13 +00:00
|
|
|
/* We could switch over to GNTTABOP_copy */
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
seg[i].buf = map[i].dev_bus_addr |
|
|
|
|
(req->u.rw.seg[i].first_sect << 9);
|
|
|
|
}
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2011-04-14 21:05:23 +00:00
|
|
|
/*
|
|
|
|
* Completion callback on the bio's. Called as bh->b_end_io()
|
2011-04-14 22:25:47 +00:00
|
|
|
*/
|
|
|
|
|
2011-04-14 21:42:07 +00:00
|
|
|
static void __end_block_io_op(struct pending_req *pending_req, int error)
|
2011-04-14 22:25:47 +00:00
|
|
|
{
|
|
|
|
/* An error fails the entire request. */
|
2011-05-04 21:07:27 +00:00
|
|
|
if ((pending_req->operation == BLKIF_OP_FLUSH_DISKCACHE) &&
|
2011-04-14 22:25:47 +00:00
|
|
|
(error == -EOPNOTSUPP)) {
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_debug(DRV_PFX "flush diskcache op failed, not supported\n");
|
2011-05-04 21:07:27 +00:00
|
|
|
xen_blkbk_flush_diskcache(XBT_NIL, pending_req->blkif->be, 0);
|
2011-04-14 22:25:47 +00:00
|
|
|
pending_req->status = BLKIF_RSP_EOPNOTSUPP;
|
|
|
|
} else if (error) {
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_debug(DRV_PFX "Buffer not up-to-date at end of operation,"
|
2011-05-12 20:42:31 +00:00
|
|
|
" error=%d\n", error);
|
2011-04-14 22:25:47 +00:00
|
|
|
pending_req->status = BLKIF_RSP_ERROR;
|
|
|
|
}
|
|
|
|
|
2011-05-11 19:57:09 +00:00
|
|
|
/*
|
|
|
|
* If all of the bio's have completed it is time to unmap
|
2011-04-14 21:05:23 +00:00
|
|
|
* the grant references associated with 'request' and provide
|
2011-04-14 21:42:07 +00:00
|
|
|
* the proper response on the ring.
|
|
|
|
*/
|
2011-04-14 22:25:47 +00:00
|
|
|
if (atomic_dec_and_test(&pending_req->pendcnt)) {
|
2011-04-15 15:50:34 +00:00
|
|
|
xen_blkbk_unmap(pending_req);
|
2011-04-14 22:25:47 +00:00
|
|
|
make_response(pending_req->blkif, pending_req->id,
|
|
|
|
pending_req->operation, pending_req->status);
|
2011-04-20 15:50:43 +00:00
|
|
|
xen_blkif_put(pending_req->blkif);
|
2011-04-14 22:25:47 +00:00
|
|
|
free_req(pending_req);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-04-14 21:05:23 +00:00
|
|
|
/*
|
|
|
|
* bio callback.
|
|
|
|
*/
|
2009-02-09 20:05:51 +00:00
|
|
|
static void end_block_io_op(struct bio *bio, int error)
|
2011-04-14 22:25:47 +00:00
|
|
|
{
|
|
|
|
__end_block_io_op(bio->bi_private, error);
|
|
|
|
bio_put(bio);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
2011-04-14 21:05:23 +00:00
|
|
|
/*
|
|
|
|
* 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.
|
2011-04-14 22:25:47 +00:00
|
|
|
*/
|
2011-05-12 20:47:48 +00:00
|
|
|
static int do_block_io_op(struct xen_blkif *blkif)
|
2011-04-14 22:25:47 +00:00
|
|
|
{
|
2009-02-09 20:05:51 +00:00
|
|
|
union blkif_back_rings *blk_rings = &blkif->blk_rings;
|
|
|
|
struct blkif_request req;
|
2011-04-14 21:42:07 +00:00
|
|
|
struct pending_req *pending_req;
|
2011-04-14 22:25:47 +00:00
|
|
|
RING_IDX rc, rp;
|
|
|
|
int more_to_do = 0;
|
|
|
|
|
|
|
|
rc = blk_rings->common.req_cons;
|
|
|
|
rp = blk_rings->common.sring->req_prod;
|
|
|
|
rmb(); /* Ensure we see queued requests up to 'rp'. */
|
|
|
|
|
|
|
|
while (rc != rp) {
|
|
|
|
|
|
|
|
if (RING_REQUEST_CONS_OVERFLOW(&blk_rings->common, rc))
|
|
|
|
break;
|
|
|
|
|
2009-03-06 08:29:15 +00:00
|
|
|
if (kthread_should_stop()) {
|
2011-04-14 22:25:47 +00:00
|
|
|
more_to_do = 1;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2009-03-06 08:29:15 +00:00
|
|
|
pending_req = alloc_req();
|
|
|
|
if (NULL == pending_req) {
|
|
|
|
blkif->st_oo_req++;
|
2011-04-14 22:25:47 +00:00
|
|
|
more_to_do = 1;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
switch (blkif->blk_protocol) {
|
|
|
|
case BLKIF_PROTOCOL_NATIVE:
|
|
|
|
memcpy(&req, RING_GET_REQUEST(&blk_rings->native, rc), sizeof(req));
|
|
|
|
break;
|
|
|
|
case BLKIF_PROTOCOL_X86_32:
|
|
|
|
blkif_get_x86_32_req(&req, RING_GET_REQUEST(&blk_rings->x86_32, rc));
|
|
|
|
break;
|
|
|
|
case BLKIF_PROTOCOL_X86_64:
|
|
|
|
blkif_get_x86_64_req(&req, RING_GET_REQUEST(&blk_rings->x86_64, rc));
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
BUG();
|
|
|
|
}
|
|
|
|
blk_rings->common.req_cons = ++rc; /* before make_response() */
|
|
|
|
|
|
|
|
/* Apply all sanity checks to /private copy/ of request. */
|
|
|
|
barrier();
|
|
|
|
|
2011-05-05 17:37:23 +00:00
|
|
|
if (dispatch_rw_block_io(blkif, &req, pending_req))
|
2011-04-14 22:25:47 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
/* Yield point for this unbounded loop. */
|
|
|
|
cond_resched();
|
|
|
|
}
|
|
|
|
|
|
|
|
return more_to_do;
|
|
|
|
}
|
|
|
|
|
2011-04-14 21:05:23 +00:00
|
|
|
/*
|
2011-05-11 19:57:09 +00:00
|
|
|
* Transmutation of the 'struct blkif_request' to a proper 'struct bio'
|
|
|
|
* and call the 'submit_bio' to pass it to the underlying storage.
|
2011-04-14 21:05:23 +00:00
|
|
|
*/
|
2011-05-12 20:47:48 +00:00
|
|
|
static int dispatch_rw_block_io(struct xen_blkif *blkif,
|
|
|
|
struct blkif_request *req,
|
|
|
|
struct pending_req *pending_req)
|
2011-04-14 22:25:47 +00:00
|
|
|
{
|
|
|
|
struct phys_req preq;
|
2011-04-15 15:35:13 +00:00
|
|
|
struct seg_buf seg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
|
2011-04-14 22:25:47 +00:00
|
|
|
unsigned int nseg;
|
|
|
|
struct bio *bio = NULL;
|
2011-04-15 14:51:27 +00:00
|
|
|
struct bio *biolist[BLKIF_MAX_SEGMENTS_PER_REQUEST];
|
2011-04-15 15:35:13 +00:00
|
|
|
int i, nbio = 0;
|
2011-04-14 22:25:47 +00:00
|
|
|
int operation;
|
2011-04-27 16:40:11 +00:00
|
|
|
struct blk_plug plug;
|
2011-04-14 22:25:47 +00:00
|
|
|
|
|
|
|
switch (req->operation) {
|
|
|
|
case BLKIF_OP_READ:
|
2011-05-05 17:37:23 +00:00
|
|
|
blkif->st_rd_req++;
|
2011-04-14 22:25:47 +00:00
|
|
|
operation = READ;
|
|
|
|
break;
|
|
|
|
case BLKIF_OP_WRITE:
|
2011-05-05 17:37:23 +00:00
|
|
|
blkif->st_wr_req++;
|
2011-04-26 20:24:18 +00:00
|
|
|
operation = WRITE_ODIRECT;
|
2011-04-14 22:25:47 +00:00
|
|
|
break;
|
2011-05-04 21:07:27 +00:00
|
|
|
case BLKIF_OP_FLUSH_DISKCACHE:
|
2011-05-05 17:37:23 +00:00
|
|
|
blkif->st_f_req++;
|
2011-05-04 21:07:27 +00:00
|
|
|
operation = WRITE_FLUSH;
|
2011-05-11 19:57:09 +00:00
|
|
|
/*
|
2011-05-12 20:53:56 +00:00
|
|
|
* The frontend likes to set this to -1, which xen_vbd_translate
|
2011-05-11 19:57:09 +00:00
|
|
|
* is alergic too.
|
|
|
|
*/
|
2011-05-04 21:07:27 +00:00
|
|
|
req->u.rw.sector_number = 0;
|
2011-04-14 22:25:47 +00:00
|
|
|
break;
|
2011-05-04 21:07:27 +00:00
|
|
|
case BLKIF_OP_WRITE_BARRIER:
|
2011-04-14 22:25:47 +00:00
|
|
|
default:
|
|
|
|
operation = 0; /* make gcc happy */
|
2011-05-05 17:37:23 +00:00
|
|
|
goto fail_response;
|
|
|
|
break;
|
2011-04-14 22:25:47 +00:00
|
|
|
}
|
|
|
|
|
2011-04-14 21:05:23 +00:00
|
|
|
/* Check that the number of segments is sane. */
|
2011-04-14 22:25:47 +00:00
|
|
|
nseg = req->nr_segments;
|
2011-05-04 21:07:27 +00:00
|
|
|
if (unlikely(nseg == 0 && operation != WRITE_FLUSH) ||
|
2011-04-14 22:25:47 +00:00
|
|
|
unlikely(nseg > BLKIF_MAX_SEGMENTS_PER_REQUEST)) {
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_debug(DRV_PFX "Bad number of segments in request (%d)\n",
|
2011-05-12 20:42:31 +00:00
|
|
|
nseg);
|
2011-04-15 15:35:13 +00:00
|
|
|
/* Haven't submitted any bio's yet. */
|
2011-04-14 22:25:47 +00:00
|
|
|
goto fail_response;
|
|
|
|
}
|
|
|
|
|
|
|
|
preq.dev = req->handle;
|
2011-03-01 21:22:28 +00:00
|
|
|
preq.sector_number = req->u.rw.sector_number;
|
2011-04-14 22:25:47 +00:00
|
|
|
preq.nr_sects = 0;
|
|
|
|
|
|
|
|
pending_req->blkif = blkif;
|
|
|
|
pending_req->id = req->id;
|
|
|
|
pending_req->operation = req->operation;
|
|
|
|
pending_req->status = BLKIF_RSP_OKAY;
|
|
|
|
pending_req->nr_pages = nseg;
|
2011-04-18 15:34:55 +00:00
|
|
|
|
2011-04-14 22:25:47 +00:00
|
|
|
for (i = 0; i < nseg; i++) {
|
2011-03-01 21:22:28 +00:00
|
|
|
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))
|
2011-04-14 22:25:47 +00:00
|
|
|
goto fail_response;
|
|
|
|
preq.nr_sects += seg[i].nsec;
|
2011-04-15 15:38:29 +00:00
|
|
|
|
2011-04-14 22:25:47 +00:00
|
|
|
}
|
|
|
|
|
2011-05-12 20:53:56 +00:00
|
|
|
if (xen_vbd_translate(&preq, blkif, operation) != 0) {
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_debug(DRV_PFX "access denied: %s of [%llu,%llu] on dev=%04x\n",
|
2011-05-12 20:42:31 +00:00
|
|
|
operation == READ ? "read" : "write",
|
|
|
|
preq.sector_number,
|
|
|
|
preq.sector_number + preq.nr_sects, preq.dev);
|
2011-04-15 15:35:13 +00:00
|
|
|
goto fail_response;
|
2011-04-14 22:25:47 +00:00
|
|
|
}
|
2011-05-11 19:57:09 +00:00
|
|
|
|
|
|
|
/*
|
2011-05-12 20:53:56 +00:00
|
|
|
* This check _MUST_ be done after xen_vbd_translate as the preq.bdev
|
2011-05-11 19:57:09 +00:00
|
|
|
* is set there.
|
|
|
|
*/
|
2011-04-18 15:34:55 +00:00
|
|
|
for (i = 0; i < nseg; i++) {
|
|
|
|
if (((int)preq.sector_number|(int)seg[i].nsec) &
|
|
|
|
((bdev_logical_block_size(preq.bdev) >> 9) - 1)) {
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_debug(DRV_PFX "Misaligned I/O request from domain %d",
|
2011-05-12 20:42:31 +00:00
|
|
|
blkif->domid);
|
2011-04-18 15:34:55 +00:00
|
|
|
goto fail_response;
|
|
|
|
}
|
|
|
|
}
|
2011-05-11 19:57:09 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* If we have failed at this point, we need to undo the M2P override,
|
2011-04-14 21:42:07 +00:00
|
|
|
* set gnttab_set_unmap_op on all of the grant references and perform
|
|
|
|
* the hypercall to unmap the grants - that is all done in
|
2011-04-15 15:50:34 +00:00
|
|
|
* xen_blkbk_unmap.
|
2011-04-14 21:42:07 +00:00
|
|
|
*/
|
2011-04-15 15:50:34 +00:00
|
|
|
if (xen_blkbk_map(req, pending_req, seg))
|
2011-04-14 22:25:47 +00:00
|
|
|
goto fail_flush;
|
|
|
|
|
2011-05-05 17:42:10 +00:00
|
|
|
/* This corresponding xen_blkif_put is done in __end_block_io_op */
|
2011-04-20 15:50:43 +00:00
|
|
|
xen_blkif_get(blkif);
|
2011-04-14 22:25:47 +00:00
|
|
|
|
|
|
|
for (i = 0; i < nseg; i++) {
|
|
|
|
while ((bio == NULL) ||
|
|
|
|
(bio_add_page(bio,
|
2011-02-25 15:51:29 +00:00
|
|
|
blkbk->pending_page(pending_req, i),
|
2011-04-14 22:25:47 +00:00
|
|
|
seg[i].nsec << 9,
|
|
|
|
seg[i].buf & ~PAGE_MASK) == 0)) {
|
2011-04-14 21:42:07 +00:00
|
|
|
|
2011-05-12 20:19:23 +00:00
|
|
|
bio = bio_alloc(GFP_KERNEL, nseg-i);
|
2011-04-14 22:25:47 +00:00
|
|
|
if (unlikely(bio == NULL))
|
|
|
|
goto fail_put_bio;
|
|
|
|
|
2011-05-12 20:19:23 +00:00
|
|
|
biolist[nbio++] = bio;
|
2011-04-14 22:25:47 +00:00
|
|
|
bio->bi_bdev = preq.bdev;
|
|
|
|
bio->bi_private = pending_req;
|
|
|
|
bio->bi_end_io = end_block_io_op;
|
|
|
|
bio->bi_sector = preq.sector_number;
|
|
|
|
}
|
|
|
|
|
|
|
|
preq.sector_number += seg[i].nsec;
|
|
|
|
}
|
|
|
|
|
2011-05-05 17:42:10 +00:00
|
|
|
/* This will be hit if the operation was a flush. */
|
2011-04-14 22:25:47 +00:00
|
|
|
if (!bio) {
|
2011-05-04 21:07:27 +00:00
|
|
|
BUG_ON(operation != WRITE_FLUSH);
|
2011-05-12 20:23:06 +00:00
|
|
|
|
2011-05-12 20:19:23 +00:00
|
|
|
bio = bio_alloc(GFP_KERNEL, 0);
|
2011-04-14 22:25:47 +00:00
|
|
|
if (unlikely(bio == NULL))
|
|
|
|
goto fail_put_bio;
|
|
|
|
|
2011-05-12 20:19:23 +00:00
|
|
|
biolist[nbio++] = bio;
|
2011-04-14 22:25:47 +00:00
|
|
|
bio->bi_bdev = preq.bdev;
|
|
|
|
bio->bi_private = pending_req;
|
|
|
|
bio->bi_end_io = end_block_io_op;
|
|
|
|
}
|
|
|
|
|
2011-05-11 19:57:09 +00:00
|
|
|
/*
|
|
|
|
* We set it one so that the last submit_bio does not have to call
|
2011-04-15 14:51:27 +00:00
|
|
|
* atomic_inc.
|
|
|
|
*/
|
|
|
|
atomic_set(&pending_req->pendcnt, nbio);
|
|
|
|
|
2011-04-27 16:40:11 +00:00
|
|
|
/* Get a reference count for the disk queue and start sending I/O */
|
|
|
|
blk_start_plug(&plug);
|
|
|
|
|
2011-04-15 14:51:27 +00:00
|
|
|
for (i = 0; i < nbio; i++)
|
|
|
|
submit_bio(operation, biolist[i]);
|
|
|
|
|
2011-04-27 16:40:11 +00:00
|
|
|
/* Let the I/Os go.. */
|
2011-05-05 17:42:10 +00:00
|
|
|
blk_finish_plug(&plug);
|
2011-04-27 16:40:11 +00:00
|
|
|
|
2011-04-14 22:25:47 +00:00
|
|
|
if (operation == READ)
|
|
|
|
blkif->st_rd_sect += preq.nr_sects;
|
2011-05-04 21:07:27 +00:00
|
|
|
else if (operation == WRITE || operation == WRITE_FLUSH)
|
2011-04-14 22:25:47 +00:00
|
|
|
blkif->st_wr_sect += preq.nr_sects;
|
|
|
|
|
2011-05-05 17:37:23 +00:00
|
|
|
return 0;
|
2011-04-14 22:25:47 +00:00
|
|
|
|
|
|
|
fail_flush:
|
2011-04-15 15:50:34 +00:00
|
|
|
xen_blkbk_unmap(pending_req);
|
2011-04-14 22:25:47 +00:00
|
|
|
fail_response:
|
2011-04-14 21:58:19 +00:00
|
|
|
/* Haven't submitted any bio's yet. */
|
2011-04-14 22:25:47 +00:00
|
|
|
make_response(blkif, req->id, req->operation, BLKIF_RSP_ERROR);
|
|
|
|
free_req(pending_req);
|
|
|
|
msleep(1); /* back off a bit */
|
2011-05-05 17:37:23 +00:00
|
|
|
return -EIO;
|
2011-04-14 22:25:47 +00:00
|
|
|
|
|
|
|
fail_put_bio:
|
2011-05-12 20:19:23 +00:00
|
|
|
for (i = 0; i < nbio; i++)
|
2011-04-15 14:51:27 +00:00
|
|
|
bio_put(biolist[i]);
|
2011-04-14 22:25:47 +00:00
|
|
|
__end_block_io_op(pending_req, -EINVAL);
|
|
|
|
msleep(1); /* back off a bit */
|
2011-05-05 17:37:23 +00:00
|
|
|
return -EIO;
|
2011-04-14 22:25:47 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
2011-04-14 21:05:23 +00:00
|
|
|
/*
|
|
|
|
* Put a response on the ring on how the operation fared.
|
2011-04-14 22:25:47 +00:00
|
|
|
*/
|
2011-05-12 20:47:48 +00:00
|
|
|
static void make_response(struct xen_blkif *blkif, u64 id,
|
2011-04-14 22:25:47 +00:00
|
|
|
unsigned short op, int st)
|
|
|
|
{
|
2009-02-09 20:05:51 +00:00
|
|
|
struct blkif_response resp;
|
2011-04-14 22:25:47 +00:00
|
|
|
unsigned long flags;
|
2009-02-09 20:05:51 +00:00
|
|
|
union blkif_back_rings *blk_rings = &blkif->blk_rings;
|
2011-04-14 22:25:47 +00:00
|
|
|
int more_to_do = 0;
|
|
|
|
int notify;
|
|
|
|
|
|
|
|
resp.id = id;
|
|
|
|
resp.operation = op;
|
|
|
|
resp.status = st;
|
|
|
|
|
|
|
|
spin_lock_irqsave(&blkif->blk_ring_lock, flags);
|
|
|
|
/* Place on the response ring for the relevant domain. */
|
|
|
|
switch (blkif->blk_protocol) {
|
|
|
|
case BLKIF_PROTOCOL_NATIVE:
|
|
|
|
memcpy(RING_GET_RESPONSE(&blk_rings->native, blk_rings->native.rsp_prod_pvt),
|
|
|
|
&resp, sizeof(resp));
|
|
|
|
break;
|
|
|
|
case BLKIF_PROTOCOL_X86_32:
|
|
|
|
memcpy(RING_GET_RESPONSE(&blk_rings->x86_32, blk_rings->x86_32.rsp_prod_pvt),
|
|
|
|
&resp, sizeof(resp));
|
|
|
|
break;
|
|
|
|
case BLKIF_PROTOCOL_X86_64:
|
|
|
|
memcpy(RING_GET_RESPONSE(&blk_rings->x86_64, blk_rings->x86_64.rsp_prod_pvt),
|
|
|
|
&resp, sizeof(resp));
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
BUG();
|
|
|
|
}
|
|
|
|
blk_rings->common.rsp_prod_pvt++;
|
|
|
|
RING_PUSH_RESPONSES_AND_CHECK_NOTIFY(&blk_rings->common, notify);
|
|
|
|
if (blk_rings->common.rsp_prod_pvt == blk_rings->common.req_cons) {
|
|
|
|
/*
|
|
|
|
* Tail check for pending requests. Allows frontend to avoid
|
|
|
|
* notifications if requests are already in flight (lower
|
|
|
|
* overheads and promotes batching).
|
|
|
|
*/
|
|
|
|
RING_FINAL_CHECK_FOR_REQUESTS(&blk_rings->common, more_to_do);
|
|
|
|
|
|
|
|
} else if (RING_HAS_UNCONSUMED_REQUESTS(&blk_rings->common)) {
|
|
|
|
more_to_do = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
spin_unlock_irqrestore(&blkif->blk_ring_lock, flags);
|
|
|
|
|
|
|
|
if (more_to_do)
|
|
|
|
blkif_notify_work(blkif);
|
|
|
|
if (notify)
|
|
|
|
notify_remote_via_irq(blkif->irq);
|
|
|
|
}
|
|
|
|
|
2011-04-20 15:50:43 +00:00
|
|
|
static int __init xen_blkif_init(void)
|
2011-04-14 22:25:47 +00:00
|
|
|
{
|
|
|
|
int i, mmap_pages;
|
2009-10-08 17:23:09 +00:00
|
|
|
int rc = 0;
|
2011-04-14 22:25:47 +00:00
|
|
|
|
2009-02-09 20:05:51 +00:00
|
|
|
if (!xen_pv_domain())
|
2011-04-14 22:25:47 +00:00
|
|
|
return -ENODEV;
|
|
|
|
|
2011-04-14 21:42:07 +00:00
|
|
|
blkbk = kzalloc(sizeof(struct xen_blkbk), GFP_KERNEL);
|
2011-02-25 15:51:29 +00:00
|
|
|
if (!blkbk) {
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_alert(DRV_PFX "%s: out of memory!\n", __func__);
|
2011-02-25 15:51:29 +00:00
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
|
2011-04-20 15:50:43 +00:00
|
|
|
mmap_pages = xen_blkif_reqs * BLKIF_MAX_SEGMENTS_PER_REQUEST;
|
2011-04-14 22:25:47 +00:00
|
|
|
|
2011-02-25 15:51:29 +00:00
|
|
|
blkbk->pending_reqs = kmalloc(sizeof(blkbk->pending_reqs[0]) *
|
2011-04-20 15:50:43 +00:00
|
|
|
xen_blkif_reqs, GFP_KERNEL);
|
2011-03-14 16:41:26 +00:00
|
|
|
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);
|
2011-04-14 22:25:47 +00:00
|
|
|
|
2011-04-14 21:42:07 +00:00
|
|
|
if (!blkbk->pending_reqs || !blkbk->pending_grant_handles ||
|
|
|
|
!blkbk->pending_pages) {
|
2009-10-08 17:23:09 +00:00
|
|
|
rc = -ENOMEM;
|
2011-04-14 22:25:47 +00:00
|
|
|
goto out_of_memory;
|
2009-10-08 17:23:09 +00:00
|
|
|
}
|
2011-04-14 22:25:47 +00:00
|
|
|
|
2011-03-01 21:26:10 +00:00
|
|
|
for (i = 0; i < mmap_pages; i++) {
|
2011-02-25 15:51:29 +00:00
|
|
|
blkbk->pending_grant_handles[i] = BLKBACK_INVALID_HANDLE;
|
2011-03-14 16:41:26 +00:00
|
|
|
blkbk->pending_pages[i] = alloc_page(GFP_KERNEL);
|
2011-03-01 21:26:10 +00:00
|
|
|
if (blkbk->pending_pages[i] == NULL) {
|
|
|
|
rc = -ENOMEM;
|
|
|
|
goto out_of_memory;
|
|
|
|
}
|
|
|
|
}
|
2011-04-20 15:50:43 +00:00
|
|
|
rc = xen_blkif_interface_init();
|
2009-10-08 17:23:09 +00:00
|
|
|
if (rc)
|
|
|
|
goto failed_init;
|
2011-04-14 22:25:47 +00:00
|
|
|
|
2011-02-25 15:51:29 +00:00
|
|
|
memset(blkbk->pending_reqs, 0, sizeof(blkbk->pending_reqs));
|
|
|
|
|
|
|
|
INIT_LIST_HEAD(&blkbk->pending_free);
|
|
|
|
spin_lock_init(&blkbk->pending_free_lock);
|
|
|
|
init_waitqueue_head(&blkbk->pending_free_wq);
|
2011-04-14 22:25:47 +00:00
|
|
|
|
2011-04-20 15:50:43 +00:00
|
|
|
for (i = 0; i < xen_blkif_reqs; i++)
|
2011-04-14 21:42:07 +00:00
|
|
|
list_add_tail(&blkbk->pending_reqs[i].free_list,
|
|
|
|
&blkbk->pending_free);
|
2011-04-14 22:25:47 +00:00
|
|
|
|
2011-04-20 15:50:43 +00:00
|
|
|
rc = xen_blkif_xenbus_init();
|
2009-10-08 17:23:09 +00:00
|
|
|
if (rc)
|
|
|
|
goto failed_init;
|
2011-04-14 22:25:47 +00:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
out_of_memory:
|
2011-05-12 20:43:12 +00:00
|
|
|
pr_alert(DRV_PFX "%s: out of memory\n", __func__);
|
2009-10-08 17:23:09 +00:00
|
|
|
failed_init:
|
2011-02-25 15:51:29 +00:00
|
|
|
kfree(blkbk->pending_reqs);
|
2011-03-14 16:41:26 +00:00
|
|
|
kfree(blkbk->pending_grant_handles);
|
2011-03-01 21:26:10 +00:00
|
|
|
for (i = 0; i < mmap_pages; i++) {
|
|
|
|
if (blkbk->pending_pages[i])
|
|
|
|
__free_page(blkbk->pending_pages[i]);
|
|
|
|
}
|
2011-03-14 16:41:26 +00:00
|
|
|
kfree(blkbk->pending_pages);
|
|
|
|
kfree(blkbk);
|
2011-02-25 15:51:29 +00:00
|
|
|
blkbk = NULL;
|
2009-10-08 17:23:09 +00:00
|
|
|
return rc;
|
2011-04-14 22:25:47 +00:00
|
|
|
}
|
|
|
|
|
2011-04-20 15:50:43 +00:00
|
|
|
module_init(xen_blkif_init);
|
2011-04-14 22:25:47 +00:00
|
|
|
|
|
|
|
MODULE_LICENSE("Dual BSD/GPL");
|