4 * XenLinux virtual block device driver.
6 * Copyright (c) 2003-2004, Keir Fraser & Steve Hand
7 * Modifications by Mark A. Williamson are (c) Intel Research Cambridge
8 * Copyright (c) 2004, Christian Limpach
9 * Copyright (c) 2004, Andrew Warfield
10 * Copyright (c) 2005, Christopher Clark
11 * Copyright (c) 2005, XenSource Ltd
13 * This program is free software; you can redistribute it and/or
14 * modify it under the terms of the GNU General Public License version 2
15 * as published by the Free Software Foundation; or, when distributed
16 * separately from the Linux kernel or incorporated into other
17 * software packages, subject to the following license:
19 * Permission is hereby granted, free of charge, to any person obtaining a copy
20 * of this source file (the "Software"), to deal in the Software without
21 * restriction, including without limitation the rights to use, copy, modify,
22 * merge, publish, distribute, sublicense, and/or sell copies of the Software,
23 * and to permit persons to whom the Software is furnished to do so, subject to
24 * the following conditions:
26 * The above copyright notice and this permission notice shall be included in
27 * all copies or substantial portions of the Software.
29 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
30 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
31 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
32 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
33 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
34 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
38 #include <linux/interrupt.h>
39 #include <linux/blkdev.h>
40 #include <linux/blk-mq.h>
41 #include <linux/hdreg.h>
42 #include <linux/cdrom.h>
43 #include <linux/module.h>
44 #include <linux/slab.h>
45 #include <linux/mutex.h>
46 #include <linux/scatterlist.h>
47 #include <linux/bitmap.h>
48 #include <linux/list.h>
51 #include <xen/xenbus.h>
52 #include <xen/grant_table.h>
53 #include <xen/events.h>
55 #include <xen/platform_pci.h>
57 #include <xen/interface/grant_table.h>
58 #include <xen/interface/io/blkif.h>
59 #include <xen/interface/io/protocols.h>
61 #include <asm/xen/hypervisor.h>
64 * The minimal size of segment supported by the block framework is PAGE_SIZE.
65 * When Linux is using a different page size than Xen, it may not be possible
66 * to put all the data in a single segment.
67 * This can happen when the backend doesn't support indirect descriptor and
68 * therefore the maximum amount of data that a request can carry is
69 * BLKIF_MAX_SEGMENTS_PER_REQUEST * XEN_PAGE_SIZE = 44KB
71 * Note that we only support one extra request. So the Linux page size
72 * should be <= ( 2 * BLKIF_MAX_SEGMENTS_PER_REQUEST * XEN_PAGE_SIZE) =
75 #define HAS_EXTRA_REQ (BLKIF_MAX_SEGMENTS_PER_REQUEST < XEN_PFN_PER_PAGE)
78 BLKIF_STATE_DISCONNECTED
,
79 BLKIF_STATE_CONNECTED
,
80 BLKIF_STATE_SUSPENDED
,
86 struct list_head node
;
97 struct blkif_request req
;
98 struct request
*request
;
99 struct grant
**grants_used
;
100 struct grant
**indirect_grants
;
101 struct scatterlist
*sg
;
103 enum blk_req_status status
;
105 #define NO_ASSOCIATED_ID ~0UL
107 * Id of the sibling if we ever need 2 requests when handling a
110 unsigned long associated_id
;
118 static DEFINE_MUTEX(blkfront_mutex
);
119 static const struct block_device_operations xlvbd_block_fops
;
122 * Maximum number of segments in indirect requests, the actual value used by
123 * the frontend driver is the minimum of this value and the value provided
124 * by the backend driver.
127 static unsigned int xen_blkif_max_segments
= 32;
128 module_param_named(max_indirect_segments
, xen_blkif_max_segments
, uint
,
130 MODULE_PARM_DESC(max_indirect_segments
,
131 "Maximum amount of segments in indirect requests (default is 32)");
133 static unsigned int xen_blkif_max_queues
= 4;
134 module_param_named(max_queues
, xen_blkif_max_queues
, uint
, S_IRUGO
);
135 MODULE_PARM_DESC(max_queues
, "Maximum number of hardware queues/rings used per virtual disk");
138 * Maximum order of pages to be used for the shared ring between front and
139 * backend, 4KB page granularity is used.
141 static unsigned int xen_blkif_max_ring_order
;
142 module_param_named(max_ring_page_order
, xen_blkif_max_ring_order
, int, S_IRUGO
);
143 MODULE_PARM_DESC(max_ring_page_order
, "Maximum order of pages to be used for the shared ring");
145 #define BLK_RING_SIZE(info) \
146 __CONST_RING_SIZE(blkif, XEN_PAGE_SIZE * (info)->nr_ring_pages)
148 #define BLK_MAX_RING_SIZE \
149 __CONST_RING_SIZE(blkif, XEN_PAGE_SIZE * XENBUS_MAX_RING_GRANTS)
152 * ring-ref%u i=(-1UL) would take 11 characters + 'ring-ref' is 8, so 19
153 * characters are enough. Define to 20 to keep consistent with backend.
155 #define RINGREF_NAME_LEN (20)
157 * queue-%u would take 7 + 10(UINT_MAX) = 17 characters.
159 #define QUEUE_NAME_LEN (17)
163 * Every blkfront device can associate with one or more blkfront_ring_info,
164 * depending on how many hardware queues/rings to be used.
166 struct blkfront_ring_info
{
167 /* Lock to protect data in every ring buffer. */
168 spinlock_t ring_lock
;
169 struct blkif_front_ring ring
;
170 unsigned int ring_ref
[XENBUS_MAX_RING_GRANTS
];
171 unsigned int evtchn
, irq
;
172 struct work_struct work
;
173 struct gnttab_free_callback callback
;
174 struct blk_shadow shadow
[BLK_MAX_RING_SIZE
];
175 struct list_head indirect_pages
;
176 struct list_head grants
;
177 unsigned int persistent_gnts_c
;
178 unsigned long shadow_free
;
179 struct blkfront_info
*dev_info
;
183 * We have one of these per vbd, whether ide, scsi or 'other'. They
184 * hang in private_data off the gendisk structure. We may end up
185 * putting all kinds of interesting stuff here :-)
190 struct xenbus_device
*xbdev
;
194 enum blkif_state connected
;
195 /* Number of pages per ring buffer. */
196 unsigned int nr_ring_pages
;
197 struct request_queue
*rq
;
198 unsigned int feature_flush
;
199 unsigned int feature_discard
:1;
200 unsigned int feature_secdiscard
:1;
201 unsigned int discard_granularity
;
202 unsigned int discard_alignment
;
203 unsigned int feature_persistent
:1;
204 /* Number of 4KB segments handled */
205 unsigned int max_indirect_segments
;
207 struct blk_mq_tag_set tag_set
;
208 struct blkfront_ring_info
*rinfo
;
209 unsigned int nr_rings
;
212 static unsigned int nr_minors
;
213 static unsigned long *minors
;
214 static DEFINE_SPINLOCK(minor_lock
);
216 #define GRANT_INVALID_REF 0
218 #define PARTS_PER_DISK 16
219 #define PARTS_PER_EXT_DISK 256
221 #define BLKIF_MAJOR(dev) ((dev)>>8)
222 #define BLKIF_MINOR(dev) ((dev) & 0xff)
225 #define EXTENDED (1<<EXT_SHIFT)
226 #define VDEV_IS_EXTENDED(dev) ((dev)&(EXTENDED))
227 #define BLKIF_MINOR_EXT(dev) ((dev)&(~EXTENDED))
228 #define EMULATED_HD_DISK_MINOR_OFFSET (0)
229 #define EMULATED_HD_DISK_NAME_OFFSET (EMULATED_HD_DISK_MINOR_OFFSET / 256)
230 #define EMULATED_SD_DISK_MINOR_OFFSET (0)
231 #define EMULATED_SD_DISK_NAME_OFFSET (EMULATED_SD_DISK_MINOR_OFFSET / 256)
233 #define DEV_NAME "xvd" /* name in /dev */
236 * Grants are always the same size as a Xen page (i.e 4KB).
237 * A physical segment is always the same size as a Linux page.
238 * Number of grants per physical segment
240 #define GRANTS_PER_PSEG (PAGE_SIZE / XEN_PAGE_SIZE)
242 #define GRANTS_PER_INDIRECT_FRAME \
243 (XEN_PAGE_SIZE / sizeof(struct blkif_request_segment))
245 #define PSEGS_PER_INDIRECT_FRAME \
246 (GRANTS_INDIRECT_FRAME / GRANTS_PSEGS)
248 #define INDIRECT_GREFS(_grants) \
249 DIV_ROUND_UP(_grants, GRANTS_PER_INDIRECT_FRAME)
251 #define GREFS(_psegs) ((_psegs) * GRANTS_PER_PSEG)
253 static int blkfront_setup_indirect(struct blkfront_ring_info
*rinfo
);
254 static void blkfront_gather_backend_features(struct blkfront_info
*info
);
256 static int get_id_from_freelist(struct blkfront_ring_info
*rinfo
)
258 unsigned long free
= rinfo
->shadow_free
;
260 BUG_ON(free
>= BLK_RING_SIZE(rinfo
->dev_info
));
261 rinfo
->shadow_free
= rinfo
->shadow
[free
].req
.u
.rw
.id
;
262 rinfo
->shadow
[free
].req
.u
.rw
.id
= 0x0fffffee; /* debug */
266 static int add_id_to_freelist(struct blkfront_ring_info
*rinfo
,
269 if (rinfo
->shadow
[id
].req
.u
.rw
.id
!= id
)
271 if (rinfo
->shadow
[id
].request
== NULL
)
273 rinfo
->shadow
[id
].req
.u
.rw
.id
= rinfo
->shadow_free
;
274 rinfo
->shadow
[id
].request
= NULL
;
275 rinfo
->shadow_free
= id
;
279 static int fill_grant_buffer(struct blkfront_ring_info
*rinfo
, int num
)
281 struct blkfront_info
*info
= rinfo
->dev_info
;
282 struct page
*granted_page
;
283 struct grant
*gnt_list_entry
, *n
;
287 gnt_list_entry
= kzalloc(sizeof(struct grant
), GFP_NOIO
);
291 if (info
->feature_persistent
) {
292 granted_page
= alloc_page(GFP_NOIO
);
294 kfree(gnt_list_entry
);
297 gnt_list_entry
->page
= granted_page
;
300 gnt_list_entry
->gref
= GRANT_INVALID_REF
;
301 list_add(&gnt_list_entry
->node
, &rinfo
->grants
);
308 list_for_each_entry_safe(gnt_list_entry
, n
,
309 &rinfo
->grants
, node
) {
310 list_del(&gnt_list_entry
->node
);
311 if (info
->feature_persistent
)
312 __free_page(gnt_list_entry
->page
);
313 kfree(gnt_list_entry
);
320 static struct grant
*get_free_grant(struct blkfront_ring_info
*rinfo
)
322 struct grant
*gnt_list_entry
;
324 BUG_ON(list_empty(&rinfo
->grants
));
325 gnt_list_entry
= list_first_entry(&rinfo
->grants
, struct grant
,
327 list_del(&gnt_list_entry
->node
);
329 if (gnt_list_entry
->gref
!= GRANT_INVALID_REF
)
330 rinfo
->persistent_gnts_c
--;
332 return gnt_list_entry
;
335 static inline void grant_foreign_access(const struct grant
*gnt_list_entry
,
336 const struct blkfront_info
*info
)
338 gnttab_page_grant_foreign_access_ref_one(gnt_list_entry
->gref
,
339 info
->xbdev
->otherend_id
,
340 gnt_list_entry
->page
,
344 static struct grant
*get_grant(grant_ref_t
*gref_head
,
346 struct blkfront_ring_info
*rinfo
)
348 struct grant
*gnt_list_entry
= get_free_grant(rinfo
);
349 struct blkfront_info
*info
= rinfo
->dev_info
;
351 if (gnt_list_entry
->gref
!= GRANT_INVALID_REF
)
352 return gnt_list_entry
;
354 /* Assign a gref to this page */
355 gnt_list_entry
->gref
= gnttab_claim_grant_reference(gref_head
);
356 BUG_ON(gnt_list_entry
->gref
== -ENOSPC
);
357 if (info
->feature_persistent
)
358 grant_foreign_access(gnt_list_entry
, info
);
360 /* Grant access to the GFN passed by the caller */
361 gnttab_grant_foreign_access_ref(gnt_list_entry
->gref
,
362 info
->xbdev
->otherend_id
,
366 return gnt_list_entry
;
369 static struct grant
*get_indirect_grant(grant_ref_t
*gref_head
,
370 struct blkfront_ring_info
*rinfo
)
372 struct grant
*gnt_list_entry
= get_free_grant(rinfo
);
373 struct blkfront_info
*info
= rinfo
->dev_info
;
375 if (gnt_list_entry
->gref
!= GRANT_INVALID_REF
)
376 return gnt_list_entry
;
378 /* Assign a gref to this page */
379 gnt_list_entry
->gref
= gnttab_claim_grant_reference(gref_head
);
380 BUG_ON(gnt_list_entry
->gref
== -ENOSPC
);
381 if (!info
->feature_persistent
) {
382 struct page
*indirect_page
;
384 /* Fetch a pre-allocated page to use for indirect grefs */
385 BUG_ON(list_empty(&rinfo
->indirect_pages
));
386 indirect_page
= list_first_entry(&rinfo
->indirect_pages
,
388 list_del(&indirect_page
->lru
);
389 gnt_list_entry
->page
= indirect_page
;
391 grant_foreign_access(gnt_list_entry
, info
);
393 return gnt_list_entry
;
396 static const char *op_name(int op
)
398 static const char *const names
[] = {
399 [BLKIF_OP_READ
] = "read",
400 [BLKIF_OP_WRITE
] = "write",
401 [BLKIF_OP_WRITE_BARRIER
] = "barrier",
402 [BLKIF_OP_FLUSH_DISKCACHE
] = "flush",
403 [BLKIF_OP_DISCARD
] = "discard" };
405 if (op
< 0 || op
>= ARRAY_SIZE(names
))
413 static int xlbd_reserve_minors(unsigned int minor
, unsigned int nr
)
415 unsigned int end
= minor
+ nr
;
418 if (end
> nr_minors
) {
419 unsigned long *bitmap
, *old
;
421 bitmap
= kcalloc(BITS_TO_LONGS(end
), sizeof(*bitmap
),
426 spin_lock(&minor_lock
);
427 if (end
> nr_minors
) {
429 memcpy(bitmap
, minors
,
430 BITS_TO_LONGS(nr_minors
) * sizeof(*bitmap
));
432 nr_minors
= BITS_TO_LONGS(end
) * BITS_PER_LONG
;
435 spin_unlock(&minor_lock
);
439 spin_lock(&minor_lock
);
440 if (find_next_bit(minors
, end
, minor
) >= end
) {
441 bitmap_set(minors
, minor
, nr
);
445 spin_unlock(&minor_lock
);
450 static void xlbd_release_minors(unsigned int minor
, unsigned int nr
)
452 unsigned int end
= minor
+ nr
;
454 BUG_ON(end
> nr_minors
);
455 spin_lock(&minor_lock
);
456 bitmap_clear(minors
, minor
, nr
);
457 spin_unlock(&minor_lock
);
460 static void blkif_restart_queue_callback(void *arg
)
462 struct blkfront_ring_info
*rinfo
= (struct blkfront_ring_info
*)arg
;
463 schedule_work(&rinfo
->work
);
466 static int blkif_getgeo(struct block_device
*bd
, struct hd_geometry
*hg
)
468 /* We don't have real geometry info, but let's at least return
469 values consistent with the size of the device */
470 sector_t nsect
= get_capacity(bd
->bd_disk
);
471 sector_t cylinders
= nsect
;
475 sector_div(cylinders
, hg
->heads
* hg
->sectors
);
476 hg
->cylinders
= cylinders
;
477 if ((sector_t
)(hg
->cylinders
+ 1) * hg
->heads
* hg
->sectors
< nsect
)
478 hg
->cylinders
= 0xffff;
482 static int blkif_ioctl(struct block_device
*bdev
, fmode_t mode
,
483 unsigned command
, unsigned long argument
)
485 struct blkfront_info
*info
= bdev
->bd_disk
->private_data
;
488 dev_dbg(&info
->xbdev
->dev
, "command: 0x%x, argument: 0x%lx\n",
489 command
, (long)argument
);
492 case CDROMMULTISESSION
:
493 dev_dbg(&info
->xbdev
->dev
, "FIXME: support multisession CDs later\n");
494 for (i
= 0; i
< sizeof(struct cdrom_multisession
); i
++)
495 if (put_user(0, (char __user
*)(argument
+ i
)))
499 case CDROM_GET_CAPABILITY
: {
500 struct gendisk
*gd
= info
->gd
;
501 if (gd
->flags
& GENHD_FL_CD
)
507 /*printk(KERN_ALERT "ioctl %08x not supported by Xen blkdev\n",
509 return -EINVAL
; /* same return as native Linux */
515 static unsigned long blkif_ring_get_request(struct blkfront_ring_info
*rinfo
,
517 struct blkif_request
**ring_req
)
521 *ring_req
= RING_GET_REQUEST(&rinfo
->ring
, rinfo
->ring
.req_prod_pvt
);
522 rinfo
->ring
.req_prod_pvt
++;
524 id
= get_id_from_freelist(rinfo
);
525 rinfo
->shadow
[id
].request
= req
;
526 rinfo
->shadow
[id
].status
= REQ_WAITING
;
527 rinfo
->shadow
[id
].associated_id
= NO_ASSOCIATED_ID
;
529 (*ring_req
)->u
.rw
.id
= id
;
534 static int blkif_queue_discard_req(struct request
*req
, struct blkfront_ring_info
*rinfo
)
536 struct blkfront_info
*info
= rinfo
->dev_info
;
537 struct blkif_request
*ring_req
;
540 /* Fill out a communications ring structure. */
541 id
= blkif_ring_get_request(rinfo
, req
, &ring_req
);
543 ring_req
->operation
= BLKIF_OP_DISCARD
;
544 ring_req
->u
.discard
.nr_sectors
= blk_rq_sectors(req
);
545 ring_req
->u
.discard
.id
= id
;
546 ring_req
->u
.discard
.sector_number
= (blkif_sector_t
)blk_rq_pos(req
);
547 if ((req
->cmd_flags
& REQ_SECURE
) && info
->feature_secdiscard
)
548 ring_req
->u
.discard
.flag
= BLKIF_DISCARD_SECURE
;
550 ring_req
->u
.discard
.flag
= 0;
552 /* Keep a private copy so we can reissue requests when recovering. */
553 rinfo
->shadow
[id
].req
= *ring_req
;
558 struct setup_rw_req
{
559 unsigned int grant_idx
;
560 struct blkif_request_segment
*segments
;
561 struct blkfront_ring_info
*rinfo
;
562 struct blkif_request
*ring_req
;
563 grant_ref_t gref_head
;
565 /* Only used when persistent grant is used and it's a read request */
567 unsigned int bvec_off
;
570 bool require_extra_req
;
571 struct blkif_request
*extra_ring_req
;
574 static void blkif_setup_rw_req_grant(unsigned long gfn
, unsigned int offset
,
575 unsigned int len
, void *data
)
577 struct setup_rw_req
*setup
= data
;
579 struct grant
*gnt_list_entry
;
580 unsigned int fsect
, lsect
;
581 /* Convenient aliases */
582 unsigned int grant_idx
= setup
->grant_idx
;
583 struct blkif_request
*ring_req
= setup
->ring_req
;
584 struct blkfront_ring_info
*rinfo
= setup
->rinfo
;
586 * We always use the shadow of the first request to store the list
587 * of grant associated to the block I/O request. This made the
588 * completion more easy to handle even if the block I/O request is
591 struct blk_shadow
*shadow
= &rinfo
->shadow
[setup
->id
];
593 if (unlikely(setup
->require_extra_req
&&
594 grant_idx
>= BLKIF_MAX_SEGMENTS_PER_REQUEST
)) {
596 * We are using the second request, setup grant_idx
597 * to be the index of the segment array.
599 grant_idx
-= BLKIF_MAX_SEGMENTS_PER_REQUEST
;
600 ring_req
= setup
->extra_ring_req
;
603 if ((ring_req
->operation
== BLKIF_OP_INDIRECT
) &&
604 (grant_idx
% GRANTS_PER_INDIRECT_FRAME
== 0)) {
606 kunmap_atomic(setup
->segments
);
608 n
= grant_idx
/ GRANTS_PER_INDIRECT_FRAME
;
609 gnt_list_entry
= get_indirect_grant(&setup
->gref_head
, rinfo
);
610 shadow
->indirect_grants
[n
] = gnt_list_entry
;
611 setup
->segments
= kmap_atomic(gnt_list_entry
->page
);
612 ring_req
->u
.indirect
.indirect_grefs
[n
] = gnt_list_entry
->gref
;
615 gnt_list_entry
= get_grant(&setup
->gref_head
, gfn
, rinfo
);
616 ref
= gnt_list_entry
->gref
;
618 * All the grants are stored in the shadow of the first
619 * request. Therefore we have to use the global index.
621 shadow
->grants_used
[setup
->grant_idx
] = gnt_list_entry
;
623 if (setup
->need_copy
) {
626 shared_data
= kmap_atomic(gnt_list_entry
->page
);
628 * this does not wipe data stored outside the
629 * range sg->offset..sg->offset+sg->length.
630 * Therefore, blkback *could* see data from
631 * previous requests. This is OK as long as
632 * persistent grants are shared with just one
633 * domain. It may need refactoring if this
636 memcpy(shared_data
+ offset
,
637 setup
->bvec_data
+ setup
->bvec_off
,
640 kunmap_atomic(shared_data
);
641 setup
->bvec_off
+= len
;
645 lsect
= fsect
+ (len
>> 9) - 1;
646 if (ring_req
->operation
!= BLKIF_OP_INDIRECT
) {
647 ring_req
->u
.rw
.seg
[grant_idx
] =
648 (struct blkif_request_segment
) {
651 .last_sect
= lsect
};
653 setup
->segments
[grant_idx
% GRANTS_PER_INDIRECT_FRAME
] =
654 (struct blkif_request_segment
) {
657 .last_sect
= lsect
};
660 (setup
->grant_idx
)++;
663 static void blkif_setup_extra_req(struct blkif_request
*first
,
664 struct blkif_request
*second
)
666 uint16_t nr_segments
= first
->u
.rw
.nr_segments
;
669 * The second request is only present when the first request uses
670 * all its segments. It's always the continuity of the first one.
672 first
->u
.rw
.nr_segments
= BLKIF_MAX_SEGMENTS_PER_REQUEST
;
674 second
->u
.rw
.nr_segments
= nr_segments
- BLKIF_MAX_SEGMENTS_PER_REQUEST
;
675 second
->u
.rw
.sector_number
= first
->u
.rw
.sector_number
+
676 (BLKIF_MAX_SEGMENTS_PER_REQUEST
* XEN_PAGE_SIZE
) / 512;
678 second
->u
.rw
.handle
= first
->u
.rw
.handle
;
679 second
->operation
= first
->operation
;
682 static int blkif_queue_rw_req(struct request
*req
, struct blkfront_ring_info
*rinfo
)
684 struct blkfront_info
*info
= rinfo
->dev_info
;
685 struct blkif_request
*ring_req
, *extra_ring_req
= NULL
;
686 unsigned long id
, extra_id
= NO_ASSOCIATED_ID
;
687 bool require_extra_req
= false;
689 struct setup_rw_req setup
= {
693 .need_copy
= rq_data_dir(req
) && info
->feature_persistent
,
697 * Used to store if we are able to queue the request by just using
698 * existing persistent grants, or if we have to get new grants,
699 * as there are not sufficiently many free.
701 struct scatterlist
*sg
;
702 int num_sg
, max_grefs
, num_grant
;
704 max_grefs
= req
->nr_phys_segments
* GRANTS_PER_PSEG
;
705 if (max_grefs
> BLKIF_MAX_SEGMENTS_PER_REQUEST
)
707 * If we are using indirect segments we need to account
708 * for the indirect grefs used in the request.
710 max_grefs
+= INDIRECT_GREFS(max_grefs
);
713 * We have to reserve 'max_grefs' grants because persistent
714 * grants are shared by all rings.
717 if (gnttab_alloc_grant_references(max_grefs
, &setup
.gref_head
) < 0) {
718 gnttab_request_free_callback(
720 blkif_restart_queue_callback
,
726 /* Fill out a communications ring structure. */
727 id
= blkif_ring_get_request(rinfo
, req
, &ring_req
);
729 num_sg
= blk_rq_map_sg(req
->q
, req
, rinfo
->shadow
[id
].sg
);
731 /* Calculate the number of grant used */
732 for_each_sg(rinfo
->shadow
[id
].sg
, sg
, num_sg
, i
)
733 num_grant
+= gnttab_count_grant(sg
->offset
, sg
->length
);
735 require_extra_req
= info
->max_indirect_segments
== 0 &&
736 num_grant
> BLKIF_MAX_SEGMENTS_PER_REQUEST
;
737 BUG_ON(!HAS_EXTRA_REQ
&& require_extra_req
);
739 rinfo
->shadow
[id
].num_sg
= num_sg
;
740 if (num_grant
> BLKIF_MAX_SEGMENTS_PER_REQUEST
&&
741 likely(!require_extra_req
)) {
743 * The indirect operation can only be a BLKIF_OP_READ or
746 BUG_ON(req
->cmd_flags
& (REQ_FLUSH
| REQ_FUA
));
747 ring_req
->operation
= BLKIF_OP_INDIRECT
;
748 ring_req
->u
.indirect
.indirect_op
= rq_data_dir(req
) ?
749 BLKIF_OP_WRITE
: BLKIF_OP_READ
;
750 ring_req
->u
.indirect
.sector_number
= (blkif_sector_t
)blk_rq_pos(req
);
751 ring_req
->u
.indirect
.handle
= info
->handle
;
752 ring_req
->u
.indirect
.nr_segments
= num_grant
;
754 ring_req
->u
.rw
.sector_number
= (blkif_sector_t
)blk_rq_pos(req
);
755 ring_req
->u
.rw
.handle
= info
->handle
;
756 ring_req
->operation
= rq_data_dir(req
) ?
757 BLKIF_OP_WRITE
: BLKIF_OP_READ
;
758 if (req
->cmd_flags
& (REQ_FLUSH
| REQ_FUA
)) {
760 * Ideally we can do an unordered flush-to-disk.
761 * In case the backend onlysupports barriers, use that.
762 * A barrier request a superset of FUA, so we can
763 * implement it the same way. (It's also a FLUSH+FUA,
764 * since it is guaranteed ordered WRT previous writes.)
766 switch (info
->feature_flush
&
767 ((REQ_FLUSH
|REQ_FUA
))) {
768 case REQ_FLUSH
|REQ_FUA
:
769 ring_req
->operation
=
770 BLKIF_OP_WRITE_BARRIER
;
773 ring_req
->operation
=
774 BLKIF_OP_FLUSH_DISKCACHE
;
777 ring_req
->operation
= 0;
780 ring_req
->u
.rw
.nr_segments
= num_grant
;
781 if (unlikely(require_extra_req
)) {
782 extra_id
= blkif_ring_get_request(rinfo
, req
,
785 * Only the first request contains the scatter-gather
788 rinfo
->shadow
[extra_id
].num_sg
= 0;
790 blkif_setup_extra_req(ring_req
, extra_ring_req
);
792 /* Link the 2 requests together */
793 rinfo
->shadow
[extra_id
].associated_id
= id
;
794 rinfo
->shadow
[id
].associated_id
= extra_id
;
798 setup
.ring_req
= ring_req
;
801 setup
.require_extra_req
= require_extra_req
;
802 if (unlikely(require_extra_req
))
803 setup
.extra_ring_req
= extra_ring_req
;
805 for_each_sg(rinfo
->shadow
[id
].sg
, sg
, num_sg
, i
) {
806 BUG_ON(sg
->offset
+ sg
->length
> PAGE_SIZE
);
808 if (setup
.need_copy
) {
809 setup
.bvec_off
= sg
->offset
;
810 setup
.bvec_data
= kmap_atomic(sg_page(sg
));
813 gnttab_foreach_grant_in_range(sg_page(sg
),
816 blkif_setup_rw_req_grant
,
820 kunmap_atomic(setup
.bvec_data
);
823 kunmap_atomic(setup
.segments
);
825 /* Keep a private copy so we can reissue requests when recovering. */
826 rinfo
->shadow
[id
].req
= *ring_req
;
827 if (unlikely(require_extra_req
))
828 rinfo
->shadow
[extra_id
].req
= *extra_ring_req
;
831 gnttab_free_grant_references(setup
.gref_head
);
837 * Generate a Xen blkfront IO request from a blk layer request. Reads
838 * and writes are handled as expected.
840 * @req: a request struct
842 static int blkif_queue_request(struct request
*req
, struct blkfront_ring_info
*rinfo
)
844 if (unlikely(rinfo
->dev_info
->connected
!= BLKIF_STATE_CONNECTED
))
847 if (unlikely(req
->cmd_flags
& (REQ_DISCARD
| REQ_SECURE
)))
848 return blkif_queue_discard_req(req
, rinfo
);
850 return blkif_queue_rw_req(req
, rinfo
);
853 static inline void flush_requests(struct blkfront_ring_info
*rinfo
)
857 RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&rinfo
->ring
, notify
);
860 notify_remote_via_irq(rinfo
->irq
);
863 static inline bool blkif_request_flush_invalid(struct request
*req
,
864 struct blkfront_info
*info
)
866 return ((req
->cmd_type
!= REQ_TYPE_FS
) ||
867 ((req
->cmd_flags
& REQ_FLUSH
) &&
868 !(info
->feature_flush
& REQ_FLUSH
)) ||
869 ((req
->cmd_flags
& REQ_FUA
) &&
870 !(info
->feature_flush
& REQ_FUA
)));
873 static int blkif_queue_rq(struct blk_mq_hw_ctx
*hctx
,
874 const struct blk_mq_queue_data
*qd
)
877 int qid
= hctx
->queue_num
;
878 struct blkfront_info
*info
= hctx
->queue
->queuedata
;
879 struct blkfront_ring_info
*rinfo
= NULL
;
881 BUG_ON(info
->nr_rings
<= qid
);
882 rinfo
= &info
->rinfo
[qid
];
883 blk_mq_start_request(qd
->rq
);
884 spin_lock_irqsave(&rinfo
->ring_lock
, flags
);
885 if (RING_FULL(&rinfo
->ring
))
888 if (blkif_request_flush_invalid(qd
->rq
, rinfo
->dev_info
))
891 if (blkif_queue_request(qd
->rq
, rinfo
))
894 flush_requests(rinfo
);
895 spin_unlock_irqrestore(&rinfo
->ring_lock
, flags
);
896 return BLK_MQ_RQ_QUEUE_OK
;
899 spin_unlock_irqrestore(&rinfo
->ring_lock
, flags
);
900 return BLK_MQ_RQ_QUEUE_ERROR
;
903 spin_unlock_irqrestore(&rinfo
->ring_lock
, flags
);
904 blk_mq_stop_hw_queue(hctx
);
905 return BLK_MQ_RQ_QUEUE_BUSY
;
908 static struct blk_mq_ops blkfront_mq_ops
= {
909 .queue_rq
= blkif_queue_rq
,
910 .map_queue
= blk_mq_map_queue
,
913 static int xlvbd_init_blk_queue(struct gendisk
*gd
, u16 sector_size
,
914 unsigned int physical_sector_size
,
915 unsigned int segments
)
917 struct request_queue
*rq
;
918 struct blkfront_info
*info
= gd
->private_data
;
920 memset(&info
->tag_set
, 0, sizeof(info
->tag_set
));
921 info
->tag_set
.ops
= &blkfront_mq_ops
;
922 info
->tag_set
.nr_hw_queues
= info
->nr_rings
;
923 if (HAS_EXTRA_REQ
&& info
->max_indirect_segments
== 0) {
925 * When indirect descriptior is not supported, the I/O request
926 * will be split between multiple request in the ring.
927 * To avoid problems when sending the request, divide by
928 * 2 the depth of the queue.
930 info
->tag_set
.queue_depth
= BLK_RING_SIZE(info
) / 2;
932 info
->tag_set
.queue_depth
= BLK_RING_SIZE(info
);
933 info
->tag_set
.numa_node
= NUMA_NO_NODE
;
934 info
->tag_set
.flags
= BLK_MQ_F_SHOULD_MERGE
| BLK_MQ_F_SG_MERGE
;
935 info
->tag_set
.cmd_size
= 0;
936 info
->tag_set
.driver_data
= info
;
938 if (blk_mq_alloc_tag_set(&info
->tag_set
))
940 rq
= blk_mq_init_queue(&info
->tag_set
);
942 blk_mq_free_tag_set(&info
->tag_set
);
946 rq
->queuedata
= info
;
947 queue_flag_set_unlocked(QUEUE_FLAG_VIRT
, rq
);
949 if (info
->feature_discard
) {
950 queue_flag_set_unlocked(QUEUE_FLAG_DISCARD
, rq
);
951 blk_queue_max_discard_sectors(rq
, get_capacity(gd
));
952 rq
->limits
.discard_granularity
= info
->discard_granularity
;
953 rq
->limits
.discard_alignment
= info
->discard_alignment
;
954 if (info
->feature_secdiscard
)
955 queue_flag_set_unlocked(QUEUE_FLAG_SECDISCARD
, rq
);
958 /* Hard sector size and max sectors impersonate the equiv. hardware. */
959 blk_queue_logical_block_size(rq
, sector_size
);
960 blk_queue_physical_block_size(rq
, physical_sector_size
);
961 blk_queue_max_hw_sectors(rq
, (segments
* XEN_PAGE_SIZE
) / 512);
963 /* Each segment in a request is up to an aligned page in size. */
964 blk_queue_segment_boundary(rq
, PAGE_SIZE
- 1);
965 blk_queue_max_segment_size(rq
, PAGE_SIZE
);
967 /* Ensure a merged request will fit in a single I/O ring slot. */
968 blk_queue_max_segments(rq
, segments
/ GRANTS_PER_PSEG
);
970 /* Make sure buffer addresses are sector-aligned. */
971 blk_queue_dma_alignment(rq
, 511);
973 /* Make sure we don't use bounce buffers. */
974 blk_queue_bounce_limit(rq
, BLK_BOUNCE_ANY
);
981 static const char *flush_info(unsigned int feature_flush
)
983 switch (feature_flush
& ((REQ_FLUSH
| REQ_FUA
))) {
984 case REQ_FLUSH
|REQ_FUA
:
985 return "barrier: enabled;";
987 return "flush diskcache: enabled;";
989 return "barrier or flush: disabled;";
993 static void xlvbd_flush(struct blkfront_info
*info
)
995 blk_queue_write_cache(info
->rq
, info
->feature_flush
& REQ_FLUSH
,
996 info
->feature_flush
& REQ_FUA
);
997 pr_info("blkfront: %s: %s %s %s %s %s\n",
998 info
->gd
->disk_name
, flush_info(info
->feature_flush
),
999 "persistent grants:", info
->feature_persistent
?
1000 "enabled;" : "disabled;", "indirect descriptors:",
1001 info
->max_indirect_segments
? "enabled;" : "disabled;");
1004 static int xen_translate_vdev(int vdevice
, int *minor
, unsigned int *offset
)
1007 major
= BLKIF_MAJOR(vdevice
);
1008 *minor
= BLKIF_MINOR(vdevice
);
1010 case XEN_IDE0_MAJOR
:
1011 *offset
= (*minor
/ 64) + EMULATED_HD_DISK_NAME_OFFSET
;
1012 *minor
= ((*minor
/ 64) * PARTS_PER_DISK
) +
1013 EMULATED_HD_DISK_MINOR_OFFSET
;
1015 case XEN_IDE1_MAJOR
:
1016 *offset
= (*minor
/ 64) + 2 + EMULATED_HD_DISK_NAME_OFFSET
;
1017 *minor
= (((*minor
/ 64) + 2) * PARTS_PER_DISK
) +
1018 EMULATED_HD_DISK_MINOR_OFFSET
;
1020 case XEN_SCSI_DISK0_MAJOR
:
1021 *offset
= (*minor
/ PARTS_PER_DISK
) + EMULATED_SD_DISK_NAME_OFFSET
;
1022 *minor
= *minor
+ EMULATED_SD_DISK_MINOR_OFFSET
;
1024 case XEN_SCSI_DISK1_MAJOR
:
1025 case XEN_SCSI_DISK2_MAJOR
:
1026 case XEN_SCSI_DISK3_MAJOR
:
1027 case XEN_SCSI_DISK4_MAJOR
:
1028 case XEN_SCSI_DISK5_MAJOR
:
1029 case XEN_SCSI_DISK6_MAJOR
:
1030 case XEN_SCSI_DISK7_MAJOR
:
1031 *offset
= (*minor
/ PARTS_PER_DISK
) +
1032 ((major
- XEN_SCSI_DISK1_MAJOR
+ 1) * 16) +
1033 EMULATED_SD_DISK_NAME_OFFSET
;
1035 ((major
- XEN_SCSI_DISK1_MAJOR
+ 1) * 16 * PARTS_PER_DISK
) +
1036 EMULATED_SD_DISK_MINOR_OFFSET
;
1038 case XEN_SCSI_DISK8_MAJOR
:
1039 case XEN_SCSI_DISK9_MAJOR
:
1040 case XEN_SCSI_DISK10_MAJOR
:
1041 case XEN_SCSI_DISK11_MAJOR
:
1042 case XEN_SCSI_DISK12_MAJOR
:
1043 case XEN_SCSI_DISK13_MAJOR
:
1044 case XEN_SCSI_DISK14_MAJOR
:
1045 case XEN_SCSI_DISK15_MAJOR
:
1046 *offset
= (*minor
/ PARTS_PER_DISK
) +
1047 ((major
- XEN_SCSI_DISK8_MAJOR
+ 8) * 16) +
1048 EMULATED_SD_DISK_NAME_OFFSET
;
1050 ((major
- XEN_SCSI_DISK8_MAJOR
+ 8) * 16 * PARTS_PER_DISK
) +
1051 EMULATED_SD_DISK_MINOR_OFFSET
;
1054 *offset
= *minor
/ PARTS_PER_DISK
;
1057 printk(KERN_WARNING
"blkfront: your disk configuration is "
1058 "incorrect, please use an xvd device instead\n");
1064 static char *encode_disk_name(char *ptr
, unsigned int n
)
1067 ptr
= encode_disk_name(ptr
, n
/ 26 - 1);
1068 *ptr
= 'a' + n
% 26;
1072 static int xlvbd_alloc_gendisk(blkif_sector_t capacity
,
1073 struct blkfront_info
*info
,
1074 u16 vdisk_info
, u16 sector_size
,
1075 unsigned int physical_sector_size
)
1080 unsigned int offset
;
1085 BUG_ON(info
->gd
!= NULL
);
1086 BUG_ON(info
->rq
!= NULL
);
1088 if ((info
->vdevice
>>EXT_SHIFT
) > 1) {
1089 /* this is above the extended range; something is wrong */
1090 printk(KERN_WARNING
"blkfront: vdevice 0x%x is above the extended range; ignoring\n", info
->vdevice
);
1094 if (!VDEV_IS_EXTENDED(info
->vdevice
)) {
1095 err
= xen_translate_vdev(info
->vdevice
, &minor
, &offset
);
1098 nr_parts
= PARTS_PER_DISK
;
1100 minor
= BLKIF_MINOR_EXT(info
->vdevice
);
1101 nr_parts
= PARTS_PER_EXT_DISK
;
1102 offset
= minor
/ nr_parts
;
1103 if (xen_hvm_domain() && offset
< EMULATED_HD_DISK_NAME_OFFSET
+ 4)
1104 printk(KERN_WARNING
"blkfront: vdevice 0x%x might conflict with "
1105 "emulated IDE disks,\n\t choose an xvd device name"
1106 "from xvde on\n", info
->vdevice
);
1108 if (minor
>> MINORBITS
) {
1109 pr_warn("blkfront: %#x's minor (%#x) out of range; ignoring\n",
1110 info
->vdevice
, minor
);
1114 if ((minor
% nr_parts
) == 0)
1115 nr_minors
= nr_parts
;
1117 err
= xlbd_reserve_minors(minor
, nr_minors
);
1122 gd
= alloc_disk(nr_minors
);
1126 strcpy(gd
->disk_name
, DEV_NAME
);
1127 ptr
= encode_disk_name(gd
->disk_name
+ sizeof(DEV_NAME
) - 1, offset
);
1128 BUG_ON(ptr
>= gd
->disk_name
+ DISK_NAME_LEN
);
1132 snprintf(ptr
, gd
->disk_name
+ DISK_NAME_LEN
- ptr
,
1133 "%d", minor
& (nr_parts
- 1));
1135 gd
->major
= XENVBD_MAJOR
;
1136 gd
->first_minor
= minor
;
1137 gd
->fops
= &xlvbd_block_fops
;
1138 gd
->private_data
= info
;
1139 gd
->driverfs_dev
= &(info
->xbdev
->dev
);
1140 set_capacity(gd
, capacity
);
1142 if (xlvbd_init_blk_queue(gd
, sector_size
, physical_sector_size
,
1143 info
->max_indirect_segments
? :
1144 BLKIF_MAX_SEGMENTS_PER_REQUEST
)) {
1149 info
->rq
= gd
->queue
;
1154 if (vdisk_info
& VDISK_READONLY
)
1157 if (vdisk_info
& VDISK_REMOVABLE
)
1158 gd
->flags
|= GENHD_FL_REMOVABLE
;
1160 if (vdisk_info
& VDISK_CDROM
)
1161 gd
->flags
|= GENHD_FL_CD
;
1166 xlbd_release_minors(minor
, nr_minors
);
1171 static void xlvbd_release_gendisk(struct blkfront_info
*info
)
1173 unsigned int minor
, nr_minors
, i
;
1175 if (info
->rq
== NULL
)
1178 /* No more blkif_request(). */
1179 blk_mq_stop_hw_queues(info
->rq
);
1181 for (i
= 0; i
< info
->nr_rings
; i
++) {
1182 struct blkfront_ring_info
*rinfo
= &info
->rinfo
[i
];
1184 /* No more gnttab callback work. */
1185 gnttab_cancel_free_callback(&rinfo
->callback
);
1187 /* Flush gnttab callback work. Must be done with no locks held. */
1188 flush_work(&rinfo
->work
);
1191 del_gendisk(info
->gd
);
1193 minor
= info
->gd
->first_minor
;
1194 nr_minors
= info
->gd
->minors
;
1195 xlbd_release_minors(minor
, nr_minors
);
1197 blk_cleanup_queue(info
->rq
);
1198 blk_mq_free_tag_set(&info
->tag_set
);
1205 /* Already hold rinfo->ring_lock. */
1206 static inline void kick_pending_request_queues_locked(struct blkfront_ring_info
*rinfo
)
1208 if (!RING_FULL(&rinfo
->ring
))
1209 blk_mq_start_stopped_hw_queues(rinfo
->dev_info
->rq
, true);
1212 static void kick_pending_request_queues(struct blkfront_ring_info
*rinfo
)
1214 unsigned long flags
;
1216 spin_lock_irqsave(&rinfo
->ring_lock
, flags
);
1217 kick_pending_request_queues_locked(rinfo
);
1218 spin_unlock_irqrestore(&rinfo
->ring_lock
, flags
);
1221 static void blkif_restart_queue(struct work_struct
*work
)
1223 struct blkfront_ring_info
*rinfo
= container_of(work
, struct blkfront_ring_info
, work
);
1225 if (rinfo
->dev_info
->connected
== BLKIF_STATE_CONNECTED
)
1226 kick_pending_request_queues(rinfo
);
1229 static void blkif_free_ring(struct blkfront_ring_info
*rinfo
)
1231 struct grant
*persistent_gnt
, *n
;
1232 struct blkfront_info
*info
= rinfo
->dev_info
;
1236 * Remove indirect pages, this only happens when using indirect
1237 * descriptors but not persistent grants
1239 if (!list_empty(&rinfo
->indirect_pages
)) {
1240 struct page
*indirect_page
, *n
;
1242 BUG_ON(info
->feature_persistent
);
1243 list_for_each_entry_safe(indirect_page
, n
, &rinfo
->indirect_pages
, lru
) {
1244 list_del(&indirect_page
->lru
);
1245 __free_page(indirect_page
);
1249 /* Remove all persistent grants. */
1250 if (!list_empty(&rinfo
->grants
)) {
1251 list_for_each_entry_safe(persistent_gnt
, n
,
1252 &rinfo
->grants
, node
) {
1253 list_del(&persistent_gnt
->node
);
1254 if (persistent_gnt
->gref
!= GRANT_INVALID_REF
) {
1255 gnttab_end_foreign_access(persistent_gnt
->gref
,
1257 rinfo
->persistent_gnts_c
--;
1259 if (info
->feature_persistent
)
1260 __free_page(persistent_gnt
->page
);
1261 kfree(persistent_gnt
);
1264 BUG_ON(rinfo
->persistent_gnts_c
!= 0);
1266 for (i
= 0; i
< BLK_RING_SIZE(info
); i
++) {
1268 * Clear persistent grants present in requests already
1269 * on the shared ring
1271 if (!rinfo
->shadow
[i
].request
)
1274 segs
= rinfo
->shadow
[i
].req
.operation
== BLKIF_OP_INDIRECT
?
1275 rinfo
->shadow
[i
].req
.u
.indirect
.nr_segments
:
1276 rinfo
->shadow
[i
].req
.u
.rw
.nr_segments
;
1277 for (j
= 0; j
< segs
; j
++) {
1278 persistent_gnt
= rinfo
->shadow
[i
].grants_used
[j
];
1279 gnttab_end_foreign_access(persistent_gnt
->gref
, 0, 0UL);
1280 if (info
->feature_persistent
)
1281 __free_page(persistent_gnt
->page
);
1282 kfree(persistent_gnt
);
1285 if (rinfo
->shadow
[i
].req
.operation
!= BLKIF_OP_INDIRECT
)
1287 * If this is not an indirect operation don't try to
1288 * free indirect segments
1292 for (j
= 0; j
< INDIRECT_GREFS(segs
); j
++) {
1293 persistent_gnt
= rinfo
->shadow
[i
].indirect_grants
[j
];
1294 gnttab_end_foreign_access(persistent_gnt
->gref
, 0, 0UL);
1295 __free_page(persistent_gnt
->page
);
1296 kfree(persistent_gnt
);
1300 kfree(rinfo
->shadow
[i
].grants_used
);
1301 rinfo
->shadow
[i
].grants_used
= NULL
;
1302 kfree(rinfo
->shadow
[i
].indirect_grants
);
1303 rinfo
->shadow
[i
].indirect_grants
= NULL
;
1304 kfree(rinfo
->shadow
[i
].sg
);
1305 rinfo
->shadow
[i
].sg
= NULL
;
1308 /* No more gnttab callback work. */
1309 gnttab_cancel_free_callback(&rinfo
->callback
);
1311 /* Flush gnttab callback work. Must be done with no locks held. */
1312 flush_work(&rinfo
->work
);
1314 /* Free resources associated with old device channel. */
1315 for (i
= 0; i
< info
->nr_ring_pages
; i
++) {
1316 if (rinfo
->ring_ref
[i
] != GRANT_INVALID_REF
) {
1317 gnttab_end_foreign_access(rinfo
->ring_ref
[i
], 0, 0);
1318 rinfo
->ring_ref
[i
] = GRANT_INVALID_REF
;
1321 free_pages((unsigned long)rinfo
->ring
.sring
, get_order(info
->nr_ring_pages
* PAGE_SIZE
));
1322 rinfo
->ring
.sring
= NULL
;
1325 unbind_from_irqhandler(rinfo
->irq
, rinfo
);
1326 rinfo
->evtchn
= rinfo
->irq
= 0;
1329 static void blkif_free(struct blkfront_info
*info
, int suspend
)
1333 /* Prevent new requests being issued until we fix things up. */
1334 info
->connected
= suspend
?
1335 BLKIF_STATE_SUSPENDED
: BLKIF_STATE_DISCONNECTED
;
1336 /* No more blkif_request(). */
1338 blk_mq_stop_hw_queues(info
->rq
);
1340 for (i
= 0; i
< info
->nr_rings
; i
++)
1341 blkif_free_ring(&info
->rinfo
[i
]);
1348 struct copy_from_grant
{
1349 const struct blk_shadow
*s
;
1350 unsigned int grant_idx
;
1351 unsigned int bvec_offset
;
1355 static void blkif_copy_from_grant(unsigned long gfn
, unsigned int offset
,
1356 unsigned int len
, void *data
)
1358 struct copy_from_grant
*info
= data
;
1360 /* Convenient aliases */
1361 const struct blk_shadow
*s
= info
->s
;
1363 shared_data
= kmap_atomic(s
->grants_used
[info
->grant_idx
]->page
);
1365 memcpy(info
->bvec_data
+ info
->bvec_offset
,
1366 shared_data
+ offset
, len
);
1368 info
->bvec_offset
+= len
;
1371 kunmap_atomic(shared_data
);
1374 static enum blk_req_status
blkif_rsp_to_req_status(int rsp
)
1378 case BLKIF_RSP_OKAY
:
1380 case BLKIF_RSP_EOPNOTSUPP
:
1381 return REQ_EOPNOTSUPP
;
1382 case BLKIF_RSP_ERROR
:
1390 * Get the final status of the block request based on two ring response
1392 static int blkif_get_final_status(enum blk_req_status s1
,
1393 enum blk_req_status s2
)
1395 BUG_ON(s1
== REQ_WAITING
);
1396 BUG_ON(s2
== REQ_WAITING
);
1398 if (s1
== REQ_ERROR
|| s2
== REQ_ERROR
)
1399 return BLKIF_RSP_ERROR
;
1400 else if (s1
== REQ_EOPNOTSUPP
|| s2
== REQ_EOPNOTSUPP
)
1401 return BLKIF_RSP_EOPNOTSUPP
;
1402 return BLKIF_RSP_OKAY
;
1405 static bool blkif_completion(unsigned long *id
,
1406 struct blkfront_ring_info
*rinfo
,
1407 struct blkif_response
*bret
)
1410 struct scatterlist
*sg
;
1411 int num_sg
, num_grant
;
1412 struct blkfront_info
*info
= rinfo
->dev_info
;
1413 struct blk_shadow
*s
= &rinfo
->shadow
[*id
];
1414 struct copy_from_grant data
= {
1418 num_grant
= s
->req
.operation
== BLKIF_OP_INDIRECT
?
1419 s
->req
.u
.indirect
.nr_segments
: s
->req
.u
.rw
.nr_segments
;
1421 /* The I/O request may be split in two. */
1422 if (unlikely(s
->associated_id
!= NO_ASSOCIATED_ID
)) {
1423 struct blk_shadow
*s2
= &rinfo
->shadow
[s
->associated_id
];
1425 /* Keep the status of the current response in shadow. */
1426 s
->status
= blkif_rsp_to_req_status(bret
->status
);
1428 /* Wait the second response if not yet here. */
1429 if (s2
->status
== REQ_WAITING
)
1432 bret
->status
= blkif_get_final_status(s
->status
,
1436 * All the grants is stored in the first shadow in order
1437 * to make the completion code simpler.
1439 num_grant
+= s2
->req
.u
.rw
.nr_segments
;
1442 * The two responses may not come in order. Only the
1443 * first request will store the scatter-gather list.
1445 if (s2
->num_sg
!= 0) {
1446 /* Update "id" with the ID of the first response. */
1447 *id
= s
->associated_id
;
1452 * We don't need anymore the second request, so recycling
1455 if (add_id_to_freelist(rinfo
, s
->associated_id
))
1456 WARN(1, "%s: can't recycle the second part (id = %ld) of the request\n",
1457 info
->gd
->disk_name
, s
->associated_id
);
1463 if (bret
->operation
== BLKIF_OP_READ
&& info
->feature_persistent
) {
1464 for_each_sg(s
->sg
, sg
, num_sg
, i
) {
1465 BUG_ON(sg
->offset
+ sg
->length
> PAGE_SIZE
);
1467 data
.bvec_offset
= sg
->offset
;
1468 data
.bvec_data
= kmap_atomic(sg_page(sg
));
1470 gnttab_foreach_grant_in_range(sg_page(sg
),
1473 blkif_copy_from_grant
,
1476 kunmap_atomic(data
.bvec_data
);
1479 /* Add the persistent grant into the list of free grants */
1480 for (i
= 0; i
< num_grant
; i
++) {
1481 if (gnttab_query_foreign_access(s
->grants_used
[i
]->gref
)) {
1483 * If the grant is still mapped by the backend (the
1484 * backend has chosen to make this grant persistent)
1485 * we add it at the head of the list, so it will be
1488 if (!info
->feature_persistent
)
1489 pr_alert_ratelimited("backed has not unmapped grant: %u\n",
1490 s
->grants_used
[i
]->gref
);
1491 list_add(&s
->grants_used
[i
]->node
, &rinfo
->grants
);
1492 rinfo
->persistent_gnts_c
++;
1495 * If the grant is not mapped by the backend we end the
1496 * foreign access and add it to the tail of the list,
1497 * so it will not be picked again unless we run out of
1498 * persistent grants.
1500 gnttab_end_foreign_access(s
->grants_used
[i
]->gref
, 0, 0UL);
1501 s
->grants_used
[i
]->gref
= GRANT_INVALID_REF
;
1502 list_add_tail(&s
->grants_used
[i
]->node
, &rinfo
->grants
);
1505 if (s
->req
.operation
== BLKIF_OP_INDIRECT
) {
1506 for (i
= 0; i
< INDIRECT_GREFS(num_grant
); i
++) {
1507 if (gnttab_query_foreign_access(s
->indirect_grants
[i
]->gref
)) {
1508 if (!info
->feature_persistent
)
1509 pr_alert_ratelimited("backed has not unmapped grant: %u\n",
1510 s
->indirect_grants
[i
]->gref
);
1511 list_add(&s
->indirect_grants
[i
]->node
, &rinfo
->grants
);
1512 rinfo
->persistent_gnts_c
++;
1514 struct page
*indirect_page
;
1516 gnttab_end_foreign_access(s
->indirect_grants
[i
]->gref
, 0, 0UL);
1518 * Add the used indirect page back to the list of
1519 * available pages for indirect grefs.
1521 if (!info
->feature_persistent
) {
1522 indirect_page
= s
->indirect_grants
[i
]->page
;
1523 list_add(&indirect_page
->lru
, &rinfo
->indirect_pages
);
1525 s
->indirect_grants
[i
]->gref
= GRANT_INVALID_REF
;
1526 list_add_tail(&s
->indirect_grants
[i
]->node
, &rinfo
->grants
);
1534 static irqreturn_t
blkif_interrupt(int irq
, void *dev_id
)
1536 struct request
*req
;
1537 struct blkif_response
*bret
;
1539 unsigned long flags
;
1540 struct blkfront_ring_info
*rinfo
= (struct blkfront_ring_info
*)dev_id
;
1541 struct blkfront_info
*info
= rinfo
->dev_info
;
1544 if (unlikely(info
->connected
!= BLKIF_STATE_CONNECTED
))
1547 spin_lock_irqsave(&rinfo
->ring_lock
, flags
);
1549 rp
= rinfo
->ring
.sring
->rsp_prod
;
1550 rmb(); /* Ensure we see queued responses up to 'rp'. */
1552 for (i
= rinfo
->ring
.rsp_cons
; i
!= rp
; i
++) {
1555 bret
= RING_GET_RESPONSE(&rinfo
->ring
, i
);
1558 * The backend has messed up and given us an id that we would
1559 * never have given to it (we stamp it up to BLK_RING_SIZE -
1560 * look in get_id_from_freelist.
1562 if (id
>= BLK_RING_SIZE(info
)) {
1563 WARN(1, "%s: response to %s has incorrect id (%ld)\n",
1564 info
->gd
->disk_name
, op_name(bret
->operation
), id
);
1565 /* We can't safely get the 'struct request' as
1566 * the id is busted. */
1569 req
= rinfo
->shadow
[id
].request
;
1571 if (bret
->operation
!= BLKIF_OP_DISCARD
) {
1573 * We may need to wait for an extra response if the
1574 * I/O request is split in 2
1576 if (!blkif_completion(&id
, rinfo
, bret
))
1580 if (add_id_to_freelist(rinfo
, id
)) {
1581 WARN(1, "%s: response to %s (id %ld) couldn't be recycled!\n",
1582 info
->gd
->disk_name
, op_name(bret
->operation
), id
);
1586 error
= (bret
->status
== BLKIF_RSP_OKAY
) ? 0 : -EIO
;
1587 switch (bret
->operation
) {
1588 case BLKIF_OP_DISCARD
:
1589 if (unlikely(bret
->status
== BLKIF_RSP_EOPNOTSUPP
)) {
1590 struct request_queue
*rq
= info
->rq
;
1591 printk(KERN_WARNING
"blkfront: %s: %s op failed\n",
1592 info
->gd
->disk_name
, op_name(bret
->operation
));
1593 error
= -EOPNOTSUPP
;
1594 info
->feature_discard
= 0;
1595 info
->feature_secdiscard
= 0;
1596 queue_flag_clear(QUEUE_FLAG_DISCARD
, rq
);
1597 queue_flag_clear(QUEUE_FLAG_SECDISCARD
, rq
);
1599 blk_mq_complete_request(req
, error
);
1601 case BLKIF_OP_FLUSH_DISKCACHE
:
1602 case BLKIF_OP_WRITE_BARRIER
:
1603 if (unlikely(bret
->status
== BLKIF_RSP_EOPNOTSUPP
)) {
1604 printk(KERN_WARNING
"blkfront: %s: %s op failed\n",
1605 info
->gd
->disk_name
, op_name(bret
->operation
));
1606 error
= -EOPNOTSUPP
;
1608 if (unlikely(bret
->status
== BLKIF_RSP_ERROR
&&
1609 rinfo
->shadow
[id
].req
.u
.rw
.nr_segments
== 0)) {
1610 printk(KERN_WARNING
"blkfront: %s: empty %s op failed\n",
1611 info
->gd
->disk_name
, op_name(bret
->operation
));
1612 error
= -EOPNOTSUPP
;
1614 if (unlikely(error
)) {
1615 if (error
== -EOPNOTSUPP
)
1617 info
->feature_flush
= 0;
1622 case BLKIF_OP_WRITE
:
1623 if (unlikely(bret
->status
!= BLKIF_RSP_OKAY
))
1624 dev_dbg(&info
->xbdev
->dev
, "Bad return from blkdev data "
1625 "request: %x\n", bret
->status
);
1627 blk_mq_complete_request(req
, error
);
1634 rinfo
->ring
.rsp_cons
= i
;
1636 if (i
!= rinfo
->ring
.req_prod_pvt
) {
1638 RING_FINAL_CHECK_FOR_RESPONSES(&rinfo
->ring
, more_to_do
);
1642 rinfo
->ring
.sring
->rsp_event
= i
+ 1;
1644 kick_pending_request_queues_locked(rinfo
);
1646 spin_unlock_irqrestore(&rinfo
->ring_lock
, flags
);
1652 static int setup_blkring(struct xenbus_device
*dev
,
1653 struct blkfront_ring_info
*rinfo
)
1655 struct blkif_sring
*sring
;
1657 struct blkfront_info
*info
= rinfo
->dev_info
;
1658 unsigned long ring_size
= info
->nr_ring_pages
* XEN_PAGE_SIZE
;
1659 grant_ref_t gref
[XENBUS_MAX_RING_GRANTS
];
1661 for (i
= 0; i
< info
->nr_ring_pages
; i
++)
1662 rinfo
->ring_ref
[i
] = GRANT_INVALID_REF
;
1664 sring
= (struct blkif_sring
*)__get_free_pages(GFP_NOIO
| __GFP_HIGH
,
1665 get_order(ring_size
));
1667 xenbus_dev_fatal(dev
, -ENOMEM
, "allocating shared ring");
1670 SHARED_RING_INIT(sring
);
1671 FRONT_RING_INIT(&rinfo
->ring
, sring
, ring_size
);
1673 err
= xenbus_grant_ring(dev
, rinfo
->ring
.sring
, info
->nr_ring_pages
, gref
);
1675 free_pages((unsigned long)sring
, get_order(ring_size
));
1676 rinfo
->ring
.sring
= NULL
;
1679 for (i
= 0; i
< info
->nr_ring_pages
; i
++)
1680 rinfo
->ring_ref
[i
] = gref
[i
];
1682 err
= xenbus_alloc_evtchn(dev
, &rinfo
->evtchn
);
1686 err
= bind_evtchn_to_irqhandler(rinfo
->evtchn
, blkif_interrupt
, 0,
1689 xenbus_dev_fatal(dev
, err
,
1690 "bind_evtchn_to_irqhandler failed");
1697 blkif_free(info
, 0);
1702 * Write out per-ring/queue nodes including ring-ref and event-channel, and each
1703 * ring buffer may have multi pages depending on ->nr_ring_pages.
1705 static int write_per_ring_nodes(struct xenbus_transaction xbt
,
1706 struct blkfront_ring_info
*rinfo
, const char *dir
)
1710 const char *message
= NULL
;
1711 struct blkfront_info
*info
= rinfo
->dev_info
;
1713 if (info
->nr_ring_pages
== 1) {
1714 err
= xenbus_printf(xbt
, dir
, "ring-ref", "%u", rinfo
->ring_ref
[0]);
1716 message
= "writing ring-ref";
1717 goto abort_transaction
;
1720 for (i
= 0; i
< info
->nr_ring_pages
; i
++) {
1721 char ring_ref_name
[RINGREF_NAME_LEN
];
1723 snprintf(ring_ref_name
, RINGREF_NAME_LEN
, "ring-ref%u", i
);
1724 err
= xenbus_printf(xbt
, dir
, ring_ref_name
,
1725 "%u", rinfo
->ring_ref
[i
]);
1727 message
= "writing ring-ref";
1728 goto abort_transaction
;
1733 err
= xenbus_printf(xbt
, dir
, "event-channel", "%u", rinfo
->evtchn
);
1735 message
= "writing event-channel";
1736 goto abort_transaction
;
1742 xenbus_transaction_end(xbt
, 1);
1744 xenbus_dev_fatal(info
->xbdev
, err
, "%s", message
);
1749 /* Common code used when first setting up, and when resuming. */
1750 static int talk_to_blkback(struct xenbus_device
*dev
,
1751 struct blkfront_info
*info
)
1753 const char *message
= NULL
;
1754 struct xenbus_transaction xbt
;
1756 unsigned int i
, max_page_order
= 0;
1757 unsigned int ring_page_order
= 0;
1759 err
= xenbus_scanf(XBT_NIL
, info
->xbdev
->otherend
,
1760 "max-ring-page-order", "%u", &max_page_order
);
1762 info
->nr_ring_pages
= 1;
1764 ring_page_order
= min(xen_blkif_max_ring_order
, max_page_order
);
1765 info
->nr_ring_pages
= 1 << ring_page_order
;
1768 for (i
= 0; i
< info
->nr_rings
; i
++) {
1769 struct blkfront_ring_info
*rinfo
= &info
->rinfo
[i
];
1771 /* Create shared ring, alloc event channel. */
1772 err
= setup_blkring(dev
, rinfo
);
1774 goto destroy_blkring
;
1778 err
= xenbus_transaction_start(&xbt
);
1780 xenbus_dev_fatal(dev
, err
, "starting transaction");
1781 goto destroy_blkring
;
1784 if (info
->nr_ring_pages
> 1) {
1785 err
= xenbus_printf(xbt
, dev
->nodename
, "ring-page-order", "%u",
1788 message
= "writing ring-page-order";
1789 goto abort_transaction
;
1793 /* We already got the number of queues/rings in _probe */
1794 if (info
->nr_rings
== 1) {
1795 err
= write_per_ring_nodes(xbt
, &info
->rinfo
[0], dev
->nodename
);
1797 goto destroy_blkring
;
1802 err
= xenbus_printf(xbt
, dev
->nodename
, "multi-queue-num-queues", "%u",
1805 message
= "writing multi-queue-num-queues";
1806 goto abort_transaction
;
1809 pathsize
= strlen(dev
->nodename
) + QUEUE_NAME_LEN
;
1810 path
= kmalloc(pathsize
, GFP_KERNEL
);
1813 message
= "ENOMEM while writing ring references";
1814 goto abort_transaction
;
1817 for (i
= 0; i
< info
->nr_rings
; i
++) {
1818 memset(path
, 0, pathsize
);
1819 snprintf(path
, pathsize
, "%s/queue-%u", dev
->nodename
, i
);
1820 err
= write_per_ring_nodes(xbt
, &info
->rinfo
[i
], path
);
1823 goto destroy_blkring
;
1828 err
= xenbus_printf(xbt
, dev
->nodename
, "protocol", "%s",
1829 XEN_IO_PROTO_ABI_NATIVE
);
1831 message
= "writing protocol";
1832 goto abort_transaction
;
1834 err
= xenbus_printf(xbt
, dev
->nodename
,
1835 "feature-persistent", "%u", 1);
1838 "writing persistent grants feature to xenbus");
1840 err
= xenbus_transaction_end(xbt
, 0);
1844 xenbus_dev_fatal(dev
, err
, "completing transaction");
1845 goto destroy_blkring
;
1848 for (i
= 0; i
< info
->nr_rings
; i
++) {
1850 struct blkfront_ring_info
*rinfo
= &info
->rinfo
[i
];
1852 for (j
= 0; j
< BLK_RING_SIZE(info
); j
++)
1853 rinfo
->shadow
[j
].req
.u
.rw
.id
= j
+ 1;
1854 rinfo
->shadow
[BLK_RING_SIZE(info
)-1].req
.u
.rw
.id
= 0x0fffffff;
1856 xenbus_switch_state(dev
, XenbusStateInitialised
);
1861 xenbus_transaction_end(xbt
, 1);
1863 xenbus_dev_fatal(dev
, err
, "%s", message
);
1865 blkif_free(info
, 0);
1868 dev_set_drvdata(&dev
->dev
, NULL
);
1873 static int negotiate_mq(struct blkfront_info
*info
)
1875 unsigned int backend_max_queues
= 0;
1879 BUG_ON(info
->nr_rings
);
1881 /* Check if backend supports multiple queues. */
1882 err
= xenbus_scanf(XBT_NIL
, info
->xbdev
->otherend
,
1883 "multi-queue-max-queues", "%u", &backend_max_queues
);
1885 backend_max_queues
= 1;
1887 info
->nr_rings
= min(backend_max_queues
, xen_blkif_max_queues
);
1888 /* We need at least one ring. */
1889 if (!info
->nr_rings
)
1892 info
->rinfo
= kzalloc(sizeof(struct blkfront_ring_info
) * info
->nr_rings
, GFP_KERNEL
);
1894 xenbus_dev_fatal(info
->xbdev
, -ENOMEM
, "allocating ring_info structure");
1898 for (i
= 0; i
< info
->nr_rings
; i
++) {
1899 struct blkfront_ring_info
*rinfo
;
1901 rinfo
= &info
->rinfo
[i
];
1902 INIT_LIST_HEAD(&rinfo
->indirect_pages
);
1903 INIT_LIST_HEAD(&rinfo
->grants
);
1904 rinfo
->dev_info
= info
;
1905 INIT_WORK(&rinfo
->work
, blkif_restart_queue
);
1906 spin_lock_init(&rinfo
->ring_lock
);
1911 * Entry point to this code when a new device is created. Allocate the basic
1912 * structures and the ring buffer for communication with the backend, and
1913 * inform the backend of the appropriate details for those. Switch to
1914 * Initialised state.
1916 static int blkfront_probe(struct xenbus_device
*dev
,
1917 const struct xenbus_device_id
*id
)
1920 struct blkfront_info
*info
;
1922 /* FIXME: Use dynamic device id if this is not set. */
1923 err
= xenbus_scanf(XBT_NIL
, dev
->nodename
,
1924 "virtual-device", "%i", &vdevice
);
1926 /* go looking in the extended area instead */
1927 err
= xenbus_scanf(XBT_NIL
, dev
->nodename
, "virtual-device-ext",
1930 xenbus_dev_fatal(dev
, err
, "reading virtual-device");
1935 if (xen_hvm_domain()) {
1938 /* no unplug has been done: do not hook devices != xen vbds */
1939 if (xen_has_pv_and_legacy_disk_devices()) {
1942 if (!VDEV_IS_EXTENDED(vdevice
))
1943 major
= BLKIF_MAJOR(vdevice
);
1945 major
= XENVBD_MAJOR
;
1947 if (major
!= XENVBD_MAJOR
) {
1949 "%s: HVM does not support vbd %d as xen block device\n",
1954 /* do not create a PV cdrom device if we are an HVM guest */
1955 type
= xenbus_read(XBT_NIL
, dev
->nodename
, "device-type", &len
);
1958 if (strncmp(type
, "cdrom", 5) == 0) {
1964 info
= kzalloc(sizeof(*info
), GFP_KERNEL
);
1966 xenbus_dev_fatal(dev
, -ENOMEM
, "allocating info structure");
1971 err
= negotiate_mq(info
);
1977 mutex_init(&info
->mutex
);
1978 info
->vdevice
= vdevice
;
1979 info
->connected
= BLKIF_STATE_DISCONNECTED
;
1981 /* Front end dir is a number, which is used as the id. */
1982 info
->handle
= simple_strtoul(strrchr(dev
->nodename
, '/')+1, NULL
, 0);
1983 dev_set_drvdata(&dev
->dev
, info
);
1988 static void split_bio_end(struct bio
*bio
)
1990 struct split_bio
*split_bio
= bio
->bi_private
;
1992 if (atomic_dec_and_test(&split_bio
->pending
)) {
1993 split_bio
->bio
->bi_phys_segments
= 0;
1994 split_bio
->bio
->bi_error
= bio
->bi_error
;
1995 bio_endio(split_bio
->bio
);
2001 static int blkif_recover(struct blkfront_info
*info
)
2003 unsigned int i
, r_index
;
2004 struct request
*req
, *n
;
2005 struct blk_shadow
*copy
;
2007 struct bio
*bio
, *cloned_bio
;
2008 struct bio_list bio_list
, merge_bio
;
2009 unsigned int segs
, offset
;
2011 struct split_bio
*split_bio
;
2012 struct list_head requests
;
2014 blkfront_gather_backend_features(info
);
2015 segs
= info
->max_indirect_segments
? : BLKIF_MAX_SEGMENTS_PER_REQUEST
;
2016 blk_queue_max_segments(info
->rq
, segs
);
2017 bio_list_init(&bio_list
);
2018 INIT_LIST_HEAD(&requests
);
2020 for (r_index
= 0; r_index
< info
->nr_rings
; r_index
++) {
2021 struct blkfront_ring_info
*rinfo
;
2023 rinfo
= &info
->rinfo
[r_index
];
2024 /* Stage 1: Make a safe copy of the shadow state. */
2025 copy
= kmemdup(rinfo
->shadow
, sizeof(rinfo
->shadow
),
2026 GFP_NOIO
| __GFP_REPEAT
| __GFP_HIGH
);
2030 /* Stage 2: Set up free list. */
2031 memset(&rinfo
->shadow
, 0, sizeof(rinfo
->shadow
));
2032 for (i
= 0; i
< BLK_RING_SIZE(info
); i
++)
2033 rinfo
->shadow
[i
].req
.u
.rw
.id
= i
+1;
2034 rinfo
->shadow_free
= rinfo
->ring
.req_prod_pvt
;
2035 rinfo
->shadow
[BLK_RING_SIZE(info
)-1].req
.u
.rw
.id
= 0x0fffffff;
2037 rc
= blkfront_setup_indirect(rinfo
);
2043 for (i
= 0; i
< BLK_RING_SIZE(info
); i
++) {
2045 if (!copy
[i
].request
)
2049 * Get the bios in the request so we can re-queue them.
2051 if (copy
[i
].request
->cmd_flags
&
2052 (REQ_FLUSH
| REQ_FUA
| REQ_DISCARD
| REQ_SECURE
)) {
2054 * Flush operations don't contain bios, so
2055 * we need to requeue the whole request
2057 list_add(©
[i
].request
->queuelist
, &requests
);
2060 merge_bio
.head
= copy
[i
].request
->bio
;
2061 merge_bio
.tail
= copy
[i
].request
->biotail
;
2062 bio_list_merge(&bio_list
, &merge_bio
);
2063 copy
[i
].request
->bio
= NULL
;
2064 blk_end_request_all(copy
[i
].request
, 0);
2069 xenbus_switch_state(info
->xbdev
, XenbusStateConnected
);
2071 /* Now safe for us to use the shared ring */
2072 info
->connected
= BLKIF_STATE_CONNECTED
;
2074 for (r_index
= 0; r_index
< info
->nr_rings
; r_index
++) {
2075 struct blkfront_ring_info
*rinfo
;
2077 rinfo
= &info
->rinfo
[r_index
];
2078 /* Kick any other new requests queued since we resumed */
2079 kick_pending_request_queues(rinfo
);
2082 list_for_each_entry_safe(req
, n
, &requests
, queuelist
) {
2083 /* Requeue pending requests (flush or discard) */
2084 list_del_init(&req
->queuelist
);
2085 BUG_ON(req
->nr_phys_segments
> segs
);
2086 blk_mq_requeue_request(req
);
2088 blk_mq_kick_requeue_list(info
->rq
);
2090 while ((bio
= bio_list_pop(&bio_list
)) != NULL
) {
2091 /* Traverse the list of pending bios and re-queue them */
2092 if (bio_segments(bio
) > segs
) {
2094 * This bio has more segments than what we can
2095 * handle, we have to split it.
2097 pending
= (bio_segments(bio
) + segs
- 1) / segs
;
2098 split_bio
= kzalloc(sizeof(*split_bio
), GFP_NOIO
);
2099 BUG_ON(split_bio
== NULL
);
2100 atomic_set(&split_bio
->pending
, pending
);
2101 split_bio
->bio
= bio
;
2102 for (i
= 0; i
< pending
; i
++) {
2103 offset
= (i
* segs
* XEN_PAGE_SIZE
) >> 9;
2104 size
= min((unsigned int)(segs
* XEN_PAGE_SIZE
) >> 9,
2105 (unsigned int)bio_sectors(bio
) - offset
);
2106 cloned_bio
= bio_clone(bio
, GFP_NOIO
);
2107 BUG_ON(cloned_bio
== NULL
);
2108 bio_trim(cloned_bio
, offset
, size
);
2109 cloned_bio
->bi_private
= split_bio
;
2110 cloned_bio
->bi_end_io
= split_bio_end
;
2111 submit_bio(cloned_bio
->bi_rw
, cloned_bio
);
2114 * Now we have to wait for all those smaller bios to
2115 * end, so we can also end the "parent" bio.
2119 /* We don't need to split this bio */
2120 submit_bio(bio
->bi_rw
, bio
);
2127 * We are reconnecting to the backend, due to a suspend/resume, or a backend
2128 * driver restart. We tear down our blkif structure and recreate it, but
2129 * leave the device-layer structures intact so that this is transparent to the
2130 * rest of the kernel.
2132 static int blkfront_resume(struct xenbus_device
*dev
)
2134 struct blkfront_info
*info
= dev_get_drvdata(&dev
->dev
);
2137 dev_dbg(&dev
->dev
, "blkfront_resume: %s\n", dev
->nodename
);
2139 blkif_free(info
, info
->connected
== BLKIF_STATE_CONNECTED
);
2141 err
= negotiate_mq(info
);
2145 err
= talk_to_blkback(dev
, info
);
2147 blk_mq_update_nr_hw_queues(&info
->tag_set
, info
->nr_rings
);
2150 * We have to wait for the backend to switch to
2151 * connected state, since we want to read which
2152 * features it supports.
2158 static void blkfront_closing(struct blkfront_info
*info
)
2160 struct xenbus_device
*xbdev
= info
->xbdev
;
2161 struct block_device
*bdev
= NULL
;
2163 mutex_lock(&info
->mutex
);
2165 if (xbdev
->state
== XenbusStateClosing
) {
2166 mutex_unlock(&info
->mutex
);
2171 bdev
= bdget_disk(info
->gd
, 0);
2173 mutex_unlock(&info
->mutex
);
2176 xenbus_frontend_closed(xbdev
);
2180 mutex_lock(&bdev
->bd_mutex
);
2182 if (bdev
->bd_openers
) {
2183 xenbus_dev_error(xbdev
, -EBUSY
,
2184 "Device in use; refusing to close");
2185 xenbus_switch_state(xbdev
, XenbusStateClosing
);
2187 xlvbd_release_gendisk(info
);
2188 xenbus_frontend_closed(xbdev
);
2191 mutex_unlock(&bdev
->bd_mutex
);
2195 static void blkfront_setup_discard(struct blkfront_info
*info
)
2198 unsigned int discard_granularity
;
2199 unsigned int discard_alignment
;
2200 unsigned int discard_secure
;
2202 info
->feature_discard
= 1;
2203 err
= xenbus_gather(XBT_NIL
, info
->xbdev
->otherend
,
2204 "discard-granularity", "%u", &discard_granularity
,
2205 "discard-alignment", "%u", &discard_alignment
,
2208 info
->discard_granularity
= discard_granularity
;
2209 info
->discard_alignment
= discard_alignment
;
2211 err
= xenbus_gather(XBT_NIL
, info
->xbdev
->otherend
,
2212 "discard-secure", "%d", &discard_secure
,
2215 info
->feature_secdiscard
= !!discard_secure
;
2218 static int blkfront_setup_indirect(struct blkfront_ring_info
*rinfo
)
2220 unsigned int psegs
, grants
;
2222 struct blkfront_info
*info
= rinfo
->dev_info
;
2224 if (info
->max_indirect_segments
== 0) {
2226 grants
= BLKIF_MAX_SEGMENTS_PER_REQUEST
;
2229 * When an extra req is required, the maximum
2230 * grants supported is related to the size of the
2231 * Linux block segment.
2233 grants
= GRANTS_PER_PSEG
;
2237 grants
= info
->max_indirect_segments
;
2238 psegs
= grants
/ GRANTS_PER_PSEG
;
2240 err
= fill_grant_buffer(rinfo
,
2241 (grants
+ INDIRECT_GREFS(grants
)) * BLK_RING_SIZE(info
));
2245 if (!info
->feature_persistent
&& info
->max_indirect_segments
) {
2247 * We are using indirect descriptors but not persistent
2248 * grants, we need to allocate a set of pages that can be
2249 * used for mapping indirect grefs
2251 int num
= INDIRECT_GREFS(grants
) * BLK_RING_SIZE(info
);
2253 BUG_ON(!list_empty(&rinfo
->indirect_pages
));
2254 for (i
= 0; i
< num
; i
++) {
2255 struct page
*indirect_page
= alloc_page(GFP_NOIO
);
2258 list_add(&indirect_page
->lru
, &rinfo
->indirect_pages
);
2262 for (i
= 0; i
< BLK_RING_SIZE(info
); i
++) {
2263 rinfo
->shadow
[i
].grants_used
= kzalloc(
2264 sizeof(rinfo
->shadow
[i
].grants_used
[0]) * grants
,
2266 rinfo
->shadow
[i
].sg
= kzalloc(sizeof(rinfo
->shadow
[i
].sg
[0]) * psegs
, GFP_NOIO
);
2267 if (info
->max_indirect_segments
)
2268 rinfo
->shadow
[i
].indirect_grants
= kzalloc(
2269 sizeof(rinfo
->shadow
[i
].indirect_grants
[0]) *
2270 INDIRECT_GREFS(grants
),
2272 if ((rinfo
->shadow
[i
].grants_used
== NULL
) ||
2273 (rinfo
->shadow
[i
].sg
== NULL
) ||
2274 (info
->max_indirect_segments
&&
2275 (rinfo
->shadow
[i
].indirect_grants
== NULL
)))
2277 sg_init_table(rinfo
->shadow
[i
].sg
, psegs
);
2284 for (i
= 0; i
< BLK_RING_SIZE(info
); i
++) {
2285 kfree(rinfo
->shadow
[i
].grants_used
);
2286 rinfo
->shadow
[i
].grants_used
= NULL
;
2287 kfree(rinfo
->shadow
[i
].sg
);
2288 rinfo
->shadow
[i
].sg
= NULL
;
2289 kfree(rinfo
->shadow
[i
].indirect_grants
);
2290 rinfo
->shadow
[i
].indirect_grants
= NULL
;
2292 if (!list_empty(&rinfo
->indirect_pages
)) {
2293 struct page
*indirect_page
, *n
;
2294 list_for_each_entry_safe(indirect_page
, n
, &rinfo
->indirect_pages
, lru
) {
2295 list_del(&indirect_page
->lru
);
2296 __free_page(indirect_page
);
2303 * Gather all backend feature-*
2305 static void blkfront_gather_backend_features(struct blkfront_info
*info
)
2308 int barrier
, flush
, discard
, persistent
;
2309 unsigned int indirect_segments
;
2311 info
->feature_flush
= 0;
2313 err
= xenbus_gather(XBT_NIL
, info
->xbdev
->otherend
,
2314 "feature-barrier", "%d", &barrier
,
2318 * If there's no "feature-barrier" defined, then it means
2319 * we're dealing with a very old backend which writes
2320 * synchronously; nothing to do.
2322 * If there are barriers, then we use flush.
2324 if (!err
&& barrier
)
2325 info
->feature_flush
= REQ_FLUSH
| REQ_FUA
;
2327 * And if there is "feature-flush-cache" use that above
2330 err
= xenbus_gather(XBT_NIL
, info
->xbdev
->otherend
,
2331 "feature-flush-cache", "%d", &flush
,
2335 info
->feature_flush
= REQ_FLUSH
;
2337 err
= xenbus_gather(XBT_NIL
, info
->xbdev
->otherend
,
2338 "feature-discard", "%d", &discard
,
2341 if (!err
&& discard
)
2342 blkfront_setup_discard(info
);
2344 err
= xenbus_gather(XBT_NIL
, info
->xbdev
->otherend
,
2345 "feature-persistent", "%u", &persistent
,
2348 info
->feature_persistent
= 0;
2350 info
->feature_persistent
= persistent
;
2352 err
= xenbus_gather(XBT_NIL
, info
->xbdev
->otherend
,
2353 "feature-max-indirect-segments", "%u", &indirect_segments
,
2356 info
->max_indirect_segments
= 0;
2358 info
->max_indirect_segments
= min(indirect_segments
,
2359 xen_blkif_max_segments
);
2363 * Invoked when the backend is finally 'ready' (and has told produced
2364 * the details about the physical device - #sectors, size, etc).
2366 static void blkfront_connect(struct blkfront_info
*info
)
2368 unsigned long long sectors
;
2369 unsigned long sector_size
;
2370 unsigned int physical_sector_size
;
2374 switch (info
->connected
) {
2375 case BLKIF_STATE_CONNECTED
:
2377 * Potentially, the back-end may be signalling
2378 * a capacity change; update the capacity.
2380 err
= xenbus_scanf(XBT_NIL
, info
->xbdev
->otherend
,
2381 "sectors", "%Lu", §ors
);
2382 if (XENBUS_EXIST_ERR(err
))
2384 printk(KERN_INFO
"Setting capacity to %Lu\n",
2386 set_capacity(info
->gd
, sectors
);
2387 revalidate_disk(info
->gd
);
2390 case BLKIF_STATE_SUSPENDED
:
2392 * If we are recovering from suspension, we need to wait
2393 * for the backend to announce it's features before
2394 * reconnecting, at least we need to know if the backend
2395 * supports indirect descriptors, and how many.
2397 blkif_recover(info
);
2404 dev_dbg(&info
->xbdev
->dev
, "%s:%s.\n",
2405 __func__
, info
->xbdev
->otherend
);
2407 err
= xenbus_gather(XBT_NIL
, info
->xbdev
->otherend
,
2408 "sectors", "%llu", §ors
,
2409 "info", "%u", &binfo
,
2410 "sector-size", "%lu", §or_size
,
2413 xenbus_dev_fatal(info
->xbdev
, err
,
2414 "reading backend fields at %s",
2415 info
->xbdev
->otherend
);
2420 * physcial-sector-size is a newer field, so old backends may not
2421 * provide this. Assume physical sector size to be the same as
2422 * sector_size in that case.
2424 err
= xenbus_scanf(XBT_NIL
, info
->xbdev
->otherend
,
2425 "physical-sector-size", "%u", &physical_sector_size
);
2427 physical_sector_size
= sector_size
;
2429 blkfront_gather_backend_features(info
);
2430 for (i
= 0; i
< info
->nr_rings
; i
++) {
2431 err
= blkfront_setup_indirect(&info
->rinfo
[i
]);
2433 xenbus_dev_fatal(info
->xbdev
, err
, "setup_indirect at %s",
2434 info
->xbdev
->otherend
);
2435 blkif_free(info
, 0);
2440 err
= xlvbd_alloc_gendisk(sectors
, info
, binfo
, sector_size
,
2441 physical_sector_size
);
2443 xenbus_dev_fatal(info
->xbdev
, err
, "xlvbd_add at %s",
2444 info
->xbdev
->otherend
);
2448 xenbus_switch_state(info
->xbdev
, XenbusStateConnected
);
2450 /* Kick pending requests. */
2451 info
->connected
= BLKIF_STATE_CONNECTED
;
2452 for (i
= 0; i
< info
->nr_rings
; i
++)
2453 kick_pending_request_queues(&info
->rinfo
[i
]);
2461 * Callback received when the backend's state changes.
2463 static void blkback_changed(struct xenbus_device
*dev
,
2464 enum xenbus_state backend_state
)
2466 struct blkfront_info
*info
= dev_get_drvdata(&dev
->dev
);
2468 dev_dbg(&dev
->dev
, "blkfront:blkback_changed to state %d.\n", backend_state
);
2470 switch (backend_state
) {
2471 case XenbusStateInitWait
:
2472 if (dev
->state
!= XenbusStateInitialising
)
2474 if (talk_to_blkback(dev
, info
))
2476 case XenbusStateInitialising
:
2477 case XenbusStateInitialised
:
2478 case XenbusStateReconfiguring
:
2479 case XenbusStateReconfigured
:
2480 case XenbusStateUnknown
:
2483 case XenbusStateConnected
:
2485 * talk_to_blkback sets state to XenbusStateInitialised
2486 * and blkfront_connect sets it to XenbusStateConnected
2487 * (if connection went OK).
2489 * If the backend (or toolstack) decides to poke at backend
2490 * state (and re-trigger the watch by setting the state repeatedly
2491 * to XenbusStateConnected (4)) we need to deal with this.
2492 * This is allowed as this is used to communicate to the guest
2493 * that the size of disk has changed!
2495 if ((dev
->state
!= XenbusStateInitialised
) &&
2496 (dev
->state
!= XenbusStateConnected
)) {
2497 if (talk_to_blkback(dev
, info
))
2501 blkfront_connect(info
);
2504 case XenbusStateClosed
:
2505 if (dev
->state
== XenbusStateClosed
)
2507 /* Missed the backend's Closing state -- fallthrough */
2508 case XenbusStateClosing
:
2510 blkfront_closing(info
);
2515 static int blkfront_remove(struct xenbus_device
*xbdev
)
2517 struct blkfront_info
*info
= dev_get_drvdata(&xbdev
->dev
);
2518 struct block_device
*bdev
= NULL
;
2519 struct gendisk
*disk
;
2521 dev_dbg(&xbdev
->dev
, "%s removed", xbdev
->nodename
);
2523 blkif_free(info
, 0);
2525 mutex_lock(&info
->mutex
);
2529 bdev
= bdget_disk(disk
, 0);
2532 mutex_unlock(&info
->mutex
);
2540 * The xbdev was removed before we reached the Closed
2541 * state. See if it's safe to remove the disk. If the bdev
2542 * isn't closed yet, we let release take care of it.
2545 mutex_lock(&bdev
->bd_mutex
);
2546 info
= disk
->private_data
;
2548 dev_warn(disk_to_dev(disk
),
2549 "%s was hot-unplugged, %d stale handles\n",
2550 xbdev
->nodename
, bdev
->bd_openers
);
2552 if (info
&& !bdev
->bd_openers
) {
2553 xlvbd_release_gendisk(info
);
2554 disk
->private_data
= NULL
;
2558 mutex_unlock(&bdev
->bd_mutex
);
2564 static int blkfront_is_ready(struct xenbus_device
*dev
)
2566 struct blkfront_info
*info
= dev_get_drvdata(&dev
->dev
);
2568 return info
->is_ready
&& info
->xbdev
;
2571 static int blkif_open(struct block_device
*bdev
, fmode_t mode
)
2573 struct gendisk
*disk
= bdev
->bd_disk
;
2574 struct blkfront_info
*info
;
2577 mutex_lock(&blkfront_mutex
);
2579 info
= disk
->private_data
;
2586 mutex_lock(&info
->mutex
);
2589 /* xbdev is closed */
2592 mutex_unlock(&info
->mutex
);
2595 mutex_unlock(&blkfront_mutex
);
2599 static void blkif_release(struct gendisk
*disk
, fmode_t mode
)
2601 struct blkfront_info
*info
= disk
->private_data
;
2602 struct block_device
*bdev
;
2603 struct xenbus_device
*xbdev
;
2605 mutex_lock(&blkfront_mutex
);
2607 bdev
= bdget_disk(disk
, 0);
2610 WARN(1, "Block device %s yanked out from us!\n", disk
->disk_name
);
2613 if (bdev
->bd_openers
)
2617 * Check if we have been instructed to close. We will have
2618 * deferred this request, because the bdev was still open.
2621 mutex_lock(&info
->mutex
);
2622 xbdev
= info
->xbdev
;
2624 if (xbdev
&& xbdev
->state
== XenbusStateClosing
) {
2625 /* pending switch to state closed */
2626 dev_info(disk_to_dev(bdev
->bd_disk
), "releasing disk\n");
2627 xlvbd_release_gendisk(info
);
2628 xenbus_frontend_closed(info
->xbdev
);
2631 mutex_unlock(&info
->mutex
);
2634 /* sudden device removal */
2635 dev_info(disk_to_dev(bdev
->bd_disk
), "releasing disk\n");
2636 xlvbd_release_gendisk(info
);
2637 disk
->private_data
= NULL
;
2644 mutex_unlock(&blkfront_mutex
);
2647 static const struct block_device_operations xlvbd_block_fops
=
2649 .owner
= THIS_MODULE
,
2651 .release
= blkif_release
,
2652 .getgeo
= blkif_getgeo
,
2653 .ioctl
= blkif_ioctl
,
2657 static const struct xenbus_device_id blkfront_ids
[] = {
2662 static struct xenbus_driver blkfront_driver
= {
2663 .ids
= blkfront_ids
,
2664 .probe
= blkfront_probe
,
2665 .remove
= blkfront_remove
,
2666 .resume
= blkfront_resume
,
2667 .otherend_changed
= blkback_changed
,
2668 .is_ready
= blkfront_is_ready
,
2671 static int __init
xlblk_init(void)
2674 int nr_cpus
= num_online_cpus();
2679 if (xen_blkif_max_ring_order
> XENBUS_MAX_RING_GRANT_ORDER
) {
2680 pr_info("Invalid max_ring_order (%d), will use default max: %d.\n",
2681 xen_blkif_max_ring_order
, XENBUS_MAX_RING_GRANT_ORDER
);
2682 xen_blkif_max_ring_order
= XENBUS_MAX_RING_GRANT_ORDER
;
2685 if (xen_blkif_max_queues
> nr_cpus
) {
2686 pr_info("Invalid max_queues (%d), will use default max: %d.\n",
2687 xen_blkif_max_queues
, nr_cpus
);
2688 xen_blkif_max_queues
= nr_cpus
;
2691 if (!xen_has_pv_disk_devices())
2694 if (register_blkdev(XENVBD_MAJOR
, DEV_NAME
)) {
2695 printk(KERN_WARNING
"xen_blk: can't get major %d with name %s\n",
2696 XENVBD_MAJOR
, DEV_NAME
);
2700 ret
= xenbus_register_frontend(&blkfront_driver
);
2702 unregister_blkdev(XENVBD_MAJOR
, DEV_NAME
);
2708 module_init(xlblk_init
);
2711 static void __exit
xlblk_exit(void)
2713 xenbus_unregister_driver(&blkfront_driver
);
2714 unregister_blkdev(XENVBD_MAJOR
, DEV_NAME
);
2717 module_exit(xlblk_exit
);
2719 MODULE_DESCRIPTION("Xen virtual block device frontend");
2720 MODULE_LICENSE("GPL");
2721 MODULE_ALIAS_BLOCKDEV_MAJOR(XENVBD_MAJOR
);
2722 MODULE_ALIAS("xen:vbd");
2723 MODULE_ALIAS("xenblk");