?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
nfsd: make export cache allocated per network namespace context
[linux-2.6/linux-acpi-2.6/ibm-acpi-2.6.git] / drivers / block / xen-blkfront.c
blob98cbeba8cd5358ca026b2e4b692a4aa55704c739
1 /*
2 * blkfront.c
3 *
4 * XenLinux virtual block device driver.
5 *
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
12 *
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:
18 *
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:
25 *
26 * The above copyright notice and this permission notice shall be included in
27 * all copies or substantial portions of the Software.
28 *
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
35 * IN THE SOFTWARE.
36 */
37
38 #include <linux/interrupt.h>
39 #include <linux/blkdev.h>
40 #include <linux/hdreg.h>
41 #include <linux/cdrom.h>
42 #include <linux/module.h>
43 #include <linux/slab.h>
44 #include <linux/mutex.h>
45 #include <linux/scatterlist.h>
46
47 #include <xen/xen.h>
48 #include <xen/xenbus.h>
49 #include <xen/grant_table.h>
50 #include <xen/events.h>
51 #include <xen/page.h>
52 #include <xen/platform_pci.h>
53
54 #include <xen/interface/grant_table.h>
55 #include <xen/interface/io/blkif.h>
56 #include <xen/interface/io/protocols.h>
57
58 #include <asm/xen/hypervisor.h>
59
60 enum blkif_state {
61 BLKIF_STATE_DISCONNECTED,
62 BLKIF_STATE_CONNECTED,
63 BLKIF_STATE_SUSPENDED,
64 };
65
66 struct blk_shadow {
67 struct blkif_request req;
68 struct request *request;
69 unsigned long frame[BLKIF_MAX_SEGMENTS_PER_REQUEST];
70 };
71
72 static DEFINE_MUTEX(blkfront_mutex);
73 static const struct block_device_operations xlvbd_block_fops;
74
75 #define BLK_RING_SIZE __CONST_RING_SIZE(blkif, PAGE_SIZE)
76
77 /*
78 * We have one of these per vbd, whether ide, scsi or 'other'. They
79 * hang in private_data off the gendisk structure. We may end up
80 * putting all kinds of interesting stuff here :-)
81 */
82 struct blkfront_info
83 {
84 struct mutex mutex;
85 struct xenbus_device *xbdev;
86 struct gendisk *gd;
87 int vdevice;
88 blkif_vdev_t handle;
89 enum blkif_state connected;
90 int ring_ref;
91 struct blkif_front_ring ring;
92 struct scatterlist sg[BLKIF_MAX_SEGMENTS_PER_REQUEST];
93 unsigned int evtchn, irq;
94 struct request_queue *rq;
95 struct work_struct work;
96 struct gnttab_free_callback callback;
97 struct blk_shadow shadow[BLK_RING_SIZE];
98 unsigned long shadow_free;
99 unsigned int feature_flush;
100 unsigned int flush_op;
101 unsigned int feature_discard:1;
102 unsigned int feature_secdiscard:1;
103 unsigned int discard_granularity;
104 unsigned int discard_alignment;
105 int is_ready;
106 };
107
108 static DEFINE_SPINLOCK(blkif_io_lock);
109
110 static unsigned int nr_minors;
111 static unsigned long *minors;
112 static DEFINE_SPINLOCK(minor_lock);
113
114 #define MAXIMUM_OUTSTANDING_BLOCK_REQS \
115 (BLKIF_MAX_SEGMENTS_PER_REQUEST * BLK_RING_SIZE)
116 #define GRANT_INVALID_REF 0
117
118 #define PARTS_PER_DISK 16
119 #define PARTS_PER_EXT_DISK 256
120
121 #define BLKIF_MAJOR(dev) ((dev)>>8)
122 #define BLKIF_MINOR(dev) ((dev) & 0xff)
123
124 #define EXT_SHIFT 28
125 #define EXTENDED (1<<EXT_SHIFT)
126 #define VDEV_IS_EXTENDED(dev) ((dev)&(EXTENDED))
127 #define BLKIF_MINOR_EXT(dev) ((dev)&(~EXTENDED))
128 #define EMULATED_HD_DISK_MINOR_OFFSET (0)
129 #define EMULATED_HD_DISK_NAME_OFFSET (EMULATED_HD_DISK_MINOR_OFFSET / 256)
130 #define EMULATED_SD_DISK_MINOR_OFFSET (0)
131 #define EMULATED_SD_DISK_NAME_OFFSET (EMULATED_SD_DISK_MINOR_OFFSET / 256)
132
133 #define DEV_NAME "xvd" /* name in /dev */
134
135 static int get_id_from_freelist(struct blkfront_info *info)
136 {
137 unsigned long free = info->shadow_free;
138 BUG_ON(free >= BLK_RING_SIZE);
139 info->shadow_free = info->shadow[free].req.u.rw.id;
140 info->shadow[free].req.u.rw.id = 0x0fffffee; /* debug */
141 return free;
142 }
143
144 static void add_id_to_freelist(struct blkfront_info *info,
145 unsigned long id)
146 {
147 info->shadow[id].req.u.rw.id = info->shadow_free;
148 info->shadow[id].request = NULL;
149 info->shadow_free = id;
150 }
151
152 static int xlbd_reserve_minors(unsigned int minor, unsigned int nr)
153 {
154 unsigned int end = minor + nr;
155 int rc;
156
157 if (end > nr_minors) {
158 unsigned long *bitmap, *old;
159
160 bitmap = kcalloc(BITS_TO_LONGS(end), sizeof(*bitmap),
161 GFP_KERNEL);
162 if (bitmap == NULL)
163 return -ENOMEM;
164
165 spin_lock(&minor_lock);
166 if (end > nr_minors) {
167 old = minors;
168 memcpy(bitmap, minors,
169 BITS_TO_LONGS(nr_minors) * sizeof(*bitmap));
170 minors = bitmap;
171 nr_minors = BITS_TO_LONGS(end) * BITS_PER_LONG;
172 } else
173 old = bitmap;
174 spin_unlock(&minor_lock);
175 kfree(old);
176 }
177
178 spin_lock(&minor_lock);
179 if (find_next_bit(minors, end, minor) >= end) {
180 for (; minor < end; ++minor)
181 __set_bit(minor, minors);
182 rc = 0;
183 } else
184 rc = -EBUSY;
185 spin_unlock(&minor_lock);
186
187 return rc;
188 }
189
190 static void xlbd_release_minors(unsigned int minor, unsigned int nr)
191 {
192 unsigned int end = minor + nr;
193
194 BUG_ON(end > nr_minors);
195 spin_lock(&minor_lock);
196 for (; minor < end; ++minor)
197 __clear_bit(minor, minors);
198 spin_unlock(&minor_lock);
199 }
200
201 static void blkif_restart_queue_callback(void *arg)
202 {
203 struct blkfront_info *info = (struct blkfront_info *)arg;
204 schedule_work(&info->work);
205 }
206
207 static int blkif_getgeo(struct block_device *bd, struct hd_geometry *hg)
208 {
209 /* We don't have real geometry info, but let's at least return
210 values consistent with the size of the device */
211 sector_t nsect = get_capacity(bd->bd_disk);
212 sector_t cylinders = nsect;
213
214 hg->heads = 0xff;
215 hg->sectors = 0x3f;
216 sector_div(cylinders, hg->heads * hg->sectors);
217 hg->cylinders = cylinders;
218 if ((sector_t)(hg->cylinders + 1) * hg->heads * hg->sectors < nsect)
219 hg->cylinders = 0xffff;
220 return 0;
221 }
222
223 static int blkif_ioctl(struct block_device *bdev, fmode_t mode,
224 unsigned command, unsigned long argument)
225 {
226 struct blkfront_info *info = bdev->bd_disk->private_data;
227 int i;
228
229 dev_dbg(&info->xbdev->dev, "command: 0x%x, argument: 0x%lx\n",
230 command, (long)argument);
231
232 switch (command) {
233 case CDROMMULTISESSION:
234 dev_dbg(&info->xbdev->dev, "FIXME: support multisession CDs later\n");
235 for (i = 0; i < sizeof(struct cdrom_multisession); i++)
236 if (put_user(0, (char __user *)(argument + i)))
237 return -EFAULT;
238 return 0;
239
240 case CDROM_GET_CAPABILITY: {
241 struct gendisk *gd = info->gd;
242 if (gd->flags & GENHD_FL_CD)
243 return 0;
244 return -EINVAL;
245 }
246
247 default:
248 /*printk(KERN_ALERT "ioctl %08x not supported by Xen blkdev\n",
249 command);*/
250 return -EINVAL; /* same return as native Linux */
251 }
252
253 return 0;
254 }
255
256 /*
257 * Generate a Xen blkfront IO request from a blk layer request. Reads
258 * and writes are handled as expected.
259 *
260 * @req: a request struct
261 */
262 static int blkif_queue_request(struct request *req)
263 {
264 struct blkfront_info *info = req->rq_disk->private_data;
265 unsigned long buffer_mfn;
266 struct blkif_request *ring_req;
267 unsigned long id;
268 unsigned int fsect, lsect;
269 int i, ref;
270 grant_ref_t gref_head;
271 struct scatterlist *sg;
272
273 if (unlikely(info->connected != BLKIF_STATE_CONNECTED))
274 return 1;
275
276 if (gnttab_alloc_grant_references(
277 BLKIF_MAX_SEGMENTS_PER_REQUEST, &gref_head) < 0) {
278 gnttab_request_free_callback(
279 &info->callback,
280 blkif_restart_queue_callback,
281 info,
282 BLKIF_MAX_SEGMENTS_PER_REQUEST);
283 return 1;
284 }
285
286 /* Fill out a communications ring structure. */
287 ring_req = RING_GET_REQUEST(&info->ring, info->ring.req_prod_pvt);
288 id = get_id_from_freelist(info);
289 info->shadow[id].request = req;
290
291 ring_req->u.rw.id = id;
292 ring_req->u.rw.sector_number = (blkif_sector_t)blk_rq_pos(req);
293 ring_req->u.rw.handle = info->handle;
294
295 ring_req->operation = rq_data_dir(req) ?
296 BLKIF_OP_WRITE : BLKIF_OP_READ;
297
298 if (req->cmd_flags & (REQ_FLUSH | REQ_FUA)) {
299 /*
300 * Ideally we can do an unordered flush-to-disk. In case the
301 * backend onlysupports barriers, use that. A barrier request
302 * a superset of FUA, so we can implement it the same
303 * way. (It's also a FLUSH+FUA, since it is
304 * guaranteed ordered WRT previous writes.)
305 */
306 ring_req->operation = info->flush_op;
307 }
308
309 if (unlikely(req->cmd_flags & (REQ_DISCARD | REQ_SECURE))) {
310 /* id, sector_number and handle are set above. */
311 ring_req->operation = BLKIF_OP_DISCARD;
312 ring_req->u.discard.nr_sectors = blk_rq_sectors(req);
313 if ((req->cmd_flags & REQ_SECURE) && info->feature_secdiscard)
314 ring_req->u.discard.flag = BLKIF_DISCARD_SECURE;
315 else
316 ring_req->u.discard.flag = 0;
317 } else {
318 ring_req->u.rw.nr_segments = blk_rq_map_sg(req->q, req,
319 info->sg);
320 BUG_ON(ring_req->u.rw.nr_segments >
321 BLKIF_MAX_SEGMENTS_PER_REQUEST);
322
323 for_each_sg(info->sg, sg, ring_req->u.rw.nr_segments, i) {
324 buffer_mfn = pfn_to_mfn(page_to_pfn(sg_page(sg)));
325 fsect = sg->offset >> 9;
326 lsect = fsect + (sg->length >> 9) - 1;
327 /* install a grant reference. */
328 ref = gnttab_claim_grant_reference(&gref_head);
329 BUG_ON(ref == -ENOSPC);
330
331 gnttab_grant_foreign_access_ref(
332 ref,
333 info->xbdev->otherend_id,
334 buffer_mfn,
335 rq_data_dir(req));
336
337 info->shadow[id].frame[i] = mfn_to_pfn(buffer_mfn);
338 ring_req->u.rw.seg[i] =
339 (struct blkif_request_segment) {
340 .gref = ref,
341 .first_sect = fsect,
342 .last_sect = lsect };
343 }
344 }
345
346 info->ring.req_prod_pvt++;
347
348 /* Keep a private copy so we can reissue requests when recovering. */
349 info->shadow[id].req = *ring_req;
350
351 gnttab_free_grant_references(gref_head);
352
353 return 0;
354 }
355
356
357 static inline void flush_requests(struct blkfront_info *info)
358 {
359 int notify;
360
361 RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&info->ring, notify);
362
363 if (notify)
364 notify_remote_via_irq(info->irq);
365 }
366
367 /*
368 * do_blkif_request
369 * read a block; request is in a request queue
370 */
371 static void do_blkif_request(struct request_queue *rq)
372 {
373 struct blkfront_info *info = NULL;
374 struct request *req;
375 int queued;
376
377 pr_debug("Entered do_blkif_request\n");
378
379 queued = 0;
380
381 while ((req = blk_peek_request(rq)) != NULL) {
382 info = req->rq_disk->private_data;
383
384 if (RING_FULL(&info->ring))
385 goto wait;
386
387 blk_start_request(req);
388
389 if ((req->cmd_type != REQ_TYPE_FS) ||
390 ((req->cmd_flags & (REQ_FLUSH | REQ_FUA)) &&
391 !info->flush_op)) {
392 __blk_end_request_all(req, -EIO);
393 continue;
394 }
395
396 pr_debug("do_blk_req %p: cmd %p, sec %lx, "
397 "(%u/%u) buffer:%p [%s]\n",
398 req, req->cmd, (unsigned long)blk_rq_pos(req),
399 blk_rq_cur_sectors(req), blk_rq_sectors(req),
400 req->buffer, rq_data_dir(req) ? "write" : "read");
401
402 if (blkif_queue_request(req)) {
403 blk_requeue_request(rq, req);
404 wait:
405 /* Avoid pointless unplugs. */
406 blk_stop_queue(rq);
407 break;
408 }
409
410 queued++;
411 }
412
413 if (queued != 0)
414 flush_requests(info);
415 }
416
417 static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size)
418 {
419 struct request_queue *rq;
420 struct blkfront_info *info = gd->private_data;
421
422 rq = blk_init_queue(do_blkif_request, &blkif_io_lock);
423 if (rq == NULL)
424 return -1;
425
426 queue_flag_set_unlocked(QUEUE_FLAG_VIRT, rq);
427
428 if (info->feature_discard) {
429 queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, rq);
430 blk_queue_max_discard_sectors(rq, get_capacity(gd));
431 rq->limits.discard_granularity = info->discard_granularity;
432 rq->limits.discard_alignment = info->discard_alignment;
433 if (info->feature_secdiscard)
434 queue_flag_set_unlocked(QUEUE_FLAG_SECDISCARD, rq);
435 }
436
437 /* Hard sector size and max sectors impersonate the equiv. hardware. */
438 blk_queue_logical_block_size(rq, sector_size);
439 blk_queue_max_hw_sectors(rq, 512);
440
441 /* Each segment in a request is up to an aligned page in size. */
442 blk_queue_segment_boundary(rq, PAGE_SIZE - 1);
443 blk_queue_max_segment_size(rq, PAGE_SIZE);
444
445 /* Ensure a merged request will fit in a single I/O ring slot. */
446 blk_queue_max_segments(rq, BLKIF_MAX_SEGMENTS_PER_REQUEST);
447
448 /* Make sure buffer addresses are sector-aligned. */
449 blk_queue_dma_alignment(rq, 511);
450
451 /* Make sure we don't use bounce buffers. */
452 blk_queue_bounce_limit(rq, BLK_BOUNCE_ANY);
453
454 gd->queue = rq;
455
456 return 0;
457 }
458
459
460 static void xlvbd_flush(struct blkfront_info *info)
461 {
462 blk_queue_flush(info->rq, info->feature_flush);
463 printk(KERN_INFO "blkfront: %s: %s: %s\n",
464 info->gd->disk_name,
465 info->flush_op == BLKIF_OP_WRITE_BARRIER ?
466 "barrier" : (info->flush_op == BLKIF_OP_FLUSH_DISKCACHE ?
467 "flush diskcache" : "barrier or flush"),
468 info->feature_flush ? "enabled" : "disabled");
469 }
470
471 static int xen_translate_vdev(int vdevice, int *minor, unsigned int *offset)
472 {
473 int major;
474 major = BLKIF_MAJOR(vdevice);
475 *minor = BLKIF_MINOR(vdevice);
476 switch (major) {
477 case XEN_IDE0_MAJOR:
478 *offset = (*minor / 64) + EMULATED_HD_DISK_NAME_OFFSET;
479 *minor = ((*minor / 64) * PARTS_PER_DISK) +
480 EMULATED_HD_DISK_MINOR_OFFSET;
481 break;
482 case XEN_IDE1_MAJOR:
483 *offset = (*minor / 64) + 2 + EMULATED_HD_DISK_NAME_OFFSET;
484 *minor = (((*minor / 64) + 2) * PARTS_PER_DISK) +
485 EMULATED_HD_DISK_MINOR_OFFSET;
486 break;
487 case XEN_SCSI_DISK0_MAJOR:
488 *offset = (*minor / PARTS_PER_DISK) + EMULATED_SD_DISK_NAME_OFFSET;
489 *minor = *minor + EMULATED_SD_DISK_MINOR_OFFSET;
490 break;
491 case XEN_SCSI_DISK1_MAJOR:
492 case XEN_SCSI_DISK2_MAJOR:
493 case XEN_SCSI_DISK3_MAJOR:
494 case XEN_SCSI_DISK4_MAJOR:
495 case XEN_SCSI_DISK5_MAJOR:
496 case XEN_SCSI_DISK6_MAJOR:
497 case XEN_SCSI_DISK7_MAJOR:
498 *offset = (*minor / PARTS_PER_DISK) +
499 ((major - XEN_SCSI_DISK1_MAJOR + 1) * 16) +
500 EMULATED_SD_DISK_NAME_OFFSET;
501 *minor = *minor +
502 ((major - XEN_SCSI_DISK1_MAJOR + 1) * 16 * PARTS_PER_DISK) +
503 EMULATED_SD_DISK_MINOR_OFFSET;
504 break;
505 case XEN_SCSI_DISK8_MAJOR:
506 case XEN_SCSI_DISK9_MAJOR:
507 case XEN_SCSI_DISK10_MAJOR:
508 case XEN_SCSI_DISK11_MAJOR:
509 case XEN_SCSI_DISK12_MAJOR:
510 case XEN_SCSI_DISK13_MAJOR:
511 case XEN_SCSI_DISK14_MAJOR:
512 case XEN_SCSI_DISK15_MAJOR:
513 *offset = (*minor / PARTS_PER_DISK) +
514 ((major - XEN_SCSI_DISK8_MAJOR + 8) * 16) +
515 EMULATED_SD_DISK_NAME_OFFSET;
516 *minor = *minor +
517 ((major - XEN_SCSI_DISK8_MAJOR + 8) * 16 * PARTS_PER_DISK) +
518 EMULATED_SD_DISK_MINOR_OFFSET;
519 break;
520 case XENVBD_MAJOR:
521 *offset = *minor / PARTS_PER_DISK;
522 break;
523 default:
524 printk(KERN_WARNING "blkfront: your disk configuration is "
525 "incorrect, please use an xvd device instead\n");
526 return -ENODEV;
527 }
528 return 0;
529 }
530
531 static int xlvbd_alloc_gendisk(blkif_sector_t capacity,
532 struct blkfront_info *info,
533 u16 vdisk_info, u16 sector_size)
534 {
535 struct gendisk *gd;
536 int nr_minors = 1;
537 int err;
538 unsigned int offset;
539 int minor;
540 int nr_parts;
541
542 BUG_ON(info->gd != NULL);
543 BUG_ON(info->rq != NULL);
544
545 if ((info->vdevice>>EXT_SHIFT) > 1) {
546 /* this is above the extended range; something is wrong */
547 printk(KERN_WARNING "blkfront: vdevice 0x%x is above the extended range; ignoring\n", info->vdevice);
548 return -ENODEV;
549 }
550
551 if (!VDEV_IS_EXTENDED(info->vdevice)) {
552 err = xen_translate_vdev(info->vdevice, &minor, &offset);
553 if (err)
554 return err;
555 nr_parts = PARTS_PER_DISK;
556 } else {
557 minor = BLKIF_MINOR_EXT(info->vdevice);
558 nr_parts = PARTS_PER_EXT_DISK;
559 offset = minor / nr_parts;
560 if (xen_hvm_domain() && offset < EMULATED_HD_DISK_NAME_OFFSET + 4)
561 printk(KERN_WARNING "blkfront: vdevice 0x%x might conflict with "
562 "emulated IDE disks,\n\t choose an xvd device name"
563 "from xvde on\n", info->vdevice);
564 }
565 err = -ENODEV;
566
567 if ((minor % nr_parts) == 0)
568 nr_minors = nr_parts;
569
570 err = xlbd_reserve_minors(minor, nr_minors);
571 if (err)
572 goto out;
573 err = -ENODEV;
574
575 gd = alloc_disk(nr_minors);
576 if (gd == NULL)
577 goto release;
578
579 if (nr_minors > 1) {
580 if (offset < 26)
581 sprintf(gd->disk_name, "%s%c", DEV_NAME, 'a' + offset);
582 else
583 sprintf(gd->disk_name, "%s%c%c", DEV_NAME,
584 'a' + ((offset / 26)-1), 'a' + (offset % 26));
585 } else {
586 if (offset < 26)
587 sprintf(gd->disk_name, "%s%c%d", DEV_NAME,
588 'a' + offset,
589 minor & (nr_parts - 1));
590 else
591 sprintf(gd->disk_name, "%s%c%c%d", DEV_NAME,
592 'a' + ((offset / 26) - 1),
593 'a' + (offset % 26),
594 minor & (nr_parts - 1));
595 }
596
597 gd->major = XENVBD_MAJOR;
598 gd->first_minor = minor;
599 gd->fops = &xlvbd_block_fops;
600 gd->private_data = info;
601 gd->driverfs_dev = &(info->xbdev->dev);
602 set_capacity(gd, capacity);
603
604 if (xlvbd_init_blk_queue(gd, sector_size)) {
605 del_gendisk(gd);
606 goto release;
607 }
608
609 info->rq = gd->queue;
610 info->gd = gd;
611
612 xlvbd_flush(info);
613
614 if (vdisk_info & VDISK_READONLY)
615 set_disk_ro(gd, 1);
616
617 if (vdisk_info & VDISK_REMOVABLE)
618 gd->flags |= GENHD_FL_REMOVABLE;
619
620 if (vdisk_info & VDISK_CDROM)
621 gd->flags |= GENHD_FL_CD;
622
623 return 0;
624
625 release:
626 xlbd_release_minors(minor, nr_minors);
627 out:
628 return err;
629 }
630
631 static void xlvbd_release_gendisk(struct blkfront_info *info)
632 {
633 unsigned int minor, nr_minors;
634 unsigned long flags;
635
636 if (info->rq == NULL)
637 return;
638
639 spin_lock_irqsave(&blkif_io_lock, flags);
640
641 /* No more blkif_request(). */
642 blk_stop_queue(info->rq);
643
644 /* No more gnttab callback work. */
645 gnttab_cancel_free_callback(&info->callback);
646 spin_unlock_irqrestore(&blkif_io_lock, flags);
647
648 /* Flush gnttab callback work. Must be done with no locks held. */
649 flush_work_sync(&info->work);
650
651 del_gendisk(info->gd);
652
653 minor = info->gd->first_minor;
654 nr_minors = info->gd->minors;
655 xlbd_release_minors(minor, nr_minors);
656
657 blk_cleanup_queue(info->rq);
658 info->rq = NULL;
659
660 put_disk(info->gd);
661 info->gd = NULL;
662 }
663
664 static void kick_pending_request_queues(struct blkfront_info *info)
665 {
666 if (!RING_FULL(&info->ring)) {
667 /* Re-enable calldowns. */
668 blk_start_queue(info->rq);
669 /* Kick things off immediately. */
670 do_blkif_request(info->rq);
671 }
672 }
673
674 static void blkif_restart_queue(struct work_struct *work)
675 {
676 struct blkfront_info *info = container_of(work, struct blkfront_info, work);
677
678 spin_lock_irq(&blkif_io_lock);
679 if (info->connected == BLKIF_STATE_CONNECTED)
680 kick_pending_request_queues(info);
681 spin_unlock_irq(&blkif_io_lock);
682 }
683
684 static void blkif_free(struct blkfront_info *info, int suspend)
685 {
686 /* Prevent new requests being issued until we fix things up. */
687 spin_lock_irq(&blkif_io_lock);
688 info->connected = suspend ?
689 BLKIF_STATE_SUSPENDED : BLKIF_STATE_DISCONNECTED;
690 /* No more blkif_request(). */
691 if (info->rq)
692 blk_stop_queue(info->rq);
693 /* No more gnttab callback work. */
694 gnttab_cancel_free_callback(&info->callback);
695 spin_unlock_irq(&blkif_io_lock);
696
697 /* Flush gnttab callback work. Must be done with no locks held. */
698 flush_work_sync(&info->work);
699
700 /* Free resources associated with old device channel. */
701 if (info->ring_ref != GRANT_INVALID_REF) {
702 gnttab_end_foreign_access(info->ring_ref, 0,
703 (unsigned long)info->ring.sring);
704 info->ring_ref = GRANT_INVALID_REF;
705 info->ring.sring = NULL;
706 }
707 if (info->irq)
708 unbind_from_irqhandler(info->irq, info);
709 info->evtchn = info->irq = 0;
710
711 }
712
713 static void blkif_completion(struct blk_shadow *s)
714 {
715 int i;
716 /* Do not let BLKIF_OP_DISCARD as nr_segment is in the same place
717 * flag. */
718 for (i = 0; i < s->req.u.rw.nr_segments; i++)
719 gnttab_end_foreign_access(s->req.u.rw.seg[i].gref, 0, 0UL);
720 }
721
722 static irqreturn_t blkif_interrupt(int irq, void *dev_id)
723 {
724 struct request *req;
725 struct blkif_response *bret;
726 RING_IDX i, rp;
727 unsigned long flags;
728 struct blkfront_info *info = (struct blkfront_info *)dev_id;
729 int error;
730
731 spin_lock_irqsave(&blkif_io_lock, flags);
732
733 if (unlikely(info->connected != BLKIF_STATE_CONNECTED)) {
734 spin_unlock_irqrestore(&blkif_io_lock, flags);
735 return IRQ_HANDLED;
736 }
737
738 again:
739 rp = info->ring.sring->rsp_prod;
740 rmb(); /* Ensure we see queued responses up to 'rp'. */
741
742 for (i = info->ring.rsp_cons; i != rp; i++) {
743 unsigned long id;
744
745 bret = RING_GET_RESPONSE(&info->ring, i);
746 id = bret->id;
747 req = info->shadow[id].request;
748
749 if (bret->operation != BLKIF_OP_DISCARD)
750 blkif_completion(&info->shadow[id]);
751
752 add_id_to_freelist(info, id);
753
754 error = (bret->status == BLKIF_RSP_OKAY) ? 0 : -EIO;
755 switch (bret->operation) {
756 case BLKIF_OP_DISCARD:
757 if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) {
758 struct request_queue *rq = info->rq;
759 printk(KERN_WARNING "blkfront: %s: discard op failed\n",
760 info->gd->disk_name);
761 error = -EOPNOTSUPP;
762 info->feature_discard = 0;
763 info->feature_secdiscard = 0;
764 queue_flag_clear(QUEUE_FLAG_DISCARD, rq);
765 queue_flag_clear(QUEUE_FLAG_SECDISCARD, rq);
766 }
767 __blk_end_request_all(req, error);
768 break;
769 case BLKIF_OP_FLUSH_DISKCACHE:
770 case BLKIF_OP_WRITE_BARRIER:
771 if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) {
772 printk(KERN_WARNING "blkfront: %s: write %s op failed\n",
773 info->flush_op == BLKIF_OP_WRITE_BARRIER ?
774 "barrier" : "flush disk cache",
775 info->gd->disk_name);
776 error = -EOPNOTSUPP;
777 }
778 if (unlikely(bret->status == BLKIF_RSP_ERROR &&
779 info->shadow[id].req.u.rw.nr_segments == 0)) {
780 printk(KERN_WARNING "blkfront: %s: empty write %s op failed\n",
781 info->flush_op == BLKIF_OP_WRITE_BARRIER ?
782 "barrier" : "flush disk cache",
783 info->gd->disk_name);
784 error = -EOPNOTSUPP;
785 }
786 if (unlikely(error)) {
787 if (error == -EOPNOTSUPP)
788 error = 0;
789 info->feature_flush = 0;
790 info->flush_op = 0;
791 xlvbd_flush(info);
792 }
793 /* fall through */
794 case BLKIF_OP_READ:
795 case BLKIF_OP_WRITE:
796 if (unlikely(bret->status != BLKIF_RSP_OKAY))
797 dev_dbg(&info->xbdev->dev, "Bad return from blkdev data "
798 "request: %x\n", bret->status);
799
800 __blk_end_request_all(req, error);
801 break;
802 default:
803 BUG();
804 }
805 }
806
807 info->ring.rsp_cons = i;
808
809 if (i != info->ring.req_prod_pvt) {
810 int more_to_do;
811 RING_FINAL_CHECK_FOR_RESPONSES(&info->ring, more_to_do);
812 if (more_to_do)
813 goto again;
814 } else
815 info->ring.sring->rsp_event = i + 1;
816
817 kick_pending_request_queues(info);
818
819 spin_unlock_irqrestore(&blkif_io_lock, flags);
820
821 return IRQ_HANDLED;
822 }
823
824
825 static int setup_blkring(struct xenbus_device *dev,
826 struct blkfront_info *info)
827 {
828 struct blkif_sring *sring;
829 int err;
830
831 info->ring_ref = GRANT_INVALID_REF;
832
833 sring = (struct blkif_sring *)__get_free_page(GFP_NOIO | __GFP_HIGH);
834 if (!sring) {
835 xenbus_dev_fatal(dev, -ENOMEM, "allocating shared ring");
836 return -ENOMEM;
837 }
838 SHARED_RING_INIT(sring);
839 FRONT_RING_INIT(&info->ring, sring, PAGE_SIZE);
840
841 sg_init_table(info->sg, BLKIF_MAX_SEGMENTS_PER_REQUEST);
842
843 err = xenbus_grant_ring(dev, virt_to_mfn(info->ring.sring));
844 if (err < 0) {
845 free_page((unsigned long)sring);
846 info->ring.sring = NULL;
847 goto fail;
848 }
849 info->ring_ref = err;
850
851 err = xenbus_alloc_evtchn(dev, &info->evtchn);
852 if (err)
853 goto fail;
854
855 err = bind_evtchn_to_irqhandler(info->evtchn,
856 blkif_interrupt,
857 IRQF_SAMPLE_RANDOM, "blkif", info);
858 if (err <= 0) {
859 xenbus_dev_fatal(dev, err,
860 "bind_evtchn_to_irqhandler failed");
861 goto fail;
862 }
863 info->irq = err;
864
865 return 0;
866 fail:
867 blkif_free(info, 0);
868 return err;
869 }
870
871
872 /* Common code used when first setting up, and when resuming. */
873 static int talk_to_blkback(struct xenbus_device *dev,
874 struct blkfront_info *info)
875 {
876 const char *message = NULL;
877 struct xenbus_transaction xbt;
878 int err;
879
880 /* Create shared ring, alloc event channel. */
881 err = setup_blkring(dev, info);
882 if (err)
883 goto out;
884
885 again:
886 err = xenbus_transaction_start(&xbt);
887 if (err) {
888 xenbus_dev_fatal(dev, err, "starting transaction");
889 goto destroy_blkring;
890 }
891
892 err = xenbus_printf(xbt, dev->nodename,
893 "ring-ref", "%u", info->ring_ref);
894 if (err) {
895 message = "writing ring-ref";
896 goto abort_transaction;
897 }
898 err = xenbus_printf(xbt, dev->nodename,
899 "event-channel", "%u", info->evtchn);
900 if (err) {
901 message = "writing event-channel";
902 goto abort_transaction;
903 }
904 err = xenbus_printf(xbt, dev->nodename, "protocol", "%s",
905 XEN_IO_PROTO_ABI_NATIVE);
906 if (err) {
907 message = "writing protocol";
908 goto abort_transaction;
909 }
910
911 err = xenbus_transaction_end(xbt, 0);
912 if (err) {
913 if (err == -EAGAIN)
914 goto again;
915 xenbus_dev_fatal(dev, err, "completing transaction");
916 goto destroy_blkring;
917 }
918
919 xenbus_switch_state(dev, XenbusStateInitialised);
920
921 return 0;
922
923 abort_transaction:
924 xenbus_transaction_end(xbt, 1);
925 if (message)
926 xenbus_dev_fatal(dev, err, "%s", message);
927 destroy_blkring:
928 blkif_free(info, 0);
929 out:
930 return err;
931 }
932
933 /**
934 * Entry point to this code when a new device is created. Allocate the basic
935 * structures and the ring buffer for communication with the backend, and
936 * inform the backend of the appropriate details for those. Switch to
937 * Initialised state.
938 */
939 static int blkfront_probe(struct xenbus_device *dev,
940 const struct xenbus_device_id *id)
941 {
942 int err, vdevice, i;
943 struct blkfront_info *info;
944
945 /* FIXME: Use dynamic device id if this is not set. */
946 err = xenbus_scanf(XBT_NIL, dev->nodename,
947 "virtual-device", "%i", &vdevice);
948 if (err != 1) {
949 /* go looking in the extended area instead */
950 err = xenbus_scanf(XBT_NIL, dev->nodename, "virtual-device-ext",
951 "%i", &vdevice);
952 if (err != 1) {
953 xenbus_dev_fatal(dev, err, "reading virtual-device");
954 return err;
955 }
956 }
957
958 if (xen_hvm_domain()) {
959 char *type;
960 int len;
961 /* no unplug has been done: do not hook devices != xen vbds */
962 if (xen_platform_pci_unplug & XEN_UNPLUG_UNNECESSARY) {
963 int major;
964
965 if (!VDEV_IS_EXTENDED(vdevice))
966 major = BLKIF_MAJOR(vdevice);
967 else
968 major = XENVBD_MAJOR;
969
970 if (major != XENVBD_MAJOR) {
971 printk(KERN_INFO
972 "%s: HVM does not support vbd %d as xen block device\n",
973 __FUNCTION__, vdevice);
974 return -ENODEV;
975 }
976 }
977 /* do not create a PV cdrom device if we are an HVM guest */
978 type = xenbus_read(XBT_NIL, dev->nodename, "device-type", &len);
979 if (IS_ERR(type))
980 return -ENODEV;
981 if (strncmp(type, "cdrom", 5) == 0) {
982 kfree(type);
983 return -ENODEV;
984 }
985 kfree(type);
986 }
987 info = kzalloc(sizeof(*info), GFP_KERNEL);
988 if (!info) {
989 xenbus_dev_fatal(dev, -ENOMEM, "allocating info structure");
990 return -ENOMEM;
991 }
992
993 mutex_init(&info->mutex);
994 info->xbdev = dev;
995 info->vdevice = vdevice;
996 info->connected = BLKIF_STATE_DISCONNECTED;
997 INIT_WORK(&info->work, blkif_restart_queue);
998
999 for (i = 0; i < BLK_RING_SIZE; i++)
1000 info->shadow[i].req.u.rw.id = i+1;
1001 info->shadow[BLK_RING_SIZE-1].req.u.rw.id = 0x0fffffff;
1002
1003 /* Front end dir is a number, which is used as the id. */
1004 info->handle = simple_strtoul(strrchr(dev->nodename, '/')+1, NULL, 0);
1005 dev_set_drvdata(&dev->dev, info);
1006
1007 err = talk_to_blkback(dev, info);
1008 if (err) {
1009 kfree(info);
1010 dev_set_drvdata(&dev->dev, NULL);
1011 return err;
1012 }
1013
1014 return 0;
1015 }
1016
1017
1018 static int blkif_recover(struct blkfront_info *info)
1019 {
1020 int i;
1021 struct blkif_request *req;
1022 struct blk_shadow *copy;
1023 int j;
1024
1025 /* Stage 1: Make a safe copy of the shadow state. */
1026 copy = kmalloc(sizeof(info->shadow),
1027 GFP_NOIO | __GFP_REPEAT | __GFP_HIGH);
1028 if (!copy)
1029 return -ENOMEM;
1030 memcpy(copy, info->shadow, sizeof(info->shadow));
1031
1032 /* Stage 2: Set up free list. */
1033 memset(&info->shadow, 0, sizeof(info->shadow));
1034 for (i = 0; i < BLK_RING_SIZE; i++)
1035 info->shadow[i].req.u.rw.id = i+1;
1036 info->shadow_free = info->ring.req_prod_pvt;
1037 info->shadow[BLK_RING_SIZE-1].req.u.rw.id = 0x0fffffff;
1038
1039 /* Stage 3: Find pending requests and requeue them. */
1040 for (i = 0; i < BLK_RING_SIZE; i++) {
1041 /* Not in use? */
1042 if (!copy[i].request)
1043 continue;
1044
1045 /* Grab a request slot and copy shadow state into it. */
1046 req = RING_GET_REQUEST(&info->ring, info->ring.req_prod_pvt);
1047 *req = copy[i].req;
1048
1049 /* We get a new request id, and must reset the shadow state. */
1050 req->u.rw.id = get_id_from_freelist(info);
1051 memcpy(&info->shadow[req->u.rw.id], &copy[i], sizeof(copy[i]));
1052
1053 if (req->operation != BLKIF_OP_DISCARD) {
1054 /* Rewrite any grant references invalidated by susp/resume. */
1055 for (j = 0; j < req->u.rw.nr_segments; j++)
1056 gnttab_grant_foreign_access_ref(
1057 req->u.rw.seg[j].gref,
1058 info->xbdev->otherend_id,
1059 pfn_to_mfn(info->shadow[req->u.rw.id].frame[j]),
1060 rq_data_dir(info->shadow[req->u.rw.id].request));
1061 }
1062 info->shadow[req->u.rw.id].req = *req;
1063
1064 info->ring.req_prod_pvt++;
1065 }
1066
1067 kfree(copy);
1068
1069 xenbus_switch_state(info->xbdev, XenbusStateConnected);
1070
1071 spin_lock_irq(&blkif_io_lock);
1072
1073 /* Now safe for us to use the shared ring */
1074 info->connected = BLKIF_STATE_CONNECTED;
1075
1076 /* Send off requeued requests */
1077 flush_requests(info);
1078
1079 /* Kick any other new requests queued since we resumed */
1080 kick_pending_request_queues(info);
1081
1082 spin_unlock_irq(&blkif_io_lock);
1083
1084 return 0;
1085 }
1086
1087 /**
1088 * We are reconnecting to the backend, due to a suspend/resume, or a backend
1089 * driver restart. We tear down our blkif structure and recreate it, but
1090 * leave the device-layer structures intact so that this is transparent to the
1091 * rest of the kernel.
1092 */
1093 static int blkfront_resume(struct xenbus_device *dev)
1094 {
1095 struct blkfront_info *info = dev_get_drvdata(&dev->dev);
1096 int err;
1097
1098 dev_dbg(&dev->dev, "blkfront_resume: %s\n", dev->nodename);
1099
1100 blkif_free(info, info->connected == BLKIF_STATE_CONNECTED);
1101
1102 err = talk_to_blkback(dev, info);
1103 if (info->connected == BLKIF_STATE_SUSPENDED && !err)
1104 err = blkif_recover(info);
1105
1106 return err;
1107 }
1108
1109 static void
1110 blkfront_closing(struct blkfront_info *info)
1111 {
1112 struct xenbus_device *xbdev = info->xbdev;
1113 struct block_device *bdev = NULL;
1114
1115 mutex_lock(&info->mutex);
1116
1117 if (xbdev->state == XenbusStateClosing) {
1118 mutex_unlock(&info->mutex);
1119 return;
1120 }
1121
1122 if (info->gd)
1123 bdev = bdget_disk(info->gd, 0);
1124
1125 mutex_unlock(&info->mutex);
1126
1127 if (!bdev) {
1128 xenbus_frontend_closed(xbdev);
1129 return;
1130 }
1131
1132 mutex_lock(&bdev->bd_mutex);
1133
1134 if (bdev->bd_openers) {
1135 xenbus_dev_error(xbdev, -EBUSY,
1136 "Device in use; refusing to close");
1137 xenbus_switch_state(xbdev, XenbusStateClosing);
1138 } else {
1139 xlvbd_release_gendisk(info);
1140 xenbus_frontend_closed(xbdev);
1141 }
1142
1143 mutex_unlock(&bdev->bd_mutex);
1144 bdput(bdev);
1145 }
1146
1147 static void blkfront_setup_discard(struct blkfront_info *info)
1148 {
1149 int err;
1150 char *type;
1151 unsigned int discard_granularity;
1152 unsigned int discard_alignment;
1153 unsigned int discard_secure;
1154
1155 type = xenbus_read(XBT_NIL, info->xbdev->otherend, "type", NULL);
1156 if (IS_ERR(type))
1157 return;
1158
1159 info->feature_secdiscard = 0;
1160 if (strncmp(type, "phy", 3) == 0) {
1161 err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
1162 "discard-granularity", "%u", &discard_granularity,
1163 "discard-alignment", "%u", &discard_alignment,
1164 NULL);
1165 if (!err) {
1166 info->feature_discard = 1;
1167 info->discard_granularity = discard_granularity;
1168 info->discard_alignment = discard_alignment;
1169 }
1170 err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
1171 "discard-secure", "%d", &discard_secure,
1172 NULL);
1173 if (!err)
1174 info->feature_secdiscard = discard_secure;
1175
1176 } else if (strncmp(type, "file", 4) == 0)
1177 info->feature_discard = 1;
1178
1179 kfree(type);
1180 }
1181
1182 /*
1183 * Invoked when the backend is finally 'ready' (and has told produced
1184 * the details about the physical device - #sectors, size, etc).
1185 */
1186 static void blkfront_connect(struct blkfront_info *info)
1187 {
1188 unsigned long long sectors;
1189 unsigned long sector_size;
1190 unsigned int binfo;
1191 int err;
1192 int barrier, flush, discard;
1193
1194 switch (info->connected) {
1195 case BLKIF_STATE_CONNECTED:
1196 /*
1197 * Potentially, the back-end may be signalling
1198 * a capacity change; update the capacity.
1199 */
1200 err = xenbus_scanf(XBT_NIL, info->xbdev->otherend,
1201 "sectors", "%Lu", &sectors);
1202 if (XENBUS_EXIST_ERR(err))
1203 return;
1204 printk(KERN_INFO "Setting capacity to %Lu\n",
1205 sectors);
1206 set_capacity(info->gd, sectors);
1207 revalidate_disk(info->gd);
1208
1209 /* fall through */
1210 case BLKIF_STATE_SUSPENDED:
1211 return;
1212
1213 default:
1214 break;
1215 }
1216
1217 dev_dbg(&info->xbdev->dev, "%s:%s.\n",
1218 __func__, info->xbdev->otherend);
1219
1220 err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
1221 "sectors", "%llu", &sectors,
1222 "info", "%u", &binfo,
1223 "sector-size", "%lu", &sector_size,
1224 NULL);
1225 if (err) {
1226 xenbus_dev_fatal(info->xbdev, err,
1227 "reading backend fields at %s",
1228 info->xbdev->otherend);
1229 return;
1230 }
1231
1232 info->feature_flush = 0;
1233 info->flush_op = 0;
1234
1235 err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
1236 "feature-barrier", "%d", &barrier,
1237 NULL);
1238
1239 /*
1240 * If there's no "feature-barrier" defined, then it means
1241 * we're dealing with a very old backend which writes
1242 * synchronously; nothing to do.
1243 *
1244 * If there are barriers, then we use flush.
1245 */
1246 if (!err && barrier) {
1247 info->feature_flush = REQ_FLUSH | REQ_FUA;
1248 info->flush_op = BLKIF_OP_WRITE_BARRIER;
1249 }
1250 /*
1251 * And if there is "feature-flush-cache" use that above
1252 * barriers.
1253 */
1254 err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
1255 "feature-flush-cache", "%d", &flush,
1256 NULL);
1257
1258 if (!err && flush) {
1259 info->feature_flush = REQ_FLUSH;
1260 info->flush_op = BLKIF_OP_FLUSH_DISKCACHE;
1261 }
1262
1263 err = xenbus_gather(XBT_NIL, info->xbdev->otherend,
1264 "feature-discard", "%d", &discard,
1265 NULL);
1266
1267 if (!err && discard)
1268 blkfront_setup_discard(info);
1269
1270 err = xlvbd_alloc_gendisk(sectors, info, binfo, sector_size);
1271 if (err) {
1272 xenbus_dev_fatal(info->xbdev, err, "xlvbd_add at %s",
1273 info->xbdev->otherend);
1274 return;
1275 }
1276
1277 xenbus_switch_state(info->xbdev, XenbusStateConnected);
1278
1279 /* Kick pending requests. */
1280 spin_lock_irq(&blkif_io_lock);
1281 info->connected = BLKIF_STATE_CONNECTED;
1282 kick_pending_request_queues(info);
1283 spin_unlock_irq(&blkif_io_lock);
1284
1285 add_disk(info->gd);
1286
1287 info->is_ready = 1;
1288 }
1289
1290 /**
1291 * Callback received when the backend's state changes.
1292 */
1293 static void blkback_changed(struct xenbus_device *dev,
1294 enum xenbus_state backend_state)
1295 {
1296 struct blkfront_info *info = dev_get_drvdata(&dev->dev);
1297
1298 dev_dbg(&dev->dev, "blkfront:blkback_changed to state %d.\n", backend_state);
1299
1300 switch (backend_state) {
1301 case XenbusStateInitialising:
1302 case XenbusStateInitWait:
1303 case XenbusStateInitialised:
1304 case XenbusStateReconfiguring:
1305 case XenbusStateReconfigured:
1306 case XenbusStateUnknown:
1307 case XenbusStateClosed:
1308 break;
1309
1310 case XenbusStateConnected:
1311 blkfront_connect(info);
1312 break;
1313
1314 case XenbusStateClosing:
1315 blkfront_closing(info);
1316 break;
1317 }
1318 }
1319
1320 static int blkfront_remove(struct xenbus_device *xbdev)
1321 {
1322 struct blkfront_info *info = dev_get_drvdata(&xbdev->dev);
1323 struct block_device *bdev = NULL;
1324 struct gendisk *disk;
1325
1326 dev_dbg(&xbdev->dev, "%s removed", xbdev->nodename);
1327
1328 blkif_free(info, 0);
1329
1330 mutex_lock(&info->mutex);
1331
1332 disk = info->gd;
1333 if (disk)
1334 bdev = bdget_disk(disk, 0);
1335
1336 info->xbdev = NULL;
1337 mutex_unlock(&info->mutex);
1338
1339 if (!bdev) {
1340 kfree(info);
1341 return 0;
1342 }
1343
1344 /*
1345 * The xbdev was removed before we reached the Closed
1346 * state. See if it's safe to remove the disk. If the bdev
1347 * isn't closed yet, we let release take care of it.
1348 */
1349
1350 mutex_lock(&bdev->bd_mutex);
1351 info = disk->private_data;
1352
1353 dev_warn(disk_to_dev(disk),
1354 "%s was hot-unplugged, %d stale handles\n",
1355 xbdev->nodename, bdev->bd_openers);
1356
1357 if (info && !bdev->bd_openers) {
1358 xlvbd_release_gendisk(info);
1359 disk->private_data = NULL;
1360 kfree(info);
1361 }
1362
1363 mutex_unlock(&bdev->bd_mutex);
1364 bdput(bdev);
1365
1366 return 0;
1367 }
1368
1369 static int blkfront_is_ready(struct xenbus_device *dev)
1370 {
1371 struct blkfront_info *info = dev_get_drvdata(&dev->dev);
1372
1373 return info->is_ready && info->xbdev;
1374 }
1375
1376 static int blkif_open(struct block_device *bdev, fmode_t mode)
1377 {
1378 struct gendisk *disk = bdev->bd_disk;
1379 struct blkfront_info *info;
1380 int err = 0;
1381
1382 mutex_lock(&blkfront_mutex);
1383
1384 info = disk->private_data;
1385 if (!info) {
1386 /* xbdev gone */
1387 err = -ERESTARTSYS;
1388 goto out;
1389 }
1390
1391 mutex_lock(&info->mutex);
1392
1393 if (!info->gd)
1394 /* xbdev is closed */
1395 err = -ERESTARTSYS;
1396
1397 mutex_unlock(&info->mutex);
1398
1399 out:
1400 mutex_unlock(&blkfront_mutex);
1401 return err;
1402 }
1403
1404 static int blkif_release(struct gendisk *disk, fmode_t mode)
1405 {
1406 struct blkfront_info *info = disk->private_data;
1407 struct block_device *bdev;
1408 struct xenbus_device *xbdev;
1409
1410 mutex_lock(&blkfront_mutex);
1411
1412 bdev = bdget_disk(disk, 0);
1413 bdput(bdev);
1414
1415 if (bdev->bd_openers)
1416 goto out;
1417
1418 /*
1419 * Check if we have been instructed to close. We will have
1420 * deferred this request, because the bdev was still open.
1421 */
1422
1423 mutex_lock(&info->mutex);
1424 xbdev = info->xbdev;
1425
1426 if (xbdev && xbdev->state == XenbusStateClosing) {
1427 /* pending switch to state closed */
1428 dev_info(disk_to_dev(bdev->bd_disk), "releasing disk\n");
1429 xlvbd_release_gendisk(info);
1430 xenbus_frontend_closed(info->xbdev);
1431 }
1432
1433 mutex_unlock(&info->mutex);
1434
1435 if (!xbdev) {
1436 /* sudden device removal */
1437 dev_info(disk_to_dev(bdev->bd_disk), "releasing disk\n");
1438 xlvbd_release_gendisk(info);
1439 disk->private_data = NULL;
1440 kfree(info);
1441 }
1442
1443 out:
1444 mutex_unlock(&blkfront_mutex);
1445 return 0;
1446 }
1447
1448 static const struct block_device_operations xlvbd_block_fops =
1449 {
1450 .owner = THIS_MODULE,
1451 .open = blkif_open,
1452 .release = blkif_release,
1453 .getgeo = blkif_getgeo,
1454 .ioctl = blkif_ioctl,
1455 };
1456
1457
1458 static const struct xenbus_device_id blkfront_ids[] = {
1459 { "vbd" },
1460 { "" }
1461 };
1462
1463 static DEFINE_XENBUS_DRIVER(blkfront, ,
1464 .probe = blkfront_probe,
1465 .remove = blkfront_remove,
1466 .resume = blkfront_resume,
1467 .otherend_changed = blkback_changed,
1468 .is_ready = blkfront_is_ready,
1469 );
1470
1471 static int __init xlblk_init(void)
1472 {
1473 int ret;
1474
1475 if (!xen_domain())
1476 return -ENODEV;
1477
1478 if (xen_hvm_domain() && !xen_platform_pci_unplug)
1479 return -ENODEV;
1480
1481 if (register_blkdev(XENVBD_MAJOR, DEV_NAME)) {
1482 printk(KERN_WARNING "xen_blk: can't get major %d with name %s\n",
1483 XENVBD_MAJOR, DEV_NAME);
1484 return -ENODEV;
1485 }
1486
1487 ret = xenbus_register_frontend(&blkfront_driver);
1488 if (ret) {
1489 unregister_blkdev(XENVBD_MAJOR, DEV_NAME);
1490 return ret;
1491 }
1492
1493 return 0;
1494 }
1495 module_init(xlblk_init);
1496
1497
1498 static void __exit xlblk_exit(void)
1499 {
1500 return xenbus_unregister_driver(&blkfront_driver);
1501 }
1502 module_exit(xlblk_exit);
1503
1504 MODULE_DESCRIPTION("Xen virtual block device frontend");
1505 MODULE_LICENSE("GPL");
1506 MODULE_ALIAS_BLOCKDEV_MAJOR(XENVBD_MAJOR);
1507 MODULE_ALIAS("xen:vbd");
1508 MODULE_ALIAS("xenblk");
Linux 2.6 ibm-acpi/thinkpad-acpi driver git tree