Commit | Line | Data |
---|---|---|
9f27ee59 JF |
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> | |
907c3eb1 | 40 | #include <linux/blk-mq.h> |
597592d9 | 41 | #include <linux/hdreg.h> |
440a01a7 | 42 | #include <linux/cdrom.h> |
9f27ee59 | 43 | #include <linux/module.h> |
5a0e3ad6 | 44 | #include <linux/slab.h> |
2a48fc0a | 45 | #include <linux/mutex.h> |
9e973e64 | 46 | #include <linux/scatterlist.h> |
34ae2e47 | 47 | #include <linux/bitmap.h> |
155b7edb | 48 | #include <linux/list.h> |
9f27ee59 | 49 | |
1ccbf534 | 50 | #include <xen/xen.h> |
9f27ee59 JF |
51 | #include <xen/xenbus.h> |
52 | #include <xen/grant_table.h> | |
53 | #include <xen/events.h> | |
54 | #include <xen/page.h> | |
c1c5413a | 55 | #include <xen/platform_pci.h> |
9f27ee59 JF |
56 | |
57 | #include <xen/interface/grant_table.h> | |
58 | #include <xen/interface/io/blkif.h> | |
3e334239 | 59 | #include <xen/interface/io/protocols.h> |
9f27ee59 JF |
60 | |
61 | #include <asm/xen/hypervisor.h> | |
62 | ||
6cc56833 JG |
63 | /* |
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 | |
70 | * | |
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) = | |
73 | * 88KB. | |
74 | */ | |
75 | #define HAS_EXTRA_REQ (BLKIF_MAX_SEGMENTS_PER_REQUEST < XEN_PFN_PER_PAGE) | |
76 | ||
9f27ee59 JF |
77 | enum blkif_state { |
78 | BLKIF_STATE_DISCONNECTED, | |
79 | BLKIF_STATE_CONNECTED, | |
80 | BLKIF_STATE_SUSPENDED, | |
81 | }; | |
82 | ||
0a8704a5 RPM |
83 | struct grant { |
84 | grant_ref_t gref; | |
a7a6df22 | 85 | struct page *page; |
155b7edb | 86 | struct list_head node; |
0a8704a5 RPM |
87 | }; |
88 | ||
6cc56833 JG |
89 | enum blk_req_status { |
90 | REQ_WAITING, | |
91 | REQ_DONE, | |
92 | REQ_ERROR, | |
93 | REQ_EOPNOTSUPP, | |
94 | }; | |
95 | ||
9f27ee59 JF |
96 | struct blk_shadow { |
97 | struct blkif_request req; | |
a945b980 | 98 | struct request *request; |
402b27f9 RPM |
99 | struct grant **grants_used; |
100 | struct grant **indirect_grants; | |
b7649158 | 101 | struct scatterlist *sg; |
c004a6fe | 102 | unsigned int num_sg; |
6cc56833 JG |
103 | enum blk_req_status status; |
104 | ||
105 | #define NO_ASSOCIATED_ID ~0UL | |
106 | /* | |
107 | * Id of the sibling if we ever need 2 requests when handling a | |
108 | * block I/O request | |
109 | */ | |
110 | unsigned long associated_id; | |
402b27f9 RPM |
111 | }; |
112 | ||
113 | struct split_bio { | |
114 | struct bio *bio; | |
115 | atomic_t pending; | |
9f27ee59 JF |
116 | }; |
117 | ||
2a48fc0a | 118 | static DEFINE_MUTEX(blkfront_mutex); |
83d5cde4 | 119 | static const struct block_device_operations xlvbd_block_fops; |
9f27ee59 | 120 | |
402b27f9 RPM |
121 | /* |
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. | |
125 | */ | |
126 | ||
127 | static unsigned int xen_blkif_max_segments = 32; | |
14e710fe JB |
128 | module_param_named(max_indirect_segments, xen_blkif_max_segments, uint, |
129 | S_IRUGO); | |
130 | MODULE_PARM_DESC(max_indirect_segments, | |
131 | "Maximum amount of segments in indirect requests (default is 32)"); | |
402b27f9 | 132 | |
28d949bc BL |
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"); | |
136 | ||
86839c56 BL |
137 | /* |
138 | * Maximum order of pages to be used for the shared ring between front and | |
139 | * backend, 4KB page granularity is used. | |
140 | */ | |
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"); | |
144 | ||
c004a6fe JG |
145 | #define BLK_RING_SIZE(info) \ |
146 | __CONST_RING_SIZE(blkif, XEN_PAGE_SIZE * (info)->nr_ring_pages) | |
147 | ||
148 | #define BLK_MAX_RING_SIZE \ | |
9cce2914 | 149 | __CONST_RING_SIZE(blkif, XEN_PAGE_SIZE * XENBUS_MAX_RING_GRANTS) |
c004a6fe | 150 | |
86839c56 | 151 | /* |
6f03a7ff KRW |
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. | |
86839c56 BL |
154 | */ |
155 | #define RINGREF_NAME_LEN (20) | |
28d949bc BL |
156 | /* |
157 | * queue-%u would take 7 + 10(UINT_MAX) = 17 characters. | |
158 | */ | |
159 | #define QUEUE_NAME_LEN (17) | |
9f27ee59 | 160 | |
81f35161 BL |
161 | /* |
162 | * Per-ring info. | |
163 | * Every blkfront device can associate with one or more blkfront_ring_info, | |
164 | * depending on how many hardware queues/rings to be used. | |
165 | */ | |
166 | struct blkfront_ring_info { | |
11659569 BL |
167 | /* Lock to protect data in every ring buffer. */ |
168 | spinlock_t ring_lock; | |
81f35161 BL |
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; | |
73716df7 BL |
176 | struct list_head grants; |
177 | unsigned int persistent_gnts_c; | |
81f35161 BL |
178 | unsigned long shadow_free; |
179 | struct blkfront_info *dev_info; | |
180 | }; | |
181 | ||
9f27ee59 JF |
182 | /* |
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 :-) | |
186 | */ | |
187 | struct blkfront_info | |
188 | { | |
b70f5fa0 | 189 | struct mutex mutex; |
9f27ee59 | 190 | struct xenbus_device *xbdev; |
9f27ee59 JF |
191 | struct gendisk *gd; |
192 | int vdevice; | |
193 | blkif_vdev_t handle; | |
194 | enum blkif_state connected; | |
3df0e505 | 195 | /* Number of pages per ring buffer. */ |
86839c56 | 196 | unsigned int nr_ring_pages; |
9f27ee59 | 197 | struct request_queue *rq; |
4913efe4 | 198 | unsigned int feature_flush; |
5ea42986 KRW |
199 | unsigned int feature_discard:1; |
200 | unsigned int feature_secdiscard:1; | |
ed30bf31 LD |
201 | unsigned int discard_granularity; |
202 | unsigned int discard_alignment; | |
0a8704a5 | 203 | unsigned int feature_persistent:1; |
c004a6fe | 204 | /* Number of 4KB segments handled */ |
402b27f9 | 205 | unsigned int max_indirect_segments; |
1d78d705 | 206 | int is_ready; |
907c3eb1 | 207 | struct blk_mq_tag_set tag_set; |
3df0e505 BL |
208 | struct blkfront_ring_info *rinfo; |
209 | unsigned int nr_rings; | |
9f27ee59 JF |
210 | }; |
211 | ||
0e345826 JB |
212 | static unsigned int nr_minors; |
213 | static unsigned long *minors; | |
214 | static DEFINE_SPINLOCK(minor_lock); | |
215 | ||
9f27ee59 JF |
216 | #define GRANT_INVALID_REF 0 |
217 | ||
218 | #define PARTS_PER_DISK 16 | |
9246b5f0 | 219 | #define PARTS_PER_EXT_DISK 256 |
9f27ee59 JF |
220 | |
221 | #define BLKIF_MAJOR(dev) ((dev)>>8) | |
222 | #define BLKIF_MINOR(dev) ((dev) & 0xff) | |
223 | ||
9246b5f0 CL |
224 | #define EXT_SHIFT 28 |
225 | #define EXTENDED (1<<EXT_SHIFT) | |
226 | #define VDEV_IS_EXTENDED(dev) ((dev)&(EXTENDED)) | |
227 | #define BLKIF_MINOR_EXT(dev) ((dev)&(~EXTENDED)) | |
c80a4209 SS |
228 | #define EMULATED_HD_DISK_MINOR_OFFSET (0) |
229 | #define EMULATED_HD_DISK_NAME_OFFSET (EMULATED_HD_DISK_MINOR_OFFSET / 256) | |
196cfe2a SB |
230 | #define EMULATED_SD_DISK_MINOR_OFFSET (0) |
231 | #define EMULATED_SD_DISK_NAME_OFFSET (EMULATED_SD_DISK_MINOR_OFFSET / 256) | |
9f27ee59 | 232 | |
9246b5f0 | 233 | #define DEV_NAME "xvd" /* name in /dev */ |
9f27ee59 | 234 | |
c004a6fe JG |
235 | /* |
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 | |
239 | */ | |
240 | #define GRANTS_PER_PSEG (PAGE_SIZE / XEN_PAGE_SIZE) | |
241 | ||
242 | #define GRANTS_PER_INDIRECT_FRAME \ | |
243 | (XEN_PAGE_SIZE / sizeof(struct blkif_request_segment)) | |
244 | ||
245 | #define PSEGS_PER_INDIRECT_FRAME \ | |
246 | (GRANTS_INDIRECT_FRAME / GRANTS_PSEGS) | |
247 | ||
248 | #define INDIRECT_GREFS(_grants) \ | |
249 | DIV_ROUND_UP(_grants, GRANTS_PER_INDIRECT_FRAME) | |
250 | ||
251 | #define GREFS(_psegs) ((_psegs) * GRANTS_PER_PSEG) | |
402b27f9 | 252 | |
81f35161 | 253 | static int blkfront_setup_indirect(struct blkfront_ring_info *rinfo); |
3df0e505 | 254 | static void blkfront_gather_backend_features(struct blkfront_info *info); |
402b27f9 | 255 | |
81f35161 | 256 | static int get_id_from_freelist(struct blkfront_ring_info *rinfo) |
9f27ee59 | 257 | { |
81f35161 BL |
258 | unsigned long free = rinfo->shadow_free; |
259 | ||
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 */ | |
9f27ee59 JF |
263 | return free; |
264 | } | |
265 | ||
81f35161 | 266 | static int add_id_to_freelist(struct blkfront_ring_info *rinfo, |
6f03a7ff | 267 | unsigned long id) |
9f27ee59 | 268 | { |
81f35161 | 269 | if (rinfo->shadow[id].req.u.rw.id != id) |
6878c32e | 270 | return -EINVAL; |
81f35161 | 271 | if (rinfo->shadow[id].request == NULL) |
6878c32e | 272 | return -EINVAL; |
81f35161 BL |
273 | rinfo->shadow[id].req.u.rw.id = rinfo->shadow_free; |
274 | rinfo->shadow[id].request = NULL; | |
275 | rinfo->shadow_free = id; | |
6878c32e | 276 | return 0; |
9f27ee59 JF |
277 | } |
278 | ||
81f35161 | 279 | static int fill_grant_buffer(struct blkfront_ring_info *rinfo, int num) |
9c1e050c | 280 | { |
81f35161 | 281 | struct blkfront_info *info = rinfo->dev_info; |
9c1e050c RPM |
282 | struct page *granted_page; |
283 | struct grant *gnt_list_entry, *n; | |
284 | int i = 0; | |
285 | ||
6f03a7ff | 286 | while (i < num) { |
9c1e050c RPM |
287 | gnt_list_entry = kzalloc(sizeof(struct grant), GFP_NOIO); |
288 | if (!gnt_list_entry) | |
289 | goto out_of_memory; | |
290 | ||
bfe11d6d RPM |
291 | if (info->feature_persistent) { |
292 | granted_page = alloc_page(GFP_NOIO); | |
293 | if (!granted_page) { | |
294 | kfree(gnt_list_entry); | |
295 | goto out_of_memory; | |
296 | } | |
a7a6df22 | 297 | gnt_list_entry->page = granted_page; |
9c1e050c RPM |
298 | } |
299 | ||
9c1e050c | 300 | gnt_list_entry->gref = GRANT_INVALID_REF; |
73716df7 | 301 | list_add(&gnt_list_entry->node, &rinfo->grants); |
9c1e050c RPM |
302 | i++; |
303 | } | |
304 | ||
305 | return 0; | |
306 | ||
307 | out_of_memory: | |
308 | list_for_each_entry_safe(gnt_list_entry, n, | |
73716df7 | 309 | &rinfo->grants, node) { |
9c1e050c | 310 | list_del(&gnt_list_entry->node); |
bfe11d6d | 311 | if (info->feature_persistent) |
a7a6df22 | 312 | __free_page(gnt_list_entry->page); |
9c1e050c RPM |
313 | kfree(gnt_list_entry); |
314 | i--; | |
315 | } | |
316 | BUG_ON(i != 0); | |
317 | return -ENOMEM; | |
318 | } | |
319 | ||
73716df7 | 320 | static struct grant *get_free_grant(struct blkfront_ring_info *rinfo) |
9c1e050c RPM |
321 | { |
322 | struct grant *gnt_list_entry; | |
9c1e050c | 323 | |
73716df7 BL |
324 | BUG_ON(list_empty(&rinfo->grants)); |
325 | gnt_list_entry = list_first_entry(&rinfo->grants, struct grant, | |
4f503fbd | 326 | node); |
9c1e050c RPM |
327 | list_del(&gnt_list_entry->node); |
328 | ||
4f503fbd | 329 | if (gnt_list_entry->gref != GRANT_INVALID_REF) |
73716df7 | 330 | rinfo->persistent_gnts_c--; |
4f503fbd JG |
331 | |
332 | return gnt_list_entry; | |
333 | } | |
334 | ||
335 | static inline void grant_foreign_access(const struct grant *gnt_list_entry, | |
336 | const struct blkfront_info *info) | |
337 | { | |
338 | gnttab_page_grant_foreign_access_ref_one(gnt_list_entry->gref, | |
339 | info->xbdev->otherend_id, | |
340 | gnt_list_entry->page, | |
341 | 0); | |
342 | } | |
343 | ||
344 | static struct grant *get_grant(grant_ref_t *gref_head, | |
345 | unsigned long gfn, | |
73716df7 | 346 | struct blkfront_ring_info *rinfo) |
4f503fbd | 347 | { |
73716df7 BL |
348 | struct grant *gnt_list_entry = get_free_grant(rinfo); |
349 | struct blkfront_info *info = rinfo->dev_info; | |
4f503fbd JG |
350 | |
351 | if (gnt_list_entry->gref != GRANT_INVALID_REF) | |
9c1e050c | 352 | return gnt_list_entry; |
4f503fbd JG |
353 | |
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); | |
359 | else { | |
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, | |
363 | gfn, 0); | |
9c1e050c RPM |
364 | } |
365 | ||
4f503fbd JG |
366 | return gnt_list_entry; |
367 | } | |
368 | ||
369 | static struct grant *get_indirect_grant(grant_ref_t *gref_head, | |
73716df7 | 370 | struct blkfront_ring_info *rinfo) |
4f503fbd | 371 | { |
73716df7 BL |
372 | struct grant *gnt_list_entry = get_free_grant(rinfo); |
373 | struct blkfront_info *info = rinfo->dev_info; | |
4f503fbd JG |
374 | |
375 | if (gnt_list_entry->gref != GRANT_INVALID_REF) | |
376 | return gnt_list_entry; | |
377 | ||
9c1e050c RPM |
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); | |
bfe11d6d | 381 | if (!info->feature_persistent) { |
4f503fbd JG |
382 | struct page *indirect_page; |
383 | ||
384 | /* Fetch a pre-allocated page to use for indirect grefs */ | |
73716df7 BL |
385 | BUG_ON(list_empty(&rinfo->indirect_pages)); |
386 | indirect_page = list_first_entry(&rinfo->indirect_pages, | |
4f503fbd JG |
387 | struct page, lru); |
388 | list_del(&indirect_page->lru); | |
389 | gnt_list_entry->page = indirect_page; | |
bfe11d6d | 390 | } |
4f503fbd JG |
391 | grant_foreign_access(gnt_list_entry, info); |
392 | ||
9c1e050c RPM |
393 | return gnt_list_entry; |
394 | } | |
395 | ||
6878c32e KRW |
396 | static const char *op_name(int op) |
397 | { | |
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" }; | |
404 | ||
405 | if (op < 0 || op >= ARRAY_SIZE(names)) | |
406 | return "unknown"; | |
407 | ||
408 | if (!names[op]) | |
409 | return "reserved"; | |
410 | ||
411 | return names[op]; | |
412 | } | |
0e345826 JB |
413 | static int xlbd_reserve_minors(unsigned int minor, unsigned int nr) |
414 | { | |
415 | unsigned int end = minor + nr; | |
416 | int rc; | |
417 | ||
418 | if (end > nr_minors) { | |
419 | unsigned long *bitmap, *old; | |
420 | ||
f094148a | 421 | bitmap = kcalloc(BITS_TO_LONGS(end), sizeof(*bitmap), |
0e345826 JB |
422 | GFP_KERNEL); |
423 | if (bitmap == NULL) | |
424 | return -ENOMEM; | |
425 | ||
426 | spin_lock(&minor_lock); | |
427 | if (end > nr_minors) { | |
428 | old = minors; | |
429 | memcpy(bitmap, minors, | |
430 | BITS_TO_LONGS(nr_minors) * sizeof(*bitmap)); | |
431 | minors = bitmap; | |
432 | nr_minors = BITS_TO_LONGS(end) * BITS_PER_LONG; | |
433 | } else | |
434 | old = bitmap; | |
435 | spin_unlock(&minor_lock); | |
436 | kfree(old); | |
437 | } | |
438 | ||
439 | spin_lock(&minor_lock); | |
440 | if (find_next_bit(minors, end, minor) >= end) { | |
34ae2e47 | 441 | bitmap_set(minors, minor, nr); |
0e345826 JB |
442 | rc = 0; |
443 | } else | |
444 | rc = -EBUSY; | |
445 | spin_unlock(&minor_lock); | |
446 | ||
447 | return rc; | |
448 | } | |
449 | ||
450 | static void xlbd_release_minors(unsigned int minor, unsigned int nr) | |
451 | { | |
452 | unsigned int end = minor + nr; | |
453 | ||
454 | BUG_ON(end > nr_minors); | |
455 | spin_lock(&minor_lock); | |
34ae2e47 | 456 | bitmap_clear(minors, minor, nr); |
0e345826 JB |
457 | spin_unlock(&minor_lock); |
458 | } | |
459 | ||
9f27ee59 JF |
460 | static void blkif_restart_queue_callback(void *arg) |
461 | { | |
81f35161 BL |
462 | struct blkfront_ring_info *rinfo = (struct blkfront_ring_info *)arg; |
463 | schedule_work(&rinfo->work); | |
9f27ee59 JF |
464 | } |
465 | ||
afe42d7d | 466 | static int blkif_getgeo(struct block_device *bd, struct hd_geometry *hg) |
597592d9 IC |
467 | { |
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; | |
472 | ||
473 | hg->heads = 0xff; | |
474 | hg->sectors = 0x3f; | |
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; | |
479 | return 0; | |
480 | } | |
481 | ||
a63c848b | 482 | static int blkif_ioctl(struct block_device *bdev, fmode_t mode, |
62aa0054 | 483 | unsigned command, unsigned long argument) |
440a01a7 | 484 | { |
a63c848b | 485 | struct blkfront_info *info = bdev->bd_disk->private_data; |
440a01a7 CL |
486 | int i; |
487 | ||
488 | dev_dbg(&info->xbdev->dev, "command: 0x%x, argument: 0x%lx\n", | |
489 | command, (long)argument); | |
490 | ||
491 | switch (command) { | |
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))) | |
496 | return -EFAULT; | |
497 | return 0; | |
498 | ||
499 | case CDROM_GET_CAPABILITY: { | |
500 | struct gendisk *gd = info->gd; | |
501 | if (gd->flags & GENHD_FL_CD) | |
502 | return 0; | |
503 | return -EINVAL; | |
504 | } | |
505 | ||
506 | default: | |
507 | /*printk(KERN_ALERT "ioctl %08x not supported by Xen blkdev\n", | |
508 | command);*/ | |
509 | return -EINVAL; /* same return as native Linux */ | |
510 | } | |
511 | ||
512 | return 0; | |
513 | } | |
514 | ||
2e073969 JG |
515 | static unsigned long blkif_ring_get_request(struct blkfront_ring_info *rinfo, |
516 | struct request *req, | |
517 | struct blkif_request **ring_req) | |
518 | { | |
519 | unsigned long id; | |
520 | ||
521 | *ring_req = RING_GET_REQUEST(&rinfo->ring, rinfo->ring.req_prod_pvt); | |
522 | rinfo->ring.req_prod_pvt++; | |
523 | ||
524 | id = get_id_from_freelist(rinfo); | |
525 | rinfo->shadow[id].request = req; | |
6cc56833 JG |
526 | rinfo->shadow[id].status = REQ_WAITING; |
527 | rinfo->shadow[id].associated_id = NO_ASSOCIATED_ID; | |
2e073969 JG |
528 | |
529 | (*ring_req)->u.rw.id = id; | |
530 | ||
531 | return id; | |
532 | } | |
533 | ||
81f35161 | 534 | static int blkif_queue_discard_req(struct request *req, struct blkfront_ring_info *rinfo) |
9f27ee59 | 535 | { |
81f35161 | 536 | struct blkfront_info *info = rinfo->dev_info; |
9f27ee59 | 537 | struct blkif_request *ring_req; |
9f27ee59 | 538 | unsigned long id; |
33204663 JG |
539 | |
540 | /* Fill out a communications ring structure. */ | |
2e073969 | 541 | id = blkif_ring_get_request(rinfo, req, &ring_req); |
33204663 JG |
542 | |
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; | |
549 | else | |
550 | ring_req->u.discard.flag = 0; | |
551 | ||
33204663 | 552 | /* Keep a private copy so we can reissue requests when recovering. */ |
81f35161 | 553 | rinfo->shadow[id].req = *ring_req; |
33204663 JG |
554 | |
555 | return 0; | |
556 | } | |
557 | ||
c004a6fe JG |
558 | struct setup_rw_req { |
559 | unsigned int grant_idx; | |
560 | struct blkif_request_segment *segments; | |
81f35161 | 561 | struct blkfront_ring_info *rinfo; |
c004a6fe JG |
562 | struct blkif_request *ring_req; |
563 | grant_ref_t gref_head; | |
564 | unsigned int id; | |
565 | /* Only used when persistent grant is used and it's a read request */ | |
566 | bool need_copy; | |
567 | unsigned int bvec_off; | |
568 | char *bvec_data; | |
6cc56833 JG |
569 | |
570 | bool require_extra_req; | |
571 | struct blkif_request *extra_ring_req; | |
c004a6fe JG |
572 | }; |
573 | ||
574 | static void blkif_setup_rw_req_grant(unsigned long gfn, unsigned int offset, | |
575 | unsigned int len, void *data) | |
576 | { | |
577 | struct setup_rw_req *setup = data; | |
578 | int n, ref; | |
579 | struct grant *gnt_list_entry; | |
9f27ee59 | 580 | unsigned int fsect, lsect; |
c004a6fe JG |
581 | /* Convenient aliases */ |
582 | unsigned int grant_idx = setup->grant_idx; | |
583 | struct blkif_request *ring_req = setup->ring_req; | |
81f35161 | 584 | struct blkfront_ring_info *rinfo = setup->rinfo; |
6cc56833 JG |
585 | /* |
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 | |
589 | * split. | |
590 | */ | |
81f35161 | 591 | struct blk_shadow *shadow = &rinfo->shadow[setup->id]; |
c004a6fe | 592 | |
6cc56833 JG |
593 | if (unlikely(setup->require_extra_req && |
594 | grant_idx >= BLKIF_MAX_SEGMENTS_PER_REQUEST)) { | |
595 | /* | |
596 | * We are using the second request, setup grant_idx | |
597 | * to be the index of the segment array. | |
598 | */ | |
599 | grant_idx -= BLKIF_MAX_SEGMENTS_PER_REQUEST; | |
600 | ring_req = setup->extra_ring_req; | |
601 | } | |
602 | ||
c004a6fe JG |
603 | if ((ring_req->operation == BLKIF_OP_INDIRECT) && |
604 | (grant_idx % GRANTS_PER_INDIRECT_FRAME == 0)) { | |
605 | if (setup->segments) | |
606 | kunmap_atomic(setup->segments); | |
607 | ||
608 | n = grant_idx / GRANTS_PER_INDIRECT_FRAME; | |
73716df7 | 609 | gnt_list_entry = get_indirect_grant(&setup->gref_head, rinfo); |
c004a6fe JG |
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; | |
613 | } | |
614 | ||
73716df7 | 615 | gnt_list_entry = get_grant(&setup->gref_head, gfn, rinfo); |
c004a6fe | 616 | ref = gnt_list_entry->gref; |
6cc56833 JG |
617 | /* |
618 | * All the grants are stored in the shadow of the first | |
619 | * request. Therefore we have to use the global index. | |
620 | */ | |
621 | shadow->grants_used[setup->grant_idx] = gnt_list_entry; | |
c004a6fe JG |
622 | |
623 | if (setup->need_copy) { | |
624 | void *shared_data; | |
625 | ||
626 | shared_data = kmap_atomic(gnt_list_entry->page); | |
627 | /* | |
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 | |
634 | * changes | |
635 | */ | |
636 | memcpy(shared_data + offset, | |
637 | setup->bvec_data + setup->bvec_off, | |
638 | len); | |
639 | ||
640 | kunmap_atomic(shared_data); | |
641 | setup->bvec_off += len; | |
642 | } | |
643 | ||
644 | fsect = offset >> 9; | |
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) { | |
649 | .gref = ref, | |
650 | .first_sect = fsect, | |
651 | .last_sect = lsect }; | |
652 | } else { | |
653 | setup->segments[grant_idx % GRANTS_PER_INDIRECT_FRAME] = | |
654 | (struct blkif_request_segment) { | |
655 | .gref = ref, | |
656 | .first_sect = fsect, | |
657 | .last_sect = lsect }; | |
658 | } | |
659 | ||
660 | (setup->grant_idx)++; | |
661 | } | |
662 | ||
6cc56833 JG |
663 | static void blkif_setup_extra_req(struct blkif_request *first, |
664 | struct blkif_request *second) | |
665 | { | |
666 | uint16_t nr_segments = first->u.rw.nr_segments; | |
667 | ||
668 | /* | |
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. | |
671 | */ | |
672 | first->u.rw.nr_segments = BLKIF_MAX_SEGMENTS_PER_REQUEST; | |
673 | ||
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; | |
677 | ||
678 | second->u.rw.handle = first->u.rw.handle; | |
679 | second->operation = first->operation; | |
680 | } | |
681 | ||
81f35161 | 682 | static int blkif_queue_rw_req(struct request *req, struct blkfront_ring_info *rinfo) |
9f27ee59 | 683 | { |
81f35161 | 684 | struct blkfront_info *info = rinfo->dev_info; |
6cc56833 JG |
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; | |
c004a6fe JG |
688 | int i; |
689 | struct setup_rw_req setup = { | |
690 | .grant_idx = 0, | |
691 | .segments = NULL, | |
81f35161 | 692 | .rinfo = rinfo, |
c004a6fe JG |
693 | .need_copy = rq_data_dir(req) && info->feature_persistent, |
694 | }; | |
0a8704a5 RPM |
695 | |
696 | /* | |
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. | |
700 | */ | |
9e973e64 | 701 | struct scatterlist *sg; |
c004a6fe | 702 | int num_sg, max_grefs, num_grant; |
9f27ee59 | 703 | |
c004a6fe | 704 | max_grefs = req->nr_phys_segments * GRANTS_PER_PSEG; |
c47206e2 RPM |
705 | if (max_grefs > BLKIF_MAX_SEGMENTS_PER_REQUEST) |
706 | /* | |
707 | * If we are using indirect segments we need to account | |
708 | * for the indirect grefs used in the request. | |
709 | */ | |
c004a6fe | 710 | max_grefs += INDIRECT_GREFS(max_grefs); |
402b27f9 | 711 | |
3df0e505 BL |
712 | /* |
713 | * We have to reserve 'max_grefs' grants because persistent | |
714 | * grants are shared by all rings. | |
715 | */ | |
716 | if (max_grefs > 0) | |
717 | if (gnttab_alloc_grant_references(max_grefs, &setup.gref_head) < 0) { | |
0a8704a5 | 718 | gnttab_request_free_callback( |
81f35161 | 719 | &rinfo->callback, |
0a8704a5 | 720 | blkif_restart_queue_callback, |
81f35161 | 721 | rinfo, |
402b27f9 | 722 | max_grefs); |
0a8704a5 RPM |
723 | return 1; |
724 | } | |
9f27ee59 JF |
725 | |
726 | /* Fill out a communications ring structure. */ | |
2e073969 | 727 | id = blkif_ring_get_request(rinfo, req, &ring_req); |
9f27ee59 | 728 | |
81f35161 | 729 | num_sg = blk_rq_map_sg(req->q, req, rinfo->shadow[id].sg); |
c004a6fe JG |
730 | num_grant = 0; |
731 | /* Calculate the number of grant used */ | |
81f35161 | 732 | for_each_sg(rinfo->shadow[id].sg, sg, num_sg, i) |
c004a6fe JG |
733 | num_grant += gnttab_count_grant(sg->offset, sg->length); |
734 | ||
6cc56833 JG |
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); | |
738 | ||
81f35161 | 739 | rinfo->shadow[id].num_sg = num_sg; |
6cc56833 JG |
740 | if (num_grant > BLKIF_MAX_SEGMENTS_PER_REQUEST && |
741 | likely(!require_extra_req)) { | |
33204663 JG |
742 | /* |
743 | * The indirect operation can only be a BLKIF_OP_READ or | |
744 | * BLKIF_OP_WRITE | |
745 | */ | |
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; | |
c004a6fe | 752 | ring_req->u.indirect.nr_segments = num_grant; |
ed30bf31 | 753 | } else { |
33204663 JG |
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)) { | |
402b27f9 | 759 | /* |
33204663 JG |
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.) | |
402b27f9 | 765 | */ |
33204663 JG |
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; | |
771 | break; | |
772 | case REQ_FLUSH: | |
773 | ring_req->operation = | |
774 | BLKIF_OP_FLUSH_DISKCACHE; | |
775 | break; | |
776 | default: | |
777 | ring_req->operation = 0; | |
402b27f9 | 778 | } |
402b27f9 | 779 | } |
c004a6fe | 780 | ring_req->u.rw.nr_segments = num_grant; |
6cc56833 JG |
781 | if (unlikely(require_extra_req)) { |
782 | extra_id = blkif_ring_get_request(rinfo, req, | |
783 | &extra_ring_req); | |
784 | /* | |
785 | * Only the first request contains the scatter-gather | |
786 | * list. | |
787 | */ | |
788 | rinfo->shadow[extra_id].num_sg = 0; | |
789 | ||
790 | blkif_setup_extra_req(ring_req, extra_ring_req); | |
791 | ||
792 | /* Link the 2 requests together */ | |
793 | rinfo->shadow[extra_id].associated_id = id; | |
794 | rinfo->shadow[id].associated_id = extra_id; | |
795 | } | |
33204663 | 796 | } |
0a8704a5 | 797 | |
c004a6fe JG |
798 | setup.ring_req = ring_req; |
799 | setup.id = id; | |
6cc56833 JG |
800 | |
801 | setup.require_extra_req = require_extra_req; | |
802 | if (unlikely(require_extra_req)) | |
803 | setup.extra_ring_req = extra_ring_req; | |
804 | ||
81f35161 | 805 | for_each_sg(rinfo->shadow[id].sg, sg, num_sg, i) { |
c004a6fe | 806 | BUG_ON(sg->offset + sg->length > PAGE_SIZE); |
0a8704a5 | 807 | |
c004a6fe JG |
808 | if (setup.need_copy) { |
809 | setup.bvec_off = sg->offset; | |
810 | setup.bvec_data = kmap_atomic(sg_page(sg)); | |
811 | } | |
0a8704a5 | 812 | |
c004a6fe JG |
813 | gnttab_foreach_grant_in_range(sg_page(sg), |
814 | sg->offset, | |
815 | sg->length, | |
816 | blkif_setup_rw_req_grant, | |
817 | &setup); | |
0a8704a5 | 818 | |
c004a6fe JG |
819 | if (setup.need_copy) |
820 | kunmap_atomic(setup.bvec_data); | |
9f27ee59 | 821 | } |
c004a6fe JG |
822 | if (setup.segments) |
823 | kunmap_atomic(setup.segments); | |
9f27ee59 | 824 | |
9f27ee59 | 825 | /* Keep a private copy so we can reissue requests when recovering. */ |
81f35161 | 826 | rinfo->shadow[id].req = *ring_req; |
6cc56833 JG |
827 | if (unlikely(require_extra_req)) |
828 | rinfo->shadow[extra_id].req = *extra_ring_req; | |
9f27ee59 | 829 | |
3df0e505 | 830 | if (max_grefs > 0) |
c004a6fe | 831 | gnttab_free_grant_references(setup.gref_head); |
9f27ee59 JF |
832 | |
833 | return 0; | |
834 | } | |
835 | ||
33204663 JG |
836 | /* |
837 | * Generate a Xen blkfront IO request from a blk layer request. Reads | |
838 | * and writes are handled as expected. | |
839 | * | |
840 | * @req: a request struct | |
841 | */ | |
81f35161 | 842 | static int blkif_queue_request(struct request *req, struct blkfront_ring_info *rinfo) |
33204663 | 843 | { |
81f35161 | 844 | if (unlikely(rinfo->dev_info->connected != BLKIF_STATE_CONNECTED)) |
33204663 JG |
845 | return 1; |
846 | ||
847 | if (unlikely(req->cmd_flags & (REQ_DISCARD | REQ_SECURE))) | |
81f35161 | 848 | return blkif_queue_discard_req(req, rinfo); |
33204663 | 849 | else |
81f35161 | 850 | return blkif_queue_rw_req(req, rinfo); |
33204663 | 851 | } |
9f27ee59 | 852 | |
81f35161 | 853 | static inline void flush_requests(struct blkfront_ring_info *rinfo) |
9f27ee59 JF |
854 | { |
855 | int notify; | |
856 | ||
81f35161 | 857 | RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&rinfo->ring, notify); |
9f27ee59 JF |
858 | |
859 | if (notify) | |
81f35161 | 860 | notify_remote_via_irq(rinfo->irq); |
9f27ee59 JF |
861 | } |
862 | ||
ad42d391 VK |
863 | static inline bool blkif_request_flush_invalid(struct request *req, |
864 | struct blkfront_info *info) | |
0f1ca65e AA |
865 | { |
866 | return ((req->cmd_type != REQ_TYPE_FS) || | |
ad42d391 VK |
867 | ((req->cmd_flags & REQ_FLUSH) && |
868 | !(info->feature_flush & REQ_FLUSH)) || | |
869 | ((req->cmd_flags & REQ_FUA) && | |
870 | !(info->feature_flush & REQ_FUA))); | |
0f1ca65e AA |
871 | } |
872 | ||
907c3eb1 | 873 | static int blkif_queue_rq(struct blk_mq_hw_ctx *hctx, |
6f03a7ff | 874 | const struct blk_mq_queue_data *qd) |
9f27ee59 | 875 | { |
11659569 | 876 | unsigned long flags; |
81f35161 | 877 | struct blkfront_ring_info *rinfo = (struct blkfront_ring_info *)hctx->driver_data; |
9f27ee59 | 878 | |
907c3eb1 | 879 | blk_mq_start_request(qd->rq); |
11659569 | 880 | spin_lock_irqsave(&rinfo->ring_lock, flags); |
81f35161 | 881 | if (RING_FULL(&rinfo->ring)) |
907c3eb1 | 882 | goto out_busy; |
9f27ee59 | 883 | |
81f35161 | 884 | if (blkif_request_flush_invalid(qd->rq, rinfo->dev_info)) |
907c3eb1 | 885 | goto out_err; |
296b2f6a | 886 | |
81f35161 | 887 | if (blkif_queue_request(qd->rq, rinfo)) |
907c3eb1 | 888 | goto out_busy; |
296b2f6a | 889 | |
81f35161 | 890 | flush_requests(rinfo); |
11659569 | 891 | spin_unlock_irqrestore(&rinfo->ring_lock, flags); |
907c3eb1 | 892 | return BLK_MQ_RQ_QUEUE_OK; |
9f27ee59 | 893 | |
907c3eb1 | 894 | out_err: |
11659569 | 895 | spin_unlock_irqrestore(&rinfo->ring_lock, flags); |
907c3eb1 | 896 | return BLK_MQ_RQ_QUEUE_ERROR; |
9f27ee59 | 897 | |
907c3eb1 | 898 | out_busy: |
11659569 | 899 | spin_unlock_irqrestore(&rinfo->ring_lock, flags); |
907c3eb1 BL |
900 | blk_mq_stop_hw_queue(hctx); |
901 | return BLK_MQ_RQ_QUEUE_BUSY; | |
9f27ee59 JF |
902 | } |
903 | ||
81f35161 BL |
904 | static int blk_mq_init_hctx(struct blk_mq_hw_ctx *hctx, void *data, |
905 | unsigned int index) | |
906 | { | |
907 | struct blkfront_info *info = (struct blkfront_info *)data; | |
908 | ||
3df0e505 BL |
909 | BUG_ON(info->nr_rings <= index); |
910 | hctx->driver_data = &info->rinfo[index]; | |
81f35161 BL |
911 | return 0; |
912 | } | |
913 | ||
907c3eb1 BL |
914 | static struct blk_mq_ops blkfront_mq_ops = { |
915 | .queue_rq = blkif_queue_rq, | |
916 | .map_queue = blk_mq_map_queue, | |
81f35161 | 917 | .init_hctx = blk_mq_init_hctx, |
907c3eb1 BL |
918 | }; |
919 | ||
402b27f9 | 920 | static int xlvbd_init_blk_queue(struct gendisk *gd, u16 sector_size, |
7c4d7d71 | 921 | unsigned int physical_sector_size, |
402b27f9 | 922 | unsigned int segments) |
9f27ee59 | 923 | { |
165125e1 | 924 | struct request_queue *rq; |
ed30bf31 | 925 | struct blkfront_info *info = gd->private_data; |
9f27ee59 | 926 | |
907c3eb1 BL |
927 | memset(&info->tag_set, 0, sizeof(info->tag_set)); |
928 | info->tag_set.ops = &blkfront_mq_ops; | |
28d949bc | 929 | info->tag_set.nr_hw_queues = info->nr_rings; |
6cc56833 JG |
930 | if (HAS_EXTRA_REQ && info->max_indirect_segments == 0) { |
931 | /* | |
932 | * When indirect descriptior is not supported, the I/O request | |
933 | * will be split between multiple request in the ring. | |
934 | * To avoid problems when sending the request, divide by | |
935 | * 2 the depth of the queue. | |
936 | */ | |
937 | info->tag_set.queue_depth = BLK_RING_SIZE(info) / 2; | |
938 | } else | |
939 | info->tag_set.queue_depth = BLK_RING_SIZE(info); | |
907c3eb1 BL |
940 | info->tag_set.numa_node = NUMA_NO_NODE; |
941 | info->tag_set.flags = BLK_MQ_F_SHOULD_MERGE | BLK_MQ_F_SG_MERGE; | |
942 | info->tag_set.cmd_size = 0; | |
943 | info->tag_set.driver_data = info; | |
944 | ||
945 | if (blk_mq_alloc_tag_set(&info->tag_set)) | |
bde21f73 | 946 | return -EINVAL; |
907c3eb1 BL |
947 | rq = blk_mq_init_queue(&info->tag_set); |
948 | if (IS_ERR(rq)) { | |
949 | blk_mq_free_tag_set(&info->tag_set); | |
bde21f73 | 950 | return PTR_ERR(rq); |
907c3eb1 | 951 | } |
9f27ee59 | 952 | |
66d352e1 | 953 | queue_flag_set_unlocked(QUEUE_FLAG_VIRT, rq); |
9f27ee59 | 954 | |
ed30bf31 LD |
955 | if (info->feature_discard) { |
956 | queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, rq); | |
957 | blk_queue_max_discard_sectors(rq, get_capacity(gd)); | |
958 | rq->limits.discard_granularity = info->discard_granularity; | |
959 | rq->limits.discard_alignment = info->discard_alignment; | |
5ea42986 KRW |
960 | if (info->feature_secdiscard) |
961 | queue_flag_set_unlocked(QUEUE_FLAG_SECDISCARD, rq); | |
ed30bf31 LD |
962 | } |
963 | ||
9f27ee59 | 964 | /* Hard sector size and max sectors impersonate the equiv. hardware. */ |
e1defc4f | 965 | blk_queue_logical_block_size(rq, sector_size); |
7c4d7d71 | 966 | blk_queue_physical_block_size(rq, physical_sector_size); |
c004a6fe | 967 | blk_queue_max_hw_sectors(rq, (segments * XEN_PAGE_SIZE) / 512); |
9f27ee59 JF |
968 | |
969 | /* Each segment in a request is up to an aligned page in size. */ | |
970 | blk_queue_segment_boundary(rq, PAGE_SIZE - 1); | |
971 | blk_queue_max_segment_size(rq, PAGE_SIZE); | |
972 | ||
973 | /* Ensure a merged request will fit in a single I/O ring slot. */ | |
c004a6fe | 974 | blk_queue_max_segments(rq, segments / GRANTS_PER_PSEG); |
9f27ee59 JF |
975 | |
976 | /* Make sure buffer addresses are sector-aligned. */ | |
977 | blk_queue_dma_alignment(rq, 511); | |
978 | ||
1c91fe1a IC |
979 | /* Make sure we don't use bounce buffers. */ |
980 | blk_queue_bounce_limit(rq, BLK_BOUNCE_ANY); | |
981 | ||
9f27ee59 JF |
982 | gd->queue = rq; |
983 | ||
984 | return 0; | |
985 | } | |
986 | ||
fdf9b965 VK |
987 | static const char *flush_info(unsigned int feature_flush) |
988 | { | |
989 | switch (feature_flush & ((REQ_FLUSH | REQ_FUA))) { | |
990 | case REQ_FLUSH|REQ_FUA: | |
991 | return "barrier: enabled;"; | |
992 | case REQ_FLUSH: | |
993 | return "flush diskcache: enabled;"; | |
994 | default: | |
995 | return "barrier or flush: disabled;"; | |
996 | } | |
997 | } | |
9f27ee59 | 998 | |
4913efe4 | 999 | static void xlvbd_flush(struct blkfront_info *info) |
9f27ee59 | 1000 | { |
4913efe4 | 1001 | blk_queue_flush(info->rq, info->feature_flush); |
fdf9b965 VK |
1002 | pr_info("blkfront: %s: %s %s %s %s %s\n", |
1003 | info->gd->disk_name, flush_info(info->feature_flush), | |
1004 | "persistent grants:", info->feature_persistent ? | |
1005 | "enabled;" : "disabled;", "indirect descriptors:", | |
1006 | info->max_indirect_segments ? "enabled;" : "disabled;"); | |
9f27ee59 JF |
1007 | } |
1008 | ||
c80a4209 SS |
1009 | static int xen_translate_vdev(int vdevice, int *minor, unsigned int *offset) |
1010 | { | |
1011 | int major; | |
1012 | major = BLKIF_MAJOR(vdevice); | |
1013 | *minor = BLKIF_MINOR(vdevice); | |
1014 | switch (major) { | |
1015 | case XEN_IDE0_MAJOR: | |
1016 | *offset = (*minor / 64) + EMULATED_HD_DISK_NAME_OFFSET; | |
1017 | *minor = ((*minor / 64) * PARTS_PER_DISK) + | |
1018 | EMULATED_HD_DISK_MINOR_OFFSET; | |
1019 | break; | |
1020 | case XEN_IDE1_MAJOR: | |
1021 | *offset = (*minor / 64) + 2 + EMULATED_HD_DISK_NAME_OFFSET; | |
1022 | *minor = (((*minor / 64) + 2) * PARTS_PER_DISK) + | |
1023 | EMULATED_HD_DISK_MINOR_OFFSET; | |
1024 | break; | |
1025 | case XEN_SCSI_DISK0_MAJOR: | |
1026 | *offset = (*minor / PARTS_PER_DISK) + EMULATED_SD_DISK_NAME_OFFSET; | |
1027 | *minor = *minor + EMULATED_SD_DISK_MINOR_OFFSET; | |
1028 | break; | |
1029 | case XEN_SCSI_DISK1_MAJOR: | |
1030 | case XEN_SCSI_DISK2_MAJOR: | |
1031 | case XEN_SCSI_DISK3_MAJOR: | |
1032 | case XEN_SCSI_DISK4_MAJOR: | |
1033 | case XEN_SCSI_DISK5_MAJOR: | |
1034 | case XEN_SCSI_DISK6_MAJOR: | |
1035 | case XEN_SCSI_DISK7_MAJOR: | |
1036 | *offset = (*minor / PARTS_PER_DISK) + | |
1037 | ((major - XEN_SCSI_DISK1_MAJOR + 1) * 16) + | |
1038 | EMULATED_SD_DISK_NAME_OFFSET; | |
1039 | *minor = *minor + | |
1040 | ((major - XEN_SCSI_DISK1_MAJOR + 1) * 16 * PARTS_PER_DISK) + | |
1041 | EMULATED_SD_DISK_MINOR_OFFSET; | |
1042 | break; | |
1043 | case XEN_SCSI_DISK8_MAJOR: | |
1044 | case XEN_SCSI_DISK9_MAJOR: | |
1045 | case XEN_SCSI_DISK10_MAJOR: | |
1046 | case XEN_SCSI_DISK11_MAJOR: | |
1047 | case XEN_SCSI_DISK12_MAJOR: | |
1048 | case XEN_SCSI_DISK13_MAJOR: | |
1049 | case XEN_SCSI_DISK14_MAJOR: | |
1050 | case XEN_SCSI_DISK15_MAJOR: | |
1051 | *offset = (*minor / PARTS_PER_DISK) + | |
1052 | ((major - XEN_SCSI_DISK8_MAJOR + 8) * 16) + | |
1053 | EMULATED_SD_DISK_NAME_OFFSET; | |
1054 | *minor = *minor + | |
1055 | ((major - XEN_SCSI_DISK8_MAJOR + 8) * 16 * PARTS_PER_DISK) + | |
1056 | EMULATED_SD_DISK_MINOR_OFFSET; | |
1057 | break; | |
1058 | case XENVBD_MAJOR: | |
1059 | *offset = *minor / PARTS_PER_DISK; | |
1060 | break; | |
1061 | default: | |
1062 | printk(KERN_WARNING "blkfront: your disk configuration is " | |
1063 | "incorrect, please use an xvd device instead\n"); | |
1064 | return -ENODEV; | |
1065 | } | |
1066 | return 0; | |
1067 | } | |
9f27ee59 | 1068 | |
e77c78c0 JB |
1069 | static char *encode_disk_name(char *ptr, unsigned int n) |
1070 | { | |
1071 | if (n >= 26) | |
1072 | ptr = encode_disk_name(ptr, n / 26 - 1); | |
1073 | *ptr = 'a' + n % 26; | |
1074 | return ptr + 1; | |
1075 | } | |
1076 | ||
9246b5f0 CL |
1077 | static int xlvbd_alloc_gendisk(blkif_sector_t capacity, |
1078 | struct blkfront_info *info, | |
7c4d7d71 SB |
1079 | u16 vdisk_info, u16 sector_size, |
1080 | unsigned int physical_sector_size) | |
9f27ee59 JF |
1081 | { |
1082 | struct gendisk *gd; | |
1083 | int nr_minors = 1; | |
c80a4209 | 1084 | int err; |
9246b5f0 CL |
1085 | unsigned int offset; |
1086 | int minor; | |
1087 | int nr_parts; | |
e77c78c0 | 1088 | char *ptr; |
9f27ee59 JF |
1089 | |
1090 | BUG_ON(info->gd != NULL); | |
1091 | BUG_ON(info->rq != NULL); | |
1092 | ||
9246b5f0 CL |
1093 | if ((info->vdevice>>EXT_SHIFT) > 1) { |
1094 | /* this is above the extended range; something is wrong */ | |
1095 | printk(KERN_WARNING "blkfront: vdevice 0x%x is above the extended range; ignoring\n", info->vdevice); | |
1096 | return -ENODEV; | |
1097 | } | |
1098 | ||
1099 | if (!VDEV_IS_EXTENDED(info->vdevice)) { | |
c80a4209 SS |
1100 | err = xen_translate_vdev(info->vdevice, &minor, &offset); |
1101 | if (err) | |
1102 | return err; | |
1103 | nr_parts = PARTS_PER_DISK; | |
9246b5f0 CL |
1104 | } else { |
1105 | minor = BLKIF_MINOR_EXT(info->vdevice); | |
1106 | nr_parts = PARTS_PER_EXT_DISK; | |
c80a4209 | 1107 | offset = minor / nr_parts; |
89153b5c | 1108 | if (xen_hvm_domain() && offset < EMULATED_HD_DISK_NAME_OFFSET + 4) |
c80a4209 SS |
1109 | printk(KERN_WARNING "blkfront: vdevice 0x%x might conflict with " |
1110 | "emulated IDE disks,\n\t choose an xvd device name" | |
1111 | "from xvde on\n", info->vdevice); | |
9246b5f0 | 1112 | } |
e77c78c0 JB |
1113 | if (minor >> MINORBITS) { |
1114 | pr_warn("blkfront: %#x's minor (%#x) out of range; ignoring\n", | |
1115 | info->vdevice, minor); | |
1116 | return -ENODEV; | |
1117 | } | |
9246b5f0 CL |
1118 | |
1119 | if ((minor % nr_parts) == 0) | |
1120 | nr_minors = nr_parts; | |
9f27ee59 | 1121 | |
0e345826 JB |
1122 | err = xlbd_reserve_minors(minor, nr_minors); |
1123 | if (err) | |
1124 | goto out; | |
1125 | err = -ENODEV; | |
1126 | ||
9f27ee59 JF |
1127 | gd = alloc_disk(nr_minors); |
1128 | if (gd == NULL) | |
0e345826 | 1129 | goto release; |
9f27ee59 | 1130 | |
e77c78c0 JB |
1131 | strcpy(gd->disk_name, DEV_NAME); |
1132 | ptr = encode_disk_name(gd->disk_name + sizeof(DEV_NAME) - 1, offset); | |
1133 | BUG_ON(ptr >= gd->disk_name + DISK_NAME_LEN); | |
1134 | if (nr_minors > 1) | |
1135 | *ptr = 0; | |
1136 | else | |
1137 | snprintf(ptr, gd->disk_name + DISK_NAME_LEN - ptr, | |
1138 | "%d", minor & (nr_parts - 1)); | |
9f27ee59 JF |
1139 | |
1140 | gd->major = XENVBD_MAJOR; | |
1141 | gd->first_minor = minor; | |
1142 | gd->fops = &xlvbd_block_fops; | |
1143 | gd->private_data = info; | |
1144 | gd->driverfs_dev = &(info->xbdev->dev); | |
1145 | set_capacity(gd, capacity); | |
1146 | ||
7c4d7d71 | 1147 | if (xlvbd_init_blk_queue(gd, sector_size, physical_sector_size, |
402b27f9 RPM |
1148 | info->max_indirect_segments ? : |
1149 | BLKIF_MAX_SEGMENTS_PER_REQUEST)) { | |
9f27ee59 | 1150 | del_gendisk(gd); |
0e345826 | 1151 | goto release; |
9f27ee59 JF |
1152 | } |
1153 | ||
1154 | info->rq = gd->queue; | |
1155 | info->gd = gd; | |
1156 | ||
4913efe4 | 1157 | xlvbd_flush(info); |
9f27ee59 JF |
1158 | |
1159 | if (vdisk_info & VDISK_READONLY) | |
1160 | set_disk_ro(gd, 1); | |
1161 | ||
1162 | if (vdisk_info & VDISK_REMOVABLE) | |
1163 | gd->flags |= GENHD_FL_REMOVABLE; | |
1164 | ||
1165 | if (vdisk_info & VDISK_CDROM) | |
1166 | gd->flags |= GENHD_FL_CD; | |
1167 | ||
1168 | return 0; | |
1169 | ||
0e345826 JB |
1170 | release: |
1171 | xlbd_release_minors(minor, nr_minors); | |
9f27ee59 JF |
1172 | out: |
1173 | return err; | |
1174 | } | |
1175 | ||
a66b5aeb DS |
1176 | static void xlvbd_release_gendisk(struct blkfront_info *info) |
1177 | { | |
3df0e505 | 1178 | unsigned int minor, nr_minors, i; |
a66b5aeb DS |
1179 | |
1180 | if (info->rq == NULL) | |
1181 | return; | |
1182 | ||
a66b5aeb | 1183 | /* No more blkif_request(). */ |
907c3eb1 | 1184 | blk_mq_stop_hw_queues(info->rq); |
a66b5aeb | 1185 | |
3df0e505 BL |
1186 | for (i = 0; i < info->nr_rings; i++) { |
1187 | struct blkfront_ring_info *rinfo = &info->rinfo[i]; | |
a66b5aeb | 1188 | |
3df0e505 BL |
1189 | /* No more gnttab callback work. */ |
1190 | gnttab_cancel_free_callback(&rinfo->callback); | |
1191 | ||
1192 | /* Flush gnttab callback work. Must be done with no locks held. */ | |
1193 | flush_work(&rinfo->work); | |
1194 | } | |
a66b5aeb DS |
1195 | |
1196 | del_gendisk(info->gd); | |
1197 | ||
1198 | minor = info->gd->first_minor; | |
1199 | nr_minors = info->gd->minors; | |
1200 | xlbd_release_minors(minor, nr_minors); | |
1201 | ||
1202 | blk_cleanup_queue(info->rq); | |
907c3eb1 | 1203 | blk_mq_free_tag_set(&info->tag_set); |
a66b5aeb DS |
1204 | info->rq = NULL; |
1205 | ||
1206 | put_disk(info->gd); | |
1207 | info->gd = NULL; | |
1208 | } | |
1209 | ||
11659569 BL |
1210 | /* Already hold rinfo->ring_lock. */ |
1211 | static inline void kick_pending_request_queues_locked(struct blkfront_ring_info *rinfo) | |
9f27ee59 | 1212 | { |
81f35161 BL |
1213 | if (!RING_FULL(&rinfo->ring)) |
1214 | blk_mq_start_stopped_hw_queues(rinfo->dev_info->rq, true); | |
9f27ee59 JF |
1215 | } |
1216 | ||
11659569 BL |
1217 | static void kick_pending_request_queues(struct blkfront_ring_info *rinfo) |
1218 | { | |
1219 | unsigned long flags; | |
1220 | ||
1221 | spin_lock_irqsave(&rinfo->ring_lock, flags); | |
1222 | kick_pending_request_queues_locked(rinfo); | |
1223 | spin_unlock_irqrestore(&rinfo->ring_lock, flags); | |
1224 | } | |
1225 | ||
9f27ee59 JF |
1226 | static void blkif_restart_queue(struct work_struct *work) |
1227 | { | |
81f35161 | 1228 | struct blkfront_ring_info *rinfo = container_of(work, struct blkfront_ring_info, work); |
9f27ee59 | 1229 | |
81f35161 BL |
1230 | if (rinfo->dev_info->connected == BLKIF_STATE_CONNECTED) |
1231 | kick_pending_request_queues(rinfo); | |
9f27ee59 JF |
1232 | } |
1233 | ||
3df0e505 | 1234 | static void blkif_free_ring(struct blkfront_ring_info *rinfo) |
9f27ee59 | 1235 | { |
73716df7 | 1236 | struct grant *persistent_gnt, *n; |
3df0e505 | 1237 | struct blkfront_info *info = rinfo->dev_info; |
402b27f9 | 1238 | int i, j, segs; |
0a8704a5 | 1239 | |
bfe11d6d RPM |
1240 | /* |
1241 | * Remove indirect pages, this only happens when using indirect | |
1242 | * descriptors but not persistent grants | |
1243 | */ | |
81f35161 | 1244 | if (!list_empty(&rinfo->indirect_pages)) { |
bfe11d6d RPM |
1245 | struct page *indirect_page, *n; |
1246 | ||
1247 | BUG_ON(info->feature_persistent); | |
81f35161 | 1248 | list_for_each_entry_safe(indirect_page, n, &rinfo->indirect_pages, lru) { |
bfe11d6d RPM |
1249 | list_del(&indirect_page->lru); |
1250 | __free_page(indirect_page); | |
1251 | } | |
1252 | } | |
1253 | ||
73716df7 BL |
1254 | /* Remove all persistent grants. */ |
1255 | if (!list_empty(&rinfo->grants)) { | |
1256 | list_for_each_entry_safe(persistent_gnt, n, | |
1257 | &rinfo->grants, node) { | |
1258 | list_del(&persistent_gnt->node); | |
1259 | if (persistent_gnt->gref != GRANT_INVALID_REF) { | |
1260 | gnttab_end_foreign_access(persistent_gnt->gref, | |
1261 | 0, 0UL); | |
1262 | rinfo->persistent_gnts_c--; | |
1263 | } | |
1264 | if (info->feature_persistent) | |
1265 | __free_page(persistent_gnt->page); | |
1266 | kfree(persistent_gnt); | |
1267 | } | |
1268 | } | |
1269 | BUG_ON(rinfo->persistent_gnts_c != 0); | |
1270 | ||
86839c56 | 1271 | for (i = 0; i < BLK_RING_SIZE(info); i++) { |
402b27f9 RPM |
1272 | /* |
1273 | * Clear persistent grants present in requests already | |
1274 | * on the shared ring | |
1275 | */ | |
81f35161 | 1276 | if (!rinfo->shadow[i].request) |
402b27f9 RPM |
1277 | goto free_shadow; |
1278 | ||
81f35161 BL |
1279 | segs = rinfo->shadow[i].req.operation == BLKIF_OP_INDIRECT ? |
1280 | rinfo->shadow[i].req.u.indirect.nr_segments : | |
1281 | rinfo->shadow[i].req.u.rw.nr_segments; | |
402b27f9 | 1282 | for (j = 0; j < segs; j++) { |
81f35161 | 1283 | persistent_gnt = rinfo->shadow[i].grants_used[j]; |
402b27f9 | 1284 | gnttab_end_foreign_access(persistent_gnt->gref, 0, 0UL); |
bfe11d6d | 1285 | if (info->feature_persistent) |
a7a6df22 | 1286 | __free_page(persistent_gnt->page); |
402b27f9 RPM |
1287 | kfree(persistent_gnt); |
1288 | } | |
1289 | ||
81f35161 | 1290 | if (rinfo->shadow[i].req.operation != BLKIF_OP_INDIRECT) |
402b27f9 RPM |
1291 | /* |
1292 | * If this is not an indirect operation don't try to | |
1293 | * free indirect segments | |
1294 | */ | |
1295 | goto free_shadow; | |
1296 | ||
1297 | for (j = 0; j < INDIRECT_GREFS(segs); j++) { | |
81f35161 | 1298 | persistent_gnt = rinfo->shadow[i].indirect_grants[j]; |
402b27f9 | 1299 | gnttab_end_foreign_access(persistent_gnt->gref, 0, 0UL); |
a7a6df22 | 1300 | __free_page(persistent_gnt->page); |
402b27f9 RPM |
1301 | kfree(persistent_gnt); |
1302 | } | |
1303 | ||
1304 | free_shadow: | |
81f35161 BL |
1305 | kfree(rinfo->shadow[i].grants_used); |
1306 | rinfo->shadow[i].grants_used = NULL; | |
1307 | kfree(rinfo->shadow[i].indirect_grants); | |
1308 | rinfo->shadow[i].indirect_grants = NULL; | |
1309 | kfree(rinfo->shadow[i].sg); | |
1310 | rinfo->shadow[i].sg = NULL; | |
402b27f9 RPM |
1311 | } |
1312 | ||
9f27ee59 | 1313 | /* No more gnttab callback work. */ |
81f35161 | 1314 | gnttab_cancel_free_callback(&rinfo->callback); |
9f27ee59 JF |
1315 | |
1316 | /* Flush gnttab callback work. Must be done with no locks held. */ | |
81f35161 | 1317 | flush_work(&rinfo->work); |
9f27ee59 JF |
1318 | |
1319 | /* Free resources associated with old device channel. */ | |
86839c56 | 1320 | for (i = 0; i < info->nr_ring_pages; i++) { |
81f35161 BL |
1321 | if (rinfo->ring_ref[i] != GRANT_INVALID_REF) { |
1322 | gnttab_end_foreign_access(rinfo->ring_ref[i], 0, 0); | |
1323 | rinfo->ring_ref[i] = GRANT_INVALID_REF; | |
86839c56 | 1324 | } |
9f27ee59 | 1325 | } |
81f35161 BL |
1326 | free_pages((unsigned long)rinfo->ring.sring, get_order(info->nr_ring_pages * PAGE_SIZE)); |
1327 | rinfo->ring.sring = NULL; | |
86839c56 | 1328 | |
81f35161 BL |
1329 | if (rinfo->irq) |
1330 | unbind_from_irqhandler(rinfo->irq, rinfo); | |
1331 | rinfo->evtchn = rinfo->irq = 0; | |
3df0e505 | 1332 | } |
9f27ee59 | 1333 | |
3df0e505 BL |
1334 | static void blkif_free(struct blkfront_info *info, int suspend) |
1335 | { | |
3df0e505 BL |
1336 | unsigned int i; |
1337 | ||
1338 | /* Prevent new requests being issued until we fix things up. */ | |
3df0e505 BL |
1339 | info->connected = suspend ? |
1340 | BLKIF_STATE_SUSPENDED : BLKIF_STATE_DISCONNECTED; | |
1341 | /* No more blkif_request(). */ | |
1342 | if (info->rq) | |
1343 | blk_mq_stop_hw_queues(info->rq); | |
1344 | ||
3df0e505 BL |
1345 | for (i = 0; i < info->nr_rings; i++) |
1346 | blkif_free_ring(&info->rinfo[i]); | |
1347 | ||
1348 | kfree(info->rinfo); | |
1349 | info->rinfo = NULL; | |
1350 | info->nr_rings = 0; | |
9f27ee59 JF |
1351 | } |
1352 | ||
c004a6fe JG |
1353 | struct copy_from_grant { |
1354 | const struct blk_shadow *s; | |
1355 | unsigned int grant_idx; | |
1356 | unsigned int bvec_offset; | |
1357 | char *bvec_data; | |
1358 | }; | |
1359 | ||
1360 | static void blkif_copy_from_grant(unsigned long gfn, unsigned int offset, | |
1361 | unsigned int len, void *data) | |
1362 | { | |
1363 | struct copy_from_grant *info = data; | |
1364 | char *shared_data; | |
1365 | /* Convenient aliases */ | |
1366 | const struct blk_shadow *s = info->s; | |
1367 | ||
1368 | shared_data = kmap_atomic(s->grants_used[info->grant_idx]->page); | |
1369 | ||
1370 | memcpy(info->bvec_data + info->bvec_offset, | |
1371 | shared_data + offset, len); | |
1372 | ||
1373 | info->bvec_offset += len; | |
1374 | info->grant_idx++; | |
1375 | ||
1376 | kunmap_atomic(shared_data); | |
1377 | } | |
1378 | ||
6cc56833 JG |
1379 | static enum blk_req_status blkif_rsp_to_req_status(int rsp) |
1380 | { | |
1381 | switch (rsp) | |
1382 | { | |
1383 | case BLKIF_RSP_OKAY: | |
1384 | return REQ_DONE; | |
1385 | case BLKIF_RSP_EOPNOTSUPP: | |
1386 | return REQ_EOPNOTSUPP; | |
1387 | case BLKIF_RSP_ERROR: | |
1388 | /* Fallthrough. */ | |
1389 | default: | |
1390 | return REQ_ERROR; | |
1391 | } | |
1392 | } | |
1393 | ||
1394 | /* | |
1395 | * Get the final status of the block request based on two ring response | |
1396 | */ | |
1397 | static int blkif_get_final_status(enum blk_req_status s1, | |
1398 | enum blk_req_status s2) | |
1399 | { | |
1400 | BUG_ON(s1 == REQ_WAITING); | |
1401 | BUG_ON(s2 == REQ_WAITING); | |
1402 | ||
1403 | if (s1 == REQ_ERROR || s2 == REQ_ERROR) | |
1404 | return BLKIF_RSP_ERROR; | |
1405 | else if (s1 == REQ_EOPNOTSUPP || s2 == REQ_EOPNOTSUPP) | |
1406 | return BLKIF_RSP_EOPNOTSUPP; | |
1407 | return BLKIF_RSP_OKAY; | |
1408 | } | |
1409 | ||
1410 | static bool blkif_completion(unsigned long *id, | |
1411 | struct blkfront_ring_info *rinfo, | |
0a8704a5 | 1412 | struct blkif_response *bret) |
9f27ee59 | 1413 | { |
d62f6918 | 1414 | int i = 0; |
b7649158 | 1415 | struct scatterlist *sg; |
c004a6fe | 1416 | int num_sg, num_grant; |
81f35161 | 1417 | struct blkfront_info *info = rinfo->dev_info; |
6cc56833 | 1418 | struct blk_shadow *s = &rinfo->shadow[*id]; |
c004a6fe | 1419 | struct copy_from_grant data = { |
c004a6fe JG |
1420 | .grant_idx = 0, |
1421 | }; | |
402b27f9 | 1422 | |
c004a6fe | 1423 | num_grant = s->req.operation == BLKIF_OP_INDIRECT ? |
402b27f9 | 1424 | s->req.u.indirect.nr_segments : s->req.u.rw.nr_segments; |
6cc56833 JG |
1425 | |
1426 | /* The I/O request may be split in two. */ | |
1427 | if (unlikely(s->associated_id != NO_ASSOCIATED_ID)) { | |
1428 | struct blk_shadow *s2 = &rinfo->shadow[s->associated_id]; | |
1429 | ||
1430 | /* Keep the status of the current response in shadow. */ | |
1431 | s->status = blkif_rsp_to_req_status(bret->status); | |
1432 | ||
1433 | /* Wait the second response if not yet here. */ | |
1434 | if (s2->status == REQ_WAITING) | |
1435 | return 0; | |
1436 | ||
1437 | bret->status = blkif_get_final_status(s->status, | |
1438 | s2->status); | |
1439 | ||
1440 | /* | |
1441 | * All the grants is stored in the first shadow in order | |
1442 | * to make the completion code simpler. | |
1443 | */ | |
1444 | num_grant += s2->req.u.rw.nr_segments; | |
1445 | ||
1446 | /* | |
1447 | * The two responses may not come in order. Only the | |
1448 | * first request will store the scatter-gather list. | |
1449 | */ | |
1450 | if (s2->num_sg != 0) { | |
1451 | /* Update "id" with the ID of the first response. */ | |
1452 | *id = s->associated_id; | |
1453 | s = s2; | |
1454 | } | |
1455 | ||
1456 | /* | |
1457 | * We don't need anymore the second request, so recycling | |
1458 | * it now. | |
1459 | */ | |
1460 | if (add_id_to_freelist(rinfo, s->associated_id)) | |
1461 | WARN(1, "%s: can't recycle the second part (id = %ld) of the request\n", | |
1462 | info->gd->disk_name, s->associated_id); | |
1463 | } | |
1464 | ||
1465 | data.s = s; | |
c004a6fe | 1466 | num_sg = s->num_sg; |
0a8704a5 | 1467 | |
bfe11d6d | 1468 | if (bret->operation == BLKIF_OP_READ && info->feature_persistent) { |
c004a6fe | 1469 | for_each_sg(s->sg, sg, num_sg, i) { |
b7649158 | 1470 | BUG_ON(sg->offset + sg->length > PAGE_SIZE); |
c004a6fe JG |
1471 | |
1472 | data.bvec_offset = sg->offset; | |
1473 | data.bvec_data = kmap_atomic(sg_page(sg)); | |
1474 | ||
1475 | gnttab_foreach_grant_in_range(sg_page(sg), | |
1476 | sg->offset, | |
1477 | sg->length, | |
1478 | blkif_copy_from_grant, | |
1479 | &data); | |
1480 | ||
1481 | kunmap_atomic(data.bvec_data); | |
0a8704a5 RPM |
1482 | } |
1483 | } | |
1484 | /* Add the persistent grant into the list of free grants */ | |
c004a6fe | 1485 | for (i = 0; i < num_grant; i++) { |
fbe363c4 RPM |
1486 | if (gnttab_query_foreign_access(s->grants_used[i]->gref)) { |
1487 | /* | |
1488 | * If the grant is still mapped by the backend (the | |
1489 | * backend has chosen to make this grant persistent) | |
1490 | * we add it at the head of the list, so it will be | |
1491 | * reused first. | |
1492 | */ | |
bfe11d6d RPM |
1493 | if (!info->feature_persistent) |
1494 | pr_alert_ratelimited("backed has not unmapped grant: %u\n", | |
1495 | s->grants_used[i]->gref); | |
73716df7 BL |
1496 | list_add(&s->grants_used[i]->node, &rinfo->grants); |
1497 | rinfo->persistent_gnts_c++; | |
fbe363c4 RPM |
1498 | } else { |
1499 | /* | |
1500 | * If the grant is not mapped by the backend we end the | |
1501 | * foreign access and add it to the tail of the list, | |
1502 | * so it will not be picked again unless we run out of | |
1503 | * persistent grants. | |
1504 | */ | |
1505 | gnttab_end_foreign_access(s->grants_used[i]->gref, 0, 0UL); | |
1506 | s->grants_used[i]->gref = GRANT_INVALID_REF; | |
73716df7 | 1507 | list_add_tail(&s->grants_used[i]->node, &rinfo->grants); |
fbe363c4 | 1508 | } |
0a8704a5 | 1509 | } |
402b27f9 | 1510 | if (s->req.operation == BLKIF_OP_INDIRECT) { |
c004a6fe | 1511 | for (i = 0; i < INDIRECT_GREFS(num_grant); i++) { |
fbe363c4 | 1512 | if (gnttab_query_foreign_access(s->indirect_grants[i]->gref)) { |
bfe11d6d RPM |
1513 | if (!info->feature_persistent) |
1514 | pr_alert_ratelimited("backed has not unmapped grant: %u\n", | |
1515 | s->indirect_grants[i]->gref); | |
73716df7 BL |
1516 | list_add(&s->indirect_grants[i]->node, &rinfo->grants); |
1517 | rinfo->persistent_gnts_c++; | |
fbe363c4 | 1518 | } else { |
bfe11d6d RPM |
1519 | struct page *indirect_page; |
1520 | ||
fbe363c4 | 1521 | gnttab_end_foreign_access(s->indirect_grants[i]->gref, 0, 0UL); |
bfe11d6d RPM |
1522 | /* |
1523 | * Add the used indirect page back to the list of | |
1524 | * available pages for indirect grefs. | |
1525 | */ | |
7b076750 | 1526 | if (!info->feature_persistent) { |
a7a6df22 | 1527 | indirect_page = s->indirect_grants[i]->page; |
81f35161 | 1528 | list_add(&indirect_page->lru, &rinfo->indirect_pages); |
7b076750 | 1529 | } |
fbe363c4 | 1530 | s->indirect_grants[i]->gref = GRANT_INVALID_REF; |
73716df7 | 1531 | list_add_tail(&s->indirect_grants[i]->node, &rinfo->grants); |
fbe363c4 | 1532 | } |
402b27f9 RPM |
1533 | } |
1534 | } | |
6cc56833 JG |
1535 | |
1536 | return 1; | |
9f27ee59 JF |
1537 | } |
1538 | ||
1539 | static irqreturn_t blkif_interrupt(int irq, void *dev_id) | |
1540 | { | |
1541 | struct request *req; | |
1542 | struct blkif_response *bret; | |
1543 | RING_IDX i, rp; | |
1544 | unsigned long flags; | |
81f35161 BL |
1545 | struct blkfront_ring_info *rinfo = (struct blkfront_ring_info *)dev_id; |
1546 | struct blkfront_info *info = rinfo->dev_info; | |
f4829a9b | 1547 | int error; |
9f27ee59 | 1548 | |
11659569 | 1549 | if (unlikely(info->connected != BLKIF_STATE_CONNECTED)) |
9f27ee59 | 1550 | return IRQ_HANDLED; |
9f27ee59 | 1551 | |
11659569 | 1552 | spin_lock_irqsave(&rinfo->ring_lock, flags); |
9f27ee59 | 1553 | again: |
81f35161 | 1554 | rp = rinfo->ring.sring->rsp_prod; |
9f27ee59 JF |
1555 | rmb(); /* Ensure we see queued responses up to 'rp'. */ |
1556 | ||
81f35161 | 1557 | for (i = rinfo->ring.rsp_cons; i != rp; i++) { |
9f27ee59 | 1558 | unsigned long id; |
9f27ee59 | 1559 | |
81f35161 | 1560 | bret = RING_GET_RESPONSE(&rinfo->ring, i); |
9f27ee59 | 1561 | id = bret->id; |
6878c32e KRW |
1562 | /* |
1563 | * The backend has messed up and given us an id that we would | |
1564 | * never have given to it (we stamp it up to BLK_RING_SIZE - | |
1565 | * look in get_id_from_freelist. | |
1566 | */ | |
86839c56 | 1567 | if (id >= BLK_RING_SIZE(info)) { |
6878c32e KRW |
1568 | WARN(1, "%s: response to %s has incorrect id (%ld)\n", |
1569 | info->gd->disk_name, op_name(bret->operation), id); | |
1570 | /* We can't safely get the 'struct request' as | |
1571 | * the id is busted. */ | |
1572 | continue; | |
1573 | } | |
81f35161 | 1574 | req = rinfo->shadow[id].request; |
9f27ee59 | 1575 | |
6cc56833 JG |
1576 | if (bret->operation != BLKIF_OP_DISCARD) { |
1577 | /* | |
1578 | * We may need to wait for an extra response if the | |
1579 | * I/O request is split in 2 | |
1580 | */ | |
1581 | if (!blkif_completion(&id, rinfo, bret)) | |
1582 | continue; | |
1583 | } | |
9f27ee59 | 1584 | |
81f35161 | 1585 | if (add_id_to_freelist(rinfo, id)) { |
6878c32e KRW |
1586 | WARN(1, "%s: response to %s (id %ld) couldn't be recycled!\n", |
1587 | info->gd->disk_name, op_name(bret->operation), id); | |
1588 | continue; | |
1589 | } | |
9f27ee59 | 1590 | |
f4829a9b | 1591 | error = (bret->status == BLKIF_RSP_OKAY) ? 0 : -EIO; |
9f27ee59 | 1592 | switch (bret->operation) { |
ed30bf31 LD |
1593 | case BLKIF_OP_DISCARD: |
1594 | if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) { | |
1595 | struct request_queue *rq = info->rq; | |
6878c32e KRW |
1596 | printk(KERN_WARNING "blkfront: %s: %s op failed\n", |
1597 | info->gd->disk_name, op_name(bret->operation)); | |
f4829a9b | 1598 | error = -EOPNOTSUPP; |
ed30bf31 | 1599 | info->feature_discard = 0; |
5ea42986 | 1600 | info->feature_secdiscard = 0; |
ed30bf31 | 1601 | queue_flag_clear(QUEUE_FLAG_DISCARD, rq); |
5ea42986 | 1602 | queue_flag_clear(QUEUE_FLAG_SECDISCARD, rq); |
ed30bf31 | 1603 | } |
f4829a9b | 1604 | blk_mq_complete_request(req, error); |
ed30bf31 | 1605 | break; |
edf6ef59 | 1606 | case BLKIF_OP_FLUSH_DISKCACHE: |
9f27ee59 JF |
1607 | case BLKIF_OP_WRITE_BARRIER: |
1608 | if (unlikely(bret->status == BLKIF_RSP_EOPNOTSUPP)) { | |
6878c32e KRW |
1609 | printk(KERN_WARNING "blkfront: %s: %s op failed\n", |
1610 | info->gd->disk_name, op_name(bret->operation)); | |
f4829a9b | 1611 | error = -EOPNOTSUPP; |
dcb8baec JF |
1612 | } |
1613 | if (unlikely(bret->status == BLKIF_RSP_ERROR && | |
81f35161 | 1614 | rinfo->shadow[id].req.u.rw.nr_segments == 0)) { |
6878c32e KRW |
1615 | printk(KERN_WARNING "blkfront: %s: empty %s op failed\n", |
1616 | info->gd->disk_name, op_name(bret->operation)); | |
f4829a9b | 1617 | error = -EOPNOTSUPP; |
dcb8baec | 1618 | } |
f4829a9b CH |
1619 | if (unlikely(error)) { |
1620 | if (error == -EOPNOTSUPP) | |
1621 | error = 0; | |
4913efe4 TH |
1622 | info->feature_flush = 0; |
1623 | xlvbd_flush(info); | |
9f27ee59 JF |
1624 | } |
1625 | /* fall through */ | |
1626 | case BLKIF_OP_READ: | |
1627 | case BLKIF_OP_WRITE: | |
1628 | if (unlikely(bret->status != BLKIF_RSP_OKAY)) | |
1629 | dev_dbg(&info->xbdev->dev, "Bad return from blkdev data " | |
1630 | "request: %x\n", bret->status); | |
1631 | ||
f4829a9b | 1632 | blk_mq_complete_request(req, error); |
9f27ee59 JF |
1633 | break; |
1634 | default: | |
1635 | BUG(); | |
1636 | } | |
1637 | } | |
1638 | ||
81f35161 | 1639 | rinfo->ring.rsp_cons = i; |
9f27ee59 | 1640 | |
81f35161 | 1641 | if (i != rinfo->ring.req_prod_pvt) { |
9f27ee59 | 1642 | int more_to_do; |
81f35161 | 1643 | RING_FINAL_CHECK_FOR_RESPONSES(&rinfo->ring, more_to_do); |
9f27ee59 JF |
1644 | if (more_to_do) |
1645 | goto again; | |
1646 | } else | |
81f35161 | 1647 | rinfo->ring.sring->rsp_event = i + 1; |
9f27ee59 | 1648 | |
11659569 | 1649 | kick_pending_request_queues_locked(rinfo); |
9f27ee59 | 1650 | |
11659569 | 1651 | spin_unlock_irqrestore(&rinfo->ring_lock, flags); |
9f27ee59 JF |
1652 | |
1653 | return IRQ_HANDLED; | |
1654 | } | |
1655 | ||
1656 | ||
1657 | static int setup_blkring(struct xenbus_device *dev, | |
81f35161 | 1658 | struct blkfront_ring_info *rinfo) |
9f27ee59 JF |
1659 | { |
1660 | struct blkif_sring *sring; | |
86839c56 | 1661 | int err, i; |
81f35161 | 1662 | struct blkfront_info *info = rinfo->dev_info; |
c004a6fe | 1663 | unsigned long ring_size = info->nr_ring_pages * XEN_PAGE_SIZE; |
9cce2914 | 1664 | grant_ref_t gref[XENBUS_MAX_RING_GRANTS]; |
9f27ee59 | 1665 | |
86839c56 | 1666 | for (i = 0; i < info->nr_ring_pages; i++) |
81f35161 | 1667 | rinfo->ring_ref[i] = GRANT_INVALID_REF; |
9f27ee59 | 1668 | |
86839c56 BL |
1669 | sring = (struct blkif_sring *)__get_free_pages(GFP_NOIO | __GFP_HIGH, |
1670 | get_order(ring_size)); | |
9f27ee59 JF |
1671 | if (!sring) { |
1672 | xenbus_dev_fatal(dev, -ENOMEM, "allocating shared ring"); | |
1673 | return -ENOMEM; | |
1674 | } | |
1675 | SHARED_RING_INIT(sring); | |
81f35161 | 1676 | FRONT_RING_INIT(&rinfo->ring, sring, ring_size); |
9e973e64 | 1677 | |
81f35161 | 1678 | err = xenbus_grant_ring(dev, rinfo->ring.sring, info->nr_ring_pages, gref); |
9f27ee59 | 1679 | if (err < 0) { |
86839c56 | 1680 | free_pages((unsigned long)sring, get_order(ring_size)); |
81f35161 | 1681 | rinfo->ring.sring = NULL; |
9f27ee59 JF |
1682 | goto fail; |
1683 | } | |
86839c56 | 1684 | for (i = 0; i < info->nr_ring_pages; i++) |
81f35161 | 1685 | rinfo->ring_ref[i] = gref[i]; |
9f27ee59 | 1686 | |
81f35161 | 1687 | err = xenbus_alloc_evtchn(dev, &rinfo->evtchn); |
9f27ee59 JF |
1688 | if (err) |
1689 | goto fail; | |
1690 | ||
81f35161 BL |
1691 | err = bind_evtchn_to_irqhandler(rinfo->evtchn, blkif_interrupt, 0, |
1692 | "blkif", rinfo); | |
9f27ee59 JF |
1693 | if (err <= 0) { |
1694 | xenbus_dev_fatal(dev, err, | |
1695 | "bind_evtchn_to_irqhandler failed"); | |
1696 | goto fail; | |
1697 | } | |
81f35161 | 1698 | rinfo->irq = err; |
9f27ee59 JF |
1699 | |
1700 | return 0; | |
1701 | fail: | |
1702 | blkif_free(info, 0); | |
1703 | return err; | |
1704 | } | |
1705 | ||
28d949bc BL |
1706 | /* |
1707 | * Write out per-ring/queue nodes including ring-ref and event-channel, and each | |
1708 | * ring buffer may have multi pages depending on ->nr_ring_pages. | |
1709 | */ | |
1710 | static int write_per_ring_nodes(struct xenbus_transaction xbt, | |
1711 | struct blkfront_ring_info *rinfo, const char *dir) | |
1712 | { | |
1713 | int err; | |
1714 | unsigned int i; | |
1715 | const char *message = NULL; | |
1716 | struct blkfront_info *info = rinfo->dev_info; | |
1717 | ||
1718 | if (info->nr_ring_pages == 1) { | |
1719 | err = xenbus_printf(xbt, dir, "ring-ref", "%u", rinfo->ring_ref[0]); | |
1720 | if (err) { | |
1721 | message = "writing ring-ref"; | |
1722 | goto abort_transaction; | |
1723 | } | |
1724 | } else { | |
1725 | for (i = 0; i < info->nr_ring_pages; i++) { | |
1726 | char ring_ref_name[RINGREF_NAME_LEN]; | |
1727 | ||
1728 | snprintf(ring_ref_name, RINGREF_NAME_LEN, "ring-ref%u", i); | |
1729 | err = xenbus_printf(xbt, dir, ring_ref_name, | |
1730 | "%u", rinfo->ring_ref[i]); | |
1731 | if (err) { | |
1732 | message = "writing ring-ref"; | |
1733 | goto abort_transaction; | |
1734 | } | |
1735 | } | |
1736 | } | |
1737 | ||
1738 | err = xenbus_printf(xbt, dir, "event-channel", "%u", rinfo->evtchn); | |
1739 | if (err) { | |
1740 | message = "writing event-channel"; | |
1741 | goto abort_transaction; | |
1742 | } | |
1743 | ||
1744 | return 0; | |
1745 | ||
1746 | abort_transaction: | |
1747 | xenbus_transaction_end(xbt, 1); | |
1748 | if (message) | |
1749 | xenbus_dev_fatal(info->xbdev, err, "%s", message); | |
1750 | ||
1751 | return err; | |
1752 | } | |
9f27ee59 JF |
1753 | |
1754 | /* Common code used when first setting up, and when resuming. */ | |
203fd61f | 1755 | static int talk_to_blkback(struct xenbus_device *dev, |
9f27ee59 JF |
1756 | struct blkfront_info *info) |
1757 | { | |
1758 | const char *message = NULL; | |
1759 | struct xenbus_transaction xbt; | |
28d949bc BL |
1760 | int err; |
1761 | unsigned int i, max_page_order = 0; | |
86839c56 BL |
1762 | unsigned int ring_page_order = 0; |
1763 | ||
1764 | err = xenbus_scanf(XBT_NIL, info->xbdev->otherend, | |
1765 | "max-ring-page-order", "%u", &max_page_order); | |
1766 | if (err != 1) | |
1767 | info->nr_ring_pages = 1; | |
1768 | else { | |
1769 | ring_page_order = min(xen_blkif_max_ring_order, max_page_order); | |
1770 | info->nr_ring_pages = 1 << ring_page_order; | |
1771 | } | |
9f27ee59 | 1772 | |
3df0e505 | 1773 | for (i = 0; i < info->nr_rings; i++) { |
28d949bc BL |
1774 | struct blkfront_ring_info *rinfo = &info->rinfo[i]; |
1775 | ||
3df0e505 BL |
1776 | /* Create shared ring, alloc event channel. */ |
1777 | err = setup_blkring(dev, rinfo); | |
1778 | if (err) | |
1779 | goto destroy_blkring; | |
1780 | } | |
9f27ee59 JF |
1781 | |
1782 | again: | |
1783 | err = xenbus_transaction_start(&xbt); | |
1784 | if (err) { | |
1785 | xenbus_dev_fatal(dev, err, "starting transaction"); | |
1786 | goto destroy_blkring; | |
1787 | } | |
1788 | ||
28d949bc BL |
1789 | if (info->nr_ring_pages > 1) { |
1790 | err = xenbus_printf(xbt, dev->nodename, "ring-page-order", "%u", | |
1791 | ring_page_order); | |
1792 | if (err) { | |
1793 | message = "writing ring-page-order"; | |
1794 | goto abort_transaction; | |
1795 | } | |
1796 | } | |
3df0e505 | 1797 | |
28d949bc BL |
1798 | /* We already got the number of queues/rings in _probe */ |
1799 | if (info->nr_rings == 1) { | |
1800 | err = write_per_ring_nodes(xbt, &info->rinfo[0], dev->nodename); | |
1801 | if (err) | |
1802 | goto destroy_blkring; | |
1803 | } else { | |
1804 | char *path; | |
1805 | size_t pathsize; | |
3df0e505 | 1806 | |
28d949bc BL |
1807 | err = xenbus_printf(xbt, dev->nodename, "multi-queue-num-queues", "%u", |
1808 | info->nr_rings); | |
3df0e505 | 1809 | if (err) { |
28d949bc | 1810 | message = "writing multi-queue-num-queues"; |
3df0e505 BL |
1811 | goto abort_transaction; |
1812 | } | |
28d949bc BL |
1813 | |
1814 | pathsize = strlen(dev->nodename) + QUEUE_NAME_LEN; | |
1815 | path = kmalloc(pathsize, GFP_KERNEL); | |
1816 | if (!path) { | |
1817 | err = -ENOMEM; | |
1818 | message = "ENOMEM while writing ring references"; | |
1819 | goto abort_transaction; | |
1820 | } | |
1821 | ||
1822 | for (i = 0; i < info->nr_rings; i++) { | |
1823 | memset(path, 0, pathsize); | |
1824 | snprintf(path, pathsize, "%s/queue-%u", dev->nodename, i); | |
1825 | err = write_per_ring_nodes(xbt, &info->rinfo[i], path); | |
1826 | if (err) { | |
1827 | kfree(path); | |
1828 | goto destroy_blkring; | |
1829 | } | |
1830 | } | |
1831 | kfree(path); | |
9f27ee59 | 1832 | } |
3e334239 MA |
1833 | err = xenbus_printf(xbt, dev->nodename, "protocol", "%s", |
1834 | XEN_IO_PROTO_ABI_NATIVE); | |
1835 | if (err) { | |
1836 | message = "writing protocol"; | |
1837 | goto abort_transaction; | |
1838 | } | |
0a8704a5 | 1839 | err = xenbus_printf(xbt, dev->nodename, |
cb5bd4d1 | 1840 | "feature-persistent", "%u", 1); |
0a8704a5 RPM |
1841 | if (err) |
1842 | dev_warn(&dev->dev, | |
1843 | "writing persistent grants feature to xenbus"); | |
9f27ee59 JF |
1844 | |
1845 | err = xenbus_transaction_end(xbt, 0); | |
1846 | if (err) { | |
1847 | if (err == -EAGAIN) | |
1848 | goto again; | |
1849 | xenbus_dev_fatal(dev, err, "completing transaction"); | |
1850 | goto destroy_blkring; | |
1851 | } | |
1852 | ||
3df0e505 BL |
1853 | for (i = 0; i < info->nr_rings; i++) { |
1854 | unsigned int j; | |
28d949bc | 1855 | struct blkfront_ring_info *rinfo = &info->rinfo[i]; |
3df0e505 BL |
1856 | |
1857 | for (j = 0; j < BLK_RING_SIZE(info); j++) | |
1858 | rinfo->shadow[j].req.u.rw.id = j + 1; | |
1859 | rinfo->shadow[BLK_RING_SIZE(info)-1].req.u.rw.id = 0x0fffffff; | |
1860 | } | |
9f27ee59 JF |
1861 | xenbus_switch_state(dev, XenbusStateInitialised); |
1862 | ||
1863 | return 0; | |
1864 | ||
1865 | abort_transaction: | |
1866 | xenbus_transaction_end(xbt, 1); | |
1867 | if (message) | |
1868 | xenbus_dev_fatal(dev, err, "%s", message); | |
1869 | destroy_blkring: | |
1870 | blkif_free(info, 0); | |
3df0e505 | 1871 | |
c31ecf6c KRW |
1872 | kfree(info); |
1873 | dev_set_drvdata(&dev->dev, NULL); | |
1874 | ||
9f27ee59 JF |
1875 | return err; |
1876 | } | |
1877 | ||
3db70a85 BL |
1878 | static int negotiate_mq(struct blkfront_info *info) |
1879 | { | |
1880 | unsigned int backend_max_queues = 0; | |
1881 | int err; | |
1882 | unsigned int i; | |
1883 | ||
1884 | BUG_ON(info->nr_rings); | |
1885 | ||
1886 | /* Check if backend supports multiple queues. */ | |
1887 | err = xenbus_scanf(XBT_NIL, info->xbdev->otherend, | |
1888 | "multi-queue-max-queues", "%u", &backend_max_queues); | |
1889 | if (err < 0) | |
1890 | backend_max_queues = 1; | |
1891 | ||
1892 | info->nr_rings = min(backend_max_queues, xen_blkif_max_queues); | |
1893 | /* We need at least one ring. */ | |
1894 | if (!info->nr_rings) | |
1895 | info->nr_rings = 1; | |
1896 | ||
1897 | info->rinfo = kzalloc(sizeof(struct blkfront_ring_info) * info->nr_rings, GFP_KERNEL); | |
1898 | if (!info->rinfo) { | |
1899 | xenbus_dev_fatal(info->xbdev, -ENOMEM, "allocating ring_info structure"); | |
1900 | return -ENOMEM; | |
1901 | } | |
1902 | ||
1903 | for (i = 0; i < info->nr_rings; i++) { | |
1904 | struct blkfront_ring_info *rinfo; | |
1905 | ||
1906 | rinfo = &info->rinfo[i]; | |
1907 | INIT_LIST_HEAD(&rinfo->indirect_pages); | |
1908 | INIT_LIST_HEAD(&rinfo->grants); | |
1909 | rinfo->dev_info = info; | |
1910 | INIT_WORK(&rinfo->work, blkif_restart_queue); | |
1911 | spin_lock_init(&rinfo->ring_lock); | |
1912 | } | |
1913 | return 0; | |
1914 | } | |
9f27ee59 JF |
1915 | /** |
1916 | * Entry point to this code when a new device is created. Allocate the basic | |
1917 | * structures and the ring buffer for communication with the backend, and | |
1918 | * inform the backend of the appropriate details for those. Switch to | |
1919 | * Initialised state. | |
1920 | */ | |
1921 | static int blkfront_probe(struct xenbus_device *dev, | |
1922 | const struct xenbus_device_id *id) | |
1923 | { | |
86839c56 | 1924 | int err, vdevice; |
9f27ee59 JF |
1925 | struct blkfront_info *info; |
1926 | ||
1927 | /* FIXME: Use dynamic device id if this is not set. */ | |
1928 | err = xenbus_scanf(XBT_NIL, dev->nodename, | |
1929 | "virtual-device", "%i", &vdevice); | |
1930 | if (err != 1) { | |
9246b5f0 CL |
1931 | /* go looking in the extended area instead */ |
1932 | err = xenbus_scanf(XBT_NIL, dev->nodename, "virtual-device-ext", | |
1933 | "%i", &vdevice); | |
1934 | if (err != 1) { | |
1935 | xenbus_dev_fatal(dev, err, "reading virtual-device"); | |
1936 | return err; | |
1937 | } | |
9f27ee59 JF |
1938 | } |
1939 | ||
b98a409b SS |
1940 | if (xen_hvm_domain()) { |
1941 | char *type; | |
1942 | int len; | |
1943 | /* no unplug has been done: do not hook devices != xen vbds */ | |
51c71a3b | 1944 | if (xen_has_pv_and_legacy_disk_devices()) { |
b98a409b SS |
1945 | int major; |
1946 | ||
1947 | if (!VDEV_IS_EXTENDED(vdevice)) | |
1948 | major = BLKIF_MAJOR(vdevice); | |
1949 | else | |
1950 | major = XENVBD_MAJOR; | |
1951 | ||
1952 | if (major != XENVBD_MAJOR) { | |
1953 | printk(KERN_INFO | |
1954 | "%s: HVM does not support vbd %d as xen block device\n", | |
02f1f217 | 1955 | __func__, vdevice); |
b98a409b SS |
1956 | return -ENODEV; |
1957 | } | |
1958 | } | |
1959 | /* do not create a PV cdrom device if we are an HVM guest */ | |
1960 | type = xenbus_read(XBT_NIL, dev->nodename, "device-type", &len); | |
1961 | if (IS_ERR(type)) | |
1962 | return -ENODEV; | |
1963 | if (strncmp(type, "cdrom", 5) == 0) { | |
1964 | kfree(type); | |
c1c5413a SS |
1965 | return -ENODEV; |
1966 | } | |
b98a409b | 1967 | kfree(type); |
c1c5413a | 1968 | } |
9f27ee59 JF |
1969 | info = kzalloc(sizeof(*info), GFP_KERNEL); |
1970 | if (!info) { | |
1971 | xenbus_dev_fatal(dev, -ENOMEM, "allocating info structure"); | |
1972 | return -ENOMEM; | |
1973 | } | |
1974 | ||
28d949bc | 1975 | info->xbdev = dev; |
3db70a85 BL |
1976 | err = negotiate_mq(info); |
1977 | if (err) { | |
3df0e505 | 1978 | kfree(info); |
3db70a85 | 1979 | return err; |
3df0e505 | 1980 | } |
81f35161 | 1981 | |
b70f5fa0 | 1982 | mutex_init(&info->mutex); |
9f27ee59 JF |
1983 | info->vdevice = vdevice; |
1984 | info->connected = BLKIF_STATE_DISCONNECTED; | |
9f27ee59 | 1985 | |
9f27ee59 JF |
1986 | /* Front end dir is a number, which is used as the id. */ |
1987 | info->handle = simple_strtoul(strrchr(dev->nodename, '/')+1, NULL, 0); | |
a1b4b12b | 1988 | dev_set_drvdata(&dev->dev, info); |
9f27ee59 | 1989 | |
9f27ee59 JF |
1990 | return 0; |
1991 | } | |
1992 | ||
4246a0b6 | 1993 | static void split_bio_end(struct bio *bio) |
402b27f9 RPM |
1994 | { |
1995 | struct split_bio *split_bio = bio->bi_private; | |
1996 | ||
402b27f9 RPM |
1997 | if (atomic_dec_and_test(&split_bio->pending)) { |
1998 | split_bio->bio->bi_phys_segments = 0; | |
4246a0b6 CH |
1999 | split_bio->bio->bi_error = bio->bi_error; |
2000 | bio_endio(split_bio->bio); | |
402b27f9 RPM |
2001 | kfree(split_bio); |
2002 | } | |
2003 | bio_put(bio); | |
2004 | } | |
9f27ee59 JF |
2005 | |
2006 | static int blkif_recover(struct blkfront_info *info) | |
2007 | { | |
3df0e505 | 2008 | unsigned int i, r_index; |
402b27f9 | 2009 | struct request *req, *n; |
9f27ee59 | 2010 | struct blk_shadow *copy; |
402b27f9 RPM |
2011 | int rc; |
2012 | struct bio *bio, *cloned_bio; | |
2013 | struct bio_list bio_list, merge_bio; | |
2014 | unsigned int segs, offset; | |
2015 | int pending, size; | |
2016 | struct split_bio *split_bio; | |
2017 | struct list_head requests; | |
402b27f9 | 2018 | |
3df0e505 | 2019 | blkfront_gather_backend_features(info); |
402b27f9 RPM |
2020 | segs = info->max_indirect_segments ? : BLKIF_MAX_SEGMENTS_PER_REQUEST; |
2021 | blk_queue_max_segments(info->rq, segs); | |
2022 | bio_list_init(&bio_list); | |
2023 | INIT_LIST_HEAD(&requests); | |
9f27ee59 | 2024 | |
3df0e505 BL |
2025 | for (r_index = 0; r_index < info->nr_rings; r_index++) { |
2026 | struct blkfront_ring_info *rinfo; | |
2027 | ||
2028 | rinfo = &info->rinfo[r_index]; | |
2029 | /* Stage 1: Make a safe copy of the shadow state. */ | |
2030 | copy = kmemdup(rinfo->shadow, sizeof(rinfo->shadow), | |
2031 | GFP_NOIO | __GFP_REPEAT | __GFP_HIGH); | |
2032 | if (!copy) | |
2033 | return -ENOMEM; | |
2034 | ||
2035 | /* Stage 2: Set up free list. */ | |
2036 | memset(&rinfo->shadow, 0, sizeof(rinfo->shadow)); | |
2037 | for (i = 0; i < BLK_RING_SIZE(info); i++) | |
2038 | rinfo->shadow[i].req.u.rw.id = i+1; | |
2039 | rinfo->shadow_free = rinfo->ring.req_prod_pvt; | |
2040 | rinfo->shadow[BLK_RING_SIZE(info)-1].req.u.rw.id = 0x0fffffff; | |
2041 | ||
2042 | rc = blkfront_setup_indirect(rinfo); | |
2043 | if (rc) { | |
2044 | kfree(copy); | |
2045 | return rc; | |
2046 | } | |
2047 | ||
2048 | for (i = 0; i < BLK_RING_SIZE(info); i++) { | |
2049 | /* Not in use? */ | |
2050 | if (!copy[i].request) | |
2051 | continue; | |
2052 | ||
402b27f9 | 2053 | /* |
3df0e505 | 2054 | * Get the bios in the request so we can re-queue them. |
402b27f9 | 2055 | */ |
3df0e505 BL |
2056 | if (copy[i].request->cmd_flags & |
2057 | (REQ_FLUSH | REQ_FUA | REQ_DISCARD | REQ_SECURE)) { | |
2058 | /* | |
2059 | * Flush operations don't contain bios, so | |
2060 | * we need to requeue the whole request | |
2061 | */ | |
2062 | list_add(©[i].request->queuelist, &requests); | |
2063 | continue; | |
2064 | } | |
2065 | merge_bio.head = copy[i].request->bio; | |
2066 | merge_bio.tail = copy[i].request->biotail; | |
2067 | bio_list_merge(&bio_list, &merge_bio); | |
2068 | copy[i].request->bio = NULL; | |
2069 | blk_end_request_all(copy[i].request, 0); | |
5ea42986 | 2070 | } |
9f27ee59 | 2071 | |
3df0e505 BL |
2072 | kfree(copy); |
2073 | } | |
9f27ee59 JF |
2074 | xenbus_switch_state(info->xbdev, XenbusStateConnected); |
2075 | ||
9f27ee59 JF |
2076 | /* Now safe for us to use the shared ring */ |
2077 | info->connected = BLKIF_STATE_CONNECTED; | |
2078 | ||
3df0e505 BL |
2079 | for (r_index = 0; r_index < info->nr_rings; r_index++) { |
2080 | struct blkfront_ring_info *rinfo; | |
2081 | ||
2082 | rinfo = &info->rinfo[r_index]; | |
2083 | /* Kick any other new requests queued since we resumed */ | |
2084 | kick_pending_request_queues(rinfo); | |
2085 | } | |
9f27ee59 | 2086 | |
402b27f9 RPM |
2087 | list_for_each_entry_safe(req, n, &requests, queuelist) { |
2088 | /* Requeue pending requests (flush or discard) */ | |
2089 | list_del_init(&req->queuelist); | |
2090 | BUG_ON(req->nr_phys_segments > segs); | |
907c3eb1 | 2091 | blk_mq_requeue_request(req); |
402b27f9 | 2092 | } |
907c3eb1 | 2093 | blk_mq_kick_requeue_list(info->rq); |
9f27ee59 | 2094 | |
402b27f9 RPM |
2095 | while ((bio = bio_list_pop(&bio_list)) != NULL) { |
2096 | /* Traverse the list of pending bios and re-queue them */ | |
2097 | if (bio_segments(bio) > segs) { | |
2098 | /* | |
2099 | * This bio has more segments than what we can | |
2100 | * handle, we have to split it. | |
2101 | */ | |
2102 | pending = (bio_segments(bio) + segs - 1) / segs; | |
2103 | split_bio = kzalloc(sizeof(*split_bio), GFP_NOIO); | |
2104 | BUG_ON(split_bio == NULL); | |
2105 | atomic_set(&split_bio->pending, pending); | |
2106 | split_bio->bio = bio; | |
2107 | for (i = 0; i < pending; i++) { | |
c004a6fe JG |
2108 | offset = (i * segs * XEN_PAGE_SIZE) >> 9; |
2109 | size = min((unsigned int)(segs * XEN_PAGE_SIZE) >> 9, | |
4f024f37 | 2110 | (unsigned int)bio_sectors(bio) - offset); |
402b27f9 RPM |
2111 | cloned_bio = bio_clone(bio, GFP_NOIO); |
2112 | BUG_ON(cloned_bio == NULL); | |
6678d83f | 2113 | bio_trim(cloned_bio, offset, size); |
402b27f9 RPM |
2114 | cloned_bio->bi_private = split_bio; |
2115 | cloned_bio->bi_end_io = split_bio_end; | |
2116 | submit_bio(cloned_bio->bi_rw, cloned_bio); | |
2117 | } | |
2118 | /* | |
2119 | * Now we have to wait for all those smaller bios to | |
2120 | * end, so we can also end the "parent" bio. | |
2121 | */ | |
2122 | continue; | |
2123 | } | |
2124 | /* We don't need to split this bio */ | |
2125 | submit_bio(bio->bi_rw, bio); | |
2126 | } | |
2127 | ||
9f27ee59 JF |
2128 | return 0; |
2129 | } | |
2130 | ||
2131 | /** | |
2132 | * We are reconnecting to the backend, due to a suspend/resume, or a backend | |
2133 | * driver restart. We tear down our blkif structure and recreate it, but | |
2134 | * leave the device-layer structures intact so that this is transparent to the | |
2135 | * rest of the kernel. | |
2136 | */ | |
2137 | static int blkfront_resume(struct xenbus_device *dev) | |
2138 | { | |
a1b4b12b | 2139 | struct blkfront_info *info = dev_get_drvdata(&dev->dev); |
3db70a85 | 2140 | int err = 0; |
9f27ee59 JF |
2141 | |
2142 | dev_dbg(&dev->dev, "blkfront_resume: %s\n", dev->nodename); | |
2143 | ||
2144 | blkif_free(info, info->connected == BLKIF_STATE_CONNECTED); | |
2145 | ||
3db70a85 BL |
2146 | err = negotiate_mq(info); |
2147 | if (err) | |
2148 | return err; | |
2149 | ||
203fd61f | 2150 | err = talk_to_blkback(dev, info); |
402b27f9 RPM |
2151 | |
2152 | /* | |
2153 | * We have to wait for the backend to switch to | |
2154 | * connected state, since we want to read which | |
2155 | * features it supports. | |
2156 | */ | |
9f27ee59 JF |
2157 | |
2158 | return err; | |
2159 | } | |
2160 | ||
6f03a7ff | 2161 | static void blkfront_closing(struct blkfront_info *info) |
b70f5fa0 DS |
2162 | { |
2163 | struct xenbus_device *xbdev = info->xbdev; | |
2164 | struct block_device *bdev = NULL; | |
2165 | ||
2166 | mutex_lock(&info->mutex); | |
2167 | ||
2168 | if (xbdev->state == XenbusStateClosing) { | |
2169 | mutex_unlock(&info->mutex); | |
2170 | return; | |
2171 | } | |
2172 | ||
2173 | if (info->gd) | |
2174 | bdev = bdget_disk(info->gd, 0); | |
2175 | ||
2176 | mutex_unlock(&info->mutex); | |
2177 | ||
2178 | if (!bdev) { | |
2179 | xenbus_frontend_closed(xbdev); | |
2180 | return; | |
2181 | } | |
2182 | ||
2183 | mutex_lock(&bdev->bd_mutex); | |
2184 | ||
7b32d104 | 2185 | if (bdev->bd_openers) { |
b70f5fa0 DS |
2186 | xenbus_dev_error(xbdev, -EBUSY, |
2187 | "Device in use; refusing to close"); | |
2188 | xenbus_switch_state(xbdev, XenbusStateClosing); | |
2189 | } else { | |
2190 | xlvbd_release_gendisk(info); | |
2191 | xenbus_frontend_closed(xbdev); | |
2192 | } | |
2193 | ||
2194 | mutex_unlock(&bdev->bd_mutex); | |
2195 | bdput(bdev); | |
2196 | } | |
9f27ee59 | 2197 | |
ed30bf31 LD |
2198 | static void blkfront_setup_discard(struct blkfront_info *info) |
2199 | { | |
2200 | int err; | |
ed30bf31 LD |
2201 | unsigned int discard_granularity; |
2202 | unsigned int discard_alignment; | |
5ea42986 | 2203 | unsigned int discard_secure; |
ed30bf31 | 2204 | |
1c8cad6c OH |
2205 | info->feature_discard = 1; |
2206 | err = xenbus_gather(XBT_NIL, info->xbdev->otherend, | |
2207 | "discard-granularity", "%u", &discard_granularity, | |
2208 | "discard-alignment", "%u", &discard_alignment, | |
2209 | NULL); | |
2210 | if (!err) { | |
2211 | info->discard_granularity = discard_granularity; | |
2212 | info->discard_alignment = discard_alignment; | |
2213 | } | |
2214 | err = xenbus_gather(XBT_NIL, info->xbdev->otherend, | |
2215 | "discard-secure", "%d", &discard_secure, | |
2216 | NULL); | |
2217 | if (!err) | |
2218 | info->feature_secdiscard = !!discard_secure; | |
ed30bf31 LD |
2219 | } |
2220 | ||
81f35161 | 2221 | static int blkfront_setup_indirect(struct blkfront_ring_info *rinfo) |
402b27f9 | 2222 | { |
c004a6fe | 2223 | unsigned int psegs, grants; |
402b27f9 | 2224 | int err, i; |
81f35161 | 2225 | struct blkfront_info *info = rinfo->dev_info; |
402b27f9 | 2226 | |
6cc56833 JG |
2227 | if (info->max_indirect_segments == 0) { |
2228 | if (!HAS_EXTRA_REQ) | |
2229 | grants = BLKIF_MAX_SEGMENTS_PER_REQUEST; | |
2230 | else { | |
2231 | /* | |
2232 | * When an extra req is required, the maximum | |
2233 | * grants supported is related to the size of the | |
2234 | * Linux block segment. | |
2235 | */ | |
2236 | grants = GRANTS_PER_PSEG; | |
2237 | } | |
2238 | } | |
d50babbe | 2239 | else |
c004a6fe JG |
2240 | grants = info->max_indirect_segments; |
2241 | psegs = grants / GRANTS_PER_PSEG; | |
402b27f9 | 2242 | |
81f35161 | 2243 | err = fill_grant_buffer(rinfo, |
c004a6fe | 2244 | (grants + INDIRECT_GREFS(grants)) * BLK_RING_SIZE(info)); |
402b27f9 RPM |
2245 | if (err) |
2246 | goto out_of_memory; | |
2247 | ||
bfe11d6d RPM |
2248 | if (!info->feature_persistent && info->max_indirect_segments) { |
2249 | /* | |
2250 | * We are using indirect descriptors but not persistent | |
2251 | * grants, we need to allocate a set of pages that can be | |
2252 | * used for mapping indirect grefs | |
2253 | */ | |
c004a6fe | 2254 | int num = INDIRECT_GREFS(grants) * BLK_RING_SIZE(info); |
bfe11d6d | 2255 | |
81f35161 | 2256 | BUG_ON(!list_empty(&rinfo->indirect_pages)); |
bfe11d6d RPM |
2257 | for (i = 0; i < num; i++) { |
2258 | struct page *indirect_page = alloc_page(GFP_NOIO); | |
2259 | if (!indirect_page) | |
2260 | goto out_of_memory; | |
81f35161 | 2261 | list_add(&indirect_page->lru, &rinfo->indirect_pages); |
bfe11d6d RPM |
2262 | } |
2263 | } | |
2264 | ||
86839c56 | 2265 | for (i = 0; i < BLK_RING_SIZE(info); i++) { |
81f35161 BL |
2266 | rinfo->shadow[i].grants_used = kzalloc( |
2267 | sizeof(rinfo->shadow[i].grants_used[0]) * grants, | |
402b27f9 | 2268 | GFP_NOIO); |
81f35161 | 2269 | rinfo->shadow[i].sg = kzalloc(sizeof(rinfo->shadow[i].sg[0]) * psegs, GFP_NOIO); |
402b27f9 | 2270 | if (info->max_indirect_segments) |
81f35161 BL |
2271 | rinfo->shadow[i].indirect_grants = kzalloc( |
2272 | sizeof(rinfo->shadow[i].indirect_grants[0]) * | |
c004a6fe | 2273 | INDIRECT_GREFS(grants), |
402b27f9 | 2274 | GFP_NOIO); |
81f35161 BL |
2275 | if ((rinfo->shadow[i].grants_used == NULL) || |
2276 | (rinfo->shadow[i].sg == NULL) || | |
402b27f9 | 2277 | (info->max_indirect_segments && |
81f35161 | 2278 | (rinfo->shadow[i].indirect_grants == NULL))) |
402b27f9 | 2279 | goto out_of_memory; |
81f35161 | 2280 | sg_init_table(rinfo->shadow[i].sg, psegs); |
402b27f9 RPM |
2281 | } |
2282 | ||
2283 | ||
2284 | return 0; | |
2285 | ||
2286 | out_of_memory: | |
86839c56 | 2287 | for (i = 0; i < BLK_RING_SIZE(info); i++) { |
81f35161 BL |
2288 | kfree(rinfo->shadow[i].grants_used); |
2289 | rinfo->shadow[i].grants_used = NULL; | |
2290 | kfree(rinfo->shadow[i].sg); | |
2291 | rinfo->shadow[i].sg = NULL; | |
2292 | kfree(rinfo->shadow[i].indirect_grants); | |
2293 | rinfo->shadow[i].indirect_grants = NULL; | |
402b27f9 | 2294 | } |
81f35161 | 2295 | if (!list_empty(&rinfo->indirect_pages)) { |
bfe11d6d | 2296 | struct page *indirect_page, *n; |
81f35161 | 2297 | list_for_each_entry_safe(indirect_page, n, &rinfo->indirect_pages, lru) { |
bfe11d6d RPM |
2298 | list_del(&indirect_page->lru); |
2299 | __free_page(indirect_page); | |
2300 | } | |
2301 | } | |
402b27f9 RPM |
2302 | return -ENOMEM; |
2303 | } | |
2304 | ||
d50babbe BL |
2305 | /* |
2306 | * Gather all backend feature-* | |
2307 | */ | |
3df0e505 | 2308 | static void blkfront_gather_backend_features(struct blkfront_info *info) |
d50babbe BL |
2309 | { |
2310 | int err; | |
2311 | int barrier, flush, discard, persistent; | |
2312 | unsigned int indirect_segments; | |
2313 | ||
2314 | info->feature_flush = 0; | |
2315 | ||
2316 | err = xenbus_gather(XBT_NIL, info->xbdev->otherend, | |
2317 | "feature-barrier", "%d", &barrier, | |
2318 | NULL); | |
2319 | ||
2320 | /* | |
2321 | * If there's no "feature-barrier" defined, then it means | |
2322 | * we're dealing with a very old backend which writes | |
2323 | * synchronously; nothing to do. | |
2324 | * | |
2325 | * If there are barriers, then we use flush. | |
2326 | */ | |
2327 | if (!err && barrier) | |
2328 | info->feature_flush = REQ_FLUSH | REQ_FUA; | |
2329 | /* | |
2330 | * And if there is "feature-flush-cache" use that above | |
2331 | * barriers. | |
2332 | */ | |
2333 | err = xenbus_gather(XBT_NIL, info->xbdev->otherend, | |
2334 | "feature-flush-cache", "%d", &flush, | |
2335 | NULL); | |
2336 | ||
2337 | if (!err && flush) | |
2338 | info->feature_flush = REQ_FLUSH; | |
2339 | ||
2340 | err = xenbus_gather(XBT_NIL, info->xbdev->otherend, | |
2341 | "feature-discard", "%d", &discard, | |
2342 | NULL); | |
2343 | ||
2344 | if (!err && discard) | |
2345 | blkfront_setup_discard(info); | |
2346 | ||
2347 | err = xenbus_gather(XBT_NIL, info->xbdev->otherend, | |
2348 | "feature-persistent", "%u", &persistent, | |
2349 | NULL); | |
2350 | if (err) | |
2351 | info->feature_persistent = 0; | |
2352 | else | |
2353 | info->feature_persistent = persistent; | |
2354 | ||
2355 | err = xenbus_gather(XBT_NIL, info->xbdev->otherend, | |
2356 | "feature-max-indirect-segments", "%u", &indirect_segments, | |
2357 | NULL); | |
2358 | if (err) | |
2359 | info->max_indirect_segments = 0; | |
2360 | else | |
2361 | info->max_indirect_segments = min(indirect_segments, | |
2362 | xen_blkif_max_segments); | |
d50babbe BL |
2363 | } |
2364 | ||
9f27ee59 JF |
2365 | /* |
2366 | * Invoked when the backend is finally 'ready' (and has told produced | |
2367 | * the details about the physical device - #sectors, size, etc). | |
2368 | */ | |
2369 | static void blkfront_connect(struct blkfront_info *info) | |
2370 | { | |
2371 | unsigned long long sectors; | |
2372 | unsigned long sector_size; | |
7c4d7d71 | 2373 | unsigned int physical_sector_size; |
9f27ee59 | 2374 | unsigned int binfo; |
3df0e505 | 2375 | int err, i; |
9f27ee59 | 2376 | |
1fa73be6 S |
2377 | switch (info->connected) { |
2378 | case BLKIF_STATE_CONNECTED: | |
2379 | /* | |
2380 | * Potentially, the back-end may be signalling | |
2381 | * a capacity change; update the capacity. | |
2382 | */ | |
2383 | err = xenbus_scanf(XBT_NIL, info->xbdev->otherend, | |
2384 | "sectors", "%Lu", §ors); | |
2385 | if (XENBUS_EXIST_ERR(err)) | |
2386 | return; | |
2387 | printk(KERN_INFO "Setting capacity to %Lu\n", | |
2388 | sectors); | |
2389 | set_capacity(info->gd, sectors); | |
2def141e | 2390 | revalidate_disk(info->gd); |
1fa73be6 | 2391 | |
402b27f9 | 2392 | return; |
1fa73be6 | 2393 | case BLKIF_STATE_SUSPENDED: |
402b27f9 RPM |
2394 | /* |
2395 | * If we are recovering from suspension, we need to wait | |
2396 | * for the backend to announce it's features before | |
2397 | * reconnecting, at least we need to know if the backend | |
2398 | * supports indirect descriptors, and how many. | |
2399 | */ | |
2400 | blkif_recover(info); | |
9f27ee59 JF |
2401 | return; |
2402 | ||
b4dddb49 JF |
2403 | default: |
2404 | break; | |
1fa73be6 | 2405 | } |
9f27ee59 JF |
2406 | |
2407 | dev_dbg(&info->xbdev->dev, "%s:%s.\n", | |
2408 | __func__, info->xbdev->otherend); | |
2409 | ||
2410 | err = xenbus_gather(XBT_NIL, info->xbdev->otherend, | |
2411 | "sectors", "%llu", §ors, | |
2412 | "info", "%u", &binfo, | |
2413 | "sector-size", "%lu", §or_size, | |
2414 | NULL); | |
2415 | if (err) { | |
2416 | xenbus_dev_fatal(info->xbdev, err, | |
2417 | "reading backend fields at %s", | |
2418 | info->xbdev->otherend); | |
2419 | return; | |
2420 | } | |
2421 | ||
7c4d7d71 SB |
2422 | /* |
2423 | * physcial-sector-size is a newer field, so old backends may not | |
2424 | * provide this. Assume physical sector size to be the same as | |
2425 | * sector_size in that case. | |
2426 | */ | |
2427 | err = xenbus_scanf(XBT_NIL, info->xbdev->otherend, | |
2428 | "physical-sector-size", "%u", &physical_sector_size); | |
2429 | if (err != 1) | |
2430 | physical_sector_size = sector_size; | |
2431 | ||
3df0e505 BL |
2432 | blkfront_gather_backend_features(info); |
2433 | for (i = 0; i < info->nr_rings; i++) { | |
2434 | err = blkfront_setup_indirect(&info->rinfo[i]); | |
2435 | if (err) { | |
2436 | xenbus_dev_fatal(info->xbdev, err, "setup_indirect at %s", | |
2437 | info->xbdev->otherend); | |
2438 | blkif_free(info, 0); | |
2439 | break; | |
2440 | } | |
402b27f9 RPM |
2441 | } |
2442 | ||
7c4d7d71 SB |
2443 | err = xlvbd_alloc_gendisk(sectors, info, binfo, sector_size, |
2444 | physical_sector_size); | |
9f27ee59 JF |
2445 | if (err) { |
2446 | xenbus_dev_fatal(info->xbdev, err, "xlvbd_add at %s", | |
2447 | info->xbdev->otherend); | |
2448 | return; | |
2449 | } | |
2450 | ||
2451 | xenbus_switch_state(info->xbdev, XenbusStateConnected); | |
2452 | ||
2453 | /* Kick pending requests. */ | |
9f27ee59 | 2454 | info->connected = BLKIF_STATE_CONNECTED; |
3df0e505 BL |
2455 | for (i = 0; i < info->nr_rings; i++) |
2456 | kick_pending_request_queues(&info->rinfo[i]); | |
9f27ee59 JF |
2457 | |
2458 | add_disk(info->gd); | |
1d78d705 CL |
2459 | |
2460 | info->is_ready = 1; | |
9f27ee59 JF |
2461 | } |
2462 | ||
9f27ee59 JF |
2463 | /** |
2464 | * Callback received when the backend's state changes. | |
2465 | */ | |
203fd61f | 2466 | static void blkback_changed(struct xenbus_device *dev, |
9f27ee59 JF |
2467 | enum xenbus_state backend_state) |
2468 | { | |
a1b4b12b | 2469 | struct blkfront_info *info = dev_get_drvdata(&dev->dev); |
9f27ee59 | 2470 | |
203fd61f | 2471 | dev_dbg(&dev->dev, "blkfront:blkback_changed to state %d.\n", backend_state); |
9f27ee59 JF |
2472 | |
2473 | switch (backend_state) { | |
9f27ee59 | 2474 | case XenbusStateInitWait: |
a9b54bb9 BL |
2475 | if (dev->state != XenbusStateInitialising) |
2476 | break; | |
c31ecf6c | 2477 | if (talk_to_blkback(dev, info)) |
8ab0144a | 2478 | break; |
8ab0144a | 2479 | case XenbusStateInitialising: |
9f27ee59 | 2480 | case XenbusStateInitialised: |
b78c9512 NI |
2481 | case XenbusStateReconfiguring: |
2482 | case XenbusStateReconfigured: | |
9f27ee59 | 2483 | case XenbusStateUnknown: |
9f27ee59 JF |
2484 | break; |
2485 | ||
2486 | case XenbusStateConnected: | |
c31ecf6c KRW |
2487 | if (dev->state != XenbusStateInitialised) { |
2488 | if (talk_to_blkback(dev, info)) | |
2489 | break; | |
2490 | } | |
9f27ee59 JF |
2491 | blkfront_connect(info); |
2492 | break; | |
2493 | ||
36613717 DV |
2494 | case XenbusStateClosed: |
2495 | if (dev->state == XenbusStateClosed) | |
2496 | break; | |
2497 | /* Missed the backend's Closing state -- fallthrough */ | |
9f27ee59 | 2498 | case XenbusStateClosing: |
a54c8f0f CA |
2499 | if (info) |
2500 | blkfront_closing(info); | |
9f27ee59 JF |
2501 | break; |
2502 | } | |
2503 | } | |
2504 | ||
fa1bd359 | 2505 | static int blkfront_remove(struct xenbus_device *xbdev) |
9f27ee59 | 2506 | { |
fa1bd359 DS |
2507 | struct blkfront_info *info = dev_get_drvdata(&xbdev->dev); |
2508 | struct block_device *bdev = NULL; | |
2509 | struct gendisk *disk; | |
9f27ee59 | 2510 | |
fa1bd359 | 2511 | dev_dbg(&xbdev->dev, "%s removed", xbdev->nodename); |
9f27ee59 JF |
2512 | |
2513 | blkif_free(info, 0); | |
2514 | ||
fa1bd359 DS |
2515 | mutex_lock(&info->mutex); |
2516 | ||
2517 | disk = info->gd; | |
2518 | if (disk) | |
2519 | bdev = bdget_disk(disk, 0); | |
2520 | ||
2521 | info->xbdev = NULL; | |
2522 | mutex_unlock(&info->mutex); | |
2523 | ||
2524 | if (!bdev) { | |
2525 | kfree(info); | |
2526 | return 0; | |
2527 | } | |
2528 | ||
2529 | /* | |
2530 | * The xbdev was removed before we reached the Closed | |
2531 | * state. See if it's safe to remove the disk. If the bdev | |
2532 | * isn't closed yet, we let release take care of it. | |
2533 | */ | |
2534 | ||
2535 | mutex_lock(&bdev->bd_mutex); | |
2536 | info = disk->private_data; | |
2537 | ||
d54142c7 DS |
2538 | dev_warn(disk_to_dev(disk), |
2539 | "%s was hot-unplugged, %d stale handles\n", | |
2540 | xbdev->nodename, bdev->bd_openers); | |
2541 | ||
7b32d104 | 2542 | if (info && !bdev->bd_openers) { |
fa1bd359 DS |
2543 | xlvbd_release_gendisk(info); |
2544 | disk->private_data = NULL; | |
0e345826 | 2545 | kfree(info); |
fa1bd359 DS |
2546 | } |
2547 | ||
2548 | mutex_unlock(&bdev->bd_mutex); | |
2549 | bdput(bdev); | |
9f27ee59 JF |
2550 | |
2551 | return 0; | |
2552 | } | |
2553 | ||
1d78d705 CL |
2554 | static int blkfront_is_ready(struct xenbus_device *dev) |
2555 | { | |
a1b4b12b | 2556 | struct blkfront_info *info = dev_get_drvdata(&dev->dev); |
1d78d705 | 2557 | |
5d7ed20e | 2558 | return info->is_ready && info->xbdev; |
1d78d705 CL |
2559 | } |
2560 | ||
a63c848b | 2561 | static int blkif_open(struct block_device *bdev, fmode_t mode) |
9f27ee59 | 2562 | { |
13961743 DS |
2563 | struct gendisk *disk = bdev->bd_disk; |
2564 | struct blkfront_info *info; | |
2565 | int err = 0; | |
6e9624b8 | 2566 | |
2a48fc0a | 2567 | mutex_lock(&blkfront_mutex); |
6e9624b8 | 2568 | |
13961743 DS |
2569 | info = disk->private_data; |
2570 | if (!info) { | |
2571 | /* xbdev gone */ | |
2572 | err = -ERESTARTSYS; | |
2573 | goto out; | |
2574 | } | |
2575 | ||
2576 | mutex_lock(&info->mutex); | |
2577 | ||
2578 | if (!info->gd) | |
2579 | /* xbdev is closed */ | |
2580 | err = -ERESTARTSYS; | |
2581 | ||
2582 | mutex_unlock(&info->mutex); | |
2583 | ||
13961743 | 2584 | out: |
2a48fc0a | 2585 | mutex_unlock(&blkfront_mutex); |
13961743 | 2586 | return err; |
9f27ee59 JF |
2587 | } |
2588 | ||
db2a144b | 2589 | static void blkif_release(struct gendisk *disk, fmode_t mode) |
9f27ee59 | 2590 | { |
a63c848b | 2591 | struct blkfront_info *info = disk->private_data; |
7fd152f4 DS |
2592 | struct block_device *bdev; |
2593 | struct xenbus_device *xbdev; | |
2594 | ||
2a48fc0a | 2595 | mutex_lock(&blkfront_mutex); |
7fd152f4 DS |
2596 | |
2597 | bdev = bdget_disk(disk, 0); | |
7fd152f4 | 2598 | |
2f089cb8 FP |
2599 | if (!bdev) { |
2600 | WARN(1, "Block device %s yanked out from us!\n", disk->disk_name); | |
2601 | goto out_mutex; | |
2602 | } | |
acfca3c6 DS |
2603 | if (bdev->bd_openers) |
2604 | goto out; | |
2605 | ||
7fd152f4 DS |
2606 | /* |
2607 | * Check if we have been instructed to close. We will have | |
2608 | * deferred this request, because the bdev was still open. | |
2609 | */ | |
2610 | ||
2611 | mutex_lock(&info->mutex); | |
2612 | xbdev = info->xbdev; | |
2613 | ||
2614 | if (xbdev && xbdev->state == XenbusStateClosing) { | |
2615 | /* pending switch to state closed */ | |
d54142c7 | 2616 | dev_info(disk_to_dev(bdev->bd_disk), "releasing disk\n"); |
7fd152f4 DS |
2617 | xlvbd_release_gendisk(info); |
2618 | xenbus_frontend_closed(info->xbdev); | |
2619 | } | |
2620 | ||
2621 | mutex_unlock(&info->mutex); | |
2622 | ||
2623 | if (!xbdev) { | |
2624 | /* sudden device removal */ | |
d54142c7 | 2625 | dev_info(disk_to_dev(bdev->bd_disk), "releasing disk\n"); |
7fd152f4 DS |
2626 | xlvbd_release_gendisk(info); |
2627 | disk->private_data = NULL; | |
2628 | kfree(info); | |
9f27ee59 | 2629 | } |
7fd152f4 | 2630 | |
a4cc14ec | 2631 | out: |
dad5cf65 | 2632 | bdput(bdev); |
2f089cb8 | 2633 | out_mutex: |
2a48fc0a | 2634 | mutex_unlock(&blkfront_mutex); |
9f27ee59 JF |
2635 | } |
2636 | ||
83d5cde4 | 2637 | static const struct block_device_operations xlvbd_block_fops = |
9f27ee59 JF |
2638 | { |
2639 | .owner = THIS_MODULE, | |
a63c848b AV |
2640 | .open = blkif_open, |
2641 | .release = blkif_release, | |
597592d9 | 2642 | .getgeo = blkif_getgeo, |
8a6cfeb6 | 2643 | .ioctl = blkif_ioctl, |
9f27ee59 JF |
2644 | }; |
2645 | ||
2646 | ||
ec9c42ec | 2647 | static const struct xenbus_device_id blkfront_ids[] = { |
9f27ee59 JF |
2648 | { "vbd" }, |
2649 | { "" } | |
2650 | }; | |
2651 | ||
95afae48 DV |
2652 | static struct xenbus_driver blkfront_driver = { |
2653 | .ids = blkfront_ids, | |
9f27ee59 JF |
2654 | .probe = blkfront_probe, |
2655 | .remove = blkfront_remove, | |
2656 | .resume = blkfront_resume, | |
203fd61f | 2657 | .otherend_changed = blkback_changed, |
1d78d705 | 2658 | .is_ready = blkfront_is_ready, |
95afae48 | 2659 | }; |
9f27ee59 JF |
2660 | |
2661 | static int __init xlblk_init(void) | |
2662 | { | |
469738e6 | 2663 | int ret; |
28d949bc | 2664 | int nr_cpus = num_online_cpus(); |
469738e6 | 2665 | |
6e833587 | 2666 | if (!xen_domain()) |
9f27ee59 JF |
2667 | return -ENODEV; |
2668 | ||
9cce2914 | 2669 | if (xen_blkif_max_ring_order > XENBUS_MAX_RING_GRANT_ORDER) { |
86839c56 | 2670 | pr_info("Invalid max_ring_order (%d), will use default max: %d.\n", |
9cce2914 | 2671 | xen_blkif_max_ring_order, XENBUS_MAX_RING_GRANT_ORDER); |
45fc8264 | 2672 | xen_blkif_max_ring_order = XENBUS_MAX_RING_GRANT_ORDER; |
86839c56 BL |
2673 | } |
2674 | ||
28d949bc BL |
2675 | if (xen_blkif_max_queues > nr_cpus) { |
2676 | pr_info("Invalid max_queues (%d), will use default max: %d.\n", | |
2677 | xen_blkif_max_queues, nr_cpus); | |
2678 | xen_blkif_max_queues = nr_cpus; | |
2679 | } | |
2680 | ||
51c71a3b | 2681 | if (!xen_has_pv_disk_devices()) |
b9136d20 IM |
2682 | return -ENODEV; |
2683 | ||
9f27ee59 JF |
2684 | if (register_blkdev(XENVBD_MAJOR, DEV_NAME)) { |
2685 | printk(KERN_WARNING "xen_blk: can't get major %d with name %s\n", | |
2686 | XENVBD_MAJOR, DEV_NAME); | |
2687 | return -ENODEV; | |
2688 | } | |
2689 | ||
73db144b | 2690 | ret = xenbus_register_frontend(&blkfront_driver); |
469738e6 LE |
2691 | if (ret) { |
2692 | unregister_blkdev(XENVBD_MAJOR, DEV_NAME); | |
2693 | return ret; | |
2694 | } | |
2695 | ||
2696 | return 0; | |
9f27ee59 JF |
2697 | } |
2698 | module_init(xlblk_init); | |
2699 | ||
2700 | ||
5a60d0cd | 2701 | static void __exit xlblk_exit(void) |
9f27ee59 | 2702 | { |
8605067f JB |
2703 | xenbus_unregister_driver(&blkfront_driver); |
2704 | unregister_blkdev(XENVBD_MAJOR, DEV_NAME); | |
2705 | kfree(minors); | |
9f27ee59 JF |
2706 | } |
2707 | module_exit(xlblk_exit); | |
2708 | ||
2709 | MODULE_DESCRIPTION("Xen virtual block device frontend"); | |
2710 | MODULE_LICENSE("GPL"); | |
2711 | MODULE_ALIAS_BLOCKDEV_MAJOR(XENVBD_MAJOR); | |
d2f0c52b | 2712 | MODULE_ALIAS("xen:vbd"); |
4f93f09b | 2713 | MODULE_ALIAS("xenblk"); |