]> git.proxmox.com Git - mirror_qemu.git/blame - block.c
task switch fixes
[mirror_qemu.git] / block.c
CommitLineData
fc01f7e7
FB
1/*
2 * QEMU System Emulator block driver
3 *
4 * Copyright (c) 2003 Fabrice Bellard
5 *
6 * Permission is hereby granted, free of charge, to any person obtaining a copy
7 * of this software and associated documentation files (the "Software"), to deal
8 * in the Software without restriction, including without limitation the rights
9 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
10 * copies of the Software, and to permit persons to whom the Software is
11 * furnished to do so, subject to the following conditions:
12 *
13 * The above copyright notice and this permission notice shall be included in
14 * all copies or substantial portions of the Software.
15 *
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
19 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
22 * THE SOFTWARE.
23 */
24#include <stdlib.h>
25#include <stdio.h>
26#include <stdarg.h>
27#include <string.h>
28#include <getopt.h>
29#include <inttypes.h>
30#include <unistd.h>
31#include <sys/mman.h>
32#include <fcntl.h>
33#include <signal.h>
34#include <time.h>
35#include <sys/time.h>
36#include <malloc.h>
37#include <termios.h>
38#include <sys/poll.h>
39#include <errno.h>
40#include <sys/wait.h>
33e3963e 41#include <netinet/in.h>
fc01f7e7
FB
42
43#include "vl.h"
44
33e3963e
FB
45#define NO_THUNK_TYPE_SIZE
46#include "thunk.h"
47
00af2b26
FB
48#include "cow.h"
49
fc01f7e7 50struct BlockDriverState {
33e3963e 51 int fd; /* if -1, only COW mappings */
fc01f7e7 52 int64_t total_sectors;
b338082b
FB
53 int read_only; /* if true, the media is read only */
54 int inserted; /* if true, the media is present */
55 int removable; /* if true, the media can be removed */
56 int locked; /* if true, the media cannot temporarily be ejected */
57 /* event callback when inserting/removing */
58 void (*change_cb)(void *opaque);
59 void *change_opaque;
33e3963e
FB
60
61 uint8_t *cow_bitmap; /* if non NULL, COW mappings are used first */
62 uint8_t *cow_bitmap_addr; /* mmap address of cow_bitmap */
63 int cow_bitmap_size;
64 int cow_fd;
65 int64_t cow_sectors_offset;
cf98951b
FB
66 int boot_sector_enabled;
67 uint8_t boot_sector_data[512];
68
33e3963e 69 char filename[1024];
b338082b
FB
70
71 /* NOTE: the following infos are only hints for real hardware
72 drivers. They are not used by the block driver */
73 int cyls, heads, secs;
74 int type;
75 char device_name[32];
76 BlockDriverState *next;
fc01f7e7
FB
77};
78
b338082b
FB
79static BlockDriverState *bdrv_first;
80
81/* create a new block device (by default it is empty) */
82BlockDriverState *bdrv_new(const char *device_name)
83{
84 BlockDriverState **pbs, *bs;
85
86 bs = qemu_mallocz(sizeof(BlockDriverState));
87 if(!bs)
88 return NULL;
89 pstrcpy(bs->device_name, sizeof(bs->device_name), device_name);
90 /* insert at the end */
91 pbs = &bdrv_first;
92 while (*pbs != NULL)
93 pbs = &(*pbs)->next;
94 *pbs = bs;
95 return bs;
96}
97
98int bdrv_open(BlockDriverState *bs, const char *filename, int snapshot)
fc01f7e7 99{
33e3963e 100 int fd, cow_fd;
fc01f7e7 101 int64_t size;
33e3963e
FB
102 char template[] = "/tmp/vl.XXXXXX";
103 struct cow_header_v2 cow_header;
104 struct stat st;
fc01f7e7 105
0849bf08 106 bs->read_only = 0;
33e3963e
FB
107 bs->fd = -1;
108 bs->cow_fd = -1;
109 bs->cow_bitmap = NULL;
110 strcpy(bs->filename, filename);
111
112 /* open standard HD image */
113 fd = open(filename, O_RDWR | O_LARGEFILE);
fc01f7e7 114 if (fd < 0) {
33e3963e
FB
115 /* read only image on disk */
116 fd = open(filename, O_RDONLY | O_LARGEFILE);
0849bf08 117 if (fd < 0) {
33e3963e
FB
118 perror(filename);
119 goto fail;
0849bf08 120 }
33e3963e
FB
121 if (!snapshot)
122 bs->read_only = 1;
fc01f7e7 123 }
fc01f7e7 124 bs->fd = fd;
33e3963e
FB
125
126 /* see if it is a cow image */
127 if (read(fd, &cow_header, sizeof(cow_header)) != sizeof(cow_header)) {
128 fprintf(stderr, "%s: could not read header\n", filename);
129 goto fail;
130 }
131 if (cow_header.magic == htonl(COW_MAGIC) &&
132 cow_header.version == htonl(COW_VERSION)) {
133 /* cow image found */
134 size = cow_header.size;
135#ifndef WORDS_BIGENDIAN
136 size = bswap64(size);
137#endif
138 bs->total_sectors = size / 512;
139
140 bs->cow_fd = fd;
141 bs->fd = -1;
142 if (cow_header.backing_file[0] != '\0') {
143 if (stat(cow_header.backing_file, &st) != 0) {
144 fprintf(stderr, "%s: could not find original disk image '%s'\n", filename, cow_header.backing_file);
145 goto fail;
146 }
147 if (st.st_mtime != htonl(cow_header.mtime)) {
148 fprintf(stderr, "%s: original raw disk image '%s' does not match saved timestamp\n", filename, cow_header.backing_file);
149 goto fail;
150 }
151 fd = open(cow_header.backing_file, O_RDONLY | O_LARGEFILE);
152 if (fd < 0)
153 goto fail;
154 bs->fd = fd;
155 }
156 /* mmap the bitmap */
157 bs->cow_bitmap_size = ((bs->total_sectors + 7) >> 3) + sizeof(cow_header);
158 bs->cow_bitmap_addr = mmap(get_mmap_addr(bs->cow_bitmap_size),
159 bs->cow_bitmap_size,
160 PROT_READ | PROT_WRITE,
161 MAP_SHARED, bs->cow_fd, 0);
162 if (bs->cow_bitmap_addr == MAP_FAILED)
163 goto fail;
164 bs->cow_bitmap = bs->cow_bitmap_addr + sizeof(cow_header);
165 bs->cow_sectors_offset = (bs->cow_bitmap_size + 511) & ~511;
166 snapshot = 0;
167 } else {
168 /* standard raw image */
169 size = lseek64(fd, 0, SEEK_END);
170 bs->total_sectors = size / 512;
171 bs->fd = fd;
172 }
173
174 if (snapshot) {
175 /* create a temporary COW file */
176 cow_fd = mkstemp(template);
177 if (cow_fd < 0)
178 goto fail;
179 bs->cow_fd = cow_fd;
180 unlink(template);
181
182 /* just need to allocate bitmap */
183 bs->cow_bitmap_size = (bs->total_sectors + 7) >> 3;
184 bs->cow_bitmap_addr = mmap(get_mmap_addr(bs->cow_bitmap_size),
185 bs->cow_bitmap_size,
186 PROT_READ | PROT_WRITE,
187 MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
188 if (bs->cow_bitmap_addr == MAP_FAILED)
189 goto fail;
190 bs->cow_bitmap = bs->cow_bitmap_addr;
191 bs->cow_sectors_offset = 0;
192 }
193
b338082b
FB
194 bs->inserted = 1;
195
196 /* call the change callback */
197 if (bs->change_cb)
198 bs->change_cb(bs->change_opaque);
199
200 return 0;
33e3963e
FB
201 fail:
202 bdrv_close(bs);
b338082b 203 return -1;
fc01f7e7
FB
204}
205
206void bdrv_close(BlockDriverState *bs)
207{
b338082b
FB
208 if (bs->inserted) {
209 /* we unmap the mapping so that it is written to the COW file */
210 if (bs->cow_bitmap_addr)
211 munmap(bs->cow_bitmap_addr, bs->cow_bitmap_size);
212 if (bs->cow_fd >= 0)
213 close(bs->cow_fd);
214 if (bs->fd >= 0)
215 close(bs->fd);
216 bs->inserted = 0;
217
218 /* call the change callback */
219 if (bs->change_cb)
220 bs->change_cb(bs->change_opaque);
221 }
222}
223
224void bdrv_delete(BlockDriverState *bs)
225{
226 bdrv_close(bs);
227 qemu_free(bs);
fc01f7e7
FB
228}
229
33e3963e
FB
230static inline void set_bit(uint8_t *bitmap, int64_t bitnum)
231{
232 bitmap[bitnum / 8] |= (1 << (bitnum%8));
233}
234
235static inline int is_bit_set(const uint8_t *bitmap, int64_t bitnum)
236{
237 return !!(bitmap[bitnum / 8] & (1 << (bitnum%8)));
238}
239
240
241/* Return true if first block has been changed (ie. current version is
242 * in COW file). Set the number of continuous blocks for which that
243 * is true. */
244static int is_changed(uint8_t *bitmap,
245 int64_t sector_num, int nb_sectors,
246 int *num_same)
247{
248 int changed;
249
250 if (!bitmap || nb_sectors == 0) {
251 *num_same = nb_sectors;
252 return 0;
253 }
254
255 changed = is_bit_set(bitmap, sector_num);
256 for (*num_same = 1; *num_same < nb_sectors; (*num_same)++) {
257 if (is_bit_set(bitmap, sector_num + *num_same) != changed)
258 break;
259 }
260
261 return changed;
262}
263
264/* commit COW file into the raw image */
265int bdrv_commit(BlockDriverState *bs)
266{
267 int64_t i;
268 uint8_t *cow_bitmap;
269
b338082b
FB
270 if (!bs->inserted)
271 return -1;
272
33e3963e
FB
273 if (!bs->cow_bitmap) {
274 fprintf(stderr, "Already committed to %s\n", bs->filename);
275 return 0;
276 }
277
278 if (bs->read_only) {
279 fprintf(stderr, "Can't commit to %s: read-only\n", bs->filename);
280 return -1;
281 }
282
283 cow_bitmap = bs->cow_bitmap;
284 for (i = 0; i < bs->total_sectors; i++) {
285 if (is_bit_set(cow_bitmap, i)) {
286 unsigned char sector[512];
287 if (bdrv_read(bs, i, sector, 1) != 0) {
288 fprintf(stderr, "Error reading sector %lli: aborting commit\n",
289 (long long)i);
290 return -1;
291 }
292
293 /* Make bdrv_write write to real file for a moment. */
294 bs->cow_bitmap = NULL;
295 if (bdrv_write(bs, i, sector, 1) != 0) {
296 fprintf(stderr, "Error writing sector %lli: aborting commit\n",
297 (long long)i);
298 bs->cow_bitmap = cow_bitmap;
299 return -1;
300 }
301 bs->cow_bitmap = cow_bitmap;
302 }
303 }
304 fprintf(stderr, "Committed snapshot to %s\n", bs->filename);
305 return 0;
306}
307
fc01f7e7
FB
308/* return -1 if error */
309int bdrv_read(BlockDriverState *bs, int64_t sector_num,
310 uint8_t *buf, int nb_sectors)
311{
33e3963e
FB
312 int ret, n, fd;
313 int64_t offset;
314
b338082b
FB
315 if (!bs->inserted)
316 return -1;
317
33e3963e
FB
318 while (nb_sectors > 0) {
319 if (is_changed(bs->cow_bitmap, sector_num, nb_sectors, &n)) {
320 fd = bs->cow_fd;
321 offset = bs->cow_sectors_offset;
cf98951b
FB
322 } else if (sector_num == 0 && bs->boot_sector_enabled) {
323 memcpy(buf, bs->boot_sector_data, 512);
324 n = 1;
325 goto next;
33e3963e
FB
326 } else {
327 fd = bs->fd;
328 offset = 0;
329 }
fc01f7e7 330
33e3963e
FB
331 if (fd < 0) {
332 /* no file, just return empty sectors */
333 memset(buf, 0, n * 512);
334 } else {
335 offset += sector_num * 512;
336 lseek64(fd, offset, SEEK_SET);
337 ret = read(fd, buf, n * 512);
338 if (ret != n * 512) {
339 return -1;
340 }
341 }
cf98951b 342 next:
33e3963e
FB
343 nb_sectors -= n;
344 sector_num += n;
345 buf += n * 512;
346 }
347 return 0;
fc01f7e7
FB
348}
349
350/* return -1 if error */
351int bdrv_write(BlockDriverState *bs, int64_t sector_num,
352 const uint8_t *buf, int nb_sectors)
353{
33e3963e
FB
354 int ret, fd, i;
355 int64_t offset, retl;
cf98951b 356
b338082b
FB
357 if (!bs->inserted)
358 return -1;
0849bf08
FB
359 if (bs->read_only)
360 return -1;
361
33e3963e
FB
362 if (bs->cow_bitmap) {
363 fd = bs->cow_fd;
364 offset = bs->cow_sectors_offset;
365 } else {
366 fd = bs->fd;
367 offset = 0;
368 }
369
370 offset += sector_num * 512;
371 retl = lseek64(fd, offset, SEEK_SET);
372 if (retl == -1) {
fc01f7e7 373 return -1;
33e3963e
FB
374 }
375 ret = write(fd, buf, nb_sectors * 512);
376 if (ret != nb_sectors * 512) {
377 return -1;
378 }
379
380 if (bs->cow_bitmap) {
381 for (i = 0; i < nb_sectors; i++)
382 set_bit(bs->cow_bitmap, sector_num + i);
383 }
384 return 0;
fc01f7e7
FB
385}
386
387void bdrv_get_geometry(BlockDriverState *bs, int64_t *nb_sectors_ptr)
388{
389 *nb_sectors_ptr = bs->total_sectors;
390}
cf98951b
FB
391
392/* force a given boot sector. */
393void bdrv_set_boot_sector(BlockDriverState *bs, const uint8_t *data, int size)
394{
395 bs->boot_sector_enabled = 1;
396 if (size > 512)
397 size = 512;
398 memcpy(bs->boot_sector_data, data, size);
399 memset(bs->boot_sector_data + size, 0, 512 - size);
400}
b338082b
FB
401
402void bdrv_set_geometry_hint(BlockDriverState *bs,
403 int cyls, int heads, int secs)
404{
405 bs->cyls = cyls;
406 bs->heads = heads;
407 bs->secs = secs;
408}
409
410void bdrv_set_type_hint(BlockDriverState *bs, int type)
411{
412 bs->type = type;
413 bs->removable = ((type == BDRV_TYPE_CDROM ||
414 type == BDRV_TYPE_FLOPPY));
415}
416
417void bdrv_get_geometry_hint(BlockDriverState *bs,
418 int *pcyls, int *pheads, int *psecs)
419{
420 *pcyls = bs->cyls;
421 *pheads = bs->heads;
422 *psecs = bs->secs;
423}
424
425int bdrv_get_type_hint(BlockDriverState *bs)
426{
427 return bs->type;
428}
429
430int bdrv_is_removable(BlockDriverState *bs)
431{
432 return bs->removable;
433}
434
435int bdrv_is_read_only(BlockDriverState *bs)
436{
437 return bs->read_only;
438}
439
440int bdrv_is_inserted(BlockDriverState *bs)
441{
442 return bs->inserted;
443}
444
445int bdrv_is_locked(BlockDriverState *bs)
446{
447 return bs->locked;
448}
449
450void bdrv_set_locked(BlockDriverState *bs, int locked)
451{
452 bs->locked = locked;
453}
454
455void bdrv_set_change_cb(BlockDriverState *bs,
456 void (*change_cb)(void *opaque), void *opaque)
457{
458 bs->change_cb = change_cb;
459 bs->change_opaque = opaque;
460}
461
462BlockDriverState *bdrv_find(const char *name)
463{
464 BlockDriverState *bs;
465
466 for (bs = bdrv_first; bs != NULL; bs = bs->next) {
467 if (!strcmp(name, bs->device_name))
468 return bs;
469 }
470 return NULL;
471}
472
473void bdrv_info(void)
474{
475 BlockDriverState *bs;
476
477 for (bs = bdrv_first; bs != NULL; bs = bs->next) {
478 term_printf("%s:", bs->device_name);
479 term_printf(" type=");
480 switch(bs->type) {
481 case BDRV_TYPE_HD:
482 term_printf("hd");
483 break;
484 case BDRV_TYPE_CDROM:
485 term_printf("cdrom");
486 break;
487 case BDRV_TYPE_FLOPPY:
488 term_printf("floppy");
489 break;
490 }
491 term_printf(" removable=%d", bs->removable);
492 if (bs->removable) {
493 term_printf(" locked=%d", bs->locked);
494 }
495 if (bs->inserted) {
496 term_printf(" file=%s", bs->filename);
497 term_printf(" ro=%d", bs->read_only);
498 } else {
499 term_printf(" [not inserted]");
500 }
501 term_printf("\n");
502 }
503}