1 /* $NetBSD: udf_strat_direct.c,v 1.9 2009/02/10 17:49:02 reinoud Exp $ */
4 * Copyright (c) 2006, 2008 Reinoud Zandijk
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
19 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
21 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
22 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
23 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
25 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
29 #include <sys/cdefs.h>
31 __KERNEL_RCSID(0, "$NetBSD: udf_strat_direct.c,v 1.9 2009/02/10 17:49:02 reinoud Exp $");
35 #if defined(_KERNEL_OPT)
36 #include "opt_compat_netbsd.h"
39 #include <sys/param.h>
40 #include <sys/systm.h>
41 #include <sys/sysctl.h>
42 #include <sys/namei.h>
44 #include <sys/kernel.h>
45 #include <sys/vnode.h>
46 #include <miscfs/genfs/genfs_node.h>
47 #include <sys/mount.h>
50 #include <sys/device.h>
51 #include <sys/disklabel.h>
52 #include <sys/ioctl.h>
53 #include <sys/malloc.h>
54 #include <sys/dirent.h>
57 #include <sys/kauth.h>
58 #include <sys/kthread.h>
59 #include <dev/clock_subr.h>
61 #include <fs/udf/ecma167-udf.h>
62 #include <fs/udf/udf_mount.h>
66 #include "udf_bswap.h"
69 #define VTOI(vnode) ((struct udf_node *) vnode->v_data)
70 #define PRIV(ump) ((struct strat_private *) ump->strategy_private)
72 /* --------------------------------------------------------------------- */
75 #define UDF_SHED_MAX 3
77 #define UDF_SHED_READING 0
78 #define UDF_SHED_WRITING 1
79 #define UDF_SHED_SEQWRITING 2
82 struct strat_private
{
83 struct pool desc_pool
; /* node descriptors */
86 /* --------------------------------------------------------------------- */
89 udf_wr_nodedscr_callback(struct buf
*buf
)
91 struct udf_node
*udf_node
;
96 /* called when write action is done */
97 DPRINTF(WRITE
, ("udf_wr_nodedscr_callback(): node written out\n"));
99 udf_node
= VTOI(buf
->b_vp
);
100 if (udf_node
== NULL
) {
102 printf("udf_wr_node_callback: NULL node?\n");
106 /* XXX right flags to mark dirty again on error? */
108 /* write error on `defect free' media??? how to solve? */
109 /* XXX lookup UDF standard for unallocatable space */
110 udf_node
->i_flags
|= IN_MODIFIED
| IN_ACCESSED
;
113 /* decrement outstanding_nodedscr */
114 KASSERT(udf_node
->outstanding_nodedscr
>= 1);
115 udf_node
->outstanding_nodedscr
--;
116 if (udf_node
->outstanding_nodedscr
== 0) {
117 /* unlock the node */
118 UDF_UNLOCK_NODE(udf_node
, 0);
119 wakeup(&udf_node
->outstanding_nodedscr
);
121 /* unreference the vnode so it can be recycled */
122 holdrele(udf_node
->vnode
);
127 /* --------------------------------------------------------------------- */
130 udf_getblank_nodedscr_direct(struct udf_strat_args
*args
)
132 union dscrptr
**dscrptr
= &args
->dscr
;
133 struct udf_mount
*ump
= args
->ump
;
134 struct strat_private
*priv
= PRIV(ump
);
137 lb_size
= udf_rw32(ump
->logical_vol
->lb_size
);
138 *dscrptr
= pool_get(&priv
->desc_pool
, PR_WAITOK
);
139 memset(*dscrptr
, 0, lb_size
);
146 udf_free_nodedscr_direct(struct udf_strat_args
*args
)
148 union dscrptr
*dscr
= args
->dscr
;
149 struct udf_mount
*ump
= args
->ump
;
150 struct strat_private
*priv
= PRIV(ump
);
152 pool_put(&priv
->desc_pool
, dscr
);
157 udf_read_nodedscr_direct(struct udf_strat_args
*args
)
159 union dscrptr
**dscrptr
= &args
->dscr
;
160 union dscrptr
*tmpdscr
;
161 struct udf_mount
*ump
= args
->ump
;
162 struct long_ad
*icb
= args
->icb
;
163 struct strat_private
*priv
= PRIV(ump
);
165 uint32_t sector
, dummy
;
168 lb_size
= udf_rw32(ump
->logical_vol
->lb_size
);
170 error
= udf_translate_vtop(ump
, icb
, §or
, &dummy
);
174 /* try to read in fe/efe */
175 error
= udf_read_phys_dscr(ump
, sector
, M_UDFTEMP
, &tmpdscr
);
179 *dscrptr
= pool_get(&priv
->desc_pool
, PR_WAITOK
);
180 memcpy(*dscrptr
, tmpdscr
, lb_size
);
181 free(tmpdscr
, M_UDFTEMP
);
188 udf_write_nodedscr_direct(struct udf_strat_args
*args
)
190 struct udf_mount
*ump
= args
->ump
;
191 struct udf_node
*udf_node
= args
->udf_node
;
192 union dscrptr
*dscr
= args
->dscr
;
193 struct long_ad
*icb
= args
->icb
;
194 int waitfor
= args
->waitfor
;
195 uint32_t logsector
, sector
, dummy
;
199 * we have to decide if we write it out sequential or at its fixed
200 * position by examining the partition its (to be) written on.
202 vpart
= udf_rw16(udf_node
->loc
.loc
.part_num
);
203 logsector
= udf_rw32(icb
->loc
.lb_num
);
204 KASSERT(ump
->vtop_tp
[vpart
] != UDF_VTOP_TYPE_VIRT
);
207 error
= udf_translate_vtop(ump
, icb
, §or
, &dummy
);
211 /* add reference to the vnode to prevent recycling */
212 vhold(udf_node
->vnode
);
215 DPRINTF(WRITE
, ("udf_write_nodedscr: sync write\n"));
217 error
= udf_write_phys_dscr_sync(ump
, udf_node
, UDF_C_NODE
,
218 dscr
, sector
, logsector
);
220 DPRINTF(WRITE
, ("udf_write_nodedscr: no wait, async write\n"));
222 error
= udf_write_phys_dscr_async(ump
, udf_node
, UDF_C_NODE
,
223 dscr
, sector
, logsector
, udf_wr_nodedscr_callback
);
224 /* will be UNLOCKED in call back */
228 holdrele(udf_node
->vnode
);
230 udf_node
->outstanding_nodedscr
--;
231 if (udf_node
->outstanding_nodedscr
== 0) {
232 UDF_UNLOCK_NODE(udf_node
, 0);
233 wakeup(&udf_node
->outstanding_nodedscr
);
239 /* --------------------------------------------------------------------- */
242 udf_queue_buf_direct(struct udf_strat_args
*args
)
244 struct udf_mount
*ump
= args
->ump
;
245 struct buf
*buf
= args
->nestbuf
;
247 struct desc_tag
*tag
;
248 struct long_ad
*node_ad_cpy
;
249 uint64_t *lmapping
, *pmapping
, *lmappos
, blknr
, run_start
;
250 uint32_t our_sectornr
, sectornr
;
251 uint32_t lb_size
, buf_offset
, rbuflen
, bpos
;
255 int sector_size
= ump
->discinfo
.sector_size
;
256 int blks
= sector_size
/ DEV_BSIZE
;
257 int len
, buf_len
, sector
, sectors
, run_length
;
258 int what
, class, queue
;
262 KASSERT(buf
->b_iodone
== nestiobuf_iodone
);
264 what
= buf
->b_udf_c_type
;
265 queue
= UDF_SHED_READING
;
266 if ((buf
->b_flags
& B_READ
) == 0) {
268 queue
= UDF_SHED_SEQWRITING
;
269 if (what
== UDF_C_ABSOLUTE
)
270 queue
= UDF_SHED_WRITING
;
271 if (what
== UDF_C_DSCR
)
272 queue
= UDF_SHED_WRITING
;
273 if (what
== UDF_C_NODE
)
274 queue
= UDF_SHED_WRITING
;
277 /* use disc sheduler */
278 class = ump
->discinfo
.mmc_class
;
279 KASSERT((class == MMC_CLASS_UNKN
) || (class == MMC_CLASS_DISC
) ||
280 (ump
->discinfo
.mmc_cur
& MMC_CAP_HW_DEFECTFREE
) ||
281 (ump
->vfs_mountp
->mnt_flag
& MNT_RDONLY
));
283 if (queue
== UDF_SHED_READING
) {
284 DPRINTF(SHEDULE
, ("\nudf_issue_buf READ %p : sector %d type %d,"
285 "b_resid %d, b_bcount %d, b_bufsize %d\n",
286 buf
, (uint32_t) buf
->b_blkno
/ blks
, buf
->b_udf_c_type
,
287 buf
->b_resid
, buf
->b_bcount
, buf
->b_bufsize
));
288 VOP_STRATEGY(ump
->devvp
, buf
);
292 /* (sectorsize == lb_size) for UDF */
293 lb_size
= udf_rw32(ump
->logical_vol
->lb_size
);
294 blknr
= buf
->b_blkno
;
295 our_sectornr
= blknr
/ blks
;
297 if (queue
== UDF_SHED_WRITING
) {
298 DPRINTF(SHEDULE
, ("\nudf_issue_buf WRITE %p : sector %d "
299 "type %d, b_resid %d, b_bcount %d, b_bufsize %d\n",
300 buf
, (uint32_t) buf
->b_blkno
/ blks
, buf
->b_udf_c_type
,
301 buf
->b_resid
, buf
->b_bcount
, buf
->b_bufsize
));
302 KASSERT(buf
->b_udf_c_type
== UDF_C_DSCR
||
303 buf
->b_udf_c_type
== UDF_C_ABSOLUTE
||
304 buf
->b_udf_c_type
== UDF_C_NODE
);
305 udf_fixup_node_internals(ump
, buf
->b_data
, buf
->b_udf_c_type
);
306 VOP_STRATEGY(ump
->devvp
, buf
);
310 /* UDF_SHED_SEQWRITING */
311 KASSERT(queue
== UDF_SHED_SEQWRITING
);
312 DPRINTF(SHEDULE
, ("\nudf_issue_buf SEQWRITE %p : sector XXXX "
313 "type %d, b_resid %d, b_bcount %d, b_bufsize %d\n",
314 buf
, buf
->b_udf_c_type
, buf
->b_resid
, buf
->b_bcount
,
318 * Buffers should not have been allocated to disc addresses yet on
319 * this queue. Note that a buffer can get multiple extents allocated.
321 * lmapping contains lb_num relative to base partition.
323 lmapping
= ump
->la_lmapping
;
324 node_ad_cpy
= ump
->la_node_ad_cpy
;
326 /* logically allocate buf and map it in the file */
327 udf_late_allocate_buf(ump
, buf
, lmapping
, node_ad_cpy
, &vpart_num
);
329 /* if we have FIDs, fixup using the new allocation table */
330 if (buf
->b_udf_c_type
== UDF_C_FIDS
) {
331 buf_len
= buf
->b_bcount
;
335 sectornr
= *lmappos
++;
336 len
= MIN(buf_len
, sector_size
);
337 fidblk
= (uint8_t *) buf
->b_data
+ bpos
;
338 udf_fixup_fid_block(fidblk
, sector_size
,
344 if (buf
->b_udf_c_type
== UDF_C_METADATA_SBM
) {
345 if (buf
->b_lblkno
== 0) {
346 /* update the tag location inside */
347 tag
= (struct desc_tag
*) buf
->b_data
;
348 tag
->tag_loc
= udf_rw32(*lmapping
);
349 udf_validate_tag_and_crc_sums(buf
->b_data
);
352 udf_fixup_node_internals(ump
, buf
->b_data
, buf
->b_udf_c_type
);
355 * Translate new mappings in lmapping to pmappings and try to
356 * conglomerate extents to reduce the number of writes.
358 * pmapping to contain lb_nums as used for disc adressing.
360 pmapping
= ump
->la_pmapping
;
361 sectors
= (buf
->b_bcount
+ sector_size
-1) / sector_size
;
362 udf_translate_vtop_list(ump
, sectors
, vpart_num
, lmapping
, pmapping
);
364 for (sector
= 0; sector
< sectors
; sector
++) {
365 buf_offset
= sector
* sector_size
;
366 DPRINTF(WRITE
, ("\tprocessing rel sector %d\n", sector
));
368 DPRINTF(WRITE
, ("\tissue write sector %"PRIu64
"\n",
371 run_start
= pmapping
[sector
];
373 while (sector
< sectors
-1) {
374 if (pmapping
[sector
+1] != pmapping
[sector
]+1)
380 /* nest an iobuf for the extent */
381 rbuflen
= run_length
* sector_size
;
382 rblk
= run_start
* (sector_size
/DEV_BSIZE
);
384 nestbuf
= getiobuf(NULL
, true);
385 nestiobuf_setup(buf
, nestbuf
, buf_offset
, rbuflen
);
386 /* nestbuf is B_ASYNC */
388 /* identify this nestbuf */
389 nestbuf
->b_lblkno
= sector
;
390 assert(nestbuf
->b_vp
== buf
->b_vp
);
392 /* CD shedules on raw blkno */
393 nestbuf
->b_blkno
= rblk
;
394 nestbuf
->b_proc
= NULL
;
395 nestbuf
->b_rawblkno
= rblk
;
396 nestbuf
->b_udf_c_type
= UDF_C_PROCESSED
;
398 VOP_STRATEGY(ump
->devvp
, nestbuf
);
404 udf_discstrat_init_direct(struct udf_strat_args
*args
)
406 struct udf_mount
*ump
= args
->ump
;
407 struct strat_private
*priv
= PRIV(ump
);
410 KASSERT(priv
== NULL
);
411 ump
->strategy_private
= malloc(sizeof(struct strat_private
),
412 M_UDFTEMP
, M_WAITOK
);
413 priv
= ump
->strategy_private
;
414 memset(priv
, 0 , sizeof(struct strat_private
));
417 * Initialise pool for descriptors associated with nodes. This is done
418 * in lb_size units though currently lb_size is dictated to be
421 memset(&priv
->desc_pool
, 0, sizeof(struct pool
));
423 lb_size
= udf_rw32(ump
->logical_vol
->lb_size
);
424 pool_init(&priv
->desc_pool
, lb_size
, 0, 0, 0, "udf_desc_pool", NULL
,
430 udf_discstrat_finish_direct(struct udf_strat_args
*args
)
432 struct udf_mount
*ump
= args
->ump
;
433 struct strat_private
*priv
= PRIV(ump
);
435 /* destroy our pool */
436 pool_destroy(&priv
->desc_pool
);
438 /* free our private space */
439 free(ump
->strategy_private
, M_UDFTEMP
);
440 ump
->strategy_private
= NULL
;
443 /* --------------------------------------------------------------------- */
445 struct udf_strategy udf_strat_direct
=
447 udf_getblank_nodedscr_direct
,
448 udf_free_nodedscr_direct
,
449 udf_read_nodedscr_direct
,
450 udf_write_nodedscr_direct
,
451 udf_queue_buf_direct
,
452 udf_discstrat_init_direct
,
453 udf_discstrat_finish_direct