NetBSD/sys/fs/udf/udf_readwrite.c
reinoud 42866dd28d Cleanup VAT writout. To prevent issues with the sequential writing strategy
trying to write on blocks that are lost due to the synchronisation, don't just
bluntly do synchronize device caches, but split out on strategies.
2016-05-24 09:55:57 +00:00

737 lines
19 KiB
C

/* $NetBSD: udf_readwrite.c,v 1.12 2016/05/24 09:55:57 reinoud Exp $ */
/*
* Copyright (c) 2007, 2008 Reinoud Zandijk
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
*/
#include <sys/cdefs.h>
#ifndef lint
__KERNEL_RCSID(0, "$NetBSD: udf_readwrite.c,v 1.12 2016/05/24 09:55:57 reinoud Exp $");
#endif /* not lint */
#if defined(_KERNEL_OPT)
#include "opt_compat_netbsd.h"
#endif
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/sysctl.h>
#include <sys/namei.h>
#include <sys/proc.h>
#include <sys/kernel.h>
#include <sys/vnode.h>
#include <miscfs/genfs/genfs_node.h>
#include <sys/mount.h>
#include <sys/buf.h>
#include <sys/file.h>
#include <sys/device.h>
#include <sys/disklabel.h>
#include <sys/ioctl.h>
#include <sys/malloc.h>
#include <sys/dirent.h>
#include <sys/stat.h>
#include <sys/conf.h>
#include <sys/kauth.h>
#include <sys/kthread.h>
#include <dev/clock_subr.h>
#include <fs/udf/ecma167-udf.h>
#include <fs/udf/udf_mount.h>
#include "udf.h"
#include "udf_subr.h"
#include "udf_bswap.h"
#define VTOI(vnode) ((struct udf_node *) vnode->v_data)
/* --------------------------------------------------------------------- */
void
udf_fixup_fid_block(uint8_t *blob, int lb_size,
int rfix_pos, int max_rfix_pos, uint32_t lb_num)
{
struct fileid_desc *fid;
uint8_t *fid_pos;
int fid_len, found;
/* needs to be word aligned */
KASSERT(rfix_pos % 4 == 0);
/* first resync with the FID stream !!! */
found = 0;
while (rfix_pos + sizeof(struct desc_tag) <= max_rfix_pos) {
fid_pos = blob + rfix_pos;
fid = (struct fileid_desc *) fid_pos;
if (udf_rw16(fid->tag.id) == TAGID_FID) {
if (udf_check_tag((union dscrptr *) fid) == 0)
found = 1;
}
if (found)
break;
/* try next location; can only be 4 bytes aligned */
rfix_pos += 4;
}
/* walk over the fids */
fid_pos = blob + rfix_pos;
while (rfix_pos + sizeof(struct desc_tag) <= max_rfix_pos) {
fid = (struct fileid_desc *) fid_pos;
if (udf_rw16(fid->tag.id) != TAGID_FID) {
/* end of FID stream; end of directory or currupted */
break;
}
/* update sector number and recalculate checkum */
fid->tag.tag_loc = udf_rw32(lb_num);
udf_validate_tag_sum((union dscrptr *) fid);
/* if the FID crosses the memory, we're done! */
if (rfix_pos + UDF_FID_SIZE >= max_rfix_pos)
break;
fid_len = udf_fidsize(fid);
fid_pos += fid_len;
rfix_pos += fid_len;
}
}
void
udf_fixup_internal_extattr(uint8_t *blob, uint32_t lb_num)
{
struct desc_tag *tag;
struct file_entry *fe;
struct extfile_entry *efe;
struct extattrhdr_desc *eahdr;
int l_ea;
/* get information from fe/efe */
tag = (struct desc_tag *) blob;
switch (udf_rw16(tag->id)) {
case TAGID_FENTRY :
fe = (struct file_entry *) blob;
l_ea = udf_rw32(fe->l_ea);
eahdr = (struct extattrhdr_desc *) fe->data;
break;
case TAGID_EXTFENTRY :
efe = (struct extfile_entry *) blob;
l_ea = udf_rw32(efe->l_ea);
eahdr = (struct extattrhdr_desc *) efe->data;
break;
case TAGID_INDIRECTENTRY :
case TAGID_ALLOCEXTENT :
case TAGID_EXTATTR_HDR :
return;
default:
panic("%s: passed bad tag\n", __func__);
}
/* something recorded here? (why am i called?) */
if (l_ea == 0)
return;
#if 0
/* check extended attribute tag */
/* TODO XXX what to do when we encounter an error here? */
error = udf_check_tag(eahdr);
if (error)
return; /* for now */
if (udf_rw16(eahdr->tag.id) != TAGID_EXTATTR_HDR)
return; /* for now */
error = udf_check_tag_payload(eahdr, sizeof(struct extattrhdr_desc));
if (error)
return; /* for now */
#endif
DPRINTF(EXTATTR, ("node fixup: found %d bytes of extended attributes\n",
l_ea));
/* fixup eahdr tag */
eahdr->tag.tag_loc = udf_rw32(lb_num);
udf_validate_tag_and_crc_sums((union dscrptr *) eahdr);
}
void
udf_fixup_node_internals(struct udf_mount *ump, uint8_t *blob, int udf_c_type)
{
struct desc_tag *tag, *sbm_tag;
struct file_entry *fe;
struct extfile_entry *efe;
struct alloc_ext_entry *ext;
uint32_t lb_size, lb_num;
uint32_t intern_pos, max_intern_pos;
int icbflags, addr_type, file_type, intern, has_fids, has_sbm, l_ea;
lb_size = udf_rw32(ump->logical_vol->lb_size);
/* if its not a node we're done */
if (udf_c_type != UDF_C_NODE)
return;
/* NOTE this could also be done in write_internal */
/* start of a descriptor */
l_ea = 0;
has_fids = 0;
has_sbm = 0;
intern = 0;
file_type = 0;
max_intern_pos = intern_pos = lb_num = 0; /* shut up gcc! */
tag = (struct desc_tag *) blob;
switch (udf_rw16(tag->id)) {
case TAGID_FENTRY :
fe = (struct file_entry *) tag;
l_ea = udf_rw32(fe->l_ea);
icbflags = udf_rw16(fe->icbtag.flags);
addr_type = (icbflags & UDF_ICB_TAG_FLAGS_ALLOC_MASK);
file_type = fe->icbtag.file_type;
intern = (addr_type == UDF_ICB_INTERN_ALLOC);
intern_pos = UDF_FENTRY_SIZE + l_ea;
max_intern_pos = intern_pos + udf_rw64(fe->inf_len);
lb_num = udf_rw32(fe->tag.tag_loc);
break;
case TAGID_EXTFENTRY :
efe = (struct extfile_entry *) tag;
l_ea = udf_rw32(efe->l_ea);
icbflags = udf_rw16(efe->icbtag.flags);
addr_type = (icbflags & UDF_ICB_TAG_FLAGS_ALLOC_MASK);
file_type = efe->icbtag.file_type;
intern = (addr_type == UDF_ICB_INTERN_ALLOC);
intern_pos = UDF_EXTFENTRY_SIZE + l_ea;
max_intern_pos = intern_pos + udf_rw64(efe->inf_len);
lb_num = udf_rw32(efe->tag.tag_loc);
break;
case TAGID_INDIRECTENTRY :
case TAGID_EXTATTR_HDR :
break;
case TAGID_ALLOCEXTENT :
/* force crclen to 8 for UDF version < 2.01 */
ext = (struct alloc_ext_entry *) tag;
if (udf_rw16(ump->logvol_info->min_udf_readver) <= 0x200)
ext->tag.desc_crc_len = udf_rw16(8);
break;
default:
panic("%s: passed bad tag\n", __func__);
break;
}
/* determine what to fix if its internally recorded */
if (intern) {
has_fids = (file_type == UDF_ICB_FILETYPE_DIRECTORY) ||
(file_type == UDF_ICB_FILETYPE_STREAMDIR);
has_sbm = (file_type == UDF_ICB_FILETYPE_META_BITMAP);
}
/* fixup internal extended attributes if present */
if (l_ea)
udf_fixup_internal_extattr(blob, lb_num);
/* fixup fids lb numbers */
if (has_fids)
udf_fixup_fid_block(blob, lb_size, intern_pos,
max_intern_pos, lb_num);
/* fixup space bitmap descriptor */
if (has_sbm) {
sbm_tag = (struct desc_tag *) (blob + intern_pos);
sbm_tag->tag_loc = tag->tag_loc;
udf_validate_tag_and_crc_sums((uint8_t *) sbm_tag);
}
udf_validate_tag_and_crc_sums(blob);
}
/* --------------------------------------------------------------------- */
/*
* Set of generic descriptor readers and writers and their helper functions.
* Descriptors inside `logical space' i.e. inside logically mapped partitions
* can never be longer than one logical sector.
*
* NOTE that these functions *can* be used by the sheduler backends to read
* node descriptors too.
*
* For reading, the size of allocated piece is returned in multiple of sector
* size due to udf_calc_udf_malloc_size().
*/
/* SYNC reading of n blocks from specified sector */
int
udf_read_phys_sectors(struct udf_mount *ump, int what, void *blob,
uint32_t start, uint32_t sectors)
{
struct buf *buf, *nestbuf;
uint32_t buf_offset;
off_t lblkno, rblkno;
int sector_size = ump->discinfo.sector_size;
int blks = sector_size / DEV_BSIZE;
int piece;
int error;
DPRINTF(READ, ("udf_intbreadn() : sectors = %d, sector_size = %d\n",
sectors, sector_size));
buf = getiobuf(ump->devvp, true);
buf->b_flags = B_READ;
buf->b_cflags = BC_BUSY; /* needed? */
buf->b_iodone = NULL;
buf->b_data = blob;
buf->b_bcount = sectors * sector_size;
buf->b_resid = buf->b_bcount;
buf->b_bufsize = buf->b_bcount;
buf->b_private = NULL; /* not needed yet */
BIO_SETPRIO(buf, BPRIO_DEFAULT);
buf->b_lblkno = buf->b_blkno = buf->b_rawblkno = start * blks;
buf->b_proc = NULL;
error = 0;
buf_offset = 0;
rblkno = start;
lblkno = 0;
while ((sectors > 0) && (error == 0)) {
piece = MIN(MAXPHYS/sector_size, sectors);
DPRINTF(READ, ("read in %d + %d\n", (uint32_t) rblkno, piece));
nestbuf = getiobuf(NULL, true);
nestiobuf_setup(buf, nestbuf, buf_offset, piece * sector_size);
/* nestbuf is B_ASYNC */
/* identify this nestbuf */
nestbuf->b_lblkno = lblkno;
/* CD shedules on raw blkno */
nestbuf->b_blkno = rblkno * blks;
nestbuf->b_proc = NULL;
nestbuf->b_rawblkno = rblkno * blks;
nestbuf->b_udf_c_type = what;
udf_discstrat_queuebuf(ump, nestbuf);
lblkno += piece;
rblkno += piece;
buf_offset += piece * sector_size;
sectors -= piece;
}
error = biowait(buf);
putiobuf(buf);
return error;
}
/* synchronous generic descriptor read */
int
udf_read_phys_dscr(struct udf_mount *ump, uint32_t sector,
struct malloc_type *mtype, union dscrptr **dstp)
{
union dscrptr *dst, *new_dst;
uint8_t *pos;
int sectors, dscrlen;
int i, error, sector_size;
sector_size = ump->discinfo.sector_size;
*dstp = dst = NULL;
dscrlen = sector_size;
/* read initial piece */
dst = malloc(sector_size, mtype, M_WAITOK);
error = udf_read_phys_sectors(ump, UDF_C_DSCR, dst, sector, 1);
DPRINTFIF(DESCRIPTOR, error, ("read error (%d)\n", error));
if (!error) {
/* check if its a valid tag */
error = udf_check_tag(dst);
if (error) {
/* check if its an empty block */
pos = (uint8_t *) dst;
for (i = 0; i < sector_size; i++, pos++) {
if (*pos) break;
}
if (i == sector_size) {
/* return no error but with no dscrptr */
/* dispose first block */
free(dst, mtype);
return 0;
}
}
/* calculate descriptor size */
dscrlen = udf_tagsize(dst, sector_size);
}
DPRINTFIF(DESCRIPTOR, error, ("bad tag checksum\n"));
if (!error && (dscrlen > sector_size)) {
DPRINTF(DESCRIPTOR, ("multi block descriptor read\n"));
/*
* Read the rest of descriptor. Since it is only used at mount
* time its overdone to define and use a specific udf_intbreadn
* for this alone.
*/
new_dst = realloc(dst, dscrlen, mtype, M_WAITOK);
if (new_dst == NULL) {
free(dst, mtype);
return ENOMEM;
}
dst = new_dst;
sectors = (dscrlen + sector_size -1) / sector_size;
DPRINTF(DESCRIPTOR, ("dscrlen = %d (%d blk)\n", dscrlen, sectors));
pos = (uint8_t *) dst + sector_size;
error = udf_read_phys_sectors(ump, UDF_C_DSCR, pos,
sector + 1, sectors-1);
DPRINTFIF(DESCRIPTOR, error, ("read error on multi (%d)\n",
error));
}
if (!error) {
error = udf_check_tag_payload(dst, dscrlen);
DPRINTFIF(DESCRIPTOR, error, ("bad payload check sum\n"));
}
if (error && dst) {
free(dst, mtype);
dst = NULL;
}
*dstp = dst;
return error;
}
static void
udf_write_phys_buf(struct udf_mount *ump, int what, struct buf *buf)
{
struct buf *nestbuf;
uint32_t buf_offset;
off_t lblkno, rblkno;
int sector_size = ump->discinfo.sector_size;
int blks = sector_size / DEV_BSIZE;
uint32_t sectors;
int piece;
int error;
sectors = buf->b_bcount / sector_size;
DPRINTF(WRITE, ("udf_intbwriten() : sectors = %d, sector_size = %d\n",
sectors, sector_size));
/* don't forget to increase pending count for the bwrite itself */
/* panic("NO WRITING\n"); */
if (buf->b_vp) {
mutex_enter(buf->b_vp->v_interlock);
buf->b_vp->v_numoutput++;
mutex_exit(buf->b_vp->v_interlock);
}
error = 0;
buf_offset = 0;
rblkno = buf->b_blkno / blks;
lblkno = 0;
while ((sectors > 0) && (error == 0)) {
piece = MIN(MAXPHYS/sector_size, sectors);
DPRINTF(WRITE, ("write out %d + %d\n",
(uint32_t) rblkno, piece));
nestbuf = getiobuf(NULL, true);
nestiobuf_setup(buf, nestbuf, buf_offset, piece * sector_size);
/* nestbuf is B_ASYNC */
/* identify this nestbuf */
nestbuf->b_lblkno = lblkno;
/* CD shedules on raw blkno */
nestbuf->b_blkno = rblkno * blks;
nestbuf->b_proc = NULL;
nestbuf->b_rawblkno = rblkno * blks;
nestbuf->b_udf_c_type = what;
udf_discstrat_queuebuf(ump, nestbuf);
lblkno += piece;
rblkno += piece;
buf_offset += piece * sector_size;
sectors -= piece;
}
}
/* SYNC writing of n blocks from specified sector */
int
udf_write_phys_sectors(struct udf_mount *ump, int what, void *blob,
uint32_t start, uint32_t sectors)
{
struct vnode *vp;
struct buf *buf;
int sector_size = ump->discinfo.sector_size;
int blks = sector_size / DEV_BSIZE;
int error;
/* get transfer buffer */
vp = ump->devvp;
buf = getiobuf(vp, true);
buf->b_flags = B_WRITE;
buf->b_cflags = BC_BUSY; /* needed? */
buf->b_iodone = NULL;
buf->b_data = blob;
buf->b_bcount = sectors * sector_size;
buf->b_resid = buf->b_bcount;
buf->b_bufsize = buf->b_bcount;
buf->b_private = NULL; /* not needed yet */
BIO_SETPRIO(buf, BPRIO_DEFAULT);
buf->b_lblkno = buf->b_blkno = buf->b_rawblkno = start * blks;
buf->b_proc = NULL;
/* do the write, wait and return error */
udf_write_phys_buf(ump, what, buf);
error = biowait(buf);
putiobuf(buf);
return error;
}
/* synchronous generic descriptor write */
int
udf_write_phys_dscr_sync(struct udf_mount *ump, struct udf_node *udf_node, int what,
union dscrptr *dscr, uint32_t sector, uint32_t logsector)
{
struct vnode *vp;
struct buf *buf;
int sector_size = ump->discinfo.sector_size;
int blks = sector_size / DEV_BSIZE;
int dscrlen;
int error;
/* set sector number in the descriptor and validate */
dscr->tag.tag_loc = udf_rw32(logsector);
udf_validate_tag_and_crc_sums(dscr);
/* calculate descriptor size */
dscrlen = udf_tagsize(dscr, sector_size);
/* get transfer buffer */
vp = udf_node ? udf_node->vnode : ump->devvp;
buf = getiobuf(vp, true);
buf->b_flags = B_WRITE;
buf->b_cflags = BC_BUSY; /* needed? */
buf->b_iodone = NULL;
buf->b_data = (void *) dscr;
buf->b_bcount = dscrlen;
buf->b_resid = buf->b_bcount;
buf->b_bufsize = buf->b_bcount;
buf->b_private = NULL; /* not needed yet */
BIO_SETPRIO(buf, BPRIO_DEFAULT);
buf->b_lblkno = buf->b_blkno = buf->b_rawblkno = sector * blks;
buf->b_proc = NULL;
/* do the write, wait and return error */
udf_write_phys_buf(ump, what, buf);
error = biowait(buf);
putiobuf(buf);
return error;
}
/* asynchronous generic descriptor write */
int
udf_write_phys_dscr_async(struct udf_mount *ump, struct udf_node *udf_node,
int what, union dscrptr *dscr,
uint32_t sector, uint32_t logsector,
void (*dscrwr_callback)(struct buf *))
{
struct vnode *vp;
struct buf *buf;
int dscrlen;
int sector_size = ump->discinfo.sector_size;
int blks = sector_size / DEV_BSIZE;
KASSERT(dscrwr_callback);
DPRINTF(NODE, ("udf_write_phys_dscr_async() called\n"));
/* set sector number in the descriptor and validate */
dscr->tag.tag_loc = udf_rw32(logsector);
udf_validate_tag_and_crc_sums(dscr);
/* calculate descriptor size */
dscrlen = udf_tagsize(dscr, sector_size);
/* get transfer buffer */
vp = udf_node ? udf_node->vnode : ump->devvp;
buf = getiobuf(vp, true);
buf->b_flags = B_WRITE; // | B_ASYNC;
buf->b_cflags = BC_BUSY;
buf->b_iodone = dscrwr_callback;
buf->b_data = dscr;
buf->b_bcount = dscrlen;
buf->b_resid = buf->b_bcount;
buf->b_bufsize = buf->b_bcount;
buf->b_private = NULL; /* not needed yet */
BIO_SETPRIO(buf, BPRIO_DEFAULT);
buf->b_lblkno = buf->b_blkno = buf->b_rawblkno = sector * blks;
buf->b_proc = NULL;
/* do the write and return no error */
udf_write_phys_buf(ump, what, buf);
return 0;
}
/* --------------------------------------------------------------------- */
/* disc strategy dispatchers */
int
udf_create_logvol_dscr(struct udf_mount *ump, struct udf_node *udf_node, struct long_ad *icb,
union dscrptr **dscrptr)
{
struct udf_strategy *strategy = ump->strategy;
struct udf_strat_args args;
int error;
KASSERT(strategy);
args.ump = ump;
args.udf_node = udf_node;
args.icb = icb;
args.dscr = NULL;
error = (strategy->create_logvol_dscr)(&args);
*dscrptr = args.dscr;
return error;
}
void
udf_free_logvol_dscr(struct udf_mount *ump, struct long_ad *icb,
void *dscr)
{
struct udf_strategy *strategy = ump->strategy;
struct udf_strat_args args;
KASSERT(strategy);
args.ump = ump;
args.icb = icb;
args.dscr = dscr;
(strategy->free_logvol_dscr)(&args);
}
int
udf_read_logvol_dscr(struct udf_mount *ump, struct long_ad *icb,
union dscrptr **dscrptr)
{
struct udf_strategy *strategy = ump->strategy;
struct udf_strat_args args;
int error;
KASSERT(strategy);
args.ump = ump;
args.icb = icb;
args.dscr = NULL;
error = (strategy->read_logvol_dscr)(&args);
*dscrptr = args.dscr;
return error;
}
int
udf_write_logvol_dscr(struct udf_node *udf_node, union dscrptr *dscr,
struct long_ad *icb, int waitfor)
{
struct udf_strategy *strategy = udf_node->ump->strategy;
struct udf_strat_args args;
int error;
KASSERT(strategy);
args.ump = udf_node->ump;
args.udf_node = udf_node;
args.icb = icb;
args.dscr = dscr;
args.waitfor = waitfor;
error = (strategy->write_logvol_dscr)(&args);
return error;
}
void
udf_discstrat_queuebuf(struct udf_mount *ump, struct buf *nestbuf)
{
struct udf_strategy *strategy = ump->strategy;
struct udf_strat_args args;
KASSERT(strategy);
args.ump = ump;
args.nestbuf = nestbuf;
(strategy->queuebuf)(&args);
}
void
udf_synchronise_caches(struct udf_mount *ump)
{
struct udf_strategy *strategy = ump->strategy;
struct udf_strat_args args;
KASSERT(strategy);
args.ump = ump;
(strategy->sync_caches)(&args);
}
void
udf_discstrat_init(struct udf_mount *ump)
{
struct udf_strategy *strategy = ump->strategy;
struct udf_strat_args args;
KASSERT(strategy);
args.ump = ump;
(strategy->discstrat_init)(&args);
}
void udf_discstrat_finish(struct udf_mount *ump)
{
struct udf_strategy *strategy = ump->strategy;
struct udf_strat_args args;
/* strategy might not have been set, so ignore if not set */
if (strategy) {
args.ump = ump;
(strategy->discstrat_finish)(&args);
}
}
/* --------------------------------------------------------------------- */