freebsd-nq/usr.sbin/makefs/msdos/msdosfs_vnops.c
Warner Losh 370e009188 makefs: make msdos creation go fast
Add missing brelse(bp). Without it the cache grows and we have a n^2
lookup.  I'm not entirely sure why we read the block before we write it
back out, since the only side effect of that is to allocate memory,
clear the memory, read it in from disk, throw it away with the contents
of the file being written out. We likely should just do a getblk() here
instead, but even with all that, this takes the time it takes to create
a 150MB msdos fs image down from 5 minutes to 30 seconds.

Old: 317.663u 0.685s 5:18.34 100.0%	198+360k 0+19io 1009pf+0w
New: 7.330u 23.841s 0:31.17 100.0%	198+360k 0+250522io 4pf+0w

See code review for how we got this. tl;dr: netbsd move brelse
into bwrite and we picked up msdos code after that, but not the
move. That change should be picked up later.

Sponsored by:		Netflix
Reviewed by:		emaste
MFC After:		1 day (13.2 is coming fast)
Differential Revision:	https://reviews.freebsd.org/D39025
2023-03-13 14:38:36 -06:00

646 lines
19 KiB
C

/* $NetBSD: msdosfs_vnops.c,v 1.19 2017/04/13 17:10:12 christos Exp $ */
/*-
* SPDX-License-Identifier: BSD-4-Clause
*
* Copyright (C) 1994, 1995, 1997 Wolfgang Solfrank.
* Copyright (C) 1994, 1995, 1997 TooLs GmbH.
* All rights reserved.
* Original code by Paul Popelka (paulp@uts.amdahl.com) (see below).
*
* 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.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by TooLs GmbH.
* 4. The name of TooLs GmbH may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY TOOLS GMBH ``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 TOOLS GMBH 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.
*/
/*-
* Written by Paul Popelka (paulp@uts.amdahl.com)
*
* You can do anything you want with this software, just don't say you wrote
* it, and don't remove this notice.
*
* This software is provided "as is".
*
* The author supplies this software to be publicly redistributed on the
* understanding that the author is not responsible for the correct
* functioning of this software in any circumstances and is not liable for
* any damages caused by this software.
*
* October 1992
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/errno.h>
#include <sys/mman.h>
#include <sys/time.h>
#include <fcntl.h>
#include <stdbool.h>
#include <stdio.h>
#include <string.h>
#include <time.h>
#include <unistd.h>
#include <fs/msdosfs/bpb.h>
#include "msdos/denode.h"
#include <fs/msdosfs/fat.h>
#include <fs/msdosfs/msdosfsmount.h>
#include "makefs.h"
#include "msdos.h"
/*
* Some general notes:
*
* In the ufs filesystem the inodes, superblocks, and indirect blocks are
* read/written using the vnode for the filesystem. Blocks that represent
* the contents of a file are read/written using the vnode for the file
* (including directories when they are read/written as files). This
* presents problems for the dos filesystem because data that should be in
* an inode (if dos had them) resides in the directory itself. Since we
* must update directory entries without the benefit of having the vnode
* for the directory we must use the vnode for the filesystem. This means
* that when a directory is actually read/written (via read, write, or
* readdir, or seek) we must use the vnode for the filesystem instead of
* the vnode for the directory as would happen in ufs. This is to insure we
* retrieve the correct block from the buffer cache since the hash value is
* based upon the vnode address and the desired block number.
*/
static int msdosfs_wfile(const char *, struct denode *, fsnode *);
static void unix2fattime(const struct timespec *tsp, uint16_t *ddp,
uint16_t *dtp);
static void
msdosfs_times(struct denode *dep, const struct stat *st)
{
if (stampst.st_ino)
st = &stampst;
#ifdef HAVE_STRUCT_STAT_BIRTHTIME
unix2fattime(&st->st_birthtim, &dep->de_CDate, &dep->de_CTime);
#else
unix2fattime(&st->st_ctim, &dep->de_CDate, &dep->de_CTime);
#endif
unix2fattime(&st->st_atim, &dep->de_ADate, NULL);
unix2fattime(&st->st_mtim, &dep->de_MDate, &dep->de_MTime);
}
static void
unix2fattime(const struct timespec *tsp, uint16_t *ddp, uint16_t *dtp)
{
time_t t1;
struct tm lt = {0};
t1 = tsp->tv_sec;
localtime_r(&t1, &lt);
unsigned long fat_time = ((lt.tm_year - 80) << 25) |
((lt.tm_mon + 1) << 21) |
(lt.tm_mday << 16) |
(lt.tm_hour << 11) |
(lt.tm_min << 5) |
(lt.tm_sec >> 1);
if (ddp != NULL)
*ddp = (uint16_t)(fat_time >> 16);
if (dtp != NULL)
*dtp = (uint16_t)fat_time;
}
/*
* When we search a directory the blocks containing directory entries are
* read and examined. The directory entries contain information that would
* normally be in the inode of a unix filesystem. This means that some of
* a directory's contents may also be in memory resident denodes (sort of
* an inode). This can cause problems if we are searching while some other
* process is modifying a directory. To prevent one process from accessing
* incompletely modified directory information we depend upon being the
* sole owner of a directory block. bread/brelse provide this service.
* This being the case, when a process modifies a directory it must first
* acquire the disk block that contains the directory entry to be modified.
* Then update the disk block and the denode, and then write the disk block
* out to disk. This way disk blocks containing directory entries and in
* memory denode's will be in synch.
*/
static int
msdosfs_findslot(struct denode *dp, struct componentname *cnp)
{
daddr_t bn;
int error;
int slotcount;
int slotoffset = 0;
int frcn;
u_long cluster;
int blkoff;
u_int diroff;
int blsize;
struct msdosfsmount *pmp;
struct m_buf *bp = 0;
struct direntry *dep;
u_char dosfilename[12];
int wincnt = 1;
int chksum = -1, chksum_ok;
int olddos = 1;
pmp = dp->de_pmp;
switch (unix2dosfn((const u_char *)cnp->cn_nameptr, dosfilename,
cnp->cn_namelen, 0)) {
case 0:
return (EINVAL);
case 1:
break;
case 2:
wincnt = winSlotCnt((const u_char *)cnp->cn_nameptr,
cnp->cn_namelen) + 1;
break;
case 3:
olddos = 0;
wincnt = winSlotCnt((const u_char *)cnp->cn_nameptr,
cnp->cn_namelen) + 1;
break;
}
if (pmp->pm_flags & MSDOSFSMNT_SHORTNAME)
wincnt = 1;
/*
* Suppress search for slots unless creating
* file and at end of pathname, in which case
* we watch for a place to put the new file in
* case it doesn't already exist.
*/
slotcount = 0;
MSDOSFS_DPRINTF(("%s(): dos filename: %s\n", __func__, dosfilename));
/*
* Search the directory pointed at by vdp for the name pointed at
* by cnp->cn_nameptr.
*/
/*
* The outer loop ranges over the clusters that make up the
* directory. Note that the root directory is different from all
* other directories. It has a fixed number of blocks that are not
* part of the pool of allocatable clusters. So, we treat it a
* little differently. The root directory starts at "cluster" 0.
*/
diroff = 0;
for (frcn = 0; diroff < dp->de_FileSize; frcn++) {
if ((error = pcbmap(dp, frcn, &bn, &cluster, &blsize)) != 0) {
if (error == E2BIG)
break;
return (error);
}
error = bread((void *)pmp->pm_devvp, bn, blsize, 0, &bp);
if (error) {
return (error);
}
for (blkoff = 0; blkoff < blsize;
blkoff += sizeof(struct direntry),
diroff += sizeof(struct direntry)) {
dep = (struct direntry *)(bp->b_data + blkoff);
/*
* If the slot is empty and we are still looking
* for an empty then remember this one. If the
* slot is not empty then check to see if it
* matches what we are looking for. If the slot
* has never been filled with anything, then the
* remainder of the directory has never been used,
* so there is no point in searching it.
*/
if (dep->deName[0] == SLOT_EMPTY ||
dep->deName[0] == SLOT_DELETED) {
/*
* Drop memory of previous long matches
*/
chksum = -1;
if (slotcount < wincnt) {
slotcount++;
slotoffset = diroff;
}
if (dep->deName[0] == SLOT_EMPTY) {
brelse(bp);
goto notfound;
}
} else {
/*
* If there wasn't enough space for our
* winentries, forget about the empty space
*/
if (slotcount < wincnt)
slotcount = 0;
/*
* Check for Win95 long filename entry
*/
if (dep->deAttributes == ATTR_WIN95) {
if (pmp->pm_flags & MSDOSFSMNT_SHORTNAME)
continue;
chksum = winChkName(
(const u_char *)cnp->cn_nameptr,
cnp->cn_namelen,
(struct winentry *)dep, chksum);
continue;
}
/*
* Ignore volume labels (anywhere, not just
* the root directory).
*/
if (dep->deAttributes & ATTR_VOLUME) {
chksum = -1;
continue;
}
/*
* Check for a checksum or name match
*/
chksum_ok = (chksum == winChksum(dep->deName));
if (!chksum_ok
&& (!olddos || memcmp(dosfilename, dep->deName, 11))) {
chksum = -1;
continue;
}
MSDOSFS_DPRINTF(("%s(): match blkoff %d, diroff %u\n",
__func__, blkoff, diroff));
/*
* Remember where this directory
* entry came from for whoever did
* this lookup.
*/
dp->de_fndoffset = diroff;
dp->de_fndcnt = 0;
return EEXIST;
}
} /* for (blkoff = 0; .... */
/*
* Release the buffer holding the directory cluster just
* searched.
*/
brelse(bp);
} /* for (frcn = 0; ; frcn++) */
notfound:
/*
* We hold no disk buffers at this point.
*/
/*
* If we get here we didn't find the entry we were looking for. But
* that's ok if we are creating or renaming and are at the end of
* the pathname and the directory hasn't been removed.
*/
MSDOSFS_DPRINTF(("%s(): refcnt %ld, slotcount %d, slotoffset %d\n",
__func__, dp->de_refcnt, slotcount, slotoffset));
/*
* Fixup the slot description to point to the place where
* we might put the new DOS direntry (putting the Win95
* long name entries before that)
*/
if (!slotcount) {
slotcount = 1;
slotoffset = diroff;
}
if (wincnt > slotcount) {
slotoffset += sizeof(struct direntry) * (wincnt - slotcount);
}
/*
* Return an indication of where the new directory
* entry should be put.
*/
dp->de_fndoffset = slotoffset;
dp->de_fndcnt = wincnt - 1;
/*
* We return with the directory locked, so that
* the parameters we set up above will still be
* valid if we actually decide to do a direnter().
* We return ni_vp == NULL to indicate that the entry
* does not currently exist; we leave a pointer to
* the (locked) directory inode in ndp->ni_dvp.
*
* NB - if the directory is unlocked, then this
* information cannot be used.
*/
return 0;
}
/*
* Create a regular file. On entry the directory to contain the file being
* created is locked. We must release before we return.
*/
struct denode *
msdosfs_mkfile(const char *path, struct denode *pdep, fsnode *node)
{
struct componentname cn;
struct denode ndirent;
struct denode *dep;
int error;
struct stat *st = &node->inode->st;
cn.cn_nameptr = node->name;
cn.cn_namelen = strlen(node->name);
MSDOSFS_DPRINTF(("%s(name %s, mode 0%o size %zu)\n",
__func__, node->name, st->st_mode, (size_t)st->st_size));
/*
* If this is the root directory and there is no space left we
* can't do anything. This is because the root directory can not
* change size.
*/
if (pdep->de_StartCluster == MSDOSFSROOT
&& pdep->de_fndoffset >= pdep->de_FileSize) {
error = ENOSPC;
goto bad;
}
/*
* Create a directory entry for the file, then call createde() to
* have it installed. NOTE: DOS files are always executable. We
* use the absence of the owner write bit to make the file
* readonly.
*/
memset(&ndirent, 0, sizeof(ndirent));
if ((error = uniqdosname(pdep, &cn, ndirent.de_Name)) != 0)
goto bad;
ndirent.de_Attributes = (st->st_mode & S_IWUSR) ?
ATTR_ARCHIVE : ATTR_ARCHIVE | ATTR_READONLY;
ndirent.de_StartCluster = 0;
ndirent.de_FileSize = 0;
ndirent.de_pmp = pdep->de_pmp;
ndirent.de_flag = DE_ACCESS | DE_CREATE | DE_UPDATE;
msdosfs_times(&ndirent, &node->inode->st);
if ((error = msdosfs_findslot(pdep, &cn)) != 0)
goto bad;
if ((error = createde(&ndirent, pdep, &dep, &cn)) != 0)
goto bad;
if ((error = msdosfs_wfile(path, dep, node)) != 0)
goto bad;
return dep;
bad:
errno = error;
return NULL;
}
static int
msdosfs_updatede(struct denode *dep)
{
struct m_buf *bp;
struct direntry *dirp;
int error;
dep->de_flag &= ~DE_MODIFIED;
error = m_readde(dep, &bp, &dirp);
if (error)
return error;
DE_EXTERNALIZE(dirp, dep);
error = bwrite(bp);
return error;
}
/*
* Write data to a file or directory.
*/
static int
msdosfs_wfile(const char *path, struct denode *dep, fsnode *node)
{
int error, fd;
size_t osize = dep->de_FileSize;
struct stat *st = &node->inode->st;
size_t nsize, offs;
struct msdosfsmount *pmp = dep->de_pmp;
struct m_buf *bp;
char *dat;
u_long cn = 0;
error = 0; /* XXX: gcc/vax */
MSDOSFS_DPRINTF(("%s(diroff %lu, dirclust %lu, startcluster %lu)\n",
__func__, dep->de_diroffset, dep->de_dirclust,
dep->de_StartCluster));
if (st->st_size == 0)
return 0;
/* Don't bother to try to write files larger than the fs limit */
if (st->st_size > MSDOSFS_FILESIZE_MAX)
return EFBIG;
nsize = st->st_size;
MSDOSFS_DPRINTF(("%s(nsize=%zu, osize=%zu)\n", __func__, nsize, osize));
if (nsize > osize) {
if ((error = deextend(dep, nsize, NULL)) != 0)
return error;
if ((error = msdosfs_updatede(dep)) != 0)
return error;
}
if ((fd = open(path, O_RDONLY)) == -1) {
error = errno;
fprintf(stderr, "open %s: %s\n", path, strerror(error));
return error;
}
if ((dat = mmap(0, nsize, PROT_READ, MAP_FILE | MAP_PRIVATE, fd, 0))
== MAP_FAILED) {
error = errno;
fprintf(stderr, "%s: mmap %s: %s\n", __func__, node->name,
strerror(error));
close(fd);
goto out;
}
close(fd);
for (offs = 0; offs < nsize;) {
int blsize, cpsize;
daddr_t bn;
u_long on = offs & pmp->pm_crbomask;
if ((error = pcbmap(dep, cn++, &bn, NULL, &blsize)) != 0) {
MSDOSFS_DPRINTF(("%s: pcbmap %lu",
__func__, (unsigned long)bn));
goto out;
}
MSDOSFS_DPRINTF(("%s(cn=%lu, bn=%llu, blsize=%d)\n",
__func__, cn, (unsigned long long)bn, blsize));
if ((error = bread((void *)pmp->pm_devvp, bn, blsize, 0,
&bp)) != 0) {
MSDOSFS_DPRINTF(("bread %d\n", error));
goto out;
}
cpsize = MIN((nsize - offs), blsize - on);
memcpy(bp->b_data + on, dat + offs, cpsize);
bwrite(bp);
brelse(bp);
offs += cpsize;
}
munmap(dat, nsize);
return 0;
out:
munmap(dat, nsize);
return error;
}
static const struct {
struct direntry dot;
struct direntry dotdot;
} dosdirtemplate = {
{ ". ", /* the . entry */
ATTR_DIRECTORY, /* file attribute */
0, /* reserved */
0, { 0, 0 }, { 0, 0 }, /* create time & date */
{ 0, 0 }, /* access date */
{ 0, 0 }, /* high bits of start cluster */
{ 210, 4 }, { 210, 4 }, /* modify time & date */
{ 0, 0 }, /* startcluster */
{ 0, 0, 0, 0 } /* filesize */
},
{ ".. ", /* the .. entry */
ATTR_DIRECTORY, /* file attribute */
0, /* reserved */
0, { 0, 0 }, { 0, 0 }, /* create time & date */
{ 0, 0 }, /* access date */
{ 0, 0 }, /* high bits of start cluster */
{ 210, 4 }, { 210, 4 }, /* modify time & date */
{ 0, 0 }, /* startcluster */
{ 0, 0, 0, 0 } /* filesize */
}
};
struct denode *
msdosfs_mkdire(const char *path __unused, struct denode *pdep, fsnode *node)
{
struct denode ndirent;
struct denode *dep;
struct componentname cn;
struct msdosfsmount *pmp = pdep->de_pmp;
int error;
u_long newcluster, pcl, bn;
struct direntry *denp;
struct m_buf *bp;
cn.cn_nameptr = node->name;
cn.cn_namelen = strlen(node->name);
/*
* If this is the root directory and there is no space left we
* can't do anything. This is because the root directory can not
* change size.
*/
if (pdep->de_StartCluster == MSDOSFSROOT
&& pdep->de_fndoffset >= pdep->de_FileSize) {
error = ENOSPC;
goto bad2;
}
/*
* Allocate a cluster to hold the about to be created directory.
*/
error = clusteralloc(pmp, 0, 1, CLUST_EOFE, &newcluster, NULL);
if (error)
goto bad2;
memset(&ndirent, 0, sizeof(ndirent));
ndirent.de_pmp = pmp;
ndirent.de_flag = DE_ACCESS | DE_CREATE | DE_UPDATE;
msdosfs_times(&ndirent, &node->inode->st);
/*
* Now fill the cluster with the "." and ".." entries. And write
* the cluster to disk. This way it is there for the parent
* directory to be pointing at if there were a crash.
*/
bn = cntobn(pmp, newcluster);
MSDOSFS_DPRINTF(("%s(newcluster %lu, bn=%lu)\n",
__func__, newcluster, bn));
/* always succeeds */
bp = getblk((void *)pmp->pm_devvp, bn, pmp->pm_bpcluster, 0, 0, 0);
memset(bp->b_data, 0, pmp->pm_bpcluster);
memcpy(bp->b_data, &dosdirtemplate, sizeof dosdirtemplate);
denp = (struct direntry *)bp->b_data;
putushort(denp[0].deStartCluster, newcluster);
putushort(denp[0].deCDate, ndirent.de_CDate);
putushort(denp[0].deCTime, ndirent.de_CTime);
denp[0].deCHundredth = ndirent.de_CHun;
putushort(denp[0].deADate, ndirent.de_ADate);
putushort(denp[0].deMDate, ndirent.de_MDate);
putushort(denp[0].deMTime, ndirent.de_MTime);
pcl = pdep->de_StartCluster;
MSDOSFS_DPRINTF(("%s(pcl %lu, rootdirblk=%lu)\n", __func__, pcl,
pmp->pm_rootdirblk));
if (FAT32(pmp) && pcl == pmp->pm_rootdirblk)
pcl = 0;
putushort(denp[1].deStartCluster, pcl);
putushort(denp[1].deCDate, ndirent.de_CDate);
putushort(denp[1].deCTime, ndirent.de_CTime);
denp[1].deCHundredth = ndirent.de_CHun;
putushort(denp[1].deADate, ndirent.de_ADate);
putushort(denp[1].deMDate, ndirent.de_MDate);
putushort(denp[1].deMTime, ndirent.de_MTime);
if (FAT32(pmp)) {
putushort(denp[0].deHighClust, newcluster >> 16);
putushort(denp[1].deHighClust, pdep->de_StartCluster >> 16);
} else {
putushort(denp[0].deHighClust, 0);
putushort(denp[1].deHighClust, 0);
}
if ((error = bwrite(bp)) != 0)
goto bad;
/*
* Now build up a directory entry pointing to the newly allocated
* cluster. This will be written to an empty slot in the parent
* directory.
*/
if ((error = uniqdosname(pdep, &cn, ndirent.de_Name)) != 0)
goto bad;
ndirent.de_Attributes = ATTR_DIRECTORY;
ndirent.de_StartCluster = newcluster;
ndirent.de_FileSize = 0;
ndirent.de_pmp = pdep->de_pmp;
if ((error = msdosfs_findslot(pdep, &cn)) != 0)
goto bad;
if ((error = createde(&ndirent, pdep, &dep, &cn)) != 0)
goto bad;
if ((error = msdosfs_updatede(dep)) != 0)
goto bad;
return dep;
bad:
clusterfree(pmp, newcluster);
bad2:
errno = error;
return NULL;
}