diff options
| -rw-r--r-- | usr.bin/Makefile | 2 | ||||
| -rw-r--r-- | usr.bin/mkuzip/Makefile | 11 | ||||
| -rw-r--r-- | usr.bin/mkuzip/mkuzip.8 | 84 | ||||
| -rw-r--r-- | usr.bin/mkuzip/mkuzip.c | 259 |
4 files changed, 356 insertions, 0 deletions
diff --git a/usr.bin/Makefile b/usr.bin/Makefile index 450110d..1949b11 100644 --- a/usr.bin/Makefile +++ b/usr.bin/Makefile @@ -118,6 +118,7 @@ SUBDIR= alias \ mklocale \ mkstr \ mktemp \ + ${_mkuzip} \ msgs \ mt \ ncal \ @@ -270,6 +271,7 @@ _usbhidctl= usbhidctl .endif .if ${MACHINE_ARCH} == "i386" +_mkuzip= mkuzip _ncplist= ncplist _ncplogin= ncplogin _smbutil= smbutil diff --git a/usr.bin/mkuzip/Makefile b/usr.bin/mkuzip/Makefile new file mode 100644 index 0000000..7d1c92a --- /dev/null +++ b/usr.bin/mkuzip/Makefile @@ -0,0 +1,11 @@ +# $FreeBSD$ + +PROG= mkuzip +MAN1= + +DPADD= ${LIBZ} +LDADD= -lz + +WARNS= 4 + +.include <bsd.prog.mk> diff --git a/usr.bin/mkuzip/mkuzip.8 b/usr.bin/mkuzip/mkuzip.8 new file mode 100644 index 0000000..277faf2 --- /dev/null +++ b/usr.bin/mkuzip/mkuzip.8 @@ -0,0 +1,84 @@ +.\" ---------------------------------------------------------------------------- +.\" "THE BEER-WARE LICENSE" (Revision 42): +.\" <sobomax@FreeBSD.ORG> wrote this file. As long as you retain this notice you +.\" can do whatever you want with this stuff. If we meet some day, and you think +.\" this stuff is worth it, you can buy me a beer in return. Maxim Sobolev +.\" ---------------------------------------------------------------------------- +.\" +.\" $FreeBSD$ +.\" +.Dd July 29, 2001 +.Dt MDGZIP 8 +.Os +.Sh NAME +.Nm mkuzip +.Nd compress a +.Xr geom_uzip 4 +image +.Sh SYNOPSIS +.Nm +.Op Fl v +.Op Fl o Ar outfile +.Op Fl s Ar cluster_size +.Ar infile +.Sh DESCRIPTION +The +.Nm +utility compresses a disk image file in such a way that the +.Xr geom_uzip 4 +class will be able to decompress resulting image in run-time +when it loaded into memory. This allows for significant reduction +of memory footprint of memory-based filesystem at the expense of +some CPU time required to decompress the data each time it is +read. Internally, operation is done in two phases as follows: +.Bl -enum +.It +An +.Ar infile +image is split into clusters and each cluster compressed using +.Xr zlib 3 . +.It +Resulting set of clusters along with headers allowing to +independently locate each individual cluster is written into +output file. +.El +.Pp +The options are: +.Bl -tag -width Fl +.It Fl o Ar outfile +Name the output file +.Ar outfile . +The default is to use the input name with the suffix +.Sq .uzip . +.It Fl s Ar cluster_size +Use +.Ar cluster_size +as the size of chunks the file being split up into. Default value +is 16K. The +.Ar cluster_size +should be multiple of block size of the +.Xr geom_uzip 4 +device (usually 512 bytes). +.It Fl v +Display verbose messages. +.El +.Sh NOTES +Compression ratio largely depends on the cluster size used. For +large cluster sizes of (16K and higher) typical compression ratios +are only 1-2% less than those achieved with the +.Xr gzip 1 +utlity. However, it should be kept in mind that larger cluster +sizes lead to higher overhead in the +.Xr geom_uzip 4 +class, as the class has to decompress the whole cluster even if +literally only several bytes from that cluster have to be read. +.Sh SEE ALSO +.Xr gzip 1 , +.Xr zlib 3 , +.Xr geom_uzip 4 , +.Xr boot 8 , +.Xr loader 8 +.Sh DIAGNOSTICS +Exit status is 0 on success and >0 on error. +.Sh AUTHORS +.An Maxim Sobolev Aq sobomax@FreeBSD.org . diff --git a/usr.bin/mkuzip/mkuzip.c b/usr.bin/mkuzip/mkuzip.c new file mode 100644 index 0000000..735529a --- /dev/null +++ b/usr.bin/mkuzip/mkuzip.c @@ -0,0 +1,259 @@ +/* + * ---------------------------------------------------------------------------- + * "THE BEER-WARE LICENSE" (Revision 42): + * <sobomax@FreeBSD.ORG> wrote this file. As long as you retain this notice you + * can do whatever you want with this stuff. If we meet some day, and you think + * this stuff is worth it, you can buy me a beer in return. Maxim Sobolev + * ---------------------------------------------------------------------------- + * + * $FreeBSD$ + * + */ + +#include <sys/types.h> +#include <sys/endian.h> +#include <sys/param.h> +#include <sys/stat.h> +#include <sys/uio.h> +#include <netinet/in.h> +#include <zlib.h> +#include <err.h> +#include <fcntl.h> +#include <signal.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <unistd.h> + +#define CLSTSIZE 16384 +#define DEFAULT_SUFX ".uzip" + +#define CLOOP_MAGIC_LEN 128 +static char CLOOP_MAGIC_START[] = "#!/bin/sh\n#V2.0 Format\n" + "m=geom_uzip\n(kldstat -n $m 2>&-||kldload $m)>&-&&" + "mount_cd9660 /dev/`mdconfig -af $0`.uzip $1\nexit $?\n"; + +/* + * Maximum allowed valid block size (to prevent foot-shooting) + */ +#define MAX_BLKSZ (MAXPHYS - MAXPHYS / 1000 - 12) + +static char *readblock(int, char *, u_int32_t); +static void usage(void); +static void *safe_malloc(size_t); +static void cleanup(void); + +static char *cleanfile = NULL; + +int main(int argc, char **argv) +{ + char *iname, *oname, *obuf, *ibuf; + uint64_t *toc; + int fdr, fdw, i, opt, verbose, tmp; + struct iovec iov[2]; + struct stat sb; + uLongf destlen; + uint64_t offset; + struct cloop_header { + char magic[CLOOP_MAGIC_LEN]; /* cloop magic */ + uint32_t blksz; /* block size */ + uint32_t nblocks; /* number of blocks */ + } hdr; + + memset(&hdr, 0, sizeof(hdr)); + hdr.blksz = CLSTSIZE; + strcpy(hdr.magic, CLOOP_MAGIC_START); + oname = NULL; + verbose = 0; + + while((opt = getopt(argc, argv, "o:s:v")) != -1) { + switch(opt) { + case 'o': + oname = optarg; + break; + + case 's': + tmp = atoi(optarg); + if (tmp <= 0) { + errx(1, "invalid cluster size specified: %s", + optarg); + /* Not reached */ + } + if (tmp % DEV_BSIZE != 0) { + errx(1, "cluster size should be multiple of %d", + DEV_BSIZE); + /* Not reached */ + } + if (tmp > MAX_BLKSZ) { + errx(1, "cluster size can't be more than %d", + MAX_BLKSZ); + /* Not reached */ + } + hdr.blksz = tmp; + break; + + case 'v': + verbose = 1; + break; + + default: + usage(); + /* Not reached */ + } + } + argc -= optind; + argv += optind; + + if (argc != 1) { + usage(); + /* Not reached */ + } + + iname = argv[0]; + if (oname == NULL) { + asprintf(&oname, "%s%s", iname, DEFAULT_SUFX); + if (oname == NULL) { + err(1, "can't allocate memory"); + /* Not reached */ + } + } + + obuf = safe_malloc(compressBound(hdr.blksz)); + ibuf = safe_malloc(hdr.blksz); + + signal(SIGHUP, exit); + signal(SIGINT, exit); + signal(SIGTERM, exit); + signal(SIGXCPU, exit); + signal(SIGXFSZ, exit); + atexit(cleanup); + + if (stat(iname, &sb) != 0) { + err(1, "%s", iname); + /* Not reached */ + } + if ((sb.st_size % hdr.blksz) != 0) { + errx(1, "%s: incorrect image: file size is not multiple of %d", + iname, hdr.blksz); + /* Not reached */ + } + hdr.nblocks = sb.st_size / hdr.blksz; + toc = safe_malloc((hdr.nblocks + 1) * sizeof(*toc)); + + fdr = open(iname, O_RDONLY); + if (fdr < 0) { + err(1, "%s", iname); + /* Not reached */ + } + fdw = open(oname, O_WRONLY | O_TRUNC | O_CREAT, + S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH); + if (fdw < 0) { + err(1, "%s", oname); + /* Not reached */ + } + cleanfile = oname; + + /* Prepare header that we will write later when we have index ready. */ + iov[0].iov_base = (char *)&hdr; + iov[0].iov_len = sizeof(hdr); + iov[1].iov_base = (char *)toc; + iov[1].iov_len = (hdr.nblocks + 1) * sizeof(*toc); + offset = iov[0].iov_len + iov[1].iov_len; + + /* Reserve space for header */ + lseek(fdw, offset, SEEK_SET); + + if (verbose != 0) + fprintf(stderr, "Data size: %ld bytes, number of clusters: " + "%ld, index lengh: %ld bytes\n", (long)sb.st_size, + (long)(hdr.nblocks), ((long)hdr.nblocks + 1) * sizeof(*toc)); + + for(i = 0; i == 0 || ibuf != NULL; i++) { + ibuf = readblock(fdr, ibuf, hdr.blksz); + if (ibuf != NULL) { + destlen = compressBound(hdr.blksz); + memset(obuf, 0, destlen); + if (compress2(obuf, &destlen, ibuf, hdr.blksz, Z_BEST_COMPRESSION) != Z_OK) { + errx(1, "can't compress data: compress2() failed"); + /* Not reached */ + } +#if 0 + /* + * We don't really need those two leading bytes. Moreover, they + * confuse oldest decompression routine presented in the + * FreeBSD kernel, so they should be omitted. + */ + destlen -= 2; +#endif + } else { + destlen = DEV_BSIZE - (offset % DEV_BSIZE); + memset(obuf, 0, destlen); + } + if (write(fdw, obuf, destlen) < 0) { + err(1, "%s", oname); + /* Not reached */ + } + toc[i] = htobe64(offset); + offset += destlen; + } + close(fdr); + + if (verbose != 0) + fprintf(stderr, "compressed data to %llu bytes.\n", offset); + + /* Convert to big endian */ + hdr.blksz = htonl(hdr.blksz); + hdr.nblocks = htonl(hdr.nblocks); + /* Write headers into pre-allocated space */ + lseek(fdw, 0, SEEK_SET); + if (writev(fdw, iov, 2) < 0) { + err(1, "%s", oname); + /* Not reached */ + } + cleanfile = NULL; + close(fdw); + + exit(0); +} + +static char * +readblock(int fd, char *ibuf, u_int32_t clstsize) { + int numread; + + bzero(ibuf, clstsize); + numread = read(fd, ibuf, clstsize); + if (numread < 0) { + err(1, "read() failed"); + /* Not reached */ + } + if (numread == 0) { + return NULL; + } + return ibuf; +} + +static void +usage(void) { + + fprintf(stderr, "usage: mkuzip [-v] [-o outfile] [-s cluster_size] infile\n"); + exit(1); +} + +static void * +safe_malloc(size_t size) { + void *retval; + + retval = malloc(size); + if (retval == NULL) { + err(1, "can't allocate memory"); + /* Not reached */ + } + return retval; +} + +static void +cleanup(void) { + + if (cleanfile != NULL) + unlink(cleanfile); +} |
