转载 2011年12月13日 16:53:25



  struct tar_header
   char name[100];
   char mode[8];
   char uid[8];
   char gid[8];
   char size[12];
   char mtime[12];
   char chksum[8];
   char typeflag;
   char linkname[100];
   char magic[6];
   char version[2];
   char uname[32];
   char gname[32];
   char devmajor[8];
   char devminor[8];
   char prefix[155];
   char padding[12];



TAR   Format
Intel   byte   order

Information   from   File   Format   List   2.0   by   Max   Maischein.

If   you   notice   any   mistakes   or   omissions,   please   let   me   know!     It   is   only
with   YOUR   help   that   the   list   can   continue   to   grow.     Please   send
all   changes   to   me   rather   than   distributing   a   modified   version   of   the   list.

This   file   has   been   authored   in   the   style   of   the   INTERxxy.*   file   list
by   Ralf   Brown,   and   uses   almost   the   same   format.

Please   read   the   file   FILEFMTS.1ST   before   asking   me   any   questions.   You   may   find
that   they   have   already   been   addressed.

                  Max   Maischein

Max   Maischein,   2:244/1106.17
Corion   on   #coders@IRC
DISCLAIMER:     THIS   MATERIAL   IS   PROVIDED   "AS   IS ".     I   verify   the   information
contained   in   this   list   to   the   best   of   my   ability,   but   I   cannot   be   held
responsible   for   any   problems   caused   by   use   or   misuse   of   the   information,
especially   for   those   file   formats   foreign   to   the   PC,   like   AMIGA   or   SUN   file
formats.   If   an   information   it   is   marked   "guesswork "   or   undocumented,   you
should   check   it   carefully   to   make   sure   your   program   will   not   break   with
an   unexpected   value   (and   please   let   me   know   whether   or   not   it   works
the   same   way).

Information   marked   with   "??? "   is   known   to   be   incomplete   or   guesswork.

Some   file   formats   were   not   released   by   their   creators,   others   are   regarded
as   proprietary,   which   means   that   if   your   programs   deal   with   them,   you   might
be   looking   for   trouble.   I   don 't   care   about   this.

The   Unix   TAR   program   is   an   archiver   program   which   stores   files   in   a   single
archive   without   compression.
OFFSET                             Count   TYPE       Description
@section   The   Standard   Format
A   @dfn{tar   tape}   or   file   contains   a   series   of   records.     Each   record
contains   @code{RECORDSIZE}   bytes.     Although   this   format   may   be
thought   of   as   being   on   magnetic   tape,   other   media   are   often   used.

Each   file   archived   is   represented   by   a   header   record   which   describes
the   file,   followed   by   zero   or   more   records   which   give   the   contents
of   the   file.     At   the   end   of   the   archive   file   there   may   be   a   record
filled   with   binary   zeros   as   an   end-of-file   marker.     A   reasonable
system   should   write   a   record   of   zeros   at   the   end,   but   must   not
assume   that   such   a   record   exists   when   reading   an   archive.

The   records   may   be   @dfn{blocked}   for   physical   I/O   operations.     Each
block   of   @var{N}   records   (where   @var{N}   is   set   by   the   @samp{-b}
option   to   @code{tar})   is   written   with   a   single   @code{write()}
operation.     On   magnetic   tapes,   the   result   of   such   a   write   is   a
single   tape   record.     When   writing   an   archive,   the   last   block   of
records   should   be   written   at   the   full   size,   with   records   after   the
zero   record   containing   all   zeroes.     When   reading   an   archive,   a
reasonable   system   should   properly   handle   an   archive   whose   last   block
is   shorter   than   the   rest,   or   which   contains   garbage   records   after   a
zero   record.

The   header   record   is   defined   in   C   as   follows:

  *   Standard   Archive   Format   -   Standard   TAR   -   USTAR
#define     RECORDSIZE     512
#define     NAMSIZ             100
#define     TUNMLEN             32
#define     TGNMLEN             32

union   record   @{
        char                 charptr[RECORDSIZE];
        struct   header   @{
                char         name[NAMSIZ];
                char         mode[8];
                char         uid[8];
                char         gid[8];
                char         size[12];
                char         mtime[12];
                char         chksum[8];
                char         linkflag;
                char         linkname[NAMSIZ];
                char         magic[8];
                char         uname[TUNMLEN];
                char         gname[TGNMLEN];
                char         devmajor[8];
                char         devminor[8];
        @}   header;

/*   The   checksum   field   is   filled   with   this   while   the   checksum   is   computed.   */
#define         CHKBLANKS         "                 "                 /*   8   blanks,   no   null   */

/*   The   magic   field   is   filled   with   this   if   uname   and   gname   are   valid.   */
#define         TMAGIC         "ustar     "                 /*   7   chars   and   a   null   */

/*   The   magic   field   is   filled   with   this   if   this   is   a   GNU   format   dump   entry   */
#define         GNUMAGIC     "GNUtar   "                 /*   7   chars   and   a   null   */

/*   The   linkflag   defines   the   type   of   file   */
#define     LF_OLDNORMAL   '/0 '               /*   Normal   disk   file,   Unix   compatible   */
#define     LF_NORMAL         '0 '                 /*   Normal   disk   file   */
#define     LF_LINK             '1 '                 /*   Link   to   previously   dumped   file   */
#define     LF_SYMLINK       '2 '                 /*   Symbolic   link   */
#define     LF_CHR               '3 '                 /*   Character   special   file   */
#define     LF_BLK               '4 '                 /*   Block   special   file   */
#define     LF_DIR               '5 '                 /*   Directory   */
#define     LF_FIFO             '6 '                 /*   FIFO   special   file   */
#define     LF_CONTIG         '7 '                 /*   Contiguous   file   */

/*   Further   link   types   may   be   defined   later.   */

/*   Bits   used   in   the   mode   field   -   values   in   octal   */
#define     TSUID         04000                 /*   Set   UID   on   execution   */
#define     TSGID         02000                 /*   Set   GID   on   execution   */
#define     TSVTX         01000                 /*   Save   text   (sticky   bit)   */

/*   File   permissions   */
#define     TUREAD       00400                 /*   read   by   owner   */
#define     TUWRITE     00200                 /*   write   by   owner   */
#define     TUEXEC       00100                 /*   execute/search   by   owner   */
#define     TGREAD       00040                 /*   read   by   group   */
#define     TGWRITE     00020                 /*   write   by   group   */
#define     TGEXEC       00010                 /*   execute/search   by   group   */
#define     TOREAD       00004                 /*   read   by   other   */
#define     TOWRITE     00002                 /*   write   by   other   */
#define     TOEXEC       00001                 /*   execute/search   by   other   */
@end   example

All   characters   in   header   records   are   represented   by   using   8-bit
characters   in   the   local   variant   of   ASCII.     Each   field   within   the
structure   is   contiguous;   that   is,   there   is   no   padding   used   within
the   structure.     Each   character   on   the   archive   medium   is   stored

Bytes   representing   the   contents   of   files   (after   the   header   record   of
each   file)   are   not   translated   in   any   way   and   are   not   constrained   to
represent   characters   in   any   character   set.     The   @code{tar}   format
does   not   distinguish   text   files   from   binary   files,   and   no
translation   of   file   contents   is   performed.

The   @code{name},   @code{linkname},   @code{magic},   @code{uname},   and
@code{gname}   are   null-terminated   character   strings.     All   other
fileds   are   zero-filled   octal   numbers   in   ASCII.     Each   numeric   field
of   width   @var{w}   contains   @var{w}@minus{}   2   digits,   a   space,   and   a   null,
except   @code{size},   and   @code{mtime},   which   do   not   contain   the
trailing   null.

The   @code{name}   field   is   the   pathname   of   the   file,   with   directory
names   (if   any)   preceding   the   file   name,   separated   by   slashes.

The   @code{mode}   field   provides   nine   bits   specifying   file   permissions
and   three   bits   to   specify   the   Set   UID,   Set   GID,   and   Save   Text
(``stick ' ')   modes.     Values   for   these   bits   are   defined   above.     When
special   permissions   are   required   to   create   a   file   with   a   given   mode,
and   the   user   restoring   files   from   the   archive   does   not   hold   such
permissions,   the   mode   bit(s)   specifying   those   special   permissions
are   ignored.     Modes   which   are   not   supported   by   the   operating   system
restoring   files   from   the   archive   will   be   ignored.     Unsupported   modes
should   be   faked   up   when   creating   or   updating   an   archive;   e.g.   the
group   permission   could   be   copied   from   the   @code{other}   permission.

The   @code{uid}   and   @code{gid}   fields   are   the   numeric   user   and   group
ID   of   the   file   owners,   respectively.     If   the   operating   system   does
not   support   numeric   user   or   group   IDs,   these   fields   should   be

The   @code{size}   field   is   the   size   of   the   file   in   bytes;   linked   files
are   archived   with   this   field   specified   as   zero.
@xref{Extraction   Options};   in   particular   the   @samp{-G}   option.@refill

The   @code{mtime}   field   is   the   modification   time   of   the   file   at   the
time   it   was   archived.     It   is   the   ASCII   representation   of   the   octal
value   of   the   last   time   the   file   was   modified,   represented   as   an
integer   number   of   seconds   since   January   1,   1970,   00:00   Coordinated
Universal   Time.

The   @code{chksum}   field   is   the   ASCII   representation   of   the   octal
value   of   the   simple   sum   of   all   bytes   in   the   header   record.     Each
8-bit   byte   in   the   header   is   added   to   an   unsigned   integer,
initialized   to   zero,   the   precision   of   which   shall   be   no   less   than
seventeen   bits.     When   calculating   the   checksum,   the   @code{chksum}
field   is   treated   as   if   it   were   all   blanks.

The   @code{typeflag}   field   specifies   the   type   of   file   archived.     If   a
particular   implementation   does   not   recognize   or   permit   the   specified
type,   the   file   will   be   extracted   as   if   it   were   a   regular   file.     As
this   action   occurs,   @code{tar}   issues   a   warning   to   the   standard

@table   @code
@item   LF_NORMAL
These   represent   a   regular   file.     In   order   to   be   compatible   with
older   versions   of   @code{tar},   a   @code{typeflag}   value   of
@code{LF_OLDNORMAL}   should   be   silently   recognized   as   a   regular
file.     New   archives   should   be   created   using   @code{LF_NORMAL}.     Also,
for   backward   compatibility,   @code{tar}   treats   a   regular   file   whose
name   ends   with   a   slash   as   a   directory.

@item   LF_LINK
This   represents   a   file   linked   to   another   file,   of   any   type,
previously   archived.     Such   files   are   identified   in   Unix   by   each   file
having   the   same   device   and   inode   number.     The   linked-to
name   is   specified   in   the   @code{linkname}   field   with   a   trailing   null.

@item   LF_SYMLINK
This   represents   a   symbolic   link   to   another   file.     The   linked-to
name   is   specified   in   the   @code{linkname}   field   with   a   trailing   null.

@item   LF_CHR
@itemx   LF_BLK
These   represent   character   special   files   and   block   special   files
respectively.     In   this   case   the   @code{devmajor}   and   @code{devminor}
fields   will   contain   the   major   and   minor   device   numbers
respectively.     Operating   systems   may   map   the   device   specifications
to   their   own   local   specification,   or   may   ignore   the   entry.

@item   LF_DIR
This   specifies   a   directory   or   sub-directory.     The   directory   name   in
the   @code{name}   field   should   end   with   a   slash.     On   systems   where
disk   allocation   is   performed   on   a   directory   basis   the   @code{size}
field   will   contain   the   maximum   number   of   bytes   (which   may   be   rounded
to   the   nearest   disk   block   allocation   unit)   which   the   directory   may
hold.     A   @code{size}   field   of   zero   indicates   no   such   limiting.
Systems   which   do   not   support   limiting   in   this   manner   should   ignore
the   @code{size}   field.

@item   LF_FIFO
This   specifies   a   FIFO   special   file.     Note   that   the   archiving   of   a
FIFO   file   archives   the   existence   of   this   file   and   not   its   contents.

@item   LF_CONTIG
This   specifies   a   contiguous   file,   which   is   the   same   as   a   normal
file   except   that,   in   operating   systems   which   support   it,
all   its   space   is   allocated   contiguously   on   the   disk.     Operating
systems   which   do   not   allow   contiguous   allocation   should   silently   treat
this   type   as   a   normal   file.

@item   'A '   @dots{}
@itemx   'Z '
These   are   reserved   for   custom   implementations.     Some   of   these   are
used   in   the   GNU   modified   format,   as   described   below.
@end   table

Other   values   are   reserved   for   specification   in   future   revisions   of
the   P1003   standard,   and   should   not   be   used   by   any   @code{tar}   program.

The   @code{magic}   field   indicates   that   this   archive   was   output   in   the
P1003   archive   format.     If   this   field   contains   @code{TMAGIC},   the
@code{uname}   and   @code{gname}   fields   will   contain   the   ASCII
representation   of   the   owner   and   group   of   the   file   respectively.     If
found,   the   user   and   group   ID   represented   by   these   names   will   be   used
rather   than   the   values   within   the   @code{uid}   and   @code{gid}   fields.

@section   GNU   Extensions   to   the   Archive   Format
The   GNU   format   uses   additional   file   types   to   describe   new   types   of
files   in   an   archive.     These   are   listed   below.

@table   @code
@item   LF_DUMPDIR
@itemx   'D '
This   represents   a   directory   and   a   list   of   files   created   by   the
@samp{-G}   option.     The   @code{size}   field   gives   the   total   size   of   the
associated   list   of   files.     Each   filename   is   preceded   by   either   a   @code{ 'Y '}
(the   file   should   be   in   this   archive)   or   an   @code{ 'N '}   (The   file   is   a
directory,   or   is   not   stored   in   the   archive).     Each   filename   is
terminated   by   a   null.     There   is   an   additional   null   after   the   last

@itemx   'M '
This   represents   a   file   continued   from   another   volume   of   a
multi-volume   archive   created   with   the   @samp{-M}   option.     The   original
type   of   the   file   is   not   given   here.     The   @code{size}   field   gives   the
maximum   size   of   this   piece   of   the   file   (assuming   the   volume   does   not
end   before   the   file   is   written   out).     The   @code{offset}   field   gives
the   offset   from   the   beginning   of   the   file   where   this   part   of   the
file   begins.     Thus   @code{size}   plus   @code{offset}   should   equal   the
original   size   of   the   file.

@item   LF_VOLHDR
@itemx   'V '
This   file   type   is   used   to   mark   the   volume   header   that   was   given   with
the   @samp{-V}   option   when   the   archive   was   created.     The   @code{name}
field   contains   the   @code{name}   given   after   the   @samp{-V}   option.
The   @code{size}   field   is   zero.     Only   the   first   file   in   each   volume
of   an   archive   should   have   this   type.

@end   table
OFFSET                             Count   TYPE       Description
0000h                                   256   byte       Other   header   info   ?
0100h                                       6   char       ID= 'ustar ',0


  • yang15225094594
  • yang15225094594
  • 2014年03月06日 17:07
  • 1407


导火索,在对打包后的数据进行MD5jiao yan
  • agony000
  • agony000
  • 2014年09月16日 19:19
  • 1033


本文将介绍一个针对Tar包解析时的优化方案,旨在优化内存、提高效率。     一、首先讲一个tar包的文件结构。(懂得可以绕开此段)       tar只是一个归档文件,并不进行压缩。   stru...
  • m694449212
  • m694449212
  • 2016年01月07日 13:33
  • 987

Aix 压缩、打包、解压、解包 tar zip gz

tar是打包 zip和gz是压缩 打包:tar -cf all.tar *.* 解包:tar -xvf *.tar 解压zip文件:jar xvf DB29.5_AIX.zip 解压gz文件...
  • luju5
  • luju5
  • 2012年11月16日 15:26
  • 1701


       -C, --directory DIR              change to directory DIR[root@hongdi lhd]# tar -cvf c.tar -C ...
  • chinalinuxzend
  • chinalinuxzend
  • 2009年02月03日 13:27
  • 9547

Linux 压缩、解压缩文件(invalid tar magic 报错)

今天在Linux-arm 设备解压*.tar 压缩文件不成功,一直报错提示invalid tar magic,但是在Ubuntu 上使用tar 压缩、解压缩相同文件并不报错,经查找,后来确定原因并找到...
  • iamplane
  • iamplane
  • 2017年03月21日 15:41
  • 4676


  • Wentasy
  • Wentasy
  • 2013年08月18日 08:52
  • 9471

tar 命令参数 详解

主操作模式:  -A, --catenate, --concatenate   追加 tar 文件至归档  -c, --create               创建一个新归档  -d,...
  • s464036801
  • s464036801
  • 2012年12月20日 21:17
  • 2310

tar命令 文件压缩与解压

  • done58
  • done58
  • 2015年12月25日 17:03
  • 8780


tar命令的使用方法如下:tar [-cxtzjvfpPN] 文件与目录参数说明:-c :建立一个打包文件;-x :解开一个打包文件;-t :查看 tar包里面的文件;-z :打包后用gzip压缩,生...
  • y970105
  • y970105
  • 2007年03月18日 00:08
  • 20800