Adding upstream version 0.17.upstream/0.17

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

1 files changed, 232 insertions, 114 deletions

diff --git a/doc/tarlz.texi b/doc/tarlz.texi
index 25acaf8..00116ee 100644
--- a/doc/tarlz.texi
+++ b/doc/tarlz.texi
@@ -6,8 +6,8 @@
 @finalout
 @c %**end of header
 
-@set UPDATED 8 October 2019
-@set VERSION 0.16
+@set UPDATED 30 July 2020
+@set VERSION 0.17
 
 @dircategory Data Compression
 @direntry
@@ -40,7 +40,8 @@ This manual is for Tarlz (version @value{VERSION}, @value{UPDATED}).
 * Portable character set::    POSIX portable filename character set
 * File format::               Detailed format of the compressed archive
 * Amendments to pax format::  The reasons for the differences with pax
-* Multi-threaded tar::        Limitations of parallel tar decoding
+* Program design::            Internal structure of tarlz
+* Multi-threaded decoding::   Limitations of parallel tar decoding
 * Minimum archive sizes::     Sizes required for full multi-threaded speed
 * Examples::                  A small tutorial with examples
 * Problems::                  Reporting bugs
@@ -48,10 +49,10 @@ This manual is for Tarlz (version @value{VERSION}, @value{UPDATED}).
 @end menu
 
 @sp 1
-Copyright @copyright{} 2013-2019 Antonio Diaz Diaz.
+Copyright @copyright{} 2013-2020 Antonio Diaz Diaz.
 
 This manual is free documentation: you have unlimited permission
-to copy, distribute and modify it.
+to copy, distribute, and modify it.
 
 
 @node Introduction
@@ -77,7 +78,8 @@ because it does not keep the members aligned.
 
 Tarlz can create tar archives with five levels of compression granularity;
 per file (---no-solid), per block (---bsolid, default), per directory
-(---dsolid), appendable solid (---asolid), and solid (---solid).
+(---dsolid), appendable solid (---asolid), and solid (---solid). It can also
+create uncompressed tar archives.
 
 @noindent
 Of course, compressing each file (or each directory) individually can't
@@ -105,16 +107,16 @@ and lziprecover can be used to recover some of the damaged members.
 @item
 A multimember tar.lz archive is usually smaller than the
 corresponding solidly compressed tar.gz archive, except when
-individually compressing files smaller than about 32 KiB.
+compressing files smaller than about 32 KiB individually.
 @end itemize
 
-Tarlz protects the extended records with a CRC in a way compatible with
-standard tar tools. @xref{crc32}.
+Tarlz protects the extended records with a Cyclic Redundancy Check (CRC) in
+a way compatible with standard tar tools. @xref{crc32}.
 
 Tarlz does not understand other tar formats like @samp{gnu}, @samp{oldgnu},
-@samp{star} or @samp{v7}. @w{@samp{tarlz -tf archive.tar.lz > /dev/null}}
-can be used to verify that the format of the archive is compatible with
-tarlz.
+@samp{star} or @samp{v7}. The command
+@w{@samp{tarlz -tf archive.tar.lz > /dev/null}} can be used to verify that
+the format of the archive is compatible with tarlz.
 
 
 @node Invoking tarlz
@@ -151,7 +153,11 @@ If several compression levels or @samp{--*solid} options are given, the last
 setting is used. For example @w{@samp{-9 --solid --uncompressed -1}} is
 equivalent to @samp{-1 --solid}
 
-tarlz supports the following options:
+tarlz supports the following
+@uref{http://www.nongnu.org/arg-parser/manual/arg_parser_manual.html#Argument-syntax,,options}:
+@ifnothtml
+@xref{Argument syntax,,,arg_parser}.
+@end ifnothtml
 
 @table @code
 @item --help
@@ -177,7 +183,7 @@ modifying the archive if no @var{files} have been specified.
 @anchor{--data-size}
 @item -B @var{bytes}
 @itemx --data-size=@var{bytes}
-Set target size of input data blocks for the @samp{--bsolid} option.
+Set target size of input data blocks for the option @samp{--bsolid}.
 @xref{--bsolid}. Valid values range from @w{8 KiB} to @w{1 GiB}. Default
 value is two times the dictionary size, except for option @samp{-0} where it
 defaults to @w{1 MiB}. @xref{Minimum archive sizes}.
@@ -210,7 +216,7 @@ standard output the differences found in type, mode (permissions), owner and
 group IDs, modification time, file size, file contents (of regular files),
 target (of symlinks) and device number (of block/character special files).
 
-As tarlz removes leading slashes from member names, the @samp{-C} option may
+As tarlz removes leading slashes from member names, the option @samp{-C} may
 be used in combination with @samp{--diff} when absolute file names were used
 on archive creation: @w{@samp{tarlz -C / -d}}. Alternatively, tarlz may be
 run from the root directory to perform the comparison.
@@ -220,14 +226,18 @@ Make @samp{--diff} ignore differences in owner and group IDs. This option is
 useful when comparing an @samp{--anonymous} archive.
 
 @item --delete
-Delete the specified files and directories from an archive in place. It
-currently can delete only from uncompressed archives and from archives with
-individually compressed files (@samp{--no-solid} archives). Note that files
-of about @samp{--data-size} or larger are compressed individually even if
+Delete files and directories from an archive in place. It currently can
+delete only from uncompressed archives and from archives with files
+compressed individually (@samp{--no-solid} archives). Note that files of
+about @samp{--data-size} or larger are compressed individually even if
 @samp{--bsolid} is used, and can therefore be deleted. Tarlz takes care to
 not delete a tar member unless it is possible to do so. For example it won't
-try to delete a tar member that is not individually compressed. To delete a
-directory without deleting the files under it, use
+try to delete a tar member that is not compressed individually. Even in the
+case of finding a corrupt member after having deleted some member(s), tarlz
+stops and copies the rest of the file as soon as corruption is found,
+leaving it just as corrupt as it was, but not worse.
+
+To delete a directory without deleting the files under it, use
 @w{@samp{tarlz --delete -f foo --exclude='dir/*' dir}}. Deleting in place
 may be dangerous. A corrupt archive, a power cut, or an I/O error may cause
 data loss.
@@ -241,14 +251,22 @@ the file name. For example, @samp{foo/*.o} matches @samp{foo/bar.o}.
 
 @item -f @var{archive}
 @itemx --file=@var{archive}
-Use archive file @var{archive}. @samp{-} used as an @var{archive} argument
-reads from standard input or writes to standard output.
+Use archive file @var{archive}. A hyphen @samp{-} used as an @var{archive}
+argument reads from standard input or writes to standard output.
 
 @item -h
 @itemx --dereference
 Follow symbolic links during archive creation, appending or comparison.
 Archive or compare the files they point to instead of the links themselves.
 
+@item --mtime=@var{date}
+When creating or appending, use @var{date} as the modification time for
+files added to the archive instead of their actual modification times. The
+value of @var{date} may be either @samp{@@} followed by the number of
+seconds since the epoch, or a date in format @w{@samp{YYYY-MM-DD HH:MM:SS}},
+or the name of an existing file starting with @samp{.} or @samp{/}. In the
+latter case, the modification time of that file is used.
+
 @item -n @var{n}
 @itemx --threads=@var{n}
 Set the number of (de)compression threads, overriding the system's default.
@@ -256,15 +274,22 @@ Valid values range from 0 to "as many as your system can support". A value
 of 0 disables threads entirely. If this option is not used, tarlz tries to
 detect the number of processors in the system and use it as default value.
 @w{@samp{tarlz --help}} shows the system's default value. See the note about
-multi-threaded archive creation in the @samp{-C} option above.
+multi-threaded archive creation in the option @samp{-C} above.
 Multi-threaded extraction of files from an archive is not yet implemented.
-@xref{Multi-threaded tar}.
+@xref{Multi-threaded decoding}.
 
 Note that the number of usable threads is limited during compression to
 @w{ceil( uncompressed_size / data_size )} (@pxref{Minimum archive sizes}),
 and during decompression to the number of lzip members in the tar.lz
 archive, which you can find by running @w{@samp{lzip -lv archive.tar.lz}}.
 
+@item -p
+@itemx --preserve-permissions
+On extraction, set file permissions as they appear in the archive. This is
+the default behavior when tarlz is run by the superuser. The default for
+other users is to subtract the umask of the user running tarlz from the
+permissions specified in the archive.
+
 @item -q
 @itemx --quiet
 Quiet operation. Suppress all messages.
@@ -298,7 +323,10 @@ Verbosely list files processed.
 Extract files from an archive. If @var{files} are given, extract only the
 @var{files} given. Else extract all the files in the archive. To extract a
 directory without extracting the files under it, use
-@w{@samp{tarlz -xf foo --exclude='dir/*' dir}}.
+@w{@samp{tarlz -xf foo --exclude='dir/*' dir}}. Tarlz will not make any
+special effort to extract a file over an incompatible type of file. For
+example, extracting a link over a directory will usually fail. (Principle of
+least surprise).
 
 @item -0 .. -9
 Set the compression level for @samp{--create} and @samp{--append}. The
@@ -411,9 +439,9 @@ keyword appearing in the same block of extended records.
 @end table
 
 Exit status: 0 for a normal exit, 1 for environmental problems (file not
-found, invalid flags, I/O errors, etc), 2 to indicate a corrupt or
-invalid input file, 3 for an internal consistency error (eg, bug) which
-caused tarlz to panic.
+found, files differ, invalid flags, I/O errors, etc), 2 to indicate a
+corrupt or invalid input file, 3 for an internal consistency error (eg, bug)
+which caused tarlz to panic.
 
 
 @node Portable character set
@@ -431,12 +459,16 @@ a b c d e f g h i j k l m n o p q r s t u v w x y z
 The last three characters are the period, underscore, and hyphen-minus
 characters, respectively.
 
+File names are identifiers. Therefore, archiving works better when file
+names use only the portable character set without spaces added.
+
 
 @node File format
 @chapter File format
 @cindex file format
 
 In the diagram below, a box like this:
+
 @verbatim
 +---+
 |   | <-- the vertical bars might be missing
@@ -444,6 +476,7 @@ In the diagram below, a box like this:
 @end verbatim
 
 represents one byte; a box like this:
+
 @verbatim
 +==============+
 |              |
@@ -486,7 +519,7 @@ Zero or more blocks that contain the contents of the file.
 Each tar member must be contiguously stored in a lzip member for the
 parallel decoding operations like @samp{--list} to work. If any tar member
 is split over two or more lzip members, the archive must be decoded
-sequentially. @xref{Multi-threaded tar}.
+sequentially. @xref{Multi-threaded decoding}.
 
 At the end of the archive file there are two 512-byte blocks filled with
 binary zeros, interpreted as an end-of-archive indicator. These EOF
@@ -530,28 +563,29 @@ An extended header just before the EOF blocks.
 @section Pax header block
 
 The pax header block is identical to the ustar header block described below
-except that the typeflag has the value @samp{x} (extended). The size field
-is the size of the extended header data in bytes. Most other fields in the
-pax header block are zeroed on archive creation to prevent trouble if the
-archive is read by an ustar tool, and are ignored by tarlz on archive
-extraction. @xref{flawed-compat}.
+except that the typeflag has the value @samp{x} (extended). The field
+@samp{size} is the size of the extended header data in bytes. Most other
+fields in the pax header block are zeroed on archive creation to prevent
+trouble if the archive is read by an ustar tool, and are ignored by tarlz on
+archive extraction. @xref{flawed-compat}.
 
 The pax extended header data consists of one or more records, each of
 them constructed as follows:@*
 @samp{"%d %s=%s\n", <length>, <keyword>, <value>}
 
-The <length>, <blank>, <keyword>, <equals-sign>, and <newline> in the
-record must be limited to the portable character set. The <length> field
-contains the decimal length of the record in bytes, including the
-trailing <newline>. The <value> field is stored as-is, without
-conversion to UTF-8 nor any other transformation.
+The fields <length> and <keyword> in the record must be limited to the
+portable character set (@pxref{Portable character set}). The field <length>
+contains the decimal length of the record in bytes, including the trailing
+newline. The field <value> is stored as-is, without conversion to UTF-8 nor
+any other transformation. The fields are separated by the ASCII characters
+space, equal-sign, and newline.
 
-These are the <keyword> fields currently supported by tarlz:
+These are the <keyword> values currently supported by tarlz:
 
 @table @code
 @item linkpath
 The pathname of a link being created to another file, of any type,
-previously archived. This record overrides the linkname field in the
+previously archived. This record overrides the field @samp{linkname} in the
 following ustar header block. The following ustar header block
 determines the type of link created. If typeflag of the following header
 block is 1, it will be a hard link. If typeflag is 2, it will be a
@@ -559,8 +593,8 @@ symbolic link and the linkpath value will be used as the contents of the
 symbolic link.
 
 @item path
-The pathname of the following file. This record overrides the name and
-prefix fields in the following ustar header block.
+The pathname of the following file. This record overrides the fields
+@samp{name} and @samp{prefix} in the following ustar header block.
 
 @item size
 The size of the file in bytes, expressed as a decimal number using
@@ -610,31 +644,30 @@ shown in the following table. All lengths and offsets are in decimal.
 All characters in the header block are coded using the ISO/IEC 646:1991
 (ASCII) standard, except in fields storing names for files, users, and
 groups. For maximum portability between implementations, names should only
-contain characters from the portable character set. But if an implementation
-supports the use of characters outside of @samp{/} and the portable
-character set in names for files, users, and groups, tarlz will use the byte
-values in these names unmodified.
+contain characters from the portable character set (@pxref{Portable
+character set}), but if an implementation supports the use of characters
+outside of @samp{/} and the portable character set in names for files,
+users, and groups, tarlz will use the byte values in these names unmodified.
 
-The fields name, linkname, and prefix are null-terminated character
-strings except when all characters in the array contain non-null
-characters including the last character.
+The fields @samp{name}, @samp{linkname}, and @samp{prefix} are
+null-terminated character strings except when all characters in the array
+contain non-null characters including the last character.
 
-The name and the prefix fields produce the pathname of the file. A new
-pathname is formed, if prefix is not an empty string (its first
+The fields @samp{prefix} and @samp{name} produce the pathname of the file. A
+new pathname is formed, if prefix is not an empty string (its first
 character is not null), by concatenating prefix (up to the first null
-character), a <slash> character, and name; otherwise, name is used
-alone. In either case, name is terminated at the first null character.
-If prefix begins with a null character, it is ignored. In this manner,
-pathnames of at most 256 characters can be supported. If a pathname does
-not fit in the space provided, an extended record is used to store the
-pathname.
-
-The linkname field does not use the prefix to produce a pathname. If the
-linkname does not fit in the 100 characters provided, an extended record
+character), a slash character, and name; otherwise, name is used alone. In
+either case, name is terminated at the first null character. If prefix
+begins with a null character, it is ignored. In this manner, pathnames of at
+most 256 characters can be supported. If a pathname does not fit in the
+space provided, an extended record is used to store the pathname.
+
+The field @samp{linkname} does not use the prefix to produce a pathname. If
+the linkname does not fit in the 100 characters provided, an extended record
 is used to store the linkname.
 
-The mode field provides 12 access permission bits. The following table
-shows the symbolic name of each bit and its octal value:
+The field @samp{mode} provides 12 access permission bits. The following
+table shows the symbolic name of each bit and its octal value:
 
 @multitable {Bit Name} {Value} {Bit Name} {Value} {Bit Name} {Value}
 @headitem Bit Name @tab Value @tab Bit Name @tab Value @tab Bit Name @tab Value
@@ -644,29 +677,28 @@ shows the symbolic name of each bit and its octal value:
 @item S_IROTH @tab 00004 @tab S_IWOTH @tab 00002 @tab S_IXOTH @tab 00001
 @end multitable
 
-The uid and gid fields are the user and group ID of the owner and group
-of the file, respectively.
+The fields @samp{uid} and @samp{gid} are the user and group IDs of the owner
+and group of the file, respectively.
 
-The size field contains the octal representation of the size of the file
-in bytes. If the typeflag field specifies a file of type '0' (regular
-file) or '7' (high performance regular file), the number of logical
+The field @samp{size} contains the octal representation of the size of the
+file in bytes. If the field @samp{typeflag} specifies a file of type '0'
+(regular file) or '7' (high performance regular file), the number of logical
 records following the header is @w{(size / 512)} rounded to the next
-integer. For all other values of typeflag, tarlz either sets the size
-field to 0 or ignores it, and does not store or expect any logical
-records following the header. If the file size is larger than
-8_589_934_591 bytes @w{(octal 77777777777)}, an extended record is used
-to store the file size.
-
-The mtime field contains the octal representation of the modification
-time of the file at the time it was archived, obtained from the stat()
-function.
-
-The chksum field contains the octal representation of the value of the
-simple sum of all bytes in the header logical record. Each byte in the
-header is treated as an unsigned value. When calculating the checksum,
-the chksum field is treated as if it were all <space> characters.
-
-The typeflag field contains a single character specifying the type of
+integer. For all other values of typeflag, tarlz either sets the size field
+to 0 or ignores it, and does not store or expect any logical records
+following the header. If the file size is larger than 8_589_934_591 bytes
+@w{(octal 77777777777)}, an extended record is used to store the file size.
+
+The field @samp{mtime} contains the octal representation of the modification
+time of the file at the time it was archived, obtained from the function
+@samp{stat}.
+
+The field @samp{chksum} contains the octal representation of the value of
+the simple sum of all bytes in the header logical record. Each byte in the
+header is treated as an unsigned value. When calculating the checksum, the
+chksum field is treated as if it were all space characters.
+
+The field @samp{typeflag} contains a single character specifying the type of
 file archived:
 
 @table @code
@@ -680,8 +712,8 @@ Hard link to another file, of any type, previously archived.
 Symbolic link.
 
 @item '3', '4'
-Character special file and block special file respectively. In this case
-the devmajor and devminor fields contain information defining the
+Character special file and block special file respectively. In this case the
+fields @samp{devmajor} and @samp{devminor} contain information defining the
 device in unspecified format.
 
 @item '5'
@@ -697,14 +729,15 @@ regular file (type 0).
 
 @end table
 
-The magic field contains the ASCII null-terminated string "ustar". The
-version field contains the characters "00" (0x30,0x30). The fields uname,
-and gname are null-terminated character strings except when all characters
-in the array contain non-null characters including the last character. Each
-numeric field contains a leading space- or zero-filled, optionally
-null-terminated octal number using digits from the ISO/IEC 646:1991 (ASCII)
-standard. Tarlz is able to decode numeric fields 1 byte longer than standard
-ustar by not requiring a terminating null character.
+The field @samp{magic} contains the ASCII null-terminated string "ustar".
+The field @samp{version} contains the characters "00" (0x30,0x30). The
+fields @samp{uname} and @samp{gname} are null-terminated character strings
+except when all characters in the array contain non-null characters
+including the last character. Each numeric field contains a leading space-
+or zero-filled, optionally null-terminated octal number using digits from
+the ISO/IEC 646:1991 (ASCII) standard. Tarlz is able to decode numeric
+fields 1 byte longer than standard ustar by not requiring a terminating null
+character.
 
 
 @node Amendments to pax format
@@ -714,10 +747,10 @@ ustar by not requiring a terminating null character.
 Tarlz creates safe archives that allow the reliable detection of invalid or
 corrupt metadata during decoding even when the integrity checking of lzip
 can't be used because the lzip members are only decompressed partially, as
-it happens in parallel @samp{--list} and @samp{--extract}. In order to
-achieve this goal, tarlz makes some changes to the variant of the pax format
-that it uses. This chapter describes these changes and the concrete reasons
-to implement them.
+it happens in parallel @samp{--diff}, @samp{--list}, and @samp{--extract}.
+In order to achieve this goal, tarlz makes some changes to the variant of
+the pax format that it uses. This chapter describes these changes and the
+concrete reasons to implement them.
 
 @sp 1
 @anchor{crc32}
@@ -735,7 +768,7 @@ Metadata like file name and file size must be always protected in an archive
 format because of the adverse effects of undetected corruption in them,
 potentially much worse that undetected corruption in the data. Even more so
 in the case of pax because the amount of metadata it stores is potentially
-large, making undetected corruption more probable.
+large, making undetected corruption and archiver misbehavior more probable.
 
 Headers and metadata must be protected separately from data because the
 integrity checking of lzip may not be able to detect the corruption before
@@ -753,12 +786,12 @@ In order to allow the extraction of pax archives by a tar utility conforming
 to the POSIX-2:1993 standard, POSIX.1-2008 recommends selecting extended
 header field values that allow such tar to create a regular file containing
 the extended header records as data. This approach is broken because if the
-extended header is needed because of a long file name, the name and prefix
-fields will be unable to contain the full pathname of the file. Therefore
-the files corresponding to both the extended header and the overridden ustar
-header will be extracted using truncated file names, perhaps overwriting
-existing files or directories. It may be a security risk to extract a file
-with a truncated file name.
+extended header is needed because of a long file name, the fields
+@samp{prefix} and @samp{name} will be unable to contain the full pathname of
+the file. Therefore the files corresponding to both the extended header and
+the overridden ustar header will be extracted using truncated file names,
+perhaps overwriting existing files or directories. It may be a security risk
+to extract a file with a truncated file name.
 
 To avoid this problem, tarlz writes extended headers with all fields zeroed
 except size, chksum, typeflag, magic and version. This prevents old tar
@@ -778,10 +811,10 @@ between the extended header and the ustar header.
 
 The tarlz format is mainly ustar. Extended pax headers are used only when
 needed because the length of a file name or link name, or the size of a file
-exceed the limits of the ustar format. Adding extended headers to each
-member just to record subsecond timestamps seems wasteful for a backup
-format. Moreover, minimizing the overhead may help recovering the archive
-with lziprecover in case of corruption.
+exceed the limits of the ustar format. Adding @w{1 KiB} of extended headers
+to each member just to record subsecond timestamps seems wasteful for a
+backup format. Moreover, minimizing the overhead may help recovering the
+archive with lziprecover in case of corruption.
 
 Global pax headers are tolerated, but not supported; they are parsed and
 ignored. Some operations may not behave as expected if the archive contains
@@ -797,7 +830,88 @@ accidental double UTF-8 conversions. If the need arises this behavior will
 be adjusted with a command line option in the future.
 
 
-@node Multi-threaded tar
+@node Program design
+@chapter Internal structure of tarlz
+@cindex program design
+
+The parts of tarlz related to sequential processing of the archive are more
+or less similar to any other tar and won't be described here. The interesting
+parts described here are those related to Multi-threaded processing.
+
+The structure of the part of tarlz performing Multi-threaded archive
+creation is somewhat similar to that of plzip with the added complication of
+the solidity levels. A grouper thread and several worker threads are
+created, acting the main thread as muxer (multiplexer) thread. A "packet
+courier" takes care of data transfers among threads and limits the maximum
+number of data blocks (packets) being processed simultaneously.
+
+The grouper traverses the directory tree, groups together the metadata of
+the files to be archived in each lzip member, and distributes them to the
+workers. The workers compress the metadata received from the grouper along
+with the file data read from the file system. The muxer collects processed
+packets from the workers, and writes them to the archive.
+
+@verbatim
+,--------,
+|    data|---> to each worker below
+|        |                    ,------------,
+| file   |                ,-->| worker   0 |--,
+| system |                |   `------------'  |
+|        |    ,---------, |   ,------------,  |   ,-------,   ,---------,
+|metadata|--->| grouper |-+-->| worker   1 |--+-->| muxer |-->| archive |
+`--------'    `---------' |   `------------'  |   `-------'   `---------'
+                          |        ...        |
+                          |   ,------------,  |
+                          `-->| worker N-1 |--'
+                              `------------'
+@end verbatim
+
+Decoding an archive is somewhat similar to how plzip decompresses a regular
+file to standard output, with the differences that it is not the data but
+only messages what is written to stdout/stderr, and that each worker may
+access files in the file system either to read them (diff) or write them
+(extract). As in plzip, each worker reads members directly from the archive.
+
+@verbatim
+,--------,
+| file   |<---> data to/from each worker below
+| system |
+`--------'
+                ,------------,
+            ,-->| worker   0 |--,
+            |   `------------'  |
+,---------, |   ,------------,  |   ,-------,   ,--------,
+| archive |-+-->| worker   1 |--+-->| muxer |-->| stdout |
+`---------' |   `------------'  |   `-------'   | stderr |
+            |        ...        |               `--------'
+            |   ,------------,  |
+            `-->| worker N-1 |--'
+                `------------'
+@end verbatim
+
+As misaligned tar.lz archives can't be decoded in parallel, and the
+misalignment can't be detected until after decoding has started, a
+"mastership request" mechanism has been designed that allows the decoding to
+continue instead of signalling an error.
+
+During parallel decoding, if a worker finds a misalignment, it requests
+mastership to decode the rest of the archive. When mastership is requested,
+an error_member_id is set, and all subsequently received packets with
+member_id > error_member_id are rejected. All workers requesting mastership
+are blocked at the request_mastership call until mastership is granted.
+Mastership is granted to the delivering worker when its queue is empty to
+make sure that all preceding packets have been processed. When mastership is
+granted, all packets are deleted and all subsequently received packets not
+coming from the master are rejected.
+
+If a worker can't continue decoding for any cause (for example lack of
+memory or finding a split tar member at the beginning of a lzip member), it
+requests mastership to print an error and terminate the program. Only if
+some other worker requests mastership in a previous lzip member can this
+error be avoided.
+
+
+@node Multi-threaded decoding
 @chapter Limitations of parallel tar decoding
 @cindex parallel tar decoding
 
@@ -827,8 +941,8 @@ decoding it safely in parallel.
 Tarlz is able to automatically decode aligned and unaligned multimember
 tar.lz archives, keeping backwards compatibility. If tarlz finds a member
 misalignment during multi-threaded decoding, it switches to single-threaded
-mode and continues decoding the archive. Currently only the @samp{--list}
-option is able to do multi-threaded decoding.
+mode and continues decoding the archive. Currently only the options
+@samp{--diff} and @samp{--list} are able to do multi-threaded decoding.
 
 If the files in the archive are large, multi-threaded @samp{--list} on a
 regular (seekable) tar.lz archive can be hundreds of times faster than
@@ -843,6 +957,10 @@ time plzip -cd silesia.tar.lz | tar -tf -           (3.256s)
 time tarlz -tf silesia.tar.lz                       (0.020s)
 @end example
 
+On the other hand, multi-threaded @samp{--list} won't detect corruption in
+the tar member data because it only decodes the part of each lzip member
+corresponding to the tar member header.
+
 
 @node Minimum archive sizes
 @chapter Minimum archive sizes required for multi-threaded block compression
@@ -860,7 +978,7 @@ least as large as the number of worker threads times the block size
 compress, and compression will be proportionally slower. The maximum speed
 increase achievable on a given archive is limited by the ratio
 @w{(uncompressed_size / data_size)}. For example, a tarball the size of gcc
-or linux will scale up to 10 or 12 processors at level -9.
+or linux will scale up to 10 or 14 processors at level -9.
 
 The following table shows the minimum uncompressed archive size needed for
 full use of N processors at a given compression level, using the default