I had hoped to keep liblzma as purely a compression
library as possible (e.g. file I/O will go into
a different library), but it seems that applications
linking agaisnt liblzma need some way to determine
the memory usage limit, and knowing the amount of RAM
is one reasonable way to help making such decisions.
Thanks to Jonathan Nieder for the original patch.
Originally the idea was that using LZMA_FULL_FLUSH
with Stream encoder would read the filter chain
from the same array that was used to intialize the
Stream encoder. Since most apps wouldn't use
LZMA_FULL_FLUSH, most apps wouldn't need to keep
the filter chain available after initializing the
Stream encoder. However, due to my mistake, it
actually required keeping the array always available.
Since setting the new filter chain via the array
used at initialization time is not a nice way to do
it for a couple of reasons, this commit ditches it
and introduces lzma_filters_update(). This new function
replaces also the "persistent" flag used by LZMA2
(and to-be-designed Subblock filter), which was also
an ugly thing to do.
Thanks to Alexey Tourbin for reminding me about the problem
that Stream encoder used to require keeping the filter
chain allocated.
This will be needed internally by liblzma once I fix
a design mistake in the encoder API. This function may
be useful to applications too so it's good to export it.
This replaces bswap.h and integer.h.
The tuklib module uses <byteswap.h> on GNU,
<sys/endian.h> on *BSDs and <sys/byteorder.h>
on Solaris, which may contain optimized code
like inline assembly.
Seems that it is a problem in some cases if the same
version of XZ Utils produces different output on different
endiannesses, so this commit fixes that problem. The output
will still vary between different XZ Utils versions, but I
cannot avoid that for now.
This commit bloatens the code on big endian systems by 1 KiB,
which should be OK since liblzma is bloated already. ;-)
and use a fix that works on all systems using
GNU assembler.
Maybe the assembler code is used e.g. on Solaris x86
but let's worry about it if this doesn't work on it.
Added lzma_nothrow for every function. It adds
throw() when the header is used in C++ code.
Some lzma_attrs were added or removed.
Lots of comments were improved.
This is a quick and slightly dirty fix to make the code
conform to the latest file format specification. Without
this patch, it's possible to make corrupt files by
specifying start offset that is not a multiple of the
filter's alignment. Custom start offset is almost never
used, so this was only a minor bug.
The xz command line tool doesn't validate the start offset,
so one will get a bit unclear error message if trying to use
an invalid start offset.
use AC_PROG_SED. We don't do anything fancy with sed,
so this should work OK. libtool 2.2 sets SED but 1.5
doesn't, so $(SED) happened to work when using libtool 2.2.
Don't use libtool convenience libraries to avoid recently
discovered long-standing subtle but somewhat severe bugs
in libtool (at least 1.5.22 and 2.2.6 are affected). It
was found when porting XZ Utils to Windows
<http://lists.gnu.org/archive/html/libtool/2009-06/msg00070.html>
but the problem is significant also e.g. on GNU/Linux.
Unless --disable-shared is passed to configure, static
library built from a set of convenience libraries will
contain PIC objects. That is, while libtool builds non-PIC
objects too, only PIC objects will be used from the
convenience libraries. On 32-bit x86 (tested on mobile XP2400+),
using PIC instead of non-PIC makes the decompressor 10 % slower
with the default CFLAGS.
So while xz was linked against static liblzma by default,
it got the slower PIC objects unless --disable-shared was
used. I tend develop and benchmark with --disable-shared
due to faster build time, so I hadn't noticed the problem
in benchmarks earlier.
This commit also adds support for building Windows resources
into liblzma and executables.
liblzma tries to avoid useless free()/malloc() pairs in
initialization when multiple files are handled using the
same lzma_stream. This didn't work with filter chains
due to comparison of wrong pointers in lzma_next_coder_init(),
making liblzma think that no memory reallocation is needed
even when it actually is.
Easy way to trigger this bug is to decompress two files with
a single xz command. The first file should have e.g. x86+LZMA2
as the filter chain, and the second file just LZMA2.
Now configure.ac will get the version number directly from
src/liblzma/api/lzma/version.h. The intent is to reduce the
number of places where the version number is duplicated. In
future, support for displaying Git commit ID may be added too.
- Use call/ret pair to get instruction pointer for PIC.
- Use PIC only if PIC or __PIC__ is #defined.
- The code should work on MinGW and Darwin in addition
to GNU/Linux and Solaris.
lzma_memlimit_encoder and lzma_memlimit_decoder to
lzma_raw_encoder_memlimit and lzma_raw_decoder_memlimit. :-(
Now it is fixed. Hopefully it doesn't cause too much trouble
to those who already thought API is stable.
Half of developers were already forgetting to use these
functions, which could have caused total breakage in some future
liblzma version or even now if --enable-small was used. Now
liblzma uses pthread_once() to do the initializations unless
it has been built with --disable-threads which make these
initializations thread-unsafe.
When --enable-small isn't used, liblzma currently gets needlessly
linked against libpthread (on systems that have it). While it is
stupid for now, liblzma will need threads in future anyway, so
this stupidity will be temporary only.
When --enable-small is used, different code CRC32 and CRC64 is
now used than without --enable-small. This made the resulting
binary slightly smaller, but the main reason was to clean it up
and to handle the lack of lzma_init_check().
The pkg-config file lzma.pc was renamed to liblzma.pc. I'm not
sure if it works correctly and portably for static linking
(Libs.private includes -pthread or other operating system
specific flags). Hopefully someone complains if it is bad.
lzma_rc_prices[] is now included as a precomputed array even
with --enable-small. It's just 128 bytes now that it uses uint8_t
instead of uint32_t. Smaller array seemed to be at least as fast
as the more bloated uint32_t array on x86; hopefully it's not bad
on other architectures.
that was related to LZMA_MODE_FAST. The original code is slightly
faster although it compresses slightly worse. But since it is fast
mode, it is better to select the faster version.
- Updated to the latest, probably final file format version.
- Command line tool reworked to not use threads anymore.
Threading will probably go into liblzma anyway.
- Memory usage limit is now about 30 % for uncompression
and about 90 % for compression.
- Progress indicator with --verbose
- Simplified --help and full --long-help
- Upgraded to the last LGPLv2.1+ getopt_long from gnulib.
- Some bug fixes
Use LZMA_PROG_ERROR instead of LZMA_HEADER_ERROR if the Filter ID
is in the reserved range. This allows Block Header encoder to
detect unallowed Filter IDs, which is good for Stream encoder.
code from block_private.h to block_decoder.c. Now the Block
encoder doesn't need compressed_size and uncompressed_size
from lzma_block structure to be initialized.
LZMA_Alone files. Decoding of concatenated LZMA_Alone files is
intentionally not supported, so it is better to put this in
auto decoder than LZMA_Alone decoder.
broken. API has changed a lot and it will still change a
little more here and there. The command line tool doesn't
have all the required changes to reflect the API changes, so
it's easy to get "internal error" or trigger assertions.
literal coder as part of the main LZMA encoder or
decoder structure.
Make the LZMA decoder to rely on the current internal API
to free the allocated memory in case an error occurs.
specification. Simplify things by removing most of the
support for known uncompressed size in most places.
There are some miscellaneous changes here and there too.
The API of liblzma has got many changes and still some
more will be done soon. While most of the code has been
updated, some things are not fixed (the command line tool
will choke with invalid filter chain, if nothing else).
Subblock filter is somewhat broken for now. It will be
updated once the encoded format of the Subblock filter
has been decided.
misunderstanding of the code. There's no tiny fix for this
problem, so I also cleaned up the code in general.
This reduces the speed of the encoder 2-5 % in the fastest
compression mode ("lzma -1"). High compression modes should
have no noticeable performance difference.
This commit breaks things (especially LZMA_SYNC_FLUSH) but I
will fix them once the new format and LZMA2 has been roughly
implemented. Plain LZMA won't support LZMA_SYNC_FLUSH at all
and won't be supported in the new .lzma format. This may
change still but this is what it looks like now.
Support for known uncompressed size (that is, LZMA or LZMA2
without EOPM) is likely to go away. This means there will
be API changes.
size. The "fix" breaks LZMA_SYNC_FLUSH at end of stream
with known uncompressed size, but since it currently seems
likely that support for encoding with known uncompressed
size will go away anyway, I'm not fixing this problem now.
once we have a stable release (won't be very soon). The
version number is no longer related to version of LZMA SDK.
Made some small Automake-related changes to toplevel
Makefile.am and configure.ac.
word literal instead of char for better consistency.
There are still some names with _char instead of _literal
in lzma_optimum, these may be changed later.
Renamed length coder variables.
This commit doesn't change the program logic.
lzma_encoder_init.c. While we don't call
fill_distances_prices() and fill_align_prices() in
lzma_lzma_encoder_init(), we still need to initialize
these two variables so that the fill functions get
called in lzma_encoder_getoptimum.c in the beginning
of a stream.
decoder. There's no danger of information leak here, so
it isn't required. Doing memzero() takes a lot of time
with large dictionaries, which could make it easier to
construct DoS attack to consume too much CPU time.
of the so called simple filters. If there is demand, limited
support for LZMA_SYNC_FLUSH may be added in future.
After this commit, using LZMA_SYNC_FLUSH shouldn't cause
undefined behavior in any situation.
encoder and decoder, and put the shared things to
block_private.h. Improved the checks a little so that
they may detect too big Compressed Size at initialization
time if lzma_options_block.total_size or .total_limit is
known.
Allow encoding and decoding Blocks with combinations of
fields that are not allowed by the file format specification.
Doing this requires that the application passes such a
combination in lzma_options_lzma; liblzma doesn't do that,
but it's not impossible that someone could find them useful
in some custom file format.
- Added lzma_memlimit_max() and lzma_memlimit_reached()
API functions.
- Added simple estimation of malloc()'s memory usage
overhead.
- Fixed integer overflow detection in lzma_memlimit_alloc().
- Made some white space cleanups and added more comments.
The description of lzma_memlimit_max() in memlimit.h is bad
and should be improved.
lzma_metadata.header_metadata_size == LZMA_VLI_VALUE_UNKNOWN
is not allowed at all. To indicate missing Header Metadata
Block, header_metadata_size must be set to zero. This is
what Metadata decoder does after this patch too.
Note that other missing fields in lzma_metadata are still
indicated with LZMA_VLI_VALUE_UNKNOWN. This isn't as
illogical as it sounds at first, because missing Size of
Header Metadata Block means that Header Metadata Block is
not present in the Stream. With other Metadata fields,
a missing field means only that the value is unknown.
liblzma as easy as using zlib, because the easy API
don't require developers to know any fancy LZMA options.
Note that Multi-Block Stream encoding is currently broken.
The easy API should be OK, the bug(s) are elsewhere.
This leaves one known alignment bug unfixed: If repeat count
doesn't fit into 28-bit integer, the encoder has to split
this to multiple Subblocks with Subblock Type `Repeating Data'.
The extra Subblocks may have wrong alignment. Correct alignment
is restored after the split Repeating Data has been completely
written out.
Since the encoder doesn't even try to fix the alignment unless
the size of Data is at least 4 bytes, to trigger this bug you
need at least 4 GiB of repeating data with sequence length of
4 or more bytes. Since the worst thing done by this bug is
misaligned data (no data corruption), this bug simply isn't
worth fixing, because a proper fix isn't simple.
The API for handing Subfilters was changed to make it
consistent with LZMA_SYNC_FLUSH.
A few sanity checks were added for Subfilter handling. Some
small bugs were fixed. More comments were added.
function is still shared between encoder and decoder, but the
actual coding is in separate files for encoder and decoder.
There are now separate functions for the actual delta
calculation depending on if Delta is the last filter in the
chain or not. If it is the last, the new code copies the
data from input to output buffer and does the delta
calculation at the same time. The old code first copied the
data, then did the delta in the target buffer, which required
reading through the data twice.
Support for LZMA_SYNC_FLUSH was added to the Delta encoder.
This doesn't change anything in the file format.
too early if we hit End of Input while decoding a Subblock of
type Repeating Data. To keep the loop termination condition
elegant, the order of enumerations in coder->sequence were
changed.
To keep the case-labels in roughly the same order as the
enumerations in coder->sequence, large chunks of code was
moved around. This made the diff big and ugly compared to
the amount of the actual changes made.
in ia64_coder_init(). It triggered assert() in
simple_coder.c, and could have caused a buffer overflow.
This error was probably a copypaste mistake, since most
of the simple filters use unfiltered_max = 4.
table-based version from LZMA SDK 4.57. This should be
fast on most systems.
A simpler and smaller alternative version is also provided.
On some CPUs this can be even a little faster than the
default table-based version (see comments in fastpos.h),
but on most systems the table-based code is faster.
These changes implement support for LZMA_SYNC_FLUSH in LZMA
encoder, and move the temporary buffer needed by range encoder
from lzma_range_encoder structure to lzma_lz_encoder.
coder->lz.stream_end_was_reached. That variable
will be removed, and the check isn't required anyway.
Rearrange the check so that it doesn't make one to
think that there could be an integer overflow.
or no inttypes.h. This is useful when the compiler has
good enough support for C99, but libc headers don't.
Changed liblzma API so that sys/types.h and inttypes.h
have to be #included before #including lzma.h. On systems
that don't have C99 inttypes.h, it's the problem of the
applications to provide the required types and macros
before #including lzma.h.
If lzma.h defined the missing types and macros, it could
conflict with third-party applications whose configure
has detected that the types are missing and defined them
in config.h already. An alternative would have been
introducing lzma_uint32 and similar types, but that would
just be an extra pain on modern systems.
that use LZMA_SYNC_FLUSH with encoder (not implemented
yet). This is a new feature in the raw LZMA format,
which isn't supported by old decoders. This shouldn't
be a problem in practice, since lzma_alone_encoder()
will not allow LZMA_SYNC_FLUSH, and thus not allow
creating files on decodable with old decoders.
Made lzma_decoder.c to require tab width of 4 characters
if one wants to fit the code in 80 columns. This makes
the code easier to read.
so that the preprocessor removes the /* */ style comments,
which are not supported by some non-GNU assemblers (Solaris)
that otherwise work with this code.
It's not strictly needed there, and just complicates the
code. LZ encoder never even had this feature.
The primary reason to have uncompressed size tracking in
filter encoders was validating that the application
doesn't give different amount of input that it had
promised. A side effect was to validate internal workings
of liblzma.
Uncompressed size tracking is still present in the Block
encoder. Maybe it should be added to LZMA_Alone and raw
encoders too. It's simpler to have one coder just to
validate the uncompressed size instead of having it
in every filter.
