Bump the version number to 5.1.4beta.

Note that this slightly changes how lzma_block_header_decode()
has been documented. Earlier it said that the .version is set
to the lowest required value, but now it says that the .version
field is kept unchanged if possible. In practice this doesn't
affect any old code, because before this commit the only
possible .version was 0.

INSTALL

@@ -26,6 +26,8 @@ XZ Utils Installation
        4.2. "No POSIX conforming shell (sh) was found."
        4.3. configure works but build fails at crc32_x86.S
        4.4. Lots of warnings about symbol visibility
+       4.5. "make check" fails
+       4.6. liblzma.so (or similar) not found when running xz
 
 
 0. Preface
@@ -251,6 +253,12 @@ XZ Utils Installation
                 Don't install the scripts xzdiff, xzgrep, xzmore, xzless,
                 and their symlinks.
 
+    --disable-doc
+                Don't install the documentation files to $docdir
+                (often /usr/doc/xz or /usr/local/doc/xz). Man pages
+                will still be installed. The $docdir can be changed
+                with --docdir=DIR.
+
     --disable-assembler
                 liblzma includes some assembler optimizations. Currently
                 there is only assembler code for CRC32 and CRC64 for
@@ -489,3 +497,26 @@ XZ Utils Installation
     resulting binaries, but fewer warnings looks nicer and may allow
     using --enable-werror.
 
+
+4.5. "make check" fails
+
+    A likely reason is that libtool links the test programs against
+    an installed version of liblzma instead of the version that was
+    just built. This is obviously a bug which seems to happen on
+    some platforms. A workaround is to uninstall the old liblzma
+    versions first.
+
+    If the problem isn't the one described above, then it's likely
+    a bug in XZ Utils or in the compiler. See the platform-specific
+    notes in this file for possible known problems. Please report
+    a bug if you cannot solve the problem. See README for contact
+    information.
+
+
+4.6. liblzma.so (or similar) not found when running xz
+
+    If you installed the package with "make install" and get an error
+    about liblzma.so (or a similarly named file) being missing, try
+    running "ldconfig" to update the run-time linker cache (if your
+    operating system has such a command).
+

Makefile.am

@@ -17,6 +17,7 @@ endif
 
 SUBDIRS += src po tests
 
+if COND_DOC
 dist_doc_DATA = \
 	AUTHORS \
 	COPYING \
@@ -42,6 +43,7 @@ examplesolddir = $(docdir)/examples_old
 dist_examplesold_DATA = \
 	doc/examples_old/xz_pipe_comp.c \
 	doc/examples_old/xz_pipe_decomp.c
+endif
 
 EXTRA_DIST = \
 	extra \

NEWS

@@ -2,6 +2,49 @@
 XZ Utils Release Notes
 ======================
 
+5.1.4beta (2014-09-14)
+
+    * All fixes from 5.0.6
+
+    * liblzma: Fixed the use of presets in threaded encoder
+      initialization.
+
+    * xz --block-list and --block-size can now be used together
+      in single-threaded mode. Previously the combination only
+      worked in multi-threaded mode.
+
+    * Added support for LZMA_IGNORE_CHECK to liblzma and made it
+      available in xz as --ignore-check.
+
+    * liblzma speed optimizations:
+
+        - Initialization of a new LZMA1 or LZMA2 encoder has been
+          optimized. (The speed of reinitializing an already-allocated
+          encoder isn't affected.) This helps when compressing many
+          small buffers with lzma_stream_buffer_encode() and other
+          similar situations where an already-allocated encoder state
+          isn't reused. This speed-up is visible in xz too if one
+          compresses many small files one at a time instead running xz
+          once and giving all files as command-line arguments.
+
+        - Buffer comparisons are now much faster when unaligned access
+          is allowed (configured with --enable-unaligned-access). This
+          speeds up encoding significantly. There is arch-specific code
+          for 32-bit and 64-bit x86 (32-bit needs SSE2 for the best
+          results and there's no run-time CPU detection for now).
+          For other archs there is only generic code which probably
+          isn't as optimal as arch-specific solutions could be.
+
+        - A few speed optimizations were made to the SHA-256 code.
+          (Note that the builtin SHA-256 code isn't used on all
+          operating systems.)
+
+    * liblzma can now be built with MSVC 2013 update 2 or later
+      using windows/config.h.
+
+    * Vietnamese translation was added.
+
+
 5.1.3alpha (2013-10-26)
 
     * All fixes from 5.0.5
@@ -121,6 +164,13 @@ XZ Utils Release Notes
       experimental and may change before it gets into a stable release.
 
 
+5.0.6 (2014-09-14)
+
+    * xzgrep now exits with status 0 if at least one file matched.
+
+    * A few minor portability and build system fixes
+
+
 5.0.5 (2013-06-30)
 
     * lzmadec and liblzma's lzma_alone_decoder(): Support decompressing

THANKS

@@ -20,6 +20,7 @@ has been important. :-) In alphabetical order:
   - Daniel Mealha Cabrita
   - Milo Casagrande
   - Marek Černocký
+  - Tomer Chachamu
   - Chris Donawa
   - Andrew Dudman
   - Markus Duft
@@ -32,10 +33,13 @@ has been important. :-) In alphabetical order:
   - Bill Glessner
   - Jason Gorski
   - Juan Manuel Guerrero
+  - Diederik de Haas
   - Joachim Henke
   - Christian Hesse
+  - Vincenzo Innocente
   - Peter Ivanov
   - Jouk Jansen
+  - Jun I Jin
   - Per Øyvind Karlsen
   - Thomas Klausner
   - Richard Koch
@@ -44,6 +48,7 @@ has been important. :-) In alphabetical order:
   - Christian Kujau
   - Stephan Kulow
   - Peter Lawler
+  - James M Leddy
   - Hin-Tak Leung
   - Andraž 'ruskie' Levstik
   - Cary Lewis
@@ -65,6 +70,7 @@ has been important. :-) In alphabetical order:
   - Diego Elio Pettenò
   - Elbert Pol
   - Mikko Pouru
+  - Trần Ngọc Quân
   - Pavel Raiskup
   - Robert Readman
   - Bernhard Reutner-Fischer

TODO

@@ -36,6 +36,15 @@ Known bugs
 Missing features
 ----------------
 
+    Add support for storing metadata in .xz files. A preliminary
+    idea is to create a new Stream type for metadata. When both
+    metadata and data are wanted in the same .xz file, two or more
+    Streams would be concatenated.
+
+    The state stored in lzma_stream should be cloneable, which would
+    be mostly useful when using a preset dictionary in LZMA2, but
+    it may have other uses too. Compare to deflateCopy() in zlib.
+
     Support LZMA_FINISH in raw decoder to indicate end of LZMA1 and
     other streams that don't have an end of payload marker.
 
@@ -68,14 +77,35 @@ Missing features
     This is tricky, because the same error codes are used with
     slightly different meanings, and this cannot be fixed anymore.
 
+    Make it possible to adjust LZMA2 options in the middle of a Block
+    so that the encoding speed vs. compression ratio can be optimized
+    when the compressed data is streamed over network.
+
+    Improved BCJ filters. The current filters are small but they aren't
+    so great when compressing binary packages that contain various file
+    types. Specifically, they make things worse if there are static
+    libraries or Linux kernel modules. The filtering could also be
+    more effective (without getting overly complex), for example,
+    streamable variant BCJ2 from 7-Zip could be implemented.
+
+    Filter that autodetects specific data types in the input stream
+    and applies appropriate filters for the corrects parts of the input.
+    Perhaps combine this with the BCJ filter improvement point above.
+
+    Long-range LZ77 method as a separate filter or as a new LZMA2
+    match finder.
+
 
 Documentation
 -------------
 
-    Some tutorial is needed for liblzma. I have planned to write some
-    extremely well commented example programs, which would work as
-    a tutorial. I suppose the Doxygen tags are quite OK as a quick
-    reference once one is familiar with the liblzma API.
+    More tutorial programs are needed for liblzma.
 
     Document the LZMA1 and LZMA2 algorithms.
 
+
+Miscellaneous
+------------
+
+    Try to get the media type for .xz registered at IANA.
+

configure.ac

@@ -435,6 +435,12 @@ AC_ARG_ENABLE([scripts], [AC_HELP_STRING([--disable-scripts],
 	[], [enable_scripts=yes])
 AM_CONDITIONAL([COND_SCRIPTS], [test x$enable_scripts != xno])
 
+AC_ARG_ENABLE([doc], [AC_HELP_STRING([--disable-doc],
+		[do not install documentation files to docdir
+		(man pages will still be installed)])],
+	[], [enable_doc=yes])
+AM_CONDITIONAL([COND_DOC], [test x$enable_doc != xno])
+
 
 #####################
 # Symbol versioning #
@@ -554,6 +560,7 @@ echo "Initializing gettext:"
 AM_GNU_GETTEXT_VERSION([0.18])
 AM_GNU_GETTEXT([external])
 
+
 ###############################################################################
 # Checks for header files.
 ###############################################################################
@@ -567,6 +574,9 @@ AC_CHECK_HEADERS([fcntl.h limits.h sys/time.h],
 	[],
 	[AC_MSG_ERROR([Required header file(s) are missing.])])
 
+# This allows the use of the intrinsic functions if they are available.
+AC_CHECK_HEADERS([immintrin.h])
+
 
 ###############################################################################
 # Checks for typedefs, structures, and compiler characteristics.
@@ -675,6 +685,15 @@ AM_CONDITIONAL([COND_INTERNAL_SHA256],
 		&& test "x$ac_cv_func_SHA256Init" != xyes \
 		&& test "x$ac_cv_func_CC_SHA256_Init" != xyes])
 
+# Check for SSE2 intrinsics.
+AC_CHECK_DECL([_mm_movemask_epi8],
+	[AC_DEFINE([HAVE__MM_MOVEMASK_EPI8], [1],
+		[Define to 1 if _mm_movemask_epi8 is available.])],
+	[],
+[#ifdef HAVE_IMMINTRIN_H
+#include <immintrin.h>
+#endif])
+
 
 ###############################################################################
 # If using GCC, set some additional AM_CFLAGS:

doc/examples/03_compress_custom.c

@@ -28,8 +28,8 @@ init_encoder(lzma_stream *strm)
 	// Use the default preset (6) for LZMA2.
 	//
 	// The lzma_options_lzma structure and the lzma_lzma_preset() function
-	// are declared in lzma/lzma.h (src/liblzma/api/lzma/lzma.h in the
-	// source package or e.g. /usr/include/lzma/lzma.h depending on
+	// are declared in lzma/lzma12.h (src/liblzma/api/lzma/lzma12.h in the
+	// source package or e.g. /usr/include/lzma/lzma12.h depending on
 	// the install prefix).
 	lzma_options_lzma opt_lzma2;
 	if (lzma_lzma_preset(&opt_lzma2, LZMA_PRESET_DEFAULT)) {
@@ -48,7 +48,7 @@ init_encoder(lzma_stream *strm)
 	// Now we could customize the LZMA2 options if we wanted. For example,
 	// we could set the the dictionary size (opt_lzma2.dict_size) to
 	// something else than the default (8 MiB) of the default preset.
-	// See lzma/lzma.h for details of all LZMA2 options.
+	// See lzma/lzma12.h for details of all LZMA2 options.
 	//
 	// The x86 BCJ filter will try to modify the x86 instruction stream so
 	// that LZMA2 can compress it better. The x86 BCJ filter doesn't need

doc/examples/04_compress_easy_mt.c

@@ -0,0 +1,184 @@
+///////////////////////////////////////////////////////////////////////////////
+//
+/// \file       04_compress_easy_mt.c
+/// \brief      Compress in multi-call mode using LZMA2 in multi-threaded mode
+///
+/// Usage:      ./04_compress_easy_mt < INFILE > OUTFILE
+///
+/// Example:    ./04_compress_easy_mt < foo > foo.xz
+//
+//  Author:     Lasse Collin
+//
+//  This file has been put into the public domain.
+//  You can do whatever you want with this file.
+//
+///////////////////////////////////////////////////////////////////////////////
+
+#include <stdbool.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <errno.h>
+#include <lzma.h>
+
+
+static bool
+init_encoder(lzma_stream *strm)
+{
+	// The threaded encoder takes the options as pointer to
+	// a lzma_mt structure.
+	lzma_mt mt = {
+		// No flags are needed.
+		.flags = 0,
+
+		// Set the number of threads to use.
+		// FIXME: Add how to autodetect a reasonable number.
+		.threads = 4,
+
+		// Let liblzma determine a sane block size.
+		.block_size = 0,
+
+		// Use no timeout for lzma_code() calls by setting timeout
+		// to zero. That is, sometimes lzma_code() might block for
+		// a long time (from several seconds to even minutes).
+		// If this is not OK, for example due to progress indicator
+		// needing updates, specify a timeout in milliseconds here.
+		// See the documentation of lzma_mt in lzma/container.h for
+		// information how to choose a reasonable timeout.
+		.timeout = 0,
+
+		// Use the default preset (6) for LZMA2.
+		// To use a preset, filters must be set to NULL.
+		.preset = LZMA_PRESET_DEFAULT,
+		.filters = NULL,
+
+		// Use CRC64 for integrity checking. See also
+		// 01_compress_easy.c about choosing the integrity check.
+		.check = LZMA_CHECK_CRC64,
+	};
+
+	// Initialize the threaded encoder.
+	lzma_ret ret = lzma_stream_encoder_mt(strm, &mt);
+
+	if (ret == LZMA_OK)
+		return true;
+
+	const char *msg;
+	switch (ret) {
+	case LZMA_MEM_ERROR:
+		msg = "Memory allocation failed";
+		break;
+
+	case LZMA_OPTIONS_ERROR:
+		// We are no longer using a plain preset so this error
+		// message has been edited accordingly compared to
+		// 01_compress_easy.c.
+		msg = "Specified filter chain is not supported";
+		break;
+
+	case LZMA_UNSUPPORTED_CHECK:
+		msg = "Specified integrity check is not supported";
+		break;
+
+	default:
+		msg = "Unknown error, possibly a bug";
+		break;
+	}
+
+	fprintf(stderr, "Error initializing the encoder: %s (error code %u)\n",
+			msg, ret);
+	return false;
+}
+
+
+// This function is identical to the one in 01_compress_easy.c.
+static bool
+compress(lzma_stream *strm, FILE *infile, FILE *outfile)
+{
+	lzma_action action = LZMA_RUN;
+
+	uint8_t inbuf[BUFSIZ];
+	uint8_t outbuf[BUFSIZ];
+
+	strm->next_in = NULL;
+	strm->avail_in = 0;
+	strm->next_out = outbuf;
+	strm->avail_out = sizeof(outbuf);
+
+	while (true) {
+		if (strm->avail_in == 0 && !feof(infile)) {
+			strm->next_in = inbuf;
+			strm->avail_in = fread(inbuf, 1, sizeof(inbuf),
+					infile);
+
+			if (ferror(infile)) {
+				fprintf(stderr, "Read error: %s\n",
+						strerror(errno));
+				return false;
+			}
+
+			if (feof(infile))
+				action = LZMA_FINISH;
+		}
+
+		lzma_ret ret = lzma_code(strm, action);
+
+		if (strm->avail_out == 0 || ret == LZMA_STREAM_END) {
+			size_t write_size = sizeof(outbuf) - strm->avail_out;
+
+			if (fwrite(outbuf, 1, write_size, outfile)
+					!= write_size) {
+				fprintf(stderr, "Write error: %s\n",
+						strerror(errno));
+				return false;
+			}
+
+			strm->next_out = outbuf;
+			strm->avail_out = sizeof(outbuf);
+		}
+
+		if (ret != LZMA_OK) {
+			if (ret == LZMA_STREAM_END)
+				return true;
+
+			const char *msg;
+			switch (ret) {
+			case LZMA_MEM_ERROR:
+				msg = "Memory allocation failed";
+				break;
+
+			case LZMA_DATA_ERROR:
+				msg = "File size limits exceeded";
+				break;
+
+			default:
+				msg = "Unknown error, possibly a bug";
+				break;
+			}
+
+			fprintf(stderr, "Encoder error: %s (error code %u)\n",
+					msg, ret);
+			return false;
+		}
+	}
+}
+
+
+extern int
+main(void)
+{
+	lzma_stream strm = LZMA_STREAM_INIT;
+
+	bool success = init_encoder(&strm);
+	if (success)
+		success = compress(&strm, stdin, stdout);
+
+	lzma_end(&strm);
+
+	if (fclose(stdout)) {
+		fprintf(stderr, "Write error: %s\n", strerror(errno));
+		success = false;
+	}
+
+	return success ? EXIT_SUCCESS : EXIT_FAILURE;
+}

doc/examples/Makefile

@@ -12,7 +12,8 @@ LDFLAGS = -llzma
 PROGS = \
 	01_compress_easy \
 	02_decompress \
-	03_compress_custom
+	03_compress_custom \
+	04_compress_easy_mt
 
 all: $(PROGS)
 

po/LINGUAS

@@ -3,3 +3,4 @@ de
 fr
 it
 pl
+vi

src/common/mythread.h

@@ -25,7 +25,7 @@
 #ifdef MYTHREAD_ENABLED
 
 ////////////////////////////////////////
-// Shared betewen all threading types //
+// Shared between all threading types //
 ////////////////////////////////////////
 
 // Locks a mutex for a duration of a block.

src/common/sysdefs.h

@@ -165,6 +165,16 @@ typedef unsigned char _Bool;
 #	include <memory.h>
 #endif
 
+// As of MSVC 2013, inline and restrict are supported with
+// non-standard keywords.
+#if defined(_WIN32) && defined(_MSC_VER)
+#	ifndef inline
+#		define inline __inline
+#	endif
+#	ifndef restrict
+#		define restrict __restrict
+#	endif
+#endif
 
 ////////////
 // Macros //

src/liblzma/Makefile.am

@@ -12,7 +12,7 @@ CLEANFILES =
 doc_DATA =
 
 lib_LTLIBRARIES = liblzma.la
-liblzma_la_SOURCES = $(top_srcdir)/src/common/tuklib_physmem.c
+liblzma_la_SOURCES =
 liblzma_la_CPPFLAGS = \
 	-I$(top_srcdir)/src/liblzma/api \
 	-I$(top_srcdir)/src/liblzma/common \
@@ -32,6 +32,12 @@ liblzma_la_LDFLAGS += \
 	-Wl,--version-script=$(top_srcdir)/src/liblzma/liblzma.map
 endif
 
+liblzma_la_SOURCES += $(top_srcdir)/src/common/tuklib_physmem.c
+
+if COND_THREADS
+liblzma_la_SOURCES += $(top_srcdir)/src/common/tuklib_cpucores.c
+endif
+
 include $(srcdir)/common/Makefile.inc
 include $(srcdir)/check/Makefile.inc
 

src/liblzma/api/Makefile.am

@@ -17,7 +17,7 @@ nobase_include_HEADERS = \
 	lzma/hardware.h \
 	lzma/index.h \
 	lzma/index_hash.h \
-	lzma/lzma.h \
+	lzma/lzma12.h \
 	lzma/stream_flags.h \
 	lzma/version.h \
 	lzma/vli.h

src/liblzma/api/lzma.h

@@ -286,7 +286,7 @@ extern "C" {
 #include "lzma/filter.h"
 #include "lzma/bcj.h"
 #include "lzma/delta.h"
-#include "lzma/lzma.h"
+#include "lzma/lzma12.h"
 
 /* Container formats */
 #include "lzma/container.h"

src/liblzma/api/lzma/block.h

@@ -31,11 +31,16 @@ typedef struct {
 	/**
 	 * \brief       Block format version
 	 *
-	 * To prevent API and ABI breakages if new features are needed in
-	 * the Block field, a version number is used to indicate which
-	 * fields in this structure are in use. For now, version must always
-	 * be zero. With non-zero version, most Block related functions will
-	 * return LZMA_OPTIONS_ERROR.
+	 * To prevent API and ABI breakages when new features are needed,
+	 * a version number is used to indicate which fields in this
+	 * structure are in use:
+	 *   - liblzma >= 5.0.0: version = 0 is supported.
+	 *   - liblzma >= 5.1.4beta: Support for version = 1 was added,
+	 *     which adds the ignore_check field.
+	 *
+	 * If version is greater than one, most Block related functions
+	 * will return LZMA_OPTIONS_ERROR (lzma_block_header_decode() works
+	 * with any version value).
 	 *
 	 * Read by:
 	 *  - All functions that take pointer to lzma_block as argument,
@@ -233,7 +238,28 @@ typedef struct {
 	lzma_reserved_enum reserved_enum2;
 	lzma_reserved_enum reserved_enum3;
 	lzma_reserved_enum reserved_enum4;
-	lzma_bool reserved_bool1;
+
+	/**
+	 * \brief       A flag to Block decoder to not verify the Check field
+	 *
+	 * This field is supported by liblzma >= 5.1.4beta if .version >= 1.
+	 *
+	 * If this is set to true, the integrity check won't be calculated
+	 * and verified. Unless you know what you are doing, you should
+	 * leave this to false. (A reason to set this to true is when the
+	 * file integrity is verified externally anyway and you want to
+	 * speed up the decompression, which matters mostly when using
+	 * SHA-256 as the integrity check.)
+	 *
+	 * If .version >= 1, read by:
+	 *   - lzma_block_decoder()
+	 *   - lzma_block_buffer_decode()
+	 *
+	 * Written by (.version is ignored):
+	 *   - lzma_block_header_decode() always sets this to false
+	 */
+	lzma_bool ignore_check;
+
 	lzma_bool reserved_bool2;
 	lzma_bool reserved_bool3;
 	lzma_bool reserved_bool4;
@@ -310,14 +336,21 @@ extern LZMA_API(lzma_ret) lzma_block_header_encode(
 /**
  * \brief       Decode Block Header
  *
- * block->version should be set to the highest value supported by the
- * application; currently the only possible version is zero. This function
- * will set version to the lowest value that still supports all the features
- * required by the Block Header.
+ * block->version should (usually) be set to the highest value supported
+ * by the application. If the application sets block->version to a value
+ * higher than supported by the current liblzma version, this function will
+ * downgrade block->version to the highest value supported by it. Thus one
+ * should check the value of block->version after calling this function if
+ * block->version was set to a non-zero value and the application doesn't
+ * otherwise know that the liblzma version being used is new enough to
+ * support the specified block->version.
  *
  * The size of the Block Header must have already been decoded with
  * lzma_block_header_size_decode() macro and stored to block->header_size.
  *
+ * The integrity check type from Stream Header must have been stored
+ * to block->check.
+ *
  * block->filters must have been allocated, but they don't need to be
  * initialized (possible existing filter options are not freed).
  *
@@ -497,7 +530,7 @@ extern LZMA_API(lzma_ret) lzma_block_buffer_encode(
 
 
 /**
- * \brief       Single-call uncompress .xz Block encoder
+ * \brief       Single-call uncompressed .xz Block encoder
  *
  * This is like lzma_block_buffer_encode() except this doesn't try to
  * compress the data and instead encodes the data using LZMA2 uncompressed

src/liblzma/api/lzma/container.h

@@ -473,6 +473,30 @@ extern LZMA_API(lzma_ret) lzma_stream_buffer_encode(
 #define LZMA_TELL_ANY_CHECK             UINT32_C(0x04)
 
 
+/**
+ * This flag makes lzma_code() not calculate and verify the integrity check
+ * of the compressed data in .xz files. This means that invalid integrity
+ * check values won't be detected and LZMA_DATA_ERROR won't be returned in
+ * such cases.
+ *
+ * This flag only affects the checks of the compressed data itself; the CRC32
+ * values in the .xz headers will still be verified normally.
+ *
+ * Don't use this flag unless you know what you are doing. Possible reasons
+ * to use this flag:
+ *
+ *   - Trying to recover data from a corrupt .xz file.
+ *
+ *   - Speeding up decompression, which matters mostly with SHA-256
+ *     or with files that have compressed extremely well. It's recommended
+ *     to not use this flag for this purpose unless the file integrity is
+ *     verified externally in some other way.
+ *
+ * Support for this flag was added in liblzma 5.1.4beta.
+ */
+#define LZMA_IGNORE_CHECK               UINT32_C(0x10)
+
+
 /**
  * This flag enables decoding of concatenated files with file formats that
  * allow concatenating compressed files as is. From the formats currently

src/liblzma/api/lzma/hardware.h

@@ -48,3 +48,17 @@
  *              of RAM on the specific operating system.
  */
 extern LZMA_API(uint64_t) lzma_physmem(void) lzma_nothrow;
+
+
+/**
+ * \brief       Get the number of processor cores or threads
+ *
+ * This function may be useful when determining how many threads to use.
+ * If the hardware supports more than one thread per CPU core, the number
+ * of hardware threads is returned if that information is available.
+ *
+ * \brief       On success, the number of available CPU threads or cores is
+ *              returned. If this information isn't available or an error
+ *              occurs, zero is returned.
+ */
+extern LZMA_API(uint32_t) lzma_cputhreads(void) lzma_nothrow;

src/liblzma/api/lzma/lzma12.h

@@ -1,5 +1,5 @@
 /**
- * \file        lzma/lzma.h
+ * \file        lzma/lzma12.h
  * \brief       LZMA1 and LZMA2 filters
  */
 

src/liblzma/api/lzma/version.h

@@ -22,8 +22,8 @@
  */
 #define LZMA_VERSION_MAJOR 5
 #define LZMA_VERSION_MINOR 1
-#define LZMA_VERSION_PATCH 3
-#define LZMA_VERSION_STABILITY LZMA_VERSION_STABILITY_ALPHA
+#define LZMA_VERSION_PATCH 4
+#define LZMA_VERSION_STABILITY LZMA_VERSION_STABILITY_BETA
 
 #ifndef LZMA_VERSION_COMMIT
 #	define LZMA_VERSION_COMMIT ""

src/liblzma/check/crc32_fast.c

@@ -20,7 +20,7 @@
 #include "crc_macros.h"
 
 
-// If you make any changes, do some bench marking! Seemingly unrelated
+// If you make any changes, do some benchmarking! Seemingly unrelated
 // changes can very easily ruin the performance (and very probably is
 // very compiler dependent).
 extern LZMA_API(uint32_t)

src/liblzma/check/sha256.c

@@ -23,20 +23,20 @@
 
 #include "check.h"
 
-// Avoid bogus warnings in transform().
-#if TUKLIB_GNUC_REQ(4, 2)
-#	pragma GCC diagnostic ignored "-Wuninitialized"
-#endif
+// Rotate a uint32_t. GCC can optimize this to a rotate instruction
+// at least on x86.
+static inline uint32_t
+rotr_32(uint32_t num, unsigned amount)
+{
+        return (num >> amount) | (num << (32 - amount));
+}
 
-// At least on x86, GCC is able to optimize this to a rotate instruction.
-#define rotr_32(num, amount) ((num) >> (amount) | (num) << (32 - (amount)))
-
-#define blk0(i) (W[i] = data[i])
+#define blk0(i) (W[i] = conv32be(data[i]))
 #define blk2(i) (W[i & 15] += s1(W[(i - 2) & 15]) + W[(i - 7) & 15] \
 		+ s0(W[(i - 15) & 15]))
 
 #define Ch(x, y, z) (z ^ (x & (y ^ z)))
-#define Maj(x, y, z) ((x & y) | (z & (x | y)))
+#define Maj(x, y, z) ((x & (y ^ z)) + (y & z))
 
 #define a(i) T[(0 - i) & 7]
 #define b(i) T[(1 - i) & 7]
@@ -47,16 +47,17 @@
 #define g(i) T[(6 - i) & 7]
 #define h(i) T[(7 - i) & 7]
 
-#define R(i) \
-	h(i) += S1(e(i)) + Ch(e(i), f(i), g(i)) + SHA256_K[i + j] \
-		+ (j ? blk2(i) : blk0(i)); \
+#define R(i, j, blk) \
+	h(i) += S1(e(i)) + Ch(e(i), f(i), g(i)) + SHA256_K[i + j] + blk; \
 	d(i) += h(i); \
 	h(i) += S0(a(i)) + Maj(a(i), b(i), c(i))
+#define R0(i) R(i, 0, blk0(i))
+#define R2(i) R(i, j, blk2(i))
 
-#define S0(x) (rotr_32(x, 2) ^ rotr_32(x, 13) ^ rotr_32(x, 22))
-#define S1(x) (rotr_32(x, 6) ^ rotr_32(x, 11) ^ rotr_32(x, 25))
-#define s0(x) (rotr_32(x, 7) ^ rotr_32(x, 18) ^ (x >> 3))
-#define s1(x) (rotr_32(x, 17) ^ rotr_32(x, 19) ^ (x >> 10))
+#define S0(x) rotr_32(x ^ rotr_32(x ^ rotr_32(x, 9), 11), 2)
+#define S1(x) rotr_32(x ^ rotr_32(x ^ rotr_32(x, 14), 5), 6)
+#define s0(x) (rotr_32(x ^ rotr_32(x, 11), 7) ^ (x >> 3))
+#define s1(x) (rotr_32(x ^ rotr_32(x, 2), 17) ^ (x >> 10))
 
 
 static const uint32_t SHA256_K[64] = {
@@ -80,7 +81,7 @@ static const uint32_t SHA256_K[64] = {
 
 
 static void
-transform(uint32_t state[static 8], const uint32_t data[static 16])
+transform(uint32_t state[8], const uint32_t data[16])
 {
 	uint32_t W[16];
 	uint32_t T[8];
@@ -88,12 +89,18 @@ transform(uint32_t state[static 8], const uint32_t data[static 16])
 	// Copy state[] to working vars.
 	memcpy(T, state, sizeof(T));
 
-	// 64 operations, partially loop unrolled
-	for (unsigned int j = 0; j < 64; j += 16) {
-		R( 0); R( 1); R( 2); R( 3);
-		R( 4); R( 5); R( 6); R( 7);
-		R( 8); R( 9); R(10); R(11);
-		R(12); R(13); R(14); R(15);
+	// The first 16 operations unrolled
+	R0( 0); R0( 1); R0( 2); R0( 3);
+	R0( 4); R0( 5); R0( 6); R0( 7);
+	R0( 8); R0( 9); R0(10); R0(11);
+	R0(12); R0(13); R0(14); R0(15);
+
+	// The remaining 48 operations partially unrolled
+	for (unsigned int j = 16; j < 64; j += 16) {
+		R2( 0); R2( 1); R2( 2); R2( 3);
+		R2( 4); R2( 5); R2( 6); R2( 7);
+		R2( 8); R2( 9); R2(10); R2(11);
+		R2(12); R2(13); R2(14); R2(15);
 	}
 
 	// Add the working vars back into state[].
@@ -111,18 +118,7 @@ transform(uint32_t state[static 8], const uint32_t data[static 16])
 static void
 process(lzma_check_state *check)
 {
-#ifdef WORDS_BIGENDIAN
 	transform(check->state.sha256.state, check->buffer.u32);
-
-#else
-	uint32_t data[16];
-
-	for (size_t i = 0; i < 16; ++i)
-		data[i] = bswap32(check->buffer.u32[i]);
-
-	transform(check->state.sha256.state, data);
-#endif
-
 	return;
 }
 

src/liblzma/common/Makefile.inc

@@ -8,6 +8,7 @@
 liblzma_la_SOURCES += \
 	common/common.c \
 	common/common.h \
+	common/memcmplen.h \
 	common/block_util.c \
 	common/easy_preset.c \
 	common/easy_preset.h \
@@ -44,6 +45,7 @@ liblzma_la_SOURCES += \
 
 if COND_THREADS
 liblzma_la_SOURCES += \
+	common/hardware_cputhreads.c \
 	common/outqueue.c \
 	common/outqueue.h \
 	common/stream_encoder_mt.c

src/liblzma/common/block_buffer_encoder.c

@@ -233,7 +233,7 @@ block_buffer_encode(lzma_block *block, const lzma_allocator *allocator,
 
 	// The contents of the structure may depend on the version so
 	// check the version before validating the contents of *block.
-	if (block->version != 0)
+	if (block->version > 1)
 		return LZMA_OPTIONS_ERROR;
 
 	if ((unsigned int)(block->check) > LZMA_CHECK_ID_MAX

src/liblzma/common/block_decoder.c

@@ -45,6 +45,9 @@ struct lzma_coder_s {
 
 	/// Check of the uncompressed data
 	lzma_check_state check;
+
+	/// True if the integrity check won't be calculated and verified.
+	bool ignore_check;
 };
 
 
@@ -97,8 +100,9 @@ block_decode(lzma_coder *coder, const lzma_allocator *allocator,
 					coder->block->uncompressed_size))
 			return LZMA_DATA_ERROR;
 
-		lzma_check_update(&coder->check, coder->block->check,
-				out + out_start, out_used);
+		if (!coder->ignore_check)
+			lzma_check_update(&coder->check, coder->block->check,
+					out + out_start, out_used);
 
 		if (ret != LZMA_STREAM_END)
 			return ret;
@@ -140,7 +144,9 @@ block_decode(lzma_coder *coder, const lzma_allocator *allocator,
 		if (coder->block->check == LZMA_CHECK_NONE)
 			return LZMA_STREAM_END;
 
-		lzma_check_finish(&coder->check, coder->block->check);
+		if (!coder->ignore_check)
+			lzma_check_finish(&coder->check, coder->block->check);
+
 		coder->sequence = SEQ_CHECK;
 
 	// Fall through
@@ -155,7 +161,8 @@ block_decode(lzma_coder *coder, const lzma_allocator *allocator,
 		// Validate the Check only if we support it.
 		// coder->check.buffer may be uninitialized
 		// when the Check ID is not supported.
-		if (lzma_check_is_supported(coder->block->check)
+		if (!coder->ignore_check
+				&& lzma_check_is_supported(coder->block->check)
 				&& memcmp(coder->block->raw_check,
 					coder->check.buffer.u8,
 					check_size) != 0)
@@ -224,6 +231,9 @@ lzma_block_decoder_init(lzma_next_coder *next, const lzma_allocator *allocator,
 	next->coder->check_pos = 0;
 	lzma_check_init(&next->coder->check, block->check);
 
+	next->coder->ignore_check = block->version >= 1
+			? block->ignore_check : false;
+
 	// Initialize the filter chain.
 	return lzma_raw_decoder_init(&next->coder->next, allocator,
 			block->filters);

src/liblzma/common/block_encoder.c

@@ -166,7 +166,7 @@ lzma_block_encoder_init(lzma_next_coder *next, const lzma_allocator *allocator,
 
 	// The contents of the structure may depend on the version so
 	// check the version first.
-	if (block->version != 0)
+	if (block->version > 1)
 		return LZMA_OPTIONS_ERROR;
 
 	// If the Check ID is not supported, we cannot calculate the check and

src/liblzma/common/block_header_decoder.c

@@ -46,8 +46,16 @@ lzma_block_header_decode(lzma_block *block,
 		block->filters[i].options = NULL;
 	}
 
-	// Always zero for now.
-	block->version = 0;
+	// Versions 0 and 1 are supported. If a newer version was specified,
+	// we need to downgrade it.
+	if (block->version > 1)
+		block->version = 1;
+
+	// This isn't a Block Header option, but since the decompressor will
+	// read it if version >= 1, it's better to initialize it here than
+	// to expect the caller to do it since in almost all cases this
+	// should be false.
+	block->ignore_check = false;
 
 	// Validate Block Header Size and Check type. The caller must have
 	// already set these, so it is a programming error if this test fails.

src/liblzma/common/block_header_encoder.c

@@ -17,7 +17,7 @@
 extern LZMA_API(lzma_ret)
 lzma_block_header_size(lzma_block *block)
 {
-	if (block->version != 0)
+	if (block->version > 1)
 		return LZMA_OPTIONS_ERROR;
 
 	// Block Header Size + Block Flags + CRC32.

src/liblzma/common/block_util.c

@@ -51,7 +51,7 @@ lzma_block_unpadded_size(const lzma_block *block)
 	// NOTE: This function is used for validation too, so it is
 	// essential that these checks are always done even if
 	// Compressed Size is unknown.
-	if (block == NULL || block->version != 0
+	if (block == NULL || block->version > 1
 			|| block->header_size < LZMA_BLOCK_HEADER_SIZE_MIN
 			|| block->header_size > LZMA_BLOCK_HEADER_SIZE_MAX
 			|| (block->header_size & 3)

src/liblzma/common/common.c

@@ -53,6 +53,27 @@ lzma_alloc(size_t size, const lzma_allocator *allocator)
 }
 
 
+extern void * lzma_attribute((__malloc__)) lzma_attr_alloc_size(1)
+lzma_alloc_zero(size_t size, const lzma_allocator *allocator)
+{
+	// Some calloc() variants return NULL if called with size == 0.
+	if (size == 0)
+		size = 1;
+
+	void *ptr;
+
+	if (allocator != NULL && allocator->alloc != NULL) {
+		ptr = allocator->alloc(allocator->opaque, 1, size);
+		if (ptr != NULL)
+			memzero(ptr, size);
+	} else {
+		ptr = calloc(1, size);
+	}
+
+	return ptr;
+}
+
+
 extern void
 lzma_free(void *ptr, const lzma_allocator *allocator)
 {

src/liblzma/common/common.h

@@ -75,6 +75,7 @@
 	( LZMA_TELL_NO_CHECK \
 	| LZMA_TELL_UNSUPPORTED_CHECK \
 	| LZMA_TELL_ANY_CHECK \
+	| LZMA_IGNORE_CHECK \
 	| LZMA_CONCATENATED )
 
 
@@ -234,6 +235,12 @@ struct lzma_internal_s {
 extern void *lzma_alloc(size_t size, const lzma_allocator *allocator)
 		lzma_attribute((__malloc__)) lzma_attr_alloc_size(1);
 
+/// Allocates memory and zeroes it (like calloc()). This can be faster
+/// than lzma_alloc() + memzero() while being backward compatible with
+/// custom allocators.
+extern void * lzma_attribute((__malloc__)) lzma_attr_alloc_size(1)
+		lzma_alloc_zero(size_t size, const lzma_allocator *allocator);
+
 /// Frees memory
 extern void lzma_free(void *ptr, const lzma_allocator *allocator);
 

src/liblzma/common/hardware_cputhreads.c

@@ -0,0 +1,22 @@
+///////////////////////////////////////////////////////////////////////////////
+//
+/// \file       hardware_cputhreads.c
+/// \brief      Get the number of CPU threads or cores
+//
+//  Author:     Lasse Collin
+//
+//  This file has been put into the public domain.
+//  You can do whatever you want with this file.
+//
+///////////////////////////////////////////////////////////////////////////////
+
+#include "common.h"
+
+#include "tuklib_cpucores.h"
+
+
+extern LZMA_API(uint32_t)
+lzma_cputhreads(void)
+{
+	return tuklib_cpucores();
+}

src/liblzma/common/memcmplen.h

@@ -0,0 +1,170 @@
+///////////////////////////////////////////////////////////////////////////////
+//
+/// \file       memcmplen.h
+/// \brief      Optimized comparison of two buffers
+//
+//  Author:     Lasse Collin
+//
+//  This file has been put into the public domain.
+//  You can do whatever you want with this file.
+//
+///////////////////////////////////////////////////////////////////////////////
+
+#ifndef LZMA_MEMCMPLEN_H
+#define LZMA_MEMCMPLEN_H
+
+#include "common.h"
+
+#ifdef HAVE_IMMINTRIN_H
+#	include <immintrin.h>
+#endif
+
+/// How many extra bytes lzma_memcmplen() may read. This depends on
+/// the method but since it is just a few bytes the biggest possible
+/// value is used here.
+#define LZMA_MEMCMPLEN_EXTRA 16
+
+
+/// Find out how many equal bytes the two buffers have.
+///
+/// \param      buf1    First buffer
+/// \param      buf2    Second buffer
+/// \param      len     How many bytes have already been compared and will
+///                     be assumed to match
+/// \param      limit   How many bytes to compare at most, including the
+///                     already-compared bytes. This must be significantly
+///                     smaller than UINT32_MAX to avoid integer overflows.
+///                     Up to LZMA_MEMCMPLEN_EXTRA bytes may be read past
+///                     the specified limit from both buf1 and buf2.
+///
+/// \return     Number of equal bytes in the buffers is returned.
+///             This is always at least len and at most limit.
+static inline uint32_t lzma_attribute((__always_inline__))
+lzma_memcmplen(const uint8_t *buf1, const uint8_t *buf2,
+		uint32_t len, uint32_t limit)
+{
+	assert(len <= limit);
+	assert(limit <= UINT32_MAX / 2);
+
+#if defined(TUKLIB_FAST_UNALIGNED_ACCESS) \
+		&& ((TUKLIB_GNUC_REQ(3, 4) && defined(__x86_64__)) \
+			|| (defined(__INTEL_COMPILER) && defined(__x86_64__)) \
+			|| (defined(__INTEL_COMPILER) && defined(_M_X64)) \
+			|| (defined(_MSC_VER) && defined(_M_X64)))
+	// NOTE: This will use 64-bit unaligned access which
+	// TUKLIB_FAST_UNALIGNED_ACCESS wasn't meant to permit, but
+	// it's convenient here at least as long as it's x86-64 only.
+	//
+	// I keep this x86-64 only for now since that's where I know this
+	// to be a good method. This may be fine on other 64-bit CPUs too.
+	// On big endian one should use xor instead of subtraction and switch
+	// to __builtin_clzll().
+	while (len < limit) {
+		const uint64_t x = *(const uint64_t *)(buf1 + len)
+				- *(const uint64_t *)(buf2 + len);
+		if (x != 0) {
+#	if defined(_M_X64) // MSVC or Intel C compiler on Windows
+			unsigned long tmp;
+			_BitScanForward64(&tmp, x);
+			len += (uint32_t)tmp >> 3;
+#	else // GCC, clang, or Intel C compiler
+			len += (uint32_t)__builtin_ctzll(x) >> 3;
+#	endif
+			return my_min(len, limit);
+		}
+
+		len += 8;
+	}
+
+	return limit;
+
+#elif defined(TUKLIB_FAST_UNALIGNED_ACCESS) \
+		&& defined(HAVE__MM_MOVEMASK_EPI8) \
+		&& ((defined(__GNUC__) && defined(__SSE2_MATH__)) \
+			|| (defined(__INTEL_COMPILER) && defined(__SSE2__)) \
+			|| (defined(_MSC_VER) && defined(_M_IX86_FP) \
+				&& _M_IX86_FP >= 2))
+	// NOTE: Like above, this will use 128-bit unaligned access which
+	// TUKLIB_FAST_UNALIGNED_ACCESS wasn't meant to permit.
+	//
+	// SSE2 version for 32-bit and 64-bit x86. On x86-64 the above
+	// version is sometimes significantly faster and sometimes
+	// slightly slower than this SSE2 version, so this SSE2
+	// version isn't used on x86-64.
+	while (len < limit) {
+		const uint32_t x = 0xFFFF ^ _mm_movemask_epi8(_mm_cmpeq_epi8(
+			_mm_loadu_si128((const __m128i *)(buf1 + len)),
+			_mm_loadu_si128((const __m128i *)(buf2 + len))));
+
+		if (x != 0) {
+#	if defined(__INTEL_COMPILER)
+			len += _bit_scan_forward(x);
+#	elif defined(_MSC_VER)
+			unsigned long tmp;
+			_BitScanForward(&tmp, x);
+			len += tmp;
+#	else
+			len += __builtin_ctz(x);
+#	endif
+			return my_min(len, limit);
+		}
+
+		len += 16;
+	}
+
+	return limit;
+
+#elif defined(TUKLIB_FAST_UNALIGNED_ACCESS) && !defined(WORDS_BIGENDIAN)
+	// Generic 32-bit little endian method
+	while (len < limit) {
+		uint32_t x = *(const uint32_t *)(buf1 + len)
+				- *(const uint32_t *)(buf2 + len);
+		if (x != 0) {
+			if ((x & 0xFFFF) == 0) {
+				len += 2;
+				x >>= 16;
+			}
+
+			if ((x & 0xFF) == 0)
+				++len;
+
+			return my_min(len, limit);
+		}
+
+		len += 4;
+	}
+
+	return limit;
+
+#elif defined(TUKLIB_FAST_UNALIGNED_ACCESS) && defined(WORDS_BIGENDIAN)
+	// Generic 32-bit big endian method
+	while (len < limit) {
+		uint32_t x = *(const uint32_t *)(buf1 + len)
+				^ *(const uint32_t *)(buf2 + len);
+		if (x != 0) {
+			if ((x & 0xFFFF0000) == 0) {
+				len += 2;
+				x <<= 16;
+			}
+
+			if ((x & 0xFF000000) == 0)
+				++len;
+
+			return my_min(len, limit);
+		}
+
+		len += 4;
+	}
+
+	return limit;
+
+#else
+	// Simple portable version that doesn't use unaligned access.
+	while (len < limit && buf1[len] == buf2[len])
+		++len;
+
+	return len;
+#endif
+}
+
+#endif

src/liblzma/common/stream_decoder.c

@@ -57,6 +57,10 @@ struct lzma_coder_s {
 	/// If true, LZMA_GET_CHECK is returned after decoding Stream Header.
 	bool tell_any_check;
 
+	/// If true, we will tell the Block decoder to skip calculating
+	/// and verifying the integrity check.
+	bool ignore_check;
+
 	/// If true, we will decode concatenated Streams that possibly have
 	/// Stream Padding between or after them. LZMA_STREAM_END is returned
 	/// once the application isn't giving us any new input, and we aren't
@@ -182,8 +186,8 @@ stream_decode(lzma_coder *coder, const lzma_allocator *allocator,
 
 		coder->pos = 0;
 
-		// Version 0 is currently the only possible version.
-		coder->block_options.version = 0;
+		// Version 1 is needed to support the .ignore_check option.
+		coder->block_options.version = 1;
 
 		// Set up a buffer to hold the filter chain. Block Header
 		// decoder will initialize all members of this array so
@@ -195,6 +199,11 @@ stream_decode(lzma_coder *coder, const lzma_allocator *allocator,
 		return_if_error(lzma_block_header_decode(&coder->block_options,
 				allocator, coder->buffer));
 
+		// If LZMA_IGNORE_CHECK was used, this flag needs to be set.
+		// It has to be set after lzma_block_header_decode() because
+		// it always resets this to false.
+		coder->block_options.ignore_check = coder->ignore_check;
+
 		// Check the memory usage limit.
 		const uint64_t memusage = lzma_raw_decoder_memusage(filters);
 		lzma_ret ret;
@@ -433,6 +442,7 @@ lzma_stream_decoder_init(
 	next->coder->tell_unsupported_check
 			= (flags & LZMA_TELL_UNSUPPORTED_CHECK) != 0;
 	next->coder->tell_any_check = (flags & LZMA_TELL_ANY_CHECK) != 0;
+	next->coder->ignore_check = (flags & LZMA_IGNORE_CHECK) != 0;
 	next->coder->concatenated = (flags & LZMA_CONCATENATED) != 0;
 	next->coder->first_stream = true;
 

src/liblzma/common/stream_encoder_mt.c

@@ -1036,8 +1036,8 @@ stream_encoder_mt_init(lzma_next_coder *next, const lzma_allocator *allocator,
 	for (size_t i = 0; next->coder->filters[i].id != LZMA_VLI_UNKNOWN; ++i)
 		lzma_free(next->coder->filters[i].options, allocator);
 
-	return_if_error(lzma_filters_copy(options->filters,
-			next->coder->filters, allocator));
+	return_if_error(lzma_filters_copy(
+			filters, next->coder->filters, allocator));
 
 	// Index
 	lzma_index_end(next->coder->index, allocator);

src/liblzma/liblzma.map

@@ -95,9 +95,10 @@ global:
 	lzma_vli_size;
 };
 
-XZ_5.1.3alpha {
+XZ_5.1.4beta {
 global:
 	lzma_block_uncomp_encode;
+	lzma_cputhreads;
 	lzma_get_progress;
 	lzma_stream_encoder_mt;
 	lzma_stream_encoder_mt_memusage;

src/liblzma/lz/lz_encoder.c

@@ -20,6 +20,8 @@
 #	include "lz_encoder_hash_table.h"
 #endif
 
+#include "memcmplen.h"
+
 
 struct lzma_coder_s {
 	/// LZ-based encoder e.g. LZMA
@@ -326,25 +328,22 @@ lz_encoder_prepare(lzma_mf *mf, const lzma_allocator *allocator,
 		hs += HASH_4_SIZE;
 */
 
-	// If the above code calculating hs is modified, make sure that
-	// this assertion stays valid (UINT32_MAX / 5 is not strictly the
-	// exact limit). If it doesn't, you need to calculate that
-	// hash_size_sum + sons_count cannot overflow.
-	assert(hs < UINT32_MAX / 5);
-
-	const uint32_t old_count = mf->hash_size_sum + mf->sons_count;
-	mf->hash_size_sum = hs;
+	const uint32_t old_hash_count = mf->hash_count;
+	const uint32_t old_sons_count = mf->sons_count;
+	mf->hash_count = hs;
 	mf->sons_count = mf->cyclic_size;
 	if (is_bt)
 		mf->sons_count *= 2;
 
-	const uint32_t new_count = mf->hash_size_sum + mf->sons_count;
-
 	// Deallocate the old hash array if it exists and has different size
 	// than what is needed now.
-	if (old_count != new_count) {
+	if (old_hash_count != mf->hash_count
+			|| old_sons_count != mf->sons_count) {
 		lzma_free(mf->hash, allocator);
 		mf->hash = NULL;
+
+		lzma_free(mf->son, allocator);
+		mf->son = NULL;
 	}
 
 	// Maximum number of match finder cycles
@@ -366,9 +365,18 @@ lz_encoder_init(lzma_mf *mf, const lzma_allocator *allocator,
 {
 	// Allocate the history buffer.
 	if (mf->buffer == NULL) {
-		mf->buffer = lzma_alloc(mf->size, allocator);
+		// lzma_memcmplen() is used for the dictionary buffer
+		// so we need to allocate a few extra bytes to prevent
+		// it from reading past the end of the buffer.
+		mf->buffer = lzma_alloc(mf->size + LZMA_MEMCMPLEN_EXTRA,
+				allocator);
 		if (mf->buffer == NULL)
 			return true;
+
+		// Keep Valgrind happy with lzma_memcmplen() and initialize
+		// the extra bytes whose value may get read but which will
+		// effectively get ignored.
+		memzero(mf->buffer + mf->size, LZMA_MEMCMPLEN_EXTRA);
 	}
 
 	// Use cyclic_size as initial mf->offset. This allows
@@ -382,44 +390,49 @@ lz_encoder_init(lzma_mf *mf, const lzma_allocator *allocator,
 	mf->write_pos = 0;
 	mf->pending = 0;
 
-	// Allocate match finder's hash array.
-	const size_t alloc_count = mf->hash_size_sum + mf->sons_count;
-
 #if UINT32_MAX >= SIZE_MAX / 4
 	// Check for integer overflow. (Huge dictionaries are not
 	// possible on 32-bit CPU.)
-	if (alloc_count > SIZE_MAX / sizeof(uint32_t))
+	if (mf->hash_count > SIZE_MAX / sizeof(uint32_t)
+			|| mf->sons_count > SIZE_MAX / sizeof(uint32_t))
 		return true;
 #endif
 
+	// Allocate and initialize the hash table. Since EMPTY_HASH_VALUE
+	// is zero, we can use lzma_alloc_zero() or memzero() for mf->hash.
+	//
+	// We don't need to initialize mf->son, but not doing that may
+	// make Valgrind complain in normalization (see normalize() in
+	// lz_encoder_mf.c). Skipping the initialization is *very* good
+	// when big dictionary is used but only small amount of data gets
+	// actually compressed: most of the mf->son won't get actually
+	// allocated by the kernel, so we avoid wasting RAM and improve
+	// initialization speed a lot.
 	if (mf->hash == NULL) {
-		mf->hash = lzma_alloc(alloc_count * sizeof(uint32_t),
+		mf->hash = lzma_alloc_zero(mf->hash_count * sizeof(uint32_t),
 				allocator);
-		if (mf->hash == NULL)
+		mf->son = lzma_alloc(mf->sons_count * sizeof(uint32_t),
+				allocator);
+
+		if (mf->hash == NULL || mf->son == NULL) {
+			lzma_free(mf->hash, allocator);
+			mf->hash = NULL;
+
+			lzma_free(mf->son, allocator);
+			mf->son = NULL;
+
 			return true;
+		}
+	} else {
+/*
+		for (uint32_t i = 0; i < mf->hash_count; ++i)
+			mf->hash[i] = EMPTY_HASH_VALUE;
+*/
+		memzero(mf->hash, mf->hash_count * sizeof(uint32_t));
 	}
 
-	mf->son = mf->hash + mf->hash_size_sum;
 	mf->cyclic_pos = 0;
 
-	// Initialize the hash table. Since EMPTY_HASH_VALUE is zero, we
-	// can use memset().
-/*
-	for (uint32_t i = 0; i < hash_size_sum; ++i)
-		mf->hash[i] = EMPTY_HASH_VALUE;
-*/
-	memzero(mf->hash, (size_t)(mf->hash_size_sum) * sizeof(uint32_t));
-
-	// We don't need to initialize mf->son, but not doing that will
-	// make Valgrind complain in normalization (see normalize() in
-	// lz_encoder_mf.c).
-	//
-	// Skipping this initialization is *very* good when big dictionary is
-	// used but only small amount of data gets actually compressed: most
-	// of the mf->hash won't get actually allocated by the kernel, so
-	// we avoid wasting RAM and improve initialization speed a lot.
-	//memzero(mf->son, (size_t)(mf->sons_count) * sizeof(uint32_t));
-
 	// Handle preset dictionary.
 	if (lz_options->preset_dict != NULL
 			&& lz_options->preset_dict_size > 0) {
@@ -446,7 +459,8 @@ lzma_lz_encoder_memusage(const lzma_lz_options *lz_options)
 	lzma_mf mf = {
 		.buffer = NULL,
 		.hash = NULL,
-		.hash_size_sum = 0,
+		.son = NULL,
+		.hash_count = 0,
 		.sons_count = 0,
 	};
 
@@ -455,9 +469,8 @@ lzma_lz_encoder_memusage(const lzma_lz_options *lz_options)
 		return UINT64_MAX;
 
 	// Calculate the memory usage.
-	return (uint64_t)(mf.hash_size_sum + mf.sons_count)
-				* sizeof(uint32_t)
-			+ (uint64_t)(mf.size) + sizeof(lzma_coder);
+	return ((uint64_t)(mf.hash_count) + mf.sons_count) * sizeof(uint32_t)
+			+ mf.size + sizeof(lzma_coder);
 }
 
 
@@ -466,6 +479,7 @@ lz_encoder_end(lzma_coder *coder, const lzma_allocator *allocator)
 {
 	lzma_next_end(&coder->next, allocator);
 
+	lzma_free(coder->mf.son, allocator);
 	lzma_free(coder->mf.hash, allocator);
 	lzma_free(coder->mf.buffer, allocator);
 
@@ -523,7 +537,8 @@ lzma_lz_encoder_init(lzma_next_coder *next, const lzma_allocator *allocator,
 
 		next->coder->mf.buffer = NULL;
 		next->coder->mf.hash = NULL;
-		next->coder->mf.hash_size_sum = 0;
+		next->coder->mf.son = NULL;
+		next->coder->mf.hash_count = 0;
 		next->coder->mf.sons_count = 0;
 
 		next->coder->next = LZMA_NEXT_CODER_INIT;

src/liblzma/lz/lz_encoder.h

@@ -119,7 +119,7 @@ struct lzma_mf_s {
 	lzma_action action;
 
 	/// Number of elements in hash[]
-	uint32_t hash_size_sum;
+	uint32_t hash_count;
 
 	/// Number of elements in son[]
 	uint32_t sons_count;

src/liblzma/lz/lz_encoder_mf.c

@@ -13,6 +13,7 @@
 
 #include "lz_encoder.h"
 #include "lz_encoder_hash.h"
+#include "memcmplen.h"
 
 
 /// \brief      Find matches starting from the current byte
@@ -65,9 +66,7 @@ lzma_mf_find(lzma_mf *mf, uint32_t *count_ptr, lzma_match *matches)
 			// here because the match distances are zero based.
 			const uint8_t *p2 = p1 - matches[count - 1].dist - 1;
 
-			while (len_best < limit
-					&& p1[len_best] == p2[len_best])
-				++len_best;
+			len_best = lzma_memcmplen(p1, p2, len_best, limit);
 		}
 	}
 
@@ -116,24 +115,27 @@ normalize(lzma_mf *mf)
 			= (MUST_NORMALIZE_POS - mf->cyclic_size);
 				// & (~(UINT32_C(1) << 10) - 1);
 
-	const uint32_t count = mf->hash_size_sum + mf->sons_count;
-	uint32_t *hash = mf->hash;
-
-	for (uint32_t i = 0; i < count; ++i) {
+	for (uint32_t i = 0; i < mf->hash_count; ++i) {
 		// If the distance is greater than the dictionary size,
 		// we can simply mark the hash element as empty.
-		//
-		// NOTE: Only the first mf->hash_size_sum elements are
-		// initialized for sure. There may be uninitialized elements
-		// in mf->son. Since we go through both mf->hash and
-		// mf->son here in normalization, Valgrind may complain
-		// that the "if" below depends on uninitialized value. In
-		// this case it is safe to ignore the warning. See also the
-		// comments in lz_encoder_init() in lz_encoder.c.
-		if (hash[i] <= subvalue)
-			hash[i] = EMPTY_HASH_VALUE;
+		if (mf->hash[i] <= subvalue)
+			mf->hash[i] = EMPTY_HASH_VALUE;
 		else
-			hash[i] -= subvalue;
+			mf->hash[i] -= subvalue;
+	}
+
+	for (uint32_t i = 0; i < mf->sons_count; ++i) {
+		// Do the same for mf->son.
+		//
+		// NOTE: There may be uninitialized elements in mf->son.
+		// Valgrind may complain that the "if" below depends on
+		// an uninitialized value. In this case it is safe to ignore
+		// the warning. See also the comments in lz_encoder_init()
+		// in lz_encoder.c.
+		if (mf->son[i] <= subvalue)
+			mf->son[i] = EMPTY_HASH_VALUE;
+		else
+			mf->son[i] -= subvalue;
 	}
 
 	// Update offset to match the new locations.
@@ -269,10 +271,7 @@ hc_find_func(
 				+ (delta > cyclic_pos ? cyclic_size : 0)];
 
 		if (pb[len_best] == cur[len_best] && pb[0] == cur[0]) {
-			uint32_t len = 0;
-			while (++len != len_limit)
-				if (pb[len] != cur[len])
-					break;
+			uint32_t len = lzma_memcmplen(pb, cur, 1, len_limit);
 
 			if (len_best < len) {
 				len_best = len;
@@ -318,9 +317,8 @@ lzma_mf_hc3_find(lzma_mf *mf, lzma_match *matches)
 	uint32_t len_best = 2;
 
 	if (delta2 < mf->cyclic_size && *(cur - delta2) == *cur) {
-		for ( ; len_best != len_limit; ++len_best)
-			if (*(cur + len_best - delta2) != cur[len_best])
-				break;
+		len_best = lzma_memcmplen(cur - delta2, cur,
+				len_best, len_limit);
 
 		matches[0].len = len_best;
 		matches[0].dist = delta2 - 1;
@@ -397,9 +395,8 @@ lzma_mf_hc4_find(lzma_mf *mf, lzma_match *matches)
 	}
 
 	if (matches_count != 0) {
-		for ( ; len_best != len_limit; ++len_best)
-			if (*(cur + len_best - delta2) != cur[len_best])
-				break;
+		len_best = lzma_memcmplen(cur - delta2, cur,
+				len_best, len_limit);
 
 		matches[matches_count - 1].len = len_best;
 
@@ -484,9 +481,7 @@ bt_find_func(
 		uint32_t len = my_min(len0, len1);
 
 		if (pb[len] == cur[len]) {
-			while (++len != len_limit)
-				if (pb[len] != cur[len])
-					break;
+			len = lzma_memcmplen(pb, cur, len + 1, len_limit);
 
 			if (len_best < len) {
 				len_best = len;
@@ -549,9 +544,7 @@ bt_skip_func(
 		uint32_t len = my_min(len0, len1);
 
 		if (pb[len] == cur[len]) {
-			while (++len != len_limit)
-				if (pb[len] != cur[len])
-					break;
+			len = lzma_memcmplen(pb, cur, len + 1, len_limit);
 
 			if (len == len_limit) {
 				*ptr1 = pair[0];
@@ -639,9 +632,8 @@ lzma_mf_bt3_find(lzma_mf *mf, lzma_match *matches)
 	uint32_t len_best = 2;
 
 	if (delta2 < mf->cyclic_size && *(cur - delta2) == *cur) {
-		for ( ; len_best != len_limit; ++len_best)
-			if (*(cur + len_best - delta2) != cur[len_best])
-				break;
+		len_best = lzma_memcmplen(
+				cur, cur - delta2, len_best, len_limit);
 
 		matches[0].len = len_best;
 		matches[0].dist = delta2 - 1;
@@ -712,9 +704,8 @@ lzma_mf_bt4_find(lzma_mf *mf, lzma_match *matches)
 	}
 
 	if (matches_count != 0) {
-		for ( ; len_best != len_limit; ++len_best)
-			if (*(cur + len_best - delta2) != cur[len_best])
-				break;
+		len_best = lzma_memcmplen(
+				cur, cur - delta2, len_best, len_limit);
 
 		matches[matches_count - 1].len = len_best;
 

src/liblzma/lzma/lzma_encoder_optimum_fast.c

@@ -10,6 +10,7 @@
 ///////////////////////////////////////////////////////////////////////////////
 
 #include "lzma_encoder_private.h"
+#include "memcmplen.h"
 
 
 #define change_pair(small_dist, big_dist) \
@@ -57,9 +58,8 @@ lzma_lzma_optimum_fast(lzma_coder *restrict coder, lzma_mf *restrict mf,
 
 		// The first two bytes matched.
 		// Calculate the length of the match.
-		uint32_t len;
-		for (len = 2; len < buf_avail
-				&& buf[len] == buf_back[len]; ++len) ;
+		const uint32_t len = lzma_memcmplen(
+				buf, buf_back, 2, buf_avail);
 
 		// If we have found a repeated match that is at least
 		// nice_len long, return it immediately.
@@ -155,16 +155,7 @@ lzma_lzma_optimum_fast(lzma_coder *restrict coder, lzma_mf *restrict mf,
 	const uint32_t limit = len_main - 1;
 
 	for (uint32_t i = 0; i < REPS; ++i) {
-		const uint8_t *const buf_back = buf - coder->reps[i] - 1;
-
-		if (not_equal_16(buf, buf_back))
-			continue;
-
-		uint32_t len;
-		for (len = 2; len < limit
-				&& buf[len] == buf_back[len]; ++len) ;
-
-		if (len >= limit) {
+		if (memcmp(buf, buf - coder->reps[i] - 1, limit) == 0) {
 			*back_res = UINT32_MAX;
 			*len_res = 1;
 			return;

src/liblzma/lzma/lzma_encoder_optimum_normal.c

@@ -11,6 +11,7 @@
 
 #include "lzma_encoder_private.h"
 #include "fastpos.h"
+#include "memcmplen.h"
 
 
 ////////////
@@ -305,13 +306,9 @@ helper1(lzma_coder *restrict coder, lzma_mf *restrict mf,
 			continue;
 		}
 
-		uint32_t len_test;
-		for (len_test = 2; len_test < buf_avail
-				&& buf[len_test] == buf_back[len_test];
-				++len_test) ;
+		rep_lens[i] = lzma_memcmplen(buf, buf_back, 2, buf_avail);
 
-		rep_lens[i] = len_test;
-		if (len_test > rep_lens[rep_max_index])
+		if (rep_lens[i] > rep_lens[rep_max_index])
 			rep_max_index = i;
 	}
 
@@ -568,11 +565,7 @@ helper2(lzma_coder *coder, uint32_t *reps, const uint8_t *buf,
 		const uint8_t *const buf_back = buf - reps[0] - 1;
 		const uint32_t limit = my_min(buf_avail_full, nice_len + 1);
 
-		uint32_t len_test = 1;
-		while (len_test < limit && buf[len_test] == buf_back[len_test])
-			++len_test;
-
-		--len_test;
+		const uint32_t len_test = lzma_memcmplen(buf, buf_back, 1, limit) - 1;
 
 		if (len_test >= 2) {
 			lzma_lzma_state state_2 = state;
@@ -612,10 +605,7 @@ helper2(lzma_coder *coder, uint32_t *reps, const uint8_t *buf,
 		if (not_equal_16(buf, buf_back))
 			continue;
 
-		uint32_t len_test;
-		for (len_test = 2; len_test < buf_avail
-				&& buf[len_test] == buf_back[len_test];
-				++len_test) ;
+		uint32_t len_test = lzma_memcmplen(buf, buf_back, 2, buf_avail);
 
 		while (len_end < cur + len_test)
 			coder->opts[++len_end].price = RC_INFINITY_PRICE;

src/liblzma/lzma/lzma_encoder_presets.c

@@ -30,14 +30,16 @@ lzma_lzma_preset(lzma_options_lzma *options, uint32_t preset)
 	options->lp = LZMA_LP_DEFAULT;
 	options->pb = LZMA_PB_DEFAULT;
 
-	options->dict_size = UINT32_C(1) << (uint8_t []){
-			18, 20, 21, 22, 22, 23, 23, 24, 25, 26 }[level];
+	static const uint8_t dict_pow2[]
+			= { 18, 20, 21, 22, 22, 23, 23, 24, 25, 26 };
+	options->dict_size = UINT32_C(1) << dict_pow2[level];
 
 	if (level <= 3) {
 		options->mode = LZMA_MODE_FAST;
 		options->mf = level == 0 ? LZMA_MF_HC3 : LZMA_MF_HC4;
 		options->nice_len = level <= 1 ? 128 : 273;
-		options->depth = (uint8_t []){ 4, 8, 24, 48 }[level];
+		static const uint8_t depths[] = { 4, 8, 24, 48 };
+		options->depth = depths[level];
 	} else {
 		options->mode = LZMA_MODE_NORMAL;
 		options->mf = LZMA_MF_BT4;

src/scripts/xzgrep.in

@@ -147,7 +147,9 @@ if test $# -eq 0; then
 fi
 
 exec 3>&1
-res=0
+
+# res=1 means that no file matched yet
+res=1
 
 for i; do
   case $i in
@@ -195,8 +197,17 @@ for i; do
     fi >&3 5>&-
   )
   r=$?
+
+  # fail occured previously, nothing worse can happen
+  test $res -gt 1 && continue
+
   test "$xz_status" -eq 0 || test "$xz_status" -eq 2 \
       || test "$(kill -l "$xz_status" 2> /dev/null)" = "PIPE" || r=2
-  test $res -lt $r && res=$r
+
+  # still no match
+  test $r -eq 1 && continue
+
+  # 0 == match, >=2 == fail
+  res=$r
 done
 exit $res

src/xz/Makefile.am

@@ -36,7 +36,6 @@ xz_SOURCES = \
 	$(top_srcdir)/src/common/tuklib_open_stdxxx.c \
 	$(top_srcdir)/src/common/tuklib_progname.c \
 	$(top_srcdir)/src/common/tuklib_exit.c \
-	$(top_srcdir)/src/common/tuklib_cpucores.c \
 	$(top_srcdir)/src/common/tuklib_mbstr_width.c \
 	$(top_srcdir)/src/common/tuklib_mbstr_fw.c
 

src/xz/args.c

@@ -22,6 +22,7 @@ bool opt_stdout = false;
 bool opt_force = false;
 bool opt_keep_original = false;
 bool opt_robot = false;
+bool opt_ignore_check = false;
 
 // We don't modify or free() this, but we need to assign it in some
 // non-const pointers.
@@ -141,6 +142,7 @@ parse_real(args_info *args, int argc, char **argv)
 		OPT_INFO_MEMORY,
 		OPT_ROBOT,
 		OPT_FLUSH_TIMEOUT,
+		OPT_IGNORE_CHECK,
 	};
 
 	static const char short_opts[]
@@ -169,6 +171,7 @@ parse_real(args_info *args, int argc, char **argv)
 		// Basic compression settings
 		{ "format",       required_argument, NULL,  'F' },
 		{ "check",        required_argument, NULL,  'C' },
+		{ "ignore-check", no_argument,       NULL,  OPT_IGNORE_CHECK },
 		{ "block-size",   required_argument, NULL,  OPT_BLOCK_SIZE },
 		{ "block-list",  required_argument, NULL,  OPT_BLOCK_LIST },
 		{ "memlimit-compress",   required_argument, NULL, OPT_MEM_COMPRESS },
@@ -438,6 +441,10 @@ parse_real(args_info *args, int argc, char **argv)
 			break;
 		}
 
+		case OPT_IGNORE_CHECK:
+			opt_ignore_check = true;
+			break;
+
 		case OPT_BLOCK_SIZE:
 			opt_block_size = str_to_uint64("block-size", optarg,
 					0, LZMA_VLI_MAX);

src/xz/args.h

@@ -36,6 +36,7 @@ extern bool opt_force;
 extern bool opt_keep_original;
 // extern bool opt_recursive;
 extern bool opt_robot;
+extern bool opt_ignore_check;
 
 extern const char stdin_filename[];
 

src/xz/coder.c

@@ -195,6 +195,29 @@ coder_set_compression_settings(void)
 	// Print the selected filter chain.
 	message_filters_show(V_DEBUG, filters);
 
+	// The --flush-timeout option requires LZMA_SYNC_FLUSH support
+	// from the filter chain. Currently threaded encoder doesn't support
+	// LZMA_SYNC_FLUSH so single-threaded mode must be used.
+	if (opt_mode == MODE_COMPRESS && opt_flush_timeout != 0) {
+		for (size_t i = 0; i < filters_count; ++i) {
+			switch (filters[i].id) {
+			case LZMA_FILTER_LZMA2:
+			case LZMA_FILTER_DELTA:
+				break;
+
+			default:
+				message_fatal(_("The filter chain is "
+					"incompatible with --flush-timeout"));
+			}
+		}
+
+		if (hardware_threads_get() > 1) {
+			message(V_WARNING, _("Switching to single-threaded "
+					"mode due to --flush-timeout"));
+			hardware_threads_set(1);
+		}
+	}
+
 	// Get the memory usage. Note that if --format=raw was used,
 	// we can be decompressing.
 	const uint64_t memory_limit = hardware_memlimit_get(opt_mode);
@@ -237,9 +260,9 @@ coder_set_compression_settings(void)
 	if (memory_usage <= memory_limit)
 		return;
 
-	// If --no-auto-adjust was used or we didn't find LZMA1 or
+	// If --no-adjust was used or we didn't find LZMA1 or
 	// LZMA2 as the last filter, give an error immediately.
-	// --format=raw implies --no-auto-adjust.
+	// --format=raw implies --no-adjust.
 	if (!opt_auto_adjust || opt_format == FORMAT_RAW)
 		memlimit_too_small(memory_usage);
 
@@ -427,7 +450,15 @@ coder_init(file_pair *pair)
 			break;
 		}
 	} else {
-		uint32_t flags = LZMA_TELL_UNSUPPORTED_CHECK;
+		uint32_t flags = 0;
+
+		// It seems silly to warn about unsupported check if the
+		// check won't be verified anyway due to --ignore-check.
+		if (opt_ignore_check)
+			flags |= LZMA_IGNORE_CHECK;
+		else
+			flags |= LZMA_TELL_UNSUPPORTED_CHECK;
+
 		if (!opt_single_stream)
 			flags |= LZMA_CONCATENATED;
 
@@ -514,6 +545,56 @@ coder_init(file_pair *pair)
 }
 
 
+/// Resolve conflicts between opt_block_size and opt_block_list in single
+/// threaded mode. We want to default to opt_block_list, except when it is
+/// larger than opt_block_size. If this is the case for the current Block
+/// at *list_pos, then we break into smaller Blocks. Otherwise advance
+/// to the next Block in opt_block_list, and break apart if needed.
+static void
+split_block(uint64_t *block_remaining,
+	    uint64_t *next_block_remaining,
+	    size_t *list_pos)
+{
+	if (*next_block_remaining > 0) {
+		// The Block at *list_pos has previously been split up.
+		assert(hardware_threads_get() == 1);
+		assert(opt_block_size > 0);
+		assert(opt_block_list != NULL);
+
+		if (*next_block_remaining > opt_block_size) {
+			// We have to split the current Block at *list_pos
+			// into another opt_block_size length Block.
+			*block_remaining = opt_block_size;
+		} else {
+			// This is the last remaining split Block for the
+			// Block at *list_pos.
+			*block_remaining = *next_block_remaining;
+		}
+
+		*next_block_remaining -= *block_remaining;
+
+	} else {
+		// The Block at *list_pos has been finished. Go to the next
+		// entry in the list. If the end of the list has been reached,
+		// reuse the size of the last Block.
+		if (opt_block_list[*list_pos + 1] != 0)
+			++*list_pos;
+
+		*block_remaining = opt_block_list[*list_pos];
+
+		// If in single-threaded mode, split up the Block if needed.
+		// This is not needed in multi-threaded mode because liblzma
+		// will do this due to how threaded encoding works.
+		if (hardware_threads_get() == 1 && opt_block_size > 0
+				&& *block_remaining > opt_block_size) {
+			*next_block_remaining
+					= *block_remaining - opt_block_size;
+			*block_remaining = opt_block_size;
+		}
+	}
+}
+
+
 /// Compress or decompress using liblzma.
 static bool
 coder_normal(file_pair *pair)
@@ -537,6 +618,10 @@ coder_normal(file_pair *pair)
 	// only a single block is created.
 	uint64_t block_remaining = UINT64_MAX;
 
+	// next_block_remining for when we are in single-threaded mode and
+	// the Block in --block-list is larger than the --block-size=SIZE.
+	uint64_t next_block_remaining = 0;
+
 	// Position in opt_block_list. Unused if --block-list wasn't used.
 	size_t list_pos = 0;
 
@@ -551,14 +636,22 @@ coder_normal(file_pair *pair)
 
 		// If --block-list was used, start with the first size.
 		//
-		// FIXME: Currently this overrides --block-size but this isn't
-		// good. For threaded case, we want --block-size to specify
-		// how big Blocks the encoder needs to be prepared to create
-		// at maximum and --block-list will simultaneously cause new
-		// Blocks to be started at specified intervals. To keep things
-		// logical, the same should be done in single-threaded mode.
-		if (opt_block_list != NULL)
-			block_remaining = opt_block_list[list_pos];
+		// For threaded case, --block-size specifies how big Blocks
+		// the encoder needs to be prepared to create at maximum
+		// and --block-list will simultaneously cause new Blocks
+		// to be started at specified intervals. To keep things
+		// logical, the same is done in single-threaded mode. The
+		// output is still not identical because in single-threaded
+		// mode the size info isn't written into Block Headers.
+		if (opt_block_list != NULL) {
+			if (block_remaining < opt_block_list[list_pos]) {
+				assert(hardware_threads_get() == 1);
+				next_block_remaining = opt_block_list[list_pos]
+						- block_remaining;
+			} else {
+				block_remaining = opt_block_list[list_pos];
+			}
+		}
 	}
 
 	strm.next_out = out_buf.u8;
@@ -622,15 +715,13 @@ coder_normal(file_pair *pair)
 			} else {
 				// Start a new Block after LZMA_FULL_BARRIER.
 				if (opt_block_list == NULL) {
+					assert(hardware_threads_get() == 1);
+					assert(opt_block_size > 0);
 					block_remaining = opt_block_size;
 				} else {
-					// FIXME: Make it work together with
-					// --block-size.
-					if (opt_block_list[list_pos + 1] != 0)
-						++list_pos;
-
-					block_remaining
-						= opt_block_list[list_pos];
+					split_block(&block_remaining,
+							&next_block_remaining,
+							&list_pos);
 				}
 			}
 

src/xz/hardware.c

@@ -11,7 +11,6 @@
 ///////////////////////////////////////////////////////////////////////////////
 
 #include "private.h"
-#include "tuklib_cpucores.h"
 
 
 /// Maximum number of worker threads. This can be set with
@@ -33,10 +32,17 @@ hardware_threads_set(uint32_t n)
 {
 	if (n == 0) {
 		// Automatic number of threads was requested.
-		// Use the number of available CPU cores.
-		threads_max = tuklib_cpucores();
+		// If threading support was enabled at build time,
+		// use the number of available CPU cores. Otherwise
+		// use one thread since disabling threading support
+		// omits lzma_cputhreads() from liblzma.
+#ifdef MYTHREAD_ENABLED
+		threads_max = lzma_cputhreads();
 		if (threads_max == 0)
 			threads_max = 1;
+#else
+		threads_max = 1;
+#endif
 	} else {
 		threads_max = n;
 	}

src/xz/message.c

@@ -1122,6 +1122,8 @@ message_help(bool long_help)
 "                      `auto' (default), `xz', `lzma', and `raw'\n"
 "  -C, --check=CHECK   integrity check type: `none' (use with caution),\n"
 "                      `crc32', `crc64' (default), or `sha256'"));
+		puts(_(
+"      --ignore-check  don't verify the integrity check when decompressing"));
 	}
 
 	puts(_(
@@ -1134,19 +1136,23 @@ message_help(bool long_help)
 
 	puts(_(
 "  -T, --threads=NUM   use at most NUM threads; the default is 1; set to 0\n"
-"                      to use the number of processor cores"));
+"                      to use as many threads as there are processor cores"));
 
 	if (long_help) {
-		// FIXME? Mention something about threading?
 		puts(_(
 "      --block-size=SIZE\n"
-"                      when compressing to the .xz format, start a new block\n"
-"                      after every SIZE bytes of input; 0=disabled (default)"));
-		// FIXME
+"                      start a new .xz block after every SIZE bytes of input;\n"
+"                      use this to set the block size for threaded compression"));
 		puts(_(
 "      --block-list=SIZES\n"
-"                      when compressing to the .xz format, start a new block\n"
-"                      after the given intervals of uncompressed data"));
+"                      start a new .xz block after the given comma-separated\n"
+"                      intervals of uncompressed data"));
+		puts(_(
+"      --flush-timeout=TIMEOUT\n"
+"                      when compressing, if more than TIMEOUT milliseconds has\n"
+"                      passed since the previous flush and reading more input\n"
+"                      would block, all pending data is flushed out"
+		));
 		puts(_( // xgettext:no-c-format
 "      --memlimit-compress=LIMIT\n"
 "      --memlimit-decompress=LIMIT\n"

src/xz/signals.c

@@ -82,17 +82,19 @@ signals_init(void)
 		sigaddset(&hooked_signals, message_progress_sigs[i]);
 #endif
 
-	struct sigaction sa;
+	// Using "my_sa" because "sa" may conflict with a sockaddr variable
+	// from system headers on Solaris.
+	struct sigaction my_sa;
 
 	// All the signals that we handle we also blocked while the signal
 	// handler runs.
-	sa.sa_mask = hooked_signals;
+	my_sa.sa_mask = hooked_signals;
 
 	// Don't set SA_RESTART, because we want EINTR so that we can check
 	// for user_abort and cleanup before exiting. We block the signals
 	// for which we have established a handler when we don't want EINTR.
-	sa.sa_flags = 0;
-	sa.sa_handler = &signal_handler;
+	my_sa.sa_flags = 0;
+	my_sa.sa_handler = &signal_handler;
 
 	for (size_t i = 0; i < ARRAY_SIZE(sigs); ++i) {
 		// If the parent process has left some signals ignored,
@@ -103,7 +105,7 @@ signals_init(void)
 			continue;
 
 		// Establish the signal handler.
-		if (sigaction(sigs[i], &sa, NULL))
+		if (sigaction(sigs[i], &my_sa, NULL))
 			message_signal_handler();
 	}
 

src/xz/suffix.c

@@ -390,7 +390,7 @@ suffix_set(const char *suffix)
 	// Empty suffix and suffixes having a directory separator are
 	// rejected. Such suffixes would break things later.
 	if (suffix[0] == '\0' || has_dir_sep(suffix))
-		message_fatal(_("%s: Invalid filename suffix"), optarg);
+		message_fatal(_("%s: Invalid filename suffix"), suffix);
 
 	// Replace the old custom_suffix (if any) with the new suffix.
 	free(custom_suffix);

src/xz/xz.1

@@ -5,7 +5,7 @@
 .\" This file has been put into the public domain.
 .\" You can do whatever you want with this file.
 .\"
-.TH XZ 1 "2013-10-25" "Tukaani" "XZ Utils"
+.TH XZ 1 "2013-11-12" "Tukaani" "XZ Utils"
 .
 .SH NAME
 xz, unxz, xzcat, lzma, unlzma, lzcat \- Compress or decompress .xz and .lzma files
@@ -609,6 +609,25 @@ Integrity of the
 headers is always verified with CRC32.
 It is not possible to change or disable it.
 .TP
+.B \-\-ignore\-check
+Don't verify the integrity check of the compressed data when decompressing.
+The CRC32 values in the
+.B .xz
+headers will still be verified normally.
+.IP ""
+.B "Do not use this option unless you know what you are doing."
+Possible reasons to use this option:
+.RS
+.IP \(bu 3
+Trying to recover data from a corrupt .xz file.
+.IP \(bu 3
+Speeding up decompression.
+This matters mostly with SHA-256 or
+with files that have compressed extremely well.
+It's recommended to not use this option for this purpose
+unless the file integrity is verified externally in some other way.
+.RE
+.TP
 .BR \-0 " ... " \-9
 Select a compression preset level.
 The default is
@@ -809,6 +828,12 @@ format, split the input data into blocks of
 bytes.
 The blocks are compressed independently from each other.
 .\" FIXME: Explain how to these can be used for random access and threading.
+.IP ""
+In multi-threaded mode the sizes of the blocks
+are stored in the block headers.
+This isn't done in single-threaded mode,
+so the encoded output won't be
+identical to that of the multi-threaded mode.
 .TP
 .BI \-\-block\-list= sizes
 When compressing to the
@@ -832,26 +857,26 @@ A special value of
 may be used as the last value to indicate that
 the rest of the file should be encoded as a single block.
 .IP ""
-If this option is used in threaded mode and one specifies
+If one specifies
 .I sizes
 that exceed the encoder's block size
-(either the default value or the value specified
-with \fB\-\-block\-size=\fIsize\fR),
+(either the default value in threaded mode or
+the value specified with \fB\-\-block\-size=\fIsize\fR),
 the encoder will create additional blocks while
 keeping the boundaries specified in
 .IR sizes .
 For example, if one specifies
-.B \-\-threads=2
 .B \-\-block\-size=10MiB
 .B \-\-block\-list=5MiB,10MiB,8MiB,12MiB,24MiB
 and the input file is 80 MiB,
 one will get 11 blocks:
 5, 10, 8, 10, 2, 10, 10, 4, 10, 10, and 1 MiB.
 .IP ""
-.\" FIXME
-.B "In single-threaded mode \-\-block\-size is ignored"
-.B "if \-\-block\-list is also specified."
-.B "This might change before 5.2.0 is released."
+In multi-threaded mode the sizes of the blocks
+are stored in the block headers.
+This isn't done in single-threaded mode,
+so the encoded output won't be
+identical to that of the multi-threaded mode.
 .TP
 .BI \-\-flush\-timeout= timeout
 When compressing, if more than

src/xzdec/Makefile.am

@@ -59,7 +59,14 @@ endif
 if COND_LZMADEC
 bin_PROGRAMS += lzmadec
 
-# FIXME: If xzdec is disabled, this will create a dangling symlink.
+# Create the symlink lzmadec.1->xzdec.1 only if xzdec.1 was installed.
+# This is better than creating a dangling symlink, especially
+# because creating the link may fail due to the directory being missing.
+#
+# FIXME: The correct solution would be to install xzdec.1 as lzmadec.1
+# but I don't know what is the sane way to do it and since this is a bit
+# unusual situation anyway, it's not that important.
+if COND_XZDEC
 install-data-hook:
 	cd $(DESTDIR)$(mandir)/man1 && \
 	target=`echo xzdec | sed '$(transform)'` && \
@@ -72,3 +79,4 @@ uninstall-hook:
 	link=`echo lzmadec | sed '$(transform)'` && \
 	rm -f $$link.1
 endif
+endif

tests/Makefile.am

@@ -13,7 +13,8 @@ EXTRA_DIST = \
 	test_scripts.sh \
 	bcj_test.c \
 	compress_prepared_bcj_sparc \
-	compress_prepared_bcj_x86
+	compress_prepared_bcj_x86 \
+	xzgrep_expected_output
 
 AM_CPPFLAGS = \
 	-I$(top_srcdir)/src/common \
@@ -52,4 +53,5 @@ TESTS += test_scripts.sh
 endif
 
 clean-local:
-	-rm -f compress_generated_*
+	-rm -f compress_generated_* \
+		xzgrep_test_output xzgrep_test_1.xz xzgrep_test_2.xz

tests/test_scripts.sh

@@ -12,16 +12,18 @@
 # If scripts weren't built, this test is skipped.
 XZ=../src/xz/xz
 XZDIFF=../src/scripts/xzdiff
-test -x "$XZ" || XZ=
-test -x "$XZDIFF" || XZDIFF=
-if test -z "$XZ" || test -z "$XZDIFF"; then
+XZGREP=../src/scripts/xzgrep
+
+for i in XZ XZDIFF XZGREP; do
+	eval test -x "\$$i" && continue
 	(exit 77)
 	exit 77
-fi
+done
 
 PATH=`pwd`/../src/xz:$PATH
 export PATH
 
+test -z "$srcdir" && srcdir=.
 preimage=$srcdir/files/good-1-check-crc32.xz
 samepostimage=$srcdir/files/good-1-check-crc64.xz
 otherpostimage=$srcdir/files/good-1-lzma2-1.xz
@@ -50,5 +52,25 @@ if test "$status" != 2 ; then
 	exit 1
 fi
 
+# The exit status must be 0 when a match was found at least from one file,
+# and 1 when no match was found in any file.
+cp "$srcdir/files/good-1-lzma2-1.xz" xzgrep_test_1.xz
+cp "$srcdir/files/good-2-lzma2.xz" xzgrep_test_2.xz
+for pattern in el Hello NOMATCH; do
+	for opts in "" "-l" "-h" "-H"; do
+		echo "=> xzgrep $opts $pattern <="
+		"$XZGREP" $opts $pattern xzgrep_test_1.xz xzgrep_test_2.xz
+		echo retval $?
+	done
+done > xzgrep_test_output 2>&1
+
+if cmp -s "$srcdir/xzgrep_expected_output" xzgrep_test_output ; then
+	:
+else
+	echo "unexpected output from xzgrep"
+	(exit 1)
+	exit 1
+fi
+
 (exit 0)
 exit 0

tests/xzgrep_expected_output

@@ -0,0 +1,39 @@
+=> xzgrep  el <=
+xzgrep_test_1.xz:elit, sed do eiusmod tempor incididunt ut 
+xzgrep_test_1.xz:in voluptate velit esse cillum dolore eu 
+xzgrep_test_2.xz:Hello
+retval 0
+=> xzgrep -l el <=
+xzgrep_test_1.xz
+xzgrep_test_2.xz
+retval 0
+=> xzgrep -h el <=
+elit, sed do eiusmod tempor incididunt ut 
+in voluptate velit esse cillum dolore eu 
+Hello
+retval 0
+=> xzgrep -H el <=
+xzgrep_test_1.xz:elit, sed do eiusmod tempor incididunt ut 
+xzgrep_test_1.xz:in voluptate velit esse cillum dolore eu 
+xzgrep_test_2.xz:Hello
+retval 0
+=> xzgrep  Hello <=
+xzgrep_test_2.xz:Hello
+retval 0
+=> xzgrep -l Hello <=
+xzgrep_test_2.xz
+retval 0
+=> xzgrep -h Hello <=
+Hello
+retval 0
+=> xzgrep -H Hello <=
+xzgrep_test_2.xz:Hello
+retval 0
+=> xzgrep  NOMATCH <=
+retval 1
+=> xzgrep -l NOMATCH <=
+retval 1
+=> xzgrep -h NOMATCH <=
+retval 1
+=> xzgrep -H NOMATCH <=
+retval 1

windows/config.h

@@ -0,0 +1,139 @@
+/* config.h for compiling liblzma (*not* the whole XZ Utils) with MSVC 2013 */
+
+/* How many MiB of RAM to assume if the real amount cannot be determined. */
+#define ASSUME_RAM 128
+
+/* Define to 1 if crc32 integrity check is enabled. */
+#define HAVE_CHECK_CRC32 1
+
+/* Define to 1 if crc64 integrity check is enabled. */
+#define HAVE_CHECK_CRC64 1
+
+/* Define to 1 if sha256 integrity check is enabled. */
+#define HAVE_CHECK_SHA256 1
+
+/* Define to 1 if arm decoder is enabled. */
+#define HAVE_DECODER_ARM 1
+
+/* Define to 1 if armthumb decoder is enabled. */
+#define HAVE_DECODER_ARMTHUMB 1
+
+/* Define to 1 if delta decoder is enabled. */
+#define HAVE_DECODER_DELTA 1
+
+/* Define to 1 if ia64 decoder is enabled. */
+#define HAVE_DECODER_IA64 1
+
+/* Define to 1 if lzma1 decoder is enabled. */
+#define HAVE_DECODER_LZMA1 1
+
+/* Define to 1 if lzma2 decoder is enabled. */
+#define HAVE_DECODER_LZMA2 1
+
+/* Define to 1 if powerpc decoder is enabled. */
+#define HAVE_DECODER_POWERPC 1
+
+/* Define to 1 if sparc decoder is enabled. */
+#define HAVE_DECODER_SPARC 1
+
+/* Define to 1 if x86 decoder is enabled. */
+#define HAVE_DECODER_X86 1
+
+/* Define to 1 if arm encoder is enabled. */
+#define HAVE_ENCODER_ARM 1
+
+/* Define to 1 if armthumb encoder is enabled. */
+#define HAVE_ENCODER_ARMTHUMB 1
+
+/* Define to 1 if delta encoder is enabled. */
+#define HAVE_ENCODER_DELTA 1
+
+/* Define to 1 if ia64 encoder is enabled. */
+#define HAVE_ENCODER_IA64 1
+
+/* Define to 1 if lzma1 encoder is enabled. */
+#define HAVE_ENCODER_LZMA1 1
+
+/* Define to 1 if lzma2 encoder is enabled. */
+#define HAVE_ENCODER_LZMA2 1
+
+/* Define to 1 if powerpc encoder is enabled. */
+#define HAVE_ENCODER_POWERPC 1
+
+/* Define to 1 if sparc encoder is enabled. */
+#define HAVE_ENCODER_SPARC 1
+
+/* Define to 1 if x86 encoder is enabled. */
+#define HAVE_ENCODER_X86 1
+
+/* Define to 1 if you have the <inttypes.h> header file. */
+#define HAVE_INTTYPES_H 1
+
+/* Define to 1 if you have the <limits.h> header file. */
+#define HAVE_LIMITS_H 1
+
+/* Define to 1 to enable bt2 match finder. */
+#define HAVE_MF_BT2 1
+
+/* Define to 1 to enable bt3 match finder. */
+#define HAVE_MF_BT3 1
+
+/* Define to 1 to enable bt4 match finder. */
+#define HAVE_MF_BT4 1
+
+/* Define to 1 to enable hc3 match finder. */
+#define HAVE_MF_HC3 1
+
+/* Define to 1 to enable hc4 match finder. */
+#define HAVE_MF_HC4 1
+
+/* Define to 1 if stdbool.h conforms to C99. */
+#define HAVE_STDBOOL_H 1
+
+/* Define to 1 if you have the <stdint.h> header file. */
+#define HAVE_STDINT_H 1
+
+/* Define to 1 if you have the <stdlib.h> header file. */
+#define HAVE_STDLIB_H 1
+
+/* Define to 1 if you have the <string.h> header file. */
+#define HAVE_STRING_H 1
+
+/* Define to 1 or 0, depending whether the compiler supports simple visibility
+   declarations. */
+#define HAVE_VISIBILITY 0
+
+/* Define to 1 if the system has the type `_Bool'. */
+#define HAVE__BOOL 1
+
+#ifdef _M_IX86
+/* Define to 1 when using Windows 95 (and thus XP) compatible threads. This
+   avoids use of features that were added in Windows Vista.
+   This is used for 32-bit x86 builds for compatibility reasons since it
+   makes no measurable difference in performance compared to Vista threads. */
+#define MYTHREAD_WIN95 1
+#else
+/* Define to 1 when using Windows Vista compatible threads. This uses features
+   that are not available on Windows XP. */
+#define MYTHREAD_VISTA 1
+#endif
+
+/* Define to 1 to disable debugging code. */
+#define NDEBUG 1
+
+/* Define to the full name of this package. */
+#define PACKAGE_NAME "XZ Utils"
+
+/* Define to the home page for this package. */
+#define PACKAGE_URL "http://tukaani.org/xz/"
+
+/* The size of `size_t', as computed by sizeof. */
+#ifdef _WIN64
+#define SIZEOF_SIZE_T 8
+#else
+#define SIZEOF_SIZE_T 4
+#endif
+
+/* Define to 1 if the system supports fast unaligned access to 16-bit and
+   32-bit integers. */
+#define TUKLIB_FAST_UNALIGNED_ACCESS 1