Commit Graph

121 Commits

Author SHA1 Message Date
Lasse Collin 06824396b2 Build: Don't put GNU/Linux-specific symbol versions into static liblzma.
It not only makes no sense to put symbol versions into a static library
but it can also cause breakage.

By default Libtool #defines PIC if building a shared library and
doesn't define it for static libraries. This is documented in the
Libtool manual. It can be overriden using --with-pic or --without-pic.
configure.ac detects if --with-pic or --without-pic is used and then
gives an error if neither --disable-shared nor --disable-static was
used at the same time. Thus, in normal situations it works to build
both shared and static library at the same time on GNU/Linux,
only --with-pic or --without-pic requires that only one type of
library is built.

Thanks to John Paul Adrian Glaubitz from Debian for reporting
the problem that occurred on ia64:
https://www.mail-archive.com/xz-devel@tukaani.org/msg00610.html
2022-11-24 14:52:44 +02:00
Lasse Collin ae1f8a723d CMake: Don't use symbol versioning with static library. 2022-11-24 00:02:31 +02:00
Jia Tan a11a2b8b5e CMake: Adds test_memlimit to CMake tests 2022-11-19 17:35:38 +02:00
Lasse Collin f644473a21 liblzma: Add fast CRC64 for 32/64-bit x86 using SSSE3 + SSE4.1 + CLMUL.
It also works on E2K as it supports these intrinsics.

On x86-64 runtime detection is used so the code keeps working on
older processors too. A CLMUL-only build can be done by using
-msse4.1 -mpclmul in CFLAGS and this will reduce the library
size since the generic implementation and its 8 KiB lookup table
will be omitted.

On 32-bit x86 this isn't used by default for now because by default
on 32-bit x86 the separate assembly file crc64_x86.S is used.
If --disable-assembler is used then this new CLMUL code is used
the same way as on 64-bit x86. However, a CLMUL-only build
(-msse4.1 -mpclmul) won't omit the 8 KiB lookup table on
32-bit x86 due to a currently-missing check for disabled
assembler usage.

The configure.ac check should be such that the code won't be
built if something in the toolchain doesn't support it but
--disable-clmul-crc option can be used to unconditionally
disable this feature.

CLMUL speeds up decompression of files that have compressed very
well (assuming CRC64 is used as a check type). It is know that
the CLMUL code is significantly slower than the generic code for
tiny inputs (especially 1-8 bytes but up to 16 bytes). If that
is a real-world problem then there is already a commented-out
variant that uses the generic version for small inputs.

Thanks to Ilya Kurdyukov for the original patch which was
derived from a white paper from Intel [1] (published in 2009)
and public domain code from [2] (released in 2016).

[1] https://www.intel.com/content/dam/www/public/us/en/documents/white-papers/fast-crc-computation-generic-polynomials-pclmulqdq-paper.pdf
[2] https://github.com/rawrunprotected/crc
2022-11-14 23:05:46 +02:00
Lasse Collin eb0f1450ad liblzma: Use __attribute__((__constructor__)) if available.
This uses it for CRC table initializations when using --disable-small.
It avoids mythread_once() overhead. It also means that then
--disable-small --disable-threads is thread-safe if this attribute
is supported.
2022-11-14 16:00:52 +02:00
Lasse Collin 781da8d6c4 CMake: Add lzip decoder files and #define to the build. 2022-11-09 14:45:05 +02:00
Lasse Collin ecb966de30 liblzma: Add experimental ARM64 BCJ filter with a temporary Filter ID.
That is, the Filter ID will be changed once the design is final.
The current version will be removed. So files created with the
tempoary Filter ID won't be supported in the future.
2022-09-19 20:23:46 +03:00
Jia Tan ba3e4ba2de CMake: Clarify a comment about Windows symlinks without file extension. 2022-09-08 15:07:00 +03:00
Lasse Collin 17485e884c CMake: Update for liblzma_*.map files and fix wrong common_w32res.rc dep.
The previous commit split liblzma.map into liblzma_linux.map and
liblzma_generic.map. This commit updates the CMake build for those.

common_w32res.rc dependency was listed under Linux/FreeBSD while
obviously it belongs to Windows when building a DLL.
2022-09-08 15:02:41 +03:00
Lasse Collin 80a1a8bb83 CMake: Add xz symlinks.
These are a minor thing especially since the xz build has
some real problems still like lack of large file support
on 32-bit systems but I'll commit this since the code exists.

Thanks to Jia Tan.
2022-08-31 16:42:04 +03:00
Lasse Collin a4193bb6d8 CMake: Put xz man page install under if(UNIX) like is for xzdec.
Thanks to Jia Tan.
2022-08-31 16:29:38 +03:00
Lasse Collin df23c31000 CMake: Add liblzma tests.
Thanks to Jia Tan for the patch.
2022-08-22 16:46:18 +03:00
Nicholas Jackson 340cf1ec39 CMake: Add missing source file to liblzma build 2022-07-19 23:21:44 +03:00
huangqinjin 2bd36c91d0 CMake: Keep compatible with Windows 95 for 32-bit build. 2022-02-06 22:49:39 +02:00
Lasse Collin e7da44d515 CMake: Use interface library for better FindLibLZMA compatibility.
https://www.mail-archive.com/xz-devel@tukaani.org/msg00446.html

Thanks to Markus Rickert.
2021-02-13 23:31:27 +02:00
Lasse Collin a61dd82ada CMake: Try to improve compatibility with the FindLibLZMA module.
The naming conflict with FindLibLZMA module gets worse.
Not avoiding it in the first place was stupid.

Normally find_package(LibLZMA) will use the module and
find_package(liblzma 5.2.5 REQUIRED CONFIG) will use the config
file even with a case insensitive file system. However, if
CMAKE_FIND_PACKAGE_PREFER_CONFIG is TRUE and the file system
is case insensitive, find_package(LibLZMA) will find our liblzma
config file instead of using FindLibLZMA module.

One big problem with this is that FindLibLZMA uses
LibLZMA::LibLZMA and we use liblzma::liblzma as the target
name. With target names CMake happens to be case sensitive.
To workaround this, this commit adds

    add_library(LibLZMA::LibLZMA ALIAS liblzma::liblzma)

to the config file. Then both spellings work.

To make the behavior consistent between case sensitive and
insensitive file systems, the config and related files are
renamed from liblzmaConfig.cmake to liblzma-config.cmake style.
With this style CMake looks for lowercase version of the package
name so find_package(LiBLzmA 5.2.5 REQUIRED CONFIG) will work
to find our config file.

There are other differences between our config file and
FindLibLZMA so it's still possible that things break for
reasons other than the spelling of the target name. Hopefully
those situations aren't too common.

When the config file is available, it should always give as good or
better results as FindLibLZMA so this commit doesn't affect the
recommendation to use find_package(liblzma 5.2.5 REQUIRED CONFIG)
which explicitly avoids FindLibLZMA.

Thanks to Markus Rickert.
2021-01-30 18:36:04 +02:00
Lasse Collin 21588ca34a Build: Don't build bundles on Apple OSes.
Thanks to Daniel Packard.
2020-12-16 18:30:14 +02:00
Lasse Collin 2f108abb3d CMake: Fix compatibility with CMake 3.13.
The syntax "if(DEFINED CACHE{FOO})" requires CMake 3.14.
In some other places the code treats the cache variables
like normal variables already (${FOO} or if(FOO) is used,
not ${CACHE{FOO}).

Thanks to ygrek for reporting the bug on IRC.
2020-11-17 21:09:39 +02:00
Lasse Collin 265daa873c Build: Make CMake build fail if tuklib_cpucores or tuklib_physmem fails. 2020-02-27 20:58:52 +02:00
Lasse Collin 474320e990 Build: Fix bugs in the CMake files.
Seems that the phrase "add more quotes" from sh/bash scripting
applies to CMake as well. E.g. passing an unquoted list ${FOO}
to a function that expects one argument results in only the
first element of the list being passed as an argument and
the rest get ignored. Adding quotes helps ("${FOO}").

list(INSERT ...) is weird. Inserting an empty string to an empty
variable results in empty list, but inserting it to a non-empty
variable does insert an empty element to the list.

Since INSERT requires at least one element,
"${CMAKE_THREAD_LIBS_INIT}" needs to be quoted in CMakeLists.txt.
It might result in an empty element in the list. It seems to not
matter as empty elements consistently get ignored in that variable.
In fact, calling cmake_check_push_state() and cmake_check_pop_state()
will strip the empty elements from CMAKE_REQUIRED_LIBRARIES!

In addition to quoting fixes, this fixes checks for the cache
variables in tuklib_cpucores.cmake and tuklib_physmem.cmake.

Thanks to Martin Matuška for testing and reporting the problems.
These fixes aren't tested yet but hopefully they soon will be.
2020-02-25 20:44:10 +02:00
Lasse Collin 7e3493d40e Build: Add very limited experimental CMake support.
This does *NOT* replace the Autotools-based build system in
the foreseeable future. See the comment in the beginning
of CMakeLists.txt.

So far this has been tested only on GNU/Linux but I commit
it anyway to make it easier for others to test. Since I
haven't played much with CMake before, it's likely that
there are things that have been done in a silly or wrong
way and need to be fixed.
2020-02-25 00:00:32 +02:00