clang supports the __has_warning macro to determine if the version of
clang compiling the code supports a given warning. If we do not define
it for other compilers, it may cause a preprocessor error.
The 32-bit build needs to be first so the configure cache only needs to
be reset one time. The 32-bit build sets the CFLAGS env variable, so any
build using that flag after will fail unless the cache is reset.
Calling coder_set_compression_settings() in list mode with verbose mode
on caused the filter chain and memory requirements to print. This was
unnecessary since the command results in an error and not consistent
with other formats like lzma and alone.
Disabling shared library generation and linking should help speed up the
runners. The shared library is still being tested in the 32 bit build
and the full feature.
Disabling nls is to check for any unexpected warnings or errors.
It's not that important. It can be annoying in builds that
disable many features since in those cases the tests programs
will correctly trigger this warning with Clang.
It doesn't warn on a 64-bit system because truncating
a ptrdiff_t (signed long) to uint32_t is diagnosed under
-Wconversion by GCC and -Wshorten-64-to-32 by Clang.
-Wstrict-aliasing was removed from the list since it is enabled
by -Wall already.
A normal build is clean with these on GNU/Linux x86-64 with
GCC 12.2.0 and Clang 14.0.6.
Explicitly casting the integer to lzma_check silences the warning.
Since such an invalid value is needed in multiple tests, a constant
INVALID_LZMA_CHECK_ID was added to tests.h.
The use of 0x1000 for lzma_block.check wasn't optimal as if
the underlying type is a char then 0x1000 will be truncated to 0.
However, in these test cases the value is ignored, thus even with
such truncation the test would have passed.
Note that assigning an unsigned int to lzma_check doesn't warn
on GNU/Linux x86-64 since the enum type is unsigned on that
platform. The enum can be signed on some other platform though
so it's best to use enumeration type lzma_check in these situations.
This is similar to 2ce4f36f17.
The actual initialization of the variables is done inside
mythread_sync() macro. Clang doesn't seem to see that
the initialization code inside the macro is always executed.
clang and gcc differ in how they handle -Wformat-nonliteral. gcc will
allow a non-literal format string as long as the function takes its
format arguments as a va_list.
This only occurs in test_filter_flags when the BCJ filters are not
configured and built. In this case, ARRAY_SIZE() returns 0 and causes a
type-limits warning with the loop variable since an unsigned number will
always be >= 0.
This affects only 32-bit x86 builds. x86-64 is OK as is.
I still cannot easily test this myself. The reporter has tested
this and it passes the tests included in the CMake build and
performance is good: raw CRC64 is 2-3 times faster than the
C version of the slice-by-four method. (Note that liblzma doesn't
include a MSVC-compatible version of the 32-bit x86 assembly code
for the slice-by-four method.)
Thanks to Iouri Kharon for figuring out a fix, testing, and
benchmarking.
This reverts commit 36edc65ab4.
It was reported that it wasn't a good enough fix and MSVC
still produced (different kind of) bad code when building
for 32-bit x86 if optimizations are enabled.
Thanks to Iouri Kharon.
On some platforms src/xz/suffix.c may need <strings.h> for
strcasecmp() but suffix.c includes the header when it needs it.
Unless there is an old system that otherwise supports enough C99
to build XZ Utils but doesn't have C89/C90-compatible <string.h>,
there should be no need to include <strings.h> in sysdefs.h.
SUSv2 and POSIX.1‐2017 declare only a few functions in <strings.h>.
Of these, strcasecmp() is used on some platforms in suffix.c.
Nothing else in the project needs <strings.h> (at least if
building on a modern system).
sysdefs.h currently includes <strings.h> if HAVE_STRINGS_H is
defined and suffix.c relied on this.
Note that dos/config.h doesn't #define HAVE_STRINGS_H even though
DJGPP does have strings.h. It isn't needed with DJGPP as strcasecmp()
is also in <string.h> in DJGPP.
It quite probably was never needed, that is, any system where memory.h
was required likely couldn't compile XZ Utils for other reasons anyway.
XZ Utils 5.2.6 and later source packages were generated using
Autoconf 2.71 which no longer defines HAVE_MEMORY_H. So the code
being removed is no longer used anyway.
At least on some systems, GNU windres needs --use-temp-file
in addition to the \x20 hack to avoid spaces in the command line
argument. Hovever, that \x20 syntax is broken with llvm-windres
version 15.0.0 (results in "XZx20Utils") but luckily it works
with a regular space. Thus it is best to limit the workarounds
to GNU toolchain on Windows.
Here are the list of the most significant issues addressed:
- Avoid using internal common.h header. It's not good to copy the
constants like this but common.h cannot be included for use outside
of liblzma. This is the quickest thing to do that could be fixed later.
- Omit the INIT_FILTER macro. Initialization should be done with just
regular designated initializers.
- Use start_offset = 257 for BCJ tests. It demonstrates that Filter
Flags encoder and decoder don't validate the options thoroughly.
257 is valid only for the x86 filter. This is a bit silly but
not a significant problem in practice because the encoder and
decoder initialization functions will catch bad alignment still.
Perhaps this should be fixed but it's not urgent and doesn't need
to be in 5.4.x.
- Various tweaks to comments such as filter id -> Filter ID
I haven't tested with MSVC myself and there doesn't seem to be
information about the problem online, so I'm relying on the bug report.
Thanks to Iouri Kharon for the bug report and the patch.
It's not needed in XZ Utils at least for now. It's good to support
it still because if such use is needed later, it wouldn't be
caught on GNU/Linux since malloc(0) from glibc returns non-NULL.
Jia Tan
Lasse Collin