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.

I had missed this when writing the commit
5db75054.

Thanks to Jun I Jin.

The Maj macro is used where multiple things are added
together, so making Maj a sum of two expressions allows
some extra freedom for the compiler to schedule the
instructions.

I learned this trick from
<http://www.hackersdelight.org/corres.txt>.

This looks weird because the rotations become sequential,
but it helps quite a bit on both 32-bit and 64-bit x86:

  - It requires fewer instructions on two-operand
    instruction sets like x86.

  - It requires one register less which matters especially
    on 32-bit x86.

I hope this doesn't hurt other archs.

I didn't invent this idea myself, but I don't remember where
I saw it first.

The unrolling in the previous commit should avoid the
situation where a compiler may think that an uninitialized
variable might be accessed.

This way a branch isn't needed for each operation
to choose between blk0 and blk2, and still the code
doesn't grow as much as it would with full unrolling.

Two locations were not changed yet because the simplest change
assumes that the initial "len" may be greater than "limit".

This doesn't change the match finder output.

This commit just adds the function. Its uses will be in
separate commits.

This hasn't been tested much yet and it's perhaps a bit early
to commit it but if there are bugs they should get found quite
quickly.

Thanks to Jun I Jin from Intel for help and for pointing out
that string comparison needs to be optimized in liblzma.

Thanks to Trần Ngọc Quân.

Updated: --threads, --block-size, and --block-list
Added: --flush-timeout

This avoids LZMA_PROG_ERROR from lzma_code() with filter chains
that don't support LZMA_SYNC_FLUSH.

Now it should be close to the functionality of the original
version by Pavel Raiskup.

This is a simplified version of Pavel Raiskup's
original patch.

Mimic the original grep behavior and return exit_success when
at least one xz compressed file matches given pattern.

Original bugreport:
https://bugzilla.redhat.com/show_bug.cgi?id=1108085

Thanks to Pavel Raiskup for the patch.

This avoids a memzero() call for a newly-allocated memory,
which can be expensive when encoding small streams with
an over-sized dictionary.

To avoid using lzma_alloc_zero() for memory that doesn't
need to be zeroed, lzma_mf.son is now allocated separately,
which requires handling it separately in normalize() too.

Thanks to Vincenzo Innocente for reporting the problem.

Thanks to Christian Hesse.

It can be confusing that two header files have the same name.
The public API file is still lzma.h.

In this case "make install" could fail if the man page directory
didn't already exist at the destination. If it did exist, a
dangling symlink was created there. Now the link is omitted
instead. This isn't the best fix but it's better than the old
behavior.

Add a note about failing "make check". The source of
the problem should be fixed in libtool (if it really is
a libtool bug and not mine) but I'm unable to spend time
on that for now. Thanks to Nelson H. F. Beebe for reporting
the issue.

Add a note about a possible need to run "ldconfig" after
"make install".

I don't know the details but I have an impression that there's
no problem in practice if using GCC since people have built xz
with GCC (without patching xz), but renaming the variable cannot
hurt either.

Thanks to Mark Ashley.