liblzma: Vaccinate against an ill patch from RHEL/CentOS 7.

RHEL/CentOS 7 shipped with 5.1.2alpha, including the threaded
encoder that is behind #ifdef LZMA_UNSTABLE in the API headers.
In 5.1.2alpha these symbols are under XZ_5.1.2alpha in liblzma.map.
API/ABI compatibility tracking isn't done between development
releases so newer releases didn't have XZ_5.1.2alpha anymore.

Later RHEL/CentOS 7 updated xz to 5.2.2 but they wanted to keep
the exported symbols compatible with 5.1.2alpha. After checking
the ABI changes it turned out that >= 5.2.0 ABI is backward
compatible with the threaded encoder functions from 5.1.2alpha
(but not vice versa as fixes and extensions to these functions
were made between 5.1.2alpha and 5.2.0).

In RHEL/CentOS 7, XZ Utils 5.2.2 was patched with
xz-5.2.2-compat-libs.patch to modify liblzma.map:

  - XZ_5.1.2alpha was added with lzma_stream_encoder_mt and
    lzma_stream_encoder_mt_memusage. This matched XZ Utils 5.1.2alpha.

  - XZ_5.2 was replaced with XZ_5.2.2. It is clear that this was
    an error; the intention was to keep using XZ_5.2 (XZ_5.2.2
    has never been used in XZ Utils). So XZ_5.2.2 lists all
    symbols that were listed under XZ_5.2 before the patch.
    lzma_stream_encoder_mt and _mt_memusage are included too so
    they are listed both here and under XZ_5.1.2alpha.

The patch didn't add any __asm__(".symver ...") lines to the .c
files. Thus the resulting liblzma.so exports the threaded encoder
functions under XZ_5.1.2alpha only. Listing the two functions
also under XZ_5.2.2 in liblzma.map has no effect without
matching .symver lines.

The lack of XZ_5.2 in RHEL/CentOS 7 means that binaries linked
against unpatched XZ Utils 5.2.x won't run on RHEL/CentOS 7.
This is unfortunate but this alone isn't too bad as the problem
is contained within RHEL/CentOS 7 and doesn't affect users
of other distributions. It could also be fixed internally in
RHEL/CentOS 7.

The second problem is more serious: In XZ Utils 5.2.2 the API
headers don't have #ifdef LZMA_UNSTABLE for obvious reasons.
This is true in RHEL/CentOS 7 version too. Thus now programs
using new APIs can be compiled without an extra #define. However,
the programs end up depending on symbol version XZ_5.1.2alpha
(and possibly also XZ_5.2.2) instead of XZ_5.2 as they would
with an unpatched XZ Utils 5.2.2. This means that such binaries
won't run on other distributions shipping XZ Utils >= 5.2.0 as
they don't provide XZ_5.1.2alpha or XZ_5.2.2; they only provide
XZ_5.2 (and XZ_5.0). (This includes RHEL/CentOS 8 as the patch
luckily isn't included there anymore with XZ Utils 5.2.4.)

Binaries built by RHEL/CentOS 7 users get distributed and then
people wonder why they don't run on some other distribution.
Seems that people have found out about the patch and been copying
it to some build scripts, seemingly curing the symptoms but
actually spreading the illness further and outside RHEL/CentOS 7.

The ill patch seems to be from late 2016 (RHEL 7.3) and in 2017 it
had spread at least to EasyBuild. I heard about the events only
recently. :-(

This commit splits liblzma.map into two versions: one for
GNU/Linux and another for other OSes that can use symbol versioning
(FreeBSD, Solaris, maybe others). The Linux-specific file and the
matching additions to .c files add full compatibility with binaries
that have been built against a RHEL/CentOS-patched liblzma. Builds
for OSes other than GNU/Linux won't get the vaccine as they should
be immune to the problem (I really hope that no build script uses
the RHEL/CentOS 7 patch outside GNU/Linux).

The RHEL/CentOS compatibility symbols XZ_5.1.2alpha and XZ_5.2.2
are intentionally put *after* XZ_5.2 in liblzma_linux.map. This way
if one forgets to #define HAVE_SYMBOL_VERSIONS_LINUX when building,
the resulting liblzma.so.5 will have lzma_stream_encoder_mt@@XZ_5.2
since XZ_5.2 {...} is the first one that lists that function.
Without HAVE_SYMBOL_VERSIONS_LINUX @XZ_5.1.2alpha and @XZ_5.2.2
will be missing but that's still a minor problem compared to
only having lzma_stream_encoder_mt@@XZ_5.1.2alpha!

The "local: *;" line was moved to XZ_5.0 so that it doesn't need
to be moved around. It doesn't matter where it is put.

Having two similar liblzma_*.map files is a bit silly as it is,
at least for now, easily possible to generate the generic one
from the Linux-specific file. But that adds extra steps and
increases the risk of mistakes when supporting more than one
build system. So I rather maintain two files in parallel and let
validate_map.sh check that they are in sync when "make mydist"
is run.

This adds .symver lines for lzma_stream_encoder_mt@XZ_5.2.2 and
lzma_stream_encoder_mt_memusage@XZ_5.2.2 even though these
weren't exported by RHEL/CentOS 7 (only @@XZ_5.1.2alpha was
for these two). I added these anyway because someone might
misunderstand the RHEL/CentOS 7 patch and think that @XZ_5.2.2
(@@XZ_5.2.2) versions were exported too.

At glance one could suggest using __typeof__ to copy the function
prototypes when making aliases. However, this doesn't work trivially
because __typeof__ won't copy attributes (lzma_nothrow, lzma_pure)
and it won't change symbol visibility from hidden to default (done
by LZMA_API()). Attributes could be copied with __copy__ attribute
but that needs GCC 9 and a fallback method would be needed anyway.

This uses __symver__ attribute with GCC >= 10 and
__asm__(".symver ...") with everything else. The attribute method
is required for LTO (-flto) support with GCC. Using -flto with
GCC older than 10 is now broken on GNU/Linux and will not be fixed
(can silently result in a broken liblzma build that has dangerously
incorrect symbol versions). LTO builds with Clang seem to work
with the traditional __asm__(".symver ...") method.

Thanks to Boud Roukema for reporting the problem and discussing
the details and testing the fix.

10 files changed, 382 insertions, 15 deletions

diff --git a/configure.ac b/configure.ac
index 7443489a..57f60f69 100644
--- a/configure.ac
+++ b/configure.ac
@@ -493,7 +493,28 @@ if test "x$enable_symbol_versions" = xauto; then
 	esac
 fi
 AC_MSG_RESULT([$enable_symbol_versions])
-AM_CONDITIONAL([COND_SYMVERS], [test "x$enable_symbol_versions" = xyes])
+
+# There are two variants for symbol versioning.
+# See src/liblzma/validate_map.sh for details.
+if test "x$enable_symbol_versions" = xyes; then
+	case $host_os in
+		linux*)
+			enable_symbol_versions=linux
+			AC_DEFINE([HAVE_SYMBOL_VERSIONS_LINUX], [1],
+				[Define to 1 to if GNU/Linux-specific details
+				are wanted for symbol versioning. This must
+				be used together with liblzma_linux.map.])
+			;;
+		*)
+			enable_symbol_versions=generic
+			;;
+	esac
+fi
+
+AM_CONDITIONAL([COND_SYMVERS_LINUX],
+	[test "x$enable_symbol_versions" = xlinux])
+AM_CONDITIONAL([COND_SYMVERS_GENERIC],
+	[test "x$enable_symbol_versions" = xgeneric])
 
 
 ##############
diff --git a/src/liblzma/Makefile.am b/src/liblzma/Makefile.am
index 6323e26a..d73d5f0a 100644
--- a/src/liblzma/Makefile.am
+++ b/src/liblzma/Makefile.am
@@ -26,10 +26,14 @@ liblzma_la_CPPFLAGS = \
 	-DTUKLIB_SYMBOL_PREFIX=lzma_
 liblzma_la_LDFLAGS = -no-undefined -version-info 8:99:3
 
-EXTRA_DIST += liblzma.map validate_map.sh
-if COND_SYMVERS
+EXTRA_DIST += liblzma_generic.map liblzma_linux.map validate_map.sh
+if COND_SYMVERS_GENERIC
 liblzma_la_LDFLAGS += \
-	-Wl,--version-script=$(top_srcdir)/src/liblzma/liblzma.map
+	-Wl,--version-script=$(top_srcdir)/src/liblzma/liblzma_generic.map
+endif
+if COND_SYMVERS_LINUX
+liblzma_la_LDFLAGS += \
+	-Wl,--version-script=$(top_srcdir)/src/liblzma/liblzma_linux.map
 endif
 
 liblzma_la_SOURCES += ../common/tuklib_physmem.c
diff --git a/src/liblzma/common/block_buffer_encoder.c b/src/liblzma/common/block_buffer_encoder.c
index 39e263aa..a47342ef 100644
--- a/src/liblzma/common/block_buffer_encoder.c
+++ b/src/liblzma/common/block_buffer_encoder.c
@@ -325,6 +325,24 @@ lzma_block_buffer_encode(lzma_block *block, const lzma_allocator *allocator,
 }
 
 
+#ifdef HAVE_SYMBOL_VERSIONS_LINUX
+// This is for compatibility with binaries linked against liblzma that
+// has been patched with xz-5.2.2-compat-libs.patch from RHEL/CentOS 7.
+LZMA_SYMVER_API("lzma_block_uncomp_encode@XZ_5.2.2",
+	lzma_ret, lzma_block_uncomp_encode_522)(lzma_block *block,
+		const uint8_t *in, size_t in_size,
+		uint8_t *out, size_t *out_pos, size_t out_size)
+		lzma_nothrow lzma_attr_warn_unused_result
+		__attribute__((__alias__("lzma_block_uncomp_encode_52")));
+
+LZMA_SYMVER_API("lzma_block_uncomp_encode@@XZ_5.2",
+	lzma_ret, lzma_block_uncomp_encode_52)(lzma_block *block,
+		const uint8_t *in, size_t in_size,
+		uint8_t *out, size_t *out_pos, size_t out_size)
+		lzma_nothrow lzma_attr_warn_unused_result;
+
+#define lzma_block_uncomp_encode lzma_block_uncomp_encode_52
+#endif
 extern LZMA_API(lzma_ret)
 lzma_block_uncomp_encode(lzma_block *block,
 		const uint8_t *in, size_t in_size,
diff --git a/src/liblzma/common/common.c b/src/liblzma/common/common.c
index 346fc7af..a708fdf1 100644
--- a/src/liblzma/common/common.c
+++ b/src/liblzma/common/common.c
@@ -374,6 +374,20 @@ lzma_end(lzma_stream *strm)
 }
 
 
+#ifdef HAVE_SYMBOL_VERSIONS_LINUX
+// This is for compatibility with binaries linked against liblzma that
+// has been patched with xz-5.2.2-compat-libs.patch from RHEL/CentOS 7.
+LZMA_SYMVER_API("lzma_get_progress@XZ_5.2.2",
+	void, lzma_get_progress_522)(lzma_stream *strm,
+		uint64_t *progress_in, uint64_t *progress_out) lzma_nothrow
+		__attribute__((__alias__("lzma_get_progress_52")));
+
+LZMA_SYMVER_API("lzma_get_progress@@XZ_5.2",
+	void, lzma_get_progress_52)(lzma_stream *strm,
+		uint64_t *progress_in, uint64_t *progress_out) lzma_nothrow;
+
+#define lzma_get_progress lzma_get_progress_52
+#endif
 extern LZMA_API(void)
 lzma_get_progress(lzma_stream *strm,
 		uint64_t *progress_in, uint64_t *progress_out)
diff --git a/src/liblzma/common/common.h b/src/liblzma/common/common.h
index 36366dbc..4783e45d 100644
--- a/src/liblzma/common/common.h
+++ b/src/liblzma/common/common.h
@@ -34,6 +34,34 @@
 
 #include "lzma.h"
 
+#ifdef HAVE_SYMBOL_VERSIONS_LINUX
+// To keep link-time optimization (LTO, -flto) working with GCC,
+// the __symver__ attribute must be used instead of __asm__(".symver ...").
+// Otherwise the symbol versions may be lost, resulting in broken liblzma
+// that has wrong default versions in the exported symbol list!
+// The attribute was added in GCC 10; LTO with older GCC is not supported.
+//
+// To keep -Wmissing-prototypes happy, use LZMA_SYMVER_API only with function
+// declarations (including those with __alias__ attribute) and LZMA_API with
+// the function definitions. This means a little bit of silly copy-and-paste
+// between declarations and definitions though.
+//
+// As of GCC 12.2, the __symver__ attribute supports only @ and @@ but the
+// very convenient @@@ isn't supported (it's supported by GNU assembler
+// since 2000). When using @@ instead of @@@, the internal name must not be
+// the same as the external name to avoid problems in some situations. This
+// is why "#define foo_52 foo" is needed for the default symbol versions.
+#	if TUKLIB_GNUC_REQ(10, 0)
+#		define LZMA_SYMVER_API(extnamever, type, intname) \
+			extern __attribute__((__symver__(extnamever))) \
+					LZMA_API(type) intname
+#	else
+#		define LZMA_SYMVER_API(extnamever, type, intname) \
+			__asm__(".symver " #intname "," extnamever); \
+			extern LZMA_API(type) intname
+#	endif
+#endif
+
 // These allow helping the compiler in some often-executed branches, whose
 // result is almost always the same.
 #ifdef __GNUC__
diff --git a/src/liblzma/common/hardware_cputhreads.c b/src/liblzma/common/hardware_cputhreads.c
index f468366a..5d246d2c 100644
--- a/src/liblzma/common/hardware_cputhreads.c
+++ b/src/liblzma/common/hardware_cputhreads.c
@@ -15,6 +15,18 @@
 #include "tuklib_cpucores.h"
 
 
+#ifdef HAVE_SYMBOL_VERSIONS_LINUX
+// This is for compatibility with binaries linked against liblzma that
+// has been patched with xz-5.2.2-compat-libs.patch from RHEL/CentOS 7.
+LZMA_SYMVER_API("lzma_cputhreads@XZ_5.2.2",
+	uint32_t, lzma_cputhreads_522)(void) lzma_nothrow
+		__attribute__((__alias__("lzma_cputhreads_52")));
+
+LZMA_SYMVER_API("lzma_cputhreads@@XZ_5.2",
+	uint32_t, lzma_cputhreads_52)(void) lzma_nothrow;
+
+#define lzma_cputhreads lzma_cputhreads_52
+#endif
 extern LZMA_API(uint32_t)
 lzma_cputhreads(void)
 {
diff --git a/src/liblzma/common/stream_encoder_mt.c b/src/liblzma/common/stream_encoder_mt.c
index 24addd40..fb56a96f 100644
--- a/src/liblzma/common/stream_encoder_mt.c
+++ b/src/liblzma/common/stream_encoder_mt.c
@@ -1096,6 +1096,31 @@ stream_encoder_mt_init(lzma_next_coder *next, const lzma_allocator *allocator,
 }
 
 
+#ifdef HAVE_SYMBOL_VERSIONS_LINUX
+// These are for compatibility with binaries linked against liblzma that
+// has been patched with xz-5.2.2-compat-libs.patch from RHEL/CentOS 7.
+// Actually that patch didn't create lzma_stream_encoder_mt@XZ_5.2.2
+// but it has been added here anyway since someone might misread the
+// RHEL patch and think both @XZ_5.1.2alpha and @XZ_5.2.2 exist.
+LZMA_SYMVER_API("lzma_stream_encoder_mt@XZ_5.1.2alpha",
+	lzma_ret, lzma_stream_encoder_mt_512a)(
+		lzma_stream *strm, const lzma_mt *options)
+		lzma_nothrow lzma_attr_warn_unused_result
+		__attribute__((__alias__("lzma_stream_encoder_mt_52")));
+
+LZMA_SYMVER_API("lzma_stream_encoder_mt@XZ_5.2.2",
+	lzma_ret, lzma_stream_encoder_mt_522)(
+		lzma_stream *strm, const lzma_mt *options)
+		lzma_nothrow lzma_attr_warn_unused_result
+		__attribute__((__alias__("lzma_stream_encoder_mt_52")));
+
+LZMA_SYMVER_API("lzma_stream_encoder_mt@@XZ_5.2",
+	lzma_ret, lzma_stream_encoder_mt_52)(
+		lzma_stream *strm, const lzma_mt *options)
+		lzma_nothrow lzma_attr_warn_unused_result;
+
+#define lzma_stream_encoder_mt lzma_stream_encoder_mt_52
+#endif
 extern LZMA_API(lzma_ret)
 lzma_stream_encoder_mt(lzma_stream *strm, const lzma_mt *options)
 {
@@ -1111,6 +1136,23 @@ lzma_stream_encoder_mt(lzma_stream *strm, const lzma_mt *options)
 }
 
 
+#ifdef HAVE_SYMBOL_VERSIONS_LINUX
+LZMA_SYMVER_API("lzma_stream_encoder_mt_memusage@XZ_5.1.2alpha",
+	uint64_t, lzma_stream_encoder_mt_memusage_512a)(
+	const lzma_mt *options) lzma_nothrow lzma_attr_pure
+	__attribute__((__alias__("lzma_stream_encoder_mt_memusage_52")));
+
+LZMA_SYMVER_API("lzma_stream_encoder_mt_memusage@XZ_5.2.2",
+	uint64_t, lzma_stream_encoder_mt_memusage_522)(
+	const lzma_mt *options) lzma_nothrow lzma_attr_pure
+	__attribute__((__alias__("lzma_stream_encoder_mt_memusage_52")));
+
+LZMA_SYMVER_API("lzma_stream_encoder_mt_memusage@@XZ_5.2",
+	uint64_t, lzma_stream_encoder_mt_memusage_52)(
+	const lzma_mt *options) lzma_nothrow lzma_attr_pure;
+
+#define lzma_stream_encoder_mt_memusage lzma_stream_encoder_mt_memusage_52
+#endif
 // This function name is a monster but it's consistent with the older
 // monster names. :-( 31 chars is the max that C99 requires so in that
 // sense it's not too long. ;-)
diff --git a/src/liblzma/liblzma.map b/src/liblzma/liblzma_generic.map
index c37f293d..0251c4c2 100644
--- a/src/liblzma/liblzma.map
+++ b/src/liblzma/liblzma_generic.map
@@ -93,6 +93,9 @@ global:
 	lzma_vli_decode;
 	lzma_vli_encode;
 	lzma_vli_size;
+
+local:
+	*;
 };
 
 XZ_5.2 {
@@ -110,7 +113,4 @@ global:
 	lzma_microlzma_encoder;
 	lzma_file_info_decoder;
 	lzma_stream_decoder_mt;
-
-local:
-	*;
 } XZ_5.2;
diff --git a/src/liblzma/liblzma_linux.map b/src/liblzma/liblzma_linux.map
new file mode 100644
index 00000000..7a09e03c
--- /dev/null
+++ b/src/liblzma/liblzma_linux.map
@@ -0,0 +1,131 @@
+XZ_5.0 {
+global:
+	lzma_alone_decoder;
+	lzma_alone_encoder;
+	lzma_auto_decoder;
+	lzma_block_buffer_bound;
+	lzma_block_buffer_decode;
+	lzma_block_buffer_encode;
+	lzma_block_compressed_size;
+	lzma_block_decoder;
+	lzma_block_encoder;
+	lzma_block_header_decode;
+	lzma_block_header_encode;
+	lzma_block_header_size;
+	lzma_block_total_size;
+	lzma_block_unpadded_size;
+	lzma_check_is_supported;
+	lzma_check_size;
+	lzma_code;
+	lzma_crc32;
+	lzma_crc64;
+	lzma_easy_buffer_encode;
+	lzma_easy_decoder_memusage;
+	lzma_easy_encoder;
+	lzma_easy_encoder_memusage;
+	lzma_end;
+	lzma_filter_decoder_is_supported;
+	lzma_filter_encoder_is_supported;
+	lzma_filter_flags_decode;
+	lzma_filter_flags_encode;
+	lzma_filter_flags_size;
+	lzma_filters_copy;
+	lzma_filters_update;
+	lzma_get_check;
+	lzma_index_append;
+	lzma_index_block_count;
+	lzma_index_buffer_decode;
+	lzma_index_buffer_encode;
+	lzma_index_cat;
+	lzma_index_checks;
+	lzma_index_decoder;
+	lzma_index_dup;
+	lzma_index_encoder;
+	lzma_index_end;
+	lzma_index_file_size;
+	lzma_index_hash_append;
+	lzma_index_hash_decode;
+	lzma_index_hash_end;
+	lzma_index_hash_init;
+	lzma_index_hash_size;
+	lzma_index_init;
+	lzma_index_iter_init;
+	lzma_index_iter_locate;
+	lzma_index_iter_next;
+	lzma_index_iter_rewind;
+	lzma_index_memusage;
+	lzma_index_memused;
+	lzma_index_size;
+	lzma_index_stream_count;
+	lzma_index_stream_flags;
+	lzma_index_stream_padding;
+	lzma_index_stream_size;
+	lzma_index_total_size;
+	lzma_index_uncompressed_size;
+	lzma_lzma_preset;
+	lzma_memlimit_get;
+	lzma_memlimit_set;
+	lzma_memusage;
+	lzma_mf_is_supported;
+	lzma_mode_is_supported;
+	lzma_physmem;
+	lzma_properties_decode;
+	lzma_properties_encode;
+	lzma_properties_size;
+	lzma_raw_buffer_decode;
+	lzma_raw_buffer_encode;
+	lzma_raw_decoder;
+	lzma_raw_decoder_memusage;
+	lzma_raw_encoder;
+	lzma_raw_encoder_memusage;
+	lzma_stream_buffer_bound;
+	lzma_stream_buffer_decode;
+	lzma_stream_buffer_encode;
+	lzma_stream_decoder;
+	lzma_stream_encoder;
+	lzma_stream_flags_compare;
+	lzma_stream_footer_decode;
+	lzma_stream_footer_encode;
+	lzma_stream_header_decode;
+	lzma_stream_header_encode;
+	lzma_version_number;
+	lzma_version_string;
+	lzma_vli_decode;
+	lzma_vli_encode;
+	lzma_vli_size;
+
+local:
+	*;
+};
+
+XZ_5.2 {
+global:
+	lzma_block_uncomp_encode;
+	lzma_cputhreads;
+	lzma_get_progress;
+	lzma_stream_encoder_mt;
+	lzma_stream_encoder_mt_memusage;
+} XZ_5.0;
+
+XZ_5.1.2alpha {
+global:
+	lzma_stream_encoder_mt;
+	lzma_stream_encoder_mt_memusage;
+} XZ_5.0;
+
+XZ_5.2.2 {
+global:
+	lzma_block_uncomp_encode;
+	lzma_cputhreads;
+	lzma_get_progress;
+	lzma_stream_encoder_mt;
+	lzma_stream_encoder_mt_memusage;
+} XZ_5.1.2alpha;
+
+XZ_5.3.3alpha {
+global:
+	lzma_microlzma_decoder;
+	lzma_microlzma_encoder;
+	lzma_file_info_decoder;
+	lzma_stream_decoder_mt;
+} XZ_5.2;
diff --git a/src/liblzma/validate_map.sh b/src/liblzma/validate_map.sh
index 3aee4668..2bf6f8b9 100644
--- a/src/liblzma/validate_map.sh
+++ b/src/liblzma/validate_map.sh
@@ -2,7 +2,79 @@
 
 ###############################################################################
 #
-# Check liblzma.map for certain types of errors
+# Check liblzma_*.map for certain types of errors.
+#
+# liblzma_generic.map is for FreeBSD and Solaris and possibly others
+# except GNU/Linux.
+#
+# liblzma_linux.map is for GNU/Linux only. This and the matching extra code
+# in the .c files make liblzma >= 5.2.7 compatible with binaries that were
+# linked against ill-patched liblzma in RHEL/CentOS 7. By providing the
+# compatibility in official XZ Utils release will hopefully prevent people
+# from further copying the broken patch to other places when they want
+# compatibility with binaries linked on RHEL/CentOS 7. The long version
+# of the story:
+#
+#     RHEL/CentOS 7 shipped with 5.1.2alpha, including the threaded
+#     encoder that is behind #ifdef LZMA_UNSTABLE in the API headers.
+#     In 5.1.2alpha these symbols are under XZ_5.1.2alpha in liblzma.map.
+#     API/ABI compatibility tracking isn't done between development
+#     releases so newer releases didn't have XZ_5.1.2alpha anymore.
+#
+#     Later RHEL/CentOS 7 updated xz to 5.2.2 but they wanted to keep
+#     the exported symbols compatible with 5.1.2alpha. After checking
+#     the ABI changes it turned out that >= 5.2.0 ABI is backward
+#     compatible with the threaded encoder functions from 5.1.2alpha
+#     (but not vice versa as fixes and extensions to these functions
+#     were made between 5.1.2alpha and 5.2.0).
+#
+#     In RHEL/CentOS 7, XZ Utils 5.2.2 was patched with
+#     xz-5.2.2-compat-libs.patch to modify liblzma.map:
+#
+#       - XZ_5.1.2alpha was added with lzma_stream_encoder_mt and
+#         lzma_stream_encoder_mt_memusage. This matched XZ Utils 5.1.2alpha.
+#
+#       - XZ_5.2 was replaced with XZ_5.2.2. It is clear that this was
+#         an error; the intention was to keep using XZ_5.2 (XZ_5.2.2
+#         has never been used in XZ Utils). So XZ_5.2.2 lists all
+#         symbols that were listed under XZ_5.2 before the patch.
+#         lzma_stream_encoder_mt and _mt_memusage are included too so
+#         they are listed both here and under XZ_5.1.2alpha.
+#
+#     The patch didn't add any __asm__(".symver ...") lines to the .c
+#     files. Thus the resulting liblzma.so exports the threaded encoder
+#     functions under XZ_5.1.2alpha only. Listing the two functions
+#     also under XZ_5.2.2 in liblzma.map has no effect without
+#     matching .symver lines.
+#
+#     The lack of XZ_5.2 in RHEL/CentOS 7 means that binaries linked
+#     against unpatched XZ Utils 5.2.x won't run on RHEL/CentOS 7.
+#     This is unfortunate but this alone isn't too bad as the problem
+#     is contained within RHEL/CentOS 7 and doesn't affect users
+#     of other distributions. It could also be fixed internally in
+#     RHEL/CentOS 7.
+#
+#     The second problem is more serious: In XZ Utils 5.2.2 the API
+#     headers don't have #ifdef LZMA_UNSTABLE for obvious reasons.
+#     This is true in RHEL/CentOS 7 version too. Thus now programs
+#     using new APIs can be compiled without an extra #define. However,
+#     the programs end up depending on symbol version XZ_5.1.2alpha
+#     (and possibly also XZ_5.2.2) instead of XZ_5.2 as they would
+#     with an unpatched XZ Utils 5.2.2. This means that such binaries
+#     won't run on other distributions shipping XZ Utils >= 5.2.0 as
+#     they don't provide XZ_5.1.2alpha or XZ_5.2.2; they only provide
+#     XZ_5.2 (and XZ_5.0). (This includes RHEL/CentOS 8 as the patch
+#     luckily isn't included there anymore with XZ Utils 5.2.4.)
+#
+#     Binaries built by RHEL/CentOS 7 users get distributed and then
+#     people wonder why they don't run on some other distribution.
+#     Seems that people have found out about the patch and been copying
+#     it to some build scripts, seemingly curing the symptoms but
+#     actually spreading the illness further and outside RHEL/CentOS 7.
+#     Adding compatibility in an official XZ Utils release should work
+#     as a vaccine against this ill patch and stop it from spreading.
+#     The vaccine is kept GNU/Linux-only as other OSes should be immune
+#     (hopefully it hasn't spread via some build script to other OSes).
 #
 # Author: Lasse Collin
 #
@@ -18,11 +90,11 @@ STATUS=0
 
 cd "$(dirname "$0")"
 
-# Get the list of symbols that aren't defined in liblzma.map.
+# Get the list of symbols that aren't defined in liblzma_generic.map.
 SYMS=$(sed -n 's/^extern LZMA_API([^)]*) \([a-z0-9_]*\)(.*$/\1;/p' \
 		api/lzma/*.h \
 	| sort \
-	| grep -Fve "$(sed '/[{}:*]/d;/^$/d;s/^	//' liblzma.map)")
+	| grep -Fve "$(sed '/[{}:*]/d;/^$/d;s/^	//' liblzma_generic.map)")
 
 # Check that there are no old alpha or beta versions listed.
 VER=$(cd ../.. && sh build-aux/version.sh)
@@ -30,21 +102,41 @@ NAMES=
 case $VER in
 	*alpha | *beta)
 		NAMES=$(sed -n 's/^.*XZ_\([^ ]*\)\(alpha\|beta\) .*$/\1\2/p' \
-			liblzma.map | grep -Fv "$VER")
+			liblzma_generic.map | grep -Fv "$VER")
 		;;
 esac
 
 # Check for duplicate lines. It can catch missing dependencies.
-DUPS=$(sort liblzma.map | sed '/^$/d;/^global:$/d' | uniq -d)
+DUPS=$(sort liblzma_generic.map | sed '/^$/d;/^global:$/d' | uniq -d)
+
+# Check that liblzma_linux.map is in sync with liblzma_generic.map.
+# The RHEL/CentOS 7 compatibility symbols are in a fixed location
+# so it makes it easy to remove them for comparison with liblzma_generic.map.
+#
+# NOTE: Putting XZ_5.2 before the compatibility symbols XZ_5.1.2alpha
+# and XZ_5.2.2 in liblzma_linux.map is important: If liblzma_linux.map is
+# incorrectly used without #define HAVE_SYMBOL_VERSIONS_LINUX, only the first
+# occurrence of each function name will be used from liblzma_linux.map;
+# the rest are ignored by the linker. Thus having XZ_5.2 before the
+# compatibility symbols means that @@XZ_5.2 will be used for the symbols
+# listed under XZ_5.2 {...} and the same function names later in
+# the file under XZ_5.1.2alpha {...} and XZ_5.2.2 {...} will be
+# ignored (@XZ_5.1.2alpha or @XZ_5.2.2 won't be added at all when
+# the #define HAVE_SYMBOL_VERSIONS_LINUX isn't used).
+IN_SYNC=
+if ! sed '109,123d' liblzma_linux.map \
+		| cmp -s - liblzma_generic.map; then
+	IN_SYNC=no
+fi
 
 # Print error messages if needed.
-if test -n "$SYMS$NAMES$DUPS"; then
+if test -n "$SYMS$NAMES$DUPS$IN_SYNC"; then
 	echo
-	echo 'validate_map.sh found problems from liblzma.map:'
+	echo 'validate_map.sh found problems from liblzma_*.map:'
 	echo
 
 	if test -n "$SYMS"; then
-		echo 'liblzma.map lacks the following symbols:'
+		echo 'liblzma_generic.map lacks the following symbols:'
 		echo "$SYMS"
 		echo
 	fi
@@ -61,6 +153,11 @@ if test -n "$SYMS$NAMES$DUPS"; then
 		echo
 	fi
 
+	if test -n "$IN_SYNC"; then
+		echo "liblzma_generic.map and liblzma_linux.map aren't in sync"
+		echo
+	fi
+
 	STATUS=1
 fi