Adding upstream version 2.39.3.upstream/2.39.3

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

82 files changed, 13049 insertions, 0 deletions

diff --git a/include/Makemodule.am b/include/Makemodule.am
new file mode 100644
index 0000000..068b0f8
--- /dev/null
+++ b/include/Makemodule.am
@@ -0,0 +1,81 @@
+
+dist_noinst_HEADERS += \
+	include/all-io.h \
+	include/audit-arch.h \
+	include/bitops.h \
+	include/blkdev.h \
+	include/buffer.h \
+	include/canonicalize.h \
+	include/carefulputc.h \
+	include/cctype.h \
+	include/c.h \
+	include/caputils.h \
+	include/closestream.h \
+	include/colors.h \
+	include/color-names.h \
+	include/cpuset.h \
+	include/crc32.h \
+	include/crc32c.h \
+	include/crc64.h \
+	include/c_strtod.h \
+	include/debug.h \
+	include/debugobj.h \
+	include/encode.h \
+	include/env.h \
+	include/exec_shell.h \
+	include/exitcodes.h \
+	include/fileeq.h \
+	include/fileutils.h \
+	include/fuzz.h \
+	include/idcache.h \
+	include/ismounted.h \
+	include/iso9660.h \
+	include/jsonwrt.h \
+	include/pwdutils.h \
+	include/linux_version.h \
+	include/list.h \
+	include/logindefs.h \
+	include/loopdev.h \
+	include/mangle.h \
+	include/match.h \
+	include/mbsalign.h \
+	include/mbsedit.h \
+	include/md5.h \
+	include/minix.h \
+	include/monotonic.h \
+	include/mount-api-utils.h \
+	include/namespace.h \
+	include/nls.h \
+	include/optutils.h \
+	include/pager.h \
+	include/partx.h \
+	include/path.h \
+	include/pathnames.h \
+	include/pidfd-utils.h \
+	include/plymouth-ctrl.h \
+	include/procfs.h \
+	include/pt-bsd.h \
+	include/pt-mbr.h \
+	include/pt-mbr-partnames.h \
+	include/pt-gpt-partnames.h \
+	include/pt-sgi.h \
+	include/pt-sun.h \
+	include/randutils.h \
+	include/rpmatch.h \
+	include/sha1.h \
+	include/sha256.h \
+	include/shells.h \
+	include/signames.h \
+	include/selinux-utils.h \
+	include/statfs_magic.h \
+	include/strutils.h \
+	include/strv.h \
+	include/swapheader.h \
+	include/swapprober.h \
+	include/sysfs.h \
+	include/timer.h \
+	include/timeutils.h \
+	include/ttyutils.h \
+	include/widechar.h \
+	include/xalloc.h \
+	include/xxhash.h
diff --git a/include/all-io.h b/include/all-io.h
new file mode 100644
index 0000000..168965a
--- /dev/null
+++ b/include/all-io.h
@@ -0,0 +1,144 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ *
+ * Written by Karel Zak <kzak@redhat.com>
+ *            Petr Uzel <petr.uzel@suse.cz>
+ */
+
+#ifndef UTIL_LINUX_ALL_IO_H
+#define UTIL_LINUX_ALL_IO_H
+
+#include <string.h>
+#include <unistd.h>
+#include <errno.h>
+#include <sys/types.h>
+#ifdef HAVE_SYS_SENDFILE_H
+# include <sys/sendfile.h>
+#endif
+
+#include "c.h"
+
+static inline int write_all(int fd, const void *buf, size_t count)
+{
+	while (count) {
+		ssize_t tmp;
+
+		errno = 0;
+		tmp = write(fd, buf, count);
+		if (tmp > 0) {
+			count -= tmp;
+			if (count)
+				buf = (const void *) ((const char *) buf + tmp);
+		} else if (errno != EINTR && errno != EAGAIN)
+			return -1;
+		if (errno == EAGAIN)	/* Try later, *sigh* */
+			xusleep(250000);
+	}
+	return 0;
+}
+
+static inline int fwrite_all(const void *ptr, size_t size,
+			     size_t nmemb, FILE *stream)
+{
+	while (nmemb) {
+		size_t tmp;
+
+		errno = 0;
+		tmp = fwrite(ptr, size, nmemb, stream);
+		if (tmp > 0) {
+			nmemb -= tmp;
+			if (nmemb)
+				ptr = (const void *) ((const char *) ptr + (tmp * size));
+		} else if (errno != EINTR && errno != EAGAIN)
+			return -1;
+		if (errno == EAGAIN)	/* Try later, *sigh* */
+			xusleep(250000);
+	}
+	return 0;
+}
+
+static inline ssize_t read_all(int fd, char *buf, size_t count)
+{
+	ssize_t ret;
+	ssize_t c = 0;
+	int tries = 0;
+
+	memset(buf, 0, count);
+	while (count > 0) {
+		ret = read(fd, buf, count);
+		if (ret < 0) {
+			if ((errno == EAGAIN || errno == EINTR) && (tries++ < 5)) {
+				xusleep(250000);
+				continue;
+			}
+			return c ? c : -1;
+		}
+		if (ret == 0)
+			return c;
+		tries = 0;
+		count -= ret;
+		buf += ret;
+		c += ret;
+	}
+	return c;
+}
+
+static inline ssize_t read_all_alloc(int fd, char **buf)
+{
+	size_t size = 1024, c = 0;
+	ssize_t ret;
+
+	*buf = malloc(size);
+	if (!*buf)
+		return -1;
+
+	while (1) {
+		ret = read_all(fd, *buf + c, size - c);
+		if (ret < 0) {
+			free(*buf);
+			*buf = NULL;
+			return -1;
+		}
+
+		if (ret == 0)
+			return c;
+
+		c += ret;
+		if (c == size) {
+			size *= 2;
+			*buf = realloc(*buf, size);
+			if (!*buf)
+				return -1;
+		}
+	}
+}
+
+static inline ssize_t sendfile_all(int out, int in, off_t *off, size_t count)
+{
+#if defined(HAVE_SENDFILE) && defined(__linux__)
+	ssize_t ret;
+	ssize_t c = 0;
+	int tries = 0;
+	while (count) {
+		ret = sendfile(out, in, off, count);
+		if (ret < 0) {
+			if ((errno == EAGAIN || errno == EINTR) && (tries++ < 5)) {
+				xusleep(250000);
+				continue;
+			}
+			return c ? c : -1;
+		}
+		if (ret == 0)
+			return c;
+		tries = 0;
+		count -= ret;
+		c += ret;
+	}
+	return c;
+#else
+	errno = ENOSYS;
+	return -1;
+#endif
+}
+#endif /* UTIL_LINUX_ALL_IO_H */
diff --git a/include/audit-arch.h b/include/audit-arch.h
new file mode 100644
index 0000000..db14560
--- /dev/null
+++ b/include/audit-arch.h
@@ -0,0 +1,66 @@
+/*
+ * SPDX-License-Identifier: LGPL-2.1-or-later
+ *
+ * Copyright (C) 2023 Thomas Weißschuh <thomas@t-8ch.de>
+ *
+ * This file may be distributed under the terms of the
+ * GNU Lesser General Public License.
+ */
+#ifndef UTIL_LINUX_AUDIT_ARCH_H
+#define UTIL_LINUX_AUDIT_ARCH_H
+
+#if __x86_64__
+#    define SECCOMP_ARCH_NATIVE AUDIT_ARCH_X86_64
+#elif __i386__
+#    define SECCOMP_ARCH_NATIVE AUDIT_ARCH_I386
+#elif __arm__
+#    define SECCOMP_ARCH_NATIVE AUDIT_ARCH_ARM
+#elif __aarch64__
+#    define SECCOMP_ARCH_NATIVE AUDIT_ARCH_AARCH64
+#elif __riscv
+#    if __riscv_xlen == 32
+#        define SECCOMP_ARCH_NATIVE AUDIT_ARCH_RISCV32
+#    elif __riscv_xlen == 64
+#        define SECCOMP_ARCH_NATIVE AUDIT_ARCH_RISCV64
+#    endif
+#elif __s390x__
+# 	 define SECCOMP_ARCH_NATIVE AUDIT_ARCH_S390X
+#elif __s390__
+# 	 define SECCOMP_ARCH_NATIVE AUDIT_ARCH_S390
+#elif __PPC64__
+#    if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
+# 	 define SECCOMP_ARCH_NATIVE AUDIT_ARCH_PPC64
+#    else
+# 	 define SECCOMP_ARCH_NATIVE AUDIT_ARCH_PPC64LE
+#    endif
+#elif __powerpc__
+#    define SECCOMP_ARCH_NATIVE AUDIT_ARCH_PPC
+#elif __mips__
+#    if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
+# 	 define SECCOMP_ARCH_NATIVE AUDIT_ARCH_MIPS
+#    else
+# 	 define SECCOMP_ARCH_NATIVE AUDIT_ARCH_MIPSEL
+#    endif
+#elif __arc__
+#    if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
+# 	 define SECCOMP_ARCH_NATIVE AUDIT_ARCH_ARCV2BE
+#    else
+# 	 define SECCOMP_ARCH_NATIVE AUDIT_ARCH_ARCV2
+#    endif
+#elif __sparc__
+#    if __SIZEOF_POINTER__ == 4
+# 	 define SECCOMP_ARCH_NATIVE AUDIT_ARCH_SPARC
+#    else
+# 	 define SECCOMP_ARCH_NATIVE AUDIT_ARCH_SPARC64
+#    endif
+#elif __loongarch__
+#    if __SIZEOF_POINTER__ == 4
+# 	 define SECCOMP_ARCH_NATIVE AUDIT_ARCH_LOONGARCH32
+#    else
+# 	 define SECCOMP_ARCH_NATIVE AUDIT_ARCH_LOONGARCH64
+#    endif
+#else
+#    error Unknown target architecture
+#endif
+
+#endif /* UTIL_LINUX_AUDIT_ARCH_H */
diff --git a/include/bitops.h b/include/bitops.h
new file mode 100644
index 0000000..16ba2ab
--- /dev/null
+++ b/include/bitops.h
@@ -0,0 +1,157 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ *
+ * Written by Karel Zak <kzak@redhat.com>
+ */
+#ifndef BITOPS_H
+#define BITOPS_H
+
+#include <stdint.h>
+#include <sys/param.h>
+
+#if defined(HAVE_BYTESWAP_H)
+# include <byteswap.h>
+#endif
+
+#if defined(HAVE_ENDIAN_H)
+#  include <endian.h>
+#elif defined(HAVE_SYS_ENDIAN_H)	/* BSDs have them here */
+#  include <sys/endian.h>
+#endif
+
+#if !(defined(HAVE_BYTESWAP_H) && defined(HAVE_ENDIAN_H))
+/*
+ * When both byteswap.h and endian.h are preseent, the proper macros are defined
+ * as those files are glibc compatible.  Otherwise, compensate for the slightly
+ * different interfaces between the different BSDs.
+ */
+#if defined(__OpenBSD__)
+# include <sys/types.h>
+# define be16toh(x) betoh16(x)
+# define be32toh(x) betoh32(x)
+# define be64toh(x) betoh64(x)
+#elif defined(__NetBSD__) || defined(__FreeBSD__) || defined(__DragonFly__)
+# define bswap_16(x) bswap16(x)
+# define bswap_32(x) bswap32(x)
+# define bswap_64(x) bswap64(x)
+#elif defined(__APPLE__)
+# include <libkern/OSByteOrder.h>
+# define htobe16(x) OSSwapHostToBigInt16(x)
+# define htole16(x) OSSwapHostToLittleInt16(x)
+# define be16toh(x) OSSwapBigToHostInt16(x)
+# define le16toh(x) OSSwapLittleToHostInt16(x)
+# define htobe32(x) OSSwapHostToBigInt32(x)
+# define htole32(x) OSSwapHostToLittleInt32(x)
+# define be32toh(x) OSSwapBigToHostInt32(x)
+# define le32toh(x) OSSwapLittleToHostInt32(x)
+# define htobe64(x) OSSwapHostToBigInt64(x)
+# define htole64(x) OSSwapHostToLittleInt64(x)
+# define be64toh(x) OSSwapBigToHostInt64(x)
+# define le64toh(x) OSSwapLittleToHostInt64(x)
+# define bswap_16(x) OSSwapInt16(x)
+# define bswap_32(x) OSSwapInt32(x)
+# define bswap_64(x) OSSwapInt64(x)
+#endif
+#endif
+
+/*
+ * Fallbacks
+ * casts are necessary for constants, because we never know how for sure
+ * how U/UL/ULL map to __u16, __u32, __u64. At least not in a portable way.
+ */
+#ifndef bswap_16
+# define bswap_16(x) ((uint16_t)( \
+	(((uint16_t)(x) & 0x00FF) << 8) | \
+	(((uint16_t)(x) & 0xFF00) >> 8)))
+#endif
+
+#ifndef bswap_32
+# define bswap_32(x) ((uint32_t)( \
+	(((uint32_t)(x) & 0x000000FF) << 24) | \
+	(((uint32_t)(x) & 0x0000FF00) << 8)  | \
+	(((uint32_t)(x) & 0x00FF0000) >> 8)  | \
+	(((uint32_t)(x) & 0xFF000000) >> 24)))
+#endif
+
+#ifndef bswap_64
+# define bswap_64(x) ((uint64_t)( \
+	(((uint64_t)(x) & 0x00000000000000FFULL) << 56) | \
+	(((uint64_t)(x) & 0x000000000000FF00ULL) << 40) | \
+	(((uint64_t)(x) & 0x0000000000FF0000ULL) << 24) | \
+	(((uint64_t)(x) & 0x00000000FF000000ULL) << 8)  | \
+	(((uint64_t)(x) & 0x000000FF00000000ULL) >> 8)  | \
+	(((uint64_t)(x) & 0x0000FF0000000000ULL) >> 24) | \
+	(((uint64_t)(x) & 0x00FF000000000000ULL) >> 40) | \
+	(((uint64_t)(x) & 0xFF00000000000000ULL) >> 56)))
+#endif
+
+#ifndef htobe16
+# if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
+#  define htobe16(x) bswap_16 (x)
+#  define htole16(x) (x)
+#  define be16toh(x) bswap_16 (x)
+#  define le16toh(x) (x)
+#  define htobe32(x) bswap_32 (x)
+#  define htole32(x) (x)
+#  define be32toh(x) bswap_32 (x)
+#  define le32toh(x) (x)
+#  define htobe64(x) bswap_64 (x)
+#  define htole64(x) (x)
+#  define be64toh(x) bswap_64 (x)
+#  define le64toh(x) (x)
+# else
+#  define htobe16(x) (x)
+#  define htole16(x) bswap_16 (x)
+#  define be16toh(x) (x)
+#  define le16toh(x) bswap_16 (x)
+#  define htobe32(x) (x)
+#  define htole32(x) bswap_32 (x)
+#  define be32toh(x) (x)
+#  define le32toh(x) bswap_32 (x)
+#  define htobe64(x) (x)
+#  define htole64(x) bswap_64 (x)
+#  define be64toh(x) (x)
+#  define le64toh(x) bswap_64 (x)
+# endif
+#endif
+
+/*
+ * Byte swab macros (based on linux/byteorder/swab.h)
+ */
+#define swab16(x) bswap_16(x)
+#define swab32(x) bswap_32(x)
+#define swab64(x) bswap_64(x)
+
+#define cpu_to_le16(x) ((uint16_t) htole16(x))
+#define cpu_to_le32(x) ((uint32_t) htole32(x))
+#define cpu_to_le64(x) ((uint64_t) htole64(x))
+
+#define cpu_to_be16(x) ((uint16_t) htobe16(x))
+#define cpu_to_be32(x) ((uint32_t) htobe32(x))
+#define cpu_to_be64(x) ((uint64_t) htobe64(x))
+
+#define le16_to_cpu(x) ((uint16_t) le16toh(x))
+#define le32_to_cpu(x) ((uint32_t) le32toh(x))
+#define le64_to_cpu(x) ((uint64_t) le64toh(x))
+
+#define be16_to_cpu(x) ((uint16_t) be16toh(x))
+#define be32_to_cpu(x) ((uint32_t) be32toh(x))
+#define be64_to_cpu(x) ((uint64_t) be64toh(x))
+
+/*
+ * Bit map related macros. Usually provided by libc.
+ */
+#ifndef NBBY
+# define NBBY            CHAR_BIT
+#endif
+
+#ifndef setbit
+# define setbit(a,i)	((a)[(i)/NBBY] |= 1<<((i)%NBBY))
+# define clrbit(a,i)	((a)[(i)/NBBY] &= ~(1<<((i)%NBBY)))
+# define isset(a,i)	((a)[(i)/NBBY] & (1<<((i)%NBBY)))
+# define isclr(a,i)	(((a)[(i)/NBBY] & (1<<((i)%NBBY))) == 0)
+#endif
+
+#endif /* BITOPS_H */
+
diff --git a/include/blkdev.h b/include/blkdev.h
new file mode 100644
index 0000000..ffecdcb
--- /dev/null
+++ b/include/blkdev.h
@@ -0,0 +1,165 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ *
+ * Written by Karel Zak <kzak@redhat.com>
+ */
+#ifndef BLKDEV_H
+#define BLKDEV_H
+
+#include <sys/types.h>
+#include <sys/ioctl.h>
+#ifdef HAVE_SYS_IOCCOM_H
+# include <sys/ioccom.h> /* for _IO macro on e.g. Solaris */
+#endif
+#include <fcntl.h>
+#include <unistd.h>
+#include <sys/stat.h>
+#include <stdint.h>
+
+#ifdef HAVE_SYS_MKDEV_H
+# include <sys/mkdev.h>		/* major and minor on Solaris */
+#endif
+
+#define DEFAULT_SECTOR_SIZE       512
+
+#ifdef __linux__
+/* very basic ioctls, should be available everywhere */
+# ifndef BLKROSET
+#  define BLKROSET   _IO(0x12,93)	/* set device read-only (0 = read-write) */
+#  define BLKROGET   _IO(0x12,94)	/* get read-only status (0 = read_write) */
+#  define BLKRRPART  _IO(0x12,95)	/* re-read partition table */
+#  define BLKGETSIZE _IO(0x12,96)	/* return device size /512 (long *arg) */
+#  define BLKFLSBUF  _IO(0x12,97)	/* flush buffer cache */
+#  define BLKRASET   _IO(0x12,98)	/* set read ahead for block device */
+#  define BLKRAGET   _IO(0x12,99)	/* get current read ahead setting */
+#  define BLKFRASET  _IO(0x12,100)	/* set filesystem (mm/filemap.c) read-ahead */
+#  define BLKFRAGET  _IO(0x12,101)	/* get filesystem (mm/filemap.c) read-ahead */
+#  define BLKSECTSET _IO(0x12,102)	/* set max sectors per request (ll_rw_blk.c) */
+#  define BLKSECTGET _IO(0x12,103)	/* get max sectors per request (ll_rw_blk.c) */
+#  define BLKSSZGET  _IO(0x12,104)	/* get block device sector size */
+
+/* ioctls introduced in 2.2.16, removed in 2.5.58 */
+#  define BLKELVGET  _IOR(0x12,106,size_t) /* elevator get */
+#  define BLKELVSET  _IOW(0x12,107,size_t) /* elevator set */
+
+#  define BLKBSZGET  _IOR(0x12,112,size_t)
+#  define BLKBSZSET  _IOW(0x12,113,size_t)
+# endif /* !BLKROSET */
+
+# ifndef BLKGETSIZE64
+#  define BLKGETSIZE64 _IOR(0x12,114,size_t) /* return device size in bytes (u64 *arg) */
+# endif
+
+/* block device topology ioctls, introduced in 2.6.32 (commit ac481c20) */
+# ifndef BLKIOMIN
+#  define BLKIOMIN   _IO(0x12,120)
+#  define BLKIOOPT   _IO(0x12,121)
+#  define BLKALIGNOFF _IO(0x12,122)
+#  define BLKPBSZGET _IO(0x12,123)
+# endif
+
+/* discard zeroes support, introduced in 2.6.33 (commit 98262f27) */
+# ifndef BLKDISCARDZEROES
+#  define BLKDISCARDZEROES _IO(0x12,124)
+# endif
+
+/* disk sequence number, introduced in 5.15 (commit 7957d93b) */
+# ifndef BLKGETDISKSEQ
+#  define BLKGETDISKSEQ _IOR(0x12, 128, uint64_t)
+# endif
+
+/* filesystem freeze, introduced in 2.6.29 (commit fcccf502) */
+# ifndef FIFREEZE
+#  define FIFREEZE   _IOWR('X', 119, int)    /* Freeze */
+#  define FITHAW     _IOWR('X', 120, int)    /* Thaw */
+# endif
+
+/* uniform CD-ROM information */
+# ifndef CDROM_GET_CAPABILITY
+#  define CDROM_GET_CAPABILITY 0x5331
+# endif
+
+#endif /* __linux */
+
+
+#ifdef APPLE_DARWIN
+# define BLKGETSIZE DKIOCGETBLOCKCOUNT32
+#endif
+
+#ifndef HDIO_GETGEO
+# ifdef __linux__
+#  define HDIO_GETGEO 0x0301
+# endif
+
+struct hd_geometry {
+	unsigned char heads;
+	unsigned char sectors;
+	unsigned short cylinders;	/* truncated */
+	unsigned long start;
+};
+#endif /* HDIO_GETGEO */
+
+
+/* are we working with block device? */
+int is_blkdev(int fd);
+
+/* open block device or file */
+int open_blkdev_or_file(const struct stat *st, const char *name, const int oflag);
+
+/* Determine size in bytes */
+off_t blkdev_find_size (int fd);
+
+/* get size in bytes */
+int blkdev_get_size(int fd, unsigned long long *bytes);
+
+/* get 512-byte sector count */
+int blkdev_get_sectors(int fd, unsigned long long *sectors);
+
+/* get hardware sector size */
+int blkdev_get_sector_size(int fd, int *sector_size);
+
+/* specifies whether or not the device is misaligned */
+int blkdev_is_misaligned(int fd);
+
+/* get physical block device size */
+int blkdev_get_physector_size(int fd, int *sector_size);
+
+/* is the device cdrom capable? */
+int blkdev_is_cdrom(int fd);
+
+/* get device's geometry - legacy */
+int blkdev_get_geometry(int fd, unsigned int *h, unsigned int *s);
+
+/* SCSI device types.  Copied almost as-is from kernel header.
+ * http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/include/scsi/scsi.h */
+#define SCSI_TYPE_DISK			0x00
+#define SCSI_TYPE_TAPE			0x01
+#define SCSI_TYPE_PRINTER		0x02
+#define SCSI_TYPE_PROCESSOR		0x03	/* HP scanners use this */
+#define SCSI_TYPE_WORM			0x04	/* Treated as ROM by our system */
+#define SCSI_TYPE_ROM			0x05
+#define SCSI_TYPE_SCANNER		0x06
+#define SCSI_TYPE_MOD			0x07	/* Magneto-optical disk - treated as SCSI_TYPE_DISK */
+#define SCSI_TYPE_MEDIUM_CHANGER	0x08
+#define SCSI_TYPE_COMM			0x09	/* Communications device */
+#define SCSI_TYPE_RAID			0x0c
+#define SCSI_TYPE_ENCLOSURE		0x0d	/* Enclosure Services Device */
+#define SCSI_TYPE_RBC			0x0e
+#define SCSI_TYPE_OSD			0x11
+#define SCSI_TYPE_NO_LUN		0x7f
+
+/* convert scsi type code to name */
+const char *blkdev_scsi_type_to_name(int type);
+
+int blkdev_lock(int fd, const char *devname, const char *lockmode);
+#ifdef HAVE_LINUX_BLKZONED_H
+struct blk_zone_report *blkdev_get_zonereport(int fd, uint64_t sector, uint32_t nzones);
+#else
+static inline struct blk_zone_report *blkdev_get_zonereport(int fd, uint64_t sector, uint32_t nzones)
+{
+	return NULL;
+}
+#endif
+
+#endif /* BLKDEV_H */
diff --git a/include/buffer.h b/include/buffer.h
new file mode 100644
index 0000000..b122da4
--- /dev/null
+++ b/include/buffer.h
@@ -0,0 +1,48 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_BUFFER
+#define UTIL_LINUX_BUFFER
+
+#include "c.h"
+
+struct ul_buffer {
+	char *begin;		/* begin of the data */
+	char *end;		/* current end of data */
+
+	size_t sz;		/* allocated space for data */
+	size_t chunksize;
+
+	char *encoded;		/* encoded data (from mbs_safe_encode_to_buffer)) */
+	size_t encoded_sz;	/* space allocated for encoded data */
+
+	char **ptrs;		/* saved pointers */
+	size_t nptrs;		/* number of saved pointers */
+};
+
+#define UL_INIT_BUFFER { .begin = NULL }
+
+void ul_buffer_reset_data(struct ul_buffer *buf);
+void ul_buffer_free_data(struct ul_buffer *buf);
+int ul_buffer_is_empty(struct ul_buffer *buf);
+void ul_buffer_set_chunksize(struct ul_buffer *buf, size_t sz);
+void ul_buffer_refer_string(struct ul_buffer *buf, char *str);
+int ul_buffer_alloc_data(struct ul_buffer *buf, size_t sz);
+int ul_buffer_append_data(struct ul_buffer *buf, const char *data, size_t sz);
+int ul_buffer_append_string(struct ul_buffer *buf, const char *str);
+int ul_buffer_append_ntimes(struct ul_buffer *buf, size_t n, const char *str);
+int ul_buffer_set_data(struct ul_buffer *buf, const char *data, size_t sz);
+
+char *ul_buffer_get_data(struct ul_buffer *buf,  size_t *sz, size_t *width);
+char *ul_buffer_get_safe_data(struct ul_buffer *buf, size_t *sz, size_t *width, const char *safechars);
+
+size_t ul_buffer_get_bufsiz(struct ul_buffer *buf);
+
+int ul_buffer_save_pointer(struct ul_buffer *buf, unsigned short ptr_idx);
+char *ul_buffer_get_pointer(struct ul_buffer *buf, unsigned short  ptr_idx);
+size_t ul_buffer_get_pointer_length(struct ul_buffer *buf, unsigned short ptr_idx);
+size_t ul_buffer_get_safe_pointer_width(struct ul_buffer *buf, unsigned short ptr_idx);
+
+
+#endif /* UTIL_LINUX_BUFFER */
diff --git a/include/c.h b/include/c.h
new file mode 100644
index 0000000..2b70e8d
--- /dev/null
+++ b/include/c.h
@@ -0,0 +1,567 @@
+/*
+ * Fundamental C definitions.
+ *
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_C_H
+#define UTIL_LINUX_C_H
+
+#include <limits.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <unistd.h>
+#include <stdarg.h>
+#include <stdlib.h>
+#include <string.h>
+#include <errno.h>
+#include <sys/types.h>
+#include <grp.h>
+
+#include <assert.h>
+
+#ifdef HAVE_ERR_H
+# include <err.h>
+#endif
+
+#ifdef HAVE_SYS_SYSMACROS_H
+# include <sys/sysmacros.h>     /* for major, minor */
+#endif
+
+#ifndef LOGIN_NAME_MAX
+# define LOGIN_NAME_MAX 256
+#endif
+
+#ifndef NAME_MAX
+# define NAME_MAX PATH_MAX
+#endif
+
+/*
+ * __GNUC_PREREQ is deprecated in favour of __has_attribute() and
+ * __has_feature(). The __has macros are supported by clang and gcc>=5.
+ */
+#ifndef __GNUC_PREREQ
+# if defined __GNUC__ && defined __GNUC_MINOR__
+#  define __GNUC_PREREQ(maj, min) \
+	((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
+# else
+#  define __GNUC_PREREQ(maj, min) 0
+# endif
+#endif
+
+#ifdef __GNUC__
+
+/* &a[0] degrades to a pointer: a different type from an array */
+# define __must_be_array(a) \
+	UL_BUILD_BUG_ON_ZERO(__builtin_types_compatible_p(__typeof__(a), __typeof__(&a[0])))
+
+# define ignore_result(x) __extension__ ({ \
+	__typeof__(x) __dummy __attribute__((__unused__)) = (x); (void) __dummy; \
+})
+
+#else /* !__GNUC__ */
+# define __must_be_array(a)	0
+# define __attribute__(_arg_)
+# define ignore_result(x) ((void) (x))
+#endif /* !__GNUC__ */
+
+
+/* "restrict" keyword fallback */
+#if __STDC__ != 1
+# define restrict __restrict /* use implementation __ format */
+#else
+# ifndef __STDC_VERSION__
+#  define restrict __restrict /* use implementation __ format */
+# else
+#  if __STDC_VERSION__ < 199901L
+#   define restrict __restrict /* use implementation __ format */
+#  endif
+# endif
+#endif
+
+
+/*
+ * It evaluates to 1 if the attribute/feature is supported by the current
+ * compilation target. Fallback for old compilers.
+ */
+#ifndef __has_attribute
+  #define __has_attribute(x) 0
+#endif
+
+#ifndef __has_feature
+  #define __has_feature(x) 0
+#endif
+
+/*
+ * Function attributes
+ */
+#ifndef __ul_alloc_size
+# if (__has_attribute(alloc_size) && __has_attribute(warn_unused_result)) || __GNUC_PREREQ (4, 3)
+#  define __ul_alloc_size(s) __attribute__((alloc_size(s), warn_unused_result))
+# else
+#  define __ul_alloc_size(s)
+# endif
+#endif
+
+#ifndef __ul_calloc_size
+# if (__has_attribute(alloc_size) && __has_attribute(warn_unused_result)) || __GNUC_PREREQ (4, 3)
+#  define __ul_calloc_size(n, s) __attribute__((alloc_size(n, s), warn_unused_result))
+# else
+#  define __ul_calloc_size(n, s)
+# endif
+#endif
+
+#if __has_attribute(returns_nonnull) || __GNUC_PREREQ (4, 9)
+# define __ul_returns_nonnull __attribute__((returns_nonnull))
+#else
+# define __ul_returns_nonnull
+#endif
+
+/*
+ * Force a compilation error if condition is true, but also produce a
+ * result (of value 0 and type size_t), so the expression can be used
+ * e.g. in a structure initializer (or wherever else comma expressions
+ * aren't permitted).
+ */
+#define UL_BUILD_BUG_ON_ZERO(e) __extension__ (sizeof(struct { int:-!!(e); }))
+#define BUILD_BUG_ON_NULL(e) ((void *)sizeof(struct { int:-!!(e); }))
+
+#ifndef ARRAY_SIZE
+# define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]) + __must_be_array(arr))
+#endif
+
+#ifndef PATH_MAX
+# define PATH_MAX 4096
+#endif
+
+#ifndef TRUE
+# define TRUE 1
+#endif
+
+#ifndef FALSE
+# define FALSE 0
+#endif
+
+#ifndef min
+# define min(x, y) __extension__ ({		\
+	__typeof__(x) _min1 = (x);		\
+	__typeof__(y) _min2 = (y);		\
+	(void) (&_min1 == &_min2);		\
+	_min1 < _min2 ? _min1 : _min2; })
+#endif
+
+#ifndef max
+# define max(x, y) __extension__ ({		\
+	__typeof__(x) _max1 = (x);		\
+	__typeof__(y) _max2 = (y);		\
+	(void) (&_max1 == &_max2);		\
+	_max1 > _max2 ? _max1 : _max2; })
+#endif
+
+#ifndef abs_diff
+# define abs_diff(x, y) __extension__ ({        \
+	__typeof__(x) _a = (x);			\
+	__typeof__(y) _b = (y);			\
+	(void) (&_a == &_b);			\
+	_a > _b ? _a - _b : _b - _a; })
+#endif
+
+#ifndef cmp_numbers
+# define cmp_numbers(x, y) __extension__ ({	\
+	__typeof__(x) _a = (x);			\
+	__typeof__(y) _b = (y);			\
+	(void) (&_a == &_b);			\
+	_a == _b ? 0 : _a > _b ? 1 : -1; })
+#endif
+
+
+#ifndef cmp_timespec
+# define cmp_timespec(a, b, CMP)		\
+	(((a)->tv_sec == (b)->tv_sec)		\
+	? ((a)->tv_nsec CMP (b)->tv_nsec)	\
+	: ((a)->tv_sec CMP (b)->tv_sec))
+#endif
+
+
+#ifndef cmp_stat_mtime
+# ifdef HAVE_STRUCT_STAT_ST_MTIM_TV_NSEC
+#  define cmp_stat_mtime(_a, _b, CMP)	cmp_timespec(&(_a)->st_mtim, &(_b)->st_mtim, CMP)
+# else
+#  define cmp_stat_mtime(_a, _b, CMP)	((_a)->st_mtime CMP (_b)->st_mtime)
+# endif
+#endif
+
+
+#ifndef offsetof
+#define offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER)
+#endif
+
+#ifndef sizeof_member
+#define sizeof_member(TYPE, MEMBER) sizeof(((TYPE *)0)->MEMBER)
+#endif
+
+/*
+ * container_of - cast a member of a structure out to the containing structure
+ * @ptr:	the pointer to the member.
+ * @type:	the type of the container struct this is embedded in.
+ * @member:	the name of the member within the struct.
+ */
+#ifndef container_of
+#define container_of(ptr, type, member) __extension__ ({	\
+	const __typeof__( ((type *)0)->member ) *__mptr = (ptr); \
+	(type *)( (char *)__mptr - offsetof(type,member) );})
+#endif
+
+#ifndef HAVE_PROGRAM_INVOCATION_SHORT_NAME
+# ifdef HAVE___PROGNAME
+extern char *__progname;
+#  define program_invocation_short_name __progname
+# else
+#  ifdef HAVE_GETEXECNAME
+#   define program_invocation_short_name \
+		prog_inv_sh_nm_from_file(getexecname(), 0)
+#  else
+#   define program_invocation_short_name \
+		prog_inv_sh_nm_from_file(__FILE__, 1)
+#  endif
+static char prog_inv_sh_nm_buf[256];
+static inline char *
+prog_inv_sh_nm_from_file(char *f, char stripext)
+{
+	char *t;
+
+	if ((t = strrchr(f, '/')) != NULL)
+		t++;
+	else
+		t = f;
+
+	strncpy(prog_inv_sh_nm_buf, t, sizeof(prog_inv_sh_nm_buf) - 1);
+	prog_inv_sh_nm_buf[sizeof(prog_inv_sh_nm_buf) - 1] = '\0';
+
+	if (stripext && (t = strrchr(prog_inv_sh_nm_buf, '.')) != NULL)
+		*t = '\0';
+
+	return prog_inv_sh_nm_buf;
+}
+# endif
+#endif
+
+
+#ifndef HAVE_ERR_H
+static inline void __attribute__ ((__format__ (__printf__, 4, 5)))
+errmsg(char doexit, int excode, char adderr, const char *fmt, ...)
+{
+	fprintf(stderr, "%s: ", program_invocation_short_name);
+	if (fmt != NULL) {
+		va_list argp;
+		va_start(argp, fmt);
+		vfprintf(stderr, fmt, argp);
+		va_end(argp);
+		if (adderr)
+			fprintf(stderr, ": ");
+	}
+	if (adderr)
+		fprintf(stderr, "%m");
+	fprintf(stderr, "\n");
+	if (doexit)
+		exit(excode);
+}
+
+#ifndef HAVE_ERR
+# define err(E, FMT...) errmsg(1, E, 1, FMT)
+#endif
+
+#ifndef HAVE_ERRX
+# define errx(E, FMT...) errmsg(1, E, 0, FMT)
+#endif
+
+#ifndef HAVE_WARN
+# define warn(FMT...) errmsg(0, 0, 1, FMT)
+#endif
+
+#ifndef HAVE_WARNX
+# define warnx(FMT...) errmsg(0, 0, 0, FMT)
+#endif
+#endif /* !HAVE_ERR_H */
+
+
+static inline
+__attribute__((__noreturn__))
+void __err_oom(const char *file, unsigned int line)
+{
+	err(EXIT_FAILURE, "%s: %u: cannot allocate memory", file, line);
+}
+#define err_oom()	__err_oom(__FILE__, __LINE__)
+
+
+/* Don't use inline function to avoid '#include "nls.h"' in c.h
+ */
+#define errtryhelp(eval) __extension__ ({ \
+	fprintf(stderr, _("Try '%s --help' for more information.\n"), \
+			program_invocation_short_name); \
+	exit(eval); \
+})
+
+/* After failed execvp() */
+#define EX_EXEC_FAILED		126	/* Program located, but not usable. */
+#define EX_EXEC_ENOENT		127	/* Could not find program to exec.  */
+#define errexec(name)	err(errno == ENOENT ? EX_EXEC_ENOENT : EX_EXEC_FAILED, \
+			_("failed to execute %s"), name)
+
+static inline __attribute__((const)) int is_power_of_2(unsigned long num)
+{
+	return (num != 0 && ((num & (num - 1)) == 0));
+}
+
+#ifndef HAVE_LOFF_T
+typedef int64_t loff_t;
+#endif
+
+#if !defined(HAVE_DIRFD) \
+    && (!defined(HAVE_DECL_DIRFD) || HAVE_DECL_DIRFD == 0) \
+    && defined(HAVE_DIR_DD_FD)
+#include <dirent.h>
+static inline int dirfd(DIR *d)
+{
+	return d->dd_fd;
+}
+#endif
+
+/*
+ * Fallback defines for old versions of glibc
+ */
+#include <fcntl.h>
+
+#ifdef O_CLOEXEC
+#define UL_CLOEXECSTR	"e"
+#else
+#define UL_CLOEXECSTR	""
+#endif
+
+#ifndef O_CLOEXEC
+#define O_CLOEXEC 0
+#endif
+
+#ifdef __FreeBSD_kernel__
+#ifndef F_DUPFD_CLOEXEC
+#define F_DUPFD_CLOEXEC	17	/* Like F_DUPFD, but FD_CLOEXEC is set */
+#endif
+#endif
+
+
+#ifndef AI_ADDRCONFIG
+#define AI_ADDRCONFIG 0x0020
+#endif
+
+#ifndef IUTF8
+#define IUTF8 0040000
+#endif
+
+/*
+ * MAXHOSTNAMELEN replacement
+ */
+static inline size_t get_hostname_max(void)
+{
+	long len = sysconf(_SC_HOST_NAME_MAX);
+
+	if (0 < len)
+		return len;
+
+#ifdef MAXHOSTNAMELEN
+	return MAXHOSTNAMELEN;
+#elif HOST_NAME_MAX
+	return HOST_NAME_MAX;
+#endif
+	return 64;
+}
+
+
+static inline int drop_permissions(void)
+{
+	errno = 0;
+
+	/* drop GID */
+	if (setgid(getgid()) < 0)
+		goto fail;
+
+	/* drop UID */
+	if (setuid(getuid()) < 0)
+		goto fail;
+
+	return 0;
+fail:
+	return errno ? -errno : -1;
+}
+
+/*
+ * The usleep function was marked obsolete in POSIX.1-2001 and was removed
+ * in POSIX.1-2008.  It was replaced with nanosleep() that provides more
+ * advantages (like no interaction with signals and other timer functions).
+ */
+#include <time.h>
+
+static inline int xusleep(useconds_t usec)
+{
+#ifdef HAVE_NANOSLEEP
+	struct timespec waittime = {
+		.tv_sec   =  usec / 1000000L,
+		.tv_nsec  = (usec % 1000000L) * 1000
+	};
+	return nanosleep(&waittime, NULL);
+#elif defined(HAVE_USLEEP)
+	return usleep(usec);
+#else
+# error	"System with usleep() or nanosleep() required!"
+#endif
+}
+
+
+#define ul_err_write(_m) ignore_result( write(STDERR_FILENO, _m, strlen(_m)) )
+
+/*
+ * warn() for signal handlers
+ */
+static inline void ul_sig_warn(const char *mesg)
+{
+	ul_err_write(program_invocation_short_name);
+	ul_err_write(": ");
+	ul_err_write(mesg);
+	ul_err_write("\n");
+}
+
+/*
+ * err() for signal handlers
+ */
+static inline void __attribute__((__noreturn__)) ul_sig_err(int excode, const char *mesg)
+{
+	ul_sig_warn(mesg);
+	_exit(excode);
+}
+
+/*
+ * Constant strings for usage() functions. For more info see
+ * Documentation/{howto-usage-function.txt,boilerplate.c}
+ */
+#define USAGE_HEADER     _("\nUsage:\n")
+#define USAGE_OPTIONS    _("\nOptions:\n")
+#define USAGE_FUNCTIONS  _("\nFunctions:\n")
+#define USAGE_COMMANDS   _("\nCommands:\n")
+#define USAGE_ARGUMENTS   _("\nArguments:\n")
+#define USAGE_COLUMNS    _("\nAvailable output columns:\n")
+#define USAGE_SEPARATOR    "\n"
+
+#define USAGE_OPTSTR_HELP     _("display this help")
+#define USAGE_OPTSTR_VERSION  _("display version")
+
+#define USAGE_HELP_OPTIONS(marg_dsc) \
+		"%-" #marg_dsc "s%s\n" \
+		"%-" #marg_dsc "s%s\n" \
+		, " -h, --help",    USAGE_OPTSTR_HELP \
+		, " -V, --version", USAGE_OPTSTR_VERSION
+
+#define USAGE_ARG_SEPARATOR    "\n"
+#define USAGE_ARG_SIZE(_name) \
+		_(" %s arguments may be followed by the suffixes for\n" \
+		  "   GiB, TiB, PiB, EiB, ZiB, and YiB (the \"iB\" is optional)\n"), _name
+
+#define USAGE_MAN_TAIL(_man)   _("\nFor more details see %s.\n"), _man
+
+#define UTIL_LINUX_VERSION _("%s from %s\n"), program_invocation_short_name, PACKAGE_STRING
+
+#define print_version(eval) __extension__ ({ \
+		printf(UTIL_LINUX_VERSION); \
+		exit(eval); \
+})
+
+static inline void print_features(const char **features, const char *prefix)
+{
+	if (features && *features) {
+		const char **p = features;
+		while (p && *p) {
+			if (prefix && p == features)
+				printf(" (%s ", prefix);
+			else
+				fputs(p == features ? " (" : ", ", stdout);
+			fputs(*p++, stdout);
+		}
+		fputc(')', stdout);
+	}
+}
+
+#define UTIL_LINUX_VERSION_NOBREAK _("%s from %s"), program_invocation_short_name, PACKAGE_STRING
+
+#define print_version_with_features(eval, features) __extension__ ({ \
+		printf(UTIL_LINUX_VERSION_NOBREAK); \
+		print_features(features, _("features:")); \
+		fputc('\n', stdout); \
+		exit(eval); \
+})
+
+/*
+ * seek stuff
+ */
+#ifndef SEEK_DATA
+# define SEEK_DATA	3
+#endif
+#ifndef SEEK_HOLE
+# define SEEK_HOLE	4
+#endif
+
+
+/*
+ * Macros to convert #define'itions to strings, for example
+ * #define XYXXY 42
+ * printf ("%s=%s\n", stringify(XYXXY), stringify_value(XYXXY));
+ */
+#define stringify_value(s) stringify(s)
+#define stringify(s) #s
+
+/* Detect if we're compiled with Address Sanitizer
+ *  - gcc (__SANITIZE_ADDRESS__)
+ *  - clang (__has_feature(address_sanitizer))
+ */
+#if !defined(HAS_FEATURE_ADDRESS_SANITIZER)
+#  ifdef __SANITIZE_ADDRESS__
+#      define HAS_FEATURE_ADDRESS_SANITIZER 1
+#  elif defined(__has_feature)
+#    if __has_feature(address_sanitizer)
+#      define HAS_FEATURE_ADDRESS_SANITIZER 1
+#    endif
+#  endif
+#  if !defined(HAS_FEATURE_ADDRESS_SANITIZER)
+#    define HAS_FEATURE_ADDRESS_SANITIZER 0
+#  endif
+#endif
+
+/*
+ * UL_ASAN_BLACKLIST is a macro to tell AddressSanitizer (a compile-time
+ * instrumentation shipped with Clang and GCC) to not instrument the
+ * annotated function.  Furthermore, it will prevent the compiler from
+ * inlining the function because inlining currently breaks the blacklisting
+ * mechanism of AddressSanitizer.
+ */
+#if __has_feature(address_sanitizer) && __has_attribute(no_sanitize_memory) && __has_attribute(no_sanitize_address)
+# define UL_ASAN_BLACKLIST __attribute__((noinline)) __attribute__((no_sanitize_memory)) __attribute__((no_sanitize_address))
+#else
+# define UL_ASAN_BLACKLIST	/* nothing */
+#endif
+
+/*
+ * Note that sysconf(_SC_GETPW_R_SIZE_MAX) returns *initial* suggested size for
+ * pwd buffer and in some cases it is not large enough. See POSIX and
+ * getpwnam_r man page for more details.
+ */
+#define UL_GETPW_BUFSIZ	(16 * 1024)
+
+/*
+ * Darwin or other BSDs may only have MAP_ANON. To get it on Darwin we must
+ * define _DARWIN_C_SOURCE before including sys/mman.h. We do this in config.h.
+ */
+#if !defined MAP_ANONYMOUS && defined MAP_ANON
+# define MAP_ANONYMOUS  (MAP_ANON)
+#endif
+
+#define SINT_MAX(t) (((t)1 << (sizeof(t) * 8 - 2)) - (t)1 + ((t)1 << (sizeof(t) * 8 - 2)))
+
+#endif /* UTIL_LINUX_C_H */
diff --git a/include/c_strtod.h b/include/c_strtod.h
new file mode 100644
index 0000000..14070cb
--- /dev/null
+++ b/include/c_strtod.h
@@ -0,0 +1,12 @@
+/*
+ * SPDX-License-Identifier: LGPL-2.1-or-later
+ *
+ * This file may be redistributed under the terms of the
+ * GNU Lesser General Public License.
+ **/
+#ifndef UTIL_LINUX_C_STRTOD_H
+#define UTIL_LINUX_C_STRTOD_H
+
+extern double c_strtod(char const *str, char **end);
+
+#endif
diff --git a/include/canonicalize.h b/include/canonicalize.h
new file mode 100644
index 0000000..ff6ef0d
--- /dev/null
+++ b/include/canonicalize.h
@@ -0,0 +1,32 @@
+/*
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU Library Public License as published by
+ * the Free Software Foundation; either version 2, or (at your option)
+ * any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU Library Public License for more details.
+ */
+#ifndef CANONICALIZE_H
+#define CANONICALIZE_H
+
+#include "c.h"	/* for PATH_MAX */
+#include "strutils.h"
+
+extern char *canonicalize_path(const char *path);
+extern char *canonicalize_path_restricted(const char *path);
+extern char *canonicalize_dm_name(const char *ptname);
+extern char *__canonicalize_dm_name(const char *prefix, const char *ptname);
+
+extern char *absolute_path(const char *path);
+
+static inline int is_relative_path(const char *path)
+{
+	if (!path || *path == '/')
+		return 0;
+	return 1;
+}
+
+#endif /* CANONICALIZE_H */
diff --git a/include/caputils.h b/include/caputils.h
new file mode 100644
index 0000000..852903a
--- /dev/null
+++ b/include/caputils.h
@@ -0,0 +1,34 @@
+/*
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the
+ * Free Software Foundation; either version 2, or (at your option) any
+ * later version.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+ */
+
+#ifndef CAPUTILS_H
+#define CAPUTILS_H
+
+#include <linux/capability.h>
+
+#ifndef PR_CAP_AMBIENT
+# define PR_CAP_AMBIENT		47
+#  define PR_CAP_AMBIENT_IS_SET	1
+#  define PR_CAP_AMBIENT_RAISE	2
+#  define PR_CAP_AMBIENT_LOWER	3
+#endif
+
+extern int capset(cap_user_header_t header, cap_user_data_t data);
+extern int capget(cap_user_header_t header, const cap_user_data_t data);
+
+extern int cap_last_cap(void);
+
+#endif /* CAPUTILS_H */
diff --git a/include/carefulputc.h b/include/carefulputc.h
new file mode 100644
index 0000000..3cc6f7f
--- /dev/null
+++ b/include/carefulputc.h
@@ -0,0 +1,126 @@
+#ifndef UTIL_LINUX_CAREFULPUTC_H
+#define UTIL_LINUX_CAREFULPUTC_H
+
+#include <stdio.h>
+#include <string.h>
+#include <ctype.h>
+#ifdef HAVE_WIDECHAR
+#include <wctype.h>
+#include <wchar.h>
+#endif
+#include <stdbool.h>
+
+#include "cctype.h"
+
+/*
+ * A puts() for use in write and wall (that sometimes are sgid tty).
+ * It avoids control and invalid characters.
+ * The locale of the recipient is nominally unknown,
+ * but it's a solid bet that it's compatible with the author's.
+ * Use soft_width=0 to disable wrapping.
+ */
+static inline int fputs_careful(const char * s, FILE *fp, const char ctrl, bool cr_lf, int soft_width)
+{
+	int ret = 0, col = 0;
+
+	for (size_t slen = strlen(s); *s; ++s, --slen) {
+		if (*s == '\t')
+			col += (7 - (col % 8)) - 1;
+		else if (*s == '\r')
+			col = -1;
+		else if (*s == '\a')
+			--col;
+
+		if ((soft_width && col >= soft_width) || *s == '\n') {
+			if (soft_width) {
+				fprintf(fp, "%*s", soft_width - col, "");
+				col = 0;
+			}
+			ret = fputs(cr_lf ? "\r\n" : "\n", fp);
+			if (*s == '\n' || ret < 0)
+				goto wrote;
+		}
+
+		if (isprint(*s) || *s == '\a' || *s == '\t' || *s == '\r') {
+			ret = putc(*s, fp);
+			++col;
+		} else if (!c_isascii(*s)) {
+#ifdef HAVE_WIDECHAR
+			wchar_t w;
+			size_t clen = mbtowc(&w, s, slen);
+			switch(clen) {
+				case (size_t)-2:  // incomplete
+				case (size_t)-1:  // EILSEQ
+					mbtowc(NULL, NULL, 0);
+				nonprint:
+					col += ret = fprintf(fp, "\\%3hho", *s);
+					break;
+				default:
+					if(!iswprint(w))
+						goto nonprint;
+					ret = fwrite(s, 1, clen, fp);
+					if (soft_width)
+						col += wcwidth(w);
+					s += clen - 1;
+					slen -= clen - 1;
+					break;
+			}
+#else
+			col += ret = fprintf(fp, "\\%3hho", *s);
+#endif
+		} else {
+			ret = fputs((char[]){ ctrl, *s ^ 0x40, '\0' }, fp);
+			col += 2;
+		}
+
+	wrote:
+		if (ret < 0)
+			return EOF;
+	}
+	return 0;
+}
+
+static inline void fputs_quoted_case(const char *data, FILE *out, int dir)
+{
+	const char *p;
+
+	fputc('"', out);
+	for (p = data; p && *p; p++) {
+		if ((unsigned char) *p == 0x22 ||		/* " */
+		    (unsigned char) *p == 0x5c ||		/* \ */
+		    (unsigned char) *p == 0x60 ||		/* ` */
+		    (unsigned char) *p == 0x24 ||		/* $ */
+		    !isprint((unsigned char) *p) ||
+		    iscntrl((unsigned char) *p)) {
+
+			fprintf(out, "\\x%02x", (unsigned char) *p);
+		} else
+			fputc(dir ==  1 ? toupper(*p) :
+			      dir == -1 ? tolower(*p) :
+			      *p, out);
+	}
+	fputc('"', out);
+}
+
+#define fputs_quoted(_d, _o)		fputs_quoted_case(_d, _o, 0)
+#define fputs_quoted_upper(_d, _o)	fputs_quoted_case(_d, _o, 1)
+#define fputs_quoted_lower(_d, _o)	fputs_quoted_case(_d, _o, -1)
+
+static inline void fputs_nonblank(const char *data, FILE *out)
+{
+	const char *p;
+
+	for (p = data; p && *p; p++) {
+		if (isblank((unsigned char) *p) ||
+		    (unsigned char) *p == 0x5c ||		/* \ */
+		    !isprint((unsigned char) *p) ||
+		    iscntrl((unsigned char) *p)) {
+
+			fprintf(out, "\\x%02x", (unsigned char) *p);
+
+		} else
+			fputc(*p, out);
+	}
+}
+
+#endif  /*  _CAREFULPUTC_H  */
diff --git a/include/cctype.h b/include/cctype.h
new file mode 100644
index 0000000..b673b6c
--- /dev/null
+++ b/include/cctype.h
@@ -0,0 +1,354 @@
+/* Character handling in C locale.
+
+   These functions work like the corresponding functions in <ctype.h>,
+   except that they have the C (POSIX) locale hardwired, whereas the
+   <ctype.h> functions' behaviour depends on the current locale set via
+   setlocale.
+
+   Copyright (C) 2000-2003, 2006, 2008-2023 Free Software Foundation, Inc.
+
+   This file is free software: you can redistribute it and/or modify
+   it under the terms of the GNU Lesser General Public License as
+   published by the Free Software Foundation; either version 2.1 of the
+   License, or (at your option) any later version.
+
+   This file is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public License
+   along with this program.  If not, see <https://www.gnu.org/licenses/>.  */
+
+#ifndef UTIL_LINUX_CCTYPE_H
+#define UTIL_LINUX_CCTYPE_H
+
+/**
+ * The functions defined in this file assume the "C" locale and a character
+ * set without diacritics (ASCII-US or EBCDIC-US or something like that).
+ * Even if the "C" locale on a particular system is an extension of the ASCII
+ * character set (like on BeOS, where it is UTF-8, or on AmigaOS, where it
+ * is ISO-8859-1), the functions in this file recognize only the ASCII
+ * characters.
+ */
+
+#if (' ' == 32) && ('!' == 33) && ('"' == 34) && ('#' == 35) \
+	&& ('%' == 37) && ('&' == 38) && ('\'' == 39) && ('(' == 40) \
+	&& (')' == 41) && ('*' == 42) && ('+' == 43) && (',' == 44) \
+	&& ('-' == 45) && ('.' == 46) && ('/' == 47) && ('0' == 48) \
+	&& ('1' == 49) && ('2' == 50) && ('3' == 51) && ('4' == 52) \
+	&& ('5' == 53) && ('6' == 54) && ('7' == 55) && ('8' == 56) \
+	&& ('9' == 57) && (':' == 58) && (';' == 59) && ('<' == 60) \
+	&& ('=' == 61) && ('>' == 62) && ('?' == 63) && ('A' == 65) \
+	&& ('B' == 66) && ('C' == 67) && ('D' == 68) && ('E' == 69) \
+	&& ('F' == 70) && ('G' == 71) && ('H' == 72) && ('I' == 73) \
+	&& ('J' == 74) && ('K' == 75) && ('L' == 76) && ('M' == 77) \
+	&& ('N' == 78) && ('O' == 79) && ('P' == 80) && ('Q' == 81) \
+	&& ('R' == 82) && ('S' == 83) && ('T' == 84) && ('U' == 85) \
+	&& ('V' == 86) && ('W' == 87) && ('X' == 88) && ('Y' == 89) \
+	&& ('Z' == 90) && ('[' == 91) && ('\\' == 92) && (']' == 93) \
+	&& ('^' == 94) && ('_' == 95) && ('a' == 97) && ('b' == 98) \
+	&& ('c' == 99) && ('d' == 100) && ('e' == 101) && ('f' == 102) \
+	&& ('g' == 103) && ('h' == 104) && ('i' == 105) && ('j' == 106) \
+	&& ('k' == 107) && ('l' == 108) && ('m' == 109) && ('n' == 110) \
+	&& ('o' == 111) && ('p' == 112) && ('q' == 113) && ('r' == 114) \
+	&& ('s' == 115) && ('t' == 116) && ('u' == 117) && ('v' == 118) \
+	&& ('w' == 119) && ('x' == 120) && ('y' == 121) && ('z' == 122) \
+	&& ('{' == 123) && ('|' == 124) && ('}' == 125) && ('~' == 126)
+
+/*
+ * The character set is ASCII or one of its variants or extensions, not EBCDIC.
+ * Testing the value of '\n' and '\r' is not relevant.
+ */
+# define C_CTYPE_ASCII 1
+#elif ! (' ' == '\x40' && '0' == '\xf0'                     \
+	 && 'A' == '\xc1' && 'J' == '\xd1' && 'S' == '\xe2' \
+	 && 'a' == '\x81' && 'j' == '\x91' && 's' == '\xa2')
+# error "Only ASCII and EBCDIC are supported"
+#endif
+
+#if 'A' < 0
+# error "EBCDIC and char is signed -- not supported"
+#endif
+
+/* Cases for control characters. */
+#define _C_CTYPE_CNTRL \
+	case '\a': case '\b': case '\f': case '\n': \
+	case '\r': case '\t': case '\v': \
+	_C_CTYPE_OTHER_CNTRL
+
+/* ASCII control characters other than those with \-letter escapes. */
+#if C_CTYPE_ASCII
+# define _C_CTYPE_OTHER_CNTRL \
+	case '\x00': case '\x01': case '\x02': case '\x03': \
+	case '\x04': case '\x05': case '\x06': case '\x0e': \
+	case '\x0f': case '\x10': case '\x11': case '\x12': \
+	case '\x13': case '\x14': case '\x15': case '\x16': \
+	case '\x17': case '\x18': case '\x19': case '\x1a': \
+	case '\x1b': case '\x1c': case '\x1d': case '\x1e': \
+	case '\x1f': case '\x7f'
+#else
+
+/*
+ * Use EBCDIC code page 1047's assignments for ASCII control chars;
+ * assume all EBCDIC code pages agree about these assignments.
+ */
+# define _C_CTYPE_OTHER_CNTRL \
+	case '\x00': case '\x01': case '\x02': case '\x03': \
+	case '\x07': case '\x0e': case '\x0f': case '\x10': \
+	case '\x11': case '\x12': case '\x13': case '\x18': \
+	case '\x19': case '\x1c': case '\x1d': case '\x1e': \
+	case '\x1f': case '\x26': case '\x27': case '\x2d': \
+	case '\x2e': case '\x32': case '\x37': case '\x3c': \
+	case '\x3d': case '\x3f'
+#endif
+
+/* Cases for lowercase hex letters, and lowercase letters, all offset by N. */
+#define _C_CTYPE_LOWER_A_THRU_F_N(N) \
+	case 'a' + (N): case 'b' + (N): case 'c' + (N): case 'd' + (N): \
+	case 'e' + (N): case 'f' + (N)
+#define _C_CTYPE_LOWER_N(N) \
+	_C_CTYPE_LOWER_A_THRU_F_N(N): \
+	case 'g' + (N): case 'h' + (N): case 'i' + (N): case 'j' + (N): \
+	case 'k' + (N): case 'l' + (N): case 'm' + (N): case 'n' + (N): \
+	case 'o' + (N): case 'p' + (N): case 'q' + (N): case 'r' + (N): \
+	case 's' + (N): case 't' + (N): case 'u' + (N): case 'v' + (N): \
+	case 'w' + (N): case 'x' + (N): case 'y' + (N): case 'z' + (N)
+
+/* Cases for hex letters, digits, lower, punct, and upper. */
+#define _C_CTYPE_A_THRU_F \
+	_C_CTYPE_LOWER_A_THRU_F_N (0): \
+	_C_CTYPE_LOWER_A_THRU_F_N ('A' - 'a')
+#define _C_CTYPE_DIGIT                     \
+	case '0': case '1': case '2': case '3': \
+	case '4': case '5': case '6': case '7': \
+	case '8': case '9'
+#define _C_CTYPE_LOWER _C_CTYPE_LOWER_N (0)
+#define _C_CTYPE_PUNCT \
+	case '!': case '"': case '#': case '$':  \
+	case '%': case '&': case '\'': case '(': \
+	case ')': case '*': case '+': case ',':  \
+	case '-': case '.': case '/': case ':':  \
+	case ';': case '<': case '=': case '>':  \
+	case '?': case '@': case '[': case '\\': \
+	case ']': case '^': case '_': case '`':  \
+	case '{': case '|': case '}': case '~'
+#define _C_CTYPE_UPPER _C_CTYPE_LOWER_N ('A' - 'a')
+
+/**
+ * Function definitions.
+ *
+ * Unlike the functions in <ctype.h>, which require an argument in the range
+ * of the 'unsigned char' type, the functions here operate on values that are
+ * in the 'unsigned char' range or in the 'char' range.  In other words,
+ * when you have a 'char' value, you need to cast it before using it as
+ * argument to a <ctype.h> function:
+ *
+ * 	const char *s = ...;
+ * 	if (isalpha ((unsigned char) *s)) ...
+ *
+ * but you don't need to cast it for the functions defined in this file:
+ *
+ * 	const char *s = ...;
+ * 	if (c_isalpha (*s)) ...
+ */
+
+static inline int c_isalnum (int c)
+{
+	switch (c) {
+	_C_CTYPE_DIGIT:
+	_C_CTYPE_LOWER:
+	_C_CTYPE_UPPER:
+		return 1;
+	default:
+		return 0;
+	}
+}
+
+static inline int c_isalpha (int c)
+{
+	switch (c) {
+	_C_CTYPE_LOWER:
+	_C_CTYPE_UPPER:
+		return 1;
+	default:
+		return 0;
+	}
+}
+
+/* The function isascii is not locale dependent.
+ * Its use in EBCDIC is questionable.
+ */
+static inline int c_isascii (int c)
+{
+	switch (c) {
+	case ' ':
+	_C_CTYPE_CNTRL:
+	_C_CTYPE_DIGIT:
+	_C_CTYPE_LOWER:
+	_C_CTYPE_PUNCT:
+	_C_CTYPE_UPPER:
+		return 1;
+	default:
+		return 0;
+	}
+}
+
+static inline int c_isblank (int c)
+{
+	return c == ' ' || c == '\t';
+}
+
+static inline int c_iscntrl (int c)
+{
+	switch (c) {
+	_C_CTYPE_CNTRL:
+		return 1;
+	default:
+		return 0;
+	}
+}
+
+static inline int c_isdigit (int c)
+{
+	switch (c) {
+	_C_CTYPE_DIGIT:
+		return 1;
+	default:
+		return 0;
+	}
+}
+
+static inline int c_isgraph (int c)
+{
+	switch (c) {
+	_C_CTYPE_DIGIT:
+	_C_CTYPE_LOWER:
+	_C_CTYPE_PUNCT:
+	_C_CTYPE_UPPER:
+		return 1;
+	default:
+		return 0;
+	}
+}
+
+static inline int c_islower (int c)
+{
+	switch (c) {
+	_C_CTYPE_LOWER:
+		return 1;
+	default:
+		return 0;
+	}
+}
+
+static inline int c_isprint (int c)
+{
+	switch (c) {
+	case ' ':
+	_C_CTYPE_DIGIT:
+	_C_CTYPE_LOWER:
+	_C_CTYPE_PUNCT:
+	_C_CTYPE_UPPER:
+		return 1;
+	default:
+		return 0;
+	}
+}
+
+static inline int c_ispunct (int c)
+{
+	switch (c) {
+	_C_CTYPE_PUNCT:
+		return 1;
+	default:
+		return 0;
+	}
+}
+
+static inline int c_isspace (int c)
+{
+	switch (c) {
+	case ' ': case '\t': case '\n': case '\v': case '\f': case '\r':
+		return 1;
+	default:
+		return 0;
+	}
+}
+
+static inline int c_isupper (int c)
+{
+	switch (c) {
+	_C_CTYPE_UPPER:
+		return 1;
+	default:
+		return 0;
+	}
+}
+
+static inline int c_isxdigit (int c)
+{
+	switch (c) {
+	_C_CTYPE_DIGIT:
+	_C_CTYPE_A_THRU_F:
+		return 1;
+	default:
+		return 0;
+	}
+}
+
+static inline int c_tolower (int c)
+{
+	switch (c) {
+	_C_CTYPE_UPPER:
+		return c - 'A' + 'a';
+	default:
+		return c;
+	}
+}
+
+static inline int c_toupper (int c)
+{
+	switch (c) {
+	_C_CTYPE_LOWER:
+		return c - 'a' + 'A';
+	default:
+		return c;
+	}
+}
+
+static inline int c_strncasecmp(const char *a, const char *b, size_t n)
+{
+	int res = 0;
+
+        for (; n > 0; a++, b++, n--) {
+		unsigned int x = (unsigned int) *a;
+		unsigned int y = (unsigned int) *b;
+
+		res = c_tolower(x) - c_tolower(y);
+                if (res)
+                        break;
+        }
+        return res;
+}
+
+static inline int c_strcasecmp(const char *a, const char *b)
+{
+	int res = 0;
+
+	if (a == b)
+		return 0;
+
+	for (; *a != '\0'; a++, b++) {
+		unsigned int x = (unsigned int) *a;
+		unsigned int y = (unsigned int) *b;
+
+		res = c_tolower(x) - c_tolower(y);
+		if (res)
+			break;
+	}
+
+	return res;
+}
+
+#endif /* UTIL_LINUX_CCTYPE_H */
diff --git a/include/closestream.h b/include/closestream.h
new file mode 100644
index 0000000..711ca5b
--- /dev/null
+++ b/include/closestream.h
@@ -0,0 +1,114 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_CLOSESTREAM_H
+#define UTIL_LINUX_CLOSESTREAM_H
+
+#include <stdio.h>
+#ifdef HAVE_STDIO_EXT_H
+#include <stdio_ext.h>
+#endif
+#include <unistd.h>
+
+#include "c.h"
+#include "nls.h"
+
+#ifndef CLOSE_EXIT_CODE
+# define CLOSE_EXIT_CODE EXIT_FAILURE
+#endif
+
+static inline int
+close_stream(FILE * stream)
+{
+#ifdef HAVE___FPENDING
+	const int some_pending = (__fpending(stream) != 0);
+#endif
+	const int prev_fail = (ferror(stream) != 0);
+	const int fclose_fail = (fclose(stream) != 0);
+
+	if (prev_fail || (fclose_fail && (
+#ifdef HAVE___FPENDING
+					  some_pending ||
+#endif
+					  errno != EBADF))) {
+		if (!fclose_fail && !(errno == EPIPE))
+			errno = 0;
+		return EOF;
+	}
+	return 0;
+}
+
+static inline int
+flush_standard_stream(FILE *stream)
+{
+	int fd;
+
+	errno = 0;
+
+	if (ferror(stream) != 0 || fflush(stream) != 0)
+		goto error;
+
+	/*
+	 * Calling fflush is not sufficient on some filesystems
+	 * like e.g. NFS, which may defer the actual flush until
+	 * close. Calling fsync would help solve this, but would
+	 * probably result in a performance hit. Thus, we work
+	 * around this issue by calling close on a dup'd file
+	 * descriptor from the stream.
+	 */
+	if ((fd = fileno(stream)) < 0 || (fd = dup(fd)) < 0 || close(fd) != 0)
+		goto error;
+
+	return 0;
+error:
+	return (errno == EBADF) ? 0 : EOF;
+}
+
+/* Meant to be used atexit(close_stdout); */
+static inline void
+close_stdout(void)
+{
+	if (flush_standard_stream(stdout) != 0 && !(errno == EPIPE)) {
+		if (errno)
+			warn(_("write error"));
+		else
+			warnx(_("write error"));
+		_exit(CLOSE_EXIT_CODE);
+	}
+
+	if (flush_standard_stream(stderr) != 0)
+		_exit(CLOSE_EXIT_CODE);
+}
+
+static inline void
+close_stdout_atexit(void)
+{
+	/*
+	 * Note that close stdout at exit disables ASAN to report memory leaks
+	 */
+#if !HAS_FEATURE_ADDRESS_SANITIZER
+	atexit(close_stdout);
+#endif
+}
+
+#ifndef HAVE_FSYNC
+static inline int
+fsync(int fd __attribute__((__unused__)))
+{
+	return 0;
+}
+#endif
+
+static inline int
+close_fd(int fd)
+{
+	const int fsync_fail = (fsync(fd) != 0);
+	const int close_fail = (close(fd) != 0);
+
+	if (fsync_fail || close_fail)
+		return EOF;
+	return 0;
+}
+
+#endif /* UTIL_LINUX_CLOSESTREAM_H */
diff --git a/include/color-names.h b/include/color-names.h
new file mode 100644
index 0000000..d6a5267
--- /dev/null
+++ b/include/color-names.h
@@ -0,0 +1,47 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ *
+ * Copyright (C) 2012-2015 Karel Zak <kzak@redhat.com>
+ */
+#ifndef UTIL_LINUX_COLOR_NAMES_H
+#define UTIL_LINUX_COLOR_NAMES_H
+
+#define UL_COLOR_RESET		"\033[0m"
+#define UL_COLOR_BOLD		"\033[1m"
+#define UL_COLOR_HALFBRIGHT	"\033[2m"
+#define UL_COLOR_UNDERSCORE	"\033[4m"
+#define UL_COLOR_BLINK		"\033[5m"
+#define UL_COLOR_REVERSE	"\033[7m"
+
+/* Standard colors */
+#define UL_COLOR_BLACK		"\033[30m"
+#define UL_COLOR_RED		"\033[31m"
+#define UL_COLOR_GREEN		"\033[32m"
+#define UL_COLOR_BROWN		"\033[33m"	/* well, brown */
+#define UL_COLOR_BLUE		"\033[34m"
+#define UL_COLOR_MAGENTA	"\033[35m"
+#define UL_COLOR_CYAN		"\033[36m"
+#define UL_COLOR_GRAY		"\033[37m"
+
+/* Bold variants */
+#define UL_COLOR_DARK_GRAY	"\033[1;30m"
+#define UL_COLOR_BOLD_RED	"\033[1;31m"
+#define UL_COLOR_BOLD_GREEN	"\033[1;32m"
+#define UL_COLOR_BOLD_YELLOW	"\033[1;33m"
+#define UL_COLOR_BOLD_BLUE	"\033[1;34m"
+#define UL_COLOR_BOLD_MAGENTA	"\033[1;35m"
+#define UL_COLOR_BOLD_CYAN	"\033[1;36m"
+
+#define UL_COLOR_WHITE		"\033[1;37m"
+
+
+/* maximal length of human readable name of ESC seq. */
+#define UL_COLORNAME_MAXSZ	32
+
+extern const char *color_sequence_from_colorname(const char *str);
+
+extern int color_is_sequence(const char *color);
+extern char *color_get_sequence(const char *color);
+
+#endif /* UTIL_LINUX_COLOR_NAMES_H */
diff --git a/include/colors.h b/include/colors.h
new file mode 100644
index 0000000..d91f5af
--- /dev/null
+++ b/include/colors.h
@@ -0,0 +1,77 @@
+/*
+ * SPDX-License-Identifier: LGPL-2.1-or-later
+ *
+ * Copyright (C) 2012 Ondrej Oprala <ooprala@redhat.com>
+ * Copyright (C) 2012-2014 Karel Zak <kzak@redhat.com>
+ *
+ * This file may be distributed under the terms of the
+ * GNU Lesser General Public License.
+ */
+#ifndef UTIL_LINUX_COLORS_H
+#define UTIL_LINUX_COLORS_H
+
+#include <stdio.h>
+#include <unistd.h>
+
+#include "color-names.h"
+
+/* --color[=WHEN] */
+enum colortmode {
+	UL_COLORMODE_AUTO = 0,
+	UL_COLORMODE_NEVER,
+	UL_COLORMODE_ALWAYS,
+	UL_COLORMODE_UNDEF,
+
+	__UL_NCOLORMODES	/* last */
+};
+
+#ifdef USE_COLORS_BY_DEFAULT
+# define USAGE_COLORS_DEFAULT	_("colors are enabled by default")
+#else
+# define USAGE_COLORS_DEFAULT   _("colors are disabled by default")
+#endif
+
+extern int colormode_from_string(const char *str);
+extern int colormode_or_err(const char *str, const char *errmsg);
+
+/* Initialize the global variable UL_COLOR_TERM_OK */
+extern int colors_init(int mode, const char *util_name);
+
+/* Returns 1 or 0 */
+extern int colors_wanted(void);
+
+/* Returns UL_COLORMODE_* */
+extern int colors_mode(void);
+
+/* temporary enable/disable colors */
+extern void colors_off(void);
+extern void colors_on(void);
+
+
+/* Set the color */
+extern void color_fenable(const char *seq, FILE *f);
+
+extern void color_scheme_fenable(const char *name, const char *dflt, FILE *f);
+extern const char *color_scheme_get_sequence(const char *name, const char *dflt);
+
+static inline void color_enable(const char *seq)
+{
+	color_fenable(seq, stdout);
+}
+
+static inline void color_scheme_enable(const char *name, const char *dflt)
+{
+	color_scheme_fenable(name, dflt, stdout);
+}
+
+/* Reset colors to default */
+extern void color_fdisable(FILE *f);
+
+static inline void color_disable(void)
+{
+	color_fdisable(stdout);
+}
+
+const char *color_get_disable_sequence(void);
+
+#endif /* UTIL_LINUX_COLORS_H */
diff --git a/include/cpuset.h b/include/cpuset.h
new file mode 100644
index 0000000..119c1a0
--- /dev/null
+++ b/include/cpuset.h
@@ -0,0 +1,101 @@
+/*
+ * SPDX-License-Identifier: LGPL-2.1-or-later
+ *
+ * This file may be redistributed under the terms of the
+ * GNU Lesser General Public License.
+ */
+#ifndef UTIL_LINUX_CPUSET_H
+#define UTIL_LINUX_CPUSET_H
+
+#include <sched.h>
+
+/*
+ * Fallback for old or obscure libcs without dynamically allocated cpusets
+ *
+ * The following macros are based on code from glibc.
+ *
+ * The GNU C Library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ */
+#if !HAVE_DECL_CPU_ALLOC
+
+# define CPU_ZERO_S(setsize, cpusetp) \
+  do {									      \
+    size_t __i;								      \
+    size_t __imax = (setsize) / sizeof (__cpu_mask);			      \
+    __cpu_mask *__bits = (cpusetp)->__bits;				      \
+    for (__i = 0; __i < __imax; ++__i)					      \
+      __bits[__i] = 0;							      \
+  } while (0)
+
+# define CPU_SET_S(cpu, setsize, cpusetp) \
+   ({ size_t __cpu = (cpu);						      \
+      __cpu < 8 * (setsize)						      \
+      ? (((__cpu_mask *) ((cpusetp)->__bits))[__CPUELT (__cpu)]		      \
+	 |= __CPUMASK (__cpu))						      \
+      : 0; })
+
+# define CPU_ISSET_S(cpu, setsize, cpusetp) \
+   ({ size_t __cpu = (cpu);						      \
+      __cpu < 8 * (setsize)						      \
+      ? ((((__cpu_mask *) ((cpusetp)->__bits))[__CPUELT (__cpu)]	      \
+	  & __CPUMASK (__cpu))) != 0					      \
+      : 0; })
+
+# define CPU_EQUAL_S(setsize, cpusetp1, cpusetp2) \
+   ({ __cpu_mask *__arr1 = (cpusetp1)->__bits;				      \
+      __cpu_mask *__arr2 = (cpusetp2)->__bits;				      \
+      size_t __imax = (setsize) / sizeof (__cpu_mask);			      \
+      size_t __i;							      \
+      for (__i = 0; __i < __imax; ++__i)				      \
+	if (__arr1[__i] != __arr2[__i])					      \
+	  break;							      \
+      __i == __imax; })
+
+extern int __cpuset_count_s(size_t setsize, const cpu_set_t *set);
+# define CPU_COUNT_S(setsize, cpusetp)	__cpuset_count_s(setsize, cpusetp)
+
+# define CPU_ALLOC_SIZE(count) \
+	  ((((count) + __NCPUBITS - 1) / __NCPUBITS) * sizeof (__cpu_mask))
+# define CPU_ALLOC(count)	(malloc(CPU_ALLOC_SIZE(count)))
+# define CPU_FREE(cpuset)	(free(cpuset))
+
+#endif /* !HAVE_DECL_CPU_ALLOC */
+
+
+#define cpuset_nbits(setsize)	(8 * (setsize))
+
+/*
+ * The @idx parameter returns an index of the first mask from @ary array where
+ * the @cpu is set.
+ *
+ * Returns: 0 if found, otherwise 1.
+ */
+static inline int cpuset_ary_isset(size_t cpu, cpu_set_t **ary, size_t nmemb,
+				   size_t setsize, size_t *idx)
+{
+	size_t i;
+
+	for (i = 0; i < nmemb; i++) {
+		if (CPU_ISSET_S(cpu, setsize, ary[i])) {
+			*idx = i;
+			return 0;
+		}
+	}
+	return 1;
+}
+
+extern int get_max_number_of_cpus(void);
+
+extern cpu_set_t *cpuset_alloc(int ncpus, size_t *setsize, size_t *nbits);
+extern void cpuset_free(cpu_set_t *set);
+
+extern char *cpulist_create(char *str, size_t len, cpu_set_t *set, size_t setsize);
+extern int cpulist_parse(const char *str, cpu_set_t *set, size_t setsize, int fail);
+
+extern char *cpumask_create(char *str, size_t len, cpu_set_t *set, size_t setsize);
+extern int cpumask_parse(const char *str, cpu_set_t *set, size_t setsize);
+
+#endif /* UTIL_LINUX_CPUSET_H */
diff --git a/include/crc32.h b/include/crc32.h
new file mode 100644
index 0000000..81c7703
--- /dev/null
+++ b/include/crc32.h
@@ -0,0 +1,16 @@
+/*
+ * No copyright is claimed. This code is in the public domain; do with it what
+ * you wish.
+ */
+#ifndef UL_CRC32_H
+#define UL_CRC32_H
+
+#include <sys/types.h>
+#include <stdint.h>
+
+extern uint32_t ul_crc32(uint32_t seed, const unsigned char *buf, size_t len);
+extern uint32_t ul_crc32_exclude_offset(uint32_t seed, const unsigned char *buf, size_t len,
+		                              size_t exclude_off, size_t exclude_len);
+
+#endif
+
diff --git a/include/crc32c.h b/include/crc32c.h
new file mode 100644
index 0000000..3d27461
--- /dev/null
+++ b/include/crc32c.h
@@ -0,0 +1,17 @@
+/*
+ * No copyright is claimed. This code is in the public domain; do with it what
+ * you wish.
+ */
+#ifndef UL_CRC32C_H
+#define UL_CRC32C_H
+
+#include <sys/types.h>
+#include <stdint.h>
+
+extern uint32_t crc32c(uint32_t crc, const void *buf, size_t size);
+extern uint32_t ul_crc32c_exclude_offset(uint32_t crc, const unsigned char *buf,
+					 size_t size, size_t exclude_off,
+					 size_t exclude_len);
+
+
+#endif /* UL_CRC32C_H */
diff --git a/include/crc64.h b/include/crc64.h
new file mode 100644
index 0000000..7cd2c4c
--- /dev/null
+++ b/include/crc64.h
@@ -0,0 +1,12 @@
+#ifndef UL_CRC64_H
+#define UL_CRC64_H
+
+#include <sys/types.h>
+#include <stdint.h>
+
+extern uint64_t ul_crc64_ecma(const unsigned char *input_str, size_t num_bytes);
+extern uint64_t ul_crc64_we(const unsigned char *input_str, size_t num_bytes);
+extern uint64_t ul_update_crc64(uint64_t crc, unsigned char c);
+
+#endif
+
diff --git a/include/debug.h b/include/debug.h
new file mode 100644
index 0000000..4497d87
--- /dev/null
+++ b/include/debug.h
@@ -0,0 +1,183 @@
+/*
+ * SPDX-License-Identifier: LGPL-2.1-or-later
+ *
+ * Copyright (C) 2014 Ondrej Oprala <ooprala@redhat.com>
+ * Copyright (C) 2014 Karel Zak <kzak@redhat.com>
+ *
+ * This file may be distributed under the terms of the
+ * GNU Lesser General Public License.
+ */
+#ifndef UTIL_LINUX_DEBUG_H
+#define UTIL_LINUX_DEBUG_H
+
+
+/*
+ * util-linux debug macros
+ *
+ * The debug stuff is based on <name>_debug_mask that controls what outputs is
+ * expected. The mask is usually initialized by <NAME>_DEBUG= env.variable
+ *
+ * After successful initialization the flag <PREFIX>_DEBUG_INIT is always set
+ * to the mask (this flag is required). The <PREFIX> is usually library API
+ * prefix (e.g. MNT_) or program name (e.g. CFDISK_)
+ *
+ * In the code is possible to use
+ *
+ *	DBG(FOO, ul_debug("this is output for foo"));
+ *
+ * where for the FOO has to be defined <PREFIX>_DEBUG_FOO.
+ *
+ * It's possible to initialize the mask by comma delimited strings with
+ * subsystem names (e.g. "LIBMOUNT_DEBUG=options,tab"). In this case is
+ * necessary to define mask names array. This functionality is optional.
+ *
+ * It's strongly recommended to use UL_* macros to define/declare/use
+ * the debug stuff.
+ *
+ * See disk-utils/cfdisk.c: cfdisk_init_debug()  for programs debug
+ *  or libmount/src/init.c: mnt_init_debug()     for library debug
+ *
+ */
+
+#include <stdarg.h>
+#include <string.h>
+
+struct ul_debug_maskname {
+	const char *name;
+	int mask;
+	const char *help;
+};
+#define UL_DEBUG_EMPTY_MASKNAMES {{ NULL, 0, NULL }}
+#define UL_DEBUG_DEFINE_MASKNAMES(m) static const struct ul_debug_maskname m ## _masknames[]
+#define UL_DEBUG_MASKNAMES(m)	m ## _masknames
+
+#define UL_DEBUG_MASK(m)         m ## _debug_mask
+#define UL_DEBUG_DEFINE_MASK(m)  int UL_DEBUG_MASK(m)
+#define UL_DEBUG_DECLARE_MASK(m) extern UL_DEBUG_DEFINE_MASK(m)
+
+/*
+ * Internal mask flags (above 0xffffff)
+ */
+#define __UL_DEBUG_FL_NOADDR	(1 << 24)	/* Don't print object address */
+
+
+/* l - library name, p - flag prefix, m - flag postfix, x - function */
+#define __UL_DBG(l, p, m, x) \
+	do { \
+		if ((p ## m) & l ## _debug_mask) { \
+			fprintf(stderr, "%d: %s: %8s: ", getpid(), # l, # m); \
+			x; \
+		} \
+	} while (0)
+
+#define __UL_DBG_CALL(l, p, m, x) \
+	do { \
+		if ((p ## m) & l ## _debug_mask) { \
+			x; \
+		} \
+	} while (0)
+
+#define __UL_DBG_FLUSH(l, p) \
+	do { \
+		if (l ## _debug_mask && \
+		    l ## _debug_mask != p ## INIT) { \
+			fflush(stderr); \
+		} \
+	} while (0)
+
+#define __UL_INIT_DEBUG_FROM_STRING(lib, pref, mask, str) \
+	do { \
+		if (lib ## _debug_mask & pref ## INIT) \
+		; \
+		else if (!mask && str) { \
+			lib ## _debug_mask = ul_debug_parse_mask(lib ## _masknames, str); \
+		} else \
+			lib ## _debug_mask = mask; \
+		if (lib ## _debug_mask) { \
+			if (getuid() != geteuid() || getgid() != getegid()) { \
+				lib ## _debug_mask |= __UL_DEBUG_FL_NOADDR; \
+				fprintf(stderr, "%d: %s: don't print memory addresses (SUID executable).\n", getpid(), # lib); \
+			} \
+		} \
+		lib ## _debug_mask |= pref ## INIT; \
+	} while (0)
+
+
+#define __UL_INIT_DEBUG_FROM_ENV(lib, pref, mask, env) \
+	do { \
+		const char *envstr = mask ? NULL : getenv(# env); \
+		__UL_INIT_DEBUG_FROM_STRING(lib, pref, mask, envstr); \
+	} while (0)
+
+
+
+static inline void __attribute__ ((__format__ (__printf__, 1, 2)))
+ul_debug(const char *mesg, ...)
+{
+	va_list ap;
+	va_start(ap, mesg);
+	vfprintf(stderr, mesg, ap);
+	va_end(ap);
+	fputc('\n', stderr);
+}
+
+static inline int ul_debug_parse_mask(
+			const struct ul_debug_maskname flagnames[],
+			const char *mask)
+{
+	int res;
+	char *ptr;
+
+	/* let's check for a numeric mask first */
+	res = strtoul(mask, &ptr, 0);
+
+	/* perhaps it's a comma-separated string? */
+	if (ptr && *ptr && flagnames && flagnames[0].name) {
+		char *msbuf, *ms, *name;
+		res = 0;
+
+		ms = msbuf = strdup(mask);
+		if (!ms)
+			return res;
+
+		while ((name = strtok_r(ms, ",", &ptr))) {
+			const struct ul_debug_maskname *d;
+			ms = ptr;
+
+			for (d = flagnames; d && d->name; d++) {
+				if (strcmp(name, d->name) == 0) {
+					res |= d->mask;
+					break;
+				}
+			}
+			/* nothing else we can do by OR-ing the mask */
+			if (res == 0xffff)
+				break;
+		}
+		free(msbuf);
+	} else if (ptr && strcmp(ptr, "all") == 0)
+		res = 0xffff;
+
+	return res;
+}
+
+static inline void ul_debug_print_masks(
+			const char *env,
+			const struct ul_debug_maskname flagnames[])
+{
+	const struct ul_debug_maskname *d;
+
+	if (!flagnames)
+		return;
+
+	fprintf(stderr, "Available \"%s=<name>[,...]|<mask>\" debug masks:\n",
+			env);
+	for (d = flagnames; d && d->name; d++) {
+		if (!d->help)
+			continue;
+		fprintf(stderr, "   %-8s [0x%06x] : %s\n",
+				d->name, d->mask, d->help);
+	}
+}
+
+#endif /* UTIL_LINUX_DEBUG_H */
diff --git a/include/debugobj.h b/include/debugobj.h
new file mode 100644
index 0000000..eac3ff9
--- /dev/null
+++ b/include/debugobj.h
@@ -0,0 +1,31 @@
+/*
+ * SPDX-License-Identifier: LGPL-2.1-or-later
+ *
+ * Copyright (C) 2014 Ondrej Oprala <ooprala@redhat.com>
+ * Copyright (C) 2014 Karel Zak <kzak@redhat.com>
+ *
+ * This file may be distributed under the terms of the
+ * GNU Lesser General Public License.
+ */
+#ifndef UTIL_LINUX_DEBUGOBJ_H
+#define UTIL_LINUX_DEBUGOBJ_H
+
+/*
+ * Include *after* debug.h and after UL_DEBUG_CURRENT_MASK define.
+ */
+
+static inline void __attribute__ ((__format__ (__printf__, 2, 3)))
+ul_debugobj(const void *handler, const char *mesg, ...)
+{
+	va_list ap;
+
+	if (handler && !(UL_DEBUG_CURRENT_MASK & __UL_DEBUG_FL_NOADDR))
+		fprintf(stderr, "[%p]: ", handler);
+
+	va_start(ap, mesg);
+	vfprintf(stderr, mesg, ap);
+	va_end(ap);
+	fputc('\n', stderr);
+}
+
+#endif /* UTIL_LINUX_DEBUGOBJ_H */
diff --git a/include/encode.h b/include/encode.h
new file mode 100644
index 0000000..d7e5a1e
--- /dev/null
+++ b/include/encode.h
@@ -0,0 +1,20 @@
+/*
+ * SPDX-License-Identifier: LGPL-2.1-or-later
+ *
+ * This file may be distributed under the terms of the
+ * GNU Lesser General Public License.
+ */
+#ifndef UTIL_LINUX_ENCODE_H
+#define UTIL_LINUX_ENCODE_H
+
+extern size_t ul_encode_to_utf8(int enc, unsigned char *dest, size_t len,
+				const unsigned char *src, size_t count)
+			__attribute__((nonnull));
+
+enum {
+	UL_ENCODE_UTF16BE = 0,
+	UL_ENCODE_UTF16LE,
+	UL_ENCODE_LATIN1
+};
+
+#endif
diff --git a/include/env.h b/include/env.h
new file mode 100644
index 0000000..9a0ec4f
--- /dev/null
+++ b/include/env.h
@@ -0,0 +1,39 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_ENV_H
+#define UTIL_LINUX_ENV_H
+
+#include "c.h"
+#include "nls.h"
+
+struct ul_env_list;
+
+extern void sanitize_env(void);
+extern void __sanitize_env(struct ul_env_list **org);
+
+extern int env_list_setenv(struct ul_env_list *ls);
+extern void env_list_free(struct ul_env_list *ls);
+extern struct ul_env_list *env_from_fd(int pid);
+
+extern char *safe_getenv(const char *arg);
+
+
+#ifndef XSETENV_EXIT_CODE
+# define XSETENV_EXIT_CODE EXIT_FAILURE
+#endif
+
+static inline void xsetenv(char const *name, char const *val, int overwrite)
+{
+	if (setenv(name, val, overwrite) != 0)
+		err(XSETENV_EXIT_CODE, _("failed to set the %s environment variable"), name);
+}
+
+static inline int remote_entry(char **argv, int remove, int last)
+{
+	memmove(argv + remove, argv + remove + 1, sizeof(char *) * (last - remove));
+	return last - 1;
+}
+
+#endif /* UTIL_LINUX_ENV_H */
diff --git a/include/exec_shell.h b/include/exec_shell.h
new file mode 100644
index 0000000..04aa089
--- /dev/null
+++ b/include/exec_shell.h
@@ -0,0 +1,6 @@
+#ifndef UTIL_LINUX_EXEC_SHELL_H
+#define UTIL_LINUX_EXEC_SHELL_H
+
+extern void exec_shell(void) __attribute__((__noreturn__));
+
+#endif /* UTIL_LINUX_EXEC_SHELL_H */
diff --git a/include/exitcodes.h b/include/exitcodes.h
new file mode 100644
index 0000000..b29c98c
--- /dev/null
+++ b/include/exitcodes.h
@@ -0,0 +1,27 @@
+#ifndef UTIL_LINUX_EXITCODES_H
+#define UTIL_LINUX_EXITCODES_H
+/*
+ * BE CAREFUL
+ *
+ * These exit codes are part of the official interface for mount,
+ * fsck, mkfs, etc. wrappers.
+ */
+
+/* Exit codes used by mkfs-type programs */
+#define MKFS_EX_OK	0	/* No errors */
+#define MKFS_EX_ERROR	8	/* Operational error */
+#define MKFS_EX_USAGE	16	/* Usage or syntax error */
+
+/* Exit codes used by fsck-type programs */
+#define FSCK_EX_OK		0	/* No errors */
+#define FSCK_EX_NONDESTRUCT	1	/* File system errors corrected */
+#define FSCK_EX_REBOOT		2	/* System should be rebooted */
+#define FSCK_EX_DESTRUCT	FSCK_EX_REBOOT	/* Alias */
+#define FSCK_EX_UNCORRECTED	4	/* File system errors left uncorrected */
+#define FSCK_EX_ERROR		8	/* Operational error */
+#define FSCK_EX_USAGE		16	/* Usage or syntax error */
+#define FSCK_EX_LIBRARY		128	/* Shared library error */
+
+#define EXIT_NOTSUPP 2
+
+#endif	/* UTIL_LINUX_EXITCODES_H */
diff --git a/include/fileeq.h b/include/fileeq.h
new file mode 100644
index 0000000..df5e6d8
--- /dev/null
+++ b/include/fileeq.h
@@ -0,0 +1,62 @@
+
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_FILEEQ
+#define UTIL_LINUX_FILEEQ
+
+#include <stdlib.h>
+#include <stdint.h>
+#include <stdbool.h>
+
+#if defined(__linux__) && defined(HAVE_LINUX_IF_ALG_H)
+# define USE_FILEEQ_CRYPTOAPI 1
+#endif
+
+/* Number of bytes from the beginning of the file we always
+ * compare by memcmp() */
+#define UL_FILEEQ_INTROSIZ	32
+
+struct ul_fileeq_data {
+	unsigned char intro[UL_FILEEQ_INTROSIZ];
+	unsigned char *blocks;
+	size_t nblocks;
+	size_t maxblocks;
+	int fd;
+	const char *name;
+	bool is_eof;
+};
+
+struct ul_fileeq {
+	int fd_api;	/* Crypto socket */
+	int fd_cip;	/* Cipher handler */
+
+	size_t readsiz;
+	uint64_t filesiz;
+	uint64_t blocksmax;
+	const struct ul_fileeq_method *method;
+
+	/* UL_FILEEQ_MEMCMP buffers */
+	unsigned char *buf_a;
+	unsigned char *buf_b;
+	unsigned char *buf_last;
+};
+
+extern int ul_fileeq_init(struct ul_fileeq *eq, const char *method);
+extern void ul_fileeq_deinit(struct ul_fileeq *eq);
+
+
+extern int ul_fileeq_data_associated(struct ul_fileeq_data *data);
+extern void ul_fileeq_data_close_file(struct ul_fileeq_data *data);
+extern void ul_fileeq_data_init(struct ul_fileeq_data *data);
+extern void ul_fileeq_data_deinit(struct ul_fileeq_data *data);
+extern void ul_fileeq_data_set_file(struct ul_fileeq_data *data,
+				    const char *name);
+extern size_t ul_fileeq_set_size(struct ul_fileeq *eq, uint64_t filesiz,
+                                 size_t readsiz, size_t memsiz);
+
+extern int ul_fileeq(struct ul_fileeq *eq,
+              struct ul_fileeq_data *a, struct ul_fileeq_data *b);
+
+#endif /* UTIL_LINUX_FILEEQ */
diff --git a/include/fileutils.h b/include/fileutils.h
new file mode 100644
index 0000000..538eab0
--- /dev/null
+++ b/include/fileutils.h
@@ -0,0 +1,118 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ **/
+#ifndef UTIL_LINUX_FILEUTILS
+#define UTIL_LINUX_FILEUTILS
+
+#include <stdio.h>
+#include <fcntl.h>
+#include <unistd.h>
+#include <dirent.h>
+#include <sys/stat.h>
+
+#include "c.h"
+
+extern int mkstemp_cloexec(char *template);
+
+extern int xmkstemp(char **tmpname, const char *dir, const char *prefix);
+
+static inline FILE *xfmkstemp(char **tmpname, const char *dir, const char *prefix)
+{
+	int fd;
+	FILE *ret;
+
+	fd = xmkstemp(tmpname, dir, prefix);
+	if (fd == -1)
+		return NULL;
+
+	if (!(ret = fdopen(fd, "w+" UL_CLOEXECSTR))) {
+		close(fd);
+		return NULL;
+	}
+	return ret;
+}
+
+#ifdef HAVE_OPENAT
+static inline FILE *fopen_at(int dir, const char *filename,
+                             int flags, const char *mode)
+{
+	int fd = openat(dir, filename, flags);
+	FILE *ret;
+
+	if (fd < 0)
+		return NULL;
+
+	ret = fdopen(fd, mode);
+	if (!ret)
+		close(fd);
+	return ret;
+}
+#endif
+
+static inline int is_same_inode(const int fd, const struct stat *st)
+{
+	struct stat f;
+
+	if (fstat(fd, &f) < 0)
+		return 0;
+	else if (f.st_dev != st->st_dev || f.st_ino != st->st_ino)
+		return 0;
+	return 1;
+}
+
+extern int dup_fd_cloexec(int oldfd, int lowfd);
+extern unsigned int get_fd_tabsize(void);
+
+extern int ul_mkdir_p(const char *path, mode_t mode);
+extern char *stripoff_last_component(char *path);
+
+/* This is readdir()-like function, but skips "." and ".." directory entries */
+static inline struct dirent *xreaddir(DIR *dp)
+{
+	struct dirent *d;
+
+	while ((d = readdir(dp))) {
+		if (!strcmp(d->d_name, ".") ||
+		    !strcmp(d->d_name, ".."))
+			continue;
+		break;
+	}
+	return d;
+}
+
+
+#ifdef HAVE_SYS_SYSCALL_H
+# include <sys/syscall.h>
+
+# if !defined(HAVE_CLOSE_RANGE) && defined(SYS_close_range)
+#  include <sys/types.h>
+static inline int close_range(unsigned int first, unsigned int last, int flags)
+{
+	return syscall(SYS_close_range, first, last, flags);
+}
+#  define HAVE_CLOSE_RANGE 1
+# endif	/* SYS_close_range */
+
+# if !defined(HAVE_STATX) && defined(HAVE_STRUCT_STATX) && defined(SYS_statx)
+static inline int statx(int fd, const char *restrict path, int flags,
+		    unsigned int mask, struct statx *stx)
+{
+	return syscall(SYS_statx, fd, path, flags, mask, stx);
+}
+#  define HAVE_STATX 1
+# endif /* SYS_statx */
+
+#endif	/* HAVE_SYS_SYSCALL_H */
+
+
+extern void ul_close_all_fds(unsigned int first, unsigned int last);
+
+#define UL_COPY_READ_ERROR (-1)
+#define UL_COPY_WRITE_ERROR (-2)
+int ul_copy_file(int from, int to);
+
+
+extern int ul_reopen(int fd, int flags);
+
+#endif /* UTIL_LINUX_FILEUTILS */
diff --git a/include/fuzz.h b/include/fuzz.h
new file mode 100644
index 0000000..0bc728d
--- /dev/null
+++ b/include/fuzz.h
@@ -0,0 +1,13 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_FUZZ_H
+#define UTIL_LINUX_FUZZ_H
+
+#include <stddef.h>
+#include <stdint.h>
+
+int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size);
+
+#endif /* UTIL_LINUX_FUZZ_H */
diff --git a/include/idcache.h b/include/idcache.h
new file mode 100644
index 0000000..518a3b1
--- /dev/null
+++ b/include/idcache.h
@@ -0,0 +1,32 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_IDCACHE_H
+#define UTIL_LINUX_IDCACHE_H
+
+#include <sys/types.h>
+#include <pwd.h>
+
+#define IDCACHE_FLAGS_NAMELEN	(1 << 1)
+
+struct identry {
+	unsigned long int	id;
+	char			*name;
+	struct identry		*next;
+};
+
+struct idcache {
+	struct identry	*ent;	/* first entry */
+	int		width;	/* name width */
+};
+
+
+extern struct idcache *new_idcache(void);
+extern void add_gid(struct idcache *cache, unsigned long int id);
+extern void add_uid(struct idcache *cache, unsigned long int id);
+
+extern void free_idcache(struct idcache *ic);
+extern struct identry *get_id(struct idcache *ic, unsigned long int id);
+
+#endif /* UTIL_LINUX_IDCACHE_H */
diff --git a/include/ismounted.h b/include/ismounted.h
new file mode 100644
index 0000000..62b1d26
--- /dev/null
+++ b/include/ismounted.h
@@ -0,0 +1,20 @@
+/*
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ *
+ * This file may be redistributed under the terms of the GNU Public
+ * License.
+ */
+#ifndef IS_MOUNTED_H
+#define IS_MOUNTED_H
+
+#define MF_MOUNTED	1
+#define MF_ISROOT	2
+#define MF_READONLY	4
+#define MF_SWAP		8
+#define MF_BUSY		16
+
+extern int is_mounted(const char *file);
+extern int check_mount_point(const char *device, int *mount_flags,
+				 char *mtpt, int mtlen);
+
+#endif /* IS_MOUNTED_H */
diff --git a/include/iso9660.h b/include/iso9660.h
new file mode 100644
index 0000000..437d6a8
--- /dev/null
+++ b/include/iso9660.h
@@ -0,0 +1,63 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_ISO_H
+#define UTIL_LINUX_ISO_H
+
+#include <stdbool.h>
+#include <stdint.h>
+
+#include "c.h"
+
+static inline uint16_t isonum_721(const unsigned char *p)
+{
+	return ((p[0] & 0xff)
+		| ((p[1] & 0xff) << 8));
+}
+
+static inline uint16_t isonum_722(const unsigned char *p)
+{
+	return ((p[1] & 0xff)
+		| ((p[0] & 0xff) << 8));
+}
+
+static inline uint16_t isonum_723(const unsigned char *p, bool check_match)
+{
+	uint16_t le = isonum_721(p);
+	uint16_t be = isonum_722(p + 2);
+
+	if (check_match && le != be)
+		/* translation is useless */
+		warnx("723error: le=%d be=%d", le, be);
+	return (le);
+}
+
+static inline uint32_t isonum_731(const unsigned char *p)
+{
+	return ((p[0] & 0xff)
+		| ((p[1] & 0xff) << 8)
+		| ((p[2] & 0xff) << 16)
+		| (((uint32_t) p[3] & 0xff) << 24));
+}
+
+static inline uint32_t isonum_732(const unsigned char *p)
+{
+	return ((p[3] & 0xff)
+		| ((p[2] & 0xff) << 8)
+		| ((p[1] & 0xff) << 16)
+		| (((uint32_t) p[0] & 0xff) << 24));
+}
+
+static inline uint32_t isonum_733(const unsigned char *p, bool check_match)
+{
+	uint32_t le = isonum_731(p);
+	uint32_t be = isonum_732(p + 4);
+
+	if (check_match && le != be)
+		/* translation is useless */
+		warnx("733error: le=%d be=%d", le, be);
+	return(le);
+}
+
+#endif
diff --git a/include/jsonwrt.h b/include/jsonwrt.h
new file mode 100644
index 0000000..421903a
--- /dev/null
+++ b/include/jsonwrt.h
@@ -0,0 +1,51 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_JSONWRT_H
+#define UTIL_LINUX_JSONWRT_H
+
+enum {
+	UL_JSON_OBJECT,
+	UL_JSON_ARRAY,
+	UL_JSON_VALUE
+};
+
+struct ul_jsonwrt {
+	FILE *out;
+	int indent;
+
+	unsigned int after_close :1;
+};
+
+void ul_jsonwrt_init(struct ul_jsonwrt *fmt, FILE *out, int indent);
+int ul_jsonwrt_is_ready(struct ul_jsonwrt *fmt);
+void ul_jsonwrt_indent(struct ul_jsonwrt *fmt);
+void ul_jsonwrt_open(struct ul_jsonwrt *fmt, const char *name, int type);
+void ul_jsonwrt_close(struct ul_jsonwrt *fmt, int type);
+
+#define ul_jsonwrt_root_open(_f)	ul_jsonwrt_open(_f, NULL, UL_JSON_OBJECT)
+#define ul_jsonwrt_root_close(_f)	ul_jsonwrt_close(_f, UL_JSON_OBJECT)
+
+#define ul_jsonwrt_array_open(_f, _n)	ul_jsonwrt_open(_f, _n, UL_JSON_ARRAY)
+#define ul_jsonwrt_array_close(_f)	ul_jsonwrt_close(_f, UL_JSON_ARRAY)
+
+#define ul_jsonwrt_object_open(_f, _n)	ul_jsonwrt_open(_f, _n, UL_JSON_OBJECT)
+#define ul_jsonwrt_object_close(_f)	ul_jsonwrt_close(_f, UL_JSON_OBJECT)
+
+#define ul_jsonwrt_value_open(_f, _n)	ul_jsonwrt_open(_f, _n, UL_JSON_VALUE)
+#define ul_jsonwrt_value_close(_f)	ul_jsonwrt_close(_f, UL_JSON_VALUE)
+
+
+void ul_jsonwrt_value_raw(struct ul_jsonwrt *fmt,
+			const char *name, const char *data);
+void ul_jsonwrt_value_s(struct ul_jsonwrt *fmt,
+			const char *name, const char *data);
+void ul_jsonwrt_value_u64(struct ul_jsonwrt *fmt,
+			const char *name, uint64_t data);
+void ul_jsonwrt_value_boolean(struct ul_jsonwrt *fmt,
+			const char *name, int data);
+void ul_jsonwrt_value_null(struct ul_jsonwrt *fmt,
+			const char *name);
+
+#endif /* UTIL_LINUX_JSONWRT_H */
diff --git a/include/linux_version.h b/include/linux_version.h
new file mode 100644
index 0000000..2fa0d65
--- /dev/null
+++ b/include/linux_version.h
@@ -0,0 +1,18 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef LINUX_VERSION_H
+#define LINUX_VERSION_H
+
+#ifdef HAVE_LINUX_VERSION_H
+# include <linux/version.h>
+#endif
+
+#ifndef KERNEL_VERSION
+# define KERNEL_VERSION(a,b,c) (((a) << 16) + ((b) << 8) + (c))
+#endif
+
+int get_linux_version(void);
+
+#endif /* LINUX_VERSION_H */
diff --git a/include/list.h b/include/list.h
new file mode 100644
index 0000000..b6bbbdd
--- /dev/null
+++ b/include/list.h
@@ -0,0 +1,380 @@
+/*
+ * Copyright (C) 2008 Karel Zak <kzak@redhat.com>
+ * Copyright (C) 1999-2008 by Theodore Ts'o
+ *
+ * This file may be redistributed under the terms of the
+ * GNU Lesser General Public License.
+ *
+ * (based on list.h from e2fsprogs)
+ * Merge sort based on kernel's implementation.
+ */
+
+#ifndef UTIL_LINUX_LIST_H
+#define UTIL_LINUX_LIST_H
+
+#include "c.h"
+
+/* TODO: use AC_C_INLINE */
+#ifdef __GNUC__
+#define _INLINE_ static __inline__
+#else                         /* For Watcom C */
+#define _INLINE_ static inline
+#endif
+
+/*
+ * Simple doubly linked list implementation.
+ *
+ * Some of the internal functions ("__xxx") are useful when
+ * manipulating whole lists rather than single entries, as
+ * sometimes we already know the next/prev entries and we can
+ * generate better code by using them directly rather than
+ * using the generic single-entry routines.
+ */
+
+struct list_head {
+	struct list_head *next, *prev;
+};
+
+#define INIT_LIST_HEAD(ptr) do { \
+	(ptr)->next = (ptr); (ptr)->prev = (ptr); \
+} while (0)
+
+/*
+ * Insert a new entry between two known consecutive entries.
+ *
+ * This is only for internal list manipulation where we know
+ * the prev/next entries already!
+ */
+_INLINE_ void __list_add(struct list_head * add,
+	struct list_head * prev,
+	struct list_head * next)
+{
+	next->prev = add;
+	add->next = next;
+	add->prev = prev;
+	prev->next = add;
+}
+
+/**
+ * list_add - add a new entry
+ * @add:	new entry to be added
+ * @head:	list head to add it after
+ *
+ * Insert a new entry after the specified head.
+ * This is good for implementing stacks.
+ */
+_INLINE_ void list_add(struct list_head *add, struct list_head *head)
+{
+	__list_add(add, head, head->next);
+}
+
+/**
+ * list_add_tail - add a new entry
+ * @add:	new entry to be added
+ * @head:	list head to add it before
+ *
+ * Insert a new entry before the specified head.
+ * This is useful for implementing queues.
+ */
+_INLINE_ void list_add_tail(struct list_head *add, struct list_head *head)
+{
+	__list_add(add, head->prev, head);
+}
+
+/*
+ * Delete a list entry by making the prev/next entries
+ * point to each other.
+ *
+ * This is only for internal list manipulation where we know
+ * the prev/next entries already!
+ */
+_INLINE_ void __list_del(struct list_head * prev,
+				  struct list_head * next)
+{
+	next->prev = prev;
+	prev->next = next;
+}
+
+/**
+ * list_del - deletes entry from list.
+ * @entry:	the element to delete from the list.
+ *
+ * list_empty() on @entry does not return true after this, @entry is
+ * in an undefined state.
+ */
+_INLINE_ void list_del(struct list_head *entry)
+{
+	__list_del(entry->prev, entry->next);
+}
+
+/**
+ * list_del_init - deletes entry from list and reinitialize it.
+ * @entry:	the element to delete from the list.
+ */
+_INLINE_ void list_del_init(struct list_head *entry)
+{
+	__list_del(entry->prev, entry->next);
+	INIT_LIST_HEAD(entry);
+}
+
+/**
+ * list_empty - tests whether a list is empty
+ * @head:	the list to test.
+ */
+_INLINE_ int list_empty(struct list_head *head)
+{
+	return head->next == head;
+}
+
+/**
+ * list_entry_is_last - tests whether is entry last in the list
+ * @entry:	the entry to test.
+ * @head:	the list to test.
+ */
+_INLINE_ int list_entry_is_last(struct list_head *entry, struct list_head *head)
+{
+	return head->prev == entry;
+}
+
+/**
+ * list_entry_is_first - tests whether is entry first in the list
+ * @entry:	the entry to test.
+ * @head:	the list to test.
+ */
+_INLINE_ int list_entry_is_first(struct list_head *entry, struct list_head *head)
+{
+	return head->next == entry;
+}
+
+/**
+ * list_splice - join two lists
+ * @list:	the new list to add.
+ * @head:	the place to add it in the first list.
+ */
+_INLINE_ void list_splice(struct list_head *list, struct list_head *head)
+{
+	struct list_head *first = list->next;
+
+	if (first != list) {
+		struct list_head *last = list->prev;
+		struct list_head *at = head->next;
+
+		first->prev = head;
+		head->next = first;
+
+		last->next = at;
+		at->prev = last;
+	}
+}
+
+/**
+ * list_entry - get the struct for this entry
+ * @ptr:	the &struct list_head pointer.
+ * @type:	the type of the struct this is embedded in.
+ * @member:	the name of the list_struct within the struct.
+ */
+#define list_entry(ptr, type, member)	container_of(ptr, type, member)
+
+#define list_first_entry(head, type, member) \
+	((head) && (head)->next != (head) ? list_entry((head)->next, type, member) : NULL)
+
+#define list_last_entry(head, type, member) \
+	((head) && (head)->prev != (head) ? list_entry((head)->prev, type, member) : NULL)
+
+/**
+ * list_for_each - iterate over elements in a list
+ * @pos:	the &struct list_head to use as a loop counter.
+ * @head:	the head for your list.
+ */
+#define list_for_each(pos, head) \
+	for (pos = (head)->next; pos != (head); pos = pos->next)
+
+/**
+ * list_for_each_backwardly - iterate over elements in a list in reverse
+ * @pos:	the &struct list_head to use as a loop counter.
+ * @head:	the head for your list.
+ */
+#define list_for_each_backwardly(pos, head) \
+	for (pos = (head)->prev; pos != (head); pos = pos->prev)
+
+/**
+ * list_for_each_safe - iterate over elements in a list, but don't dereference
+ *                      pos after the body is done (in case it is freed)
+ * @pos:	the &struct list_head to use as a loop counter.
+ * @pnext:	the &struct list_head to use as a pointer to the next item.
+ * @head:	the head for your list (not included in iteration).
+ */
+#define list_for_each_safe(pos, pnext, head) \
+	for (pos = (head)->next, pnext = pos->next; pos != (head); \
+	     pos = pnext, pnext = pos->next)
+
+/**
+ * list_free - remove all entries from list and call freefunc()
+ *             for each entry
+ * @head:       the head for your list
+ * @type:       the type of the struct this is embedded in.
+ * @member:     the name of the list_struct within the struct.
+ * @freefunc:   the list entry deallocator
+ */
+#define list_free(head, type, member, freefunc)				\
+	do {								\
+		struct list_head *__p, *__pnext;			\
+									\
+		list_for_each_safe (__p, __pnext, (head)) {		\
+			type *__elt = list_entry(__p, type, member);	\
+			list_del(__p);					\
+			freefunc(__elt);			\
+		}							\
+	} while (0)
+
+_INLINE_ size_t list_count_entries(struct list_head *head)
+{
+	struct list_head *pos;
+	size_t ct = 0;
+
+	list_for_each(pos, head)
+		ct++;
+
+	return ct;
+}
+
+#define MAX_LIST_LENGTH_BITS 20
+
+/*
+ * Returns a list organized in an intermediate format suited
+ * to chaining of merge() calls: null-terminated, no reserved or
+ * sentinel head node, "prev" links not maintained.
+ */
+_INLINE_ struct list_head *merge(int (*cmp)(struct list_head *a,
+					  struct list_head *b,
+					  void *data),
+			       void *data,
+			       struct list_head *a, struct list_head *b)
+{
+	struct list_head head, *tail = &head;
+
+	while (a && b) {
+		/* if equal, take 'a' -- important for sort stability */
+		if ((*cmp)(a, b, data) <= 0) {
+			tail->next = a;
+			a = a->next;
+		} else {
+			tail->next = b;
+			b = b->next;
+		}
+		tail = tail->next;
+	}
+	tail->next = a ? a : b;
+	return head.next;
+}
+
+/*
+ * Combine final list merge with restoration of standard doubly-linked
+ * list structure.  This approach duplicates code from merge(), but
+ * runs faster than the tidier alternatives of either a separate final
+ * prev-link restoration pass, or maintaining the prev links
+ * throughout.
+ */
+_INLINE_ void merge_and_restore_back_links(int (*cmp)(struct list_head *a,
+						    struct list_head *b,
+						    void *data),
+					 void *data,
+					 struct list_head *head,
+					 struct list_head *a, struct list_head *b)
+{
+	struct list_head *tail = head;
+
+	while (a && b) {
+		/* if equal, take 'a' -- important for sort stability */
+		if ((*cmp)(a, b, data) <= 0) {
+			tail->next = a;
+			a->prev = tail;
+			a = a->next;
+		} else {
+			tail->next = b;
+			b->prev = tail;
+			b = b->next;
+		}
+		tail = tail->next;
+	}
+	tail->next = a ? a : b;
+
+	do {
+		/*
+		 * In worst cases this loop may run many iterations.
+		 * Continue callbacks to the client even though no
+		 * element comparison is needed, so the client's cmp()
+		 * routine can invoke cond_resched() periodically.
+		 */
+		(*cmp)(tail->next, tail->next, data);
+
+		tail->next->prev = tail;
+		tail = tail->next;
+	} while (tail->next);
+
+	tail->next = head;
+	head->prev = tail;
+}
+
+
+/**
+ * list_sort - sort a list
+ * @head: the list to sort
+ * @cmp: the elements comparison function
+ *
+ * This function implements "merge sort", which has O(nlog(n))
+ * complexity.
+ *
+ * The comparison function @cmp must return a negative value if @a
+ * should sort before @b, and a positive value if @a should sort after
+ * @b. If @a and @b are equivalent, and their original relative
+ * ordering is to be preserved, @cmp must return 0.
+ */
+_INLINE_ void list_sort(struct list_head *head,
+			int (*cmp)(struct list_head *a,
+				   struct list_head *b,
+				   void *data),
+			void *data)
+{
+	struct list_head *part[MAX_LIST_LENGTH_BITS+1]; /* sorted partial lists
+							   -- last slot is a sentinel */
+	size_t lev;  /* index into part[] */
+	size_t max_lev = 0;
+	struct list_head *list;
+
+	if (list_empty(head))
+		return;
+
+	memset(part, 0, sizeof(part));
+
+	head->prev->next = NULL;
+	list = head->next;
+
+	while (list) {
+		struct list_head *cur = list;
+		list = list->next;
+		cur->next = NULL;
+
+		for (lev = 0; part[lev]; lev++) {
+			cur = merge(cmp, data, part[lev], cur);
+			part[lev] = NULL;
+		}
+		if (lev > max_lev) {
+			/* list passed to list_sort() too long for efficiency */
+			if (lev >= ARRAY_SIZE(part) - 1)
+				lev--;
+			max_lev = lev;
+		}
+		part[lev] = cur;
+	}
+
+	for (lev = 0; lev < max_lev; lev++)
+		if (part[lev])
+			list = merge(cmp, data, part[lev], list);
+
+	merge_and_restore_back_links(cmp, data, head, part[max_lev], list);
+}
+
+#undef _INLINE_
+
+#endif /* UTIL_LINUX_LIST_H */
diff --git a/include/logindefs.h b/include/logindefs.h
new file mode 100644
index 0000000..fa6aa2f
--- /dev/null
+++ b/include/logindefs.h
@@ -0,0 +1,35 @@
+/*
+ * Copyright (C) 2003, 2004, 2005 Thorsten Kukuk
+ * Author: Thorsten Kukuk <kukuk@suse.de>
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * 1. Redistributions of source code must retain any existing copyright
+ *    notice, and this entire permission notice in its entirety,
+ *    including the disclaimer of warranties.
+ *
+ * 2. Redistributions in binary form must reproduce all prior and current
+ *    copyright notices, this list of conditions, and the following
+ *    disclaimer in the documentation and/or other materials provided
+ *    with the distribution.
+ *
+ * 3. The name of any author may not be used to endorse or promote
+ *    products derived from this software without their specific prior
+ *   written permission.
+ */
+#ifndef UTIL_LINUX_LOGINDEFS_H
+#define UTIL_LINUX_LOGINDEFS_H
+
+extern void logindefs_load_file(const char *filename);
+extern void logindefs_set_loader(void (*loader)(void *data), void *data);
+extern int getlogindefs_bool(const char *name, int dflt);
+extern unsigned long getlogindefs_num(const char *name, unsigned long dflt);
+extern const char *getlogindefs_str(const char *name, const char *dflt);
+extern void free_getlogindefs_data(void);
+extern int logindefs_setenv(const char *name, const char *conf, const char *dflt);
+extern int effective_access(const char *path, int mode);
+extern int get_hushlogin_status(struct passwd *pwd, int force_check);
+
+#endif /* UTIL_LINUX_LOGINDEFS_H */
diff --git a/include/loopdev.h b/include/loopdev.h
new file mode 100644
index 0000000..903adc4
--- /dev/null
+++ b/include/loopdev.h
@@ -0,0 +1,226 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_LOOPDEV_H
+#define UTIL_LINUX_LOOPDEV_H
+
+#include "sysfs.h"
+
+/*
+ * loop_info.lo_encrypt_type
+ */
+#define LO_CRYPT_NONE	0
+#define LO_CRYPT_XOR	1
+#define LO_CRYPT_DES	2
+#define LO_CRYPT_CRYPTOAPI 18
+
+#define LOOP_SET_FD		0x4C00
+#define LOOP_CLR_FD		0x4C01
+/*
+ * Obsolete (kernel < 2.6)
+ *
+ * #define LOOP_SET_STATUS	0x4C02
+ * #define LOOP_GET_STATUS	0x4C03
+ */
+#define LOOP_SET_STATUS64	0x4C04
+#define LOOP_GET_STATUS64	0x4C05
+/* #define LOOP_CHANGE_FD	0x4C06 */
+#define LOOP_SET_CAPACITY	0x4C07
+#define LOOP_SET_DIRECT_IO	0x4C08
+#define LOOP_SET_BLOCK_SIZE	0x4C09
+
+/* /dev/loop-control interface */
+#ifndef LOOP_CTL_ADD
+# define LOOP_CTL_ADD		0x4C80
+# define LOOP_CTL_REMOVE	0x4C81
+# define LOOP_CTL_GET_FREE	0x4C82
+#endif
+
+/*
+ * loop_info.lo_flags
+ */
+enum {
+	LO_FLAGS_READ_ONLY  = 1,
+	LO_FLAGS_USE_AOPS   = 2,
+	LO_FLAGS_AUTOCLEAR  = 4,	/* kernel >= 2.6.25 */
+	LO_FLAGS_PARTSCAN   = 8,	/* kernel >= 3.2 */
+	LO_FLAGS_DIRECT_IO  = 16,	/* kernel >= 4.2 */
+};
+
+#define LO_NAME_SIZE	64
+#define LO_KEY_SIZE	32
+
+/*
+ * Linux LOOP_{SET,GET}_STATUS64 ioctl struct
+ */
+struct loop_info64 {
+	uint64_t	lo_device;
+	uint64_t	lo_inode;
+	uint64_t	lo_rdevice;
+	uint64_t	lo_offset;
+	uint64_t	lo_sizelimit; /* bytes, 0 == max available */
+	uint32_t	lo_number;
+	uint32_t	lo_encrypt_type;
+	uint32_t	lo_encrypt_key_size;
+	uint32_t	lo_flags;
+	uint8_t		lo_file_name[LO_NAME_SIZE];
+	uint8_t		lo_crypt_name[LO_NAME_SIZE];
+	uint8_t		lo_encrypt_key[LO_KEY_SIZE];
+	uint64_t	lo_init[2];
+};
+
+#ifndef LOOP_CONFIGURE
+/*
+ * Since Linux v5.8-rc1 (commit 3448914e8cc550ba792d4ccc74471d1ca4293aae)
+ */
+# define LOOP_CONFIGURE		0x4C0A
+struct loop_config {
+  uint32_t fd;
+  uint32_t block_size;
+  struct loop_info64 info;
+  uint64_t __reserved[8];
+};
+#endif
+
+#define LOOPDEV_MAJOR		7	/* loop major number */
+#define LOOPDEV_DEFAULT_NNODES	8	/* default number of loop devices */
+
+struct loopdev_iter {
+	FILE		*proc;		/* /proc/partitions */
+	DIR		*sysblock;	/* /sys/block */
+	int		ncur;		/* current position */
+	int		*minors;	/* ary of minor numbers (when scan whole /dev) */
+	int		nminors;	/* number of items in *minors */
+	int		ct_perm;	/* count permission problems */
+	int		ct_succ;	/* count number of detected devices */
+
+	unsigned int	done:1;		/* scanning done */
+	unsigned int	default_check:1;/* check first LOOPDEV_NLOOPS */
+	int		flags;		/* LOOPITER_FL_* flags */
+};
+
+enum {
+	LOOPITER_FL_FREE	= (1 << 0),
+	LOOPITER_FL_USED	= (1 << 1)
+};
+
+/*
+ * handler for work with loop devices
+ */
+struct loopdev_cxt {
+	char		device[128];	/* device path (e.g. /dev/loop<N>) */
+	char		*filename;	/* backing file for loopcxt_set_... */
+	int		fd;		/* open(/dev/looo<N>) */
+	int		mode;		/* fd mode O_{RDONLY,RDWR} */
+	uint64_t	blocksize;	/* used by loopcxt_setup_device() */
+
+	int		flags;		/* LOOPDEV_FL_* flags */
+	unsigned int	has_info:1;	/* .info contains data */
+	unsigned int	extra_check:1;	/* unusual stuff for iterator */
+	unsigned int	info_failed:1;	/* LOOP_GET_STATUS ioctl failed */
+	unsigned int    control_ok:1;	/* /dev/loop-control success */
+
+	struct path_cxt		*sysfs; /* pointer to /sys/dev/block/<maj:min>/ */
+	struct loop_config 	config;	/* for GET/SET ioctl */
+	struct loopdev_iter	iter;	/* scans /sys or /dev for used/free devices */
+};
+
+#define UL_LOOPDEVCXT_EMPTY { .fd = -1  }
+
+/*
+ * loopdev_cxt.flags
+ */
+enum {
+	LOOPDEV_FL_RDONLY	= (1 << 0),	/* open(/dev/loop) mode; default */
+	LOOPDEV_FL_RDWR		= (1 << 1),	/* necessary for loop setup only */
+	LOOPDEV_FL_OFFSET	= (1 << 4),
+	LOOPDEV_FL_NOSYSFS	= (1 << 5),
+	LOOPDEV_FL_NOIOCTL	= (1 << 6),
+	LOOPDEV_FL_DEVSUBDIR	= (1 << 7),
+	LOOPDEV_FL_CONTROL	= (1 << 8),	/* system with /dev/loop-control */
+	LOOPDEV_FL_SIZELIMIT	= (1 << 9)
+};
+
+/*
+ * High-level
+ */
+extern int loopmod_supports_partscan(void);
+
+extern int is_loopdev(const char *device);
+extern int loopdev_is_autoclear(const char *device);
+
+extern char *loopdev_get_backing_file(const char *device);
+extern int loopdev_has_backing_file(const char *device);
+extern int loopdev_is_used(const char *device, const char *filename,
+			   uint64_t offset, uint64_t sizelimit, int flags);
+extern char *loopdev_find_by_backing_file(const char *filename,
+				uint64_t offset, uint64_t sizelimit, int flags);
+extern int loopcxt_find_unused(struct loopdev_cxt *lc);
+extern int loopdev_delete(const char *device);
+extern int loopdev_count_by_backing_file(const char *filename, char **loopdev);
+
+/*
+ * Low-level
+ */
+extern int loopcxt_init(struct loopdev_cxt *lc, int flags)
+				__attribute__ ((warn_unused_result));
+extern void loopcxt_deinit(struct loopdev_cxt *lc);
+
+extern int loopcxt_set_device(struct loopdev_cxt *lc, const char *device)
+				__attribute__ ((warn_unused_result));
+extern int loopcxt_has_device(struct loopdev_cxt *lc);
+extern int loopcxt_add_device(struct loopdev_cxt *lc);
+extern char *loopcxt_strdup_device(struct loopdev_cxt *lc);
+extern const char *loopcxt_get_device(struct loopdev_cxt *lc);
+extern struct loop_info64 *loopcxt_get_info(struct loopdev_cxt *lc);
+
+extern int loopcxt_get_fd(struct loopdev_cxt *lc);
+extern int loopcxt_set_fd(struct loopdev_cxt *lc, int fd, int mode);
+
+extern int loopcxt_init_iterator(struct loopdev_cxt *lc, int flags);
+extern int loopcxt_deinit_iterator(struct loopdev_cxt *lc);
+extern int loopcxt_next(struct loopdev_cxt *lc);
+
+extern int loopcxt_setup_device(struct loopdev_cxt *lc);
+extern int loopcxt_delete_device(struct loopdev_cxt *lc);
+
+extern int loopcxt_ioctl_status(struct loopdev_cxt *lc);
+extern int loopcxt_ioctl_capacity(struct loopdev_cxt *lc);
+extern int loopcxt_ioctl_dio(struct loopdev_cxt *lc, unsigned long use_dio);
+extern int loopcxt_ioctl_blocksize(struct loopdev_cxt *lc, uint64_t blocksize);
+
+int loopcxt_set_offset(struct loopdev_cxt *lc, uint64_t offset);
+int loopcxt_set_sizelimit(struct loopdev_cxt *lc, uint64_t sizelimit);
+int loopcxt_set_blocksize(struct loopdev_cxt *lc, uint64_t blocksize);
+int loopcxt_set_flags(struct loopdev_cxt *lc, uint32_t flags);
+int loopcxt_set_backing_file(struct loopdev_cxt *lc, const char *filename);
+
+extern char *loopcxt_get_backing_file(struct loopdev_cxt *lc);
+extern int loopcxt_get_backing_devno(struct loopdev_cxt *lc, dev_t *devno);
+extern int loopcxt_get_backing_inode(struct loopdev_cxt *lc, ino_t *ino);
+extern int loopcxt_get_offset(struct loopdev_cxt *lc, uint64_t *offset);
+extern int loopcxt_get_blocksize(struct loopdev_cxt *lc, uint64_t *blocksize);
+extern int loopcxt_get_sizelimit(struct loopdev_cxt *lc, uint64_t *size);
+extern int loopcxt_get_encrypt_type(struct loopdev_cxt *lc, uint32_t *type);
+extern const char *loopcxt_get_crypt_name(struct loopdev_cxt *lc);
+extern int loopcxt_is_autoclear(struct loopdev_cxt *lc);
+extern int loopcxt_is_readonly(struct loopdev_cxt *lc);
+extern int loopcxt_is_dio(struct loopdev_cxt *lc);
+extern int loopcxt_is_partscan(struct loopdev_cxt *lc);
+extern int loopcxt_find_by_backing_file(struct loopdev_cxt *lc,
+				const char *filename,
+				uint64_t offset, uint64_t sizelimit,
+				int flags);
+extern int loopcxt_find_overlap(struct loopdev_cxt *lc,
+				const char *filename,
+				uint64_t offset, uint64_t sizelimit);
+
+extern int loopcxt_is_used(struct loopdev_cxt *lc,
+                    struct stat *st,
+                    const char *backing_file,
+                    uint64_t offset,
+                    uint64_t sizelimit,
+                    int flags);
+
+#endif /* UTIL_LINUX_LOOPDEV_H */
diff --git a/include/mangle.h b/include/mangle.h
new file mode 100644
index 0000000..5939691
--- /dev/null
+++ b/include/mangle.h
@@ -0,0 +1,32 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_MANGLE_H
+#define UTIL_LINUX_MANGLE_H
+
+/*
+ * Functions for \oct encoding used in mtab/fstab/swaps/etc.
+ */
+
+extern char *mangle(const char *s);
+
+extern void unmangle_to_buffer(const char *s, char *buf, size_t len);
+extern size_t unhexmangle_to_buffer(const char *s, char *buf, size_t len);
+
+extern char *unmangle(const char *s, const char **end);
+
+static inline void unmangle_string(char *s)
+{
+	if (s)
+		unmangle_to_buffer(s, s, strlen(s) + 1);
+}
+
+static inline void unhexmangle_string(char *s)
+{
+	if (s)
+		unhexmangle_to_buffer(s, s, strlen(s) + 1);
+}
+
+#endif /* UTIL_LINUX_MANGLE_H */
+
diff --git a/include/match.h b/include/match.h
new file mode 100644
index 0000000..94440c2
--- /dev/null
+++ b/include/match.h
@@ -0,0 +1,12 @@
+/*
+ * Copyright (C) 2011 Karel Zak <kzak@redhat.com>
+ *
+ * This file may be redistributed under the terms of the
+ * GNU Lesser General Public License.
+ */
+#ifndef UTIL_LINUX_MATCH_H
+#define UTIL_LINUX_MATCH_H
+
+extern int match_fstype(const char *type, const char *pattern);
+
+#endif /* UTIL_LINUX_MATCH_H */
diff --git a/include/mbsalign.h b/include/mbsalign.h
new file mode 100644
index 0000000..6e56755
--- /dev/null
+++ b/include/mbsalign.h
@@ -0,0 +1,62 @@
+/*
+ * SPDX-License-Identifier: LGPL-2.1-or-later
+ *
+ * Align/Truncate a string in a given screen width
+ * Copyright (C) 2009-2010 Free Software Foundation, Inc.
+ * Copyright (C) 2010-2013 Karel Zak <kzak@redhat.com>
+ *
+ * This program is free software: you can redistribute it and/or modify it
+ * under the terms of the GNU Lesser General Public License as published by the
+ * Free Software Foundation, either version 2.1 of the License, or (at your
+ * option) any later version.
+ */
+#ifndef UTIL_LINUX_MBSALIGN_H
+# define UTIL_LINUX_MBSALIGN_H
+# include <stddef.h>
+
+typedef enum { MBS_ALIGN_LEFT, MBS_ALIGN_RIGHT, MBS_ALIGN_CENTER } mbs_align_t;
+
+enum {
+  /* Use unibyte mode for invalid multibyte strings or
+     or when heap memory is exhausted.  */
+  MBA_UNIBYTE_FALLBACK = 0x0001,
+
+#if 0 /* Other possible options.  */
+  /* Skip invalid multibyte chars rather than failing  */
+  MBA_IGNORE_INVALID   = 0x0002,
+
+  /* Align multibyte strings using "figure space" (\u2007)  */
+  MBA_USE_FIGURE_SPACE = 0x0004,
+
+  /* Don't add any padding  */
+  MBA_TRUNCATE_ONLY    = 0x0008,
+
+  /* Don't truncate  */
+  MBA_PAD_ONLY         = 0x0010,
+#endif
+};
+
+extern size_t mbs_truncate(char *str, size_t *width);
+
+extern size_t mbsalign (const char *src, char *dest,
+			size_t dest_size,  size_t *width,
+			mbs_align_t align, int flags);
+
+extern size_t mbsalign_with_padding (const char *src, char *dest, size_t dest_size,
+	               size_t *width, mbs_align_t align, int flags,
+		       int padchar);
+
+extern size_t mbs_safe_nwidth(const char *buf, size_t bufsz, size_t *sz);
+extern size_t mbs_safe_width(const char *s);
+
+extern size_t mbs_nwidth(const char *buf, size_t bufsz);
+extern size_t mbs_width(const char *s);
+
+extern char *mbs_safe_encode(const char *s, size_t *width);
+extern char *mbs_safe_encode_to_buffer(const char *s, size_t *width, char *buf, const char *safechars);
+extern size_t mbs_safe_encode_size(size_t bytes);
+
+extern char *mbs_invalid_encode(const char *s, size_t *width);
+extern char *mbs_invalid_encode_to_buffer(const char *s, size_t *width, char *buf);
+
+#endif /* UTIL_LINUX_MBSALIGN_H */
diff --git a/include/mbsedit.h b/include/mbsedit.h
new file mode 100644
index 0000000..ed7960f
--- /dev/null
+++ b/include/mbsedit.h
@@ -0,0 +1,38 @@
+/*
+ * SPDX-License-Identifier: LGPL-2.1-or-later
+ *
+ * This file may be distributed under the terms of the
+ * GNU Lesser General Public License.
+ */
+#ifndef UTIL_LINUX_MBSEDIT_H
+# define UTIL_LINUX_MBSEDIT_H
+
+#include "mbsalign.h"
+#include "widechar.h"
+
+struct mbs_editor {
+	char	*buf;		/* buffer */
+	size_t	max_bytes;	/* size of the buffer */
+	size_t	max_cells;	/* maximal allowed number of cells */
+	size_t	cur_cells;	/* number of cells to print the buffer */
+	size_t  cur_bytes;	/* number of chars in bytes */
+	size_t  cursor;		/* cursor position in bytes */
+	size_t  cursor_cells;	/* cursor position in cells */
+};
+
+enum {
+	MBS_EDIT_LEFT,
+	MBS_EDIT_RIGHT,
+	MBS_EDIT_END,
+	MBS_EDIT_HOME
+};
+
+struct mbs_editor *mbs_new_edit(char *buf, size_t bufsz, size_t ncells);
+char *mbs_free_edit(struct mbs_editor *edit);
+
+int mbs_edit_goto(struct mbs_editor *edit, int where);
+int mbs_edit_delete(struct mbs_editor *edit);
+int mbs_edit_backspace(struct mbs_editor *edit);
+int mbs_edit_insert(struct mbs_editor *edit, wint_t c);
+
+#endif /* UTIL_LINUX_MBSEDIT_H */
diff --git a/include/md5.h b/include/md5.h
new file mode 100644
index 0000000..791b532
--- /dev/null
+++ b/include/md5.h
@@ -0,0 +1,28 @@
+/*
+ * No copyright is claimed. This code is in the public domain; do with it what
+ * you wish.
+ */
+#ifndef UTIL_LINUX_MD5_H
+#define UTIL_LINUX_MD5_H
+
+#include <stdint.h>
+
+#define UL_MD5LENGTH 16
+
+struct UL_MD5Context {
+	uint32_t buf[4];
+	uint32_t bits[2];
+	unsigned char in[64];
+};
+
+void ul_MD5Init(struct UL_MD5Context *ctx);
+void ul_MD5Update(struct UL_MD5Context *ctx, unsigned char const *buf, unsigned len);
+void ul_MD5Final(unsigned char digest[UL_MD5LENGTH], struct UL_MD5Context *ctx);
+void ul_MD5Transform(uint32_t buf[4], uint32_t const in[16]);
+
+/*
+ * This is needed to make RSAREF happy on some MS-DOS compilers.
+ */
+typedef struct UL_MD5Context UL_MD5_CTX;
+
+#endif /* !UTIL_LINUX_MD5_H */
diff --git a/include/meson.build b/include/meson.build
new file mode 100644
index 0000000..6162ff0
--- /dev/null
+++ b/include/meson.build
@@ -0,0 +1,3 @@
+dir_include = include_directories('.')
+
+list_h = files('list.h')
diff --git a/include/minix.h b/include/minix.h
new file mode 100644
index 0000000..571d06e
--- /dev/null
+++ b/include/minix.h
@@ -0,0 +1,89 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_MINIX_H
+#define UTIL_LINUX_MINIX_H
+
+#include <stdint.h>
+
+struct minix_inode {
+	uint16_t i_mode;
+	uint16_t i_uid;
+	uint32_t i_size;
+	uint32_t i_time;
+	uint8_t  i_gid;
+	uint8_t  i_nlinks;
+	uint16_t i_zone[9];
+};
+
+struct minix2_inode {
+	uint16_t i_mode;
+	uint16_t i_nlinks;
+	uint16_t i_uid;
+	uint16_t i_gid;
+	uint32_t i_size;
+	uint32_t i_atime;
+	uint32_t i_mtime;
+	uint32_t i_ctime;
+	uint32_t i_zone[10];
+};
+
+struct minix_super_block {
+	uint16_t s_ninodes;
+	uint16_t s_nzones;
+	uint16_t s_imap_blocks;
+	uint16_t s_zmap_blocks;
+	uint16_t s_firstdatazone;
+	uint16_t s_log_zone_size;
+	uint32_t s_max_size;
+	uint16_t s_magic;
+	uint16_t s_state;
+	uint32_t s_zones;
+};
+
+/* V3 minix super-block data on disk */
+struct minix3_super_block {
+	uint32_t s_ninodes;
+	uint16_t s_pad0;
+	uint16_t s_imap_blocks;
+	uint16_t s_zmap_blocks;
+	uint16_t s_firstdatazone;
+	uint16_t s_log_zone_size;
+	uint16_t s_pad1;
+	uint32_t s_max_size;
+	uint32_t s_zones;
+	uint16_t s_magic;
+	uint16_t s_pad2;
+	uint16_t s_blocksize;
+	uint8_t  s_disk_version;
+};
+
+/*
+ * Minix subpartitions are always within primary dos partition.
+ */
+#define MINIX_MAXPARTITIONS  4
+
+#define MINIX_BLOCK_SIZE_BITS 10
+#define MINIX_BLOCK_SIZE     (1 << MINIX_BLOCK_SIZE_BITS)
+
+#define MINIX_NAME_MAX       255             /* # chars in a file name */
+#define MINIX_MAX_INODES     65535
+
+#define MINIX_INODES_PER_BLOCK ((MINIX_BLOCK_SIZE)/(sizeof (struct minix_inode)))
+#define MINIX2_INODES_PER_BLOCK ((MINIX_BLOCK_SIZE)/(sizeof (struct minix2_inode)))
+
+/* minix_super_block.s_state */
+#define MINIX_VALID_FS       0x0001          /* Clean fs. */
+#define MINIX_ERROR_FS       0x0002          /* fs has errors. */
+
+
+#define MINIX_SUPER_MAGIC    0x137F          /* minix V1 fs, 14 char names */
+#define MINIX_SUPER_MAGIC2   0x138F          /* minix V1 fs, 30 char names */
+
+#define MINIX2_SUPER_MAGIC   0x2468	     /* minix V2 fs, 14 char names */
+#define MINIX2_SUPER_MAGIC2  0x2478	     /* minix V2 fs, 30 char names */
+
+#define MINIX3_SUPER_MAGIC   0x4d5a          /* minix V3 fs (60 char names) */
+
+#endif /* UTIL_LINUX_MINIX_H */
diff --git a/include/monotonic.h b/include/monotonic.h
new file mode 100644
index 0000000..5c5c2c1
--- /dev/null
+++ b/include/monotonic.h
@@ -0,0 +1,24 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_MONOTONIC_H
+#define UTIL_LINUX_MONOTONIC_H
+
+# ifdef CLOCK_MONOTONIC_RAW
+#  define UL_CLOCK_MONOTONIC	CLOCK_MONOTONIC_RAW
+# else
+#  define UL_CLOCK_MONOTONIC	CLOCK_MONOTONIC
+# endif
+
+#include <sys/time.h>
+
+#include "timeutils.h"
+
+extern int get_boot_time(struct timeval *boot_time);
+
+extern usec_t get_suspended_time(void);
+
+extern int gettime_monotonic(struct timeval *tv);
+
+#endif /* UTIL_LINUX_MONOTONIC_H */
diff --git a/include/mount-api-utils.h b/include/mount-api-utils.h
new file mode 100644
index 0000000..8ae546b
--- /dev/null
+++ b/include/mount-api-utils.h
@@ -0,0 +1,208 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_MOUNT_API_UTILS
+#define UTIL_LINUX_MOUNT_API_UTILS
+
+#if defined(HAVE_MOUNTFD_API) && defined(HAVE_LINUX_MOUNT_H)
+
+#include <sys/syscall.h>
+#include <linux/mount.h>
+
+/* Accepted by both open_tree() and mount_setattr(). */
+#ifndef AT_RECURSIVE
+# define AT_RECURSIVE 0x8000
+#endif
+
+#ifndef OPEN_TREE_CLONE
+# define OPEN_TREE_CLONE 1
+#endif
+
+#ifndef OPEN_TREE_CLOEXEC
+# define OPEN_TREE_CLOEXEC O_CLOEXEC
+#endif
+
+#if !defined(HAVE_OPEN_TREE) && defined(SYS_open_tree)
+static inline int open_tree(int dfd, const char *filename, unsigned int flags)
+{
+	return syscall(SYS_open_tree, dfd, filename, flags);
+}
+#endif
+
+#ifndef MOVE_MOUNT_F_SYMLINKS
+# define MOVE_MOUNT_F_SYMLINKS 0x00000001 /* Follow symlinks on from path */
+#endif
+
+#ifndef MOVE_MOUNT_F_AUTOMOUNTS
+# define MOVE_MOUNT_F_AUTOMOUNTS 0x00000002 /* Follow automounts on from path */
+#endif
+
+#ifndef MOVE_MOUNT_F_EMPTY_PATH
+# define MOVE_MOUNT_F_EMPTY_PATH 0x00000004 /* Empty from path permitted */
+#endif
+
+#ifndef MOVE_MOUNT_T_SYMLINKS
+# define MOVE_MOUNT_T_SYMLINKS 0x00000010 /* Follow symlinks on to path */
+#endif
+
+#ifndef MOVE_MOUNT_T_AUTOMOUNTS
+# define MOVE_MOUNT_T_AUTOMOUNTS 0x00000020 /* Follow automounts on to path */
+#endif
+
+#ifndef MOVE_MOUNT_T_EMPTY_PATH
+# define MOVE_MOUNT_T_EMPTY_PATH 0x00000040 /* Empty to path permitted */
+#endif
+
+#ifndef MOVE_MOUNT_SET_GROUP
+# define MOVE_MOUNT_SET_GROUP	0x00000100 /* Set sharing group instead */
+#endif
+
+#ifndef MOVE_MOUNT__MASK
+# define MOVE_MOUNT__MASK 0x00000077
+#endif
+
+#if !defined(HAVE_MOVE_MOUNT) && defined(SYS_move_mount)
+static inline int move_mount(int from_dfd, const char *from_pathname, int to_dfd,
+			     const char *to_pathname, unsigned int flags)
+{
+	return syscall(SYS_move_mount, from_dfd, from_pathname, to_dfd,
+		       to_pathname, flags);
+}
+#endif
+
+#ifndef MOUNT_ATTR_RDONLY
+# define MOUNT_ATTR_RDONLY 0x00000001
+#endif
+
+#ifndef MOUNT_ATTR_NOSUID
+# define MOUNT_ATTR_NOSUID 0x00000002
+#endif
+
+#ifndef MOUNT_ATTR_NODEV
+# define MOUNT_ATTR_NODEV 0x00000004
+#endif
+
+#ifndef MOUNT_ATTR_NOEXEC
+# define MOUNT_ATTR_NOEXEC 0x00000008
+#endif
+
+#ifndef MOUNT_ATTR__ATIME
+# define MOUNT_ATTR__ATIME 0x00000070
+#endif
+
+#ifndef MOUNT_ATTR_RELATIME
+# define MOUNT_ATTR_RELATIME 0x00000000
+#endif
+
+#ifndef MOUNT_ATTR_NOATIME
+# define MOUNT_ATTR_NOATIME 0x00000010
+#endif
+
+#ifndef MOUNT_ATTR_STRICTATIME
+# define MOUNT_ATTR_STRICTATIME 0x00000020
+#endif
+
+#ifndef MOUNT_ATTR_NODIRATIME
+# define MOUNT_ATTR_NODIRATIME 0x00000080
+#endif
+
+#ifndef MOUNT_ATTR_IDMAP
+# define MOUNT_ATTR_IDMAP 0x00100000
+#endif
+
+#ifndef MOUNT_ATTR_NOSYMFOLLOW
+# define MOUNT_ATTR_NOSYMFOLLOW 0x00200000
+#endif
+
+#ifndef HAVE_STRUCT_MOUNT_ATTR
+# ifndef MOUNT_ATTR_SIZE_VER0 /* For case mount.h comes from a place invisible for autotools/meson */
+# include <inttypes.h>
+struct mount_attr {
+	uint64_t attr_set;
+	uint64_t attr_clr;
+	uint64_t propagation;
+	uint64_t userns_fd;
+};
+# endif
+#endif
+
+#if !defined(HAVE_MOUNT_SETATTR) && defined(SYS_mount_setattr)
+static inline int mount_setattr(int dfd, const char *path, unsigned int flags,
+				struct mount_attr *attr, size_t size)
+{
+	return syscall(SYS_mount_setattr, dfd, path, flags, attr, size);
+}
+#endif
+
+#ifndef HAVE_ENUM_FSCONFIG_COMMAND
+# ifndef FSOPEN_CLOEXEC /* For case mount.h comes from a place invisible for autotools/meson */
+enum fsconfig_command {
+	FSCONFIG_SET_FLAG	= 0,	/* Set parameter, supplying no value */
+	FSCONFIG_SET_STRING	= 1,	/* Set parameter, supplying a string value */
+	FSCONFIG_SET_BINARY	= 2,	/* Set parameter, supplying a binary blob value */
+	FSCONFIG_SET_PATH	= 3,	/* Set parameter, supplying an object by path */
+	FSCONFIG_SET_PATH_EMPTY	= 4,	/* Set parameter, supplying an object by (empty) path */
+	FSCONFIG_SET_FD		= 5,	/* Set parameter, supplying an object by fd */
+	FSCONFIG_CMD_CREATE	= 6,	/* Invoke superblock creation */
+	FSCONFIG_CMD_RECONFIGURE = 7,	/* Invoke superblock reconfiguration */
+};
+# endif
+#endif
+
+#if !defined(HAVE_FSCONFIG) && defined(SYS_fsconfig)
+static inline int fsconfig(int fd, unsigned int cmd, const char *key,
+                    const void *value, int aux)
+{
+        return syscall(SYS_fsconfig, fd, cmd, key, value, aux);
+}
+#endif
+
+#ifndef FSOPEN_CLOEXEC
+# define FSOPEN_CLOEXEC          0x00000001
+#endif
+
+#if !defined(HAVE_FSOPEN) && defined(SYS_fsopen)
+static inline int fsopen(const char *fsname, unsigned int flags)
+{
+        return syscall(SYS_fsopen, fsname, flags);
+}
+#endif
+
+#ifndef FSMOUNT_CLOEXEC
+# define FSMOUNT_CLOEXEC         0x00000001
+#endif
+
+#if !defined(HAVE_FSMOUNT) && defined(SYS_fsmount)
+static inline int fsmount(int fd, unsigned int flags, unsigned int mount_attrs)
+{
+        return syscall(SYS_fsmount, fd, flags, mount_attrs);
+}
+#endif
+
+#ifndef FSPICK_CLOEXEC
+# define FSPICK_CLOEXEC          0x00000001
+#endif
+
+#ifndef FSPICK_SYMLINK_NOFOLLOW
+# define FSPICK_SYMLINK_NOFOLLOW 0x00000002
+#endif
+
+#ifndef FSPICK_NO_AUTOMOUNT
+# define FSPICK_NO_AUTOMOUNT     0x00000004
+#endif
+
+#ifdef FSPICK_EMPTY_PATH
+# define FSPICK_EMPTY_PATH       0x00000008
+#endif
+
+#if !defined(HAVE_FSPICK) && defined(SYS_fspick)
+static inline int fspick(int dfd, const char *pathname, unsigned int flags)
+{
+        return syscall(SYS_fspick, dfd, pathname, flags);
+}
+#endif
+
+#endif /* HAVE_MOUNTFD_API && HAVE_LINUX_MOUNT_H */
+#endif /* UTIL_LINUX_MOUNT_API_UTILS */
+
diff --git a/include/namespace.h b/include/namespace.h
new file mode 100644
index 0000000..2d0a56e
--- /dev/null
+++ b/include/namespace.h
@@ -0,0 +1,56 @@
+
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ *
+ * Compat code so unshare and setns can be used with older libcs
+ */
+#ifndef UTIL_LINUX_NAMESPACE_H
+# define UTIL_LINUX_NAMESPACE_H
+
+# include <sched.h>
+
+# ifndef CLONE_NEWNS
+#  define CLONE_NEWNS 0x00020000
+# endif
+# ifndef CLONE_NEWCGROUP
+#  define CLONE_NEWCGROUP 0x02000000
+# endif
+# ifndef CLONE_NEWUTS
+#  define CLONE_NEWUTS 0x04000000
+# endif
+# ifndef CLONE_NEWIPC
+#  define CLONE_NEWIPC 0x08000000
+# endif
+# ifndef CLONE_NEWNET
+#  define CLONE_NEWNET 0x40000000
+# endif
+# ifndef CLONE_NEWUSER
+#  define CLONE_NEWUSER 0x10000000
+# endif
+# ifndef CLONE_NEWPID
+#  define CLONE_NEWPID 0x20000000
+# endif
+# ifndef CLONE_NEWTIME
+#  define CLONE_NEWTIME 0x00000080
+# endif
+
+# if !defined(HAVE_UNSHARE) || !defined(HAVE_SETNS)
+#  include <sys/syscall.h>
+# endif
+
+# if !defined(HAVE_UNSHARE) && defined(SYS_unshare)
+static inline int unshare(int flags)
+{
+	return syscall(SYS_unshare, flags);
+}
+# endif
+
+# if !defined(HAVE_SETNS) && defined(SYS_setns)
+static inline int setns(int fd, int nstype)
+{
+	return syscall(SYS_setns, fd, nstype);
+}
+# endif
+
+#endif	/* UTIL_LINUX_NAMESPACE_H */
diff --git a/include/nls.h b/include/nls.h
new file mode 100644
index 0000000..377c5a1
--- /dev/null
+++ b/include/nls.h
@@ -0,0 +1,157 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_NLS_H
+#define UTIL_LINUX_NLS_H
+
+#ifndef LOCALEDIR
+#define LOCALEDIR "/usr/share/locale"
+#endif
+
+#ifdef HAVE_LOCALE_H
+# include <locale.h>
+#else
+# undef setlocale
+# define setlocale(Category, Locale) /* empty */
+struct lconv
+{
+	char *decimal_point;
+};
+# undef localeconv
+# define localeconv() NULL
+#endif
+
+
+#ifdef ENABLE_NLS
+# include <libintl.h>
+/*
+ * For NLS support in the public shared libraries we have to specify text
+ * domain name to be independent on the main program. For this purpose define
+ * UL_TEXTDOMAIN_EXPLICIT before you include nls.h to your shared library code.
+ */
+# ifdef UL_TEXTDOMAIN_EXPLICIT
+#  define _(Text) dgettext (UL_TEXTDOMAIN_EXPLICIT, Text)
+# else
+#  define _(Text) gettext (Text)
+# endif
+# ifdef gettext_noop
+#  define N_(String) gettext_noop (String)
+# else
+#  define N_(String) (String)
+# endif
+# define P_(Singular, Plural, n) ngettext (Singular, Plural, n)
+#else
+# undef bindtextdomain
+# define bindtextdomain(Domain, Directory) /* empty */
+# undef textdomain
+# define textdomain(Domain) /* empty */
+# define _(Text) (Text)
+# define N_(Text) (Text)
+# define P_(Singular, Plural, n) ((n) == 1 ? (Singular) : (Plural))
+#endif /* ENABLE_NLS */
+
+#ifdef HAVE_LANGINFO_H
+# include <langinfo.h>
+#else
+
+typedef int nl_item;
+extern char *langinfo_fallback(nl_item item);
+
+# define nl_langinfo	langinfo_fallback
+
+enum {
+	CODESET = 1,
+	RADIXCHAR,
+	THOUSEP,
+	D_T_FMT,
+	D_FMT,
+	T_FMT,
+	T_FMT_AMPM,
+	AM_STR,
+	PM_STR,
+
+	DAY_1,
+	DAY_2,
+	DAY_3,
+	DAY_4,
+	DAY_5,
+	DAY_6,
+	DAY_7,
+
+	ABDAY_1,
+	ABDAY_2,
+	ABDAY_3,
+	ABDAY_4,
+	ABDAY_5,
+	ABDAY_6,
+	ABDAY_7,
+
+	MON_1,
+	MON_2,
+	MON_3,
+	MON_4,
+	MON_5,
+	MON_6,
+	MON_7,
+	MON_8,
+	MON_9,
+	MON_10,
+	MON_11,
+	MON_12,
+
+	ABMON_1,
+	ABMON_2,
+	ABMON_3,
+	ABMON_4,
+	ABMON_5,
+	ABMON_6,
+	ABMON_7,
+	ABMON_8,
+	ABMON_9,
+	ABMON_10,
+	ABMON_11,
+	ABMON_12,
+
+	ERA_D_FMT,
+	ERA_D_T_FMT,
+	ERA_T_FMT,
+	ALT_DIGITS,
+	CRNCYSTR,
+	YESEXPR,
+	NOEXPR
+};
+
+#endif /* !HAVE_LANGINFO_H */
+
+#ifndef HAVE_LANGINFO_ALTMON
+# define ALTMON_1 MON_1
+# define ALTMON_2 MON_2
+# define ALTMON_3 MON_3
+# define ALTMON_4 MON_4
+# define ALTMON_5 MON_5
+# define ALTMON_6 MON_6
+# define ALTMON_7 MON_7
+# define ALTMON_8 MON_8
+# define ALTMON_9 MON_9
+# define ALTMON_10 MON_10
+# define ALTMON_11 MON_11
+# define ALTMON_12 MON_12
+#endif /* !HAVE_LANGINFO_ALTMON */
+
+#ifndef HAVE_LANGINFO_NL_ABALTMON
+# define _NL_ABALTMON_1 ABMON_1
+# define _NL_ABALTMON_2 ABMON_2
+# define _NL_ABALTMON_3 ABMON_3
+# define _NL_ABALTMON_4 ABMON_4
+# define _NL_ABALTMON_5 ABMON_5
+# define _NL_ABALTMON_6 ABMON_6
+# define _NL_ABALTMON_7 ABMON_7
+# define _NL_ABALTMON_8 ABMON_8
+# define _NL_ABALTMON_9 ABMON_9
+# define _NL_ABALTMON_10 ABMON_10
+# define _NL_ABALTMON_11 ABMON_11
+# define _NL_ABALTMON_12 ABMON_12
+#endif /* !HAVE_LANGINFO_NL_ABALTMON */
+
+#endif /* UTIL_LINUX_NLS_H */
diff --git a/include/optutils.h b/include/optutils.h
new file mode 100644
index 0000000..6fe4aeb
--- /dev/null
+++ b/include/optutils.h
@@ -0,0 +1,111 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_OPTUTILS_H
+#define UTIL_LINUX_OPTUTILS_H
+
+#include <assert.h>
+
+#include "c.h"
+#include "nls.h"
+#include "cctype.h"
+
+static inline const char *option_to_longopt(int c, const struct option *opts)
+{
+	const struct option *o;
+
+	assert(!(opts == NULL));
+	for (o = opts; o->name; o++)
+		if (o->val == c)
+			return o->name;
+	return NULL;
+}
+
+#ifndef OPTUTILS_EXIT_CODE
+# define OPTUTILS_EXIT_CODE EXIT_FAILURE
+#endif
+
+/*
+ * Check collisions between options.
+ *
+ * The conflicts between options are described in ul_excl_t array. The
+ * array contains groups of mutually exclusive options. For example
+ *
+ *	static const ul_excl_t excl[] = {
+ *		{ 'Z','b','c' },		// first group
+ *		{ 'b','x' },			// second group
+ *		{ 0 }
+ *	};
+ *
+ *	int excl_st[ARRAY_SIZE(excl)] = UL_EXCL_STATUS_INIT;
+ *
+ *	while ((c = getopt_long(argc, argv, "Zbcx", longopts, NULL)) != -1) {
+ *
+ *		err_exclusive_options(c, longopts, excl, excl_st);
+ *
+ *		switch (c) {
+ *		case 'Z':
+ *		   ....
+ *		}
+ *	}
+ *
+ * The array excl[] defines two groups of the mutually exclusive options. The
+ * option '-b' is in the both groups.
+ *
+ * Note that the options in the group have to be in ASCII order (ABC..abc..) and
+ * groups have to be also in ASCII order.
+ *
+ * The maximal number of the options in the group is 15 (size of the array is
+ * 16, last is zero).
+ *
+ * The current status of options is stored in excl_st array. The size of the array
+ * must be the same as number of the groups in the ul_excl_t array.
+ *
+ * If you're unsure then see sys-utils/mount.c or misc-utils/findmnt.c.
+ */
+#define UL_EXCL_STATUS_INIT	{ 0 }
+typedef int ul_excl_t[16];
+
+static inline void err_exclusive_options(
+			int c,
+			const struct option *opts,
+			const ul_excl_t *excl,
+			int *status)
+{
+	int e;
+
+	for (e = 0; excl[e][0] && excl[e][0] <= c; e++) {
+		const int *op = excl[e];
+
+		for (; *op && *op <= c; op++) {
+			if (*op != c)
+				continue;
+			if (status[e] == 0)
+				status[e] = c;
+			else if (status[e] != c) {
+				size_t ct = 0;
+
+				fprintf(stderr, _("%s: mutually exclusive "
+						  "arguments:"),
+						program_invocation_short_name);
+
+				for (op = excl[e];
+				     ct + 1 < ARRAY_SIZE(excl[0]) && *op;
+				     op++, ct++) {
+					const char *n = option_to_longopt(*op, opts);
+					if (n)
+						fprintf(stderr, " --%s", n);
+					else if (c_isgraph(*op))
+						fprintf(stderr, " -%c", *op);
+				}
+				fputc('\n', stderr);
+				exit(OPTUTILS_EXIT_CODE);
+			}
+			break;
+		}
+	}
+}
+
+#endif
+
diff --git a/include/pager.h b/include/pager.h
new file mode 100644
index 0000000..6d8fdd7
--- /dev/null
+++ b/include/pager.h
@@ -0,0 +1,14 @@
+/*
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ *
+ * This file may be redistributed under the terms of the GNU Public License.
+ */
+#ifndef UTIL_LINUX_PAGER
+#define UTIL_LINUX_PAGER
+
+void pager_redirect(void);
+
+void pager_open(void);
+void pager_close(void);
+
+#endif
diff --git a/include/partx.h b/include/partx.h
new file mode 100644
index 0000000..7924aae
--- /dev/null
+++ b/include/partx.h
@@ -0,0 +1,67 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_PARTX_H
+#define UTIL_LINUX_PARTX_H
+
+#include <sys/ioctl.h>
+#include <linux/blkpg.h>
+#include <stdint.h>
+
+#ifndef BLKPG_ADD_PARTITION
+# define BLKPG_ADD_PARTITION	1
+#endif
+
+#ifndef BLKPG_DEL_PARTITION
+# define BLKPG_DEL_PARTITION	2
+#endif
+
+#ifndef BLKPG_RESIZE_PARTITION
+# define BLKPG_RESIZE_PARTITION	3		/* since Linux 3.6 */
+#endif
+
+
+#define INIT_BLKPG_PARTITION(_partno, _start, _size) {	\
+		.pno    = (_partno),			\
+		.start  = (_start) << 9,		\
+		.length = (_size) << 9,			\
+		.devname[0] = 0,			\
+		.volname[0] = 0				\
+}
+
+#define INIT_BLKPG_ARG(_action, _part) {		\
+		.op      = (_action),			\
+		.flags   = 0,				\
+		.datalen = sizeof(*(_part)),		\
+		.data = (_part)				\
+}
+
+
+static inline int partx_del_partition(int fd, unsigned int partno)
+{
+	struct blkpg_partition p = INIT_BLKPG_PARTITION(partno, 0, 0);
+	struct blkpg_ioctl_arg a = INIT_BLKPG_ARG(BLKPG_DEL_PARTITION, &p);
+
+	return ioctl(fd, BLKPG, &a);
+}
+
+static inline int partx_add_partition(int fd, int partno,
+			uint64_t start, uint64_t size)
+{
+	struct blkpg_partition p = INIT_BLKPG_PARTITION(partno, start, size);
+	struct blkpg_ioctl_arg a = INIT_BLKPG_ARG(BLKPG_ADD_PARTITION, &p);
+
+	return ioctl(fd, BLKPG, &a);
+}
+
+static inline int partx_resize_partition(int fd, int partno,
+			uint64_t start, uint64_t size)
+{
+	struct blkpg_partition p = INIT_BLKPG_PARTITION(partno, start, size);
+	struct blkpg_ioctl_arg a = INIT_BLKPG_ARG(BLKPG_RESIZE_PARTITION, &p);
+
+	return ioctl(fd, BLKPG, &a);
+}
+
+#endif /*  UTIL_LINUX_PARTX_H */
diff --git a/include/path.h b/include/path.h
new file mode 100644
index 0000000..19c1be6
--- /dev/null
+++ b/include/path.h
@@ -0,0 +1,148 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_PATH_H
+#define UTIL_LINUX_PATH_H
+
+#include <stdio.h>
+#include <stdint.h>
+#include <sys/stat.h>
+#include <sys/types.h>
+#include <dirent.h>
+
+#include "c.h"
+
+struct path_cxt {
+	int	dir_fd;
+	char	*dir_path;
+
+	int	refcount;
+
+	char *prefix;
+	char path_buffer[PATH_MAX];
+
+	void	*dialect;
+	void	(*free_dialect)(struct path_cxt *);
+	int	(*redirect_on_enoent)(struct path_cxt *, const char *, int *);
+};
+
+struct path_cxt *ul_new_path(const char *dir, ...)
+			__attribute__ ((__format__ (__printf__, 1, 2)));
+void ul_unref_path(struct path_cxt *pc);
+void ul_ref_path(struct path_cxt *pc);
+
+void ul_path_init_debug(void);
+
+int ul_path_set_prefix(struct path_cxt *pc, const char *prefix);
+const char *ul_path_get_prefix(struct path_cxt *pc);
+
+int ul_path_set_dir(struct path_cxt *pc, const char *dir);
+const char *ul_path_get_dir(struct path_cxt *pc);
+
+int ul_path_set_dialect(struct path_cxt *pc, void *data, void free_data(struct path_cxt *));
+void *ul_path_get_dialect(struct path_cxt *pc);
+
+int ul_path_set_enoent_redirect(struct path_cxt *pc, int (*func)(struct path_cxt *, const char *, int *));
+int ul_path_get_dirfd(struct path_cxt *pc);
+void ul_path_close_dirfd(struct path_cxt *pc);
+int ul_path_isopen_dirfd(struct path_cxt *pc);
+int ul_path_is_accessible(struct path_cxt *pc);
+
+char *ul_path_get_abspath(struct path_cxt *pc, char *buf, size_t bufsz, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 4, 5)));
+
+int ul_path_stat(struct path_cxt *pc, struct stat *sb, int flags, const char *path);
+int ul_path_access(struct path_cxt *pc, int mode, const char *path);
+int ul_path_accessf(struct path_cxt *pc, int mode, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 3, 4)));
+
+int ul_path_open(struct path_cxt *pc, int flags, const char *path);
+int ul_path_openf(struct path_cxt *pc, int flags, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 3, 4)));
+int ul_path_vopenf(struct path_cxt *pc, int flags, const char *path, va_list ap)
+				__attribute__ ((__format__ (__printf__, 3, 0)));
+
+FILE *ul_path_fopen(struct path_cxt *pc, const char *mode, const char *path);
+FILE *ul_path_fopenf(struct path_cxt *pc, const char *mode, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 3, 4)));
+FILE *ul_path_vfopenf(struct path_cxt *pc, const char *mode, const char *path, va_list ap)
+				__attribute__ ((__format__ (__printf__, 3, 0)));
+
+DIR *ul_path_opendir(struct path_cxt *pc, const char *path);
+DIR *ul_path_vopendirf(struct path_cxt *pc, const char *path, va_list ap)
+				__attribute__ ((__format__ (__printf__, 2, 0)));
+DIR *ul_path_opendirf(struct path_cxt *pc, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 2, 3)));
+
+ssize_t ul_path_readlink(struct path_cxt *pc, char *buf, size_t bufsiz, const char *path);
+ssize_t ul_path_readlinkf(struct path_cxt *pc, char *buf, size_t bufsiz, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 4, 5)));
+
+int ul_path_read(struct path_cxt *pc, char *buf, size_t len, const char *path);
+int ul_path_vreadf(struct path_cxt *pc, char *buf, size_t len, const char *path, va_list ap)
+				__attribute__ ((__format__ (__printf__, 4, 0)));
+int ul_path_readf(struct path_cxt *pc, char *buf, size_t len, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 4, 5)));
+
+int ul_path_read_string(struct path_cxt *pc, char **str, const char *path);
+int ul_path_readf_string(struct path_cxt *pc, char **str, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 3, 4)));
+
+int ul_path_read_buffer(struct path_cxt *pc, char *buf, size_t bufsz, const char *path);
+int ul_path_readf_buffer(struct path_cxt *pc, char *buf, size_t bufsz, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 4, 5)));
+
+int ul_path_scanf(struct path_cxt *pc, const char *path, const char *fmt, ...)
+				__attribute__ ((__format__ (__scanf__, 3, 4)));
+int ul_path_scanff(struct path_cxt *pc, const char *path, va_list ap, const char *fmt, ...)
+				__attribute__ ((__format__ (__printf__, 2, 0)))
+				__attribute__ ((__format__ (__scanf__, 4, 5)));
+
+int ul_path_read_majmin(struct path_cxt *pc, dev_t *res, const char *path);
+int ul_path_readf_majmin(struct path_cxt *pc, dev_t *res, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 3, 4)));
+
+int ul_path_read_u32(struct path_cxt *pc, uint32_t *res, const char *path);
+int ul_path_readf_u32(struct path_cxt *pc, uint32_t *res, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 3, 4)));
+
+int ul_path_read_s32(struct path_cxt *pc, int32_t *res, const char *path);
+int ul_path_readf_s32(struct path_cxt *pc, int32_t *res, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 3, 4)));
+
+int ul_path_read_u64(struct path_cxt *pc, uint64_t *res, const char *path);
+int ul_path_readf_u64(struct path_cxt *pc, uint64_t *res, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 3, 4)));
+
+int ul_path_read_s64(struct path_cxt *pc, int64_t *res, const char *path);
+int ul_path_readf_s64(struct path_cxt *pc, int64_t *res, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 3, 4)));
+
+int ul_path_write_string(struct path_cxt *pc, const char *str, const char *path);
+int ul_path_writef_string(struct path_cxt *pc, const char *str, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 3, 4)));
+
+int ul_path_write_s64(struct path_cxt *pc, int64_t num, const char *path);
+int ul_path_write_u64(struct path_cxt *pc, uint64_t num, const char *path);
+int ul_path_writef_u64(struct path_cxt *pc, uint64_t num, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 3, 4)));
+
+int ul_path_count_dirents(struct path_cxt *pc, const char *path);
+int ul_path_countf_dirents(struct path_cxt *pc, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 2, 3)));
+
+int ul_path_next_dirent(struct path_cxt *pc, DIR **sub, const char *dirname, struct dirent **d);
+
+FILE *ul_prefix_fopen(const char *prefix, const char *path, const char *mode);
+
+
+#ifdef HAVE_CPU_SET_T
+# include "cpuset.h"
+int ul_path_readf_cpuset(struct path_cxt *pc, cpu_set_t **set, int maxcpus, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 4, 5)));
+
+int ul_path_readf_cpulist(struct path_cxt *pc, cpu_set_t **set, int maxcpus, const char *path, ...)
+				__attribute__ ((__format__ (__printf__, 4, 5)));
+#endif /* HAVE_CPU_SET_T */
+#endif /* UTIL_LINUX_PATH_H */
diff --git a/include/pathnames.h b/include/pathnames.h
new file mode 100644
index 0000000..56f64c3
--- /dev/null
+++ b/include/pathnames.h
@@ -0,0 +1,230 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef PATHNAMES_H
+#define PATHNAMES_H
+
+#ifdef HAVE_PATHS_H
+# include <paths.h>
+#endif
+
+#ifndef __STDC__
+# error "we need an ANSI compiler"
+#endif
+
+/* used by kernel in /proc (e.g. /proc/swaps) for deleted files */
+#define PATH_DELETED_SUFFIX	" (deleted)"
+
+/* DEFPATHs from <paths.h> don't include /usr/local */
+#undef _PATH_DEFPATH
+
+#ifdef USE_USRDIR_PATHS_ONLY
+# define _PATH_DEFPATH	        "/usr/local/bin:/usr/bin"
+#else
+# define _PATH_DEFPATH	        "/usr/local/bin:/bin:/usr/bin"
+#endif
+
+#undef _PATH_DEFPATH_ROOT
+
+#ifdef USE_USRDIR_PATHS_ONLY
+# define _PATH_DEFPATH_ROOT	"/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin"
+#else
+# define _PATH_DEFPATH_ROOT	"/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin"
+#endif
+
+#define	_PATH_HUSHLOGIN		".hushlogin"
+#define	_PATH_HUSHLOGINS	"/etc/hushlogins"
+
+#define _PATH_NOLOGIN_TXT	"/etc/nologin.txt"
+
+#ifndef _PATH_MAILDIR
+# define _PATH_MAILDIR		"/var/spool/mail"
+#endif
+#define	_PATH_MOTDFILE		"/usr/share/misc/motd:/run/motd:/etc/motd"
+#ifndef _PATH_NOLOGIN
+# define _PATH_NOLOGIN		"/etc/nologin"
+#endif
+#define	_PATH_VAR_NOLOGIN	"/var/run/nologin"
+
+#ifndef _PATH_LOGIN
+# define _PATH_LOGIN		"/bin/login"
+#endif
+#define _PATH_SHUTDOWN		"/sbin/shutdown"
+#define _PATH_POWEROFF		"/sbin/poweroff"
+
+#define _PATH_TERMCOLORS_DIRNAME "terminal-colors.d"
+#define _PATH_TERMCOLORS_DIR	"/etc/" _PATH_TERMCOLORS_DIRNAME
+
+/* login paths */
+#define _PATH_PASSWD		"/etc/passwd"
+#define _PATH_GSHADOW		"/etc/gshadow"
+#define _PATH_GROUP		"/etc/group"
+#define _PATH_SHADOW_PASSWD	"/etc/shadow"
+#define _PATH_SHELLS		"/etc/shells"
+
+#ifndef _PATH_TMP
+# define _PATH_TMP		"/tmp/"
+#endif
+
+#ifndef _PATH_BTMP
+# define _PATH_BTMP		"/var/log/btmp"
+#endif
+
+#define _PATH_ISSUE_FILENAME	"issue"
+#define _PATH_ISSUE_DIRNAME	_PATH_ISSUE_FILENAME ".d"
+
+#define _PATH_ISSUE		"/etc/" _PATH_ISSUE_FILENAME
+#define _PATH_ISSUEDIR		"/etc/" _PATH_ISSUE_DIRNAME
+
+#define _PATH_OS_RELEASE_ETC	"/etc/os-release"
+#define _PATH_OS_RELEASE_USR	"/usr/lib/os-release"
+#define _PATH_NUMLOCK_ON	_PATH_RUNSTATEDIR "/numlock-on"
+#define _PATH_LOGINDEFS		"/etc/login.defs"
+
+#define _PATH_SD_UNITSLOAD	_PATH_RUNSTATEDIR "/systemd/systemd-units-load"
+
+/* misc paths */
+#define _PATH_WORDS             "/usr/share/dict/words"
+#define _PATH_WORDS_ALT         "/usr/share/dict/web2"
+
+/* mount paths */
+#define _PATH_FILESYSTEMS	"/etc/filesystems"
+#define _PATH_PROC		"/proc"
+#define _PATH_PROC_SWAPS	"/proc/swaps"
+#define _PATH_PROC_FILESYSTEMS	"/proc/filesystems"
+#define _PATH_PROC_MOUNTS	"/proc/mounts"
+#define _PATH_PROC_PARTITIONS	"/proc/partitions"
+#define _PATH_PROC_DEVICES	"/proc/devices"
+#define _PATH_PROC_MOUNTINFO	"/proc/self/mountinfo"
+#define _PATH_PROC_LOCKS        "/proc/locks"
+#define _PATH_PROC_CDROMINFO	"/proc/sys/dev/cdrom/info"
+
+/* unshare paths */
+#define _PATH_SUBUID		"/etc/subuid"
+#define _PATH_SUBGID		"/etc/subgid"
+#define _PATH_PROC_UIDMAP	"/proc/self/uid_map"
+#define _PATH_PROC_GIDMAP	"/proc/self/gid_map"
+#define _PATH_PROC_SETGROUPS	"/proc/self/setgroups"
+
+#define _PATH_PROC_FDDIR	"/proc/self/fd"
+
+#define _PATH_PROC_ATTR_CURRENT	"/proc/self/attr/current"
+#define _PATH_PROC_ATTR_EXEC	"/proc/self/attr/exec"
+#define _PATH_PROC_CAPLASTCAP	"/proc/sys/kernel/cap_last_cap"
+
+
+#define _PATH_SYS_BLOCK		"/sys/block"
+#define _PATH_SYS_DEVBLOCK	"/sys/dev/block"
+#define _PATH_SYS_DEVCHAR	"/sys/dev/char"
+#define _PATH_SYS_CLASS		"/sys/class"
+#define _PATH_SYS_SCSI		"/sys/bus/scsi"
+#define _PATH_SYS_CPU_BYTEORDER	"/sys/kernel/cpu_byteorder"
+#define _PATH_SYS_ADDRESS_BITS	"/sys/kernel/address_bits"
+
+#define _PATH_SYS_SELINUX	"/sys/fs/selinux"
+#define _PATH_SYS_APPARMOR	"/sys/kernel/security/apparmor"
+
+#ifndef _PATH_MOUNTED
+# ifdef MOUNTED					/* deprecated */
+#  define _PATH_MOUNTED		MOUNTED
+# else
+#  define _PATH_MOUNTED		"/etc/mtab"
+# endif
+#endif
+
+#ifndef _PATH_MNTTAB
+# ifdef MNTTAB					/* deprecated */
+#  define _PATH_MNTTAB		MNTTAB
+# else
+#  define _PATH_MNTTAB		"/etc/fstab"
+# endif
+#endif
+
+#ifndef _PATH_DEV
+  /*
+   * The tailing '/' in _PATH_DEV is there for compatibility with libc.
+   */
+# define _PATH_DEV		"/dev/"
+#endif
+
+#define _PATH_DEV_MAPPER	"/dev/mapper"
+
+#define _PATH_DEV_MEM		"/dev/mem"
+
+#define _PATH_DEV_LOOP		"/dev/loop"
+#define _PATH_DEV_LOOPCTL	"/dev/loop-control"
+
+/* udev paths */
+#define _PATH_DEV_BYLABEL	"/dev/disk/by-label"
+#define _PATH_DEV_BYUUID	"/dev/disk/by-uuid"
+#define _PATH_DEV_BYID		"/dev/disk/by-id"
+#define _PATH_DEV_BYPATH	"/dev/disk/by-path"
+#define _PATH_DEV_BYPARTLABEL	"/dev/disk/by-partlabel"
+#define _PATH_DEV_BYPARTUUID	"/dev/disk/by-partuuid"
+
+/* hwclock paths */
+#ifdef CONFIG_ADJTIME_PATH
+# define _PATH_ADJTIME		CONFIG_ADJTIME_PATH
+#else
+# define _PATH_ADJTIME		"/etc/adjtime"
+#endif
+
+#ifdef __ia64__
+# define _PATH_RTC_DEV		"/dev/efirtc"
+#else
+# define _PATH_RTC_DEV		"/dev/rtc0"
+#endif
+
+/* raw paths*/
+#define _PATH_RAWDEVDIR		"/dev/raw/"
+#define _PATH_RAWDEVCTL		_PATH_RAWDEVDIR "rawctl"
+/* deprecated */
+#define _PATH_RAWDEVCTL_OLD	"/dev/rawctl"
+
+#define _PATH_PROC_KERNEL	"/proc/sys/kernel"
+
+/* ipc paths */
+#define _PATH_PROC_SYSV_MSG	"/proc/sysvipc/msg"
+#define _PATH_PROC_SYSV_SEM	"/proc/sysvipc/sem"
+#define _PATH_PROC_SYSV_SHM	"/proc/sysvipc/shm"
+#define _PATH_PROC_IPC_MSGMAX	_PATH_PROC_KERNEL "/msgmax"
+#define _PATH_PROC_IPC_MSGMNB	_PATH_PROC_KERNEL "/msgmnb"
+#define _PATH_PROC_IPC_MSGMNI	_PATH_PROC_KERNEL "/msgmni"
+#define _PATH_PROC_IPC_SEM	_PATH_PROC_KERNEL "/sem"
+#define _PATH_PROC_IPC_SHMALL	_PATH_PROC_KERNEL "/shmall"
+#define _PATH_PROC_IPC_SHMMAX	_PATH_PROC_KERNEL "/shmmax"
+#define _PATH_PROC_IPC_SHMMNI	_PATH_PROC_KERNEL "/shmmni"
+
+/* util clamp */
+#define _PATH_PROC_UCLAMP_MIN	_PATH_PROC_KERNEL "/sched_util_clamp_min"
+#define _PATH_PROC_UCLAMP_MAX	_PATH_PROC_KERNEL "/sched_util_clamp_max"
+
+/* sysctl fs paths */
+#define _PATH_PROC_SYS_FS	"/proc/sys/fs"
+#define _PATH_PROC_PIPE_MAX_SIZE	_PATH_PROC_SYS_FS "/pipe-max-size"
+
+/* irqtop paths */
+#define _PATH_PROC_INTERRUPTS	"/proc/interrupts"
+#define _PATH_PROC_SOFTIRQS	"/proc/softirqs"
+#define _PATH_PROC_UPTIME	"/proc/uptime"
+
+/* kernel command line */
+#define _PATH_PROC_CMDLINE	"/proc/cmdline"
+
+
+/* logger paths */
+#define _PATH_DEVLOG		"/dev/log"
+
+/* ctrlaltdel paths */
+#define _PATH_PROC_CTRL_ALT_DEL	"/proc/sys/kernel/ctrl-alt-del"
+
+/* lscpu paths */
+#define _PATH_PROC_CPUINFO	"/proc/cpuinfo"
+
+/* rfkill paths */
+#define _PATH_DEV_RFKILL	"/dev/rfkill"
+#define _PATH_SYS_RFKILL	"/sys/class/rfkill"
+
+
+#endif /* PATHNAMES_H */
diff --git a/include/pidfd-utils.h b/include/pidfd-utils.h
new file mode 100644
index 0000000..ff0bc4c
--- /dev/null
+++ b/include/pidfd-utils.h
@@ -0,0 +1,49 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_PIDFD_UTILS
+#define UTIL_LINUX_PIDFD_UTILS
+
+#ifdef HAVE_SYS_SYSCALL_H
+# include <sys/syscall.h>
+
+/*
+ * If the kernel headers are too old to provide the syscall numbers, let's
+ * define them ourselves. This can be helpful while cross-compiling.
+ */
+#ifndef __NR_pidfd_send_signal
+#define __NR_pidfd_send_signal 424
+#define SYS_pidfd_send_signal __NR_pidfd_send_signal
+#endif
+#ifndef __NR_pidfd_open
+#define __NR_pidfd_open 434
+#define SYS_pidfd_open __NR_pidfd_open
+#endif
+
+# if defined(SYS_pidfd_send_signal) && defined(SYS_pidfd_open)
+#  ifdef HAVE_SYS_PIDFD_H
+#   include <sys/pidfd.h>
+#  endif
+#  include <sys/types.h>
+#  ifndef HAVE_PIDFD_SEND_SIGNAL
+#   include <sys/wait.h>
+static inline int pidfd_send_signal(int pidfd, int sig, siginfo_t *info,
+				    unsigned int flags)
+{
+	return syscall(SYS_pidfd_send_signal, pidfd, sig, info, flags);
+}
+#  endif
+
+#  ifndef HAVE_PIDFD_OPEN
+static inline int pidfd_open(pid_t pid, unsigned int flags)
+{
+	return syscall(SYS_pidfd_open, pid, flags);
+}
+#  endif
+
+#  define UL_HAVE_PIDFD 1
+
+# endif	/* SYS_pidfd_send_signal */
+#endif /* HAVE_SYS_SYSCALL_H */
+#endif /* UTIL_LINUX_PIDFD_UTILS */
diff --git a/include/plymouth-ctrl.h b/include/plymouth-ctrl.h
new file mode 100644
index 0000000..b6f1299
--- /dev/null
+++ b/include/plymouth-ctrl.h
@@ -0,0 +1,65 @@
+/*
+ * plymouth-ctrl.h	Header file for communications with plymouthd
+ *
+ * Copyright (c) 2016 SUSE Linux GmbH, All rights reserved.
+ * Copyright (c) 2016 Werner Fink <werner@suse.de>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2, or (at your option)
+ * any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program (see the file COPYING); if not, write to the
+ * Free Software Foundation, Inc., 51 Franklin St, Fifth Floor, Boston,
+ * MA 02110-1301, USA.
+ *
+ * Author: Werner Fink <werner@suse.de>
+ */
+
+/*
+ * Taken from plymouth 0.9.0 src/ply-boot-protocol.h
+ */
+
+#ifndef UTIL_LINUX_PLYMOUTH_CTRL_H
+#define UTIL_LINUX_PLYMOUTH_CTRL_H
+
+#define PLYMOUTH_SOCKET_PATH	"\0/org/freedesktop/plymouthd"
+#define ANSWER_TYP		'\x2'
+#define ANSWER_ENQ		'\x5'
+#define ANSWER_ACK		'\x6'
+#define ANSWER_MLT		'\t'
+#define ANSWER_NCK		'\x15'
+
+#define MAGIC_PRG_STOP		'A'
+#define MAGIC_PRG_CONT		'a'
+#define MAGIC_UPDATE		'U'
+#define MAGIC_SYS_UPDATE	'u'
+#define MAGIC_SYS_INIT		'S'
+#define MAGIC_DEACTIVATE	'D'
+#define MAGIC_REACTIVATE	'r'
+#define MAGIC_SHOW_SPLASH	'$'
+#define MAGIC_HIDE_SPLASH	'H'
+#define MAGIC_CHMOD		'C'
+#define MAGIC_CHROOT		'R'
+#define MAGIC_ACTIVE_VT		'V'
+#define MAGIC_QUESTION		'W'
+#define MAGIC_SHOW_MSG		'M'
+#define MAGIC_HIDE_MSG		'm'
+#define MAGIC_KEYSTROKE		'K'
+#define MAGIC_KEYSTROKE_RM	'L'
+#define MAGIC_PING		'P'
+#define MAGIC_QUIT		'Q'
+#define MAGIC_CACHED_PWD	'c'
+#define MAGIC_ASK_PWD		'*'
+#define MAGIC_DETAILS		'!'
+
+#define PLYMOUTH_TERMIOS_FLAGS_DELAY	30
+extern int plymouth_command(int cmd, ...);
+
+#endif /* UTIL_LINUX_PLYMOUTH_CTRL_H */
diff --git a/include/procfs.h b/include/procfs.h
new file mode 100644
index 0000000..25be513
--- /dev/null
+++ b/include/procfs.h
@@ -0,0 +1,67 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ *
+ * Copyright (C) 2021 Karel Zak <kzak@redhat.com>
+ */
+#ifndef UTIL_LINUX_PROCFS_H
+#define UTIL_LINUX_PROCFS_H
+
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <stdint.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <string.h>
+#include <inttypes.h>
+#include <dirent.h>
+
+#include "path.h"
+
+struct procfs_process {
+	pid_t pid;
+};
+
+extern void ul_procfs_init_debug(void);
+extern struct path_cxt *ul_new_procfs_path(pid_t pid, const char *prefix);
+extern int procfs_process_init_path(struct path_cxt *pc, pid_t pid);
+
+extern int procfs_process_get_uid(struct path_cxt *pc, uid_t *uid);
+extern ssize_t procfs_process_get_cmdline(struct path_cxt *pc, char *buf, size_t bufsz);
+extern ssize_t procfs_process_get_cmdname(struct path_cxt *pc, char *buf, size_t bufsz);
+extern ssize_t procfs_process_get_stat(struct path_cxt *pc, char *buf, size_t bufsz);
+
+extern int procfs_process_get_stat_nth(struct path_cxt *pc, int n, uintmax_t *re);
+
+static inline ssize_t procfs_process_get_exe(struct path_cxt *pc, char *buf, size_t bufsz)
+{
+	return ul_path_readlink(pc, buf, bufsz, "exe");
+}
+
+static inline ssize_t procfs_process_get_root(struct path_cxt *pc, char *buf, size_t bufsz)
+{
+	return ul_path_readlink(pc, buf, bufsz, "root");
+}
+
+static inline ssize_t procfs_process_get_cwd(struct path_cxt *pc, char *buf, size_t bufsz)
+{
+	return ul_path_readlink(pc, buf, bufsz, "cwd");
+}
+
+extern int procfs_process_next_tid(struct path_cxt *pc, DIR **sub, pid_t *tid);
+extern int procfs_process_next_fd(struct path_cxt *pc, DIR **sub, int *fd);
+
+extern int procfs_dirent_is_process(struct dirent *d);
+extern int procfs_dirent_get_pid(struct dirent *d, pid_t *pid);
+extern int procfs_dirent_get_uid(DIR *procfs, struct dirent *d, uid_t *uid);
+extern int procfs_dirent_match_uid(DIR *procfs, struct dirent *d, uid_t uid);
+extern int procfs_dirent_get_name(DIR *procfs, struct dirent *d, char *buf, size_t bufsz);
+extern int procfs_dirent_match_name(DIR *procfs, struct dirent *d, const char *name);
+
+extern int fd_is_procfs(int fd);
+extern char *pid_get_cmdname(pid_t pid);
+extern char *pid_get_cmdline(pid_t pid);
+
+#endif /* UTIL_LINUX_PROCFS_H */
diff --git a/include/pt-bsd.h b/include/pt-bsd.h
new file mode 100644
index 0000000..20c24b7
--- /dev/null
+++ b/include/pt-bsd.h
@@ -0,0 +1,160 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_PT_BSD_H
+#define UTIL_LINUX_PT_BSD_H
+
+#define BSD_MAXPARTITIONS	16
+#define BSD_FS_UNUSED		0
+
+#ifndef BSD_DISKMAGIC
+# define BSD_DISKMAGIC     ((uint32_t) 0x82564557)
+#endif
+
+#define BSD_LINUX_BOOTDIR "/usr/ucb/mdec"
+
+#if defined (__alpha__) || defined (__powerpc__) || \
+    defined (__ia64__) || defined (__hppa__)
+# define BSD_LABELSECTOR   0
+# define BSD_LABELOFFSET   64
+#else
+# define BSD_LABELSECTOR   1
+# define BSD_LABELOFFSET   0
+#endif
+
+#define	BSD_BBSIZE        8192		/* size of boot area, with label */
+#define	BSD_SBSIZE        8192		/* max size of fs superblock */
+
+struct bsd_disklabel {
+	uint32_t	d_magic;		/* the magic number */
+	int16_t		d_type;			/* drive type */
+	int16_t		d_subtype;		/* controller/d_type specific */
+	char		d_typename[16];		/* type name, e.g. "eagle" */
+	char		d_packname[16];		/* pack identifier */
+
+			/* disk geometry: */
+	uint32_t	d_secsize;		/* # of bytes per sector */
+	uint32_t	d_nsectors;		/* # of data sectors per track */
+	uint32_t	d_ntracks;		/* # of tracks per cylinder */
+	uint32_t	d_ncylinders;		/* # of data cylinders per unit */
+	uint32_t	d_secpercyl;		/* # of data sectors per cylinder */
+	uint32_t	d_secperunit;		/* # of data sectors per unit */
+
+	/*
+	 * Spares (bad sector replacements) below
+	 * are not counted in d_nsectors or d_secpercyl.
+	 * Spare sectors are assumed to be physical sectors
+	 * which occupy space at the end of each track and/or cylinder.
+	 */
+	uint16_t	d_sparespertrack;	/* # of spare sectors per track */
+	uint16_t	d_sparespercyl;		/* # of spare sectors per cylinder */
+
+	/*
+	 * Alternate cylinders include maintenance, replacement,
+	 * configuration description areas, etc.
+	 */
+	uint32_t	d_acylinders;		/* # of alt. cylinders per unit */
+
+			/* hardware characteristics: */
+	/*
+	 * d_interleave, d_trackskew and d_cylskew describe perturbations
+	 * in the media format used to compensate for a slow controller.
+	 * Interleave is physical sector interleave, set up by the formatter
+	 * or controller when formatting.  When interleaving is in use,
+	 * logically adjacent sectors are not physically contiguous,
+	 * but instead are separated by some number of sectors.
+	 * It is specified as the ratio of physical sectors traversed
+	 * per logical sector.  Thus an interleave of 1:1 implies contiguous
+	 * layout, while 2:1 implies that logical sector 0 is separated
+	 * by one sector from logical sector 1.
+	 * d_trackskew is the offset of sector 0 on track N
+	 * relative to sector 0 on track N-1 on the same cylinder.
+	 * Finally, d_cylskew is the offset of sector 0 on cylinder N
+	 * relative to sector 0 on cylinder N-1.
+	 */
+	uint16_t	d_rpm;			/* rotational speed */
+	uint16_t	d_interleave;		/* hardware sector interleave */
+	uint16_t	d_trackskew;		/* sector 0 skew, per track */
+	uint16_t	d_cylskew;		/* sector 0 skew, per cylinder */
+	uint32_t	d_headswitch;		/* head switch time, usec */
+	uint32_t	d_trkseek;		/* track-to-track seek, usec */
+	uint32_t	d_flags;		/* generic flags */
+	uint32_t	d_drivedata[5];		/* drive-type specific information */
+	uint32_t	d_spare[5];		/* reserved for future use */
+	uint32_t	d_magic2;		/* the magic number (again) */
+	uint16_t	d_checksum;		/* xor of data incl. partitions */
+
+			/* filesystem and partition information: */
+	uint16_t	d_npartitions;	        /* number of partitions in following */
+	uint32_t	d_bbsize;	        /* size of boot area at sn0, bytes */
+	uint32_t	d_sbsize;	        /* max size of fs superblock, bytes */
+
+	struct bsd_partition	 {		/* the partition table */
+		uint32_t	p_size;	        /* number of sectors in partition */
+		uint32_t	p_offset;       /* starting sector */
+		uint32_t	p_fsize;        /* filesystem basic fragment size */
+		uint8_t		p_fstype;	/* filesystem type, see below */
+		uint8_t		p_frag;		/* filesystem fragments per block */
+		uint16_t	p_cpg;	        /* filesystem cylinders per group */
+	} __attribute__((packed)) d_partitions[BSD_MAXPARTITIONS];	/* actually may be more */
+} __attribute__((packed));
+
+
+/* d_type values: */
+#define	BSD_DTYPE_SMD		1		/* SMD, XSMD; VAX hp/up */
+#define	BSD_DTYPE_MSCP		2		/* MSCP */
+#define	BSD_DTYPE_DEC		3		/* other DEC (rk, rl) */
+#define	BSD_DTYPE_SCSI		4		/* SCSI */
+#define	BSD_DTYPE_ESDI		5		/* ESDI interface */
+#define	BSD_DTYPE_ST506		6		/* ST506 etc. */
+#define	BSD_DTYPE_HPIB		7		/* CS/80 on HP-IB */
+#define BSD_DTYPE_HPFL		8		/* HP Fiber-link */
+#define	BSD_DTYPE_FLOPPY	10		/* floppy */
+
+/* d_subtype values: */
+#define BSD_DSTYPE_INDOSPART	0x8		/* is inside dos partition */
+#define BSD_DSTYPE_DOSPART(s)	((s) & 3)	/* dos partition number */
+#define BSD_DSTYPE_GEOMETRY	0x10		/* drive params in label */
+
+/*
+ * Filesystem type and version.
+ * Used to interpret other filesystem-specific
+ * per-partition information.
+ */
+#define	BSD_FS_UNUSED	0		/* unused */
+#define	BSD_FS_SWAP    	1		/* swap */
+#define	BSD_FS_V6      	2		/* Sixth Edition */
+#define	BSD_FS_V7      	3		/* Seventh Edition */
+#define	BSD_FS_SYSV    	4		/* System V */
+#define	BSD_FS_V71K    	5		/* V7 with 1K blocks (4.1, 2.9) */
+#define	BSD_FS_V8      	6		/* Eighth Edition, 4K blocks */
+#define	BSD_FS_BSDFFS	7		/* 4.2BSD fast file system */
+#define	BSD_FS_BSDLFS	9		/* 4.4BSD log-structured file system */
+#define	BSD_FS_OTHER	10		/* in use, but unknown/unsupported */
+#define	BSD_FS_HPFS	11		/* OS/2 high-performance file system */
+#define	BSD_FS_ISO9660	12		/* ISO-9660 filesystem (cdrom) */
+#define BSD_FS_ISOFS	BSD_FS_ISO9660
+#define	BSD_FS_BOOT	13		/* partition contains bootstrap */
+#define BSD_FS_ADOS	14		/* AmigaDOS fast file system */
+#define BSD_FS_HFS	15		/* Macintosh HFS */
+#define BSD_FS_ADVFS	16		/* Digital Unix AdvFS */
+
+/* this is annoying, but it's also the way it is :-( */
+#ifdef __alpha__
+#define	BSD_FS_EXT2	8		/* ext2 file system */
+#else
+#define	BSD_FS_MSDOS	8		/* MS-DOS file system */
+#endif
+
+/*
+ * flags shared by various drives:
+ */
+#define	BSD_D_REMOVABLE	0x01		/* removable media */
+#define	BSD_D_ECC      	0x02		/* supports ECC */
+#define	BSD_D_BADSECT	0x04		/* supports bad sector forw. */
+#define	BSD_D_RAMDISK	0x08		/* disk emulator */
+#define	BSD_D_CHAIN    	0x10		/* can do back-back transfers */
+#define	BSD_D_DOSPART	0x20		/* within MSDOS partition */
+
+#endif /* UTIL_LINUX_PT_BSD_H */
diff --git a/include/pt-gpt-partnames.h b/include/pt-gpt-partnames.h
new file mode 100644
index 0000000..2ee9068
--- /dev/null
+++ b/include/pt-gpt-partnames.h
@@ -0,0 +1,288 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ *
+ * Written by Karel Zak <kzak@redhat.com>
+ *
+ * Probably the most complete list of the GUIDs are at:
+ *	 https://wikipedia.org/wiki/GUID_Partition_Table
+ *
+ * The macro DEF_GUID() has to be defined in your code according to array where
+ * you want to include this file.
+ */
+
+/* Generic OS */
+DEF_GUID("C12A7328-F81F-11D2-BA4B-00A0C93EC93B", N_("EFI System")),
+
+DEF_GUID("024DEE41-33E7-11D3-9D69-0008C781F39F", N_("MBR partition scheme")),
+DEF_GUID("D3BFE2DE-3DAF-11DF-BA40-E3A556D89593", N_("Intel Fast Flash")),
+
+/* Hah!IdontneedEFI */
+DEF_GUID("21686148-6449-6E6F-744E-656564454649", N_("BIOS boot")),
+
+/* NIH syndrome */
+DEF_GUID("F4019732-066E-4E12-8273-346C5641494F", N_("Sony boot partition")),
+DEF_GUID("BFBFAFE7-A34F-448A-9A5B-6213EB736C22", N_("Lenovo boot partition")),
+
+/* PowerPC reference platform boot partition */
+DEF_GUID("9E1A2D38-C612-4316-AA26-8B49521E5A8B", N_("PowerPC PReP boot")),
+
+/* Open Network Install Environment */
+DEF_GUID("7412F7D5-A156-4B13-81DC-867174929325", N_("ONIE boot")),
+DEF_GUID("D4E6E2CD-4469-46F3-B5CB-1BFF57AFC149", N_("ONIE config")),
+
+/* Windows */
+DEF_GUID("E3C9E316-0B5C-4DB8-817D-F92DF00215AE", N_("Microsoft reserved")),
+DEF_GUID("EBD0A0A2-B9E5-4433-87C0-68B6B72699C7", N_("Microsoft basic data")),
+DEF_GUID("5808C8AA-7E8F-42E0-85D2-E1E90434CFB3", N_("Microsoft LDM metadata")),
+DEF_GUID("AF9B60A0-1431-4F62-BC68-3311714A69AD", N_("Microsoft LDM data")),
+DEF_GUID("DE94BBA4-06D1-4D40-A16A-BFD50179D6AC", N_("Windows recovery environment")),
+DEF_GUID("37AFFC90-EF7D-4E96-91C3-2D7AE055B174", N_("IBM General Parallel Fs")),
+DEF_GUID("E75CAF8F-F680-4CEE-AFA3-B001E56EFC2D", N_("Microsoft Storage Spaces")),
+
+/* HP-UX */
+DEF_GUID("75894C1E-3AEB-11D3-B7C1-7B03A0000000", N_("HP-UX data")),
+DEF_GUID("E2A1E728-32E3-11D6-A682-7B03A0000000", N_("HP-UX service")),
+
+/* Linux (https://systemd.io/DISCOVERABLE_PARTITIONS/) */
+DEF_GUID("0657FD6D-A4AB-43C4-84E5-0933C84B4F4F", N_("Linux swap")),
+DEF_GUID("0FC63DAF-8483-4772-8E79-3D69D8477DE4", N_("Linux filesystem")),
+DEF_GUID("3B8F8425-20E0-4F3B-907F-1A25A76F98E8", N_("Linux server data")),
+DEF_GUID("44479540-F297-41B2-9AF7-D131D5F0458A", N_("Linux root (x86)")),
+DEF_GUID("4F68BCE3-E8CD-4DB1-96E7-FBCAF984B709", N_("Linux root (x86-64)")),
+DEF_GUID("6523F8AE-3EB1-4E2A-A05A-18B695AE656F", N_("Linux root (Alpha)")),
+DEF_GUID("D27F46ED-2919-4CB8-BD25-9531F3C16534", N_("Linux root (ARC)")),
+DEF_GUID("69DAD710-2CE4-4E3C-B16C-21A1D49ABED3", N_("Linux root (ARM)")),
+DEF_GUID("B921B045-1DF0-41C3-AF44-4C6F280D3FAE", N_("Linux root (ARM-64)")),
+DEF_GUID("993D8D3D-F80E-4225-855A-9DAF8ED7EA97", N_("Linux root (IA-64)")),
+DEF_GUID("77055800-792C-4F94-B39A-98C91B762BB6", N_("Linux root (LoongArch-64)")),
+DEF_GUID("37C58C8A-D913-4156-A25F-48B1B64E07F0", N_("Linux root (MIPS-32 LE)")),
+DEF_GUID("700BDA43-7A34-4507-B179-EEB93D7A7CA3", N_("Linux root (MIPS-64 LE)")),
+DEF_GUID("1AACDB3B-5444-4138-BD9E-E5C2239B2346", N_("Linux root (HPPA/PARISC)")),
+DEF_GUID("1DE3F1EF-FA98-47B5-8DCD-4A860A654D78", N_("Linux root (PPC)")),
+DEF_GUID("912ADE1D-A839-4913-8964-A10EEE08FBD2", N_("Linux root (PPC64)")),
+DEF_GUID("C31C45E6-3F39-412E-80FB-4809C4980599", N_("Linux root (PPC64LE)")),
+DEF_GUID("60D5A7FE-8E7D-435C-B714-3DD8162144E1", N_("Linux root (RISC-V-32)")),
+DEF_GUID("72EC70A6-CF74-40E6-BD49-4BDA08E8F224", N_("Linux root (RISC-V-64)")),
+DEF_GUID("08A7ACEA-624C-4A20-91E8-6E0FA67D23F9", N_("Linux root (S390)")),
+DEF_GUID("5EEAD9A9-FE09-4A1E-A1D7-520D00531306", N_("Linux root (S390X)")),
+DEF_GUID("C50CDD70-3862-4CC3-90E1-809A8C93EE2C", N_("Linux root (TILE-Gx)")),
+DEF_GUID("8DA63339-0007-60C0-C436-083AC8230908", N_("Linux reserved")),
+DEF_GUID("933AC7E1-2EB4-4F13-B844-0E14E2AEF915", N_("Linux home")),
+DEF_GUID("A19D880F-05FC-4D3B-A006-743F0F84911E", N_("Linux RAID")),
+DEF_GUID("E6D6D379-F507-44C2-A23C-238F2A3DF928", N_("Linux LVM")),
+DEF_GUID("4D21B016-B534-45C2-A9FB-5C16E091FD2D", N_("Linux variable data")),
+DEF_GUID("7EC6F557-3BC5-4ACA-B293-16EF5DF639D1", N_("Linux temporary data")),
+DEF_GUID("75250D76-8CC6-458E-BD66-BD47CC81A812", N_("Linux /usr (x86)")),
+DEF_GUID("8484680C-9521-48C6-9C11-B0720656F69E", N_("Linux /usr (x86-64)")),
+DEF_GUID("E18CF08C-33EC-4C0D-8246-C6C6FB3DA024", N_("Linux /usr (Alpha)")),
+DEF_GUID("7978A683-6316-4922-BBEE-38BFF5A2FECC", N_("Linux /usr (ARC)")),
+DEF_GUID("7D0359A3-02B3-4F0A-865C-654403E70625", N_("Linux /usr (ARM)")),
+DEF_GUID("B0E01050-EE5F-4390-949A-9101B17104E9", N_("Linux /usr (ARM-64)")),
+DEF_GUID("4301D2A6-4E3B-4B2A-BB94-9E0B2C4225EA", N_("Linux /usr (IA-64)")),
+DEF_GUID("E611C702-575C-4CBE-9A46-434FA0BF7E3F", N_("Linux /usr (LoongArch-64)")),
+DEF_GUID("0F4868E9-9952-4706-979F-3ED3A473E947", N_("Linux /usr (MIPS-32 LE)")),
+DEF_GUID("C97C1F32-BA06-40B4-9F22-236061B08AA8", N_("Linux /usr (MIPS-64 LE)")),
+DEF_GUID("DC4A4480-6917-4262-A4EC-DB9384949F25", N_("Linux /usr (HPPA/PARISC)")),
+DEF_GUID("7D14FEC5-CC71-415D-9D6C-06BF0B3C3EAF", N_("Linux /usr (PPC)")),
+DEF_GUID("2C9739E2-F068-46B3-9FD0-01C5A9AFBCCA", N_("Linux /usr (PPC64)")),
+DEF_GUID("15BB03AF-77E7-4D4A-B12B-C0D084F7491C", N_("Linux /usr (PPC64LE)")),
+DEF_GUID("B933FB22-5C3F-4F91-AF90-E2BB0FA50702", N_("Linux /usr (RISC-V-32)")),
+DEF_GUID("BEAEC34B-8442-439B-A40B-984381ED097D", N_("Linux /usr (RISC-V-64)")),
+DEF_GUID("CD0F869B-D0FB-4CA0-B141-9EA87CC78D66", N_("Linux /usr (S390)")),
+DEF_GUID("8A4F5770-50AA-4ED3-874A-99B710DB6FEA", N_("Linux /usr (S390X)")),
+DEF_GUID("55497029-C7C1-44CC-AA39-815ED1558630", N_("Linux /usr (TILE-Gx)")),
+DEF_GUID("D13C5D3B-B5D1-422A-B29F-9454FDC89D76", N_("Linux root verity (x86)")),
+DEF_GUID("2C7357ED-EBD2-46D9-AEC1-23D437EC2BF5", N_("Linux root verity (x86-64)")),
+DEF_GUID("FC56D9E9-E6E5-4C06-BE32-E74407CE09A5", N_("Linux root verity (Alpha)")),
+DEF_GUID("24B2D975-0F97-4521-AFA1-CD531E421B8D", N_("Linux root verity (ARC)")),
+DEF_GUID("7386CDF2-203C-47A9-A498-F2ECCE45A2D6", N_("Linux root verity (ARM)")),
+DEF_GUID("DF3300CE-D69F-4C92-978C-9BFB0F38D820", N_("Linux root verity (ARM-64)")),
+DEF_GUID("86ED10D5-B607-45BB-8957-D350F23D0571", N_("Linux root verity (IA-64)")),
+DEF_GUID("F3393B22-E9AF-4613-A948-9D3BFBD0C535", N_("Linux root verity (LoongArch-64)")),
+DEF_GUID("D7D150D2-2A04-4A33-8F12-16651205FF7B", N_("Linux root verity (MIPS-32 LE)")),
+DEF_GUID("16B417F8-3E06-4F57-8DD2-9B5232F41AA6", N_("Linux root verity (MIPS-64 LE)")),
+DEF_GUID("D212A430-FBC5-49F9-A983-A7FEEF2B8D0E", N_("Linux root verity (HPPA/PARISC)")),
+DEF_GUID("98CFE649-1588-46DC-B2F0-ADD147424925", N_("Linux root verity (PPC)")),
+DEF_GUID("9225A9A3-3C19-4D89-B4F6-EEFF88F17631", N_("Linux root verity (PPC64)")),
+DEF_GUID("906BD944-4589-4AAE-A4E4-DD983917446A", N_("Linux root verity (PPC64LE)")),
+DEF_GUID("AE0253BE-1167-4007-AC68-43926C14C5DE", N_("Linux root verity (RISC-V-32)")),
+DEF_GUID("B6ED5582-440B-4209-B8DA-5FF7C419EA3D", N_("Linux root verity (RISC-V-64)")),
+DEF_GUID("7AC63B47-B25C-463B-8DF8-B4A94E6C90E1", N_("Linux root verity (S390)")),
+DEF_GUID("B325BFBE-C7BE-4AB8-8357-139E652D2F6B", N_("Linux root verity (S390X)")),
+DEF_GUID("966061EC-28E4-4B2E-B4A5-1F0A825A1D84", N_("Linux root verity (TILE-Gx)")),
+DEF_GUID("8F461B0D-14EE-4E81-9AA9-049B6FB97ABD", N_("Linux /usr verity (x86)")),
+DEF_GUID("77FF5F63-E7B6-4633-ACF4-1565B864C0E6", N_("Linux /usr verity (x86-64)")),
+DEF_GUID("8CCE0D25-C0D0-4A44-BD87-46331BF1DF67", N_("Linux /usr verity (Alpha)")),
+DEF_GUID("FCA0598C-D880-4591-8C16-4EDA05C7347C", N_("Linux /usr verity (ARC)")),
+DEF_GUID("C215D751-7BCD-4649-BE90-6627490A4C05", N_("Linux /usr verity (ARM)")),
+DEF_GUID("6E11A4E7-FBCA-4DED-B9E9-E1A512BB664E", N_("Linux /usr verity (ARM-64)")),
+DEF_GUID("6A491E03-3BE7-4545-8E38-83320E0EA880", N_("Linux /usr verity (IA-64)")),
+DEF_GUID("F46B2C26-59AE-48F0-9106-C50ED47F673D", N_("Linux /usr verity (LoongArch-64)")),
+DEF_GUID("46B98D8D-B55C-4E8F-AAB3-37FCA7F80752", N_("Linux /usr verity (MIPS-32 LE)")),
+DEF_GUID("3C3D61FE-B5F3-414D-BB71-8739A694A4EF", N_("Linux /usr verity (MIPS-64 LE)")),
+DEF_GUID("5843D618-EC37-48D7-9F12-CEA8E08768B2", N_("Linux /usr verity (HPPA/PARISC)")),
+DEF_GUID("DF765D00-270E-49E5-BC75-F47BB2118B09", N_("Linux /usr verity (PPC)")),
+DEF_GUID("BDB528A5-A259-475F-A87D-DA53FA736A07", N_("Linux /usr verity (PPC64)")),
+DEF_GUID("EE2B9983-21E8-4153-86D9-B6901A54D1CE", N_("Linux /usr verity (PPC64LE)")),
+DEF_GUID("CB1EE4E3-8CD0-4136-A0A4-AA61A32E8730", N_("Linux /usr verity (RISC-V-32)")),
+DEF_GUID("8F1056BE-9B05-47C4-81D6-BE53128E5B54", N_("Linux /usr verity (RISC-V-64)")),
+DEF_GUID("B663C618-E7BC-4D6D-90AA-11B756BB1797", N_("Linux /usr verity (S390)")),
+DEF_GUID("31741CC4-1A2A-4111-A581-E00B447D2D06", N_("Linux /usr verity (S390X)")),
+DEF_GUID("2FB4BF56-07FA-42DA-8132-6B139F2026AE", N_("Linux /usr verity (TILE-Gx)")),
+DEF_GUID("5996FC05-109C-48DE-808B-23FA0830B676", N_("Linux root verity sign. (x86)")),
+DEF_GUID("41092B05-9FC8-4523-994F-2DEF0408B176", N_("Linux root verity sign. (x86-64)")),
+DEF_GUID("D46495B7-A053-414F-80F7-700C99921EF8", N_("Linux root verity sign. (Alpha)")),
+DEF_GUID("143A70BA-CBD3-4F06-919F-6C05683A78BC", N_("Linux root verity sign. (ARC)")),
+DEF_GUID("42B0455F-EB11-491D-98D3-56145BA9D037", N_("Linux root verity sign. (ARM)")),
+DEF_GUID("6DB69DE6-29F4-4758-A7A5-962190F00CE3", N_("Linux root verity sign. (ARM-64)")),
+DEF_GUID("E98B36EE-32BA-4882-9B12-0CE14655F46A", N_("Linux root verity sign. (IA-64)")),
+DEF_GUID("5AFB67EB-ECC8-4F85-AE8E-AC1E7C50E7D0", N_("Linux root verity sign. (LoongArch-64)")),
+DEF_GUID("C919CC1F-4456-4EFF-918C-F75E94525CA5", N_("Linux root verity sign. (MIPS-32 LE)")),
+DEF_GUID("904E58EF-5C65-4A31-9C57-6AF5FC7C5DE7", N_("Linux root verity sign. (MIPS-64 LE)")),
+DEF_GUID("15DE6170-65D3-431C-916E-B0DCD8393F25", N_("Linux root verity sign. (HPPA/PARISC)")),
+DEF_GUID("1B31B5AA-ADD9-463A-B2ED-BD467FC857E7", N_("Linux root verity sign. (PPC)")),
+DEF_GUID("F5E2C20C-45B2-4FFA-BCE9-2A60737E1AAF", N_("Linux root verity sign. (PPC64)")),
+DEF_GUID("D4A236E7-E873-4C07-BF1D-BF6CF7F1C3C6", N_("Linux root verity sign. (PPC64LE)")),
+DEF_GUID("3A112A75-8729-4380-B4CF-764D79934448", N_("Linux root verity sign. (RISC-V-32)")),
+DEF_GUID("EFE0F087-EA8D-4469-821A-4C2A96A8386A", N_("Linux root verity sign. (RISC-V-64)")),
+DEF_GUID("3482388E-4254-435A-A241-766A065F9960", N_("Linux root verity sign. (S390)")),
+DEF_GUID("C80187A5-73A3-491A-901A-017C3FA953E9", N_("Linux root verity sign. (S390X)")),
+DEF_GUID("B3671439-97B0-4A53-90F7-2D5A8F3AD47B", N_("Linux root verity sign. (TILE-Gx)")),
+DEF_GUID("974A71C0-DE41-43C3-BE5D-5C5CCD1AD2C0", N_("Linux /usr verity sign. (x86)")),
+DEF_GUID("E7BB33FB-06CF-4E81-8273-E543B413E2E2", N_("Linux /usr verity sign. (x86-64)")),
+DEF_GUID("5C6E1C76-076A-457A-A0FE-F3B4CD21CE6E", N_("Linux /usr verity sign. (Alpha)")),
+DEF_GUID("94F9A9A1-9971-427A-A400-50CB297F0F35", N_("Linux /usr verity sign. (ARC)")),
+DEF_GUID("D7FF812F-37D1-4902-A810-D76BA57B975A", N_("Linux /usr verity sign. (ARM)")),
+DEF_GUID("C23CE4FF-44BD-4B00-B2D4-B41B3419E02A", N_("Linux /usr verity sign. (ARM-64)")),
+DEF_GUID("8DE58BC2-2A43-460D-B14E-A76E4A17B47F", N_("Linux /usr verity sign. (IA-64)")),
+DEF_GUID("B024F315-D330-444C-8461-44BBDE524E99", N_("Linux /usr verity sign. (LoongArch-64)")),
+DEF_GUID("3E23CA0B-A4BC-4B4E-8087-5AB6A26AA8A9", N_("Linux /usr verity sign. (MIPS-32 LE)")),
+DEF_GUID("F2C2C7EE-ADCC-4351-B5C6-EE9816B66E16", N_("Linux /usr verity sign. (MIPS-64 LE)")),
+DEF_GUID("450DD7D1-3224-45EC-9CF2-A43A346D71EE", N_("Linux /usr verity sign. (HPPA/PARISC)")),
+DEF_GUID("7007891D-D371-4A80-86A4-5CB875B9302E", N_("Linux /usr verity sign. (PPC)")),
+DEF_GUID("0B888863-D7F8-4D9E-9766-239FCE4D58AF", N_("Linux /usr verity sign. (PPC64)")),
+DEF_GUID("C8BFBD1E-268E-4521-8BBA-BF314C399557", N_("Linux /usr verity sign. (PPC64LE)")),
+DEF_GUID("C3836A13-3137-45BA-B583-B16C50FE5EB4", N_("Linux /usr verity sign. (RISC-V-32)")),
+DEF_GUID("D2F9000A-7A18-453F-B5CD-4D32F77A7B32", N_("Linux /usr verity sign. (RISC-V-64)")),
+DEF_GUID("17440E4F-A8D0-467F-A46E-3912AE6EF2C5", N_("Linux /usr verity sign. (S390)")),
+DEF_GUID("3F324816-667B-46AE-86EE-9B0C0C6C11B4", N_("Linux /usr verity sign. (S390X)")),
+DEF_GUID("4EDE75E2-6CCC-4CC8-B9C7-70334B087510", N_("Linux /usr verity sign. (TILE-Gx)")),
+/* ... too crazy, ignore for now:
+DEF_GUID("7FFEC5C9-2D00-49B7-8941-3EA10A5586B7", N_("Linux plain dm-crypt")),
+DEF_GUID("CA7D7CCB-63ED-4C53-861C-1742536059CC", N_("Linux LUKS")),
+*/
+/* Linux https://systemd.io/BOOT_LOADER_SPECIFICATION/ */
+DEF_GUID("BC13C2FF-59E6-4262-A352-B275FD6F7172", N_("Linux extended boot")),
+
+/* Linux https://systemd.io/HOME_DIRECTORY/ */
+DEF_GUID("773f91ef-66d4-49b5-bd83-d683bf40ad16", N_("Linux user's home")),
+
+/* FreeBSD */
+DEF_GUID("516E7CB4-6ECF-11D6-8FF8-00022D09712B", N_("FreeBSD data")),
+DEF_GUID("83BD6B9D-7F41-11DC-BE0B-001560B84F0F", N_("FreeBSD boot")),
+DEF_GUID("516E7CB5-6ECF-11D6-8FF8-00022D09712B", N_("FreeBSD swap")),
+DEF_GUID("516E7CB6-6ECF-11D6-8FF8-00022D09712B", N_("FreeBSD UFS")),
+DEF_GUID("516E7CBA-6ECF-11D6-8FF8-00022D09712B", N_("FreeBSD ZFS")),
+DEF_GUID("516E7CB8-6ECF-11D6-8FF8-00022D09712B", N_("FreeBSD Vinum")),
+
+/* Apple OSX */
+DEF_GUID("48465300-0000-11AA-AA11-00306543ECAC", N_("Apple HFS/HFS+")),
+DEF_GUID("7C3457EF-0000-11AA-AA11-00306543ECAC", N_("Apple APFS")),
+DEF_GUID("55465300-0000-11AA-AA11-00306543ECAC", N_("Apple UFS")),
+DEF_GUID("52414944-0000-11AA-AA11-00306543ECAC", N_("Apple RAID")),
+DEF_GUID("52414944-5F4F-11AA-AA11-00306543ECAC", N_("Apple RAID offline")),
+DEF_GUID("426F6F74-0000-11AA-AA11-00306543ECAC", N_("Apple boot")),
+DEF_GUID("4C616265-6C00-11AA-AA11-00306543ECAC", N_("Apple label")),
+DEF_GUID("5265636F-7665-11AA-AA11-00306543ECAC", N_("Apple TV recovery")),
+DEF_GUID("53746F72-6167-11AA-AA11-00306543ECAC", N_("Apple Core storage")),
+DEF_GUID("69646961-6700-11AA-AA11-00306543ECAC", N_("Apple Silicon boot")),
+DEF_GUID("52637672-7900-11AA-AA11-00306543ECAC", N_("Apple Silicon recovery")),
+
+/* Solaris */
+DEF_GUID("6A82CB45-1DD2-11B2-99A6-080020736631", N_("Solaris boot")),
+DEF_GUID("6A85CF4D-1DD2-11B2-99A6-080020736631", N_("Solaris root")),
+/* same as Apple ZFS */
+DEF_GUID("6A898CC3-1DD2-11B2-99A6-080020736631", N_("Solaris /usr & Apple ZFS")),
+DEF_GUID("6A87C46F-1DD2-11B2-99A6-080020736631", N_("Solaris swap")),
+DEF_GUID("6A8B642B-1DD2-11B2-99A6-080020736631", N_("Solaris backup")),
+DEF_GUID("6A8EF2E9-1DD2-11B2-99A6-080020736631", N_("Solaris /var")),
+DEF_GUID("6A90BA39-1DD2-11B2-99A6-080020736631", N_("Solaris /home")),
+DEF_GUID("6A9283A5-1DD2-11B2-99A6-080020736631", N_("Solaris alternate sector")),
+DEF_GUID("6A945A3B-1DD2-11B2-99A6-080020736631", N_("Solaris reserved 1")),
+DEF_GUID("6A9630D1-1DD2-11B2-99A6-080020736631", N_("Solaris reserved 2")),
+DEF_GUID("6A980767-1DD2-11B2-99A6-080020736631", N_("Solaris reserved 3")),
+DEF_GUID("6A96237F-1DD2-11B2-99A6-080020736631", N_("Solaris reserved 4")),
+DEF_GUID("6A8D2AC7-1DD2-11B2-99A6-080020736631", N_("Solaris reserved 5")),
+
+/* NetBSD */
+/* References:
+ *    Current: https://ftp.netbsd.org/pub/NetBSD/NetBSD-current/src/sys/sys/disklabel_gpt.h
+ *  Release 9: https://ftp.netbsd.org/pub/NetBSD/NetBSD-release-9/src/sys/sys/disklabel_gpt.h
+ * Latest CVS: http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/sys/disklabel_gpt.h?rev=.&content-type=text/x-cvsweb-markup
+ */
+DEF_GUID("49F48D32-B10E-11DC-B99B-0019D1879648", N_("NetBSD swap")),
+DEF_GUID("49F48D5A-B10E-11DC-B99B-0019D1879648", N_("NetBSD FFS")),
+DEF_GUID("49F48D82-B10E-11DC-B99B-0019D1879648", N_("NetBSD LFS")),
+DEF_GUID("2DB519C4-B10F-11DC-B99B-0019D1879648", N_("NetBSD concatenated")),
+DEF_GUID("2DB519EC-B10F-11DC-B99B-0019D1879648", N_("NetBSD encrypted")),
+DEF_GUID("49F48DAA-B10E-11DC-B99B-0019D1879648", N_("NetBSD RAID")),
+
+/* ChromeOS */
+DEF_GUID("FE3A2A5D-4F32-41A7-B725-ACCC3285A309", N_("ChromeOS kernel")),
+DEF_GUID("3CB8E202-3B7E-47DD-8A3C-7FF2A13CFCEC", N_("ChromeOS root fs")),
+DEF_GUID("2E0A753D-9E48-43B0-8337-B15192CB1B5E", N_("ChromeOS reserved")),
+
+/* MidnightBSD */
+DEF_GUID("85D5E45A-237C-11E1-B4B3-E89A8F7FC3A7", N_("MidnightBSD data")),
+DEF_GUID("85D5E45E-237C-11E1-B4B3-E89A8F7FC3A7", N_("MidnightBSD boot")),
+DEF_GUID("85D5E45B-237C-11E1-B4B3-E89A8F7FC3A7", N_("MidnightBSD swap")),
+DEF_GUID("0394EF8B-237E-11E1-B4B3-E89A8F7FC3A7", N_("MidnightBSD UFS")),
+DEF_GUID("85D5E45D-237C-11E1-B4B3-E89A8F7FC3A7", N_("MidnightBSD ZFS")),
+DEF_GUID("85D5E45C-237C-11E1-B4B3-E89A8F7FC3A7", N_("MidnightBSD Vinum")),
+
+/* Ceph */
+DEF_GUID("45B0969E-9B03-4F30-B4C6-B4B80CEFF106", N_("Ceph Journal")),
+DEF_GUID("45B0969E-9B03-4F30-B4C6-5EC00CEFF106", N_("Ceph Encrypted Journal")),
+DEF_GUID("4FBD7E29-9D25-41B8-AFD0-062C0CEFF05D", N_("Ceph OSD")),
+DEF_GUID("4FBD7E29-9D25-41B8-AFD0-5EC00CEFF05D", N_("Ceph crypt OSD")),
+DEF_GUID("89C57F98-2FE5-4DC0-89C1-F3AD0CEFF2BE", N_("Ceph disk in creation")),
+DEF_GUID("89C57F98-2FE5-4DC0-89C1-5EC00CEFF2BE", N_("Ceph crypt disk in creation")),
+
+/* VMware */
+DEF_GUID("AA31E02A-400F-11DB-9590-000C2911D1B8", N_("VMware VMFS")),
+DEF_GUID("9D275380-40AD-11DB-BF97-000C2911D1B8", N_("VMware Diagnostic")),
+DEF_GUID("381CFCCC-7288-11E0-92EE-000C2911D0B2", N_("VMware Virtual SAN")),
+DEF_GUID("77719A0C-A4A0-11E3-A47E-000C29745A24", N_("VMware Virsto")),
+DEF_GUID("9198EFFC-31C0-11DB-8F78-000C2911D1B8", N_("VMware Reserved")),
+
+/* OpenBSD */
+DEF_GUID("824CC7A0-36A8-11E3-890A-952519AD3F61", N_("OpenBSD data")),
+
+/* QNX */
+DEF_GUID("CEF5A9AD-73BC-4601-89F3-CDEEEEE321A1", N_("QNX6 file system")),
+
+/* Plan 9 */
+DEF_GUID("C91818F9-8025-47AF-89D2-F030D7000C2C", N_("Plan 9 partition")),
+
+/* HiFive bootloaders */
+DEF_GUID("5B193300-FC78-40CD-8002-E86C45580B47", N_("HiFive FSBL")),
+DEF_GUID("2E54B353-1271-4842-806F-E436D6AF6985", N_("HiFive BBL")),
+
+/* Haiku */
+DEF_GUID("42465331-3BA3-10F1-802A-4861696B7521", N_("Haiku BFS")),
+
+/* Marvell */
+DEF_GUID("6828311A-BA55-42A4-BCDE-A89BB5EDECAE", N_("Marvell Armada 3700 Boot partition")),
+
+/* DragonFly BSD */
+/* https://github.com/DragonFlyBSD/DragonFlyBSD/blob/master/etc/defaults/uuids */
+DEF_GUID("9D087404-1CA5-11DC-8817-01301BB8A9F5", N_("DragonFlyBSD Label32")),
+DEF_GUID("9D58FDBD-1CA5-11DC-8817-01301BB8A9F5", N_("DragonFlyBSD Swap")),
+DEF_GUID("9D94CE7C-1CA5-11DC-8817-01301BB8A9F5", N_("DragonFlyBSD UFS1")),
+DEF_GUID("9DD4478F-1CA5-11DC-8817-01301BB8A9F5", N_("DragonFlyBSD Vinum")),
+DEF_GUID("DBD5211B-1CA5-11DC-8817-01301BB8A9F5", N_("DragonFlyBSD CCD")),
+DEF_GUID("3D48CE54-1D16-11DC-8696-01301BB8A9F5", N_("DragonFlyBSD Label64")),
+DEF_GUID("BD215AB2-1D16-11DC-8696-01301BB8A9F5", N_("DragonFlyBSD Legacy")),
+DEF_GUID("61DC63AC-6E38-11DC-8513-01301BB8A9F5", N_("DragonFlyBSD HAMMER")),
+DEF_GUID("5CBB9AD1-862D-11DC-A94D-01301BB8A9F5", N_("DragonFlyBSD HAMMER2")),
diff --git a/include/pt-mbr-partnames.h b/include/pt-mbr-partnames.h
new file mode 100644
index 0000000..0a65daa
--- /dev/null
+++ b/include/pt-mbr-partnames.h
@@ -0,0 +1,117 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+	{0x00, N_("Empty")},
+	{0x01, N_("FAT12")},
+	{0x02, N_("XENIX root")},
+	{0x03, N_("XENIX usr")},
+	{0x04, N_("FAT16 <32M")},
+	{0x05, N_("Extended")},		/* DOS 3.3+ extended partition */
+	{0x06, N_("FAT16")},		/* DOS 16-bit >=32M */
+	{0x07, N_("HPFS/NTFS/exFAT")},	/* OS/2 IFS, eg, HPFS or NTFS or QNX or exFAT */
+	{0x08, N_("AIX")},		/* AIX boot (AIX -- PS/2 port) or SplitDrive */
+	{0x09, N_("AIX bootable")},	/* AIX data or Coherent */
+	{0x0a, N_("OS/2 Boot Manager")},/* OS/2 Boot Manager */
+	{0x0b, N_("W95 FAT32")},
+	{0x0c, N_("W95 FAT32 (LBA)")},/* LBA really is `Extended Int 13h' */
+	{0x0e, N_("W95 FAT16 (LBA)")},
+	{0x0f, N_("W95 Ext'd (LBA)")},
+	{0x10, N_("OPUS")},
+	{0x11, N_("Hidden FAT12")},
+	{0x12, N_("Compaq diagnostics")},
+	{0x14, N_("Hidden FAT16 <32M")},
+	{0x16, N_("Hidden FAT16")},
+	{0x17, N_("Hidden HPFS/NTFS")},
+	{0x18, N_("AST SmartSleep")},
+	{0x1b, N_("Hidden W95 FAT32")},
+	{0x1c, N_("Hidden W95 FAT32 (LBA)")},
+	{0x1e, N_("Hidden W95 FAT16 (LBA)")},
+	{0x24, N_("NEC DOS")},
+	{0x27, N_("Hidden NTFS WinRE")},
+	{0x39, N_("Plan 9")},
+	{0x3c, N_("PartitionMagic recovery")},
+	{0x40, N_("Venix 80286")},
+	{0x41, N_("PPC PReP Boot")},
+	{0x42, N_("SFS")},
+	{0x4d, N_("QNX4.x")},
+	{0x4e, N_("QNX4.x 2nd part")},
+	{0x4f, N_("QNX4.x 3rd part")},
+	{0x50, N_("OnTrack DM")},
+	{0x51, N_("OnTrack DM6 Aux1")},	/* (or Novell) */
+	{0x52, N_("CP/M")},		/* CP/M or Microport SysV/AT */
+	{0x53, N_("OnTrack DM6 Aux3")},
+	{0x54, N_("OnTrackDM6")},
+	{0x55, N_("EZ-Drive")},
+	{0x56, N_("Golden Bow")},
+	{0x5c, N_("Priam Edisk")},
+	{0x61, N_("SpeedStor")},
+	{0x63, N_("GNU HURD or SysV")},	/* GNU HURD or Mach or Sys V/386 (such as ISC UNIX) */
+	{0x64, N_("Novell Netware 286")},
+	{0x65, N_("Novell Netware 386")},
+	{0x70, N_("DiskSecure Multi-Boot")},
+	{0x75, N_("PC/IX")},
+	{0x80, N_("Old Minix")},	/* Minix 1.4a and earlier */
+	{0x81, N_("Minix / old Linux")},/* Minix 1.4b and later */
+	{0x82, N_("Linux swap / Solaris")},
+	{0x83, N_("Linux")},
+	{0x84, N_("OS/2 hidden or Intel hibernation")},/* OS/2 hidden C: drive,
+					   hibernation type Microsoft APM
+					   or hibernation Intel Rapid Start */
+	{0x85, N_("Linux extended")},
+	{0x86, N_("NTFS volume set")},
+	{0x87, N_("NTFS volume set")},
+	{0x88, N_("Linux plaintext")},
+	{0x8e, N_("Linux LVM")},
+	{0x93, N_("Amoeba")},
+	{0x94, N_("Amoeba BBT")},	/* (bad block table) */
+	{0x9f, N_("BSD/OS")},		/* BSDI */
+	{0xa0, N_("IBM Thinkpad hibernation")},
+	{0xa5, N_("FreeBSD")},		/* various BSD flavours */
+	{0xa6, N_("OpenBSD")},
+	{0xa7, N_("NeXTSTEP")},
+	{0xa8, N_("Darwin UFS")},
+	{0xa9, N_("NetBSD")},
+	{0xab, N_("Darwin boot")},
+	{0xaf, N_("HFS / HFS+")},
+	{0xb7, N_("BSDI fs")},
+	{0xb8, N_("BSDI swap")},
+	{0xbb, N_("Boot Wizard hidden")},
+	{0xbc, N_("Acronis FAT32 LBA")},/* hidden (+0xb0) Acronis Secure Zone (backup software) */
+	{0xbe, N_("Solaris boot")},
+	{0xbf, N_("Solaris")},
+	{0xc1, N_("DRDOS/sec (FAT-12)")},
+	{0xc4, N_("DRDOS/sec (FAT-16 < 32M)")},
+	{0xc6, N_("DRDOS/sec (FAT-16)")},
+	{0xc7, N_("Syrinx")},
+	{0xda, N_("Non-FS data")},
+	{0xdb, N_("CP/M / CTOS / ...")},/* CP/M or Concurrent CP/M or
+					   Concurrent DOS or CTOS */
+	{0xde, N_("Dell Utility")},	/* Dell PowerEdge Server utilities */
+	{0xdf, N_("BootIt")},		/* BootIt EMBRM */
+	{0xe1, N_("DOS access")},	/* DOS access or SpeedStor 12-bit FAT
+					   extended partition */
+	{0xe3, N_("DOS R/O")},		/* DOS R/O or SpeedStor */
+	{0xe4, N_("SpeedStor")},	/* SpeedStor 16-bit FAT extended
+					   partition < 1024 cyl. */
+
+	/* Linux https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec/ */
+	{0xea, N_("Linux extended boot")},
+
+	{0xeb, N_("BeOS fs")},
+	{0xee, N_("GPT")},		/* Intel EFI GUID Partition Table */
+	{0xef, N_("EFI (FAT-12/16/32)")},/* Intel EFI System Partition */
+	{0xf0, N_("Linux/PA-RISC boot")},/* Linux/PA-RISC boot loader */
+	{0xf1, N_("SpeedStor")},
+	{0xf4, N_("SpeedStor")},	/* SpeedStor large partition */
+	{0xf2, N_("DOS secondary")},	/* DOS 3.3+ secondary */
+	{0xf8, N_("EBBR protective")},	/* Arm EBBR firmware protective partition */
+	{0xfb, N_("VMware VMFS")},
+	{0xfc, N_("VMware VMKCORE")},	/* VMware kernel dump partition */
+	{0xfd, N_("Linux raid autodetect")},/* Linux raid partition with
+					       autodetect using persistent
+					       superblock */
+	{0xfe, N_("LANstep")},		/* SpeedStor >1024 cyl. or LANstep */
+	{0xff, N_("BBT")},		/* Xenix Bad Block Table */
+
+	{ 0, NULL }
diff --git a/include/pt-mbr.h b/include/pt-mbr.h
new file mode 100644
index 0000000..93a6def
--- /dev/null
+++ b/include/pt-mbr.h
@@ -0,0 +1,214 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_PT_MBR_H
+#define UTIL_LINUX_PT_MBR_H
+
+#include <assert.h>
+
+struct dos_partition {
+	unsigned char boot_ind;		/* 0x80 - active */
+	unsigned char bh, bs, bc;	/* begin CHS */
+	unsigned char sys_ind;
+	unsigned char eh, es, ec;	/* end CHS */
+	unsigned char start_sect[4];
+	unsigned char nr_sects[4];
+} __attribute__((packed));
+
+#define MBR_PT_OFFSET		0x1be
+#define MBR_PT_BOOTBITS_SIZE	440
+
+static inline struct dos_partition *mbr_get_partition(unsigned char *mbr, int i)
+{
+	return (struct dos_partition *)
+		(mbr + MBR_PT_OFFSET + (i * sizeof(struct dos_partition)));
+}
+
+/* assemble badly aligned little endian integer */
+static inline uint32_t __dos_assemble_4le(const unsigned char *p)
+{
+	uint32_t last_byte = p[3];
+
+	return p[0] | (p[1] << 8) | (p[2] << 16) | (last_byte << 24);
+}
+
+static inline void __dos_store_4le(unsigned char *p, unsigned int val)
+{
+	assert(!(p == NULL));
+	p[0] = (val & 0xff);
+	p[1] = ((val >> 8) & 0xff);
+	p[2] = ((val >> 16) & 0xff);
+	p[3] = ((val >> 24) & 0xff);
+}
+
+static inline unsigned int dos_partition_get_start(struct dos_partition *p)
+{
+	return __dos_assemble_4le(&(p->start_sect[0]));
+}
+
+static inline void dos_partition_set_start(struct dos_partition *p, unsigned int n)
+{
+	__dos_store_4le(p->start_sect, n);
+}
+
+static inline unsigned int dos_partition_get_size(struct dos_partition *p)
+{
+	return __dos_assemble_4le(&(p->nr_sects[0]));
+}
+
+static inline void dos_partition_set_size(struct dos_partition *p, unsigned int n)
+{
+	__dos_store_4le(p->nr_sects, n);
+}
+
+static inline void dos_partition_sync_chs(struct dos_partition *p, unsigned long long int part_offset, unsigned int geom_sectors, unsigned int geom_heads)
+{
+	unsigned long long int start = part_offset + dos_partition_get_start(p);
+	unsigned long long int stop = start + dos_partition_get_size(p) - 1;
+	unsigned int spc = geom_heads * geom_sectors;
+
+	if (start / spc > 1023)
+		start = spc * 1024 - 1;
+	if (stop / spc > 1023)
+		stop = spc * 1024 - 1;
+
+	p->bc = (start / spc) & 0xff;
+	p->bh = (start / geom_sectors) % geom_heads;
+	p->bs = ((start % geom_sectors + 1) & 0x3f) |
+		(((start / spc) >> 2) & 0xc0);
+
+	p->ec = (stop / spc) & 0xff;
+	p->eh = (stop / geom_sectors) % geom_heads;
+	p->es = ((stop % geom_sectors + 1) & 0x3f) |
+		(((stop / spc) >> 2) & 0xc0);
+}
+
+static inline int mbr_is_valid_magic(const unsigned char *mbr)
+{
+	return mbr[510] == 0x55 && mbr[511] == 0xaa ? 1 : 0;
+}
+
+static inline void mbr_set_magic(unsigned char *b)
+{
+	b[510] = 0x55;
+	b[511] = 0xaa;
+}
+
+static inline unsigned int mbr_get_id(const unsigned char *mbr)
+{
+	return __dos_assemble_4le(&mbr[440]);
+}
+
+static inline void mbr_set_id(unsigned char *b, unsigned int id)
+{
+	__dos_store_4le(&b[440], id);
+}
+
+enum {
+	MBR_EMPTY_PARTITION		= 0x00,
+	MBR_FAT12_PARTITION		= 0x01,
+	MBR_XENIX_ROOT_PARTITION	= 0x02,
+	MBR_XENIX_USR_PARTITION		= 0x03,
+	MBR_FAT16_LESS32M_PARTITION	= 0x04,
+	MBR_DOS_EXTENDED_PARTITION	= 0x05,
+	MBR_FAT16_PARTITION		= 0x06, /* DOS 16-bit >=32M */
+	MBR_HPFS_NTFS_PARTITION		= 0x07, /* OS/2 IFS, eg, HPFS or NTFS or QNX */
+	MBR_AIX_PARTITION		= 0x08, /* AIX boot (AIX -- PS/2 port) or SplitDrive */
+	MBR_AIX_BOOTABLE_PARTITION	= 0x09, /* AIX data or Coherent */
+	MBR_OS2_BOOTMNGR_PARTITION	= 0x0a, /* OS/2 Boot Manager */
+	MBR_W95_FAT32_PARTITION		= 0x0b,
+	MBR_W95_FAT32_LBA_PARTITION	= 0x0c, /* LBA really is `Extended Int 13h' */
+	MBR_W95_FAT16_LBA_PARTITION	= 0x0e,
+	MBR_W95_EXTENDED_PARTITION	= 0x0f,
+	MBR_OPUS_PARTITION		= 0x10,
+	MBR_HIDDEN_FAT12_PARTITION	= 0x11,
+	MBR_COMPAQ_DIAGNOSTICS_PARTITION = 0x12,
+	MBR_HIDDEN_FAT16_L32M_PARTITION	= 0x14,
+	MBR_HIDDEN_FAT16_PARTITION	= 0x16,
+	MBR_HIDDEN_HPFS_NTFS_PARTITION	= 0x17,
+	MBR_AST_SMARTSLEEP_PARTITION	= 0x18,
+	MBR_HIDDEN_W95_FAT32_PARTITION	= 0x1b,
+	MBR_HIDDEN_W95_FAT32LBA_PARTITION = 0x1c,
+	MBR_HIDDEN_W95_FAT16LBA_PARTITION = 0x1e,
+	MBR_NEC_DOS_PARTITION		= 0x24,
+	MBR_PLAN9_PARTITION		= 0x39,
+	MBR_PARTITIONMAGIC_PARTITION	= 0x3c,
+	MBR_VENIX80286_PARTITION	= 0x40,
+	MBR_PPC_PREP_BOOT_PARTITION	= 0x41,
+	MBR_SFS_PARTITION		= 0x42,
+	MBR_QNX_4X_PARTITION		= 0x4d,
+	MBR_QNX_4X_2ND_PARTITION	= 0x4e,
+	MBR_QNX_4X_3RD_PARTITION	= 0x4f,
+	MBR_DM_PARTITION		= 0x50,
+	MBR_DM6_AUX1_PARTITION		= 0x51, /* (or Novell) */
+	MBR_CPM_PARTITION		= 0x52, /* CP/M or Microport SysV/AT */
+	MBR_DM6_AUX3_PARTITION		= 0x53,
+	MBR_DM6_PARTITION		= 0x54,
+	MBR_EZ_DRIVE_PARTITION		= 0x55,
+	MBR_GOLDEN_BOW_PARTITION	= 0x56,
+	MBR_PRIAM_EDISK_PARTITION	= 0x5c,
+	MBR_SPEEDSTOR_PARTITION		= 0x61,
+	MBR_GNU_HURD_PARTITION		= 0x63, /* GNU HURD or Mach or Sys V/386 (such as ISC UNIX) */
+	MBR_UNIXWARE_PARTITION		= MBR_GNU_HURD_PARTITION,
+	MBR_NETWARE_286_PARTITION	= 0x64,
+	MBR_NETWARE_386_PARTITION	= 0x65,
+	MBR_DISKSECURE_MULTIBOOT_PARTITION = 0x70,
+	MBR_PC_IX_PARTITION		= 0x75,
+	MBR_OLD_MINIX_PARTITION		= 0x80, /* Minix 1.4a and earlier */
+	MBR_MINIX_PARTITION		= 0x81, /* Minix 1.4b and later */
+	MBR_LINUX_SWAP_PARTITION	= 0x82,
+	MBR_SOLARIS_X86_PARTITION	= MBR_LINUX_SWAP_PARTITION,
+	MBR_LINUX_DATA_PARTITION	= 0x83,
+	MBR_OS2_HIDDEN_DRIVE_PARTITION	= 0x84, /* also hibernation MS APM, Intel Rapid Start */
+	MBR_INTEL_HIBERNATION_PARTITION	= MBR_OS2_HIDDEN_DRIVE_PARTITION,
+	MBR_LINUX_EXTENDED_PARTITION	= 0x85,
+	MBR_NTFS_VOL_SET1_PARTITION	= 0x86,
+	MBR_NTFS_VOL_SET2_PARTITION	= 0x87,
+	MBR_LINUX_PLAINTEXT_PARTITION	= 0x88,
+	MBR_LINUX_LVM_PARTITION		= 0x8e,
+	MBR_AMOEBA_PARTITION		= 0x93,
+	MBR_AMOEBA_BBT_PARTITION	= 0x94, /* (bad block table) */
+	MBR_BSD_OS_PARTITION		= 0x9f, /* BSDI */
+	MBR_THINKPAD_HIBERNATION_PARTITION = 0xa0,
+	MBR_FREEBSD_PARTITION		= 0xa5, /* various BSD flavours */
+	MBR_OPENBSD_PARTITION		= 0xa6,
+	MBR_NEXTSTEP_PARTITION		= 0xa7,
+	MBR_DARWIN_UFS_PARTITION	= 0xa8,
+	MBR_NETBSD_PARTITION		= 0xa9,
+	MBR_DARWIN_BOOT_PARTITION	= 0xab,
+	MBR_HFS_HFS_PARTITION		= 0xaf,
+	MBR_BSDI_FS_PARTITION		= 0xb7,
+	MBR_BSDI_SWAP_PARTITION		= 0xb8,
+	MBR_BOOTWIZARD_HIDDEN_PARTITION	= 0xbb,
+	MBR_ACRONIS_FAT32LBA_PARTITION  = 0xbc, /* Acronis Secure Zone with ipl for loader F11.SYS */
+	MBR_SOLARIS_BOOT_PARTITION	= 0xbe,
+	MBR_SOLARIS_PARTITION		= 0xbf,
+	MBR_DRDOS_FAT12_PARTITION	= 0xc1,
+	MBR_DRDOS_FAT16_L32M_PARTITION	= 0xc4,
+	MBR_DRDOS_FAT16_PARTITION	= 0xc6,
+	MBR_SYRINX_PARTITION		= 0xc7,
+	MBR_NONFS_DATA_PARTITION	= 0xda,
+	MBR_CPM_CTOS_PARTITION		= 0xdb, /* CP/M or Concurrent CP/M or Concurrent DOS or CTOS */
+	MBR_DELL_UTILITY_PARTITION	= 0xde, /* Dell PowerEdge Server utilities */
+	MBR_BOOTIT_PARTITION		= 0xdf, /* BootIt EMBRM */
+	MBR_DOS_ACCESS_PARTITION	= 0xe1, /* DOS access or SpeedStor 12-bit FAT extended partition */
+	MBR_DOS_RO_PARTITION		= 0xe3, /* DOS R/O or SpeedStor */
+	MBR_SPEEDSTOR_EXTENDED_PARTITION = 0xe4, /* SpeedStor 16-bit FAT extended partition < 1024 cyl. */
+	MBR_RUFUS_EXTRA_PARTITION	= 0xea, /* Rufus extra partition for alignment */
+	MBR_BEOS_FS_PARTITION		= 0xeb,
+	MBR_GPT_PARTITION		= 0xee, /* Intel EFI GUID Partition Table */
+	MBR_EFI_SYSTEM_PARTITION	= 0xef, /* Intel EFI System Partition */
+	MBR_LINUX_PARISC_BOOT_PARTITION	= 0xf0, /* Linux/PA-RISC boot loader */
+	MBR_SPEEDSTOR1_PARTITION	= 0xf1,
+	MBR_SPEEDSTOR2_PARTITION	= 0xf4, /* SpeedStor large partition */
+	MBR_DOS_SECONDARY_PARTITION	= 0xf2, /* DOS 3.3+ secondary */
+	MBR_EBBR_PROTECTIVE_PARTITION	= 0xf8, /* Arm EBBR firmware protective partition */
+	MBR_VMWARE_VMFS_PARTITION	= 0xfb,
+	MBR_VMWARE_VMKCORE_PARTITION	= 0xfc, /* VMware kernel dump partition */
+	MBR_LINUX_RAID_PARTITION	= 0xfd, /* Linux raid partition with autodetect using persistent superblock */
+	MBR_LANSTEP_PARTITION		= 0xfe, /* SpeedStor >1024 cyl. or LANstep */
+	MBR_XENIX_BBT_PARTITION		= 0xff, /* Xenix Bad Block Table */
+};
+
+#endif /* UTIL_LINUX_PT_MBR_H */
diff --git a/include/pt-sgi.h b/include/pt-sgi.h
new file mode 100644
index 0000000..3bfba84
--- /dev/null
+++ b/include/pt-sgi.h
@@ -0,0 +1,119 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_PT_SGI_H
+#define UTIL_LINUX_PT_SGI_H
+
+#include <stdint.h>
+
+#define	SGI_LABEL_MAGIC		0x0be5a941
+
+#define SGI_MAXPARTITIONS	16
+#define SGI_MAXVOLUMES		15
+
+/* partition types */
+enum {
+	SGI_TYPE_VOLHDR		= 0x00,
+	SGI_TYPE_TRKREPL	= 0x01,
+	SGI_TYPE_SECREPL	= 0x02,
+	SGI_TYPE_SWAP		= 0x03,
+	SGI_TYPE_BSD		= 0x04,
+	SGI_TYPE_SYSV		= 0x05,
+	SGI_TYPE_ENTIRE_DISK	= 0x06,
+	SGI_TYPE_EFS		= 0x07,
+	SGI_TYPE_LVOL		= 0x08,
+	SGI_TYPE_RLVOL		= 0x09,
+	SGI_TYPE_XFS		= 0x0a,
+	SGI_TYPE_XFSLOG		= 0x0b,
+	SGI_TYPE_XLV		= 0x0c,
+	SGI_TYPE_XVM		= 0x0d
+};
+
+struct sgi_device_parameter {
+	unsigned char skew;
+	unsigned char gap1;
+	unsigned char gap2;
+	unsigned char sparecyl;
+
+	uint16_t pcylcount;
+	uint16_t head_vol0;
+	uint16_t ntrks;		/* tracks in cyl 0 or vol 0 */
+
+	unsigned char cmd_tag_queue_depth;
+	unsigned char unused0;
+
+	uint16_t unused1;
+	uint16_t nsect;		/* sectors/tracks in cyl 0 or vol 0 */
+	uint16_t bytes;
+	uint16_t ilfact;
+	uint32_t flags;		/* SGI_DEVPARAM_* controller flags */
+	uint32_t datarate;
+	uint32_t retries_on_error;
+	uint32_t ms_per_word;
+	uint16_t xylogics_gap1;
+	uint16_t xylogics_syncdelay;
+	uint16_t xylogics_readdelay;
+	uint16_t xylogics_gap2;
+	uint16_t xylogics_readgate;
+	uint16_t xylogics_writecont;
+} __attribute__((packed));
+
+enum {
+	SGI_DEVPARAM_SECTOR_SLIP	= 0x01,
+	SGI_DEVPARAM_SECTOR_FWD		= 0x02,
+	SGI_DEVPARAM_TRACK_FWD		= 0x04,
+	SGI_DEVPARAM_TRACK_MULTIVOL	= 0x08,
+	SGI_DEVPARAM_IGNORE_ERRORS	= 0x10,
+	SGI_DEVPARAM_RESEEK		= 0x20,
+	SGI_DEVPARAM_CMDTAGQ_ENABLE	= 0x40
+};
+
+
+struct sgi_disklabel {
+	uint32_t magic;			/* magic number */
+	uint16_t root_part_num;		/* # root partition */
+	uint16_t swap_part_num;		/* # swap partition */
+	unsigned char boot_file[16];	/* name of boot file */
+
+	struct sgi_device_parameter	devparam;	/* not used now */
+
+	struct sgi_volume {
+		unsigned char name[8];	/* name of volume */
+		uint32_t block_num;	/* logical block number */
+		uint32_t num_bytes;	/* how big, in bytes */
+	} __attribute__((packed)) volume[SGI_MAXVOLUMES];
+
+	struct sgi_partition {
+		uint32_t num_blocks;	/* size in logical blocks */
+		uint32_t first_block;	/* first logical block */
+		uint32_t type;		/* type of this partition */
+	} __attribute__((packed)) partitions[SGI_MAXPARTITIONS];
+
+	/* checksum is the 32bit 2's complement sum of the disklabel */
+	uint32_t csum;			/* disk label checksum */
+	uint32_t padding;		/* padding */
+} __attribute__((packed));
+
+static inline uint32_t sgi_pt_checksum(struct sgi_disklabel *label)
+{
+	int count;
+	uint32_t sum = 0;
+	unsigned char *ptr = (unsigned char *) label;
+
+	count = sizeof(*label) / sizeof(uint32_t);
+	ptr += sizeof(uint32_t) * (count - 1);
+
+	while (count--) {
+		uint32_t val;
+
+		memcpy(&val, ptr, sizeof(uint32_t));
+		sum -= be32_to_cpu(val);
+
+		ptr -= sizeof(uint32_t);
+	}
+
+	return sum;
+}
+
+#endif /* UTIL_LINUX_PT_SGI_H */
diff --git a/include/pt-sun.h b/include/pt-sun.h
new file mode 100644
index 0000000..acb4b56
--- /dev/null
+++ b/include/pt-sun.h
@@ -0,0 +1,94 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_PT_SUN_H
+#define UTIL_LINUX_PT_SUN_H
+
+#include <stdint.h>
+
+#define SUN_LABEL_MAGIC		0xDABE
+
+/* Supported VTOC setting */
+#define SUN_VTOC_SANITY		0x600DDEEE	/* magic number */
+#define SUN_VTOC_VERSION	1
+#define SUN_MAXPARTITIONS	8
+
+struct sun_disklabel {
+	unsigned char label_id[128];   /* Informative text string */
+
+	struct sun_vtoc {
+		uint32_t version;     /* version */
+		char	 volume_id[8];/* volume name */
+		uint16_t nparts;      /* num of partitions */
+
+		struct sun_info {        /* partition information */
+			uint16_t id;     /* SUN_TAG_*  */
+			uint16_t flags;  /* SUN_FLAG_* */
+		} __attribute__ ((packed)) infos[8];
+
+		uint16_t padding;      /* padding */
+		uint32_t bootinfo[3];  /* info needed by mboot */
+		uint32_t sanity;       /* magic number */
+		uint32_t reserved[10]; /* padding */
+		uint32_t timestamp[8]; /* partition timestamp */
+	} __attribute__ ((packed)) vtoc;
+
+	uint32_t write_reinstruct;     /* sectors to skip, writes */
+	uint32_t read_reinstruct;      /* sectors to skip, reads */
+	unsigned char spare[148];      /* padding */
+	uint16_t rpm;                  /* disk rotational speed */
+	uint16_t pcyl;                 /* physical cylinder count */
+	uint16_t apc;                  /* extra sects per cylinder */
+	uint16_t obs1;
+	uint16_t obs2;
+	uint16_t intrlv;               /* interleave factor */
+	uint16_t ncyl;                 /* data cylinder count */
+	uint16_t acyl;                 /* alt. cylinder count */
+	uint16_t nhead;                /* tracks per cylinder   <---- */
+	uint16_t nsect;                /* sectors per track     <---- */
+	uint16_t obs3;
+	uint16_t obs4;
+
+	struct sun_partition {         /* partitions */
+		uint32_t start_cylinder;
+		uint32_t num_sectors;
+	} __attribute__ ((packed)) partitions[8];
+
+	uint16_t magic;                /* magic number */
+	uint16_t csum;                 /* label xor'd checksum */
+} __attribute__ ((packed));
+
+
+#define SUN_TAG_UNASSIGNED	0x00	/* Unassigned partition */
+#define SUN_TAG_BOOT		0x01	/* Boot partition	*/
+#define SUN_TAG_ROOT		0x02	/* Root filesystem	*/
+#define SUN_TAG_SWAP		0x03	/* Swap partition	*/
+#define SUN_TAG_USR		0x04	/* /usr filesystem	*/
+#define SUN_TAG_WHOLEDISK	0x05	/* Full-disk slice	*/
+#define SUN_TAG_STAND		0x06	/* Stand partition	*/
+#define SUN_TAG_VAR		0x07	/* /var filesystem	*/
+#define SUN_TAG_HOME		0x08	/* /home filesystem	*/
+#define SUN_TAG_ALTSCTR		0x09	/* Alt sector partition	*/
+#define SUN_TAG_CACHE		0x0a	/* Cachefs partition	*/
+#define SUN_TAG_RESERVED	0x0b	/* SMI reserved data	*/
+#define SUN_TAG_LINUX_SWAP	0x82	/* Linux SWAP		*/
+#define SUN_TAG_LINUX_NATIVE	0x83	/* Linux filesystem	*/
+#define SUN_TAG_LINUX_LVM	0x8e	/* Linux LVM		*/
+#define SUN_TAG_LINUX_RAID	0xfd	/* LInux RAID		*/
+
+#define SUN_FLAG_UNMNT		0x01	/* Unmountable partition*/
+#define SUN_FLAG_RONLY		0x10	/* Read only		*/
+
+static inline uint16_t sun_pt_checksum(const struct sun_disklabel *label)
+{
+	const uint16_t *ptr = ((const uint16_t *) (label + 1)) - 1;
+	uint16_t sum;
+
+	for (sum = 0; ptr >= ((const uint16_t *) label);)
+		sum ^= *ptr--;
+
+	return sum;
+}
+
+#endif /* UTIL_LINUX_PT_SUN_H */
diff --git a/include/pty-session.h b/include/pty-session.h
new file mode 100644
index 0000000..bece863
--- /dev/null
+++ b/include/pty-session.h
@@ -0,0 +1,120 @@
+/*
+ * This code is in the public domain; do with it what you wish.
+ *
+ * Written by Karel Zak <kzak@redhat.com> in Jul 2019
+ */
+#ifndef UTIL_LINUX_PTY_SESSION_H
+#define UTIL_LINUX_PTY_SESSION_H
+
+#include <pty.h>
+#include <termios.h>
+#include <signal.h>
+#include <sys/time.h>
+#include <sys/stat.h>
+
+#include <sys/signalfd.h>
+
+/*
+ * Callbacks -- the first argument is always callback data, see
+ * ul_pty_set_callback_data().
+ */
+struct ul_pty_callbacks {
+	/*
+	 * Optional. Executed on SIGCHLD when ssi_code is EXITED, KILLED or
+	 * DUMPED; The callback has to call ul_pty_set_child(pty, (pid_t) -1)
+	 * if child is no more alive.
+	 */
+	void (*child_wait)(void *, pid_t);
+
+	/*
+	 * Used when child_wait() undefined to informa about child status
+	 */
+	void (*child_die)(void *, pid_t, int);
+
+	/*
+	 * Executed on SIGCHLD when ssi_status is SIGSTOP
+	 */
+	void (*child_sigstop)(void *, pid_t);
+
+	/*
+	 * Executed in master loop before ul_pty enter poll() and in time set by
+	 * ul_pty_set_mainloop_time(). The callback is no used when time is not set.
+	 */
+	int (*mainloop)(void *);
+
+	/*
+	 * Executed on master or stdin activity, arguments:
+	 *   2nd - file descriptor
+	 *   3rd - buffer with data
+	 *   4th - size of the data
+	 */
+	int (*log_stream_activity)(void *, int, char *, size_t);
+
+	/*
+	 * Executed on signal, arguments:
+	 *   2nd - signal info
+	 *   3rd - NULL or signal specific data (e.g. struct winsize on SIGWINCH
+	 */
+	int (*log_signal)(void *, struct signalfd_siginfo *, void *);
+
+	/*
+	 * Executed on SIGUSR1
+	 */
+	int (*flush_logs)(void *);
+};
+
+struct ul_pty {
+	struct termios	stdin_attrs;	/* stdin and slave terminal runtime attributes */
+	int		master;		/* parent side */
+	int		slave;		/* child side */
+	int		sigfd;		/* signalfd() */
+	int		poll_timeout;
+	struct winsize	win;		/* terminal window size */
+	sigset_t	orgsig;		/* original signal mask */
+
+	int		delivered_signal;
+
+	struct ul_pty_callbacks	callbacks;
+	void			*callback_data;
+
+	pid_t		child;
+
+	struct timeval	next_callback_time;
+
+	struct ul_pty_child_buffer {
+		struct ul_pty_child_buffer *next;
+		char buf[BUFSIZ];
+		size_t size, cursor;
+		unsigned int final_input:1;	/* drain child before writing */
+	} *child_buffer_head, *child_buffer_tail, *free_buffers;
+
+	unsigned int isterm:1,		/* is stdin terminal? */
+		     slave_echo:1;	/* keep ECHO on pty slave */
+};
+
+void ul_pty_init_debug(int mask);
+struct ul_pty *ul_new_pty(int is_stdin_tty);
+void ul_free_pty(struct ul_pty *pty);
+
+void ul_pty_slave_echo(struct ul_pty *pty, int enable);
+int ul_pty_get_delivered_signal(struct ul_pty *pty);
+
+void ul_pty_set_callback_data(struct ul_pty *pty, void *data);
+void ul_pty_set_child(struct ul_pty *pty, pid_t child);
+
+struct ul_pty_callbacks *ul_pty_get_callbacks(struct ul_pty *pty);
+int ul_pty_is_running(struct ul_pty *pty);
+int ul_pty_setup(struct ul_pty *pty);
+int ul_pty_signals_setup(struct ul_pty *pty);
+void ul_pty_cleanup(struct ul_pty *pty);
+int ul_pty_chownmod_slave(struct ul_pty *pty, uid_t uid, gid_t gid, mode_t mode);
+void ul_pty_init_slave(struct ul_pty *pty);
+int ul_pty_proxy_master(struct ul_pty *pty);
+
+void ul_pty_set_mainloop_time(struct ul_pty *pty, struct timeval *tv);
+int ul_pty_get_childfd(struct ul_pty *pty);
+void ul_pty_wait_for_child(struct ul_pty *pty);
+pid_t ul_pty_get_child(struct ul_pty *pty);
+void ul_pty_write_eof_to_child(struct ul_pty *pty);
+
+#endif /* UTIL_LINUX_PTY_H */
diff --git a/include/pwdutils.h b/include/pwdutils.h
new file mode 100644
index 0000000..1d44242
--- /dev/null
+++ b/include/pwdutils.h
@@ -0,0 +1,18 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_PWDUTILS_H
+#define UTIL_LINUX_PWDUTILS_H
+
+#include <sys/types.h>
+#include <pwd.h>
+#include <grp.h>
+
+extern struct passwd *xgetpwnam(const char *username, char **pwdbuf);
+extern struct group *xgetgrnam(const char *groupname, char **grpbuf);
+extern struct passwd *xgetpwuid(uid_t uid, char **pwdbuf);
+extern char *xgetlogin(void);
+
+#endif /* UTIL_LINUX_PWDUTILS_H */
+
diff --git a/include/randutils.h b/include/randutils.h
new file mode 100644
index 0000000..a42af98
--- /dev/null
+++ b/include/randutils.h
@@ -0,0 +1,20 @@
+/*
+ * SPDX-License-Identifier: BSD-3-Clause
+ */
+#ifndef UTIL_LINUX_RANDUTILS
+#define UTIL_LINUX_RANDUTILS
+
+#ifdef HAVE_SRANDOM
+#define srand(x)	srandom(x)
+#define rand()		random()
+#endif
+
+/* rand() based */
+extern int rand_get_number(int low_n, int high_n);
+
+/* /dev/urandom based with fallback to rand() */
+extern int random_get_fd(void);
+extern int ul_random_get_bytes(void *buf, size_t nbytes);
+extern const char *random_tell_source(void);
+
+#endif
diff --git a/include/rpmatch.h b/include/rpmatch.h
new file mode 100644
index 0000000..ee24904
--- /dev/null
+++ b/include/rpmatch.h
@@ -0,0 +1,17 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_RPMATCH_H
+#define UTIL_LINUX_RPMATCH_H
+
+#ifndef HAVE_RPMATCH
+#define rpmatch(r) \
+	(*r == 'y' || *r == 'Y' ? 1 : *r == 'n' || *r == 'N' ? 0 : -1)
+#endif
+
+#define RPMATCH_YES	 1
+#define RPMATCH_NO	 0
+#define RPMATCH_INVALID	-1
+
+#endif /* UTIL_LINUX_RPMATCH_H */
diff --git a/include/selinux-utils.h b/include/selinux-utils.h
new file mode 100644
index 0000000..0379561
--- /dev/null
+++ b/include/selinux-utils.h
@@ -0,0 +1,12 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_SELINUX_UTILS_H
+#define UTIL_LINUX_SELINUX_UTILS_H
+
+extern int ul_setfscreatecon_from_file(char *orig_file);
+extern int ul_selinux_has_access(const char *classstr, const char *perm, char **user_cxt);
+extern int ul_selinux_get_default_context(const char *path, int st_mode, char **cxt);
+
+#endif
diff --git a/include/sha1.h b/include/sha1.h
new file mode 100644
index 0000000..8ec74c9
--- /dev/null
+++ b/include/sha1.h
@@ -0,0 +1,31 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_SHA1_H
+#define UTIL_LINUX_SHA1_H
+
+/*
+   SHA-1 in C
+   By Steve Reid <steve@edmweb.com>
+   100% Public Domain
+ */
+
+#include "stdint.h"
+
+#define UL_SHA1LENGTH		20
+
+typedef struct
+{
+    uint32_t	state[5];
+    uint32_t	count[2];
+    unsigned	char buffer[64];
+} UL_SHA1_CTX;
+
+void ul_SHA1Transform(uint32_t state[5], const unsigned char buffer[64]);
+void ul_SHA1Init(UL_SHA1_CTX *context);
+void ul_SHA1Update(UL_SHA1_CTX *context, const unsigned char *data, uint32_t len);
+void ul_SHA1Final(unsigned char digest[UL_SHA1LENGTH], UL_SHA1_CTX *context);
+void ul_SHA1(char *hash_out, const char *str, unsigned len);
+
+#endif /* UTIL_LINUX_SHA1_H */
diff --git a/include/sha256.h b/include/sha256.h
new file mode 100644
index 0000000..bf878e5
--- /dev/null
+++ b/include/sha256.h
@@ -0,0 +1,14 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_SHA256_H
+#define UTIL_LINUX_SHA256_H
+
+#include <stddef.h>
+
+#define UL_SHA256LENGTH		32
+
+extern void ul_SHA256(unsigned char hash_out[UL_SHA256LENGTH], const unsigned char *str, size_t len);
+
+#endif /* UTIL_LINUX_SHA256_H */
diff --git a/include/shells.h b/include/shells.h
new file mode 100644
index 0000000..ba59ba4
--- /dev/null
+++ b/include/shells.h
@@ -0,0 +1,9 @@
+/*
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ */
+#ifndef UTIL_LINUX_SHELLS_H
+#define UTIL_LINUX_SHELLS_H
+
+extern int is_known_shell(const char *shell_name);
+
+#endif /* UTIL_LINUX_SHELLS_H */
diff --git a/include/signames.h b/include/signames.h
new file mode 100644
index 0000000..3f1043a
--- /dev/null
+++ b/include/signames.h
@@ -0,0 +1,12 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef SIGNAMES_H
+#define SIGNAMES_H
+
+int signame_to_signum(const char *sig);
+const char *signum_to_signame(int signum);
+int get_signame_by_idx(size_t idx, const char **signame, int *signum);
+
+#endif /* SIGNAMES_H */
diff --git a/include/statfs_magic.h b/include/statfs_magic.h
new file mode 100644
index 0000000..e87b784
--- /dev/null
+++ b/include/statfs_magic.h
@@ -0,0 +1,106 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_STATFS_MAGIC_H
+#define UTIL_LINUX_STATFS_MAGIC_H
+
+#ifdef HAVE_SYS_STATFS_H
+# include <sys/statfs.h>
+#endif
+
+/*
+ * If possible then don't depend on internal libc __SWORD_TYPE type.
+ */
+#ifdef __GNUC__
+#define F_TYPE_EQUAL(a, b) (a == (__typeof__(a)) b)
+#else
+#define F_TYPE_EQUAL(a, b) (a == (__SWORD_TYPE) b)
+#endif
+
+/*
+ *  Unfortunately, Linux kernel header file <linux/magic.h> is incomplete
+ *  mess and kernel returns by statfs f_type many numbers that are nowhere
+ *  specified (in API).
+ *
+ *  This is collection of the magic numbers.
+ */
+#define STATFS_ADFS_MAGIC	0xadf5
+#define STATFS_AFFS_MAGIC	0xadff
+#define STATFS_AFS_MAGIC	0x5346414F
+#define STATFS_AUTOFS_MAGIC	0x0187
+#define STATFS_BDEVFS_MAGIC	0x62646576
+#define STATFS_BEFS_MAGIC	0x42465331
+#define STATFS_BFS_MAGIC	0x1BADFACE
+#define STATFS_BINFMTFS_MAGIC	0x42494e4d
+#define STATFS_BTRFS_MAGIC	0x9123683E
+#define STATFS_CEPH_MAGIC	0x00c36400
+#define STATFS_CGROUP_MAGIC	0x27e0eb
+#define STATFS_CGROUP2_MAGIC	0x63677270
+#define STATFS_CIFS_MAGIC	0xff534d42
+#define STATFS_CODA_MAGIC	0x73757245
+#define STATFS_CONFIGFS_MAGIC	0x62656570
+#define STATFS_CRAMFS_MAGIC	0x28cd3d45
+#define STATFS_DEBUGFS_MAGIC	0x64626720
+#define STATFS_DEVPTS_MAGIC	0x1cd1
+#define STATFS_ECRYPTFS_MAGIC	0xf15f
+#define STATFS_EFIVARFS_MAGIC	0xde5e81e4
+#define STATFS_EFS_MAGIC	0x414A53
+#define STATFS_EXOFS_MAGIC	0x5DF5
+#define STATFS_EXT2_MAGIC	0xEF53
+#define STATFS_EXT3_MAGIC	0xEF53
+#define STATFS_EXT4_MAGIC	0xEF53
+#define STATFS_F2FS_MAGIC	0xF2F52010
+#define STATFS_FUSE_MAGIC	0x65735546
+#define STATFS_FUTEXFS_MAGIC	0xBAD1DEA
+#define STATFS_GFS2_MAGIC	0x01161970
+#define STATFS_HFSPLUS_MAGIC	0x482b
+#define STATFS_HOSTFS_MAGIC	0x00c0ffee
+#define STATFS_HPFS_MAGIC	0xf995e849
+#define STATFS_HPPFS_MAGIC	0xb00000ee
+#define STATFS_HUGETLBFS_MAGIC	0x958458f6
+#define STATFS_ISOFS_MAGIC	0x9660
+#define STATFS_JFFS2_MAGIC	0x72b6
+#define STATFS_JFS_MAGIC	0x3153464a
+#define STATFS_LOGFS_MAGIC	0xc97e8168
+#define STATFS_MINIX2_MAGIC	0x2468
+#define STATFS_MINIX2_MAGIC2	0x2478
+#define STATFS_MINIX3_MAGIC	0x4d5a
+#define STATFS_MINIX_MAGIC	0x137F
+#define STATFS_MINIX_MAGIC2	0x138F
+#define STATFS_MQUEUE_MAGIC	0x19800202
+#define STATFS_MSDOS_MAGIC	0x4d44
+#define STATFS_NCP_MAGIC	0x564c
+#define STATFS_NFS_MAGIC	0x6969
+#define STATFS_NILFS_MAGIC	0x3434
+#define STATFS_NTFS_MAGIC	0x5346544e
+#define STATFS_OCFS2_MAGIC	0x7461636f
+#define STATFS_OMFS_MAGIC	0xC2993D87
+#define STATFS_OPENPROMFS_MAGIC	0x9fa1
+#define STATFS_PIPEFS_MAGIC	0x50495045
+#define STATFS_PROC_MAGIC	0x9fa0
+#define STATFS_PSTOREFS_MAGIC	0x6165676C
+#define STATFS_QNX4_MAGIC	0x002f
+#define STATFS_QNX6_MAGIC	0x68191122
+#define STATFS_RAMFS_MAGIC	0x858458f6
+#define STATFS_REISERFS_MAGIC	0x52654973
+#define STATFS_ROMFS_MAGIC	0x7275
+#define STATFS_SECURITYFS_MAGIC	0x73636673
+#define STATFS_SELINUXFS_MAGIC	0xf97cff8c
+#define STATFS_SMACKFS_MAGIC	0x43415d53
+#define STATFS_SMB_MAGIC	0x517B
+#define STATFS_SOCKFS_MAGIC	0x534F434B
+#define STATFS_SQUASHFS_MAGIC	0x73717368
+#define STATFS_SYSFS_MAGIC	0x62656572
+#define STATFS_TMPFS_MAGIC	0x01021994
+#define STATFS_UBIFS_MAGIC	0x24051905
+#define STATFS_UDF_MAGIC	0x15013346
+#define STATFS_UFS2_MAGIC	0x19540119
+#define STATFS_UFS_MAGIC	0x00011954
+#define STATFS_V9FS_MAGIC	0x01021997
+#define STATFS_VXFS_MAGIC	0xa501FCF5
+#define STATFS_XENFS_MAGIC	0xabba1974
+#define STATFS_XFS_MAGIC	0x58465342
+
+#endif /* UTIL_LINUX_STATFS_MAGIC_H */
+
diff --git a/include/strutils.h b/include/strutils.h
new file mode 100644
index 0000000..07aa656
--- /dev/null
+++ b/include/strutils.h
@@ -0,0 +1,407 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_STRUTILS
+#define UTIL_LINUX_STRUTILS
+
+#include <stdlib.h>
+#include <inttypes.h>
+#include <string.h>
+#include <sys/types.h>
+#include <ctype.h>
+#include <stdio.h>
+#include <errno.h>
+#include <time.h>
+
+#include "c.h"
+
+/* initialize a custom exit code for all *_or_err functions */
+extern void strutils_set_exitcode(int exit_code);
+
+extern int parse_size(const char *str, uintmax_t *res, int *power);
+extern int strtosize(const char *str, uintmax_t *res);
+extern uintmax_t strtosize_or_err(const char *str, const char *errmesg);
+
+extern int ul_strtos64(const char *str, int64_t *num, int base);
+extern int ul_strtou64(const char *str, uint64_t *num, int base);
+extern int ul_strtos32(const char *str, int32_t *num, int base);
+extern int ul_strtou32(const char *str, uint32_t *num, int base);
+
+extern int64_t str2num_or_err(const char *str, int base, const char *errmesg, int64_t low, int64_t up);
+extern uint64_t str2unum_or_err(const char *str, int base, const char *errmesg, uint64_t up);
+
+#define strtos64_or_err(_s, _e)	str2num_or_err(_s, 10, _e, 0, 0)
+#define strtou64_or_err(_s, _e)	str2unum_or_err(_s, 10, _e, 0)
+#define strtox64_or_err(_s, _e)	str2unum_or_err(_s, 16, _e, 0)
+
+#define strtos32_or_err(_s, _e)	(int32_t) str2num_or_err(_s, 10, _e, INT32_MIN, INT32_MAX)
+#define strtou32_or_err(_s, _e)	(uint32_t) str2unum_or_err(_s, 10, _e, UINT32_MAX)
+#define strtox32_or_err(_s, _e)	(uint32_t) str2unum_or_err(_s, 16, _e, UINT32_MAX)
+
+#define strtos16_or_err(_s, _e)	(int16_t) str2num_or_err(_s, 10, _e, INT16_MIN, INT16_MAX)
+#define strtou16_or_err(_s, _e)	(uint16_t) str2unum_or_err(_s, 10, _e, UINT16_MAX)
+#define strtox16_or_err(_s, _e)	(uint16_t) str2unum_or_err(_s, 16, _e, UINT16_MAX)
+
+extern double strtod_or_err(const char *str, const char *errmesg);
+extern long double strtold_or_err(const char *str, const char *errmesg);
+
+#define strtol_or_err(_s, _e)	(long) str2num_or_err(_s, 10, _e, LONG_MIN, LONG_MAX)
+#define strtopid_or_err(_s, _e)	(pid_t) str2num_or_err(_s, 10, _e, 1, SINT_MAX(pid_t))
+#define strtoul_or_err(_s, _e)	(unsigned long) str2unum_or_err(_s, 10, _e, ULONG_MAX)
+
+extern void strtotimeval_or_err(const char *str, struct timeval *tv,
+		const char *errmesg);
+extern void strtotimespec_or_err(const char *str, struct timespec *ts,
+		const char *errmesg);
+extern time_t strtotime_or_err(const char *str, const char *errmesg);
+
+extern int isdigit_strend(const char *str, const char **end);
+#define isdigit_string(_s)	isdigit_strend(_s, NULL)
+
+extern int isxdigit_strend(const char *str, const char **end);
+#define isxdigit_string(_s)	isxdigit_strend(_s, NULL)
+
+
+extern int parse_switch(const char *arg, const char *errmesg, ...);
+
+#ifndef HAVE_MEMPCPY
+extern void *mempcpy(void *restrict dest, const void *restrict src, size_t n);
+#endif
+#ifndef HAVE_STRNLEN
+extern size_t strnlen(const char *s, size_t maxlen);
+#endif
+#ifndef HAVE_STRNDUP
+extern char *strndup(const char *s, size_t n);
+#endif
+#ifndef HAVE_STRNCHR
+extern char *strnchr(const char *s, size_t maxlen, int c);
+#endif
+
+/* caller guarantees n > 0 */
+static inline void xstrncpy(char *dest, const char *src, size_t n)
+{
+	size_t len = src ? strlen(src) : 0;
+
+	if (!len)
+		return;
+	len = min(len, n - 1);
+	memcpy(dest, src, len);
+	dest[len] = 0;
+}
+
+/* This is like strncpy(), but based on memcpy(), so compilers and static
+ * analyzers do not complain when sizeof(destination) is the same as 'n' and
+ * result is not terminated by zero.
+ *
+ * Use this function to copy string to logs with fixed sizes (wtmp/utmp. ...)
+ * where string terminator is optional.
+ */
+static inline void * __attribute__((nonnull (1)))
+str2memcpy(void *dest, const char *src, size_t n)
+{
+	size_t bytes = strlen(src) + 1;
+
+	if (bytes > n)
+		bytes = n;
+
+	memcpy(dest, src, bytes);
+	return dest;
+}
+
+static inline char * __attribute__((nonnull (1)))
+mem2strcpy(char *dest, const void *src, size_t n, size_t nmax)
+{
+	if (n + 1 > nmax)
+		n = nmax - 1;
+
+	memset(dest, '\0', nmax);
+	memcpy(dest, src, n);
+	return dest;
+}
+
+/* Reallocate @str according to @newstr and copy @newstr to @str; returns new @str.
+ * The @str is not modified if reallocation failed (like classic realloc()).
+ */
+static inline char * __attribute__((warn_unused_result))
+strrealloc(char *str, const char *newstr)
+{
+	size_t nsz, osz;
+
+	if (!str)
+		return newstr ? strdup(newstr) : NULL;
+	if (!newstr)
+		return NULL;
+
+	osz = strlen(str);
+	nsz = strlen(newstr);
+
+	if (nsz > osz)
+		str = realloc(str, nsz + 1);
+	if (str)
+		memcpy(str, newstr, nsz + 1);
+	return str;
+}
+
+/* Copy string @str to struct @stru to member addressed by @offset */
+static inline int strdup_to_offset(void *stru, size_t offset, const char *str)
+{
+	char **o;
+	char *p = NULL;
+
+	if (!stru)
+		return -EINVAL;
+
+	o = (char **) ((char *) stru + offset);
+	if (str) {
+		p = strdup(str);
+		if (!p)
+			return -ENOMEM;
+	}
+
+	free(*o);
+	*o = p;
+	return 0;
+}
+
+/* Copy string __str to struct member _m of the struct _s */
+#define strdup_to_struct_member(_s, _m, _str) \
+		strdup_to_offset((void *) _s, offsetof(__typeof__(*(_s)), _m), _str)
+
+/* Copy string addressed by @offset between two structs */
+static inline int strdup_between_offsets(void *stru_dst, void *stru_src, size_t offset)
+{
+	char **src;
+	char **dst;
+	char *p = NULL;
+
+	if (!stru_src || !stru_dst)
+		return -EINVAL;
+
+	src = (char **) ((char *) stru_src + offset);
+	dst = (char **) ((char *) stru_dst + offset);
+
+	if (*src) {
+		p = strdup(*src);
+		if (!p)
+			return -ENOMEM;
+	}
+
+	free(*dst);
+	*dst = p;
+	return 0;
+}
+
+/* Copy string addressed by struct member between two instances of the same
+ * struct type */
+#define strdup_between_structs(_dst, _src, _m) \
+		strdup_between_offsets((void *)_dst, (void *)_src, offsetof(__typeof__(*(_src)), _m))
+
+
+extern char *xstrmode(mode_t mode, char *str);
+
+/* Options for size_to_human_string() */
+enum
+{
+	SIZE_SUFFIX_1LETTER  = 0,
+	SIZE_SUFFIX_3LETTER  = (1 << 0),
+	SIZE_SUFFIX_SPACE    = (1 << 1),
+	SIZE_DECIMAL_2DIGITS = (1 << 2)
+};
+
+extern char *size_to_human_string(int options, uint64_t bytes);
+
+extern int string_to_idarray(const char *list, int ary[], size_t arysz,
+			   int (name2id)(const char *, size_t));
+extern int string_add_to_idarray(const char *list, int ary[],
+				 size_t arysz, size_t *ary_pos,
+				 int (name2id)(const char *, size_t));
+
+extern int string_to_bitarray(const char *list, char *ary,
+			    int (*name2bit)(const char *, size_t),
+			    size_t allow_range);
+
+extern int string_to_bitmask(const char *list,
+			     unsigned long *mask,
+			     long (*name2flag)(const char *, size_t));
+extern int parse_range(const char *str, int *lower, int *upper, int def);
+
+extern int streq_paths(const char *a, const char *b);
+
+/*
+ * Match string beginning.
+ */
+static inline const char *startswith(const char *s, const char *prefix)
+{
+	size_t sz = prefix ? strlen(prefix) : 0;
+
+        if (s && sz && strncmp(s, prefix, sz) == 0)
+                return s + sz;
+	return NULL;
+}
+
+/*
+ * Case insensitive match string beginning.
+ */
+static inline const char *startswith_no_case(const char *s, const char *prefix)
+{
+	size_t sz = prefix ? strlen(prefix) : 0;
+
+        if (s && sz && strncasecmp(s, prefix, sz) == 0)
+                return s + sz;
+	return NULL;
+}
+
+/*
+ * Match string ending.
+ */
+static inline const char *endswith(const char *s, const char *postfix)
+{
+	size_t sl = s ? strlen(s) : 0;
+	size_t pl = postfix ? strlen(postfix) : 0;
+
+	if (pl == 0)
+		return s + sl;
+	if (sl < pl)
+		return NULL;
+	if (memcmp(s + sl - pl, postfix, pl) != 0)
+		return NULL;
+	return s + sl - pl;
+}
+
+/*
+ * Skip leading white space.
+ */
+static inline const char *skip_space(const char *p)
+{
+	while (isspace(*p))
+		++p;
+	return p;
+}
+
+static inline const char *skip_blank(const char *p)
+{
+	while (isblank(*p))
+		++p;
+	return p;
+}
+
+
+/* Removes whitespace from the right-hand side of a string (trailing
+ * whitespace).
+ *
+ * Returns size of the new string (without \0).
+ */
+static inline size_t rtrim_whitespace(unsigned char *str)
+{
+	size_t i;
+
+	if (!str)
+		return 0;
+	i = strlen((char *) str);
+	while (i) {
+		i--;
+		if (!isspace(str[i])) {
+			i++;
+			break;
+		}
+	}
+	str[i] = '\0';
+	return i;
+}
+
+/* Removes whitespace from the left-hand side of a string.
+ *
+ * Returns size of the new string (without \0).
+ */
+static inline size_t ltrim_whitespace(unsigned char *str)
+{
+	size_t len;
+	unsigned char *p;
+
+	if (!str)
+		return 0;
+	for (p = str; *p && isspace(*p); p++);
+
+	len = strlen((char *) p);
+
+	if (p > str)
+		memmove(str, p, len + 1);
+
+	return len;
+}
+
+/* Removes left-hand, right-hand and repeating whitespaces.
+ */
+static inline size_t __normalize_whitespace(
+				const unsigned char *src,
+				size_t sz,
+				unsigned char *dst,
+				size_t len)
+{
+	size_t i, x = 0;
+	int nsp = 0, intext = 0;
+
+	if (!sz)
+		goto done;
+
+	for (i = 0, x = 0; i < sz && x < len - 1;  ) {
+		if (isspace(src[i]))
+			nsp++;
+		else
+			nsp = 0, intext = 1;
+
+		if (nsp > 1 || (nsp && !intext))
+			i++;
+		else
+			dst[x++] = src[i++];
+	}
+	if (nsp && x > 0)		/* tailing space */
+		x--;
+done:
+	dst[x] = '\0';
+	return x;
+}
+
+static inline size_t normalize_whitespace(unsigned char *str)
+{
+	size_t sz = strlen((char *) str);
+	return __normalize_whitespace(str, sz, str, sz + 1);
+}
+
+static inline void strrep(char *s, int find, int replace)
+{
+	while (s && *s && (s = strchr(s, find)) != NULL)
+		*s++ = replace;
+}
+
+static inline void strrem(char *s, int rem)
+{
+	char *p;
+
+	if (!s)
+		return;
+	for (p = s; *s; s++) {
+		if (*s != rem)
+			*p++ = *s;
+	}
+	*p = '\0';
+}
+
+extern char *strnconcat(const char *s, const char *suffix, size_t b);
+extern char *strconcat(const char *s, const char *suffix);
+extern char *strfconcat(const char *s, const char *format, ...)
+		 __attribute__ ((__format__ (__printf__, 2, 3)));
+
+extern int strappend(char **a, const char *b);
+
+extern const char *split(const char **state, size_t *l, const char *separator, int quoted);
+
+extern char *ul_strchr_escaped(const char *s, int c);
+
+extern int skip_fline(FILE *fp);
+extern int ul_stralnumcmp(const char *p1, const char *p2);
+
+extern int ul_optstr_next(char **optstr, char **name, size_t *namesz, char **value, size_t *valsz);
+
+#endif
diff --git a/include/strv.h b/include/strv.h
new file mode 100644
index 0000000..6091fce
--- /dev/null
+++ b/include/strv.h
@@ -0,0 +1,61 @@
+/*
+ * SPDX-License-Identifier: LGPL-2.1-or-later
+ */
+#ifndef UTIL_LINUX_STRV
+#define UTIL_LINUX_STRV
+
+#include <stdarg.h>
+
+#include "c.h"
+
+char **strv_free(char **l);
+void strv_clear(char **l);
+char **strv_copy(char * const *l);
+unsigned strv_length(char * const *l);
+
+int strv_extend_strv(char ***a, char **b);
+int strv_extend_strv_concat(char ***a, char **b, const char *suffix);
+int strv_extend(char ***l, const char *value);
+
+int strv_extendv(char ***l, const char *format, va_list ap)
+		__attribute__ ((__format__ (__printf__, 2, 0)));
+int strv_extendf(char ***l, const char *format, ...)
+		__attribute__ ((__format__ (__printf__, 2, 3)));
+
+int strv_push(char ***l, char *value);
+int strv_push_prepend(char ***l, char *value);
+int strv_consume(char ***l, char *value);
+int strv_consume_prepend(char ***l, char *value);
+
+char **strv_remove(char **l, const char *s);
+
+char **strv_new(const char *x, ...);
+char **strv_new_ap(const char *x, va_list ap);
+
+static inline const char* STRV_IFNOTNULL(const char *x) {
+        return x ? x : (const char *) -1;
+}
+
+static inline int strv_isempty(char * const *l) {
+        return !l || !*l;
+}
+
+char **strv_split(const char *s, const char *separator);
+char *strv_join(char **l, const char *separator);
+
+#define STRV_FOREACH(s, l)                      \
+        for ((s) = (l); (s) && *(s); (s)++)
+
+#define STRV_FOREACH_BACKWARDS(s, l)            \
+        STRV_FOREACH(s, l)                      \
+                ;                               \
+        for ((s)--; (l) && ((s) >= (l)); (s)--)
+
+
+#define STRV_MAKE_EMPTY ((char*[1]) { NULL })
+
+char **strv_reverse(char **l);
+
+#endif /* UTIL_LINUX_STRV */
+
+
diff --git a/include/swapheader.h b/include/swapheader.h
new file mode 100644
index 0000000..223bb56
--- /dev/null
+++ b/include/swapheader.h
@@ -0,0 +1,27 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef _SWAPHEADER_H
+#define _SWAPHEADER_H
+
+#define SWAP_VERSION 1
+#define SWAP_UUID_LENGTH 16
+#define SWAP_LABEL_LENGTH 16
+#define SWAP_SIGNATURE "SWAPSPACE2"
+#define SWAP_SIGNATURE_SZ (sizeof(SWAP_SIGNATURE) - 1)
+
+#include <stdint.h>
+
+struct swap_header_v1_2 {
+	char	      bootbits[1024];    /* Space for disklabel etc. */
+	uint32_t      version;
+	uint32_t      last_page;
+	uint32_t      nr_badpages;
+	unsigned char uuid[SWAP_UUID_LENGTH];
+	char	      volume_name[SWAP_LABEL_LENGTH];
+	uint32_t      padding[117];
+	uint32_t      badpages[1];
+};
+
+#endif /* _SWAPHEADER_H */
diff --git a/include/swapprober.h b/include/swapprober.h
new file mode 100644
index 0000000..5107700
--- /dev/null
+++ b/include/swapprober.h
@@ -0,0 +1,9 @@
+#ifndef UTIL_LINUX_SWAP_PROBER_H
+#define UTIL_LINUX_SWAP_PROBER_H
+
+#include <blkid.h>
+
+blkid_probe get_swap_prober(const char *devname);
+
+#endif /* UTIL_LINUX_SWAP_PROBER_H */
+
diff --git a/include/sysfs.h b/include/sysfs.h
new file mode 100644
index 0000000..8d77f75
--- /dev/null
+++ b/include/sysfs.h
@@ -0,0 +1,126 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ *
+ * Copyright (C) 2011 Karel Zak <kzak@redhat.com>
+ */
+#ifndef UTIL_LINUX_SYSFS_H
+#define UTIL_LINUX_SYSFS_H
+
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <stdint.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <fcntl.h>
+#include <string.h>
+#include <inttypes.h>
+#include <dirent.h>
+
+#include "path.h"
+
+/**
+ * sysfs_devname_sys_to_dev:
+ * @name: devname to be converted in place
+ *
+ * Linux kernel linux/drivers/base/core.c: device_get_devnode()
+ * defines a replacement of '!' in the /sys device name by '/' in the
+ * /dev device name. This helper replaces all occurrences of '!' in
+ * @name by '/' to convert from /sys to /dev.
+ */
+static inline void sysfs_devname_sys_to_dev(char *name)
+{
+	char *c;
+
+	if (name)
+		while ((c = strchr(name, '!')))
+			c[0] = '/';
+}
+
+/**
+ * sysfs_devname_dev_to_sys:
+ * @name: devname to be converted in place
+ *
+ * See sysfs_devname_sys_to_dev().
+ */
+static inline void sysfs_devname_dev_to_sys(char *name)
+{
+	char *c;
+
+	if (name)
+		while ((c = strchr(name, '/')))
+			c[0] = '!';
+}
+
+struct sysfs_blkdev {
+	dev_t	devno;
+	struct path_cxt *parent;
+
+	unsigned int	scsi_host,
+			scsi_channel,
+			scsi_target,
+			scsi_lun;
+
+	unsigned int	has_hctl   : 1,
+			hctl_error : 1 ;
+};
+
+void ul_sysfs_init_debug(void);
+
+struct path_cxt *ul_new_sysfs_path(dev_t devno, struct path_cxt *parent, const char *prefix);
+int sysfs_blkdev_init_path(struct path_cxt *pc, dev_t devno, struct path_cxt *parent);
+
+int sysfs_blkdev_set_parent(struct path_cxt *pc, struct path_cxt *parent);
+struct path_cxt *sysfs_blkdev_get_parent(struct path_cxt *pc);
+
+char *sysfs_blkdev_get_name(struct path_cxt *pc, char *buf, size_t bufsiz);
+int sysfs_blkdev_is_partition_dirent(DIR *dir, struct dirent *d, const char *parent_name);
+int sysfs_blkdev_count_partitions(struct path_cxt *pc, const char *devname);
+dev_t sysfs_blkdev_partno_to_devno(struct path_cxt *pc, int partno);
+char *sysfs_blkdev_get_slave(struct path_cxt *pc);
+char *sysfs_blkdev_get_path(struct path_cxt *pc, char *buf, size_t bufsiz);
+dev_t sysfs_blkdev_get_devno(struct path_cxt *pc);
+
+char *sysfs_blkdev_get_devchain(struct path_cxt *pc, char *buf, size_t bufsz);
+int sysfs_blkdev_next_subsystem(struct path_cxt *pc __attribute__((unused)), char *devchain, char **subsys);
+
+int sysfs_blkdev_is_hotpluggable(struct path_cxt *pc);
+int sysfs_blkdev_is_removable(struct path_cxt *pc);
+int sysfs_blkdev_get_wholedisk( struct path_cxt *pc,
+                                char *diskname,
+                                size_t len,
+                                dev_t *diskdevno);
+
+int sysfs_devno_to_wholedisk(dev_t dev, char *diskname,
+                             size_t len, dev_t *diskdevno);
+int sysfs_devno_is_dm_private(dev_t devno, char **uuid);
+int sysfs_devno_is_wholedisk(dev_t devno);
+
+dev_t sysfs_devname_to_devno(const char *name);
+dev_t __sysfs_devname_to_devno(const char *prefix, const char *name, const char *parent);
+int sysfs_devname_is_hidden(const char *prefix, const char *name);
+
+char *sysfs_devno_to_devpath(dev_t devno, char *buf, size_t bufsiz);
+char *sysfs_devno_to_devname(dev_t devno, char *buf, size_t bufsiz);
+int sysfs_devno_count_partitions(dev_t devno);
+
+int sysfs_blkdev_scsi_get_hctl(struct path_cxt *pc, int *h, int *c, int *t, int *l);
+char *sysfs_blkdev_scsi_host_strdup_attribute(struct path_cxt *pc,
+                        const char *type, const char *attr);
+int sysfs_blkdev_scsi_host_is(struct path_cxt *pc, const char *type);
+int sysfs_blkdev_scsi_has_attribute(struct path_cxt *pc, const char *attr);
+int sysfs_blkdev_scsi_path_contains(struct path_cxt *pc, const char *pattern);
+
+char *sysfs_chrdev_devno_to_devname(dev_t devno, char *buf, size_t bufsiz);
+
+enum sysfs_byteorder {
+	SYSFS_BYTEORDER_LITTLE,
+	SYSFS_BYTEORDER_BIG,
+};
+
+extern enum sysfs_byteorder sysfs_get_byteorder(struct path_cxt *pc);
+extern int sysfs_get_address_bits(struct path_cxt *pc);
+
+#endif /* UTIL_LINUX_SYSFS_H */
diff --git a/include/timer.h b/include/timer.h
new file mode 100644
index 0000000..dd044a3
--- /dev/null
+++ b/include/timer.h
@@ -0,0 +1,26 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+#ifndef UTIL_LINUX_TIMER_H
+#define UTIL_LINUX_TIMER_H
+
+#include <signal.h>
+#include <sys/time.h>
+
+#ifdef HAVE_TIMER_CREATE
+struct ul_timer {
+	timer_t t_id;
+};
+#else
+struct ul_timer {
+	struct itimerval old_timer;
+	struct sigaction old_sa;
+};
+#endif
+
+extern int setup_timer(struct ul_timer *timer, struct itimerval *timeout,
+		       void (*timeout_handler)(int, siginfo_t *, void *));
+extern void cancel_timer(struct ul_timer *timer);
+
+#endif /* UTIL_LINUX_TIMER_H */
diff --git a/include/timeutils.h b/include/timeutils.h
new file mode 100644
index 0000000..a195df8
--- /dev/null
+++ b/include/timeutils.h
@@ -0,0 +1,105 @@
+/*
+ * SPDX-License-Identifier: LGPL-2.1-or-later
+ *
+ * First set of functions in this file are part of systemd, and were
+ * copied to util-linux at August 2013.
+ *
+ * Copyright 2010 Lennart Poettering
+ * Copyright (C) 2014 Karel Zak <kzak@redhat.com>
+ *
+ * This is free software; you can redistribute it and/or modify it under the
+ * terms of the GNU Lesser General Public License as published by the Free
+ * Software Foundation; either version 2.1 of the License, or (at your option)
+ * any later version.
+ */
+#ifndef UTIL_LINUX_TIME_UTIL_H
+#define UTIL_LINUX_TIME_UTIL_H
+
+#include <stdio.h>
+#include <inttypes.h>
+#include <sys/time.h>
+
+typedef uint64_t usec_t;
+typedef uint64_t nsec_t;
+
+#define MSEC_PER_SEC  1000ULL
+#define USEC_PER_SEC  1000000ULL
+#define USEC_PER_MSEC 1000ULL
+#define NSEC_PER_SEC  1000000000ULL
+#define NSEC_PER_MSEC 1000000ULL
+#define NSEC_PER_USEC 1000ULL
+
+#define USEC_PER_MINUTE	(60ULL*USEC_PER_SEC)
+#define NSEC_PER_MINUTE	(60ULL*NSEC_PER_SEC)
+#define USEC_PER_HOUR	(60ULL*USEC_PER_MINUTE)
+#define NSEC_PER_HOUR	(60ULL*NSEC_PER_MINUTE)
+#define USEC_PER_DAY	(24ULL*USEC_PER_HOUR)
+#define NSEC_PER_DAY	(24ULL*NSEC_PER_HOUR)
+#define USEC_PER_WEEK	(7ULL*USEC_PER_DAY)
+#define NSEC_PER_WEEK	(7ULL*NSEC_PER_DAY)
+#define USEC_PER_MONTH	(2629800ULL*USEC_PER_SEC)
+#define NSEC_PER_MONTH	(2629800ULL*NSEC_PER_SEC)
+#define USEC_PER_YEAR	(31557600ULL*USEC_PER_SEC)
+#define NSEC_PER_YEAR	(31557600ULL*NSEC_PER_SEC)
+
+#define FORMAT_TIMESTAMP_MAX ((4*4+1)+11+9+4+1)	/* weekdays can be unicode */
+#define FORMAT_TIMESTAMP_RELATIVE_MAX 256
+#define FORMAT_TIMESPAN_MAX 64
+
+int parse_timestamp(const char *t, usec_t *usec);
+int get_gmtoff(const struct tm *tp);
+
+/* flags and masks for strxxx_iso() functions */
+enum {
+	ISO_DATE		= (1 << 0),
+	ISO_TIME		= (1 << 1),
+	ISO_TIMEZONE		= (1 << 2),
+	ISO_DOTUSEC		= (1 << 3),
+	ISO_COMMAUSEC		= (1 << 4),
+	ISO_T			= (1 << 5),
+	ISO_GMTIME		= (1 << 6),
+	ISO_TIMESTAMP		= ISO_DATE | ISO_TIME | ISO_TIMEZONE,
+	ISO_TIMESTAMP_T		= ISO_TIMESTAMP | ISO_T,
+	ISO_TIMESTAMP_DOT	= ISO_TIMESTAMP | ISO_DOTUSEC,
+	ISO_TIMESTAMP_DOT_T	= ISO_TIMESTAMP_DOT | ISO_T,
+	ISO_TIMESTAMP_COMMA	= ISO_TIMESTAMP | ISO_COMMAUSEC,
+	ISO_TIMESTAMP_COMMA_T	= ISO_TIMESTAMP_COMMA | ISO_T,
+	ISO_TIMESTAMP_COMMA_G	= ISO_TIMESTAMP_COMMA | ISO_GMTIME,
+	ISO_TIMESTAMP_COMMA_GT  = ISO_TIMESTAMP_COMMA_G | ISO_T
+};
+
+#define CTIME_BUFSIZ	26
+#define ISO_BUFSIZ	42
+
+int strtimeval_iso(struct timeval *tv, int flags, char *buf, size_t bufsz);
+int strtm_iso(struct tm *tm, int flags, char *buf, size_t bufsz);
+int strtime_iso(const time_t *t, int flags, char *buf, size_t bufsz);
+
+#define UL_SHORTTIME_THISYEAR_HHMM (1 << 1)
+
+int strtime_short(const time_t *t, struct timeval *now, int flags, char *buf, size_t bufsz);
+
+#ifndef HAVE_TIMEGM
+extern time_t timegm(struct tm *tm);
+#endif
+
+static inline usec_t timeval_to_usec(const struct timeval *t)
+{
+	return t->tv_sec * USEC_PER_SEC + t->tv_usec;
+}
+
+static inline usec_t timespec_to_usec(const struct timespec *t)
+{
+	return t->tv_sec * USEC_PER_SEC + t->tv_nsec / NSEC_PER_USEC;
+}
+
+static inline struct timeval usec_to_timeval(usec_t t)
+{
+	struct timeval r = {
+		.tv_sec = t / USEC_PER_SEC,
+		.tv_usec = t % USEC_PER_SEC,
+	};
+	return r;
+}
+
+#endif /* UTIL_LINUX_TIME_UTIL_H */
diff --git a/include/ttyutils.h b/include/ttyutils.h
new file mode 100644
index 0000000..e28a2b0
--- /dev/null
+++ b/include/ttyutils.h
@@ -0,0 +1,212 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ *
+ * Written by Karel Zak <kzak@redhat.com>
+ */
+#ifndef UTIL_LINUX_TTYUTILS_H
+#define UTIL_LINUX_TTYUTILS_H
+
+#include <stdlib.h>
+#include <termios.h>
+#include <limits.h>
+#ifdef HAVE_SYS_IOCTL_H
+#include <sys/ioctl.h>
+#endif
+#ifdef HAVE_SYS_TTYDEFAULTS_H
+#include <sys/ttydefaults.h>
+#endif
+
+#ifdef USE_TTY_GROUP
+# define TTY_MODE 0620
+#else
+# define TTY_MODE 0600
+#endif
+#define TTYGRPNAME      "tty"   /* name of group to own ttys */
+
+/* Some shorthands for control characters. */
+#define CTL(x)		((x) ^ 0100)	/* Assumes ASCII dialect */
+#define CR		CTL('M')	/* carriage return */
+#define NL		CTL('J')	/* line feed */
+#define BS		CTL('H')	/* back space */
+#define DEL		CTL('?')	/* delete */
+
+/* Defaults for line-editing etc. characters; you may want to change these. */
+#define DEF_ERASE	DEL		/* default erase character */
+#define DEF_INTR	CTL('C')	/* default interrupt character */
+#define DEF_QUIT	CTL('\\')	/* default quit char */
+#define DEF_KILL	CTL('U')	/* default kill char */
+#define DEF_EOF		CTL('D')	/* default EOF char */
+#define DEF_EOL		0
+#define DEF_SWITCH	0		/* default switch char */
+
+/* Fallback for termios->c_cc[] */
+#ifndef CREPRINT
+# define CREPRINT	('r' & 037)
+#endif
+#ifndef CDISCARD
+# define CDISCARD	('o' & 037)
+#endif
+
+/* Default termios->iflag */
+#ifndef TTYDEF_IFLAG
+# define TTYDEF_IFLAG	(BRKINT | ICRNL | IMAXBEL | IXON | IXANY)
+#endif
+
+/* Default termios->oflag */
+#ifndef TTYDEF_OFLAG
+# define TTYDEF_OFLAG	(OPOST | ONLCR /*| OXTABS*/)
+#endif
+
+/* Default termios->lflag */
+#ifndef TTYDEF_LFLAG
+# define TTYDEF_LFLAG	(ECHO | ICANON | ISIG | IEXTEN | ECHOE|ECHOKE|ECHOCTL)
+#endif
+
+/* Default termios->cflag */
+#ifndef TTYDEF_CFLAG
+# define TTYDEF_CFLAG	(CREAD | CS8 | HUPCL)
+#endif
+
+/* Storage for things detected while the login name was read. */
+struct chardata {
+	int erase;		/* erase character */
+	int kill;		/* kill character */
+	int eol;		/* end-of-line character */
+	int parity;		/* what parity did we see */
+	int capslock;		/* upper case without lower case */
+};
+
+#define INIT_CHARDATA(ptr) do {              \
+		(ptr)->erase    = DEF_ERASE; \
+		(ptr)->kill     = DEF_KILL;  \
+		(ptr)->eol      = CR;        \
+	        (ptr)->parity   = 0;         \
+	        (ptr)->capslock = 0;         \
+	} while (0)
+
+extern int get_terminal_dimension(int *cols, int *lines);
+extern int get_terminal_width(int default_width);
+extern int get_terminal_type(const char **type);
+extern int get_terminal_stdfd(void);
+extern int get_terminal_name(const char **path, const char **name,
+			     const char **number);
+
+#define UL_TTY_KEEPCFLAGS	(1 << 1)
+#define UL_TTY_UTF8		(1 << 2)
+
+static inline void reset_virtual_console(struct termios *tp, int flags)
+{
+	/* Use defaults of <sys/ttydefaults.h> for base settings */
+	tp->c_iflag |= TTYDEF_IFLAG;
+	tp->c_oflag |= TTYDEF_OFLAG;
+	tp->c_lflag |= TTYDEF_LFLAG;
+
+	if ((flags & UL_TTY_KEEPCFLAGS) == 0) {
+#ifdef CBAUD
+		tp->c_lflag &= ~CBAUD;
+#endif
+		tp->c_cflag |= (B38400 | TTYDEF_CFLAG);
+	}
+
+	/* Sane setting, allow eight bit characters, no carriage return delay
+	 * the same result as `stty sane cr0 pass8'
+	 */
+#ifndef IUCLC
+# define IUCLC 0
+#endif
+#ifndef NL0
+# define NL0 0
+#endif
+#ifndef CR0
+# define CR0 0
+#endif
+#ifndef BS0
+# define BS0 0
+#endif
+#ifndef VT0
+# define VT0 0
+#endif
+#ifndef FF0
+# define FF0 0
+#endif
+#ifndef OLCUC
+# define OLCUC 0
+#endif
+#ifndef OFILL
+# define OFILL 0
+#endif
+#ifndef NLDLY
+# define NLDLY 0
+#endif
+#ifndef CRDLY
+# define CRDLY 0
+#endif
+#ifndef BSDLY
+# define BSDLY 0
+#endif
+#ifndef VTDLY
+# define VTDLY 0
+#endif
+#ifndef FFDLY
+# define FFDLY 0
+#endif
+#ifndef TAB0
+# define TAB0 0
+#endif
+#ifndef TABDLY
+# define TABDLY 0
+#endif
+
+	tp->c_iflag |=  (BRKINT | ICRNL | IMAXBEL);
+	tp->c_iflag &= ~(IGNBRK | INLCR | IGNCR | IXOFF | IUCLC | IXANY | ISTRIP);
+	tp->c_oflag |=  (OPOST | ONLCR | NL0 | CR0 | TAB0 | BS0 | VT0 | FF0);
+	tp->c_oflag &= ~(OLCUC | OCRNL | ONOCR | ONLRET | OFILL | \
+			    NLDLY|CRDLY|TABDLY|BSDLY|VTDLY|FFDLY);
+	tp->c_lflag |=  (ISIG | ICANON | IEXTEN | ECHO|ECHOE|ECHOK|ECHOKE|ECHOCTL);
+	tp->c_lflag &= ~(ECHONL|ECHOPRT | NOFLSH | TOSTOP);
+
+	if ((flags & UL_TTY_KEEPCFLAGS) == 0) {
+		tp->c_cflag |=  (CREAD | CS8 | HUPCL);
+		tp->c_cflag &= ~(PARODD | PARENB);
+	}
+#ifdef OFDEL
+	tp->c_oflag &= ~OFDEL;
+#endif
+#ifdef XCASE
+	tp->c_lflag &= ~XCASE;
+#endif
+#ifdef IUTF8
+	if (flags & UL_TTY_UTF8)
+		tp->c_iflag |= IUTF8;	    /* Set UTF-8 input flag */
+	else
+		tp->c_iflag &= ~IUTF8;
+#endif
+	/* VTIME and VMIN can overlap with VEOF and VEOL since they are
+	 * only used for non-canonical mode. We just set the at the
+	 * beginning, so nothing bad should happen.
+	 */
+	tp->c_cc[VTIME]    = 0;
+	tp->c_cc[VMIN]     = 1;
+	tp->c_cc[VINTR]    = CINTR;
+	tp->c_cc[VQUIT]    = CQUIT;
+	tp->c_cc[VERASE]   = CERASE; /* ASCII DEL (0177) */
+	tp->c_cc[VKILL]    = CKILL;
+	tp->c_cc[VEOF]     = CEOF;
+#ifdef VSWTC
+	tp->c_cc[VSWTC]    = _POSIX_VDISABLE;
+#elif defined(VSWTCH)
+	tp->c_cc[VSWTCH]   = _POSIX_VDISABLE;
+#endif
+	tp->c_cc[VSTART]   = CSTART;
+	tp->c_cc[VSTOP]    = CSTOP;
+	tp->c_cc[VSUSP]    = CSUSP;
+	tp->c_cc[VEOL]     = _POSIX_VDISABLE;
+	tp->c_cc[VREPRINT] = CREPRINT;
+	tp->c_cc[VDISCARD] = CDISCARD;
+	tp->c_cc[VWERASE]  = CWERASE;
+	tp->c_cc[VLNEXT]   = CLNEXT;
+	tp->c_cc[VEOL2]    = _POSIX_VDISABLE;
+}
+
+#endif /* UTIL_LINUX_TTYUTILS_H */
diff --git a/include/widechar.h b/include/widechar.h
new file mode 100644
index 0000000..bbccc88
--- /dev/null
+++ b/include/widechar.h
@@ -0,0 +1,52 @@
+/*
+ * No copyright is claimed.  This code is in the public domain; do with
+ * it what you wish.
+ */
+
+/* Declarations for wide characters */
+/* This file must be included last because the redefinition of wchar_t may
+   cause conflicts when system include files were included after it. */
+
+#ifdef HAVE_WIDECHAR
+
+# include <wchar.h>
+# include <wctype.h>
+
+#else /* !HAVE_WIDECHAR */
+
+# include <ctype.h>
+  /* Fallback for types */
+# define wchar_t char
+# define wint_t int
+# ifndef WEOF
+#  define WEOF EOF
+# endif
+
+ /* Fallback for input operations */
+# define fgetwc fgetc
+# define getwc getc
+# define getwchar getchar
+# define fgetws fgets
+
+  /* Fallback for output operations */
+# define fputwc fputc
+# define putwc putc
+# define putwchar putchar
+# define fputws fputs
+
+  /* Fallback for character classification */
+# define iswgraph isgraph
+# define iswprint isprint
+# define iswspace isspace
+
+  /* Fallback for string functions */
+# define wcschr strchr
+# define wcsdup strdup
+# define wcslen strlen
+# define wcspbrk strpbrk
+
+# define wcwidth(c) (1)
+# define wmemset memset
+# define ungetwc ungetc
+
+#endif /* HAVE_WIDECHAR */
diff --git a/include/xalloc.h b/include/xalloc.h
new file mode 100644
index 0000000..1dee5fe
--- /dev/null
+++ b/include/xalloc.h
@@ -0,0 +1,133 @@
+/*
+ * SPDX-License-Identifier: LGPL-2.1-or-later
+ *
+ * Copyright (C) 2010 Davidlohr Bueso <dave@gnu.org>
+ * Copyright (C) 2010-2022 Karel Zak <kzak@redhat.com>
+ *
+ * This file may be redistributed under the terms of the
+ * GNU Lesser General Public License.
+ *
+ * General memory allocation wrappers for malloc, realloc, calloc and strdup
+ */
+
+#ifndef UTIL_LINUX_XALLOC_H
+#define UTIL_LINUX_XALLOC_H
+
+#include <stdlib.h>
+#include <string.h>
+
+#include "c.h"
+
+#ifndef XALLOC_EXIT_CODE
+# define XALLOC_EXIT_CODE EXIT_FAILURE
+#endif
+
+static inline
+__ul_alloc_size(1)
+__ul_returns_nonnull
+void *xmalloc(const size_t size)
+{
+	void *ret = malloc(size);
+
+	if (!ret && size)
+		err(XALLOC_EXIT_CODE, "cannot allocate %zu bytes", size);
+	return ret;
+}
+
+static inline
+__ul_alloc_size(2)
+__ul_returns_nonnull
+void *xrealloc(void *ptr, const size_t size)
+{
+	void *ret = realloc(ptr, size);
+
+	if (!ret && size)
+		err(XALLOC_EXIT_CODE, "cannot allocate %zu bytes", size);
+	return ret;
+}
+
+static inline
+__ul_calloc_size(1, 2)
+__ul_returns_nonnull
+void *xcalloc(const size_t nelems, const size_t size)
+{
+	void *ret = calloc(nelems, size);
+
+	if (!ret && size && nelems)
+		err(XALLOC_EXIT_CODE, "cannot allocate %zu bytes", size);
+	return ret;
+}
+
+static inline
+__attribute__((warn_unused_result))
+__ul_returns_nonnull
+char *xstrdup(const char *str)
+{
+	char *ret;
+
+	assert(str);
+	ret = strdup(str);
+	if (!ret)
+		err(XALLOC_EXIT_CODE, "cannot duplicate string");
+	return ret;
+}
+
+static inline
+__attribute__((warn_unused_result))
+__ul_returns_nonnull
+char *xstrndup(const char *str, size_t size)
+{
+	char *ret;
+
+	assert(str);
+	ret = strndup(str, size);
+	if (!ret)
+		err(XALLOC_EXIT_CODE, "cannot duplicate string");
+	return ret;
+}
+
+
+static inline
+__attribute__((__format__(printf, 2, 3)))
+int xasprintf(char **strp, const char *fmt, ...)
+{
+	int ret;
+	va_list args;
+
+	va_start(args, fmt);
+	ret = vasprintf(&(*strp), fmt, args);
+	va_end(args);
+	if (ret < 0)
+		err(XALLOC_EXIT_CODE, "cannot allocate string");
+	return ret;
+}
+
+static inline
+__attribute__((__format__(printf, 2, 0)))
+int xvasprintf(char **strp, const char *fmt, va_list ap)
+{
+	int ret = vasprintf(&(*strp), fmt, ap);
+
+	if (ret < 0)
+		err(XALLOC_EXIT_CODE, "cannot allocate string");
+	return ret;
+}
+
+
+static inline
+__attribute__((warn_unused_result))
+char *xgethostname(void)
+{
+	char *name;
+	size_t sz = get_hostname_max() + 1;
+
+	name = xmalloc(sizeof(char) * sz);
+	if (gethostname(name, sz) != 0) {
+		free(name);
+		return NULL;
+	}
+	name[sz - 1] = '\0';
+	return name;
+}
+
+#endif
diff --git a/include/xxhash.h b/include/xxhash.h
new file mode 100644
index 0000000..9451967
--- /dev/null
+++ b/include/xxhash.h
@@ -0,0 +1,5586 @@
+/*
+ * SPDX-License-Identifier: BSD-2-Clause
+ *
+ * xxHash - Extremely Fast Hash algorithm
+ * Header File
+ * Copyright (C) 2012-2020 Yann Collet
+ *
+ * BSD 2-Clause License (https://www.opensource.org/licenses/bsd-license.php)
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *    * Redistributions of source code must retain the above copyright
+ *      notice, this list of conditions and the following disclaimer.
+ *    * Redistributions in binary form must reproduce the above
+ *      copyright notice, this list of conditions and the following disclaimer
+ *      in the documentation and/or other materials provided with the
+ *      distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * You can contact the author at:
+ *   - xxHash homepage: https://www.xxhash.com
+ *   - xxHash source repository: https://github.com/Cyan4973/xxHash
+ */
+/*!
+ * @mainpage xxHash
+ *
+ * @file xxhash.h
+ * xxHash prototypes and implementation
+ */
+/* TODO: update */
+/* Notice extracted from xxHash homepage:
+
+xxHash is an extremely fast hash algorithm, running at RAM speed limits.
+It also successfully passes all tests from the SMHasher suite.
+
+Comparison (single thread, Windows Seven 32 bits, using SMHasher on a Core 2 Duo @3GHz)
+
+Name            Speed       Q.Score   Author
+xxHash          5.4 GB/s     10
+CrapWow         3.2 GB/s      2       Andrew
+MurmurHash 3a   2.7 GB/s     10       Austin Appleby
+SpookyHash      2.0 GB/s     10       Bob Jenkins
+SBox            1.4 GB/s      9       Bret Mulvey
+Lookup3         1.2 GB/s      9       Bob Jenkins
+SuperFastHash   1.2 GB/s      1       Paul Hsieh
+CityHash64      1.05 GB/s    10       Pike & Alakuijala
+FNV             0.55 GB/s     5       Fowler, Noll, Vo
+CRC32           0.43 GB/s     9
+MD5-32          0.33 GB/s    10       Ronald L. Rivest
+SHA1-32         0.28 GB/s    10
+
+Q.Score is a measure of quality of the hash function.
+It depends on successfully passing SMHasher test set.
+10 is a perfect score.
+
+Note: SMHasher's CRC32 implementation is not the fastest one.
+Other speed-oriented implementations can be faster,
+especially in combination with PCLMUL instruction:
+https://fastcompression.blogspot.com/2019/03/presenting-xxh3.html?showComment=1552696407071#c3490092340461170735
+
+A 64-bit version, named XXH64, is available since r35.
+It offers much better speed, but for 64-bit applications only.
+Name     Speed on 64 bits    Speed on 32 bits
+XXH64       13.8 GB/s            1.9 GB/s
+XXH32        6.8 GB/s            6.0 GB/s
+*/
+
+/* util-linux customizations */
+#define XXH_NO_XXH3
+#define XXH_NAMESPACE ul_
+
+#if defined (__cplusplus)
+extern "C" {
+#endif
+
+/* ****************************
+ *  INLINE mode
+ ******************************/
+/*!
+ * XXH_INLINE_ALL (and XXH_PRIVATE_API)
+ * Use these build macros to inline xxhash into the target unit.
+ * Inlining improves performance on small inputs, especially when the length is
+ * expressed as a compile-time constant:
+ *
+ *      https://fastcompression.blogspot.com/2018/03/xxhash-for-small-keys-impressive-power.html
+ *
+ * It also keeps xxHash symbols private to the unit, so they are not exported.
+ *
+ * Usage:
+ *     #define XXH_INLINE_ALL
+ *     #include "xxhash.h"
+ *
+ * Do not compile and link xxhash.o as a separate object, as it is not useful.
+ */
+#if (defined(XXH_INLINE_ALL) || defined(XXH_PRIVATE_API)) \
+    && !defined(XXH_INLINE_ALL_31684351384)
+   /* this section should be traversed only once */
+#  define XXH_INLINE_ALL_31684351384
+   /* give access to the advanced API, required to compile implementations */
+#  undef XXH_STATIC_LINKING_ONLY   /* avoid macro redef */
+#  define XXH_STATIC_LINKING_ONLY
+   /* make all functions private */
+#  undef XXH_PUBLIC_API
+#  if defined(__GNUC__)
+#    define XXH_PUBLIC_API static __inline __attribute__((unused))
+#  elif defined (__cplusplus) || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */)
+#    define XXH_PUBLIC_API static inline
+#  elif defined(_MSC_VER)
+#    define XXH_PUBLIC_API static __inline
+#  else
+     /* note: this version may generate warnings for unused static functions */
+#    define XXH_PUBLIC_API static
+#  endif
+
+   /*
+    * This part deals with the special case where a unit wants to inline xxHash,
+    * but "xxhash.h" has previously been included without XXH_INLINE_ALL,
+    * such as part of some previously included *.h header file.
+    * Without further action, the new include would just be ignored,
+    * and functions would effectively _not_ be inlined (silent failure).
+    * The following macros solve this situation by prefixing all inlined names,
+    * avoiding naming collision with previous inclusions.
+    */
+   /* Before that, we unconditionally #undef all symbols,
+    * in case they were already defined with XXH_NAMESPACE.
+    * They will then be redefined for XXH_INLINE_ALL
+    */
+#  undef XXH_versionNumber
+    /* XXH32 */
+#  undef XXH32
+#  undef XXH32_createState
+#  undef XXH32_freeState
+#  undef XXH32_reset
+#  undef XXH32_update
+#  undef XXH32_digest
+#  undef XXH32_copyState
+#  undef XXH32_canonicalFromHash
+#  undef XXH32_hashFromCanonical
+    /* XXH64 */
+#  undef XXH64
+#  undef XXH64_createState
+#  undef XXH64_freeState
+#  undef XXH64_reset
+#  undef XXH64_update
+#  undef XXH64_digest
+#  undef XXH64_copyState
+#  undef XXH64_canonicalFromHash
+#  undef XXH64_hashFromCanonical
+    /* XXH3_64bits */
+#  undef XXH3_64bits
+#  undef XXH3_64bits_withSecret
+#  undef XXH3_64bits_withSeed
+#  undef XXH3_64bits_withSecretandSeed
+#  undef XXH3_createState
+#  undef XXH3_freeState
+#  undef XXH3_copyState
+#  undef XXH3_64bits_reset
+#  undef XXH3_64bits_reset_withSeed
+#  undef XXH3_64bits_reset_withSecret
+#  undef XXH3_64bits_update
+#  undef XXH3_64bits_digest
+#  undef XXH3_generateSecret
+    /* XXH3_128bits */
+#  undef XXH128
+#  undef XXH3_128bits
+#  undef XXH3_128bits_withSeed
+#  undef XXH3_128bits_withSecret
+#  undef XXH3_128bits_reset
+#  undef XXH3_128bits_reset_withSeed
+#  undef XXH3_128bits_reset_withSecret
+#  undef XXH3_128bits_reset_withSecretandSeed
+#  undef XXH3_128bits_update
+#  undef XXH3_128bits_digest
+#  undef XXH128_isEqual
+#  undef XXH128_cmp
+#  undef XXH128_canonicalFromHash
+#  undef XXH128_hashFromCanonical
+    /* Finally, free the namespace itself */
+#  undef XXH_NAMESPACE
+
+    /* employ the namespace for XXH_INLINE_ALL */
+#  define XXH_NAMESPACE XXH_INLINE_
+   /*
+    * Some identifiers (enums, type names) are not symbols,
+    * but they must nonetheless be renamed to avoid redeclaration.
+    * Alternative solution: do not redeclare them.
+    * However, this requires some #ifdefs, and has a more dispersed impact.
+    * Meanwhile, renaming can be achieved in a single place.
+    */
+#  define XXH_IPREF(Id)   XXH_NAMESPACE ## Id
+#  define XXH_OK XXH_IPREF(XXH_OK)
+#  define XXH_ERROR XXH_IPREF(XXH_ERROR)
+#  define XXH_errorcode XXH_IPREF(XXH_errorcode)
+#  define XXH32_canonical_t  XXH_IPREF(XXH32_canonical_t)
+#  define XXH64_canonical_t  XXH_IPREF(XXH64_canonical_t)
+#  define XXH128_canonical_t XXH_IPREF(XXH128_canonical_t)
+#  define XXH32_state_s XXH_IPREF(XXH32_state_s)
+#  define XXH32_state_t XXH_IPREF(XXH32_state_t)
+#  define XXH64_state_s XXH_IPREF(XXH64_state_s)
+#  define XXH64_state_t XXH_IPREF(XXH64_state_t)
+#  define XXH3_state_s  XXH_IPREF(XXH3_state_s)
+#  define XXH3_state_t  XXH_IPREF(XXH3_state_t)
+#  define XXH128_hash_t XXH_IPREF(XXH128_hash_t)
+   /* Ensure the header is parsed again, even if it was previously included */
+#  undef XXHASH_H_5627135585666179
+#  undef XXHASH_H_STATIC_13879238742
+#endif /* XXH_INLINE_ALL || XXH_PRIVATE_API */
+
+
+
+/* ****************************************************************
+ *  Stable API
+ *****************************************************************/
+#ifndef XXHASH_H_5627135585666179
+#define XXHASH_H_5627135585666179 1
+
+
+/*!
+ * @defgroup public Public API
+ * Contains details on the public xxHash functions.
+ * @{
+ */
+/* specific declaration modes for Windows */
+#if !defined(XXH_INLINE_ALL) && !defined(XXH_PRIVATE_API)
+#  if defined(WIN32) && defined(_MSC_VER) && (defined(XXH_IMPORT) || defined(XXH_EXPORT))
+#    ifdef XXH_EXPORT
+#      define XXH_PUBLIC_API __declspec(dllexport)
+#    elif XXH_IMPORT
+#      define XXH_PUBLIC_API __declspec(dllimport)
+#    endif
+#  else
+#    define XXH_PUBLIC_API   /* do nothing */
+#  endif
+#endif
+
+#ifdef XXH_DOXYGEN
+/*!
+ * @brief Emulate a namespace by transparently prefixing all symbols.
+ *
+ * If you want to include _and expose_ xxHash functions from within your own
+ * library, but also want to avoid symbol collisions with other libraries which
+ * may also include xxHash, you can use XXH_NAMESPACE to automatically prefix
+ * any public symbol from xxhash library with the value of XXH_NAMESPACE
+ * (therefore, avoid empty or numeric values).
+ *
+ * Note that no change is required within the calling program as long as it
+ * includes `xxhash.h`: Regular symbol names will be automatically translated
+ * by this header.
+ */
+#  define XXH_NAMESPACE /* YOUR NAME HERE */
+#  undef XXH_NAMESPACE
+#endif
+
+#ifdef XXH_NAMESPACE
+#  define XXH_CAT(A,B) A##B
+#  define XXH_NAME2(A,B) XXH_CAT(A,B)
+#  define XXH_versionNumber XXH_NAME2(XXH_NAMESPACE, XXH_versionNumber)
+/* XXH32 */
+#  define XXH32 XXH_NAME2(XXH_NAMESPACE, XXH32)
+#  define XXH32_createState XXH_NAME2(XXH_NAMESPACE, XXH32_createState)
+#  define XXH32_freeState XXH_NAME2(XXH_NAMESPACE, XXH32_freeState)
+#  define XXH32_reset XXH_NAME2(XXH_NAMESPACE, XXH32_reset)
+#  define XXH32_update XXH_NAME2(XXH_NAMESPACE, XXH32_update)
+#  define XXH32_digest XXH_NAME2(XXH_NAMESPACE, XXH32_digest)
+#  define XXH32_copyState XXH_NAME2(XXH_NAMESPACE, XXH32_copyState)
+#  define XXH32_canonicalFromHash XXH_NAME2(XXH_NAMESPACE, XXH32_canonicalFromHash)
+#  define XXH32_hashFromCanonical XXH_NAME2(XXH_NAMESPACE, XXH32_hashFromCanonical)
+/* XXH64 */
+#  define XXH64 XXH_NAME2(XXH_NAMESPACE, XXH64)
+#  define XXH64_createState XXH_NAME2(XXH_NAMESPACE, XXH64_createState)
+#  define XXH64_freeState XXH_NAME2(XXH_NAMESPACE, XXH64_freeState)
+#  define XXH64_reset XXH_NAME2(XXH_NAMESPACE, XXH64_reset)
+#  define XXH64_update XXH_NAME2(XXH_NAMESPACE, XXH64_update)
+#  define XXH64_digest XXH_NAME2(XXH_NAMESPACE, XXH64_digest)
+#  define XXH64_copyState XXH_NAME2(XXH_NAMESPACE, XXH64_copyState)
+#  define XXH64_canonicalFromHash XXH_NAME2(XXH_NAMESPACE, XXH64_canonicalFromHash)
+#  define XXH64_hashFromCanonical XXH_NAME2(XXH_NAMESPACE, XXH64_hashFromCanonical)
+/* XXH3_64bits */
+#  define XXH3_64bits XXH_NAME2(XXH_NAMESPACE, XXH3_64bits)
+#  define XXH3_64bits_withSecret XXH_NAME2(XXH_NAMESPACE, XXH3_64bits_withSecret)
+#  define XXH3_64bits_withSeed XXH_NAME2(XXH_NAMESPACE, XXH3_64bits_withSeed)
+#  define XXH3_64bits_withSecretandSeed XXH_NAME2(XXH_NAMESPACE, XXH3_64bits_withSecretandSeed)
+#  define XXH3_createState XXH_NAME2(XXH_NAMESPACE, XXH3_createState)
+#  define XXH3_freeState XXH_NAME2(XXH_NAMESPACE, XXH3_freeState)
+#  define XXH3_copyState XXH_NAME2(XXH_NAMESPACE, XXH3_copyState)
+#  define XXH3_64bits_reset XXH_NAME2(XXH_NAMESPACE, XXH3_64bits_reset)
+#  define XXH3_64bits_reset_withSeed XXH_NAME2(XXH_NAMESPACE, XXH3_64bits_reset_withSeed)
+#  define XXH3_64bits_reset_withSecret XXH_NAME2(XXH_NAMESPACE, XXH3_64bits_reset_withSecret)
+#  define XXH3_64bits_reset_withSecretandSeed XXH_NAME2(XXH_NAMESPACE, XXH3_64bits_reset_withSecretandSeed)
+#  define XXH3_64bits_update XXH_NAME2(XXH_NAMESPACE, XXH3_64bits_update)
+#  define XXH3_64bits_digest XXH_NAME2(XXH_NAMESPACE, XXH3_64bits_digest)
+#  define XXH3_generateSecret XXH_NAME2(XXH_NAMESPACE, XXH3_generateSecret)
+#  define XXH3_generateSecret_fromSeed XXH_NAME2(XXH_NAMESPACE, XXH3_generateSecret_fromSeed)
+/* XXH3_128bits */
+#  define XXH128 XXH_NAME2(XXH_NAMESPACE, XXH128)
+#  define XXH3_128bits XXH_NAME2(XXH_NAMESPACE, XXH3_128bits)
+#  define XXH3_128bits_withSeed XXH_NAME2(XXH_NAMESPACE, XXH3_128bits_withSeed)
+#  define XXH3_128bits_withSecret XXH_NAME2(XXH_NAMESPACE, XXH3_128bits_withSecret)
+#  define XXH3_128bits_withSecretandSeed XXH_NAME2(XXH_NAMESPACE, XXH3_128bits_withSecretandSeed)
+#  define XXH3_128bits_reset XXH_NAME2(XXH_NAMESPACE, XXH3_128bits_reset)
+#  define XXH3_128bits_reset_withSeed XXH_NAME2(XXH_NAMESPACE, XXH3_128bits_reset_withSeed)
+#  define XXH3_128bits_reset_withSecret XXH_NAME2(XXH_NAMESPACE, XXH3_128bits_reset_withSecret)
+#  define XXH3_128bits_reset_withSecretandSeed XXH_NAME2(XXH_NAMESPACE, XXH3_128bits_reset_withSecretandSeed)
+#  define XXH3_128bits_update XXH_NAME2(XXH_NAMESPACE, XXH3_128bits_update)
+#  define XXH3_128bits_digest XXH_NAME2(XXH_NAMESPACE, XXH3_128bits_digest)
+#  define XXH128_isEqual XXH_NAME2(XXH_NAMESPACE, XXH128_isEqual)
+#  define XXH128_cmp     XXH_NAME2(XXH_NAMESPACE, XXH128_cmp)
+#  define XXH128_canonicalFromHash XXH_NAME2(XXH_NAMESPACE, XXH128_canonicalFromHash)
+#  define XXH128_hashFromCanonical XXH_NAME2(XXH_NAMESPACE, XXH128_hashFromCanonical)
+#endif
+
+
+/* *************************************
+*  Version
+***************************************/
+#define XXH_VERSION_MAJOR    0
+#define XXH_VERSION_MINOR    8
+#define XXH_VERSION_RELEASE  1
+#define XXH_VERSION_NUMBER  (XXH_VERSION_MAJOR *100*100 + XXH_VERSION_MINOR *100 + XXH_VERSION_RELEASE)
+
+/*!
+ * @brief Obtains the xxHash version.
+ *
+ * This is mostly useful when xxHash is compiled as a shared library,
+ * since the returned value comes from the library, as opposed to header file.
+ *
+ * @return `XXH_VERSION_NUMBER` of the invoked library.
+ */
+XXH_PUBLIC_API unsigned XXH_versionNumber (void);
+
+
+/* ****************************
+*  Common basic types
+******************************/
+#include <stddef.h>   /* size_t */
+typedef enum { XXH_OK=0, XXH_ERROR } XXH_errorcode;
+
+
+/*-**********************************************************************
+*  32-bit hash
+************************************************************************/
+#if defined(XXH_DOXYGEN) /* Don't show <stdint.h> include */
+/*!
+ * @brief An unsigned 32-bit integer.
+ *
+ * Not necessarily defined to `uint32_t` but functionally equivalent.
+ */
+typedef uint32_t XXH32_hash_t;
+
+#elif !defined (__VMS) \
+  && (defined (__cplusplus) \
+  || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) )
+#   include <stdint.h>
+    typedef uint32_t XXH32_hash_t;
+
+#else
+#   include <limits.h>
+#   if UINT_MAX == 0xFFFFFFFFUL
+      typedef unsigned int XXH32_hash_t;
+#   else
+#     if ULONG_MAX == 0xFFFFFFFFUL
+        typedef unsigned long XXH32_hash_t;
+#     else
+#       error "unsupported platform: need a 32-bit type"
+#     endif
+#   endif
+#endif
+
+/*!
+ * @}
+ *
+ * @defgroup xxh32_family XXH32 family
+ * @ingroup public
+ * Contains functions used in the classic 32-bit xxHash algorithm.
+ *
+ * @note
+ *   XXH32 is useful for older platforms, with no or poor 64-bit performance.
+ *   Note that @ref xxh3_family provides competitive speed
+ *   for both 32-bit and 64-bit systems, and offers true 64/128 bit hash results.
+ *
+ * @see @ref xxh64_family, @ref xxh3_family : Other xxHash families
+ * @see @ref xxh32_impl for implementation details
+ * @{
+ */
+
+/*!
+ * @brief Calculates the 32-bit hash of @p input using xxHash32.
+ *
+ * Speed on Core 2 Duo @ 3 GHz (single thread, SMHasher benchmark): 5.4 GB/s
+ *
+ * @param input The block of data to be hashed, at least @p length bytes in size.
+ * @param length The length of @p input, in bytes.
+ * @param seed The 32-bit seed to alter the hash's output predictably.
+ *
+ * @pre
+ *   The memory between @p input and @p input + @p length must be valid,
+ *   readable, contiguous memory. However, if @p length is `0`, @p input may be
+ *   `NULL`. In C++, this also must be *TriviallyCopyable*.
+ *
+ * @return The calculated 32-bit hash value.
+ *
+ * @see
+ *    XXH64(), XXH3_64bits_withSeed(), XXH3_128bits_withSeed(), XXH128():
+ *    Direct equivalents for the other variants of xxHash.
+ * @see
+ *    XXH32_createState(), XXH32_update(), XXH32_digest(): Streaming version.
+ */
+XXH_PUBLIC_API XXH32_hash_t XXH32 (const void* input, size_t length, XXH32_hash_t seed);
+
+/*!
+ * Streaming functions generate the xxHash value from an incremental input.
+ * This method is slower than single-call functions, due to state management.
+ * For small inputs, prefer `XXH32()` and `XXH64()`, which are better optimized.
+ *
+ * An XXH state must first be allocated using `XXH*_createState()`.
+ *
+ * Start a new hash by initializing the state with a seed using `XXH*_reset()`.
+ *
+ * Then, feed the hash state by calling `XXH*_update()` as many times as necessary.
+ *
+ * The function returns an error code, with 0 meaning OK, and any other value
+ * meaning there is an error.
+ *
+ * Finally, a hash value can be produced anytime, by using `XXH*_digest()`.
+ * This function returns the nn-bits hash as an int or long long.
+ *
+ * It's still possible to continue inserting input into the hash state after a
+ * digest, and generate new hash values later on by invoking `XXH*_digest()`.
+ *
+ * When done, release the state using `XXH*_freeState()`.
+ *
+ * Example code for incrementally hashing a file:
+ * @code{.c}
+ *    #include <stdio.h>
+ *    #include <xxhash.h>
+ *    #define BUFFER_SIZE 256
+ *
+ *    // Note: XXH64 and XXH3 use the same interface.
+ *    XXH32_hash_t
+ *    hashFile(FILE* stream)
+ *    {
+ *        XXH32_state_t* state;
+ *        unsigned char buf[BUFFER_SIZE];
+ *        size_t amt;
+ *        XXH32_hash_t hash;
+ *
+ *        state = XXH32_createState();       // Create a state
+ *        assert(state != NULL);             // Error check here
+ *        XXH32_reset(state, 0xbaad5eed);    // Reset state with our seed
+ *        while ((amt = fread(buf, 1, sizeof(buf), stream)) != 0) {
+ *            XXH32_update(state, buf, amt); // Hash the file in chunks
+ *        }
+ *        hash = XXH32_digest(state);        // Finalize the hash
+ *        XXH32_freeState(state);            // Clean up
+ *        return hash;
+ *    }
+ * @endcode
+ */
+
+/*!
+ * @typedef struct XXH32_state_s XXH32_state_t
+ * @brief The opaque state struct for the XXH32 streaming API.
+ *
+ * @see XXH32_state_s for details.
+ */
+typedef struct XXH32_state_s XXH32_state_t;
+
+/*!
+ * @brief Allocates an @ref XXH32_state_t.
+ *
+ * Must be freed with XXH32_freeState().
+ * @return An allocated XXH32_state_t on success, `NULL` on failure.
+ */
+XXH_PUBLIC_API XXH32_state_t* XXH32_createState(void);
+/*!
+ * @brief Frees an @ref XXH32_state_t.
+ *
+ * Must be allocated with XXH32_createState().
+ * @param statePtr A pointer to an @ref XXH32_state_t allocated with @ref XXH32_createState().
+ * @return XXH_OK.
+ */
+XXH_PUBLIC_API XXH_errorcode  XXH32_freeState(XXH32_state_t* statePtr);
+/*!
+ * @brief Copies one @ref XXH32_state_t to another.
+ *
+ * @param dst_state The state to copy to.
+ * @param src_state The state to copy from.
+ * @pre
+ *   @p dst_state and @p src_state must not be `NULL` and must not overlap.
+ */
+XXH_PUBLIC_API void XXH32_copyState(XXH32_state_t* dst_state, const XXH32_state_t* src_state);
+
+/*!
+ * @brief Resets an @ref XXH32_state_t to begin a new hash.
+ *
+ * This function resets and seeds a state. Call it before @ref XXH32_update().
+ *
+ * @param statePtr The state struct to reset.
+ * @param seed The 32-bit seed to alter the hash result predictably.
+ *
+ * @pre
+ *   @p statePtr must not be `NULL`.
+ *
+ * @return @ref XXH_OK on success, @ref XXH_ERROR on failure.
+ */
+XXH_PUBLIC_API XXH_errorcode XXH32_reset  (XXH32_state_t* statePtr, XXH32_hash_t seed);
+
+/*!
+ * @brief Consumes a block of @p input to an @ref XXH32_state_t.
+ *
+ * Call this to incrementally consume blocks of data.
+ *
+ * @param statePtr The state struct to update.
+ * @param input The block of data to be hashed, at least @p length bytes in size.
+ * @param length The length of @p input, in bytes.
+ *
+ * @pre
+ *   @p statePtr must not be `NULL`.
+ * @pre
+ *   The memory between @p input and @p input + @p length must be valid,
+ *   readable, contiguous memory. However, if @p length is `0`, @p input may be
+ *   `NULL`. In C++, this also must be *TriviallyCopyable*.
+ *
+ * @return @ref XXH_OK on success, @ref XXH_ERROR on failure.
+ */
+XXH_PUBLIC_API XXH_errorcode XXH32_update (XXH32_state_t* statePtr, const void* input, size_t length);
+
+/*!
+ * @brief Returns the calculated hash value from an @ref XXH32_state_t.
+ *
+ * @note
+ *   Calling XXH32_digest() will not affect @p statePtr, so you can update,
+ *   digest, and update again.
+ *
+ * @param statePtr The state struct to calculate the hash from.
+ *
+ * @pre
+ *  @p statePtr must not be `NULL`.
+ *
+ * @return The calculated xxHash32 value from that state.
+ */
+XXH_PUBLIC_API XXH32_hash_t  XXH32_digest (const XXH32_state_t* statePtr);
+
+/*******   Canonical representation   *******/
+
+/*
+ * The default return values from XXH functions are unsigned 32 and 64 bit
+ * integers.
+ * This the simplest and fastest format for further post-processing.
+ *
+ * However, this leaves open the question of what is the order on the byte level,
+ * since little and big endian conventions will store the same number differently.
+ *
+ * The canonical representation settles this issue by mandating big-endian
+ * convention, the same convention as human-readable numbers (large digits first).
+ *
+ * When writing hash values to storage, sending them over a network, or printing
+ * them, it's highly recommended to use the canonical representation to ensure
+ * portability across a wider range of systems, present and future.
+ *
+ * The following functions allow transformation of hash values to and from
+ * canonical format.
+ */
+
+/*!
+ * @brief Canonical (big endian) representation of @ref XXH32_hash_t.
+ */
+typedef struct {
+    unsigned char digest[4]; /*!< Hash bytes, big endian */
+} XXH32_canonical_t;
+
+/*!
+ * @brief Converts an @ref XXH32_hash_t to a big endian @ref XXH32_canonical_t.
+ *
+ * @param dst The @ref XXH32_canonical_t pointer to be stored to.
+ * @param hash The @ref XXH32_hash_t to be converted.
+ *
+ * @pre
+ *   @p dst must not be `NULL`.
+ */
+XXH_PUBLIC_API void XXH32_canonicalFromHash(XXH32_canonical_t* dst, XXH32_hash_t hash);
+
+/*!
+ * @brief Converts an @ref XXH32_canonical_t to a native @ref XXH32_hash_t.
+ *
+ * @param src The @ref XXH32_canonical_t to convert.
+ *
+ * @pre
+ *   @p src must not be `NULL`.
+ *
+ * @return The converted hash.
+ */
+XXH_PUBLIC_API XXH32_hash_t XXH32_hashFromCanonical(const XXH32_canonical_t* src);
+
+
+#ifdef __has_attribute
+# define XXH_HAS_ATTRIBUTE(x) __has_attribute(x)
+#else
+# define XXH_HAS_ATTRIBUTE(x) 0
+#endif
+
+/* C-language Attributes are added in C23. */
+#if defined(__STDC_VERSION__) && (__STDC_VERSION__ > 201710L) && defined(__has_c_attribute)
+# define XXH_HAS_C_ATTRIBUTE(x) __has_c_attribute(x)
+#else
+# define XXH_HAS_C_ATTRIBUTE(x) 0
+#endif
+
+#if defined(__cplusplus) && defined(__has_cpp_attribute)
+# define XXH_HAS_CPP_ATTRIBUTE(x) __has_cpp_attribute(x)
+#else
+# define XXH_HAS_CPP_ATTRIBUTE(x) 0
+#endif
+
+/*
+Define XXH_FALLTHROUGH macro for annotating switch case with the 'fallthrough' attribute
+introduced in CPP17 and C23.
+CPP17 : https://en.cppreference.com/w/cpp/language/attributes/fallthrough
+C23   : https://en.cppreference.com/w/c/language/attributes/fallthrough
+*/
+#if XXH_HAS_C_ATTRIBUTE(x)
+# define XXH_FALLTHROUGH [[fallthrough]]
+#elif XXH_HAS_CPP_ATTRIBUTE(x)
+# define XXH_FALLTHROUGH [[fallthrough]]
+#elif XXH_HAS_ATTRIBUTE(__fallthrough__)
+# define XXH_FALLTHROUGH __attribute__ ((fallthrough))
+#else
+# define XXH_FALLTHROUGH
+#endif
+
+/*!
+ * @}
+ * @ingroup public
+ * @{
+ */
+
+#ifndef XXH_NO_LONG_LONG
+/*-**********************************************************************
+*  64-bit hash
+************************************************************************/
+#if defined(XXH_DOXYGEN) /* don't include <stdint.h> */
+/*!
+ * @brief An unsigned 64-bit integer.
+ *
+ * Not necessarily defined to `uint64_t` but functionally equivalent.
+ */
+typedef uint64_t XXH64_hash_t;
+#elif !defined (__VMS) \
+  && (defined (__cplusplus) \
+  || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) )
+#  include <stdint.h>
+   typedef uint64_t XXH64_hash_t;
+#else
+#  include <limits.h>
+#  if defined(__LP64__) && ULONG_MAX == 0xFFFFFFFFFFFFFFFFULL
+     /* LP64 ABI says uint64_t is unsigned long */
+     typedef unsigned long XXH64_hash_t;
+#  else
+     /* the following type must have a width of 64-bit */
+     typedef unsigned long long XXH64_hash_t;
+#  endif
+#endif
+
+/*!
+ * @}
+ *
+ * @defgroup xxh64_family XXH64 family
+ * @ingroup public
+ * @{
+ * Contains functions used in the classic 64-bit xxHash algorithm.
+ *
+ * @note
+ *   XXH3 provides competitive speed for both 32-bit and 64-bit systems,
+ *   and offers true 64/128 bit hash results.
+ *   It provides better speed for systems with vector processing capabilities.
+ */
+
+
+/*!
+ * @brief Calculates the 64-bit hash of @p input using xxHash64.
+ *
+ * This function usually runs faster on 64-bit systems, but slower on 32-bit
+ * systems (see benchmark).
+ *
+ * @param input The block of data to be hashed, at least @p length bytes in size.
+ * @param length The length of @p input, in bytes.
+ * @param seed The 64-bit seed to alter the hash's output predictably.
+ *
+ * @pre
+ *   The memory between @p input and @p input + @p length must be valid,
+ *   readable, contiguous memory. However, if @p length is `0`, @p input may be
+ *   `NULL`. In C++, this also must be *TriviallyCopyable*.
+ *
+ * @return The calculated 64-bit hash.
+ *
+ * @see
+ *    XXH32(), XXH3_64bits_withSeed(), XXH3_128bits_withSeed(), XXH128():
+ *    Direct equivalents for the other variants of xxHash.
+ * @see
+ *    XXH64_createState(), XXH64_update(), XXH64_digest(): Streaming version.
+ */
+XXH_PUBLIC_API XXH64_hash_t XXH64(const void* input, size_t length, XXH64_hash_t seed);
+
+/*******   Streaming   *******/
+/*!
+ * @brief The opaque state struct for the XXH64 streaming API.
+ *
+ * @see XXH64_state_s for details.
+ */
+typedef struct XXH64_state_s XXH64_state_t;   /* incomplete type */
+XXH_PUBLIC_API XXH64_state_t* XXH64_createState(void);
+XXH_PUBLIC_API XXH_errorcode  XXH64_freeState(XXH64_state_t* statePtr);
+XXH_PUBLIC_API void XXH64_copyState(XXH64_state_t* dst_state, const XXH64_state_t* src_state);
+
+XXH_PUBLIC_API XXH_errorcode XXH64_reset  (XXH64_state_t* statePtr, XXH64_hash_t seed);
+XXH_PUBLIC_API XXH_errorcode XXH64_update (XXH64_state_t* statePtr, const void* input, size_t length);
+XXH_PUBLIC_API XXH64_hash_t  XXH64_digest (const XXH64_state_t* statePtr);
+
+/*******   Canonical representation   *******/
+typedef struct { unsigned char digest[sizeof(XXH64_hash_t)]; } XXH64_canonical_t;
+XXH_PUBLIC_API void XXH64_canonicalFromHash(XXH64_canonical_t* dst, XXH64_hash_t hash);
+XXH_PUBLIC_API XXH64_hash_t XXH64_hashFromCanonical(const XXH64_canonical_t* src);
+
+/*!
+ * @}
+ * ************************************************************************
+ * @defgroup xxh3_family XXH3 family
+ * @ingroup public
+ * @{
+ *
+ * XXH3 is a more recent hash algorithm featuring:
+ *  - Improved speed for both small and large inputs
+ *  - True 64-bit and 128-bit outputs
+ *  - SIMD acceleration
+ *  - Improved 32-bit viability
+ *
+ * Speed analysis methodology is explained here:
+ *
+ *    https://fastcompression.blogspot.com/2019/03/presenting-xxh3.html
+ *
+ * Compared to XXH64, expect XXH3 to run approximately
+ * ~2x faster on large inputs and >3x faster on small ones,
+ * exact differences vary depending on platform.
+ *
+ * XXH3's speed benefits greatly from SIMD and 64-bit arithmetic,
+ * but does not require it.
+ * Any 32-bit and 64-bit targets that can run XXH32 smoothly
+ * can run XXH3 at competitive speeds, even without vector support.
+ * Further details are explained in the implementation.
+ *
+ * Optimized implementations are provided for AVX512, AVX2, SSE2, NEON, POWER8,
+ * ZVector and scalar targets. This can be controlled via the XXH_VECTOR macro.
+ *
+ * XXH3 implementation is portable:
+ * it has a generic C90 formulation that can be compiled on any platform,
+ * all implementations generage exactly the same hash value on all platforms.
+ * Starting from v0.8.0, it's also labelled "stable", meaning that
+ * any future version will also generate the same hash value.
+ *
+ * XXH3 offers 2 variants, _64bits and _128bits.
+ *
+ * When only 64 bits are needed, prefer invoking the _64bits variant, as it
+ * reduces the amount of mixing, resulting in faster speed on small inputs.
+ * It's also generally simpler to manipulate a scalar return type than a struct.
+ *
+ * The API supports one-shot hashing, streaming mode, and custom secrets.
+ */
+
+/*-**********************************************************************
+*  XXH3 64-bit variant
+************************************************************************/
+
+/* XXH3_64bits():
+ * default 64-bit variant, using default secret and default seed of 0.
+ * It's the fastest variant. */
+XXH_PUBLIC_API XXH64_hash_t XXH3_64bits(const void* data, size_t len);
+
+/*
+ * XXH3_64bits_withSeed():
+ * This variant generates a custom secret on the fly
+ * based on default secret altered using the `seed` value.
+ * While this operation is decently fast, note that it's not completely free.
+ * Note: seed==0 produces the same results as XXH3_64bits().
+ */
+XXH_PUBLIC_API XXH64_hash_t XXH3_64bits_withSeed(const void* data, size_t len, XXH64_hash_t seed);
+
+/*!
+ * The bare minimum size for a custom secret.
+ *
+ * @see
+ *  XXH3_64bits_withSecret(), XXH3_64bits_reset_withSecret(),
+ *  XXH3_128bits_withSecret(), XXH3_128bits_reset_withSecret().
+ */
+#define XXH3_SECRET_SIZE_MIN 136
+
+/*
+ * XXH3_64bits_withSecret():
+ * It's possible to provide any blob of bytes as a "secret" to generate the hash.
+ * This makes it more difficult for an external actor to prepare an intentional collision.
+ * The main condition is that secretSize *must* be large enough (>= XXH3_SECRET_SIZE_MIN).
+ * However, the quality of the secret impacts the dispersion of the hash algorithm.
+ * Therefore, the secret _must_ look like a bunch of random bytes.
+ * Avoid "trivial" or structured data such as repeated sequences or a text document.
+ * Whenever in doubt about the "randomness" of the blob of bytes,
+ * consider employing "XXH3_generateSecret()" instead (see below).
+ * It will generate a proper high entropy secret derived from the blob of bytes.
+ * Another advantage of using XXH3_generateSecret() is that
+ * it guarantees that all bits within the initial blob of bytes
+ * will impact every bit of the output.
+ * This is not necessarily the case when using the blob of bytes directly
+ * because, when hashing _small_ inputs, only a portion of the secret is employed.
+ */
+XXH_PUBLIC_API XXH64_hash_t XXH3_64bits_withSecret(const void* data, size_t len, const void* secret, size_t secretSize);
+
+
+/*******   Streaming   *******/
+/*
+ * Streaming requires state maintenance.
+ * This operation costs memory and CPU.
+ * As a consequence, streaming is slower than one-shot hashing.
+ * For better performance, prefer one-shot functions whenever applicable.
+ */
+
+/*!
+ * @brief The state struct for the XXH3 streaming API.
+ *
+ * @see XXH3_state_s for details.
+ */
+typedef struct XXH3_state_s XXH3_state_t;
+XXH_PUBLIC_API XXH3_state_t* XXH3_createState(void);
+XXH_PUBLIC_API XXH_errorcode XXH3_freeState(XXH3_state_t* statePtr);
+XXH_PUBLIC_API void XXH3_copyState(XXH3_state_t* dst_state, const XXH3_state_t* src_state);
+
+/*
+ * XXH3_64bits_reset():
+ * Initialize with default parameters.
+ * digest will be equivalent to `XXH3_64bits()`.
+ */
+XXH_PUBLIC_API XXH_errorcode XXH3_64bits_reset(XXH3_state_t* statePtr);
+/*
+ * XXH3_64bits_reset_withSeed():
+ * Generate a custom secret from `seed`, and store it into `statePtr`.
+ * digest will be equivalent to `XXH3_64bits_withSeed()`.
+ */
+XXH_PUBLIC_API XXH_errorcode XXH3_64bits_reset_withSeed(XXH3_state_t* statePtr, XXH64_hash_t seed);
+/*
+ * XXH3_64bits_reset_withSecret():
+ * `secret` is referenced, it _must outlive_ the hash streaming session.
+ * Similar to one-shot API, `secretSize` must be >= `XXH3_SECRET_SIZE_MIN`,
+ * and the quality of produced hash values depends on secret's entropy
+ * (secret's content should look like a bunch of random bytes).
+ * When in doubt about the randomness of a candidate `secret`,
+ * consider employing `XXH3_generateSecret()` instead (see below).
+ */
+XXH_PUBLIC_API XXH_errorcode XXH3_64bits_reset_withSecret(XXH3_state_t* statePtr, const void* secret, size_t secretSize);
+
+XXH_PUBLIC_API XXH_errorcode XXH3_64bits_update (XXH3_state_t* statePtr, const void* input, size_t length);
+XXH_PUBLIC_API XXH64_hash_t  XXH3_64bits_digest (const XXH3_state_t* statePtr);
+
+/* note : canonical representation of XXH3 is the same as XXH64
+ * since they both produce XXH64_hash_t values */
+
+
+/*-**********************************************************************
+*  XXH3 128-bit variant
+************************************************************************/
+
+/*!
+ * @brief The return value from 128-bit hashes.
+ *
+ * Stored in little endian order, although the fields themselves are in native
+ * endianness.
+ */
+typedef struct {
+    XXH64_hash_t low64;   /*!< `value & 0xFFFFFFFFFFFFFFFF` */
+    XXH64_hash_t high64;  /*!< `value >> 64` */
+} XXH128_hash_t;
+
+XXH_PUBLIC_API XXH128_hash_t XXH3_128bits(const void* data, size_t len);
+XXH_PUBLIC_API XXH128_hash_t XXH3_128bits_withSeed(const void* data, size_t len, XXH64_hash_t seed);
+XXH_PUBLIC_API XXH128_hash_t XXH3_128bits_withSecret(const void* data, size_t len, const void* secret, size_t secretSize);
+
+/*******   Streaming   *******/
+/*
+ * Streaming requires state maintenance.
+ * This operation costs memory and CPU.
+ * As a consequence, streaming is slower than one-shot hashing.
+ * For better performance, prefer one-shot functions whenever applicable.
+ *
+ * XXH3_128bits uses the same XXH3_state_t as XXH3_64bits().
+ * Use already declared XXH3_createState() and XXH3_freeState().
+ *
+ * All reset and streaming functions have same meaning as their 64-bit counterpart.
+ */
+
+XXH_PUBLIC_API XXH_errorcode XXH3_128bits_reset(XXH3_state_t* statePtr);
+XXH_PUBLIC_API XXH_errorcode XXH3_128bits_reset_withSeed(XXH3_state_t* statePtr, XXH64_hash_t seed);
+XXH_PUBLIC_API XXH_errorcode XXH3_128bits_reset_withSecret(XXH3_state_t* statePtr, const void* secret, size_t secretSize);
+
+XXH_PUBLIC_API XXH_errorcode XXH3_128bits_update (XXH3_state_t* statePtr, const void* input, size_t length);
+XXH_PUBLIC_API XXH128_hash_t XXH3_128bits_digest (const XXH3_state_t* statePtr);
+
+/* Following helper functions make it possible to compare XXH128_hast_t values.
+ * Since XXH128_hash_t is a structure, this capability is not offered by the language.
+ * Note: For better performance, these functions can be inlined using XXH_INLINE_ALL */
+
+/*!
+ * XXH128_isEqual():
+ * Return: 1 if `h1` and `h2` are equal, 0 if they are not.
+ */
+XXH_PUBLIC_API int XXH128_isEqual(XXH128_hash_t h1, XXH128_hash_t h2);
+
+/*!
+ * XXH128_cmp():
+ *
+ * This comparator is compatible with stdlib's `qsort()`/`bsearch()`.
+ *
+ * return: >0 if *h128_1  > *h128_2
+ *         =0 if *h128_1 == *h128_2
+ *         <0 if *h128_1  < *h128_2
+ */
+XXH_PUBLIC_API int XXH128_cmp(const void* h128_1, const void* h128_2);
+
+
+/*******   Canonical representation   *******/
+typedef struct { unsigned char digest[sizeof(XXH128_hash_t)]; } XXH128_canonical_t;
+XXH_PUBLIC_API void XXH128_canonicalFromHash(XXH128_canonical_t* dst, XXH128_hash_t hash);
+XXH_PUBLIC_API XXH128_hash_t XXH128_hashFromCanonical(const XXH128_canonical_t* src);
+
+
+#endif  /* XXH_NO_LONG_LONG */
+
+/*!
+ * @}
+ */
+#endif /* XXHASH_H_5627135585666179 */
+
+
+
+#if defined(XXH_STATIC_LINKING_ONLY) && !defined(XXHASH_H_STATIC_13879238742)
+#define XXHASH_H_STATIC_13879238742
+/* ****************************************************************************
+ * This section contains declarations which are not guaranteed to remain stable.
+ * They may change in future versions, becoming incompatible with a different
+ * version of the library.
+ * These declarations should only be used with static linking.
+ * Never use them in association with dynamic linking!
+ ***************************************************************************** */
+
+/*
+ * These definitions are only present to allow static allocation
+ * of XXH states, on stack or in a struct, for example.
+ * Never **ever** access their members directly.
+ */
+
+/*!
+ * @internal
+ * @brief Structure for XXH32 streaming API.
+ *
+ * @note This is only defined when @ref XXH_STATIC_LINKING_ONLY,
+ * @ref XXH_INLINE_ALL, or @ref XXH_IMPLEMENTATION is defined. Otherwise it is
+ * an opaque type. This allows fields to safely be changed.
+ *
+ * Typedef'd to @ref XXH32_state_t.
+ * Do not access the members of this struct directly.
+ * @see XXH64_state_s, XXH3_state_s
+ */
+struct XXH32_state_s {
+   XXH32_hash_t total_len_32; /*!< Total length hashed, modulo 2^32 */
+   XXH32_hash_t large_len;    /*!< Whether the hash is >= 16 (handles @ref total_len_32 overflow) */
+   XXH32_hash_t v[4];         /*!< Accumulator lanes */
+   XXH32_hash_t mem32[4];     /*!< Internal buffer for partial reads. Treated as unsigned char[16]. */
+   XXH32_hash_t memsize;      /*!< Amount of data in @ref mem32 */
+   XXH32_hash_t reserved;     /*!< Reserved field. Do not read or write to it, it may be removed. */
+};   /* typedef'd to XXH32_state_t */
+
+
+#ifndef XXH_NO_LONG_LONG  /* defined when there is no 64-bit support */
+
+/*!
+ * @internal
+ * @brief Structure for XXH64 streaming API.
+ *
+ * @note This is only defined when @ref XXH_STATIC_LINKING_ONLY,
+ * @ref XXH_INLINE_ALL, or @ref XXH_IMPLEMENTATION is defined. Otherwise it is
+ * an opaque type. This allows fields to safely be changed.
+ *
+ * Typedef'd to @ref XXH64_state_t.
+ * Do not access the members of this struct directly.
+ * @see XXH32_state_s, XXH3_state_s
+ */
+struct XXH64_state_s {
+   XXH64_hash_t total_len;    /*!< Total length hashed. This is always 64-bit. */
+   XXH64_hash_t v[4];         /*!< Accumulator lanes */
+   XXH64_hash_t mem64[4];     /*!< Internal buffer for partial reads. Treated as unsigned char[32]. */
+   XXH32_hash_t memsize;      /*!< Amount of data in @ref mem64 */
+   XXH32_hash_t reserved32;   /*!< Reserved field, needed for padding anyways*/
+   XXH64_hash_t reserved64;   /*!< Reserved field. Do not read or write to it, it may be removed. */
+};   /* typedef'd to XXH64_state_t */
+
+#if defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L) /* >= C11 */
+#  include <stdalign.h>
+#  define XXH_ALIGN(n)      alignas(n)
+#elif defined(__cplusplus) && (__cplusplus >= 201103L) /* >= C++11 */
+/* In C++ alignas() is a keyword */
+#  define XXH_ALIGN(n)      alignas(n)
+#elif defined(__GNUC__)
+#  define XXH_ALIGN(n)      __attribute__ ((aligned(n)))
+#elif defined(_MSC_VER)
+#  define XXH_ALIGN(n)      __declspec(align(n))
+#else
+#  define XXH_ALIGN(n)   /* disabled */
+#endif
+
+/* Old GCC versions only accept the attribute after the type in structures. */
+#if !(defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L))   /* C11+ */ \
+    && ! (defined(__cplusplus) && (__cplusplus >= 201103L)) /* >= C++11 */ \
+    && defined(__GNUC__)
+#   define XXH_ALIGN_MEMBER(align, type) type XXH_ALIGN(align)
+#else
+#   define XXH_ALIGN_MEMBER(align, type) XXH_ALIGN(align) type
+#endif
+
+/*!
+ * @brief The size of the internal XXH3 buffer.
+ *
+ * This is the optimal update size for incremental hashing.
+ *
+ * @see XXH3_64b_update(), XXH3_128b_update().
+ */
+#define XXH3_INTERNALBUFFER_SIZE 256
+
+/*!
+ * @brief Default size of the secret buffer (and @ref XXH3_kSecret).
+ *
+ * This is the size used in @ref XXH3_kSecret and the seeded functions.
+ *
+ * Not to be confused with @ref XXH3_SECRET_SIZE_MIN.
+ */
+#define XXH3_SECRET_DEFAULT_SIZE 192
+
+/*!
+ * @internal
+ * @brief Structure for XXH3 streaming API.
+ *
+ * @note This is only defined when @ref XXH_STATIC_LINKING_ONLY,
+ * @ref XXH_INLINE_ALL, or @ref XXH_IMPLEMENTATION is defined.
+ * Otherwise it is an opaque type.
+ * Never use this definition in combination with dynamic library.
+ * This allows fields to safely be changed in the future.
+ *
+ * @note ** This structure has a strict alignment requirement of 64 bytes!! **
+ * Do not allocate this with `malloc()` or `new`,
+ * it will not be sufficiently aligned.
+ * Use @ref XXH3_createState() and @ref XXH3_freeState(), or stack allocation.
+ *
+ * Typedef'd to @ref XXH3_state_t.
+ * Do never access the members of this struct directly.
+ *
+ * @see XXH3_INITSTATE() for stack initialization.
+ * @see XXH3_createState(), XXH3_freeState().
+ * @see XXH32_state_s, XXH64_state_s
+ */
+struct XXH3_state_s {
+   XXH_ALIGN_MEMBER(64, XXH64_hash_t acc[8]);
+       /*!< The 8 accumulators. Similar to `vN` in @ref XXH32_state_s::v1 and @ref XXH64_state_s */
+   XXH_ALIGN_MEMBER(64, unsigned char customSecret[XXH3_SECRET_DEFAULT_SIZE]);
+       /*!< Used to store a custom secret generated from a seed. */
+   XXH_ALIGN_MEMBER(64, unsigned char buffer[XXH3_INTERNALBUFFER_SIZE]);
+       /*!< The internal buffer. @see XXH32_state_s::mem32 */
+   XXH32_hash_t bufferedSize;
+       /*!< The amount of memory in @ref buffer, @see XXH32_state_s::memsize */
+   XXH32_hash_t useSeed;
+       /*!< Reserved field. Needed for padding on 64-bit. */
+   size_t nbStripesSoFar;
+       /*!< Number or stripes processed. */
+   XXH64_hash_t totalLen;
+       /*!< Total length hashed. 64-bit even on 32-bit targets. */
+   size_t nbStripesPerBlock;
+       /*!< Number of stripes per block. */
+   size_t secretLimit;
+       /*!< Size of @ref customSecret or @ref extSecret */
+   XXH64_hash_t seed;
+       /*!< Seed for _withSeed variants. Must be zero otherwise, @see XXH3_INITSTATE() */
+   XXH64_hash_t reserved64;
+       /*!< Reserved field. */
+   const unsigned char* extSecret;
+       /*!< Reference to an external secret for the _withSecret variants, NULL
+        *   for other variants. */
+   /* note: there may be some padding at the end due to alignment on 64 bytes */
+}; /* typedef'd to XXH3_state_t */
+
+#undef XXH_ALIGN_MEMBER
+
+/*!
+ * @brief Initializes a stack-allocated `XXH3_state_s`.
+ *
+ * When the @ref XXH3_state_t structure is merely emplaced on stack,
+ * it should be initialized with XXH3_INITSTATE() or a memset()
+ * in case its first reset uses XXH3_NNbits_reset_withSeed().
+ * This init can be omitted if the first reset uses default or _withSecret mode.
+ * This operation isn't necessary when the state is created with XXH3_createState().
+ * Note that this doesn't prepare the state for a streaming operation,
+ * it's still necessary to use XXH3_NNbits_reset*() afterwards.
+ */
+#define XXH3_INITSTATE(XXH3_state_ptr)   { (XXH3_state_ptr)->seed = 0; }
+
+
+/* XXH128() :
+ * simple alias to pre-selected XXH3_128bits variant
+ */
+XXH_PUBLIC_API XXH128_hash_t XXH128(const void* data, size_t len, XXH64_hash_t seed);
+
+
+/* ===   Experimental API   === */
+/* Symbols defined below must be considered tied to a specific library version. */
+
+/*
+ * XXH3_generateSecret():
+ *
+ * Derive a high-entropy secret from any user-defined content, named customSeed.
+ * The generated secret can be used in combination with `*_withSecret()` functions.
+ * The `_withSecret()` variants are useful to provide a higher level of protection than 64-bit seed,
+ * as it becomes much more difficult for an external actor to guess how to impact the calculation logic.
+ *
+ * The function accepts as input a custom seed of any length and any content,
+ * and derives from it a high-entropy secret of length @secretSize
+ * into an already allocated buffer @secretBuffer.
+ * @secretSize must be >= XXH3_SECRET_SIZE_MIN
+ *
+ * The generated secret can then be used with any `*_withSecret()` variant.
+ * Functions `XXH3_128bits_withSecret()`, `XXH3_64bits_withSecret()`,
+ * `XXH3_128bits_reset_withSecret()` and `XXH3_64bits_reset_withSecret()`
+ * are part of this list. They all accept a `secret` parameter
+ * which must be large enough for implementation reasons (>= XXH3_SECRET_SIZE_MIN)
+ * _and_ feature very high entropy (consist of random-looking bytes).
+ * These conditions can be a high bar to meet, so
+ * XXH3_generateSecret() can be employed to ensure proper quality.
+ *
+ * customSeed can be anything. It can have any size, even small ones,
+ * and its content can be anything, even "poor entropy" sources such as a bunch of zeroes.
+ * The resulting `secret` will nonetheless provide all required qualities.
+ *
+ * When customSeedSize > 0, supplying NULL as customSeed is undefined behavior.
+ */
+XXH_PUBLIC_API XXH_errorcode XXH3_generateSecret(void* secretBuffer, size_t secretSize, const void* customSeed, size_t customSeedSize);
+
+
+/*
+ * XXH3_generateSecret_fromSeed():
+ *
+ * Generate the same secret as the _withSeed() variants.
+ *
+ * The resulting secret has a length of XXH3_SECRET_DEFAULT_SIZE (necessarily).
+ * @secretBuffer must be already allocated, of size at least XXH3_SECRET_DEFAULT_SIZE bytes.
+ *
+ * The generated secret can be used in combination with
+ *`*_withSecret()` and `_withSecretandSeed()` variants.
+ * This generator is notably useful in combination with `_withSecretandSeed()`,
+ * as a way to emulate a faster `_withSeed()` variant.
+ */
+XXH_PUBLIC_API void XXH3_generateSecret_fromSeed(void* secretBuffer, XXH64_hash_t seed);
+
+/*
+ * *_withSecretandSeed() :
+ * These variants generate hash values using either
+ * @seed for "short" keys (< XXH3_MIDSIZE_MAX = 240 bytes)
+ * or @secret for "large" keys (>= XXH3_MIDSIZE_MAX).
+ *
+ * This generally benefits speed, compared to `_withSeed()` or `_withSecret()`.
+ * `_withSeed()` has to generate the secret on the fly for "large" keys.
+ * It's fast, but can be perceptible for "not so large" keys (< 1 KB).
+ * `_withSecret()` has to generate the masks on the fly for "small" keys,
+ * which requires more instructions than _withSeed() variants.
+ * Therefore, _withSecretandSeed variant combines the best of both worlds.
+ *
+ * When @secret has been generated by XXH3_generateSecret_fromSeed(),
+ * this variant produces *exactly* the same results as `_withSeed()` variant,
+ * hence offering only a pure speed benefit on "large" input,
+ * by skipping the need to regenerate the secret for every large input.
+ *
+ * Another usage scenario is to hash the secret to a 64-bit hash value,
+ * for example with XXH3_64bits(), which then becomes the seed,
+ * and then employ both the seed and the secret in _withSecretandSeed().
+ * On top of speed, an added benefit is that each bit in the secret
+ * has a 50% chance to swap each bit in the output,
+ * via its impact to the seed.
+ * This is not guaranteed when using the secret directly in "small data" scenarios,
+ * because only portions of the secret are employed for small data.
+ */
+XXH_PUBLIC_API XXH64_hash_t
+XXH3_64bits_withSecretandSeed(const void* data, size_t len,
+                              const void* secret, size_t secretSize,
+                              XXH64_hash_t seed);
+
+XXH_PUBLIC_API XXH128_hash_t
+XXH3_128bits_withSecretandSeed(const void* data, size_t len,
+                               const void* secret, size_t secretSize,
+                               XXH64_hash_t seed64);
+
+XXH_PUBLIC_API XXH_errorcode
+XXH3_64bits_reset_withSecretandSeed(XXH3_state_t* statePtr,
+                                    const void* secret, size_t secretSize,
+                                    XXH64_hash_t seed64);
+
+XXH_PUBLIC_API XXH_errorcode
+XXH3_128bits_reset_withSecretandSeed(XXH3_state_t* statePtr,
+                                     const void* secret, size_t secretSize,
+                                     XXH64_hash_t seed64);
+
+
+#endif  /* XXH_NO_LONG_LONG */
+#if defined(XXH_INLINE_ALL) || defined(XXH_PRIVATE_API)
+#  define XXH_IMPLEMENTATION
+#endif
+
+#endif  /* defined(XXH_STATIC_LINKING_ONLY) && !defined(XXHASH_H_STATIC_13879238742) */
+
+
+/* ======================================================================== */
+/* ======================================================================== */
+/* ======================================================================== */
+
+
+/*-**********************************************************************
+ * xxHash implementation
+ *-**********************************************************************
+ * xxHash's implementation used to be hosted inside xxhash.c.
+ *
+ * However, inlining requires implementation to be visible to the compiler,
+ * hence be included alongside the header.
+ * Previously, implementation was hosted inside xxhash.c,
+ * which was then #included when inlining was activated.
+ * This construction created issues with a few build and install systems,
+ * as it required xxhash.c to be stored in /include directory.
+ *
+ * xxHash implementation is now directly integrated within xxhash.h.
+ * As a consequence, xxhash.c is no longer needed in /include.
+ *
+ * xxhash.c is still available and is still useful.
+ * In a "normal" setup, when xxhash is not inlined,
+ * xxhash.h only exposes the prototypes and public symbols,
+ * while xxhash.c can be built into an object file xxhash.o
+ * which can then be linked into the final binary.
+ ************************************************************************/
+
+#if ( defined(XXH_INLINE_ALL) || defined(XXH_PRIVATE_API) \
+   || defined(XXH_IMPLEMENTATION) ) && !defined(XXH_IMPLEM_13a8737387)
+#  define XXH_IMPLEM_13a8737387
+
+/* *************************************
+*  Tuning parameters
+***************************************/
+
+/*!
+ * @defgroup tuning Tuning parameters
+ * @{
+ *
+ * Various macros to control xxHash's behavior.
+ */
+#ifdef XXH_DOXYGEN
+/*!
+ * @brief Define this to disable 64-bit code.
+ *
+ * Useful if only using the @ref xxh32_family and you have a strict C90 compiler.
+ */
+#  define XXH_NO_LONG_LONG
+#  undef XXH_NO_LONG_LONG /* don't actually */
+/*!
+ * @brief Controls how unaligned memory is accessed.
+ *
+ * By default, access to unaligned memory is controlled by `memcpy()`, which is
+ * safe and portable.
+ *
+ * Unfortunately, on some target/compiler combinations, the generated assembly
+ * is sub-optimal.
+ *
+ * The below switch allow selection of a different access method
+ * in the search for improved performance.
+ *
+ * @par Possible options:
+ *
+ *  - `XXH_FORCE_MEMORY_ACCESS=0` (default): `memcpy`
+ *   @par
+ *     Use `memcpy()`. Safe and portable. Note that most modern compilers will
+ *     eliminate the function call and treat it as an unaligned access.
+ *
+ *  - `XXH_FORCE_MEMORY_ACCESS=1`: `__attribute__((packed))`
+ *   @par
+ *     Depends on compiler extensions and is therefore not portable.
+ *     This method is safe _if_ your compiler supports it,
+ *     and *generally* as fast or faster than `memcpy`.
+ *
+ *  - `XXH_FORCE_MEMORY_ACCESS=2`: Direct cast
+ *  @par
+ *     Casts directly and dereferences. This method doesn't depend on the
+ *     compiler, but it violates the C standard as it directly dereferences an
+ *     unaligned pointer. It can generate buggy code on targets which do not
+ *     support unaligned memory accesses, but in some circumstances, it's the
+ *     only known way to get the most performance.
+ *
+ *  - `XXH_FORCE_MEMORY_ACCESS=3`: Byteshift
+ *  @par
+ *     Also portable. This can generate the best code on old compilers which don't
+ *     inline small `memcpy()` calls, and it might also be faster on big-endian
+ *     systems which lack a native byteswap instruction. However, some compilers
+ *     will emit literal byteshifts even if the target supports unaligned access.
+ *  .
+ *
+ * @warning
+ *   Methods 1 and 2 rely on implementation-defined behavior. Use these with
+ *   care, as what works on one compiler/platform/optimization level may cause
+ *   another to read garbage data or even crash.
+ *
+ * See http://fastcompression.blogspot.com/2015/08/accessing-unaligned-memory.html for details.
+ *
+ * Prefer these methods in priority order (0 > 3 > 1 > 2)
+ */
+#  define XXH_FORCE_MEMORY_ACCESS 0
+
+/*!
+ * @def XXH_FORCE_ALIGN_CHECK
+ * @brief If defined to non-zero, adds a special path for aligned inputs (XXH32()
+ * and XXH64() only).
+ *
+ * This is an important performance trick for architectures without decent
+ * unaligned memory access performance.
+ *
+ * It checks for input alignment, and when conditions are met, uses a "fast
+ * path" employing direct 32-bit/64-bit reads, resulting in _dramatically
+ * faster_ read speed.
+ *
+ * The check costs one initial branch per hash, which is generally negligible,
+ * but not zero.
+ *
+ * Moreover, it's not useful to generate an additional code path if memory
+ * access uses the same instruction for both aligned and unaligned
+ * addresses (e.g. x86 and aarch64).
+ *
+ * In these cases, the alignment check can be removed by setting this macro to 0.
+ * Then the code will always use unaligned memory access.
+ * Align check is automatically disabled on x86, x64 & arm64,
+ * which are platforms known to offer good unaligned memory accesses performance.
+ *
+ * This option does not affect XXH3 (only XXH32 and XXH64).
+ */
+#  define XXH_FORCE_ALIGN_CHECK 0
+
+/*!
+ * @def XXH_NO_INLINE_HINTS
+ * @brief When non-zero, sets all functions to `static`.
+ *
+ * By default, xxHash tries to force the compiler to inline almost all internal
+ * functions.
+ *
+ * This can usually improve performance due to reduced jumping and improved
+ * constant folding, but significantly increases the size of the binary which
+ * might not be favorable.
+ *
+ * Additionally, sometimes the forced inlining can be detrimental to performance,
+ * depending on the architecture.
+ *
+ * XXH_NO_INLINE_HINTS marks all internal functions as static, giving the
+ * compiler full control on whether to inline or not.
+ *
+ * When not optimizing (-O0), optimizing for size (-Os, -Oz), or using
+ * -fno-inline with GCC or Clang, this will automatically be defined.
+ */
+#  define XXH_NO_INLINE_HINTS 0
+
+/*!
+ * @def XXH32_ENDJMP
+ * @brief Whether to use a jump for `XXH32_finalize`.
+ *
+ * For performance, `XXH32_finalize` uses multiple branches in the finalizer.
+ * This is generally preferable for performance,
+ * but depending on exact architecture, a jmp may be preferable.
+ *
+ * This setting is only possibly making a difference for very small inputs.
+ */
+#  define XXH32_ENDJMP 0
+
+/*!
+ * @internal
+ * @brief Redefines old internal names.
+ *
+ * For compatibility with code that uses xxHash's internals before the names
+ * were changed to improve namespacing. There is no other reason to use this.
+ */
+#  define XXH_OLD_NAMES
+#  undef XXH_OLD_NAMES /* don't actually use, it is ugly. */
+#endif /* XXH_DOXYGEN */
+/*!
+ * @}
+ */
+
+#ifndef XXH_FORCE_MEMORY_ACCESS   /* can be defined externally, on command line for example */
+   /* prefer __packed__ structures (method 1) for gcc on armv7+ and mips */
+#  if !defined(__clang__) && \
+( \
+    (defined(__INTEL_COMPILER) && !defined(_WIN32)) || \
+    ( \
+        defined(__GNUC__) && ( \
+            (defined(__ARM_ARCH) && __ARM_ARCH >= 7) || \
+            ( \
+                defined(__mips__) && \
+                (__mips <= 5 || __mips_isa_rev < 6) && \
+                (!defined(__mips16) || defined(__mips_mips16e2)) \
+            ) \
+        ) \
+    ) \
+)
+#    define XXH_FORCE_MEMORY_ACCESS 1
+#  endif
+#endif
+
+#ifndef XXH_FORCE_ALIGN_CHECK  /* can be defined externally */
+#  if defined(__i386)  || defined(__x86_64__) || defined(__aarch64__) \
+   || defined(_M_IX86) || defined(_M_X64)     || defined(_M_ARM64) /* visual */
+#    define XXH_FORCE_ALIGN_CHECK 0
+#  else
+#    define XXH_FORCE_ALIGN_CHECK 1
+#  endif
+#endif
+
+#ifndef XXH_NO_INLINE_HINTS
+#  if defined(__OPTIMIZE_SIZE__) /* -Os, -Oz */ \
+   || defined(__NO_INLINE__)     /* -O0, -fno-inline */
+#    define XXH_NO_INLINE_HINTS 1
+#  else
+#    define XXH_NO_INLINE_HINTS 0
+#  endif
+#endif
+
+#ifndef XXH32_ENDJMP
+/* generally preferable for performance */
+#  define XXH32_ENDJMP 0
+#endif
+
+/*!
+ * @defgroup impl Implementation
+ * @{
+ */
+
+
+/* *************************************
+*  Includes & Memory related functions
+***************************************/
+/*
+ * Modify the local functions below should you wish to use
+ * different memory routines for malloc() and free()
+ */
+#include <stdlib.h>
+
+/*!
+ * @internal
+ * @brief Modify this function to use a different routine than malloc().
+ */
+static void* XXH_malloc(size_t s) { return malloc(s); }
+
+/*!
+ * @internal
+ * @brief Modify this function to use a different routine than free().
+ */
+static void XXH_free(void* p) { free(p); }
+
+#include <string.h>
+
+/*!
+ * @internal
+ * @brief Modify this function to use a different routine than memcpy().
+ */
+static void* XXH_memcpy(void* dest, const void* src, size_t size)
+{
+    return memcpy(dest,src,size);
+}
+
+#include <limits.h>   /* ULLONG_MAX */
+
+
+/* *************************************
+*  Compiler Specific Options
+***************************************/
+#ifdef _MSC_VER /* Visual Studio warning fix */
+#  pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */
+#endif
+
+#if XXH_NO_INLINE_HINTS  /* disable inlining hints */
+#  if defined(__GNUC__) || defined(__clang__)
+#    define XXH_FORCE_INLINE static __attribute__((unused))
+#  else
+#    define XXH_FORCE_INLINE static
+#  endif
+#  define XXH_NO_INLINE static
+/* enable inlining hints */
+#elif defined(__GNUC__) || defined(__clang__)
+#  define XXH_FORCE_INLINE static __inline__ __attribute__((always_inline, unused))
+#  define XXH_NO_INLINE static __attribute__((noinline))
+#elif defined(_MSC_VER)  /* Visual Studio */
+#  define XXH_FORCE_INLINE static __forceinline
+#  define XXH_NO_INLINE static __declspec(noinline)
+#elif defined (__cplusplus) \
+  || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L))   /* C99 */
+#  define XXH_FORCE_INLINE static inline
+#  define XXH_NO_INLINE static
+#else
+#  define XXH_FORCE_INLINE static
+#  define XXH_NO_INLINE static
+#endif
+
+
+
+/* *************************************
+*  Debug
+***************************************/
+/*!
+ * @ingroup tuning
+ * @def XXH_DEBUGLEVEL
+ * @brief Sets the debugging level.
+ *
+ * XXH_DEBUGLEVEL is expected to be defined externally, typically via the
+ * compiler's command line options. The value must be a number.
+ */
+#ifndef XXH_DEBUGLEVEL
+#  ifdef DEBUGLEVEL /* backwards compat */
+#    define XXH_DEBUGLEVEL DEBUGLEVEL
+#  else
+#    define XXH_DEBUGLEVEL 0
+#  endif
+#endif
+
+#if (XXH_DEBUGLEVEL>=1)
+#  include <assert.h>   /* note: can still be disabled with NDEBUG */
+#  define XXH_ASSERT(c)   assert(c)
+#else
+#  define XXH_ASSERT(c)   ((void)0)
+#endif
+
+/* note: use after variable declarations */
+#ifndef XXH_STATIC_ASSERT
+#  if defined(__STDC_VERSION__) && (__STDC_VERSION__ >= 201112L)    /* C11 */
+#    include <assert.h>
+#    define XXH_STATIC_ASSERT_WITH_MESSAGE(c,m) do { static_assert((c),m); } while(0)
+#  elif defined(__cplusplus) && (__cplusplus >= 201103L)            /* C++11 */
+#    define XXH_STATIC_ASSERT_WITH_MESSAGE(c,m) do { static_assert((c),m); } while(0)
+#  else
+#    define XXH_STATIC_ASSERT_WITH_MESSAGE(c,m) do { struct xxh_sa { char x[(c) ? 1 : -1]; }; } while(0)
+#  endif
+#  define XXH_STATIC_ASSERT(c) XXH_STATIC_ASSERT_WITH_MESSAGE((c),#c)
+#endif
+
+/*!
+ * @internal
+ * @def XXH_COMPILER_GUARD(var)
+ * @brief Used to prevent unwanted optimizations for @p var.
+ *
+ * It uses an empty GCC inline assembly statement with a register constraint
+ * which forces @p var into a general purpose register (eg eax, ebx, ecx
+ * on x86) and marks it as modified.
+ *
+ * This is used in a few places to avoid unwanted autovectorization (e.g.
+ * XXH32_round()). All vectorization we want is explicit via intrinsics,
+ * and _usually_ isn't wanted elsewhere.
+ *
+ * We also use it to prevent unwanted constant folding for AArch64 in
+ * XXH3_initCustomSecret_scalar().
+ */
+#if defined(__GNUC__) || defined(__clang__)
+#  define XXH_COMPILER_GUARD(var) __asm__ __volatile__("" : "+r" (var))
+#else
+#  define XXH_COMPILER_GUARD(var) ((void)0)
+#endif
+
+/* *************************************
+*  Basic Types
+***************************************/
+#if !defined (__VMS) \
+ && (defined (__cplusplus) \
+ || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) )
+# include <stdint.h>
+  typedef uint8_t xxh_u8;
+#else
+  typedef unsigned char xxh_u8;
+#endif
+typedef XXH32_hash_t xxh_u32;
+
+#ifdef XXH_OLD_NAMES
+#  define BYTE xxh_u8
+#  define U8   xxh_u8
+#  define U32  xxh_u32
+#endif
+
+/* ***   Memory access   *** */
+
+/*!
+ * @internal
+ * @fn xxh_u32 XXH_read32(const void* ptr)
+ * @brief Reads an unaligned 32-bit integer from @p ptr in native endianness.
+ *
+ * Affected by @ref XXH_FORCE_MEMORY_ACCESS.
+ *
+ * @param ptr The pointer to read from.
+ * @return The 32-bit native endian integer from the bytes at @p ptr.
+ */
+
+/*!
+ * @internal
+ * @fn xxh_u32 XXH_readLE32(const void* ptr)
+ * @brief Reads an unaligned 32-bit little endian integer from @p ptr.
+ *
+ * Affected by @ref XXH_FORCE_MEMORY_ACCESS.
+ *
+ * @param ptr The pointer to read from.
+ * @return The 32-bit little endian integer from the bytes at @p ptr.
+ */
+
+/*!
+ * @internal
+ * @fn xxh_u32 XXH_readBE32(const void* ptr)
+ * @brief Reads an unaligned 32-bit big endian integer from @p ptr.
+ *
+ * Affected by @ref XXH_FORCE_MEMORY_ACCESS.
+ *
+ * @param ptr The pointer to read from.
+ * @return The 32-bit big endian integer from the bytes at @p ptr.
+ */
+
+/*!
+ * @internal
+ * @fn xxh_u32 XXH_readLE32_align(const void* ptr, XXH_alignment align)
+ * @brief Like @ref XXH_readLE32(), but has an option for aligned reads.
+ *
+ * Affected by @ref XXH_FORCE_MEMORY_ACCESS.
+ * Note that when @ref XXH_FORCE_ALIGN_CHECK == 0, the @p align parameter is
+ * always @ref XXH_alignment::XXH_unaligned.
+ *
+ * @param ptr The pointer to read from.
+ * @param align Whether @p ptr is aligned.
+ * @pre
+ *   If @p align == @ref XXH_alignment::XXH_aligned, @p ptr must be 4 byte
+ *   aligned.
+ * @return The 32-bit little endian integer from the bytes at @p ptr.
+ */
+
+#if (defined(XXH_FORCE_MEMORY_ACCESS) && (XXH_FORCE_MEMORY_ACCESS==3))
+/*
+ * Manual byteshift. Best for old compilers which don't inline memcpy.
+ * We actually directly use XXH_readLE32 and XXH_readBE32.
+ */
+#elif (defined(XXH_FORCE_MEMORY_ACCESS) && (XXH_FORCE_MEMORY_ACCESS==2))
+
+/*
+ * Force direct memory access. Only works on CPU which support unaligned memory
+ * access in hardware.
+ */
+static xxh_u32 XXH_read32(const void* memPtr) { return *(const xxh_u32*) memPtr; }
+
+#elif (defined(XXH_FORCE_MEMORY_ACCESS) && (XXH_FORCE_MEMORY_ACCESS==1))
+
+/*
+ * __pack instructions are safer but compiler specific, hence potentially
+ * problematic for some compilers.
+ *
+ * Currently only defined for GCC and ICC.
+ */
+#ifdef XXH_OLD_NAMES
+typedef union { xxh_u32 u32; } __attribute__((packed)) unalign;
+#endif
+static xxh_u32 XXH_read32(const void* ptr)
+{
+    typedef union { xxh_u32 u32; } __attribute__((packed)) xxh_unalign;
+    return ((const xxh_unalign*)ptr)->u32;
+}
+
+#else
+
+/*
+ * Portable and safe solution. Generally efficient.
+ * see: http://fastcompression.blogspot.com/2015/08/accessing-unaligned-memory.html
+ */
+static xxh_u32 XXH_read32(const void* memPtr)
+{
+    xxh_u32 val;
+    XXH_memcpy(&val, memPtr, sizeof(val));
+    return val;
+}
+
+#endif   /* XXH_FORCE_DIRECT_MEMORY_ACCESS */
+
+
+/* ***   Endianness   *** */
+
+/*!
+ * @ingroup tuning
+ * @def XXH_CPU_LITTLE_ENDIAN
+ * @brief Whether the target is little endian.
+ *
+ * Defined to 1 if the target is little endian, or 0 if it is big endian.
+ * It can be defined externally, for example on the compiler command line.
+ *
+ * If it is not defined,
+ * a runtime check (which is usually constant folded) is used instead.
+ *
+ * @note
+ *   This is not necessarily defined to an integer constant.
+ *
+ * @see XXH_isLittleEndian() for the runtime check.
+ */
+#ifndef XXH_CPU_LITTLE_ENDIAN
+/*
+ * Try to detect endianness automatically, to avoid the nonstandard behavior
+ * in `XXH_isLittleEndian()`
+ */
+#  if defined(_WIN32) /* Windows is always little endian */ \
+     || defined(__LITTLE_ENDIAN__) \
+     || (defined(__BYTE_ORDER__) && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__)
+#    define XXH_CPU_LITTLE_ENDIAN 1
+#  elif defined(__BIG_ENDIAN__) \
+     || (defined(__BYTE_ORDER__) && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__)
+#    define XXH_CPU_LITTLE_ENDIAN 0
+#  else
+/*!
+ * @internal
+ * @brief Runtime check for @ref XXH_CPU_LITTLE_ENDIAN.
+ *
+ * Most compilers will constant fold this.
+ */
+static int XXH_isLittleEndian(void)
+{
+    /*
+     * Portable and well-defined behavior.
+     * Don't use static: it is detrimental to performance.
+     */
+    const union { xxh_u32 u; xxh_u8 c[4]; } one = { 1 };
+    return one.c[0];
+}
+#   define XXH_CPU_LITTLE_ENDIAN   XXH_isLittleEndian()
+#  endif
+#endif
+
+
+
+
+/* ****************************************
+*  Compiler-specific Functions and Macros
+******************************************/
+#define XXH_GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__)
+
+#ifdef __has_builtin
+#  define XXH_HAS_BUILTIN(x) __has_builtin(x)
+#else
+#  define XXH_HAS_BUILTIN(x) 0
+#endif
+
+/*!
+ * @internal
+ * @def XXH_rotl32(x,r)
+ * @brief 32-bit rotate left.
+ *
+ * @param x The 32-bit integer to be rotated.
+ * @param r The number of bits to rotate.
+ * @pre
+ *   @p r > 0 && @p r < 32
+ * @note
+ *   @p x and @p r may be evaluated multiple times.
+ * @return The rotated result.
+ */
+#if !defined(NO_CLANG_BUILTIN) && XXH_HAS_BUILTIN(__builtin_rotateleft32) \
+                               && XXH_HAS_BUILTIN(__builtin_rotateleft64)
+#  define XXH_rotl32 __builtin_rotateleft32
+#  define XXH_rotl64 __builtin_rotateleft64
+/* Note: although _rotl exists for minGW (GCC under windows), performance seems poor */
+#elif defined(_MSC_VER)
+#  define XXH_rotl32(x,r) _rotl(x,r)
+#  define XXH_rotl64(x,r) _rotl64(x,r)
+#else
+#  define XXH_rotl32(x,r) (((x) << (r)) | ((x) >> (32 - (r))))
+#  define XXH_rotl64(x,r) (((x) << (r)) | ((x) >> (64 - (r))))
+#endif
+
+/*!
+ * @internal
+ * @fn xxh_u32 XXH_swap32(xxh_u32 x)
+ * @brief A 32-bit byteswap.
+ *
+ * @param x The 32-bit integer to byteswap.
+ * @return @p x, byteswapped.
+ */
+#if defined(_MSC_VER)     /* Visual Studio */
+#  define XXH_swap32 _byteswap_ulong
+#elif XXH_GCC_VERSION >= 403
+#  define XXH_swap32 __builtin_bswap32
+#else
+static xxh_u32 XXH_swap32 (xxh_u32 x)
+{
+    return  ((x << 24) & 0xff000000 ) |
+            ((x <<  8) & 0x00ff0000 ) |
+            ((x >>  8) & 0x0000ff00 ) |
+            ((x >> 24) & 0x000000ff );
+}
+#endif
+
+
+/* ***************************
+*  Memory reads
+*****************************/
+
+/*!
+ * @internal
+ * @brief Enum to indicate whether a pointer is aligned.
+ */
+typedef enum {
+    XXH_aligned,  /*!< Aligned */
+    XXH_unaligned /*!< Possibly unaligned */
+} XXH_alignment;
+
+/*
+ * XXH_FORCE_MEMORY_ACCESS==3 is an endian-independent byteshift load.
+ *
+ * This is ideal for older compilers which don't inline memcpy.
+ */
+#if (defined(XXH_FORCE_MEMORY_ACCESS) && (XXH_FORCE_MEMORY_ACCESS==3))
+
+XXH_FORCE_INLINE xxh_u32 XXH_readLE32(const void* memPtr)
+{
+    const xxh_u8* bytePtr = (const xxh_u8 *)memPtr;
+    return bytePtr[0]
+         | ((xxh_u32)bytePtr[1] << 8)
+         | ((xxh_u32)bytePtr[2] << 16)
+         | ((xxh_u32)bytePtr[3] << 24);
+}
+
+XXH_FORCE_INLINE xxh_u32 XXH_readBE32(const void* memPtr)
+{
+    const xxh_u8* bytePtr = (const xxh_u8 *)memPtr;
+    return bytePtr[3]
+         | ((xxh_u32)bytePtr[2] << 8)
+         | ((xxh_u32)bytePtr[1] << 16)
+         | ((xxh_u32)bytePtr[0] << 24);
+}
+
+#else
+XXH_FORCE_INLINE xxh_u32 XXH_readLE32(const void* ptr)
+{
+    return XXH_CPU_LITTLE_ENDIAN ? XXH_read32(ptr) : XXH_swap32(XXH_read32(ptr));
+}
+
+static xxh_u32 XXH_readBE32(const void* ptr)
+{
+    return XXH_CPU_LITTLE_ENDIAN ? XXH_swap32(XXH_read32(ptr)) : XXH_read32(ptr);
+}
+#endif
+
+XXH_FORCE_INLINE xxh_u32
+XXH_readLE32_align(const void* ptr, XXH_alignment align)
+{
+    if (align==XXH_unaligned) {
+        return XXH_readLE32(ptr);
+    } else {
+        return XXH_CPU_LITTLE_ENDIAN ? *(const xxh_u32*)ptr : XXH_swap32(*(const xxh_u32*)ptr);
+    }
+}
+
+
+/* *************************************
+*  Misc
+***************************************/
+/*! @ingroup public */
+XXH_PUBLIC_API unsigned XXH_versionNumber (void) { return XXH_VERSION_NUMBER; }
+
+
+/* *******************************************************************
+*  32-bit hash functions
+*********************************************************************/
+/*!
+ * @}
+ * @defgroup xxh32_impl XXH32 implementation
+ * @ingroup impl
+ * @{
+ */
+ /* #define instead of static const, to be used as initializers */
+#define XXH_PRIME32_1  0x9E3779B1U  /*!< 0b10011110001101110111100110110001 */
+#define XXH_PRIME32_2  0x85EBCA77U  /*!< 0b10000101111010111100101001110111 */
+#define XXH_PRIME32_3  0xC2B2AE3DU  /*!< 0b11000010101100101010111000111101 */
+#define XXH_PRIME32_4  0x27D4EB2FU  /*!< 0b00100111110101001110101100101111 */
+#define XXH_PRIME32_5  0x165667B1U  /*!< 0b00010110010101100110011110110001 */
+
+#ifdef XXH_OLD_NAMES
+#  define PRIME32_1 XXH_PRIME32_1
+#  define PRIME32_2 XXH_PRIME32_2
+#  define PRIME32_3 XXH_PRIME32_3
+#  define PRIME32_4 XXH_PRIME32_4
+#  define PRIME32_5 XXH_PRIME32_5
+#endif
+
+/*!
+ * @internal
+ * @brief Normal stripe processing routine.
+ *
+ * This shuffles the bits so that any bit from @p input impacts several bits in
+ * @p acc.
+ *
+ * @param acc The accumulator lane.
+ * @param input The stripe of input to mix.
+ * @return The mixed accumulator lane.
+ */
+static xxh_u32 XXH32_round(xxh_u32 acc, xxh_u32 input)
+{
+    acc += input * XXH_PRIME32_2;
+    acc  = XXH_rotl32(acc, 13);
+    acc *= XXH_PRIME32_1;
+#if (defined(__SSE4_1__) || defined(__aarch64__)) && !defined(XXH_ENABLE_AUTOVECTORIZE)
+    /*
+     * UGLY HACK:
+     * A compiler fence is the only thing that prevents GCC and Clang from
+     * autovectorizing the XXH32 loop (pragmas and attributes don't work for some
+     * reason) without globally disabling SSE4.1.
+     *
+     * The reason we want to avoid vectorization is because despite working on
+     * 4 integers at a time, there are multiple factors slowing XXH32 down on
+     * SSE4:
+     * - There's a ridiculous amount of lag from pmulld (10 cycles of latency on
+     *   newer chips!) making it slightly slower to multiply four integers at
+     *   once compared to four integers independently. Even when pmulld was
+     *   fastest, Sandy/Ivy Bridge, it is still not worth it to go into SSE
+     *   just to multiply unless doing a long operation.
+     *
+     * - Four instructions are required to rotate,
+     *      movqda tmp,  v // not required with VEX encoding
+     *      pslld  tmp, 13 // tmp <<= 13
+     *      psrld  v,   19 // x >>= 19
+     *      por    v,  tmp // x |= tmp
+     *   compared to one for scalar:
+     *      roll   v, 13    // reliably fast across the board
+     *      shldl  v, v, 13 // Sandy Bridge and later prefer this for some reason
+     *
+     * - Instruction level parallelism is actually more beneficial here because
+     *   the SIMD actually serializes this operation: While v1 is rotating, v2
+     *   can load data, while v3 can multiply. SSE forces them to operate
+     *   together.
+     *
+     * This is also enabled on AArch64, as Clang autovectorizes it incorrectly
+     * and it is pointless writing a NEON implementation that is basically the
+     * same speed as scalar for XXH32.
+     */
+    XXH_COMPILER_GUARD(acc);
+#endif
+    return acc;
+}
+
+/*!
+ * @internal
+ * @brief Mixes all bits to finalize the hash.
+ *
+ * The final mix ensures that all input bits have a chance to impact any bit in
+ * the output digest, resulting in an unbiased distribution.
+ *
+ * @param h32 The hash to avalanche.
+ * @return The avalanched hash.
+ */
+static xxh_u32 XXH32_avalanche(xxh_u32 h32)
+{
+    h32 ^= h32 >> 15;
+    h32 *= XXH_PRIME32_2;
+    h32 ^= h32 >> 13;
+    h32 *= XXH_PRIME32_3;
+    h32 ^= h32 >> 16;
+    return(h32);
+}
+
+#define XXH_get32bits(p) XXH_readLE32_align(p, align)
+
+/*!
+ * @internal
+ * @brief Processes the last 0-15 bytes of @p ptr.
+ *
+ * There may be up to 15 bytes remaining to consume from the input.
+ * This final stage will digest them to ensure that all input bytes are present
+ * in the final mix.
+ *
+ * @param h32 The hash to finalize.
+ * @param ptr The pointer to the remaining input.
+ * @param len The remaining length, modulo 16.
+ * @param align Whether @p ptr is aligned.
+ * @return The finalized hash.
+ */
+static xxh_u32
+XXH32_finalize(xxh_u32 h32, const xxh_u8* ptr, size_t len, XXH_alignment align)
+{
+#define XXH_PROCESS1 do {                           \
+    h32 += (*ptr++) * XXH_PRIME32_5;                \
+    h32 = XXH_rotl32(h32, 11) * XXH_PRIME32_1;      \
+} while (0)
+
+#define XXH_PROCESS4 do {                           \
+    h32 += XXH_get32bits(ptr) * XXH_PRIME32_3;      \
+    ptr += 4;                                   \
+    h32  = XXH_rotl32(h32, 17) * XXH_PRIME32_4;     \
+} while (0)
+
+    if (ptr==NULL) XXH_ASSERT(len == 0);
+
+    /* Compact rerolled version; generally faster */
+    if (!XXH32_ENDJMP) {
+        len &= 15;
+        while (len >= 4) {
+            XXH_PROCESS4;
+            len -= 4;
+        }
+        while (len > 0) {
+            XXH_PROCESS1;
+            --len;
+        }
+        return XXH32_avalanche(h32);
+    } else {
+         switch(len&15) /* or switch(bEnd - p) */ {
+           case 12:      XXH_PROCESS4;
+                         XXH_FALLTHROUGH;
+           case 8:       XXH_PROCESS4;
+                         XXH_FALLTHROUGH;
+           case 4:       XXH_PROCESS4;
+                         return XXH32_avalanche(h32);
+
+           case 13:      XXH_PROCESS4;
+                         XXH_FALLTHROUGH;
+           case 9:       XXH_PROCESS4;
+                         XXH_FALLTHROUGH;
+           case 5:       XXH_PROCESS4;
+                         XXH_PROCESS1;
+                         return XXH32_avalanche(h32);
+
+           case 14:      XXH_PROCESS4;
+                         XXH_FALLTHROUGH;
+           case 10:      XXH_PROCESS4;
+                         XXH_FALLTHROUGH;
+           case 6:       XXH_PROCESS4;
+                         XXH_PROCESS1;
+                         XXH_PROCESS1;
+                         return XXH32_avalanche(h32);
+
+           case 15:      XXH_PROCESS4;
+                         XXH_FALLTHROUGH;
+           case 11:      XXH_PROCESS4;
+                         XXH_FALLTHROUGH;
+           case 7:       XXH_PROCESS4;
+                         XXH_FALLTHROUGH;
+           case 3:       XXH_PROCESS1;
+                         XXH_FALLTHROUGH;
+           case 2:       XXH_PROCESS1;
+                         XXH_FALLTHROUGH;
+           case 1:       XXH_PROCESS1;
+                         XXH_FALLTHROUGH;
+           case 0:       return XXH32_avalanche(h32);
+        }
+        XXH_ASSERT(0);
+        return h32;   /* reaching this point is deemed impossible */
+    }
+}
+
+#ifdef XXH_OLD_NAMES
+#  define PROCESS1 XXH_PROCESS1
+#  define PROCESS4 XXH_PROCESS4
+#else
+#  undef XXH_PROCESS1
+#  undef XXH_PROCESS4
+#endif
+
+/*!
+ * @internal
+ * @brief The implementation for @ref XXH32().
+ *
+ * @param input , len , seed Directly passed from @ref XXH32().
+ * @param align Whether @p input is aligned.
+ * @return The calculated hash.
+ */
+XXH_FORCE_INLINE xxh_u32
+XXH32_endian_align(const xxh_u8* input, size_t len, xxh_u32 seed, XXH_alignment align)
+{
+    xxh_u32 h32;
+
+    if (input==NULL) XXH_ASSERT(len == 0);
+
+    if (len>=16) {
+        const xxh_u8* const bEnd = input + len;
+        const xxh_u8* const limit = bEnd - 15;
+        xxh_u32 v1 = seed + XXH_PRIME32_1 + XXH_PRIME32_2;
+        xxh_u32 v2 = seed + XXH_PRIME32_2;
+        xxh_u32 v3 = seed + 0;
+        xxh_u32 v4 = seed - XXH_PRIME32_1;
+
+        do {
+            v1 = XXH32_round(v1, XXH_get32bits(input)); input += 4;
+            v2 = XXH32_round(v2, XXH_get32bits(input)); input += 4;
+            v3 = XXH32_round(v3, XXH_get32bits(input)); input += 4;
+            v4 = XXH32_round(v4, XXH_get32bits(input)); input += 4;
+        } while (input < limit);
+
+        h32 = XXH_rotl32(v1, 1)  + XXH_rotl32(v2, 7)
+            + XXH_rotl32(v3, 12) + XXH_rotl32(v4, 18);
+    } else {
+        h32  = seed + XXH_PRIME32_5;
+    }
+
+    h32 += (xxh_u32)len;
+
+    return XXH32_finalize(h32, input, len&15, align);
+}
+
+/*! @ingroup xxh32_family */
+XXH_PUBLIC_API XXH32_hash_t XXH32 (const void* input, size_t len, XXH32_hash_t seed)
+{
+#if 0
+    /* Simple version, good for code maintenance, but unfortunately slow for small inputs */
+    XXH32_state_t state;
+    XXH32_reset(&state, seed);
+    XXH32_update(&state, (const xxh_u8*)input, len);
+    return XXH32_digest(&state);
+#else
+    if (XXH_FORCE_ALIGN_CHECK) {
+        if ((((size_t)input) & 3) == 0) {   /* Input is 4-bytes aligned, leverage the speed benefit */
+            return XXH32_endian_align((const xxh_u8*)input, len, seed, XXH_aligned);
+    }   }
+
+    return XXH32_endian_align((const xxh_u8*)input, len, seed, XXH_unaligned);
+#endif
+}
+
+
+
+/*******   Hash streaming   *******/
+/*!
+ * @ingroup xxh32_family
+ */
+XXH_PUBLIC_API XXH32_state_t* XXH32_createState(void)
+{
+    return (XXH32_state_t*)XXH_malloc(sizeof(XXH32_state_t));
+}
+/*! @ingroup xxh32_family */
+XXH_PUBLIC_API XXH_errorcode XXH32_freeState(XXH32_state_t* statePtr)
+{
+    XXH_free(statePtr);
+    return XXH_OK;
+}
+
+/*! @ingroup xxh32_family */
+XXH_PUBLIC_API void XXH32_copyState(XXH32_state_t* dstState, const XXH32_state_t* srcState)
+{
+    XXH_memcpy(dstState, srcState, sizeof(*dstState));
+}
+
+/*! @ingroup xxh32_family */
+XXH_PUBLIC_API XXH_errorcode XXH32_reset(XXH32_state_t* statePtr, XXH32_hash_t seed)
+{
+    XXH32_state_t state;   /* using a local state to memcpy() in order to avoid strict-aliasing warnings */
+    memset(&state, 0, sizeof(state));
+    state.v[0] = seed + XXH_PRIME32_1 + XXH_PRIME32_2;
+    state.v[1] = seed + XXH_PRIME32_2;
+    state.v[2] = seed + 0;
+    state.v[3] = seed - XXH_PRIME32_1;
+    /* do not write into reserved, planned to be removed in a future version */
+    XXH_memcpy(statePtr, &state, sizeof(state) - sizeof(state.reserved));
+    return XXH_OK;
+}
+
+
+/*! @ingroup xxh32_family */
+XXH_PUBLIC_API XXH_errorcode
+XXH32_update(XXH32_state_t* state, const void* input, size_t len)
+{
+    if (input==NULL) {
+        XXH_ASSERT(len == 0);
+        return XXH_OK;
+    }
+
+    {   const xxh_u8* p = (const xxh_u8*)input;
+        const xxh_u8* const bEnd = p + len;
+
+        state->total_len_32 += (XXH32_hash_t)len;
+        state->large_len |= (XXH32_hash_t)((len>=16) | (state->total_len_32>=16));
+
+        if (state->memsize + len < 16)  {   /* fill in tmp buffer */
+            XXH_memcpy((xxh_u8*)(state->mem32) + state->memsize, input, len);
+            state->memsize += (XXH32_hash_t)len;
+            return XXH_OK;
+        }
+
+        if (state->memsize) {   /* some data left from previous update */
+            XXH_memcpy((xxh_u8*)(state->mem32) + state->memsize, input, 16-state->memsize);
+            {   const xxh_u32* p32 = state->mem32;
+                state->v[0] = XXH32_round(state->v[0], XXH_readLE32(p32)); p32++;
+                state->v[1] = XXH32_round(state->v[1], XXH_readLE32(p32)); p32++;
+                state->v[2] = XXH32_round(state->v[2], XXH_readLE32(p32)); p32++;
+                state->v[3] = XXH32_round(state->v[3], XXH_readLE32(p32));
+            }
+            p += 16-state->memsize;
+            state->memsize = 0;
+        }
+
+        if (p <= bEnd-16) {
+            const xxh_u8* const limit = bEnd - 16;
+
+            do {
+                state->v[0] = XXH32_round(state->v[0], XXH_readLE32(p)); p+=4;
+                state->v[1] = XXH32_round(state->v[1], XXH_readLE32(p)); p+=4;
+                state->v[2] = XXH32_round(state->v[2], XXH_readLE32(p)); p+=4;
+                state->v[3] = XXH32_round(state->v[3], XXH_readLE32(p)); p+=4;
+            } while (p<=limit);
+
+        }
+
+        if (p < bEnd) {
+            XXH_memcpy(state->mem32, p, (size_t)(bEnd-p));
+            state->memsize = (unsigned)(bEnd-p);
+        }
+    }
+
+    return XXH_OK;
+}
+
+
+/*! @ingroup xxh32_family */
+XXH_PUBLIC_API XXH32_hash_t XXH32_digest(const XXH32_state_t* state)
+{
+    xxh_u32 h32;
+
+    if (state->large_len) {
+        h32 = XXH_rotl32(state->v[0], 1)
+            + XXH_rotl32(state->v[1], 7)
+            + XXH_rotl32(state->v[2], 12)
+            + XXH_rotl32(state->v[3], 18);
+    } else {
+        h32 = state->v[2] /* == seed */ + XXH_PRIME32_5;
+    }
+
+    h32 += state->total_len_32;
+
+    return XXH32_finalize(h32, (const xxh_u8*)state->mem32, state->memsize, XXH_aligned);
+}
+
+
+/*******   Canonical representation   *******/
+
+/*!
+ * @ingroup xxh32_family
+ * The default return values from XXH functions are unsigned 32 and 64 bit
+ * integers.
+ *
+ * The canonical representation uses big endian convention, the same convention
+ * as human-readable numbers (large digits first).
+ *
+ * This way, hash values can be written into a file or buffer, remaining
+ * comparable across different systems.
+ *
+ * The following functions allow transformation of hash values to and from their
+ * canonical format.
+ */
+XXH_PUBLIC_API void XXH32_canonicalFromHash(XXH32_canonical_t* dst, XXH32_hash_t hash)
+{
+    XXH_STATIC_ASSERT(sizeof(XXH32_canonical_t) == sizeof(XXH32_hash_t));
+    if (XXH_CPU_LITTLE_ENDIAN) hash = XXH_swap32(hash);
+    XXH_memcpy(dst, &hash, sizeof(*dst));
+}
+/*! @ingroup xxh32_family */
+XXH_PUBLIC_API XXH32_hash_t XXH32_hashFromCanonical(const XXH32_canonical_t* src)
+{
+    return XXH_readBE32(src);
+}
+
+
+#ifndef XXH_NO_LONG_LONG
+
+/* *******************************************************************
+*  64-bit hash functions
+*********************************************************************/
+/*!
+ * @}
+ * @ingroup impl
+ * @{
+ */
+/*******   Memory access   *******/
+
+typedef XXH64_hash_t xxh_u64;
+
+#ifdef XXH_OLD_NAMES
+#  define U64 xxh_u64
+#endif
+
+#if (defined(XXH_FORCE_MEMORY_ACCESS) && (XXH_FORCE_MEMORY_ACCESS==3))
+/*
+ * Manual byteshift. Best for old compilers which don't inline memcpy.
+ * We actually directly use XXH_readLE64 and XXH_readBE64.
+ */
+#elif (defined(XXH_FORCE_MEMORY_ACCESS) && (XXH_FORCE_MEMORY_ACCESS==2))
+
+/* Force direct memory access. Only works on CPU which support unaligned memory access in hardware */
+static xxh_u64 XXH_read64(const void* memPtr)
+{
+    return *(const xxh_u64*) memPtr;
+}
+
+#elif (defined(XXH_FORCE_MEMORY_ACCESS) && (XXH_FORCE_MEMORY_ACCESS==1))
+
+/*
+ * __pack instructions are safer, but compiler specific, hence potentially
+ * problematic for some compilers.
+ *
+ * Currently only defined for GCC and ICC.
+ */
+#ifdef XXH_OLD_NAMES
+typedef union { xxh_u32 u32; xxh_u64 u64; } __attribute__((packed)) unalign64;
+#endif
+static xxh_u64 XXH_read64(const void* ptr)
+{
+    typedef union { xxh_u32 u32; xxh_u64 u64; } __attribute__((packed)) xxh_unalign64;
+    return ((const xxh_unalign64*)ptr)->u64;
+}
+
+#else
+
+/*
+ * Portable and safe solution. Generally efficient.
+ * see: http://fastcompression.blogspot.com/2015/08/accessing-unaligned-memory.html
+ */
+static xxh_u64 XXH_read64(const void* memPtr)
+{
+    xxh_u64 val;
+    XXH_memcpy(&val, memPtr, sizeof(val));
+    return val;
+}
+
+#endif   /* XXH_FORCE_DIRECT_MEMORY_ACCESS */
+
+#if defined(_MSC_VER)     /* Visual Studio */
+#  define XXH_swap64 _byteswap_uint64
+#elif XXH_GCC_VERSION >= 403
+#  define XXH_swap64 __builtin_bswap64
+#else
+static xxh_u64 XXH_swap64(xxh_u64 x)
+{
+    return  ((x << 56) & 0xff00000000000000ULL) |
+            ((x << 40) & 0x00ff000000000000ULL) |
+            ((x << 24) & 0x0000ff0000000000ULL) |
+            ((x << 8)  & 0x000000ff00000000ULL) |
+            ((x >> 8)  & 0x00000000ff000000ULL) |
+            ((x >> 24) & 0x0000000000ff0000ULL) |
+            ((x >> 40) & 0x000000000000ff00ULL) |
+            ((x >> 56) & 0x00000000000000ffULL);
+}
+#endif
+
+
+/* XXH_FORCE_MEMORY_ACCESS==3 is an endian-independent byteshift load. */
+#if (defined(XXH_FORCE_MEMORY_ACCESS) && (XXH_FORCE_MEMORY_ACCESS==3))
+
+XXH_FORCE_INLINE xxh_u64 XXH_readLE64(const void* memPtr)
+{
+    const xxh_u8* bytePtr = (const xxh_u8 *)memPtr;
+    return bytePtr[0]
+         | ((xxh_u64)bytePtr[1] << 8)
+         | ((xxh_u64)bytePtr[2] << 16)
+         | ((xxh_u64)bytePtr[3] << 24)
+         | ((xxh_u64)bytePtr[4] << 32)
+         | ((xxh_u64)bytePtr[5] << 40)
+         | ((xxh_u64)bytePtr[6] << 48)
+         | ((xxh_u64)bytePtr[7] << 56);
+}
+
+XXH_FORCE_INLINE xxh_u64 XXH_readBE64(const void* memPtr)
+{
+    const xxh_u8* bytePtr = (const xxh_u8 *)memPtr;
+    return bytePtr[7]
+         | ((xxh_u64)bytePtr[6] << 8)
+         | ((xxh_u64)bytePtr[5] << 16)
+         | ((xxh_u64)bytePtr[4] << 24)
+         | ((xxh_u64)bytePtr[3] << 32)
+         | ((xxh_u64)bytePtr[2] << 40)
+         | ((xxh_u64)bytePtr[1] << 48)
+         | ((xxh_u64)bytePtr[0] << 56);
+}
+
+#else
+XXH_FORCE_INLINE xxh_u64 XXH_readLE64(const void* ptr)
+{
+    return XXH_CPU_LITTLE_ENDIAN ? XXH_read64(ptr) : XXH_swap64(XXH_read64(ptr));
+}
+
+static xxh_u64 XXH_readBE64(const void* ptr)
+{
+    return XXH_CPU_LITTLE_ENDIAN ? XXH_swap64(XXH_read64(ptr)) : XXH_read64(ptr);
+}
+#endif
+
+XXH_FORCE_INLINE xxh_u64
+XXH_readLE64_align(const void* ptr, XXH_alignment align)
+{
+    if (align==XXH_unaligned)
+        return XXH_readLE64(ptr);
+    else
+        return XXH_CPU_LITTLE_ENDIAN ? *(const xxh_u64*)ptr : XXH_swap64(*(const xxh_u64*)ptr);
+}
+
+
+/*******   xxh64   *******/
+/*!
+ * @}
+ * @defgroup xxh64_impl XXH64 implementation
+ * @ingroup impl
+ * @{
+ */
+/* #define rather that static const, to be used as initializers */
+#define XXH_PRIME64_1  0x9E3779B185EBCA87ULL  /*!< 0b1001111000110111011110011011000110000101111010111100101010000111 */
+#define XXH_PRIME64_2  0xC2B2AE3D27D4EB4FULL  /*!< 0b1100001010110010101011100011110100100111110101001110101101001111 */
+#define XXH_PRIME64_3  0x165667B19E3779F9ULL  /*!< 0b0001011001010110011001111011000110011110001101110111100111111001 */
+#define XXH_PRIME64_4  0x85EBCA77C2B2AE63ULL  /*!< 0b1000010111101011110010100111011111000010101100101010111001100011 */
+#define XXH_PRIME64_5  0x27D4EB2F165667C5ULL  /*!< 0b0010011111010100111010110010111100010110010101100110011111000101 */
+
+#ifdef XXH_OLD_NAMES
+#  define PRIME64_1 XXH_PRIME64_1
+#  define PRIME64_2 XXH_PRIME64_2
+#  define PRIME64_3 XXH_PRIME64_3
+#  define PRIME64_4 XXH_PRIME64_4
+#  define PRIME64_5 XXH_PRIME64_5
+#endif
+
+static xxh_u64 XXH64_round(xxh_u64 acc, xxh_u64 input)
+{
+    acc += input * XXH_PRIME64_2;
+    acc  = XXH_rotl64(acc, 31);
+    acc *= XXH_PRIME64_1;
+    return acc;
+}
+
+static xxh_u64 XXH64_mergeRound(xxh_u64 acc, xxh_u64 val)
+{
+    val  = XXH64_round(0, val);
+    acc ^= val;
+    acc  = acc * XXH_PRIME64_1 + XXH_PRIME64_4;
+    return acc;
+}
+
+static xxh_u64 XXH64_avalanche(xxh_u64 h64)
+{
+    h64 ^= h64 >> 33;
+    h64 *= XXH_PRIME64_2;
+    h64 ^= h64 >> 29;
+    h64 *= XXH_PRIME64_3;
+    h64 ^= h64 >> 32;
+    return h64;
+}
+
+
+#define XXH_get64bits(p) XXH_readLE64_align(p, align)
+
+static xxh_u64
+XXH64_finalize(xxh_u64 h64, const xxh_u8* ptr, size_t len, XXH_alignment align)
+{
+    if (ptr==NULL) XXH_ASSERT(len == 0);
+    len &= 31;
+    while (len >= 8) {
+        xxh_u64 const k1 = XXH64_round(0, XXH_get64bits(ptr));
+        ptr += 8;
+        h64 ^= k1;
+        h64  = XXH_rotl64(h64,27) * XXH_PRIME64_1 + XXH_PRIME64_4;
+        len -= 8;
+    }
+    if (len >= 4) {
+        h64 ^= (xxh_u64)(XXH_get32bits(ptr)) * XXH_PRIME64_1;
+        ptr += 4;
+        h64 = XXH_rotl64(h64, 23) * XXH_PRIME64_2 + XXH_PRIME64_3;
+        len -= 4;
+    }
+    while (len > 0) {
+        h64 ^= (*ptr++) * XXH_PRIME64_5;
+        h64 = XXH_rotl64(h64, 11) * XXH_PRIME64_1;
+        --len;
+    }
+    return  XXH64_avalanche(h64);
+}
+
+#ifdef XXH_OLD_NAMES
+#  define PROCESS1_64 XXH_PROCESS1_64
+#  define PROCESS4_64 XXH_PROCESS4_64
+#  define PROCESS8_64 XXH_PROCESS8_64
+#else
+#  undef XXH_PROCESS1_64
+#  undef XXH_PROCESS4_64
+#  undef XXH_PROCESS8_64
+#endif
+
+XXH_FORCE_INLINE xxh_u64
+XXH64_endian_align(const xxh_u8* input, size_t len, xxh_u64 seed, XXH_alignment align)
+{
+    xxh_u64 h64;
+    if (input==NULL) XXH_ASSERT(len == 0);
+
+    if (len>=32) {
+        const xxh_u8* const bEnd = input + len;
+        const xxh_u8* const limit = bEnd - 31;
+        xxh_u64 v1 = seed + XXH_PRIME64_1 + XXH_PRIME64_2;
+        xxh_u64 v2 = seed + XXH_PRIME64_2;
+        xxh_u64 v3 = seed + 0;
+        xxh_u64 v4 = seed - XXH_PRIME64_1;
+
+        do {
+            v1 = XXH64_round(v1, XXH_get64bits(input)); input+=8;
+            v2 = XXH64_round(v2, XXH_get64bits(input)); input+=8;
+            v3 = XXH64_round(v3, XXH_get64bits(input)); input+=8;
+            v4 = XXH64_round(v4, XXH_get64bits(input)); input+=8;
+        } while (input<limit);
+
+        h64 = XXH_rotl64(v1, 1) + XXH_rotl64(v2, 7) + XXH_rotl64(v3, 12) + XXH_rotl64(v4, 18);
+        h64 = XXH64_mergeRound(h64, v1);
+        h64 = XXH64_mergeRound(h64, v2);
+        h64 = XXH64_mergeRound(h64, v3);
+        h64 = XXH64_mergeRound(h64, v4);
+
+    } else {
+        h64  = seed + XXH_PRIME64_5;
+    }
+
+    h64 += (xxh_u64) len;
+
+    return XXH64_finalize(h64, input, len, align);
+}
+
+
+/*! @ingroup xxh64_family */
+XXH_PUBLIC_API XXH64_hash_t XXH64 (const void* input, size_t len, XXH64_hash_t seed)
+{
+#if 0
+    /* Simple version, good for code maintenance, but unfortunately slow for small inputs */
+    XXH64_state_t state;
+    XXH64_reset(&state, seed);
+    XXH64_update(&state, (const xxh_u8*)input, len);
+    return XXH64_digest(&state);
+#else
+    if (XXH_FORCE_ALIGN_CHECK) {
+        if ((((size_t)input) & 7)==0) {  /* Input is aligned, let's leverage the speed advantage */
+            return XXH64_endian_align((const xxh_u8*)input, len, seed, XXH_aligned);
+    }   }
+
+    return XXH64_endian_align((const xxh_u8*)input, len, seed, XXH_unaligned);
+
+#endif
+}
+
+/*******   Hash Streaming   *******/
+
+/*! @ingroup xxh64_family*/
+XXH_PUBLIC_API XXH64_state_t* XXH64_createState(void)
+{
+    return (XXH64_state_t*)XXH_malloc(sizeof(XXH64_state_t));
+}
+/*! @ingroup xxh64_family */
+XXH_PUBLIC_API XXH_errorcode XXH64_freeState(XXH64_state_t* statePtr)
+{
+    XXH_free(statePtr);
+    return XXH_OK;
+}
+
+/*! @ingroup xxh64_family */
+XXH_PUBLIC_API void XXH64_copyState(XXH64_state_t* dstState, const XXH64_state_t* srcState)
+{
+    XXH_memcpy(dstState, srcState, sizeof(*dstState));
+}
+
+/*! @ingroup xxh64_family */
+XXH_PUBLIC_API XXH_errorcode XXH64_reset(XXH64_state_t* statePtr, XXH64_hash_t seed)
+{
+    XXH64_state_t state;   /* use a local state to memcpy() in order to avoid strict-aliasing warnings */
+    memset(&state, 0, sizeof(state));
+    state.v[0] = seed + XXH_PRIME64_1 + XXH_PRIME64_2;
+    state.v[1] = seed + XXH_PRIME64_2;
+    state.v[2] = seed + 0;
+    state.v[3] = seed - XXH_PRIME64_1;
+     /* do not write into reserved64, might be removed in a future version */
+    XXH_memcpy(statePtr, &state, sizeof(state) - sizeof(state.reserved64));
+    return XXH_OK;
+}
+
+/*! @ingroup xxh64_family */
+XXH_PUBLIC_API XXH_errorcode
+XXH64_update (XXH64_state_t* state, const void* input, size_t len)
+{
+    if (input==NULL) {
+        XXH_ASSERT(len == 0);
+        return XXH_OK;
+    }
+
+    {   const xxh_u8* p = (const xxh_u8*)input;
+        const xxh_u8* const bEnd = p + len;
+
+        state->total_len += len;
+
+        if (state->memsize + len < 32) {  /* fill in tmp buffer */
+            XXH_memcpy(((xxh_u8*)state->mem64) + state->memsize, input, len);
+            state->memsize += (xxh_u32)len;
+            return XXH_OK;
+        }
+
+        if (state->memsize) {   /* tmp buffer is full */
+            XXH_memcpy(((xxh_u8*)state->mem64) + state->memsize, input, 32-state->memsize);
+            state->v[0] = XXH64_round(state->v[0], XXH_readLE64(state->mem64+0));
+            state->v[1] = XXH64_round(state->v[1], XXH_readLE64(state->mem64+1));
+            state->v[2] = XXH64_round(state->v[2], XXH_readLE64(state->mem64+2));
+            state->v[3] = XXH64_round(state->v[3], XXH_readLE64(state->mem64+3));
+            p += 32 - state->memsize;
+            state->memsize = 0;
+        }
+
+        if (p+32 <= bEnd) {
+            const xxh_u8* const limit = bEnd - 32;
+
+            do {
+                state->v[0] = XXH64_round(state->v[0], XXH_readLE64(p)); p+=8;
+                state->v[1] = XXH64_round(state->v[1], XXH_readLE64(p)); p+=8;
+                state->v[2] = XXH64_round(state->v[2], XXH_readLE64(p)); p+=8;
+                state->v[3] = XXH64_round(state->v[3], XXH_readLE64(p)); p+=8;
+            } while (p<=limit);
+
+        }
+
+        if (p < bEnd) {
+            XXH_memcpy(state->mem64, p, (size_t)(bEnd-p));
+            state->memsize = (unsigned)(bEnd-p);
+        }
+    }
+
+    return XXH_OK;
+}
+
+
+/*! @ingroup xxh64_family */
+XXH_PUBLIC_API XXH64_hash_t XXH64_digest(const XXH64_state_t* state)
+{
+    xxh_u64 h64;
+
+    if (state->total_len >= 32) {
+        h64 = XXH_rotl64(state->v[0], 1) + XXH_rotl64(state->v[1], 7) + XXH_rotl64(state->v[2], 12) + XXH_rotl64(state->v[3], 18);
+        h64 = XXH64_mergeRound(h64, state->v[0]);
+        h64 = XXH64_mergeRound(h64, state->v[1]);
+        h64 = XXH64_mergeRound(h64, state->v[2]);
+        h64 = XXH64_mergeRound(h64, state->v[3]);
+    } else {
+        h64  = state->v[2] /*seed*/ + XXH_PRIME64_5;
+    }
+
+    h64 += (xxh_u64) state->total_len;
+
+    return XXH64_finalize(h64, (const xxh_u8*)state->mem64, (size_t)state->total_len, XXH_aligned);
+}
+
+
+/******* Canonical representation   *******/
+
+/*! @ingroup xxh64_family */
+XXH_PUBLIC_API void XXH64_canonicalFromHash(XXH64_canonical_t* dst, XXH64_hash_t hash)
+{
+    XXH_STATIC_ASSERT(sizeof(XXH64_canonical_t) == sizeof(XXH64_hash_t));
+    if (XXH_CPU_LITTLE_ENDIAN) hash = XXH_swap64(hash);
+    XXH_memcpy(dst, &hash, sizeof(*dst));
+}
+
+/*! @ingroup xxh64_family */
+XXH_PUBLIC_API XXH64_hash_t XXH64_hashFromCanonical(const XXH64_canonical_t* src)
+{
+    return XXH_readBE64(src);
+}
+
+#ifndef XXH_NO_XXH3
+
+/* *********************************************************************
+*  XXH3
+*  New generation hash designed for speed on small keys and vectorization
+************************************************************************ */
+/*!
+ * @}
+ * @defgroup xxh3_impl XXH3 implementation
+ * @ingroup impl
+ * @{
+ */
+
+/* ===   Compiler specifics   === */
+
+#if ((defined(sun) || defined(__sun)) && __cplusplus) /* Solaris includes __STDC_VERSION__ with C++. Tested with GCC 5.5 */
+#  define XXH_RESTRICT /* disable */
+#elif defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L   /* >= C99 */
+#  define XXH_RESTRICT   restrict
+#else
+/* Note: it might be useful to define __restrict or __restrict__ for some C++ compilers */
+#  define XXH_RESTRICT   /* disable */
+#endif
+
+#if (defined(__GNUC__) && (__GNUC__ >= 3))  \
+  || (defined(__INTEL_COMPILER) && (__INTEL_COMPILER >= 800)) \
+  || defined(__clang__)
+#    define XXH_likely(x) __builtin_expect(x, 1)
+#    define XXH_unlikely(x) __builtin_expect(x, 0)
+#else
+#    define XXH_likely(x) (x)
+#    define XXH_unlikely(x) (x)
+#endif
+
+#if defined(__GNUC__)
+#  if defined(__AVX2__)
+#    include <immintrin.h>
+#  elif defined(__SSE2__)
+#    include <emmintrin.h>
+#  elif defined(__ARM_NEON__) || defined(__ARM_NEON)
+#    define inline __inline__  /* circumvent a clang bug */
+#    include <arm_neon.h>
+#    undef inline
+#  endif
+#elif defined(_MSC_VER)
+#  include <intrin.h>
+#endif
+
+/*
+ * One goal of XXH3 is to make it fast on both 32-bit and 64-bit, while
+ * remaining a true 64-bit/128-bit hash function.
+ *
+ * This is done by prioritizing a subset of 64-bit operations that can be
+ * emulated without too many steps on the average 32-bit machine.
+ *
+ * For example, these two lines seem similar, and run equally fast on 64-bit:
+ *
+ *   xxh_u64 x;
+ *   x ^= (x >> 47); // good
+ *   x ^= (x >> 13); // bad
+ *
+ * However, to a 32-bit machine, there is a major difference.
+ *
+ * x ^= (x >> 47) looks like this:
+ *
+ *   x.lo ^= (x.hi >> (47 - 32));
+ *
+ * while x ^= (x >> 13) looks like this:
+ *
+ *   // note: funnel shifts are not usually cheap.
+ *   x.lo ^= (x.lo >> 13) | (x.hi << (32 - 13));
+ *   x.hi ^= (x.hi >> 13);
+ *
+ * The first one is significantly faster than the second, simply because the
+ * shift is larger than 32. This means:
+ *  - All the bits we need are in the upper 32 bits, so we can ignore the lower
+ *    32 bits in the shift.
+ *  - The shift result will always fit in the lower 32 bits, and therefore,
+ *    we can ignore the upper 32 bits in the xor.
+ *
+ * Thanks to this optimization, XXH3 only requires these features to be efficient:
+ *
+ *  - Usable unaligned access
+ *  - A 32-bit or 64-bit ALU
+ *      - If 32-bit, a decent ADC instruction
+ *  - A 32 or 64-bit multiply with a 64-bit result
+ *  - For the 128-bit variant, a decent byteswap helps short inputs.
+ *
+ * The first two are already required by XXH32, and almost all 32-bit and 64-bit
+ * platforms which can run XXH32 can run XXH3 efficiently.
+ *
+ * Thumb-1, the classic 16-bit only subset of ARM's instruction set, is one
+ * notable exception.
+ *
+ * First of all, Thumb-1 lacks support for the UMULL instruction which
+ * performs the important long multiply. This means numerous __aeabi_lmul
+ * calls.
+ *
+ * Second of all, the 8 functional registers are just not enough.
+ * Setup for __aeabi_lmul, byteshift loads, pointers, and all arithmetic need
+ * Lo registers, and this shuffling results in thousands more MOVs than A32.
+ *
+ * A32 and T32 don't have this limitation. They can access all 14 registers,
+ * do a 32->64 multiply with UMULL, and the flexible operand allowing free
+ * shifts is helpful, too.
+ *
+ * Therefore, we do a quick sanity check.
+ *
+ * If compiling Thumb-1 for a target which supports ARM instructions, we will
+ * emit a warning, as it is not a "sane" platform to compile for.
+ *
+ * Usually, if this happens, it is because of an accident and you probably need
+ * to specify -march, as you likely meant to compile for a newer architecture.
+ *
+ * Credit: large sections of the vectorial and asm source code paths
+ *         have been contributed by @easyaspi314
+ */
+#if defined(__thumb__) && !defined(__thumb2__) && defined(__ARM_ARCH_ISA_ARM)
+#   warning "XXH3 is highly inefficient without ARM or Thumb-2."
+#endif
+
+/* ==========================================
+ * Vectorization detection
+ * ========================================== */
+
+#ifdef XXH_DOXYGEN
+/*!
+ * @ingroup tuning
+ * @brief Overrides the vectorization implementation chosen for XXH3.
+ *
+ * Can be defined to 0 to disable SIMD or any of the values mentioned in
+ * @ref XXH_VECTOR_TYPE.
+ *
+ * If this is not defined, it uses predefined macros to determine the best
+ * implementation.
+ */
+#  define XXH_VECTOR XXH_SCALAR
+/*!
+ * @ingroup tuning
+ * @brief Possible values for @ref XXH_VECTOR.
+ *
+ * Note that these are actually implemented as macros.
+ *
+ * If this is not defined, it is detected automatically.
+ * @ref XXH_X86DISPATCH overrides this.
+ */
+enum XXH_VECTOR_TYPE /* fake enum */ {
+    XXH_SCALAR = 0,  /*!< Portable scalar version */
+    XXH_SSE2   = 1,  /*!<
+                      * SSE2 for Pentium 4, Opteron, all x86_64.
+                      *
+                      * @note SSE2 is also guaranteed on Windows 10, macOS, and
+                      * Android x86.
+                      */
+    XXH_AVX2   = 2,  /*!< AVX2 for Haswell and Bulldozer */
+    XXH_AVX512 = 3,  /*!< AVX512 for Skylake and Icelake */
+    XXH_NEON   = 4,  /*!< NEON for most ARMv7-A and all AArch64 */
+    XXH_VSX    = 5,  /*!< VSX and ZVector for POWER8/z13 (64-bit) */
+};
+/*!
+ * @ingroup tuning
+ * @brief Selects the minimum alignment for XXH3's accumulators.
+ *
+ * When using SIMD, this should match the alignment reqired for said vector
+ * type, so, for example, 32 for AVX2.
+ *
+ * Default: Auto detected.
+ */
+#  define XXH_ACC_ALIGN 8
+#endif
+
+/* Actual definition */
+#ifndef XXH_DOXYGEN
+#  define XXH_SCALAR 0
+#  define XXH_SSE2   1
+#  define XXH_AVX2   2
+#  define XXH_AVX512 3
+#  define XXH_NEON   4
+#  define XXH_VSX    5
+#endif
+
+#ifndef XXH_VECTOR    /* can be defined on command line */
+#  if defined(__AVX512F__)
+#    define XXH_VECTOR XXH_AVX512
+#  elif defined(__AVX2__)
+#    define XXH_VECTOR XXH_AVX2
+#  elif defined(__SSE2__) || defined(_M_AMD64) || defined(_M_X64) || (defined(_M_IX86_FP) && (_M_IX86_FP == 2))
+#    define XXH_VECTOR XXH_SSE2
+#  elif ( \
+        defined(__ARM_NEON__) || defined(__ARM_NEON) /* gcc */ \
+     || defined(_M_ARM64) || defined(_M_ARM_ARMV7VE) /* msvc */ \
+   ) && ( \
+        defined(_WIN32) || defined(__LITTLE_ENDIAN__) /* little endian only */ \
+    || (defined(__BYTE_ORDER__) && __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__) \
+   )
+#    define XXH_VECTOR XXH_NEON
+#  elif (defined(__PPC64__) && defined(__POWER8_VECTOR__)) \
+     || (defined(__s390x__) && defined(__VEC__)) \
+     && defined(__GNUC__) /* TODO: IBM XL */
+#    define XXH_VECTOR XXH_VSX
+#  else
+#    define XXH_VECTOR XXH_SCALAR
+#  endif
+#endif
+
+/*
+ * Controls the alignment of the accumulator,
+ * for compatibility with aligned vector loads, which are usually faster.
+ */
+#ifndef XXH_ACC_ALIGN
+#  if defined(XXH_X86DISPATCH)
+#     define XXH_ACC_ALIGN 64  /* for compatibility with avx512 */
+#  elif XXH_VECTOR == XXH_SCALAR  /* scalar */
+#     define XXH_ACC_ALIGN 8
+#  elif XXH_VECTOR == XXH_SSE2  /* sse2 */
+#     define XXH_ACC_ALIGN 16
+#  elif XXH_VECTOR == XXH_AVX2  /* avx2 */
+#     define XXH_ACC_ALIGN 32
+#  elif XXH_VECTOR == XXH_NEON  /* neon */
+#     define XXH_ACC_ALIGN 16
+#  elif XXH_VECTOR == XXH_VSX   /* vsx */
+#     define XXH_ACC_ALIGN 16
+#  elif XXH_VECTOR == XXH_AVX512  /* avx512 */
+#     define XXH_ACC_ALIGN 64
+#  endif
+#endif
+
+#if defined(XXH_X86DISPATCH) || XXH_VECTOR == XXH_SSE2 \
+    || XXH_VECTOR == XXH_AVX2 || XXH_VECTOR == XXH_AVX512
+#  define XXH_SEC_ALIGN XXH_ACC_ALIGN
+#else
+#  define XXH_SEC_ALIGN 8
+#endif
+
+/*
+ * UGLY HACK:
+ * GCC usually generates the best code with -O3 for xxHash.
+ *
+ * However, when targeting AVX2, it is overzealous in its unrolling resulting
+ * in code roughly 3/4 the speed of Clang.
+ *
+ * There are other issues, such as GCC splitting _mm256_loadu_si256 into
+ * _mm_loadu_si128 + _mm256_inserti128_si256. This is an optimization which
+ * only applies to Sandy and Ivy Bridge... which don't even support AVX2.
+ *
+ * That is why when compiling the AVX2 version, it is recommended to use either
+ *   -O2 -mavx2 -march=haswell
+ * or
+ *   -O2 -mavx2 -mno-avx256-split-unaligned-load
+ * for decent performance, or to use Clang instead.
+ *
+ * Fortunately, we can control the first one with a pragma that forces GCC into
+ * -O2, but the other one we can't control without "failed to inline always
+ * inline function due to target mismatch" warnings.
+ */
+#if XXH_VECTOR == XXH_AVX2 /* AVX2 */ \
+  && defined(__GNUC__) && !defined(__clang__) /* GCC, not Clang */ \
+  && defined(__OPTIMIZE__) && !defined(__OPTIMIZE_SIZE__) /* respect -O0 and -Os */
+#  pragma GCC push_options
+#  pragma GCC optimize("-O2")
+#endif
+
+
+#if XXH_VECTOR == XXH_NEON
+/*
+ * NEON's setup for vmlal_u32 is a little more complicated than it is on
+ * SSE2, AVX2, and VSX.
+ *
+ * While PMULUDQ and VMULEUW both perform a mask, VMLAL.U32 performs an upcast.
+ *
+ * To do the same operation, the 128-bit 'Q' register needs to be split into
+ * two 64-bit 'D' registers, performing this operation::
+ *
+ *   [                a                 |                 b                ]
+ *            |              '---------. .--------'                |
+ *            |                         x                          |
+ *            |              .---------' '--------.                |
+ *   [ a & 0xFFFFFFFF | b & 0xFFFFFFFF ],[    a >> 32     |     b >> 32    ]
+ *
+ * Due to significant changes in aarch64, the fastest method for aarch64 is
+ * completely different than the fastest method for ARMv7-A.
+ *
+ * ARMv7-A treats D registers as unions overlaying Q registers, so modifying
+ * D11 will modify the high half of Q5. This is similar to how modifying AH
+ * will only affect bits 8-15 of AX on x86.
+ *
+ * VZIP takes two registers, and puts even lanes in one register and odd lanes
+ * in the other.
+ *
+ * On ARMv7-A, this strangely modifies both parameters in place instead of
+ * taking the usual 3-operand form.
+ *
+ * Therefore, if we want to do this, we can simply use a D-form VZIP.32 on the
+ * lower and upper halves of the Q register to end up with the high and low
+ * halves where we want - all in one instruction.
+ *
+ *   vzip.32   d10, d11       @ d10 = { d10[0], d11[0] }; d11 = { d10[1], d11[1] }
+ *
+ * Unfortunately we need inline assembly for this: Instructions modifying two
+ * registers at once is not possible in GCC or Clang's IR, and they have to
+ * create a copy.
+ *
+ * aarch64 requires a different approach.
+ *
+ * In order to make it easier to write a decent compiler for aarch64, many
+ * quirks were removed, such as conditional execution.
+ *
+ * NEON was also affected by this.
+ *
+ * aarch64 cannot access the high bits of a Q-form register, and writes to a
+ * D-form register zero the high bits, similar to how writes to W-form scalar
+ * registers (or DWORD registers on x86_64) work.
+ *
+ * The formerly free vget_high intrinsics now require a vext (with a few
+ * exceptions)
+ *
+ * Additionally, VZIP was replaced by ZIP1 and ZIP2, which are the equivalent
+ * of PUNPCKL* and PUNPCKH* in SSE, respectively, in order to only modify one
+ * operand.
+ *
+ * The equivalent of the VZIP.32 on the lower and upper halves would be this
+ * mess:
+ *
+ *   ext     v2.4s, v0.4s, v0.4s, #2 // v2 = { v0[2], v0[3], v0[0], v0[1] }
+ *   zip1    v1.2s, v0.2s, v2.2s     // v1 = { v0[0], v2[0] }
+ *   zip2    v0.2s, v0.2s, v1.2s     // v0 = { v0[1], v2[1] }
+ *
+ * Instead, we use a literal downcast, vmovn_u64 (XTN), and vshrn_n_u64 (SHRN):
+ *
+ *   shrn    v1.2s, v0.2d, #32  // v1 = (uint32x2_t)(v0 >> 32);
+ *   xtn     v0.2s, v0.2d       // v0 = (uint32x2_t)(v0 & 0xFFFFFFFF);
+ *
+ * This is available on ARMv7-A, but is less efficient than a single VZIP.32.
+ */
+
+/*!
+ * Function-like macro:
+ * void XXH_SPLIT_IN_PLACE(uint64x2_t &in, uint32x2_t &outLo, uint32x2_t &outHi)
+ * {
+ *     outLo = (uint32x2_t)(in & 0xFFFFFFFF);
+ *     outHi = (uint32x2_t)(in >> 32);
+ *     in = UNDEFINED;
+ * }
+ */
+# if !defined(XXH_NO_VZIP_HACK) /* define to disable */ \
+   && defined(__GNUC__) \
+   && !defined(__aarch64__) && !defined(__arm64__) && !defined(_M_ARM64)
+#  define XXH_SPLIT_IN_PLACE(in, outLo, outHi)                                              \
+    do {                                                                                    \
+      /* Undocumented GCC/Clang operand modifier: %e0 = lower D half, %f0 = upper D half */ \
+      /* https://github.com/gcc-mirror/gcc/blob/38cf91e5/gcc/config/arm/arm.c#L22486 */     \
+      /* https://github.com/llvm-mirror/llvm/blob/2c4ca683/lib/Target/ARM/ARMAsmPrinter.cpp#L399 */ \
+      __asm__("vzip.32  %e0, %f0" : "+w" (in));                                             \
+      (outLo) = vget_low_u32 (vreinterpretq_u32_u64(in));                                   \
+      (outHi) = vget_high_u32(vreinterpretq_u32_u64(in));                                   \
+   } while (0)
+# else
+#  define XXH_SPLIT_IN_PLACE(in, outLo, outHi)                                            \
+    do {                                                                                  \
+      (outLo) = vmovn_u64    (in);                                                        \
+      (outHi) = vshrn_n_u64  ((in), 32);                                                  \
+    } while (0)
+# endif
+#endif  /* XXH_VECTOR == XXH_NEON */
+
+/*
+ * VSX and Z Vector helpers.
+ *
+ * This is very messy, and any pull requests to clean this up are welcome.
+ *
+ * There are a lot of problems with supporting VSX and s390x, due to
+ * inconsistent intrinsics, spotty coverage, and multiple endiannesses.
+ */
+#if XXH_VECTOR == XXH_VSX
+#  if defined(__s390x__)
+#    include <s390intrin.h>
+#  else
+/* gcc's altivec.h can have the unwanted consequence to unconditionally
+ * #define bool, vector, and pixel keywords,
+ * with bad consequences for programs already using these keywords for other purposes.
+ * The paragraph defining these macros is skipped when __APPLE_ALTIVEC__ is defined.
+ * __APPLE_ALTIVEC__ is _generally_ defined automatically by the compiler,
+ * but it seems that, in some cases, it isn't.
+ * Force the build macro to be defined, so that keywords are not altered.
+ */
+#    if defined(__GNUC__) && !defined(__APPLE_ALTIVEC__)
+#      define __APPLE_ALTIVEC__
+#    endif
+#    include <altivec.h>
+#  endif
+
+typedef __vector unsigned long long xxh_u64x2;
+typedef __vector unsigned char xxh_u8x16;
+typedef __vector unsigned xxh_u32x4;
+
+# ifndef XXH_VSX_BE
+#  if defined(__BIG_ENDIAN__) \
+  || (defined(__BYTE_ORDER__) && __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__)
+#    define XXH_VSX_BE 1
+#  elif defined(__VEC_ELEMENT_REG_ORDER__) && __VEC_ELEMENT_REG_ORDER__ == __ORDER_BIG_ENDIAN__
+#    warning "-maltivec=be is not recommended. Please use native endianness."
+#    define XXH_VSX_BE 1
+#  else
+#    define XXH_VSX_BE 0
+#  endif
+# endif /* !defined(XXH_VSX_BE) */
+
+# if XXH_VSX_BE
+#  if defined(__POWER9_VECTOR__) || (defined(__clang__) && defined(__s390x__))
+#    define XXH_vec_revb vec_revb
+#  else
+/*!
+ * A polyfill for POWER9's vec_revb().
+ */
+XXH_FORCE_INLINE xxh_u64x2 XXH_vec_revb(xxh_u64x2 val)
+{
+    xxh_u8x16 const vByteSwap = { 0x07, 0x06, 0x05, 0x04, 0x03, 0x02, 0x01, 0x00,
+                                  0x0F, 0x0E, 0x0D, 0x0C, 0x0B, 0x0A, 0x09, 0x08 };
+    return vec_perm(val, val, vByteSwap);
+}
+#  endif
+# endif /* XXH_VSX_BE */
+
+/*!
+ * Performs an unaligned vector load and byte swaps it on big endian.
+ */
+XXH_FORCE_INLINE xxh_u64x2 XXH_vec_loadu(const void *ptr)
+{
+    xxh_u64x2 ret;
+    XXH_memcpy(&ret, ptr, sizeof(xxh_u64x2));
+# if XXH_VSX_BE
+    ret = XXH_vec_revb(ret);
+# endif
+    return ret;
+}
+
+/*
+ * vec_mulo and vec_mule are very problematic intrinsics on PowerPC
+ *
+ * These intrinsics weren't added until GCC 8, despite existing for a while,
+ * and they are endian dependent. Also, their meaning swap depending on version.
+ * */
+# if defined(__s390x__)
+ /* s390x is always big endian, no issue on this platform */
+#  define XXH_vec_mulo vec_mulo
+#  define XXH_vec_mule vec_mule
+# elif defined(__clang__) && XXH_HAS_BUILTIN(__builtin_altivec_vmuleuw)
+/* Clang has a better way to control this, we can just use the builtin which doesn't swap. */
+#  define XXH_vec_mulo __builtin_altivec_vmulouw
+#  define XXH_vec_mule __builtin_altivec_vmuleuw
+# else
+/* gcc needs inline assembly */
+/* Adapted from https://github.com/google/highwayhash/blob/master/highwayhash/hh_vsx.h. */
+XXH_FORCE_INLINE xxh_u64x2 XXH_vec_mulo(xxh_u32x4 a, xxh_u32x4 b)
+{
+    xxh_u64x2 result;
+    __asm__("vmulouw %0, %1, %2" : "=v" (result) : "v" (a), "v" (b));
+    return result;
+}
+XXH_FORCE_INLINE xxh_u64x2 XXH_vec_mule(xxh_u32x4 a, xxh_u32x4 b)
+{
+    xxh_u64x2 result;
+    __asm__("vmuleuw %0, %1, %2" : "=v" (result) : "v" (a), "v" (b));
+    return result;
+}
+# endif /* XXH_vec_mulo, XXH_vec_mule */
+#endif /* XXH_VECTOR == XXH_VSX */
+
+
+/* prefetch
+ * can be disabled, by declaring XXH_NO_PREFETCH build macro */
+#if defined(XXH_NO_PREFETCH)
+#  define XXH_PREFETCH(ptr)  (void)(ptr)  /* disabled */
+#else
+#  if defined(_MSC_VER) && (defined(_M_X64) || defined(_M_IX86))  /* _mm_prefetch() not defined outside of x86/x64 */
+#    include <mmintrin.h>   /* https://msdn.microsoft.com/fr-fr/library/84szxsww(v=vs.90).aspx */
+#    define XXH_PREFETCH(ptr)  _mm_prefetch((const char*)(ptr), _MM_HINT_T0)
+#  elif defined(__GNUC__) && ( (__GNUC__ >= 4) || ( (__GNUC__ == 3) && (__GNUC_MINOR__ >= 1) ) )
+#    define XXH_PREFETCH(ptr)  __builtin_prefetch((ptr), 0 /* rw==read */, 3 /* locality */)
+#  else
+#    define XXH_PREFETCH(ptr) (void)(ptr)  /* disabled */
+#  endif
+#endif  /* XXH_NO_PREFETCH */
+
+
+/* ==========================================
+ * XXH3 default settings
+ * ========================================== */
+
+#define XXH_SECRET_DEFAULT_SIZE 192   /* minimum XXH3_SECRET_SIZE_MIN */
+
+#if (XXH_SECRET_DEFAULT_SIZE < XXH3_SECRET_SIZE_MIN)
+#  error "default keyset is not large enough"
+#endif
+
+/*! Pseudorandom secret taken directly from FARSH. */
+XXH_ALIGN(64) static const xxh_u8 XXH3_kSecret[XXH_SECRET_DEFAULT_SIZE] = {
+    0xb8, 0xfe, 0x6c, 0x39, 0x23, 0xa4, 0x4b, 0xbe, 0x7c, 0x01, 0x81, 0x2c, 0xf7, 0x21, 0xad, 0x1c,
+    0xde, 0xd4, 0x6d, 0xe9, 0x83, 0x90, 0x97, 0xdb, 0x72, 0x40, 0xa4, 0xa4, 0xb7, 0xb3, 0x67, 0x1f,
+    0xcb, 0x79, 0xe6, 0x4e, 0xcc, 0xc0, 0xe5, 0x78, 0x82, 0x5a, 0xd0, 0x7d, 0xcc, 0xff, 0x72, 0x21,
+    0xb8, 0x08, 0x46, 0x74, 0xf7, 0x43, 0x24, 0x8e, 0xe0, 0x35, 0x90, 0xe6, 0x81, 0x3a, 0x26, 0x4c,
+    0x3c, 0x28, 0x52, 0xbb, 0x91, 0xc3, 0x00, 0xcb, 0x88, 0xd0, 0x65, 0x8b, 0x1b, 0x53, 0x2e, 0xa3,
+    0x71, 0x64, 0x48, 0x97, 0xa2, 0x0d, 0xf9, 0x4e, 0x38, 0x19, 0xef, 0x46, 0xa9, 0xde, 0xac, 0xd8,
+    0xa8, 0xfa, 0x76, 0x3f, 0xe3, 0x9c, 0x34, 0x3f, 0xf9, 0xdc, 0xbb, 0xc7, 0xc7, 0x0b, 0x4f, 0x1d,
+    0x8a, 0x51, 0xe0, 0x4b, 0xcd, 0xb4, 0x59, 0x31, 0xc8, 0x9f, 0x7e, 0xc9, 0xd9, 0x78, 0x73, 0x64,
+    0xea, 0xc5, 0xac, 0x83, 0x34, 0xd3, 0xeb, 0xc3, 0xc5, 0x81, 0xa0, 0xff, 0xfa, 0x13, 0x63, 0xeb,
+    0x17, 0x0d, 0xdd, 0x51, 0xb7, 0xf0, 0xda, 0x49, 0xd3, 0x16, 0x55, 0x26, 0x29, 0xd4, 0x68, 0x9e,
+    0x2b, 0x16, 0xbe, 0x58, 0x7d, 0x47, 0xa1, 0xfc, 0x8f, 0xf8, 0xb8, 0xd1, 0x7a, 0xd0, 0x31, 0xce,
+    0x45, 0xcb, 0x3a, 0x8f, 0x95, 0x16, 0x04, 0x28, 0xaf, 0xd7, 0xfb, 0xca, 0xbb, 0x4b, 0x40, 0x7e,
+};
+
+
+#ifdef XXH_OLD_NAMES
+#  define kSecret XXH3_kSecret
+#endif
+
+#ifdef XXH_DOXYGEN
+/*!
+ * @brief Calculates a 32-bit to 64-bit long multiply.
+ *
+ * Implemented as a macro.
+ *
+ * Wraps `__emulu` on MSVC x86 because it tends to call `__allmul` when it doesn't
+ * need to (but it shouldn't need to anyways, it is about 7 instructions to do
+ * a 64x64 multiply...). Since we know that this will _always_ emit `MULL`, we
+ * use that instead of the normal method.
+ *
+ * If you are compiling for platforms like Thumb-1 and don't have a better option,
+ * you may also want to write your own long multiply routine here.
+ *
+ * @param x, y Numbers to be multiplied
+ * @return 64-bit product of the low 32 bits of @p x and @p y.
+ */
+XXH_FORCE_INLINE xxh_u64
+XXH_mult32to64(xxh_u64 x, xxh_u64 y)
+{
+   return (x & 0xFFFFFFFF) * (y & 0xFFFFFFFF);
+}
+#elif defined(_MSC_VER) && defined(_M_IX86)
+#    include <intrin.h>
+#    define XXH_mult32to64(x, y) __emulu((unsigned)(x), (unsigned)(y))
+#else
+/*
+ * Downcast + upcast is usually better than masking on older compilers like
+ * GCC 4.2 (especially 32-bit ones), all without affecting newer compilers.
+ *
+ * The other method, (x & 0xFFFFFFFF) * (y & 0xFFFFFFFF), will AND both operands
+ * and perform a full 64x64 multiply -- entirely redundant on 32-bit.
+ */
+#    define XXH_mult32to64(x, y) ((xxh_u64)(xxh_u32)(x) * (xxh_u64)(xxh_u32)(y))
+#endif
+
+/*!
+ * @brief Calculates a 64->128-bit long multiply.
+ *
+ * Uses `__uint128_t` and `_umul128` if available, otherwise uses a scalar
+ * version.
+ *
+ * @param lhs , rhs The 64-bit integers to be multiplied
+ * @return The 128-bit result represented in an @ref XXH128_hash_t.
+ */
+static XXH128_hash_t
+XXH_mult64to128(xxh_u64 lhs, xxh_u64 rhs)
+{
+    /*
+     * GCC/Clang __uint128_t method.
+     *
+     * On most 64-bit targets, GCC and Clang define a __uint128_t type.
+     * This is usually the best way as it usually uses a native long 64-bit
+     * multiply, such as MULQ on x86_64 or MUL + UMULH on aarch64.
+     *
+     * Usually.
+     *
+     * Despite being a 32-bit platform, Clang (and emscripten) define this type
+     * despite not having the arithmetic for it. This results in a laggy
+     * compiler builtin call which calculates a full 128-bit multiply.
+     * In that case it is best to use the portable one.
+     * https://github.com/Cyan4973/xxHash/issues/211#issuecomment-515575677
+     */
+#if defined(__GNUC__) && !defined(__wasm__) \
+    && defined(__SIZEOF_INT128__) \
+    || (defined(_INTEGRAL_MAX_BITS) && _INTEGRAL_MAX_BITS >= 128)
+
+    __uint128_t const product = (__uint128_t)lhs * (__uint128_t)rhs;
+    XXH128_hash_t r128;
+    r128.low64  = (xxh_u64)(product);
+    r128.high64 = (xxh_u64)(product >> 64);
+    return r128;
+
+    /*
+     * MSVC for x64's _umul128 method.
+     *
+     * xxh_u64 _umul128(xxh_u64 Multiplier, xxh_u64 Multiplicand, xxh_u64 *HighProduct);
+     *
+     * This compiles to single operand MUL on x64.
+     */
+#elif defined(_M_X64) || defined(_M_IA64)
+
+#ifndef _MSC_VER
+#   pragma intrinsic(_umul128)
+#endif
+    xxh_u64 product_high;
+    xxh_u64 const product_low = _umul128(lhs, rhs, &product_high);
+    XXH128_hash_t r128;
+    r128.low64  = product_low;
+    r128.high64 = product_high;
+    return r128;
+
+    /*
+     * MSVC for ARM64's __umulh method.
+     *
+     * This compiles to the same MUL + UMULH as GCC/Clang's __uint128_t method.
+     */
+#elif defined(_M_ARM64)
+
+#ifndef _MSC_VER
+#   pragma intrinsic(__umulh)
+#endif
+    XXH128_hash_t r128;
+    r128.low64  = lhs * rhs;
+    r128.high64 = __umulh(lhs, rhs);
+    return r128;
+
+#else
+    /*
+     * Portable scalar method. Optimized for 32-bit and 64-bit ALUs.
+     *
+     * This is a fast and simple grade school multiply, which is shown below
+     * with base 10 arithmetic instead of base 0x100000000.
+     *
+     *           9 3 // D2 lhs = 93
+     *         x 7 5 // D2 rhs = 75
+     *     ----------
+     *           1 5 // D2 lo_lo = (93 % 10) * (75 % 10) = 15
+     *         4 5 | // D2 hi_lo = (93 / 10) * (75 % 10) = 45
+     *         2 1 | // D2 lo_hi = (93 % 10) * (75 / 10) = 21
+     *     + 6 3 | | // D2 hi_hi = (93 / 10) * (75 / 10) = 63
+     *     ---------
+     *         2 7 | // D2 cross = (15 / 10) + (45 % 10) + 21 = 27
+     *     + 6 7 | | // D2 upper = (27 / 10) + (45 / 10) + 63 = 67
+     *     ---------
+     *       6 9 7 5 // D4 res = (27 * 10) + (15 % 10) + (67 * 100) = 6975
+     *
+     * The reasons for adding the products like this are:
+     *  1. It avoids manual carry tracking. Just like how
+     *     (9 * 9) + 9 + 9 = 99, the same applies with this for UINT64_MAX.
+     *     This avoids a lot of complexity.
+     *
+     *  2. It hints for, and on Clang, compiles to, the powerful UMAAL
+     *     instruction available in ARM's Digital Signal Processing extension
+     *     in 32-bit ARMv6 and later, which is shown below:
+     *
+     *         void UMAAL(xxh_u32 *RdLo, xxh_u32 *RdHi, xxh_u32 Rn, xxh_u32 Rm)
+     *         {
+     *             xxh_u64 product = (xxh_u64)*RdLo * (xxh_u64)*RdHi + Rn + Rm;
+     *             *RdLo = (xxh_u32)(product & 0xFFFFFFFF);
+     *             *RdHi = (xxh_u32)(product >> 32);
+     *         }
+     *
+     *     This instruction was designed for efficient long multiplication, and
+     *     allows this to be calculated in only 4 instructions at speeds
+     *     comparable to some 64-bit ALUs.
+     *
+     *  3. It isn't terrible on other platforms. Usually this will be a couple
+     *     of 32-bit ADD/ADCs.
+     */
+
+    /* First calculate all of the cross products. */
+    xxh_u64 const lo_lo = XXH_mult32to64(lhs & 0xFFFFFFFF, rhs & 0xFFFFFFFF);
+    xxh_u64 const hi_lo = XXH_mult32to64(lhs >> 32,        rhs & 0xFFFFFFFF);
+    xxh_u64 const lo_hi = XXH_mult32to64(lhs & 0xFFFFFFFF, rhs >> 32);
+    xxh_u64 const hi_hi = XXH_mult32to64(lhs >> 32,        rhs >> 32);
+
+    /* Now add the products together. These will never overflow. */
+    xxh_u64 const cross = (lo_lo >> 32) + (hi_lo & 0xFFFFFFFF) + lo_hi;
+    xxh_u64 const upper = (hi_lo >> 32) + (cross >> 32)        + hi_hi;
+    xxh_u64 const lower = (cross << 32) | (lo_lo & 0xFFFFFFFF);
+
+    XXH128_hash_t r128;
+    r128.low64  = lower;
+    r128.high64 = upper;
+    return r128;
+#endif
+}
+
+/*!
+ * @brief Calculates a 64-bit to 128-bit multiply, then XOR folds it.
+ *
+ * The reason for the separate function is to prevent passing too many structs
+ * around by value. This will hopefully inline the multiply, but we don't force it.
+ *
+ * @param lhs , rhs The 64-bit integers to multiply
+ * @return The low 64 bits of the product XOR'd by the high 64 bits.
+ * @see XXH_mult64to128()
+ */
+static xxh_u64
+XXH3_mul128_fold64(xxh_u64 lhs, xxh_u64 rhs)
+{
+    XXH128_hash_t product = XXH_mult64to128(lhs, rhs);
+    return product.low64 ^ product.high64;
+}
+
+/*! Seems to produce slightly better code on GCC for some reason. */
+XXH_FORCE_INLINE xxh_u64 XXH_xorshift64(xxh_u64 v64, int shift)
+{
+    XXH_ASSERT(0 <= shift && shift < 64);
+    return v64 ^ (v64 >> shift);
+}
+
+/*
+ * This is a fast avalanche stage,
+ * suitable when input bits are already partially mixed
+ */
+static XXH64_hash_t XXH3_avalanche(xxh_u64 h64)
+{
+    h64 = XXH_xorshift64(h64, 37);
+    h64 *= 0x165667919E3779F9ULL;
+    h64 = XXH_xorshift64(h64, 32);
+    return h64;
+}
+
+/*
+ * This is a stronger avalanche,
+ * inspired by Pelle Evensen's rrmxmx
+ * preferable when input has not been previously mixed
+ */
+static XXH64_hash_t XXH3_rrmxmx(xxh_u64 h64, xxh_u64 len)
+{
+    /* this mix is inspired by Pelle Evensen's rrmxmx */
+    h64 ^= XXH_rotl64(h64, 49) ^ XXH_rotl64(h64, 24);
+    h64 *= 0x9FB21C651E98DF25ULL;
+    h64 ^= (h64 >> 35) + len ;
+    h64 *= 0x9FB21C651E98DF25ULL;
+    return XXH_xorshift64(h64, 28);
+}
+
+
+/* ==========================================
+ * Short keys
+ * ==========================================
+ * One of the shortcomings of XXH32 and XXH64 was that their performance was
+ * sub-optimal on short lengths. It used an iterative algorithm which strongly
+ * favored lengths that were a multiple of 4 or 8.
+ *
+ * Instead of iterating over individual inputs, we use a set of single shot
+ * functions which piece together a range of lengths and operate in constant time.
+ *
+ * Additionally, the number of multiplies has been significantly reduced. This
+ * reduces latency, especially when emulating 64-bit multiplies on 32-bit.
+ *
+ * Depending on the platform, this may or may not be faster than XXH32, but it
+ * is almost guaranteed to be faster than XXH64.
+ */
+
+/*
+ * At very short lengths, there isn't enough input to fully hide secrets, or use
+ * the entire secret.
+ *
+ * There is also only a limited amount of mixing we can do before significantly
+ * impacting performance.
+ *
+ * Therefore, we use different sections of the secret and always mix two secret
+ * samples with an XOR. This should have no effect on performance on the
+ * seedless or withSeed variants because everything _should_ be constant folded
+ * by modern compilers.
+ *
+ * The XOR mixing hides individual parts of the secret and increases entropy.
+ *
+ * This adds an extra layer of strength for custom secrets.
+ */
+XXH_FORCE_INLINE XXH64_hash_t
+XXH3_len_1to3_64b(const xxh_u8* input, size_t len, const xxh_u8* secret, XXH64_hash_t seed)
+{
+    XXH_ASSERT(input != NULL);
+    XXH_ASSERT(1 <= len && len <= 3);
+    XXH_ASSERT(secret != NULL);
+    /*
+     * len = 1: combined = { input[0], 0x01, input[0], input[0] }
+     * len = 2: combined = { input[1], 0x02, input[0], input[1] }
+     * len = 3: combined = { input[2], 0x03, input[0], input[1] }
+     */
+    {   xxh_u8  const c1 = input[0];
+        xxh_u8  const c2 = input[len >> 1];
+        xxh_u8  const c3 = input[len - 1];
+        xxh_u32 const combined = ((xxh_u32)c1 << 16) | ((xxh_u32)c2  << 24)
+                               | ((xxh_u32)c3 <<  0) | ((xxh_u32)len << 8);
+        xxh_u64 const bitflip = (XXH_readLE32(secret) ^ XXH_readLE32(secret+4)) + seed;
+        xxh_u64 const keyed = (xxh_u64)combined ^ bitflip;
+        return XXH64_avalanche(keyed);
+    }
+}
+
+XXH_FORCE_INLINE XXH64_hash_t
+XXH3_len_4to8_64b(const xxh_u8* input, size_t len, const xxh_u8* secret, XXH64_hash_t seed)
+{
+    XXH_ASSERT(input != NULL);
+    XXH_ASSERT(secret != NULL);
+    XXH_ASSERT(4 <= len && len <= 8);
+    seed ^= (xxh_u64)XXH_swap32((xxh_u32)seed) << 32;
+    {   xxh_u32 const input1 = XXH_readLE32(input);
+        xxh_u32 const input2 = XXH_readLE32(input + len - 4);
+        xxh_u64 const bitflip = (XXH_readLE64(secret+8) ^ XXH_readLE64(secret+16)) - seed;
+        xxh_u64 const input64 = input2 + (((xxh_u64)input1) << 32);
+        xxh_u64 const keyed = input64 ^ bitflip;
+        return XXH3_rrmxmx(keyed, len);
+    }
+}
+
+XXH_FORCE_INLINE XXH64_hash_t
+XXH3_len_9to16_64b(const xxh_u8* input, size_t len, const xxh_u8* secret, XXH64_hash_t seed)
+{
+    XXH_ASSERT(input != NULL);
+    XXH_ASSERT(secret != NULL);
+    XXH_ASSERT(9 <= len && len <= 16);
+    {   xxh_u64 const bitflip1 = (XXH_readLE64(secret+24) ^ XXH_readLE64(secret+32)) + seed;
+        xxh_u64 const bitflip2 = (XXH_readLE64(secret+40) ^ XXH_readLE64(secret+48)) - seed;
+        xxh_u64 const input_lo = XXH_readLE64(input)           ^ bitflip1;
+        xxh_u64 const input_hi = XXH_readLE64(input + len - 8) ^ bitflip2;
+        xxh_u64 const acc = len
+                          + XXH_swap64(input_lo) + input_hi
+                          + XXH3_mul128_fold64(input_lo, input_hi);
+        return XXH3_avalanche(acc);
+    }
+}
+
+XXH_FORCE_INLINE XXH64_hash_t
+XXH3_len_0to16_64b(const xxh_u8* input, size_t len, const xxh_u8* secret, XXH64_hash_t seed)
+{
+    XXH_ASSERT(len <= 16);
+    {   if (XXH_likely(len >  8)) return XXH3_len_9to16_64b(input, len, secret, seed);
+        if (XXH_likely(len >= 4)) return XXH3_len_4to8_64b(input, len, secret, seed);
+        if (len) return XXH3_len_1to3_64b(input, len, secret, seed);
+        return XXH64_avalanche(seed ^ (XXH_readLE64(secret+56) ^ XXH_readLE64(secret+64)));
+    }
+}
+
+/*
+ * DISCLAIMER: There are known *seed-dependent* multicollisions here due to
+ * multiplication by zero, affecting hashes of lengths 17 to 240.
+ *
+ * However, they are very unlikely.
+ *
+ * Keep this in mind when using the unseeded XXH3_64bits() variant: As with all
+ * unseeded non-cryptographic hashes, it does not attempt to defend itself
+ * against specially crafted inputs, only random inputs.
+ *
+ * Compared to classic UMAC where a 1 in 2^31 chance of 4 consecutive bytes
+ * cancelling out the secret is taken an arbitrary number of times (addressed
+ * in XXH3_accumulate_512), this collision is very unlikely with random inputs
+ * and/or proper seeding:
+ *
+ * This only has a 1 in 2^63 chance of 8 consecutive bytes cancelling out, in a
+ * function that is only called up to 16 times per hash with up to 240 bytes of
+ * input.
+ *
+ * This is not too bad for a non-cryptographic hash function, especially with
+ * only 64 bit outputs.
+ *
+ * The 128-bit variant (which trades some speed for strength) is NOT affected
+ * by this, although it is always a good idea to use a proper seed if you care
+ * about strength.
+ */
+XXH_FORCE_INLINE xxh_u64 XXH3_mix16B(const xxh_u8* XXH_RESTRICT input,
+                                     const xxh_u8* XXH_RESTRICT secret, xxh_u64 seed64)
+{
+#if defined(__GNUC__) && !defined(__clang__) /* GCC, not Clang */ \
+  && defined(__i386__) && defined(__SSE2__)  /* x86 + SSE2 */ \
+  && !defined(XXH_ENABLE_AUTOVECTORIZE)      /* Define to disable like XXH32 hack */
+    /*
+     * UGLY HACK:
+     * GCC for x86 tends to autovectorize the 128-bit multiply, resulting in
+     * slower code.
+     *
+     * By forcing seed64 into a register, we disrupt the cost model and
+     * cause it to scalarize. See `XXH32_round()`
+     *
+     * FIXME: Clang's output is still _much_ faster -- On an AMD Ryzen 3600,
+     * XXH3_64bits @ len=240 runs at 4.6 GB/s with Clang 9, but 3.3 GB/s on
+     * GCC 9.2, despite both emitting scalar code.
+     *
+     * GCC generates much better scalar code than Clang for the rest of XXH3,
+     * which is why finding a more optimal codepath is an interest.
+     */
+    XXH_COMPILER_GUARD(seed64);
+#endif
+    {   xxh_u64 const input_lo = XXH_readLE64(input);
+        xxh_u64 const input_hi = XXH_readLE64(input+8);
+        return XXH3_mul128_fold64(
+            input_lo ^ (XXH_readLE64(secret)   + seed64),
+            input_hi ^ (XXH_readLE64(secret+8) - seed64)
+        );
+    }
+}
+
+/* For mid range keys, XXH3 uses a Mum-hash variant. */
+XXH_FORCE_INLINE XXH64_hash_t
+XXH3_len_17to128_64b(const xxh_u8* XXH_RESTRICT input, size_t len,
+                     const xxh_u8* XXH_RESTRICT secret, size_t secretSize,
+                     XXH64_hash_t seed)
+{
+    XXH_ASSERT(secretSize >= XXH3_SECRET_SIZE_MIN); (void)secretSize;
+    XXH_ASSERT(16 < len && len <= 128);
+
+    {   xxh_u64 acc = len * XXH_PRIME64_1;
+        if (len > 32) {
+            if (len > 64) {
+                if (len > 96) {
+                    acc += XXH3_mix16B(input+48, secret+96, seed);
+                    acc += XXH3_mix16B(input+len-64, secret+112, seed);
+                }
+                acc += XXH3_mix16B(input+32, secret+64, seed);
+                acc += XXH3_mix16B(input+len-48, secret+80, seed);
+            }
+            acc += XXH3_mix16B(input+16, secret+32, seed);
+            acc += XXH3_mix16B(input+len-32, secret+48, seed);
+        }
+        acc += XXH3_mix16B(input+0, secret+0, seed);
+        acc += XXH3_mix16B(input+len-16, secret+16, seed);
+
+        return XXH3_avalanche(acc);
+    }
+}
+
+#define XXH3_MIDSIZE_MAX 240
+
+XXH_NO_INLINE XXH64_hash_t
+XXH3_len_129to240_64b(const xxh_u8* XXH_RESTRICT input, size_t len,
+                      const xxh_u8* XXH_RESTRICT secret, size_t secretSize,
+                      XXH64_hash_t seed)
+{
+    XXH_ASSERT(secretSize >= XXH3_SECRET_SIZE_MIN); (void)secretSize;
+    XXH_ASSERT(128 < len && len <= XXH3_MIDSIZE_MAX);
+
+    #define XXH3_MIDSIZE_STARTOFFSET 3
+    #define XXH3_MIDSIZE_LASTOFFSET  17
+
+    {   xxh_u64 acc = len * XXH_PRIME64_1;
+        int const nbRounds = (int)len / 16;
+        int i;
+        for (i=0; i<8; i++) {
+            acc += XXH3_mix16B(input+(16*i), secret+(16*i), seed);
+        }
+        acc = XXH3_avalanche(acc);
+        XXH_ASSERT(nbRounds >= 8);
+#if defined(__clang__)                                /* Clang */ \
+    && (defined(__ARM_NEON) || defined(__ARM_NEON__)) /* NEON */ \
+    && !defined(XXH_ENABLE_AUTOVECTORIZE)             /* Define to disable */
+        /*
+         * UGLY HACK:
+         * Clang for ARMv7-A tries to vectorize this loop, similar to GCC x86.
+         * In everywhere else, it uses scalar code.
+         *
+         * For 64->128-bit multiplies, even if the NEON was 100% optimal, it
+         * would still be slower than UMAAL (see XXH_mult64to128).
+         *
+         * Unfortunately, Clang doesn't handle the long multiplies properly and
+         * converts them to the nonexistent "vmulq_u64" intrinsic, which is then
+         * scalarized into an ugly mess of VMOV.32 instructions.
+         *
+         * This mess is difficult to avoid without turning autovectorization
+         * off completely, but they are usually relatively minor and/or not
+         * worth it to fix.
+         *
+         * This loop is the easiest to fix, as unlike XXH32, this pragma
+         * _actually works_ because it is a loop vectorization instead of an
+         * SLP vectorization.
+         */
+        #pragma clang loop vectorize(disable)
+#endif
+        for (i=8 ; i < nbRounds; i++) {
+            acc += XXH3_mix16B(input+(16*i), secret+(16*(i-8)) + XXH3_MIDSIZE_STARTOFFSET, seed);
+        }
+        /* last bytes */
+        acc += XXH3_mix16B(input + len - 16, secret + XXH3_SECRET_SIZE_MIN - XXH3_MIDSIZE_LASTOFFSET, seed);
+        return XXH3_avalanche(acc);
+    }
+}
+
+
+/* =======     Long Keys     ======= */
+
+#define XXH_STRIPE_LEN 64
+#define XXH_SECRET_CONSUME_RATE 8   /* nb of secret bytes consumed at each accumulation */
+#define XXH_ACC_NB (XXH_STRIPE_LEN / sizeof(xxh_u64))
+
+#ifdef XXH_OLD_NAMES
+#  define STRIPE_LEN XXH_STRIPE_LEN
+#  define ACC_NB XXH_ACC_NB
+#endif
+
+XXH_FORCE_INLINE void XXH_writeLE64(void* dst, xxh_u64 v64)
+{
+    if (!XXH_CPU_LITTLE_ENDIAN) v64 = XXH_swap64(v64);
+    XXH_memcpy(dst, &v64, sizeof(v64));
+}
+
+/* Several intrinsic functions below are supposed to accept __int64 as argument,
+ * as documented in https://software.intel.com/sites/landingpage/IntrinsicsGuide/ .
+ * However, several environments do not define __int64 type,
+ * requiring a workaround.
+ */
+#if !defined (__VMS) \
+  && (defined (__cplusplus) \
+  || (defined (__STDC_VERSION__) && (__STDC_VERSION__ >= 199901L) /* C99 */) )
+    typedef int64_t xxh_i64;
+#else
+    /* the following type must have a width of 64-bit */
+    typedef long long xxh_i64;
+#endif
+
+/*
+ * XXH3_accumulate_512 is the tightest loop for long inputs, and it is the most optimized.
+ *
+ * It is a hardened version of UMAC, based off of FARSH's implementation.
+ *
+ * This was chosen because it adapts quite well to 32-bit, 64-bit, and SIMD
+ * implementations, and it is ridiculously fast.
+ *
+ * We harden it by mixing the original input to the accumulators as well as the product.
+ *
+ * This means that in the (relatively likely) case of a multiply by zero, the
+ * original input is preserved.
+ *
+ * On 128-bit inputs, we swap 64-bit pairs when we add the input to improve
+ * cross-pollination, as otherwise the upper and lower halves would be
+ * essentially independent.
+ *
+ * This doesn't matter on 64-bit hashes since they all get merged together in
+ * the end, so we skip the extra step.
+ *
+ * Both XXH3_64bits and XXH3_128bits use this subroutine.
+ */
+
+#if (XXH_VECTOR == XXH_AVX512) \
+     || (defined(XXH_DISPATCH_AVX512) && XXH_DISPATCH_AVX512 != 0)
+
+#ifndef XXH_TARGET_AVX512
+# define XXH_TARGET_AVX512  /* disable attribute target */
+#endif
+
+XXH_FORCE_INLINE XXH_TARGET_AVX512 void
+XXH3_accumulate_512_avx512(void* XXH_RESTRICT acc,
+                     const void* XXH_RESTRICT input,
+                     const void* XXH_RESTRICT secret)
+{
+    __m512i* const xacc = (__m512i *) acc;
+    XXH_ASSERT((((size_t)acc) & 63) == 0);
+    XXH_STATIC_ASSERT(XXH_STRIPE_LEN == sizeof(__m512i));
+
+    {
+        /* data_vec    = input[0]; */
+        __m512i const data_vec    = _mm512_loadu_si512   (input);
+        /* key_vec     = secret[0]; */
+        __m512i const key_vec     = _mm512_loadu_si512   (secret);
+        /* data_key    = data_vec ^ key_vec; */
+        __m512i const data_key    = _mm512_xor_si512     (data_vec, key_vec);
+        /* data_key_lo = data_key >> 32; */
+        __m512i const data_key_lo = _mm512_shuffle_epi32 (data_key, (_MM_PERM_ENUM)_MM_SHUFFLE(0, 3, 0, 1));
+        /* product     = (data_key & 0xffffffff) * (data_key_lo & 0xffffffff); */
+        __m512i const product     = _mm512_mul_epu32     (data_key, data_key_lo);
+        /* xacc[0] += swap(data_vec); */
+        __m512i const data_swap = _mm512_shuffle_epi32(data_vec, (_MM_PERM_ENUM)_MM_SHUFFLE(1, 0, 3, 2));
+        __m512i const sum       = _mm512_add_epi64(*xacc, data_swap);
+        /* xacc[0] += product; */
+        *xacc = _mm512_add_epi64(product, sum);
+    }
+}
+
+/*
+ * XXH3_scrambleAcc: Scrambles the accumulators to improve mixing.
+ *
+ * Multiplication isn't perfect, as explained by Google in HighwayHash:
+ *
+ *  // Multiplication mixes/scrambles bytes 0-7 of the 64-bit result to
+ *  // varying degrees. In descending order of goodness, bytes
+ *  // 3 4 2 5 1 6 0 7 have quality 228 224 164 160 100 96 36 32.
+ *  // As expected, the upper and lower bytes are much worse.
+ *
+ * Source: https://github.com/google/highwayhash/blob/0aaf66b/highwayhash/hh_avx2.h#L291
+ *
+ * Since our algorithm uses a pseudorandom secret to add some variance into the
+ * mix, we don't need to (or want to) mix as often or as much as HighwayHash does.
+ *
+ * This isn't as tight as XXH3_accumulate, but still written in SIMD to avoid
+ * extraction.
+ *
+ * Both XXH3_64bits and XXH3_128bits use this subroutine.
+ */
+
+XXH_FORCE_INLINE XXH_TARGET_AVX512 void
+XXH3_scrambleAcc_avx512(void* XXH_RESTRICT acc, const void* XXH_RESTRICT secret)
+{
+    XXH_ASSERT((((size_t)acc) & 63) == 0);
+    XXH_STATIC_ASSERT(XXH_STRIPE_LEN == sizeof(__m512i));
+    {   __m512i* const xacc = (__m512i*) acc;
+        const __m512i prime32 = _mm512_set1_epi32((int)XXH_PRIME32_1);
+
+        /* xacc[0] ^= (xacc[0] >> 47) */
+        __m512i const acc_vec     = *xacc;
+        __m512i const shifted     = _mm512_srli_epi64    (acc_vec, 47);
+        __m512i const data_vec    = _mm512_xor_si512     (acc_vec, shifted);
+        /* xacc[0] ^= secret; */
+        __m512i const key_vec     = _mm512_loadu_si512   (secret);
+        __m512i const data_key    = _mm512_xor_si512     (data_vec, key_vec);
+
+        /* xacc[0] *= XXH_PRIME32_1; */
+        __m512i const data_key_hi = _mm512_shuffle_epi32 (data_key, (_MM_PERM_ENUM)_MM_SHUFFLE(0, 3, 0, 1));
+        __m512i const prod_lo     = _mm512_mul_epu32     (data_key, prime32);
+        __m512i const prod_hi     = _mm512_mul_epu32     (data_key_hi, prime32);
+        *xacc = _mm512_add_epi64(prod_lo, _mm512_slli_epi64(prod_hi, 32));
+    }
+}
+
+XXH_FORCE_INLINE XXH_TARGET_AVX512 void
+XXH3_initCustomSecret_avx512(void* XXH_RESTRICT customSecret, xxh_u64 seed64)
+{
+    XXH_STATIC_ASSERT((XXH_SECRET_DEFAULT_SIZE & 63) == 0);
+    XXH_STATIC_ASSERT(XXH_SEC_ALIGN == 64);
+    XXH_ASSERT(((size_t)customSecret & 63) == 0);
+    (void)(&XXH_writeLE64);
+    {   int const nbRounds = XXH_SECRET_DEFAULT_SIZE / sizeof(__m512i);
+        __m512i const seed = _mm512_mask_set1_epi64(_mm512_set1_epi64((xxh_i64)seed64), 0xAA, (xxh_i64)(0U - seed64));
+
+        const __m512i* const src  = (const __m512i*) ((const void*) XXH3_kSecret);
+              __m512i* const dest = (      __m512i*) customSecret;
+        int i;
+        XXH_ASSERT(((size_t)src & 63) == 0); /* control alignment */
+        XXH_ASSERT(((size_t)dest & 63) == 0);
+        for (i=0; i < nbRounds; ++i) {
+            /* GCC has a bug, _mm512_stream_load_si512 accepts 'void*', not 'void const*',
+             * this will warn "discards 'const' qualifier". */
+            union {
+                const __m512i* cp;
+                void* p;
+            } remote_const_void;
+            remote_const_void.cp = src + i;
+            dest[i] = _mm512_add_epi64(_mm512_stream_load_si512(remote_const_void.p), seed);
+    }   }
+}
+
+#endif
+
+#if (XXH_VECTOR == XXH_AVX2) \
+    || (defined(XXH_DISPATCH_AVX2) && XXH_DISPATCH_AVX2 != 0)
+
+#ifndef XXH_TARGET_AVX2
+# define XXH_TARGET_AVX2  /* disable attribute target */
+#endif
+
+XXH_FORCE_INLINE XXH_TARGET_AVX2 void
+XXH3_accumulate_512_avx2( void* XXH_RESTRICT acc,
+                    const void* XXH_RESTRICT input,
+                    const void* XXH_RESTRICT secret)
+{
+    XXH_ASSERT((((size_t)acc) & 31) == 0);
+    {   __m256i* const xacc    =       (__m256i *) acc;
+        /* Unaligned. This is mainly for pointer arithmetic, and because
+         * _mm256_loadu_si256 requires  a const __m256i * pointer for some reason. */
+        const         __m256i* const xinput  = (const __m256i *) input;
+        /* Unaligned. This is mainly for pointer arithmetic, and because
+         * _mm256_loadu_si256 requires a const __m256i * pointer for some reason. */
+        const         __m256i* const xsecret = (const __m256i *) secret;
+
+        size_t i;
+        for (i=0; i < XXH_STRIPE_LEN/sizeof(__m256i); i++) {
+            /* data_vec    = xinput[i]; */
+            __m256i const data_vec    = _mm256_loadu_si256    (xinput+i);
+            /* key_vec     = xsecret[i]; */
+            __m256i const key_vec     = _mm256_loadu_si256   (xsecret+i);
+            /* data_key    = data_vec ^ key_vec; */
+            __m256i const data_key    = _mm256_xor_si256     (data_vec, key_vec);
+            /* data_key_lo = data_key >> 32; */
+            __m256i const data_key_lo = _mm256_shuffle_epi32 (data_key, _MM_SHUFFLE(0, 3, 0, 1));
+            /* product     = (data_key & 0xffffffff) * (data_key_lo & 0xffffffff); */
+            __m256i const product     = _mm256_mul_epu32     (data_key, data_key_lo);
+            /* xacc[i] += swap(data_vec); */
+            __m256i const data_swap = _mm256_shuffle_epi32(data_vec, _MM_SHUFFLE(1, 0, 3, 2));
+            __m256i const sum       = _mm256_add_epi64(xacc[i], data_swap);
+            /* xacc[i] += product; */
+            xacc[i] = _mm256_add_epi64(product, sum);
+    }   }
+}
+
+XXH_FORCE_INLINE XXH_TARGET_AVX2 void
+XXH3_scrambleAcc_avx2(void* XXH_RESTRICT acc, const void* XXH_RESTRICT secret)
+{
+    XXH_ASSERT((((size_t)acc) & 31) == 0);
+    {   __m256i* const xacc = (__m256i*) acc;
+        /* Unaligned. This is mainly for pointer arithmetic, and because
+         * _mm256_loadu_si256 requires a const __m256i * pointer for some reason. */
+        const         __m256i* const xsecret = (const __m256i *) secret;
+        const __m256i prime32 = _mm256_set1_epi32((int)XXH_PRIME32_1);
+
+        size_t i;
+        for (i=0; i < XXH_STRIPE_LEN/sizeof(__m256i); i++) {
+            /* xacc[i] ^= (xacc[i] >> 47) */
+            __m256i const acc_vec     = xacc[i];
+            __m256i const shifted     = _mm256_srli_epi64    (acc_vec, 47);
+            __m256i const data_vec    = _mm256_xor_si256     (acc_vec, shifted);
+            /* xacc[i] ^= xsecret; */
+            __m256i const key_vec     = _mm256_loadu_si256   (xsecret+i);
+            __m256i const data_key    = _mm256_xor_si256     (data_vec, key_vec);
+
+            /* xacc[i] *= XXH_PRIME32_1; */
+            __m256i const data_key_hi = _mm256_shuffle_epi32 (data_key, _MM_SHUFFLE(0, 3, 0, 1));
+            __m256i const prod_lo     = _mm256_mul_epu32     (data_key, prime32);
+            __m256i const prod_hi     = _mm256_mul_epu32     (data_key_hi, prime32);
+            xacc[i] = _mm256_add_epi64(prod_lo, _mm256_slli_epi64(prod_hi, 32));
+        }
+    }
+}
+
+XXH_FORCE_INLINE XXH_TARGET_AVX2 void XXH3_initCustomSecret_avx2(void* XXH_RESTRICT customSecret, xxh_u64 seed64)
+{
+    XXH_STATIC_ASSERT((XXH_SECRET_DEFAULT_SIZE & 31) == 0);
+    XXH_STATIC_ASSERT((XXH_SECRET_DEFAULT_SIZE / sizeof(__m256i)) == 6);
+    XXH_STATIC_ASSERT(XXH_SEC_ALIGN <= 64);
+    (void)(&XXH_writeLE64);
+    XXH_PREFETCH(customSecret);
+    {   __m256i const seed = _mm256_set_epi64x((xxh_i64)(0U - seed64), (xxh_i64)seed64, (xxh_i64)(0U - seed64), (xxh_i64)seed64);
+
+        const __m256i* const src  = (const __m256i*) ((const void*) XXH3_kSecret);
+              __m256i*       dest = (      __m256i*) customSecret;
+
+#       if defined(__GNUC__) || defined(__clang__)
+        /*
+         * On GCC & Clang, marking 'dest' as modified will cause the compiler:
+         *   - do not extract the secret from sse registers in the internal loop
+         *   - use less common registers, and avoid pushing these reg into stack
+         */
+        XXH_COMPILER_GUARD(dest);
+#       endif
+        XXH_ASSERT(((size_t)src & 31) == 0); /* control alignment */
+        XXH_ASSERT(((size_t)dest & 31) == 0);
+
+        /* GCC -O2 need unroll loop manually */
+        dest[0] = _mm256_add_epi64(_mm256_stream_load_si256(src+0), seed);
+        dest[1] = _mm256_add_epi64(_mm256_stream_load_si256(src+1), seed);
+        dest[2] = _mm256_add_epi64(_mm256_stream_load_si256(src+2), seed);
+        dest[3] = _mm256_add_epi64(_mm256_stream_load_si256(src+3), seed);
+        dest[4] = _mm256_add_epi64(_mm256_stream_load_si256(src+4), seed);
+        dest[5] = _mm256_add_epi64(_mm256_stream_load_si256(src+5), seed);
+    }
+}
+
+#endif
+
+/* x86dispatch always generates SSE2 */
+#if (XXH_VECTOR == XXH_SSE2) || defined(XXH_X86DISPATCH)
+
+#ifndef XXH_TARGET_SSE2
+# define XXH_TARGET_SSE2  /* disable attribute target */
+#endif
+
+XXH_FORCE_INLINE XXH_TARGET_SSE2 void
+XXH3_accumulate_512_sse2( void* XXH_RESTRICT acc,
+                    const void* XXH_RESTRICT input,
+                    const void* XXH_RESTRICT secret)
+{
+    /* SSE2 is just a half-scale version of the AVX2 version. */
+    XXH_ASSERT((((size_t)acc) & 15) == 0);
+    {   __m128i* const xacc    =       (__m128i *) acc;
+        /* Unaligned. This is mainly for pointer arithmetic, and because
+         * _mm_loadu_si128 requires a const __m128i * pointer for some reason. */
+        const         __m128i* const xinput  = (const __m128i *) input;
+        /* Unaligned. This is mainly for pointer arithmetic, and because
+         * _mm_loadu_si128 requires a const __m128i * pointer for some reason. */
+        const         __m128i* const xsecret = (const __m128i *) secret;
+
+        size_t i;
+        for (i=0; i < XXH_STRIPE_LEN/sizeof(__m128i); i++) {
+            /* data_vec    = xinput[i]; */
+            __m128i const data_vec    = _mm_loadu_si128   (xinput+i);
+            /* key_vec     = xsecret[i]; */
+            __m128i const key_vec     = _mm_loadu_si128   (xsecret+i);
+            /* data_key    = data_vec ^ key_vec; */
+            __m128i const data_key    = _mm_xor_si128     (data_vec, key_vec);
+            /* data_key_lo = data_key >> 32; */
+            __m128i const data_key_lo = _mm_shuffle_epi32 (data_key, _MM_SHUFFLE(0, 3, 0, 1));
+            /* product     = (data_key & 0xffffffff) * (data_key_lo & 0xffffffff); */
+            __m128i const product     = _mm_mul_epu32     (data_key, data_key_lo);
+            /* xacc[i] += swap(data_vec); */
+            __m128i const data_swap = _mm_shuffle_epi32(data_vec, _MM_SHUFFLE(1,0,3,2));
+            __m128i const sum       = _mm_add_epi64(xacc[i], data_swap);
+            /* xacc[i] += product; */
+            xacc[i] = _mm_add_epi64(product, sum);
+    }   }
+}
+
+XXH_FORCE_INLINE XXH_TARGET_SSE2 void
+XXH3_scrambleAcc_sse2(void* XXH_RESTRICT acc, const void* XXH_RESTRICT secret)
+{
+    XXH_ASSERT((((size_t)acc) & 15) == 0);
+    {   __m128i* const xacc = (__m128i*) acc;
+        /* Unaligned. This is mainly for pointer arithmetic, and because
+         * _mm_loadu_si128 requires a const __m128i * pointer for some reason. */
+        const         __m128i* const xsecret = (const __m128i *) secret;
+        const __m128i prime32 = _mm_set1_epi32((int)XXH_PRIME32_1);
+
+        size_t i;
+        for (i=0; i < XXH_STRIPE_LEN/sizeof(__m128i); i++) {
+            /* xacc[i] ^= (xacc[i] >> 47) */
+            __m128i const acc_vec     = xacc[i];
+            __m128i const shifted     = _mm_srli_epi64    (acc_vec, 47);
+            __m128i const data_vec    = _mm_xor_si128     (acc_vec, shifted);
+            /* xacc[i] ^= xsecret[i]; */
+            __m128i const key_vec     = _mm_loadu_si128   (xsecret+i);
+            __m128i const data_key    = _mm_xor_si128     (data_vec, key_vec);
+
+            /* xacc[i] *= XXH_PRIME32_1; */
+            __m128i const data_key_hi = _mm_shuffle_epi32 (data_key, _MM_SHUFFLE(0, 3, 0, 1));
+            __m128i const prod_lo     = _mm_mul_epu32     (data_key, prime32);
+            __m128i const prod_hi     = _mm_mul_epu32     (data_key_hi, prime32);
+            xacc[i] = _mm_add_epi64(prod_lo, _mm_slli_epi64(prod_hi, 32));
+        }
+    }
+}
+
+XXH_FORCE_INLINE XXH_TARGET_SSE2 void XXH3_initCustomSecret_sse2(void* XXH_RESTRICT customSecret, xxh_u64 seed64)
+{
+    XXH_STATIC_ASSERT((XXH_SECRET_DEFAULT_SIZE & 15) == 0);
+    (void)(&XXH_writeLE64);
+    {   int const nbRounds = XXH_SECRET_DEFAULT_SIZE / sizeof(__m128i);
+
+#       if defined(_MSC_VER) && defined(_M_IX86) && _MSC_VER < 1900
+        /* MSVC 32bit mode does not support _mm_set_epi64x before 2015 */
+        XXH_ALIGN(16) const xxh_i64 seed64x2[2] = { (xxh_i64)seed64, (xxh_i64)(0U - seed64) };
+        __m128i const seed = _mm_load_si128((__m128i const*)seed64x2);
+#       else
+        __m128i const seed = _mm_set_epi64x((xxh_i64)(0U - seed64), (xxh_i64)seed64);
+#       endif
+        int i;
+
+        const void* const src16 = XXH3_kSecret;
+        __m128i* dst16 = (__m128i*) customSecret;
+#       if defined(__GNUC__) || defined(__clang__)
+        /*
+         * On GCC & Clang, marking 'dest' as modified will cause the compiler:
+         *   - do not extract the secret from sse registers in the internal loop
+         *   - use less common registers, and avoid pushing these reg into stack
+         */
+        XXH_COMPILER_GUARD(dst16);
+#       endif
+        XXH_ASSERT(((size_t)src16 & 15) == 0); /* control alignment */
+        XXH_ASSERT(((size_t)dst16 & 15) == 0);
+
+        for (i=0; i < nbRounds; ++i) {
+            dst16[i] = _mm_add_epi64(_mm_load_si128((const __m128i *)src16+i), seed);
+    }   }
+}
+
+#endif
+
+#if (XXH_VECTOR == XXH_NEON)
+
+XXH_FORCE_INLINE void
+XXH3_accumulate_512_neon( void* XXH_RESTRICT acc,
+                    const void* XXH_RESTRICT input,
+                    const void* XXH_RESTRICT secret)
+{
+    XXH_ASSERT((((size_t)acc) & 15) == 0);
+    {
+        uint64x2_t* const xacc = (uint64x2_t *) acc;
+        /* We don't use a uint32x4_t pointer because it causes bus errors on ARMv7. */
+        uint8_t const* const xinput = (const uint8_t *) input;
+        uint8_t const* const xsecret  = (const uint8_t *) secret;
+
+        size_t i;
+        for (i=0; i < XXH_STRIPE_LEN / sizeof(uint64x2_t); i++) {
+            /* data_vec = xinput[i]; */
+            uint8x16_t data_vec    = vld1q_u8(xinput  + (i * 16));
+            /* key_vec  = xsecret[i];  */
+            uint8x16_t key_vec     = vld1q_u8(xsecret + (i * 16));
+            uint64x2_t data_key;
+            uint32x2_t data_key_lo, data_key_hi;
+            /* xacc[i] += swap(data_vec); */
+            uint64x2_t const data64  = vreinterpretq_u64_u8(data_vec);
+            uint64x2_t const swapped = vextq_u64(data64, data64, 1);
+            xacc[i] = vaddq_u64 (xacc[i], swapped);
+            /* data_key = data_vec ^ key_vec; */
+            data_key = vreinterpretq_u64_u8(veorq_u8(data_vec, key_vec));
+            /* data_key_lo = (uint32x2_t) (data_key & 0xFFFFFFFF);
+             * data_key_hi = (uint32x2_t) (data_key >> 32);
+             * data_key = UNDEFINED; */
+            XXH_SPLIT_IN_PLACE(data_key, data_key_lo, data_key_hi);
+            /* xacc[i] += (uint64x2_t) data_key_lo * (uint64x2_t) data_key_hi; */
+            xacc[i] = vmlal_u32 (xacc[i], data_key_lo, data_key_hi);
+
+        }
+    }
+}
+
+XXH_FORCE_INLINE void
+XXH3_scrambleAcc_neon(void* XXH_RESTRICT acc, const void* XXH_RESTRICT secret)
+{
+    XXH_ASSERT((((size_t)acc) & 15) == 0);
+
+    {   uint64x2_t* xacc       = (uint64x2_t*) acc;
+        uint8_t const* xsecret = (uint8_t const*) secret;
+        uint32x2_t prime       = vdup_n_u32 (XXH_PRIME32_1);
+
+        size_t i;
+        for (i=0; i < XXH_STRIPE_LEN/sizeof(uint64x2_t); i++) {
+            /* xacc[i] ^= (xacc[i] >> 47); */
+            uint64x2_t acc_vec  = xacc[i];
+            uint64x2_t shifted  = vshrq_n_u64 (acc_vec, 47);
+            uint64x2_t data_vec = veorq_u64   (acc_vec, shifted);
+
+            /* xacc[i] ^= xsecret[i]; */
+            uint8x16_t key_vec  = vld1q_u8    (xsecret + (i * 16));
+            uint64x2_t data_key = veorq_u64   (data_vec, vreinterpretq_u64_u8(key_vec));
+
+            /* xacc[i] *= XXH_PRIME32_1 */
+            uint32x2_t data_key_lo, data_key_hi;
+            /* data_key_lo = (uint32x2_t) (xacc[i] & 0xFFFFFFFF);
+             * data_key_hi = (uint32x2_t) (xacc[i] >> 32);
+             * xacc[i] = UNDEFINED; */
+            XXH_SPLIT_IN_PLACE(data_key, data_key_lo, data_key_hi);
+            {   /*
+                 * prod_hi = (data_key >> 32) * XXH_PRIME32_1;
+                 *
+                 * Avoid vmul_u32 + vshll_n_u32 since Clang 6 and 7 will
+                 * incorrectly "optimize" this:
+                 *   tmp     = vmul_u32(vmovn_u64(a), vmovn_u64(b));
+                 *   shifted = vshll_n_u32(tmp, 32);
+                 * to this:
+                 *   tmp     = "vmulq_u64"(a, b); // no such thing!
+                 *   shifted = vshlq_n_u64(tmp, 32);
+                 *
+                 * However, unlike SSE, Clang lacks a 64-bit multiply routine
+                 * for NEON, and it scalarizes two 64-bit multiplies instead.
+                 *
+                 * vmull_u32 has the same timing as vmul_u32, and it avoids
+                 * this bug completely.
+                 * See https://bugs.llvm.org/show_bug.cgi?id=39967
+                 */
+                uint64x2_t prod_hi = vmull_u32 (data_key_hi, prime);
+                /* xacc[i] = prod_hi << 32; */
+                xacc[i] = vshlq_n_u64(prod_hi, 32);
+                /* xacc[i] += (prod_hi & 0xFFFFFFFF) * XXH_PRIME32_1; */
+                xacc[i] = vmlal_u32(xacc[i], data_key_lo, prime);
+            }
+    }   }
+}
+
+#endif
+
+#if (XXH_VECTOR == XXH_VSX)
+
+XXH_FORCE_INLINE void
+XXH3_accumulate_512_vsx(  void* XXH_RESTRICT acc,
+                    const void* XXH_RESTRICT input,
+                    const void* XXH_RESTRICT secret)
+{
+    /* presumed aligned */
+    unsigned long long* const xacc = (unsigned long long*) acc;
+    xxh_u64x2 const* const xinput   = (xxh_u64x2 const*) input;   /* no alignment restriction */
+    xxh_u64x2 const* const xsecret  = (xxh_u64x2 const*) secret;    /* no alignment restriction */
+    xxh_u64x2 const v32 = { 32, 32 };
+    size_t i;
+    for (i = 0; i < XXH_STRIPE_LEN / sizeof(xxh_u64x2); i++) {
+        /* data_vec = xinput[i]; */
+        xxh_u64x2 const data_vec = XXH_vec_loadu(xinput + i);
+        /* key_vec = xsecret[i]; */
+        xxh_u64x2 const key_vec  = XXH_vec_loadu(xsecret + i);
+        xxh_u64x2 const data_key = data_vec ^ key_vec;
+        /* shuffled = (data_key << 32) | (data_key >> 32); */
+        xxh_u32x4 const shuffled = (xxh_u32x4)vec_rl(data_key, v32);
+        /* product = ((xxh_u64x2)data_key & 0xFFFFFFFF) * ((xxh_u64x2)shuffled & 0xFFFFFFFF); */
+        xxh_u64x2 const product  = XXH_vec_mulo((xxh_u32x4)data_key, shuffled);
+        /* acc_vec = xacc[i]; */
+        xxh_u64x2 acc_vec        = vec_xl(0, xacc + 2 * i);
+        acc_vec += product;
+
+        /* swap high and low halves */
+#ifdef __s390x__
+        acc_vec += vec_permi(data_vec, data_vec, 2);
+#else
+        acc_vec += vec_xxpermdi(data_vec, data_vec, 2);
+#endif
+        /* xacc[i] = acc_vec; */
+        vec_xst(acc_vec, 0, xacc + 2 * i);
+    }
+}
+
+XXH_FORCE_INLINE void
+XXH3_scrambleAcc_vsx(void* XXH_RESTRICT acc, const void* XXH_RESTRICT secret)
+{
+    XXH_ASSERT((((size_t)acc) & 15) == 0);
+
+    {         xxh_u64x2* const xacc    =       (xxh_u64x2*) acc;
+        const xxh_u64x2* const xsecret = (const xxh_u64x2*) secret;
+        /* constants */
+        xxh_u64x2 const v32  = { 32, 32 };
+        xxh_u64x2 const v47 = { 47, 47 };
+        xxh_u32x4 const prime = { XXH_PRIME32_1, XXH_PRIME32_1, XXH_PRIME32_1, XXH_PRIME32_1 };
+        size_t i;
+        for (i = 0; i < XXH_STRIPE_LEN / sizeof(xxh_u64x2); i++) {
+            /* xacc[i] ^= (xacc[i] >> 47); */
+            xxh_u64x2 const acc_vec  = xacc[i];
+            xxh_u64x2 const data_vec = acc_vec ^ (acc_vec >> v47);
+
+            /* xacc[i] ^= xsecret[i]; */
+            xxh_u64x2 const key_vec  = XXH_vec_loadu(xsecret + i);
+            xxh_u64x2 const data_key = data_vec ^ key_vec;
+
+            /* xacc[i] *= XXH_PRIME32_1 */
+            /* prod_lo = ((xxh_u64x2)data_key & 0xFFFFFFFF) * ((xxh_u64x2)prime & 0xFFFFFFFF);  */
+            xxh_u64x2 const prod_even  = XXH_vec_mule((xxh_u32x4)data_key, prime);
+            /* prod_hi = ((xxh_u64x2)data_key >> 32) * ((xxh_u64x2)prime >> 32);  */
+            xxh_u64x2 const prod_odd  = XXH_vec_mulo((xxh_u32x4)data_key, prime);
+            xacc[i] = prod_odd + (prod_even << v32);
+    }   }
+}
+
+#endif
+
+/* scalar variants - universal */
+
+XXH_FORCE_INLINE void
+XXH3_accumulate_512_scalar(void* XXH_RESTRICT acc,
+                     const void* XXH_RESTRICT input,
+                     const void* XXH_RESTRICT secret)
+{
+    xxh_u64* const xacc = (xxh_u64*) acc; /* presumed aligned */
+    const xxh_u8* const xinput  = (const xxh_u8*) input;  /* no alignment restriction */
+    const xxh_u8* const xsecret = (const xxh_u8*) secret;   /* no alignment restriction */
+    size_t i;
+    XXH_ASSERT(((size_t)acc & (XXH_ACC_ALIGN-1)) == 0);
+    for (i=0; i < XXH_ACC_NB; i++) {
+        xxh_u64 const data_val = XXH_readLE64(xinput + 8*i);
+        xxh_u64 const data_key = data_val ^ XXH_readLE64(xsecret + i*8);
+        xacc[i ^ 1] += data_val; /* swap adjacent lanes */
+        xacc[i] += XXH_mult32to64(data_key & 0xFFFFFFFF, data_key >> 32);
+    }
+}
+
+XXH_FORCE_INLINE void
+XXH3_scrambleAcc_scalar(void* XXH_RESTRICT acc, const void* XXH_RESTRICT secret)
+{
+    xxh_u64* const xacc = (xxh_u64*) acc;   /* presumed aligned */
+    const xxh_u8* const xsecret = (const xxh_u8*) secret;   /* no alignment restriction */
+    size_t i;
+    XXH_ASSERT((((size_t)acc) & (XXH_ACC_ALIGN-1)) == 0);
+    for (i=0; i < XXH_ACC_NB; i++) {
+        xxh_u64 const key64 = XXH_readLE64(xsecret + 8*i);
+        xxh_u64 acc64 = xacc[i];
+        acc64 = XXH_xorshift64(acc64, 47);
+        acc64 ^= key64;
+        acc64 *= XXH_PRIME32_1;
+        xacc[i] = acc64;
+    }
+}
+
+XXH_FORCE_INLINE void
+XXH3_initCustomSecret_scalar(void* XXH_RESTRICT customSecret, xxh_u64 seed64)
+{
+    /*
+     * We need a separate pointer for the hack below,
+     * which requires a non-const pointer.
+     * Any decent compiler will optimize this out otherwise.
+     */
+    const xxh_u8* kSecretPtr = XXH3_kSecret;
+    XXH_STATIC_ASSERT((XXH_SECRET_DEFAULT_SIZE & 15) == 0);
+
+#if defined(__clang__) && defined(__aarch64__)
+    /*
+     * UGLY HACK:
+     * Clang generates a bunch of MOV/MOVK pairs for aarch64, and they are
+     * placed sequentially, in order, at the top of the unrolled loop.
+     *
+     * While MOVK is great for generating constants (2 cycles for a 64-bit
+     * constant compared to 4 cycles for LDR), long MOVK chains stall the
+     * integer pipelines:
+     *   I   L   S
+     * MOVK
+     * MOVK
+     * MOVK
+     * MOVK
+     * ADD
+     * SUB      STR
+     *          STR
+     * By forcing loads from memory (as the asm line causes Clang to assume
+     * that XXH3_kSecretPtr has been changed), the pipelines are used more
+     * efficiently:
+     *   I   L   S
+     *      LDR
+     *  ADD LDR
+     *  SUB     STR
+     *          STR
+     * XXH3_64bits_withSeed, len == 256, Snapdragon 835
+     *   without hack: 2654.4 MB/s
+     *   with hack:    3202.9 MB/s
+     */
+    XXH_COMPILER_GUARD(kSecretPtr);
+#endif
+    /*
+     * Note: in debug mode, this overrides the asm optimization
+     * and Clang will emit MOVK chains again.
+     */
+    XXH_ASSERT(kSecretPtr == XXH3_kSecret);
+
+    {   int const nbRounds = XXH_SECRET_DEFAULT_SIZE / 16;
+        int i;
+        for (i=0; i < nbRounds; i++) {
+            /*
+             * The asm hack causes Clang to assume that kSecretPtr aliases with
+             * customSecret, and on aarch64, this prevented LDP from merging two
+             * loads together for free. Putting the loads together before the stores
+             * properly generates LDP.
+             */
+            xxh_u64 lo = XXH_readLE64(kSecretPtr + 16*i)     + seed64;
+            xxh_u64 hi = XXH_readLE64(kSecretPtr + 16*i + 8) - seed64;
+            XXH_writeLE64((xxh_u8*)customSecret + 16*i,     lo);
+            XXH_writeLE64((xxh_u8*)customSecret + 16*i + 8, hi);
+    }   }
+}
+
+
+typedef void (*XXH3_f_accumulate_512)(void* XXH_RESTRICT, const void*, const void*);
+typedef void (*XXH3_f_scrambleAcc)(void* XXH_RESTRICT, const void*);
+typedef void (*XXH3_f_initCustomSecret)(void* XXH_RESTRICT, xxh_u64);
+
+
+#if (XXH_VECTOR == XXH_AVX512)
+
+#define XXH3_accumulate_512 XXH3_accumulate_512_avx512
+#define XXH3_scrambleAcc    XXH3_scrambleAcc_avx512
+#define XXH3_initCustomSecret XXH3_initCustomSecret_avx512
+
+#elif (XXH_VECTOR == XXH_AVX2)
+
+#define XXH3_accumulate_512 XXH3_accumulate_512_avx2
+#define XXH3_scrambleAcc    XXH3_scrambleAcc_avx2
+#define XXH3_initCustomSecret XXH3_initCustomSecret_avx2
+
+#elif (XXH_VECTOR == XXH_SSE2)
+
+#define XXH3_accumulate_512 XXH3_accumulate_512_sse2
+#define XXH3_scrambleAcc    XXH3_scrambleAcc_sse2
+#define XXH3_initCustomSecret XXH3_initCustomSecret_sse2
+
+#elif (XXH_VECTOR == XXH_NEON)
+
+#define XXH3_accumulate_512 XXH3_accumulate_512_neon
+#define XXH3_scrambleAcc    XXH3_scrambleAcc_neon
+#define XXH3_initCustomSecret XXH3_initCustomSecret_scalar
+
+#elif (XXH_VECTOR == XXH_VSX)
+
+#define XXH3_accumulate_512 XXH3_accumulate_512_vsx
+#define XXH3_scrambleAcc    XXH3_scrambleAcc_vsx
+#define XXH3_initCustomSecret XXH3_initCustomSecret_scalar
+
+#else /* scalar */
+
+#define XXH3_accumulate_512 XXH3_accumulate_512_scalar
+#define XXH3_scrambleAcc    XXH3_scrambleAcc_scalar
+#define XXH3_initCustomSecret XXH3_initCustomSecret_scalar
+
+#endif
+
+
+
+#ifndef XXH_PREFETCH_DIST
+#  ifdef __clang__
+#    define XXH_PREFETCH_DIST 320
+#  else
+#    if (XXH_VECTOR == XXH_AVX512)
+#      define XXH_PREFETCH_DIST 512
+#    else
+#      define XXH_PREFETCH_DIST 384
+#    endif
+#  endif  /* __clang__ */
+#endif  /* XXH_PREFETCH_DIST */
+
+/*
+ * XXH3_accumulate()
+ * Loops over XXH3_accumulate_512().
+ * Assumption: nbStripes will not overflow the secret size
+ */
+XXH_FORCE_INLINE void
+XXH3_accumulate(     xxh_u64* XXH_RESTRICT acc,
+                const xxh_u8* XXH_RESTRICT input,
+                const xxh_u8* XXH_RESTRICT secret,
+                      size_t nbStripes,
+                      XXH3_f_accumulate_512 f_acc512)
+{
+    size_t n;
+    for (n = 0; n < nbStripes; n++ ) {
+        const xxh_u8* const in = input + n*XXH_STRIPE_LEN;
+        XXH_PREFETCH(in + XXH_PREFETCH_DIST);
+        f_acc512(acc,
+                 in,
+                 secret + n*XXH_SECRET_CONSUME_RATE);
+    }
+}
+
+XXH_FORCE_INLINE void
+XXH3_hashLong_internal_loop(xxh_u64* XXH_RESTRICT acc,
+                      const xxh_u8* XXH_RESTRICT input, size_t len,
+                      const xxh_u8* XXH_RESTRICT secret, size_t secretSize,
+                            XXH3_f_accumulate_512 f_acc512,
+                            XXH3_f_scrambleAcc f_scramble)
+{
+    size_t const nbStripesPerBlock = (secretSize - XXH_STRIPE_LEN) / XXH_SECRET_CONSUME_RATE;
+    size_t const block_len = XXH_STRIPE_LEN * nbStripesPerBlock;
+    size_t const nb_blocks = (len - 1) / block_len;
+
+    size_t n;
+
+    XXH_ASSERT(secretSize >= XXH3_SECRET_SIZE_MIN);
+
+    for (n = 0; n < nb_blocks; n++) {
+        XXH3_accumulate(acc, input + n*block_len, secret, nbStripesPerBlock, f_acc512);
+        f_scramble(acc, secret + secretSize - XXH_STRIPE_LEN);
+    }
+
+    /* last partial block */
+    XXH_ASSERT(len > XXH_STRIPE_LEN);
+    {   size_t const nbStripes = ((len - 1) - (block_len * nb_blocks)) / XXH_STRIPE_LEN;
+        XXH_ASSERT(nbStripes <= (secretSize / XXH_SECRET_CONSUME_RATE));
+        XXH3_accumulate(acc, input + nb_blocks*block_len, secret, nbStripes, f_acc512);
+
+        /* last stripe */
+        {   const xxh_u8* const p = input + len - XXH_STRIPE_LEN;
+#define XXH_SECRET_LASTACC_START 7  /* not aligned on 8, last secret is different from acc & scrambler */
+            f_acc512(acc, p, secret + secretSize - XXH_STRIPE_LEN - XXH_SECRET_LASTACC_START);
+    }   }
+}
+
+XXH_FORCE_INLINE xxh_u64
+XXH3_mix2Accs(const xxh_u64* XXH_RESTRICT acc, const xxh_u8* XXH_RESTRICT secret)
+{
+    return XXH3_mul128_fold64(
+               acc[0] ^ XXH_readLE64(secret),
+               acc[1] ^ XXH_readLE64(secret+8) );
+}
+
+static XXH64_hash_t
+XXH3_mergeAccs(const xxh_u64* XXH_RESTRICT acc, const xxh_u8* XXH_RESTRICT secret, xxh_u64 start)
+{
+    xxh_u64 result64 = start;
+    size_t i = 0;
+
+    for (i = 0; i < 4; i++) {
+        result64 += XXH3_mix2Accs(acc+2*i, secret + 16*i);
+#if defined(__clang__)                                /* Clang */ \
+    && (defined(__arm__) || defined(__thumb__))       /* ARMv7 */ \
+    && (defined(__ARM_NEON) || defined(__ARM_NEON__)) /* NEON */  \
+    && !defined(XXH_ENABLE_AUTOVECTORIZE)             /* Define to disable */
+        /*
+         * UGLY HACK:
+         * Prevent autovectorization on Clang ARMv7-a. Exact same problem as
+         * the one in XXH3_len_129to240_64b. Speeds up shorter keys > 240b.
+         * XXH3_64bits, len == 256, Snapdragon 835:
+         *   without hack: 2063.7 MB/s
+         *   with hack:    2560.7 MB/s
+         */
+        XXH_COMPILER_GUARD(result64);
+#endif
+    }
+
+    return XXH3_avalanche(result64);
+}
+
+#define XXH3_INIT_ACC { XXH_PRIME32_3, XXH_PRIME64_1, XXH_PRIME64_2, XXH_PRIME64_3, \
+                        XXH_PRIME64_4, XXH_PRIME32_2, XXH_PRIME64_5, XXH_PRIME32_1 }
+
+XXH_FORCE_INLINE XXH64_hash_t
+XXH3_hashLong_64b_internal(const void* XXH_RESTRICT input, size_t len,
+                           const void* XXH_RESTRICT secret, size_t secretSize,
+                           XXH3_f_accumulate_512 f_acc512,
+                           XXH3_f_scrambleAcc f_scramble)
+{
+    XXH_ALIGN(XXH_ACC_ALIGN) xxh_u64 acc[XXH_ACC_NB] = XXH3_INIT_ACC;
+
+    XXH3_hashLong_internal_loop(acc, (const xxh_u8*)input, len, (const xxh_u8*)secret, secretSize, f_acc512, f_scramble);
+
+    /* converge into final hash */
+    XXH_STATIC_ASSERT(sizeof(acc) == 64);
+    /* do not align on 8, so that the secret is different from the accumulator */
+#define XXH_SECRET_MERGEACCS_START 11
+    XXH_ASSERT(secretSize >= sizeof(acc) + XXH_SECRET_MERGEACCS_START);
+    return XXH3_mergeAccs(acc, (const xxh_u8*)secret + XXH_SECRET_MERGEACCS_START, (xxh_u64)len * XXH_PRIME64_1);
+}
+
+/*
+ * It's important for performance to transmit secret's size (when it's static)
+ * so that the compiler can properly optimize the vectorized loop.
+ * This makes a big performance difference for "medium" keys (<1 KB) when using AVX instruction set.
+ */
+XXH_FORCE_INLINE XXH64_hash_t
+XXH3_hashLong_64b_withSecret(const void* XXH_RESTRICT input, size_t len,
+                             XXH64_hash_t seed64, const xxh_u8* XXH_RESTRICT secret, size_t secretLen)
+{
+    (void)seed64;
+    return XXH3_hashLong_64b_internal(input, len, secret, secretLen, XXH3_accumulate_512, XXH3_scrambleAcc);
+}
+
+/*
+ * It's preferable for performance that XXH3_hashLong is not inlined,
+ * as it results in a smaller function for small data, easier to the instruction cache.
+ * Note that inside this no_inline function, we do inline the internal loop,
+ * and provide a statically defined secret size to allow optimization of vector loop.
+ */
+XXH_NO_INLINE XXH64_hash_t
+XXH3_hashLong_64b_default(const void* XXH_RESTRICT input, size_t len,
+                          XXH64_hash_t seed64, const xxh_u8* XXH_RESTRICT secret, size_t secretLen)
+{
+    (void)seed64; (void)secret; (void)secretLen;
+    return XXH3_hashLong_64b_internal(input, len, XXH3_kSecret, sizeof(XXH3_kSecret), XXH3_accumulate_512, XXH3_scrambleAcc);
+}
+
+/*
+ * XXH3_hashLong_64b_withSeed():
+ * Generate a custom key based on alteration of default XXH3_kSecret with the seed,
+ * and then use this key for long mode hashing.
+ *
+ * This operation is decently fast but nonetheless costs a little bit of time.
+ * Try to avoid it whenever possible (typically when seed==0).
+ *
+ * It's important for performance that XXH3_hashLong is not inlined. Not sure
+ * why (uop cache maybe?), but the difference is large and easily measurable.
+ */
+XXH_FORCE_INLINE XXH64_hash_t
+XXH3_hashLong_64b_withSeed_internal(const void* input, size_t len,
+                                    XXH64_hash_t seed,
+                                    XXH3_f_accumulate_512 f_acc512,
+                                    XXH3_f_scrambleAcc f_scramble,
+                                    XXH3_f_initCustomSecret f_initSec)
+{
+    if (seed == 0)
+        return XXH3_hashLong_64b_internal(input, len,
+                                          XXH3_kSecret, sizeof(XXH3_kSecret),
+                                          f_acc512, f_scramble);
+    {   XXH_ALIGN(XXH_SEC_ALIGN) xxh_u8 secret[XXH_SECRET_DEFAULT_SIZE];
+        f_initSec(secret, seed);
+        return XXH3_hashLong_64b_internal(input, len, secret, sizeof(secret),
+                                          f_acc512, f_scramble);
+    }
+}
+
+/*
+ * It's important for performance that XXH3_hashLong is not inlined.
+ */
+XXH_NO_INLINE XXH64_hash_t
+XXH3_hashLong_64b_withSeed(const void* input, size_t len,
+                           XXH64_hash_t seed, const xxh_u8* secret, size_t secretLen)
+{
+    (void)secret; (void)secretLen;
+    return XXH3_hashLong_64b_withSeed_internal(input, len, seed,
+                XXH3_accumulate_512, XXH3_scrambleAcc, XXH3_initCustomSecret);
+}
+
+
+typedef XXH64_hash_t (*XXH3_hashLong64_f)(const void* XXH_RESTRICT, size_t,
+                                          XXH64_hash_t, const xxh_u8* XXH_RESTRICT, size_t);
+
+XXH_FORCE_INLINE XXH64_hash_t
+XXH3_64bits_internal(const void* XXH_RESTRICT input, size_t len,
+                     XXH64_hash_t seed64, const void* XXH_RESTRICT secret, size_t secretLen,
+                     XXH3_hashLong64_f f_hashLong)
+{
+    XXH_ASSERT(secretLen >= XXH3_SECRET_SIZE_MIN);
+    /*
+     * If an action is to be taken if `secretLen` condition is not respected,
+     * it should be done here.
+     * For now, it's a contract pre-condition.
+     * Adding a check and a branch here would cost performance at every hash.
+     * Also, note that function signature doesn't offer room to return an error.
+     */
+    if (len <= 16)
+        return XXH3_len_0to16_64b((const xxh_u8*)input, len, (const xxh_u8*)secret, seed64);
+    if (len <= 128)
+        return XXH3_len_17to128_64b((const xxh_u8*)input, len, (const xxh_u8*)secret, secretLen, seed64);
+    if (len <= XXH3_MIDSIZE_MAX)
+        return XXH3_len_129to240_64b((const xxh_u8*)input, len, (const xxh_u8*)secret, secretLen, seed64);
+    return f_hashLong(input, len, seed64, (const xxh_u8*)secret, secretLen);
+}
+
+
+/* ===   Public entry point   === */
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH64_hash_t XXH3_64bits(const void* input, size_t len)
+{
+    return XXH3_64bits_internal(input, len, 0, XXH3_kSecret, sizeof(XXH3_kSecret), XXH3_hashLong_64b_default);
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH64_hash_t
+XXH3_64bits_withSecret(const void* input, size_t len, const void* secret, size_t secretSize)
+{
+    return XXH3_64bits_internal(input, len, 0, secret, secretSize, XXH3_hashLong_64b_withSecret);
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH64_hash_t
+XXH3_64bits_withSeed(const void* input, size_t len, XXH64_hash_t seed)
+{
+    return XXH3_64bits_internal(input, len, seed, XXH3_kSecret, sizeof(XXH3_kSecret), XXH3_hashLong_64b_withSeed);
+}
+
+XXH_PUBLIC_API XXH64_hash_t
+XXH3_64bits_withSecretandSeed(const void* input, size_t len, const void* secret, size_t secretSize, XXH64_hash_t seed)
+{
+    if (len <= XXH3_MIDSIZE_MAX)
+        return XXH3_64bits_internal(input, len, seed, XXH3_kSecret, sizeof(XXH3_kSecret), NULL);
+    return XXH3_hashLong_64b_withSecret(input, len, seed, (const xxh_u8*)secret, secretSize);
+}
+
+
+/* ===   XXH3 streaming   === */
+
+/*
+ * Malloc's a pointer that is always aligned to align.
+ *
+ * This must be freed with `XXH_alignedFree()`.
+ *
+ * malloc typically guarantees 16 byte alignment on 64-bit systems and 8 byte
+ * alignment on 32-bit. This isn't enough for the 32 byte aligned loads in AVX2
+ * or on 32-bit, the 16 byte aligned loads in SSE2 and NEON.
+ *
+ * This underalignment previously caused a rather obvious crash which went
+ * completely unnoticed due to XXH3_createState() not actually being tested.
+ * Credit to RedSpah for noticing this bug.
+ *
+ * The alignment is done manually: Functions like posix_memalign or _mm_malloc
+ * are avoided: To maintain portability, we would have to write a fallback
+ * like this anyways, and besides, testing for the existence of library
+ * functions without relying on external build tools is impossible.
+ *
+ * The method is simple: Overallocate, manually align, and store the offset
+ * to the original behind the returned pointer.
+ *
+ * Align must be a power of 2 and 8 <= align <= 128.
+ */
+static void* XXH_alignedMalloc(size_t s, size_t align)
+{
+    XXH_ASSERT(align <= 128 && align >= 8); /* range check */
+    XXH_ASSERT((align & (align-1)) == 0);   /* power of 2 */
+    XXH_ASSERT(s != 0 && s < (s + align));  /* empty/overflow */
+    {   /* Overallocate to make room for manual realignment and an offset byte */
+        xxh_u8* base = (xxh_u8*)XXH_malloc(s + align);
+        if (base != NULL) {
+            /*
+             * Get the offset needed to align this pointer.
+             *
+             * Even if the returned pointer is aligned, there will always be
+             * at least one byte to store the offset to the original pointer.
+             */
+            size_t offset = align - ((size_t)base & (align - 1)); /* base % align */
+            /* Add the offset for the now-aligned pointer */
+            xxh_u8* ptr = base + offset;
+
+            XXH_ASSERT((size_t)ptr % align == 0);
+
+            /* Store the offset immediately before the returned pointer. */
+            ptr[-1] = (xxh_u8)offset;
+            return ptr;
+        }
+        return NULL;
+    }
+}
+/*
+ * Frees an aligned pointer allocated by XXH_alignedMalloc(). Don't pass
+ * normal malloc'd pointers, XXH_alignedMalloc has a specific data layout.
+ */
+static void XXH_alignedFree(void* p)
+{
+    if (p != NULL) {
+        xxh_u8* ptr = (xxh_u8*)p;
+        /* Get the offset byte we added in XXH_malloc. */
+        xxh_u8 offset = ptr[-1];
+        /* Free the original malloc'd pointer */
+        xxh_u8* base = ptr - offset;
+        XXH_free(base);
+    }
+}
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH3_state_t* XXH3_createState(void)
+{
+    XXH3_state_t* const state = (XXH3_state_t*)XXH_alignedMalloc(sizeof(XXH3_state_t), 64);
+    if (state==NULL) return NULL;
+    XXH3_INITSTATE(state);
+    return state;
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH_errorcode XXH3_freeState(XXH3_state_t* statePtr)
+{
+    XXH_alignedFree(statePtr);
+    return XXH_OK;
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API void
+XXH3_copyState(XXH3_state_t* dst_state, const XXH3_state_t* src_state)
+{
+    XXH_memcpy(dst_state, src_state, sizeof(*dst_state));
+}
+
+static void
+XXH3_reset_internal(XXH3_state_t* statePtr,
+                    XXH64_hash_t seed,
+                    const void* secret, size_t secretSize)
+{
+    size_t const initStart = offsetof(XXH3_state_t, bufferedSize);
+    size_t const initLength = offsetof(XXH3_state_t, nbStripesPerBlock) - initStart;
+    XXH_ASSERT(offsetof(XXH3_state_t, nbStripesPerBlock) > initStart);
+    XXH_ASSERT(statePtr != NULL);
+    /* set members from bufferedSize to nbStripesPerBlock (excluded) to 0 */
+    memset((char*)statePtr + initStart, 0, initLength);
+    statePtr->acc[0] = XXH_PRIME32_3;
+    statePtr->acc[1] = XXH_PRIME64_1;
+    statePtr->acc[2] = XXH_PRIME64_2;
+    statePtr->acc[3] = XXH_PRIME64_3;
+    statePtr->acc[4] = XXH_PRIME64_4;
+    statePtr->acc[5] = XXH_PRIME32_2;
+    statePtr->acc[6] = XXH_PRIME64_5;
+    statePtr->acc[7] = XXH_PRIME32_1;
+    statePtr->seed = seed;
+    statePtr->useSeed = (seed != 0);
+    statePtr->extSecret = (const unsigned char*)secret;
+    XXH_ASSERT(secretSize >= XXH3_SECRET_SIZE_MIN);
+    statePtr->secretLimit = secretSize - XXH_STRIPE_LEN;
+    statePtr->nbStripesPerBlock = statePtr->secretLimit / XXH_SECRET_CONSUME_RATE;
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH_errorcode
+XXH3_64bits_reset(XXH3_state_t* statePtr)
+{
+    if (statePtr == NULL) return XXH_ERROR;
+    XXH3_reset_internal(statePtr, 0, XXH3_kSecret, XXH_SECRET_DEFAULT_SIZE);
+    return XXH_OK;
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH_errorcode
+XXH3_64bits_reset_withSecret(XXH3_state_t* statePtr, const void* secret, size_t secretSize)
+{
+    if (statePtr == NULL) return XXH_ERROR;
+    XXH3_reset_internal(statePtr, 0, secret, secretSize);
+    if (secret == NULL) return XXH_ERROR;
+    if (secretSize < XXH3_SECRET_SIZE_MIN) return XXH_ERROR;
+    return XXH_OK;
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH_errorcode
+XXH3_64bits_reset_withSeed(XXH3_state_t* statePtr, XXH64_hash_t seed)
+{
+    if (statePtr == NULL) return XXH_ERROR;
+    if (seed==0) return XXH3_64bits_reset(statePtr);
+    if ((seed != statePtr->seed) || (statePtr->extSecret != NULL))
+        XXH3_initCustomSecret(statePtr->customSecret, seed);
+    XXH3_reset_internal(statePtr, seed, NULL, XXH_SECRET_DEFAULT_SIZE);
+    return XXH_OK;
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH_errorcode
+XXH3_64bits_reset_withSecretandSeed(XXH3_state_t* statePtr, const void* secret, size_t secretSize, XXH64_hash_t seed64)
+{
+    if (statePtr == NULL) return XXH_ERROR;
+    if (secret == NULL) return XXH_ERROR;
+    if (secretSize < XXH3_SECRET_SIZE_MIN) return XXH_ERROR;
+    XXH3_reset_internal(statePtr, seed64, secret, secretSize);
+    statePtr->useSeed = 1; /* always, even if seed64==0 */
+    return XXH_OK;
+}
+
+/* Note : when XXH3_consumeStripes() is invoked,
+ * there must be a guarantee that at least one more byte must be consumed from input
+ * so that the function can blindly consume all stripes using the "normal" secret segment */
+XXH_FORCE_INLINE void
+XXH3_consumeStripes(xxh_u64* XXH_RESTRICT acc,
+                    size_t* XXH_RESTRICT nbStripesSoFarPtr, size_t nbStripesPerBlock,
+                    const xxh_u8* XXH_RESTRICT input, size_t nbStripes,
+                    const xxh_u8* XXH_RESTRICT secret, size_t secretLimit,
+                    XXH3_f_accumulate_512 f_acc512,
+                    XXH3_f_scrambleAcc f_scramble)
+{
+    XXH_ASSERT(nbStripes <= nbStripesPerBlock);  /* can handle max 1 scramble per invocation */
+    XXH_ASSERT(*nbStripesSoFarPtr < nbStripesPerBlock);
+    if (nbStripesPerBlock - *nbStripesSoFarPtr <= nbStripes) {
+        /* need a scrambling operation */
+        size_t const nbStripesToEndofBlock = nbStripesPerBlock - *nbStripesSoFarPtr;
+        size_t const nbStripesAfterBlock = nbStripes - nbStripesToEndofBlock;
+        XXH3_accumulate(acc, input, secret + nbStripesSoFarPtr[0] * XXH_SECRET_CONSUME_RATE, nbStripesToEndofBlock, f_acc512);
+        f_scramble(acc, secret + secretLimit);
+        XXH3_accumulate(acc, input + nbStripesToEndofBlock * XXH_STRIPE_LEN, secret, nbStripesAfterBlock, f_acc512);
+        *nbStripesSoFarPtr = nbStripesAfterBlock;
+    } else {
+        XXH3_accumulate(acc, input, secret + nbStripesSoFarPtr[0] * XXH_SECRET_CONSUME_RATE, nbStripes, f_acc512);
+        *nbStripesSoFarPtr += nbStripes;
+    }
+}
+
+#ifndef XXH3_STREAM_USE_STACK
+# ifndef __clang__ /* clang doesn't need additional stack space */
+#   define XXH3_STREAM_USE_STACK 1
+# endif
+#endif
+/*
+ * Both XXH3_64bits_update and XXH3_128bits_update use this routine.
+ */
+XXH_FORCE_INLINE XXH_errorcode
+XXH3_update(XXH3_state_t* XXH_RESTRICT const state,
+            const xxh_u8* XXH_RESTRICT input, size_t len,
+            XXH3_f_accumulate_512 f_acc512,
+            XXH3_f_scrambleAcc f_scramble)
+{
+    if (input==NULL) {
+        XXH_ASSERT(len == 0);
+        return XXH_OK;
+    }
+
+    XXH_ASSERT(state != NULL);
+    {   const xxh_u8* const bEnd = input + len;
+        const unsigned char* const secret = (state->extSecret == NULL) ? state->customSecret : state->extSecret;
+#if defined(XXH3_STREAM_USE_STACK) && XXH3_STREAM_USE_STACK >= 1
+        /* For some reason, gcc and MSVC seem to suffer greatly
+         * when operating accumulators directly into state.
+         * Operating into stack space seems to enable proper optimization.
+         * clang, on the other hand, doesn't seem to need this trick */
+        XXH_ALIGN(XXH_ACC_ALIGN) xxh_u64 acc[8]; memcpy(acc, state->acc, sizeof(acc));
+#else
+        xxh_u64* XXH_RESTRICT const acc = state->acc;
+#endif
+        state->totalLen += len;
+        XXH_ASSERT(state->bufferedSize <= XXH3_INTERNALBUFFER_SIZE);
+
+        /* small input : just fill in tmp buffer */
+        if (state->bufferedSize + len <= XXH3_INTERNALBUFFER_SIZE) {
+            XXH_memcpy(state->buffer + state->bufferedSize, input, len);
+            state->bufferedSize += (XXH32_hash_t)len;
+            return XXH_OK;
+        }
+
+        /* total input is now > XXH3_INTERNALBUFFER_SIZE */
+        #define XXH3_INTERNALBUFFER_STRIPES (XXH3_INTERNALBUFFER_SIZE / XXH_STRIPE_LEN)
+        XXH_STATIC_ASSERT(XXH3_INTERNALBUFFER_SIZE % XXH_STRIPE_LEN == 0);   /* clean multiple */
+
+        /*
+         * Internal buffer is partially filled (always, except at beginning)
+         * Complete it, then consume it.
+         */
+        if (state->bufferedSize) {
+            size_t const loadSize = XXH3_INTERNALBUFFER_SIZE - state->bufferedSize;
+            XXH_memcpy(state->buffer + state->bufferedSize, input, loadSize);
+            input += loadSize;
+            XXH3_consumeStripes(acc,
+                               &state->nbStripesSoFar, state->nbStripesPerBlock,
+                                state->buffer, XXH3_INTERNALBUFFER_STRIPES,
+                                secret, state->secretLimit,
+                                f_acc512, f_scramble);
+            state->bufferedSize = 0;
+        }
+        XXH_ASSERT(input < bEnd);
+
+        /* large input to consume : ingest per full block */
+        if ((size_t)(bEnd - input) > state->nbStripesPerBlock * XXH_STRIPE_LEN) {
+            size_t nbStripes = (size_t)(bEnd - 1 - input) / XXH_STRIPE_LEN;
+            XXH_ASSERT(state->nbStripesPerBlock >= state->nbStripesSoFar);
+            /* join to current block's end */
+            {   size_t const nbStripesToEnd = state->nbStripesPerBlock - state->nbStripesSoFar;
+                XXH_ASSERT(nbStripes <= nbStripes);
+                XXH3_accumulate(acc, input, secret + state->nbStripesSoFar * XXH_SECRET_CONSUME_RATE, nbStripesToEnd, f_acc512);
+                f_scramble(acc, secret + state->secretLimit);
+                state->nbStripesSoFar = 0;
+                input += nbStripesToEnd * XXH_STRIPE_LEN;
+                nbStripes -= nbStripesToEnd;
+            }
+            /* consume per entire blocks */
+            while(nbStripes >= state->nbStripesPerBlock) {
+                XXH3_accumulate(acc, input, secret, state->nbStripesPerBlock, f_acc512);
+                f_scramble(acc, secret + state->secretLimit);
+                input += state->nbStripesPerBlock * XXH_STRIPE_LEN;
+                nbStripes -= state->nbStripesPerBlock;
+            }
+            /* consume last partial block */
+            XXH3_accumulate(acc, input, secret, nbStripes, f_acc512);
+            input += nbStripes * XXH_STRIPE_LEN;
+            XXH_ASSERT(input < bEnd);  /* at least some bytes left */
+            state->nbStripesSoFar = nbStripes;
+            /* buffer predecessor of last partial stripe */
+            XXH_memcpy(state->buffer + sizeof(state->buffer) - XXH_STRIPE_LEN, input - XXH_STRIPE_LEN, XXH_STRIPE_LEN);
+            XXH_ASSERT(bEnd - input <= XXH_STRIPE_LEN);
+        } else {
+            /* content to consume <= block size */
+            /* Consume input by a multiple of internal buffer size */
+            if (bEnd - input > XXH3_INTERNALBUFFER_SIZE) {
+                const xxh_u8* const limit = bEnd - XXH3_INTERNALBUFFER_SIZE;
+                do {
+                    XXH3_consumeStripes(acc,
+                                       &state->nbStripesSoFar, state->nbStripesPerBlock,
+                                        input, XXH3_INTERNALBUFFER_STRIPES,
+                                        secret, state->secretLimit,
+                                        f_acc512, f_scramble);
+                    input += XXH3_INTERNALBUFFER_SIZE;
+                } while (input<limit);
+                /* buffer predecessor of last partial stripe */
+                XXH_memcpy(state->buffer + sizeof(state->buffer) - XXH_STRIPE_LEN, input - XXH_STRIPE_LEN, XXH_STRIPE_LEN);
+            }
+        }
+
+        /* Some remaining input (always) : buffer it */
+        XXH_ASSERT(input < bEnd);
+        XXH_ASSERT(bEnd - input <= XXH3_INTERNALBUFFER_SIZE);
+        XXH_ASSERT(state->bufferedSize == 0);
+        XXH_memcpy(state->buffer, input, (size_t)(bEnd-input));
+        state->bufferedSize = (XXH32_hash_t)(bEnd-input);
+#if defined(XXH3_STREAM_USE_STACK) && XXH3_STREAM_USE_STACK >= 1
+        /* save stack accumulators into state */
+        memcpy(state->acc, acc, sizeof(acc));
+#endif
+    }
+
+    return XXH_OK;
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH_errorcode
+XXH3_64bits_update(XXH3_state_t* state, const void* input, size_t len)
+{
+    return XXH3_update(state, (const xxh_u8*)input, len,
+                       XXH3_accumulate_512, XXH3_scrambleAcc);
+}
+
+
+XXH_FORCE_INLINE void
+XXH3_digest_long (XXH64_hash_t* acc,
+                  const XXH3_state_t* state,
+                  const unsigned char* secret)
+{
+    /*
+     * Digest on a local copy. This way, the state remains unaltered, and it can
+     * continue ingesting more input afterwards.
+     */
+    XXH_memcpy(acc, state->acc, sizeof(state->acc));
+    if (state->bufferedSize >= XXH_STRIPE_LEN) {
+        size_t const nbStripes = (state->bufferedSize - 1) / XXH_STRIPE_LEN;
+        size_t nbStripesSoFar = state->nbStripesSoFar;
+        XXH3_consumeStripes(acc,
+                           &nbStripesSoFar, state->nbStripesPerBlock,
+                            state->buffer, nbStripes,
+                            secret, state->secretLimit,
+                            XXH3_accumulate_512, XXH3_scrambleAcc);
+        /* last stripe */
+        XXH3_accumulate_512(acc,
+                            state->buffer + state->bufferedSize - XXH_STRIPE_LEN,
+                            secret + state->secretLimit - XXH_SECRET_LASTACC_START);
+    } else {  /* bufferedSize < XXH_STRIPE_LEN */
+        xxh_u8 lastStripe[XXH_STRIPE_LEN];
+        size_t const catchupSize = XXH_STRIPE_LEN - state->bufferedSize;
+        XXH_ASSERT(state->bufferedSize > 0);  /* there is always some input buffered */
+        XXH_memcpy(lastStripe, state->buffer + sizeof(state->buffer) - catchupSize, catchupSize);
+        XXH_memcpy(lastStripe + catchupSize, state->buffer, state->bufferedSize);
+        XXH3_accumulate_512(acc,
+                            lastStripe,
+                            secret + state->secretLimit - XXH_SECRET_LASTACC_START);
+    }
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH64_hash_t XXH3_64bits_digest (const XXH3_state_t* state)
+{
+    const unsigned char* const secret = (state->extSecret == NULL) ? state->customSecret : state->extSecret;
+    if (state->totalLen > XXH3_MIDSIZE_MAX) {
+        XXH_ALIGN(XXH_ACC_ALIGN) XXH64_hash_t acc[XXH_ACC_NB];
+        XXH3_digest_long(acc, state, secret);
+        return XXH3_mergeAccs(acc,
+                              secret + XXH_SECRET_MERGEACCS_START,
+                              (xxh_u64)state->totalLen * XXH_PRIME64_1);
+    }
+    /* totalLen <= XXH3_MIDSIZE_MAX: digesting a short input */
+    if (state->useSeed)
+        return XXH3_64bits_withSeed(state->buffer, (size_t)state->totalLen, state->seed);
+    return XXH3_64bits_withSecret(state->buffer, (size_t)(state->totalLen),
+                                  secret, state->secretLimit + XXH_STRIPE_LEN);
+}
+
+
+
+/* ==========================================
+ * XXH3 128 bits (a.k.a XXH128)
+ * ==========================================
+ * XXH3's 128-bit variant has better mixing and strength than the 64-bit variant,
+ * even without counting the significantly larger output size.
+ *
+ * For example, extra steps are taken to avoid the seed-dependent collisions
+ * in 17-240 byte inputs (See XXH3_mix16B and XXH128_mix32B).
+ *
+ * This strength naturally comes at the cost of some speed, especially on short
+ * lengths. Note that longer hashes are about as fast as the 64-bit version
+ * due to it using only a slight modification of the 64-bit loop.
+ *
+ * XXH128 is also more oriented towards 64-bit machines. It is still extremely
+ * fast for a _128-bit_ hash on 32-bit (it usually clears XXH64).
+ */
+
+XXH_FORCE_INLINE XXH128_hash_t
+XXH3_len_1to3_128b(const xxh_u8* input, size_t len, const xxh_u8* secret, XXH64_hash_t seed)
+{
+    /* A doubled version of 1to3_64b with different constants. */
+    XXH_ASSERT(input != NULL);
+    XXH_ASSERT(1 <= len && len <= 3);
+    XXH_ASSERT(secret != NULL);
+    /*
+     * len = 1: combinedl = { input[0], 0x01, input[0], input[0] }
+     * len = 2: combinedl = { input[1], 0x02, input[0], input[1] }
+     * len = 3: combinedl = { input[2], 0x03, input[0], input[1] }
+     */
+    {   xxh_u8 const c1 = input[0];
+        xxh_u8 const c2 = input[len >> 1];
+        xxh_u8 const c3 = input[len - 1];
+        xxh_u32 const combinedl = ((xxh_u32)c1 <<16) | ((xxh_u32)c2 << 24)
+                                | ((xxh_u32)c3 << 0) | ((xxh_u32)len << 8);
+        xxh_u32 const combinedh = XXH_rotl32(XXH_swap32(combinedl), 13);
+        xxh_u64 const bitflipl = (XXH_readLE32(secret) ^ XXH_readLE32(secret+4)) + seed;
+        xxh_u64 const bitfliph = (XXH_readLE32(secret+8) ^ XXH_readLE32(secret+12)) - seed;
+        xxh_u64 const keyed_lo = (xxh_u64)combinedl ^ bitflipl;
+        xxh_u64 const keyed_hi = (xxh_u64)combinedh ^ bitfliph;
+        XXH128_hash_t h128;
+        h128.low64  = XXH64_avalanche(keyed_lo);
+        h128.high64 = XXH64_avalanche(keyed_hi);
+        return h128;
+    }
+}
+
+XXH_FORCE_INLINE XXH128_hash_t
+XXH3_len_4to8_128b(const xxh_u8* input, size_t len, const xxh_u8* secret, XXH64_hash_t seed)
+{
+    XXH_ASSERT(input != NULL);
+    XXH_ASSERT(secret != NULL);
+    XXH_ASSERT(4 <= len && len <= 8);
+    seed ^= (xxh_u64)XXH_swap32((xxh_u32)seed) << 32;
+    {   xxh_u32 const input_lo = XXH_readLE32(input);
+        xxh_u32 const input_hi = XXH_readLE32(input + len - 4);
+        xxh_u64 const input_64 = input_lo + ((xxh_u64)input_hi << 32);
+        xxh_u64 const bitflip = (XXH_readLE64(secret+16) ^ XXH_readLE64(secret+24)) + seed;
+        xxh_u64 const keyed = input_64 ^ bitflip;
+
+        /* Shift len to the left to ensure it is even, this avoids even multiplies. */
+        XXH128_hash_t m128 = XXH_mult64to128(keyed, XXH_PRIME64_1 + (len << 2));
+
+        m128.high64 += (m128.low64 << 1);
+        m128.low64  ^= (m128.high64 >> 3);
+
+        m128.low64   = XXH_xorshift64(m128.low64, 35);
+        m128.low64  *= 0x9FB21C651E98DF25ULL;
+        m128.low64   = XXH_xorshift64(m128.low64, 28);
+        m128.high64  = XXH3_avalanche(m128.high64);
+        return m128;
+    }
+}
+
+XXH_FORCE_INLINE XXH128_hash_t
+XXH3_len_9to16_128b(const xxh_u8* input, size_t len, const xxh_u8* secret, XXH64_hash_t seed)
+{
+    XXH_ASSERT(input != NULL);
+    XXH_ASSERT(secret != NULL);
+    XXH_ASSERT(9 <= len && len <= 16);
+    {   xxh_u64 const bitflipl = (XXH_readLE64(secret+32) ^ XXH_readLE64(secret+40)) - seed;
+        xxh_u64 const bitfliph = (XXH_readLE64(secret+48) ^ XXH_readLE64(secret+56)) + seed;
+        xxh_u64 const input_lo = XXH_readLE64(input);
+        xxh_u64       input_hi = XXH_readLE64(input + len - 8);
+        XXH128_hash_t m128 = XXH_mult64to128(input_lo ^ input_hi ^ bitflipl, XXH_PRIME64_1);
+        /*
+         * Put len in the middle of m128 to ensure that the length gets mixed to
+         * both the low and high bits in the 128x64 multiply below.
+         */
+        m128.low64 += (xxh_u64)(len - 1) << 54;
+        input_hi   ^= bitfliph;
+        /*
+         * Add the high 32 bits of input_hi to the high 32 bits of m128, then
+         * add the long product of the low 32 bits of input_hi and XXH_PRIME32_2 to
+         * the high 64 bits of m128.
+         *
+         * The best approach to this operation is different on 32-bit and 64-bit.
+         */
+        if (sizeof(void *) < sizeof(xxh_u64)) { /* 32-bit */
+            /*
+             * 32-bit optimized version, which is more readable.
+             *
+             * On 32-bit, it removes an ADC and delays a dependency between the two
+             * halves of m128.high64, but it generates an extra mask on 64-bit.
+             */
+            m128.high64 += (input_hi & 0xFFFFFFFF00000000ULL) + XXH_mult32to64((xxh_u32)input_hi, XXH_PRIME32_2);
+        } else {
+            /*
+             * 64-bit optimized (albeit more confusing) version.
+             *
+             * Uses some properties of addition and multiplication to remove the mask:
+             *
+             * Let:
+             *    a = input_hi.lo = (input_hi & 0x00000000FFFFFFFF)
+             *    b = input_hi.hi = (input_hi & 0xFFFFFFFF00000000)
+             *    c = XXH_PRIME32_2
+             *
+             *    a + (b * c)
+             * Inverse Property: x + y - x == y
+             *    a + (b * (1 + c - 1))
+             * Distributive Property: x * (y + z) == (x * y) + (x * z)
+             *    a + (b * 1) + (b * (c - 1))
+             * Identity Property: x * 1 == x
+             *    a + b + (b * (c - 1))
+             *
+             * Substitute a, b, and c:
+             *    input_hi.hi + input_hi.lo + ((xxh_u64)input_hi.lo * (XXH_PRIME32_2 - 1))
+             *
+             * Since input_hi.hi + input_hi.lo == input_hi, we get this:
+             *    input_hi + ((xxh_u64)input_hi.lo * (XXH_PRIME32_2 - 1))
+             */
+            m128.high64 += input_hi + XXH_mult32to64((xxh_u32)input_hi, XXH_PRIME32_2 - 1);
+        }
+        /* m128 ^= XXH_swap64(m128 >> 64); */
+        m128.low64  ^= XXH_swap64(m128.high64);
+
+        {   /* 128x64 multiply: h128 = m128 * XXH_PRIME64_2; */
+            XXH128_hash_t h128 = XXH_mult64to128(m128.low64, XXH_PRIME64_2);
+            h128.high64 += m128.high64 * XXH_PRIME64_2;
+
+            h128.low64   = XXH3_avalanche(h128.low64);
+            h128.high64  = XXH3_avalanche(h128.high64);
+            return h128;
+    }   }
+}
+
+/*
+ * Assumption: `secret` size is >= XXH3_SECRET_SIZE_MIN
+ */
+XXH_FORCE_INLINE XXH128_hash_t
+XXH3_len_0to16_128b(const xxh_u8* input, size_t len, const xxh_u8* secret, XXH64_hash_t seed)
+{
+    XXH_ASSERT(len <= 16);
+    {   if (len > 8) return XXH3_len_9to16_128b(input, len, secret, seed);
+        if (len >= 4) return XXH3_len_4to8_128b(input, len, secret, seed);
+        if (len) return XXH3_len_1to3_128b(input, len, secret, seed);
+        {   XXH128_hash_t h128;
+            xxh_u64 const bitflipl = XXH_readLE64(secret+64) ^ XXH_readLE64(secret+72);
+            xxh_u64 const bitfliph = XXH_readLE64(secret+80) ^ XXH_readLE64(secret+88);
+            h128.low64 = XXH64_avalanche(seed ^ bitflipl);
+            h128.high64 = XXH64_avalanche( seed ^ bitfliph);
+            return h128;
+    }   }
+}
+
+/*
+ * A bit slower than XXH3_mix16B, but handles multiply by zero better.
+ */
+XXH_FORCE_INLINE XXH128_hash_t
+XXH128_mix32B(XXH128_hash_t acc, const xxh_u8* input_1, const xxh_u8* input_2,
+              const xxh_u8* secret, XXH64_hash_t seed)
+{
+    acc.low64  += XXH3_mix16B (input_1, secret+0, seed);
+    acc.low64  ^= XXH_readLE64(input_2) + XXH_readLE64(input_2 + 8);
+    acc.high64 += XXH3_mix16B (input_2, secret+16, seed);
+    acc.high64 ^= XXH_readLE64(input_1) + XXH_readLE64(input_1 + 8);
+    return acc;
+}
+
+
+XXH_FORCE_INLINE XXH128_hash_t
+XXH3_len_17to128_128b(const xxh_u8* XXH_RESTRICT input, size_t len,
+                      const xxh_u8* XXH_RESTRICT secret, size_t secretSize,
+                      XXH64_hash_t seed)
+{
+    XXH_ASSERT(secretSize >= XXH3_SECRET_SIZE_MIN); (void)secretSize;
+    XXH_ASSERT(16 < len && len <= 128);
+
+    {   XXH128_hash_t acc;
+        acc.low64 = len * XXH_PRIME64_1;
+        acc.high64 = 0;
+        if (len > 32) {
+            if (len > 64) {
+                if (len > 96) {
+                    acc = XXH128_mix32B(acc, input+48, input+len-64, secret+96, seed);
+                }
+                acc = XXH128_mix32B(acc, input+32, input+len-48, secret+64, seed);
+            }
+            acc = XXH128_mix32B(acc, input+16, input+len-32, secret+32, seed);
+        }
+        acc = XXH128_mix32B(acc, input, input+len-16, secret, seed);
+        {   XXH128_hash_t h128;
+            h128.low64  = acc.low64 + acc.high64;
+            h128.high64 = (acc.low64    * XXH_PRIME64_1)
+                        + (acc.high64   * XXH_PRIME64_4)
+                        + ((len - seed) * XXH_PRIME64_2);
+            h128.low64  = XXH3_avalanche(h128.low64);
+            h128.high64 = (XXH64_hash_t)0 - XXH3_avalanche(h128.high64);
+            return h128;
+        }
+    }
+}
+
+XXH_NO_INLINE XXH128_hash_t
+XXH3_len_129to240_128b(const xxh_u8* XXH_RESTRICT input, size_t len,
+                       const xxh_u8* XXH_RESTRICT secret, size_t secretSize,
+                       XXH64_hash_t seed)
+{
+    XXH_ASSERT(secretSize >= XXH3_SECRET_SIZE_MIN); (void)secretSize;
+    XXH_ASSERT(128 < len && len <= XXH3_MIDSIZE_MAX);
+
+    {   XXH128_hash_t acc;
+        int const nbRounds = (int)len / 32;
+        int i;
+        acc.low64 = len * XXH_PRIME64_1;
+        acc.high64 = 0;
+        for (i=0; i<4; i++) {
+            acc = XXH128_mix32B(acc,
+                                input  + (32 * i),
+                                input  + (32 * i) + 16,
+                                secret + (32 * i),
+                                seed);
+        }
+        acc.low64 = XXH3_avalanche(acc.low64);
+        acc.high64 = XXH3_avalanche(acc.high64);
+        XXH_ASSERT(nbRounds >= 4);
+        for (i=4 ; i < nbRounds; i++) {
+            acc = XXH128_mix32B(acc,
+                                input + (32 * i),
+                                input + (32 * i) + 16,
+                                secret + XXH3_MIDSIZE_STARTOFFSET + (32 * (i - 4)),
+                                seed);
+        }
+        /* last bytes */
+        acc = XXH128_mix32B(acc,
+                            input + len - 16,
+                            input + len - 32,
+                            secret + XXH3_SECRET_SIZE_MIN - XXH3_MIDSIZE_LASTOFFSET - 16,
+                            0ULL - seed);
+
+        {   XXH128_hash_t h128;
+            h128.low64  = acc.low64 + acc.high64;
+            h128.high64 = (acc.low64    * XXH_PRIME64_1)
+                        + (acc.high64   * XXH_PRIME64_4)
+                        + ((len - seed) * XXH_PRIME64_2);
+            h128.low64  = XXH3_avalanche(h128.low64);
+            h128.high64 = (XXH64_hash_t)0 - XXH3_avalanche(h128.high64);
+            return h128;
+        }
+    }
+}
+
+XXH_FORCE_INLINE XXH128_hash_t
+XXH3_hashLong_128b_internal(const void* XXH_RESTRICT input, size_t len,
+                            const xxh_u8* XXH_RESTRICT secret, size_t secretSize,
+                            XXH3_f_accumulate_512 f_acc512,
+                            XXH3_f_scrambleAcc f_scramble)
+{
+    XXH_ALIGN(XXH_ACC_ALIGN) xxh_u64 acc[XXH_ACC_NB] = XXH3_INIT_ACC;
+
+    XXH3_hashLong_internal_loop(acc, (const xxh_u8*)input, len, secret, secretSize, f_acc512, f_scramble);
+
+    /* converge into final hash */
+    XXH_STATIC_ASSERT(sizeof(acc) == 64);
+    XXH_ASSERT(secretSize >= sizeof(acc) + XXH_SECRET_MERGEACCS_START);
+    {   XXH128_hash_t h128;
+        h128.low64  = XXH3_mergeAccs(acc,
+                                     secret + XXH_SECRET_MERGEACCS_START,
+                                     (xxh_u64)len * XXH_PRIME64_1);
+        h128.high64 = XXH3_mergeAccs(acc,
+                                     secret + secretSize
+                                            - sizeof(acc) - XXH_SECRET_MERGEACCS_START,
+                                     ~((xxh_u64)len * XXH_PRIME64_2));
+        return h128;
+    }
+}
+
+/*
+ * It's important for performance that XXH3_hashLong is not inlined.
+ */
+XXH_NO_INLINE XXH128_hash_t
+XXH3_hashLong_128b_default(const void* XXH_RESTRICT input, size_t len,
+                           XXH64_hash_t seed64,
+                           const void* XXH_RESTRICT secret, size_t secretLen)
+{
+    (void)seed64; (void)secret; (void)secretLen;
+    return XXH3_hashLong_128b_internal(input, len, XXH3_kSecret, sizeof(XXH3_kSecret),
+                                       XXH3_accumulate_512, XXH3_scrambleAcc);
+}
+
+/*
+ * It's important for performance to pass @secretLen (when it's static)
+ * to the compiler, so that it can properly optimize the vectorized loop.
+ */
+XXH_FORCE_INLINE XXH128_hash_t
+XXH3_hashLong_128b_withSecret(const void* XXH_RESTRICT input, size_t len,
+                              XXH64_hash_t seed64,
+                              const void* XXH_RESTRICT secret, size_t secretLen)
+{
+    (void)seed64;
+    return XXH3_hashLong_128b_internal(input, len, (const xxh_u8*)secret, secretLen,
+                                       XXH3_accumulate_512, XXH3_scrambleAcc);
+}
+
+XXH_FORCE_INLINE XXH128_hash_t
+XXH3_hashLong_128b_withSeed_internal(const void* XXH_RESTRICT input, size_t len,
+                                XXH64_hash_t seed64,
+                                XXH3_f_accumulate_512 f_acc512,
+                                XXH3_f_scrambleAcc f_scramble,
+                                XXH3_f_initCustomSecret f_initSec)
+{
+    if (seed64 == 0)
+        return XXH3_hashLong_128b_internal(input, len,
+                                           XXH3_kSecret, sizeof(XXH3_kSecret),
+                                           f_acc512, f_scramble);
+    {   XXH_ALIGN(XXH_SEC_ALIGN) xxh_u8 secret[XXH_SECRET_DEFAULT_SIZE];
+        f_initSec(secret, seed64);
+        return XXH3_hashLong_128b_internal(input, len, (const xxh_u8*)secret, sizeof(secret),
+                                           f_acc512, f_scramble);
+    }
+}
+
+/*
+ * It's important for performance that XXH3_hashLong is not inlined.
+ */
+XXH_NO_INLINE XXH128_hash_t
+XXH3_hashLong_128b_withSeed(const void* input, size_t len,
+                            XXH64_hash_t seed64, const void* XXH_RESTRICT secret, size_t secretLen)
+{
+    (void)secret; (void)secretLen;
+    return XXH3_hashLong_128b_withSeed_internal(input, len, seed64,
+                XXH3_accumulate_512, XXH3_scrambleAcc, XXH3_initCustomSecret);
+}
+
+typedef XXH128_hash_t (*XXH3_hashLong128_f)(const void* XXH_RESTRICT, size_t,
+                                            XXH64_hash_t, const void* XXH_RESTRICT, size_t);
+
+XXH_FORCE_INLINE XXH128_hash_t
+XXH3_128bits_internal(const void* input, size_t len,
+                      XXH64_hash_t seed64, const void* XXH_RESTRICT secret, size_t secretLen,
+                      XXH3_hashLong128_f f_hl128)
+{
+    XXH_ASSERT(secretLen >= XXH3_SECRET_SIZE_MIN);
+    /*
+     * If an action is to be taken if `secret` conditions are not respected,
+     * it should be done here.
+     * For now, it's a contract pre-condition.
+     * Adding a check and a branch here would cost performance at every hash.
+     */
+    if (len <= 16)
+        return XXH3_len_0to16_128b((const xxh_u8*)input, len, (const xxh_u8*)secret, seed64);
+    if (len <= 128)
+        return XXH3_len_17to128_128b((const xxh_u8*)input, len, (const xxh_u8*)secret, secretLen, seed64);
+    if (len <= XXH3_MIDSIZE_MAX)
+        return XXH3_len_129to240_128b((const xxh_u8*)input, len, (const xxh_u8*)secret, secretLen, seed64);
+    return f_hl128(input, len, seed64, secret, secretLen);
+}
+
+
+/* ===   Public XXH128 API   === */
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH128_hash_t XXH3_128bits(const void* input, size_t len)
+{
+    return XXH3_128bits_internal(input, len, 0,
+                                 XXH3_kSecret, sizeof(XXH3_kSecret),
+                                 XXH3_hashLong_128b_default);
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH128_hash_t
+XXH3_128bits_withSecret(const void* input, size_t len, const void* secret, size_t secretSize)
+{
+    return XXH3_128bits_internal(input, len, 0,
+                                 (const xxh_u8*)secret, secretSize,
+                                 XXH3_hashLong_128b_withSecret);
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH128_hash_t
+XXH3_128bits_withSeed(const void* input, size_t len, XXH64_hash_t seed)
+{
+    return XXH3_128bits_internal(input, len, seed,
+                                 XXH3_kSecret, sizeof(XXH3_kSecret),
+                                 XXH3_hashLong_128b_withSeed);
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH128_hash_t
+XXH3_128bits_withSecretandSeed(const void* input, size_t len, const void* secret, size_t secretSize, XXH64_hash_t seed)
+{
+    if (len <= XXH3_MIDSIZE_MAX)
+        return XXH3_128bits_internal(input, len, seed, XXH3_kSecret, sizeof(XXH3_kSecret), NULL);
+    return XXH3_hashLong_128b_withSecret(input, len, seed, secret, secretSize);
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH128_hash_t
+XXH128(const void* input, size_t len, XXH64_hash_t seed)
+{
+    return XXH3_128bits_withSeed(input, len, seed);
+}
+
+
+/* ===   XXH3 128-bit streaming   === */
+
+/*
+ * All initialization and update functions are identical to 64-bit streaming variant.
+ * The only difference is the finalization routine.
+ */
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH_errorcode
+XXH3_128bits_reset(XXH3_state_t* statePtr)
+{
+    return XXH3_64bits_reset(statePtr);
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH_errorcode
+XXH3_128bits_reset_withSecret(XXH3_state_t* statePtr, const void* secret, size_t secretSize)
+{
+    return XXH3_64bits_reset_withSecret(statePtr, secret, secretSize);
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH_errorcode
+XXH3_128bits_reset_withSeed(XXH3_state_t* statePtr, XXH64_hash_t seed)
+{
+    return XXH3_64bits_reset_withSeed(statePtr, seed);
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH_errorcode
+XXH3_128bits_reset_withSecretandSeed(XXH3_state_t* statePtr, const void* secret, size_t secretSize, XXH64_hash_t seed)
+{
+    return XXH3_64bits_reset_withSecretandSeed(statePtr, secret, secretSize, seed);
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH_errorcode
+XXH3_128bits_update(XXH3_state_t* state, const void* input, size_t len)
+{
+    return XXH3_update(state, (const xxh_u8*)input, len,
+                       XXH3_accumulate_512, XXH3_scrambleAcc);
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH128_hash_t XXH3_128bits_digest (const XXH3_state_t* state)
+{
+    const unsigned char* const secret = (state->extSecret == NULL) ? state->customSecret : state->extSecret;
+    if (state->totalLen > XXH3_MIDSIZE_MAX) {
+        XXH_ALIGN(XXH_ACC_ALIGN) XXH64_hash_t acc[XXH_ACC_NB];
+        XXH3_digest_long(acc, state, secret);
+        XXH_ASSERT(state->secretLimit + XXH_STRIPE_LEN >= sizeof(acc) + XXH_SECRET_MERGEACCS_START);
+        {   XXH128_hash_t h128;
+            h128.low64  = XXH3_mergeAccs(acc,
+                                         secret + XXH_SECRET_MERGEACCS_START,
+                                         (xxh_u64)state->totalLen * XXH_PRIME64_1);
+            h128.high64 = XXH3_mergeAccs(acc,
+                                         secret + state->secretLimit + XXH_STRIPE_LEN
+                                                - sizeof(acc) - XXH_SECRET_MERGEACCS_START,
+                                         ~((xxh_u64)state->totalLen * XXH_PRIME64_2));
+            return h128;
+        }
+    }
+    /* len <= XXH3_MIDSIZE_MAX : short code */
+    if (state->seed)
+        return XXH3_128bits_withSeed(state->buffer, (size_t)state->totalLen, state->seed);
+    return XXH3_128bits_withSecret(state->buffer, (size_t)(state->totalLen),
+                                   secret, state->secretLimit + XXH_STRIPE_LEN);
+}
+
+/* 128-bit utility functions */
+
+#include <string.h>   /* memcmp, memcpy */
+
+/* return : 1 is equal, 0 if different */
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API int XXH128_isEqual(XXH128_hash_t h1, XXH128_hash_t h2)
+{
+    /* note : XXH128_hash_t is compact, it has no padding byte */
+    return !(memcmp(&h1, &h2, sizeof(h1)));
+}
+
+/* This prototype is compatible with stdlib's qsort().
+ * return : >0 if *h128_1  > *h128_2
+ *          <0 if *h128_1  < *h128_2
+ *          =0 if *h128_1 == *h128_2  */
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API int XXH128_cmp(const void* h128_1, const void* h128_2)
+{
+    XXH128_hash_t const h1 = *(const XXH128_hash_t*)h128_1;
+    XXH128_hash_t const h2 = *(const XXH128_hash_t*)h128_2;
+    int const hcmp = (h1.high64 > h2.high64) - (h2.high64 > h1.high64);
+    /* note : bets that, in most cases, hash values are different */
+    if (hcmp) return hcmp;
+    return (h1.low64 > h2.low64) - (h2.low64 > h1.low64);
+}
+
+
+/*======   Canonical representation   ======*/
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API void
+XXH128_canonicalFromHash(XXH128_canonical_t* dst, XXH128_hash_t hash)
+{
+    XXH_STATIC_ASSERT(sizeof(XXH128_canonical_t) == sizeof(XXH128_hash_t));
+    if (XXH_CPU_LITTLE_ENDIAN) {
+        hash.high64 = XXH_swap64(hash.high64);
+        hash.low64  = XXH_swap64(hash.low64);
+    }
+    XXH_memcpy(dst, &hash.high64, sizeof(hash.high64));
+    XXH_memcpy((char*)dst + sizeof(hash.high64), &hash.low64, sizeof(hash.low64));
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH128_hash_t
+XXH128_hashFromCanonical(const XXH128_canonical_t* src)
+{
+    XXH128_hash_t h;
+    h.high64 = XXH_readBE64(src);
+    h.low64  = XXH_readBE64(src->digest + 8);
+    return h;
+}
+
+
+
+/* ==========================================
+ * Secret generators
+ * ==========================================
+ */
+#define XXH_MIN(x, y) (((x) > (y)) ? (y) : (x))
+
+static void XXH3_combine16(void* dst, XXH128_hash_t h128)
+{
+    XXH_writeLE64( dst, XXH_readLE64(dst) ^ h128.low64 );
+    XXH_writeLE64( (char*)dst+8, XXH_readLE64((char*)dst+8) ^ h128.high64 );
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API XXH_errorcode
+XXH3_generateSecret(void* secretBuffer, size_t secretSize, const void* customSeed, size_t customSeedSize)
+{
+    XXH_ASSERT(secretBuffer != NULL);
+    if (secretBuffer == NULL) return XXH_ERROR;
+    XXH_ASSERT(secretSize >= XXH3_SECRET_SIZE_MIN);
+    if (secretSize < XXH3_SECRET_SIZE_MIN) return XXH_ERROR;
+    if (customSeedSize == 0) {
+        customSeed = XXH3_kSecret;
+        customSeedSize = XXH_SECRET_DEFAULT_SIZE;
+    }
+    XXH_ASSERT(customSeed != NULL);
+    if (customSeed == NULL) return XXH_ERROR;
+
+    /* Fill secretBuffer with a copy of customSeed - repeat as needed */
+    {   size_t pos = 0;
+        while (pos < secretSize) {
+            size_t const toCopy = XXH_MIN((secretSize - pos), customSeedSize);
+            memcpy((char*)secretBuffer + pos, customSeed, toCopy);
+            pos += toCopy;
+    }   }
+
+    {   size_t const nbSeg16 = secretSize / 16;
+        size_t n;
+        XXH128_canonical_t scrambler;
+        XXH128_canonicalFromHash(&scrambler, XXH128(customSeed, customSeedSize, 0));
+        for (n=0; n<nbSeg16; n++) {
+            XXH128_hash_t const h128 = XXH128(&scrambler, sizeof(scrambler), n);
+            XXH3_combine16((char*)secretBuffer + n*16, h128);
+        }
+        /* last segment */
+        XXH3_combine16((char*)secretBuffer + secretSize - 16, XXH128_hashFromCanonical(&scrambler));
+    }
+    return XXH_OK;
+}
+
+/*! @ingroup xxh3_family */
+XXH_PUBLIC_API void
+XXH3_generateSecret_fromSeed(void* secretBuffer, XXH64_hash_t seed)
+{
+    XXH_ALIGN(XXH_SEC_ALIGN) xxh_u8 secret[XXH_SECRET_DEFAULT_SIZE];
+    XXH3_initCustomSecret(secret, seed);
+    XXH_ASSERT(secretBuffer != NULL);
+    memcpy(secretBuffer, secret, XXH_SECRET_DEFAULT_SIZE);
+}
+
+
+
+/* Pop our optimization override from above */
+#if XXH_VECTOR == XXH_AVX2 /* AVX2 */ \
+  && defined(__GNUC__) && !defined(__clang__) /* GCC, not Clang */ \
+  && defined(__OPTIMIZE__) && !defined(__OPTIMIZE_SIZE__) /* respect -O0 and -Os */
+#  pragma GCC pop_options
+#endif
+
+#endif  /* XXH_NO_LONG_LONG */
+
+#endif  /* XXH_NO_XXH3 */
+
+/*!
+ * @}
+ */
+#endif  /* XXH_IMPLEMENTATION */
+
+
+#if defined (__cplusplus)
+}
+#endif