/* SPDX-License-Identifier: LGPL-2.1-or-later */

#include <fcntl.h>
#include <linux/btrfs.h>
#include <linux/fsverity.h>
#include <stddef.h>
#include <stdio.h>
#include <sys/file.h>
#include <sys/ioctl.h>
#include <sys/sendfile.h>
#include <sys/sysmacros.h>
#include <unistd.h>

#include "alloc-util.h"
#include "btrfs-util.h"
#include "chattr-util.h"
#include "copy.h"
#include "dirent-util.h"
#include "errno-util.h"
#include "fd-util.h"
#include "fileio.h"
#include "fs-util.h"
#include "hashmap.h"
#include "log.h"
#include "mountpoint-util.h"
#include "nulstr-util.h"
#include "path-util.h"
#include "recurse-dir.h"
#include "rm-rf.h"
#include "selinux-util.h"
#include "signal-util.h"
#include "stat-util.h"
#include "stdio-util.h"
#include "string-util.h"
#include "sync-util.h"
#include "time-util.h"
#include "tmpfile-util.h"
#include "umask-util.h"
#include "user-util.h"
#include "xattr-util.h"

/* If we copy via a userspace buffer, size it to 64K */
#define COPY_BUFFER_SIZE (64U*U64_KB)

/* If a byte progress function is specified during copying, never try to copy more than 1M, so that we can
 * reasonably call the progress function still */
#define PROGRESS_STEP_SIZE (1U*U64_MB)

/* A safety net for descending recursively into file system trees to copy. On Linux PATH_MAX is 4096, which means the
 * deepest valid path one can build is around 2048, which we hence use as a safety net here, to not spin endlessly in
 * case of bind mount cycles and suchlike. */
#define COPY_DEPTH_MAX 2048U

static ssize_t try_copy_file_range(
                int fd_in, loff_t *off_in,
                int fd_out, loff_t *off_out,
                size_t len,
                unsigned flags) {

        static int have = -1;
        ssize_t r;

        if (have == 0)
                return -ENOSYS;

        r = copy_file_range(fd_in, off_in, fd_out, off_out, len, flags);
        if (have < 0)
                have = r >= 0 || errno != ENOSYS;
        if (r < 0)
                return -errno;

        return r;
}

static int look_for_signals(CopyFlags copy_flags) {
        int r;

        if ((copy_flags & (COPY_SIGINT|COPY_SIGTERM)) == 0)
                return 0;

        r = pop_pending_signal(copy_flags & COPY_SIGINT ? SIGINT : 0,
                               copy_flags & COPY_SIGTERM ? SIGTERM : 0);
        if (r < 0)
                return r;
        if (r != 0)
                return log_debug_errno(SYNTHETIC_ERRNO(EINTR),
                                       "Got %s, cancelling copy operation.", signal_to_string(r));

        return 0;
}

static int create_hole(int fd, off_t size) {
        off_t offset;
        off_t end;

        offset = lseek(fd, 0, SEEK_CUR);
        if (offset < 0)
                return -errno;

        end = lseek(fd, 0, SEEK_END);
        if (end < 0)
                return -errno;

        /* If we're not at the end of the target file, try to punch a hole in the existing space using fallocate(). */

        if (offset < end &&
            fallocate(fd, FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE, offset, MIN(size, end - offset)) < 0 &&
            !ERRNO_IS_NOT_SUPPORTED(errno))
                return -errno;

        if (end - offset >= size) {
                /* If we've created the full hole, set the file pointer to the end of the hole we created and exit. */
                if (lseek(fd, offset + size, SEEK_SET) < 0)
                        return -errno;

                return 0;
        }

        /* If we haven't created the full hole, use ftruncate() to grow the file (and the hole) to the
         * required size and move the file pointer to the end of the file. */

        size -= end - offset;

        if (ftruncate(fd, end + size) < 0)
                return -errno;

        if (lseek(fd, 0, SEEK_END) < 0)
                return -errno;

        return 0;
}

static int try_reflink_copy_bytes(
                int fdf,
                int fdt,
                uint64_t max_bytes,
                CopyFlags copy_flags) {

        int r;

        assert(fdf >= 0);
        assert(fdt >= 0);

        if (max_bytes == 0)
                return -EOPNOTSUPP;

        off_t foffset = FLAGS_SET(copy_flags, COPY_SEEK0_SOURCE) ? 0 : lseek(fdf, 0, SEEK_CUR);
        if (foffset < 0)
                return -EOPNOTSUPP;

        off_t toffset = FLAGS_SET(copy_flags, COPY_SEEK0_TARGET) ? 0 : lseek(fdt, 0, SEEK_CUR);
        if (toffset < 0)
                return -EOPNOTSUPP;

        r = reflink_range(fdf, foffset, fdt, toffset, max_bytes);
        if (r < 0)
                return -EOPNOTSUPP;

        if (max_bytes == UINT64_MAX) {
                off_t end;

                end = lseek(fdf, 0, SEEK_END);
                if (end < 0)
                        return -errno;
                if (end < foffset)
                        return -ESPIPE;

                if (lseek(fdt, toffset + (end - foffset), SEEK_SET) < 0)
                        return -errno;
        } else {
                if (lseek(fdf, foffset + max_bytes, SEEK_SET) < 0)
                        return -errno;
                if (lseek(fdt, toffset + max_bytes, SEEK_SET) < 0)
                        return -errno;
        }

        if (FLAGS_SET(copy_flags, COPY_VERIFY_LINKED)) {
                r = fd_verify_linked(fdf);
                if (r < 0)
                        return r;
        }

        return max_bytes == UINT64_MAX ? 0 /* we copied until EOF */
                                       : 1 /* we copied the requested range */;
}

int copy_bytes_full(
                int fdf, int fdt,
                uint64_t max_bytes,
                CopyFlags copy_flags,
                void **ret_remains,
                size_t *ret_remains_size,
                copy_progress_bytes_t progress,
                void *userdata) {

        _cleanup_close_ int fdf_opened = -EBADF, fdt_opened = -EBADF;
        bool try_cfr = true, try_sendfile = true;
        uint64_t copied_total = 0;
        int r;

        assert(fdf >= 0);
        assert(fdt >= 0);
        assert(!FLAGS_SET(copy_flags, COPY_LOCK_BSD));

        /* Tries to copy bytes from the file descriptor 'fdf' to 'fdt' in the smartest possible way. Copies a
         * maximum of 'max_bytes', which may be specified as UINT64_MAX, in which no maximum is applied.
         * Returns negative on error, zero if EOF is hit before the bytes limit is hit and positive
         * otherwise. If the copy fails for some reason but we read but didn't yet write some data and
         * ret_remains/ret_remains_size is not NULL, then it will be initialized with an allocated buffer
         * containing this "remaining" data. Note that these two parameters are initialized with a valid
         * buffer only on failure and only if there's actually data already read. Otherwise these parameters
         * if non-NULL are set to NULL. */

        if (ret_remains)
                *ret_remains = NULL;
        if (ret_remains_size)
                *ret_remains_size = 0;

        fdf = fd_reopen_condition(fdf, O_CLOEXEC | O_NOCTTY | O_RDONLY, O_PATH, &fdf_opened);
        if (fdf < 0)
                return fdf;
        fdt = fd_reopen_condition(fdt, O_CLOEXEC | O_NOCTTY | O_RDWR, O_PATH, &fdt_opened);
        if (fdt < 0)
                return fdt;

        if (FLAGS_SET(copy_flags, COPY_SEEK0_SOURCE) &&
            lseek(fdf, 0, SEEK_SET) < 0)
                return -errno;

        if (FLAGS_SET(copy_flags, COPY_SEEK0_TARGET) &&
            lseek(fdt, 0, SEEK_SET) < 0)
                return -errno;

        r = try_reflink_copy_bytes(fdf, fdt, max_bytes, copy_flags);
        if (r != -EOPNOTSUPP)
                return r;

        usec_t start_timestamp = USEC_INFINITY;
        if (progress)
                start_timestamp = now(CLOCK_MONOTONIC);

        for (;;) {
                ssize_t n;
                size_t m;

                if (max_bytes <= 0)
                        break;

                r = look_for_signals(copy_flags);
                if (r < 0)
                        return r;

                /* sendfile() accepts at most SSIZE_MAX-offset bytes to copy, hence let's subtract how much
                 * copied so far from SSIZE_MAX as maximum of what we want to copy. */
                if (try_sendfile) {
                        assert(copied_total < SSIZE_MAX);
                        m = (uint64_t) SSIZE_MAX - copied_total;
                } else
                        m = SSIZE_MAX;

                if (max_bytes != UINT64_MAX && m > max_bytes)
                        m = max_bytes;

                if (progress && m > PROGRESS_STEP_SIZE)
                        m = PROGRESS_STEP_SIZE;

                if (copy_flags & COPY_HOLES) {
                        off_t c, e;

                        c = lseek(fdf, 0, SEEK_CUR);
                        if (c < 0)
                                return -errno;

                        /* To see if we're in a hole, we search for the next data offset. */
                        e = lseek(fdf, c, SEEK_DATA);
                        if (e < 0 && errno == ENXIO)
                                /* If errno == ENXIO, that means we've reached the final hole of the file and
                                * that hole isn't followed by more data. */
                                e = lseek(fdf, 0, SEEK_END);
                        if (e < 0)
                                return -errno;

                        /* If we're in a hole (current offset is not a data offset), create a hole of the
                         * same size in the target file. */
                        if (e > c) {
                                /* Make sure our new hole doesn't go over the maximum size we're allowed to copy. */
                                n = MIN(max_bytes, (uint64_t) e - c);
                                r = create_hole(fdt, n);
                                if (r < 0)
                                        return r;

                                /* Make sure holes are taken into account in the maximum size we're supposed to copy. */
                                if (max_bytes != UINT64_MAX) {
                                        max_bytes -= n;
                                        if (max_bytes <= 0)
                                                break;
                                }

                                /* Update the size we're supposed to copy in this iteration if needed. */
                                if (m > max_bytes)
                                        m = max_bytes;
                        }

                        c = e; /* Set c to the start of the data segment. */

                        /* After copying a potential hole, find the end of the data segment by looking for
                         * the next hole. If we get ENXIO, we're at EOF. */
                        e = lseek(fdf, c, SEEK_HOLE);
                        if (e < 0) {
                                if (errno == ENXIO)
                                        break;
                                return -errno;
                        }
                        if (e == c)
                                /* Empty data segment? Maybe concurrent hole punching taking place? Or EOF?
                                 * Let's figure this out by copying the smallest amount possible */
                                m = 1;
                        else {
                                assert(e > c);

                                /* SEEK_HOLE modifies the file offset so we need to move back to the initial offset. */
                                if (lseek(fdf, c, SEEK_SET) < 0)
                                        return -errno;

                                /* Make sure we're not copying more than the current data segment. */
                                m = MIN(m, (size_t) e - c);
                                assert(m > 0);
                        }
                }

                /* First, try copy_file_range(), unless we already tried */
                if (try_cfr) {
                        n = try_copy_file_range(fdf, NULL, fdt, NULL, m, 0u);
                        if (n < 0) {
                                if (!IN_SET(n, -EINVAL, -ENOSYS, -EXDEV, -EBADF, -EOPNOTSUPP))
                                        return n;

                                try_cfr = false;
                                /* use fallback below */
                        } else if (n == 0) { /* likely EOF */

                                if (copied_total > 0)
                                        break;

                                /* So, we hit EOF immediately, without having copied a single byte. This
                                 * could indicate two things: the file is actually empty, or we are on some
                                 * virtual file system such as procfs/sysfs where the syscall actually
                                 * doesn't work but doesn't return an error. Try to handle that, by falling
                                 * back to simple read()s in case we encounter empty files.
                                 *
                                 * See: https://lwn.net/Articles/846403/ */
                                try_cfr = try_sendfile = false;
                        } else
                                /* Success! */
                                goto next;
                }

                /* Second, try sendfile(), unless we already tried */
                if (try_sendfile) {
                        n = sendfile(fdt, fdf, NULL, m);
                        if (n < 0) {
                                if (!IN_SET(errno, EINVAL, ENOSYS))
                                        return -errno;

                                try_sendfile = false;
                                /* use fallback below */
                        } else if (n == 0) { /* likely EOF */

                                if (copied_total > 0)
                                        break;

                                try_sendfile = false; /* same logic as above for copy_file_range() */
                        } else
                                /* Success! */
                                goto next;
                }

                /* Third, as a fallback just copy bits by hand */
                {
                        uint8_t buf[MIN(m, COPY_BUFFER_SIZE)], *p = buf;
                        ssize_t z;

                        n = read(fdf, buf, sizeof buf);
                        if (n < 0)
                                return -errno;
                        if (n == 0) /* EOF */
                                break;

                        z = (size_t) n;
                        do {
                                ssize_t k;

                                k = write(fdt, p, z);
                                if (k < 0) {
                                        r = -errno;

                                        if (ret_remains) {
                                                void *copy;

                                                copy = memdup(p, z);
                                                if (!copy)
                                                        return -ENOMEM;

                                                *ret_remains = copy;
                                        }

                                        if (ret_remains_size)
                                                *ret_remains_size = z;

                                        return r;
                                }

                                assert(k <= z);
                                z -= k;
                                p += k;
                        } while (z > 0);
                }

        next:
                copied_total += n;

                /* Disable sendfile() in case we are getting too close to it's SSIZE_MAX-offset limit */
                if (copied_total > SSIZE_MAX - COPY_BUFFER_SIZE)
                        try_sendfile = false;

                if (progress) {
                        usec_t t = now(CLOCK_MONOTONIC);
                        usec_t d = usec_sub_unsigned(t, start_timestamp);
                        uint64_t bps = UINT64_MAX;
                        if (d > USEC_PER_SEC * 3U)
                                bps = (uint64_t) (copied_total / ((double) d / USEC_PER_SEC));

                        r = progress(n, bps, userdata);
                        if (r < 0)
                                return r;
                }

                if (max_bytes != UINT64_MAX) {
                        assert(max_bytes >= (uint64_t) n);
                        max_bytes -= n;
                }
        }

        if (FLAGS_SET(copy_flags, COPY_VERIFY_LINKED)) {
                r = fd_verify_linked(fdf);
                if (r < 0)
                        return r;
        }

        if (copy_flags & COPY_TRUNCATE) {
                off_t off = lseek(fdt, 0, SEEK_CUR);
                if (off < 0)
                        return -errno;

                if (ftruncate(fdt, off) < 0)
                        return -errno;
        }

        return max_bytes <= 0; /* return 0 if we hit EOF earlier than the size limit */
}

/* Check whether *ts is within [epoch; clamp] */
static bool timespec_within(const struct timespec *ts, usec_t ts_clamp) {
        assert(ts);

        return ts->tv_sec < 0 ||              /* Date pre-1970, cannot clamp these */
               timespec_load(ts) <= ts_clamp; /* Date in [epoch; clamp] */
}

/* See ts_clamp in copy.h. */
static void clamp_inode_times(const struct stat *st, usec_t ts_clamp, struct timespec ret[static 2]) {
        assert(st);
        assert(ret);

        ret[0] = timespec_within(&st->st_atim, ts_clamp) ? st->st_atim : *TIMESPEC_STORE(ts_clamp);
        ret[1] = timespec_within(&st->st_mtim, ts_clamp) ? st->st_mtim : *TIMESPEC_STORE(ts_clamp);
}

/* Put *st's clamped times on the inode referred to by fdt */
static int clamp_futimens(int fdt, const struct stat *st, usec_t ts_clamp) {
        struct timespec times[2];

        assert(fdt >= 0);
        assert(st);

        clamp_inode_times(st, ts_clamp, times);
        return RET_NERRNO(futimens(fdt, times));
}

/* Same for (dt, to) */
static int clamp_utimensat(int dt, const char *to, const struct stat *st, usec_t ts_clamp) {
        struct timespec times[2];

        assert(dt >= 0 || dt == AT_FDCWD);
        assert(to);
        assert(st);

        clamp_inode_times(st, ts_clamp, times);
        return RET_NERRNO(utimensat(dt, to, times, AT_SYMLINK_NOFOLLOW));
}

static int fd_copy_symlink(
                int df,
                const char *from,
                const struct stat *st,
                int dt,
                const char *to,
                uid_t override_uid,
                gid_t override_gid,
                CopyFlags copy_flags,
                usec_t ts_clamp) {

        _cleanup_free_ char *target = NULL;
        int r;

        assert(st);
        assert(to);

        r = readlinkat_malloc(df, from, &target);
        if (r < 0)
                return r;

        if (copy_flags & COPY_MAC_CREATE) {
                r = mac_selinux_create_file_prepare_at(dt, to, S_IFLNK, /* label_context= */ NULL);
                if (r < 0)
                        return r;
        }
        r = RET_NERRNO(symlinkat(target, dt, to));
        if (copy_flags & COPY_MAC_CREATE)
                mac_selinux_create_file_clear();
        if (r < 0) {
                if (FLAGS_SET(copy_flags, COPY_GRACEFUL_WARN) && (ERRNO_IS_PRIVILEGE(r) || ERRNO_IS_NOT_SUPPORTED(r))) {
                        log_notice_errno(r, "Failed to copy symlink%s%s%s, ignoring: %m",
                                         isempty(from) ? "" : " '",
                                         strempty(from),
                                         isempty(from) ? "" : "'");
                        return 0;
                }

                return r;
        }

        if (fchownat(dt, to,
                     uid_is_valid(override_uid) ? override_uid : st->st_uid,
                     gid_is_valid(override_gid) ? override_gid : st->st_gid,
                     AT_SYMLINK_NOFOLLOW) < 0)
                r = -errno;

        (void) copy_xattr(df, from, dt, to, copy_flags);

        (void) clamp_utimensat(dt, to, st, ts_clamp);
        return r;
}

/* Encapsulates the database we store potential hardlink targets in */
typedef struct HardlinkContext {
        int dir_fd;    /* An fd to the directory we use as lookup table. Never AT_FDCWD. Lazily created, when
                        * we add the first entry. */

        /* These two fields are used to create the hardlink repository directory above — via
         * mkdirat(parent_fd, subdir) — and are kept so that we can automatically remove the directory again
         * when we are done. */
        int parent_fd; /* Possibly AT_FDCWD */
        char *subdir;
} HardlinkContext;

static int hardlink_context_setup(
                HardlinkContext *c,
                int dt,
                const char *to,
                CopyFlags copy_flags) {

        _cleanup_close_ int dt_copy = -EBADF;
        int r;

        assert(c);
        assert(c->dir_fd < 0 && c->dir_fd != AT_FDCWD);
        assert(c->parent_fd < 0);
        assert(!c->subdir);

        /* If hardlink recreation is requested we have to maintain a database of inodes that are potential
         * hardlink sources. Given that generally disk sizes have to be assumed to be larger than what fits
         * into physical RAM we cannot maintain that database in dynamic memory alone. Here we opt to
         * maintain it on disk, to simplify things: inside the destination directory we'll maintain a
         * temporary directory consisting of hardlinks of every inode we copied that might be subject of
         * hardlinks. We can then use that as hardlink source later on. Yes, this means additional disk IO
         * but thankfully Linux is optimized for this kind of thing. If this ever becomes a performance
         * bottleneck we can certainly place an in-memory hash table in front of this, but for the beginning,
         * let's keep things simple, and just use the disk as lookup table for inodes.
         *
         * Note that this should have zero performance impact as long as .n_link of all files copied remains
         * <= 0, because in that case we will not actually allocate the hardlink inode lookup table directory
         * on disk (we do so lazily, when the first candidate with .n_link > 1 is seen). This means, in the
         * common case where hardlinks are not used at all or only for few files the fact that we store the
         * table on disk shouldn't matter performance-wise. */

        if (!FLAGS_SET(copy_flags, COPY_HARDLINKS))
                return 0;

        if (dt == AT_FDCWD)
                dt_copy = AT_FDCWD;
        else if (dt < 0)
                return -EBADF;
        else {
                dt_copy = fcntl(dt, F_DUPFD_CLOEXEC, 3);
                if (dt_copy < 0)
                        return -errno;
        }

        if (to)
                r = tempfn_random_child(to, "hardlink", &c->subdir);
        else
                r = tempfn_random("hardlink", /* extra= */ NULL, &c->subdir);
        if (r < 0)
                return r;

        c->parent_fd = TAKE_FD(dt_copy);

        /* We don't actually create the directory we keep the table in here, that's done on-demand when the
         * first entry is added, using hardlink_context_realize() below. */
        return 1;
}

static int hardlink_context_realize(HardlinkContext *c) {
        if (!c)
                return 0;

        if (c->dir_fd >= 0) /* Already realized */
                return 1;

        if (c->parent_fd < 0 && c->parent_fd != AT_FDCWD) /* Not configured */
                return 0;

        assert(c->subdir);

        c->dir_fd = open_mkdir_at(c->parent_fd, c->subdir, O_EXCL|O_CLOEXEC, 0700);
        if (c->dir_fd < 0)
                return c->dir_fd;

        return 1;
}

static void hardlink_context_destroy(HardlinkContext *c) {
        int r;

        assert(c);

        /* Automatically remove the hardlink lookup table directory again after we are done. This is used via
         * _cleanup_() so that we really delete this, even on failure. */

        if (c->dir_fd >= 0) {
                /* <dir_fd> might be have already been used for reading, so we need to rewind it. */
                if (lseek(c->dir_fd, 0, SEEK_SET) < 0)
                        log_debug_errno(errno, "Failed to lseek on file descriptor, ignoring: %m");

                r = rm_rf_children(TAKE_FD(c->dir_fd), REMOVE_PHYSICAL, NULL); /* consumes dir_fd in all cases, even on failure */
                if (r < 0)
                        log_debug_errno(r, "Failed to remove hardlink store (%s) contents, ignoring: %m", c->subdir);

                assert(c->parent_fd >= 0 || c->parent_fd == AT_FDCWD);
                assert(c->subdir);

                if (unlinkat(c->parent_fd, c->subdir, AT_REMOVEDIR) < 0)
                        log_debug_errno(errno, "Failed to remove hardlink store (%s) directory, ignoring: %m", c->subdir);
        }

        assert_cc(AT_FDCWD < 0);
        c->parent_fd = safe_close(c->parent_fd);

        c->subdir = mfree(c->subdir);
}

static int try_hardlink(
                HardlinkContext *c,
                const struct stat *st,
                int dt,
                const char *to) {

        char dev_ino[DECIMAL_STR_MAX(dev_t)*2 + DECIMAL_STR_MAX(uint64_t) + 4];

        assert(st);
        assert(dt >= 0 || dt == AT_FDCWD);
        assert(to);

        if (!c) /* No temporary hardlink directory, don't bother */
                return 0;

        if (st->st_nlink <= 1) /* Source not hardlinked, don't bother */
                return 0;

        if (c->dir_fd < 0) /* not yet realized, hence empty */
                return 0;

        xsprintf(dev_ino, "%u:%u:%" PRIu64, major(st->st_dev), minor(st->st_dev), (uint64_t) st->st_ino);
        if (linkat(c->dir_fd, dev_ino, dt, to, 0) < 0)  {
                if (errno != ENOENT) /* doesn't exist in store yet */
                        log_debug_errno(errno, "Failed to hardlink %s to %s, ignoring: %m", dev_ino, to);
                return 0;
        }

        return 1;
}

static int memorize_hardlink(
                HardlinkContext *c,
                const struct stat *st,
                int dt,
                const char *to) {

        char dev_ino[DECIMAL_STR_MAX(dev_t)*2 + DECIMAL_STR_MAX(uint64_t) + 4];
        int r;

        assert(st);
        assert(dt >= 0 || dt == AT_FDCWD);
        assert(to);

        if (!c) /* No temporary hardlink directory, don't bother */
                return 0;

        if (st->st_nlink <= 1) /* Source not hardlinked, don't bother */
                return 0;

        r = hardlink_context_realize(c); /* Create the hardlink store lazily */
        if (r < 0)
                return r;

        xsprintf(dev_ino, "%u:%u:%" PRIu64, major(st->st_dev), minor(st->st_dev), (uint64_t) st->st_ino);
        if (linkat(dt, to, c->dir_fd, dev_ino, 0) < 0) {
                log_debug_errno(errno, "Failed to hardlink %s to %s, ignoring: %m", to, dev_ino);
                return 0;
        }

        return 1;
}

static int prepare_nocow(int fdf, const char *from, int fdt, unsigned *chattr_mask, unsigned *chattr_flags) {
        unsigned attrs = 0;
        int r;

        assert(fdf >= 0 || fdf == AT_FDCWD);
        assert(fdt >= 0);
        assert(!!chattr_mask == !!chattr_flags);

        /* If caller explicitly requested NOCOW to be set or unset, let's not interfere. */
        if (chattr_mask && FLAGS_SET(*chattr_mask, FS_NOCOW_FL))
                return 0;

        r = read_attr_at(fdf, from, &attrs);
        if (r < 0 && !ERRNO_IS_IOCTL_NOT_SUPPORTED(r) && r != -ELOOP) /* If the source is a symlink we get ELOOP */
                return r;

        if (FLAGS_SET(attrs, FS_NOCOW_FL)) {
                if (chattr_mask && chattr_flags) {
                        *chattr_mask |= FS_NOCOW_FL;
                        *chattr_flags |= FS_NOCOW_FL;
                } else
                        /* If the NOCOW flag is set on the source, make the copy NOCOW as well. If the source
                         * is not NOCOW, don't do anything in particular with the copy. */
                        (void) chattr_fd(fdt, FS_NOCOW_FL, FS_NOCOW_FL);
        }

        return 0;
}

/* Copies fs-verity status.  May re-open fdt to do its job. */
static int copy_fs_verity(int fdf, int *fdt) {
        int r;

        assert(fdf >= 0);
        assert(fdt);
        assert(*fdt >= 0);

        r = fd_verify_regular(fdf);
        if (r < 0)
                return r;

        struct fsverity_descriptor desc = {};
        struct fsverity_read_metadata_arg read_arg = {
                .metadata_type = FS_VERITY_METADATA_TYPE_DESCRIPTOR,
                .buf_ptr = (uintptr_t) &desc,
                .length = sizeof(desc),
        };

        r = ioctl(fdf, FS_IOC_READ_VERITY_METADATA, &read_arg);
        if (r < 0) {
                /* ENODATA means that the file doesn't have fs-verity,
                 * so the correct thing to do is to do nothing at all. */
                if (errno == ENODATA)
                        return 0;
                log_error_errno(errno, "Failed to read fs-verity metadata from source file: %m");
                /* For cases where fs-verity is unsupported we return a special error code */
                if (ERRNO_IS_NOT_SUPPORTED(errno))
                        return -ESOCKTNOSUPPORT;
                return -errno;
        }

        /* Make sure that the descriptor is completely initialized */
        assert(r == (int) sizeof desc);

        r = fd_verify_regular(*fdt);
        if (r < 0)
                return r;

        /* Okay. We're doing this now. We need to re-open fdt as read-only because
         * we can't enable fs-verity while writable file descriptors are outstanding. */
        _cleanup_close_ int reopened_fd = -EBADF;
        r = fd_reopen_condition(*fdt, O_RDONLY|O_CLOEXEC|O_NOCTTY, O_ACCMODE_STRICT|O_PATH, &reopened_fd);
        if (r < 0)
                return r;
        if (reopened_fd >= 0)
                close_and_replace(*fdt, reopened_fd);

        struct fsverity_enable_arg enable_arg = {
                .version = desc.version,
                .hash_algorithm = desc.hash_algorithm,
                .block_size = UINT32_C(1) << desc.log_blocksize,
                .salt_size = desc.salt_size,
                .salt_ptr = (uintptr_t) &desc.salt,
        };

        if (ioctl(*fdt, FS_IOC_ENABLE_VERITY, &enable_arg) < 0) {
                log_error_errno(errno, "Failed to set fs-verity metadata: %m");
                /* For cases where fs-verity is unsupported we return a special error code */
                if (ERRNO_IS_NOT_SUPPORTED(errno))
                        return -ESOCKTNOSUPPORT;
                return -errno;
        }

        return 0;
}

static int fd_copy_tree_generic(
                int df,
                const char *from,
                const struct stat *st,
                int dt,
                const char *to,
                dev_t original_device,
                unsigned depth_left,
                uid_t override_uid,
                gid_t override_gid,
                CopyFlags copy_flags,
                usec_t ts_clamp,
                Hashmap *denylist,
                Hashmap *subvolumes,
                HardlinkContext *hardlink_context,
                const char *display_path,
                copy_progress_path_t progress_path,
                copy_progress_bytes_t progress_bytes,
                void *userdata);

static int fd_copy_regular(
                int df,
                const char *from,
                const struct stat *st,
                int dt,
                const char *to,
                uid_t override_uid,
                gid_t override_gid,
                CopyFlags copy_flags,
                usec_t ts_clamp,
                HardlinkContext *hardlink_context,
                copy_progress_bytes_t progress,
                void *userdata) {

        _cleanup_close_ int fdf = -EBADF, fdt = -EBADF;
        int r, q;

        assert(st);
        assert(to);

        r = try_hardlink(hardlink_context, st, dt, to);
        if (r < 0)
                return r;
        if (r > 0) /* worked! */
                return 0;

        fdf = xopenat_full(df, from, O_RDONLY|O_CLOEXEC|O_NOCTTY|O_NOFOLLOW, XO_REGULAR, 0);
        if (fdf < 0)
                return fdf;

        if (copy_flags & COPY_MAC_CREATE) {
                r = mac_selinux_create_file_prepare_at(dt, to, S_IFREG, /* label_context= */ NULL);
                if (r < 0)
                        return r;
        }
        fdt = openat(dt, to, O_WRONLY|O_CREAT|O_EXCL|O_CLOEXEC|O_NOCTTY|O_NOFOLLOW, st->st_mode & 07777);
        if (copy_flags & COPY_MAC_CREATE)
                mac_selinux_create_file_clear();
        if (fdt < 0)
                return -errno;

        r = prepare_nocow(fdf, /* from= */ NULL, fdt, /* chattr_mask= */ NULL, /* chattr_flags= */ NULL);
        if (r < 0)
                return r;

        r = copy_bytes_full(fdf, fdt, UINT64_MAX, copy_flags, NULL, NULL, progress, userdata);
        if (r < 0)
                goto fail;

        if (fchown(fdt,
                   uid_is_valid(override_uid) ? override_uid : st->st_uid,
                   gid_is_valid(override_gid) ? override_gid : st->st_gid) < 0)
                r = -errno;

        if (fchmod(fdt, st->st_mode & 07777) < 0)
                r = -errno;

        (void) clamp_futimens(fdt, st, ts_clamp);
        (void) copy_xattr(fdf, NULL, fdt, NULL, copy_flags);

        if (FLAGS_SET(copy_flags, COPY_VERIFY_LINKED)) {
                r = fd_verify_linked(fdf);
                if (r < 0)
                        return r;
        }

        /* NB: fs-verity cannot be enabled when a writable file descriptor is outstanding.
         * copy_fs_verity() may well re-open 'fdt' as O_RDONLY. All code below this point
         * needs to be able to work with a read-only file descriptor. */
        if (FLAGS_SET(copy_flags, COPY_PRESERVE_FS_VERITY)) {
                r = copy_fs_verity(fdf, &fdt);
                if (r < 0)
                        goto fail;
        }

        if (copy_flags & COPY_FSYNC) {
                if (fsync(fdt) < 0) {
                        r = -errno;
                        goto fail;
                }
        }

        q = close_nointr(TAKE_FD(fdt)); /* even if this fails, the fd is now invalidated */
        if (q < 0) {
                r = q;
                goto fail;
        }

        (void) memorize_hardlink(hardlink_context, st, dt, to);
        return r;

fail:
        (void) unlinkat(dt, to, 0);
        return r;
}

static int fd_copy_fifo(
                int df,
                const char *from,
                const struct stat *st,
                int dt,
                const char *to,
                uid_t override_uid,
                gid_t override_gid,
                CopyFlags copy_flags,
                usec_t ts_clamp,
                HardlinkContext *hardlink_context) {
        int r;

        assert(st);
        assert(to);

        r = try_hardlink(hardlink_context, st, dt, to);
        if (r < 0)
                return r;
        if (r > 0) /* worked! */
                return 0;

        if (copy_flags & COPY_MAC_CREATE) {
                r = mac_selinux_create_file_prepare_at(dt, to, S_IFIFO, /* label_context= */ NULL);
                if (r < 0)
                        return r;
        }
        r = RET_NERRNO(mkfifoat(dt, to, st->st_mode & 07777));
        if (copy_flags & COPY_MAC_CREATE)
                mac_selinux_create_file_clear();
        if (FLAGS_SET(copy_flags, COPY_GRACEFUL_WARN) && (ERRNO_IS_NEG_PRIVILEGE(r) || ERRNO_IS_NEG_NOT_SUPPORTED(r))) {
                log_notice_errno(r, "Failed to copy fifo%s%s%s, ignoring: %m",
                                 isempty(from) ? "" : " '",
                                 strempty(from),
                                 isempty(from) ? "" : "'");
                return 0;
        } else if (r < 0)
                return r;

        if (fchownat(dt, to,
                     uid_is_valid(override_uid) ? override_uid : st->st_uid,
                     gid_is_valid(override_gid) ? override_gid : st->st_gid,
                     AT_SYMLINK_NOFOLLOW) < 0)
                r = -errno;

        if (fchmodat(dt, to, st->st_mode & 07777, AT_SYMLINK_NOFOLLOW) < 0)
                r = -errno;

        (void) clamp_utimensat(dt, to, st, ts_clamp);

        (void) memorize_hardlink(hardlink_context, st, dt, to);
        return r;
}

static int fd_copy_node(
                int df,
                const char *from,
                const struct stat *st,
                int dt,
                const char *to,
                uid_t override_uid,
                gid_t override_gid,
                CopyFlags copy_flags,
                usec_t ts_clamp,
                HardlinkContext *hardlink_context) {
        int r;

        assert(st);
        assert(to);

        r = try_hardlink(hardlink_context, st, dt, to);
        if (r < 0)
                return r;
        if (r > 0) /* worked! */
                return 0;

        if (copy_flags & COPY_MAC_CREATE) {
                r = mac_selinux_create_file_prepare_at(dt, to, st->st_mode & S_IFMT, /* label_context= */ NULL);
                if (r < 0)
                        return r;
        }
        r = RET_NERRNO(mknodat(dt, to, st->st_mode, st->st_rdev));
        if (copy_flags & COPY_MAC_CREATE)
                mac_selinux_create_file_clear();
        if (FLAGS_SET(copy_flags, COPY_GRACEFUL_WARN) && (ERRNO_IS_NEG_PRIVILEGE(r) || ERRNO_IS_NEG_NOT_SUPPORTED(r))) {
                log_notice_errno(r, "Failed to copy node%s%s%s, ignoring: %m",
                                 isempty(from) ? "" : " '",
                                 strempty(from),
                                 isempty(from) ? "" : "'");
                return 0;
        } else if (r < 0)
                return r;

        if (fchownat(dt, to,
                     uid_is_valid(override_uid) ? override_uid : st->st_uid,
                     gid_is_valid(override_gid) ? override_gid : st->st_gid,
                     AT_SYMLINK_NOFOLLOW) < 0)
                r = -errno;

        if (fchmodat(dt, to, st->st_mode & 07777, AT_SYMLINK_NOFOLLOW) < 0)
                r = -errno;

        (void) clamp_utimensat(dt, to, st, ts_clamp);

        (void) memorize_hardlink(hardlink_context, st, dt, to);
        return r;
}

static int fd_copy_directory(
                int df,
                const char *from,
                const struct stat *st,
                int dt,
                const char *to,
                dev_t original_device,
                unsigned depth_left,
                uid_t override_uid,
                gid_t override_gid,
                CopyFlags copy_flags,
                usec_t ts_clamp,
                Hashmap *denylist,
                Hashmap *subvolumes,
                HardlinkContext *hardlink_context,
                const char *display_path,
                copy_progress_path_t progress_path,
                copy_progress_bytes_t progress_bytes,
                void *userdata) {

        _cleanup_(hardlink_context_destroy) HardlinkContext our_hardlink_context = {
                .dir_fd = -EBADF,
                .parent_fd = -EBADF,
        };

        _cleanup_free_ DirectoryEntries *des = NULL;
        _cleanup_close_ int fdf = -EBADF, fdt = -EBADF;
        _cleanup_closedir_ DIR *d = NULL;
        struct stat dt_st;
        bool exists;
        int r;

        assert(st);

        if (depth_left == 0)
                return -ENAMETOOLONG;

        fdf = xopenat(df, from, O_RDONLY|O_DIRECTORY|O_CLOEXEC|O_NOCTTY|O_NOFOLLOW);
        if (fdf < 0)
                return fdf;

        if (!hardlink_context) {
                /* If recreating hardlinks is requested let's set up a context for that now. */
                r = hardlink_context_setup(&our_hardlink_context, dt, to, copy_flags);
                if (r < 0)
                        return r;
                if (r > 0) /* It's enabled and allocated, let's now use the same context for all recursive
                            * invocations from here down */
                        hardlink_context = &our_hardlink_context;
        }

        d = take_fdopendir(&fdf);
        if (!d)
                return -errno;

        r = dir_is_empty_at(dt, to, /* ignore_hidden_or_backup= */ false);
        if (r < 0 && r != -ENOENT)
                return r;
        if ((r > 0 && !(copy_flags & (COPY_MERGE|COPY_MERGE_EMPTY))) || (r == 0 && !FLAGS_SET(copy_flags, COPY_MERGE)))
                return -EEXIST;

        exists = r >= 0;

        XOpenFlags flags = copy_flags & COPY_MAC_CREATE ? XO_LABEL : 0;
        if (hashmap_contains(subvolumes, st)) {
                flags |= XO_SUBVOLUME;
                if ((PTR_TO_INT(hashmap_get(subvolumes, st)) & BTRFS_SUBVOL_NODATACOW))
                        flags |= XO_NOCOW;
        }

        fdt = xopenat_lock_full(dt, to,
                                O_RDONLY|O_DIRECTORY|O_CLOEXEC|O_NOCTTY|O_NOFOLLOW|(exists ? 0 : O_CREAT|O_EXCL),
                                flags,
                                st->st_mode & 07777,
                                copy_flags & COPY_LOCK_BSD ? LOCK_BSD : LOCK_NONE,
                                LOCK_EX);
        if (fdt < 0)
                return fdt;

        if (exists && FLAGS_SET(copy_flags, COPY_RESTORE_DIRECTORY_TIMESTAMPS) && fstat(fdt, &dt_st) < 0)
                return -errno;

        int ret = 0;

        if (PTR_TO_INT(hashmap_get(denylist, st)) == DENY_CONTENTS) {
                log_debug("%s is in the denylist, not recursing", from ?: "file to copy");
                goto finish;
        }

        /* Walk children in deterministic (alphabetical) order. The natural readdir() order depends on the
         * source filesystem's directory storage (e.g. ext4 dir hash) and varies across hosts, which leaks
         * into the destination when it records entries in insertion order (e.g. vfat). Sorting here keeps
         * copy_tree() reproducible regardless of the source filesystem layout. */
        r = readdir_all(dirfd(d), RECURSE_DIR_SORT, &des);
        if (r < 0)
                return r;

        FOREACH_ARRAY(i, des->entries, des->n_entries) {
                struct dirent *de = *i;
                const char *child_display_path = NULL;
                _cleanup_free_ char *dp = NULL;
                struct stat buf;

                r = look_for_signals(copy_flags);
                if (r < 0)
                        return r;

                if (fstatat(dirfd(d), de->d_name, &buf, AT_SYMLINK_NOFOLLOW) < 0) {
                        RET_GATHER(ret, -errno);
                        continue;
                }

                if (progress_path) {
                        if (display_path)
                                child_display_path = dp = path_join(display_path, de->d_name);
                        else
                                child_display_path = de->d_name;

                        r = progress_path(child_display_path, &buf, userdata);
                        if (r < 0)
                                return r;
                }

                if (PTR_TO_INT(hashmap_get(denylist, &buf)) == DENY_INODE) {
                        log_debug("%s%s%s is in the denylist, ignoring",
                                  strempty(from), isempty(from) ? "" : "/", de->d_name);
                        continue;
                }

                if (S_ISDIR(buf.st_mode)) {
                        /*
                         * Don't descend into directories on other file systems, if this is requested. We do a simple
                         * .st_dev check here, which basically comes for free. Note that we do this check only on
                         * directories, not other kind of file system objects, for two reason:
                         *
                         * • The kernel's overlayfs pseudo file system that overlays multiple real file systems
                         *   propagates the .st_dev field of the file system a file originates from all the way up
                         *   through the stack to stat(). It doesn't do that for directories however. This means that
                         *   comparing .st_dev on non-directories suggests that they all are mount points. To avoid
                         *   confusion we hence avoid relying on this check for regular files.
                         *
                         * • The main reason we do this check at all is to protect ourselves from bind mount cycles,
                         *   where we really want to avoid descending down in all eternity. However the .st_dev check
                         *   is usually not sufficient for this protection anyway, as bind mount cycles from the same
                         *   file system onto itself can't be detected that way. (Note we also do a recursion depth
                         *   check, which is probably the better protection in this regard, which is why
                         *   COPY_SAME_MOUNT is optional).
                         */

                        if (FLAGS_SET(copy_flags, COPY_SAME_MOUNT)) {
                                if (buf.st_dev != original_device)
                                        continue;

                                r = is_mount_point_at(dirfd(d), de->d_name, 0);
                                if (r < 0)
                                        return r;
                                if (r > 0)
                                        continue;
                        }
                }

                r = fd_copy_tree_generic(dirfd(d), de->d_name, &buf, fdt, de->d_name, original_device,
                                         depth_left-1, override_uid, override_gid, copy_flags & ~COPY_LOCK_BSD,
                                         ts_clamp, denylist, subvolumes, hardlink_context, child_display_path, progress_path,
                                         progress_bytes, userdata);

                /* Propagate SIGINT/SIGTERM, ENOSPC, and fs-verity fails up instantly */
                if (IN_SET(r, -EINTR, -ENOSPC, -ESOCKTNOSUPPORT))
                        return r;
                if (r == -EEXIST && (copy_flags & COPY_MERGE))
                        r = 0;
                RET_GATHER(ret, r);
        }

finish:
        if (FLAGS_SET(copy_flags, COPY_MERGE_APPLY_STAT) || !exists) {
                if (fchown(fdt,
                           uid_is_valid(override_uid) ? override_uid : st->st_uid,
                           gid_is_valid(override_gid) ? override_gid : st->st_gid) < 0)
                        RET_GATHER(ret, -errno);

                if (fchmod(fdt, st->st_mode & 07777) < 0)
                        RET_GATHER(ret, -errno);

                /* Run hardlink context cleanup now because it potentially changes timestamps */
                hardlink_context_destroy(&our_hardlink_context);
                (void) copy_xattr(dirfd(d), NULL, fdt, NULL, copy_flags);

                (void) clamp_futimens(fdt, st, ts_clamp);
        } else if (FLAGS_SET(copy_flags, COPY_RESTORE_DIRECTORY_TIMESTAMPS)) {
                /* Run hardlink context cleanup now because it potentially changes timestamps */
                hardlink_context_destroy(&our_hardlink_context);
                /* If the directory already exists, make sure the timestamps stay the same as before. */
                (void) futimens(fdt, (struct timespec[]) { dt_st.st_atim, dt_st.st_mtim });
        }

        if (copy_flags & COPY_FSYNC_FULL) {
                if (fsync(fdt) < 0)
                        return -errno;
        }

        if (ret < 0)
                return ret;

        return copy_flags & COPY_LOCK_BSD ? TAKE_FD(fdt) : 0;
}

static int fd_copy_leaf(
                int df,
                const char *from,
                const struct stat *st,
                int dt,
                const char *to,
                uid_t override_uid,
                gid_t override_gid,
                CopyFlags copy_flags,
                usec_t ts_clamp,
                HardlinkContext *hardlink_context,
                const char *display_path,
                copy_progress_bytes_t progress_bytes,
                void *userdata) {
        int r;

        if (S_ISREG(st->st_mode))
                r = fd_copy_regular(df, from, st, dt, to, override_uid, override_gid, copy_flags, ts_clamp, hardlink_context, progress_bytes, userdata);
        else if (S_ISLNK(st->st_mode))
                r = fd_copy_symlink(df, from, st, dt, to, override_uid, override_gid, copy_flags, ts_clamp);
        else if (S_ISFIFO(st->st_mode))
                r = fd_copy_fifo(df, from, st, dt, to, override_uid, override_gid, copy_flags, ts_clamp, hardlink_context);
        else if (S_ISBLK(st->st_mode) || S_ISCHR(st->st_mode) || S_ISSOCK(st->st_mode))
                r = fd_copy_node(df, from, st, dt, to, override_uid, override_gid, copy_flags, ts_clamp, hardlink_context);
        else
                r = -EOPNOTSUPP;

        return r;
}

static int fd_copy_tree_generic(
                int df,
                const char *from,
                const struct stat *st,
                int dt,
                const char *to,
                dev_t original_device,
                unsigned depth_left,
                uid_t override_uid,
                gid_t override_gid,
                CopyFlags copy_flags,
                usec_t ts_clamp,
                Hashmap *denylist,
                Hashmap *subvolumes,
                HardlinkContext *hardlink_context,
                const char *display_path,
                copy_progress_path_t progress_path,
                copy_progress_bytes_t progress_bytes,
                void *userdata) {

        int r;

        assert(!FLAGS_SET(copy_flags, COPY_LOCK_BSD));

        if (S_ISDIR(st->st_mode))
                return fd_copy_directory(df, from, st, dt, to, original_device, depth_left-1, override_uid,
                                         override_gid, copy_flags, ts_clamp, denylist, subvolumes, hardlink_context,
                                         display_path, progress_path, progress_bytes, userdata);

        /* Only if we are copying a directory we are fine if the target dir is referenced by fd only */
        if (!to)
                return -ENOTDIR;

        DenyType t = PTR_TO_INT(hashmap_get(denylist, st));
        if (t == DENY_INODE) {
                log_debug("%s is in the denylist, ignoring", from ?: "file to copy");
                return 0;
        } else if (t == DENY_CONTENTS)
                log_debug("%s is configured to have its contents excluded, but is not a directory", from ?: "file to copy");

        r = fd_copy_leaf(df, from, st, dt, to, override_uid, override_gid, copy_flags, ts_clamp, hardlink_context, display_path, progress_bytes, userdata);
        /* We just tried to copy a leaf node of the tree. If it failed because the node already exists *and* the COPY_REPLACE flag has been provided, we should unlink the node and re-copy. */
        if (r == -EEXIST && (copy_flags & COPY_REPLACE)) {
                /* This codepath is us trying to address an error to copy, if the unlink fails, lets just return the original error. */
                if (unlinkat(dt, to, 0) < 0)
                        return r;

                r = fd_copy_leaf(df, from, st, dt, to, override_uid, override_gid, copy_flags, ts_clamp, hardlink_context, display_path, progress_bytes, userdata);
        }

        return r;
}

int copy_tree_at_full(
                int fdf,
                const char *from,
                int fdt,
                const char *to,
                uid_t override_uid,
                gid_t override_gid,
                CopyFlags copy_flags,
                usec_t ts_clamp,
                Hashmap *denylist,
                Hashmap *subvolumes,
                copy_progress_path_t progress_path,
                copy_progress_bytes_t progress_bytes,
                void *userdata) {

        struct stat st;
        int r;

        assert(!FLAGS_SET(copy_flags, COPY_LOCK_BSD));

        if (fstatat(fdf, strempty(from), &st, AT_SYMLINK_NOFOLLOW | (isempty(from) ? AT_EMPTY_PATH : 0)) < 0)
                return -errno;

        r = fd_copy_tree_generic(fdf, from, &st, fdt, to, st.st_dev, COPY_DEPTH_MAX, override_uid,
                                 override_gid, copy_flags, ts_clamp, denylist, subvolumes, NULL, NULL, progress_path,
                                 progress_bytes, userdata);
        if (r < 0)
                return r;

        if (S_ISDIR(st.st_mode) && (copy_flags & COPY_SYNCFS)) {
                /* If the top-level inode is a directory run syncfs() now. */
                r = syncfs_path(fdt, to);
                if (r < 0)
                        return r;
        } else if ((copy_flags & (COPY_FSYNC_FULL|COPY_SYNCFS)) != 0) {
                /* fsync() the parent dir of what we just copied if COPY_FSYNC_FULL is set. Also do this in
                 * case COPY_SYNCFS is set but the top-level inode wasn't actually a directory. We do this so that
                 * COPY_SYNCFS provides reasonable synchronization semantics on any kind of inode: when the
                 * copy operation is done the whole inode — regardless of its type — and all its children
                 * will be synchronized to disk. */
                r = fsync_parent_at(fdt, to);
                if (r < 0)
                        return r;
        }

        return 0;
}

static int sync_dir_by_flags(int dir_fd, const char *path, CopyFlags copy_flags) {
        assert(dir_fd >= 0 || dir_fd == AT_FDCWD);
        assert(path);

        if (copy_flags & COPY_SYNCFS)
                return syncfs_path(dir_fd, path);
        if (copy_flags & COPY_FSYNC_FULL)
                return fsync_parent_at(dir_fd, path);

        return 0;
}

int copy_directory_at_full(
                int dir_fdf,
                const char *from,
                int dir_fdt,
                const char *to,
                uid_t override_uid,
                gid_t override_gid,
                CopyFlags copy_flags,
                copy_progress_path_t progress_path,
                copy_progress_bytes_t progress_bytes,
                void *userdata) {

        _cleanup_close_ int fdt = -EBADF;
        struct stat st;
        int r;

        assert(dir_fdf >= 0 || dir_fdf == AT_FDCWD);
        assert(dir_fdt >= 0 || dir_fdt == AT_FDCWD);
        assert(to);

        if (fstatat(dir_fdf, strempty(from), &st, AT_SYMLINK_NOFOLLOW|(isempty(from) ? AT_EMPTY_PATH : 0)) < 0)
                return -errno;

        r = stat_verify_directory(&st);
        if (r < 0)
                return r;

        r = fd_copy_directory(
                        dir_fdf, from,
                        &st,
                        dir_fdt, to,
                        st.st_dev,
                        COPY_DEPTH_MAX,
                        override_uid,
                        override_gid,
                        copy_flags,
                        /* ts_clamp= */ USEC_INFINITY,
                        /* denylist= */ NULL,
                        /* subvolumes= */ NULL,
                        /* hardlink_context= */ NULL,
                        /* display_path= */ NULL,
                        progress_path,
                        progress_bytes,
                        userdata);
        if (r < 0)
                return r;

        if (FLAGS_SET(copy_flags, COPY_LOCK_BSD))
                fdt = r;

        r = sync_dir_by_flags(dir_fdt, to, copy_flags);
        if (r < 0)
                return r;

        return FLAGS_SET(copy_flags, COPY_LOCK_BSD) ? TAKE_FD(fdt) : 0;
}

int copy_file_fd_at_full(
                int dir_fdf,
                const char *from,
                int fdt,
                CopyFlags copy_flags,
                copy_progress_bytes_t progress_bytes,
                void *userdata) {

        _cleanup_close_ int fdf = -EBADF;
        struct stat st;
        int r;

        assert(dir_fdf >= 0 || dir_fdf == AT_FDCWD);
        assert(fdt >= 0);
        assert(!FLAGS_SET(copy_flags, COPY_LOCK_BSD));

        fdf = xopenat_full(dir_fdf, from, O_RDONLY|O_CLOEXEC|O_NOCTTY, XO_REGULAR, 0);
        if (fdf < 0)
                return fdf;

        if (fstat(fdt, &st) < 0)
                return -errno;

        r = copy_bytes_full(fdf, fdt, UINT64_MAX, copy_flags, NULL, NULL, progress_bytes, userdata);
        if (r < 0)
                return r;

        /* Make sure to copy file attributes only over if target is a regular
         * file (so that copying a file to /dev/null won't alter the access
         * mode/ownership of that device node...) */
        if (S_ISREG(st.st_mode)) {
                (void) copy_times(fdf, fdt, copy_flags);
                (void) copy_xattr(fdf, NULL, fdt, NULL, copy_flags);
        }

        if (FLAGS_SET(copy_flags, COPY_VERIFY_LINKED)) {
                r = fd_verify_linked(fdf);
                if (r < 0)
                        return r;
        }

        if (copy_flags & COPY_FSYNC_FULL) {
                r = fsync_full(fdt);
                if (r < 0)
                        return r;
        } else if (copy_flags & COPY_FSYNC) {
                if (fsync(fdt) < 0)
                        return -errno;
        }

        return 0;
}

int copy_file_at_full(
                int dir_fdf,
                const char *from,
                int dir_fdt,
                const char *to,
                int flags,
                mode_t mode,
                unsigned chattr_flags,
                unsigned chattr_mask,
                CopyFlags copy_flags,
                copy_progress_bytes_t progress_bytes,
                void *userdata) {

        _cleanup_close_ int fdf = -EBADF, fdt = -EBADF;
        int r;

        assert(dir_fdf >= 0 || dir_fdf == AT_FDCWD);
        assert(dir_fdt >= 0 || dir_fdt == AT_FDCWD);
        assert(to);

        fdf = xopenat_full(dir_fdf, from, O_RDONLY|O_CLOEXEC|O_NOCTTY, XO_REGULAR, 0);
        if (fdf < 0)
                return fdf;

        if (mode == MODE_INVALID) {
                struct stat st;

                if (fstat(fdf, &st) < 0)
                        return -errno;

                mode = st.st_mode;
        }

        WITH_UMASK(0000) {
                fdt = xopenat_lock_full(dir_fdt, to,
                                        flags|O_WRONLY|O_CREAT|O_CLOEXEC|O_NOCTTY,
                                        XO_REGULAR | (copy_flags & COPY_MAC_CREATE ? XO_LABEL : 0),
                                        mode,
                                        copy_flags & COPY_LOCK_BSD ? LOCK_BSD : LOCK_NONE, LOCK_EX);
                if (fdt < 0)
                        return fdt;
        }

        r = prepare_nocow(fdf, /* from= */ NULL, fdt, &chattr_mask, &chattr_flags);
        if (r < 0)
                return r;

        if ((chattr_mask & CHATTR_EARLY_FL) != 0)
                (void) chattr_fd(fdt, chattr_flags, chattr_mask & CHATTR_EARLY_FL);

        r = copy_bytes_full(fdf, fdt, UINT64_MAX, copy_flags & ~COPY_LOCK_BSD, NULL, NULL, progress_bytes, userdata);
        if (r < 0)
                goto fail;

        (void) copy_times(fdf, fdt, copy_flags);
        (void) copy_xattr(fdf, NULL, fdt, NULL, copy_flags);

        if (FLAGS_SET(copy_flags, COPY_VERIFY_LINKED)) {
                r = fd_verify_linked(fdf);
                if (r < 0)
                        goto fail;
        }

        unsigned nocow = FLAGS_SET(copy_flags, COPY_NOCOW_AFTER) ? FS_NOCOW_FL : 0;
        if (((chattr_mask & ~CHATTR_EARLY_FL) | nocow) != 0)
                (void) chattr_fd(fdt, chattr_flags | nocow, (chattr_mask & ~CHATTR_EARLY_FL) | nocow);

        if (copy_flags & (COPY_FSYNC|COPY_FSYNC_FULL)) {
                if (fsync(fdt) < 0) {
                        r = -errno;
                        goto fail;
                }
        }

        if (!FLAGS_SET(copy_flags, COPY_LOCK_BSD)) {
                r = close_nointr(TAKE_FD(fdt)); /* even if this fails, the fd is now invalidated */
                if (r < 0)
                        goto fail;
        }

        if (copy_flags & COPY_FSYNC_FULL) {
                r = fsync_parent_at(dir_fdt, to);
                if (r < 0)
                        goto fail;
        }

        return copy_flags & COPY_LOCK_BSD ? TAKE_FD(fdt) : 0;

fail:
        /* Only unlink if we definitely are the ones who created the file */
        if (FLAGS_SET(flags, O_EXCL))
                (void) unlinkat(dir_fdt, to, 0);

        return r;
}

int copy_file_atomic_at_full(
                int dir_fdf,
                const char *from,
                int dir_fdt,
                const char *to,
                mode_t mode,
                unsigned chattr_flags,
                unsigned chattr_mask,
                CopyFlags copy_flags,
                copy_progress_bytes_t progress_bytes,
                void *userdata) {

        _cleanup_(unlink_and_freep) char *t = NULL;
        _cleanup_close_ int fdt = -EBADF;
        int r;

        assert(to);
        assert(!FLAGS_SET(copy_flags, COPY_LOCK_BSD));

        if (copy_flags & COPY_MAC_CREATE) {
                r = mac_selinux_create_file_prepare_at(dir_fdt, to, S_IFREG, /* label_context= */ NULL);
                if (r < 0)
                        return r;
        }
        fdt = open_tmpfile_linkable_at(dir_fdt, to, O_WRONLY|O_CLOEXEC, &t);
        if (copy_flags & COPY_MAC_CREATE)
                mac_selinux_create_file_clear();
        if (fdt < 0)
                return fdt;

        r = prepare_nocow(dir_fdf, from, fdt, &chattr_mask, &chattr_flags);
        if (r < 0)
                return r;

        if ((chattr_mask & CHATTR_EARLY_FL) != 0)
                (void) chattr_fd(fdt, chattr_flags, chattr_mask & CHATTR_EARLY_FL);

        r = copy_file_fd_at_full(dir_fdf, from, fdt, copy_flags, progress_bytes, userdata);
        if (r < 0)
                return r;

        if (fchmod(fdt, mode) < 0)
                return -errno;

        if ((copy_flags & (COPY_FSYNC|COPY_FSYNC_FULL))) {
                /* Sync the file */
                if (fsync(fdt) < 0)
                        return -errno;
        }

        r = link_tmpfile_at(fdt, dir_fdt, t, to, (copy_flags & COPY_REPLACE) ? LINK_TMPFILE_REPLACE : 0);
        if (r < 0)
                return r;

        t = mfree(t);

        unsigned nocow = FLAGS_SET(copy_flags, COPY_NOCOW_AFTER) ? FS_NOCOW_FL : 0;
        if (((chattr_mask & ~CHATTR_EARLY_FL) | nocow) != 0)
                (void) chattr_fd(fdt, chattr_flags | nocow, (chattr_mask & ~CHATTR_EARLY_FL) | nocow);

        r = close_nointr(TAKE_FD(fdt)); /* even if this fails, the fd is now invalidated */
        if (r < 0)
                goto fail;

        if (copy_flags & COPY_FSYNC_FULL) {
                /* Sync the parent directory */
                r = fsync_parent_at(dir_fdt, to);
                if (r < 0)
                        goto fail;
        }

        return 0;

fail:
        /* link_tmpfile_at() succeeded, so 'to' is now published. In replacement mode, do not
         * remove it again, as that may delete a pre-existing target replaced by this copy. */
        if (!FLAGS_SET(copy_flags, COPY_REPLACE))
                (void) unlinkat(dir_fdt, to, 0);

        return r;
}

int copy_times_full(int fdf, int fdt, CopyFlags flags, usec_t ts_clamp) {
        struct stat st;
        int r;

        assert(fdf >= 0);
        assert(fdt >= 0);

        if (fstat(fdf, &st) < 0)
                return -errno;

        r = clamp_futimens(fdt, &st, ts_clamp);
        if (r < 0)
                return r;

        if (FLAGS_SET(flags, COPY_CRTIME)) {
                usec_t crtime;

                if (fd_getcrtime(fdf, &crtime) >= 0)
                        (void) fd_setcrtime(fdt, MIN(crtime, ts_clamp));
        }

        return 0;
}

int copy_access(int fdf, int fdt) {
        struct stat st;

        assert(fdf >= 0);
        assert(fdt >= 0);

        /* Copies just the access mode (and not the ownership) from fdf to fdt */

        if (fstat(fdf, &st) < 0)
                return -errno;

        return RET_NERRNO(fchmod(fdt, st.st_mode & 07777));
}

int copy_owner(int fdf, int fdt) {
        struct stat st;

        assert(fdf >= 0);
        assert(fdt >= 0);

        if (fstat(fdf, &st) < 0)
                return -errno;

        return RET_NERRNO(fchown(fdt, st.st_uid, st.st_gid));
}

int copy_rights_with_fallback(int fdf, int fdt, const char *patht) {
        struct stat st;

        assert(fdf >= 0);
        assert(fdt >= 0);

        /* Copies both access mode and ownership from fdf to fdt */

        if (fstat(fdf, &st) < 0)
                return -errno;

        return fchmod_and_chown_with_fallback(fdt, patht, st.st_mode & 07777, st.st_uid, st.st_gid);
}

int copy_xattr(int df, const char *from, int dt, const char *to, CopyFlags copy_flags) {
        _cleanup_free_ char *names = NULL;
        int ret = 0, r;

        r = listxattr_at_malloc(df, from, 0, &names);
        if (r < 0)
                return r;

        NULSTR_FOREACH(p, names) {
                if (!FLAGS_SET(copy_flags, COPY_ALL_XATTRS) && !startswith(p, "user."))
                        continue;

                _cleanup_free_ char *value = NULL;
                size_t value_size;
                r = getxattr_at_malloc(df, from, p, 0, &value, &value_size);
                if (r == -ENODATA)
                        continue; /* gone by now */
                if (r < 0)
                        return r;

                RET_GATHER(ret, xsetxattr_full(dt, to, /* at_flags= */ 0, p, value, value_size, /* xattr_flags= */ 0));
        }

        return ret;
}

int reflink(int infd, int outfd) {
        int r;

        assert(infd >= 0);
        assert(outfd >= 0);

        /* Make sure we invoke the ioctl on a regular file, so that no device driver accidentally gets it. */

        r = fd_verify_regular(outfd);
        if (r < 0)
                return r;

        /* FICLONE was introduced in Linux 4.5 but it uses the same number as BTRFS_IOC_CLONE introduced earlier */

        assert_cc(FICLONE == BTRFS_IOC_CLONE);

        return RET_NERRNO(ioctl(outfd, FICLONE, infd));
}

assert_cc(sizeof(struct file_clone_range) == sizeof(struct btrfs_ioctl_clone_range_args));

int reflink_range(int infd, uint64_t in_offset, int outfd, uint64_t out_offset, uint64_t size) {
        int r;

        assert(infd >= 0);
        assert(outfd >= 0);

        /* size==UINT64_MAX mean "clone everything". Translate to 0 for the kernel. */
        if (size == UINT64_MAX)
                size = 0;

        /* Inside the kernel, FICLONE is identical to FICLONERANGE with offsets and size set to zero, let's
         * simplify things and use the simple ioctl in that case. */
        if (in_offset == 0 && out_offset == 0 && size == 0)
                return reflink(infd, outfd);

        r = fd_verify_regular(outfd);
        if (r < 0)
                return r;

        assert_cc(FICLONERANGE == BTRFS_IOC_CLONE_RANGE);

        struct file_clone_range args = {
                .src_fd = infd,
                .src_offset = in_offset,
                .src_length = size,
                .dest_offset = out_offset,
        };
        return RET_NERRNO(ioctl(outfd, FICLONERANGE, &args));
}
