posix.c

// Part of the Wasmtime Project, under the Apache License v2.0 with LLVM
// Exceptions. See
// https://github.com/bytecodealliance/wasmtime/blob/main/LICENSE for license
// information.
//
// Significant parts of this file are derived from cloudabi-utils. See
// https://github.com/bytecodealliance/wasmtime/blob/main/lib/wasi/sandboxed-system-primitives/src/LICENSE
// for license information.
//
// The upstream file contains the following copyright notice:
//
// Copyright (c) 2016-2018 Nuxi, https://nuxi.nl/

#include "ssp_config.h"
#include "bh_platform.h"
#include "blocking_op.h"
#include "wasmtime_ssp.h"
#include "libc_errno.h"
#include "locking.h"
#include "posix.h"
#include "random.h"
#include "refcount.h"
#include "rights.h"
#include "str.h"

/* Some platforms (e.g. Windows) already define `min()` macro.
 We're undefing it here to make sure the `min` call does exactly
 what we want it to do. */
#ifdef min
#undef min
#endif
static inline size_t
min(size_t a, size_t b)
{
    return a > b ? b : a;
}

#if 0 /* TODO: -std=gnu99 causes compile error, comment them first */
// struct iovec must have the same layout as __wasi_iovec_t.
static_assert(offsetof(struct iovec, iov_base) ==
                  offsetof(__wasi_iovec_t, buf),
              "Offset mismatch");
static_assert(sizeof(((struct iovec *)0)->iov_base) ==
                  sizeof(((__wasi_iovec_t *)0)->buf),
              "Size mismatch");
static_assert(offsetof(struct iovec, iov_len) ==
                  offsetof(__wasi_iovec_t, buf_len),
              "Offset mismatch");
static_assert(sizeof(((struct iovec *)0)->iov_len) ==
                  sizeof(((__wasi_iovec_t *)0)->buf_len),
              "Size mismatch");
static_assert(sizeof(struct iovec) == sizeof(__wasi_iovec_t),
              "Size mismatch");

// struct iovec must have the same layout as __wasi_ciovec_t.
static_assert(offsetof(struct iovec, iov_base) ==
                  offsetof(__wasi_ciovec_t, buf),
              "Offset mismatch");
static_assert(sizeof(((struct iovec *)0)->iov_base) ==
                  sizeof(((__wasi_ciovec_t *)0)->buf),
              "Size mismatch");
static_assert(offsetof(struct iovec, iov_len) ==
                  offsetof(__wasi_ciovec_t, buf_len),
              "Offset mismatch");
static_assert(sizeof(((struct iovec *)0)->iov_len) ==
                  sizeof(((__wasi_ciovec_t *)0)->buf_len),
              "Size mismatch");
static_assert(sizeof(struct iovec) == sizeof(__wasi_ciovec_t),
              "Size mismatch");
#endif

static bool
ns_lookup_list_search(char **list, const char *host)
{
    size_t host_len = strlen(host), suffix_len;

    while (*list) {
        if (*list[0] == '*') {
            suffix_len = strlen(*list) - 1;
            if (suffix_len <= host_len
                && strncmp(host + host_len - suffix_len, *list + 1, suffix_len)
                       == 0) {
                return true;
            }
        }
        else {
            if (strcmp(*list, host) == 0) {
                return true;
            }
        }
        list++;
    }

    return false;
}

#if !defined(BH_PLATFORM_WINDOWS) && CONFIG_HAS_CLOCK_NANOSLEEP
static bool
wasi_clockid_to_clockid(__wasi_clockid_t in, clockid_t *out)
{
    switch (in) {
        case __WASI_CLOCK_MONOTONIC:
            *out = CLOCK_MONOTONIC;
            return true;
#if defined(CLOCK_PROCESS_CPUTIME_ID)
        case __WASI_CLOCK_PROCESS_CPUTIME_ID:
            *out = CLOCK_PROCESS_CPUTIME_ID;
            return true;
#endif
        case __WASI_CLOCK_REALTIME:
            *out = CLOCK_REALTIME;
            return true;
#if defined(CLOCK_THREAD_CPUTIME_ID)
        case __WASI_CLOCK_THREAD_CPUTIME_ID:
            *out = CLOCK_THREAD_CPUTIME_ID;
            return true;
#endif
        default:
            return false;
    }
}
#endif

static void
wasi_addr_to_bh_sockaddr(const __wasi_addr_t *wasi_addr,
                         bh_sockaddr_t *sockaddr)
{
    if (wasi_addr->kind == IPv4) {
        sockaddr->addr_buffer.ipv4 = (wasi_addr->addr.ip4.addr.n0 << 24)
                                     | (wasi_addr->addr.ip4.addr.n1 << 16)
                                     | (wasi_addr->addr.ip4.addr.n2 << 8)
                                     | wasi_addr->addr.ip4.addr.n3;
        sockaddr->is_ipv4 = true;
        sockaddr->port = wasi_addr->addr.ip4.port;
    }
    else {
        sockaddr->addr_buffer.ipv6[0] = wasi_addr->addr.ip6.addr.n0;
        sockaddr->addr_buffer.ipv6[1] = wasi_addr->addr.ip6.addr.n1;
        sockaddr->addr_buffer.ipv6[2] = wasi_addr->addr.ip6.addr.n2;
        sockaddr->addr_buffer.ipv6[3] = wasi_addr->addr.ip6.addr.n3;
        sockaddr->addr_buffer.ipv6[4] = wasi_addr->addr.ip6.addr.h0;
        sockaddr->addr_buffer.ipv6[5] = wasi_addr->addr.ip6.addr.h1;
        sockaddr->addr_buffer.ipv6[6] = wasi_addr->addr.ip6.addr.h2;
        sockaddr->addr_buffer.ipv6[7] = wasi_addr->addr.ip6.addr.h3;
        sockaddr->is_ipv4 = false;
        sockaddr->port = wasi_addr->addr.ip6.port;
    }
}

// Converts an IPv6 binary address object to WASI address object.
static void
bh_sockaddr_to_wasi_addr(const bh_sockaddr_t *sockaddr,
                         __wasi_addr_t *wasi_addr)
{
    if (sockaddr->is_ipv4) {
        wasi_addr->kind = IPv4;
        wasi_addr->addr.ip4.port = sockaddr->port;
        wasi_addr->addr.ip4.addr.n0 =
            (sockaddr->addr_buffer.ipv4 & 0xFF000000) >> 24;
        wasi_addr->addr.ip4.addr.n1 =
            (sockaddr->addr_buffer.ipv4 & 0x00FF0000) >> 16;
        wasi_addr->addr.ip4.addr.n2 =
            (sockaddr->addr_buffer.ipv4 & 0x0000FF00) >> 8;
        wasi_addr->addr.ip4.addr.n3 = (sockaddr->addr_buffer.ipv4 & 0x000000FF);
    }
    else {
        wasi_addr->kind = IPv6;
        wasi_addr->addr.ip6.port = sockaddr->port;
        wasi_addr->addr.ip6.addr.n0 = sockaddr->addr_buffer.ipv6[0];
        wasi_addr->addr.ip6.addr.n1 = sockaddr->addr_buffer.ipv6[1];
        wasi_addr->addr.ip6.addr.n2 = sockaddr->addr_buffer.ipv6[2];
        wasi_addr->addr.ip6.addr.n3 = sockaddr->addr_buffer.ipv6[3];
        wasi_addr->addr.ip6.addr.h0 = sockaddr->addr_buffer.ipv6[4];
        wasi_addr->addr.ip6.addr.h1 = sockaddr->addr_buffer.ipv6[5];
        wasi_addr->addr.ip6.addr.h2 = sockaddr->addr_buffer.ipv6[6];
        wasi_addr->addr.ip6.addr.h3 = sockaddr->addr_buffer.ipv6[7];
    }
}

static void
wasi_addr_ip_to_bh_ip_addr_buffer(__wasi_addr_ip_t *addr,
                                  bh_ip_addr_buffer_t *out)
{
    if (addr->kind == IPv4) {
        out->ipv4 = htonl((addr->addr.ip4.n0 << 24) | (addr->addr.ip4.n1 << 16)
                          | (addr->addr.ip4.n2 << 8) | addr->addr.ip4.n3);
    }
    else {
        out->ipv6[0] = htons(addr->addr.ip6.n0);
        out->ipv6[1] = htons(addr->addr.ip6.n1);
        out->ipv6[2] = htons(addr->addr.ip6.n2);
        out->ipv6[3] = htons(addr->addr.ip6.n3);
        out->ipv6[4] = htons(addr->addr.ip6.h0);
        out->ipv6[5] = htons(addr->addr.ip6.h1);
        out->ipv6[6] = htons(addr->addr.ip6.h2);
        out->ipv6[7] = htons(addr->addr.ip6.h3);
    }
}

struct fd_prestat {
    const char *dir;
};

bool
fd_prestats_init(struct fd_prestats *pt)
{
    if (!rwlock_initialize(&pt->lock))
        return false;
    pt->prestats = NULL;
    pt->size = 0;
    pt->used = 0;
    return true;
}

// Grows the preopened resource table to a required lower bound and a
// minimum number of free preopened resource table entries.
static __wasi_errno_t
fd_prestats_grow(struct fd_prestats *pt, size_t min, size_t incr)
    REQUIRES_EXCLUSIVE(pt->lock)
{
    if (pt->size <= min || pt->size < (pt->used + incr) * 2) {
        // Keep on doubling the table size until we've met our constraints.
        size_t size = pt->size == 0 ? 1 : pt->size;
        while (size <= min || size < (pt->used + incr) * 2)
            size *= 2;

        // Grow the file descriptor table's allocation.
        struct fd_prestat *prestats =
            wasm_runtime_malloc((uint32)(sizeof(*prestats) * size));
        if (prestats == NULL)
            return __WASI_ENOMEM;

        if (pt->prestats && pt->size > 0) {
            bh_memcpy_s(prestats, (uint32)(sizeof(*prestats) * size),
                        pt->prestats, (uint32)(sizeof(*prestats) * pt->size));
        }

        if (pt->prestats)
            wasm_runtime_free(pt->prestats);

        // Mark all new file descriptors as unused.
        for (size_t i = pt->size; i < size; ++i)
            prestats[i].dir = NULL;
        pt->prestats = prestats;
        pt->size = size;
    }
    return __WASI_ESUCCESS;
}

static __wasi_errno_t
fd_prestats_insert_locked(struct fd_prestats *pt, const char *dir,
                          __wasi_fd_t fd)
{
    // Grow the preopened resource table if needed.
    __wasi_errno_t error = fd_prestats_grow(pt, fd, 1);

    if (error != __WASI_ESUCCESS) {
        return error;
    }

    pt->prestats[fd].dir = bh_strdup(dir);

    if (pt->prestats[fd].dir == NULL)
        return __WASI_ENOMEM;

    return __WASI_ESUCCESS;
}

// Inserts a preopened resource record into the preopened resource table.
bool
fd_prestats_insert(struct fd_prestats *pt, const char *dir, __wasi_fd_t fd)
{
    rwlock_wrlock(&pt->lock);

    __wasi_errno_t error = fd_prestats_insert_locked(pt, dir, fd);

    rwlock_unlock(&pt->lock);

    return error == __WASI_ESUCCESS;
}

// Looks up a preopened resource table entry by number.
static __wasi_errno_t
fd_prestats_get_entry(struct fd_prestats *pt, __wasi_fd_t fd,
                      struct fd_prestat **ret) REQUIRES_SHARED(pt->lock)
{
    // Test for file descriptor existence.
    if ((size_t)fd >= pt->size)
        return __WASI_EBADF;
    struct fd_prestat *prestat = &pt->prestats[fd];
    if (prestat->dir == NULL)
        return __WASI_EBADF;

    *ret = prestat;
    return 0;
}

// Remove a preopened resource record from the preopened resource table by
// number
static __wasi_errno_t
fd_prestats_remove_entry(struct fd_prestats *pt, __wasi_fd_t fd)
{
    // Test for file descriptor existence.
    if ((size_t)fd >= pt->size)
        return __WASI_EBADF;
    struct fd_prestat *prestat = &pt->prestats[fd];

    if (prestat->dir != NULL) {
        wasm_runtime_free((void *)prestat->dir);
        prestat->dir = NULL;
    }

    return __WASI_ESUCCESS;
}

struct fd_object {
    struct refcount refcount;
    __wasi_filetype_t type;
    os_file_handle file_handle;

    // Keep track of whether this fd object refers to a stdio stream so we know
    // whether to close the underlying file handle when releasing the object.
    bool is_stdio;
    union {
        // Data associated with directory file descriptors.
        struct {
            struct mutex lock;         // Lock to protect members below.
            os_dir_stream handle;      // Directory handle.
            __wasi_dircookie_t offset; // Offset of the directory.
        } directory;
    };
};

struct fd_entry {
    struct fd_object *object;
    __wasi_rights_t rights_base;
    __wasi_rights_t rights_inheriting;
};

bool
fd_table_init(struct fd_table *ft)
{
    if (!rwlock_initialize(&ft->lock))
        return false;
    ft->entries = NULL;
    ft->size = 0;
    ft->used = 0;
    return true;
}

// Looks up a file descriptor table entry by number and required rights.
static __wasi_errno_t
fd_table_get_entry(struct fd_table *ft, __wasi_fd_t fd,
                   __wasi_rights_t rights_base,
                   __wasi_rights_t rights_inheriting, struct fd_entry **ret)
    REQUIRES_SHARED(ft->lock)
{
    // Test for file descriptor existence.
    if ((size_t)fd >= ft->size) {
        return __WASI_EBADF;
    }

    struct fd_entry *fe = &ft->entries[fd];
    if (fe->object == NULL) {
        return __WASI_EBADF;
    }

    // Validate rights.
    if ((~fe->rights_base & rights_base) != 0
        || (~fe->rights_inheriting & rights_inheriting) != 0) {
        return __WASI_ENOTCAPABLE;
    }
    *ret = fe;
    return 0;
}

// Grows the file descriptor table to a required lower bound and a
// minimum number of free file descriptor table entries.
static bool
fd_table_grow(struct fd_table *ft, size_t min, size_t incr)
    REQUIRES_EXCLUSIVE(ft->lock)
{
    if (ft->size <= min || ft->size < (ft->used + incr) * 2) {
        // Keep on doubling the table size until we've met our constraints.
        size_t size = ft->size == 0 ? 1 : ft->size;
        while (size <= min || size < (ft->used + incr) * 2)
            size *= 2;

        // Grow the file descriptor table's allocation.
        struct fd_entry *entries =
            wasm_runtime_malloc((uint32)(sizeof(*entries) * size));
        if (entries == NULL)
            return false;

        if (ft->entries && ft->size > 0) {
            bh_memcpy_s(entries, (uint32)(sizeof(*entries) * size), ft->entries,
                        (uint32)(sizeof(*entries) * ft->size));
        }

        if (ft->entries)
            wasm_runtime_free(ft->entries);

        // Mark all new file descriptors as unused.
        for (size_t i = ft->size; i < size; ++i)
            entries[i].object = NULL;
        ft->entries = entries;
        ft->size = size;
    }
    return true;
}

// Allocates a new file descriptor object.
static __wasi_errno_t
fd_object_new(__wasi_filetype_t type, bool is_stdio, struct fd_object **fo)
    TRYLOCKS_SHARED(0, (*fo)->refcount)
{
    *fo = wasm_runtime_malloc(sizeof(**fo));
    if (*fo == NULL)
        return __WASI_ENOMEM;
    refcount_init(&(*fo)->refcount, 1);
    (*fo)->type = type;
    (*fo)->file_handle = os_get_invalid_handle();
    (*fo)->is_stdio = is_stdio;
    return 0;
}

// Attaches a file descriptor to the file descriptor table.
static void
fd_table_attach(struct fd_table *ft, __wasi_fd_t fd, struct fd_object *fo,
                __wasi_rights_t rights_base, __wasi_rights_t rights_inheriting)
    REQUIRES_EXCLUSIVE(ft->lock) CONSUMES(fo->refcount)
{
    bh_assert(ft->size > fd && "File descriptor table too small");
    struct fd_entry *fe = &ft->entries[fd];
    bh_assert(fe->object == NULL
              && "Attempted to overwrite an existing descriptor");
    fe->object = fo;
    fe->rights_base = rights_base;
    fe->rights_inheriting = rights_inheriting;
    ++ft->used;
    bh_assert(ft->size >= ft->used * 2 && "File descriptor too full");
}

// Detaches a file descriptor from the file descriptor table.
static void
fd_table_detach(struct fd_table *ft, __wasi_fd_t fd, struct fd_object **fo)
    REQUIRES_EXCLUSIVE(ft->lock) PRODUCES((*fo)->refcount)
{
    bh_assert(ft->size > fd && "File descriptor table too small");
    struct fd_entry *fe = &ft->entries[fd];
    *fo = fe->object;
    bh_assert(*fo != NULL && "Attempted to detach nonexistent descriptor");
    fe->object = NULL;
    bh_assert(ft->used > 0 && "Reference count mismatch");
    --ft->used;
}

// Determines the type of a file descriptor and its maximum set of
// rights that should be attached to it.
static __wasi_errno_t
fd_determine_type_rights(os_file_handle fd, __wasi_filetype_t *type,
                         __wasi_rights_t *rights_base,
                         __wasi_rights_t *rights_inheriting)
{
    struct __wasi_filestat_t buf;
    __wasi_errno_t error;

    if (os_is_stdin_handle(fd)) {
        *rights_base = RIGHTS_STDIN;
        *rights_inheriting = RIGHTS_STDIN;
        return __WASI_ESUCCESS;
    }

    if (os_is_stdout_handle(fd)) {
        *rights_base = RIGHTS_STDOUT;
        *rights_inheriting = RIGHTS_STDOUT;
        return __WASI_ESUCCESS;
    }

    if (os_is_stderr_handle(fd)) {
        *rights_base = RIGHTS_STDERR;
        *rights_inheriting = RIGHTS_STDERR;
        return __WASI_ESUCCESS;
    }

    error = os_fstat(fd, &buf);
    if (error != __WASI_ESUCCESS)
        return error;

    *type = buf.st_filetype;

    switch (buf.st_filetype) {
        case __WASI_FILETYPE_BLOCK_DEVICE:
            *rights_base = RIGHTS_BLOCK_DEVICE_BASE;
            *rights_inheriting = RIGHTS_BLOCK_DEVICE_INHERITING;
            break;
        case __WASI_FILETYPE_CHARACTER_DEVICE:
            error = os_isatty(fd);

            if (error == __WASI_ESUCCESS) {
                *rights_base = RIGHTS_TTY_BASE;
                *rights_inheriting = RIGHTS_TTY_INHERITING;
            }
            else {
                *rights_base = RIGHTS_CHARACTER_DEVICE_BASE;
                *rights_inheriting = RIGHTS_CHARACTER_DEVICE_INHERITING;
            }
            break;
        case __WASI_FILETYPE_DIRECTORY:
            *rights_base = RIGHTS_DIRECTORY_BASE;
            *rights_inheriting = RIGHTS_DIRECTORY_INHERITING;
            break;
        case __WASI_FILETYPE_REGULAR_FILE:
            *rights_base = RIGHTS_REGULAR_FILE_BASE;
            *rights_inheriting = RIGHTS_REGULAR_FILE_INHERITING;
            break;
        case __WASI_FILETYPE_SOCKET_DGRAM:
        case __WASI_FILETYPE_SOCKET_STREAM:
            *rights_base = RIGHTS_SOCKET_BASE;
            *rights_inheriting = RIGHTS_SOCKET_INHERITING;
            break;
        case __WASI_FILETYPE_SYMBOLIC_LINK:
        case __WASI_FILETYPE_UNKNOWN:
            // If we don't know the type, allow for the maximum set of
            // rights
            *rights_base = RIGHTS_ALL;
            *rights_inheriting = RIGHTS_ALL;
            break;
        default:
            return __WASI_EINVAL;
    }

    wasi_libc_file_access_mode access_mode;
    error = os_file_get_access_mode(fd, &access_mode);

    if (error != __WASI_ESUCCESS)
        return error;

    // Strip off read/write bits based on the access mode.
    switch (access_mode) {
        case WASI_LIBC_ACCESS_MODE_READ_ONLY:
            *rights_base &= ~(__wasi_rights_t)__WASI_RIGHT_FD_WRITE;
            break;
        case WASI_LIBC_ACCESS_MODE_WRITE_ONLY:
            *rights_base &= ~(__wasi_rights_t)__WASI_RIGHT_FD_READ;
            break;
    }

    return error;
}

// Lowers the reference count on a file descriptor object. When the
// reference count reaches zero, its resources are cleaned up.
static __wasi_errno_t
fd_object_release(wasm_exec_env_t env, struct fd_object *fo)
    UNLOCKS(fo->refcount)
{
    __wasi_errno_t error = __WASI_ESUCCESS;

    if (refcount_release(&fo->refcount)) {
        int saved_errno = errno;
        switch (fo->type) {
            case __WASI_FILETYPE_DIRECTORY:
                // For directories we may keep track of a DIR object.
                // Calling os_closedir() on it also closes the underlying file
                // descriptor.
                mutex_destroy(&fo->directory.lock);
                if (os_is_dir_stream_valid(&fo->directory.handle)) {
                    error = os_closedir(fo->directory.handle);
                    break;
                }
                // Fallthrough.
            default:
                // The env == NULL case is for
                // fd_table_destroy, path_get, path_put,
                // fd_table_insert_existing
                error = (env == NULL) ? os_close(fo->file_handle, fo->is_stdio)
                                      : blocking_op_close(env, fo->file_handle,
                                                          fo->is_stdio);
                break;
        }
        wasm_runtime_free(fo);
        errno = saved_errno;
    }
    return error;
}

// Inserts an already existing file descriptor into the file descriptor
// table.
bool
fd_table_insert_existing(struct fd_table *ft, __wasi_fd_t in,
                         os_file_handle out, bool is_stdio)
{
    __wasi_filetype_t type = __WASI_FILETYPE_UNKNOWN;
    __wasi_rights_t rights_base = 0, rights_inheriting = 0;
    struct fd_object *fo;
    __wasi_errno_t error;

    error =
        fd_determine_type_rights(out, &type, &rights_base, &rights_inheriting);
    if (error != 0) {
#ifdef BH_PLATFORM_EGO
        /**
         * since it is an already opened file and we can assume the opened
         * file has all necessary rights no matter how to get
         */
        if (error != __WASI_ENOTSUP)
            return false;
#else
        return false;
#endif
    }

    error = fd_object_new(type, is_stdio, &fo);
    if (error != 0)
        return false;
    fo->file_handle = out;
    if (type == __WASI_FILETYPE_DIRECTORY) {
        if (!mutex_init(&fo->directory.lock)) {
            fd_object_release(NULL, fo);
            return false;
        }
        fo->directory.handle = os_get_invalid_dir_stream();
    }

    // Grow the file descriptor table if needed.
    rwlock_wrlock(&ft->lock);
    if (!fd_table_grow(ft, in, 1)) {
        rwlock_unlock(&ft->lock);
        fd_object_release(NULL, fo);
        return false;
    }

    fd_table_attach(ft, in, fo, rights_base, rights_inheriting);
    rwlock_unlock(&ft->lock);
    return true;
}

// Picks an unused slot from the file descriptor table.
static __wasi_errno_t
fd_table_unused(struct fd_table *ft, __wasi_fd_t *out) REQUIRES_SHARED(ft->lock)
{
    bh_assert(ft->size > ft->used && "File descriptor table has no free slots");
    for (;;) {
        uintmax_t random_fd = 0;
        __wasi_errno_t error = random_uniform(ft->size, &random_fd);

        if (error != __WASI_ESUCCESS)
            return error;

        if (ft->entries[(__wasi_fd_t)random_fd].object == NULL) {
            *out = (__wasi_fd_t)random_fd;
            return error;
        }
    }
}

// Inserts a file descriptor object into an unused slot of the file
// descriptor table.
static __wasi_errno_t
fd_table_insert(wasm_exec_env_t exec_env, struct fd_table *ft,
                struct fd_object *fo, __wasi_rights_t rights_base,
                __wasi_rights_t rights_inheriting, __wasi_fd_t *out)
    REQUIRES_UNLOCKED(ft->lock) UNLOCKS(fo->refcount)
{
    // Grow the file descriptor table if needed.
    rwlock_wrlock(&ft->lock);
    if (!fd_table_grow(ft, 0, 1)) {
        rwlock_unlock(&ft->lock);
        fd_object_release(exec_env, fo);
        return convert_errno(errno);
    }

    __wasi_errno_t error = fd_table_unused(ft, out);

    if (error != __WASI_ESUCCESS) {
        rwlock_unlock(&ft->lock);
        return error;
    }

    fd_table_attach(ft, *out, fo, rights_base, rights_inheriting);
    rwlock_unlock(&ft->lock);
    return error;
}

// Inserts a numerical file descriptor into the file descriptor table.
static __wasi_errno_t
fd_table_insert_fd(wasm_exec_env_t exec_env, struct fd_table *ft,
                   os_file_handle in, __wasi_filetype_t type,
                   __wasi_rights_t rights_base,
                   __wasi_rights_t rights_inheriting, __wasi_fd_t *out)
    REQUIRES_UNLOCKED(ft->lock)
{
    struct fd_object *fo;

    __wasi_errno_t error = fd_object_new(type, false, &fo);
    if (error != 0) {
        os_close(in, false);
        return error;
    }

    fo->file_handle = in;
    if (type == __WASI_FILETYPE_DIRECTORY) {
        if (!mutex_init(&fo->directory.lock)) {
            fd_object_release(exec_env, fo);
            return (__wasi_errno_t)-1;
        }
        fo->directory.handle = os_get_invalid_dir_stream();
    }
    return fd_table_insert(exec_env, ft, fo, rights_base, rights_inheriting,
                           out);
}

__wasi_errno_t
wasmtime_ssp_fd_prestat_get(struct fd_prestats *prestats, __wasi_fd_t fd,
                            __wasi_prestat_t *buf)
{
    rwlock_rdlock(&prestats->lock);
    struct fd_prestat *prestat;
    __wasi_errno_t error = fd_prestats_get_entry(prestats, fd, &prestat);
    if (error != 0) {
        rwlock_unlock(&prestats->lock);
        return error;
    }

    *buf = (__wasi_prestat_t){
        .pr_type = __WASI_PREOPENTYPE_DIR,
    };

    buf->u.dir.pr_name_len = strlen(prestat->dir);

    rwlock_unlock(&prestats->lock);

    return 0;
}

__wasi_errno_t
wasmtime_ssp_fd_prestat_dir_name(struct fd_prestats *prestats, __wasi_fd_t fd,
                                 char *path, size_t path_len)
{
    rwlock_rdlock(&prestats->lock);
    struct fd_prestat *prestat;
    __wasi_errno_t error = fd_prestats_get_entry(prestats, fd, &prestat);
    if (error != 0) {
        rwlock_unlock(&prestats->lock);
        return error;
    }

    const size_t prestat_dir_len = strlen(prestat->dir);
    if (path_len < prestat_dir_len) {
        rwlock_unlock(&prestats->lock);
        return __WASI_EINVAL;
    }

    bh_memcpy_s(path, (uint32)path_len, prestat->dir, (uint32)prestat_dir_len);

    rwlock_unlock(&prestats->lock);

    return 0;
}

__wasi_errno_t
wasmtime_ssp_fd_close(wasm_exec_env_t exec_env, struct fd_table *curfds,
                      struct fd_prestats *prestats, __wasi_fd_t fd)
{
    // Validate the file descriptor.
    struct fd_table *ft = curfds;
    rwlock_wrlock(&ft->lock);
    rwlock_wrlock(&prestats->lock);

    struct fd_entry *fe;
    __wasi_errno_t error = fd_table_get_entry(ft, fd, 0, 0, &fe);
    if (error != 0) {
        rwlock_unlock(&prestats->lock);
        rwlock_unlock(&ft->lock);
        return error;
    }

    // Remove it from the file descriptor table.
    struct fd_object *fo;
    fd_table_detach(ft, fd, &fo);

    // Remove it from the preopened resource table if it exists
    error = fd_prestats_remove_entry(prestats, fd);

    rwlock_unlock(&prestats->lock);
    rwlock_unlock(&ft->lock);
    fd_object_release(exec_env, fo);

    // Ignore the error if there is no preopen associated with this fd
    if (error == __WASI_EBADF) {
        return __WASI_ESUCCESS;
    }

    return error;
}

// Look up a file descriptor object in a locked file descriptor table
// and increases its reference count.
static __wasi_errno_t
fd_object_get_locked(struct fd_object **fo, struct fd_table *ft, __wasi_fd_t fd,
                     __wasi_rights_t rights_base,
                     __wasi_rights_t rights_inheriting)
    TRYLOCKS_EXCLUSIVE(0, (*fo)->refcount) REQUIRES_EXCLUSIVE(ft->lock)
{
    // Test whether the file descriptor number is valid.
    struct fd_entry *fe;
    __wasi_errno_t error =
        fd_table_get_entry(ft, fd, rights_base, rights_inheriting, &fe);
    if (error != 0)
        return error;

    // Increase the reference count on the file descriptor object. A copy
    // of the rights are also stored, so callers can still access those if
    // needed.
    *fo = fe->object;
    refcount_acquire(&(*fo)->refcount);
    return 0;
}

// Temporarily locks the file descriptor table to look up a file
// descriptor object, increases its reference count and drops the lock.
static __wasi_errno_t
fd_object_get(struct fd_table *curfds, struct fd_object **fo, __wasi_fd_t fd,
              __wasi_rights_t rights_base, __wasi_rights_t rights_inheriting)
    TRYLOCKS_EXCLUSIVE(0, (*fo)->refcount)
{
    struct fd_table *ft = curfds;
    rwlock_rdlock(&ft->lock);
    __wasi_errno_t error =
        fd_object_get_locked(fo, ft, fd, rights_base, rights_inheriting);
    rwlock_unlock(&ft->lock);
    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_datasync(wasm_exec_env_t exec_env, struct fd_table *curfds,
                         __wasi_fd_t fd)
{
    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_FD_DATASYNC, 0);
    if (error != 0)
        return error;

    error = os_fdatasync(fo->file_handle);

    fd_object_release(exec_env, fo);

    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_pread(wasm_exec_env_t exec_env, struct fd_table *curfds,
                      __wasi_fd_t fd, const __wasi_iovec_t *iov, size_t iovcnt,
                      __wasi_filesize_t offset, size_t *nread)
{
    if (iovcnt == 0)
        return __WASI_EINVAL;

    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_FD_READ, 0);

    if (error != 0)
        return error;

    error = blocking_op_preadv(exec_env, fo->file_handle, iov, (int)iovcnt,
                               offset, nread);

    fd_object_release(exec_env, fo);

    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_pwrite(wasm_exec_env_t exec_env, struct fd_table *curfds,
                       __wasi_fd_t fd, const __wasi_ciovec_t *iov,
                       size_t iovcnt, __wasi_filesize_t offset,
                       size_t *nwritten)
{
    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_FD_WRITE, 0);

    if (error != 0)
        return error;

    error = blocking_op_pwritev(exec_env, fo->file_handle, iov, (int)iovcnt,
                                offset, nwritten);
    fd_object_release(exec_env, fo);

    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_read(wasm_exec_env_t exec_env, struct fd_table *curfds,
                     __wasi_fd_t fd, const __wasi_iovec_t *iov, size_t iovcnt,
                     size_t *nread)
{
    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_FD_READ, 0);

    if (error != 0)
        return error;

    error =
        blocking_op_readv(exec_env, fo->file_handle, iov, (int)iovcnt, nread);

    fd_object_release(exec_env, fo);

    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_renumber(wasm_exec_env_t exec_env, struct fd_table *curfds,
                         struct fd_prestats *prestats, __wasi_fd_t from,
                         __wasi_fd_t to)
{
    struct fd_table *ft = curfds;
    rwlock_wrlock(&ft->lock);
    rwlock_wrlock(&prestats->lock);

    struct fd_entry *fe_from;
    __wasi_errno_t error = fd_table_get_entry(ft, from, 0, 0, &fe_from);
    if (error != 0) {
        rwlock_unlock(&prestats->lock);
        rwlock_unlock(&ft->lock);
        return error;
    }
    struct fd_entry *fe_to;
    error = fd_table_get_entry(ft, to, 0, 0, &fe_to);
    if (error != 0) {
        rwlock_unlock(&prestats->lock);
        rwlock_unlock(&ft->lock);
        return error;
    }

    struct fd_object *fo;
    fd_table_detach(ft, to, &fo);
    refcount_acquire(&fe_from->object->refcount);
    fd_table_attach(ft, to, fe_from->object, fe_from->rights_base,
                    fe_from->rights_inheriting);
    fd_object_release(exec_env, fo);

    // Remove the old fd from the file descriptor table.
    fd_table_detach(ft, from, &fo);
    fd_object_release(exec_env, fo);
    --ft->used;

    // Handle renumbering of any preopened resources
    struct fd_prestat *prestat_from;
    __wasi_errno_t prestat_from_error =
        fd_prestats_get_entry(prestats, from, &prestat_from);

    struct fd_prestat *prestat_to;
    __wasi_errno_t prestat_to_error =
        fd_prestats_get_entry(prestats, to, &prestat_to);

    // Renumbering over two preopened resources.
    if (prestat_from_error == __WASI_ESUCCESS
        && prestat_to_error == __WASI_ESUCCESS) {
        (void)fd_prestats_remove_entry(prestats, to);

        error = fd_prestats_insert_locked(prestats, prestat_from->dir, to);

        if (error == __WASI_ESUCCESS) {
            (void)fd_prestats_remove_entry(prestats, from);
        }
        else {
            (void)fd_prestats_remove_entry(prestats, to);
        }
    }
    // Renumbering from a non-preopened fd to a preopened fd. In this case,
    // we can't a keep the destination fd entry in the preopened table so
    // remove it entirely.
    else if (prestat_from_error != __WASI_ESUCCESS
             && prestat_to_error == __WASI_ESUCCESS) {
        (void)fd_prestats_remove_entry(prestats, to);
    }
    // Renumbering from a preopened fd to a non-preopened fd
    else if (prestat_from_error == __WASI_ESUCCESS
             && prestat_to_error != __WASI_ESUCCESS) {
        error = fd_prestats_insert_locked(prestats, prestat_from->dir, to);

        if (error == __WASI_ESUCCESS) {
            (void)fd_prestats_remove_entry(prestats, from);
        }
        else {
            (void)fd_prestats_remove_entry(prestats, to);
        }
    }

    rwlock_unlock(&prestats->lock);
    rwlock_unlock(&ft->lock);

    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_seek(wasm_exec_env_t exec_env, struct fd_table *curfds,
                     __wasi_fd_t fd, __wasi_filedelta_t offset,
                     __wasi_whence_t whence, __wasi_filesize_t *newoffset)
{
    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd,
                      offset == 0 && whence == __WASI_WHENCE_CUR
                          ? __WASI_RIGHT_FD_TELL
                          : __WASI_RIGHT_FD_SEEK | __WASI_RIGHT_FD_TELL,
                      0);
    if (error != 0)
        return error;

    error = os_lseek(fo->file_handle, offset, whence, newoffset);

    fd_object_release(exec_env, fo);

    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_tell(wasm_exec_env_t exec_env, struct fd_table *curfds,
                     __wasi_fd_t fd, __wasi_filesize_t *newoffset)
{
    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_FD_TELL, 0);
    if (error != 0)
        return error;

    error = os_lseek(fo->file_handle, 0, __WASI_WHENCE_CUR, newoffset);

    fd_object_release(exec_env, fo);

    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_fdstat_get(wasm_exec_env_t exec_env, struct fd_table *curfds,
                           __wasi_fd_t fd, __wasi_fdstat_t *buf)
{
    struct fd_table *ft = curfds;
    struct fd_entry *fe;
    __wasi_errno_t error;

    (void)exec_env;

    rwlock_rdlock(&ft->lock);
    error = fd_table_get_entry(ft, fd, 0, 0, &fe);
    if (error != __WASI_ESUCCESS) {
        rwlock_unlock(&ft->lock);
        return error;
    }

    // Extract file descriptor type and rights.
    struct fd_object *fo = fe->object;

    __wasi_fdflags_t flags;
    error = os_file_get_fdflags(fo->file_handle, &flags);

    if (error != __WASI_ESUCCESS) {
        rwlock_unlock(&ft->lock);
        return error;
    }

    *buf = (__wasi_fdstat_t){ .fs_filetype = fo->type,
                              .fs_rights_base = fe->rights_base,
                              .fs_rights_inheriting = fe->rights_inheriting,
                              .fs_flags = flags };

    rwlock_unlock(&ft->lock);
    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_fdstat_set_flags(wasm_exec_env_t exec_env,
                                 struct fd_table *curfds, __wasi_fd_t fd,
                                 __wasi_fdflags_t fs_flags)
{
    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_FD_FDSTAT_SET_FLAGS, 0);

    if (error != 0)
        return error;

    error = os_file_set_fdflags(fo->file_handle, fs_flags);

    fd_object_release(exec_env, fo);

    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_fdstat_set_rights(wasm_exec_env_t exec_env,
                                  struct fd_table *curfds, __wasi_fd_t fd,
                                  __wasi_rights_t fs_rights_base,
                                  __wasi_rights_t fs_rights_inheriting)
{
    struct fd_table *ft = curfds;
    struct fd_entry *fe;
    __wasi_errno_t error;

    (void)exec_env;

    rwlock_wrlock(&ft->lock);
    error =
        fd_table_get_entry(ft, fd, fs_rights_base, fs_rights_inheriting, &fe);
    if (error != 0) {
        rwlock_unlock(&ft->lock);
        return error;
    }

    // Restrict the rights on the file descriptor.
    fe->rights_base = fs_rights_base;
    fe->rights_inheriting = fs_rights_inheriting;
    rwlock_unlock(&ft->lock);
    return 0;
}

__wasi_errno_t
wasmtime_ssp_fd_sync(wasm_exec_env_t exec_env, struct fd_table *curfds,
                     __wasi_fd_t fd)
{
    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_FD_SYNC, 0);

    if (error != 0)
        return error;

    error = os_fsync(fo->file_handle);

    fd_object_release(exec_env, fo);

    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_write(wasm_exec_env_t exec_env, struct fd_table *curfds,
                      __wasi_fd_t fd, const __wasi_ciovec_t *iov, size_t iovcnt,
                      size_t *nwritten)
{
    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_FD_WRITE, 0);
    if (error != 0)
        return error;

#ifndef BH_VPRINTF
    error = blocking_op_writev(exec_env, fo->file_handle, iov, (int)iovcnt,
                               nwritten);
#else
    /* redirect stdout/stderr output to BH_VPRINTF function */
    if (fo->is_stdio) {
        int i;
        *nwritten = 0;
        for (i = 0; i < (int)iovcnt; i++) {
            if (iov[i].buf_len > 0 && iov[i].buf != NULL) {
                char format[16];

                /* make up format string "%.ns" */
                snprintf(format, sizeof(format), "%%.%ds", (int)iov[i].buf_len);
                *nwritten += (size_t)os_printf(format, iov[i].buf);
            }
        }
    }
    else {
        error = blocking_op_writev(exec_env, fo->file_handle, iov, (int)iovcnt,
                                   nwritten);
    }
#endif /* end of BH_VPRINTF */
    fd_object_release(exec_env, fo);

    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_advise(wasm_exec_env_t exec_env, struct fd_table *curfds,
                       __wasi_fd_t fd, __wasi_filesize_t offset,
                       __wasi_filesize_t len, __wasi_advice_t advice)
{
    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_FD_ADVISE, 0);
    if (error != 0)
        return error;

    if (fo->type == __WASI_FILETYPE_DIRECTORY) {
        fd_object_release(exec_env, fo);
        return __WASI_EBADF;
    }

    error = os_fadvise(fo->file_handle, offset, len, advice);

    fd_object_release(exec_env, fo);

    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_allocate(wasm_exec_env_t exec_env, struct fd_table *curfds,
                         __wasi_fd_t fd, __wasi_filesize_t offset,
                         __wasi_filesize_t len)
{
    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_FD_ALLOCATE, 0);
    if (error != __WASI_ESUCCESS)
        return error;

    error = os_fallocate(fo->file_handle, offset, len);

    fd_object_release(exec_env, fo);

    return error;
}

// Reads the entire contents of a symbolic link, returning the contents
// in an allocated buffer. The allocated buffer is large enough to fit
// at least one extra byte, so the caller may append a trailing slash to
// it. This is needed by path_get().
__wasi_errno_t
readlinkat_dup(os_file_handle handle, const char *path, size_t *p_len,
               char **out_buf)
{
    __wasi_errno_t error;
    struct __wasi_filestat_t stat = { 0 };
    size_t buf_len;

    /*
     * use fstatat to get a better estimation
     * If path is a symbolic link, do not dereference it:
     * instead return information about the link itself,
     * like lstat().
     */
    error = os_fstatat(handle, path, &stat, 0);
    if (error != __WASI_ESUCCESS) {
        stat.st_size = 0;
    }

    /*
     * Some magic symlinks report `st_size` as zero. In that case, take
     * 32 as the initial buffer size. Otherwise, use `st_size + 1`.
     */
    buf_len = stat.st_size ? stat.st_size + 1 : 32;
    for (;;) {
        size_t bytes_read = 0;
        char *buf;

        buf = wasm_runtime_malloc((uint32)buf_len);
        if (buf == NULL) {
            *out_buf = NULL;
            return __WASI_ENOMEM;
        }

        error = os_readlinkat(handle, path, buf, buf_len, &bytes_read);
        if (error != __WASI_ESUCCESS) {
            wasm_runtime_free(buf);
            *p_len = 0;
            *out_buf = NULL;
            return error;
        }

        /* not truncated */
        if (bytes_read < buf_len) {
            buf[bytes_read] = '\0';
            *p_len = bytes_read + 1;
            *out_buf = buf;
            return __WASI_ESUCCESS;
        }

        /* truncated, try again with a bigger buf */
        wasm_runtime_free(buf);
        buf = NULL;
        buf_len *= 2;
    }
}

// Lease to a directory, so a path underneath it can be accessed.
//
// This structure is used by system calls that operate on pathnames. In
// this environment, pathnames always consist of a pair of a file
// descriptor representing the directory where the lookup needs to start
// and the actual pathname string.
struct path_access {
    os_file_handle fd;           // Directory file descriptor.
    const char *path;            // Pathname.
    bool follow;                 // Whether symbolic links should be followed.
    char *path_start;            // Internal: pathname to free.
    struct fd_object *fd_object; // Internal: directory file descriptor object.
};

// Creates a lease to a file descriptor and pathname pair. If the
// operating system does not implement Capsicum, it also normalizes the
// pathname to ensure the target path is placed underneath the
// directory.
static __wasi_errno_t
path_get(wasm_exec_env_t exec_env, struct fd_table *curfds,
         struct path_access *pa, __wasi_fd_t fd, __wasi_lookupflags_t flags,
         const char *upath, size_t upathlen, __wasi_rights_t rights_base,
         __wasi_rights_t rights_inheriting, bool needs_final_component)
    TRYLOCKS_EXCLUSIVE(0, pa->fd_object->refcount)
{
    char *path = str_nullterminate(upath, upathlen);
    if (path == NULL)
        return convert_errno(errno);

    // Fetch the directory file descriptor.
    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, rights_base, rights_inheriting);
    if (error != 0) {
        wasm_runtime_free(path);
        return error;
    }

#if CONFIG_HAS_CAP_ENTER
    // Rely on the kernel to constrain access to automatically constrain
    // access to files stored underneath this directory.
    pa->fd = fo->file_handle;
    pa->path = pa->path_start = path;
    pa->follow = (flags & __WASI_LOOKUP_SYMLINK_FOLLOW) != 0;
    pa->fd_object = fo;
    return 0;
#else
    // The implementation provides no mechanism to constrain lookups to a
    // directory automatically. Emulate this logic by resolving the
    // pathname manually.

    // Stack of directory file descriptors. Index 0 always corresponds
    // with the directory provided to this function. Entering a directory
    // causes a file descriptor to be pushed, while handling ".." entries
    // causes an entry to be popped. Index 0 cannot be popped, as this
    // would imply escaping the base directory.
    os_file_handle fds[128];
    fds[0] = fo->file_handle;
    size_t curfd = 0;

    // Stack of pathname strings used for symlink expansion. By using a
    // stack, there is no need to concatenate any pathname strings while
    // expanding symlinks.
    char *paths[32];
    char *paths_start[32];
    paths[0] = paths_start[0] = path;
    size_t curpath = 0;
    size_t expansions = 0;
    char *symlink = NULL;
    size_t symlink_len;
#ifdef BH_PLATFORM_WINDOWS
#define PATH_SEPARATORS "/\\"
#else
#define PATH_SEPARATORS "/"
#endif

    for (;;) {
        // Extract the next pathname component from 'paths[curpath]', null
        // terminate it and store it in 'file'. 'ends_with_slashes' stores
        // whether the pathname component is followed by one or more
        // trailing slashes, as this requires it to be a directory.
        char *file = paths[curpath];
        char *file_end = file + strcspn(file, PATH_SEPARATORS);
        paths[curpath] = file_end + strspn(file_end, PATH_SEPARATORS);
        bool ends_with_slashes =
            (*file_end != '\0' && strchr(PATH_SEPARATORS, *file_end));
        *file_end = '\0';

        // Test for empty pathname strings and absolute paths.
        if (file == file_end) {
            error = ends_with_slashes ? __WASI_ENOTCAPABLE : __WASI_ENOENT;
            goto fail;
        }

        if (strcmp(file, ".") == 0) {
            // Skip component.
        }
        else if (strcmp(file, "..") == 0) {
            // Pop a directory off the stack.
            if (curfd == 0) {
                // Attempted to go to parent directory of the directory file
                // descriptor.
                error = __WASI_ENOTCAPABLE;
                goto fail;
            }
            error = os_close(fds[curfd--], false);

            if (error != __WASI_ESUCCESS)
                goto fail;
        }
        else if (curpath > 0 || *paths[curpath] != '\0'
                 || (ends_with_slashes && !needs_final_component)) {
            // A pathname component whose name we're not interested in that is
            // followed by a slash or is followed by other pathname
            // components. In other words, a pathname component that must be a
            // directory. First attempt to obtain a directory file descriptor
            // for it.
            os_file_handle newdir;
            error = blocking_op_openat(
                exec_env, fds[curfd], file, __WASI_O_DIRECTORY, 0, 0,
                WASI_LIBC_ACCESS_MODE_READ_ONLY, &newdir);
            if (error == __WASI_ESUCCESS) {
                // Success. Push it onto the directory stack.
                if (curfd + 1 == sizeof(fds) / sizeof(fds[0])) {
                    os_close(newdir, false);
                    error = __WASI_ENAMETOOLONG;
                    goto fail;
                }
                fds[++curfd] = newdir;
            }
            else {
                // Failed to open it. Attempt symlink expansion.
                if (error != __WASI_ELOOP && error != __WASI_EMLINK
                    && error != __WASI_ENOTDIR) {
                    goto fail;
                }
                error =
                    readlinkat_dup(fds[curfd], file, &symlink_len, &symlink);

                if (error == __WASI_ESUCCESS) {
                    bh_assert(symlink != NULL);
                    goto push_symlink;
                }

                // readlink returns EINVAL if the path isn't a symlink. In that
                // case, it's more informative to return ENOTDIR.
                if (error == __WASI_EINVAL)
                    error = __WASI_ENOTDIR;

                goto fail;
            }
        }
        else {
            // The final pathname component. Depending on whether it ends with
            // a slash or the symlink-follow flag is set, perform symlink
            // expansion.
            if (ends_with_slashes
                || (flags & __WASI_LOOKUP_SYMLINK_FOLLOW) != 0) {
                error =
                    readlinkat_dup(fds[curfd], file, &symlink_len, &symlink);
                if (error == __WASI_ESUCCESS) {
                    bh_assert(symlink != NULL);
                    goto push_symlink;
                }
                if (error != __WASI_EINVAL && error != __WASI_ENOENT) {
                    goto fail;
                }
            }

            // Not a symlink, meaning we're done. Return the filename,
            // together with the directory containing this file.
            //
            // If the file was followed by a trailing slash, we must retain
            // it, to ensure system calls properly return ENOTDIR.
            // Unfortunately, this opens up a race condition, because this
            // means that users of path_get() will perform symlink expansion a
            // second time. There is nothing we can do to mitigate this, as
            // far as I know.
            if (ends_with_slashes)
                *file_end = '/';
            pa->path = file;
            pa->path_start = paths_start[0];
            goto success;
        }

        if (*paths[curpath] == '\0') {
            if (curpath == 0) {
                // No further pathname components to process. We may end up here
                // when called on paths like ".", "a/..", but also if the path
                // had trailing slashes and the caller is not interested in the
                // name of the pathname component.
                wasm_runtime_free(paths_start[0]);
                pa->path = ".";
                pa->path_start = NULL;
                goto success;
            }

            // Finished expanding symlink. Continue processing along the
            // original path.
            wasm_runtime_free(paths_start[curpath--]);
        }
        continue;

    push_symlink:
        // Prevent infinite loops by placing an upper limit on the number of
        // symlink expansions.
        if (++expansions == 128) {
            wasm_runtime_free(symlink);
            error = __WASI_ELOOP;
            goto fail;
        }

        if (*paths[curpath] == '\0') {
            // The original path already finished processing. Replace it by
            // this symlink entirely.
            wasm_runtime_free(paths_start[curpath]);
        }
        else if (curpath + 1 == sizeof(paths) / sizeof(paths[0])) {
            // Too many nested symlinks. Stop processing.
            wasm_runtime_free(symlink);
            error = __WASI_ELOOP;
            goto fail;
        }
        else {
            // The original path still has components left. Retain the
            // components that remain, so we can process them afterwards.
            ++curpath;
        }

        // Append a trailing slash to the symlink if the path leading up to
        // it also contained one. Otherwise we would not throw ENOTDIR if
        // the target is not a directory.
        if (ends_with_slashes)
            bh_strcat_s(symlink, (uint32)symlink_len, "/");
        paths[curpath] = paths_start[curpath] = symlink;
    }

success:
    // Return the lease. Close all directories, except the one the caller
    // needs to use.
    for (size_t i = 1; i < curfd; ++i)
        os_close(fds[i], false);
    pa->fd = fds[curfd];
    pa->follow = false;
    pa->fd_object = fo;
    return 0;

fail:
    // Failure. Free all resources.
    for (size_t i = 1; i <= curfd; ++i)
        os_close(fds[i], false);
    for (size_t i = 0; i <= curpath; ++i)
        wasm_runtime_free(paths_start[i]);
    fd_object_release(NULL, fo);
    return error;
#endif
}

static __wasi_errno_t
path_get_nofollow(wasm_exec_env_t exec_env, struct fd_table *curfds,
                  struct path_access *pa, __wasi_fd_t fd, const char *path,
                  size_t pathlen, __wasi_rights_t rights_base,
                  __wasi_rights_t rights_inheriting, bool needs_final_component)
    TRYLOCKS_EXCLUSIVE(0, pa->fd_object->refcount)
{
    __wasi_lookupflags_t flags = 0;
    return path_get(exec_env, curfds, pa, fd, flags, path, pathlen, rights_base,
                    rights_inheriting, needs_final_component);
}

static void
path_put(struct path_access *pa) UNLOCKS(pa->fd_object->refcount)
{
    if (pa->path_start)
        wasm_runtime_free(pa->path_start);
    /* Can't use `!=` operator when `os_file_handle` is a struct */
    if (!os_compare_file_handle(pa->fd_object->file_handle, pa->fd))
        os_close(pa->fd, false);
    fd_object_release(NULL, pa->fd_object);
}

__wasi_errno_t
wasmtime_ssp_path_create_directory(wasm_exec_env_t exec_env,
                                   struct fd_table *curfds, __wasi_fd_t fd,
                                   const char *path, size_t pathlen)
{
    struct path_access pa;
    __wasi_errno_t error =
        path_get_nofollow(exec_env, curfds, &pa, fd, path, pathlen,
                          __WASI_RIGHT_PATH_CREATE_DIRECTORY, 0, true);
    if (error != 0)
        return error;

    error = os_mkdirat(pa.fd, pa.path);
    path_put(&pa);

    return error;
}

static bool
validate_path(const char *path, struct fd_prestats *pt)
{
    size_t i;
    char path_resolved[PATH_MAX], prestat_dir_resolved[PATH_MAX];
    char *path_real, *prestat_dir_real;

    if (!(path_real = os_realpath(path, path_resolved)))
        /* path doesn't exist, creating a link to this file
           is allowed: if this file is to be created in
           the future, WASI will strictly check whether it
           can be created or not. */
        return true;

    for (i = 0; i < pt->size; i++) {
        if (pt->prestats[i].dir) {
            if (!(prestat_dir_real =
                      os_realpath(pt->prestats[i].dir, prestat_dir_resolved)))
                return false;
            if (!strncmp(path_real, prestat_dir_real, strlen(prestat_dir_real)))
                return true;
        }
    }

    return false;
}

__wasi_errno_t
wasmtime_ssp_path_link(wasm_exec_env_t exec_env, struct fd_table *curfds,
                       struct fd_prestats *prestats, __wasi_fd_t old_fd,
                       __wasi_lookupflags_t old_flags, const char *old_path,
                       size_t old_path_len, __wasi_fd_t new_fd,
                       const char *new_path, size_t new_path_len)
{
    struct path_access old_pa;
    __wasi_errno_t error =
        path_get(exec_env, curfds, &old_pa, old_fd, old_flags, old_path,
                 old_path_len, __WASI_RIGHT_PATH_LINK_SOURCE, 0, false);
    if (error != 0)
        return error;

    struct path_access new_pa;
    error =
        path_get_nofollow(exec_env, curfds, &new_pa, new_fd, new_path,
                          new_path_len, __WASI_RIGHT_PATH_LINK_TARGET, 0, true);
    if (error != 0) {
        path_put(&old_pa);
        return error;
    }

    rwlock_rdlock(&prestats->lock);
    if (!validate_path(old_pa.path, prestats)
        || !validate_path(new_pa.path, prestats)) {
        rwlock_unlock(&prestats->lock);
        return __WASI_EBADF;
    }
    rwlock_unlock(&prestats->lock);

    error = os_linkat(old_pa.fd, old_pa.path, new_pa.fd, new_pa.path,
                      old_pa.follow ? __WASI_LOOKUP_SYMLINK_FOLLOW : 0);

#if defined(__APPLE__)
    if (error == __WASI_ENOTSUP && !old_pa.follow) {
        // OS X doesn't allow creating hardlinks to symbolic links.
        // Duplicate the symbolic link instead.
        size_t target_len;
        char *target = NULL;
        error = readlinkat_dup(old_pa.fd, old_pa.path, &target_len, &target);
        if (error == __WASI_ESUCCESS) {
            bh_assert(target != NULL);
            bh_assert(target[target_len] == '\0');
            rwlock_rdlock(&prestats->lock);
            if (!validate_path(target, prestats)) {
                rwlock_unlock(&prestats->lock);
                wasm_runtime_free(target);
                return __WASI_EBADF;
            }
            rwlock_unlock(&prestats->lock);
            error = os_symlinkat(target, new_pa.fd, new_pa.path);
            wasm_runtime_free(target);
        }
    }
#endif

    path_put(&old_pa);
    path_put(&new_pa);

    return error;
}

__wasi_errno_t
wasmtime_ssp_path_open(wasm_exec_env_t exec_env, struct fd_table *curfds,
                       __wasi_fd_t dirfd, __wasi_lookupflags_t dirflags,
                       const char *path, size_t pathlen, __wasi_oflags_t oflags,
                       __wasi_rights_t fs_rights_base,
                       __wasi_rights_t fs_rights_inheriting,
                       __wasi_fdflags_t fs_flags, __wasi_fd_t *fd)
{
    // Rights that should be installed on the new file descriptor.
    __wasi_rights_t rights_base = fs_rights_base;
    __wasi_rights_t rights_inheriting = fs_rights_inheriting;

    // Which open() mode should be used to satisfy the needed rights.
    bool read =
        (rights_base & (__WASI_RIGHT_FD_READ | __WASI_RIGHT_FD_READDIR)) != 0;
    bool write =
        (rights_base
         & (__WASI_RIGHT_FD_DATASYNC | __WASI_RIGHT_FD_WRITE
            | __WASI_RIGHT_FD_ALLOCATE | __WASI_RIGHT_FD_FILESTAT_SET_SIZE))
        != 0;

    wasi_libc_file_access_mode access_mode =
        write ? read ? WASI_LIBC_ACCESS_MODE_READ_WRITE
                     : WASI_LIBC_ACCESS_MODE_WRITE_ONLY
              : WASI_LIBC_ACCESS_MODE_READ_ONLY;

    // Which rights are needed on the directory file descriptor.
    __wasi_rights_t needed_base = __WASI_RIGHT_PATH_OPEN;
    __wasi_rights_t needed_inheriting = rights_base | rights_inheriting;

    // Convert open flags.
    if ((oflags & __WASI_O_CREAT) != 0) {
        needed_base |= __WASI_RIGHT_PATH_CREATE_FILE;
    }
    if ((oflags & __WASI_O_TRUNC) != 0) {
        needed_base |= __WASI_RIGHT_PATH_FILESTAT_SET_SIZE;
    }

    // Convert file descriptor flags.
    if ((fs_flags & __WASI_FDFLAG_SYNC) != 0) {
        needed_inheriting |= __WASI_RIGHT_FD_SYNC;
    }
    if ((fs_flags & __WASI_FDFLAG_RSYNC) != 0) {
        needed_inheriting |= __WASI_RIGHT_FD_SYNC;
    }
    if ((fs_flags & __WASI_FDFLAG_DSYNC) != 0) {
        needed_inheriting |= __WASI_RIGHT_FD_DATASYNC;
    }

    if (write
        && !((fs_flags & __WASI_FDFLAG_APPEND) || (__WASI_O_TRUNC & oflags)))
        needed_inheriting |= __WASI_RIGHT_FD_SEEK;

    struct path_access pa;
    __wasi_errno_t error = path_get(
        exec_env, curfds, &pa, dirfd, dirflags, path, pathlen, needed_base,
        needed_inheriting, (oflags & __WASI_O_CREAT) != 0);

    if (error != 0)
        return error;

    os_file_handle handle;
    error = blocking_op_openat(exec_env, pa.fd, pa.path, oflags, fs_flags,
                               dirflags, access_mode, &handle);

    path_put(&pa);

    if (error != __WASI_ESUCCESS)
        return error;

    // Determine the type of the new file descriptor and which rights
    // contradict with this type.
    __wasi_filetype_t type;
    __wasi_rights_t max_base, max_inheriting;

    error = fd_determine_type_rights(handle, &type, &max_base, &max_inheriting);

    if (error != __WASI_ESUCCESS) {
        os_close(handle, false);
        return error;
    }

    return fd_table_insert_fd(exec_env, curfds, handle, type,
                              rights_base & max_base,
                              rights_inheriting & max_inheriting, fd);
}

// Copies out directory entry metadata or filename, potentially
// truncating it in the process.
static void
fd_readdir_put(void *buf, size_t bufsize, size_t *bufused, const void *elem,
               size_t elemsize)
{
    size_t bufavail = bufsize - *bufused;
    if (elemsize > bufavail)
        elemsize = bufavail;
    bh_memcpy_s((char *)buf + *bufused, (uint32)bufavail, elem,
                (uint32)elemsize);
    *bufused += elemsize;
}

__wasi_errno_t
wasmtime_ssp_fd_readdir(wasm_exec_env_t exec_env, struct fd_table *curfds,
                        __wasi_fd_t fd, void *buf, size_t nbyte,
                        __wasi_dircookie_t cookie, size_t *bufused)
{
    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_FD_READDIR, 0);
    if (error != 0) {
        return error;
    }

    // Create a directory handle if none has been opened yet.
    mutex_lock(&fo->directory.lock);
    if (!os_is_dir_stream_valid(&fo->directory.handle)) {
        error = os_fdopendir(fo->file_handle, &fo->directory.handle);
        if (error != __WASI_ESUCCESS) {
            mutex_unlock(&fo->directory.lock);
            fd_object_release(exec_env, fo);
            return error;
        }
        fo->directory.offset = __WASI_DIRCOOKIE_START;
    }

    // Seek to the right position if the requested offset does not match
    // the current offset.
    if (fo->directory.offset != cookie) {
        if (cookie == __WASI_DIRCOOKIE_START)
            os_rewinddir(fo->directory.handle);
        else
            os_seekdir(fo->directory.handle, cookie);
        fo->directory.offset = cookie;
    }

    *bufused = 0;
    while (*bufused < nbyte) {
        // Read the next directory entry.
        __wasi_dirent_t cde;
        const char *d_name = NULL;

        error = os_readdir(fo->directory.handle, &cde, &d_name);
        if (d_name == NULL) {
            mutex_unlock(&fo->directory.lock);
            fd_object_release(exec_env, fo);

            return *bufused > 0 ? __WASI_ESUCCESS : error;
        }

        fo->directory.offset = cde.d_next;

        fd_readdir_put(buf, nbyte, bufused, &cde, sizeof(cde));
        fd_readdir_put(buf, nbyte, bufused, d_name, cde.d_namlen);
    }
    mutex_unlock(&fo->directory.lock);
    fd_object_release(exec_env, fo);
    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasmtime_ssp_path_readlink(wasm_exec_env_t exec_env, struct fd_table *curfds,
                           __wasi_fd_t fd, const char *path, size_t pathlen,
                           char *buf, size_t bufsize, size_t *bufused)
{
    struct path_access pa;
    __wasi_errno_t error =
        path_get_nofollow(exec_env, curfds, &pa, fd, path, pathlen,
                          __WASI_RIGHT_PATH_READLINK, 0, false);

    if (error != 0)
        return error;

    error = os_readlinkat(pa.fd, pa.path, buf, bufsize, bufused);

    path_put(&pa);

    return error;
}

__wasi_errno_t
wasmtime_ssp_path_rename(wasm_exec_env_t exec_env, struct fd_table *curfds,
                         __wasi_fd_t old_fd, const char *old_path,
                         size_t old_path_len, __wasi_fd_t new_fd,
                         const char *new_path, size_t new_path_len)
{
    struct path_access old_pa;
    __wasi_errno_t error = path_get_nofollow(
        exec_env, curfds, &old_pa, old_fd, old_path, old_path_len,
        __WASI_RIGHT_PATH_RENAME_SOURCE, 0, true);
    if (error != 0)
        return error;

    struct path_access new_pa;
    error = path_get_nofollow(exec_env, curfds, &new_pa, new_fd, new_path,
                              new_path_len, __WASI_RIGHT_PATH_RENAME_TARGET, 0,
                              true);
    if (error != 0) {
        path_put(&old_pa);
        return error;
    }

    error = os_renameat(old_pa.fd, old_pa.path, new_pa.fd, new_pa.path);

    path_put(&old_pa);
    path_put(&new_pa);

    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_filestat_get(wasm_exec_env_t exec_env, struct fd_table *curfds,
                             __wasi_fd_t fd, __wasi_filestat_t *buf)
{
    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_FD_FILESTAT_GET, 0);

    if (error != 0)
        return error;

    error = os_fstat(fo->file_handle, buf);

    fd_object_release(exec_env, fo);

    return error;
}

static void
convert_timestamp(__wasi_timestamp_t in, os_timespec *out)
{
    // Store sub-second remainder.
#if defined(__SYSCALL_SLONG_TYPE)
    out->tv_nsec = (__SYSCALL_SLONG_TYPE)(in % 1000000000);
#else
    out->tv_nsec = (long)(in % 1000000000);
#endif
    __wasi_timestamp_t temp = in / 1000000000;

    // Clamp to the maximum in case it would overflow our system's time_t.
    out->tv_sec = (time_t)temp < BH_TIME_T_MAX ? (time_t)temp : BH_TIME_T_MAX;
}

__wasi_errno_t
wasmtime_ssp_fd_filestat_set_size(wasm_exec_env_t exec_env,
                                  struct fd_table *curfds, __wasi_fd_t fd,
                                  __wasi_filesize_t st_size)
{
    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_FD_FILESTAT_SET_SIZE, 0);

    if (error != 0)
        return error;

    error = os_ftruncate(fo->file_handle, st_size);
    fd_object_release(exec_env, fo);

    return error;
}

__wasi_errno_t
wasmtime_ssp_fd_filestat_set_times(wasm_exec_env_t exec_env,
                                   struct fd_table *curfds, __wasi_fd_t fd,
                                   __wasi_timestamp_t st_atim,
                                   __wasi_timestamp_t st_mtim,
                                   __wasi_fstflags_t fstflags)
{
    if ((fstflags
         & ~(__WASI_FILESTAT_SET_ATIM | __WASI_FILESTAT_SET_ATIM_NOW
             | __WASI_FILESTAT_SET_MTIM | __WASI_FILESTAT_SET_MTIM_NOW))
            != 0
        || (fstflags
            & (__WASI_FILESTAT_SET_ATIM | __WASI_FILESTAT_SET_ATIM_NOW))
               == (__WASI_FILESTAT_SET_ATIM | __WASI_FILESTAT_SET_ATIM_NOW)
        || (fstflags
            & (__WASI_FILESTAT_SET_MTIM | __WASI_FILESTAT_SET_MTIM_NOW))
               == (__WASI_FILESTAT_SET_MTIM | __WASI_FILESTAT_SET_MTIM_NOW))
        return __WASI_EINVAL;

    struct fd_object *fo;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_FD_FILESTAT_SET_TIMES, 0);
    if (error != 0)
        return error;

    error = os_futimens(fo->file_handle, st_atim, st_mtim, fstflags);

    fd_object_release(exec_env, fo);

    return error;
}

__wasi_errno_t
wasmtime_ssp_path_filestat_get(wasm_exec_env_t exec_env,
                               struct fd_table *curfds, __wasi_fd_t fd,
                               __wasi_lookupflags_t flags, const char *path,
                               size_t pathlen, __wasi_filestat_t *buf)
{
    struct path_access pa;
    __wasi_errno_t error =
        path_get(exec_env, curfds, &pa, fd, flags, path, pathlen,
                 __WASI_RIGHT_PATH_FILESTAT_GET, 0, false);
    if (error != 0)
        return error;

    error = os_fstatat(pa.fd, pa.path, buf,
                       pa.follow ? __WASI_LOOKUP_SYMLINK_FOLLOW : 0);

    path_put(&pa);

    return error;
}

__wasi_errno_t
wasmtime_ssp_path_filestat_set_times(wasm_exec_env_t exec_env,
                                     struct fd_table *curfds, __wasi_fd_t fd,
                                     __wasi_lookupflags_t flags,
                                     const char *path, size_t pathlen,
                                     __wasi_timestamp_t st_atim,
                                     __wasi_timestamp_t st_mtim,
                                     __wasi_fstflags_t fstflags)
{
    if (((fstflags
          & ~(__WASI_FILESTAT_SET_ATIM | __WASI_FILESTAT_SET_ATIM_NOW
              | __WASI_FILESTAT_SET_MTIM | __WASI_FILESTAT_SET_MTIM_NOW))
         != 0)
        /* ATIM & ATIM_NOW can't be set at the same time */
        || ((fstflags & __WASI_FILESTAT_SET_ATIM) != 0
            && (fstflags & __WASI_FILESTAT_SET_ATIM_NOW) != 0)
        /* MTIM & MTIM_NOW can't be set at the same time */
        || ((fstflags & __WASI_FILESTAT_SET_MTIM) != 0
            && (fstflags & __WASI_FILESTAT_SET_MTIM_NOW) != 0))
        return __WASI_EINVAL;

    struct path_access pa;
    __wasi_errno_t error =
        path_get(exec_env, curfds, &pa, fd, flags, path, pathlen,
                 __WASI_RIGHT_PATH_FILESTAT_SET_TIMES, 0, false);
    if (error != 0)
        return error;

    error = os_utimensat(pa.fd, pa.path, st_atim, st_mtim, fstflags,
                         pa.follow ? __WASI_LOOKUP_SYMLINK_FOLLOW : 0);

    path_put(&pa);
    return error;
}

__wasi_errno_t
wasmtime_ssp_path_symlink(wasm_exec_env_t exec_env, struct fd_table *curfds,
                          struct fd_prestats *prestats, const char *old_path,
                          size_t old_path_len, __wasi_fd_t fd,
                          const char *new_path, size_t new_path_len)
{
    char *target = str_nullterminate(old_path, old_path_len);
    if (target == NULL)
        return convert_errno(errno);

    struct path_access pa;
    __wasi_errno_t error =
        path_get_nofollow(exec_env, curfds, &pa, fd, new_path, new_path_len,
                          __WASI_RIGHT_PATH_SYMLINK, 0, true);
    if (error != 0) {
        wasm_runtime_free(target);
        return error;
    }

    rwlock_rdlock(&prestats->lock);
    if (!validate_path(target, prestats)) {
        rwlock_unlock(&prestats->lock);
        wasm_runtime_free(target);
        return __WASI_EBADF;
    }
    rwlock_unlock(&prestats->lock);

    error = os_symlinkat(target, pa.fd, pa.path);

    path_put(&pa);
    wasm_runtime_free(target);

    return error;
}

__wasi_errno_t
wasmtime_ssp_path_unlink_file(wasm_exec_env_t exec_env, struct fd_table *curfds,
                              __wasi_fd_t fd, const char *path, size_t pathlen)
{
    struct path_access pa;
    __wasi_errno_t error =
        path_get_nofollow(exec_env, curfds, &pa, fd, path, pathlen,
                          __WASI_RIGHT_PATH_UNLINK_FILE, 0, true);
    if (error != __WASI_ESUCCESS)
        return error;

    error = os_unlinkat(pa.fd, pa.path, false);

    path_put(&pa);

    return error;
}

__wasi_errno_t
wasmtime_ssp_path_remove_directory(wasm_exec_env_t exec_env,
                                   struct fd_table *curfds, __wasi_fd_t fd,
                                   const char *path, size_t pathlen)
{
    struct path_access pa;
    __wasi_errno_t error =
        path_get_nofollow(exec_env, curfds, &pa, fd, path, pathlen,
                          __WASI_RIGHT_PATH_REMOVE_DIRECTORY, 0, true);
    if (error != 0)
        return error;

    error = os_unlinkat(pa.fd, pa.path, true);

    path_put(&pa);

    return error;
}

__wasi_errno_t
wasmtime_ssp_poll_oneoff(wasm_exec_env_t exec_env, struct fd_table *curfds,
                         const __wasi_subscription_t *in, __wasi_event_t *out,
                         size_t nsubscriptions,
                         size_t *nevents) NO_LOCK_ANALYSIS
{
#if defined(BH_PLATFORM_WINDOWS)
    return __WASI_ENOSYS;
#else
    // Sleeping.
    if (nsubscriptions == 1 && in[0].u.type == __WASI_EVENTTYPE_CLOCK) {
        out[0] = (__wasi_event_t){
            .userdata = in[0].userdata,
            .type = in[0].u.type,
        };
#if CONFIG_HAS_CLOCK_NANOSLEEP
        clockid_t clock_id;
        if (wasi_clockid_to_clockid(in[0].u.u.clock.clock_id, &clock_id)) {
            os_timespec ts;
            convert_timestamp(in[0].u.u.clock.timeout, &ts);
            int ret = clock_nanosleep(
                clock_id,
                (in[0].u.u.clock.flags & __WASI_SUBSCRIPTION_CLOCK_ABSTIME) != 0
                    ? TIMER_ABSTIME
                    : 0,
                &ts, NULL);
            if (ret != 0)
                out[0].error = convert_errno(ret);
        }
        else {
            out[0].error = __WASI_ENOTSUP;
        }
#else
        switch (in[0].u.u.clock.clock_id) {
            case __WASI_CLOCK_MONOTONIC:
                if ((in[0].u.u.clock.flags & __WASI_SUBSCRIPTION_CLOCK_ABSTIME)
                    != 0) {
                    // TODO(ed): Implement.
                    fputs("Unimplemented absolute sleep on monotonic clock\n",
                          stderr);
                    out[0].error = __WASI_ENOSYS;
                }
                else {
                    // Perform relative sleeps on the monotonic clock also using
                    // nanosleep(). This is incorrect, but good enough for now.
                    os_timespec ts;
                    convert_timestamp(in[0].u.u.clock.timeout, &ts);
                    os_nanosleep(&ts, NULL);
                }
                break;
            case __WASI_CLOCK_REALTIME:
                if ((in[0].u.u.clock.flags & __WASI_SUBSCRIPTION_CLOCK_ABSTIME)
                    != 0) {
                    // Sleeping to an absolute point in time can only be done
                    // by waiting on a condition variable.
                    struct mutex mutex;
                    struct cond cond;

                    if (!mutex_init(&mutex))
                        return -1;
                    if (!cond_init_realtime(&cond)) {
                        mutex_destroy(&mutex);
                        return -1;
                    }
                    mutex_lock(&mutex);
                    cond_timedwait(&cond, &mutex, in[0].u.u.clock.timeout,
                                   true);
                    mutex_unlock(&mutex);
                    mutex_destroy(&mutex);
                    cond_destroy(&cond);
                }
                else {
                    // Relative sleeps can be done using nanosleep().
                    os_timespec ts;
                    convert_timestamp(in[0].u.u.clock.timeout, &ts);
                    os_nanosleep(&ts, NULL);
                }
                break;
            default:
                out[0].error = __WASI_ENOTSUP;
                break;
        }
#endif
        *nevents = 1;
        if (out[0].error != 0)
            return convert_errno(out[0].error);
        return 0;
    }

    // Last option: call into poll(). This can only be done in case all
    // subscriptions consist of __WASI_EVENTTYPE_FD_READ and
    // __WASI_EVENTTYPE_FD_WRITE entries. There may be up to one
    // __WASI_EVENTTYPE_CLOCK entry to act as a timeout. These are also
    // the subscriptions generate by cloudlibc's poll() and select().
    struct fd_object **fos =
        wasm_runtime_malloc((uint32)(nsubscriptions * sizeof(*fos)));
    if (fos == NULL)
        return __WASI_ENOMEM;
    os_poll_file_handle *pfds =
        wasm_runtime_malloc((uint32)(nsubscriptions * sizeof(*pfds)));
    if (pfds == NULL) {
        wasm_runtime_free(fos);
        return __WASI_ENOMEM;
    }

    // Convert subscriptions to pollfd entries. Increase the reference
    // count on the file descriptors to ensure they remain valid across
    // the call to poll().
    struct fd_table *ft = curfds;
    rwlock_rdlock(&ft->lock);
    *nevents = 0;
    const __wasi_subscription_t *clock_subscription = NULL;
    for (size_t i = 0; i < nsubscriptions; ++i) {
        const __wasi_subscription_t *s = &in[i];
        switch (s->u.type) {
            case __WASI_EVENTTYPE_FD_READ:
            case __WASI_EVENTTYPE_FD_WRITE:
            {
                __wasi_errno_t error =
                    fd_object_get_locked(&fos[i], ft, s->u.u.fd_readwrite.fd,
                                         __WASI_RIGHT_POLL_FD_READWRITE, 0);
                if (error == 0) {

// Temporary workaround (see PR#4377)
#ifdef BH_PLATFORM_ZEPHYR
                    os_file_handle tfd = fos[i]->file_handle->fd;
#else
                    os_file_handle tfd = fos[i]->file_handle;
#endif
                    // Proper file descriptor on which we can poll().
                    pfds[i] = (os_poll_file_handle){
                        .fd = tfd,
                        .events = s->u.type == __WASI_EVENTTYPE_FD_READ
                                      ? POLLIN
                                      : POLLOUT,
                    };
                }
                else {
                    // Invalid file descriptor or rights missing.
                    fos[i] = NULL;
                    pfds[i] = (os_poll_file_handle){ .fd = -1 };
                    out[(*nevents)++] = (__wasi_event_t){
                        .userdata = s->userdata,
                        .error = error,
                        .type = s->u.type,
                    };
                }
                break;
            }
            case __WASI_EVENTTYPE_CLOCK:
                if (clock_subscription == NULL
                    && (s->u.u.clock.flags & __WASI_SUBSCRIPTION_CLOCK_ABSTIME)
                           == 0) {
                    // Relative timeout.
                    fos[i] = NULL;
                    pfds[i] = (os_poll_file_handle){ .fd = -1 };
                    clock_subscription = s;
                    break;
                }
            // Fallthrough.
            default:
                // Unsupported event.
                fos[i] = NULL;
                pfds[i] = (os_poll_file_handle){ .fd = -1 };
                out[(*nevents)++] = (__wasi_event_t){
                    .userdata = s->userdata,
                    .error = __WASI_ENOSYS,
                    .type = s->u.type,
                };
                break;
        }
    }
    rwlock_unlock(&ft->lock);

    // Use a zero-second timeout in case we've already generated events in
    // the loop above.
    int timeout;
    if (*nevents != 0) {
        timeout = 0;
    }
    else if (clock_subscription != NULL) {
        __wasi_timestamp_t ts = clock_subscription->u.u.clock.timeout / 1000000;
        timeout = ts > INT_MAX ? -1 : (int)ts;
    }
    else {
        timeout = -1;
    }

    int ret;
    int error = blocking_op_poll(exec_env, pfds, nsubscriptions, timeout, &ret);
    if (error != 0) {
        /* got an error */
    }
    else if (ret == 0 && *nevents == 0 && clock_subscription != NULL) {
        // No events triggered. Trigger the clock event.
        out[(*nevents)++] = (__wasi_event_t){
            .userdata = clock_subscription->userdata,
            .type = __WASI_EVENTTYPE_CLOCK,
        };
    }
    else {
        // Events got triggered. Don't trigger the clock event.
        for (size_t i = 0; i < nsubscriptions; ++i) {
            if (pfds[i].fd >= 0) {
                __wasi_filesize_t nbytes = 0;
                if (in[i].u.type == __WASI_EVENTTYPE_FD_READ) {
                    int l;
                    if (os_ioctl(fos[i]->file_handle, FIONREAD, &l) == 0)
                        nbytes = (__wasi_filesize_t)l;
                }
                if ((pfds[i].revents & POLLNVAL) != 0) {
                    // Bad file descriptor. This normally cannot occur, as
                    // referencing the file descriptor object will always ensure
                    // the descriptor is valid. Still, macOS may sometimes
                    // return this on FIFOs when reaching end-of-file.
                    out[(*nevents)++] = (__wasi_event_t){
                        .userdata = in[i].userdata,
#ifdef __APPLE__
                        .u.fd_readwrite.nbytes = nbytes,
                        .u.fd_readwrite.flags =
                            __WASI_EVENT_FD_READWRITE_HANGUP,
#else
                        .error = __WASI_EBADF,
#endif
                        .type = in[i].u.type,
                    };
                }
                else if ((pfds[i].revents & POLLERR) != 0) {
                    // File descriptor is in an error state.
                    out[(*nevents)++] = (__wasi_event_t){
                        .userdata = in[i].userdata,
                        .error = __WASI_EIO,
                        .type = in[i].u.type,
                    };
                }
                else if ((pfds[i].revents & POLLHUP) != 0) {
                    // End-of-file.
                    out[(*nevents)++] = (__wasi_event_t){
                        .userdata = in[i].userdata,
                        .type = in[i].u.type,
                        .u.fd_readwrite.nbytes = nbytes,
                        .u.fd_readwrite.flags =
                            __WASI_EVENT_FD_READWRITE_HANGUP,
                    };
                }
                else if ((pfds[i].revents & (POLLIN | POLLOUT)) != 0) {
                    // Read or write possible.
                    out[(*nevents)++] = (__wasi_event_t){
                        .userdata = in[i].userdata,
                        .type = in[i].u.type,
                        .u.fd_readwrite.nbytes = nbytes,
                    };
                }
            }
        }
    }

    for (size_t i = 0; i < nsubscriptions; ++i)
        if (fos[i] != NULL)
            fd_object_release(exec_env, fos[i]);
    wasm_runtime_free(fos);
    wasm_runtime_free(pfds);
    return error;
#endif
}

__wasi_errno_t
wasmtime_ssp_random_get(void *buf, size_t nbyte)
{
    return random_buf(buf, nbyte);
}

__wasi_errno_t
wasi_ssp_sock_accept(wasm_exec_env_t exec_env, struct fd_table *curfds,
                     __wasi_fd_t fd, __wasi_fdflags_t flags,
                     __wasi_fd_t *fd_new)
{
    __wasi_filetype_t wasi_type;
    __wasi_rights_t max_base, max_inheriting;
    struct fd_object *fo;
    bh_socket_t new_sock = os_get_invalid_handle();
    int ret;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_SOCK_ACCEPT, 0);
    if (error != __WASI_ESUCCESS) {
        goto fail;
    }

    ret = blocking_op_socket_accept(exec_env, fo->file_handle, &new_sock, NULL,
                                    NULL);
    fd_object_release(exec_env, fo);
    if (BHT_OK != ret) {
        error = convert_errno(errno);
        goto fail;
    }

    error = fd_determine_type_rights(new_sock, &wasi_type, &max_base,
                                     &max_inheriting);
    if (error != __WASI_ESUCCESS) {
        goto fail;
    }

    error = fd_table_insert_fd(exec_env, curfds, new_sock, wasi_type, max_base,
                               max_inheriting, fd_new);
    if (error != __WASI_ESUCCESS) {
        /* released in fd_table_insert_fd() */
        new_sock = os_get_invalid_handle();
        goto fail;
    }

    return __WASI_ESUCCESS;

fail:
    if (os_is_handle_valid(&new_sock)) {
        os_socket_close(new_sock);
    }
    return error;
}

__wasi_errno_t
wasi_ssp_sock_addr_local(wasm_exec_env_t exec_env, struct fd_table *curfds,
                         __wasi_fd_t fd, __wasi_addr_t *addr)
{
    struct fd_object *fo;
    bh_sockaddr_t bh_addr;
    int ret;

    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_SOCK_ADDR_LOCAL, 0);
    if (error != __WASI_ESUCCESS)
        return error;

    ret = os_socket_addr_local(fo->file_handle, &bh_addr);
    fd_object_release(exec_env, fo);
    if (ret != BHT_OK) {
        return convert_errno(errno);
    }

    bh_sockaddr_to_wasi_addr(&bh_addr, addr);

    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasi_ssp_sock_addr_remote(wasm_exec_env_t exec_env, struct fd_table *curfds,
                          __wasi_fd_t fd, __wasi_addr_t *addr)
{
    struct fd_object *fo;
    bh_sockaddr_t bh_addr;
    int ret;

    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_SOCK_ADDR_LOCAL, 0);
    if (error != __WASI_ESUCCESS)
        return error;

    ret = os_socket_addr_remote(fo->file_handle, &bh_addr);
    fd_object_release(exec_env, fo);
    if (ret != BHT_OK) {
        return convert_errno(errno);
    }

    bh_sockaddr_to_wasi_addr(&bh_addr, addr);

    return __WASI_ESUCCESS;
}

static bool
wasi_addr_to_string(const __wasi_addr_t *addr, char *buf, size_t buflen)
{
    if (addr->kind == IPv4) {
        const char *format = "%u.%u.%u.%u";

        bh_assert(buflen >= 16);

        snprintf(buf, buflen, format, addr->addr.ip4.addr.n0,
                 addr->addr.ip4.addr.n1, addr->addr.ip4.addr.n2,
                 addr->addr.ip4.addr.n3);

        return true;
    }
    else if (addr->kind == IPv6) {
        const char *format = "%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x";
        __wasi_addr_ip6_t ipv6 = addr->addr.ip6.addr;

        bh_assert(buflen >= 40);

        snprintf(buf, buflen, format, ipv6.n0, ipv6.n1, ipv6.n2, ipv6.n3,
                 ipv6.h0, ipv6.h1, ipv6.h2, ipv6.h3);

        return true;
    }
    return false;
}

__wasi_errno_t
wasi_ssp_sock_bind(wasm_exec_env_t exec_env, struct fd_table *curfds,
                   struct addr_pool *addr_pool, __wasi_fd_t fd,
                   __wasi_addr_t *addr)
{
    char buf[48] = { 0 };
    struct fd_object *fo;
    __wasi_errno_t error;
    int port = addr->kind == IPv4 ? addr->addr.ip4.port : addr->addr.ip6.port;
    int ret;

    if (!wasi_addr_to_string(addr, buf, sizeof(buf))) {
        return __WASI_EPROTONOSUPPORT;
    }

    if (!addr_pool_search(addr_pool, buf)) {
        return __WASI_EACCES;
    }

    error = fd_object_get(curfds, &fo, fd, __WASI_RIGHT_SOCK_BIND, 0);
    if (error != __WASI_ESUCCESS)
        return error;

    ret = os_socket_bind(fo->file_handle, buf, &port);
    fd_object_release(exec_env, fo);
    if (BHT_OK != ret) {
        return convert_errno(errno);
    }

    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasi_ssp_sock_addr_resolve(wasm_exec_env_t exec_env, struct fd_table *curfds,
                           char **ns_lookup_list, const char *host,
                           const char *service, __wasi_addr_info_hints_t *hints,
                           __wasi_addr_info_t *addr_info,
                           __wasi_size_t addr_info_size,
                           __wasi_size_t *max_info_size)
{
    bh_addr_info_t *wamr_addr_info =
        wasm_runtime_malloc(addr_info_size * sizeof(bh_addr_info_t));
    uint8_t hints_is_ipv4 = hints->family == INET4;
    uint8_t hints_is_tcp = hints->type == SOCKET_STREAM;
    size_t _max_info_size;
    size_t actual_info_size;

    if (!wamr_addr_info) {
        return __WASI_ENOMEM;
    }

    if (!ns_lookup_list_search(ns_lookup_list, host)) {
        wasm_runtime_free(wamr_addr_info);
        return __WASI_EACCES;
    }

    int ret = blocking_op_socket_addr_resolve(
        exec_env, host, service,
        hints->hints_enabled && hints->type != SOCKET_ANY ? &hints_is_tcp
                                                          : NULL,
        hints->hints_enabled && hints->family != INET_UNSPEC ? &hints_is_ipv4
                                                             : NULL,
        wamr_addr_info, addr_info_size, &_max_info_size);

    if (ret != BHT_OK) {
        wasm_runtime_free(wamr_addr_info);
        return convert_errno(errno);
    }

    *max_info_size = _max_info_size;
    actual_info_size =
        addr_info_size < *max_info_size ? addr_info_size : *max_info_size;

    for (size_t i = 0; i < actual_info_size; i++) {
        addr_info[i].type =
            wamr_addr_info[i].is_tcp ? SOCKET_STREAM : SOCKET_DGRAM;
        bh_sockaddr_to_wasi_addr(&wamr_addr_info[i].sockaddr,
                                 &addr_info[i].addr);
    }

    wasm_runtime_free(wamr_addr_info);
    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasi_ssp_sock_connect(wasm_exec_env_t exec_env, struct fd_table *curfds,
                      struct addr_pool *addr_pool, __wasi_fd_t fd,
                      __wasi_addr_t *addr)
{
    char buf[48] = { 0 };
    struct fd_object *fo;
    __wasi_errno_t error;
    int ret;

    if (!wasi_addr_to_string(addr, buf, sizeof(buf))) {
        return __WASI_EPROTONOSUPPORT;
    }

    if (!addr_pool_search(addr_pool, buf)) {
        return __WASI_EACCES;
    }

    error = fd_object_get(curfds, &fo, fd, __WASI_RIGHT_SOCK_BIND, 0);
    if (error != __WASI_ESUCCESS) {
        return error;
    }

    /* Consume __wasi_addr_t */
    ret = blocking_op_socket_connect(exec_env, fo->file_handle, buf,
                                     addr->kind == IPv4 ? addr->addr.ip4.port
                                                        : addr->addr.ip6.port);
    fd_object_release(exec_env, fo);
    if (BHT_OK != ret) {
        return convert_errno(errno);
    }

    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasi_ssp_sock_get_recv_buf_size(wasm_exec_env_t exec_env,
                                struct fd_table *curfds, __wasi_fd_t fd,
                                __wasi_size_t *size)
{
    struct fd_object *fo;
    __wasi_errno_t error = fd_object_get(curfds, &fo, fd, 0, 0);
    if (error != __WASI_ESUCCESS)
        return error;

    size_t bufsize = 0;
    int ret = os_socket_get_recv_buf_size(fo->file_handle, &bufsize);
    fd_object_release(exec_env, fo);
    if (BHT_OK != ret) {
        return convert_errno(errno);
    }

    *size = (__wasi_size_t)bufsize;

    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasi_ssp_sock_get_reuse_addr(wasm_exec_env_t exec_env, struct fd_table *curfds,
                             __wasi_fd_t fd, uint8_t *reuse)
{
    struct fd_object *fo;
    __wasi_errno_t error = fd_object_get(curfds, &fo, fd, 0, 0);
    if (error != __WASI_ESUCCESS)
        return error;

    bool enabled = false;

    int ret = os_socket_get_reuse_addr(fo->file_handle, &enabled);
    fd_object_release(exec_env, fo);
    if (BHT_OK != ret) {
        return convert_errno(errno);
    }

    *reuse = (uint8_t)enabled;

    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasi_ssp_sock_get_reuse_port(wasm_exec_env_t exec_env, struct fd_table *curfds,
                             __wasi_fd_t fd, uint8_t *reuse)
{
    struct fd_object *fo;
    __wasi_errno_t error = fd_object_get(curfds, &fo, fd, 0, 0);
    if (error != __WASI_ESUCCESS)
        return error;

    bool enabled = false;
    int ret = os_socket_get_reuse_port(fo->file_handle, &enabled);

    fd_object_release(exec_env, fo);
    if (BHT_OK != ret) {
        return convert_errno(errno);
    }

    *reuse = (uint8_t)enabled;

    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasi_ssp_sock_get_send_buf_size(wasm_exec_env_t exec_env,
                                struct fd_table *curfds, __wasi_fd_t fd,
                                __wasi_size_t *size)
{
    struct fd_object *fo;
    __wasi_errno_t error = fd_object_get(curfds, &fo, fd, 0, 0);
    if (error != __WASI_ESUCCESS)
        return error;

    size_t bufsize = 0;
    int ret = os_socket_get_send_buf_size(fo->file_handle, &bufsize);

    fd_object_release(exec_env, fo);
    if (BHT_OK != ret) {
        return convert_errno(errno);
    }

    *size = (__wasi_size_t)bufsize;

    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasi_ssp_sock_listen(wasm_exec_env_t exec_env, struct fd_table *curfds,
                     __wasi_fd_t fd, __wasi_size_t backlog)
{
    struct fd_object *fo;
    int ret;
    __wasi_errno_t error =
        fd_object_get(curfds, &fo, fd, __WASI_RIGHT_SOCK_LISTEN, 0);
    if (error != __WASI_ESUCCESS)
        return error;

    ret = os_socket_listen(fo->file_handle, backlog);
    fd_object_release(exec_env, fo);
    if (BHT_OK != ret) {
        return convert_errno(errno);
    }

    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasi_ssp_sock_open(wasm_exec_env_t exec_env, struct fd_table *curfds,
                   __wasi_fd_t poolfd, __wasi_address_family_t af,
                   __wasi_sock_type_t socktype, __wasi_fd_t *sockfd)
{
    bh_socket_t sock;
    bool is_tcp = SOCKET_DGRAM == socktype ? false : true;
    bool is_ipv4 = INET6 == af ? false : true;
    int ret;
    __wasi_filetype_t wasi_type = __WASI_FILETYPE_UNKNOWN;
    __wasi_rights_t max_base = 0, max_inheriting = 0;
    __wasi_errno_t error;

    (void)poolfd;

    ret = os_socket_create(&sock, is_ipv4, is_tcp);
    if (BHT_OK != ret) {
        return convert_errno(errno);
    }

    error =
        fd_determine_type_rights(sock, &wasi_type, &max_base, &max_inheriting);
    if (error != __WASI_ESUCCESS) {
        os_socket_close(sock);
        return error;
    }

    if (SOCKET_DGRAM == socktype) {
        bh_assert(wasi_type == __WASI_FILETYPE_SOCKET_DGRAM);
    }
    else {
        bh_assert(wasi_type == __WASI_FILETYPE_SOCKET_STREAM);
    }

    // TODO: base rights and inheriting rights ?
    error = fd_table_insert_fd(exec_env, curfds, sock, wasi_type, max_base,
                               max_inheriting, sockfd);
    if (error != __WASI_ESUCCESS) {
        return error;
    }

    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasi_ssp_sock_set_recv_buf_size(wasm_exec_env_t exec_env,
                                struct fd_table *curfds, __wasi_fd_t fd,
                                __wasi_size_t size)
{
    struct fd_object *fo;
    __wasi_errno_t error = fd_object_get(curfds, &fo, fd, 0, 0);
    if (error != __WASI_ESUCCESS)
        return error;

    int ret = os_socket_set_recv_buf_size(fo->file_handle, size);

    fd_object_release(exec_env, fo);
    if (BHT_OK != ret) {
        return convert_errno(errno);
    }

    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasi_ssp_sock_set_reuse_addr(wasm_exec_env_t exec_env, struct fd_table *curfds,
                             __wasi_fd_t fd, uint8_t reuse)
{
    struct fd_object *fo;
    __wasi_errno_t error = fd_object_get(curfds, &fo, fd, 0, 0);
    if (error != __WASI_ESUCCESS)
        return error;

    int ret = os_socket_set_reuse_addr(fo->file_handle, (bool)reuse);

    fd_object_release(exec_env, fo);
    if (BHT_OK != ret) {
        return convert_errno(errno);
    }

    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasi_ssp_sock_set_reuse_port(wasm_exec_env_t exec_env, struct fd_table *curfds,
                             __wasi_fd_t fd, uint8_t reuse)
{
    struct fd_object *fo;
    __wasi_errno_t error = fd_object_get(curfds, &fo, fd, 0, 0);
    if (error != __WASI_ESUCCESS)
        return error;

    int ret = os_socket_set_reuse_port(fo->file_handle, (bool)reuse);

    fd_object_release(exec_env, fo);
    if (BHT_OK != ret) {
        return convert_errno(errno);
    }

    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasi_ssp_sock_set_send_buf_size(wasm_exec_env_t exec_env,
                                struct fd_table *curfds, __wasi_fd_t fd,
                                __wasi_size_t size)
{
    struct fd_object *fo;
    __wasi_errno_t error = fd_object_get(curfds, &fo, fd, 0, 0);
    if (error != __WASI_ESUCCESS)
        return error;

    int ret = os_socket_set_send_buf_size(fo->file_handle, size);

    fd_object_release(exec_env, fo);
    if (BHT_OK != ret) {
        return convert_errno(errno);
    }

    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasmtime_ssp_sock_recv(wasm_exec_env_t exec_env, struct fd_table *curfds,
                       __wasi_fd_t sock, void *buf, size_t buf_len,
                       size_t *recv_len)
{
    __wasi_addr_t src_addr;

    return wasmtime_ssp_sock_recv_from(exec_env, curfds, sock, buf, buf_len, 0,
                                       &src_addr, recv_len);
}

__wasi_errno_t
wasmtime_ssp_sock_recv_from(wasm_exec_env_t exec_env, struct fd_table *curfds,
                            __wasi_fd_t sock, void *buf, size_t buf_len,
                            __wasi_riflags_t ri_flags, __wasi_addr_t *src_addr,
                            size_t *recv_len)
{
    struct fd_object *fo;
    __wasi_errno_t error;
    bh_sockaddr_t sockaddr, *sockaddr_ptr = NULL;
    int ret;

    error = fd_object_get(curfds, &fo, sock, __WASI_RIGHT_FD_READ, 0);
    if (error != 0) {
        return error;
    }

    // If the source address is not NULL, the caller is requesting the source
    // address to be returned if the protocol supports it.  If the value is
    // NULL, the POSIX standard states that the address is not returned.
    if (src_addr != NULL) {
        sockaddr_ptr = &sockaddr;
    }

    /* Consume bh_sockaddr_t instead of __wasi_addr_t */
    ret = blocking_op_socket_recv_from(exec_env, fo->file_handle, buf, buf_len,
                                       0, sockaddr_ptr);
    fd_object_release(exec_env, fo);
    if (-1 == ret) {
        return convert_errno(errno);
    }

    // If the source address is not NULL, we need to convert the sockaddr
    // back to __wasi_addr_t format.
    if (src_addr != NULL) {
        bh_sockaddr_to_wasi_addr(sockaddr_ptr, src_addr);
    }

    *recv_len = (size_t)ret;
    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasmtime_ssp_sock_send(wasm_exec_env_t exec_env, struct fd_table *curfds,
                       __wasi_fd_t sock, const void *buf, size_t buf_len,
                       size_t *sent_len)
{
    struct fd_object *fo;
    __wasi_errno_t error;
    int ret;

    error = fd_object_get(curfds, &fo, sock, __WASI_RIGHT_FD_WRITE, 0);
    if (error != 0) {
        return error;
    }

    ret = os_socket_send(fo->file_handle, buf, buf_len);
    fd_object_release(exec_env, fo);
    if (-1 == ret) {
        return convert_errno(errno);
    }

    *sent_len = (size_t)ret;
    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasmtime_ssp_sock_send_to(wasm_exec_env_t exec_env, struct fd_table *curfds,
                          struct addr_pool *addr_pool, __wasi_fd_t sock,
                          const void *buf, size_t buf_len,
                          __wasi_siflags_t si_flags,
                          const __wasi_addr_t *dest_addr, size_t *sent_len)
{
    char addr_buf[48] = { 0 };
    struct fd_object *fo;
    __wasi_errno_t error;
    int ret;
    bh_sockaddr_t sockaddr;

    if (!wasi_addr_to_string(dest_addr, addr_buf, sizeof(addr_buf))) {
        return __WASI_EPROTONOSUPPORT;
    }

    if (!addr_pool_search(addr_pool, addr_buf)) {
        return __WASI_EACCES;
    }

    error = fd_object_get(curfds, &fo, sock, __WASI_RIGHT_FD_WRITE, 0);
    if (error != 0) {
        return error;
    }

    wasi_addr_to_bh_sockaddr(dest_addr, &sockaddr);

    /* Consume bh_sockaddr instead of __wasi_addr_t */
    ret = blocking_op_socket_send_to(exec_env, fo->file_handle, buf, buf_len, 0,
                                     &sockaddr);
    fd_object_release(exec_env, fo);
    if (-1 == ret) {
        return convert_errno(errno);
    }

    *sent_len = (size_t)ret;
    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasmtime_ssp_sock_shutdown(wasm_exec_env_t exec_env, struct fd_table *curfds,
                           __wasi_fd_t sock)
{
    struct fd_object *fo;
    __wasi_errno_t error;

    error = fd_object_get(curfds, &fo, sock, 0, 0);
    if (error != 0)
        return error;

    error = os_socket_shutdown(fo->file_handle);
    fd_object_release(exec_env, fo);

    return error;
}

__wasi_errno_t
wasmtime_ssp_sched_yield(void)
{
#if defined(BH_PLATFORM_WINDOWS)
    SwitchToThread();
#elif defined(BH_PLATFORM_ZEPHYR)
    k_yield();
#else
    if (sched_yield() < 0)
        return convert_errno(errno);
#endif
    return 0;
}

__wasi_errno_t
wasmtime_ssp_args_get(struct argv_environ_values *argv_environ, char **argv,
                      char *argv_buf)
{
    for (size_t i = 0; i < argv_environ->argc; ++i) {
        argv[i] =
            argv_buf + (argv_environ->argv_list[i] - argv_environ->argv_buf);
    }
    argv[argv_environ->argc] = NULL;
    bh_memcpy_s(argv_buf, (uint32)argv_environ->argv_buf_size,
                argv_environ->argv_buf, (uint32)argv_environ->argv_buf_size);
    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasmtime_ssp_args_sizes_get(struct argv_environ_values *argv_environ,
                            size_t *argc, size_t *argv_buf_size)
{
    *argc = argv_environ->argc;
    *argv_buf_size = argv_environ->argv_buf_size;
    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasmtime_ssp_environ_get(struct argv_environ_values *argv_environ,
                         char **environs, char *environ_buf)
{
    for (size_t i = 0; i < argv_environ->environ_count; ++i) {
        environs[i] =
            environ_buf
            + (argv_environ->environ_list[i] - argv_environ->environ_buf);
    }
    environs[argv_environ->environ_count] = NULL;
    bh_memcpy_s(environ_buf, (uint32)argv_environ->environ_buf_size,
                argv_environ->environ_buf,
                (uint32)argv_environ->environ_buf_size);
    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasmtime_ssp_environ_sizes_get(struct argv_environ_values *argv_environ,
                               size_t *environ_count, size_t *environ_buf_size)
{
    *environ_count = argv_environ->environ_count;
    *environ_buf_size = argv_environ->environ_buf_size;
    return __WASI_ESUCCESS;
}

bool
argv_environ_init(struct argv_environ_values *argv_environ, char *argv_buf,
                  size_t argv_buf_size, char **argv_list, size_t argc,
                  char *environ_buf, size_t environ_buf_size,
                  char **environ_list, size_t environ_count)
{
    memset(argv_environ, 0, sizeof(struct argv_environ_values));

    argv_environ->argv_buf = argv_buf;
    argv_environ->argv_buf_size = argv_buf_size;
    argv_environ->argv_list = argv_list;
    argv_environ->argc = argc;
    argv_environ->environ_buf = environ_buf;
    argv_environ->environ_buf_size = environ_buf_size;
    argv_environ->environ_list = environ_list;
    argv_environ->environ_count = environ_count;
    return true;
}

void
argv_environ_destroy(struct argv_environ_values *argv_environ)
{
    (void)argv_environ;
}

void
fd_table_destroy(struct fd_table *ft)
{
    if (ft->entries) {
        for (uint32 i = 0; i < ft->size; i++) {
            if (ft->entries[i].object != NULL) {
                fd_object_release(NULL, ft->entries[i].object);
            }
        }
        wasm_runtime_free(ft->entries);
    }
    rwlock_destroy(&ft->lock);
}

void
fd_prestats_destroy(struct fd_prestats *pt)
{
    if (pt->prestats) {
        for (uint32 i = 0; i < pt->size; i++) {
            if (pt->prestats[i].dir != NULL) {
                wasm_runtime_free((void *)pt->prestats[i].dir);
            }
        }
        wasm_runtime_free(pt->prestats);
    }
    rwlock_destroy(&pt->lock);
}

bool
addr_pool_init(struct addr_pool *addr_pool)
{
    memset(addr_pool, 0, sizeof(*addr_pool));

    return true;
}

bool
addr_pool_insert(struct addr_pool *addr_pool, const char *addr, uint8 mask)
{
    struct addr_pool *cur = addr_pool;
    struct addr_pool *next;
    bh_ip_addr_buffer_t target;

    if (!addr_pool) {
        return false;
    }

    if (!(next = wasm_runtime_malloc(sizeof(struct addr_pool)))) {
        return false;
    }

    next->next = NULL;

    if (os_socket_inet_network(true, addr, &target) != BHT_OK) {
        // If parsing IPv4 fails, try IPv6
        if (os_socket_inet_network(false, addr, &target) != BHT_OK) {
            wasm_runtime_free(next);
            return false;
        }
        next->type = IPv6;
        bh_memcpy_s(next->addr.ip6, sizeof(next->addr.ip6), target.ipv6,
                    sizeof(target.ipv6));
        if (mask > 128) {
            wasm_runtime_free(next);
            return false;
        }
        next->mask = mask;
    }
    else {
        next->type = IPv4;
        next->addr.ip4 = target.ipv4;
        if (mask > 32) {
            wasm_runtime_free(next);
            return false;
        }
        next->mask = mask;
    }

    /* attach with */
    while (cur->next) {
        cur = cur->next;
    }
    cur->next = next;
    return true;
}

static void
init_address_mask(uint8_t *buf, size_t buflen, size_t mask)
{
    size_t element_size = sizeof(uint8_t) * 8;

    for (size_t i = 0; i < buflen; i++) {
        if (mask <= i * element_size) {
            buf[i] = 0;
        }
        else {
            size_t offset = min(mask - i * element_size, element_size);
            buf[i] = (~0u) << (element_size - offset);
        }
    }
}

/* target must be in network byte order */
static bool
compare_address(const struct addr_pool *addr_pool_entry,
                bh_ip_addr_buffer_t *target)
{
    uint8_t maskbuf[16] = { 0 };
    uint8_t basebuf[16] = { 0 };
    size_t addr_size;
    uint8_t max_addr_mask;

    if (addr_pool_entry->type == IPv4) {
        uint32_t addr_ip4 = htonl(addr_pool_entry->addr.ip4);
        bh_memcpy_s(basebuf, sizeof(addr_ip4), &addr_ip4, sizeof(addr_ip4));
        addr_size = 4;
    }
    else {
        uint16_t partial_addr_ip6;
        for (int i = 0; i < 8; i++) {
            partial_addr_ip6 = htons(addr_pool_entry->addr.ip6[i]);
            bh_memcpy_s(&basebuf[i * sizeof(partial_addr_ip6)],
                        sizeof(partial_addr_ip6), &partial_addr_ip6,
                        sizeof(partial_addr_ip6));
        }
        addr_size = 16;
    }
    max_addr_mask = (uint8)(addr_size * 8);

    /* IPv4 0.0.0.0 or IPv6 :: means any address */
    if (basebuf[0] == 0 && !memcmp(basebuf, basebuf + 1, addr_size - 1)) {
        return true;
    }

    /* No support for invalid mask value */
    if (addr_pool_entry->mask > max_addr_mask) {
        return false;
    }

    init_address_mask(maskbuf, addr_size, addr_pool_entry->mask);

    for (size_t i = 0; i < addr_size; i++) {
        uint8_t addr_mask = target->data[i] & maskbuf[i];
        uint8_t range_mask = basebuf[i] & maskbuf[i];
        if (addr_mask != range_mask) {
            return false;
        }
    }

    return true;
}

bool
addr_pool_search(struct addr_pool *addr_pool, const char *addr)
{
    struct addr_pool *cur = addr_pool->next;
    bh_ip_addr_buffer_t target;
    __wasi_addr_type_t addr_type;

    if (os_socket_inet_network(true, addr, &target) != BHT_OK) {
        size_t i;

        if (os_socket_inet_network(false, addr, &target) != BHT_OK) {
            return false;
        }
        addr_type = IPv6;
        for (i = 0; i < sizeof(target.ipv6) / sizeof(target.ipv6[0]); i++) {
            target.ipv6[i] = htons(target.ipv6[i]);
        }
    }
    else {
        addr_type = IPv4;
        target.ipv4 = htonl(target.ipv4);
    }

    while (cur) {
        if (cur->type == addr_type && compare_address(cur, &target)) {
            return true;
        }

        cur = cur->next;
    }

    return false;
}

void
addr_pool_destroy(struct addr_pool *addr_pool)
{
    struct addr_pool *cur = addr_pool->next;

    while (cur) {
        struct addr_pool *next = cur->next;
        wasm_runtime_free(cur);
        cur = next;
    }
}

#define WASMTIME_SSP_PASSTHROUGH_FD_TABLE struct fd_table *curfds,

// Defines a function that passes through the socket option to the OS
// implementation
#define WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(FUNC_NAME, OPTION_TYPE) \
    __wasi_errno_t wasmtime_ssp_sock_##FUNC_NAME(                      \
        wasm_exec_env_t exec_env,                                      \
        WASMTIME_SSP_PASSTHROUGH_FD_TABLE __wasi_fd_t sock,            \
        OPTION_TYPE option)                                            \
    {                                                                  \
        struct fd_object *fo;                                          \
        __wasi_errno_t error;                                          \
        int ret;                                                       \
        error = fd_object_get(curfds, &fo, sock, 0, 0);                \
        if (error != 0)                                                \
            return error;                                              \
        ret = os_socket_##FUNC_NAME(fo->file_handle, option);          \
        fd_object_release(exec_env, fo);                               \
        if (BHT_OK != ret)                                             \
            return convert_errno(errno);                               \
        return __WASI_ESUCCESS;                                        \
    }

WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_send_timeout, uint64)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_send_timeout, uint64 *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_recv_timeout, uint64)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_recv_timeout, uint64 *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_send_buf_size, size_t)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_send_buf_size, size_t *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_recv_buf_size, size_t)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_recv_buf_size, size_t *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_broadcast, bool)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_broadcast, bool *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_keep_alive, bool)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_keep_alive, bool *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_reuse_addr, bool)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_reuse_addr, bool *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_reuse_port, bool)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_reuse_port, bool *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_tcp_no_delay, bool)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_tcp_no_delay, bool *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_tcp_quick_ack, bool)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_tcp_quick_ack, bool *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_tcp_keep_idle, uint32)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_tcp_keep_idle, uint32 *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_tcp_keep_intvl, uint32)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_tcp_keep_intvl, uint32 *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_tcp_fastopen_connect, bool)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_tcp_fastopen_connect, bool *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_ip_ttl, uint8_t)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_ip_ttl, uint8_t *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_ip_multicast_ttl, uint8_t)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_ip_multicast_ttl, uint8_t *)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(set_ipv6_only, bool)
WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION(get_ipv6_only, bool *)

#undef WASMTIME_SSP_PASSTHROUGH_FD_TABLE
#undef WASMTIME_SSP_PASSTHROUGH_SOCKET_OPTION

__wasi_errno_t
wasmtime_ssp_sock_set_linger(wasm_exec_env_t exec_env, struct fd_table *curfds,
                             __wasi_fd_t sock, bool is_enabled, int linger_s)
{
    struct fd_object *fo;
    __wasi_errno_t error;
    int ret;
    error = fd_object_get(curfds, &fo, sock, 0, 0);
    if (error != 0)
        return error;

    ret = os_socket_set_linger(fo->file_handle, is_enabled, linger_s);
    fd_object_release(exec_env, fo);
    if (BHT_OK != ret)
        return convert_errno(errno);
    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasmtime_ssp_sock_get_linger(wasm_exec_env_t exec_env, struct fd_table *curfds,
                             __wasi_fd_t sock, bool *is_enabled, int *linger_s)
{
    struct fd_object *fo;
    __wasi_errno_t error;
    int ret;
    error = fd_object_get(curfds, &fo, sock, 0, 0);
    if (error != 0)
        return error;

    ret = os_socket_get_linger(fo->file_handle, is_enabled, linger_s);
    fd_object_release(exec_env, fo);
    if (BHT_OK != ret)
        return convert_errno(errno);

    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasmtime_ssp_sock_set_ip_add_membership(wasm_exec_env_t exec_env,
                                        struct fd_table *curfds,
                                        __wasi_fd_t sock,
                                        __wasi_addr_ip_t *imr_multiaddr,
                                        uint32_t imr_interface)
{
    struct fd_object *fo;
    __wasi_errno_t error;
    int ret;
    bh_ip_addr_buffer_t addr_info;
    bool is_ipv6;
    error = fd_object_get(curfds, &fo, sock, 0, 0);
    if (error != 0)
        return error;

    wasi_addr_ip_to_bh_ip_addr_buffer(imr_multiaddr, &addr_info);
    is_ipv6 = imr_multiaddr->kind == IPv6;
    ret = os_socket_set_ip_add_membership(fo->file_handle, &addr_info,
                                          imr_interface, is_ipv6);
    fd_object_release(exec_env, fo);
    if (BHT_OK != ret)
        return convert_errno(errno);
    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasmtime_ssp_sock_set_ip_drop_membership(wasm_exec_env_t exec_env,
                                         struct fd_table *curfds,
                                         __wasi_fd_t sock,
                                         __wasi_addr_ip_t *imr_multiaddr,
                                         uint32_t imr_interface)
{
    struct fd_object *fo;
    __wasi_errno_t error;
    int ret;
    bh_ip_addr_buffer_t addr_info;
    bool is_ipv6;
    error = fd_object_get(curfds, &fo, sock, 0, 0);
    if (error != 0)
        return error;

    wasi_addr_ip_to_bh_ip_addr_buffer(imr_multiaddr, &addr_info);
    is_ipv6 = imr_multiaddr->kind == IPv6;
    ret = os_socket_set_ip_drop_membership(fo->file_handle, &addr_info,
                                           imr_interface, is_ipv6);
    fd_object_release(exec_env, fo);
    if (BHT_OK != ret)
        return convert_errno(errno);
    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasmtime_ssp_sock_set_ip_multicast_loop(wasm_exec_env_t exec_env,
                                        struct fd_table *curfds,
                                        __wasi_fd_t sock, bool ipv6,
                                        bool is_enabled)
{
    struct fd_object *fo;
    __wasi_errno_t error;
    int ret;
    error = fd_object_get(curfds, &fo, sock, 0, 0);
    if (error != 0)
        return error;

    ret = os_socket_set_ip_multicast_loop(fo->file_handle, ipv6, is_enabled);
    fd_object_release(exec_env, fo);
    if (BHT_OK != ret)
        return convert_errno(errno);
    return __WASI_ESUCCESS;
}

__wasi_errno_t
wasmtime_ssp_sock_get_ip_multicast_loop(wasm_exec_env_t exec_env,
                                        struct fd_table *curfds,
                                        __wasi_fd_t sock, bool ipv6,
                                        bool *is_enabled)
{
    struct fd_object *fo;
    __wasi_errno_t error;
    int ret;
    error = fd_object_get(curfds, &fo, sock, 0, 0);
    if (error != 0)
        return error;

    ret = os_socket_get_ip_multicast_loop(fo->file_handle, ipv6, is_enabled);
    fd_object_release(exec_env, fo);
    if (BHT_OK != ret)
        return convert_errno(errno);

    return __WASI_ESUCCESS;
}