graphene/LibOS/shim/src/sys/shim_open.c

/* -*- mode:c; c-file-style:"k&r"; c-basic-offset: 4; tab-width:4; indent-tabs-mode:nil; mode:auto-fill; fill-column:78; -*- */
/* vim: set ts=4 sw=4 et tw=78 fo=cqt wm=0: */

/* Copyright (C) 2014 Stony Brook University
   This file is part of Graphene Library OS.

   Graphene Library OS is free software: you can redistribute it and/or
   modify it under the terms of the GNU Lesser General Public License
   as published by the Free Software Foundation, either version 3 of the
   License, or (at your option) any later version.

   Graphene Library OS is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   GNU Lesser General Public License for more details.

   You should have received a copy of the GNU Lesser General Public License
   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */

/*
 * shim_open.c
 *
 * Implementation of system call "read", "write", "open", "creat", "openat",
 * "close", "lseek", "pread64", "pwrite64", "getdents", "getdents64",
 * "fsync", "truncate" and "ftruncate".
 */

#include <shim_internal.h>
#include <shim_utils.h>
#include <shim_table.h>
#include <shim_thread.h>
#include <shim_handle.h>
#include <shim_fs.h>
#include <shim_profile.h>

#include <pal.h>
#include <pal_error.h>

#include <errno.h>
#include <dirent.h>

#include <linux/stat.h>
#include <linux/fcntl.h>

int do_handle_read (struct shim_handle * hdl, void * buf, int count)
{
    if (!(hdl->acc_mode & MAY_READ))
        return -EACCES;

    struct shim_mount * fs = hdl->fs;
    assert (fs && fs->fs_ops);

    if (!fs->fs_ops->read)
        return -EBADF;

    if (hdl->type == TYPE_DIR)
        return -EISDIR;

    return fs->fs_ops->read(hdl, buf, count);
}

size_t shim_do_read (int fd, void * buf, size_t count)
{
    struct shim_handle * hdl = get_fd_handle(fd, NULL, NULL);
    if (!hdl)
        return -EBADF;
    int ret = do_handle_read(hdl, buf, count);
    put_handle(hdl);
    return ret;
}

int do_handle_write (struct shim_handle * hdl, const void * buf, int count)
{
    if (!(hdl->acc_mode & MAY_WRITE))
        return -EACCES;

    struct shim_mount * fs = hdl->fs;
    assert (fs && fs->fs_ops);

    if (!fs->fs_ops->write)
        return -EBADF;

    if (hdl->type == TYPE_DIR)
        return -EISDIR;

    return fs->fs_ops->write(hdl, buf, count);
}

size_t shim_do_write (int fd, const void * buf, size_t count)
{
    struct shim_handle * hdl = get_fd_handle(fd, NULL, NULL);
    if (!hdl)
        return -EBADF;

    int ret = do_handle_write(hdl, buf, count);
    put_handle(hdl);
    return ret;
}

int shim_do_open (const char * file, int flags, mode_t mode)
{
    if (!file || !(*file))
        return -EINVAL;

    struct shim_handle * hdl = get_new_handle();
    if (!hdl)
        return -ENOMEM;

    int ret = 0;
    ret = open_namei(hdl, NULL, file, flags, mode, NULL);
    if (ret < 0)
        goto out;

    ret = set_new_fd_handle(hdl, flags & O_CLOEXEC ? FD_CLOEXEC : 0, NULL);
out:
    put_handle(hdl);
    return ret;
}

int shim_do_creat (const char * path, mode_t mode)
{
    return shim_do_open(path, O_CREAT|O_TRUNC|O_WRONLY, mode);
}

int shim_do_openat (int dfd, const char * filename, int flags, int mode)
{
    if (!filename)
        return -EINVAL;

    if (*filename == '/')
        return shim_do_open(filename, flags, mode);

    struct shim_dentry * dir = NULL;
    int ret = 0;

    if ((ret = path_startat(dfd, &dir)) < 0)
        return ret;

    struct shim_handle * hdl = get_new_handle();
    if (!hdl) {
        ret = -ENOMEM;
        goto out;
    }

    ret = open_namei(hdl, dir, filename, flags, mode, NULL);
    if (ret < 0)
        goto out_hdl;

    ret = set_new_fd_handle(hdl, flags & O_CLOEXEC ? FD_CLOEXEC : 0, NULL);

out_hdl:
    put_handle(hdl);
out:
    put_dentry(dir);
    return ret;
}

int shim_do_close (int fd)
{
    struct shim_handle * handle = detach_fd_handle(fd, NULL, NULL);
    if (!handle)
        return -EBADF;

    close_handle(handle);
    return 0;
}

/* lseek is simply doing arithmetic on the offset, no PAL call here */
off_t shim_do_lseek (int fd, off_t offset, int origin)
{
    if (origin != SEEK_SET && origin != SEEK_CUR && origin != SEEK_END)
        return -EINVAL;

    struct shim_handle * hdl = get_fd_handle(fd, NULL, NULL);
    if (!hdl)
        return -EBADF;

    int ret = 0;
    struct shim_mount * fs = hdl->fs;
    assert(fs && fs->fs_ops);

    if (!fs->fs_ops->seek) {
        ret = -ESPIPE;
        goto out;
    }

    if (hdl->type == TYPE_DIR) {
        /* TODO: handle lseek'ing of directories */
        ret = -ENOSYS;
        goto out;
    }

    ret = fs->fs_ops->seek(hdl, offset, origin);
out:
    put_handle(hdl);
    return ret;
}

size_t shim_do_pread64 (int fd, char * buf, size_t count, loff_t pos)
{
    struct shim_handle * hdl = get_fd_handle(fd, NULL, NULL);
    if (!hdl)
        return -EBADF;

    struct shim_mount * fs = hdl->fs;
    size_t ret = -EACCES;

    if (!fs || !fs->fs_ops)
        goto out;

    if (!fs->fs_ops->seek) {
        ret = -ESPIPE;
        goto out;
    }

    if (!fs->fs_ops->read)
        goto out;

    if (hdl->type == TYPE_DIR)
        goto out;

    int offset = fs->fs_ops->seek(hdl, 0, SEEK_CUR);
    if (offset < 0) {
        ret = offset;
        goto out;
    }

    ret = fs->fs_ops->seek(hdl, pos, SEEK_SET);
    if (ret < 0)
        goto out;

    int bytes = fs->fs_ops->read(hdl, buf, count);

    ret = fs->fs_ops->seek(hdl, offset, SEEK_SET);
    if (ret < 0)
        goto out;

    ret = bytes;
out:
    put_handle(hdl);
    return ret;
}

size_t shim_do_pwrite64 (int fd, char * buf, size_t count, loff_t pos)
{
    struct shim_handle * hdl = get_fd_handle(fd, NULL, NULL);
    if (!hdl)
        return -EBADF;

    struct shim_mount * fs = hdl->fs;
    size_t ret = -EACCES;

    if (!fs || !fs->fs_ops)
        goto out;

    if (!fs->fs_ops->seek) {
        ret = -ESPIPE;
        goto out;
    }

    if (!fs->fs_ops->write)
        goto out;

    if (hdl->type == TYPE_DIR)
        goto out;

    int offset = fs->fs_ops->seek(hdl, 0, SEEK_CUR);
    if (offset < 0) {
        ret = offset;
        goto out;
    }

    ret = fs->fs_ops->seek(hdl, pos, SEEK_SET);
    if (ret < 0)
        goto out;

    int bytes = fs->fs_ops->write(hdl, buf, count);

    ret = fs->fs_ops->seek(hdl, offset, SEEK_SET);
    if (ret < 0)
        goto out;

    ret = bytes;
out:
    put_handle(hdl);
    return ret;
}

static inline int get_dirent_type (mode_t type)
{
    switch (type) {
        case S_IFLNK:
            return LINUX_DT_LNK;
        case S_IFREG:
            return LINUX_DT_REG;
        case S_IFDIR:
            return LINUX_DT_DIR;
        case S_IFCHR:
            return LINUX_DT_CHR;
        case S_IFBLK:
            return LINUX_DT_BLK;
        case S_IFIFO:
            return LINUX_DT_FIFO;
        case S_IFSOCK:
            return LINUX_DT_SOCK;
        default:
            return LINUX_DT_UNKNOWN;
    }
}

size_t shim_do_getdents (int fd, struct linux_dirent * buf, size_t count)
{
    struct shim_handle * hdl = get_fd_handle(fd, NULL, NULL);
    if (!hdl)
        return -EBADF;

    int ret = -EACCES;

    if (hdl->type != TYPE_DIR) {
        ret = -ENOTDIR;
        goto out_no_unlock;
    }

    /* DEP 3/3/17: Properly handle an unlinked directory */
    if (hdl->dentry->state & DENTRY_NEGATIVE) {
        ret = -ENOENT;
        goto out_no_unlock;
    }
    
    /* we are grabbing the lock because the handle content is actually
       updated */
    lock(hdl->lock);

    struct shim_dir_handle * dirhdl = &hdl->info.dir;
    struct shim_dentry * dent = hdl->dentry;
    struct linux_dirent * b = buf;
    int bytes = 0;

    /* If we haven't listed the directory, do this first */
    if (!(dent->state & DENTRY_LISTED)) {
        ret = list_directory_dentry(dent);
        if (ret < 0)
            goto out;
    }

#define DIRENT_SIZE(len)  (sizeof(struct linux_dirent) +                \
                           sizeof(struct linux_dirent_tail) + (len) + 1)

#define ASSIGN_DIRENT(dent, name, type)                                 \
        do {                                                            \
            int len = strlen(name);                                     \
            if (bytes + DIRENT_SIZE(len) > count)                       \
                goto done;                                              \
                                                                        \
            struct linux_dirent_tail * bt                               \
                    = (void *) b + sizeof(struct linux_dirent) + len + 1; \
                                                                        \
            b->d_ino = dent->ino;                                       \
            b->d_off = ++dirhdl->offset;                                \
            b->d_reclen = DIRENT_SIZE(len);                             \
                                                                        \
            memcpy(b->d_name, name, len + 1);                           \
                                                                        \
            bt->pad = 0;                                                \
            bt->d_type = type ? : get_dirent_type(dent->mode);          \
                                                                        \
            b = (void *) bt + sizeof(struct linux_dirent_tail);         \
            bytes += DIRENT_SIZE(len);                                  \
        } while(0)


    if (dirhdl->dot) {
        ASSIGN_DIRENT(dirhdl->dot, ".", LINUX_DT_DIR);
        put_dentry(dirhdl->dot);
        dirhdl->dot = NULL;
    }

    if (dirhdl->dotdot) {
        ASSIGN_DIRENT(dirhdl->dotdot, "..", LINUX_DT_DIR);
        put_dentry(dirhdl->dotdot);
        dirhdl->dotdot = NULL;
    }

    if (dirhdl->ptr == (void *) -1) {
        ret = list_directory_handle(dent, hdl);
        if (ret < 0)
            goto out;
    }
    
    while (dirhdl->ptr && *dirhdl->ptr) {
        dent = *dirhdl->ptr;
        /* DEP 3/3/17: We need to filter negative dentries */
        if (!(dent->state & DENTRY_NEGATIVE))
            ASSIGN_DIRENT(dent, dentry_get_name(dent), 0);
        put_dentry(dent);
        *(dirhdl->ptr++) = NULL;
    }

#undef DIRENT_SIZE
#undef ASSIGN_DIRENT

done:
    ret = bytes;
    /* DEP 3/3/17: Properly detect EINVAL case, where buffer is too small to
     * hold anything */
    if (bytes == 0 && (dirhdl->dot || dirhdl->dotdot || 
                       (dirhdl->ptr && *dirhdl->ptr)))
        ret = -EINVAL;
out:
    unlock(hdl->lock);
out_no_unlock:
    put_handle(hdl);
    return ret;
}

size_t shim_do_getdents64 (int fd, struct linux_dirent64 * buf, size_t count)
{
    struct shim_handle * hdl = get_fd_handle(fd, NULL, NULL);
    if (!hdl)
        return -EBADF;

    int ret = -EACCES;

    if (hdl->type != TYPE_DIR) {
        ret = -ENOTDIR;
        goto out;
    }

    /* DEP 3/3/17: Properly handle an unlinked directory */
    if (hdl->dentry->state & DENTRY_NEGATIVE) {
        ret = -ENOENT;
        goto out;
    }
    
    lock(hdl->lock);

    struct shim_dir_handle * dirhdl = &hdl->info.dir;
    struct shim_dentry * dent = hdl->dentry;
    struct linux_dirent64 * b = buf;
    int bytes = 0;

    /* If we haven't listed the directory, do this first */
    if (!(dent->state & DENTRY_LISTED)) {
        ret = list_directory_dentry(dent);
        if (ret) goto out;
    }
    
#define DIRENT_SIZE(len)  (sizeof(struct linux_dirent64) + (len) + 1)

#define ASSIGN_DIRENT(dent, name, type)                                 \
        do {                                                            \
            int len = strlen(name);                                     \
            if (bytes + DIRENT_SIZE(len) > count)                       \
                goto done;                                              \
                                                                        \
            b->d_ino = dent->ino;                                       \
            b->d_off = ++dirhdl->offset;                                \
            b->d_reclen = DIRENT_SIZE(len);                             \
            b->d_type = type ? : get_dirent_type(dent->mode);           \
                                                                        \
            memcpy(b->d_name, name, len + 1);                           \
                                                                        \
            b = (void *) b + DIRENT_SIZE(len);                          \
            bytes += DIRENT_SIZE(len);                                  \
        } while(0)

    if (dirhdl->dot) {
        ASSIGN_DIRENT(dirhdl->dot, ".", LINUX_DT_DIR);
        put_dentry(dirhdl->dot);
        dirhdl->dot = NULL;
    }

    if (dirhdl->dotdot) {
        ASSIGN_DIRENT(dirhdl->dotdot, "..", LINUX_DT_DIR);
        put_dentry(dirhdl->dotdot);
        dirhdl->dotdot = NULL;
    }

    if (dirhdl->ptr == (void *) -1) {
        ret = list_directory_handle(dent, hdl);
        if (ret) goto out;
    }
    
    while (dirhdl->ptr && *dirhdl->ptr) {
        dent = *dirhdl->ptr;
        /* DEP 3/3/17: We need to filter negative dentries */
        if (!(dent->state & DENTRY_NEGATIVE))
            ASSIGN_DIRENT(dent, dentry_get_name(dent), 0);
        put_dentry(dent);
        *(dirhdl->ptr++) = NULL;
    }

#undef DIRENT_SIZE
#undef ASSIGN_DIRENT

done:
    ret = bytes;
    /* DEP 3/3/17: Properly detect EINVAL case, where buffer is too small to
     * hold anything */
    if (bytes == 0 && (dirhdl->dot || dirhdl->dotdot || 
                       (dirhdl->ptr && *dirhdl->ptr)))
        ret = -EINVAL;
    unlock(hdl->lock);
out:
    put_handle(hdl);
    return ret;
}

int shim_do_fsync (int fd)
{
    struct shim_handle * hdl = get_fd_handle(fd, NULL, NULL);
    if (!hdl)
        return -EBADF;

    int ret = -EACCES;
    struct shim_mount * fs = hdl->fs;

    if (!fs || !fs->fs_ops)
        goto out;

    if (hdl->type == TYPE_DIR)
        goto out;

    if (!fs->fs_ops->flush) {
        ret = -EROFS;
        goto out;
    }

    ret = fs->fs_ops->flush(hdl);
out:
    put_handle(hdl);
    return ret;
}


// DEP 10/20/16: Assuming fsync >> fdatasync for now
//  and no app depends on only syncing data for correctness.
int shim_do_fdatasync (int fd)
{
    return shim_do_fsync(fd);
}


int shim_do_truncate (const char * path, loff_t length)
{
    struct shim_dentry * dent = NULL;
    int ret = 0;

    if ((ret = path_lookupat(NULL, path, 0, &dent, NULL)) < 0)
        return ret;

    struct shim_mount * fs = dent->fs;

    if (!fs || !fs->d_ops || !fs->d_ops->open) {
        ret = -EBADF;
        goto out;
    }

    if (!fs->fs_ops->truncate) {
        ret = -EROFS;
        goto out;
    }

    struct shim_handle * hdl = get_new_handle();

    if (!hdl) {
        ret = -ENOMEM;
        goto out;
    }

    hdl->fs = fs;

    if ((ret = fs->d_ops->open(hdl, dent, O_WRONLY)) < 0)
        goto out_handle;

    ret = fs->fs_ops->truncate(hdl, length);
    flush_handle(hdl);
out_handle:
    put_handle(hdl);
out:
    return ret;
}

int shim_do_ftruncate (int fd, loff_t length)
{
    struct shim_handle * hdl = get_fd_handle(fd, NULL, NULL);
    if (!hdl)
        return -EBADF;

    struct shim_mount * fs = hdl->fs;
    int ret = -EINVAL;

    if (!fs || !fs->fs_ops)
        goto out;

    if (hdl->type == TYPE_DIR ||
        !fs->fs_ops->truncate)
        goto out;

    ret = fs->fs_ops->truncate(hdl, length);
out:
    put_handle(hdl);
    return ret;
}