| 1 | /* $NetBSD: linux_file64.c,v 1.55 2013/12/27 14:17:11 njoly Exp $ */ |
| 2 | |
| 3 | /*- |
| 4 | * Copyright (c) 1995, 1998, 2000, 2008 The NetBSD Foundation, Inc. |
| 5 | * All rights reserved. |
| 6 | * |
| 7 | * This code is derived from software contributed to The NetBSD Foundation |
| 8 | * by Frank van der Linden and Eric Haszlakiewicz. |
| 9 | * |
| 10 | * Redistribution and use in source and binary forms, with or without |
| 11 | * modification, are permitted provided that the following conditions |
| 12 | * are met: |
| 13 | * 1. Redistributions of source code must retain the above copyright |
| 14 | * notice, this list of conditions and the following disclaimer. |
| 15 | * 2. Redistributions in binary form must reproduce the above copyright |
| 16 | * notice, this list of conditions and the following disclaimer in the |
| 17 | * documentation and/or other materials provided with the distribution. |
| 18 | * |
| 19 | * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS |
| 20 | * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED |
| 21 | * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR |
| 22 | * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS |
| 23 | * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR |
| 24 | * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF |
| 25 | * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS |
| 26 | * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN |
| 27 | * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) |
| 28 | * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE |
| 29 | * POSSIBILITY OF SUCH DAMAGE. |
| 30 | */ |
| 31 | |
| 32 | /* |
| 33 | * Linux 64bit filesystem calls. Used on 32bit archs, not used on 64bit ones. |
| 34 | */ |
| 35 | |
| 36 | #include <sys/cdefs.h> |
| 37 | __KERNEL_RCSID(0, "$NetBSD: linux_file64.c,v 1.55 2013/12/27 14:17:11 njoly Exp $" ); |
| 38 | |
| 39 | #include <sys/param.h> |
| 40 | #include <sys/systm.h> |
| 41 | #include <sys/namei.h> |
| 42 | #include <sys/proc.h> |
| 43 | #include <sys/dirent.h> |
| 44 | #include <sys/file.h> |
| 45 | #include <sys/stat.h> |
| 46 | #include <sys/filedesc.h> |
| 47 | #include <sys/ioctl.h> |
| 48 | #include <sys/kernel.h> |
| 49 | #include <sys/mount.h> |
| 50 | #include <sys/malloc.h> |
| 51 | #include <sys/namei.h> |
| 52 | #include <sys/vfs_syscalls.h> |
| 53 | #include <sys/vnode.h> |
| 54 | #include <sys/tty.h> |
| 55 | #include <sys/conf.h> |
| 56 | |
| 57 | #include <sys/syscallargs.h> |
| 58 | |
| 59 | #include <compat/linux/common/linux_types.h> |
| 60 | #include <compat/linux/common/linux_signal.h> |
| 61 | #include <compat/linux/common/linux_fcntl.h> |
| 62 | #include <compat/linux/common/linux_util.h> |
| 63 | #include <compat/linux/common/linux_machdep.h> |
| 64 | #include <compat/linux/common/linux_dirent.h> |
| 65 | #include <compat/linux/common/linux_ipc.h> |
| 66 | #include <compat/linux/common/linux_sem.h> |
| 67 | |
| 68 | #include <compat/linux/linux_syscallargs.h> |
| 69 | |
| 70 | static void bsd_to_linux_stat(struct stat *, struct linux_stat64 *); |
| 71 | |
| 72 | /* |
| 73 | * Convert a NetBSD stat structure to a Linux stat structure. |
| 74 | * Only the order of the fields and the padding in the structure |
| 75 | * is different. linux_fakedev is a machine-dependent function |
| 76 | * which optionally converts device driver major/minor numbers |
| 77 | * (XXX horrible, but what can you do against code that compares |
| 78 | * things against constant major device numbers? sigh) |
| 79 | */ |
| 80 | static void |
| 81 | bsd_to_linux_stat(struct stat *bsp, struct linux_stat64 *lsp) |
| 82 | { |
| 83 | lsp->lst_dev = linux_fakedev(bsp->st_dev, 0); |
| 84 | lsp->lst_ino = bsp->st_ino; |
| 85 | lsp->lst_mode = (linux_mode_t)bsp->st_mode; |
| 86 | if (bsp->st_nlink >= (1 << 15)) |
| 87 | lsp->lst_nlink = (1 << 15) - 1; |
| 88 | else |
| 89 | lsp->lst_nlink = (linux_nlink_t)bsp->st_nlink; |
| 90 | lsp->lst_uid = bsp->st_uid; |
| 91 | lsp->lst_gid = bsp->st_gid; |
| 92 | lsp->lst_rdev = linux_fakedev(bsp->st_rdev, 1); |
| 93 | lsp->lst_size = bsp->st_size; |
| 94 | lsp->lst_blksize = bsp->st_blksize; |
| 95 | lsp->lst_blocks = bsp->st_blocks; |
| 96 | lsp->lst_atime = bsp->st_atime; |
| 97 | lsp->lst_mtime = bsp->st_mtime; |
| 98 | lsp->lst_ctime = bsp->st_ctime; |
| 99 | # ifdef LINUX_STAT64_HAS_NSEC |
| 100 | lsp->lst_atime_nsec = bsp->st_atimensec; |
| 101 | lsp->lst_mtime_nsec = bsp->st_mtimensec; |
| 102 | lsp->lst_ctime_nsec = bsp->st_ctimensec; |
| 103 | # endif |
| 104 | # if LINUX_STAT64_HAS_BROKEN_ST_INO |
| 105 | lsp->__lst_ino = (linux_ino_t) bsp->st_ino; |
| 106 | # endif |
| 107 | } |
| 108 | |
| 109 | /* |
| 110 | * The stat functions below are plain sailing. stat and lstat are handled |
| 111 | * by one function to avoid code duplication. |
| 112 | */ |
| 113 | int |
| 114 | linux_sys_fstat64(struct lwp *l, const struct linux_sys_fstat64_args *uap, register_t *retval) |
| 115 | { |
| 116 | /* { |
| 117 | syscallarg(int) fd; |
| 118 | syscallarg(struct linux_stat64 *) sp; |
| 119 | } */ |
| 120 | struct linux_stat64 tmplst; |
| 121 | struct stat tmpst; |
| 122 | int error; |
| 123 | |
| 124 | error = do_sys_fstat(SCARG(uap, fd), &tmpst); |
| 125 | if (error != 0) |
| 126 | return error; |
| 127 | |
| 128 | bsd_to_linux_stat(&tmpst, &tmplst); |
| 129 | |
| 130 | return copyout(&tmplst, SCARG(uap, sp), sizeof tmplst); |
| 131 | } |
| 132 | |
| 133 | static int |
| 134 | linux_do_stat64(struct lwp *l, const struct linux_sys_stat64_args *uap, register_t *retval, int flags) |
| 135 | { |
| 136 | struct linux_stat64 tmplst; |
| 137 | struct stat tmpst; |
| 138 | int error; |
| 139 | |
| 140 | error = do_sys_stat(SCARG(uap, path), flags, &tmpst); |
| 141 | if (error != 0) |
| 142 | return error; |
| 143 | |
| 144 | bsd_to_linux_stat(&tmpst, &tmplst); |
| 145 | |
| 146 | return copyout(&tmplst, SCARG(uap, sp), sizeof tmplst); |
| 147 | } |
| 148 | |
| 149 | int |
| 150 | linux_sys_stat64(struct lwp *l, const struct linux_sys_stat64_args *uap, register_t *retval) |
| 151 | { |
| 152 | /* { |
| 153 | syscallarg(const char *) path; |
| 154 | syscallarg(struct linux_stat64 *) sp; |
| 155 | } */ |
| 156 | |
| 157 | return linux_do_stat64(l, uap, retval, FOLLOW); |
| 158 | } |
| 159 | |
| 160 | int |
| 161 | linux_sys_lstat64(struct lwp *l, const struct linux_sys_lstat64_args *uap, register_t *retval) |
| 162 | { |
| 163 | /* { |
| 164 | syscallarg(const char *) path; |
| 165 | syscallarg(struct linux_stat64 *) sp; |
| 166 | } */ |
| 167 | |
| 168 | return linux_do_stat64(l, (const void *)uap, retval, NOFOLLOW); |
| 169 | } |
| 170 | |
| 171 | int |
| 172 | linux_sys_fstatat64(struct lwp *l, const struct linux_sys_fstatat64_args *uap, register_t *retval) |
| 173 | { |
| 174 | /* { |
| 175 | syscallarg(int) fd; |
| 176 | syscallarg(const char *) path; |
| 177 | syscallarg(struct linux_stat64 *) sp; |
| 178 | syscallarg(int) flag; |
| 179 | } */ |
| 180 | struct linux_stat64 tmplst; |
| 181 | struct stat tmpst; |
| 182 | int error, nd_flag; |
| 183 | |
| 184 | if (SCARG(uap, flag) & LINUX_AT_SYMLINK_NOFOLLOW) |
| 185 | nd_flag = NOFOLLOW; |
| 186 | else |
| 187 | nd_flag = FOLLOW; |
| 188 | |
| 189 | error = do_sys_statat(l, SCARG(uap, fd), SCARG(uap, path), nd_flag, &tmpst); |
| 190 | if (error != 0) |
| 191 | return error; |
| 192 | |
| 193 | bsd_to_linux_stat(&tmpst, &tmplst); |
| 194 | |
| 195 | return copyout(&tmplst, SCARG(uap, sp), sizeof tmplst); |
| 196 | } |
| 197 | |
| 198 | #ifndef __alpha__ |
| 199 | int |
| 200 | linux_sys_truncate64(struct lwp *l, const struct linux_sys_truncate64_args *uap, register_t *retval) |
| 201 | { |
| 202 | /* { |
| 203 | syscallarg(const char *) path; |
| 204 | syscallarg(off_t) length; |
| 205 | } */ |
| 206 | struct sys_truncate_args ta; |
| 207 | |
| 208 | /* Linux doesn't have the 'pad' pseudo-parameter */ |
| 209 | SCARG(&ta, path) = SCARG(uap, path); |
| 210 | SCARG(&ta, PAD) = 0; |
| 211 | SCARG(&ta, length) = SCARG(uap, length); |
| 212 | |
| 213 | return sys_truncate(l, &ta, retval); |
| 214 | } |
| 215 | |
| 216 | int |
| 217 | linux_sys_ftruncate64(struct lwp *l, const struct linux_sys_ftruncate64_args *uap, register_t *retval) |
| 218 | { |
| 219 | /* { |
| 220 | syscallarg(unsigned int) fd; |
| 221 | syscallarg(off_t) length; |
| 222 | } */ |
| 223 | struct sys_ftruncate_args ta; |
| 224 | |
| 225 | /* Linux doesn't have the 'pad' pseudo-parameter */ |
| 226 | SCARG(&ta, fd) = SCARG(uap, fd); |
| 227 | SCARG(&ta, PAD) = 0; |
| 228 | SCARG(&ta, length) = SCARG(uap, length); |
| 229 | |
| 230 | return sys_ftruncate(l, &ta, retval); |
| 231 | } |
| 232 | #endif /* __alpha__ */ |
| 233 | |
| 234 | /* |
| 235 | * Linux 'readdir' call. This code is mostly taken from the |
| 236 | * SunOS getdents call (see compat/sunos/sunos_misc.c), though |
| 237 | * an attempt has been made to keep it a little cleaner. |
| 238 | * |
| 239 | * The d_off field contains the offset of the next valid entry, |
| 240 | * unless the older Linux getdents(2), which used to have it set |
| 241 | * to the offset of the entry itself. This function also doesn't |
| 242 | * need to deal with the old count == 1 glibc problem. |
| 243 | * |
| 244 | * Read in BSD-style entries, convert them, and copy them out. |
| 245 | * |
| 246 | * Note that this doesn't handle union-mounted filesystems. |
| 247 | */ |
| 248 | int |
| 249 | linux_sys_getdents64(struct lwp *l, const struct linux_sys_getdents64_args *uap, register_t *retval) |
| 250 | { |
| 251 | /* { |
| 252 | syscallarg(int) fd; |
| 253 | syscallarg(struct linux_dirent64 *) dent; |
| 254 | syscallarg(unsigned int) count; |
| 255 | } */ |
| 256 | struct dirent *bdp; |
| 257 | struct vnode *vp; |
| 258 | char *inp, *tbuf; /* BSD-format */ |
| 259 | int len, reclen; /* BSD-format */ |
| 260 | char *outp; /* Linux-format */ |
| 261 | int resid, linux_reclen = 0; /* Linux-format */ |
| 262 | file_t *fp; |
| 263 | struct uio auio; |
| 264 | struct iovec aiov; |
| 265 | struct linux_dirent64 idb; |
| 266 | off_t off; /* true file offset */ |
| 267 | int buflen, error, eofflag, nbytes; |
| 268 | struct vattr va; |
| 269 | off_t *cookiebuf = NULL, *cookie; |
| 270 | int ncookies; |
| 271 | |
| 272 | /* fd_getvnode() will use the descriptor for us */ |
| 273 | if ((error = fd_getvnode(SCARG(uap, fd), &fp)) != 0) |
| 274 | return (error); |
| 275 | |
| 276 | if ((fp->f_flag & FREAD) == 0) { |
| 277 | error = EBADF; |
| 278 | goto out1; |
| 279 | } |
| 280 | |
| 281 | vp = (struct vnode *)fp->f_data; |
| 282 | if (vp->v_type != VDIR) { |
| 283 | error = ENOTDIR; |
| 284 | goto out1; |
| 285 | } |
| 286 | |
| 287 | vn_lock(vp, LK_SHARED | LK_RETRY); |
| 288 | error = VOP_GETATTR(vp, &va, l->l_cred); |
| 289 | VOP_UNLOCK(vp); |
| 290 | if (error) |
| 291 | goto out1; |
| 292 | |
| 293 | nbytes = SCARG(uap, count); |
| 294 | buflen = min(MAXBSIZE, nbytes); |
| 295 | if (buflen < va.va_blocksize) |
| 296 | buflen = va.va_blocksize; |
| 297 | tbuf = malloc(buflen, M_TEMP, M_WAITOK); |
| 298 | |
| 299 | vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); |
| 300 | off = fp->f_offset; |
| 301 | again: |
| 302 | aiov.iov_base = tbuf; |
| 303 | aiov.iov_len = buflen; |
| 304 | auio.uio_iov = &aiov; |
| 305 | auio.uio_iovcnt = 1; |
| 306 | auio.uio_rw = UIO_READ; |
| 307 | auio.uio_resid = buflen; |
| 308 | auio.uio_offset = off; |
| 309 | UIO_SETUP_SYSSPACE(&auio); |
| 310 | /* |
| 311 | * First we read into the malloc'ed buffer, then |
| 312 | * we massage it into user space, one record at a time. |
| 313 | */ |
| 314 | error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, &cookiebuf, |
| 315 | &ncookies); |
| 316 | if (error) |
| 317 | goto out; |
| 318 | |
| 319 | inp = tbuf; |
| 320 | outp = (void *)SCARG(uap, dent); |
| 321 | resid = nbytes; |
| 322 | if ((len = buflen - auio.uio_resid) == 0) |
| 323 | goto eof; |
| 324 | |
| 325 | for (cookie = cookiebuf; len > 0; len -= reclen) { |
| 326 | bdp = (struct dirent *)inp; |
| 327 | reclen = bdp->d_reclen; |
| 328 | if (reclen & 3) |
| 329 | panic("linux_readdir" ); |
| 330 | if (bdp->d_fileno == 0) { |
| 331 | inp += reclen; /* it is a hole; squish it out */ |
| 332 | if (cookie) |
| 333 | off = *cookie++; |
| 334 | else |
| 335 | off += reclen; |
| 336 | continue; |
| 337 | } |
| 338 | linux_reclen = LINUX_RECLEN(&idb, bdp->d_namlen); |
| 339 | if (reclen > len || resid < linux_reclen) { |
| 340 | /* entry too big for buffer, so just stop */ |
| 341 | outp++; |
| 342 | break; |
| 343 | } |
| 344 | if (cookie) |
| 345 | off = *cookie++; /* each entry points to next */ |
| 346 | else |
| 347 | off += reclen; |
| 348 | /* |
| 349 | * Massage in place to make a Linux-shaped dirent (otherwise |
| 350 | * we have to worry about touching user memory outside of |
| 351 | * the copyout() call). |
| 352 | */ |
| 353 | idb.d_ino = bdp->d_fileno; |
| 354 | idb.d_type = bdp->d_type; |
| 355 | idb.d_off = off; |
| 356 | idb.d_reclen = (u_short)linux_reclen; |
| 357 | strcpy(idb.d_name, bdp->d_name); |
| 358 | if ((error = copyout((void *)&idb, outp, linux_reclen))) |
| 359 | goto out; |
| 360 | /* advance past this real entry */ |
| 361 | inp += reclen; |
| 362 | /* advance output past Linux-shaped entry */ |
| 363 | outp += linux_reclen; |
| 364 | resid -= linux_reclen; |
| 365 | } |
| 366 | |
| 367 | /* if we squished out the whole block, try again */ |
| 368 | if (outp == (void *)SCARG(uap, dent)) { |
| 369 | if (cookiebuf) |
| 370 | free(cookiebuf, M_TEMP); |
| 371 | cookiebuf = NULL; |
| 372 | goto again; |
| 373 | } |
| 374 | fp->f_offset = off; /* update the vnode offset */ |
| 375 | |
| 376 | eof: |
| 377 | *retval = nbytes - resid; |
| 378 | out: |
| 379 | VOP_UNLOCK(vp); |
| 380 | if (cookiebuf) |
| 381 | free(cookiebuf, M_TEMP); |
| 382 | free(tbuf, M_TEMP); |
| 383 | out1: |
| 384 | fd_putfile(SCARG(uap, fd)); |
| 385 | return error; |
| 386 | } |
| 387 | |