linux_file64.c revision 1.19 1 /* $NetBSD: linux_file64.c,v 1.19 2003/01/18 08:02:52 thorpej Exp $ */
2
3 /*-
4 * Copyright (c) 1995, 1998, 2000 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 * 3. All advertising materials mentioning features or use of this software
19 * must display the following acknowledgement:
20 * This product includes software developed by the NetBSD
21 * Foundation, Inc. and its contributors.
22 * 4. Neither the name of The NetBSD Foundation nor the names of its
23 * contributors may be used to endorse or promote products derived
24 * from this software without specific prior written permission.
25 *
26 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
27 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
28 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
29 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
30 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
31 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
32 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
33 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
34 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
35 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
36 * POSSIBILITY OF SUCH DAMAGE.
37 */
38
39 /*
40 * Linux 64bit filesystem calls. Used on 32bit archs, not used on 64bit ones.
41 */
42
43 #include <sys/cdefs.h>
44 __KERNEL_RCSID(0, "$NetBSD: linux_file64.c,v 1.19 2003/01/18 08:02:52 thorpej Exp $");
45
46 #include <sys/param.h>
47 #include <sys/systm.h>
48 #include <sys/namei.h>
49 #include <sys/proc.h>
50 #include <sys/dirent.h>
51 #include <sys/file.h>
52 #include <sys/stat.h>
53 #include <sys/filedesc.h>
54 #include <sys/ioctl.h>
55 #include <sys/kernel.h>
56 #include <sys/mount.h>
57 #include <sys/malloc.h>
58 #include <sys/vnode.h>
59 #include <sys/tty.h>
60 #include <sys/conf.h>
61
62 #include <sys/sa.h>
63 #include <sys/syscallargs.h>
64
65 #include <compat/linux/common/linux_types.h>
66 #include <compat/linux/common/linux_signal.h>
67 #include <compat/linux/common/linux_fcntl.h>
68 #include <compat/linux/common/linux_util.h>
69 #include <compat/linux/common/linux_machdep.h>
70 #include <compat/linux/common/linux_dirent.h>
71
72 #include <compat/linux/linux_syscallargs.h>
73
74 #ifndef alpha
75
76 static void bsd_to_linux_stat __P((struct stat *, struct linux_stat64 *));
77 static int linux_do_stat64 __P((struct lwp *, void *, register_t *, int));
78
79 /*
80 * Convert a NetBSD stat structure to a Linux stat structure.
81 * Only the order of the fields and the padding in the structure
82 * is different. linux_fakedev is a machine-dependent function
83 * which optionally converts device driver major/minor numbers
84 * (XXX horrible, but what can you do against code that compares
85 * things against constant major device numbers? sigh)
86 */
87 static void
88 bsd_to_linux_stat(bsp, lsp)
89 struct stat *bsp;
90 struct linux_stat64 *lsp;
91 {
92 lsp->lst_dev = linux_fakedev(bsp->st_dev, 0);
93 lsp->lst_ino = bsp->st_ino;
94 lsp->lst_mode = (linux_mode_t)bsp->st_mode;
95 if (bsp->st_nlink >= (1 << 15))
96 lsp->lst_nlink = (1 << 15) - 1;
97 else
98 lsp->lst_nlink = (linux_nlink_t)bsp->st_nlink;
99 lsp->lst_uid = bsp->st_uid;
100 lsp->lst_gid = bsp->st_gid;
101 lsp->lst_rdev = linux_fakedev(bsp->st_rdev, 1);
102 lsp->lst_size = bsp->st_size;
103 lsp->lst_blksize = bsp->st_blksize;
104 lsp->lst_blocks = bsp->st_blocks;
105 lsp->lst_atime = bsp->st_atime;
106 lsp->lst_mtime = bsp->st_mtime;
107 lsp->lst_ctime = bsp->st_ctime;
108 #if LINUX_STAT64_HAS_BROKEN_ST_INO
109 lsp->__lst_ino = (linux_ino_t) bsp->st_ino;
110 #endif
111 }
112
113 /*
114 * The stat functions below are plain sailing. stat and lstat are handled
115 * by one function to avoid code duplication.
116 */
117 int
118 linux_sys_fstat64(l, v, retval)
119 struct lwp *l;
120 void *v;
121 register_t *retval;
122 {
123 struct linux_sys_fstat64_args /* {
124 syscallarg(int) fd;
125 syscallarg(struct linux_stat64 *) sp;
126 } */ *uap = v;
127 struct proc *p = l->l_proc;
128 struct sys___fstat13_args fsa;
129 struct linux_stat64 tmplst;
130 struct stat *st,tmpst;
131 caddr_t sg;
132 int error;
133
134 sg = stackgap_init(p, 0);
135
136 st = stackgap_alloc(p, &sg, sizeof (struct stat));
137
138 SCARG(&fsa, fd) = SCARG(uap, fd);
139 SCARG(&fsa, sb) = st;
140
141 if ((error = sys___fstat13(l, &fsa, retval)))
142 return error;
143
144 if ((error = copyin(st, &tmpst, sizeof tmpst)))
145 return error;
146
147 bsd_to_linux_stat(&tmpst, &tmplst);
148
149 if ((error = copyout(&tmplst, SCARG(uap, sp), sizeof tmplst)))
150 return error;
151
152 return 0;
153 }
154
155 static int
156 linux_do_stat64(l, v, retval, dolstat)
157 struct lwp *l;
158 void *v;
159 register_t *retval;
160 int dolstat;
161 {
162 struct proc *p = l->l_proc;
163 struct sys___stat13_args sa;
164 struct linux_stat64 tmplst;
165 struct stat *st, tmpst;
166 caddr_t sg;
167 int error;
168 struct linux_sys_stat64_args *uap = v;
169
170 sg = stackgap_init(p, 0);
171 st = stackgap_alloc(p, &sg, sizeof (struct stat));
172 CHECK_ALT_EXIST(p, &sg, SCARG(uap, path));
173
174 SCARG(&sa, ub) = st;
175 SCARG(&sa, path) = SCARG(uap, path);
176
177 if ((error = (dolstat ? sys___lstat13(l, &sa, retval) :
178 sys___stat13(l, &sa, retval))))
179 return error;
180
181 if ((error = copyin(st, &tmpst, sizeof tmpst)))
182 return error;
183
184 bsd_to_linux_stat(&tmpst, &tmplst);
185
186 if ((error = copyout(&tmplst, SCARG(uap, sp), sizeof tmplst)))
187 return error;
188
189 return 0;
190 }
191
192 int
193 linux_sys_stat64(l, v, retval)
194 struct lwp *l;
195 void *v;
196 register_t *retval;
197 {
198 struct linux_sys_stat64_args /* {
199 syscallarg(const char *) path;
200 syscallarg(struct linux_stat64 *) sp;
201 } */ *uap = v;
202
203 return linux_do_stat64(l, uap, retval, 0);
204 }
205
206 int
207 linux_sys_lstat64(l, v, retval)
208 struct lwp *l;
209 void *v;
210 register_t *retval;
211 {
212 struct linux_sys_lstat64_args /* {
213 syscallarg(const char *) path;
214 syscallarg(struct linux_stat64 *) sp;
215 } */ *uap = v;
216
217 return linux_do_stat64(l, uap, retval, 1);
218 }
219
220 int
221 linux_sys_truncate64(l, v, retval)
222 struct lwp *l;
223 void *v;
224 register_t *retval;
225 {
226 struct linux_sys_truncate64_args /* {
227 syscallarg(const char *) path;
228 syscallarg(off_t) length;
229 } */ *uap = v;
230 struct proc *p = l->l_proc;
231 caddr_t sg = stackgap_init(p, 0);
232
233 CHECK_ALT_EXIST(p, &sg, SCARG(uap, path));
234
235 return sys_truncate(l, uap, retval);
236 }
237
238 #if !defined(__m68k__)
239 static void bsd_to_linux_flock64 __P((struct linux_flock64 *,
240 const struct flock *));
241 static void linux_to_bsd_flock64 __P((struct flock *,
242 const struct linux_flock64 *));
243
244 static void
245 bsd_to_linux_flock64(lfp, bfp)
246 struct linux_flock64 *lfp;
247 const struct flock *bfp;
248 {
249
250 lfp->l_start = bfp->l_start;
251 lfp->l_len = bfp->l_len;
252 lfp->l_pid = bfp->l_pid;
253 lfp->l_whence = bfp->l_whence;
254 switch (bfp->l_type) {
255 case F_RDLCK:
256 lfp->l_type = LINUX_F_RDLCK;
257 break;
258 case F_UNLCK:
259 lfp->l_type = LINUX_F_UNLCK;
260 break;
261 case F_WRLCK:
262 lfp->l_type = LINUX_F_WRLCK;
263 break;
264 }
265 }
266
267 static void
268 linux_to_bsd_flock64(bfp, lfp)
269 struct flock *bfp;
270 const struct linux_flock64 *lfp;
271 {
272
273 bfp->l_start = lfp->l_start;
274 bfp->l_len = lfp->l_len;
275 bfp->l_pid = lfp->l_pid;
276 bfp->l_whence = lfp->l_whence;
277 switch (lfp->l_type) {
278 case LINUX_F_RDLCK:
279 bfp->l_type = F_RDLCK;
280 break;
281 case LINUX_F_UNLCK:
282 bfp->l_type = F_UNLCK;
283 break;
284 case LINUX_F_WRLCK:
285 bfp->l_type = F_WRLCK;
286 break;
287 }
288 }
289
290 int
291 linux_sys_fcntl64(l, v, retval)
292 struct lwp *l;
293 void *v;
294 register_t *retval;
295 {
296 struct linux_sys_fcntl64_args /* {
297 syscallarg(int) fd;
298 syscallarg(int) cmd;
299 syscallarg(void *) arg;
300 } */ *uap = v;
301 struct proc *p = l->l_proc;
302 struct sys_fcntl_args fca;
303 struct linux_flock64 lfl;
304 struct flock bfl, *bfp;
305 int error;
306 caddr_t sg;
307 void *arg = SCARG(uap, arg);
308 int cmd = SCARG(uap, cmd);
309 int fd = SCARG(uap, fd);
310
311 switch (cmd) {
312 case LINUX_F_GETLK64:
313 sg = stackgap_init(p, 0);
314 bfp = (struct flock *) stackgap_alloc(p, &sg, sizeof *bfp);
315 if ((error = copyin(arg, &lfl, sizeof lfl)) != 0)
316 return error;
317 linux_to_bsd_flock64(&bfl, &lfl);
318 if ((error = copyout(&bfl, bfp, sizeof bfl)) != 0)
319 return error;
320 SCARG(&fca, fd) = fd;
321 SCARG(&fca, cmd) = F_GETLK;
322 SCARG(&fca, arg) = bfp;
323 if ((error = sys_fcntl(l, &fca, retval)) != 0)
324 return error;
325 if ((error = copyin(bfp, &bfl, sizeof bfl)) != 0)
326 return error;
327 bsd_to_linux_flock64(&lfl, &bfl);
328 return copyout(&lfl, arg, sizeof lfl);
329 case LINUX_F_SETLK64:
330 case LINUX_F_SETLKW64:
331 cmd = (cmd == LINUX_F_SETLK64 ? F_SETLK : F_SETLKW);
332 if ((error = copyin(arg, &lfl, sizeof lfl)) != 0)
333 return error;
334 linux_to_bsd_flock64(&bfl, &lfl);
335 sg = stackgap_init(p, 0);
336 bfp = (struct flock *) stackgap_alloc(p, &sg, sizeof *bfp);
337 if ((error = copyout(&bfl, bfp, sizeof bfl)) != 0)
338 return error;
339 SCARG(&fca, fd) = fd;
340 SCARG(&fca, cmd) = cmd;
341 SCARG(&fca, arg) = bfp;
342 return sys_fcntl(l, &fca, retval);
343 default:
344 return linux_sys_fcntl(l, v, retval);
345 }
346 }
347 #endif /* !m68k */
348
349 #endif /* !alpha */
350
351 /*
352 * Linux 'readdir' call. This code is mostly taken from the
353 * SunOS getdents call (see compat/sunos/sunos_misc.c), though
354 * an attempt has been made to keep it a little cleaner.
355 *
356 * The d_off field contains the offset of the next valid entry,
357 * unless the older Linux getdents(2), which used to have it set
358 * to the offset of the entry itself. This function also doesn't
359 * need to deal with the old count == 1 glibc problem.
360 *
361 * Read in BSD-style entries, convert them, and copy them out.
362 *
363 * Note that this doesn't handle union-mounted filesystems.
364 */
365 int
366 linux_sys_getdents64(l, v, retval)
367 struct lwp *l;
368 void *v;
369 register_t *retval;
370 {
371 struct linux_sys_getdents_args /* {
372 syscallarg(int) fd;
373 syscallarg(struct linux_dirent64 *) dent;
374 syscallarg(unsigned int) count;
375 } */ *uap = v;
376 struct proc *p = l->l_proc;
377 struct dirent *bdp;
378 struct vnode *vp;
379 caddr_t inp, buf; /* BSD-format */
380 int len, reclen; /* BSD-format */
381 caddr_t outp; /* Linux-format */
382 int resid, linux_reclen = 0; /* Linux-format */
383 struct file *fp;
384 struct uio auio;
385 struct iovec aiov;
386 struct linux_dirent64 idb;
387 off_t off; /* true file offset */
388 int buflen, error, eofflag, nbytes;
389 struct vattr va;
390 off_t *cookiebuf = NULL, *cookie;
391 int ncookies;
392
393 /* getvnode() will use the descriptor for us */
394 if ((error = getvnode(p->p_fd, SCARG(uap, fd), &fp)) != 0)
395 return (error);
396
397 if ((fp->f_flag & FREAD) == 0) {
398 error = EBADF;
399 goto out1;
400 }
401
402 vp = (struct vnode *)fp->f_data;
403 if (vp->v_type != VDIR) {
404 error = EINVAL;
405 goto out1;
406 }
407
408 if ((error = VOP_GETATTR(vp, &va, p->p_ucred, p)))
409 goto out1;
410
411 nbytes = SCARG(uap, count);
412 buflen = min(MAXBSIZE, nbytes);
413 if (buflen < va.va_blocksize)
414 buflen = va.va_blocksize;
415 buf = malloc(buflen, M_TEMP, M_WAITOK);
416
417 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
418 off = fp->f_offset;
419 again:
420 aiov.iov_base = buf;
421 aiov.iov_len = buflen;
422 auio.uio_iov = &aiov;
423 auio.uio_iovcnt = 1;
424 auio.uio_rw = UIO_READ;
425 auio.uio_segflg = UIO_SYSSPACE;
426 auio.uio_procp = p;
427 auio.uio_resid = buflen;
428 auio.uio_offset = off;
429 /*
430 * First we read into the malloc'ed buffer, then
431 * we massage it into user space, one record at a time.
432 */
433 error = VOP_READDIR(vp, &auio, fp->f_cred, &eofflag, &cookiebuf,
434 &ncookies);
435 if (error)
436 goto out;
437
438 inp = buf;
439 outp = (caddr_t)SCARG(uap, dent);
440 resid = nbytes;
441 if ((len = buflen - auio.uio_resid) == 0)
442 goto eof;
443
444 for (cookie = cookiebuf; len > 0; len -= reclen) {
445 bdp = (struct dirent *)inp;
446 reclen = bdp->d_reclen;
447 if (reclen & 3)
448 panic("linux_readdir");
449 if (bdp->d_fileno == 0) {
450 inp += reclen; /* it is a hole; squish it out */
451 off = *cookie++;
452 continue;
453 }
454 linux_reclen = LINUX_RECLEN(&idb, bdp->d_namlen);
455 if (reclen > len || resid < linux_reclen) {
456 /* entry too big for buffer, so just stop */
457 outp++;
458 break;
459 }
460 off = *cookie++; /* each entry points to next */
461 /*
462 * Massage in place to make a Linux-shaped dirent (otherwise
463 * we have to worry about touching user memory outside of
464 * the copyout() call).
465 */
466 idb.d_ino = bdp->d_fileno;
467 idb.d_type = bdp->d_type;
468 idb.d_off = off;
469 idb.d_reclen = (u_short)linux_reclen;
470 strcpy(idb.d_name, bdp->d_name);
471 if ((error = copyout((caddr_t)&idb, outp, linux_reclen)))
472 goto out;
473 /* advance past this real entry */
474 inp += reclen;
475 /* advance output past Linux-shaped entry */
476 outp += linux_reclen;
477 resid -= linux_reclen;
478 }
479
480 /* if we squished out the whole block, try again */
481 if (outp == (caddr_t)SCARG(uap, dent))
482 goto again;
483 fp->f_offset = off; /* update the vnode offset */
484
485 eof:
486 *retval = nbytes - resid;
487 out:
488 VOP_UNLOCK(vp, 0);
489 if (cookiebuf)
490 free(cookiebuf, M_TEMP);
491 free(buf, M_TEMP);
492 out1:
493 FILE_UNUSE(fp, p);
494 return error;
495 }
496