1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
2 * vim:expandtab:shiftwidth=8:tabstop=8:
4 * Copyright (c) 2002, 2003 Cluster File Systems, Inc.
5 * Author: Peter Braam <braam@clusterfs.com>
6 * Author: Phil Schwan <phil@clusterfs.com>
7 * Author: Andreas Dilger <adilger@clusterfs.com>
9 * This file is part of Lustre, http://www.lustre.org.
11 * Lustre is free software; you can redistribute it and/or
12 * modify it under the terms of version 2 of the GNU General Public
13 * License as published by the Free Software Foundation.
15 * Lustre is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 * GNU General Public License for more details.
20 * You should have received a copy of the GNU General Public License
21 * along with Lustre; if not, write to the Free Software
22 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
25 #define DEBUG_SUBSYSTEM S_LLITE
26 #include <linux/lustre_dlm.h>
27 #include <linux/lustre_lite.h>
28 #include <linux/pagemap.h>
29 #include <linux/file.h>
30 #include <linux/lustre_acl.h>
31 #if (LINUX_VERSION_CODE < KERNEL_VERSION(2,5,0))
32 #include <linux/lustre_compat25.h>
34 #include "llite_internal.h"
35 #include <linux/obd_lov.h>
37 int ll_md_och_close(struct obd_export *md_exp, struct inode *inode,
38 struct obd_client_handle *och)
40 struct ptlrpc_request *req = NULL;
41 struct obdo *obdo = NULL;
42 struct obd_device *obd;
46 obd = class_exp2obd(md_exp);
48 CERROR("Invalid MDC connection handle "LPX64"\n",
49 md_exp->exp_handle.h_cookie);
55 * here we check if this is forced umount. If so this is called on
56 * canceling "open lock" and we do not call md_close() in this case , as
57 * it will not successful, as import is already deactivated.
59 if (obd->obd_no_recov)
62 /* closing opened file */
67 obdo->o_id = inode->i_ino;
68 obdo->o_valid = OBD_MD_FLID;
69 obdo_from_inode(obdo, inode, (OBD_MD_FLTYPE | OBD_MD_FLMODE |
70 OBD_MD_FLSIZE | OBD_MD_FLBLOCKS |
71 OBD_MD_FLATIME | OBD_MD_FLMTIME |
73 if (0 /* ll_is_inode_dirty(inode) */) {
74 obdo->o_flags = MDS_BFLAG_UNCOMMITTED_WRITES;
75 obdo->o_valid |= OBD_MD_FLFLAGS;
77 obdo->o_fid = id_fid(&ll_i2info(inode)->lli_id);
78 obdo->o_mds = id_group(&ll_i2info(inode)->lli_id);
79 rc = md_close(md_exp, obdo, och, &req);
84 * we are the last writer, so the MDS has instructed us to get
85 * the file size and any write cookies, then close again.
88 //ll_queue_done_writing(inode);
91 CERROR("inode %lu mdc close failed: rc = %d\n",
92 (unsigned long)inode->i_ino, rc);
95 /* objects are destroed on OST only if metadata close was
98 rc = ll_objects_destroy(req, inode, 1);
100 CERROR("inode %lu ll_objects destroy: rc = %d\n",
104 ptlrpc_req_finished(req);
107 mdc_clear_open_replay_data(md_exp, och);
108 och->och_fh.cookie = DEAD_HANDLE_MAGIC;
109 OBD_FREE(och, sizeof *och);
113 int ll_md_real_close(struct obd_export *md_exp,
114 struct inode *inode, int flags)
116 struct ll_inode_info *lli = ll_i2info(inode);
117 struct obd_client_handle **och_p;
118 struct obd_client_handle *och;
123 if (flags & FMODE_WRITE) {
124 och_p = &lli->lli_mds_write_och;
125 och_usecount = &lli->lli_open_fd_write_count;
126 } else if (flags & FMODE_EXEC) {
127 och_p = &lli->lli_mds_exec_och;
128 och_usecount = &lli->lli_open_fd_exec_count;
130 och_p = &lli->lli_mds_read_och;
131 och_usecount = &lli->lli_open_fd_read_count;
134 down(&lli->lli_och_sem);
135 if (*och_usecount) { /* There are still users of this handle, so
137 up(&lli->lli_och_sem);
143 up(&lli->lli_och_sem);
146 * there might be a race and somebody have freed this och
147 * already. Another way to have this twice called is if file closing
148 * will fail due to netwok problems and on umount lock will be canceled
149 * and this will be called from block_ast callack.
151 if (och && och->och_fh.cookie != DEAD_HANDLE_MAGIC)
152 rc = ll_md_och_close(md_exp, inode, och);
157 int ll_md_close(struct obd_export *md_exp, struct inode *inode,
160 struct ll_file_data *fd = file->private_data;
161 struct ll_inode_info *lli = ll_i2info(inode);
165 /* clear group lock, if present */
166 if (fd->fd_flags & LL_FILE_GROUP_LOCKED) {
167 struct lov_stripe_md *lsm = ll_i2info(inode)->lli_smd;
168 fd->fd_flags &= ~(LL_FILE_GROUP_LOCKED|LL_FILE_IGNORE_LOCK);
169 rc = ll_extent_unlock(fd, inode, lsm, LCK_GROUP,
173 /* Let's see if we have good enough OPEN lock on the file and if
174 we can skip talking to MDS */
175 if (file->f_dentry->d_inode) {
177 struct obd_device *obddev;
178 struct lustre_handle lockh;
179 int flags = LDLM_FL_BLOCK_GRANTED;
180 struct ldlm_res_id file_res_id = {.name = {id_fid(&lli->lli_id),
181 id_group(&lli->lli_id)}};
182 ldlm_policy_data_t policy = {.l_inodebits={MDS_INODELOCK_OPEN}};
184 down(&lli->lli_och_sem);
185 if (fd->fd_omode & FMODE_WRITE) {
187 LASSERT(lli->lli_open_fd_write_count);
188 lli->lli_open_fd_write_count--;
189 } else if (fd->fd_omode & FMODE_EXEC) {
191 LASSERT(lli->lli_open_fd_exec_count);
192 lli->lli_open_fd_exec_count--;
195 LASSERT(lli->lli_open_fd_read_count);
196 lli->lli_open_fd_read_count--;
198 up(&lli->lli_och_sem);
200 obddev = md_get_real_obd(md_exp, &lli->lli_id);
201 if (!ldlm_lock_match(obddev->obd_namespace, flags, &file_res_id,
202 LDLM_IBITS, &policy, lockmode, &lockh))
204 rc = ll_md_real_close(md_exp, file->f_dentry->d_inode,
207 ldlm_lock_decref(&lockh, lockmode);
211 file->private_data = NULL;
212 OBD_SLAB_FREE(fd, ll_file_data_slab, sizeof(*fd));
216 /* While this returns an error code, fput() the caller does not, so we need
217 * to make every effort to clean up all of our state here. Also, applications
218 * rarely check close errors and even if an error is returned they will not
219 * re-try the close call.
221 int ll_file_release(struct inode *inode, struct file *file)
223 struct ll_file_data *fd;
224 struct ll_sb_info *sbi = ll_i2sbi(inode);
228 CDEBUG(D_VFSTRACE, "VFS Op:inode="DLID4"(%p)\n",
229 OLID4(&ll_i2info(inode)->lli_id), inode);
231 /* don't do anything for / */
232 if (inode->i_sb->s_root == file->f_dentry)
235 lprocfs_counter_incr(sbi->ll_stats, LPROC_LL_RELEASE);
236 fd = (struct ll_file_data *)file->private_data;
239 rc = ll_md_close(sbi->ll_md_exp, inode, file);
243 static int ll_intent_file_open(struct file *file, void *lmm,
244 int lmmsize, struct lookup_intent *itp)
246 struct ll_sb_info *sbi = ll_i2sbi(file->f_dentry->d_inode);
247 struct dentry *parent = file->f_dentry->d_parent;
248 const char *name = file->f_dentry->d_name.name;
249 const int len = file->f_dentry->d_name.len;
250 struct lustre_handle lockh;
251 struct mdc_op_data *op_data;
257 OBD_ALLOC(op_data, sizeof(*op_data));
261 ll_prepare_mdc_data(op_data, parent->d_inode, NULL,
264 rc = md_enqueue(sbi->ll_md_exp, LDLM_IBITS, itp, LCK_PR, op_data,
265 &lockh, lmm, lmmsize, ldlm_completion_ast,
266 ll_mdc_blocking_ast, NULL);
267 OBD_FREE(op_data, sizeof(*op_data));
269 if (LUSTRE_IT(itp)->it_lock_mode)
270 memcpy(&LUSTRE_IT(itp)->it_lock_handle,
271 &lockh, sizeof(lockh));
274 CERROR("lock enqueue: err: %d\n", rc);
279 void ll_och_fill(struct inode *inode, struct lookup_intent *it,
280 struct obd_client_handle *och)
282 struct ptlrpc_request *req = LUSTRE_IT(it)->it_data;
283 struct ll_inode_info *lli = ll_i2info(inode);
284 struct mds_body *body;
287 body = lustre_msg_buf (req->rq_repmsg, 1, sizeof (*body));
288 LASSERT (body != NULL); /* reply already checked out */
289 LASSERT_REPSWABBED (req, 1); /* and swabbed down */
291 memcpy(&och->och_fh, &body->handle, sizeof(body->handle));
292 och->och_magic = OBD_CLIENT_HANDLE_MAGIC;
293 lli->lli_io_epoch = body->io_epoch;
294 mdc_set_open_replay_data(ll_i2mdexp(inode), och,
295 LUSTRE_IT(it)->it_data);
298 int ll_local_open(struct file *file, struct lookup_intent *it,
299 struct obd_client_handle *och)
301 struct ll_file_data *fd;
305 ll_och_fill(file->f_dentry->d_inode, it, och);
307 LASSERTF(file->private_data == NULL, "file %.*s/%.*s ino %lu/%u (%o)\n",
308 file->f_dentry->d_name.len, file->f_dentry->d_name.name,
309 file->f_dentry->d_parent->d_name.len,
310 file->f_dentry->d_parent->d_name.name,
311 file->f_dentry->d_inode->i_ino,
312 file->f_dentry->d_inode->i_generation,
313 file->f_dentry->d_inode->i_mode);
315 OBD_SLAB_ALLOC(fd, ll_file_data_slab, SLAB_KERNEL, sizeof *fd);
317 /* We can't handle this well without reorganizing ll_file_open and
318 * ll_md_close(), so don't even try right now. */
321 file->private_data = fd;
322 ll_readahead_init(file->f_dentry->d_inode, &fd->fd_ras);
323 fd->fd_omode = it->it_flags;
327 /* Open a file, and (for the very first open) create objects on the OSTs at
328 * this time. If opened with O_LOV_DELAY_CREATE, then we don't do the object
329 * creation or open until ll_lov_setstripe() ioctl is called. We grab
330 * lli_open_sem to ensure no other process will create objects, send the
331 * stripe MD to the MDS, or try to destroy the objects if that fails.
333 * If we already have the stripe MD locally then we don't request it in
334 * mdc_open(), by passing a lmm_size = 0.
336 * It is up to the application to ensure no other processes open this file
337 * in the O_LOV_DELAY_CREATE case, or the default striping pattern will be
338 * used. We might be able to avoid races of that sort by getting lli_open_sem
339 * before returning in the O_LOV_DELAY_CREATE case and dropping it here
340 * or in ll_file_release(), but I'm not sure that is desirable/necessary.
342 int ll_file_open(struct inode *inode, struct file *file)
344 struct ll_inode_info *lli = ll_i2info(inode);
345 struct lookup_intent *it, oit = { .it_op = IT_OPEN,
346 .it_flags = file->f_flags };
347 struct lov_stripe_md *lsm;
348 struct ptlrpc_request *req;
350 struct obd_client_handle **och_p;
354 CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p), flags %o\n",
355 inode->i_ino, inode->i_generation, inode, file->f_flags);
357 /* don't do anything for / */
358 if (inode->i_sb->s_root == file->f_dentry)
361 if ((file->f_flags+1) & O_ACCMODE)
363 if (file->f_flags & O_TRUNC)
369 * sometimes LUSTRE_IT(it) may not be allocated like opening file by
370 * dentry_open() from GNS stuff.
372 if (!it || !LUSTRE_IT(it)) {
374 rc = ll_intent_alloc(it);
379 lprocfs_counter_incr(ll_i2sbi(inode)->ll_stats, LPROC_LL_OPEN);
382 * mdc_intent_lock() didn't get a request ref if there was an open
383 * error, so don't do cleanup on the * request here (bug 3430)
385 if (LUSTRE_IT(it)->it_disposition) {
386 rc = it_open_error(DISP_OPEN_OPEN, it);
391 /* Let's see if we have file open on MDS already. */
392 if (it->it_flags & FMODE_WRITE) {
393 och_p = &lli->lli_mds_write_och;
394 och_usecount = &lli->lli_open_fd_write_count;
395 } else if (it->it_flags & FMODE_EXEC) {
396 och_p = &lli->lli_mds_exec_och;
397 och_usecount = &lli->lli_open_fd_exec_count;
399 och_p = &lli->lli_mds_read_och;
400 och_usecount = &lli->lli_open_fd_read_count;
403 down(&lli->lli_och_sem);
404 if (*och_p) { /* Open handle is present */
405 if (LUSTRE_IT(it)->it_disposition) {
406 struct obd_client_handle *och;
407 /* Well, there's extra open request that we do not need,
408 let's close it somehow*/
409 OBD_ALLOC(och, sizeof (struct obd_client_handle));
411 up(&lli->lli_och_sem);
415 ll_och_fill(inode, it, och);
416 /* ll_md_och_close() will free och */
417 ll_md_och_close(ll_i2mdexp(inode), inode, och);
421 rc = ll_local_open(file, it, NULL);
425 LASSERT(*och_usecount == 0);
426 OBD_ALLOC(*och_p, sizeof (struct obd_client_handle));
428 GOTO(out, rc = -ENOMEM);
431 if (!it || !LUSTRE_IT(it) || !LUSTRE_IT(it)->it_disposition) {
432 /* We are going to replace intent here, and that may
433 possibly change access mode (FMODE_EXEC can only be
434 set in intent), but I hope it never happens (I was
435 not able to trigger it yet at least) -- green */
436 /* FIXME: FMODE_EXEC is not covered by O_ACCMODE! */
437 LASSERT(!(it->it_flags & FMODE_EXEC));
438 LASSERTF((it->it_flags & O_ACCMODE) ==
439 (oit.it_flags & O_ACCMODE), "Changing intent "
440 "flags %x to incompatible %x\n",it->it_flags,
443 rc = ll_intent_file_open(file, NULL, 0, it);
446 rc = it_open_error(DISP_OPEN_OPEN, it);
448 GOTO(out_och_free, rc);
450 mdc_set_lock_data(NULL, &LUSTRE_IT(it)->it_lock_handle,
451 file->f_dentry->d_inode);
453 lprocfs_counter_incr(ll_i2sbi(inode)->ll_stats, LPROC_LL_OPEN);
454 rc = ll_local_open(file, it, *och_p);
455 LASSERTF(rc == 0, "rc = %d\n", rc);
457 up(&lli->lli_och_sem);
458 /* Must do this outside lli_och_sem lock to prevent deadlock where
459 different kind of OPEN lock for this same inode gets cancelled
460 by ldlm_cancel_lru */
462 if (!S_ISREG(inode->i_mode))
467 if (file->f_flags & O_LOV_DELAY_CREATE ||
468 !(file->f_mode & FMODE_WRITE)) {
469 CDEBUG(D_INODE, "object creation was delayed\n");
473 file->f_flags &= ~O_LOV_DELAY_CREATE;
476 req = LUSTRE_IT(it)->it_data;
477 ll_intent_drop_lock(it);
478 ll_intent_release(it);
479 ptlrpc_req_finished(req);
481 ll_open_complete(inode);
485 OBD_FREE(*och_p, sizeof (struct obd_client_handle));
486 *och_p = NULL; /* OBD_FREE writes some magic there */
489 up(&lli->lli_och_sem);
495 /* Fills the obdo with the attributes for the inode defined by lsm */
496 int ll_lsm_getattr(struct obd_export *exp, struct lov_stripe_md *lsm,
499 struct ptlrpc_request_set *set;
503 LASSERT(lsm != NULL);
505 memset(oa, 0, sizeof *oa);
506 oa->o_id = lsm->lsm_object_id;
507 oa->o_gr = lsm->lsm_object_gr;
508 oa->o_mode = S_IFREG;
509 oa->o_valid = OBD_MD_FLID | OBD_MD_FLTYPE | OBD_MD_FLSIZE |
510 OBD_MD_FLBLOCKS | OBD_MD_FLBLKSZ | OBD_MD_FLMTIME |
511 OBD_MD_FLCTIME | OBD_MD_FLGROUP;
513 set = ptlrpc_prep_set();
517 rc = obd_getattr_async(exp, oa, lsm, set);
519 rc = ptlrpc_set_wait(set);
520 ptlrpc_set_destroy(set);
525 oa->o_valid &= (OBD_MD_FLBLOCKS | OBD_MD_FLBLKSZ | OBD_MD_FLMTIME |
526 OBD_MD_FLCTIME | OBD_MD_FLSIZE);
530 static inline void ll_remove_suid(struct inode *inode)
534 /* set S_IGID if S_IXGRP is set, and always set S_ISUID */
535 mode = (inode->i_mode & S_IXGRP)*(S_ISGID/S_IXGRP) | S_ISUID;
537 /* was any of the uid bits set? */
538 mode &= inode->i_mode;
539 if (mode && !capable(CAP_FSETID)) {
540 inode->i_mode &= ~mode;
541 // XXX careful here - we cannot change the size
545 static int ll_lock_to_stripe_offset(struct inode *inode, struct ldlm_lock *lock)
547 struct ll_inode_info *lli = ll_i2info(inode);
548 struct lov_stripe_md *lsm = lli->lli_smd;
549 struct obd_export *exp = ll_i2dtexp(inode);
552 struct ldlm_lock *lock;
553 struct lov_stripe_md *lsm;
554 } key = { .name = "lock_to_stripe", .lock = lock, .lsm = lsm };
555 __u32 stripe, vallen = sizeof(stripe);
559 if (lsm->lsm_stripe_count == 1)
560 GOTO(check, stripe = 0);
562 /* get our offset in the lov */
563 rc = obd_get_info(exp, sizeof(key), &key, &vallen, &stripe);
565 CERROR("obd_get_info: rc = %d\n", rc);
568 LASSERT(stripe < lsm->lsm_stripe_count);
571 if (lsm->lsm_oinfo[stripe].loi_id != lock->l_resource->lr_name.name[0]||
572 lsm->lsm_oinfo[stripe].loi_gr != lock->l_resource->lr_name.name[2]){
573 LDLM_ERROR(lock, "resource doesn't match object "LPU64"/"LPU64
574 " inode=%lu/%u (%p)\n",
575 lsm->lsm_oinfo[stripe].loi_id,
576 lsm->lsm_oinfo[stripe].loi_gr,
577 inode->i_ino, inode->i_generation, inode);
578 return -ELDLM_NO_LOCK_DATA;
584 /* Flush the page cache for an extent as its canceled. When we're on an LOV,
585 * we get a lock cancellation for each stripe, so we have to map the obd's
586 * region back onto the stripes in the file that it held.
588 * No one can dirty the extent until we've finished our work and they can
589 * enqueue another lock. The DLM protects us from ll_file_read/write here,
590 * but other kernel actors could have pages locked.
592 * Called with the DLM lock held. */
593 void ll_pgcache_remove_extent(struct inode *inode, struct lov_stripe_md *lsm,
594 struct ldlm_lock *lock, __u32 stripe)
596 ldlm_policy_data_t tmpex;
597 unsigned long start, end, count, skip, i, j;
599 int rc, rc2, discard = lock->l_flags & LDLM_FL_DISCARD_DATA;
600 struct lustre_handle lockh;
603 memcpy(&tmpex, &lock->l_policy_data, sizeof(tmpex));
604 CDEBUG(D_INODE|D_PAGE, "inode %lu(%p) ["LPU64"->"LPU64"] size: %llu\n",
605 inode->i_ino, inode, tmpex.l_extent.start, tmpex.l_extent.end,
608 /* our locks are page granular thanks to osc_enqueue, we invalidate the
610 LASSERT((tmpex.l_extent.start & ~PAGE_CACHE_MASK) == 0);
611 LASSERT(((tmpex.l_extent.end + 1) & ~PAGE_CACHE_MASK) == 0);
615 start = tmpex.l_extent.start >> PAGE_CACHE_SHIFT;
616 end = tmpex.l_extent.end >> PAGE_CACHE_SHIFT;
617 if (lsm->lsm_stripe_count > 1) {
618 count = lsm->lsm_stripe_size >> PAGE_CACHE_SHIFT;
619 skip = (lsm->lsm_stripe_count - 1) * count;
620 start += start/count * skip + stripe * count;
622 end += end/count * skip + stripe * count;
624 if (end < tmpex.l_extent.end >> PAGE_CACHE_SHIFT)
627 i = inode->i_size ? (inode->i_size - 1) >> PAGE_CACHE_SHIFT : 0;
631 CDEBUG(D_INODE|D_PAGE, "walking page indices start: %lu j: %lu "
632 "count: %lu skip: %lu end: %lu%s\n", start, start % count,
633 count, skip, end, discard ? " (DISCARDING)" : "");
635 /* walk through the vmas on the inode and tear down mmaped pages that
636 * intersect with the lock. this stops immediately if there are no
637 * mmap()ed regions of the file. This is not efficient at all and
638 * should be short lived. We'll associate mmap()ed pages with the lock
639 * and will be able to find them directly */
641 for (i = start; i <= end; i += (j + skip)) {
642 j = min(count - (i % count), end - i + 1);
644 LASSERT(inode->i_mapping);
645 if (ll_teardown_mmaps(inode->i_mapping, i << PAGE_CACHE_SHIFT,
646 ((i+j) << PAGE_CACHE_SHIFT) - 1) )
650 /* this is the simplistic implementation of page eviction at
651 * cancelation. It is careful to get races with other page
652 * lockers handled correctly. fixes from bug 20 will make it
653 * more efficient by associating locks with pages and with
654 * batching writeback under the lock explicitly. */
655 for (i = start, j = start % count; i <= end;
656 j++, i++, tmpex.l_extent.start += PAGE_CACHE_SIZE) {
658 CDEBUG(D_PAGE, "skip index %lu to %lu\n", i, i + skip);
664 LASSERTF(tmpex.l_extent.start< lock->l_policy_data.l_extent.end,
665 LPU64" >= "LPU64" start %lu i %lu end %lu\n",
666 tmpex.l_extent.start, lock->l_policy_data.l_extent.end,
669 if (!mapping_has_pages(inode->i_mapping)) {
670 CDEBUG(D_INODE|D_PAGE, "nothing left\n");
676 page = find_get_page(inode->i_mapping, i);
679 LL_CDEBUG_PAGE(D_PAGE, page, "lock page idx %lu ext "LPU64"\n",
680 i, tmpex.l_extent.start);
683 /* page->mapping to check with racing against teardown */
684 if (!discard && clear_page_dirty_for_io(page)) {
685 rc = ll_call_writepage(inode, page);
687 CERROR("writepage of page %p failed: %d\n",
689 /* either waiting for io to complete or reacquiring
690 * the lock that the failed writepage released */
694 tmpex.l_extent.end = tmpex.l_extent.start + PAGE_CACHE_SIZE - 1;
695 /* check to see if another DLM lock covers this page */
696 rc2 = ldlm_lock_match(lock->l_resource->lr_namespace,
697 LDLM_FL_BLOCK_GRANTED|LDLM_FL_CBPENDING |
699 &lock->l_resource->lr_name, LDLM_EXTENT,
700 &tmpex, LCK_PR | LCK_PW, &lockh);
701 if (rc2 == 0 && page->mapping != NULL) {
702 // checking again to account for writeback's lock_page()
703 LL_CDEBUG_PAGE(D_PAGE, page, "truncating\n");
704 ll_ra_accounting(page, inode->i_mapping);
705 ll_truncate_complete_page(page);
708 page_cache_release(page);
710 LASSERTF(tmpex.l_extent.start <=
711 (lock->l_policy_data.l_extent.end == ~0ULL ? ~0ULL :
712 lock->l_policy_data.l_extent.end + 1),
713 "loop too long "LPU64" > "LPU64" start %lu i %lu end %lu\n",
714 tmpex.l_extent.start, lock->l_policy_data.l_extent.end,
719 static int ll_extent_lock_callback(struct ldlm_lock *lock,
720 struct ldlm_lock_desc *new, void *data,
723 struct lustre_handle lockh = { 0 };
727 if ((unsigned long)data > 0 && (unsigned long)data < 0x1000) {
728 LDLM_ERROR(lock, "cancelling lock with bad data %p", data);
733 case LDLM_CB_BLOCKING:
734 ldlm_lock2handle(lock, &lockh);
735 rc = ldlm_cli_cancel(&lockh);
737 CERROR("ldlm_cli_cancel failed: %d\n", rc);
739 case LDLM_CB_CANCELING: {
741 struct ll_inode_info *lli;
742 struct lov_stripe_md *lsm;
746 /* This lock wasn't granted, don't try to evict pages */
747 if (lock->l_req_mode != lock->l_granted_mode)
750 inode = ll_inode_from_lock(lock);
753 lli = ll_i2info(inode);
756 if (lli->lli_smd == NULL)
760 stripe = ll_lock_to_stripe_offset(inode, lock);
763 ll_pgcache_remove_extent(inode, lsm, lock, stripe);
765 l_lock(&lock->l_resource->lr_namespace->ns_lock);
766 down(&lli->lli_size_sem);
767 kms = ldlm_extent_shift_kms(lock,
768 lsm->lsm_oinfo[stripe].loi_kms);
770 if (lsm->lsm_oinfo[stripe].loi_kms != kms)
771 LDLM_DEBUG(lock, "updating kms from "LPU64" to "LPU64,
772 lsm->lsm_oinfo[stripe].loi_kms, kms);
773 lsm->lsm_oinfo[stripe].loi_kms = kms;
774 up(&lli->lli_size_sem);
775 l_unlock(&lock->l_resource->lr_namespace->ns_lock);
776 //ll_try_done_writing(inode);
789 int ll_async_completion_ast(struct ldlm_lock *lock, int flags, void *data)
791 /* XXX ALLOCATE - 160 bytes */
792 struct inode *inode = ll_inode_from_lock(lock);
793 struct ll_inode_info *lli = ll_i2info(inode);
794 struct lustre_handle lockh = { 0 };
799 if (flags & (LDLM_FL_BLOCK_WAIT | LDLM_FL_BLOCK_GRANTED |
800 LDLM_FL_BLOCK_CONV)) {
801 LBUG(); /* not expecting any blocked async locks yet */
802 LDLM_DEBUG(lock, "client-side async enqueue returned a blocked "
804 ldlm_lock_dump(D_OTHER, lock, 0);
805 ldlm_reprocess_all(lock->l_resource);
809 LDLM_DEBUG(lock, "client-side async enqueue: granted/glimpsed");
811 stripe = ll_lock_to_stripe_offset(inode, lock);
815 if (lock->l_lvb_len) {
816 struct lov_stripe_md *lsm = lli->lli_smd;
818 lvb = lock->l_lvb_data;
819 lsm->lsm_oinfo[stripe].loi_rss = lvb->lvb_size;
821 l_lock(&lock->l_resource->lr_namespace->ns_lock);
823 kms = MAX(lsm->lsm_oinfo[stripe].loi_kms, lvb->lvb_size);
824 kms = ldlm_extent_shift_kms(NULL, kms);
825 if (lsm->lsm_oinfo[stripe].loi_kms != kms)
826 LDLM_DEBUG(lock, "updating kms from "LPU64" to "LPU64,
827 lsm->lsm_oinfo[stripe].loi_kms, kms);
828 lsm->lsm_oinfo[stripe].loi_kms = kms;
830 l_unlock(&lock->l_resource->lr_namespace->ns_lock);
835 wake_up(&lock->l_waitq);
837 ldlm_lock2handle(lock, &lockh);
838 ldlm_lock_decref(&lockh, LCK_PR);
843 static int ll_glimpse_callback(struct ldlm_lock *lock, void *reqp)
845 struct ptlrpc_request *req = reqp;
846 struct inode *inode = ll_inode_from_lock(lock);
847 struct ll_inode_info *lli;
849 struct lov_stripe_md *lsm;
850 int rc, size = sizeof(*lvb), stripe;
854 GOTO(out, rc = -ELDLM_NO_LOCK_DATA);
855 lli = ll_i2info(inode);
857 GOTO(iput, rc = -ELDLM_NO_LOCK_DATA);
861 GOTO(iput, rc = -ELDLM_NO_LOCK_DATA);
863 /* First, find out which stripe index this lock corresponds to. */
864 stripe = ll_lock_to_stripe_offset(inode, lock);
866 GOTO(iput, rc = -ELDLM_NO_LOCK_DATA);
868 rc = lustre_pack_reply(req, 1, &size, NULL);
870 CERROR("lustre_pack_reply: %d\n", rc);
874 lvb = lustre_msg_buf(req->rq_repmsg, 0, sizeof(*lvb));
875 lvb->lvb_size = lli->lli_smd->lsm_oinfo[stripe].loi_kms;
876 lvb->lvb_mtime = LTIME_S(inode->i_mtime);
877 lvb->lvb_atime = LTIME_S(inode->i_atime);
878 lvb->lvb_ctime = LTIME_S(inode->i_ctime);
880 LDLM_DEBUG(lock, "i_size: %llu -> stripe number %u -> kms "LPU64,
881 inode->i_size, stripe, lvb->lvb_size);
887 /* These errors are normal races, so we don't want to fill the console
888 * with messages by calling ptlrpc_error() */
889 if (rc == -ELDLM_NO_LOCK_DATA)
890 lustre_pack_reply(req, 0, NULL, NULL);
896 __u64 lov_merge_size(struct lov_stripe_md *lsm, int kms);
897 __u64 lov_merge_blocks(struct lov_stripe_md *lsm);
898 __u64 lov_merge_mtime(struct lov_stripe_md *lsm, __u64 current_time);
900 /* NB: lov_merge_size will prefer locally cached writes if they extend the
901 * file (because it prefers KMS over RSS when larger) */
902 int ll_glimpse_size(struct inode *inode)
904 struct ll_inode_info *lli = ll_i2info(inode);
905 struct ll_sb_info *sbi = ll_i2sbi(inode);
906 ldlm_policy_data_t policy = { .l_extent = { 0, OBD_OBJECT_EOF } };
907 struct lustre_handle lockh = { 0 };
908 int rc, flags = LDLM_FL_HAS_INTENT;
911 CDEBUG(D_DLMTRACE, "Glimpsing inode %lu\n", inode->i_ino);
913 rc = obd_enqueue(sbi->ll_dt_exp, lli->lli_smd, LDLM_EXTENT, &policy,
914 LCK_PR, &flags, ll_extent_lock_callback,
915 ldlm_completion_ast, ll_glimpse_callback, inode,
916 sizeof(struct ost_lvb), lustre_swab_ost_lvb, &lockh);
921 CERROR("obd_enqueue returned rc %d, returning -EIO\n", rc);
922 RETURN(rc > 0 ? -EIO : rc);
925 down(&lli->lli_size_sem);
926 inode->i_size = lov_merge_size(lli->lli_smd, 0);
927 inode->i_blocks = lov_merge_blocks(lli->lli_smd);
928 up(&lli->lli_size_sem);
930 LTIME_S(inode->i_mtime) = lov_merge_mtime(lli->lli_smd,
931 LTIME_S(inode->i_mtime));
933 CDEBUG(D_DLMTRACE, "glimpse: size: "LPU64", blocks: "LPU64"\n",
934 (__u64)inode->i_size, (__u64)inode->i_blocks);
936 obd_cancel(sbi->ll_dt_exp, lli->lli_smd, LCK_PR, &lockh);
940 void ll_stime_record(struct ll_sb_info *sbi, struct timeval *start,
941 struct obd_service_time *stime)
944 do_gettimeofday(&stop);
946 spin_lock(&sbi->ll_lock);
947 lprocfs_stime_record(stime, &stop, start);
948 spin_unlock(&sbi->ll_lock);
951 int ll_extent_lock(struct ll_file_data *fd, struct inode *inode,
952 struct lov_stripe_md *lsm, int mode,
953 ldlm_policy_data_t *policy, struct lustre_handle *lockh,
954 int ast_flags, struct obd_service_time *stime)
956 struct ll_inode_info *lli = ll_i2info(inode);
957 struct ll_sb_info *sbi = ll_i2sbi(inode);
958 struct timeval start;
962 LASSERT(lockh->cookie == 0);
964 /* XXX phil: can we do this? won't it screw the file size up? */
965 if ((fd && (fd->fd_flags & LL_FILE_IGNORE_LOCK)) ||
966 (sbi->ll_flags & LL_SBI_NOLCK))
969 CDEBUG(D_DLMTRACE, "Locking inode %lu, start "LPU64" end "LPU64"\n",
970 inode->i_ino, policy->l_extent.start, policy->l_extent.end);
972 do_gettimeofday(&start);
973 rc = obd_enqueue(sbi->ll_dt_exp, lsm, LDLM_EXTENT, policy, mode,
974 &ast_flags, ll_extent_lock_callback,
975 ldlm_completion_ast, ll_glimpse_callback, inode,
976 sizeof(struct ost_lvb), lustre_swab_ost_lvb, lockh);
980 ll_stime_record(sbi, &start, stime);
982 if (policy->l_extent.start == 0 &&
983 policy->l_extent.end == OBD_OBJECT_EOF) {
984 /* vmtruncate()->ll_truncate() first sets the i_size and then
985 * the kms under both a DLM lock and the i_sem. If we don't
986 * get the i_sem here we can match the DLM lock and reset
987 * i_size from the kms before the truncating path has updated
988 * the kms. generic_file_write can then trust the stale i_size
989 * when doing appending writes and effectively cancel the
990 * result of the truncate. Getting the i_sem after the enqueue
991 * maintains the DLM -> i_sem acquiry order. */
992 down(&lli->lli_size_sem);
993 inode->i_size = lov_merge_size(lsm, 1);
994 up(&lli->lli_size_sem);
998 LTIME_S(inode->i_mtime) =
999 lov_merge_mtime(lsm, LTIME_S(inode->i_mtime));
1005 int ll_extent_unlock(struct ll_file_data *fd, struct inode *inode,
1006 struct lov_stripe_md *lsm, int mode,
1007 struct lustre_handle *lockh)
1009 struct ll_sb_info *sbi = ll_i2sbi(inode);
1013 /* XXX phil: can we do this? won't it screw the file size up? */
1014 if ((fd && (fd->fd_flags & LL_FILE_IGNORE_LOCK)) ||
1015 (sbi->ll_flags & LL_SBI_NOLCK))
1018 rc = obd_cancel(sbi->ll_dt_exp, lsm, mode, lockh);
1023 static ssize_t ll_file_read(struct file *file, char *buf, size_t count,
1026 struct inode *inode = file->f_dentry->d_inode;
1027 struct ll_inode_info *lli = ll_i2info(inode);
1028 struct lov_stripe_md *lsm = lli->lli_smd;
1029 struct ll_lock_tree tree;
1030 struct ll_lock_tree_node *node;
1035 CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p),size="LPSZ",offset=%Ld\n",
1036 inode->i_ino, inode->i_generation, inode, count, *ppos);
1038 /* "If nbyte is 0, read() will return 0 and have no other results."
1039 * -- Single Unix Spec */
1043 lprocfs_counter_add(ll_i2sbi(inode)->ll_stats, LPROC_LL_READ_BYTES,
1049 node = ll_node_from_inode(inode, *ppos, *ppos + count - 1,
1052 tree.lt_fd = file->private_data;
1054 rc = ll_tree_lock(&tree, node, inode, buf, count,
1055 file->f_flags & O_NONBLOCK ? LDLM_FL_BLOCK_NOWAIT :0);
1059 down(&lli->lli_size_sem);
1060 kms = lov_merge_size(lsm, 1);
1061 if (*ppos + count - 1 > kms) {
1062 /* A glimpse is necessary to determine whether we return a short
1063 * read or some zeroes at the end of the buffer */
1064 up(&lli->lli_size_sem);
1065 retval = ll_glimpse_size(inode);
1069 inode->i_size = kms;
1070 up(&lli->lli_size_sem);
1073 CDEBUG(D_INFO, "Read ino %lu, "LPSZ" bytes, offset %lld, i_size %llu\n",
1074 inode->i_ino, count, *ppos, inode->i_size);
1076 /* turn off the kernel's read-ahead */
1077 #if (LINUX_VERSION_CODE < KERNEL_VERSION(2,5,0))
1080 file->f_ra.ra_pages = 0;
1082 retval = generic_file_read(file, buf, count, ppos);
1085 ll_tree_unlock(&tree, inode);
1090 * Write to a file (through the page cache).
1092 static ssize_t ll_file_write(struct file *file, const char *buf,
1093 size_t count, loff_t *ppos)
1095 struct inode *inode = file->f_dentry->d_inode;
1096 loff_t maxbytes = ll_file_maxbytes(inode);
1097 struct ll_lock_tree tree;
1098 struct ll_lock_tree_node *node;
1103 CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p),size="LPSZ",offset=%Ld\n",
1104 inode->i_ino, inode->i_generation, inode, count, *ppos);
1106 SIGNAL_MASK_ASSERT(); /* XXX BUG 1511 */
1108 /* POSIX, but surprised the VFS doesn't check this already */
1112 /* If file was opened for LL_IOC_LOV_SETSTRIPE but the ioctl wasn't
1113 * called on the file, don't fail the below assertion (bug 2388). */
1114 if (file->f_flags & O_LOV_DELAY_CREATE &&
1115 ll_i2info(inode)->lli_smd == NULL)
1118 LASSERT(ll_i2info(inode)->lli_smd != NULL);
1120 if (file->f_flags & O_APPEND)
1121 node = ll_node_from_inode(inode, 0, OBD_OBJECT_EOF, LCK_PW);
1123 node = ll_node_from_inode(inode, *ppos, *ppos + count - 1,
1127 RETURN(PTR_ERR(node));
1129 tree.lt_fd = file->private_data;
1131 rc = ll_tree_lock(&tree, node, inode, buf, count,
1132 file->f_flags & O_NONBLOCK ? LDLM_FL_BLOCK_NOWAIT :0);
1136 /* this is ok, g_f_w will overwrite this under i_sem if it races
1137 * with a local truncate, it just makes our maxbyte checking easier */
1138 if (file->f_flags & O_APPEND)
1139 *ppos = inode->i_size;
1141 if (*ppos >= maxbytes) {
1142 if (count || *ppos > maxbytes) {
1143 send_sig(SIGXFSZ, current, 0);
1144 GOTO(out, retval = -EFBIG);
1147 if (*ppos + count > maxbytes)
1148 count = maxbytes - *ppos;
1150 CDEBUG(D_INFO, "Writing inode %lu, "LPSZ" bytes, offset %Lu\n",
1151 inode->i_ino, count, *ppos);
1153 /* generic_file_write handles O_APPEND after getting i_sem */
1154 retval = generic_file_write(file, buf, count, ppos);
1157 ll_tree_unlock(&tree, inode);
1158 /* serialize with mmap/munmap/mremap */
1159 lprocfs_counter_add(ll_i2sbi(inode)->ll_stats, LPROC_LL_WRITE_BYTES,
1160 retval > 0 ? retval : 0);
1164 static int ll_lov_recreate_obj(struct inode *inode, struct file *file,
1167 struct ll_inode_info *lli = ll_i2info(inode);
1168 struct obd_export *exp = ll_i2dtexp(inode);
1169 struct ll_recreate_obj ucreatp;
1170 struct obd_trans_info oti = { 0 };
1171 struct obdo *oa = NULL;
1174 struct lov_stripe_md *lsm, *lsm2;
1177 if (!capable (CAP_SYS_ADMIN))
1180 rc = copy_from_user(&ucreatp, (struct ll_recreate_obj *)arg,
1181 sizeof(struct ll_recreate_obj));
1189 down(&lli->lli_open_sem);
1192 GOTO(out, rc = -ENOENT);
1193 lsm_size = sizeof(*lsm) + (sizeof(struct lov_oinfo) *
1194 (lsm->lsm_stripe_count));
1196 OBD_ALLOC(lsm2, lsm_size);
1198 GOTO(out, rc = -ENOMEM);
1200 oa->o_id = ucreatp.lrc_id;
1201 oa->o_nlink = ucreatp.lrc_ost_idx;
1202 oa->o_gr = ucreatp.lrc_group;
1203 oa->o_valid = OBD_MD_FLID | OBD_MD_FLGROUP | OBD_MD_FLFLAGS;
1204 oa->o_flags |= OBD_FL_RECREATE_OBJS;
1205 obdo_from_inode(oa, inode, OBD_MD_FLTYPE | OBD_MD_FLATIME |
1206 OBD_MD_FLMTIME | OBD_MD_FLCTIME);
1208 oti.oti_objid = NULL;
1209 memcpy(lsm2, lsm, lsm_size);
1210 rc = obd_create(exp, oa, &lsm2, &oti);
1212 OBD_FREE(lsm2, lsm_size);
1215 up(&lli->lli_open_sem);
1220 static int ll_lov_setstripe_ea_info(struct inode *inode, struct file *file,
1221 int flags, struct lov_user_md *lum,
1224 struct ll_inode_info *lli = ll_i2info(inode);
1226 struct obd_export *exp = ll_i2dtexp(inode);
1227 struct lov_stripe_md *lsm;
1228 struct lookup_intent oit = {.it_op = IT_OPEN, .it_flags = flags};
1229 struct ptlrpc_request *req = NULL;
1231 struct lustre_md md;
1232 struct obd_client_handle *och;
1236 if ((file->f_flags+1) & O_ACCMODE)
1238 if (file->f_flags & O_TRUNC)
1241 down(&lli->lli_open_sem);
1244 up(&lli->lli_open_sem);
1245 CDEBUG(D_IOCTL, "stripe already exists for ino %lu\n",
1250 f = get_empty_filp();
1254 f->f_dentry = file->f_dentry;
1255 f->f_vfsmnt = file->f_vfsmnt;
1258 rc = ll_intent_alloc(&oit);
1262 rc = ll_intent_file_open(f, lum, lum_size, &oit);
1265 if (it_disposition(&oit, DISP_LOOKUP_NEG))
1268 req = LUSTRE_IT(&oit)->it_data;
1269 rc = LUSTRE_IT(&oit)->it_status;
1274 rc = mdc_req2lustre_md(ll_i2mdexp(inode), req, 1, exp, &md);
1277 ll_update_inode(f->f_dentry->d_inode, &md);
1279 OBD_ALLOC(och, sizeof(struct obd_client_handle));
1280 rc = ll_local_open(f, &oit, och);
1281 if (rc) { /* Actually ll_local_open cannot fail! */
1284 if (LUSTRE_IT(&oit)->it_lock_mode) {
1285 ldlm_lock_decref_and_cancel((struct lustre_handle *)
1286 &LUSTRE_IT(&oit)->it_lock_handle,
1287 LUSTRE_IT(&oit)->it_lock_mode);
1288 LUSTRE_IT(&oit)->it_lock_mode = 0;
1291 ll_intent_release(&oit);
1293 /* ll_file_release will decrease the count, but won't free anything
1294 because we have at least one more reference coming from actual open
1296 down(&lli->lli_och_sem);
1297 lli->lli_open_fd_write_count++;
1298 up(&lli->lli_och_sem);
1299 rc = ll_file_release(f->f_dentry->d_inode, f);
1301 /* Now also destroy our supplemental och */
1302 ll_md_och_close(ll_i2mdexp(inode), f->f_dentry->d_inode, och);
1305 ll_intent_release(&oit);
1308 up(&lli->lli_open_sem);
1310 ptlrpc_req_finished(req);
1314 static int ll_lov_setea(struct inode *inode, struct file *file,
1317 int flags = MDS_OPEN_HAS_OBJS | FMODE_WRITE;
1318 struct lov_user_md *lump;
1319 int lum_size = sizeof(struct lov_user_md) +
1320 sizeof(struct lov_user_ost_data);
1324 if (!capable (CAP_SYS_ADMIN))
1327 OBD_ALLOC(lump, lum_size);
1331 rc = copy_from_user(lump, (struct lov_user_md *)arg, lum_size);
1333 OBD_FREE(lump, lum_size);
1337 rc = ll_lov_setstripe_ea_info(inode, file, flags, lump, lum_size);
1339 OBD_FREE(lump, lum_size);
1343 static int ll_lov_setstripe(struct inode *inode, struct file *file,
1346 struct lov_user_md lum, *lump = (struct lov_user_md *)arg;
1348 int flags = FMODE_WRITE;
1351 /* Bug 1152: copy properly when this is no longer true */
1352 LASSERT(sizeof(lum) == sizeof(*lump));
1353 LASSERT(sizeof(lum.lmm_objects[0]) == sizeof(lump->lmm_objects[0]));
1354 rc = copy_from_user(&lum, lump, sizeof(lum));
1358 rc = ll_lov_setstripe_ea_info(inode, file, flags, &lum, sizeof(lum));
1360 put_user(0, &lump->lmm_stripe_count);
1361 rc = obd_iocontrol(LL_IOC_LOV_GETSTRIPE, ll_i2dtexp(inode),
1362 0, ll_i2info(inode)->lli_smd, lump);
1367 static int ll_lov_getstripe(struct inode *inode, unsigned long arg)
1369 struct lov_stripe_md *lsm = ll_i2info(inode)->lli_smd;
1374 return obd_iocontrol(LL_IOC_LOV_GETSTRIPE, ll_i2dtexp(inode), 0, lsm,
1378 static int ll_get_grouplock(struct inode *inode, struct file *file,
1381 struct ll_file_data *fd = file->private_data;
1382 ldlm_policy_data_t policy = { .l_extent = { .start = 0,
1383 .end = OBD_OBJECT_EOF}};
1384 struct lustre_handle lockh = { 0 };
1385 struct ll_inode_info *lli = ll_i2info(inode);
1386 struct lov_stripe_md *lsm = lli->lli_smd;
1390 if (fd->fd_flags & LL_FILE_GROUP_LOCKED) {
1394 policy.l_extent.gid = arg;
1395 if (file->f_flags & O_NONBLOCK)
1396 flags = LDLM_FL_BLOCK_NOWAIT;
1398 rc = ll_extent_lock(fd, inode, lsm, LCK_GROUP, &policy, &lockh, flags,
1399 &ll_i2sbi(inode)->ll_grouplock_stime);
1403 fd->fd_flags |= LL_FILE_GROUP_LOCKED|LL_FILE_IGNORE_LOCK;
1405 memcpy(&fd->fd_cwlockh, &lockh, sizeof(lockh));
1410 static int ll_put_grouplock(struct inode *inode, struct file *file,
1413 struct ll_file_data *fd = file->private_data;
1414 struct ll_inode_info *lli = ll_i2info(inode);
1415 struct lov_stripe_md *lsm = lli->lli_smd;
1419 if (!(fd->fd_flags & LL_FILE_GROUP_LOCKED)) {
1420 /* Ugh, it's already unlocked. */
1424 if (fd->fd_gid != arg) /* Ugh? Unlocking with different gid? */
1427 fd->fd_flags &= ~(LL_FILE_GROUP_LOCKED|LL_FILE_IGNORE_LOCK);
1429 rc = ll_extent_unlock(fd, inode, lsm, LCK_GROUP, &fd->fd_cwlockh);
1434 memset(&fd->fd_cwlockh, 0, sizeof(fd->fd_cwlockh));
1439 int ll_file_ioctl(struct inode *inode, struct file *file, unsigned int cmd,
1442 struct ll_file_data *fd = file->private_data;
1443 struct ll_sb_info *sbi = ll_i2sbi(inode);
1447 CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p),cmd=%x\n", inode->i_ino,
1448 inode->i_generation, inode, cmd);
1450 if (_IOC_TYPE(cmd) == 'T') /* tty ioctls */
1453 lprocfs_counter_incr(ll_i2sbi(inode)->ll_stats, LPROC_LL_IOCTL);
1455 case LL_IOC_GETFLAGS:
1456 /* Get the current value of the file flags */
1457 return put_user(fd->fd_flags, (int *)arg);
1458 case LL_IOC_SETFLAGS:
1459 case LL_IOC_CLRFLAGS:
1460 /* Set or clear specific file flags */
1461 /* XXX This probably needs checks to ensure the flags are
1462 * not abused, and to handle any flag side effects.
1464 if (get_user(flags, (int *) arg))
1467 if (cmd == LL_IOC_SETFLAGS)
1468 fd->fd_flags |= flags;
1470 fd->fd_flags &= ~flags;
1472 case LL_IOC_LOV_SETSTRIPE:
1473 RETURN(ll_lov_setstripe(inode, file, arg));
1474 case LL_IOC_LOV_SETEA:
1475 RETURN(ll_lov_setea(inode, file, arg));
1476 case IOC_MDC_SHOWFID: {
1477 struct lustre_id *idp = (struct lustre_id *)arg;
1478 struct lustre_id id;
1482 filename = getname((const char *)arg);
1483 if (IS_ERR(filename))
1484 RETURN(PTR_ERR(filename));
1486 ll_inode2id(&id, inode);
1488 rc = ll_get_fid(sbi->ll_md_exp, &id, filename, &id);
1490 GOTO(out_filename, rc);
1492 rc = copy_to_user(idp, &id, sizeof(*idp));
1494 GOTO(out_filename, rc = -EFAULT);
1501 case LL_IOC_LOV_GETSTRIPE:
1502 RETURN(ll_lov_getstripe(inode, arg));
1503 case LL_IOC_RECREATE_OBJ:
1504 RETURN(ll_lov_recreate_obj(inode, file, arg));
1505 case EXT3_IOC_GETFLAGS:
1506 case EXT3_IOC_SETFLAGS:
1507 RETURN( ll_iocontrol(inode, file, cmd, arg) );
1508 case LL_IOC_GROUP_LOCK:
1509 RETURN(ll_get_grouplock(inode, file, arg));
1510 case LL_IOC_GROUP_UNLOCK:
1511 RETURN(ll_put_grouplock(inode, file, arg));
1512 case EXT3_IOC_GETVERSION_OLD:
1513 case EXT3_IOC_GETVERSION:
1514 return put_user(inode->i_generation, (int *) arg);
1515 /* We need to special case any other ioctls we want to handle,
1516 * to send them to the MDS/OST as appropriate and to properly
1517 * network encode the arg field.
1518 case EXT2_IOC_GETVERSION_OLD:
1519 case EXT2_IOC_GETVERSION_NEW:
1520 case EXT2_IOC_SETVERSION_OLD:
1521 case EXT2_IOC_SETVERSION_NEW:
1522 case EXT3_IOC_SETVERSION_OLD:
1523 case EXT3_IOC_SETVERSION:
1526 RETURN( obd_iocontrol(cmd, ll_i2dtexp(inode), 0, NULL,
1531 loff_t ll_file_seek(struct file *file, loff_t offset, int origin)
1533 struct inode *inode = file->f_dentry->d_inode;
1534 struct ll_file_data *fd = file->private_data;
1535 struct lov_stripe_md *lsm = ll_i2info(inode)->lli_smd;
1536 struct lustre_handle lockh = {0};
1539 CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p),to=%llu\n", inode->i_ino,
1540 inode->i_generation, inode,
1541 offset + ((origin==2) ? inode->i_size : file->f_pos));
1543 lprocfs_counter_incr(ll_i2sbi(inode)->ll_stats, LPROC_LL_LLSEEK);
1544 if (origin == 2) { /* SEEK_END */
1545 ldlm_policy_data_t policy = { .l_extent = {0, OBD_OBJECT_EOF }};
1546 struct ll_inode_info *lli = ll_i2info(inode);
1547 int nonblock = 0, rc;
1549 if (file->f_flags & O_NONBLOCK)
1550 nonblock = LDLM_FL_BLOCK_NOWAIT;
1552 rc = ll_extent_lock(fd, inode, lsm, LCK_PR, &policy, &lockh,
1553 nonblock, &ll_i2sbi(inode)->ll_seek_stime);
1557 down(&lli->lli_size_sem);
1558 offset += inode->i_size;
1559 up(&lli->lli_size_sem);
1560 } else if (origin == 1) { /* SEEK_CUR */
1561 offset += file->f_pos;
1565 if (offset >= 0 && offset <= ll_file_maxbytes(inode)) {
1566 if (offset != file->f_pos) {
1567 file->f_pos = offset;
1568 #if (LINUX_VERSION_CODE < KERNEL_VERSION(2,5,0))
1570 file->f_version = ++event;
1577 ll_extent_unlock(fd, inode, lsm, LCK_PR, &lockh);
1581 int ll_fsync(struct file *file, struct dentry *dentry, int data)
1583 struct inode *inode = dentry->d_inode;
1584 struct lov_stripe_md *lsm = ll_i2info(inode)->lli_smd;
1585 struct lustre_id id;
1586 struct ptlrpc_request *req;
1589 CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p)\n", inode->i_ino,
1590 inode->i_generation, inode);
1592 lprocfs_counter_incr(ll_i2sbi(inode)->ll_stats, LPROC_LL_FSYNC);
1594 /* fsync's caller has already called _fdata{sync,write}, we want
1595 * that IO to finish before calling the osc and mdc sync methods */
1596 rc = filemap_fdatawait(inode->i_mapping);
1598 ll_inode2id(&id, inode);
1599 err = md_sync(ll_i2sbi(inode)->ll_md_exp, &id, &req);
1603 ptlrpc_req_finished(req);
1606 struct obdo *oa = obdo_alloc();
1609 RETURN(rc ? rc : -ENOMEM);
1611 oa->o_id = lsm->lsm_object_id;
1612 oa->o_gr = lsm->lsm_object_gr;
1613 oa->o_valid = OBD_MD_FLID;
1614 obdo_from_inode(oa, inode, (OBD_MD_FLTYPE | OBD_MD_FLATIME |
1615 OBD_MD_FLMTIME | OBD_MD_FLCTIME |
1618 err = obd_sync(ll_i2sbi(inode)->ll_dt_exp, oa, lsm,
1628 int ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock)
1630 struct inode *inode = file->f_dentry->d_inode;
1631 struct ll_inode_info *li = ll_i2info(inode);
1632 struct ll_sb_info *sbi = ll_i2sbi(inode);
1633 struct obd_device *obddev;
1634 struct ldlm_res_id res_id =
1635 { .name = {id_fid(&li->lli_id), id_group(&li->lli_id), LDLM_FLOCK} };
1636 struct lustre_handle lockh = {0};
1637 ldlm_policy_data_t flock;
1638 ldlm_mode_t mode = 0;
1643 CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu file_lock=%p\n",
1644 inode->i_ino, file_lock);
1646 flock.l_flock.pid = file_lock->fl_pid;
1647 flock.l_flock.start = file_lock->fl_start;
1648 flock.l_flock.end = file_lock->fl_end;
1650 switch (file_lock->fl_type) {
1655 /* An unlock request may or may not have any relation to
1656 * existing locks so we may not be able to pass a lock handle
1657 * via a normal ldlm_lock_cancel() request. The request may even
1658 * unlock a byte range in the middle of an existing lock. In
1659 * order to process an unlock request we need all of the same
1660 * information that is given with a normal read or write record
1661 * lock request. To avoid creating another ldlm unlock (cancel)
1662 * message we'll treat a LCK_NL flock request as an unlock. */
1669 CERROR("unknown fcntl lock type: %d\n", file_lock->fl_type);
1684 flags = LDLM_FL_BLOCK_NOWAIT;
1690 flags = LDLM_FL_TEST_LOCK;
1691 /* Save the old mode so that if the mode in the lock changes we
1692 * can decrement the appropriate reader or writer refcount. */
1693 file_lock->fl_type = mode;
1696 CERROR("unknown fcntl lock command: %d\n", cmd);
1700 CDEBUG(D_DLMTRACE, "inode=%lu, pid="LPU64", flags=%#x, mode=%u, "
1701 "start="LPU64", end="LPU64"\n", inode->i_ino, flock.l_flock.pid,
1702 flags, mode, flock.l_flock.start, flock.l_flock.end);
1704 obddev = md_get_real_obd(sbi->ll_md_exp, &li->lli_id);
1705 rc = ldlm_cli_enqueue(obddev->obd_self_export, NULL,
1706 obddev->obd_namespace,
1707 res_id, LDLM_FLOCK, &flock, mode, &flags,
1708 NULL, ldlm_flock_completion_ast, NULL, file_lock,
1709 NULL, 0, NULL, &lockh);
1713 int ll_inode_revalidate_it(struct dentry *dentry)
1715 struct lookup_intent oit = { .it_op = IT_GETATTR };
1716 struct inode *inode = dentry->d_inode;
1717 struct ptlrpc_request *req = NULL;
1718 struct ll_inode_info *lli;
1719 struct lov_stripe_md *lsm;
1720 struct ll_sb_info *sbi;
1721 struct lustre_id id;
1726 CERROR("REPORT THIS LINE TO PETER\n");
1730 sbi = ll_i2sbi(inode);
1732 ll_inode2id(&id, inode);
1733 lli = ll_i2info(inode);
1734 LASSERT(id_fid(&id) != 0);
1736 CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p), name=%s(%p)\n",
1737 inode->i_ino, inode->i_generation, inode, dentry->d_name.name,
1740 #if (LINUX_VERSION_CODE <= KERNEL_VERSION(2,5,0))
1741 lprocfs_counter_incr(ll_i2sbi(inode)->ll_stats, LPROC_LL_REVALIDATE);
1744 rc = ll_intent_alloc(&oit);
1748 rc = md_intent_lock(sbi->ll_md_exp, &id, NULL, 0, NULL, 0, &id,
1749 &oit, 0, &req, ll_mdc_blocking_ast);
1753 rc = revalidate_it_finish(req, 1, &oit, dentry);
1758 ll_lookup_finish_locks(&oit, dentry);
1761 if (lsm == NULL) /* object not yet allocated, don't validate size */
1765 * ll_glimpse_size() will prefer locally cached writes if they extend
1768 rc = ll_glimpse_size(inode);
1771 ll_intent_release(&oit);
1773 ptlrpc_req_finished(req);
1777 #if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0))
1778 int ll_getattr(struct vfsmount *mnt, struct dentry *de, struct kstat *stat)
1781 struct inode *inode = de->d_inode;
1782 struct ll_inode_info *lli = ll_i2info(inode);
1784 res = ll_inode_revalidate_it(de);
1785 lprocfs_counter_incr(ll_i2sbi(inode)->ll_stats, LPROC_LL_GETATTR);
1790 stat->ino = inode->i_ino;
1791 stat->mode = inode->i_mode;
1792 stat->nlink = inode->i_nlink;
1793 stat->uid = inode->i_uid;
1794 stat->gid = inode->i_gid;
1795 stat->atime = inode->i_atime;
1796 stat->mtime = inode->i_mtime;
1797 stat->ctime = inode->i_ctime;
1798 stat->blksize = inode->i_blksize;
1800 down(&lli->lli_size_sem);
1801 stat->size = inode->i_size;
1802 stat->blocks = inode->i_blocks;
1803 up(&lli->lli_size_sem);
1805 stat->rdev = kdev_t_to_nr(inode->i_rdev);
1806 stat->dev = id_group(&ll_i2info(inode)->lli_id);
1812 int ll_setxattr_internal(struct inode *inode, const char *name,
1813 const void *value, size_t size, int flags,
1816 struct ll_sb_info *sbi = ll_i2sbi(inode);
1817 struct ptlrpc_request *request = NULL;
1818 struct mdc_op_data op_data;
1823 CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu\n", inode->i_ino);
1824 lprocfs_counter_incr(ll_i2sbi(inode)->ll_stats, LPROC_LL_SETXATTR);
1826 memset(&attr, 0x0, sizeof(attr));
1827 attr.ia_valid |= valid;
1828 attr.ia_attr_flags = flags;
1830 ll_prepare_mdc_data(&op_data, inode, NULL, NULL, 0, 0);
1832 rc = md_setattr(sbi->ll_md_exp, &op_data, &attr,
1833 (void*) name, strnlen(name, XATTR_NAME_MAX)+1,
1834 (void*) value, size, &request);
1836 CERROR("md_setattr fails: rc = %d\n", rc);
1841 ptlrpc_req_finished(request);
1845 int ll_setxattr(struct dentry *dentry, const char *name, const void *value,
1846 size_t size, int flags)
1849 struct posix_acl *acl;
1850 struct ll_inode_info *lli;
1853 rc = ll_setxattr_internal(dentry->d_inode, name, value, size,
1856 /* update inode's acl info */
1857 if (rc == 0 && strcmp(name, XATTR_NAME_ACL_ACCESS) == 0) {
1859 acl = posix_acl_from_xattr(value, size);
1861 CERROR("convert from xattr to acl error: %ld",
1865 error = posix_acl_valid(acl);
1867 CERROR("acl valid error: %d", error);
1868 posix_acl_release(acl);
1876 lli = ll_i2info(dentry->d_inode);
1877 spin_lock(&lli->lli_lock);
1878 if (lli->lli_acl_access != NULL)
1879 posix_acl_release(lli->lli_acl_access);
1880 lli->lli_acl_access = acl;
1881 spin_unlock(&lli->lli_lock);
1888 int ll_removexattr(struct dentry *dentry, const char *name)
1890 return ll_setxattr_internal(dentry->d_inode, name, NULL, 0, 0,
1895 int ll_getxattr_internal(struct inode *inode, const char *name, int namelen,
1896 void *value, size_t size, __u64 valid)
1898 struct ptlrpc_request *request = NULL;
1899 struct ll_sb_info *sbi = ll_i2sbi(inode);
1900 struct lustre_id id;
1901 struct mds_body *body;
1906 lprocfs_counter_incr(ll_i2sbi(inode)->ll_stats, LPROC_LL_GETXATTR);
1908 ll_inode2id(&id, inode);
1909 rc = md_getattr(sbi->ll_md_exp, &id, valid, name, namelen,
1912 if (rc != -ENODATA && rc != -EOPNOTSUPP)
1913 CERROR("md_getattr fails: rc = %d\n", rc);
1917 body = lustre_msg_buf(request->rq_repmsg, 0, sizeof(*body));
1918 LASSERT(body != NULL);
1919 LASSERT_REPSWABBED(request, 0);
1921 ea_size = body->eadatasize;
1922 LASSERT(ea_size <= request->rq_repmsg->buflens[0]);
1925 GOTO(out, rc = ea_size);
1927 ea_data = lustre_msg_buf(request->rq_repmsg, 1, ea_size);
1928 LASSERT(ea_data != NULL);
1929 LASSERT_REPSWABBED(request, 1);
1932 memcpy(value, ea_data, ea_size);
1935 ptlrpc_req_finished(request);
1939 int ll_getxattr(struct dentry *dentry, const char *name, void *value,
1942 return ll_getxattr_internal(dentry->d_inode, name, strlen(name) + 1,
1943 value, size, OBD_MD_FLEA);
1946 int ll_listxattr(struct dentry *dentry, char *list, size_t size)
1948 return ll_getxattr_internal(dentry->d_inode, NULL, 0, list, size,
1952 int ll_inode_permission(struct inode *inode, int mask, struct nameidata *nd)
1954 struct lookup_intent it = { .it_op = IT_GETATTR };
1955 int mode = inode->i_mode;
1957 struct ll_sb_info *sbi;
1958 struct lustre_id id;
1959 struct ptlrpc_request *req = NULL;
1963 sbi = ll_i2sbi(inode);
1964 ll_inode2id(&id, inode);
1966 /* Nobody gets write access to a read-only fs */
1967 if ((mask & MAY_WRITE) && IS_RDONLY(inode) &&
1968 (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)))
1970 /* Nobody gets write access to an immutable file */
1971 if ((mask & MAY_WRITE) && IS_IMMUTABLE(inode))
1973 if (current->fsuid == inode->i_uid) {
1976 struct ll_inode_info *lli = ll_i2info(inode);
1977 struct posix_acl *acl;
1979 /* The access ACL cannot grant access if the group class
1980 permission bits don't contain all requested permissions. */
1981 if (((mode >> 3) & mask & S_IRWXO) != mask)
1984 if (ll_intent_alloc(&it))
1988 rc = md_intent_lock(sbi->ll_md_exp, &id, NULL, 0, NULL, 0, &id,
1989 &it, 0, &req, ll_mdc_blocking_ast);
1991 ll_intent_free(&it);
1995 rc = revalidate_it_finish(req, 1, &it, &de);
1997 ll_intent_release(&it);
2001 ll_lookup_finish_locks(&it, &de);
2002 ll_intent_free(&it);
2004 spin_lock(&lli->lli_lock);
2005 acl = posix_acl_dup(ll_i2info(inode)->lli_acl_access);
2006 spin_unlock(&lli->lli_lock);
2011 rc = posix_acl_permission(inode, acl, mask);
2012 posix_acl_release(acl);
2014 goto check_capabilities;
2018 if (in_group_p(inode->i_gid))
2021 if ((mode & mask & S_IRWXO) == mask)
2026 /* Allowed to override Discretionary Access Control? */
2027 if (!(mask & MAY_EXEC) ||
2028 (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode))
2029 if (capable(CAP_DAC_OVERRIDE))
2031 /* Read and search granted if capable(CAP_DAC_READ_SEARCH) */
2032 if (capable(CAP_DAC_READ_SEARCH) && ((mask == MAY_READ) ||
2033 (S_ISDIR(inode->i_mode) && !(mask & MAY_WRITE))))
2037 ptlrpc_req_finished(req);
2042 struct file_operations ll_file_operations = {
2043 .read = ll_file_read,
2044 .write = ll_file_write,
2045 .ioctl = ll_file_ioctl,
2046 .open = ll_file_open,
2047 .release = ll_file_release,
2048 .mmap = ll_file_mmap,
2049 .llseek = ll_file_seek,
2050 #if (LINUX_VERSION_CODE >= KERNEL_VERSION(2,5,0))
2051 .sendfile = generic_file_sendfile,
2054 .lock = ll_file_flock
2057 struct inode_operations ll_file_inode_operations = {
2058 .setattr = ll_setattr,
2059 .truncate = ll_truncate,
2060 #if (LINUX_VERSION_CODE > KERNEL_VERSION(2,5,0))
2061 .getattr = ll_getattr,
2063 .revalidate_it = ll_inode_revalidate_it,
2065 .setxattr = ll_setxattr,
2066 .getxattr = ll_getxattr,
2067 .listxattr = ll_listxattr,
2068 .removexattr = ll_removexattr,
2069 .permission = ll_inode_permission,