1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
2 * vim:expandtab:shiftwidth=8:tabstop=8:
4 * linux/fs/obdfilter/filter_io.c
6 * Copyright (c) 2001-2003 Cluster File Systems, Inc.
7 * Author: Peter Braam <braam@clusterfs.com>
8 * Author: Andreas Dilger <adilger@clusterfs.com>
9 * Author: Phil Schwan <phil@clusterfs.com>
11 * This file is part of Lustre, http://www.lustre.org.
13 * Lustre is free software; you can redistribute it and/or
14 * modify it under the terms of version 2 of the GNU General Public
15 * License as published by the Free Software Foundation.
17 * Lustre is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20 * GNU General Public License for more details.
22 * You should have received a copy of the GNU General Public License
23 * along with Lustre; if not, write to the Free Software
24 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
27 #define DEBUG_SUBSYSTEM S_FILTER
29 #include <linux/config.h>
30 #include <linux/module.h>
31 #include <linux/pagemap.h> // XXX kill me soon
32 #include <linux/version.h>
34 #include <linux/obd_class.h>
35 #include <linux/lustre_fsfilt.h>
36 #include "filter_internal.h"
38 static int filter_start_page_read(struct inode *inode, struct niobuf_local *lnb)
40 struct address_space *mapping = inode->i_mapping;
42 unsigned long index = lnb->offset >> PAGE_SHIFT;
45 page = grab_cache_page(mapping, index); /* locked page */
47 return lnb->rc = -ENOMEM;
49 LASSERT(page->mapping == mapping);
53 if (inode->i_size < lnb->offset + lnb->len - 1)
54 lnb->rc = inode->i_size - lnb->offset;
58 if (PageUptodate(page)) {
63 rc = mapping->a_ops->readpage(NULL, page);
65 CERROR("page index %lu, rc = %d\n", index, rc);
67 page_cache_release(page);
74 static int filter_finish_page_read(struct niobuf_local *lnb)
76 if (lnb->page == NULL)
79 if (PageUptodate(lnb->page))
82 wait_on_page(lnb->page);
83 if (!PageUptodate(lnb->page)) {
84 CERROR("page index %lu/offset "LPX64" not uptodate\n",
85 lnb->page->index, lnb->offset);
86 GOTO(err_page, lnb->rc = -EIO);
88 if (PageError(lnb->page)) {
89 CERROR("page index %lu/offset "LPX64" has error\n",
90 lnb->page->index, lnb->offset);
91 GOTO(err_page, lnb->rc = -EIO);
97 page_cache_release(lnb->page);
102 static int filter_preprw_read(int cmd, struct obd_export *exp, struct obdo *oa,
103 int objcount, struct obd_ioobj *obj,
104 int niocount, struct niobuf_remote *nb,
105 struct niobuf_local *res,
106 struct obd_trans_info *oti)
108 struct obd_run_ctxt saved;
110 struct niobuf_remote *rnb;
111 struct niobuf_local *lnb = NULL;
112 struct fsfilt_objinfo *fso;
113 struct dentry *dentry;
115 int rc = 0, i, j, tot_bytes = 0, cleanup_phase = 0;
116 unsigned long now = jiffies;
119 /* We are currently not supporting multi-obj BRW_READ RPCS at all.
120 * When we do this function's dentry cleanup will need to be fixed */
121 LASSERT(objcount == 1);
123 OBD_ALLOC(fso, objcount * sizeof(*fso));
127 memset(res, 0, niocount * sizeof(*res));
129 push_ctxt(&saved, &exp->exp_obd->obd_ctxt, NULL);
130 for (i = 0, o = obj; i < objcount; i++, o++) {
131 LASSERT(o->ioo_bufcnt);
133 dentry = filter_oa2dentry(exp->exp_obd, oa);
135 GOTO(cleanup, rc = PTR_ERR(dentry));
137 if (dentry->d_inode == NULL) {
138 CERROR("trying to BRW to non-existent file "LPU64"\n",
141 GOTO(cleanup, rc = -ENOENT);
144 fso[i].fso_dentry = dentry;
145 fso[i].fso_bufcnt = o->ioo_bufcnt;
148 if (time_after(jiffies, now + 15 * HZ))
149 CERROR("slow preprw_read setup %lus\n", (jiffies - now) / HZ);
151 CDEBUG(D_INFO, "preprw_read setup: %lu jiffies\n",
154 for (i = 0, o = obj, rnb = nb, lnb = res; i < objcount; i++, o++) {
155 dentry = fso[i].fso_dentry;
156 inode = dentry->d_inode;
158 for (j = 0; j < o->ioo_bufcnt; j++, rnb++, lnb++) {
159 lnb->dentry = dentry;
160 lnb->offset = rnb->offset;
162 lnb->flags = rnb->flags;
163 lnb->start = jiffies;
165 if (inode->i_size <= rnb->offset) {
166 /* If there's no more data, abort early.
167 * lnb->page == NULL and lnb->rc == 0, so it's
168 * easy to detect later. */
171 rc = filter_start_page_read(inode, lnb);
175 CDEBUG(rc == -ENOSPC ? D_INODE : D_ERROR,
176 "page err %u@"LPU64" %u/%u %p: rc %d\n",
177 lnb->len, lnb->offset, j, o->ioo_bufcnt,
183 tot_bytes += lnb->rc;
184 if (lnb->rc < lnb->len) {
185 /* short read, be sure to wait on it */
192 if (time_after(jiffies, now + 15 * HZ))
193 CERROR("slow start_page_read %lus\n", (jiffies - now) / HZ);
195 CDEBUG(D_INFO, "start_page_read: %lu jiffies\n",
198 lprocfs_counter_add(exp->exp_obd->obd_stats, LPROC_FILTER_READ_BYTES,
200 while (lnb-- > res) {
201 rc = filter_finish_page_read(lnb);
203 CERROR("error page %u@"LPU64" %u %p: rc %d\n", lnb->len,
204 lnb->offset, (int)(lnb - res), lnb->dentry, rc);
210 if (time_after(jiffies, now + 15 * HZ))
211 CERROR("slow finish_page_read %lus\n", (jiffies - now) / HZ);
213 CDEBUG(D_INFO, "finish_page_read: %lu jiffies\n",
216 filter_tally_read(&exp->exp_obd->u.filter, res, niocount);
221 switch (cleanup_phase) {
223 for (lnb = res; lnb < (res + niocount); lnb++) {
225 page_cache_release(lnb->page);
227 if (res->dentry != NULL)
230 CERROR("NULL dentry in cleanup -- tell CFS\n");
232 OBD_FREE(fso, objcount * sizeof(*fso));
233 pop_ctxt(&saved, &exp->exp_obd->obd_ctxt, NULL);
238 static int filter_start_page_write(struct inode *inode,
239 struct niobuf_local *lnb)
241 struct page *page = alloc_pages(GFP_HIGHUSER, 0);
243 CERROR("no memory for a temp page\n");
244 RETURN(lnb->rc = -ENOMEM);
246 POISON_PAGE(page, 0xf1);
247 page->index = lnb->offset >> PAGE_SHIFT;
253 /* If we ever start to support multi-object BRW RPCs, we will need to get locks
254 * on mulitple inodes. That isn't all, because there still exists the
255 * possibility of a truncate starting a new transaction while holding the ext3
256 * rwsem = write while some writes (which have started their transactions here)
257 * blocking on the ext3 rwsem = read => lock inversion.
259 * The handling gets very ugly when dealing with locked pages. It may be easier
260 * to just get rid of the locked page code (which has problems of its own) and
261 * either discover we do not need it anymore (i.e. it was a symptom of another
262 * bug) or ensure we get the page locks in an appropriate order. */
263 static int filter_preprw_write(int cmd, struct obd_export *exp, struct obdo *oa,
264 int objcount, struct obd_ioobj *obj,
265 int niocount, struct niobuf_remote *nb,
266 struct niobuf_local *res,
267 struct obd_trans_info *oti)
269 struct obd_run_ctxt saved;
270 struct niobuf_remote *rnb;
271 struct niobuf_local *lnb = NULL;
272 struct fsfilt_objinfo fso;
273 struct dentry *dentry;
274 int rc = 0, i, tot_bytes = 0;
275 unsigned long now = jiffies;
277 LASSERT(objcount == 1);
278 LASSERT(obj->ioo_bufcnt > 0);
280 memset(res, 0, niocount * sizeof(*res));
282 push_ctxt(&saved, &exp->exp_obd->obd_ctxt, NULL);
283 dentry = filter_fid2dentry(exp->exp_obd, NULL, obj->ioo_gr,
286 GOTO(cleanup, rc = PTR_ERR(dentry));
288 if (dentry->d_inode == NULL) {
289 CERROR("trying to BRW to non-existent file "LPU64"\n",
292 GOTO(cleanup, rc = -ENOENT);
295 fso.fso_dentry = dentry;
296 fso.fso_bufcnt = obj->ioo_bufcnt;
298 if (time_after(jiffies, now + 15 * HZ))
299 CERROR("slow preprw_write setup %lus\n", (jiffies - now) / HZ);
301 CDEBUG(D_INFO, "preprw_write setup: %lu jiffies\n",
304 for (i = 0, rnb = nb, lnb = res; i < obj->ioo_bufcnt;
306 lnb->dentry = dentry;
307 lnb->offset = rnb->offset;
309 lnb->flags = rnb->flags;
310 lnb->start = jiffies;
312 rc = filter_start_page_write(dentry->d_inode, lnb);
314 CDEBUG(rc == -ENOSPC ? D_INODE : D_ERROR, "page err %u@"
315 LPU64" %u/%u %p: rc %d\n", lnb->len, lnb->offset,
316 i, obj->ioo_bufcnt, dentry, rc);
318 __free_pages(lnb->page, 0);
322 tot_bytes += lnb->len;
325 if (time_after(jiffies, now + 15 * HZ))
326 CERROR("slow start_page_write %lus\n", (jiffies - now) / HZ);
328 CDEBUG(D_INFO, "start_page_write: %lu jiffies\n",
331 lprocfs_counter_add(exp->exp_obd->obd_stats, LPROC_FILTER_WRITE_BYTES,
335 pop_ctxt(&saved, &exp->exp_obd->obd_ctxt, NULL);
339 int filter_preprw(int cmd, struct obd_export *exp, struct obdo *oa,
340 int objcount, struct obd_ioobj *obj, int niocount,
341 struct niobuf_remote *nb, struct niobuf_local *res,
342 struct obd_trans_info *oti)
344 if (cmd == OBD_BRW_WRITE)
345 return filter_preprw_write(cmd, exp, oa, objcount, obj,
346 niocount, nb, res, oti);
348 if (cmd == OBD_BRW_READ)
349 return filter_preprw_read(cmd, exp, oa, objcount, obj,
350 niocount, nb, res, oti);
356 static int filter_commitrw_read(struct obd_export *exp, struct obdo *oa,
357 int objcount, struct obd_ioobj *obj,
358 int niocount, struct niobuf_local *res,
359 struct obd_trans_info *oti)
362 struct niobuf_local *lnb;
366 if (res->dentry != NULL)
367 drop = (res->dentry->d_inode->i_size >
368 exp->exp_obd->u.filter.fo_readcache_max_filesize);
370 for (i = 0, o = obj, lnb = res; i < objcount; i++, o++) {
371 for (j = 0 ; j < o->ioo_bufcnt ; j++, lnb++) {
372 if (lnb->page == NULL)
374 /* drop from cache like truncate_list_pages() */
375 if (drop && !TryLockPage(lnb->page)) {
376 if (lnb->page->mapping)
377 truncate_complete_page(lnb->page);
378 unlock_page(lnb->page);
380 page_cache_release(lnb->page);
383 if (res->dentry != NULL)
388 void flip_into_page_cache(struct inode *inode, struct page *new_page)
390 struct page *old_page;
394 /* the dlm is protecting us from read/write concurrency, so we
395 * expect this find_lock_page to return quickly. even if we
396 * race with another writer it won't be doing much work with
397 * the page locked. we do this 'cause t_c_p expects a
398 * locked page, and it wants to grab the pagecache lock
400 old_page = find_lock_page(inode->i_mapping, new_page->index);
402 #if (LINUX_VERSION_CODE < KERNEL_VERSION(2,5,0))
403 truncate_complete_page(old_page);
405 truncate_complete_page(old_page->mapping, old_page);
407 unlock_page(old_page);
408 page_cache_release(old_page);
411 #if 0 /* this should be a /proc tunable someday */
412 /* racing o_directs (no locking ioctl) could race adding
413 * their pages, so we repeat the page invalidation unless
414 * we successfully added our new page */
415 rc = add_to_page_cache_unique(new_page, inode->i_mapping,
417 page_hash(inode->i_mapping,
420 /* add_to_page_cache clears uptodate|dirty and locks
422 SetPageUptodate(new_page);
423 unlock_page(new_page);
431 /* XXX needs to trickle its oa down */
432 int filter_commitrw(int cmd, struct obd_export *exp, struct obdo *oa,
433 int objcount, struct obd_ioobj *obj, int niocount,
434 struct niobuf_local *res, struct obd_trans_info *oti)
436 if (cmd == OBD_BRW_WRITE)
437 return filter_commitrw_write(exp, oa, objcount, obj, niocount,
439 if (cmd == OBD_BRW_READ)
440 return filter_commitrw_read(exp, oa, objcount, obj, niocount,
446 int filter_brw(int cmd, struct obd_export *exp, struct obdo *oa,
447 struct lov_stripe_md *lsm, obd_count oa_bufs,
448 struct brw_page *pga, struct obd_trans_info *oti)
450 struct obd_ioobj ioo;
451 struct niobuf_local *lnb;
452 struct niobuf_remote *rnb;
457 OBD_ALLOC(lnb, oa_bufs * sizeof(struct niobuf_local));
458 OBD_ALLOC(rnb, oa_bufs * sizeof(struct niobuf_remote));
460 if (lnb == NULL || rnb == NULL)
461 GOTO(out, ret = -ENOMEM);
463 for (i = 0; i < oa_bufs; i++) {
464 rnb[i].offset = pga[i].off;
465 rnb[i].len = pga[i].count;
468 obdo_to_ioobj(oa, &ioo);
469 ioo.ioo_bufcnt = oa_bufs;
471 ret = filter_preprw(cmd, exp, oa, 1, &ioo, oa_bufs, rnb, lnb, oti);
475 for (i = 0; i < oa_bufs; i++) {
476 void *virt = kmap(pga[i].pg);
477 obd_off off = pga[i].off & ~PAGE_MASK;
478 void *addr = kmap(lnb[i].page);
480 /* 2 kmaps == vanishingly small deadlock opportunity */
482 if (cmd & OBD_BRW_WRITE)
483 memcpy(addr + off, virt + off, pga[i].count);
485 memcpy(virt + off, addr + off, pga[i].count);
491 ret = filter_commitrw(cmd, exp, oa, 1, &ioo, oa_bufs, lnb, oti);
495 OBD_FREE(lnb, oa_bufs * sizeof(struct niobuf_local));
497 OBD_FREE(rnb, oa_bufs * sizeof(struct niobuf_remote));