blob: 4144caf2f9d3a5ef95a7a9162b1de0cfa326cf4e [file] [log] [blame]
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -07001#include <linux/ceph/ceph_debug.h>
Sage Weil1d3576f2009-10-06 11:31:09 -07002
3#include <linux/backing-dev.h>
4#include <linux/fs.h>
5#include <linux/mm.h>
6#include <linux/pagemap.h>
7#include <linux/writeback.h> /* generic_writepages */
Tejun Heo5a0e3ad2010-03-24 17:04:11 +09008#include <linux/slab.h>
Sage Weil1d3576f2009-10-06 11:31:09 -07009#include <linux/pagevec.h>
10#include <linux/task_io_accounting_ops.h>
11
12#include "super.h"
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -070013#include "mds_client.h"
14#include <linux/ceph/osd_client.h>
Sage Weil1d3576f2009-10-06 11:31:09 -070015
16/*
17 * Ceph address space ops.
18 *
19 * There are a few funny things going on here.
20 *
21 * The page->private field is used to reference a struct
22 * ceph_snap_context for _every_ dirty page. This indicates which
23 * snapshot the page was logically dirtied in, and thus which snap
24 * context needs to be associated with the osd write during writeback.
25 *
26 * Similarly, struct ceph_inode_info maintains a set of counters to
Lucas De Marchi25985ed2011-03-30 22:57:33 -030027 * count dirty pages on the inode. In the absence of snapshots,
Sage Weil1d3576f2009-10-06 11:31:09 -070028 * i_wrbuffer_ref == i_wrbuffer_ref_head == the dirty page count.
29 *
30 * When a snapshot is taken (that is, when the client receives
31 * notification that a snapshot was taken), each inode with caps and
32 * with dirty pages (dirty pages implies there is a cap) gets a new
33 * ceph_cap_snap in the i_cap_snaps list (which is sorted in ascending
34 * order, new snaps go to the tail). The i_wrbuffer_ref_head count is
35 * moved to capsnap->dirty. (Unless a sync write is currently in
36 * progress. In that case, the capsnap is said to be "pending", new
37 * writes cannot start, and the capsnap isn't "finalized" until the
38 * write completes (or fails) and a final size/mtime for the inode for
39 * that snap can be settled upon.) i_wrbuffer_ref_head is reset to 0.
40 *
41 * On writeback, we must submit writes to the osd IN SNAP ORDER. So,
42 * we look for the first capsnap in i_cap_snaps and write out pages in
43 * that snap context _only_. Then we move on to the next capsnap,
44 * eventually reaching the "live" or "head" context (i.e., pages that
45 * are not yet snapped) and are writing the most recently dirtied
46 * pages.
47 *
48 * Invalidate and so forth must take care to ensure the dirty page
49 * accounting is preserved.
50 */
51
Yehuda Sadeh2baba252009-12-18 13:51:57 -080052#define CONGESTION_ON_THRESH(congestion_kb) (congestion_kb >> (PAGE_SHIFT-10))
53#define CONGESTION_OFF_THRESH(congestion_kb) \
54 (CONGESTION_ON_THRESH(congestion_kb) - \
55 (CONGESTION_ON_THRESH(congestion_kb) >> 2))
56
57
Sage Weil1d3576f2009-10-06 11:31:09 -070058
59/*
60 * Dirty a page. Optimistically adjust accounting, on the assumption
61 * that we won't race with invalidate. If we do, readjust.
62 */
63static int ceph_set_page_dirty(struct page *page)
64{
65 struct address_space *mapping = page->mapping;
66 struct inode *inode;
67 struct ceph_inode_info *ci;
68 int undo = 0;
69 struct ceph_snap_context *snapc;
70
71 if (unlikely(!mapping))
72 return !TestSetPageDirty(page);
73
74 if (TestSetPageDirty(page)) {
75 dout("%p set_page_dirty %p idx %lu -- already dirty\n",
76 mapping->host, page, page->index);
77 return 0;
78 }
79
80 inode = mapping->host;
81 ci = ceph_inode(inode);
82
83 /*
84 * Note that we're grabbing a snapc ref here without holding
85 * any locks!
86 */
87 snapc = ceph_get_snap_context(ci->i_snap_realm->cached_context);
88
89 /* dirty the head */
90 spin_lock(&inode->i_lock);
Sage Weil7d8cb262010-08-24 08:44:16 -070091 if (ci->i_head_snapc == NULL)
Sage Weil1d3576f2009-10-06 11:31:09 -070092 ci->i_head_snapc = ceph_get_snap_context(snapc);
93 ++ci->i_wrbuffer_ref_head;
94 if (ci->i_wrbuffer_ref == 0)
Dave Chinner0444d762011-03-29 18:08:50 +110095 ihold(inode);
Sage Weil1d3576f2009-10-06 11:31:09 -070096 ++ci->i_wrbuffer_ref;
97 dout("%p set_page_dirty %p idx %lu head %d/%d -> %d/%d "
98 "snapc %p seq %lld (%d snaps)\n",
99 mapping->host, page, page->index,
100 ci->i_wrbuffer_ref-1, ci->i_wrbuffer_ref_head-1,
101 ci->i_wrbuffer_ref, ci->i_wrbuffer_ref_head,
102 snapc, snapc->seq, snapc->num_snaps);
103 spin_unlock(&inode->i_lock);
104
105 /* now adjust page */
106 spin_lock_irq(&mapping->tree_lock);
107 if (page->mapping) { /* Race with truncate? */
108 WARN_ON_ONCE(!PageUptodate(page));
Michael Rubin679ceac2010-08-20 02:31:26 -0700109 account_page_dirtied(page, page->mapping);
Sage Weil1d3576f2009-10-06 11:31:09 -0700110 radix_tree_tag_set(&mapping->page_tree,
111 page_index(page), PAGECACHE_TAG_DIRTY);
112
113 /*
114 * Reference snap context in page->private. Also set
115 * PagePrivate so that we get invalidatepage callback.
116 */
117 page->private = (unsigned long)snapc;
118 SetPagePrivate(page);
119 } else {
120 dout("ANON set_page_dirty %p (raced truncate?)\n", page);
121 undo = 1;
122 }
123
124 spin_unlock_irq(&mapping->tree_lock);
125
126 if (undo)
127 /* whoops, we failed to dirty the page */
128 ceph_put_wrbuffer_cap_refs(ci, 1, snapc);
129
130 __mark_inode_dirty(mapping->host, I_DIRTY_PAGES);
131
132 BUG_ON(!PageDirty(page));
133 return 1;
134}
135
136/*
137 * If we are truncating the full page (i.e. offset == 0), adjust the
138 * dirty page counters appropriately. Only called if there is private
139 * data on the page.
140 */
141static void ceph_invalidatepage(struct page *page, unsigned long offset)
142{
Alexander Beregalov4ce1e9a2010-02-22 17:17:44 +0300143 struct inode *inode;
Sage Weil1d3576f2009-10-06 11:31:09 -0700144 struct ceph_inode_info *ci;
145 struct ceph_snap_context *snapc = (void *)page->private;
146
147 BUG_ON(!PageLocked(page));
148 BUG_ON(!page->private);
149 BUG_ON(!PagePrivate(page));
150 BUG_ON(!page->mapping);
151
Alexander Beregalov4ce1e9a2010-02-22 17:17:44 +0300152 inode = page->mapping->host;
153
Sage Weil1d3576f2009-10-06 11:31:09 -0700154 /*
155 * We can get non-dirty pages here due to races between
156 * set_page_dirty and truncate_complete_page; just spit out a
157 * warning, in case we end up with accounting problems later.
158 */
159 if (!PageDirty(page))
160 pr_err("%p invalidatepage %p page not dirty\n", inode, page);
161
162 if (offset == 0)
163 ClearPageChecked(page);
164
165 ci = ceph_inode(inode);
166 if (offset == 0) {
167 dout("%p invalidatepage %p idx %lu full dirty page %lu\n",
168 inode, page, page->index, offset);
169 ceph_put_wrbuffer_cap_refs(ci, 1, snapc);
170 ceph_put_snap_context(snapc);
171 page->private = 0;
172 ClearPagePrivate(page);
173 } else {
174 dout("%p invalidatepage %p idx %lu partial dirty page\n",
175 inode, page, page->index);
176 }
177}
178
179/* just a sanity check */
180static int ceph_releasepage(struct page *page, gfp_t g)
181{
182 struct inode *inode = page->mapping ? page->mapping->host : NULL;
183 dout("%p releasepage %p idx %lu\n", inode, page, page->index);
184 WARN_ON(PageDirty(page));
185 WARN_ON(page->private);
186 WARN_ON(PagePrivate(page));
187 return 0;
188}
189
190/*
191 * read a single page, without unlocking it.
192 */
193static int readpage_nounlock(struct file *filp, struct page *page)
194{
195 struct inode *inode = filp->f_dentry->d_inode;
196 struct ceph_inode_info *ci = ceph_inode(inode);
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700197 struct ceph_osd_client *osdc =
198 &ceph_inode_to_client(inode)->client->osdc;
Sage Weil1d3576f2009-10-06 11:31:09 -0700199 int err = 0;
200 u64 len = PAGE_CACHE_SIZE;
201
202 dout("readpage inode %p file %p page %p index %lu\n",
203 inode, filp, page, page->index);
204 err = ceph_osdc_readpages(osdc, ceph_vino(inode), &ci->i_layout,
205 page->index << PAGE_CACHE_SHIFT, &len,
206 ci->i_truncate_seq, ci->i_truncate_size,
Sage Weilb7495fc2010-11-09 12:43:12 -0800207 &page, 1, 0);
Sage Weil1d3576f2009-10-06 11:31:09 -0700208 if (err == -ENOENT)
209 err = 0;
210 if (err < 0) {
211 SetPageError(page);
212 goto out;
213 } else if (err < PAGE_CACHE_SIZE) {
214 /* zero fill remainder of page */
215 zero_user_segment(page, err, PAGE_CACHE_SIZE);
216 }
217 SetPageUptodate(page);
218
219out:
220 return err < 0 ? err : 0;
221}
222
223static int ceph_readpage(struct file *filp, struct page *page)
224{
225 int r = readpage_nounlock(filp, page);
226 unlock_page(page);
227 return r;
228}
229
230/*
Sage Weil7c272192011-08-03 09:58:09 -0700231 * Finish an async read(ahead) op.
Sage Weil1d3576f2009-10-06 11:31:09 -0700232 */
Sage Weil7c272192011-08-03 09:58:09 -0700233static void finish_read(struct ceph_osd_request *req, struct ceph_msg *msg)
Sage Weil1d3576f2009-10-06 11:31:09 -0700234{
Sage Weil7c272192011-08-03 09:58:09 -0700235 struct inode *inode = req->r_inode;
236 struct ceph_osd_reply_head *replyhead;
237 int rc, bytes;
238 int i;
239
240 /* parse reply */
241 replyhead = msg->front.iov_base;
242 WARN_ON(le32_to_cpu(replyhead->num_ops) == 0);
243 rc = le32_to_cpu(replyhead->result);
244 bytes = le32_to_cpu(msg->hdr.data_len);
245
246 dout("finish_read %p req %p rc %d bytes %d\n", inode, req, rc, bytes);
247
248 /* unlock all pages, zeroing any data we didn't read */
249 for (i = 0; i < req->r_num_pages; i++, bytes -= PAGE_CACHE_SIZE) {
250 struct page *page = req->r_pages[i];
251
252 if (bytes < (int)PAGE_CACHE_SIZE) {
253 /* zero (remainder of) page */
254 int s = bytes < 0 ? 0 : bytes;
255 zero_user_segment(page, s, PAGE_CACHE_SIZE);
256 }
257 dout("finish_read %p uptodate %p idx %lu\n", inode, page,
258 page->index);
259 flush_dcache_page(page);
260 SetPageUptodate(page);
261 unlock_page(page);
262 page_cache_release(page);
263 }
264 kfree(req->r_pages);
265}
266
267/*
268 * start an async read(ahead) operation. return nr_pages we submitted
269 * a read for on success, or negative error code.
270 */
Sage Weil0d66a482011-08-04 08:21:30 -0700271static int start_read(struct inode *inode, struct list_head *page_list, int max)
Sage Weil7c272192011-08-03 09:58:09 -0700272{
273 struct ceph_osd_client *osdc =
274 &ceph_inode_to_client(inode)->client->osdc;
275 struct ceph_inode_info *ci = ceph_inode(inode);
276 struct page *page = list_entry(page_list->prev, struct page, lru);
277 struct ceph_osd_request *req;
278 u64 off;
279 u64 len;
280 int i;
Sage Weil1d3576f2009-10-06 11:31:09 -0700281 struct page **pages;
Sage Weil7c272192011-08-03 09:58:09 -0700282 pgoff_t next_index;
283 int nr_pages = 0;
284 int ret;
285
286 off = page->index << PAGE_CACHE_SHIFT;
287
288 /* count pages */
289 next_index = page->index;
290 list_for_each_entry_reverse(page, page_list, lru) {
291 if (page->index != next_index)
292 break;
293 nr_pages++;
294 next_index++;
Sage Weil0d66a482011-08-04 08:21:30 -0700295 if (max && nr_pages == max)
296 break;
Sage Weil7c272192011-08-03 09:58:09 -0700297 }
298 len = nr_pages << PAGE_CACHE_SHIFT;
299 dout("start_read %p nr_pages %d is %lld~%lld\n", inode, nr_pages,
300 off, len);
301
302 req = ceph_osdc_new_request(osdc, &ci->i_layout, ceph_vino(inode),
303 off, &len,
304 CEPH_OSD_OP_READ, CEPH_OSD_FLAG_READ,
305 NULL, 0,
306 ci->i_truncate_seq, ci->i_truncate_size,
307 NULL, false, 1, 0);
308 if (!req)
309 return -ENOMEM;
Sage Weil1d3576f2009-10-06 11:31:09 -0700310
311 /* build page vector */
Sage Weil7c272192011-08-03 09:58:09 -0700312 nr_pages = len >> PAGE_CACHE_SHIFT;
313 pages = kmalloc(sizeof(*pages) * nr_pages, GFP_NOFS);
314 ret = -ENOMEM;
Sage Weil1d3576f2009-10-06 11:31:09 -0700315 if (!pages)
Sage Weil7c272192011-08-03 09:58:09 -0700316 goto out;
317 for (i = 0; i < nr_pages; ++i) {
318 page = list_entry(page_list->prev, struct page, lru);
319 BUG_ON(PageLocked(page));
320 list_del(&page->lru);
321
322 dout("start_read %p adding %p idx %lu\n", inode, page,
323 page->index);
324 if (add_to_page_cache_lru(page, &inode->i_data, page->index,
325 GFP_NOFS)) {
326 page_cache_release(page);
327 dout("start_read %p add_to_page_cache failed %p\n",
328 inode, page);
329 nr_pages = i;
330 goto out_pages;
Sage Weil1d3576f2009-10-06 11:31:09 -0700331 }
Sage Weil7c272192011-08-03 09:58:09 -0700332 pages[i] = page;
Sage Weil1d3576f2009-10-06 11:31:09 -0700333 }
Sage Weil7c272192011-08-03 09:58:09 -0700334 req->r_pages = pages;
335 req->r_num_pages = nr_pages;
336 req->r_callback = finish_read;
337 req->r_inode = inode;
338
339 dout("start_read %p starting %p %lld~%lld\n", inode, req, off, len);
340 ret = ceph_osdc_start_request(osdc, req, false);
341 if (ret < 0)
342 goto out_pages;
343 ceph_osdc_put_request(req);
344 return nr_pages;
345
346out_pages:
347 ceph_release_page_vector(pages, nr_pages);
Sage Weil7c272192011-08-03 09:58:09 -0700348out:
349 ceph_osdc_put_request(req);
350 return ret;
Sage Weil1d3576f2009-10-06 11:31:09 -0700351}
352
Sage Weil7c272192011-08-03 09:58:09 -0700353
Sage Weil1d3576f2009-10-06 11:31:09 -0700354/*
355 * Read multiple pages. Leave pages we don't read + unlock in page_list;
356 * the caller (VM) cleans them up.
357 */
358static int ceph_readpages(struct file *file, struct address_space *mapping,
359 struct list_head *page_list, unsigned nr_pages)
360{
361 struct inode *inode = file->f_dentry->d_inode;
Sage Weil0d66a482011-08-04 08:21:30 -0700362 struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
Sage Weil1d3576f2009-10-06 11:31:09 -0700363 int rc = 0;
Sage Weil0d66a482011-08-04 08:21:30 -0700364 int max = 0;
Sage Weil1d3576f2009-10-06 11:31:09 -0700365
Sage Weil0d66a482011-08-04 08:21:30 -0700366 if (fsc->mount_options->rsize >= PAGE_CACHE_SIZE)
367 max = (fsc->mount_options->rsize + PAGE_CACHE_SIZE - 1)
368 >> PAGE_SHIFT;
369
370 dout("readpages %p file %p nr_pages %d max %d\n", inode, file, nr_pages,
371 max);
Sage Weil7c272192011-08-03 09:58:09 -0700372 while (!list_empty(page_list)) {
Sage Weil0d66a482011-08-04 08:21:30 -0700373 rc = start_read(inode, page_list, max);
Sage Weil7c272192011-08-03 09:58:09 -0700374 if (rc < 0)
375 goto out;
376 BUG_ON(rc == 0);
Sage Weil1d3576f2009-10-06 11:31:09 -0700377 }
Sage Weil1d3576f2009-10-06 11:31:09 -0700378out:
Sage Weil7c272192011-08-03 09:58:09 -0700379 dout("readpages %p file %p ret %d\n", inode, file, rc);
Sage Weil1d3576f2009-10-06 11:31:09 -0700380 return rc;
381}
382
383/*
384 * Get ref for the oldest snapc for an inode with dirty data... that is, the
385 * only snap context we are allowed to write back.
Sage Weil1d3576f2009-10-06 11:31:09 -0700386 */
Sage Weil6298a332010-03-31 22:01:38 -0700387static struct ceph_snap_context *get_oldest_context(struct inode *inode,
388 u64 *snap_size)
Sage Weil1d3576f2009-10-06 11:31:09 -0700389{
390 struct ceph_inode_info *ci = ceph_inode(inode);
391 struct ceph_snap_context *snapc = NULL;
392 struct ceph_cap_snap *capsnap = NULL;
393
Sage Weil6298a332010-03-31 22:01:38 -0700394 spin_lock(&inode->i_lock);
Sage Weil1d3576f2009-10-06 11:31:09 -0700395 list_for_each_entry(capsnap, &ci->i_cap_snaps, ci_item) {
396 dout(" cap_snap %p snapc %p has %d dirty pages\n", capsnap,
397 capsnap->context, capsnap->dirty_pages);
398 if (capsnap->dirty_pages) {
399 snapc = ceph_get_snap_context(capsnap->context);
400 if (snap_size)
401 *snap_size = capsnap->size;
402 break;
403 }
404 }
Sage Weil7d8cb262010-08-24 08:44:16 -0700405 if (!snapc && ci->i_wrbuffer_ref_head) {
Sage Weil80e755f2010-03-31 21:52:10 -0700406 snapc = ceph_get_snap_context(ci->i_head_snapc);
Sage Weil1d3576f2009-10-06 11:31:09 -0700407 dout(" head snapc %p has %d dirty pages\n",
408 snapc, ci->i_wrbuffer_ref_head);
409 }
Sage Weil1d3576f2009-10-06 11:31:09 -0700410 spin_unlock(&inode->i_lock);
411 return snapc;
412}
413
414/*
415 * Write a single page, but leave the page locked.
416 *
417 * If we get a write error, set the page error bit, but still adjust the
418 * dirty page accounting (i.e., page is no longer dirty).
419 */
420static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
421{
422 struct inode *inode;
423 struct ceph_inode_info *ci;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700424 struct ceph_fs_client *fsc;
Sage Weil1d3576f2009-10-06 11:31:09 -0700425 struct ceph_osd_client *osdc;
426 loff_t page_off = page->index << PAGE_CACHE_SHIFT;
427 int len = PAGE_CACHE_SIZE;
428 loff_t i_size;
429 int err = 0;
Sage Weil6298a332010-03-31 22:01:38 -0700430 struct ceph_snap_context *snapc, *oldest;
Sage Weil1d3576f2009-10-06 11:31:09 -0700431 u64 snap_size = 0;
Yehuda Sadeh2baba252009-12-18 13:51:57 -0800432 long writeback_stat;
Sage Weil1d3576f2009-10-06 11:31:09 -0700433
434 dout("writepage %p idx %lu\n", page, page->index);
435
436 if (!page->mapping || !page->mapping->host) {
437 dout("writepage %p - no mapping\n", page);
438 return -EFAULT;
439 }
440 inode = page->mapping->host;
441 ci = ceph_inode(inode);
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700442 fsc = ceph_inode_to_client(inode);
443 osdc = &fsc->client->osdc;
Sage Weil1d3576f2009-10-06 11:31:09 -0700444
445 /* verify this is a writeable snap context */
446 snapc = (void *)page->private;
447 if (snapc == NULL) {
448 dout("writepage %p page %p not dirty?\n", inode, page);
449 goto out;
450 }
Sage Weil6298a332010-03-31 22:01:38 -0700451 oldest = get_oldest_context(inode, &snap_size);
452 if (snapc->seq > oldest->seq) {
Sage Weil1d3576f2009-10-06 11:31:09 -0700453 dout("writepage %p page %p snapc %p not writeable - noop\n",
454 inode, page, (void *)page->private);
455 /* we should only noop if called by kswapd */
456 WARN_ON((current->flags & PF_MEMALLOC) == 0);
Sage Weil6298a332010-03-31 22:01:38 -0700457 ceph_put_snap_context(oldest);
Sage Weil1d3576f2009-10-06 11:31:09 -0700458 goto out;
459 }
Sage Weil6298a332010-03-31 22:01:38 -0700460 ceph_put_snap_context(oldest);
Sage Weil1d3576f2009-10-06 11:31:09 -0700461
462 /* is this a partial page at end of file? */
463 if (snap_size)
464 i_size = snap_size;
465 else
466 i_size = i_size_read(inode);
467 if (i_size < page_off + len)
468 len = i_size - page_off;
469
Sage Weilae00d4f2010-09-16 16:26:51 -0700470 dout("writepage %p page %p index %lu on %llu~%u snapc %p\n",
471 inode, page, page->index, page_off, len, snapc);
Sage Weil1d3576f2009-10-06 11:31:09 -0700472
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700473 writeback_stat = atomic_long_inc_return(&fsc->writeback_count);
Yehuda Sadeh2baba252009-12-18 13:51:57 -0800474 if (writeback_stat >
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700475 CONGESTION_ON_THRESH(fsc->mount_options->congestion_kb))
476 set_bdi_congested(&fsc->backing_dev_info, BLK_RW_ASYNC);
Yehuda Sadeh2baba252009-12-18 13:51:57 -0800477
Sage Weil1d3576f2009-10-06 11:31:09 -0700478 set_page_writeback(page);
479 err = ceph_osdc_writepages(osdc, ceph_vino(inode),
480 &ci->i_layout, snapc,
481 page_off, len,
482 ci->i_truncate_seq, ci->i_truncate_size,
483 &inode->i_mtime,
484 &page, 1, 0, 0, true);
485 if (err < 0) {
486 dout("writepage setting page/mapping error %d %p\n", err, page);
487 SetPageError(page);
488 mapping_set_error(&inode->i_data, err);
489 if (wbc)
490 wbc->pages_skipped++;
491 } else {
492 dout("writepage cleaned page %p\n", page);
493 err = 0; /* vfs expects us to return 0 */
494 }
495 page->private = 0;
496 ClearPagePrivate(page);
497 end_page_writeback(page);
498 ceph_put_wrbuffer_cap_refs(ci, 1, snapc);
Sage Weil6298a332010-03-31 22:01:38 -0700499 ceph_put_snap_context(snapc); /* page's reference */
Sage Weil1d3576f2009-10-06 11:31:09 -0700500out:
501 return err;
502}
503
504static int ceph_writepage(struct page *page, struct writeback_control *wbc)
505{
Yehuda Sadehdbd646a2009-12-16 14:51:06 -0800506 int err;
507 struct inode *inode = page->mapping->host;
508 BUG_ON(!inode);
Sage Weil70b666c2011-05-27 09:24:26 -0700509 ihold(inode);
Yehuda Sadehdbd646a2009-12-16 14:51:06 -0800510 err = writepage_nounlock(page, wbc);
Sage Weil1d3576f2009-10-06 11:31:09 -0700511 unlock_page(page);
Yehuda Sadehdbd646a2009-12-16 14:51:06 -0800512 iput(inode);
Sage Weil1d3576f2009-10-06 11:31:09 -0700513 return err;
514}
515
516
517/*
518 * lame release_pages helper. release_pages() isn't exported to
519 * modules.
520 */
521static void ceph_release_pages(struct page **pages, int num)
522{
523 struct pagevec pvec;
524 int i;
525
526 pagevec_init(&pvec, 0);
527 for (i = 0; i < num; i++) {
528 if (pagevec_add(&pvec, pages[i]) == 0)
529 pagevec_release(&pvec);
530 }
531 pagevec_release(&pvec);
532}
533
534
535/*
536 * async writeback completion handler.
537 *
538 * If we get an error, set the mapping error bit, but not the individual
539 * page error bits.
540 */
541static void writepages_finish(struct ceph_osd_request *req,
542 struct ceph_msg *msg)
543{
544 struct inode *inode = req->r_inode;
545 struct ceph_osd_reply_head *replyhead;
546 struct ceph_osd_op *op;
547 struct ceph_inode_info *ci = ceph_inode(inode);
548 unsigned wrote;
Sage Weil1d3576f2009-10-06 11:31:09 -0700549 struct page *page;
550 int i;
551 struct ceph_snap_context *snapc = req->r_snapc;
552 struct address_space *mapping = inode->i_mapping;
Sage Weil1d3576f2009-10-06 11:31:09 -0700553 __s32 rc = -EIO;
554 u64 bytes = 0;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700555 struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
Yehuda Sadeh2baba252009-12-18 13:51:57 -0800556 long writeback_stat;
Sage Weil7ff899d2010-04-23 10:25:33 -0700557 unsigned issued = ceph_caps_issued(ci);
Sage Weil1d3576f2009-10-06 11:31:09 -0700558
559 /* parse reply */
560 replyhead = msg->front.iov_base;
561 WARN_ON(le32_to_cpu(replyhead->num_ops) == 0);
562 op = (void *)(replyhead + 1);
563 rc = le32_to_cpu(replyhead->result);
564 bytes = le64_to_cpu(op->extent.length);
565
566 if (rc >= 0) {
Sage Weil79788c62010-02-02 16:34:04 -0800567 /*
568 * Assume we wrote the pages we originally sent. The
569 * osd might reply with fewer pages if our writeback
570 * raced with a truncation and was adjusted at the osd,
571 * so don't believe the reply.
572 */
573 wrote = req->r_num_pages;
Sage Weil1d3576f2009-10-06 11:31:09 -0700574 } else {
575 wrote = 0;
576 mapping_set_error(mapping, rc);
577 }
578 dout("writepages_finish %p rc %d bytes %llu wrote %d (pages)\n",
579 inode, rc, bytes, wrote);
580
581 /* clean all pages */
582 for (i = 0; i < req->r_num_pages; i++) {
583 page = req->r_pages[i];
584 BUG_ON(!page);
585 WARN_ON(!PageUptodate(page));
586
Yehuda Sadeh2baba252009-12-18 13:51:57 -0800587 writeback_stat =
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700588 atomic_long_dec_return(&fsc->writeback_count);
Yehuda Sadeh2baba252009-12-18 13:51:57 -0800589 if (writeback_stat <
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700590 CONGESTION_OFF_THRESH(fsc->mount_options->congestion_kb))
591 clear_bdi_congested(&fsc->backing_dev_info,
Yehuda Sadeh2baba252009-12-18 13:51:57 -0800592 BLK_RW_ASYNC);
593
Sage Weil80e755f2010-03-31 21:52:10 -0700594 ceph_put_snap_context((void *)page->private);
Sage Weil1d3576f2009-10-06 11:31:09 -0700595 page->private = 0;
596 ClearPagePrivate(page);
Sage Weil1d3576f2009-10-06 11:31:09 -0700597 dout("unlocking %d %p\n", i, page);
598 end_page_writeback(page);
Yehuda Sadehe63dc5c2010-02-19 00:07:01 +0000599
600 /*
601 * We lost the cache cap, need to truncate the page before
602 * it is unlocked, otherwise we'd truncate it later in the
603 * page truncation thread, possibly losing some data that
604 * raced its way in
605 */
Sage Weil29625072010-05-27 10:40:43 -0700606 if ((issued & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) == 0)
Yehuda Sadehe63dc5c2010-02-19 00:07:01 +0000607 generic_error_remove_page(inode->i_mapping, page);
608
Sage Weil1d3576f2009-10-06 11:31:09 -0700609 unlock_page(page);
610 }
611 dout("%p wrote+cleaned %d pages\n", inode, wrote);
612 ceph_put_wrbuffer_cap_refs(ci, req->r_num_pages, snapc);
613
614 ceph_release_pages(req->r_pages, req->r_num_pages);
615 if (req->r_pages_from_pool)
616 mempool_free(req->r_pages,
Cheng Renquan640ef792010-03-26 17:40:33 +0800617 ceph_sb_to_client(inode->i_sb)->wb_pagevec_pool);
Sage Weil1d3576f2009-10-06 11:31:09 -0700618 else
619 kfree(req->r_pages);
620 ceph_osdc_put_request(req);
621}
622
623/*
624 * allocate a page vec, either directly, or if necessary, via a the
625 * mempool. we avoid the mempool if we can because req->r_num_pages
626 * may be less than the maximum write size.
627 */
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700628static void alloc_page_vec(struct ceph_fs_client *fsc,
Sage Weil1d3576f2009-10-06 11:31:09 -0700629 struct ceph_osd_request *req)
630{
631 req->r_pages = kmalloc(sizeof(struct page *) * req->r_num_pages,
632 GFP_NOFS);
633 if (!req->r_pages) {
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700634 req->r_pages = mempool_alloc(fsc->wb_pagevec_pool, GFP_NOFS);
Sage Weil1d3576f2009-10-06 11:31:09 -0700635 req->r_pages_from_pool = 1;
636 WARN_ON(!req->r_pages);
637 }
638}
639
640/*
641 * initiate async writeback
642 */
643static int ceph_writepages_start(struct address_space *mapping,
644 struct writeback_control *wbc)
645{
646 struct inode *inode = mapping->host;
Sage Weil1d3576f2009-10-06 11:31:09 -0700647 struct ceph_inode_info *ci = ceph_inode(inode);
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700648 struct ceph_fs_client *fsc;
Sage Weil1d3576f2009-10-06 11:31:09 -0700649 pgoff_t index, start, end;
650 int range_whole = 0;
651 int should_loop = 1;
652 pgoff_t max_pages = 0, max_pages_ever = 0;
Sage Weil80e755f2010-03-31 21:52:10 -0700653 struct ceph_snap_context *snapc = NULL, *last_snapc = NULL, *pgsnapc;
Sage Weil1d3576f2009-10-06 11:31:09 -0700654 struct pagevec pvec;
655 int done = 0;
656 int rc = 0;
657 unsigned wsize = 1 << inode->i_blkbits;
658 struct ceph_osd_request *req = NULL;
659 int do_sync;
660 u64 snap_size = 0;
661
662 /*
663 * Include a 'sync' in the OSD request if this is a data
664 * integrity write (e.g., O_SYNC write or fsync()), or if our
665 * cap is being revoked.
666 */
667 do_sync = wbc->sync_mode == WB_SYNC_ALL;
668 if (ceph_caps_revoking(ci, CEPH_CAP_FILE_BUFFER))
669 do_sync = 1;
670 dout("writepages_start %p dosync=%d (mode=%s)\n",
671 inode, do_sync,
672 wbc->sync_mode == WB_SYNC_NONE ? "NONE" :
673 (wbc->sync_mode == WB_SYNC_ALL ? "ALL" : "HOLD"));
674
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700675 fsc = ceph_inode_to_client(inode);
676 if (fsc->mount_state == CEPH_MOUNT_SHUTDOWN) {
Sage Weil1d3576f2009-10-06 11:31:09 -0700677 pr_warning("writepage_start %p on forced umount\n", inode);
678 return -EIO; /* we're in a forced umount, don't write! */
679 }
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700680 if (fsc->mount_options->wsize && fsc->mount_options->wsize < wsize)
681 wsize = fsc->mount_options->wsize;
Sage Weil1d3576f2009-10-06 11:31:09 -0700682 if (wsize < PAGE_CACHE_SIZE)
683 wsize = PAGE_CACHE_SIZE;
684 max_pages_ever = wsize >> PAGE_CACHE_SHIFT;
685
686 pagevec_init(&pvec, 0);
687
Sage Weil1d3576f2009-10-06 11:31:09 -0700688 /* where to start/end? */
689 if (wbc->range_cyclic) {
690 start = mapping->writeback_index; /* Start from prev offset */
691 end = -1;
692 dout(" cyclic, start at %lu\n", start);
693 } else {
694 start = wbc->range_start >> PAGE_CACHE_SHIFT;
695 end = wbc->range_end >> PAGE_CACHE_SHIFT;
696 if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX)
697 range_whole = 1;
698 should_loop = 0;
699 dout(" not cyclic, %lu to %lu\n", start, end);
700 }
701 index = start;
702
703retry:
704 /* find oldest snap context with dirty data */
705 ceph_put_snap_context(snapc);
706 snapc = get_oldest_context(inode, &snap_size);
707 if (!snapc) {
708 /* hmm, why does writepages get called when there
709 is no dirty data? */
710 dout(" no snap context with dirty data?\n");
711 goto out;
712 }
713 dout(" oldest snapc is %p seq %lld (%d snaps)\n",
714 snapc, snapc->seq, snapc->num_snaps);
715 if (last_snapc && snapc != last_snapc) {
716 /* if we switched to a newer snapc, restart our scan at the
717 * start of the original file range. */
718 dout(" snapc differs from last pass, restarting at %lu\n",
719 index);
720 index = start;
721 }
722 last_snapc = snapc;
723
724 while (!done && index <= end) {
725 unsigned i;
726 int first;
727 pgoff_t next;
728 int pvec_pages, locked_pages;
729 struct page *page;
730 int want;
731 u64 offset, len;
732 struct ceph_osd_request_head *reqhead;
733 struct ceph_osd_op *op;
Yehuda Sadeh2baba252009-12-18 13:51:57 -0800734 long writeback_stat;
Sage Weil1d3576f2009-10-06 11:31:09 -0700735
736 next = 0;
737 locked_pages = 0;
738 max_pages = max_pages_ever;
739
740get_more_pages:
741 first = -1;
742 want = min(end - index,
743 min((pgoff_t)PAGEVEC_SIZE,
744 max_pages - (pgoff_t)locked_pages) - 1)
745 + 1;
746 pvec_pages = pagevec_lookup_tag(&pvec, mapping, &index,
747 PAGECACHE_TAG_DIRTY,
748 want);
749 dout("pagevec_lookup_tag got %d\n", pvec_pages);
750 if (!pvec_pages && !locked_pages)
751 break;
752 for (i = 0; i < pvec_pages && locked_pages < max_pages; i++) {
753 page = pvec.pages[i];
754 dout("? %p idx %lu\n", page, page->index);
755 if (locked_pages == 0)
756 lock_page(page); /* first page */
757 else if (!trylock_page(page))
758 break;
759
760 /* only dirty pages, or our accounting breaks */
761 if (unlikely(!PageDirty(page)) ||
762 unlikely(page->mapping != mapping)) {
763 dout("!dirty or !mapping %p\n", page);
764 unlock_page(page);
765 break;
766 }
767 if (!wbc->range_cyclic && page->index > end) {
768 dout("end of range %p\n", page);
769 done = 1;
770 unlock_page(page);
771 break;
772 }
773 if (next && (page->index != next)) {
774 dout("not consecutive %p\n", page);
775 unlock_page(page);
776 break;
777 }
778 if (wbc->sync_mode != WB_SYNC_NONE) {
779 dout("waiting on writeback %p\n", page);
780 wait_on_page_writeback(page);
781 }
782 if ((snap_size && page_offset(page) > snap_size) ||
783 (!snap_size &&
784 page_offset(page) > i_size_read(inode))) {
785 dout("%p page eof %llu\n", page, snap_size ?
786 snap_size : i_size_read(inode));
787 done = 1;
788 unlock_page(page);
789 break;
790 }
791 if (PageWriteback(page)) {
792 dout("%p under writeback\n", page);
793 unlock_page(page);
794 break;
795 }
796
797 /* only if matching snap context */
Sage Weil80e755f2010-03-31 21:52:10 -0700798 pgsnapc = (void *)page->private;
799 if (pgsnapc->seq > snapc->seq) {
800 dout("page snapc %p %lld > oldest %p %lld\n",
801 pgsnapc, pgsnapc->seq, snapc, snapc->seq);
Sage Weil1d3576f2009-10-06 11:31:09 -0700802 unlock_page(page);
803 if (!locked_pages)
804 continue; /* keep looking for snap */
805 break;
806 }
807
808 if (!clear_page_dirty_for_io(page)) {
809 dout("%p !clear_page_dirty_for_io\n", page);
810 unlock_page(page);
811 break;
812 }
813
814 /* ok */
815 if (locked_pages == 0) {
816 /* prepare async write request */
Sage Weila77d9f72010-09-11 10:55:25 -0700817 offset = (unsigned long long)page->index
818 << PAGE_CACHE_SHIFT;
Sage Weil1d3576f2009-10-06 11:31:09 -0700819 len = wsize;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700820 req = ceph_osdc_new_request(&fsc->client->osdc,
Sage Weil1d3576f2009-10-06 11:31:09 -0700821 &ci->i_layout,
822 ceph_vino(inode),
823 offset, &len,
824 CEPH_OSD_OP_WRITE,
825 CEPH_OSD_FLAG_WRITE |
826 CEPH_OSD_FLAG_ONDISK,
827 snapc, do_sync,
828 ci->i_truncate_seq,
829 ci->i_truncate_size,
Sage Weilb7495fc2010-11-09 12:43:12 -0800830 &inode->i_mtime, true, 1, 0);
Henry C Chang8c718972011-05-03 09:45:16 +0000831
832 if (!req) {
833 rc = -ENOMEM;
834 unlock_page(page);
835 break;
836 }
837
Sage Weil1d3576f2009-10-06 11:31:09 -0700838 max_pages = req->r_num_pages;
839
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700840 alloc_page_vec(fsc, req);
Sage Weil1d3576f2009-10-06 11:31:09 -0700841 req->r_callback = writepages_finish;
842 req->r_inode = inode;
Sage Weil1d3576f2009-10-06 11:31:09 -0700843 }
844
845 /* note position of first page in pvec */
846 if (first < 0)
847 first = i;
848 dout("%p will write page %p idx %lu\n",
849 inode, page, page->index);
Yehuda Sadeh2baba252009-12-18 13:51:57 -0800850
Sage Weil213c99e2010-08-03 10:25:11 -0700851 writeback_stat =
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700852 atomic_long_inc_return(&fsc->writeback_count);
Sage Weil213c99e2010-08-03 10:25:11 -0700853 if (writeback_stat > CONGESTION_ON_THRESH(
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700854 fsc->mount_options->congestion_kb)) {
855 set_bdi_congested(&fsc->backing_dev_info,
Sage Weil213c99e2010-08-03 10:25:11 -0700856 BLK_RW_ASYNC);
Yehuda Sadeh2baba252009-12-18 13:51:57 -0800857 }
858
Sage Weil1d3576f2009-10-06 11:31:09 -0700859 set_page_writeback(page);
860 req->r_pages[locked_pages] = page;
861 locked_pages++;
862 next = page->index + 1;
863 }
864
865 /* did we get anything? */
866 if (!locked_pages)
867 goto release_pvec_pages;
868 if (i) {
869 int j;
870 BUG_ON(!locked_pages || first < 0);
871
872 if (pvec_pages && i == pvec_pages &&
873 locked_pages < max_pages) {
874 dout("reached end pvec, trying for more\n");
875 pagevec_reinit(&pvec);
876 goto get_more_pages;
877 }
878
879 /* shift unused pages over in the pvec... we
880 * will need to release them below. */
881 for (j = i; j < pvec_pages; j++) {
882 dout(" pvec leftover page %p\n",
883 pvec.pages[j]);
884 pvec.pages[j-i+first] = pvec.pages[j];
885 }
886 pvec.nr -= i-first;
887 }
888
889 /* submit the write */
890 offset = req->r_pages[0]->index << PAGE_CACHE_SHIFT;
891 len = min((snap_size ? snap_size : i_size_read(inode)) - offset,
892 (u64)locked_pages << PAGE_CACHE_SHIFT);
893 dout("writepages got %d pages at %llu~%llu\n",
894 locked_pages, offset, len);
895
896 /* revise final length, page count */
897 req->r_num_pages = locked_pages;
898 reqhead = req->r_request->front.iov_base;
899 op = (void *)(reqhead + 1);
900 op->extent.length = cpu_to_le64(len);
901 op->payload_len = cpu_to_le32(len);
902 req->r_request->hdr.data_len = cpu_to_le32(len);
903
Sage Weil9d6fcb02011-05-12 15:48:16 -0700904 rc = ceph_osdc_start_request(&fsc->client->osdc, req, true);
905 BUG_ON(rc);
Sage Weil1d3576f2009-10-06 11:31:09 -0700906 req = NULL;
907
908 /* continue? */
909 index = next;
910 wbc->nr_to_write -= locked_pages;
911 if (wbc->nr_to_write <= 0)
912 done = 1;
913
914release_pvec_pages:
915 dout("pagevec_release on %d pages (%p)\n", (int)pvec.nr,
916 pvec.nr ? pvec.pages[0] : NULL);
917 pagevec_release(&pvec);
918
919 if (locked_pages && !done)
920 goto retry;
921 }
922
923 if (should_loop && !done) {
924 /* more to do; loop back to beginning of file */
925 dout("writepages looping back to beginning of file\n");
926 should_loop = 0;
927 index = 0;
928 goto retry;
929 }
930
931 if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0))
932 mapping->writeback_index = index;
933
934out:
935 if (req)
936 ceph_osdc_put_request(req);
Sage Weil1d3576f2009-10-06 11:31:09 -0700937 ceph_put_snap_context(snapc);
938 dout("writepages done, rc = %d\n", rc);
Sage Weil1d3576f2009-10-06 11:31:09 -0700939 return rc;
940}
941
942
943
944/*
945 * See if a given @snapc is either writeable, or already written.
946 */
947static int context_is_writeable_or_written(struct inode *inode,
948 struct ceph_snap_context *snapc)
949{
950 struct ceph_snap_context *oldest = get_oldest_context(inode, NULL);
Sage Weil6298a332010-03-31 22:01:38 -0700951 int ret = !oldest || snapc->seq <= oldest->seq;
952
953 ceph_put_snap_context(oldest);
954 return ret;
Sage Weil1d3576f2009-10-06 11:31:09 -0700955}
956
957/*
958 * We are only allowed to write into/dirty the page if the page is
959 * clean, or already dirty within the same snap context.
Sage Weil8f883c22010-03-19 13:27:53 -0700960 *
961 * called with page locked.
962 * return success with page locked,
963 * or any failure (incl -EAGAIN) with page unlocked.
Sage Weil1d3576f2009-10-06 11:31:09 -0700964 */
Yehuda Sadeh4af6b222010-02-09 11:02:51 -0800965static int ceph_update_writeable_page(struct file *file,
966 loff_t pos, unsigned len,
967 struct page *page)
Sage Weil1d3576f2009-10-06 11:31:09 -0700968{
969 struct inode *inode = file->f_dentry->d_inode;
970 struct ceph_inode_info *ci = ceph_inode(inode);
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -0700971 struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
Sage Weil1d3576f2009-10-06 11:31:09 -0700972 loff_t page_off = pos & PAGE_CACHE_MASK;
973 int pos_in_page = pos & ~PAGE_CACHE_MASK;
974 int end_in_page = pos_in_page + len;
975 loff_t i_size;
Sage Weil1d3576f2009-10-06 11:31:09 -0700976 int r;
Sage Weil80e755f2010-03-31 21:52:10 -0700977 struct ceph_snap_context *snapc, *oldest;
Sage Weil1d3576f2009-10-06 11:31:09 -0700978
Sage Weil1d3576f2009-10-06 11:31:09 -0700979retry_locked:
980 /* writepages currently holds page lock, but if we change that later, */
981 wait_on_page_writeback(page);
982
983 /* check snap context */
984 BUG_ON(!ci->i_snap_realm);
985 down_read(&mdsc->snap_rwsem);
986 BUG_ON(!ci->i_snap_realm->cached_context);
Sage Weil80e755f2010-03-31 21:52:10 -0700987 snapc = (void *)page->private;
988 if (snapc && snapc != ci->i_head_snapc) {
Sage Weil1d3576f2009-10-06 11:31:09 -0700989 /*
990 * this page is already dirty in another (older) snap
991 * context! is it writeable now?
992 */
Sage Weil80e755f2010-03-31 21:52:10 -0700993 oldest = get_oldest_context(inode, NULL);
Sage Weil1d3576f2009-10-06 11:31:09 -0700994 up_read(&mdsc->snap_rwsem);
995
Sage Weil80e755f2010-03-31 21:52:10 -0700996 if (snapc->seq > oldest->seq) {
Sage Weil6298a332010-03-31 22:01:38 -0700997 ceph_put_snap_context(oldest);
Sage Weil1d3576f2009-10-06 11:31:09 -0700998 dout(" page %p snapc %p not current or oldest\n",
Sage Weil6298a332010-03-31 22:01:38 -0700999 page, snapc);
Sage Weil1d3576f2009-10-06 11:31:09 -07001000 /*
1001 * queue for writeback, and wait for snapc to
1002 * be writeable or written
1003 */
Sage Weil6298a332010-03-31 22:01:38 -07001004 snapc = ceph_get_snap_context(snapc);
Sage Weil1d3576f2009-10-06 11:31:09 -07001005 unlock_page(page);
Sage Weil3c6f6b72010-02-09 15:24:44 -08001006 ceph_queue_writeback(inode);
Sage Weil8f883c22010-03-19 13:27:53 -07001007 r = wait_event_interruptible(ci->i_cap_wq,
Sage Weil1d3576f2009-10-06 11:31:09 -07001008 context_is_writeable_or_written(inode, snapc));
1009 ceph_put_snap_context(snapc);
Sage Weil8f883c22010-03-19 13:27:53 -07001010 if (r == -ERESTARTSYS)
1011 return r;
Yehuda Sadeh4af6b222010-02-09 11:02:51 -08001012 return -EAGAIN;
Sage Weil1d3576f2009-10-06 11:31:09 -07001013 }
Sage Weil6298a332010-03-31 22:01:38 -07001014 ceph_put_snap_context(oldest);
Sage Weil1d3576f2009-10-06 11:31:09 -07001015
1016 /* yay, writeable, do it now (without dropping page lock) */
1017 dout(" page %p snapc %p not current, but oldest\n",
1018 page, snapc);
1019 if (!clear_page_dirty_for_io(page))
1020 goto retry_locked;
1021 r = writepage_nounlock(page, NULL);
1022 if (r < 0)
1023 goto fail_nosnap;
1024 goto retry_locked;
1025 }
1026
1027 if (PageUptodate(page)) {
1028 dout(" page %p already uptodate\n", page);
1029 return 0;
1030 }
1031
1032 /* full page? */
1033 if (pos_in_page == 0 && len == PAGE_CACHE_SIZE)
1034 return 0;
1035
1036 /* past end of file? */
1037 i_size = inode->i_size; /* caller holds i_mutex */
1038
1039 if (i_size + len > inode->i_sb->s_maxbytes) {
1040 /* file is too big */
1041 r = -EINVAL;
1042 goto fail;
1043 }
1044
1045 if (page_off >= i_size ||
1046 (pos_in_page == 0 && (pos+len) >= i_size &&
1047 end_in_page - pos_in_page != PAGE_CACHE_SIZE)) {
1048 dout(" zeroing %p 0 - %d and %d - %d\n",
1049 page, pos_in_page, end_in_page, (int)PAGE_CACHE_SIZE);
1050 zero_user_segments(page,
1051 0, pos_in_page,
1052 end_in_page, PAGE_CACHE_SIZE);
1053 return 0;
1054 }
1055
1056 /* we need to read it. */
1057 up_read(&mdsc->snap_rwsem);
1058 r = readpage_nounlock(file, page);
1059 if (r < 0)
1060 goto fail_nosnap;
1061 goto retry_locked;
1062
1063fail:
1064 up_read(&mdsc->snap_rwsem);
1065fail_nosnap:
1066 unlock_page(page);
1067 return r;
1068}
1069
1070/*
Yehuda Sadeh4af6b222010-02-09 11:02:51 -08001071 * We are only allowed to write into/dirty the page if the page is
1072 * clean, or already dirty within the same snap context.
1073 */
1074static int ceph_write_begin(struct file *file, struct address_space *mapping,
1075 loff_t pos, unsigned len, unsigned flags,
1076 struct page **pagep, void **fsdata)
1077{
1078 struct inode *inode = file->f_dentry->d_inode;
1079 struct page *page;
1080 pgoff_t index = pos >> PAGE_CACHE_SHIFT;
1081 int r;
1082
1083 do {
Sage Weil8f883c22010-03-19 13:27:53 -07001084 /* get a page */
Yehuda Sadeh4af6b222010-02-09 11:02:51 -08001085 page = grab_cache_page_write_begin(mapping, index, 0);
1086 if (!page)
1087 return -ENOMEM;
1088 *pagep = page;
1089
1090 dout("write_begin file %p inode %p page %p %d~%d\n", file,
Sage Weil213c99e2010-08-03 10:25:11 -07001091 inode, page, (int)pos, (int)len);
Yehuda Sadeh4af6b222010-02-09 11:02:51 -08001092
1093 r = ceph_update_writeable_page(file, pos, len, page);
1094 } while (r == -EAGAIN);
1095
1096 return r;
1097}
1098
1099/*
Sage Weil1d3576f2009-10-06 11:31:09 -07001100 * we don't do anything in here that simple_write_end doesn't do
1101 * except adjust dirty page accounting and drop read lock on
1102 * mdsc->snap_rwsem.
1103 */
1104static int ceph_write_end(struct file *file, struct address_space *mapping,
1105 loff_t pos, unsigned len, unsigned copied,
1106 struct page *page, void *fsdata)
1107{
1108 struct inode *inode = file->f_dentry->d_inode;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -07001109 struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
1110 struct ceph_mds_client *mdsc = fsc->mdsc;
Sage Weil1d3576f2009-10-06 11:31:09 -07001111 unsigned from = pos & (PAGE_CACHE_SIZE - 1);
1112 int check_cap = 0;
1113
1114 dout("write_end file %p inode %p page %p %d~%d (%d)\n", file,
1115 inode, page, (int)pos, (int)copied, (int)len);
1116
1117 /* zero the stale part of the page if we did a short copy */
1118 if (copied < len)
1119 zero_user_segment(page, from+copied, len);
1120
1121 /* did file size increase? */
1122 /* (no need for i_size_read(); we caller holds i_mutex */
1123 if (pos+copied > inode->i_size)
1124 check_cap = ceph_inode_set_size(inode, pos+copied);
1125
1126 if (!PageUptodate(page))
1127 SetPageUptodate(page);
1128
1129 set_page_dirty(page);
1130
1131 unlock_page(page);
1132 up_read(&mdsc->snap_rwsem);
1133 page_cache_release(page);
1134
1135 if (check_cap)
1136 ceph_check_caps(ceph_inode(inode), CHECK_CAPS_AUTHONLY, NULL);
1137
1138 return copied;
1139}
1140
1141/*
1142 * we set .direct_IO to indicate direct io is supported, but since we
1143 * intercept O_DIRECT reads and writes early, this function should
1144 * never get called.
1145 */
1146static ssize_t ceph_direct_io(int rw, struct kiocb *iocb,
1147 const struct iovec *iov,
1148 loff_t pos, unsigned long nr_segs)
1149{
1150 WARN_ON(1);
1151 return -EINVAL;
1152}
1153
1154const struct address_space_operations ceph_aops = {
1155 .readpage = ceph_readpage,
1156 .readpages = ceph_readpages,
1157 .writepage = ceph_writepage,
1158 .writepages = ceph_writepages_start,
1159 .write_begin = ceph_write_begin,
1160 .write_end = ceph_write_end,
1161 .set_page_dirty = ceph_set_page_dirty,
1162 .invalidatepage = ceph_invalidatepage,
1163 .releasepage = ceph_releasepage,
1164 .direct_IO = ceph_direct_io,
1165};
1166
1167
1168/*
1169 * vm ops
1170 */
1171
1172/*
1173 * Reuse write_begin here for simplicity.
1174 */
1175static int ceph_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
1176{
1177 struct inode *inode = vma->vm_file->f_dentry->d_inode;
1178 struct page *page = vmf->page;
Yehuda Sadeh3d14c5d2010-04-06 15:14:15 -07001179 struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
Sage Weil1d3576f2009-10-06 11:31:09 -07001180 loff_t off = page->index << PAGE_CACHE_SHIFT;
1181 loff_t size, len;
Sage Weil1d3576f2009-10-06 11:31:09 -07001182 int ret;
1183
1184 size = i_size_read(inode);
1185 if (off + PAGE_CACHE_SIZE <= size)
1186 len = PAGE_CACHE_SIZE;
1187 else
1188 len = size & ~PAGE_CACHE_MASK;
1189
1190 dout("page_mkwrite %p %llu~%llu page %p idx %lu\n", inode,
1191 off, len, page, page->index);
Yehuda Sadeh4af6b222010-02-09 11:02:51 -08001192
1193 lock_page(page);
1194
1195 ret = VM_FAULT_NOPAGE;
1196 if ((off > size) ||
1197 (page->mapping != inode->i_mapping))
1198 goto out;
1199
1200 ret = ceph_update_writeable_page(vma->vm_file, off, len, page);
1201 if (ret == 0) {
1202 /* success. we'll keep the page locked. */
Sage Weil1d3576f2009-10-06 11:31:09 -07001203 set_page_dirty(page);
1204 up_read(&mdsc->snap_rwsem);
Sage Weil1d3576f2009-10-06 11:31:09 -07001205 ret = VM_FAULT_LOCKED;
1206 } else {
Yehuda Sadeh4af6b222010-02-09 11:02:51 -08001207 if (ret == -ENOMEM)
1208 ret = VM_FAULT_OOM;
1209 else
1210 ret = VM_FAULT_SIGBUS;
Sage Weil1d3576f2009-10-06 11:31:09 -07001211 }
Yehuda Sadeh4af6b222010-02-09 11:02:51 -08001212out:
Sage Weil1d3576f2009-10-06 11:31:09 -07001213 dout("page_mkwrite %p %llu~%llu = %d\n", inode, off, len, ret);
Yehuda Sadeh4af6b222010-02-09 11:02:51 -08001214 if (ret != VM_FAULT_LOCKED)
1215 unlock_page(page);
Sage Weil1d3576f2009-10-06 11:31:09 -07001216 return ret;
1217}
1218
1219static struct vm_operations_struct ceph_vmops = {
1220 .fault = filemap_fault,
1221 .page_mkwrite = ceph_page_mkwrite,
1222};
1223
1224int ceph_mmap(struct file *file, struct vm_area_struct *vma)
1225{
1226 struct address_space *mapping = file->f_mapping;
1227
1228 if (!mapping->a_ops->readpage)
1229 return -ENOEXEC;
1230 file_accessed(file);
1231 vma->vm_ops = &ceph_vmops;
1232 vma->vm_flags |= VM_CAN_NONLINEAR;
1233 return 0;
1234}