]>
Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * linux/fs/nfs/direct.c | |
3 | * | |
4 | * Copyright (C) 2003 by Chuck Lever <cel@netapp.com> | |
5 | * | |
6 | * High-performance uncached I/O for the Linux NFS client | |
7 | * | |
8 | * There are important applications whose performance or correctness | |
9 | * depends on uncached access to file data. Database clusters | |
88467055 | 10 | * (multiple copies of the same instance running on separate hosts) |
1da177e4 | 11 | * implement their own cache coherency protocol that subsumes file |
88467055 CL |
12 | * system cache protocols. Applications that process datasets |
13 | * considerably larger than the client's memory do not always benefit | |
14 | * from a local cache. A streaming video server, for instance, has no | |
1da177e4 LT |
15 | * need to cache the contents of a file. |
16 | * | |
17 | * When an application requests uncached I/O, all read and write requests | |
18 | * are made directly to the server; data stored or fetched via these | |
19 | * requests is not cached in the Linux page cache. The client does not | |
20 | * correct unaligned requests from applications. All requested bytes are | |
21 | * held on permanent storage before a direct write system call returns to | |
22 | * an application. | |
23 | * | |
24 | * Solaris implements an uncached I/O facility called directio() that | |
25 | * is used for backups and sequential I/O to very large files. Solaris | |
26 | * also supports uncaching whole NFS partitions with "-o forcedirectio," | |
27 | * an undocumented mount option. | |
28 | * | |
29 | * Designed by Jeff Kimmel, Chuck Lever, and Trond Myklebust, with | |
30 | * help from Andrew Morton. | |
31 | * | |
32 | * 18 Dec 2001 Initial implementation for 2.4 --cel | |
33 | * 08 Jul 2002 Version for 2.4.19, with bug fixes --trondmy | |
34 | * 08 Jun 2003 Port to 2.5 APIs --cel | |
35 | * 31 Mar 2004 Handle direct I/O without VFS support --cel | |
36 | * 15 Sep 2004 Parallel async reads --cel | |
88467055 | 37 | * 04 May 2005 support O_DIRECT with aio --cel |
1da177e4 LT |
38 | * |
39 | */ | |
40 | ||
1da177e4 LT |
41 | #include <linux/errno.h> |
42 | #include <linux/sched.h> | |
43 | #include <linux/kernel.h> | |
1da177e4 LT |
44 | #include <linux/file.h> |
45 | #include <linux/pagemap.h> | |
46 | #include <linux/kref.h> | |
5a0e3ad6 | 47 | #include <linux/slab.h> |
7ec10f26 | 48 | #include <linux/task_io_accounting_ops.h> |
6296556f | 49 | #include <linux/module.h> |
1da177e4 LT |
50 | |
51 | #include <linux/nfs_fs.h> | |
52 | #include <linux/nfs_page.h> | |
53 | #include <linux/sunrpc/clnt.h> | |
54 | ||
1da177e4 | 55 | #include <asm/uaccess.h> |
60063497 | 56 | #include <linux/atomic.h> |
1da177e4 | 57 | |
8d5658c9 | 58 | #include "internal.h" |
91d5b470 | 59 | #include "iostat.h" |
1763da12 | 60 | #include "pnfs.h" |
91d5b470 | 61 | |
1da177e4 | 62 | #define NFSDBG_FACILITY NFSDBG_VFS |
1da177e4 | 63 | |
e18b890b | 64 | static struct kmem_cache *nfs_direct_cachep; |
1da177e4 LT |
65 | |
66 | /* | |
67 | * This represents a set of asynchronous requests that we're waiting on | |
68 | */ | |
69 | struct nfs_direct_req { | |
70 | struct kref kref; /* release manager */ | |
15ce4a0c CL |
71 | |
72 | /* I/O parameters */ | |
a8881f5a | 73 | struct nfs_open_context *ctx; /* file open context info */ |
f11ac8db | 74 | struct nfs_lock_context *l_ctx; /* Lock context info */ |
99514f8f | 75 | struct kiocb * iocb; /* controlling i/o request */ |
88467055 | 76 | struct inode * inode; /* target file of i/o */ |
15ce4a0c CL |
77 | |
78 | /* completion state */ | |
607f31e8 | 79 | atomic_t io_count; /* i/os we're waiting for */ |
15ce4a0c | 80 | spinlock_t lock; /* protect completion state */ |
15ce4a0c | 81 | ssize_t count, /* bytes actually processed */ |
35754bc0 | 82 | bytes_left, /* bytes left to be sent */ |
1da177e4 | 83 | error; /* any reported error */ |
d72b7a6b | 84 | struct completion completion; /* wait for i/o completion */ |
fad61490 TM |
85 | |
86 | /* commit state */ | |
1763da12 FI |
87 | struct nfs_mds_commit_info mds_cinfo; /* Storage for cinfo */ |
88 | struct pnfs_ds_commit_info ds_cinfo; /* Storage for cinfo */ | |
89 | struct work_struct work; | |
fad61490 TM |
90 | int flags; |
91 | #define NFS_ODIRECT_DO_COMMIT (1) /* an unstable reply was received */ | |
92 | #define NFS_ODIRECT_RESCHED_WRITES (2) /* write verification failed */ | |
93 | struct nfs_writeverf verf; /* unstable write verifier */ | |
1da177e4 LT |
94 | }; |
95 | ||
1763da12 FI |
96 | static const struct nfs_pgio_completion_ops nfs_direct_write_completion_ops; |
97 | static const struct nfs_commit_completion_ops nfs_direct_commit_completion_ops; | |
fad61490 | 98 | static void nfs_direct_write_complete(struct nfs_direct_req *dreq, struct inode *inode); |
1763da12 | 99 | static void nfs_direct_write_schedule_work(struct work_struct *work); |
607f31e8 TM |
100 | |
101 | static inline void get_dreq(struct nfs_direct_req *dreq) | |
102 | { | |
103 | atomic_inc(&dreq->io_count); | |
104 | } | |
105 | ||
106 | static inline int put_dreq(struct nfs_direct_req *dreq) | |
107 | { | |
108 | return atomic_dec_and_test(&dreq->io_count); | |
109 | } | |
110 | ||
1da177e4 | 111 | /** |
b8a32e2b CL |
112 | * nfs_direct_IO - NFS address space operation for direct I/O |
113 | * @rw: direction (read or write) | |
114 | * @iocb: target I/O control block | |
115 | * @iov: array of vectors that define I/O buffer | |
116 | * @pos: offset in file to begin the operation | |
117 | * @nr_segs: size of iovec array | |
118 | * | |
119 | * The presence of this routine in the address space ops vector means | |
a564b8f0 MG |
120 | * the NFS client supports direct I/O. However, for most direct IO, we |
121 | * shunt off direct read and write requests before the VFS gets them, | |
122 | * so this method is only ever called for swap. | |
1da177e4 | 123 | */ |
b8a32e2b CL |
124 | ssize_t nfs_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov, loff_t pos, unsigned long nr_segs) |
125 | { | |
a564b8f0 | 126 | #ifndef CONFIG_NFS_SWAP |
b8a32e2b | 127 | dprintk("NFS: nfs_direct_IO (%s) off/no(%Ld/%lu) EINVAL\n", |
01cce933 | 128 | iocb->ki_filp->f_path.dentry->d_name.name, |
e99170ff | 129 | (long long) pos, nr_segs); |
b8a32e2b CL |
130 | |
131 | return -EINVAL; | |
a564b8f0 MG |
132 | #else |
133 | VM_BUG_ON(iocb->ki_left != PAGE_SIZE); | |
134 | VM_BUG_ON(iocb->ki_nbytes != PAGE_SIZE); | |
135 | ||
136 | if (rw == READ || rw == KERNEL_READ) | |
137 | return nfs_file_direct_read(iocb, iov, nr_segs, pos, | |
138 | rw == READ ? true : false); | |
139 | return nfs_file_direct_write(iocb, iov, nr_segs, pos, | |
140 | rw == WRITE ? true : false); | |
141 | #endif /* CONFIG_NFS_SWAP */ | |
b8a32e2b CL |
142 | } |
143 | ||
749e146e | 144 | static void nfs_direct_release_pages(struct page **pages, unsigned int npages) |
9c93ab7d | 145 | { |
749e146e | 146 | unsigned int i; |
607f31e8 TM |
147 | for (i = 0; i < npages; i++) |
148 | page_cache_release(pages[i]); | |
6b45d858 TM |
149 | } |
150 | ||
1763da12 FI |
151 | void nfs_init_cinfo_from_dreq(struct nfs_commit_info *cinfo, |
152 | struct nfs_direct_req *dreq) | |
153 | { | |
154 | cinfo->lock = &dreq->lock; | |
155 | cinfo->mds = &dreq->mds_cinfo; | |
156 | cinfo->ds = &dreq->ds_cinfo; | |
157 | cinfo->dreq = dreq; | |
158 | cinfo->completion_ops = &nfs_direct_commit_completion_ops; | |
159 | } | |
160 | ||
93619e59 | 161 | static inline struct nfs_direct_req *nfs_direct_req_alloc(void) |
1da177e4 | 162 | { |
93619e59 CL |
163 | struct nfs_direct_req *dreq; |
164 | ||
292f3eee | 165 | dreq = kmem_cache_zalloc(nfs_direct_cachep, GFP_KERNEL); |
93619e59 CL |
166 | if (!dreq) |
167 | return NULL; | |
168 | ||
169 | kref_init(&dreq->kref); | |
607f31e8 | 170 | kref_get(&dreq->kref); |
d72b7a6b | 171 | init_completion(&dreq->completion); |
1763da12 FI |
172 | INIT_LIST_HEAD(&dreq->mds_cinfo.list); |
173 | INIT_WORK(&dreq->work, nfs_direct_write_schedule_work); | |
15ce4a0c | 174 | spin_lock_init(&dreq->lock); |
93619e59 CL |
175 | |
176 | return dreq; | |
1da177e4 LT |
177 | } |
178 | ||
b4946ffb | 179 | static void nfs_direct_req_free(struct kref *kref) |
1da177e4 LT |
180 | { |
181 | struct nfs_direct_req *dreq = container_of(kref, struct nfs_direct_req, kref); | |
a8881f5a | 182 | |
f11ac8db TM |
183 | if (dreq->l_ctx != NULL) |
184 | nfs_put_lock_context(dreq->l_ctx); | |
a8881f5a TM |
185 | if (dreq->ctx != NULL) |
186 | put_nfs_open_context(dreq->ctx); | |
1da177e4 LT |
187 | kmem_cache_free(nfs_direct_cachep, dreq); |
188 | } | |
189 | ||
b4946ffb TM |
190 | static void nfs_direct_req_release(struct nfs_direct_req *dreq) |
191 | { | |
192 | kref_put(&dreq->kref, nfs_direct_req_free); | |
193 | } | |
194 | ||
6296556f PT |
195 | ssize_t nfs_dreq_bytes_left(struct nfs_direct_req *dreq) |
196 | { | |
197 | return dreq->bytes_left; | |
198 | } | |
199 | EXPORT_SYMBOL_GPL(nfs_dreq_bytes_left); | |
200 | ||
bc0fb201 CL |
201 | /* |
202 | * Collects and returns the final error value/byte-count. | |
203 | */ | |
204 | static ssize_t nfs_direct_wait(struct nfs_direct_req *dreq) | |
205 | { | |
15ce4a0c | 206 | ssize_t result = -EIOCBQUEUED; |
bc0fb201 CL |
207 | |
208 | /* Async requests don't wait here */ | |
209 | if (dreq->iocb) | |
210 | goto out; | |
211 | ||
150030b7 | 212 | result = wait_for_completion_killable(&dreq->completion); |
bc0fb201 CL |
213 | |
214 | if (!result) | |
15ce4a0c | 215 | result = dreq->error; |
bc0fb201 | 216 | if (!result) |
15ce4a0c | 217 | result = dreq->count; |
bc0fb201 CL |
218 | |
219 | out: | |
bc0fb201 CL |
220 | return (ssize_t) result; |
221 | } | |
222 | ||
63ab46ab | 223 | /* |
607f31e8 TM |
224 | * Synchronous I/O uses a stack-allocated iocb. Thus we can't trust |
225 | * the iocb is still valid here if this is a synchronous request. | |
63ab46ab CL |
226 | */ |
227 | static void nfs_direct_complete(struct nfs_direct_req *dreq) | |
228 | { | |
63ab46ab | 229 | if (dreq->iocb) { |
15ce4a0c | 230 | long res = (long) dreq->error; |
63ab46ab | 231 | if (!res) |
15ce4a0c | 232 | res = (long) dreq->count; |
63ab46ab | 233 | aio_complete(dreq->iocb, res, 0); |
d72b7a6b TM |
234 | } |
235 | complete_all(&dreq->completion); | |
63ab46ab | 236 | |
b4946ffb | 237 | nfs_direct_req_release(dreq); |
63ab46ab CL |
238 | } |
239 | ||
1385b811 | 240 | static void nfs_direct_readpage_release(struct nfs_page *req) |
1da177e4 | 241 | { |
584aa810 FI |
242 | dprintk("NFS: direct read done (%s/%lld %d@%lld)\n", |
243 | req->wb_context->dentry->d_inode->i_sb->s_id, | |
244 | (long long)NFS_FILEID(req->wb_context->dentry->d_inode), | |
245 | req->wb_bytes, | |
246 | (long long)req_offset(req)); | |
247 | nfs_release_request(req); | |
fdd1e74c TM |
248 | } |
249 | ||
584aa810 | 250 | static void nfs_direct_read_completion(struct nfs_pgio_header *hdr) |
fdd1e74c | 251 | { |
584aa810 FI |
252 | unsigned long bytes = 0; |
253 | struct nfs_direct_req *dreq = hdr->dreq; | |
fdd1e74c | 254 | |
584aa810 FI |
255 | if (test_bit(NFS_IOHDR_REDO, &hdr->flags)) |
256 | goto out_put; | |
15ce4a0c CL |
257 | |
258 | spin_lock(&dreq->lock); | |
584aa810 FI |
259 | if (test_bit(NFS_IOHDR_ERROR, &hdr->flags) && (hdr->good_bytes == 0)) |
260 | dreq->error = hdr->error; | |
261 | else | |
262 | dreq->count += hdr->good_bytes; | |
263 | spin_unlock(&dreq->lock); | |
264 | ||
4bd8b010 TM |
265 | while (!list_empty(&hdr->pages)) { |
266 | struct nfs_page *req = nfs_list_entry(hdr->pages.next); | |
267 | struct page *page = req->wb_page; | |
268 | ||
be7e9858 JL |
269 | if (!PageCompound(page) && bytes < hdr->good_bytes) |
270 | set_page_dirty(page); | |
4bd8b010 TM |
271 | bytes += req->wb_bytes; |
272 | nfs_list_remove_request(req); | |
273 | nfs_direct_readpage_release(req); | |
d4a8f367 | 274 | } |
584aa810 | 275 | out_put: |
607f31e8 TM |
276 | if (put_dreq(dreq)) |
277 | nfs_direct_complete(dreq); | |
584aa810 | 278 | hdr->release(hdr); |
1da177e4 LT |
279 | } |
280 | ||
3e9e0ca3 | 281 | static void nfs_read_sync_pgio_error(struct list_head *head) |
cd841605 | 282 | { |
584aa810 | 283 | struct nfs_page *req; |
cd841605 | 284 | |
584aa810 FI |
285 | while (!list_empty(head)) { |
286 | req = nfs_list_entry(head->next); | |
287 | nfs_list_remove_request(req); | |
288 | nfs_release_request(req); | |
289 | } | |
cd841605 FI |
290 | } |
291 | ||
584aa810 FI |
292 | static void nfs_direct_pgio_init(struct nfs_pgio_header *hdr) |
293 | { | |
294 | get_dreq(hdr->dreq); | |
295 | } | |
296 | ||
297 | static const struct nfs_pgio_completion_ops nfs_direct_read_completion_ops = { | |
3e9e0ca3 | 298 | .error_cleanup = nfs_read_sync_pgio_error, |
584aa810 FI |
299 | .init_hdr = nfs_direct_pgio_init, |
300 | .completion = nfs_direct_read_completion, | |
301 | }; | |
302 | ||
d4cc948b | 303 | /* |
607f31e8 TM |
304 | * For each rsize'd chunk of the user's buffer, dispatch an NFS READ |
305 | * operation. If nfs_readdata_alloc() or get_user_pages() fails, | |
306 | * bail and stop sending more reads. Read length accounting is | |
307 | * handled automatically by nfs_direct_read_result(). Otherwise, if | |
308 | * no requests have been sent, just return an error. | |
1da177e4 | 309 | */ |
584aa810 | 310 | static ssize_t nfs_direct_read_schedule_segment(struct nfs_pageio_descriptor *desc, |
02fe4946 | 311 | const struct iovec *iov, |
a564b8f0 | 312 | loff_t pos, bool uio) |
1da177e4 | 313 | { |
584aa810 | 314 | struct nfs_direct_req *dreq = desc->pg_dreq; |
a8881f5a | 315 | struct nfs_open_context *ctx = dreq->ctx; |
3d4ff43d | 316 | struct inode *inode = ctx->dentry->d_inode; |
02fe4946 CL |
317 | unsigned long user_addr = (unsigned long)iov->iov_base; |
318 | size_t count = iov->iov_len; | |
5dd602f2 | 319 | size_t rsize = NFS_SERVER(inode)->rsize; |
607f31e8 TM |
320 | unsigned int pgbase; |
321 | int result; | |
322 | ssize_t started = 0; | |
584aa810 FI |
323 | struct page **pagevec = NULL; |
324 | unsigned int npages; | |
607f31e8 | 325 | |
1da177e4 | 326 | do { |
5dd602f2 | 327 | size_t bytes; |
584aa810 | 328 | int i; |
1da177e4 | 329 | |
e9f7bee1 | 330 | pgbase = user_addr & ~PAGE_MASK; |
bf5fc402 | 331 | bytes = min(max_t(size_t, rsize, PAGE_SIZE), count); |
e9f7bee1 | 332 | |
607f31e8 | 333 | result = -ENOMEM; |
584aa810 FI |
334 | npages = nfs_page_array_len(pgbase, bytes); |
335 | if (!pagevec) | |
336 | pagevec = kmalloc(npages * sizeof(struct page *), | |
337 | GFP_KERNEL); | |
338 | if (!pagevec) | |
4db6e0b7 | 339 | break; |
a564b8f0 MG |
340 | if (uio) { |
341 | down_read(¤t->mm->mmap_sem); | |
342 | result = get_user_pages(current, current->mm, user_addr, | |
584aa810 | 343 | npages, 1, 0, pagevec, NULL); |
a564b8f0 MG |
344 | up_read(¤t->mm->mmap_sem); |
345 | if (result < 0) | |
346 | break; | |
347 | } else { | |
348 | WARN_ON(npages != 1); | |
349 | result = get_kernel_page(user_addr, 1, pagevec); | |
350 | if (WARN_ON(result != 1)) | |
351 | break; | |
352 | } | |
353 | ||
584aa810 | 354 | if ((unsigned)result < npages) { |
d9df8d6b TM |
355 | bytes = result * PAGE_SIZE; |
356 | if (bytes <= pgbase) { | |
584aa810 | 357 | nfs_direct_release_pages(pagevec, result); |
d9df8d6b TM |
358 | break; |
359 | } | |
360 | bytes -= pgbase; | |
584aa810 | 361 | npages = result; |
607f31e8 TM |
362 | } |
363 | ||
584aa810 FI |
364 | for (i = 0; i < npages; i++) { |
365 | struct nfs_page *req; | |
bf5fc402 | 366 | unsigned int req_len = min_t(size_t, bytes, PAGE_SIZE - pgbase); |
584aa810 FI |
367 | /* XXX do we need to do the eof zeroing found in async_filler? */ |
368 | req = nfs_create_request(dreq->ctx, dreq->inode, | |
369 | pagevec[i], | |
370 | pgbase, req_len); | |
371 | if (IS_ERR(req)) { | |
584aa810 FI |
372 | result = PTR_ERR(req); |
373 | break; | |
374 | } | |
375 | req->wb_index = pos >> PAGE_SHIFT; | |
376 | req->wb_offset = pos & ~PAGE_MASK; | |
377 | if (!nfs_pageio_add_request(desc, req)) { | |
378 | result = desc->pg_error; | |
379 | nfs_release_request(req); | |
584aa810 FI |
380 | break; |
381 | } | |
382 | pgbase = 0; | |
383 | bytes -= req_len; | |
384 | started += req_len; | |
385 | user_addr += req_len; | |
386 | pos += req_len; | |
387 | count -= req_len; | |
35754bc0 | 388 | dreq->bytes_left -= req_len; |
584aa810 | 389 | } |
6d74743b TM |
390 | /* The nfs_page now hold references to these pages */ |
391 | nfs_direct_release_pages(pagevec, npages); | |
71e8cc00 | 392 | } while (count != 0 && result >= 0); |
607f31e8 | 393 | |
584aa810 FI |
394 | kfree(pagevec); |
395 | ||
607f31e8 | 396 | if (started) |
c216fd70 | 397 | return started; |
607f31e8 | 398 | return result < 0 ? (ssize_t) result : -EFAULT; |
1da177e4 LT |
399 | } |
400 | ||
19f73787 CL |
401 | static ssize_t nfs_direct_read_schedule_iovec(struct nfs_direct_req *dreq, |
402 | const struct iovec *iov, | |
403 | unsigned long nr_segs, | |
a564b8f0 | 404 | loff_t pos, bool uio) |
19f73787 | 405 | { |
584aa810 | 406 | struct nfs_pageio_descriptor desc; |
19f73787 CL |
407 | ssize_t result = -EINVAL; |
408 | size_t requested_bytes = 0; | |
409 | unsigned long seg; | |
410 | ||
59948db3 | 411 | NFS_PROTO(dreq->inode)->read_pageio_init(&desc, dreq->inode, |
584aa810 | 412 | &nfs_direct_read_completion_ops); |
19f73787 | 413 | get_dreq(dreq); |
584aa810 | 414 | desc.pg_dreq = dreq; |
19f73787 CL |
415 | |
416 | for (seg = 0; seg < nr_segs; seg++) { | |
417 | const struct iovec *vec = &iov[seg]; | |
a564b8f0 | 418 | result = nfs_direct_read_schedule_segment(&desc, vec, pos, uio); |
19f73787 CL |
419 | if (result < 0) |
420 | break; | |
421 | requested_bytes += result; | |
422 | if ((size_t)result < vec->iov_len) | |
423 | break; | |
424 | pos += vec->iov_len; | |
425 | } | |
426 | ||
584aa810 FI |
427 | nfs_pageio_complete(&desc); |
428 | ||
839f7ad6 CL |
429 | /* |
430 | * If no bytes were started, return the error, and let the | |
431 | * generic layer handle the completion. | |
432 | */ | |
433 | if (requested_bytes == 0) { | |
434 | nfs_direct_req_release(dreq); | |
435 | return result < 0 ? result : -EIO; | |
436 | } | |
437 | ||
19f73787 CL |
438 | if (put_dreq(dreq)) |
439 | nfs_direct_complete(dreq); | |
839f7ad6 | 440 | return 0; |
19f73787 CL |
441 | } |
442 | ||
c216fd70 | 443 | static ssize_t nfs_direct_read(struct kiocb *iocb, const struct iovec *iov, |
a564b8f0 | 444 | unsigned long nr_segs, loff_t pos, bool uio) |
1da177e4 | 445 | { |
f11ac8db | 446 | ssize_t result = -ENOMEM; |
99514f8f | 447 | struct inode *inode = iocb->ki_filp->f_mapping->host; |
1da177e4 | 448 | struct nfs_direct_req *dreq; |
b3c54de6 | 449 | struct nfs_lock_context *l_ctx; |
1da177e4 | 450 | |
607f31e8 | 451 | dreq = nfs_direct_req_alloc(); |
f11ac8db TM |
452 | if (dreq == NULL) |
453 | goto out; | |
1da177e4 | 454 | |
91d5b470 | 455 | dreq->inode = inode; |
35754bc0 | 456 | dreq->bytes_left = iov_length(iov, nr_segs); |
cd3758e3 | 457 | dreq->ctx = get_nfs_open_context(nfs_file_open_context(iocb->ki_filp)); |
b3c54de6 TM |
458 | l_ctx = nfs_get_lock_context(dreq->ctx); |
459 | if (IS_ERR(l_ctx)) { | |
460 | result = PTR_ERR(l_ctx); | |
f11ac8db | 461 | goto out_release; |
b3c54de6 TM |
462 | } |
463 | dreq->l_ctx = l_ctx; | |
487b8372 CL |
464 | if (!is_sync_kiocb(iocb)) |
465 | dreq->iocb = iocb; | |
1da177e4 | 466 | |
7acdb026 | 467 | NFS_I(inode)->read_io += iov_length(iov, nr_segs); |
a564b8f0 | 468 | result = nfs_direct_read_schedule_iovec(dreq, iov, nr_segs, pos, uio); |
607f31e8 TM |
469 | if (!result) |
470 | result = nfs_direct_wait(dreq); | |
f11ac8db | 471 | out_release: |
b4946ffb | 472 | nfs_direct_req_release(dreq); |
f11ac8db | 473 | out: |
1da177e4 LT |
474 | return result; |
475 | } | |
476 | ||
1d59d61f TM |
477 | static void nfs_inode_dio_write_done(struct inode *inode) |
478 | { | |
479 | nfs_zap_mapping(inode, inode->i_mapping); | |
480 | inode_dio_done(inode); | |
481 | } | |
482 | ||
89d77c8f | 483 | #if IS_ENABLED(CONFIG_NFS_V3) || IS_ENABLED(CONFIG_NFS_V4) |
fad61490 TM |
484 | static void nfs_direct_write_reschedule(struct nfs_direct_req *dreq) |
485 | { | |
1763da12 FI |
486 | struct nfs_pageio_descriptor desc; |
487 | struct nfs_page *req, *tmp; | |
488 | LIST_HEAD(reqs); | |
489 | struct nfs_commit_info cinfo; | |
490 | LIST_HEAD(failed); | |
491 | ||
492 | nfs_init_cinfo_from_dreq(&cinfo, dreq); | |
493 | pnfs_recover_commit_reqs(dreq->inode, &reqs, &cinfo); | |
494 | spin_lock(cinfo.lock); | |
495 | nfs_scan_commit_list(&cinfo.mds->list, &reqs, &cinfo, 0); | |
496 | spin_unlock(cinfo.lock); | |
1da177e4 | 497 | |
fad61490 | 498 | dreq->count = 0; |
607f31e8 TM |
499 | get_dreq(dreq); |
500 | ||
c95908e4 | 501 | NFS_PROTO(dreq->inode)->write_pageio_init(&desc, dreq->inode, FLUSH_STABLE, |
1763da12 FI |
502 | &nfs_direct_write_completion_ops); |
503 | desc.pg_dreq = dreq; | |
fedb595c | 504 | |
1763da12 FI |
505 | list_for_each_entry_safe(req, tmp, &reqs, wb_list) { |
506 | if (!nfs_pageio_add_request(&desc, req)) { | |
4035c248 | 507 | nfs_list_remove_request(req); |
1763da12 FI |
508 | nfs_list_add_request(req, &failed); |
509 | spin_lock(cinfo.lock); | |
510 | dreq->flags = 0; | |
511 | dreq->error = -EIO; | |
512 | spin_unlock(cinfo.lock); | |
513 | } | |
5a695da2 | 514 | nfs_release_request(req); |
1763da12 FI |
515 | } |
516 | nfs_pageio_complete(&desc); | |
fad61490 | 517 | |
4035c248 TM |
518 | while (!list_empty(&failed)) { |
519 | req = nfs_list_entry(failed.next); | |
520 | nfs_list_remove_request(req); | |
1d1afcbc | 521 | nfs_unlock_and_release_request(req); |
4035c248 | 522 | } |
fad61490 | 523 | |
1763da12 FI |
524 | if (put_dreq(dreq)) |
525 | nfs_direct_write_complete(dreq, dreq->inode); | |
c9d8f89d TM |
526 | } |
527 | ||
1763da12 | 528 | static void nfs_direct_commit_complete(struct nfs_commit_data *data) |
c9d8f89d | 529 | { |
0b7c0153 | 530 | struct nfs_direct_req *dreq = data->dreq; |
1763da12 FI |
531 | struct nfs_commit_info cinfo; |
532 | struct nfs_page *req; | |
c9d8f89d TM |
533 | int status = data->task.tk_status; |
534 | ||
1763da12 | 535 | nfs_init_cinfo_from_dreq(&cinfo, dreq); |
c9d8f89d | 536 | if (status < 0) { |
60fa3f76 | 537 | dprintk("NFS: %5u commit failed with error %d.\n", |
1763da12 | 538 | data->task.tk_pid, status); |
fad61490 | 539 | dreq->flags = NFS_ODIRECT_RESCHED_WRITES; |
60fa3f76 | 540 | } else if (memcmp(&dreq->verf, &data->verf, sizeof(data->verf))) { |
c9d8f89d | 541 | dprintk("NFS: %5u commit verify failed\n", data->task.tk_pid); |
fad61490 | 542 | dreq->flags = NFS_ODIRECT_RESCHED_WRITES; |
1da177e4 LT |
543 | } |
544 | ||
c9d8f89d | 545 | dprintk("NFS: %5u commit returned %d\n", data->task.tk_pid, status); |
1763da12 FI |
546 | while (!list_empty(&data->pages)) { |
547 | req = nfs_list_entry(data->pages.next); | |
548 | nfs_list_remove_request(req); | |
549 | if (dreq->flags == NFS_ODIRECT_RESCHED_WRITES) { | |
550 | /* Note the rewrite will go through mds */ | |
551 | nfs_mark_request_commit(req, NULL, &cinfo); | |
906369e4 FI |
552 | } else |
553 | nfs_release_request(req); | |
1d1afcbc | 554 | nfs_unlock_and_release_request(req); |
1763da12 FI |
555 | } |
556 | ||
557 | if (atomic_dec_and_test(&cinfo.mds->rpcs_out)) | |
558 | nfs_direct_write_complete(dreq, data->inode); | |
1da177e4 LT |
559 | } |
560 | ||
1763da12 FI |
561 | static void nfs_direct_error_cleanup(struct nfs_inode *nfsi) |
562 | { | |
563 | /* There is no lock to clear */ | |
564 | } | |
565 | ||
566 | static const struct nfs_commit_completion_ops nfs_direct_commit_completion_ops = { | |
567 | .completion = nfs_direct_commit_complete, | |
568 | .error_cleanup = nfs_direct_error_cleanup, | |
fad61490 TM |
569 | }; |
570 | ||
571 | static void nfs_direct_commit_schedule(struct nfs_direct_req *dreq) | |
1da177e4 | 572 | { |
1763da12 FI |
573 | int res; |
574 | struct nfs_commit_info cinfo; | |
575 | LIST_HEAD(mds_list); | |
576 | ||
577 | nfs_init_cinfo_from_dreq(&cinfo, dreq); | |
578 | nfs_scan_commit(dreq->inode, &mds_list, &cinfo); | |
579 | res = nfs_generic_commit_list(dreq->inode, &mds_list, 0, &cinfo); | |
580 | if (res < 0) /* res == -ENOMEM */ | |
581 | nfs_direct_write_reschedule(dreq); | |
fad61490 | 582 | } |
1da177e4 | 583 | |
1763da12 | 584 | static void nfs_direct_write_schedule_work(struct work_struct *work) |
fad61490 | 585 | { |
1763da12 | 586 | struct nfs_direct_req *dreq = container_of(work, struct nfs_direct_req, work); |
fad61490 | 587 | int flags = dreq->flags; |
1da177e4 | 588 | |
fad61490 TM |
589 | dreq->flags = 0; |
590 | switch (flags) { | |
591 | case NFS_ODIRECT_DO_COMMIT: | |
592 | nfs_direct_commit_schedule(dreq); | |
1da177e4 | 593 | break; |
fad61490 TM |
594 | case NFS_ODIRECT_RESCHED_WRITES: |
595 | nfs_direct_write_reschedule(dreq); | |
596 | break; | |
597 | default: | |
1d59d61f | 598 | nfs_inode_dio_write_done(dreq->inode); |
fad61490 TM |
599 | nfs_direct_complete(dreq); |
600 | } | |
601 | } | |
1da177e4 | 602 | |
1763da12 | 603 | static void nfs_direct_write_complete(struct nfs_direct_req *dreq, struct inode *inode) |
fad61490 | 604 | { |
1763da12 | 605 | schedule_work(&dreq->work); /* Calls nfs_direct_write_schedule_work */ |
fad61490 | 606 | } |
1763da12 | 607 | |
fad61490 | 608 | #else |
24fc9211 BS |
609 | static void nfs_direct_write_schedule_work(struct work_struct *work) |
610 | { | |
611 | } | |
1da177e4 | 612 | |
fad61490 TM |
613 | static void nfs_direct_write_complete(struct nfs_direct_req *dreq, struct inode *inode) |
614 | { | |
1d59d61f | 615 | nfs_inode_dio_write_done(inode); |
fad61490 TM |
616 | nfs_direct_complete(dreq); |
617 | } | |
618 | #endif | |
1da177e4 | 619 | |
c9d8f89d TM |
620 | /* |
621 | * NB: Return the value of the first error return code. Subsequent | |
622 | * errors after the first one are ignored. | |
623 | */ | |
462d5b32 | 624 | /* |
607f31e8 TM |
625 | * For each wsize'd chunk of the user's buffer, dispatch an NFS WRITE |
626 | * operation. If nfs_writedata_alloc() or get_user_pages() fails, | |
627 | * bail and stop sending more writes. Write length accounting is | |
628 | * handled automatically by nfs_direct_write_result(). Otherwise, if | |
629 | * no requests have been sent, just return an error. | |
462d5b32 | 630 | */ |
1763da12 | 631 | static ssize_t nfs_direct_write_schedule_segment(struct nfs_pageio_descriptor *desc, |
02fe4946 | 632 | const struct iovec *iov, |
a564b8f0 | 633 | loff_t pos, bool uio) |
462d5b32 | 634 | { |
1763da12 | 635 | struct nfs_direct_req *dreq = desc->pg_dreq; |
a8881f5a | 636 | struct nfs_open_context *ctx = dreq->ctx; |
3d4ff43d | 637 | struct inode *inode = ctx->dentry->d_inode; |
02fe4946 CL |
638 | unsigned long user_addr = (unsigned long)iov->iov_base; |
639 | size_t count = iov->iov_len; | |
462d5b32 | 640 | size_t wsize = NFS_SERVER(inode)->wsize; |
607f31e8 TM |
641 | unsigned int pgbase; |
642 | int result; | |
643 | ssize_t started = 0; | |
1763da12 FI |
644 | struct page **pagevec = NULL; |
645 | unsigned int npages; | |
82b145c5 | 646 | |
1da177e4 | 647 | do { |
462d5b32 | 648 | size_t bytes; |
1763da12 | 649 | int i; |
462d5b32 | 650 | |
e9f7bee1 | 651 | pgbase = user_addr & ~PAGE_MASK; |
bf5fc402 | 652 | bytes = min(max_t(size_t, wsize, PAGE_SIZE), count); |
e9f7bee1 | 653 | |
607f31e8 | 654 | result = -ENOMEM; |
1763da12 FI |
655 | npages = nfs_page_array_len(pgbase, bytes); |
656 | if (!pagevec) | |
657 | pagevec = kmalloc(npages * sizeof(struct page *), GFP_KERNEL); | |
658 | if (!pagevec) | |
607f31e8 TM |
659 | break; |
660 | ||
a564b8f0 MG |
661 | if (uio) { |
662 | down_read(¤t->mm->mmap_sem); | |
663 | result = get_user_pages(current, current->mm, user_addr, | |
664 | npages, 0, 0, pagevec, NULL); | |
665 | up_read(¤t->mm->mmap_sem); | |
666 | if (result < 0) | |
667 | break; | |
668 | } else { | |
669 | WARN_ON(npages != 1); | |
670 | result = get_kernel_page(user_addr, 0, pagevec); | |
671 | if (WARN_ON(result != 1)) | |
672 | break; | |
673 | } | |
1763da12 FI |
674 | |
675 | if ((unsigned)result < npages) { | |
d9df8d6b TM |
676 | bytes = result * PAGE_SIZE; |
677 | if (bytes <= pgbase) { | |
1763da12 | 678 | nfs_direct_release_pages(pagevec, result); |
d9df8d6b TM |
679 | break; |
680 | } | |
681 | bytes -= pgbase; | |
1763da12 | 682 | npages = result; |
607f31e8 TM |
683 | } |
684 | ||
1763da12 FI |
685 | for (i = 0; i < npages; i++) { |
686 | struct nfs_page *req; | |
bf5fc402 | 687 | unsigned int req_len = min_t(size_t, bytes, PAGE_SIZE - pgbase); |
1da177e4 | 688 | |
1763da12 FI |
689 | req = nfs_create_request(dreq->ctx, dreq->inode, |
690 | pagevec[i], | |
691 | pgbase, req_len); | |
692 | if (IS_ERR(req)) { | |
1763da12 FI |
693 | result = PTR_ERR(req); |
694 | break; | |
695 | } | |
696 | nfs_lock_request(req); | |
697 | req->wb_index = pos >> PAGE_SHIFT; | |
698 | req->wb_offset = pos & ~PAGE_MASK; | |
699 | if (!nfs_pageio_add_request(desc, req)) { | |
700 | result = desc->pg_error; | |
1d1afcbc | 701 | nfs_unlock_and_release_request(req); |
71e8cc00 | 702 | break; |
1763da12 FI |
703 | } |
704 | pgbase = 0; | |
705 | bytes -= req_len; | |
706 | started += req_len; | |
707 | user_addr += req_len; | |
708 | pos += req_len; | |
709 | count -= req_len; | |
35754bc0 | 710 | dreq->bytes_left -= req_len; |
1763da12 | 711 | } |
6d74743b TM |
712 | /* The nfs_page now hold references to these pages */ |
713 | nfs_direct_release_pages(pagevec, npages); | |
71e8cc00 | 714 | } while (count != 0 && result >= 0); |
607f31e8 | 715 | |
1763da12 FI |
716 | kfree(pagevec); |
717 | ||
607f31e8 | 718 | if (started) |
c216fd70 | 719 | return started; |
607f31e8 | 720 | return result < 0 ? (ssize_t) result : -EFAULT; |
462d5b32 | 721 | } |
1da177e4 | 722 | |
1763da12 FI |
723 | static void nfs_direct_write_completion(struct nfs_pgio_header *hdr) |
724 | { | |
725 | struct nfs_direct_req *dreq = hdr->dreq; | |
726 | struct nfs_commit_info cinfo; | |
727 | int bit = -1; | |
728 | struct nfs_page *req = nfs_list_entry(hdr->pages.next); | |
729 | ||
730 | if (test_bit(NFS_IOHDR_REDO, &hdr->flags)) | |
731 | goto out_put; | |
732 | ||
733 | nfs_init_cinfo_from_dreq(&cinfo, dreq); | |
734 | ||
735 | spin_lock(&dreq->lock); | |
736 | ||
737 | if (test_bit(NFS_IOHDR_ERROR, &hdr->flags)) { | |
738 | dreq->flags = 0; | |
739 | dreq->error = hdr->error; | |
740 | } | |
741 | if (dreq->error != 0) | |
742 | bit = NFS_IOHDR_ERROR; | |
743 | else { | |
744 | dreq->count += hdr->good_bytes; | |
745 | if (test_bit(NFS_IOHDR_NEED_RESCHED, &hdr->flags)) { | |
746 | dreq->flags = NFS_ODIRECT_RESCHED_WRITES; | |
747 | bit = NFS_IOHDR_NEED_RESCHED; | |
748 | } else if (test_bit(NFS_IOHDR_NEED_COMMIT, &hdr->flags)) { | |
749 | if (dreq->flags == NFS_ODIRECT_RESCHED_WRITES) | |
750 | bit = NFS_IOHDR_NEED_RESCHED; | |
751 | else if (dreq->flags == 0) { | |
9bce008b | 752 | memcpy(&dreq->verf, hdr->verf, |
1763da12 FI |
753 | sizeof(dreq->verf)); |
754 | bit = NFS_IOHDR_NEED_COMMIT; | |
755 | dreq->flags = NFS_ODIRECT_DO_COMMIT; | |
756 | } else if (dreq->flags == NFS_ODIRECT_DO_COMMIT) { | |
9bce008b | 757 | if (memcmp(&dreq->verf, hdr->verf, sizeof(dreq->verf))) { |
1763da12 FI |
758 | dreq->flags = NFS_ODIRECT_RESCHED_WRITES; |
759 | bit = NFS_IOHDR_NEED_RESCHED; | |
760 | } else | |
761 | bit = NFS_IOHDR_NEED_COMMIT; | |
762 | } | |
763 | } | |
764 | } | |
765 | spin_unlock(&dreq->lock); | |
766 | ||
767 | while (!list_empty(&hdr->pages)) { | |
768 | req = nfs_list_entry(hdr->pages.next); | |
769 | nfs_list_remove_request(req); | |
770 | switch (bit) { | |
771 | case NFS_IOHDR_NEED_RESCHED: | |
772 | case NFS_IOHDR_NEED_COMMIT: | |
04277086 | 773 | kref_get(&req->wb_kref); |
1763da12 | 774 | nfs_mark_request_commit(req, hdr->lseg, &cinfo); |
1763da12 | 775 | } |
1d1afcbc | 776 | nfs_unlock_and_release_request(req); |
1763da12 FI |
777 | } |
778 | ||
779 | out_put: | |
780 | if (put_dreq(dreq)) | |
781 | nfs_direct_write_complete(dreq, hdr->inode); | |
782 | hdr->release(hdr); | |
783 | } | |
784 | ||
3e9e0ca3 TM |
785 | static void nfs_write_sync_pgio_error(struct list_head *head) |
786 | { | |
787 | struct nfs_page *req; | |
788 | ||
789 | while (!list_empty(head)) { | |
790 | req = nfs_list_entry(head->next); | |
791 | nfs_list_remove_request(req); | |
1d1afcbc | 792 | nfs_unlock_and_release_request(req); |
3e9e0ca3 TM |
793 | } |
794 | } | |
795 | ||
1763da12 | 796 | static const struct nfs_pgio_completion_ops nfs_direct_write_completion_ops = { |
3e9e0ca3 | 797 | .error_cleanup = nfs_write_sync_pgio_error, |
1763da12 FI |
798 | .init_hdr = nfs_direct_pgio_init, |
799 | .completion = nfs_direct_write_completion, | |
800 | }; | |
801 | ||
19f73787 CL |
802 | static ssize_t nfs_direct_write_schedule_iovec(struct nfs_direct_req *dreq, |
803 | const struct iovec *iov, | |
804 | unsigned long nr_segs, | |
a564b8f0 | 805 | loff_t pos, bool uio) |
19f73787 | 806 | { |
1763da12 | 807 | struct nfs_pageio_descriptor desc; |
1d59d61f | 808 | struct inode *inode = dreq->inode; |
19f73787 CL |
809 | ssize_t result = 0; |
810 | size_t requested_bytes = 0; | |
811 | unsigned long seg; | |
812 | ||
c95908e4 | 813 | NFS_PROTO(inode)->write_pageio_init(&desc, inode, FLUSH_COND_STABLE, |
1763da12 FI |
814 | &nfs_direct_write_completion_ops); |
815 | desc.pg_dreq = dreq; | |
19f73787 | 816 | get_dreq(dreq); |
1d59d61f | 817 | atomic_inc(&inode->i_dio_count); |
19f73787 | 818 | |
7acdb026 | 819 | NFS_I(dreq->inode)->write_io += iov_length(iov, nr_segs); |
19f73787 CL |
820 | for (seg = 0; seg < nr_segs; seg++) { |
821 | const struct iovec *vec = &iov[seg]; | |
a564b8f0 | 822 | result = nfs_direct_write_schedule_segment(&desc, vec, pos, uio); |
19f73787 CL |
823 | if (result < 0) |
824 | break; | |
825 | requested_bytes += result; | |
826 | if ((size_t)result < vec->iov_len) | |
827 | break; | |
828 | pos += vec->iov_len; | |
829 | } | |
1763da12 | 830 | nfs_pageio_complete(&desc); |
19f73787 | 831 | |
839f7ad6 CL |
832 | /* |
833 | * If no bytes were started, return the error, and let the | |
834 | * generic layer handle the completion. | |
835 | */ | |
836 | if (requested_bytes == 0) { | |
1d59d61f | 837 | inode_dio_done(inode); |
839f7ad6 CL |
838 | nfs_direct_req_release(dreq); |
839 | return result < 0 ? result : -EIO; | |
840 | } | |
841 | ||
19f73787 CL |
842 | if (put_dreq(dreq)) |
843 | nfs_direct_write_complete(dreq, dreq->inode); | |
839f7ad6 | 844 | return 0; |
19f73787 CL |
845 | } |
846 | ||
c216fd70 CL |
847 | static ssize_t nfs_direct_write(struct kiocb *iocb, const struct iovec *iov, |
848 | unsigned long nr_segs, loff_t pos, | |
a564b8f0 | 849 | size_t count, bool uio) |
462d5b32 | 850 | { |
f11ac8db | 851 | ssize_t result = -ENOMEM; |
c89f2ee5 | 852 | struct inode *inode = iocb->ki_filp->f_mapping->host; |
462d5b32 | 853 | struct nfs_direct_req *dreq; |
b3c54de6 | 854 | struct nfs_lock_context *l_ctx; |
1da177e4 | 855 | |
607f31e8 | 856 | dreq = nfs_direct_req_alloc(); |
462d5b32 | 857 | if (!dreq) |
f11ac8db | 858 | goto out; |
1da177e4 | 859 | |
c89f2ee5 | 860 | dreq->inode = inode; |
35754bc0 | 861 | dreq->bytes_left = count; |
cd3758e3 | 862 | dreq->ctx = get_nfs_open_context(nfs_file_open_context(iocb->ki_filp)); |
b3c54de6 TM |
863 | l_ctx = nfs_get_lock_context(dreq->ctx); |
864 | if (IS_ERR(l_ctx)) { | |
865 | result = PTR_ERR(l_ctx); | |
f11ac8db | 866 | goto out_release; |
b3c54de6 TM |
867 | } |
868 | dreq->l_ctx = l_ctx; | |
c89f2ee5 CL |
869 | if (!is_sync_kiocb(iocb)) |
870 | dreq->iocb = iocb; | |
1da177e4 | 871 | |
a564b8f0 | 872 | result = nfs_direct_write_schedule_iovec(dreq, iov, nr_segs, pos, uio); |
607f31e8 TM |
873 | if (!result) |
874 | result = nfs_direct_wait(dreq); | |
f11ac8db | 875 | out_release: |
b4946ffb | 876 | nfs_direct_req_release(dreq); |
f11ac8db | 877 | out: |
1da177e4 LT |
878 | return result; |
879 | } | |
880 | ||
881 | /** | |
882 | * nfs_file_direct_read - file direct read operation for NFS files | |
883 | * @iocb: target I/O control block | |
027445c3 BP |
884 | * @iov: vector of user buffers into which to read data |
885 | * @nr_segs: size of iov vector | |
88467055 | 886 | * @pos: byte offset in file where reading starts |
1da177e4 LT |
887 | * |
888 | * We use this function for direct reads instead of calling | |
889 | * generic_file_aio_read() in order to avoid gfar's check to see if | |
890 | * the request starts before the end of the file. For that check | |
891 | * to work, we must generate a GETATTR before each direct read, and | |
892 | * even then there is a window between the GETATTR and the subsequent | |
88467055 | 893 | * READ where the file size could change. Our preference is simply |
1da177e4 LT |
894 | * to do all reads the application wants, and the server will take |
895 | * care of managing the end of file boundary. | |
88467055 | 896 | * |
1da177e4 LT |
897 | * This function also eliminates unnecessarily updating the file's |
898 | * atime locally, as the NFS server sets the file's atime, and this | |
899 | * client must read the updated atime from the server back into its | |
900 | * cache. | |
901 | */ | |
027445c3 | 902 | ssize_t nfs_file_direct_read(struct kiocb *iocb, const struct iovec *iov, |
a564b8f0 | 903 | unsigned long nr_segs, loff_t pos, bool uio) |
1da177e4 LT |
904 | { |
905 | ssize_t retval = -EINVAL; | |
1da177e4 | 906 | struct file *file = iocb->ki_filp; |
1da177e4 | 907 | struct address_space *mapping = file->f_mapping; |
c216fd70 CL |
908 | size_t count; |
909 | ||
910 | count = iov_length(iov, nr_segs); | |
911 | nfs_add_stats(mapping->host, NFSIOS_DIRECTREADBYTES, count); | |
1da177e4 | 912 | |
6da24bc9 | 913 | dfprintk(FILE, "NFS: direct read(%s/%s, %zd@%Ld)\n", |
01cce933 JJS |
914 | file->f_path.dentry->d_parent->d_name.name, |
915 | file->f_path.dentry->d_name.name, | |
c216fd70 | 916 | count, (long long) pos); |
1da177e4 | 917 | |
1da177e4 LT |
918 | retval = 0; |
919 | if (!count) | |
920 | goto out; | |
921 | ||
29884df0 TM |
922 | retval = nfs_sync_mapping(mapping); |
923 | if (retval) | |
924 | goto out; | |
1da177e4 | 925 | |
7ec10f26 KK |
926 | task_io_account_read(count); |
927 | ||
a564b8f0 | 928 | retval = nfs_direct_read(iocb, iov, nr_segs, pos, uio); |
1da177e4 | 929 | if (retval > 0) |
0cdd80d0 | 930 | iocb->ki_pos = pos + retval; |
1da177e4 LT |
931 | |
932 | out: | |
933 | return retval; | |
934 | } | |
935 | ||
936 | /** | |
937 | * nfs_file_direct_write - file direct write operation for NFS files | |
938 | * @iocb: target I/O control block | |
027445c3 BP |
939 | * @iov: vector of user buffers from which to write data |
940 | * @nr_segs: size of iov vector | |
88467055 | 941 | * @pos: byte offset in file where writing starts |
1da177e4 LT |
942 | * |
943 | * We use this function for direct writes instead of calling | |
944 | * generic_file_aio_write() in order to avoid taking the inode | |
945 | * semaphore and updating the i_size. The NFS server will set | |
946 | * the new i_size and this client must read the updated size | |
947 | * back into its cache. We let the server do generic write | |
948 | * parameter checking and report problems. | |
949 | * | |
1da177e4 LT |
950 | * We eliminate local atime updates, see direct read above. |
951 | * | |
952 | * We avoid unnecessary page cache invalidations for normal cached | |
953 | * readers of this file. | |
954 | * | |
955 | * Note that O_APPEND is not supported for NFS direct writes, as there | |
956 | * is no atomic O_APPEND write facility in the NFS protocol. | |
957 | */ | |
027445c3 | 958 | ssize_t nfs_file_direct_write(struct kiocb *iocb, const struct iovec *iov, |
a564b8f0 | 959 | unsigned long nr_segs, loff_t pos, bool uio) |
1da177e4 | 960 | { |
070ea602 | 961 | ssize_t retval = -EINVAL; |
1da177e4 | 962 | struct file *file = iocb->ki_filp; |
1da177e4 | 963 | struct address_space *mapping = file->f_mapping; |
c216fd70 | 964 | size_t count; |
1da177e4 | 965 | |
c216fd70 CL |
966 | count = iov_length(iov, nr_segs); |
967 | nfs_add_stats(mapping->host, NFSIOS_DIRECTWRITTENBYTES, count); | |
968 | ||
6da24bc9 | 969 | dfprintk(FILE, "NFS: direct write(%s/%s, %zd@%Ld)\n", |
01cce933 JJS |
970 | file->f_path.dentry->d_parent->d_name.name, |
971 | file->f_path.dentry->d_name.name, | |
c216fd70 | 972 | count, (long long) pos); |
027445c3 | 973 | |
ce1a8e67 CL |
974 | retval = generic_write_checks(file, &pos, &count, 0); |
975 | if (retval) | |
1da177e4 | 976 | goto out; |
ce1a8e67 CL |
977 | |
978 | retval = -EINVAL; | |
979 | if ((ssize_t) count < 0) | |
1da177e4 | 980 | goto out; |
1da177e4 LT |
981 | retval = 0; |
982 | if (!count) | |
983 | goto out; | |
ce1a8e67 | 984 | |
29884df0 TM |
985 | retval = nfs_sync_mapping(mapping); |
986 | if (retval) | |
987 | goto out; | |
1da177e4 | 988 | |
7ec10f26 KK |
989 | task_io_account_write(count); |
990 | ||
a564b8f0 | 991 | retval = nfs_direct_write(iocb, iov, nr_segs, pos, count, uio); |
1763da12 FI |
992 | if (retval > 0) { |
993 | struct inode *inode = mapping->host; | |
9eafa8cc | 994 | |
ce1a8e67 | 995 | iocb->ki_pos = pos + retval; |
1763da12 FI |
996 | spin_lock(&inode->i_lock); |
997 | if (i_size_read(inode) < iocb->ki_pos) | |
998 | i_size_write(inode, iocb->ki_pos); | |
999 | spin_unlock(&inode->i_lock); | |
1000 | } | |
1da177e4 LT |
1001 | out: |
1002 | return retval; | |
1003 | } | |
1004 | ||
88467055 CL |
1005 | /** |
1006 | * nfs_init_directcache - create a slab cache for nfs_direct_req structures | |
1007 | * | |
1008 | */ | |
f7b422b1 | 1009 | int __init nfs_init_directcache(void) |
1da177e4 LT |
1010 | { |
1011 | nfs_direct_cachep = kmem_cache_create("nfs_direct_cache", | |
1012 | sizeof(struct nfs_direct_req), | |
fffb60f9 PJ |
1013 | 0, (SLAB_RECLAIM_ACCOUNT| |
1014 | SLAB_MEM_SPREAD), | |
20c2df83 | 1015 | NULL); |
1da177e4 LT |
1016 | if (nfs_direct_cachep == NULL) |
1017 | return -ENOMEM; | |
1018 | ||
1019 | return 0; | |
1020 | } | |
1021 | ||
88467055 | 1022 | /** |
f7b422b1 | 1023 | * nfs_destroy_directcache - destroy the slab cache for nfs_direct_req structures |
88467055 CL |
1024 | * |
1025 | */ | |
266bee88 | 1026 | void nfs_destroy_directcache(void) |
1da177e4 | 1027 | { |
1a1d92c1 | 1028 | kmem_cache_destroy(nfs_direct_cachep); |
1da177e4 | 1029 | } |