2 * linux/fs/nfs/direct.c
4 * Copyright (C) 2003 by Chuck Lever <cel@netapp.com>
6 * High-performance uncached I/O for the Linux NFS client
8 * There are important applications whose performance or correctness
9 * depends on uncached access to file data. Database clusters
10 * (multiple copies of the same instance running on separate hosts)
11 * implement their own cache coherency protocol that subsumes file
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
15 * need to cache the contents of a file.
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
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.
29 * Designed by Jeff Kimmel, Chuck Lever, and Trond Myklebust, with
30 * help from Andrew Morton.
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
37 * 04 May 2005 support O_DIRECT with aio --cel
41 #include <linux/errno.h>
42 #include <linux/sched.h>
43 #include <linux/kernel.h>
44 #include <linux/file.h>
45 #include <linux/pagemap.h>
46 #include <linux/kref.h>
47 #include <linux/slab.h>
48 #include <linux/task_io_accounting_ops.h>
50 #include <linux/nfs_fs.h>
51 #include <linux/nfs_page.h>
52 #include <linux/sunrpc/clnt.h>
54 #include <asm/uaccess.h>
55 #include <linux/atomic.h>
61 #define NFSDBG_FACILITY NFSDBG_VFS
63 static struct kmem_cache
*nfs_direct_cachep
;
66 * This represents a set of asynchronous requests that we're waiting on
68 struct nfs_direct_req
{
69 struct kref kref
; /* release manager */
72 struct nfs_open_context
*ctx
; /* file open context info */
73 struct nfs_lock_context
*l_ctx
; /* Lock context info */
74 struct kiocb
* iocb
; /* controlling i/o request */
75 struct inode
* inode
; /* target file of i/o */
77 /* completion state */
78 atomic_t io_count
; /* i/os we're waiting for */
79 spinlock_t lock
; /* protect completion state */
80 ssize_t count
, /* bytes actually processed */
81 error
; /* any reported error */
82 struct completion completion
; /* wait for i/o completion */
85 struct nfs_mds_commit_info mds_cinfo
; /* Storage for cinfo */
86 struct pnfs_ds_commit_info ds_cinfo
; /* Storage for cinfo */
87 struct work_struct work
;
89 #define NFS_ODIRECT_DO_COMMIT (1) /* an unstable reply was received */
90 #define NFS_ODIRECT_RESCHED_WRITES (2) /* write verification failed */
91 struct nfs_writeverf verf
; /* unstable write verifier */
94 static const struct nfs_pgio_completion_ops nfs_direct_write_completion_ops
;
95 static const struct nfs_commit_completion_ops nfs_direct_commit_completion_ops
;
96 static void nfs_direct_write_complete(struct nfs_direct_req
*dreq
, struct inode
*inode
);
97 static void nfs_direct_write_schedule_work(struct work_struct
*work
);
99 static inline void get_dreq(struct nfs_direct_req
*dreq
)
101 atomic_inc(&dreq
->io_count
);
104 static inline int put_dreq(struct nfs_direct_req
*dreq
)
106 return atomic_dec_and_test(&dreq
->io_count
);
110 * nfs_direct_IO - NFS address space operation for direct I/O
111 * @rw: direction (read or write)
112 * @iocb: target I/O control block
113 * @iov: array of vectors that define I/O buffer
114 * @pos: offset in file to begin the operation
115 * @nr_segs: size of iovec array
117 * The presence of this routine in the address space ops vector means
118 * the NFS client supports direct I/O. However, for most direct IO, we
119 * shunt off direct read and write requests before the VFS gets them,
120 * so this method is only ever called for swap.
122 ssize_t
nfs_direct_IO(int rw
, struct kiocb
*iocb
, const struct iovec
*iov
, loff_t pos
, unsigned long nr_segs
)
124 #ifndef CONFIG_NFS_SWAP
125 dprintk("NFS: nfs_direct_IO (%s) off/no(%Ld/%lu) EINVAL\n",
126 iocb
->ki_filp
->f_path
.dentry
->d_name
.name
,
127 (long long) pos
, nr_segs
);
131 VM_BUG_ON(iocb
->ki_left
!= PAGE_SIZE
);
132 VM_BUG_ON(iocb
->ki_nbytes
!= PAGE_SIZE
);
134 if (rw
== READ
|| rw
== KERNEL_READ
)
135 return nfs_file_direct_read(iocb
, iov
, nr_segs
, pos
,
136 rw
== READ
? true : false);
137 return nfs_file_direct_write(iocb
, iov
, nr_segs
, pos
,
138 rw
== WRITE
? true : false);
139 #endif /* CONFIG_NFS_SWAP */
142 static void nfs_direct_release_pages(struct page
**pages
, unsigned int npages
)
145 for (i
= 0; i
< npages
; i
++)
146 page_cache_release(pages
[i
]);
149 void nfs_init_cinfo_from_dreq(struct nfs_commit_info
*cinfo
,
150 struct nfs_direct_req
*dreq
)
152 cinfo
->lock
= &dreq
->lock
;
153 cinfo
->mds
= &dreq
->mds_cinfo
;
154 cinfo
->ds
= &dreq
->ds_cinfo
;
156 cinfo
->completion_ops
= &nfs_direct_commit_completion_ops
;
159 static inline struct nfs_direct_req
*nfs_direct_req_alloc(void)
161 struct nfs_direct_req
*dreq
;
163 dreq
= kmem_cache_zalloc(nfs_direct_cachep
, GFP_KERNEL
);
167 kref_init(&dreq
->kref
);
168 kref_get(&dreq
->kref
);
169 init_completion(&dreq
->completion
);
170 INIT_LIST_HEAD(&dreq
->mds_cinfo
.list
);
171 INIT_WORK(&dreq
->work
, nfs_direct_write_schedule_work
);
172 spin_lock_init(&dreq
->lock
);
177 static void nfs_direct_req_free(struct kref
*kref
)
179 struct nfs_direct_req
*dreq
= container_of(kref
, struct nfs_direct_req
, kref
);
181 if (dreq
->l_ctx
!= NULL
)
182 nfs_put_lock_context(dreq
->l_ctx
);
183 if (dreq
->ctx
!= NULL
)
184 put_nfs_open_context(dreq
->ctx
);
185 kmem_cache_free(nfs_direct_cachep
, dreq
);
188 static void nfs_direct_req_release(struct nfs_direct_req
*dreq
)
190 kref_put(&dreq
->kref
, nfs_direct_req_free
);
194 * Collects and returns the final error value/byte-count.
196 static ssize_t
nfs_direct_wait(struct nfs_direct_req
*dreq
)
198 ssize_t result
= -EIOCBQUEUED
;
200 /* Async requests don't wait here */
204 result
= wait_for_completion_killable(&dreq
->completion
);
207 result
= dreq
->error
;
209 result
= dreq
->count
;
212 return (ssize_t
) result
;
216 * Synchronous I/O uses a stack-allocated iocb. Thus we can't trust
217 * the iocb is still valid here if this is a synchronous request.
219 static void nfs_direct_complete(struct nfs_direct_req
*dreq
)
222 long res
= (long) dreq
->error
;
224 res
= (long) dreq
->count
;
225 aio_complete(dreq
->iocb
, res
, 0);
227 complete_all(&dreq
->completion
);
229 nfs_direct_req_release(dreq
);
232 static void nfs_direct_readpage_release(struct nfs_page
*req
)
234 dprintk("NFS: direct read done (%s/%lld %d@%lld)\n",
235 req
->wb_context
->dentry
->d_inode
->i_sb
->s_id
,
236 (long long)NFS_FILEID(req
->wb_context
->dentry
->d_inode
),
238 (long long)req_offset(req
));
239 nfs_release_request(req
);
242 static void nfs_direct_read_completion(struct nfs_pgio_header
*hdr
)
244 unsigned long bytes
= 0;
245 struct nfs_direct_req
*dreq
= hdr
->dreq
;
247 if (test_bit(NFS_IOHDR_REDO
, &hdr
->flags
))
250 spin_lock(&dreq
->lock
);
251 if (test_bit(NFS_IOHDR_ERROR
, &hdr
->flags
) && (hdr
->good_bytes
== 0))
252 dreq
->error
= hdr
->error
;
254 dreq
->count
+= hdr
->good_bytes
;
255 spin_unlock(&dreq
->lock
);
257 while (!list_empty(&hdr
->pages
)) {
258 struct nfs_page
*req
= nfs_list_entry(hdr
->pages
.next
);
259 struct page
*page
= req
->wb_page
;
261 if (test_bit(NFS_IOHDR_EOF
, &hdr
->flags
)) {
262 if (bytes
> hdr
->good_bytes
)
263 zero_user(page
, 0, PAGE_SIZE
);
264 else if (hdr
->good_bytes
- bytes
< PAGE_SIZE
)
265 zero_user_segment(page
,
266 hdr
->good_bytes
& ~PAGE_MASK
,
269 if (!PageCompound(page
)) {
270 if (test_bit(NFS_IOHDR_ERROR
, &hdr
->flags
)) {
271 if (bytes
< hdr
->good_bytes
)
272 set_page_dirty(page
);
274 set_page_dirty(page
);
276 bytes
+= req
->wb_bytes
;
277 nfs_list_remove_request(req
);
278 nfs_direct_readpage_release(req
);
282 nfs_direct_complete(dreq
);
286 static void nfs_read_sync_pgio_error(struct list_head
*head
)
288 struct nfs_page
*req
;
290 while (!list_empty(head
)) {
291 req
= nfs_list_entry(head
->next
);
292 nfs_list_remove_request(req
);
293 nfs_release_request(req
);
297 static void nfs_direct_pgio_init(struct nfs_pgio_header
*hdr
)
302 static const struct nfs_pgio_completion_ops nfs_direct_read_completion_ops
= {
303 .error_cleanup
= nfs_read_sync_pgio_error
,
304 .init_hdr
= nfs_direct_pgio_init
,
305 .completion
= nfs_direct_read_completion
,
309 * For each rsize'd chunk of the user's buffer, dispatch an NFS READ
310 * operation. If nfs_readdata_alloc() or get_user_pages() fails,
311 * bail and stop sending more reads. Read length accounting is
312 * handled automatically by nfs_direct_read_result(). Otherwise, if
313 * no requests have been sent, just return an error.
315 static ssize_t
nfs_direct_read_schedule_segment(struct nfs_pageio_descriptor
*desc
,
316 const struct iovec
*iov
,
317 loff_t pos
, bool uio
)
319 struct nfs_direct_req
*dreq
= desc
->pg_dreq
;
320 struct nfs_open_context
*ctx
= dreq
->ctx
;
321 struct inode
*inode
= ctx
->dentry
->d_inode
;
322 unsigned long user_addr
= (unsigned long)iov
->iov_base
;
323 size_t count
= iov
->iov_len
;
324 size_t rsize
= NFS_SERVER(inode
)->rsize
;
328 struct page
**pagevec
= NULL
;
335 pgbase
= user_addr
& ~PAGE_MASK
;
336 bytes
= min(max_t(size_t, rsize
, PAGE_SIZE
), count
);
339 npages
= nfs_page_array_len(pgbase
, bytes
);
341 pagevec
= kmalloc(npages
* sizeof(struct page
*),
346 down_read(¤t
->mm
->mmap_sem
);
347 result
= get_user_pages(current
, current
->mm
, user_addr
,
348 npages
, 1, 0, pagevec
, NULL
);
349 up_read(¤t
->mm
->mmap_sem
);
353 WARN_ON(npages
!= 1);
354 result
= get_kernel_page(user_addr
, 1, pagevec
);
355 if (WARN_ON(result
!= 1))
359 if ((unsigned)result
< npages
) {
360 bytes
= result
* PAGE_SIZE
;
361 if (bytes
<= pgbase
) {
362 nfs_direct_release_pages(pagevec
, result
);
369 for (i
= 0; i
< npages
; i
++) {
370 struct nfs_page
*req
;
371 unsigned int req_len
= min_t(size_t, bytes
, PAGE_SIZE
- pgbase
);
372 /* XXX do we need to do the eof zeroing found in async_filler? */
373 req
= nfs_create_request(dreq
->ctx
, dreq
->inode
,
377 result
= PTR_ERR(req
);
380 req
->wb_index
= pos
>> PAGE_SHIFT
;
381 req
->wb_offset
= pos
& ~PAGE_MASK
;
382 if (!nfs_pageio_add_request(desc
, req
)) {
383 result
= desc
->pg_error
;
384 nfs_release_request(req
);
390 user_addr
+= req_len
;
394 /* The nfs_page now hold references to these pages */
395 nfs_direct_release_pages(pagevec
, npages
);
396 } while (count
!= 0 && result
>= 0);
402 return result
< 0 ? (ssize_t
) result
: -EFAULT
;
405 static ssize_t
nfs_direct_read_schedule_iovec(struct nfs_direct_req
*dreq
,
406 const struct iovec
*iov
,
407 unsigned long nr_segs
,
408 loff_t pos
, bool uio
)
410 struct nfs_pageio_descriptor desc
;
411 ssize_t result
= -EINVAL
;
412 size_t requested_bytes
= 0;
415 NFS_PROTO(dreq
->inode
)->read_pageio_init(&desc
, dreq
->inode
,
416 &nfs_direct_read_completion_ops
);
420 for (seg
= 0; seg
< nr_segs
; seg
++) {
421 const struct iovec
*vec
= &iov
[seg
];
422 result
= nfs_direct_read_schedule_segment(&desc
, vec
, pos
, uio
);
425 requested_bytes
+= result
;
426 if ((size_t)result
< vec
->iov_len
)
431 nfs_pageio_complete(&desc
);
434 * If no bytes were started, return the error, and let the
435 * generic layer handle the completion.
437 if (requested_bytes
== 0) {
438 nfs_direct_req_release(dreq
);
439 return result
< 0 ? result
: -EIO
;
443 nfs_direct_complete(dreq
);
447 static ssize_t
nfs_direct_read(struct kiocb
*iocb
, const struct iovec
*iov
,
448 unsigned long nr_segs
, loff_t pos
, bool uio
)
450 ssize_t result
= -ENOMEM
;
451 struct inode
*inode
= iocb
->ki_filp
->f_mapping
->host
;
452 struct nfs_direct_req
*dreq
;
454 dreq
= nfs_direct_req_alloc();
459 dreq
->ctx
= get_nfs_open_context(nfs_file_open_context(iocb
->ki_filp
));
460 dreq
->l_ctx
= nfs_get_lock_context(dreq
->ctx
);
461 if (dreq
->l_ctx
== NULL
)
463 if (!is_sync_kiocb(iocb
))
466 result
= nfs_direct_read_schedule_iovec(dreq
, iov
, nr_segs
, pos
, uio
);
468 result
= nfs_direct_wait(dreq
);
469 NFS_I(inode
)->read_io
+= result
;
471 nfs_direct_req_release(dreq
);
476 static void nfs_inode_dio_write_done(struct inode
*inode
)
478 nfs_zap_mapping(inode
, inode
->i_mapping
);
479 inode_dio_done(inode
);
482 #if IS_ENABLED(CONFIG_NFS_V3) || IS_ENABLED(CONFIG_NFS_V4)
483 static void nfs_direct_write_reschedule(struct nfs_direct_req
*dreq
)
485 struct nfs_pageio_descriptor desc
;
486 struct nfs_page
*req
, *tmp
;
488 struct nfs_commit_info cinfo
;
491 nfs_init_cinfo_from_dreq(&cinfo
, dreq
);
492 pnfs_recover_commit_reqs(dreq
->inode
, &reqs
, &cinfo
);
493 spin_lock(cinfo
.lock
);
494 nfs_scan_commit_list(&cinfo
.mds
->list
, &reqs
, &cinfo
, 0);
495 spin_unlock(cinfo
.lock
);
500 NFS_PROTO(dreq
->inode
)->write_pageio_init(&desc
, dreq
->inode
, FLUSH_STABLE
,
501 &nfs_direct_write_completion_ops
);
504 list_for_each_entry_safe(req
, tmp
, &reqs
, wb_list
) {
505 if (!nfs_pageio_add_request(&desc
, req
)) {
506 nfs_list_remove_request(req
);
507 nfs_list_add_request(req
, &failed
);
508 spin_lock(cinfo
.lock
);
511 spin_unlock(cinfo
.lock
);
513 nfs_release_request(req
);
515 nfs_pageio_complete(&desc
);
517 while (!list_empty(&failed
)) {
518 req
= nfs_list_entry(failed
.next
);
519 nfs_list_remove_request(req
);
520 nfs_unlock_and_release_request(req
);
524 nfs_direct_write_complete(dreq
, dreq
->inode
);
527 static void nfs_direct_commit_complete(struct nfs_commit_data
*data
)
529 struct nfs_direct_req
*dreq
= data
->dreq
;
530 struct nfs_commit_info cinfo
;
531 struct nfs_page
*req
;
532 int status
= data
->task
.tk_status
;
534 nfs_init_cinfo_from_dreq(&cinfo
, dreq
);
536 dprintk("NFS: %5u commit failed with error %d.\n",
537 data
->task
.tk_pid
, status
);
538 dreq
->flags
= NFS_ODIRECT_RESCHED_WRITES
;
539 } else if (memcmp(&dreq
->verf
, &data
->verf
, sizeof(data
->verf
))) {
540 dprintk("NFS: %5u commit verify failed\n", data
->task
.tk_pid
);
541 dreq
->flags
= NFS_ODIRECT_RESCHED_WRITES
;
544 dprintk("NFS: %5u commit returned %d\n", data
->task
.tk_pid
, status
);
545 while (!list_empty(&data
->pages
)) {
546 req
= nfs_list_entry(data
->pages
.next
);
547 nfs_list_remove_request(req
);
548 if (dreq
->flags
== NFS_ODIRECT_RESCHED_WRITES
) {
549 /* Note the rewrite will go through mds */
550 nfs_mark_request_commit(req
, NULL
, &cinfo
);
552 nfs_release_request(req
);
553 nfs_unlock_and_release_request(req
);
556 if (atomic_dec_and_test(&cinfo
.mds
->rpcs_out
))
557 nfs_direct_write_complete(dreq
, data
->inode
);
560 static void nfs_direct_error_cleanup(struct nfs_inode
*nfsi
)
562 /* There is no lock to clear */
565 static const struct nfs_commit_completion_ops nfs_direct_commit_completion_ops
= {
566 .completion
= nfs_direct_commit_complete
,
567 .error_cleanup
= nfs_direct_error_cleanup
,
570 static void nfs_direct_commit_schedule(struct nfs_direct_req
*dreq
)
573 struct nfs_commit_info cinfo
;
576 nfs_init_cinfo_from_dreq(&cinfo
, dreq
);
577 nfs_scan_commit(dreq
->inode
, &mds_list
, &cinfo
);
578 res
= nfs_generic_commit_list(dreq
->inode
, &mds_list
, 0, &cinfo
);
579 if (res
< 0) /* res == -ENOMEM */
580 nfs_direct_write_reschedule(dreq
);
583 static void nfs_direct_write_schedule_work(struct work_struct
*work
)
585 struct nfs_direct_req
*dreq
= container_of(work
, struct nfs_direct_req
, work
);
586 int flags
= dreq
->flags
;
590 case NFS_ODIRECT_DO_COMMIT
:
591 nfs_direct_commit_schedule(dreq
);
593 case NFS_ODIRECT_RESCHED_WRITES
:
594 nfs_direct_write_reschedule(dreq
);
597 nfs_inode_dio_write_done(dreq
->inode
);
598 nfs_direct_complete(dreq
);
602 static void nfs_direct_write_complete(struct nfs_direct_req
*dreq
, struct inode
*inode
)
604 schedule_work(&dreq
->work
); /* Calls nfs_direct_write_schedule_work */
608 static void nfs_direct_write_schedule_work(struct work_struct
*work
)
612 static void nfs_direct_write_complete(struct nfs_direct_req
*dreq
, struct inode
*inode
)
614 nfs_inode_dio_write_done(inode
);
615 nfs_direct_complete(dreq
);
620 * NB: Return the value of the first error return code. Subsequent
621 * errors after the first one are ignored.
624 * For each wsize'd chunk of the user's buffer, dispatch an NFS WRITE
625 * operation. If nfs_writedata_alloc() or get_user_pages() fails,
626 * bail and stop sending more writes. Write length accounting is
627 * handled automatically by nfs_direct_write_result(). Otherwise, if
628 * no requests have been sent, just return an error.
630 static ssize_t
nfs_direct_write_schedule_segment(struct nfs_pageio_descriptor
*desc
,
631 const struct iovec
*iov
,
632 loff_t pos
, bool uio
)
634 struct nfs_direct_req
*dreq
= desc
->pg_dreq
;
635 struct nfs_open_context
*ctx
= dreq
->ctx
;
636 struct inode
*inode
= ctx
->dentry
->d_inode
;
637 unsigned long user_addr
= (unsigned long)iov
->iov_base
;
638 size_t count
= iov
->iov_len
;
639 size_t wsize
= NFS_SERVER(inode
)->wsize
;
643 struct page
**pagevec
= NULL
;
650 pgbase
= user_addr
& ~PAGE_MASK
;
651 bytes
= min(max_t(size_t, wsize
, PAGE_SIZE
), count
);
654 npages
= nfs_page_array_len(pgbase
, bytes
);
656 pagevec
= kmalloc(npages
* sizeof(struct page
*), GFP_KERNEL
);
661 down_read(¤t
->mm
->mmap_sem
);
662 result
= get_user_pages(current
, current
->mm
, user_addr
,
663 npages
, 0, 0, pagevec
, NULL
);
664 up_read(¤t
->mm
->mmap_sem
);
668 WARN_ON(npages
!= 1);
669 result
= get_kernel_page(user_addr
, 0, pagevec
);
670 if (WARN_ON(result
!= 1))
674 if ((unsigned)result
< npages
) {
675 bytes
= result
* PAGE_SIZE
;
676 if (bytes
<= pgbase
) {
677 nfs_direct_release_pages(pagevec
, result
);
684 for (i
= 0; i
< npages
; i
++) {
685 struct nfs_page
*req
;
686 unsigned int req_len
= min_t(size_t, bytes
, PAGE_SIZE
- pgbase
);
688 req
= nfs_create_request(dreq
->ctx
, dreq
->inode
,
692 result
= PTR_ERR(req
);
695 nfs_lock_request(req
);
696 req
->wb_index
= pos
>> PAGE_SHIFT
;
697 req
->wb_offset
= pos
& ~PAGE_MASK
;
698 if (!nfs_pageio_add_request(desc
, req
)) {
699 result
= desc
->pg_error
;
700 nfs_unlock_and_release_request(req
);
706 user_addr
+= req_len
;
710 /* The nfs_page now hold references to these pages */
711 nfs_direct_release_pages(pagevec
, npages
);
712 } while (count
!= 0 && result
>= 0);
718 return result
< 0 ? (ssize_t
) result
: -EFAULT
;
721 static void nfs_direct_write_completion(struct nfs_pgio_header
*hdr
)
723 struct nfs_direct_req
*dreq
= hdr
->dreq
;
724 struct nfs_commit_info cinfo
;
726 struct nfs_page
*req
= nfs_list_entry(hdr
->pages
.next
);
728 if (test_bit(NFS_IOHDR_REDO
, &hdr
->flags
))
731 nfs_init_cinfo_from_dreq(&cinfo
, dreq
);
733 spin_lock(&dreq
->lock
);
735 if (test_bit(NFS_IOHDR_ERROR
, &hdr
->flags
)) {
737 dreq
->error
= hdr
->error
;
739 if (dreq
->error
!= 0)
740 bit
= NFS_IOHDR_ERROR
;
742 dreq
->count
+= hdr
->good_bytes
;
743 if (test_bit(NFS_IOHDR_NEED_RESCHED
, &hdr
->flags
)) {
744 dreq
->flags
= NFS_ODIRECT_RESCHED_WRITES
;
745 bit
= NFS_IOHDR_NEED_RESCHED
;
746 } else if (test_bit(NFS_IOHDR_NEED_COMMIT
, &hdr
->flags
)) {
747 if (dreq
->flags
== NFS_ODIRECT_RESCHED_WRITES
)
748 bit
= NFS_IOHDR_NEED_RESCHED
;
749 else if (dreq
->flags
== 0) {
750 memcpy(&dreq
->verf
, hdr
->verf
,
752 bit
= NFS_IOHDR_NEED_COMMIT
;
753 dreq
->flags
= NFS_ODIRECT_DO_COMMIT
;
754 } else if (dreq
->flags
== NFS_ODIRECT_DO_COMMIT
) {
755 if (memcmp(&dreq
->verf
, hdr
->verf
, sizeof(dreq
->verf
))) {
756 dreq
->flags
= NFS_ODIRECT_RESCHED_WRITES
;
757 bit
= NFS_IOHDR_NEED_RESCHED
;
759 bit
= NFS_IOHDR_NEED_COMMIT
;
763 spin_unlock(&dreq
->lock
);
765 while (!list_empty(&hdr
->pages
)) {
766 req
= nfs_list_entry(hdr
->pages
.next
);
767 nfs_list_remove_request(req
);
769 case NFS_IOHDR_NEED_RESCHED
:
770 case NFS_IOHDR_NEED_COMMIT
:
771 kref_get(&req
->wb_kref
);
772 nfs_mark_request_commit(req
, hdr
->lseg
, &cinfo
);
774 nfs_unlock_and_release_request(req
);
779 nfs_direct_write_complete(dreq
, hdr
->inode
);
783 static void nfs_write_sync_pgio_error(struct list_head
*head
)
785 struct nfs_page
*req
;
787 while (!list_empty(head
)) {
788 req
= nfs_list_entry(head
->next
);
789 nfs_list_remove_request(req
);
790 nfs_unlock_and_release_request(req
);
794 static const struct nfs_pgio_completion_ops nfs_direct_write_completion_ops
= {
795 .error_cleanup
= nfs_write_sync_pgio_error
,
796 .init_hdr
= nfs_direct_pgio_init
,
797 .completion
= nfs_direct_write_completion
,
800 static ssize_t
nfs_direct_write_schedule_iovec(struct nfs_direct_req
*dreq
,
801 const struct iovec
*iov
,
802 unsigned long nr_segs
,
803 loff_t pos
, bool uio
)
805 struct nfs_pageio_descriptor desc
;
806 struct inode
*inode
= dreq
->inode
;
808 size_t requested_bytes
= 0;
811 NFS_PROTO(inode
)->write_pageio_init(&desc
, inode
, FLUSH_COND_STABLE
,
812 &nfs_direct_write_completion_ops
);
815 atomic_inc(&inode
->i_dio_count
);
817 for (seg
= 0; seg
< nr_segs
; seg
++) {
818 const struct iovec
*vec
= &iov
[seg
];
819 result
= nfs_direct_write_schedule_segment(&desc
, vec
, pos
, uio
);
822 requested_bytes
+= result
;
823 if ((size_t)result
< vec
->iov_len
)
827 nfs_pageio_complete(&desc
);
828 NFS_I(dreq
->inode
)->write_io
+= desc
.pg_bytes_written
;
831 * If no bytes were started, return the error, and let the
832 * generic layer handle the completion.
834 if (requested_bytes
== 0) {
835 inode_dio_done(inode
);
836 nfs_direct_req_release(dreq
);
837 return result
< 0 ? result
: -EIO
;
841 nfs_direct_write_complete(dreq
, dreq
->inode
);
845 static ssize_t
nfs_direct_write(struct kiocb
*iocb
, const struct iovec
*iov
,
846 unsigned long nr_segs
, loff_t pos
,
847 size_t count
, bool uio
)
849 ssize_t result
= -ENOMEM
;
850 struct inode
*inode
= iocb
->ki_filp
->f_mapping
->host
;
851 struct nfs_direct_req
*dreq
;
853 dreq
= nfs_direct_req_alloc();
858 dreq
->ctx
= get_nfs_open_context(nfs_file_open_context(iocb
->ki_filp
));
859 dreq
->l_ctx
= nfs_get_lock_context(dreq
->ctx
);
860 if (dreq
->l_ctx
== NULL
)
862 if (!is_sync_kiocb(iocb
))
865 result
= nfs_direct_write_schedule_iovec(dreq
, iov
, nr_segs
, pos
, uio
);
867 result
= nfs_direct_wait(dreq
);
869 nfs_direct_req_release(dreq
);
875 * nfs_file_direct_read - file direct read operation for NFS files
876 * @iocb: target I/O control block
877 * @iov: vector of user buffers into which to read data
878 * @nr_segs: size of iov vector
879 * @pos: byte offset in file where reading starts
881 * We use this function for direct reads instead of calling
882 * generic_file_aio_read() in order to avoid gfar's check to see if
883 * the request starts before the end of the file. For that check
884 * to work, we must generate a GETATTR before each direct read, and
885 * even then there is a window between the GETATTR and the subsequent
886 * READ where the file size could change. Our preference is simply
887 * to do all reads the application wants, and the server will take
888 * care of managing the end of file boundary.
890 * This function also eliminates unnecessarily updating the file's
891 * atime locally, as the NFS server sets the file's atime, and this
892 * client must read the updated atime from the server back into its
895 ssize_t
nfs_file_direct_read(struct kiocb
*iocb
, const struct iovec
*iov
,
896 unsigned long nr_segs
, loff_t pos
, bool uio
)
898 ssize_t retval
= -EINVAL
;
899 struct file
*file
= iocb
->ki_filp
;
900 struct address_space
*mapping
= file
->f_mapping
;
903 count
= iov_length(iov
, nr_segs
);
904 nfs_add_stats(mapping
->host
, NFSIOS_DIRECTREADBYTES
, count
);
906 dfprintk(FILE, "NFS: direct read(%s/%s, %zd@%Ld)\n",
907 file
->f_path
.dentry
->d_parent
->d_name
.name
,
908 file
->f_path
.dentry
->d_name
.name
,
909 count
, (long long) pos
);
915 retval
= nfs_sync_mapping(mapping
);
919 task_io_account_read(count
);
921 retval
= nfs_direct_read(iocb
, iov
, nr_segs
, pos
, uio
);
923 iocb
->ki_pos
= pos
+ retval
;
930 * nfs_file_direct_write - file direct write operation for NFS files
931 * @iocb: target I/O control block
932 * @iov: vector of user buffers from which to write data
933 * @nr_segs: size of iov vector
934 * @pos: byte offset in file where writing starts
936 * We use this function for direct writes instead of calling
937 * generic_file_aio_write() in order to avoid taking the inode
938 * semaphore and updating the i_size. The NFS server will set
939 * the new i_size and this client must read the updated size
940 * back into its cache. We let the server do generic write
941 * parameter checking and report problems.
943 * We eliminate local atime updates, see direct read above.
945 * We avoid unnecessary page cache invalidations for normal cached
946 * readers of this file.
948 * Note that O_APPEND is not supported for NFS direct writes, as there
949 * is no atomic O_APPEND write facility in the NFS protocol.
951 ssize_t
nfs_file_direct_write(struct kiocb
*iocb
, const struct iovec
*iov
,
952 unsigned long nr_segs
, loff_t pos
, bool uio
)
954 ssize_t retval
= -EINVAL
;
955 struct file
*file
= iocb
->ki_filp
;
956 struct address_space
*mapping
= file
->f_mapping
;
959 count
= iov_length(iov
, nr_segs
);
960 nfs_add_stats(mapping
->host
, NFSIOS_DIRECTWRITTENBYTES
, count
);
962 dfprintk(FILE, "NFS: direct write(%s/%s, %zd@%Ld)\n",
963 file
->f_path
.dentry
->d_parent
->d_name
.name
,
964 file
->f_path
.dentry
->d_name
.name
,
965 count
, (long long) pos
);
967 retval
= generic_write_checks(file
, &pos
, &count
, 0);
972 if ((ssize_t
) count
< 0)
978 retval
= nfs_sync_mapping(mapping
);
982 task_io_account_write(count
);
984 retval
= nfs_direct_write(iocb
, iov
, nr_segs
, pos
, count
, uio
);
986 struct inode
*inode
= mapping
->host
;
988 iocb
->ki_pos
= pos
+ retval
;
989 spin_lock(&inode
->i_lock
);
990 if (i_size_read(inode
) < iocb
->ki_pos
)
991 i_size_write(inode
, iocb
->ki_pos
);
992 spin_unlock(&inode
->i_lock
);
999 * nfs_init_directcache - create a slab cache for nfs_direct_req structures
1002 int __init
nfs_init_directcache(void)
1004 nfs_direct_cachep
= kmem_cache_create("nfs_direct_cache",
1005 sizeof(struct nfs_direct_req
),
1006 0, (SLAB_RECLAIM_ACCOUNT
|
1009 if (nfs_direct_cachep
== NULL
)
1016 * nfs_destroy_directcache - destroy the slab cache for nfs_direct_req structures
1019 void nfs_destroy_directcache(void)
1021 kmem_cache_destroy(nfs_direct_cachep
);