summaryrefslogtreecommitdiff
path: root/source3/lib/recvfile.c
blob: e1eb241d7bd8ce168443882d3cc7cad145cb6590 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
/*
 Unix SMB/Netbios implementation.
 Version 3.2.x
 recvfile implementations.
 Copyright (C) Jeremy Allison 2007.

 This program is free software; you can redistribute it and/or modify
 it under the terms of the GNU General Public License as published by
 the Free Software Foundation; either version 3 of the License, or
 (at your option) any later version.
 This program is distributed in the hope that it will be useful,
 but WITHOUT ANY WARRANTY; without even the implied warranty of
 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 GNU General Public License for more details.

 You should have received a copy of the GNU General Public License
 along with this program; if not, see <http://www.gnu.org/licenses/>.
*/

/*
 * This file handles the OS dependent recvfile implementations.
 * The API is such that it returns -1 on error, else returns the
 * number of bytes written.
 */

#include "includes.h"
#include "system/filesys.h"
#include "lib/util/sys_rw.h"

/* Do this on our own in TRANSFER_BUF_SIZE chunks.
 * It's safe to make direct syscalls to lseek/write here
 * as we're below the Samba vfs layer.
 *
 * Returns -1 on short reads from fromfd (read error)
 * and sets errno.
 *
 * Returns number of bytes written to 'tofd'
 * return != count then sets errno.
 * Returns count if complete success.
 */

#ifndef TRANSFER_BUF_SIZE
#define TRANSFER_BUF_SIZE (128*1024)
#endif

static ssize_t default_sys_recvfile(int fromfd,
			int tofd,
			off_t offset,
			size_t count)
{
	int saved_errno = 0;
	size_t total = 0;
	size_t bufsize = MIN(TRANSFER_BUF_SIZE,count);
	size_t total_written = 0;
	char buffer[bufsize];

	DEBUG(10,("default_sys_recvfile: from = %d, to = %d, "
		"offset=%.0f, count = %lu\n",
		fromfd, tofd, (double)offset,
		(unsigned long)count));

	if (count == 0) {
		return 0;
	}

	if (tofd != -1 && offset != (off_t)-1) {
		if (lseek(tofd, offset, SEEK_SET) == -1) {
			if (errno != ESPIPE) {
				return -1;
			}
		}
	}

	while (total < count) {
		size_t num_written = 0;
		ssize_t read_ret;
		size_t toread = MIN(bufsize,count - total);

		/*
		 * Read from socket - ignore EINTR.
		 * Can't use sys_read() as that also
		 * ignores EAGAIN and EWOULDBLOCK.
		 */
		do {
			read_ret = read(fromfd, buffer, toread);
		} while (read_ret == -1 && errno == EINTR);

		if (read_ret == -1 && (errno == EAGAIN || errno == EWOULDBLOCK)) {
			/*
			 * fromfd socket is in non-blocking mode.
			 * If we already read some and wrote
			 * it successfully, return that.
			 * Only return -1 if this is the first read
			 * attempt. Caller will handle both cases.
			 */
			if (total_written != 0) {
				return total_written;
			}
			return -1;
		}

		if (read_ret <= 0) {
			/* EOF or socket error. */
			return -1;
		}

		num_written = 0;

		/* Don't write any more after a write error. */
		while (tofd != -1 && (num_written < read_ret)) {
			ssize_t write_ret;

			/* Write to file - ignore EINTR. */
			write_ret = sys_write(tofd,
					buffer + num_written,
					read_ret - num_written);

			if (write_ret <= 0) {
				/* write error - stop writing. */
				tofd = -1;
                                if (total_written == 0) {
					/* Ensure we return
					   -1 if the first
					   write failed. */
                                        total_written = -1;
                                }
				saved_errno = errno;
				break;
			}

			num_written += (size_t)write_ret;
			total_written += (size_t)write_ret;
		}

		total += read_ret;
	}

	if (saved_errno) {
		/* Return the correct write error. */
		errno = saved_errno;
	}
	return (ssize_t)total_written;
}

#if defined(HAVE_LINUX_SPLICE)

/*
 * Try and use the Linux system call to do this.
 * Remember we only return -1 if the socket read
 * failed. Else we return the number of bytes
 * actually written. We always read count bytes
 * from the network in the case of return != -1.
 */


ssize_t sys_recvfile(int fromfd,
			int tofd,
			off_t offset,
			size_t count)
{
	static int pipefd[2] = { -1, -1 };
	static bool try_splice_call = false;
	size_t total_written = 0;
	loff_t splice_offset = offset;

	DEBUG(10,("sys_recvfile: from = %d, to = %d, "
		"offset=%.0f, count = %lu\n",
		fromfd, tofd, (double)offset,
		(unsigned long)count));

	if (count == 0) {
		return 0;
	}

	/*
	 * Older Linux kernels have splice for sendfile,
	 * but it fails for recvfile. Ensure we only try
	 * this once and always fall back to the userspace
	 * implementation if recvfile splice fails. JRA.
	 */

	if (!try_splice_call) {
		return default_sys_recvfile(fromfd,
				tofd,
				offset,
				count);
	}

	if ((pipefd[0] == -1) && (pipe(pipefd) == -1)) {
		try_splice_call = false;
		return default_sys_recvfile(fromfd, tofd, offset, count);
	}

	while (count > 0) {
		int nread, to_write;

		nread = splice(fromfd, NULL, pipefd[1], NULL,
			       MIN(count, 16384), SPLICE_F_MOVE);
		if (nread == -1) {
			if (errno == EINTR) {
				continue;
			}
			if (total_written == 0 &&
			    (errno == EBADF || errno == EINVAL)) {
				try_splice_call = false;
				return default_sys_recvfile(fromfd, tofd,
							    offset, count);
			}
			if (errno == EAGAIN || errno == EWOULDBLOCK) {
				/*
				 * fromfd socket is in non-blocking mode.
				 * If we already read some and wrote
				 * it successfully, return that.
				 * Only return -1 if this is the first read
				 * attempt. Caller will handle both cases.
				 */
				if (total_written != 0) {
					return total_written;
				}
				return -1;
			}
			break;
		}

		to_write = nread;
		while (to_write > 0) {
			int thistime;
			thistime = splice(pipefd[0], NULL, tofd,
					  &splice_offset, to_write,
					  SPLICE_F_MOVE);
			if (thistime == -1) {
				goto done;
			}
			to_write -= thistime;
		}

		total_written += nread;
		count -= nread;
	}

 done:
	if (count) {
		int saved_errno = errno;
		if (drain_socket(fromfd, count) != count) {
			/* socket is dead. */
			return -1;
		}
		errno = saved_errno;
	}

	return total_written;
}
#else

/*****************************************************************
 No recvfile system call - use the default 128 chunk implementation.
*****************************************************************/

ssize_t sys_recvfile(int fromfd,
			int tofd,
			off_t offset,
			size_t count)
{
	return default_sys_recvfile(fromfd, tofd, offset, count);
}
#endif

/*****************************************************************
 Throw away "count" bytes from the client socket.
 Returns count or -1 on error.
 Must only operate on a blocking socket.
*****************************************************************/

ssize_t drain_socket(int sockfd, size_t count)
{
	size_t total = 0;
	size_t bufsize = MIN(TRANSFER_BUF_SIZE,count);
	char buffer[bufsize];
	int old_flags = 0;

	if (count == 0) {
		return 0;
	}

	old_flags = fcntl(sockfd, F_GETFL, 0);
	if (set_blocking(sockfd, true) == -1) {
		return -1;
	}

	while (total < count) {
		ssize_t read_ret;
		size_t toread = MIN(bufsize,count - total);

		/* Read from socket - ignore EINTR. */
		read_ret = sys_read(sockfd, buffer, toread);
		if (read_ret <= 0) {
			/* EOF or socket error. */
			count = (size_t)-1;
			goto out;
		}
		total += read_ret;
	}

  out:

	if (fcntl(sockfd, F_SETFL, old_flags) == -1) {
		return -1;
	}
	return count;
}