summaryrefslogtreecommitdiffhomepage
path: root/libbb/read.c
blob: 288358d79bfbc9a9bc54692e85d5be0e52188bf0 (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
/* vi: set sw=4 ts=4: */
/*
 * Utility routines.
 *
 * Copyright (C) 1999-2004 by Erik Andersen <andersen@codepoet.org>
 *
 * Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
 */

#include "libbb.h"

ssize_t safe_read(int fd, void *buf, size_t count)
{
	ssize_t n;

	do {
		n = read(fd, buf, count);
	} while (n < 0 && errno == EINTR);

	return n;
}

/* Suppose that you are a shell. You start child processes.
 * They work and eventually exit. You want to get user input.
 * You read stdin. But what happens if last child switched
 * its stdin into O_NONBLOCK mode?
 *
 * *** SURPRISE! It will affect the parent too! ***
 * *** BIG SURPRISE! It stays even after child exits! ***
 *
 * This is a design bug in UNIX API.
 *      fcntl(0, F_SETFL, fcntl(0, F_GETFL, 0) | O_NONBLOCK);
 * will set nonblocking mode not only on _your_ stdin, but
 * also on stdin of your parent, etc.
 *
 * In general,
 *      fd2 = dup(fd1);
 *      fcntl(fd2, F_SETFL, fcntl(fd2, F_GETFL, 0) | O_NONBLOCK);
 * sets both fd1 and fd2 to O_NONBLOCK. This includes cases
 * where duping is done implicitly by fork() etc.
 *
 * We need
 *      fcntl(fd2, F_SETFD, fcntl(fd2, F_GETFD, 0) | O_NONBLOCK);
 * (note SETFD, not SETFL!) but such thing doesn't exist.
 *
 * Alternatively, we need nonblocking_read(fd, ...) which doesn't
 * require O_NONBLOCK dance at all. Actually, it exists:
 *      n = recv(fd, buf, len, MSG_DONTWAIT);
 *      "MSG_DONTWAIT:
 *      Enables non-blocking operation; if the operation
 *      would block, EAGAIN is returned."
 * but recv() works only for sockets!
 *
 * So far I don't see any good solution, I can only propose
 * that affected readers should be careful and use this routine,
 * which detects EAGAIN and uses poll() to wait on the fd.
 * Thankfully, poll() doesn't care about O_NONBLOCK flag.
 */
ssize_t nonblock_safe_read(int fd, void *buf, size_t count)
{
	struct pollfd pfd[1];
	ssize_t n;

	while (1) {
		n = safe_read(fd, buf, count);
		if (n >= 0 || errno != EAGAIN)
			return n;
		/* fd is in O_NONBLOCK mode. Wait using poll and repeat */
		pfd[0].fd = fd;
		pfd[0].events = POLLIN;
		safe_poll(pfd, 1, -1);
	}
}

/*
 * Read all of the supplied buffer from a file.
 * This does multiple reads as necessary.
 * Returns the amount read, or -1 on an error.
 * A short read is returned on an end of file.
 */
ssize_t full_read(int fd, void *buf, size_t len)
{
	ssize_t cc;
	ssize_t total;

	total = 0;

	while (len) {
		cc = safe_read(fd, buf, len);

		if (cc < 0) {
			if (total) {
				/* we already have some! */
				/* user can do another read to know the error code */
				return total;
			}
			return cc; /* read() returns -1 on failure. */
		}
		if (cc == 0)
			break;
		buf = ((char *)buf) + cc;
		total += cc;
		len -= cc;
	}

	return total;
}

// Die with an error message if we can't read the entire buffer.
void xread(int fd, void *buf, size_t count)
{
	if (count) {
		ssize_t size = full_read(fd, buf, count);
		if (size != count)
			bb_error_msg_and_die("short read");
	}
}

// Die with an error message if we can't read one character.
unsigned char xread_char(int fd)
{
	char tmp;
	xread(fd, &tmp, 1);
	return tmp;
}

// Read one line a-la fgets. Works only on seekable streams
char *reads(int fd, char *buffer, size_t size)
{
	char *p;

	if (size < 2)
		return NULL;
	size = full_read(fd, buffer, size-1);
	if ((ssize_t)size <= 0)
		return NULL;

	buffer[size] = '\0';
	p = strchr(buffer, '\n');
	if (p) {
		off_t offset;
		*p++ = '\0';
		// avoid incorrect (unsigned) widening
		offset = (off_t)(p - buffer) - (off_t)size;
		// set fd position right after '\n'
		if (offset && lseek(fd, offset, SEEK_CUR) == (off_t)-1)
			return NULL;
	}
	return buffer;
}

// Reads one line a-la fgets (but doesn't save terminating '\n').
// Reads byte-by-byte. Useful when it is important to not read ahead.
// Bytes are appended to pfx (which must be malloced, or NULL).
char *xmalloc_reads(int fd, char *buf, size_t *maxsz_p)
{
	char *p;
	size_t sz = buf ? strlen(buf) : 0;
	size_t maxsz = maxsz_p ? *maxsz_p : MAXINT(size_t);

	goto jump_in;
	while (sz < maxsz) {
		if (p - buf == sz) {
 jump_in:
			buf = xrealloc(buf, sz + 128);
			p = buf + sz;
			sz += 128;
		}
		/* nonblock_safe_read() because we are used by e.g. shells */
		if (nonblock_safe_read(fd, p, 1) != 1) { /* EOF/error */
			if (p == buf) { /* we read nothing */
				free(buf);
				return NULL;
			}
			break;
		}
		if (*p == '\n')
			break;
		p++;
	}
	*p = '\0';
	if (maxsz_p)
		*maxsz_p  = p - buf;
	p++;
	return xrealloc(buf, p - buf);
}

ssize_t read_close(int fd, void *buf, size_t size)
{
	/*int e;*/
	size = full_read(fd, buf, size);
	/*e = errno;*/
	close(fd);
	/*errno = e;*/
	return size;
}

ssize_t open_read_close(const char *filename, void *buf, size_t size)
{
	int fd = open(filename, O_RDONLY);
	if (fd < 0)
		return fd;
	return read_close(fd, buf, size);
}

// Read (potentially big) files in one go. File size is estimated by
// lseek to end.
void *xmalloc_open_read_close(const char *filename, size_t *sizep)
{
	char *buf;
	size_t size;
	int fd;
	off_t len;
	struct stat st;

	fd = open(filename, O_RDONLY);
	if (fd < 0)
		return NULL;

	st.st_size = 0; /* in case fstat fail, define to 0 */
	fstat(fd, &st);
	/* /proc/N/stat files report len 0 here */
	/* In order to make such files readable, we add small const */
	len = st.st_size | 0x3ff; /* read only 1k on unseekable files */
	size = sizep ? *sizep : INT_MAX;
	if (len < size)
		size = len;
	buf = xmalloc(size + 1);
	size = read_close(fd, buf, size);
	if ((ssize_t)size < 0) {
		free(buf);
		return NULL;
	}
	xrealloc(buf, size + 1);
	buf[size] = '\0';

	if (sizep)
		*sizep = size;
	return buf;
}

#ifdef USING_LSEEK_TO_GET_SIZE
/* Alternatively, file size can be obtained by lseek to the end.
 * The code is slightly bigger. Retained in case fstat approach
 * will not work for some weird cases (/proc, block devices, etc).
 * (NB: lseek also can fail to work for some weird files) */

// Read (potentially big) files in one go. File size is estimated by
// lseek to end.
void *xmalloc_open_read_close(const char *filename, size_t *sizep)
{
	char *buf;
	size_t size;
	int fd;
	off_t len;

	fd = open(filename, O_RDONLY);
	if (fd < 0)
		return NULL;

	/* /proc/N/stat files report len 0 here */
	/* In order to make such files readable, we add small const */
	size = 0x3ff; /* read only 1k on unseekable files */
	len = lseek(fd, 0, SEEK_END) | 0x3ff; /* + up to 1k */
	if (len != (off_t)-1) {
		xlseek(fd, 0, SEEK_SET);
		size = sizep ? *sizep : INT_MAX;
		if (len < size)
			size = len;
	}

	buf = xmalloc(size + 1);
	size = read_close(fd, buf, size);
	if ((ssize_t)size < 0) {
		free(buf);
		return NULL;
	}
	xrealloc(buf, size + 1);
	buf[size] = '\0';

	if (sizep)
		*sizep = size;
	return buf;
}
#endif

void *xmalloc_xopen_read_close(const char *filename, size_t *sizep)
{
	void *buf = xmalloc_open_read_close(filename, sizep);
	if (!buf)
		bb_perror_msg_and_die("can't read '%s'", filename);
	return buf;
}