xref: /linux/lib/string_helpers.c (revision 63307d015b91e626c97bb82e88054af3d0b74643)
1 // SPDX-License-Identifier: GPL-2.0-only
2 /*
3  * Helpers for formatting and printing strings
4  *
5  * Copyright 31 August 2008 James Bottomley
6  * Copyright (C) 2013, Intel Corporation
7  */
8 #include <linux/bug.h>
9 #include <linux/kernel.h>
10 #include <linux/math64.h>
11 #include <linux/export.h>
12 #include <linux/ctype.h>
13 #include <linux/errno.h>
14 #include <linux/fs.h>
15 #include <linux/limits.h>
16 #include <linux/mm.h>
17 #include <linux/slab.h>
18 #include <linux/string.h>
19 #include <linux/string_helpers.h>
20 
21 /**
22  * string_get_size - get the size in the specified units
23  * @size:	The size to be converted in blocks
24  * @blk_size:	Size of the block (use 1 for size in bytes)
25  * @units:	units to use (powers of 1000 or 1024)
26  * @buf:	buffer to format to
27  * @len:	length of buffer
28  *
29  * This function returns a string formatted to 3 significant figures
30  * giving the size in the required units.  @buf should have room for
31  * at least 9 bytes and will always be zero terminated.
32  *
33  */
34 void string_get_size(u64 size, u64 blk_size, const enum string_size_units units,
35 		     char *buf, int len)
36 {
37 	static const char *const units_10[] = {
38 		"B", "kB", "MB", "GB", "TB", "PB", "EB", "ZB", "YB"
39 	};
40 	static const char *const units_2[] = {
41 		"B", "KiB", "MiB", "GiB", "TiB", "PiB", "EiB", "ZiB", "YiB"
42 	};
43 	static const char *const *const units_str[] = {
44 		[STRING_UNITS_10] = units_10,
45 		[STRING_UNITS_2] = units_2,
46 	};
47 	static const unsigned int divisor[] = {
48 		[STRING_UNITS_10] = 1000,
49 		[STRING_UNITS_2] = 1024,
50 	};
51 	static const unsigned int rounding[] = { 500, 50, 5 };
52 	int i = 0, j;
53 	u32 remainder = 0, sf_cap;
54 	char tmp[8];
55 	const char *unit;
56 
57 	tmp[0] = '\0';
58 
59 	if (blk_size == 0)
60 		size = 0;
61 	if (size == 0)
62 		goto out;
63 
64 	/* This is Napier's algorithm.  Reduce the original block size to
65 	 *
66 	 * coefficient * divisor[units]^i
67 	 *
68 	 * we do the reduction so both coefficients are just under 32 bits so
69 	 * that multiplying them together won't overflow 64 bits and we keep
70 	 * as much precision as possible in the numbers.
71 	 *
72 	 * Note: it's safe to throw away the remainders here because all the
73 	 * precision is in the coefficients.
74 	 */
75 	while (blk_size >> 32) {
76 		do_div(blk_size, divisor[units]);
77 		i++;
78 	}
79 
80 	while (size >> 32) {
81 		do_div(size, divisor[units]);
82 		i++;
83 	}
84 
85 	/* now perform the actual multiplication keeping i as the sum of the
86 	 * two logarithms */
87 	size *= blk_size;
88 
89 	/* and logarithmically reduce it until it's just under the divisor */
90 	while (size >= divisor[units]) {
91 		remainder = do_div(size, divisor[units]);
92 		i++;
93 	}
94 
95 	/* work out in j how many digits of precision we need from the
96 	 * remainder */
97 	sf_cap = size;
98 	for (j = 0; sf_cap*10 < 1000; j++)
99 		sf_cap *= 10;
100 
101 	if (units == STRING_UNITS_2) {
102 		/* express the remainder as a decimal.  It's currently the
103 		 * numerator of a fraction whose denominator is
104 		 * divisor[units], which is 1 << 10 for STRING_UNITS_2 */
105 		remainder *= 1000;
106 		remainder >>= 10;
107 	}
108 
109 	/* add a 5 to the digit below what will be printed to ensure
110 	 * an arithmetical round up and carry it through to size */
111 	remainder += rounding[j];
112 	if (remainder >= 1000) {
113 		remainder -= 1000;
114 		size += 1;
115 	}
116 
117 	if (j) {
118 		snprintf(tmp, sizeof(tmp), ".%03u", remainder);
119 		tmp[j+1] = '\0';
120 	}
121 
122  out:
123 	if (i >= ARRAY_SIZE(units_2))
124 		unit = "UNK";
125 	else
126 		unit = units_str[units][i];
127 
128 	snprintf(buf, len, "%u%s %s", (u32)size,
129 		 tmp, unit);
130 }
131 EXPORT_SYMBOL(string_get_size);
132 
133 static bool unescape_space(char **src, char **dst)
134 {
135 	char *p = *dst, *q = *src;
136 
137 	switch (*q) {
138 	case 'n':
139 		*p = '\n';
140 		break;
141 	case 'r':
142 		*p = '\r';
143 		break;
144 	case 't':
145 		*p = '\t';
146 		break;
147 	case 'v':
148 		*p = '\v';
149 		break;
150 	case 'f':
151 		*p = '\f';
152 		break;
153 	default:
154 		return false;
155 	}
156 	*dst += 1;
157 	*src += 1;
158 	return true;
159 }
160 
161 static bool unescape_octal(char **src, char **dst)
162 {
163 	char *p = *dst, *q = *src;
164 	u8 num;
165 
166 	if (isodigit(*q) == 0)
167 		return false;
168 
169 	num = (*q++) & 7;
170 	while (num < 32 && isodigit(*q) && (q - *src < 3)) {
171 		num <<= 3;
172 		num += (*q++) & 7;
173 	}
174 	*p = num;
175 	*dst += 1;
176 	*src = q;
177 	return true;
178 }
179 
180 static bool unescape_hex(char **src, char **dst)
181 {
182 	char *p = *dst, *q = *src;
183 	int digit;
184 	u8 num;
185 
186 	if (*q++ != 'x')
187 		return false;
188 
189 	num = digit = hex_to_bin(*q++);
190 	if (digit < 0)
191 		return false;
192 
193 	digit = hex_to_bin(*q);
194 	if (digit >= 0) {
195 		q++;
196 		num = (num << 4) | digit;
197 	}
198 	*p = num;
199 	*dst += 1;
200 	*src = q;
201 	return true;
202 }
203 
204 static bool unescape_special(char **src, char **dst)
205 {
206 	char *p = *dst, *q = *src;
207 
208 	switch (*q) {
209 	case '\"':
210 		*p = '\"';
211 		break;
212 	case '\\':
213 		*p = '\\';
214 		break;
215 	case 'a':
216 		*p = '\a';
217 		break;
218 	case 'e':
219 		*p = '\e';
220 		break;
221 	default:
222 		return false;
223 	}
224 	*dst += 1;
225 	*src += 1;
226 	return true;
227 }
228 
229 /**
230  * string_unescape - unquote characters in the given string
231  * @src:	source buffer (escaped)
232  * @dst:	destination buffer (unescaped)
233  * @size:	size of the destination buffer (0 to unlimit)
234  * @flags:	combination of the flags (bitwise OR):
235  *	%UNESCAPE_SPACE:
236  *		'\f' - form feed
237  *		'\n' - new line
238  *		'\r' - carriage return
239  *		'\t' - horizontal tab
240  *		'\v' - vertical tab
241  *	%UNESCAPE_OCTAL:
242  *		'\NNN' - byte with octal value NNN (1 to 3 digits)
243  *	%UNESCAPE_HEX:
244  *		'\xHH' - byte with hexadecimal value HH (1 to 2 digits)
245  *	%UNESCAPE_SPECIAL:
246  *		'\"' - double quote
247  *		'\\' - backslash
248  *		'\a' - alert (BEL)
249  *		'\e' - escape
250  *	%UNESCAPE_ANY:
251  *		all previous together
252  *
253  * Description:
254  * The function unquotes characters in the given string.
255  *
256  * Because the size of the output will be the same as or less than the size of
257  * the input, the transformation may be performed in place.
258  *
259  * Caller must provide valid source and destination pointers. Be aware that
260  * destination buffer will always be NULL-terminated. Source string must be
261  * NULL-terminated as well.
262  *
263  * Return:
264  * The amount of the characters processed to the destination buffer excluding
265  * trailing '\0' is returned.
266  */
267 int string_unescape(char *src, char *dst, size_t size, unsigned int flags)
268 {
269 	char *out = dst;
270 
271 	while (*src && --size) {
272 		if (src[0] == '\\' && src[1] != '\0' && size > 1) {
273 			src++;
274 			size--;
275 
276 			if (flags & UNESCAPE_SPACE &&
277 					unescape_space(&src, &out))
278 				continue;
279 
280 			if (flags & UNESCAPE_OCTAL &&
281 					unescape_octal(&src, &out))
282 				continue;
283 
284 			if (flags & UNESCAPE_HEX &&
285 					unescape_hex(&src, &out))
286 				continue;
287 
288 			if (flags & UNESCAPE_SPECIAL &&
289 					unescape_special(&src, &out))
290 				continue;
291 
292 			*out++ = '\\';
293 		}
294 		*out++ = *src++;
295 	}
296 	*out = '\0';
297 
298 	return out - dst;
299 }
300 EXPORT_SYMBOL(string_unescape);
301 
302 static bool escape_passthrough(unsigned char c, char **dst, char *end)
303 {
304 	char *out = *dst;
305 
306 	if (out < end)
307 		*out = c;
308 	*dst = out + 1;
309 	return true;
310 }
311 
312 static bool escape_space(unsigned char c, char **dst, char *end)
313 {
314 	char *out = *dst;
315 	unsigned char to;
316 
317 	switch (c) {
318 	case '\n':
319 		to = 'n';
320 		break;
321 	case '\r':
322 		to = 'r';
323 		break;
324 	case '\t':
325 		to = 't';
326 		break;
327 	case '\v':
328 		to = 'v';
329 		break;
330 	case '\f':
331 		to = 'f';
332 		break;
333 	default:
334 		return false;
335 	}
336 
337 	if (out < end)
338 		*out = '\\';
339 	++out;
340 	if (out < end)
341 		*out = to;
342 	++out;
343 
344 	*dst = out;
345 	return true;
346 }
347 
348 static bool escape_special(unsigned char c, char **dst, char *end)
349 {
350 	char *out = *dst;
351 	unsigned char to;
352 
353 	switch (c) {
354 	case '\\':
355 		to = '\\';
356 		break;
357 	case '\a':
358 		to = 'a';
359 		break;
360 	case '\e':
361 		to = 'e';
362 		break;
363 	default:
364 		return false;
365 	}
366 
367 	if (out < end)
368 		*out = '\\';
369 	++out;
370 	if (out < end)
371 		*out = to;
372 	++out;
373 
374 	*dst = out;
375 	return true;
376 }
377 
378 static bool escape_null(unsigned char c, char **dst, char *end)
379 {
380 	char *out = *dst;
381 
382 	if (c)
383 		return false;
384 
385 	if (out < end)
386 		*out = '\\';
387 	++out;
388 	if (out < end)
389 		*out = '0';
390 	++out;
391 
392 	*dst = out;
393 	return true;
394 }
395 
396 static bool escape_octal(unsigned char c, char **dst, char *end)
397 {
398 	char *out = *dst;
399 
400 	if (out < end)
401 		*out = '\\';
402 	++out;
403 	if (out < end)
404 		*out = ((c >> 6) & 0x07) + '0';
405 	++out;
406 	if (out < end)
407 		*out = ((c >> 3) & 0x07) + '0';
408 	++out;
409 	if (out < end)
410 		*out = ((c >> 0) & 0x07) + '0';
411 	++out;
412 
413 	*dst = out;
414 	return true;
415 }
416 
417 static bool escape_hex(unsigned char c, char **dst, char *end)
418 {
419 	char *out = *dst;
420 
421 	if (out < end)
422 		*out = '\\';
423 	++out;
424 	if (out < end)
425 		*out = 'x';
426 	++out;
427 	if (out < end)
428 		*out = hex_asc_hi(c);
429 	++out;
430 	if (out < end)
431 		*out = hex_asc_lo(c);
432 	++out;
433 
434 	*dst = out;
435 	return true;
436 }
437 
438 /**
439  * string_escape_mem - quote characters in the given memory buffer
440  * @src:	source buffer (unescaped)
441  * @isz:	source buffer size
442  * @dst:	destination buffer (escaped)
443  * @osz:	destination buffer size
444  * @flags:	combination of the flags (bitwise OR):
445  *	%ESCAPE_SPACE: (special white space, not space itself)
446  *		'\f' - form feed
447  *		'\n' - new line
448  *		'\r' - carriage return
449  *		'\t' - horizontal tab
450  *		'\v' - vertical tab
451  *	%ESCAPE_SPECIAL:
452  *		'\\' - backslash
453  *		'\a' - alert (BEL)
454  *		'\e' - escape
455  *	%ESCAPE_NULL:
456  *		'\0' - null
457  *	%ESCAPE_OCTAL:
458  *		'\NNN' - byte with octal value NNN (3 digits)
459  *	%ESCAPE_ANY:
460  *		all previous together
461  *	%ESCAPE_NP:
462  *		escape only non-printable characters (checked by isprint)
463  *	%ESCAPE_ANY_NP:
464  *		all previous together
465  *	%ESCAPE_HEX:
466  *		'\xHH' - byte with hexadecimal value HH (2 digits)
467  * @only:	NULL-terminated string containing characters used to limit
468  *		the selected escape class. If characters are included in @only
469  *		that would not normally be escaped by the classes selected
470  *		in @flags, they will be copied to @dst unescaped.
471  *
472  * Description:
473  * The process of escaping byte buffer includes several parts. They are applied
474  * in the following sequence.
475  *	1. The character is matched to the printable class, if asked, and in
476  *	   case of match it passes through to the output.
477  *	2. The character is not matched to the one from @only string and thus
478  *	   must go as-is to the output.
479  *	3. The character is checked if it falls into the class given by @flags.
480  *	   %ESCAPE_OCTAL and %ESCAPE_HEX are going last since they cover any
481  *	   character. Note that they actually can't go together, otherwise
482  *	   %ESCAPE_HEX will be ignored.
483  *
484  * Caller must provide valid source and destination pointers. Be aware that
485  * destination buffer will not be NULL-terminated, thus caller have to append
486  * it if needs.
487  *
488  * Return:
489  * The total size of the escaped output that would be generated for
490  * the given input and flags. To check whether the output was
491  * truncated, compare the return value to osz. There is room left in
492  * dst for a '\0' terminator if and only if ret < osz.
493  */
494 int string_escape_mem(const char *src, size_t isz, char *dst, size_t osz,
495 		      unsigned int flags, const char *only)
496 {
497 	char *p = dst;
498 	char *end = p + osz;
499 	bool is_dict = only && *only;
500 
501 	while (isz--) {
502 		unsigned char c = *src++;
503 
504 		/*
505 		 * Apply rules in the following sequence:
506 		 *	- the character is printable, when @flags has
507 		 *	  %ESCAPE_NP bit set
508 		 *	- the @only string is supplied and does not contain a
509 		 *	  character under question
510 		 *	- the character doesn't fall into a class of symbols
511 		 *	  defined by given @flags
512 		 * In these cases we just pass through a character to the
513 		 * output buffer.
514 		 */
515 		if ((flags & ESCAPE_NP && isprint(c)) ||
516 		    (is_dict && !strchr(only, c))) {
517 			/* do nothing */
518 		} else {
519 			if (flags & ESCAPE_SPACE && escape_space(c, &p, end))
520 				continue;
521 
522 			if (flags & ESCAPE_SPECIAL && escape_special(c, &p, end))
523 				continue;
524 
525 			if (flags & ESCAPE_NULL && escape_null(c, &p, end))
526 				continue;
527 
528 			/* ESCAPE_OCTAL and ESCAPE_HEX always go last */
529 			if (flags & ESCAPE_OCTAL && escape_octal(c, &p, end))
530 				continue;
531 
532 			if (flags & ESCAPE_HEX && escape_hex(c, &p, end))
533 				continue;
534 		}
535 
536 		escape_passthrough(c, &p, end);
537 	}
538 
539 	return p - dst;
540 }
541 EXPORT_SYMBOL(string_escape_mem);
542 
543 /*
544  * Return an allocated string that has been escaped of special characters
545  * and double quotes, making it safe to log in quotes.
546  */
547 char *kstrdup_quotable(const char *src, gfp_t gfp)
548 {
549 	size_t slen, dlen;
550 	char *dst;
551 	const int flags = ESCAPE_HEX;
552 	const char esc[] = "\f\n\r\t\v\a\e\\\"";
553 
554 	if (!src)
555 		return NULL;
556 	slen = strlen(src);
557 
558 	dlen = string_escape_mem(src, slen, NULL, 0, flags, esc);
559 	dst = kmalloc(dlen + 1, gfp);
560 	if (!dst)
561 		return NULL;
562 
563 	WARN_ON(string_escape_mem(src, slen, dst, dlen, flags, esc) != dlen);
564 	dst[dlen] = '\0';
565 
566 	return dst;
567 }
568 EXPORT_SYMBOL_GPL(kstrdup_quotable);
569 
570 /*
571  * Returns allocated NULL-terminated string containing process
572  * command line, with inter-argument NULLs replaced with spaces,
573  * and other special characters escaped.
574  */
575 char *kstrdup_quotable_cmdline(struct task_struct *task, gfp_t gfp)
576 {
577 	char *buffer, *quoted;
578 	int i, res;
579 
580 	buffer = kmalloc(PAGE_SIZE, GFP_KERNEL);
581 	if (!buffer)
582 		return NULL;
583 
584 	res = get_cmdline(task, buffer, PAGE_SIZE - 1);
585 	buffer[res] = '\0';
586 
587 	/* Collapse trailing NULLs, leave res pointing to last non-NULL. */
588 	while (--res >= 0 && buffer[res] == '\0')
589 		;
590 
591 	/* Replace inter-argument NULLs. */
592 	for (i = 0; i <= res; i++)
593 		if (buffer[i] == '\0')
594 			buffer[i] = ' ';
595 
596 	/* Make sure result is printable. */
597 	quoted = kstrdup_quotable(buffer, gfp);
598 	kfree(buffer);
599 	return quoted;
600 }
601 EXPORT_SYMBOL_GPL(kstrdup_quotable_cmdline);
602 
603 /*
604  * Returns allocated NULL-terminated string containing pathname,
605  * with special characters escaped, able to be safely logged. If
606  * there is an error, the leading character will be "<".
607  */
608 char *kstrdup_quotable_file(struct file *file, gfp_t gfp)
609 {
610 	char *temp, *pathname;
611 
612 	if (!file)
613 		return kstrdup("<unknown>", gfp);
614 
615 	/* We add 11 spaces for ' (deleted)' to be appended */
616 	temp = kmalloc(PATH_MAX + 11, GFP_KERNEL);
617 	if (!temp)
618 		return kstrdup("<no_memory>", gfp);
619 
620 	pathname = file_path(file, temp, PATH_MAX + 11);
621 	if (IS_ERR(pathname))
622 		pathname = kstrdup("<too_long>", gfp);
623 	else
624 		pathname = kstrdup_quotable(pathname, gfp);
625 
626 	kfree(temp);
627 	return pathname;
628 }
629 EXPORT_SYMBOL_GPL(kstrdup_quotable_file);
630