coreboot-libre-fam15h-rdimm/3rdparty/chromeec/common/util.c

656 lines
12 KiB
C
Raw Permalink Normal View History

2024-03-04 11:14:53 +01:00
/* Copyright 2013 The Chromium OS Authors. All rights reserved.
* Use of this source code is governed by a BSD-style license that can be
* found in the LICENSE file.
*/
/* Utility functions for Chrome EC */
#include "common.h"
#include "console.h"
#include "util.h"
__stdlib_compat size_t strlen(const char *s)
{
int len = 0;
while (*s++)
len++;
return len;
}
__stdlib_compat size_t strnlen(const char *s, size_t maxlen)
{
size_t len = 0;
while (len < maxlen && *s) {
s++;
len++;
}
return len;
}
__stdlib_compat int isspace(int c)
{
return c == ' ' || c == '\t' || c == '\r' || c == '\n';
}
__stdlib_compat int isdigit(int c)
{
return c >= '0' && c <= '9';
}
__stdlib_compat int isalpha(int c)
{
return (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z');
}
__stdlib_compat int isupper(int c)
{
return c >= 'A' && c <= 'Z';
}
__stdlib_compat int isprint(int c)
{
return c >= ' ' && c <= '~';
}
__stdlib_compat int tolower(int c)
{
return c >= 'A' && c <= 'Z' ? c + 'a' - 'A' : c;
}
__stdlib_compat int strcasecmp(const char *s1, const char *s2)
{
int diff;
do {
diff = tolower(*s1) - tolower(*s2);
if (diff)
return diff;
} while (*(s1++) && *(s2++));
return 0;
}
__stdlib_compat int strncasecmp(const char *s1, const char *s2, size_t size)
{
int diff;
if (!size)
return 0;
do {
diff = tolower(*s1) - tolower(*s2);
if (diff)
return diff;
} while (*(s1++) && *(s2++) && --size);
return 0;
}
__stdlib_compat char *strstr(const char *s1, const char *s2)
{
const char *p, *q, *r;
size_t len1 = strlen(s1);
size_t len2 = strlen(s2);
if (len1 == 0 || len2 == 0 || len1 < len2)
return NULL;
r = s1 + len1 - len2 + 1;
for (; s1 < r; s1++) {
if (*s1 == *s2) {
p = s1 + 1;
q = s2 + 1;
for (; q < s2 + len2;) {
if (*p++ != *q++)
break;
}
if (*q == '\0')
return (char *)s1;
}
}
return NULL;
}
__stdlib_compat int atoi(const char *nptr)
{
int result = 0;
int neg = 0;
char c = '\0';
while ((c = *nptr++) && isspace(c))
;
if (c == '-') {
neg = 1;
c = *nptr++;
}
while (isdigit(c)) {
result = result * 10 + (c - '0');
c = *nptr++;
}
return neg ? -result : result;
}
static int find_base(int base, int *c, const char **nptr) {
if ((base == 0 || base == 16) && *c == '0'
&& (**nptr == 'x' || **nptr == 'X')) {
*c = (*nptr)[1];
(*nptr) += 2;
base = 16;
} else if (base == 0) {
base = *c == '0' ? 8 : 10;
}
return base;
}
/* Like strtol(), but for integers */
__stdlib_compat int strtoi(const char *nptr, char **endptr, int base)
{
int result = 0;
int neg = 0;
int c = '\0';
while ((c = *nptr++) && isspace(c))
;
if (c == '+') {
c = *nptr++;
} else if (c == '-') {
neg = 1;
c = *nptr++;
}
base = find_base(base, &c, &nptr);
while (c) {
if (c >= '0' && c < '0' + MIN(base, 10))
result = result * base + (c - '0');
else if (c >= 'A' && c < 'A' + base - 10)
result = result * base + (c - 'A' + 10);
else if (c >= 'a' && c < 'a' + base - 10)
result = result * base + (c - 'a' + 10);
else
break;
c = *nptr++;
}
if (endptr)
*endptr = (char *)nptr - 1;
return neg ? -result : result;
}
__stdlib_compat uint64_t strtoul(const char *nptr, char **endptr, int base)
{
uint64_t result = 0;
int c = '\0';
while ((c = *nptr++) && isspace(c))
;
if (c == '+') {
c = *nptr++;
} else if (c == '-') {
if (endptr)
*endptr = (char *)nptr - 1;
return result;
}
base = find_base(base, &c, &nptr);
while (c) {
if (c >= '0' && c < '0' + MIN(base, 10))
result = result * base + (c - '0');
else if (c >= 'A' && c < 'A' + base - 10)
result = result * base + (c - 'A' + 10);
else if (c >= 'a' && c < 'a' + base - 10)
result = result * base + (c - 'a' + 10);
else
break;
c = *nptr++;
}
if (endptr)
*endptr = (char *)nptr - 1;
return result;
}
__stdlib_compat int parse_bool(const char *s, int *dest)
{
/* off, disable, false, no */
if (!strcasecmp(s, "off") || !strncasecmp(s, "dis", 3) ||
tolower(*s) == 'f' || tolower(*s) == 'n') {
*dest = 0;
return 1;
}
/* on, enable, true, yes */
if (!strcasecmp(s, "on") || !strncasecmp(s, "ena", 3) ||
tolower(*s) == 't' || tolower(*s) == 'y') {
*dest = 1;
return 1;
}
/* dunno */
return 0;
}
__stdlib_compat int memcmp(const void *s1, const void *s2, size_t len)
{
const char *sa = s1;
const char *sb = s2;
int diff = 0;
while (len-- > 0) {
diff = *(sa++) - *(sb++);
if (diff)
return diff;
}
return 0;
}
/* Constant-time memory comparison */
int safe_memcmp(const void *s1, const void *s2, size_t size)
{
const uint8_t *us1 = s1;
const uint8_t *us2 = s2;
int result = 0;
if (size == 0)
return 0;
/*
* Code snippet without data-dependent branch due to Nate Lawson
* (nate@root.org) of Root Labs.
*/
while (size--)
result |= *us1++ ^ *us2++;
return result != 0;
}
__stdlib_compat void *memcpy(void *dest, const void *src, size_t len)
{
char *d = (char *)dest;
const char *s = (const char *)src;
uint32_t *dw;
const uint32_t *sw;
char *head;
char * const tail = (char *)dest + len;
/* Set 'body' to the last word boundary */
uint32_t * const body = (uint32_t *)((uintptr_t)tail & ~3);
if (((uintptr_t)dest & 3) != ((uintptr_t)src & 3)) {
/* Misaligned. no body, no tail. */
head = tail;
} else {
/* Aligned */
if ((uintptr_t)tail < (((uintptr_t)d + 3) & ~3))
/* len is shorter than the first word boundary */
head = tail;
else
/* Set 'head' to the first word boundary */
head = (char *)(((uintptr_t)d + 3) & ~3);
}
/* Copy head */
while (d < head)
*(d++) = *(s++);
/* Copy body */
dw = (uint32_t *)d;
sw = (uint32_t *)s;
while (dw < body)
*(dw++) = *(sw++);
/* Copy tail */
d = (char *)dw;
s = (const char *)sw;
while (d < tail)
*(d++) = *(s++);
return dest;
}
__stdlib_compat __visible void *memset(void *dest, int c, size_t len)
{
char *d = (char *)dest;
uint32_t cccc;
uint32_t *dw;
char *head;
char * const tail = (char *)dest + len;
/* Set 'body' to the last word boundary */
uint32_t * const body = (uint32_t *)((uintptr_t)tail & ~3);
c &= 0xff; /* Clear upper bits before ORing below */
cccc = c | (c << 8) | (c << 16) | (c << 24);
if ((uintptr_t)tail < (((uintptr_t)d + 3) & ~3))
/* len is shorter than the first word boundary */
head = tail;
else
/* Set 'head' to the first word boundary */
head = (char *)(((uintptr_t)d + 3) & ~3);
/* Copy head */
while (d < head)
*(d++) = c;
/* Copy body */
dw = (uint32_t *)d;
while (dw < body)
*(dw++) = cccc;
/* Copy tail */
d = (char *)dw;
while (d < tail)
*(d++) = c;
return dest;
}
__stdlib_compat void *memmove(void *dest, const void *src, size_t len)
{
if ((uintptr_t)dest <= (uintptr_t)src ||
(uintptr_t)dest >= (uintptr_t)src + len) {
/* Start of destination doesn't overlap source, so just use
* memcpy(). */
return memcpy(dest, src, len);
} else {
/* Need to copy from tail because there is overlap. */
char *d = (char *)dest + len;
const char *s = (const char *)src + len;
uint32_t *dw;
const uint32_t *sw;
char *head;
char * const tail = (char *)dest;
/* Set 'body' to the last word boundary */
uint32_t * const body = (uint32_t *)(((uintptr_t)tail+3) & ~3);
if (((uintptr_t)dest & 3) != ((uintptr_t)src & 3)) {
/* Misaligned. no body, no tail. */
head = tail;
} else {
/* Aligned */
if ((uintptr_t)tail > ((uintptr_t)d & ~3))
/* Shorter than the first word boundary */
head = tail;
else
/* Set 'head' to the first word boundary */
head = (char *)((uintptr_t)d & ~3);
}
/* Copy head */
while (d > head)
*(--d) = *(--s);
/* Copy body */
dw = (uint32_t *)d;
sw = (uint32_t *)s;
while (dw > body)
*(--dw) = *(--sw);
/* Copy tail */
d = (char *)dw;
s = (const char *)sw;
while (d > tail)
*(--d) = *(--s);
return dest;
}
}
__stdlib_compat void *memchr(const void *buffer, int c, size_t n)
{
char *current = (char *)buffer;
char *end = current + n;
while (current != end) {
if (*current == c)
return current;
current++;
}
return NULL;
}
void reverse(void *dest, size_t len)
{
int i;
uint8_t *start = dest;
uint8_t *end = start + len;
for (i = 0; i < len / 2; ++i) {
uint8_t tmp = *start;
*start++ = *--end;
*end = tmp;
}
}
__stdlib_compat char *strzcpy(char *dest, const char *src, int len)
{
char *d = dest;
if (len <= 0)
return dest;
while (len > 1 && *src) {
*(d++) = *(src++);
len--;
}
*d = '\0';
return dest;
}
__stdlib_compat char *strncpy(char *dest, const char *src, size_t n)
{
char *d = dest;
while (n && *src) {
*d++ = *src++;
n--;
}
if (n)
*d = '\0';
return dest;
}
__stdlib_compat int strncmp(const char *s1, const char *s2, size_t n)
{
while (n--) {
if (*s1 != *s2)
return *s1 - *s2;
if (!*s1)
break;
s1++;
s2++;
}
return 0;
}
int uint64divmod(uint64_t *n, int d)
{
uint64_t q = 0, mask;
int r = 0;
/* Divide-by-zero returns zero */
if (!d) {
*n = 0;
return 0;
}
/* Common powers of 2 = simple shifts */
if (d == 2) {
r = *n & 1;
*n >>= 1;
return r;
} else if (d == 16) {
r = *n & 0xf;
*n >>= 4;
return r;
}
/* If v fits in 32-bit, we're done. */
if (*n <= 0xffffffff) {
uint32_t v32 = *n;
r = v32 % d;
*n = v32 / d;
return r;
}
/* Otherwise do integer division the slow way. */
for (mask = (1ULL << 63); mask; mask >>= 1) {
r <<= 1;
if (*n & mask)
r |= 1;
if (r >= d) {
r -= d;
q |= mask;
}
}
*n = q;
return r;
}
int get_next_bit(uint32_t *mask)
{
int bit = 31 - __builtin_clz(*mask);
*mask &= ~BIT(bit);
return bit;
}
/****************************************************************************/
/* stateful conditional stuff */
enum cond_internal_bits {
COND_CURR_MASK = BIT(0), /* current value */
COND_RISE_MASK = BIT(1), /* set if 0->1 */
COND_FALL_MASK = BIT(2), /* set if 1->0 */
};
void cond_init(cond_t *c, int val)
{
if (val)
*c = COND_CURR_MASK;
else
*c = 0;
}
int cond_is(cond_t *c, int val)
{
if (val)
return *c & COND_CURR_MASK;
else
return !(*c & COND_CURR_MASK);
}
void cond_set(cond_t *c, int val)
{
if (val && cond_is(c, 0))
*c |= COND_RISE_MASK;
else if (!val && cond_is(c, 1))
*c |= COND_FALL_MASK;
if (val)
*c |= COND_CURR_MASK;
else
*c &= ~COND_CURR_MASK;
}
int cond_went(cond_t *c, int val)
{
int ret;
if (val) {
ret = *c & COND_RISE_MASK;
*c &= ~COND_RISE_MASK;
} else {
ret = *c & COND_FALL_MASK;
*c &= ~COND_FALL_MASK;
}
return ret;
}
/****************************************************************************/
/* console command parsing */
/**
* Parse offset and size from command line argv[shift] and argv[shift+1]
*
* Default values: If argc<=shift, leaves offset unchanged, returning error if
* *offset<0. If argc<shift+1, leaves size unchanged, returning error if
* *size<0.
*/
int parse_offset_size(int argc, char **argv, int shift,
int *offset, int *size)
{
char *e;
int i;
if (argc > shift) {
i = (uint32_t)strtoi(argv[shift], &e, 0);
if (*e)
return EC_ERROR_PARAM1;
*offset = i;
} else if (*offset < 0)
return EC_ERROR_PARAM_COUNT;
if (argc > shift + 1) {
i = (uint32_t)strtoi(argv[shift + 1], &e, 0);
if (*e)
return EC_ERROR_PARAM2;
*size = i;
} else if (*size < 0)
return EC_ERROR_PARAM_COUNT;
return EC_SUCCESS;
}
void hexdump(const uint8_t *data, int len)
{
int i, j;
if (!data || !len)
return;
for (i = 0; i < len; i += 16) {
/* Left column (Hex) */
for (j = i; j < i + 16; j++) {
if (j < len)
ccprintf(" %02x", data[j]);
else
ccprintf(" ");
}
/* Right column (ASCII) */
ccprintf(" |");
for (j = i; j < i + 16; j++) {
int c = j < len ? data[j] : ' ';
ccprintf("%c", isprint(c) ? c : '.');
}
ccprintf("|\n");
}
}