The branch main has been updated by arichardson: URL: https://cgit.FreeBSD.org/src/commit/?id=d053fb22f6d3b6bb0f4e47e4507fefc20cbe0e32
commit d053fb22f6d3b6bb0f4e47e4507fefc20cbe0e32 Author: Alex Richardson <[email protected]> AuthorDate: 2021-07-05 13:32:48 +0000 Commit: Alex Richardson <[email protected]> CommitDate: 2021-07-06 09:51:05 +0000 usr.bin/sort: Avoid UBSan errors UBSan complains about out-of-bounds accesses for zero-length arrays. To avoid this we can use flexible array members. However, the C standard does not allow for structures that only contain flexible array members, so we move the length parameters into that structure too. Split out from D28233. Reviewed By: markj MFC after: 1 week Differential Revision: https://reviews.freebsd.org/D31009 --- usr.bin/sort/Makefile | 2 +- usr.bin/sort/bwstring.c | 267 ++++++++++++++++++++++++------------------------ usr.bin/sort/bwstring.h | 31 +++--- 3 files changed, 155 insertions(+), 145 deletions(-) diff --git a/usr.bin/sort/Makefile b/usr.bin/sort/Makefile index d38e036b90c0..3f7b607a177a 100644 --- a/usr.bin/sort/Makefile +++ b/usr.bin/sort/Makefile @@ -3,8 +3,8 @@ .include <src.opts.mk> PROG= sort - SRCS= bwstring.c coll.c file.c mem.c radixsort.c sort.c vsort.c +CSTD= c11 sort.1: sort.1.in sed ${MAN_SUB} ${.ALLSRC} >${.TARGET} diff --git a/usr.bin/sort/bwstring.c b/usr.bin/sort/bwstring.c index f6200c53c83e..c31cb859cb37 100644 --- a/usr.bin/sort/bwstring.c +++ b/usr.bin/sort/bwstring.c @@ -113,7 +113,7 @@ initialise_months(void) static int wide_str_coll(const wchar_t *s1, const wchar_t *s2) { - int ret = 0; + int ret; errno = 0; ret = wcscoll(s1, s2); @@ -144,41 +144,44 @@ bwsprintf(FILE *f, struct bwstring *bws, const char *prefix, const char *suffix) { if (mb_cur_max == 1) - fprintf(f, "%s%s%s", prefix, bws->data.cstr, suffix); + fprintf(f, "%s%s%s", prefix, bws->cdata.str, suffix); else - fprintf(f, "%s%S%s", prefix, bws->data.wstr, suffix); + fprintf(f, "%s%S%s", prefix, bws->wdata.str, suffix); } const void* bwsrawdata(const struct bwstring *bws) { - return (&(bws->data)); + return (&(bws->wdata)); } size_t bwsrawlen(const struct bwstring *bws) { - return ((mb_cur_max == 1) ? bws->len : SIZEOF_WCHAR_STRING(bws->len)); + return ((mb_cur_max == 1) ? bws->cdata.len : + SIZEOF_WCHAR_STRING(bws->wdata.len)); } size_t bws_memsize(const struct bwstring *bws) { - return ((mb_cur_max == 1) ? (bws->len + 2 + sizeof(struct bwstring)) : - (SIZEOF_WCHAR_STRING(bws->len + 1) + sizeof(struct bwstring))); + return ((mb_cur_max == 1) ? + (bws->cdata.len + 2 + sizeof(struct bwstring)) : + (SIZEOF_WCHAR_STRING(bws->wdata.len + 1) + sizeof(struct bwstring))); } void bws_setlen(struct bwstring *bws, size_t newlen) { - if (bws && newlen != bws->len && newlen <= bws->len) { - bws->len = newlen; - if (mb_cur_max == 1) - bws->data.cstr[newlen] = '\0'; - else - bws->data.wstr[newlen] = L'\0'; + if (mb_cur_max == 1 && bws && newlen != bws->cdata.len && + newlen <= bws->cdata.len) { + bws->cdata.len = newlen; + bws->cdata.str[newlen] = '\0'; + } else if (bws && newlen != bws->wdata.len && newlen <= bws->wdata.len) { + bws->wdata.len = newlen; + bws->wdata.str[newlen] = L'\0'; } } @@ -190,17 +193,16 @@ bwsalloc(size_t sz) { struct bwstring *ret; - if (mb_cur_max == 1) + if (mb_cur_max == 1) { ret = sort_malloc(sizeof(struct bwstring) + 1 + sz); - else - ret = sort_malloc(sizeof(struct bwstring) + - SIZEOF_WCHAR_STRING(sz + 1)); - ret->len = sz; - - if (mb_cur_max == 1) - ret->data.cstr[ret->len] = '\0'; - else - ret->data.wstr[ret->len] = L'\0'; + ret->cdata.len = sz; + ret->cdata.str[sz] = '\0'; + } else { + ret = sort_malloc( + sizeof(struct bwstring) + SIZEOF_WCHAR_STRING(sz + 1)); + ret->wdata.len = sz; + ret->wdata.str[sz] = L'\0'; + } return (ret); } @@ -216,13 +218,13 @@ bwsdup(const struct bwstring *s) if (s == NULL) return (NULL); else { - struct bwstring *ret = bwsalloc(s->len); + struct bwstring *ret = bwsalloc(BWSLEN(s)); if (mb_cur_max == 1) - memcpy(ret->data.cstr, s->data.cstr, (s->len)); + memcpy(ret->cdata.str, s->cdata.str, (s->cdata.len)); else - memcpy(ret->data.wstr, s->data.wstr, - SIZEOF_WCHAR_STRING(s->len)); + memcpy(ret->wdata.str, s->wdata.str, + SIZEOF_WCHAR_STRING(s->wdata.len)); return (ret); } @@ -244,9 +246,9 @@ bwssbdup(const wchar_t *str, size_t len) if (mb_cur_max == 1) for (size_t i = 0; i < len; ++i) - ret->data.cstr[i] = (unsigned char) str[i]; + ret->cdata.str[i] = (char)str[i]; else - memcpy(ret->data.wstr, str, SIZEOF_WCHAR_STRING(len)); + memcpy(ret->wdata.str, str, SIZEOF_WCHAR_STRING(len)); return (ret); } @@ -264,7 +266,7 @@ bwscsbdup(const unsigned char *str, size_t len) if (str) { if (mb_cur_max == 1) - memcpy(ret->data.cstr, str, len); + memcpy(ret->cdata.str, str, len); else { mbstate_t mbs; const char *s; @@ -288,12 +290,12 @@ bwscsbdup(const unsigned char *str, size_t len) case (size_t) -1: /* FALLTHROUGH */ case (size_t) -2: - ret->data.wstr[chars++] = + ret->wdata.str[chars++] = (unsigned char) s[cptr]; ++cptr; break; default: - n = mbrtowc(ret->data.wstr + (chars++), + n = mbrtowc(ret->wdata.str + (chars++), s + cptr, charlen, &mbs); if ((n == (size_t)-1) || (n == (size_t)-2)) /* NOTREACHED */ @@ -302,8 +304,8 @@ bwscsbdup(const unsigned char *str, size_t len) } } - ret->len = chars; - ret->data.wstr[ret->len] = L'\0'; + ret->wdata.len = chars; + ret->wdata.str[ret->wdata.len] = L'\0'; } } return (ret); @@ -328,19 +330,20 @@ bwsfree(const struct bwstring *s) size_t bwscpy(struct bwstring *dst, const struct bwstring *src) { - size_t nums = src->len; + size_t nums = BWSLEN(src); - if (nums > dst->len) - nums = dst->len; - dst->len = nums; + if (nums > BWSLEN(dst)) + nums = BWSLEN(dst); if (mb_cur_max == 1) { - memcpy(dst->data.cstr, src->data.cstr, nums); - dst->data.cstr[dst->len] = '\0'; + memcpy(dst->cdata.str, src->cdata.str, nums); + dst->cdata.len = nums; + dst->cdata.str[dst->cdata.len] = '\0'; } else { - memcpy(dst->data.wstr, src->data.wstr, - SIZEOF_WCHAR_STRING(nums + 1)); - dst->data.wstr[dst->len] = L'\0'; + memcpy(dst->wdata.str, src->wdata.str, + SIZEOF_WCHAR_STRING(nums)); + dst->wdata.len = nums; + dst->wdata.str[nums] = L'\0'; } return (nums); @@ -355,21 +358,22 @@ bwscpy(struct bwstring *dst, const struct bwstring *src) struct bwstring * bwsncpy(struct bwstring *dst, const struct bwstring *src, size_t size) { - size_t nums = src->len; + size_t nums = BWSLEN(src); - if (nums > dst->len) - nums = dst->len; + if (nums > BWSLEN(dst)) + nums = BWSLEN(dst); if (nums > size) nums = size; - dst->len = nums; if (mb_cur_max == 1) { - memcpy(dst->data.cstr, src->data.cstr, nums); - dst->data.cstr[dst->len] = '\0'; + memcpy(dst->cdata.str, src->cdata.str, nums); + dst->cdata.len = nums; + dst->cdata.str[nums] = '\0'; } else { - memcpy(dst->data.wstr, src->data.wstr, - SIZEOF_WCHAR_STRING(nums + 1)); - dst->data.wstr[dst->len] = L'\0'; + memcpy(dst->wdata.str, src->wdata.str, + SIZEOF_WCHAR_STRING(nums)); + dst->wdata.len = nums; + dst->wdata.str[nums] = L'\0'; } return (dst); @@ -387,25 +391,24 @@ bwsnocpy(struct bwstring *dst, const struct bwstring *src, size_t offset, size_t size) { - if (offset >= src->len) { - dst->data.wstr[0] = 0; - dst->len = 0; + if (offset >= BWSLEN(src)) { + bws_setlen(dst, 0); } else { - size_t nums = src->len - offset; + size_t nums = BWSLEN(src) - offset; - if (nums > dst->len) - nums = dst->len; + if (nums > BWSLEN(dst)) + nums = BWSLEN(dst); if (nums > size) nums = size; - dst->len = nums; if (mb_cur_max == 1) { - memcpy(dst->data.cstr, src->data.cstr + offset, - (nums)); - dst->data.cstr[dst->len] = '\0'; + memcpy(dst->cdata.str, src->cdata.str + offset, nums); + dst->cdata.len = nums; + dst->cdata.str[nums] = '\0'; } else { - memcpy(dst->data.wstr, src->data.wstr + offset, + memcpy(dst->wdata.str, src->wdata.str + offset, SIZEOF_WCHAR_STRING(nums)); - dst->data.wstr[dst->len] = L'\0'; + dst->wdata.len = nums; + dst->wdata.str[nums] = L'\0'; } } return (dst); @@ -421,16 +424,16 @@ bwsfwrite(struct bwstring *bws, FILE *f, bool zero_ended) { if (mb_cur_max == 1) { - size_t len = bws->len; + size_t len = bws->cdata.len; if (!zero_ended) { - bws->data.cstr[len] = '\n'; + bws->cdata.str[len] = '\n'; - if (fwrite(bws->data.cstr, len + 1, 1, f) < 1) + if (fwrite(bws->cdata.str, len + 1, 1, f) < 1) err(2, NULL); - bws->data.cstr[len] = '\0'; - } else if (fwrite(bws->data.cstr, len + 1, 1, f) < 1) + bws->cdata.str[len] = '\0'; + } else if (fwrite(bws->cdata.str, len + 1, 1, f) < 1) err(2, NULL); return (len + 1); @@ -442,7 +445,7 @@ bwsfwrite(struct bwstring *bws, FILE *f, bool zero_ended) eols = zero_ended ? btowc('\0') : btowc('\n'); while (printed < BWSLEN(bws)) { - const wchar_t *s = bws->data.wstr + printed; + const wchar_t *s = bws->wdata.str + printed; if (*s == L'\0') { int nums; @@ -508,7 +511,7 @@ bwsfgetln(FILE *f, size_t *len, bool zero_ended, struct reader_buffer *rb) if (ret[*len - 1] == '\n') --(*len); } - return (bwscsbdup((unsigned char*)ret, *len)); + return (bwscsbdup((unsigned char *)ret, *len)); } else { *len = 0; @@ -548,7 +551,7 @@ bwsfgetln(FILE *f, size_t *len, bool zero_ended, struct reader_buffer *rb) } else while (!feof(f)) { - wint_t c = 0; + wint_t c; c = fgetwc(f); @@ -581,10 +584,10 @@ bwsncmp(const struct bwstring *bws1, const struct bwstring *bws2, size_t offset, size_t len) { size_t cmp_len, len1, len2; - int res = 0; + int res; - len1 = bws1->len; - len2 = bws2->len; + len1 = BWSLEN(bws1); + len2 = BWSLEN(bws2); if (len1 <= offset) { return ((len2 <= offset) ? 0 : -1); @@ -604,18 +607,18 @@ bwsncmp(const struct bwstring *bws1, const struct bwstring *bws2, cmp_len = len; if (mb_cur_max == 1) { - const unsigned char *s1, *s2; + const char *s1, *s2; - s1 = bws1->data.cstr + offset; - s2 = bws2->data.cstr + offset; + s1 = bws1->cdata.str + offset; + s2 = bws2->cdata.str + offset; res = memcmp(s1, s2, cmp_len); } else { const wchar_t *s1, *s2; - s1 = bws1->data.wstr + offset; - s2 = bws2->data.wstr + offset; + s1 = bws1->wdata.str + offset; + s2 = bws2->wdata.str + offset; res = memcmp(s1, s2, SIZEOF_WCHAR_STRING(cmp_len)); } @@ -638,8 +641,8 @@ bwscmp(const struct bwstring *bws1, const struct bwstring *bws2, size_t offset) size_t len1, len2, cmp_len; int res; - len1 = bws1->len; - len2 = bws2->len; + len1 = BWSLEN(bws1); + len2 = BWSLEN(bws2); len1 -= offset; len2 -= offset; @@ -665,7 +668,7 @@ int bws_iterator_cmp(bwstring_iterator iter1, bwstring_iterator iter2, size_t len) { wchar_t c1, c2; - size_t i = 0; + size_t i; for (i = 0; i < len; ++i) { c1 = bws_get_iter_value(iter1); @@ -684,8 +687,8 @@ bwscoll(const struct bwstring *bws1, const struct bwstring *bws2, size_t offset) { size_t len1, len2; - len1 = bws1->len; - len2 = bws2->len; + len1 = BWSLEN(bws1); + len2 = BWSLEN(bws2); if (len1 <= offset) return ((len2 <= offset) ? 0 : -1); @@ -697,13 +700,13 @@ bwscoll(const struct bwstring *bws1, const struct bwstring *bws2, size_t offset) len2 -= offset; if (mb_cur_max == 1) { - const unsigned char *s1, *s2; + const char *s1, *s2; - s1 = bws1->data.cstr + offset; - s2 = bws2->data.cstr + offset; + s1 = bws1->cdata.str + offset; + s2 = bws2->cdata.str + offset; if (byte_sort) { - int res = 0; + int res; if (len1 > len2) { res = memcmp(s1, s2, len2); @@ -719,7 +722,7 @@ bwscoll(const struct bwstring *bws1, const struct bwstring *bws2, size_t offset) return (res); } else { - int res = 0; + int res; size_t i, maxlen; i = 0; @@ -780,10 +783,10 @@ bwscoll(const struct bwstring *bws1, const struct bwstring *bws2, size_t offset) } else { const wchar_t *s1, *s2; size_t i, maxlen; - int res = 0; + int res; - s1 = bws1->data.wstr + offset; - s2 = bws2->data.wstr + offset; + s1 = bws1->wdata.str + offset; + s2 = bws2->wdata.str + offset; i = 0; maxlen = len1; @@ -850,14 +853,14 @@ bwscoll(const struct bwstring *bws1, const struct bwstring *bws2, size_t offset) double bwstod(struct bwstring *s0, bool *empty) { - double ret = 0; + double ret; if (mb_cur_max == 1) { - unsigned char *end, *s; + char *end, *s; char *ep; - s = s0->data.cstr; - end = s + s0->len; + s = s0->cdata.str; + end = s + s0->cdata.len; ep = NULL; while (isblank(*s) && s < end) @@ -869,15 +872,15 @@ bwstod(struct bwstring *s0, bool *empty) } ret = strtod((char*)s, &ep); - if ((unsigned char*) ep == s) { + if (ep == s) { *empty = true; return (0); } } else { wchar_t *end, *ep, *s; - s = s0->data.wstr; - end = s + s0->len; + s = s0->wdata.str; + end = s + s0->wdata.len; ep = NULL; while (iswblank(*s) && s < end) @@ -910,25 +913,25 @@ bws_month_score(const struct bwstring *s0) { if (mb_cur_max == 1) { - const unsigned char *end, *s; + const char *end, *s; - s = s0->data.cstr; - end = s + s0->len; + s = s0->cdata.str; + end = s + s0->cdata.len; while (isblank(*s) && s < end) ++s; for (int i = 11; i >= 0; --i) { if (cmonths[i] && - (s == (unsigned char*)strstr((const char*)s, (char*)(cmonths[i])))) + (s == strstr(s, cmonths[i]))) return (i); } } else { const wchar_t *end, *s; - s = s0->data.wstr; - end = s + s0->len; + s = s0->wdata.str; + end = s + s0->wdata.len; while (iswblank(*s) && s < end) ++s; @@ -950,11 +953,11 @@ ignore_leading_blanks(struct bwstring *str) { if (mb_cur_max == 1) { - unsigned char *dst, *end, *src; + char *dst, *end, *src; - src = str->data.cstr; + src = str->cdata.str; dst = src; - end = src + str->len; + end = src + str->cdata.len; while (src < end && isblank(*src)) ++src; @@ -974,9 +977,9 @@ ignore_leading_blanks(struct bwstring *str) } else { wchar_t *dst, *end, *src; - src = str->data.wstr; + src = str->wdata.str; dst = src; - end = src + str->len; + end = src + str->wdata.len; while (src < end && iswblank(*src)) ++src; @@ -1003,15 +1006,15 @@ ignore_leading_blanks(struct bwstring *str) struct bwstring * ignore_nonprinting(struct bwstring *str) { - size_t newlen = str->len; + size_t newlen = BWSLEN(str); if (mb_cur_max == 1) { - unsigned char *dst, *end, *src; - unsigned char c; + char *dst, *end, *src; + char c; - src = str->data.cstr; + src = str->cdata.str; dst = src; - end = src + str->len; + end = src + str->cdata.len; while (src < end) { c = *src; @@ -1028,9 +1031,9 @@ ignore_nonprinting(struct bwstring *str) wchar_t *dst, *end, *src; wchar_t c; - src = str->data.wstr; + src = str->wdata.str; dst = src; - end = src + str->len; + end = src + str->wdata.len; while (src < end) { c = *src; @@ -1056,15 +1059,15 @@ ignore_nonprinting(struct bwstring *str) struct bwstring * dictionary_order(struct bwstring *str) { - size_t newlen = str->len; + size_t newlen = BWSLEN(str); if (mb_cur_max == 1) { - unsigned char *dst, *end, *src; - unsigned char c; + char *dst, *end, *src; + char c; - src = str->data.cstr; + src = str->cdata.str; dst = src; - end = src + str->len; + end = src + str->cdata.len; while (src < end) { c = *src; @@ -1081,9 +1084,9 @@ dictionary_order(struct bwstring *str) wchar_t *dst, *end, *src; wchar_t c; - src = str->data.wstr; + src = str->wdata.str; dst = src; - end = src + str->len; + end = src + str->wdata.len; while (src < end) { c = *src; @@ -1110,10 +1113,10 @@ ignore_case(struct bwstring *str) { if (mb_cur_max == 1) { - unsigned char *end, *s; + char *end, *s; - s = str->data.cstr; - end = s + str->len; + s = str->cdata.str; + end = s + str->cdata.len; while (s < end) { *s = toupper(*s); @@ -1122,8 +1125,8 @@ ignore_case(struct bwstring *str) } else { wchar_t *end, *s; - s = str->data.wstr; - end = s + str->len; + s = str->wdata.str; + end = s + str->wdata.len; while (s < end) { *s = towupper(*s); @@ -1138,7 +1141,7 @@ bws_disorder_warnx(struct bwstring *s, const char *fn, size_t pos) { if (mb_cur_max == 1) - warnx("%s:%zu: disorder: %s", fn, pos + 1, s->data.cstr); + warnx("%s:%zu: disorder: %s", fn, pos + 1, s->cdata.str); else - warnx("%s:%zu: disorder: %ls", fn, pos + 1, s->data.wstr); + warnx("%s:%zu: disorder: %ls", fn, pos + 1, s->wdata.str); } diff --git a/usr.bin/sort/bwstring.h b/usr.bin/sort/bwstring.h index b63bb97ab93f..09a0dbf2fac2 100644 --- a/usr.bin/sort/bwstring.h +++ b/usr.bin/sort/bwstring.h @@ -46,17 +46,25 @@ extern bool byte_sort; /* wchar_t is of 4 bytes: */ #define SIZEOF_WCHAR_STRING(LEN) ((LEN)*sizeof(wchar_t)) +struct wstr { + size_t len; + wchar_t str[]; +}; + +struct cstr { + size_t len; + char str[]; +}; + /* * Binary "wide" string */ struct bwstring { - size_t len; - union - { - wchar_t wstr[0]; - unsigned char cstr[0]; - } data; + union { + struct wstr wdata; + struct cstr cdata; + }; }; struct reader_buffer @@ -67,8 +75,7 @@ struct reader_buffer typedef void *bwstring_iterator; -#define BWSLEN(s) ((s)->len) - +#define BWSLEN(s) ((mb_cur_max == 1) ? (s)->cdata.len : (s)->wdata.len) struct bwstring *bwsalloc(size_t sz); size_t bwsrawlen(const struct bwstring *bws); @@ -103,7 +110,7 @@ static inline bwstring_iterator bws_begin(struct bwstring *bws) { - return (bwstring_iterator) (&(bws->data)); + return ((bwstring_iterator)bws->wdata.str); } static inline bwstring_iterator @@ -111,8 +118,8 @@ bws_end(struct bwstring *bws) { return ((mb_cur_max == 1) ? - (bwstring_iterator) (bws->data.cstr + bws->len) : - (bwstring_iterator) (bws->data.wstr + bws->len)); + (bwstring_iterator) (bws->cdata.str + bws->cdata.len) : + (bwstring_iterator) (bws->wdata.str + bws->wdata.len)); } static inline bwstring_iterator @@ -138,7 +145,7 @@ bws_get_iter_value(bwstring_iterator iter) int bws_iterator_cmp(bwstring_iterator iter1, bwstring_iterator iter2, size_t len); -#define BWS_GET(bws, pos) ((mb_cur_max == 1) ? ((bws)->data.cstr[(pos)]) : (bws)->data.wstr[(pos)]) +#define BWS_GET(bws, pos) ((mb_cur_max == 1) ? (bws->cdata.str[(pos)]) : bws->wdata.str[(pos)]) void initialise_months(void); _______________________________________________ [email protected] mailing list https://lists.freebsd.org/mailman/listinfo/dev-commits-src-main To unsubscribe, send any mail to "[email protected]"
