From patchwork Thu May 24 12:36:44 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Andreas Schwab X-Patchwork-Id: 27488 Received: (qmail 58722 invoked by alias); 24 May 2018 12:36:51 -0000 Mailing-List: contact libc-alpha-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Unsubscribe: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: libc-alpha-owner@sourceware.org Delivered-To: mailing list libc-alpha@sourceware.org Received: (qmail 57129 invoked by uid 89); 24 May 2018 12:36:49 -0000 Authentication-Results: sourceware.org; auth=none X-Virus-Found: No X-Spam-SWARE-Status: No, score=-26.9 required=5.0 tests=BAYES_00, GIT_PATCH_0, GIT_PATCH_1, GIT_PATCH_2, GIT_PATCH_3, KAM_SHORT, SPF_PASS autolearn=ham version=3.3.2 spammy= X-HELO: mx2.suse.de From: Andreas Schwab To: libc-alpha@sourceware.org Subject: [PATCH] Fix fnmatch handling of collating elements (bug 17396, bug 16976) X-Yow: If I am elected, the concrete barriers around the WHITE HOUSE will be replaced by tasteful foam replicas of ANN MARGARET! Date: Thu, 24 May 2018 14:36:44 +0200 Message-ID: User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/26.1 (gnu/linux) MIME-Version: 1.0 This fixes the same bug that was fixed by commit 7e2f0d2 for regexp matching. As a side effect it also removes the use of an unbound VLA. [BZ #16976] [BZ #17396] * posix/fnmatch_loop.c (internal_fnmatch, internal_fnwmatch): When looking up collating elements match against (wide) character sequence instead of name. Correct alignment adjustment. * posix/fnmatch.c: Don't include "../locale/elem-hash.h". * posix/Makefile (tests): Add tst-fnmatch4 and tst-fnmatch5. (LOCALES): Add cs_CZ.ISO-8859-2. * posix/tst-fnmatch4.c: New file. * posix/tst-fnmatch5.c: New file. --- posix/Makefile | 4 +- posix/fnmatch.c | 1 - posix/fnmatch_loop.c | 227 +++++++++++++++++-------------------------- posix/tst-fnmatch4.c | 50 ++++++++++ posix/tst-fnmatch5.c | 52 ++++++++++ 5 files changed, 192 insertions(+), 142 deletions(-) create mode 100644 posix/tst-fnmatch4.c create mode 100644 posix/tst-fnmatch5.c diff --git a/posix/Makefile b/posix/Makefile index e9730eedf0..d87e245fc9 100644 --- a/posix/Makefile +++ b/posix/Makefile @@ -92,6 +92,7 @@ tests := test-errno tstgetopt testfnm runtests runptests \ bug-getopt5 tst-getopt_long1 bug-regex34 bug-regex35 \ tst-pathconf tst-rxspencer-no-utf8 \ tst-fnmatch3 bug-regex36 \ + tst-fnmatch4 tst-fnmatch5 \ tst-posix_spawn-fd tst-posix_spawn-setsid \ tst-posix_fadvise tst-posix_fadvise64 \ tst-sysconf-empty-chroot tst-glob_symlinks tst-fexecve \ @@ -165,7 +166,8 @@ $(objpfx)wordexp-tst.out: wordexp-tst.sh $(objpfx)wordexp-test endif LOCALES := cs_CZ.UTF-8 da_DK.ISO-8859-1 de_DE.ISO-8859-1 de_DE.UTF-8 \ - en_US.UTF-8 es_US.ISO-8859-1 es_US.UTF-8 ja_JP.EUC-JP tr_TR.UTF-8 + en_US.UTF-8 es_US.ISO-8859-1 es_US.UTF-8 ja_JP.EUC-JP tr_TR.UTF-8 \ + cs_CZ.ISO-8859-2 include ../gen-locales.mk $(objpfx)bug-regex1.out: $(gen-locales) diff --git a/posix/fnmatch.c b/posix/fnmatch.c index a9b762624f..2ef00f1e2f 100644 --- a/posix/fnmatch.c +++ b/posix/fnmatch.c @@ -53,7 +53,6 @@ we support a correct implementation only in glibc. */ #ifdef _LIBC # include "../locale/localeinfo.h" -# include "../locale/elem-hash.h" # include "../locale/coll-lookup.h" # include diff --git a/posix/fnmatch_loop.c b/posix/fnmatch_loop.c index e298cac5dc..30f3be1aba 100644 --- a/posix/fnmatch_loop.c +++ b/posix/fnmatch_loop.c @@ -494,26 +494,12 @@ FCT (const CHAR *pattern, const CHAR *string, const CHAR *string_end, { int32_t table_size; const int32_t *symb_table; -# if WIDE_CHAR_VERSION - char str[c1]; - unsigned int strcnt; -# else -# define str (startp + 1) -# endif const unsigned char *extra; int32_t idx; int32_t elem; - int32_t second; - int32_t hash; - # if WIDE_CHAR_VERSION - /* We have to convert the name to a single-byte - string. This is possible since the names - consist of ASCII characters and the internal - representation is UCS4. */ - for (strcnt = 0; strcnt < c1; ++strcnt) - str[strcnt] = startp[1 + strcnt]; -#endif + int32_t *wextra; +# endif table_size = _NL_CURRENT_WORD (LC_COLLATE, @@ -525,71 +511,55 @@ FCT (const CHAR *pattern, const CHAR *string, const CHAR *string_end, _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB); - /* Locate the character in the hashing table. */ - hash = elem_hash (str, c1); - - idx = 0; - elem = hash % table_size; - if (symb_table[2 * elem] != 0) - { - second = hash % (table_size - 2) + 1; - - do - { - /* First compare the hashing value. */ - if (symb_table[2 * elem] == hash - && (c1 - == extra[symb_table[2 * elem + 1]]) - && memcmp (str, - &extra[symb_table[2 * elem - + 1] - + 1], c1) == 0) - { - /* Yep, this is the entry. */ - idx = symb_table[2 * elem + 1]; - idx += 1 + extra[idx]; - break; - } - - /* Next entry. */ - elem += second; - } - while (symb_table[2 * elem] != 0); - } + for (elem = 0; elem < table_size; elem++) + if (symb_table[2 * elem] != 0) + { + idx = symb_table[2 * elem + 1]; + /* Skip the name of collating element. */ + idx += 1 + extra[idx]; +# if WIDE_CHAR_VERSION + /* Skip the byte sequence of the + collating element. */ + idx += 1 + extra[idx]; + /* Adjust for the alignment. */ + idx = (idx + 3) & ~3; + + wextra = (int32_t *) &extra[idx + 4]; + + if (/* Compare the length of the sequence. */ + c1 == wextra[0] + /* Compare the wide char sequence. */ + && memcmp (startp + 1, &wextra[1], + c1 * sizeof (UCHAR)) == 0) + /* Yep, this is the entry. */ + break; +# else + if (/* Compare the length of the sequence. */ + c1 == extra[idx] + /* Compare the byte sequence. */ + && memcmp (startp + 1, + &extra[idx + 1], c1) == 0) + /* Yep, this is the entry. */ + break; +# endif + } - if (symb_table[2 * elem] != 0) + if (elem < table_size) { /* Compare the byte sequence but only if this is not part of a range. */ -# if WIDE_CHAR_VERSION - int32_t *wextra; + if (! is_range - idx += 1 + extra[idx]; - /* Adjust for the alignment. */ - idx = (idx + 3) & ~3; - - wextra = (int32_t *) &extra[idx + 4]; -# endif - - if (! is_range) - { # if WIDE_CHAR_VERSION - for (c1 = 0; - (int32_t) c1 < wextra[idx]; - ++c1) - if (n[c1] != wextra[1 + c1]) - break; - - if ((int32_t) c1 == wextra[idx]) - goto matched; + && memcmp (n, &wextra[1], + c1 * sizeof (UCHAR)) == 0 # else - for (c1 = 0; c1 < extra[idx]; ++c1) - if (n[c1] != extra[1 + c1]) - break; - - if (c1 == extra[idx]) - goto matched; + && memcmp (n, &extra[idx + 1], c1) == 0 # endif + ) + { + n += c1 - 1; + goto matched; } /* Get the collation sequence value. */ @@ -597,9 +567,9 @@ FCT (const CHAR *pattern, const CHAR *string, const CHAR *string_end, # if WIDE_CHAR_VERSION cold = wextra[1 + wextra[idx]]; # else - /* Adjust for the alignment. */ idx += 1 + extra[idx]; - idx = (idx + 3) & ~4; + /* Adjust for the alignment. */ + idx = (idx + 3) & ~3; cold = *((int32_t *) &extra[idx]); # endif @@ -609,10 +579,10 @@ FCT (const CHAR *pattern, const CHAR *string, const CHAR *string_end, { /* No valid character. Match it as a single byte. */ - if (!is_range && *n == str[0]) + if (!is_range && *n == startp[1]) goto matched; - cold = str[0]; + cold = startp[1]; c = *p++; } else @@ -620,7 +590,6 @@ FCT (const CHAR *pattern, const CHAR *string, const CHAR *string_end, } } else -# undef str #endif { c = FOLD (c); @@ -712,25 +681,11 @@ FCT (const CHAR *pattern, const CHAR *string, const CHAR *string_end, { int32_t table_size; const int32_t *symb_table; -# if WIDE_CHAR_VERSION - char str[c1]; - unsigned int strcnt; -# else -# define str (startp + 1) -# endif const unsigned char *extra; int32_t idx; int32_t elem; - int32_t second; - int32_t hash; - # if WIDE_CHAR_VERSION - /* We have to convert the name to a single-byte - string. This is possible since the names - consist of ASCII characters and the internal - representation is UCS4. */ - for (strcnt = 0; strcnt < c1; ++strcnt) - str[strcnt] = startp[1 + strcnt]; + int32_t *wextra; # endif table_size = @@ -743,51 +698,44 @@ FCT (const CHAR *pattern, const CHAR *string, const CHAR *string_end, _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB); - /* Locate the character in the hashing - table. */ - hash = elem_hash (str, c1); - - idx = 0; - elem = hash % table_size; - if (symb_table[2 * elem] != 0) - { - second = hash % (table_size - 2) + 1; - - do - { - /* First compare the hashing value. */ - if (symb_table[2 * elem] == hash - && (c1 - == extra[symb_table[2 * elem + 1]]) - && memcmp (str, - &extra[symb_table[2 * elem + 1] - + 1], c1) == 0) - { - /* Yep, this is the entry. */ - idx = symb_table[2 * elem + 1]; - idx += 1 + extra[idx]; - break; - } - - /* Next entry. */ - elem += second; - } - while (symb_table[2 * elem] != 0); - } - - if (symb_table[2 * elem] != 0) - { - /* Compare the byte sequence but only if - this is not part of a range. */ + for (elem = 0; elem < table_size; elem++) + if (symb_table[2 * elem] != 0) + { + idx = symb_table[2 * elem + 1]; + /* Skip the name of collating + element. */ + idx += 1 + extra[idx]; # if WIDE_CHAR_VERSION - int32_t *wextra; - - idx += 1 + extra[idx]; - /* Adjust for the alignment. */ - idx = (idx + 3) & ~4; - - wextra = (int32_t *) &extra[idx + 4]; + /* Skip the byte sequence of the + collating element. */ + idx += 1 + extra[idx]; + /* Adjust for the alignment. */ + idx = (idx + 3) & ~3; + + wextra = (int32_t *) &extra[idx + 4]; + + if (/* Compare the length of the + sequence. */ + c1 == wextra[0] + /* Compare the wide char sequence. */ + && memcmp (startp + 1, &wextra[1], + c1 * sizeof (int32_t)) == 0) + /* Yep, this is the entry. */ + break; +# else + if (/* Compare the length of the + sequence. */ + c1 == extra[idx] + /* Compare the byte sequence. */ + && memcmp (startp + 1, + &extra[idx + 1], c1) == 0) + /* Yep, this is the entry. */ + break; # endif + } + + if (elem < table_size) + { /* Get the collation sequence value. */ is_seqval = 1; # if WIDE_CHAR_VERSION @@ -795,19 +743,18 @@ FCT (const CHAR *pattern, const CHAR *string, const CHAR *string_end, # else /* Adjust for the alignment. */ idx += 1 + extra[idx]; - idx = (idx + 3) & ~4; + idx = (idx + 3) & ~3; cend = *((int32_t *) &extra[idx]); # endif } - else if (symb_table[2 * elem] != 0 && c1 == 1) + else if (c1 == 1) { - cend = str[0]; + cend = startp[1]; c = *p++; } else return FNM_NOMATCH; } -# undef str } else { diff --git a/posix/tst-fnmatch4.c b/posix/tst-fnmatch4.c new file mode 100644 index 0000000000..9edeeb19db --- /dev/null +++ b/posix/tst-fnmatch4.c @@ -0,0 +1,50 @@ +/* Test for fnmatch handling of collating elements + Copyright (C) 2018 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + The GNU C Library 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 + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with the GNU C Library; if not, see + . */ + +#include +#include +#include + +static int +do_test_locale (const char *locale) +{ + const char *pattern = "[[.ch.]]"; + + if (setlocale (LC_ALL, locale) == NULL) + { + printf ("could not set locale %s\n", locale); + return 1; + } + + if (fnmatch (pattern, "ch", 0) != 0) + { + printf ("%s didn't match in locale %s\n", pattern, locale); + return 1; + } + + return 0; +} + +static int +do_test (void) +{ + return (do_test_locale ("cs_CZ.ISO-8859-2") + || do_test_locale ("cs_CZ.UTF-8")); +} + +#include diff --git a/posix/tst-fnmatch5.c b/posix/tst-fnmatch5.c new file mode 100644 index 0000000000..c01abc9db5 --- /dev/null +++ b/posix/tst-fnmatch5.c @@ -0,0 +1,52 @@ +/* Test for fnmatch handling of collating elements + Copyright (C) 2018 Free Software Foundation, Inc. + This file is part of the GNU C Library. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Lesser General Public + License as published by the Free Software Foundation; either + version 2.1 of the License, or (at your option) any later version. + + The GNU C Library 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 + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public + License along with the GNU C Library; if not, see + . */ + +#include +#include +#include +#include + +#define LENGTH 20000000 + +static char pattern[LENGTH + 7]; + +static int +do_test (void) +{ + if (setlocale (LC_ALL, "en_US.UTF-8") == NULL) + { + puts ("could not set locale"); + return 1; + } + pattern[0] = '['; + pattern[1] = '['; + pattern[2] = '.'; + memset (pattern + 3, 'a', LENGTH); + pattern[LENGTH + 3] = '.'; + pattern[LENGTH + 4] = ']'; + pattern[LENGTH + 5] = ']'; + int ret = fnmatch (pattern, "a", 0); + if (ret == 0) + { + puts ("fnmatch returned 0 for invalid pattern"); + return 1; + } + return 0; +} + +#include