* [PATCH 1/4] regex: avoid copying of uninitialized storage
@ 2019-10-10 1:34 Paul Eggert
2019-10-10 1:34 ` [PATCH 2/4] regex: simplify by assuming C99 Paul Eggert
` (2 more replies)
0 siblings, 3 replies; 5+ messages in thread
From: Paul Eggert @ 2019-10-10 1:34 UTC (permalink / raw)
To: bug-gnulib; +Cc: Paul Eggert
* config/srclist.txt: Comment out regcomp.c temporarily.
* lib/regcomp.c (build_charclass_op, create_tree) [! (GCC_LINT||lint)]:
Initialize even when not checking for lint, as the behavior is
arguably undefined otherwise and Coverity warns about it.
---
ChangeLog | 8 ++++++++
config/srclist.txt | 2 +-
lib/regcomp.c | 13 ++-----------
3 files changed, 11 insertions(+), 12 deletions(-)
diff --git a/ChangeLog b/ChangeLog
index 0fbb2d439..8251b8396 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,11 @@
+2019-10-09 Paul Eggert <eggert@cs.ucla.edu>
+
+ regex: avoid copying of uninitialized storage
+ * config/srclist.txt: Comment out regcomp.c temporarily.
+ * lib/regcomp.c (build_charclass_op, create_tree) [! (GCC_LINT||lint)]:
+ Initialize even when not checking for lint, as the behavior is
+ arguably undefined otherwise and Coverity warns about it.
+
2019-10-06 Bruno Haible <bruno@clisp.org>
access tests: Fix test failure when run as root.
diff --git a/config/srclist.txt b/config/srclist.txt
index 4a3a5a7af..bceaee863 100644
--- a/config/srclist.txt
+++ b/config/srclist.txt
@@ -52,7 +52,7 @@ $LIBCSRC malloc/scratch_buffer_grow_preserve.c lib/malloc
$LIBCSRC malloc/scratch_buffer_set_array_size.c lib/malloc
# Temporarily newer in Gnulib than in glibc.
#$LIBCSRC include/intprops.h lib
-$LIBCSRC posix/regcomp.c lib
+#$LIBCSRC posix/regcomp.c lib
$LIBCSRC posix/regex.c lib
$LIBCSRC posix/regex.h lib
$LIBCSRC posix/regex_internal.c lib
diff --git a/lib/regcomp.c b/lib/regcomp.c
index 7525355a9..c1f7f2b2a 100644
--- a/lib/regcomp.c
+++ b/lib/regcomp.c
@@ -3662,7 +3662,6 @@ build_charclass_op (re_dfa_t *dfa, RE_TRANSLATE_TYPE trans,
Idx alloc = 0;
#endif /* not RE_ENABLE_I18N */
reg_errcode_t ret;
- re_token_t br_token;
bin_tree_t *tree;
sbcset = (re_bitset_ptr_t) calloc (sizeof (bitset_t), 1);
@@ -3713,11 +3712,7 @@ build_charclass_op (re_dfa_t *dfa, RE_TRANSLATE_TYPE trans,
#endif
/* Build a tree for simple bracket. */
-#if defined GCC_LINT || defined lint
- memset (&br_token, 0, sizeof br_token);
-#endif
- br_token.type = SIMPLE_BRACKET;
- br_token.opr.sbcset = sbcset;
+ re_token_t br_token = { .type = SIMPLE_BRACKET, .opr.sbcset = sbcset };
tree = create_token_tree (dfa, NULL, NULL, &br_token);
if (__glibc_unlikely (tree == NULL))
goto build_word_op_espace;
@@ -3808,11 +3803,7 @@ static bin_tree_t *
create_tree (re_dfa_t *dfa, bin_tree_t *left, bin_tree_t *right,
re_token_type_t type)
{
- re_token_t t;
-#if defined GCC_LINT || defined lint
- memset (&t, 0, sizeof t);
-#endif
- t.type = type;
+ re_token_t t = { .type = type };
return create_token_tree (dfa, left, right, &t);
}
--
2.21.0
^ permalink raw reply related [flat|nested] 5+ messages in thread* [PATCH 2/4] regex: simplify by assuming C99
2019-10-10 1:34 [PATCH 1/4] regex: avoid copying of uninitialized storage Paul Eggert
@ 2019-10-10 1:34 ` Paul Eggert
2019-10-12 10:41 ` Bruno Haible
2019-10-10 1:34 ` [PATCH 3/4] regex: tell compiler there’s at most 256 arcs out Paul Eggert
2019-10-10 1:34 ` [PATCH 4/4] regex: omit debug assignment when not debugging Paul Eggert
2 siblings, 1 reply; 5+ messages in thread
From: Paul Eggert @ 2019-10-10 1:34 UTC (permalink / raw)
To: bug-gnulib; +Cc: Paul Eggert
* config/srclist.txt: Comment out regex_internal.h and regexec.c
temporarily.
* lib/regex_internal.h (lock_define, re_match_context_t):
Simplify by assuming C99 macros and const.
* lib/regexec.c (re_search_internal): Simplify by assuming C99
initializers. Remove unnecessary assignment, as mctx is now
safely initialized earlier.
---
ChangeLog | 9 +++++++++
config/srclist.txt | 4 ++--
lib/regex_internal.h | 21 +--------------------
lib/regexec.c | 11 -----------
4 files changed, 12 insertions(+), 33 deletions(-)
diff --git a/ChangeLog b/ChangeLog
index 8251b8396..4f96c8cbf 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,14 @@
2019-10-09 Paul Eggert <eggert@cs.ucla.edu>
+ regex: simplify by assuming C99
+ * config/srclist.txt: Comment out regex_internal.h and regexec.c
+ temporarily.
+ * lib/regex_internal.h (lock_define, re_match_context_t):
+ Simplify by assuming C99 macros and const.
+ * lib/regexec.c (re_search_internal): Simplify by assuming C99
+ initializers. Remove unnecessary assignment, as mctx is now
+ safely initialized earlier.
+
regex: avoid copying of uninitialized storage
* config/srclist.txt: Comment out regcomp.c temporarily.
* lib/regcomp.c (build_charclass_op, create_tree) [! (GCC_LINT||lint)]:
diff --git a/config/srclist.txt b/config/srclist.txt
index bceaee863..c53fb90a8 100644
--- a/config/srclist.txt
+++ b/config/srclist.txt
@@ -56,8 +56,8 @@ $LIBCSRC malloc/scratch_buffer_set_array_size.c lib/malloc
$LIBCSRC posix/regex.c lib
$LIBCSRC posix/regex.h lib
$LIBCSRC posix/regex_internal.c lib
-$LIBCSRC posix/regex_internal.h lib
-$LIBCSRC posix/regexec.c lib
+#$LIBCSRC posix/regex_internal.h lib
+#$LIBCSRC posix/regexec.c lib
$LIBCSRC time/timegm.c lib
$LIBCSRC time/mktime.c lib
$LIBCSRC time/mktime-internal.h lib
diff --git a/lib/regex_internal.h b/lib/regex_internal.h
index 5462419b7..9c0e1f1d1 100644
--- a/lib/regex_internal.h
+++ b/lib/regex_internal.h
@@ -44,22 +44,7 @@
# define lock_unlock(lock) __libc_lock_unlock (lock)
#elif defined GNULIB_LOCK && !defined USE_UNLOCKED_IO
# include "glthread/lock.h"
- /* Use gl_lock_define if empty macro arguments are known to work.
- Otherwise, fall back on less-portable substitutes. */
-# if ((defined __GNUC__ && !defined __STRICT_ANSI__) \
- || (defined __STDC_VERSION__ && 199901L <= __STDC_VERSION__))
-# define lock_define(name) gl_lock_define (, name)
-# elif USE_POSIX_THREADS
-# define lock_define(name) pthread_mutex_t name;
-# elif USE_PTH_THREADS
-# define lock_define(name) pth_mutex_t name;
-# elif USE_SOLARIS_THREADS
-# define lock_define(name) mutex_t name;
-# elif USE_WINDOWS_THREADS
-# define lock_define(name) gl_lock_t name;
-# else
-# define lock_define(name)
-# endif
+# define lock_define(name) gl_lock_define (, name)
# define lock_init(lock) glthread_lock_init (&(lock))
# define lock_fini(lock) glthread_lock_destroy (&(lock))
# define lock_lock(lock) glthread_lock_lock (&(lock))
@@ -618,11 +603,7 @@ typedef struct
{
/* The string object corresponding to the input string. */
re_string_t input;
-#if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
const re_dfa_t *const dfa;
-#else
- const re_dfa_t *dfa;
-#endif
/* EFLAGS of the argument of regexec. */
int eflags;
/* Where the matching ends. */
diff --git a/lib/regexec.c b/lib/regexec.c
index 4ff30a79c..f7d2b5b20 100644
--- a/lib/regexec.c
+++ b/lib/regexec.c
@@ -597,21 +597,12 @@ re_search_internal (const regex_t *preg, const char *string, Idx length,
Idx extra_nmatch;
bool sb;
int ch;
-#if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)
re_match_context_t mctx = { .dfa = dfa };
-#else
- re_match_context_t mctx;
-#endif
char *fastmap = ((preg->fastmap != NULL && preg->fastmap_accurate
&& start != last_start && !preg->can_be_null)
? preg->fastmap : NULL);
RE_TRANSLATE_TYPE t = preg->translate;
-#if !(defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L))
- memset (&mctx, '\0', sizeof (re_match_context_t));
- mctx.dfa = dfa;
-#endif
-
extra_nmatch = (nmatch > preg->re_nsub) ? nmatch - (preg->re_nsub + 1) : 0;
nmatch -= extra_nmatch;
@@ -677,8 +668,6 @@ re_search_internal (const regex_t *preg, const char *string, Idx length,
goto free_return;
}
}
- else
- mctx.state_log = NULL;
match_first = start;
mctx.input.tip_context = (eflags & REG_NOTBOL) ? CONTEXT_BEGBUF
--
2.21.0
^ permalink raw reply related [flat|nested] 5+ messages in thread* [PATCH 3/4] regex: tell compiler there’s at most 256 arcs out
2019-10-10 1:34 [PATCH 1/4] regex: avoid copying of uninitialized storage Paul Eggert
2019-10-10 1:34 ` [PATCH 2/4] regex: simplify by assuming C99 Paul Eggert
@ 2019-10-10 1:34 ` Paul Eggert
2019-10-10 1:34 ` [PATCH 4/4] regex: omit debug assignment when not debugging Paul Eggert
2 siblings, 0 replies; 5+ messages in thread
From: Paul Eggert @ 2019-10-10 1:34 UTC (permalink / raw)
To: bug-gnulib; +Cc: Paul Eggert
Partly this is to help the reader (and maybe help GCC);
partly this is to pacify Coverity.
* lib/regex_internal.h: Include verify.h.
* lib/regexec.c (group_nodes_into_DFAstates):
Tell the compiler that ndests cannot exceed SBC_MAX.
* modules/regex (Depends-on): Add ‘verify’.
---
ChangeLog | 8 ++++++++
lib/regex_internal.h | 1 +
lib/regexec.c | 1 +
modules/regex | 1 +
4 files changed, 11 insertions(+)
diff --git a/ChangeLog b/ChangeLog
index 4f96c8cbf..360973ba4 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,13 @@
2019-10-09 Paul Eggert <eggert@cs.ucla.edu>
+ regex: tell compiler there’s at most 256 arcs out
+ Partly this is to help the reader (and maybe help GCC);
+ partly this is to pacify Coverity.
+ * lib/regex_internal.h: Include verify.h.
+ * lib/regexec.c (group_nodes_into_DFAstates):
+ Tell the compiler that ndests cannot exceed SBC_MAX.
+ * modules/regex (Depends-on): Add ‘verify’.
+
regex: simplify by assuming C99
* config/srclist.txt: Comment out regex_internal.h and regexec.c
temporarily.
diff --git a/lib/regex_internal.h b/lib/regex_internal.h
index 9c0e1f1d1..b6eeba32d 100644
--- a/lib/regex_internal.h
+++ b/lib/regex_internal.h
@@ -34,6 +34,7 @@
#include <stdint.h>
#include <intprops.h>
+#include <verify.h>
#ifdef _LIBC
# include <libc-lock.h>
diff --git a/lib/regexec.c b/lib/regexec.c
index f7d2b5b20..0d32e0139 100644
--- a/lib/regexec.c
+++ b/lib/regexec.c
@@ -3684,6 +3684,7 @@ group_nodes_into_DFAstates (const re_dfa_t *dfa, const re_dfastate_t *state,
bitset_empty (accepts);
}
}
+ assume (ndests <= SBC_MAX);
return ndests;
error_return:
for (j = 0; j < ndests; ++j)
diff --git a/modules/regex b/modules/regex
index 7f50916ce..dbfd37982 100644
--- a/modules/regex
+++ b/modules/regex
@@ -30,6 +30,7 @@ mbsinit [test $ac_use_included_regex = yes]
nl_langinfo [test $ac_use_included_regex = yes]
stdbool [test $ac_use_included_regex = yes]
stdint [test $ac_use_included_regex = yes]
+verify [test $ac_use_included_regex = yes]
wchar [test $ac_use_included_regex = yes]
wcrtomb [test $ac_use_included_regex = yes]
wctype-h [test $ac_use_included_regex = yes]
--
2.21.0
^ permalink raw reply related [flat|nested] 5+ messages in thread* [PATCH 4/4] regex: omit debug assignment when not debugging
2019-10-10 1:34 [PATCH 1/4] regex: avoid copying of uninitialized storage Paul Eggert
2019-10-10 1:34 ` [PATCH 2/4] regex: simplify by assuming C99 Paul Eggert
2019-10-10 1:34 ` [PATCH 3/4] regex: tell compiler there’s at most 256 arcs out Paul Eggert
@ 2019-10-10 1:34 ` Paul Eggert
2 siblings, 0 replies; 5+ messages in thread
From: Paul Eggert @ 2019-10-10 1:34 UTC (permalink / raw)
To: bug-gnulib; +Cc: Paul Eggert
* lib/regexec.c (re_search_internal) [!DEBUG]:
Remove unnecessary assignment to pacify Coverity.
---
ChangeLog | 4 ++++
lib/regexec.c | 2 ++
2 files changed, 6 insertions(+)
diff --git a/ChangeLog b/ChangeLog
index 360973ba4..e9e337666 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,9 @@
2019-10-09 Paul Eggert <eggert@cs.ucla.edu>
+ regex: omit debug assignment when not debugging
+ * lib/regexec.c (re_search_internal) [!DEBUG]:
+ Remove unnecessary assignment.
+
regex: tell compiler there’s at most 256 arcs out
Partly this is to help the reader (and maybe help GCC);
partly this is to pacify Coverity.
diff --git a/lib/regexec.c b/lib/regexec.c
index 0d32e0139..809f89e26 100644
--- a/lib/regexec.c
+++ b/lib/regexec.c
@@ -817,7 +817,9 @@ re_search_internal (const regex_t *preg, const char *string, Idx length,
break;
if (__glibc_unlikely (err != REG_NOMATCH))
goto free_return;
+#ifdef DEBUG
match_last = -1;
+#endif
}
else
break; /* We found a match. */
--
2.21.0
^ permalink raw reply related [flat|nested] 5+ messages in thread
end of thread, other threads:[~2019-10-12 10:41 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-10-10 1:34 [PATCH 1/4] regex: avoid copying of uninitialized storage Paul Eggert
2019-10-10 1:34 ` [PATCH 2/4] regex: simplify by assuming C99 Paul Eggert
2019-10-12 10:41 ` Bruno Haible
2019-10-10 1:34 ` [PATCH 3/4] regex: tell compiler there’s at most 256 arcs out Paul Eggert
2019-10-10 1:34 ` [PATCH 4/4] regex: omit debug assignment when not debugging Paul Eggert
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).