gmane.comp.lib.gnulib.bugs

Re: regexp_exec thread-unsafe

Ludovic Courtès — 2013-05-20T14:20:08

Paul Eggert <eggert< at >cs.ucla.edu> skribis:


Thank you for the quick reply.  I haven’t tested yet, but I have a question:


Seems like this is going to add -lpthread to LDFLAGS, right?

That would raise two issues: first, it should add -pthread to both
CFLAGS and LDFLAGS, not just -lpthread.

Second, Guile can be compiled with and without thread support.  In the
latter case, we wouldn’t want Gnulib to stealthily pull in pthreads.
How can this be addressed?

Thanks!

Ludo’.

Re: regexp_exec thread-unsafe

Paul Eggert — 2013-05-19T21:30:55


Yes, thanks, that sounds right.  I pushed the following patch into gnulib;
could you please see whether it fixes the problem for you?  It'd be
helpful to try it on non-glibc systems too, if possible.

---
 ChangeLog            | 19 +++++++++++++++++++
 lib/regcomp.c        | 11 ++++++++---
 lib/regex_internal.h | 23 +++++++++++++++++------
 lib/regexec.c        |  4 ----
 modules/regex        |  1 +
 5 files changed, 45 insertions(+), 13 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index 08042c5..35b6dd2 100644
--- a/ChangeLog
+++ b/ChangeLog
< at >< at > -1,5 +1,24 < at >< at >
 2013-05-19  Paul Eggert  <eggert< at >cs.ucla.edu>
 
+regex: fix dfa race in multithreaded uses
+Problem reported by Ludovic Courtès in
+<http://lists.gnu.org/archive/html/bug-gnulib/2013-05/msg00058.html>.
+* lib/regex_internal.h (lock_define, lock_init, lock_fini):
+New macros.  All uses of __libc_lock_define, __libc_lock_init
+changed to use the first two of these.
+(__libc_lock_lock, __libc_lock_unlock): New macros, for
+non-glibc platfor

[PATCH] malloca: port to compilers that reject size-zero arrays

Paul Eggert — 2013-05-19T20:06:39

---
 ChangeLog     | 11 +++++++++++
 lib/malloca.c | 20 +++++++++++++-------
 2 files changed, 24 insertions(+), 7 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index 6963994..08042c5 100644
--- a/ChangeLog
+++ b/ChangeLog
< at >< at > -1,3 +1,14 < at >< at >
+2013-05-19  Paul Eggert  <eggert< at >cs.ucla.edu>
+
+malloca: port to compilers that reject size-zero arrays
+This fixes a bug introduced in my previous patch.
+* lib/malloca.c (struct preliminary_header): Use an int
+rather than a character array of size int; that's simpler.
+(struct header): Remove, replacing with ...
+(union header): New type.  This avoids the need for declaring a
+character array of size zero, which is not allowed on some platforms.
+All uses changed.
+
 2013-05-18  Paul Eggert  <eggert< at >cs.ucla.edu>
 
 parse-datetime, tests: don't use "string" + int
diff --git a/lib/malloca.c b/lib/malloca.c
index 53ecb81..7a4b0cd 100644
--- a/lib/malloca.c
+++ b/lib/malloca.c
< at >< at > -49,12 +49,18 < at >< at >
 #define MAGIC_SIZE sizeof (int)
 /* This is how the header info

[PATCH] parse-datetime, tests: don't use "string" + int

Paul Eggert — 2013-05-19T00:22:26

---
 ChangeLog                            | 18 ++++++++++++++++++
 lib/parse-datetime.y                 |  2 +-
 tests/test-fchdir.c                  |  2 +-
 tests/test-snprintf-posix.h          |  2 +-
 tests/test-snprintf.c                |  2 +-
 tests/test-vasnprintf-posix.c        |  2 +-
 tests/test-vasnprintf.c              |  2 +-
 tests/test-vsnprintf.c               |  2 +-
 tests/unistdio/test-ulc-asnprintf1.h |  2 +-
 9 files changed, 26 insertions(+), 8 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index de7f539..6963994 100644
--- a/ChangeLog
+++ b/ChangeLog
< at >< at > -1,3 +1,21 < at >< at >
+2013-05-18  Paul Eggert  <eggert< at >cs.ucla.edu>
+
+parse-datetime, tests: don't use "string" + int
+Recent versions of 'clang' complain about C source code that
+uses expressions of the form '"string literal" + integer',
+I guess on the theory that it's confusing for readers who are
+used to C++.  On those grounds I suppose it's OK to make this
+minor style change.
+* lib/parse-datetime.y (parse_datetime):
+* tes

Re: [PATCH 2/2] argmatch: port to C++

Eric Blake — 2013-05-17T17:14:37

Thanks; I've pushed both of these, after adding a ChangeLog entry.

[PATCH 2/2] argmatch: port to C++

Alexandre Duret-Lutz — 2013-05-17T17:01:15

* lib/argmatch.h [__cplusplus]: Add extern "C".
---
 lib/argmatch.h |    8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/lib/argmatch.h b/lib/argmatch.h
index 2e0275c..e4c8027 100644
--- a/lib/argmatch.h
+++ b/lib/argmatch.h
< at >< at > -26,6 +26,10 < at >< at >
 
 # include "verify.h"
 
+#ifdef  __cplusplus
+extern "C" {
+#endif
+
 # define ARRAY_CARDINALITY(Array) (sizeof (Array) / sizeof *(Array))
 
 /* Assert there are as many real arguments as there are values
< at >< at > -100,4 +104,8 < at >< at > char const *argmatch_to_argument (char const *value,
   argmatch_to_argument (Value, Arglist,                                 \
                         (char const *) (Vallist), sizeof *(Vallist))
 
+#ifdef  __cplusplus
+}
+#endif
+
 #endif /* ARGMATCH_H_ */

[PATCH 1/2] * lib/argp-help.c: Typo in comment.

Alexandre Duret-Lutz — 2013-05-17T17:01:14

---
 lib/argp-help.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/argp-help.c b/lib/argp-help.c
index 354f1e2..85def44 100644
--- a/lib/argp-help.c
+++ b/lib/argp-help.c
< at >< at > -649,7 +649,7 < at >< at > hol_find_entry (struct hol *hol, const char *name)
   return 0;
 }
 
-/* If an entry with the long option NAME occurs in HOL, set it's special
+/* If an entry with the long option NAME occurs in HOL, set its special
    sort position to GROUP.  */
 static void
 hol_set_group (struct hol *hol, const char *name, int group)

Re: [libvirt] [PATCH] Define __USE_MINGW_ANSI_STDIO in config.h

Eric Blake — 2013-05-17T14:35:48

[adding gnulib]

On 05/17/2013 08:14 AM, Daniel P. Berrange wrote:

Yes, gnulib should be taking advantage of this new mingw development.

Please don't push this to libvirt until gnulib has decided what to do
about it.

[PATCH] manywarnings: update for GCC 4.8.0

Paul Eggert — 2013-05-16T06:22:31

 ChangeLog          | 8 ++++++++
 m4/manywarnings.m4 | 6 ++----
 2 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index 171fa5d..c03bf8c 100644
--- a/ChangeLog
+++ b/ChangeLog
< at >< at > -1,5 +1,13 < at >< at >
 2013-05-15  Paul Eggert  <eggert< at >cs.ucla.edu>
 
+manywarnings: update for GCC 4.8.0
+* m4/manywarnings.m4 (gl_MANYWARN_ALL_GCC):
+Add -Waggressive-loop-optimizations, -Wreturn-local-addr, which
+are new to GCC 4.8.  Remove -Wformat=2, -Wmissing-format-attribute,
+-Wmissing-noreturn, as they are duplicates of other warnings.
+Remove -Wunreachable-code, as it is removed in GCC 4.8 and
+was documented to be flaky in earlier versions of GCC.
+
 spawn-tests, sys_socket-tests, sys_wait-tests: port to clang
 * tests/test-spawn.c (main):
 * tests/test-sys_socket.c (main):
diff --git a/m4/manywarnings.m4 b/m4/manywarnings.m4
index 45a30af..be6d4c9 100644
--- a/m4/manywarnings.m4
+++ b/m4/manywarnings.m4
< at >< at > -98,6 +98,7 < at >< at > AC_DEFUN([gl_MANYWARN_ALL_GCC],
     -W \
     -Wabi \

Re: [PATCH 1/4] stdio: use __REDIRECT for fwrite, fwrite_unlocked

Eric Blake — 2013-05-16T04:18:51

My first worry is that __REDIRECT is non-standard; but then I realized:


We only get here on glibc, where system headers give us __REDIRECT, so


this usage only has to work for glibc, which it does.  Interesting solution.

[PATCH 4/4] spawn-tests, sys_socket-tests, sys_wait-tests: port toclang

Paul Eggert — 2013-05-16T03:54:17

 ChangeLog               | 6 ++++++
 tests/test-spawn.c      | 2 +-
 tests/test-sys_socket.c | 2 +-
 tests/test-sys_wait.c   | 6 +++---
 4 files changed, 11 insertions(+), 5 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index f9212fe..171fa5d 100644
--- a/ChangeLog
+++ b/ChangeLog
< at >< at > -1,5 +1,11 < at >< at >
 2013-05-15  Paul Eggert  <eggert< at >cs.ucla.edu>
 
+spawn-tests, sys_socket-tests, sys_wait-tests: port to clang
+* tests/test-spawn.c (main):
+* tests/test-sys_socket.c (main):
+* tests/test-sys_wait.c (main):
+Don't have a switch value that isn't covered by a case.
+
 getaddrinfo-tests: port --enable-gcc-warnings to clang
 * tests/test-getaddrinfo.c (simple):
 Avoid casts from looser to stricter-aligned pointers.
diff --git a/tests/test-spawn.c b/tests/test-spawn.c
index cfa9dc6..5b68171 100644
--- a/tests/test-spawn.c
+++ b/tests/test-spawn.c
< at >< at > -40,7 +40,7 < at >< at > extern void f (struct sched_param *g);
 int
 main (void)
 {
-  switch (0)
+  switch (POSIX_SPAWN_RESETIDS)
     {
     case POSIX_SPAWN_RESETIDS:

[PATCH 3/4] getaddrinfo-tests: port --enable-gcc-warnings to clang

Paul Eggert — 2013-05-16T03:53:39

 ChangeLog                | 4 ++++
 tests/test-getaddrinfo.c | 5 +++--
 2 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index fbd47fc..f9212fe 100644
--- a/ChangeLog
+++ b/ChangeLog
< at >< at > -1,5 +1,9 < at >< at >
 2013-05-15  Paul Eggert  <eggert< at >cs.ucla.edu>
 
+getaddrinfo-tests: port --enable-gcc-warnings to clang
+* tests/test-getaddrinfo.c (simple):
+Avoid casts from looser to stricter-aligned pointers.
+
 thread: port --enable-gcc-warnings to clang
 * lib/glthread/thread.h [__clang__ && USE_POSIX_THREADS_WEAK]:
 Include <signal.h>, to pacify a warning about pthread_sigmask.
diff --git a/tests/test-getaddrinfo.c b/tests/test-getaddrinfo.c
index 1b9892f..5c27d3b 100644
--- a/tests/test-getaddrinfo.c
+++ b/tests/test-getaddrinfo.c
< at >< at > -114,6 +114,8 < at >< at > simple (char const *host, char const *service)
 
   for (ai = ai0; ai; ai = ai->ai_next)
     {
+      void *ai_addr = ai->ai_addr;
+      struct sockaddr_in *sock_addr = ai_addr;
       dbgprintf ("\tflags %x\n", ai->ai_flags);

[PATCH 2/4] thread: port --enable-gcc-warnings to clang

Paul Eggert — 2013-05-16T03:53:01

 ChangeLog             | 4 ++++
 lib/glthread/thread.h | 6 ++++++
 2 files changed, 10 insertions(+)

diff --git a/ChangeLog b/ChangeLog
index 3544972..fbd47fc 100644
--- a/ChangeLog
+++ b/ChangeLog
< at >< at > -1,5 +1,9 < at >< at >
 2013-05-15  Paul Eggert  <eggert< at >cs.ucla.edu>
 
+thread: port --enable-gcc-warnings to clang
+* lib/glthread/thread.h [__clang__ && USE_POSIX_THREADS_WEAK]:
+Include <signal.h>, to pacify a warning about pthread_sigmask.
+
 stdio: use __REDIRECT for fwrite, fwrite_unlocked
 * lib/stdio.in.h (fwrite):
 When working around bug 11959, use __REDIRECT rather than '#define
diff --git a/lib/glthread/thread.h b/lib/glthread/thread.h
index 166a50c..e52b615 100644
--- a/lib/glthread/thread.h
+++ b/lib/glthread/thread.h
< at >< at > -120,7 +120,13 < at >< at > extern int glthread_in_use (void);
    address of a function in libpthread that we don't use.  */
 
 #  pragma weak pthread_create
+
+#  ifdef __clang__
+  /* Without this, clang complains that pthread_sigmask is never declared.  */
+#   include <signal.h>
+#  endi

[PATCH 1/4] stdio: use __REDIRECT for fwrite, fwrite_unlocked

Paul Eggert — 2013-05-16T03:52:17

---
 ChangeLog      | 10 ++++++++++
 lib/stdio.in.h | 19 +++++++++++++------
 2 files changed, 23 insertions(+), 6 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index dfb2e5c..3544972 100644
--- a/ChangeLog
+++ b/ChangeLog
< at >< at > -1,5 +1,15 < at >< at >
 2013-05-15  Paul Eggert  <eggert< at >cs.ucla.edu>
 
+stdio: use __REDIRECT for fwrite, fwrite_unlocked
+* lib/stdio.in.h (fwrite):
+When working around bug 11959, use __REDIRECT rather than '#define
+fwrite(...) ... fwrite (...) ...'.  This is a more-targeted way to
+fix the -Wunused-value issue with clang, and it works with GCC too.
+Problem with targeting reported by Eric Blake in
+<http://lists.gnu.org/archive/html/bug-gnulib/2013-05/msg00067.html>.
+(fwrite_unlocked): Treat like fwrite.  I ran into this issue while
+debugging the fwrite issue.
+
 stdio: port --enable-gcc-warnings to clang
 * lib/stdio.in.h (fwrite) [__clang__]: Ignore -Wunused-value entirely,
 since the GCC workaround for fwrite does not pacify clang.
diff --git a/lib/stdio.in.h b/lib/stdio

Re: [PATCH 8/8] stdio: port --enable-gcc-warnings to clang

Paul Eggert — 2013-05-16T03:50:44


Yes.  Unfortunately they don't work here, even with _Pragma.


Unfortunately no.


I think that's because in other places we simply do a (void) fwrite (...),
which pacifies 'clang'.  But here we can't do that, since the replacement
fwrite must return a value.  None of the obvious workarounds worked for me.

But I did think of a workaround that wasn't obvious (at least, not to me),
and installed it.  I'll follow up with a message
"[PATCH 1/4] stdio: use __REDIRECT for fwrite, fwrite_unlocked"
about that.  Three other clang-related patches will follow.

Re: [PATCH 8/8] stdio: port --enable-gcc-warnings to clang

Eric Blake — 2013-05-15T11:57:44


Does clang support push/pop of pragmas?  Can it be silenced by adding a
cast to void instead of globally turning off this normally-useful warning?


Here, we were just inlining the gcc version if "ignore-value.h" (so that
our replacement <stdio.h> doesn't have to drag in the actual
ignore-value.h header); but the non-gcc counterpart is a cast to void,
and clang doesn't seem to be issuing unused result warnings for other
uses of ignore_value().

[PATCH 8/8] stdio: port --enable-gcc-warnings to clang

Paul Eggert — 2013-05-15T07:39:18

* lib/stdio.in.h (fwrite) [__clang__]: Ignore -Wunused-value entirely,
since the GCC workaround for fwrite does not pacify clang.
---
 ChangeLog      |  4 ++++
 lib/stdio.in.h | 11 +++++++----
 2 files changed, 11 insertions(+), 4 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index baa594e..dfb2e5c 100644
--- a/ChangeLog
+++ b/ChangeLog
< at >< at > -1,5 +1,9 < at >< at >
 2013-05-15  Paul Eggert  <eggert< at >cs.ucla.edu>
 
+stdio: port --enable-gcc-warnings to clang
+* lib/stdio.in.h (fwrite) [__clang__]: Ignore -Wunused-value entirely,
+since the GCC workaround for fwrite does not pacify clang.
+
 sig2str: port --enable-gcc-warnings to clang
 * lib/sig2str.c (sig2str): Avoid warning about unused printf argument.
 
diff --git a/lib/stdio.in.h b/lib/stdio.in.h
index c345499..115c976 100644
--- a/lib/stdio.in.h
+++ b/lib/stdio.in.h
< at >< at > -579,13 +579,16 < at >< at > _GL_CXXALIAS_SYS (fwrite, size_t,
    <http://sources.redhat.com/bugzilla/show_bug.cgi?id=11959>,
    which sometimes causes an unwanted diagnostic for fwrite calls.
    This

[PATCH 7/8] sig2str: port --enable-gcc-warnings to clang

Paul Eggert — 2013-05-15T07:38:31

* lib/sig2str.c (sig2str): Avoid warning about unused printf argument.
---
 ChangeLog     |  3 +++
 lib/sig2str.c | 12 ++++++++----
 2 files changed, 11 insertions(+), 4 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index 35a57b4..baa594e 100644
--- a/ChangeLog
+++ b/ChangeLog
< at >< at > -1,5 +1,8 < at >< at >
 2013-05-15  Paul Eggert  <eggert< at >cs.ucla.edu>
 
+sig2str: port --enable-gcc-warnings to clang
+* lib/sig2str.c (sig2str): Avoid warning about unused printf argument.
+
 obstack: port --enable-gcc-warnings to clang
 * lib/obstack.h (obstack_ptr_grow_fast, obstack_int_grow_fast):
 Avoid casts from looser to stricter-aligned pointers.
diff --git a/lib/sig2str.c b/lib/sig2str.c
index 8b36e2f..6ead2a7 100644
--- a/lib/sig2str.c
+++ b/lib/sig2str.c
< at >< at > -325,21 +325,25 < at >< at > sig2str (int signum, char *signame)
   {
     int rtmin = SIGRTMIN;
     int rtmax = SIGRTMAX;
+    int base, delta;
 
     if (! (rtmin <= signum && signum <= rtmax))
       return -1;
 
     if (signum <= rtmin + (rtmax - rtmin) / 2)
       {
-

[PATCH 6/8] obstack: port --enable-gcc-warnings to clang

Paul Eggert — 2013-05-15T07:37:59

* lib/obstack.h (obstack_ptr_grow_fast, obstack_int_grow_fast):
Avoid casts from looser to stricter-aligned pointers.
---
 ChangeLog     | 4 ++++
 lib/obstack.h | 6 ++++--
 2 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index f637cda..35a57b4 100644
--- a/ChangeLog
+++ b/ChangeLog
< at >< at > -1,5 +1,9 < at >< at >
 2013-05-15  Paul Eggert  <eggert< at >cs.ucla.edu>
 
+obstack: port --enable-gcc-warnings to clang
+* lib/obstack.h (obstack_ptr_grow_fast, obstack_int_grow_fast):
+Avoid casts from looser to stricter-aligned pointers.
+
 memchr2: port --enable-gcc-warnings to clang
 * lib/memchr2.c (memchr2):
 Avoid casts from looser to stricter-aligned pointers.
diff --git a/lib/obstack.h b/lib/obstack.h
index 159cfa2..7cf98ed 100644
--- a/lib/obstack.h
+++ b/lib/obstack.h
< at >< at > -342,14 +342,16 < at >< at > __extension__                                                           \
 # define obstack_ptr_grow_fast(OBSTACK,aptr)                            \
 __extension__

[PATCH 5/8] memchr2: port --enable-gcc-warnings to clang

Paul Eggert — 2013-05-15T07:37:29

* lib/memchr2.c (memchr2):
Avoid casts from looser to stricter-aligned pointers.
---
 ChangeLog     |  4 ++++
 lib/memchr2.c | 21 +++++++++++++--------
 2 files changed, 17 insertions(+), 8 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index 1a8925a..f637cda 100644
--- a/ChangeLog
+++ b/ChangeLog
< at >< at > -1,5 +1,9 < at >< at >
 2013-05-15  Paul Eggert  <eggert< at >cs.ucla.edu>
 
+memchr2: port --enable-gcc-warnings to clang
+* lib/memchr2.c (memchr2):
+Avoid casts from looser to stricter-aligned pointers.
+
 mbsstr: port --enable-gcc-warnings to clang
 * lib/mbsstr.c (knuth_morris_pratt_multibyte):
 Avoid casts from looser to stricter-aligned pointers.
diff --git a/lib/memchr2.c b/lib/memchr2.c
index 3912e02..8b105b7 100644
--- a/lib/memchr2.c
+++ b/lib/memchr2.c
< at >< at > -43,6 +43,7 < at >< at > memchr2 (void const *s, int c1_in, int c2_in, size_t n)
   typedef unsigned long int longword;
 
   const unsigned char *char_ptr;
+  void const *void_ptr;
   const longword *longword_ptr;
   longword repeated_one;
   longword repeated_c1;
< at >

[PATCH 4/8] mbsstr: port --enable-gcc-warnings to clang

Paul Eggert — 2013-05-15T07:36:58

* lib/mbsstr.c (knuth_morris_pratt_multibyte):
Avoid casts from looser to stricter-aligned pointers.
---
 ChangeLog    | 4 ++++
 lib/mbsstr.c | 7 ++++---
 2 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index 508458e..1a8925a 100644
--- a/ChangeLog
+++ b/ChangeLog
< at >< at > -1,5 +1,9 < at >< at >
 2013-05-15  Paul Eggert  <eggert< at >cs.ucla.edu>
 
+mbsstr: port --enable-gcc-warnings to clang
+* lib/mbsstr.c (knuth_morris_pratt_multibyte):
+Avoid casts from looser to stricter-aligned pointers.
+
 malloca: port --enable-gcc-warnings to clang
 * lib/malloca.c (struct header): New member 'magic', to avoid casts.
 (mmalloca): Avoid casts from looser to stricter-aligned pointers.
diff --git a/lib/mbsstr.c b/lib/mbsstr.c
index f84e689..24cff25 100644
--- a/lib/mbsstr.c
+++ b/lib/mbsstr.c
< at >< at > -45,11 +45,12 < at >< at > knuth_morris_pratt_multibyte (const char *haystack, const char *needle,
   size_t *table;
 
   /* Allocate room for needle_mbchars and the table.  */
-  char *memory = (char *) nmalloca (m