Re: [PATCH] m4/pthread-cond.m4: fix build without threads

[Bruno Haible]

Hi,

Fabrice Fontaine wrote:
> HAVE_PTHREAD_COND_xxx variables are wrongly set to 1 instead of 0 if
> HAVE_PTHREAD_H is set to 0 since
> https://git.savannah.gnu.org/cgit/gnulib.git/commit/?id=e08989ca11a13c8a09cb0c34f797af5a40753379

Indeed, that's obviously a copy&paste mistake.

> This will result in the following build failure if threads are not
> available, for example with coreutils in version 8.32:

Thanks for the analysis and fix. Applied with a ChangeLog entry:


2021-04-04  Fabrice Fontaine  <fontaine.fabrice@gmail.com>

        pthread-cond: Fix wrong configure results when <pthread.h> is absent.
        * m4/pthread-cond.m4 (gl_PTHREAD_COND): When <pthread.h> does not exist,
        set HAVE_PTHREAD_COND_* to 0, not to 1.

diff --git a/m4/pthread-cond.m4 b/m4/pthread-cond.m4
index 75e37c6..8922a8a 100644
--- a/m4/pthread-cond.m4
+++ b/m4/pthread-cond.m4
@@ -1,4 +1,4 @@
-# pthread-cond.m4 serial 1
+# pthread-cond.m4 serial 2
 dnl Copyright (C) 2019-2021 Free Software Foundation, Inc.
 dnl This file is free software; the Free Software Foundation
 dnl gives unlimited permission to copy and/or distribute it,
@@ -23,14 +23,14 @@ AC_DEFUN([gl_PTHREAD_COND],
     REPLACE_PTHREAD_COND_DESTROY=1
   else
     if test $HAVE_PTHREAD_H = 0; then
-      HAVE_PTHREAD_COND_INIT=1
-      HAVE_PTHREAD_CONDATTR_INIT=1
-      HAVE_PTHREAD_CONDATTR_DESTROY=1
-      HAVE_PTHREAD_COND_WAIT=1
-      HAVE_PTHREAD_COND_TIMEDWAIT=1
-      HAVE_PTHREAD_COND_SIGNAL=1
-      HAVE_PTHREAD_COND_BROADCAST=1
-      HAVE_PTHREAD_COND_DESTROY=1
+      HAVE_PTHREAD_COND_INIT=0
+      HAVE_PTHREAD_CONDATTR_INIT=0
+      HAVE_PTHREAD_CONDATTR_DESTROY=0
+      HAVE_PTHREAD_COND_WAIT=0
+      HAVE_PTHREAD_COND_TIMEDWAIT=0
+      HAVE_PTHREAD_COND_SIGNAL=0
+      HAVE_PTHREAD_COND_BROADCAST=0
+      HAVE_PTHREAD_COND_DESTROY=0
     fi
   fi
 ])