malloc(0) problem again

[Bruno Haible]

Unit tests revealed that some Unicode string functions were dependent on
the result of malloc(0). This fixes it, and documents the portability problem.


2009-04-26  Bruno Haible  <address@hidden>

        Avoid test failures on AIX and OSF/1.
        * lib/uniconv/u8-conv-from-enc.c (u8_conv_from_encoding): Avoid calling
        malloc(0).
        * lib/uniconv/u8-conv-to-enc.c (u8_conv_to_encoding): Likewise.
        * lib/unilbrk/ulc-possible-linebreaks.c (ulc_possible_linebreaks):
        Likewise.
        * lib/unilbrk/ulc-width-linebreaks.c (ulc_width_linebreaks): Likewise.
        * lib/uniwbrk/ulc-wordbreaks.c (ulc_wordbreaks): Likewise.
        * lib/uniconv/u-conv-to-enc.h (FUNC): Likewise. Fix memory leak.
        * lib/unistr/u-cpy-alloc.h (FUNC): Call malloc(1) instead of malloc(0).
        * doc/posix-functions/malloc.texi: Document the portability problem
        related to malloc(0).

--- doc/posix-functions/malloc.texi.orig        2009-04-26 12:14:01.000000000 
+0200
+++ doc/posix-functions/malloc.texi     2009-04-26 12:13:18.000000000 +0200
@@ -16,7 +16,10 @@
 
 Portability problems not fixed by Gnulib:
 @itemize
address@hidden (0)} always returns a NULL pointer on some platforms:
+AIX 5.1, OSF/1 5.1.
 @end itemize
 
 Extension: Gnulib provides a module @samp{malloc} that substitutes a
address@hidden implementation that behaves more like the glibc implementation.
address@hidden implementation that behaves more like the glibc implementation,
+regarding the result of @code{malloc (0)}.
--- lib/uniconv/u-conv-to-enc.h.orig    2009-04-26 12:14:01.000000000 +0200
+++ lib/uniconv/u-conv-to-enc.h 2009-04-26 11:26:01.000000000 +0200
@@ -1,5 +1,5 @@
 /* Conversion from UTF-16/UTF-32 to legacy encodings.
-   Copyright (C) 2002, 2006-2008 Free Software Foundation, Inc.
+   Copyright (C) 2002, 2006-2009 Free Software Foundation, Inc.
 
    This program is free software: you can redistribute it and/or modify it
    under the terms of the GNU Lesser General Public License as published
@@ -25,7 +25,7 @@
   size_t *scaled_offsets;
   int retval;
 
-  if (offsets != NULL)
+  if (offsets != NULL && srclen > 0)
     {
       scaled_offsets =
        (size_t *) malloc (srclen * sizeof (UNIT) * sizeof (size_t));
@@ -70,7 +70,7 @@
     return -1;
   utf8_srclen = tmpbufsize;
 
-  if (offsets != NULL)
+  if (offsets != NULL && utf8_srclen > 0)
     {
       scaled_offsets = (size_t *) malloc (utf8_srclen * sizeof (size_t));
       if (scaled_offsets == NULL)
@@ -88,12 +88,11 @@
                                scaled_offsets, resultp, lengthp);
   if (retval < 0)
     {
+      int saved_errno = errno;
+      free (scaled_offsets);
       if (utf8_src != tmpbuf)
-       {
-         int saved_errno = errno;
-         free (utf8_src);
-         errno = saved_errno;
-       }
+       free (utf8_src);
+      errno = saved_errno;
       return -1;
     }
   if (offsets != NULL)
--- lib/uniconv/u8-conv-from-enc.c.orig 2009-04-26 12:14:01.000000000 +0200
+++ lib/uniconv/u8-conv-from-enc.c      2009-04-26 11:50:35.000000000 +0200
@@ -1,5 +1,5 @@
 /* Conversion to UTF-8 from legacy encodings.
-   Copyright (C) 2002, 2006-2007 Free Software Foundation, Inc.
+   Copyright (C) 2002, 2006-2007, 2009 Free Software Foundation, Inc.
 
    This program is free software: you can redistribute it and/or modify it
    under the terms of the GNU Lesser General Public License as published
@@ -65,7 +65,7 @@
        }
 
       /* Memory allocation.  */
-      if (*resultp != NULL && *lengthp >= srclen)
+      if ((*resultp != NULL && *lengthp >= srclen) || srclen == 0)
        result = *resultp;
       else
        {
--- lib/uniconv/u8-conv-to-enc.c.orig   2009-04-26 12:14:01.000000000 +0200
+++ lib/uniconv/u8-conv-to-enc.c        2009-04-26 11:49:32.000000000 +0200
@@ -1,5 +1,5 @@
 /* Conversion from UTF-8 to legacy encodings.
-   Copyright (C) 2002, 2006-2007 Free Software Foundation, Inc.
+   Copyright (C) 2002, 2006-2007, 2009 Free Software Foundation, Inc.
 
    This program is free software: you can redistribute it and/or modify it
    under the terms of the GNU Lesser General Public License as published
@@ -50,7 +50,7 @@
 #endif
 
       /* Memory allocation.  */
-      if (*resultp != NULL && *lengthp >= srclen)
+      if ((*resultp != NULL && *lengthp >= srclen) || srclen == 0)
        result = *resultp;
       else
        {
--- lib/unilbrk/ulc-possible-linebreaks.c.orig  2009-04-26 12:14:01.000000000 
+0200
+++ lib/unilbrk/ulc-possible-linebreaks.c       2009-04-26 11:52:19.000000000 
+0200
@@ -1,5 +1,5 @@
 /* Line breaking of strings.
-   Copyright (C) 2001-2003, 2006-2008 Free Software Foundation, Inc.
+   Copyright (C) 2001-2003, 2006-2009 Free Software Foundation, Inc.
    Written by Bruno Haible <address@hidden>, 2001.
 
    This program is free software: you can redistribute it and/or modify it
@@ -62,9 +62,9 @@
                                         s, n, offsets, &t, &m)
                  == 0)
                {
-                 char *q = (char *) malloc (m);
+                 char *q = (char *) (m > 0 ? malloc (m) : NULL);
 
-                 if (q != NULL)
+                 if (m == 0 || q != NULL)
                    {
                      size_t i;
 
--- lib/unilbrk/ulc-width-linebreaks.c.orig     2009-04-26 12:14:01.000000000 
+0200
+++ lib/unilbrk/ulc-width-linebreaks.c  2009-04-26 11:54:37.000000000 +0200
@@ -1,5 +1,5 @@
 /* Line breaking of strings.
-   Copyright (C) 2001-2003, 2006-2008 Free Software Foundation, Inc.
+   Copyright (C) 2001-2003, 2006-2009 Free Software Foundation, Inc.
    Written by Bruno Haible <address@hidden>, 2001.
 
    This program is free software: you can redistribute it and/or modify it
@@ -64,9 +64,10 @@
                                         s, n, offsets, &t, &m)
                  == 0)
                {
-                 char *memory = (char *) malloc (m + (o != NULL ? m : 0));
+                 char *memory =
+                   (char *) (m > 0 ? malloc (m + (o != NULL ? m : 0)) : NULL);
 
-                 if (memory != NULL)
+                 if (m == 0 || memory != NULL)
                    {
                      char *q = (char *) memory;
                      char *o8 = (o != NULL ? (char *) (q + m) : NULL);
--- lib/unistr/u-cpy-alloc.h.orig       2009-04-26 12:14:01.000000000 +0200
+++ lib/unistr/u-cpy-alloc.h    2009-04-26 11:56:20.000000000 +0200
@@ -1,5 +1,5 @@
 /* Copy piece of UTF-8/UTF-16/UTF-32 string.
-   Copyright (C) 1999, 2002, 2006-2007 Free Software Foundation, Inc.
+   Copyright (C) 1999, 2002, 2006-2007, 2009 Free Software Foundation, Inc.
    Written by Bruno Haible <address@hidden>, 2002.
 
    This program is free software: you can redistribute it and/or modify it
@@ -23,7 +23,7 @@
 {
   UNIT *dest;
 
-  dest = (UNIT *) malloc (n * sizeof (UNIT));
+  dest = (UNIT *) malloc (n > 0 ? n * sizeof (UNIT) : 1);
   if (dest != NULL)
     {
 #if 0
--- lib/uniwbrk/ulc-wordbreaks.c.orig   2009-04-26 12:14:01.000000000 +0200
+++ lib/uniwbrk/ulc-wordbreaks.c        2009-04-26 11:55:32.000000000 +0200
@@ -64,9 +64,9 @@
                                         s, n, offsets, &t, &m)
                  == 0)
                {
-                 char *q = (char *) malloc (m);
+                 char *q = (char *) (m > 0 ? malloc (m) : NULL);
 
-                 if (q != NULL)
+                 if (m == 0 || q != NULL)
                    {
                      size_t i;