[PATCH RFA #2] utils.c: Fix xcalloc (0, 0) behavior

[PATCH RFA #2] utils.c: Fix xcalloc (0, 0) behavior

[Kevin Buettner]

The patch below supercedes the patch that I submitted in

    http://sources.redhat.com/ml/gdb-patches/2001-03/msg00052.html

The patch below is much more aggressive in that it will cause NULL to
be returned for *all* platforms when a zero-sized request is made to
one of GDB's memory allocation functions.  The patch in the message
referenced above only returned NULL when the underlying allocation
functions (upon which GDB's are implemented) chose to return NULL.  I
think the consensus opinion is that it would be better to make the
behavior consistent across platforms so that bugs will be caught
sooner.

I have tested this patch on i386-unknown-freebsd4.2 and
i686-pc-linux-gnu and do not see any regressions.

	* utils.c (xmrealloc, xcalloc): Return NULL for zero-sized requests.

Index: utils.c
===================================================================
RCS file: /cvs/src/src/gdb/utils.c,v
retrieving revision 1.31
diff -u -p -r1.31 utils.c
--- utils.c	2001/02/25 04:45:11	1.31
+++ utils.c	2001/03/05 23:24:05
@@ -1042,18 +1042,27 @@ xmrealloc (PTR md, PTR ptr, long size)
 {
   register PTR val;
 
-  if (ptr != NULL)
+  if (size == 0)
     {
-      val = mrealloc (md, ptr, size);
+      if (ptr != NULL)
+	mfree (md, ptr);
+      val = NULL;
     }
   else
     {
-      val = mmalloc (md, size);
+      if (ptr != NULL)
+	{
+	  val = mrealloc (md, ptr, size);
+	}
+      else
+	{
+	  val = mmalloc (md, size);
+	}
+      if (val == NULL)
+	{
+	  nomem (size);
+	}
     }
-  if (val == NULL)
-    {
-      nomem (size);
-    }
   return (val);
 }
 
@@ -1071,9 +1080,16 @@ xmalloc (size_t size)
 PTR
 xcalloc (size_t number, size_t size)
 {
-  void *mem = mcalloc (NULL, number, size);
-  if (mem == NULL)
-    nomem (number * size);
+  void *mem;
+
+  if (number == 0 || size == 0)
+    mem = NULL;
+  else
+    {
+      mem = mcalloc (NULL, number, size);
+      if (mem == NULL)
+	nomem (number * size);
+    }
   return mem;
 }
 

Re: [PATCH RFA #2] utils.c: Fix xcalloc (0, 0) behavior

[Kevin Buettner]

On Mar 5,  5:06pm, Kevin Buettner wrote:

> I have tested this patch on i386-unknown-freebsd4.2 and
> i686-pc-linux-gnu and do not see any regressions.
> 
> 	* utils.c (xmrealloc, xcalloc): Return NULL for zero-sized requests.

Another data point: I've just tested this patch on sparc-sun-solaris2.7
and all's well...

Kevin

Re: [PATCH RFA #2] utils.c: Fix xcalloc (0, 0) behavior

[Andrew Cagney]

How ironic :-)

Go to all this effort to get everything using a function that doesn't
return NULL only to find that we need to go to even more effort so that
it can return NULL. Ha!

One aside.  From memory, the asprintf() functions in ../libiberty tend
to not like:
	asprintf ("%s", NULL);
giving either:
	(null)
  or	Segmentation Fault, Core dumped.
I submitted a patch to make the behavour but had it rejected.

The thing I would take from this is that we've a GDB convention. 
Getting these specific semantics merged into ../libiberty is separate.

M'kay with me.  Anyone?

	Andrew