Fix stupid bug in brush model indix block size calculations.

We're taking indices and converting them to pointer relative to the
hunks base. Yes, that's dirty. Since the indices are stored as 32 bit
values and hunks are generally small using 32 bit pointers is enough,
even on 64 bit platforms. So the code took the size of void* / 2...
See the problem? Yes, that's not a good idea on 32 bit platforms. Bite
the bullet and just take the size of void*. Shouldn't be a problem,
because the indices are the first thing that's loaded and the hunk is
trimmed right after it anyways. If, and just if, we really need each and
every byte in the early stages of map loading we need two cases. One for
64 bit and one for 32 bit.

This fixes issue #346. Kudos to @ricardosdl for the analysis.

src/client/refresh/gl1/gl1_model.c

@@ -828,9 +828,15 @@ Mod_LoadBrushModel(model_t *mod, void *buffer, int modfilelen)
 	dheader_t *header;
 	mmodel_t *bm;
 
-	// map use short indexes that we convert to pointers
-	// pointer size / 2 bytes
-	loadmodel->extradata = Hunk_Begin(modfilelen * sizeof(void*) / 2);
+	/* Because Quake II is is sometimes so ... "optimized" this
+	 * is going to be somewhat dirty. The map data contains indices
+	 * that we're converting into pointers. Yeah. No comments. The
+	 * indices are 32 bit long, they just encode the offset between
+	 * the hunks base address and the position in the hunk, so 32 bit
+	 * pointers should be enough. But let's play save, waste some
+	 * allocations and just take the plattforms pointer size instead
+	 * of relying on assumptions. */
+	loadmodel->extradata = Hunk_Begin(modfilelen * sizeof(void*));
 	loadmodel->type = mod_brush;
 
 	if (loadmodel != mod_known)

src/client/refresh/gl3/gl3_model.c

@@ -714,9 +714,15 @@ Mod_LoadBrushModel(gl3model_t *mod, void *buffer, int modfilelen)
 	dheader_t *header;
 	mmodel_t *bm;
 
-	// map use short indexes that we convert to pointers
-	// pointer size / 2 bytes
-	loadmodel->extradata = Hunk_Begin(modfilelen * sizeof(void*) / 2);
+	/* Because Quake II is is sometimes so ... "optimized" this
+	 * is going to be somewhat dirty. The map data contains indices
+	 * that we're converting into pointers. Yeah. No comments. The
+	 * indices are 32 bit long, they just encode the offset between
+	 * the hunks base address and the position in the hunk, so 32 bit
+	 * pointers should be enough. But let's play save, waste some
+	 * allocations and just take the plattforms pointer size instead
+	 * of relying on assumptions. */
+	loadmodel->extradata = Hunk_Begin(modfilelen * sizeof(void*));
 	loadmodel->type = mod_brush;
 
 	if (loadmodel != mod_known)

src/client/refresh/soft/sw_model.c

@@ -882,8 +882,14 @@ Mod_LoadBrushModel(model_t *mod, void *buffer, int modfilelen)
 	dheader_t	*header;
 	dmodel_t 	*bm;
 
-	// map use short indexes that we convert to pointers
-	// pointer size / 2 bytes
+	/* Because Quake II is is sometimes so ... "optimized" this
+	 * is going to be somewhat dirty. The map data contains indices
+	 * that we're converting into pointers. Yeah. No comments. The
+	 * indices are 32 bit long, we're loading the map somewhere at
+	 * the lower end of the address space (at least on the common
+	 * platforms) so 32 bit pointers should be enough. But let's
+	 * play save, waste some memory and just take the plattforms
+	 * pointer size instead of relying on assumptions. */
 	loadmodel->extradata = Hunk_Begin(modfilelen * sizeof(void*) / 2);
 
 	loadmodel->type = mod_brush;