]> code.delx.au - gnu-emacs/blobdiff - src/ChangeLog
* alloc.c (allocate_pseudovector): Don't use EMACS_INT when int would do.
[gnu-emacs] / src / ChangeLog
index c747a325139466787ba0fb5d177dacfe8325a44a..0c3028fb94a82e9baa9b1868551ebc4270f2796e 100644 (file)
@@ -1,5 +1,24 @@
 2011-06-08  Paul Eggert  <eggert@cs.ucla.edu>
 
+       * alloc.c: Use EMACS_INT, not int, to count objects.
+       (total_conses, total_markers, total_symbols, total_vector_size)
+       (total_free_conses, total_free_markers, total_free_symbols)
+       (total_free_floats, total_floats, total_free_intervals, total_intervals)
+       (total_strings, total_free_strings):
+       Now EMACS_INT, not int.  All uses changed.
+       (Fgarbage_collect): Compute overall total using a double, so that
+       integer overflow is less likely to be a problem.  Check for overflow
+       when converting back to an integer.
+       (n_interval_blocks, n_string_blocks, n_float_blocks, n_cons_blocks)
+       (n_vectors, n_symbol_blocks, n_marker_blocks): Remove.
+       These were 'int' variables that could overflow on 64-bit hosts;
+       they were never used, so remove them instead of repairing them.
+       (nzombies, ngcs, max_live, max_zombies): Now EMACS_INT, not 'int'.
+       (inhibit_garbage_collection): Set gc_cons_threshold to max value.
+       Previously, this ceilinged at INT_MAX, but that doesn't work on
+       64-bit machines.
+       (allocate_pseudovector): Don't use EMACS_INT when int would do.
+
        * alloc.c (Fmake_bool_vector): Don't assume vector size fits in int.
        (allocate_vectorlike): Check for ptrdiff_t overflow.
        (mark_vectorlike, mark_char_table, mark_object): Avoid EMACS_UINT