Change to how locations are passed around, for the sake of generational

GC. The issue being solved here is the accuracy of the gc_set function.
The existing impelmentation is too conservative. It has no generation
information about the memory location being stored, and so it assumes
the worst: that it is a location in the middle of a gen 1 object.
This is sub-optimal, creating unacceptable pressure against the
checkobj array and, worse, as a consequence causing unreachable gen 0
objects to be tenured into gen 1.

To solve this problem, we replace "val *" pointers with a structure
of type "loc" which keeps track of the object too, which lets us
discover the generation.

I tried another approach: using just a pointer with a bitfield
indicating the generation.  This turned out to have a serious issue:
such a bitfield goes stale when the object is moved to a different
generation. The object holding the memory location is in gen 1, but the
annotated pointer still indicates gen 0. The gc_set function then
makes the wrong decision, and premature reclamation takes place.

* combi.c (perm_init_common, comb_gen_fun_common,
rcomb_gen_fun_common, rcomb_list_gen_fun): Update to new interfaces
for managing mutation.

* debug.c (debug): Update to new interfaces for managing mutation.
Avoid loc variable name.

* eval.c (env_fbind, env_fbind): Update to new interfaces
for managing mutation.
(lookup_var_l, dwim_loc): Return loc type and update to new interfaces.
(apply_frob_args, op_modplace, op_dohash, transform_op, mapcarv,
mappendv, repeat_infinite_func, repeat_times_func): Update to new
interfaces for managing mutation.

* eval.h (lookup_var_l): Declaration updated.

* filter.c (trie_add, trie_compress, trie_compress_intrinsic,
* build_filter, built_filter_from_list, filter_init): Update to new
* interfaces.

* gc.c (gc_set): Rewritten to use loc type which provides the exact
generation. We do not need the in_malloc_range hack any more, since
we have the backpointer to the object.
(gc_push): Take loc rather than raw pointer.

* gc.h (gc_set, gc_push): Declarations updated.

* hash.c (struct hash): The acons* functions use loc instead
of val * now.
(hash_equal_op, copy_hash, gethash_c, inhash, gethash_n, pushhash,

Change to how locations are passed around, for the sake of generational
GC. The issue being solved here is the accuracy of the gc_set function.
The existing impelmentation is too conservative. It has no generation
information about the memory location being stored, and so it assumes
the worst: that it is a location in the middle of a gen 1 object.
This is sub-optimal, creating unacceptable pressure against the
checkobj array and, worse, as a consequence causing unreachable gen 0
objects to be tenured into gen 1.

To solve this problem, we replace "val *" pointers with a structure
of type "loc" which keeps track of the object too, which lets us
discover the generation.

I tried another approach: using just a pointer with a bitfield
indicating the generation.  This turned out to have a serious issue:
such a bitfield goes stale when the object is moved to a different
generation. The object holding the memory location is in gen 1, but the
annotated pointer still indicates gen 0. The gc_set function then
makes the wrong decision, and premature reclamation takes place.

* combi.c (perm_init_common, comb_gen_fun_common,
rcomb_gen_fun_common, rcomb_list_gen_fun): Update to new interfaces
for managing mutation.

* debug.c (debug): Update to new interfaces for managing mutation.
Avoid loc variable name.

* eval.c (env_fbind, env_fbind): Update to new interfaces
for managing mutation.
(lookup_var_l, dwim_loc): Return loc type and update to new interfaces.
(apply_frob_args, op_modplace, op_dohash, transform_op, mapcarv,
mappendv, repeat_infinite_func, repeat_times_func): Update to new
interfaces for managing mutation.

* eval.h (lookup_var_l): Declaration updated.

* filter.c (trie_add, trie_compress, trie_compress_intrinsic,
* build_filter, built_filter_from_list, filter_init): Update to new
* interfaces.

* gc.c (gc_set): Rewritten to use loc type which provides the exact
generation. We do not need the in_malloc_range hack any more, since
we have the backpointer to the object.
(gc_push): Take loc rather than raw pointer.

* gc.h (gc_set, gc_push): Declarations updated.

* hash.c (struct hash): The acons* functions use loc instead
of val * now.
(hash_equal_op, copy_hash, gethash_c, inhash, gethash_n, pushhash,

1 files changed, 10 insertions, 10 deletions

diff --git a/match.c b/match.c
index ae37a56d..25b7c4db 100644
--- a/match.c
+++ b/match.c
@@ -252,7 +252,7 @@ static val dest_set(val spec, val bindings, val pattern, val value)
       sem_error(spec, lit("~s cannot be used as a variable"), pattern, nao);
     if (!existing)
       sem_error(spec, lit("cannot set unbound variable ~s"), pattern, nao);
-    set(*cdr_l(existing), value);
+    set(cdr_l(existing), value);
   } else if (consp(pattern)) {
     if (first(pattern) == var_s) {
       uw_throwf(query_error_s, 
@@ -751,7 +751,7 @@ static val h_coll(match_line_ctx *c)
   val maxtimes = txeval(elem, getplist(args, maxtimes_k), c->bindings);
   val chars = txeval(elem, getplist(args, chars_k), c->bindings);
   val have_vars;
-  val vars = getplist_f(args, vars_k, &have_vars);
+  val vars = getplist_f(args, vars_k, mkcloc(have_vars));
   cnum cmax = fixnump(gap) ? c_num(gap) : (fixnump(max) ? c_num(max) : 0);
   cnum cmin = fixnump(gap) ? c_num(gap) : (fixnump(min) ? c_num(min) : 0);
   cnum mincounter = cmin, maxcounter = 0;
@@ -2586,7 +2586,7 @@ static val v_gather(match_files_ctx *c)
                  match_files(mf_spec(*c, nested_spec)));
 
       if (!success) {
-        *cdr_l(iter) = nil;
+        deref(cdr_l(iter)) = nil;
         ptail = list_collect_nconc(ptail, iter);
       } else if (success == t) {
         c->bindings = new_bindings;
@@ -2689,7 +2689,7 @@ static val v_collect(match_files_ctx *c)
   val maxtimes = txeval(specline, getplist(args, maxtimes_k), c->bindings);
   val lines = txeval(specline, getplist(args, lines_k), c->bindings);
   val have_vars;
-  volatile val vars = getplist_f(args, vars_k, &have_vars);
+  volatile val vars = getplist_f(args, vars_k, mkcloc(have_vars));
   cnum cmax = fixnump(gap) ? c_num(gap) : (fixnump(max) ? c_num(max) : 0);
   cnum cmin = fixnump(gap) ? c_num(gap) : (fixnump(min) ? c_num(min) : 0);
   cnum mincounter = cmin, maxcounter = 0;
@@ -2816,7 +2816,7 @@ static val v_collect(match_files_ctx *c)
 
           c->data = new_data;
           c->data_lineno = new_line;
-          *car_l(success) = nil;
+          deref(car_l(success)) = nil;
         } else {
           debuglf(specline, lit("collect consumed entire file"), nao);
           c->data = nil;
@@ -2897,7 +2897,7 @@ static val v_flatten(match_files_ctx *c)
       val existing = assoc(sym, c->bindings);
 
       if (existing)
-        set(*cdr_l(existing), flatten(cdr(existing)));
+        set(cdr_l(existing), flatten(cdr(existing)));
     }
   }
 
@@ -3056,7 +3056,7 @@ static val v_cat(match_files_ctx *c)
     if (existing) {
       val sep = if3(sep_form, txeval(specline, sep_form, c->bindings),
                     lit(" "));
-      set(*cdr_l(existing), cat_str(flatten(cdr(existing)), sep));
+      set(cdr_l(existing), cat_str(flatten(cdr(existing)), sep));
     } else {
       sem_error(specline, lit("cat: unbound variable ~s"), sym, nao);
     }
@@ -3145,9 +3145,9 @@ static val v_output(match_files_ctx *c)
 
         if (existing) {
           if (append) {
-            set(*cdr_l(existing), append2(flatten(cdr(existing)), list_out));
+            set(cdr_l(existing), append2(flatten(cdr(existing)), list_out));
           } else {
-            set(*cdr_l(existing), list_out);
+            set(cdr_l(existing), list_out);
           }
         } else {
           c->bindings = acons(into_var, list_out, c->bindings);
@@ -3462,7 +3462,7 @@ static val v_filter(match_files_ctx *c)
     if (!existing)
       sem_error(specline, lit("filter: variable ~a is unbound"), var, nao);
 
-    set(*cdr_l(existing), filter_string_tree(filter, cdr(existing)));
+    set(cdr_l(existing), filter_string_tree(filter, cdr(existing)));
   }
 
   uw_env_end;