[RFC/PATCH] workaround breakage from urcu v0.11.4

[Eric Wong <e@80x24.org>]

urcu v0.11.4+ introduced commit
7ca7fe9c03 (Make temporary variable in _rcu_dereference non-const, 2021-07-29)
which conflicts with our use of _LGPL_SOURCE.  In retrospect,
CMM_LOAD_SHARED and CMM_STORE_SHARED seem sufficient for our use
of the `totals' cds_lfht pointer since the constructur should always
fire before any threads are running.

Link: https://lore.kernel.org/lttng-dev/20220809181927.GA3718@dcvr/
---
 I'm nearly certain this change is correct and prior use of
 rcu_dereference was overkill; but confirmation from a URCU
 hacker would be nice.

 ext/mwrap/mwrap.c | 19 +++++++++----------
 1 file changed, 9 insertions(+), 10 deletions(-)

diff --git a/ext/mwrap/mwrap.c b/ext/mwrap/mwrap.c
index 4575e34..477b1cb 100644
--- a/ext/mwrap/mwrap.c
+++ b/ext/mwrap/mwrap.c
@@ -139,8 +139,8 @@ __attribute__((constructor)) static void resolve_malloc(void)
 		_exit(1);
 	}
 #endif /* !FreeBSD */
-	totals = lfht_new();
-	if (!totals)
+	CMM_STORE_SHARED(totals, lfht_new());
+	if (!CMM_LOAD_SHARED(totals))
 		fprintf(stderr, "failed to allocate totals table\n");
 
 	err = pthread_atfork(call_rcu_before_fork,
@@ -375,7 +375,7 @@ static struct src_loc *totals_add_rcu(struct src_loc *k)
 	struct cds_lfht *t;
 
 again:
-	t = rcu_dereference(totals);
+	t = CMM_LOAD_SHARED(totals);
 	if (!t) goto out_unlock;
 	cds_lfht_lookup(t, k->hval, loc_eq, k, &iter);
 	cur = cds_lfht_iter_get_node(&iter);
@@ -417,7 +417,7 @@ static struct src_loc *update_stats_rcu_lock(size_t size, uintptr_t caller)
 	static const size_t xlen = sizeof(caller);
 	char *dst;
 
-	if (caa_unlikely(!totals)) return 0;
+	if (caa_unlikely(!CMM_LOAD_SHARED(totals))) return 0;
 	if (locating++) goto out; /* do not recurse into another *alloc */
 
 	uatomic_add(&total_bytes_inc, size);
@@ -808,7 +808,7 @@ static void *dump_to_file(void *x)
 
 	++locating;
 	rcu_read_lock();
-	t = rcu_dereference(totals);
+	t = CMM_LOAD_SHARED(totals);
 	if (!t)
 		goto out_unlock;
 	cds_lfht_for_each_entry(t, &iter, l, hnode) {
@@ -877,7 +877,7 @@ static void *totals_reset(void *ign)
 	uatomic_set(&total_bytes_dec, 0);
 
 	rcu_read_lock();
-	t = rcu_dereference(totals);
+	t = CMM_LOAD_SHARED(totals);
 	cds_lfht_for_each_entry(t, &iter, l, hnode) {
 		uatomic_set(&l->total, 0);
 		uatomic_set(&l->allocations, 0);
@@ -945,7 +945,7 @@ static VALUE dump_each_rcu(VALUE x)
 	struct cds_lfht_iter iter;
 	struct src_loc *l;
 
-	t = rcu_dereference(totals);
+	t = CMM_LOAD_SHARED(totals);
 	cds_lfht_for_each_entry(t, &iter, l, hnode) {
 		VALUE v[6];
 		if (l->total <= a->min) continue;
@@ -1049,9 +1049,9 @@ static VALUE mwrap_aref(VALUE mod, VALUE loc)
 
 	if (!k) return val;
 
+	t = CMM_LOAD_SHARED(totals);
+	if (!t) return val;
 	rcu_read_lock();
-	t = rcu_dereference(totals);
-	if (!t) goto out_unlock;
 
 	cds_lfht_lookup(t, k->hval, loc_eq, k, &iter);
 	cur = cds_lfht_iter_get_node(&iter);
@@ -1059,7 +1059,6 @@ static VALUE mwrap_aref(VALUE mod, VALUE loc)
 		l = caa_container_of(cur, struct src_loc, hnode);
 		val = TypedData_Wrap_Struct(cSrcLoc, &src_loc_type, l);
 	}
-out_unlock:
 	rcu_read_unlock();
 	return val;
 }