Re: [PATCH v2 11/14] futex: Implement FUTEX2_NUMA

On Mon, Aug 07, 2023 at 11:11:04PM +0200, Thomas Gleixner wrote:

On Mon, Aug 07 2023 at 14:18, Peter Zijlstra wrote:

quoted

 /**
  * futex_hash - Return the hash bucket in the global hash
  * @key:	Pointer to the futex key for which the hash is calculated

@@ -114,10 +137,29 @@ late_initcall(fail_futex_debugfs);
  */
 struct futex_hash_bucket *futex_hash(union futex_key *key)
 {
-	u32 hash = jhash2((u32 *)key, offsetof(typeof(*key), both.offset) / 4,
+	u32 hash = jhash2((u32 *)key,
+			  offsetof(typeof(*key), both.offset) / sizeof(u32),
 			  key->both.offset);
+	int node = key->both.node;
+
+	if (node == -1) {

NUMA_NO_NODE please all over the place.

Ah, so our (futex2) ABI states this needs to be -1, but in theory
someone could come along and change the kernel internal NUMA_NO_NODE to
something else.

That is, I explicitly chose not to use it. I can of course, because as
of now these values do match.

quoted

+		/*
+		 * In case of !FLAGS_NUMA, use some unused hash bits to pick a
+		 * node -- this ensures regular futexes are interleaved across
+		 * the nodes and avoids having to allocate multiple
+		 * hash-tables.
+		 *
+		 * NOTE: this isn't perfectly uniform, but it is fast and
+		 * handles sparse node masks.
+		 */
+		node = (hash >> futex_hashshift) % nr_node_ids;
+		if (!node_possible(node)) {
+			node = find_next_bit_wrap(node_possible_map.bits,
+						  nr_node_ids, node);
+		}

Smart.

Thanks :-)

quoted

 
+static inline unsigned int futex_size(unsigned int flags)
+{
+	return 1 << (flags & FLAGS_SIZE_MASK);
+}
+
 static inline bool futex_flags_valid(unsigned int flags)

If you reorder these two functions in the patch which introduces them,
this diff gets readable :)

Durr, I knew I was forgetting something, will do.