[libc-commits] [libc] [libc] Introduce MCS-based Flat-Combining Lambda Lock (PR #101916)

[Schrodinger ZHU Yifan via libc-commits]

================
@@ -0,0 +1,325 @@
+//===-- MCS-based Flat-Combining Lambda Lock --------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_SUPPORT_THREADS_LAMBDA_LOCK_H
+#define LLVM_LIBC_SRC_SUPPORT_THREADS_LAMBDA_LOCK_H
+
+#include "src/__support/CPP/atomic.h"
+#include "src/__support/CPP/limits.h"
+#include "src/__support/common.h"
+#include "src/__support/libc_assert.h"
+#include "src/__support/macros/attributes.h"
+#include "src/__support/threads/linux/futex_utils.h"
+#include "src/__support/threads/linux/futex_word.h"
+#include "src/__support/threads/sleep.h"
+#include "src/__support/threads/spin_lock.h"
+
+// This file contains an implementation of a flat combining lock based on MCS
+// queue.
+//
+// What is an MCS queue?
+// =====================
+// An MCS queue is a queue-based lock that is designed to be cache-friendly.
+// Each thread only spin on its local node, with minimal traffic across threads.
+// The thread-local node is not nessarily a node inside TLS storage. Rather, the
+// node can be allocated on stack. It is assumed that, during the lifespan of
+// the node, the thread is waiting in its locking routine, thus the stack space
+// is always valid.
+//
+// --------        ----------------------      ----------------------
----------------
SchrodingerZhu wrote:

I have Redrawn the box manually just now. I found online tools not easy to use.  Any recommendation?

https://github.com/llvm/llvm-project/pull/101916