Load-link/store-conditional

在计算机科学中,load-linked/store-conditional(LL/SC)，也会被称作load-reserved/store-conditional (LR/SC)，load-link与store-conditional (LL/SC)是一对用于并发同步访问内存的CPU指令。Load-link返回内存位置处的当前值，随后的store-conditional在该内存位置处保存新值（如果从load-link后没有被修改）。这被用于实现无锁算法（英语：Non-blocking_algorithm）与read-modify-write原子操作。

历史

LL/SC最初是Jensen, Hagensen, Broughton在劳伦斯利佛摩国家实验室为S-1 AAP multiprocessor （页面存档备份，存于互联网档案馆）而设计。

实现

LL/SC指令被下述架构支持：

Alpha: ldl_l/stl_c , ldq_l/stq_c
PowerPC: lwarx/stwcx , ldarx/stdcx
MIPS架构: ll/sc
ARM架构: ldrex/strex (ARMv6 , v7), ldxr/stxr (ARM version 8)
RISC-V: lr/sc

LL/SC的硬件实现典型地不允许嵌套LL/SC对。^[1] 嵌套的LL/SC机制用于提供MCAS原语（多字Compare-And-Swap，字可以是分布的）。^[2] 2013年Trevor Brown, Faith Ellen, Eric Ruppert软件实现了多地址LL/SC扩展。^[3]这被用来优化二叉搜索树。^[4]

参见

非阻塞同步（英语：Non-blocking synchronization）
事务内存（Transactional memory）

参考文献

^ Larus, James R.; Rajwar, Ravi. Transactional Memory. Morgan & Claypool. 2007: 55. ISBN 978-1-59829-124-7.
^ Fraser, Keir. Practical lock-freedom (PDF) (技术报告). University of Cambridge Computer Laboratory: 20. February 2004 [2017-11-22]. UCAM-CL-TR-579. （原始内容存档 (PDF)于2020-11-12）.
^ Brown, Trevor; Ellen, Faith; Ruppert, Eric. Pragmatic primitives for non-blocking data structures (PDF). PODC '13 Proceedings of the 2013 ACM symposium on Principles of distributed computing. ACM. 2013: 13–22 [2017-11-22]. ISBN 978-1-4503-2065-8. doi:10.1145/2484239.2484273. （原始内容 (PDF)存档于2015-02-03）. See also slides （页面存档备份，存于互联网档案馆）
^ Brown, Trevor; Ellen, Faith; Ruppert, Eric. A general technique for non-blocking trees (PDF). PPoPP '14 ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming. ACM. 2014: 329–342 [2017-11-22]. ISBN 978-1-4503-2656-8. doi:10.1145/2555243.2555267. （原始内容存档 (PDF)于2019-10-24）.

Jensen, Eric H.; Hagensen, Gary W.; Broughton, Jeffrey M. A New Approach to Exclusive Data Access in Shared Memory Multiprocessors (PDF) (技术报告). Lawrence Livermore National Laboratory. November 1987 [2017-11-22]. UCRL-97663. （原始内容 (PDF)存档于2017-02-02）.
Bruner, John D.; Hagensen, Gary W.; Jensen, Eric H.; Pattin, Jay C.; Broughton, Jeffrey M. Cache Coherence on the S-1 AAP (PDF) (技术报告). Lawrence Livermore National Laboratory. 11 November 1987 [2017-11-22]. UCRL-97646. （原始内容 (PDF)存档于2017-02-02）.
Detlefs, D.; Martin, P.; Moir, M.; Steele, Jr., Guy L. Lock-free reference counting. PODC '01 Proceedings of the twentieth annual ACM symposium on Principles of distributed computing. ACM. 2001: 190–9. ISBN 1-58113-383-9. doi:10.1145/383962.384016.
Reinholtz, Kirk. Atomic Reference Counting Pointers. C/C++ Users Journal. December 2004. ^{[永久失效链接]}
Sites, R. L. Alpha AXP architecture. Comm. ACM. February 1993, 36 (2): 33–44. doi:10.1145/151220.151226.

[LarusRajwar2007-1] Larus, James R.; Rajwar, Ravi. Transactional Memory. Morgan & Claypool. 2007: 55. ISBN 978-1-59829-124-7.

[2] Fraser, Keir. Practical lock-freedom (PDF) (技术报告). University of Cambridge Computer Laboratory: 20. February 2004 [2017-11-22]. UCAM-CL-TR-579. （原始内容存档 (PDF)于2020-11-12）.

[3] Brown, Trevor; Ellen, Faith; Ruppert, Eric. Pragmatic primitives for non-blocking data structures (PDF). PODC '13 Proceedings of the 2013 ACM symposium on Principles of distributed computing. ACM. 2013: 13–22 [2017-11-22]. ISBN 978-1-4503-2065-8. doi:10.1145/2484239.2484273. （原始内容 (PDF)存档于2015-02-03）. See also slides （页面存档备份，存于互联网档案馆）

[4] Brown, Trevor; Ellen, Faith; Ruppert, Eric. A general technique for non-blocking trees (PDF). PPoPP '14 ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming. ACM. 2014: 329–342 [2017-11-22]. ISBN 978-1-4503-2656-8. doi:10.1145/2555243.2555267. （原始内容存档 (PDF)于2019-10-24）.

[1]

[2]

[3]

[4]