{"id":4855,"date":"2016-08-18T07:27:29","date_gmt":"2016-08-18T07:27:29","guid":{"rendered":"https:\/\/www.modernescpp.com\/index.php\/ongoing-optimization-a-data-race-with-cppmem\/"},"modified":"2023-06-26T12:45:38","modified_gmt":"2023-06-26T12:45:38","slug":"ongoing-optimization-a-data-race-with-cppmem","status":"publish","type":"post","link":"https:\/\/www.modernescpp.com\/index.php\/ongoing-optimization-a-data-race-with-cppmem\/","title":{"rendered":"Ongoing Optimization: A Data Race with CppMem"},"content":{"rendered":"

But we can improve and further improve the acquire-release semantics of the last post<\/a>. Why should x be atomic? There is no reason. That was my first but incorrect assumption. See why?<\/p>\n

<\/p>\n

A typical misunderstanding<\/a> in applying the acquire-release semantic is to assume that the acquire operation is waiting for the release operation. So based on this assumption, you may think that x has not be an atomic variable. So we can further optimize the program.<\/p>\n

<\/p>\n

\n\n\n\n
\n
 1\r\n 2\r\n 3\r\n 4\r\n 5\r\n 6\r\n 7\r\n 8\r\n 9\r\n10\r\n11\r\n12\r\n13\r\n14\r\n15\r\n16\r\n17\r\n18\r\n19\r\n20\r\n21\r\n22\r\n23\r\n24\r\n25<\/pre>\n<\/td>\n
\n
\/\/ ongoingOptimizationAcquireReleaseBroken.cpp<\/span>\r\n\r\n#include <atomic><\/span>\r\n#include <iostream><\/span>\r\n#include <thread><\/span>\r\n\r\nint<\/span> x= 0;\r\nstd::atomic<int<\/span>> y{0};\r\n\r\nvoid<\/span> writing(){  \r\n  x= 2000;  \r\n  y.store(11,std::memory_order_release);\r\n}\r\n\r\nvoid<\/span> reading(){  \r\n  std::cout << y.load(std::memory_order_acquire) << \" \"<\/span>;  \r\n  std::cout << x << std::endl;\r\n}\r\n\r\nint<\/span> main(){\r\n  std::thread<\/span> thread1(writing);\r\n  std::thread<\/span> thread2(reading);\r\n  thread1.join();\r\n  thread2.join();\r\n};\r\n<\/pre>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n
 <\/p>\n
The program has a data race<\/a> on x and has, therefore, undefined behavior. If y.store(11,std::memory_order_release)<\/span> (line 12) is executed before  y.load(std::memory_order_acquire)<\/span> (line 16), the acquire-release semantic guarantees that x= 2000<\/span> (line 11) is executed before the reading of x in line 17. But if not. In this case, the reading of x will be executed simultaneously as the writing of x. So we have concurrent access to a shared variable, one of them is a write. That’s, per definition, a data race.  <\/span><\/p>\n<\/p>\n
The table puts it in a nutshell.<\/p>\n
\"undefinedEng\"<\/p>\n
 <\/p>\n
I made this mistake in my presentation “Multithreading done right?<\/a>” in Berlin<\/a>. In Moscow<\/a>, I did it right. I never claimed that the C++ memory model is a piece of cake.<\/p>\n
Now it’s time for CppMem. Let’s see what CppMem finds out.<\/p>\n
CppMem<\/h3>\n
 <\/p>\n
<\/p>\n
\n
int<\/span> main(){\r\n  int<\/span> x= 0;\r\n  atomic_int y= 0;\r\n  {{{ { \r\n      x= 2000;\r\n      y.store(11,memory_order_release);\r\n      }\r\n  ||| {\r\n      y.load(memory_order_acquire);\r\n      x;\r\n      }\r\n  }}}\r\n}\r\n<\/pre>\n<\/div>\n
The data race<\/a> occurs if one thread writes x= 2000 and the other reads x. The graph shows a dr<\/strong><\/span> symbol (data race) on the arrow. <\/em><\/p>\n
\"raceAcquireRelease\"<\/p>\n
What’s next?<\/h2>\n
The ultimate step in the process of ongoing optimization is still missing. In the next post<\/a>, I will use the relaxed semantic.<\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n
 <\/p>\n","protected":false},"excerpt":{"rendered":"
But we can improve and further improve the acquire-release semantics of the last post. Why should x be atomic? There is no reason. That was my first but incorrect assumption. See why?<\/p>\n","protected":false},"author":21,"featured_media":4840,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[369],"tags":[505,434,486,521],"class_list":["post-4855","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-multithreading-application","tag-acquire-release-semantic","tag-atomics","tag-cppmem","tag-ongoing-optimization"],"_links":{"self":[{"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/posts\/4855","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/users\/21"}],"replies":[{"embeddable":true,"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/comments?post=4855"}],"version-history":[{"count":1,"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/posts\/4855\/revisions"}],"predecessor-version":[{"id":6954,"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/posts\/4855\/revisions\/6954"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/media\/4840"}],"wp:attachment":[{"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/media?parent=4855"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/categories?post=4855"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.modernescpp.com\/index.php\/wp-json\/wp\/v2\/tags?post=4855"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}