CINXE.COM

<?xml version="1.0" encoding="UTF-8" standalone="yes"?> <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>LKML: linux@horizon ...: Re: [PATCH 1/19] MUTEX: Introduce simple mutex implementation</title><link href="/css/message.css" rel="stylesheet" type="text/css" /><link href="/css/wrap.css" rel="alternate stylesheet" type="text/css" title="wrap" /><link href="/css/nowrap.css" rel="stylesheet" type="text/css" title="nowrap" /><link href="/favicon.ico" rel="shortcut icon" /><script src="/js/simple-calendar.js" type="text/javascript"></script><script src="/js/styleswitcher.js" type="text/javascript"></script><link rel="alternate" type="application/rss+xml" title="lkml.org : last 100 messages" href="/rss.php" /><link rel="alternate" type="application/rss+xml" title="lkml.org : last messages by 	linux@horizon ..." href="/groupie.php?aid=25483" /><script> var _paq = window._paq = window._paq || []; /* tracker methods like "setCustomDimension" should be called before "trackPageView" */ _paq.push(["setDoNotTrack", true]); _paq.push(["disableCookies"]); _paq.push(['trackPageView']); _paq.push(['enableLinkTracking']); (function() { var u="//m.lkml.org/"; _paq.push(['setTrackerUrl', u+'matomo.php']); _paq.push(['setSiteId', '1']); var d=document, g=d.createElement('script'), s=d.getElementsByTagName('script')[0]; g.async=true; g.src=u+'matomo.js'; s.parentNode.insertBefore(g,s); })(); </script></head><body onload="es.jasper.simpleCalendar.init();" itemscope="itemscope" itemtype="http://schema.org/BlogPosting"><table border="0" cellpadding="0" cellspacing="0"><tr><td width="180" align="center"><a href="/"><img style="border:0;width:135px;height:32px" src="/images/toprowlk.gif" alt="lkml.org" /></a></td><td width="32">聽</td><td class="nb"><div><a class="nb" href="/lkml"> [lkml]</a> 聽 <a class="nb" href="/lkml/2005"> [2005]</a> 聽 <a class="nb" href="/lkml/2005/12"> [Dec]</a> 聽 <a class="nb" href="/lkml/2005/12/15"> [15]</a> 聽 <a class="nb" href="/lkml/last100"> [last100]</a> 聽 <a href="/rss.php"><img src="/images/rss-or.gif" border="0" alt="RSS Feed" /></a></div><div>Views: <a href="#" class="nowrap" onclick="setActiveStyleSheet('wrap');return false;">[wrap]</a><a href="#" class="wrap" onclick="setActiveStyleSheet('nowrap');return false;">[no wrap]</a> 聽 <a class="nb" href="/lkml/mheaders/2005/12/15/149" onclick="this.href='/lkml/headers'+'/2005/12/15/149';">[headers]</a>聽 <a href="/lkml/bounce/2005/12/15/149">[forward]</a>聽 </div></td><td width="32">聽</td></tr><tr><td valign="top"><div class="es-jasper-simpleCalendar" baseurl="/lkml/"></div><div class="threadlist">Messages in this thread</div><ul class="threadlist"><li class="root"><a href="/lkml/2005/12/15/149">First message in thread</a></li><li class="origin"><a href="/lkml/2005/12/15/201"> linux@horizon ...</a><ul><li><a href="/lkml/2005/12/15/201">Linus Torvalds</a><ul><li><a href="/lkml/2005/12/15/216">Erik Mouw</a><ul><li><a href="/lkml/2005/12/15/224">(Dick Streefland)</a></li><li><a href="/lkml/2005/12/16/61">Erik Mouw</a><ul><li><a href="/lkml/2005/12/17/22">Sander</a></li></ul></li></ul></li><li><a href="/lkml/2005/12/15/269">Nikita Danilov</a></li><li><a href="/lkml/2005/12/15/274"> linux@horizon ...</a><ul><li><a href="/lkml/2005/12/15/286">Linus Torvalds</a><ul><li><a href="/lkml/2005/12/15/422"> linux@horizon ...</a></li></ul></li><li><a href="/lkml/2005/12/15/315">Steven Rostedt</a></li></ul></li><li><a href="/lkml/2005/12/15/304">Steven Rostedt</a></li></ul></li></ul></li></ul></td><td width="32" rowspan="2" class="c" valign="top"><img src="/images/icornerl.gif" width="32" height="32" alt="/" /></td><td class="c" rowspan="2" valign="top" style="padding-top: 1em"><table><tr><td><table><tr><td class="lp">Date</td><td class="rp" itemprop="datePublished">15 Dec 2005 08:58:12 -0500</td></tr><tr><td class="lp">From</td><td class="rp" itemprop="author"> linux@horizon ...</td></tr><tr><td class="lp">Subject</td><td class="rp" itemprop="name">Re: [PATCH 1/19] MUTEX: Introduce simple mutex implementation</td></tr></table></td><td></td></tr></table><pre itemprop="articleBody">Our Fearless Leader, in a fit of madness, intoned:<br />> A real semaphore is counting. <br />> <br />> Dammit, unless the pure mutex has a _huge_ performance advantage on major <br />> architectures, we're not changing it. There's absolutely zero point. A <br />> counting semaphore is a perfectly fine mutex - the fact that it can _also_ <br />> be used to allow more than 1 user into a critical region and generally do <br />> other things is totally immaterial.<br /><br />You're being thick today. Pay attention to the arguments.<br /><br />A counting semaphore is NOT a perfectly fine mutex, and it SHOULD be changed.<br /><br />People are indeed unhappy with the naming, and whether patching 95%<br />of the callers of up() and down() is a good idea is a valid and active<br />subject of debate. (For an out-of-tree -rt patch, is was certaintly<br />an extremely practical solution.)<br /><br />But regardless of the eventual naming convention, mutexes are a good idea.<br />A mutex is *safer* than a counting semaphore. That's the main benefit.<br />Indeed, unless there's a performance advantage to a counting semaphore,<br />you should use a mutex!<br /><br />It documents in the source what you're doing. Using a counting semaphore<br />for a mutex is as silly as using a float for a loop index. Even if<br />there isn't much speed penalty on modern processors.<br /><br />Or perhaps I should compare it to using void * everywhere. That's a<br />perfectly fine pointer; why type-check it?<br /><br />A separate mutex type allows extra error-checking. You can keep track<br />of the current holder (since there can be only one) and check that the<br />same person released it and didn't try to double-acquire it.<br /><br />You can do priority inheritance, which is the main motivation for doing<br />this work in the -rt patches.<br /><br />This isn't about speed, it's about bug-free code. And having a mutex<br />abstraction distinct from a general counting semaphore is damn useful<br />error-checking, even if it is simply a thin wrapper over a counting<br />semaphore. The only reason the code is full of counting semaphores<br />right now is that that's all people had.<br /><br />Better to give them the right tool.<br />-<br />To unsubscribe from this list: send the line "unsubscribe linux-kernel" in<br />the body of a message to majordomo@vger.kernel.org<br />More majordomo info at <a href="http://vger.kernel.org/majordomo-info.html">http://vger.kernel.org/majordomo-info.html</a><br />Please read the FAQ at <a href="http://www.tux.org/lkml/">http://www.tux.org/lkml/</a><br /><br /></pre></td><td width="32" rowspan="2" class="c" valign="top"><img src="/images/icornerr.gif" width="32" height="32" alt="\" /></td></tr><tr><td align="right" valign="bottom"> 聽 </td></tr><tr><td align="right" valign="bottom">聽</td><td class="c" valign="bottom" style="padding-bottom: 0px"><img src="/images/bcornerl.gif" width="32" height="32" alt="\" /></td><td class="c">聽</td><td class="c" valign="bottom" style="padding-bottom: 0px"><img src="/images/bcornerr.gif" width="32" height="32" alt="/" /></td></tr><tr><td align="right" valign="top" colspan="2"> 聽 </td><td class="lm">Last update: 2005-12-15 15:00 聽聽 [from the cache]<br />漏2003-2020 <a href="http://blog.jasper.es/"><span itemprop="editor">Jasper Spaans</span></a>|hosted at <a href="https://www.digitalocean.com/?refcode=9a8e99d24cf9">Digital Ocean</a> and my Meterkast|<a href="http://blog.jasper.es/categories.html#lkml-ref">Read the blog</a></td><td>聽</td></tr></table><script language="javascript" src="/js/styleswitcher.js" type="text/javascript"></script></body></html>