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: Benjamin Herrenschmidt: Re: [PATCH] PCI: Add pci shutdown ability</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 Benjamin Herrenschmidt" href="/groupie.php?aid=447" /><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/2005/4"> [Apr]</a> 聽 <a class="nb" href="/lkml/2005/4/25"> [25]</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/2005/4/25/297" onclick="this.href='/lkml/headers'+'/2005/4/25/297';">[headers]</a>聽 <a href="/lkml/bounce/2005/4/25/297">[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/2005/4/25/154">First message in thread</a></li><li><a href="/lkml/2005/4/25/181">Alan Stern</a><ul><li><a href="/lkml/2005/4/25/199">Alexander Nyberg</a><ul><li><a href="/lkml/2005/4/25/207">Alan Stern</a><ul><li><a href="/lkml/2005/4/26/195">Grant Grundler</a><ul><li><a href="/lkml/2005/4/26/200">Alan Stern</a></li></ul></li><li><a href="/lkml/2005/5/11/7">Vivek Goyal</a><ul><li><a href="/lkml/2005/5/11/90">Alan Stern</a></li></ul></li></ul></li></ul></li><li><a href="/lkml/2005/4/25/225">Andrew Morton</a><ul><li><a href="/lkml/2005/4/25/228">Dave Jones</a><ul><li><a href="/lkml/2005/4/25/253">Adam Belay</a><ul><li><a href="/lkml/2005/4/26/14">Benjamin Herrenschmidt</a></li><li><a href="/lkml/2005/4/26/92">Pavel Machek</a></li></ul></li><li><a href="/lkml/2005/4/25/300">Benjamin Herrenschmidt</a><ul><li><a href="/lkml/2005/4/26/177">Alan Stern</a></li></ul></li><li><a href="/lkml/2005/4/26/91">Pavel Machek</a><ul><li><a href="/lkml/2005/4/26/243">Dave Jones</a></li></ul></li></ul></li><li class="origin"><a href="">Benjamin Herrenschmidt</a></li><li><a href="/lkml/2005/4/26/175">Alan Stern</a><ul><li><a href="/lkml/2005/4/26/199">Alexander Nyberg</a></li></ul></li></ul></li><li><a href="/lkml/2005/4/26/187">Grant Grundler</a><ul><li><a href="/lkml/2005/4/26/204">"Richard B. Johnson"</a><ul><li><a href="/lkml/2005/4/26/207">Grant Grundler</a><ul><li><a href="/lkml/2005/4/26/224">Alan Stern</a></li></ul></li></ul></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">Subject</td><td class="rp" itemprop="name">Re: [PATCH] PCI: Add pci shutdown ability</td></tr><tr><td class="lp">From</td><td class="rp" itemprop="author">Benjamin Herrenschmidt <></td></tr><tr><td class="lp">Date</td><td class="rp" itemprop="datePublished">Tue, 26 Apr 2005 13:45:43 +1000</td></tr></table></td><td></td></tr></table><pre itemprop="articleBody">On Mon, 2005-04-25 at 14:58 -0700, Andrew Morton wrote:<br />> Alan Stern <stern@rowland.harvard.edu> wrote:<br />> ><br />> > On Mon, 25 Apr 2005, Alexander Nyberg wrote:<br />> > <br />> > > Not sure what you mean by "make kexec work nicer" but if it is because<br />> > > some devices don't work after a kexec I have some objections.<br />> > <br />> > That was indeed the reason, at least in my case. The newly-rebooted<br />> > kernel doesn't work too well when there are active devices, with no driver<br />> > loaded, doing DMA and issuing IRQs because they were never shut down.<br />> <br />> I have vague memories of this being discussed at some length last year. <br />> Nothing comprehensive came of it, except that perhaps the kdump code should<br />> spin with irqs off for a couple of seconds so the DMA and IRQs stop.<br /><br />That i bogus, USB will never stop DMA unless told to do so for example.<br /><br />> (Ongoing DMA is not a problem actually, because the kdump kernel won't be<br />> using that memory anyway)<br /><br />Ok, good. So kdump don't need to call PMSG_FREEZE, normal kexec still<br />does though.<br /><br />> > > What about the kexec-on-panic?<br />> > > <br />> > > In the end at least every storage device should work after a<br />> > > kexec-on-panic or else there might be cases where we cannot get dumps of<br />> > > what happened. My guess is that having access to the network might come<br />> > > in handy after a kexec-on-panic as well.<br />> > > <br />> > > So if this patch is because some devices don't work across kexec I don't<br />> > > think this is a good idea because the same devices won't work after a<br />> > > kexec-on-panic.<br />> > <br />> > Do I understand your argument correctly? You seem to be saying that <br />> > because this new facility sometimes won't work (the kexec-on-panic case) <br />> > it shouldn't be added at all. What about all the other times when it will <br />> > work?<br />> <br />> For kdump we really don't want to be doing fancy driver shutdown inside a<br />> crashed kernel. It would be better to just jump to the new kernel and<br />> to reset the hardware from there. DMA doesn't matter, and maybe IRQs can<br />> be handled with a sustained local_irq_disable() (hard).<br /><br />Yup.<br /><br />> But for the normal kexec path, yes, graceful device shutdown is desirable.<br />> <br />> So the requirements for the two different kexec scenarios are quite<br />> different and it seems unlikely that any single approach to device shutdown<br />> will suit both situations.<br />> <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 />Benjamin Herrenschmidt <benh@kernel.crashing.org><br /><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: 2005-04-26 05:52 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>