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: Grant Grundler: Re: pci-sysfs resource mmap broken</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 Grant Grundler" href="/groupie.php?aid=1424" /><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/2005/4/26"> [26]</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/2005/4/26/212" onclick="this.href='/lkml/headers'+'/2005/4/26/212';">[headers]</a>聽 <a href="/lkml/bounce/2005/4/26/212">[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/2005/4/26/25">First message in thread</a></li><li><a href="/lkml/2005/4/26/25">Benjamin Herrenschmidt</a><ul><li><a href="/lkml/2005/4/26/35">Benjamin Herrenschmidt</a><ul><li><a href="/lkml/2005/4/26/41">Greg KH</a></li><li><a href="/lkml/2005/4/26/86">Russell King</a></li></ul></li><li class="origin"><a href="/lkml/2005/4/26/306">Grant Grundler</a><ul><li><a href="/lkml/2005/4/26/306">Benjamin Herrenschmidt</a><ul><li><a href="/lkml/2005/4/26/345">Grant Grundler</a><ul><li><a href="/lkml/2005/4/27/3">Benjamin Herrenschmidt</a></li></ul></li><li><a href="/lkml/2005/4/27/8">Benjamin Herrenschmidt</a></li><li><a href="/lkml/2005/4/27/263">Benjamin Herrenschmidt</a><ul><li><a href="/lkml/2005/4/28/12">Grant Grundler</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">Date</td><td class="rp" itemprop="datePublished">Tue, 26 Apr 2005 10:30:42 -0600</td></tr><tr><td class="lp">From</td><td class="rp" itemprop="author">Grant Grundler <></td></tr><tr><td class="lp">Subject</td><td class="rp" itemprop="name">Re: pci-sysfs resource mmap broken</td></tr></table></td><td></td></tr></table><pre itemprop="articleBody">On Tue, Apr 26, 2005 at 03:33:29PM +1000, Benjamin Herrenschmidt wrote:<br />...<br />> The problem is that can only work ... on architectures where the<br />> resources contain the same thing as the BAR values. On ppc, where this<br />> is not the case, it will not work. On ppc, resources are "fixed up" in<br />> various ways (for example, PReP adds a fixed offset to all memory<br />> resources to match the HW translation since PCI isn't 1:1 on those, and<br />> all PPCs with more than one domain play tricks with IO resources).<br /><br />IIRC, alpha, sparc, and parisc also are broken then.<br /><br />> In a similar vein, the "resource" is exposing directly to userland the<br />> content of "struct resource". This doesn't mean anything. The kernel is<br />> internally playing all sort of offset tricks on these values, so they<br />> can't be used for anything useful, either via /dev/mem, or for io port<br />> accesses, or whatever.<br /><br />There are two "views" of a PCI resource and the names I've used in<br />the past are "IO View" and "CPU View". The raw BAR value is the "IO View"<br />since that's what that devices on that PCI bus need to use for Peer-to-Peer<br />reads/writes. IIRC, sym2 driver directly reads the BAR for telling the scripts<br />engine where onboard RAM lives. The "CPU View" is what drivers/user space<br />needs to use when accessing the device. This is what we should be<br />exporting to user space and I'm pretty sure that's what X.org/XF86<br />should be using too. Bjorn, have I got that right?<br /><br />> Shouldn't we expose the BAR values & size rather here ? That is,<br />> reconsitutes non-offset'd resources, possibly with arch help, or just<br />> reading BAR to get base, and apply resource size & flags ?<br />> <br />> Unless you are on x86 of course ...<br />> <br />> There is some serious brokenness in there, it needs to be fixed if we<br />> want things like X.org to be ever properly adapted (and we'll have to<br />> deal with existing broken kernels, gack).<br /><br />ISTR Bjorn was looking at this and the VGA routing issues.<br /><br />thanks,<br />grant<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 18:38 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>