<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:taxo="http://purl.org/rss/1.0/modules/taxonomy/" version="2.0">
  <channel>
    <title>topic I managed to compile the mpss in Software Archive</title>
    <link>https://community.intel.com/t5/Software-Archive/KNC-on-Fedora-23-will-it-work/m-p/1105876#M70452</link>
    <description>&lt;P&gt;I managed to compile the mpss kernel module for kernel 4.2.6 using the attached patch. Lot of the stuff is similar to the patch here &lt;A href="https://github.com/abusse/xeon-phi-overlay/blob/master/sys-kernel/mpss-modules/files/linux-3.18.7.patch" target="_blank"&gt;https://github.com/abusse/xeon-phi-overlay/blob/master/sys-kernel/mpss-modules/files/linux-3.18.7.patch&lt;/A&gt; and look pretty innocent, but there are some kernel 4.2 specifics in my patch I am not exactly sure about:&lt;/P&gt;

&lt;P&gt;&lt;STRONG&gt;1. &lt;/STRONG&gt;rcu_dereference_index_check() in host/vhost/vhost.h:&lt;BR /&gt;
	This has been completely removed in 4.2, so I replaced it with&amp;nbsp;smp_load_acquire(&amp;amp;(dev-&amp;gt;acked_features)) - similar to how it was done here:&lt;BR /&gt;
	&lt;A href="http://lxr.free-electrons.com/source/arch/x86/kernel/cpu/mcheck/mce.c?v=4.1#L56" target="_blank"&gt;http://lxr.free-electrons.com/source/arch/x86/kernel/cpu/mcheck/mce.c?v=4.1#L56&lt;/A&gt; vs. &lt;A href="http://lxr.free-electrons.com/source/arch/x86/kernel/cpu/mcheck/mce.c?v=4.2#L60" target="_blank"&gt;http://lxr.free-electrons.com/source/arch/x86/kernel/cpu/mcheck/mce.c?v=4.2#L60&lt;/A&gt;&lt;/P&gt;

&lt;P&gt;&lt;STRONG&gt;2. &lt;/STRONG&gt;set_mb() in micscif/micscif_select.c:&lt;BR /&gt;
	Replaced with smp_store_mb(pwq-&amp;gt;triggered, 0), as seen in &lt;A href="http://www.spinics.net/lists/linux-tip-commits/msg28183.html" target="_blank"&gt;http://www.spinics.net/lists/linux-tip-commits/msg28183.html&lt;/A&gt;&lt;/P&gt;

&lt;P&gt;I won't have a chance to actually install Xeon Phi in the machine until Sunday, so I can't check whether it works now.&lt;/P&gt;

&lt;P&gt;Do you think the modifications I made are safe? (I'm mostly worried about the rcu_dereference_index_check since I have no idea what was it supposed to do. The rest seems ok)&lt;/P&gt;</description>
    <pubDate>Fri, 11 Dec 2015 12:38:00 GMT</pubDate>
    <dc:creator>Jiří_V_</dc:creator>
    <dc:date>2015-12-11T12:38:00Z</dc:date>
    <item>
      <title>KNC on Fedora 23 - will it work?</title>
      <link>https://community.intel.com/t5/Software-Archive/KNC-on-Fedora-23-will-it-work/m-p/1105875#M70451</link>
      <description>&lt;P&gt;Hello,&lt;/P&gt;

&lt;P&gt;my Xeon Phi 31S1P should arrive today, After I fit an appropriate cooler on it, I want to install it in a workstation running &lt;STRONG&gt;Fedora 23&lt;/STRONG&gt; (a dual E5-2630 v2 on a Supermicro X9DAi board).&lt;/P&gt;

&lt;P&gt;Preparing for the installation, I understant that to run the Phi, I will need to install the Intel MPSS stack which seems to be officially supported only on kernels up to 3.12 (or whatever is in SLES or RHEL - 3.something in any case). Fedora 23 runs on 4.2.6. I gather the MPSS stack consits of some tools packed in RPM packages, and a kernel module which could be possibly recompiled, and that there were people on this forum who successfully installed it on Fedora 20, but I found no one mentioning kernel 4 and up.&lt;/P&gt;

&lt;P&gt;Do you think it will possible to recompile/repack/whatever to make it work under Fedora 23? Is there anything else I might need to do besides repacking the RPMs for Fedora, and recompiling the kernel modules?&lt;/P&gt;

&lt;P&gt;Thanks, Jiri&lt;/P&gt;</description>
      <pubDate>Fri, 11 Dec 2015 07:55:30 GMT</pubDate>
      <guid>https://community.intel.com/t5/Software-Archive/KNC-on-Fedora-23-will-it-work/m-p/1105875#M70451</guid>
      <dc:creator>Jiří_V_</dc:creator>
      <dc:date>2015-12-11T07:55:30Z</dc:date>
    </item>
    <item>
      <title>I managed to compile the mpss</title>
      <link>https://community.intel.com/t5/Software-Archive/KNC-on-Fedora-23-will-it-work/m-p/1105876#M70452</link>
      <description>&lt;P&gt;I managed to compile the mpss kernel module for kernel 4.2.6 using the attached patch. Lot of the stuff is similar to the patch here &lt;A href="https://github.com/abusse/xeon-phi-overlay/blob/master/sys-kernel/mpss-modules/files/linux-3.18.7.patch" target="_blank"&gt;https://github.com/abusse/xeon-phi-overlay/blob/master/sys-kernel/mpss-modules/files/linux-3.18.7.patch&lt;/A&gt; and look pretty innocent, but there are some kernel 4.2 specifics in my patch I am not exactly sure about:&lt;/P&gt;

&lt;P&gt;&lt;STRONG&gt;1. &lt;/STRONG&gt;rcu_dereference_index_check() in host/vhost/vhost.h:&lt;BR /&gt;
	This has been completely removed in 4.2, so I replaced it with&amp;nbsp;smp_load_acquire(&amp;amp;(dev-&amp;gt;acked_features)) - similar to how it was done here:&lt;BR /&gt;
	&lt;A href="http://lxr.free-electrons.com/source/arch/x86/kernel/cpu/mcheck/mce.c?v=4.1#L56" target="_blank"&gt;http://lxr.free-electrons.com/source/arch/x86/kernel/cpu/mcheck/mce.c?v=4.1#L56&lt;/A&gt; vs. &lt;A href="http://lxr.free-electrons.com/source/arch/x86/kernel/cpu/mcheck/mce.c?v=4.2#L60" target="_blank"&gt;http://lxr.free-electrons.com/source/arch/x86/kernel/cpu/mcheck/mce.c?v=4.2#L60&lt;/A&gt;&lt;/P&gt;

&lt;P&gt;&lt;STRONG&gt;2. &lt;/STRONG&gt;set_mb() in micscif/micscif_select.c:&lt;BR /&gt;
	Replaced with smp_store_mb(pwq-&amp;gt;triggered, 0), as seen in &lt;A href="http://www.spinics.net/lists/linux-tip-commits/msg28183.html" target="_blank"&gt;http://www.spinics.net/lists/linux-tip-commits/msg28183.html&lt;/A&gt;&lt;/P&gt;

&lt;P&gt;I won't have a chance to actually install Xeon Phi in the machine until Sunday, so I can't check whether it works now.&lt;/P&gt;

&lt;P&gt;Do you think the modifications I made are safe? (I'm mostly worried about the rcu_dereference_index_check since I have no idea what was it supposed to do. The rest seems ok)&lt;/P&gt;</description>
      <pubDate>Fri, 11 Dec 2015 12:38:00 GMT</pubDate>
      <guid>https://community.intel.com/t5/Software-Archive/KNC-on-Fedora-23-will-it-work/m-p/1105876#M70452</guid>
      <dc:creator>Jiří_V_</dc:creator>
      <dc:date>2015-12-11T12:38:00Z</dc:date>
    </item>
    <item>
      <title>I finally got a chance to</title>
      <link>https://community.intel.com/t5/Software-Archive/KNC-on-Fedora-23-will-it-work/m-p/1105877#M70453</link>
      <description>&lt;P&gt;I finally got a chance to install the Xeon Phi in&amp;nbsp; my machine. The cooling solution (three 40mm Airen fans with a DIY duct, they are a bit loud though) keeps the MIC at 60C on 196W.&lt;/P&gt;

&lt;P&gt;I added &lt;EM&gt;rcu_lockdep_assert&lt;/EM&gt; to my patch. I still don't know what's that part doing. I would be glad if someone could reassure me it won't make the board burn down my office while I'm away for Christmas :-)&lt;/P&gt;

&lt;P&gt;The rest was just following the readme.txt. Now I can log in to the MIC which I suppose means it works.&lt;/P&gt;</description>
      <pubDate>Tue, 15 Dec 2015 09:28:07 GMT</pubDate>
      <guid>https://community.intel.com/t5/Software-Archive/KNC-on-Fedora-23-will-it-work/m-p/1105877#M70453</guid>
      <dc:creator>Jiří_V_</dc:creator>
      <dc:date>2015-12-15T09:28:07Z</dc:date>
    </item>
  </channel>
</rss>

