Date: Fri, 30 Sep 2005 10:14:02 -0700 From: David Kirchner <dpk@dpk.net> To: "Tamouh H." <hakmi@rogers.com> Cc: freebsd-questions@freebsd.org Subject: Re: Weird Real memory reported with 5.4 PAE support Message-ID: <35c231bf0509301014k3d54fa24u4effc3273e0a1262@mail.gmail.com> In-Reply-To: <20050930153855.204C943D49@mx1.FreeBSD.org> References: <20050930153855.204C943D49@mx1.FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On 9/30/05, Tamouh H. <hakmi@rogers.com> wrote: > > Hello, > > We have a Supermicro that has Winston 4GB of ECC Registered DDR333 RAM. D= ual > Xeon Processors. > > With PAE enabled, it seems the system reports 6GB of realmemory instead o= f > 4GB, as below: We've seen similar stuff with Supermicro servers. PCI/PCI-X/PCI-E (or something like that) uses memory addresses in the upper end of the 4GB barrier. The BIOS remaps any real memory above that. The more PCI slots you have the more is reserved, and the higher the BIOS has to remap it. (Someone can correct me if I'm wrong here, I may be oversimplifying). The panics you are seeing are a known bug in FreeBSD 5.4-RELEASE. The bug is fixed in 5-STABLE: http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/i386/i386/pmap.c.diff?r1=3D1.= 494.2.9&r2=3D1.494.2.10&f=3Dh For an unknown reason, the patch has not been commited to the 5.4-RELEASE-p* branches, or 5.4-STABLE. If you don't want to move all the way to 5-STABLE you can just make the above changes manually (it's just 2 lines) and rebuild, and it'll work fine.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?35c231bf0509301014k3d54fa24u4effc3273e0a1262>