Date: Tue, 08 Mar 2005 17:55:34 -0700 From: Scott Long <scottl@samsco.org> To: David Sze <dsze@alumni.uwaterloo.ca> Cc: freebsd-stable@freebsd.org Subject: Re: Can FreeBSD be installed on DellPowerEdge2800 ? Message-ID: <422E4986.6080409@samsco.org> In-Reply-To: <20050308235853.GA81150@mail.distrust.net> References: <BA84A761ED41C54893BED691A4A574CF977226@EXCH5.eurowings.com> <4224F19C.2090207@rackspace.com> <6.2.1.2.2.20050301234017.03e93e68@pop3.tellurian.com> <4225CCE4.3080304@rackspace.com> <20050308235853.GA81150@mail.distrust.net>
next in thread | previous in thread | raw e-mail | index | archive | help
David Sze wrote: > On Wed, Mar 02, 2005 at 08:25:40AM -0600, Art Mason wrote: > >>Vinny Abello wrote: >> >>>At 05:50 PM 3/1/2005, Art Mason wrote: >>> >>>>Indeed, just installed on a customer's server last night, and I build >>>>SMP support into the GENERIC kernel this morning. Only issues I >>>>encountered has to do w/ recognizing the full 4GB of RAM, but this >>>>apparently has some compatibility issues w/ the amr driver. >>>>As per Holger's comments about disabling USB in BIOS, this is a good >>>>suggestion, since the DRAC apparently causes some IRQ issues w/ the >>>>PS/2 keyboard controller, or something along those lines. Regardless, >>>>these 2850 boxes are stupid fast, and I'm looking forward to >>>>benchmarking 5.3 on them. >>> >>> >>>FYI, the architecture of the 2850 (assuming it is similar to your 2800) >>>requires that you enable PAE support in your kernel of whatever OS you >>>run to address the full 4GB of RAM. This is due in some part because of >>>the memory mapping that was done for the PCI express bus and/or onboard >>>Perc controller (per Dell tech support). We've run into this with all of >>>our 2850's with various OS's and enabling PAE (per Dell's advice) fixes it. >> >> >>Yeah, I cam across that on some mailing list archives, but saw some >>additional references that this can cause issues with the amr driver. >>Have you experienced any instability issues with the amr driver and PAE >>enabled? If not, then I'll give this a shot on one of our lab boxes for >>testing. Thanks for the suggestion, BTW. > > > > On my 2850 the upper 1GB of RAM cannot be addressed without PAE support > in the kernel. However, amr in 4.11-RELEASE does indeed seem to have > trouble with PAE. The result being that with the option in the kernel, > the root device (ufs:/dev/amrd0s1a) cannot be found. > > I backported scottl's amr PAE fixes from RELENG_5; the diffs applied > cleanly (with offsets). I'm happy to report that my 2850 can now > address its full 4GB and see its amr volumes. > > The diff against 4.11-RELEASE is attached, hopefully someone will be > willing to commit it to RELENG_4. > I don't have the cycles to watch after RELENG_4, so I'll have to take your word that this works. What date is this patch against? I committed some follow-up fixes last week that are required for management apps to work (and yes, LSI is getting ready to support FreeBSD with management apps). Scott
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?422E4986.6080409>