Date: Tue, 15 Jan 2013 18:19:25 -0800 From: matt <sendtomatt@gmail.com> To: Michael Copeland <michael@kryptos-security.com> Cc: FreeBSD PowerPC ML <freebsd-ppc@freebsd.org> Subject: Re: FreeBSD 9.1 freezes randomly on G5 Message-ID: <50F60E2D.4060407@gmail.com> In-Reply-To: <CAKLtBChWh9VKCRciNCzWaczcoEa0k5nwypwBhQcx65t0Za4kLA@mail.gmail.com> References: <50F4BADB.5@swissmail.org> <CAHSQbTBiy%2BxxYWu2neKNQgUvFUz3tpQDB_SaZm1F2GgrJx8ZkQ@mail.gmail.com> <50F4DB79.4050900@swissmail.org> <50F5F6F4.9070803@swissmail.org> <50F60564.8060307@gmail.com> <CAKLtBChWh9VKCRciNCzWaczcoEa0k5nwypwBhQcx65t0Za4kLA@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 01/15/2013 17:52, Michael Copeland wrote: > Those of you that saw my posts from a few months ago regarding the > same machine as the one in question, I had the same issues he is > having. The fans would go full blast, the system would hang without > any form of console message. It all stopped when I ran the pmu/smu/fan > script that some people in this mailing list have used. So I'm > thinking it is definitely an issue with the smu driver or the smu > itself on this machine. That also makes sense, but I assumed Nathan's comment meant that it's probably not always a problem on this model. I know the only air-cooled G5 I ever repaired under warranty came in with these exact symptoms under OS X. That is just a single sample, of course. In my mind there are 4 major systems that can produce these symptoms in an OS agnostic manner: 1. Power supply (Apple used AcBel...they had major issues with the G4s for example) 2. CPU (IBM was reducing die reliability and increasing clock speed, on an expensive separate production run...) 3. RAM 4. SMU...it can fail to operate fans properly (I can't remember if it also controls voltages) which produce symptoms in any of the above The way to rule out the above is to see the machine working fine under Tiger (10.4). 10.5 had some of its own issues, it's not the gold standard of PPC reliability usually. If the problem can only be reproduced on FreeBSD, it's very likely smu related (or at least fixable there by increasing cooling). You have to understand Apple was reeling after the "Windtunnel" G4 noise problems...they wanted to have a powerful but quiet machine, so fan control became much more complex. The SMU is at the heart of that matter, and I'm sure most of the machine's thermal design assumes it is working properly and in full control of the fans. Matt
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?50F60E2D.4060407>