Date: Mon, 25 Aug 2008 18:28:01 -0400 From: John Baldwin <jhb@freebsd.org> To: Dimitry Andric <dimitry@andric.com> Cc: Renato Botelho <rbgarga@gmail.com>, cvs-src@freebsd.org, src-committers@freebsd.org, cvs-all@freebsd.org Subject: Re: cvs commit: src/sys/i386/cpufreq est.c Message-ID: <200808251828.02139.jhb@freebsd.org> In-Reply-To: <48B31456.5030009@andric.com> References: <200808231253.m7NCrkLp093604@repoman.freebsd.org> <747dc8f30808251227v286ab480mea5b0d7dff0311fa@mail.gmail.com> <48B31456.5030009@andric.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Monday 25 August 2008 04:21:42 pm Dimitry Andric wrote: > On 2008-08-25 21:27, Renato Botelho wrote: > > cpu0: <ACPI CPU> on acpi0 > > est0: <Enhanced SpeedStep Frequency Control> on cpu0 > > est0: Guessed bus clock (high) of 200 MHz > > > > Fatal trap 18: integer divide fault while in kernel mode > > cpuid = 0; acpi id = 00 > > ... > > panic: integer divide fault > > Unfortunately, there are CPU models around that have MSR_PERF_STATUS > bits that are inconsistent, e.g: > > - The low and high multipliers (bits 31:24 and 15:8) are equal, so if > you subtract them and then divide... boom :) > - Either the low or high multipliers (or both) are zero, which is also > not good, at least not with the current code. This latter is what is probably the problem. I think I will make this feature just be off by default and people can enable it with a tunable if it helps on their system. -- John Baldwin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200808251828.02139.jhb>