On 06/29/09 11:13, John Wythe wrote: > Hi Bill, > > Is this a limitation of the processor power management in the kernel? > The processor itself supports up to C4E according to Intel: > http://processorfinder.intel.com/details.aspx?sSpec=SLAY7
This is a limitation of the Time Stamp Counter in the CPU and Solaris's reliance on a very low latency, high precision timer. > > I can't say I know what the TSC is, but are there any plans to widen > support for these deep C-states? There are plans. http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6700898 Thank you, Bill > Thanks, > John > > > On Mon, Jun 29, 2009 at 2:10 PM, Bill Holler <Bill.Holler at sun.com > <mailto:Bill.Holler at sun.com>> wrote: > > C-states are not supported on hardware which stop the TSC. > C-state are only supported on Nehalem/CoreI7. > > Regards, > Bill > > > > On 06/29/09 10:13, bitspace at gmail.com <mailto:bitspace at gmail.com> > wrote: > > Hello, > > My apologies if this is the incorrect list. > > I am having trouble getting deep C-States beyond C1 working on > OpenSolaris nv_111b (2009.06 release). I am running an Intel > Core 2 dual-core E5200 with a Supermicro X7SBL-LN2 motherboard > that supports C1E and deep C-states in the BIOS. > > This shows up in var/adm/messages: > > Jun 29 12:16:05 databank unix: [ID 950921 kern.info > <http://kern.info> > <http://kern.info>] cpu0: x86 (chipid 0x0 GenuineIntel 1067A > > family 6 model 23 step 10 clock 2500 MHz) > Jun 29 12:16:05 databank unix: [ID 950921 kern.info > <http://kern.info> > <http://kern.info>] cpu0: Pentium(r) Dual-Core CPU > E5200 @ > > 2.50GHz > Jun 29 12:16:05 databank unix: [ID 950921 kern.info > <http://kern.info> > <http://kern.info>] cpu1: x86 (chipid 0x0 GenuineIntel 1067A > > family 6 model 23 step 10 clock 2500 MHz) > Jun 29 12:16:05 databank unix: [ID 950921 kern.info > <http://kern.info> > <http://kern.info>] cpu1: Pentium(r) Dual-Core CPU > E5200 @ > > 2.50GHz > Jun 29 12:16:05 databank acpica: [ID 423971 kern.notice] ACPI: > SSDT @ 0xbfe740cb/0x0234 (v 1 PmRef Cpu1Ist 0x00003000 INTL > 0x20050228) > Jun 29 12:16:05 databank acpica: [ID 395089 kern.notice] ACPI: > SSDT @ 0xbfe73a85/0x0085 (v 1 PmRef Cpu1Cst 0x00003000 INTL > 0x20050228) > Jun 29 12:16:05 databank unix: [ID 395387 kern.info > <http://kern.info> > <http://kern.info>] NOTICE: cpu_acpi: _CST invalid count 1 < 2 > > Jun 29 12:16:05 databank unix: [ID 518192 kern.warning] > WARNING: > cpu_acpi: error parsing _CST for CPU 1 > Jun 29 12:16:05 databank unix: [ID 726263 kern.info > <http://kern.info> > <http://kern.info>] NOTICE: cpu_idle_init: Failed to cache > ACPI > > C-state data > Jun 29 12:16:05 databank unix: [ID 763091 kern.warning] > WARNING: > cpupm_init: processor 1: unable to initialize C-state support > > > > The same occurs for CPU 0 but earlier in the log. > > There also appears to be something strange going on with the > p-states as well: > > $ kstat -m cpu_info -s current_clock_Hz > module: cpu_info instance: 0 > name: cpu_info0 class: misc > current_clock_Hz 2500000000 > > module: cpu_info instance: 1 > name: cpu_info1 class: misc > current_clock_Hz 2500000000 > > > > As you can see, the current_clock_Hz is reported by kstat as > 2.5GHz, but in powertop it is reported differently: > > Cn Avg residency P-states > (frequencies) > C0 (cpu running) (3.2%) 1200 Mhz > 100.0% > C1 2.4ms (96.8%) 1600 Mhz > 0.0% > 2000 Mhz > 0.0% > 2400 Mhz > 0.0% > 2500 Mhz > 0.0% > > > > Powertop says the processor is running at 1.2GHz not 2.5GHz, > and of course no deep c-states. I am unsure as why they are > reporting differently. > > This is my current power.conf: > > autopm enable > autoS3 default > cpu-threshold 1s > # Auto-Shutdown Idle(min) Start/Finish(hh:mm) > Behavior > autoshutdown 30 9:00 9:00 > noshutdown > cpupm enable > cpu_deep_idle enable > > device-thresholds /pci at 0,0/pci15d9,d880 at 1f,2/disk at > 0,0 > 45m > device-thresholds /pci at 0,0/pci15d9,d880 at 1f,2/disk at > 1,0 > 45m > device-thresholds /pci at 0,0/pci15d9,d880 at 1f,2/disk at > 2,0 > 45m > > > > I would appreciate any insights anyone may have. I looked to > see if this has been a reported bug and if there is a fix but > have not found anything. I am new enough to OpenSolaris that I > still have not found out where the changelogs are for the > packages in the dev repository either, so I am unsure if there > are patches to the C-state initialization for other bugs that > may fix this. > > Cheers, > John > > > > > > > > > > ------------------------------------------------------------------------ > > _______________________________________________ > tesla-dev mailing list > tesla-dev at opensolaris.org <mailto:tesla-dev at opensolaris.org> > http://mail.opensolaris.org/mailman/listinfo/tesla-dev > > > >
