FreeBSD source tree https://cgit-dev.freebsd.org/src/atom?h=release%2F12.3.0 2010-02-25T14:13:39Z 2010-02-25T14:13:39Z Attilio Rao attilio@FreeBSD.org 2010-02-25T14:13:39Z urn:sha1:32580301441a6c4f89a20c20f53e45e5557a8b64 shared and generalized between our current amd64, i386 and pc98. This is just an initial step that should lead to a more complete effort. For the moment, a very simple porting of cpufreq modules, BIOS calls and the whole MD specific ISA bus part is added to the sub-tree but ideally a lot of code might be added and more shared support should grow. Sponsored by: Sandvine Incorporated Reviewed by: emaste, kib, jhb, imp Discussed on: arch MFC: 3 weeks 2010-01-15T16:04:30Z Attilio Rao attilio@FreeBSD.org 2010-01-15T16:04:30Z urn:sha1:a26cb6d54748fc3366394ddfe3805bb96440a498 LAPIC may lead to aliasing for softclock and profclock because frequencies are sized in order to fit mainly hardclock. atrtc used to take care of the softclock and profclock and it does still do, if the LAPIC can't handle the clocks properly. Revert the change when the LAPIC started taking charge of all three of them and let atrtc handle softclock and profclock if not explicitly requested. Such request can be made setting != 0 the new tunable machdep.lapic_allclocks or if the new device ATPIC is not present within the i386 kernel config (atrtc is linked to atpic presence). Diagnosed by: Sandvine Incorporated Reviewed by: jhb, emaste Sponsored by: Sandvine Incorporated MFC: 3 weeks 2009-06-23T23:16:37Z Alexander Motin mav@FreeBSD.org 2009-06-23T23:16:37Z urn:sha1:9f23a6caa4f90ac3347968748f071a47b2e6820d 2009-06-23T22:08:25Z Alexander Motin mav@FreeBSD.org 2009-06-23T22:08:25Z urn:sha1:9e9be269064ed95f90402eaa7c4e93a0d4225197 2009-06-23T21:45:33Z Alexander Motin mav@FreeBSD.org 2009-06-23T21:45:33Z urn:sha1:96c5d068d8e78a4ab287f773cb3f9b61f0815851 While general idea of patch was good, it was not working properly due the way it was implemented. When we are using same timer interrupt for several of hard/prof/stat purposes we should not send several IPIs same time to other CPUs. Sending several IPIs same time leads to terrible accounting/profiling results due to strong synchronization effect, when the second interrupt handler accounts processing of the first one. Interlink timer events in a such way, that no more then one IPI is sent for any original timer interrupt. 2009-06-09T09:47:02Z Bjoern A. Zeeb bz@FreeBSD.org 2009-06-09T09:47:02Z urn:sha1:78c7c6adbc03e8b6d2998ea699ff9c1940407399 2009-06-09T07:26:52Z Ariff Abdullah ariff@FreeBSD.org 2009-06-09T07:26:52Z urn:sha1:b65cb1db3c90960f2509826c5b1a4a14c7f13090 - Interpolate stat/prof clock using clkintr() in a similar fashion to local APIC timer, since statclock usually run slower. - Liberate hardclockintr() from taking the burden of handling both stat and prof clock interrupt. Instead, send IPIs within clkintr() to handle those. 2009-05-03T17:47:21Z Alexander Motin mav@FreeBSD.org 2009-05-03T17:47:21Z urn:sha1:1703f2b4249486c90020bba27d17fb815a2640fc and hide it inside of atrtc driver. Add new tunable hint.atrtc.0.clock controlling it. Setting it to 0 disables using RTC clock as stat-/ profclock sources. Teach i386 and amd64 SMP platforms to emulate stat-/profclocks using i8254 hardclock, when LAPIC and RTC clocks are disabled. This allows to reduce global interrupt rate of idle system down to about 100 interrupts per core, permitting C3 and deeper C-states provide maximum CPU power efficiency. 2009-05-02T12:20:43Z Alexander Motin mav@FreeBSD.org 2009-05-02T12:20:43Z urn:sha1:6a3a164d6e1a5b3c59daefd59ac885cd467c3caf system. Redistribute hard-/stat-/profclock events to other CPUs using IPIs. 2009-05-01T21:43:04Z Alexander Motin mav@FreeBSD.org 2009-05-01T21:43:04Z urn:sha1:58a2bb4996aec6aed823b4a710ab56e6a5f830f4