From owner-svn-src-all@freebsd.org Wed Jan 22 23:28:44 2020 Return-Path: Delivered-To: svn-src-all@mailman.nyi.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2610:1c1:1:606c::19:1]) by mailman.nyi.freebsd.org (Postfix) with ESMTP id 8A36222202D; Wed, 22 Jan 2020 23:28:44 +0000 (UTC) (envelope-from cem@FreeBSD.org) Received: from mxrelay.nyi.freebsd.org (mxrelay.nyi.freebsd.org [IPv6:2610:1c1:1:606c::19:3]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) server-signature RSA-PSS (4096 bits) client-signature RSA-PSS (4096 bits) client-digest SHA256) (Client CN "mxrelay.nyi.freebsd.org", Issuer "Let's Encrypt Authority X3" (verified OK)) by mx1.freebsd.org (Postfix) with ESMTPS id 4831n83vFzz3C8D; Wed, 22 Jan 2020 23:28:44 +0000 (UTC) (envelope-from cem@FreeBSD.org) Received: from repo.freebsd.org (repo.freebsd.org [IPv6:2610:1c1:1:6068::e6a:0]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (Client did not present a certificate) by mxrelay.nyi.freebsd.org (Postfix) with ESMTPS id 80DBF5599; Wed, 22 Jan 2020 23:28:44 +0000 (UTC) (envelope-from cem@FreeBSD.org) Received: from repo.freebsd.org ([127.0.1.37]) by repo.freebsd.org (8.15.2/8.15.2) with ESMTP id 00MNSisS058367; Wed, 22 Jan 2020 23:28:44 GMT (envelope-from cem@FreeBSD.org) Received: (from cem@localhost) by repo.freebsd.org (8.15.2/8.15.2/Submit) id 00MNSh8a058358; Wed, 22 Jan 2020 23:28:43 GMT (envelope-from cem@FreeBSD.org) Message-Id: <202001222328.00MNSh8a058358@repo.freebsd.org> X-Authentication-Warning: repo.freebsd.org: cem set sender to cem@FreeBSD.org using -f From: Conrad Meyer Date: Wed, 22 Jan 2020 23:28:43 +0000 (UTC) To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: svn commit: r357002 - in head: share/man/man4 sys/conf sys/kern sys/modules/cpufreq sys/sys sys/x86/cpufreq X-SVN-Group: head X-SVN-Commit-Author: cem X-SVN-Commit-Paths: in head: share/man/man4 sys/conf sys/kern sys/modules/cpufreq sys/sys sys/x86/cpufreq X-SVN-Commit-Revision: 357002 X-SVN-Commit-Repository: base MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit X-BeenThere: svn-src-all@freebsd.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: "SVN commit messages for the entire src tree \(except for " user" and " projects" \)" List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 22 Jan 2020 23:28:44 -0000 Author: cem Date: Wed Jan 22 23:28:42 2020 New Revision: 357002 URL: https://svnweb.freebsd.org/changeset/base/357002 Log: cpufreq(4): Add support for Intel Speed Shift Intel Speed Shift is Intel's technology to control frequency in hardware, with hints from software. Let's get a working version of this in the tree and we can refine it from here. Submitted by: bwidawsk, scottph Reviewed by: bcr (manpages), myself Discussed with: jhb, kib (earlier versions) With feedback from: Greg V, gallatin, freebsdnewbie AT freenet.de Relnotes: yes Differential Revision: https://reviews.freebsd.org/D18028 Added: head/share/man/man4/hwpstate_intel.4 (contents, props changed) head/sys/x86/cpufreq/hwpstate_amd.c - copied unchanged from r357001, head/sys/x86/cpufreq/hwpstate.c head/sys/x86/cpufreq/hwpstate_intel.c (contents, props changed) head/sys/x86/cpufreq/hwpstate_intel_internal.h (contents, props changed) Deleted: head/sys/x86/cpufreq/hwpstate.c Modified: head/share/man/man4/cpufreq.4 head/sys/conf/files.x86 head/sys/kern/kern_cpu.c head/sys/modules/cpufreq/Makefile head/sys/sys/cpu.h head/sys/x86/cpufreq/est.c Modified: head/share/man/man4/cpufreq.4 ============================================================================== --- head/share/man/man4/cpufreq.4 Wed Jan 22 22:51:55 2020 (r357001) +++ head/share/man/man4/cpufreq.4 Wed Jan 22 23:28:42 2020 (r357002) @@ -24,7 +24,7 @@ .\" .\" $FreeBSD$ .\" -.Dd March 3, 2006 +.Dd January 22, 2020 .Dt CPUFREQ 4 .Os .Sh NAME @@ -85,6 +85,10 @@ sysctl entry. .Bl -tag -width indent .It Va dev.cpu.%d.freq Current active CPU frequency in MHz. +.It Va dev.cpu.%d.freq_driver +The specific +.Nm +driver used by this cpu. .It Va dev.cpu.%d.freq_levels Currently available levels for the CPU (frequency/power usage). Values are in units of MHz and milliwatts. Added: head/share/man/man4/hwpstate_intel.4 ============================================================================== --- /dev/null 00:00:00 1970 (empty, because file is newly added) +++ head/share/man/man4/hwpstate_intel.4 Wed Jan 22 23:28:42 2020 (r357002) @@ -0,0 +1,89 @@ +.\" +.\" Copyright (c) 2019 Intel Corporation +.\" +.\" Redistribution and use in source and binary forms, with or without +.\" modification, are permitted provided that the following conditions +.\" are met: +.\" 1. Redistributions of source code must retain the above copyright +.\" notice, this list of conditions and the following disclaimer. +.\" 2. Redistributions in binary form must reproduce the above copyright +.\" notice, this list of conditions and the following disclaimer in the +.\" documentation and/or other materials provided with the distribution. +.\" +.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND +.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE +.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE +.\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE +.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS +.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) +.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT +.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY +.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF +.\" SUCH DAMAGE. +.\" +.\" $FreeBSD$ +.\" +.Dd January 22, 2020 +.Dt HWPSTATE_INTEL 4 +.Os +.Sh NAME +.Nm hwpstate_intel +.Nd Intel Speed Shift Technology driver +.Sh SYNOPSIS +To compile this driver into your kernel +place the following line in your kernel +configuration file: +.Bd -ragged -offset indent +.Cd "device cpufreq" +.Ed +.Sh DESCRIPTION +The +.Nm +driver provides support for hardware-controlled performance states on Intel +platforms, also known as Intel Speed Shift Technology. +.Sh LOADER TUNABLES +.Bl -tag -width indent +.It Va hint.hwpstate_intel.0.disabled +Can be used to disable +.Nm , +allowing other compatible drivers to manage performance states, like +.Xr est 4 . +.Pq default 0 +.El +.Sh SYSCTL VARIABLES +The following +.Xr sysctl 8 +values are available +.Bl -tag -width indent +.It Va dev.hwpstate_intel.%d.\%desc +Describes the attached driver +.It dev.hwpstate_intel.0.%desc: Intel Speed Shift +.It Va dev.hwpstate_intel.%d.\%driver +Driver in use, always hwpstate_intel. +.It dev.hwpstate_intel.0.%driver: hwpstate_intel +.It Va dev.hwpstate_intel.%d.\%parent +.It dev.hwpstate_intel.0.%parent: cpu0 +The cpu that is exposing these frequencies. +For example +.Va cpu0 . +.It Va dev.hwpstate_intel.%d.epp +Energy/Performance Preference. +Valid values range from 0 to 100. +Setting this field conveys a hint to the hardware regarding a preference towards +performance (at value 0), energy efficiency (at value 100), or somewhere in +between. +.It dev.hwpstate_intel.0.epp: 0 +.El +.Sh COMPATIBILITY +.Nm +is only found on supported Intel CPUs. +.Sh SEE ALSO +.Xr cpufreq 4 +.Rs +.%T "Intel 64 and IA-32 Architectures Software Developer Manuals" +.%U "http://www.intel.com/content/www/us/en/processors/architectures-software-developer-manuals.html" +.Re +.Sh AUTHORS +This manual page was written by +.An D Scott Phillips Aq Mt scottph@FreeBSD.org . Modified: head/sys/conf/files.x86 ============================================================================== --- head/sys/conf/files.x86 Wed Jan 22 22:51:55 2020 (r357001) +++ head/sys/conf/files.x86 Wed Jan 22 23:28:42 2020 (r357002) @@ -290,7 +290,8 @@ x86/acpica/srat.c optional acpi x86/bios/smbios.c optional smbios x86/bios/vpd.c optional vpd x86/cpufreq/est.c optional cpufreq -x86/cpufreq/hwpstate.c optional cpufreq +x86/cpufreq/hwpstate_amd.c optional cpufreq +x86/cpufreq/hwpstate_intel.c optional cpufreq x86/cpufreq/p4tcc.c optional cpufreq x86/cpufreq/powernow.c optional cpufreq x86/iommu/busdma_dmar.c optional acpi acpi_dmar pci Modified: head/sys/kern/kern_cpu.c ============================================================================== --- head/sys/kern/kern_cpu.c Wed Jan 22 22:51:55 2020 (r357001) +++ head/sys/kern/kern_cpu.c Wed Jan 22 23:28:42 2020 (r357002) @@ -76,6 +76,7 @@ struct cpufreq_softc { int all_count; int max_mhz; device_t dev; + device_t cf_drv_dev; struct sysctl_ctx_list sysctl_ctx; struct task startup_task; struct cf_level *levels_buf; @@ -142,6 +143,11 @@ SYSCTL_INT(_debug_cpufreq, OID_AUTO, lowest, CTLFLAG_R SYSCTL_INT(_debug_cpufreq, OID_AUTO, verbose, CTLFLAG_RWTUN, &cf_verbose, 1, "Print verbose debugging messages"); +/* + * This is called as the result of a hardware specific frequency control driver + * calling cpufreq_register. It provides a general interface for system wide + * frequency controls and operates on a per cpu basis. + */ static int cpufreq_attach(device_t dev) { @@ -149,7 +155,6 @@ cpufreq_attach(device_t dev) struct pcpu *pc; device_t parent; uint64_t rate; - int numdevs; CF_DEBUG("initializing %s\n", device_get_nameunit(dev)); sc = device_get_softc(dev); @@ -164,6 +169,7 @@ cpufreq_attach(device_t dev) sc->max_mhz = cpu_get_nominal_mhz(dev); /* If that fails, try to measure the current rate */ if (sc->max_mhz <= 0) { + CF_DEBUG("Unable to obtain nominal frequency.\n"); pc = cpu_get_pcpu(dev); if (cpu_est_clockrate(pc->pc_cpuid, &rate) == 0) sc->max_mhz = rate / 1000000; @@ -171,15 +177,6 @@ cpufreq_attach(device_t dev) sc->max_mhz = CPUFREQ_VAL_UNKNOWN; } - /* - * Only initialize one set of sysctls for all CPUs. In the future, - * if multiple CPUs can have different settings, we can move these - * sysctls to be under every CPU instead of just the first one. - */ - numdevs = devclass_get_count(cpufreq_dc); - if (numdevs > 1) - return (0); - CF_DEBUG("initializing one-time data for %s\n", device_get_nameunit(dev)); sc->levels_buf = malloc(CF_MAX_LEVELS * sizeof(*sc->levels_buf), @@ -216,7 +213,6 @@ cpufreq_detach(device_t dev) { struct cpufreq_softc *sc; struct cf_saved_freq *saved_freq; - int numdevs; CF_DEBUG("shutdown %s\n", device_get_nameunit(dev)); sc = device_get_softc(dev); @@ -227,12 +223,7 @@ cpufreq_detach(device_t dev) free(saved_freq, M_TEMP); } - /* Only clean up these resources when the last device is detaching. */ - numdevs = devclass_get_count(cpufreq_dc); - if (numdevs == 1) { - CF_DEBUG("final shutdown for %s\n", device_get_nameunit(dev)); - free(sc->levels_buf, M_DEVBUF); - } + free(sc->levels_buf, M_DEVBUF); return (0); } @@ -422,25 +413,74 @@ out: } static int +cpufreq_get_frequency(device_t dev) +{ + struct cf_setting set; + + if (CPUFREQ_DRV_GET(dev, &set) != 0) + return (-1); + + return (set.freq); +} + +/* Returns the index into *levels with the match */ +static int +cpufreq_get_level(device_t dev, struct cf_level *levels, int count) +{ + int i, freq; + + if ((freq = cpufreq_get_frequency(dev)) < 0) + return (-1); + for (i = 0; i < count; i++) + if (freq == levels[i].total_set.freq) + return (i); + + return (-1); +} + +/* + * Used by the cpufreq core, this function will populate *level with the current + * frequency as either determined by a cached value sc->curr_level, or in the + * case the lower level driver has set the CPUFREQ_FLAG_UNCACHED flag, it will + * obtain the frequency from the driver itself. + */ +static int cf_get_method(device_t dev, struct cf_level *level) { struct cpufreq_softc *sc; struct cf_level *levels; - struct cf_setting *curr_set, set; + struct cf_setting *curr_set; struct pcpu *pc; - device_t *devs; - int bdiff, count, diff, error, i, n, numdevs; + int bdiff, count, diff, error, i, type; uint64_t rate; sc = device_get_softc(dev); error = 0; levels = NULL; - /* If we already know the current frequency, we're done. */ + /* + * If we already know the current frequency, and the driver didn't ask + * for uncached usage, we're done. + */ CF_MTX_LOCK(&sc->lock); curr_set = &sc->curr_level.total_set; - if (curr_set->freq != CPUFREQ_VAL_UNKNOWN) { + error = CPUFREQ_DRV_TYPE(sc->cf_drv_dev, &type); + if (error == 0 && (type & CPUFREQ_FLAG_UNCACHED)) { + struct cf_setting set; + + /* + * If the driver wants to always report back the real frequency, + * first try the driver and if that fails, fall back to + * estimating. + */ + if (CPUFREQ_DRV_GET(sc->cf_drv_dev, &set) != 0) + goto estimate; + sc->curr_level.total_set = set; + CF_DEBUG("get returning immediate freq %d\n", curr_set->freq); + goto out; + } else if (curr_set->freq != CPUFREQ_VAL_UNKNOWN) { CF_DEBUG("get returning known freq %d\n", curr_set->freq); + error = 0; goto out; } CF_MTX_UNLOCK(&sc->lock); @@ -461,11 +501,6 @@ cf_get_method(device_t dev, struct cf_level *level) free(levels, M_TEMP); return (error); } - error = device_get_children(device_get_parent(dev), &devs, &numdevs); - if (error) { - free(levels, M_TEMP); - return (error); - } /* * Reacquire the lock and search for the given level. @@ -476,24 +511,21 @@ cf_get_method(device_t dev, struct cf_level *level) * The estimation code below catches this case though. */ CF_MTX_LOCK(&sc->lock); - for (n = 0; n < numdevs && curr_set->freq == CPUFREQ_VAL_UNKNOWN; n++) { - if (!device_is_attached(devs[n])) - continue; - if (CPUFREQ_DRV_GET(devs[n], &set) != 0) - continue; - for (i = 0; i < count; i++) { - if (set.freq == levels[i].total_set.freq) { - sc->curr_level = levels[i]; - break; - } - } - } - free(devs, M_TEMP); + i = cpufreq_get_level(sc->cf_drv_dev, levels, count); + if (i >= 0) + sc->curr_level = levels[i]; + else + CF_DEBUG("Couldn't find supported level for %s\n", + device_get_nameunit(sc->cf_drv_dev)); + if (curr_set->freq != CPUFREQ_VAL_UNKNOWN) { CF_DEBUG("get matched freq %d from drivers\n", curr_set->freq); goto out; } +estimate: + CF_MTX_ASSERT(&sc->lock); + /* * We couldn't find an exact match, so attempt to estimate and then * match against a level. @@ -525,17 +557,82 @@ out: return (error); } +/* + * Either directly obtain settings from the cpufreq driver, or build a list of + * relative settings to be integrated later against an absolute max. + */ static int +cpufreq_add_levels(device_t cf_dev, struct cf_setting_lst *rel_sets) +{ + struct cf_setting_array *set_arr; + struct cf_setting *sets; + device_t dev; + struct cpufreq_softc *sc; + int type, set_count, error; + + sc = device_get_softc(cf_dev); + dev = sc->cf_drv_dev; + + /* Skip devices that aren't ready. */ + if (!device_is_attached(cf_dev)) + return (0); + + /* + * Get settings, skipping drivers that offer no settings or + * provide settings for informational purposes only. + */ + error = CPUFREQ_DRV_TYPE(dev, &type); + if (error != 0 || (type & CPUFREQ_FLAG_INFO_ONLY)) { + if (error == 0) { + CF_DEBUG("skipping info-only driver %s\n", + device_get_nameunit(cf_dev)); + } + return (error); + } + + sets = malloc(MAX_SETTINGS * sizeof(*sets), M_TEMP, M_NOWAIT); + if (sets == NULL) + return (ENOMEM); + + set_count = MAX_SETTINGS; + error = CPUFREQ_DRV_SETTINGS(dev, sets, &set_count); + if (error != 0 || set_count == 0) + goto out; + + /* Add the settings to our absolute/relative lists. */ + switch (type & CPUFREQ_TYPE_MASK) { + case CPUFREQ_TYPE_ABSOLUTE: + error = cpufreq_insert_abs(sc, sets, set_count); + break; + case CPUFREQ_TYPE_RELATIVE: + CF_DEBUG("adding %d relative settings\n", set_count); + set_arr = malloc(sizeof(*set_arr), M_TEMP, M_NOWAIT); + if (set_arr == NULL) { + error = ENOMEM; + goto out; + } + bcopy(sets, set_arr->sets, set_count * sizeof(*sets)); + set_arr->count = set_count; + TAILQ_INSERT_TAIL(rel_sets, set_arr, link); + break; + default: + error = EINVAL; + } + +out: + free(sets, M_TEMP); + return (error); +} + +static int cf_levels_method(device_t dev, struct cf_level *levels, int *count) { struct cf_setting_array *set_arr; struct cf_setting_lst rel_sets; struct cpufreq_softc *sc; struct cf_level *lev; - struct cf_setting *sets; struct pcpu *pc; - device_t *devs; - int error, i, numdevs, set_count, type; + int error, i; uint64_t rate; if (levels == NULL || count == NULL) @@ -543,67 +640,21 @@ cf_levels_method(device_t dev, struct cf_level *levels TAILQ_INIT(&rel_sets); sc = device_get_softc(dev); - error = device_get_children(device_get_parent(dev), &devs, &numdevs); - if (error) - return (error); - sets = malloc(MAX_SETTINGS * sizeof(*sets), M_TEMP, M_NOWAIT); - if (sets == NULL) { - free(devs, M_TEMP); - return (ENOMEM); - } - /* Get settings from all cpufreq drivers. */ CF_MTX_LOCK(&sc->lock); - for (i = 0; i < numdevs; i++) { - /* Skip devices that aren't ready. */ - if (!device_is_attached(devs[i])) - continue; + error = cpufreq_add_levels(sc->dev, &rel_sets); + if (error) + goto out; - /* - * Get settings, skipping drivers that offer no settings or - * provide settings for informational purposes only. - */ - error = CPUFREQ_DRV_TYPE(devs[i], &type); - if (error || (type & CPUFREQ_FLAG_INFO_ONLY)) { - if (error == 0) { - CF_DEBUG("skipping info-only driver %s\n", - device_get_nameunit(devs[i])); - } - continue; - } - set_count = MAX_SETTINGS; - error = CPUFREQ_DRV_SETTINGS(devs[i], sets, &set_count); - if (error || set_count == 0) - continue; - - /* Add the settings to our absolute/relative lists. */ - switch (type & CPUFREQ_TYPE_MASK) { - case CPUFREQ_TYPE_ABSOLUTE: - error = cpufreq_insert_abs(sc, sets, set_count); - break; - case CPUFREQ_TYPE_RELATIVE: - CF_DEBUG("adding %d relative settings\n", set_count); - set_arr = malloc(sizeof(*set_arr), M_TEMP, M_NOWAIT); - if (set_arr == NULL) { - error = ENOMEM; - goto out; - } - bcopy(sets, set_arr->sets, set_count * sizeof(*sets)); - set_arr->count = set_count; - TAILQ_INSERT_TAIL(&rel_sets, set_arr, link); - break; - default: - error = EINVAL; - } - if (error) - goto out; - } - /* * If there are no absolute levels, create a fake one at 100%. We * then cache the clockrate for later use as our base frequency. */ if (TAILQ_EMPTY(&sc->all_levels)) { + struct cf_setting set; + + CF_DEBUG("No absolute levels returned by driver\n"); + if (sc->max_mhz == CPUFREQ_VAL_UNKNOWN) { sc->max_mhz = cpu_get_nominal_mhz(dev); /* @@ -617,10 +668,10 @@ cf_levels_method(device_t dev, struct cf_level *levels sc->max_mhz = rate / 1000000; } } - memset(&sets[0], CPUFREQ_VAL_UNKNOWN, sizeof(*sets)); - sets[0].freq = sc->max_mhz; - sets[0].dev = NULL; - error = cpufreq_insert_abs(sc, sets, 1); + memset(&set, CPUFREQ_VAL_UNKNOWN, sizeof(set)); + set.freq = sc->max_mhz; + set.dev = NULL; + error = cpufreq_insert_abs(sc, &set, 1); if (error) goto out; } @@ -665,8 +716,6 @@ out: TAILQ_REMOVE(&rel_sets, set_arr, link); free(set_arr, M_TEMP); } - free(devs, M_TEMP); - free(sets, M_TEMP); return (error); } @@ -1011,11 +1060,24 @@ out: return (error); } +static void +cpufreq_add_freq_driver_sysctl(device_t cf_dev) +{ + struct cpufreq_softc *sc; + + sc = device_get_softc(cf_dev); + SYSCTL_ADD_CONST_STRING(&sc->sysctl_ctx, + SYSCTL_CHILDREN(device_get_sysctl_tree(cf_dev)), OID_AUTO, + "freq_driver", CTLFLAG_RD, device_get_nameunit(sc->cf_drv_dev), + "cpufreq driver used by this cpu"); +} + int cpufreq_register(device_t dev) { struct cpufreq_softc *sc; device_t cf_dev, cpu_dev; + int error; /* Add a sysctl to get each driver's settings separately. */ SYSCTL_ADD_PROC(device_get_sysctl_ctx(dev), @@ -1031,6 +1093,7 @@ cpufreq_register(device_t dev) if ((cf_dev = device_find_child(cpu_dev, "cpufreq", -1))) { sc = device_get_softc(cf_dev); sc->max_mhz = CPUFREQ_VAL_UNKNOWN; + MPASS(sc->cf_drv_dev != NULL); return (0); } @@ -1040,40 +1103,36 @@ cpufreq_register(device_t dev) return (ENOMEM); device_quiet(cf_dev); - return (device_probe_and_attach(cf_dev)); + error = device_probe_and_attach(cf_dev); + if (error) + return (error); + + sc = device_get_softc(cf_dev); + sc->cf_drv_dev = dev; + cpufreq_add_freq_driver_sysctl(cf_dev); + return (error); } int cpufreq_unregister(device_t dev) { - device_t cf_dev, *devs; - int cfcount, devcount, error, i, type; + device_t cf_dev; + struct cpufreq_softc *sc; /* * If this is the last cpufreq child device, remove the control * device as well. We identify cpufreq children by calling a method * they support. */ - error = device_get_children(device_get_parent(dev), &devs, &devcount); - if (error) - return (error); cf_dev = device_find_child(device_get_parent(dev), "cpufreq", -1); if (cf_dev == NULL) { device_printf(dev, "warning: cpufreq_unregister called with no cpufreq device active\n"); - free(devs, M_TEMP); return (0); } - cfcount = 0; - for (i = 0; i < devcount; i++) { - if (!device_is_attached(devs[i])) - continue; - if (CPUFREQ_DRV_TYPE(devs[i], &type) == 0) - cfcount++; - } - if (cfcount <= 1) - device_delete_child(device_get_parent(cf_dev), cf_dev); - free(devs, M_TEMP); + sc = device_get_softc(cf_dev); + MPASS(sc->cf_drv_dev == dev); + device_delete_child(device_get_parent(cf_dev), cf_dev); return (0); } Modified: head/sys/modules/cpufreq/Makefile ============================================================================== --- head/sys/modules/cpufreq/Makefile Wed Jan 22 22:51:55 2020 (r357001) +++ head/sys/modules/cpufreq/Makefile Wed Jan 22 23:28:42 2020 (r357002) @@ -11,7 +11,7 @@ SRCS+= bus_if.h cpufreq_if.h device_if.h pci_if.h .PATH: ${SRCTOP}/sys/x86/cpufreq SRCS+= acpi_if.h opt_acpi.h -SRCS+= est.c hwpstate.c p4tcc.c powernow.c +SRCS+= est.c hwpstate_amd.c p4tcc.c powernow.c hwpstate_intel.c .endif .if ${MACHINE} == "i386" Modified: head/sys/sys/cpu.h ============================================================================== --- head/sys/sys/cpu.h Wed Jan 22 22:51:55 2020 (r357001) +++ head/sys/sys/cpu.h Wed Jan 22 23:28:42 2020 (r357002) @@ -120,11 +120,16 @@ TAILQ_HEAD(cf_level_lst, cf_level); * information about settings but rely on another machine-dependent driver * for actually performing the frequency transition (e.g., ACPI performance * states of type "functional fixed hardware.") + * + * The "uncached" flag tells CPUFREQ_DRV_GET to try obtaining the real + * instantaneous frequency from the underlying hardware regardless of cached + * state. It is probably a bug to not combine this with "info only" */ #define CPUFREQ_TYPE_MASK 0xffff #define CPUFREQ_TYPE_RELATIVE (1<<0) #define CPUFREQ_TYPE_ABSOLUTE (1<<1) #define CPUFREQ_FLAG_INFO_ONLY (1<<16) +#define CPUFREQ_FLAG_UNCACHED (1<<17) /* * When setting a level, the caller indicates the priority of this request. Modified: head/sys/x86/cpufreq/est.c ============================================================================== --- head/sys/x86/cpufreq/est.c Wed Jan 22 22:51:55 2020 (r357001) +++ head/sys/x86/cpufreq/est.c Wed Jan 22 23:28:42 2020 (r357002) @@ -50,6 +50,8 @@ __FBSDID("$FreeBSD$"); #include #include "acpi_if.h" +#include + /* Status/control registers (from the IA-32 System Programming Guide). */ #define MSR_PERF_STATUS 0x198 #define MSR_PERF_CTL 0x199 @@ -898,6 +900,7 @@ static driver_t est_driver = { static devclass_t est_devclass; DRIVER_MODULE(est, cpu, est_driver, est_devclass, 0, 0); +MODULE_DEPEND(est, hwpstate_intel, 1, 1, 1); static int est_features(driver_t *driver, u_int *features) @@ -915,6 +918,15 @@ static void est_identify(driver_t *driver, device_t parent) { device_t child; + + /* + * Defer to hwpstate if it is present. This priority logic + * should be replaced with normal newbus probing in the + * future. + */ + intel_hwpstate_identify(NULL, parent); + if (device_find_child(parent, "hwpstate_intel", -1) != NULL) + return; /* Make sure we're not being doubly invoked. */ if (device_find_child(parent, "est", -1) != NULL) Copied: head/sys/x86/cpufreq/hwpstate_amd.c (from r357001, head/sys/x86/cpufreq/hwpstate.c) ============================================================================== --- /dev/null 00:00:00 1970 (empty, because file is newly added) +++ head/sys/x86/cpufreq/hwpstate_amd.c Wed Jan 22 23:28:42 2020 (r357002, copy of r357001, head/sys/x86/cpufreq/hwpstate.c) @@ -0,0 +1,543 @@ +/*- + * SPDX-License-Identifier: BSD-2-Clause-FreeBSD + * + * Copyright (c) 2005 Nate Lawson + * Copyright (c) 2004 Colin Percival + * Copyright (c) 2004-2005 Bruno Durcot + * Copyright (c) 2004 FUKUDA Nobuhiko + * Copyright (c) 2009 Michael Reifenberger + * Copyright (c) 2009 Norikatsu Shigemura + * Copyright (c) 2008-2009 Gen Otsuji + * + * This code is depending on kern_cpu.c, est.c, powernow.c, p4tcc.c, smist.c + * in various parts. The authors of these files are Nate Lawson, + * Colin Percival, Bruno Durcot, and FUKUDA Nobuhiko. + * This code contains patches by Michael Reifenberger and Norikatsu Shigemura. + * Thank you. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted providing that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED + * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY + * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, + * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING + * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + */ + +/* + * For more info: + * BIOS and Kernel Developer's Guide(BKDG) for AMD Family 10h Processors + * 31116 Rev 3.20 February 04, 2009 + * BIOS and Kernel Developer's Guide(BKDG) for AMD Family 11h Processors + * 41256 Rev 3.00 - July 07, 2008 + */ + +#include +__FBSDID("$FreeBSD$"); + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include + +#include + +#include + +#include "acpi_if.h" +#include "cpufreq_if.h" + +#define MSR_AMD_10H_11H_LIMIT 0xc0010061 +#define MSR_AMD_10H_11H_CONTROL 0xc0010062 +#define MSR_AMD_10H_11H_STATUS 0xc0010063 +#define MSR_AMD_10H_11H_CONFIG 0xc0010064 + +#define AMD_10H_11H_MAX_STATES 16 + +/* for MSR_AMD_10H_11H_LIMIT C001_0061 */ +#define AMD_10H_11H_GET_PSTATE_MAX_VAL(msr) (((msr) >> 4) & 0x7) +#define AMD_10H_11H_GET_PSTATE_LIMIT(msr) (((msr)) & 0x7) +/* for MSR_AMD_10H_11H_CONFIG 10h:C001_0064:68 / 11h:C001_0064:6B */ +#define AMD_10H_11H_CUR_VID(msr) (((msr) >> 9) & 0x7F) +#define AMD_10H_11H_CUR_DID(msr) (((msr) >> 6) & 0x07) +#define AMD_10H_11H_CUR_FID(msr) ((msr) & 0x3F) + +#define AMD_17H_CUR_VID(msr) (((msr) >> 14) & 0xFF) +#define AMD_17H_CUR_DID(msr) (((msr) >> 8) & 0x3F) +#define AMD_17H_CUR_FID(msr) ((msr) & 0xFF) + +#define HWPSTATE_DEBUG(dev, msg...) \ + do { \ + if (hwpstate_verbose) \ + device_printf(dev, msg); \ + } while (0) + +struct hwpstate_setting { + int freq; /* CPU clock in Mhz or 100ths of a percent. */ + int volts; /* Voltage in mV. */ + int power; /* Power consumed in mW. */ + int lat; /* Transition latency in us. */ + int pstate_id; /* P-State id */ +}; + +struct hwpstate_softc { + device_t dev; + struct hwpstate_setting hwpstate_settings[AMD_10H_11H_MAX_STATES]; + int cfnum; +}; + +static void hwpstate_identify(driver_t *driver, device_t parent); +static int hwpstate_probe(device_t dev); +static int hwpstate_attach(device_t dev); +static int hwpstate_detach(device_t dev); +static int hwpstate_set(device_t dev, const struct cf_setting *cf); +static int hwpstate_get(device_t dev, struct cf_setting *cf); +static int hwpstate_settings(device_t dev, struct cf_setting *sets, int *count); +static int hwpstate_type(device_t dev, int *type); +static int hwpstate_shutdown(device_t dev); +static int hwpstate_features(driver_t *driver, u_int *features); +static int hwpstate_get_info_from_acpi_perf(device_t dev, device_t perf_dev); +static int hwpstate_get_info_from_msr(device_t dev); +static int hwpstate_goto_pstate(device_t dev, int pstate_id); + +static int hwpstate_verbose; +SYSCTL_INT(_debug, OID_AUTO, hwpstate_verbose, CTLFLAG_RWTUN, + &hwpstate_verbose, 0, "Debug hwpstate"); + +static int hwpstate_verify; +SYSCTL_INT(_debug, OID_AUTO, hwpstate_verify, CTLFLAG_RWTUN, + &hwpstate_verify, 0, "Verify P-state after setting"); + +static device_method_t hwpstate_methods[] = { + /* Device interface */ + DEVMETHOD(device_identify, hwpstate_identify), + DEVMETHOD(device_probe, hwpstate_probe), + DEVMETHOD(device_attach, hwpstate_attach), + DEVMETHOD(device_detach, hwpstate_detach), + DEVMETHOD(device_shutdown, hwpstate_shutdown), + + /* cpufreq interface */ + DEVMETHOD(cpufreq_drv_set, hwpstate_set), + DEVMETHOD(cpufreq_drv_get, hwpstate_get), + DEVMETHOD(cpufreq_drv_settings, hwpstate_settings), + DEVMETHOD(cpufreq_drv_type, hwpstate_type), + + /* ACPI interface */ + DEVMETHOD(acpi_get_features, hwpstate_features), + + {0, 0} +}; + +static devclass_t hwpstate_devclass; +static driver_t hwpstate_driver = { + "hwpstate", + hwpstate_methods, + sizeof(struct hwpstate_softc), +}; + +DRIVER_MODULE(hwpstate, cpu, hwpstate_driver, hwpstate_devclass, 0, 0); + +/* + * Go to Px-state on all cpus considering the limit. + */ +static int +hwpstate_goto_pstate(device_t dev, int id) +{ + sbintime_t sbt; + uint64_t msr; + int cpu, i, j, limit; + + /* get the current pstate limit */ + msr = rdmsr(MSR_AMD_10H_11H_LIMIT); + limit = AMD_10H_11H_GET_PSTATE_LIMIT(msr); + if (limit > id) + id = limit; + + cpu = curcpu; + HWPSTATE_DEBUG(dev, "setting P%d-state on cpu%d\n", id, cpu); + /* Go To Px-state */ + wrmsr(MSR_AMD_10H_11H_CONTROL, id); + + /* + * We are going to the same Px-state on all cpus. + * Probably should take _PSD into account. + */ + CPU_FOREACH(i) { + if (i == cpu) + continue; + + /* Bind to each cpu. */ + thread_lock(curthread); + sched_bind(curthread, i); + thread_unlock(curthread); + HWPSTATE_DEBUG(dev, "setting P%d-state on cpu%d\n", id, i); + /* Go To Px-state */ + wrmsr(MSR_AMD_10H_11H_CONTROL, id); + } + + /* + * Verify whether each core is in the requested P-state. + */ + if (hwpstate_verify) { + CPU_FOREACH(i) { + thread_lock(curthread); + sched_bind(curthread, i); + thread_unlock(curthread); + /* wait loop (100*100 usec is enough ?) */ + for (j = 0; j < 100; j++) { + /* get the result. not assure msr=id */ + msr = rdmsr(MSR_AMD_10H_11H_STATUS); + if (msr == id) + break; + sbt = SBT_1MS / 10; + tsleep_sbt(dev, PZERO, "pstate_goto", sbt, + sbt >> tc_precexp, 0); + } + HWPSTATE_DEBUG(dev, "result: P%d-state on cpu%d\n", + (int)msr, i); + if (msr != id) { + HWPSTATE_DEBUG(dev, + "error: loop is not enough.\n"); + return (ENXIO); + } + } + } + + return (0); +} + +static int +hwpstate_set(device_t dev, const struct cf_setting *cf) +{ + struct hwpstate_softc *sc; + struct hwpstate_setting *set; + int i; + + if (cf == NULL) + return (EINVAL); + sc = device_get_softc(dev); + set = sc->hwpstate_settings; + for (i = 0; i < sc->cfnum; i++) + if (CPUFREQ_CMP(cf->freq, set[i].freq)) + break; + if (i == sc->cfnum) + return (EINVAL); + + return (hwpstate_goto_pstate(dev, set[i].pstate_id)); +} + +static int +hwpstate_get(device_t dev, struct cf_setting *cf) +{ + struct hwpstate_softc *sc; + struct hwpstate_setting set; + uint64_t msr; + + sc = device_get_softc(dev); + if (cf == NULL) + return (EINVAL); + msr = rdmsr(MSR_AMD_10H_11H_STATUS); + if (msr >= sc->cfnum) + return (EINVAL); + set = sc->hwpstate_settings[msr]; + + cf->freq = set.freq; + cf->volts = set.volts; + cf->power = set.power; + cf->lat = set.lat; + cf->dev = dev; + return (0); +} + +static int +hwpstate_settings(device_t dev, struct cf_setting *sets, int *count) +{ + struct hwpstate_softc *sc; + struct hwpstate_setting set; + int i; + + if (sets == NULL || count == NULL) + return (EINVAL); + sc = device_get_softc(dev); + if (*count < sc->cfnum) + return (E2BIG); + for (i = 0; i < sc->cfnum; i++, sets++) { + set = sc->hwpstate_settings[i]; + sets->freq = set.freq; + sets->volts = set.volts; + sets->power = set.power; + sets->lat = set.lat; + sets->dev = dev; + } + *count = sc->cfnum; + + return (0); +} + +static int +hwpstate_type(device_t dev, int *type) +{ + + if (type == NULL) + return (EINVAL); + + *type = CPUFREQ_TYPE_ABSOLUTE; + return (0); +} + +static void +hwpstate_identify(driver_t *driver, device_t parent) +{ + + if (device_find_child(parent, "hwpstate", -1) != NULL) + return; + + if ((cpu_vendor_id != CPU_VENDOR_AMD || CPUID_TO_FAMILY(cpu_id) < 0x10) && + cpu_vendor_id != CPU_VENDOR_HYGON) + return; + + /* + * Check if hardware pstate enable bit is set. + */ + if ((amd_pminfo & AMDPM_HW_PSTATE) == 0) { + HWPSTATE_DEBUG(parent, "hwpstate enable bit is not set.\n"); + return; + } + + if (resource_disabled("hwpstate", 0)) + return; *** DIFF OUTPUT TRUNCATED AT 1000 LINES ***