Merge tag 'pm-4.13-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull power management updates from Rafael Wysocki:
 "The big ticket items here are the rework of suspend-to-idle in order
  to add proper support for power button wakeup from it on recent Dell
  laptops and the rework of interfaces exporting the current CPU
  frequency on x86.

  In addition to that, support for a few new pieces of hardware is
  added, the PCI/ACPI device wakeup infrastructure is simplified
  significantly and the wakeup IRQ framework is fixed to unbreak the IRQ
  bus locking infrastructure.

  Also, there are some functional improvements for intel_pstate, tools
  updates and small fixes and cleanups all over.

  Specifics:

   - Rework suspend-to-idle to allow it to take wakeup events signaled
     by the EC into account on ACPI-based platforms in order to properly
     support power button wakeup from suspend-to-idle on recent Dell
     laptops (Rafael Wysocki).

     That includes the core suspend-to-idle code rework, support for the
     Low Power S0 _DSM interface, and support for the ACPI INT0002
     Virtual GPIO device from Hans de Goede (required for USB keyboard
     wakeup from suspend-to-idle to work on some machines).

   - Stop trying to export the current CPU frequency via /proc/cpuinfo
     on x86 as that is inaccurate and confusing (Len Brown).

   - Rework the way in which the current CPU frequency is exported by
     the kernel (over the cpufreq sysfs interface) on x86 systems with
     the APERF and MPERF registers by always using values read from
     these registers, when available, to compute the current frequency
     regardless of which cpufreq driver is in use (Len Brown).

   - Rework the PCI/ACPI device wakeup infrastructure to remove the
     questionable and artificial distinction between "devices that can
     wake up the system from sleep states" and "devices that can
     generate wakeup signals in the working state" from it, which allows
     the code to be simplified quite a bit (Rafael Wysocki).

   - Fix the wakeup IRQ framework by making it use SRCU instead of RCU
     which doesn't allow sleeping in the read-side critical sections,
     but which in turn is expected to be allowed by the IRQ bus locking
     infrastructure (Thomas Gleixner).

   - Modify some computations in the intel_pstate driver to avoid
     rounding errors resulting from them (Srinivas Pandruvada).

   - Reduce the overhead of the intel_pstate driver in the HWP
     (hardware-managed P-states) mode and when the "performance" P-state
     selection algorithm is in use by making it avoid registering
     scheduler callbacks in those cases (Len Brown).

   - Rework the energy_performance_preference sysfs knob in intel_pstate
     by changing the values that correspond to different symbolic hint
     names used by it (Len Brown).

   - Make it possible to use more than one cpuidle driver at the same
     time on ARM (Daniel Lezcano).

   - Make it possible to prevent the cpuidle menu governor from using
     the 0 state by disabling it via sysfs (Nicholas Piggin).

   - Add support for FFH (Fixed Functional Hardware) MWAIT in ACPI C1 on
     AMD systems (Yazen Ghannam).

   - Make the CPPC cpufreq driver take the lowest nonlinear performance
     information into account (Prashanth Prakash).

   - Add support for hi3660 to the cpufreq-dt driver, fix the imx6q
     driver and clean up the sfi, exynos5440 and intel_pstate drivers
     (Colin Ian King, Krzysztof Kozlowski, Octavian Purdila, Rafael
     Wysocki, Tao Wang).

   - Fix a few minor issues in the generic power domains (genpd)
     framework and clean it up somewhat (Krzysztof Kozlowski, Mikko
     Perttunen, Viresh Kumar).

   - Fix a couple of minor issues in the operating performance points
     (OPP) framework and clean it up somewhat (Viresh Kumar).

   - Fix a CONFIG dependency in the hibernation core and clean it up
     slightly (Balbir Singh, Arvind Yadav, BaoJun Luo).

   - Add rk3228 support to the rockchip-io adaptive voltage scaling
     (AVS) driver (David Wu).

   - Fix an incorrect bit shift operation in the RAPL power capping
     driver (Adam Lessnau).

   - Add support for the EPP field in the HWP (hardware managed
     P-states) control register, HWP.EPP, to the x86_energy_perf_policy
     tool and update msr-index.h with HWP.EPP values (Len Brown).

   - Fix some minor issues in the turbostat tool (Len Brown).

   - Add support for AMD family 0x17 CPUs to the cpupower tool and fix a
     minor issue in it (Sherry Hurwitz).

   - Assorted cleanups, mostly related to the constification of some
     data structures (Arvind Yadav, Joe Perches, Kees Cook, Krzysztof
     Kozlowski)"

* tag 'pm-4.13-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (69 commits)
  cpufreq: Update scaling_cur_freq documentation
  cpufreq: intel_pstate: Clean up after performance governor changes
  PM: hibernate: constify attribute_group structures.
  cpuidle: menu: allow state 0 to be disabled
  intel_idle: Use more common logging style
  PM / Domains: Fix missing default_power_down_ok comment
  PM / Domains: Fix unsafe iteration over modified list of domains
  PM / Domains: Fix unsafe iteration over modified list of domain providers
  PM / Domains: Fix unsafe iteration over modified list of device links
  PM / Domains: Handle safely genpd_syscore_switch() call on non-genpd device
  PM / Domains: Call driver's noirq callbacks
  PM / core: Drop run_wake flag from struct dev_pm_info
  PCI / PM: Simplify device wakeup settings code
  PCI / PM: Drop pme_interrupt flag from struct pci_dev
  ACPI / PM: Consolidate device wakeup settings code
  ACPI / PM: Drop run_wake from struct acpi_device_wakeup_flags
  PM / QoS: constify *_attribute_group.
  PM / AVS: rockchip-io: add io selectors and supplies for rk3228
  powercap/RAPL: prevent overridding bits outside of the mask
  PM / sysfs: Constify attribute groups
  ...

7 files changed, 1621 insertions, 301 deletions

diff --git a/tools/power/cpupower/utils/helpers/amd.c b/tools/power/cpupower/utils/helpers/amd.c
index 6437ef39aeea..5fd5c5b8c7b8 100644
--- a/tools/power/cpupower/utils/helpers/amd.c
+++ b/tools/power/cpupower/utils/helpers/amd.c
@@ -26,6 +26,15 @@ union msr_pstate {
 		unsigned res3:21;
 		unsigned en:1;
 	} bits;
+	struct {
+		unsigned fid:8;
+		unsigned did:6;
+		unsigned vid:8;
+		unsigned iddval:8;
+		unsigned idddiv:2;
+		unsigned res1:30;
+		unsigned en:1;
+	} fam17h_bits;
 	unsigned long long val;
 };
 
@@ -35,6 +44,8 @@ static int get_did(int family, union msr_pstate pstate)
 
 	if (family == 0x12)
 		t = pstate.val & 0xf;
+	else if (family == 0x17)
+		t = pstate.fam17h_bits.did;
 	else
 		t = pstate.bits.did;
 
@@ -44,16 +55,20 @@ static int get_did(int family, union msr_pstate pstate)
 static int get_cof(int family, union msr_pstate pstate)
 {
 	int t;
-	int fid, did;
+	int fid, did, cof;
 
 	did = get_did(family, pstate);
-
-	t = 0x10;
-	fid = pstate.bits.fid;
-	if (family == 0x11)
-		t = 0x8;
-
-	return (100 * (fid + t)) >> did;
+	if (family == 0x17) {
+		fid = pstate.fam17h_bits.fid;
+		cof = 200 * fid / did;
+	} else {
+		t = 0x10;
+		fid = pstate.bits.fid;
+		if (family == 0x11)
+			t = 0x8;
+		cof = (100 * (fid + t)) >> did;
+	}
+	return cof;
 }
 
 /* Needs:
diff --git a/tools/power/cpupower/utils/helpers/helpers.h b/tools/power/cpupower/utils/helpers/helpers.h
index afb66f80554e..799a18be60aa 100644
--- a/tools/power/cpupower/utils/helpers/helpers.h
+++ b/tools/power/cpupower/utils/helpers/helpers.h
@@ -70,6 +70,8 @@ enum cpupower_cpu_vendor {X86_VENDOR_UNKNOWN = 0, X86_VENDOR_INTEL,
 #define CPUPOWER_CAP_IS_SNB		0x00000020
 #define CPUPOWER_CAP_INTEL_IDA		0x00000040
 
+#define CPUPOWER_AMD_CPBDIS		0x02000000
+
 #define MAX_HW_PSTATES 10
 
 struct cpupower_cpu_info {
diff --git a/tools/power/cpupower/utils/helpers/misc.c b/tools/power/cpupower/utils/helpers/misc.c
index 1609243f5c64..601d719d4e60 100644
--- a/tools/power/cpupower/utils/helpers/misc.c
+++ b/tools/power/cpupower/utils/helpers/misc.c
@@ -2,11 +2,14 @@
 
 #include "helpers/helpers.h"
 
+#define MSR_AMD_HWCR	0xc0010015
+
 int cpufreq_has_boost_support(unsigned int cpu, int *support, int *active,
 			int *states)
 {
 	struct cpupower_cpu_info cpu_info;
 	int ret;
+	unsigned long long val;
 
 	*support = *active = *states = 0;
 
@@ -16,10 +19,22 @@ int cpufreq_has_boost_support(unsigned int cpu, int *support, int *active,
 
 	if (cpupower_cpu_info.caps & CPUPOWER_CAP_AMD_CBP) {
 		*support = 1;
-		amd_pci_get_num_boost_states(active, states);
-		if (ret <= 0)
-			return ret;
-		*support = 1;
+
+		/* AMD Family 0x17 does not utilize PCI D18F4 like prior
+		 * families and has no fixed discrete boost states but
+		 * has Hardware determined variable increments instead.
+		 */
+
+		if (cpu_info.family == 0x17) {
+			if (!read_msr(cpu, MSR_AMD_HWCR, &val)) {
+				if (!(val & CPUPOWER_AMD_CPBDIS))
+					*active = 1;
+			}
+		} else {
+			ret = amd_pci_get_num_boost_states(active, states);
+			if (ret)
+				return ret;
+		}
 	} else if (cpupower_cpu_info.caps & CPUPOWER_CAP_INTEL_IDA)
 		*support = *active = 1;
 	return 0;
diff --git a/tools/power/x86/turbostat/turbostat.c b/tools/power/x86/turbostat/turbostat.c
index b11294730771..0dafba2c1e7d 100644
--- a/tools/power/x86/turbostat/turbostat.c
+++ b/tools/power/x86/turbostat/turbostat.c
@@ -57,7 +57,6 @@ unsigned int list_header_only;
 unsigned int dump_only;
 unsigned int do_snb_cstates;
 unsigned int do_knl_cstates;
-unsigned int do_skl_residency;
 unsigned int do_slm_cstates;
 unsigned int use_c1_residency_msr;
 unsigned int has_aperf;
@@ -93,6 +92,7 @@ unsigned int do_ring_perf_limit_reasons;
 unsigned int crystal_hz;
 unsigned long long tsc_hz;
 int base_cpu;
+int do_migrate;
 double discover_bclk(unsigned int family, unsigned int model);
 unsigned int has_hwp;	/* IA32_PM_ENABLE, IA32_HWP_CAPABILITIES */
 			/* IA32_HWP_REQUEST, IA32_HWP_STATUS */
@@ -151,6 +151,8 @@ size_t cpu_present_setsize, cpu_affinity_setsize, cpu_subset_size;
 #define MAX_ADDED_COUNTERS 16
 
 struct thread_data {
+	struct timeval tv_begin;
+	struct timeval tv_end;
 	unsigned long long tsc;
 	unsigned long long aperf;
 	unsigned long long mperf;
@@ -301,6 +303,9 @@ int for_all_cpus(int (func)(struct thread_data *, struct core_data *, struct pkg
 
 int cpu_migrate(int cpu)
 {
+	if (!do_migrate)
+		return 0;
+
 	CPU_ZERO_S(cpu_affinity_setsize, cpu_affinity_set);
 	CPU_SET_S(cpu, cpu_affinity_setsize, cpu_affinity_set);
 	if (sched_setaffinity(0, cpu_affinity_setsize, cpu_affinity_set) == -1)
@@ -384,8 +389,14 @@ struct msr_counter bic[] = {
 	{ 0x0, "CPU" },
 	{ 0x0, "Mod%c6" },
 	{ 0x0, "sysfs" },
+	{ 0x0, "Totl%C0" },
+	{ 0x0, "Any%C0" },
+	{ 0x0, "GFX%C0" },
+	{ 0x0, "CPUGFX%" },
 };
 
+
+
 #define MAX_BIC (sizeof(bic) / sizeof(struct msr_counter))
 #define	BIC_Package	(1ULL << 0)
 #define	BIC_Avg_MHz	(1ULL << 1)
@@ -426,6 +437,10 @@ struct msr_counter bic[] = {
 #define	BIC_CPU		(1ULL << 36)
 #define	BIC_Mod_c6	(1ULL << 37)
 #define	BIC_sysfs	(1ULL << 38)
+#define	BIC_Totl_c0	(1ULL << 39)
+#define	BIC_Any_c0	(1ULL << 40)
+#define	BIC_GFX_c0	(1ULL << 41)
+#define	BIC_CPUGFX	(1ULL << 42)
 
 unsigned long long bic_enabled = 0xFFFFFFFFFFFFFFFFULL;
 unsigned long long bic_present = BIC_sysfs;
@@ -521,6 +536,8 @@ void print_header(char *delim)
 	struct msr_counter *mp;
 	int printed = 0;
 
+	if (debug)
+		outp += sprintf(outp, "usec %s", delim);
 	if (DO_BIC(BIC_Package))
 		outp += sprintf(outp, "%sPackage", (printed++ ? delim : ""));
 	if (DO_BIC(BIC_Core))
@@ -599,12 +616,14 @@ void print_header(char *delim)
 	if (DO_BIC(BIC_GFXMHz))
 		outp += sprintf(outp, "%sGFXMHz", (printed++ ? delim : ""));
 
-	if (do_skl_residency) {
+	if (DO_BIC(BIC_Totl_c0))
 		outp += sprintf(outp, "%sTotl%%C0", (printed++ ? delim : ""));
+	if (DO_BIC(BIC_Any_c0))
 		outp += sprintf(outp, "%sAny%%C0", (printed++ ? delim : ""));
+	if (DO_BIC(BIC_GFX_c0))
 		outp += sprintf(outp, "%sGFX%%C0", (printed++ ? delim : ""));
+	if (DO_BIC(BIC_CPUGFX))
 		outp += sprintf(outp, "%sCPUGFX%%", (printed++ ? delim : ""));
-	}
 
 	if (DO_BIC(BIC_Pkgpc2))
 		outp += sprintf(outp, "%sPkg%%pc2", (printed++ ? delim : ""));
@@ -771,6 +790,14 @@ int format_counters(struct thread_data *t, struct core_data *c,
 		(cpu_subset && !CPU_ISSET_S(t->cpu_id, cpu_subset_size, cpu_subset)))
 		return 0;
 
+	if (debug) {
+		/* on each row, print how many usec each timestamp took to gather */
+		struct timeval tv;
+
+		timersub(&t->tv_end, &t->tv_begin, &tv);
+		outp += sprintf(outp, "%5ld\t", tv.tv_sec * 1000000 + tv.tv_usec);
+	}
+
 	interval_float = tv_delta.tv_sec + tv_delta.tv_usec/1000000.0;
 
 	tsc = t->tsc * tsc_tweak;
@@ -912,12 +939,14 @@ int format_counters(struct thread_data *t, struct core_data *c,
 		outp += sprintf(outp, "%s%d", (printed++ ? delim : ""), p->gfx_mhz);
 
 	/* Totl%C0, Any%C0 GFX%C0 CPUGFX% */
-	if (do_skl_residency) {
+	if (DO_BIC(BIC_Totl_c0))
 		outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pkg_wtd_core_c0/tsc);
+	if (DO_BIC(BIC_Any_c0))
 		outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pkg_any_core_c0/tsc);
+	if (DO_BIC(BIC_GFX_c0))
 		outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pkg_any_gfxe_c0/tsc);
+	if (DO_BIC(BIC_CPUGFX))
 		outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pkg_both_core_gfxe_c0/tsc);
-	}
 
 	if (DO_BIC(BIC_Pkgpc2))
 		outp += sprintf(outp, "%s%.2f", (printed++ ? delim : ""), 100.0 * p->pc2/tsc);
@@ -1038,12 +1067,16 @@ delta_package(struct pkg_data *new, struct pkg_data *old)
 	int i;
 	struct msr_counter *mp;
 
-	if (do_skl_residency) {
+
+	if (DO_BIC(BIC_Totl_c0))
 		old->pkg_wtd_core_c0 = new->pkg_wtd_core_c0 - old->pkg_wtd_core_c0;
+	if (DO_BIC(BIC_Any_c0))
 		old->pkg_any_core_c0 = new->pkg_any_core_c0 - old->pkg_any_core_c0;
+	if (DO_BIC(BIC_GFX_c0))
 		old->pkg_any_gfxe_c0 = new->pkg_any_gfxe_c0 - old->pkg_any_gfxe_c0;
+	if (DO_BIC(BIC_CPUGFX))
 		old->pkg_both_core_gfxe_c0 = new->pkg_both_core_gfxe_c0 - old->pkg_both_core_gfxe_c0;
-	}
+
 	old->pc2 = new->pc2 - old->pc2;
 	if (DO_BIC(BIC_Pkgpc3))
 		old->pc3 = new->pc3 - old->pc3;
@@ -1292,12 +1325,14 @@ int sum_counters(struct thread_data *t, struct core_data *c,
 	if (!(t->flags & CPU_IS_FIRST_CORE_IN_PACKAGE))
 		return 0;
 
-	if (do_skl_residency) {
+	if (DO_BIC(BIC_Totl_c0))
 		average.packages.pkg_wtd_core_c0 += p->pkg_wtd_core_c0;
+	if (DO_BIC(BIC_Any_c0))
 		average.packages.pkg_any_core_c0 += p->pkg_any_core_c0;
+	if (DO_BIC(BIC_GFX_c0))
 		average.packages.pkg_any_gfxe_c0 += p->pkg_any_gfxe_c0;
+	if (DO_BIC(BIC_CPUGFX))
 		average.packages.pkg_both_core_gfxe_c0 += p->pkg_both_core_gfxe_c0;
-	}
 
 	average.packages.pc2 += p->pc2;
 	if (DO_BIC(BIC_Pkgpc3))
@@ -1357,12 +1392,14 @@ void compute_average(struct thread_data *t, struct core_data *c,
 	average.cores.c7 /= topo.num_cores;
 	average.cores.mc6_us /= topo.num_cores;
 
-	if (do_skl_residency) {
+	if (DO_BIC(BIC_Totl_c0))
 		average.packages.pkg_wtd_core_c0 /= topo.num_packages;
+	if (DO_BIC(BIC_Any_c0))
 		average.packages.pkg_any_core_c0 /= topo.num_packages;
+	if (DO_BIC(BIC_GFX_c0))
 		average.packages.pkg_any_gfxe_c0 /= topo.num_packages;
+	if (DO_BIC(BIC_CPUGFX))
 		average.packages.pkg_both_core_gfxe_c0 /= topo.num_packages;
-	}
 
 	average.packages.pc2 /= topo.num_packages;
 	if (DO_BIC(BIC_Pkgpc3))
@@ -1482,6 +1519,9 @@ int get_counters(struct thread_data *t, struct core_data *c, struct pkg_data *p)
 	struct msr_counter *mp;
 	int i;
 
+
+	gettimeofday(&t->tv_begin, (struct timezone *)NULL);
+
 	if (cpu_migrate(cpu)) {
 		fprintf(outf, "Could not migrate to CPU %d\n", cpu);
 		return -1;
@@ -1565,7 +1605,7 @@ retry:
 
 	/* collect core counters only for 1st thread in core */
 	if (!(t->flags & CPU_IS_FIRST_THREAD_IN_CORE))
-		return 0;
+		goto done;
 
 	if (DO_BIC(BIC_CPU_c3) && !do_slm_cstates && !do_knl_cstates) {
 		if (get_msr(cpu, MSR_CORE_C3_RESIDENCY, &c->c3))
@@ -1601,15 +1641,21 @@ retry:
 
 	/* collect package counters only for 1st core in package */
 	if (!(t->flags & CPU_IS_FIRST_CORE_IN_PACKAGE))
-		return 0;
+		goto done;
 
-	if (do_skl_residency) {
+	if (DO_BIC(BIC_Totl_c0)) {
 		if (get_msr(cpu, MSR_PKG_WEIGHTED_CORE_C0_RES, &p->pkg_wtd_core_c0))
 			return -10;
+	}
+	if (DO_BIC(BIC_Any_c0)) {
 		if (get_msr(cpu, MSR_PKG_ANY_CORE_C0_RES, &p->pkg_any_core_c0))
 			return -11;
+	}
+	if (DO_BIC(BIC_GFX_c0)) {
 		if (get_msr(cpu, MSR_PKG_ANY_GFXE_C0_RES, &p->pkg_any_gfxe_c0))
 			return -12;
+	}
+	if (DO_BIC(BIC_CPUGFX)) {
 		if (get_msr(cpu, MSR_PKG_BOTH_CORE_GFXE_C0_RES, &p->pkg_both_core_gfxe_c0))
 			return -13;
 	}
@@ -1688,6 +1734,8 @@ retry:
 		if (get_mp(cpu, mp, &p->counter[i]))
 			return -10;
 	}
+done:
+	gettimeofday(&t->tv_end, (struct timezone *)NULL);
 
 	return 0;
 }
@@ -3895,6 +3943,9 @@ void decode_misc_enable_msr(void)
 {
 	unsigned long long msr;
 
+	if (!genuine_intel)
+		return;
+
 	if (!get_msr(base_cpu, MSR_IA32_MISC_ENABLE, &msr))
 		fprintf(outf, "cpu%d: MSR_IA32_MISC_ENABLE: 0x%08llx (%sTCC %sEIST %sMWAIT %sPREFETCH %sTURBO)\n",
 			base_cpu, msr,
@@ -4198,7 +4249,12 @@ void process_cpuid()
 		BIC_PRESENT(BIC_Pkgpc10);
 	}
 	do_irtl_hsw = has_hsw_msrs(family, model);
-	do_skl_residency = has_skl_msrs(family, model);
+	if (has_skl_msrs(family, model)) {
+		BIC_PRESENT(BIC_Totl_c0);
+		BIC_PRESENT(BIC_Any_c0);
+		BIC_PRESENT(BIC_GFX_c0);
+		BIC_PRESENT(BIC_CPUGFX);
+	}
 	do_slm_cstates = is_slm(family, model);
 	do_knl_cstates  = is_knl(family, model);
 
@@ -4578,7 +4634,7 @@ int get_and_dump_counters(void)
 }
 
 void print_version() {
-	fprintf(outf, "turbostat version 17.04.12"
+	fprintf(outf, "turbostat version 17.06.23"
 		" - Len Brown <lenb@kernel.org>\n");
 }
 
@@ -4951,6 +5007,7 @@ void cmdline(int argc, char **argv)
 		{"hide",	required_argument,	0, 'H'},	// meh, -h taken by --help
 		{"Joules",	no_argument,		0, 'J'},
 		{"list",	no_argument,		0, 'l'},
+		{"migrate",	no_argument,		0, 'm'},
 		{"out",		required_argument,	0, 'o'},
 		{"quiet",	no_argument,		0, 'q'},
 		{"show",	required_argument,	0, 's'},
@@ -4962,7 +5019,7 @@ void cmdline(int argc, char **argv)
 
 	progname = argv[0];
 
-	while ((opt = getopt_long_only(argc, argv, "+C:c:Ddhi:JM:m:o:qST:v",
+	while ((opt = getopt_long_only(argc, argv, "+C:c:Ddhi:Jmo:qST:v",
 				long_options, &option_index)) != -1) {
 		switch (opt) {
 		case 'a':
@@ -5005,6 +5062,9 @@ void cmdline(int argc, char **argv)
 			list_header_only++;
 			quiet++;
 			break;
+		case 'm':
+			do_migrate = 1;
+			break;
 		case 'o':
 			outf = fopen_or_die(optarg, "w");
 			break;
diff --git a/tools/power/x86/x86_energy_perf_policy/Makefile b/tools/power/x86/x86_energy_perf_policy/Makefile
index 971c9ffdcb50..a711eec0c895 100644
--- a/tools/power/x86/x86_energy_perf_policy/Makefile
+++ b/tools/power/x86/x86_energy_perf_policy/Makefile
@@ -1,10 +1,27 @@
-DESTDIR ?=
+CC		= $(CROSS_COMPILE)gcc
+BUILD_OUTPUT    := $(CURDIR)
+PREFIX		:= /usr
+DESTDIR		:=
+
+ifeq ("$(origin O)", "command line")
+	BUILD_OUTPUT := $(O)
+endif
 
 x86_energy_perf_policy : x86_energy_perf_policy.c
+CFLAGS +=	-Wall
+CFLAGS +=	-DMSRHEADER='"../../../../arch/x86/include/asm/msr-index.h"'
+
+%: %.c
+	@mkdir -p $(BUILD_OUTPUT)
+	$(CC) $(CFLAGS) $< -o $(BUILD_OUTPUT)/$@
 
+.PHONY : clean
 clean :
-	rm -f x86_energy_perf_policy
+	@rm -f $(BUILD_OUTPUT)/x86_energy_perf_policy
+
+install : x86_energy_perf_policy
+	install -d  $(DESTDIR)$(PREFIX)/bin
+	install $(BUILD_OUTPUT)/x86_energy_perf_policy $(DESTDIR)$(PREFIX)/bin/x86_energy_perf_policy
+	install -d  $(DESTDIR)$(PREFIX)/share/man/man8
+	install x86_energy_perf_policy.8 $(DESTDIR)$(PREFIX)/share/man/man8
 
-install :
-	install x86_energy_perf_policy ${DESTDIR}/usr/bin/
-	install x86_energy_perf_policy.8 ${DESTDIR}/usr/share/man/man8/
diff --git a/tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.8 b/tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.8
index 8eaaad648cdb..17db1c3af4d0 100644
--- a/tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.8
+++ b/tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.8
@@ -1,104 +1,213 @@
-.\"  This page Copyright (C) 2010 Len Brown <len.brown@intel.com>
+.\"  This page Copyright (C) 2010 - 2015 Len Brown <len.brown@intel.com>
 .\"  Distributed under the GPL, Copyleft 1994.
 .TH X86_ENERGY_PERF_POLICY 8
 .SH NAME
-x86_energy_perf_policy \- read or write MSR_IA32_ENERGY_PERF_BIAS
+x86_energy_perf_policy \- Manage Energy vs. Performance Policy via x86 Model Specific Registers
 .SH SYNOPSIS
-.ft B
 .B x86_energy_perf_policy
-.RB [ "\-c cpu" ]
-.RB [ "\-v" ]
-.RB "\-r"
+.RB "[ options ] [ scope ] [field \ value]"
 .br
-.B x86_energy_perf_policy
-.RB [ "\-c cpu" ]
-.RB [ "\-v" ]
-.RB 'performance'
+.RB "scope: \-\-cpu\ cpu-list | \-\-pkg\ pkg-list"
 .br
-.B x86_energy_perf_policy
-.RB [ "\-c cpu" ]
-.RB [ "\-v" ]
-.RB 'normal'
+.RB "cpu-list, pkg-list: # | #,# | #-# | all"
 .br
-.B x86_energy_perf_policy
-.RB [ "\-c cpu" ]
-.RB [ "\-v" ]
-.RB 'powersave'
+.RB "field: \-\-all | \-\-epb | \-\-hwp-epp | \-\-hwp-min | \-\-hwp-max | \-\-hwp-desired"
 .br
-.B x86_energy_perf_policy
-.RB [ "\-c cpu" ]
-.RB [ "\-v" ]
-.RB n
+.RB "other: (\-\-force | \-\-hwp-enable | \-\-turbo-enable)  value)"
 .br
+.RB "value: # | default | performance | balance-performance | balance-power | power"
 .SH DESCRIPTION
 \fBx86_energy_perf_policy\fP
-allows software to convey
-its policy for the relative importance of performance
-versus energy savings to the processor.
+displays and updates energy-performance policy settings specific to
+Intel Architecture Processors.  Settings are accessed via Model Specific Register (MSR)
+updates, no matter if the Linux cpufreq sub-system is enabled or not.
 
-The processor uses this information in model-specific ways
-when it must select trade-offs between performance and
-energy efficiency.
+Policy in MSR_IA32_ENERGY_PERF_BIAS (EPB)
+may affect a wide range of hardware decisions,
+such as how aggressively the hardware enters and exits CPU idle states (C-states)
+and Processor Performance States (P-states).
+This policy hint does not replace explicit OS C-state and P-state selection.
+Rather, it tells the hardware how aggressively to implement those selections.
+Further, it allows the OS to influence energy/performance trade-offs where there
+is no software interface, such as in the opportunistic "turbo-mode" P-state range.
+Note that MSR_IA32_ENERGY_PERF_BIAS is defined per CPU,
+but some implementations
+share a single MSR among all CPUs in each processor package.
+On those systems, a write to EPB on one processor will
+be visible, and will have an effect, on all CPUs
+in the same processor package.
 
-This policy hint does not supersede Processor Performance states
-(P-states) or CPU Idle power states (C-states), but allows
-software to have influence where it would otherwise be unable
-to express a preference.
+Hardware P-States (HWP) are effectively an expansion of hardware
+P-state control from the opportunistic turbo-mode P-state range
+to include the entire range of available P-states.
+On Broadwell Xeon, the initial HWP implementation, EBP influenced HWP.
+That influence was removed in subsequent generations,
+where it was moved to the
+Energy_Performance_Preference (EPP) field in
+a pair of dedicated MSRs -- MSR_IA32_HWP_REQUEST and MSR_IA32_HWP_REQUEST_PKG.
 
-For example, this setting may tell the hardware how
-aggressively or conservatively to control frequency
-in the "turbo range" above the explicitly OS-controlled
-P-state frequency range.  It may also tell the hardware
-how aggressively is should enter the OS requested C-states.
+EPP is the most commonly managed knob in HWP mode,
+but MSR_IA32_HWP_REQUEST also allows the user to specify
+minimum-frequency for Quality-of-Service,
+and maximum-frequency for power-capping.
+MSR_IA32_HWP_REQUEST is defined per-CPU.
 
-Support for this feature is indicated by CPUID.06H.ECX.bit3
-per the Intel Architectures Software Developer's Manual.
+MSR_IA32_HWP_REQUEST_PKG has the same capability as MSR_IA32_HWP_REQUEST,
+but it can simultaneously set the default policy for all CPUs within a package.
+A bit in per-CPU MSR_IA32_HWP_REQUEST indicates whether it is
+over-ruled-by or exempt-from MSR_IA32_HWP_REQUEST_PKG.
 
-.SS Options
-\fB-c\fP limits operation to a single CPU.
-The default is to operate on all CPUs.
-Note that MSR_IA32_ENERGY_PERF_BIAS is defined per
-logical processor, but that the initial implementations
-of the MSR were shared among all processors in each package.
-.PP
-\fB-v\fP increases verbosity.  By default
-x86_energy_perf_policy is silent.
-.PP
-\fB-r\fP is for "read-only" mode - the unchanged state
-is read and displayed.
+MSR_HWP_CAPABILITIES shows the default values for the fields
+in MSR_IA32_HWP_REQUEST.  It is displayed when no values
+are being written.
+
+.SS SCOPE OPTIONS
 .PP
-.I performance
-Set a policy where performance is paramount.
-The processor will be unwilling to sacrifice any performance
-for the sake of energy saving. This is the hardware default.
+\fB-c, --cpu\fP Operate on the MSR_IA32_HWP_REQUEST for each CPU in a CPU-list.
+The CPU-list may be comma-separated CPU numbers, with dash for range
+or the string "all".  Eg. '--cpu 1,4,6-8' or '--cpu all'.
+When --cpu is used, \fB--hwp-use-pkg\fP is available, which specifies whether the per-cpu
+MSR_IA32_HWP_REQUEST should be over-ruled by MSR_IA32_HWP_REQUEST_PKG (1),
+or exempt from MSR_IA32_HWP_REQUEST_PKG (0).
+
+\fB-p, --pkg\fP Operate on the MSR_IA32_HWP_REQUEST_PKG for each package in the package-list.
+The list is a string of individual package numbers separated
+by commas, and or ranges of package numbers separated by a dash,
+or the string "all".
+For example '--pkg 1,3' or '--pkg all'
+
+.SS VALUE OPTIONS
 .PP
-.I normal
+.I normal | default
 Set a policy with a normal balance between performance and energy efficiency.
 The processor will tolerate minor performance compromise
 for potentially significant energy savings.
-This reasonable default for most desktops and servers.
+This is a reasonable default for most desktops and servers.
+"default" is a synonym for "normal".
 .PP
-.I powersave
+.I performance
+Set a policy for maximum performance,
+accepting no performance sacrifice for the benefit of energy efficiency.
+.PP
+.I balance-performance
+Set a policy with a high priority on performance,
+but allowing some performance loss to benefit energy efficiency.
+.PP
+.I balance-power
+Set a policy where the performance and power are balanced.
+This is the default.
+.PP
+.I power
 Set a policy where the processor can accept
-a measurable performance hit to maximize energy efficiency.
+a measurable performance impact to maximize energy efficiency.
+
 .PP
-.I n
-Set MSR_IA32_ENERGY_PERF_BIAS to the specified number.
-The range of valid numbers is 0-15, where 0 is maximum
-performance and 15 is maximum energy efficiency.
+The following table shows the mapping from the value strings above to actual MSR values.
+This mapping is defined in the Linux-kernel header, msr-index.h.
 
+.nf
+VALUE STRING      	EPB	EPP
+performance       	0	0
+balance-performance	4	128
+normal, default		6	128
+balance-power	 	8	192
+power       		15	255
+.fi
+.PP
+For MSR_IA32_HWP_REQUEST performance fields
+(--hwp-min, --hwp-max, --hwp-desired), the value option
+is in units of 100 MHz, Eg. 12 signifies 1200 MHz.
+
+.SS FIELD OPTIONS
+\fB-a, --all value-string\fP Sets all EPB and EPP and HWP limit fields to the value associated with
+the value-string.  In addition, enables turbo-mode and HWP-mode, if they were previous disabled.
+Thus "--all normal" will set a system without cpufreq into a well known configuration.
+.PP
+\fB-B, --epb\fP set EPB per-core or per-package.
+See value strings in the table above.
+.PP
+\fB-d, --debug\fP debug increases verbosity.  By default
+x86_energy_perf_policy is silent for updates,
+and verbose for read-only mode.
+.PP
+\fB-P, --hwp-epp\fP set HWP.EPP per-core or per-package.
+See value strings in the table above.
+.PP
+\fB-m, --hwp-min\fP request HWP to not go below the specified core/bus ratio.
+The "default" is the value found in IA32_HWP_CAPABILITIES.min.
+.PP
+\fB-M, --hwp-max\fP request HWP not exceed a the specified core/bus ratio.
+The "default" is the value found in IA32_HWP_CAPABILITIES.max.
+.PP
+\fB-D, --hwp-desired\fP request HWP 'desired' frequency.
+The "normal" setting is 0, which
+corresponds to 'full autonomous' HWP control.
+Non-zero performance values request a specific performance
+level on this processor, specified in multiples of 100 MHz.
+.PP
+\fB-w, --hwp-window\fP specify integer number of microsec
+in the sliding window that HWP uses to maintain average frequency.
+This parameter is meaningful only when the "desired" field above is non-zero.
+Default is 0, allowing the HW to choose.
+.SH OTHER OPTIONS
+.PP
+\fB-f, --force\fP writes the specified values without bounds checking.
+.PP
+\fB-U, --hwp-use-pkg\fP (0 | 1), when used in conjunction with --cpu,
+indicates whether the per-CPU MSR_IA32_HWP_REQUEST should be overruled (1)
+or exempt (0) from per-Package MSR_IA32_HWP_REQUEST_PKG settings.
+The default is exempt.
+.PP
+\fB-H, --hwp-enable\fP enable HardWare-P-state (HWP) mode.  Once enabled, system RESET is required to disable HWP mode.
+.PP
+\fB-t, --turbo-enable\fP enable (1) or disable (0) turbo mode.
+.PP
+\fB-v, --version\fP print version and exit.
+.PP
+If no request to change policy is made,
+the default behavior is to read
+and display the current system state,
+including the default capabilities.
+.SH WARNING
+.PP
+This utility writes directly to Model Specific Registers.
+There is no locking or coordination should this utility
+be used to modify HWP limit fields at the same time that
+intel_pstate's sysfs attributes access the same MSRs.
+.PP
+Note that --hwp-desired and --hwp-window are considered experimental.
+Future versions of Linux reserve the right to access these
+fields internally -- potentially conflicting with user-space access.
+.SH EXAMPLE
+.nf
+# sudo x86_energy_perf_policy
+cpu0: EPB 6
+cpu0: HWP_REQ: min 6 max 35 des 0 epp 128 window 0x0 (0*10^0us) use_pkg 0
+cpu0: HWP_CAP: low 1 eff 8 guar 27 high 35
+cpu1: EPB 6
+cpu1: HWP_REQ: min 6 max 35 des 0 epp 128 window 0x0 (0*10^0us) use_pkg 0
+cpu1: HWP_CAP: low 1 eff 8 guar 27 high 35
+cpu2: EPB 6
+cpu2: HWP_REQ: min 6 max 35 des 0 epp 128 window 0x0 (0*10^0us) use_pkg 0
+cpu2: HWP_CAP: low 1 eff 8 guar 27 high 35
+cpu3: EPB 6
+cpu3: HWP_REQ: min 6 max 35 des 0 epp 128 window 0x0 (0*10^0us) use_pkg 0
+cpu3: HWP_CAP: low 1 eff 8 guar 27 high 35
+.fi
 .SH NOTES
-.B "x86_energy_perf_policy "
+.B "x86_energy_perf_policy"
 runs only as root.
 .SH FILES
 .ta
 .nf
 /dev/cpu/*/msr
 .fi
-
 .SH "SEE ALSO"
+.nf
 msr(4)
+Intel(R) 64 and IA-32 Architectures Software Developer's Manual
+.fi
 .PP
 .SH AUTHORS
 .nf
-Written by Len Brown <len.brown@intel.com>
+Len Brown
diff --git a/tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.c b/tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.c
index 40b3e5482f8a..65bbe627a425 100644
--- a/tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.c
+++ b/tools/power/x86/x86_energy_perf_policy/x86_energy_perf_policy.c
@@ -3,322 +3,1424 @@
  * policy preference bias on recent X86 processors.
  */
 /*
- * Copyright (c) 2010, Intel Corporation.
+ * Copyright (c) 2010 - 2017 Intel Corporation.
  * Len Brown <len.brown@intel.com>
  *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms and conditions of the GNU General Public License,
- * version 2, as published by the Free Software Foundation.
- *
- * This program is distributed in the hope it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
- * more details.
- *
- * You should have received a copy of the GNU General Public License along with
- * this program; if not, write to the Free Software Foundation, Inc.,
- * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ * This program is released under GPL v2
  */
 
+#define _GNU_SOURCE
+#include MSRHEADER
 #include <stdio.h>
 #include <unistd.h>
 #include <sys/types.h>
+#include <sched.h>
 #include <sys/stat.h>
 #include <sys/resource.h>
+#include <getopt.h>
+#include <err.h>
 #include <fcntl.h>
 #include <signal.h>
 #include <sys/time.h>
+#include <limits.h>
 #include <stdlib.h>
 #include <string.h>
+#include <cpuid.h>
+#include <errno.h>
+
+#define	OPTARG_NORMAL			(INT_MAX - 1)
+#define	OPTARG_POWER			(INT_MAX - 2)
+#define	OPTARG_BALANCE_POWER		(INT_MAX - 3)
+#define	OPTARG_BALANCE_PERFORMANCE	(INT_MAX - 4)
+#define	OPTARG_PERFORMANCE		(INT_MAX - 5)
+
+struct msr_hwp_cap {
+	unsigned char highest;
+	unsigned char guaranteed;
+	unsigned char efficient;
+	unsigned char lowest;
+};
 
-unsigned int verbose;		/* set with -v */
-unsigned int read_only;		/* set with -r */
+struct msr_hwp_request {
+	unsigned char hwp_min;
+	unsigned char hwp_max;
+	unsigned char hwp_desired;
+	unsigned char hwp_epp;
+	unsigned int hwp_window;
+	unsigned char hwp_use_pkg;
+} req_update;
+
+unsigned int debug;
+unsigned int verbose;
+unsigned int force;
 char *progname;
-unsigned long long new_bias;
-int cpu = -1;
+int base_cpu;
+unsigned char update_epb;
+unsigned long long new_epb;
+unsigned char turbo_is_enabled;
+unsigned char update_turbo;
+unsigned char turbo_update_value;
+unsigned char update_hwp_epp;
+unsigned char update_hwp_min;
+unsigned char update_hwp_max;
+unsigned char update_hwp_desired;
+unsigned char update_hwp_window;
+unsigned char update_hwp_use_pkg;
+unsigned char update_hwp_enable;
+#define hwp_update_enabled() (update_hwp_enable | update_hwp_epp | update_hwp_max | update_hwp_min | update_hwp_desired | update_hwp_window | update_hwp_use_pkg)
+int max_cpu_num;
+int max_pkg_num;
+#define MAX_PACKAGES 64
+unsigned int first_cpu_in_pkg[MAX_PACKAGES];
+unsigned long long pkg_present_set;
+unsigned long long pkg_selected_set;
+cpu_set_t *cpu_present_set;
+cpu_set_t *cpu_selected_set;
+int genuine_intel;
+
+size_t cpu_setsize;
+
+char *proc_stat = "/proc/stat";
+
+unsigned int has_epb;	/* MSR_IA32_ENERGY_PERF_BIAS */
+unsigned int has_hwp;	/* IA32_PM_ENABLE, IA32_HWP_CAPABILITIES */
+			/* IA32_HWP_REQUEST, IA32_HWP_STATUS */
+unsigned int has_hwp_notify;		/* IA32_HWP_INTERRUPT */
+unsigned int has_hwp_activity_window;	/* IA32_HWP_REQUEST[bits 41:32] */
+unsigned int has_hwp_epp;	/* IA32_HWP_REQUEST[bits 31:24] */
+unsigned int has_hwp_request_pkg;	/* IA32_HWP_REQUEST_PKG */
+
+unsigned int bdx_highest_ratio;
 
 /*
- * Usage:
- *
- * -c cpu: limit action to a single CPU (default is all CPUs)
- * -v: verbose output (can invoke more than once)
- * -r: read-only, don't change any settings
- *
- *  performance
- *	Performance is paramount.
- *	Unwilling to sacrifice any performance
- *	for the sake of energy saving. (hardware default)
- *
- *  normal
- *	Can tolerate minor performance compromise
- *	for potentially significant energy savings.
- *	(reasonable default for most desktops and servers)
- *
- *  powersave
- *	Can tolerate significant performance hit
- *	to maximize energy savings.
- *
- * n
- *	a numerical value to write to the underlying MSR.
+ * maintain compatibility with original implementation, but don't document it:
  */
 void usage(void)
 {
-	printf("%s: [-c cpu] [-v] "
-		"(-r | 'performance' | 'normal' | 'powersave' | n)\n",
-		progname);
+	fprintf(stderr, "%s [options] [scope][field value]\n", progname);
+	fprintf(stderr, "scope: --cpu cpu-list [--hwp-use-pkg #] | --pkg pkg-list\n");
+	fprintf(stderr, "field: --all | --epb | --hwp-epp | --hwp-min | --hwp-max | --hwp-desired\n");
+	fprintf(stderr, "other: --hwp-enable | --turbo-enable (0 | 1) | --help | --force\n");
+	fprintf(stderr,
+		"value: ( # | \"normal\" | \"performance\" | \"balance-performance\" | \"balance-power\"| \"power\")\n");
+	fprintf(stderr, "--hwp-window usec\n");
+
+	fprintf(stderr, "Specify only Energy Performance BIAS (legacy usage):\n");
+	fprintf(stderr, "%s: [-c cpu] [-v] (-r | policy-value )\n", progname);
+
 	exit(1);
 }
 
-#define MSR_IA32_ENERGY_PERF_BIAS	0x000001b0
+/*
+ * If bdx_highest_ratio is set,
+ * then we must translate between MSR format and simple ratio
+ * used on the cmdline.
+ */
+int ratio_2_msr_perf(int ratio)
+{
+	int msr_perf;
+
+	if (!bdx_highest_ratio)
+		return ratio;
+
+	msr_perf = ratio * 255 / bdx_highest_ratio;
+
+	if (debug)
+		fprintf(stderr, "%d = ratio_to_msr_perf(%d)\n", msr_perf, ratio);
+
+	return msr_perf;
+}
+int msr_perf_2_ratio(int msr_perf)
+{
+	int ratio;
+	double d;
+
+	if (!bdx_highest_ratio)
+		return msr_perf;
+
+	d = (double)msr_perf * (double) bdx_highest_ratio / 255.0;
+	d = d + 0.5;	/* round */
+	ratio = (int)d;
+
+	if (debug)
+		fprintf(stderr, "%d = msr_perf_ratio(%d) {%f}\n", ratio, msr_perf, d);
+
+	return ratio;
+}
+int parse_cmdline_epb(int i)
+{
+	if (!has_epb)
+		errx(1, "EPB not enabled on this platform");
+
+	update_epb = 1;
+
+	switch (i) {
+	case OPTARG_POWER:
+		return ENERGY_PERF_BIAS_POWERSAVE;
+	case OPTARG_BALANCE_POWER:
+		return ENERGY_PERF_BIAS_BALANCE_POWERSAVE;
+	case OPTARG_NORMAL:
+		return ENERGY_PERF_BIAS_NORMAL;
+	case OPTARG_BALANCE_PERFORMANCE:
+		return ENERGY_PERF_BIAS_BALANCE_PERFORMANCE;
+	case OPTARG_PERFORMANCE:
+		return ENERGY_PERF_BIAS_PERFORMANCE;
+	}
+	if (i < 0 || i > ENERGY_PERF_BIAS_POWERSAVE)
+		errx(1, "--epb must be from 0 to 15");
+	return i;
+}
+
+#define HWP_CAP_LOWEST 0
+#define HWP_CAP_HIGHEST 255
+
+/*
+ * "performance" changes hwp_min to cap.highest
+ * All others leave it at cap.lowest
+ */
+int parse_cmdline_hwp_min(int i)
+{
+	update_hwp_min = 1;
+
+	switch (i) {
+	case OPTARG_POWER:
+	case OPTARG_BALANCE_POWER:
+	case OPTARG_NORMAL:
+	case OPTARG_BALANCE_PERFORMANCE:
+		return HWP_CAP_LOWEST;
+	case OPTARG_PERFORMANCE:
+		return HWP_CAP_HIGHEST;
+	}
+	return i;
+}
+/*
+ * "power" changes hwp_max to cap.lowest
+ * All others leave it at cap.highest
+ */
+int parse_cmdline_hwp_max(int i)
+{
+	update_hwp_max = 1;
+
+	switch (i) {
+	case OPTARG_POWER:
+		return HWP_CAP_LOWEST;
+	case OPTARG_NORMAL:
+	case OPTARG_BALANCE_POWER:
+	case OPTARG_BALANCE_PERFORMANCE:
+	case OPTARG_PERFORMANCE:
+		return HWP_CAP_HIGHEST;
+	}
+	return i;
+}
+/*
+ * for --hwp-des, all strings leave it in autonomous mode
+ * If you want to change it, you need to explicitly pick a value
+ */
+int parse_cmdline_hwp_desired(int i)
+{
+	update_hwp_desired = 1;
+
+	switch (i) {
+	case OPTARG_POWER:
+	case OPTARG_BALANCE_POWER:
+	case OPTARG_BALANCE_PERFORMANCE:
+	case OPTARG_NORMAL:
+	case OPTARG_PERFORMANCE:
+		return 0;	/* autonomous */
+	}
+	return i;
+}
+
+int parse_cmdline_hwp_window(int i)
+{
+	unsigned int exponent;
+
+	update_hwp_window = 1;
+
+	switch (i) {
+	case OPTARG_POWER:
+	case OPTARG_BALANCE_POWER:
+	case OPTARG_NORMAL:
+	case OPTARG_BALANCE_PERFORMANCE:
+	case OPTARG_PERFORMANCE:
+		return 0;
+	}
+	if (i < 0 || i > 1270000000) {
+		fprintf(stderr, "--hwp-window: 0 for auto; 1 - 1270000000 usec for window duration\n");
+		usage();
+	}
+	for (exponent = 0; ; ++exponent) {
+		if (debug)
+			printf("%d 10^%d\n", i, exponent);
+
+		if (i <= 127)
+			break;
+
+		i = i / 10;
+	}
+	if (debug)
+		fprintf(stderr, "%d*10^%d: 0x%x\n", i, exponent, (exponent << 7) | i);
+
+	return (exponent << 7) | i;
+}
+int parse_cmdline_hwp_epp(int i)
+{
+	update_hwp_epp = 1;
+
+	switch (i) {
+	case OPTARG_POWER:
+		return HWP_EPP_POWERSAVE;
+	case OPTARG_BALANCE_POWER:
+		return HWP_EPP_BALANCE_POWERSAVE;
+	case OPTARG_NORMAL:
+	case OPTARG_BALANCE_PERFORMANCE:
+		return HWP_EPP_BALANCE_PERFORMANCE;
+	case OPTARG_PERFORMANCE:
+		return HWP_EPP_PERFORMANCE;
+	}
+	if (i < 0 || i > 0xff) {
+		fprintf(stderr, "--hwp-epp must be from 0 to 0xff\n");
+		usage();
+	}
+	return i;
+}
+int parse_cmdline_turbo(int i)
+{
+	update_turbo = 1;
+
+	switch (i) {
+	case OPTARG_POWER:
+		return 0;
+	case OPTARG_NORMAL:
+	case OPTARG_BALANCE_POWER:
+	case OPTARG_BALANCE_PERFORMANCE:
+	case OPTARG_PERFORMANCE:
+		return 1;
+	}
+	if (i < 0 || i > 1) {
+		fprintf(stderr, "--turbo-enable: 1 to enable, 0 to disable\n");
+		usage();
+	}
+	return i;
+}
+
+int parse_optarg_string(char *s)
+{
+	int i;
+	char *endptr;
+
+	if (!strncmp(s, "default", 7))
+		return OPTARG_NORMAL;
+
+	if (!strncmp(s, "normal", 6))
+		return OPTARG_NORMAL;
+
+	if (!strncmp(s, "power", 9))
+		return OPTARG_POWER;
+
+	if (!strncmp(s, "balance-power", 17))
+		return OPTARG_BALANCE_POWER;
+
+	if (!strncmp(s, "balance-performance", 19))
+		return OPTARG_BALANCE_PERFORMANCE;
+
+	if (!strncmp(s, "performance", 11))
+		return OPTARG_PERFORMANCE;
+
+	i = strtol(s, &endptr, 0);
+	if (s == endptr) {
+		fprintf(stderr, "no digits in \"%s\"\n", s);
+		usage();
+	}
+	if (i == LONG_MIN || i == LONG_MAX)
+		errx(-1, "%s", s);
+
+	if (i > 0xFF)
+		errx(-1, "%d (0x%x) must be < 256", i, i);
+
+	if (i < 0)
+		errx(-1, "%d (0x%x) must be >= 0", i, i);
+	return i;
+}
+
+void parse_cmdline_all(char *s)
+{
+	force++;
+	update_hwp_enable = 1;
+	req_update.hwp_min = parse_cmdline_hwp_min(parse_optarg_string(s));
+	req_update.hwp_max = parse_cmdline_hwp_max(parse_optarg_string(s));
+	req_update.hwp_epp = parse_cmdline_hwp_epp(parse_optarg_string(s));
+	if (has_epb)
+		new_epb = parse_cmdline_epb(parse_optarg_string(s));
+	turbo_update_value = parse_cmdline_turbo(parse_optarg_string(s));
+	req_update.hwp_desired = parse_cmdline_hwp_desired(parse_optarg_string(s));
+	req_update.hwp_window = parse_cmdline_hwp_window(parse_optarg_string(s));
+}
+
+void validate_cpu_selected_set(void)
+{
+	int cpu;
+
+	if (CPU_COUNT_S(cpu_setsize, cpu_selected_set) == 0)
+		errx(0, "no CPUs requested");
+
+	for (cpu = 0; cpu <= max_cpu_num; ++cpu) {
+		if (CPU_ISSET_S(cpu, cpu_setsize, cpu_selected_set))
+			if (!CPU_ISSET_S(cpu, cpu_setsize, cpu_present_set))
+				errx(1, "Requested cpu% is not present", cpu);
+	}
+}
+
+void parse_cmdline_cpu(char *s)
+{
+	char *startp, *endp;
+	int cpu = 0;
+
+	if (pkg_selected_set) {
+		usage();
+		errx(1, "--cpu | --pkg");
+	}
+	cpu_selected_set = CPU_ALLOC((max_cpu_num + 1));
+	if (cpu_selected_set == NULL)
+		err(1, "cpu_selected_set");
+	CPU_ZERO_S(cpu_setsize, cpu_selected_set);
+
+	for (startp = s; startp && *startp;) {
+
+		if (*startp == ',') {
+			startp++;
+			continue;
+		}
+
+		if (*startp == '-') {
+			int end_cpu;
 
-#define	BIAS_PERFORMANCE		0
-#define BIAS_BALANCE			6
-#define	BIAS_POWERSAVE			15
+			startp++;
+			end_cpu = strtol(startp, &endp, 10);
+			if (startp == endp)
+				continue;
+
+			while (cpu <= end_cpu) {
+				if (cpu > max_cpu_num)
+					errx(1, "Requested cpu%d exceeds max cpu%d", cpu, max_cpu_num);
+				CPU_SET_S(cpu, cpu_setsize, cpu_selected_set);
+				cpu++;
+			}
+			startp = endp;
+			continue;
+		}
+
+		if (strncmp(startp, "all", 3) == 0) {
+			for (cpu = 0; cpu <= max_cpu_num; cpu += 1) {
+				if (CPU_ISSET_S(cpu, cpu_setsize, cpu_present_set))
+					CPU_SET_S(cpu, cpu_setsize, cpu_selected_set);
+			}
+			startp += 3;
+			if (*startp == 0)
+				break;
+		}
+		/* "--cpu even" is not documented */
+		if (strncmp(startp, "even", 4) == 0) {
+			for (cpu = 0; cpu <= max_cpu_num; cpu += 2) {
+				if (CPU_ISSET_S(cpu, cpu_setsize, cpu_present_set))
+					CPU_SET_S(cpu, cpu_setsize, cpu_selected_set);
+			}
+			startp += 4;
+			if (*startp == 0)
+				break;
+		}
+
+		/* "--cpu odd" is not documented */
+		if (strncmp(startp, "odd", 3) == 0) {
+			for (cpu = 1; cpu <= max_cpu_num; cpu += 2) {
+				if (CPU_ISSET_S(cpu, cpu_setsize, cpu_present_set))
+					CPU_SET_S(cpu, cpu_setsize, cpu_selected_set);
+			}
+			startp += 3;
+			if (*startp == 0)
+				break;
+		}
+
+		cpu = strtol(startp, &endp, 10);
+		if (startp == endp)
+			errx(1, "--cpu cpu-set: confused by '%s'", startp);
+		if (cpu > max_cpu_num)
+			errx(1, "Requested cpu%d exceeds max cpu%d", cpu, max_cpu_num);
+		CPU_SET_S(cpu, cpu_setsize, cpu_selected_set);
+		startp = endp;
+	}
+
+	validate_cpu_selected_set();
+
+}
+
+void parse_cmdline_pkg(char *s)
+{
+	char *startp, *endp;
+	int pkg = 0;
+
+	if (cpu_selected_set) {
+		usage();
+		errx(1, "--pkg | --cpu");
+	}
+	pkg_selected_set = 0;
+
+	for (startp = s; startp && *startp;) {
+
+		if (*startp == ',') {
+			startp++;
+			continue;
+		}
+
+		if (*startp == '-') {
+			int end_pkg;
+
+			startp++;
+			end_pkg = strtol(startp, &endp, 10);
+			if (startp == endp)
+				continue;
+
+			while (pkg <= end_pkg) {
+				if (pkg > max_pkg_num)
+					errx(1, "Requested pkg%d exceeds max pkg%d", pkg, max_pkg_num);
+				pkg_selected_set |= 1 << pkg;
+				pkg++;
+			}
+			startp = endp;
+			continue;
+		}
+
+		if (strncmp(startp, "all", 3) == 0) {
+			pkg_selected_set = pkg_present_set;
+			return;
+		}
+
+		pkg = strtol(startp, &endp, 10);
+		if (pkg > max_pkg_num)
+			errx(1, "Requested pkg%d Exceeds max pkg%d", pkg, max_pkg_num);
+		pkg_selected_set |= 1 << pkg;
+		startp = endp;
+	}
+}
+
+void for_packages(unsigned long long pkg_set, int (func)(int))
+{
+	int pkg_num;
+
+	for (pkg_num = 0; pkg_num <= max_pkg_num; ++pkg_num) {
+		if (pkg_set & (1UL << pkg_num))
+			func(pkg_num);
+	}
+}
+
+void print_version(void)
+{
+	printf("x86_energy_perf_policy 17.05.11 (C) Len Brown <len.brown@intel.com>\n");
+}
 
 void cmdline(int argc, char **argv)
 {
 	int opt;
+	int option_index = 0;
+
+	static struct option long_options[] = {
+		{"all",		required_argument,	0, 'a'},
+		{"cpu",		required_argument,	0, 'c'},
+		{"pkg",		required_argument,	0, 'p'},
+		{"debug",	no_argument,		0, 'd'},
+		{"hwp-desired",	required_argument,	0, 'D'},
+		{"epb",	required_argument,	0, 'B'},
+		{"force",	no_argument,	0, 'f'},
+		{"hwp-enable",	no_argument,	0, 'e'},
+		{"help",	no_argument,	0, 'h'},
+		{"hwp-epp",	required_argument,	0, 'P'},
+		{"hwp-min",	required_argument,	0, 'm'},
+		{"hwp-max",	required_argument,	0, 'M'},
+		{"read",	no_argument,		0, 'r'},
+		{"turbo-enable",	required_argument,	0, 't'},
+		{"hwp-use-pkg",	required_argument,	0, 'u'},
+		{"version",	no_argument,		0, 'v'},
+		{"hwp-window",	required_argument,	0, 'w'},
+		{0,		0,			0, 0 }
+	};
 
 	progname = argv[0];
 
-	while ((opt = getopt(argc, argv, "+rvc:")) != -1) {
+	while ((opt = getopt_long_only(argc, argv, "+a:c:dD:E:e:f:m:M:rt:u:vw",
+				long_options, &option_index)) != -1) {
 		switch (opt) {
+		case 'a':
+			parse_cmdline_all(optarg);
+			break;
+		case 'B':
+			new_epb = parse_cmdline_epb(parse_optarg_string(optarg));
+			break;
 		case 'c':
-			cpu = atoi(optarg);
+			parse_cmdline_cpu(optarg);
+			break;
+		case 'e':
+			update_hwp_enable = 1;
+			break;
+		case 'h':
+			usage();
+			break;
+		case 'd':
+			debug++;
+			verbose++;
+			break;
+		case 'f':
+			force++;
+			break;
+		case 'D':
+			req_update.hwp_desired = parse_cmdline_hwp_desired(parse_optarg_string(optarg));
+			break;
+		case 'm':
+			req_update.hwp_min = parse_cmdline_hwp_min(parse_optarg_string(optarg));
+			break;
+		case 'M':
+			req_update.hwp_max = parse_cmdline_hwp_max(parse_optarg_string(optarg));
+			break;
+		case 'p':
+			parse_cmdline_pkg(optarg);
+			break;
+		case 'P':
+			req_update.hwp_epp = parse_cmdline_hwp_epp(parse_optarg_string(optarg));
 			break;
 		case 'r':
-			read_only = 1;
+			/* v1 used -r to specify read-only mode, now the default */
+			break;
+		case 't':
+			turbo_update_value = parse_cmdline_turbo(parse_optarg_string(optarg));
+			break;
+		case 'u':
+			update_hwp_use_pkg++;
+			if (atoi(optarg) == 0)
+				req_update.hwp_use_pkg = 0;
+			else
+				req_update.hwp_use_pkg = 1;
 			break;
 		case 'v':
-			verbose++;
+			print_version();
+			exit(0);
+			break;
+		case 'w':
+			req_update.hwp_window = parse_cmdline_hwp_window(parse_optarg_string(optarg));
 			break;
 		default:
 			usage();
 		}
 	}
-	/* if -r, then should be no additional optind */
-	if (read_only && (argc > optind))
-		usage();
-
 	/*
-	 * if no -r , then must be one additional optind
+	 * v1 allowed "performance"|"normal"|"power" with no policy specifier
+	 * to update BIAS.  Continue to support that, even though no longer documented.
 	 */
-	if (!read_only) {
+	if (argc == optind + 1)
+		new_epb = parse_cmdline_epb(parse_optarg_string(argv[optind]));
 
-		if (argc != optind + 1) {
-			printf("must supply -r or policy param\n");
-			usage();
-			}
+	if (argc > optind + 1) {
+		fprintf(stderr, "stray parameter '%s'\n", argv[optind + 1]);
+		usage();
+	}
+}
 
-		if (!strcmp("performance", argv[optind])) {
-			new_bias = BIAS_PERFORMANCE;
-		} else if (!strcmp("normal", argv[optind])) {
-			new_bias = BIAS_BALANCE;
-		} else if (!strcmp("powersave", argv[optind])) {
-			new_bias = BIAS_POWERSAVE;
-		} else {
-			char *endptr;
-
-			new_bias = strtoull(argv[optind], &endptr, 0);
-			if (endptr == argv[optind] ||
-				new_bias > BIAS_POWERSAVE) {
-					fprintf(stderr, "invalid value: %s\n",
-						argv[optind]);
-				usage();
-			}
-		}
+
+int get_msr(int cpu, int offset, unsigned long long *msr)
+{
+	int retval;
+	char pathname[32];
+	int fd;
+
+	sprintf(pathname, "/dev/cpu/%d/msr", cpu);
+	fd = open(pathname, O_RDONLY);
+	if (fd < 0)
+		err(-1, "%s open failed, try chown or chmod +r /dev/cpu/*/msr, or run as root", pathname);
+
+	retval = pread(fd, msr, sizeof(*msr), offset);
+	if (retval != sizeof(*msr))
+		err(-1, "%s offset 0x%llx read failed", pathname, (unsigned long long)offset);
+
+	if (debug > 1)
+		fprintf(stderr, "get_msr(cpu%d, 0x%X, 0x%llX)\n", cpu, offset, *msr);
+
+	close(fd);
+	return 0;
+}
+
+int put_msr(int cpu, int offset, unsigned long long new_msr)
+{
+	char pathname[32];
+	int retval;
+	int fd;
+
+	sprintf(pathname, "/dev/cpu/%d/msr", cpu);
+	fd = open(pathname, O_RDWR);
+	if (fd < 0)
+		err(-1, "%s open failed, try chown or chmod +r /dev/cpu/*/msr, or run as root", pathname);
+
+	retval = pwrite(fd, &new_msr, sizeof(new_msr), offset);
+	if (retval != sizeof(new_msr))
+		err(-2, "pwrite(cpu%d, offset 0x%x, 0x%llx) = %d", cpu, offset, new_msr, retval);
+
+	close(fd);
+
+	if (debug > 1)
+		fprintf(stderr, "put_msr(cpu%d, 0x%X, 0x%llX)\n", cpu, offset, new_msr);
+
+	return 0;
+}
+
+void print_hwp_cap(int cpu, struct msr_hwp_cap *cap, char *str)
+{
+	if (cpu != -1)
+		printf("cpu%d: ", cpu);
+
+	printf("HWP_CAP: low %d eff %d guar %d high %d\n",
+		cap->lowest, cap->efficient, cap->guaranteed, cap->highest);
+}
+void read_hwp_cap(int cpu, struct msr_hwp_cap *cap, unsigned int msr_offset)
+{
+	unsigned long long msr;
+
+	get_msr(cpu, msr_offset, &msr);
+
+	cap->highest = msr_perf_2_ratio(HWP_HIGHEST_PERF(msr));
+	cap->guaranteed = msr_perf_2_ratio(HWP_GUARANTEED_PERF(msr));
+	cap->efficient = msr_perf_2_ratio(HWP_MOSTEFFICIENT_PERF(msr));
+	cap->lowest = msr_perf_2_ratio(HWP_LOWEST_PERF(msr));
+}
+
+void print_hwp_request(int cpu, struct msr_hwp_request *h, char *str)
+{
+	if (cpu != -1)
+		printf("cpu%d: ", cpu);
+
+	if (str)
+		printf("%s", str);
+
+	printf("HWP_REQ: min %d max %d des %d epp %d window 0x%x (%d*10^%dus) use_pkg %d\n",
+		h->hwp_min, h->hwp_max, h->hwp_desired, h->hwp_epp,
+		h->hwp_window, h->hwp_window & 0x7F, (h->hwp_window >> 7) & 0x7, h->hwp_use_pkg);
+}
+void print_hwp_request_pkg(int pkg, struct msr_hwp_request *h, char *str)
+{
+	printf("pkg%d: ", pkg);
+
+	if (str)
+		printf("%s", str);
+
+	printf("HWP_REQ_PKG: min %d max %d des %d epp %d window 0x%x (%d*10^%dus)\n",
+		h->hwp_min, h->hwp_max, h->hwp_desired, h->hwp_epp,
+		h->hwp_window, h->hwp_window & 0x7F, (h->hwp_window >> 7) & 0x7);
+}
+void read_hwp_request(int cpu, struct msr_hwp_request *hwp_req, unsigned int msr_offset)
+{
+	unsigned long long msr;
+
+	get_msr(cpu, msr_offset, &msr);
+
+	hwp_req->hwp_min = msr_perf_2_ratio((((msr) >> 0) & 0xff));
+	hwp_req->hwp_max = msr_perf_2_ratio((((msr) >> 8) & 0xff));
+	hwp_req->hwp_desired = msr_perf_2_ratio((((msr) >> 16) & 0xff));
+	hwp_req->hwp_epp = (((msr) >> 24) & 0xff);
+	hwp_req->hwp_window = (((msr) >> 32) & 0x3ff);
+	hwp_req->hwp_use_pkg = (((msr) >> 42) & 0x1);
+}
+
+void write_hwp_request(int cpu, struct msr_hwp_request *hwp_req, unsigned int msr_offset)
+{
+	unsigned long long msr = 0;
+
+	if (debug > 1)
+		printf("cpu%d: requesting min %d max %d des %d epp %d window 0x%0x use_pkg %d\n",
+			cpu, hwp_req->hwp_min, hwp_req->hwp_max,
+			hwp_req->hwp_desired, hwp_req->hwp_epp,
+			hwp_req->hwp_window, hwp_req->hwp_use_pkg);
+
+	msr |= HWP_MIN_PERF(ratio_2_msr_perf(hwp_req->hwp_min));
+	msr |= HWP_MAX_PERF(ratio_2_msr_perf(hwp_req->hwp_max));
+	msr |= HWP_DESIRED_PERF(ratio_2_msr_perf(hwp_req->hwp_desired));
+	msr |= HWP_ENERGY_PERF_PREFERENCE(hwp_req->hwp_epp);
+	msr |= HWP_ACTIVITY_WINDOW(hwp_req->hwp_window);
+	msr |= HWP_PACKAGE_CONTROL(hwp_req->hwp_use_pkg);
+
+	put_msr(cpu, msr_offset, msr);
+}
+
+int print_cpu_msrs(int cpu)
+{
+	unsigned long long msr;
+	struct msr_hwp_request req;
+	struct msr_hwp_cap cap;
+
+	if (has_epb) {
+		get_msr(cpu, MSR_IA32_ENERGY_PERF_BIAS, &msr);
+
+		printf("cpu%d: EPB %u\n", cpu, (unsigned int) msr);
 	}
+
+	if (!has_hwp)
+		return 0;
+
+	read_hwp_request(cpu, &req, MSR_HWP_REQUEST);
+	print_hwp_request(cpu, &req, "");
+
+	read_hwp_cap(cpu, &cap, MSR_HWP_CAPABILITIES);
+	print_hwp_cap(cpu, &cap, "");
+
+	return 0;
+}
+
+int print_pkg_msrs(int pkg)
+{
+	struct msr_hwp_request req;
+	unsigned long long msr;
+
+	if (!has_hwp)
+		return 0;
+
+	read_hwp_request(first_cpu_in_pkg[pkg], &req, MSR_HWP_REQUEST_PKG);
+	print_hwp_request_pkg(pkg, &req, "");
+
+	if (has_hwp_notify) {
+		get_msr(first_cpu_in_pkg[pkg], MSR_HWP_INTERRUPT, &msr);
+		fprintf(stderr,
+		"pkg%d: MSR_HWP_INTERRUPT: 0x%08llx (Excursion_Min-%sabled, Guaranteed_Perf_Change-%sabled)\n",
+		pkg, msr,
+		((msr) & 0x2) ? "EN" : "Dis",
+		((msr) & 0x1) ? "EN" : "Dis");
+	}
+	get_msr(first_cpu_in_pkg[pkg], MSR_HWP_STATUS, &msr);
+	fprintf(stderr,
+		"pkg%d: MSR_HWP_STATUS: 0x%08llx (%sExcursion_Min, %sGuaranteed_Perf_Change)\n",
+		pkg, msr,
+		((msr) & 0x4) ? "" : "No-",
+		((msr) & 0x1) ? "" : "No-");
+
+	return 0;
 }
 
 /*
- * validate_cpuid()
- * returns on success, quietly exits on failure (make verbose with -v)
+ * Assumption: All HWP systems have 100 MHz bus clock
  */
-void validate_cpuid(void)
+int ratio_2_sysfs_khz(int ratio)
 {
-	unsigned int eax, ebx, ecx, edx, max_level;
-	unsigned int fms, family, model, stepping;
+	int bclk_khz = 100 * 1000;	/* 100,000 KHz = 100 MHz */
 
-	eax = ebx = ecx = edx = 0;
+	return ratio * bclk_khz;
+}
+/*
+ * If HWP is enabled and cpufreq sysfs attribtes are present,
+ * then update sysfs, so that it will not become
+ * stale when we write to MSRs.
+ * (intel_pstate's max_perf_pct and min_perf_pct will follow cpufreq,
+ *  so we don't have to touch that.)
+ */
+void update_cpufreq_scaling_freq(int is_max, int cpu, unsigned int ratio)
+{
+	char pathname[64];
+	FILE *fp;
+	int retval;
+	int khz;
 
-	asm("cpuid" : "=a" (max_level), "=b" (ebx), "=c" (ecx),
-		"=d" (edx) : "a" (0));
+	sprintf(pathname, "/sys/devices/system/cpu/cpu%d/cpufreq/scaling_%s_freq",
+		cpu, is_max ? "max" : "min");
 
-	if (ebx != 0x756e6547 || edx != 0x49656e69 || ecx != 0x6c65746e) {
-		if (verbose)
-			fprintf(stderr, "%.4s%.4s%.4s != GenuineIntel",
-				(char *)&ebx, (char *)&edx, (char *)&ecx);
-		exit(1);
+	fp = fopen(pathname, "w");
+	if (!fp) {
+		if (debug)
+			perror(pathname);
+		return;
 	}
 
-	asm("cpuid" : "=a" (fms), "=c" (ecx), "=d" (edx) : "a" (1) : "ebx");
-	family = (fms >> 8) & 0xf;
-	model = (fms >> 4) & 0xf;
-	stepping = fms & 0xf;
-	if (family == 6 || family == 0xf)
-		model += ((fms >> 16) & 0xf) << 4;
+	khz = ratio_2_sysfs_khz(ratio);
+	retval = fprintf(fp, "%d", khz);
+	if (retval < 0)
+		if (debug)
+			perror("fprintf");
+	if (debug)
+		printf("echo %d > %s\n", khz, pathname);
 
-	if (verbose > 1)
-		printf("CPUID %d levels family:model:stepping "
-			"0x%x:%x:%x (%d:%d:%d)\n", max_level,
-			family, model, stepping, family, model, stepping);
+	fclose(fp);
+}
 
-	if (!(edx & (1 << 5))) {
-		if (verbose)
-			printf("CPUID: no MSR\n");
-		exit(1);
+/*
+ * We update all sysfs before updating any MSRs because of
+ * bugs in cpufreq/intel_pstate where the sysfs writes
+ * for a CPU may change the min/max values on other CPUS.
+ */
+
+int update_sysfs(int cpu)
+{
+	if (!has_hwp)
+		return 0;
+
+	if (!hwp_update_enabled())
+		return 0;
+
+	if (access("/sys/devices/system/cpu/cpu0/cpufreq", F_OK))
+		return 0;
+
+	if (update_hwp_min)
+		update_cpufreq_scaling_freq(0, cpu, req_update.hwp_min);
+
+	if (update_hwp_max)
+		update_cpufreq_scaling_freq(1, cpu, req_update.hwp_max);
+
+	return 0;
+}
+
+int verify_hwp_req_self_consistency(int cpu, struct msr_hwp_request *req)
+{
+	/* fail if min > max requested */
+	if (req->hwp_min > req->hwp_max) {
+		errx(1, "cpu%d: requested hwp-min %d > hwp_max %d",
+			cpu, req->hwp_min, req->hwp_max);
 	}
 
-	/*
-	 * Support for MSR_IA32_ENERGY_PERF_BIAS
-	 * is indicated by CPUID.06H.ECX.bit3
-	 */
-	asm("cpuid" : "=a" (eax), "=b" (ebx), "=c" (ecx), "=d" (edx) : "a" (6));
-	if (verbose)
-		printf("CPUID.06H.ECX: 0x%x\n", ecx);
-	if (!(ecx & (1 << 3))) {
-		if (verbose)
-			printf("CPUID: No MSR_IA32_ENERGY_PERF_BIAS\n");
-		exit(1);
+	/* fail if desired > max requestd */
+	if (req->hwp_desired && (req->hwp_desired > req->hwp_max)) {
+		errx(1, "cpu%d: requested hwp-desired %d > hwp_max %d",
+			cpu, req->hwp_desired, req->hwp_max);
 	}
-	return;	/* success */
+	/* fail if desired < min requestd */
+	if (req->hwp_desired && (req->hwp_desired < req->hwp_min)) {
+		errx(1, "cpu%d: requested hwp-desired %d < requested hwp_min %d",
+			cpu, req->hwp_desired, req->hwp_min);
+	}
+
+	return 0;
 }
 
-unsigned long long get_msr(int cpu, int offset)
+int check_hwp_request_v_hwp_capabilities(int cpu, struct msr_hwp_request *req, struct msr_hwp_cap *cap)
 {
-	unsigned long long msr;
-	char msr_path[32];
-	int retval;
-	int fd;
+	if (update_hwp_max) {
+		if (req->hwp_max > cap->highest)
+			errx(1, "cpu%d: requested max %d > capabilities highest %d, use --force?",
+				cpu, req->hwp_max, cap->highest);
+		if (req->hwp_max < cap->lowest)
+			errx(1, "cpu%d: requested max %d < capabilities lowest %d, use --force?",
+				cpu, req->hwp_max, cap->lowest);
+	}
 
-	sprintf(msr_path, "/dev/cpu/%d/msr", cpu);
-	fd = open(msr_path, O_RDONLY);
-	if (fd < 0) {
-		printf("Try \"# modprobe msr\"\n");
-		perror(msr_path);
-		exit(1);
+	if (update_hwp_min) {
+		if (req->hwp_min > cap->highest)
+			errx(1, "cpu%d: requested min %d > capabilities highest %d, use --force?",
+				cpu, req->hwp_min, cap->highest);
+		if (req->hwp_min < cap->lowest)
+			errx(1, "cpu%d: requested min %d < capabilities lowest %d, use --force?",
+				cpu, req->hwp_min, cap->lowest);
 	}
 
-	retval = pread(fd, &msr, sizeof msr, offset);
+	if (update_hwp_min && update_hwp_max && (req->hwp_min > req->hwp_max))
+		errx(1, "cpu%d: requested min %d > requested max %d",
+			cpu, req->hwp_min, req->hwp_max);
 
-	if (retval != sizeof msr) {
-		printf("pread cpu%d 0x%x = %d\n", cpu, offset, retval);
-		exit(-2);
+	if (update_hwp_desired && req->hwp_desired) {
+		if (req->hwp_desired > req->hwp_max)
+			errx(1, "cpu%d: requested desired %d > requested max %d, use --force?",
+				cpu, req->hwp_desired, req->hwp_max);
+		if (req->hwp_desired < req->hwp_min)
+			errx(1, "cpu%d: requested desired %d < requested min %d, use --force?",
+				cpu, req->hwp_desired, req->hwp_min);
+		if (req->hwp_desired < cap->lowest)
+			errx(1, "cpu%d: requested desired %d < capabilities lowest %d, use --force?",
+				cpu, req->hwp_desired, cap->lowest);
+		if (req->hwp_desired > cap->highest)
+			errx(1, "cpu%d: requested desired %d > capabilities highest %d, use --force?",
+				cpu, req->hwp_desired, cap->highest);
 	}
-	close(fd);
-	return msr;
+
+	return 0;
 }
 
-unsigned long long  put_msr(int cpu, unsigned long long new_msr, int offset)
+int update_hwp_request(int cpu)
 {
-	unsigned long long old_msr;
-	char msr_path[32];
-	int retval;
-	int fd;
+	struct msr_hwp_request req;
+	struct msr_hwp_cap cap;
+
+	int msr_offset = MSR_HWP_REQUEST;
+
+	read_hwp_request(cpu, &req, msr_offset);
+	if (debug)
+		print_hwp_request(cpu, &req, "old: ");
+
+	if (update_hwp_min)
+		req.hwp_min = req_update.hwp_min;
+
+	if (update_hwp_max)
+		req.hwp_max = req_update.hwp_max;
+
+	if (update_hwp_desired)
+		req.hwp_desired = req_update.hwp_desired;
+
+	if (update_hwp_window)
+		req.hwp_window = req_update.hwp_window;
+
+	if (update_hwp_epp)
+		req.hwp_epp = req_update.hwp_epp;
+
+	req.hwp_use_pkg = req_update.hwp_use_pkg;
+
+	read_hwp_cap(cpu, &cap, MSR_HWP_CAPABILITIES);
+	if (debug)
+		print_hwp_cap(cpu, &cap, "");
+
+	if (!force)
+		check_hwp_request_v_hwp_capabilities(cpu, &req, &cap);
+
+	verify_hwp_req_self_consistency(cpu, &req);
 
-	sprintf(msr_path, "/dev/cpu/%d/msr", cpu);
-	fd = open(msr_path, O_RDWR);
-	if (fd < 0) {
-		perror(msr_path);
-		exit(1);
+	write_hwp_request(cpu, &req, msr_offset);
+
+	if (debug) {
+		read_hwp_request(cpu, &req, msr_offset);
+		print_hwp_request(cpu, &req, "new: ");
 	}
+	return 0;
+}
+int update_hwp_request_pkg(int pkg)
+{
+	struct msr_hwp_request req;
+	struct msr_hwp_cap cap;
+	int cpu = first_cpu_in_pkg[pkg];
+
+	int msr_offset = MSR_HWP_REQUEST_PKG;
+
+	read_hwp_request(cpu, &req, msr_offset);
+	if (debug)
+		print_hwp_request_pkg(pkg, &req, "old: ");
+
+	if (update_hwp_min)
+		req.hwp_min = req_update.hwp_min;
+
+	if (update_hwp_max)
+		req.hwp_max = req_update.hwp_max;
+
+	if (update_hwp_desired)
+		req.hwp_desired = req_update.hwp_desired;
+
+	if (update_hwp_window)
+		req.hwp_window = req_update.hwp_window;
+
+	if (update_hwp_epp)
+		req.hwp_epp = req_update.hwp_epp;
+
+	read_hwp_cap(cpu, &cap, MSR_HWP_CAPABILITIES);
+	if (debug)
+		print_hwp_cap(cpu, &cap, "");
+
+	if (!force)
+		check_hwp_request_v_hwp_capabilities(cpu, &req, &cap);
+
+	verify_hwp_req_self_consistency(cpu, &req);
+
+	write_hwp_request(cpu, &req, msr_offset);
 
-	retval = pread(fd, &old_msr, sizeof old_msr, offset);
-	if (retval != sizeof old_msr) {
-		perror("pwrite");
-		printf("pread cpu%d 0x%x = %d\n", cpu, offset, retval);
-		exit(-2);
+	if (debug) {
+		read_hwp_request(cpu, &req, msr_offset);
+		print_hwp_request_pkg(pkg, &req, "new: ");
 	}
+	return 0;
+}
+
+int enable_hwp_on_cpu(int cpu)
+{
+	unsigned long long msr;
+
+	get_msr(cpu, MSR_PM_ENABLE, &msr);
+	put_msr(cpu, MSR_PM_ENABLE, 1);
+
+	if (verbose)
+		printf("cpu%d: MSR_PM_ENABLE old: %d new: %d\n", cpu, (unsigned int) msr, 1);
+
+	return 0;
+}
+
+int update_cpu_msrs(int cpu)
+{
+	unsigned long long msr;
+
 
-	retval = pwrite(fd, &new_msr, sizeof new_msr, offset);
-	if (retval != sizeof new_msr) {
-		perror("pwrite");
-		printf("pwrite cpu%d 0x%x = %d\n", cpu, offset, retval);
-		exit(-2);
+	if (update_epb) {
+		get_msr(cpu, MSR_IA32_ENERGY_PERF_BIAS, &msr);
+		put_msr(cpu, MSR_IA32_ENERGY_PERF_BIAS, new_epb);
+
+		if (verbose)
+			printf("cpu%d: ENERGY_PERF_BIAS old: %d new: %d\n",
+				cpu, (unsigned int) msr, (unsigned int) new_epb);
 	}
 
-	close(fd);
+	if (update_turbo) {
+		int turbo_is_present_and_disabled;
+
+		get_msr(cpu, MSR_IA32_MISC_ENABLE, &msr);
+
+		turbo_is_present_and_disabled = ((msr & MSR_IA32_MISC_ENABLE_TURBO_DISABLE) != 0);
+
+		if (turbo_update_value == 1)	{
+			if (turbo_is_present_and_disabled) {
+				msr &= ~MSR_IA32_MISC_ENABLE_TURBO_DISABLE;
+				put_msr(cpu, MSR_IA32_MISC_ENABLE, msr);
+				if (verbose)
+					printf("cpu%d: turbo ENABLE\n", cpu);
+			}
+		} else {
+			/*
+			 * if "turbo_is_enabled" were known to be describe this cpu
+			 * then we could use it here to skip redundant disable requests.
+			 * but cpu may be in a different package, so we always write.
+			 */
+			msr |= MSR_IA32_MISC_ENABLE_TURBO_DISABLE;
+			put_msr(cpu, MSR_IA32_MISC_ENABLE, msr);
+			if (verbose)
+				printf("cpu%d: turbo DISABLE\n", cpu);
+		}
+	}
+
+	if (!has_hwp)
+		return 0;
+
+	if (!hwp_update_enabled())
+		return 0;
+
+	update_hwp_request(cpu);
+	return 0;
+}
+
+/*
+ * Open a file, and exit on failure
+ */
+FILE *fopen_or_die(const char *path, const char *mode)
+{
+	FILE *filep = fopen(path, "r");
 
-	return old_msr;
+	if (!filep)
+		err(1, "%s: open failed", path);
+	return filep;
 }
 
-void print_msr(int cpu)
+unsigned int get_pkg_num(int cpu)
 {
-	printf("cpu%d: 0x%016llx\n",
-		cpu, get_msr(cpu, MSR_IA32_ENERGY_PERF_BIAS));
+	FILE *fp;
+	char pathname[128];
+	unsigned int pkg;
+	int retval;
+
+	sprintf(pathname, "/sys/devices/system/cpu/cpu%d/topology/physical_package_id", cpu);
+
+	fp = fopen_or_die(pathname, "r");
+	retval = fscanf(fp, "%d\n", &pkg);
+	if (retval != 1)
+		errx(1, "%s: failed to parse", pathname);
+	return pkg;
 }
 
-void update_msr(int cpu)
+int set_max_cpu_pkg_num(int cpu)
 {
-	unsigned long long previous_msr;
+	unsigned int pkg;
 
-	previous_msr = put_msr(cpu, new_bias, MSR_IA32_ENERGY_PERF_BIAS);
+	if (max_cpu_num < cpu)
+		max_cpu_num = cpu;
 
-	if (verbose)
-		printf("cpu%d  msr0x%x 0x%016llx -> 0x%016llx\n",
-			cpu, MSR_IA32_ENERGY_PERF_BIAS, previous_msr, new_bias);
+	pkg = get_pkg_num(cpu);
+
+	if (pkg >= MAX_PACKAGES)
+		errx(1, "cpu%d: %d >= MAX_PACKAGES (%d)", cpu, pkg, MAX_PACKAGES);
+
+	if (pkg > max_pkg_num)
+		max_pkg_num = pkg;
 
-	return;
+	if ((pkg_present_set & (1ULL << pkg)) == 0) {
+		pkg_present_set |= (1ULL << pkg);
+		first_cpu_in_pkg[pkg] = cpu;
+	}
+
+	return 0;
+}
+int mark_cpu_present(int cpu)
+{
+	CPU_SET_S(cpu, cpu_setsize, cpu_present_set);
+	return 0;
 }
 
-char *proc_stat = "/proc/stat";
 /*
- * run func() on every cpu in /dev/cpu
+ * run func(cpu) on every cpu in /proc/stat
+ * return max_cpu number
  */
-void for_every_cpu(void (func)(int))
+int for_all_proc_cpus(int (func)(int))
 {
 	FILE *fp;
+	int cpu_num;
 	int retval;
 
-	fp = fopen(proc_stat, "r");
-	if (fp == NULL) {
-		perror(proc_stat);
-		exit(1);
-	}
+	fp = fopen_or_die(proc_stat, "r");
 
 	retval = fscanf(fp, "cpu %*d %*d %*d %*d %*d %*d %*d %*d %*d %*d\n");
-	if (retval != 0) {
-		perror("/proc/stat format");
-		exit(1);
-	}
+	if (retval != 0)
+		err(1, "%s: failed to parse format", proc_stat);
 
 	while (1) {
-		int cpu;
-
-		retval = fscanf(fp,
-			"cpu%u %*d %*d %*d %*d %*d %*d %*d %*d %*d %*d\n",
-			&cpu);
+		retval = fscanf(fp, "cpu%u %*d %*d %*d %*d %*d %*d %*d %*d %*d %*d\n", &cpu_num);
 		if (retval != 1)
 			break;
 
-		func(cpu);
+		retval = func(cpu_num);
+		if (retval) {
+			fclose(fp);
+			return retval;
+		}
 	}
 	fclose(fp);
+	return 0;
+}
+
+void for_all_cpus_in_set(size_t set_size, cpu_set_t *cpu_set, int (func)(int))
+{
+	int cpu_num;
+
+	for (cpu_num = 0; cpu_num <= max_cpu_num; ++cpu_num)
+		if (CPU_ISSET_S(cpu_num, set_size, cpu_set))
+			func(cpu_num);
+}
+
+void init_data_structures(void)
+{
+	for_all_proc_cpus(set_max_cpu_pkg_num);
+
+	cpu_setsize = CPU_ALLOC_SIZE((max_cpu_num + 1));
+
+	cpu_present_set = CPU_ALLOC((max_cpu_num + 1));
+	if (cpu_present_set == NULL)
+		err(3, "CPU_ALLOC");
+	CPU_ZERO_S(cpu_setsize, cpu_present_set);
+	for_all_proc_cpus(mark_cpu_present);
+}
+
+/* clear has_hwp if it is not enable (or being enabled) */
+
+void verify_hwp_is_enabled(void)
+{
+	unsigned long long msr;
+
+	if (!has_hwp)	/* set in early_cpuid() */
+		return;
+
+	/* MSR_PM_ENABLE[1] == 1 if HWP is enabled and MSRs visible */
+	get_msr(base_cpu, MSR_PM_ENABLE, &msr);
+	if ((msr & 1) == 0) {
+		fprintf(stderr, "HWP can be enabled using '--hwp-enable'\n");
+		has_hwp = 0;
+		return;
+	}
+}
+
+int req_update_bounds_check(void)
+{
+	if (!hwp_update_enabled())
+		return 0;
+
+	/* fail if min > max requested */
+	if ((update_hwp_max && update_hwp_min) &&
+	    (req_update.hwp_min > req_update.hwp_max)) {
+		printf("hwp-min %d > hwp_max %d\n", req_update.hwp_min, req_update.hwp_max);
+		return -EINVAL;
+	}
+
+	/* fail if desired > max requestd */
+	if (req_update.hwp_desired && update_hwp_max &&
+	    (req_update.hwp_desired > req_update.hwp_max)) {
+		printf("hwp-desired cannot be greater than hwp_max\n");
+		return -EINVAL;
+	}
+	/* fail if desired < min requestd */
+	if (req_update.hwp_desired && update_hwp_min &&
+	    (req_update.hwp_desired < req_update.hwp_min)) {
+		printf("hwp-desired cannot be less than hwp_min\n");
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+void set_base_cpu(void)
+{
+	base_cpu = sched_getcpu();
+	if (base_cpu < 0)
+		err(-ENODEV, "No valid cpus found");
+}
+
+
+void probe_dev_msr(void)
+{
+	struct stat sb;
+	char pathname[32];
+
+	sprintf(pathname, "/dev/cpu/%d/msr", base_cpu);
+	if (stat(pathname, &sb))
+		if (system("/sbin/modprobe msr > /dev/null 2>&1"))
+			err(-5, "no /dev/cpu/0/msr, Try \"# modprobe msr\" ");
+}
+/*
+ * early_cpuid()
+ * initialize turbo_is_enabled, has_hwp, has_epb
+ * before cmdline is parsed
+ */
+void early_cpuid(void)
+{
+	unsigned int eax, ebx, ecx, edx, max_level;
+	unsigned int fms, family, model;
+
+	__get_cpuid(0, &max_level, &ebx, &ecx, &edx);
+
+	if (max_level < 6)
+		errx(1, "Processor not supported\n");
+
+	__get_cpuid(1, &fms, &ebx, &ecx, &edx);
+	family = (fms >> 8) & 0xf;
+	model = (fms >> 4) & 0xf;
+	if (family == 6 || family == 0xf)
+		model += ((fms >> 16) & 0xf) << 4;
+
+	if (model == 0x4F) {
+		unsigned long long msr;
+
+		get_msr(base_cpu, MSR_TURBO_RATIO_LIMIT, &msr);
+
+		bdx_highest_ratio = msr & 0xFF;
+	}
+
+	__get_cpuid(0x6, &eax, &ebx, &ecx, &edx);
+	turbo_is_enabled = (eax >> 1) & 1;
+	has_hwp = (eax >> 7) & 1;
+	has_epb = (ecx >> 3) & 1;
+}
+
+/*
+ * parse_cpuid()
+ * set
+ * has_hwp, has_hwp_notify, has_hwp_activity_window, has_hwp_epp, has_hwp_request_pkg, has_epb
+ */
+void parse_cpuid(void)
+{
+	unsigned int eax, ebx, ecx, edx, max_level;
+	unsigned int fms, family, model, stepping;
+
+	eax = ebx = ecx = edx = 0;
+
+	__get_cpuid(0, &max_level, &ebx, &ecx, &edx);
+
+	if (ebx == 0x756e6547 && edx == 0x49656e69 && ecx == 0x6c65746e)
+		genuine_intel = 1;
+
+	if (debug)
+		fprintf(stderr, "CPUID(0): %.4s%.4s%.4s ",
+			(char *)&ebx, (char *)&edx, (char *)&ecx);
+
+	__get_cpuid(1, &fms, &ebx, &ecx, &edx);
+	family = (fms >> 8) & 0xf;
+	model = (fms >> 4) & 0xf;
+	stepping = fms & 0xf;
+	if (family == 6 || family == 0xf)
+		model += ((fms >> 16) & 0xf) << 4;
+
+	if (debug) {
+		fprintf(stderr, "%d CPUID levels; family:model:stepping 0x%x:%x:%x (%d:%d:%d)\n",
+			max_level, family, model, stepping, family, model, stepping);
+		fprintf(stderr, "CPUID(1): %s %s %s %s %s %s %s %s\n",
+			ecx & (1 << 0) ? "SSE3" : "-",
+			ecx & (1 << 3) ? "MONITOR" : "-",
+			ecx & (1 << 7) ? "EIST" : "-",
+			ecx & (1 << 8) ? "TM2" : "-",
+			edx & (1 << 4) ? "TSC" : "-",
+			edx & (1 << 5) ? "MSR" : "-",
+			edx & (1 << 22) ? "ACPI-TM" : "-",
+			edx & (1 << 29) ? "TM" : "-");
+	}
+
+	if (!(edx & (1 << 5)))
+		errx(1, "CPUID: no MSR");
+
+
+	__get_cpuid(0x6, &eax, &ebx, &ecx, &edx);
+	/* turbo_is_enabled already set */
+	/* has_hwp already set */
+	has_hwp_notify = eax & (1 << 8);
+	has_hwp_activity_window = eax & (1 << 9);
+	has_hwp_epp = eax & (1 << 10);
+	has_hwp_request_pkg = eax & (1 << 11);
+
+	if (!has_hwp_request_pkg && update_hwp_use_pkg)
+		errx(1, "--hwp-use-pkg is not available on this hardware");
+
+	/* has_epb already set */
+
+	if (debug)
+		fprintf(stderr,
+			"CPUID(6): %sTURBO, %sHWP, %sHWPnotify, %sHWPwindow, %sHWPepp, %sHWPpkg, %sEPB\n",
+			turbo_is_enabled ? "" : "No-",
+			has_hwp ? "" : "No-",
+			has_hwp_notify ? "" : "No-",
+			has_hwp_activity_window ? "" : "No-",
+			has_hwp_epp ? "" : "No-",
+			has_hwp_request_pkg ? "" : "No-",
+			has_epb ? "" : "No-");
+
+	return;	/* success */
 }
 
 int main(int argc, char **argv)
 {
+	set_base_cpu();
+	probe_dev_msr();
+	init_data_structures();
+
+	early_cpuid();	/* initial cpuid parse before cmdline */
+
 	cmdline(argc, argv);
 
-	if (verbose > 1)
-		printf("x86_energy_perf_policy Nov 24, 2010"
-				" - Len Brown <lenb@kernel.org>\n");
-	if (verbose > 1 && !read_only)
-		printf("new_bias %lld\n", new_bias);
-
-	validate_cpuid();
-
-	if (cpu != -1) {
-		if (read_only)
-			print_msr(cpu);
-		else
-			update_msr(cpu);
-	} else {
-		if (read_only)
-			for_every_cpu(print_msr);
-		else
-			for_every_cpu(update_msr);
+	if (debug)
+		print_version();
+
+	parse_cpuid();
+
+	 /* If CPU-set and PKG-set are not initialized, default to all CPUs */
+	if ((cpu_selected_set == 0) && (pkg_selected_set == 0))
+		cpu_selected_set = cpu_present_set;
+
+	/*
+	 * If HWP is being enabled, do it now, so that subsequent operations
+	 * that access HWP registers can work.
+	 */
+	if (update_hwp_enable)
+		for_all_cpus_in_set(cpu_setsize, cpu_selected_set, enable_hwp_on_cpu);
+
+	/* If HWP present, but disabled, warn and ignore from here forward */
+	verify_hwp_is_enabled();
+
+	if (req_update_bounds_check())
+		return -EINVAL;
+
+	/* display information only, no updates to settings */
+	if (!update_epb && !update_turbo && !hwp_update_enabled()) {
+		if (cpu_selected_set)
+			for_all_cpus_in_set(cpu_setsize, cpu_selected_set, print_cpu_msrs);
+
+		if (has_hwp_request_pkg) {
+			if (pkg_selected_set == 0)
+				pkg_selected_set = pkg_present_set;
+
+			for_packages(pkg_selected_set, print_pkg_msrs);
+		}
+
+		return 0;
 	}
 
+	/* update CPU set */
+	if (cpu_selected_set) {
+		for_all_cpus_in_set(cpu_setsize, cpu_selected_set, update_sysfs);
+		for_all_cpus_in_set(cpu_setsize, cpu_selected_set, update_cpu_msrs);
+	} else if (pkg_selected_set)
+		for_packages(pkg_selected_set, update_hwp_request_pkg);
+
 	return 0;
 }