summaryrefslogtreecommitdiff
path: root/src/soc/intel/xeon_sp/skx/cpu.c
blob: fcdb2d7ea04aefb6a5cc4bcd97014d95a50062ca (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
/* SPDX-License-Identifier: GPL-2.0-or-later */

#include <console/console.h>
#include <console/debug.h>
#include <intelblocks/cpulib.h>
#include <cpu/cpu.h>
#include <cpu/x86/mtrr.h>
#include <cpu/x86/mp.h>
#include <cpu/intel/turbo.h>
#include <soc/msr.h>
#include <soc/cpu.h>
#include <soc/soc_util.h>
#include <soc/smmrelocate.h>
#include <soc/util.h>
#include <assert.h>
#include "chip.h"
#include <cpu/intel/smm_reloc.h>
#include <cpu/intel/em64t101_save_state.h>
#include <types.h>

static const config_t *chip_config = NULL;

static void xeon_configure_mca(void)
{
	msr_t msr;
	struct cpuid_result cpuid_regs;

	/* Check feature flag in CPUID.(EAX=1):EDX[7]==1  MCE
	 *                   and CPUID.(EAX=1):EDX[14]==1 MCA*/
	cpuid_regs = cpuid(1);
	if ((cpuid_regs.edx & (1<<7 | 1<<14)) != (1<<7 | 1<<14))
		return;

	msr = rdmsr(IA32_MCG_CAP);
	if (msr.lo & IA32_MCG_CAP_CTL_P_MASK) {
		/* Enable all error logging */
		msr.lo = msr.hi = 0xffffffff;
		wrmsr(IA32_MCG_CTL, msr);
	}

	/* TODO(adurbin): This should only be done on a cold boot. Also, some
	   of these banks are core vs package scope. For now every CPU clears
	   every bank. */
	mca_configure();
}

static void xeon_sp_core_init(struct device *cpu)
{
	msr_t msr;

	printk(BIOS_INFO, "%s dev: %s, cpu: %d, apic_id: 0x%x\n",
		__func__, dev_path(cpu), cpu_index(), cpu->path.apic.apic_id);
	assert(chip_config != NULL);

	/* set MSR_PKG_CST_CONFIG_CONTROL - scope per core*/
	msr.hi = 0;
	msr.lo = (PKG_CSTATE_NO_LIMIT | CFG_LOCK_ENABLE);
	wrmsr(MSR_PKG_CST_CONFIG_CONTROL, msr);

	/* Enable Energy Perf Bias Access, Dynamic switching and lock MSR */
	msr = rdmsr(MSR_POWER_CTL);
	msr.lo |= (ENERGY_PERF_BIAS_ACCESS_ENABLE | PWR_PERF_TUNING_DYN_SWITCHING_ENABLE
		| PROCHOT_LOCK_ENABLE);
	wrmsr(MSR_POWER_CTL, msr);

	/* Set P-State ratio */
	msr = rdmsr(MSR_IA32_PERF_CTRL);
	msr.lo &= ~PSTATE_REQ_MASK;
	msr.lo |= (chip_config->pstate_req_ratio << PSTATE_REQ_SHIFT);
	wrmsr(MSR_IA32_PERF_CTRL, msr);

	/*
	 * Set HWP base feature, EPP reg enumeration, lock thermal and msr
	 * TODO: Set LOCK_MISC_PWR_MGMT_MSR, Unexpected Exception if you
	 * lock & issue wrmsr on every thread
	 * This is package level MSR. Need to check if it updates correctly on
	 * multi-socket platform.
	 */
	msr = rdmsr(MSR_MISC_PWR_MGMT);
	if (!(msr.lo & LOCK_MISC_PWR_MGMT_MSR)) { /* if already locked skip update */
		msr.lo = (HWP_ENUM_ENABLE | HWP_EPP_ENUM_ENABLE | LOCK_MISC_PWR_MGMT_MSR |
			LOCK_THERM_INT);
		wrmsr(MSR_MISC_PWR_MGMT, msr);
	}

	/* TODO MSR_VR_MISC_CONFIG */

	/* Set current limit lock */
	msr = rdmsr(MSR_VR_CURRENT_CONFIG);
	msr.lo |= CURRENT_LIMIT_LOCK;
	wrmsr(MSR_VR_CURRENT_CONFIG, msr);

	/* Set Turbo Ratio Limits */
	msr.lo = chip_config->turbo_ratio_limit & 0xffffffff;
	msr.hi = (chip_config->turbo_ratio_limit >> 32) & 0xffffffff;
	wrmsr(MSR_TURBO_RATIO_LIMIT, msr);

	/* Set Turbo Ratio Limit Cores */
	msr.lo = chip_config->turbo_ratio_limit_cores & 0xffffffff;
	msr.hi = (chip_config->turbo_ratio_limit_cores >> 32) & 0xffffffff;
	wrmsr(MSR_TURBO_RATIO_LIMIT_CORES, msr);

	/* set Turbo Activation ratio */
	msr.hi = 0;
	msr = rdmsr(MSR_TURBO_ACTIVATION_RATIO);
	msr.lo |= MAX_NON_TURBO_RATIO;
	wrmsr(MSR_TURBO_ACTIVATION_RATIO, msr);

	/* Enable Fast Strings */
	msr = rdmsr(IA32_MISC_ENABLE);
	msr.lo |= FAST_STRINGS_ENABLE_BIT;
	wrmsr(IA32_MISC_ENABLE, msr);

	/* Set energy policy */
	msr_t msr1 = rdmsr(MSR_ENERGY_PERF_BIAS_CONFIG);
	msr.lo = (msr1.lo & EPB_ENERGY_POLICY_MASK) >> EPB_ENERGY_POLICY_SHIFT;
	msr.hi = 0;
	wrmsr(MSR_IA32_ENERGY_PERF_BIAS, msr);

	/* Enable Turbo */
	enable_turbo();

	/* Enable speed step. */
	if (get_turbo_state() == TURBO_ENABLED) {
		msr = rdmsr(IA32_MISC_ENABLE);
		msr.lo |= SPEED_STEP_ENABLE_BIT;
		wrmsr(IA32_MISC_ENABLE, msr);
	}

	/* Clear out pending MCEs */
	xeon_configure_mca();
}

static struct device_operations cpu_dev_ops = {
	.init = xeon_sp_core_init,
};

static const struct cpu_device_id cpu_table[] = {
	/* Skylake-SP A0/A1 CPUID 0x506f0*/
	{X86_VENDOR_INTEL, CPUID_SKYLAKE_SP_A0_A1},
	/* Skylake-SP B0 CPUID 0x506f1*/
	{X86_VENDOR_INTEL, CPUID_SKYLAKE_SP_B0},
	/* Skylake-SP 4 CPUID 0x50654*/
	{X86_VENDOR_INTEL, CPUID_SKYLAKE_SP_4},
	{0, 0},
};

static const struct cpu_driver driver __cpu_driver = {
	.ops = &cpu_dev_ops,
	.id_table = cpu_table,
};

static void set_max_turbo_freq(void)
{
	msr_t msr, perf_ctl;

	FUNC_ENTER();
	perf_ctl.hi = 0;

	/* Check for configurable TDP option */
	if (get_turbo_state() == TURBO_ENABLED) {
		msr = rdmsr(MSR_TURBO_RATIO_LIMIT);
		perf_ctl.lo = (msr.lo & 0xff) << 8;
	} else if (cpu_config_tdp_levels()) {
		/* Set to nominal TDP ratio */
		msr = rdmsr(MSR_CONFIG_TDP_NOMINAL);
		perf_ctl.lo = (msr.lo & 0xff) << 8;
	} else {
		/* Platform Info bits 15:8 give max ratio */
		msr = rdmsr(MSR_PLATFORM_INFO);
		perf_ctl.lo = msr.lo & 0xff00;
	}
	wrmsr(IA32_PERF_CTL, perf_ctl);

	printk(BIOS_DEBUG, "cpu: frequency set to %d\n",
	       ((perf_ctl.lo >> 8) & 0xff) * CPU_BCLK);
	FUNC_EXIT();
}

/*
 * Do essential initialization tasks before APs can be fired up
 *
 * Prevent race condition in MTRR solution. Enable MTRRs on the BSP. This
 * creates the MTRR solution that the APs will use. Otherwise APs will try to
 * apply the incomplete solution as the BSP is calculating it.
 */
static void pre_mp_init(void)
{
	printk(BIOS_DEBUG, "%s: entry\n", __func__);

	x86_setup_mtrrs_with_detect();
	x86_mtrr_check();
}

static void post_mp_init(void)
{
	/* Set Max Ratio */
	set_max_turbo_freq();

	if (CONFIG(HAVE_SMI_HANDLER))
		global_smi_enable();
}

/*
 * CPU initialization recipe
 *
 * Note that no microcode update is passed to the init function. CSE updates
 * the microcode on all cores before releasing them from reset. That means that
 * the BSP and all APs will come up with the same microcode revision.
 */
static const struct mp_ops mp_ops = {
	.pre_mp_init = pre_mp_init,
	.get_cpu_count = get_platform_thread_count,
	.get_smm_info = get_smm_info,
	.pre_mp_smm_init = smm_southbridge_clear_state,
	.relocation_handler = smm_relocation_handler,
	.post_mp_init = post_mp_init,
};

void xeon_sp_init_cpus(struct device *dev)
{
	FUNC_ENTER();

	/*
	 * This gets used in cpu device callback. Other than cpu 0,
	 * rest of the CPU devices do not have
	 * chip_info updated. Global chip_config is used as workaround
	 */
	chip_config = dev->chip_info;

	config_reset_cpl3_csrs();

	/* calls src/cpu/x86/mp_init.c */
	/* TODO: Handle mp_init_with_smm failure? */
	mp_init_with_smm(dev->link_list, &mp_ops);

	/* update numa domain for all cpu devices */
	xeonsp_init_cpu_config();

	FUNC_EXIT();
}