xref: /openbmc/linux/arch/arm64/include/asm/kvm_host.h (revision 680ef72a)
1 /*
2  * Copyright (C) 2012,2013 - ARM Ltd
3  * Author: Marc Zyngier <marc.zyngier@arm.com>
4  *
5  * Derived from arch/arm/include/asm/kvm_host.h:
6  * Copyright (C) 2012 - Virtual Open Systems and Columbia University
7  * Author: Christoffer Dall <c.dall@virtualopensystems.com>
8  *
9  * This program is free software; you can redistribute it and/or modify
10  * it under the terms of the GNU General Public License version 2 as
11  * published by the Free Software Foundation.
12  *
13  * This program is distributed in the hope that it will be useful,
14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
16  * GNU General Public License for more details.
17  *
18  * You should have received a copy of the GNU General Public License
19  * along with this program.  If not, see <http://www.gnu.org/licenses/>.
20  */
21 
22 #ifndef __ARM64_KVM_HOST_H__
23 #define __ARM64_KVM_HOST_H__
24 
25 #include <linux/types.h>
26 #include <linux/kvm_types.h>
27 #include <asm/cpufeature.h>
28 #include <asm/fpsimd.h>
29 #include <asm/kvm.h>
30 #include <asm/kvm_asm.h>
31 #include <asm/kvm_mmio.h>
32 
33 #define __KVM_HAVE_ARCH_INTC_INITIALIZED
34 
35 #define KVM_USER_MEM_SLOTS 512
36 #define KVM_HALT_POLL_NS_DEFAULT 500000
37 
38 #include <kvm/arm_vgic.h>
39 #include <kvm/arm_arch_timer.h>
40 #include <kvm/arm_pmu.h>
41 
42 #define KVM_MAX_VCPUS VGIC_V3_MAX_CPUS
43 
44 #define KVM_VCPU_MAX_FEATURES 4
45 
46 #define KVM_REQ_SLEEP \
47 	KVM_ARCH_REQ_FLAGS(0, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
48 #define KVM_REQ_IRQ_PENDING	KVM_ARCH_REQ(1)
49 
50 int __attribute_const__ kvm_target_cpu(void);
51 int kvm_reset_vcpu(struct kvm_vcpu *vcpu);
52 int kvm_arch_dev_ioctl_check_extension(struct kvm *kvm, long ext);
53 void __extended_idmap_trampoline(phys_addr_t boot_pgd, phys_addr_t idmap_start);
54 
55 struct kvm_arch {
56 	/* The VMID generation used for the virt. memory system */
57 	u64    vmid_gen;
58 	u32    vmid;
59 
60 	/* 1-level 2nd stage table and lock */
61 	spinlock_t pgd_lock;
62 	pgd_t *pgd;
63 
64 	/* VTTBR value associated with above pgd and vmid */
65 	u64    vttbr;
66 
67 	/* The last vcpu id that ran on each physical CPU */
68 	int __percpu *last_vcpu_ran;
69 
70 	/* The maximum number of vCPUs depends on the used GIC model */
71 	int max_vcpus;
72 
73 	/* Interrupt controller */
74 	struct vgic_dist	vgic;
75 };
76 
77 #define KVM_NR_MEM_OBJS     40
78 
79 /*
80  * We don't want allocation failures within the mmu code, so we preallocate
81  * enough memory for a single page fault in a cache.
82  */
83 struct kvm_mmu_memory_cache {
84 	int nobjs;
85 	void *objects[KVM_NR_MEM_OBJS];
86 };
87 
88 struct kvm_vcpu_fault_info {
89 	u32 esr_el2;		/* Hyp Syndrom Register */
90 	u64 far_el2;		/* Hyp Fault Address Register */
91 	u64 hpfar_el2;		/* Hyp IPA Fault Address Register */
92 };
93 
94 /*
95  * 0 is reserved as an invalid value.
96  * Order should be kept in sync with the save/restore code.
97  */
98 enum vcpu_sysreg {
99 	__INVALID_SYSREG__,
100 	MPIDR_EL1,	/* MultiProcessor Affinity Register */
101 	CSSELR_EL1,	/* Cache Size Selection Register */
102 	SCTLR_EL1,	/* System Control Register */
103 	ACTLR_EL1,	/* Auxiliary Control Register */
104 	CPACR_EL1,	/* Coprocessor Access Control */
105 	TTBR0_EL1,	/* Translation Table Base Register 0 */
106 	TTBR1_EL1,	/* Translation Table Base Register 1 */
107 	TCR_EL1,	/* Translation Control Register */
108 	ESR_EL1,	/* Exception Syndrome Register */
109 	AFSR0_EL1,	/* Auxiliary Fault Status Register 0 */
110 	AFSR1_EL1,	/* Auxiliary Fault Status Register 1 */
111 	FAR_EL1,	/* Fault Address Register */
112 	MAIR_EL1,	/* Memory Attribute Indirection Register */
113 	VBAR_EL1,	/* Vector Base Address Register */
114 	CONTEXTIDR_EL1,	/* Context ID Register */
115 	TPIDR_EL0,	/* Thread ID, User R/W */
116 	TPIDRRO_EL0,	/* Thread ID, User R/O */
117 	TPIDR_EL1,	/* Thread ID, Privileged */
118 	AMAIR_EL1,	/* Aux Memory Attribute Indirection Register */
119 	CNTKCTL_EL1,	/* Timer Control Register (EL1) */
120 	PAR_EL1,	/* Physical Address Register */
121 	MDSCR_EL1,	/* Monitor Debug System Control Register */
122 	MDCCINT_EL1,	/* Monitor Debug Comms Channel Interrupt Enable Reg */
123 
124 	/* Performance Monitors Registers */
125 	PMCR_EL0,	/* Control Register */
126 	PMSELR_EL0,	/* Event Counter Selection Register */
127 	PMEVCNTR0_EL0,	/* Event Counter Register (0-30) */
128 	PMEVCNTR30_EL0 = PMEVCNTR0_EL0 + 30,
129 	PMCCNTR_EL0,	/* Cycle Counter Register */
130 	PMEVTYPER0_EL0,	/* Event Type Register (0-30) */
131 	PMEVTYPER30_EL0 = PMEVTYPER0_EL0 + 30,
132 	PMCCFILTR_EL0,	/* Cycle Count Filter Register */
133 	PMCNTENSET_EL0,	/* Count Enable Set Register */
134 	PMINTENSET_EL1,	/* Interrupt Enable Set Register */
135 	PMOVSSET_EL0,	/* Overflow Flag Status Set Register */
136 	PMSWINC_EL0,	/* Software Increment Register */
137 	PMUSERENR_EL0,	/* User Enable Register */
138 
139 	/* 32bit specific registers. Keep them at the end of the range */
140 	DACR32_EL2,	/* Domain Access Control Register */
141 	IFSR32_EL2,	/* Instruction Fault Status Register */
142 	FPEXC32_EL2,	/* Floating-Point Exception Control Register */
143 	DBGVCR32_EL2,	/* Debug Vector Catch Register */
144 
145 	NR_SYS_REGS	/* Nothing after this line! */
146 };
147 
148 /* 32bit mapping */
149 #define c0_MPIDR	(MPIDR_EL1 * 2)	/* MultiProcessor ID Register */
150 #define c0_CSSELR	(CSSELR_EL1 * 2)/* Cache Size Selection Register */
151 #define c1_SCTLR	(SCTLR_EL1 * 2)	/* System Control Register */
152 #define c1_ACTLR	(ACTLR_EL1 * 2)	/* Auxiliary Control Register */
153 #define c1_CPACR	(CPACR_EL1 * 2)	/* Coprocessor Access Control */
154 #define c2_TTBR0	(TTBR0_EL1 * 2)	/* Translation Table Base Register 0 */
155 #define c2_TTBR0_high	(c2_TTBR0 + 1)	/* TTBR0 top 32 bits */
156 #define c2_TTBR1	(TTBR1_EL1 * 2)	/* Translation Table Base Register 1 */
157 #define c2_TTBR1_high	(c2_TTBR1 + 1)	/* TTBR1 top 32 bits */
158 #define c2_TTBCR	(TCR_EL1 * 2)	/* Translation Table Base Control R. */
159 #define c3_DACR		(DACR32_EL2 * 2)/* Domain Access Control Register */
160 #define c5_DFSR		(ESR_EL1 * 2)	/* Data Fault Status Register */
161 #define c5_IFSR		(IFSR32_EL2 * 2)/* Instruction Fault Status Register */
162 #define c5_ADFSR	(AFSR0_EL1 * 2)	/* Auxiliary Data Fault Status R */
163 #define c5_AIFSR	(AFSR1_EL1 * 2)	/* Auxiliary Instr Fault Status R */
164 #define c6_DFAR		(FAR_EL1 * 2)	/* Data Fault Address Register */
165 #define c6_IFAR		(c6_DFAR + 1)	/* Instruction Fault Address Register */
166 #define c7_PAR		(PAR_EL1 * 2)	/* Physical Address Register */
167 #define c7_PAR_high	(c7_PAR + 1)	/* PAR top 32 bits */
168 #define c10_PRRR	(MAIR_EL1 * 2)	/* Primary Region Remap Register */
169 #define c10_NMRR	(c10_PRRR + 1)	/* Normal Memory Remap Register */
170 #define c12_VBAR	(VBAR_EL1 * 2)	/* Vector Base Address Register */
171 #define c13_CID		(CONTEXTIDR_EL1 * 2)	/* Context ID Register */
172 #define c13_TID_URW	(TPIDR_EL0 * 2)	/* Thread ID, User R/W */
173 #define c13_TID_URO	(TPIDRRO_EL0 * 2)/* Thread ID, User R/O */
174 #define c13_TID_PRIV	(TPIDR_EL1 * 2)	/* Thread ID, Privileged */
175 #define c10_AMAIR0	(AMAIR_EL1 * 2)	/* Aux Memory Attr Indirection Reg */
176 #define c10_AMAIR1	(c10_AMAIR0 + 1)/* Aux Memory Attr Indirection Reg */
177 #define c14_CNTKCTL	(CNTKCTL_EL1 * 2) /* Timer Control Register (PL1) */
178 
179 #define cp14_DBGDSCRext	(MDSCR_EL1 * 2)
180 #define cp14_DBGBCR0	(DBGBCR0_EL1 * 2)
181 #define cp14_DBGBVR0	(DBGBVR0_EL1 * 2)
182 #define cp14_DBGBXVR0	(cp14_DBGBVR0 + 1)
183 #define cp14_DBGWCR0	(DBGWCR0_EL1 * 2)
184 #define cp14_DBGWVR0	(DBGWVR0_EL1 * 2)
185 #define cp14_DBGDCCINT	(MDCCINT_EL1 * 2)
186 
187 #define NR_COPRO_REGS	(NR_SYS_REGS * 2)
188 
189 struct kvm_cpu_context {
190 	struct kvm_regs	gp_regs;
191 	union {
192 		u64 sys_regs[NR_SYS_REGS];
193 		u32 copro[NR_COPRO_REGS];
194 	};
195 };
196 
197 typedef struct kvm_cpu_context kvm_cpu_context_t;
198 
199 struct kvm_vcpu_arch {
200 	struct kvm_cpu_context ctxt;
201 
202 	/* HYP configuration */
203 	u64 hcr_el2;
204 	u32 mdcr_el2;
205 
206 	/* Exception Information */
207 	struct kvm_vcpu_fault_info fault;
208 
209 	/* Guest debug state */
210 	u64 debug_flags;
211 
212 	/*
213 	 * We maintain more than a single set of debug registers to support
214 	 * debugging the guest from the host and to maintain separate host and
215 	 * guest state during world switches. vcpu_debug_state are the debug
216 	 * registers of the vcpu as the guest sees them.  host_debug_state are
217 	 * the host registers which are saved and restored during
218 	 * world switches. external_debug_state contains the debug
219 	 * values we want to debug the guest. This is set via the
220 	 * KVM_SET_GUEST_DEBUG ioctl.
221 	 *
222 	 * debug_ptr points to the set of debug registers that should be loaded
223 	 * onto the hardware when running the guest.
224 	 */
225 	struct kvm_guest_debug_arch *debug_ptr;
226 	struct kvm_guest_debug_arch vcpu_debug_state;
227 	struct kvm_guest_debug_arch external_debug_state;
228 
229 	/* Pointer to host CPU context */
230 	kvm_cpu_context_t *host_cpu_context;
231 	struct {
232 		/* {Break,watch}point registers */
233 		struct kvm_guest_debug_arch regs;
234 		/* Statistical profiling extension */
235 		u64 pmscr_el1;
236 	} host_debug_state;
237 
238 	/* VGIC state */
239 	struct vgic_cpu vgic_cpu;
240 	struct arch_timer_cpu timer_cpu;
241 	struct kvm_pmu pmu;
242 
243 	/*
244 	 * Anything that is not used directly from assembly code goes
245 	 * here.
246 	 */
247 
248 	/*
249 	 * Guest registers we preserve during guest debugging.
250 	 *
251 	 * These shadow registers are updated by the kvm_handle_sys_reg
252 	 * trap handler if the guest accesses or updates them while we
253 	 * are using guest debug.
254 	 */
255 	struct {
256 		u32	mdscr_el1;
257 	} guest_debug_preserved;
258 
259 	/* vcpu power-off state */
260 	bool power_off;
261 
262 	/* Don't run the guest (internal implementation need) */
263 	bool pause;
264 
265 	/* IO related fields */
266 	struct kvm_decode mmio_decode;
267 
268 	/* Interrupt related fields */
269 	u64 irq_lines;		/* IRQ and FIQ levels */
270 
271 	/* Cache some mmu pages needed inside spinlock regions */
272 	struct kvm_mmu_memory_cache mmu_page_cache;
273 
274 	/* Target CPU and feature flags */
275 	int target;
276 	DECLARE_BITMAP(features, KVM_VCPU_MAX_FEATURES);
277 
278 	/* Detect first run of a vcpu */
279 	bool has_run_once;
280 };
281 
282 #define vcpu_gp_regs(v)		(&(v)->arch.ctxt.gp_regs)
283 #define vcpu_sys_reg(v,r)	((v)->arch.ctxt.sys_regs[(r)])
284 /*
285  * CP14 and CP15 live in the same array, as they are backed by the
286  * same system registers.
287  */
288 #define vcpu_cp14(v,r)		((v)->arch.ctxt.copro[(r)])
289 #define vcpu_cp15(v,r)		((v)->arch.ctxt.copro[(r)])
290 
291 #ifdef CONFIG_CPU_BIG_ENDIAN
292 #define vcpu_cp15_64_high(v,r)	vcpu_cp15((v),(r))
293 #define vcpu_cp15_64_low(v,r)	vcpu_cp15((v),(r) + 1)
294 #else
295 #define vcpu_cp15_64_high(v,r)	vcpu_cp15((v),(r) + 1)
296 #define vcpu_cp15_64_low(v,r)	vcpu_cp15((v),(r))
297 #endif
298 
299 struct kvm_vm_stat {
300 	ulong remote_tlb_flush;
301 };
302 
303 struct kvm_vcpu_stat {
304 	u64 halt_successful_poll;
305 	u64 halt_attempted_poll;
306 	u64 halt_poll_invalid;
307 	u64 halt_wakeup;
308 	u64 hvc_exit_stat;
309 	u64 wfe_exit_stat;
310 	u64 wfi_exit_stat;
311 	u64 mmio_exit_user;
312 	u64 mmio_exit_kernel;
313 	u64 exits;
314 };
315 
316 int kvm_vcpu_preferred_target(struct kvm_vcpu_init *init);
317 unsigned long kvm_arm_num_regs(struct kvm_vcpu *vcpu);
318 int kvm_arm_copy_reg_indices(struct kvm_vcpu *vcpu, u64 __user *indices);
319 int kvm_arm_get_reg(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg);
320 int kvm_arm_set_reg(struct kvm_vcpu *vcpu, const struct kvm_one_reg *reg);
321 
322 #define KVM_ARCH_WANT_MMU_NOTIFIER
323 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
324 int kvm_unmap_hva_range(struct kvm *kvm,
325 			unsigned long start, unsigned long end);
326 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
327 int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end);
328 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
329 
330 struct kvm_vcpu *kvm_arm_get_running_vcpu(void);
331 struct kvm_vcpu * __percpu *kvm_get_running_vcpus(void);
332 void kvm_arm_halt_guest(struct kvm *kvm);
333 void kvm_arm_resume_guest(struct kvm *kvm);
334 
335 u64 __kvm_call_hyp(void *hypfn, ...);
336 #define kvm_call_hyp(f, ...) __kvm_call_hyp(kvm_ksym_ref(f), ##__VA_ARGS__)
337 
338 void force_vm_exit(const cpumask_t *mask);
339 void kvm_mmu_wp_memory_region(struct kvm *kvm, int slot);
340 
341 int handle_exit(struct kvm_vcpu *vcpu, struct kvm_run *run,
342 		int exception_index);
343 
344 int kvm_perf_init(void);
345 int kvm_perf_teardown(void);
346 
347 struct kvm_vcpu *kvm_mpidr_to_vcpu(struct kvm *kvm, unsigned long mpidr);
348 
349 static inline void __cpu_init_hyp_mode(phys_addr_t pgd_ptr,
350 				       unsigned long hyp_stack_ptr,
351 				       unsigned long vector_ptr)
352 {
353 	/*
354 	 * Call initialization code, and switch to the full blown HYP code.
355 	 * If the cpucaps haven't been finalized yet, something has gone very
356 	 * wrong, and hyp will crash and burn when it uses any
357 	 * cpus_have_const_cap() wrapper.
358 	 */
359 	BUG_ON(!static_branch_likely(&arm64_const_caps_ready));
360 	__kvm_call_hyp((void *)pgd_ptr, hyp_stack_ptr, vector_ptr);
361 }
362 
363 static inline void kvm_arch_hardware_unsetup(void) {}
364 static inline void kvm_arch_sync_events(struct kvm *kvm) {}
365 static inline void kvm_arch_vcpu_uninit(struct kvm_vcpu *vcpu) {}
366 static inline void kvm_arch_sched_in(struct kvm_vcpu *vcpu, int cpu) {}
367 static inline void kvm_arch_vcpu_block_finish(struct kvm_vcpu *vcpu) {}
368 
369 void kvm_arm_init_debug(void);
370 void kvm_arm_setup_debug(struct kvm_vcpu *vcpu);
371 void kvm_arm_clear_debug(struct kvm_vcpu *vcpu);
372 void kvm_arm_reset_debug_ptr(struct kvm_vcpu *vcpu);
373 int kvm_arm_vcpu_arch_set_attr(struct kvm_vcpu *vcpu,
374 			       struct kvm_device_attr *attr);
375 int kvm_arm_vcpu_arch_get_attr(struct kvm_vcpu *vcpu,
376 			       struct kvm_device_attr *attr);
377 int kvm_arm_vcpu_arch_has_attr(struct kvm_vcpu *vcpu,
378 			       struct kvm_device_attr *attr);
379 
380 static inline void __cpu_init_stage2(void)
381 {
382 	u32 parange = kvm_call_hyp(__init_stage2_translation);
383 
384 	WARN_ONCE(parange < 40,
385 		  "PARange is %d bits, unsupported configuration!", parange);
386 }
387 
388 /*
389  * All host FP/SIMD state is restored on guest exit, so nothing needs
390  * doing here except in the SVE case:
391 */
392 static inline void kvm_fpsimd_flush_cpu_state(void)
393 {
394 	if (system_supports_sve())
395 		sve_flush_cpu_state();
396 }
397 
398 #endif /* __ARM64_KVM_HOST_H__ */
399