^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1) // SPDX-License-Identifier: GPL-2.0-only
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3) * Copyright 2011 Paul Mackerras, IBM Corp. <paulus@au1.ibm.com>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4) * Copyright (C) 2009. SUSE Linux Products GmbH. All rights reserved.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 6) * Authors:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 7) * Paul Mackerras <paulus@au1.ibm.com>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 8) * Alexander Graf <agraf@suse.de>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 9) * Kevin Wolf <mail@kevin-wolf.de>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 10) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 11) * Description: KVM functions specific to running on Book 3S
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 12) * processors in hypervisor mode (specifically POWER7 and later).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 13) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 14) * This file is derived from arch/powerpc/kvm/book3s.c,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 15) * by Alexander Graf <agraf@suse.de>.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 16) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 17)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 18) #include <linux/kvm_host.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 19) #include <linux/kernel.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 20) #include <linux/err.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 21) #include <linux/slab.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 22) #include <linux/preempt.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 23) #include <linux/sched/signal.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 24) #include <linux/sched/stat.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 25) #include <linux/delay.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 26) #include <linux/export.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 27) #include <linux/fs.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 28) #include <linux/anon_inodes.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 29) #include <linux/cpu.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 30) #include <linux/cpumask.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 31) #include <linux/spinlock.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 32) #include <linux/page-flags.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 33) #include <linux/srcu.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 34) #include <linux/miscdevice.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 35) #include <linux/debugfs.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 36) #include <linux/gfp.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 37) #include <linux/vmalloc.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 38) #include <linux/highmem.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 39) #include <linux/hugetlb.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 40) #include <linux/kvm_irqfd.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 41) #include <linux/irqbypass.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 42) #include <linux/module.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 43) #include <linux/compiler.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 44) #include <linux/of.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 45)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 46) #include <asm/ftrace.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 47) #include <asm/reg.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 48) #include <asm/ppc-opcode.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 49) #include <asm/asm-prototypes.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 50) #include <asm/archrandom.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 51) #include <asm/debug.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 52) #include <asm/disassemble.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 53) #include <asm/cputable.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 54) #include <asm/cacheflush.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 55) #include <linux/uaccess.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 56) #include <asm/io.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 57) #include <asm/kvm_ppc.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 58) #include <asm/kvm_book3s.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 59) #include <asm/mmu_context.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 60) #include <asm/lppaca.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 61) #include <asm/pmc.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 62) #include <asm/processor.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 63) #include <asm/cputhreads.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 64) #include <asm/page.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 65) #include <asm/hvcall.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 66) #include <asm/switch_to.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 67) #include <asm/smp.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 68) #include <asm/dbell.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 69) #include <asm/hmi.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 70) #include <asm/pnv-pci.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 71) #include <asm/mmu.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 72) #include <asm/opal.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 73) #include <asm/xics.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 74) #include <asm/xive.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 75) #include <asm/hw_breakpoint.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 76) #include <asm/kvm_book3s_uvmem.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 77) #include <asm/ultravisor.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 78) #include <asm/dtl.h>
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 79)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 80) #include "book3s.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 81)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 82) #define CREATE_TRACE_POINTS
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 83) #include "trace_hv.h"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 84)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 85) /* #define EXIT_DEBUG */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 86) /* #define EXIT_DEBUG_SIMPLE */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 87) /* #define EXIT_DEBUG_INT */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 88)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 89) /* Used to indicate that a guest page fault needs to be handled */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 90) #define RESUME_PAGE_FAULT (RESUME_GUEST | RESUME_FLAG_ARCH1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 91) /* Used to indicate that a guest passthrough interrupt needs to be handled */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 92) #define RESUME_PASSTHROUGH (RESUME_GUEST | RESUME_FLAG_ARCH2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 93)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 94) /* Used as a "null" value for timebase values */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 95) #define TB_NIL (~(u64)0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 96)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 97) static DECLARE_BITMAP(default_enabled_hcalls, MAX_HCALL_OPCODE/4 + 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 98)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 99) static int dynamic_mt_modes = 6;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 100) module_param(dynamic_mt_modes, int, 0644);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 101) MODULE_PARM_DESC(dynamic_mt_modes, "Set of allowed dynamic micro-threading modes: 0 (= none), 2, 4, or 6 (= 2 or 4)");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 102) static int target_smt_mode;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 103) module_param(target_smt_mode, int, 0644);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 104) MODULE_PARM_DESC(target_smt_mode, "Target threads per core (0 = max)");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 105)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 106) static bool indep_threads_mode = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 107) module_param(indep_threads_mode, bool, S_IRUGO | S_IWUSR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 108) MODULE_PARM_DESC(indep_threads_mode, "Independent-threads mode (only on POWER9)");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 109)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 110) static bool one_vm_per_core;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 111) module_param(one_vm_per_core, bool, S_IRUGO | S_IWUSR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 112) MODULE_PARM_DESC(one_vm_per_core, "Only run vCPUs from the same VM on a core (requires indep_threads_mode=N)");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 113)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 114) #ifdef CONFIG_KVM_XICS
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 115) static const struct kernel_param_ops module_param_ops = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 116) .set = param_set_int,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 117) .get = param_get_int,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 118) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 119)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 120) module_param_cb(kvm_irq_bypass, &module_param_ops, &kvm_irq_bypass, 0644);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 121) MODULE_PARM_DESC(kvm_irq_bypass, "Bypass passthrough interrupt optimization");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 122)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 123) module_param_cb(h_ipi_redirect, &module_param_ops, &h_ipi_redirect, 0644);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 124) MODULE_PARM_DESC(h_ipi_redirect, "Redirect H_IPI wakeup to a free host core");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 125) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 126)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 127) /* If set, guests are allowed to create and control nested guests */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 128) static bool nested = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 129) module_param(nested, bool, S_IRUGO | S_IWUSR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 130) MODULE_PARM_DESC(nested, "Enable nested virtualization (only on POWER9)");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 131)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 132) static inline bool nesting_enabled(struct kvm *kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 133) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 134) return kvm->arch.nested_enable && kvm_is_radix(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 135) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 136)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 137) /* If set, the threads on each CPU core have to be in the same MMU mode */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 138) static bool no_mixing_hpt_and_radix;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 139)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 140) static int kvmppc_hv_setup_htab_rma(struct kvm_vcpu *vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 141)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 142) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 143) * RWMR values for POWER8. These control the rate at which PURR
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 144) * and SPURR count and should be set according to the number of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 145) * online threads in the vcore being run.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 146) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 147) #define RWMR_RPA_P8_1THREAD 0x164520C62609AECAUL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 148) #define RWMR_RPA_P8_2THREAD 0x7FFF2908450D8DA9UL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 149) #define RWMR_RPA_P8_3THREAD 0x164520C62609AECAUL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 150) #define RWMR_RPA_P8_4THREAD 0x199A421245058DA9UL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 151) #define RWMR_RPA_P8_5THREAD 0x164520C62609AECAUL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 152) #define RWMR_RPA_P8_6THREAD 0x164520C62609AECAUL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 153) #define RWMR_RPA_P8_7THREAD 0x164520C62609AECAUL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 154) #define RWMR_RPA_P8_8THREAD 0x164520C62609AECAUL
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 155)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 156) static unsigned long p8_rwmr_values[MAX_SMT_THREADS + 1] = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 157) RWMR_RPA_P8_1THREAD,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 158) RWMR_RPA_P8_1THREAD,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 159) RWMR_RPA_P8_2THREAD,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 160) RWMR_RPA_P8_3THREAD,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 161) RWMR_RPA_P8_4THREAD,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 162) RWMR_RPA_P8_5THREAD,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 163) RWMR_RPA_P8_6THREAD,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 164) RWMR_RPA_P8_7THREAD,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 165) RWMR_RPA_P8_8THREAD,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 166) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 167)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 168) static inline struct kvm_vcpu *next_runnable_thread(struct kvmppc_vcore *vc,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 169) int *ip)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 170) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 171) int i = *ip;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 172) struct kvm_vcpu *vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 173)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 174) while (++i < MAX_SMT_THREADS) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 175) vcpu = READ_ONCE(vc->runnable_threads[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 176) if (vcpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 177) *ip = i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 178) return vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 179) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 180) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 181) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 182) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 183)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 184) /* Used to traverse the list of runnable threads for a given vcore */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 185) #define for_each_runnable_thread(i, vcpu, vc) \
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 186) for (i = -1; (vcpu = next_runnable_thread(vc, &i)); )
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 187)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 188) static bool kvmppc_ipi_thread(int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 189) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 190) unsigned long msg = PPC_DBELL_TYPE(PPC_DBELL_SERVER);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 191)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 192) /* If we're a nested hypervisor, fall back to ordinary IPIs for now */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 193) if (kvmhv_on_pseries())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 194) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 195)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 196) /* On POWER9 we can use msgsnd to IPI any cpu */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 197) if (cpu_has_feature(CPU_FTR_ARCH_300)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 198) msg |= get_hard_smp_processor_id(cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 199) smp_mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 200) __asm__ __volatile__ (PPC_MSGSND(%0) : : "r" (msg));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 201) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 202) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 203)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 204) /* On POWER8 for IPIs to threads in the same core, use msgsnd */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 205) if (cpu_has_feature(CPU_FTR_ARCH_207S)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 206) preempt_disable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 207) if (cpu_first_thread_sibling(cpu) ==
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 208) cpu_first_thread_sibling(smp_processor_id())) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 209) msg |= cpu_thread_in_core(cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 210) smp_mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 211) __asm__ __volatile__ (PPC_MSGSND(%0) : : "r" (msg));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 212) preempt_enable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 213) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 214) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 215) preempt_enable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 216) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 217)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 218) #if defined(CONFIG_PPC_ICP_NATIVE) && defined(CONFIG_SMP)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 219) if (cpu >= 0 && cpu < nr_cpu_ids) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 220) if (paca_ptrs[cpu]->kvm_hstate.xics_phys) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 221) xics_wake_cpu(cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 222) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 223) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 224) opal_int_set_mfrr(get_hard_smp_processor_id(cpu), IPI_PRIORITY);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 225) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 226) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 227) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 228)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 229) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 230) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 231)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 232) static void kvmppc_fast_vcpu_kick_hv(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 233) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 234) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 235) struct rcuwait *waitp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 236)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 237) waitp = kvm_arch_vcpu_get_wait(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 238) if (rcuwait_wake_up(waitp))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 239) ++vcpu->stat.halt_wakeup;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 240)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 241) cpu = READ_ONCE(vcpu->arch.thread_cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 242) if (cpu >= 0 && kvmppc_ipi_thread(cpu))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 243) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 244)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 245) /* CPU points to the first thread of the core */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 246) cpu = vcpu->cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 247) if (cpu >= 0 && cpu < nr_cpu_ids && cpu_online(cpu))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 248) smp_send_reschedule(cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 249) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 250)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 251) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 252) * We use the vcpu_load/put functions to measure stolen time.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 253) * Stolen time is counted as time when either the vcpu is able to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 254) * run as part of a virtual core, but the task running the vcore
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 255) * is preempted or sleeping, or when the vcpu needs something done
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 256) * in the kernel by the task running the vcpu, but that task is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 257) * preempted or sleeping. Those two things have to be counted
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 258) * separately, since one of the vcpu tasks will take on the job
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 259) * of running the core, and the other vcpu tasks in the vcore will
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 260) * sleep waiting for it to do that, but that sleep shouldn't count
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 261) * as stolen time.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 262) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 263) * Hence we accumulate stolen time when the vcpu can run as part of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 264) * a vcore using vc->stolen_tb, and the stolen time when the vcpu
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 265) * needs its task to do other things in the kernel (for example,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 266) * service a page fault) in busy_stolen. We don't accumulate
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 267) * stolen time for a vcore when it is inactive, or for a vcpu
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 268) * when it is in state RUNNING or NOTREADY. NOTREADY is a bit of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 269) * a misnomer; it means that the vcpu task is not executing in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 270) * the KVM_VCPU_RUN ioctl, i.e. it is in userspace or elsewhere in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 271) * the kernel. We don't have any way of dividing up that time
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 272) * between time that the vcpu is genuinely stopped, time that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 273) * the task is actively working on behalf of the vcpu, and time
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 274) * that the task is preempted, so we don't count any of it as
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 275) * stolen.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 276) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 277) * Updates to busy_stolen are protected by arch.tbacct_lock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 278) * updates to vc->stolen_tb are protected by the vcore->stoltb_lock
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 279) * lock. The stolen times are measured in units of timebase ticks.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 280) * (Note that the != TB_NIL checks below are purely defensive;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 281) * they should never fail.)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 282) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 283)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 284) static void kvmppc_core_start_stolen(struct kvmppc_vcore *vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 285) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 286) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 287)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 288) spin_lock_irqsave(&vc->stoltb_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 289) vc->preempt_tb = mftb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 290) spin_unlock_irqrestore(&vc->stoltb_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 291) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 292)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 293) static void kvmppc_core_end_stolen(struct kvmppc_vcore *vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 294) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 295) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 296)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 297) spin_lock_irqsave(&vc->stoltb_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 298) if (vc->preempt_tb != TB_NIL) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 299) vc->stolen_tb += mftb() - vc->preempt_tb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 300) vc->preempt_tb = TB_NIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 301) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 302) spin_unlock_irqrestore(&vc->stoltb_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 303) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 304)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 305) static void kvmppc_core_vcpu_load_hv(struct kvm_vcpu *vcpu, int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 306) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 307) struct kvmppc_vcore *vc = vcpu->arch.vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 308) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 309)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 310) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 311) * We can test vc->runner without taking the vcore lock,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 312) * because only this task ever sets vc->runner to this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 313) * vcpu, and once it is set to this vcpu, only this task
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 314) * ever sets it to NULL.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 315) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 316) if (vc->runner == vcpu && vc->vcore_state >= VCORE_SLEEPING)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 317) kvmppc_core_end_stolen(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 318)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 319) spin_lock_irqsave(&vcpu->arch.tbacct_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 320) if (vcpu->arch.state == KVMPPC_VCPU_BUSY_IN_HOST &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 321) vcpu->arch.busy_preempt != TB_NIL) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 322) vcpu->arch.busy_stolen += mftb() - vcpu->arch.busy_preempt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 323) vcpu->arch.busy_preempt = TB_NIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 324) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 325) spin_unlock_irqrestore(&vcpu->arch.tbacct_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 326) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 327)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 328) static void kvmppc_core_vcpu_put_hv(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 329) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 330) struct kvmppc_vcore *vc = vcpu->arch.vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 331) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 332)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 333) if (vc->runner == vcpu && vc->vcore_state >= VCORE_SLEEPING)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 334) kvmppc_core_start_stolen(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 335)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 336) spin_lock_irqsave(&vcpu->arch.tbacct_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 337) if (vcpu->arch.state == KVMPPC_VCPU_BUSY_IN_HOST)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 338) vcpu->arch.busy_preempt = mftb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 339) spin_unlock_irqrestore(&vcpu->arch.tbacct_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 340) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 341)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 342) static void kvmppc_set_pvr_hv(struct kvm_vcpu *vcpu, u32 pvr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 343) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 344) vcpu->arch.pvr = pvr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 345) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 346)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 347) /* Dummy value used in computing PCR value below */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 348) #define PCR_ARCH_31 (PCR_ARCH_300 << 1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 349)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 350) static int kvmppc_set_arch_compat(struct kvm_vcpu *vcpu, u32 arch_compat)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 351) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 352) unsigned long host_pcr_bit = 0, guest_pcr_bit = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 353) struct kvmppc_vcore *vc = vcpu->arch.vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 354)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 355) /* We can (emulate) our own architecture version and anything older */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 356) if (cpu_has_feature(CPU_FTR_ARCH_31))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 357) host_pcr_bit = PCR_ARCH_31;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 358) else if (cpu_has_feature(CPU_FTR_ARCH_300))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 359) host_pcr_bit = PCR_ARCH_300;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 360) else if (cpu_has_feature(CPU_FTR_ARCH_207S))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 361) host_pcr_bit = PCR_ARCH_207;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 362) else if (cpu_has_feature(CPU_FTR_ARCH_206))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 363) host_pcr_bit = PCR_ARCH_206;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 364) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 365) host_pcr_bit = PCR_ARCH_205;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 366)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 367) /* Determine lowest PCR bit needed to run guest in given PVR level */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 368) guest_pcr_bit = host_pcr_bit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 369) if (arch_compat) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 370) switch (arch_compat) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 371) case PVR_ARCH_205:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 372) guest_pcr_bit = PCR_ARCH_205;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 373) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 374) case PVR_ARCH_206:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 375) case PVR_ARCH_206p:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 376) guest_pcr_bit = PCR_ARCH_206;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 377) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 378) case PVR_ARCH_207:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 379) guest_pcr_bit = PCR_ARCH_207;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 380) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 381) case PVR_ARCH_300:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 382) guest_pcr_bit = PCR_ARCH_300;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 383) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 384) case PVR_ARCH_31:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 385) guest_pcr_bit = PCR_ARCH_31;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 386) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 387) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 388) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 389) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 390) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 391)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 392) /* Check requested PCR bits don't exceed our capabilities */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 393) if (guest_pcr_bit > host_pcr_bit)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 394) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 395)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 396) spin_lock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 397) vc->arch_compat = arch_compat;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 398) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 399) * Set all PCR bits for which guest_pcr_bit <= bit < host_pcr_bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 400) * Also set all reserved PCR bits
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 401) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 402) vc->pcr = (host_pcr_bit - guest_pcr_bit) | PCR_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 403) spin_unlock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 404)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 405) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 406) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 407)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 408) static void kvmppc_dump_regs(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 409) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 410) int r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 411)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 412) pr_err("vcpu %p (%d):\n", vcpu, vcpu->vcpu_id);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 413) pr_err("pc = %.16lx msr = %.16llx trap = %x\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 414) vcpu->arch.regs.nip, vcpu->arch.shregs.msr, vcpu->arch.trap);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 415) for (r = 0; r < 16; ++r)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 416) pr_err("r%2d = %.16lx r%d = %.16lx\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 417) r, kvmppc_get_gpr(vcpu, r),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 418) r+16, kvmppc_get_gpr(vcpu, r+16));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 419) pr_err("ctr = %.16lx lr = %.16lx\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 420) vcpu->arch.regs.ctr, vcpu->arch.regs.link);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 421) pr_err("srr0 = %.16llx srr1 = %.16llx\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 422) vcpu->arch.shregs.srr0, vcpu->arch.shregs.srr1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 423) pr_err("sprg0 = %.16llx sprg1 = %.16llx\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 424) vcpu->arch.shregs.sprg0, vcpu->arch.shregs.sprg1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 425) pr_err("sprg2 = %.16llx sprg3 = %.16llx\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 426) vcpu->arch.shregs.sprg2, vcpu->arch.shregs.sprg3);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 427) pr_err("cr = %.8lx xer = %.16lx dsisr = %.8x\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 428) vcpu->arch.regs.ccr, vcpu->arch.regs.xer, vcpu->arch.shregs.dsisr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 429) pr_err("dar = %.16llx\n", vcpu->arch.shregs.dar);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 430) pr_err("fault dar = %.16lx dsisr = %.8x\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 431) vcpu->arch.fault_dar, vcpu->arch.fault_dsisr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 432) pr_err("SLB (%d entries):\n", vcpu->arch.slb_max);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 433) for (r = 0; r < vcpu->arch.slb_max; ++r)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 434) pr_err(" ESID = %.16llx VSID = %.16llx\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 435) vcpu->arch.slb[r].orige, vcpu->arch.slb[r].origv);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 436) pr_err("lpcr = %.16lx sdr1 = %.16lx last_inst = %.8x\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 437) vcpu->arch.vcore->lpcr, vcpu->kvm->arch.sdr1,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 438) vcpu->arch.last_inst);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 439) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 440)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 441) static struct kvm_vcpu *kvmppc_find_vcpu(struct kvm *kvm, int id)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 442) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 443) return kvm_get_vcpu_by_id(kvm, id);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 444) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 445)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 446) static void init_vpa(struct kvm_vcpu *vcpu, struct lppaca *vpa)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 447) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 448) vpa->__old_status |= LPPACA_OLD_SHARED_PROC;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 449) vpa->yield_count = cpu_to_be32(1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 450) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 451)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 452) static int set_vpa(struct kvm_vcpu *vcpu, struct kvmppc_vpa *v,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 453) unsigned long addr, unsigned long len)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 454) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 455) /* check address is cacheline aligned */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 456) if (addr & (L1_CACHE_BYTES - 1))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 457) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 458) spin_lock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 459) if (v->next_gpa != addr || v->len != len) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 460) v->next_gpa = addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 461) v->len = addr ? len : 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 462) v->update_pending = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 463) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 464) spin_unlock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 465) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 466) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 467)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 468) /* Length for a per-processor buffer is passed in at offset 4 in the buffer */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 469) struct reg_vpa {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 470) u32 dummy;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 471) union {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 472) __be16 hword;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 473) __be32 word;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 474) } length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 475) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 476)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 477) static int vpa_is_registered(struct kvmppc_vpa *vpap)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 478) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 479) if (vpap->update_pending)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 480) return vpap->next_gpa != 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 481) return vpap->pinned_addr != NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 482) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 483)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 484) static unsigned long do_h_register_vpa(struct kvm_vcpu *vcpu,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 485) unsigned long flags,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 486) unsigned long vcpuid, unsigned long vpa)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 487) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 488) struct kvm *kvm = vcpu->kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 489) unsigned long len, nb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 490) void *va;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 491) struct kvm_vcpu *tvcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 492) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 493) int subfunc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 494) struct kvmppc_vpa *vpap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 495)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 496) tvcpu = kvmppc_find_vcpu(kvm, vcpuid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 497) if (!tvcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 498) return H_PARAMETER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 499)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 500) subfunc = (flags >> H_VPA_FUNC_SHIFT) & H_VPA_FUNC_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 501) if (subfunc == H_VPA_REG_VPA || subfunc == H_VPA_REG_DTL ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 502) subfunc == H_VPA_REG_SLB) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 503) /* Registering new area - address must be cache-line aligned */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 504) if ((vpa & (L1_CACHE_BYTES - 1)) || !vpa)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 505) return H_PARAMETER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 506)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 507) /* convert logical addr to kernel addr and read length */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 508) va = kvmppc_pin_guest_page(kvm, vpa, &nb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 509) if (va == NULL)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 510) return H_PARAMETER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 511) if (subfunc == H_VPA_REG_VPA)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 512) len = be16_to_cpu(((struct reg_vpa *)va)->length.hword);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 513) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 514) len = be32_to_cpu(((struct reg_vpa *)va)->length.word);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 515) kvmppc_unpin_guest_page(kvm, va, vpa, false);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 516)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 517) /* Check length */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 518) if (len > nb || len < sizeof(struct reg_vpa))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 519) return H_PARAMETER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 520) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 521) vpa = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 522) len = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 523) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 524)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 525) err = H_PARAMETER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 526) vpap = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 527) spin_lock(&tvcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 528)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 529) switch (subfunc) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 530) case H_VPA_REG_VPA: /* register VPA */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 531) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 532) * The size of our lppaca is 1kB because of the way we align
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 533) * it for the guest to avoid crossing a 4kB boundary. We only
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 534) * use 640 bytes of the structure though, so we should accept
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 535) * clients that set a size of 640.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 536) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 537) BUILD_BUG_ON(sizeof(struct lppaca) != 640);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 538) if (len < sizeof(struct lppaca))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 539) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 540) vpap = &tvcpu->arch.vpa;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 541) err = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 542) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 543)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 544) case H_VPA_REG_DTL: /* register DTL */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 545) if (len < sizeof(struct dtl_entry))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 546) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 547) len -= len % sizeof(struct dtl_entry);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 548)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 549) /* Check that they have previously registered a VPA */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 550) err = H_RESOURCE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 551) if (!vpa_is_registered(&tvcpu->arch.vpa))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 552) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 553)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 554) vpap = &tvcpu->arch.dtl;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 555) err = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 556) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 557)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 558) case H_VPA_REG_SLB: /* register SLB shadow buffer */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 559) /* Check that they have previously registered a VPA */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 560) err = H_RESOURCE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 561) if (!vpa_is_registered(&tvcpu->arch.vpa))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 562) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 563)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 564) vpap = &tvcpu->arch.slb_shadow;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 565) err = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 566) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 567)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 568) case H_VPA_DEREG_VPA: /* deregister VPA */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 569) /* Check they don't still have a DTL or SLB buf registered */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 570) err = H_RESOURCE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 571) if (vpa_is_registered(&tvcpu->arch.dtl) ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 572) vpa_is_registered(&tvcpu->arch.slb_shadow))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 573) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 574)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 575) vpap = &tvcpu->arch.vpa;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 576) err = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 577) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 578)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 579) case H_VPA_DEREG_DTL: /* deregister DTL */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 580) vpap = &tvcpu->arch.dtl;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 581) err = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 582) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 583)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 584) case H_VPA_DEREG_SLB: /* deregister SLB shadow buffer */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 585) vpap = &tvcpu->arch.slb_shadow;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 586) err = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 587) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 588) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 589)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 590) if (vpap) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 591) vpap->next_gpa = vpa;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 592) vpap->len = len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 593) vpap->update_pending = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 594) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 595)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 596) spin_unlock(&tvcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 597)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 598) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 599) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 600)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 601) static void kvmppc_update_vpa(struct kvm_vcpu *vcpu, struct kvmppc_vpa *vpap)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 602) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 603) struct kvm *kvm = vcpu->kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 604) void *va;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 605) unsigned long nb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 606) unsigned long gpa;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 607)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 608) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 609) * We need to pin the page pointed to by vpap->next_gpa,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 610) * but we can't call kvmppc_pin_guest_page under the lock
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 611) * as it does get_user_pages() and down_read(). So we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 612) * have to drop the lock, pin the page, then get the lock
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 613) * again and check that a new area didn't get registered
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 614) * in the meantime.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 615) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 616) for (;;) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 617) gpa = vpap->next_gpa;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 618) spin_unlock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 619) va = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 620) nb = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 621) if (gpa)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 622) va = kvmppc_pin_guest_page(kvm, gpa, &nb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 623) spin_lock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 624) if (gpa == vpap->next_gpa)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 625) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 626) /* sigh... unpin that one and try again */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 627) if (va)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 628) kvmppc_unpin_guest_page(kvm, va, gpa, false);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 629) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 630)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 631) vpap->update_pending = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 632) if (va && nb < vpap->len) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 633) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 634) * If it's now too short, it must be that userspace
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 635) * has changed the mappings underlying guest memory,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 636) * so unregister the region.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 637) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 638) kvmppc_unpin_guest_page(kvm, va, gpa, false);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 639) va = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 640) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 641) if (vpap->pinned_addr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 642) kvmppc_unpin_guest_page(kvm, vpap->pinned_addr, vpap->gpa,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 643) vpap->dirty);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 644) vpap->gpa = gpa;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 645) vpap->pinned_addr = va;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 646) vpap->dirty = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 647) if (va)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 648) vpap->pinned_end = va + vpap->len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 649) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 650)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 651) static void kvmppc_update_vpas(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 652) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 653) if (!(vcpu->arch.vpa.update_pending ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 654) vcpu->arch.slb_shadow.update_pending ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 655) vcpu->arch.dtl.update_pending))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 656) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 657)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 658) spin_lock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 659) if (vcpu->arch.vpa.update_pending) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 660) kvmppc_update_vpa(vcpu, &vcpu->arch.vpa);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 661) if (vcpu->arch.vpa.pinned_addr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 662) init_vpa(vcpu, vcpu->arch.vpa.pinned_addr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 663) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 664) if (vcpu->arch.dtl.update_pending) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 665) kvmppc_update_vpa(vcpu, &vcpu->arch.dtl);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 666) vcpu->arch.dtl_ptr = vcpu->arch.dtl.pinned_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 667) vcpu->arch.dtl_index = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 668) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 669) if (vcpu->arch.slb_shadow.update_pending)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 670) kvmppc_update_vpa(vcpu, &vcpu->arch.slb_shadow);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 671) spin_unlock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 672) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 673)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 674) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 675) * Return the accumulated stolen time for the vcore up until `now'.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 676) * The caller should hold the vcore lock.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 677) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 678) static u64 vcore_stolen_time(struct kvmppc_vcore *vc, u64 now)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 679) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 680) u64 p;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 681) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 682)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 683) spin_lock_irqsave(&vc->stoltb_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 684) p = vc->stolen_tb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 685) if (vc->vcore_state != VCORE_INACTIVE &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 686) vc->preempt_tb != TB_NIL)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 687) p += now - vc->preempt_tb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 688) spin_unlock_irqrestore(&vc->stoltb_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 689) return p;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 690) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 691)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 692) static void kvmppc_create_dtl_entry(struct kvm_vcpu *vcpu,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 693) struct kvmppc_vcore *vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 694) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 695) struct dtl_entry *dt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 696) struct lppaca *vpa;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 697) unsigned long stolen;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 698) unsigned long core_stolen;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 699) u64 now;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 700) unsigned long flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 701)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 702) dt = vcpu->arch.dtl_ptr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 703) vpa = vcpu->arch.vpa.pinned_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 704) now = mftb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 705) core_stolen = vcore_stolen_time(vc, now);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 706) stolen = core_stolen - vcpu->arch.stolen_logged;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 707) vcpu->arch.stolen_logged = core_stolen;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 708) spin_lock_irqsave(&vcpu->arch.tbacct_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 709) stolen += vcpu->arch.busy_stolen;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 710) vcpu->arch.busy_stolen = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 711) spin_unlock_irqrestore(&vcpu->arch.tbacct_lock, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 712) if (!dt || !vpa)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 713) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 714) memset(dt, 0, sizeof(struct dtl_entry));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 715) dt->dispatch_reason = 7;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 716) dt->processor_id = cpu_to_be16(vc->pcpu + vcpu->arch.ptid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 717) dt->timebase = cpu_to_be64(now + vc->tb_offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 718) dt->enqueue_to_dispatch_time = cpu_to_be32(stolen);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 719) dt->srr0 = cpu_to_be64(kvmppc_get_pc(vcpu));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 720) dt->srr1 = cpu_to_be64(vcpu->arch.shregs.msr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 721) ++dt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 722) if (dt == vcpu->arch.dtl.pinned_end)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 723) dt = vcpu->arch.dtl.pinned_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 724) vcpu->arch.dtl_ptr = dt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 725) /* order writing *dt vs. writing vpa->dtl_idx */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 726) smp_wmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 727) vpa->dtl_idx = cpu_to_be64(++vcpu->arch.dtl_index);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 728) vcpu->arch.dtl.dirty = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 729) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 730)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 731) /* See if there is a doorbell interrupt pending for a vcpu */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 732) static bool kvmppc_doorbell_pending(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 733) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 734) int thr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 735) struct kvmppc_vcore *vc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 736)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 737) if (vcpu->arch.doorbell_request)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 738) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 739) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 740) * Ensure that the read of vcore->dpdes comes after the read
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 741) * of vcpu->doorbell_request. This barrier matches the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 742) * smp_wmb() in kvmppc_guest_entry_inject().
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 743) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 744) smp_rmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 745) vc = vcpu->arch.vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 746) thr = vcpu->vcpu_id - vc->first_vcpuid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 747) return !!(vc->dpdes & (1 << thr));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 748) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 749)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 750) static bool kvmppc_power8_compatible(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 751) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 752) if (vcpu->arch.vcore->arch_compat >= PVR_ARCH_207)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 753) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 754) if ((!vcpu->arch.vcore->arch_compat) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 755) cpu_has_feature(CPU_FTR_ARCH_207S))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 756) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 757) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 758) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 759)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 760) static int kvmppc_h_set_mode(struct kvm_vcpu *vcpu, unsigned long mflags,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 761) unsigned long resource, unsigned long value1,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 762) unsigned long value2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 763) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 764) switch (resource) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 765) case H_SET_MODE_RESOURCE_SET_CIABR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 766) if (!kvmppc_power8_compatible(vcpu))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 767) return H_P2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 768) if (value2)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 769) return H_P4;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 770) if (mflags)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 771) return H_UNSUPPORTED_FLAG_START;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 772) /* Guests can't breakpoint the hypervisor */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 773) if ((value1 & CIABR_PRIV) == CIABR_PRIV_HYPER)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 774) return H_P3;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 775) vcpu->arch.ciabr = value1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 776) return H_SUCCESS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 777) case H_SET_MODE_RESOURCE_SET_DAWR0:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 778) if (!kvmppc_power8_compatible(vcpu))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 779) return H_P2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 780) if (!ppc_breakpoint_available())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 781) return H_P2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 782) if (mflags)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 783) return H_UNSUPPORTED_FLAG_START;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 784) if (value2 & DABRX_HYP)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 785) return H_P4;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 786) vcpu->arch.dawr = value1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 787) vcpu->arch.dawrx = value2;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 788) return H_SUCCESS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 789) case H_SET_MODE_RESOURCE_ADDR_TRANS_MODE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 790) /* KVM does not support mflags=2 (AIL=2) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 791) if (mflags != 0 && mflags != 3)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 792) return H_UNSUPPORTED_FLAG_START;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 793) return H_TOO_HARD;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 794) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 795) return H_TOO_HARD;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 796) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 797) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 798)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 799) /* Copy guest memory in place - must reside within a single memslot */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 800) static int kvmppc_copy_guest(struct kvm *kvm, gpa_t to, gpa_t from,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 801) unsigned long len)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 802) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 803) struct kvm_memory_slot *to_memslot = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 804) struct kvm_memory_slot *from_memslot = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 805) unsigned long to_addr, from_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 806) int r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 807)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 808) /* Get HPA for from address */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 809) from_memslot = gfn_to_memslot(kvm, from >> PAGE_SHIFT);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 810) if (!from_memslot)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 811) return -EFAULT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 812) if ((from + len) >= ((from_memslot->base_gfn + from_memslot->npages)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 813) << PAGE_SHIFT))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 814) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 815) from_addr = gfn_to_hva_memslot(from_memslot, from >> PAGE_SHIFT);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 816) if (kvm_is_error_hva(from_addr))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 817) return -EFAULT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 818) from_addr |= (from & (PAGE_SIZE - 1));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 819)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 820) /* Get HPA for to address */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 821) to_memslot = gfn_to_memslot(kvm, to >> PAGE_SHIFT);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 822) if (!to_memslot)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 823) return -EFAULT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 824) if ((to + len) >= ((to_memslot->base_gfn + to_memslot->npages)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 825) << PAGE_SHIFT))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 826) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 827) to_addr = gfn_to_hva_memslot(to_memslot, to >> PAGE_SHIFT);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 828) if (kvm_is_error_hva(to_addr))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 829) return -EFAULT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 830) to_addr |= (to & (PAGE_SIZE - 1));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 831)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 832) /* Perform copy */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 833) r = raw_copy_in_user((void __user *)to_addr, (void __user *)from_addr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 834) len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 835) if (r)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 836) return -EFAULT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 837) mark_page_dirty(kvm, to >> PAGE_SHIFT);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 838) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 839) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 840)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 841) static long kvmppc_h_page_init(struct kvm_vcpu *vcpu, unsigned long flags,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 842) unsigned long dest, unsigned long src)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 843) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 844) u64 pg_sz = SZ_4K; /* 4K page size */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 845) u64 pg_mask = SZ_4K - 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 846) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 847)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 848) /* Check for invalid flags (H_PAGE_SET_LOANED covers all CMO flags) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 849) if (flags & ~(H_ICACHE_INVALIDATE | H_ICACHE_SYNCHRONIZE |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 850) H_ZERO_PAGE | H_COPY_PAGE | H_PAGE_SET_LOANED))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 851) return H_PARAMETER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 852)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 853) /* dest (and src if copy_page flag set) must be page aligned */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 854) if ((dest & pg_mask) || ((flags & H_COPY_PAGE) && (src & pg_mask)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 855) return H_PARAMETER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 856)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 857) /* zero and/or copy the page as determined by the flags */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 858) if (flags & H_COPY_PAGE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 859) ret = kvmppc_copy_guest(vcpu->kvm, dest, src, pg_sz);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 860) if (ret < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 861) return H_PARAMETER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 862) } else if (flags & H_ZERO_PAGE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 863) ret = kvm_clear_guest(vcpu->kvm, dest, pg_sz);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 864) if (ret < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 865) return H_PARAMETER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 866) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 867)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 868) /* We can ignore the remaining flags */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 869)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 870) return H_SUCCESS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 871) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 872)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 873) static int kvm_arch_vcpu_yield_to(struct kvm_vcpu *target)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 874) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 875) struct kvmppc_vcore *vcore = target->arch.vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 876)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 877) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 878) * We expect to have been called by the real mode handler
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 879) * (kvmppc_rm_h_confer()) which would have directly returned
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 880) * H_SUCCESS if the source vcore wasn't idle (e.g. if it may
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 881) * have useful work to do and should not confer) so we don't
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 882) * recheck that here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 883) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 884)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 885) spin_lock(&vcore->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 886) if (target->arch.state == KVMPPC_VCPU_RUNNABLE &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 887) vcore->vcore_state != VCORE_INACTIVE &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 888) vcore->runner)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 889) target = vcore->runner;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 890) spin_unlock(&vcore->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 891)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 892) return kvm_vcpu_yield_to(target);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 893) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 894)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 895) static int kvmppc_get_yield_count(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 896) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 897) int yield_count = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 898) struct lppaca *lppaca;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 899)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 900) spin_lock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 901) lppaca = (struct lppaca *)vcpu->arch.vpa.pinned_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 902) if (lppaca)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 903) yield_count = be32_to_cpu(lppaca->yield_count);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 904) spin_unlock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 905) return yield_count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 906) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 907)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 908) int kvmppc_pseries_do_hcall(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 909) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 910) unsigned long req = kvmppc_get_gpr(vcpu, 3);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 911) unsigned long target, ret = H_SUCCESS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 912) int yield_count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 913) struct kvm_vcpu *tvcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 914) int idx, rc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 915)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 916) if (req <= MAX_HCALL_OPCODE &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 917) !test_bit(req/4, vcpu->kvm->arch.enabled_hcalls))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 918) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 919)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 920) switch (req) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 921) case H_CEDE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 922) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 923) case H_PROD:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 924) target = kvmppc_get_gpr(vcpu, 4);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 925) tvcpu = kvmppc_find_vcpu(vcpu->kvm, target);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 926) if (!tvcpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 927) ret = H_PARAMETER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 928) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 929) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 930) tvcpu->arch.prodded = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 931) smp_mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 932) if (tvcpu->arch.ceded)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 933) kvmppc_fast_vcpu_kick_hv(tvcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 934) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 935) case H_CONFER:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 936) target = kvmppc_get_gpr(vcpu, 4);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 937) if (target == -1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 938) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 939) tvcpu = kvmppc_find_vcpu(vcpu->kvm, target);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 940) if (!tvcpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 941) ret = H_PARAMETER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 942) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 943) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 944) yield_count = kvmppc_get_gpr(vcpu, 5);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 945) if (kvmppc_get_yield_count(tvcpu) != yield_count)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 946) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 947) kvm_arch_vcpu_yield_to(tvcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 948) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 949) case H_REGISTER_VPA:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 950) ret = do_h_register_vpa(vcpu, kvmppc_get_gpr(vcpu, 4),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 951) kvmppc_get_gpr(vcpu, 5),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 952) kvmppc_get_gpr(vcpu, 6));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 953) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 954) case H_RTAS:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 955) if (list_empty(&vcpu->kvm->arch.rtas_tokens))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 956) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 957)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 958) idx = srcu_read_lock(&vcpu->kvm->srcu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 959) rc = kvmppc_rtas_hcall(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 960) srcu_read_unlock(&vcpu->kvm->srcu, idx);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 961)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 962) if (rc == -ENOENT)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 963) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 964) else if (rc == 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 965) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 966)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 967) /* Send the error out to userspace via KVM_RUN */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 968) return rc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 969) case H_LOGICAL_CI_LOAD:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 970) ret = kvmppc_h_logical_ci_load(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 971) if (ret == H_TOO_HARD)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 972) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 973) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 974) case H_LOGICAL_CI_STORE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 975) ret = kvmppc_h_logical_ci_store(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 976) if (ret == H_TOO_HARD)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 977) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 978) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 979) case H_SET_MODE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 980) ret = kvmppc_h_set_mode(vcpu, kvmppc_get_gpr(vcpu, 4),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 981) kvmppc_get_gpr(vcpu, 5),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 982) kvmppc_get_gpr(vcpu, 6),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 983) kvmppc_get_gpr(vcpu, 7));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 984) if (ret == H_TOO_HARD)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 985) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 986) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 987) case H_XIRR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 988) case H_CPPR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 989) case H_EOI:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 990) case H_IPI:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 991) case H_IPOLL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 992) case H_XIRR_X:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 993) if (kvmppc_xics_enabled(vcpu)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 994) if (xics_on_xive()) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 995) ret = H_NOT_AVAILABLE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 996) return RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 997) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 998) ret = kvmppc_xics_hcall(vcpu, req);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 999) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1000) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1001) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1002) case H_SET_DABR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1003) ret = kvmppc_h_set_dabr(vcpu, kvmppc_get_gpr(vcpu, 4));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1004) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1005) case H_SET_XDABR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1006) ret = kvmppc_h_set_xdabr(vcpu, kvmppc_get_gpr(vcpu, 4),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1007) kvmppc_get_gpr(vcpu, 5));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1008) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1009) #ifdef CONFIG_SPAPR_TCE_IOMMU
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1010) case H_GET_TCE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1011) ret = kvmppc_h_get_tce(vcpu, kvmppc_get_gpr(vcpu, 4),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1012) kvmppc_get_gpr(vcpu, 5));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1013) if (ret == H_TOO_HARD)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1014) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1015) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1016) case H_PUT_TCE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1017) ret = kvmppc_h_put_tce(vcpu, kvmppc_get_gpr(vcpu, 4),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1018) kvmppc_get_gpr(vcpu, 5),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1019) kvmppc_get_gpr(vcpu, 6));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1020) if (ret == H_TOO_HARD)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1021) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1022) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1023) case H_PUT_TCE_INDIRECT:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1024) ret = kvmppc_h_put_tce_indirect(vcpu, kvmppc_get_gpr(vcpu, 4),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1025) kvmppc_get_gpr(vcpu, 5),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1026) kvmppc_get_gpr(vcpu, 6),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1027) kvmppc_get_gpr(vcpu, 7));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1028) if (ret == H_TOO_HARD)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1029) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1030) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1031) case H_STUFF_TCE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1032) ret = kvmppc_h_stuff_tce(vcpu, kvmppc_get_gpr(vcpu, 4),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1033) kvmppc_get_gpr(vcpu, 5),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1034) kvmppc_get_gpr(vcpu, 6),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1035) kvmppc_get_gpr(vcpu, 7));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1036) if (ret == H_TOO_HARD)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1037) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1038) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1039) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1040) case H_RANDOM:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1041) if (!powernv_get_random_long(&vcpu->arch.regs.gpr[4]))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1042) ret = H_HARDWARE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1043) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1044)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1045) case H_SET_PARTITION_TABLE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1046) ret = H_FUNCTION;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1047) if (nesting_enabled(vcpu->kvm))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1048) ret = kvmhv_set_partition_table(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1049) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1050) case H_ENTER_NESTED:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1051) ret = H_FUNCTION;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1052) if (!nesting_enabled(vcpu->kvm))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1053) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1054) ret = kvmhv_enter_nested_guest(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1055) if (ret == H_INTERRUPT) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1056) kvmppc_set_gpr(vcpu, 3, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1057) vcpu->arch.hcall_needed = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1058) return -EINTR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1059) } else if (ret == H_TOO_HARD) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1060) kvmppc_set_gpr(vcpu, 3, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1061) vcpu->arch.hcall_needed = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1062) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1063) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1064) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1065) case H_TLB_INVALIDATE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1066) ret = H_FUNCTION;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1067) if (nesting_enabled(vcpu->kvm))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1068) ret = kvmhv_do_nested_tlbie(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1069) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1070) case H_COPY_TOFROM_GUEST:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1071) ret = H_FUNCTION;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1072) if (nesting_enabled(vcpu->kvm))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1073) ret = kvmhv_copy_tofrom_guest_nested(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1074) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1075) case H_PAGE_INIT:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1076) ret = kvmppc_h_page_init(vcpu, kvmppc_get_gpr(vcpu, 4),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1077) kvmppc_get_gpr(vcpu, 5),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1078) kvmppc_get_gpr(vcpu, 6));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1079) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1080) case H_SVM_PAGE_IN:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1081) ret = H_UNSUPPORTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1082) if (kvmppc_get_srr1(vcpu) & MSR_S)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1083) ret = kvmppc_h_svm_page_in(vcpu->kvm,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1084) kvmppc_get_gpr(vcpu, 4),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1085) kvmppc_get_gpr(vcpu, 5),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1086) kvmppc_get_gpr(vcpu, 6));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1087) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1088) case H_SVM_PAGE_OUT:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1089) ret = H_UNSUPPORTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1090) if (kvmppc_get_srr1(vcpu) & MSR_S)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1091) ret = kvmppc_h_svm_page_out(vcpu->kvm,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1092) kvmppc_get_gpr(vcpu, 4),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1093) kvmppc_get_gpr(vcpu, 5),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1094) kvmppc_get_gpr(vcpu, 6));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1095) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1096) case H_SVM_INIT_START:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1097) ret = H_UNSUPPORTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1098) if (kvmppc_get_srr1(vcpu) & MSR_S)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1099) ret = kvmppc_h_svm_init_start(vcpu->kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1100) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1101) case H_SVM_INIT_DONE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1102) ret = H_UNSUPPORTED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1103) if (kvmppc_get_srr1(vcpu) & MSR_S)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1104) ret = kvmppc_h_svm_init_done(vcpu->kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1105) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1106) case H_SVM_INIT_ABORT:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1107) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1108) * Even if that call is made by the Ultravisor, the SSR1 value
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1109) * is the guest context one, with the secure bit clear as it has
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1110) * not yet been secured. So we can't check it here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1111) * Instead the kvm->arch.secure_guest flag is checked inside
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1112) * kvmppc_h_svm_init_abort().
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1113) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1114) ret = kvmppc_h_svm_init_abort(vcpu->kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1115) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1116)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1117) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1118) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1119) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1120) kvmppc_set_gpr(vcpu, 3, ret);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1121) vcpu->arch.hcall_needed = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1122) return RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1123) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1124)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1125) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1126) * Handle H_CEDE in the nested virtualization case where we haven't
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1127) * called the real-mode hcall handlers in book3s_hv_rmhandlers.S.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1128) * This has to be done early, not in kvmppc_pseries_do_hcall(), so
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1129) * that the cede logic in kvmppc_run_single_vcpu() works properly.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1130) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1131) static void kvmppc_nested_cede(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1132) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1133) vcpu->arch.shregs.msr |= MSR_EE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1134) vcpu->arch.ceded = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1135) smp_mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1136) if (vcpu->arch.prodded) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1137) vcpu->arch.prodded = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1138) smp_mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1139) vcpu->arch.ceded = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1140) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1141) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1142)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1143) static int kvmppc_hcall_impl_hv(unsigned long cmd)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1144) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1145) switch (cmd) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1146) case H_CEDE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1147) case H_PROD:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1148) case H_CONFER:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1149) case H_REGISTER_VPA:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1150) case H_SET_MODE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1151) case H_LOGICAL_CI_LOAD:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1152) case H_LOGICAL_CI_STORE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1153) #ifdef CONFIG_KVM_XICS
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1154) case H_XIRR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1155) case H_CPPR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1156) case H_EOI:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1157) case H_IPI:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1158) case H_IPOLL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1159) case H_XIRR_X:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1160) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1161) case H_PAGE_INIT:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1162) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1163) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1164)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1165) /* See if it's in the real-mode table */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1166) return kvmppc_hcall_impl_hv_realmode(cmd);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1167) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1168)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1169) static int kvmppc_emulate_debug_inst(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1170) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1171) u32 last_inst;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1172)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1173) if (kvmppc_get_last_inst(vcpu, INST_GENERIC, &last_inst) !=
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1174) EMULATE_DONE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1175) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1176) * Fetch failed, so return to guest and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1177) * try executing it again.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1178) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1179) return RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1180) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1181)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1182) if (last_inst == KVMPPC_INST_SW_BREAKPOINT) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1183) vcpu->run->exit_reason = KVM_EXIT_DEBUG;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1184) vcpu->run->debug.arch.address = kvmppc_get_pc(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1185) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1186) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1187) kvmppc_core_queue_program(vcpu, SRR1_PROGILL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1188) return RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1189) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1190) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1191)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1192) static void do_nothing(void *x)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1193) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1194) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1195)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1196) static unsigned long kvmppc_read_dpdes(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1197) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1198) int thr, cpu, pcpu, nthreads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1199) struct kvm_vcpu *v;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1200) unsigned long dpdes;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1201)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1202) nthreads = vcpu->kvm->arch.emul_smt_mode;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1203) dpdes = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1204) cpu = vcpu->vcpu_id & ~(nthreads - 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1205) for (thr = 0; thr < nthreads; ++thr, ++cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1206) v = kvmppc_find_vcpu(vcpu->kvm, cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1207) if (!v)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1208) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1209) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1210) * If the vcpu is currently running on a physical cpu thread,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1211) * interrupt it in order to pull it out of the guest briefly,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1212) * which will update its vcore->dpdes value.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1213) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1214) pcpu = READ_ONCE(v->cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1215) if (pcpu >= 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1216) smp_call_function_single(pcpu, do_nothing, NULL, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1217) if (kvmppc_doorbell_pending(v))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1218) dpdes |= 1 << thr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1219) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1220) return dpdes;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1221) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1222)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1223) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1224) * On POWER9, emulate doorbell-related instructions in order to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1225) * give the guest the illusion of running on a multi-threaded core.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1226) * The instructions emulated are msgsndp, msgclrp, mfspr TIR,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1227) * and mfspr DPDES.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1228) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1229) static int kvmppc_emulate_doorbell_instr(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1230) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1231) u32 inst, rb, thr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1232) unsigned long arg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1233) struct kvm *kvm = vcpu->kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1234) struct kvm_vcpu *tvcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1235)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1236) if (kvmppc_get_last_inst(vcpu, INST_GENERIC, &inst) != EMULATE_DONE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1237) return RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1238) if (get_op(inst) != 31)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1239) return EMULATE_FAIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1240) rb = get_rb(inst);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1241) thr = vcpu->vcpu_id & (kvm->arch.emul_smt_mode - 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1242) switch (get_xop(inst)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1243) case OP_31_XOP_MSGSNDP:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1244) arg = kvmppc_get_gpr(vcpu, rb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1245) if (((arg >> 27) & 0xf) != PPC_DBELL_SERVER)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1246) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1247) arg &= 0x3f;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1248) if (arg >= kvm->arch.emul_smt_mode)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1249) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1250) tvcpu = kvmppc_find_vcpu(kvm, vcpu->vcpu_id - thr + arg);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1251) if (!tvcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1252) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1253) if (!tvcpu->arch.doorbell_request) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1254) tvcpu->arch.doorbell_request = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1255) kvmppc_fast_vcpu_kick_hv(tvcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1256) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1257) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1258) case OP_31_XOP_MSGCLRP:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1259) arg = kvmppc_get_gpr(vcpu, rb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1260) if (((arg >> 27) & 0xf) != PPC_DBELL_SERVER)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1261) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1262) vcpu->arch.vcore->dpdes = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1263) vcpu->arch.doorbell_request = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1264) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1265) case OP_31_XOP_MFSPR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1266) switch (get_sprn(inst)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1267) case SPRN_TIR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1268) arg = thr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1269) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1270) case SPRN_DPDES:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1271) arg = kvmppc_read_dpdes(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1272) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1273) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1274) return EMULATE_FAIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1275) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1276) kvmppc_set_gpr(vcpu, get_rt(inst), arg);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1277) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1278) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1279) return EMULATE_FAIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1280) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1281) kvmppc_set_pc(vcpu, kvmppc_get_pc(vcpu) + 4);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1282) return RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1283) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1284)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1285) static int kvmppc_handle_exit_hv(struct kvm_vcpu *vcpu,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1286) struct task_struct *tsk)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1287) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1288) struct kvm_run *run = vcpu->run;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1289) int r = RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1290)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1291) vcpu->stat.sum_exits++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1292)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1293) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1294) * This can happen if an interrupt occurs in the last stages
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1295) * of guest entry or the first stages of guest exit (i.e. after
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1296) * setting paca->kvm_hstate.in_guest to KVM_GUEST_MODE_GUEST_HV
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1297) * and before setting it to KVM_GUEST_MODE_HOST_HV).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1298) * That can happen due to a bug, or due to a machine check
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1299) * occurring at just the wrong time.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1300) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1301) if (vcpu->arch.shregs.msr & MSR_HV) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1302) printk(KERN_EMERG "KVM trap in HV mode!\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1303) printk(KERN_EMERG "trap=0x%x | pc=0x%lx | msr=0x%llx\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1304) vcpu->arch.trap, kvmppc_get_pc(vcpu),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1305) vcpu->arch.shregs.msr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1306) kvmppc_dump_regs(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1307) run->exit_reason = KVM_EXIT_INTERNAL_ERROR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1308) run->hw.hardware_exit_reason = vcpu->arch.trap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1309) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1310) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1311) run->exit_reason = KVM_EXIT_UNKNOWN;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1312) run->ready_for_interrupt_injection = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1313) switch (vcpu->arch.trap) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1314) /* We're good on these - the host merely wanted to get our attention */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1315) case BOOK3S_INTERRUPT_HV_DECREMENTER:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1316) vcpu->stat.dec_exits++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1317) r = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1318) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1319) case BOOK3S_INTERRUPT_EXTERNAL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1320) case BOOK3S_INTERRUPT_H_DOORBELL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1321) case BOOK3S_INTERRUPT_H_VIRT:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1322) vcpu->stat.ext_intr_exits++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1323) r = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1324) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1325) /* SR/HMI/PMI are HV interrupts that host has handled. Resume guest.*/
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1326) case BOOK3S_INTERRUPT_HMI:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1327) case BOOK3S_INTERRUPT_PERFMON:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1328) case BOOK3S_INTERRUPT_SYSTEM_RESET:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1329) r = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1330) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1331) case BOOK3S_INTERRUPT_MACHINE_CHECK:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1332) /* Print the MCE event to host console. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1333) machine_check_print_event_info(&vcpu->arch.mce_evt, false, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1334)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1335) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1336) * If the guest can do FWNMI, exit to userspace so it can
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1337) * deliver a FWNMI to the guest.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1338) * Otherwise we synthesize a machine check for the guest
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1339) * so that it knows that the machine check occurred.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1340) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1341) if (!vcpu->kvm->arch.fwnmi_enabled) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1342) ulong flags = vcpu->arch.shregs.msr & 0x083c0000;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1343) kvmppc_core_queue_machine_check(vcpu, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1344) r = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1345) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1346) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1347)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1348) /* Exit to guest with KVM_EXIT_NMI as exit reason */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1349) run->exit_reason = KVM_EXIT_NMI;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1350) run->hw.hardware_exit_reason = vcpu->arch.trap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1351) /* Clear out the old NMI status from run->flags */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1352) run->flags &= ~KVM_RUN_PPC_NMI_DISP_MASK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1353) /* Now set the NMI status */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1354) if (vcpu->arch.mce_evt.disposition == MCE_DISPOSITION_RECOVERED)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1355) run->flags |= KVM_RUN_PPC_NMI_DISP_FULLY_RECOV;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1356) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1357) run->flags |= KVM_RUN_PPC_NMI_DISP_NOT_RECOV;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1358)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1359) r = RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1360) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1361) case BOOK3S_INTERRUPT_PROGRAM:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1362) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1363) ulong flags;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1364) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1365) * Normally program interrupts are delivered directly
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1366) * to the guest by the hardware, but we can get here
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1367) * as a result of a hypervisor emulation interrupt
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1368) * (e40) getting turned into a 700 by BML RTAS.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1369) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1370) flags = vcpu->arch.shregs.msr & 0x1f0000ull;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1371) kvmppc_core_queue_program(vcpu, flags);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1372) r = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1373) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1374) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1375) case BOOK3S_INTERRUPT_SYSCALL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1376) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1377) /* hcall - punt to userspace */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1378) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1379)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1380) /* hypercall with MSR_PR has already been handled in rmode,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1381) * and never reaches here.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1382) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1383)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1384) run->papr_hcall.nr = kvmppc_get_gpr(vcpu, 3);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1385) for (i = 0; i < 9; ++i)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1386) run->papr_hcall.args[i] = kvmppc_get_gpr(vcpu, 4 + i);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1387) run->exit_reason = KVM_EXIT_PAPR_HCALL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1388) vcpu->arch.hcall_needed = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1389) r = RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1390) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1391) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1392) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1393) * We get these next two if the guest accesses a page which it thinks
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1394) * it has mapped but which is not actually present, either because
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1395) * it is for an emulated I/O device or because the corresonding
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1396) * host page has been paged out. Any other HDSI/HISI interrupts
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1397) * have been handled already.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1398) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1399) case BOOK3S_INTERRUPT_H_DATA_STORAGE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1400) r = RESUME_PAGE_FAULT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1401) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1402) case BOOK3S_INTERRUPT_H_INST_STORAGE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1403) vcpu->arch.fault_dar = kvmppc_get_pc(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1404) vcpu->arch.fault_dsisr = vcpu->arch.shregs.msr &
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1405) DSISR_SRR1_MATCH_64S;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1406) if (vcpu->arch.shregs.msr & HSRR1_HISI_WRITE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1407) vcpu->arch.fault_dsisr |= DSISR_ISSTORE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1408) r = RESUME_PAGE_FAULT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1409) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1410) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1411) * This occurs if the guest executes an illegal instruction.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1412) * If the guest debug is disabled, generate a program interrupt
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1413) * to the guest. If guest debug is enabled, we need to check
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1414) * whether the instruction is a software breakpoint instruction.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1415) * Accordingly return to Guest or Host.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1416) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1417) case BOOK3S_INTERRUPT_H_EMUL_ASSIST:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1418) if (vcpu->arch.emul_inst != KVM_INST_FETCH_FAILED)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1419) vcpu->arch.last_inst = kvmppc_need_byteswap(vcpu) ?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1420) swab32(vcpu->arch.emul_inst) :
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1421) vcpu->arch.emul_inst;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1422) if (vcpu->guest_debug & KVM_GUESTDBG_USE_SW_BP) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1423) r = kvmppc_emulate_debug_inst(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1424) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1425) kvmppc_core_queue_program(vcpu, SRR1_PROGILL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1426) r = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1427) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1428) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1429) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1430) * This occurs if the guest (kernel or userspace), does something that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1431) * is prohibited by HFSCR.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1432) * On POWER9, this could be a doorbell instruction that we need
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1433) * to emulate.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1434) * Otherwise, we just generate a program interrupt to the guest.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1435) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1436) case BOOK3S_INTERRUPT_H_FAC_UNAVAIL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1437) r = EMULATE_FAIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1438) if (((vcpu->arch.hfscr >> 56) == FSCR_MSGP_LG) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1439) cpu_has_feature(CPU_FTR_ARCH_300))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1440) r = kvmppc_emulate_doorbell_instr(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1441) if (r == EMULATE_FAIL) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1442) kvmppc_core_queue_program(vcpu, SRR1_PROGILL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1443) r = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1444) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1445) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1446)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1447) #ifdef CONFIG_PPC_TRANSACTIONAL_MEM
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1448) case BOOK3S_INTERRUPT_HV_SOFTPATCH:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1449) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1450) * This occurs for various TM-related instructions that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1451) * we need to emulate on POWER9 DD2.2. We have already
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1452) * handled the cases where the guest was in real-suspend
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1453) * mode and was transitioning to transactional state.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1454) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1455) r = kvmhv_p9_tm_emulation(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1456) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1457) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1458)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1459) case BOOK3S_INTERRUPT_HV_RM_HARD:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1460) r = RESUME_PASSTHROUGH;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1461) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1462) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1463) kvmppc_dump_regs(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1464) printk(KERN_EMERG "trap=0x%x | pc=0x%lx | msr=0x%llx\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1465) vcpu->arch.trap, kvmppc_get_pc(vcpu),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1466) vcpu->arch.shregs.msr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1467) run->hw.hardware_exit_reason = vcpu->arch.trap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1468) r = RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1469) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1470) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1471)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1472) return r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1473) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1474)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1475) static int kvmppc_handle_nested_exit(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1476) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1477) int r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1478) int srcu_idx;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1479)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1480) vcpu->stat.sum_exits++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1481)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1482) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1483) * This can happen if an interrupt occurs in the last stages
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1484) * of guest entry or the first stages of guest exit (i.e. after
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1485) * setting paca->kvm_hstate.in_guest to KVM_GUEST_MODE_GUEST_HV
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1486) * and before setting it to KVM_GUEST_MODE_HOST_HV).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1487) * That can happen due to a bug, or due to a machine check
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1488) * occurring at just the wrong time.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1489) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1490) if (vcpu->arch.shregs.msr & MSR_HV) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1491) pr_emerg("KVM trap in HV mode while nested!\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1492) pr_emerg("trap=0x%x | pc=0x%lx | msr=0x%llx\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1493) vcpu->arch.trap, kvmppc_get_pc(vcpu),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1494) vcpu->arch.shregs.msr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1495) kvmppc_dump_regs(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1496) return RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1497) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1498) switch (vcpu->arch.trap) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1499) /* We're good on these - the host merely wanted to get our attention */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1500) case BOOK3S_INTERRUPT_HV_DECREMENTER:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1501) vcpu->stat.dec_exits++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1502) r = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1503) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1504) case BOOK3S_INTERRUPT_EXTERNAL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1505) vcpu->stat.ext_intr_exits++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1506) r = RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1507) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1508) case BOOK3S_INTERRUPT_H_DOORBELL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1509) case BOOK3S_INTERRUPT_H_VIRT:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1510) vcpu->stat.ext_intr_exits++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1511) r = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1512) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1513) /* SR/HMI/PMI are HV interrupts that host has handled. Resume guest.*/
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1514) case BOOK3S_INTERRUPT_HMI:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1515) case BOOK3S_INTERRUPT_PERFMON:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1516) case BOOK3S_INTERRUPT_SYSTEM_RESET:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1517) r = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1518) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1519) case BOOK3S_INTERRUPT_MACHINE_CHECK:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1520) /* Pass the machine check to the L1 guest */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1521) r = RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1522) /* Print the MCE event to host console. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1523) machine_check_print_event_info(&vcpu->arch.mce_evt, false, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1524) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1525) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1526) * We get these next two if the guest accesses a page which it thinks
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1527) * it has mapped but which is not actually present, either because
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1528) * it is for an emulated I/O device or because the corresonding
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1529) * host page has been paged out.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1530) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1531) case BOOK3S_INTERRUPT_H_DATA_STORAGE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1532) srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1533) r = kvmhv_nested_page_fault(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1534) srcu_read_unlock(&vcpu->kvm->srcu, srcu_idx);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1535) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1536) case BOOK3S_INTERRUPT_H_INST_STORAGE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1537) vcpu->arch.fault_dar = kvmppc_get_pc(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1538) vcpu->arch.fault_dsisr = kvmppc_get_msr(vcpu) &
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1539) DSISR_SRR1_MATCH_64S;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1540) if (vcpu->arch.shregs.msr & HSRR1_HISI_WRITE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1541) vcpu->arch.fault_dsisr |= DSISR_ISSTORE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1542) srcu_idx = srcu_read_lock(&vcpu->kvm->srcu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1543) r = kvmhv_nested_page_fault(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1544) srcu_read_unlock(&vcpu->kvm->srcu, srcu_idx);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1545) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1546)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1547) #ifdef CONFIG_PPC_TRANSACTIONAL_MEM
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1548) case BOOK3S_INTERRUPT_HV_SOFTPATCH:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1549) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1550) * This occurs for various TM-related instructions that
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1551) * we need to emulate on POWER9 DD2.2. We have already
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1552) * handled the cases where the guest was in real-suspend
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1553) * mode and was transitioning to transactional state.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1554) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1555) r = kvmhv_p9_tm_emulation(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1556) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1557) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1558)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1559) case BOOK3S_INTERRUPT_HV_RM_HARD:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1560) vcpu->arch.trap = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1561) r = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1562) if (!xics_on_xive())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1563) kvmppc_xics_rm_complete(vcpu, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1564) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1565) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1566) r = RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1567) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1568) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1569)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1570) return r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1571) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1572)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1573) static int kvm_arch_vcpu_ioctl_get_sregs_hv(struct kvm_vcpu *vcpu,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1574) struct kvm_sregs *sregs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1575) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1576) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1577)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1578) memset(sregs, 0, sizeof(struct kvm_sregs));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1579) sregs->pvr = vcpu->arch.pvr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1580) for (i = 0; i < vcpu->arch.slb_max; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1581) sregs->u.s.ppc64.slb[i].slbe = vcpu->arch.slb[i].orige;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1582) sregs->u.s.ppc64.slb[i].slbv = vcpu->arch.slb[i].origv;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1583) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1584)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1585) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1586) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1587)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1588) static int kvm_arch_vcpu_ioctl_set_sregs_hv(struct kvm_vcpu *vcpu,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1589) struct kvm_sregs *sregs)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1590) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1591) int i, j;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1592)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1593) /* Only accept the same PVR as the host's, since we can't spoof it */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1594) if (sregs->pvr != vcpu->arch.pvr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1595) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1596)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1597) j = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1598) for (i = 0; i < vcpu->arch.slb_nr; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1599) if (sregs->u.s.ppc64.slb[i].slbe & SLB_ESID_V) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1600) vcpu->arch.slb[j].orige = sregs->u.s.ppc64.slb[i].slbe;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1601) vcpu->arch.slb[j].origv = sregs->u.s.ppc64.slb[i].slbv;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1602) ++j;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1603) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1604) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1605) vcpu->arch.slb_max = j;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1606)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1607) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1608) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1609)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1610) static void kvmppc_set_lpcr(struct kvm_vcpu *vcpu, u64 new_lpcr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1611) bool preserve_top32)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1612) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1613) struct kvm *kvm = vcpu->kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1614) struct kvmppc_vcore *vc = vcpu->arch.vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1615) u64 mask;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1616)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1617) spin_lock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1618) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1619) * If ILE (interrupt little-endian) has changed, update the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1620) * MSR_LE bit in the intr_msr for each vcpu in this vcore.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1621) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1622) if ((new_lpcr & LPCR_ILE) != (vc->lpcr & LPCR_ILE)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1623) struct kvm_vcpu *vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1624) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1625)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1626) kvm_for_each_vcpu(i, vcpu, kvm) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1627) if (vcpu->arch.vcore != vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1628) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1629) if (new_lpcr & LPCR_ILE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1630) vcpu->arch.intr_msr |= MSR_LE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1631) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1632) vcpu->arch.intr_msr &= ~MSR_LE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1633) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1634) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1635)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1636) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1637) * Userspace can only modify DPFD (default prefetch depth),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1638) * ILE (interrupt little-endian) and TC (translation control).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1639) * On POWER8 and POWER9 userspace can also modify AIL (alt. interrupt loc.).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1640) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1641) mask = LPCR_DPFD | LPCR_ILE | LPCR_TC;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1642) if (cpu_has_feature(CPU_FTR_ARCH_207S))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1643) mask |= LPCR_AIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1644) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1645) * On POWER9, allow userspace to enable large decrementer for the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1646) * guest, whether or not the host has it enabled.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1647) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1648) if (cpu_has_feature(CPU_FTR_ARCH_300))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1649) mask |= LPCR_LD;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1650)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1651) /* Broken 32-bit version of LPCR must not clear top bits */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1652) if (preserve_top32)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1653) mask &= 0xFFFFFFFF;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1654) vc->lpcr = (vc->lpcr & ~mask) | (new_lpcr & mask);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1655) spin_unlock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1656) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1657)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1658) static int kvmppc_get_one_reg_hv(struct kvm_vcpu *vcpu, u64 id,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1659) union kvmppc_one_reg *val)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1660) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1661) int r = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1662) long int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1663)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1664) switch (id) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1665) case KVM_REG_PPC_DEBUG_INST:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1666) *val = get_reg_val(id, KVMPPC_INST_SW_BREAKPOINT);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1667) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1668) case KVM_REG_PPC_HIOR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1669) *val = get_reg_val(id, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1670) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1671) case KVM_REG_PPC_DABR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1672) *val = get_reg_val(id, vcpu->arch.dabr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1673) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1674) case KVM_REG_PPC_DABRX:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1675) *val = get_reg_val(id, vcpu->arch.dabrx);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1676) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1677) case KVM_REG_PPC_DSCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1678) *val = get_reg_val(id, vcpu->arch.dscr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1679) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1680) case KVM_REG_PPC_PURR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1681) *val = get_reg_val(id, vcpu->arch.purr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1682) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1683) case KVM_REG_PPC_SPURR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1684) *val = get_reg_val(id, vcpu->arch.spurr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1685) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1686) case KVM_REG_PPC_AMR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1687) *val = get_reg_val(id, vcpu->arch.amr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1688) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1689) case KVM_REG_PPC_UAMOR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1690) *val = get_reg_val(id, vcpu->arch.uamor);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1691) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1692) case KVM_REG_PPC_MMCR0 ... KVM_REG_PPC_MMCR1:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1693) i = id - KVM_REG_PPC_MMCR0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1694) *val = get_reg_val(id, vcpu->arch.mmcr[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1695) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1696) case KVM_REG_PPC_MMCR2:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1697) *val = get_reg_val(id, vcpu->arch.mmcr[2]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1698) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1699) case KVM_REG_PPC_MMCRA:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1700) *val = get_reg_val(id, vcpu->arch.mmcra);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1701) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1702) case KVM_REG_PPC_MMCRS:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1703) *val = get_reg_val(id, vcpu->arch.mmcrs);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1704) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1705) case KVM_REG_PPC_MMCR3:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1706) *val = get_reg_val(id, vcpu->arch.mmcr[3]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1707) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1708) case KVM_REG_PPC_PMC1 ... KVM_REG_PPC_PMC8:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1709) i = id - KVM_REG_PPC_PMC1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1710) *val = get_reg_val(id, vcpu->arch.pmc[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1711) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1712) case KVM_REG_PPC_SPMC1 ... KVM_REG_PPC_SPMC2:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1713) i = id - KVM_REG_PPC_SPMC1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1714) *val = get_reg_val(id, vcpu->arch.spmc[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1715) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1716) case KVM_REG_PPC_SIAR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1717) *val = get_reg_val(id, vcpu->arch.siar);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1718) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1719) case KVM_REG_PPC_SDAR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1720) *val = get_reg_val(id, vcpu->arch.sdar);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1721) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1722) case KVM_REG_PPC_SIER:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1723) *val = get_reg_val(id, vcpu->arch.sier[0]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1724) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1725) case KVM_REG_PPC_SIER2:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1726) *val = get_reg_val(id, vcpu->arch.sier[1]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1727) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1728) case KVM_REG_PPC_SIER3:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1729) *val = get_reg_val(id, vcpu->arch.sier[2]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1730) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1731) case KVM_REG_PPC_IAMR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1732) *val = get_reg_val(id, vcpu->arch.iamr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1733) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1734) case KVM_REG_PPC_PSPB:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1735) *val = get_reg_val(id, vcpu->arch.pspb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1736) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1737) case KVM_REG_PPC_DPDES:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1738) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1739) * On POWER9, where we are emulating msgsndp etc.,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1740) * we return 1 bit for each vcpu, which can come from
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1741) * either vcore->dpdes or doorbell_request.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1742) * On POWER8, doorbell_request is 0.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1743) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1744) *val = get_reg_val(id, vcpu->arch.vcore->dpdes |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1745) vcpu->arch.doorbell_request);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1746) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1747) case KVM_REG_PPC_VTB:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1748) *val = get_reg_val(id, vcpu->arch.vcore->vtb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1749) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1750) case KVM_REG_PPC_DAWR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1751) *val = get_reg_val(id, vcpu->arch.dawr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1752) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1753) case KVM_REG_PPC_DAWRX:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1754) *val = get_reg_val(id, vcpu->arch.dawrx);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1755) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1756) case KVM_REG_PPC_CIABR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1757) *val = get_reg_val(id, vcpu->arch.ciabr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1758) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1759) case KVM_REG_PPC_CSIGR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1760) *val = get_reg_val(id, vcpu->arch.csigr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1761) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1762) case KVM_REG_PPC_TACR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1763) *val = get_reg_val(id, vcpu->arch.tacr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1764) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1765) case KVM_REG_PPC_TCSCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1766) *val = get_reg_val(id, vcpu->arch.tcscr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1767) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1768) case KVM_REG_PPC_PID:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1769) *val = get_reg_val(id, vcpu->arch.pid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1770) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1771) case KVM_REG_PPC_ACOP:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1772) *val = get_reg_val(id, vcpu->arch.acop);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1773) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1774) case KVM_REG_PPC_WORT:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1775) *val = get_reg_val(id, vcpu->arch.wort);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1776) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1777) case KVM_REG_PPC_TIDR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1778) *val = get_reg_val(id, vcpu->arch.tid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1779) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1780) case KVM_REG_PPC_PSSCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1781) *val = get_reg_val(id, vcpu->arch.psscr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1782) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1783) case KVM_REG_PPC_VPA_ADDR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1784) spin_lock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1785) *val = get_reg_val(id, vcpu->arch.vpa.next_gpa);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1786) spin_unlock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1787) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1788) case KVM_REG_PPC_VPA_SLB:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1789) spin_lock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1790) val->vpaval.addr = vcpu->arch.slb_shadow.next_gpa;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1791) val->vpaval.length = vcpu->arch.slb_shadow.len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1792) spin_unlock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1793) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1794) case KVM_REG_PPC_VPA_DTL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1795) spin_lock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1796) val->vpaval.addr = vcpu->arch.dtl.next_gpa;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1797) val->vpaval.length = vcpu->arch.dtl.len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1798) spin_unlock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1799) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1800) case KVM_REG_PPC_TB_OFFSET:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1801) *val = get_reg_val(id, vcpu->arch.vcore->tb_offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1802) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1803) case KVM_REG_PPC_LPCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1804) case KVM_REG_PPC_LPCR_64:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1805) *val = get_reg_val(id, vcpu->arch.vcore->lpcr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1806) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1807) case KVM_REG_PPC_PPR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1808) *val = get_reg_val(id, vcpu->arch.ppr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1809) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1810) #ifdef CONFIG_PPC_TRANSACTIONAL_MEM
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1811) case KVM_REG_PPC_TFHAR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1812) *val = get_reg_val(id, vcpu->arch.tfhar);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1813) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1814) case KVM_REG_PPC_TFIAR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1815) *val = get_reg_val(id, vcpu->arch.tfiar);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1816) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1817) case KVM_REG_PPC_TEXASR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1818) *val = get_reg_val(id, vcpu->arch.texasr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1819) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1820) case KVM_REG_PPC_TM_GPR0 ... KVM_REG_PPC_TM_GPR31:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1821) i = id - KVM_REG_PPC_TM_GPR0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1822) *val = get_reg_val(id, vcpu->arch.gpr_tm[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1823) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1824) case KVM_REG_PPC_TM_VSR0 ... KVM_REG_PPC_TM_VSR63:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1825) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1826) int j;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1827) i = id - KVM_REG_PPC_TM_VSR0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1828) if (i < 32)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1829) for (j = 0; j < TS_FPRWIDTH; j++)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1830) val->vsxval[j] = vcpu->arch.fp_tm.fpr[i][j];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1831) else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1832) if (cpu_has_feature(CPU_FTR_ALTIVEC))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1833) val->vval = vcpu->arch.vr_tm.vr[i-32];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1834) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1835) r = -ENXIO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1836) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1837) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1838) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1839) case KVM_REG_PPC_TM_CR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1840) *val = get_reg_val(id, vcpu->arch.cr_tm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1841) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1842) case KVM_REG_PPC_TM_XER:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1843) *val = get_reg_val(id, vcpu->arch.xer_tm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1844) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1845) case KVM_REG_PPC_TM_LR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1846) *val = get_reg_val(id, vcpu->arch.lr_tm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1847) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1848) case KVM_REG_PPC_TM_CTR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1849) *val = get_reg_val(id, vcpu->arch.ctr_tm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1850) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1851) case KVM_REG_PPC_TM_FPSCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1852) *val = get_reg_val(id, vcpu->arch.fp_tm.fpscr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1853) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1854) case KVM_REG_PPC_TM_AMR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1855) *val = get_reg_val(id, vcpu->arch.amr_tm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1856) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1857) case KVM_REG_PPC_TM_PPR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1858) *val = get_reg_val(id, vcpu->arch.ppr_tm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1859) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1860) case KVM_REG_PPC_TM_VRSAVE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1861) *val = get_reg_val(id, vcpu->arch.vrsave_tm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1862) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1863) case KVM_REG_PPC_TM_VSCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1864) if (cpu_has_feature(CPU_FTR_ALTIVEC))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1865) *val = get_reg_val(id, vcpu->arch.vr_tm.vscr.u[3]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1866) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1867) r = -ENXIO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1868) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1869) case KVM_REG_PPC_TM_DSCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1870) *val = get_reg_val(id, vcpu->arch.dscr_tm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1871) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1872) case KVM_REG_PPC_TM_TAR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1873) *val = get_reg_val(id, vcpu->arch.tar_tm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1874) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1875) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1876) case KVM_REG_PPC_ARCH_COMPAT:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1877) *val = get_reg_val(id, vcpu->arch.vcore->arch_compat);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1878) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1879) case KVM_REG_PPC_DEC_EXPIRY:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1880) *val = get_reg_val(id, vcpu->arch.dec_expires +
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1881) vcpu->arch.vcore->tb_offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1882) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1883) case KVM_REG_PPC_ONLINE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1884) *val = get_reg_val(id, vcpu->arch.online);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1885) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1886) case KVM_REG_PPC_PTCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1887) *val = get_reg_val(id, vcpu->kvm->arch.l1_ptcr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1888) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1889) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1890) r = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1891) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1892) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1893)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1894) return r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1895) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1896)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1897) static int kvmppc_set_one_reg_hv(struct kvm_vcpu *vcpu, u64 id,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1898) union kvmppc_one_reg *val)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1899) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1900) int r = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1901) long int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1902) unsigned long addr, len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1903)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1904) switch (id) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1905) case KVM_REG_PPC_HIOR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1906) /* Only allow this to be set to zero */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1907) if (set_reg_val(id, *val))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1908) r = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1909) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1910) case KVM_REG_PPC_DABR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1911) vcpu->arch.dabr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1912) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1913) case KVM_REG_PPC_DABRX:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1914) vcpu->arch.dabrx = set_reg_val(id, *val) & ~DABRX_HYP;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1915) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1916) case KVM_REG_PPC_DSCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1917) vcpu->arch.dscr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1918) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1919) case KVM_REG_PPC_PURR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1920) vcpu->arch.purr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1921) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1922) case KVM_REG_PPC_SPURR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1923) vcpu->arch.spurr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1924) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1925) case KVM_REG_PPC_AMR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1926) vcpu->arch.amr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1927) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1928) case KVM_REG_PPC_UAMOR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1929) vcpu->arch.uamor = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1930) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1931) case KVM_REG_PPC_MMCR0 ... KVM_REG_PPC_MMCR1:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1932) i = id - KVM_REG_PPC_MMCR0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1933) vcpu->arch.mmcr[i] = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1934) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1935) case KVM_REG_PPC_MMCR2:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1936) vcpu->arch.mmcr[2] = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1937) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1938) case KVM_REG_PPC_MMCRA:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1939) vcpu->arch.mmcra = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1940) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1941) case KVM_REG_PPC_MMCRS:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1942) vcpu->arch.mmcrs = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1943) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1944) case KVM_REG_PPC_MMCR3:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1945) *val = get_reg_val(id, vcpu->arch.mmcr[3]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1946) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1947) case KVM_REG_PPC_PMC1 ... KVM_REG_PPC_PMC8:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1948) i = id - KVM_REG_PPC_PMC1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1949) vcpu->arch.pmc[i] = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1950) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1951) case KVM_REG_PPC_SPMC1 ... KVM_REG_PPC_SPMC2:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1952) i = id - KVM_REG_PPC_SPMC1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1953) vcpu->arch.spmc[i] = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1954) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1955) case KVM_REG_PPC_SIAR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1956) vcpu->arch.siar = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1957) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1958) case KVM_REG_PPC_SDAR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1959) vcpu->arch.sdar = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1960) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1961) case KVM_REG_PPC_SIER:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1962) vcpu->arch.sier[0] = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1963) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1964) case KVM_REG_PPC_SIER2:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1965) vcpu->arch.sier[1] = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1966) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1967) case KVM_REG_PPC_SIER3:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1968) vcpu->arch.sier[2] = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1969) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1970) case KVM_REG_PPC_IAMR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1971) vcpu->arch.iamr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1972) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1973) case KVM_REG_PPC_PSPB:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1974) vcpu->arch.pspb = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1975) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1976) case KVM_REG_PPC_DPDES:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1977) vcpu->arch.vcore->dpdes = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1978) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1979) case KVM_REG_PPC_VTB:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1980) vcpu->arch.vcore->vtb = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1981) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1982) case KVM_REG_PPC_DAWR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1983) vcpu->arch.dawr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1984) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1985) case KVM_REG_PPC_DAWRX:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1986) vcpu->arch.dawrx = set_reg_val(id, *val) & ~DAWRX_HYP;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1987) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1988) case KVM_REG_PPC_CIABR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1989) vcpu->arch.ciabr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1990) /* Don't allow setting breakpoints in hypervisor code */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1991) if ((vcpu->arch.ciabr & CIABR_PRIV) == CIABR_PRIV_HYPER)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1992) vcpu->arch.ciabr &= ~CIABR_PRIV; /* disable */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1993) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1994) case KVM_REG_PPC_CSIGR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1995) vcpu->arch.csigr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1996) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1997) case KVM_REG_PPC_TACR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1998) vcpu->arch.tacr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 1999) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2000) case KVM_REG_PPC_TCSCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2001) vcpu->arch.tcscr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2002) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2003) case KVM_REG_PPC_PID:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2004) vcpu->arch.pid = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2005) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2006) case KVM_REG_PPC_ACOP:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2007) vcpu->arch.acop = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2008) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2009) case KVM_REG_PPC_WORT:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2010) vcpu->arch.wort = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2011) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2012) case KVM_REG_PPC_TIDR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2013) vcpu->arch.tid = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2014) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2015) case KVM_REG_PPC_PSSCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2016) vcpu->arch.psscr = set_reg_val(id, *val) & PSSCR_GUEST_VIS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2017) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2018) case KVM_REG_PPC_VPA_ADDR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2019) addr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2020) r = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2021) if (!addr && (vcpu->arch.slb_shadow.next_gpa ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2022) vcpu->arch.dtl.next_gpa))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2023) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2024) r = set_vpa(vcpu, &vcpu->arch.vpa, addr, sizeof(struct lppaca));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2025) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2026) case KVM_REG_PPC_VPA_SLB:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2027) addr = val->vpaval.addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2028) len = val->vpaval.length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2029) r = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2030) if (addr && !vcpu->arch.vpa.next_gpa)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2031) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2032) r = set_vpa(vcpu, &vcpu->arch.slb_shadow, addr, len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2033) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2034) case KVM_REG_PPC_VPA_DTL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2035) addr = val->vpaval.addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2036) len = val->vpaval.length;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2037) r = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2038) if (addr && (len < sizeof(struct dtl_entry) ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2039) !vcpu->arch.vpa.next_gpa))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2040) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2041) len -= len % sizeof(struct dtl_entry);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2042) r = set_vpa(vcpu, &vcpu->arch.dtl, addr, len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2043) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2044) case KVM_REG_PPC_TB_OFFSET:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2045) /* round up to multiple of 2^24 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2046) vcpu->arch.vcore->tb_offset =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2047) ALIGN(set_reg_val(id, *val), 1UL << 24);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2048) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2049) case KVM_REG_PPC_LPCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2050) kvmppc_set_lpcr(vcpu, set_reg_val(id, *val), true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2051) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2052) case KVM_REG_PPC_LPCR_64:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2053) kvmppc_set_lpcr(vcpu, set_reg_val(id, *val), false);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2054) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2055) case KVM_REG_PPC_PPR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2056) vcpu->arch.ppr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2057) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2058) #ifdef CONFIG_PPC_TRANSACTIONAL_MEM
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2059) case KVM_REG_PPC_TFHAR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2060) vcpu->arch.tfhar = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2061) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2062) case KVM_REG_PPC_TFIAR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2063) vcpu->arch.tfiar = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2064) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2065) case KVM_REG_PPC_TEXASR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2066) vcpu->arch.texasr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2067) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2068) case KVM_REG_PPC_TM_GPR0 ... KVM_REG_PPC_TM_GPR31:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2069) i = id - KVM_REG_PPC_TM_GPR0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2070) vcpu->arch.gpr_tm[i] = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2071) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2072) case KVM_REG_PPC_TM_VSR0 ... KVM_REG_PPC_TM_VSR63:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2073) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2074) int j;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2075) i = id - KVM_REG_PPC_TM_VSR0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2076) if (i < 32)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2077) for (j = 0; j < TS_FPRWIDTH; j++)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2078) vcpu->arch.fp_tm.fpr[i][j] = val->vsxval[j];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2079) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2080) if (cpu_has_feature(CPU_FTR_ALTIVEC))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2081) vcpu->arch.vr_tm.vr[i-32] = val->vval;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2082) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2083) r = -ENXIO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2084) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2085) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2086) case KVM_REG_PPC_TM_CR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2087) vcpu->arch.cr_tm = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2088) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2089) case KVM_REG_PPC_TM_XER:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2090) vcpu->arch.xer_tm = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2091) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2092) case KVM_REG_PPC_TM_LR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2093) vcpu->arch.lr_tm = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2094) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2095) case KVM_REG_PPC_TM_CTR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2096) vcpu->arch.ctr_tm = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2097) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2098) case KVM_REG_PPC_TM_FPSCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2099) vcpu->arch.fp_tm.fpscr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2100) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2101) case KVM_REG_PPC_TM_AMR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2102) vcpu->arch.amr_tm = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2103) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2104) case KVM_REG_PPC_TM_PPR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2105) vcpu->arch.ppr_tm = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2106) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2107) case KVM_REG_PPC_TM_VRSAVE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2108) vcpu->arch.vrsave_tm = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2109) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2110) case KVM_REG_PPC_TM_VSCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2111) if (cpu_has_feature(CPU_FTR_ALTIVEC))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2112) vcpu->arch.vr.vscr.u[3] = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2113) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2114) r = - ENXIO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2115) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2116) case KVM_REG_PPC_TM_DSCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2117) vcpu->arch.dscr_tm = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2118) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2119) case KVM_REG_PPC_TM_TAR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2120) vcpu->arch.tar_tm = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2121) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2122) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2123) case KVM_REG_PPC_ARCH_COMPAT:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2124) r = kvmppc_set_arch_compat(vcpu, set_reg_val(id, *val));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2125) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2126) case KVM_REG_PPC_DEC_EXPIRY:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2127) vcpu->arch.dec_expires = set_reg_val(id, *val) -
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2128) vcpu->arch.vcore->tb_offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2129) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2130) case KVM_REG_PPC_ONLINE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2131) i = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2132) if (i && !vcpu->arch.online)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2133) atomic_inc(&vcpu->arch.vcore->online_count);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2134) else if (!i && vcpu->arch.online)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2135) atomic_dec(&vcpu->arch.vcore->online_count);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2136) vcpu->arch.online = i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2137) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2138) case KVM_REG_PPC_PTCR:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2139) vcpu->kvm->arch.l1_ptcr = set_reg_val(id, *val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2140) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2141) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2142) r = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2143) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2144) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2145)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2146) return r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2147) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2148)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2149) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2150) * On POWER9, threads are independent and can be in different partitions.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2151) * Therefore we consider each thread to be a subcore.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2152) * There is a restriction that all threads have to be in the same
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2153) * MMU mode (radix or HPT), unfortunately, but since we only support
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2154) * HPT guests on a HPT host so far, that isn't an impediment yet.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2155) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2156) static int threads_per_vcore(struct kvm *kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2157) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2158) if (kvm->arch.threads_indep)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2159) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2160) return threads_per_subcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2161) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2162)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2163) static struct kvmppc_vcore *kvmppc_vcore_create(struct kvm *kvm, int id)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2164) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2165) struct kvmppc_vcore *vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2166)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2167) vcore = kzalloc(sizeof(struct kvmppc_vcore), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2168)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2169) if (vcore == NULL)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2170) return NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2171)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2172) spin_lock_init(&vcore->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2173) spin_lock_init(&vcore->stoltb_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2174) rcuwait_init(&vcore->wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2175) vcore->preempt_tb = TB_NIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2176) vcore->lpcr = kvm->arch.lpcr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2177) vcore->first_vcpuid = id;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2178) vcore->kvm = kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2179) INIT_LIST_HEAD(&vcore->preempt_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2180)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2181) return vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2182) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2183)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2184) #ifdef CONFIG_KVM_BOOK3S_HV_EXIT_TIMING
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2185) static struct debugfs_timings_element {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2186) const char *name;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2187) size_t offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2188) } timings[] = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2189) {"rm_entry", offsetof(struct kvm_vcpu, arch.rm_entry)},
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2190) {"rm_intr", offsetof(struct kvm_vcpu, arch.rm_intr)},
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2191) {"rm_exit", offsetof(struct kvm_vcpu, arch.rm_exit)},
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2192) {"guest", offsetof(struct kvm_vcpu, arch.guest_time)},
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2193) {"cede", offsetof(struct kvm_vcpu, arch.cede_time)},
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2194) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2195)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2196) #define N_TIMINGS (ARRAY_SIZE(timings))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2197)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2198) struct debugfs_timings_state {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2199) struct kvm_vcpu *vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2200) unsigned int buflen;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2201) char buf[N_TIMINGS * 100];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2202) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2203)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2204) static int debugfs_timings_open(struct inode *inode, struct file *file)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2205) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2206) struct kvm_vcpu *vcpu = inode->i_private;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2207) struct debugfs_timings_state *p;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2208)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2209) p = kzalloc(sizeof(*p), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2210) if (!p)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2211) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2212)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2213) kvm_get_kvm(vcpu->kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2214) p->vcpu = vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2215) file->private_data = p;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2216)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2217) return nonseekable_open(inode, file);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2218) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2219)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2220) static int debugfs_timings_release(struct inode *inode, struct file *file)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2221) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2222) struct debugfs_timings_state *p = file->private_data;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2223)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2224) kvm_put_kvm(p->vcpu->kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2225) kfree(p);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2226) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2227) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2228)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2229) static ssize_t debugfs_timings_read(struct file *file, char __user *buf,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2230) size_t len, loff_t *ppos)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2231) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2232) struct debugfs_timings_state *p = file->private_data;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2233) struct kvm_vcpu *vcpu = p->vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2234) char *s, *buf_end;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2235) struct kvmhv_tb_accumulator tb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2236) u64 count;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2237) loff_t pos;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2238) ssize_t n;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2239) int i, loops;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2240) bool ok;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2241)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2242) if (!p->buflen) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2243) s = p->buf;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2244) buf_end = s + sizeof(p->buf);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2245) for (i = 0; i < N_TIMINGS; ++i) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2246) struct kvmhv_tb_accumulator *acc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2247)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2248) acc = (struct kvmhv_tb_accumulator *)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2249) ((unsigned long)vcpu + timings[i].offset);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2250) ok = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2251) for (loops = 0; loops < 1000; ++loops) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2252) count = acc->seqcount;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2253) if (!(count & 1)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2254) smp_rmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2255) tb = *acc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2256) smp_rmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2257) if (count == acc->seqcount) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2258) ok = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2259) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2260) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2261) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2262) udelay(1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2263) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2264) if (!ok)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2265) snprintf(s, buf_end - s, "%s: stuck\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2266) timings[i].name);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2267) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2268) snprintf(s, buf_end - s,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2269) "%s: %llu %llu %llu %llu\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2270) timings[i].name, count / 2,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2271) tb_to_ns(tb.tb_total),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2272) tb_to_ns(tb.tb_min),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2273) tb_to_ns(tb.tb_max));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2274) s += strlen(s);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2275) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2276) p->buflen = s - p->buf;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2277) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2278)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2279) pos = *ppos;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2280) if (pos >= p->buflen)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2281) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2282) if (len > p->buflen - pos)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2283) len = p->buflen - pos;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2284) n = copy_to_user(buf, p->buf + pos, len);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2285) if (n) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2286) if (n == len)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2287) return -EFAULT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2288) len -= n;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2289) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2290) *ppos = pos + len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2291) return len;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2292) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2293)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2294) static ssize_t debugfs_timings_write(struct file *file, const char __user *buf,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2295) size_t len, loff_t *ppos)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2296) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2297) return -EACCES;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2298) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2299)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2300) static const struct file_operations debugfs_timings_ops = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2301) .owner = THIS_MODULE,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2302) .open = debugfs_timings_open,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2303) .release = debugfs_timings_release,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2304) .read = debugfs_timings_read,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2305) .write = debugfs_timings_write,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2306) .llseek = generic_file_llseek,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2307) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2308)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2309) /* Create a debugfs directory for the vcpu */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2310) static void debugfs_vcpu_init(struct kvm_vcpu *vcpu, unsigned int id)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2311) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2312) char buf[16];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2313) struct kvm *kvm = vcpu->kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2314)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2315) snprintf(buf, sizeof(buf), "vcpu%u", id);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2316) vcpu->arch.debugfs_dir = debugfs_create_dir(buf, kvm->arch.debugfs_dir);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2317) debugfs_create_file("timings", 0444, vcpu->arch.debugfs_dir, vcpu,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2318) &debugfs_timings_ops);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2319) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2320)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2321) #else /* CONFIG_KVM_BOOK3S_HV_EXIT_TIMING */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2322) static void debugfs_vcpu_init(struct kvm_vcpu *vcpu, unsigned int id)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2323) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2324) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2325) #endif /* CONFIG_KVM_BOOK3S_HV_EXIT_TIMING */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2326)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2327) static int kvmppc_core_vcpu_create_hv(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2328) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2329) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2330) int core;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2331) struct kvmppc_vcore *vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2332) struct kvm *kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2333) unsigned int id;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2334)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2335) kvm = vcpu->kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2336) id = vcpu->vcpu_id;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2337)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2338) vcpu->arch.shared = &vcpu->arch.shregs;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2339) #ifdef CONFIG_KVM_BOOK3S_PR_POSSIBLE
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2340) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2341) * The shared struct is never shared on HV,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2342) * so we can always use host endianness
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2343) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2344) #ifdef __BIG_ENDIAN__
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2345) vcpu->arch.shared_big_endian = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2346) #else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2347) vcpu->arch.shared_big_endian = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2348) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2349) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2350) vcpu->arch.mmcr[0] = MMCR0_FC;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2351) vcpu->arch.ctrl = CTRL_RUNLATCH;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2352) /* default to host PVR, since we can't spoof it */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2353) kvmppc_set_pvr_hv(vcpu, mfspr(SPRN_PVR));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2354) spin_lock_init(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2355) spin_lock_init(&vcpu->arch.tbacct_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2356) vcpu->arch.busy_preempt = TB_NIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2357) vcpu->arch.intr_msr = MSR_SF | MSR_ME;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2358)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2359) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2360) * Set the default HFSCR for the guest from the host value.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2361) * This value is only used on POWER9.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2362) * On POWER9, we want to virtualize the doorbell facility, so we
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2363) * don't set the HFSCR_MSGP bit, and that causes those instructions
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2364) * to trap and then we emulate them.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2365) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2366) vcpu->arch.hfscr = HFSCR_TAR | HFSCR_EBB | HFSCR_PM | HFSCR_BHRB |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2367) HFSCR_DSCR | HFSCR_VECVSX | HFSCR_FP | HFSCR_PREFIX;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2368) if (cpu_has_feature(CPU_FTR_HVMODE)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2369) vcpu->arch.hfscr &= mfspr(SPRN_HFSCR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2370) #ifdef CONFIG_PPC_TRANSACTIONAL_MEM
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2371) if (cpu_has_feature(CPU_FTR_P9_TM_HV_ASSIST))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2372) vcpu->arch.hfscr |= HFSCR_TM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2373) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2374) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2375) if (cpu_has_feature(CPU_FTR_TM_COMP))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2376) vcpu->arch.hfscr |= HFSCR_TM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2377)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2378) kvmppc_mmu_book3s_hv_init(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2379)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2380) vcpu->arch.state = KVMPPC_VCPU_NOTREADY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2381)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2382) init_waitqueue_head(&vcpu->arch.cpu_run);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2383)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2384) mutex_lock(&kvm->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2385) vcore = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2386) err = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2387) if (cpu_has_feature(CPU_FTR_ARCH_300)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2388) if (id >= (KVM_MAX_VCPUS * kvm->arch.emul_smt_mode)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2389) pr_devel("KVM: VCPU ID too high\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2390) core = KVM_MAX_VCORES;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2391) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2392) BUG_ON(kvm->arch.smt_mode != 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2393) core = kvmppc_pack_vcpu_id(kvm, id);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2394) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2395) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2396) core = id / kvm->arch.smt_mode;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2397) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2398) if (core < KVM_MAX_VCORES) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2399) vcore = kvm->arch.vcores[core];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2400) if (vcore && cpu_has_feature(CPU_FTR_ARCH_300)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2401) pr_devel("KVM: collision on id %u", id);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2402) vcore = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2403) } else if (!vcore) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2404) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2405) * Take mmu_setup_lock for mutual exclusion
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2406) * with kvmppc_update_lpcr().
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2407) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2408) err = -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2409) vcore = kvmppc_vcore_create(kvm,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2410) id & ~(kvm->arch.smt_mode - 1));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2411) mutex_lock(&kvm->arch.mmu_setup_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2412) kvm->arch.vcores[core] = vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2413) kvm->arch.online_vcores++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2414) mutex_unlock(&kvm->arch.mmu_setup_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2415) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2416) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2417) mutex_unlock(&kvm->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2418)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2419) if (!vcore)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2420) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2421)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2422) spin_lock(&vcore->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2423) ++vcore->num_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2424) spin_unlock(&vcore->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2425) vcpu->arch.vcore = vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2426) vcpu->arch.ptid = vcpu->vcpu_id - vcore->first_vcpuid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2427) vcpu->arch.thread_cpu = -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2428) vcpu->arch.prev_cpu = -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2429)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2430) vcpu->arch.cpu_type = KVM_CPU_3S_64;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2431) kvmppc_sanity_check(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2432)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2433) debugfs_vcpu_init(vcpu, id);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2434)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2435) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2436) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2437)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2438) static int kvmhv_set_smt_mode(struct kvm *kvm, unsigned long smt_mode,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2439) unsigned long flags)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2440) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2441) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2442) int esmt = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2443)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2444) if (flags)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2445) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2446) if (smt_mode > MAX_SMT_THREADS || !is_power_of_2(smt_mode))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2447) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2448) if (!cpu_has_feature(CPU_FTR_ARCH_300)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2449) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2450) * On POWER8 (or POWER7), the threading mode is "strict",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2451) * so we pack smt_mode vcpus per vcore.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2452) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2453) if (smt_mode > threads_per_subcore)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2454) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2455) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2456) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2457) * On POWER9, the threading mode is "loose",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2458) * so each vcpu gets its own vcore.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2459) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2460) esmt = smt_mode;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2461) smt_mode = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2462) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2463) mutex_lock(&kvm->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2464) err = -EBUSY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2465) if (!kvm->arch.online_vcores) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2466) kvm->arch.smt_mode = smt_mode;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2467) kvm->arch.emul_smt_mode = esmt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2468) err = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2469) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2470) mutex_unlock(&kvm->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2471)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2472) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2473) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2474)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2475) static void unpin_vpa(struct kvm *kvm, struct kvmppc_vpa *vpa)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2476) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2477) if (vpa->pinned_addr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2478) kvmppc_unpin_guest_page(kvm, vpa->pinned_addr, vpa->gpa,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2479) vpa->dirty);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2480) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2481)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2482) static void kvmppc_core_vcpu_free_hv(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2483) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2484) spin_lock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2485) unpin_vpa(vcpu->kvm, &vcpu->arch.dtl);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2486) unpin_vpa(vcpu->kvm, &vcpu->arch.slb_shadow);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2487) unpin_vpa(vcpu->kvm, &vcpu->arch.vpa);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2488) spin_unlock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2489) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2490)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2491) static int kvmppc_core_check_requests_hv(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2492) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2493) /* Indicate we want to get back into the guest */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2494) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2495) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2496)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2497) static void kvmppc_set_timer(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2498) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2499) unsigned long dec_nsec, now;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2500)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2501) now = get_tb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2502) if (now > vcpu->arch.dec_expires) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2503) /* decrementer has already gone negative */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2504) kvmppc_core_queue_dec(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2505) kvmppc_core_prepare_to_enter(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2506) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2507) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2508) dec_nsec = tb_to_ns(vcpu->arch.dec_expires - now);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2509) hrtimer_start(&vcpu->arch.dec_timer, dec_nsec, HRTIMER_MODE_REL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2510) vcpu->arch.timer_running = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2511) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2512)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2513) extern int __kvmppc_vcore_entry(void);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2514)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2515) static void kvmppc_remove_runnable(struct kvmppc_vcore *vc,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2516) struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2517) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2518) u64 now;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2519)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2520) if (vcpu->arch.state != KVMPPC_VCPU_RUNNABLE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2521) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2522) spin_lock_irq(&vcpu->arch.tbacct_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2523) now = mftb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2524) vcpu->arch.busy_stolen += vcore_stolen_time(vc, now) -
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2525) vcpu->arch.stolen_logged;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2526) vcpu->arch.busy_preempt = now;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2527) vcpu->arch.state = KVMPPC_VCPU_BUSY_IN_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2528) spin_unlock_irq(&vcpu->arch.tbacct_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2529) --vc->n_runnable;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2530) WRITE_ONCE(vc->runnable_threads[vcpu->arch.ptid], NULL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2531) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2532)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2533) static int kvmppc_grab_hwthread(int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2534) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2535) struct paca_struct *tpaca;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2536) long timeout = 10000;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2537)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2538) tpaca = paca_ptrs[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2539)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2540) /* Ensure the thread won't go into the kernel if it wakes */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2541) tpaca->kvm_hstate.kvm_vcpu = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2542) tpaca->kvm_hstate.kvm_vcore = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2543) tpaca->kvm_hstate.napping = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2544) smp_wmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2545) tpaca->kvm_hstate.hwthread_req = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2546)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2547) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2548) * If the thread is already executing in the kernel (e.g. handling
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2549) * a stray interrupt), wait for it to get back to nap mode.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2550) * The smp_mb() is to ensure that our setting of hwthread_req
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2551) * is visible before we look at hwthread_state, so if this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2552) * races with the code at system_reset_pSeries and the thread
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2553) * misses our setting of hwthread_req, we are sure to see its
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2554) * setting of hwthread_state, and vice versa.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2555) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2556) smp_mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2557) while (tpaca->kvm_hstate.hwthread_state == KVM_HWTHREAD_IN_KERNEL) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2558) if (--timeout <= 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2559) pr_err("KVM: couldn't grab cpu %d\n", cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2560) return -EBUSY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2561) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2562) udelay(1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2563) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2564) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2565) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2566)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2567) static void kvmppc_release_hwthread(int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2568) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2569) struct paca_struct *tpaca;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2570)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2571) tpaca = paca_ptrs[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2572) tpaca->kvm_hstate.hwthread_req = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2573) tpaca->kvm_hstate.kvm_vcpu = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2574) tpaca->kvm_hstate.kvm_vcore = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2575) tpaca->kvm_hstate.kvm_split_mode = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2576) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2577)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2578) static void radix_flush_cpu(struct kvm *kvm, int cpu, struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2579) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2580) struct kvm_nested_guest *nested = vcpu->arch.nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2581) cpumask_t *cpu_in_guest;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2582) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2583)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2584) cpu = cpu_first_tlb_thread_sibling(cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2585) if (nested) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2586) cpumask_set_cpu(cpu, &nested->need_tlb_flush);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2587) cpu_in_guest = &nested->cpu_in_guest;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2588) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2589) cpumask_set_cpu(cpu, &kvm->arch.need_tlb_flush);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2590) cpu_in_guest = &kvm->arch.cpu_in_guest;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2591) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2592) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2593) * Make sure setting of bit in need_tlb_flush precedes
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2594) * testing of cpu_in_guest bits. The matching barrier on
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2595) * the other side is the first smp_mb() in kvmppc_run_core().
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2596) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2597) smp_mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2598) for (i = cpu; i <= cpu_last_tlb_thread_sibling(cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2599) i += cpu_tlb_thread_sibling_step())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2600) if (cpumask_test_cpu(i, cpu_in_guest))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2601) smp_call_function_single(i, do_nothing, NULL, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2602) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2603)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2604) static void kvmppc_prepare_radix_vcpu(struct kvm_vcpu *vcpu, int pcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2605) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2606) struct kvm_nested_guest *nested = vcpu->arch.nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2607) struct kvm *kvm = vcpu->kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2608) int prev_cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2609)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2610) if (!cpu_has_feature(CPU_FTR_HVMODE))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2611) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2612)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2613) if (nested)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2614) prev_cpu = nested->prev_cpu[vcpu->arch.nested_vcpu_id];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2615) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2616) prev_cpu = vcpu->arch.prev_cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2617)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2618) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2619) * With radix, the guest can do TLB invalidations itself,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2620) * and it could choose to use the local form (tlbiel) if
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2621) * it is invalidating a translation that has only ever been
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2622) * used on one vcpu. However, that doesn't mean it has
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2623) * only ever been used on one physical cpu, since vcpus
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2624) * can move around between pcpus. To cope with this, when
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2625) * a vcpu moves from one pcpu to another, we need to tell
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2626) * any vcpus running on the same core as this vcpu previously
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2627) * ran to flush the TLB. The TLB is shared between threads,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2628) * so we use a single bit in .need_tlb_flush for all 4 threads.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2629) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2630) if (prev_cpu != pcpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2631) if (prev_cpu >= 0 &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2632) cpu_first_tlb_thread_sibling(prev_cpu) !=
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2633) cpu_first_tlb_thread_sibling(pcpu))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2634) radix_flush_cpu(kvm, prev_cpu, vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2635) if (nested)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2636) nested->prev_cpu[vcpu->arch.nested_vcpu_id] = pcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2637) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2638) vcpu->arch.prev_cpu = pcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2639) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2640) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2641)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2642) static void kvmppc_start_thread(struct kvm_vcpu *vcpu, struct kvmppc_vcore *vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2643) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2644) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2645) struct paca_struct *tpaca;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2646) struct kvm *kvm = vc->kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2647)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2648) cpu = vc->pcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2649) if (vcpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2650) if (vcpu->arch.timer_running) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2651) hrtimer_try_to_cancel(&vcpu->arch.dec_timer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2652) vcpu->arch.timer_running = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2653) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2654) cpu += vcpu->arch.ptid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2655) vcpu->cpu = vc->pcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2656) vcpu->arch.thread_cpu = cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2657) cpumask_set_cpu(cpu, &kvm->arch.cpu_in_guest);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2658) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2659) tpaca = paca_ptrs[cpu];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2660) tpaca->kvm_hstate.kvm_vcpu = vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2661) tpaca->kvm_hstate.ptid = cpu - vc->pcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2662) tpaca->kvm_hstate.fake_suspend = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2663) /* Order stores to hstate.kvm_vcpu etc. before store to kvm_vcore */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2664) smp_wmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2665) tpaca->kvm_hstate.kvm_vcore = vc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2666) if (cpu != smp_processor_id())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2667) kvmppc_ipi_thread(cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2668) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2669)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2670) static void kvmppc_wait_for_nap(int n_threads)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2671) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2672) int cpu = smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2673) int i, loops;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2674)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2675) if (n_threads <= 1)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2676) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2677) for (loops = 0; loops < 1000000; ++loops) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2678) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2679) * Check if all threads are finished.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2680) * We set the vcore pointer when starting a thread
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2681) * and the thread clears it when finished, so we look
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2682) * for any threads that still have a non-NULL vcore ptr.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2683) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2684) for (i = 1; i < n_threads; ++i)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2685) if (paca_ptrs[cpu + i]->kvm_hstate.kvm_vcore)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2686) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2687) if (i == n_threads) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2688) HMT_medium();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2689) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2690) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2691) HMT_low();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2692) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2693) HMT_medium();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2694) for (i = 1; i < n_threads; ++i)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2695) if (paca_ptrs[cpu + i]->kvm_hstate.kvm_vcore)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2696) pr_err("KVM: CPU %d seems to be stuck\n", cpu + i);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2697) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2698)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2699) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2700) * Check that we are on thread 0 and that any other threads in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2701) * this core are off-line. Then grab the threads so they can't
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2702) * enter the kernel.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2703) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2704) static int on_primary_thread(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2705) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2706) int cpu = smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2707) int thr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2708)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2709) /* Are we on a primary subcore? */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2710) if (cpu_thread_in_subcore(cpu))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2711) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2712)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2713) thr = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2714) while (++thr < threads_per_subcore)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2715) if (cpu_online(cpu + thr))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2716) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2717)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2718) /* Grab all hw threads so they can't go into the kernel */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2719) for (thr = 1; thr < threads_per_subcore; ++thr) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2720) if (kvmppc_grab_hwthread(cpu + thr)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2721) /* Couldn't grab one; let the others go */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2722) do {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2723) kvmppc_release_hwthread(cpu + thr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2724) } while (--thr > 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2725) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2726) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2727) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2728) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2729) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2730)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2731) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2732) * A list of virtual cores for each physical CPU.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2733) * These are vcores that could run but their runner VCPU tasks are
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2734) * (or may be) preempted.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2735) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2736) struct preempted_vcore_list {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2737) struct list_head list;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2738) spinlock_t lock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2739) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2740)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2741) static DEFINE_PER_CPU(struct preempted_vcore_list, preempted_vcores);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2742)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2743) static void init_vcore_lists(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2744) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2745) int cpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2746)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2747) for_each_possible_cpu(cpu) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2748) struct preempted_vcore_list *lp = &per_cpu(preempted_vcores, cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2749) spin_lock_init(&lp->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2750) INIT_LIST_HEAD(&lp->list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2751) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2752) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2753)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2754) static void kvmppc_vcore_preempt(struct kvmppc_vcore *vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2755) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2756) struct preempted_vcore_list *lp = this_cpu_ptr(&preempted_vcores);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2757)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2758) vc->vcore_state = VCORE_PREEMPT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2759) vc->pcpu = smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2760) if (vc->num_threads < threads_per_vcore(vc->kvm)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2761) spin_lock(&lp->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2762) list_add_tail(&vc->preempt_list, &lp->list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2763) spin_unlock(&lp->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2764) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2765)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2766) /* Start accumulating stolen time */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2767) kvmppc_core_start_stolen(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2768) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2769)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2770) static void kvmppc_vcore_end_preempt(struct kvmppc_vcore *vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2771) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2772) struct preempted_vcore_list *lp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2773)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2774) kvmppc_core_end_stolen(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2775) if (!list_empty(&vc->preempt_list)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2776) lp = &per_cpu(preempted_vcores, vc->pcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2777) spin_lock(&lp->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2778) list_del_init(&vc->preempt_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2779) spin_unlock(&lp->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2780) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2781) vc->vcore_state = VCORE_INACTIVE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2782) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2783)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2784) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2785) * This stores information about the virtual cores currently
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2786) * assigned to a physical core.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2787) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2788) struct core_info {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2789) int n_subcores;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2790) int max_subcore_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2791) int total_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2792) int subcore_threads[MAX_SUBCORES];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2793) struct kvmppc_vcore *vc[MAX_SUBCORES];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2794) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2795)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2796) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2797) * This mapping means subcores 0 and 1 can use threads 0-3 and 4-7
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2798) * respectively in 2-way micro-threading (split-core) mode on POWER8.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2799) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2800) static int subcore_thread_map[MAX_SUBCORES] = { 0, 4, 2, 6 };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2801)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2802) static void init_core_info(struct core_info *cip, struct kvmppc_vcore *vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2803) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2804) memset(cip, 0, sizeof(*cip));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2805) cip->n_subcores = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2806) cip->max_subcore_threads = vc->num_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2807) cip->total_threads = vc->num_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2808) cip->subcore_threads[0] = vc->num_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2809) cip->vc[0] = vc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2810) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2811)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2812) static bool subcore_config_ok(int n_subcores, int n_threads)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2813) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2814) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2815) * POWER9 "SMT4" cores are permanently in what is effectively a 4-way
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2816) * split-core mode, with one thread per subcore.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2817) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2818) if (cpu_has_feature(CPU_FTR_ARCH_300))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2819) return n_subcores <= 4 && n_threads == 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2820)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2821) /* On POWER8, can only dynamically split if unsplit to begin with */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2822) if (n_subcores > 1 && threads_per_subcore < MAX_SMT_THREADS)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2823) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2824) if (n_subcores > MAX_SUBCORES)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2825) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2826) if (n_subcores > 1) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2827) if (!(dynamic_mt_modes & 2))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2828) n_subcores = 4;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2829) if (n_subcores > 2 && !(dynamic_mt_modes & 4))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2830) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2831) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2832)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2833) return n_subcores * roundup_pow_of_two(n_threads) <= MAX_SMT_THREADS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2834) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2835)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2836) static void init_vcore_to_run(struct kvmppc_vcore *vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2837) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2838) vc->entry_exit_map = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2839) vc->in_guest = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2840) vc->napping_threads = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2841) vc->conferring_threads = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2842) vc->tb_offset_applied = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2843) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2844)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2845) static bool can_dynamic_split(struct kvmppc_vcore *vc, struct core_info *cip)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2846) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2847) int n_threads = vc->num_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2848) int sub;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2849)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2850) if (!cpu_has_feature(CPU_FTR_ARCH_207S))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2851) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2852)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2853) /* In one_vm_per_core mode, require all vcores to be from the same vm */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2854) if (one_vm_per_core && vc->kvm != cip->vc[0]->kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2855) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2856)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2857) /* Some POWER9 chips require all threads to be in the same MMU mode */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2858) if (no_mixing_hpt_and_radix &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2859) kvm_is_radix(vc->kvm) != kvm_is_radix(cip->vc[0]->kvm))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2860) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2861)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2862) if (n_threads < cip->max_subcore_threads)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2863) n_threads = cip->max_subcore_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2864) if (!subcore_config_ok(cip->n_subcores + 1, n_threads))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2865) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2866) cip->max_subcore_threads = n_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2867)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2868) sub = cip->n_subcores;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2869) ++cip->n_subcores;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2870) cip->total_threads += vc->num_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2871) cip->subcore_threads[sub] = vc->num_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2872) cip->vc[sub] = vc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2873) init_vcore_to_run(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2874) list_del_init(&vc->preempt_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2875)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2876) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2877) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2878)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2879) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2880) * Work out whether it is possible to piggyback the execution of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2881) * vcore *pvc onto the execution of the other vcores described in *cip.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2882) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2883) static bool can_piggyback(struct kvmppc_vcore *pvc, struct core_info *cip,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2884) int target_threads)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2885) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2886) if (cip->total_threads + pvc->num_threads > target_threads)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2887) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2888)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2889) return can_dynamic_split(pvc, cip);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2890) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2891)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2892) static void prepare_threads(struct kvmppc_vcore *vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2893) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2894) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2895) struct kvm_vcpu *vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2896)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2897) for_each_runnable_thread(i, vcpu, vc) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2898) if (signal_pending(vcpu->arch.run_task))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2899) vcpu->arch.ret = -EINTR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2900) else if (vcpu->arch.vpa.update_pending ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2901) vcpu->arch.slb_shadow.update_pending ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2902) vcpu->arch.dtl.update_pending)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2903) vcpu->arch.ret = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2904) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2905) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2906) kvmppc_remove_runnable(vc, vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2907) wake_up(&vcpu->arch.cpu_run);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2908) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2909) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2910)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2911) static void collect_piggybacks(struct core_info *cip, int target_threads)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2912) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2913) struct preempted_vcore_list *lp = this_cpu_ptr(&preempted_vcores);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2914) struct kvmppc_vcore *pvc, *vcnext;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2915)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2916) spin_lock(&lp->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2917) list_for_each_entry_safe(pvc, vcnext, &lp->list, preempt_list) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2918) if (!spin_trylock(&pvc->lock))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2919) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2920) prepare_threads(pvc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2921) if (!pvc->n_runnable || !pvc->kvm->arch.mmu_ready) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2922) list_del_init(&pvc->preempt_list);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2923) if (pvc->runner == NULL) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2924) pvc->vcore_state = VCORE_INACTIVE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2925) kvmppc_core_end_stolen(pvc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2926) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2927) spin_unlock(&pvc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2928) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2929) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2930) if (!can_piggyback(pvc, cip, target_threads)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2931) spin_unlock(&pvc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2932) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2933) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2934) kvmppc_core_end_stolen(pvc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2935) pvc->vcore_state = VCORE_PIGGYBACK;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2936) if (cip->total_threads >= target_threads)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2937) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2938) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2939) spin_unlock(&lp->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2940) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2941)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2942) static bool recheck_signals_and_mmu(struct core_info *cip)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2943) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2944) int sub, i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2945) struct kvm_vcpu *vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2946) struct kvmppc_vcore *vc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2947)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2948) for (sub = 0; sub < cip->n_subcores; ++sub) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2949) vc = cip->vc[sub];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2950) if (!vc->kvm->arch.mmu_ready)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2951) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2952) for_each_runnable_thread(i, vcpu, vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2953) if (signal_pending(vcpu->arch.run_task))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2954) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2955) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2956) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2957) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2958)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2959) static void post_guest_process(struct kvmppc_vcore *vc, bool is_master)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2960) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2961) int still_running = 0, i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2962) u64 now;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2963) long ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2964) struct kvm_vcpu *vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2965)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2966) spin_lock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2967) now = get_tb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2968) for_each_runnable_thread(i, vcpu, vc) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2969) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2970) * It's safe to unlock the vcore in the loop here, because
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2971) * for_each_runnable_thread() is safe against removal of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2972) * the vcpu, and the vcore state is VCORE_EXITING here,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2973) * so any vcpus becoming runnable will have their arch.trap
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2974) * set to zero and can't actually run in the guest.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2975) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2976) spin_unlock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2977) /* cancel pending dec exception if dec is positive */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2978) if (now < vcpu->arch.dec_expires &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2979) kvmppc_core_pending_dec(vcpu))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2980) kvmppc_core_dequeue_dec(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2981)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2982) trace_kvm_guest_exit(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2983)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2984) ret = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2985) if (vcpu->arch.trap)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2986) ret = kvmppc_handle_exit_hv(vcpu,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2987) vcpu->arch.run_task);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2988)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2989) vcpu->arch.ret = ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2990) vcpu->arch.trap = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2991)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2992) spin_lock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2993) if (is_kvmppc_resume_guest(vcpu->arch.ret)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2994) if (vcpu->arch.pending_exceptions)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2995) kvmppc_core_prepare_to_enter(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2996) if (vcpu->arch.ceded)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2997) kvmppc_set_timer(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2998) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 2999) ++still_running;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3000) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3001) kvmppc_remove_runnable(vc, vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3002) wake_up(&vcpu->arch.cpu_run);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3003) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3004) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3005) if (!is_master) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3006) if (still_running > 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3007) kvmppc_vcore_preempt(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3008) } else if (vc->runner) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3009) vc->vcore_state = VCORE_PREEMPT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3010) kvmppc_core_start_stolen(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3011) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3012) vc->vcore_state = VCORE_INACTIVE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3013) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3014) if (vc->n_runnable > 0 && vc->runner == NULL) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3015) /* make sure there's a candidate runner awake */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3016) i = -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3017) vcpu = next_runnable_thread(vc, &i);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3018) wake_up(&vcpu->arch.cpu_run);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3019) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3020) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3021) spin_unlock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3022) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3023)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3024) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3025) * Clear core from the list of active host cores as we are about to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3026) * enter the guest. Only do this if it is the primary thread of the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3027) * core (not if a subcore) that is entering the guest.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3028) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3029) static inline int kvmppc_clear_host_core(unsigned int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3030) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3031) int core;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3032)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3033) if (!kvmppc_host_rm_ops_hv || cpu_thread_in_core(cpu))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3034) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3035) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3036) * Memory barrier can be omitted here as we will do a smp_wmb()
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3037) * later in kvmppc_start_thread and we need ensure that state is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3038) * visible to other CPUs only after we enter guest.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3039) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3040) core = cpu >> threads_shift;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3041) kvmppc_host_rm_ops_hv->rm_core[core].rm_state.in_host = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3042) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3043) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3044)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3045) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3046) * Advertise this core as an active host core since we exited the guest
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3047) * Only need to do this if it is the primary thread of the core that is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3048) * exiting.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3049) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3050) static inline int kvmppc_set_host_core(unsigned int cpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3051) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3052) int core;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3053)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3054) if (!kvmppc_host_rm_ops_hv || cpu_thread_in_core(cpu))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3055) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3056)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3057) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3058) * Memory barrier can be omitted here because we do a spin_unlock
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3059) * immediately after this which provides the memory barrier.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3060) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3061) core = cpu >> threads_shift;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3062) kvmppc_host_rm_ops_hv->rm_core[core].rm_state.in_host = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3063) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3064) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3065)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3066) static void set_irq_happened(int trap)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3067) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3068) switch (trap) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3069) case BOOK3S_INTERRUPT_EXTERNAL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3070) local_paca->irq_happened |= PACA_IRQ_EE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3071) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3072) case BOOK3S_INTERRUPT_H_DOORBELL:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3073) local_paca->irq_happened |= PACA_IRQ_DBELL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3074) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3075) case BOOK3S_INTERRUPT_HMI:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3076) local_paca->irq_happened |= PACA_IRQ_HMI;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3077) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3078) case BOOK3S_INTERRUPT_SYSTEM_RESET:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3079) replay_system_reset();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3080) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3081) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3082) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3083)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3084) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3085) * Run a set of guest threads on a physical core.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3086) * Called with vc->lock held.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3087) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3088) static noinline void kvmppc_run_core(struct kvmppc_vcore *vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3089) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3090) struct kvm_vcpu *vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3091) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3092) int srcu_idx;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3093) struct core_info core_info;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3094) struct kvmppc_vcore *pvc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3095) struct kvm_split_mode split_info, *sip;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3096) int split, subcore_size, active;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3097) int sub;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3098) bool thr0_done;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3099) unsigned long cmd_bit, stat_bit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3100) int pcpu, thr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3101) int target_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3102) int controlled_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3103) int trap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3104) bool is_power8;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3105) bool hpt_on_radix;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3106)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3107) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3108) * Remove from the list any threads that have a signal pending
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3109) * or need a VPA update done
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3110) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3111) prepare_threads(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3112)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3113) /* if the runner is no longer runnable, let the caller pick a new one */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3114) if (vc->runner->arch.state != KVMPPC_VCPU_RUNNABLE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3115) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3116)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3117) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3118) * Initialize *vc.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3119) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3120) init_vcore_to_run(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3121) vc->preempt_tb = TB_NIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3122)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3123) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3124) * Number of threads that we will be controlling: the same as
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3125) * the number of threads per subcore, except on POWER9,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3126) * where it's 1 because the threads are (mostly) independent.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3127) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3128) controlled_threads = threads_per_vcore(vc->kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3129)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3130) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3131) * Make sure we are running on primary threads, and that secondary
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3132) * threads are offline. Also check if the number of threads in this
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3133) * guest are greater than the current system threads per guest.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3134) * On POWER9, we need to be not in independent-threads mode if
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3135) * this is a HPT guest on a radix host machine where the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3136) * CPU threads may not be in different MMU modes.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3137) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3138) hpt_on_radix = no_mixing_hpt_and_radix && radix_enabled() &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3139) !kvm_is_radix(vc->kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3140) if (((controlled_threads > 1) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3141) ((vc->num_threads > threads_per_subcore) || !on_primary_thread())) ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3142) (hpt_on_radix && vc->kvm->arch.threads_indep)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3143) for_each_runnable_thread(i, vcpu, vc) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3144) vcpu->arch.ret = -EBUSY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3145) kvmppc_remove_runnable(vc, vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3146) wake_up(&vcpu->arch.cpu_run);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3147) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3148) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3149) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3150)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3151) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3152) * See if we could run any other vcores on the physical core
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3153) * along with this one.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3154) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3155) init_core_info(&core_info, vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3156) pcpu = smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3157) target_threads = controlled_threads;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3158) if (target_smt_mode && target_smt_mode < target_threads)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3159) target_threads = target_smt_mode;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3160) if (vc->num_threads < target_threads)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3161) collect_piggybacks(&core_info, target_threads);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3162)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3163) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3164) * On radix, arrange for TLB flushing if necessary.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3165) * This has to be done before disabling interrupts since
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3166) * it uses smp_call_function().
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3167) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3168) pcpu = smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3169) if (kvm_is_radix(vc->kvm)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3170) for (sub = 0; sub < core_info.n_subcores; ++sub)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3171) for_each_runnable_thread(i, vcpu, core_info.vc[sub])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3172) kvmppc_prepare_radix_vcpu(vcpu, pcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3173) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3174)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3175) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3176) * Hard-disable interrupts, and check resched flag and signals.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3177) * If we need to reschedule or deliver a signal, clean up
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3178) * and return without going into the guest(s).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3179) * If the mmu_ready flag has been cleared, don't go into the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3180) * guest because that means a HPT resize operation is in progress.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3181) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3182) local_irq_disable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3183) hard_irq_disable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3184) if (lazy_irq_pending() || need_resched() ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3185) recheck_signals_and_mmu(&core_info)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3186) local_irq_enable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3187) vc->vcore_state = VCORE_INACTIVE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3188) /* Unlock all except the primary vcore */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3189) for (sub = 1; sub < core_info.n_subcores; ++sub) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3190) pvc = core_info.vc[sub];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3191) /* Put back on to the preempted vcores list */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3192) kvmppc_vcore_preempt(pvc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3193) spin_unlock(&pvc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3194) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3195) for (i = 0; i < controlled_threads; ++i)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3196) kvmppc_release_hwthread(pcpu + i);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3197) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3198) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3199)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3200) kvmppc_clear_host_core(pcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3201)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3202) /* Decide on micro-threading (split-core) mode */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3203) subcore_size = threads_per_subcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3204) cmd_bit = stat_bit = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3205) split = core_info.n_subcores;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3206) sip = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3207) is_power8 = cpu_has_feature(CPU_FTR_ARCH_207S)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3208) && !cpu_has_feature(CPU_FTR_ARCH_300);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3209)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3210) if (split > 1 || hpt_on_radix) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3211) sip = &split_info;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3212) memset(&split_info, 0, sizeof(split_info));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3213) for (sub = 0; sub < core_info.n_subcores; ++sub)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3214) split_info.vc[sub] = core_info.vc[sub];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3215)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3216) if (is_power8) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3217) if (split == 2 && (dynamic_mt_modes & 2)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3218) cmd_bit = HID0_POWER8_1TO2LPAR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3219) stat_bit = HID0_POWER8_2LPARMODE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3220) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3221) split = 4;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3222) cmd_bit = HID0_POWER8_1TO4LPAR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3223) stat_bit = HID0_POWER8_4LPARMODE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3224) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3225) subcore_size = MAX_SMT_THREADS / split;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3226) split_info.rpr = mfspr(SPRN_RPR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3227) split_info.pmmar = mfspr(SPRN_PMMAR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3228) split_info.ldbar = mfspr(SPRN_LDBAR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3229) split_info.subcore_size = subcore_size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3230) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3231) split_info.subcore_size = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3232) if (hpt_on_radix) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3233) /* Use the split_info for LPCR/LPIDR changes */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3234) split_info.lpcr_req = vc->lpcr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3235) split_info.lpidr_req = vc->kvm->arch.lpid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3236) split_info.host_lpcr = vc->kvm->arch.host_lpcr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3237) split_info.do_set = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3238) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3239) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3240)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3241) /* order writes to split_info before kvm_split_mode pointer */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3242) smp_wmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3243) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3244)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3245) for (thr = 0; thr < controlled_threads; ++thr) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3246) struct paca_struct *paca = paca_ptrs[pcpu + thr];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3247)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3248) paca->kvm_hstate.tid = thr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3249) paca->kvm_hstate.napping = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3250) paca->kvm_hstate.kvm_split_mode = sip;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3251) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3252)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3253) /* Initiate micro-threading (split-core) on POWER8 if required */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3254) if (cmd_bit) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3255) unsigned long hid0 = mfspr(SPRN_HID0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3256)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3257) hid0 |= cmd_bit | HID0_POWER8_DYNLPARDIS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3258) mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3259) mtspr(SPRN_HID0, hid0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3260) isync();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3261) for (;;) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3262) hid0 = mfspr(SPRN_HID0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3263) if (hid0 & stat_bit)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3264) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3265) cpu_relax();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3266) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3267) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3268)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3269) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3270) * On POWER8, set RWMR register.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3271) * Since it only affects PURR and SPURR, it doesn't affect
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3272) * the host, so we don't save/restore the host value.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3273) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3274) if (is_power8) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3275) unsigned long rwmr_val = RWMR_RPA_P8_8THREAD;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3276) int n_online = atomic_read(&vc->online_count);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3277)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3278) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3279) * Use the 8-thread value if we're doing split-core
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3280) * or if the vcore's online count looks bogus.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3281) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3282) if (split == 1 && threads_per_subcore == MAX_SMT_THREADS &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3283) n_online >= 1 && n_online <= MAX_SMT_THREADS)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3284) rwmr_val = p8_rwmr_values[n_online];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3285) mtspr(SPRN_RWMR, rwmr_val);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3286) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3287)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3288) /* Start all the threads */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3289) active = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3290) for (sub = 0; sub < core_info.n_subcores; ++sub) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3291) thr = is_power8 ? subcore_thread_map[sub] : sub;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3292) thr0_done = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3293) active |= 1 << thr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3294) pvc = core_info.vc[sub];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3295) pvc->pcpu = pcpu + thr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3296) for_each_runnable_thread(i, vcpu, pvc) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3297) kvmppc_start_thread(vcpu, pvc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3298) kvmppc_create_dtl_entry(vcpu, pvc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3299) trace_kvm_guest_enter(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3300) if (!vcpu->arch.ptid)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3301) thr0_done = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3302) active |= 1 << (thr + vcpu->arch.ptid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3303) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3304) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3305) * We need to start the first thread of each subcore
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3306) * even if it doesn't have a vcpu.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3307) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3308) if (!thr0_done)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3309) kvmppc_start_thread(NULL, pvc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3310) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3311)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3312) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3313) * Ensure that split_info.do_nap is set after setting
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3314) * the vcore pointer in the PACA of the secondaries.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3315) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3316) smp_mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3317)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3318) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3319) * When doing micro-threading, poke the inactive threads as well.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3320) * This gets them to the nap instruction after kvm_do_nap,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3321) * which reduces the time taken to unsplit later.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3322) * For POWER9 HPT guest on radix host, we need all the secondary
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3323) * threads woken up so they can do the LPCR/LPIDR change.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3324) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3325) if (cmd_bit || hpt_on_radix) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3326) split_info.do_nap = 1; /* ask secondaries to nap when done */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3327) for (thr = 1; thr < threads_per_subcore; ++thr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3328) if (!(active & (1 << thr)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3329) kvmppc_ipi_thread(pcpu + thr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3330) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3331)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3332) vc->vcore_state = VCORE_RUNNING;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3333) preempt_disable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3334)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3335) trace_kvmppc_run_core(vc, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3336)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3337) for (sub = 0; sub < core_info.n_subcores; ++sub)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3338) spin_unlock(&core_info.vc[sub]->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3339)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3340) guest_enter_irqoff();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3341)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3342) srcu_idx = srcu_read_lock(&vc->kvm->srcu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3343)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3344) this_cpu_disable_ftrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3345)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3346) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3347) * Interrupts will be enabled once we get into the guest,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3348) * so tell lockdep that we're about to enable interrupts.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3349) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3350) trace_hardirqs_on();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3351)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3352) trap = __kvmppc_vcore_entry();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3353)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3354) trace_hardirqs_off();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3355)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3356) this_cpu_enable_ftrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3357)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3358) srcu_read_unlock(&vc->kvm->srcu, srcu_idx);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3359)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3360) set_irq_happened(trap);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3361)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3362) spin_lock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3363) /* prevent other vcpu threads from doing kvmppc_start_thread() now */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3364) vc->vcore_state = VCORE_EXITING;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3365)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3366) /* wait for secondary threads to finish writing their state to memory */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3367) kvmppc_wait_for_nap(controlled_threads);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3368)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3369) /* Return to whole-core mode if we split the core earlier */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3370) if (cmd_bit) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3371) unsigned long hid0 = mfspr(SPRN_HID0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3372) unsigned long loops = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3373)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3374) hid0 &= ~HID0_POWER8_DYNLPARDIS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3375) stat_bit = HID0_POWER8_2LPARMODE | HID0_POWER8_4LPARMODE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3376) mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3377) mtspr(SPRN_HID0, hid0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3378) isync();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3379) for (;;) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3380) hid0 = mfspr(SPRN_HID0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3381) if (!(hid0 & stat_bit))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3382) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3383) cpu_relax();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3384) ++loops;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3385) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3386) } else if (hpt_on_radix) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3387) /* Wait for all threads to have seen final sync */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3388) for (thr = 1; thr < controlled_threads; ++thr) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3389) struct paca_struct *paca = paca_ptrs[pcpu + thr];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3390)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3391) while (paca->kvm_hstate.kvm_split_mode) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3392) HMT_low();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3393) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3394) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3395) HMT_medium();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3396) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3397) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3398) split_info.do_nap = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3399)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3400) kvmppc_set_host_core(pcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3401)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3402) local_irq_enable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3403) guest_exit();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3404)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3405) /* Let secondaries go back to the offline loop */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3406) for (i = 0; i < controlled_threads; ++i) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3407) kvmppc_release_hwthread(pcpu + i);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3408) if (sip && sip->napped[i])
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3409) kvmppc_ipi_thread(pcpu + i);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3410) cpumask_clear_cpu(pcpu + i, &vc->kvm->arch.cpu_in_guest);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3411) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3412)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3413) spin_unlock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3414)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3415) /* make sure updates to secondary vcpu structs are visible now */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3416) smp_mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3417)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3418) preempt_enable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3419)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3420) for (sub = 0; sub < core_info.n_subcores; ++sub) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3421) pvc = core_info.vc[sub];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3422) post_guest_process(pvc, pvc == vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3423) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3424)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3425) spin_lock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3426)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3427) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3428) vc->vcore_state = VCORE_INACTIVE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3429) trace_kvmppc_run_core(vc, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3430) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3431)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3432) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3433) * Load up hypervisor-mode registers on P9.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3434) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3435) static int kvmhv_load_hv_regs_and_go(struct kvm_vcpu *vcpu, u64 time_limit,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3436) unsigned long lpcr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3437) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3438) struct kvmppc_vcore *vc = vcpu->arch.vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3439) s64 hdec;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3440) u64 tb, purr, spurr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3441) int trap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3442) unsigned long host_hfscr = mfspr(SPRN_HFSCR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3443) unsigned long host_ciabr = mfspr(SPRN_CIABR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3444) unsigned long host_dawr = mfspr(SPRN_DAWR0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3445) unsigned long host_dawrx = mfspr(SPRN_DAWRX0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3446) unsigned long host_psscr = mfspr(SPRN_PSSCR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3447) unsigned long host_pidr = mfspr(SPRN_PID);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3448)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3449) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3450) * P8 and P9 suppress the HDEC exception when LPCR[HDICE] = 0,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3451) * so set HDICE before writing HDEC.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3452) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3453) mtspr(SPRN_LPCR, vcpu->kvm->arch.host_lpcr | LPCR_HDICE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3454) isync();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3455)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3456) hdec = time_limit - mftb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3457) if (hdec < 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3458) mtspr(SPRN_LPCR, vcpu->kvm->arch.host_lpcr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3459) isync();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3460) return BOOK3S_INTERRUPT_HV_DECREMENTER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3461) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3462) mtspr(SPRN_HDEC, hdec);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3463)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3464) if (vc->tb_offset) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3465) u64 new_tb = mftb() + vc->tb_offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3466) mtspr(SPRN_TBU40, new_tb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3467) tb = mftb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3468) if ((tb & 0xffffff) < (new_tb & 0xffffff))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3469) mtspr(SPRN_TBU40, new_tb + 0x1000000);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3470) vc->tb_offset_applied = vc->tb_offset;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3471) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3472)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3473) if (vc->pcr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3474) mtspr(SPRN_PCR, vc->pcr | PCR_MASK);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3475) mtspr(SPRN_DPDES, vc->dpdes);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3476) mtspr(SPRN_VTB, vc->vtb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3477)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3478) local_paca->kvm_hstate.host_purr = mfspr(SPRN_PURR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3479) local_paca->kvm_hstate.host_spurr = mfspr(SPRN_SPURR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3480) mtspr(SPRN_PURR, vcpu->arch.purr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3481) mtspr(SPRN_SPURR, vcpu->arch.spurr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3482)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3483) if (dawr_enabled()) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3484) mtspr(SPRN_DAWR0, vcpu->arch.dawr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3485) mtspr(SPRN_DAWRX0, vcpu->arch.dawrx);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3486) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3487) mtspr(SPRN_CIABR, vcpu->arch.ciabr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3488) mtspr(SPRN_IC, vcpu->arch.ic);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3489) mtspr(SPRN_PID, vcpu->arch.pid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3490)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3491) mtspr(SPRN_PSSCR, vcpu->arch.psscr | PSSCR_EC |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3492) (local_paca->kvm_hstate.fake_suspend << PSSCR_FAKE_SUSPEND_LG));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3493)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3494) mtspr(SPRN_HFSCR, vcpu->arch.hfscr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3495)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3496) mtspr(SPRN_SPRG0, vcpu->arch.shregs.sprg0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3497) mtspr(SPRN_SPRG1, vcpu->arch.shregs.sprg1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3498) mtspr(SPRN_SPRG2, vcpu->arch.shregs.sprg2);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3499) mtspr(SPRN_SPRG3, vcpu->arch.shregs.sprg3);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3500)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3501) mtspr(SPRN_AMOR, ~0UL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3502)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3503) mtspr(SPRN_LPCR, lpcr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3504) isync();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3505)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3506) kvmppc_xive_push_vcpu(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3507)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3508) mtspr(SPRN_SRR0, vcpu->arch.shregs.srr0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3509) mtspr(SPRN_SRR1, vcpu->arch.shregs.srr1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3510)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3511) trap = __kvmhv_vcpu_entry_p9(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3512)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3513) /* Advance host PURR/SPURR by the amount used by guest */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3514) purr = mfspr(SPRN_PURR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3515) spurr = mfspr(SPRN_SPURR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3516) mtspr(SPRN_PURR, local_paca->kvm_hstate.host_purr +
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3517) purr - vcpu->arch.purr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3518) mtspr(SPRN_SPURR, local_paca->kvm_hstate.host_spurr +
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3519) spurr - vcpu->arch.spurr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3520) vcpu->arch.purr = purr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3521) vcpu->arch.spurr = spurr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3522)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3523) vcpu->arch.ic = mfspr(SPRN_IC);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3524) vcpu->arch.pid = mfspr(SPRN_PID);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3525) vcpu->arch.psscr = mfspr(SPRN_PSSCR) & PSSCR_GUEST_VIS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3526)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3527) vcpu->arch.shregs.sprg0 = mfspr(SPRN_SPRG0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3528) vcpu->arch.shregs.sprg1 = mfspr(SPRN_SPRG1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3529) vcpu->arch.shregs.sprg2 = mfspr(SPRN_SPRG2);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3530) vcpu->arch.shregs.sprg3 = mfspr(SPRN_SPRG3);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3531)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3532) /* Preserve PSSCR[FAKE_SUSPEND] until we've called kvmppc_save_tm_hv */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3533) mtspr(SPRN_PSSCR, host_psscr |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3534) (local_paca->kvm_hstate.fake_suspend << PSSCR_FAKE_SUSPEND_LG));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3535) mtspr(SPRN_HFSCR, host_hfscr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3536) mtspr(SPRN_CIABR, host_ciabr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3537) mtspr(SPRN_DAWR0, host_dawr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3538) mtspr(SPRN_DAWRX0, host_dawrx);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3539) mtspr(SPRN_PID, host_pidr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3540)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3541) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3542) * Since this is radix, do a eieio; tlbsync; ptesync sequence in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3543) * case we interrupted the guest between a tlbie and a ptesync.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3544) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3545) asm volatile("eieio; tlbsync; ptesync");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3546)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3547) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3548) * cp_abort is required if the processor supports local copy-paste
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3549) * to clear the copy buffer that was under control of the guest.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3550) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3551) if (cpu_has_feature(CPU_FTR_ARCH_31))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3552) asm volatile(PPC_CP_ABORT);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3553)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3554) mtspr(SPRN_LPID, vcpu->kvm->arch.host_lpid); /* restore host LPID */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3555) isync();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3556)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3557) vc->dpdes = mfspr(SPRN_DPDES);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3558) vc->vtb = mfspr(SPRN_VTB);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3559) mtspr(SPRN_DPDES, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3560) if (vc->pcr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3561) mtspr(SPRN_PCR, PCR_MASK);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3562)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3563) if (vc->tb_offset_applied) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3564) u64 new_tb = mftb() - vc->tb_offset_applied;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3565) mtspr(SPRN_TBU40, new_tb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3566) tb = mftb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3567) if ((tb & 0xffffff) < (new_tb & 0xffffff))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3568) mtspr(SPRN_TBU40, new_tb + 0x1000000);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3569) vc->tb_offset_applied = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3570) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3571)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3572) mtspr(SPRN_HDEC, 0x7fffffff);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3573) mtspr(SPRN_LPCR, vcpu->kvm->arch.host_lpcr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3574)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3575) return trap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3576) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3577)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3578) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3579) * Virtual-mode guest entry for POWER9 and later when the host and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3580) * guest are both using the radix MMU. The LPIDR has already been set.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3581) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3582) static int kvmhv_p9_guest_entry(struct kvm_vcpu *vcpu, u64 time_limit,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3583) unsigned long lpcr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3584) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3585) struct kvmppc_vcore *vc = vcpu->arch.vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3586) unsigned long host_dscr = mfspr(SPRN_DSCR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3587) unsigned long host_tidr = mfspr(SPRN_TIDR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3588) unsigned long host_iamr = mfspr(SPRN_IAMR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3589) unsigned long host_amr = mfspr(SPRN_AMR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3590) unsigned long host_fscr = mfspr(SPRN_FSCR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3591) s64 dec;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3592) u64 tb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3593) int trap, save_pmu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3594)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3595) dec = mfspr(SPRN_DEC);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3596) tb = mftb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3597) if (dec < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3598) return BOOK3S_INTERRUPT_HV_DECREMENTER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3599) local_paca->kvm_hstate.dec_expires = dec + tb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3600) if (local_paca->kvm_hstate.dec_expires < time_limit)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3601) time_limit = local_paca->kvm_hstate.dec_expires;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3602)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3603) vcpu->arch.ceded = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3604)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3605) kvmhv_save_host_pmu(); /* saves it to PACA kvm_hstate */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3606)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3607) kvmppc_subcore_enter_guest();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3608)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3609) vc->entry_exit_map = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3610) vc->in_guest = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3611)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3612) if (vcpu->arch.vpa.pinned_addr) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3613) struct lppaca *lp = vcpu->arch.vpa.pinned_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3614) u32 yield_count = be32_to_cpu(lp->yield_count) + 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3615) lp->yield_count = cpu_to_be32(yield_count);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3616) vcpu->arch.vpa.dirty = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3617) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3618)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3619) if (cpu_has_feature(CPU_FTR_TM) ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3620) cpu_has_feature(CPU_FTR_P9_TM_HV_ASSIST))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3621) kvmppc_restore_tm_hv(vcpu, vcpu->arch.shregs.msr, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3622)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3623) #ifdef CONFIG_PPC_PSERIES
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3624) if (kvmhv_on_pseries()) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3625) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3626) if (vcpu->arch.vpa.pinned_addr) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3627) struct lppaca *lp = vcpu->arch.vpa.pinned_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3628) get_lppaca()->pmcregs_in_use = lp->pmcregs_in_use;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3629) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3630) get_lppaca()->pmcregs_in_use = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3631) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3632) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3633) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3634) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3635) kvmhv_load_guest_pmu(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3636)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3637) msr_check_and_set(MSR_FP | MSR_VEC | MSR_VSX);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3638) load_fp_state(&vcpu->arch.fp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3639) #ifdef CONFIG_ALTIVEC
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3640) load_vr_state(&vcpu->arch.vr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3641) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3642) mtspr(SPRN_VRSAVE, vcpu->arch.vrsave);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3643)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3644) mtspr(SPRN_DSCR, vcpu->arch.dscr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3645) mtspr(SPRN_IAMR, vcpu->arch.iamr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3646) mtspr(SPRN_PSPB, vcpu->arch.pspb);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3647) mtspr(SPRN_FSCR, vcpu->arch.fscr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3648) mtspr(SPRN_TAR, vcpu->arch.tar);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3649) mtspr(SPRN_EBBHR, vcpu->arch.ebbhr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3650) mtspr(SPRN_EBBRR, vcpu->arch.ebbrr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3651) mtspr(SPRN_BESCR, vcpu->arch.bescr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3652) mtspr(SPRN_WORT, vcpu->arch.wort);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3653) mtspr(SPRN_TIDR, vcpu->arch.tid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3654) mtspr(SPRN_DAR, vcpu->arch.shregs.dar);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3655) mtspr(SPRN_DSISR, vcpu->arch.shregs.dsisr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3656) mtspr(SPRN_AMR, vcpu->arch.amr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3657) mtspr(SPRN_UAMOR, vcpu->arch.uamor);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3658)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3659) if (!(vcpu->arch.ctrl & 1))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3660) mtspr(SPRN_CTRLT, mfspr(SPRN_CTRLF) & ~1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3661)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3662) mtspr(SPRN_DEC, vcpu->arch.dec_expires - mftb());
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3663)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3664) if (kvmhv_on_pseries()) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3665) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3666) * We need to save and restore the guest visible part of the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3667) * psscr (i.e. using SPRN_PSSCR_PR) since the hypervisor
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3668) * doesn't do this for us. Note only required if pseries since
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3669) * this is done in kvmhv_load_hv_regs_and_go() below otherwise.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3670) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3671) unsigned long host_psscr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3672) /* call our hypervisor to load up HV regs and go */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3673) struct hv_guest_state hvregs;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3674)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3675) host_psscr = mfspr(SPRN_PSSCR_PR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3676) mtspr(SPRN_PSSCR_PR, vcpu->arch.psscr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3677) kvmhv_save_hv_regs(vcpu, &hvregs);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3678) hvregs.lpcr = lpcr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3679) vcpu->arch.regs.msr = vcpu->arch.shregs.msr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3680) hvregs.version = HV_GUEST_STATE_VERSION;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3681) if (vcpu->arch.nested) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3682) hvregs.lpid = vcpu->arch.nested->shadow_lpid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3683) hvregs.vcpu_token = vcpu->arch.nested_vcpu_id;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3684) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3685) hvregs.lpid = vcpu->kvm->arch.lpid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3686) hvregs.vcpu_token = vcpu->vcpu_id;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3687) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3688) hvregs.hdec_expiry = time_limit;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3689) trap = plpar_hcall_norets(H_ENTER_NESTED, __pa(&hvregs),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3690) __pa(&vcpu->arch.regs));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3691) kvmhv_restore_hv_return_state(vcpu, &hvregs);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3692) vcpu->arch.shregs.msr = vcpu->arch.regs.msr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3693) vcpu->arch.shregs.dar = mfspr(SPRN_DAR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3694) vcpu->arch.shregs.dsisr = mfspr(SPRN_DSISR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3695) vcpu->arch.psscr = mfspr(SPRN_PSSCR_PR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3696) mtspr(SPRN_PSSCR_PR, host_psscr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3697)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3698) /* H_CEDE has to be handled now, not later */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3699) if (trap == BOOK3S_INTERRUPT_SYSCALL && !vcpu->arch.nested &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3700) kvmppc_get_gpr(vcpu, 3) == H_CEDE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3701) kvmppc_nested_cede(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3702) kvmppc_set_gpr(vcpu, 3, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3703) trap = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3704) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3705) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3706) trap = kvmhv_load_hv_regs_and_go(vcpu, time_limit, lpcr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3707) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3708)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3709) vcpu->arch.slb_max = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3710) dec = mfspr(SPRN_DEC);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3711) if (!(lpcr & LPCR_LD)) /* Sign extend if not using large decrementer */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3712) dec = (s32) dec;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3713) tb = mftb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3714) vcpu->arch.dec_expires = dec + tb;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3715) vcpu->cpu = -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3716) vcpu->arch.thread_cpu = -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3717) /* Save guest CTRL register, set runlatch to 1 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3718) vcpu->arch.ctrl = mfspr(SPRN_CTRLF);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3719) if (!(vcpu->arch.ctrl & 1))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3720) mtspr(SPRN_CTRLT, vcpu->arch.ctrl | 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3721)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3722) vcpu->arch.iamr = mfspr(SPRN_IAMR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3723) vcpu->arch.pspb = mfspr(SPRN_PSPB);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3724) vcpu->arch.fscr = mfspr(SPRN_FSCR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3725) vcpu->arch.tar = mfspr(SPRN_TAR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3726) vcpu->arch.ebbhr = mfspr(SPRN_EBBHR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3727) vcpu->arch.ebbrr = mfspr(SPRN_EBBRR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3728) vcpu->arch.bescr = mfspr(SPRN_BESCR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3729) vcpu->arch.wort = mfspr(SPRN_WORT);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3730) vcpu->arch.tid = mfspr(SPRN_TIDR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3731) vcpu->arch.amr = mfspr(SPRN_AMR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3732) vcpu->arch.uamor = mfspr(SPRN_UAMOR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3733) vcpu->arch.dscr = mfspr(SPRN_DSCR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3734)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3735) mtspr(SPRN_PSPB, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3736) mtspr(SPRN_WORT, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3737) mtspr(SPRN_UAMOR, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3738) mtspr(SPRN_DSCR, host_dscr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3739) mtspr(SPRN_TIDR, host_tidr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3740) mtspr(SPRN_IAMR, host_iamr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3741) mtspr(SPRN_PSPB, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3742)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3743) if (host_amr != vcpu->arch.amr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3744) mtspr(SPRN_AMR, host_amr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3745)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3746) if (host_fscr != vcpu->arch.fscr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3747) mtspr(SPRN_FSCR, host_fscr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3748)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3749) msr_check_and_set(MSR_FP | MSR_VEC | MSR_VSX);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3750) store_fp_state(&vcpu->arch.fp);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3751) #ifdef CONFIG_ALTIVEC
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3752) store_vr_state(&vcpu->arch.vr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3753) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3754) vcpu->arch.vrsave = mfspr(SPRN_VRSAVE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3755)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3756) if (cpu_has_feature(CPU_FTR_TM) ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3757) cpu_has_feature(CPU_FTR_P9_TM_HV_ASSIST))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3758) kvmppc_save_tm_hv(vcpu, vcpu->arch.shregs.msr, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3759)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3760) save_pmu = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3761) if (vcpu->arch.vpa.pinned_addr) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3762) struct lppaca *lp = vcpu->arch.vpa.pinned_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3763) u32 yield_count = be32_to_cpu(lp->yield_count) + 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3764) lp->yield_count = cpu_to_be32(yield_count);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3765) vcpu->arch.vpa.dirty = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3766) save_pmu = lp->pmcregs_in_use;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3767) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3768) /* Must save pmu if this guest is capable of running nested guests */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3769) save_pmu |= nesting_enabled(vcpu->kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3770)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3771) kvmhv_save_guest_pmu(vcpu, save_pmu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3772) #ifdef CONFIG_PPC_PSERIES
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3773) if (kvmhv_on_pseries()) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3774) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3775) get_lppaca()->pmcregs_in_use = ppc_get_pmu_inuse();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3776) barrier();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3777) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3778) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3779)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3780) vc->entry_exit_map = 0x101;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3781) vc->in_guest = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3782)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3783) mtspr(SPRN_DEC, local_paca->kvm_hstate.dec_expires - mftb());
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3784) mtspr(SPRN_SPRG_VDSO_WRITE, local_paca->sprg_vdso);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3785)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3786) kvmhv_load_host_pmu();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3787)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3788) kvmppc_subcore_exit_guest();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3789)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3790) return trap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3791) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3792)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3793) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3794) * Wait for some other vcpu thread to execute us, and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3795) * wake us up when we need to handle something in the host.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3796) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3797) static void kvmppc_wait_for_exec(struct kvmppc_vcore *vc,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3798) struct kvm_vcpu *vcpu, int wait_state)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3799) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3800) DEFINE_WAIT(wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3801)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3802) prepare_to_wait(&vcpu->arch.cpu_run, &wait, wait_state);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3803) if (vcpu->arch.state == KVMPPC_VCPU_RUNNABLE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3804) spin_unlock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3805) schedule();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3806) spin_lock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3807) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3808) finish_wait(&vcpu->arch.cpu_run, &wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3809) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3810)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3811) static void grow_halt_poll_ns(struct kvmppc_vcore *vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3812) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3813) if (!halt_poll_ns_grow)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3814) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3815)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3816) vc->halt_poll_ns *= halt_poll_ns_grow;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3817) if (vc->halt_poll_ns < halt_poll_ns_grow_start)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3818) vc->halt_poll_ns = halt_poll_ns_grow_start;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3819) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3820)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3821) static void shrink_halt_poll_ns(struct kvmppc_vcore *vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3822) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3823) if (halt_poll_ns_shrink == 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3824) vc->halt_poll_ns = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3825) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3826) vc->halt_poll_ns /= halt_poll_ns_shrink;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3827) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3828)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3829) #ifdef CONFIG_KVM_XICS
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3830) static inline bool xive_interrupt_pending(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3831) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3832) if (!xics_on_xive())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3833) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3834) return vcpu->arch.irq_pending || vcpu->arch.xive_saved_state.pipr <
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3835) vcpu->arch.xive_saved_state.cppr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3836) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3837) #else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3838) static inline bool xive_interrupt_pending(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3839) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3840) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3841) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3842) #endif /* CONFIG_KVM_XICS */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3843)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3844) static bool kvmppc_vcpu_woken(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3845) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3846) if (vcpu->arch.pending_exceptions || vcpu->arch.prodded ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3847) kvmppc_doorbell_pending(vcpu) || xive_interrupt_pending(vcpu))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3848) return true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3849)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3850) return false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3851) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3852)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3853) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3854) * Check to see if any of the runnable vcpus on the vcore have pending
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3855) * exceptions or are no longer ceded
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3856) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3857) static int kvmppc_vcore_check_block(struct kvmppc_vcore *vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3858) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3859) struct kvm_vcpu *vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3860) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3861)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3862) for_each_runnable_thread(i, vcpu, vc) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3863) if (!vcpu->arch.ceded || kvmppc_vcpu_woken(vcpu))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3864) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3865) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3866)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3867) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3868) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3869)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3870) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3871) * All the vcpus in this vcore are idle, so wait for a decrementer
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3872) * or external interrupt to one of the vcpus. vc->lock is held.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3873) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3874) static void kvmppc_vcore_blocked(struct kvmppc_vcore *vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3875) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3876) ktime_t cur, start_poll, start_wait;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3877) int do_sleep = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3878) u64 block_ns;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3879)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3880) /* Poll for pending exceptions and ceded state */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3881) cur = start_poll = ktime_get();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3882) if (vc->halt_poll_ns) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3883) ktime_t stop = ktime_add_ns(start_poll, vc->halt_poll_ns);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3884) ++vc->runner->stat.halt_attempted_poll;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3885)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3886) vc->vcore_state = VCORE_POLLING;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3887) spin_unlock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3888)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3889) do {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3890) if (kvmppc_vcore_check_block(vc)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3891) do_sleep = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3892) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3893) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3894) cur = ktime_get();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3895) } while (single_task_running() && ktime_before(cur, stop));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3896)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3897) spin_lock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3898) vc->vcore_state = VCORE_INACTIVE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3899)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3900) if (!do_sleep) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3901) ++vc->runner->stat.halt_successful_poll;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3902) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3903) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3904) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3905)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3906) prepare_to_rcuwait(&vc->wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3907) set_current_state(TASK_INTERRUPTIBLE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3908) if (kvmppc_vcore_check_block(vc)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3909) finish_rcuwait(&vc->wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3910) do_sleep = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3911) /* If we polled, count this as a successful poll */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3912) if (vc->halt_poll_ns)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3913) ++vc->runner->stat.halt_successful_poll;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3914) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3915) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3916)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3917) start_wait = ktime_get();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3918)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3919) vc->vcore_state = VCORE_SLEEPING;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3920) trace_kvmppc_vcore_blocked(vc, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3921) spin_unlock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3922) schedule();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3923) finish_rcuwait(&vc->wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3924) spin_lock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3925) vc->vcore_state = VCORE_INACTIVE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3926) trace_kvmppc_vcore_blocked(vc, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3927) ++vc->runner->stat.halt_successful_wait;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3928)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3929) cur = ktime_get();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3930)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3931) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3932) block_ns = ktime_to_ns(cur) - ktime_to_ns(start_poll);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3933)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3934) /* Attribute wait time */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3935) if (do_sleep) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3936) vc->runner->stat.halt_wait_ns +=
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3937) ktime_to_ns(cur) - ktime_to_ns(start_wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3938) /* Attribute failed poll time */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3939) if (vc->halt_poll_ns)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3940) vc->runner->stat.halt_poll_fail_ns +=
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3941) ktime_to_ns(start_wait) -
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3942) ktime_to_ns(start_poll);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3943) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3944) /* Attribute successful poll time */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3945) if (vc->halt_poll_ns)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3946) vc->runner->stat.halt_poll_success_ns +=
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3947) ktime_to_ns(cur) -
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3948) ktime_to_ns(start_poll);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3949) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3950)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3951) /* Adjust poll time */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3952) if (halt_poll_ns) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3953) if (block_ns <= vc->halt_poll_ns)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3954) ;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3955) /* We slept and blocked for longer than the max halt time */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3956) else if (vc->halt_poll_ns && block_ns > halt_poll_ns)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3957) shrink_halt_poll_ns(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3958) /* We slept and our poll time is too small */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3959) else if (vc->halt_poll_ns < halt_poll_ns &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3960) block_ns < halt_poll_ns)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3961) grow_halt_poll_ns(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3962) if (vc->halt_poll_ns > halt_poll_ns)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3963) vc->halt_poll_ns = halt_poll_ns;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3964) } else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3965) vc->halt_poll_ns = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3966)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3967) trace_kvmppc_vcore_wakeup(do_sleep, block_ns);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3968) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3969)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3970) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3971) * This never fails for a radix guest, as none of the operations it does
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3972) * for a radix guest can fail or have a way to report failure.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3973) * kvmhv_run_single_vcpu() relies on this fact.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3974) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3975) static int kvmhv_setup_mmu(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3976) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3977) int r = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3978) struct kvm *kvm = vcpu->kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3979)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3980) mutex_lock(&kvm->arch.mmu_setup_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3981) if (!kvm->arch.mmu_ready) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3982) if (!kvm_is_radix(kvm))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3983) r = kvmppc_hv_setup_htab_rma(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3984) if (!r) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3985) if (cpu_has_feature(CPU_FTR_ARCH_300))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3986) kvmppc_setup_partition_table(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3987) kvm->arch.mmu_ready = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3988) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3989) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3990) mutex_unlock(&kvm->arch.mmu_setup_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3991) return r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3992) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3993)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3994) static int kvmppc_run_vcpu(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3995) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3996) struct kvm_run *run = vcpu->run;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3997) int n_ceded, i, r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3998) struct kvmppc_vcore *vc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 3999) struct kvm_vcpu *v;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4000)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4001) trace_kvmppc_run_vcpu_enter(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4002)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4003) run->exit_reason = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4004) vcpu->arch.ret = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4005) vcpu->arch.trap = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4006) kvmppc_update_vpas(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4007)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4008) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4009) * Synchronize with other threads in this virtual core
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4010) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4011) vc = vcpu->arch.vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4012) spin_lock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4013) vcpu->arch.ceded = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4014) vcpu->arch.run_task = current;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4015) vcpu->arch.stolen_logged = vcore_stolen_time(vc, mftb());
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4016) vcpu->arch.state = KVMPPC_VCPU_RUNNABLE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4017) vcpu->arch.busy_preempt = TB_NIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4018) WRITE_ONCE(vc->runnable_threads[vcpu->arch.ptid], vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4019) ++vc->n_runnable;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4020)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4021) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4022) * This happens the first time this is called for a vcpu.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4023) * If the vcore is already running, we may be able to start
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4024) * this thread straight away and have it join in.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4025) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4026) if (!signal_pending(current)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4027) if ((vc->vcore_state == VCORE_PIGGYBACK ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4028) vc->vcore_state == VCORE_RUNNING) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4029) !VCORE_IS_EXITING(vc)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4030) kvmppc_create_dtl_entry(vcpu, vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4031) kvmppc_start_thread(vcpu, vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4032) trace_kvm_guest_enter(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4033) } else if (vc->vcore_state == VCORE_SLEEPING) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4034) rcuwait_wake_up(&vc->wait);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4035) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4036)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4037) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4038)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4039) while (vcpu->arch.state == KVMPPC_VCPU_RUNNABLE &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4040) !signal_pending(current)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4041) /* See if the MMU is ready to go */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4042) if (!vcpu->kvm->arch.mmu_ready) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4043) spin_unlock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4044) r = kvmhv_setup_mmu(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4045) spin_lock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4046) if (r) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4047) run->exit_reason = KVM_EXIT_FAIL_ENTRY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4048) run->fail_entry.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4049) hardware_entry_failure_reason = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4050) vcpu->arch.ret = r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4051) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4052) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4053) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4054)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4055) if (vc->vcore_state == VCORE_PREEMPT && vc->runner == NULL)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4056) kvmppc_vcore_end_preempt(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4057)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4058) if (vc->vcore_state != VCORE_INACTIVE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4059) kvmppc_wait_for_exec(vc, vcpu, TASK_INTERRUPTIBLE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4060) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4061) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4062) for_each_runnable_thread(i, v, vc) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4063) kvmppc_core_prepare_to_enter(v);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4064) if (signal_pending(v->arch.run_task)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4065) kvmppc_remove_runnable(vc, v);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4066) v->stat.signal_exits++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4067) v->run->exit_reason = KVM_EXIT_INTR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4068) v->arch.ret = -EINTR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4069) wake_up(&v->arch.cpu_run);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4070) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4071) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4072) if (!vc->n_runnable || vcpu->arch.state != KVMPPC_VCPU_RUNNABLE)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4073) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4074) n_ceded = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4075) for_each_runnable_thread(i, v, vc) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4076) if (!kvmppc_vcpu_woken(v))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4077) n_ceded += v->arch.ceded;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4078) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4079) v->arch.ceded = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4080) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4081) vc->runner = vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4082) if (n_ceded == vc->n_runnable) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4083) kvmppc_vcore_blocked(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4084) } else if (need_resched()) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4085) kvmppc_vcore_preempt(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4086) /* Let something else run */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4087) cond_resched_lock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4088) if (vc->vcore_state == VCORE_PREEMPT)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4089) kvmppc_vcore_end_preempt(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4090) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4091) kvmppc_run_core(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4092) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4093) vc->runner = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4094) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4095)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4096) while (vcpu->arch.state == KVMPPC_VCPU_RUNNABLE &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4097) (vc->vcore_state == VCORE_RUNNING ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4098) vc->vcore_state == VCORE_EXITING ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4099) vc->vcore_state == VCORE_PIGGYBACK))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4100) kvmppc_wait_for_exec(vc, vcpu, TASK_UNINTERRUPTIBLE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4101)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4102) if (vc->vcore_state == VCORE_PREEMPT && vc->runner == NULL)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4103) kvmppc_vcore_end_preempt(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4104)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4105) if (vcpu->arch.state == KVMPPC_VCPU_RUNNABLE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4106) kvmppc_remove_runnable(vc, vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4107) vcpu->stat.signal_exits++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4108) run->exit_reason = KVM_EXIT_INTR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4109) vcpu->arch.ret = -EINTR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4110) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4111)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4112) if (vc->n_runnable && vc->vcore_state == VCORE_INACTIVE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4113) /* Wake up some vcpu to run the core */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4114) i = -1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4115) v = next_runnable_thread(vc, &i);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4116) wake_up(&v->arch.cpu_run);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4117) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4118)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4119) trace_kvmppc_run_vcpu_exit(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4120) spin_unlock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4121) return vcpu->arch.ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4122) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4123)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4124) int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4125) unsigned long lpcr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4126) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4127) struct kvm_run *run = vcpu->run;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4128) int trap, r, pcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4129) int srcu_idx, lpid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4130) struct kvmppc_vcore *vc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4131) struct kvm *kvm = vcpu->kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4132) struct kvm_nested_guest *nested = vcpu->arch.nested;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4133)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4134) trace_kvmppc_run_vcpu_enter(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4135)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4136) run->exit_reason = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4137) vcpu->arch.ret = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4138) vcpu->arch.trap = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4139)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4140) vc = vcpu->arch.vcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4141) vcpu->arch.ceded = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4142) vcpu->arch.run_task = current;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4143) vcpu->arch.stolen_logged = vcore_stolen_time(vc, mftb());
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4144) vcpu->arch.state = KVMPPC_VCPU_RUNNABLE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4145) vcpu->arch.busy_preempt = TB_NIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4146) vcpu->arch.last_inst = KVM_INST_FETCH_FAILED;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4147) vc->runnable_threads[0] = vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4148) vc->n_runnable = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4149) vc->runner = vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4150)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4151) /* See if the MMU is ready to go */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4152) if (!kvm->arch.mmu_ready)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4153) kvmhv_setup_mmu(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4154)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4155) if (need_resched())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4156) cond_resched();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4157)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4158) kvmppc_update_vpas(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4159)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4160) init_vcore_to_run(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4161) vc->preempt_tb = TB_NIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4162)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4163) preempt_disable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4164) pcpu = smp_processor_id();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4165) vc->pcpu = pcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4166) kvmppc_prepare_radix_vcpu(vcpu, pcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4167)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4168) local_irq_disable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4169) hard_irq_disable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4170) if (signal_pending(current))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4171) goto sigpend;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4172) if (lazy_irq_pending() || need_resched() || !kvm->arch.mmu_ready)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4173) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4174)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4175) if (!nested) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4176) kvmppc_core_prepare_to_enter(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4177) if (vcpu->arch.doorbell_request) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4178) vc->dpdes = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4179) smp_wmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4180) vcpu->arch.doorbell_request = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4181) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4182) if (test_bit(BOOK3S_IRQPRIO_EXTERNAL,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4183) &vcpu->arch.pending_exceptions))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4184) lpcr |= LPCR_MER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4185) } else if (vcpu->arch.pending_exceptions ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4186) vcpu->arch.doorbell_request ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4187) xive_interrupt_pending(vcpu)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4188) vcpu->arch.ret = RESUME_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4189) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4190) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4191)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4192) kvmppc_clear_host_core(pcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4193)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4194) local_paca->kvm_hstate.tid = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4195) local_paca->kvm_hstate.napping = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4196) local_paca->kvm_hstate.kvm_split_mode = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4197) kvmppc_start_thread(vcpu, vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4198) kvmppc_create_dtl_entry(vcpu, vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4199) trace_kvm_guest_enter(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4200)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4201) vc->vcore_state = VCORE_RUNNING;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4202) trace_kvmppc_run_core(vc, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4203)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4204) if (cpu_has_feature(CPU_FTR_HVMODE)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4205) lpid = nested ? nested->shadow_lpid : kvm->arch.lpid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4206) mtspr(SPRN_LPID, lpid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4207) isync();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4208) kvmppc_check_need_tlb_flush(kvm, pcpu, nested);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4209) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4210)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4211) guest_enter_irqoff();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4212)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4213) srcu_idx = srcu_read_lock(&kvm->srcu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4214)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4215) this_cpu_disable_ftrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4216)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4217) /* Tell lockdep that we're about to enable interrupts */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4218) trace_hardirqs_on();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4219)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4220) trap = kvmhv_p9_guest_entry(vcpu, time_limit, lpcr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4221) vcpu->arch.trap = trap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4222)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4223) trace_hardirqs_off();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4224)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4225) this_cpu_enable_ftrace();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4226)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4227) srcu_read_unlock(&kvm->srcu, srcu_idx);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4228)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4229) if (cpu_has_feature(CPU_FTR_HVMODE)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4230) mtspr(SPRN_LPID, kvm->arch.host_lpid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4231) isync();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4232) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4233)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4234) set_irq_happened(trap);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4235)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4236) kvmppc_set_host_core(pcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4237)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4238) local_irq_enable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4239) guest_exit();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4240)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4241) cpumask_clear_cpu(pcpu, &kvm->arch.cpu_in_guest);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4242)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4243) preempt_enable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4244)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4245) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4246) * cancel pending decrementer exception if DEC is now positive, or if
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4247) * entering a nested guest in which case the decrementer is now owned
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4248) * by L2 and the L1 decrementer is provided in hdec_expires
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4249) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4250) if (kvmppc_core_pending_dec(vcpu) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4251) ((get_tb() < vcpu->arch.dec_expires) ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4252) (trap == BOOK3S_INTERRUPT_SYSCALL &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4253) kvmppc_get_gpr(vcpu, 3) == H_ENTER_NESTED)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4254) kvmppc_core_dequeue_dec(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4255)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4256) trace_kvm_guest_exit(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4257) r = RESUME_GUEST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4258) if (trap) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4259) if (!nested)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4260) r = kvmppc_handle_exit_hv(vcpu, current);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4261) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4262) r = kvmppc_handle_nested_exit(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4263) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4264) vcpu->arch.ret = r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4265)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4266) if (is_kvmppc_resume_guest(r) && vcpu->arch.ceded &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4267) !kvmppc_vcpu_woken(vcpu)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4268) kvmppc_set_timer(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4269) while (vcpu->arch.ceded && !kvmppc_vcpu_woken(vcpu)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4270) if (signal_pending(current)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4271) vcpu->stat.signal_exits++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4272) run->exit_reason = KVM_EXIT_INTR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4273) vcpu->arch.ret = -EINTR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4274) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4275) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4276) spin_lock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4277) kvmppc_vcore_blocked(vc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4278) spin_unlock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4279) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4280) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4281) vcpu->arch.ceded = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4282)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4283) vc->vcore_state = VCORE_INACTIVE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4284) trace_kvmppc_run_core(vc, 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4285)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4286) done:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4287) kvmppc_remove_runnable(vc, vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4288) trace_kvmppc_run_vcpu_exit(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4289)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4290) return vcpu->arch.ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4291)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4292) sigpend:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4293) vcpu->stat.signal_exits++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4294) run->exit_reason = KVM_EXIT_INTR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4295) vcpu->arch.ret = -EINTR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4296) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4297) local_irq_enable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4298) preempt_enable();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4299) goto done;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4300) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4301)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4302) static int kvmppc_vcpu_run_hv(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4303) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4304) struct kvm_run *run = vcpu->run;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4305) int r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4306) int srcu_idx;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4307) unsigned long ebb_regs[3] = {}; /* shut up GCC */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4308) unsigned long user_tar = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4309) unsigned int user_vrsave;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4310) struct kvm *kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4311)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4312) if (!vcpu->arch.sane) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4313) run->exit_reason = KVM_EXIT_INTERNAL_ERROR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4314) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4315) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4316)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4317) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4318) * Don't allow entry with a suspended transaction, because
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4319) * the guest entry/exit code will lose it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4320) * If the guest has TM enabled, save away their TM-related SPRs
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4321) * (they will get restored by the TM unavailable interrupt).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4322) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4323) #ifdef CONFIG_PPC_TRANSACTIONAL_MEM
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4324) if (cpu_has_feature(CPU_FTR_TM) && current->thread.regs &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4325) (current->thread.regs->msr & MSR_TM)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4326) if (MSR_TM_ACTIVE(current->thread.regs->msr)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4327) run->exit_reason = KVM_EXIT_FAIL_ENTRY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4328) run->fail_entry.hardware_entry_failure_reason = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4329) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4330) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4331) /* Enable TM so we can read the TM SPRs */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4332) mtmsr(mfmsr() | MSR_TM);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4333) current->thread.tm_tfhar = mfspr(SPRN_TFHAR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4334) current->thread.tm_tfiar = mfspr(SPRN_TFIAR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4335) current->thread.tm_texasr = mfspr(SPRN_TEXASR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4336) current->thread.regs->msr &= ~MSR_TM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4337) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4338) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4339)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4340) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4341) * Force online to 1 for the sake of old userspace which doesn't
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4342) * set it.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4343) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4344) if (!vcpu->arch.online) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4345) atomic_inc(&vcpu->arch.vcore->online_count);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4346) vcpu->arch.online = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4347) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4348)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4349) kvmppc_core_prepare_to_enter(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4350)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4351) /* No need to go into the guest when all we'll do is come back out */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4352) if (signal_pending(current)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4353) run->exit_reason = KVM_EXIT_INTR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4354) return -EINTR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4355) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4356)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4357) kvm = vcpu->kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4358) atomic_inc(&kvm->arch.vcpus_running);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4359) /* Order vcpus_running vs. mmu_ready, see kvmppc_alloc_reset_hpt */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4360) smp_mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4361)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4362) flush_all_to_thread(current);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4363)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4364) /* Save userspace EBB and other register values */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4365) if (cpu_has_feature(CPU_FTR_ARCH_207S)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4366) ebb_regs[0] = mfspr(SPRN_EBBHR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4367) ebb_regs[1] = mfspr(SPRN_EBBRR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4368) ebb_regs[2] = mfspr(SPRN_BESCR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4369) user_tar = mfspr(SPRN_TAR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4370) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4371) user_vrsave = mfspr(SPRN_VRSAVE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4372)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4373) vcpu->arch.waitp = &vcpu->arch.vcore->wait;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4374) vcpu->arch.pgdir = kvm->mm->pgd;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4375) vcpu->arch.state = KVMPPC_VCPU_BUSY_IN_HOST;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4376)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4377) do {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4378) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4379) * The early POWER9 chips that can't mix radix and HPT threads
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4380) * on the same core also need the workaround for the problem
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4381) * where the TLB would prefetch entries in the guest exit path
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4382) * for radix guests using the guest PIDR value and LPID 0.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4383) * The workaround is in the old path (kvmppc_run_vcpu())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4384) * but not the new path (kvmhv_run_single_vcpu()).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4385) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4386) if (kvm->arch.threads_indep && kvm_is_radix(kvm) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4387) !no_mixing_hpt_and_radix)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4388) r = kvmhv_run_single_vcpu(vcpu, ~(u64)0,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4389) vcpu->arch.vcore->lpcr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4390) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4391) r = kvmppc_run_vcpu(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4392)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4393) if (run->exit_reason == KVM_EXIT_PAPR_HCALL &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4394) !(vcpu->arch.shregs.msr & MSR_PR)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4395) trace_kvm_hcall_enter(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4396) r = kvmppc_pseries_do_hcall(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4397) trace_kvm_hcall_exit(vcpu, r);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4398) kvmppc_core_prepare_to_enter(vcpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4399) } else if (r == RESUME_PAGE_FAULT) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4400) srcu_idx = srcu_read_lock(&kvm->srcu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4401) r = kvmppc_book3s_hv_page_fault(vcpu,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4402) vcpu->arch.fault_dar, vcpu->arch.fault_dsisr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4403) srcu_read_unlock(&kvm->srcu, srcu_idx);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4404) } else if (r == RESUME_PASSTHROUGH) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4405) if (WARN_ON(xics_on_xive()))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4406) r = H_SUCCESS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4407) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4408) r = kvmppc_xics_rm_complete(vcpu, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4409) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4410) } while (is_kvmppc_resume_guest(r));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4411)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4412) /* Restore userspace EBB and other register values */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4413) if (cpu_has_feature(CPU_FTR_ARCH_207S)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4414) mtspr(SPRN_EBBHR, ebb_regs[0]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4415) mtspr(SPRN_EBBRR, ebb_regs[1]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4416) mtspr(SPRN_BESCR, ebb_regs[2]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4417) mtspr(SPRN_TAR, user_tar);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4418) mtspr(SPRN_FSCR, current->thread.fscr);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4419) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4420) mtspr(SPRN_VRSAVE, user_vrsave);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4421)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4422) vcpu->arch.state = KVMPPC_VCPU_NOTREADY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4423) atomic_dec(&kvm->arch.vcpus_running);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4424) return r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4425) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4426)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4427) static void kvmppc_add_seg_page_size(struct kvm_ppc_one_seg_page_size **sps,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4428) int shift, int sllp)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4429) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4430) (*sps)->page_shift = shift;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4431) (*sps)->slb_enc = sllp;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4432) (*sps)->enc[0].page_shift = shift;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4433) (*sps)->enc[0].pte_enc = kvmppc_pgsize_lp_encoding(shift, shift);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4434) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4435) * Add 16MB MPSS support (may get filtered out by userspace)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4436) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4437) if (shift != 24) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4438) int penc = kvmppc_pgsize_lp_encoding(shift, 24);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4439) if (penc != -1) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4440) (*sps)->enc[1].page_shift = 24;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4441) (*sps)->enc[1].pte_enc = penc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4442) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4443) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4444) (*sps)++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4445) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4446)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4447) static int kvm_vm_ioctl_get_smmu_info_hv(struct kvm *kvm,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4448) struct kvm_ppc_smmu_info *info)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4449) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4450) struct kvm_ppc_one_seg_page_size *sps;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4451)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4452) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4453) * POWER7, POWER8 and POWER9 all support 32 storage keys for data.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4454) * POWER7 doesn't support keys for instruction accesses,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4455) * POWER8 and POWER9 do.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4456) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4457) info->data_keys = 32;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4458) info->instr_keys = cpu_has_feature(CPU_FTR_ARCH_207S) ? 32 : 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4459)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4460) /* POWER7, 8 and 9 all have 1T segments and 32-entry SLB */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4461) info->flags = KVM_PPC_PAGE_SIZES_REAL | KVM_PPC_1T_SEGMENTS;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4462) info->slb_size = 32;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4463)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4464) /* We only support these sizes for now, and no muti-size segments */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4465) sps = &info->sps[0];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4466) kvmppc_add_seg_page_size(&sps, 12, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4467) kvmppc_add_seg_page_size(&sps, 16, SLB_VSID_L | SLB_VSID_LP_01);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4468) kvmppc_add_seg_page_size(&sps, 24, SLB_VSID_L);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4469)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4470) /* If running as a nested hypervisor, we don't support HPT guests */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4471) if (kvmhv_on_pseries())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4472) info->flags |= KVM_PPC_NO_HASH;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4473)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4474) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4475) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4476)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4477) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4478) * Get (and clear) the dirty memory log for a memory slot.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4479) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4480) static int kvm_vm_ioctl_get_dirty_log_hv(struct kvm *kvm,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4481) struct kvm_dirty_log *log)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4482) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4483) struct kvm_memslots *slots;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4484) struct kvm_memory_slot *memslot;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4485) int i, r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4486) unsigned long n;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4487) unsigned long *buf, *p;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4488) struct kvm_vcpu *vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4489)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4490) mutex_lock(&kvm->slots_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4491)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4492) r = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4493) if (log->slot >= KVM_USER_MEM_SLOTS)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4494) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4495)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4496) slots = kvm_memslots(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4497) memslot = id_to_memslot(slots, log->slot);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4498) r = -ENOENT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4499) if (!memslot || !memslot->dirty_bitmap)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4500) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4501)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4502) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4503) * Use second half of bitmap area because both HPT and radix
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4504) * accumulate bits in the first half.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4505) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4506) n = kvm_dirty_bitmap_bytes(memslot);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4507) buf = memslot->dirty_bitmap + n / sizeof(long);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4508) memset(buf, 0, n);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4509)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4510) if (kvm_is_radix(kvm))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4511) r = kvmppc_hv_get_dirty_log_radix(kvm, memslot, buf);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4512) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4513) r = kvmppc_hv_get_dirty_log_hpt(kvm, memslot, buf);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4514) if (r)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4515) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4516)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4517) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4518) * We accumulate dirty bits in the first half of the
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4519) * memslot's dirty_bitmap area, for when pages are paged
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4520) * out or modified by the host directly. Pick up these
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4521) * bits and add them to the map.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4522) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4523) p = memslot->dirty_bitmap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4524) for (i = 0; i < n / sizeof(long); ++i)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4525) buf[i] |= xchg(&p[i], 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4526)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4527) /* Harvest dirty bits from VPA and DTL updates */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4528) /* Note: we never modify the SLB shadow buffer areas */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4529) kvm_for_each_vcpu(i, vcpu, kvm) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4530) spin_lock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4531) kvmppc_harvest_vpa_dirty(&vcpu->arch.vpa, memslot, buf);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4532) kvmppc_harvest_vpa_dirty(&vcpu->arch.dtl, memslot, buf);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4533) spin_unlock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4534) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4535)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4536) r = -EFAULT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4537) if (copy_to_user(log->dirty_bitmap, buf, n))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4538) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4539)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4540) r = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4541) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4542) mutex_unlock(&kvm->slots_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4543) return r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4544) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4545)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4546) static void kvmppc_core_free_memslot_hv(struct kvm_memory_slot *slot)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4547) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4548) vfree(slot->arch.rmap);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4549) slot->arch.rmap = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4550) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4551)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4552) static int kvmppc_core_prepare_memory_region_hv(struct kvm *kvm,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4553) struct kvm_memory_slot *slot,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4554) const struct kvm_userspace_memory_region *mem,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4555) enum kvm_mr_change change)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4556) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4557) unsigned long npages = mem->memory_size >> PAGE_SHIFT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4558)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4559) if (change == KVM_MR_CREATE) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4560) unsigned long size = array_size(npages, sizeof(*slot->arch.rmap));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4561)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4562) if ((size >> PAGE_SHIFT) > totalram_pages())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4563) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4564)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4565) slot->arch.rmap = vzalloc(size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4566) if (!slot->arch.rmap)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4567) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4568) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4569)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4570) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4571) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4572)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4573) static void kvmppc_core_commit_memory_region_hv(struct kvm *kvm,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4574) const struct kvm_userspace_memory_region *mem,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4575) const struct kvm_memory_slot *old,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4576) const struct kvm_memory_slot *new,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4577) enum kvm_mr_change change)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4578) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4579) unsigned long npages = mem->memory_size >> PAGE_SHIFT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4580)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4581) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4582) * If we are making a new memslot, it might make
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4583) * some address that was previously cached as emulated
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4584) * MMIO be no longer emulated MMIO, so invalidate
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4585) * all the caches of emulated MMIO translations.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4586) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4587) if (npages)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4588) atomic64_inc(&kvm->arch.mmio_update);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4589)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4590) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4591) * For change == KVM_MR_MOVE or KVM_MR_DELETE, higher levels
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4592) * have already called kvm_arch_flush_shadow_memslot() to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4593) * flush shadow mappings. For KVM_MR_CREATE we have no
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4594) * previous mappings. So the only case to handle is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4595) * KVM_MR_FLAGS_ONLY when the KVM_MEM_LOG_DIRTY_PAGES bit
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4596) * has been changed.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4597) * For radix guests, we flush on setting KVM_MEM_LOG_DIRTY_PAGES
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4598) * to get rid of any THP PTEs in the partition-scoped page tables
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4599) * so we can track dirtiness at the page level; we flush when
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4600) * clearing KVM_MEM_LOG_DIRTY_PAGES so that we can go back to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4601) * using THP PTEs.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4602) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4603) if (change == KVM_MR_FLAGS_ONLY && kvm_is_radix(kvm) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4604) ((new->flags ^ old->flags) & KVM_MEM_LOG_DIRTY_PAGES))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4605) kvmppc_radix_flush_memslot(kvm, old);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4606) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4607) * If UV hasn't yet called H_SVM_INIT_START, don't register memslots.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4608) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4609) if (!kvm->arch.secure_guest)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4610) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4611)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4612) switch (change) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4613) case KVM_MR_CREATE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4614) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4615) * @TODO kvmppc_uvmem_memslot_create() can fail and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4616) * return error. Fix this.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4617) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4618) kvmppc_uvmem_memslot_create(kvm, new);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4619) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4620) case KVM_MR_DELETE:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4621) kvmppc_uvmem_memslot_delete(kvm, old);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4622) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4623) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4624) /* TODO: Handle KVM_MR_MOVE */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4625) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4626) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4627) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4628)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4629) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4630) * Update LPCR values in kvm->arch and in vcores.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4631) * Caller must hold kvm->arch.mmu_setup_lock (for mutual exclusion
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4632) * of kvm->arch.lpcr update).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4633) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4634) void kvmppc_update_lpcr(struct kvm *kvm, unsigned long lpcr, unsigned long mask)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4635) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4636) long int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4637) u32 cores_done = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4638)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4639) if ((kvm->arch.lpcr & mask) == lpcr)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4640) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4641)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4642) kvm->arch.lpcr = (kvm->arch.lpcr & ~mask) | lpcr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4643)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4644) for (i = 0; i < KVM_MAX_VCORES; ++i) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4645) struct kvmppc_vcore *vc = kvm->arch.vcores[i];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4646) if (!vc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4647) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4648) spin_lock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4649) vc->lpcr = (vc->lpcr & ~mask) | lpcr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4650) spin_unlock(&vc->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4651) if (++cores_done >= kvm->arch.online_vcores)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4652) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4653) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4654) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4655)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4656) void kvmppc_setup_partition_table(struct kvm *kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4657) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4658) unsigned long dw0, dw1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4659)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4660) if (!kvm_is_radix(kvm)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4661) /* PS field - page size for VRMA */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4662) dw0 = ((kvm->arch.vrma_slb_v & SLB_VSID_L) >> 1) |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4663) ((kvm->arch.vrma_slb_v & SLB_VSID_LP) << 1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4664) /* HTABSIZE and HTABORG fields */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4665) dw0 |= kvm->arch.sdr1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4666)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4667) /* Second dword as set by userspace */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4668) dw1 = kvm->arch.process_table;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4669) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4670) dw0 = PATB_HR | radix__get_tree_size() |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4671) __pa(kvm->arch.pgtable) | RADIX_PGD_INDEX_SIZE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4672) dw1 = PATB_GR | kvm->arch.process_table;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4673) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4674) kvmhv_set_ptbl_entry(kvm->arch.lpid, dw0, dw1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4675) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4676)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4677) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4678) * Set up HPT (hashed page table) and RMA (real-mode area).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4679) * Must be called with kvm->arch.mmu_setup_lock held.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4680) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4681) static int kvmppc_hv_setup_htab_rma(struct kvm_vcpu *vcpu)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4682) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4683) int err = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4684) struct kvm *kvm = vcpu->kvm;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4685) unsigned long hva;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4686) struct kvm_memory_slot *memslot;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4687) struct vm_area_struct *vma;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4688) unsigned long lpcr = 0, senc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4689) unsigned long psize, porder;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4690) int srcu_idx;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4691)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4692) /* Allocate hashed page table (if not done already) and reset it */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4693) if (!kvm->arch.hpt.virt) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4694) int order = KVM_DEFAULT_HPT_ORDER;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4695) struct kvm_hpt_info info;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4696)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4697) err = kvmppc_allocate_hpt(&info, order);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4698) /* If we get here, it means userspace didn't specify a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4699) * size explicitly. So, try successively smaller
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4700) * sizes if the default failed. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4701) while ((err == -ENOMEM) && --order >= PPC_MIN_HPT_ORDER)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4702) err = kvmppc_allocate_hpt(&info, order);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4703)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4704) if (err < 0) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4705) pr_err("KVM: Couldn't alloc HPT\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4706) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4707) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4708)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4709) kvmppc_set_hpt(kvm, &info);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4710) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4711)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4712) /* Look up the memslot for guest physical address 0 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4713) srcu_idx = srcu_read_lock(&kvm->srcu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4714) memslot = gfn_to_memslot(kvm, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4715)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4716) /* We must have some memory at 0 by now */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4717) err = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4718) if (!memslot || (memslot->flags & KVM_MEMSLOT_INVALID))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4719) goto out_srcu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4720)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4721) /* Look up the VMA for the start of this memory slot */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4722) hva = memslot->userspace_addr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4723) mmap_read_lock(kvm->mm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4724) vma = find_vma(kvm->mm, hva);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4725) if (!vma || vma->vm_start > hva || (vma->vm_flags & VM_IO))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4726) goto up_out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4727)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4728) psize = vma_kernel_pagesize(vma);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4729)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4730) mmap_read_unlock(kvm->mm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4731)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4732) /* We can handle 4k, 64k or 16M pages in the VRMA */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4733) if (psize >= 0x1000000)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4734) psize = 0x1000000;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4735) else if (psize >= 0x10000)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4736) psize = 0x10000;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4737) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4738) psize = 0x1000;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4739) porder = __ilog2(psize);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4740)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4741) senc = slb_pgsize_encoding(psize);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4742) kvm->arch.vrma_slb_v = senc | SLB_VSID_B_1T |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4743) (VRMA_VSID << SLB_VSID_SHIFT_1T);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4744) /* Create HPTEs in the hash page table for the VRMA */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4745) kvmppc_map_vrma(vcpu, memslot, porder);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4746)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4747) /* Update VRMASD field in the LPCR */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4748) if (!cpu_has_feature(CPU_FTR_ARCH_300)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4749) /* the -4 is to account for senc values starting at 0x10 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4750) lpcr = senc << (LPCR_VRMASD_SH - 4);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4751) kvmppc_update_lpcr(kvm, lpcr, LPCR_VRMASD);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4752) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4753)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4754) /* Order updates to kvm->arch.lpcr etc. vs. mmu_ready */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4755) smp_wmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4756) err = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4757) out_srcu:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4758) srcu_read_unlock(&kvm->srcu, srcu_idx);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4759) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4760) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4761)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4762) up_out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4763) mmap_read_unlock(kvm->mm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4764) goto out_srcu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4765) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4766)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4767) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4768) * Must be called with kvm->arch.mmu_setup_lock held and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4769) * mmu_ready = 0 and no vcpus running.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4770) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4771) int kvmppc_switch_mmu_to_hpt(struct kvm *kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4772) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4773) if (nesting_enabled(kvm))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4774) kvmhv_release_all_nested(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4775) kvmppc_rmap_reset(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4776) kvm->arch.process_table = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4777) /* Mutual exclusion with kvm_unmap_hva_range etc. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4778) spin_lock(&kvm->mmu_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4779) kvm->arch.radix = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4780) spin_unlock(&kvm->mmu_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4781) kvmppc_free_radix(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4782) kvmppc_update_lpcr(kvm, LPCR_VPM1,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4783) LPCR_VPM1 | LPCR_UPRT | LPCR_GTSE | LPCR_HR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4784) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4785) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4786)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4787) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4788) * Must be called with kvm->arch.mmu_setup_lock held and
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4789) * mmu_ready = 0 and no vcpus running.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4790) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4791) int kvmppc_switch_mmu_to_radix(struct kvm *kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4792) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4793) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4794)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4795) err = kvmppc_init_vm_radix(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4796) if (err)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4797) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4798) kvmppc_rmap_reset(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4799) /* Mutual exclusion with kvm_unmap_hva_range etc. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4800) spin_lock(&kvm->mmu_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4801) kvm->arch.radix = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4802) spin_unlock(&kvm->mmu_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4803) kvmppc_free_hpt(&kvm->arch.hpt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4804) kvmppc_update_lpcr(kvm, LPCR_UPRT | LPCR_GTSE | LPCR_HR,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4805) LPCR_VPM1 | LPCR_UPRT | LPCR_GTSE | LPCR_HR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4806) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4807) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4808)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4809) #ifdef CONFIG_KVM_XICS
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4810) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4811) * Allocate a per-core structure for managing state about which cores are
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4812) * running in the host versus the guest and for exchanging data between
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4813) * real mode KVM and CPU running in the host.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4814) * This is only done for the first VM.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4815) * The allocated structure stays even if all VMs have stopped.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4816) * It is only freed when the kvm-hv module is unloaded.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4817) * It's OK for this routine to fail, we just don't support host
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4818) * core operations like redirecting H_IPI wakeups.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4819) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4820) void kvmppc_alloc_host_rm_ops(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4821) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4822) struct kvmppc_host_rm_ops *ops;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4823) unsigned long l_ops;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4824) int cpu, core;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4825) int size;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4826)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4827) /* Not the first time here ? */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4828) if (kvmppc_host_rm_ops_hv != NULL)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4829) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4830)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4831) ops = kzalloc(sizeof(struct kvmppc_host_rm_ops), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4832) if (!ops)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4833) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4834)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4835) size = cpu_nr_cores() * sizeof(struct kvmppc_host_rm_core);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4836) ops->rm_core = kzalloc(size, GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4837)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4838) if (!ops->rm_core) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4839) kfree(ops);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4840) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4841) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4842)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4843) cpus_read_lock();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4844)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4845) for (cpu = 0; cpu < nr_cpu_ids; cpu += threads_per_core) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4846) if (!cpu_online(cpu))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4847) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4848)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4849) core = cpu >> threads_shift;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4850) ops->rm_core[core].rm_state.in_host = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4851) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4852)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4853) ops->vcpu_kick = kvmppc_fast_vcpu_kick_hv;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4854)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4855) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4856) * Make the contents of the kvmppc_host_rm_ops structure visible
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4857) * to other CPUs before we assign it to the global variable.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4858) * Do an atomic assignment (no locks used here), but if someone
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4859) * beats us to it, just free our copy and return.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4860) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4861) smp_wmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4862) l_ops = (unsigned long) ops;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4863)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4864) if (cmpxchg64((unsigned long *)&kvmppc_host_rm_ops_hv, 0, l_ops)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4865) cpus_read_unlock();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4866) kfree(ops->rm_core);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4867) kfree(ops);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4868) return;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4869) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4870)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4871) cpuhp_setup_state_nocalls_cpuslocked(CPUHP_KVM_PPC_BOOK3S_PREPARE,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4872) "ppc/kvm_book3s:prepare",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4873) kvmppc_set_host_core,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4874) kvmppc_clear_host_core);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4875) cpus_read_unlock();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4876) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4877)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4878) void kvmppc_free_host_rm_ops(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4879) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4880) if (kvmppc_host_rm_ops_hv) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4881) cpuhp_remove_state_nocalls(CPUHP_KVM_PPC_BOOK3S_PREPARE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4882) kfree(kvmppc_host_rm_ops_hv->rm_core);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4883) kfree(kvmppc_host_rm_ops_hv);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4884) kvmppc_host_rm_ops_hv = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4885) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4886) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4887) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4888)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4889) static int kvmppc_core_init_vm_hv(struct kvm *kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4890) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4891) unsigned long lpcr, lpid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4892) char buf[32];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4893) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4894)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4895) mutex_init(&kvm->arch.uvmem_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4896) INIT_LIST_HEAD(&kvm->arch.uvmem_pfns);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4897) mutex_init(&kvm->arch.mmu_setup_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4898)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4899) /* Allocate the guest's logical partition ID */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4900)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4901) lpid = kvmppc_alloc_lpid();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4902) if ((long)lpid < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4903) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4904) kvm->arch.lpid = lpid;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4905)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4906) kvmppc_alloc_host_rm_ops();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4907)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4908) kvmhv_vm_nested_init(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4909)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4910) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4911) * Since we don't flush the TLB when tearing down a VM,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4912) * and this lpid might have previously been used,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4913) * make sure we flush on each core before running the new VM.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4914) * On POWER9, the tlbie in mmu_partition_table_set_entry()
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4915) * does this flush for us.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4916) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4917) if (!cpu_has_feature(CPU_FTR_ARCH_300))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4918) cpumask_setall(&kvm->arch.need_tlb_flush);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4919)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4920) /* Start out with the default set of hcalls enabled */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4921) memcpy(kvm->arch.enabled_hcalls, default_enabled_hcalls,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4922) sizeof(kvm->arch.enabled_hcalls));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4923)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4924) if (!cpu_has_feature(CPU_FTR_ARCH_300))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4925) kvm->arch.host_sdr1 = mfspr(SPRN_SDR1);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4926)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4927) /* Init LPCR for virtual RMA mode */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4928) if (cpu_has_feature(CPU_FTR_HVMODE)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4929) kvm->arch.host_lpid = mfspr(SPRN_LPID);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4930) kvm->arch.host_lpcr = lpcr = mfspr(SPRN_LPCR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4931) lpcr &= LPCR_PECE | LPCR_LPES;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4932) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4933) lpcr = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4934) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4935) lpcr |= (4UL << LPCR_DPFD_SH) | LPCR_HDICE |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4936) LPCR_VPM0 | LPCR_VPM1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4937) kvm->arch.vrma_slb_v = SLB_VSID_B_1T |
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4938) (VRMA_VSID << SLB_VSID_SHIFT_1T);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4939) /* On POWER8 turn on online bit to enable PURR/SPURR */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4940) if (cpu_has_feature(CPU_FTR_ARCH_207S))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4941) lpcr |= LPCR_ONL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4942) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4943) * On POWER9, VPM0 bit is reserved (VPM0=1 behaviour is assumed)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4944) * Set HVICE bit to enable hypervisor virtualization interrupts.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4945) * Set HEIC to prevent OS interrupts to go to hypervisor (should
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4946) * be unnecessary but better safe than sorry in case we re-enable
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4947) * EE in HV mode with this LPCR still set)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4948) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4949) if (cpu_has_feature(CPU_FTR_ARCH_300)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4950) lpcr &= ~LPCR_VPM0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4951) lpcr |= LPCR_HVICE | LPCR_HEIC;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4952)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4953) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4954) * If xive is enabled, we route 0x500 interrupts directly
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4955) * to the guest.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4956) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4957) if (xics_on_xive())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4958) lpcr |= LPCR_LPES;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4959) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4960)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4961) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4962) * If the host uses radix, the guest starts out as radix.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4963) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4964) if (radix_enabled()) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4965) kvm->arch.radix = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4966) kvm->arch.mmu_ready = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4967) lpcr &= ~LPCR_VPM1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4968) lpcr |= LPCR_UPRT | LPCR_GTSE | LPCR_HR;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4969) ret = kvmppc_init_vm_radix(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4970) if (ret) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4971) kvmppc_free_lpid(kvm->arch.lpid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4972) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4973) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4974) kvmppc_setup_partition_table(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4975) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4976)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4977) kvm->arch.lpcr = lpcr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4978)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4979) /* Initialization for future HPT resizes */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4980) kvm->arch.resize_hpt = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4981)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4982) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4983) * Work out how many sets the TLB has, for the use of
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4984) * the TLB invalidation loop in book3s_hv_rmhandlers.S.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4985) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4986) if (radix_enabled())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4987) kvm->arch.tlb_sets = POWER9_TLB_SETS_RADIX; /* 128 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4988) else if (cpu_has_feature(CPU_FTR_ARCH_300))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4989) kvm->arch.tlb_sets = POWER9_TLB_SETS_HASH; /* 256 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4990) else if (cpu_has_feature(CPU_FTR_ARCH_207S))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4991) kvm->arch.tlb_sets = POWER8_TLB_SETS; /* 512 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4992) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4993) kvm->arch.tlb_sets = POWER7_TLB_SETS; /* 128 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4994)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4995) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4996) * Track that we now have a HV mode VM active. This blocks secondary
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4997) * CPU threads from coming online.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4998) * On POWER9, we only need to do this if the "indep_threads_mode"
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 4999) * module parameter has been set to N.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5000) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5001) if (cpu_has_feature(CPU_FTR_ARCH_300)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5002) if (!indep_threads_mode && !cpu_has_feature(CPU_FTR_HVMODE)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5003) pr_warn("KVM: Ignoring indep_threads_mode=N in nested hypervisor\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5004) kvm->arch.threads_indep = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5005) } else {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5006) kvm->arch.threads_indep = indep_threads_mode;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5007) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5008) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5009) if (!kvm->arch.threads_indep)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5010) kvm_hv_vm_activated();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5011)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5012) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5013) * Initialize smt_mode depending on processor.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5014) * POWER8 and earlier have to use "strict" threading, where
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5015) * all vCPUs in a vcore have to run on the same (sub)core,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5016) * whereas on POWER9 the threads can each run a different
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5017) * guest.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5018) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5019) if (!cpu_has_feature(CPU_FTR_ARCH_300))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5020) kvm->arch.smt_mode = threads_per_subcore;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5021) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5022) kvm->arch.smt_mode = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5023) kvm->arch.emul_smt_mode = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5024)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5025) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5026) * Create a debugfs directory for the VM
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5027) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5028) snprintf(buf, sizeof(buf), "vm%d", current->pid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5029) kvm->arch.debugfs_dir = debugfs_create_dir(buf, kvm_debugfs_dir);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5030) kvmppc_mmu_debugfs_init(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5031) if (radix_enabled())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5032) kvmhv_radix_debugfs_init(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5033)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5034) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5035) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5036)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5037) static void kvmppc_free_vcores(struct kvm *kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5038) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5039) long int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5040)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5041) for (i = 0; i < KVM_MAX_VCORES; ++i)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5042) kfree(kvm->arch.vcores[i]);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5043) kvm->arch.online_vcores = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5044) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5045)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5046) static void kvmppc_core_destroy_vm_hv(struct kvm *kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5047) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5048) debugfs_remove_recursive(kvm->arch.debugfs_dir);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5049)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5050) if (!kvm->arch.threads_indep)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5051) kvm_hv_vm_deactivated();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5052)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5053) kvmppc_free_vcores(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5054)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5055)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5056) if (kvm_is_radix(kvm))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5057) kvmppc_free_radix(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5058) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5059) kvmppc_free_hpt(&kvm->arch.hpt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5060)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5061) /* Perform global invalidation and return lpid to the pool */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5062) if (cpu_has_feature(CPU_FTR_ARCH_300)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5063) if (nesting_enabled(kvm))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5064) kvmhv_release_all_nested(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5065) kvm->arch.process_table = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5066) if (kvm->arch.secure_guest)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5067) uv_svm_terminate(kvm->arch.lpid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5068) kvmhv_set_ptbl_entry(kvm->arch.lpid, 0, 0);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5069) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5070)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5071) kvmppc_free_lpid(kvm->arch.lpid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5072)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5073) kvmppc_free_pimap(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5074) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5075)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5076) /* We don't need to emulate any privileged instructions or dcbz */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5077) static int kvmppc_core_emulate_op_hv(struct kvm_vcpu *vcpu,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5078) unsigned int inst, int *advance)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5079) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5080) return EMULATE_FAIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5081) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5082)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5083) static int kvmppc_core_emulate_mtspr_hv(struct kvm_vcpu *vcpu, int sprn,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5084) ulong spr_val)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5085) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5086) return EMULATE_FAIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5087) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5088)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5089) static int kvmppc_core_emulate_mfspr_hv(struct kvm_vcpu *vcpu, int sprn,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5090) ulong *spr_val)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5091) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5092) return EMULATE_FAIL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5093) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5094)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5095) static int kvmppc_core_check_processor_compat_hv(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5096) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5097) if (cpu_has_feature(CPU_FTR_HVMODE) &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5098) cpu_has_feature(CPU_FTR_ARCH_206))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5099) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5100)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5101) /* POWER9 in radix mode is capable of being a nested hypervisor. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5102) if (cpu_has_feature(CPU_FTR_ARCH_300) && radix_enabled())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5103) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5104)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5105) return -EIO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5106) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5107)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5108) #ifdef CONFIG_KVM_XICS
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5109)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5110) void kvmppc_free_pimap(struct kvm *kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5111) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5112) kfree(kvm->arch.pimap);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5113) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5114)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5115) static struct kvmppc_passthru_irqmap *kvmppc_alloc_pimap(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5116) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5117) return kzalloc(sizeof(struct kvmppc_passthru_irqmap), GFP_KERNEL);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5118) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5119)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5120) static int kvmppc_set_passthru_irq(struct kvm *kvm, int host_irq, int guest_gsi)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5121) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5122) struct irq_desc *desc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5123) struct kvmppc_irq_map *irq_map;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5124) struct kvmppc_passthru_irqmap *pimap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5125) struct irq_chip *chip;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5126) int i, rc = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5127)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5128) if (!kvm_irq_bypass)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5129) return 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5130)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5131) desc = irq_to_desc(host_irq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5132) if (!desc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5133) return -EIO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5134)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5135) mutex_lock(&kvm->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5136)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5137) pimap = kvm->arch.pimap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5138) if (pimap == NULL) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5139) /* First call, allocate structure to hold IRQ map */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5140) pimap = kvmppc_alloc_pimap();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5141) if (pimap == NULL) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5142) mutex_unlock(&kvm->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5143) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5144) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5145) kvm->arch.pimap = pimap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5146) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5147)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5148) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5149) * For now, we only support interrupts for which the EOI operation
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5150) * is an OPAL call followed by a write to XIRR, since that's
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5151) * what our real-mode EOI code does, or a XIVE interrupt
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5152) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5153) chip = irq_data_get_irq_chip(&desc->irq_data);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5154) if (!chip || !(is_pnv_opal_msi(chip) || is_xive_irq(chip))) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5155) pr_warn("kvmppc_set_passthru_irq_hv: Could not assign IRQ map for (%d,%d)\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5156) host_irq, guest_gsi);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5157) mutex_unlock(&kvm->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5158) return -ENOENT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5159) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5160)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5161) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5162) * See if we already have an entry for this guest IRQ number.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5163) * If it's mapped to a hardware IRQ number, that's an error,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5164) * otherwise re-use this entry.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5165) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5166) for (i = 0; i < pimap->n_mapped; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5167) if (guest_gsi == pimap->mapped[i].v_hwirq) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5168) if (pimap->mapped[i].r_hwirq) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5169) mutex_unlock(&kvm->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5170) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5171) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5172) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5173) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5174) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5175)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5176) if (i == KVMPPC_PIRQ_MAPPED) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5177) mutex_unlock(&kvm->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5178) return -EAGAIN; /* table is full */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5179) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5180)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5181) irq_map = &pimap->mapped[i];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5182)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5183) irq_map->v_hwirq = guest_gsi;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5184) irq_map->desc = desc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5185)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5186) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5187) * Order the above two stores before the next to serialize with
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5188) * the KVM real mode handler.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5189) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5190) smp_wmb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5191) irq_map->r_hwirq = desc->irq_data.hwirq;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5192)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5193) if (i == pimap->n_mapped)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5194) pimap->n_mapped++;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5195)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5196) if (xics_on_xive())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5197) rc = kvmppc_xive_set_mapped(kvm, guest_gsi, desc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5198) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5199) kvmppc_xics_set_mapped(kvm, guest_gsi, desc->irq_data.hwirq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5200) if (rc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5201) irq_map->r_hwirq = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5202)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5203) mutex_unlock(&kvm->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5204)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5205) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5206) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5207)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5208) static int kvmppc_clr_passthru_irq(struct kvm *kvm, int host_irq, int guest_gsi)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5209) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5210) struct irq_desc *desc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5211) struct kvmppc_passthru_irqmap *pimap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5212) int i, rc = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5213)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5214) if (!kvm_irq_bypass)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5215) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5216)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5217) desc = irq_to_desc(host_irq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5218) if (!desc)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5219) return -EIO;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5220)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5221) mutex_lock(&kvm->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5222) if (!kvm->arch.pimap)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5223) goto unlock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5224)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5225) pimap = kvm->arch.pimap;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5226)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5227) for (i = 0; i < pimap->n_mapped; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5228) if (guest_gsi == pimap->mapped[i].v_hwirq)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5229) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5230) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5231)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5232) if (i == pimap->n_mapped) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5233) mutex_unlock(&kvm->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5234) return -ENODEV;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5235) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5236)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5237) if (xics_on_xive())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5238) rc = kvmppc_xive_clr_mapped(kvm, guest_gsi, pimap->mapped[i].desc);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5239) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5240) kvmppc_xics_clr_mapped(kvm, guest_gsi, pimap->mapped[i].r_hwirq);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5241)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5242) /* invalidate the entry (what do do on error from the above ?) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5243) pimap->mapped[i].r_hwirq = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5244)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5245) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5246) * We don't free this structure even when the count goes to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5247) * zero. The structure is freed when we destroy the VM.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5248) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5249) unlock:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5250) mutex_unlock(&kvm->lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5251) return rc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5252) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5253)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5254) static int kvmppc_irq_bypass_add_producer_hv(struct irq_bypass_consumer *cons,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5255) struct irq_bypass_producer *prod)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5256) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5257) int ret = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5258) struct kvm_kernel_irqfd *irqfd =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5259) container_of(cons, struct kvm_kernel_irqfd, consumer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5260)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5261) irqfd->producer = prod;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5262)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5263) ret = kvmppc_set_passthru_irq(irqfd->kvm, prod->irq, irqfd->gsi);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5264) if (ret)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5265) pr_info("kvmppc_set_passthru_irq (irq %d, gsi %d) fails: %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5266) prod->irq, irqfd->gsi, ret);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5267)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5268) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5269) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5270)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5271) static void kvmppc_irq_bypass_del_producer_hv(struct irq_bypass_consumer *cons,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5272) struct irq_bypass_producer *prod)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5273) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5274) int ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5275) struct kvm_kernel_irqfd *irqfd =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5276) container_of(cons, struct kvm_kernel_irqfd, consumer);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5277)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5278) irqfd->producer = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5279)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5280) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5281) * When producer of consumer is unregistered, we change back to
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5282) * default external interrupt handling mode - KVM real mode
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5283) * will switch back to host.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5284) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5285) ret = kvmppc_clr_passthru_irq(irqfd->kvm, prod->irq, irqfd->gsi);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5286) if (ret)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5287) pr_warn("kvmppc_clr_passthru_irq (irq %d, gsi %d) fails: %d\n",
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5288) prod->irq, irqfd->gsi, ret);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5289) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5290) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5291)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5292) static long kvm_arch_vm_ioctl_hv(struct file *filp,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5293) unsigned int ioctl, unsigned long arg)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5294) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5295) struct kvm *kvm __maybe_unused = filp->private_data;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5296) void __user *argp = (void __user *)arg;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5297) long r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5298)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5299) switch (ioctl) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5300)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5301) case KVM_PPC_ALLOCATE_HTAB: {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5302) u32 htab_order;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5303)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5304) /* If we're a nested hypervisor, we currently only support radix */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5305) if (kvmhv_on_pseries()) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5306) r = -EOPNOTSUPP;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5307) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5308) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5309)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5310) r = -EFAULT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5311) if (get_user(htab_order, (u32 __user *)argp))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5312) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5313) r = kvmppc_alloc_reset_hpt(kvm, htab_order);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5314) if (r)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5315) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5316) r = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5317) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5318) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5319)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5320) case KVM_PPC_GET_HTAB_FD: {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5321) struct kvm_get_htab_fd ghf;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5322)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5323) r = -EFAULT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5324) if (copy_from_user(&ghf, argp, sizeof(ghf)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5325) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5326) r = kvm_vm_ioctl_get_htab_fd(kvm, &ghf);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5327) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5328) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5329)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5330) case KVM_PPC_RESIZE_HPT_PREPARE: {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5331) struct kvm_ppc_resize_hpt rhpt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5332)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5333) r = -EFAULT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5334) if (copy_from_user(&rhpt, argp, sizeof(rhpt)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5335) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5336)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5337) r = kvm_vm_ioctl_resize_hpt_prepare(kvm, &rhpt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5338) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5339) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5340)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5341) case KVM_PPC_RESIZE_HPT_COMMIT: {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5342) struct kvm_ppc_resize_hpt rhpt;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5343)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5344) r = -EFAULT;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5345) if (copy_from_user(&rhpt, argp, sizeof(rhpt)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5346) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5347)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5348) r = kvm_vm_ioctl_resize_hpt_commit(kvm, &rhpt);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5349) break;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5350) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5351)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5352) default:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5353) r = -ENOTTY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5354) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5355)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5356) return r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5357) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5358)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5359) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5360) * List of hcall numbers to enable by default.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5361) * For compatibility with old userspace, we enable by default
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5362) * all hcalls that were implemented before the hcall-enabling
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5363) * facility was added. Note this list should not include H_RTAS.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5364) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5365) static unsigned int default_hcall_list[] = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5366) H_REMOVE,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5367) H_ENTER,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5368) H_READ,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5369) H_PROTECT,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5370) H_BULK_REMOVE,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5371) H_GET_TCE,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5372) H_PUT_TCE,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5373) H_SET_DABR,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5374) H_SET_XDABR,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5375) H_CEDE,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5376) H_PROD,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5377) H_CONFER,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5378) H_REGISTER_VPA,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5379) #ifdef CONFIG_KVM_XICS
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5380) H_EOI,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5381) H_CPPR,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5382) H_IPI,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5383) H_IPOLL,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5384) H_XIRR,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5385) H_XIRR_X,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5386) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5387) 0
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5388) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5389)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5390) static void init_default_hcalls(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5391) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5392) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5393) unsigned int hcall;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5394)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5395) for (i = 0; default_hcall_list[i]; ++i) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5396) hcall = default_hcall_list[i];
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5397) WARN_ON(!kvmppc_hcall_impl_hv(hcall));
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5398) __set_bit(hcall / 4, default_enabled_hcalls);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5399) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5400) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5401)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5402) static int kvmhv_configure_mmu(struct kvm *kvm, struct kvm_ppc_mmuv3_cfg *cfg)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5403) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5404) unsigned long lpcr;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5405) int radix;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5406) int err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5407)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5408) /* If not on a POWER9, reject it */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5409) if (!cpu_has_feature(CPU_FTR_ARCH_300))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5410) return -ENODEV;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5411)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5412) /* If any unknown flags set, reject it */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5413) if (cfg->flags & ~(KVM_PPC_MMUV3_RADIX | KVM_PPC_MMUV3_GTSE))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5414) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5415)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5416) /* GR (guest radix) bit in process_table field must match */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5417) radix = !!(cfg->flags & KVM_PPC_MMUV3_RADIX);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5418) if (!!(cfg->process_table & PATB_GR) != radix)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5419) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5420)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5421) /* Process table size field must be reasonable, i.e. <= 24 */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5422) if ((cfg->process_table & PRTS_MASK) > 24)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5423) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5424)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5425) /* We can change a guest to/from radix now, if the host is radix */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5426) if (radix && !radix_enabled())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5427) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5428)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5429) /* If we're a nested hypervisor, we currently only support radix */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5430) if (kvmhv_on_pseries() && !radix)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5431) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5432)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5433) mutex_lock(&kvm->arch.mmu_setup_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5434) if (radix != kvm_is_radix(kvm)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5435) if (kvm->arch.mmu_ready) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5436) kvm->arch.mmu_ready = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5437) /* order mmu_ready vs. vcpus_running */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5438) smp_mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5439) if (atomic_read(&kvm->arch.vcpus_running)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5440) kvm->arch.mmu_ready = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5441) err = -EBUSY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5442) goto out_unlock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5443) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5444) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5445) if (radix)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5446) err = kvmppc_switch_mmu_to_radix(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5447) else
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5448) err = kvmppc_switch_mmu_to_hpt(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5449) if (err)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5450) goto out_unlock;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5451) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5452)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5453) kvm->arch.process_table = cfg->process_table;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5454) kvmppc_setup_partition_table(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5455)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5456) lpcr = (cfg->flags & KVM_PPC_MMUV3_GTSE) ? LPCR_GTSE : 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5457) kvmppc_update_lpcr(kvm, lpcr, LPCR_GTSE);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5458) err = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5459)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5460) out_unlock:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5461) mutex_unlock(&kvm->arch.mmu_setup_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5462) return err;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5463) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5464)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5465) static int kvmhv_enable_nested(struct kvm *kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5466) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5467) if (!nested)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5468) return -EPERM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5469) if (!cpu_has_feature(CPU_FTR_ARCH_300) || no_mixing_hpt_and_radix)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5470) return -ENODEV;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5471)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5472) /* kvm == NULL means the caller is testing if the capability exists */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5473) if (kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5474) kvm->arch.nested_enable = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5475) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5476) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5477)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5478) static int kvmhv_load_from_eaddr(struct kvm_vcpu *vcpu, ulong *eaddr, void *ptr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5479) int size)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5480) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5481) int rc = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5482)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5483) if (kvmhv_vcpu_is_radix(vcpu)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5484) rc = kvmhv_copy_from_guest_radix(vcpu, *eaddr, ptr, size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5485)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5486) if (rc > 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5487) rc = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5488) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5489)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5490) /* For now quadrants are the only way to access nested guest memory */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5491) if (rc && vcpu->arch.nested)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5492) rc = -EAGAIN;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5493)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5494) return rc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5495) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5496)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5497) static int kvmhv_store_to_eaddr(struct kvm_vcpu *vcpu, ulong *eaddr, void *ptr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5498) int size)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5499) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5500) int rc = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5501)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5502) if (kvmhv_vcpu_is_radix(vcpu)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5503) rc = kvmhv_copy_to_guest_radix(vcpu, *eaddr, ptr, size);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5504)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5505) if (rc > 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5506) rc = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5507) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5508)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5509) /* For now quadrants are the only way to access nested guest memory */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5510) if (rc && vcpu->arch.nested)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5511) rc = -EAGAIN;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5512)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5513) return rc;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5514) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5515)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5516) static void unpin_vpa_reset(struct kvm *kvm, struct kvmppc_vpa *vpa)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5517) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5518) unpin_vpa(kvm, vpa);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5519) vpa->gpa = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5520) vpa->pinned_addr = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5521) vpa->dirty = false;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5522) vpa->update_pending = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5523) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5524)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5525) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5526) * Enable a guest to become a secure VM, or test whether
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5527) * that could be enabled.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5528) * Called when the KVM_CAP_PPC_SECURE_GUEST capability is
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5529) * tested (kvm == NULL) or enabled (kvm != NULL).
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5530) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5531) static int kvmhv_enable_svm(struct kvm *kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5532) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5533) if (!kvmppc_uvmem_available())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5534) return -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5535) if (kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5536) kvm->arch.svm_enabled = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5537) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5538) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5539)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5540) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5541) * IOCTL handler to turn off secure mode of guest
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5542) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5543) * - Release all device pages
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5544) * - Issue ucall to terminate the guest on the UV side
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5545) * - Unpin the VPA pages.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5546) * - Reinit the partition scoped page tables
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5547) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5548) static int kvmhv_svm_off(struct kvm *kvm)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5549) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5550) struct kvm_vcpu *vcpu;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5551) int mmu_was_ready;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5552) int srcu_idx;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5553) int ret = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5554) int i;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5555)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5556) if (!(kvm->arch.secure_guest & KVMPPC_SECURE_INIT_START))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5557) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5558)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5559) mutex_lock(&kvm->arch.mmu_setup_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5560) mmu_was_ready = kvm->arch.mmu_ready;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5561) if (kvm->arch.mmu_ready) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5562) kvm->arch.mmu_ready = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5563) /* order mmu_ready vs. vcpus_running */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5564) smp_mb();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5565) if (atomic_read(&kvm->arch.vcpus_running)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5566) kvm->arch.mmu_ready = 1;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5567) ret = -EBUSY;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5568) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5569) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5570) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5571)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5572) srcu_idx = srcu_read_lock(&kvm->srcu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5573) for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5574) struct kvm_memory_slot *memslot;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5575) struct kvm_memslots *slots = __kvm_memslots(kvm, i);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5576)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5577) if (!slots)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5578) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5579)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5580) kvm_for_each_memslot(memslot, slots) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5581) kvmppc_uvmem_drop_pages(memslot, kvm, true);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5582) uv_unregister_mem_slot(kvm->arch.lpid, memslot->id);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5583) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5584) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5585) srcu_read_unlock(&kvm->srcu, srcu_idx);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5586)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5587) ret = uv_svm_terminate(kvm->arch.lpid);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5588) if (ret != U_SUCCESS) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5589) ret = -EINVAL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5590) goto out;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5591) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5592)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5593) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5594) * When secure guest is reset, all the guest pages are sent
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5595) * to UV via UV_PAGE_IN before the non-boot vcpus get a
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5596) * chance to run and unpin their VPA pages. Unpinning of all
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5597) * VPA pages is done here explicitly so that VPA pages
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5598) * can be migrated to the secure side.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5599) *
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5600) * This is required to for the secure SMP guest to reboot
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5601) * correctly.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5602) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5603) kvm_for_each_vcpu(i, vcpu, kvm) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5604) spin_lock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5605) unpin_vpa_reset(kvm, &vcpu->arch.dtl);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5606) unpin_vpa_reset(kvm, &vcpu->arch.slb_shadow);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5607) unpin_vpa_reset(kvm, &vcpu->arch.vpa);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5608) spin_unlock(&vcpu->arch.vpa_update_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5609) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5610)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5611) kvmppc_setup_partition_table(kvm);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5612) kvm->arch.secure_guest = 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5613) kvm->arch.mmu_ready = mmu_was_ready;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5614) out:
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5615) mutex_unlock(&kvm->arch.mmu_setup_lock);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5616) return ret;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5617) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5618)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5619) static struct kvmppc_ops kvm_ops_hv = {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5620) .get_sregs = kvm_arch_vcpu_ioctl_get_sregs_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5621) .set_sregs = kvm_arch_vcpu_ioctl_set_sregs_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5622) .get_one_reg = kvmppc_get_one_reg_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5623) .set_one_reg = kvmppc_set_one_reg_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5624) .vcpu_load = kvmppc_core_vcpu_load_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5625) .vcpu_put = kvmppc_core_vcpu_put_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5626) .inject_interrupt = kvmppc_inject_interrupt_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5627) .set_msr = kvmppc_set_msr_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5628) .vcpu_run = kvmppc_vcpu_run_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5629) .vcpu_create = kvmppc_core_vcpu_create_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5630) .vcpu_free = kvmppc_core_vcpu_free_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5631) .check_requests = kvmppc_core_check_requests_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5632) .get_dirty_log = kvm_vm_ioctl_get_dirty_log_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5633) .flush_memslot = kvmppc_core_flush_memslot_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5634) .prepare_memory_region = kvmppc_core_prepare_memory_region_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5635) .commit_memory_region = kvmppc_core_commit_memory_region_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5636) .unmap_hva_range = kvm_unmap_hva_range_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5637) .age_hva = kvm_age_hva_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5638) .test_age_hva = kvm_test_age_hva_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5639) .set_spte_hva = kvm_set_spte_hva_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5640) .free_memslot = kvmppc_core_free_memslot_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5641) .init_vm = kvmppc_core_init_vm_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5642) .destroy_vm = kvmppc_core_destroy_vm_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5643) .get_smmu_info = kvm_vm_ioctl_get_smmu_info_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5644) .emulate_op = kvmppc_core_emulate_op_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5645) .emulate_mtspr = kvmppc_core_emulate_mtspr_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5646) .emulate_mfspr = kvmppc_core_emulate_mfspr_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5647) .fast_vcpu_kick = kvmppc_fast_vcpu_kick_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5648) .arch_vm_ioctl = kvm_arch_vm_ioctl_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5649) .hcall_implemented = kvmppc_hcall_impl_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5650) #ifdef CONFIG_KVM_XICS
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5651) .irq_bypass_add_producer = kvmppc_irq_bypass_add_producer_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5652) .irq_bypass_del_producer = kvmppc_irq_bypass_del_producer_hv,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5653) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5654) .configure_mmu = kvmhv_configure_mmu,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5655) .get_rmmu_info = kvmhv_get_rmmu_info,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5656) .set_smt_mode = kvmhv_set_smt_mode,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5657) .enable_nested = kvmhv_enable_nested,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5658) .load_from_eaddr = kvmhv_load_from_eaddr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5659) .store_to_eaddr = kvmhv_store_to_eaddr,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5660) .enable_svm = kvmhv_enable_svm,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5661) .svm_off = kvmhv_svm_off,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5662) };
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5663)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5664) static int kvm_init_subcore_bitmap(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5665) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5666) int i, j;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5667) int nr_cores = cpu_nr_cores();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5668) struct sibling_subcore_state *sibling_subcore_state;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5669)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5670) for (i = 0; i < nr_cores; i++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5671) int first_cpu = i * threads_per_core;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5672) int node = cpu_to_node(first_cpu);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5673)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5674) /* Ignore if it is already allocated. */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5675) if (paca_ptrs[first_cpu]->sibling_subcore_state)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5676) continue;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5677)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5678) sibling_subcore_state =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5679) kzalloc_node(sizeof(struct sibling_subcore_state),
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5680) GFP_KERNEL, node);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5681) if (!sibling_subcore_state)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5682) return -ENOMEM;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5683)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5684)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5685) for (j = 0; j < threads_per_core; j++) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5686) int cpu = first_cpu + j;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5687)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5688) paca_ptrs[cpu]->sibling_subcore_state =
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5689) sibling_subcore_state;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5690) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5691) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5692) return 0;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5693) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5694)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5695) static int kvmppc_radix_possible(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5696) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5697) return cpu_has_feature(CPU_FTR_ARCH_300) && radix_enabled();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5698) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5699)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5700) static int kvmppc_book3s_init_hv(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5701) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5702) int r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5703)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5704) if (!tlbie_capable) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5705) pr_err("KVM-HV: Host does not support TLBIE\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5706) return -ENODEV;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5707) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5708)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5709) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5710) * FIXME!! Do we need to check on all cpus ?
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5711) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5712) r = kvmppc_core_check_processor_compat_hv();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5713) if (r < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5714) return -ENODEV;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5715)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5716) r = kvmhv_nested_init();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5717) if (r)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5718) return r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5719)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5720) r = kvm_init_subcore_bitmap();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5721) if (r)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5722) return r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5723)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5724) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5725) * We need a way of accessing the XICS interrupt controller,
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5726) * either directly, via paca_ptrs[cpu]->kvm_hstate.xics_phys, or
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5727) * indirectly, via OPAL.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5728) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5729) #ifdef CONFIG_SMP
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5730) if (!xics_on_xive() && !kvmhv_on_pseries() &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5731) !local_paca->kvm_hstate.xics_phys) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5732) struct device_node *np;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5733)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5734) np = of_find_compatible_node(NULL, NULL, "ibm,opal-intc");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5735) if (!np) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5736) pr_err("KVM-HV: Cannot determine method for accessing XICS\n");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5737) return -ENODEV;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5738) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5739) /* presence of intc confirmed - node can be dropped again */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5740) of_node_put(np);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5741) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5742) #endif
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5743)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5744) kvm_ops_hv.owner = THIS_MODULE;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5745) kvmppc_hv_ops = &kvm_ops_hv;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5746)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5747) init_default_hcalls();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5748)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5749) init_vcore_lists();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5750)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5751) r = kvmppc_mmu_hv_init();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5752) if (r)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5753) return r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5754)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5755) if (kvmppc_radix_possible()) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5756) r = kvmppc_radix_init();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5757) if (r)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5758) return r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5759) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5760)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5761) /*
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5762) * POWER9 chips before version 2.02 can't have some threads in
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5763) * HPT mode and some in radix mode on the same core.
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5764) */
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5765) if (cpu_has_feature(CPU_FTR_ARCH_300)) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5766) unsigned int pvr = mfspr(SPRN_PVR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5767) if ((pvr >> 16) == PVR_POWER9 &&
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5768) (((pvr & 0xe000) == 0 && (pvr & 0xfff) < 0x202) ||
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5769) ((pvr & 0xe000) == 0x2000 && (pvr & 0xfff) < 0x101)))
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5770) no_mixing_hpt_and_radix = true;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5771) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5772)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5773) r = kvmppc_uvmem_init();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5774) if (r < 0)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5775) pr_err("KVM-HV: kvmppc_uvmem_init failed %d\n", r);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5776)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5777) return r;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5778) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5779)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5780) static void kvmppc_book3s_exit_hv(void)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5781) {
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5782) kvmppc_uvmem_free();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5783) kvmppc_free_host_rm_ops();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5784) if (kvmppc_radix_possible())
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5785) kvmppc_radix_exit();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5786) kvmppc_hv_ops = NULL;
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5787) kvmhv_nested_exit();
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5788) }
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5789)
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5790) module_init(kvmppc_book3s_init_hv);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5791) module_exit(kvmppc_book3s_exit_hv);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5792) MODULE_LICENSE("GPL");
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5793) MODULE_ALIAS_MISCDEV(KVM_MINOR);
^8f3ce5b39 (kx 2023-10-28 12:00:06 +0300 5794) MODULE_ALIAS("devname:kvm");
Orange Pi5 kernel
Deprecated Linux kernel 5.10.110 for OrangePi 5/5B/5+ boards
3 Commits
0 Branches
0 Tags