| .. | .. |
|---|
| 1 | +// SPDX-License-Identifier: GPL-2.0-only |
|---|
| 1 | 2 | /* |
|---|
| 2 | 3 | * kvm asynchronous fault support |
|---|
| 3 | 4 | * |
|---|
| .. | .. |
|---|
| 5 | 6 | * |
|---|
| 6 | 7 | * Author: |
|---|
| 7 | 8 | * Gleb Natapov <gleb@redhat.com> |
|---|
| 8 | | - * |
|---|
| 9 | | - * This file is free software; you can redistribute it and/or modify |
|---|
| 10 | | - * it under the terms of version 2 of the GNU General Public License |
|---|
| 11 | | - * as published by the Free Software Foundation. |
|---|
| 12 | | - * |
|---|
| 13 | | - * This program is distributed in the hope that it will be useful, |
|---|
| 14 | | - * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|---|
| 15 | | - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
|---|
| 16 | | - * GNU General Public License for more details. |
|---|
| 17 | | - * |
|---|
| 18 | | - * You should have received a copy of the GNU General Public License |
|---|
| 19 | | - * along with this program; if not, write to the Free Software Foundation, |
|---|
| 20 | | - * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA. |
|---|
| 21 | 9 | */ |
|---|
| 22 | 10 | |
|---|
| 23 | 11 | #include <linux/kvm_host.h> |
|---|
| .. | .. |
|---|
| 28 | 16 | |
|---|
| 29 | 17 | #include "async_pf.h" |
|---|
| 30 | 18 | #include <trace/events/kvm.h> |
|---|
| 31 | | - |
|---|
| 32 | | -static inline void kvm_async_page_present_sync(struct kvm_vcpu *vcpu, |
|---|
| 33 | | - struct kvm_async_pf *work) |
|---|
| 34 | | -{ |
|---|
| 35 | | -#ifdef CONFIG_KVM_ASYNC_PF_SYNC |
|---|
| 36 | | - kvm_arch_async_page_present(vcpu, work); |
|---|
| 37 | | -#endif |
|---|
| 38 | | -} |
|---|
| 39 | | -static inline void kvm_async_page_present_async(struct kvm_vcpu *vcpu, |
|---|
| 40 | | - struct kvm_async_pf *work) |
|---|
| 41 | | -{ |
|---|
| 42 | | -#ifndef CONFIG_KVM_ASYNC_PF_SYNC |
|---|
| 43 | | - kvm_arch_async_page_present(vcpu, work); |
|---|
| 44 | | -#endif |
|---|
| 45 | | -} |
|---|
| 46 | 19 | |
|---|
| 47 | 20 | static struct kmem_cache *async_pf_cache; |
|---|
| 48 | 21 | |
|---|
| .. | .. |
|---|
| 78 | 51 | unsigned long addr = apf->addr; |
|---|
| 79 | 52 | gpa_t cr2_or_gpa = apf->cr2_or_gpa; |
|---|
| 80 | 53 | int locked = 1; |
|---|
| 54 | + bool first; |
|---|
| 81 | 55 | |
|---|
| 82 | 56 | might_sleep(); |
|---|
| 83 | 57 | |
|---|
| 84 | 58 | /* |
|---|
| 85 | | - * This work is run asynchromously to the task which owns |
|---|
| 59 | + * This work is run asynchronously to the task which owns |
|---|
| 86 | 60 | * mm and might be done in another context, so we must |
|---|
| 87 | 61 | * access remotely. |
|---|
| 88 | 62 | */ |
|---|
| 89 | | - down_read(&mm->mmap_sem); |
|---|
| 90 | | - get_user_pages_remote(NULL, mm, addr, 1, FOLL_WRITE, NULL, NULL, |
|---|
| 63 | + mmap_read_lock(mm); |
|---|
| 64 | + get_user_pages_remote(mm, addr, 1, FOLL_WRITE, NULL, NULL, |
|---|
| 91 | 65 | &locked); |
|---|
| 92 | 66 | if (locked) |
|---|
| 93 | | - up_read(&mm->mmap_sem); |
|---|
| 67 | + mmap_read_unlock(mm); |
|---|
| 94 | 68 | |
|---|
| 95 | | - kvm_async_page_present_sync(vcpu, apf); |
|---|
| 69 | + if (IS_ENABLED(CONFIG_KVM_ASYNC_PF_SYNC)) |
|---|
| 70 | + kvm_arch_async_page_present(vcpu, apf); |
|---|
| 96 | 71 | |
|---|
| 97 | 72 | spin_lock(&vcpu->async_pf.lock); |
|---|
| 73 | + first = list_empty(&vcpu->async_pf.done); |
|---|
| 98 | 74 | list_add_tail(&apf->link, &vcpu->async_pf.done); |
|---|
| 99 | 75 | apf->vcpu = NULL; |
|---|
| 100 | 76 | spin_unlock(&vcpu->async_pf.lock); |
|---|
| 77 | + |
|---|
| 78 | + if (!IS_ENABLED(CONFIG_KVM_ASYNC_PF_SYNC) && first) |
|---|
| 79 | + kvm_arch_async_page_present_queued(vcpu); |
|---|
| 101 | 80 | |
|---|
| 102 | 81 | /* |
|---|
| 103 | 82 | * apf may be freed by kvm_check_async_pf_completion() after |
|---|
| .. | .. |
|---|
| 106 | 85 | |
|---|
| 107 | 86 | trace_kvm_async_pf_completed(addr, cr2_or_gpa); |
|---|
| 108 | 87 | |
|---|
| 109 | | - if (swq_has_sleeper(&vcpu->wq)) |
|---|
| 110 | | - swake_up_one(&vcpu->wq); |
|---|
| 88 | + rcuwait_wake_up(&vcpu->wait); |
|---|
| 111 | 89 | |
|---|
| 112 | 90 | mmput(mm); |
|---|
| 113 | 91 | kvm_put_kvm(vcpu->kvm); |
|---|
| .. | .. |
|---|
| 161 | 139 | struct kvm_async_pf *work; |
|---|
| 162 | 140 | |
|---|
| 163 | 141 | while (!list_empty_careful(&vcpu->async_pf.done) && |
|---|
| 164 | | - kvm_arch_can_inject_async_page_present(vcpu)) { |
|---|
| 142 | + kvm_arch_can_dequeue_async_page_present(vcpu)) { |
|---|
| 165 | 143 | spin_lock(&vcpu->async_pf.lock); |
|---|
| 166 | 144 | work = list_first_entry(&vcpu->async_pf.done, typeof(*work), |
|---|
| 167 | 145 | link); |
|---|
| .. | .. |
|---|
| 169 | 147 | spin_unlock(&vcpu->async_pf.lock); |
|---|
| 170 | 148 | |
|---|
| 171 | 149 | kvm_arch_async_page_ready(vcpu, work); |
|---|
| 172 | | - kvm_async_page_present_async(vcpu, work); |
|---|
| 150 | + if (!IS_ENABLED(CONFIG_KVM_ASYNC_PF_SYNC)) |
|---|
| 151 | + kvm_arch_async_page_present(vcpu, work); |
|---|
| 173 | 152 | |
|---|
| 174 | 153 | list_del(&work->queue); |
|---|
| 175 | 154 | vcpu->async_pf.queued--; |
|---|
| .. | .. |
|---|
| 177 | 156 | } |
|---|
| 178 | 157 | } |
|---|
| 179 | 158 | |
|---|
| 180 | | -int kvm_setup_async_pf(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa, |
|---|
| 181 | | - unsigned long hva, struct kvm_arch_async_pf *arch) |
|---|
| 159 | +/* |
|---|
| 160 | + * Try to schedule a job to handle page fault asynchronously. Returns 'true' on |
|---|
| 161 | + * success, 'false' on failure (page fault has to be handled synchronously). |
|---|
| 162 | + */ |
|---|
| 163 | +bool kvm_setup_async_pf(struct kvm_vcpu *vcpu, gpa_t cr2_or_gpa, |
|---|
| 164 | + unsigned long hva, struct kvm_arch_async_pf *arch) |
|---|
| 182 | 165 | { |
|---|
| 183 | 166 | struct kvm_async_pf *work; |
|---|
| 184 | 167 | |
|---|
| 185 | 168 | if (vcpu->async_pf.queued >= ASYNC_PF_PER_VCPU) |
|---|
| 186 | | - return 0; |
|---|
| 169 | + return false; |
|---|
| 187 | 170 | |
|---|
| 188 | | - /* setup delayed work */ |
|---|
| 171 | + /* Arch specific code should not do async PF in this case */ |
|---|
| 172 | + if (unlikely(kvm_is_error_hva(hva))) |
|---|
| 173 | + return false; |
|---|
| 189 | 174 | |
|---|
| 190 | 175 | /* |
|---|
| 191 | 176 | * do alloc nowait since if we are going to sleep anyway we |
|---|
| .. | .. |
|---|
| 193 | 178 | */ |
|---|
| 194 | 179 | work = kmem_cache_zalloc(async_pf_cache, GFP_NOWAIT | __GFP_NOWARN); |
|---|
| 195 | 180 | if (!work) |
|---|
| 196 | | - return 0; |
|---|
| 181 | + return false; |
|---|
| 197 | 182 | |
|---|
| 198 | 183 | work->wakeup_all = false; |
|---|
| 199 | 184 | work->vcpu = vcpu; |
|---|
| .. | .. |
|---|
| 204 | 189 | mmget(work->mm); |
|---|
| 205 | 190 | kvm_get_kvm(work->vcpu->kvm); |
|---|
| 206 | 191 | |
|---|
| 207 | | - /* this can't really happen otherwise gfn_to_pfn_async |
|---|
| 208 | | - would succeed */ |
|---|
| 209 | | - if (unlikely(kvm_is_error_hva(work->addr))) |
|---|
| 210 | | - goto retry_sync; |
|---|
| 211 | | - |
|---|
| 212 | 192 | INIT_WORK(&work->work, async_pf_execute); |
|---|
| 213 | | - if (!schedule_work(&work->work)) |
|---|
| 214 | | - goto retry_sync; |
|---|
| 215 | 193 | |
|---|
| 216 | 194 | list_add_tail(&work->queue, &vcpu->async_pf.queue); |
|---|
| 217 | 195 | vcpu->async_pf.queued++; |
|---|
| 218 | | - kvm_arch_async_page_not_present(vcpu, work); |
|---|
| 219 | | - return 1; |
|---|
| 220 | | -retry_sync: |
|---|
| 221 | | - kvm_put_kvm(work->vcpu->kvm); |
|---|
| 222 | | - mmput(work->mm); |
|---|
| 223 | | - kmem_cache_free(async_pf_cache, work); |
|---|
| 224 | | - return 0; |
|---|
| 196 | + work->notpresent_injected = kvm_arch_async_page_not_present(vcpu, work); |
|---|
| 197 | + |
|---|
| 198 | + schedule_work(&work->work); |
|---|
| 199 | + |
|---|
| 200 | + return true; |
|---|
| 225 | 201 | } |
|---|
| 226 | 202 | |
|---|
| 227 | 203 | int kvm_async_pf_wakeup_all(struct kvm_vcpu *vcpu) |
|---|
| 228 | 204 | { |
|---|
| 229 | 205 | struct kvm_async_pf *work; |
|---|
| 206 | + bool first; |
|---|
| 230 | 207 | |
|---|
| 231 | 208 | if (!list_empty_careful(&vcpu->async_pf.done)) |
|---|
| 232 | 209 | return 0; |
|---|
| .. | .. |
|---|
| 239 | 216 | INIT_LIST_HEAD(&work->queue); /* for list_del to work */ |
|---|
| 240 | 217 | |
|---|
| 241 | 218 | spin_lock(&vcpu->async_pf.lock); |
|---|
| 219 | + first = list_empty(&vcpu->async_pf.done); |
|---|
| 242 | 220 | list_add_tail(&work->link, &vcpu->async_pf.done); |
|---|
| 243 | 221 | spin_unlock(&vcpu->async_pf.lock); |
|---|
| 244 | 222 | |
|---|
| 223 | + if (!IS_ENABLED(CONFIG_KVM_ASYNC_PF_SYNC) && first) |
|---|
| 224 | + kvm_arch_async_page_present_queued(vcpu); |
|---|
| 225 | + |
|---|
| 245 | 226 | vcpu->async_pf.queued++; |
|---|
| 246 | 227 | return 0; |
|---|
| 247 | 228 | } |
|---|