vcpu-requests.rst (13936B)
1.. SPDX-License-Identifier: GPL-2.0 2 3================= 4KVM VCPU Requests 5================= 6 7Overview 8======== 9 10KVM supports an internal API enabling threads to request a VCPU thread to 11perform some activity. For example, a thread may request a VCPU to flush 12its TLB with a VCPU request. The API consists of the following functions:: 13 14 /* Check if any requests are pending for VCPU @vcpu. */ 15 bool kvm_request_pending(struct kvm_vcpu *vcpu); 16 17 /* Check if VCPU @vcpu has request @req pending. */ 18 bool kvm_test_request(int req, struct kvm_vcpu *vcpu); 19 20 /* Clear request @req for VCPU @vcpu. */ 21 void kvm_clear_request(int req, struct kvm_vcpu *vcpu); 22 23 /* 24 * Check if VCPU @vcpu has request @req pending. When the request is 25 * pending it will be cleared and a memory barrier, which pairs with 26 * another in kvm_make_request(), will be issued. 27 */ 28 bool kvm_check_request(int req, struct kvm_vcpu *vcpu); 29 30 /* 31 * Make request @req of VCPU @vcpu. Issues a memory barrier, which pairs 32 * with another in kvm_check_request(), prior to setting the request. 33 */ 34 void kvm_make_request(int req, struct kvm_vcpu *vcpu); 35 36 /* Make request @req of all VCPUs of the VM with struct kvm @kvm. */ 37 bool kvm_make_all_cpus_request(struct kvm *kvm, unsigned int req); 38 39Typically a requester wants the VCPU to perform the activity as soon 40as possible after making the request. This means most requests 41(kvm_make_request() calls) are followed by a call to kvm_vcpu_kick(), 42and kvm_make_all_cpus_request() has the kicking of all VCPUs built 43into it. 44 45VCPU Kicks 46---------- 47 48The goal of a VCPU kick is to bring a VCPU thread out of guest mode in 49order to perform some KVM maintenance. To do so, an IPI is sent, forcing 50a guest mode exit. However, a VCPU thread may not be in guest mode at the 51time of the kick. Therefore, depending on the mode and state of the VCPU 52thread, there are two other actions a kick may take. All three actions 53are listed below: 54 551) Send an IPI. This forces a guest mode exit. 562) Waking a sleeping VCPU. Sleeping VCPUs are VCPU threads outside guest 57 mode that wait on waitqueues. Waking them removes the threads from 58 the waitqueues, allowing the threads to run again. This behavior 59 may be suppressed, see KVM_REQUEST_NO_WAKEUP below. 603) Nothing. When the VCPU is not in guest mode and the VCPU thread is not 61 sleeping, then there is nothing to do. 62 63VCPU Mode 64--------- 65 66VCPUs have a mode state, ``vcpu->mode``, that is used to track whether the 67guest is running in guest mode or not, as well as some specific 68outside guest mode states. The architecture may use ``vcpu->mode`` to 69ensure VCPU requests are seen by VCPUs (see "Ensuring Requests Are Seen"), 70as well as to avoid sending unnecessary IPIs (see "IPI Reduction"), and 71even to ensure IPI acknowledgements are waited upon (see "Waiting for 72Acknowledgements"). The following modes are defined: 73 74OUTSIDE_GUEST_MODE 75 76 The VCPU thread is outside guest mode. 77 78IN_GUEST_MODE 79 80 The VCPU thread is in guest mode. 81 82EXITING_GUEST_MODE 83 84 The VCPU thread is transitioning from IN_GUEST_MODE to 85 OUTSIDE_GUEST_MODE. 86 87READING_SHADOW_PAGE_TABLES 88 89 The VCPU thread is outside guest mode, but it wants the sender of 90 certain VCPU requests, namely KVM_REQ_TLB_FLUSH, to wait until the VCPU 91 thread is done reading the page tables. 92 93VCPU Request Internals 94====================== 95 96VCPU requests are simply bit indices of the ``vcpu->requests`` bitmap. 97This means general bitops, like those documented in [atomic-ops]_ could 98also be used, e.g. :: 99 100 clear_bit(KVM_REQ_UNHALT & KVM_REQUEST_MASK, &vcpu->requests); 101 102However, VCPU request users should refrain from doing so, as it would 103break the abstraction. The first 8 bits are reserved for architecture 104independent requests, all additional bits are available for architecture 105dependent requests. 106 107Architecture Independent Requests 108--------------------------------- 109 110KVM_REQ_TLB_FLUSH 111 112 KVM's common MMU notifier may need to flush all of a guest's TLB 113 entries, calling kvm_flush_remote_tlbs() to do so. Architectures that 114 choose to use the common kvm_flush_remote_tlbs() implementation will 115 need to handle this VCPU request. 116 117KVM_REQ_VM_DEAD 118 119 This request informs all VCPUs that the VM is dead and unusable, e.g. due to 120 fatal error or because the VM's state has been intentionally destroyed. 121 122KVM_REQ_UNBLOCK 123 124 This request informs the vCPU to exit kvm_vcpu_block. It is used for 125 example from timer handlers that run on the host on behalf of a vCPU, 126 or in order to update the interrupt routing and ensure that assigned 127 devices will wake up the vCPU. 128 129KVM_REQ_UNHALT 130 131 This request may be made from the KVM common function kvm_vcpu_block(), 132 which is used to emulate an instruction that causes a CPU to halt until 133 one of an architectural specific set of events and/or interrupts is 134 received (determined by checking kvm_arch_vcpu_runnable()). When that 135 event or interrupt arrives kvm_vcpu_block() makes the request. This is 136 in contrast to when kvm_vcpu_block() returns due to any other reason, 137 such as a pending signal, which does not indicate the VCPU's halt 138 emulation should stop, and therefore does not make the request. 139 140KVM_REQ_OUTSIDE_GUEST_MODE 141 142 This "request" ensures the target vCPU has exited guest mode prior to the 143 sender of the request continuing on. No action needs be taken by the target, 144 and so no request is actually logged for the target. This request is similar 145 to a "kick", but unlike a kick it guarantees the vCPU has actually exited 146 guest mode. A kick only guarantees the vCPU will exit at some point in the 147 future, e.g. a previous kick may have started the process, but there's no 148 guarantee the to-be-kicked vCPU has fully exited guest mode. 149 150KVM_REQUEST_MASK 151---------------- 152 153VCPU requests should be masked by KVM_REQUEST_MASK before using them with 154bitops. This is because only the lower 8 bits are used to represent the 155request's number. The upper bits are used as flags. Currently only two 156flags are defined. 157 158VCPU Request Flags 159------------------ 160 161KVM_REQUEST_NO_WAKEUP 162 163 This flag is applied to requests that only need immediate attention 164 from VCPUs running in guest mode. That is, sleeping VCPUs do not need 165 to be awaken for these requests. Sleeping VCPUs will handle the 166 requests when they are awaken later for some other reason. 167 168KVM_REQUEST_WAIT 169 170 When requests with this flag are made with kvm_make_all_cpus_request(), 171 then the caller will wait for each VCPU to acknowledge its IPI before 172 proceeding. This flag only applies to VCPUs that would receive IPIs. 173 If, for example, the VCPU is sleeping, so no IPI is necessary, then 174 the requesting thread does not wait. This means that this flag may be 175 safely combined with KVM_REQUEST_NO_WAKEUP. See "Waiting for 176 Acknowledgements" for more information about requests with 177 KVM_REQUEST_WAIT. 178 179VCPU Requests with Associated State 180=================================== 181 182Requesters that want the receiving VCPU to handle new state need to ensure 183the newly written state is observable to the receiving VCPU thread's CPU 184by the time it observes the request. This means a write memory barrier 185must be inserted after writing the new state and before setting the VCPU 186request bit. Additionally, on the receiving VCPU thread's side, a 187corresponding read barrier must be inserted after reading the request bit 188and before proceeding to read the new state associated with it. See 189scenario 3, Message and Flag, of [lwn-mb]_ and the kernel documentation 190[memory-barriers]_. 191 192The pair of functions, kvm_check_request() and kvm_make_request(), provide 193the memory barriers, allowing this requirement to be handled internally by 194the API. 195 196Ensuring Requests Are Seen 197========================== 198 199When making requests to VCPUs, we want to avoid the receiving VCPU 200executing in guest mode for an arbitrary long time without handling the 201request. We can be sure this won't happen as long as we ensure the VCPU 202thread checks kvm_request_pending() before entering guest mode and that a 203kick will send an IPI to force an exit from guest mode when necessary. 204Extra care must be taken to cover the period after the VCPU thread's last 205kvm_request_pending() check and before it has entered guest mode, as kick 206IPIs will only trigger guest mode exits for VCPU threads that are in guest 207mode or at least have already disabled interrupts in order to prepare to 208enter guest mode. This means that an optimized implementation (see "IPI 209Reduction") must be certain when it's safe to not send the IPI. One 210solution, which all architectures except s390 apply, is to: 211 212- set ``vcpu->mode`` to IN_GUEST_MODE between disabling the interrupts and 213 the last kvm_request_pending() check; 214- enable interrupts atomically when entering the guest. 215 216This solution also requires memory barriers to be placed carefully in both 217the requesting thread and the receiving VCPU. With the memory barriers we 218can exclude the possibility of a VCPU thread observing 219!kvm_request_pending() on its last check and then not receiving an IPI for 220the next request made of it, even if the request is made immediately after 221the check. This is done by way of the Dekker memory barrier pattern 222(scenario 10 of [lwn-mb]_). As the Dekker pattern requires two variables, 223this solution pairs ``vcpu->mode`` with ``vcpu->requests``. Substituting 224them into the pattern gives:: 225 226 CPU1 CPU2 227 ================= ================= 228 local_irq_disable(); 229 WRITE_ONCE(vcpu->mode, IN_GUEST_MODE); kvm_make_request(REQ, vcpu); 230 smp_mb(); smp_mb(); 231 if (kvm_request_pending(vcpu)) { if (READ_ONCE(vcpu->mode) == 232 IN_GUEST_MODE) { 233 ...abort guest entry... ...send IPI... 234 } } 235 236As stated above, the IPI is only useful for VCPU threads in guest mode or 237that have already disabled interrupts. This is why this specific case of 238the Dekker pattern has been extended to disable interrupts before setting 239``vcpu->mode`` to IN_GUEST_MODE. WRITE_ONCE() and READ_ONCE() are used to 240pedantically implement the memory barrier pattern, guaranteeing the 241compiler doesn't interfere with ``vcpu->mode``'s carefully planned 242accesses. 243 244IPI Reduction 245------------- 246 247As only one IPI is needed to get a VCPU to check for any/all requests, 248then they may be coalesced. This is easily done by having the first IPI 249sending kick also change the VCPU mode to something !IN_GUEST_MODE. The 250transitional state, EXITING_GUEST_MODE, is used for this purpose. 251 252Waiting for Acknowledgements 253---------------------------- 254 255Some requests, those with the KVM_REQUEST_WAIT flag set, require IPIs to 256be sent, and the acknowledgements to be waited upon, even when the target 257VCPU threads are in modes other than IN_GUEST_MODE. For example, one case 258is when a target VCPU thread is in READING_SHADOW_PAGE_TABLES mode, which 259is set after disabling interrupts. To support these cases, the 260KVM_REQUEST_WAIT flag changes the condition for sending an IPI from 261checking that the VCPU is IN_GUEST_MODE to checking that it is not 262OUTSIDE_GUEST_MODE. 263 264Request-less VCPU Kicks 265----------------------- 266 267As the determination of whether or not to send an IPI depends on the 268two-variable Dekker memory barrier pattern, then it's clear that 269request-less VCPU kicks are almost never correct. Without the assurance 270that a non-IPI generating kick will still result in an action by the 271receiving VCPU, as the final kvm_request_pending() check does for 272request-accompanying kicks, then the kick may not do anything useful at 273all. If, for instance, a request-less kick was made to a VCPU that was 274just about to set its mode to IN_GUEST_MODE, meaning no IPI is sent, then 275the VCPU thread may continue its entry without actually having done 276whatever it was the kick was meant to initiate. 277 278One exception is x86's posted interrupt mechanism. In this case, however, 279even the request-less VCPU kick is coupled with the same 280local_irq_disable() + smp_mb() pattern described above; the ON bit 281(Outstanding Notification) in the posted interrupt descriptor takes the 282role of ``vcpu->requests``. When sending a posted interrupt, PIR.ON is 283set before reading ``vcpu->mode``; dually, in the VCPU thread, 284vmx_sync_pir_to_irr() reads PIR after setting ``vcpu->mode`` to 285IN_GUEST_MODE. 286 287Additional Considerations 288========================= 289 290Sleeping VCPUs 291-------------- 292 293VCPU threads may need to consider requests before and/or after calling 294functions that may put them to sleep, e.g. kvm_vcpu_block(). Whether they 295do or not, and, if they do, which requests need consideration, is 296architecture dependent. kvm_vcpu_block() calls kvm_arch_vcpu_runnable() 297to check if it should awaken. One reason to do so is to provide 298architectures a function where requests may be checked if necessary. 299 300Clearing Requests 301----------------- 302 303Generally it only makes sense for the receiving VCPU thread to clear a 304request. However, in some circumstances, such as when the requesting 305thread and the receiving VCPU thread are executed serially, such as when 306they are the same thread, or when they are using some form of concurrency 307control to temporarily execute synchronously, then it's possible to know 308that the request may be cleared immediately, rather than waiting for the 309receiving VCPU thread to handle the request in VCPU RUN. The only current 310examples of this are kvm_vcpu_block() calls made by VCPUs to block 311themselves. A possible side-effect of that call is to make the 312KVM_REQ_UNHALT request, which may then be cleared immediately when the 313VCPU returns from the call. 314 315References 316========== 317 318.. [atomic-ops] Documentation/atomic_bitops.txt and Documentation/atomic_t.txt 319.. [memory-barriers] Documentation/memory-barriers.txt 320.. [lwn-mb] https://lwn.net/Articles/573436/