| 1 |
.. SPDX-License-Identifier: GPL-2.0 |
| 2 |
|
| 3 |
=================== |
| 4 |
Linux KVM Hypercall |
| 5 |
=================== |
| 6 |
|
| 7 |
X86: |
| 8 |
KVM Hypercalls have a three-byte sequence of either the vmcall or the vmmcall |
| 9 |
instruction. The hypervisor can replace it with instructions that are |
| 10 |
guaranteed to be supported. |
| 11 |
|
| 12 |
Up to four arguments may be passed in rbx, rcx, rdx, and rsi respectively. |
| 13 |
The hypercall number should be placed in rax and the return value will be |
| 14 |
placed in rax. No other registers will be clobbered unless explicitly stated |
| 15 |
by the particular hypercall. |
| 16 |
|
| 17 |
S390: |
| 18 |
R2-R7 are used for parameters 1-6. In addition, R1 is used for hypercall |
| 19 |
number. The return value is written to R2. |
| 20 |
|
| 21 |
S390 uses diagnose instruction as hypercall (0x500) along with hypercall |
| 22 |
number in R1. |
| 23 |
|
| 24 |
For further information on the S390 diagnose call as supported by KVM, |
| 25 |
refer to Documentation/virt/kvm/s390/s390-diag.rst. |
| 26 |
|
| 27 |
PowerPC: |
| 28 |
It uses R3-R10 and hypercall number in R11. R4-R11 are used as output registers. |
| 29 |
Return value is placed in R3. |
| 30 |
|
| 31 |
KVM hypercalls uses 4 byte opcode, that are patched with 'hypercall-instructions' |
| 32 |
property inside the device tree's /hypervisor node. |
| 33 |
For more information refer to Documentation/virt/kvm/ppc-pv.rst |
| 34 |
|
| 35 |
MIPS: |
| 36 |
KVM hypercalls use the HYPCALL instruction with code 0 and the hypercall |
| 37 |
number in $2 (v0). Up to four arguments may be placed in $4-$7 (a0-a3) and |
| 38 |
the return value is placed in $2 (v0). |
| 39 |
|
| 40 |
KVM Hypercalls Documentation |
| 41 |
============================ |
| 42 |
|
| 43 |
The template for each hypercall is: |
| 44 |
1. Hypercall name. |
| 45 |
2. Architecture(s) |
| 46 |
3. Status (deprecated, obsolete, active) |
| 47 |
4. Purpose |
| 48 |
|
| 49 |
1. KVM_HC_VAPIC_POLL_IRQ |
| 50 |
------------------------ |
| 51 |
|
| 52 |
:Architecture: x86 |
| 53 |
:Status: active |
| 54 |
:Purpose: Trigger guest exit so that the host can check for pending |
| 55 |
interrupts on reentry. |
| 56 |
|
| 57 |
2. KVM_HC_MMU_OP |
| 58 |
---------------- |
| 59 |
|
| 60 |
:Architecture: x86 |
| 61 |
:Status: deprecated. |
| 62 |
:Purpose: Support MMU operations such as writing to PTE, |
| 63 |
flushing TLB, release PT. |
| 64 |
|
| 65 |
3. KVM_HC_FEATURES |
| 66 |
------------------ |
| 67 |
|
| 68 |
:Architecture: PPC |
| 69 |
:Status: active |
| 70 |
:Purpose: Expose hypercall availability to the guest. On x86 platforms, cpuid |
| 71 |
used to enumerate which hypercalls are available. On PPC, either |
| 72 |
device tree based lookup ( which is also what EPAPR dictates) |
| 73 |
OR KVM specific enumeration mechanism (which is this hypercall) |
| 74 |
can be used. |
| 75 |
|
| 76 |
4. KVM_HC_PPC_MAP_MAGIC_PAGE |
| 77 |
---------------------------- |
| 78 |
|
| 79 |
:Architecture: PPC |
| 80 |
:Status: active |
| 81 |
:Purpose: To enable communication between the hypervisor and guest there is a |
| 82 |
shared page that contains parts of supervisor visible register state. |
| 83 |
The guest can map this shared page to access its supervisor register |
| 84 |
through memory using this hypercall. |
| 85 |
|
| 86 |
5. KVM_HC_KICK_CPU |
| 87 |
------------------ |
| 88 |
|
| 89 |
:Architecture: x86 |
| 90 |
:Status: active |
| 91 |
:Purpose: Hypercall used to wakeup a vcpu from HLT state |
| 92 |
:Usage example: |
| 93 |
A vcpu of a paravirtualized guest that is busywaiting in guest |
| 94 |
kernel mode for an event to occur (ex: a spinlock to become available) can |
| 95 |
execute HLT instruction once it has busy-waited for more than a threshold |
| 96 |
time-interval. Execution of HLT instruction would cause the hypervisor to put |
| 97 |
the vcpu to sleep until occurrence of an appropriate event. Another vcpu of the |
| 98 |
same guest can wakeup the sleeping vcpu by issuing KVM_HC_KICK_CPU hypercall, |
| 99 |
specifying APIC ID (a1) of the vcpu to be woken up. An additional argument (a0) |
| 100 |
is used in the hypercall for future use. |
| 101 |
|
| 102 |
|
| 103 |
6. KVM_HC_CLOCK_PAIRING |
| 104 |
----------------------- |
| 105 |
:Architecture: x86 |
| 106 |
:Status: active |
| 107 |
:Purpose: Hypercall used to synchronize host and guest clocks. |
| 108 |
|
| 109 |
Usage: |
| 110 |
|
| 111 |
a0: guest physical address where host copies |
| 112 |
"struct kvm_clock_offset" structure. |
| 113 |
|
| 114 |
a1: clock_type, ATM only KVM_CLOCK_PAIRING_WALLCLOCK (0) |
| 115 |
is supported (corresponding to the host's CLOCK_REALTIME clock). |
| 116 |
|
| 117 |
:: |
| 118 |
|
| 119 |
struct kvm_clock_pairing { |
| 120 |
__s64 sec; |
| 121 |
__s64 nsec; |
| 122 |
__u64 tsc; |
| 123 |
__u32 flags; |
| 124 |
__u32 pad[9]; |
| 125 |
}; |
| 126 |
|
| 127 |
Where: |
| 128 |
* sec: seconds from clock_type clock. |
| 129 |
* nsec: nanoseconds from clock_type clock. |
| 130 |
* tsc: guest TSC value used to calculate sec/nsec pair |
| 131 |
* flags: flags, unused (0) at the moment. |
| 132 |
|
| 133 |
The hypercall lets a guest compute a precise timestamp across |
| 134 |
host and guest. The guest can use the returned TSC value to |
| 135 |
compute the CLOCK_REALTIME for its clock, at the same instant. |
| 136 |
|
| 137 |
Returns KVM_EOPNOTSUPP if the host does not use TSC clocksource, |
| 138 |
or if clock type is different than KVM_CLOCK_PAIRING_WALLCLOCK. |
| 139 |
|
| 140 |
7. KVM_HC_SEND_IPI |
| 141 |
------------------ |
| 142 |
|
| 143 |
:Architecture: x86 |
| 144 |
:Status: active |
| 145 |
:Purpose: Send IPIs to multiple vCPUs. |
| 146 |
|
| 147 |
- a0: lower part of the bitmap of destination APIC IDs |
| 148 |
- a1: higher part of the bitmap of destination APIC IDs |
| 149 |
- a2: the lowest APIC ID in bitmap |
| 150 |
- a3: APIC ICR |
| 151 |
|
| 152 |
The hypercall lets a guest send multicast IPIs, with at most 128 |
| 153 |
128 destinations per hypercall in 64-bit mode and 64 vCPUs per |
| 154 |
hypercall in 32-bit mode. The destinations are represented by a |
| 155 |
bitmap contained in the first two arguments (a0 and a1). Bit 0 of |
| 156 |
a0 corresponds to the APIC ID in the third argument (a2), bit 1 |
| 157 |
corresponds to the APIC ID a2+1, and so on. |
| 158 |
|
| 159 |
Returns the number of CPUs to which the IPIs were delivered successfully. |
| 160 |
|
| 161 |
8. KVM_HC_SCHED_YIELD |
| 162 |
--------------------- |
| 163 |
|
| 164 |
:Architecture: x86 |
| 165 |
:Status: active |
| 166 |
:Purpose: Hypercall used to yield if the IPI target vCPU is preempted |
| 167 |
|
| 168 |
a0: destination APIC ID |
| 169 |
|
| 170 |
:Usage example: When sending a call-function IPI-many to vCPUs, yield if |
| 171 |
any of the IPI target vCPUs was preempted. |
| 172 |
|
| 173 |
9. KVM_HC_MAP_GPA_RANGE |
| 174 |
------------------------- |
| 175 |
:Architecture: x86 |
| 176 |
:Status: active |
| 177 |
:Purpose: Request KVM to map a GPA range with the specified attributes. |
| 178 |
|
| 179 |
a0: the guest physical address of the start page |
| 180 |
a1: the number of (4kb) pages (must be contiguous in GPA space) |
| 181 |
a2: attributes |
| 182 |
|
| 183 |
Where 'attributes' : |
| 184 |
* bits 3:0 - preferred page size encoding 0 = 4kb, 1 = 2mb, 2 = 1gb, etc... |
| 185 |
* bit 4 - plaintext = 0, encrypted = 1 |
| 186 |
* bits 63:5 - reserved (must be zero) |
| 187 |
|
| 188 |
**Implementation note**: this hypercall is implemented in userspace via |
| 189 |
the KVM_CAP_EXIT_HYPERCALL capability. Userspace must enable that capability |
| 190 |
before advertising KVM_FEATURE_HC_MAP_GPA_RANGE in the guest CPUID. In |
| 191 |
addition, if the guest supports KVM_FEATURE_MIGRATION_CONTROL, userspace |
| 192 |
must also set up an MSR filter to process writes to MSR_KVM_MIGRATION_CONTROL. |