1 ARM Virtual Generic Interrupt Controller v3 and later (VGICv3)
2 ==============================================================
5 Device types supported:
6 KVM_DEV_TYPE_ARM_VGIC_V3 ARM Generic Interrupt Controller v3.0
8 Only one VGIC instance may be instantiated through this API. The created VGIC
9 will act as the VM interrupt controller, requiring emulated user-space devices
10 to inject interrupts to the VGIC instead of directly to CPUs. It is not
11 possible to create both a GICv3 and GICv2 on the same VM.
13 Creating a guest GICv3 device requires a host GICv3 as well.
17 KVM_DEV_ARM_VGIC_GRP_ADDR
19 KVM_VGIC_V3_ADDR_TYPE_DIST (rw, 64-bit)
20 Base address in the guest physical address space of the GICv3 distributor
21 register mappings. Only valid for KVM_DEV_TYPE_ARM_VGIC_V3.
22 This address needs to be 64K aligned and the region covers 64 KByte.
24 KVM_VGIC_V3_ADDR_TYPE_REDIST (rw, 64-bit)
25 Base address in the guest physical address space of the GICv3
26 redistributor register mappings. There are two 64K pages for each
27 VCPU and all of the redistributor pages are contiguous.
28 Only valid for KVM_DEV_TYPE_ARM_VGIC_V3.
29 This address needs to be 64K aligned.
31 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION (rw, 64-bit)
32 The attribute data pointed to by kvm_device_attr.addr is a __u64 value:
33 bits: | 63 .... 52 | 51 .... 16 | 15 - 12 |11 - 0
34 values: | count | base | flags | index
35 - index encodes the unique redistributor region index
36 - flags: reserved for future use, currently 0
37 - base field encodes bits [51:16] of the guest physical base address
38 of the first redistributor in the region.
39 - count encodes the number of redistributors in the region. Must be
41 There are two 64K pages for each redistributor in the region and
42 redistributors are laid out contiguously within the region. Regions
43 are filled with redistributors in the index order. The sum of all
44 region count fields must be greater than or equal to the number of
45 VCPUs. Redistributor regions must be registered in the incremental
46 index order, starting from index 0.
47 The characteristics of a specific redistributor region can be read
48 by presetting the index field in the attr data.
49 Only valid for KVM_DEV_TYPE_ARM_VGIC_V3.
51 It is invalid to mix calls with KVM_VGIC_V3_ADDR_TYPE_REDIST and
52 KVM_VGIC_V3_ADDR_TYPE_REDIST_REGION attributes.
55 -E2BIG: Address outside of addressable IPA range
56 -EINVAL: Incorrectly aligned address, bad redistributor region
57 count/index, mixed redistributor region attribute usage
58 -EEXIST: Address already configured
59 -ENOENT: Attempt to read the characteristics of a non existing
61 -ENXIO: The group or attribute is unknown/unsupported for this device
62 or hardware support is missing.
63 -EFAULT: Invalid user pointer for attr->addr.
66 KVM_DEV_ARM_VGIC_GRP_DIST_REGS
67 KVM_DEV_ARM_VGIC_GRP_REDIST_REGS
69 The attr field of kvm_device_attr encodes two values:
70 bits: | 63 .... 32 | 31 .... 0 |
71 values: | mpidr | offset |
73 All distributor regs are (rw, 32-bit) and kvm_device_attr.addr points to a
74 __u32 value. 64-bit registers must be accessed by separately accessing the
75 lower and higher word.
77 Writes to read-only registers are ignored by the kernel.
79 KVM_DEV_ARM_VGIC_GRP_DIST_REGS accesses the main distributor registers.
80 KVM_DEV_ARM_VGIC_GRP_REDIST_REGS accesses the redistributor of the CPU
81 specified by the mpidr.
83 The offset is relative to the "[Re]Distributor base address" as defined
84 in the GICv3/4 specs. Getting or setting such a register has the same
85 effect as reading or writing the register on real hardware, except for the
86 following registers: GICD_STATUSR, GICR_STATUSR, GICD_ISPENDR,
87 GICR_ISPENDR0, GICD_ICPENDR, and GICR_ICPENDR0. These registers behave
88 differently when accessed via this interface compared to their
89 architecturally defined behavior to allow software a full view of the
90 VGIC's internal state.
92 The mpidr field is used to specify which
93 redistributor is accessed. The mpidr is ignored for the distributor.
95 The mpidr encoding is based on the affinity information in the
96 architecture defined MPIDR, and the field is encoded as follows:
97 | 63 .... 56 | 55 .... 48 | 47 .... 40 | 39 .... 32 |
98 | Aff3 | Aff2 | Aff1 | Aff0 |
100 Note that distributor fields are not banked, but return the same value
101 regardless of the mpidr used to access the register.
103 GICD_IIDR.Revision is updated when the KVM implementation is changed in a
104 way directly observable by the guest or userspace. Userspace should read
105 GICD_IIDR from KVM and write back the read value to confirm its expected
106 behavior is aligned with the KVM implementation. Userspace should set
107 GICD_IIDR before setting any other registers to ensure the expected
111 The GICD_STATUSR and GICR_STATUSR registers are architecturally defined such
112 that a write of a clear bit has no effect, whereas a write with a set bit
113 clears that value. To allow userspace to freely set the values of these two
114 registers, setting the attributes with the register offsets for these two
115 registers simply sets the non-reserved bits to the value written.
118 Accesses (reads and writes) to the GICD_ISPENDR register region and
119 GICR_ISPENDR0 registers get/set the value of the latched pending state for
122 This is identical to the value returned by a guest read from ISPENDR for an
123 edge triggered interrupt, but may differ for level triggered interrupts.
124 For edge triggered interrupts, once an interrupt becomes pending (whether
125 because of an edge detected on the input line or because of a guest write
126 to ISPENDR) this state is "latched", and only cleared when either the
127 interrupt is activated or when the guest writes to ICPENDR. A level
128 triggered interrupt may be pending either because the level input is held
129 high by a device, or because of a guest write to the ISPENDR register. Only
130 ISPENDR writes are latched; if the device lowers the line level then the
131 interrupt is no longer pending unless the guest also wrote to ISPENDR, and
132 conversely writes to ICPENDR or activations of the interrupt do not clear
133 the pending status if the line level is still being held high. (These
134 rules are documented in the GICv3 specification descriptions of the ICPENDR
135 and ISPENDR registers.) For a level triggered interrupt the value accessed
136 here is that of the latch which is set by ISPENDR and cleared by ICPENDR or
137 interrupt activation, whereas the value returned by a guest read from
138 ISPENDR is the logical OR of the latch value and the input line level.
140 Raw access to the latch state is provided to userspace so that it can save
141 and restore the entire GIC internal state (which is defined by the
142 combination of the current input line level and the latch state, and cannot
143 be deduced from purely the line level and the value of the ISPENDR
146 Accesses to GICD_ICPENDR register region and GICR_ICPENDR0 registers have
147 RAZ/WI semantics, meaning that reads always return 0 and writes are always
151 -ENXIO: Getting or setting this register is not yet supported
152 -EBUSY: One or more VCPUs are running
155 KVM_DEV_ARM_VGIC_GRP_CPU_SYSREGS
157 The attr field of kvm_device_attr encodes two values:
158 bits: | 63 .... 32 | 31 .... 16 | 15 .... 0 |
159 values: | mpidr | RES | instr |
161 The mpidr field encodes the CPU ID based on the affinity information in the
162 architecture defined MPIDR, and the field is encoded as follows:
163 | 63 .... 56 | 55 .... 48 | 47 .... 40 | 39 .... 32 |
164 | Aff3 | Aff2 | Aff1 | Aff0 |
166 The instr field encodes the system register to access based on the fields
167 defined in the A64 instruction set encoding for system register access
168 (RES means the bits are reserved for future use and should be zero):
170 | 15 ... 14 | 13 ... 11 | 10 ... 7 | 6 ... 3 | 2 ... 0 |
171 | Op 0 | Op1 | CRn | CRm | Op2 |
173 All system regs accessed through this API are (rw, 64-bit) and
174 kvm_device_attr.addr points to a __u64 value.
176 KVM_DEV_ARM_VGIC_GRP_CPU_SYSREGS accesses the CPU interface registers for the
177 CPU specified by the mpidr field.
179 CPU interface registers access is not implemented for AArch32 mode.
180 Error -ENXIO is returned when accessed in AArch32 mode.
182 -ENXIO: Getting or setting this register is not yet supported
183 -EBUSY: VCPU is running
184 -EINVAL: Invalid mpidr or register value supplied
187 KVM_DEV_ARM_VGIC_GRP_NR_IRQS
189 A value describing the number of interrupts (SGI, PPI and SPI) for
190 this GIC instance, ranging from 64 to 1024, in increments of 32.
192 kvm_device_attr.addr points to a __u32 value.
195 -EINVAL: Value set is out of the expected range
196 -EBUSY: Value has already be set.
199 KVM_DEV_ARM_VGIC_GRP_CTRL
201 KVM_DEV_ARM_VGIC_CTRL_INIT
202 request the initialization of the VGIC, no additional parameter in
203 kvm_device_attr.addr.
204 KVM_DEV_ARM_VGIC_SAVE_PENDING_TABLES
205 save all LPI pending bits into guest RAM pending tables.
207 The first kB of the pending table is not altered by this operation.
209 -ENXIO: VGIC not properly configured as required prior to calling
211 -ENODEV: no online VCPU
212 -ENOMEM: memory shortage when allocating vgic internal data
213 -EFAULT: Invalid guest ram access
214 -EBUSY: One or more VCPUS are running
217 KVM_DEV_ARM_VGIC_GRP_LEVEL_INFO
219 The attr field of kvm_device_attr encodes the following values:
220 bits: | 63 .... 32 | 31 .... 10 | 9 .... 0 |
221 values: | mpidr | info | vINTID |
223 The vINTID specifies which set of IRQs is reported on.
225 The info field specifies which information userspace wants to get or set
226 using this interface. Currently we support the following info values:
228 VGIC_LEVEL_INFO_LINE_LEVEL:
229 Get/Set the input level of the IRQ line for a set of 32 contiguously
231 vINTID must be a multiple of 32.
233 kvm_device_attr.addr points to a __u32 value which will contain a
234 bitmap where a set bit means the interrupt level is asserted.
236 Bit[n] indicates the status for interrupt vINTID + n.
238 SGIs and any interrupt with a higher ID than the number of interrupts
239 supported, will be RAZ/WI. LPIs are always edge-triggered and are
240 therefore not supported by this interface.
242 PPIs are reported per VCPU as specified in the mpidr field, and SPIs are
243 reported with the same value regardless of the mpidr specified.
245 The mpidr field encodes the CPU ID based on the affinity information in the
246 architecture defined MPIDR, and the field is encoded as follows:
247 | 63 .... 56 | 55 .... 48 | 47 .... 40 | 39 .... 32 |
248 | Aff3 | Aff2 | Aff1 | Aff0 |
250 -EINVAL: vINTID is not multiple of 32 or
251 info field is not VGIC_LEVEL_INFO_LINE_LEVEL