arm_vcpu 0.5.5

Aarch64 VCPU implementation for Arceos Hypervisor
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
// Copyright 2025 The Axvisor Team
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//     http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

use aarch64_cpu::registers::*;
use ax_errno::AxResult;
use axaddrspace::{GuestPhysAddr, HostPhysAddr, device::SysRegAddr};
use axvcpu::{AxArchVCpu, AxVCpuExitReason};

use crate::{
    TrapFrame,
    context_frame::GuestSystemRegisters,
    exception::{TrapKind, handle_exception_sync},
    exception_utils::exception_class_value,
};

#[ax_percpu::def_percpu]
static HOST_SP_EL0: u64 = 0;

/// Save host's `SP_EL0` to the current ax-percpu region.
unsafe fn save_host_sp_el0() {
    unsafe { HOST_SP_EL0.write_current_raw(SP_EL0.get()) }
}

/// Restore host's `SP_EL0` from the current ax-percpu region.
unsafe fn restore_host_sp_el0() {
    SP_EL0.set(unsafe { HOST_SP_EL0.read_current_raw() });
}

/// (v)CPU register state that must be saved or restored when entering/exiting a VM or switching
/// between VMs.
#[repr(C)]
#[derive(Clone, Debug, Copy, Default)]
pub struct VmCpuRegisters {
    /// guest trap context
    pub trap_context_regs: TrapFrame,
    /// virtual machine system regs setting
    pub vm_system_regs: GuestSystemRegisters,
}

/// A virtual CPU within a guest
#[repr(C)]
#[derive(Debug)]
pub struct Aarch64VCpu {
    // DO NOT modify `guest_regs` and `host_stack_top` and their order unless you do know what you are doing!
    // DO NOT add anything before or between them unless you do know what you are doing!
    ctx: TrapFrame,
    host_stack_top: u64,
    guest_system_regs: GuestSystemRegisters,
    /// The MPIDR_EL1 value for the vCPU.
    mpidr: u64,
}

/// Configuration for creating a new `Aarch64VCpu`
#[derive(Clone, Debug, Default)]
pub struct Aarch64VCpuCreateConfig {
    /// The MPIDR_EL1 value for the new vCPU,
    /// which is used to identify the CPU in a multiprocessor system.
    /// Note: mind CPU cluster.
    // FIXME: Handle its interaction with the virtual GIC.
    pub mpidr_el1: u64,
    /// The address of the device tree blob.
    pub dtb_addr: usize,
}

/// Configuration for setting up a new `Aarch64VCpu`
#[derive(Clone, Debug, Default)]
pub struct Aarch64VCpuSetupConfig {
    /// Should the hypervisor passthrough interrupts to the guest?
    pub passthrough_interrupt: bool,
    /// Should the hypervisor passthrough timers to the guest?
    pub passthrough_timer: bool,
}

impl axvcpu::AxArchVCpu for Aarch64VCpu {
    type CreateConfig = Aarch64VCpuCreateConfig;

    type SetupConfig = Aarch64VCpuSetupConfig;

    fn new(_vm_id: usize, _vcpu_id: usize, config: Self::CreateConfig) -> AxResult<Self> {
        let mut ctx = TrapFrame::default();
        ctx.set_argument(config.dtb_addr);

        Ok(Self {
            ctx,
            host_stack_top: 0,
            guest_system_regs: GuestSystemRegisters::default(),
            mpidr: config.mpidr_el1,
        })
    }

    fn setup(&mut self, config: Self::SetupConfig) -> AxResult {
        self.init_hv(config);
        Ok(())
    }

    fn set_entry(&mut self, entry: GuestPhysAddr) -> AxResult {
        debug!("set vcpu entry:{entry:?}");
        self.set_elr(entry.as_usize());
        Ok(())
    }

    fn set_ept_root(&mut self, ept_root: HostPhysAddr) -> AxResult {
        debug!("set vcpu ept root:{ept_root:#x}");
        self.guest_system_regs.vttbr_el2 = ept_root.as_usize() as u64;
        Ok(())
    }

    fn run(&mut self) -> AxResult<AxVCpuExitReason> {
        unsafe {
            core::arch::asm!("msr daifset, #2");
        }

        // Run guest.
        let exit_reson = unsafe {
            // Save host SP_EL0 to the ctx becase it's used as current task ptr.
            // This has to be done before vm system regs are restored.
            save_host_sp_el0();
            self.restore_vm_system_regs();
            self.run_guest()
        };

        unsafe {
            core::arch::asm!("msr daifclr, #2");
        }

        let trap_kind = TrapKind::try_from(exit_reson as u8).expect("Invalid TrapKind");
        self.vmexit_handler(trap_kind)
    }

    fn bind(&mut self) -> AxResult {
        Ok(())
    }

    fn unbind(&mut self) -> AxResult {
        Ok(())
    }

    fn set_gpr(&mut self, idx: usize, val: usize) {
        self.ctx.set_gpr(idx, val);
    }

    fn inject_interrupt(&mut self, vector: usize) -> AxResult {
        axvisor_api::arch::hardware_inject_virtual_interrupt(vector as u8);
        Ok(())
    }

    fn set_return_value(&mut self, val: usize) {
        // Return value is stored in x0.
        self.ctx.set_argument(val);
    }
}

// Private function
impl Aarch64VCpu {
    fn init_hv(&mut self, config: Aarch64VCpuSetupConfig) {
        self.ctx.spsr = (SPSR_EL1::M::EL1h
            + SPSR_EL1::I::Masked
            + SPSR_EL1::F::Masked
            + SPSR_EL1::A::Masked
            + SPSR_EL1::D::Masked)
            .value;
        self.init_vm_context(config);
    }

    /// Init guest context. Also set some el2 register value.
    fn init_vm_context(&mut self, config: Aarch64VCpuSetupConfig) {
        // CNTHCTL_EL2.modify(CNTHCTL_EL2::EL1PCEN::SET + CNTHCTL_EL2::EL1PCTEN::SET);
        // Set CNTVOFF_EL2 to the current physical counter so the guest's
        // virtual counter (CNTVCT_EL0 = CNTPCT_EL0 - CNTVOFF_EL2) starts near zero.
        let cntpct: u64;
        unsafe { core::arch::asm!("mrs {0}, CNTPCT_EL0", out(reg) cntpct) };
        self.guest_system_regs.cntvoff_el2 = cntpct;
        self.guest_system_regs.cntkctl_el1 = 0;
        self.guest_system_regs.cnthctl_el2 = if config.passthrough_timer {
            (CNTHCTL_EL2::EL1PCEN::SET + CNTHCTL_EL2::EL1PCTEN::SET).into()
        } else {
            (CNTHCTL_EL2::EL1PCEN::CLEAR + CNTHCTL_EL2::EL1PCTEN::CLEAR).into()
        };

        self.guest_system_regs.sctlr_el1 = 0x30C50830;
        self.guest_system_regs.pmcr_el0 = 0;

        self.guest_system_regs.vtcr_el2 = probe_vtcr_support()
            + (VTCR_EL2::TG0::Granule4KB
                + VTCR_EL2::SH0::Inner
                + VTCR_EL2::ORGN0::NormalWBRAWA
                + VTCR_EL2::IRGN0::NormalWBRAWA)
                .value;

        let mut hcr_el2 =
            HCR_EL2::VM::Enable + HCR_EL2::TSC::EnableTrapEl1SmcToEl2 + HCR_EL2::RW::EL1IsAarch64;

        if !config.passthrough_interrupt {
            // Set HCR_EL2.IMO will trap IRQs to EL2 while enabling virtual IRQs.
            //
            // We must choose one of the two:
            // - Enable virtual IRQs and trap physical IRQs to EL2.
            // - Disable virtual IRQs and pass through physical IRQs to EL1.
            hcr_el2 += HCR_EL2::IMO::EnableVirtualIRQ + HCR_EL2::FMO::EnableVirtualFIQ;
        }

        self.guest_system_regs.hcr_el2 = hcr_el2.into();

        // Set VMPIDR_EL2, which provides the value of the Virtualization Multiprocessor ID.
        // This is the value returned by Non-secure EL1 reads of MPIDR.
        let mut vmpidr = 1 << 31;
        // Note: mind CPU cluster here.
        vmpidr |= self.mpidr;
        self.guest_system_regs.vmpidr_el2 = vmpidr;
    }

    /// Set exception return pc
    fn set_elr(&mut self, elr: usize) {
        self.ctx.set_exception_pc(elr);
    }

    /// Get general purpose register
    #[allow(unused)]
    fn get_gpr(&self, idx: usize) {
        self.ctx.gpr(idx);
    }
}

/// Private functions related to vcpu runtime control flow.
impl Aarch64VCpu {
    /// Save host context and run guest.
    ///
    /// When a VM-Exit happens when guest's vCpu is running,
    /// the control flow will be redirected to this function through `return_run_guest`.
    #[unsafe(naked)]
    unsafe extern "C" fn run_guest(&mut self) -> usize {
        // Fixes: https://github.com/arceos-hypervisor/arm_vcpu/issues/22
        //
        // The original issue seems to be caused by an unexpected compiler optimization that takes
        // the dummy return value `0` of `run_guest` as the actual return value. By replacing the
        // original `run_guest` with the current naked one, we eliminate the dummy code path of the
        // original version, and ensure that the compiler does not perform any unexpected return
        // value optimization.
        core::arch::naked_asm!(
            // Save host context.
            save_regs_to_stack!(),
            // Save current host stack top to `self.host_stack_top`.
            //
            // 'extern "C"' here specifies the aapcs64 calling convention, according to which
            // the first and only parameter, the pointer of self, should be in x0:
            "mov x9, sp",
            "add x0, x0, {host_stack_top_offset}",
            "str x9, [x0]",
            // Go to `context_vm_entry`.
            "b context_vm_entry",
            // Panic if the control flow comes back here, which should never happen.
            "b {run_guest_panic}",
            host_stack_top_offset = const core::mem::size_of::<TrapFrame>(),
            run_guest_panic = sym Self::run_guest_panic,
        );
    }

    /// This function is called when the control flow comes back to `run_guest`. To provide a error
    /// message for debugging purposes.
    ///
    /// This function may fail as the stack may have been corrupted when this function is called.
    /// But we won't handle it here for now.
    unsafe fn run_guest_panic() -> ! {
        panic!("run_guest_panic");
    }

    /// Restores guest system control registers.
    unsafe fn restore_vm_system_regs(&mut self) {
        unsafe {
            // load system regs
            core::arch::asm!(
                "
                mov x3, xzr           // Trap nothing from EL1 to El2.
                msr cptr_el2, x3"
            );
            self.guest_system_regs.restore();
            core::arch::asm!(
                "
                ic  iallu
                tlbi	alle2
                tlbi	alle1         // Flush tlb
                dsb	nsh
                isb"
            );
        }
    }

    /// Handle VM-Exits.
    ///
    /// Parameters:
    /// - `exit_reason`: The reason why the VM-Exit happened in [`TrapKind`].
    ///
    /// Returns:
    /// - [`AxVCpuExitReason`]: a wrappered VM-Exit reason needed to be handled by the hypervisor.
    ///
    /// This function may panic for unhandled exceptions.
    fn vmexit_handler(&mut self, exit_reason: TrapKind) -> AxResult<AxVCpuExitReason> {
        trace!(
            "Aarch64VCpu vmexit_handler() esr:{:#x} ctx:{:#x?}",
            exception_class_value(),
            self.ctx
        );

        unsafe {
            // Store guest system regs
            self.guest_system_regs.store();

            // Store guest `SP_EL0` into the `Aarch64VCpu` struct,
            // which will be restored when the guest is resumed in `exception_return_el2`.
            self.ctx.sp_el0 = self.guest_system_regs.sp_el0;

            // Restore host `SP_EL0`.
            // This has to be done after guest's SP_EL0 is stored by `ext_regs_store`.
            restore_host_sp_el0();
        }

        let result = match exit_reason {
            TrapKind::Synchronous => handle_exception_sync(&mut self.ctx),
            TrapKind::Irq => Ok(AxVCpuExitReason::ExternalInterrupt {
                vector: axvisor_api::arch::fetch_irq(),
            }),
            _ => panic!("Unhandled exception {:?}", exit_reason),
        };

        match result {
            Ok(AxVCpuExitReason::SysRegRead { addr, reg }) => {
                if let Some(exit_reason) =
                    self.builtin_sysreg_access_handler(addr, false, 0, reg)?
                {
                    return Ok(exit_reason);
                }

                result
            }
            Ok(AxVCpuExitReason::SysRegWrite { addr, value }) => {
                if let Some(exit_reason) =
                    self.builtin_sysreg_access_handler(addr, true, value, 0)?
                {
                    return Ok(exit_reason);
                }

                result
            }
            r => r,
        }
    }

    /// Handle system register access that can and should be handled by the VCpu itself.
    ///
    /// Return `Ok(None)` if the system register access is not handled by the VCpu itself,
    fn builtin_sysreg_access_handler(
        &mut self,
        addr: SysRegAddr,
        write: bool,
        value: u64,
        reg: usize,
    ) -> AxResult<Option<AxVCpuExitReason>> {
        const SYSREG_ICC_SGI1R_EL1: SysRegAddr = SysRegAddr::new(0x3A_3016); // ICC_SGI1R_EL1

        match (addr, write) {
            (SYSREG_ICC_SGI1R_EL1, true) => {
                debug!("arm_vcpu ICC_SGI1R_EL1 write: {value:#x}");

                // TODO: support RangeSelector

                let intid = (value >> 24) & 0b1111;
                let irm = ((value >> 40) & 0b1) != 0;

                // IRM == 1 => send to all except self
                if irm {
                    debug!("arm_vcpu ICC_SGI1R_EL1 write: irm == 1, send to all except self");

                    return Ok(Some(AxVCpuExitReason::SendIPI {
                        target_cpu: 0,
                        target_cpu_aux: 0,
                        send_to_all: true,
                        send_to_self: false,
                        vector: intid,
                    }));
                }

                let aff3 = (value >> 48) & 0xff;
                let aff2 = (value >> 32) & 0xff;
                let aff1 = (value >> 16) & 0xff;
                let target_list = value & 0xffff;

                debug!(
                    "arm_vcpu ICC_SGI1R_EL1 write: aff3:{aff3:#x} aff2:{aff2:#x} aff1:{aff1:#x} \
                     intid:{intid:#x} target_list:{target_list:#x}"
                );

                Ok(Some(AxVCpuExitReason::SendIPI {
                    target_cpu: (aff3 << 24) | (aff2 << 16) | (aff1 << 8),
                    target_cpu_aux: target_list,
                    send_to_all: false,
                    send_to_self: false,
                    vector: intid,
                }))
            }
            (SYSREG_ICC_SGI1R_EL1, false) => {
                // ICC_SGI1R_EL1 is WO, we take it as RAZ.
                self.set_gpr(reg, 0);
                Ok(Some(AxVCpuExitReason::Nothing))
            }
            _ => {
                // If the system register access is not handled by the VCpu itself,
                // we return None to let the hypervisor handle it.
                Ok(None)
            }
        }
    }
}

pub(crate) fn pa_bits() -> usize {
    match ID_AA64MMFR0_EL1.read_as_enum(ID_AA64MMFR0_EL1::PARange) {
        Some(ID_AA64MMFR0_EL1::PARange::Value::Bits_32) => 32,
        Some(ID_AA64MMFR0_EL1::PARange::Value::Bits_36) => 36,
        Some(ID_AA64MMFR0_EL1::PARange::Value::Bits_40) => 40,
        Some(ID_AA64MMFR0_EL1::PARange::Value::Bits_42) => 42,
        Some(ID_AA64MMFR0_EL1::PARange::Value::Bits_44) => 44,
        Some(ID_AA64MMFR0_EL1::PARange::Value::Bits_48) => 48,
        Some(ID_AA64MMFR0_EL1::PARange::Value::Bits_52) => 52,
        _ => 32,
    }
}

#[allow(dead_code)]
pub(crate) fn current_gpt_level() -> usize {
    let t0sz = VTCR_EL2.read(VTCR_EL2::T0SZ) as usize;
    match t0sz {
        16..=25 => 4,
        26..=35 => 3,
        _ => 2,
    }
}

pub(crate) fn max_gpt_level(pa_bits: usize) -> usize {
    match pa_bits {
        44.. => 4,
        _ => 3,
    }
}

fn probe_vtcr_support() -> u64 {
    let pa_bits = pa_bits();

    let mut val = match max_gpt_level(pa_bits) {
        4 => VTCR_EL2::SL0::Granule4KBLevel0 + VTCR_EL2::T0SZ.val(64 - 48),
        _ => VTCR_EL2::SL0::Granule4KBLevel1 + VTCR_EL2::T0SZ.val(64 - 39),
    };

    match pa_bits {
        52..=64 => val += VTCR_EL2::PS::PA_52B_4PB,
        48..=51 => val += VTCR_EL2::PS::PA_48B_256TB,
        44..=47 => val += VTCR_EL2::PS::PA_44B_16TB,
        42..=43 => val += VTCR_EL2::PS::PA_42B_4TB,
        40..=41 => val += VTCR_EL2::PS::PA_40B_1TB,
        36..=39 => val += VTCR_EL2::PS::PA_36B_64GB,
        _ => val += VTCR_EL2::PS::PA_32B_4GB,
    }

    val.value
}