kvm: Add basic support for ARM

Architecture specific limitations:
 * LPAE is currently not supported by gem5. We therefore panic if LPAE
   is enabled when returning to gem5.
 * The co-processor based interface to the architected timer is
   unsupported. We can't support this due to limitations in the KVM
   API on ARM.
 * M5 ops are currently not supported. This requires either a kernel
   hack or a memory mapped device that handles the guest<->m5
   interface.

4 files changed, 913 insertions, 0 deletions

diff --git a/src/cpu/kvm/ArmKvmCPU.py b/src/cpu/kvm/ArmKvmCPU.py
new file mode 100644
index 000000000..2bdc05820
--- /dev/null
+++ b/src/cpu/kvm/ArmKvmCPU.py
@@ -0,0 +1,43 @@
+# Copyright (c) 2012 ARM Limited
+# All rights reserved.
+#
+# The license below extends only to copyright in the software and shall
+# not be construed as granting a license to any other intellectual
+# property including but not limited to intellectual property relating
+# to a hardware implementation of the functionality of the software
+# licensed hereunder.  You may use the software subject to the license
+# terms below provided that you ensure that this notice is replicated
+# unmodified and in its entirety in all distributions of the software,
+# modified or unmodified, in source code or in binary form.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met: redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer;
+# redistributions in binary form must reproduce the above copyright
+# notice, this list of conditions and the following disclaimer in the
+# documentation and/or other materials provided with the distribution;
+# neither the name of the copyright holders nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+# Authors: Andreas Sandberg
+
+from m5.params import *
+from BaseKvmCPU import BaseKvmCPU
+
+class ArmKvmCPU(BaseKvmCPU):
+    type = 'ArmKvmCPU'
+    cxx_header = "cpu/kvm/arm_cpu.hh"
diff --git a/src/cpu/kvm/SConscript b/src/cpu/kvm/SConscript
index a567720fa..56cc660cb 100644
--- a/src/cpu/kvm/SConscript
+++ b/src/cpu/kvm/SConscript
@@ -48,6 +48,10 @@ if env['USE_KVM']:
     Source('perfevent.cc')
     Source('timer.cc')
 
+    if env['TARGET_ISA'] == 'arm':
+        SimObject('ArmKvmCPU.py')
+        Source('arm_cpu.cc')
+
     DebugFlag('Kvm', 'Basic KVM Functionality')
     DebugFlag('KvmContext', 'KVM/gem5 context synchronization')
     DebugFlag('KvmIO', 'KVM MMIO diagnostics')
diff --git a/src/cpu/kvm/arm_cpu.cc b/src/cpu/kvm/arm_cpu.cc
new file mode 100644
index 000000000..92851520f
--- /dev/null
+++ b/src/cpu/kvm/arm_cpu.cc
@@ -0,0 +1,705 @@
+/*
+ * Copyright (c) 2012 ARM Limited
+ * All rights reserved
+ *
+ * The license below extends only to copyright in the software and shall
+ * not be construed as granting a license to any other intellectual
+ * property including but not limited to intellectual property relating
+ * to a hardware implementation of the functionality of the software
+ * licensed hereunder.  You may use the software subject to the license
+ * terms below provided that you ensure that this notice is replicated
+ * unmodified and in its entirety in all distributions of the software,
+ * modified or unmodified, in source code or in binary form.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met: redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer;
+ * redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution;
+ * neither the name of the copyright holders nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * Authors: Andreas Sandberg
+ */
+
+#include <linux/kvm.h>
+
+#include <algorithm>
+#include <cerrno>
+#include <memory>
+
+#include "arch/registers.hh"
+#include "cpu/kvm/arm_cpu.hh"
+#include "cpu/kvm/base.hh"
+#include "debug/Kvm.hh"
+#include "debug/KvmContext.hh"
+#include "debug/KvmInt.hh"
+
+using namespace ArmISA;
+
+#define EXTRACT_FIELD(val, mask, shift)         \
+    (((val) & (mask)) >> (shift))
+
+#define REG_IS_ARM(id)                          \
+    (((id) & KVM_REG_ARCH_MASK) == KVM_REG_ARM)
+
+#define REG_IS_32BIT(id)                                \
+    (((id) & KVM_REG_SIZE_MASK) == KVM_REG_SIZE_U32)
+
+#define REG_IS_64BIT(id)                                \
+    (((id) & KVM_REG_SIZE_MASK) == KVM_REG_SIZE_U64)
+
+#define REG_IS_CP(id, cp)                       \
+    (((id) & KVM_REG_ARM_COPROC_MASK) == (cp))
+
+#define REG_IS_CORE(id) REG_IS_CP((id), KVM_REG_ARM_CORE)
+
+#define REG_IS_VFP(id) REG_IS_CP((id), KVM_REG_ARM_VFP)
+#define REG_VFP_REG(id) ((id) & KVM_REG_ARM_VFP_MASK)
+// HACK: These aren't really defined in any of the headers, so we'll
+// assume some reasonable values for now.
+#define REG_IS_VFP_REG(id) (REG_VFP_REG(id) < 0x100)
+#define REG_IS_VFP_CTRL(id) (REG_VFP_REG(id) >= 0x100)
+
+#define REG_IS_DEMUX(id) REG_IS_CP((id), KVM_REG_ARM_DEMUX)
+
+
+// There is no constant in the kernel headers defining the mask to use
+// to get the core register index. We'll just do what they do
+// internally.
+#define REG_CORE_IDX(id)                                                \
+    (~(KVM_REG_ARCH_MASK | KVM_REG_SIZE_MASK | KVM_REG_ARM_CORE))
+
+#define REG_CP(id)                                                      \
+    EXTRACT_FIELD(id, KVM_REG_ARM_COPROC_MASK, KVM_REG_ARM_COPROC_SHIFT)
+
+#define REG_CRN(id)                                                     \
+    EXTRACT_FIELD(id, KVM_REG_ARM_32_CRN_MASK, KVM_REG_ARM_32_CRN_SHIFT)
+
+#define REG_OPC1(id)                                                    \
+    EXTRACT_FIELD(id, KVM_REG_ARM_OPC1_MASK, KVM_REG_ARM_OPC1_SHIFT)
+
+#define REG_CRM(id)                                                     \
+    EXTRACT_FIELD(id, KVM_REG_ARM_CRM_MASK, KVM_REG_ARM_CRM_SHIFT)
+
+#define REG_OPC2(id)                                                    \
+    EXTRACT_FIELD(id, KVM_REG_ARM_32_OPC2_MASK, KVM_REG_ARM_32_OPC2_SHIFT)
+
+#define REG_CP32(cpnum, crn, opc1, crm, opc2) ( \
+        (KVM_REG_ARM | KVM_REG_SIZE_U32) |      \
+        ((cpnum) << KVM_REG_ARM_COPROC_SHIFT) | \
+        ((crn) << KVM_REG_ARM_32_CRN_SHIFT) |   \
+        ((opc1) << KVM_REG_ARM_OPC1_SHIFT) |    \
+        ((crm) << KVM_REG_ARM_CRM_SHIFT) |      \
+        ((opc2) << KVM_REG_ARM_32_OPC2_SHIFT))
+
+#define REG_CP64(cpnum, opc1, crm) (            \
+        (KVM_REG_ARM | KVM_REG_SIZE_U64) |      \
+        ((cpnum) << KVM_REG_ARM_COPROC_SHIFT) | \
+        ((opc1) << KVM_REG_ARM_OPC1_SHIFT) |    \
+        ((crm) << KVM_REG_ARM_CRM_SHIFT))
+
+#define REG_CORE32(kname) (                     \
+        (KVM_REG_ARM | KVM_REG_SIZE_U32) |      \
+        (KVM_REG_ARM_CORE) |                    \
+        (KVM_REG_ARM_CORE_REG(kname)))
+
+#define REG_VFP32(regno) (                      \
+        (KVM_REG_ARM | KVM_REG_SIZE_U32) |      \
+        KVM_REG_ARM_VFP | (regno))
+
+#define REG_VFP64(regno) (                      \
+        (KVM_REG_ARM | KVM_REG_SIZE_U64) |      \
+        KVM_REG_ARM_VFP | (regno))
+
+#define REG_DEMUX32(dmxid, val) (               \
+        (KVM_REG_ARM | KVM_REG_SIZE_U32) |      \
+        (dmxid) | (val))
+
+// Some of the co-processor registers are invariants and must have the
+// same value on both the host and the guest. We need to keep a list
+// of these to prevent gem5 from fiddling with them on the guest.
+static uint64_t invariant_reg_vector[] = {
+    REG_CP32(15, 0, 0, 0, 0), // MIDR
+    REG_CP32(15, 0, 0, 0, 1), // CTR
+    REG_CP32(15, 0, 0, 0, 2), // TCMTR
+    REG_CP32(15, 0, 0, 0, 3), // TLBTR
+    REG_CP32(15, 0, 0, 0, 6), // REVIDR
+
+    REG_CP32(15, 0, 0, 1, 0), // ID_PFR0
+    REG_CP32(15, 0, 0, 1, 1), // ID_PFR1
+    REG_CP32(15, 0, 0, 1, 2), // ID_DFR0
+    REG_CP32(15, 0, 0, 1, 3), // ID_AFR0
+    REG_CP32(15, 0, 0, 1, 4), // ID_MMFR0
+    REG_CP32(15, 0, 0, 1, 5), // ID_MMFR1
+    REG_CP32(15, 0, 0, 1, 6), // ID_MMFR2
+    REG_CP32(15, 0, 0, 1, 7), // ID_MMFR3
+
+    REG_CP32(15, 0, 0, 2, 0), // ID_ISAR0
+    REG_CP32(15, 0, 0, 2, 1), // ID_ISAR1
+    REG_CP32(15, 0, 0, 2, 2), // ID_ISAR2
+    REG_CP32(15, 0, 0, 2, 3), // ID_ISAR3
+    REG_CP32(15, 0, 0, 2, 4), // ID_ISAR4
+    REG_CP32(15, 0, 0, 2, 5), // ID_ISAR5
+
+    REG_CP32(15, 0, 1, 0, 0), // CSSIDR
+    REG_CP32(15, 0, 1, 0, 1), // CLIDR
+    REG_CP32(15, 0, 1, 0, 7), // AIDR
+
+    REG_VFP32(KVM_REG_ARM_VFP_MVFR0),
+    REG_VFP32(KVM_REG_ARM_VFP_MVFR1),
+    REG_VFP32(KVM_REG_ARM_VFP_FPSID),
+
+    REG_DEMUX32(KVM_REG_ARM_DEMUX_ID_CCSIDR, 0),
+};
+
+const static uint64_t KVM_REG64_TTBR0(REG_CP64(15, 0, 2));
+const static uint64_t KVM_REG64_TTBR1(REG_CP64(15, 1, 2));
+
+#define INTERRUPT_ID(type, vcpu, irq) (                    \
+        ((type) << KVM_ARM_IRQ_TYPE_SHIFT) |               \
+        ((vcpu) << KVM_ARM_IRQ_VCPU_SHIFT) |               \
+        ((irq) << KVM_ARM_IRQ_NUM_SHIFT))
+
+#define INTERRUPT_VCPU_IRQ(vcpu)                                \
+    INTERRUPT_ID(KVM_ARM_IRQ_TYPE_CPU, vcpu, KVM_ARM_IRQ_CPU_IRQ)
+
+#define INTERRUPT_VCPU_FIQ(vcpu)                                \
+    INTERRUPT_ID(KVM_ARM_IRQ_TYPE_CPU, vcpu, KVM_ARM_IRQ_CPU_FIQ)
+
+
+#define COUNT_OF(l) (sizeof(l) / sizeof(*l))
+
+const std::set<uint64_t> ArmKvmCPU::invariant_regs(
+    invariant_reg_vector,
+    invariant_reg_vector + COUNT_OF(invariant_reg_vector));
+
+
+ArmKvmCPU::KvmIntRegInfo ArmKvmCPU::kvmIntRegs[] = {
+    { REG_CORE32(usr_regs.ARM_r0), INTREG_R0, "R0" },
+    { REG_CORE32(usr_regs.ARM_r1), INTREG_R1, "R1" },
+    { REG_CORE32(usr_regs.ARM_r2), INTREG_R2, "R2" },
+    { REG_CORE32(usr_regs.ARM_r3), INTREG_R3, "R3" },
+    { REG_CORE32(usr_regs.ARM_r4), INTREG_R4, "R4" },
+    { REG_CORE32(usr_regs.ARM_r5), INTREG_R5, "R5" },
+    { REG_CORE32(usr_regs.ARM_r6), INTREG_R6, "R6" },
+    { REG_CORE32(usr_regs.ARM_r7), INTREG_R7, "R7" },
+    { REG_CORE32(usr_regs.ARM_r8), INTREG_R8, "R8" },
+    { REG_CORE32(usr_regs.ARM_r9), INTREG_R9, "R9" },
+    { REG_CORE32(usr_regs.ARM_r10), INTREG_R10, "R10" },
+    { REG_CORE32(usr_regs.ARM_fp), INTREG_R11, "R11" },
+    { REG_CORE32(usr_regs.ARM_ip), INTREG_R12, "R12" },
+    { REG_CORE32(usr_regs.ARM_sp), INTREG_R13, "R13(USR)" },
+    { REG_CORE32(usr_regs.ARM_lr), INTREG_R14, "R14(USR)" },
+
+    { REG_CORE32(svc_regs[0]), INTREG_SP_SVC, "R13(SVC)" },
+    { REG_CORE32(svc_regs[1]), INTREG_LR_SVC, "R14(SVC)" },
+
+    { REG_CORE32(abt_regs[0]), INTREG_SP_ABT, "R13(ABT)" },
+    { REG_CORE32(abt_regs[1]), INTREG_LR_ABT, "R14(ABT)" },
+
+    { REG_CORE32(und_regs[0]), INTREG_SP_UND, "R13(UND)" },
+    { REG_CORE32(und_regs[1]), INTREG_LR_UND, "R14(UND)" },
+
+    { REG_CORE32(irq_regs[0]), INTREG_SP_IRQ, "R13(IRQ)" },
+    { REG_CORE32(irq_regs[1]), INTREG_LR_IRQ, "R14(IRQ)" },
+
+
+    { REG_CORE32(fiq_regs[0]), INTREG_R8_FIQ, "R8(FIQ)" },
+    { REG_CORE32(fiq_regs[1]), INTREG_R9_FIQ, "R9(FIQ)" },
+    { REG_CORE32(fiq_regs[2]), INTREG_R10_FIQ, "R10(FIQ)" },
+    { REG_CORE32(fiq_regs[3]), INTREG_R11_FIQ, "R11(FIQ)" },
+    { REG_CORE32(fiq_regs[4]), INTREG_R12_FIQ, "R12(FIQ)" },
+    { REG_CORE32(fiq_regs[5]), INTREG_R13_FIQ, "R13(FIQ)" },
+    { REG_CORE32(fiq_regs[6]), INTREG_R14_FIQ, "R14(FIQ)" },
+    { 0, NUM_INTREGS, NULL }
+};
+
+ArmKvmCPU::KvmCoreMiscRegInfo ArmKvmCPU::kvmCoreMiscRegs[] = {
+    { REG_CORE32(usr_regs.ARM_cpsr), MISCREG_CPSR, "CPSR" },
+    { REG_CORE32(svc_regs[2]), MISCREG_SPSR_SVC, "SPSR(SVC)" },
+    { REG_CORE32(abt_regs[2]), MISCREG_SPSR_ABT, "SPSR(ABT)" },
+    { REG_CORE32(und_regs[2]), MISCREG_SPSR_UND, "SPSR(UND)" },
+    { REG_CORE32(irq_regs[2]), MISCREG_SPSR_IRQ, "SPSR(IRQ)" },
+    { REG_CORE32(fiq_regs[2]), MISCREG_SPSR_FIQ, "SPSR(FIQ)" },
+    { 0, NUM_MISCREGS }
+};
+
+ArmKvmCPU::ArmKvmCPU(ArmKvmCPUParams *params)
+    : BaseKvmCPU(params),
+      irqAsserted(false), fiqAsserted(false)
+{
+}
+
+ArmKvmCPU::~ArmKvmCPU()
+{
+}
+
+void
+ArmKvmCPU::startup()
+{
+    BaseKvmCPU::startup();
+
+    /* TODO: This needs to be moved when we start to support VMs with
+     * multiple threads since kvmArmVCpuInit requires that all CPUs in
+     * the VM have been created.
+     */
+    /* TODO: The CPU type needs to be configurable once KVM on ARM
+     * starts to support more CPUs.
+     */
+    kvmArmVCpuInit(KVM_ARM_TARGET_CORTEX_A15);
+}
+
+void
+ArmKvmCPU::tick()
+{
+    bool simFIQ(interrupts->checkRaw(INT_FIQ));
+    bool simIRQ(interrupts->checkRaw(INT_IRQ));
+
+    if (fiqAsserted != simFIQ) {
+        fiqAsserted = simFIQ;
+        DPRINTF(KvmInt, "KVM: Update FIQ state: %i\n", simFIQ);
+        vm.setIRQLine(INTERRUPT_VCPU_FIQ(vcpuID), simFIQ);
+    }
+    if (irqAsserted != simIRQ) {
+        irqAsserted = simIRQ;
+        DPRINTF(KvmInt, "KVM: Update IRQ state: %i\n", simIRQ);
+        vm.setIRQLine(INTERRUPT_VCPU_IRQ(vcpuID), simIRQ);
+    }
+
+    BaseKvmCPU::tick();
+}
+
+void
+ArmKvmCPU::updateKvmState()
+{
+    DPRINTF(KvmContext, "Updating KVM state...\n");
+
+    updateKvmStateCore();
+    updateKvmStateMisc();
+}
+
+void
+ArmKvmCPU::updateThreadContext()
+{
+    DPRINTF(KvmContext, "Updating gem5 state...\n");
+
+    updateTCStateCore();
+    updateTCStateMisc();
+}
+
+const ArmKvmCPU::RegIndexVector &
+ArmKvmCPU::getRegList() const
+{
+    if (_regIndexList.size() == 0) {
+        std::unique_ptr<struct kvm_reg_list> regs;
+        uint64_t i(1);
+
+        do {
+            i <<= 1;
+            regs.reset((struct kvm_reg_list *)
+                       operator new(sizeof(struct kvm_reg_list) +
+                                    i * sizeof(uint64_t)));
+            regs->n = i;
+        } while (!getRegList(*regs));
+        _regIndexList.assign(regs->reg,
+                             regs->reg + regs->n);
+    }
+
+    return _regIndexList;
+}
+
+void
+ArmKvmCPU::kvmArmVCpuInit(uint32_t target)
+{
+    struct kvm_vcpu_init init;
+
+    memset(&init, 0, sizeof(init));
+
+    init.target = target;
+
+    kvmArmVCpuInit(init);
+}
+
+void
+ArmKvmCPU::kvmArmVCpuInit(const struct kvm_vcpu_init &init)
+{
+    if (ioctl(KVM_ARM_VCPU_INIT, (void *)&init) == -1)
+        panic("KVM: Failed to initialize vCPU\n");
+}
+
+MiscRegIndex
+ArmKvmCPU::decodeCoProcReg(uint64_t id) const
+{
+    const unsigned cp(REG_CP(id));
+    const bool is_reg32(REG_IS_32BIT(id));
+    const bool is_reg64(REG_IS_64BIT(id));
+
+    // CP numbers larger than 15 are reserved for KVM extensions
+    if (cp > 15)
+        return NUM_MISCREGS;
+
+    const unsigned crm(REG_CRM(id));
+    const unsigned crn(REG_CRN(id));
+    const unsigned opc1(REG_OPC1(id));
+    const unsigned opc2(REG_OPC2(id));
+
+    if (is_reg32) {
+        switch (cp) {
+          case 14:
+            return decodeCP14Reg(crn, opc1, crm, opc2);
+
+          case 15:
+            return decodeCP15Reg(crn, opc1, crm, opc2);
+
+          default:
+            return NUM_MISCREGS;
+        }
+    } else if(is_reg64) {
+        return NUM_MISCREGS;
+    } else {
+        warn("Unhandled register length, register (0x%x) ignored.\n");
+        return NUM_MISCREGS;
+    }
+}
+
+ArmISA::MiscRegIndex
+ArmKvmCPU::decodeVFPCtrlReg(uint64_t id) const
+{
+    if (!REG_IS_ARM(id) || !REG_IS_VFP(id) || !REG_IS_VFP_CTRL(id))
+        return NUM_MISCREGS;
+
+    const unsigned vfp_reg(REG_VFP_REG(id));
+    switch (vfp_reg) {
+      case KVM_REG_ARM_VFP_FPSID: return MISCREG_FPSID;
+      case KVM_REG_ARM_VFP_FPSCR: return MISCREG_FPSCR;
+      case KVM_REG_ARM_VFP_MVFR0: return MISCREG_MVFR0;
+      case KVM_REG_ARM_VFP_MVFR1: return MISCREG_MVFR1;
+      case KVM_REG_ARM_VFP_FPEXC: return MISCREG_FPEXC;
+
+      case KVM_REG_ARM_VFP_FPINST:
+      case KVM_REG_ARM_VFP_FPINST2:
+        warn_once("KVM: FPINST not implemented.\n");
+        return NUM_MISCREGS;
+
+      default:
+        return NUM_MISCREGS;
+    }
+}
+
+bool
+ArmKvmCPU::isInvariantReg(uint64_t id)
+{
+    /* Mask away the value field from multiplexed registers, we assume
+     * that entire groups of multiplexed registers can be treated as
+     * invariant. */
+    if (REG_IS_ARM(id) && REG_IS_DEMUX(id))
+        id &= ~KVM_REG_ARM_DEMUX_VAL_MASK;
+
+    return invariant_regs.find(id) != invariant_regs.end();
+}
+
+bool
+ArmKvmCPU::getRegList(struct kvm_reg_list &regs) const
+{
+    if (ioctl(KVM_GET_REG_LIST, (void *)&regs) == -1) {
+        if (errno == E2BIG) {
+            return false;
+        } else {
+            panic("KVM: Failed to get vCPU register list (errno: %i)\n",
+                  errno);
+        }
+    } else {
+        return true;
+    }
+}
+
+void
+ArmKvmCPU::updateKvmStateCore()
+{
+    for (const KvmIntRegInfo *ri(kvmIntRegs);
+         ri->idx != NUM_INTREGS; ++ri) {
+
+        uint64_t value(tc->readIntRegFlat(ri->idx));
+        DPRINTF(KvmContext, "kvm(%s) := 0x%x\n", ri->name, value);
+        setOneReg(ri->id, value);
+    }
+
+    DPRINTF(KvmContext, "kvm(PC) := 0x%x\n", tc->instAddr());
+    setOneReg(REG_CORE32(usr_regs.ARM_pc), tc->instAddr());
+
+    for (const KvmCoreMiscRegInfo *ri(kvmCoreMiscRegs);
+         ri->idx != NUM_MISCREGS; ++ri) {
+
+        uint64_t value(tc->readMiscReg(ri->idx));
+        DPRINTF(KvmContext, "kvm(%s) := 0x%x\n", ri->name, value);
+        setOneReg(ri->id, value);
+    }
+}
+
+void
+ArmKvmCPU::updateKvmStateMisc()
+{
+    static bool warned(false); // We can't use warn_once since we want
+                               // to show /all/ registers
+
+    const RegIndexVector &regs(getRegList());
+
+    for (RegIndexVector::const_iterator it(regs.begin());
+         it != regs.end();
+         ++it) {
+
+        if (!REG_IS_ARM(*it)) {
+            if (!warned)
+                warn("Skipping non-ARM register: 0x%x\n", *it);
+        } else if (isInvariantReg(*it)) {
+            DPRINTF(Kvm, "Skipping invariant register: 0x%x\n", *it);
+        } else if (REG_IS_CORE(*it)) {
+            // Core registers are handled in updateKvmStateCore
+            continue;
+        } else if (REG_CP(*it) <= 15) {
+            updateKvmStateCoProc(*it, !warned);
+        } else if (REG_IS_VFP(*it)) {
+            updateKvmStateVFP(*it, !warned);
+        } else {
+            if (!warned) {
+                warn("Skipping register with unknown CP (%i) id: 0x%x\n",
+                     REG_CP(*it), *it);
+            }
+        }
+
+    }
+
+    warned = true;
+}
+
+void
+ArmKvmCPU::updateKvmStateCoProc(uint64_t id, bool show_warnings)
+{
+    MiscRegIndex reg(decodeCoProcReg(id));
+
+    assert(REG_IS_ARM(id));
+    assert(REG_CP(id) <= 15);
+
+    if (id == KVM_REG64_TTBR0 || id == KVM_REG64_TTBR1) {
+        // HACK HACK HACK: Workaround for 64-bit TTBRx
+        reg = (id == KVM_REG64_TTBR0 ? MISCREG_TTBR0 : MISCREG_TTBR1);
+        if (show_warnings)
+            hack("KVM: 64-bit TTBBRx workaround\n");
+    }
+
+    if (reg == NUM_MISCREGS) {
+        if (show_warnings) {
+            warn("KVM: Ignoring unknown KVM co-processor register (0x%.8x):\n",
+                 id);
+            warn("\t0x%x: [CP: %i 64: %i CRn: c%i opc1: %.2i CRm: c%i"
+                 " opc2: %i]\n",
+                 id, REG_CP(id), REG_IS_64BIT(id), REG_CRN(id),
+                 REG_OPC1(id), REG_CRM(id), REG_OPC2(id));
+        }
+    } else if (reg >= MISCREG_CP15_UNIMP_START && reg < MISCREG_CP15_END) {
+        if (show_warnings)
+            warn("KVM: Co-processor reg. %s not implemented by gem5.\n",
+                 miscRegName[reg]);
+    } else {
+        setOneReg(id, tc->readMiscRegNoEffect(reg));
+    }
+}
+
+
+void
+ArmKvmCPU::updateKvmStateVFP(uint64_t id, bool show_warnings)
+{
+    assert(REG_IS_ARM(id));
+    assert(REG_IS_VFP(id));
+
+    if (REG_IS_VFP_REG(id)) {
+        if (!REG_IS_64BIT(id)) {
+            if (show_warnings)
+                warn("Unexpected VFP register length (reg: 0x%x).\n", id);
+            return;
+        }
+        const unsigned idx(id & KVM_REG_ARM_VFP_MASK);
+        const unsigned idx_base(idx << 1);
+        const unsigned idx_hi(idx_base + 1);
+        const unsigned idx_lo(idx_base + 0);
+        uint64_t value(
+            ((uint64_t)tc->readFloatRegBitsFlat(idx_hi) << 32) |
+            tc->readFloatRegBitsFlat(idx_lo));
+
+        setOneReg(id, value);
+    } else if (REG_IS_VFP_CTRL(id)) {
+        MiscRegIndex idx(decodeVFPCtrlReg(id));
+        if (idx == NUM_MISCREGS) {
+            if (show_warnings)
+                warn("Unhandled VFP control register: 0x%x\n", id);
+            return;
+        }
+        if (!REG_IS_32BIT(id)) {
+            if (show_warnings)
+                warn("Ignoring VFP control register (%s) with "
+                     "unexpected size.\n",
+                     miscRegName[idx]);
+            return;
+        }
+        setOneReg(id, (uint32_t)tc->readMiscReg(idx));
+    } else {
+        if (show_warnings)
+            warn("Unhandled VFP register: 0x%x\n", id);
+    }
+}
+
+void
+ArmKvmCPU::updateTCStateCore()
+{
+    for (const KvmIntRegInfo *ri(kvmIntRegs);
+         ri->idx != NUM_INTREGS; ++ri) {
+
+        tc->setIntRegFlat(ri->idx, getOneRegU32(ri->id));
+    }
+
+    for (const KvmCoreMiscRegInfo *ri(kvmCoreMiscRegs);
+         ri->idx != NUM_MISCREGS; ++ri) {
+
+        tc->setMiscRegNoEffect(ri->idx, getOneRegU32(ri->id));
+    }
+
+    /* We want the simulator to execute all side-effects of the CPSR
+     * update since this updates PC state and register maps.
+     */
+    tc->setMiscReg(MISCREG_CPSR, tc->readMiscRegNoEffect(MISCREG_CPSR));
+
+    // We update the PC state after we have updated the CPSR the
+    // contents of the CPSR affects how the npc is updated.
+    PCState pc(tc->pcState());
+    pc.set(getOneRegU32(REG_CORE32(usr_regs.ARM_pc)));
+    tc->pcState(pc);
+}
+
+void
+ArmKvmCPU::updateTCStateMisc()
+{
+    static bool warned(false); // We can't use warn_once since we want
+                               // to show /all/ registers
+
+    const RegIndexVector &reg_ids(getRegList());;
+    for (RegIndexVector::const_iterator it(reg_ids.begin());
+         it != reg_ids.end(); ++it) {
+
+        if (!REG_IS_ARM(*it)) {
+            if (!warned)
+                warn("Skipping non-ARM register: 0x%x\n", *it);
+        } else if (REG_IS_CORE(*it)) {
+            // Core registers are handled in updateKvmStateCore
+        } else if (REG_CP(*it) <= 15) {
+            updateTCStateCoProc(*it, !warned);
+        } else if (REG_IS_VFP(*it)) {
+            updateTCStateVFP(*it, !warned);
+        } else {
+            if (!warned) {
+                warn("Skipping register with unknown CP (%i) id: 0x%x\n",
+                     REG_CP(*it), *it);
+            }
+        }
+    }
+
+    warned = true;
+}
+
+void
+ArmKvmCPU::updateTCStateCoProc(uint64_t id, bool show_warnings)
+{
+    MiscRegIndex reg(decodeCoProcReg(id));
+
+    assert(REG_IS_ARM(id));
+    assert(REG_CP(id) <= 15);
+
+    if (id == KVM_REG64_TTBR0 || id == KVM_REG64_TTBR1) {
+        // HACK HACK HACK: We don't currently support 64-bit TTBR0/TTBR1
+        hack_once("KVM: 64-bit TTBRx workaround\n");
+        tc->setMiscRegNoEffect(
+            id == KVM_REG64_TTBR0 ? MISCREG_TTBR0 : MISCREG_TTBR1,
+            (uint32_t)(getOneRegU64(id) & 0xFFFFFFFF));
+    } else if (reg == MISCREG_TTBCR) {
+        uint32_t value(getOneRegU64(id));
+        if (value & 0x80000000)
+            panic("KVM: Guest tried to enable LPAE.\n");
+        tc->setMiscRegNoEffect(reg, value);
+    } else if (reg == NUM_MISCREGS) {
+        if (show_warnings) {
+            warn("KVM: Ignoring unknown KVM co-processor register:\n", id);
+            warn("\t0x%x: [CP: %i 64: %i CRn: c%i opc1: %.2i CRm: c%i"
+                 " opc2: %i]\n",
+                 id, REG_CP(id), REG_IS_64BIT(id), REG_CRN(id),
+                 REG_OPC1(id), REG_CRM(id), REG_OPC2(id));
+        }
+    } else if (reg >= MISCREG_CP15_UNIMP_START && reg < MISCREG_CP15_END) {
+        if (show_warnings)
+            warn_once("KVM: Co-processor reg. %s not implemented by gem5.\n",
+                      miscRegName[reg]);
+    } else {
+        tc->setMiscRegNoEffect(reg, getOneRegU32(id));
+    }
+}
+
+void
+ArmKvmCPU::updateTCStateVFP(uint64_t id, bool show_warnings)
+{
+    assert(REG_IS_ARM(id));
+    assert(REG_IS_VFP(id));
+
+    if (REG_IS_VFP_REG(id)) {
+        if (!REG_IS_64BIT(id)) {
+            if (show_warnings)
+                warn("Unexpected VFP register length (reg: 0x%x).\n", id);
+            return;
+        }
+        const unsigned idx(id & KVM_REG_ARM_VFP_MASK);
+        const unsigned idx_base(idx << 1);
+        const unsigned idx_hi(idx_base + 1);
+        const unsigned idx_lo(idx_base + 0);
+        uint64_t value(getOneRegU64(id));
+
+        tc->setFloatRegBitsFlat(idx_hi, (value >> 32) & 0xFFFFFFFF);
+        tc->setFloatRegBitsFlat(idx_lo, value & 0xFFFFFFFF);
+    } else if (REG_IS_VFP_CTRL(id)) {
+        MiscRegIndex idx(decodeVFPCtrlReg(id));
+        if (idx == NUM_MISCREGS) {
+            if (show_warnings)
+                warn("Unhandled VFP control register: 0x%x\n", id);
+            return;
+        }
+        if (!REG_IS_32BIT(id)) {
+            if (show_warnings)
+                warn("Ignoring VFP control register (%s) with "
+                     "unexpected size.\n",
+                     miscRegName[idx]);
+            return;
+        }
+        tc->setMiscReg(idx, getOneRegU64(id));
+    } else {
+        if (show_warnings)
+            warn("Unhandled VFP register: 0x%x\n", id);
+    }
+}
+
+ArmKvmCPU *
+ArmKvmCPUParams::create()
+{
+    return new ArmKvmCPU(this);
+}
diff --git a/src/cpu/kvm/arm_cpu.hh b/src/cpu/kvm/arm_cpu.hh
new file mode 100644
index 000000000..b9d18b7f1
--- /dev/null
+++ b/src/cpu/kvm/arm_cpu.hh
@@ -0,0 +1,161 @@
+/*
+ * Copyright (c) 2012 ARM Limited
+ * All rights reserved
+ *
+ * The license below extends only to copyright in the software and shall
+ * not be construed as granting a license to any other intellectual
+ * property including but not limited to intellectual property relating
+ * to a hardware implementation of the functionality of the software
+ * licensed hereunder.  You may use the software subject to the license
+ * terms below provided that you ensure that this notice is replicated
+ * unmodified and in its entirety in all distributions of the software,
+ * modified or unmodified, in source code or in binary form.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met: redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer;
+ * redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution;
+ * neither the name of the copyright holders nor the names of its
+ * contributors may be used to endorse or promote products derived from
+ * this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * Authors: Andreas Sandberg
+ */
+
+#ifndef __CPU_KVM_ARM_CPU_HH__
+#define __CPU_KVM_ARM_CPU_HH__
+
+#include <set>
+#include <vector>
+
+#include "cpu/kvm/base.hh"
+#include "params/ArmKvmCPU.hh"
+
+/**
+ * ARM implementation of a KVM-based hardware virtualized CPU.
+ * Architecture specific limitations:
+ *  * LPAE is currently not supported by gem5. We therefore panic if LPAE
+ *    is enabled when returning to gem5.
+ *  * The co-processor based interface to the architected timer is
+ *    unsupported. We can't support this due to limitations in the KVM
+ *    API on ARM.
+ *  * M5 ops are currently not supported. This requires either a kernel
+ *  hack or a memory mapped device that handles the guest<->m5
+ *  interface.
+ */
+class ArmKvmCPU : public BaseKvmCPU
+{
+  public:
+    ArmKvmCPU(ArmKvmCPUParams *params);
+    virtual ~ArmKvmCPU();
+
+    void startup();
+
+  protected:
+    struct KvmIntRegInfo {
+        /** KVM ID */
+        const uint64_t id;
+        /** gem5 index */
+        const IntRegIndex idx;
+        /** Name in debug output */
+        const char *name;
+    };
+
+    struct KvmCoreMiscRegInfo {
+        /** KVM ID */
+        const uint64_t id;
+        /** gem5 index */
+        const MiscRegIndex idx;
+        /** Name in debug output */
+        const char *name;
+    };
+
+    typedef std::vector<uint64_t> RegIndexVector;
+
+    void tick();
+
+    void updateKvmState();
+    void updateThreadContext();
+
+    /**
+     * Get a list of registers supported by getOneReg() and setOneReg().
+     */
+    const RegIndexVector &getRegList() const;
+
+    void kvmArmVCpuInit(uint32_t target);
+    void kvmArmVCpuInit(const struct kvm_vcpu_init &init);
+
+    ArmISA::MiscRegIndex decodeCoProcReg(uint64_t id) const;
+
+    ArmISA::MiscRegIndex decodeVFPCtrlReg(uint64_t id) const;
+
+    /**
+     * Determine if a register is invariant.
+     *
+     * Some registers must have the same value in both the host and
+     * the guest. Such registers are referred to as "invariant"
+     * registers in KVM. This is a restriction imposed by KVM as
+     * having different values in ID registers (e.g., the cache
+     * identification registers) would confuse the guest kernel.
+     */
+    bool isInvariantReg(uint64_t id);
+
+    static KvmIntRegInfo kvmIntRegs[];
+    static KvmCoreMiscRegInfo kvmCoreMiscRegs[];
+
+  private:
+    /**
+     * Get a list of registers supported by getOneReg() and setOneReg().
+     *
+     * @return False if the number of elements allocated in the list
+     * is too small to hold the complete register list (the required
+     * value is written into n in this case). True on success.
+     */
+    bool getRegList(struct kvm_reg_list &regs) const;
+
+    void updateKvmStateCore();
+    void updateKvmStateMisc();
+    void updateKvmStateCoProc(uint64_t id, bool show_warnings);
+    void updateKvmStateVFP(uint64_t id, bool show_warnings);
+
+    void updateTCStateCore();
+    void updateTCStateMisc();
+    void updateTCStateCoProc(uint64_t id, bool show_warnings);
+    void updateTCStateVFP(uint64_t id, bool show_warnings);
+
+
+    /** Cached state of the IRQ line */
+    bool irqAsserted;
+    /** Cached state of the FIQ line */
+    bool fiqAsserted;
+
+    /**
+     * Cached copy of the list of co-processor registers supported by
+     * KVM
+     */
+    mutable RegIndexVector _regIndexList;
+
+    /**
+     * List of co-processor registers that KVM requires to be
+     * identical on both the host and the guest. KVM does not allow
+     * writes to these registers.
+     */
+    static const std::set<uint64_t> invariant_regs;
+};
+
+#endif