qemu/hw/i386/pc.c
<<
>>
Prefs
   1/*
   2 * QEMU PC System Emulator
   3 *
   4 * Copyright (c) 2003-2004 Fabrice Bellard
   5 *
   6 * Permission is hereby granted, free of charge, to any person obtaining a copy
   7 * of this software and associated documentation files (the "Software"), to deal
   8 * in the Software without restriction, including without limitation the rights
   9 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
  10 * copies of the Software, and to permit persons to whom the Software is
  11 * furnished to do so, subject to the following conditions:
  12 *
  13 * The above copyright notice and this permission notice shall be included in
  14 * all copies or substantial portions of the Software.
  15 *
  16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
  19 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
  21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
  22 * THE SOFTWARE.
  23 */
  24
  25#include "qemu/osdep.h"
  26#include "qemu/units.h"
  27#include "hw/i386/x86.h"
  28#include "hw/i386/pc.h"
  29#include "hw/char/serial.h"
  30#include "hw/char/parallel.h"
  31#include "hw/i386/topology.h"
  32#include "hw/i386/fw_cfg.h"
  33#include "hw/i386/vmport.h"
  34#include "sysemu/cpus.h"
  35#include "hw/block/fdc.h"
  36#include "hw/ide/internal.h"
  37#include "hw/ide/isa.h"
  38#include "hw/pci/pci.h"
  39#include "hw/pci/pci_bus.h"
  40#include "hw/pci-bridge/pci_expander_bridge.h"
  41#include "hw/nvram/fw_cfg.h"
  42#include "hw/timer/hpet.h"
  43#include "hw/firmware/smbios.h"
  44#include "hw/loader.h"
  45#include "elf.h"
  46#include "migration/vmstate.h"
  47#include "multiboot.h"
  48#include "hw/rtc/mc146818rtc.h"
  49#include "hw/intc/i8259.h"
  50#include "hw/intc/ioapic.h"
  51#include "hw/timer/i8254.h"
  52#include "hw/input/i8042.h"
  53#include "hw/irq.h"
  54#include "hw/audio/pcspk.h"
  55#include "hw/pci/msi.h"
  56#include "hw/sysbus.h"
  57#include "sysemu/sysemu.h"
  58#include "sysemu/tcg.h"
  59#include "sysemu/numa.h"
  60#include "sysemu/kvm.h"
  61#include "sysemu/xen.h"
  62#include "sysemu/reset.h"
  63#include "sysemu/runstate.h"
  64#include "kvm/kvm_i386.h"
  65#include "hw/xen/xen.h"
  66#include "hw/xen/start_info.h"
  67#include "ui/qemu-spice.h"
  68#include "exec/memory.h"
  69#include "qemu/bitmap.h"
  70#include "qemu/config-file.h"
  71#include "qemu/error-report.h"
  72#include "qemu/option.h"
  73#include "qemu/cutils.h"
  74#include "hw/acpi/acpi.h"
  75#include "hw/acpi/cpu_hotplug.h"
  76#include "acpi-build.h"
  77#include "hw/mem/pc-dimm.h"
  78#include "hw/mem/nvdimm.h"
  79#include "hw/cxl/cxl.h"
  80#include "hw/cxl/cxl_host.h"
  81#include "qapi/error.h"
  82#include "qapi/qapi-visit-common.h"
  83#include "qapi/qapi-visit-machine.h"
  84#include "qapi/visitor.h"
  85#include "hw/core/cpu.h"
  86#include "hw/usb.h"
  87#include "hw/i386/intel_iommu.h"
  88#include "hw/net/ne2000-isa.h"
  89#include "standard-headers/asm-x86/bootparam.h"
  90#include "hw/virtio/virtio-iommu.h"
  91#include "hw/virtio/virtio-pmem-pci.h"
  92#include "hw/virtio/virtio-mem-pci.h"
  93#include "hw/i386/kvm/xen_overlay.h"
  94#include "hw/i386/kvm/xen_evtchn.h"
  95#include "hw/i386/kvm/xen_gnttab.h"
  96#include "hw/i386/kvm/xen_xenstore.h"
  97#include "hw/mem/memory-device.h"
  98#include "sysemu/replay.h"
  99#include "target/i386/cpu.h"
 100#include "e820_memory_layout.h"
 101#include "fw_cfg.h"
 102#include "trace.h"
 103#include CONFIG_DEVICES
 104
 105#ifdef CONFIG_XEN_EMU
 106#include "hw/xen/xen-legacy-backend.h"
 107#include "hw/xen/xen-bus.h"
 108#endif
 109
 110/*
 111 * Helper for setting model-id for CPU models that changed model-id
 112 * depending on QEMU versions up to QEMU 2.4.
 113 */
 114#define PC_CPU_MODEL_IDS(v) \
 115    { "qemu32-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
 116    { "qemu64-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
 117    { "athlon-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },
 118
 119GlobalProperty pc_compat_7_2[] = {
 120    { "ICH9-LPC", "noreboot", "true" },
 121};
 122const size_t pc_compat_7_2_len = G_N_ELEMENTS(pc_compat_7_2);
 123
 124GlobalProperty pc_compat_7_1[] = {};
 125const size_t pc_compat_7_1_len = G_N_ELEMENTS(pc_compat_7_1);
 126
 127GlobalProperty pc_compat_7_0[] = {};
 128const size_t pc_compat_7_0_len = G_N_ELEMENTS(pc_compat_7_0);
 129
 130GlobalProperty pc_compat_6_2[] = {
 131    { "virtio-mem", "unplugged-inaccessible", "off" },
 132};
 133const size_t pc_compat_6_2_len = G_N_ELEMENTS(pc_compat_6_2);
 134
 135GlobalProperty pc_compat_6_1[] = {
 136    { TYPE_X86_CPU, "hv-version-id-build", "0x1bbc" },
 137    { TYPE_X86_CPU, "hv-version-id-major", "0x0006" },
 138    { TYPE_X86_CPU, "hv-version-id-minor", "0x0001" },
 139    { "ICH9-LPC", "x-keep-pci-slot-hpc", "false" },
 140};
 141const size_t pc_compat_6_1_len = G_N_ELEMENTS(pc_compat_6_1);
 142
 143GlobalProperty pc_compat_6_0[] = {
 144    { "qemu64" "-" TYPE_X86_CPU, "family", "6" },
 145    { "qemu64" "-" TYPE_X86_CPU, "model", "6" },
 146    { "qemu64" "-" TYPE_X86_CPU, "stepping", "3" },
 147    { TYPE_X86_CPU, "x-vendor-cpuid-only", "off" },
 148    { "ICH9-LPC", ACPI_PM_PROP_ACPI_PCIHP_BRIDGE, "off" },
 149    { "ICH9-LPC", "x-keep-pci-slot-hpc", "true" },
 150};
 151const size_t pc_compat_6_0_len = G_N_ELEMENTS(pc_compat_6_0);
 152
 153GlobalProperty pc_compat_5_2[] = {
 154    { "ICH9-LPC", "x-smi-cpu-hotunplug", "off" },
 155};
 156const size_t pc_compat_5_2_len = G_N_ELEMENTS(pc_compat_5_2);
 157
 158GlobalProperty pc_compat_5_1[] = {
 159    { "ICH9-LPC", "x-smi-cpu-hotplug", "off" },
 160    { TYPE_X86_CPU, "kvm-msi-ext-dest-id", "off" },
 161};
 162const size_t pc_compat_5_1_len = G_N_ELEMENTS(pc_compat_5_1);
 163
 164GlobalProperty pc_compat_5_0[] = {
 165};
 166const size_t pc_compat_5_0_len = G_N_ELEMENTS(pc_compat_5_0);
 167
 168GlobalProperty pc_compat_4_2[] = {
 169    { "mch", "smbase-smram", "off" },
 170};
 171const size_t pc_compat_4_2_len = G_N_ELEMENTS(pc_compat_4_2);
 172
 173GlobalProperty pc_compat_4_1[] = {};
 174const size_t pc_compat_4_1_len = G_N_ELEMENTS(pc_compat_4_1);
 175
 176GlobalProperty pc_compat_4_0[] = {};
 177const size_t pc_compat_4_0_len = G_N_ELEMENTS(pc_compat_4_0);
 178
 179GlobalProperty pc_compat_3_1[] = {
 180    { "intel-iommu", "dma-drain", "off" },
 181    { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "off" },
 182    { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "off" },
 183    { "Opteron_G4" "-" TYPE_X86_CPU, "npt", "off" },
 184    { "Opteron_G4" "-" TYPE_X86_CPU, "nrip-save", "off" },
 185    { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "off" },
 186    { "Opteron_G5" "-" TYPE_X86_CPU, "npt", "off" },
 187    { "Opteron_G5" "-" TYPE_X86_CPU, "nrip-save", "off" },
 188    { "EPYC" "-" TYPE_X86_CPU, "npt", "off" },
 189    { "EPYC" "-" TYPE_X86_CPU, "nrip-save", "off" },
 190    { "EPYC-IBPB" "-" TYPE_X86_CPU, "npt", "off" },
 191    { "EPYC-IBPB" "-" TYPE_X86_CPU, "nrip-save", "off" },
 192    { "Skylake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
 193    { "Skylake-Client-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
 194    { "Skylake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
 195    { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
 196    { "Cascadelake-Server" "-" TYPE_X86_CPU,  "mpx", "on" },
 197    { "Icelake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
 198    { "Icelake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
 199    { "Cascadelake-Server" "-" TYPE_X86_CPU, "stepping", "5" },
 200    { TYPE_X86_CPU, "x-intel-pt-auto-level", "off" },
 201};
 202const size_t pc_compat_3_1_len = G_N_ELEMENTS(pc_compat_3_1);
 203
 204GlobalProperty pc_compat_3_0[] = {
 205    { TYPE_X86_CPU, "x-hv-synic-kvm-only", "on" },
 206    { "Skylake-Server" "-" TYPE_X86_CPU, "pku", "off" },
 207    { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "pku", "off" },
 208};
 209const size_t pc_compat_3_0_len = G_N_ELEMENTS(pc_compat_3_0);
 210
 211GlobalProperty pc_compat_2_12[] = {
 212    { TYPE_X86_CPU, "legacy-cache", "on" },
 213    { TYPE_X86_CPU, "topoext", "off" },
 214    { "EPYC-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
 215    { "EPYC-IBPB-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
 216};
 217const size_t pc_compat_2_12_len = G_N_ELEMENTS(pc_compat_2_12);
 218
 219GlobalProperty pc_compat_2_11[] = {
 220    { TYPE_X86_CPU, "x-migrate-smi-count", "off" },
 221    { "Skylake-Server" "-" TYPE_X86_CPU, "clflushopt", "off" },
 222};
 223const size_t pc_compat_2_11_len = G_N_ELEMENTS(pc_compat_2_11);
 224
 225GlobalProperty pc_compat_2_10[] = {
 226    { TYPE_X86_CPU, "x-hv-max-vps", "0x40" },
 227    { "i440FX-pcihost", "x-pci-hole64-fix", "off" },
 228    { "q35-pcihost", "x-pci-hole64-fix", "off" },
 229};
 230const size_t pc_compat_2_10_len = G_N_ELEMENTS(pc_compat_2_10);
 231
 232GlobalProperty pc_compat_2_9[] = {
 233    { "mch", "extended-tseg-mbytes", "0" },
 234};
 235const size_t pc_compat_2_9_len = G_N_ELEMENTS(pc_compat_2_9);
 236
 237GlobalProperty pc_compat_2_8[] = {
 238    { TYPE_X86_CPU, "tcg-cpuid", "off" },
 239    { "kvmclock", "x-mach-use-reliable-get-clock", "off" },
 240    { "ICH9-LPC", "x-smi-broadcast", "off" },
 241    { TYPE_X86_CPU, "vmware-cpuid-freq", "off" },
 242    { "Haswell-" TYPE_X86_CPU, "stepping", "1" },
 243};
 244const size_t pc_compat_2_8_len = G_N_ELEMENTS(pc_compat_2_8);
 245
 246GlobalProperty pc_compat_2_7[] = {
 247    { TYPE_X86_CPU, "l3-cache", "off" },
 248    { TYPE_X86_CPU, "full-cpuid-auto-level", "off" },
 249    { "Opteron_G3" "-" TYPE_X86_CPU, "family", "15" },
 250    { "Opteron_G3" "-" TYPE_X86_CPU, "model", "6" },
 251    { "Opteron_G3" "-" TYPE_X86_CPU, "stepping", "1" },
 252    { "isa-pcspk", "migrate", "off" },
 253};
 254const size_t pc_compat_2_7_len = G_N_ELEMENTS(pc_compat_2_7);
 255
 256GlobalProperty pc_compat_2_6[] = {
 257    { TYPE_X86_CPU, "cpuid-0xb", "off" },
 258    { "vmxnet3", "romfile", "" },
 259    { TYPE_X86_CPU, "fill-mtrr-mask", "off" },
 260    { "apic-common", "legacy-instance-id", "on", }
 261};
 262const size_t pc_compat_2_6_len = G_N_ELEMENTS(pc_compat_2_6);
 263
 264GlobalProperty pc_compat_2_5[] = {};
 265const size_t pc_compat_2_5_len = G_N_ELEMENTS(pc_compat_2_5);
 266
 267GlobalProperty pc_compat_2_4[] = {
 268    PC_CPU_MODEL_IDS("2.4.0")
 269    { "Haswell-" TYPE_X86_CPU, "abm", "off" },
 270    { "Haswell-noTSX-" TYPE_X86_CPU, "abm", "off" },
 271    { "Broadwell-" TYPE_X86_CPU, "abm", "off" },
 272    { "Broadwell-noTSX-" TYPE_X86_CPU, "abm", "off" },
 273    { "host" "-" TYPE_X86_CPU, "host-cache-info", "on" },
 274    { TYPE_X86_CPU, "check", "off" },
 275    { "qemu64" "-" TYPE_X86_CPU, "sse4a", "on" },
 276    { "qemu64" "-" TYPE_X86_CPU, "abm", "on" },
 277    { "qemu64" "-" TYPE_X86_CPU, "popcnt", "on" },
 278    { "qemu32" "-" TYPE_X86_CPU, "popcnt", "on" },
 279    { "Opteron_G2" "-" TYPE_X86_CPU, "rdtscp", "on" },
 280    { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "on" },
 281    { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "on" },
 282    { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "on", }
 283};
 284const size_t pc_compat_2_4_len = G_N_ELEMENTS(pc_compat_2_4);
 285
 286GlobalProperty pc_compat_2_3[] = {
 287    PC_CPU_MODEL_IDS("2.3.0")
 288    { TYPE_X86_CPU, "arat", "off" },
 289    { "qemu64" "-" TYPE_X86_CPU, "min-level", "4" },
 290    { "kvm64" "-" TYPE_X86_CPU, "min-level", "5" },
 291    { "pentium3" "-" TYPE_X86_CPU, "min-level", "2" },
 292    { "n270" "-" TYPE_X86_CPU, "min-level", "5" },
 293    { "Conroe" "-" TYPE_X86_CPU, "min-level", "4" },
 294    { "Penryn" "-" TYPE_X86_CPU, "min-level", "4" },
 295    { "Nehalem" "-" TYPE_X86_CPU, "min-level", "4" },
 296    { "n270" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
 297    { "Penryn" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
 298    { "Conroe" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
 299    { "Nehalem" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
 300    { "Westmere" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
 301    { "SandyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
 302    { "IvyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
 303    { "Haswell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
 304    { "Haswell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
 305    { "Broadwell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
 306    { "Broadwell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
 307    { TYPE_X86_CPU, "kvm-no-smi-migration", "on" },
 308};
 309const size_t pc_compat_2_3_len = G_N_ELEMENTS(pc_compat_2_3);
 310
 311GlobalProperty pc_compat_2_2[] = {
 312    PC_CPU_MODEL_IDS("2.2.0")
 313    { "kvm64" "-" TYPE_X86_CPU, "vme", "off" },
 314    { "kvm32" "-" TYPE_X86_CPU, "vme", "off" },
 315    { "Conroe" "-" TYPE_X86_CPU, "vme", "off" },
 316    { "Penryn" "-" TYPE_X86_CPU, "vme", "off" },
 317    { "Nehalem" "-" TYPE_X86_CPU, "vme", "off" },
 318    { "Westmere" "-" TYPE_X86_CPU, "vme", "off" },
 319    { "SandyBridge" "-" TYPE_X86_CPU, "vme", "off" },
 320    { "Haswell" "-" TYPE_X86_CPU, "vme", "off" },
 321    { "Broadwell" "-" TYPE_X86_CPU, "vme", "off" },
 322    { "Opteron_G1" "-" TYPE_X86_CPU, "vme", "off" },
 323    { "Opteron_G2" "-" TYPE_X86_CPU, "vme", "off" },
 324    { "Opteron_G3" "-" TYPE_X86_CPU, "vme", "off" },
 325    { "Opteron_G4" "-" TYPE_X86_CPU, "vme", "off" },
 326    { "Opteron_G5" "-" TYPE_X86_CPU, "vme", "off" },
 327    { "Haswell" "-" TYPE_X86_CPU, "f16c", "off" },
 328    { "Haswell" "-" TYPE_X86_CPU, "rdrand", "off" },
 329    { "Broadwell" "-" TYPE_X86_CPU, "f16c", "off" },
 330    { "Broadwell" "-" TYPE_X86_CPU, "rdrand", "off" },
 331};
 332const size_t pc_compat_2_2_len = G_N_ELEMENTS(pc_compat_2_2);
 333
 334GlobalProperty pc_compat_2_1[] = {
 335    PC_CPU_MODEL_IDS("2.1.0")
 336    { "coreduo" "-" TYPE_X86_CPU, "vmx", "on" },
 337    { "core2duo" "-" TYPE_X86_CPU, "vmx", "on" },
 338};
 339const size_t pc_compat_2_1_len = G_N_ELEMENTS(pc_compat_2_1);
 340
 341GlobalProperty pc_compat_2_0[] = {
 342    PC_CPU_MODEL_IDS("2.0.0")
 343    { "virtio-scsi-pci", "any_layout", "off" },
 344    { "PIIX4_PM", "memory-hotplug-support", "off" },
 345    { "apic", "version", "0x11" },
 346    { "nec-usb-xhci", "superspeed-ports-first", "off" },
 347    { "nec-usb-xhci", "force-pcie-endcap", "on" },
 348    { "pci-serial", "prog_if", "0" },
 349    { "pci-serial-2x", "prog_if", "0" },
 350    { "pci-serial-4x", "prog_if", "0" },
 351    { "virtio-net-pci", "guest_announce", "off" },
 352    { "ICH9-LPC", "memory-hotplug-support", "off" },
 353};
 354const size_t pc_compat_2_0_len = G_N_ELEMENTS(pc_compat_2_0);
 355
 356GlobalProperty pc_compat_1_7[] = {
 357    PC_CPU_MODEL_IDS("1.7.0")
 358    { TYPE_USB_DEVICE, "msos-desc", "no" },
 359    { "PIIX4_PM", ACPI_PM_PROP_ACPI_PCIHP_BRIDGE, "off" },
 360    { "hpet", HPET_INTCAP, "4" },
 361};
 362const size_t pc_compat_1_7_len = G_N_ELEMENTS(pc_compat_1_7);
 363
 364GlobalProperty pc_compat_1_6[] = {
 365    PC_CPU_MODEL_IDS("1.6.0")
 366    { "e1000", "mitigation", "off" },
 367    { "qemu64-" TYPE_X86_CPU, "model", "2" },
 368    { "qemu32-" TYPE_X86_CPU, "model", "3" },
 369    { "i440FX-pcihost", "short_root_bus", "1" },
 370    { "q35-pcihost", "short_root_bus", "1" },
 371};
 372const size_t pc_compat_1_6_len = G_N_ELEMENTS(pc_compat_1_6);
 373
 374GlobalProperty pc_compat_1_5[] = {
 375    PC_CPU_MODEL_IDS("1.5.0")
 376    { "Conroe-" TYPE_X86_CPU, "model", "2" },
 377    { "Conroe-" TYPE_X86_CPU, "min-level", "2" },
 378    { "Penryn-" TYPE_X86_CPU, "model", "2" },
 379    { "Penryn-" TYPE_X86_CPU, "min-level", "2" },
 380    { "Nehalem-" TYPE_X86_CPU, "model", "2" },
 381    { "Nehalem-" TYPE_X86_CPU, "min-level", "2" },
 382    { "virtio-net-pci", "any_layout", "off" },
 383    { TYPE_X86_CPU, "pmu", "on" },
 384    { "i440FX-pcihost", "short_root_bus", "0" },
 385    { "q35-pcihost", "short_root_bus", "0" },
 386};
 387const size_t pc_compat_1_5_len = G_N_ELEMENTS(pc_compat_1_5);
 388
 389GlobalProperty pc_compat_1_4[] = {
 390    PC_CPU_MODEL_IDS("1.4.0")
 391    { "scsi-hd", "discard_granularity", "0" },
 392    { "scsi-cd", "discard_granularity", "0" },
 393    { "ide-hd", "discard_granularity", "0" },
 394    { "ide-cd", "discard_granularity", "0" },
 395    { "virtio-blk-pci", "discard_granularity", "0" },
 396    /* DEV_NVECTORS_UNSPECIFIED as a uint32_t string: */
 397    { "virtio-serial-pci", "vectors", "0xFFFFFFFF" },
 398    { "virtio-net-pci", "ctrl_guest_offloads", "off" },
 399    { "e1000", "romfile", "pxe-e1000.rom" },
 400    { "ne2k_pci", "romfile", "pxe-ne2k_pci.rom" },
 401    { "pcnet", "romfile", "pxe-pcnet.rom" },
 402    { "rtl8139", "romfile", "pxe-rtl8139.rom" },
 403    { "virtio-net-pci", "romfile", "pxe-virtio.rom" },
 404    { "486-" TYPE_X86_CPU, "model", "0" },
 405    { "n270" "-" TYPE_X86_CPU, "movbe", "off" },
 406    { "Westmere" "-" TYPE_X86_CPU, "pclmulqdq", "off" },
 407};
 408const size_t pc_compat_1_4_len = G_N_ELEMENTS(pc_compat_1_4);
 409
 410GSIState *pc_gsi_create(qemu_irq **irqs, bool pci_enabled)
 411{
 412    GSIState *s;
 413
 414    s = g_new0(GSIState, 1);
 415    if (kvm_ioapic_in_kernel()) {
 416        kvm_pc_setup_irq_routing(pci_enabled);
 417    }
 418    *irqs = qemu_allocate_irqs(gsi_handler, s, IOAPIC_NUM_PINS);
 419
 420    return s;
 421}
 422
 423static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
 424                           unsigned size)
 425{
 426}
 427
 428static uint64_t ioport80_read(void *opaque, hwaddr addr, unsigned size)
 429{
 430    return 0xffffffffffffffffULL;
 431}
 432
 433/* MSDOS compatibility mode FPU exception support */
 434static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
 435                           unsigned size)
 436{
 437    if (tcg_enabled()) {
 438        cpu_set_ignne();
 439    }
 440}
 441
 442static uint64_t ioportF0_read(void *opaque, hwaddr addr, unsigned size)
 443{
 444    return 0xffffffffffffffffULL;
 445}
 446
 447/* PC cmos mappings */
 448
 449#define REG_EQUIPMENT_BYTE          0x14
 450
 451static void cmos_init_hd(MC146818RtcState *s, int type_ofs, int info_ofs,
 452                         int16_t cylinders, int8_t heads, int8_t sectors)
 453{
 454    mc146818rtc_set_cmos_data(s, type_ofs, 47);
 455    mc146818rtc_set_cmos_data(s, info_ofs, cylinders);
 456    mc146818rtc_set_cmos_data(s, info_ofs + 1, cylinders >> 8);
 457    mc146818rtc_set_cmos_data(s, info_ofs + 2, heads);
 458    mc146818rtc_set_cmos_data(s, info_ofs + 3, 0xff);
 459    mc146818rtc_set_cmos_data(s, info_ofs + 4, 0xff);
 460    mc146818rtc_set_cmos_data(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
 461    mc146818rtc_set_cmos_data(s, info_ofs + 6, cylinders);
 462    mc146818rtc_set_cmos_data(s, info_ofs + 7, cylinders >> 8);
 463    mc146818rtc_set_cmos_data(s, info_ofs + 8, sectors);
 464}
 465
 466/* convert boot_device letter to something recognizable by the bios */
 467static int boot_device2nibble(char boot_device)
 468{
 469    switch(boot_device) {
 470    case 'a':
 471    case 'b':
 472        return 0x01; /* floppy boot */
 473    case 'c':
 474        return 0x02; /* hard drive boot */
 475    case 'd':
 476        return 0x03; /* CD-ROM boot */
 477    case 'n':
 478        return 0x04; /* Network boot */
 479    }
 480    return 0;
 481}
 482
 483static void set_boot_dev(MC146818RtcState *s, const char *boot_device,
 484                         Error **errp)
 485{
 486#define PC_MAX_BOOT_DEVICES 3
 487    int nbds, bds[3] = { 0, };
 488    int i;
 489
 490    nbds = strlen(boot_device);
 491    if (nbds > PC_MAX_BOOT_DEVICES) {
 492        error_setg(errp, "Too many boot devices for PC");
 493        return;
 494    }
 495    for (i = 0; i < nbds; i++) {
 496        bds[i] = boot_device2nibble(boot_device[i]);
 497        if (bds[i] == 0) {
 498            error_setg(errp, "Invalid boot device for PC: '%c'",
 499                       boot_device[i]);
 500            return;
 501        }
 502    }
 503    mc146818rtc_set_cmos_data(s, 0x3d, (bds[1] << 4) | bds[0]);
 504    mc146818rtc_set_cmos_data(s, 0x38, (bds[2] << 4) | (fd_bootchk ? 0x0 : 0x1));
 505}
 506
 507static void pc_boot_set(void *opaque, const char *boot_device, Error **errp)
 508{
 509    set_boot_dev(opaque, boot_device, errp);
 510}
 511
 512static void pc_cmos_init_floppy(MC146818RtcState *rtc_state, ISADevice *floppy)
 513{
 514    int val, nb, i;
 515    FloppyDriveType fd_type[2] = { FLOPPY_DRIVE_TYPE_NONE,
 516                                   FLOPPY_DRIVE_TYPE_NONE };
 517
 518    /* floppy type */
 519    if (floppy) {
 520        for (i = 0; i < 2; i++) {
 521            fd_type[i] = isa_fdc_get_drive_type(floppy, i);
 522        }
 523    }
 524    val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
 525        cmos_get_fd_drive_type(fd_type[1]);
 526    mc146818rtc_set_cmos_data(rtc_state, 0x10, val);
 527
 528    val = mc146818rtc_get_cmos_data(rtc_state, REG_EQUIPMENT_BYTE);
 529    nb = 0;
 530    if (fd_type[0] != FLOPPY_DRIVE_TYPE_NONE) {
 531        nb++;
 532    }
 533    if (fd_type[1] != FLOPPY_DRIVE_TYPE_NONE) {
 534        nb++;
 535    }
 536    switch (nb) {
 537    case 0:
 538        break;
 539    case 1:
 540        val |= 0x01; /* 1 drive, ready for boot */
 541        break;
 542    case 2:
 543        val |= 0x41; /* 2 drives, ready for boot */
 544        break;
 545    }
 546    mc146818rtc_set_cmos_data(rtc_state, REG_EQUIPMENT_BYTE, val);
 547}
 548
 549typedef struct pc_cmos_init_late_arg {
 550    MC146818RtcState *rtc_state;
 551    BusState *idebus[2];
 552} pc_cmos_init_late_arg;
 553
 554typedef struct check_fdc_state {
 555    ISADevice *floppy;
 556    bool multiple;
 557} CheckFdcState;
 558
 559static int check_fdc(Object *obj, void *opaque)
 560{
 561    CheckFdcState *state = opaque;
 562    Object *fdc;
 563    uint32_t iobase;
 564    Error *local_err = NULL;
 565
 566    fdc = object_dynamic_cast(obj, TYPE_ISA_FDC);
 567    if (!fdc) {
 568        return 0;
 569    }
 570
 571    iobase = object_property_get_uint(obj, "iobase", &local_err);
 572    if (local_err || iobase != 0x3f0) {
 573        error_free(local_err);
 574        return 0;
 575    }
 576
 577    if (state->floppy) {
 578        state->multiple = true;
 579    } else {
 580        state->floppy = ISA_DEVICE(obj);
 581    }
 582    return 0;
 583}
 584
 585static const char * const fdc_container_path[] = {
 586    "/unattached", "/peripheral", "/peripheral-anon"
 587};
 588
 589/*
 590 * Locate the FDC at IO address 0x3f0, in order to configure the CMOS registers
 591 * and ACPI objects.
 592 */
 593static ISADevice *pc_find_fdc0(void)
 594{
 595    int i;
 596    Object *container;
 597    CheckFdcState state = { 0 };
 598
 599    for (i = 0; i < ARRAY_SIZE(fdc_container_path); i++) {
 600        container = container_get(qdev_get_machine(), fdc_container_path[i]);
 601        object_child_foreach(container, check_fdc, &state);
 602    }
 603
 604    if (state.multiple) {
 605        warn_report("multiple floppy disk controllers with "
 606                    "iobase=0x3f0 have been found");
 607        error_printf("the one being picked for CMOS setup might not reflect "
 608                     "your intent");
 609    }
 610
 611    return state.floppy;
 612}
 613
 614static void pc_cmos_init_late(void *opaque)
 615{
 616    pc_cmos_init_late_arg *arg = opaque;
 617    MC146818RtcState *s = arg->rtc_state;
 618    int16_t cylinders;
 619    int8_t heads, sectors;
 620    int val;
 621    int i, trans;
 622
 623    val = 0;
 624    if (arg->idebus[0] && ide_get_geometry(arg->idebus[0], 0,
 625                                           &cylinders, &heads, &sectors) >= 0) {
 626        cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
 627        val |= 0xf0;
 628    }
 629    if (arg->idebus[0] && ide_get_geometry(arg->idebus[0], 1,
 630                                           &cylinders, &heads, &sectors) >= 0) {
 631        cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
 632        val |= 0x0f;
 633    }
 634    mc146818rtc_set_cmos_data(s, 0x12, val);
 635
 636    val = 0;
 637    for (i = 0; i < 4; i++) {
 638        /* NOTE: ide_get_geometry() returns the physical
 639           geometry.  It is always such that: 1 <= sects <= 63, 1
 640           <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
 641           geometry can be different if a translation is done. */
 642        if (arg->idebus[i / 2] &&
 643            ide_get_geometry(arg->idebus[i / 2], i % 2,
 644                             &cylinders, &heads, &sectors) >= 0) {
 645            trans = ide_get_bios_chs_trans(arg->idebus[i / 2], i % 2) - 1;
 646            assert((trans & ~3) == 0);
 647            val |= trans << (i * 2);
 648        }
 649    }
 650    mc146818rtc_set_cmos_data(s, 0x39, val);
 651
 652    pc_cmos_init_floppy(s, pc_find_fdc0());
 653
 654    qemu_unregister_reset(pc_cmos_init_late, opaque);
 655}
 656
 657void pc_cmos_init(PCMachineState *pcms,
 658                  BusState *idebus0, BusState *idebus1,
 659                  ISADevice *rtc)
 660{
 661    int val;
 662    static pc_cmos_init_late_arg arg;
 663    X86MachineState *x86ms = X86_MACHINE(pcms);
 664    MC146818RtcState *s = MC146818_RTC(rtc);
 665
 666    /* various important CMOS locations needed by PC/Bochs bios */
 667
 668    /* memory size */
 669    /* base memory (first MiB) */
 670    val = MIN(x86ms->below_4g_mem_size / KiB, 640);
 671    mc146818rtc_set_cmos_data(s, 0x15, val);
 672    mc146818rtc_set_cmos_data(s, 0x16, val >> 8);
 673    /* extended memory (next 64MiB) */
 674    if (x86ms->below_4g_mem_size > 1 * MiB) {
 675        val = (x86ms->below_4g_mem_size - 1 * MiB) / KiB;
 676    } else {
 677        val = 0;
 678    }
 679    if (val > 65535)
 680        val = 65535;
 681    mc146818rtc_set_cmos_data(s, 0x17, val);
 682    mc146818rtc_set_cmos_data(s, 0x18, val >> 8);
 683    mc146818rtc_set_cmos_data(s, 0x30, val);
 684    mc146818rtc_set_cmos_data(s, 0x31, val >> 8);
 685    /* memory between 16MiB and 4GiB */
 686    if (x86ms->below_4g_mem_size > 16 * MiB) {
 687        val = (x86ms->below_4g_mem_size - 16 * MiB) / (64 * KiB);
 688    } else {
 689        val = 0;
 690    }
 691    if (val > 65535)
 692        val = 65535;
 693    mc146818rtc_set_cmos_data(s, 0x34, val);
 694    mc146818rtc_set_cmos_data(s, 0x35, val >> 8);
 695    /* memory above 4GiB */
 696    val = x86ms->above_4g_mem_size / 65536;
 697    mc146818rtc_set_cmos_data(s, 0x5b, val);
 698    mc146818rtc_set_cmos_data(s, 0x5c, val >> 8);
 699    mc146818rtc_set_cmos_data(s, 0x5d, val >> 16);
 700
 701    object_property_add_link(OBJECT(pcms), "rtc_state",
 702                             TYPE_ISA_DEVICE,
 703                             (Object **)&x86ms->rtc,
 704                             object_property_allow_set_link,
 705                             OBJ_PROP_LINK_STRONG);
 706    object_property_set_link(OBJECT(pcms), "rtc_state", OBJECT(s),
 707                             &error_abort);
 708
 709    set_boot_dev(s, MACHINE(pcms)->boot_config.order, &error_fatal);
 710
 711    val = 0;
 712    val |= 0x02; /* FPU is there */
 713    val |= 0x04; /* PS/2 mouse installed */
 714    mc146818rtc_set_cmos_data(s, REG_EQUIPMENT_BYTE, val);
 715
 716    /* hard drives and FDC */
 717    arg.rtc_state = s;
 718    arg.idebus[0] = idebus0;
 719    arg.idebus[1] = idebus1;
 720    qemu_register_reset(pc_cmos_init_late, &arg);
 721}
 722
 723static void handle_a20_line_change(void *opaque, int irq, int level)
 724{
 725    X86CPU *cpu = opaque;
 726
 727    /* XXX: send to all CPUs ? */
 728    /* XXX: add logic to handle multiple A20 line sources */
 729    x86_cpu_set_a20(cpu, level);
 730}
 731
 732#define NE2000_NB_MAX 6
 733
 734static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
 735                                              0x280, 0x380 };
 736static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
 737
 738static void pc_init_ne2k_isa(ISABus *bus, NICInfo *nd)
 739{
 740    static int nb_ne2k = 0;
 741
 742    if (nb_ne2k == NE2000_NB_MAX)
 743        return;
 744    isa_ne2000_init(bus, ne2000_io[nb_ne2k],
 745                    ne2000_irq[nb_ne2k], nd);
 746    nb_ne2k++;
 747}
 748
 749void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
 750{
 751    X86CPU *cpu = opaque;
 752
 753    if (level) {
 754        cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
 755    }
 756}
 757
 758static
 759void pc_machine_done(Notifier *notifier, void *data)
 760{
 761    PCMachineState *pcms = container_of(notifier,
 762                                        PCMachineState, machine_done);
 763    X86MachineState *x86ms = X86_MACHINE(pcms);
 764
 765    cxl_hook_up_pxb_registers(pcms->bus, &pcms->cxl_devices_state,
 766                              &error_fatal);
 767
 768    if (pcms->cxl_devices_state.is_enabled) {
 769        cxl_fmws_link_targets(&pcms->cxl_devices_state, &error_fatal);
 770    }
 771
 772    /* set the number of CPUs */
 773    x86_rtc_set_cpus_count(x86ms->rtc, x86ms->boot_cpus);
 774
 775    fw_cfg_add_extra_pci_roots(pcms->bus, x86ms->fw_cfg);
 776
 777    acpi_setup();
 778    if (x86ms->fw_cfg) {
 779        fw_cfg_build_smbios(MACHINE(pcms), x86ms->fw_cfg);
 780        fw_cfg_build_feature_control(MACHINE(pcms), x86ms->fw_cfg);
 781        /* update FW_CFG_NB_CPUS to account for -device added CPUs */
 782        fw_cfg_modify_i16(x86ms->fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
 783    }
 784}
 785
 786void pc_guest_info_init(PCMachineState *pcms)
 787{
 788    X86MachineState *x86ms = X86_MACHINE(pcms);
 789
 790    x86ms->apic_xrupt_override = true;
 791    pcms->machine_done.notify = pc_machine_done;
 792    qemu_add_machine_init_done_notifier(&pcms->machine_done);
 793}
 794
 795/* setup pci memory address space mapping into system address space */
 796void pc_pci_as_mapping_init(MemoryRegion *system_memory,
 797                            MemoryRegion *pci_address_space)
 798{
 799    /* Set to lower priority than RAM */
 800    memory_region_add_subregion_overlap(system_memory, 0x0,
 801                                        pci_address_space, -1);
 802}
 803
 804void xen_load_linux(PCMachineState *pcms)
 805{
 806    int i;
 807    FWCfgState *fw_cfg;
 808    PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
 809    X86MachineState *x86ms = X86_MACHINE(pcms);
 810
 811    assert(MACHINE(pcms)->kernel_filename != NULL);
 812
 813    fw_cfg = fw_cfg_init_io(FW_CFG_IO_BASE);
 814    fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
 815    rom_set_fw(fw_cfg);
 816
 817    x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
 818                   pcmc->pvh_enabled);
 819    for (i = 0; i < nb_option_roms; i++) {
 820        assert(!strcmp(option_rom[i].name, "linuxboot.bin") ||
 821               !strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
 822               !strcmp(option_rom[i].name, "pvh.bin") ||
 823               !strcmp(option_rom[i].name, "multiboot.bin") ||
 824               !strcmp(option_rom[i].name, "multiboot_dma.bin"));
 825        rom_add_option(option_rom[i].name, option_rom[i].bootindex);
 826    }
 827    x86ms->fw_cfg = fw_cfg;
 828}
 829
 830#define PC_ROM_MIN_VGA     0xc0000
 831#define PC_ROM_MIN_OPTION  0xc8000
 832#define PC_ROM_MAX         0xe0000
 833#define PC_ROM_ALIGN       0x800
 834#define PC_ROM_SIZE        (PC_ROM_MAX - PC_ROM_MIN_VGA)
 835
 836static hwaddr pc_above_4g_end(PCMachineState *pcms)
 837{
 838    X86MachineState *x86ms = X86_MACHINE(pcms);
 839
 840    if (pcms->sgx_epc.size != 0) {
 841        return sgx_epc_above_4g_end(&pcms->sgx_epc);
 842    }
 843
 844    return x86ms->above_4g_mem_start + x86ms->above_4g_mem_size;
 845}
 846
 847static void pc_get_device_memory_range(PCMachineState *pcms,
 848                                       hwaddr *base,
 849                                       ram_addr_t *device_mem_size)
 850{
 851    PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
 852    MachineState *machine = MACHINE(pcms);
 853    ram_addr_t size;
 854    hwaddr addr;
 855
 856    size = machine->maxram_size - machine->ram_size;
 857    addr = ROUND_UP(pc_above_4g_end(pcms), 1 * GiB);
 858
 859    if (pcmc->enforce_aligned_dimm) {
 860        /* size device region assuming 1G page max alignment per slot */
 861        size += (1 * GiB) * machine->ram_slots;
 862    }
 863
 864    *base = addr;
 865    *device_mem_size = size;
 866}
 867
 868static uint64_t pc_get_cxl_range_start(PCMachineState *pcms)
 869{
 870    PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
 871    hwaddr cxl_base;
 872    ram_addr_t size;
 873
 874    if (pcmc->has_reserved_memory) {
 875        pc_get_device_memory_range(pcms, &cxl_base, &size);
 876        cxl_base += size;
 877    } else {
 878        cxl_base = pc_above_4g_end(pcms);
 879    }
 880
 881    return cxl_base;
 882}
 883
 884static uint64_t pc_get_cxl_range_end(PCMachineState *pcms)
 885{
 886    uint64_t start = pc_get_cxl_range_start(pcms) + MiB;
 887
 888    if (pcms->cxl_devices_state.fixed_windows) {
 889        GList *it;
 890
 891        start = ROUND_UP(start, 256 * MiB);
 892        for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
 893            CXLFixedWindow *fw = it->data;
 894            start += fw->size;
 895        }
 896    }
 897
 898    return start;
 899}
 900
 901static hwaddr pc_max_used_gpa(PCMachineState *pcms, uint64_t pci_hole64_size)
 902{
 903    X86CPU *cpu = X86_CPU(first_cpu);
 904
 905    /* 32-bit systems don't have hole64 thus return max CPU address */
 906    if (cpu->phys_bits <= 32) {
 907        return ((hwaddr)1 << cpu->phys_bits) - 1;
 908    }
 909
 910    return pc_pci_hole64_start() + pci_hole64_size - 1;
 911}
 912
 913/*
 914 * AMD systems with an IOMMU have an additional hole close to the
 915 * 1Tb, which are special GPAs that cannot be DMA mapped. Depending
 916 * on kernel version, VFIO may or may not let you DMA map those ranges.
 917 * Starting Linux v5.4 we validate it, and can't create guests on AMD machines
 918 * with certain memory sizes. It's also wrong to use those IOVA ranges
 919 * in detriment of leading to IOMMU INVALID_DEVICE_REQUEST or worse.
 920 * The ranges reserved for Hyper-Transport are:
 921 *
 922 * FD_0000_0000h - FF_FFFF_FFFFh
 923 *
 924 * The ranges represent the following:
 925 *
 926 * Base Address   Top Address  Use
 927 *
 928 * FD_0000_0000h FD_F7FF_FFFFh Reserved interrupt address space
 929 * FD_F800_0000h FD_F8FF_FFFFh Interrupt/EOI IntCtl
 930 * FD_F900_0000h FD_F90F_FFFFh Legacy PIC IACK
 931 * FD_F910_0000h FD_F91F_FFFFh System Management
 932 * FD_F920_0000h FD_FAFF_FFFFh Reserved Page Tables
 933 * FD_FB00_0000h FD_FBFF_FFFFh Address Translation
 934 * FD_FC00_0000h FD_FDFF_FFFFh I/O Space
 935 * FD_FE00_0000h FD_FFFF_FFFFh Configuration
 936 * FE_0000_0000h FE_1FFF_FFFFh Extended Configuration/Device Messages
 937 * FE_2000_0000h FF_FFFF_FFFFh Reserved
 938 *
 939 * See AMD IOMMU spec, section 2.1.2 "IOMMU Logical Topology",
 940 * Table 3: Special Address Controls (GPA) for more information.
 941 */
 942#define AMD_HT_START         0xfd00000000UL
 943#define AMD_HT_END           0xffffffffffUL
 944#define AMD_ABOVE_1TB_START  (AMD_HT_END + 1)
 945#define AMD_HT_SIZE          (AMD_ABOVE_1TB_START - AMD_HT_START)
 946
 947void pc_memory_init(PCMachineState *pcms,
 948                    MemoryRegion *system_memory,
 949                    MemoryRegion *rom_memory,
 950                    MemoryRegion **ram_memory,
 951                    uint64_t pci_hole64_size)
 952{
 953    int linux_boot, i;
 954    MemoryRegion *option_rom_mr;
 955    MemoryRegion *ram_below_4g, *ram_above_4g;
 956    FWCfgState *fw_cfg;
 957    MachineState *machine = MACHINE(pcms);
 958    MachineClass *mc = MACHINE_GET_CLASS(machine);
 959    PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
 960    X86MachineState *x86ms = X86_MACHINE(pcms);
 961    hwaddr maxphysaddr, maxusedaddr;
 962    hwaddr cxl_base, cxl_resv_end = 0;
 963    X86CPU *cpu = X86_CPU(first_cpu);
 964
 965    assert(machine->ram_size == x86ms->below_4g_mem_size +
 966                                x86ms->above_4g_mem_size);
 967
 968    linux_boot = (machine->kernel_filename != NULL);
 969
 970    /*
 971     * The HyperTransport range close to the 1T boundary is unique to AMD
 972     * hosts with IOMMUs enabled. Restrict the ram-above-4g relocation
 973     * to above 1T to AMD vCPUs only. @enforce_amd_1tb_hole is only false in
 974     * older machine types (<= 7.0) for compatibility purposes.
 975     */
 976    if (IS_AMD_CPU(&cpu->env) && pcmc->enforce_amd_1tb_hole) {
 977        /* Bail out if max possible address does not cross HT range */
 978        if (pc_max_used_gpa(pcms, pci_hole64_size) >= AMD_HT_START) {
 979            x86ms->above_4g_mem_start = AMD_ABOVE_1TB_START;
 980        }
 981
 982        /*
 983         * Advertise the HT region if address space covers the reserved
 984         * region or if we relocate.
 985         */
 986        if (cpu->phys_bits >= 40) {
 987            e820_add_entry(AMD_HT_START, AMD_HT_SIZE, E820_RESERVED);
 988        }
 989    }
 990
 991    /*
 992     * phys-bits is required to be appropriately configured
 993     * to make sure max used GPA is reachable.
 994     */
 995    maxusedaddr = pc_max_used_gpa(pcms, pci_hole64_size);
 996    maxphysaddr = ((hwaddr)1 << cpu->phys_bits) - 1;
 997    if (maxphysaddr < maxusedaddr) {
 998        error_report("Address space limit 0x%"PRIx64" < 0x%"PRIx64
 999                     " phys-bits too low (%u)",
1000                     maxphysaddr, maxusedaddr, cpu->phys_bits);
1001        exit(EXIT_FAILURE);
1002    }
1003
1004    /*
1005     * Split single memory region and use aliases to address portions of it,
1006     * done for backwards compatibility with older qemus.
1007     */
1008    *ram_memory = machine->ram;
1009    ram_below_4g = g_malloc(sizeof(*ram_below_4g));
1010    memory_region_init_alias(ram_below_4g, NULL, "ram-below-4g", machine->ram,
1011                             0, x86ms->below_4g_mem_size);
1012    memory_region_add_subregion(system_memory, 0, ram_below_4g);
1013    e820_add_entry(0, x86ms->below_4g_mem_size, E820_RAM);
1014    if (x86ms->above_4g_mem_size > 0) {
1015        ram_above_4g = g_malloc(sizeof(*ram_above_4g));
1016        memory_region_init_alias(ram_above_4g, NULL, "ram-above-4g",
1017                                 machine->ram,
1018                                 x86ms->below_4g_mem_size,
1019                                 x86ms->above_4g_mem_size);
1020        memory_region_add_subregion(system_memory, x86ms->above_4g_mem_start,
1021                                    ram_above_4g);
1022        e820_add_entry(x86ms->above_4g_mem_start, x86ms->above_4g_mem_size,
1023                       E820_RAM);
1024    }
1025
1026    if (pcms->sgx_epc.size != 0) {
1027        e820_add_entry(pcms->sgx_epc.base, pcms->sgx_epc.size, E820_RESERVED);
1028    }
1029
1030    if (!pcmc->has_reserved_memory &&
1031        (machine->ram_slots ||
1032         (machine->maxram_size > machine->ram_size))) {
1033
1034        error_report("\"-memory 'slots|maxmem'\" is not supported by: %s",
1035                     mc->name);
1036        exit(EXIT_FAILURE);
1037    }
1038
1039    /* always allocate the device memory information */
1040    machine->device_memory = g_malloc0(sizeof(*machine->device_memory));
1041
1042    /* initialize device memory address space */
1043    if (pcmc->has_reserved_memory &&
1044        (machine->ram_size < machine->maxram_size)) {
1045        ram_addr_t device_mem_size;
1046
1047        if (machine->ram_slots > ACPI_MAX_RAM_SLOTS) {
1048            error_report("unsupported amount of memory slots: %"PRIu64,
1049                         machine->ram_slots);
1050            exit(EXIT_FAILURE);
1051        }
1052
1053        if (QEMU_ALIGN_UP(machine->maxram_size,
1054                          TARGET_PAGE_SIZE) != machine->maxram_size) {
1055            error_report("maximum memory size must by aligned to multiple of "
1056                         "%d bytes", TARGET_PAGE_SIZE);
1057            exit(EXIT_FAILURE);
1058        }
1059
1060        pc_get_device_memory_range(pcms, &machine->device_memory->base, &device_mem_size);
1061
1062        if ((machine->device_memory->base + device_mem_size) <
1063            device_mem_size) {
1064            error_report("unsupported amount of maximum memory: " RAM_ADDR_FMT,
1065                         machine->maxram_size);
1066            exit(EXIT_FAILURE);
1067        }
1068
1069        memory_region_init(&machine->device_memory->mr, OBJECT(pcms),
1070                           "device-memory", device_mem_size);
1071        memory_region_add_subregion(system_memory, machine->device_memory->base,
1072                                    &machine->device_memory->mr);
1073    }
1074
1075    if (pcms->cxl_devices_state.is_enabled) {
1076        MemoryRegion *mr = &pcms->cxl_devices_state.host_mr;
1077        hwaddr cxl_size = MiB;
1078
1079        cxl_base = pc_get_cxl_range_start(pcms);
1080        memory_region_init(mr, OBJECT(machine), "cxl_host_reg", cxl_size);
1081        memory_region_add_subregion(system_memory, cxl_base, mr);
1082        cxl_resv_end = cxl_base + cxl_size;
1083        if (pcms->cxl_devices_state.fixed_windows) {
1084            hwaddr cxl_fmw_base;
1085            GList *it;
1086
1087            cxl_fmw_base = ROUND_UP(cxl_base + cxl_size, 256 * MiB);
1088            for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
1089                CXLFixedWindow *fw = it->data;
1090
1091                fw->base = cxl_fmw_base;
1092                memory_region_init_io(&fw->mr, OBJECT(machine), &cfmws_ops, fw,
1093                                      "cxl-fixed-memory-region", fw->size);
1094                memory_region_add_subregion(system_memory, fw->base, &fw->mr);
1095                cxl_fmw_base += fw->size;
1096                cxl_resv_end = cxl_fmw_base;
1097            }
1098        }
1099    }
1100
1101    /* Initialize PC system firmware */
1102    pc_system_firmware_init(pcms, rom_memory);
1103
1104    option_rom_mr = g_malloc(sizeof(*option_rom_mr));
1105    memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE,
1106                           &error_fatal);
1107    if (pcmc->pci_enabled) {
1108        memory_region_set_readonly(option_rom_mr, true);
1109    }
1110    memory_region_add_subregion_overlap(rom_memory,
1111                                        PC_ROM_MIN_VGA,
1112                                        option_rom_mr,
1113                                        1);
1114
1115    fw_cfg = fw_cfg_arch_create(machine,
1116                                x86ms->boot_cpus, x86ms->apic_id_limit);
1117
1118    rom_set_fw(fw_cfg);
1119
1120    if (pcmc->has_reserved_memory && machine->device_memory->base) {
1121        uint64_t *val = g_malloc(sizeof(*val));
1122        PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1123        uint64_t res_mem_end = machine->device_memory->base;
1124
1125        if (!pcmc->broken_reserved_end) {
1126            res_mem_end += memory_region_size(&machine->device_memory->mr);
1127        }
1128
1129        if (pcms->cxl_devices_state.is_enabled) {
1130            res_mem_end = cxl_resv_end;
1131        }
1132        *val = cpu_to_le64(ROUND_UP(res_mem_end, 1 * GiB));
1133        fw_cfg_add_file(fw_cfg, "etc/reserved-memory-end", val, sizeof(*val));
1134    }
1135
1136    if (linux_boot) {
1137        x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
1138                       pcmc->pvh_enabled);
1139    }
1140
1141    for (i = 0; i < nb_option_roms; i++) {
1142        rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1143    }
1144    x86ms->fw_cfg = fw_cfg;
1145
1146    /* Init default IOAPIC address space */
1147    x86ms->ioapic_as = &address_space_memory;
1148
1149    /* Init ACPI memory hotplug IO base address */
1150    pcms->memhp_io_base = ACPI_MEMORY_HOTPLUG_BASE;
1151}
1152
1153/*
1154 * The 64bit pci hole starts after "above 4G RAM" and
1155 * potentially the space reserved for memory hotplug.
1156 */
1157uint64_t pc_pci_hole64_start(void)
1158{
1159    PCMachineState *pcms = PC_MACHINE(qdev_get_machine());
1160    PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1161    MachineState *ms = MACHINE(pcms);
1162    uint64_t hole64_start = 0;
1163    ram_addr_t size = 0;
1164
1165    if (pcms->cxl_devices_state.is_enabled) {
1166        hole64_start = pc_get_cxl_range_end(pcms);
1167    } else if (pcmc->has_reserved_memory && (ms->ram_size < ms->maxram_size)) {
1168        pc_get_device_memory_range(pcms, &hole64_start, &size);
1169        if (!pcmc->broken_reserved_end) {
1170            hole64_start += size;
1171        }
1172    } else {
1173        hole64_start = pc_above_4g_end(pcms);
1174    }
1175
1176    return ROUND_UP(hole64_start, 1 * GiB);
1177}
1178
1179DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1180{
1181    DeviceState *dev = NULL;
1182
1183    rom_set_order_override(FW_CFG_ORDER_OVERRIDE_VGA);
1184    if (pci_bus) {
1185        PCIDevice *pcidev = pci_vga_init(pci_bus);
1186        dev = pcidev ? &pcidev->qdev : NULL;
1187    } else if (isa_bus) {
1188        ISADevice *isadev = isa_vga_init(isa_bus);
1189        dev = isadev ? DEVICE(isadev) : NULL;
1190    }
1191    rom_reset_order_override();
1192    return dev;
1193}
1194
1195static const MemoryRegionOps ioport80_io_ops = {
1196    .write = ioport80_write,
1197    .read = ioport80_read,
1198    .endianness = DEVICE_NATIVE_ENDIAN,
1199    .impl = {
1200        .min_access_size = 1,
1201        .max_access_size = 1,
1202    },
1203};
1204
1205static const MemoryRegionOps ioportF0_io_ops = {
1206    .write = ioportF0_write,
1207    .read = ioportF0_read,
1208    .endianness = DEVICE_NATIVE_ENDIAN,
1209    .impl = {
1210        .min_access_size = 1,
1211        .max_access_size = 1,
1212    },
1213};
1214
1215static void pc_superio_init(ISABus *isa_bus, bool create_fdctrl,
1216                            bool create_i8042, bool no_vmport)
1217{
1218    int i;
1219    DriveInfo *fd[MAX_FD];
1220    qemu_irq *a20_line;
1221    ISADevice *fdc, *i8042, *port92, *vmmouse;
1222
1223    serial_hds_isa_init(isa_bus, 0, MAX_ISA_SERIAL_PORTS);
1224    parallel_hds_isa_init(isa_bus, MAX_PARALLEL_PORTS);
1225
1226    for (i = 0; i < MAX_FD; i++) {
1227        fd[i] = drive_get(IF_FLOPPY, 0, i);
1228        create_fdctrl |= !!fd[i];
1229    }
1230    if (create_fdctrl) {
1231        fdc = isa_new(TYPE_ISA_FDC);
1232        if (fdc) {
1233            isa_realize_and_unref(fdc, isa_bus, &error_fatal);
1234            isa_fdc_init_drives(fdc, fd);
1235        }
1236    }
1237
1238    if (!create_i8042) {
1239        return;
1240    }
1241
1242    i8042 = isa_create_simple(isa_bus, TYPE_I8042);
1243    if (!no_vmport) {
1244        isa_create_simple(isa_bus, TYPE_VMPORT);
1245        vmmouse = isa_try_new("vmmouse");
1246    } else {
1247        vmmouse = NULL;
1248    }
1249    if (vmmouse) {
1250        object_property_set_link(OBJECT(vmmouse), TYPE_I8042, OBJECT(i8042),
1251                                 &error_abort);
1252        isa_realize_and_unref(vmmouse, isa_bus, &error_fatal);
1253    }
1254    port92 = isa_create_simple(isa_bus, TYPE_PORT92);
1255
1256    a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1257    i8042_setup_a20_line(i8042, a20_line[0]);
1258    qdev_connect_gpio_out_named(DEVICE(port92),
1259                                PORT92_A20_LINE, 0, a20_line[1]);
1260    g_free(a20_line);
1261}
1262
1263void pc_basic_device_init(struct PCMachineState *pcms,
1264                          ISABus *isa_bus, qemu_irq *gsi,
1265                          ISADevice **rtc_state,
1266                          bool create_fdctrl,
1267                          uint32_t hpet_irqs)
1268{
1269    int i;
1270    DeviceState *hpet = NULL;
1271    int pit_isa_irq = 0;
1272    qemu_irq pit_alt_irq = NULL;
1273    qemu_irq rtc_irq = NULL;
1274    ISADevice *pit = NULL;
1275    MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1276    MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
1277    X86MachineState *x86ms = X86_MACHINE(pcms);
1278
1279    memory_region_init_io(ioport80_io, NULL, &ioport80_io_ops, NULL, "ioport80", 1);
1280    memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
1281
1282    memory_region_init_io(ioportF0_io, NULL, &ioportF0_io_ops, NULL, "ioportF0", 1);
1283    memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
1284
1285    /*
1286     * Check if an HPET shall be created.
1287     *
1288     * Without KVM_CAP_PIT_STATE2, we cannot switch off the in-kernel PIT
1289     * when the HPET wants to take over. Thus we have to disable the latter.
1290     */
1291    if (pcms->hpet_enabled && (!kvm_irqchip_in_kernel() ||
1292                               kvm_has_pit_state2())) {
1293        hpet = qdev_try_new(TYPE_HPET);
1294        if (!hpet) {
1295            error_report("couldn't create HPET device");
1296            exit(1);
1297        }
1298        /*
1299         * For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-1.7 and
1300         * earlier, use IRQ2 for compat. Otherwise, use IRQ16~23, IRQ8 and
1301         * IRQ2.
1302         */
1303        uint8_t compat = object_property_get_uint(OBJECT(hpet),
1304                HPET_INTCAP, NULL);
1305        if (!compat) {
1306            qdev_prop_set_uint32(hpet, HPET_INTCAP, hpet_irqs);
1307        }
1308        sysbus_realize_and_unref(SYS_BUS_DEVICE(hpet), &error_fatal);
1309        sysbus_mmio_map(SYS_BUS_DEVICE(hpet), 0, HPET_BASE);
1310
1311        for (i = 0; i < IOAPIC_NUM_PINS; i++) {
1312            sysbus_connect_irq(SYS_BUS_DEVICE(hpet), i, gsi[i]);
1313        }
1314        pit_isa_irq = -1;
1315        pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1316        rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1317    }
1318    *rtc_state = ISA_DEVICE(mc146818_rtc_init(isa_bus, 2000, rtc_irq));
1319
1320#ifdef CONFIG_XEN_EMU
1321    if (xen_mode == XEN_EMULATE) {
1322        xen_evtchn_connect_gsis(gsi);
1323        if (pcms->bus) {
1324            pci_create_simple(pcms->bus, -1, "xen-platform");
1325        }
1326        xen_bus_init();
1327        xen_be_init();
1328    }
1329#endif
1330
1331    qemu_register_boot_set(pc_boot_set, *rtc_state);
1332
1333    if (!xen_enabled() &&
1334        (x86ms->pit == ON_OFF_AUTO_AUTO || x86ms->pit == ON_OFF_AUTO_ON)) {
1335        if (kvm_pit_in_kernel()) {
1336            pit = kvm_pit_init(isa_bus, 0x40);
1337        } else {
1338            pit = i8254_pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1339        }
1340        if (hpet) {
1341            /* connect PIT to output control line of the HPET */
1342            qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(DEVICE(pit), 0));
1343        }
1344        pcspk_init(pcms->pcspk, isa_bus, pit);
1345    }
1346
1347    /* Super I/O */
1348    pc_superio_init(isa_bus, create_fdctrl, pcms->i8042_enabled,
1349                    pcms->vmport != ON_OFF_AUTO_ON);
1350}
1351
1352void pc_nic_init(PCMachineClass *pcmc, ISABus *isa_bus, PCIBus *pci_bus)
1353{
1354    int i;
1355
1356    rom_set_order_override(FW_CFG_ORDER_OVERRIDE_NIC);
1357    for (i = 0; i < nb_nics; i++) {
1358        NICInfo *nd = &nd_table[i];
1359        const char *model = nd->model ? nd->model : pcmc->default_nic_model;
1360
1361        if (g_str_equal(model, "ne2k_isa")) {
1362            pc_init_ne2k_isa(isa_bus, nd);
1363        } else {
1364            pci_nic_init_nofail(nd, pci_bus, model, NULL);
1365        }
1366    }
1367    rom_reset_order_override();
1368}
1369
1370void pc_i8259_create(ISABus *isa_bus, qemu_irq *i8259_irqs)
1371{
1372    qemu_irq *i8259;
1373
1374    if (kvm_pic_in_kernel()) {
1375        i8259 = kvm_i8259_init(isa_bus);
1376    } else if (xen_enabled()) {
1377        i8259 = xen_interrupt_controller_init();
1378    } else {
1379        i8259 = i8259_init(isa_bus, x86_allocate_cpu_irq());
1380    }
1381
1382    for (size_t i = 0; i < ISA_NUM_IRQS; i++) {
1383        i8259_irqs[i] = i8259[i];
1384    }
1385
1386    g_free(i8259);
1387}
1388
1389static void pc_memory_pre_plug(HotplugHandler *hotplug_dev, DeviceState *dev,
1390                               Error **errp)
1391{
1392    const PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1393    const X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1394    const PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1395    const MachineState *ms = MACHINE(hotplug_dev);
1396    const bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1397    const uint64_t legacy_align = TARGET_PAGE_SIZE;
1398    Error *local_err = NULL;
1399
1400    /*
1401     * When -no-acpi is used with Q35 machine type, no ACPI is built,
1402     * but pcms->acpi_dev is still created. Check !acpi_enabled in
1403     * addition to cover this case.
1404     */
1405    if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1406        error_setg(errp,
1407                   "memory hotplug is not enabled: missing acpi device or acpi disabled");
1408        return;
1409    }
1410
1411    if (is_nvdimm && !ms->nvdimms_state->is_enabled) {
1412        error_setg(errp, "nvdimm is not enabled: missing 'nvdimm' in '-M'");
1413        return;
1414    }
1415
1416    hotplug_handler_pre_plug(x86ms->acpi_dev, dev, &local_err);
1417    if (local_err) {
1418        error_propagate(errp, local_err);
1419        return;
1420    }
1421
1422    pc_dimm_pre_plug(PC_DIMM(dev), MACHINE(hotplug_dev),
1423                     pcmc->enforce_aligned_dimm ? NULL : &legacy_align, errp);
1424}
1425
1426static void pc_memory_plug(HotplugHandler *hotplug_dev,
1427                           DeviceState *dev, Error **errp)
1428{
1429    PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1430    X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1431    MachineState *ms = MACHINE(hotplug_dev);
1432    bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1433
1434    pc_dimm_plug(PC_DIMM(dev), MACHINE(pcms));
1435
1436    if (is_nvdimm) {
1437        nvdimm_plug(ms->nvdimms_state);
1438    }
1439
1440    hotplug_handler_plug(x86ms->acpi_dev, dev, &error_abort);
1441}
1442
1443static void pc_memory_unplug_request(HotplugHandler *hotplug_dev,
1444                                     DeviceState *dev, Error **errp)
1445{
1446    X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1447
1448    /*
1449     * When -no-acpi is used with Q35 machine type, no ACPI is built,
1450     * but pcms->acpi_dev is still created. Check !acpi_enabled in
1451     * addition to cover this case.
1452     */
1453    if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1454        error_setg(errp,
1455                   "memory hotplug is not enabled: missing acpi device or acpi disabled");
1456        return;
1457    }
1458
1459    if (object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM)) {
1460        error_setg(errp, "nvdimm device hot unplug is not supported yet.");
1461        return;
1462    }
1463
1464    hotplug_handler_unplug_request(x86ms->acpi_dev, dev,
1465                                   errp);
1466}
1467
1468static void pc_memory_unplug(HotplugHandler *hotplug_dev,
1469                             DeviceState *dev, Error **errp)
1470{
1471    PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1472    X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1473    Error *local_err = NULL;
1474
1475    hotplug_handler_unplug(x86ms->acpi_dev, dev, &local_err);
1476    if (local_err) {
1477        goto out;
1478    }
1479
1480    pc_dimm_unplug(PC_DIMM(dev), MACHINE(pcms));
1481    qdev_unrealize(dev);
1482 out:
1483    error_propagate(errp, local_err);
1484}
1485
1486static void pc_virtio_md_pci_pre_plug(HotplugHandler *hotplug_dev,
1487                                      DeviceState *dev, Error **errp)
1488{
1489    HotplugHandler *hotplug_dev2 = qdev_get_bus_hotplug_handler(dev);
1490    Error *local_err = NULL;
1491
1492    if (!hotplug_dev2 && dev->hotplugged) {
1493        /*
1494         * Without a bus hotplug handler, we cannot control the plug/unplug
1495         * order. We should never reach this point when hotplugging on x86,
1496         * however, better add a safety net.
1497         */
1498        error_setg(errp, "hotplug of virtio based memory devices not supported"
1499                   " on this bus.");
1500        return;
1501    }
1502    /*
1503     * First, see if we can plug this memory device at all. If that
1504     * succeeds, branch of to the actual hotplug handler.
1505     */
1506    memory_device_pre_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev), NULL,
1507                           &local_err);
1508    if (!local_err && hotplug_dev2) {
1509        hotplug_handler_pre_plug(hotplug_dev2, dev, &local_err);
1510    }
1511    error_propagate(errp, local_err);
1512}
1513
1514static void pc_virtio_md_pci_plug(HotplugHandler *hotplug_dev,
1515                                  DeviceState *dev, Error **errp)
1516{
1517    HotplugHandler *hotplug_dev2 = qdev_get_bus_hotplug_handler(dev);
1518    Error *local_err = NULL;
1519
1520    /*
1521     * Plug the memory device first and then branch off to the actual
1522     * hotplug handler. If that one fails, we can easily undo the memory
1523     * device bits.
1524     */
1525    memory_device_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev));
1526    if (hotplug_dev2) {
1527        hotplug_handler_plug(hotplug_dev2, dev, &local_err);
1528        if (local_err) {
1529            memory_device_unplug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev));
1530        }
1531    }
1532    error_propagate(errp, local_err);
1533}
1534
1535static void pc_virtio_md_pci_unplug_request(HotplugHandler *hotplug_dev,
1536                                            DeviceState *dev, Error **errp)
1537{
1538    /* We don't support hot unplug of virtio based memory devices */
1539    error_setg(errp, "virtio based memory devices cannot be unplugged.");
1540}
1541
1542static void pc_virtio_md_pci_unplug(HotplugHandler *hotplug_dev,
1543                                    DeviceState *dev, Error **errp)
1544{
1545    /* We don't support hot unplug of virtio based memory devices */
1546}
1547
1548static void pc_machine_device_pre_plug_cb(HotplugHandler *hotplug_dev,
1549                                          DeviceState *dev, Error **errp)
1550{
1551    if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1552        pc_memory_pre_plug(hotplug_dev, dev, errp);
1553    } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1554        x86_cpu_pre_plug(hotplug_dev, dev, errp);
1555    } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI) ||
1556               object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MEM_PCI)) {
1557        pc_virtio_md_pci_pre_plug(hotplug_dev, dev, errp);
1558    } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1559        /* Declare the APIC range as the reserved MSI region */
1560        char *resv_prop_str = g_strdup_printf("0xfee00000:0xfeefffff:%d",
1561                                              VIRTIO_IOMMU_RESV_MEM_T_MSI);
1562
1563        object_property_set_uint(OBJECT(dev), "len-reserved-regions", 1, errp);
1564        object_property_set_str(OBJECT(dev), "reserved-regions[0]",
1565                                resv_prop_str, errp);
1566        g_free(resv_prop_str);
1567    }
1568
1569    if (object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE) ||
1570        object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1571        PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1572
1573        if (pcms->iommu) {
1574            error_setg(errp, "QEMU does not support multiple vIOMMUs "
1575                       "for x86 yet.");
1576            return;
1577        }
1578        pcms->iommu = dev;
1579    }
1580}
1581
1582static void pc_machine_device_plug_cb(HotplugHandler *hotplug_dev,
1583                                      DeviceState *dev, Error **errp)
1584{
1585    if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1586        pc_memory_plug(hotplug_dev, dev, errp);
1587    } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1588        x86_cpu_plug(hotplug_dev, dev, errp);
1589    } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI) ||
1590               object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MEM_PCI)) {
1591        pc_virtio_md_pci_plug(hotplug_dev, dev, errp);
1592    }
1593}
1594
1595static void pc_machine_device_unplug_request_cb(HotplugHandler *hotplug_dev,
1596                                                DeviceState *dev, Error **errp)
1597{
1598    if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1599        pc_memory_unplug_request(hotplug_dev, dev, errp);
1600    } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1601        x86_cpu_unplug_request_cb(hotplug_dev, dev, errp);
1602    } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI) ||
1603               object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MEM_PCI)) {
1604        pc_virtio_md_pci_unplug_request(hotplug_dev, dev, errp);
1605    } else {
1606        error_setg(errp, "acpi: device unplug request for not supported device"
1607                   " type: %s", object_get_typename(OBJECT(dev)));
1608    }
1609}
1610
1611static void pc_machine_device_unplug_cb(HotplugHandler *hotplug_dev,
1612                                        DeviceState *dev, Error **errp)
1613{
1614    if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1615        pc_memory_unplug(hotplug_dev, dev, errp);
1616    } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1617        x86_cpu_unplug_cb(hotplug_dev, dev, errp);
1618    } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI) ||
1619               object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MEM_PCI)) {
1620        pc_virtio_md_pci_unplug(hotplug_dev, dev, errp);
1621    } else {
1622        error_setg(errp, "acpi: device unplug for not supported device"
1623                   " type: %s", object_get_typename(OBJECT(dev)));
1624    }
1625}
1626
1627static HotplugHandler *pc_get_hotplug_handler(MachineState *machine,
1628                                             DeviceState *dev)
1629{
1630    if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM) ||
1631        object_dynamic_cast(OBJECT(dev), TYPE_CPU) ||
1632        object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI) ||
1633        object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MEM_PCI) ||
1634        object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI) ||
1635        object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE)) {
1636        return HOTPLUG_HANDLER(machine);
1637    }
1638
1639    return NULL;
1640}
1641
1642static void
1643pc_machine_get_device_memory_region_size(Object *obj, Visitor *v,
1644                                         const char *name, void *opaque,
1645                                         Error **errp)
1646{
1647    MachineState *ms = MACHINE(obj);
1648    int64_t value = 0;
1649
1650    if (ms->device_memory) {
1651        value = memory_region_size(&ms->device_memory->mr);
1652    }
1653
1654    visit_type_int(v, name, &value, errp);
1655}
1656
1657static void pc_machine_get_vmport(Object *obj, Visitor *v, const char *name,
1658                                  void *opaque, Error **errp)
1659{
1660    PCMachineState *pcms = PC_MACHINE(obj);
1661    OnOffAuto vmport = pcms->vmport;
1662
1663    visit_type_OnOffAuto(v, name, &vmport, errp);
1664}
1665
1666static void pc_machine_set_vmport(Object *obj, Visitor *v, const char *name,
1667                                  void *opaque, Error **errp)
1668{
1669    PCMachineState *pcms = PC_MACHINE(obj);
1670
1671    visit_type_OnOffAuto(v, name, &pcms->vmport, errp);
1672}
1673
1674static bool pc_machine_get_smbus(Object *obj, Error **errp)
1675{
1676    PCMachineState *pcms = PC_MACHINE(obj);
1677
1678    return pcms->smbus_enabled;
1679}
1680
1681static void pc_machine_set_smbus(Object *obj, bool value, Error **errp)
1682{
1683    PCMachineState *pcms = PC_MACHINE(obj);
1684
1685    pcms->smbus_enabled = value;
1686}
1687
1688static bool pc_machine_get_sata(Object *obj, Error **errp)
1689{
1690    PCMachineState *pcms = PC_MACHINE(obj);
1691
1692    return pcms->sata_enabled;
1693}
1694
1695static void pc_machine_set_sata(Object *obj, bool value, Error **errp)
1696{
1697    PCMachineState *pcms = PC_MACHINE(obj);
1698
1699    pcms->sata_enabled = value;
1700}
1701
1702static bool pc_machine_get_hpet(Object *obj, Error **errp)
1703{
1704    PCMachineState *pcms = PC_MACHINE(obj);
1705
1706    return pcms->hpet_enabled;
1707}
1708
1709static void pc_machine_set_hpet(Object *obj, bool value, Error **errp)
1710{
1711    PCMachineState *pcms = PC_MACHINE(obj);
1712
1713    pcms->hpet_enabled = value;
1714}
1715
1716static bool pc_machine_get_i8042(Object *obj, Error **errp)
1717{
1718    PCMachineState *pcms = PC_MACHINE(obj);
1719
1720    return pcms->i8042_enabled;
1721}
1722
1723static void pc_machine_set_i8042(Object *obj, bool value, Error **errp)
1724{
1725    PCMachineState *pcms = PC_MACHINE(obj);
1726
1727    pcms->i8042_enabled = value;
1728}
1729
1730static bool pc_machine_get_default_bus_bypass_iommu(Object *obj, Error **errp)
1731{
1732    PCMachineState *pcms = PC_MACHINE(obj);
1733
1734    return pcms->default_bus_bypass_iommu;
1735}
1736
1737static void pc_machine_set_default_bus_bypass_iommu(Object *obj, bool value,
1738                                                    Error **errp)
1739{
1740    PCMachineState *pcms = PC_MACHINE(obj);
1741
1742    pcms->default_bus_bypass_iommu = value;
1743}
1744
1745static void pc_machine_get_smbios_ep(Object *obj, Visitor *v, const char *name,
1746                                     void *opaque, Error **errp)
1747{
1748    PCMachineState *pcms = PC_MACHINE(obj);
1749    SmbiosEntryPointType smbios_entry_point_type = pcms->smbios_entry_point_type;
1750
1751    visit_type_SmbiosEntryPointType(v, name, &smbios_entry_point_type, errp);
1752}
1753
1754static void pc_machine_set_smbios_ep(Object *obj, Visitor *v, const char *name,
1755                                     void *opaque, Error **errp)
1756{
1757    PCMachineState *pcms = PC_MACHINE(obj);
1758
1759    visit_type_SmbiosEntryPointType(v, name, &pcms->smbios_entry_point_type, errp);
1760}
1761
1762static void pc_machine_get_max_ram_below_4g(Object *obj, Visitor *v,
1763                                            const char *name, void *opaque,
1764                                            Error **errp)
1765{
1766    PCMachineState *pcms = PC_MACHINE(obj);
1767    uint64_t value = pcms->max_ram_below_4g;
1768
1769    visit_type_size(v, name, &value, errp);
1770}
1771
1772static void pc_machine_set_max_ram_below_4g(Object *obj, Visitor *v,
1773                                            const char *name, void *opaque,
1774                                            Error **errp)
1775{
1776    PCMachineState *pcms = PC_MACHINE(obj);
1777    uint64_t value;
1778
1779    if (!visit_type_size(v, name, &value, errp)) {
1780        return;
1781    }
1782    if (value > 4 * GiB) {
1783        error_setg(errp,
1784                   "Machine option 'max-ram-below-4g=%"PRIu64
1785                   "' expects size less than or equal to 4G", value);
1786        return;
1787    }
1788
1789    if (value < 1 * MiB) {
1790        warn_report("Only %" PRIu64 " bytes of RAM below the 4GiB boundary,"
1791                    "BIOS may not work with less than 1MiB", value);
1792    }
1793
1794    pcms->max_ram_below_4g = value;
1795}
1796
1797static void pc_machine_get_max_fw_size(Object *obj, Visitor *v,
1798                                       const char *name, void *opaque,
1799                                       Error **errp)
1800{
1801    PCMachineState *pcms = PC_MACHINE(obj);
1802    uint64_t value = pcms->max_fw_size;
1803
1804    visit_type_size(v, name, &value, errp);
1805}
1806
1807static void pc_machine_set_max_fw_size(Object *obj, Visitor *v,
1808                                       const char *name, void *opaque,
1809                                       Error **errp)
1810{
1811    PCMachineState *pcms = PC_MACHINE(obj);
1812    uint64_t value;
1813
1814    if (!visit_type_size(v, name, &value, errp)) {
1815        return;
1816    }
1817
1818    /*
1819    * We don't have a theoretically justifiable exact lower bound on the base
1820    * address of any flash mapping. In practice, the IO-APIC MMIO range is
1821    * [0xFEE00000..0xFEE01000] -- see IO_APIC_DEFAULT_ADDRESS --, leaving free
1822    * only 18MB-4KB below 4G. For now, restrict the cumulative mapping to 8MB in
1823    * size.
1824    */
1825    if (value > 16 * MiB) {
1826        error_setg(errp,
1827                   "User specified max allowed firmware size %" PRIu64 " is "
1828                   "greater than 16MiB. If combined firwmare size exceeds "
1829                   "16MiB the system may not boot, or experience intermittent"
1830                   "stability issues.",
1831                   value);
1832        return;
1833    }
1834
1835    pcms->max_fw_size = value;
1836}
1837
1838
1839static void pc_machine_initfn(Object *obj)
1840{
1841    PCMachineState *pcms = PC_MACHINE(obj);
1842
1843#ifdef CONFIG_VMPORT
1844    pcms->vmport = ON_OFF_AUTO_AUTO;
1845#else
1846    pcms->vmport = ON_OFF_AUTO_OFF;
1847#endif /* CONFIG_VMPORT */
1848    pcms->max_ram_below_4g = 0; /* use default */
1849    pcms->smbios_entry_point_type = SMBIOS_ENTRY_POINT_TYPE_32;
1850
1851    /* acpi build is enabled by default if machine supports it */
1852    pcms->acpi_build_enabled = PC_MACHINE_GET_CLASS(pcms)->has_acpi_build;
1853    pcms->smbus_enabled = true;
1854    pcms->sata_enabled = true;
1855    pcms->i8042_enabled = true;
1856    pcms->max_fw_size = 8 * MiB;
1857#ifdef CONFIG_HPET
1858    pcms->hpet_enabled = true;
1859#endif
1860    pcms->default_bus_bypass_iommu = false;
1861
1862    pc_system_flash_create(pcms);
1863    pcms->pcspk = isa_new(TYPE_PC_SPEAKER);
1864    object_property_add_alias(OBJECT(pcms), "pcspk-audiodev",
1865                              OBJECT(pcms->pcspk), "audiodev");
1866    cxl_machine_init(obj, &pcms->cxl_devices_state);
1867}
1868
1869int pc_machine_kvm_type(MachineState *machine, const char *kvm_type)
1870{
1871#ifdef CONFIG_XEN_EMU
1872    if (xen_mode == XEN_EMULATE) {
1873        xen_overlay_create();
1874        xen_evtchn_create();
1875        xen_gnttab_create();
1876        xen_xenstore_create();
1877    }
1878#endif
1879    return 0;
1880}
1881
1882static void pc_machine_reset(MachineState *machine, ShutdownCause reason)
1883{
1884    CPUState *cs;
1885    X86CPU *cpu;
1886
1887    qemu_devices_reset(reason);
1888
1889    /* Reset APIC after devices have been reset to cancel
1890     * any changes that qemu_devices_reset() might have done.
1891     */
1892    CPU_FOREACH(cs) {
1893        cpu = X86_CPU(cs);
1894
1895        x86_cpu_after_reset(cpu);
1896    }
1897}
1898
1899static void pc_machine_wakeup(MachineState *machine)
1900{
1901    cpu_synchronize_all_states();
1902    pc_machine_reset(machine, SHUTDOWN_CAUSE_NONE);
1903    cpu_synchronize_all_post_reset();
1904}
1905
1906static bool pc_hotplug_allowed(MachineState *ms, DeviceState *dev, Error **errp)
1907{
1908    X86IOMMUState *iommu = x86_iommu_get_default();
1909    IntelIOMMUState *intel_iommu;
1910
1911    if (iommu &&
1912        object_dynamic_cast((Object *)iommu, TYPE_INTEL_IOMMU_DEVICE) &&
1913        object_dynamic_cast((Object *)dev, "vfio-pci")) {
1914        intel_iommu = INTEL_IOMMU_DEVICE(iommu);
1915        if (!intel_iommu->caching_mode) {
1916            error_setg(errp, "Device assignment is not allowed without "
1917                       "enabling caching-mode=on for Intel IOMMU.");
1918            return false;
1919        }
1920    }
1921
1922    return true;
1923}
1924
1925static void pc_machine_class_init(ObjectClass *oc, void *data)
1926{
1927    MachineClass *mc = MACHINE_CLASS(oc);
1928    PCMachineClass *pcmc = PC_MACHINE_CLASS(oc);
1929    HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
1930
1931    pcmc->pci_enabled = true;
1932    pcmc->has_acpi_build = true;
1933    pcmc->rsdp_in_ram = true;
1934    pcmc->smbios_defaults = true;
1935    pcmc->smbios_uuid_encoded = true;
1936    pcmc->gigabyte_align = true;
1937    pcmc->has_reserved_memory = true;
1938    pcmc->kvmclock_enabled = true;
1939    pcmc->enforce_aligned_dimm = true;
1940    pcmc->enforce_amd_1tb_hole = true;
1941    /* BIOS ACPI tables: 128K. Other BIOS datastructures: less than 4K reported
1942     * to be used at the moment, 32K should be enough for a while.  */
1943    pcmc->acpi_data_size = 0x20000 + 0x8000;
1944    pcmc->pvh_enabled = true;
1945    pcmc->kvmclock_create_always = true;
1946    assert(!mc->get_hotplug_handler);
1947    mc->get_hotplug_handler = pc_get_hotplug_handler;
1948    mc->hotplug_allowed = pc_hotplug_allowed;
1949    mc->cpu_index_to_instance_props = x86_cpu_index_to_props;
1950    mc->get_default_cpu_node_id = x86_get_default_cpu_node_id;
1951    mc->possible_cpu_arch_ids = x86_possible_cpu_arch_ids;
1952    mc->auto_enable_numa_with_memhp = true;
1953    mc->auto_enable_numa_with_memdev = true;
1954    mc->has_hotpluggable_cpus = true;
1955    mc->default_boot_order = "cad";
1956    mc->block_default_type = IF_IDE;
1957    mc->max_cpus = 255;
1958    mc->reset = pc_machine_reset;
1959    mc->wakeup = pc_machine_wakeup;
1960    hc->pre_plug = pc_machine_device_pre_plug_cb;
1961    hc->plug = pc_machine_device_plug_cb;
1962    hc->unplug_request = pc_machine_device_unplug_request_cb;
1963    hc->unplug = pc_machine_device_unplug_cb;
1964    mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE;
1965    mc->nvdimm_supported = true;
1966    mc->smp_props.dies_supported = true;
1967    mc->default_ram_id = "pc.ram";
1968
1969    object_class_property_add(oc, PC_MACHINE_MAX_RAM_BELOW_4G, "size",
1970        pc_machine_get_max_ram_below_4g, pc_machine_set_max_ram_below_4g,
1971        NULL, NULL);
1972    object_class_property_set_description(oc, PC_MACHINE_MAX_RAM_BELOW_4G,
1973        "Maximum ram below the 4G boundary (32bit boundary)");
1974
1975    object_class_property_add(oc, PC_MACHINE_DEVMEM_REGION_SIZE, "int",
1976        pc_machine_get_device_memory_region_size, NULL,
1977        NULL, NULL);
1978
1979    object_class_property_add(oc, PC_MACHINE_VMPORT, "OnOffAuto",
1980        pc_machine_get_vmport, pc_machine_set_vmport,
1981        NULL, NULL);
1982    object_class_property_set_description(oc, PC_MACHINE_VMPORT,
1983        "Enable vmport (pc & q35)");
1984
1985    object_class_property_add_bool(oc, PC_MACHINE_SMBUS,
1986        pc_machine_get_smbus, pc_machine_set_smbus);
1987    object_class_property_set_description(oc, PC_MACHINE_SMBUS,
1988        "Enable/disable system management bus");
1989
1990    object_class_property_add_bool(oc, PC_MACHINE_SATA,
1991        pc_machine_get_sata, pc_machine_set_sata);
1992    object_class_property_set_description(oc, PC_MACHINE_SATA,
1993        "Enable/disable Serial ATA bus");
1994
1995    object_class_property_add_bool(oc, "hpet",
1996        pc_machine_get_hpet, pc_machine_set_hpet);
1997    object_class_property_set_description(oc, "hpet",
1998        "Enable/disable high precision event timer emulation");
1999
2000    object_class_property_add_bool(oc, PC_MACHINE_I8042,
2001        pc_machine_get_i8042, pc_machine_set_i8042);
2002
2003    object_class_property_add_bool(oc, "default-bus-bypass-iommu",
2004        pc_machine_get_default_bus_bypass_iommu,
2005        pc_machine_set_default_bus_bypass_iommu);
2006
2007    object_class_property_add(oc, PC_MACHINE_MAX_FW_SIZE, "size",
2008        pc_machine_get_max_fw_size, pc_machine_set_max_fw_size,
2009        NULL, NULL);
2010    object_class_property_set_description(oc, PC_MACHINE_MAX_FW_SIZE,
2011        "Maximum combined firmware size");
2012
2013    object_class_property_add(oc, PC_MACHINE_SMBIOS_EP, "str",
2014        pc_machine_get_smbios_ep, pc_machine_set_smbios_ep,
2015        NULL, NULL);
2016    object_class_property_set_description(oc, PC_MACHINE_SMBIOS_EP,
2017        "SMBIOS Entry Point type [32, 64]");
2018}
2019
2020static const TypeInfo pc_machine_info = {
2021    .name = TYPE_PC_MACHINE,
2022    .parent = TYPE_X86_MACHINE,
2023    .abstract = true,
2024    .instance_size = sizeof(PCMachineState),
2025    .instance_init = pc_machine_initfn,
2026    .class_size = sizeof(PCMachineClass),
2027    .class_init = pc_machine_class_init,
2028    .interfaces = (InterfaceInfo[]) {
2029         { TYPE_HOTPLUG_HANDLER },
2030         { }
2031    },
2032};
2033
2034static void pc_machine_register_types(void)
2035{
2036    type_register_static(&pc_machine_info);
2037}
2038
2039type_init(pc_machine_register_types)
2040