xref: /qemu/hw/i386/pc.c (revision 6a015046606ebf260950605ec48fc6420422f43c)
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.h"
37 #include "hw/pci/pci.h"
38 #include "hw/pci/pci_bus.h"
39 #include "hw/pci-bridge/pci_expander_bridge.h"
40 #include "hw/nvram/fw_cfg.h"
41 #include "hw/timer/hpet.h"
42 #include "hw/firmware/smbios.h"
43 #include "hw/loader.h"
44 #include "elf.h"
45 #include "migration/vmstate.h"
46 #include "multiboot.h"
47 #include "hw/rtc/mc146818rtc.h"
48 #include "hw/intc/i8259.h"
49 #include "hw/intc/ioapic.h"
50 #include "hw/timer/i8254.h"
51 #include "hw/input/i8042.h"
52 #include "hw/irq.h"
53 #include "hw/audio/pcspk.h"
54 #include "hw/pci/msi.h"
55 #include "hw/sysbus.h"
56 #include "sysemu/sysemu.h"
57 #include "sysemu/tcg.h"
58 #include "sysemu/numa.h"
59 #include "sysemu/kvm.h"
60 #include "sysemu/xen.h"
61 #include "sysemu/reset.h"
62 #include "sysemu/runstate.h"
63 #include "kvm/kvm_i386.h"
64 #include "hw/xen/xen.h"
65 #include "hw/xen/start_info.h"
66 #include "ui/qemu-spice.h"
67 #include "exec/memory.h"
68 #include "qemu/bitmap.h"
69 #include "qemu/config-file.h"
70 #include "qemu/error-report.h"
71 #include "qemu/option.h"
72 #include "qemu/cutils.h"
73 #include "hw/acpi/acpi.h"
74 #include "hw/acpi/cpu_hotplug.h"
75 #include "acpi-build.h"
76 #include "hw/mem/pc-dimm.h"
77 #include "hw/mem/nvdimm.h"
78 #include "hw/cxl/cxl.h"
79 #include "hw/cxl/cxl_host.h"
80 #include "qapi/error.h"
81 #include "qapi/qapi-visit-common.h"
82 #include "qapi/qapi-visit-machine.h"
83 #include "qapi/visitor.h"
84 #include "hw/core/cpu.h"
85 #include "hw/usb.h"
86 #include "hw/i386/intel_iommu.h"
87 #include "hw/net/ne2000-isa.h"
88 #include "standard-headers/asm-x86/bootparam.h"
89 #include "hw/virtio/virtio-iommu.h"
90 #include "hw/virtio/virtio-pmem-pci.h"
91 #include "hw/virtio/virtio-mem-pci.h"
92 #include "hw/mem/memory-device.h"
93 #include "sysemu/replay.h"
94 #include "target/i386/cpu.h"
95 #include "e820_memory_layout.h"
96 #include "fw_cfg.h"
97 #include "trace.h"
98 #include CONFIG_DEVICES
99 
100 /*
101  * Helper for setting model-id for CPU models that changed model-id
102  * depending on QEMU versions up to QEMU 2.4.
103  */
104 #define PC_CPU_MODEL_IDS(v) \
105     { "qemu32-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
106     { "qemu64-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
107     { "athlon-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },
108 
109 GlobalProperty pc_compat_7_2[] = {
110     { "ICH9-LPC", "noreboot", "true" },
111 };
112 const size_t pc_compat_7_2_len = G_N_ELEMENTS(pc_compat_7_2);
113 
114 GlobalProperty pc_compat_7_1[] = {};
115 const size_t pc_compat_7_1_len = G_N_ELEMENTS(pc_compat_7_1);
116 
117 GlobalProperty pc_compat_7_0[] = {};
118 const size_t pc_compat_7_0_len = G_N_ELEMENTS(pc_compat_7_0);
119 
120 GlobalProperty pc_compat_6_2[] = {
121     { "virtio-mem", "unplugged-inaccessible", "off" },
122 };
123 const size_t pc_compat_6_2_len = G_N_ELEMENTS(pc_compat_6_2);
124 
125 GlobalProperty pc_compat_6_1[] = {
126     { TYPE_X86_CPU, "hv-version-id-build", "0x1bbc" },
127     { TYPE_X86_CPU, "hv-version-id-major", "0x0006" },
128     { TYPE_X86_CPU, "hv-version-id-minor", "0x0001" },
129     { "ICH9-LPC", "x-keep-pci-slot-hpc", "false" },
130 };
131 const size_t pc_compat_6_1_len = G_N_ELEMENTS(pc_compat_6_1);
132 
133 GlobalProperty pc_compat_6_0[] = {
134     { "qemu64" "-" TYPE_X86_CPU, "family", "6" },
135     { "qemu64" "-" TYPE_X86_CPU, "model", "6" },
136     { "qemu64" "-" TYPE_X86_CPU, "stepping", "3" },
137     { TYPE_X86_CPU, "x-vendor-cpuid-only", "off" },
138     { "ICH9-LPC", ACPI_PM_PROP_ACPI_PCIHP_BRIDGE, "off" },
139     { "ICH9-LPC", "x-keep-pci-slot-hpc", "true" },
140 };
141 const size_t pc_compat_6_0_len = G_N_ELEMENTS(pc_compat_6_0);
142 
143 GlobalProperty pc_compat_5_2[] = {
144     { "ICH9-LPC", "x-smi-cpu-hotunplug", "off" },
145 };
146 const size_t pc_compat_5_2_len = G_N_ELEMENTS(pc_compat_5_2);
147 
148 GlobalProperty pc_compat_5_1[] = {
149     { "ICH9-LPC", "x-smi-cpu-hotplug", "off" },
150     { TYPE_X86_CPU, "kvm-msi-ext-dest-id", "off" },
151 };
152 const size_t pc_compat_5_1_len = G_N_ELEMENTS(pc_compat_5_1);
153 
154 GlobalProperty pc_compat_5_0[] = {
155 };
156 const size_t pc_compat_5_0_len = G_N_ELEMENTS(pc_compat_5_0);
157 
158 GlobalProperty pc_compat_4_2[] = {
159     { "mch", "smbase-smram", "off" },
160 };
161 const size_t pc_compat_4_2_len = G_N_ELEMENTS(pc_compat_4_2);
162 
163 GlobalProperty pc_compat_4_1[] = {};
164 const size_t pc_compat_4_1_len = G_N_ELEMENTS(pc_compat_4_1);
165 
166 GlobalProperty pc_compat_4_0[] = {};
167 const size_t pc_compat_4_0_len = G_N_ELEMENTS(pc_compat_4_0);
168 
169 GlobalProperty pc_compat_3_1[] = {
170     { "intel-iommu", "dma-drain", "off" },
171     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "off" },
172     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "off" },
173     { "Opteron_G4" "-" TYPE_X86_CPU, "npt", "off" },
174     { "Opteron_G4" "-" TYPE_X86_CPU, "nrip-save", "off" },
175     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "off" },
176     { "Opteron_G5" "-" TYPE_X86_CPU, "npt", "off" },
177     { "Opteron_G5" "-" TYPE_X86_CPU, "nrip-save", "off" },
178     { "EPYC" "-" TYPE_X86_CPU, "npt", "off" },
179     { "EPYC" "-" TYPE_X86_CPU, "nrip-save", "off" },
180     { "EPYC-IBPB" "-" TYPE_X86_CPU, "npt", "off" },
181     { "EPYC-IBPB" "-" TYPE_X86_CPU, "nrip-save", "off" },
182     { "Skylake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
183     { "Skylake-Client-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
184     { "Skylake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
185     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
186     { "Cascadelake-Server" "-" TYPE_X86_CPU,  "mpx", "on" },
187     { "Icelake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
188     { "Icelake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
189     { "Cascadelake-Server" "-" TYPE_X86_CPU, "stepping", "5" },
190     { TYPE_X86_CPU, "x-intel-pt-auto-level", "off" },
191 };
192 const size_t pc_compat_3_1_len = G_N_ELEMENTS(pc_compat_3_1);
193 
194 GlobalProperty pc_compat_3_0[] = {
195     { TYPE_X86_CPU, "x-hv-synic-kvm-only", "on" },
196     { "Skylake-Server" "-" TYPE_X86_CPU, "pku", "off" },
197     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "pku", "off" },
198 };
199 const size_t pc_compat_3_0_len = G_N_ELEMENTS(pc_compat_3_0);
200 
201 GlobalProperty pc_compat_2_12[] = {
202     { TYPE_X86_CPU, "legacy-cache", "on" },
203     { TYPE_X86_CPU, "topoext", "off" },
204     { "EPYC-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
205     { "EPYC-IBPB-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
206 };
207 const size_t pc_compat_2_12_len = G_N_ELEMENTS(pc_compat_2_12);
208 
209 GlobalProperty pc_compat_2_11[] = {
210     { TYPE_X86_CPU, "x-migrate-smi-count", "off" },
211     { "Skylake-Server" "-" TYPE_X86_CPU, "clflushopt", "off" },
212 };
213 const size_t pc_compat_2_11_len = G_N_ELEMENTS(pc_compat_2_11);
214 
215 GlobalProperty pc_compat_2_10[] = {
216     { TYPE_X86_CPU, "x-hv-max-vps", "0x40" },
217     { "i440FX-pcihost", "x-pci-hole64-fix", "off" },
218     { "q35-pcihost", "x-pci-hole64-fix", "off" },
219 };
220 const size_t pc_compat_2_10_len = G_N_ELEMENTS(pc_compat_2_10);
221 
222 GlobalProperty pc_compat_2_9[] = {
223     { "mch", "extended-tseg-mbytes", "0" },
224 };
225 const size_t pc_compat_2_9_len = G_N_ELEMENTS(pc_compat_2_9);
226 
227 GlobalProperty pc_compat_2_8[] = {
228     { TYPE_X86_CPU, "tcg-cpuid", "off" },
229     { "kvmclock", "x-mach-use-reliable-get-clock", "off" },
230     { "ICH9-LPC", "x-smi-broadcast", "off" },
231     { TYPE_X86_CPU, "vmware-cpuid-freq", "off" },
232     { "Haswell-" TYPE_X86_CPU, "stepping", "1" },
233 };
234 const size_t pc_compat_2_8_len = G_N_ELEMENTS(pc_compat_2_8);
235 
236 GlobalProperty pc_compat_2_7[] = {
237     { TYPE_X86_CPU, "l3-cache", "off" },
238     { TYPE_X86_CPU, "full-cpuid-auto-level", "off" },
239     { "Opteron_G3" "-" TYPE_X86_CPU, "family", "15" },
240     { "Opteron_G3" "-" TYPE_X86_CPU, "model", "6" },
241     { "Opteron_G3" "-" TYPE_X86_CPU, "stepping", "1" },
242     { "isa-pcspk", "migrate", "off" },
243 };
244 const size_t pc_compat_2_7_len = G_N_ELEMENTS(pc_compat_2_7);
245 
246 GlobalProperty pc_compat_2_6[] = {
247     { TYPE_X86_CPU, "cpuid-0xb", "off" },
248     { "vmxnet3", "romfile", "" },
249     { TYPE_X86_CPU, "fill-mtrr-mask", "off" },
250     { "apic-common", "legacy-instance-id", "on", }
251 };
252 const size_t pc_compat_2_6_len = G_N_ELEMENTS(pc_compat_2_6);
253 
254 GlobalProperty pc_compat_2_5[] = {};
255 const size_t pc_compat_2_5_len = G_N_ELEMENTS(pc_compat_2_5);
256 
257 GlobalProperty pc_compat_2_4[] = {
258     PC_CPU_MODEL_IDS("2.4.0")
259     { "Haswell-" TYPE_X86_CPU, "abm", "off" },
260     { "Haswell-noTSX-" TYPE_X86_CPU, "abm", "off" },
261     { "Broadwell-" TYPE_X86_CPU, "abm", "off" },
262     { "Broadwell-noTSX-" TYPE_X86_CPU, "abm", "off" },
263     { "host" "-" TYPE_X86_CPU, "host-cache-info", "on" },
264     { TYPE_X86_CPU, "check", "off" },
265     { "qemu64" "-" TYPE_X86_CPU, "sse4a", "on" },
266     { "qemu64" "-" TYPE_X86_CPU, "abm", "on" },
267     { "qemu64" "-" TYPE_X86_CPU, "popcnt", "on" },
268     { "qemu32" "-" TYPE_X86_CPU, "popcnt", "on" },
269     { "Opteron_G2" "-" TYPE_X86_CPU, "rdtscp", "on" },
270     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "on" },
271     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "on" },
272     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "on", }
273 };
274 const size_t pc_compat_2_4_len = G_N_ELEMENTS(pc_compat_2_4);
275 
276 GlobalProperty pc_compat_2_3[] = {
277     PC_CPU_MODEL_IDS("2.3.0")
278     { TYPE_X86_CPU, "arat", "off" },
279     { "qemu64" "-" TYPE_X86_CPU, "min-level", "4" },
280     { "kvm64" "-" TYPE_X86_CPU, "min-level", "5" },
281     { "pentium3" "-" TYPE_X86_CPU, "min-level", "2" },
282     { "n270" "-" TYPE_X86_CPU, "min-level", "5" },
283     { "Conroe" "-" TYPE_X86_CPU, "min-level", "4" },
284     { "Penryn" "-" TYPE_X86_CPU, "min-level", "4" },
285     { "Nehalem" "-" TYPE_X86_CPU, "min-level", "4" },
286     { "n270" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
287     { "Penryn" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
288     { "Conroe" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
289     { "Nehalem" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
290     { "Westmere" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
291     { "SandyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
292     { "IvyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
293     { "Haswell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
294     { "Haswell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
295     { "Broadwell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
296     { "Broadwell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
297     { TYPE_X86_CPU, "kvm-no-smi-migration", "on" },
298 };
299 const size_t pc_compat_2_3_len = G_N_ELEMENTS(pc_compat_2_3);
300 
301 GlobalProperty pc_compat_2_2[] = {
302     PC_CPU_MODEL_IDS("2.2.0")
303     { "kvm64" "-" TYPE_X86_CPU, "vme", "off" },
304     { "kvm32" "-" TYPE_X86_CPU, "vme", "off" },
305     { "Conroe" "-" TYPE_X86_CPU, "vme", "off" },
306     { "Penryn" "-" TYPE_X86_CPU, "vme", "off" },
307     { "Nehalem" "-" TYPE_X86_CPU, "vme", "off" },
308     { "Westmere" "-" TYPE_X86_CPU, "vme", "off" },
309     { "SandyBridge" "-" TYPE_X86_CPU, "vme", "off" },
310     { "Haswell" "-" TYPE_X86_CPU, "vme", "off" },
311     { "Broadwell" "-" TYPE_X86_CPU, "vme", "off" },
312     { "Opteron_G1" "-" TYPE_X86_CPU, "vme", "off" },
313     { "Opteron_G2" "-" TYPE_X86_CPU, "vme", "off" },
314     { "Opteron_G3" "-" TYPE_X86_CPU, "vme", "off" },
315     { "Opteron_G4" "-" TYPE_X86_CPU, "vme", "off" },
316     { "Opteron_G5" "-" TYPE_X86_CPU, "vme", "off" },
317     { "Haswell" "-" TYPE_X86_CPU, "f16c", "off" },
318     { "Haswell" "-" TYPE_X86_CPU, "rdrand", "off" },
319     { "Broadwell" "-" TYPE_X86_CPU, "f16c", "off" },
320     { "Broadwell" "-" TYPE_X86_CPU, "rdrand", "off" },
321 };
322 const size_t pc_compat_2_2_len = G_N_ELEMENTS(pc_compat_2_2);
323 
324 GlobalProperty pc_compat_2_1[] = {
325     PC_CPU_MODEL_IDS("2.1.0")
326     { "coreduo" "-" TYPE_X86_CPU, "vmx", "on" },
327     { "core2duo" "-" TYPE_X86_CPU, "vmx", "on" },
328 };
329 const size_t pc_compat_2_1_len = G_N_ELEMENTS(pc_compat_2_1);
330 
331 GlobalProperty pc_compat_2_0[] = {
332     PC_CPU_MODEL_IDS("2.0.0")
333     { "virtio-scsi-pci", "any_layout", "off" },
334     { "PIIX4_PM", "memory-hotplug-support", "off" },
335     { "apic", "version", "0x11" },
336     { "nec-usb-xhci", "superspeed-ports-first", "off" },
337     { "nec-usb-xhci", "force-pcie-endcap", "on" },
338     { "pci-serial", "prog_if", "0" },
339     { "pci-serial-2x", "prog_if", "0" },
340     { "pci-serial-4x", "prog_if", "0" },
341     { "virtio-net-pci", "guest_announce", "off" },
342     { "ICH9-LPC", "memory-hotplug-support", "off" },
343 };
344 const size_t pc_compat_2_0_len = G_N_ELEMENTS(pc_compat_2_0);
345 
346 GlobalProperty pc_compat_1_7[] = {
347     PC_CPU_MODEL_IDS("1.7.0")
348     { TYPE_USB_DEVICE, "msos-desc", "no" },
349     { "PIIX4_PM", ACPI_PM_PROP_ACPI_PCIHP_BRIDGE, "off" },
350     { "hpet", HPET_INTCAP, "4" },
351 };
352 const size_t pc_compat_1_7_len = G_N_ELEMENTS(pc_compat_1_7);
353 
354 GlobalProperty pc_compat_1_6[] = {
355     PC_CPU_MODEL_IDS("1.6.0")
356     { "e1000", "mitigation", "off" },
357     { "qemu64-" TYPE_X86_CPU, "model", "2" },
358     { "qemu32-" TYPE_X86_CPU, "model", "3" },
359     { "i440FX-pcihost", "short_root_bus", "1" },
360     { "q35-pcihost", "short_root_bus", "1" },
361 };
362 const size_t pc_compat_1_6_len = G_N_ELEMENTS(pc_compat_1_6);
363 
364 GlobalProperty pc_compat_1_5[] = {
365     PC_CPU_MODEL_IDS("1.5.0")
366     { "Conroe-" TYPE_X86_CPU, "model", "2" },
367     { "Conroe-" TYPE_X86_CPU, "min-level", "2" },
368     { "Penryn-" TYPE_X86_CPU, "model", "2" },
369     { "Penryn-" TYPE_X86_CPU, "min-level", "2" },
370     { "Nehalem-" TYPE_X86_CPU, "model", "2" },
371     { "Nehalem-" TYPE_X86_CPU, "min-level", "2" },
372     { "virtio-net-pci", "any_layout", "off" },
373     { TYPE_X86_CPU, "pmu", "on" },
374     { "i440FX-pcihost", "short_root_bus", "0" },
375     { "q35-pcihost", "short_root_bus", "0" },
376 };
377 const size_t pc_compat_1_5_len = G_N_ELEMENTS(pc_compat_1_5);
378 
379 GlobalProperty pc_compat_1_4[] = {
380     PC_CPU_MODEL_IDS("1.4.0")
381     { "scsi-hd", "discard_granularity", "0" },
382     { "scsi-cd", "discard_granularity", "0" },
383     { "ide-hd", "discard_granularity", "0" },
384     { "ide-cd", "discard_granularity", "0" },
385     { "virtio-blk-pci", "discard_granularity", "0" },
386     /* DEV_NVECTORS_UNSPECIFIED as a uint32_t string: */
387     { "virtio-serial-pci", "vectors", "0xFFFFFFFF" },
388     { "virtio-net-pci", "ctrl_guest_offloads", "off" },
389     { "e1000", "romfile", "pxe-e1000.rom" },
390     { "ne2k_pci", "romfile", "pxe-ne2k_pci.rom" },
391     { "pcnet", "romfile", "pxe-pcnet.rom" },
392     { "rtl8139", "romfile", "pxe-rtl8139.rom" },
393     { "virtio-net-pci", "romfile", "pxe-virtio.rom" },
394     { "486-" TYPE_X86_CPU, "model", "0" },
395     { "n270" "-" TYPE_X86_CPU, "movbe", "off" },
396     { "Westmere" "-" TYPE_X86_CPU, "pclmulqdq", "off" },
397 };
398 const size_t pc_compat_1_4_len = G_N_ELEMENTS(pc_compat_1_4);
399 
400 GSIState *pc_gsi_create(qemu_irq **irqs, bool pci_enabled)
401 {
402     GSIState *s;
403 
404     s = g_new0(GSIState, 1);
405     if (kvm_ioapic_in_kernel()) {
406         kvm_pc_setup_irq_routing(pci_enabled);
407     }
408     *irqs = qemu_allocate_irqs(gsi_handler, s, IOAPIC_NUM_PINS);
409 
410     return s;
411 }
412 
413 static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
414                            unsigned size)
415 {
416 }
417 
418 static uint64_t ioport80_read(void *opaque, hwaddr addr, unsigned size)
419 {
420     return 0xffffffffffffffffULL;
421 }
422 
423 /* MSDOS compatibility mode FPU exception support */
424 static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
425                            unsigned size)
426 {
427     if (tcg_enabled()) {
428         cpu_set_ignne();
429     }
430 }
431 
432 static uint64_t ioportF0_read(void *opaque, hwaddr addr, unsigned size)
433 {
434     return 0xffffffffffffffffULL;
435 }
436 
437 /* PC cmos mappings */
438 
439 #define REG_EQUIPMENT_BYTE          0x14
440 
441 static void cmos_init_hd(MC146818RtcState *s, int type_ofs, int info_ofs,
442                          int16_t cylinders, int8_t heads, int8_t sectors)
443 {
444     mc146818rtc_set_cmos_data(s, type_ofs, 47);
445     mc146818rtc_set_cmos_data(s, info_ofs, cylinders);
446     mc146818rtc_set_cmos_data(s, info_ofs + 1, cylinders >> 8);
447     mc146818rtc_set_cmos_data(s, info_ofs + 2, heads);
448     mc146818rtc_set_cmos_data(s, info_ofs + 3, 0xff);
449     mc146818rtc_set_cmos_data(s, info_ofs + 4, 0xff);
450     mc146818rtc_set_cmos_data(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
451     mc146818rtc_set_cmos_data(s, info_ofs + 6, cylinders);
452     mc146818rtc_set_cmos_data(s, info_ofs + 7, cylinders >> 8);
453     mc146818rtc_set_cmos_data(s, info_ofs + 8, sectors);
454 }
455 
456 /* convert boot_device letter to something recognizable by the bios */
457 static int boot_device2nibble(char boot_device)
458 {
459     switch(boot_device) {
460     case 'a':
461     case 'b':
462         return 0x01; /* floppy boot */
463     case 'c':
464         return 0x02; /* hard drive boot */
465     case 'd':
466         return 0x03; /* CD-ROM boot */
467     case 'n':
468         return 0x04; /* Network boot */
469     }
470     return 0;
471 }
472 
473 static void set_boot_dev(MC146818RtcState *s, const char *boot_device,
474                          Error **errp)
475 {
476 #define PC_MAX_BOOT_DEVICES 3
477     int nbds, bds[3] = { 0, };
478     int i;
479 
480     nbds = strlen(boot_device);
481     if (nbds > PC_MAX_BOOT_DEVICES) {
482         error_setg(errp, "Too many boot devices for PC");
483         return;
484     }
485     for (i = 0; i < nbds; i++) {
486         bds[i] = boot_device2nibble(boot_device[i]);
487         if (bds[i] == 0) {
488             error_setg(errp, "Invalid boot device for PC: '%c'",
489                        boot_device[i]);
490             return;
491         }
492     }
493     mc146818rtc_set_cmos_data(s, 0x3d, (bds[1] << 4) | bds[0]);
494     mc146818rtc_set_cmos_data(s, 0x38, (bds[2] << 4) | (fd_bootchk ? 0x0 : 0x1));
495 }
496 
497 static void pc_boot_set(void *opaque, const char *boot_device, Error **errp)
498 {
499     set_boot_dev(opaque, boot_device, errp);
500 }
501 
502 static void pc_cmos_init_floppy(MC146818RtcState *rtc_state, ISADevice *floppy)
503 {
504     int val, nb, i;
505     FloppyDriveType fd_type[2] = { FLOPPY_DRIVE_TYPE_NONE,
506                                    FLOPPY_DRIVE_TYPE_NONE };
507 
508     /* floppy type */
509     if (floppy) {
510         for (i = 0; i < 2; i++) {
511             fd_type[i] = isa_fdc_get_drive_type(floppy, i);
512         }
513     }
514     val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
515         cmos_get_fd_drive_type(fd_type[1]);
516     mc146818rtc_set_cmos_data(rtc_state, 0x10, val);
517 
518     val = mc146818rtc_get_cmos_data(rtc_state, REG_EQUIPMENT_BYTE);
519     nb = 0;
520     if (fd_type[0] != FLOPPY_DRIVE_TYPE_NONE) {
521         nb++;
522     }
523     if (fd_type[1] != FLOPPY_DRIVE_TYPE_NONE) {
524         nb++;
525     }
526     switch (nb) {
527     case 0:
528         break;
529     case 1:
530         val |= 0x01; /* 1 drive, ready for boot */
531         break;
532     case 2:
533         val |= 0x41; /* 2 drives, ready for boot */
534         break;
535     }
536     mc146818rtc_set_cmos_data(rtc_state, REG_EQUIPMENT_BYTE, val);
537 }
538 
539 typedef struct pc_cmos_init_late_arg {
540     MC146818RtcState *rtc_state;
541     BusState *idebus[2];
542 } pc_cmos_init_late_arg;
543 
544 typedef struct check_fdc_state {
545     ISADevice *floppy;
546     bool multiple;
547 } CheckFdcState;
548 
549 static int check_fdc(Object *obj, void *opaque)
550 {
551     CheckFdcState *state = opaque;
552     Object *fdc;
553     uint32_t iobase;
554     Error *local_err = NULL;
555 
556     fdc = object_dynamic_cast(obj, TYPE_ISA_FDC);
557     if (!fdc) {
558         return 0;
559     }
560 
561     iobase = object_property_get_uint(obj, "iobase", &local_err);
562     if (local_err || iobase != 0x3f0) {
563         error_free(local_err);
564         return 0;
565     }
566 
567     if (state->floppy) {
568         state->multiple = true;
569     } else {
570         state->floppy = ISA_DEVICE(obj);
571     }
572     return 0;
573 }
574 
575 static const char * const fdc_container_path[] = {
576     "/unattached", "/peripheral", "/peripheral-anon"
577 };
578 
579 /*
580  * Locate the FDC at IO address 0x3f0, in order to configure the CMOS registers
581  * and ACPI objects.
582  */
583 static ISADevice *pc_find_fdc0(void)
584 {
585     int i;
586     Object *container;
587     CheckFdcState state = { 0 };
588 
589     for (i = 0; i < ARRAY_SIZE(fdc_container_path); i++) {
590         container = container_get(qdev_get_machine(), fdc_container_path[i]);
591         object_child_foreach(container, check_fdc, &state);
592     }
593 
594     if (state.multiple) {
595         warn_report("multiple floppy disk controllers with "
596                     "iobase=0x3f0 have been found");
597         error_printf("the one being picked for CMOS setup might not reflect "
598                      "your intent");
599     }
600 
601     return state.floppy;
602 }
603 
604 static void pc_cmos_init_late(void *opaque)
605 {
606     pc_cmos_init_late_arg *arg = opaque;
607     MC146818RtcState *s = arg->rtc_state;
608     int16_t cylinders;
609     int8_t heads, sectors;
610     int val;
611     int i, trans;
612 
613     val = 0;
614     if (arg->idebus[0] && ide_get_geometry(arg->idebus[0], 0,
615                                            &cylinders, &heads, &sectors) >= 0) {
616         cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
617         val |= 0xf0;
618     }
619     if (arg->idebus[0] && ide_get_geometry(arg->idebus[0], 1,
620                                            &cylinders, &heads, &sectors) >= 0) {
621         cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
622         val |= 0x0f;
623     }
624     mc146818rtc_set_cmos_data(s, 0x12, val);
625 
626     val = 0;
627     for (i = 0; i < 4; i++) {
628         /* NOTE: ide_get_geometry() returns the physical
629            geometry.  It is always such that: 1 <= sects <= 63, 1
630            <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
631            geometry can be different if a translation is done. */
632         if (arg->idebus[i / 2] &&
633             ide_get_geometry(arg->idebus[i / 2], i % 2,
634                              &cylinders, &heads, &sectors) >= 0) {
635             trans = ide_get_bios_chs_trans(arg->idebus[i / 2], i % 2) - 1;
636             assert((trans & ~3) == 0);
637             val |= trans << (i * 2);
638         }
639     }
640     mc146818rtc_set_cmos_data(s, 0x39, val);
641 
642     pc_cmos_init_floppy(s, pc_find_fdc0());
643 
644     qemu_unregister_reset(pc_cmos_init_late, opaque);
645 }
646 
647 void pc_cmos_init(PCMachineState *pcms,
648                   BusState *idebus0, BusState *idebus1,
649                   ISADevice *rtc)
650 {
651     int val;
652     static pc_cmos_init_late_arg arg;
653     X86MachineState *x86ms = X86_MACHINE(pcms);
654     MC146818RtcState *s = MC146818_RTC(rtc);
655 
656     /* various important CMOS locations needed by PC/Bochs bios */
657 
658     /* memory size */
659     /* base memory (first MiB) */
660     val = MIN(x86ms->below_4g_mem_size / KiB, 640);
661     mc146818rtc_set_cmos_data(s, 0x15, val);
662     mc146818rtc_set_cmos_data(s, 0x16, val >> 8);
663     /* extended memory (next 64MiB) */
664     if (x86ms->below_4g_mem_size > 1 * MiB) {
665         val = (x86ms->below_4g_mem_size - 1 * MiB) / KiB;
666     } else {
667         val = 0;
668     }
669     if (val > 65535)
670         val = 65535;
671     mc146818rtc_set_cmos_data(s, 0x17, val);
672     mc146818rtc_set_cmos_data(s, 0x18, val >> 8);
673     mc146818rtc_set_cmos_data(s, 0x30, val);
674     mc146818rtc_set_cmos_data(s, 0x31, val >> 8);
675     /* memory between 16MiB and 4GiB */
676     if (x86ms->below_4g_mem_size > 16 * MiB) {
677         val = (x86ms->below_4g_mem_size - 16 * MiB) / (64 * KiB);
678     } else {
679         val = 0;
680     }
681     if (val > 65535)
682         val = 65535;
683     mc146818rtc_set_cmos_data(s, 0x34, val);
684     mc146818rtc_set_cmos_data(s, 0x35, val >> 8);
685     /* memory above 4GiB */
686     val = x86ms->above_4g_mem_size / 65536;
687     mc146818rtc_set_cmos_data(s, 0x5b, val);
688     mc146818rtc_set_cmos_data(s, 0x5c, val >> 8);
689     mc146818rtc_set_cmos_data(s, 0x5d, val >> 16);
690 
691     object_property_add_link(OBJECT(pcms), "rtc_state",
692                              TYPE_ISA_DEVICE,
693                              (Object **)&x86ms->rtc,
694                              object_property_allow_set_link,
695                              OBJ_PROP_LINK_STRONG);
696     object_property_set_link(OBJECT(pcms), "rtc_state", OBJECT(s),
697                              &error_abort);
698 
699     set_boot_dev(s, MACHINE(pcms)->boot_config.order, &error_fatal);
700 
701     val = 0;
702     val |= 0x02; /* FPU is there */
703     val |= 0x04; /* PS/2 mouse installed */
704     mc146818rtc_set_cmos_data(s, REG_EQUIPMENT_BYTE, val);
705 
706     /* hard drives and FDC */
707     arg.rtc_state = s;
708     arg.idebus[0] = idebus0;
709     arg.idebus[1] = idebus1;
710     qemu_register_reset(pc_cmos_init_late, &arg);
711 }
712 
713 static void handle_a20_line_change(void *opaque, int irq, int level)
714 {
715     X86CPU *cpu = opaque;
716 
717     /* XXX: send to all CPUs ? */
718     /* XXX: add logic to handle multiple A20 line sources */
719     x86_cpu_set_a20(cpu, level);
720 }
721 
722 #define NE2000_NB_MAX 6
723 
724 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
725                                               0x280, 0x380 };
726 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
727 
728 static void pc_init_ne2k_isa(ISABus *bus, NICInfo *nd)
729 {
730     static int nb_ne2k = 0;
731 
732     if (nb_ne2k == NE2000_NB_MAX)
733         return;
734     isa_ne2000_init(bus, ne2000_io[nb_ne2k],
735                     ne2000_irq[nb_ne2k], nd);
736     nb_ne2k++;
737 }
738 
739 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
740 {
741     X86CPU *cpu = opaque;
742 
743     if (level) {
744         cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
745     }
746 }
747 
748 static
749 void pc_machine_done(Notifier *notifier, void *data)
750 {
751     PCMachineState *pcms = container_of(notifier,
752                                         PCMachineState, machine_done);
753     X86MachineState *x86ms = X86_MACHINE(pcms);
754 
755     cxl_hook_up_pxb_registers(pcms->bus, &pcms->cxl_devices_state,
756                               &error_fatal);
757 
758     if (pcms->cxl_devices_state.is_enabled) {
759         cxl_fmws_link_targets(&pcms->cxl_devices_state, &error_fatal);
760     }
761 
762     /* set the number of CPUs */
763     x86_rtc_set_cpus_count(x86ms->rtc, x86ms->boot_cpus);
764 
765     fw_cfg_add_extra_pci_roots(pcms->bus, x86ms->fw_cfg);
766 
767     acpi_setup();
768     if (x86ms->fw_cfg) {
769         fw_cfg_build_smbios(MACHINE(pcms), x86ms->fw_cfg);
770         fw_cfg_build_feature_control(MACHINE(pcms), x86ms->fw_cfg);
771         /* update FW_CFG_NB_CPUS to account for -device added CPUs */
772         fw_cfg_modify_i16(x86ms->fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
773     }
774 }
775 
776 void pc_guest_info_init(PCMachineState *pcms)
777 {
778     X86MachineState *x86ms = X86_MACHINE(pcms);
779 
780     x86ms->apic_xrupt_override = true;
781     pcms->machine_done.notify = pc_machine_done;
782     qemu_add_machine_init_done_notifier(&pcms->machine_done);
783 }
784 
785 /* setup pci memory address space mapping into system address space */
786 void pc_pci_as_mapping_init(MemoryRegion *system_memory,
787                             MemoryRegion *pci_address_space)
788 {
789     /* Set to lower priority than RAM */
790     memory_region_add_subregion_overlap(system_memory, 0x0,
791                                         pci_address_space, -1);
792 }
793 
794 void xen_load_linux(PCMachineState *pcms)
795 {
796     int i;
797     FWCfgState *fw_cfg;
798     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
799     X86MachineState *x86ms = X86_MACHINE(pcms);
800 
801     assert(MACHINE(pcms)->kernel_filename != NULL);
802 
803     fw_cfg = fw_cfg_init_io(FW_CFG_IO_BASE);
804     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
805     rom_set_fw(fw_cfg);
806 
807     x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
808                    pcmc->pvh_enabled, pcmc->legacy_no_rng_seed);
809     for (i = 0; i < nb_option_roms; i++) {
810         assert(!strcmp(option_rom[i].name, "linuxboot.bin") ||
811                !strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
812                !strcmp(option_rom[i].name, "pvh.bin") ||
813                !strcmp(option_rom[i].name, "multiboot.bin") ||
814                !strcmp(option_rom[i].name, "multiboot_dma.bin"));
815         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
816     }
817     x86ms->fw_cfg = fw_cfg;
818 }
819 
820 #define PC_ROM_MIN_VGA     0xc0000
821 #define PC_ROM_MIN_OPTION  0xc8000
822 #define PC_ROM_MAX         0xe0000
823 #define PC_ROM_ALIGN       0x800
824 #define PC_ROM_SIZE        (PC_ROM_MAX - PC_ROM_MIN_VGA)
825 
826 static hwaddr pc_above_4g_end(PCMachineState *pcms)
827 {
828     X86MachineState *x86ms = X86_MACHINE(pcms);
829 
830     if (pcms->sgx_epc.size != 0) {
831         return sgx_epc_above_4g_end(&pcms->sgx_epc);
832     }
833 
834     return x86ms->above_4g_mem_start + x86ms->above_4g_mem_size;
835 }
836 
837 static void pc_get_device_memory_range(PCMachineState *pcms,
838                                        hwaddr *base,
839                                        ram_addr_t *device_mem_size)
840 {
841     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
842     MachineState *machine = MACHINE(pcms);
843     ram_addr_t size;
844     hwaddr addr;
845 
846     size = machine->maxram_size - machine->ram_size;
847     addr = ROUND_UP(pc_above_4g_end(pcms), 1 * GiB);
848 
849     if (pcmc->enforce_aligned_dimm) {
850         /* size device region assuming 1G page max alignment per slot */
851         size += (1 * GiB) * machine->ram_slots;
852     }
853 
854     *base = addr;
855     *device_mem_size = size;
856 }
857 
858 static uint64_t pc_get_cxl_range_start(PCMachineState *pcms)
859 {
860     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
861     hwaddr cxl_base;
862     ram_addr_t size;
863 
864     if (pcmc->has_reserved_memory) {
865         pc_get_device_memory_range(pcms, &cxl_base, &size);
866         cxl_base += size;
867     } else {
868         cxl_base = pc_above_4g_end(pcms);
869     }
870 
871     return cxl_base;
872 }
873 
874 static uint64_t pc_get_cxl_range_end(PCMachineState *pcms)
875 {
876     uint64_t start = pc_get_cxl_range_start(pcms) + MiB;
877 
878     if (pcms->cxl_devices_state.fixed_windows) {
879         GList *it;
880 
881         start = ROUND_UP(start, 256 * MiB);
882         for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
883             CXLFixedWindow *fw = it->data;
884             start += fw->size;
885         }
886     }
887 
888     return start;
889 }
890 
891 static hwaddr pc_max_used_gpa(PCMachineState *pcms, uint64_t pci_hole64_size)
892 {
893     X86CPU *cpu = X86_CPU(first_cpu);
894 
895     /* 32-bit systems don't have hole64 thus return max CPU address */
896     if (cpu->phys_bits <= 32) {
897         return ((hwaddr)1 << cpu->phys_bits) - 1;
898     }
899 
900     return pc_pci_hole64_start() + pci_hole64_size - 1;
901 }
902 
903 /*
904  * AMD systems with an IOMMU have an additional hole close to the
905  * 1Tb, which are special GPAs that cannot be DMA mapped. Depending
906  * on kernel version, VFIO may or may not let you DMA map those ranges.
907  * Starting Linux v5.4 we validate it, and can't create guests on AMD machines
908  * with certain memory sizes. It's also wrong to use those IOVA ranges
909  * in detriment of leading to IOMMU INVALID_DEVICE_REQUEST or worse.
910  * The ranges reserved for Hyper-Transport are:
911  *
912  * FD_0000_0000h - FF_FFFF_FFFFh
913  *
914  * The ranges represent the following:
915  *
916  * Base Address   Top Address  Use
917  *
918  * FD_0000_0000h FD_F7FF_FFFFh Reserved interrupt address space
919  * FD_F800_0000h FD_F8FF_FFFFh Interrupt/EOI IntCtl
920  * FD_F900_0000h FD_F90F_FFFFh Legacy PIC IACK
921  * FD_F910_0000h FD_F91F_FFFFh System Management
922  * FD_F920_0000h FD_FAFF_FFFFh Reserved Page Tables
923  * FD_FB00_0000h FD_FBFF_FFFFh Address Translation
924  * FD_FC00_0000h FD_FDFF_FFFFh I/O Space
925  * FD_FE00_0000h FD_FFFF_FFFFh Configuration
926  * FE_0000_0000h FE_1FFF_FFFFh Extended Configuration/Device Messages
927  * FE_2000_0000h FF_FFFF_FFFFh Reserved
928  *
929  * See AMD IOMMU spec, section 2.1.2 "IOMMU Logical Topology",
930  * Table 3: Special Address Controls (GPA) for more information.
931  */
932 #define AMD_HT_START         0xfd00000000UL
933 #define AMD_HT_END           0xffffffffffUL
934 #define AMD_ABOVE_1TB_START  (AMD_HT_END + 1)
935 #define AMD_HT_SIZE          (AMD_ABOVE_1TB_START - AMD_HT_START)
936 
937 void pc_memory_init(PCMachineState *pcms,
938                     MemoryRegion *system_memory,
939                     MemoryRegion *rom_memory,
940                     MemoryRegion **ram_memory,
941                     uint64_t pci_hole64_size)
942 {
943     int linux_boot, i;
944     MemoryRegion *option_rom_mr;
945     MemoryRegion *ram_below_4g, *ram_above_4g;
946     FWCfgState *fw_cfg;
947     MachineState *machine = MACHINE(pcms);
948     MachineClass *mc = MACHINE_GET_CLASS(machine);
949     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
950     X86MachineState *x86ms = X86_MACHINE(pcms);
951     hwaddr maxphysaddr, maxusedaddr;
952     hwaddr cxl_base, cxl_resv_end = 0;
953     X86CPU *cpu = X86_CPU(first_cpu);
954 
955     assert(machine->ram_size == x86ms->below_4g_mem_size +
956                                 x86ms->above_4g_mem_size);
957 
958     linux_boot = (machine->kernel_filename != NULL);
959 
960     /*
961      * The HyperTransport range close to the 1T boundary is unique to AMD
962      * hosts with IOMMUs enabled. Restrict the ram-above-4g relocation
963      * to above 1T to AMD vCPUs only. @enforce_amd_1tb_hole is only false in
964      * older machine types (<= 7.0) for compatibility purposes.
965      */
966     if (IS_AMD_CPU(&cpu->env) && pcmc->enforce_amd_1tb_hole) {
967         /* Bail out if max possible address does not cross HT range */
968         if (pc_max_used_gpa(pcms, pci_hole64_size) >= AMD_HT_START) {
969             x86ms->above_4g_mem_start = AMD_ABOVE_1TB_START;
970         }
971 
972         /*
973          * Advertise the HT region if address space covers the reserved
974          * region or if we relocate.
975          */
976         if (cpu->phys_bits >= 40) {
977             e820_add_entry(AMD_HT_START, AMD_HT_SIZE, E820_RESERVED);
978         }
979     }
980 
981     /*
982      * phys-bits is required to be appropriately configured
983      * to make sure max used GPA is reachable.
984      */
985     maxusedaddr = pc_max_used_gpa(pcms, pci_hole64_size);
986     maxphysaddr = ((hwaddr)1 << cpu->phys_bits) - 1;
987     if (maxphysaddr < maxusedaddr) {
988         error_report("Address space limit 0x%"PRIx64" < 0x%"PRIx64
989                      " phys-bits too low (%u)",
990                      maxphysaddr, maxusedaddr, cpu->phys_bits);
991         exit(EXIT_FAILURE);
992     }
993 
994     /*
995      * Split single memory region and use aliases to address portions of it,
996      * done for backwards compatibility with older qemus.
997      */
998     *ram_memory = machine->ram;
999     ram_below_4g = g_malloc(sizeof(*ram_below_4g));
1000     memory_region_init_alias(ram_below_4g, NULL, "ram-below-4g", machine->ram,
1001                              0, x86ms->below_4g_mem_size);
1002     memory_region_add_subregion(system_memory, 0, ram_below_4g);
1003     e820_add_entry(0, x86ms->below_4g_mem_size, E820_RAM);
1004     if (x86ms->above_4g_mem_size > 0) {
1005         ram_above_4g = g_malloc(sizeof(*ram_above_4g));
1006         memory_region_init_alias(ram_above_4g, NULL, "ram-above-4g",
1007                                  machine->ram,
1008                                  x86ms->below_4g_mem_size,
1009                                  x86ms->above_4g_mem_size);
1010         memory_region_add_subregion(system_memory, x86ms->above_4g_mem_start,
1011                                     ram_above_4g);
1012         e820_add_entry(x86ms->above_4g_mem_start, x86ms->above_4g_mem_size,
1013                        E820_RAM);
1014     }
1015 
1016     if (pcms->sgx_epc.size != 0) {
1017         e820_add_entry(pcms->sgx_epc.base, pcms->sgx_epc.size, E820_RESERVED);
1018     }
1019 
1020     if (!pcmc->has_reserved_memory &&
1021         (machine->ram_slots ||
1022          (machine->maxram_size > machine->ram_size))) {
1023 
1024         error_report("\"-memory 'slots|maxmem'\" is not supported by: %s",
1025                      mc->name);
1026         exit(EXIT_FAILURE);
1027     }
1028 
1029     /* always allocate the device memory information */
1030     machine->device_memory = g_malloc0(sizeof(*machine->device_memory));
1031 
1032     /* initialize device memory address space */
1033     if (pcmc->has_reserved_memory &&
1034         (machine->ram_size < machine->maxram_size)) {
1035         ram_addr_t device_mem_size;
1036 
1037         if (machine->ram_slots > ACPI_MAX_RAM_SLOTS) {
1038             error_report("unsupported amount of memory slots: %"PRIu64,
1039                          machine->ram_slots);
1040             exit(EXIT_FAILURE);
1041         }
1042 
1043         if (QEMU_ALIGN_UP(machine->maxram_size,
1044                           TARGET_PAGE_SIZE) != machine->maxram_size) {
1045             error_report("maximum memory size must by aligned to multiple of "
1046                          "%d bytes", TARGET_PAGE_SIZE);
1047             exit(EXIT_FAILURE);
1048         }
1049 
1050         pc_get_device_memory_range(pcms, &machine->device_memory->base, &device_mem_size);
1051 
1052         if ((machine->device_memory->base + device_mem_size) <
1053             device_mem_size) {
1054             error_report("unsupported amount of maximum memory: " RAM_ADDR_FMT,
1055                          machine->maxram_size);
1056             exit(EXIT_FAILURE);
1057         }
1058 
1059         memory_region_init(&machine->device_memory->mr, OBJECT(pcms),
1060                            "device-memory", device_mem_size);
1061         memory_region_add_subregion(system_memory, machine->device_memory->base,
1062                                     &machine->device_memory->mr);
1063     }
1064 
1065     if (pcms->cxl_devices_state.is_enabled) {
1066         MemoryRegion *mr = &pcms->cxl_devices_state.host_mr;
1067         hwaddr cxl_size = MiB;
1068 
1069         cxl_base = pc_get_cxl_range_start(pcms);
1070         memory_region_init(mr, OBJECT(machine), "cxl_host_reg", cxl_size);
1071         memory_region_add_subregion(system_memory, cxl_base, mr);
1072         cxl_resv_end = cxl_base + cxl_size;
1073         if (pcms->cxl_devices_state.fixed_windows) {
1074             hwaddr cxl_fmw_base;
1075             GList *it;
1076 
1077             cxl_fmw_base = ROUND_UP(cxl_base + cxl_size, 256 * MiB);
1078             for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
1079                 CXLFixedWindow *fw = it->data;
1080 
1081                 fw->base = cxl_fmw_base;
1082                 memory_region_init_io(&fw->mr, OBJECT(machine), &cfmws_ops, fw,
1083                                       "cxl-fixed-memory-region", fw->size);
1084                 memory_region_add_subregion(system_memory, fw->base, &fw->mr);
1085                 cxl_fmw_base += fw->size;
1086                 cxl_resv_end = cxl_fmw_base;
1087             }
1088         }
1089     }
1090 
1091     /* Initialize PC system firmware */
1092     pc_system_firmware_init(pcms, rom_memory);
1093 
1094     option_rom_mr = g_malloc(sizeof(*option_rom_mr));
1095     memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE,
1096                            &error_fatal);
1097     if (pcmc->pci_enabled) {
1098         memory_region_set_readonly(option_rom_mr, true);
1099     }
1100     memory_region_add_subregion_overlap(rom_memory,
1101                                         PC_ROM_MIN_VGA,
1102                                         option_rom_mr,
1103                                         1);
1104 
1105     fw_cfg = fw_cfg_arch_create(machine,
1106                                 x86ms->boot_cpus, x86ms->apic_id_limit);
1107 
1108     rom_set_fw(fw_cfg);
1109 
1110     if (pcmc->has_reserved_memory && machine->device_memory->base) {
1111         uint64_t *val = g_malloc(sizeof(*val));
1112         PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1113         uint64_t res_mem_end = machine->device_memory->base;
1114 
1115         if (!pcmc->broken_reserved_end) {
1116             res_mem_end += memory_region_size(&machine->device_memory->mr);
1117         }
1118 
1119         if (pcms->cxl_devices_state.is_enabled) {
1120             res_mem_end = cxl_resv_end;
1121         }
1122         *val = cpu_to_le64(ROUND_UP(res_mem_end, 1 * GiB));
1123         fw_cfg_add_file(fw_cfg, "etc/reserved-memory-end", val, sizeof(*val));
1124     }
1125 
1126     if (linux_boot) {
1127         x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
1128                        pcmc->pvh_enabled, pcmc->legacy_no_rng_seed);
1129     }
1130 
1131     for (i = 0; i < nb_option_roms; i++) {
1132         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1133     }
1134     x86ms->fw_cfg = fw_cfg;
1135 
1136     /* Init default IOAPIC address space */
1137     x86ms->ioapic_as = &address_space_memory;
1138 
1139     /* Init ACPI memory hotplug IO base address */
1140     pcms->memhp_io_base = ACPI_MEMORY_HOTPLUG_BASE;
1141 }
1142 
1143 /*
1144  * The 64bit pci hole starts after "above 4G RAM" and
1145  * potentially the space reserved for memory hotplug.
1146  */
1147 uint64_t pc_pci_hole64_start(void)
1148 {
1149     PCMachineState *pcms = PC_MACHINE(qdev_get_machine());
1150     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1151     MachineState *ms = MACHINE(pcms);
1152     uint64_t hole64_start = 0;
1153     ram_addr_t size = 0;
1154 
1155     if (pcms->cxl_devices_state.is_enabled) {
1156         hole64_start = pc_get_cxl_range_end(pcms);
1157     } else if (pcmc->has_reserved_memory && (ms->ram_size < ms->maxram_size)) {
1158         pc_get_device_memory_range(pcms, &hole64_start, &size);
1159         if (!pcmc->broken_reserved_end) {
1160             hole64_start += size;
1161         }
1162     } else {
1163         hole64_start = pc_above_4g_end(pcms);
1164     }
1165 
1166     return ROUND_UP(hole64_start, 1 * GiB);
1167 }
1168 
1169 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1170 {
1171     DeviceState *dev = NULL;
1172 
1173     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_VGA);
1174     if (pci_bus) {
1175         PCIDevice *pcidev = pci_vga_init(pci_bus);
1176         dev = pcidev ? &pcidev->qdev : NULL;
1177     } else if (isa_bus) {
1178         ISADevice *isadev = isa_vga_init(isa_bus);
1179         dev = isadev ? DEVICE(isadev) : NULL;
1180     }
1181     rom_reset_order_override();
1182     return dev;
1183 }
1184 
1185 static const MemoryRegionOps ioport80_io_ops = {
1186     .write = ioport80_write,
1187     .read = ioport80_read,
1188     .endianness = DEVICE_NATIVE_ENDIAN,
1189     .impl = {
1190         .min_access_size = 1,
1191         .max_access_size = 1,
1192     },
1193 };
1194 
1195 static const MemoryRegionOps ioportF0_io_ops = {
1196     .write = ioportF0_write,
1197     .read = ioportF0_read,
1198     .endianness = DEVICE_NATIVE_ENDIAN,
1199     .impl = {
1200         .min_access_size = 1,
1201         .max_access_size = 1,
1202     },
1203 };
1204 
1205 static void pc_superio_init(ISABus *isa_bus, bool create_fdctrl,
1206                             bool create_i8042, bool no_vmport)
1207 {
1208     int i;
1209     DriveInfo *fd[MAX_FD];
1210     qemu_irq *a20_line;
1211     ISADevice *fdc, *i8042, *port92, *vmmouse;
1212 
1213     serial_hds_isa_init(isa_bus, 0, MAX_ISA_SERIAL_PORTS);
1214     parallel_hds_isa_init(isa_bus, MAX_PARALLEL_PORTS);
1215 
1216     for (i = 0; i < MAX_FD; i++) {
1217         fd[i] = drive_get(IF_FLOPPY, 0, i);
1218         create_fdctrl |= !!fd[i];
1219     }
1220     if (create_fdctrl) {
1221         fdc = isa_new(TYPE_ISA_FDC);
1222         if (fdc) {
1223             isa_realize_and_unref(fdc, isa_bus, &error_fatal);
1224             isa_fdc_init_drives(fdc, fd);
1225         }
1226     }
1227 
1228     if (!create_i8042) {
1229         return;
1230     }
1231 
1232     i8042 = isa_create_simple(isa_bus, TYPE_I8042);
1233     if (!no_vmport) {
1234         isa_create_simple(isa_bus, TYPE_VMPORT);
1235         vmmouse = isa_try_new("vmmouse");
1236     } else {
1237         vmmouse = NULL;
1238     }
1239     if (vmmouse) {
1240         object_property_set_link(OBJECT(vmmouse), TYPE_I8042, OBJECT(i8042),
1241                                  &error_abort);
1242         isa_realize_and_unref(vmmouse, isa_bus, &error_fatal);
1243     }
1244     port92 = isa_create_simple(isa_bus, TYPE_PORT92);
1245 
1246     a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1247     i8042_setup_a20_line(i8042, a20_line[0]);
1248     qdev_connect_gpio_out_named(DEVICE(port92),
1249                                 PORT92_A20_LINE, 0, a20_line[1]);
1250     g_free(a20_line);
1251 }
1252 
1253 void pc_basic_device_init(struct PCMachineState *pcms,
1254                           ISABus *isa_bus, qemu_irq *gsi,
1255                           ISADevice **rtc_state,
1256                           bool create_fdctrl,
1257                           uint32_t hpet_irqs)
1258 {
1259     int i;
1260     DeviceState *hpet = NULL;
1261     int pit_isa_irq = 0;
1262     qemu_irq pit_alt_irq = NULL;
1263     qemu_irq rtc_irq = NULL;
1264     ISADevice *pit = NULL;
1265     MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1266     MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
1267     X86MachineState *x86ms = X86_MACHINE(pcms);
1268 
1269     memory_region_init_io(ioport80_io, NULL, &ioport80_io_ops, NULL, "ioport80", 1);
1270     memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
1271 
1272     memory_region_init_io(ioportF0_io, NULL, &ioportF0_io_ops, NULL, "ioportF0", 1);
1273     memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
1274 
1275     /*
1276      * Check if an HPET shall be created.
1277      *
1278      * Without KVM_CAP_PIT_STATE2, we cannot switch off the in-kernel PIT
1279      * when the HPET wants to take over. Thus we have to disable the latter.
1280      */
1281     if (pcms->hpet_enabled && (!kvm_irqchip_in_kernel() ||
1282                                kvm_has_pit_state2())) {
1283         hpet = qdev_try_new(TYPE_HPET);
1284         if (!hpet) {
1285             error_report("couldn't create HPET device");
1286             exit(1);
1287         }
1288         /*
1289          * For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-1.7 and
1290          * earlier, use IRQ2 for compat. Otherwise, use IRQ16~23, IRQ8 and
1291          * IRQ2.
1292          */
1293         uint8_t compat = object_property_get_uint(OBJECT(hpet),
1294                 HPET_INTCAP, NULL);
1295         if (!compat) {
1296             qdev_prop_set_uint32(hpet, HPET_INTCAP, hpet_irqs);
1297         }
1298         sysbus_realize_and_unref(SYS_BUS_DEVICE(hpet), &error_fatal);
1299         sysbus_mmio_map(SYS_BUS_DEVICE(hpet), 0, HPET_BASE);
1300 
1301         for (i = 0; i < IOAPIC_NUM_PINS; i++) {
1302             sysbus_connect_irq(SYS_BUS_DEVICE(hpet), i, gsi[i]);
1303         }
1304         pit_isa_irq = -1;
1305         pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1306         rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1307     }
1308     *rtc_state = ISA_DEVICE(mc146818_rtc_init(isa_bus, 2000, rtc_irq));
1309 
1310     qemu_register_boot_set(pc_boot_set, *rtc_state);
1311 
1312     if (!xen_enabled() &&
1313         (x86ms->pit == ON_OFF_AUTO_AUTO || x86ms->pit == ON_OFF_AUTO_ON)) {
1314         if (kvm_pit_in_kernel()) {
1315             pit = kvm_pit_init(isa_bus, 0x40);
1316         } else {
1317             pit = i8254_pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1318         }
1319         if (hpet) {
1320             /* connect PIT to output control line of the HPET */
1321             qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(DEVICE(pit), 0));
1322         }
1323         pcspk_init(pcms->pcspk, isa_bus, pit);
1324     }
1325 
1326     /* Super I/O */
1327     pc_superio_init(isa_bus, create_fdctrl, pcms->i8042_enabled,
1328                     pcms->vmport != ON_OFF_AUTO_ON);
1329 }
1330 
1331 void pc_nic_init(PCMachineClass *pcmc, ISABus *isa_bus, PCIBus *pci_bus)
1332 {
1333     int i;
1334 
1335     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_NIC);
1336     for (i = 0; i < nb_nics; i++) {
1337         NICInfo *nd = &nd_table[i];
1338         const char *model = nd->model ? nd->model : pcmc->default_nic_model;
1339 
1340         if (g_str_equal(model, "ne2k_isa")) {
1341             pc_init_ne2k_isa(isa_bus, nd);
1342         } else {
1343             pci_nic_init_nofail(nd, pci_bus, model, NULL);
1344         }
1345     }
1346     rom_reset_order_override();
1347 }
1348 
1349 void pc_i8259_create(ISABus *isa_bus, qemu_irq *i8259_irqs)
1350 {
1351     qemu_irq *i8259;
1352 
1353     if (kvm_pic_in_kernel()) {
1354         i8259 = kvm_i8259_init(isa_bus);
1355     } else if (xen_enabled()) {
1356         i8259 = xen_interrupt_controller_init();
1357     } else {
1358         i8259 = i8259_init(isa_bus, x86_allocate_cpu_irq());
1359     }
1360 
1361     for (size_t i = 0; i < ISA_NUM_IRQS; i++) {
1362         i8259_irqs[i] = i8259[i];
1363     }
1364 
1365     g_free(i8259);
1366 }
1367 
1368 static void pc_memory_pre_plug(HotplugHandler *hotplug_dev, DeviceState *dev,
1369                                Error **errp)
1370 {
1371     const PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1372     const X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1373     const PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1374     const MachineState *ms = MACHINE(hotplug_dev);
1375     const bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1376     const uint64_t legacy_align = TARGET_PAGE_SIZE;
1377     Error *local_err = NULL;
1378 
1379     /*
1380      * When -no-acpi is used with Q35 machine type, no ACPI is built,
1381      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1382      * addition to cover this case.
1383      */
1384     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1385         error_setg(errp,
1386                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1387         return;
1388     }
1389 
1390     if (is_nvdimm && !ms->nvdimms_state->is_enabled) {
1391         error_setg(errp, "nvdimm is not enabled: missing 'nvdimm' in '-M'");
1392         return;
1393     }
1394 
1395     hotplug_handler_pre_plug(x86ms->acpi_dev, dev, &local_err);
1396     if (local_err) {
1397         error_propagate(errp, local_err);
1398         return;
1399     }
1400 
1401     pc_dimm_pre_plug(PC_DIMM(dev), MACHINE(hotplug_dev),
1402                      pcmc->enforce_aligned_dimm ? NULL : &legacy_align, errp);
1403 }
1404 
1405 static void pc_memory_plug(HotplugHandler *hotplug_dev,
1406                            DeviceState *dev, Error **errp)
1407 {
1408     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1409     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1410     MachineState *ms = MACHINE(hotplug_dev);
1411     bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1412 
1413     pc_dimm_plug(PC_DIMM(dev), MACHINE(pcms));
1414 
1415     if (is_nvdimm) {
1416         nvdimm_plug(ms->nvdimms_state);
1417     }
1418 
1419     hotplug_handler_plug(x86ms->acpi_dev, dev, &error_abort);
1420 }
1421 
1422 static void pc_memory_unplug_request(HotplugHandler *hotplug_dev,
1423                                      DeviceState *dev, Error **errp)
1424 {
1425     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1426 
1427     /*
1428      * When -no-acpi is used with Q35 machine type, no ACPI is built,
1429      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1430      * addition to cover this case.
1431      */
1432     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1433         error_setg(errp,
1434                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1435         return;
1436     }
1437 
1438     if (object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM)) {
1439         error_setg(errp, "nvdimm device hot unplug is not supported yet.");
1440         return;
1441     }
1442 
1443     hotplug_handler_unplug_request(x86ms->acpi_dev, dev,
1444                                    errp);
1445 }
1446 
1447 static void pc_memory_unplug(HotplugHandler *hotplug_dev,
1448                              DeviceState *dev, Error **errp)
1449 {
1450     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1451     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1452     Error *local_err = NULL;
1453 
1454     hotplug_handler_unplug(x86ms->acpi_dev, dev, &local_err);
1455     if (local_err) {
1456         goto out;
1457     }
1458 
1459     pc_dimm_unplug(PC_DIMM(dev), MACHINE(pcms));
1460     qdev_unrealize(dev);
1461  out:
1462     error_propagate(errp, local_err);
1463 }
1464 
1465 static void pc_virtio_md_pci_pre_plug(HotplugHandler *hotplug_dev,
1466                                       DeviceState *dev, Error **errp)
1467 {
1468     HotplugHandler *hotplug_dev2 = qdev_get_bus_hotplug_handler(dev);
1469     Error *local_err = NULL;
1470 
1471     if (!hotplug_dev2 && dev->hotplugged) {
1472         /*
1473          * Without a bus hotplug handler, we cannot control the plug/unplug
1474          * order. We should never reach this point when hotplugging on x86,
1475          * however, better add a safety net.
1476          */
1477         error_setg(errp, "hotplug of virtio based memory devices not supported"
1478                    " on this bus.");
1479         return;
1480     }
1481     /*
1482      * First, see if we can plug this memory device at all. If that
1483      * succeeds, branch of to the actual hotplug handler.
1484      */
1485     memory_device_pre_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev), NULL,
1486                            &local_err);
1487     if (!local_err && hotplug_dev2) {
1488         hotplug_handler_pre_plug(hotplug_dev2, dev, &local_err);
1489     }
1490     error_propagate(errp, local_err);
1491 }
1492 
1493 static void pc_virtio_md_pci_plug(HotplugHandler *hotplug_dev,
1494                                   DeviceState *dev, Error **errp)
1495 {
1496     HotplugHandler *hotplug_dev2 = qdev_get_bus_hotplug_handler(dev);
1497     Error *local_err = NULL;
1498 
1499     /*
1500      * Plug the memory device first and then branch off to the actual
1501      * hotplug handler. If that one fails, we can easily undo the memory
1502      * device bits.
1503      */
1504     memory_device_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev));
1505     if (hotplug_dev2) {
1506         hotplug_handler_plug(hotplug_dev2, dev, &local_err);
1507         if (local_err) {
1508             memory_device_unplug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev));
1509         }
1510     }
1511     error_propagate(errp, local_err);
1512 }
1513 
1514 static void pc_virtio_md_pci_unplug_request(HotplugHandler *hotplug_dev,
1515                                             DeviceState *dev, Error **errp)
1516 {
1517     /* We don't support hot unplug of virtio based memory devices */
1518     error_setg(errp, "virtio based memory devices cannot be unplugged.");
1519 }
1520 
1521 static void pc_virtio_md_pci_unplug(HotplugHandler *hotplug_dev,
1522                                     DeviceState *dev, Error **errp)
1523 {
1524     /* We don't support hot unplug of virtio based memory devices */
1525 }
1526 
1527 static void pc_machine_device_pre_plug_cb(HotplugHandler *hotplug_dev,
1528                                           DeviceState *dev, Error **errp)
1529 {
1530     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1531         pc_memory_pre_plug(hotplug_dev, dev, errp);
1532     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1533         x86_cpu_pre_plug(hotplug_dev, dev, errp);
1534     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI) ||
1535                object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MEM_PCI)) {
1536         pc_virtio_md_pci_pre_plug(hotplug_dev, dev, errp);
1537     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1538         /* Declare the APIC range as the reserved MSI region */
1539         char *resv_prop_str = g_strdup_printf("0xfee00000:0xfeefffff:%d",
1540                                               VIRTIO_IOMMU_RESV_MEM_T_MSI);
1541 
1542         object_property_set_uint(OBJECT(dev), "len-reserved-regions", 1, errp);
1543         object_property_set_str(OBJECT(dev), "reserved-regions[0]",
1544                                 resv_prop_str, errp);
1545         g_free(resv_prop_str);
1546     }
1547 
1548     if (object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE) ||
1549         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1550         PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1551 
1552         if (pcms->iommu) {
1553             error_setg(errp, "QEMU does not support multiple vIOMMUs "
1554                        "for x86 yet.");
1555             return;
1556         }
1557         pcms->iommu = dev;
1558     }
1559 }
1560 
1561 static void pc_machine_device_plug_cb(HotplugHandler *hotplug_dev,
1562                                       DeviceState *dev, Error **errp)
1563 {
1564     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1565         pc_memory_plug(hotplug_dev, dev, errp);
1566     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1567         x86_cpu_plug(hotplug_dev, dev, errp);
1568     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI) ||
1569                object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MEM_PCI)) {
1570         pc_virtio_md_pci_plug(hotplug_dev, dev, errp);
1571     }
1572 }
1573 
1574 static void pc_machine_device_unplug_request_cb(HotplugHandler *hotplug_dev,
1575                                                 DeviceState *dev, Error **errp)
1576 {
1577     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1578         pc_memory_unplug_request(hotplug_dev, dev, errp);
1579     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1580         x86_cpu_unplug_request_cb(hotplug_dev, dev, errp);
1581     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI) ||
1582                object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MEM_PCI)) {
1583         pc_virtio_md_pci_unplug_request(hotplug_dev, dev, errp);
1584     } else {
1585         error_setg(errp, "acpi: device unplug request for not supported device"
1586                    " type: %s", object_get_typename(OBJECT(dev)));
1587     }
1588 }
1589 
1590 static void pc_machine_device_unplug_cb(HotplugHandler *hotplug_dev,
1591                                         DeviceState *dev, Error **errp)
1592 {
1593     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1594         pc_memory_unplug(hotplug_dev, dev, errp);
1595     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1596         x86_cpu_unplug_cb(hotplug_dev, dev, errp);
1597     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI) ||
1598                object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MEM_PCI)) {
1599         pc_virtio_md_pci_unplug(hotplug_dev, dev, errp);
1600     } else {
1601         error_setg(errp, "acpi: device unplug for not supported device"
1602                    " type: %s", object_get_typename(OBJECT(dev)));
1603     }
1604 }
1605 
1606 static HotplugHandler *pc_get_hotplug_handler(MachineState *machine,
1607                                              DeviceState *dev)
1608 {
1609     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM) ||
1610         object_dynamic_cast(OBJECT(dev), TYPE_CPU) ||
1611         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI) ||
1612         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MEM_PCI) ||
1613         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI) ||
1614         object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE)) {
1615         return HOTPLUG_HANDLER(machine);
1616     }
1617 
1618     return NULL;
1619 }
1620 
1621 static void
1622 pc_machine_get_device_memory_region_size(Object *obj, Visitor *v,
1623                                          const char *name, void *opaque,
1624                                          Error **errp)
1625 {
1626     MachineState *ms = MACHINE(obj);
1627     int64_t value = 0;
1628 
1629     if (ms->device_memory) {
1630         value = memory_region_size(&ms->device_memory->mr);
1631     }
1632 
1633     visit_type_int(v, name, &value, errp);
1634 }
1635 
1636 static void pc_machine_get_vmport(Object *obj, Visitor *v, const char *name,
1637                                   void *opaque, Error **errp)
1638 {
1639     PCMachineState *pcms = PC_MACHINE(obj);
1640     OnOffAuto vmport = pcms->vmport;
1641 
1642     visit_type_OnOffAuto(v, name, &vmport, errp);
1643 }
1644 
1645 static void pc_machine_set_vmport(Object *obj, Visitor *v, const char *name,
1646                                   void *opaque, Error **errp)
1647 {
1648     PCMachineState *pcms = PC_MACHINE(obj);
1649 
1650     visit_type_OnOffAuto(v, name, &pcms->vmport, errp);
1651 }
1652 
1653 static bool pc_machine_get_smbus(Object *obj, Error **errp)
1654 {
1655     PCMachineState *pcms = PC_MACHINE(obj);
1656 
1657     return pcms->smbus_enabled;
1658 }
1659 
1660 static void pc_machine_set_smbus(Object *obj, bool value, Error **errp)
1661 {
1662     PCMachineState *pcms = PC_MACHINE(obj);
1663 
1664     pcms->smbus_enabled = value;
1665 }
1666 
1667 static bool pc_machine_get_sata(Object *obj, Error **errp)
1668 {
1669     PCMachineState *pcms = PC_MACHINE(obj);
1670 
1671     return pcms->sata_enabled;
1672 }
1673 
1674 static void pc_machine_set_sata(Object *obj, bool value, Error **errp)
1675 {
1676     PCMachineState *pcms = PC_MACHINE(obj);
1677 
1678     pcms->sata_enabled = value;
1679 }
1680 
1681 static bool pc_machine_get_hpet(Object *obj, Error **errp)
1682 {
1683     PCMachineState *pcms = PC_MACHINE(obj);
1684 
1685     return pcms->hpet_enabled;
1686 }
1687 
1688 static void pc_machine_set_hpet(Object *obj, bool value, Error **errp)
1689 {
1690     PCMachineState *pcms = PC_MACHINE(obj);
1691 
1692     pcms->hpet_enabled = value;
1693 }
1694 
1695 static bool pc_machine_get_i8042(Object *obj, Error **errp)
1696 {
1697     PCMachineState *pcms = PC_MACHINE(obj);
1698 
1699     return pcms->i8042_enabled;
1700 }
1701 
1702 static void pc_machine_set_i8042(Object *obj, bool value, Error **errp)
1703 {
1704     PCMachineState *pcms = PC_MACHINE(obj);
1705 
1706     pcms->i8042_enabled = value;
1707 }
1708 
1709 static bool pc_machine_get_default_bus_bypass_iommu(Object *obj, Error **errp)
1710 {
1711     PCMachineState *pcms = PC_MACHINE(obj);
1712 
1713     return pcms->default_bus_bypass_iommu;
1714 }
1715 
1716 static void pc_machine_set_default_bus_bypass_iommu(Object *obj, bool value,
1717                                                     Error **errp)
1718 {
1719     PCMachineState *pcms = PC_MACHINE(obj);
1720 
1721     pcms->default_bus_bypass_iommu = value;
1722 }
1723 
1724 static void pc_machine_get_smbios_ep(Object *obj, Visitor *v, const char *name,
1725                                      void *opaque, Error **errp)
1726 {
1727     PCMachineState *pcms = PC_MACHINE(obj);
1728     SmbiosEntryPointType smbios_entry_point_type = pcms->smbios_entry_point_type;
1729 
1730     visit_type_SmbiosEntryPointType(v, name, &smbios_entry_point_type, errp);
1731 }
1732 
1733 static void pc_machine_set_smbios_ep(Object *obj, Visitor *v, const char *name,
1734                                      void *opaque, Error **errp)
1735 {
1736     PCMachineState *pcms = PC_MACHINE(obj);
1737 
1738     visit_type_SmbiosEntryPointType(v, name, &pcms->smbios_entry_point_type, errp);
1739 }
1740 
1741 static void pc_machine_get_max_ram_below_4g(Object *obj, Visitor *v,
1742                                             const char *name, void *opaque,
1743                                             Error **errp)
1744 {
1745     PCMachineState *pcms = PC_MACHINE(obj);
1746     uint64_t value = pcms->max_ram_below_4g;
1747 
1748     visit_type_size(v, name, &value, errp);
1749 }
1750 
1751 static void pc_machine_set_max_ram_below_4g(Object *obj, Visitor *v,
1752                                             const char *name, void *opaque,
1753                                             Error **errp)
1754 {
1755     PCMachineState *pcms = PC_MACHINE(obj);
1756     uint64_t value;
1757 
1758     if (!visit_type_size(v, name, &value, errp)) {
1759         return;
1760     }
1761     if (value > 4 * GiB) {
1762         error_setg(errp,
1763                    "Machine option 'max-ram-below-4g=%"PRIu64
1764                    "' expects size less than or equal to 4G", value);
1765         return;
1766     }
1767 
1768     if (value < 1 * MiB) {
1769         warn_report("Only %" PRIu64 " bytes of RAM below the 4GiB boundary,"
1770                     "BIOS may not work with less than 1MiB", value);
1771     }
1772 
1773     pcms->max_ram_below_4g = value;
1774 }
1775 
1776 static void pc_machine_get_max_fw_size(Object *obj, Visitor *v,
1777                                        const char *name, void *opaque,
1778                                        Error **errp)
1779 {
1780     PCMachineState *pcms = PC_MACHINE(obj);
1781     uint64_t value = pcms->max_fw_size;
1782 
1783     visit_type_size(v, name, &value, errp);
1784 }
1785 
1786 static void pc_machine_set_max_fw_size(Object *obj, Visitor *v,
1787                                        const char *name, void *opaque,
1788                                        Error **errp)
1789 {
1790     PCMachineState *pcms = PC_MACHINE(obj);
1791     uint64_t value;
1792 
1793     if (!visit_type_size(v, name, &value, errp)) {
1794         return;
1795     }
1796 
1797     /*
1798     * We don't have a theoretically justifiable exact lower bound on the base
1799     * address of any flash mapping. In practice, the IO-APIC MMIO range is
1800     * [0xFEE00000..0xFEE01000] -- see IO_APIC_DEFAULT_ADDRESS --, leaving free
1801     * only 18MB-4KB below 4G. For now, restrict the cumulative mapping to 8MB in
1802     * size.
1803     */
1804     if (value > 16 * MiB) {
1805         error_setg(errp,
1806                    "User specified max allowed firmware size %" PRIu64 " is "
1807                    "greater than 16MiB. If combined firwmare size exceeds "
1808                    "16MiB the system may not boot, or experience intermittent"
1809                    "stability issues.",
1810                    value);
1811         return;
1812     }
1813 
1814     pcms->max_fw_size = value;
1815 }
1816 
1817 
1818 static void pc_machine_initfn(Object *obj)
1819 {
1820     PCMachineState *pcms = PC_MACHINE(obj);
1821 
1822 #ifdef CONFIG_VMPORT
1823     pcms->vmport = ON_OFF_AUTO_AUTO;
1824 #else
1825     pcms->vmport = ON_OFF_AUTO_OFF;
1826 #endif /* CONFIG_VMPORT */
1827     pcms->max_ram_below_4g = 0; /* use default */
1828     pcms->smbios_entry_point_type = SMBIOS_ENTRY_POINT_TYPE_32;
1829 
1830     /* acpi build is enabled by default if machine supports it */
1831     pcms->acpi_build_enabled = PC_MACHINE_GET_CLASS(pcms)->has_acpi_build;
1832     pcms->smbus_enabled = true;
1833     pcms->sata_enabled = true;
1834     pcms->i8042_enabled = true;
1835     pcms->max_fw_size = 8 * MiB;
1836 #ifdef CONFIG_HPET
1837     pcms->hpet_enabled = true;
1838 #endif
1839     pcms->default_bus_bypass_iommu = false;
1840 
1841     pc_system_flash_create(pcms);
1842     pcms->pcspk = isa_new(TYPE_PC_SPEAKER);
1843     object_property_add_alias(OBJECT(pcms), "pcspk-audiodev",
1844                               OBJECT(pcms->pcspk), "audiodev");
1845     cxl_machine_init(obj, &pcms->cxl_devices_state);
1846 }
1847 
1848 static void pc_machine_reset(MachineState *machine, ShutdownCause reason)
1849 {
1850     CPUState *cs;
1851     X86CPU *cpu;
1852 
1853     qemu_devices_reset(reason);
1854 
1855     /* Reset APIC after devices have been reset to cancel
1856      * any changes that qemu_devices_reset() might have done.
1857      */
1858     CPU_FOREACH(cs) {
1859         cpu = X86_CPU(cs);
1860 
1861         x86_cpu_after_reset(cpu);
1862     }
1863 }
1864 
1865 static void pc_machine_wakeup(MachineState *machine)
1866 {
1867     cpu_synchronize_all_states();
1868     pc_machine_reset(machine, SHUTDOWN_CAUSE_NONE);
1869     cpu_synchronize_all_post_reset();
1870 }
1871 
1872 static bool pc_hotplug_allowed(MachineState *ms, DeviceState *dev, Error **errp)
1873 {
1874     X86IOMMUState *iommu = x86_iommu_get_default();
1875     IntelIOMMUState *intel_iommu;
1876 
1877     if (iommu &&
1878         object_dynamic_cast((Object *)iommu, TYPE_INTEL_IOMMU_DEVICE) &&
1879         object_dynamic_cast((Object *)dev, "vfio-pci")) {
1880         intel_iommu = INTEL_IOMMU_DEVICE(iommu);
1881         if (!intel_iommu->caching_mode) {
1882             error_setg(errp, "Device assignment is not allowed without "
1883                        "enabling caching-mode=on for Intel IOMMU.");
1884             return false;
1885         }
1886     }
1887 
1888     return true;
1889 }
1890 
1891 static void pc_machine_class_init(ObjectClass *oc, void *data)
1892 {
1893     MachineClass *mc = MACHINE_CLASS(oc);
1894     PCMachineClass *pcmc = PC_MACHINE_CLASS(oc);
1895     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
1896 
1897     pcmc->pci_enabled = true;
1898     pcmc->has_acpi_build = true;
1899     pcmc->rsdp_in_ram = true;
1900     pcmc->smbios_defaults = true;
1901     pcmc->smbios_uuid_encoded = true;
1902     pcmc->gigabyte_align = true;
1903     pcmc->has_reserved_memory = true;
1904     pcmc->kvmclock_enabled = true;
1905     pcmc->enforce_aligned_dimm = true;
1906     pcmc->enforce_amd_1tb_hole = true;
1907     /* BIOS ACPI tables: 128K. Other BIOS datastructures: less than 4K reported
1908      * to be used at the moment, 32K should be enough for a while.  */
1909     pcmc->acpi_data_size = 0x20000 + 0x8000;
1910     pcmc->pvh_enabled = true;
1911     pcmc->kvmclock_create_always = true;
1912     assert(!mc->get_hotplug_handler);
1913     mc->get_hotplug_handler = pc_get_hotplug_handler;
1914     mc->hotplug_allowed = pc_hotplug_allowed;
1915     mc->cpu_index_to_instance_props = x86_cpu_index_to_props;
1916     mc->get_default_cpu_node_id = x86_get_default_cpu_node_id;
1917     mc->possible_cpu_arch_ids = x86_possible_cpu_arch_ids;
1918     mc->auto_enable_numa_with_memhp = true;
1919     mc->auto_enable_numa_with_memdev = true;
1920     mc->has_hotpluggable_cpus = true;
1921     mc->default_boot_order = "cad";
1922     mc->block_default_type = IF_IDE;
1923     mc->max_cpus = 255;
1924     mc->reset = pc_machine_reset;
1925     mc->wakeup = pc_machine_wakeup;
1926     hc->pre_plug = pc_machine_device_pre_plug_cb;
1927     hc->plug = pc_machine_device_plug_cb;
1928     hc->unplug_request = pc_machine_device_unplug_request_cb;
1929     hc->unplug = pc_machine_device_unplug_cb;
1930     mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE;
1931     mc->nvdimm_supported = true;
1932     mc->smp_props.dies_supported = true;
1933     mc->default_ram_id = "pc.ram";
1934 
1935     object_class_property_add(oc, PC_MACHINE_MAX_RAM_BELOW_4G, "size",
1936         pc_machine_get_max_ram_below_4g, pc_machine_set_max_ram_below_4g,
1937         NULL, NULL);
1938     object_class_property_set_description(oc, PC_MACHINE_MAX_RAM_BELOW_4G,
1939         "Maximum ram below the 4G boundary (32bit boundary)");
1940 
1941     object_class_property_add(oc, PC_MACHINE_DEVMEM_REGION_SIZE, "int",
1942         pc_machine_get_device_memory_region_size, NULL,
1943         NULL, NULL);
1944 
1945     object_class_property_add(oc, PC_MACHINE_VMPORT, "OnOffAuto",
1946         pc_machine_get_vmport, pc_machine_set_vmport,
1947         NULL, NULL);
1948     object_class_property_set_description(oc, PC_MACHINE_VMPORT,
1949         "Enable vmport (pc & q35)");
1950 
1951     object_class_property_add_bool(oc, PC_MACHINE_SMBUS,
1952         pc_machine_get_smbus, pc_machine_set_smbus);
1953     object_class_property_set_description(oc, PC_MACHINE_SMBUS,
1954         "Enable/disable system management bus");
1955 
1956     object_class_property_add_bool(oc, PC_MACHINE_SATA,
1957         pc_machine_get_sata, pc_machine_set_sata);
1958     object_class_property_set_description(oc, PC_MACHINE_SATA,
1959         "Enable/disable Serial ATA bus");
1960 
1961     object_class_property_add_bool(oc, "hpet",
1962         pc_machine_get_hpet, pc_machine_set_hpet);
1963     object_class_property_set_description(oc, "hpet",
1964         "Enable/disable high precision event timer emulation");
1965 
1966     object_class_property_add_bool(oc, PC_MACHINE_I8042,
1967         pc_machine_get_i8042, pc_machine_set_i8042);
1968 
1969     object_class_property_add_bool(oc, "default-bus-bypass-iommu",
1970         pc_machine_get_default_bus_bypass_iommu,
1971         pc_machine_set_default_bus_bypass_iommu);
1972 
1973     object_class_property_add(oc, PC_MACHINE_MAX_FW_SIZE, "size",
1974         pc_machine_get_max_fw_size, pc_machine_set_max_fw_size,
1975         NULL, NULL);
1976     object_class_property_set_description(oc, PC_MACHINE_MAX_FW_SIZE,
1977         "Maximum combined firmware size");
1978 
1979     object_class_property_add(oc, PC_MACHINE_SMBIOS_EP, "str",
1980         pc_machine_get_smbios_ep, pc_machine_set_smbios_ep,
1981         NULL, NULL);
1982     object_class_property_set_description(oc, PC_MACHINE_SMBIOS_EP,
1983         "SMBIOS Entry Point type [32, 64]");
1984 }
1985 
1986 static const TypeInfo pc_machine_info = {
1987     .name = TYPE_PC_MACHINE,
1988     .parent = TYPE_X86_MACHINE,
1989     .abstract = true,
1990     .instance_size = sizeof(PCMachineState),
1991     .instance_init = pc_machine_initfn,
1992     .class_size = sizeof(PCMachineClass),
1993     .class_init = pc_machine_class_init,
1994     .interfaces = (InterfaceInfo[]) {
1995          { TYPE_HOTPLUG_HANDLER },
1996          { }
1997     },
1998 };
1999 
2000 static void pc_machine_register_types(void)
2001 {
2002     type_register_static(&pc_machine_info);
2003 }
2004 
2005 type_init(pc_machine_register_types)
2006