xref: /qemu/hw/i386/pc.c (revision b64b7ed8)
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/pc.h"
28 #include "hw/char/serial.h"
29 #include "hw/char/parallel.h"
30 #include "hw/hyperv/hv-balloon.h"
31 #include "hw/i386/fw_cfg.h"
32 #include "hw/i386/vmport.h"
33 #include "sysemu/cpus.h"
34 #include "hw/ide/ide-bus.h"
35 #include "hw/timer/hpet.h"
36 #include "hw/loader.h"
37 #include "hw/rtc/mc146818rtc.h"
38 #include "hw/intc/i8259.h"
39 #include "hw/timer/i8254.h"
40 #include "hw/input/i8042.h"
41 #include "hw/audio/pcspk.h"
42 #include "sysemu/sysemu.h"
43 #include "sysemu/xen.h"
44 #include "sysemu/reset.h"
45 #include "kvm/kvm_i386.h"
46 #include "hw/xen/xen.h"
47 #include "qapi/qmp/qlist.h"
48 #include "qemu/error-report.h"
49 #include "hw/acpi/cpu_hotplug.h"
50 #include "acpi-build.h"
51 #include "hw/mem/nvdimm.h"
52 #include "hw/cxl/cxl_host.h"
53 #include "hw/usb.h"
54 #include "hw/i386/intel_iommu.h"
55 #include "hw/net/ne2000-isa.h"
56 #include "hw/virtio/virtio-iommu.h"
57 #include "hw/virtio/virtio-md-pci.h"
58 #include "hw/i386/kvm/xen_overlay.h"
59 #include "hw/i386/kvm/xen_evtchn.h"
60 #include "hw/i386/kvm/xen_gnttab.h"
61 #include "hw/i386/kvm/xen_xenstore.h"
62 #include "hw/mem/memory-device.h"
63 #include "e820_memory_layout.h"
64 #include "trace.h"
65 #include CONFIG_DEVICES
66 
67 #ifdef CONFIG_XEN_EMU
68 #include "hw/xen/xen-legacy-backend.h"
69 #include "hw/xen/xen-bus.h"
70 #endif
71 
72 /*
73  * Helper for setting model-id for CPU models that changed model-id
74  * depending on QEMU versions up to QEMU 2.4.
75  */
76 #define PC_CPU_MODEL_IDS(v) \
77     { "qemu32-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
78     { "qemu64-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
79     { "athlon-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },
80 
81 GlobalProperty pc_compat_8_2[] = {};
82 const size_t pc_compat_8_2_len = G_N_ELEMENTS(pc_compat_8_2);
83 
84 GlobalProperty pc_compat_8_1[] = {};
85 const size_t pc_compat_8_1_len = G_N_ELEMENTS(pc_compat_8_1);
86 
87 GlobalProperty pc_compat_8_0[] = {
88     { "virtio-mem", "unplugged-inaccessible", "auto" },
89 };
90 const size_t pc_compat_8_0_len = G_N_ELEMENTS(pc_compat_8_0);
91 
92 GlobalProperty pc_compat_7_2[] = {
93     { "ICH9-LPC", "noreboot", "true" },
94 };
95 const size_t pc_compat_7_2_len = G_N_ELEMENTS(pc_compat_7_2);
96 
97 GlobalProperty pc_compat_7_1[] = {};
98 const size_t pc_compat_7_1_len = G_N_ELEMENTS(pc_compat_7_1);
99 
100 GlobalProperty pc_compat_7_0[] = {};
101 const size_t pc_compat_7_0_len = G_N_ELEMENTS(pc_compat_7_0);
102 
103 GlobalProperty pc_compat_6_2[] = {
104     { "virtio-mem", "unplugged-inaccessible", "off" },
105 };
106 const size_t pc_compat_6_2_len = G_N_ELEMENTS(pc_compat_6_2);
107 
108 GlobalProperty pc_compat_6_1[] = {
109     { TYPE_X86_CPU, "hv-version-id-build", "0x1bbc" },
110     { TYPE_X86_CPU, "hv-version-id-major", "0x0006" },
111     { TYPE_X86_CPU, "hv-version-id-minor", "0x0001" },
112     { "ICH9-LPC", "x-keep-pci-slot-hpc", "false" },
113 };
114 const size_t pc_compat_6_1_len = G_N_ELEMENTS(pc_compat_6_1);
115 
116 GlobalProperty pc_compat_6_0[] = {
117     { "qemu64" "-" TYPE_X86_CPU, "family", "6" },
118     { "qemu64" "-" TYPE_X86_CPU, "model", "6" },
119     { "qemu64" "-" TYPE_X86_CPU, "stepping", "3" },
120     { TYPE_X86_CPU, "x-vendor-cpuid-only", "off" },
121     { "ICH9-LPC", ACPI_PM_PROP_ACPI_PCIHP_BRIDGE, "off" },
122     { "ICH9-LPC", "x-keep-pci-slot-hpc", "true" },
123 };
124 const size_t pc_compat_6_0_len = G_N_ELEMENTS(pc_compat_6_0);
125 
126 GlobalProperty pc_compat_5_2[] = {
127     { "ICH9-LPC", "x-smi-cpu-hotunplug", "off" },
128 };
129 const size_t pc_compat_5_2_len = G_N_ELEMENTS(pc_compat_5_2);
130 
131 GlobalProperty pc_compat_5_1[] = {
132     { "ICH9-LPC", "x-smi-cpu-hotplug", "off" },
133     { TYPE_X86_CPU, "kvm-msi-ext-dest-id", "off" },
134 };
135 const size_t pc_compat_5_1_len = G_N_ELEMENTS(pc_compat_5_1);
136 
137 GlobalProperty pc_compat_5_0[] = {
138 };
139 const size_t pc_compat_5_0_len = G_N_ELEMENTS(pc_compat_5_0);
140 
141 GlobalProperty pc_compat_4_2[] = {
142     { "mch", "smbase-smram", "off" },
143 };
144 const size_t pc_compat_4_2_len = G_N_ELEMENTS(pc_compat_4_2);
145 
146 GlobalProperty pc_compat_4_1[] = {};
147 const size_t pc_compat_4_1_len = G_N_ELEMENTS(pc_compat_4_1);
148 
149 GlobalProperty pc_compat_4_0[] = {};
150 const size_t pc_compat_4_0_len = G_N_ELEMENTS(pc_compat_4_0);
151 
152 GlobalProperty pc_compat_3_1[] = {
153     { "intel-iommu", "dma-drain", "off" },
154     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "off" },
155     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "off" },
156     { "Opteron_G4" "-" TYPE_X86_CPU, "npt", "off" },
157     { "Opteron_G4" "-" TYPE_X86_CPU, "nrip-save", "off" },
158     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "off" },
159     { "Opteron_G5" "-" TYPE_X86_CPU, "npt", "off" },
160     { "Opteron_G5" "-" TYPE_X86_CPU, "nrip-save", "off" },
161     { "EPYC" "-" TYPE_X86_CPU, "npt", "off" },
162     { "EPYC" "-" TYPE_X86_CPU, "nrip-save", "off" },
163     { "EPYC-IBPB" "-" TYPE_X86_CPU, "npt", "off" },
164     { "EPYC-IBPB" "-" TYPE_X86_CPU, "nrip-save", "off" },
165     { "Skylake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
166     { "Skylake-Client-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
167     { "Skylake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
168     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
169     { "Cascadelake-Server" "-" TYPE_X86_CPU,  "mpx", "on" },
170     { "Icelake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
171     { "Icelake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
172     { "Cascadelake-Server" "-" TYPE_X86_CPU, "stepping", "5" },
173     { TYPE_X86_CPU, "x-intel-pt-auto-level", "off" },
174 };
175 const size_t pc_compat_3_1_len = G_N_ELEMENTS(pc_compat_3_1);
176 
177 GlobalProperty pc_compat_3_0[] = {
178     { TYPE_X86_CPU, "x-hv-synic-kvm-only", "on" },
179     { "Skylake-Server" "-" TYPE_X86_CPU, "pku", "off" },
180     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "pku", "off" },
181 };
182 const size_t pc_compat_3_0_len = G_N_ELEMENTS(pc_compat_3_0);
183 
184 GlobalProperty pc_compat_2_12[] = {
185     { TYPE_X86_CPU, "legacy-cache", "on" },
186     { TYPE_X86_CPU, "topoext", "off" },
187     { "EPYC-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
188     { "EPYC-IBPB-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
189 };
190 const size_t pc_compat_2_12_len = G_N_ELEMENTS(pc_compat_2_12);
191 
192 GlobalProperty pc_compat_2_11[] = {
193     { TYPE_X86_CPU, "x-migrate-smi-count", "off" },
194     { "Skylake-Server" "-" TYPE_X86_CPU, "clflushopt", "off" },
195 };
196 const size_t pc_compat_2_11_len = G_N_ELEMENTS(pc_compat_2_11);
197 
198 GlobalProperty pc_compat_2_10[] = {
199     { TYPE_X86_CPU, "x-hv-max-vps", "0x40" },
200     { "i440FX-pcihost", "x-pci-hole64-fix", "off" },
201     { "q35-pcihost", "x-pci-hole64-fix", "off" },
202 };
203 const size_t pc_compat_2_10_len = G_N_ELEMENTS(pc_compat_2_10);
204 
205 GlobalProperty pc_compat_2_9[] = {
206     { "mch", "extended-tseg-mbytes", "0" },
207 };
208 const size_t pc_compat_2_9_len = G_N_ELEMENTS(pc_compat_2_9);
209 
210 GlobalProperty pc_compat_2_8[] = {
211     { TYPE_X86_CPU, "tcg-cpuid", "off" },
212     { "kvmclock", "x-mach-use-reliable-get-clock", "off" },
213     { "ICH9-LPC", "x-smi-broadcast", "off" },
214     { TYPE_X86_CPU, "vmware-cpuid-freq", "off" },
215     { "Haswell-" TYPE_X86_CPU, "stepping", "1" },
216 };
217 const size_t pc_compat_2_8_len = G_N_ELEMENTS(pc_compat_2_8);
218 
219 GlobalProperty pc_compat_2_7[] = {
220     { TYPE_X86_CPU, "l3-cache", "off" },
221     { TYPE_X86_CPU, "full-cpuid-auto-level", "off" },
222     { "Opteron_G3" "-" TYPE_X86_CPU, "family", "15" },
223     { "Opteron_G3" "-" TYPE_X86_CPU, "model", "6" },
224     { "Opteron_G3" "-" TYPE_X86_CPU, "stepping", "1" },
225     { "isa-pcspk", "migrate", "off" },
226 };
227 const size_t pc_compat_2_7_len = G_N_ELEMENTS(pc_compat_2_7);
228 
229 GlobalProperty pc_compat_2_6[] = {
230     { TYPE_X86_CPU, "cpuid-0xb", "off" },
231     { "vmxnet3", "romfile", "" },
232     { TYPE_X86_CPU, "fill-mtrr-mask", "off" },
233     { "apic-common", "legacy-instance-id", "on", }
234 };
235 const size_t pc_compat_2_6_len = G_N_ELEMENTS(pc_compat_2_6);
236 
237 GlobalProperty pc_compat_2_5[] = {};
238 const size_t pc_compat_2_5_len = G_N_ELEMENTS(pc_compat_2_5);
239 
240 GlobalProperty pc_compat_2_4[] = {
241     PC_CPU_MODEL_IDS("2.4.0")
242     { "Haswell-" TYPE_X86_CPU, "abm", "off" },
243     { "Haswell-noTSX-" TYPE_X86_CPU, "abm", "off" },
244     { "Broadwell-" TYPE_X86_CPU, "abm", "off" },
245     { "Broadwell-noTSX-" TYPE_X86_CPU, "abm", "off" },
246     { "host" "-" TYPE_X86_CPU, "host-cache-info", "on" },
247     { TYPE_X86_CPU, "check", "off" },
248     { "qemu64" "-" TYPE_X86_CPU, "sse4a", "on" },
249     { "qemu64" "-" TYPE_X86_CPU, "abm", "on" },
250     { "qemu64" "-" TYPE_X86_CPU, "popcnt", "on" },
251     { "qemu32" "-" TYPE_X86_CPU, "popcnt", "on" },
252     { "Opteron_G2" "-" TYPE_X86_CPU, "rdtscp", "on" },
253     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "on" },
254     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "on" },
255     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "on", }
256 };
257 const size_t pc_compat_2_4_len = G_N_ELEMENTS(pc_compat_2_4);
258 
259 GlobalProperty pc_compat_2_3[] = {
260     PC_CPU_MODEL_IDS("2.3.0")
261     { TYPE_X86_CPU, "arat", "off" },
262     { "qemu64" "-" TYPE_X86_CPU, "min-level", "4" },
263     { "kvm64" "-" TYPE_X86_CPU, "min-level", "5" },
264     { "pentium3" "-" TYPE_X86_CPU, "min-level", "2" },
265     { "n270" "-" TYPE_X86_CPU, "min-level", "5" },
266     { "Conroe" "-" TYPE_X86_CPU, "min-level", "4" },
267     { "Penryn" "-" TYPE_X86_CPU, "min-level", "4" },
268     { "Nehalem" "-" TYPE_X86_CPU, "min-level", "4" },
269     { "n270" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
270     { "Penryn" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
271     { "Conroe" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
272     { "Nehalem" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
273     { "Westmere" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
274     { "SandyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
275     { "IvyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
276     { "Haswell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
277     { "Haswell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
278     { "Broadwell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
279     { "Broadwell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
280     { TYPE_X86_CPU, "kvm-no-smi-migration", "on" },
281 };
282 const size_t pc_compat_2_3_len = G_N_ELEMENTS(pc_compat_2_3);
283 
284 GlobalProperty pc_compat_2_2[] = {
285     PC_CPU_MODEL_IDS("2.2.0")
286     { "kvm64" "-" TYPE_X86_CPU, "vme", "off" },
287     { "kvm32" "-" TYPE_X86_CPU, "vme", "off" },
288     { "Conroe" "-" TYPE_X86_CPU, "vme", "off" },
289     { "Penryn" "-" TYPE_X86_CPU, "vme", "off" },
290     { "Nehalem" "-" TYPE_X86_CPU, "vme", "off" },
291     { "Westmere" "-" TYPE_X86_CPU, "vme", "off" },
292     { "SandyBridge" "-" TYPE_X86_CPU, "vme", "off" },
293     { "Haswell" "-" TYPE_X86_CPU, "vme", "off" },
294     { "Broadwell" "-" TYPE_X86_CPU, "vme", "off" },
295     { "Opteron_G1" "-" TYPE_X86_CPU, "vme", "off" },
296     { "Opteron_G2" "-" TYPE_X86_CPU, "vme", "off" },
297     { "Opteron_G3" "-" TYPE_X86_CPU, "vme", "off" },
298     { "Opteron_G4" "-" TYPE_X86_CPU, "vme", "off" },
299     { "Opteron_G5" "-" TYPE_X86_CPU, "vme", "off" },
300     { "Haswell" "-" TYPE_X86_CPU, "f16c", "off" },
301     { "Haswell" "-" TYPE_X86_CPU, "rdrand", "off" },
302     { "Broadwell" "-" TYPE_X86_CPU, "f16c", "off" },
303     { "Broadwell" "-" TYPE_X86_CPU, "rdrand", "off" },
304 };
305 const size_t pc_compat_2_2_len = G_N_ELEMENTS(pc_compat_2_2);
306 
307 GlobalProperty pc_compat_2_1[] = {
308     PC_CPU_MODEL_IDS("2.1.0")
309     { "coreduo" "-" TYPE_X86_CPU, "vmx", "on" },
310     { "core2duo" "-" TYPE_X86_CPU, "vmx", "on" },
311 };
312 const size_t pc_compat_2_1_len = G_N_ELEMENTS(pc_compat_2_1);
313 
314 GlobalProperty pc_compat_2_0[] = {
315     PC_CPU_MODEL_IDS("2.0.0")
316     { "virtio-scsi-pci", "any_layout", "off" },
317     { "PIIX4_PM", "memory-hotplug-support", "off" },
318     { "apic", "version", "0x11" },
319     { "nec-usb-xhci", "superspeed-ports-first", "off" },
320     { "nec-usb-xhci", "force-pcie-endcap", "on" },
321     { "pci-serial", "prog_if", "0" },
322     { "pci-serial-2x", "prog_if", "0" },
323     { "pci-serial-4x", "prog_if", "0" },
324     { "virtio-net-pci", "guest_announce", "off" },
325     { "ICH9-LPC", "memory-hotplug-support", "off" },
326 };
327 const size_t pc_compat_2_0_len = G_N_ELEMENTS(pc_compat_2_0);
328 
329 GSIState *pc_gsi_create(qemu_irq **irqs, bool pci_enabled)
330 {
331     GSIState *s;
332 
333     s = g_new0(GSIState, 1);
334     if (kvm_ioapic_in_kernel()) {
335         kvm_pc_setup_irq_routing(pci_enabled);
336     }
337     *irqs = qemu_allocate_irqs(gsi_handler, s, IOAPIC_NUM_PINS);
338 
339     return s;
340 }
341 
342 static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
343                            unsigned size)
344 {
345 }
346 
347 static uint64_t ioport80_read(void *opaque, hwaddr addr, unsigned size)
348 {
349     return 0xffffffffffffffffULL;
350 }
351 
352 /* MS-DOS compatibility mode FPU exception support */
353 static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
354                            unsigned size)
355 {
356     if (tcg_enabled()) {
357         cpu_set_ignne();
358     }
359 }
360 
361 static uint64_t ioportF0_read(void *opaque, hwaddr addr, unsigned size)
362 {
363     return 0xffffffffffffffffULL;
364 }
365 
366 /* PC cmos mappings */
367 
368 #define REG_EQUIPMENT_BYTE          0x14
369 
370 static void cmos_init_hd(MC146818RtcState *s, int type_ofs, int info_ofs,
371                          int16_t cylinders, int8_t heads, int8_t sectors)
372 {
373     mc146818rtc_set_cmos_data(s, type_ofs, 47);
374     mc146818rtc_set_cmos_data(s, info_ofs, cylinders);
375     mc146818rtc_set_cmos_data(s, info_ofs + 1, cylinders >> 8);
376     mc146818rtc_set_cmos_data(s, info_ofs + 2, heads);
377     mc146818rtc_set_cmos_data(s, info_ofs + 3, 0xff);
378     mc146818rtc_set_cmos_data(s, info_ofs + 4, 0xff);
379     mc146818rtc_set_cmos_data(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
380     mc146818rtc_set_cmos_data(s, info_ofs + 6, cylinders);
381     mc146818rtc_set_cmos_data(s, info_ofs + 7, cylinders >> 8);
382     mc146818rtc_set_cmos_data(s, info_ofs + 8, sectors);
383 }
384 
385 /* convert boot_device letter to something recognizable by the bios */
386 static int boot_device2nibble(char boot_device)
387 {
388     switch(boot_device) {
389     case 'a':
390     case 'b':
391         return 0x01; /* floppy boot */
392     case 'c':
393         return 0x02; /* hard drive boot */
394     case 'd':
395         return 0x03; /* CD-ROM boot */
396     case 'n':
397         return 0x04; /* Network boot */
398     }
399     return 0;
400 }
401 
402 static void set_boot_dev(PCMachineState *pcms, MC146818RtcState *s,
403                          const char *boot_device, Error **errp)
404 {
405 #define PC_MAX_BOOT_DEVICES 3
406     int nbds, bds[3] = { 0, };
407     int i;
408 
409     nbds = strlen(boot_device);
410     if (nbds > PC_MAX_BOOT_DEVICES) {
411         error_setg(errp, "Too many boot devices for PC");
412         return;
413     }
414     for (i = 0; i < nbds; i++) {
415         bds[i] = boot_device2nibble(boot_device[i]);
416         if (bds[i] == 0) {
417             error_setg(errp, "Invalid boot device for PC: '%c'",
418                        boot_device[i]);
419             return;
420         }
421     }
422     mc146818rtc_set_cmos_data(s, 0x3d, (bds[1] << 4) | bds[0]);
423     mc146818rtc_set_cmos_data(s, 0x38, (bds[2] << 4) | !pcms->fd_bootchk);
424 }
425 
426 static void pc_boot_set(void *opaque, const char *boot_device, Error **errp)
427 {
428     PCMachineState *pcms = opaque;
429     X86MachineState *x86ms = X86_MACHINE(pcms);
430 
431     set_boot_dev(pcms, MC146818_RTC(x86ms->rtc), boot_device, errp);
432 }
433 
434 static void pc_cmos_init_floppy(MC146818RtcState *rtc_state, ISADevice *floppy)
435 {
436     int val, nb, i;
437     FloppyDriveType fd_type[2] = { FLOPPY_DRIVE_TYPE_NONE,
438                                    FLOPPY_DRIVE_TYPE_NONE };
439 
440     /* floppy type */
441     if (floppy) {
442         for (i = 0; i < 2; i++) {
443             fd_type[i] = isa_fdc_get_drive_type(floppy, i);
444         }
445     }
446     val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
447         cmos_get_fd_drive_type(fd_type[1]);
448     mc146818rtc_set_cmos_data(rtc_state, 0x10, val);
449 
450     val = mc146818rtc_get_cmos_data(rtc_state, REG_EQUIPMENT_BYTE);
451     nb = 0;
452     if (fd_type[0] != FLOPPY_DRIVE_TYPE_NONE) {
453         nb++;
454     }
455     if (fd_type[1] != FLOPPY_DRIVE_TYPE_NONE) {
456         nb++;
457     }
458     switch (nb) {
459     case 0:
460         break;
461     case 1:
462         val |= 0x01; /* 1 drive, ready for boot */
463         break;
464     case 2:
465         val |= 0x41; /* 2 drives, ready for boot */
466         break;
467     }
468     mc146818rtc_set_cmos_data(rtc_state, REG_EQUIPMENT_BYTE, val);
469 }
470 
471 typedef struct check_fdc_state {
472     ISADevice *floppy;
473     bool multiple;
474 } CheckFdcState;
475 
476 static int check_fdc(Object *obj, void *opaque)
477 {
478     CheckFdcState *state = opaque;
479     Object *fdc;
480     uint32_t iobase;
481     Error *local_err = NULL;
482 
483     fdc = object_dynamic_cast(obj, TYPE_ISA_FDC);
484     if (!fdc) {
485         return 0;
486     }
487 
488     iobase = object_property_get_uint(obj, "iobase", &local_err);
489     if (local_err || iobase != 0x3f0) {
490         error_free(local_err);
491         return 0;
492     }
493 
494     if (state->floppy) {
495         state->multiple = true;
496     } else {
497         state->floppy = ISA_DEVICE(obj);
498     }
499     return 0;
500 }
501 
502 static const char * const fdc_container_path[] = {
503     "/unattached", "/peripheral", "/peripheral-anon"
504 };
505 
506 /*
507  * Locate the FDC at IO address 0x3f0, in order to configure the CMOS registers
508  * and ACPI objects.
509  */
510 static ISADevice *pc_find_fdc0(void)
511 {
512     int i;
513     Object *container;
514     CheckFdcState state = { 0 };
515 
516     for (i = 0; i < ARRAY_SIZE(fdc_container_path); i++) {
517         container = container_get(qdev_get_machine(), fdc_container_path[i]);
518         object_child_foreach(container, check_fdc, &state);
519     }
520 
521     if (state.multiple) {
522         warn_report("multiple floppy disk controllers with "
523                     "iobase=0x3f0 have been found");
524         error_printf("the one being picked for CMOS setup might not reflect "
525                      "your intent");
526     }
527 
528     return state.floppy;
529 }
530 
531 static void pc_cmos_init_late(PCMachineState *pcms)
532 {
533     X86MachineState *x86ms = X86_MACHINE(pcms);
534     MC146818RtcState *s = MC146818_RTC(x86ms->rtc);
535     int16_t cylinders;
536     int8_t heads, sectors;
537     int val;
538     int i, trans;
539 
540     val = 0;
541     if (pcms->idebus[0] &&
542         ide_get_geometry(pcms->idebus[0], 0,
543                          &cylinders, &heads, &sectors) >= 0) {
544         cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
545         val |= 0xf0;
546     }
547     if (pcms->idebus[0] &&
548         ide_get_geometry(pcms->idebus[0], 1,
549                          &cylinders, &heads, &sectors) >= 0) {
550         cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
551         val |= 0x0f;
552     }
553     mc146818rtc_set_cmos_data(s, 0x12, val);
554 
555     val = 0;
556     for (i = 0; i < 4; i++) {
557         /* NOTE: ide_get_geometry() returns the physical
558            geometry.  It is always such that: 1 <= sects <= 63, 1
559            <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
560            geometry can be different if a translation is done. */
561         BusState *idebus = pcms->idebus[i / 2];
562         if (idebus &&
563             ide_get_geometry(idebus, i % 2,
564                              &cylinders, &heads, &sectors) >= 0) {
565             trans = ide_get_bios_chs_trans(idebus, i % 2) - 1;
566             assert((trans & ~3) == 0);
567             val |= trans << (i * 2);
568         }
569     }
570     mc146818rtc_set_cmos_data(s, 0x39, val);
571 
572     pc_cmos_init_floppy(s, pc_find_fdc0());
573 
574     /* various important CMOS locations needed by PC/Bochs bios */
575 
576     /* memory size */
577     /* base memory (first MiB) */
578     val = MIN(x86ms->below_4g_mem_size / KiB, 640);
579     mc146818rtc_set_cmos_data(s, 0x15, val);
580     mc146818rtc_set_cmos_data(s, 0x16, val >> 8);
581     /* extended memory (next 64MiB) */
582     if (x86ms->below_4g_mem_size > 1 * MiB) {
583         val = (x86ms->below_4g_mem_size - 1 * MiB) / KiB;
584     } else {
585         val = 0;
586     }
587     if (val > 65535)
588         val = 65535;
589     mc146818rtc_set_cmos_data(s, 0x17, val);
590     mc146818rtc_set_cmos_data(s, 0x18, val >> 8);
591     mc146818rtc_set_cmos_data(s, 0x30, val);
592     mc146818rtc_set_cmos_data(s, 0x31, val >> 8);
593     /* memory between 16MiB and 4GiB */
594     if (x86ms->below_4g_mem_size > 16 * MiB) {
595         val = (x86ms->below_4g_mem_size - 16 * MiB) / (64 * KiB);
596     } else {
597         val = 0;
598     }
599     if (val > 65535)
600         val = 65535;
601     mc146818rtc_set_cmos_data(s, 0x34, val);
602     mc146818rtc_set_cmos_data(s, 0x35, val >> 8);
603     /* memory above 4GiB */
604     val = x86ms->above_4g_mem_size / 65536;
605     mc146818rtc_set_cmos_data(s, 0x5b, val);
606     mc146818rtc_set_cmos_data(s, 0x5c, val >> 8);
607     mc146818rtc_set_cmos_data(s, 0x5d, val >> 16);
608 
609     val = 0;
610     val |= 0x02; /* FPU is there */
611     val |= 0x04; /* PS/2 mouse installed */
612     mc146818rtc_set_cmos_data(s, REG_EQUIPMENT_BYTE, val);
613 }
614 
615 static void handle_a20_line_change(void *opaque, int irq, int level)
616 {
617     X86CPU *cpu = opaque;
618 
619     /* XXX: send to all CPUs ? */
620     /* XXX: add logic to handle multiple A20 line sources */
621     x86_cpu_set_a20(cpu, level);
622 }
623 
624 #define NE2000_NB_MAX 6
625 
626 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
627                                               0x280, 0x380 };
628 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
629 
630 static gboolean pc_init_ne2k_isa(ISABus *bus, NICInfo *nd, Error **errp)
631 {
632     static int nb_ne2k = 0;
633 
634     if (nb_ne2k == NE2000_NB_MAX) {
635         error_setg(errp,
636                    "maximum number of ISA NE2000 devices exceeded");
637         return false;
638     }
639     isa_ne2000_init(bus, ne2000_io[nb_ne2k],
640                     ne2000_irq[nb_ne2k], nd);
641     nb_ne2k++;
642     return true;
643 }
644 
645 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
646 {
647     X86CPU *cpu = opaque;
648 
649     if (level) {
650         cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
651     }
652 }
653 
654 static
655 void pc_machine_done(Notifier *notifier, void *data)
656 {
657     PCMachineState *pcms = container_of(notifier,
658                                         PCMachineState, machine_done);
659     X86MachineState *x86ms = X86_MACHINE(pcms);
660 
661     cxl_hook_up_pxb_registers(pcms->pcibus, &pcms->cxl_devices_state,
662                               &error_fatal);
663 
664     if (pcms->cxl_devices_state.is_enabled) {
665         cxl_fmws_link_targets(&pcms->cxl_devices_state, &error_fatal);
666     }
667 
668     /* set the number of CPUs */
669     x86_rtc_set_cpus_count(x86ms->rtc, x86ms->boot_cpus);
670 
671     fw_cfg_add_extra_pci_roots(pcms->pcibus, x86ms->fw_cfg);
672 
673     acpi_setup();
674     if (x86ms->fw_cfg) {
675         fw_cfg_build_smbios(pcms, x86ms->fw_cfg);
676         fw_cfg_build_feature_control(MACHINE(pcms), x86ms->fw_cfg);
677         /* update FW_CFG_NB_CPUS to account for -device added CPUs */
678         fw_cfg_modify_i16(x86ms->fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
679     }
680 
681     pc_cmos_init_late(pcms);
682 }
683 
684 /* setup pci memory address space mapping into system address space */
685 void pc_pci_as_mapping_init(MemoryRegion *system_memory,
686                             MemoryRegion *pci_address_space)
687 {
688     /* Set to lower priority than RAM */
689     memory_region_add_subregion_overlap(system_memory, 0x0,
690                                         pci_address_space, -1);
691 }
692 
693 void xen_load_linux(PCMachineState *pcms)
694 {
695     int i;
696     FWCfgState *fw_cfg;
697     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
698     X86MachineState *x86ms = X86_MACHINE(pcms);
699 
700     assert(MACHINE(pcms)->kernel_filename != NULL);
701 
702     fw_cfg = fw_cfg_init_io(FW_CFG_IO_BASE);
703     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
704     rom_set_fw(fw_cfg);
705 
706     x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
707                    pcmc->pvh_enabled);
708     for (i = 0; i < nb_option_roms; i++) {
709         assert(!strcmp(option_rom[i].name, "linuxboot.bin") ||
710                !strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
711                !strcmp(option_rom[i].name, "pvh.bin") ||
712                !strcmp(option_rom[i].name, "multiboot.bin") ||
713                !strcmp(option_rom[i].name, "multiboot_dma.bin"));
714         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
715     }
716     x86ms->fw_cfg = fw_cfg;
717 }
718 
719 #define PC_ROM_MIN_VGA     0xc0000
720 #define PC_ROM_MIN_OPTION  0xc8000
721 #define PC_ROM_MAX         0xe0000
722 #define PC_ROM_ALIGN       0x800
723 #define PC_ROM_SIZE        (PC_ROM_MAX - PC_ROM_MIN_VGA)
724 
725 static hwaddr pc_above_4g_end(PCMachineState *pcms)
726 {
727     X86MachineState *x86ms = X86_MACHINE(pcms);
728 
729     if (pcms->sgx_epc.size != 0) {
730         return sgx_epc_above_4g_end(&pcms->sgx_epc);
731     }
732 
733     return x86ms->above_4g_mem_start + x86ms->above_4g_mem_size;
734 }
735 
736 static void pc_get_device_memory_range(PCMachineState *pcms,
737                                        hwaddr *base,
738                                        ram_addr_t *device_mem_size)
739 {
740     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
741     MachineState *machine = MACHINE(pcms);
742     ram_addr_t size;
743     hwaddr addr;
744 
745     size = machine->maxram_size - machine->ram_size;
746     addr = ROUND_UP(pc_above_4g_end(pcms), 1 * GiB);
747 
748     if (pcmc->enforce_aligned_dimm) {
749         /* size device region assuming 1G page max alignment per slot */
750         size += (1 * GiB) * machine->ram_slots;
751     }
752 
753     *base = addr;
754     *device_mem_size = size;
755 }
756 
757 static uint64_t pc_get_cxl_range_start(PCMachineState *pcms)
758 {
759     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
760     MachineState *ms = MACHINE(pcms);
761     hwaddr cxl_base;
762     ram_addr_t size;
763 
764     if (pcmc->has_reserved_memory &&
765         (ms->ram_size < ms->maxram_size)) {
766         pc_get_device_memory_range(pcms, &cxl_base, &size);
767         cxl_base += size;
768     } else {
769         cxl_base = pc_above_4g_end(pcms);
770     }
771 
772     return cxl_base;
773 }
774 
775 static uint64_t pc_get_cxl_range_end(PCMachineState *pcms)
776 {
777     uint64_t start = pc_get_cxl_range_start(pcms) + MiB;
778 
779     if (pcms->cxl_devices_state.fixed_windows) {
780         GList *it;
781 
782         start = ROUND_UP(start, 256 * MiB);
783         for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
784             CXLFixedWindow *fw = it->data;
785             start += fw->size;
786         }
787     }
788 
789     return start;
790 }
791 
792 static hwaddr pc_max_used_gpa(PCMachineState *pcms, uint64_t pci_hole64_size)
793 {
794     X86CPU *cpu = X86_CPU(first_cpu);
795     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
796     MachineState *ms = MACHINE(pcms);
797 
798     if (cpu->env.features[FEAT_8000_0001_EDX] & CPUID_EXT2_LM) {
799         /* 64-bit systems */
800         return pc_pci_hole64_start() + pci_hole64_size - 1;
801     }
802 
803     /* 32-bit systems */
804     if (pcmc->broken_32bit_mem_addr_check) {
805         /* old value for compatibility reasons */
806         return ((hwaddr)1 << cpu->phys_bits) - 1;
807     }
808 
809     /*
810      * 32-bit systems don't have hole64 but they might have a region for
811      * memory devices. Even if additional hotplugged memory devices might
812      * not be usable by most guest OSes, we need to still consider them for
813      * calculating the highest possible GPA so that we can properly report
814      * if someone configures them on a CPU that cannot possibly address them.
815      */
816     if (pcmc->has_reserved_memory &&
817         (ms->ram_size < ms->maxram_size)) {
818         hwaddr devmem_start;
819         ram_addr_t devmem_size;
820 
821         pc_get_device_memory_range(pcms, &devmem_start, &devmem_size);
822         devmem_start += devmem_size;
823         return devmem_start - 1;
824     }
825 
826     /* configuration without any memory hotplug */
827     return pc_above_4g_end(pcms) - 1;
828 }
829 
830 /*
831  * AMD systems with an IOMMU have an additional hole close to the
832  * 1Tb, which are special GPAs that cannot be DMA mapped. Depending
833  * on kernel version, VFIO may or may not let you DMA map those ranges.
834  * Starting Linux v5.4 we validate it, and can't create guests on AMD machines
835  * with certain memory sizes. It's also wrong to use those IOVA ranges
836  * in detriment of leading to IOMMU INVALID_DEVICE_REQUEST or worse.
837  * The ranges reserved for Hyper-Transport are:
838  *
839  * FD_0000_0000h - FF_FFFF_FFFFh
840  *
841  * The ranges represent the following:
842  *
843  * Base Address   Top Address  Use
844  *
845  * FD_0000_0000h FD_F7FF_FFFFh Reserved interrupt address space
846  * FD_F800_0000h FD_F8FF_FFFFh Interrupt/EOI IntCtl
847  * FD_F900_0000h FD_F90F_FFFFh Legacy PIC IACK
848  * FD_F910_0000h FD_F91F_FFFFh System Management
849  * FD_F920_0000h FD_FAFF_FFFFh Reserved Page Tables
850  * FD_FB00_0000h FD_FBFF_FFFFh Address Translation
851  * FD_FC00_0000h FD_FDFF_FFFFh I/O Space
852  * FD_FE00_0000h FD_FFFF_FFFFh Configuration
853  * FE_0000_0000h FE_1FFF_FFFFh Extended Configuration/Device Messages
854  * FE_2000_0000h FF_FFFF_FFFFh Reserved
855  *
856  * See AMD IOMMU spec, section 2.1.2 "IOMMU Logical Topology",
857  * Table 3: Special Address Controls (GPA) for more information.
858  */
859 #define AMD_HT_START         0xfd00000000UL
860 #define AMD_HT_END           0xffffffffffUL
861 #define AMD_ABOVE_1TB_START  (AMD_HT_END + 1)
862 #define AMD_HT_SIZE          (AMD_ABOVE_1TB_START - AMD_HT_START)
863 
864 void pc_memory_init(PCMachineState *pcms,
865                     MemoryRegion *system_memory,
866                     MemoryRegion *rom_memory,
867                     uint64_t pci_hole64_size)
868 {
869     int linux_boot, i;
870     MemoryRegion *option_rom_mr;
871     MemoryRegion *ram_below_4g, *ram_above_4g;
872     FWCfgState *fw_cfg;
873     MachineState *machine = MACHINE(pcms);
874     MachineClass *mc = MACHINE_GET_CLASS(machine);
875     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
876     X86MachineState *x86ms = X86_MACHINE(pcms);
877     hwaddr maxphysaddr, maxusedaddr;
878     hwaddr cxl_base, cxl_resv_end = 0;
879     X86CPU *cpu = X86_CPU(first_cpu);
880 
881     assert(machine->ram_size == x86ms->below_4g_mem_size +
882                                 x86ms->above_4g_mem_size);
883 
884     linux_boot = (machine->kernel_filename != NULL);
885 
886     /*
887      * The HyperTransport range close to the 1T boundary is unique to AMD
888      * hosts with IOMMUs enabled. Restrict the ram-above-4g relocation
889      * to above 1T to AMD vCPUs only. @enforce_amd_1tb_hole is only false in
890      * older machine types (<= 7.0) for compatibility purposes.
891      */
892     if (IS_AMD_CPU(&cpu->env) && pcmc->enforce_amd_1tb_hole) {
893         /* Bail out if max possible address does not cross HT range */
894         if (pc_max_used_gpa(pcms, pci_hole64_size) >= AMD_HT_START) {
895             x86ms->above_4g_mem_start = AMD_ABOVE_1TB_START;
896         }
897 
898         /*
899          * Advertise the HT region if address space covers the reserved
900          * region or if we relocate.
901          */
902         if (cpu->phys_bits >= 40) {
903             e820_add_entry(AMD_HT_START, AMD_HT_SIZE, E820_RESERVED);
904         }
905     }
906 
907     /*
908      * phys-bits is required to be appropriately configured
909      * to make sure max used GPA is reachable.
910      */
911     maxusedaddr = pc_max_used_gpa(pcms, pci_hole64_size);
912     maxphysaddr = ((hwaddr)1 << cpu->phys_bits) - 1;
913     if (maxphysaddr < maxusedaddr) {
914         error_report("Address space limit 0x%"PRIx64" < 0x%"PRIx64
915                      " phys-bits too low (%u)",
916                      maxphysaddr, maxusedaddr, cpu->phys_bits);
917         exit(EXIT_FAILURE);
918     }
919 
920     /*
921      * Split single memory region and use aliases to address portions of it,
922      * done for backwards compatibility with older qemus.
923      */
924     ram_below_4g = g_malloc(sizeof(*ram_below_4g));
925     memory_region_init_alias(ram_below_4g, NULL, "ram-below-4g", machine->ram,
926                              0, x86ms->below_4g_mem_size);
927     memory_region_add_subregion(system_memory, 0, ram_below_4g);
928     e820_add_entry(0, x86ms->below_4g_mem_size, E820_RAM);
929     if (x86ms->above_4g_mem_size > 0) {
930         ram_above_4g = g_malloc(sizeof(*ram_above_4g));
931         memory_region_init_alias(ram_above_4g, NULL, "ram-above-4g",
932                                  machine->ram,
933                                  x86ms->below_4g_mem_size,
934                                  x86ms->above_4g_mem_size);
935         memory_region_add_subregion(system_memory, x86ms->above_4g_mem_start,
936                                     ram_above_4g);
937         e820_add_entry(x86ms->above_4g_mem_start, x86ms->above_4g_mem_size,
938                        E820_RAM);
939     }
940 
941     if (pcms->sgx_epc.size != 0) {
942         e820_add_entry(pcms->sgx_epc.base, pcms->sgx_epc.size, E820_RESERVED);
943     }
944 
945     if (!pcmc->has_reserved_memory &&
946         (machine->ram_slots ||
947          (machine->maxram_size > machine->ram_size))) {
948 
949         error_report("\"-memory 'slots|maxmem'\" is not supported by: %s",
950                      mc->name);
951         exit(EXIT_FAILURE);
952     }
953 
954     /* initialize device memory address space */
955     if (pcmc->has_reserved_memory &&
956         (machine->ram_size < machine->maxram_size)) {
957         ram_addr_t device_mem_size;
958         hwaddr device_mem_base;
959 
960         if (machine->ram_slots > ACPI_MAX_RAM_SLOTS) {
961             error_report("unsupported amount of memory slots: %"PRIu64,
962                          machine->ram_slots);
963             exit(EXIT_FAILURE);
964         }
965 
966         if (QEMU_ALIGN_UP(machine->maxram_size,
967                           TARGET_PAGE_SIZE) != machine->maxram_size) {
968             error_report("maximum memory size must by aligned to multiple of "
969                          "%d bytes", TARGET_PAGE_SIZE);
970             exit(EXIT_FAILURE);
971         }
972 
973         pc_get_device_memory_range(pcms, &device_mem_base, &device_mem_size);
974 
975         if (device_mem_base + device_mem_size < device_mem_size) {
976             error_report("unsupported amount of maximum memory: " RAM_ADDR_FMT,
977                          machine->maxram_size);
978             exit(EXIT_FAILURE);
979         }
980         machine_memory_devices_init(machine, device_mem_base, device_mem_size);
981     }
982 
983     if (pcms->cxl_devices_state.is_enabled) {
984         MemoryRegion *mr = &pcms->cxl_devices_state.host_mr;
985         hwaddr cxl_size = MiB;
986 
987         cxl_base = pc_get_cxl_range_start(pcms);
988         memory_region_init(mr, OBJECT(machine), "cxl_host_reg", cxl_size);
989         memory_region_add_subregion(system_memory, cxl_base, mr);
990         cxl_resv_end = cxl_base + cxl_size;
991         if (pcms->cxl_devices_state.fixed_windows) {
992             hwaddr cxl_fmw_base;
993             GList *it;
994 
995             cxl_fmw_base = ROUND_UP(cxl_base + cxl_size, 256 * MiB);
996             for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
997                 CXLFixedWindow *fw = it->data;
998 
999                 fw->base = cxl_fmw_base;
1000                 memory_region_init_io(&fw->mr, OBJECT(machine), &cfmws_ops, fw,
1001                                       "cxl-fixed-memory-region", fw->size);
1002                 memory_region_add_subregion(system_memory, fw->base, &fw->mr);
1003                 cxl_fmw_base += fw->size;
1004                 cxl_resv_end = cxl_fmw_base;
1005             }
1006         }
1007     }
1008 
1009     /* Initialize PC system firmware */
1010     pc_system_firmware_init(pcms, rom_memory);
1011 
1012     option_rom_mr = g_malloc(sizeof(*option_rom_mr));
1013     memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE,
1014                            &error_fatal);
1015     if (pcmc->pci_enabled) {
1016         memory_region_set_readonly(option_rom_mr, true);
1017     }
1018     memory_region_add_subregion_overlap(rom_memory,
1019                                         PC_ROM_MIN_VGA,
1020                                         option_rom_mr,
1021                                         1);
1022 
1023     fw_cfg = fw_cfg_arch_create(machine,
1024                                 x86ms->boot_cpus, x86ms->apic_id_limit);
1025 
1026     rom_set_fw(fw_cfg);
1027 
1028     if (machine->device_memory) {
1029         uint64_t *val = g_malloc(sizeof(*val));
1030         uint64_t res_mem_end = machine->device_memory->base;
1031 
1032         if (!pcmc->broken_reserved_end) {
1033             res_mem_end += memory_region_size(&machine->device_memory->mr);
1034         }
1035 
1036         if (pcms->cxl_devices_state.is_enabled) {
1037             res_mem_end = cxl_resv_end;
1038         }
1039         *val = cpu_to_le64(ROUND_UP(res_mem_end, 1 * GiB));
1040         fw_cfg_add_file(fw_cfg, "etc/reserved-memory-end", val, sizeof(*val));
1041     }
1042 
1043     if (linux_boot) {
1044         x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
1045                        pcmc->pvh_enabled);
1046     }
1047 
1048     for (i = 0; i < nb_option_roms; i++) {
1049         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1050     }
1051     x86ms->fw_cfg = fw_cfg;
1052 
1053     /* Init default IOAPIC address space */
1054     x86ms->ioapic_as = &address_space_memory;
1055 
1056     /* Init ACPI memory hotplug IO base address */
1057     pcms->memhp_io_base = ACPI_MEMORY_HOTPLUG_BASE;
1058 }
1059 
1060 /*
1061  * The 64bit pci hole starts after "above 4G RAM" and
1062  * potentially the space reserved for memory hotplug.
1063  */
1064 uint64_t pc_pci_hole64_start(void)
1065 {
1066     PCMachineState *pcms = PC_MACHINE(qdev_get_machine());
1067     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1068     MachineState *ms = MACHINE(pcms);
1069     uint64_t hole64_start = 0;
1070     ram_addr_t size = 0;
1071 
1072     if (pcms->cxl_devices_state.is_enabled) {
1073         hole64_start = pc_get_cxl_range_end(pcms);
1074     } else if (pcmc->has_reserved_memory && (ms->ram_size < ms->maxram_size)) {
1075         pc_get_device_memory_range(pcms, &hole64_start, &size);
1076         if (!pcmc->broken_reserved_end) {
1077             hole64_start += size;
1078         }
1079     } else {
1080         hole64_start = pc_above_4g_end(pcms);
1081     }
1082 
1083     return ROUND_UP(hole64_start, 1 * GiB);
1084 }
1085 
1086 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1087 {
1088     DeviceState *dev = NULL;
1089 
1090     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_VGA);
1091     if (pci_bus) {
1092         PCIDevice *pcidev = pci_vga_init(pci_bus);
1093         dev = pcidev ? &pcidev->qdev : NULL;
1094     } else if (isa_bus) {
1095         ISADevice *isadev = isa_vga_init(isa_bus);
1096         dev = isadev ? DEVICE(isadev) : NULL;
1097     }
1098     rom_reset_order_override();
1099     return dev;
1100 }
1101 
1102 static const MemoryRegionOps ioport80_io_ops = {
1103     .write = ioport80_write,
1104     .read = ioport80_read,
1105     .endianness = DEVICE_NATIVE_ENDIAN,
1106     .impl = {
1107         .min_access_size = 1,
1108         .max_access_size = 1,
1109     },
1110 };
1111 
1112 static const MemoryRegionOps ioportF0_io_ops = {
1113     .write = ioportF0_write,
1114     .read = ioportF0_read,
1115     .endianness = DEVICE_NATIVE_ENDIAN,
1116     .impl = {
1117         .min_access_size = 1,
1118         .max_access_size = 1,
1119     },
1120 };
1121 
1122 static void pc_superio_init(ISABus *isa_bus, bool create_fdctrl,
1123                             bool create_i8042, bool no_vmport)
1124 {
1125     int i;
1126     DriveInfo *fd[MAX_FD];
1127     qemu_irq *a20_line;
1128     ISADevice *fdc, *i8042, *port92, *vmmouse;
1129 
1130     serial_hds_isa_init(isa_bus, 0, MAX_ISA_SERIAL_PORTS);
1131     parallel_hds_isa_init(isa_bus, MAX_PARALLEL_PORTS);
1132 
1133     for (i = 0; i < MAX_FD; i++) {
1134         fd[i] = drive_get(IF_FLOPPY, 0, i);
1135         create_fdctrl |= !!fd[i];
1136     }
1137     if (create_fdctrl) {
1138         fdc = isa_new(TYPE_ISA_FDC);
1139         if (fdc) {
1140             isa_realize_and_unref(fdc, isa_bus, &error_fatal);
1141             isa_fdc_init_drives(fdc, fd);
1142         }
1143     }
1144 
1145     if (!create_i8042) {
1146         return;
1147     }
1148 
1149     i8042 = isa_create_simple(isa_bus, TYPE_I8042);
1150     if (!no_vmport) {
1151         isa_create_simple(isa_bus, TYPE_VMPORT);
1152         vmmouse = isa_try_new("vmmouse");
1153     } else {
1154         vmmouse = NULL;
1155     }
1156     if (vmmouse) {
1157         object_property_set_link(OBJECT(vmmouse), TYPE_I8042, OBJECT(i8042),
1158                                  &error_abort);
1159         isa_realize_and_unref(vmmouse, isa_bus, &error_fatal);
1160     }
1161     port92 = isa_create_simple(isa_bus, TYPE_PORT92);
1162 
1163     a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1164     qdev_connect_gpio_out_named(DEVICE(i8042),
1165                                 I8042_A20_LINE, 0, a20_line[0]);
1166     qdev_connect_gpio_out_named(DEVICE(port92),
1167                                 PORT92_A20_LINE, 0, a20_line[1]);
1168     g_free(a20_line);
1169 }
1170 
1171 void pc_basic_device_init(struct PCMachineState *pcms,
1172                           ISABus *isa_bus, qemu_irq *gsi,
1173                           ISADevice *rtc_state,
1174                           bool create_fdctrl,
1175                           uint32_t hpet_irqs)
1176 {
1177     int i;
1178     DeviceState *hpet = NULL;
1179     int pit_isa_irq = 0;
1180     qemu_irq pit_alt_irq = NULL;
1181     ISADevice *pit = NULL;
1182     MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1183     MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
1184     X86MachineState *x86ms = X86_MACHINE(pcms);
1185 
1186     memory_region_init_io(ioport80_io, NULL, &ioport80_io_ops, NULL, "ioport80", 1);
1187     memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
1188 
1189     memory_region_init_io(ioportF0_io, NULL, &ioportF0_io_ops, NULL, "ioportF0", 1);
1190     memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
1191 
1192     /*
1193      * Check if an HPET shall be created.
1194      */
1195     if (pcms->hpet_enabled) {
1196         qemu_irq rtc_irq;
1197 
1198         hpet = qdev_try_new(TYPE_HPET);
1199         if (!hpet) {
1200             error_report("couldn't create HPET device");
1201             exit(1);
1202         }
1203         /*
1204          * For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-*,
1205          * use IRQ16~23, IRQ8 and IRQ2.  If the user has already set
1206          * the property, use whatever mask they specified.
1207          */
1208         uint8_t compat = object_property_get_uint(OBJECT(hpet),
1209                 HPET_INTCAP, NULL);
1210         if (!compat) {
1211             qdev_prop_set_uint32(hpet, HPET_INTCAP, hpet_irqs);
1212         }
1213         sysbus_realize_and_unref(SYS_BUS_DEVICE(hpet), &error_fatal);
1214         sysbus_mmio_map(SYS_BUS_DEVICE(hpet), 0, HPET_BASE);
1215 
1216         for (i = 0; i < IOAPIC_NUM_PINS; i++) {
1217             sysbus_connect_irq(SYS_BUS_DEVICE(hpet), i, gsi[i]);
1218         }
1219         pit_isa_irq = -1;
1220         pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1221         rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1222 
1223         /* overwrite connection created by south bridge */
1224         qdev_connect_gpio_out(DEVICE(rtc_state), 0, rtc_irq);
1225     }
1226 
1227     object_property_add_alias(OBJECT(pcms), "rtc-time", OBJECT(rtc_state),
1228                               "date");
1229 
1230 #ifdef CONFIG_XEN_EMU
1231     if (xen_mode == XEN_EMULATE) {
1232         xen_overlay_create();
1233         xen_evtchn_create(IOAPIC_NUM_PINS, gsi);
1234         xen_gnttab_create();
1235         xen_xenstore_create();
1236         if (pcms->pcibus) {
1237             pci_create_simple(pcms->pcibus, -1, "xen-platform");
1238         }
1239         xen_bus_init();
1240         xen_be_init();
1241     }
1242 #endif
1243 
1244     qemu_register_boot_set(pc_boot_set, pcms);
1245     set_boot_dev(pcms, MC146818_RTC(rtc_state),
1246                  MACHINE(pcms)->boot_config.order, &error_fatal);
1247 
1248     if (!xen_enabled() &&
1249         (x86ms->pit == ON_OFF_AUTO_AUTO || x86ms->pit == ON_OFF_AUTO_ON)) {
1250         if (kvm_pit_in_kernel()) {
1251             pit = kvm_pit_init(isa_bus, 0x40);
1252         } else {
1253             pit = i8254_pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1254         }
1255         if (hpet) {
1256             /* connect PIT to output control line of the HPET */
1257             qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(DEVICE(pit), 0));
1258         }
1259         object_property_set_link(OBJECT(pcms->pcspk), "pit",
1260                                  OBJECT(pit), &error_fatal);
1261         isa_realize_and_unref(pcms->pcspk, isa_bus, &error_fatal);
1262     }
1263 
1264     /* Super I/O */
1265     pc_superio_init(isa_bus, create_fdctrl, pcms->i8042_enabled,
1266                     pcms->vmport != ON_OFF_AUTO_ON);
1267 }
1268 
1269 void pc_nic_init(PCMachineClass *pcmc, ISABus *isa_bus, PCIBus *pci_bus)
1270 {
1271     MachineClass *mc = MACHINE_CLASS(pcmc);
1272     bool default_is_ne2k = g_str_equal(mc->default_nic, TYPE_ISA_NE2000);
1273     NICInfo *nd;
1274 
1275     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_NIC);
1276 
1277     while ((nd = qemu_find_nic_info(TYPE_ISA_NE2000, default_is_ne2k, NULL))) {
1278         pc_init_ne2k_isa(isa_bus, nd, &error_fatal);
1279     }
1280 
1281     /* Anything remaining should be a PCI NIC */
1282     pci_init_nic_devices(pci_bus, mc->default_nic);
1283 
1284     rom_reset_order_override();
1285 }
1286 
1287 void pc_i8259_create(ISABus *isa_bus, qemu_irq *i8259_irqs)
1288 {
1289     qemu_irq *i8259;
1290 
1291     if (kvm_pic_in_kernel()) {
1292         i8259 = kvm_i8259_init(isa_bus);
1293     } else if (xen_enabled()) {
1294         i8259 = xen_interrupt_controller_init();
1295     } else {
1296         i8259 = i8259_init(isa_bus, x86_allocate_cpu_irq());
1297     }
1298 
1299     for (size_t i = 0; i < ISA_NUM_IRQS; i++) {
1300         i8259_irqs[i] = i8259[i];
1301     }
1302 
1303     g_free(i8259);
1304 }
1305 
1306 static void pc_memory_pre_plug(HotplugHandler *hotplug_dev, DeviceState *dev,
1307                                Error **errp)
1308 {
1309     const PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1310     const X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1311     const PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1312     const MachineState *ms = MACHINE(hotplug_dev);
1313     const bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1314     const uint64_t legacy_align = TARGET_PAGE_SIZE;
1315     Error *local_err = NULL;
1316 
1317     /*
1318      * When "acpi=off" is used with the Q35 machine type, no ACPI is built,
1319      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1320      * addition to cover this case.
1321      */
1322     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1323         error_setg(errp,
1324                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1325         return;
1326     }
1327 
1328     if (is_nvdimm && !ms->nvdimms_state->is_enabled) {
1329         error_setg(errp, "nvdimm is not enabled: missing 'nvdimm' in '-M'");
1330         return;
1331     }
1332 
1333     hotplug_handler_pre_plug(x86ms->acpi_dev, dev, &local_err);
1334     if (local_err) {
1335         error_propagate(errp, local_err);
1336         return;
1337     }
1338 
1339     pc_dimm_pre_plug(PC_DIMM(dev), MACHINE(hotplug_dev),
1340                      pcmc->enforce_aligned_dimm ? NULL : &legacy_align, errp);
1341 }
1342 
1343 static void pc_memory_plug(HotplugHandler *hotplug_dev,
1344                            DeviceState *dev, Error **errp)
1345 {
1346     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1347     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1348     MachineState *ms = MACHINE(hotplug_dev);
1349     bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1350 
1351     pc_dimm_plug(PC_DIMM(dev), MACHINE(pcms));
1352 
1353     if (is_nvdimm) {
1354         nvdimm_plug(ms->nvdimms_state);
1355     }
1356 
1357     hotplug_handler_plug(x86ms->acpi_dev, dev, &error_abort);
1358 }
1359 
1360 static void pc_memory_unplug_request(HotplugHandler *hotplug_dev,
1361                                      DeviceState *dev, Error **errp)
1362 {
1363     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1364 
1365     /*
1366      * When "acpi=off" is used with the Q35 machine type, no ACPI is built,
1367      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1368      * addition to cover this case.
1369      */
1370     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1371         error_setg(errp,
1372                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1373         return;
1374     }
1375 
1376     if (object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM)) {
1377         error_setg(errp, "nvdimm device hot unplug is not supported yet.");
1378         return;
1379     }
1380 
1381     hotplug_handler_unplug_request(x86ms->acpi_dev, dev,
1382                                    errp);
1383 }
1384 
1385 static void pc_memory_unplug(HotplugHandler *hotplug_dev,
1386                              DeviceState *dev, Error **errp)
1387 {
1388     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1389     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1390     Error *local_err = NULL;
1391 
1392     hotplug_handler_unplug(x86ms->acpi_dev, dev, &local_err);
1393     if (local_err) {
1394         goto out;
1395     }
1396 
1397     pc_dimm_unplug(PC_DIMM(dev), MACHINE(pcms));
1398     qdev_unrealize(dev);
1399  out:
1400     error_propagate(errp, local_err);
1401 }
1402 
1403 static void pc_hv_balloon_pre_plug(HotplugHandler *hotplug_dev,
1404                                    DeviceState *dev, Error **errp)
1405 {
1406     /* The vmbus handler has no hotplug handler; we should never end up here. */
1407     g_assert(!dev->hotplugged);
1408     memory_device_pre_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev), NULL,
1409                            errp);
1410 }
1411 
1412 static void pc_hv_balloon_plug(HotplugHandler *hotplug_dev,
1413                                DeviceState *dev, Error **errp)
1414 {
1415     memory_device_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev));
1416 }
1417 
1418 static void pc_machine_device_pre_plug_cb(HotplugHandler *hotplug_dev,
1419                                           DeviceState *dev, Error **errp)
1420 {
1421     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1422         pc_memory_pre_plug(hotplug_dev, dev, errp);
1423     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1424         x86_cpu_pre_plug(hotplug_dev, dev, errp);
1425     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1426         virtio_md_pci_pre_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1427     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1428         /* Declare the APIC range as the reserved MSI region */
1429         char *resv_prop_str = g_strdup_printf("0xfee00000:0xfeefffff:%d",
1430                                               VIRTIO_IOMMU_RESV_MEM_T_MSI);
1431         QList *reserved_regions = qlist_new();
1432 
1433         qlist_append_str(reserved_regions, resv_prop_str);
1434         qdev_prop_set_array(dev, "reserved-regions", reserved_regions);
1435 
1436         g_free(resv_prop_str);
1437     }
1438 
1439     if (object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE) ||
1440         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1441         PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1442 
1443         if (pcms->iommu) {
1444             error_setg(errp, "QEMU does not support multiple vIOMMUs "
1445                        "for x86 yet.");
1446             return;
1447         }
1448         pcms->iommu = dev;
1449     } else if (object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON)) {
1450         pc_hv_balloon_pre_plug(hotplug_dev, dev, errp);
1451     }
1452 }
1453 
1454 static void pc_machine_device_plug_cb(HotplugHandler *hotplug_dev,
1455                                       DeviceState *dev, Error **errp)
1456 {
1457     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1458         pc_memory_plug(hotplug_dev, dev, errp);
1459     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1460         x86_cpu_plug(hotplug_dev, dev, errp);
1461     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1462         virtio_md_pci_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1463     } else if (object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON)) {
1464         pc_hv_balloon_plug(hotplug_dev, dev, errp);
1465     }
1466 }
1467 
1468 static void pc_machine_device_unplug_request_cb(HotplugHandler *hotplug_dev,
1469                                                 DeviceState *dev, Error **errp)
1470 {
1471     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1472         pc_memory_unplug_request(hotplug_dev, dev, errp);
1473     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1474         x86_cpu_unplug_request_cb(hotplug_dev, dev, errp);
1475     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1476         virtio_md_pci_unplug_request(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev),
1477                                      errp);
1478     } else {
1479         error_setg(errp, "acpi: device unplug request for not supported device"
1480                    " type: %s", object_get_typename(OBJECT(dev)));
1481     }
1482 }
1483 
1484 static void pc_machine_device_unplug_cb(HotplugHandler *hotplug_dev,
1485                                         DeviceState *dev, Error **errp)
1486 {
1487     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1488         pc_memory_unplug(hotplug_dev, dev, errp);
1489     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1490         x86_cpu_unplug_cb(hotplug_dev, dev, errp);
1491     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1492         virtio_md_pci_unplug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1493     } else {
1494         error_setg(errp, "acpi: device unplug for not supported device"
1495                    " type: %s", object_get_typename(OBJECT(dev)));
1496     }
1497 }
1498 
1499 static HotplugHandler *pc_get_hotplug_handler(MachineState *machine,
1500                                              DeviceState *dev)
1501 {
1502     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM) ||
1503         object_dynamic_cast(OBJECT(dev), TYPE_CPU) ||
1504         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI) ||
1505         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI) ||
1506         object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON) ||
1507         object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE)) {
1508         return HOTPLUG_HANDLER(machine);
1509     }
1510 
1511     return NULL;
1512 }
1513 
1514 static void pc_machine_get_vmport(Object *obj, Visitor *v, const char *name,
1515                                   void *opaque, Error **errp)
1516 {
1517     PCMachineState *pcms = PC_MACHINE(obj);
1518     OnOffAuto vmport = pcms->vmport;
1519 
1520     visit_type_OnOffAuto(v, name, &vmport, errp);
1521 }
1522 
1523 static void pc_machine_set_vmport(Object *obj, Visitor *v, const char *name,
1524                                   void *opaque, Error **errp)
1525 {
1526     PCMachineState *pcms = PC_MACHINE(obj);
1527 
1528     visit_type_OnOffAuto(v, name, &pcms->vmport, errp);
1529 }
1530 
1531 static bool pc_machine_get_fd_bootchk(Object *obj, Error **errp)
1532 {
1533     PCMachineState *pcms = PC_MACHINE(obj);
1534 
1535     return pcms->fd_bootchk;
1536 }
1537 
1538 static void pc_machine_set_fd_bootchk(Object *obj, bool value, Error **errp)
1539 {
1540     PCMachineState *pcms = PC_MACHINE(obj);
1541 
1542     pcms->fd_bootchk = value;
1543 }
1544 
1545 static bool pc_machine_get_smbus(Object *obj, Error **errp)
1546 {
1547     PCMachineState *pcms = PC_MACHINE(obj);
1548 
1549     return pcms->smbus_enabled;
1550 }
1551 
1552 static void pc_machine_set_smbus(Object *obj, bool value, Error **errp)
1553 {
1554     PCMachineState *pcms = PC_MACHINE(obj);
1555 
1556     pcms->smbus_enabled = value;
1557 }
1558 
1559 static bool pc_machine_get_sata(Object *obj, Error **errp)
1560 {
1561     PCMachineState *pcms = PC_MACHINE(obj);
1562 
1563     return pcms->sata_enabled;
1564 }
1565 
1566 static void pc_machine_set_sata(Object *obj, bool value, Error **errp)
1567 {
1568     PCMachineState *pcms = PC_MACHINE(obj);
1569 
1570     pcms->sata_enabled = value;
1571 }
1572 
1573 static bool pc_machine_get_hpet(Object *obj, Error **errp)
1574 {
1575     PCMachineState *pcms = PC_MACHINE(obj);
1576 
1577     return pcms->hpet_enabled;
1578 }
1579 
1580 static void pc_machine_set_hpet(Object *obj, bool value, Error **errp)
1581 {
1582     PCMachineState *pcms = PC_MACHINE(obj);
1583 
1584     pcms->hpet_enabled = value;
1585 }
1586 
1587 static bool pc_machine_get_i8042(Object *obj, Error **errp)
1588 {
1589     PCMachineState *pcms = PC_MACHINE(obj);
1590 
1591     return pcms->i8042_enabled;
1592 }
1593 
1594 static void pc_machine_set_i8042(Object *obj, bool value, Error **errp)
1595 {
1596     PCMachineState *pcms = PC_MACHINE(obj);
1597 
1598     pcms->i8042_enabled = value;
1599 }
1600 
1601 static bool pc_machine_get_default_bus_bypass_iommu(Object *obj, Error **errp)
1602 {
1603     PCMachineState *pcms = PC_MACHINE(obj);
1604 
1605     return pcms->default_bus_bypass_iommu;
1606 }
1607 
1608 static void pc_machine_set_default_bus_bypass_iommu(Object *obj, bool value,
1609                                                     Error **errp)
1610 {
1611     PCMachineState *pcms = PC_MACHINE(obj);
1612 
1613     pcms->default_bus_bypass_iommu = value;
1614 }
1615 
1616 static void pc_machine_get_smbios_ep(Object *obj, Visitor *v, const char *name,
1617                                      void *opaque, Error **errp)
1618 {
1619     PCMachineState *pcms = PC_MACHINE(obj);
1620     SmbiosEntryPointType smbios_entry_point_type = pcms->smbios_entry_point_type;
1621 
1622     visit_type_SmbiosEntryPointType(v, name, &smbios_entry_point_type, errp);
1623 }
1624 
1625 static void pc_machine_set_smbios_ep(Object *obj, Visitor *v, const char *name,
1626                                      void *opaque, Error **errp)
1627 {
1628     PCMachineState *pcms = PC_MACHINE(obj);
1629 
1630     visit_type_SmbiosEntryPointType(v, name, &pcms->smbios_entry_point_type, errp);
1631 }
1632 
1633 static void pc_machine_get_max_ram_below_4g(Object *obj, Visitor *v,
1634                                             const char *name, void *opaque,
1635                                             Error **errp)
1636 {
1637     PCMachineState *pcms = PC_MACHINE(obj);
1638     uint64_t value = pcms->max_ram_below_4g;
1639 
1640     visit_type_size(v, name, &value, errp);
1641 }
1642 
1643 static void pc_machine_set_max_ram_below_4g(Object *obj, Visitor *v,
1644                                             const char *name, void *opaque,
1645                                             Error **errp)
1646 {
1647     PCMachineState *pcms = PC_MACHINE(obj);
1648     uint64_t value;
1649 
1650     if (!visit_type_size(v, name, &value, errp)) {
1651         return;
1652     }
1653     if (value > 4 * GiB) {
1654         error_setg(errp,
1655                    "Machine option 'max-ram-below-4g=%"PRIu64
1656                    "' expects size less than or equal to 4G", value);
1657         return;
1658     }
1659 
1660     if (value < 1 * MiB) {
1661         warn_report("Only %" PRIu64 " bytes of RAM below the 4GiB boundary,"
1662                     "BIOS may not work with less than 1MiB", value);
1663     }
1664 
1665     pcms->max_ram_below_4g = value;
1666 }
1667 
1668 static void pc_machine_get_max_fw_size(Object *obj, Visitor *v,
1669                                        const char *name, void *opaque,
1670                                        Error **errp)
1671 {
1672     PCMachineState *pcms = PC_MACHINE(obj);
1673     uint64_t value = pcms->max_fw_size;
1674 
1675     visit_type_size(v, name, &value, errp);
1676 }
1677 
1678 static void pc_machine_set_max_fw_size(Object *obj, Visitor *v,
1679                                        const char *name, void *opaque,
1680                                        Error **errp)
1681 {
1682     PCMachineState *pcms = PC_MACHINE(obj);
1683     uint64_t value;
1684 
1685     if (!visit_type_size(v, name, &value, errp)) {
1686         return;
1687     }
1688 
1689     /*
1690      * We don't have a theoretically justifiable exact lower bound on the base
1691      * address of any flash mapping. In practice, the IO-APIC MMIO range is
1692      * [0xFEE00000..0xFEE01000] -- see IO_APIC_DEFAULT_ADDRESS --, leaving free
1693      * only 18MiB-4KiB below 4GiB. For now, restrict the cumulative mapping to
1694      * 16MiB in size.
1695      */
1696     if (value > 16 * MiB) {
1697         error_setg(errp,
1698                    "User specified max allowed firmware size %" PRIu64 " is "
1699                    "greater than 16MiB. If combined firmware size exceeds "
1700                    "16MiB the system may not boot, or experience intermittent"
1701                    "stability issues.",
1702                    value);
1703         return;
1704     }
1705 
1706     pcms->max_fw_size = value;
1707 }
1708 
1709 
1710 static void pc_machine_initfn(Object *obj)
1711 {
1712     PCMachineState *pcms = PC_MACHINE(obj);
1713     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1714 
1715 #ifdef CONFIG_VMPORT
1716     pcms->vmport = ON_OFF_AUTO_AUTO;
1717 #else
1718     pcms->vmport = ON_OFF_AUTO_OFF;
1719 #endif /* CONFIG_VMPORT */
1720     pcms->max_ram_below_4g = 0; /* use default */
1721     pcms->smbios_entry_point_type = pcmc->default_smbios_ep_type;
1722     pcms->south_bridge = pcmc->default_south_bridge;
1723 
1724     /* acpi build is enabled by default if machine supports it */
1725     pcms->acpi_build_enabled = pcmc->has_acpi_build;
1726     pcms->smbus_enabled = true;
1727     pcms->sata_enabled = true;
1728     pcms->i8042_enabled = true;
1729     pcms->max_fw_size = 8 * MiB;
1730 #ifdef CONFIG_HPET
1731     pcms->hpet_enabled = true;
1732 #endif
1733     pcms->fd_bootchk = true;
1734     pcms->default_bus_bypass_iommu = false;
1735 
1736     pc_system_flash_create(pcms);
1737     pcms->pcspk = isa_new(TYPE_PC_SPEAKER);
1738     object_property_add_alias(OBJECT(pcms), "pcspk-audiodev",
1739                               OBJECT(pcms->pcspk), "audiodev");
1740     cxl_machine_init(obj, &pcms->cxl_devices_state);
1741 
1742     pcms->machine_done.notify = pc_machine_done;
1743     qemu_add_machine_init_done_notifier(&pcms->machine_done);
1744 }
1745 
1746 static void pc_machine_reset(MachineState *machine, ShutdownCause reason)
1747 {
1748     CPUState *cs;
1749     X86CPU *cpu;
1750 
1751     qemu_devices_reset(reason);
1752 
1753     /* Reset APIC after devices have been reset to cancel
1754      * any changes that qemu_devices_reset() might have done.
1755      */
1756     CPU_FOREACH(cs) {
1757         cpu = X86_CPU(cs);
1758 
1759         x86_cpu_after_reset(cpu);
1760     }
1761 }
1762 
1763 static void pc_machine_wakeup(MachineState *machine)
1764 {
1765     cpu_synchronize_all_states();
1766     pc_machine_reset(machine, SHUTDOWN_CAUSE_NONE);
1767     cpu_synchronize_all_post_reset();
1768 }
1769 
1770 static bool pc_hotplug_allowed(MachineState *ms, DeviceState *dev, Error **errp)
1771 {
1772     X86IOMMUState *iommu = x86_iommu_get_default();
1773     IntelIOMMUState *intel_iommu;
1774 
1775     if (iommu &&
1776         object_dynamic_cast((Object *)iommu, TYPE_INTEL_IOMMU_DEVICE) &&
1777         object_dynamic_cast((Object *)dev, "vfio-pci")) {
1778         intel_iommu = INTEL_IOMMU_DEVICE(iommu);
1779         if (!intel_iommu->caching_mode) {
1780             error_setg(errp, "Device assignment is not allowed without "
1781                        "enabling caching-mode=on for Intel IOMMU.");
1782             return false;
1783         }
1784     }
1785 
1786     return true;
1787 }
1788 
1789 static void pc_machine_class_init(ObjectClass *oc, void *data)
1790 {
1791     MachineClass *mc = MACHINE_CLASS(oc);
1792     X86MachineClass *x86mc = X86_MACHINE_CLASS(oc);
1793     PCMachineClass *pcmc = PC_MACHINE_CLASS(oc);
1794     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
1795 
1796     pcmc->pci_enabled = true;
1797     pcmc->has_acpi_build = true;
1798     pcmc->rsdp_in_ram = true;
1799     pcmc->smbios_defaults = true;
1800     pcmc->smbios_uuid_encoded = true;
1801     pcmc->gigabyte_align = true;
1802     pcmc->has_reserved_memory = true;
1803     pcmc->enforce_aligned_dimm = true;
1804     pcmc->enforce_amd_1tb_hole = true;
1805     /* BIOS ACPI tables: 128K. Other BIOS datastructures: less than 4K reported
1806      * to be used at the moment, 32K should be enough for a while.  */
1807     pcmc->acpi_data_size = 0x20000 + 0x8000;
1808     pcmc->pvh_enabled = true;
1809     pcmc->kvmclock_create_always = true;
1810     pcmc->resizable_acpi_blob = true;
1811     x86mc->apic_xrupt_override = true;
1812     assert(!mc->get_hotplug_handler);
1813     mc->get_hotplug_handler = pc_get_hotplug_handler;
1814     mc->hotplug_allowed = pc_hotplug_allowed;
1815     mc->cpu_index_to_instance_props = x86_cpu_index_to_props;
1816     mc->get_default_cpu_node_id = x86_get_default_cpu_node_id;
1817     mc->possible_cpu_arch_ids = x86_possible_cpu_arch_ids;
1818     mc->auto_enable_numa_with_memhp = true;
1819     mc->auto_enable_numa_with_memdev = true;
1820     mc->has_hotpluggable_cpus = true;
1821     mc->default_boot_order = "cad";
1822     mc->block_default_type = IF_IDE;
1823     mc->max_cpus = 255;
1824     mc->reset = pc_machine_reset;
1825     mc->wakeup = pc_machine_wakeup;
1826     hc->pre_plug = pc_machine_device_pre_plug_cb;
1827     hc->plug = pc_machine_device_plug_cb;
1828     hc->unplug_request = pc_machine_device_unplug_request_cb;
1829     hc->unplug = pc_machine_device_unplug_cb;
1830     mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE;
1831     mc->nvdimm_supported = true;
1832     mc->smp_props.dies_supported = true;
1833     mc->default_ram_id = "pc.ram";
1834     pcmc->default_smbios_ep_type = SMBIOS_ENTRY_POINT_TYPE_64;
1835 
1836     object_class_property_add(oc, PC_MACHINE_MAX_RAM_BELOW_4G, "size",
1837         pc_machine_get_max_ram_below_4g, pc_machine_set_max_ram_below_4g,
1838         NULL, NULL);
1839     object_class_property_set_description(oc, PC_MACHINE_MAX_RAM_BELOW_4G,
1840         "Maximum ram below the 4G boundary (32bit boundary)");
1841 
1842     object_class_property_add(oc, PC_MACHINE_VMPORT, "OnOffAuto",
1843         pc_machine_get_vmport, pc_machine_set_vmport,
1844         NULL, NULL);
1845     object_class_property_set_description(oc, PC_MACHINE_VMPORT,
1846         "Enable vmport (pc & q35)");
1847 
1848     object_class_property_add_bool(oc, PC_MACHINE_SMBUS,
1849         pc_machine_get_smbus, pc_machine_set_smbus);
1850     object_class_property_set_description(oc, PC_MACHINE_SMBUS,
1851         "Enable/disable system management bus");
1852 
1853     object_class_property_add_bool(oc, PC_MACHINE_SATA,
1854         pc_machine_get_sata, pc_machine_set_sata);
1855     object_class_property_set_description(oc, PC_MACHINE_SATA,
1856         "Enable/disable Serial ATA bus");
1857 
1858     object_class_property_add_bool(oc, "hpet",
1859         pc_machine_get_hpet, pc_machine_set_hpet);
1860     object_class_property_set_description(oc, "hpet",
1861         "Enable/disable high precision event timer emulation");
1862 
1863     object_class_property_add_bool(oc, PC_MACHINE_I8042,
1864         pc_machine_get_i8042, pc_machine_set_i8042);
1865 
1866     object_class_property_add_bool(oc, "default-bus-bypass-iommu",
1867         pc_machine_get_default_bus_bypass_iommu,
1868         pc_machine_set_default_bus_bypass_iommu);
1869 
1870     object_class_property_add(oc, PC_MACHINE_MAX_FW_SIZE, "size",
1871         pc_machine_get_max_fw_size, pc_machine_set_max_fw_size,
1872         NULL, NULL);
1873     object_class_property_set_description(oc, PC_MACHINE_MAX_FW_SIZE,
1874         "Maximum combined firmware size");
1875 
1876     object_class_property_add(oc, PC_MACHINE_SMBIOS_EP, "str",
1877         pc_machine_get_smbios_ep, pc_machine_set_smbios_ep,
1878         NULL, NULL);
1879     object_class_property_set_description(oc, PC_MACHINE_SMBIOS_EP,
1880         "SMBIOS Entry Point type [32, 64]");
1881 
1882     object_class_property_add_bool(oc, "fd-bootchk",
1883         pc_machine_get_fd_bootchk,
1884         pc_machine_set_fd_bootchk);
1885 }
1886 
1887 static const TypeInfo pc_machine_info = {
1888     .name = TYPE_PC_MACHINE,
1889     .parent = TYPE_X86_MACHINE,
1890     .abstract = true,
1891     .instance_size = sizeof(PCMachineState),
1892     .instance_init = pc_machine_initfn,
1893     .class_size = sizeof(PCMachineClass),
1894     .class_init = pc_machine_class_init,
1895     .interfaces = (InterfaceInfo[]) {
1896          { TYPE_HOTPLUG_HANDLER },
1897          { }
1898     },
1899 };
1900 
1901 static void pc_machine_register_types(void)
1902 {
1903     type_register_static(&pc_machine_info);
1904 }
1905 
1906 type_init(pc_machine_register_types)
1907