xref: /openbmc/qemu/hw/i386/pc.c (revision ab0c7fb2)
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_9_0[] = {
82     { TYPE_X86_CPU, "guest-phys-bits", "0" },
83 };
84 const size_t pc_compat_9_0_len = G_N_ELEMENTS(pc_compat_9_0);
85 
86 GlobalProperty pc_compat_8_2[] = {};
87 const size_t pc_compat_8_2_len = G_N_ELEMENTS(pc_compat_8_2);
88 
89 GlobalProperty pc_compat_8_1[] = {};
90 const size_t pc_compat_8_1_len = G_N_ELEMENTS(pc_compat_8_1);
91 
92 GlobalProperty pc_compat_8_0[] = {
93     { "virtio-mem", "unplugged-inaccessible", "auto" },
94 };
95 const size_t pc_compat_8_0_len = G_N_ELEMENTS(pc_compat_8_0);
96 
97 GlobalProperty pc_compat_7_2[] = {
98     { "ICH9-LPC", "noreboot", "true" },
99 };
100 const size_t pc_compat_7_2_len = G_N_ELEMENTS(pc_compat_7_2);
101 
102 GlobalProperty pc_compat_7_1[] = {};
103 const size_t pc_compat_7_1_len = G_N_ELEMENTS(pc_compat_7_1);
104 
105 GlobalProperty pc_compat_7_0[] = {};
106 const size_t pc_compat_7_0_len = G_N_ELEMENTS(pc_compat_7_0);
107 
108 GlobalProperty pc_compat_6_2[] = {
109     { "virtio-mem", "unplugged-inaccessible", "off" },
110 };
111 const size_t pc_compat_6_2_len = G_N_ELEMENTS(pc_compat_6_2);
112 
113 GlobalProperty pc_compat_6_1[] = {
114     { TYPE_X86_CPU, "hv-version-id-build", "0x1bbc" },
115     { TYPE_X86_CPU, "hv-version-id-major", "0x0006" },
116     { TYPE_X86_CPU, "hv-version-id-minor", "0x0001" },
117     { "ICH9-LPC", "x-keep-pci-slot-hpc", "false" },
118 };
119 const size_t pc_compat_6_1_len = G_N_ELEMENTS(pc_compat_6_1);
120 
121 GlobalProperty pc_compat_6_0[] = {
122     { "qemu64" "-" TYPE_X86_CPU, "family", "6" },
123     { "qemu64" "-" TYPE_X86_CPU, "model", "6" },
124     { "qemu64" "-" TYPE_X86_CPU, "stepping", "3" },
125     { TYPE_X86_CPU, "x-vendor-cpuid-only", "off" },
126     { "ICH9-LPC", ACPI_PM_PROP_ACPI_PCIHP_BRIDGE, "off" },
127     { "ICH9-LPC", "x-keep-pci-slot-hpc", "true" },
128 };
129 const size_t pc_compat_6_0_len = G_N_ELEMENTS(pc_compat_6_0);
130 
131 GlobalProperty pc_compat_5_2[] = {
132     { "ICH9-LPC", "x-smi-cpu-hotunplug", "off" },
133 };
134 const size_t pc_compat_5_2_len = G_N_ELEMENTS(pc_compat_5_2);
135 
136 GlobalProperty pc_compat_5_1[] = {
137     { "ICH9-LPC", "x-smi-cpu-hotplug", "off" },
138     { TYPE_X86_CPU, "kvm-msi-ext-dest-id", "off" },
139 };
140 const size_t pc_compat_5_1_len = G_N_ELEMENTS(pc_compat_5_1);
141 
142 GlobalProperty pc_compat_5_0[] = {
143 };
144 const size_t pc_compat_5_0_len = G_N_ELEMENTS(pc_compat_5_0);
145 
146 GlobalProperty pc_compat_4_2[] = {
147     { "mch", "smbase-smram", "off" },
148 };
149 const size_t pc_compat_4_2_len = G_N_ELEMENTS(pc_compat_4_2);
150 
151 GlobalProperty pc_compat_4_1[] = {};
152 const size_t pc_compat_4_1_len = G_N_ELEMENTS(pc_compat_4_1);
153 
154 GlobalProperty pc_compat_4_0[] = {};
155 const size_t pc_compat_4_0_len = G_N_ELEMENTS(pc_compat_4_0);
156 
157 GlobalProperty pc_compat_3_1[] = {
158     { "intel-iommu", "dma-drain", "off" },
159     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "off" },
160     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "off" },
161     { "Opteron_G4" "-" TYPE_X86_CPU, "npt", "off" },
162     { "Opteron_G4" "-" TYPE_X86_CPU, "nrip-save", "off" },
163     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "off" },
164     { "Opteron_G5" "-" TYPE_X86_CPU, "npt", "off" },
165     { "Opteron_G5" "-" TYPE_X86_CPU, "nrip-save", "off" },
166     { "EPYC" "-" TYPE_X86_CPU, "npt", "off" },
167     { "EPYC" "-" TYPE_X86_CPU, "nrip-save", "off" },
168     { "EPYC-IBPB" "-" TYPE_X86_CPU, "npt", "off" },
169     { "EPYC-IBPB" "-" TYPE_X86_CPU, "nrip-save", "off" },
170     { "Skylake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
171     { "Skylake-Client-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
172     { "Skylake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
173     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
174     { "Cascadelake-Server" "-" TYPE_X86_CPU,  "mpx", "on" },
175     { "Icelake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
176     { "Icelake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
177     { "Cascadelake-Server" "-" TYPE_X86_CPU, "stepping", "5" },
178     { TYPE_X86_CPU, "x-intel-pt-auto-level", "off" },
179 };
180 const size_t pc_compat_3_1_len = G_N_ELEMENTS(pc_compat_3_1);
181 
182 GlobalProperty pc_compat_3_0[] = {
183     { TYPE_X86_CPU, "x-hv-synic-kvm-only", "on" },
184     { "Skylake-Server" "-" TYPE_X86_CPU, "pku", "off" },
185     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "pku", "off" },
186 };
187 const size_t pc_compat_3_0_len = G_N_ELEMENTS(pc_compat_3_0);
188 
189 GlobalProperty pc_compat_2_12[] = {
190     { TYPE_X86_CPU, "legacy-cache", "on" },
191     { TYPE_X86_CPU, "topoext", "off" },
192     { "EPYC-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
193     { "EPYC-IBPB-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
194 };
195 const size_t pc_compat_2_12_len = G_N_ELEMENTS(pc_compat_2_12);
196 
197 GlobalProperty pc_compat_2_11[] = {
198     { TYPE_X86_CPU, "x-migrate-smi-count", "off" },
199     { "Skylake-Server" "-" TYPE_X86_CPU, "clflushopt", "off" },
200 };
201 const size_t pc_compat_2_11_len = G_N_ELEMENTS(pc_compat_2_11);
202 
203 GlobalProperty pc_compat_2_10[] = {
204     { TYPE_X86_CPU, "x-hv-max-vps", "0x40" },
205     { "i440FX-pcihost", "x-pci-hole64-fix", "off" },
206     { "q35-pcihost", "x-pci-hole64-fix", "off" },
207 };
208 const size_t pc_compat_2_10_len = G_N_ELEMENTS(pc_compat_2_10);
209 
210 GlobalProperty pc_compat_2_9[] = {
211     { "mch", "extended-tseg-mbytes", "0" },
212 };
213 const size_t pc_compat_2_9_len = G_N_ELEMENTS(pc_compat_2_9);
214 
215 GlobalProperty pc_compat_2_8[] = {
216     { TYPE_X86_CPU, "tcg-cpuid", "off" },
217     { "kvmclock", "x-mach-use-reliable-get-clock", "off" },
218     { "ICH9-LPC", "x-smi-broadcast", "off" },
219     { TYPE_X86_CPU, "vmware-cpuid-freq", "off" },
220     { "Haswell-" TYPE_X86_CPU, "stepping", "1" },
221 };
222 const size_t pc_compat_2_8_len = G_N_ELEMENTS(pc_compat_2_8);
223 
224 GlobalProperty pc_compat_2_7[] = {
225     { TYPE_X86_CPU, "l3-cache", "off" },
226     { TYPE_X86_CPU, "full-cpuid-auto-level", "off" },
227     { "Opteron_G3" "-" TYPE_X86_CPU, "family", "15" },
228     { "Opteron_G3" "-" TYPE_X86_CPU, "model", "6" },
229     { "Opteron_G3" "-" TYPE_X86_CPU, "stepping", "1" },
230     { "isa-pcspk", "migrate", "off" },
231 };
232 const size_t pc_compat_2_7_len = G_N_ELEMENTS(pc_compat_2_7);
233 
234 GlobalProperty pc_compat_2_6[] = {
235     { TYPE_X86_CPU, "cpuid-0xb", "off" },
236     { "vmxnet3", "romfile", "" },
237     { TYPE_X86_CPU, "fill-mtrr-mask", "off" },
238     { "apic-common", "legacy-instance-id", "on", }
239 };
240 const size_t pc_compat_2_6_len = G_N_ELEMENTS(pc_compat_2_6);
241 
242 GlobalProperty pc_compat_2_5[] = {};
243 const size_t pc_compat_2_5_len = G_N_ELEMENTS(pc_compat_2_5);
244 
245 GlobalProperty pc_compat_2_4[] = {
246     PC_CPU_MODEL_IDS("2.4.0")
247     { "Haswell-" TYPE_X86_CPU, "abm", "off" },
248     { "Haswell-noTSX-" TYPE_X86_CPU, "abm", "off" },
249     { "Broadwell-" TYPE_X86_CPU, "abm", "off" },
250     { "Broadwell-noTSX-" TYPE_X86_CPU, "abm", "off" },
251     { "host" "-" TYPE_X86_CPU, "host-cache-info", "on" },
252     { TYPE_X86_CPU, "check", "off" },
253     { "qemu64" "-" TYPE_X86_CPU, "sse4a", "on" },
254     { "qemu64" "-" TYPE_X86_CPU, "abm", "on" },
255     { "qemu64" "-" TYPE_X86_CPU, "popcnt", "on" },
256     { "qemu32" "-" TYPE_X86_CPU, "popcnt", "on" },
257     { "Opteron_G2" "-" TYPE_X86_CPU, "rdtscp", "on" },
258     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "on" },
259     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "on" },
260     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "on", }
261 };
262 const size_t pc_compat_2_4_len = G_N_ELEMENTS(pc_compat_2_4);
263 
264 GlobalProperty pc_compat_2_3[] = {
265     PC_CPU_MODEL_IDS("2.3.0")
266     { TYPE_X86_CPU, "arat", "off" },
267     { "qemu64" "-" TYPE_X86_CPU, "min-level", "4" },
268     { "kvm64" "-" TYPE_X86_CPU, "min-level", "5" },
269     { "pentium3" "-" TYPE_X86_CPU, "min-level", "2" },
270     { "n270" "-" TYPE_X86_CPU, "min-level", "5" },
271     { "Conroe" "-" TYPE_X86_CPU, "min-level", "4" },
272     { "Penryn" "-" TYPE_X86_CPU, "min-level", "4" },
273     { "Nehalem" "-" TYPE_X86_CPU, "min-level", "4" },
274     { "n270" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
275     { "Penryn" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
276     { "Conroe" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
277     { "Nehalem" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
278     { "Westmere" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
279     { "SandyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
280     { "IvyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
281     { "Haswell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
282     { "Haswell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
283     { "Broadwell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
284     { "Broadwell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
285     { TYPE_X86_CPU, "kvm-no-smi-migration", "on" },
286 };
287 const size_t pc_compat_2_3_len = G_N_ELEMENTS(pc_compat_2_3);
288 
289 GlobalProperty pc_compat_2_2[] = {
290     PC_CPU_MODEL_IDS("2.2.0")
291     { "kvm64" "-" TYPE_X86_CPU, "vme", "off" },
292     { "kvm32" "-" TYPE_X86_CPU, "vme", "off" },
293     { "Conroe" "-" TYPE_X86_CPU, "vme", "off" },
294     { "Penryn" "-" TYPE_X86_CPU, "vme", "off" },
295     { "Nehalem" "-" TYPE_X86_CPU, "vme", "off" },
296     { "Westmere" "-" TYPE_X86_CPU, "vme", "off" },
297     { "SandyBridge" "-" TYPE_X86_CPU, "vme", "off" },
298     { "Haswell" "-" TYPE_X86_CPU, "vme", "off" },
299     { "Broadwell" "-" TYPE_X86_CPU, "vme", "off" },
300     { "Opteron_G1" "-" TYPE_X86_CPU, "vme", "off" },
301     { "Opteron_G2" "-" TYPE_X86_CPU, "vme", "off" },
302     { "Opteron_G3" "-" TYPE_X86_CPU, "vme", "off" },
303     { "Opteron_G4" "-" TYPE_X86_CPU, "vme", "off" },
304     { "Opteron_G5" "-" TYPE_X86_CPU, "vme", "off" },
305     { "Haswell" "-" TYPE_X86_CPU, "f16c", "off" },
306     { "Haswell" "-" TYPE_X86_CPU, "rdrand", "off" },
307     { "Broadwell" "-" TYPE_X86_CPU, "f16c", "off" },
308     { "Broadwell" "-" TYPE_X86_CPU, "rdrand", "off" },
309 };
310 const size_t pc_compat_2_2_len = G_N_ELEMENTS(pc_compat_2_2);
311 
312 GlobalProperty pc_compat_2_1[] = {
313     PC_CPU_MODEL_IDS("2.1.0")
314     { "coreduo" "-" TYPE_X86_CPU, "vmx", "on" },
315     { "core2duo" "-" TYPE_X86_CPU, "vmx", "on" },
316 };
317 const size_t pc_compat_2_1_len = G_N_ELEMENTS(pc_compat_2_1);
318 
319 GlobalProperty pc_compat_2_0[] = {
320     PC_CPU_MODEL_IDS("2.0.0")
321     { "virtio-scsi-pci", "any_layout", "off" },
322     { "PIIX4_PM", "memory-hotplug-support", "off" },
323     { "apic", "version", "0x11" },
324     { "nec-usb-xhci", "superspeed-ports-first", "off" },
325     { "nec-usb-xhci", "force-pcie-endcap", "on" },
326     { "pci-serial", "prog_if", "0" },
327     { "pci-serial-2x", "prog_if", "0" },
328     { "pci-serial-4x", "prog_if", "0" },
329     { "virtio-net-pci", "guest_announce", "off" },
330     { "ICH9-LPC", "memory-hotplug-support", "off" },
331 };
332 const size_t pc_compat_2_0_len = G_N_ELEMENTS(pc_compat_2_0);
333 
334 GSIState *pc_gsi_create(qemu_irq **irqs, bool pci_enabled)
335 {
336     GSIState *s;
337 
338     s = g_new0(GSIState, 1);
339     if (kvm_ioapic_in_kernel()) {
340         kvm_pc_setup_irq_routing(pci_enabled);
341     }
342     *irqs = qemu_allocate_irqs(gsi_handler, s, IOAPIC_NUM_PINS);
343 
344     return s;
345 }
346 
347 static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
348                            unsigned size)
349 {
350 }
351 
352 static uint64_t ioport80_read(void *opaque, hwaddr addr, unsigned size)
353 {
354     return 0xffffffffffffffffULL;
355 }
356 
357 /* MS-DOS compatibility mode FPU exception support */
358 static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
359                            unsigned size)
360 {
361     if (tcg_enabled()) {
362         cpu_set_ignne();
363     }
364 }
365 
366 static uint64_t ioportF0_read(void *opaque, hwaddr addr, unsigned size)
367 {
368     return 0xffffffffffffffffULL;
369 }
370 
371 /* PC cmos mappings */
372 
373 #define REG_EQUIPMENT_BYTE          0x14
374 
375 static void cmos_init_hd(MC146818RtcState *s, int type_ofs, int info_ofs,
376                          int16_t cylinders, int8_t heads, int8_t sectors)
377 {
378     mc146818rtc_set_cmos_data(s, type_ofs, 47);
379     mc146818rtc_set_cmos_data(s, info_ofs, cylinders);
380     mc146818rtc_set_cmos_data(s, info_ofs + 1, cylinders >> 8);
381     mc146818rtc_set_cmos_data(s, info_ofs + 2, heads);
382     mc146818rtc_set_cmos_data(s, info_ofs + 3, 0xff);
383     mc146818rtc_set_cmos_data(s, info_ofs + 4, 0xff);
384     mc146818rtc_set_cmos_data(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
385     mc146818rtc_set_cmos_data(s, info_ofs + 6, cylinders);
386     mc146818rtc_set_cmos_data(s, info_ofs + 7, cylinders >> 8);
387     mc146818rtc_set_cmos_data(s, info_ofs + 8, sectors);
388 }
389 
390 /* convert boot_device letter to something recognizable by the bios */
391 static int boot_device2nibble(char boot_device)
392 {
393     switch(boot_device) {
394     case 'a':
395     case 'b':
396         return 0x01; /* floppy boot */
397     case 'c':
398         return 0x02; /* hard drive boot */
399     case 'd':
400         return 0x03; /* CD-ROM boot */
401     case 'n':
402         return 0x04; /* Network boot */
403     }
404     return 0;
405 }
406 
407 static void set_boot_dev(PCMachineState *pcms, MC146818RtcState *s,
408                          const char *boot_device, Error **errp)
409 {
410 #define PC_MAX_BOOT_DEVICES 3
411     int nbds, bds[3] = { 0, };
412     int i;
413 
414     nbds = strlen(boot_device);
415     if (nbds > PC_MAX_BOOT_DEVICES) {
416         error_setg(errp, "Too many boot devices for PC");
417         return;
418     }
419     for (i = 0; i < nbds; i++) {
420         bds[i] = boot_device2nibble(boot_device[i]);
421         if (bds[i] == 0) {
422             error_setg(errp, "Invalid boot device for PC: '%c'",
423                        boot_device[i]);
424             return;
425         }
426     }
427     mc146818rtc_set_cmos_data(s, 0x3d, (bds[1] << 4) | bds[0]);
428     mc146818rtc_set_cmos_data(s, 0x38, (bds[2] << 4) | !pcms->fd_bootchk);
429 }
430 
431 static void pc_boot_set(void *opaque, const char *boot_device, Error **errp)
432 {
433     PCMachineState *pcms = opaque;
434     X86MachineState *x86ms = X86_MACHINE(pcms);
435 
436     set_boot_dev(pcms, MC146818_RTC(x86ms->rtc), boot_device, errp);
437 }
438 
439 static void pc_cmos_init_floppy(MC146818RtcState *rtc_state, ISADevice *floppy)
440 {
441     int val, nb, i;
442     FloppyDriveType fd_type[2] = { FLOPPY_DRIVE_TYPE_NONE,
443                                    FLOPPY_DRIVE_TYPE_NONE };
444 
445     /* floppy type */
446     if (floppy) {
447         for (i = 0; i < 2; i++) {
448             fd_type[i] = isa_fdc_get_drive_type(floppy, i);
449         }
450     }
451     val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
452         cmos_get_fd_drive_type(fd_type[1]);
453     mc146818rtc_set_cmos_data(rtc_state, 0x10, val);
454 
455     val = mc146818rtc_get_cmos_data(rtc_state, REG_EQUIPMENT_BYTE);
456     nb = 0;
457     if (fd_type[0] != FLOPPY_DRIVE_TYPE_NONE) {
458         nb++;
459     }
460     if (fd_type[1] != FLOPPY_DRIVE_TYPE_NONE) {
461         nb++;
462     }
463     switch (nb) {
464     case 0:
465         break;
466     case 1:
467         val |= 0x01; /* 1 drive, ready for boot */
468         break;
469     case 2:
470         val |= 0x41; /* 2 drives, ready for boot */
471         break;
472     }
473     mc146818rtc_set_cmos_data(rtc_state, REG_EQUIPMENT_BYTE, val);
474 }
475 
476 typedef struct check_fdc_state {
477     ISADevice *floppy;
478     bool multiple;
479 } CheckFdcState;
480 
481 static int check_fdc(Object *obj, void *opaque)
482 {
483     CheckFdcState *state = opaque;
484     Object *fdc;
485     uint32_t iobase;
486     Error *local_err = NULL;
487 
488     fdc = object_dynamic_cast(obj, TYPE_ISA_FDC);
489     if (!fdc) {
490         return 0;
491     }
492 
493     iobase = object_property_get_uint(obj, "iobase", &local_err);
494     if (local_err || iobase != 0x3f0) {
495         error_free(local_err);
496         return 0;
497     }
498 
499     if (state->floppy) {
500         state->multiple = true;
501     } else {
502         state->floppy = ISA_DEVICE(obj);
503     }
504     return 0;
505 }
506 
507 static const char * const fdc_container_path[] = {
508     "/unattached", "/peripheral", "/peripheral-anon"
509 };
510 
511 /*
512  * Locate the FDC at IO address 0x3f0, in order to configure the CMOS registers
513  * and ACPI objects.
514  */
515 static ISADevice *pc_find_fdc0(void)
516 {
517     int i;
518     Object *container;
519     CheckFdcState state = { 0 };
520 
521     for (i = 0; i < ARRAY_SIZE(fdc_container_path); i++) {
522         container = container_get(qdev_get_machine(), fdc_container_path[i]);
523         object_child_foreach(container, check_fdc, &state);
524     }
525 
526     if (state.multiple) {
527         warn_report("multiple floppy disk controllers with "
528                     "iobase=0x3f0 have been found");
529         error_printf("the one being picked for CMOS setup might not reflect "
530                      "your intent");
531     }
532 
533     return state.floppy;
534 }
535 
536 static void pc_cmos_init_late(PCMachineState *pcms)
537 {
538     X86MachineState *x86ms = X86_MACHINE(pcms);
539     MC146818RtcState *s = MC146818_RTC(x86ms->rtc);
540     int16_t cylinders;
541     int8_t heads, sectors;
542     int val;
543     int i, trans;
544 
545     val = 0;
546     if (pcms->idebus[0] &&
547         ide_get_geometry(pcms->idebus[0], 0,
548                          &cylinders, &heads, &sectors) >= 0) {
549         cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
550         val |= 0xf0;
551     }
552     if (pcms->idebus[0] &&
553         ide_get_geometry(pcms->idebus[0], 1,
554                          &cylinders, &heads, &sectors) >= 0) {
555         cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
556         val |= 0x0f;
557     }
558     mc146818rtc_set_cmos_data(s, 0x12, val);
559 
560     val = 0;
561     for (i = 0; i < 4; i++) {
562         /* NOTE: ide_get_geometry() returns the physical
563            geometry.  It is always such that: 1 <= sects <= 63, 1
564            <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
565            geometry can be different if a translation is done. */
566         BusState *idebus = pcms->idebus[i / 2];
567         if (idebus &&
568             ide_get_geometry(idebus, i % 2,
569                              &cylinders, &heads, &sectors) >= 0) {
570             trans = ide_get_bios_chs_trans(idebus, i % 2) - 1;
571             assert((trans & ~3) == 0);
572             val |= trans << (i * 2);
573         }
574     }
575     mc146818rtc_set_cmos_data(s, 0x39, val);
576 
577     pc_cmos_init_floppy(s, pc_find_fdc0());
578 
579     /* various important CMOS locations needed by PC/Bochs bios */
580 
581     /* memory size */
582     /* base memory (first MiB) */
583     val = MIN(x86ms->below_4g_mem_size / KiB, 640);
584     mc146818rtc_set_cmos_data(s, 0x15, val);
585     mc146818rtc_set_cmos_data(s, 0x16, val >> 8);
586     /* extended memory (next 64MiB) */
587     if (x86ms->below_4g_mem_size > 1 * MiB) {
588         val = (x86ms->below_4g_mem_size - 1 * MiB) / KiB;
589     } else {
590         val = 0;
591     }
592     if (val > 65535)
593         val = 65535;
594     mc146818rtc_set_cmos_data(s, 0x17, val);
595     mc146818rtc_set_cmos_data(s, 0x18, val >> 8);
596     mc146818rtc_set_cmos_data(s, 0x30, val);
597     mc146818rtc_set_cmos_data(s, 0x31, val >> 8);
598     /* memory between 16MiB and 4GiB */
599     if (x86ms->below_4g_mem_size > 16 * MiB) {
600         val = (x86ms->below_4g_mem_size - 16 * MiB) / (64 * KiB);
601     } else {
602         val = 0;
603     }
604     if (val > 65535)
605         val = 65535;
606     mc146818rtc_set_cmos_data(s, 0x34, val);
607     mc146818rtc_set_cmos_data(s, 0x35, val >> 8);
608     /* memory above 4GiB */
609     val = x86ms->above_4g_mem_size / 65536;
610     mc146818rtc_set_cmos_data(s, 0x5b, val);
611     mc146818rtc_set_cmos_data(s, 0x5c, val >> 8);
612     mc146818rtc_set_cmos_data(s, 0x5d, val >> 16);
613 
614     val = 0;
615     val |= 0x02; /* FPU is there */
616     val |= 0x04; /* PS/2 mouse installed */
617     mc146818rtc_set_cmos_data(s, REG_EQUIPMENT_BYTE, val);
618 }
619 
620 static void handle_a20_line_change(void *opaque, int irq, int level)
621 {
622     X86CPU *cpu = opaque;
623 
624     /* XXX: send to all CPUs ? */
625     /* XXX: add logic to handle multiple A20 line sources */
626     x86_cpu_set_a20(cpu, level);
627 }
628 
629 #define NE2000_NB_MAX 6
630 
631 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
632                                               0x280, 0x380 };
633 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
634 
635 static gboolean pc_init_ne2k_isa(ISABus *bus, NICInfo *nd, Error **errp)
636 {
637     static int nb_ne2k = 0;
638 
639     if (nb_ne2k == NE2000_NB_MAX) {
640         error_setg(errp,
641                    "maximum number of ISA NE2000 devices exceeded");
642         return false;
643     }
644     isa_ne2000_init(bus, ne2000_io[nb_ne2k],
645                     ne2000_irq[nb_ne2k], nd);
646     nb_ne2k++;
647     return true;
648 }
649 
650 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
651 {
652     X86CPU *cpu = opaque;
653 
654     if (level) {
655         cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
656     }
657 }
658 
659 static
660 void pc_machine_done(Notifier *notifier, void *data)
661 {
662     PCMachineState *pcms = container_of(notifier,
663                                         PCMachineState, machine_done);
664     X86MachineState *x86ms = X86_MACHINE(pcms);
665 
666     cxl_hook_up_pxb_registers(pcms->pcibus, &pcms->cxl_devices_state,
667                               &error_fatal);
668 
669     if (pcms->cxl_devices_state.is_enabled) {
670         cxl_fmws_link_targets(&pcms->cxl_devices_state, &error_fatal);
671     }
672 
673     /* set the number of CPUs */
674     x86_rtc_set_cpus_count(x86ms->rtc, x86ms->boot_cpus);
675 
676     fw_cfg_add_extra_pci_roots(pcms->pcibus, x86ms->fw_cfg);
677 
678     acpi_setup();
679     if (x86ms->fw_cfg) {
680         fw_cfg_build_smbios(pcms, x86ms->fw_cfg, pcms->smbios_entry_point_type);
681         fw_cfg_build_feature_control(MACHINE(pcms), x86ms->fw_cfg);
682         /* update FW_CFG_NB_CPUS to account for -device added CPUs */
683         fw_cfg_modify_i16(x86ms->fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
684     }
685 
686     pc_cmos_init_late(pcms);
687 }
688 
689 /* setup pci memory address space mapping into system address space */
690 void pc_pci_as_mapping_init(MemoryRegion *system_memory,
691                             MemoryRegion *pci_address_space)
692 {
693     /* Set to lower priority than RAM */
694     memory_region_add_subregion_overlap(system_memory, 0x0,
695                                         pci_address_space, -1);
696 }
697 
698 void xen_load_linux(PCMachineState *pcms)
699 {
700     int i;
701     FWCfgState *fw_cfg;
702     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
703     X86MachineState *x86ms = X86_MACHINE(pcms);
704 
705     assert(MACHINE(pcms)->kernel_filename != NULL);
706 
707     fw_cfg = fw_cfg_init_io_dma(FW_CFG_IO_BASE, FW_CFG_IO_BASE + 4,
708                                 &address_space_memory);
709     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
710     rom_set_fw(fw_cfg);
711 
712     x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
713                    pcmc->pvh_enabled);
714     for (i = 0; i < nb_option_roms; i++) {
715         assert(!strcmp(option_rom[i].name, "linuxboot.bin") ||
716                !strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
717                !strcmp(option_rom[i].name, "pvh.bin") ||
718                !strcmp(option_rom[i].name, "multiboot.bin") ||
719                !strcmp(option_rom[i].name, "multiboot_dma.bin"));
720         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
721     }
722     x86ms->fw_cfg = fw_cfg;
723 }
724 
725 #define PC_ROM_MIN_VGA     0xc0000
726 #define PC_ROM_MIN_OPTION  0xc8000
727 #define PC_ROM_MAX         0xe0000
728 #define PC_ROM_ALIGN       0x800
729 #define PC_ROM_SIZE        (PC_ROM_MAX - PC_ROM_MIN_VGA)
730 
731 static hwaddr pc_above_4g_end(PCMachineState *pcms)
732 {
733     X86MachineState *x86ms = X86_MACHINE(pcms);
734 
735     if (pcms->sgx_epc.size != 0) {
736         return sgx_epc_above_4g_end(&pcms->sgx_epc);
737     }
738 
739     return x86ms->above_4g_mem_start + x86ms->above_4g_mem_size;
740 }
741 
742 static void pc_get_device_memory_range(PCMachineState *pcms,
743                                        hwaddr *base,
744                                        ram_addr_t *device_mem_size)
745 {
746     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
747     MachineState *machine = MACHINE(pcms);
748     ram_addr_t size;
749     hwaddr addr;
750 
751     size = machine->maxram_size - machine->ram_size;
752     addr = ROUND_UP(pc_above_4g_end(pcms), 1 * GiB);
753 
754     if (pcmc->enforce_aligned_dimm) {
755         /* size device region assuming 1G page max alignment per slot */
756         size += (1 * GiB) * machine->ram_slots;
757     }
758 
759     *base = addr;
760     *device_mem_size = size;
761 }
762 
763 static uint64_t pc_get_cxl_range_start(PCMachineState *pcms)
764 {
765     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
766     MachineState *ms = MACHINE(pcms);
767     hwaddr cxl_base;
768     ram_addr_t size;
769 
770     if (pcmc->has_reserved_memory &&
771         (ms->ram_size < ms->maxram_size)) {
772         pc_get_device_memory_range(pcms, &cxl_base, &size);
773         cxl_base += size;
774     } else {
775         cxl_base = pc_above_4g_end(pcms);
776     }
777 
778     return cxl_base;
779 }
780 
781 static uint64_t pc_get_cxl_range_end(PCMachineState *pcms)
782 {
783     uint64_t start = pc_get_cxl_range_start(pcms) + MiB;
784 
785     if (pcms->cxl_devices_state.fixed_windows) {
786         GList *it;
787 
788         start = ROUND_UP(start, 256 * MiB);
789         for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
790             CXLFixedWindow *fw = it->data;
791             start += fw->size;
792         }
793     }
794 
795     return start;
796 }
797 
798 static hwaddr pc_max_used_gpa(PCMachineState *pcms, uint64_t pci_hole64_size)
799 {
800     X86CPU *cpu = X86_CPU(first_cpu);
801     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
802     MachineState *ms = MACHINE(pcms);
803 
804     if (cpu->env.features[FEAT_8000_0001_EDX] & CPUID_EXT2_LM) {
805         /* 64-bit systems */
806         return pc_pci_hole64_start() + pci_hole64_size - 1;
807     }
808 
809     /* 32-bit systems */
810     if (pcmc->broken_32bit_mem_addr_check) {
811         /* old value for compatibility reasons */
812         return ((hwaddr)1 << cpu->phys_bits) - 1;
813     }
814 
815     /*
816      * 32-bit systems don't have hole64 but they might have a region for
817      * memory devices. Even if additional hotplugged memory devices might
818      * not be usable by most guest OSes, we need to still consider them for
819      * calculating the highest possible GPA so that we can properly report
820      * if someone configures them on a CPU that cannot possibly address them.
821      */
822     if (pcmc->has_reserved_memory &&
823         (ms->ram_size < ms->maxram_size)) {
824         hwaddr devmem_start;
825         ram_addr_t devmem_size;
826 
827         pc_get_device_memory_range(pcms, &devmem_start, &devmem_size);
828         devmem_start += devmem_size;
829         return devmem_start - 1;
830     }
831 
832     /* configuration without any memory hotplug */
833     return pc_above_4g_end(pcms) - 1;
834 }
835 
836 /*
837  * AMD systems with an IOMMU have an additional hole close to the
838  * 1Tb, which are special GPAs that cannot be DMA mapped. Depending
839  * on kernel version, VFIO may or may not let you DMA map those ranges.
840  * Starting Linux v5.4 we validate it, and can't create guests on AMD machines
841  * with certain memory sizes. It's also wrong to use those IOVA ranges
842  * in detriment of leading to IOMMU INVALID_DEVICE_REQUEST or worse.
843  * The ranges reserved for Hyper-Transport are:
844  *
845  * FD_0000_0000h - FF_FFFF_FFFFh
846  *
847  * The ranges represent the following:
848  *
849  * Base Address   Top Address  Use
850  *
851  * FD_0000_0000h FD_F7FF_FFFFh Reserved interrupt address space
852  * FD_F800_0000h FD_F8FF_FFFFh Interrupt/EOI IntCtl
853  * FD_F900_0000h FD_F90F_FFFFh Legacy PIC IACK
854  * FD_F910_0000h FD_F91F_FFFFh System Management
855  * FD_F920_0000h FD_FAFF_FFFFh Reserved Page Tables
856  * FD_FB00_0000h FD_FBFF_FFFFh Address Translation
857  * FD_FC00_0000h FD_FDFF_FFFFh I/O Space
858  * FD_FE00_0000h FD_FFFF_FFFFh Configuration
859  * FE_0000_0000h FE_1FFF_FFFFh Extended Configuration/Device Messages
860  * FE_2000_0000h FF_FFFF_FFFFh Reserved
861  *
862  * See AMD IOMMU spec, section 2.1.2 "IOMMU Logical Topology",
863  * Table 3: Special Address Controls (GPA) for more information.
864  */
865 #define AMD_HT_START         0xfd00000000UL
866 #define AMD_HT_END           0xffffffffffUL
867 #define AMD_ABOVE_1TB_START  (AMD_HT_END + 1)
868 #define AMD_HT_SIZE          (AMD_ABOVE_1TB_START - AMD_HT_START)
869 
870 void pc_memory_init(PCMachineState *pcms,
871                     MemoryRegion *system_memory,
872                     MemoryRegion *rom_memory,
873                     uint64_t pci_hole64_size)
874 {
875     int linux_boot, i;
876     MemoryRegion *option_rom_mr;
877     MemoryRegion *ram_below_4g, *ram_above_4g;
878     FWCfgState *fw_cfg;
879     MachineState *machine = MACHINE(pcms);
880     MachineClass *mc = MACHINE_GET_CLASS(machine);
881     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
882     X86MachineState *x86ms = X86_MACHINE(pcms);
883     hwaddr maxphysaddr, maxusedaddr;
884     hwaddr cxl_base, cxl_resv_end = 0;
885     X86CPU *cpu = X86_CPU(first_cpu);
886 
887     assert(machine->ram_size == x86ms->below_4g_mem_size +
888                                 x86ms->above_4g_mem_size);
889 
890     linux_boot = (machine->kernel_filename != NULL);
891 
892     /*
893      * The HyperTransport range close to the 1T boundary is unique to AMD
894      * hosts with IOMMUs enabled. Restrict the ram-above-4g relocation
895      * to above 1T to AMD vCPUs only. @enforce_amd_1tb_hole is only false in
896      * older machine types (<= 7.0) for compatibility purposes.
897      */
898     if (IS_AMD_CPU(&cpu->env) && pcmc->enforce_amd_1tb_hole) {
899         /* Bail out if max possible address does not cross HT range */
900         if (pc_max_used_gpa(pcms, pci_hole64_size) >= AMD_HT_START) {
901             x86ms->above_4g_mem_start = AMD_ABOVE_1TB_START;
902         }
903 
904         /*
905          * Advertise the HT region if address space covers the reserved
906          * region or if we relocate.
907          */
908         if (cpu->phys_bits >= 40) {
909             e820_add_entry(AMD_HT_START, AMD_HT_SIZE, E820_RESERVED);
910         }
911     }
912 
913     /*
914      * phys-bits is required to be appropriately configured
915      * to make sure max used GPA is reachable.
916      */
917     maxusedaddr = pc_max_used_gpa(pcms, pci_hole64_size);
918     maxphysaddr = ((hwaddr)1 << cpu->phys_bits) - 1;
919     if (maxphysaddr < maxusedaddr) {
920         error_report("Address space limit 0x%"PRIx64" < 0x%"PRIx64
921                      " phys-bits too low (%u)",
922                      maxphysaddr, maxusedaddr, cpu->phys_bits);
923         exit(EXIT_FAILURE);
924     }
925 
926     /*
927      * Split single memory region and use aliases to address portions of it,
928      * done for backwards compatibility with older qemus.
929      */
930     ram_below_4g = g_malloc(sizeof(*ram_below_4g));
931     memory_region_init_alias(ram_below_4g, NULL, "ram-below-4g", machine->ram,
932                              0, x86ms->below_4g_mem_size);
933     memory_region_add_subregion(system_memory, 0, ram_below_4g);
934     e820_add_entry(0, x86ms->below_4g_mem_size, E820_RAM);
935     if (x86ms->above_4g_mem_size > 0) {
936         ram_above_4g = g_malloc(sizeof(*ram_above_4g));
937         memory_region_init_alias(ram_above_4g, NULL, "ram-above-4g",
938                                  machine->ram,
939                                  x86ms->below_4g_mem_size,
940                                  x86ms->above_4g_mem_size);
941         memory_region_add_subregion(system_memory, x86ms->above_4g_mem_start,
942                                     ram_above_4g);
943         e820_add_entry(x86ms->above_4g_mem_start, x86ms->above_4g_mem_size,
944                        E820_RAM);
945     }
946 
947     if (pcms->sgx_epc.size != 0) {
948         e820_add_entry(pcms->sgx_epc.base, pcms->sgx_epc.size, E820_RESERVED);
949     }
950 
951     if (!pcmc->has_reserved_memory &&
952         (machine->ram_slots ||
953          (machine->maxram_size > machine->ram_size))) {
954 
955         error_report("\"-memory 'slots|maxmem'\" is not supported by: %s",
956                      mc->name);
957         exit(EXIT_FAILURE);
958     }
959 
960     /* initialize device memory address space */
961     if (pcmc->has_reserved_memory &&
962         (machine->ram_size < machine->maxram_size)) {
963         ram_addr_t device_mem_size;
964         hwaddr device_mem_base;
965 
966         if (machine->ram_slots > ACPI_MAX_RAM_SLOTS) {
967             error_report("unsupported amount of memory slots: %"PRIu64,
968                          machine->ram_slots);
969             exit(EXIT_FAILURE);
970         }
971 
972         if (QEMU_ALIGN_UP(machine->maxram_size,
973                           TARGET_PAGE_SIZE) != machine->maxram_size) {
974             error_report("maximum memory size must by aligned to multiple of "
975                          "%d bytes", TARGET_PAGE_SIZE);
976             exit(EXIT_FAILURE);
977         }
978 
979         pc_get_device_memory_range(pcms, &device_mem_base, &device_mem_size);
980 
981         if (device_mem_base + device_mem_size < device_mem_size) {
982             error_report("unsupported amount of maximum memory: " RAM_ADDR_FMT,
983                          machine->maxram_size);
984             exit(EXIT_FAILURE);
985         }
986         machine_memory_devices_init(machine, device_mem_base, device_mem_size);
987     }
988 
989     if (pcms->cxl_devices_state.is_enabled) {
990         MemoryRegion *mr = &pcms->cxl_devices_state.host_mr;
991         hwaddr cxl_size = MiB;
992 
993         cxl_base = pc_get_cxl_range_start(pcms);
994         memory_region_init(mr, OBJECT(machine), "cxl_host_reg", cxl_size);
995         memory_region_add_subregion(system_memory, cxl_base, mr);
996         cxl_resv_end = cxl_base + cxl_size;
997         if (pcms->cxl_devices_state.fixed_windows) {
998             hwaddr cxl_fmw_base;
999             GList *it;
1000 
1001             cxl_fmw_base = ROUND_UP(cxl_base + cxl_size, 256 * MiB);
1002             for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
1003                 CXLFixedWindow *fw = it->data;
1004 
1005                 fw->base = cxl_fmw_base;
1006                 memory_region_init_io(&fw->mr, OBJECT(machine), &cfmws_ops, fw,
1007                                       "cxl-fixed-memory-region", fw->size);
1008                 memory_region_add_subregion(system_memory, fw->base, &fw->mr);
1009                 cxl_fmw_base += fw->size;
1010                 cxl_resv_end = cxl_fmw_base;
1011             }
1012         }
1013     }
1014 
1015     /* Initialize PC system firmware */
1016     pc_system_firmware_init(pcms, rom_memory);
1017 
1018     option_rom_mr = g_malloc(sizeof(*option_rom_mr));
1019     memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE,
1020                            &error_fatal);
1021     if (pcmc->pci_enabled) {
1022         memory_region_set_readonly(option_rom_mr, true);
1023     }
1024     memory_region_add_subregion_overlap(rom_memory,
1025                                         PC_ROM_MIN_VGA,
1026                                         option_rom_mr,
1027                                         1);
1028 
1029     fw_cfg = fw_cfg_arch_create(machine,
1030                                 x86ms->boot_cpus, x86ms->apic_id_limit);
1031 
1032     rom_set_fw(fw_cfg);
1033 
1034     if (machine->device_memory) {
1035         uint64_t *val = g_malloc(sizeof(*val));
1036         uint64_t res_mem_end = machine->device_memory->base;
1037 
1038         if (!pcmc->broken_reserved_end) {
1039             res_mem_end += memory_region_size(&machine->device_memory->mr);
1040         }
1041 
1042         if (pcms->cxl_devices_state.is_enabled) {
1043             res_mem_end = cxl_resv_end;
1044         }
1045         *val = cpu_to_le64(ROUND_UP(res_mem_end, 1 * GiB));
1046         fw_cfg_add_file(fw_cfg, "etc/reserved-memory-end", val, sizeof(*val));
1047     }
1048 
1049     if (linux_boot) {
1050         x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
1051                        pcmc->pvh_enabled);
1052     }
1053 
1054     for (i = 0; i < nb_option_roms; i++) {
1055         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1056     }
1057     x86ms->fw_cfg = fw_cfg;
1058 
1059     /* Init default IOAPIC address space */
1060     x86ms->ioapic_as = &address_space_memory;
1061 
1062     /* Init ACPI memory hotplug IO base address */
1063     pcms->memhp_io_base = ACPI_MEMORY_HOTPLUG_BASE;
1064 }
1065 
1066 /*
1067  * The 64bit pci hole starts after "above 4G RAM" and
1068  * potentially the space reserved for memory hotplug.
1069  */
1070 uint64_t pc_pci_hole64_start(void)
1071 {
1072     PCMachineState *pcms = PC_MACHINE(qdev_get_machine());
1073     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1074     MachineState *ms = MACHINE(pcms);
1075     uint64_t hole64_start = 0;
1076     ram_addr_t size = 0;
1077 
1078     if (pcms->cxl_devices_state.is_enabled) {
1079         hole64_start = pc_get_cxl_range_end(pcms);
1080     } else if (pcmc->has_reserved_memory && (ms->ram_size < ms->maxram_size)) {
1081         pc_get_device_memory_range(pcms, &hole64_start, &size);
1082         if (!pcmc->broken_reserved_end) {
1083             hole64_start += size;
1084         }
1085     } else {
1086         hole64_start = pc_above_4g_end(pcms);
1087     }
1088 
1089     return ROUND_UP(hole64_start, 1 * GiB);
1090 }
1091 
1092 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1093 {
1094     DeviceState *dev = NULL;
1095 
1096     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_VGA);
1097     if (pci_bus) {
1098         PCIDevice *pcidev = pci_vga_init(pci_bus);
1099         dev = pcidev ? &pcidev->qdev : NULL;
1100     } else if (isa_bus) {
1101         ISADevice *isadev = isa_vga_init(isa_bus);
1102         dev = isadev ? DEVICE(isadev) : NULL;
1103     }
1104     rom_reset_order_override();
1105     return dev;
1106 }
1107 
1108 static const MemoryRegionOps ioport80_io_ops = {
1109     .write = ioport80_write,
1110     .read = ioport80_read,
1111     .endianness = DEVICE_NATIVE_ENDIAN,
1112     .impl = {
1113         .min_access_size = 1,
1114         .max_access_size = 1,
1115     },
1116 };
1117 
1118 static const MemoryRegionOps ioportF0_io_ops = {
1119     .write = ioportF0_write,
1120     .read = ioportF0_read,
1121     .endianness = DEVICE_NATIVE_ENDIAN,
1122     .impl = {
1123         .min_access_size = 1,
1124         .max_access_size = 1,
1125     },
1126 };
1127 
1128 static void pc_superio_init(ISABus *isa_bus, bool create_fdctrl,
1129                             bool create_i8042, bool no_vmport)
1130 {
1131     int i;
1132     DriveInfo *fd[MAX_FD];
1133     qemu_irq *a20_line;
1134     ISADevice *fdc, *i8042, *port92, *vmmouse;
1135 
1136     serial_hds_isa_init(isa_bus, 0, MAX_ISA_SERIAL_PORTS);
1137     parallel_hds_isa_init(isa_bus, MAX_PARALLEL_PORTS);
1138 
1139     for (i = 0; i < MAX_FD; i++) {
1140         fd[i] = drive_get(IF_FLOPPY, 0, i);
1141         create_fdctrl |= !!fd[i];
1142     }
1143     if (create_fdctrl) {
1144         fdc = isa_new(TYPE_ISA_FDC);
1145         if (fdc) {
1146             isa_realize_and_unref(fdc, isa_bus, &error_fatal);
1147             isa_fdc_init_drives(fdc, fd);
1148         }
1149     }
1150 
1151     if (!create_i8042) {
1152         return;
1153     }
1154 
1155     i8042 = isa_create_simple(isa_bus, TYPE_I8042);
1156     if (!no_vmport) {
1157         isa_create_simple(isa_bus, TYPE_VMPORT);
1158         vmmouse = isa_try_new("vmmouse");
1159     } else {
1160         vmmouse = NULL;
1161     }
1162     if (vmmouse) {
1163         object_property_set_link(OBJECT(vmmouse), TYPE_I8042, OBJECT(i8042),
1164                                  &error_abort);
1165         isa_realize_and_unref(vmmouse, isa_bus, &error_fatal);
1166     }
1167     port92 = isa_create_simple(isa_bus, TYPE_PORT92);
1168 
1169     a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1170     qdev_connect_gpio_out_named(DEVICE(i8042),
1171                                 I8042_A20_LINE, 0, a20_line[0]);
1172     qdev_connect_gpio_out_named(DEVICE(port92),
1173                                 PORT92_A20_LINE, 0, a20_line[1]);
1174     g_free(a20_line);
1175 }
1176 
1177 void pc_basic_device_init(struct PCMachineState *pcms,
1178                           ISABus *isa_bus, qemu_irq *gsi,
1179                           ISADevice *rtc_state,
1180                           bool create_fdctrl,
1181                           uint32_t hpet_irqs)
1182 {
1183     int i;
1184     DeviceState *hpet = NULL;
1185     int pit_isa_irq = 0;
1186     qemu_irq pit_alt_irq = NULL;
1187     ISADevice *pit = NULL;
1188     MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1189     MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
1190     X86MachineState *x86ms = X86_MACHINE(pcms);
1191 
1192     memory_region_init_io(ioport80_io, NULL, &ioport80_io_ops, NULL, "ioport80", 1);
1193     memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
1194 
1195     memory_region_init_io(ioportF0_io, NULL, &ioportF0_io_ops, NULL, "ioportF0", 1);
1196     memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
1197 
1198     /*
1199      * Check if an HPET shall be created.
1200      */
1201     if (pcms->hpet_enabled) {
1202         qemu_irq rtc_irq;
1203 
1204         hpet = qdev_try_new(TYPE_HPET);
1205         if (!hpet) {
1206             error_report("couldn't create HPET device");
1207             exit(1);
1208         }
1209         /*
1210          * For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-*,
1211          * use IRQ16~23, IRQ8 and IRQ2.  If the user has already set
1212          * the property, use whatever mask they specified.
1213          */
1214         uint8_t compat = object_property_get_uint(OBJECT(hpet),
1215                 HPET_INTCAP, NULL);
1216         if (!compat) {
1217             qdev_prop_set_uint32(hpet, HPET_INTCAP, hpet_irqs);
1218         }
1219         sysbus_realize_and_unref(SYS_BUS_DEVICE(hpet), &error_fatal);
1220         sysbus_mmio_map(SYS_BUS_DEVICE(hpet), 0, HPET_BASE);
1221 
1222         for (i = 0; i < IOAPIC_NUM_PINS; i++) {
1223             sysbus_connect_irq(SYS_BUS_DEVICE(hpet), i, gsi[i]);
1224         }
1225         pit_isa_irq = -1;
1226         pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1227         rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1228 
1229         /* overwrite connection created by south bridge */
1230         qdev_connect_gpio_out(DEVICE(rtc_state), 0, rtc_irq);
1231     }
1232 
1233     object_property_add_alias(OBJECT(pcms), "rtc-time", OBJECT(rtc_state),
1234                               "date");
1235 
1236 #ifdef CONFIG_XEN_EMU
1237     if (xen_mode == XEN_EMULATE) {
1238         xen_overlay_create();
1239         xen_evtchn_create(IOAPIC_NUM_PINS, gsi);
1240         xen_gnttab_create();
1241         xen_xenstore_create();
1242         if (pcms->pcibus) {
1243             pci_create_simple(pcms->pcibus, -1, "xen-platform");
1244         }
1245         xen_bus_init();
1246         xen_be_init();
1247     }
1248 #endif
1249 
1250     qemu_register_boot_set(pc_boot_set, pcms);
1251     set_boot_dev(pcms, MC146818_RTC(rtc_state),
1252                  MACHINE(pcms)->boot_config.order, &error_fatal);
1253 
1254     if (!xen_enabled() &&
1255         (x86ms->pit == ON_OFF_AUTO_AUTO || x86ms->pit == ON_OFF_AUTO_ON)) {
1256         if (kvm_pit_in_kernel()) {
1257             pit = kvm_pit_init(isa_bus, 0x40);
1258         } else {
1259             pit = i8254_pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1260         }
1261         if (hpet) {
1262             /* connect PIT to output control line of the HPET */
1263             qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(DEVICE(pit), 0));
1264         }
1265         object_property_set_link(OBJECT(pcms->pcspk), "pit",
1266                                  OBJECT(pit), &error_fatal);
1267         isa_realize_and_unref(pcms->pcspk, isa_bus, &error_fatal);
1268     }
1269 
1270     /* Super I/O */
1271     pc_superio_init(isa_bus, create_fdctrl, pcms->i8042_enabled,
1272                     pcms->vmport != ON_OFF_AUTO_ON);
1273 }
1274 
1275 void pc_nic_init(PCMachineClass *pcmc, ISABus *isa_bus, PCIBus *pci_bus)
1276 {
1277     MachineClass *mc = MACHINE_CLASS(pcmc);
1278     bool default_is_ne2k = g_str_equal(mc->default_nic, TYPE_ISA_NE2000);
1279     NICInfo *nd;
1280 
1281     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_NIC);
1282 
1283     while ((nd = qemu_find_nic_info(TYPE_ISA_NE2000, default_is_ne2k, NULL))) {
1284         pc_init_ne2k_isa(isa_bus, nd, &error_fatal);
1285     }
1286 
1287     /* Anything remaining should be a PCI NIC */
1288     pci_init_nic_devices(pci_bus, mc->default_nic);
1289 
1290     rom_reset_order_override();
1291 }
1292 
1293 void pc_i8259_create(ISABus *isa_bus, qemu_irq *i8259_irqs)
1294 {
1295     qemu_irq *i8259;
1296 
1297     if (kvm_pic_in_kernel()) {
1298         i8259 = kvm_i8259_init(isa_bus);
1299     } else if (xen_enabled()) {
1300         i8259 = xen_interrupt_controller_init();
1301     } else {
1302         i8259 = i8259_init(isa_bus, x86_allocate_cpu_irq());
1303     }
1304 
1305     for (size_t i = 0; i < ISA_NUM_IRQS; i++) {
1306         i8259_irqs[i] = i8259[i];
1307     }
1308 
1309     g_free(i8259);
1310 }
1311 
1312 static void pc_memory_pre_plug(HotplugHandler *hotplug_dev, DeviceState *dev,
1313                                Error **errp)
1314 {
1315     const PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1316     const X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1317     const PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1318     const MachineState *ms = MACHINE(hotplug_dev);
1319     const bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1320     const uint64_t legacy_align = TARGET_PAGE_SIZE;
1321     Error *local_err = NULL;
1322 
1323     /*
1324      * When "acpi=off" is used with the Q35 machine type, no ACPI is built,
1325      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1326      * addition to cover this case.
1327      */
1328     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1329         error_setg(errp,
1330                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1331         return;
1332     }
1333 
1334     if (is_nvdimm && !ms->nvdimms_state->is_enabled) {
1335         error_setg(errp, "nvdimm is not enabled: missing 'nvdimm' in '-M'");
1336         return;
1337     }
1338 
1339     hotplug_handler_pre_plug(x86ms->acpi_dev, dev, &local_err);
1340     if (local_err) {
1341         error_propagate(errp, local_err);
1342         return;
1343     }
1344 
1345     pc_dimm_pre_plug(PC_DIMM(dev), MACHINE(hotplug_dev),
1346                      pcmc->enforce_aligned_dimm ? NULL : &legacy_align, errp);
1347 }
1348 
1349 static void pc_memory_plug(HotplugHandler *hotplug_dev,
1350                            DeviceState *dev, Error **errp)
1351 {
1352     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1353     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1354     MachineState *ms = MACHINE(hotplug_dev);
1355     bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1356 
1357     pc_dimm_plug(PC_DIMM(dev), MACHINE(pcms));
1358 
1359     if (is_nvdimm) {
1360         nvdimm_plug(ms->nvdimms_state);
1361     }
1362 
1363     hotplug_handler_plug(x86ms->acpi_dev, dev, &error_abort);
1364 }
1365 
1366 static void pc_memory_unplug_request(HotplugHandler *hotplug_dev,
1367                                      DeviceState *dev, Error **errp)
1368 {
1369     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1370 
1371     /*
1372      * When "acpi=off" is used with the Q35 machine type, no ACPI is built,
1373      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1374      * addition to cover this case.
1375      */
1376     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1377         error_setg(errp,
1378                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1379         return;
1380     }
1381 
1382     if (object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM)) {
1383         error_setg(errp, "nvdimm device hot unplug is not supported yet.");
1384         return;
1385     }
1386 
1387     hotplug_handler_unplug_request(x86ms->acpi_dev, dev,
1388                                    errp);
1389 }
1390 
1391 static void pc_memory_unplug(HotplugHandler *hotplug_dev,
1392                              DeviceState *dev, Error **errp)
1393 {
1394     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1395     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1396     Error *local_err = NULL;
1397 
1398     hotplug_handler_unplug(x86ms->acpi_dev, dev, &local_err);
1399     if (local_err) {
1400         goto out;
1401     }
1402 
1403     pc_dimm_unplug(PC_DIMM(dev), MACHINE(pcms));
1404     qdev_unrealize(dev);
1405  out:
1406     error_propagate(errp, local_err);
1407 }
1408 
1409 static void pc_hv_balloon_pre_plug(HotplugHandler *hotplug_dev,
1410                                    DeviceState *dev, Error **errp)
1411 {
1412     /* The vmbus handler has no hotplug handler; we should never end up here. */
1413     g_assert(!dev->hotplugged);
1414     memory_device_pre_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev), NULL,
1415                            errp);
1416 }
1417 
1418 static void pc_hv_balloon_plug(HotplugHandler *hotplug_dev,
1419                                DeviceState *dev, Error **errp)
1420 {
1421     memory_device_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev));
1422 }
1423 
1424 static void pc_machine_device_pre_plug_cb(HotplugHandler *hotplug_dev,
1425                                           DeviceState *dev, Error **errp)
1426 {
1427     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1428         pc_memory_pre_plug(hotplug_dev, dev, errp);
1429     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1430         x86_cpu_pre_plug(hotplug_dev, dev, errp);
1431     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1432         virtio_md_pci_pre_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1433     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1434         /* Declare the APIC range as the reserved MSI region */
1435         char *resv_prop_str = g_strdup_printf("0xfee00000:0xfeefffff:%d",
1436                                               VIRTIO_IOMMU_RESV_MEM_T_MSI);
1437         QList *reserved_regions = qlist_new();
1438 
1439         qlist_append_str(reserved_regions, resv_prop_str);
1440         qdev_prop_set_array(dev, "reserved-regions", reserved_regions);
1441 
1442         g_free(resv_prop_str);
1443     }
1444 
1445     if (object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE) ||
1446         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1447         PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1448 
1449         if (pcms->iommu) {
1450             error_setg(errp, "QEMU does not support multiple vIOMMUs "
1451                        "for x86 yet.");
1452             return;
1453         }
1454         pcms->iommu = dev;
1455     } else if (object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON)) {
1456         pc_hv_balloon_pre_plug(hotplug_dev, dev, errp);
1457     }
1458 }
1459 
1460 static void pc_machine_device_plug_cb(HotplugHandler *hotplug_dev,
1461                                       DeviceState *dev, Error **errp)
1462 {
1463     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1464         pc_memory_plug(hotplug_dev, dev, errp);
1465     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1466         x86_cpu_plug(hotplug_dev, dev, errp);
1467     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1468         virtio_md_pci_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1469     } else if (object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON)) {
1470         pc_hv_balloon_plug(hotplug_dev, dev, errp);
1471     }
1472 }
1473 
1474 static void pc_machine_device_unplug_request_cb(HotplugHandler *hotplug_dev,
1475                                                 DeviceState *dev, Error **errp)
1476 {
1477     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1478         pc_memory_unplug_request(hotplug_dev, dev, errp);
1479     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1480         x86_cpu_unplug_request_cb(hotplug_dev, dev, errp);
1481     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1482         virtio_md_pci_unplug_request(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev),
1483                                      errp);
1484     } else {
1485         error_setg(errp, "acpi: device unplug request for not supported device"
1486                    " type: %s", object_get_typename(OBJECT(dev)));
1487     }
1488 }
1489 
1490 static void pc_machine_device_unplug_cb(HotplugHandler *hotplug_dev,
1491                                         DeviceState *dev, Error **errp)
1492 {
1493     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1494         pc_memory_unplug(hotplug_dev, dev, errp);
1495     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1496         x86_cpu_unplug_cb(hotplug_dev, dev, errp);
1497     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1498         virtio_md_pci_unplug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1499     } else {
1500         error_setg(errp, "acpi: device unplug for not supported device"
1501                    " type: %s", object_get_typename(OBJECT(dev)));
1502     }
1503 }
1504 
1505 static HotplugHandler *pc_get_hotplug_handler(MachineState *machine,
1506                                              DeviceState *dev)
1507 {
1508     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM) ||
1509         object_dynamic_cast(OBJECT(dev), TYPE_CPU) ||
1510         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI) ||
1511         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI) ||
1512         object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON) ||
1513         object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE)) {
1514         return HOTPLUG_HANDLER(machine);
1515     }
1516 
1517     return NULL;
1518 }
1519 
1520 static void pc_machine_get_vmport(Object *obj, Visitor *v, const char *name,
1521                                   void *opaque, Error **errp)
1522 {
1523     PCMachineState *pcms = PC_MACHINE(obj);
1524     OnOffAuto vmport = pcms->vmport;
1525 
1526     visit_type_OnOffAuto(v, name, &vmport, errp);
1527 }
1528 
1529 static void pc_machine_set_vmport(Object *obj, Visitor *v, const char *name,
1530                                   void *opaque, Error **errp)
1531 {
1532     PCMachineState *pcms = PC_MACHINE(obj);
1533 
1534     visit_type_OnOffAuto(v, name, &pcms->vmport, errp);
1535 }
1536 
1537 static bool pc_machine_get_fd_bootchk(Object *obj, Error **errp)
1538 {
1539     PCMachineState *pcms = PC_MACHINE(obj);
1540 
1541     return pcms->fd_bootchk;
1542 }
1543 
1544 static void pc_machine_set_fd_bootchk(Object *obj, bool value, Error **errp)
1545 {
1546     PCMachineState *pcms = PC_MACHINE(obj);
1547 
1548     pcms->fd_bootchk = value;
1549 }
1550 
1551 static bool pc_machine_get_smbus(Object *obj, Error **errp)
1552 {
1553     PCMachineState *pcms = PC_MACHINE(obj);
1554 
1555     return pcms->smbus_enabled;
1556 }
1557 
1558 static void pc_machine_set_smbus(Object *obj, bool value, Error **errp)
1559 {
1560     PCMachineState *pcms = PC_MACHINE(obj);
1561 
1562     pcms->smbus_enabled = value;
1563 }
1564 
1565 static bool pc_machine_get_sata(Object *obj, Error **errp)
1566 {
1567     PCMachineState *pcms = PC_MACHINE(obj);
1568 
1569     return pcms->sata_enabled;
1570 }
1571 
1572 static void pc_machine_set_sata(Object *obj, bool value, Error **errp)
1573 {
1574     PCMachineState *pcms = PC_MACHINE(obj);
1575 
1576     pcms->sata_enabled = value;
1577 }
1578 
1579 static bool pc_machine_get_hpet(Object *obj, Error **errp)
1580 {
1581     PCMachineState *pcms = PC_MACHINE(obj);
1582 
1583     return pcms->hpet_enabled;
1584 }
1585 
1586 static void pc_machine_set_hpet(Object *obj, bool value, Error **errp)
1587 {
1588     PCMachineState *pcms = PC_MACHINE(obj);
1589 
1590     pcms->hpet_enabled = value;
1591 }
1592 
1593 static bool pc_machine_get_i8042(Object *obj, Error **errp)
1594 {
1595     PCMachineState *pcms = PC_MACHINE(obj);
1596 
1597     return pcms->i8042_enabled;
1598 }
1599 
1600 static void pc_machine_set_i8042(Object *obj, bool value, Error **errp)
1601 {
1602     PCMachineState *pcms = PC_MACHINE(obj);
1603 
1604     pcms->i8042_enabled = value;
1605 }
1606 
1607 static bool pc_machine_get_default_bus_bypass_iommu(Object *obj, Error **errp)
1608 {
1609     PCMachineState *pcms = PC_MACHINE(obj);
1610 
1611     return pcms->default_bus_bypass_iommu;
1612 }
1613 
1614 static void pc_machine_set_default_bus_bypass_iommu(Object *obj, bool value,
1615                                                     Error **errp)
1616 {
1617     PCMachineState *pcms = PC_MACHINE(obj);
1618 
1619     pcms->default_bus_bypass_iommu = value;
1620 }
1621 
1622 static void pc_machine_get_smbios_ep(Object *obj, Visitor *v, const char *name,
1623                                      void *opaque, Error **errp)
1624 {
1625     PCMachineState *pcms = PC_MACHINE(obj);
1626     SmbiosEntryPointType smbios_entry_point_type = pcms->smbios_entry_point_type;
1627 
1628     visit_type_SmbiosEntryPointType(v, name, &smbios_entry_point_type, errp);
1629 }
1630 
1631 static void pc_machine_set_smbios_ep(Object *obj, Visitor *v, const char *name,
1632                                      void *opaque, Error **errp)
1633 {
1634     PCMachineState *pcms = PC_MACHINE(obj);
1635 
1636     visit_type_SmbiosEntryPointType(v, name, &pcms->smbios_entry_point_type, errp);
1637 }
1638 
1639 static void pc_machine_get_max_ram_below_4g(Object *obj, Visitor *v,
1640                                             const char *name, void *opaque,
1641                                             Error **errp)
1642 {
1643     PCMachineState *pcms = PC_MACHINE(obj);
1644     uint64_t value = pcms->max_ram_below_4g;
1645 
1646     visit_type_size(v, name, &value, errp);
1647 }
1648 
1649 static void pc_machine_set_max_ram_below_4g(Object *obj, Visitor *v,
1650                                             const char *name, void *opaque,
1651                                             Error **errp)
1652 {
1653     PCMachineState *pcms = PC_MACHINE(obj);
1654     uint64_t value;
1655 
1656     if (!visit_type_size(v, name, &value, errp)) {
1657         return;
1658     }
1659     if (value > 4 * GiB) {
1660         error_setg(errp,
1661                    "Machine option 'max-ram-below-4g=%"PRIu64
1662                    "' expects size less than or equal to 4G", value);
1663         return;
1664     }
1665 
1666     if (value < 1 * MiB) {
1667         warn_report("Only %" PRIu64 " bytes of RAM below the 4GiB boundary,"
1668                     "BIOS may not work with less than 1MiB", value);
1669     }
1670 
1671     pcms->max_ram_below_4g = value;
1672 }
1673 
1674 static void pc_machine_get_max_fw_size(Object *obj, Visitor *v,
1675                                        const char *name, void *opaque,
1676                                        Error **errp)
1677 {
1678     PCMachineState *pcms = PC_MACHINE(obj);
1679     uint64_t value = pcms->max_fw_size;
1680 
1681     visit_type_size(v, name, &value, errp);
1682 }
1683 
1684 static void pc_machine_set_max_fw_size(Object *obj, Visitor *v,
1685                                        const char *name, void *opaque,
1686                                        Error **errp)
1687 {
1688     PCMachineState *pcms = PC_MACHINE(obj);
1689     uint64_t value;
1690 
1691     if (!visit_type_size(v, name, &value, errp)) {
1692         return;
1693     }
1694 
1695     /*
1696      * We don't have a theoretically justifiable exact lower bound on the base
1697      * address of any flash mapping. In practice, the IO-APIC MMIO range is
1698      * [0xFEE00000..0xFEE01000] -- see IO_APIC_DEFAULT_ADDRESS --, leaving free
1699      * only 18MiB-4KiB below 4GiB. For now, restrict the cumulative mapping to
1700      * 16MiB in size.
1701      */
1702     if (value > 16 * MiB) {
1703         error_setg(errp,
1704                    "User specified max allowed firmware size %" PRIu64 " is "
1705                    "greater than 16MiB. If combined firmware size exceeds "
1706                    "16MiB the system may not boot, or experience intermittent"
1707                    "stability issues.",
1708                    value);
1709         return;
1710     }
1711 
1712     pcms->max_fw_size = value;
1713 }
1714 
1715 
1716 static void pc_machine_initfn(Object *obj)
1717 {
1718     PCMachineState *pcms = PC_MACHINE(obj);
1719     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1720 
1721 #ifdef CONFIG_VMPORT
1722     pcms->vmport = ON_OFF_AUTO_AUTO;
1723 #else
1724     pcms->vmport = ON_OFF_AUTO_OFF;
1725 #endif /* CONFIG_VMPORT */
1726     pcms->max_ram_below_4g = 0; /* use default */
1727     pcms->smbios_entry_point_type = pcmc->default_smbios_ep_type;
1728     pcms->south_bridge = pcmc->default_south_bridge;
1729 
1730     /* acpi build is enabled by default if machine supports it */
1731     pcms->acpi_build_enabled = pcmc->has_acpi_build;
1732     pcms->smbus_enabled = true;
1733     pcms->sata_enabled = true;
1734     pcms->i8042_enabled = true;
1735     pcms->max_fw_size = 8 * MiB;
1736 #ifdef CONFIG_HPET
1737     pcms->hpet_enabled = true;
1738 #endif
1739     pcms->fd_bootchk = true;
1740     pcms->default_bus_bypass_iommu = false;
1741 
1742     pc_system_flash_create(pcms);
1743     pcms->pcspk = isa_new(TYPE_PC_SPEAKER);
1744     object_property_add_alias(OBJECT(pcms), "pcspk-audiodev",
1745                               OBJECT(pcms->pcspk), "audiodev");
1746     if (pcmc->pci_enabled) {
1747         cxl_machine_init(obj, &pcms->cxl_devices_state);
1748     }
1749 
1750     pcms->machine_done.notify = pc_machine_done;
1751     qemu_add_machine_init_done_notifier(&pcms->machine_done);
1752 }
1753 
1754 static void pc_machine_reset(MachineState *machine, ShutdownCause reason)
1755 {
1756     CPUState *cs;
1757     X86CPU *cpu;
1758 
1759     qemu_devices_reset(reason);
1760 
1761     /* Reset APIC after devices have been reset to cancel
1762      * any changes that qemu_devices_reset() might have done.
1763      */
1764     CPU_FOREACH(cs) {
1765         cpu = X86_CPU(cs);
1766 
1767         x86_cpu_after_reset(cpu);
1768     }
1769 }
1770 
1771 static void pc_machine_wakeup(MachineState *machine)
1772 {
1773     cpu_synchronize_all_states();
1774     pc_machine_reset(machine, SHUTDOWN_CAUSE_NONE);
1775     cpu_synchronize_all_post_reset();
1776 }
1777 
1778 static bool pc_hotplug_allowed(MachineState *ms, DeviceState *dev, Error **errp)
1779 {
1780     X86IOMMUState *iommu = x86_iommu_get_default();
1781     IntelIOMMUState *intel_iommu;
1782 
1783     if (iommu &&
1784         object_dynamic_cast((Object *)iommu, TYPE_INTEL_IOMMU_DEVICE) &&
1785         object_dynamic_cast((Object *)dev, "vfio-pci")) {
1786         intel_iommu = INTEL_IOMMU_DEVICE(iommu);
1787         if (!intel_iommu->caching_mode) {
1788             error_setg(errp, "Device assignment is not allowed without "
1789                        "enabling caching-mode=on for Intel IOMMU.");
1790             return false;
1791         }
1792     }
1793 
1794     return true;
1795 }
1796 
1797 static void pc_machine_class_init(ObjectClass *oc, void *data)
1798 {
1799     MachineClass *mc = MACHINE_CLASS(oc);
1800     X86MachineClass *x86mc = X86_MACHINE_CLASS(oc);
1801     PCMachineClass *pcmc = PC_MACHINE_CLASS(oc);
1802     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
1803 
1804     pcmc->pci_enabled = true;
1805     pcmc->has_acpi_build = true;
1806     pcmc->rsdp_in_ram = true;
1807     pcmc->smbios_defaults = true;
1808     pcmc->smbios_uuid_encoded = true;
1809     pcmc->gigabyte_align = true;
1810     pcmc->has_reserved_memory = true;
1811     pcmc->enforce_aligned_dimm = true;
1812     pcmc->enforce_amd_1tb_hole = true;
1813     /* BIOS ACPI tables: 128K. Other BIOS datastructures: less than 4K reported
1814      * to be used at the moment, 32K should be enough for a while.  */
1815     pcmc->acpi_data_size = 0x20000 + 0x8000;
1816     pcmc->pvh_enabled = true;
1817     pcmc->kvmclock_create_always = true;
1818     pcmc->resizable_acpi_blob = true;
1819     x86mc->apic_xrupt_override = true;
1820     assert(!mc->get_hotplug_handler);
1821     mc->get_hotplug_handler = pc_get_hotplug_handler;
1822     mc->hotplug_allowed = pc_hotplug_allowed;
1823     mc->cpu_index_to_instance_props = x86_cpu_index_to_props;
1824     mc->get_default_cpu_node_id = x86_get_default_cpu_node_id;
1825     mc->possible_cpu_arch_ids = x86_possible_cpu_arch_ids;
1826     mc->auto_enable_numa_with_memhp = true;
1827     mc->auto_enable_numa_with_memdev = true;
1828     mc->has_hotpluggable_cpus = true;
1829     mc->default_boot_order = "cad";
1830     mc->block_default_type = IF_IDE;
1831     mc->max_cpus = 255;
1832     mc->reset = pc_machine_reset;
1833     mc->wakeup = pc_machine_wakeup;
1834     hc->pre_plug = pc_machine_device_pre_plug_cb;
1835     hc->plug = pc_machine_device_plug_cb;
1836     hc->unplug_request = pc_machine_device_unplug_request_cb;
1837     hc->unplug = pc_machine_device_unplug_cb;
1838     mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE;
1839     mc->nvdimm_supported = true;
1840     mc->smp_props.dies_supported = true;
1841     mc->default_ram_id = "pc.ram";
1842     pcmc->default_smbios_ep_type = SMBIOS_ENTRY_POINT_TYPE_AUTO;
1843 
1844     object_class_property_add(oc, PC_MACHINE_MAX_RAM_BELOW_4G, "size",
1845         pc_machine_get_max_ram_below_4g, pc_machine_set_max_ram_below_4g,
1846         NULL, NULL);
1847     object_class_property_set_description(oc, PC_MACHINE_MAX_RAM_BELOW_4G,
1848         "Maximum ram below the 4G boundary (32bit boundary)");
1849 
1850     object_class_property_add(oc, PC_MACHINE_VMPORT, "OnOffAuto",
1851         pc_machine_get_vmport, pc_machine_set_vmport,
1852         NULL, NULL);
1853     object_class_property_set_description(oc, PC_MACHINE_VMPORT,
1854         "Enable vmport (pc & q35)");
1855 
1856     object_class_property_add_bool(oc, PC_MACHINE_SMBUS,
1857         pc_machine_get_smbus, pc_machine_set_smbus);
1858     object_class_property_set_description(oc, PC_MACHINE_SMBUS,
1859         "Enable/disable system management bus");
1860 
1861     object_class_property_add_bool(oc, PC_MACHINE_SATA,
1862         pc_machine_get_sata, pc_machine_set_sata);
1863     object_class_property_set_description(oc, PC_MACHINE_SATA,
1864         "Enable/disable Serial ATA bus");
1865 
1866     object_class_property_add_bool(oc, "hpet",
1867         pc_machine_get_hpet, pc_machine_set_hpet);
1868     object_class_property_set_description(oc, "hpet",
1869         "Enable/disable high precision event timer emulation");
1870 
1871     object_class_property_add_bool(oc, PC_MACHINE_I8042,
1872         pc_machine_get_i8042, pc_machine_set_i8042);
1873 
1874     object_class_property_add_bool(oc, "default-bus-bypass-iommu",
1875         pc_machine_get_default_bus_bypass_iommu,
1876         pc_machine_set_default_bus_bypass_iommu);
1877 
1878     object_class_property_add(oc, PC_MACHINE_MAX_FW_SIZE, "size",
1879         pc_machine_get_max_fw_size, pc_machine_set_max_fw_size,
1880         NULL, NULL);
1881     object_class_property_set_description(oc, PC_MACHINE_MAX_FW_SIZE,
1882         "Maximum combined firmware size");
1883 
1884     object_class_property_add(oc, PC_MACHINE_SMBIOS_EP, "str",
1885         pc_machine_get_smbios_ep, pc_machine_set_smbios_ep,
1886         NULL, NULL);
1887     object_class_property_set_description(oc, PC_MACHINE_SMBIOS_EP,
1888         "SMBIOS Entry Point type [32, 64]");
1889 
1890     object_class_property_add_bool(oc, "fd-bootchk",
1891         pc_machine_get_fd_bootchk,
1892         pc_machine_set_fd_bootchk);
1893 }
1894 
1895 static const TypeInfo pc_machine_info = {
1896     .name = TYPE_PC_MACHINE,
1897     .parent = TYPE_X86_MACHINE,
1898     .abstract = true,
1899     .instance_size = sizeof(PCMachineState),
1900     .instance_init = pc_machine_initfn,
1901     .class_size = sizeof(PCMachineClass),
1902     .class_init = pc_machine_class_init,
1903     .interfaces = (InterfaceInfo[]) {
1904          { TYPE_HOTPLUG_HANDLER },
1905          { }
1906     },
1907 };
1908 
1909 static void pc_machine_register_types(void)
1910 {
1911     type_register_static(&pc_machine_info);
1912 }
1913 
1914 type_init(pc_machine_register_types)
1915