xref: /openbmc/qemu/hw/i386/pc.c (revision c1774bdb)
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/i386/fw_cfg.h"
31 #include "hw/i386/vmport.h"
32 #include "sysemu/cpus.h"
33 #include "hw/ide/internal.h"
34 #include "hw/timer/hpet.h"
35 #include "hw/loader.h"
36 #include "hw/rtc/mc146818rtc.h"
37 #include "hw/intc/i8259.h"
38 #include "hw/timer/i8254.h"
39 #include "hw/input/i8042.h"
40 #include "hw/audio/pcspk.h"
41 #include "sysemu/sysemu.h"
42 #include "sysemu/xen.h"
43 #include "sysemu/reset.h"
44 #include "kvm/kvm_i386.h"
45 #include "hw/xen/xen.h"
46 #include "qemu/error-report.h"
47 #include "hw/acpi/cpu_hotplug.h"
48 #include "acpi-build.h"
49 #include "hw/mem/nvdimm.h"
50 #include "hw/cxl/cxl_host.h"
51 #include "hw/usb.h"
52 #include "hw/i386/intel_iommu.h"
53 #include "hw/net/ne2000-isa.h"
54 #include "hw/virtio/virtio-iommu.h"
55 #include "hw/virtio/virtio-md-pci.h"
56 #include "hw/i386/kvm/xen_overlay.h"
57 #include "hw/i386/kvm/xen_evtchn.h"
58 #include "hw/i386/kvm/xen_gnttab.h"
59 #include "hw/i386/kvm/xen_xenstore.h"
60 #include "e820_memory_layout.h"
61 #include "trace.h"
62 #include CONFIG_DEVICES
63 
64 #ifdef CONFIG_XEN_EMU
65 #include "hw/xen/xen-legacy-backend.h"
66 #include "hw/xen/xen-bus.h"
67 #endif
68 
69 /*
70  * Helper for setting model-id for CPU models that changed model-id
71  * depending on QEMU versions up to QEMU 2.4.
72  */
73 #define PC_CPU_MODEL_IDS(v) \
74     { "qemu32-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
75     { "qemu64-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
76     { "athlon-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },
77 
78 GlobalProperty pc_compat_8_1[] = {};
79 const size_t pc_compat_8_1_len = G_N_ELEMENTS(pc_compat_8_1);
80 
81 GlobalProperty pc_compat_8_0[] = {
82     { "virtio-mem", "unplugged-inaccessible", "auto" },
83 };
84 const size_t pc_compat_8_0_len = G_N_ELEMENTS(pc_compat_8_0);
85 
86 GlobalProperty pc_compat_7_2[] = {
87     { "ICH9-LPC", "noreboot", "true" },
88 };
89 const size_t pc_compat_7_2_len = G_N_ELEMENTS(pc_compat_7_2);
90 
91 GlobalProperty pc_compat_7_1[] = {};
92 const size_t pc_compat_7_1_len = G_N_ELEMENTS(pc_compat_7_1);
93 
94 GlobalProperty pc_compat_7_0[] = {};
95 const size_t pc_compat_7_0_len = G_N_ELEMENTS(pc_compat_7_0);
96 
97 GlobalProperty pc_compat_6_2[] = {
98     { "virtio-mem", "unplugged-inaccessible", "off" },
99 };
100 const size_t pc_compat_6_2_len = G_N_ELEMENTS(pc_compat_6_2);
101 
102 GlobalProperty pc_compat_6_1[] = {
103     { TYPE_X86_CPU, "hv-version-id-build", "0x1bbc" },
104     { TYPE_X86_CPU, "hv-version-id-major", "0x0006" },
105     { TYPE_X86_CPU, "hv-version-id-minor", "0x0001" },
106     { "ICH9-LPC", "x-keep-pci-slot-hpc", "false" },
107 };
108 const size_t pc_compat_6_1_len = G_N_ELEMENTS(pc_compat_6_1);
109 
110 GlobalProperty pc_compat_6_0[] = {
111     { "qemu64" "-" TYPE_X86_CPU, "family", "6" },
112     { "qemu64" "-" TYPE_X86_CPU, "model", "6" },
113     { "qemu64" "-" TYPE_X86_CPU, "stepping", "3" },
114     { TYPE_X86_CPU, "x-vendor-cpuid-only", "off" },
115     { "ICH9-LPC", ACPI_PM_PROP_ACPI_PCIHP_BRIDGE, "off" },
116     { "ICH9-LPC", "x-keep-pci-slot-hpc", "true" },
117 };
118 const size_t pc_compat_6_0_len = G_N_ELEMENTS(pc_compat_6_0);
119 
120 GlobalProperty pc_compat_5_2[] = {
121     { "ICH9-LPC", "x-smi-cpu-hotunplug", "off" },
122 };
123 const size_t pc_compat_5_2_len = G_N_ELEMENTS(pc_compat_5_2);
124 
125 GlobalProperty pc_compat_5_1[] = {
126     { "ICH9-LPC", "x-smi-cpu-hotplug", "off" },
127     { TYPE_X86_CPU, "kvm-msi-ext-dest-id", "off" },
128 };
129 const size_t pc_compat_5_1_len = G_N_ELEMENTS(pc_compat_5_1);
130 
131 GlobalProperty pc_compat_5_0[] = {
132 };
133 const size_t pc_compat_5_0_len = G_N_ELEMENTS(pc_compat_5_0);
134 
135 GlobalProperty pc_compat_4_2[] = {
136     { "mch", "smbase-smram", "off" },
137 };
138 const size_t pc_compat_4_2_len = G_N_ELEMENTS(pc_compat_4_2);
139 
140 GlobalProperty pc_compat_4_1[] = {};
141 const size_t pc_compat_4_1_len = G_N_ELEMENTS(pc_compat_4_1);
142 
143 GlobalProperty pc_compat_4_0[] = {};
144 const size_t pc_compat_4_0_len = G_N_ELEMENTS(pc_compat_4_0);
145 
146 GlobalProperty pc_compat_3_1[] = {
147     { "intel-iommu", "dma-drain", "off" },
148     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "off" },
149     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "off" },
150     { "Opteron_G4" "-" TYPE_X86_CPU, "npt", "off" },
151     { "Opteron_G4" "-" TYPE_X86_CPU, "nrip-save", "off" },
152     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "off" },
153     { "Opteron_G5" "-" TYPE_X86_CPU, "npt", "off" },
154     { "Opteron_G5" "-" TYPE_X86_CPU, "nrip-save", "off" },
155     { "EPYC" "-" TYPE_X86_CPU, "npt", "off" },
156     { "EPYC" "-" TYPE_X86_CPU, "nrip-save", "off" },
157     { "EPYC-IBPB" "-" TYPE_X86_CPU, "npt", "off" },
158     { "EPYC-IBPB" "-" TYPE_X86_CPU, "nrip-save", "off" },
159     { "Skylake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
160     { "Skylake-Client-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
161     { "Skylake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
162     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
163     { "Cascadelake-Server" "-" TYPE_X86_CPU,  "mpx", "on" },
164     { "Icelake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
165     { "Icelake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
166     { "Cascadelake-Server" "-" TYPE_X86_CPU, "stepping", "5" },
167     { TYPE_X86_CPU, "x-intel-pt-auto-level", "off" },
168 };
169 const size_t pc_compat_3_1_len = G_N_ELEMENTS(pc_compat_3_1);
170 
171 GlobalProperty pc_compat_3_0[] = {
172     { TYPE_X86_CPU, "x-hv-synic-kvm-only", "on" },
173     { "Skylake-Server" "-" TYPE_X86_CPU, "pku", "off" },
174     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "pku", "off" },
175 };
176 const size_t pc_compat_3_0_len = G_N_ELEMENTS(pc_compat_3_0);
177 
178 GlobalProperty pc_compat_2_12[] = {
179     { TYPE_X86_CPU, "legacy-cache", "on" },
180     { TYPE_X86_CPU, "topoext", "off" },
181     { "EPYC-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
182     { "EPYC-IBPB-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
183 };
184 const size_t pc_compat_2_12_len = G_N_ELEMENTS(pc_compat_2_12);
185 
186 GlobalProperty pc_compat_2_11[] = {
187     { TYPE_X86_CPU, "x-migrate-smi-count", "off" },
188     { "Skylake-Server" "-" TYPE_X86_CPU, "clflushopt", "off" },
189 };
190 const size_t pc_compat_2_11_len = G_N_ELEMENTS(pc_compat_2_11);
191 
192 GlobalProperty pc_compat_2_10[] = {
193     { TYPE_X86_CPU, "x-hv-max-vps", "0x40" },
194     { "i440FX-pcihost", "x-pci-hole64-fix", "off" },
195     { "q35-pcihost", "x-pci-hole64-fix", "off" },
196 };
197 const size_t pc_compat_2_10_len = G_N_ELEMENTS(pc_compat_2_10);
198 
199 GlobalProperty pc_compat_2_9[] = {
200     { "mch", "extended-tseg-mbytes", "0" },
201 };
202 const size_t pc_compat_2_9_len = G_N_ELEMENTS(pc_compat_2_9);
203 
204 GlobalProperty pc_compat_2_8[] = {
205     { TYPE_X86_CPU, "tcg-cpuid", "off" },
206     { "kvmclock", "x-mach-use-reliable-get-clock", "off" },
207     { "ICH9-LPC", "x-smi-broadcast", "off" },
208     { TYPE_X86_CPU, "vmware-cpuid-freq", "off" },
209     { "Haswell-" TYPE_X86_CPU, "stepping", "1" },
210 };
211 const size_t pc_compat_2_8_len = G_N_ELEMENTS(pc_compat_2_8);
212 
213 GlobalProperty pc_compat_2_7[] = {
214     { TYPE_X86_CPU, "l3-cache", "off" },
215     { TYPE_X86_CPU, "full-cpuid-auto-level", "off" },
216     { "Opteron_G3" "-" TYPE_X86_CPU, "family", "15" },
217     { "Opteron_G3" "-" TYPE_X86_CPU, "model", "6" },
218     { "Opteron_G3" "-" TYPE_X86_CPU, "stepping", "1" },
219     { "isa-pcspk", "migrate", "off" },
220 };
221 const size_t pc_compat_2_7_len = G_N_ELEMENTS(pc_compat_2_7);
222 
223 GlobalProperty pc_compat_2_6[] = {
224     { TYPE_X86_CPU, "cpuid-0xb", "off" },
225     { "vmxnet3", "romfile", "" },
226     { TYPE_X86_CPU, "fill-mtrr-mask", "off" },
227     { "apic-common", "legacy-instance-id", "on", }
228 };
229 const size_t pc_compat_2_6_len = G_N_ELEMENTS(pc_compat_2_6);
230 
231 GlobalProperty pc_compat_2_5[] = {};
232 const size_t pc_compat_2_5_len = G_N_ELEMENTS(pc_compat_2_5);
233 
234 GlobalProperty pc_compat_2_4[] = {
235     PC_CPU_MODEL_IDS("2.4.0")
236     { "Haswell-" TYPE_X86_CPU, "abm", "off" },
237     { "Haswell-noTSX-" TYPE_X86_CPU, "abm", "off" },
238     { "Broadwell-" TYPE_X86_CPU, "abm", "off" },
239     { "Broadwell-noTSX-" TYPE_X86_CPU, "abm", "off" },
240     { "host" "-" TYPE_X86_CPU, "host-cache-info", "on" },
241     { TYPE_X86_CPU, "check", "off" },
242     { "qemu64" "-" TYPE_X86_CPU, "sse4a", "on" },
243     { "qemu64" "-" TYPE_X86_CPU, "abm", "on" },
244     { "qemu64" "-" TYPE_X86_CPU, "popcnt", "on" },
245     { "qemu32" "-" TYPE_X86_CPU, "popcnt", "on" },
246     { "Opteron_G2" "-" TYPE_X86_CPU, "rdtscp", "on" },
247     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "on" },
248     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "on" },
249     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "on", }
250 };
251 const size_t pc_compat_2_4_len = G_N_ELEMENTS(pc_compat_2_4);
252 
253 GlobalProperty pc_compat_2_3[] = {
254     PC_CPU_MODEL_IDS("2.3.0")
255     { TYPE_X86_CPU, "arat", "off" },
256     { "qemu64" "-" TYPE_X86_CPU, "min-level", "4" },
257     { "kvm64" "-" TYPE_X86_CPU, "min-level", "5" },
258     { "pentium3" "-" TYPE_X86_CPU, "min-level", "2" },
259     { "n270" "-" TYPE_X86_CPU, "min-level", "5" },
260     { "Conroe" "-" TYPE_X86_CPU, "min-level", "4" },
261     { "Penryn" "-" TYPE_X86_CPU, "min-level", "4" },
262     { "Nehalem" "-" TYPE_X86_CPU, "min-level", "4" },
263     { "n270" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
264     { "Penryn" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
265     { "Conroe" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
266     { "Nehalem" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
267     { "Westmere" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
268     { "SandyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
269     { "IvyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
270     { "Haswell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
271     { "Haswell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
272     { "Broadwell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
273     { "Broadwell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
274     { TYPE_X86_CPU, "kvm-no-smi-migration", "on" },
275 };
276 const size_t pc_compat_2_3_len = G_N_ELEMENTS(pc_compat_2_3);
277 
278 GlobalProperty pc_compat_2_2[] = {
279     PC_CPU_MODEL_IDS("2.2.0")
280     { "kvm64" "-" TYPE_X86_CPU, "vme", "off" },
281     { "kvm32" "-" TYPE_X86_CPU, "vme", "off" },
282     { "Conroe" "-" TYPE_X86_CPU, "vme", "off" },
283     { "Penryn" "-" TYPE_X86_CPU, "vme", "off" },
284     { "Nehalem" "-" TYPE_X86_CPU, "vme", "off" },
285     { "Westmere" "-" TYPE_X86_CPU, "vme", "off" },
286     { "SandyBridge" "-" TYPE_X86_CPU, "vme", "off" },
287     { "Haswell" "-" TYPE_X86_CPU, "vme", "off" },
288     { "Broadwell" "-" TYPE_X86_CPU, "vme", "off" },
289     { "Opteron_G1" "-" TYPE_X86_CPU, "vme", "off" },
290     { "Opteron_G2" "-" TYPE_X86_CPU, "vme", "off" },
291     { "Opteron_G3" "-" TYPE_X86_CPU, "vme", "off" },
292     { "Opteron_G4" "-" TYPE_X86_CPU, "vme", "off" },
293     { "Opteron_G5" "-" TYPE_X86_CPU, "vme", "off" },
294     { "Haswell" "-" TYPE_X86_CPU, "f16c", "off" },
295     { "Haswell" "-" TYPE_X86_CPU, "rdrand", "off" },
296     { "Broadwell" "-" TYPE_X86_CPU, "f16c", "off" },
297     { "Broadwell" "-" TYPE_X86_CPU, "rdrand", "off" },
298 };
299 const size_t pc_compat_2_2_len = G_N_ELEMENTS(pc_compat_2_2);
300 
301 GlobalProperty pc_compat_2_1[] = {
302     PC_CPU_MODEL_IDS("2.1.0")
303     { "coreduo" "-" TYPE_X86_CPU, "vmx", "on" },
304     { "core2duo" "-" TYPE_X86_CPU, "vmx", "on" },
305 };
306 const size_t pc_compat_2_1_len = G_N_ELEMENTS(pc_compat_2_1);
307 
308 GlobalProperty pc_compat_2_0[] = {
309     PC_CPU_MODEL_IDS("2.0.0")
310     { "virtio-scsi-pci", "any_layout", "off" },
311     { "PIIX4_PM", "memory-hotplug-support", "off" },
312     { "apic", "version", "0x11" },
313     { "nec-usb-xhci", "superspeed-ports-first", "off" },
314     { "nec-usb-xhci", "force-pcie-endcap", "on" },
315     { "pci-serial", "prog_if", "0" },
316     { "pci-serial-2x", "prog_if", "0" },
317     { "pci-serial-4x", "prog_if", "0" },
318     { "virtio-net-pci", "guest_announce", "off" },
319     { "ICH9-LPC", "memory-hotplug-support", "off" },
320 };
321 const size_t pc_compat_2_0_len = G_N_ELEMENTS(pc_compat_2_0);
322 
323 GSIState *pc_gsi_create(qemu_irq **irqs, bool pci_enabled)
324 {
325     GSIState *s;
326 
327     s = g_new0(GSIState, 1);
328     if (kvm_ioapic_in_kernel()) {
329         kvm_pc_setup_irq_routing(pci_enabled);
330     }
331     *irqs = qemu_allocate_irqs(gsi_handler, s, IOAPIC_NUM_PINS);
332 
333     return s;
334 }
335 
336 static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
337                            unsigned size)
338 {
339 }
340 
341 static uint64_t ioport80_read(void *opaque, hwaddr addr, unsigned size)
342 {
343     return 0xffffffffffffffffULL;
344 }
345 
346 /* MS-DOS compatibility mode FPU exception support */
347 static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
348                            unsigned size)
349 {
350     if (tcg_enabled()) {
351         cpu_set_ignne();
352     }
353 }
354 
355 static uint64_t ioportF0_read(void *opaque, hwaddr addr, unsigned size)
356 {
357     return 0xffffffffffffffffULL;
358 }
359 
360 /* PC cmos mappings */
361 
362 #define REG_EQUIPMENT_BYTE          0x14
363 
364 static void cmos_init_hd(MC146818RtcState *s, int type_ofs, int info_ofs,
365                          int16_t cylinders, int8_t heads, int8_t sectors)
366 {
367     mc146818rtc_set_cmos_data(s, type_ofs, 47);
368     mc146818rtc_set_cmos_data(s, info_ofs, cylinders);
369     mc146818rtc_set_cmos_data(s, info_ofs + 1, cylinders >> 8);
370     mc146818rtc_set_cmos_data(s, info_ofs + 2, heads);
371     mc146818rtc_set_cmos_data(s, info_ofs + 3, 0xff);
372     mc146818rtc_set_cmos_data(s, info_ofs + 4, 0xff);
373     mc146818rtc_set_cmos_data(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
374     mc146818rtc_set_cmos_data(s, info_ofs + 6, cylinders);
375     mc146818rtc_set_cmos_data(s, info_ofs + 7, cylinders >> 8);
376     mc146818rtc_set_cmos_data(s, info_ofs + 8, sectors);
377 }
378 
379 /* convert boot_device letter to something recognizable by the bios */
380 static int boot_device2nibble(char boot_device)
381 {
382     switch(boot_device) {
383     case 'a':
384     case 'b':
385         return 0x01; /* floppy boot */
386     case 'c':
387         return 0x02; /* hard drive boot */
388     case 'd':
389         return 0x03; /* CD-ROM boot */
390     case 'n':
391         return 0x04; /* Network boot */
392     }
393     return 0;
394 }
395 
396 static void set_boot_dev(MC146818RtcState *s, const char *boot_device,
397                          Error **errp)
398 {
399 #define PC_MAX_BOOT_DEVICES 3
400     int nbds, bds[3] = { 0, };
401     int i;
402 
403     nbds = strlen(boot_device);
404     if (nbds > PC_MAX_BOOT_DEVICES) {
405         error_setg(errp, "Too many boot devices for PC");
406         return;
407     }
408     for (i = 0; i < nbds; i++) {
409         bds[i] = boot_device2nibble(boot_device[i]);
410         if (bds[i] == 0) {
411             error_setg(errp, "Invalid boot device for PC: '%c'",
412                        boot_device[i]);
413             return;
414         }
415     }
416     mc146818rtc_set_cmos_data(s, 0x3d, (bds[1] << 4) | bds[0]);
417     mc146818rtc_set_cmos_data(s, 0x38, (bds[2] << 4) | (fd_bootchk ? 0x0 : 0x1));
418 }
419 
420 static void pc_boot_set(void *opaque, const char *boot_device, Error **errp)
421 {
422     set_boot_dev(opaque, boot_device, errp);
423 }
424 
425 static void pc_cmos_init_floppy(MC146818RtcState *rtc_state, ISADevice *floppy)
426 {
427     int val, nb, i;
428     FloppyDriveType fd_type[2] = { FLOPPY_DRIVE_TYPE_NONE,
429                                    FLOPPY_DRIVE_TYPE_NONE };
430 
431     /* floppy type */
432     if (floppy) {
433         for (i = 0; i < 2; i++) {
434             fd_type[i] = isa_fdc_get_drive_type(floppy, i);
435         }
436     }
437     val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
438         cmos_get_fd_drive_type(fd_type[1]);
439     mc146818rtc_set_cmos_data(rtc_state, 0x10, val);
440 
441     val = mc146818rtc_get_cmos_data(rtc_state, REG_EQUIPMENT_BYTE);
442     nb = 0;
443     if (fd_type[0] != FLOPPY_DRIVE_TYPE_NONE) {
444         nb++;
445     }
446     if (fd_type[1] != FLOPPY_DRIVE_TYPE_NONE) {
447         nb++;
448     }
449     switch (nb) {
450     case 0:
451         break;
452     case 1:
453         val |= 0x01; /* 1 drive, ready for boot */
454         break;
455     case 2:
456         val |= 0x41; /* 2 drives, ready for boot */
457         break;
458     }
459     mc146818rtc_set_cmos_data(rtc_state, REG_EQUIPMENT_BYTE, val);
460 }
461 
462 typedef struct pc_cmos_init_late_arg {
463     MC146818RtcState *rtc_state;
464     BusState *idebus[2];
465 } pc_cmos_init_late_arg;
466 
467 typedef struct check_fdc_state {
468     ISADevice *floppy;
469     bool multiple;
470 } CheckFdcState;
471 
472 static int check_fdc(Object *obj, void *opaque)
473 {
474     CheckFdcState *state = opaque;
475     Object *fdc;
476     uint32_t iobase;
477     Error *local_err = NULL;
478 
479     fdc = object_dynamic_cast(obj, TYPE_ISA_FDC);
480     if (!fdc) {
481         return 0;
482     }
483 
484     iobase = object_property_get_uint(obj, "iobase", &local_err);
485     if (local_err || iobase != 0x3f0) {
486         error_free(local_err);
487         return 0;
488     }
489 
490     if (state->floppy) {
491         state->multiple = true;
492     } else {
493         state->floppy = ISA_DEVICE(obj);
494     }
495     return 0;
496 }
497 
498 static const char * const fdc_container_path[] = {
499     "/unattached", "/peripheral", "/peripheral-anon"
500 };
501 
502 /*
503  * Locate the FDC at IO address 0x3f0, in order to configure the CMOS registers
504  * and ACPI objects.
505  */
506 static ISADevice *pc_find_fdc0(void)
507 {
508     int i;
509     Object *container;
510     CheckFdcState state = { 0 };
511 
512     for (i = 0; i < ARRAY_SIZE(fdc_container_path); i++) {
513         container = container_get(qdev_get_machine(), fdc_container_path[i]);
514         object_child_foreach(container, check_fdc, &state);
515     }
516 
517     if (state.multiple) {
518         warn_report("multiple floppy disk controllers with "
519                     "iobase=0x3f0 have been found");
520         error_printf("the one being picked for CMOS setup might not reflect "
521                      "your intent");
522     }
523 
524     return state.floppy;
525 }
526 
527 static void pc_cmos_init_late(void *opaque)
528 {
529     pc_cmos_init_late_arg *arg = opaque;
530     MC146818RtcState *s = arg->rtc_state;
531     int16_t cylinders;
532     int8_t heads, sectors;
533     int val;
534     int i, trans;
535 
536     val = 0;
537     if (arg->idebus[0] && ide_get_geometry(arg->idebus[0], 0,
538                                            &cylinders, &heads, &sectors) >= 0) {
539         cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
540         val |= 0xf0;
541     }
542     if (arg->idebus[0] && ide_get_geometry(arg->idebus[0], 1,
543                                            &cylinders, &heads, &sectors) >= 0) {
544         cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
545         val |= 0x0f;
546     }
547     mc146818rtc_set_cmos_data(s, 0x12, val);
548 
549     val = 0;
550     for (i = 0; i < 4; i++) {
551         /* NOTE: ide_get_geometry() returns the physical
552            geometry.  It is always such that: 1 <= sects <= 63, 1
553            <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
554            geometry can be different if a translation is done. */
555         if (arg->idebus[i / 2] &&
556             ide_get_geometry(arg->idebus[i / 2], i % 2,
557                              &cylinders, &heads, &sectors) >= 0) {
558             trans = ide_get_bios_chs_trans(arg->idebus[i / 2], i % 2) - 1;
559             assert((trans & ~3) == 0);
560             val |= trans << (i * 2);
561         }
562     }
563     mc146818rtc_set_cmos_data(s, 0x39, val);
564 
565     pc_cmos_init_floppy(s, pc_find_fdc0());
566 
567     qemu_unregister_reset(pc_cmos_init_late, opaque);
568 }
569 
570 void pc_cmos_init(PCMachineState *pcms,
571                   BusState *idebus0, BusState *idebus1,
572                   ISADevice *rtc)
573 {
574     int val;
575     static pc_cmos_init_late_arg arg;
576     X86MachineState *x86ms = X86_MACHINE(pcms);
577     MC146818RtcState *s = MC146818_RTC(rtc);
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     object_property_add_link(OBJECT(pcms), "rtc_state",
615                              TYPE_ISA_DEVICE,
616                              (Object **)&x86ms->rtc,
617                              object_property_allow_set_link,
618                              OBJ_PROP_LINK_STRONG);
619     object_property_set_link(OBJECT(pcms), "rtc_state", OBJECT(s),
620                              &error_abort);
621 
622     set_boot_dev(s, MACHINE(pcms)->boot_config.order, &error_fatal);
623 
624     val = 0;
625     val |= 0x02; /* FPU is there */
626     val |= 0x04; /* PS/2 mouse installed */
627     mc146818rtc_set_cmos_data(s, REG_EQUIPMENT_BYTE, val);
628 
629     /* hard drives and FDC */
630     arg.rtc_state = s;
631     arg.idebus[0] = idebus0;
632     arg.idebus[1] = idebus1;
633     qemu_register_reset(pc_cmos_init_late, &arg);
634 }
635 
636 static void handle_a20_line_change(void *opaque, int irq, int level)
637 {
638     X86CPU *cpu = opaque;
639 
640     /* XXX: send to all CPUs ? */
641     /* XXX: add logic to handle multiple A20 line sources */
642     x86_cpu_set_a20(cpu, level);
643 }
644 
645 #define NE2000_NB_MAX 6
646 
647 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
648                                               0x280, 0x380 };
649 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
650 
651 static void pc_init_ne2k_isa(ISABus *bus, NICInfo *nd)
652 {
653     static int nb_ne2k = 0;
654 
655     if (nb_ne2k == NE2000_NB_MAX)
656         return;
657     isa_ne2000_init(bus, ne2000_io[nb_ne2k],
658                     ne2000_irq[nb_ne2k], nd);
659     nb_ne2k++;
660 }
661 
662 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
663 {
664     X86CPU *cpu = opaque;
665 
666     if (level) {
667         cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
668     }
669 }
670 
671 static
672 void pc_machine_done(Notifier *notifier, void *data)
673 {
674     PCMachineState *pcms = container_of(notifier,
675                                         PCMachineState, machine_done);
676     X86MachineState *x86ms = X86_MACHINE(pcms);
677 
678     cxl_hook_up_pxb_registers(pcms->bus, &pcms->cxl_devices_state,
679                               &error_fatal);
680 
681     if (pcms->cxl_devices_state.is_enabled) {
682         cxl_fmws_link_targets(&pcms->cxl_devices_state, &error_fatal);
683     }
684 
685     /* set the number of CPUs */
686     x86_rtc_set_cpus_count(x86ms->rtc, x86ms->boot_cpus);
687 
688     fw_cfg_add_extra_pci_roots(pcms->bus, x86ms->fw_cfg);
689 
690     acpi_setup();
691     if (x86ms->fw_cfg) {
692         fw_cfg_build_smbios(MACHINE(pcms), x86ms->fw_cfg);
693         fw_cfg_build_feature_control(MACHINE(pcms), x86ms->fw_cfg);
694         /* update FW_CFG_NB_CPUS to account for -device added CPUs */
695         fw_cfg_modify_i16(x86ms->fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
696     }
697 }
698 
699 void pc_guest_info_init(PCMachineState *pcms)
700 {
701     X86MachineState *x86ms = X86_MACHINE(pcms);
702 
703     x86ms->apic_xrupt_override = true;
704     pcms->machine_done.notify = pc_machine_done;
705     qemu_add_machine_init_done_notifier(&pcms->machine_done);
706 }
707 
708 /* setup pci memory address space mapping into system address space */
709 void pc_pci_as_mapping_init(MemoryRegion *system_memory,
710                             MemoryRegion *pci_address_space)
711 {
712     /* Set to lower priority than RAM */
713     memory_region_add_subregion_overlap(system_memory, 0x0,
714                                         pci_address_space, -1);
715 }
716 
717 void xen_load_linux(PCMachineState *pcms)
718 {
719     int i;
720     FWCfgState *fw_cfg;
721     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
722     X86MachineState *x86ms = X86_MACHINE(pcms);
723 
724     assert(MACHINE(pcms)->kernel_filename != NULL);
725 
726     fw_cfg = fw_cfg_init_io(FW_CFG_IO_BASE);
727     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
728     rom_set_fw(fw_cfg);
729 
730     x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
731                    pcmc->pvh_enabled);
732     for (i = 0; i < nb_option_roms; i++) {
733         assert(!strcmp(option_rom[i].name, "linuxboot.bin") ||
734                !strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
735                !strcmp(option_rom[i].name, "pvh.bin") ||
736                !strcmp(option_rom[i].name, "multiboot.bin") ||
737                !strcmp(option_rom[i].name, "multiboot_dma.bin"));
738         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
739     }
740     x86ms->fw_cfg = fw_cfg;
741 }
742 
743 #define PC_ROM_MIN_VGA     0xc0000
744 #define PC_ROM_MIN_OPTION  0xc8000
745 #define PC_ROM_MAX         0xe0000
746 #define PC_ROM_ALIGN       0x800
747 #define PC_ROM_SIZE        (PC_ROM_MAX - PC_ROM_MIN_VGA)
748 
749 static hwaddr pc_above_4g_end(PCMachineState *pcms)
750 {
751     X86MachineState *x86ms = X86_MACHINE(pcms);
752 
753     if (pcms->sgx_epc.size != 0) {
754         return sgx_epc_above_4g_end(&pcms->sgx_epc);
755     }
756 
757     return x86ms->above_4g_mem_start + x86ms->above_4g_mem_size;
758 }
759 
760 static void pc_get_device_memory_range(PCMachineState *pcms,
761                                        hwaddr *base,
762                                        ram_addr_t *device_mem_size)
763 {
764     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
765     MachineState *machine = MACHINE(pcms);
766     ram_addr_t size;
767     hwaddr addr;
768 
769     size = machine->maxram_size - machine->ram_size;
770     addr = ROUND_UP(pc_above_4g_end(pcms), 1 * GiB);
771 
772     if (pcmc->enforce_aligned_dimm) {
773         /* size device region assuming 1G page max alignment per slot */
774         size += (1 * GiB) * machine->ram_slots;
775     }
776 
777     *base = addr;
778     *device_mem_size = size;
779 }
780 
781 static uint64_t pc_get_cxl_range_start(PCMachineState *pcms)
782 {
783     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
784     hwaddr cxl_base;
785     ram_addr_t size;
786 
787     if (pcmc->has_reserved_memory) {
788         pc_get_device_memory_range(pcms, &cxl_base, &size);
789         cxl_base += size;
790     } else {
791         cxl_base = pc_above_4g_end(pcms);
792     }
793 
794     return cxl_base;
795 }
796 
797 static uint64_t pc_get_cxl_range_end(PCMachineState *pcms)
798 {
799     uint64_t start = pc_get_cxl_range_start(pcms) + MiB;
800 
801     if (pcms->cxl_devices_state.fixed_windows) {
802         GList *it;
803 
804         start = ROUND_UP(start, 256 * MiB);
805         for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
806             CXLFixedWindow *fw = it->data;
807             start += fw->size;
808         }
809     }
810 
811     return start;
812 }
813 
814 static hwaddr pc_max_used_gpa(PCMachineState *pcms, uint64_t pci_hole64_size)
815 {
816     X86CPU *cpu = X86_CPU(first_cpu);
817     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
818     MachineState *ms = MACHINE(pcms);
819 
820     if (cpu->env.features[FEAT_8000_0001_EDX] & CPUID_EXT2_LM) {
821         /* 64-bit systems */
822         return pc_pci_hole64_start() + pci_hole64_size - 1;
823     }
824 
825     /* 32-bit systems */
826     if (pcmc->broken_32bit_mem_addr_check) {
827         /* old value for compatibility reasons */
828         return ((hwaddr)1 << cpu->phys_bits) - 1;
829     }
830 
831     /*
832      * 32-bit systems don't have hole64 but they might have a region for
833      * memory devices. Even if additional hotplugged memory devices might
834      * not be usable by most guest OSes, we need to still consider them for
835      * calculating the highest possible GPA so that we can properly report
836      * if someone configures them on a CPU that cannot possibly address them.
837      */
838     if (pcmc->has_reserved_memory &&
839         (ms->ram_size < ms->maxram_size)) {
840         hwaddr devmem_start;
841         ram_addr_t devmem_size;
842 
843         pc_get_device_memory_range(pcms, &devmem_start, &devmem_size);
844         devmem_start += devmem_size;
845         return devmem_start - 1;
846     }
847 
848     /* configuration without any memory hotplug */
849     return pc_above_4g_end(pcms) - 1;
850 }
851 
852 /*
853  * AMD systems with an IOMMU have an additional hole close to the
854  * 1Tb, which are special GPAs that cannot be DMA mapped. Depending
855  * on kernel version, VFIO may or may not let you DMA map those ranges.
856  * Starting Linux v5.4 we validate it, and can't create guests on AMD machines
857  * with certain memory sizes. It's also wrong to use those IOVA ranges
858  * in detriment of leading to IOMMU INVALID_DEVICE_REQUEST or worse.
859  * The ranges reserved for Hyper-Transport are:
860  *
861  * FD_0000_0000h - FF_FFFF_FFFFh
862  *
863  * The ranges represent the following:
864  *
865  * Base Address   Top Address  Use
866  *
867  * FD_0000_0000h FD_F7FF_FFFFh Reserved interrupt address space
868  * FD_F800_0000h FD_F8FF_FFFFh Interrupt/EOI IntCtl
869  * FD_F900_0000h FD_F90F_FFFFh Legacy PIC IACK
870  * FD_F910_0000h FD_F91F_FFFFh System Management
871  * FD_F920_0000h FD_FAFF_FFFFh Reserved Page Tables
872  * FD_FB00_0000h FD_FBFF_FFFFh Address Translation
873  * FD_FC00_0000h FD_FDFF_FFFFh I/O Space
874  * FD_FE00_0000h FD_FFFF_FFFFh Configuration
875  * FE_0000_0000h FE_1FFF_FFFFh Extended Configuration/Device Messages
876  * FE_2000_0000h FF_FFFF_FFFFh Reserved
877  *
878  * See AMD IOMMU spec, section 2.1.2 "IOMMU Logical Topology",
879  * Table 3: Special Address Controls (GPA) for more information.
880  */
881 #define AMD_HT_START         0xfd00000000UL
882 #define AMD_HT_END           0xffffffffffUL
883 #define AMD_ABOVE_1TB_START  (AMD_HT_END + 1)
884 #define AMD_HT_SIZE          (AMD_ABOVE_1TB_START - AMD_HT_START)
885 
886 void pc_memory_init(PCMachineState *pcms,
887                     MemoryRegion *system_memory,
888                     MemoryRegion *rom_memory,
889                     uint64_t pci_hole64_size)
890 {
891     int linux_boot, i;
892     MemoryRegion *option_rom_mr;
893     MemoryRegion *ram_below_4g, *ram_above_4g;
894     FWCfgState *fw_cfg;
895     MachineState *machine = MACHINE(pcms);
896     MachineClass *mc = MACHINE_GET_CLASS(machine);
897     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
898     X86MachineState *x86ms = X86_MACHINE(pcms);
899     hwaddr maxphysaddr, maxusedaddr;
900     hwaddr cxl_base, cxl_resv_end = 0;
901     X86CPU *cpu = X86_CPU(first_cpu);
902 
903     assert(machine->ram_size == x86ms->below_4g_mem_size +
904                                 x86ms->above_4g_mem_size);
905 
906     linux_boot = (machine->kernel_filename != NULL);
907 
908     /*
909      * The HyperTransport range close to the 1T boundary is unique to AMD
910      * hosts with IOMMUs enabled. Restrict the ram-above-4g relocation
911      * to above 1T to AMD vCPUs only. @enforce_amd_1tb_hole is only false in
912      * older machine types (<= 7.0) for compatibility purposes.
913      */
914     if (IS_AMD_CPU(&cpu->env) && pcmc->enforce_amd_1tb_hole) {
915         /* Bail out if max possible address does not cross HT range */
916         if (pc_max_used_gpa(pcms, pci_hole64_size) >= AMD_HT_START) {
917             x86ms->above_4g_mem_start = AMD_ABOVE_1TB_START;
918         }
919 
920         /*
921          * Advertise the HT region if address space covers the reserved
922          * region or if we relocate.
923          */
924         if (cpu->phys_bits >= 40) {
925             e820_add_entry(AMD_HT_START, AMD_HT_SIZE, E820_RESERVED);
926         }
927     }
928 
929     /*
930      * phys-bits is required to be appropriately configured
931      * to make sure max used GPA is reachable.
932      */
933     maxusedaddr = pc_max_used_gpa(pcms, pci_hole64_size);
934     maxphysaddr = ((hwaddr)1 << cpu->phys_bits) - 1;
935     if (maxphysaddr < maxusedaddr) {
936         error_report("Address space limit 0x%"PRIx64" < 0x%"PRIx64
937                      " phys-bits too low (%u)",
938                      maxphysaddr, maxusedaddr, cpu->phys_bits);
939         exit(EXIT_FAILURE);
940     }
941 
942     /*
943      * Split single memory region and use aliases to address portions of it,
944      * done for backwards compatibility with older qemus.
945      */
946     ram_below_4g = g_malloc(sizeof(*ram_below_4g));
947     memory_region_init_alias(ram_below_4g, NULL, "ram-below-4g", machine->ram,
948                              0, x86ms->below_4g_mem_size);
949     memory_region_add_subregion(system_memory, 0, ram_below_4g);
950     e820_add_entry(0, x86ms->below_4g_mem_size, E820_RAM);
951     if (x86ms->above_4g_mem_size > 0) {
952         ram_above_4g = g_malloc(sizeof(*ram_above_4g));
953         memory_region_init_alias(ram_above_4g, NULL, "ram-above-4g",
954                                  machine->ram,
955                                  x86ms->below_4g_mem_size,
956                                  x86ms->above_4g_mem_size);
957         memory_region_add_subregion(system_memory, x86ms->above_4g_mem_start,
958                                     ram_above_4g);
959         e820_add_entry(x86ms->above_4g_mem_start, x86ms->above_4g_mem_size,
960                        E820_RAM);
961     }
962 
963     if (pcms->sgx_epc.size != 0) {
964         e820_add_entry(pcms->sgx_epc.base, pcms->sgx_epc.size, E820_RESERVED);
965     }
966 
967     if (!pcmc->has_reserved_memory &&
968         (machine->ram_slots ||
969          (machine->maxram_size > machine->ram_size))) {
970 
971         error_report("\"-memory 'slots|maxmem'\" is not supported by: %s",
972                      mc->name);
973         exit(EXIT_FAILURE);
974     }
975 
976     /* initialize device memory address space */
977     if (pcmc->has_reserved_memory &&
978         (machine->ram_size < machine->maxram_size)) {
979         ram_addr_t device_mem_size;
980         hwaddr device_mem_base;
981 
982         if (machine->ram_slots > ACPI_MAX_RAM_SLOTS) {
983             error_report("unsupported amount of memory slots: %"PRIu64,
984                          machine->ram_slots);
985             exit(EXIT_FAILURE);
986         }
987 
988         if (QEMU_ALIGN_UP(machine->maxram_size,
989                           TARGET_PAGE_SIZE) != machine->maxram_size) {
990             error_report("maximum memory size must by aligned to multiple of "
991                          "%d bytes", TARGET_PAGE_SIZE);
992             exit(EXIT_FAILURE);
993         }
994 
995         pc_get_device_memory_range(pcms, &device_mem_base, &device_mem_size);
996 
997         if (device_mem_base + device_mem_size < device_mem_size) {
998             error_report("unsupported amount of maximum memory: " RAM_ADDR_FMT,
999                          machine->maxram_size);
1000             exit(EXIT_FAILURE);
1001         }
1002         machine_memory_devices_init(machine, device_mem_base, device_mem_size);
1003     }
1004 
1005     if (pcms->cxl_devices_state.is_enabled) {
1006         MemoryRegion *mr = &pcms->cxl_devices_state.host_mr;
1007         hwaddr cxl_size = MiB;
1008 
1009         cxl_base = pc_get_cxl_range_start(pcms);
1010         memory_region_init(mr, OBJECT(machine), "cxl_host_reg", cxl_size);
1011         memory_region_add_subregion(system_memory, cxl_base, mr);
1012         cxl_resv_end = cxl_base + cxl_size;
1013         if (pcms->cxl_devices_state.fixed_windows) {
1014             hwaddr cxl_fmw_base;
1015             GList *it;
1016 
1017             cxl_fmw_base = ROUND_UP(cxl_base + cxl_size, 256 * MiB);
1018             for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
1019                 CXLFixedWindow *fw = it->data;
1020 
1021                 fw->base = cxl_fmw_base;
1022                 memory_region_init_io(&fw->mr, OBJECT(machine), &cfmws_ops, fw,
1023                                       "cxl-fixed-memory-region", fw->size);
1024                 memory_region_add_subregion(system_memory, fw->base, &fw->mr);
1025                 cxl_fmw_base += fw->size;
1026                 cxl_resv_end = cxl_fmw_base;
1027             }
1028         }
1029     }
1030 
1031     /* Initialize PC system firmware */
1032     pc_system_firmware_init(pcms, rom_memory);
1033 
1034     option_rom_mr = g_malloc(sizeof(*option_rom_mr));
1035     memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE,
1036                            &error_fatal);
1037     if (pcmc->pci_enabled) {
1038         memory_region_set_readonly(option_rom_mr, true);
1039     }
1040     memory_region_add_subregion_overlap(rom_memory,
1041                                         PC_ROM_MIN_VGA,
1042                                         option_rom_mr,
1043                                         1);
1044 
1045     fw_cfg = fw_cfg_arch_create(machine,
1046                                 x86ms->boot_cpus, x86ms->apic_id_limit);
1047 
1048     rom_set_fw(fw_cfg);
1049 
1050     if (machine->device_memory) {
1051         uint64_t *val = g_malloc(sizeof(*val));
1052         PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1053         uint64_t res_mem_end = machine->device_memory->base;
1054 
1055         if (!pcmc->broken_reserved_end) {
1056             res_mem_end += memory_region_size(&machine->device_memory->mr);
1057         }
1058 
1059         if (pcms->cxl_devices_state.is_enabled) {
1060             res_mem_end = cxl_resv_end;
1061         }
1062         *val = cpu_to_le64(ROUND_UP(res_mem_end, 1 * GiB));
1063         fw_cfg_add_file(fw_cfg, "etc/reserved-memory-end", val, sizeof(*val));
1064     }
1065 
1066     if (linux_boot) {
1067         x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
1068                        pcmc->pvh_enabled);
1069     }
1070 
1071     for (i = 0; i < nb_option_roms; i++) {
1072         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1073     }
1074     x86ms->fw_cfg = fw_cfg;
1075 
1076     /* Init default IOAPIC address space */
1077     x86ms->ioapic_as = &address_space_memory;
1078 
1079     /* Init ACPI memory hotplug IO base address */
1080     pcms->memhp_io_base = ACPI_MEMORY_HOTPLUG_BASE;
1081 }
1082 
1083 /*
1084  * The 64bit pci hole starts after "above 4G RAM" and
1085  * potentially the space reserved for memory hotplug.
1086  */
1087 uint64_t pc_pci_hole64_start(void)
1088 {
1089     PCMachineState *pcms = PC_MACHINE(qdev_get_machine());
1090     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1091     MachineState *ms = MACHINE(pcms);
1092     uint64_t hole64_start = 0;
1093     ram_addr_t size = 0;
1094 
1095     if (pcms->cxl_devices_state.is_enabled) {
1096         hole64_start = pc_get_cxl_range_end(pcms);
1097     } else if (pcmc->has_reserved_memory && (ms->ram_size < ms->maxram_size)) {
1098         pc_get_device_memory_range(pcms, &hole64_start, &size);
1099         if (!pcmc->broken_reserved_end) {
1100             hole64_start += size;
1101         }
1102     } else {
1103         hole64_start = pc_above_4g_end(pcms);
1104     }
1105 
1106     return ROUND_UP(hole64_start, 1 * GiB);
1107 }
1108 
1109 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1110 {
1111     DeviceState *dev = NULL;
1112 
1113     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_VGA);
1114     if (pci_bus) {
1115         PCIDevice *pcidev = pci_vga_init(pci_bus);
1116         dev = pcidev ? &pcidev->qdev : NULL;
1117     } else if (isa_bus) {
1118         ISADevice *isadev = isa_vga_init(isa_bus);
1119         dev = isadev ? DEVICE(isadev) : NULL;
1120     }
1121     rom_reset_order_override();
1122     return dev;
1123 }
1124 
1125 static const MemoryRegionOps ioport80_io_ops = {
1126     .write = ioport80_write,
1127     .read = ioport80_read,
1128     .endianness = DEVICE_NATIVE_ENDIAN,
1129     .impl = {
1130         .min_access_size = 1,
1131         .max_access_size = 1,
1132     },
1133 };
1134 
1135 static const MemoryRegionOps ioportF0_io_ops = {
1136     .write = ioportF0_write,
1137     .read = ioportF0_read,
1138     .endianness = DEVICE_NATIVE_ENDIAN,
1139     .impl = {
1140         .min_access_size = 1,
1141         .max_access_size = 1,
1142     },
1143 };
1144 
1145 static void pc_superio_init(ISABus *isa_bus, bool create_fdctrl,
1146                             bool create_i8042, bool no_vmport)
1147 {
1148     int i;
1149     DriveInfo *fd[MAX_FD];
1150     qemu_irq *a20_line;
1151     ISADevice *fdc, *i8042, *port92, *vmmouse;
1152 
1153     serial_hds_isa_init(isa_bus, 0, MAX_ISA_SERIAL_PORTS);
1154     parallel_hds_isa_init(isa_bus, MAX_PARALLEL_PORTS);
1155 
1156     for (i = 0; i < MAX_FD; i++) {
1157         fd[i] = drive_get(IF_FLOPPY, 0, i);
1158         create_fdctrl |= !!fd[i];
1159     }
1160     if (create_fdctrl) {
1161         fdc = isa_new(TYPE_ISA_FDC);
1162         if (fdc) {
1163             isa_realize_and_unref(fdc, isa_bus, &error_fatal);
1164             isa_fdc_init_drives(fdc, fd);
1165         }
1166     }
1167 
1168     if (!create_i8042) {
1169         return;
1170     }
1171 
1172     i8042 = isa_create_simple(isa_bus, TYPE_I8042);
1173     if (!no_vmport) {
1174         isa_create_simple(isa_bus, TYPE_VMPORT);
1175         vmmouse = isa_try_new("vmmouse");
1176     } else {
1177         vmmouse = NULL;
1178     }
1179     if (vmmouse) {
1180         object_property_set_link(OBJECT(vmmouse), TYPE_I8042, OBJECT(i8042),
1181                                  &error_abort);
1182         isa_realize_and_unref(vmmouse, isa_bus, &error_fatal);
1183     }
1184     port92 = isa_create_simple(isa_bus, TYPE_PORT92);
1185 
1186     a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1187     i8042_setup_a20_line(i8042, a20_line[0]);
1188     qdev_connect_gpio_out_named(DEVICE(port92),
1189                                 PORT92_A20_LINE, 0, a20_line[1]);
1190     g_free(a20_line);
1191 }
1192 
1193 void pc_basic_device_init(struct PCMachineState *pcms,
1194                           ISABus *isa_bus, qemu_irq *gsi,
1195                           ISADevice *rtc_state,
1196                           bool create_fdctrl,
1197                           uint32_t hpet_irqs)
1198 {
1199     int i;
1200     DeviceState *hpet = NULL;
1201     int pit_isa_irq = 0;
1202     qemu_irq pit_alt_irq = NULL;
1203     qemu_irq rtc_irq = NULL;
1204     ISADevice *pit = NULL;
1205     MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1206     MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
1207     X86MachineState *x86ms = X86_MACHINE(pcms);
1208 
1209     memory_region_init_io(ioport80_io, NULL, &ioport80_io_ops, NULL, "ioport80", 1);
1210     memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
1211 
1212     memory_region_init_io(ioportF0_io, NULL, &ioportF0_io_ops, NULL, "ioportF0", 1);
1213     memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
1214 
1215     /*
1216      * Check if an HPET shall be created.
1217      *
1218      * Without KVM_CAP_PIT_STATE2, we cannot switch off the in-kernel PIT
1219      * when the HPET wants to take over. Thus we have to disable the latter.
1220      */
1221     if (pcms->hpet_enabled && (!kvm_irqchip_in_kernel() ||
1222                                kvm_has_pit_state2())) {
1223         hpet = qdev_try_new(TYPE_HPET);
1224         if (!hpet) {
1225             error_report("couldn't create HPET device");
1226             exit(1);
1227         }
1228         /*
1229          * For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-*,
1230          * use IRQ16~23, IRQ8 and IRQ2.  If the user has already set
1231          * the property, use whatever mask they specified.
1232          */
1233         uint8_t compat = object_property_get_uint(OBJECT(hpet),
1234                 HPET_INTCAP, NULL);
1235         if (!compat) {
1236             qdev_prop_set_uint32(hpet, HPET_INTCAP, hpet_irqs);
1237         }
1238         sysbus_realize_and_unref(SYS_BUS_DEVICE(hpet), &error_fatal);
1239         sysbus_mmio_map(SYS_BUS_DEVICE(hpet), 0, HPET_BASE);
1240 
1241         for (i = 0; i < IOAPIC_NUM_PINS; i++) {
1242             sysbus_connect_irq(SYS_BUS_DEVICE(hpet), i, gsi[i]);
1243         }
1244         pit_isa_irq = -1;
1245         pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1246         rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1247     }
1248 
1249     if (rtc_irq) {
1250         qdev_connect_gpio_out(DEVICE(rtc_state), 0, rtc_irq);
1251     } else {
1252         uint32_t irq = object_property_get_uint(OBJECT(rtc_state),
1253                                                 "irq",
1254                                                 &error_fatal);
1255         isa_connect_gpio_out(rtc_state, 0, irq);
1256     }
1257     object_property_add_alias(OBJECT(pcms), "rtc-time", OBJECT(rtc_state),
1258                               "date");
1259 
1260 #ifdef CONFIG_XEN_EMU
1261     if (xen_mode == XEN_EMULATE) {
1262         xen_overlay_create();
1263         xen_evtchn_create(IOAPIC_NUM_PINS, gsi);
1264         xen_gnttab_create();
1265         xen_xenstore_create();
1266         if (pcms->bus) {
1267             pci_create_simple(pcms->bus, -1, "xen-platform");
1268         }
1269         xen_bus_init();
1270         xen_be_init();
1271     }
1272 #endif
1273 
1274     qemu_register_boot_set(pc_boot_set, rtc_state);
1275 
1276     if (!xen_enabled() &&
1277         (x86ms->pit == ON_OFF_AUTO_AUTO || x86ms->pit == ON_OFF_AUTO_ON)) {
1278         if (kvm_pit_in_kernel()) {
1279             pit = kvm_pit_init(isa_bus, 0x40);
1280         } else {
1281             pit = i8254_pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1282         }
1283         if (hpet) {
1284             /* connect PIT to output control line of the HPET */
1285             qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(DEVICE(pit), 0));
1286         }
1287         pcspk_init(pcms->pcspk, isa_bus, pit);
1288     }
1289 
1290     /* Super I/O */
1291     pc_superio_init(isa_bus, create_fdctrl, pcms->i8042_enabled,
1292                     pcms->vmport != ON_OFF_AUTO_ON);
1293 }
1294 
1295 void pc_nic_init(PCMachineClass *pcmc, ISABus *isa_bus, PCIBus *pci_bus)
1296 {
1297     MachineClass *mc = MACHINE_CLASS(pcmc);
1298     int i;
1299 
1300     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_NIC);
1301     for (i = 0; i < nb_nics; i++) {
1302         NICInfo *nd = &nd_table[i];
1303         const char *model = nd->model ? nd->model : mc->default_nic;
1304 
1305         if (g_str_equal(model, "ne2k_isa")) {
1306             pc_init_ne2k_isa(isa_bus, nd);
1307         } else {
1308             pci_nic_init_nofail(nd, pci_bus, model, NULL);
1309         }
1310     }
1311     rom_reset_order_override();
1312 }
1313 
1314 void pc_i8259_create(ISABus *isa_bus, qemu_irq *i8259_irqs)
1315 {
1316     qemu_irq *i8259;
1317 
1318     if (kvm_pic_in_kernel()) {
1319         i8259 = kvm_i8259_init(isa_bus);
1320     } else if (xen_enabled()) {
1321         i8259 = xen_interrupt_controller_init();
1322     } else {
1323         i8259 = i8259_init(isa_bus, x86_allocate_cpu_irq());
1324     }
1325 
1326     for (size_t i = 0; i < ISA_NUM_IRQS; i++) {
1327         i8259_irqs[i] = i8259[i];
1328     }
1329 
1330     g_free(i8259);
1331 }
1332 
1333 static void pc_memory_pre_plug(HotplugHandler *hotplug_dev, DeviceState *dev,
1334                                Error **errp)
1335 {
1336     const PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1337     const X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1338     const PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1339     const MachineState *ms = MACHINE(hotplug_dev);
1340     const bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1341     const uint64_t legacy_align = TARGET_PAGE_SIZE;
1342     Error *local_err = NULL;
1343 
1344     /*
1345      * When -no-acpi is used with Q35 machine type, no ACPI is built,
1346      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1347      * addition to cover this case.
1348      */
1349     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1350         error_setg(errp,
1351                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1352         return;
1353     }
1354 
1355     if (is_nvdimm && !ms->nvdimms_state->is_enabled) {
1356         error_setg(errp, "nvdimm is not enabled: missing 'nvdimm' in '-M'");
1357         return;
1358     }
1359 
1360     hotplug_handler_pre_plug(x86ms->acpi_dev, dev, &local_err);
1361     if (local_err) {
1362         error_propagate(errp, local_err);
1363         return;
1364     }
1365 
1366     pc_dimm_pre_plug(PC_DIMM(dev), MACHINE(hotplug_dev),
1367                      pcmc->enforce_aligned_dimm ? NULL : &legacy_align, errp);
1368 }
1369 
1370 static void pc_memory_plug(HotplugHandler *hotplug_dev,
1371                            DeviceState *dev, Error **errp)
1372 {
1373     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1374     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1375     MachineState *ms = MACHINE(hotplug_dev);
1376     bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1377 
1378     pc_dimm_plug(PC_DIMM(dev), MACHINE(pcms));
1379 
1380     if (is_nvdimm) {
1381         nvdimm_plug(ms->nvdimms_state);
1382     }
1383 
1384     hotplug_handler_plug(x86ms->acpi_dev, dev, &error_abort);
1385 }
1386 
1387 static void pc_memory_unplug_request(HotplugHandler *hotplug_dev,
1388                                      DeviceState *dev, Error **errp)
1389 {
1390     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1391 
1392     /*
1393      * When -no-acpi is used with Q35 machine type, no ACPI is built,
1394      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1395      * addition to cover this case.
1396      */
1397     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1398         error_setg(errp,
1399                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1400         return;
1401     }
1402 
1403     if (object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM)) {
1404         error_setg(errp, "nvdimm device hot unplug is not supported yet.");
1405         return;
1406     }
1407 
1408     hotplug_handler_unplug_request(x86ms->acpi_dev, dev,
1409                                    errp);
1410 }
1411 
1412 static void pc_memory_unplug(HotplugHandler *hotplug_dev,
1413                              DeviceState *dev, Error **errp)
1414 {
1415     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1416     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1417     Error *local_err = NULL;
1418 
1419     hotplug_handler_unplug(x86ms->acpi_dev, dev, &local_err);
1420     if (local_err) {
1421         goto out;
1422     }
1423 
1424     pc_dimm_unplug(PC_DIMM(dev), MACHINE(pcms));
1425     qdev_unrealize(dev);
1426  out:
1427     error_propagate(errp, local_err);
1428 }
1429 
1430 static void pc_machine_device_pre_plug_cb(HotplugHandler *hotplug_dev,
1431                                           DeviceState *dev, Error **errp)
1432 {
1433     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1434         pc_memory_pre_plug(hotplug_dev, dev, errp);
1435     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1436         x86_cpu_pre_plug(hotplug_dev, dev, errp);
1437     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1438         virtio_md_pci_pre_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1439     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1440         /* Declare the APIC range as the reserved MSI region */
1441         char *resv_prop_str = g_strdup_printf("0xfee00000:0xfeefffff:%d",
1442                                               VIRTIO_IOMMU_RESV_MEM_T_MSI);
1443 
1444         object_property_set_uint(OBJECT(dev), "len-reserved-regions", 1, errp);
1445         object_property_set_str(OBJECT(dev), "reserved-regions[0]",
1446                                 resv_prop_str, errp);
1447         g_free(resv_prop_str);
1448     }
1449 
1450     if (object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE) ||
1451         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1452         PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1453 
1454         if (pcms->iommu) {
1455             error_setg(errp, "QEMU does not support multiple vIOMMUs "
1456                        "for x86 yet.");
1457             return;
1458         }
1459         pcms->iommu = dev;
1460     }
1461 }
1462 
1463 static void pc_machine_device_plug_cb(HotplugHandler *hotplug_dev,
1464                                       DeviceState *dev, Error **errp)
1465 {
1466     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1467         pc_memory_plug(hotplug_dev, dev, errp);
1468     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1469         x86_cpu_plug(hotplug_dev, dev, errp);
1470     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1471         virtio_md_pci_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1472     }
1473 }
1474 
1475 static void pc_machine_device_unplug_request_cb(HotplugHandler *hotplug_dev,
1476                                                 DeviceState *dev, Error **errp)
1477 {
1478     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1479         pc_memory_unplug_request(hotplug_dev, dev, errp);
1480     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1481         x86_cpu_unplug_request_cb(hotplug_dev, dev, errp);
1482     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1483         virtio_md_pci_unplug_request(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev),
1484                                      errp);
1485     } else {
1486         error_setg(errp, "acpi: device unplug request for not supported device"
1487                    " type: %s", object_get_typename(OBJECT(dev)));
1488     }
1489 }
1490 
1491 static void pc_machine_device_unplug_cb(HotplugHandler *hotplug_dev,
1492                                         DeviceState *dev, Error **errp)
1493 {
1494     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1495         pc_memory_unplug(hotplug_dev, dev, errp);
1496     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1497         x86_cpu_unplug_cb(hotplug_dev, dev, errp);
1498     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1499         virtio_md_pci_unplug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1500     } else {
1501         error_setg(errp, "acpi: device unplug for not supported device"
1502                    " type: %s", object_get_typename(OBJECT(dev)));
1503     }
1504 }
1505 
1506 static HotplugHandler *pc_get_hotplug_handler(MachineState *machine,
1507                                              DeviceState *dev)
1508 {
1509     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM) ||
1510         object_dynamic_cast(OBJECT(dev), TYPE_CPU) ||
1511         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI) ||
1512         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI) ||
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_smbus(Object *obj, Error **errp)
1538 {
1539     PCMachineState *pcms = PC_MACHINE(obj);
1540 
1541     return pcms->smbus_enabled;
1542 }
1543 
1544 static void pc_machine_set_smbus(Object *obj, bool value, Error **errp)
1545 {
1546     PCMachineState *pcms = PC_MACHINE(obj);
1547 
1548     pcms->smbus_enabled = value;
1549 }
1550 
1551 static bool pc_machine_get_sata(Object *obj, Error **errp)
1552 {
1553     PCMachineState *pcms = PC_MACHINE(obj);
1554 
1555     return pcms->sata_enabled;
1556 }
1557 
1558 static void pc_machine_set_sata(Object *obj, bool value, Error **errp)
1559 {
1560     PCMachineState *pcms = PC_MACHINE(obj);
1561 
1562     pcms->sata_enabled = value;
1563 }
1564 
1565 static bool pc_machine_get_hpet(Object *obj, Error **errp)
1566 {
1567     PCMachineState *pcms = PC_MACHINE(obj);
1568 
1569     return pcms->hpet_enabled;
1570 }
1571 
1572 static void pc_machine_set_hpet(Object *obj, bool value, Error **errp)
1573 {
1574     PCMachineState *pcms = PC_MACHINE(obj);
1575 
1576     pcms->hpet_enabled = value;
1577 }
1578 
1579 static bool pc_machine_get_i8042(Object *obj, Error **errp)
1580 {
1581     PCMachineState *pcms = PC_MACHINE(obj);
1582 
1583     return pcms->i8042_enabled;
1584 }
1585 
1586 static void pc_machine_set_i8042(Object *obj, bool value, Error **errp)
1587 {
1588     PCMachineState *pcms = PC_MACHINE(obj);
1589 
1590     pcms->i8042_enabled = value;
1591 }
1592 
1593 static bool pc_machine_get_default_bus_bypass_iommu(Object *obj, Error **errp)
1594 {
1595     PCMachineState *pcms = PC_MACHINE(obj);
1596 
1597     return pcms->default_bus_bypass_iommu;
1598 }
1599 
1600 static void pc_machine_set_default_bus_bypass_iommu(Object *obj, bool value,
1601                                                     Error **errp)
1602 {
1603     PCMachineState *pcms = PC_MACHINE(obj);
1604 
1605     pcms->default_bus_bypass_iommu = value;
1606 }
1607 
1608 static void pc_machine_get_smbios_ep(Object *obj, Visitor *v, const char *name,
1609                                      void *opaque, Error **errp)
1610 {
1611     PCMachineState *pcms = PC_MACHINE(obj);
1612     SmbiosEntryPointType smbios_entry_point_type = pcms->smbios_entry_point_type;
1613 
1614     visit_type_SmbiosEntryPointType(v, name, &smbios_entry_point_type, errp);
1615 }
1616 
1617 static void pc_machine_set_smbios_ep(Object *obj, Visitor *v, const char *name,
1618                                      void *opaque, Error **errp)
1619 {
1620     PCMachineState *pcms = PC_MACHINE(obj);
1621 
1622     visit_type_SmbiosEntryPointType(v, name, &pcms->smbios_entry_point_type, errp);
1623 }
1624 
1625 static void pc_machine_get_max_ram_below_4g(Object *obj, Visitor *v,
1626                                             const char *name, void *opaque,
1627                                             Error **errp)
1628 {
1629     PCMachineState *pcms = PC_MACHINE(obj);
1630     uint64_t value = pcms->max_ram_below_4g;
1631 
1632     visit_type_size(v, name, &value, errp);
1633 }
1634 
1635 static void pc_machine_set_max_ram_below_4g(Object *obj, Visitor *v,
1636                                             const char *name, void *opaque,
1637                                             Error **errp)
1638 {
1639     PCMachineState *pcms = PC_MACHINE(obj);
1640     uint64_t value;
1641 
1642     if (!visit_type_size(v, name, &value, errp)) {
1643         return;
1644     }
1645     if (value > 4 * GiB) {
1646         error_setg(errp,
1647                    "Machine option 'max-ram-below-4g=%"PRIu64
1648                    "' expects size less than or equal to 4G", value);
1649         return;
1650     }
1651 
1652     if (value < 1 * MiB) {
1653         warn_report("Only %" PRIu64 " bytes of RAM below the 4GiB boundary,"
1654                     "BIOS may not work with less than 1MiB", value);
1655     }
1656 
1657     pcms->max_ram_below_4g = value;
1658 }
1659 
1660 static void pc_machine_get_max_fw_size(Object *obj, Visitor *v,
1661                                        const char *name, void *opaque,
1662                                        Error **errp)
1663 {
1664     PCMachineState *pcms = PC_MACHINE(obj);
1665     uint64_t value = pcms->max_fw_size;
1666 
1667     visit_type_size(v, name, &value, errp);
1668 }
1669 
1670 static void pc_machine_set_max_fw_size(Object *obj, Visitor *v,
1671                                        const char *name, void *opaque,
1672                                        Error **errp)
1673 {
1674     PCMachineState *pcms = PC_MACHINE(obj);
1675     uint64_t value;
1676 
1677     if (!visit_type_size(v, name, &value, errp)) {
1678         return;
1679     }
1680 
1681     /*
1682      * We don't have a theoretically justifiable exact lower bound on the base
1683      * address of any flash mapping. In practice, the IO-APIC MMIO range is
1684      * [0xFEE00000..0xFEE01000] -- see IO_APIC_DEFAULT_ADDRESS --, leaving free
1685      * only 18MiB-4KiB below 4GiB. For now, restrict the cumulative mapping to
1686      * 16MiB in size.
1687      */
1688     if (value > 16 * MiB) {
1689         error_setg(errp,
1690                    "User specified max allowed firmware size %" PRIu64 " is "
1691                    "greater than 16MiB. If combined firmware size exceeds "
1692                    "16MiB the system may not boot, or experience intermittent"
1693                    "stability issues.",
1694                    value);
1695         return;
1696     }
1697 
1698     pcms->max_fw_size = value;
1699 }
1700 
1701 
1702 static void pc_machine_initfn(Object *obj)
1703 {
1704     PCMachineState *pcms = PC_MACHINE(obj);
1705     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1706 
1707 #ifdef CONFIG_VMPORT
1708     pcms->vmport = ON_OFF_AUTO_AUTO;
1709 #else
1710     pcms->vmport = ON_OFF_AUTO_OFF;
1711 #endif /* CONFIG_VMPORT */
1712     pcms->max_ram_below_4g = 0; /* use default */
1713     pcms->smbios_entry_point_type = pcmc->default_smbios_ep_type;
1714 
1715     /* acpi build is enabled by default if machine supports it */
1716     pcms->acpi_build_enabled = pcmc->has_acpi_build;
1717     pcms->smbus_enabled = true;
1718     pcms->sata_enabled = true;
1719     pcms->i8042_enabled = true;
1720     pcms->max_fw_size = 8 * MiB;
1721 #ifdef CONFIG_HPET
1722     pcms->hpet_enabled = true;
1723 #endif
1724     pcms->default_bus_bypass_iommu = false;
1725 
1726     pc_system_flash_create(pcms);
1727     pcms->pcspk = isa_new(TYPE_PC_SPEAKER);
1728     object_property_add_alias(OBJECT(pcms), "pcspk-audiodev",
1729                               OBJECT(pcms->pcspk), "audiodev");
1730     cxl_machine_init(obj, &pcms->cxl_devices_state);
1731 }
1732 
1733 int pc_machine_kvm_type(MachineState *machine, const char *kvm_type)
1734 {
1735     return 0;
1736 }
1737 
1738 static void pc_machine_reset(MachineState *machine, ShutdownCause reason)
1739 {
1740     CPUState *cs;
1741     X86CPU *cpu;
1742 
1743     qemu_devices_reset(reason);
1744 
1745     /* Reset APIC after devices have been reset to cancel
1746      * any changes that qemu_devices_reset() might have done.
1747      */
1748     CPU_FOREACH(cs) {
1749         cpu = X86_CPU(cs);
1750 
1751         x86_cpu_after_reset(cpu);
1752     }
1753 }
1754 
1755 static void pc_machine_wakeup(MachineState *machine)
1756 {
1757     cpu_synchronize_all_states();
1758     pc_machine_reset(machine, SHUTDOWN_CAUSE_NONE);
1759     cpu_synchronize_all_post_reset();
1760 }
1761 
1762 static bool pc_hotplug_allowed(MachineState *ms, DeviceState *dev, Error **errp)
1763 {
1764     X86IOMMUState *iommu = x86_iommu_get_default();
1765     IntelIOMMUState *intel_iommu;
1766 
1767     if (iommu &&
1768         object_dynamic_cast((Object *)iommu, TYPE_INTEL_IOMMU_DEVICE) &&
1769         object_dynamic_cast((Object *)dev, "vfio-pci")) {
1770         intel_iommu = INTEL_IOMMU_DEVICE(iommu);
1771         if (!intel_iommu->caching_mode) {
1772             error_setg(errp, "Device assignment is not allowed without "
1773                        "enabling caching-mode=on for Intel IOMMU.");
1774             return false;
1775         }
1776     }
1777 
1778     return true;
1779 }
1780 
1781 static void pc_machine_class_init(ObjectClass *oc, void *data)
1782 {
1783     MachineClass *mc = MACHINE_CLASS(oc);
1784     PCMachineClass *pcmc = PC_MACHINE_CLASS(oc);
1785     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
1786 
1787     pcmc->pci_enabled = true;
1788     pcmc->has_acpi_build = true;
1789     pcmc->rsdp_in_ram = true;
1790     pcmc->smbios_defaults = true;
1791     pcmc->smbios_uuid_encoded = true;
1792     pcmc->gigabyte_align = true;
1793     pcmc->has_reserved_memory = true;
1794     pcmc->kvmclock_enabled = true;
1795     pcmc->enforce_aligned_dimm = true;
1796     pcmc->enforce_amd_1tb_hole = true;
1797     /* BIOS ACPI tables: 128K. Other BIOS datastructures: less than 4K reported
1798      * to be used at the moment, 32K should be enough for a while.  */
1799     pcmc->acpi_data_size = 0x20000 + 0x8000;
1800     pcmc->pvh_enabled = true;
1801     pcmc->kvmclock_create_always = true;
1802     pcmc->resizable_acpi_blob = true;
1803     assert(!mc->get_hotplug_handler);
1804     mc->get_hotplug_handler = pc_get_hotplug_handler;
1805     mc->hotplug_allowed = pc_hotplug_allowed;
1806     mc->cpu_index_to_instance_props = x86_cpu_index_to_props;
1807     mc->get_default_cpu_node_id = x86_get_default_cpu_node_id;
1808     mc->possible_cpu_arch_ids = x86_possible_cpu_arch_ids;
1809     mc->auto_enable_numa_with_memhp = true;
1810     mc->auto_enable_numa_with_memdev = true;
1811     mc->has_hotpluggable_cpus = true;
1812     mc->default_boot_order = "cad";
1813     mc->block_default_type = IF_IDE;
1814     mc->max_cpus = 255;
1815     mc->reset = pc_machine_reset;
1816     mc->wakeup = pc_machine_wakeup;
1817     hc->pre_plug = pc_machine_device_pre_plug_cb;
1818     hc->plug = pc_machine_device_plug_cb;
1819     hc->unplug_request = pc_machine_device_unplug_request_cb;
1820     hc->unplug = pc_machine_device_unplug_cb;
1821     mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE;
1822     mc->nvdimm_supported = true;
1823     mc->smp_props.dies_supported = true;
1824     mc->default_ram_id = "pc.ram";
1825     pcmc->default_smbios_ep_type = SMBIOS_ENTRY_POINT_TYPE_64;
1826 
1827     object_class_property_add(oc, PC_MACHINE_MAX_RAM_BELOW_4G, "size",
1828         pc_machine_get_max_ram_below_4g, pc_machine_set_max_ram_below_4g,
1829         NULL, NULL);
1830     object_class_property_set_description(oc, PC_MACHINE_MAX_RAM_BELOW_4G,
1831         "Maximum ram below the 4G boundary (32bit boundary)");
1832 
1833     object_class_property_add(oc, PC_MACHINE_VMPORT, "OnOffAuto",
1834         pc_machine_get_vmport, pc_machine_set_vmport,
1835         NULL, NULL);
1836     object_class_property_set_description(oc, PC_MACHINE_VMPORT,
1837         "Enable vmport (pc & q35)");
1838 
1839     object_class_property_add_bool(oc, PC_MACHINE_SMBUS,
1840         pc_machine_get_smbus, pc_machine_set_smbus);
1841     object_class_property_set_description(oc, PC_MACHINE_SMBUS,
1842         "Enable/disable system management bus");
1843 
1844     object_class_property_add_bool(oc, PC_MACHINE_SATA,
1845         pc_machine_get_sata, pc_machine_set_sata);
1846     object_class_property_set_description(oc, PC_MACHINE_SATA,
1847         "Enable/disable Serial ATA bus");
1848 
1849     object_class_property_add_bool(oc, "hpet",
1850         pc_machine_get_hpet, pc_machine_set_hpet);
1851     object_class_property_set_description(oc, "hpet",
1852         "Enable/disable high precision event timer emulation");
1853 
1854     object_class_property_add_bool(oc, PC_MACHINE_I8042,
1855         pc_machine_get_i8042, pc_machine_set_i8042);
1856 
1857     object_class_property_add_bool(oc, "default-bus-bypass-iommu",
1858         pc_machine_get_default_bus_bypass_iommu,
1859         pc_machine_set_default_bus_bypass_iommu);
1860 
1861     object_class_property_add(oc, PC_MACHINE_MAX_FW_SIZE, "size",
1862         pc_machine_get_max_fw_size, pc_machine_set_max_fw_size,
1863         NULL, NULL);
1864     object_class_property_set_description(oc, PC_MACHINE_MAX_FW_SIZE,
1865         "Maximum combined firmware size");
1866 
1867     object_class_property_add(oc, PC_MACHINE_SMBIOS_EP, "str",
1868         pc_machine_get_smbios_ep, pc_machine_set_smbios_ep,
1869         NULL, NULL);
1870     object_class_property_set_description(oc, PC_MACHINE_SMBIOS_EP,
1871         "SMBIOS Entry Point type [32, 64]");
1872 }
1873 
1874 static const TypeInfo pc_machine_info = {
1875     .name = TYPE_PC_MACHINE,
1876     .parent = TYPE_X86_MACHINE,
1877     .abstract = true,
1878     .instance_size = sizeof(PCMachineState),
1879     .instance_init = pc_machine_initfn,
1880     .class_size = sizeof(PCMachineClass),
1881     .class_init = pc_machine_class_init,
1882     .interfaces = (InterfaceInfo[]) {
1883          { TYPE_HOTPLUG_HANDLER },
1884          { }
1885     },
1886 };
1887 
1888 static void pc_machine_register_types(void)
1889 {
1890     type_register_static(&pc_machine_info);
1891 }
1892 
1893 type_init(pc_machine_register_types)
1894