xref: /openbmc/qemu/hw/i386/pc.c (revision 84aacddd)
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, "x-l1-cache-per-thread", "false" },
83     { TYPE_X86_CPU, "guest-phys-bits", "0" },
84     { "sev-guest", "legacy-vm-type", "true" },
85     { TYPE_X86_CPU, "legacy-multi-node", "on" },
86 };
87 const size_t pc_compat_9_0_len = G_N_ELEMENTS(pc_compat_9_0);
88 
89 GlobalProperty pc_compat_8_2[] = {};
90 const size_t pc_compat_8_2_len = G_N_ELEMENTS(pc_compat_8_2);
91 
92 GlobalProperty pc_compat_8_1[] = {};
93 const size_t pc_compat_8_1_len = G_N_ELEMENTS(pc_compat_8_1);
94 
95 GlobalProperty pc_compat_8_0[] = {
96     { "virtio-mem", "unplugged-inaccessible", "auto" },
97 };
98 const size_t pc_compat_8_0_len = G_N_ELEMENTS(pc_compat_8_0);
99 
100 GlobalProperty pc_compat_7_2[] = {
101     { "ICH9-LPC", "noreboot", "true" },
102 };
103 const size_t pc_compat_7_2_len = G_N_ELEMENTS(pc_compat_7_2);
104 
105 GlobalProperty pc_compat_7_1[] = {};
106 const size_t pc_compat_7_1_len = G_N_ELEMENTS(pc_compat_7_1);
107 
108 GlobalProperty pc_compat_7_0[] = {};
109 const size_t pc_compat_7_0_len = G_N_ELEMENTS(pc_compat_7_0);
110 
111 GlobalProperty pc_compat_6_2[] = {
112     { "virtio-mem", "unplugged-inaccessible", "off" },
113 };
114 const size_t pc_compat_6_2_len = G_N_ELEMENTS(pc_compat_6_2);
115 
116 GlobalProperty pc_compat_6_1[] = {
117     { TYPE_X86_CPU, "hv-version-id-build", "0x1bbc" },
118     { TYPE_X86_CPU, "hv-version-id-major", "0x0006" },
119     { TYPE_X86_CPU, "hv-version-id-minor", "0x0001" },
120     { "ICH9-LPC", "x-keep-pci-slot-hpc", "false" },
121 };
122 const size_t pc_compat_6_1_len = G_N_ELEMENTS(pc_compat_6_1);
123 
124 GlobalProperty pc_compat_6_0[] = {
125     { "qemu64" "-" TYPE_X86_CPU, "family", "6" },
126     { "qemu64" "-" TYPE_X86_CPU, "model", "6" },
127     { "qemu64" "-" TYPE_X86_CPU, "stepping", "3" },
128     { TYPE_X86_CPU, "x-vendor-cpuid-only", "off" },
129     { "ICH9-LPC", ACPI_PM_PROP_ACPI_PCIHP_BRIDGE, "off" },
130     { "ICH9-LPC", "x-keep-pci-slot-hpc", "true" },
131 };
132 const size_t pc_compat_6_0_len = G_N_ELEMENTS(pc_compat_6_0);
133 
134 GlobalProperty pc_compat_5_2[] = {
135     { "ICH9-LPC", "x-smi-cpu-hotunplug", "off" },
136 };
137 const size_t pc_compat_5_2_len = G_N_ELEMENTS(pc_compat_5_2);
138 
139 GlobalProperty pc_compat_5_1[] = {
140     { "ICH9-LPC", "x-smi-cpu-hotplug", "off" },
141     { TYPE_X86_CPU, "kvm-msi-ext-dest-id", "off" },
142 };
143 const size_t pc_compat_5_1_len = G_N_ELEMENTS(pc_compat_5_1);
144 
145 GlobalProperty pc_compat_5_0[] = {
146 };
147 const size_t pc_compat_5_0_len = G_N_ELEMENTS(pc_compat_5_0);
148 
149 GlobalProperty pc_compat_4_2[] = {
150     { "mch", "smbase-smram", "off" },
151 };
152 const size_t pc_compat_4_2_len = G_N_ELEMENTS(pc_compat_4_2);
153 
154 GlobalProperty pc_compat_4_1[] = {};
155 const size_t pc_compat_4_1_len = G_N_ELEMENTS(pc_compat_4_1);
156 
157 GlobalProperty pc_compat_4_0[] = {};
158 const size_t pc_compat_4_0_len = G_N_ELEMENTS(pc_compat_4_0);
159 
160 GlobalProperty pc_compat_3_1[] = {
161     { "intel-iommu", "dma-drain", "off" },
162     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "off" },
163     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "off" },
164     { "Opteron_G4" "-" TYPE_X86_CPU, "npt", "off" },
165     { "Opteron_G4" "-" TYPE_X86_CPU, "nrip-save", "off" },
166     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "off" },
167     { "Opteron_G5" "-" TYPE_X86_CPU, "npt", "off" },
168     { "Opteron_G5" "-" TYPE_X86_CPU, "nrip-save", "off" },
169     { "EPYC" "-" TYPE_X86_CPU, "npt", "off" },
170     { "EPYC" "-" TYPE_X86_CPU, "nrip-save", "off" },
171     { "EPYC-IBPB" "-" TYPE_X86_CPU, "npt", "off" },
172     { "EPYC-IBPB" "-" TYPE_X86_CPU, "nrip-save", "off" },
173     { "Skylake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
174     { "Skylake-Client-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
175     { "Skylake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
176     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
177     { "Cascadelake-Server" "-" TYPE_X86_CPU,  "mpx", "on" },
178     { "Icelake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
179     { "Icelake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
180     { "Cascadelake-Server" "-" TYPE_X86_CPU, "stepping", "5" },
181     { TYPE_X86_CPU, "x-intel-pt-auto-level", "off" },
182 };
183 const size_t pc_compat_3_1_len = G_N_ELEMENTS(pc_compat_3_1);
184 
185 GlobalProperty pc_compat_3_0[] = {
186     { TYPE_X86_CPU, "x-hv-synic-kvm-only", "on" },
187     { "Skylake-Server" "-" TYPE_X86_CPU, "pku", "off" },
188     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "pku", "off" },
189 };
190 const size_t pc_compat_3_0_len = G_N_ELEMENTS(pc_compat_3_0);
191 
192 GlobalProperty pc_compat_2_12[] = {
193     { TYPE_X86_CPU, "legacy-cache", "on" },
194     { TYPE_X86_CPU, "topoext", "off" },
195     { "EPYC-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
196     { "EPYC-IBPB-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
197 };
198 const size_t pc_compat_2_12_len = G_N_ELEMENTS(pc_compat_2_12);
199 
200 GlobalProperty pc_compat_2_11[] = {
201     { TYPE_X86_CPU, "x-migrate-smi-count", "off" },
202     { "Skylake-Server" "-" TYPE_X86_CPU, "clflushopt", "off" },
203 };
204 const size_t pc_compat_2_11_len = G_N_ELEMENTS(pc_compat_2_11);
205 
206 GlobalProperty pc_compat_2_10[] = {
207     { TYPE_X86_CPU, "x-hv-max-vps", "0x40" },
208     { "i440FX-pcihost", "x-pci-hole64-fix", "off" },
209     { "q35-pcihost", "x-pci-hole64-fix", "off" },
210 };
211 const size_t pc_compat_2_10_len = G_N_ELEMENTS(pc_compat_2_10);
212 
213 GlobalProperty pc_compat_2_9[] = {
214     { "mch", "extended-tseg-mbytes", "0" },
215 };
216 const size_t pc_compat_2_9_len = G_N_ELEMENTS(pc_compat_2_9);
217 
218 GlobalProperty pc_compat_2_8[] = {
219     { TYPE_X86_CPU, "tcg-cpuid", "off" },
220     { "kvmclock", "x-mach-use-reliable-get-clock", "off" },
221     { "ICH9-LPC", "x-smi-broadcast", "off" },
222     { TYPE_X86_CPU, "vmware-cpuid-freq", "off" },
223     { "Haswell-" TYPE_X86_CPU, "stepping", "1" },
224 };
225 const size_t pc_compat_2_8_len = G_N_ELEMENTS(pc_compat_2_8);
226 
227 GlobalProperty pc_compat_2_7[] = {
228     { TYPE_X86_CPU, "l3-cache", "off" },
229     { TYPE_X86_CPU, "full-cpuid-auto-level", "off" },
230     { "Opteron_G3" "-" TYPE_X86_CPU, "family", "15" },
231     { "Opteron_G3" "-" TYPE_X86_CPU, "model", "6" },
232     { "Opteron_G3" "-" TYPE_X86_CPU, "stepping", "1" },
233     { "isa-pcspk", "migrate", "off" },
234 };
235 const size_t pc_compat_2_7_len = G_N_ELEMENTS(pc_compat_2_7);
236 
237 GlobalProperty pc_compat_2_6[] = {
238     { TYPE_X86_CPU, "cpuid-0xb", "off" },
239     { "vmxnet3", "romfile", "" },
240     { TYPE_X86_CPU, "fill-mtrr-mask", "off" },
241     { "apic-common", "legacy-instance-id", "on", }
242 };
243 const size_t pc_compat_2_6_len = G_N_ELEMENTS(pc_compat_2_6);
244 
245 GlobalProperty pc_compat_2_5[] = {};
246 const size_t pc_compat_2_5_len = G_N_ELEMENTS(pc_compat_2_5);
247 
248 GlobalProperty pc_compat_2_4[] = {
249     PC_CPU_MODEL_IDS("2.4.0")
250     { "Haswell-" TYPE_X86_CPU, "abm", "off" },
251     { "Haswell-noTSX-" TYPE_X86_CPU, "abm", "off" },
252     { "Broadwell-" TYPE_X86_CPU, "abm", "off" },
253     { "Broadwell-noTSX-" TYPE_X86_CPU, "abm", "off" },
254     { "host" "-" TYPE_X86_CPU, "host-cache-info", "on" },
255     { TYPE_X86_CPU, "check", "off" },
256     { "qemu64" "-" TYPE_X86_CPU, "sse4a", "on" },
257     { "qemu64" "-" TYPE_X86_CPU, "abm", "on" },
258     { "qemu64" "-" TYPE_X86_CPU, "popcnt", "on" },
259     { "qemu32" "-" TYPE_X86_CPU, "popcnt", "on" },
260     { "Opteron_G2" "-" TYPE_X86_CPU, "rdtscp", "on" },
261     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "on" },
262     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "on" },
263     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "on", }
264 };
265 const size_t pc_compat_2_4_len = G_N_ELEMENTS(pc_compat_2_4);
266 
267 GlobalProperty pc_compat_2_3[] = {
268     PC_CPU_MODEL_IDS("2.3.0")
269     { TYPE_X86_CPU, "arat", "off" },
270     { "qemu64" "-" TYPE_X86_CPU, "min-level", "4" },
271     { "kvm64" "-" TYPE_X86_CPU, "min-level", "5" },
272     { "pentium3" "-" TYPE_X86_CPU, "min-level", "2" },
273     { "n270" "-" TYPE_X86_CPU, "min-level", "5" },
274     { "Conroe" "-" TYPE_X86_CPU, "min-level", "4" },
275     { "Penryn" "-" TYPE_X86_CPU, "min-level", "4" },
276     { "Nehalem" "-" TYPE_X86_CPU, "min-level", "4" },
277     { "n270" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
278     { "Penryn" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
279     { "Conroe" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
280     { "Nehalem" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
281     { "Westmere" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
282     { "SandyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
283     { "IvyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
284     { "Haswell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
285     { "Haswell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
286     { "Broadwell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
287     { "Broadwell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
288     { TYPE_X86_CPU, "kvm-no-smi-migration", "on" },
289 };
290 const size_t pc_compat_2_3_len = G_N_ELEMENTS(pc_compat_2_3);
291 
292 GlobalProperty pc_compat_2_2[] = {
293     PC_CPU_MODEL_IDS("2.2.0")
294     { "kvm64" "-" TYPE_X86_CPU, "vme", "off" },
295     { "kvm32" "-" TYPE_X86_CPU, "vme", "off" },
296     { "Conroe" "-" TYPE_X86_CPU, "vme", "off" },
297     { "Penryn" "-" TYPE_X86_CPU, "vme", "off" },
298     { "Nehalem" "-" TYPE_X86_CPU, "vme", "off" },
299     { "Westmere" "-" TYPE_X86_CPU, "vme", "off" },
300     { "SandyBridge" "-" TYPE_X86_CPU, "vme", "off" },
301     { "Haswell" "-" TYPE_X86_CPU, "vme", "off" },
302     { "Broadwell" "-" TYPE_X86_CPU, "vme", "off" },
303     { "Opteron_G1" "-" TYPE_X86_CPU, "vme", "off" },
304     { "Opteron_G2" "-" TYPE_X86_CPU, "vme", "off" },
305     { "Opteron_G3" "-" TYPE_X86_CPU, "vme", "off" },
306     { "Opteron_G4" "-" TYPE_X86_CPU, "vme", "off" },
307     { "Opteron_G5" "-" TYPE_X86_CPU, "vme", "off" },
308     { "Haswell" "-" TYPE_X86_CPU, "f16c", "off" },
309     { "Haswell" "-" TYPE_X86_CPU, "rdrand", "off" },
310     { "Broadwell" "-" TYPE_X86_CPU, "f16c", "off" },
311     { "Broadwell" "-" TYPE_X86_CPU, "rdrand", "off" },
312 };
313 const size_t pc_compat_2_2_len = G_N_ELEMENTS(pc_compat_2_2);
314 
315 GlobalProperty pc_compat_2_1[] = {
316     PC_CPU_MODEL_IDS("2.1.0")
317     { "coreduo" "-" TYPE_X86_CPU, "vmx", "on" },
318     { "core2duo" "-" TYPE_X86_CPU, "vmx", "on" },
319 };
320 const size_t pc_compat_2_1_len = G_N_ELEMENTS(pc_compat_2_1);
321 
322 GlobalProperty pc_compat_2_0[] = {
323     PC_CPU_MODEL_IDS("2.0.0")
324     { "virtio-scsi-pci", "any_layout", "off" },
325     { "PIIX4_PM", "memory-hotplug-support", "off" },
326     { "apic", "version", "0x11" },
327     { "nec-usb-xhci", "superspeed-ports-first", "off" },
328     { "nec-usb-xhci", "force-pcie-endcap", "on" },
329     { "pci-serial", "prog_if", "0" },
330     { "pci-serial-2x", "prog_if", "0" },
331     { "pci-serial-4x", "prog_if", "0" },
332     { "virtio-net-pci", "guest_announce", "off" },
333     { "ICH9-LPC", "memory-hotplug-support", "off" },
334 };
335 const size_t pc_compat_2_0_len = G_N_ELEMENTS(pc_compat_2_0);
336 
337 GSIState *pc_gsi_create(qemu_irq **irqs, bool pci_enabled)
338 {
339     GSIState *s;
340 
341     s = g_new0(GSIState, 1);
342     if (kvm_ioapic_in_kernel()) {
343         kvm_pc_setup_irq_routing(pci_enabled);
344     }
345     *irqs = qemu_allocate_irqs(gsi_handler, s, IOAPIC_NUM_PINS);
346 
347     return s;
348 }
349 
350 static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
351                            unsigned size)
352 {
353 }
354 
355 static uint64_t ioport80_read(void *opaque, hwaddr addr, unsigned size)
356 {
357     return 0xffffffffffffffffULL;
358 }
359 
360 /* MS-DOS compatibility mode FPU exception support */
361 static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
362                            unsigned size)
363 {
364     if (tcg_enabled()) {
365         cpu_set_ignne();
366     }
367 }
368 
369 static uint64_t ioportF0_read(void *opaque, hwaddr addr, unsigned size)
370 {
371     return 0xffffffffffffffffULL;
372 }
373 
374 /* PC cmos mappings */
375 
376 #define REG_EQUIPMENT_BYTE          0x14
377 
378 static void cmos_init_hd(MC146818RtcState *s, int type_ofs, int info_ofs,
379                          int16_t cylinders, int8_t heads, int8_t sectors)
380 {
381     mc146818rtc_set_cmos_data(s, type_ofs, 47);
382     mc146818rtc_set_cmos_data(s, info_ofs, cylinders);
383     mc146818rtc_set_cmos_data(s, info_ofs + 1, cylinders >> 8);
384     mc146818rtc_set_cmos_data(s, info_ofs + 2, heads);
385     mc146818rtc_set_cmos_data(s, info_ofs + 3, 0xff);
386     mc146818rtc_set_cmos_data(s, info_ofs + 4, 0xff);
387     mc146818rtc_set_cmos_data(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
388     mc146818rtc_set_cmos_data(s, info_ofs + 6, cylinders);
389     mc146818rtc_set_cmos_data(s, info_ofs + 7, cylinders >> 8);
390     mc146818rtc_set_cmos_data(s, info_ofs + 8, sectors);
391 }
392 
393 /* convert boot_device letter to something recognizable by the bios */
394 static int boot_device2nibble(char boot_device)
395 {
396     switch(boot_device) {
397     case 'a':
398     case 'b':
399         return 0x01; /* floppy boot */
400     case 'c':
401         return 0x02; /* hard drive boot */
402     case 'd':
403         return 0x03; /* CD-ROM boot */
404     case 'n':
405         return 0x04; /* Network boot */
406     }
407     return 0;
408 }
409 
410 static void set_boot_dev(PCMachineState *pcms, MC146818RtcState *s,
411                          const char *boot_device, Error **errp)
412 {
413 #define PC_MAX_BOOT_DEVICES 3
414     int nbds, bds[3] = { 0, };
415     int i;
416 
417     nbds = strlen(boot_device);
418     if (nbds > PC_MAX_BOOT_DEVICES) {
419         error_setg(errp, "Too many boot devices for PC");
420         return;
421     }
422     for (i = 0; i < nbds; i++) {
423         bds[i] = boot_device2nibble(boot_device[i]);
424         if (bds[i] == 0) {
425             error_setg(errp, "Invalid boot device for PC: '%c'",
426                        boot_device[i]);
427             return;
428         }
429     }
430     mc146818rtc_set_cmos_data(s, 0x3d, (bds[1] << 4) | bds[0]);
431     mc146818rtc_set_cmos_data(s, 0x38, (bds[2] << 4) | !pcms->fd_bootchk);
432 }
433 
434 static void pc_boot_set(void *opaque, const char *boot_device, Error **errp)
435 {
436     PCMachineState *pcms = opaque;
437     X86MachineState *x86ms = X86_MACHINE(pcms);
438 
439     set_boot_dev(pcms, MC146818_RTC(x86ms->rtc), boot_device, errp);
440 }
441 
442 static void pc_cmos_init_floppy(MC146818RtcState *rtc_state, ISADevice *floppy)
443 {
444     int val, nb;
445     FloppyDriveType fd_type[2] = { FLOPPY_DRIVE_TYPE_NONE,
446                                    FLOPPY_DRIVE_TYPE_NONE };
447 
448 #ifdef CONFIG_FDC_ISA
449     /* floppy type */
450     if (floppy) {
451         for (int i = 0; i < 2; i++) {
452             fd_type[i] = isa_fdc_get_drive_type(floppy, i);
453         }
454     }
455 #endif
456 
457     val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
458         cmos_get_fd_drive_type(fd_type[1]);
459     mc146818rtc_set_cmos_data(rtc_state, 0x10, val);
460 
461     val = mc146818rtc_get_cmos_data(rtc_state, REG_EQUIPMENT_BYTE);
462     nb = 0;
463     if (fd_type[0] != FLOPPY_DRIVE_TYPE_NONE) {
464         nb++;
465     }
466     if (fd_type[1] != FLOPPY_DRIVE_TYPE_NONE) {
467         nb++;
468     }
469     switch (nb) {
470     case 0:
471         break;
472     case 1:
473         val |= 0x01; /* 1 drive, ready for boot */
474         break;
475     case 2:
476         val |= 0x41; /* 2 drives, ready for boot */
477         break;
478     }
479     mc146818rtc_set_cmos_data(rtc_state, REG_EQUIPMENT_BYTE, val);
480 }
481 
482 typedef struct check_fdc_state {
483     ISADevice *floppy;
484     bool multiple;
485 } CheckFdcState;
486 
487 static int check_fdc(Object *obj, void *opaque)
488 {
489     CheckFdcState *state = opaque;
490     Object *fdc;
491     uint32_t iobase;
492     Error *local_err = NULL;
493 
494     fdc = object_dynamic_cast(obj, TYPE_ISA_FDC);
495     if (!fdc) {
496         return 0;
497     }
498 
499     iobase = object_property_get_uint(obj, "iobase", &local_err);
500     if (local_err || iobase != 0x3f0) {
501         error_free(local_err);
502         return 0;
503     }
504 
505     if (state->floppy) {
506         state->multiple = true;
507     } else {
508         state->floppy = ISA_DEVICE(obj);
509     }
510     return 0;
511 }
512 
513 static const char * const fdc_container_path[] = {
514     "/unattached", "/peripheral", "/peripheral-anon"
515 };
516 
517 /*
518  * Locate the FDC at IO address 0x3f0, in order to configure the CMOS registers
519  * and ACPI objects.
520  */
521 static ISADevice *pc_find_fdc0(void)
522 {
523     int i;
524     Object *container;
525     CheckFdcState state = { 0 };
526 
527     for (i = 0; i < ARRAY_SIZE(fdc_container_path); i++) {
528         container = container_get(qdev_get_machine(), fdc_container_path[i]);
529         object_child_foreach(container, check_fdc, &state);
530     }
531 
532     if (state.multiple) {
533         warn_report("multiple floppy disk controllers with "
534                     "iobase=0x3f0 have been found");
535         error_printf("the one being picked for CMOS setup might not reflect "
536                      "your intent");
537     }
538 
539     return state.floppy;
540 }
541 
542 static void pc_cmos_init_late(PCMachineState *pcms)
543 {
544     X86MachineState *x86ms = X86_MACHINE(pcms);
545     MC146818RtcState *s = MC146818_RTC(x86ms->rtc);
546     int16_t cylinders;
547     int8_t heads, sectors;
548     int val;
549     int i, trans;
550 
551     val = 0;
552     if (pcms->idebus[0] &&
553         ide_get_geometry(pcms->idebus[0], 0,
554                          &cylinders, &heads, &sectors) >= 0) {
555         cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
556         val |= 0xf0;
557     }
558     if (pcms->idebus[0] &&
559         ide_get_geometry(pcms->idebus[0], 1,
560                          &cylinders, &heads, &sectors) >= 0) {
561         cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
562         val |= 0x0f;
563     }
564     mc146818rtc_set_cmos_data(s, 0x12, val);
565 
566     val = 0;
567     for (i = 0; i < 4; i++) {
568         /* NOTE: ide_get_geometry() returns the physical
569            geometry.  It is always such that: 1 <= sects <= 63, 1
570            <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
571            geometry can be different if a translation is done. */
572         BusState *idebus = pcms->idebus[i / 2];
573         if (idebus &&
574             ide_get_geometry(idebus, i % 2,
575                              &cylinders, &heads, &sectors) >= 0) {
576             trans = ide_get_bios_chs_trans(idebus, i % 2) - 1;
577             assert((trans & ~3) == 0);
578             val |= trans << (i * 2);
579         }
580     }
581     mc146818rtc_set_cmos_data(s, 0x39, val);
582 
583     pc_cmos_init_floppy(s, pc_find_fdc0());
584 
585     /* various important CMOS locations needed by PC/Bochs bios */
586 
587     /* memory size */
588     /* base memory (first MiB) */
589     val = MIN(x86ms->below_4g_mem_size / KiB, 640);
590     mc146818rtc_set_cmos_data(s, 0x15, val);
591     mc146818rtc_set_cmos_data(s, 0x16, val >> 8);
592     /* extended memory (next 64MiB) */
593     if (x86ms->below_4g_mem_size > 1 * MiB) {
594         val = (x86ms->below_4g_mem_size - 1 * MiB) / KiB;
595     } else {
596         val = 0;
597     }
598     if (val > 65535)
599         val = 65535;
600     mc146818rtc_set_cmos_data(s, 0x17, val);
601     mc146818rtc_set_cmos_data(s, 0x18, val >> 8);
602     mc146818rtc_set_cmos_data(s, 0x30, val);
603     mc146818rtc_set_cmos_data(s, 0x31, val >> 8);
604     /* memory between 16MiB and 4GiB */
605     if (x86ms->below_4g_mem_size > 16 * MiB) {
606         val = (x86ms->below_4g_mem_size - 16 * MiB) / (64 * KiB);
607     } else {
608         val = 0;
609     }
610     if (val > 65535)
611         val = 65535;
612     mc146818rtc_set_cmos_data(s, 0x34, val);
613     mc146818rtc_set_cmos_data(s, 0x35, val >> 8);
614     /* memory above 4GiB */
615     val = x86ms->above_4g_mem_size / 65536;
616     mc146818rtc_set_cmos_data(s, 0x5b, val);
617     mc146818rtc_set_cmos_data(s, 0x5c, val >> 8);
618     mc146818rtc_set_cmos_data(s, 0x5d, val >> 16);
619 
620     val = 0;
621     val |= 0x02; /* FPU is there */
622     val |= 0x04; /* PS/2 mouse installed */
623     mc146818rtc_set_cmos_data(s, REG_EQUIPMENT_BYTE, val);
624 }
625 
626 static void handle_a20_line_change(void *opaque, int irq, int level)
627 {
628     X86CPU *cpu = opaque;
629 
630     /* XXX: send to all CPUs ? */
631     /* XXX: add logic to handle multiple A20 line sources */
632     x86_cpu_set_a20(cpu, level);
633 }
634 
635 #define NE2000_NB_MAX 6
636 
637 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
638                                               0x280, 0x380 };
639 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
640 
641 static gboolean pc_init_ne2k_isa(ISABus *bus, NICInfo *nd, Error **errp)
642 {
643     static int nb_ne2k = 0;
644 
645     if (nb_ne2k == NE2000_NB_MAX) {
646         error_setg(errp,
647                    "maximum number of ISA NE2000 devices exceeded");
648         return false;
649     }
650     isa_ne2000_init(bus, ne2000_io[nb_ne2k],
651                     ne2000_irq[nb_ne2k], nd);
652     nb_ne2k++;
653     return true;
654 }
655 
656 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
657 {
658     X86CPU *cpu = opaque;
659 
660     if (level) {
661         cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
662     }
663 }
664 
665 static
666 void pc_machine_done(Notifier *notifier, void *data)
667 {
668     PCMachineState *pcms = container_of(notifier,
669                                         PCMachineState, machine_done);
670     X86MachineState *x86ms = X86_MACHINE(pcms);
671 
672     cxl_hook_up_pxb_registers(pcms->pcibus, &pcms->cxl_devices_state,
673                               &error_fatal);
674 
675     if (pcms->cxl_devices_state.is_enabled) {
676         cxl_fmws_link_targets(&pcms->cxl_devices_state, &error_fatal);
677     }
678 
679     /* set the number of CPUs */
680     x86_rtc_set_cpus_count(x86ms->rtc, x86ms->boot_cpus);
681 
682     fw_cfg_add_extra_pci_roots(pcms->pcibus, x86ms->fw_cfg);
683 
684     acpi_setup();
685     if (x86ms->fw_cfg) {
686         fw_cfg_build_smbios(pcms, x86ms->fw_cfg, pcms->smbios_entry_point_type);
687         fw_cfg_build_feature_control(MACHINE(pcms), x86ms->fw_cfg);
688         /* update FW_CFG_NB_CPUS to account for -device added CPUs */
689         fw_cfg_modify_i16(x86ms->fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
690     }
691 
692     pc_cmos_init_late(pcms);
693 }
694 
695 /* setup pci memory address space mapping into system address space */
696 void pc_pci_as_mapping_init(MemoryRegion *system_memory,
697                             MemoryRegion *pci_address_space)
698 {
699     /* Set to lower priority than RAM */
700     memory_region_add_subregion_overlap(system_memory, 0x0,
701                                         pci_address_space, -1);
702 }
703 
704 void xen_load_linux(PCMachineState *pcms)
705 {
706     int i;
707     FWCfgState *fw_cfg;
708     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
709     X86MachineState *x86ms = X86_MACHINE(pcms);
710 
711     assert(MACHINE(pcms)->kernel_filename != NULL);
712 
713     fw_cfg = fw_cfg_init_io_dma(FW_CFG_IO_BASE, FW_CFG_IO_BASE + 4,
714                                 &address_space_memory);
715     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
716     rom_set_fw(fw_cfg);
717 
718     x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
719                    pcmc->pvh_enabled);
720     for (i = 0; i < nb_option_roms; i++) {
721         assert(!strcmp(option_rom[i].name, "linuxboot.bin") ||
722                !strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
723                !strcmp(option_rom[i].name, "pvh.bin") ||
724                !strcmp(option_rom[i].name, "multiboot.bin") ||
725                !strcmp(option_rom[i].name, "multiboot_dma.bin"));
726         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
727     }
728     x86ms->fw_cfg = fw_cfg;
729 }
730 
731 #define PC_ROM_MIN_VGA     0xc0000
732 #define PC_ROM_MIN_OPTION  0xc8000
733 #define PC_ROM_MAX         0xe0000
734 #define PC_ROM_ALIGN       0x800
735 #define PC_ROM_SIZE        (PC_ROM_MAX - PC_ROM_MIN_VGA)
736 
737 static hwaddr pc_above_4g_end(PCMachineState *pcms)
738 {
739     X86MachineState *x86ms = X86_MACHINE(pcms);
740 
741     if (pcms->sgx_epc.size != 0) {
742         return sgx_epc_above_4g_end(&pcms->sgx_epc);
743     }
744 
745     return x86ms->above_4g_mem_start + x86ms->above_4g_mem_size;
746 }
747 
748 static void pc_get_device_memory_range(PCMachineState *pcms,
749                                        hwaddr *base,
750                                        ram_addr_t *device_mem_size)
751 {
752     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
753     MachineState *machine = MACHINE(pcms);
754     ram_addr_t size;
755     hwaddr addr;
756 
757     size = machine->maxram_size - machine->ram_size;
758     addr = ROUND_UP(pc_above_4g_end(pcms), 1 * GiB);
759 
760     if (pcmc->enforce_aligned_dimm) {
761         /* size device region assuming 1G page max alignment per slot */
762         size += (1 * GiB) * machine->ram_slots;
763     }
764 
765     *base = addr;
766     *device_mem_size = size;
767 }
768 
769 static uint64_t pc_get_cxl_range_start(PCMachineState *pcms)
770 {
771     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
772     MachineState *ms = MACHINE(pcms);
773     hwaddr cxl_base;
774     ram_addr_t size;
775 
776     if (pcmc->has_reserved_memory &&
777         (ms->ram_size < ms->maxram_size)) {
778         pc_get_device_memory_range(pcms, &cxl_base, &size);
779         cxl_base += size;
780     } else {
781         cxl_base = pc_above_4g_end(pcms);
782     }
783 
784     return cxl_base;
785 }
786 
787 static uint64_t pc_get_cxl_range_end(PCMachineState *pcms)
788 {
789     uint64_t start = pc_get_cxl_range_start(pcms) + MiB;
790 
791     if (pcms->cxl_devices_state.fixed_windows) {
792         GList *it;
793 
794         start = ROUND_UP(start, 256 * MiB);
795         for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
796             CXLFixedWindow *fw = it->data;
797             start += fw->size;
798         }
799     }
800 
801     return start;
802 }
803 
804 static hwaddr pc_max_used_gpa(PCMachineState *pcms, uint64_t pci_hole64_size)
805 {
806     X86CPU *cpu = X86_CPU(first_cpu);
807     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
808     MachineState *ms = MACHINE(pcms);
809 
810     if (cpu->env.features[FEAT_8000_0001_EDX] & CPUID_EXT2_LM) {
811         /* 64-bit systems */
812         return pc_pci_hole64_start() + pci_hole64_size - 1;
813     }
814 
815     /* 32-bit systems */
816     if (pcmc->broken_32bit_mem_addr_check) {
817         /* old value for compatibility reasons */
818         return ((hwaddr)1 << cpu->phys_bits) - 1;
819     }
820 
821     /*
822      * 32-bit systems don't have hole64 but they might have a region for
823      * memory devices. Even if additional hotplugged memory devices might
824      * not be usable by most guest OSes, we need to still consider them for
825      * calculating the highest possible GPA so that we can properly report
826      * if someone configures them on a CPU that cannot possibly address them.
827      */
828     if (pcmc->has_reserved_memory &&
829         (ms->ram_size < ms->maxram_size)) {
830         hwaddr devmem_start;
831         ram_addr_t devmem_size;
832 
833         pc_get_device_memory_range(pcms, &devmem_start, &devmem_size);
834         devmem_start += devmem_size;
835         return devmem_start - 1;
836     }
837 
838     /* configuration without any memory hotplug */
839     return pc_above_4g_end(pcms) - 1;
840 }
841 
842 /*
843  * AMD systems with an IOMMU have an additional hole close to the
844  * 1Tb, which are special GPAs that cannot be DMA mapped. Depending
845  * on kernel version, VFIO may or may not let you DMA map those ranges.
846  * Starting Linux v5.4 we validate it, and can't create guests on AMD machines
847  * with certain memory sizes. It's also wrong to use those IOVA ranges
848  * in detriment of leading to IOMMU INVALID_DEVICE_REQUEST or worse.
849  * The ranges reserved for Hyper-Transport are:
850  *
851  * FD_0000_0000h - FF_FFFF_FFFFh
852  *
853  * The ranges represent the following:
854  *
855  * Base Address   Top Address  Use
856  *
857  * FD_0000_0000h FD_F7FF_FFFFh Reserved interrupt address space
858  * FD_F800_0000h FD_F8FF_FFFFh Interrupt/EOI IntCtl
859  * FD_F900_0000h FD_F90F_FFFFh Legacy PIC IACK
860  * FD_F910_0000h FD_F91F_FFFFh System Management
861  * FD_F920_0000h FD_FAFF_FFFFh Reserved Page Tables
862  * FD_FB00_0000h FD_FBFF_FFFFh Address Translation
863  * FD_FC00_0000h FD_FDFF_FFFFh I/O Space
864  * FD_FE00_0000h FD_FFFF_FFFFh Configuration
865  * FE_0000_0000h FE_1FFF_FFFFh Extended Configuration/Device Messages
866  * FE_2000_0000h FF_FFFF_FFFFh Reserved
867  *
868  * See AMD IOMMU spec, section 2.1.2 "IOMMU Logical Topology",
869  * Table 3: Special Address Controls (GPA) for more information.
870  */
871 #define AMD_HT_START         0xfd00000000UL
872 #define AMD_HT_END           0xffffffffffUL
873 #define AMD_ABOVE_1TB_START  (AMD_HT_END + 1)
874 #define AMD_HT_SIZE          (AMD_ABOVE_1TB_START - AMD_HT_START)
875 
876 void pc_memory_init(PCMachineState *pcms,
877                     MemoryRegion *system_memory,
878                     MemoryRegion *rom_memory,
879                     uint64_t pci_hole64_size)
880 {
881     int linux_boot, i;
882     MemoryRegion *option_rom_mr;
883     MemoryRegion *ram_below_4g, *ram_above_4g;
884     FWCfgState *fw_cfg;
885     MachineState *machine = MACHINE(pcms);
886     MachineClass *mc = MACHINE_GET_CLASS(machine);
887     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
888     X86MachineState *x86ms = X86_MACHINE(pcms);
889     hwaddr maxphysaddr, maxusedaddr;
890     hwaddr cxl_base, cxl_resv_end = 0;
891     X86CPU *cpu = X86_CPU(first_cpu);
892 
893     assert(machine->ram_size == x86ms->below_4g_mem_size +
894                                 x86ms->above_4g_mem_size);
895 
896     linux_boot = (machine->kernel_filename != NULL);
897 
898     /*
899      * The HyperTransport range close to the 1T boundary is unique to AMD
900      * hosts with IOMMUs enabled. Restrict the ram-above-4g relocation
901      * to above 1T to AMD vCPUs only. @enforce_amd_1tb_hole is only false in
902      * older machine types (<= 7.0) for compatibility purposes.
903      */
904     if (IS_AMD_CPU(&cpu->env) && pcmc->enforce_amd_1tb_hole) {
905         /* Bail out if max possible address does not cross HT range */
906         if (pc_max_used_gpa(pcms, pci_hole64_size) >= AMD_HT_START) {
907             x86ms->above_4g_mem_start = AMD_ABOVE_1TB_START;
908         }
909 
910         /*
911          * Advertise the HT region if address space covers the reserved
912          * region or if we relocate.
913          */
914         if (cpu->phys_bits >= 40) {
915             e820_add_entry(AMD_HT_START, AMD_HT_SIZE, E820_RESERVED);
916         }
917     }
918 
919     /*
920      * phys-bits is required to be appropriately configured
921      * to make sure max used GPA is reachable.
922      */
923     maxusedaddr = pc_max_used_gpa(pcms, pci_hole64_size);
924     maxphysaddr = ((hwaddr)1 << cpu->phys_bits) - 1;
925     if (maxphysaddr < maxusedaddr) {
926         error_report("Address space limit 0x%"PRIx64" < 0x%"PRIx64
927                      " phys-bits too low (%u)",
928                      maxphysaddr, maxusedaddr, cpu->phys_bits);
929         exit(EXIT_FAILURE);
930     }
931 
932     /*
933      * Split single memory region and use aliases to address portions of it,
934      * done for backwards compatibility with older qemus.
935      */
936     ram_below_4g = g_malloc(sizeof(*ram_below_4g));
937     memory_region_init_alias(ram_below_4g, NULL, "ram-below-4g", machine->ram,
938                              0, x86ms->below_4g_mem_size);
939     memory_region_add_subregion(system_memory, 0, ram_below_4g);
940     e820_add_entry(0, x86ms->below_4g_mem_size, E820_RAM);
941     if (x86ms->above_4g_mem_size > 0) {
942         ram_above_4g = g_malloc(sizeof(*ram_above_4g));
943         memory_region_init_alias(ram_above_4g, NULL, "ram-above-4g",
944                                  machine->ram,
945                                  x86ms->below_4g_mem_size,
946                                  x86ms->above_4g_mem_size);
947         memory_region_add_subregion(system_memory, x86ms->above_4g_mem_start,
948                                     ram_above_4g);
949         e820_add_entry(x86ms->above_4g_mem_start, x86ms->above_4g_mem_size,
950                        E820_RAM);
951     }
952 
953     if (pcms->sgx_epc.size != 0) {
954         e820_add_entry(pcms->sgx_epc.base, pcms->sgx_epc.size, E820_RESERVED);
955     }
956 
957     if (!pcmc->has_reserved_memory &&
958         (machine->ram_slots ||
959          (machine->maxram_size > machine->ram_size))) {
960 
961         error_report("\"-memory 'slots|maxmem'\" is not supported by: %s",
962                      mc->name);
963         exit(EXIT_FAILURE);
964     }
965 
966     /* initialize device memory address space */
967     if (pcmc->has_reserved_memory &&
968         (machine->ram_size < machine->maxram_size)) {
969         ram_addr_t device_mem_size;
970         hwaddr device_mem_base;
971 
972         if (machine->ram_slots > ACPI_MAX_RAM_SLOTS) {
973             error_report("unsupported amount of memory slots: %"PRIu64,
974                          machine->ram_slots);
975             exit(EXIT_FAILURE);
976         }
977 
978         if (QEMU_ALIGN_UP(machine->maxram_size,
979                           TARGET_PAGE_SIZE) != machine->maxram_size) {
980             error_report("maximum memory size must by aligned to multiple of "
981                          "%d bytes", TARGET_PAGE_SIZE);
982             exit(EXIT_FAILURE);
983         }
984 
985         pc_get_device_memory_range(pcms, &device_mem_base, &device_mem_size);
986 
987         if (device_mem_base + device_mem_size < device_mem_size) {
988             error_report("unsupported amount of maximum memory: " RAM_ADDR_FMT,
989                          machine->maxram_size);
990             exit(EXIT_FAILURE);
991         }
992         machine_memory_devices_init(machine, device_mem_base, device_mem_size);
993     }
994 
995     if (pcms->cxl_devices_state.is_enabled) {
996         MemoryRegion *mr = &pcms->cxl_devices_state.host_mr;
997         hwaddr cxl_size = MiB;
998 
999         cxl_base = pc_get_cxl_range_start(pcms);
1000         memory_region_init(mr, OBJECT(machine), "cxl_host_reg", cxl_size);
1001         memory_region_add_subregion(system_memory, cxl_base, mr);
1002         cxl_resv_end = cxl_base + cxl_size;
1003         if (pcms->cxl_devices_state.fixed_windows) {
1004             hwaddr cxl_fmw_base;
1005             GList *it;
1006 
1007             cxl_fmw_base = ROUND_UP(cxl_base + cxl_size, 256 * MiB);
1008             for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
1009                 CXLFixedWindow *fw = it->data;
1010 
1011                 fw->base = cxl_fmw_base;
1012                 memory_region_init_io(&fw->mr, OBJECT(machine), &cfmws_ops, fw,
1013                                       "cxl-fixed-memory-region", fw->size);
1014                 memory_region_add_subregion(system_memory, fw->base, &fw->mr);
1015                 cxl_fmw_base += fw->size;
1016                 cxl_resv_end = cxl_fmw_base;
1017             }
1018         }
1019     }
1020 
1021     /* Initialize PC system firmware */
1022     pc_system_firmware_init(pcms, rom_memory);
1023 
1024     option_rom_mr = g_malloc(sizeof(*option_rom_mr));
1025     memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE,
1026                            &error_fatal);
1027     if (pcmc->pci_enabled) {
1028         memory_region_set_readonly(option_rom_mr, true);
1029     }
1030     memory_region_add_subregion_overlap(rom_memory,
1031                                         PC_ROM_MIN_VGA,
1032                                         option_rom_mr,
1033                                         1);
1034 
1035     fw_cfg = fw_cfg_arch_create(machine,
1036                                 x86ms->boot_cpus, x86ms->apic_id_limit);
1037 
1038     rom_set_fw(fw_cfg);
1039 
1040     if (machine->device_memory) {
1041         uint64_t *val = g_malloc(sizeof(*val));
1042         uint64_t res_mem_end = machine->device_memory->base;
1043 
1044         if (!pcmc->broken_reserved_end) {
1045             res_mem_end += memory_region_size(&machine->device_memory->mr);
1046         }
1047 
1048         if (pcms->cxl_devices_state.is_enabled) {
1049             res_mem_end = cxl_resv_end;
1050         }
1051         *val = cpu_to_le64(ROUND_UP(res_mem_end, 1 * GiB));
1052         fw_cfg_add_file(fw_cfg, "etc/reserved-memory-end", val, sizeof(*val));
1053     }
1054 
1055     if (linux_boot) {
1056         x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
1057                        pcmc->pvh_enabled);
1058     }
1059 
1060     for (i = 0; i < nb_option_roms; i++) {
1061         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1062     }
1063     x86ms->fw_cfg = fw_cfg;
1064 
1065     /* Init default IOAPIC address space */
1066     x86ms->ioapic_as = &address_space_memory;
1067 
1068     /* Init ACPI memory hotplug IO base address */
1069     pcms->memhp_io_base = ACPI_MEMORY_HOTPLUG_BASE;
1070 }
1071 
1072 /*
1073  * The 64bit pci hole starts after "above 4G RAM" and
1074  * potentially the space reserved for memory hotplug.
1075  */
1076 uint64_t pc_pci_hole64_start(void)
1077 {
1078     PCMachineState *pcms = PC_MACHINE(qdev_get_machine());
1079     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1080     MachineState *ms = MACHINE(pcms);
1081     uint64_t hole64_start = 0;
1082     ram_addr_t size = 0;
1083 
1084     if (pcms->cxl_devices_state.is_enabled) {
1085         hole64_start = pc_get_cxl_range_end(pcms);
1086     } else if (pcmc->has_reserved_memory && (ms->ram_size < ms->maxram_size)) {
1087         pc_get_device_memory_range(pcms, &hole64_start, &size);
1088         if (!pcmc->broken_reserved_end) {
1089             hole64_start += size;
1090         }
1091     } else {
1092         hole64_start = pc_above_4g_end(pcms);
1093     }
1094 
1095     return ROUND_UP(hole64_start, 1 * GiB);
1096 }
1097 
1098 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1099 {
1100     DeviceState *dev = NULL;
1101 
1102     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_VGA);
1103     if (pci_bus) {
1104         PCIDevice *pcidev = pci_vga_init(pci_bus);
1105         dev = pcidev ? &pcidev->qdev : NULL;
1106     } else if (isa_bus) {
1107         ISADevice *isadev = isa_vga_init(isa_bus);
1108         dev = isadev ? DEVICE(isadev) : NULL;
1109     }
1110     rom_reset_order_override();
1111     return dev;
1112 }
1113 
1114 static const MemoryRegionOps ioport80_io_ops = {
1115     .write = ioport80_write,
1116     .read = ioport80_read,
1117     .endianness = DEVICE_NATIVE_ENDIAN,
1118     .impl = {
1119         .min_access_size = 1,
1120         .max_access_size = 1,
1121     },
1122 };
1123 
1124 static const MemoryRegionOps ioportF0_io_ops = {
1125     .write = ioportF0_write,
1126     .read = ioportF0_read,
1127     .endianness = DEVICE_NATIVE_ENDIAN,
1128     .impl = {
1129         .min_access_size = 1,
1130         .max_access_size = 1,
1131     },
1132 };
1133 
1134 static void pc_superio_init(ISABus *isa_bus, bool create_fdctrl,
1135                             bool create_i8042, bool no_vmport)
1136 {
1137     int i;
1138     DriveInfo *fd[MAX_FD];
1139     qemu_irq *a20_line;
1140     ISADevice *i8042, *port92, *vmmouse;
1141 
1142     serial_hds_isa_init(isa_bus, 0, MAX_ISA_SERIAL_PORTS);
1143     parallel_hds_isa_init(isa_bus, MAX_PARALLEL_PORTS);
1144 
1145     for (i = 0; i < MAX_FD; i++) {
1146         fd[i] = drive_get(IF_FLOPPY, 0, i);
1147         create_fdctrl |= !!fd[i];
1148     }
1149     if (create_fdctrl) {
1150 #ifdef CONFIG_FDC_ISA
1151         ISADevice *fdc = isa_new(TYPE_ISA_FDC);
1152         if (fdc) {
1153             isa_realize_and_unref(fdc, isa_bus, &error_fatal);
1154             isa_fdc_init_drives(fdc, fd);
1155         }
1156 #endif
1157     }
1158 
1159     if (!create_i8042) {
1160         return;
1161     }
1162 
1163     i8042 = isa_create_simple(isa_bus, TYPE_I8042);
1164     if (!no_vmport) {
1165         isa_create_simple(isa_bus, TYPE_VMPORT);
1166         vmmouse = isa_try_new("vmmouse");
1167     } else {
1168         vmmouse = NULL;
1169     }
1170     if (vmmouse) {
1171         object_property_set_link(OBJECT(vmmouse), TYPE_I8042, OBJECT(i8042),
1172                                  &error_abort);
1173         isa_realize_and_unref(vmmouse, isa_bus, &error_fatal);
1174     }
1175     port92 = isa_create_simple(isa_bus, TYPE_PORT92);
1176 
1177     a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1178     qdev_connect_gpio_out_named(DEVICE(i8042),
1179                                 I8042_A20_LINE, 0, a20_line[0]);
1180     qdev_connect_gpio_out_named(DEVICE(port92),
1181                                 PORT92_A20_LINE, 0, a20_line[1]);
1182     g_free(a20_line);
1183 }
1184 
1185 void pc_basic_device_init(struct PCMachineState *pcms,
1186                           ISABus *isa_bus, qemu_irq *gsi,
1187                           ISADevice *rtc_state,
1188                           bool create_fdctrl,
1189                           uint32_t hpet_irqs)
1190 {
1191     int i;
1192     DeviceState *hpet = NULL;
1193     int pit_isa_irq = 0;
1194     qemu_irq pit_alt_irq = NULL;
1195     ISADevice *pit = NULL;
1196     MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1197     MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
1198     X86MachineState *x86ms = X86_MACHINE(pcms);
1199 
1200     memory_region_init_io(ioport80_io, NULL, &ioport80_io_ops, NULL, "ioport80", 1);
1201     memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
1202 
1203     memory_region_init_io(ioportF0_io, NULL, &ioportF0_io_ops, NULL, "ioportF0", 1);
1204     memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
1205 
1206     /*
1207      * Check if an HPET shall be created.
1208      */
1209     if (pcms->hpet_enabled) {
1210         qemu_irq rtc_irq;
1211 
1212         hpet = qdev_try_new(TYPE_HPET);
1213         if (!hpet) {
1214             error_report("couldn't create HPET device");
1215             exit(1);
1216         }
1217         /*
1218          * For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-*,
1219          * use IRQ16~23, IRQ8 and IRQ2.  If the user has already set
1220          * the property, use whatever mask they specified.
1221          */
1222         uint8_t compat = object_property_get_uint(OBJECT(hpet),
1223                 HPET_INTCAP, NULL);
1224         if (!compat) {
1225             qdev_prop_set_uint32(hpet, HPET_INTCAP, hpet_irqs);
1226         }
1227         sysbus_realize_and_unref(SYS_BUS_DEVICE(hpet), &error_fatal);
1228         sysbus_mmio_map(SYS_BUS_DEVICE(hpet), 0, HPET_BASE);
1229 
1230         for (i = 0; i < IOAPIC_NUM_PINS; i++) {
1231             sysbus_connect_irq(SYS_BUS_DEVICE(hpet), i, gsi[i]);
1232         }
1233         pit_isa_irq = -1;
1234         pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1235         rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1236 
1237         /* overwrite connection created by south bridge */
1238         qdev_connect_gpio_out(DEVICE(rtc_state), 0, rtc_irq);
1239     }
1240 
1241     object_property_add_alias(OBJECT(pcms), "rtc-time", OBJECT(rtc_state),
1242                               "date");
1243 
1244 #ifdef CONFIG_XEN_EMU
1245     if (xen_mode == XEN_EMULATE) {
1246         xen_overlay_create();
1247         xen_evtchn_create(IOAPIC_NUM_PINS, gsi);
1248         xen_gnttab_create();
1249         xen_xenstore_create();
1250         if (pcms->pcibus) {
1251             pci_create_simple(pcms->pcibus, -1, "xen-platform");
1252         }
1253         xen_bus_init();
1254     }
1255 #endif
1256 
1257     qemu_register_boot_set(pc_boot_set, pcms);
1258     set_boot_dev(pcms, MC146818_RTC(rtc_state),
1259                  MACHINE(pcms)->boot_config.order, &error_fatal);
1260 
1261     if (!xen_enabled() &&
1262         (x86ms->pit == ON_OFF_AUTO_AUTO || x86ms->pit == ON_OFF_AUTO_ON)) {
1263         if (kvm_pit_in_kernel()) {
1264             pit = kvm_pit_init(isa_bus, 0x40);
1265         } else {
1266             pit = i8254_pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1267         }
1268         if (hpet) {
1269             /* connect PIT to output control line of the HPET */
1270             qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(DEVICE(pit), 0));
1271         }
1272         object_property_set_link(OBJECT(pcms->pcspk), "pit",
1273                                  OBJECT(pit), &error_fatal);
1274         isa_realize_and_unref(pcms->pcspk, isa_bus, &error_fatal);
1275     }
1276 
1277     /* Super I/O */
1278     pc_superio_init(isa_bus, create_fdctrl, pcms->i8042_enabled,
1279                     pcms->vmport != ON_OFF_AUTO_ON);
1280 }
1281 
1282 void pc_nic_init(PCMachineClass *pcmc, ISABus *isa_bus, PCIBus *pci_bus)
1283 {
1284     MachineClass *mc = MACHINE_CLASS(pcmc);
1285     bool default_is_ne2k = g_str_equal(mc->default_nic, TYPE_ISA_NE2000);
1286     NICInfo *nd;
1287 
1288     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_NIC);
1289 
1290     while ((nd = qemu_find_nic_info(TYPE_ISA_NE2000, default_is_ne2k, NULL))) {
1291         pc_init_ne2k_isa(isa_bus, nd, &error_fatal);
1292     }
1293 
1294     /* Anything remaining should be a PCI NIC */
1295     pci_init_nic_devices(pci_bus, mc->default_nic);
1296 
1297     rom_reset_order_override();
1298 }
1299 
1300 void pc_i8259_create(ISABus *isa_bus, qemu_irq *i8259_irqs)
1301 {
1302     qemu_irq *i8259;
1303 
1304     if (kvm_pic_in_kernel()) {
1305         i8259 = kvm_i8259_init(isa_bus);
1306     } else if (xen_enabled()) {
1307         i8259 = xen_interrupt_controller_init();
1308     } else {
1309         i8259 = i8259_init(isa_bus, x86_allocate_cpu_irq());
1310     }
1311 
1312     for (size_t i = 0; i < ISA_NUM_IRQS; i++) {
1313         i8259_irqs[i] = i8259[i];
1314     }
1315 
1316     g_free(i8259);
1317 }
1318 
1319 static void pc_memory_pre_plug(HotplugHandler *hotplug_dev, DeviceState *dev,
1320                                Error **errp)
1321 {
1322     const PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1323     const X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1324     const PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1325     const MachineState *ms = MACHINE(hotplug_dev);
1326     const bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1327     const uint64_t legacy_align = TARGET_PAGE_SIZE;
1328     Error *local_err = NULL;
1329 
1330     /*
1331      * When "acpi=off" is used with the Q35 machine type, no ACPI is built,
1332      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1333      * addition to cover this case.
1334      */
1335     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1336         error_setg(errp,
1337                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1338         return;
1339     }
1340 
1341     if (is_nvdimm && !ms->nvdimms_state->is_enabled) {
1342         error_setg(errp, "nvdimm is not enabled: missing 'nvdimm' in '-M'");
1343         return;
1344     }
1345 
1346     hotplug_handler_pre_plug(x86ms->acpi_dev, dev, &local_err);
1347     if (local_err) {
1348         error_propagate(errp, local_err);
1349         return;
1350     }
1351 
1352     pc_dimm_pre_plug(PC_DIMM(dev), MACHINE(hotplug_dev),
1353                      pcmc->enforce_aligned_dimm ? NULL : &legacy_align, errp);
1354 }
1355 
1356 static void pc_memory_plug(HotplugHandler *hotplug_dev,
1357                            DeviceState *dev, Error **errp)
1358 {
1359     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1360     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1361     MachineState *ms = MACHINE(hotplug_dev);
1362     bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1363 
1364     pc_dimm_plug(PC_DIMM(dev), MACHINE(pcms));
1365 
1366     if (is_nvdimm) {
1367         nvdimm_plug(ms->nvdimms_state);
1368     }
1369 
1370     hotplug_handler_plug(x86ms->acpi_dev, dev, &error_abort);
1371 }
1372 
1373 static void pc_memory_unplug_request(HotplugHandler *hotplug_dev,
1374                                      DeviceState *dev, Error **errp)
1375 {
1376     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1377 
1378     /*
1379      * When "acpi=off" is used with the Q35 machine type, no ACPI is built,
1380      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1381      * addition to cover this case.
1382      */
1383     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1384         error_setg(errp,
1385                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1386         return;
1387     }
1388 
1389     if (object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM)) {
1390         error_setg(errp, "nvdimm device hot unplug is not supported yet.");
1391         return;
1392     }
1393 
1394     hotplug_handler_unplug_request(x86ms->acpi_dev, dev,
1395                                    errp);
1396 }
1397 
1398 static void pc_memory_unplug(HotplugHandler *hotplug_dev,
1399                              DeviceState *dev, Error **errp)
1400 {
1401     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1402     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1403     Error *local_err = NULL;
1404 
1405     hotplug_handler_unplug(x86ms->acpi_dev, dev, &local_err);
1406     if (local_err) {
1407         goto out;
1408     }
1409 
1410     pc_dimm_unplug(PC_DIMM(dev), MACHINE(pcms));
1411     qdev_unrealize(dev);
1412  out:
1413     error_propagate(errp, local_err);
1414 }
1415 
1416 static void pc_hv_balloon_pre_plug(HotplugHandler *hotplug_dev,
1417                                    DeviceState *dev, Error **errp)
1418 {
1419     /* The vmbus handler has no hotplug handler; we should never end up here. */
1420     g_assert(!dev->hotplugged);
1421     memory_device_pre_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev), NULL,
1422                            errp);
1423 }
1424 
1425 static void pc_hv_balloon_plug(HotplugHandler *hotplug_dev,
1426                                DeviceState *dev, Error **errp)
1427 {
1428     memory_device_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev));
1429 }
1430 
1431 static void pc_machine_device_pre_plug_cb(HotplugHandler *hotplug_dev,
1432                                           DeviceState *dev, Error **errp)
1433 {
1434     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1435         pc_memory_pre_plug(hotplug_dev, dev, errp);
1436     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1437         x86_cpu_pre_plug(hotplug_dev, dev, errp);
1438     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1439         virtio_md_pci_pre_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1440     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1441         /* Declare the APIC range as the reserved MSI region */
1442         char *resv_prop_str = g_strdup_printf("0xfee00000:0xfeefffff:%d",
1443                                               VIRTIO_IOMMU_RESV_MEM_T_MSI);
1444         QList *reserved_regions = qlist_new();
1445 
1446         qlist_append_str(reserved_regions, resv_prop_str);
1447         qdev_prop_set_array(dev, "reserved-regions", reserved_regions);
1448 
1449         g_free(resv_prop_str);
1450     }
1451 
1452     if (object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE) ||
1453         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1454         PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1455 
1456         if (pcms->iommu) {
1457             error_setg(errp, "QEMU does not support multiple vIOMMUs "
1458                        "for x86 yet.");
1459             return;
1460         }
1461         pcms->iommu = dev;
1462     } else if (object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON)) {
1463         pc_hv_balloon_pre_plug(hotplug_dev, dev, errp);
1464     }
1465 }
1466 
1467 static void pc_machine_device_plug_cb(HotplugHandler *hotplug_dev,
1468                                       DeviceState *dev, Error **errp)
1469 {
1470     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1471         pc_memory_plug(hotplug_dev, dev, errp);
1472     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1473         x86_cpu_plug(hotplug_dev, dev, errp);
1474     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1475         virtio_md_pci_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1476     } else if (object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON)) {
1477         pc_hv_balloon_plug(hotplug_dev, dev, errp);
1478     }
1479 }
1480 
1481 static void pc_machine_device_unplug_request_cb(HotplugHandler *hotplug_dev,
1482                                                 DeviceState *dev, Error **errp)
1483 {
1484     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1485         pc_memory_unplug_request(hotplug_dev, dev, errp);
1486     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1487         x86_cpu_unplug_request_cb(hotplug_dev, dev, errp);
1488     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1489         virtio_md_pci_unplug_request(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev),
1490                                      errp);
1491     } else {
1492         error_setg(errp, "acpi: device unplug request for not supported device"
1493                    " type: %s", object_get_typename(OBJECT(dev)));
1494     }
1495 }
1496 
1497 static void pc_machine_device_unplug_cb(HotplugHandler *hotplug_dev,
1498                                         DeviceState *dev, Error **errp)
1499 {
1500     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1501         pc_memory_unplug(hotplug_dev, dev, errp);
1502     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1503         x86_cpu_unplug_cb(hotplug_dev, dev, errp);
1504     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1505         virtio_md_pci_unplug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1506     } else {
1507         error_setg(errp, "acpi: device unplug for not supported device"
1508                    " type: %s", object_get_typename(OBJECT(dev)));
1509     }
1510 }
1511 
1512 static HotplugHandler *pc_get_hotplug_handler(MachineState *machine,
1513                                              DeviceState *dev)
1514 {
1515     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM) ||
1516         object_dynamic_cast(OBJECT(dev), TYPE_CPU) ||
1517         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI) ||
1518         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI) ||
1519         object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON) ||
1520         object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE)) {
1521         return HOTPLUG_HANDLER(machine);
1522     }
1523 
1524     return NULL;
1525 }
1526 
1527 static void pc_machine_get_vmport(Object *obj, Visitor *v, const char *name,
1528                                   void *opaque, Error **errp)
1529 {
1530     PCMachineState *pcms = PC_MACHINE(obj);
1531     OnOffAuto vmport = pcms->vmport;
1532 
1533     visit_type_OnOffAuto(v, name, &vmport, errp);
1534 }
1535 
1536 static void pc_machine_set_vmport(Object *obj, Visitor *v, const char *name,
1537                                   void *opaque, Error **errp)
1538 {
1539     PCMachineState *pcms = PC_MACHINE(obj);
1540 
1541     visit_type_OnOffAuto(v, name, &pcms->vmport, errp);
1542 }
1543 
1544 static bool pc_machine_get_fd_bootchk(Object *obj, Error **errp)
1545 {
1546     PCMachineState *pcms = PC_MACHINE(obj);
1547 
1548     return pcms->fd_bootchk;
1549 }
1550 
1551 static void pc_machine_set_fd_bootchk(Object *obj, bool value, Error **errp)
1552 {
1553     PCMachineState *pcms = PC_MACHINE(obj);
1554 
1555     pcms->fd_bootchk = value;
1556 }
1557 
1558 static bool pc_machine_get_smbus(Object *obj, Error **errp)
1559 {
1560     PCMachineState *pcms = PC_MACHINE(obj);
1561 
1562     return pcms->smbus_enabled;
1563 }
1564 
1565 static void pc_machine_set_smbus(Object *obj, bool value, Error **errp)
1566 {
1567     PCMachineState *pcms = PC_MACHINE(obj);
1568 
1569     pcms->smbus_enabled = value;
1570 }
1571 
1572 static bool pc_machine_get_sata(Object *obj, Error **errp)
1573 {
1574     PCMachineState *pcms = PC_MACHINE(obj);
1575 
1576     return pcms->sata_enabled;
1577 }
1578 
1579 static void pc_machine_set_sata(Object *obj, bool value, Error **errp)
1580 {
1581     PCMachineState *pcms = PC_MACHINE(obj);
1582 
1583     pcms->sata_enabled = value;
1584 }
1585 
1586 static bool pc_machine_get_hpet(Object *obj, Error **errp)
1587 {
1588     PCMachineState *pcms = PC_MACHINE(obj);
1589 
1590     return pcms->hpet_enabled;
1591 }
1592 
1593 static void pc_machine_set_hpet(Object *obj, bool value, Error **errp)
1594 {
1595     PCMachineState *pcms = PC_MACHINE(obj);
1596 
1597     pcms->hpet_enabled = value;
1598 }
1599 
1600 static bool pc_machine_get_i8042(Object *obj, Error **errp)
1601 {
1602     PCMachineState *pcms = PC_MACHINE(obj);
1603 
1604     return pcms->i8042_enabled;
1605 }
1606 
1607 static void pc_machine_set_i8042(Object *obj, bool value, Error **errp)
1608 {
1609     PCMachineState *pcms = PC_MACHINE(obj);
1610 
1611     pcms->i8042_enabled = value;
1612 }
1613 
1614 static bool pc_machine_get_default_bus_bypass_iommu(Object *obj, Error **errp)
1615 {
1616     PCMachineState *pcms = PC_MACHINE(obj);
1617 
1618     return pcms->default_bus_bypass_iommu;
1619 }
1620 
1621 static void pc_machine_set_default_bus_bypass_iommu(Object *obj, bool value,
1622                                                     Error **errp)
1623 {
1624     PCMachineState *pcms = PC_MACHINE(obj);
1625 
1626     pcms->default_bus_bypass_iommu = value;
1627 }
1628 
1629 static void pc_machine_get_smbios_ep(Object *obj, Visitor *v, const char *name,
1630                                      void *opaque, Error **errp)
1631 {
1632     PCMachineState *pcms = PC_MACHINE(obj);
1633     SmbiosEntryPointType smbios_entry_point_type = pcms->smbios_entry_point_type;
1634 
1635     visit_type_SmbiosEntryPointType(v, name, &smbios_entry_point_type, errp);
1636 }
1637 
1638 static void pc_machine_set_smbios_ep(Object *obj, Visitor *v, const char *name,
1639                                      void *opaque, Error **errp)
1640 {
1641     PCMachineState *pcms = PC_MACHINE(obj);
1642 
1643     visit_type_SmbiosEntryPointType(v, name, &pcms->smbios_entry_point_type, errp);
1644 }
1645 
1646 static void pc_machine_get_max_ram_below_4g(Object *obj, Visitor *v,
1647                                             const char *name, void *opaque,
1648                                             Error **errp)
1649 {
1650     PCMachineState *pcms = PC_MACHINE(obj);
1651     uint64_t value = pcms->max_ram_below_4g;
1652 
1653     visit_type_size(v, name, &value, errp);
1654 }
1655 
1656 static void pc_machine_set_max_ram_below_4g(Object *obj, Visitor *v,
1657                                             const char *name, void *opaque,
1658                                             Error **errp)
1659 {
1660     PCMachineState *pcms = PC_MACHINE(obj);
1661     uint64_t value;
1662 
1663     if (!visit_type_size(v, name, &value, errp)) {
1664         return;
1665     }
1666     if (value > 4 * GiB) {
1667         error_setg(errp,
1668                    "Machine option 'max-ram-below-4g=%"PRIu64
1669                    "' expects size less than or equal to 4G", value);
1670         return;
1671     }
1672 
1673     if (value < 1 * MiB) {
1674         warn_report("Only %" PRIu64 " bytes of RAM below the 4GiB boundary,"
1675                     "BIOS may not work with less than 1MiB", value);
1676     }
1677 
1678     pcms->max_ram_below_4g = value;
1679 }
1680 
1681 static void pc_machine_get_max_fw_size(Object *obj, Visitor *v,
1682                                        const char *name, void *opaque,
1683                                        Error **errp)
1684 {
1685     PCMachineState *pcms = PC_MACHINE(obj);
1686     uint64_t value = pcms->max_fw_size;
1687 
1688     visit_type_size(v, name, &value, errp);
1689 }
1690 
1691 static void pc_machine_set_max_fw_size(Object *obj, Visitor *v,
1692                                        const char *name, void *opaque,
1693                                        Error **errp)
1694 {
1695     PCMachineState *pcms = PC_MACHINE(obj);
1696     uint64_t value;
1697 
1698     if (!visit_type_size(v, name, &value, errp)) {
1699         return;
1700     }
1701 
1702     /*
1703      * We don't have a theoretically justifiable exact lower bound on the base
1704      * address of any flash mapping. In practice, the IO-APIC MMIO range is
1705      * [0xFEE00000..0xFEE01000] -- see IO_APIC_DEFAULT_ADDRESS --, leaving free
1706      * only 18MiB-4KiB below 4GiB. For now, restrict the cumulative mapping to
1707      * 16MiB in size.
1708      */
1709     if (value > 16 * MiB) {
1710         error_setg(errp,
1711                    "User specified max allowed firmware size %" PRIu64 " is "
1712                    "greater than 16MiB. If combined firmware size exceeds "
1713                    "16MiB the system may not boot, or experience intermittent"
1714                    "stability issues.",
1715                    value);
1716         return;
1717     }
1718 
1719     pcms->max_fw_size = value;
1720 }
1721 
1722 
1723 static void pc_machine_initfn(Object *obj)
1724 {
1725     PCMachineState *pcms = PC_MACHINE(obj);
1726     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1727 
1728 #ifdef CONFIG_VMPORT
1729     pcms->vmport = ON_OFF_AUTO_AUTO;
1730 #else
1731     pcms->vmport = ON_OFF_AUTO_OFF;
1732 #endif /* CONFIG_VMPORT */
1733     pcms->max_ram_below_4g = 0; /* use default */
1734     pcms->smbios_entry_point_type = pcmc->default_smbios_ep_type;
1735     pcms->south_bridge = pcmc->default_south_bridge;
1736 
1737     /* acpi build is enabled by default if machine supports it */
1738     pcms->acpi_build_enabled = pcmc->has_acpi_build;
1739     pcms->smbus_enabled = true;
1740     pcms->sata_enabled = true;
1741     pcms->i8042_enabled = true;
1742     pcms->max_fw_size = 8 * MiB;
1743 #ifdef CONFIG_HPET
1744     pcms->hpet_enabled = true;
1745 #endif
1746     pcms->fd_bootchk = true;
1747     pcms->default_bus_bypass_iommu = false;
1748 
1749     pc_system_flash_create(pcms);
1750     pcms->pcspk = isa_new(TYPE_PC_SPEAKER);
1751     object_property_add_alias(OBJECT(pcms), "pcspk-audiodev",
1752                               OBJECT(pcms->pcspk), "audiodev");
1753     if (pcmc->pci_enabled) {
1754         cxl_machine_init(obj, &pcms->cxl_devices_state);
1755     }
1756 
1757     pcms->machine_done.notify = pc_machine_done;
1758     qemu_add_machine_init_done_notifier(&pcms->machine_done);
1759 }
1760 
1761 static void pc_machine_reset(MachineState *machine, ShutdownCause reason)
1762 {
1763     CPUState *cs;
1764     X86CPU *cpu;
1765 
1766     qemu_devices_reset(reason);
1767 
1768     /* Reset APIC after devices have been reset to cancel
1769      * any changes that qemu_devices_reset() might have done.
1770      */
1771     CPU_FOREACH(cs) {
1772         cpu = X86_CPU(cs);
1773 
1774         x86_cpu_after_reset(cpu);
1775     }
1776 }
1777 
1778 static void pc_machine_wakeup(MachineState *machine)
1779 {
1780     cpu_synchronize_all_states();
1781     pc_machine_reset(machine, SHUTDOWN_CAUSE_NONE);
1782     cpu_synchronize_all_post_reset();
1783 }
1784 
1785 static bool pc_hotplug_allowed(MachineState *ms, DeviceState *dev, Error **errp)
1786 {
1787     X86IOMMUState *iommu = x86_iommu_get_default();
1788     IntelIOMMUState *intel_iommu;
1789 
1790     if (iommu &&
1791         object_dynamic_cast((Object *)iommu, TYPE_INTEL_IOMMU_DEVICE) &&
1792         object_dynamic_cast((Object *)dev, "vfio-pci")) {
1793         intel_iommu = INTEL_IOMMU_DEVICE(iommu);
1794         if (!intel_iommu->caching_mode) {
1795             error_setg(errp, "Device assignment is not allowed without "
1796                        "enabling caching-mode=on for Intel IOMMU.");
1797             return false;
1798         }
1799     }
1800 
1801     return true;
1802 }
1803 
1804 static void pc_machine_class_init(ObjectClass *oc, void *data)
1805 {
1806     MachineClass *mc = MACHINE_CLASS(oc);
1807     X86MachineClass *x86mc = X86_MACHINE_CLASS(oc);
1808     PCMachineClass *pcmc = PC_MACHINE_CLASS(oc);
1809     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
1810 
1811     pcmc->pci_enabled = true;
1812     pcmc->has_acpi_build = true;
1813     pcmc->rsdp_in_ram = true;
1814     pcmc->smbios_defaults = true;
1815     pcmc->smbios_uuid_encoded = true;
1816     pcmc->gigabyte_align = true;
1817     pcmc->has_reserved_memory = true;
1818     pcmc->enforce_aligned_dimm = true;
1819     pcmc->enforce_amd_1tb_hole = true;
1820     pcmc->isa_bios_alias = true;
1821     /* BIOS ACPI tables: 128K. Other BIOS datastructures: less than 4K reported
1822      * to be used at the moment, 32K should be enough for a while.  */
1823     pcmc->acpi_data_size = 0x20000 + 0x8000;
1824     pcmc->pvh_enabled = true;
1825     pcmc->kvmclock_create_always = true;
1826     pcmc->resizable_acpi_blob = true;
1827     x86mc->apic_xrupt_override = true;
1828     assert(!mc->get_hotplug_handler);
1829     mc->get_hotplug_handler = pc_get_hotplug_handler;
1830     mc->hotplug_allowed = pc_hotplug_allowed;
1831     mc->auto_enable_numa_with_memhp = true;
1832     mc->auto_enable_numa_with_memdev = true;
1833     mc->has_hotpluggable_cpus = true;
1834     mc->default_boot_order = "cad";
1835     mc->block_default_type = IF_IDE;
1836     mc->max_cpus = 255;
1837     mc->reset = pc_machine_reset;
1838     mc->wakeup = pc_machine_wakeup;
1839     hc->pre_plug = pc_machine_device_pre_plug_cb;
1840     hc->plug = pc_machine_device_plug_cb;
1841     hc->unplug_request = pc_machine_device_unplug_request_cb;
1842     hc->unplug = pc_machine_device_unplug_cb;
1843     mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE;
1844     mc->nvdimm_supported = true;
1845     mc->smp_props.dies_supported = true;
1846     mc->smp_props.modules_supported = true;
1847     mc->default_ram_id = "pc.ram";
1848     pcmc->default_smbios_ep_type = SMBIOS_ENTRY_POINT_TYPE_AUTO;
1849 
1850     object_class_property_add(oc, PC_MACHINE_MAX_RAM_BELOW_4G, "size",
1851         pc_machine_get_max_ram_below_4g, pc_machine_set_max_ram_below_4g,
1852         NULL, NULL);
1853     object_class_property_set_description(oc, PC_MACHINE_MAX_RAM_BELOW_4G,
1854         "Maximum ram below the 4G boundary (32bit boundary)");
1855 
1856     object_class_property_add(oc, PC_MACHINE_VMPORT, "OnOffAuto",
1857         pc_machine_get_vmport, pc_machine_set_vmport,
1858         NULL, NULL);
1859     object_class_property_set_description(oc, PC_MACHINE_VMPORT,
1860         "Enable vmport (pc & q35)");
1861 
1862     object_class_property_add_bool(oc, PC_MACHINE_SMBUS,
1863         pc_machine_get_smbus, pc_machine_set_smbus);
1864     object_class_property_set_description(oc, PC_MACHINE_SMBUS,
1865         "Enable/disable system management bus");
1866 
1867     object_class_property_add_bool(oc, PC_MACHINE_SATA,
1868         pc_machine_get_sata, pc_machine_set_sata);
1869     object_class_property_set_description(oc, PC_MACHINE_SATA,
1870         "Enable/disable Serial ATA bus");
1871 
1872     object_class_property_add_bool(oc, "hpet",
1873         pc_machine_get_hpet, pc_machine_set_hpet);
1874     object_class_property_set_description(oc, "hpet",
1875         "Enable/disable high precision event timer emulation");
1876 
1877     object_class_property_add_bool(oc, PC_MACHINE_I8042,
1878         pc_machine_get_i8042, pc_machine_set_i8042);
1879 
1880     object_class_property_add_bool(oc, "default-bus-bypass-iommu",
1881         pc_machine_get_default_bus_bypass_iommu,
1882         pc_machine_set_default_bus_bypass_iommu);
1883 
1884     object_class_property_add(oc, PC_MACHINE_MAX_FW_SIZE, "size",
1885         pc_machine_get_max_fw_size, pc_machine_set_max_fw_size,
1886         NULL, NULL);
1887     object_class_property_set_description(oc, PC_MACHINE_MAX_FW_SIZE,
1888         "Maximum combined firmware size");
1889 
1890     object_class_property_add(oc, PC_MACHINE_SMBIOS_EP, "str",
1891         pc_machine_get_smbios_ep, pc_machine_set_smbios_ep,
1892         NULL, NULL);
1893     object_class_property_set_description(oc, PC_MACHINE_SMBIOS_EP,
1894         "SMBIOS Entry Point type [32, 64]");
1895 
1896     object_class_property_add_bool(oc, "fd-bootchk",
1897         pc_machine_get_fd_bootchk,
1898         pc_machine_set_fd_bootchk);
1899 }
1900 
1901 static const TypeInfo pc_machine_info = {
1902     .name = TYPE_PC_MACHINE,
1903     .parent = TYPE_X86_MACHINE,
1904     .abstract = true,
1905     .instance_size = sizeof(PCMachineState),
1906     .instance_init = pc_machine_initfn,
1907     .class_size = sizeof(PCMachineClass),
1908     .class_init = pc_machine_class_init,
1909     .interfaces = (InterfaceInfo[]) {
1910          { TYPE_HOTPLUG_HANDLER },
1911          { }
1912     },
1913 };
1914 
1915 static void pc_machine_register_types(void)
1916 {
1917     type_register_static(&pc_machine_info);
1918 }
1919 
1920 type_init(pc_machine_register_types)
1921