xref: /openbmc/qemu/hw/i386/pc.c (revision 16c04166ae71c78ec36a9e2914c46c2289a58503)
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 "exec/target_page.h"
28 #include "hw/i386/pc.h"
29 #include "hw/char/serial-isa.h"
30 #include "hw/char/parallel.h"
31 #include "hw/hyperv/hv-balloon.h"
32 #include "hw/i386/fw_cfg.h"
33 #include "hw/i386/vmport.h"
34 #include "system/cpus.h"
35 #include "hw/ide/ide-bus.h"
36 #include "hw/timer/hpet.h"
37 #include "hw/loader.h"
38 #include "hw/rtc/mc146818rtc.h"
39 #include "hw/intc/i8259.h"
40 #include "hw/timer/i8254.h"
41 #include "hw/input/i8042.h"
42 #include "hw/audio/pcspk.h"
43 #include "system/system.h"
44 #include "system/xen.h"
45 #include "system/reset.h"
46 #include "kvm/kvm_i386.h"
47 #include "hw/xen/xen.h"
48 #include "qobject/qlist.h"
49 #include "qemu/error-report.h"
50 #include "hw/acpi/cpu_hotplug.h"
51 #include "acpi-build.h"
52 #include "hw/mem/nvdimm.h"
53 #include "hw/cxl/cxl_host.h"
54 #include "hw/usb.h"
55 #include "hw/i386/intel_iommu.h"
56 #include "hw/net/ne2000-isa.h"
57 #include "hw/virtio/virtio-iommu.h"
58 #include "hw/virtio/virtio-md-pci.h"
59 #include "hw/i386/kvm/xen_overlay.h"
60 #include "hw/i386/kvm/xen_evtchn.h"
61 #include "hw/i386/kvm/xen_gnttab.h"
62 #include "hw/i386/kvm/xen_xenstore.h"
63 #include "hw/mem/memory-device.h"
64 #include "e820_memory_layout.h"
65 #include "trace.h"
66 #include "sev.h"
67 #include CONFIG_DEVICES
68 
69 #ifdef CONFIG_XEN_EMU
70 #include "hw/xen/xen-legacy-backend.h"
71 #include "hw/xen/xen-bus.h"
72 #endif
73 
74 /*
75  * Helper for setting model-id for CPU models that changed model-id
76  * depending on QEMU versions up to QEMU 2.4.
77  */
78 #define PC_CPU_MODEL_IDS(v) \
79     { "qemu32-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
80     { "qemu64-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
81     { "athlon-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },
82 
83 GlobalProperty pc_compat_10_0[] = {};
84 const size_t pc_compat_10_0_len = G_N_ELEMENTS(pc_compat_10_0);
85 
86 GlobalProperty pc_compat_9_2[] = {};
87 const size_t pc_compat_9_2_len = G_N_ELEMENTS(pc_compat_9_2);
88 
89 GlobalProperty pc_compat_9_1[] = {
90     { "ICH9-LPC", "x-smi-swsmi-timer", "off" },
91     { "ICH9-LPC", "x-smi-periodic-timer", "off" },
92     { TYPE_INTEL_IOMMU_DEVICE, "stale-tm", "on" },
93     { TYPE_INTEL_IOMMU_DEVICE, "aw-bits", "39" },
94 };
95 const size_t pc_compat_9_1_len = G_N_ELEMENTS(pc_compat_9_1);
96 
97 GlobalProperty pc_compat_9_0[] = {
98     { TYPE_X86_CPU, "x-amd-topoext-features-only", "false" },
99     { TYPE_X86_CPU, "x-l1-cache-per-thread", "false" },
100     { TYPE_X86_CPU, "guest-phys-bits", "0" },
101     { "sev-guest", "legacy-vm-type", "on" },
102     { TYPE_X86_CPU, "legacy-multi-node", "on" },
103 };
104 const size_t pc_compat_9_0_len = G_N_ELEMENTS(pc_compat_9_0);
105 
106 GlobalProperty pc_compat_8_2[] = {};
107 const size_t pc_compat_8_2_len = G_N_ELEMENTS(pc_compat_8_2);
108 
109 GlobalProperty pc_compat_8_1[] = {};
110 const size_t pc_compat_8_1_len = G_N_ELEMENTS(pc_compat_8_1);
111 
112 GlobalProperty pc_compat_8_0[] = {
113     { "virtio-mem", "unplugged-inaccessible", "auto" },
114 };
115 const size_t pc_compat_8_0_len = G_N_ELEMENTS(pc_compat_8_0);
116 
117 GlobalProperty pc_compat_7_2[] = {
118     { "ICH9-LPC", "noreboot", "true" },
119 };
120 const size_t pc_compat_7_2_len = G_N_ELEMENTS(pc_compat_7_2);
121 
122 GlobalProperty pc_compat_7_1[] = {};
123 const size_t pc_compat_7_1_len = G_N_ELEMENTS(pc_compat_7_1);
124 
125 GlobalProperty pc_compat_7_0[] = {};
126 const size_t pc_compat_7_0_len = G_N_ELEMENTS(pc_compat_7_0);
127 
128 GlobalProperty pc_compat_6_2[] = {
129     { "virtio-mem", "unplugged-inaccessible", "off" },
130 };
131 const size_t pc_compat_6_2_len = G_N_ELEMENTS(pc_compat_6_2);
132 
133 GlobalProperty pc_compat_6_1[] = {
134     { TYPE_X86_CPU, "hv-version-id-build", "0x1bbc" },
135     { TYPE_X86_CPU, "hv-version-id-major", "0x0006" },
136     { TYPE_X86_CPU, "hv-version-id-minor", "0x0001" },
137     { "ICH9-LPC", "x-keep-pci-slot-hpc", "false" },
138 };
139 const size_t pc_compat_6_1_len = G_N_ELEMENTS(pc_compat_6_1);
140 
141 GlobalProperty pc_compat_6_0[] = {
142     { "qemu64" "-" TYPE_X86_CPU, "family", "6" },
143     { "qemu64" "-" TYPE_X86_CPU, "model", "6" },
144     { "qemu64" "-" TYPE_X86_CPU, "stepping", "3" },
145     { TYPE_X86_CPU, "x-vendor-cpuid-only", "off" },
146     { "ICH9-LPC", ACPI_PM_PROP_ACPI_PCIHP_BRIDGE, "off" },
147     { "ICH9-LPC", "x-keep-pci-slot-hpc", "true" },
148 };
149 const size_t pc_compat_6_0_len = G_N_ELEMENTS(pc_compat_6_0);
150 
151 GlobalProperty pc_compat_5_2[] = {
152     { "ICH9-LPC", "x-smi-cpu-hotunplug", "off" },
153 };
154 const size_t pc_compat_5_2_len = G_N_ELEMENTS(pc_compat_5_2);
155 
156 GlobalProperty pc_compat_5_1[] = {
157     { "ICH9-LPC", "x-smi-cpu-hotplug", "off" },
158     { TYPE_X86_CPU, "kvm-msi-ext-dest-id", "off" },
159 };
160 const size_t pc_compat_5_1_len = G_N_ELEMENTS(pc_compat_5_1);
161 
162 GlobalProperty pc_compat_5_0[] = {
163 };
164 const size_t pc_compat_5_0_len = G_N_ELEMENTS(pc_compat_5_0);
165 
166 GlobalProperty pc_compat_4_2[] = {
167     { "mch", "smbase-smram", "off" },
168 };
169 const size_t pc_compat_4_2_len = G_N_ELEMENTS(pc_compat_4_2);
170 
171 GlobalProperty pc_compat_4_1[] = {};
172 const size_t pc_compat_4_1_len = G_N_ELEMENTS(pc_compat_4_1);
173 
174 GlobalProperty pc_compat_4_0[] = {};
175 const size_t pc_compat_4_0_len = G_N_ELEMENTS(pc_compat_4_0);
176 
177 GlobalProperty pc_compat_3_1[] = {
178     { "intel-iommu", "dma-drain", "off" },
179     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "off" },
180     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "off" },
181     { "Opteron_G4" "-" TYPE_X86_CPU, "npt", "off" },
182     { "Opteron_G4" "-" TYPE_X86_CPU, "nrip-save", "off" },
183     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "off" },
184     { "Opteron_G5" "-" TYPE_X86_CPU, "npt", "off" },
185     { "Opteron_G5" "-" TYPE_X86_CPU, "nrip-save", "off" },
186     { "EPYC" "-" TYPE_X86_CPU, "npt", "off" },
187     { "EPYC" "-" TYPE_X86_CPU, "nrip-save", "off" },
188     { "EPYC-IBPB" "-" TYPE_X86_CPU, "npt", "off" },
189     { "EPYC-IBPB" "-" TYPE_X86_CPU, "nrip-save", "off" },
190     { "Skylake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
191     { "Skylake-Client-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
192     { "Skylake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
193     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
194     { "Cascadelake-Server" "-" TYPE_X86_CPU,  "mpx", "on" },
195     { "Icelake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
196     { "Icelake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
197     { "Cascadelake-Server" "-" TYPE_X86_CPU, "stepping", "5" },
198     { TYPE_X86_CPU, "x-intel-pt-auto-level", "off" },
199 };
200 const size_t pc_compat_3_1_len = G_N_ELEMENTS(pc_compat_3_1);
201 
202 GlobalProperty pc_compat_3_0[] = {
203     { TYPE_X86_CPU, "x-hv-synic-kvm-only", "on" },
204     { "Skylake-Server" "-" TYPE_X86_CPU, "pku", "off" },
205     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "pku", "off" },
206 };
207 const size_t pc_compat_3_0_len = G_N_ELEMENTS(pc_compat_3_0);
208 
209 GlobalProperty pc_compat_2_12[] = {
210     { TYPE_X86_CPU, "legacy-cache", "on" },
211     { TYPE_X86_CPU, "topoext", "off" },
212     { "EPYC-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
213     { "EPYC-IBPB-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
214 };
215 const size_t pc_compat_2_12_len = G_N_ELEMENTS(pc_compat_2_12);
216 
217 GlobalProperty pc_compat_2_11[] = {
218     { TYPE_X86_CPU, "x-migrate-smi-count", "off" },
219     { "Skylake-Server" "-" TYPE_X86_CPU, "clflushopt", "off" },
220 };
221 const size_t pc_compat_2_11_len = G_N_ELEMENTS(pc_compat_2_11);
222 
223 GlobalProperty pc_compat_2_10[] = {
224     { TYPE_X86_CPU, "x-hv-max-vps", "0x40" },
225     { "i440FX-pcihost", "x-pci-hole64-fix", "off" },
226     { "q35-pcihost", "x-pci-hole64-fix", "off" },
227 };
228 const size_t pc_compat_2_10_len = G_N_ELEMENTS(pc_compat_2_10);
229 
230 GlobalProperty pc_compat_2_9[] = {
231     { "mch", "extended-tseg-mbytes", "0" },
232 };
233 const size_t pc_compat_2_9_len = G_N_ELEMENTS(pc_compat_2_9);
234 
235 GlobalProperty pc_compat_2_8[] = {
236     { TYPE_X86_CPU, "tcg-cpuid", "off" },
237     { "kvmclock", "x-mach-use-reliable-get-clock", "off" },
238     { "ICH9-LPC", "x-smi-broadcast", "off" },
239     { TYPE_X86_CPU, "vmware-cpuid-freq", "off" },
240     { "Haswell-" TYPE_X86_CPU, "stepping", "1" },
241 };
242 const size_t pc_compat_2_8_len = G_N_ELEMENTS(pc_compat_2_8);
243 
244 GlobalProperty pc_compat_2_7[] = {
245     { TYPE_X86_CPU, "l3-cache", "off" },
246     { TYPE_X86_CPU, "full-cpuid-auto-level", "off" },
247     { "Opteron_G3" "-" TYPE_X86_CPU, "family", "15" },
248     { "Opteron_G3" "-" TYPE_X86_CPU, "model", "6" },
249     { "Opteron_G3" "-" TYPE_X86_CPU, "stepping", "1" },
250     { "isa-pcspk", "migrate", "off" },
251 };
252 const size_t pc_compat_2_7_len = G_N_ELEMENTS(pc_compat_2_7);
253 
254 GlobalProperty pc_compat_2_6[] = {
255     { TYPE_X86_CPU, "cpuid-0xb", "off" },
256     { "vmxnet3", "romfile", "" },
257     { TYPE_X86_CPU, "fill-mtrr-mask", "off" },
258     { "apic-common", "legacy-instance-id", "on", }
259 };
260 const size_t pc_compat_2_6_len = G_N_ELEMENTS(pc_compat_2_6);
261 
262 /*
263  * @PC_FW_DATA:
264  * Size of the chunk of memory at the top of RAM for the BIOS ACPI tables
265  * and other BIOS datastructures.
266  *
267  * BIOS ACPI tables: 128K. Other BIOS datastructures: less than 4K
268  * reported to be used at the moment, 32K should be enough for a while.
269  */
270 #define PC_FW_DATA (0x20000 + 0x8000)
271 
272 GSIState *pc_gsi_create(qemu_irq **irqs, bool pci_enabled)
273 {
274     GSIState *s;
275 
276     s = g_new0(GSIState, 1);
277     if (kvm_ioapic_in_kernel()) {
278         kvm_pc_setup_irq_routing(pci_enabled);
279     }
280     *irqs = qemu_allocate_irqs(gsi_handler, s, IOAPIC_NUM_PINS);
281 
282     return s;
283 }
284 
285 static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
286                            unsigned size)
287 {
288 }
289 
290 static uint64_t ioport80_read(void *opaque, hwaddr addr, unsigned size)
291 {
292     return 0xffffffffffffffffULL;
293 }
294 
295 /* MS-DOS compatibility mode FPU exception support */
296 static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
297                            unsigned size)
298 {
299     if (tcg_enabled()) {
300         cpu_set_ignne();
301     }
302 }
303 
304 static uint64_t ioportF0_read(void *opaque, hwaddr addr, unsigned size)
305 {
306     return 0xffffffffffffffffULL;
307 }
308 
309 /* PC cmos mappings */
310 
311 #define REG_EQUIPMENT_BYTE          0x14
312 
313 static void cmos_init_hd(MC146818RtcState *s, int type_ofs, int info_ofs,
314                          int16_t cylinders, int8_t heads, int8_t sectors)
315 {
316     mc146818rtc_set_cmos_data(s, type_ofs, 47);
317     mc146818rtc_set_cmos_data(s, info_ofs, cylinders);
318     mc146818rtc_set_cmos_data(s, info_ofs + 1, cylinders >> 8);
319     mc146818rtc_set_cmos_data(s, info_ofs + 2, heads);
320     mc146818rtc_set_cmos_data(s, info_ofs + 3, 0xff);
321     mc146818rtc_set_cmos_data(s, info_ofs + 4, 0xff);
322     mc146818rtc_set_cmos_data(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
323     mc146818rtc_set_cmos_data(s, info_ofs + 6, cylinders);
324     mc146818rtc_set_cmos_data(s, info_ofs + 7, cylinders >> 8);
325     mc146818rtc_set_cmos_data(s, info_ofs + 8, sectors);
326 }
327 
328 /* convert boot_device letter to something recognizable by the bios */
329 static int boot_device2nibble(char boot_device)
330 {
331     switch(boot_device) {
332     case 'a':
333     case 'b':
334         return 0x01; /* floppy boot */
335     case 'c':
336         return 0x02; /* hard drive boot */
337     case 'd':
338         return 0x03; /* CD-ROM boot */
339     case 'n':
340         return 0x04; /* Network boot */
341     }
342     return 0;
343 }
344 
345 static void set_boot_dev(PCMachineState *pcms, MC146818RtcState *s,
346                          const char *boot_device, Error **errp)
347 {
348 #define PC_MAX_BOOT_DEVICES 3
349     int nbds, bds[3] = { 0, };
350     int i;
351 
352     nbds = strlen(boot_device);
353     if (nbds > PC_MAX_BOOT_DEVICES) {
354         error_setg(errp, "Too many boot devices for PC");
355         return;
356     }
357     for (i = 0; i < nbds; i++) {
358         bds[i] = boot_device2nibble(boot_device[i]);
359         if (bds[i] == 0) {
360             error_setg(errp, "Invalid boot device for PC: '%c'",
361                        boot_device[i]);
362             return;
363         }
364     }
365     mc146818rtc_set_cmos_data(s, 0x3d, (bds[1] << 4) | bds[0]);
366     mc146818rtc_set_cmos_data(s, 0x38, (bds[2] << 4) | !pcms->fd_bootchk);
367 }
368 
369 static void pc_boot_set(void *opaque, const char *boot_device, Error **errp)
370 {
371     PCMachineState *pcms = opaque;
372     X86MachineState *x86ms = X86_MACHINE(pcms);
373 
374     set_boot_dev(pcms, MC146818_RTC(x86ms->rtc), boot_device, errp);
375 }
376 
377 static void pc_cmos_init_floppy(MC146818RtcState *rtc_state, ISADevice *floppy)
378 {
379     int val, nb;
380     FloppyDriveType fd_type[2] = { FLOPPY_DRIVE_TYPE_NONE,
381                                    FLOPPY_DRIVE_TYPE_NONE };
382 
383 #ifdef CONFIG_FDC_ISA
384     /* floppy type */
385     if (floppy) {
386         for (int i = 0; i < 2; i++) {
387             fd_type[i] = isa_fdc_get_drive_type(floppy, i);
388         }
389     }
390 #endif
391 
392     val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
393         cmos_get_fd_drive_type(fd_type[1]);
394     mc146818rtc_set_cmos_data(rtc_state, 0x10, val);
395 
396     val = mc146818rtc_get_cmos_data(rtc_state, REG_EQUIPMENT_BYTE);
397     nb = 0;
398     if (fd_type[0] != FLOPPY_DRIVE_TYPE_NONE) {
399         nb++;
400     }
401     if (fd_type[1] != FLOPPY_DRIVE_TYPE_NONE) {
402         nb++;
403     }
404     switch (nb) {
405     case 0:
406         break;
407     case 1:
408         val |= 0x01; /* 1 drive, ready for boot */
409         break;
410     case 2:
411         val |= 0x41; /* 2 drives, ready for boot */
412         break;
413     }
414     mc146818rtc_set_cmos_data(rtc_state, REG_EQUIPMENT_BYTE, val);
415 }
416 
417 typedef struct check_fdc_state {
418     ISADevice *floppy;
419     bool multiple;
420 } CheckFdcState;
421 
422 static int check_fdc(Object *obj, void *opaque)
423 {
424     CheckFdcState *state = opaque;
425     Object *fdc;
426     uint32_t iobase;
427     Error *local_err = NULL;
428 
429     fdc = object_dynamic_cast(obj, TYPE_ISA_FDC);
430     if (!fdc) {
431         return 0;
432     }
433 
434     iobase = object_property_get_uint(obj, "iobase", &local_err);
435     if (local_err || iobase != 0x3f0) {
436         error_free(local_err);
437         return 0;
438     }
439 
440     if (state->floppy) {
441         state->multiple = true;
442     } else {
443         state->floppy = ISA_DEVICE(obj);
444     }
445     return 0;
446 }
447 
448 static const char * const fdc_container_path[] = {
449     "unattached", "peripheral", "peripheral-anon"
450 };
451 
452 /*
453  * Locate the FDC at IO address 0x3f0, in order to configure the CMOS registers
454  * and ACPI objects.
455  */
456 static ISADevice *pc_find_fdc0(void)
457 {
458     int i;
459     Object *container;
460     CheckFdcState state = { 0 };
461 
462     for (i = 0; i < ARRAY_SIZE(fdc_container_path); i++) {
463         container = machine_get_container(fdc_container_path[i]);
464         object_child_foreach(container, check_fdc, &state);
465     }
466 
467     if (state.multiple) {
468         warn_report("multiple floppy disk controllers with "
469                     "iobase=0x3f0 have been found");
470         error_printf("the one being picked for CMOS setup might not reflect "
471                      "your intent");
472     }
473 
474     return state.floppy;
475 }
476 
477 static void pc_cmos_init_late(PCMachineState *pcms)
478 {
479     X86MachineState *x86ms = X86_MACHINE(pcms);
480     MC146818RtcState *s = MC146818_RTC(x86ms->rtc);
481     int16_t cylinders;
482     int8_t heads, sectors;
483     int val;
484     int i, trans;
485 
486     val = 0;
487     if (pcms->idebus[0] &&
488         ide_get_geometry(pcms->idebus[0], 0,
489                          &cylinders, &heads, &sectors) >= 0) {
490         cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
491         val |= 0xf0;
492     }
493     if (pcms->idebus[0] &&
494         ide_get_geometry(pcms->idebus[0], 1,
495                          &cylinders, &heads, &sectors) >= 0) {
496         cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
497         val |= 0x0f;
498     }
499     mc146818rtc_set_cmos_data(s, 0x12, val);
500 
501     val = 0;
502     for (i = 0; i < 4; i++) {
503         /* NOTE: ide_get_geometry() returns the physical
504            geometry.  It is always such that: 1 <= sects <= 63, 1
505            <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
506            geometry can be different if a translation is done. */
507         BusState *idebus = pcms->idebus[i / 2];
508         if (idebus &&
509             ide_get_geometry(idebus, i % 2,
510                              &cylinders, &heads, &sectors) >= 0) {
511             trans = ide_get_bios_chs_trans(idebus, i % 2) - 1;
512             assert((trans & ~3) == 0);
513             val |= trans << (i * 2);
514         }
515     }
516     mc146818rtc_set_cmos_data(s, 0x39, val);
517 
518     pc_cmos_init_floppy(s, pc_find_fdc0());
519 
520     /* various important CMOS locations needed by PC/Bochs bios */
521 
522     /* memory size */
523     /* base memory (first MiB) */
524     val = MIN(x86ms->below_4g_mem_size / KiB, 640);
525     mc146818rtc_set_cmos_data(s, 0x15, val);
526     mc146818rtc_set_cmos_data(s, 0x16, val >> 8);
527     /* extended memory (next 64MiB) */
528     if (x86ms->below_4g_mem_size > 1 * MiB) {
529         val = (x86ms->below_4g_mem_size - 1 * MiB) / KiB;
530     } else {
531         val = 0;
532     }
533     if (val > 65535)
534         val = 65535;
535     mc146818rtc_set_cmos_data(s, 0x17, val);
536     mc146818rtc_set_cmos_data(s, 0x18, val >> 8);
537     mc146818rtc_set_cmos_data(s, 0x30, val);
538     mc146818rtc_set_cmos_data(s, 0x31, val >> 8);
539     /* memory between 16MiB and 4GiB */
540     if (x86ms->below_4g_mem_size > 16 * MiB) {
541         val = (x86ms->below_4g_mem_size - 16 * MiB) / (64 * KiB);
542     } else {
543         val = 0;
544     }
545     if (val > 65535)
546         val = 65535;
547     mc146818rtc_set_cmos_data(s, 0x34, val);
548     mc146818rtc_set_cmos_data(s, 0x35, val >> 8);
549     /* memory above 4GiB */
550     val = x86ms->above_4g_mem_size / 65536;
551     mc146818rtc_set_cmos_data(s, 0x5b, val);
552     mc146818rtc_set_cmos_data(s, 0x5c, val >> 8);
553     mc146818rtc_set_cmos_data(s, 0x5d, val >> 16);
554 
555     val = 0;
556     val |= 0x02; /* FPU is there */
557     val |= 0x04; /* PS/2 mouse installed */
558     mc146818rtc_set_cmos_data(s, REG_EQUIPMENT_BYTE, val);
559 }
560 
561 static void handle_a20_line_change(void *opaque, int irq, int level)
562 {
563     X86CPU *cpu = opaque;
564 
565     /* XXX: send to all CPUs ? */
566     /* XXX: add logic to handle multiple A20 line sources */
567     x86_cpu_set_a20(cpu, level);
568 }
569 
570 #define NE2000_NB_MAX 6
571 
572 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
573                                               0x280, 0x380 };
574 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
575 
576 static gboolean pc_init_ne2k_isa(ISABus *bus, NICInfo *nd, Error **errp)
577 {
578     static int nb_ne2k = 0;
579 
580     if (nb_ne2k == NE2000_NB_MAX) {
581         error_setg(errp,
582                    "maximum number of ISA NE2000 devices exceeded");
583         return false;
584     }
585     isa_ne2000_init(bus, ne2000_io[nb_ne2k],
586                     ne2000_irq[nb_ne2k], nd);
587     nb_ne2k++;
588     return true;
589 }
590 
591 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
592 {
593     X86CPU *cpu = opaque;
594 
595     if (level) {
596         cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
597     }
598 }
599 
600 static
601 void pc_machine_done(Notifier *notifier, void *data)
602 {
603     PCMachineState *pcms = container_of(notifier,
604                                         PCMachineState, machine_done);
605     X86MachineState *x86ms = X86_MACHINE(pcms);
606 
607     cxl_hook_up_pxb_registers(pcms->pcibus, &pcms->cxl_devices_state,
608                               &error_fatal);
609 
610     if (pcms->cxl_devices_state.is_enabled) {
611         cxl_fmws_link_targets(&pcms->cxl_devices_state, &error_fatal);
612     }
613 
614     /* set the number of CPUs */
615     x86_rtc_set_cpus_count(x86ms->rtc, x86ms->boot_cpus);
616 
617     pci_bus_add_fw_cfg_extra_pci_roots(x86ms->fw_cfg, pcms->pcibus,
618                                        &error_abort);
619 
620     acpi_setup();
621     if (x86ms->fw_cfg) {
622         fw_cfg_build_smbios(pcms, x86ms->fw_cfg, pcms->smbios_entry_point_type);
623         fw_cfg_add_e820(x86ms->fw_cfg);
624         fw_cfg_build_feature_control(MACHINE(pcms), x86ms->fw_cfg);
625         /* update FW_CFG_NB_CPUS to account for -device added CPUs */
626         fw_cfg_modify_i16(x86ms->fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
627     }
628 
629     pc_cmos_init_late(pcms);
630 }
631 
632 /* setup pci memory address space mapping into system address space */
633 void pc_pci_as_mapping_init(MemoryRegion *system_memory,
634                             MemoryRegion *pci_address_space)
635 {
636     /* Set to lower priority than RAM */
637     memory_region_add_subregion_overlap(system_memory, 0x0,
638                                         pci_address_space, -1);
639 }
640 
641 void xen_load_linux(PCMachineState *pcms)
642 {
643     int i;
644     FWCfgState *fw_cfg;
645     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
646     X86MachineState *x86ms = X86_MACHINE(pcms);
647 
648     assert(MACHINE(pcms)->kernel_filename != NULL);
649 
650     fw_cfg = fw_cfg_init_io_dma(FW_CFG_IO_BASE, FW_CFG_IO_BASE + 4,
651                                 &address_space_memory);
652     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
653     rom_set_fw(fw_cfg);
654 
655     x86_load_linux(x86ms, fw_cfg, PC_FW_DATA, pcmc->pvh_enabled);
656     for (i = 0; i < nb_option_roms; i++) {
657         assert(!strcmp(option_rom[i].name, "linuxboot.bin") ||
658                !strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
659                !strcmp(option_rom[i].name, "pvh.bin") ||
660                !strcmp(option_rom[i].name, "multiboot.bin") ||
661                !strcmp(option_rom[i].name, "multiboot_dma.bin"));
662         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
663     }
664     x86ms->fw_cfg = fw_cfg;
665 }
666 
667 #define PC_ROM_MIN_VGA     0xc0000
668 #define PC_ROM_MIN_OPTION  0xc8000
669 #define PC_ROM_MAX         0xe0000
670 #define PC_ROM_ALIGN       0x800
671 #define PC_ROM_SIZE        (PC_ROM_MAX - PC_ROM_MIN_VGA)
672 
673 static hwaddr pc_above_4g_end(PCMachineState *pcms)
674 {
675     X86MachineState *x86ms = X86_MACHINE(pcms);
676 
677     if (pcms->sgx_epc.size != 0) {
678         return sgx_epc_above_4g_end(&pcms->sgx_epc);
679     }
680 
681     return x86ms->above_4g_mem_start + x86ms->above_4g_mem_size;
682 }
683 
684 static void pc_get_device_memory_range(PCMachineState *pcms,
685                                        hwaddr *base,
686                                        ram_addr_t *device_mem_size)
687 {
688     MachineState *machine = MACHINE(pcms);
689     ram_addr_t size;
690     hwaddr addr;
691 
692     size = machine->maxram_size - machine->ram_size;
693     addr = ROUND_UP(pc_above_4g_end(pcms), 1 * GiB);
694 
695     /* size device region assuming 1G page max alignment per slot */
696     size += (1 * GiB) * machine->ram_slots;
697 
698     *base = addr;
699     *device_mem_size = size;
700 }
701 
702 static uint64_t pc_get_cxl_range_start(PCMachineState *pcms)
703 {
704     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
705     MachineState *ms = MACHINE(pcms);
706     hwaddr cxl_base;
707     ram_addr_t size;
708 
709     if (pcmc->has_reserved_memory &&
710         (ms->ram_size < ms->maxram_size)) {
711         pc_get_device_memory_range(pcms, &cxl_base, &size);
712         cxl_base += size;
713     } else {
714         cxl_base = pc_above_4g_end(pcms);
715     }
716 
717     return cxl_base;
718 }
719 
720 static uint64_t pc_get_cxl_range_end(PCMachineState *pcms)
721 {
722     uint64_t start = pc_get_cxl_range_start(pcms) + MiB;
723 
724     if (pcms->cxl_devices_state.fixed_windows) {
725         GList *it;
726 
727         start = ROUND_UP(start, 256 * MiB);
728         for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
729             CXLFixedWindow *fw = it->data;
730             start += fw->size;
731         }
732     }
733 
734     return start;
735 }
736 
737 static hwaddr pc_max_used_gpa(PCMachineState *pcms, uint64_t pci_hole64_size)
738 {
739     X86CPU *cpu = X86_CPU(first_cpu);
740     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
741     MachineState *ms = MACHINE(pcms);
742 
743     if (cpu->env.features[FEAT_8000_0001_EDX] & CPUID_EXT2_LM) {
744         /* 64-bit systems */
745         return pc_pci_hole64_start() + pci_hole64_size - 1;
746     }
747 
748     /* 32-bit systems */
749     if (pcmc->broken_32bit_mem_addr_check) {
750         /* old value for compatibility reasons */
751         return ((hwaddr)1 << cpu->phys_bits) - 1;
752     }
753 
754     /*
755      * 32-bit systems don't have hole64 but they might have a region for
756      * memory devices. Even if additional hotplugged memory devices might
757      * not be usable by most guest OSes, we need to still consider them for
758      * calculating the highest possible GPA so that we can properly report
759      * if someone configures them on a CPU that cannot possibly address them.
760      */
761     if (pcmc->has_reserved_memory &&
762         (ms->ram_size < ms->maxram_size)) {
763         hwaddr devmem_start;
764         ram_addr_t devmem_size;
765 
766         pc_get_device_memory_range(pcms, &devmem_start, &devmem_size);
767         devmem_start += devmem_size;
768         return devmem_start - 1;
769     }
770 
771     /* configuration without any memory hotplug */
772     return pc_above_4g_end(pcms) - 1;
773 }
774 
775 /*
776  * AMD systems with an IOMMU have an additional hole close to the
777  * 1Tb, which are special GPAs that cannot be DMA mapped. Depending
778  * on kernel version, VFIO may or may not let you DMA map those ranges.
779  * Starting Linux v5.4 we validate it, and can't create guests on AMD machines
780  * with certain memory sizes. It's also wrong to use those IOVA ranges
781  * in detriment of leading to IOMMU INVALID_DEVICE_REQUEST or worse.
782  * The ranges reserved for Hyper-Transport are:
783  *
784  * FD_0000_0000h - FF_FFFF_FFFFh
785  *
786  * The ranges represent the following:
787  *
788  * Base Address   Top Address  Use
789  *
790  * FD_0000_0000h FD_F7FF_FFFFh Reserved interrupt address space
791  * FD_F800_0000h FD_F8FF_FFFFh Interrupt/EOI IntCtl
792  * FD_F900_0000h FD_F90F_FFFFh Legacy PIC IACK
793  * FD_F910_0000h FD_F91F_FFFFh System Management
794  * FD_F920_0000h FD_FAFF_FFFFh Reserved Page Tables
795  * FD_FB00_0000h FD_FBFF_FFFFh Address Translation
796  * FD_FC00_0000h FD_FDFF_FFFFh I/O Space
797  * FD_FE00_0000h FD_FFFF_FFFFh Configuration
798  * FE_0000_0000h FE_1FFF_FFFFh Extended Configuration/Device Messages
799  * FE_2000_0000h FF_FFFF_FFFFh Reserved
800  *
801  * See AMD IOMMU spec, section 2.1.2 "IOMMU Logical Topology",
802  * Table 3: Special Address Controls (GPA) for more information.
803  */
804 #define AMD_HT_START         0xfd00000000UL
805 #define AMD_HT_END           0xffffffffffUL
806 #define AMD_ABOVE_1TB_START  (AMD_HT_END + 1)
807 #define AMD_HT_SIZE          (AMD_ABOVE_1TB_START - AMD_HT_START)
808 
809 void pc_memory_init(PCMachineState *pcms,
810                     MemoryRegion *system_memory,
811                     MemoryRegion *rom_memory,
812                     uint64_t pci_hole64_size)
813 {
814     int linux_boot, i;
815     MemoryRegion *option_rom_mr;
816     MemoryRegion *ram_below_4g, *ram_above_4g;
817     FWCfgState *fw_cfg;
818     MachineState *machine = MACHINE(pcms);
819     MachineClass *mc = MACHINE_GET_CLASS(machine);
820     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
821     X86MachineState *x86ms = X86_MACHINE(pcms);
822     hwaddr maxphysaddr, maxusedaddr;
823     hwaddr cxl_base, cxl_resv_end = 0;
824     X86CPU *cpu = X86_CPU(first_cpu);
825 
826     assert(machine->ram_size == x86ms->below_4g_mem_size +
827                                 x86ms->above_4g_mem_size);
828 
829     linux_boot = (machine->kernel_filename != NULL);
830 
831     /*
832      * The HyperTransport range close to the 1T boundary is unique to AMD
833      * hosts with IOMMUs enabled. Restrict the ram-above-4g relocation
834      * to above 1T to AMD vCPUs only. @enforce_amd_1tb_hole is only false in
835      * older machine types (<= 7.0) for compatibility purposes.
836      */
837     if (IS_AMD_CPU(&cpu->env) && pcmc->enforce_amd_1tb_hole) {
838         /* Bail out if max possible address does not cross HT range */
839         if (pc_max_used_gpa(pcms, pci_hole64_size) >= AMD_HT_START) {
840             x86ms->above_4g_mem_start = AMD_ABOVE_1TB_START;
841         }
842 
843         /*
844          * Advertise the HT region if address space covers the reserved
845          * region or if we relocate.
846          */
847         if (cpu->phys_bits >= 40) {
848             e820_add_entry(AMD_HT_START, AMD_HT_SIZE, E820_RESERVED);
849         }
850     }
851 
852     /*
853      * phys-bits is required to be appropriately configured
854      * to make sure max used GPA is reachable.
855      */
856     maxusedaddr = pc_max_used_gpa(pcms, pci_hole64_size);
857     maxphysaddr = ((hwaddr)1 << cpu->phys_bits) - 1;
858     if (maxphysaddr < maxusedaddr) {
859         error_report("Address space limit 0x%"PRIx64" < 0x%"PRIx64
860                      " phys-bits too low (%u)",
861                      maxphysaddr, maxusedaddr, cpu->phys_bits);
862         exit(EXIT_FAILURE);
863     }
864 
865     /*
866      * Split single memory region and use aliases to address portions of it,
867      * done for backwards compatibility with older qemus.
868      */
869     ram_below_4g = g_malloc(sizeof(*ram_below_4g));
870     memory_region_init_alias(ram_below_4g, NULL, "ram-below-4g", machine->ram,
871                              0, x86ms->below_4g_mem_size);
872     memory_region_add_subregion(system_memory, 0, ram_below_4g);
873     e820_add_entry(0, x86ms->below_4g_mem_size, E820_RAM);
874     if (x86ms->above_4g_mem_size > 0) {
875         ram_above_4g = g_malloc(sizeof(*ram_above_4g));
876         memory_region_init_alias(ram_above_4g, NULL, "ram-above-4g",
877                                  machine->ram,
878                                  x86ms->below_4g_mem_size,
879                                  x86ms->above_4g_mem_size);
880         memory_region_add_subregion(system_memory, x86ms->above_4g_mem_start,
881                                     ram_above_4g);
882         e820_add_entry(x86ms->above_4g_mem_start, x86ms->above_4g_mem_size,
883                        E820_RAM);
884     }
885 
886     if (pcms->sgx_epc.size != 0) {
887         e820_add_entry(pcms->sgx_epc.base, pcms->sgx_epc.size, E820_RESERVED);
888     }
889 
890     if (!pcmc->has_reserved_memory &&
891         (machine->ram_slots ||
892          (machine->maxram_size > machine->ram_size))) {
893 
894         error_report("\"-memory 'slots|maxmem'\" is not supported by: %s",
895                      mc->name);
896         exit(EXIT_FAILURE);
897     }
898 
899     /* initialize device memory address space */
900     if (pcmc->has_reserved_memory &&
901         (machine->ram_size < machine->maxram_size)) {
902         ram_addr_t device_mem_size;
903         hwaddr device_mem_base;
904 
905         if (machine->ram_slots > ACPI_MAX_RAM_SLOTS) {
906             error_report("unsupported amount of memory slots: %"PRIu64,
907                          machine->ram_slots);
908             exit(EXIT_FAILURE);
909         }
910 
911         if (QEMU_ALIGN_UP(machine->maxram_size,
912                           TARGET_PAGE_SIZE) != machine->maxram_size) {
913             error_report("maximum memory size must by aligned to multiple of "
914                          "%d bytes", TARGET_PAGE_SIZE);
915             exit(EXIT_FAILURE);
916         }
917 
918         pc_get_device_memory_range(pcms, &device_mem_base, &device_mem_size);
919 
920         if (device_mem_base + device_mem_size < device_mem_size) {
921             error_report("unsupported amount of maximum memory: " RAM_ADDR_FMT,
922                          machine->maxram_size);
923             exit(EXIT_FAILURE);
924         }
925         machine_memory_devices_init(machine, device_mem_base, device_mem_size);
926     }
927 
928     if (pcms->cxl_devices_state.is_enabled) {
929         MemoryRegion *mr = &pcms->cxl_devices_state.host_mr;
930         hwaddr cxl_size = MiB;
931 
932         cxl_base = pc_get_cxl_range_start(pcms);
933         memory_region_init(mr, OBJECT(machine), "cxl_host_reg", cxl_size);
934         memory_region_add_subregion(system_memory, cxl_base, mr);
935         cxl_resv_end = cxl_base + cxl_size;
936         if (pcms->cxl_devices_state.fixed_windows) {
937             hwaddr cxl_fmw_base;
938             GList *it;
939 
940             cxl_fmw_base = ROUND_UP(cxl_base + cxl_size, 256 * MiB);
941             for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
942                 CXLFixedWindow *fw = it->data;
943 
944                 fw->base = cxl_fmw_base;
945                 memory_region_init_io(&fw->mr, OBJECT(machine), &cfmws_ops, fw,
946                                       "cxl-fixed-memory-region", fw->size);
947                 memory_region_add_subregion(system_memory, fw->base, &fw->mr);
948                 cxl_fmw_base += fw->size;
949                 cxl_resv_end = cxl_fmw_base;
950             }
951         }
952     }
953 
954     /* Initialize PC system firmware */
955     pc_system_firmware_init(pcms, rom_memory);
956 
957     option_rom_mr = g_malloc(sizeof(*option_rom_mr));
958     if (machine_require_guest_memfd(machine)) {
959         memory_region_init_ram_guest_memfd(option_rom_mr, NULL, "pc.rom",
960                                            PC_ROM_SIZE, &error_fatal);
961     } else {
962         memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE,
963                                &error_fatal);
964         if (pcmc->pci_enabled) {
965             memory_region_set_readonly(option_rom_mr, true);
966         }
967     }
968     memory_region_add_subregion_overlap(rom_memory,
969                                         PC_ROM_MIN_VGA,
970                                         option_rom_mr,
971                                         1);
972 
973     fw_cfg = fw_cfg_arch_create(machine,
974                                 x86ms->boot_cpus, x86ms->apic_id_limit);
975 
976     rom_set_fw(fw_cfg);
977 
978     if (machine->device_memory) {
979         uint64_t *val = g_malloc(sizeof(*val));
980         uint64_t res_mem_end;
981 
982         if (pcms->cxl_devices_state.is_enabled) {
983             res_mem_end = cxl_resv_end;
984         } else {
985             res_mem_end = machine->device_memory->base
986                           + memory_region_size(&machine->device_memory->mr);
987         }
988         *val = cpu_to_le64(ROUND_UP(res_mem_end, 1 * GiB));
989         fw_cfg_add_file(fw_cfg, "etc/reserved-memory-end", val, sizeof(*val));
990     }
991 
992     if (linux_boot) {
993         x86_load_linux(x86ms, fw_cfg, PC_FW_DATA, pcmc->pvh_enabled);
994     }
995 
996     for (i = 0; i < nb_option_roms; i++) {
997         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
998     }
999     x86ms->fw_cfg = fw_cfg;
1000 
1001     /* Init default IOAPIC address space */
1002     x86ms->ioapic_as = &address_space_memory;
1003 
1004     /* Init ACPI memory hotplug IO base address */
1005     pcms->memhp_io_base = ACPI_MEMORY_HOTPLUG_BASE;
1006 }
1007 
1008 /*
1009  * The 64bit pci hole starts after "above 4G RAM" and
1010  * potentially the space reserved for memory hotplug.
1011  */
1012 uint64_t pc_pci_hole64_start(void)
1013 {
1014     PCMachineState *pcms = PC_MACHINE(qdev_get_machine());
1015     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1016     MachineState *ms = MACHINE(pcms);
1017     uint64_t hole64_start = 0;
1018     ram_addr_t size = 0;
1019 
1020     if (pcms->cxl_devices_state.is_enabled) {
1021         hole64_start = pc_get_cxl_range_end(pcms);
1022     } else if (pcmc->has_reserved_memory && (ms->ram_size < ms->maxram_size)) {
1023         pc_get_device_memory_range(pcms, &hole64_start, &size);
1024         hole64_start += size;
1025     } else {
1026         hole64_start = pc_above_4g_end(pcms);
1027     }
1028 
1029     return ROUND_UP(hole64_start, 1 * GiB);
1030 }
1031 
1032 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1033 {
1034     DeviceState *dev = NULL;
1035 
1036     if (pci_bus) {
1037         PCIDevice *pcidev = pci_vga_init(pci_bus);
1038         dev = pcidev ? &pcidev->qdev : NULL;
1039     } else if (isa_bus) {
1040         ISADevice *isadev = isa_vga_init(isa_bus);
1041         dev = isadev ? DEVICE(isadev) : NULL;
1042     }
1043 
1044     return dev;
1045 }
1046 
1047 static const MemoryRegionOps ioport80_io_ops = {
1048     .write = ioport80_write,
1049     .read = ioport80_read,
1050     .endianness = DEVICE_LITTLE_ENDIAN,
1051     .impl = {
1052         .min_access_size = 1,
1053         .max_access_size = 1,
1054     },
1055 };
1056 
1057 static const MemoryRegionOps ioportF0_io_ops = {
1058     .write = ioportF0_write,
1059     .read = ioportF0_read,
1060     .endianness = DEVICE_LITTLE_ENDIAN,
1061     .impl = {
1062         .min_access_size = 1,
1063         .max_access_size = 1,
1064     },
1065 };
1066 
1067 static void pc_superio_init(ISABus *isa_bus, bool create_fdctrl,
1068                             bool create_i8042, bool no_vmport, Error **errp)
1069 {
1070     int i;
1071     DriveInfo *fd[MAX_FD];
1072     qemu_irq *a20_line;
1073     ISADevice *i8042, *port92, *vmmouse;
1074 
1075     serial_hds_isa_init(isa_bus, 0, MAX_ISA_SERIAL_PORTS);
1076     parallel_hds_isa_init(isa_bus, MAX_PARALLEL_PORTS);
1077 
1078     for (i = 0; i < MAX_FD; i++) {
1079         fd[i] = drive_get(IF_FLOPPY, 0, i);
1080         create_fdctrl |= !!fd[i];
1081     }
1082     if (create_fdctrl) {
1083 #ifdef CONFIG_FDC_ISA
1084         ISADevice *fdc = isa_new(TYPE_ISA_FDC);
1085         if (fdc) {
1086             isa_realize_and_unref(fdc, isa_bus, &error_fatal);
1087             isa_fdc_init_drives(fdc, fd);
1088         }
1089 #endif
1090     }
1091 
1092     if (!create_i8042) {
1093         if (!no_vmport) {
1094             error_setg(errp,
1095                        "vmport requires the i8042 controller to be enabled");
1096         }
1097         return;
1098     }
1099 
1100     i8042 = isa_create_simple(isa_bus, TYPE_I8042);
1101     if (!no_vmport) {
1102         isa_create_simple(isa_bus, TYPE_VMPORT);
1103         vmmouse = isa_try_new("vmmouse");
1104     } else {
1105         vmmouse = NULL;
1106     }
1107     if (vmmouse) {
1108         object_property_set_link(OBJECT(vmmouse), TYPE_I8042, OBJECT(i8042),
1109                                  &error_abort);
1110         isa_realize_and_unref(vmmouse, isa_bus, &error_fatal);
1111     }
1112     port92 = isa_create_simple(isa_bus, TYPE_PORT92);
1113 
1114     a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1115     qdev_connect_gpio_out_named(DEVICE(i8042),
1116                                 I8042_A20_LINE, 0, a20_line[0]);
1117     qdev_connect_gpio_out_named(DEVICE(port92),
1118                                 PORT92_A20_LINE, 0, a20_line[1]);
1119     g_free(a20_line);
1120 }
1121 
1122 void pc_basic_device_init(struct PCMachineState *pcms,
1123                           ISABus *isa_bus, qemu_irq *gsi,
1124                           ISADevice *rtc_state,
1125                           bool create_fdctrl,
1126                           uint32_t hpet_irqs)
1127 {
1128     int i;
1129     DeviceState *hpet = NULL;
1130     int pit_isa_irq = 0;
1131     qemu_irq pit_alt_irq = NULL;
1132     ISADevice *pit = NULL;
1133     MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1134     MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
1135     X86MachineState *x86ms = X86_MACHINE(pcms);
1136 
1137     memory_region_init_io(ioport80_io, NULL, &ioport80_io_ops, NULL, "ioport80", 1);
1138     memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
1139 
1140     memory_region_init_io(ioportF0_io, NULL, &ioportF0_io_ops, NULL, "ioportF0", 1);
1141     memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
1142 
1143     /*
1144      * Check if an HPET shall be created.
1145      */
1146     if (pcms->hpet_enabled) {
1147         qemu_irq rtc_irq;
1148 
1149         hpet = qdev_try_new(TYPE_HPET);
1150         if (!hpet) {
1151             error_report("couldn't create HPET device");
1152             exit(1);
1153         }
1154         /*
1155          * For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-*,
1156          * use IRQ16~23, IRQ8 and IRQ2.  If the user has already set
1157          * the property, use whatever mask they specified.
1158          */
1159         uint8_t compat = object_property_get_uint(OBJECT(hpet),
1160                 HPET_INTCAP, NULL);
1161         if (!compat) {
1162             qdev_prop_set_uint32(hpet, HPET_INTCAP, hpet_irqs);
1163         }
1164         sysbus_realize_and_unref(SYS_BUS_DEVICE(hpet), &error_fatal);
1165         sysbus_mmio_map(SYS_BUS_DEVICE(hpet), 0, HPET_BASE);
1166 
1167         for (i = 0; i < IOAPIC_NUM_PINS; i++) {
1168             sysbus_connect_irq(SYS_BUS_DEVICE(hpet), i, gsi[i]);
1169         }
1170         pit_isa_irq = -1;
1171         pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1172         rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1173 
1174         /* overwrite connection created by south bridge */
1175         qdev_connect_gpio_out(DEVICE(rtc_state), 0, rtc_irq);
1176     }
1177 
1178     object_property_add_alias(OBJECT(pcms), "rtc-time", OBJECT(rtc_state),
1179                               "date");
1180 
1181 #ifdef CONFIG_XEN_EMU
1182     if (xen_mode == XEN_EMULATE) {
1183         xen_overlay_create();
1184         xen_evtchn_create(IOAPIC_NUM_PINS, gsi);
1185         xen_gnttab_create();
1186         xen_xenstore_create();
1187         if (pcms->pcibus) {
1188             pci_create_simple(pcms->pcibus, -1, "xen-platform");
1189         }
1190         xen_bus_init();
1191     }
1192 #endif
1193 
1194     qemu_register_boot_set(pc_boot_set, pcms);
1195     set_boot_dev(pcms, MC146818_RTC(rtc_state),
1196                  MACHINE(pcms)->boot_config.order, &error_fatal);
1197 
1198     if (!xen_enabled() &&
1199         (x86ms->pit == ON_OFF_AUTO_AUTO || x86ms->pit == ON_OFF_AUTO_ON)) {
1200         if (kvm_pit_in_kernel()) {
1201             pit = kvm_pit_init(isa_bus, 0x40);
1202         } else {
1203             pit = i8254_pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1204         }
1205         if (hpet) {
1206             /* connect PIT to output control line of the HPET */
1207             qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(DEVICE(pit), 0));
1208         }
1209         object_property_set_link(OBJECT(pcms->pcspk), "pit",
1210                                  OBJECT(pit), &error_fatal);
1211         isa_realize_and_unref(pcms->pcspk, isa_bus, &error_fatal);
1212     }
1213 
1214     if (pcms->vmport == ON_OFF_AUTO_AUTO) {
1215         pcms->vmport = (xen_enabled() || !pcms->i8042_enabled)
1216             ? ON_OFF_AUTO_OFF : ON_OFF_AUTO_ON;
1217     }
1218 
1219     /* Super I/O */
1220     pc_superio_init(isa_bus, create_fdctrl, pcms->i8042_enabled,
1221                     pcms->vmport != ON_OFF_AUTO_ON, &error_fatal);
1222 
1223     pcms->machine_done.notify = pc_machine_done;
1224     qemu_add_machine_init_done_notifier(&pcms->machine_done);
1225 }
1226 
1227 void pc_nic_init(PCMachineClass *pcmc, ISABus *isa_bus, PCIBus *pci_bus)
1228 {
1229     MachineClass *mc = MACHINE_CLASS(pcmc);
1230     bool default_is_ne2k = g_str_equal(mc->default_nic, TYPE_ISA_NE2000);
1231     NICInfo *nd;
1232 
1233     while ((nd = qemu_find_nic_info(TYPE_ISA_NE2000, default_is_ne2k, NULL))) {
1234         pc_init_ne2k_isa(isa_bus, nd, &error_fatal);
1235     }
1236 
1237     /* Anything remaining should be a PCI NIC */
1238     if (pci_bus) {
1239         pci_init_nic_devices(pci_bus, mc->default_nic);
1240     }
1241 }
1242 
1243 void pc_i8259_create(ISABus *isa_bus, qemu_irq *i8259_irqs)
1244 {
1245     qemu_irq *i8259;
1246 
1247     if (kvm_pic_in_kernel()) {
1248         i8259 = kvm_i8259_init(isa_bus);
1249     } else if (xen_enabled()) {
1250         i8259 = xen_interrupt_controller_init();
1251     } else {
1252         i8259 = i8259_init(isa_bus, x86_allocate_cpu_irq());
1253     }
1254 
1255     for (size_t i = 0; i < ISA_NUM_IRQS; i++) {
1256         i8259_irqs[i] = i8259[i];
1257     }
1258 
1259     g_free(i8259);
1260 }
1261 
1262 static void pc_memory_pre_plug(HotplugHandler *hotplug_dev, DeviceState *dev,
1263                                Error **errp)
1264 {
1265     const X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1266     const MachineState *ms = MACHINE(hotplug_dev);
1267     const bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1268     Error *local_err = NULL;
1269 
1270     /*
1271      * When "acpi=off" is used with the Q35 machine type, no ACPI is built,
1272      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1273      * addition to cover this case.
1274      */
1275     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1276         error_setg(errp,
1277                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1278         return;
1279     }
1280 
1281     if (is_nvdimm && !ms->nvdimms_state->is_enabled) {
1282         error_setg(errp, "nvdimm is not enabled: missing 'nvdimm' in '-M'");
1283         return;
1284     }
1285 
1286     hotplug_handler_pre_plug(x86ms->acpi_dev, dev, &local_err);
1287     if (local_err) {
1288         error_propagate(errp, local_err);
1289         return;
1290     }
1291 
1292     pc_dimm_pre_plug(PC_DIMM(dev), MACHINE(hotplug_dev), errp);
1293 }
1294 
1295 static void pc_memory_plug(HotplugHandler *hotplug_dev,
1296                            DeviceState *dev, Error **errp)
1297 {
1298     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1299     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1300     MachineState *ms = MACHINE(hotplug_dev);
1301     bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1302 
1303     pc_dimm_plug(PC_DIMM(dev), MACHINE(pcms));
1304 
1305     if (is_nvdimm) {
1306         nvdimm_plug(ms->nvdimms_state);
1307     }
1308 
1309     hotplug_handler_plug(x86ms->acpi_dev, dev, &error_abort);
1310 }
1311 
1312 static void pc_memory_unplug_request(HotplugHandler *hotplug_dev,
1313                                      DeviceState *dev, Error **errp)
1314 {
1315     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1316 
1317     /*
1318      * When "acpi=off" is used with the Q35 machine type, no ACPI is built,
1319      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1320      * addition to cover this case.
1321      */
1322     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1323         error_setg(errp,
1324                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1325         return;
1326     }
1327 
1328     if (object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM)) {
1329         error_setg(errp, "nvdimm device hot unplug is not supported yet.");
1330         return;
1331     }
1332 
1333     hotplug_handler_unplug_request(x86ms->acpi_dev, dev,
1334                                    errp);
1335 }
1336 
1337 static void pc_memory_unplug(HotplugHandler *hotplug_dev,
1338                              DeviceState *dev, Error **errp)
1339 {
1340     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1341     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1342     Error *local_err = NULL;
1343 
1344     hotplug_handler_unplug(x86ms->acpi_dev, dev, &local_err);
1345     if (local_err) {
1346         goto out;
1347     }
1348 
1349     pc_dimm_unplug(PC_DIMM(dev), MACHINE(pcms));
1350     qdev_unrealize(dev);
1351  out:
1352     error_propagate(errp, local_err);
1353 }
1354 
1355 static void pc_hv_balloon_pre_plug(HotplugHandler *hotplug_dev,
1356                                    DeviceState *dev, Error **errp)
1357 {
1358     /* The vmbus handler has no hotplug handler; we should never end up here. */
1359     g_assert(!dev->hotplugged);
1360     memory_device_pre_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev), errp);
1361 }
1362 
1363 static void pc_hv_balloon_plug(HotplugHandler *hotplug_dev,
1364                                DeviceState *dev, Error **errp)
1365 {
1366     memory_device_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev));
1367 }
1368 
1369 static void pc_machine_device_pre_plug_cb(HotplugHandler *hotplug_dev,
1370                                           DeviceState *dev, Error **errp)
1371 {
1372     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1373         pc_memory_pre_plug(hotplug_dev, dev, errp);
1374     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1375         x86_cpu_pre_plug(hotplug_dev, dev, errp);
1376     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1377         virtio_md_pci_pre_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1378     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1379         /* Declare the APIC range as the reserved MSI region */
1380         char *resv_prop_str = g_strdup_printf("0xfee00000:0xfeefffff:%d",
1381                                               VIRTIO_IOMMU_RESV_MEM_T_MSI);
1382         QList *reserved_regions = qlist_new();
1383 
1384         qlist_append_str(reserved_regions, resv_prop_str);
1385         qdev_prop_set_array(dev, "reserved-regions", reserved_regions);
1386 
1387         g_free(resv_prop_str);
1388     }
1389 
1390     if (object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE) ||
1391         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1392         PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1393 
1394         if (pcms->iommu) {
1395             error_setg(errp, "QEMU does not support multiple vIOMMUs "
1396                        "for x86 yet.");
1397             return;
1398         }
1399         pcms->iommu = dev;
1400     } else if (object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON)) {
1401         pc_hv_balloon_pre_plug(hotplug_dev, dev, errp);
1402     }
1403 }
1404 
1405 static void pc_machine_device_plug_cb(HotplugHandler *hotplug_dev,
1406                                       DeviceState *dev, Error **errp)
1407 {
1408     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1409         pc_memory_plug(hotplug_dev, dev, errp);
1410     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1411         x86_cpu_plug(hotplug_dev, dev, errp);
1412     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1413         virtio_md_pci_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1414     } else if (object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON)) {
1415         pc_hv_balloon_plug(hotplug_dev, dev, errp);
1416     }
1417 }
1418 
1419 static void pc_machine_device_unplug_request_cb(HotplugHandler *hotplug_dev,
1420                                                 DeviceState *dev, Error **errp)
1421 {
1422     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1423         pc_memory_unplug_request(hotplug_dev, dev, errp);
1424     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1425         x86_cpu_unplug_request_cb(hotplug_dev, dev, errp);
1426     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1427         virtio_md_pci_unplug_request(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev),
1428                                      errp);
1429     } else {
1430         error_setg(errp, "acpi: device unplug request for not supported device"
1431                    " type: %s", object_get_typename(OBJECT(dev)));
1432     }
1433 }
1434 
1435 static void pc_machine_device_unplug_cb(HotplugHandler *hotplug_dev,
1436                                         DeviceState *dev, Error **errp)
1437 {
1438     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1439         pc_memory_unplug(hotplug_dev, dev, errp);
1440     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1441         x86_cpu_unplug_cb(hotplug_dev, dev, errp);
1442     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1443         virtio_md_pci_unplug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1444     } else {
1445         error_setg(errp, "acpi: device unplug for not supported device"
1446                    " type: %s", object_get_typename(OBJECT(dev)));
1447     }
1448 }
1449 
1450 static HotplugHandler *pc_get_hotplug_handler(MachineState *machine,
1451                                              DeviceState *dev)
1452 {
1453     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM) ||
1454         object_dynamic_cast(OBJECT(dev), TYPE_CPU) ||
1455         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI) ||
1456         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI) ||
1457         object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON) ||
1458         object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE)) {
1459         return HOTPLUG_HANDLER(machine);
1460     }
1461 
1462     return NULL;
1463 }
1464 
1465 static void pc_machine_get_vmport(Object *obj, Visitor *v, const char *name,
1466                                   void *opaque, Error **errp)
1467 {
1468     PCMachineState *pcms = PC_MACHINE(obj);
1469     OnOffAuto vmport = pcms->vmport;
1470 
1471     visit_type_OnOffAuto(v, name, &vmport, errp);
1472 }
1473 
1474 static void pc_machine_set_vmport(Object *obj, Visitor *v, const char *name,
1475                                   void *opaque, Error **errp)
1476 {
1477     PCMachineState *pcms = PC_MACHINE(obj);
1478 
1479     visit_type_OnOffAuto(v, name, &pcms->vmport, errp);
1480 }
1481 
1482 static bool pc_machine_get_fd_bootchk(Object *obj, Error **errp)
1483 {
1484     PCMachineState *pcms = PC_MACHINE(obj);
1485 
1486     return pcms->fd_bootchk;
1487 }
1488 
1489 static void pc_machine_set_fd_bootchk(Object *obj, bool value, Error **errp)
1490 {
1491     PCMachineState *pcms = PC_MACHINE(obj);
1492 
1493     pcms->fd_bootchk = value;
1494 }
1495 
1496 static bool pc_machine_get_smbus(Object *obj, Error **errp)
1497 {
1498     PCMachineState *pcms = PC_MACHINE(obj);
1499 
1500     return pcms->smbus_enabled;
1501 }
1502 
1503 static void pc_machine_set_smbus(Object *obj, bool value, Error **errp)
1504 {
1505     PCMachineState *pcms = PC_MACHINE(obj);
1506 
1507     pcms->smbus_enabled = value;
1508 }
1509 
1510 static bool pc_machine_get_sata(Object *obj, Error **errp)
1511 {
1512     PCMachineState *pcms = PC_MACHINE(obj);
1513 
1514     return pcms->sata_enabled;
1515 }
1516 
1517 static void pc_machine_set_sata(Object *obj, bool value, Error **errp)
1518 {
1519     PCMachineState *pcms = PC_MACHINE(obj);
1520 
1521     pcms->sata_enabled = value;
1522 }
1523 
1524 static bool pc_machine_get_hpet(Object *obj, Error **errp)
1525 {
1526     PCMachineState *pcms = PC_MACHINE(obj);
1527 
1528     return pcms->hpet_enabled;
1529 }
1530 
1531 static void pc_machine_set_hpet(Object *obj, bool value, Error **errp)
1532 {
1533     PCMachineState *pcms = PC_MACHINE(obj);
1534 
1535     pcms->hpet_enabled = value;
1536 }
1537 
1538 static bool pc_machine_get_i8042(Object *obj, Error **errp)
1539 {
1540     PCMachineState *pcms = PC_MACHINE(obj);
1541 
1542     return pcms->i8042_enabled;
1543 }
1544 
1545 static void pc_machine_set_i8042(Object *obj, bool value, Error **errp)
1546 {
1547     PCMachineState *pcms = PC_MACHINE(obj);
1548 
1549     pcms->i8042_enabled = value;
1550 }
1551 
1552 static bool pc_machine_get_default_bus_bypass_iommu(Object *obj, Error **errp)
1553 {
1554     PCMachineState *pcms = PC_MACHINE(obj);
1555 
1556     return pcms->default_bus_bypass_iommu;
1557 }
1558 
1559 static void pc_machine_set_default_bus_bypass_iommu(Object *obj, bool value,
1560                                                     Error **errp)
1561 {
1562     PCMachineState *pcms = PC_MACHINE(obj);
1563 
1564     pcms->default_bus_bypass_iommu = value;
1565 }
1566 
1567 static void pc_machine_get_smbios_ep(Object *obj, Visitor *v, const char *name,
1568                                      void *opaque, Error **errp)
1569 {
1570     PCMachineState *pcms = PC_MACHINE(obj);
1571     SmbiosEntryPointType smbios_entry_point_type = pcms->smbios_entry_point_type;
1572 
1573     visit_type_SmbiosEntryPointType(v, name, &smbios_entry_point_type, errp);
1574 }
1575 
1576 static void pc_machine_set_smbios_ep(Object *obj, Visitor *v, const char *name,
1577                                      void *opaque, Error **errp)
1578 {
1579     PCMachineState *pcms = PC_MACHINE(obj);
1580 
1581     visit_type_SmbiosEntryPointType(v, name, &pcms->smbios_entry_point_type, errp);
1582 }
1583 
1584 static void pc_machine_get_max_ram_below_4g(Object *obj, Visitor *v,
1585                                             const char *name, void *opaque,
1586                                             Error **errp)
1587 {
1588     PCMachineState *pcms = PC_MACHINE(obj);
1589     uint64_t value = pcms->max_ram_below_4g;
1590 
1591     visit_type_size(v, name, &value, errp);
1592 }
1593 
1594 static void pc_machine_set_max_ram_below_4g(Object *obj, Visitor *v,
1595                                             const char *name, void *opaque,
1596                                             Error **errp)
1597 {
1598     PCMachineState *pcms = PC_MACHINE(obj);
1599     uint64_t value;
1600 
1601     if (!visit_type_size(v, name, &value, errp)) {
1602         return;
1603     }
1604     if (value > 4 * GiB) {
1605         error_setg(errp,
1606                    "Machine option 'max-ram-below-4g=%"PRIu64
1607                    "' expects size less than or equal to 4G", value);
1608         return;
1609     }
1610 
1611     if (value < 1 * MiB) {
1612         warn_report("Only %" PRIu64 " bytes of RAM below the 4GiB boundary,"
1613                     "BIOS may not work with less than 1MiB", value);
1614     }
1615 
1616     pcms->max_ram_below_4g = value;
1617 }
1618 
1619 static void pc_machine_get_max_fw_size(Object *obj, Visitor *v,
1620                                        const char *name, void *opaque,
1621                                        Error **errp)
1622 {
1623     PCMachineState *pcms = PC_MACHINE(obj);
1624     uint64_t value = pcms->max_fw_size;
1625 
1626     visit_type_size(v, name, &value, errp);
1627 }
1628 
1629 static void pc_machine_set_max_fw_size(Object *obj, Visitor *v,
1630                                        const char *name, void *opaque,
1631                                        Error **errp)
1632 {
1633     PCMachineState *pcms = PC_MACHINE(obj);
1634     uint64_t value;
1635 
1636     if (!visit_type_size(v, name, &value, errp)) {
1637         return;
1638     }
1639 
1640     /*
1641      * We don't have a theoretically justifiable exact lower bound on the base
1642      * address of any flash mapping. In practice, the IO-APIC MMIO range is
1643      * [0xFEE00000..0xFEE01000] -- see IO_APIC_DEFAULT_ADDRESS --, leaving free
1644      * only 18MiB-4KiB below 4GiB. For now, restrict the cumulative mapping to
1645      * 16MiB in size.
1646      */
1647     if (value > 16 * MiB) {
1648         error_setg(errp,
1649                    "User specified max allowed firmware size %" PRIu64 " is "
1650                    "greater than 16MiB. If combined firmware size exceeds "
1651                    "16MiB the system may not boot, or experience intermittent"
1652                    "stability issues.",
1653                    value);
1654         return;
1655     }
1656 
1657     pcms->max_fw_size = value;
1658 }
1659 
1660 
1661 static void pc_machine_initfn(Object *obj)
1662 {
1663     PCMachineState *pcms = PC_MACHINE(obj);
1664     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1665 
1666 #ifdef CONFIG_VMPORT
1667     pcms->vmport = ON_OFF_AUTO_AUTO;
1668 #else
1669     pcms->vmport = ON_OFF_AUTO_OFF;
1670 #endif /* CONFIG_VMPORT */
1671     pcms->max_ram_below_4g = 0; /* use default */
1672     pcms->smbios_entry_point_type = pcmc->default_smbios_ep_type;
1673     pcms->south_bridge = pcmc->default_south_bridge;
1674 
1675     /* acpi build is enabled by default if machine supports it */
1676     pcms->acpi_build_enabled = pcmc->has_acpi_build;
1677     pcms->smbus_enabled = true;
1678     pcms->sata_enabled = true;
1679     pcms->i8042_enabled = true;
1680     pcms->max_fw_size = 8 * MiB;
1681 #if defined(CONFIG_HPET)
1682     pcms->hpet_enabled = true;
1683 #endif
1684     pcms->fd_bootchk = true;
1685     pcms->default_bus_bypass_iommu = false;
1686 
1687     pc_system_flash_create(pcms);
1688     pcms->pcspk = isa_new(TYPE_PC_SPEAKER);
1689     object_property_add_alias(OBJECT(pcms), "pcspk-audiodev",
1690                               OBJECT(pcms->pcspk), "audiodev");
1691     if (pcmc->pci_enabled) {
1692         cxl_machine_init(obj, &pcms->cxl_devices_state);
1693     }
1694 }
1695 
1696 static void pc_machine_reset(MachineState *machine, ResetType type)
1697 {
1698     CPUState *cs;
1699     X86CPU *cpu;
1700 
1701     qemu_devices_reset(type);
1702 
1703     /* Reset APIC after devices have been reset to cancel
1704      * any changes that qemu_devices_reset() might have done.
1705      */
1706     CPU_FOREACH(cs) {
1707         cpu = X86_CPU(cs);
1708 
1709         x86_cpu_after_reset(cpu);
1710     }
1711 }
1712 
1713 static void pc_machine_wakeup(MachineState *machine)
1714 {
1715     cpu_synchronize_all_states();
1716     pc_machine_reset(machine, RESET_TYPE_WAKEUP);
1717     cpu_synchronize_all_post_reset();
1718 }
1719 
1720 static bool pc_hotplug_allowed(MachineState *ms, DeviceState *dev, Error **errp)
1721 {
1722     X86IOMMUState *iommu = x86_iommu_get_default();
1723     IntelIOMMUState *intel_iommu;
1724 
1725     if (iommu &&
1726         object_dynamic_cast((Object *)iommu, TYPE_INTEL_IOMMU_DEVICE) &&
1727         object_dynamic_cast((Object *)dev, "vfio-pci")) {
1728         intel_iommu = INTEL_IOMMU_DEVICE(iommu);
1729         if (!intel_iommu->caching_mode) {
1730             error_setg(errp, "Device assignment is not allowed without "
1731                        "enabling caching-mode=on for Intel IOMMU.");
1732             return false;
1733         }
1734     }
1735 
1736     return true;
1737 }
1738 
1739 static void pc_machine_class_init(ObjectClass *oc, const void *data)
1740 {
1741     MachineClass *mc = MACHINE_CLASS(oc);
1742     X86MachineClass *x86mc = X86_MACHINE_CLASS(oc);
1743     PCMachineClass *pcmc = PC_MACHINE_CLASS(oc);
1744     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
1745 
1746     pcmc->pci_enabled = true;
1747     pcmc->has_acpi_build = true;
1748     pcmc->smbios_defaults = true;
1749     pcmc->gigabyte_align = true;
1750     pcmc->has_reserved_memory = true;
1751     pcmc->enforce_amd_1tb_hole = true;
1752     pcmc->isa_bios_alias = true;
1753     pcmc->pvh_enabled = true;
1754     pcmc->kvmclock_create_always = true;
1755     x86mc->apic_xrupt_override = true;
1756     assert(!mc->get_hotplug_handler);
1757     mc->get_hotplug_handler = pc_get_hotplug_handler;
1758     mc->hotplug_allowed = pc_hotplug_allowed;
1759     mc->auto_enable_numa_with_memhp = true;
1760     mc->auto_enable_numa_with_memdev = true;
1761     mc->has_hotpluggable_cpus = true;
1762     mc->default_boot_order = "cad";
1763     mc->block_default_type = IF_IDE;
1764     mc->max_cpus = 255;
1765     mc->reset = pc_machine_reset;
1766     mc->wakeup = pc_machine_wakeup;
1767     hc->pre_plug = pc_machine_device_pre_plug_cb;
1768     hc->plug = pc_machine_device_plug_cb;
1769     hc->unplug_request = pc_machine_device_unplug_request_cb;
1770     hc->unplug = pc_machine_device_unplug_cb;
1771     mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE;
1772     mc->nvdimm_supported = true;
1773     mc->smp_props.dies_supported = true;
1774     mc->smp_props.modules_supported = true;
1775     mc->smp_props.cache_supported[CACHE_LEVEL_AND_TYPE_L1D] = true;
1776     mc->smp_props.cache_supported[CACHE_LEVEL_AND_TYPE_L1I] = true;
1777     mc->smp_props.cache_supported[CACHE_LEVEL_AND_TYPE_L2] = true;
1778     mc->smp_props.cache_supported[CACHE_LEVEL_AND_TYPE_L3] = true;
1779     mc->default_ram_id = "pc.ram";
1780     pcmc->default_smbios_ep_type = SMBIOS_ENTRY_POINT_TYPE_AUTO;
1781 
1782     object_class_property_add(oc, PC_MACHINE_MAX_RAM_BELOW_4G, "size",
1783         pc_machine_get_max_ram_below_4g, pc_machine_set_max_ram_below_4g,
1784         NULL, NULL);
1785     object_class_property_set_description(oc, PC_MACHINE_MAX_RAM_BELOW_4G,
1786         "Maximum ram below the 4G boundary (32bit boundary)");
1787 
1788     object_class_property_add(oc, PC_MACHINE_VMPORT, "OnOffAuto",
1789         pc_machine_get_vmport, pc_machine_set_vmport,
1790         NULL, NULL);
1791     object_class_property_set_description(oc, PC_MACHINE_VMPORT,
1792         "Enable vmport (pc & q35)");
1793 
1794     object_class_property_add_bool(oc, PC_MACHINE_SMBUS,
1795         pc_machine_get_smbus, pc_machine_set_smbus);
1796     object_class_property_set_description(oc, PC_MACHINE_SMBUS,
1797         "Enable/disable system management bus");
1798 
1799     object_class_property_add_bool(oc, PC_MACHINE_SATA,
1800         pc_machine_get_sata, pc_machine_set_sata);
1801     object_class_property_set_description(oc, PC_MACHINE_SATA,
1802         "Enable/disable Serial ATA bus");
1803 
1804     object_class_property_add_bool(oc, "hpet",
1805         pc_machine_get_hpet, pc_machine_set_hpet);
1806     object_class_property_set_description(oc, "hpet",
1807         "Enable/disable high precision event timer emulation");
1808 
1809     object_class_property_add_bool(oc, PC_MACHINE_I8042,
1810         pc_machine_get_i8042, pc_machine_set_i8042);
1811     object_class_property_set_description(oc, PC_MACHINE_I8042,
1812         "Enable/disable Intel 8042 PS/2 controller emulation");
1813 
1814     object_class_property_add_bool(oc, "default-bus-bypass-iommu",
1815         pc_machine_get_default_bus_bypass_iommu,
1816         pc_machine_set_default_bus_bypass_iommu);
1817 
1818     object_class_property_add(oc, PC_MACHINE_MAX_FW_SIZE, "size",
1819         pc_machine_get_max_fw_size, pc_machine_set_max_fw_size,
1820         NULL, NULL);
1821     object_class_property_set_description(oc, PC_MACHINE_MAX_FW_SIZE,
1822         "Maximum combined firmware size");
1823 
1824     object_class_property_add(oc, PC_MACHINE_SMBIOS_EP, "str",
1825         pc_machine_get_smbios_ep, pc_machine_set_smbios_ep,
1826         NULL, NULL);
1827     object_class_property_set_description(oc, PC_MACHINE_SMBIOS_EP,
1828         "SMBIOS Entry Point type [32, 64]");
1829 
1830     object_class_property_add_bool(oc, "fd-bootchk",
1831         pc_machine_get_fd_bootchk,
1832         pc_machine_set_fd_bootchk);
1833 }
1834 
1835 static const TypeInfo pc_machine_info = {
1836     .name = TYPE_PC_MACHINE,
1837     .parent = TYPE_X86_MACHINE,
1838     .abstract = true,
1839     .instance_size = sizeof(PCMachineState),
1840     .instance_init = pc_machine_initfn,
1841     .class_size = sizeof(PCMachineClass),
1842     .class_init = pc_machine_class_init,
1843     .interfaces = (const InterfaceInfo[]) {
1844          { TYPE_HOTPLUG_HANDLER },
1845          { }
1846     },
1847 };
1848 
1849 static void pc_machine_register_types(void)
1850 {
1851     type_register_static(&pc_machine_info);
1852 }
1853 
1854 type_init(pc_machine_register_types)
1855