xref: /openbmc/qemu/hw/i386/pc.c (revision 33a24910)
1 /*
2  * QEMU PC System Emulator
3  *
4  * Copyright (c) 2003-2004 Fabrice Bellard
5  *
6  * Permission is hereby granted, free of charge, to any person obtaining a copy
7  * of this software and associated documentation files (the "Software"), to deal
8  * in the Software without restriction, including without limitation the rights
9  * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
10  * copies of the Software, and to permit persons to whom the Software is
11  * furnished to do so, subject to the following conditions:
12  *
13  * The above copyright notice and this permission notice shall be included in
14  * all copies or substantial portions of the Software.
15  *
16  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
19  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
22  * THE SOFTWARE.
23  */
24 
25 #include "qemu/osdep.h"
26 #include "qemu/units.h"
27 #include "hw/i386/pc.h"
28 #include "hw/char/serial.h"
29 #include "hw/char/parallel.h"
30 #include "hw/hyperv/hv-balloon.h"
31 #include "hw/i386/fw_cfg.h"
32 #include "hw/i386/vmport.h"
33 #include "sysemu/cpus.h"
34 #include "hw/ide/ide-bus.h"
35 #include "hw/timer/hpet.h"
36 #include "hw/loader.h"
37 #include "hw/rtc/mc146818rtc.h"
38 #include "hw/intc/i8259.h"
39 #include "hw/timer/i8254.h"
40 #include "hw/input/i8042.h"
41 #include "hw/audio/pcspk.h"
42 #include "sysemu/sysemu.h"
43 #include "sysemu/xen.h"
44 #include "sysemu/reset.h"
45 #include "kvm/kvm_i386.h"
46 #include "hw/xen/xen.h"
47 #include "qapi/qmp/qlist.h"
48 #include "qemu/error-report.h"
49 #include "hw/acpi/cpu_hotplug.h"
50 #include "acpi-build.h"
51 #include "hw/mem/nvdimm.h"
52 #include "hw/cxl/cxl_host.h"
53 #include "hw/usb.h"
54 #include "hw/i386/intel_iommu.h"
55 #include "hw/net/ne2000-isa.h"
56 #include "hw/virtio/virtio-iommu.h"
57 #include "hw/virtio/virtio-md-pci.h"
58 #include "hw/i386/kvm/xen_overlay.h"
59 #include "hw/i386/kvm/xen_evtchn.h"
60 #include "hw/i386/kvm/xen_gnttab.h"
61 #include "hw/i386/kvm/xen_xenstore.h"
62 #include "hw/mem/memory-device.h"
63 #include "e820_memory_layout.h"
64 #include "trace.h"
65 #include CONFIG_DEVICES
66 
67 #ifdef CONFIG_XEN_EMU
68 #include "hw/xen/xen-legacy-backend.h"
69 #include "hw/xen/xen-bus.h"
70 #endif
71 
72 /*
73  * Helper for setting model-id for CPU models that changed model-id
74  * depending on QEMU versions up to QEMU 2.4.
75  */
76 #define PC_CPU_MODEL_IDS(v) \
77     { "qemu32-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
78     { "qemu64-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
79     { "athlon-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },
80 
81 GlobalProperty pc_compat_8_2[] = {};
82 const size_t pc_compat_8_2_len = G_N_ELEMENTS(pc_compat_8_2);
83 
84 GlobalProperty pc_compat_8_1[] = {};
85 const size_t pc_compat_8_1_len = G_N_ELEMENTS(pc_compat_8_1);
86 
87 GlobalProperty pc_compat_8_0[] = {
88     { "virtio-mem", "unplugged-inaccessible", "auto" },
89 };
90 const size_t pc_compat_8_0_len = G_N_ELEMENTS(pc_compat_8_0);
91 
92 GlobalProperty pc_compat_7_2[] = {
93     { "ICH9-LPC", "noreboot", "true" },
94 };
95 const size_t pc_compat_7_2_len = G_N_ELEMENTS(pc_compat_7_2);
96 
97 GlobalProperty pc_compat_7_1[] = {};
98 const size_t pc_compat_7_1_len = G_N_ELEMENTS(pc_compat_7_1);
99 
100 GlobalProperty pc_compat_7_0[] = {};
101 const size_t pc_compat_7_0_len = G_N_ELEMENTS(pc_compat_7_0);
102 
103 GlobalProperty pc_compat_6_2[] = {
104     { "virtio-mem", "unplugged-inaccessible", "off" },
105 };
106 const size_t pc_compat_6_2_len = G_N_ELEMENTS(pc_compat_6_2);
107 
108 GlobalProperty pc_compat_6_1[] = {
109     { TYPE_X86_CPU, "hv-version-id-build", "0x1bbc" },
110     { TYPE_X86_CPU, "hv-version-id-major", "0x0006" },
111     { TYPE_X86_CPU, "hv-version-id-minor", "0x0001" },
112     { "ICH9-LPC", "x-keep-pci-slot-hpc", "false" },
113 };
114 const size_t pc_compat_6_1_len = G_N_ELEMENTS(pc_compat_6_1);
115 
116 GlobalProperty pc_compat_6_0[] = {
117     { "qemu64" "-" TYPE_X86_CPU, "family", "6" },
118     { "qemu64" "-" TYPE_X86_CPU, "model", "6" },
119     { "qemu64" "-" TYPE_X86_CPU, "stepping", "3" },
120     { TYPE_X86_CPU, "x-vendor-cpuid-only", "off" },
121     { "ICH9-LPC", ACPI_PM_PROP_ACPI_PCIHP_BRIDGE, "off" },
122     { "ICH9-LPC", "x-keep-pci-slot-hpc", "true" },
123 };
124 const size_t pc_compat_6_0_len = G_N_ELEMENTS(pc_compat_6_0);
125 
126 GlobalProperty pc_compat_5_2[] = {
127     { "ICH9-LPC", "x-smi-cpu-hotunplug", "off" },
128 };
129 const size_t pc_compat_5_2_len = G_N_ELEMENTS(pc_compat_5_2);
130 
131 GlobalProperty pc_compat_5_1[] = {
132     { "ICH9-LPC", "x-smi-cpu-hotplug", "off" },
133     { TYPE_X86_CPU, "kvm-msi-ext-dest-id", "off" },
134 };
135 const size_t pc_compat_5_1_len = G_N_ELEMENTS(pc_compat_5_1);
136 
137 GlobalProperty pc_compat_5_0[] = {
138 };
139 const size_t pc_compat_5_0_len = G_N_ELEMENTS(pc_compat_5_0);
140 
141 GlobalProperty pc_compat_4_2[] = {
142     { "mch", "smbase-smram", "off" },
143 };
144 const size_t pc_compat_4_2_len = G_N_ELEMENTS(pc_compat_4_2);
145 
146 GlobalProperty pc_compat_4_1[] = {};
147 const size_t pc_compat_4_1_len = G_N_ELEMENTS(pc_compat_4_1);
148 
149 GlobalProperty pc_compat_4_0[] = {};
150 const size_t pc_compat_4_0_len = G_N_ELEMENTS(pc_compat_4_0);
151 
152 GlobalProperty pc_compat_3_1[] = {
153     { "intel-iommu", "dma-drain", "off" },
154     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "off" },
155     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "off" },
156     { "Opteron_G4" "-" TYPE_X86_CPU, "npt", "off" },
157     { "Opteron_G4" "-" TYPE_X86_CPU, "nrip-save", "off" },
158     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "off" },
159     { "Opteron_G5" "-" TYPE_X86_CPU, "npt", "off" },
160     { "Opteron_G5" "-" TYPE_X86_CPU, "nrip-save", "off" },
161     { "EPYC" "-" TYPE_X86_CPU, "npt", "off" },
162     { "EPYC" "-" TYPE_X86_CPU, "nrip-save", "off" },
163     { "EPYC-IBPB" "-" TYPE_X86_CPU, "npt", "off" },
164     { "EPYC-IBPB" "-" TYPE_X86_CPU, "nrip-save", "off" },
165     { "Skylake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
166     { "Skylake-Client-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
167     { "Skylake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
168     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
169     { "Cascadelake-Server" "-" TYPE_X86_CPU,  "mpx", "on" },
170     { "Icelake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
171     { "Icelake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
172     { "Cascadelake-Server" "-" TYPE_X86_CPU, "stepping", "5" },
173     { TYPE_X86_CPU, "x-intel-pt-auto-level", "off" },
174 };
175 const size_t pc_compat_3_1_len = G_N_ELEMENTS(pc_compat_3_1);
176 
177 GlobalProperty pc_compat_3_0[] = {
178     { TYPE_X86_CPU, "x-hv-synic-kvm-only", "on" },
179     { "Skylake-Server" "-" TYPE_X86_CPU, "pku", "off" },
180     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "pku", "off" },
181 };
182 const size_t pc_compat_3_0_len = G_N_ELEMENTS(pc_compat_3_0);
183 
184 GlobalProperty pc_compat_2_12[] = {
185     { TYPE_X86_CPU, "legacy-cache", "on" },
186     { TYPE_X86_CPU, "topoext", "off" },
187     { "EPYC-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
188     { "EPYC-IBPB-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
189 };
190 const size_t pc_compat_2_12_len = G_N_ELEMENTS(pc_compat_2_12);
191 
192 GlobalProperty pc_compat_2_11[] = {
193     { TYPE_X86_CPU, "x-migrate-smi-count", "off" },
194     { "Skylake-Server" "-" TYPE_X86_CPU, "clflushopt", "off" },
195 };
196 const size_t pc_compat_2_11_len = G_N_ELEMENTS(pc_compat_2_11);
197 
198 GlobalProperty pc_compat_2_10[] = {
199     { TYPE_X86_CPU, "x-hv-max-vps", "0x40" },
200     { "i440FX-pcihost", "x-pci-hole64-fix", "off" },
201     { "q35-pcihost", "x-pci-hole64-fix", "off" },
202 };
203 const size_t pc_compat_2_10_len = G_N_ELEMENTS(pc_compat_2_10);
204 
205 GlobalProperty pc_compat_2_9[] = {
206     { "mch", "extended-tseg-mbytes", "0" },
207 };
208 const size_t pc_compat_2_9_len = G_N_ELEMENTS(pc_compat_2_9);
209 
210 GlobalProperty pc_compat_2_8[] = {
211     { TYPE_X86_CPU, "tcg-cpuid", "off" },
212     { "kvmclock", "x-mach-use-reliable-get-clock", "off" },
213     { "ICH9-LPC", "x-smi-broadcast", "off" },
214     { TYPE_X86_CPU, "vmware-cpuid-freq", "off" },
215     { "Haswell-" TYPE_X86_CPU, "stepping", "1" },
216 };
217 const size_t pc_compat_2_8_len = G_N_ELEMENTS(pc_compat_2_8);
218 
219 GlobalProperty pc_compat_2_7[] = {
220     { TYPE_X86_CPU, "l3-cache", "off" },
221     { TYPE_X86_CPU, "full-cpuid-auto-level", "off" },
222     { "Opteron_G3" "-" TYPE_X86_CPU, "family", "15" },
223     { "Opteron_G3" "-" TYPE_X86_CPU, "model", "6" },
224     { "Opteron_G3" "-" TYPE_X86_CPU, "stepping", "1" },
225     { "isa-pcspk", "migrate", "off" },
226 };
227 const size_t pc_compat_2_7_len = G_N_ELEMENTS(pc_compat_2_7);
228 
229 GlobalProperty pc_compat_2_6[] = {
230     { TYPE_X86_CPU, "cpuid-0xb", "off" },
231     { "vmxnet3", "romfile", "" },
232     { TYPE_X86_CPU, "fill-mtrr-mask", "off" },
233     { "apic-common", "legacy-instance-id", "on", }
234 };
235 const size_t pc_compat_2_6_len = G_N_ELEMENTS(pc_compat_2_6);
236 
237 GlobalProperty pc_compat_2_5[] = {};
238 const size_t pc_compat_2_5_len = G_N_ELEMENTS(pc_compat_2_5);
239 
240 GlobalProperty pc_compat_2_4[] = {
241     PC_CPU_MODEL_IDS("2.4.0")
242     { "Haswell-" TYPE_X86_CPU, "abm", "off" },
243     { "Haswell-noTSX-" TYPE_X86_CPU, "abm", "off" },
244     { "Broadwell-" TYPE_X86_CPU, "abm", "off" },
245     { "Broadwell-noTSX-" TYPE_X86_CPU, "abm", "off" },
246     { "host" "-" TYPE_X86_CPU, "host-cache-info", "on" },
247     { TYPE_X86_CPU, "check", "off" },
248     { "qemu64" "-" TYPE_X86_CPU, "sse4a", "on" },
249     { "qemu64" "-" TYPE_X86_CPU, "abm", "on" },
250     { "qemu64" "-" TYPE_X86_CPU, "popcnt", "on" },
251     { "qemu32" "-" TYPE_X86_CPU, "popcnt", "on" },
252     { "Opteron_G2" "-" TYPE_X86_CPU, "rdtscp", "on" },
253     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "on" },
254     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "on" },
255     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "on", }
256 };
257 const size_t pc_compat_2_4_len = G_N_ELEMENTS(pc_compat_2_4);
258 
259 GlobalProperty pc_compat_2_3[] = {
260     PC_CPU_MODEL_IDS("2.3.0")
261     { TYPE_X86_CPU, "arat", "off" },
262     { "qemu64" "-" TYPE_X86_CPU, "min-level", "4" },
263     { "kvm64" "-" TYPE_X86_CPU, "min-level", "5" },
264     { "pentium3" "-" TYPE_X86_CPU, "min-level", "2" },
265     { "n270" "-" TYPE_X86_CPU, "min-level", "5" },
266     { "Conroe" "-" TYPE_X86_CPU, "min-level", "4" },
267     { "Penryn" "-" TYPE_X86_CPU, "min-level", "4" },
268     { "Nehalem" "-" TYPE_X86_CPU, "min-level", "4" },
269     { "n270" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
270     { "Penryn" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
271     { "Conroe" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
272     { "Nehalem" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
273     { "Westmere" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
274     { "SandyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
275     { "IvyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
276     { "Haswell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
277     { "Haswell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
278     { "Broadwell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
279     { "Broadwell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
280     { TYPE_X86_CPU, "kvm-no-smi-migration", "on" },
281 };
282 const size_t pc_compat_2_3_len = G_N_ELEMENTS(pc_compat_2_3);
283 
284 GlobalProperty pc_compat_2_2[] = {
285     PC_CPU_MODEL_IDS("2.2.0")
286     { "kvm64" "-" TYPE_X86_CPU, "vme", "off" },
287     { "kvm32" "-" TYPE_X86_CPU, "vme", "off" },
288     { "Conroe" "-" TYPE_X86_CPU, "vme", "off" },
289     { "Penryn" "-" TYPE_X86_CPU, "vme", "off" },
290     { "Nehalem" "-" TYPE_X86_CPU, "vme", "off" },
291     { "Westmere" "-" TYPE_X86_CPU, "vme", "off" },
292     { "SandyBridge" "-" TYPE_X86_CPU, "vme", "off" },
293     { "Haswell" "-" TYPE_X86_CPU, "vme", "off" },
294     { "Broadwell" "-" TYPE_X86_CPU, "vme", "off" },
295     { "Opteron_G1" "-" TYPE_X86_CPU, "vme", "off" },
296     { "Opteron_G2" "-" TYPE_X86_CPU, "vme", "off" },
297     { "Opteron_G3" "-" TYPE_X86_CPU, "vme", "off" },
298     { "Opteron_G4" "-" TYPE_X86_CPU, "vme", "off" },
299     { "Opteron_G5" "-" TYPE_X86_CPU, "vme", "off" },
300     { "Haswell" "-" TYPE_X86_CPU, "f16c", "off" },
301     { "Haswell" "-" TYPE_X86_CPU, "rdrand", "off" },
302     { "Broadwell" "-" TYPE_X86_CPU, "f16c", "off" },
303     { "Broadwell" "-" TYPE_X86_CPU, "rdrand", "off" },
304 };
305 const size_t pc_compat_2_2_len = G_N_ELEMENTS(pc_compat_2_2);
306 
307 GlobalProperty pc_compat_2_1[] = {
308     PC_CPU_MODEL_IDS("2.1.0")
309     { "coreduo" "-" TYPE_X86_CPU, "vmx", "on" },
310     { "core2duo" "-" TYPE_X86_CPU, "vmx", "on" },
311 };
312 const size_t pc_compat_2_1_len = G_N_ELEMENTS(pc_compat_2_1);
313 
314 GlobalProperty pc_compat_2_0[] = {
315     PC_CPU_MODEL_IDS("2.0.0")
316     { "virtio-scsi-pci", "any_layout", "off" },
317     { "PIIX4_PM", "memory-hotplug-support", "off" },
318     { "apic", "version", "0x11" },
319     { "nec-usb-xhci", "superspeed-ports-first", "off" },
320     { "nec-usb-xhci", "force-pcie-endcap", "on" },
321     { "pci-serial", "prog_if", "0" },
322     { "pci-serial-2x", "prog_if", "0" },
323     { "pci-serial-4x", "prog_if", "0" },
324     { "virtio-net-pci", "guest_announce", "off" },
325     { "ICH9-LPC", "memory-hotplug-support", "off" },
326 };
327 const size_t pc_compat_2_0_len = G_N_ELEMENTS(pc_compat_2_0);
328 
329 GSIState *pc_gsi_create(qemu_irq **irqs, bool pci_enabled)
330 {
331     GSIState *s;
332 
333     s = g_new0(GSIState, 1);
334     if (kvm_ioapic_in_kernel()) {
335         kvm_pc_setup_irq_routing(pci_enabled);
336     }
337     *irqs = qemu_allocate_irqs(gsi_handler, s, IOAPIC_NUM_PINS);
338 
339     return s;
340 }
341 
342 static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
343                            unsigned size)
344 {
345 }
346 
347 static uint64_t ioport80_read(void *opaque, hwaddr addr, unsigned size)
348 {
349     return 0xffffffffffffffffULL;
350 }
351 
352 /* MS-DOS compatibility mode FPU exception support */
353 static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
354                            unsigned size)
355 {
356     if (tcg_enabled()) {
357         cpu_set_ignne();
358     }
359 }
360 
361 static uint64_t ioportF0_read(void *opaque, hwaddr addr, unsigned size)
362 {
363     return 0xffffffffffffffffULL;
364 }
365 
366 /* PC cmos mappings */
367 
368 #define REG_EQUIPMENT_BYTE          0x14
369 
370 static void cmos_init_hd(MC146818RtcState *s, int type_ofs, int info_ofs,
371                          int16_t cylinders, int8_t heads, int8_t sectors)
372 {
373     mc146818rtc_set_cmos_data(s, type_ofs, 47);
374     mc146818rtc_set_cmos_data(s, info_ofs, cylinders);
375     mc146818rtc_set_cmos_data(s, info_ofs + 1, cylinders >> 8);
376     mc146818rtc_set_cmos_data(s, info_ofs + 2, heads);
377     mc146818rtc_set_cmos_data(s, info_ofs + 3, 0xff);
378     mc146818rtc_set_cmos_data(s, info_ofs + 4, 0xff);
379     mc146818rtc_set_cmos_data(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
380     mc146818rtc_set_cmos_data(s, info_ofs + 6, cylinders);
381     mc146818rtc_set_cmos_data(s, info_ofs + 7, cylinders >> 8);
382     mc146818rtc_set_cmos_data(s, info_ofs + 8, sectors);
383 }
384 
385 /* convert boot_device letter to something recognizable by the bios */
386 static int boot_device2nibble(char boot_device)
387 {
388     switch(boot_device) {
389     case 'a':
390     case 'b':
391         return 0x01; /* floppy boot */
392     case 'c':
393         return 0x02; /* hard drive boot */
394     case 'd':
395         return 0x03; /* CD-ROM boot */
396     case 'n':
397         return 0x04; /* Network boot */
398     }
399     return 0;
400 }
401 
402 static void set_boot_dev(MC146818RtcState *s, const char *boot_device,
403                          Error **errp)
404 {
405 #define PC_MAX_BOOT_DEVICES 3
406     int nbds, bds[3] = { 0, };
407     int i;
408 
409     nbds = strlen(boot_device);
410     if (nbds > PC_MAX_BOOT_DEVICES) {
411         error_setg(errp, "Too many boot devices for PC");
412         return;
413     }
414     for (i = 0; i < nbds; i++) {
415         bds[i] = boot_device2nibble(boot_device[i]);
416         if (bds[i] == 0) {
417             error_setg(errp, "Invalid boot device for PC: '%c'",
418                        boot_device[i]);
419             return;
420         }
421     }
422     mc146818rtc_set_cmos_data(s, 0x3d, (bds[1] << 4) | bds[0]);
423     mc146818rtc_set_cmos_data(s, 0x38, (bds[2] << 4) | (fd_bootchk ? 0x0 : 0x1));
424 }
425 
426 static void pc_boot_set(void *opaque, const char *boot_device, Error **errp)
427 {
428     set_boot_dev(opaque, boot_device, errp);
429 }
430 
431 static void pc_cmos_init_floppy(MC146818RtcState *rtc_state, ISADevice *floppy)
432 {
433     int val, nb, i;
434     FloppyDriveType fd_type[2] = { FLOPPY_DRIVE_TYPE_NONE,
435                                    FLOPPY_DRIVE_TYPE_NONE };
436 
437     /* floppy type */
438     if (floppy) {
439         for (i = 0; i < 2; i++) {
440             fd_type[i] = isa_fdc_get_drive_type(floppy, i);
441         }
442     }
443     val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
444         cmos_get_fd_drive_type(fd_type[1]);
445     mc146818rtc_set_cmos_data(rtc_state, 0x10, val);
446 
447     val = mc146818rtc_get_cmos_data(rtc_state, REG_EQUIPMENT_BYTE);
448     nb = 0;
449     if (fd_type[0] != FLOPPY_DRIVE_TYPE_NONE) {
450         nb++;
451     }
452     if (fd_type[1] != FLOPPY_DRIVE_TYPE_NONE) {
453         nb++;
454     }
455     switch (nb) {
456     case 0:
457         break;
458     case 1:
459         val |= 0x01; /* 1 drive, ready for boot */
460         break;
461     case 2:
462         val |= 0x41; /* 2 drives, ready for boot */
463         break;
464     }
465     mc146818rtc_set_cmos_data(rtc_state, REG_EQUIPMENT_BYTE, val);
466 }
467 
468 typedef struct check_fdc_state {
469     ISADevice *floppy;
470     bool multiple;
471 } CheckFdcState;
472 
473 static int check_fdc(Object *obj, void *opaque)
474 {
475     CheckFdcState *state = opaque;
476     Object *fdc;
477     uint32_t iobase;
478     Error *local_err = NULL;
479 
480     fdc = object_dynamic_cast(obj, TYPE_ISA_FDC);
481     if (!fdc) {
482         return 0;
483     }
484 
485     iobase = object_property_get_uint(obj, "iobase", &local_err);
486     if (local_err || iobase != 0x3f0) {
487         error_free(local_err);
488         return 0;
489     }
490 
491     if (state->floppy) {
492         state->multiple = true;
493     } else {
494         state->floppy = ISA_DEVICE(obj);
495     }
496     return 0;
497 }
498 
499 static const char * const fdc_container_path[] = {
500     "/unattached", "/peripheral", "/peripheral-anon"
501 };
502 
503 /*
504  * Locate the FDC at IO address 0x3f0, in order to configure the CMOS registers
505  * and ACPI objects.
506  */
507 static ISADevice *pc_find_fdc0(void)
508 {
509     int i;
510     Object *container;
511     CheckFdcState state = { 0 };
512 
513     for (i = 0; i < ARRAY_SIZE(fdc_container_path); i++) {
514         container = container_get(qdev_get_machine(), fdc_container_path[i]);
515         object_child_foreach(container, check_fdc, &state);
516     }
517 
518     if (state.multiple) {
519         warn_report("multiple floppy disk controllers with "
520                     "iobase=0x3f0 have been found");
521         error_printf("the one being picked for CMOS setup might not reflect "
522                      "your intent");
523     }
524 
525     return state.floppy;
526 }
527 
528 static void pc_cmos_init_late(PCMachineState *pcms)
529 {
530     X86MachineState *x86ms = X86_MACHINE(pcms);
531     MC146818RtcState *s = MC146818_RTC(x86ms->rtc);
532     int16_t cylinders;
533     int8_t heads, sectors;
534     int val;
535     int i, trans;
536 
537     val = 0;
538     if (pcms->idebus[0] &&
539         ide_get_geometry(pcms->idebus[0], 0,
540                          &cylinders, &heads, &sectors) >= 0) {
541         cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
542         val |= 0xf0;
543     }
544     if (pcms->idebus[0] &&
545         ide_get_geometry(pcms->idebus[0], 1,
546                          &cylinders, &heads, &sectors) >= 0) {
547         cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
548         val |= 0x0f;
549     }
550     mc146818rtc_set_cmos_data(s, 0x12, val);
551 
552     val = 0;
553     for (i = 0; i < 4; i++) {
554         /* NOTE: ide_get_geometry() returns the physical
555            geometry.  It is always such that: 1 <= sects <= 63, 1
556            <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
557            geometry can be different if a translation is done. */
558         BusState *idebus = pcms->idebus[i / 2];
559         if (idebus &&
560             ide_get_geometry(idebus, i % 2,
561                              &cylinders, &heads, &sectors) >= 0) {
562             trans = ide_get_bios_chs_trans(idebus, i % 2) - 1;
563             assert((trans & ~3) == 0);
564             val |= trans << (i * 2);
565         }
566     }
567     mc146818rtc_set_cmos_data(s, 0x39, val);
568 
569     pc_cmos_init_floppy(s, pc_find_fdc0());
570 }
571 
572 void pc_cmos_init(PCMachineState *pcms,
573                   ISADevice *rtc)
574 {
575     int val;
576     X86MachineState *x86ms = X86_MACHINE(pcms);
577     MC146818RtcState *s = MC146818_RTC(rtc);
578 
579     /* various important CMOS locations needed by PC/Bochs bios */
580 
581     /* memory size */
582     /* base memory (first MiB) */
583     val = MIN(x86ms->below_4g_mem_size / KiB, 640);
584     mc146818rtc_set_cmos_data(s, 0x15, val);
585     mc146818rtc_set_cmos_data(s, 0x16, val >> 8);
586     /* extended memory (next 64MiB) */
587     if (x86ms->below_4g_mem_size > 1 * MiB) {
588         val = (x86ms->below_4g_mem_size - 1 * MiB) / KiB;
589     } else {
590         val = 0;
591     }
592     if (val > 65535)
593         val = 65535;
594     mc146818rtc_set_cmos_data(s, 0x17, val);
595     mc146818rtc_set_cmos_data(s, 0x18, val >> 8);
596     mc146818rtc_set_cmos_data(s, 0x30, val);
597     mc146818rtc_set_cmos_data(s, 0x31, val >> 8);
598     /* memory between 16MiB and 4GiB */
599     if (x86ms->below_4g_mem_size > 16 * MiB) {
600         val = (x86ms->below_4g_mem_size - 16 * MiB) / (64 * KiB);
601     } else {
602         val = 0;
603     }
604     if (val > 65535)
605         val = 65535;
606     mc146818rtc_set_cmos_data(s, 0x34, val);
607     mc146818rtc_set_cmos_data(s, 0x35, val >> 8);
608     /* memory above 4GiB */
609     val = x86ms->above_4g_mem_size / 65536;
610     mc146818rtc_set_cmos_data(s, 0x5b, val);
611     mc146818rtc_set_cmos_data(s, 0x5c, val >> 8);
612     mc146818rtc_set_cmos_data(s, 0x5d, val >> 16);
613 
614     set_boot_dev(s, MACHINE(pcms)->boot_config.order, &error_fatal);
615 
616     val = 0;
617     val |= 0x02; /* FPU is there */
618     val |= 0x04; /* PS/2 mouse installed */
619     mc146818rtc_set_cmos_data(s, REG_EQUIPMENT_BYTE, val);
620 
621     /* hard drives and FDC are handled by pc_cmos_init_late() */
622 }
623 
624 static void handle_a20_line_change(void *opaque, int irq, int level)
625 {
626     X86CPU *cpu = opaque;
627 
628     /* XXX: send to all CPUs ? */
629     /* XXX: add logic to handle multiple A20 line sources */
630     x86_cpu_set_a20(cpu, level);
631 }
632 
633 #define NE2000_NB_MAX 6
634 
635 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
636                                               0x280, 0x380 };
637 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
638 
639 static gboolean pc_init_ne2k_isa(ISABus *bus, NICInfo *nd, Error **errp)
640 {
641     static int nb_ne2k = 0;
642 
643     if (nb_ne2k == NE2000_NB_MAX) {
644         error_setg(errp,
645                    "maximum number of ISA NE2000 devices exceeded");
646         return false;
647     }
648     isa_ne2000_init(bus, ne2000_io[nb_ne2k],
649                     ne2000_irq[nb_ne2k], nd);
650     nb_ne2k++;
651     return true;
652 }
653 
654 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
655 {
656     X86CPU *cpu = opaque;
657 
658     if (level) {
659         cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
660     }
661 }
662 
663 static
664 void pc_machine_done(Notifier *notifier, void *data)
665 {
666     PCMachineState *pcms = container_of(notifier,
667                                         PCMachineState, machine_done);
668     X86MachineState *x86ms = X86_MACHINE(pcms);
669 
670     cxl_hook_up_pxb_registers(pcms->pcibus, &pcms->cxl_devices_state,
671                               &error_fatal);
672 
673     if (pcms->cxl_devices_state.is_enabled) {
674         cxl_fmws_link_targets(&pcms->cxl_devices_state, &error_fatal);
675     }
676 
677     /* set the number of CPUs */
678     x86_rtc_set_cpus_count(x86ms->rtc, x86ms->boot_cpus);
679 
680     fw_cfg_add_extra_pci_roots(pcms->pcibus, x86ms->fw_cfg);
681 
682     acpi_setup();
683     if (x86ms->fw_cfg) {
684         fw_cfg_build_smbios(pcms, x86ms->fw_cfg);
685         fw_cfg_build_feature_control(MACHINE(pcms), x86ms->fw_cfg);
686         /* update FW_CFG_NB_CPUS to account for -device added CPUs */
687         fw_cfg_modify_i16(x86ms->fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
688     }
689 
690     pc_cmos_init_late(pcms);
691 }
692 
693 /* setup pci memory address space mapping into system address space */
694 void pc_pci_as_mapping_init(MemoryRegion *system_memory,
695                             MemoryRegion *pci_address_space)
696 {
697     /* Set to lower priority than RAM */
698     memory_region_add_subregion_overlap(system_memory, 0x0,
699                                         pci_address_space, -1);
700 }
701 
702 void xen_load_linux(PCMachineState *pcms)
703 {
704     int i;
705     FWCfgState *fw_cfg;
706     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
707     X86MachineState *x86ms = X86_MACHINE(pcms);
708 
709     assert(MACHINE(pcms)->kernel_filename != NULL);
710 
711     fw_cfg = fw_cfg_init_io(FW_CFG_IO_BASE);
712     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
713     rom_set_fw(fw_cfg);
714 
715     x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
716                    pcmc->pvh_enabled);
717     for (i = 0; i < nb_option_roms; i++) {
718         assert(!strcmp(option_rom[i].name, "linuxboot.bin") ||
719                !strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
720                !strcmp(option_rom[i].name, "pvh.bin") ||
721                !strcmp(option_rom[i].name, "multiboot.bin") ||
722                !strcmp(option_rom[i].name, "multiboot_dma.bin"));
723         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
724     }
725     x86ms->fw_cfg = fw_cfg;
726 }
727 
728 #define PC_ROM_MIN_VGA     0xc0000
729 #define PC_ROM_MIN_OPTION  0xc8000
730 #define PC_ROM_MAX         0xe0000
731 #define PC_ROM_ALIGN       0x800
732 #define PC_ROM_SIZE        (PC_ROM_MAX - PC_ROM_MIN_VGA)
733 
734 static hwaddr pc_above_4g_end(PCMachineState *pcms)
735 {
736     X86MachineState *x86ms = X86_MACHINE(pcms);
737 
738     if (pcms->sgx_epc.size != 0) {
739         return sgx_epc_above_4g_end(&pcms->sgx_epc);
740     }
741 
742     return x86ms->above_4g_mem_start + x86ms->above_4g_mem_size;
743 }
744 
745 static void pc_get_device_memory_range(PCMachineState *pcms,
746                                        hwaddr *base,
747                                        ram_addr_t *device_mem_size)
748 {
749     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
750     MachineState *machine = MACHINE(pcms);
751     ram_addr_t size;
752     hwaddr addr;
753 
754     size = machine->maxram_size - machine->ram_size;
755     addr = ROUND_UP(pc_above_4g_end(pcms), 1 * GiB);
756 
757     if (pcmc->enforce_aligned_dimm) {
758         /* size device region assuming 1G page max alignment per slot */
759         size += (1 * GiB) * machine->ram_slots;
760     }
761 
762     *base = addr;
763     *device_mem_size = size;
764 }
765 
766 static uint64_t pc_get_cxl_range_start(PCMachineState *pcms)
767 {
768     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
769     MachineState *ms = MACHINE(pcms);
770     hwaddr cxl_base;
771     ram_addr_t size;
772 
773     if (pcmc->has_reserved_memory &&
774         (ms->ram_size < ms->maxram_size)) {
775         pc_get_device_memory_range(pcms, &cxl_base, &size);
776         cxl_base += size;
777     } else {
778         cxl_base = pc_above_4g_end(pcms);
779     }
780 
781     return cxl_base;
782 }
783 
784 static uint64_t pc_get_cxl_range_end(PCMachineState *pcms)
785 {
786     uint64_t start = pc_get_cxl_range_start(pcms) + MiB;
787 
788     if (pcms->cxl_devices_state.fixed_windows) {
789         GList *it;
790 
791         start = ROUND_UP(start, 256 * MiB);
792         for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
793             CXLFixedWindow *fw = it->data;
794             start += fw->size;
795         }
796     }
797 
798     return start;
799 }
800 
801 static hwaddr pc_max_used_gpa(PCMachineState *pcms, uint64_t pci_hole64_size)
802 {
803     X86CPU *cpu = X86_CPU(first_cpu);
804     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
805     MachineState *ms = MACHINE(pcms);
806 
807     if (cpu->env.features[FEAT_8000_0001_EDX] & CPUID_EXT2_LM) {
808         /* 64-bit systems */
809         return pc_pci_hole64_start() + pci_hole64_size - 1;
810     }
811 
812     /* 32-bit systems */
813     if (pcmc->broken_32bit_mem_addr_check) {
814         /* old value for compatibility reasons */
815         return ((hwaddr)1 << cpu->phys_bits) - 1;
816     }
817 
818     /*
819      * 32-bit systems don't have hole64 but they might have a region for
820      * memory devices. Even if additional hotplugged memory devices might
821      * not be usable by most guest OSes, we need to still consider them for
822      * calculating the highest possible GPA so that we can properly report
823      * if someone configures them on a CPU that cannot possibly address them.
824      */
825     if (pcmc->has_reserved_memory &&
826         (ms->ram_size < ms->maxram_size)) {
827         hwaddr devmem_start;
828         ram_addr_t devmem_size;
829 
830         pc_get_device_memory_range(pcms, &devmem_start, &devmem_size);
831         devmem_start += devmem_size;
832         return devmem_start - 1;
833     }
834 
835     /* configuration without any memory hotplug */
836     return pc_above_4g_end(pcms) - 1;
837 }
838 
839 /*
840  * AMD systems with an IOMMU have an additional hole close to the
841  * 1Tb, which are special GPAs that cannot be DMA mapped. Depending
842  * on kernel version, VFIO may or may not let you DMA map those ranges.
843  * Starting Linux v5.4 we validate it, and can't create guests on AMD machines
844  * with certain memory sizes. It's also wrong to use those IOVA ranges
845  * in detriment of leading to IOMMU INVALID_DEVICE_REQUEST or worse.
846  * The ranges reserved for Hyper-Transport are:
847  *
848  * FD_0000_0000h - FF_FFFF_FFFFh
849  *
850  * The ranges represent the following:
851  *
852  * Base Address   Top Address  Use
853  *
854  * FD_0000_0000h FD_F7FF_FFFFh Reserved interrupt address space
855  * FD_F800_0000h FD_F8FF_FFFFh Interrupt/EOI IntCtl
856  * FD_F900_0000h FD_F90F_FFFFh Legacy PIC IACK
857  * FD_F910_0000h FD_F91F_FFFFh System Management
858  * FD_F920_0000h FD_FAFF_FFFFh Reserved Page Tables
859  * FD_FB00_0000h FD_FBFF_FFFFh Address Translation
860  * FD_FC00_0000h FD_FDFF_FFFFh I/O Space
861  * FD_FE00_0000h FD_FFFF_FFFFh Configuration
862  * FE_0000_0000h FE_1FFF_FFFFh Extended Configuration/Device Messages
863  * FE_2000_0000h FF_FFFF_FFFFh Reserved
864  *
865  * See AMD IOMMU spec, section 2.1.2 "IOMMU Logical Topology",
866  * Table 3: Special Address Controls (GPA) for more information.
867  */
868 #define AMD_HT_START         0xfd00000000UL
869 #define AMD_HT_END           0xffffffffffUL
870 #define AMD_ABOVE_1TB_START  (AMD_HT_END + 1)
871 #define AMD_HT_SIZE          (AMD_ABOVE_1TB_START - AMD_HT_START)
872 
873 void pc_memory_init(PCMachineState *pcms,
874                     MemoryRegion *system_memory,
875                     MemoryRegion *rom_memory,
876                     uint64_t pci_hole64_size)
877 {
878     int linux_boot, i;
879     MemoryRegion *option_rom_mr;
880     MemoryRegion *ram_below_4g, *ram_above_4g;
881     FWCfgState *fw_cfg;
882     MachineState *machine = MACHINE(pcms);
883     MachineClass *mc = MACHINE_GET_CLASS(machine);
884     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
885     X86MachineState *x86ms = X86_MACHINE(pcms);
886     hwaddr maxphysaddr, maxusedaddr;
887     hwaddr cxl_base, cxl_resv_end = 0;
888     X86CPU *cpu = X86_CPU(first_cpu);
889 
890     assert(machine->ram_size == x86ms->below_4g_mem_size +
891                                 x86ms->above_4g_mem_size);
892 
893     linux_boot = (machine->kernel_filename != NULL);
894 
895     /*
896      * The HyperTransport range close to the 1T boundary is unique to AMD
897      * hosts with IOMMUs enabled. Restrict the ram-above-4g relocation
898      * to above 1T to AMD vCPUs only. @enforce_amd_1tb_hole is only false in
899      * older machine types (<= 7.0) for compatibility purposes.
900      */
901     if (IS_AMD_CPU(&cpu->env) && pcmc->enforce_amd_1tb_hole) {
902         /* Bail out if max possible address does not cross HT range */
903         if (pc_max_used_gpa(pcms, pci_hole64_size) >= AMD_HT_START) {
904             x86ms->above_4g_mem_start = AMD_ABOVE_1TB_START;
905         }
906 
907         /*
908          * Advertise the HT region if address space covers the reserved
909          * region or if we relocate.
910          */
911         if (cpu->phys_bits >= 40) {
912             e820_add_entry(AMD_HT_START, AMD_HT_SIZE, E820_RESERVED);
913         }
914     }
915 
916     /*
917      * phys-bits is required to be appropriately configured
918      * to make sure max used GPA is reachable.
919      */
920     maxusedaddr = pc_max_used_gpa(pcms, pci_hole64_size);
921     maxphysaddr = ((hwaddr)1 << cpu->phys_bits) - 1;
922     if (maxphysaddr < maxusedaddr) {
923         error_report("Address space limit 0x%"PRIx64" < 0x%"PRIx64
924                      " phys-bits too low (%u)",
925                      maxphysaddr, maxusedaddr, cpu->phys_bits);
926         exit(EXIT_FAILURE);
927     }
928 
929     /*
930      * Split single memory region and use aliases to address portions of it,
931      * done for backwards compatibility with older qemus.
932      */
933     ram_below_4g = g_malloc(sizeof(*ram_below_4g));
934     memory_region_init_alias(ram_below_4g, NULL, "ram-below-4g", machine->ram,
935                              0, x86ms->below_4g_mem_size);
936     memory_region_add_subregion(system_memory, 0, ram_below_4g);
937     e820_add_entry(0, x86ms->below_4g_mem_size, E820_RAM);
938     if (x86ms->above_4g_mem_size > 0) {
939         ram_above_4g = g_malloc(sizeof(*ram_above_4g));
940         memory_region_init_alias(ram_above_4g, NULL, "ram-above-4g",
941                                  machine->ram,
942                                  x86ms->below_4g_mem_size,
943                                  x86ms->above_4g_mem_size);
944         memory_region_add_subregion(system_memory, x86ms->above_4g_mem_start,
945                                     ram_above_4g);
946         e820_add_entry(x86ms->above_4g_mem_start, x86ms->above_4g_mem_size,
947                        E820_RAM);
948     }
949 
950     if (pcms->sgx_epc.size != 0) {
951         e820_add_entry(pcms->sgx_epc.base, pcms->sgx_epc.size, E820_RESERVED);
952     }
953 
954     if (!pcmc->has_reserved_memory &&
955         (machine->ram_slots ||
956          (machine->maxram_size > machine->ram_size))) {
957 
958         error_report("\"-memory 'slots|maxmem'\" is not supported by: %s",
959                      mc->name);
960         exit(EXIT_FAILURE);
961     }
962 
963     /* initialize device memory address space */
964     if (pcmc->has_reserved_memory &&
965         (machine->ram_size < machine->maxram_size)) {
966         ram_addr_t device_mem_size;
967         hwaddr device_mem_base;
968 
969         if (machine->ram_slots > ACPI_MAX_RAM_SLOTS) {
970             error_report("unsupported amount of memory slots: %"PRIu64,
971                          machine->ram_slots);
972             exit(EXIT_FAILURE);
973         }
974 
975         if (QEMU_ALIGN_UP(machine->maxram_size,
976                           TARGET_PAGE_SIZE) != machine->maxram_size) {
977             error_report("maximum memory size must by aligned to multiple of "
978                          "%d bytes", TARGET_PAGE_SIZE);
979             exit(EXIT_FAILURE);
980         }
981 
982         pc_get_device_memory_range(pcms, &device_mem_base, &device_mem_size);
983 
984         if (device_mem_base + device_mem_size < device_mem_size) {
985             error_report("unsupported amount of maximum memory: " RAM_ADDR_FMT,
986                          machine->maxram_size);
987             exit(EXIT_FAILURE);
988         }
989         machine_memory_devices_init(machine, device_mem_base, device_mem_size);
990     }
991 
992     if (pcms->cxl_devices_state.is_enabled) {
993         MemoryRegion *mr = &pcms->cxl_devices_state.host_mr;
994         hwaddr cxl_size = MiB;
995 
996         cxl_base = pc_get_cxl_range_start(pcms);
997         memory_region_init(mr, OBJECT(machine), "cxl_host_reg", cxl_size);
998         memory_region_add_subregion(system_memory, cxl_base, mr);
999         cxl_resv_end = cxl_base + cxl_size;
1000         if (pcms->cxl_devices_state.fixed_windows) {
1001             hwaddr cxl_fmw_base;
1002             GList *it;
1003 
1004             cxl_fmw_base = ROUND_UP(cxl_base + cxl_size, 256 * MiB);
1005             for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
1006                 CXLFixedWindow *fw = it->data;
1007 
1008                 fw->base = cxl_fmw_base;
1009                 memory_region_init_io(&fw->mr, OBJECT(machine), &cfmws_ops, fw,
1010                                       "cxl-fixed-memory-region", fw->size);
1011                 memory_region_add_subregion(system_memory, fw->base, &fw->mr);
1012                 cxl_fmw_base += fw->size;
1013                 cxl_resv_end = cxl_fmw_base;
1014             }
1015         }
1016     }
1017 
1018     /* Initialize PC system firmware */
1019     pc_system_firmware_init(pcms, rom_memory);
1020 
1021     option_rom_mr = g_malloc(sizeof(*option_rom_mr));
1022     memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE,
1023                            &error_fatal);
1024     if (pcmc->pci_enabled) {
1025         memory_region_set_readonly(option_rom_mr, true);
1026     }
1027     memory_region_add_subregion_overlap(rom_memory,
1028                                         PC_ROM_MIN_VGA,
1029                                         option_rom_mr,
1030                                         1);
1031 
1032     fw_cfg = fw_cfg_arch_create(machine,
1033                                 x86ms->boot_cpus, x86ms->apic_id_limit);
1034 
1035     rom_set_fw(fw_cfg);
1036 
1037     if (machine->device_memory) {
1038         uint64_t *val = g_malloc(sizeof(*val));
1039         uint64_t res_mem_end = machine->device_memory->base;
1040 
1041         if (!pcmc->broken_reserved_end) {
1042             res_mem_end += memory_region_size(&machine->device_memory->mr);
1043         }
1044 
1045         if (pcms->cxl_devices_state.is_enabled) {
1046             res_mem_end = cxl_resv_end;
1047         }
1048         *val = cpu_to_le64(ROUND_UP(res_mem_end, 1 * GiB));
1049         fw_cfg_add_file(fw_cfg, "etc/reserved-memory-end", val, sizeof(*val));
1050     }
1051 
1052     if (linux_boot) {
1053         x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
1054                        pcmc->pvh_enabled);
1055     }
1056 
1057     for (i = 0; i < nb_option_roms; i++) {
1058         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1059     }
1060     x86ms->fw_cfg = fw_cfg;
1061 
1062     /* Init default IOAPIC address space */
1063     x86ms->ioapic_as = &address_space_memory;
1064 
1065     /* Init ACPI memory hotplug IO base address */
1066     pcms->memhp_io_base = ACPI_MEMORY_HOTPLUG_BASE;
1067 }
1068 
1069 /*
1070  * The 64bit pci hole starts after "above 4G RAM" and
1071  * potentially the space reserved for memory hotplug.
1072  */
1073 uint64_t pc_pci_hole64_start(void)
1074 {
1075     PCMachineState *pcms = PC_MACHINE(qdev_get_machine());
1076     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1077     MachineState *ms = MACHINE(pcms);
1078     uint64_t hole64_start = 0;
1079     ram_addr_t size = 0;
1080 
1081     if (pcms->cxl_devices_state.is_enabled) {
1082         hole64_start = pc_get_cxl_range_end(pcms);
1083     } else if (pcmc->has_reserved_memory && (ms->ram_size < ms->maxram_size)) {
1084         pc_get_device_memory_range(pcms, &hole64_start, &size);
1085         if (!pcmc->broken_reserved_end) {
1086             hole64_start += size;
1087         }
1088     } else {
1089         hole64_start = pc_above_4g_end(pcms);
1090     }
1091 
1092     return ROUND_UP(hole64_start, 1 * GiB);
1093 }
1094 
1095 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1096 {
1097     DeviceState *dev = NULL;
1098 
1099     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_VGA);
1100     if (pci_bus) {
1101         PCIDevice *pcidev = pci_vga_init(pci_bus);
1102         dev = pcidev ? &pcidev->qdev : NULL;
1103     } else if (isa_bus) {
1104         ISADevice *isadev = isa_vga_init(isa_bus);
1105         dev = isadev ? DEVICE(isadev) : NULL;
1106     }
1107     rom_reset_order_override();
1108     return dev;
1109 }
1110 
1111 static const MemoryRegionOps ioport80_io_ops = {
1112     .write = ioport80_write,
1113     .read = ioport80_read,
1114     .endianness = DEVICE_NATIVE_ENDIAN,
1115     .impl = {
1116         .min_access_size = 1,
1117         .max_access_size = 1,
1118     },
1119 };
1120 
1121 static const MemoryRegionOps ioportF0_io_ops = {
1122     .write = ioportF0_write,
1123     .read = ioportF0_read,
1124     .endianness = DEVICE_NATIVE_ENDIAN,
1125     .impl = {
1126         .min_access_size = 1,
1127         .max_access_size = 1,
1128     },
1129 };
1130 
1131 static void pc_superio_init(ISABus *isa_bus, bool create_fdctrl,
1132                             bool create_i8042, bool no_vmport)
1133 {
1134     int i;
1135     DriveInfo *fd[MAX_FD];
1136     qemu_irq *a20_line;
1137     ISADevice *fdc, *i8042, *port92, *vmmouse;
1138 
1139     serial_hds_isa_init(isa_bus, 0, MAX_ISA_SERIAL_PORTS);
1140     parallel_hds_isa_init(isa_bus, MAX_PARALLEL_PORTS);
1141 
1142     for (i = 0; i < MAX_FD; i++) {
1143         fd[i] = drive_get(IF_FLOPPY, 0, i);
1144         create_fdctrl |= !!fd[i];
1145     }
1146     if (create_fdctrl) {
1147         fdc = isa_new(TYPE_ISA_FDC);
1148         if (fdc) {
1149             isa_realize_and_unref(fdc, isa_bus, &error_fatal);
1150             isa_fdc_init_drives(fdc, fd);
1151         }
1152     }
1153 
1154     if (!create_i8042) {
1155         return;
1156     }
1157 
1158     i8042 = isa_create_simple(isa_bus, TYPE_I8042);
1159     if (!no_vmport) {
1160         isa_create_simple(isa_bus, TYPE_VMPORT);
1161         vmmouse = isa_try_new("vmmouse");
1162     } else {
1163         vmmouse = NULL;
1164     }
1165     if (vmmouse) {
1166         object_property_set_link(OBJECT(vmmouse), TYPE_I8042, OBJECT(i8042),
1167                                  &error_abort);
1168         isa_realize_and_unref(vmmouse, isa_bus, &error_fatal);
1169     }
1170     port92 = isa_create_simple(isa_bus, TYPE_PORT92);
1171 
1172     a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1173     qdev_connect_gpio_out_named(DEVICE(i8042),
1174                                 I8042_A20_LINE, 0, a20_line[0]);
1175     qdev_connect_gpio_out_named(DEVICE(port92),
1176                                 PORT92_A20_LINE, 0, a20_line[1]);
1177     g_free(a20_line);
1178 }
1179 
1180 void pc_basic_device_init(struct PCMachineState *pcms,
1181                           ISABus *isa_bus, qemu_irq *gsi,
1182                           ISADevice *rtc_state,
1183                           bool create_fdctrl,
1184                           uint32_t hpet_irqs)
1185 {
1186     int i;
1187     DeviceState *hpet = NULL;
1188     int pit_isa_irq = 0;
1189     qemu_irq pit_alt_irq = NULL;
1190     ISADevice *pit = NULL;
1191     MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1192     MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
1193     X86MachineState *x86ms = X86_MACHINE(pcms);
1194 
1195     memory_region_init_io(ioport80_io, NULL, &ioport80_io_ops, NULL, "ioport80", 1);
1196     memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
1197 
1198     memory_region_init_io(ioportF0_io, NULL, &ioportF0_io_ops, NULL, "ioportF0", 1);
1199     memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
1200 
1201     /*
1202      * Check if an HPET shall be created.
1203      */
1204     if (pcms->hpet_enabled) {
1205         qemu_irq rtc_irq;
1206 
1207         hpet = qdev_try_new(TYPE_HPET);
1208         if (!hpet) {
1209             error_report("couldn't create HPET device");
1210             exit(1);
1211         }
1212         /*
1213          * For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-*,
1214          * use IRQ16~23, IRQ8 and IRQ2.  If the user has already set
1215          * the property, use whatever mask they specified.
1216          */
1217         uint8_t compat = object_property_get_uint(OBJECT(hpet),
1218                 HPET_INTCAP, NULL);
1219         if (!compat) {
1220             qdev_prop_set_uint32(hpet, HPET_INTCAP, hpet_irqs);
1221         }
1222         sysbus_realize_and_unref(SYS_BUS_DEVICE(hpet), &error_fatal);
1223         sysbus_mmio_map(SYS_BUS_DEVICE(hpet), 0, HPET_BASE);
1224 
1225         for (i = 0; i < IOAPIC_NUM_PINS; i++) {
1226             sysbus_connect_irq(SYS_BUS_DEVICE(hpet), i, gsi[i]);
1227         }
1228         pit_isa_irq = -1;
1229         pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1230         rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1231 
1232         /* overwrite connection created by south bridge */
1233         qdev_connect_gpio_out(DEVICE(rtc_state), 0, rtc_irq);
1234     }
1235 
1236     object_property_add_alias(OBJECT(pcms), "rtc-time", OBJECT(rtc_state),
1237                               "date");
1238 
1239 #ifdef CONFIG_XEN_EMU
1240     if (xen_mode == XEN_EMULATE) {
1241         xen_overlay_create();
1242         xen_evtchn_create(IOAPIC_NUM_PINS, gsi);
1243         xen_gnttab_create();
1244         xen_xenstore_create();
1245         if (pcms->pcibus) {
1246             pci_create_simple(pcms->pcibus, -1, "xen-platform");
1247         }
1248         xen_bus_init();
1249         xen_be_init();
1250     }
1251 #endif
1252 
1253     qemu_register_boot_set(pc_boot_set, rtc_state);
1254 
1255     if (!xen_enabled() &&
1256         (x86ms->pit == ON_OFF_AUTO_AUTO || x86ms->pit == ON_OFF_AUTO_ON)) {
1257         if (kvm_pit_in_kernel()) {
1258             pit = kvm_pit_init(isa_bus, 0x40);
1259         } else {
1260             pit = i8254_pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1261         }
1262         if (hpet) {
1263             /* connect PIT to output control line of the HPET */
1264             qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(DEVICE(pit), 0));
1265         }
1266         object_property_set_link(OBJECT(pcms->pcspk), "pit",
1267                                  OBJECT(pit), &error_fatal);
1268         isa_realize_and_unref(pcms->pcspk, isa_bus, &error_fatal);
1269     }
1270 
1271     /* Super I/O */
1272     pc_superio_init(isa_bus, create_fdctrl, pcms->i8042_enabled,
1273                     pcms->vmport != ON_OFF_AUTO_ON);
1274 }
1275 
1276 void pc_nic_init(PCMachineClass *pcmc, ISABus *isa_bus, PCIBus *pci_bus)
1277 {
1278     MachineClass *mc = MACHINE_CLASS(pcmc);
1279     bool default_is_ne2k = g_str_equal(mc->default_nic, TYPE_ISA_NE2000);
1280     NICInfo *nd;
1281 
1282     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_NIC);
1283 
1284     while ((nd = qemu_find_nic_info(TYPE_ISA_NE2000, default_is_ne2k, NULL))) {
1285         pc_init_ne2k_isa(isa_bus, nd, &error_fatal);
1286     }
1287 
1288     /* Anything remaining should be a PCI NIC */
1289     pci_init_nic_devices(pci_bus, mc->default_nic);
1290 
1291     rom_reset_order_override();
1292 }
1293 
1294 void pc_i8259_create(ISABus *isa_bus, qemu_irq *i8259_irqs)
1295 {
1296     qemu_irq *i8259;
1297 
1298     if (kvm_pic_in_kernel()) {
1299         i8259 = kvm_i8259_init(isa_bus);
1300     } else if (xen_enabled()) {
1301         i8259 = xen_interrupt_controller_init();
1302     } else {
1303         i8259 = i8259_init(isa_bus, x86_allocate_cpu_irq());
1304     }
1305 
1306     for (size_t i = 0; i < ISA_NUM_IRQS; i++) {
1307         i8259_irqs[i] = i8259[i];
1308     }
1309 
1310     g_free(i8259);
1311 }
1312 
1313 static void pc_memory_pre_plug(HotplugHandler *hotplug_dev, DeviceState *dev,
1314                                Error **errp)
1315 {
1316     const PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1317     const X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1318     const PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1319     const MachineState *ms = MACHINE(hotplug_dev);
1320     const bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1321     const uint64_t legacy_align = TARGET_PAGE_SIZE;
1322     Error *local_err = NULL;
1323 
1324     /*
1325      * When "acpi=off" is used with the Q35 machine type, no ACPI is built,
1326      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1327      * addition to cover this case.
1328      */
1329     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1330         error_setg(errp,
1331                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1332         return;
1333     }
1334 
1335     if (is_nvdimm && !ms->nvdimms_state->is_enabled) {
1336         error_setg(errp, "nvdimm is not enabled: missing 'nvdimm' in '-M'");
1337         return;
1338     }
1339 
1340     hotplug_handler_pre_plug(x86ms->acpi_dev, dev, &local_err);
1341     if (local_err) {
1342         error_propagate(errp, local_err);
1343         return;
1344     }
1345 
1346     pc_dimm_pre_plug(PC_DIMM(dev), MACHINE(hotplug_dev),
1347                      pcmc->enforce_aligned_dimm ? NULL : &legacy_align, errp);
1348 }
1349 
1350 static void pc_memory_plug(HotplugHandler *hotplug_dev,
1351                            DeviceState *dev, Error **errp)
1352 {
1353     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1354     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1355     MachineState *ms = MACHINE(hotplug_dev);
1356     bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1357 
1358     pc_dimm_plug(PC_DIMM(dev), MACHINE(pcms));
1359 
1360     if (is_nvdimm) {
1361         nvdimm_plug(ms->nvdimms_state);
1362     }
1363 
1364     hotplug_handler_plug(x86ms->acpi_dev, dev, &error_abort);
1365 }
1366 
1367 static void pc_memory_unplug_request(HotplugHandler *hotplug_dev,
1368                                      DeviceState *dev, Error **errp)
1369 {
1370     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1371 
1372     /*
1373      * When "acpi=off" is used with the Q35 machine type, no ACPI is built,
1374      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1375      * addition to cover this case.
1376      */
1377     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1378         error_setg(errp,
1379                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1380         return;
1381     }
1382 
1383     if (object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM)) {
1384         error_setg(errp, "nvdimm device hot unplug is not supported yet.");
1385         return;
1386     }
1387 
1388     hotplug_handler_unplug_request(x86ms->acpi_dev, dev,
1389                                    errp);
1390 }
1391 
1392 static void pc_memory_unplug(HotplugHandler *hotplug_dev,
1393                              DeviceState *dev, Error **errp)
1394 {
1395     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1396     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1397     Error *local_err = NULL;
1398 
1399     hotplug_handler_unplug(x86ms->acpi_dev, dev, &local_err);
1400     if (local_err) {
1401         goto out;
1402     }
1403 
1404     pc_dimm_unplug(PC_DIMM(dev), MACHINE(pcms));
1405     qdev_unrealize(dev);
1406  out:
1407     error_propagate(errp, local_err);
1408 }
1409 
1410 static void pc_hv_balloon_pre_plug(HotplugHandler *hotplug_dev,
1411                                    DeviceState *dev, Error **errp)
1412 {
1413     /* The vmbus handler has no hotplug handler; we should never end up here. */
1414     g_assert(!dev->hotplugged);
1415     memory_device_pre_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev), NULL,
1416                            errp);
1417 }
1418 
1419 static void pc_hv_balloon_plug(HotplugHandler *hotplug_dev,
1420                                DeviceState *dev, Error **errp)
1421 {
1422     memory_device_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev));
1423 }
1424 
1425 static void pc_machine_device_pre_plug_cb(HotplugHandler *hotplug_dev,
1426                                           DeviceState *dev, Error **errp)
1427 {
1428     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1429         pc_memory_pre_plug(hotplug_dev, dev, errp);
1430     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1431         x86_cpu_pre_plug(hotplug_dev, dev, errp);
1432     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1433         virtio_md_pci_pre_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1434     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1435         /* Declare the APIC range as the reserved MSI region */
1436         char *resv_prop_str = g_strdup_printf("0xfee00000:0xfeefffff:%d",
1437                                               VIRTIO_IOMMU_RESV_MEM_T_MSI);
1438         QList *reserved_regions = qlist_new();
1439 
1440         qlist_append_str(reserved_regions, resv_prop_str);
1441         qdev_prop_set_array(dev, "reserved-regions", reserved_regions);
1442 
1443         g_free(resv_prop_str);
1444     }
1445 
1446     if (object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE) ||
1447         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1448         PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1449 
1450         if (pcms->iommu) {
1451             error_setg(errp, "QEMU does not support multiple vIOMMUs "
1452                        "for x86 yet.");
1453             return;
1454         }
1455         pcms->iommu = dev;
1456     } else if (object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON)) {
1457         pc_hv_balloon_pre_plug(hotplug_dev, dev, errp);
1458     }
1459 }
1460 
1461 static void pc_machine_device_plug_cb(HotplugHandler *hotplug_dev,
1462                                       DeviceState *dev, Error **errp)
1463 {
1464     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1465         pc_memory_plug(hotplug_dev, dev, errp);
1466     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1467         x86_cpu_plug(hotplug_dev, dev, errp);
1468     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1469         virtio_md_pci_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1470     } else if (object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON)) {
1471         pc_hv_balloon_plug(hotplug_dev, dev, errp);
1472     }
1473 }
1474 
1475 static void pc_machine_device_unplug_request_cb(HotplugHandler *hotplug_dev,
1476                                                 DeviceState *dev, Error **errp)
1477 {
1478     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1479         pc_memory_unplug_request(hotplug_dev, dev, errp);
1480     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1481         x86_cpu_unplug_request_cb(hotplug_dev, dev, errp);
1482     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1483         virtio_md_pci_unplug_request(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev),
1484                                      errp);
1485     } else {
1486         error_setg(errp, "acpi: device unplug request for not supported device"
1487                    " type: %s", object_get_typename(OBJECT(dev)));
1488     }
1489 }
1490 
1491 static void pc_machine_device_unplug_cb(HotplugHandler *hotplug_dev,
1492                                         DeviceState *dev, Error **errp)
1493 {
1494     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1495         pc_memory_unplug(hotplug_dev, dev, errp);
1496     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1497         x86_cpu_unplug_cb(hotplug_dev, dev, errp);
1498     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1499         virtio_md_pci_unplug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1500     } else {
1501         error_setg(errp, "acpi: device unplug for not supported device"
1502                    " type: %s", object_get_typename(OBJECT(dev)));
1503     }
1504 }
1505 
1506 static HotplugHandler *pc_get_hotplug_handler(MachineState *machine,
1507                                              DeviceState *dev)
1508 {
1509     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM) ||
1510         object_dynamic_cast(OBJECT(dev), TYPE_CPU) ||
1511         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI) ||
1512         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI) ||
1513         object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON) ||
1514         object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE)) {
1515         return HOTPLUG_HANDLER(machine);
1516     }
1517 
1518     return NULL;
1519 }
1520 
1521 static void pc_machine_get_vmport(Object *obj, Visitor *v, const char *name,
1522                                   void *opaque, Error **errp)
1523 {
1524     PCMachineState *pcms = PC_MACHINE(obj);
1525     OnOffAuto vmport = pcms->vmport;
1526 
1527     visit_type_OnOffAuto(v, name, &vmport, errp);
1528 }
1529 
1530 static void pc_machine_set_vmport(Object *obj, Visitor *v, const char *name,
1531                                   void *opaque, Error **errp)
1532 {
1533     PCMachineState *pcms = PC_MACHINE(obj);
1534 
1535     visit_type_OnOffAuto(v, name, &pcms->vmport, errp);
1536 }
1537 
1538 static bool pc_machine_get_smbus(Object *obj, Error **errp)
1539 {
1540     PCMachineState *pcms = PC_MACHINE(obj);
1541 
1542     return pcms->smbus_enabled;
1543 }
1544 
1545 static void pc_machine_set_smbus(Object *obj, bool value, Error **errp)
1546 {
1547     PCMachineState *pcms = PC_MACHINE(obj);
1548 
1549     pcms->smbus_enabled = value;
1550 }
1551 
1552 static bool pc_machine_get_sata(Object *obj, Error **errp)
1553 {
1554     PCMachineState *pcms = PC_MACHINE(obj);
1555 
1556     return pcms->sata_enabled;
1557 }
1558 
1559 static void pc_machine_set_sata(Object *obj, bool value, Error **errp)
1560 {
1561     PCMachineState *pcms = PC_MACHINE(obj);
1562 
1563     pcms->sata_enabled = value;
1564 }
1565 
1566 static bool pc_machine_get_hpet(Object *obj, Error **errp)
1567 {
1568     PCMachineState *pcms = PC_MACHINE(obj);
1569 
1570     return pcms->hpet_enabled;
1571 }
1572 
1573 static void pc_machine_set_hpet(Object *obj, bool value, Error **errp)
1574 {
1575     PCMachineState *pcms = PC_MACHINE(obj);
1576 
1577     pcms->hpet_enabled = value;
1578 }
1579 
1580 static bool pc_machine_get_i8042(Object *obj, Error **errp)
1581 {
1582     PCMachineState *pcms = PC_MACHINE(obj);
1583 
1584     return pcms->i8042_enabled;
1585 }
1586 
1587 static void pc_machine_set_i8042(Object *obj, bool value, Error **errp)
1588 {
1589     PCMachineState *pcms = PC_MACHINE(obj);
1590 
1591     pcms->i8042_enabled = value;
1592 }
1593 
1594 static bool pc_machine_get_default_bus_bypass_iommu(Object *obj, Error **errp)
1595 {
1596     PCMachineState *pcms = PC_MACHINE(obj);
1597 
1598     return pcms->default_bus_bypass_iommu;
1599 }
1600 
1601 static void pc_machine_set_default_bus_bypass_iommu(Object *obj, bool value,
1602                                                     Error **errp)
1603 {
1604     PCMachineState *pcms = PC_MACHINE(obj);
1605 
1606     pcms->default_bus_bypass_iommu = value;
1607 }
1608 
1609 static void pc_machine_get_smbios_ep(Object *obj, Visitor *v, const char *name,
1610                                      void *opaque, Error **errp)
1611 {
1612     PCMachineState *pcms = PC_MACHINE(obj);
1613     SmbiosEntryPointType smbios_entry_point_type = pcms->smbios_entry_point_type;
1614 
1615     visit_type_SmbiosEntryPointType(v, name, &smbios_entry_point_type, errp);
1616 }
1617 
1618 static void pc_machine_set_smbios_ep(Object *obj, Visitor *v, const char *name,
1619                                      void *opaque, Error **errp)
1620 {
1621     PCMachineState *pcms = PC_MACHINE(obj);
1622 
1623     visit_type_SmbiosEntryPointType(v, name, &pcms->smbios_entry_point_type, errp);
1624 }
1625 
1626 static void pc_machine_get_max_ram_below_4g(Object *obj, Visitor *v,
1627                                             const char *name, void *opaque,
1628                                             Error **errp)
1629 {
1630     PCMachineState *pcms = PC_MACHINE(obj);
1631     uint64_t value = pcms->max_ram_below_4g;
1632 
1633     visit_type_size(v, name, &value, errp);
1634 }
1635 
1636 static void pc_machine_set_max_ram_below_4g(Object *obj, Visitor *v,
1637                                             const char *name, void *opaque,
1638                                             Error **errp)
1639 {
1640     PCMachineState *pcms = PC_MACHINE(obj);
1641     uint64_t value;
1642 
1643     if (!visit_type_size(v, name, &value, errp)) {
1644         return;
1645     }
1646     if (value > 4 * GiB) {
1647         error_setg(errp,
1648                    "Machine option 'max-ram-below-4g=%"PRIu64
1649                    "' expects size less than or equal to 4G", value);
1650         return;
1651     }
1652 
1653     if (value < 1 * MiB) {
1654         warn_report("Only %" PRIu64 " bytes of RAM below the 4GiB boundary,"
1655                     "BIOS may not work with less than 1MiB", value);
1656     }
1657 
1658     pcms->max_ram_below_4g = value;
1659 }
1660 
1661 static void pc_machine_get_max_fw_size(Object *obj, Visitor *v,
1662                                        const char *name, void *opaque,
1663                                        Error **errp)
1664 {
1665     PCMachineState *pcms = PC_MACHINE(obj);
1666     uint64_t value = pcms->max_fw_size;
1667 
1668     visit_type_size(v, name, &value, errp);
1669 }
1670 
1671 static void pc_machine_set_max_fw_size(Object *obj, Visitor *v,
1672                                        const char *name, void *opaque,
1673                                        Error **errp)
1674 {
1675     PCMachineState *pcms = PC_MACHINE(obj);
1676     uint64_t value;
1677 
1678     if (!visit_type_size(v, name, &value, errp)) {
1679         return;
1680     }
1681 
1682     /*
1683      * We don't have a theoretically justifiable exact lower bound on the base
1684      * address of any flash mapping. In practice, the IO-APIC MMIO range is
1685      * [0xFEE00000..0xFEE01000] -- see IO_APIC_DEFAULT_ADDRESS --, leaving free
1686      * only 18MiB-4KiB below 4GiB. For now, restrict the cumulative mapping to
1687      * 16MiB in size.
1688      */
1689     if (value > 16 * MiB) {
1690         error_setg(errp,
1691                    "User specified max allowed firmware size %" PRIu64 " is "
1692                    "greater than 16MiB. If combined firmware size exceeds "
1693                    "16MiB the system may not boot, or experience intermittent"
1694                    "stability issues.",
1695                    value);
1696         return;
1697     }
1698 
1699     pcms->max_fw_size = value;
1700 }
1701 
1702 
1703 static void pc_machine_initfn(Object *obj)
1704 {
1705     PCMachineState *pcms = PC_MACHINE(obj);
1706     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1707 
1708 #ifdef CONFIG_VMPORT
1709     pcms->vmport = ON_OFF_AUTO_AUTO;
1710 #else
1711     pcms->vmport = ON_OFF_AUTO_OFF;
1712 #endif /* CONFIG_VMPORT */
1713     pcms->max_ram_below_4g = 0; /* use default */
1714     pcms->smbios_entry_point_type = pcmc->default_smbios_ep_type;
1715     pcms->south_bridge = pcmc->default_south_bridge;
1716 
1717     /* acpi build is enabled by default if machine supports it */
1718     pcms->acpi_build_enabled = pcmc->has_acpi_build;
1719     pcms->smbus_enabled = true;
1720     pcms->sata_enabled = true;
1721     pcms->i8042_enabled = true;
1722     pcms->max_fw_size = 8 * MiB;
1723 #ifdef CONFIG_HPET
1724     pcms->hpet_enabled = true;
1725 #endif
1726     pcms->default_bus_bypass_iommu = false;
1727 
1728     pcms->pcspk = isa_new(TYPE_PC_SPEAKER);
1729     object_property_add_alias(OBJECT(pcms), "pcspk-audiodev",
1730                               OBJECT(pcms->pcspk), "audiodev");
1731     cxl_machine_init(obj, &pcms->cxl_devices_state);
1732 
1733     pcms->machine_done.notify = pc_machine_done;
1734     qemu_add_machine_init_done_notifier(&pcms->machine_done);
1735 }
1736 
1737 static void pc_machine_reset(MachineState *machine, ShutdownCause reason)
1738 {
1739     CPUState *cs;
1740     X86CPU *cpu;
1741 
1742     qemu_devices_reset(reason);
1743 
1744     /* Reset APIC after devices have been reset to cancel
1745      * any changes that qemu_devices_reset() might have done.
1746      */
1747     CPU_FOREACH(cs) {
1748         cpu = X86_CPU(cs);
1749 
1750         x86_cpu_after_reset(cpu);
1751     }
1752 }
1753 
1754 static void pc_machine_wakeup(MachineState *machine)
1755 {
1756     cpu_synchronize_all_states();
1757     pc_machine_reset(machine, SHUTDOWN_CAUSE_NONE);
1758     cpu_synchronize_all_post_reset();
1759 }
1760 
1761 static bool pc_hotplug_allowed(MachineState *ms, DeviceState *dev, Error **errp)
1762 {
1763     X86IOMMUState *iommu = x86_iommu_get_default();
1764     IntelIOMMUState *intel_iommu;
1765 
1766     if (iommu &&
1767         object_dynamic_cast((Object *)iommu, TYPE_INTEL_IOMMU_DEVICE) &&
1768         object_dynamic_cast((Object *)dev, "vfio-pci")) {
1769         intel_iommu = INTEL_IOMMU_DEVICE(iommu);
1770         if (!intel_iommu->caching_mode) {
1771             error_setg(errp, "Device assignment is not allowed without "
1772                        "enabling caching-mode=on for Intel IOMMU.");
1773             return false;
1774         }
1775     }
1776 
1777     return true;
1778 }
1779 
1780 static void pc_machine_class_init(ObjectClass *oc, void *data)
1781 {
1782     MachineClass *mc = MACHINE_CLASS(oc);
1783     X86MachineClass *x86mc = X86_MACHINE_CLASS(oc);
1784     PCMachineClass *pcmc = PC_MACHINE_CLASS(oc);
1785     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
1786 
1787     pcmc->pci_enabled = true;
1788     pcmc->has_acpi_build = true;
1789     pcmc->rsdp_in_ram = true;
1790     pcmc->smbios_defaults = true;
1791     pcmc->smbios_uuid_encoded = true;
1792     pcmc->gigabyte_align = true;
1793     pcmc->has_reserved_memory = true;
1794     pcmc->enforce_aligned_dimm = true;
1795     pcmc->enforce_amd_1tb_hole = true;
1796     /* BIOS ACPI tables: 128K. Other BIOS datastructures: less than 4K reported
1797      * to be used at the moment, 32K should be enough for a while.  */
1798     pcmc->acpi_data_size = 0x20000 + 0x8000;
1799     pcmc->pvh_enabled = true;
1800     pcmc->kvmclock_create_always = true;
1801     pcmc->resizable_acpi_blob = true;
1802     x86mc->apic_xrupt_override = true;
1803     assert(!mc->get_hotplug_handler);
1804     mc->get_hotplug_handler = pc_get_hotplug_handler;
1805     mc->hotplug_allowed = pc_hotplug_allowed;
1806     mc->cpu_index_to_instance_props = x86_cpu_index_to_props;
1807     mc->get_default_cpu_node_id = x86_get_default_cpu_node_id;
1808     mc->possible_cpu_arch_ids = x86_possible_cpu_arch_ids;
1809     mc->auto_enable_numa_with_memhp = true;
1810     mc->auto_enable_numa_with_memdev = true;
1811     mc->has_hotpluggable_cpus = true;
1812     mc->default_boot_order = "cad";
1813     mc->block_default_type = IF_IDE;
1814     mc->max_cpus = 255;
1815     mc->reset = pc_machine_reset;
1816     mc->wakeup = pc_machine_wakeup;
1817     hc->pre_plug = pc_machine_device_pre_plug_cb;
1818     hc->plug = pc_machine_device_plug_cb;
1819     hc->unplug_request = pc_machine_device_unplug_request_cb;
1820     hc->unplug = pc_machine_device_unplug_cb;
1821     mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE;
1822     mc->nvdimm_supported = true;
1823     mc->smp_props.dies_supported = true;
1824     mc->default_ram_id = "pc.ram";
1825     pcmc->default_smbios_ep_type = SMBIOS_ENTRY_POINT_TYPE_64;
1826 
1827     object_class_property_add(oc, PC_MACHINE_MAX_RAM_BELOW_4G, "size",
1828         pc_machine_get_max_ram_below_4g, pc_machine_set_max_ram_below_4g,
1829         NULL, NULL);
1830     object_class_property_set_description(oc, PC_MACHINE_MAX_RAM_BELOW_4G,
1831         "Maximum ram below the 4G boundary (32bit boundary)");
1832 
1833     object_class_property_add(oc, PC_MACHINE_VMPORT, "OnOffAuto",
1834         pc_machine_get_vmport, pc_machine_set_vmport,
1835         NULL, NULL);
1836     object_class_property_set_description(oc, PC_MACHINE_VMPORT,
1837         "Enable vmport (pc & q35)");
1838 
1839     object_class_property_add_bool(oc, PC_MACHINE_SMBUS,
1840         pc_machine_get_smbus, pc_machine_set_smbus);
1841     object_class_property_set_description(oc, PC_MACHINE_SMBUS,
1842         "Enable/disable system management bus");
1843 
1844     object_class_property_add_bool(oc, PC_MACHINE_SATA,
1845         pc_machine_get_sata, pc_machine_set_sata);
1846     object_class_property_set_description(oc, PC_MACHINE_SATA,
1847         "Enable/disable Serial ATA bus");
1848 
1849     object_class_property_add_bool(oc, "hpet",
1850         pc_machine_get_hpet, pc_machine_set_hpet);
1851     object_class_property_set_description(oc, "hpet",
1852         "Enable/disable high precision event timer emulation");
1853 
1854     object_class_property_add_bool(oc, PC_MACHINE_I8042,
1855         pc_machine_get_i8042, pc_machine_set_i8042);
1856 
1857     object_class_property_add_bool(oc, "default-bus-bypass-iommu",
1858         pc_machine_get_default_bus_bypass_iommu,
1859         pc_machine_set_default_bus_bypass_iommu);
1860 
1861     object_class_property_add(oc, PC_MACHINE_MAX_FW_SIZE, "size",
1862         pc_machine_get_max_fw_size, pc_machine_set_max_fw_size,
1863         NULL, NULL);
1864     object_class_property_set_description(oc, PC_MACHINE_MAX_FW_SIZE,
1865         "Maximum combined firmware size");
1866 
1867     object_class_property_add(oc, PC_MACHINE_SMBIOS_EP, "str",
1868         pc_machine_get_smbios_ep, pc_machine_set_smbios_ep,
1869         NULL, NULL);
1870     object_class_property_set_description(oc, PC_MACHINE_SMBIOS_EP,
1871         "SMBIOS Entry Point type [32, 64]");
1872 }
1873 
1874 static const TypeInfo pc_machine_info = {
1875     .name = TYPE_PC_MACHINE,
1876     .parent = TYPE_X86_MACHINE,
1877     .abstract = true,
1878     .instance_size = sizeof(PCMachineState),
1879     .instance_init = pc_machine_initfn,
1880     .class_size = sizeof(PCMachineClass),
1881     .class_init = pc_machine_class_init,
1882     .interfaces = (InterfaceInfo[]) {
1883          { TYPE_HOTPLUG_HANDLER },
1884          { }
1885     },
1886 };
1887 
1888 static void pc_machine_register_types(void)
1889 {
1890     type_register_static(&pc_machine_info);
1891 }
1892 
1893 type_init(pc_machine_register_types)
1894