xref: /openbmc/qemu/hw/i386/pc.c (revision 70c66701f3915d009d66a223c98914e34a093a61)
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 "kvm/tdx.h"
48 #include "hw/xen/xen.h"
49 #include "qobject/qlist.h"
50 #include "qemu/error-report.h"
51 #include "hw/acpi/cpu_hotplug.h"
52 #include "acpi-build.h"
53 #include "hw/mem/nvdimm.h"
54 #include "hw/cxl/cxl_host.h"
55 #include "hw/usb.h"
56 #include "hw/i386/intel_iommu.h"
57 #include "hw/net/ne2000-isa.h"
58 #include "hw/virtio/virtio-iommu.h"
59 #include "hw/virtio/virtio-md-pci.h"
60 #include "hw/i386/kvm/xen_overlay.h"
61 #include "hw/i386/kvm/xen_evtchn.h"
62 #include "hw/i386/kvm/xen_gnttab.h"
63 #include "hw/i386/kvm/xen_xenstore.h"
64 #include "hw/mem/memory-device.h"
65 #include "e820_memory_layout.h"
66 #include "trace.h"
67 #include "sev.h"
68 #include CONFIG_DEVICES
69 
70 #ifdef CONFIG_XEN_EMU
71 #include "hw/xen/xen-legacy-backend.h"
72 #include "hw/xen/xen-bus.h"
73 #endif
74 
75 /*
76  * Helper for setting model-id for CPU models that changed model-id
77  * depending on QEMU versions up to QEMU 2.4.
78  */
79 #define PC_CPU_MODEL_IDS(v) \
80     { "qemu32-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
81     { "qemu64-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
82     { "athlon-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },
83 
84 GlobalProperty pc_compat_10_0[] = {
85     { TYPE_X86_CPU, "x-consistent-cache", "false" },
86     { TYPE_X86_CPU, "x-vendor-cpuid-only-v2", "false" },
87     { TYPE_X86_CPU, "x-arch-cap-always-on", "true" },
88     { TYPE_X86_CPU, "x-pdcm-on-even-without-pmu", "true" },
89 };
90 const size_t pc_compat_10_0_len = G_N_ELEMENTS(pc_compat_10_0);
91 
92 GlobalProperty pc_compat_9_2[] = {};
93 const size_t pc_compat_9_2_len = G_N_ELEMENTS(pc_compat_9_2);
94 
95 GlobalProperty pc_compat_9_1[] = {
96     { "ICH9-LPC", "x-smi-swsmi-timer", "off" },
97     { "ICH9-LPC", "x-smi-periodic-timer", "off" },
98     { TYPE_INTEL_IOMMU_DEVICE, "stale-tm", "on" },
99     { TYPE_INTEL_IOMMU_DEVICE, "aw-bits", "39" },
100 };
101 const size_t pc_compat_9_1_len = G_N_ELEMENTS(pc_compat_9_1);
102 
103 GlobalProperty pc_compat_9_0[] = {
104     { TYPE_X86_CPU, "x-amd-topoext-features-only", "false" },
105     { TYPE_X86_CPU, "x-l1-cache-per-thread", "false" },
106     { TYPE_X86_CPU, "guest-phys-bits", "0" },
107     { "sev-guest", "legacy-vm-type", "on" },
108     { TYPE_X86_CPU, "legacy-multi-node", "on" },
109 };
110 const size_t pc_compat_9_0_len = G_N_ELEMENTS(pc_compat_9_0);
111 
112 GlobalProperty pc_compat_8_2[] = {};
113 const size_t pc_compat_8_2_len = G_N_ELEMENTS(pc_compat_8_2);
114 
115 GlobalProperty pc_compat_8_1[] = {};
116 const size_t pc_compat_8_1_len = G_N_ELEMENTS(pc_compat_8_1);
117 
118 GlobalProperty pc_compat_8_0[] = {
119     { "virtio-mem", "unplugged-inaccessible", "auto" },
120 };
121 const size_t pc_compat_8_0_len = G_N_ELEMENTS(pc_compat_8_0);
122 
123 GlobalProperty pc_compat_7_2[] = {
124     { "ICH9-LPC", "noreboot", "true" },
125 };
126 const size_t pc_compat_7_2_len = G_N_ELEMENTS(pc_compat_7_2);
127 
128 GlobalProperty pc_compat_7_1[] = {};
129 const size_t pc_compat_7_1_len = G_N_ELEMENTS(pc_compat_7_1);
130 
131 GlobalProperty pc_compat_7_0[] = {};
132 const size_t pc_compat_7_0_len = G_N_ELEMENTS(pc_compat_7_0);
133 
134 GlobalProperty pc_compat_6_2[] = {
135     { "virtio-mem", "unplugged-inaccessible", "off" },
136 };
137 const size_t pc_compat_6_2_len = G_N_ELEMENTS(pc_compat_6_2);
138 
139 GlobalProperty pc_compat_6_1[] = {
140     { TYPE_X86_CPU, "hv-version-id-build", "0x1bbc" },
141     { TYPE_X86_CPU, "hv-version-id-major", "0x0006" },
142     { TYPE_X86_CPU, "hv-version-id-minor", "0x0001" },
143     { "ICH9-LPC", "x-keep-pci-slot-hpc", "false" },
144 };
145 const size_t pc_compat_6_1_len = G_N_ELEMENTS(pc_compat_6_1);
146 
147 GlobalProperty pc_compat_6_0[] = {
148     { "qemu64" "-" TYPE_X86_CPU, "family", "6" },
149     { "qemu64" "-" TYPE_X86_CPU, "model", "6" },
150     { "qemu64" "-" TYPE_X86_CPU, "stepping", "3" },
151     { TYPE_X86_CPU, "x-vendor-cpuid-only", "off" },
152     { "ICH9-LPC", ACPI_PM_PROP_ACPI_PCIHP_BRIDGE, "off" },
153     { "ICH9-LPC", "x-keep-pci-slot-hpc", "true" },
154 };
155 const size_t pc_compat_6_0_len = G_N_ELEMENTS(pc_compat_6_0);
156 
157 GlobalProperty pc_compat_5_2[] = {
158     { "ICH9-LPC", "x-smi-cpu-hotunplug", "off" },
159 };
160 const size_t pc_compat_5_2_len = G_N_ELEMENTS(pc_compat_5_2);
161 
162 GlobalProperty pc_compat_5_1[] = {
163     { "ICH9-LPC", "x-smi-cpu-hotplug", "off" },
164     { TYPE_X86_CPU, "kvm-msi-ext-dest-id", "off" },
165 };
166 const size_t pc_compat_5_1_len = G_N_ELEMENTS(pc_compat_5_1);
167 
168 GlobalProperty pc_compat_5_0[] = {
169 };
170 const size_t pc_compat_5_0_len = G_N_ELEMENTS(pc_compat_5_0);
171 
172 GlobalProperty pc_compat_4_2[] = {
173     { "mch", "smbase-smram", "off" },
174 };
175 const size_t pc_compat_4_2_len = G_N_ELEMENTS(pc_compat_4_2);
176 
177 GlobalProperty pc_compat_4_1[] = {};
178 const size_t pc_compat_4_1_len = G_N_ELEMENTS(pc_compat_4_1);
179 
180 GlobalProperty pc_compat_4_0[] = {};
181 const size_t pc_compat_4_0_len = G_N_ELEMENTS(pc_compat_4_0);
182 
183 GlobalProperty pc_compat_3_1[] = {
184     { "intel-iommu", "dma-drain", "off" },
185     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "off" },
186     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "off" },
187     { "Opteron_G4" "-" TYPE_X86_CPU, "npt", "off" },
188     { "Opteron_G4" "-" TYPE_X86_CPU, "nrip-save", "off" },
189     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "off" },
190     { "Opteron_G5" "-" TYPE_X86_CPU, "npt", "off" },
191     { "Opteron_G5" "-" TYPE_X86_CPU, "nrip-save", "off" },
192     { "EPYC" "-" TYPE_X86_CPU, "npt", "off" },
193     { "EPYC" "-" TYPE_X86_CPU, "nrip-save", "off" },
194     { "EPYC-IBPB" "-" TYPE_X86_CPU, "npt", "off" },
195     { "EPYC-IBPB" "-" TYPE_X86_CPU, "nrip-save", "off" },
196     { "Skylake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
197     { "Skylake-Client-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
198     { "Skylake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
199     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
200     { "Cascadelake-Server" "-" TYPE_X86_CPU,  "mpx", "on" },
201     { "Icelake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
202     { "Icelake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
203     { "Cascadelake-Server" "-" TYPE_X86_CPU, "stepping", "5" },
204     { TYPE_X86_CPU, "x-intel-pt-auto-level", "off" },
205 };
206 const size_t pc_compat_3_1_len = G_N_ELEMENTS(pc_compat_3_1);
207 
208 GlobalProperty pc_compat_3_0[] = {
209     { TYPE_X86_CPU, "x-hv-synic-kvm-only", "on" },
210     { "Skylake-Server" "-" TYPE_X86_CPU, "pku", "off" },
211     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "pku", "off" },
212 };
213 const size_t pc_compat_3_0_len = G_N_ELEMENTS(pc_compat_3_0);
214 
215 GlobalProperty pc_compat_2_12[] = {
216     { TYPE_X86_CPU, "legacy-cache", "on" },
217     { TYPE_X86_CPU, "topoext", "off" },
218     { "EPYC-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
219     { "EPYC-IBPB-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
220 };
221 const size_t pc_compat_2_12_len = G_N_ELEMENTS(pc_compat_2_12);
222 
223 GlobalProperty pc_compat_2_11[] = {
224     { TYPE_X86_CPU, "x-migrate-smi-count", "off" },
225     { "Skylake-Server" "-" TYPE_X86_CPU, "clflushopt", "off" },
226 };
227 const size_t pc_compat_2_11_len = G_N_ELEMENTS(pc_compat_2_11);
228 
229 GlobalProperty pc_compat_2_10[] = {
230     { TYPE_X86_CPU, "x-hv-max-vps", "0x40" },
231     { "i440FX-pcihost", "x-pci-hole64-fix", "off" },
232     { "q35-pcihost", "x-pci-hole64-fix", "off" },
233 };
234 const size_t pc_compat_2_10_len = G_N_ELEMENTS(pc_compat_2_10);
235 
236 GlobalProperty pc_compat_2_9[] = {
237     { "mch", "extended-tseg-mbytes", "0" },
238 };
239 const size_t pc_compat_2_9_len = G_N_ELEMENTS(pc_compat_2_9);
240 
241 GlobalProperty pc_compat_2_8[] = {
242     { TYPE_X86_CPU, "tcg-cpuid", "off" },
243     { "kvmclock", "x-mach-use-reliable-get-clock", "off" },
244     { "ICH9-LPC", "x-smi-broadcast", "off" },
245     { TYPE_X86_CPU, "vmware-cpuid-freq", "off" },
246     { "Haswell-" TYPE_X86_CPU, "stepping", "1" },
247 };
248 const size_t pc_compat_2_8_len = G_N_ELEMENTS(pc_compat_2_8);
249 
250 GlobalProperty pc_compat_2_7[] = {
251     { TYPE_X86_CPU, "l3-cache", "off" },
252     { TYPE_X86_CPU, "full-cpuid-auto-level", "off" },
253     { "Opteron_G3" "-" TYPE_X86_CPU, "family", "15" },
254     { "Opteron_G3" "-" TYPE_X86_CPU, "model", "6" },
255     { "Opteron_G3" "-" TYPE_X86_CPU, "stepping", "1" },
256     { "isa-pcspk", "migrate", "off" },
257 };
258 const size_t pc_compat_2_7_len = G_N_ELEMENTS(pc_compat_2_7);
259 
260 GlobalProperty pc_compat_2_6[] = {
261     { TYPE_X86_CPU, "cpuid-0xb", "off" },
262     { "vmxnet3", "romfile", "" },
263     { TYPE_X86_CPU, "fill-mtrr-mask", "off" },
264     { "apic-common", "legacy-instance-id", "on", }
265 };
266 const size_t pc_compat_2_6_len = G_N_ELEMENTS(pc_compat_2_6);
267 
268 /*
269  * @PC_FW_DATA:
270  * Size of the chunk of memory at the top of RAM for the BIOS ACPI tables
271  * and other BIOS datastructures.
272  *
273  * BIOS ACPI tables: 128K. Other BIOS datastructures: less than 4K
274  * reported to be used at the moment, 32K should be enough for a while.
275  */
276 #define PC_FW_DATA (0x20000 + 0x8000)
277 
pc_gsi_create(qemu_irq ** irqs,bool pci_enabled)278 GSIState *pc_gsi_create(qemu_irq **irqs, bool pci_enabled)
279 {
280     GSIState *s;
281 
282     s = g_new0(GSIState, 1);
283     if (kvm_ioapic_in_kernel()) {
284         kvm_pc_setup_irq_routing(pci_enabled);
285     }
286     *irqs = qemu_allocate_irqs(gsi_handler, s, IOAPIC_NUM_PINS);
287 
288     return s;
289 }
290 
ioport80_write(void * opaque,hwaddr addr,uint64_t data,unsigned size)291 static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
292                            unsigned size)
293 {
294 }
295 
ioport80_read(void * opaque,hwaddr addr,unsigned size)296 static uint64_t ioport80_read(void *opaque, hwaddr addr, unsigned size)
297 {
298     return 0xffffffffffffffffULL;
299 }
300 
301 /* MS-DOS compatibility mode FPU exception support */
ioportF0_write(void * opaque,hwaddr addr,uint64_t data,unsigned size)302 static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
303                            unsigned size)
304 {
305     if (tcg_enabled()) {
306         cpu_set_ignne();
307     }
308 }
309 
ioportF0_read(void * opaque,hwaddr addr,unsigned size)310 static uint64_t ioportF0_read(void *opaque, hwaddr addr, unsigned size)
311 {
312     return 0xffffffffffffffffULL;
313 }
314 
315 /* PC cmos mappings */
316 
317 #define REG_EQUIPMENT_BYTE          0x14
318 
cmos_init_hd(MC146818RtcState * s,int type_ofs,int info_ofs,int16_t cylinders,int8_t heads,int8_t sectors)319 static void cmos_init_hd(MC146818RtcState *s, int type_ofs, int info_ofs,
320                          int16_t cylinders, int8_t heads, int8_t sectors)
321 {
322     mc146818rtc_set_cmos_data(s, type_ofs, 47);
323     mc146818rtc_set_cmos_data(s, info_ofs, cylinders);
324     mc146818rtc_set_cmos_data(s, info_ofs + 1, cylinders >> 8);
325     mc146818rtc_set_cmos_data(s, info_ofs + 2, heads);
326     mc146818rtc_set_cmos_data(s, info_ofs + 3, 0xff);
327     mc146818rtc_set_cmos_data(s, info_ofs + 4, 0xff);
328     mc146818rtc_set_cmos_data(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
329     mc146818rtc_set_cmos_data(s, info_ofs + 6, cylinders);
330     mc146818rtc_set_cmos_data(s, info_ofs + 7, cylinders >> 8);
331     mc146818rtc_set_cmos_data(s, info_ofs + 8, sectors);
332 }
333 
334 /* convert boot_device letter to something recognizable by the bios */
boot_device2nibble(char boot_device)335 static int boot_device2nibble(char boot_device)
336 {
337     switch(boot_device) {
338     case 'a':
339     case 'b':
340         return 0x01; /* floppy boot */
341     case 'c':
342         return 0x02; /* hard drive boot */
343     case 'd':
344         return 0x03; /* CD-ROM boot */
345     case 'n':
346         return 0x04; /* Network boot */
347     }
348     return 0;
349 }
350 
set_boot_dev(PCMachineState * pcms,MC146818RtcState * s,const char * boot_device,Error ** errp)351 static void set_boot_dev(PCMachineState *pcms, MC146818RtcState *s,
352                          const char *boot_device, Error **errp)
353 {
354 #define PC_MAX_BOOT_DEVICES 3
355     int nbds, bds[3] = { 0, };
356     int i;
357 
358     nbds = strlen(boot_device);
359     if (nbds > PC_MAX_BOOT_DEVICES) {
360         error_setg(errp, "Too many boot devices for PC");
361         return;
362     }
363     for (i = 0; i < nbds; i++) {
364         bds[i] = boot_device2nibble(boot_device[i]);
365         if (bds[i] == 0) {
366             error_setg(errp, "Invalid boot device for PC: '%c'",
367                        boot_device[i]);
368             return;
369         }
370     }
371     mc146818rtc_set_cmos_data(s, 0x3d, (bds[1] << 4) | bds[0]);
372     mc146818rtc_set_cmos_data(s, 0x38, (bds[2] << 4) | !pcms->fd_bootchk);
373 }
374 
pc_boot_set(void * opaque,const char * boot_device,Error ** errp)375 static void pc_boot_set(void *opaque, const char *boot_device, Error **errp)
376 {
377     PCMachineState *pcms = opaque;
378     X86MachineState *x86ms = X86_MACHINE(pcms);
379 
380     set_boot_dev(pcms, MC146818_RTC(x86ms->rtc), boot_device, errp);
381 }
382 
pc_cmos_init_floppy(MC146818RtcState * rtc_state,ISADevice * floppy)383 static void pc_cmos_init_floppy(MC146818RtcState *rtc_state, ISADevice *floppy)
384 {
385     int val, nb;
386     FloppyDriveType fd_type[2] = { FLOPPY_DRIVE_TYPE_NONE,
387                                    FLOPPY_DRIVE_TYPE_NONE };
388 
389 #ifdef CONFIG_FDC_ISA
390     /* floppy type */
391     if (floppy) {
392         for (int i = 0; i < 2; i++) {
393             fd_type[i] = isa_fdc_get_drive_type(floppy, i);
394         }
395     }
396 #endif
397 
398     val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
399         cmos_get_fd_drive_type(fd_type[1]);
400     mc146818rtc_set_cmos_data(rtc_state, 0x10, val);
401 
402     val = mc146818rtc_get_cmos_data(rtc_state, REG_EQUIPMENT_BYTE);
403     nb = 0;
404     if (fd_type[0] != FLOPPY_DRIVE_TYPE_NONE) {
405         nb++;
406     }
407     if (fd_type[1] != FLOPPY_DRIVE_TYPE_NONE) {
408         nb++;
409     }
410     switch (nb) {
411     case 0:
412         break;
413     case 1:
414         val |= 0x01; /* 1 drive, ready for boot */
415         break;
416     case 2:
417         val |= 0x41; /* 2 drives, ready for boot */
418         break;
419     }
420     mc146818rtc_set_cmos_data(rtc_state, REG_EQUIPMENT_BYTE, val);
421 }
422 
423 typedef struct check_fdc_state {
424     ISADevice *floppy;
425     bool multiple;
426 } CheckFdcState;
427 
check_fdc(Object * obj,void * opaque)428 static int check_fdc(Object *obj, void *opaque)
429 {
430     CheckFdcState *state = opaque;
431     Object *fdc;
432     uint32_t iobase;
433     Error *local_err = NULL;
434 
435     fdc = object_dynamic_cast(obj, TYPE_ISA_FDC);
436     if (!fdc) {
437         return 0;
438     }
439 
440     iobase = object_property_get_uint(obj, "iobase", &local_err);
441     if (local_err || iobase != 0x3f0) {
442         error_free(local_err);
443         return 0;
444     }
445 
446     if (state->floppy) {
447         state->multiple = true;
448     } else {
449         state->floppy = ISA_DEVICE(obj);
450     }
451     return 0;
452 }
453 
454 static const char * const fdc_container_path[] = {
455     "unattached", "peripheral", "peripheral-anon"
456 };
457 
458 /*
459  * Locate the FDC at IO address 0x3f0, in order to configure the CMOS registers
460  * and ACPI objects.
461  */
pc_find_fdc0(void)462 static ISADevice *pc_find_fdc0(void)
463 {
464     int i;
465     Object *container;
466     CheckFdcState state = { 0 };
467 
468     for (i = 0; i < ARRAY_SIZE(fdc_container_path); i++) {
469         container = machine_get_container(fdc_container_path[i]);
470         object_child_foreach(container, check_fdc, &state);
471     }
472 
473     if (state.multiple) {
474         warn_report("multiple floppy disk controllers with "
475                     "iobase=0x3f0 have been found");
476         error_printf("the one being picked for CMOS setup might not reflect "
477                      "your intent");
478     }
479 
480     return state.floppy;
481 }
482 
pc_cmos_init_late(PCMachineState * pcms)483 static void pc_cmos_init_late(PCMachineState *pcms)
484 {
485     X86MachineState *x86ms = X86_MACHINE(pcms);
486     MC146818RtcState *s = MC146818_RTC(x86ms->rtc);
487     int16_t cylinders;
488     int8_t heads, sectors;
489     int val;
490     int i, trans;
491 
492     val = 0;
493     if (pcms->idebus[0] &&
494         ide_get_geometry(pcms->idebus[0], 0,
495                          &cylinders, &heads, &sectors) >= 0) {
496         cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
497         val |= 0xf0;
498     }
499     if (pcms->idebus[0] &&
500         ide_get_geometry(pcms->idebus[0], 1,
501                          &cylinders, &heads, &sectors) >= 0) {
502         cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
503         val |= 0x0f;
504     }
505     mc146818rtc_set_cmos_data(s, 0x12, val);
506 
507     val = 0;
508     for (i = 0; i < 4; i++) {
509         /* NOTE: ide_get_geometry() returns the physical
510            geometry.  It is always such that: 1 <= sects <= 63, 1
511            <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
512            geometry can be different if a translation is done. */
513         BusState *idebus = pcms->idebus[i / 2];
514         if (idebus &&
515             ide_get_geometry(idebus, i % 2,
516                              &cylinders, &heads, &sectors) >= 0) {
517             trans = ide_get_bios_chs_trans(idebus, i % 2) - 1;
518             assert((trans & ~3) == 0);
519             val |= trans << (i * 2);
520         }
521     }
522     mc146818rtc_set_cmos_data(s, 0x39, val);
523 
524     pc_cmos_init_floppy(s, pc_find_fdc0());
525 
526     /* various important CMOS locations needed by PC/Bochs bios */
527 
528     /* memory size */
529     /* base memory (first MiB) */
530     val = MIN(x86ms->below_4g_mem_size / KiB, 640);
531     mc146818rtc_set_cmos_data(s, 0x15, val);
532     mc146818rtc_set_cmos_data(s, 0x16, val >> 8);
533     /* extended memory (next 64MiB) */
534     if (x86ms->below_4g_mem_size > 1 * MiB) {
535         val = (x86ms->below_4g_mem_size - 1 * MiB) / KiB;
536     } else {
537         val = 0;
538     }
539     if (val > 65535)
540         val = 65535;
541     mc146818rtc_set_cmos_data(s, 0x17, val);
542     mc146818rtc_set_cmos_data(s, 0x18, val >> 8);
543     mc146818rtc_set_cmos_data(s, 0x30, val);
544     mc146818rtc_set_cmos_data(s, 0x31, val >> 8);
545     /* memory between 16MiB and 4GiB */
546     if (x86ms->below_4g_mem_size > 16 * MiB) {
547         val = (x86ms->below_4g_mem_size - 16 * MiB) / (64 * KiB);
548     } else {
549         val = 0;
550     }
551     if (val > 65535)
552         val = 65535;
553     mc146818rtc_set_cmos_data(s, 0x34, val);
554     mc146818rtc_set_cmos_data(s, 0x35, val >> 8);
555     /* memory above 4GiB */
556     val = x86ms->above_4g_mem_size / 65536;
557     mc146818rtc_set_cmos_data(s, 0x5b, val);
558     mc146818rtc_set_cmos_data(s, 0x5c, val >> 8);
559     mc146818rtc_set_cmos_data(s, 0x5d, val >> 16);
560 
561     val = 0;
562     val |= 0x02; /* FPU is there */
563     val |= 0x04; /* PS/2 mouse installed */
564     mc146818rtc_set_cmos_data(s, REG_EQUIPMENT_BYTE, val);
565 }
566 
handle_a20_line_change(void * opaque,int irq,int level)567 static void handle_a20_line_change(void *opaque, int irq, int level)
568 {
569     X86CPU *cpu = opaque;
570 
571     /* XXX: send to all CPUs ? */
572     /* XXX: add logic to handle multiple A20 line sources */
573     x86_cpu_set_a20(cpu, level);
574 }
575 
576 #define NE2000_NB_MAX 6
577 
578 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
579                                               0x280, 0x380 };
580 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
581 
pc_init_ne2k_isa(ISABus * bus,NICInfo * nd,Error ** errp)582 static gboolean pc_init_ne2k_isa(ISABus *bus, NICInfo *nd, Error **errp)
583 {
584     static int nb_ne2k = 0;
585 
586     if (nb_ne2k == NE2000_NB_MAX) {
587         error_setg(errp,
588                    "maximum number of ISA NE2000 devices exceeded");
589         return false;
590     }
591     isa_ne2000_init(bus, ne2000_io[nb_ne2k],
592                     ne2000_irq[nb_ne2k], nd);
593     nb_ne2k++;
594     return true;
595 }
596 
pc_acpi_smi_interrupt(void * opaque,int irq,int level)597 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
598 {
599     X86CPU *cpu = opaque;
600 
601     if (level) {
602         cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
603     }
604 }
605 
606 static
pc_machine_done(Notifier * notifier,void * data)607 void pc_machine_done(Notifier *notifier, void *data)
608 {
609     PCMachineState *pcms = container_of(notifier,
610                                         PCMachineState, machine_done);
611     X86MachineState *x86ms = X86_MACHINE(pcms);
612 
613     cxl_hook_up_pxb_registers(pcms->pcibus, &pcms->cxl_devices_state,
614                               &error_fatal);
615 
616     if (pcms->cxl_devices_state.is_enabled) {
617         cxl_fmws_link_targets(&error_fatal);
618     }
619 
620     /* set the number of CPUs */
621     x86_rtc_set_cpus_count(x86ms->rtc, x86ms->boot_cpus);
622 
623     pci_bus_add_fw_cfg_extra_pci_roots(x86ms->fw_cfg, pcms->pcibus,
624                                        &error_abort);
625 
626     acpi_setup();
627     if (x86ms->fw_cfg) {
628         fw_cfg_build_smbios(pcms, x86ms->fw_cfg, pcms->smbios_entry_point_type);
629         fw_cfg_add_e820(x86ms->fw_cfg);
630         fw_cfg_build_feature_control(MACHINE(pcms), x86ms->fw_cfg);
631         /* update FW_CFG_NB_CPUS to account for -device added CPUs */
632         fw_cfg_modify_i16(x86ms->fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
633     }
634 
635     pc_cmos_init_late(pcms);
636 }
637 
638 /* setup pci memory address space mapping into system address space */
pc_pci_as_mapping_init(MemoryRegion * system_memory,MemoryRegion * pci_address_space)639 void pc_pci_as_mapping_init(MemoryRegion *system_memory,
640                             MemoryRegion *pci_address_space)
641 {
642     /* Set to lower priority than RAM */
643     memory_region_add_subregion_overlap(system_memory, 0x0,
644                                         pci_address_space, -1);
645 }
646 
xen_load_linux(PCMachineState * pcms)647 void xen_load_linux(PCMachineState *pcms)
648 {
649     int i;
650     FWCfgState *fw_cfg;
651     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
652     X86MachineState *x86ms = X86_MACHINE(pcms);
653 
654     assert(MACHINE(pcms)->kernel_filename != NULL);
655 
656     fw_cfg = fw_cfg_init_io_dma(FW_CFG_IO_BASE, FW_CFG_IO_BASE + 4,
657                                 &address_space_memory);
658     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
659     rom_set_fw(fw_cfg);
660 
661     x86_load_linux(x86ms, fw_cfg, PC_FW_DATA, pcmc->pvh_enabled);
662     for (i = 0; i < nb_option_roms; i++) {
663         assert(!strcmp(option_rom[i].name, "linuxboot.bin") ||
664                !strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
665                !strcmp(option_rom[i].name, "pvh.bin") ||
666                !strcmp(option_rom[i].name, "multiboot.bin") ||
667                !strcmp(option_rom[i].name, "multiboot_dma.bin"));
668         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
669     }
670     x86ms->fw_cfg = fw_cfg;
671 }
672 
673 #define PC_ROM_MIN_VGA     0xc0000
674 #define PC_ROM_MIN_OPTION  0xc8000
675 #define PC_ROM_MAX         0xe0000
676 #define PC_ROM_ALIGN       0x800
677 #define PC_ROM_SIZE        (PC_ROM_MAX - PC_ROM_MIN_VGA)
678 
pc_above_4g_end(PCMachineState * pcms)679 static hwaddr pc_above_4g_end(PCMachineState *pcms)
680 {
681     X86MachineState *x86ms = X86_MACHINE(pcms);
682 
683     if (pcms->sgx_epc.size != 0) {
684         return sgx_epc_above_4g_end(&pcms->sgx_epc);
685     }
686 
687     return x86ms->above_4g_mem_start + x86ms->above_4g_mem_size;
688 }
689 
pc_get_device_memory_range(PCMachineState * pcms,hwaddr * base,ram_addr_t * device_mem_size)690 static void pc_get_device_memory_range(PCMachineState *pcms,
691                                        hwaddr *base,
692                                        ram_addr_t *device_mem_size)
693 {
694     MachineState *machine = MACHINE(pcms);
695     ram_addr_t size;
696     hwaddr addr;
697 
698     size = machine->maxram_size - machine->ram_size;
699     addr = ROUND_UP(pc_above_4g_end(pcms), 1 * GiB);
700 
701     /* size device region assuming 1G page max alignment per slot */
702     size += (1 * GiB) * machine->ram_slots;
703 
704     *base = addr;
705     *device_mem_size = size;
706 }
707 
pc_get_cxl_range_start(PCMachineState * pcms)708 static uint64_t pc_get_cxl_range_start(PCMachineState *pcms)
709 {
710     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
711     MachineState *ms = MACHINE(pcms);
712     hwaddr cxl_base;
713     ram_addr_t size;
714 
715     if (pcmc->has_reserved_memory &&
716         (ms->ram_size < ms->maxram_size)) {
717         pc_get_device_memory_range(pcms, &cxl_base, &size);
718         cxl_base += size;
719     } else {
720         cxl_base = pc_above_4g_end(pcms);
721     }
722 
723     return cxl_base;
724 }
725 
cxl_get_fmw_end(Object * obj,void * opaque)726 static int cxl_get_fmw_end(Object *obj, void *opaque)
727 {
728     struct CXLFixedWindow *fw;
729     uint64_t *start = opaque;
730 
731     if (!object_dynamic_cast(obj, TYPE_CXL_FMW)) {
732         return 0;
733     }
734     fw = CXL_FMW(obj);
735 
736     *start += fw->size;
737 
738     return 0;
739 }
740 
pc_get_cxl_range_end(PCMachineState * pcms)741 static uint64_t pc_get_cxl_range_end(PCMachineState *pcms)
742 {
743     uint64_t start = pc_get_cxl_range_start(pcms) + MiB;
744 
745     /* Ordering doesn't matter so no need to build a sorted list */
746     object_child_foreach_recursive(object_get_root(), cxl_get_fmw_end,
747                                    &start);
748     return start;
749 }
750 
pc_max_used_gpa(PCMachineState * pcms,uint64_t pci_hole64_size)751 static hwaddr pc_max_used_gpa(PCMachineState *pcms, uint64_t pci_hole64_size)
752 {
753     X86CPU *cpu = X86_CPU(first_cpu);
754     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
755     MachineState *ms = MACHINE(pcms);
756 
757     if (cpu->env.features[FEAT_8000_0001_EDX] & CPUID_EXT2_LM) {
758         /* 64-bit systems */
759         return pc_pci_hole64_start() + pci_hole64_size - 1;
760     }
761 
762     /* 32-bit systems */
763     if (pcmc->broken_32bit_mem_addr_check) {
764         /* old value for compatibility reasons */
765         return ((hwaddr)1 << cpu->phys_bits) - 1;
766     }
767 
768     /*
769      * 32-bit systems don't have hole64 but they might have a region for
770      * memory devices. Even if additional hotplugged memory devices might
771      * not be usable by most guest OSes, we need to still consider them for
772      * calculating the highest possible GPA so that we can properly report
773      * if someone configures them on a CPU that cannot possibly address them.
774      */
775     if (pcmc->has_reserved_memory &&
776         (ms->ram_size < ms->maxram_size)) {
777         hwaddr devmem_start;
778         ram_addr_t devmem_size;
779 
780         pc_get_device_memory_range(pcms, &devmem_start, &devmem_size);
781         devmem_start += devmem_size;
782         return devmem_start - 1;
783     }
784 
785     /* configuration without any memory hotplug */
786     return pc_above_4g_end(pcms) - 1;
787 }
788 
789 /*
790  * AMD systems with an IOMMU have an additional hole close to the
791  * 1Tb, which are special GPAs that cannot be DMA mapped. Depending
792  * on kernel version, VFIO may or may not let you DMA map those ranges.
793  * Starting Linux v5.4 we validate it, and can't create guests on AMD machines
794  * with certain memory sizes. It's also wrong to use those IOVA ranges
795  * in detriment of leading to IOMMU INVALID_DEVICE_REQUEST or worse.
796  * The ranges reserved for Hyper-Transport are:
797  *
798  * FD_0000_0000h - FF_FFFF_FFFFh
799  *
800  * The ranges represent the following:
801  *
802  * Base Address   Top Address  Use
803  *
804  * FD_0000_0000h FD_F7FF_FFFFh Reserved interrupt address space
805  * FD_F800_0000h FD_F8FF_FFFFh Interrupt/EOI IntCtl
806  * FD_F900_0000h FD_F90F_FFFFh Legacy PIC IACK
807  * FD_F910_0000h FD_F91F_FFFFh System Management
808  * FD_F920_0000h FD_FAFF_FFFFh Reserved Page Tables
809  * FD_FB00_0000h FD_FBFF_FFFFh Address Translation
810  * FD_FC00_0000h FD_FDFF_FFFFh I/O Space
811  * FD_FE00_0000h FD_FFFF_FFFFh Configuration
812  * FE_0000_0000h FE_1FFF_FFFFh Extended Configuration/Device Messages
813  * FE_2000_0000h FF_FFFF_FFFFh Reserved
814  *
815  * See AMD IOMMU spec, section 2.1.2 "IOMMU Logical Topology",
816  * Table 3: Special Address Controls (GPA) for more information.
817  */
818 #define AMD_HT_START         0xfd00000000UL
819 #define AMD_HT_END           0xffffffffffUL
820 #define AMD_ABOVE_1TB_START  (AMD_HT_END + 1)
821 #define AMD_HT_SIZE          (AMD_ABOVE_1TB_START - AMD_HT_START)
822 
pc_memory_init(PCMachineState * pcms,MemoryRegion * system_memory,MemoryRegion * rom_memory,uint64_t pci_hole64_size)823 void pc_memory_init(PCMachineState *pcms,
824                     MemoryRegion *system_memory,
825                     MemoryRegion *rom_memory,
826                     uint64_t pci_hole64_size)
827 {
828     int linux_boot, i;
829     MemoryRegion *option_rom_mr;
830     MemoryRegion *ram_below_4g, *ram_above_4g;
831     FWCfgState *fw_cfg;
832     MachineState *machine = MACHINE(pcms);
833     MachineClass *mc = MACHINE_GET_CLASS(machine);
834     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
835     X86MachineState *x86ms = X86_MACHINE(pcms);
836     hwaddr maxphysaddr, maxusedaddr;
837     hwaddr cxl_base, cxl_resv_end = 0;
838     X86CPU *cpu = X86_CPU(first_cpu);
839     uint64_t res_mem_end;
840 
841     assert(machine->ram_size == x86ms->below_4g_mem_size +
842                                 x86ms->above_4g_mem_size);
843 
844     linux_boot = (machine->kernel_filename != NULL);
845 
846     /*
847      * The HyperTransport range close to the 1T boundary is unique to AMD
848      * hosts with IOMMUs enabled. Restrict the ram-above-4g relocation
849      * to above 1T to AMD vCPUs only. @enforce_amd_1tb_hole is only false in
850      * older machine types (<= 7.0) for compatibility purposes.
851      */
852     if (IS_AMD_CPU(&cpu->env) && pcmc->enforce_amd_1tb_hole) {
853         /* Bail out if max possible address does not cross HT range */
854         if (pc_max_used_gpa(pcms, pci_hole64_size) >= AMD_HT_START) {
855             x86ms->above_4g_mem_start = AMD_ABOVE_1TB_START;
856         }
857 
858         /*
859          * Advertise the HT region if address space covers the reserved
860          * region or if we relocate.
861          */
862         if (cpu->phys_bits >= 40) {
863             e820_add_entry(AMD_HT_START, AMD_HT_SIZE, E820_RESERVED);
864         }
865     }
866 
867     /*
868      * phys-bits is required to be appropriately configured
869      * to make sure max used GPA is reachable.
870      */
871     maxusedaddr = pc_max_used_gpa(pcms, pci_hole64_size);
872     maxphysaddr = ((hwaddr)1 << cpu->phys_bits) - 1;
873     if (maxphysaddr < maxusedaddr) {
874         error_report("Address space limit 0x%"PRIx64" < 0x%"PRIx64
875                      " phys-bits too low (%u)",
876                      maxphysaddr, maxusedaddr, cpu->phys_bits);
877         exit(EXIT_FAILURE);
878     }
879 
880     /*
881      * Split single memory region and use aliases to address portions of it,
882      * done for backwards compatibility with older qemus.
883      */
884     ram_below_4g = g_malloc(sizeof(*ram_below_4g));
885     memory_region_init_alias(ram_below_4g, NULL, "ram-below-4g", machine->ram,
886                              0, x86ms->below_4g_mem_size);
887     memory_region_add_subregion(system_memory, 0, ram_below_4g);
888     e820_add_entry(0, x86ms->below_4g_mem_size, E820_RAM);
889     if (x86ms->above_4g_mem_size > 0) {
890         ram_above_4g = g_malloc(sizeof(*ram_above_4g));
891         memory_region_init_alias(ram_above_4g, NULL, "ram-above-4g",
892                                  machine->ram,
893                                  x86ms->below_4g_mem_size,
894                                  x86ms->above_4g_mem_size);
895         memory_region_add_subregion(system_memory, x86ms->above_4g_mem_start,
896                                     ram_above_4g);
897         e820_add_entry(x86ms->above_4g_mem_start, x86ms->above_4g_mem_size,
898                        E820_RAM);
899     }
900 
901     if (pcms->sgx_epc.size != 0) {
902         e820_add_entry(pcms->sgx_epc.base, pcms->sgx_epc.size, E820_RESERVED);
903     }
904 
905     if (!pcmc->has_reserved_memory &&
906         (machine->ram_slots ||
907          (machine->maxram_size > machine->ram_size))) {
908 
909         error_report("\"-memory 'slots|maxmem'\" is not supported by: %s",
910                      mc->name);
911         exit(EXIT_FAILURE);
912     }
913 
914     /* initialize device memory address space */
915     if (pcmc->has_reserved_memory &&
916         (machine->ram_size < machine->maxram_size)) {
917         ram_addr_t device_mem_size;
918         hwaddr device_mem_base;
919 
920         if (machine->ram_slots > ACPI_MAX_RAM_SLOTS) {
921             error_report("unsupported amount of memory slots: %"PRIu64,
922                          machine->ram_slots);
923             exit(EXIT_FAILURE);
924         }
925 
926         if (QEMU_ALIGN_UP(machine->maxram_size,
927                           TARGET_PAGE_SIZE) != machine->maxram_size) {
928             error_report("maximum memory size must by aligned to multiple of "
929                          "%d bytes", TARGET_PAGE_SIZE);
930             exit(EXIT_FAILURE);
931         }
932 
933         pc_get_device_memory_range(pcms, &device_mem_base, &device_mem_size);
934 
935         if (device_mem_base + device_mem_size < device_mem_size) {
936             error_report("unsupported amount of maximum memory: " RAM_ADDR_FMT,
937                          machine->maxram_size);
938             exit(EXIT_FAILURE);
939         }
940         machine_memory_devices_init(machine, device_mem_base, device_mem_size);
941     }
942 
943     if (pcms->cxl_devices_state.is_enabled) {
944         MemoryRegion *mr = &pcms->cxl_devices_state.host_mr;
945         hwaddr cxl_size = MiB;
946 
947         cxl_base = pc_get_cxl_range_start(pcms);
948         memory_region_init(mr, OBJECT(machine), "cxl_host_reg", cxl_size);
949         memory_region_add_subregion(system_memory, cxl_base, mr);
950         cxl_base = ROUND_UP(cxl_base + cxl_size, 256 * MiB);
951         cxl_resv_end = cxl_fmws_set_memmap(cxl_base, maxphysaddr);
952         cxl_fmws_update_mmio();
953     }
954 
955     /* Initialize PC system firmware */
956     pc_system_firmware_init(pcms, rom_memory);
957 
958     if (!is_tdx_vm()) {
959         option_rom_mr = g_malloc(sizeof(*option_rom_mr));
960         if (machine_require_guest_memfd(machine)) {
961             memory_region_init_ram_guest_memfd(option_rom_mr, NULL, "pc.rom",
962                                             PC_ROM_SIZE, &error_fatal);
963         } else {
964             memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE,
965                                 &error_fatal);
966             if (pcmc->pci_enabled) {
967                 memory_region_set_readonly(option_rom_mr, true);
968             }
969         }
970         memory_region_add_subregion_overlap(rom_memory,
971                                             PC_ROM_MIN_VGA,
972                                             option_rom_mr,
973                                             1);
974     }
975 
976     fw_cfg = fw_cfg_arch_create(machine,
977                                 x86ms->boot_cpus, x86ms->apic_id_limit);
978 
979     rom_set_fw(fw_cfg);
980 
981     if (pcms->cxl_devices_state.is_enabled) {
982         res_mem_end = cxl_resv_end;
983     } else if (machine->device_memory) {
984         res_mem_end = machine->device_memory->base
985                       + memory_region_size(&machine->device_memory->mr);
986     } else {
987         res_mem_end = 0;
988     }
989 
990     if (res_mem_end) {
991         uint64_t *val = g_malloc(sizeof(*val));
992         *val = cpu_to_le64(ROUND_UP(res_mem_end, 1 * GiB));
993         fw_cfg_add_file(fw_cfg, "etc/reserved-memory-end", val, sizeof(*val));
994     }
995 
996     if (linux_boot) {
997         x86_load_linux(x86ms, fw_cfg, PC_FW_DATA, pcmc->pvh_enabled);
998     }
999 
1000     for (i = 0; i < nb_option_roms; i++) {
1001         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1002     }
1003     x86ms->fw_cfg = fw_cfg;
1004 
1005     /* Init default IOAPIC address space */
1006     x86ms->ioapic_as = &address_space_memory;
1007 
1008     /* Init ACPI memory hotplug IO base address */
1009     pcms->memhp_io_base = ACPI_MEMORY_HOTPLUG_BASE;
1010 }
1011 
1012 /*
1013  * The 64bit pci hole starts after "above 4G RAM" and
1014  * potentially the space reserved for memory hotplug.
1015  */
pc_pci_hole64_start(void)1016 uint64_t pc_pci_hole64_start(void)
1017 {
1018     PCMachineState *pcms = PC_MACHINE(qdev_get_machine());
1019     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1020     MachineState *ms = MACHINE(pcms);
1021     uint64_t hole64_start = 0;
1022     ram_addr_t size = 0;
1023 
1024     if (pcms->cxl_devices_state.is_enabled) {
1025         hole64_start = pc_get_cxl_range_end(pcms);
1026     } else if (pcmc->has_reserved_memory && (ms->ram_size < ms->maxram_size)) {
1027         pc_get_device_memory_range(pcms, &hole64_start, &size);
1028         hole64_start += size;
1029     } else {
1030         hole64_start = pc_above_4g_end(pcms);
1031     }
1032 
1033     return ROUND_UP(hole64_start, 1 * GiB);
1034 }
1035 
pc_vga_init(ISABus * isa_bus,PCIBus * pci_bus)1036 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1037 {
1038     DeviceState *dev = NULL;
1039 
1040     if (pci_bus) {
1041         PCIDevice *pcidev = pci_vga_init(pci_bus);
1042         dev = pcidev ? &pcidev->qdev : NULL;
1043     } else if (isa_bus) {
1044         ISADevice *isadev = isa_vga_init(isa_bus);
1045         dev = isadev ? DEVICE(isadev) : NULL;
1046     }
1047 
1048     return dev;
1049 }
1050 
1051 static const MemoryRegionOps ioport80_io_ops = {
1052     .write = ioport80_write,
1053     .read = ioport80_read,
1054     .endianness = DEVICE_LITTLE_ENDIAN,
1055     .impl = {
1056         .min_access_size = 1,
1057         .max_access_size = 1,
1058     },
1059 };
1060 
1061 static const MemoryRegionOps ioportF0_io_ops = {
1062     .write = ioportF0_write,
1063     .read = ioportF0_read,
1064     .endianness = DEVICE_LITTLE_ENDIAN,
1065     .impl = {
1066         .min_access_size = 1,
1067         .max_access_size = 1,
1068     },
1069 };
1070 
pc_superio_init(ISABus * isa_bus,bool create_fdctrl,bool create_i8042,bool no_vmport,Error ** errp)1071 static void pc_superio_init(ISABus *isa_bus, bool create_fdctrl,
1072                             bool create_i8042, bool no_vmport, Error **errp)
1073 {
1074     int i;
1075     DriveInfo *fd[MAX_FD];
1076     qemu_irq *a20_line;
1077     ISADevice *i8042, *port92, *vmmouse;
1078 
1079     serial_hds_isa_init(isa_bus, 0, MAX_ISA_SERIAL_PORTS);
1080     parallel_hds_isa_init(isa_bus, MAX_PARALLEL_PORTS);
1081 
1082     for (i = 0; i < MAX_FD; i++) {
1083         fd[i] = drive_get(IF_FLOPPY, 0, i);
1084         create_fdctrl |= !!fd[i];
1085     }
1086     if (create_fdctrl) {
1087 #ifdef CONFIG_FDC_ISA
1088         ISADevice *fdc = isa_new(TYPE_ISA_FDC);
1089         if (fdc) {
1090             isa_realize_and_unref(fdc, isa_bus, &error_fatal);
1091             isa_fdc_init_drives(fdc, fd);
1092         }
1093 #endif
1094     }
1095 
1096     if (!create_i8042) {
1097         if (!no_vmport) {
1098             error_setg(errp,
1099                        "vmport requires the i8042 controller to be enabled");
1100         }
1101         return;
1102     }
1103 
1104     i8042 = isa_create_simple(isa_bus, TYPE_I8042);
1105     if (!no_vmport) {
1106         isa_create_simple(isa_bus, TYPE_VMPORT);
1107         vmmouse = isa_try_new("vmmouse");
1108     } else {
1109         vmmouse = NULL;
1110     }
1111     if (vmmouse) {
1112         object_property_set_link(OBJECT(vmmouse), TYPE_I8042, OBJECT(i8042),
1113                                  &error_abort);
1114         isa_realize_and_unref(vmmouse, isa_bus, &error_fatal);
1115     }
1116     port92 = isa_create_simple(isa_bus, TYPE_PORT92);
1117 
1118     a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1119     qdev_connect_gpio_out_named(DEVICE(i8042),
1120                                 I8042_A20_LINE, 0, a20_line[0]);
1121     qdev_connect_gpio_out_named(DEVICE(port92),
1122                                 PORT92_A20_LINE, 0, a20_line[1]);
1123     g_free(a20_line);
1124 }
1125 
pc_basic_device_init(struct PCMachineState * pcms,ISABus * isa_bus,qemu_irq * gsi,ISADevice * rtc_state,bool create_fdctrl,uint32_t hpet_irqs)1126 void pc_basic_device_init(struct PCMachineState *pcms,
1127                           ISABus *isa_bus, qemu_irq *gsi,
1128                           ISADevice *rtc_state,
1129                           bool create_fdctrl,
1130                           uint32_t hpet_irqs)
1131 {
1132     int i;
1133     DeviceState *hpet = NULL;
1134     int pit_isa_irq = 0;
1135     qemu_irq pit_alt_irq = NULL;
1136     ISADevice *pit = NULL;
1137     MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1138     MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
1139     X86MachineState *x86ms = X86_MACHINE(pcms);
1140 
1141     memory_region_init_io(ioport80_io, NULL, &ioport80_io_ops, NULL, "ioport80", 1);
1142     memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
1143 
1144     memory_region_init_io(ioportF0_io, NULL, &ioportF0_io_ops, NULL, "ioportF0", 1);
1145     memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
1146 
1147     /*
1148      * Check if an HPET shall be created.
1149      */
1150     if (pcms->hpet_enabled) {
1151         qemu_irq rtc_irq;
1152 
1153         hpet = qdev_try_new(TYPE_HPET);
1154         if (!hpet) {
1155             error_report("couldn't create HPET device");
1156             exit(1);
1157         }
1158         /*
1159          * For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-*,
1160          * use IRQ16~23, IRQ8 and IRQ2.  If the user has already set
1161          * the property, use whatever mask they specified.
1162          */
1163         uint8_t compat = object_property_get_uint(OBJECT(hpet),
1164                 HPET_INTCAP, NULL);
1165         if (!compat) {
1166             qdev_prop_set_uint32(hpet, HPET_INTCAP, hpet_irqs);
1167         }
1168         sysbus_realize_and_unref(SYS_BUS_DEVICE(hpet), &error_fatal);
1169         sysbus_mmio_map(SYS_BUS_DEVICE(hpet), 0, HPET_BASE);
1170 
1171         for (i = 0; i < IOAPIC_NUM_PINS; i++) {
1172             sysbus_connect_irq(SYS_BUS_DEVICE(hpet), i, gsi[i]);
1173         }
1174         pit_isa_irq = -1;
1175         pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1176         rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1177 
1178         /* overwrite connection created by south bridge */
1179         qdev_connect_gpio_out(DEVICE(rtc_state), 0, rtc_irq);
1180     }
1181 
1182     object_property_add_alias(OBJECT(pcms), "rtc-time", OBJECT(rtc_state),
1183                               "date");
1184 
1185 #ifdef CONFIG_XEN_EMU
1186     if (xen_mode == XEN_EMULATE) {
1187         xen_overlay_create();
1188         xen_evtchn_create(IOAPIC_NUM_PINS, gsi);
1189         xen_gnttab_create();
1190         xen_xenstore_create();
1191         if (pcms->pcibus) {
1192             pci_create_simple(pcms->pcibus, -1, "xen-platform");
1193         }
1194         xen_bus_init();
1195     }
1196 #endif
1197 
1198     qemu_register_boot_set(pc_boot_set, pcms);
1199     set_boot_dev(pcms, MC146818_RTC(rtc_state),
1200                  MACHINE(pcms)->boot_config.order, &error_fatal);
1201 
1202     if (!xen_enabled() &&
1203         (x86ms->pit == ON_OFF_AUTO_AUTO || x86ms->pit == ON_OFF_AUTO_ON)) {
1204         if (kvm_pit_in_kernel()) {
1205             pit = kvm_pit_init(isa_bus, 0x40);
1206         } else {
1207             pit = i8254_pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1208         }
1209         if (hpet) {
1210             /* connect PIT to output control line of the HPET */
1211             qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(DEVICE(pit), 0));
1212         }
1213         object_property_set_link(OBJECT(pcms->pcspk), "pit",
1214                                  OBJECT(pit), &error_fatal);
1215         isa_realize_and_unref(pcms->pcspk, isa_bus, &error_fatal);
1216     }
1217 
1218     if (pcms->vmport == ON_OFF_AUTO_AUTO) {
1219         pcms->vmport = (xen_enabled() || !pcms->i8042_enabled)
1220             ? ON_OFF_AUTO_OFF : ON_OFF_AUTO_ON;
1221     }
1222 
1223     /* Super I/O */
1224     pc_superio_init(isa_bus, create_fdctrl, pcms->i8042_enabled,
1225                     pcms->vmport != ON_OFF_AUTO_ON, &error_fatal);
1226 
1227     pcms->machine_done.notify = pc_machine_done;
1228     qemu_add_machine_init_done_notifier(&pcms->machine_done);
1229 }
1230 
pc_nic_init(PCMachineClass * pcmc,ISABus * isa_bus,PCIBus * pci_bus)1231 void pc_nic_init(PCMachineClass *pcmc, ISABus *isa_bus, PCIBus *pci_bus)
1232 {
1233     MachineClass *mc = MACHINE_CLASS(pcmc);
1234     bool default_is_ne2k = g_str_equal(mc->default_nic, TYPE_ISA_NE2000);
1235     NICInfo *nd;
1236 
1237     while ((nd = qemu_find_nic_info(TYPE_ISA_NE2000, default_is_ne2k, NULL))) {
1238         pc_init_ne2k_isa(isa_bus, nd, &error_fatal);
1239     }
1240 
1241     /* Anything remaining should be a PCI NIC */
1242     if (pci_bus) {
1243         pci_init_nic_devices(pci_bus, mc->default_nic);
1244     }
1245 }
1246 
pc_i8259_create(ISABus * isa_bus,qemu_irq * i8259_irqs)1247 void pc_i8259_create(ISABus *isa_bus, qemu_irq *i8259_irqs)
1248 {
1249     qemu_irq *i8259;
1250 
1251     if (kvm_pic_in_kernel()) {
1252         i8259 = kvm_i8259_init(isa_bus);
1253     } else if (xen_enabled()) {
1254         i8259 = xen_interrupt_controller_init();
1255     } else {
1256         i8259 = i8259_init(isa_bus, x86_allocate_cpu_irq());
1257     }
1258 
1259     for (size_t i = 0; i < ISA_NUM_IRQS; i++) {
1260         i8259_irqs[i] = i8259[i];
1261     }
1262 
1263     g_free(i8259);
1264 }
1265 
pc_memory_pre_plug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1266 static void pc_memory_pre_plug(HotplugHandler *hotplug_dev, DeviceState *dev,
1267                                Error **errp)
1268 {
1269     const X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1270     const MachineState *ms = MACHINE(hotplug_dev);
1271     const bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1272     Error *local_err = NULL;
1273 
1274     /*
1275      * When "acpi=off" is used with the Q35 machine type, no ACPI is built,
1276      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1277      * addition to cover this case.
1278      */
1279     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1280         error_setg(errp,
1281                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1282         return;
1283     }
1284 
1285     if (is_nvdimm && !ms->nvdimms_state->is_enabled) {
1286         error_setg(errp, "nvdimm is not enabled: missing 'nvdimm' in '-M'");
1287         return;
1288     }
1289 
1290     hotplug_handler_pre_plug(x86ms->acpi_dev, dev, &local_err);
1291     if (local_err) {
1292         error_propagate(errp, local_err);
1293         return;
1294     }
1295 
1296     pc_dimm_pre_plug(PC_DIMM(dev), MACHINE(hotplug_dev), errp);
1297 }
1298 
pc_memory_plug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1299 static void pc_memory_plug(HotplugHandler *hotplug_dev,
1300                            DeviceState *dev, Error **errp)
1301 {
1302     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1303     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1304     MachineState *ms = MACHINE(hotplug_dev);
1305     bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1306 
1307     pc_dimm_plug(PC_DIMM(dev), MACHINE(pcms));
1308 
1309     if (is_nvdimm) {
1310         nvdimm_plug(ms->nvdimms_state);
1311     }
1312 
1313     hotplug_handler_plug(x86ms->acpi_dev, dev, &error_abort);
1314 }
1315 
pc_memory_unplug_request(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1316 static void pc_memory_unplug_request(HotplugHandler *hotplug_dev,
1317                                      DeviceState *dev, Error **errp)
1318 {
1319     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1320 
1321     /*
1322      * When "acpi=off" is used with the Q35 machine type, no ACPI is built,
1323      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1324      * addition to cover this case.
1325      */
1326     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1327         error_setg(errp,
1328                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1329         return;
1330     }
1331 
1332     if (object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM)) {
1333         error_setg(errp, "nvdimm device hot unplug is not supported yet.");
1334         return;
1335     }
1336 
1337     hotplug_handler_unplug_request(x86ms->acpi_dev, dev,
1338                                    errp);
1339 }
1340 
pc_memory_unplug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1341 static void pc_memory_unplug(HotplugHandler *hotplug_dev,
1342                              DeviceState *dev, Error **errp)
1343 {
1344     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1345     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1346     Error *local_err = NULL;
1347 
1348     hotplug_handler_unplug(x86ms->acpi_dev, dev, &local_err);
1349     if (local_err) {
1350         goto out;
1351     }
1352 
1353     pc_dimm_unplug(PC_DIMM(dev), MACHINE(pcms));
1354     qdev_unrealize(dev);
1355  out:
1356     error_propagate(errp, local_err);
1357 }
1358 
pc_hv_balloon_pre_plug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1359 static void pc_hv_balloon_pre_plug(HotplugHandler *hotplug_dev,
1360                                    DeviceState *dev, Error **errp)
1361 {
1362     /* The vmbus handler has no hotplug handler; we should never end up here. */
1363     g_assert(!dev->hotplugged);
1364     memory_device_pre_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev), errp);
1365 }
1366 
pc_hv_balloon_plug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1367 static void pc_hv_balloon_plug(HotplugHandler *hotplug_dev,
1368                                DeviceState *dev, Error **errp)
1369 {
1370     memory_device_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev));
1371 }
1372 
pc_machine_device_pre_plug_cb(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1373 static void pc_machine_device_pre_plug_cb(HotplugHandler *hotplug_dev,
1374                                           DeviceState *dev, Error **errp)
1375 {
1376     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1377         pc_memory_pre_plug(hotplug_dev, dev, errp);
1378     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1379         x86_cpu_pre_plug(hotplug_dev, dev, errp);
1380     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1381         virtio_md_pci_pre_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1382     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1383         /* Declare the APIC range as the reserved MSI region */
1384         char *resv_prop_str = g_strdup_printf("0xfee00000:0xfeefffff:%d",
1385                                               VIRTIO_IOMMU_RESV_MEM_T_MSI);
1386         QList *reserved_regions = qlist_new();
1387 
1388         qlist_append_str(reserved_regions, resv_prop_str);
1389         qdev_prop_set_array(dev, "reserved-regions", reserved_regions);
1390 
1391         g_free(resv_prop_str);
1392     }
1393 
1394     if (object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE) ||
1395         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1396         PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1397 
1398         if (pcms->iommu) {
1399             error_setg(errp, "QEMU does not support multiple vIOMMUs "
1400                        "for x86 yet.");
1401             return;
1402         }
1403         pcms->iommu = dev;
1404     } else if (object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON)) {
1405         pc_hv_balloon_pre_plug(hotplug_dev, dev, errp);
1406     }
1407 }
1408 
pc_machine_device_plug_cb(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1409 static void pc_machine_device_plug_cb(HotplugHandler *hotplug_dev,
1410                                       DeviceState *dev, Error **errp)
1411 {
1412     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1413         pc_memory_plug(hotplug_dev, dev, errp);
1414     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1415         x86_cpu_plug(hotplug_dev, dev, errp);
1416     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1417         virtio_md_pci_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1418     } else if (object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON)) {
1419         pc_hv_balloon_plug(hotplug_dev, dev, errp);
1420     }
1421 }
1422 
pc_machine_device_unplug_request_cb(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1423 static void pc_machine_device_unplug_request_cb(HotplugHandler *hotplug_dev,
1424                                                 DeviceState *dev, Error **errp)
1425 {
1426     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1427         pc_memory_unplug_request(hotplug_dev, dev, errp);
1428     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1429         x86_cpu_unplug_request_cb(hotplug_dev, dev, errp);
1430     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1431         virtio_md_pci_unplug_request(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev),
1432                                      errp);
1433     } else {
1434         error_setg(errp, "acpi: device unplug request for not supported device"
1435                    " type: %s", object_get_typename(OBJECT(dev)));
1436     }
1437 }
1438 
pc_machine_device_unplug_cb(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1439 static void pc_machine_device_unplug_cb(HotplugHandler *hotplug_dev,
1440                                         DeviceState *dev, Error **errp)
1441 {
1442     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1443         pc_memory_unplug(hotplug_dev, dev, errp);
1444     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1445         x86_cpu_unplug_cb(hotplug_dev, dev, errp);
1446     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1447         virtio_md_pci_unplug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1448     } else {
1449         error_setg(errp, "acpi: device unplug for not supported device"
1450                    " type: %s", object_get_typename(OBJECT(dev)));
1451     }
1452 }
1453 
pc_get_hotplug_handler(MachineState * machine,DeviceState * dev)1454 static HotplugHandler *pc_get_hotplug_handler(MachineState *machine,
1455                                              DeviceState *dev)
1456 {
1457     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM) ||
1458         object_dynamic_cast(OBJECT(dev), TYPE_CPU) ||
1459         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI) ||
1460         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI) ||
1461         object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON) ||
1462         object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE)) {
1463         return HOTPLUG_HANDLER(machine);
1464     }
1465 
1466     return NULL;
1467 }
1468 
pc_machine_get_vmport(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1469 static void pc_machine_get_vmport(Object *obj, Visitor *v, const char *name,
1470                                   void *opaque, Error **errp)
1471 {
1472     PCMachineState *pcms = PC_MACHINE(obj);
1473     OnOffAuto vmport = pcms->vmport;
1474 
1475     visit_type_OnOffAuto(v, name, &vmport, errp);
1476 }
1477 
pc_machine_set_vmport(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1478 static void pc_machine_set_vmport(Object *obj, Visitor *v, const char *name,
1479                                   void *opaque, Error **errp)
1480 {
1481     PCMachineState *pcms = PC_MACHINE(obj);
1482 
1483     visit_type_OnOffAuto(v, name, &pcms->vmport, errp);
1484 }
1485 
pc_machine_get_fd_bootchk(Object * obj,Error ** errp)1486 static bool pc_machine_get_fd_bootchk(Object *obj, Error **errp)
1487 {
1488     PCMachineState *pcms = PC_MACHINE(obj);
1489 
1490     return pcms->fd_bootchk;
1491 }
1492 
pc_machine_set_fd_bootchk(Object * obj,bool value,Error ** errp)1493 static void pc_machine_set_fd_bootchk(Object *obj, bool value, Error **errp)
1494 {
1495     PCMachineState *pcms = PC_MACHINE(obj);
1496 
1497     pcms->fd_bootchk = value;
1498 }
1499 
pc_machine_get_smbus(Object * obj,Error ** errp)1500 static bool pc_machine_get_smbus(Object *obj, Error **errp)
1501 {
1502     PCMachineState *pcms = PC_MACHINE(obj);
1503 
1504     return pcms->smbus_enabled;
1505 }
1506 
pc_machine_set_smbus(Object * obj,bool value,Error ** errp)1507 static void pc_machine_set_smbus(Object *obj, bool value, Error **errp)
1508 {
1509     PCMachineState *pcms = PC_MACHINE(obj);
1510 
1511     pcms->smbus_enabled = value;
1512 }
1513 
pc_machine_get_sata(Object * obj,Error ** errp)1514 static bool pc_machine_get_sata(Object *obj, Error **errp)
1515 {
1516     PCMachineState *pcms = PC_MACHINE(obj);
1517 
1518     return pcms->sata_enabled;
1519 }
1520 
pc_machine_set_sata(Object * obj,bool value,Error ** errp)1521 static void pc_machine_set_sata(Object *obj, bool value, Error **errp)
1522 {
1523     PCMachineState *pcms = PC_MACHINE(obj);
1524 
1525     pcms->sata_enabled = value;
1526 }
1527 
pc_machine_get_hpet(Object * obj,Error ** errp)1528 static bool pc_machine_get_hpet(Object *obj, Error **errp)
1529 {
1530     PCMachineState *pcms = PC_MACHINE(obj);
1531 
1532     return pcms->hpet_enabled;
1533 }
1534 
pc_machine_set_hpet(Object * obj,bool value,Error ** errp)1535 static void pc_machine_set_hpet(Object *obj, bool value, Error **errp)
1536 {
1537     PCMachineState *pcms = PC_MACHINE(obj);
1538 
1539     pcms->hpet_enabled = value;
1540 }
1541 
pc_machine_get_i8042(Object * obj,Error ** errp)1542 static bool pc_machine_get_i8042(Object *obj, Error **errp)
1543 {
1544     PCMachineState *pcms = PC_MACHINE(obj);
1545 
1546     return pcms->i8042_enabled;
1547 }
1548 
pc_machine_set_i8042(Object * obj,bool value,Error ** errp)1549 static void pc_machine_set_i8042(Object *obj, bool value, Error **errp)
1550 {
1551     PCMachineState *pcms = PC_MACHINE(obj);
1552 
1553     pcms->i8042_enabled = value;
1554 }
1555 
pc_machine_get_default_bus_bypass_iommu(Object * obj,Error ** errp)1556 static bool pc_machine_get_default_bus_bypass_iommu(Object *obj, Error **errp)
1557 {
1558     PCMachineState *pcms = PC_MACHINE(obj);
1559 
1560     return pcms->default_bus_bypass_iommu;
1561 }
1562 
pc_machine_set_default_bus_bypass_iommu(Object * obj,bool value,Error ** errp)1563 static void pc_machine_set_default_bus_bypass_iommu(Object *obj, bool value,
1564                                                     Error **errp)
1565 {
1566     PCMachineState *pcms = PC_MACHINE(obj);
1567 
1568     pcms->default_bus_bypass_iommu = value;
1569 }
1570 
pc_machine_get_smbios_ep(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1571 static void pc_machine_get_smbios_ep(Object *obj, Visitor *v, const char *name,
1572                                      void *opaque, Error **errp)
1573 {
1574     PCMachineState *pcms = PC_MACHINE(obj);
1575     SmbiosEntryPointType smbios_entry_point_type = pcms->smbios_entry_point_type;
1576 
1577     visit_type_SmbiosEntryPointType(v, name, &smbios_entry_point_type, errp);
1578 }
1579 
pc_machine_set_smbios_ep(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1580 static void pc_machine_set_smbios_ep(Object *obj, Visitor *v, const char *name,
1581                                      void *opaque, Error **errp)
1582 {
1583     PCMachineState *pcms = PC_MACHINE(obj);
1584 
1585     visit_type_SmbiosEntryPointType(v, name, &pcms->smbios_entry_point_type, errp);
1586 }
1587 
pc_machine_get_max_ram_below_4g(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1588 static void pc_machine_get_max_ram_below_4g(Object *obj, Visitor *v,
1589                                             const char *name, void *opaque,
1590                                             Error **errp)
1591 {
1592     PCMachineState *pcms = PC_MACHINE(obj);
1593     uint64_t value = pcms->max_ram_below_4g;
1594 
1595     visit_type_size(v, name, &value, errp);
1596 }
1597 
pc_machine_set_max_ram_below_4g(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1598 static void pc_machine_set_max_ram_below_4g(Object *obj, Visitor *v,
1599                                             const char *name, void *opaque,
1600                                             Error **errp)
1601 {
1602     PCMachineState *pcms = PC_MACHINE(obj);
1603     uint64_t value;
1604 
1605     if (!visit_type_size(v, name, &value, errp)) {
1606         return;
1607     }
1608     if (value > 4 * GiB) {
1609         error_setg(errp,
1610                    "Machine option 'max-ram-below-4g=%"PRIu64
1611                    "' expects size less than or equal to 4G", value);
1612         return;
1613     }
1614 
1615     if (value < 1 * MiB) {
1616         warn_report("Only %" PRIu64 " bytes of RAM below the 4GiB boundary,"
1617                     "BIOS may not work with less than 1MiB", value);
1618     }
1619 
1620     pcms->max_ram_below_4g = value;
1621 }
1622 
pc_machine_get_max_fw_size(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1623 static void pc_machine_get_max_fw_size(Object *obj, Visitor *v,
1624                                        const char *name, void *opaque,
1625                                        Error **errp)
1626 {
1627     PCMachineState *pcms = PC_MACHINE(obj);
1628     uint64_t value = pcms->max_fw_size;
1629 
1630     visit_type_size(v, name, &value, errp);
1631 }
1632 
pc_machine_set_max_fw_size(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1633 static void pc_machine_set_max_fw_size(Object *obj, Visitor *v,
1634                                        const char *name, void *opaque,
1635                                        Error **errp)
1636 {
1637     PCMachineState *pcms = PC_MACHINE(obj);
1638     uint64_t value;
1639 
1640     if (!visit_type_size(v, name, &value, errp)) {
1641         return;
1642     }
1643 
1644     /*
1645      * We don't have a theoretically justifiable exact lower bound on the base
1646      * address of any flash mapping. In practice, the IO-APIC MMIO range is
1647      * [0xFEE00000..0xFEE01000] -- see IO_APIC_DEFAULT_ADDRESS --, leaving free
1648      * only 18MiB-4KiB below 4GiB. For now, restrict the cumulative mapping to
1649      * 16MiB in size.
1650      */
1651     if (value > 16 * MiB) {
1652         error_setg(errp,
1653                    "User specified max allowed firmware size %" PRIu64 " is "
1654                    "greater than 16MiB. If combined firmware size exceeds "
1655                    "16MiB the system may not boot, or experience intermittent"
1656                    "stability issues.",
1657                    value);
1658         return;
1659     }
1660 
1661     pcms->max_fw_size = value;
1662 }
1663 
1664 
pc_machine_initfn(Object * obj)1665 static void pc_machine_initfn(Object *obj)
1666 {
1667     PCMachineState *pcms = PC_MACHINE(obj);
1668     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1669 
1670 #ifdef CONFIG_VMPORT
1671     pcms->vmport = ON_OFF_AUTO_AUTO;
1672 #else
1673     pcms->vmport = ON_OFF_AUTO_OFF;
1674 #endif /* CONFIG_VMPORT */
1675     pcms->max_ram_below_4g = 0; /* use default */
1676     pcms->smbios_entry_point_type = pcmc->default_smbios_ep_type;
1677     pcms->south_bridge = pcmc->default_south_bridge;
1678 
1679     /* acpi build is enabled by default if machine supports it */
1680     pcms->acpi_build_enabled = pcmc->has_acpi_build;
1681     pcms->smbus_enabled = true;
1682     pcms->sata_enabled = true;
1683     pcms->i8042_enabled = true;
1684     pcms->max_fw_size = 8 * MiB;
1685 #if defined(CONFIG_HPET)
1686     pcms->hpet_enabled = true;
1687 #endif
1688     pcms->fd_bootchk = true;
1689     pcms->default_bus_bypass_iommu = false;
1690 
1691     pc_system_flash_create(pcms);
1692     pcms->pcspk = isa_new(TYPE_PC_SPEAKER);
1693     object_property_add_alias(OBJECT(pcms), "pcspk-audiodev",
1694                               OBJECT(pcms->pcspk), "audiodev");
1695     if (pcmc->pci_enabled) {
1696         cxl_machine_init(obj, &pcms->cxl_devices_state);
1697     }
1698 }
1699 
pc_machine_reset(MachineState * machine,ResetType type)1700 static void pc_machine_reset(MachineState *machine, ResetType type)
1701 {
1702     CPUState *cs;
1703     X86CPU *cpu;
1704 
1705     qemu_devices_reset(type);
1706 
1707     /* Reset APIC after devices have been reset to cancel
1708      * any changes that qemu_devices_reset() might have done.
1709      */
1710     CPU_FOREACH(cs) {
1711         cpu = X86_CPU(cs);
1712 
1713         x86_cpu_after_reset(cpu);
1714     }
1715 }
1716 
pc_machine_wakeup(MachineState * machine)1717 static void pc_machine_wakeup(MachineState *machine)
1718 {
1719     cpu_synchronize_all_states();
1720     pc_machine_reset(machine, RESET_TYPE_WAKEUP);
1721     cpu_synchronize_all_post_reset();
1722 }
1723 
pc_hotplug_allowed(MachineState * ms,DeviceState * dev,Error ** errp)1724 static bool pc_hotplug_allowed(MachineState *ms, DeviceState *dev, Error **errp)
1725 {
1726     X86IOMMUState *iommu = x86_iommu_get_default();
1727     IntelIOMMUState *intel_iommu;
1728 
1729     if (iommu &&
1730         object_dynamic_cast((Object *)iommu, TYPE_INTEL_IOMMU_DEVICE) &&
1731         object_dynamic_cast((Object *)dev, "vfio-pci")) {
1732         intel_iommu = INTEL_IOMMU_DEVICE(iommu);
1733         if (!intel_iommu->caching_mode) {
1734             error_setg(errp, "Device assignment is not allowed without "
1735                        "enabling caching-mode=on for Intel IOMMU.");
1736             return false;
1737         }
1738     }
1739 
1740     return true;
1741 }
1742 
pc_machine_class_init(ObjectClass * oc,const void * data)1743 static void pc_machine_class_init(ObjectClass *oc, const void *data)
1744 {
1745     MachineClass *mc = MACHINE_CLASS(oc);
1746     X86MachineClass *x86mc = X86_MACHINE_CLASS(oc);
1747     PCMachineClass *pcmc = PC_MACHINE_CLASS(oc);
1748     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
1749 
1750     pcmc->pci_enabled = true;
1751     pcmc->has_acpi_build = true;
1752     pcmc->smbios_defaults = true;
1753     pcmc->gigabyte_align = true;
1754     pcmc->has_reserved_memory = true;
1755     pcmc->enforce_amd_1tb_hole = true;
1756     pcmc->isa_bios_alias = true;
1757     pcmc->pvh_enabled = true;
1758     pcmc->kvmclock_create_always = true;
1759     x86mc->apic_xrupt_override = true;
1760     assert(!mc->get_hotplug_handler);
1761     mc->get_hotplug_handler = pc_get_hotplug_handler;
1762     mc->hotplug_allowed = pc_hotplug_allowed;
1763     mc->auto_enable_numa_with_memhp = true;
1764     mc->auto_enable_numa_with_memdev = true;
1765     mc->has_hotpluggable_cpus = true;
1766     mc->default_boot_order = "cad";
1767     mc->block_default_type = IF_IDE;
1768     mc->max_cpus = 255;
1769     mc->reset = pc_machine_reset;
1770     mc->wakeup = pc_machine_wakeup;
1771     hc->pre_plug = pc_machine_device_pre_plug_cb;
1772     hc->plug = pc_machine_device_plug_cb;
1773     hc->unplug_request = pc_machine_device_unplug_request_cb;
1774     hc->unplug = pc_machine_device_unplug_cb;
1775     mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE;
1776     mc->nvdimm_supported = true;
1777     mc->smp_props.dies_supported = true;
1778     mc->smp_props.modules_supported = true;
1779     mc->smp_props.cache_supported[CACHE_LEVEL_AND_TYPE_L1D] = true;
1780     mc->smp_props.cache_supported[CACHE_LEVEL_AND_TYPE_L1I] = true;
1781     mc->smp_props.cache_supported[CACHE_LEVEL_AND_TYPE_L2] = true;
1782     mc->smp_props.cache_supported[CACHE_LEVEL_AND_TYPE_L3] = true;
1783     mc->default_ram_id = "pc.ram";
1784     pcmc->default_smbios_ep_type = SMBIOS_ENTRY_POINT_TYPE_AUTO;
1785 
1786     object_class_property_add(oc, PC_MACHINE_MAX_RAM_BELOW_4G, "size",
1787         pc_machine_get_max_ram_below_4g, pc_machine_set_max_ram_below_4g,
1788         NULL, NULL);
1789     object_class_property_set_description(oc, PC_MACHINE_MAX_RAM_BELOW_4G,
1790         "Maximum ram below the 4G boundary (32bit boundary)");
1791 
1792     object_class_property_add(oc, PC_MACHINE_VMPORT, "OnOffAuto",
1793         pc_machine_get_vmport, pc_machine_set_vmport,
1794         NULL, NULL);
1795     object_class_property_set_description(oc, PC_MACHINE_VMPORT,
1796         "Enable vmport (pc & q35)");
1797 
1798     object_class_property_add_bool(oc, PC_MACHINE_SMBUS,
1799         pc_machine_get_smbus, pc_machine_set_smbus);
1800     object_class_property_set_description(oc, PC_MACHINE_SMBUS,
1801         "Enable/disable system management bus");
1802 
1803     object_class_property_add_bool(oc, PC_MACHINE_SATA,
1804         pc_machine_get_sata, pc_machine_set_sata);
1805     object_class_property_set_description(oc, PC_MACHINE_SATA,
1806         "Enable/disable Serial ATA bus");
1807 
1808     object_class_property_add_bool(oc, "hpet",
1809         pc_machine_get_hpet, pc_machine_set_hpet);
1810     object_class_property_set_description(oc, "hpet",
1811         "Enable/disable high precision event timer emulation");
1812 
1813     object_class_property_add_bool(oc, PC_MACHINE_I8042,
1814         pc_machine_get_i8042, pc_machine_set_i8042);
1815     object_class_property_set_description(oc, PC_MACHINE_I8042,
1816         "Enable/disable Intel 8042 PS/2 controller emulation");
1817 
1818     object_class_property_add_bool(oc, "default-bus-bypass-iommu",
1819         pc_machine_get_default_bus_bypass_iommu,
1820         pc_machine_set_default_bus_bypass_iommu);
1821 
1822     object_class_property_add(oc, PC_MACHINE_MAX_FW_SIZE, "size",
1823         pc_machine_get_max_fw_size, pc_machine_set_max_fw_size,
1824         NULL, NULL);
1825     object_class_property_set_description(oc, PC_MACHINE_MAX_FW_SIZE,
1826         "Maximum combined firmware size");
1827 
1828     object_class_property_add(oc, PC_MACHINE_SMBIOS_EP, "str",
1829         pc_machine_get_smbios_ep, pc_machine_set_smbios_ep,
1830         NULL, NULL);
1831     object_class_property_set_description(oc, PC_MACHINE_SMBIOS_EP,
1832         "SMBIOS Entry Point type [32, 64]");
1833 
1834     object_class_property_add_bool(oc, "fd-bootchk",
1835         pc_machine_get_fd_bootchk,
1836         pc_machine_set_fd_bootchk);
1837 
1838 #if defined(CONFIG_IGVM)
1839     object_class_property_add_link(oc, "igvm-cfg",
1840                                    TYPE_IGVM_CFG,
1841                                    offsetof(X86MachineState, igvm),
1842                                    object_property_allow_set_link,
1843                                    OBJ_PROP_LINK_STRONG);
1844     object_class_property_set_description(oc, "igvm-cfg",
1845                                           "Set IGVM configuration");
1846 #endif
1847 
1848 
1849 }
1850 
1851 static const TypeInfo pc_machine_info = {
1852     .name = TYPE_PC_MACHINE,
1853     .parent = TYPE_X86_MACHINE,
1854     .abstract = true,
1855     .instance_size = sizeof(PCMachineState),
1856     .instance_init = pc_machine_initfn,
1857     .class_size = sizeof(PCMachineClass),
1858     .class_init = pc_machine_class_init,
1859     .interfaces = (const InterfaceInfo[]) {
1860          { TYPE_HOTPLUG_HANDLER },
1861          { }
1862     },
1863 };
1864 
pc_machine_register_types(void)1865 static void pc_machine_register_types(void)
1866 {
1867     type_register_static(&pc_machine_info);
1868 }
1869 
1870 type_init(pc_machine_register_types)
1871