xref: /openbmc/qemu/hw/i386/pc.c (revision 702cbdc4)
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 "sev.h"
66 #include CONFIG_DEVICES
67 
68 #ifdef CONFIG_XEN_EMU
69 #include "hw/xen/xen-legacy-backend.h"
70 #include "hw/xen/xen-bus.h"
71 #endif
72 
73 /*
74  * Helper for setting model-id for CPU models that changed model-id
75  * depending on QEMU versions up to QEMU 2.4.
76  */
77 #define PC_CPU_MODEL_IDS(v) \
78     { "qemu32-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
79     { "qemu64-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },\
80     { "athlon-" TYPE_X86_CPU, "model-id", "QEMU Virtual CPU version " v, },
81 
82 GlobalProperty pc_compat_9_0[] = {
83     { TYPE_X86_CPU, "x-amd-topoext-features-only", "false" },
84     { TYPE_X86_CPU, "x-l1-cache-per-thread", "false" },
85     { TYPE_X86_CPU, "guest-phys-bits", "0" },
86     { "sev-guest", "legacy-vm-type", "on" },
87     { TYPE_X86_CPU, "legacy-multi-node", "on" },
88 };
89 const size_t pc_compat_9_0_len = G_N_ELEMENTS(pc_compat_9_0);
90 
91 GlobalProperty pc_compat_8_2[] = {};
92 const size_t pc_compat_8_2_len = G_N_ELEMENTS(pc_compat_8_2);
93 
94 GlobalProperty pc_compat_8_1[] = {};
95 const size_t pc_compat_8_1_len = G_N_ELEMENTS(pc_compat_8_1);
96 
97 GlobalProperty pc_compat_8_0[] = {
98     { "virtio-mem", "unplugged-inaccessible", "auto" },
99 };
100 const size_t pc_compat_8_0_len = G_N_ELEMENTS(pc_compat_8_0);
101 
102 GlobalProperty pc_compat_7_2[] = {
103     { "ICH9-LPC", "noreboot", "true" },
104 };
105 const size_t pc_compat_7_2_len = G_N_ELEMENTS(pc_compat_7_2);
106 
107 GlobalProperty pc_compat_7_1[] = {};
108 const size_t pc_compat_7_1_len = G_N_ELEMENTS(pc_compat_7_1);
109 
110 GlobalProperty pc_compat_7_0[] = {};
111 const size_t pc_compat_7_0_len = G_N_ELEMENTS(pc_compat_7_0);
112 
113 GlobalProperty pc_compat_6_2[] = {
114     { "virtio-mem", "unplugged-inaccessible", "off" },
115 };
116 const size_t pc_compat_6_2_len = G_N_ELEMENTS(pc_compat_6_2);
117 
118 GlobalProperty pc_compat_6_1[] = {
119     { TYPE_X86_CPU, "hv-version-id-build", "0x1bbc" },
120     { TYPE_X86_CPU, "hv-version-id-major", "0x0006" },
121     { TYPE_X86_CPU, "hv-version-id-minor", "0x0001" },
122     { "ICH9-LPC", "x-keep-pci-slot-hpc", "false" },
123 };
124 const size_t pc_compat_6_1_len = G_N_ELEMENTS(pc_compat_6_1);
125 
126 GlobalProperty pc_compat_6_0[] = {
127     { "qemu64" "-" TYPE_X86_CPU, "family", "6" },
128     { "qemu64" "-" TYPE_X86_CPU, "model", "6" },
129     { "qemu64" "-" TYPE_X86_CPU, "stepping", "3" },
130     { TYPE_X86_CPU, "x-vendor-cpuid-only", "off" },
131     { "ICH9-LPC", ACPI_PM_PROP_ACPI_PCIHP_BRIDGE, "off" },
132     { "ICH9-LPC", "x-keep-pci-slot-hpc", "true" },
133 };
134 const size_t pc_compat_6_0_len = G_N_ELEMENTS(pc_compat_6_0);
135 
136 GlobalProperty pc_compat_5_2[] = {
137     { "ICH9-LPC", "x-smi-cpu-hotunplug", "off" },
138 };
139 const size_t pc_compat_5_2_len = G_N_ELEMENTS(pc_compat_5_2);
140 
141 GlobalProperty pc_compat_5_1[] = {
142     { "ICH9-LPC", "x-smi-cpu-hotplug", "off" },
143     { TYPE_X86_CPU, "kvm-msi-ext-dest-id", "off" },
144 };
145 const size_t pc_compat_5_1_len = G_N_ELEMENTS(pc_compat_5_1);
146 
147 GlobalProperty pc_compat_5_0[] = {
148 };
149 const size_t pc_compat_5_0_len = G_N_ELEMENTS(pc_compat_5_0);
150 
151 GlobalProperty pc_compat_4_2[] = {
152     { "mch", "smbase-smram", "off" },
153 };
154 const size_t pc_compat_4_2_len = G_N_ELEMENTS(pc_compat_4_2);
155 
156 GlobalProperty pc_compat_4_1[] = {};
157 const size_t pc_compat_4_1_len = G_N_ELEMENTS(pc_compat_4_1);
158 
159 GlobalProperty pc_compat_4_0[] = {};
160 const size_t pc_compat_4_0_len = G_N_ELEMENTS(pc_compat_4_0);
161 
162 GlobalProperty pc_compat_3_1[] = {
163     { "intel-iommu", "dma-drain", "off" },
164     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "off" },
165     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "off" },
166     { "Opteron_G4" "-" TYPE_X86_CPU, "npt", "off" },
167     { "Opteron_G4" "-" TYPE_X86_CPU, "nrip-save", "off" },
168     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "off" },
169     { "Opteron_G5" "-" TYPE_X86_CPU, "npt", "off" },
170     { "Opteron_G5" "-" TYPE_X86_CPU, "nrip-save", "off" },
171     { "EPYC" "-" TYPE_X86_CPU, "npt", "off" },
172     { "EPYC" "-" TYPE_X86_CPU, "nrip-save", "off" },
173     { "EPYC-IBPB" "-" TYPE_X86_CPU, "npt", "off" },
174     { "EPYC-IBPB" "-" TYPE_X86_CPU, "nrip-save", "off" },
175     { "Skylake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
176     { "Skylake-Client-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
177     { "Skylake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
178     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
179     { "Cascadelake-Server" "-" TYPE_X86_CPU,  "mpx", "on" },
180     { "Icelake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
181     { "Icelake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
182     { "Cascadelake-Server" "-" TYPE_X86_CPU, "stepping", "5" },
183     { TYPE_X86_CPU, "x-intel-pt-auto-level", "off" },
184 };
185 const size_t pc_compat_3_1_len = G_N_ELEMENTS(pc_compat_3_1);
186 
187 GlobalProperty pc_compat_3_0[] = {
188     { TYPE_X86_CPU, "x-hv-synic-kvm-only", "on" },
189     { "Skylake-Server" "-" TYPE_X86_CPU, "pku", "off" },
190     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "pku", "off" },
191 };
192 const size_t pc_compat_3_0_len = G_N_ELEMENTS(pc_compat_3_0);
193 
194 GlobalProperty pc_compat_2_12[] = {
195     { TYPE_X86_CPU, "legacy-cache", "on" },
196     { TYPE_X86_CPU, "topoext", "off" },
197     { "EPYC-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
198     { "EPYC-IBPB-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
199 };
200 const size_t pc_compat_2_12_len = G_N_ELEMENTS(pc_compat_2_12);
201 
202 GlobalProperty pc_compat_2_11[] = {
203     { TYPE_X86_CPU, "x-migrate-smi-count", "off" },
204     { "Skylake-Server" "-" TYPE_X86_CPU, "clflushopt", "off" },
205 };
206 const size_t pc_compat_2_11_len = G_N_ELEMENTS(pc_compat_2_11);
207 
208 GlobalProperty pc_compat_2_10[] = {
209     { TYPE_X86_CPU, "x-hv-max-vps", "0x40" },
210     { "i440FX-pcihost", "x-pci-hole64-fix", "off" },
211     { "q35-pcihost", "x-pci-hole64-fix", "off" },
212 };
213 const size_t pc_compat_2_10_len = G_N_ELEMENTS(pc_compat_2_10);
214 
215 GlobalProperty pc_compat_2_9[] = {
216     { "mch", "extended-tseg-mbytes", "0" },
217 };
218 const size_t pc_compat_2_9_len = G_N_ELEMENTS(pc_compat_2_9);
219 
220 GlobalProperty pc_compat_2_8[] = {
221     { TYPE_X86_CPU, "tcg-cpuid", "off" },
222     { "kvmclock", "x-mach-use-reliable-get-clock", "off" },
223     { "ICH9-LPC", "x-smi-broadcast", "off" },
224     { TYPE_X86_CPU, "vmware-cpuid-freq", "off" },
225     { "Haswell-" TYPE_X86_CPU, "stepping", "1" },
226 };
227 const size_t pc_compat_2_8_len = G_N_ELEMENTS(pc_compat_2_8);
228 
229 GlobalProperty pc_compat_2_7[] = {
230     { TYPE_X86_CPU, "l3-cache", "off" },
231     { TYPE_X86_CPU, "full-cpuid-auto-level", "off" },
232     { "Opteron_G3" "-" TYPE_X86_CPU, "family", "15" },
233     { "Opteron_G3" "-" TYPE_X86_CPU, "model", "6" },
234     { "Opteron_G3" "-" TYPE_X86_CPU, "stepping", "1" },
235     { "isa-pcspk", "migrate", "off" },
236 };
237 const size_t pc_compat_2_7_len = G_N_ELEMENTS(pc_compat_2_7);
238 
239 GlobalProperty pc_compat_2_6[] = {
240     { TYPE_X86_CPU, "cpuid-0xb", "off" },
241     { "vmxnet3", "romfile", "" },
242     { TYPE_X86_CPU, "fill-mtrr-mask", "off" },
243     { "apic-common", "legacy-instance-id", "on", }
244 };
245 const size_t pc_compat_2_6_len = G_N_ELEMENTS(pc_compat_2_6);
246 
247 GlobalProperty pc_compat_2_5[] = {};
248 const size_t pc_compat_2_5_len = G_N_ELEMENTS(pc_compat_2_5);
249 
250 GlobalProperty pc_compat_2_4[] = {
251     PC_CPU_MODEL_IDS("2.4.0")
252     { "Haswell-" TYPE_X86_CPU, "abm", "off" },
253     { "Haswell-noTSX-" TYPE_X86_CPU, "abm", "off" },
254     { "Broadwell-" TYPE_X86_CPU, "abm", "off" },
255     { "Broadwell-noTSX-" TYPE_X86_CPU, "abm", "off" },
256     { "host" "-" TYPE_X86_CPU, "host-cache-info", "on" },
257     { TYPE_X86_CPU, "check", "off" },
258     { "qemu64" "-" TYPE_X86_CPU, "sse4a", "on" },
259     { "qemu64" "-" TYPE_X86_CPU, "abm", "on" },
260     { "qemu64" "-" TYPE_X86_CPU, "popcnt", "on" },
261     { "qemu32" "-" TYPE_X86_CPU, "popcnt", "on" },
262     { "Opteron_G2" "-" TYPE_X86_CPU, "rdtscp", "on" },
263     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "on" },
264     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "on" },
265     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "on", }
266 };
267 const size_t pc_compat_2_4_len = G_N_ELEMENTS(pc_compat_2_4);
268 
269 /*
270  * @PC_FW_DATA:
271  * Size of the chunk of memory at the top of RAM for the BIOS ACPI tables
272  * and other BIOS datastructures.
273  *
274  * BIOS ACPI tables: 128K. Other BIOS datastructures: less than 4K
275  * reported to be used at the moment, 32K should be enough for a while.
276  */
277 #define PC_FW_DATA (0x20000 + 0x8000)
278 
pc_gsi_create(qemu_irq ** irqs,bool pci_enabled)279 GSIState *pc_gsi_create(qemu_irq **irqs, bool pci_enabled)
280 {
281     GSIState *s;
282 
283     s = g_new0(GSIState, 1);
284     if (kvm_ioapic_in_kernel()) {
285         kvm_pc_setup_irq_routing(pci_enabled);
286     }
287     *irqs = qemu_allocate_irqs(gsi_handler, s, IOAPIC_NUM_PINS);
288 
289     return s;
290 }
291 
ioport80_write(void * opaque,hwaddr addr,uint64_t data,unsigned size)292 static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
293                            unsigned size)
294 {
295 }
296 
ioport80_read(void * opaque,hwaddr addr,unsigned size)297 static uint64_t ioport80_read(void *opaque, hwaddr addr, unsigned size)
298 {
299     return 0xffffffffffffffffULL;
300 }
301 
302 /* MS-DOS compatibility mode FPU exception support */
ioportF0_write(void * opaque,hwaddr addr,uint64_t data,unsigned size)303 static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
304                            unsigned size)
305 {
306     if (tcg_enabled()) {
307         cpu_set_ignne();
308     }
309 }
310 
ioportF0_read(void * opaque,hwaddr addr,unsigned size)311 static uint64_t ioportF0_read(void *opaque, hwaddr addr, unsigned size)
312 {
313     return 0xffffffffffffffffULL;
314 }
315 
316 /* PC cmos mappings */
317 
318 #define REG_EQUIPMENT_BYTE          0x14
319 
cmos_init_hd(MC146818RtcState * s,int type_ofs,int info_ofs,int16_t cylinders,int8_t heads,int8_t sectors)320 static void cmos_init_hd(MC146818RtcState *s, int type_ofs, int info_ofs,
321                          int16_t cylinders, int8_t heads, int8_t sectors)
322 {
323     mc146818rtc_set_cmos_data(s, type_ofs, 47);
324     mc146818rtc_set_cmos_data(s, info_ofs, cylinders);
325     mc146818rtc_set_cmos_data(s, info_ofs + 1, cylinders >> 8);
326     mc146818rtc_set_cmos_data(s, info_ofs + 2, heads);
327     mc146818rtc_set_cmos_data(s, info_ofs + 3, 0xff);
328     mc146818rtc_set_cmos_data(s, info_ofs + 4, 0xff);
329     mc146818rtc_set_cmos_data(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
330     mc146818rtc_set_cmos_data(s, info_ofs + 6, cylinders);
331     mc146818rtc_set_cmos_data(s, info_ofs + 7, cylinders >> 8);
332     mc146818rtc_set_cmos_data(s, info_ofs + 8, sectors);
333 }
334 
335 /* convert boot_device letter to something recognizable by the bios */
boot_device2nibble(char boot_device)336 static int boot_device2nibble(char boot_device)
337 {
338     switch(boot_device) {
339     case 'a':
340     case 'b':
341         return 0x01; /* floppy boot */
342     case 'c':
343         return 0x02; /* hard drive boot */
344     case 'd':
345         return 0x03; /* CD-ROM boot */
346     case 'n':
347         return 0x04; /* Network boot */
348     }
349     return 0;
350 }
351 
set_boot_dev(PCMachineState * pcms,MC146818RtcState * s,const char * boot_device,Error ** errp)352 static void set_boot_dev(PCMachineState *pcms, MC146818RtcState *s,
353                          const char *boot_device, Error **errp)
354 {
355 #define PC_MAX_BOOT_DEVICES 3
356     int nbds, bds[3] = { 0, };
357     int i;
358 
359     nbds = strlen(boot_device);
360     if (nbds > PC_MAX_BOOT_DEVICES) {
361         error_setg(errp, "Too many boot devices for PC");
362         return;
363     }
364     for (i = 0; i < nbds; i++) {
365         bds[i] = boot_device2nibble(boot_device[i]);
366         if (bds[i] == 0) {
367             error_setg(errp, "Invalid boot device for PC: '%c'",
368                        boot_device[i]);
369             return;
370         }
371     }
372     mc146818rtc_set_cmos_data(s, 0x3d, (bds[1] << 4) | bds[0]);
373     mc146818rtc_set_cmos_data(s, 0x38, (bds[2] << 4) | !pcms->fd_bootchk);
374 }
375 
pc_boot_set(void * opaque,const char * boot_device,Error ** errp)376 static void pc_boot_set(void *opaque, const char *boot_device, Error **errp)
377 {
378     PCMachineState *pcms = opaque;
379     X86MachineState *x86ms = X86_MACHINE(pcms);
380 
381     set_boot_dev(pcms, MC146818_RTC(x86ms->rtc), boot_device, errp);
382 }
383 
pc_cmos_init_floppy(MC146818RtcState * rtc_state,ISADevice * floppy)384 static void pc_cmos_init_floppy(MC146818RtcState *rtc_state, ISADevice *floppy)
385 {
386     int val, nb;
387     FloppyDriveType fd_type[2] = { FLOPPY_DRIVE_TYPE_NONE,
388                                    FLOPPY_DRIVE_TYPE_NONE };
389 
390 #ifdef CONFIG_FDC_ISA
391     /* floppy type */
392     if (floppy) {
393         for (int i = 0; i < 2; i++) {
394             fd_type[i] = isa_fdc_get_drive_type(floppy, i);
395         }
396     }
397 #endif
398 
399     val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
400         cmos_get_fd_drive_type(fd_type[1]);
401     mc146818rtc_set_cmos_data(rtc_state, 0x10, val);
402 
403     val = mc146818rtc_get_cmos_data(rtc_state, REG_EQUIPMENT_BYTE);
404     nb = 0;
405     if (fd_type[0] != FLOPPY_DRIVE_TYPE_NONE) {
406         nb++;
407     }
408     if (fd_type[1] != FLOPPY_DRIVE_TYPE_NONE) {
409         nb++;
410     }
411     switch (nb) {
412     case 0:
413         break;
414     case 1:
415         val |= 0x01; /* 1 drive, ready for boot */
416         break;
417     case 2:
418         val |= 0x41; /* 2 drives, ready for boot */
419         break;
420     }
421     mc146818rtc_set_cmos_data(rtc_state, REG_EQUIPMENT_BYTE, val);
422 }
423 
424 typedef struct check_fdc_state {
425     ISADevice *floppy;
426     bool multiple;
427 } CheckFdcState;
428 
check_fdc(Object * obj,void * opaque)429 static int check_fdc(Object *obj, void *opaque)
430 {
431     CheckFdcState *state = opaque;
432     Object *fdc;
433     uint32_t iobase;
434     Error *local_err = NULL;
435 
436     fdc = object_dynamic_cast(obj, TYPE_ISA_FDC);
437     if (!fdc) {
438         return 0;
439     }
440 
441     iobase = object_property_get_uint(obj, "iobase", &local_err);
442     if (local_err || iobase != 0x3f0) {
443         error_free(local_err);
444         return 0;
445     }
446 
447     if (state->floppy) {
448         state->multiple = true;
449     } else {
450         state->floppy = ISA_DEVICE(obj);
451     }
452     return 0;
453 }
454 
455 static const char * const fdc_container_path[] = {
456     "/unattached", "/peripheral", "/peripheral-anon"
457 };
458 
459 /*
460  * Locate the FDC at IO address 0x3f0, in order to configure the CMOS registers
461  * and ACPI objects.
462  */
pc_find_fdc0(void)463 static ISADevice *pc_find_fdc0(void)
464 {
465     int i;
466     Object *container;
467     CheckFdcState state = { 0 };
468 
469     for (i = 0; i < ARRAY_SIZE(fdc_container_path); i++) {
470         container = container_get(qdev_get_machine(), fdc_container_path[i]);
471         object_child_foreach(container, check_fdc, &state);
472     }
473 
474     if (state.multiple) {
475         warn_report("multiple floppy disk controllers with "
476                     "iobase=0x3f0 have been found");
477         error_printf("the one being picked for CMOS setup might not reflect "
478                      "your intent");
479     }
480 
481     return state.floppy;
482 }
483 
pc_cmos_init_late(PCMachineState * pcms)484 static void pc_cmos_init_late(PCMachineState *pcms)
485 {
486     X86MachineState *x86ms = X86_MACHINE(pcms);
487     MC146818RtcState *s = MC146818_RTC(x86ms->rtc);
488     int16_t cylinders;
489     int8_t heads, sectors;
490     int val;
491     int i, trans;
492 
493     val = 0;
494     if (pcms->idebus[0] &&
495         ide_get_geometry(pcms->idebus[0], 0,
496                          &cylinders, &heads, &sectors) >= 0) {
497         cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
498         val |= 0xf0;
499     }
500     if (pcms->idebus[0] &&
501         ide_get_geometry(pcms->idebus[0], 1,
502                          &cylinders, &heads, &sectors) >= 0) {
503         cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
504         val |= 0x0f;
505     }
506     mc146818rtc_set_cmos_data(s, 0x12, val);
507 
508     val = 0;
509     for (i = 0; i < 4; i++) {
510         /* NOTE: ide_get_geometry() returns the physical
511            geometry.  It is always such that: 1 <= sects <= 63, 1
512            <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
513            geometry can be different if a translation is done. */
514         BusState *idebus = pcms->idebus[i / 2];
515         if (idebus &&
516             ide_get_geometry(idebus, i % 2,
517                              &cylinders, &heads, &sectors) >= 0) {
518             trans = ide_get_bios_chs_trans(idebus, i % 2) - 1;
519             assert((trans & ~3) == 0);
520             val |= trans << (i * 2);
521         }
522     }
523     mc146818rtc_set_cmos_data(s, 0x39, val);
524 
525     pc_cmos_init_floppy(s, pc_find_fdc0());
526 
527     /* various important CMOS locations needed by PC/Bochs bios */
528 
529     /* memory size */
530     /* base memory (first MiB) */
531     val = MIN(x86ms->below_4g_mem_size / KiB, 640);
532     mc146818rtc_set_cmos_data(s, 0x15, val);
533     mc146818rtc_set_cmos_data(s, 0x16, val >> 8);
534     /* extended memory (next 64MiB) */
535     if (x86ms->below_4g_mem_size > 1 * MiB) {
536         val = (x86ms->below_4g_mem_size - 1 * MiB) / KiB;
537     } else {
538         val = 0;
539     }
540     if (val > 65535)
541         val = 65535;
542     mc146818rtc_set_cmos_data(s, 0x17, val);
543     mc146818rtc_set_cmos_data(s, 0x18, val >> 8);
544     mc146818rtc_set_cmos_data(s, 0x30, val);
545     mc146818rtc_set_cmos_data(s, 0x31, val >> 8);
546     /* memory between 16MiB and 4GiB */
547     if (x86ms->below_4g_mem_size > 16 * MiB) {
548         val = (x86ms->below_4g_mem_size - 16 * MiB) / (64 * KiB);
549     } else {
550         val = 0;
551     }
552     if (val > 65535)
553         val = 65535;
554     mc146818rtc_set_cmos_data(s, 0x34, val);
555     mc146818rtc_set_cmos_data(s, 0x35, val >> 8);
556     /* memory above 4GiB */
557     val = x86ms->above_4g_mem_size / 65536;
558     mc146818rtc_set_cmos_data(s, 0x5b, val);
559     mc146818rtc_set_cmos_data(s, 0x5c, val >> 8);
560     mc146818rtc_set_cmos_data(s, 0x5d, val >> 16);
561 
562     val = 0;
563     val |= 0x02; /* FPU is there */
564     val |= 0x04; /* PS/2 mouse installed */
565     mc146818rtc_set_cmos_data(s, REG_EQUIPMENT_BYTE, val);
566 }
567 
handle_a20_line_change(void * opaque,int irq,int level)568 static void handle_a20_line_change(void *opaque, int irq, int level)
569 {
570     X86CPU *cpu = opaque;
571 
572     /* XXX: send to all CPUs ? */
573     /* XXX: add logic to handle multiple A20 line sources */
574     x86_cpu_set_a20(cpu, level);
575 }
576 
577 #define NE2000_NB_MAX 6
578 
579 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
580                                               0x280, 0x380 };
581 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
582 
pc_init_ne2k_isa(ISABus * bus,NICInfo * nd,Error ** errp)583 static gboolean pc_init_ne2k_isa(ISABus *bus, NICInfo *nd, Error **errp)
584 {
585     static int nb_ne2k = 0;
586 
587     if (nb_ne2k == NE2000_NB_MAX) {
588         error_setg(errp,
589                    "maximum number of ISA NE2000 devices exceeded");
590         return false;
591     }
592     isa_ne2000_init(bus, ne2000_io[nb_ne2k],
593                     ne2000_irq[nb_ne2k], nd);
594     nb_ne2k++;
595     return true;
596 }
597 
pc_acpi_smi_interrupt(void * opaque,int irq,int level)598 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
599 {
600     X86CPU *cpu = opaque;
601 
602     if (level) {
603         cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
604     }
605 }
606 
607 static
pc_machine_done(Notifier * notifier,void * data)608 void pc_machine_done(Notifier *notifier, void *data)
609 {
610     PCMachineState *pcms = container_of(notifier,
611                                         PCMachineState, machine_done);
612     X86MachineState *x86ms = X86_MACHINE(pcms);
613 
614     cxl_hook_up_pxb_registers(pcms->pcibus, &pcms->cxl_devices_state,
615                               &error_fatal);
616 
617     if (pcms->cxl_devices_state.is_enabled) {
618         cxl_fmws_link_targets(&pcms->cxl_devices_state, &error_fatal);
619     }
620 
621     /* set the number of CPUs */
622     x86_rtc_set_cpus_count(x86ms->rtc, x86ms->boot_cpus);
623 
624     fw_cfg_add_extra_pci_roots(pcms->pcibus, x86ms->fw_cfg);
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 
pc_get_cxl_range_end(PCMachineState * pcms)726 static uint64_t pc_get_cxl_range_end(PCMachineState *pcms)
727 {
728     uint64_t start = pc_get_cxl_range_start(pcms) + MiB;
729 
730     if (pcms->cxl_devices_state.fixed_windows) {
731         GList *it;
732 
733         start = ROUND_UP(start, 256 * MiB);
734         for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
735             CXLFixedWindow *fw = it->data;
736             start += fw->size;
737         }
738     }
739 
740     return start;
741 }
742 
pc_max_used_gpa(PCMachineState * pcms,uint64_t pci_hole64_size)743 static hwaddr pc_max_used_gpa(PCMachineState *pcms, uint64_t pci_hole64_size)
744 {
745     X86CPU *cpu = X86_CPU(first_cpu);
746     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
747     MachineState *ms = MACHINE(pcms);
748 
749     if (cpu->env.features[FEAT_8000_0001_EDX] & CPUID_EXT2_LM) {
750         /* 64-bit systems */
751         return pc_pci_hole64_start() + pci_hole64_size - 1;
752     }
753 
754     /* 32-bit systems */
755     if (pcmc->broken_32bit_mem_addr_check) {
756         /* old value for compatibility reasons */
757         return ((hwaddr)1 << cpu->phys_bits) - 1;
758     }
759 
760     /*
761      * 32-bit systems don't have hole64 but they might have a region for
762      * memory devices. Even if additional hotplugged memory devices might
763      * not be usable by most guest OSes, we need to still consider them for
764      * calculating the highest possible GPA so that we can properly report
765      * if someone configures them on a CPU that cannot possibly address them.
766      */
767     if (pcmc->has_reserved_memory &&
768         (ms->ram_size < ms->maxram_size)) {
769         hwaddr devmem_start;
770         ram_addr_t devmem_size;
771 
772         pc_get_device_memory_range(pcms, &devmem_start, &devmem_size);
773         devmem_start += devmem_size;
774         return devmem_start - 1;
775     }
776 
777     /* configuration without any memory hotplug */
778     return pc_above_4g_end(pcms) - 1;
779 }
780 
781 /*
782  * AMD systems with an IOMMU have an additional hole close to the
783  * 1Tb, which are special GPAs that cannot be DMA mapped. Depending
784  * on kernel version, VFIO may or may not let you DMA map those ranges.
785  * Starting Linux v5.4 we validate it, and can't create guests on AMD machines
786  * with certain memory sizes. It's also wrong to use those IOVA ranges
787  * in detriment of leading to IOMMU INVALID_DEVICE_REQUEST or worse.
788  * The ranges reserved for Hyper-Transport are:
789  *
790  * FD_0000_0000h - FF_FFFF_FFFFh
791  *
792  * The ranges represent the following:
793  *
794  * Base Address   Top Address  Use
795  *
796  * FD_0000_0000h FD_F7FF_FFFFh Reserved interrupt address space
797  * FD_F800_0000h FD_F8FF_FFFFh Interrupt/EOI IntCtl
798  * FD_F900_0000h FD_F90F_FFFFh Legacy PIC IACK
799  * FD_F910_0000h FD_F91F_FFFFh System Management
800  * FD_F920_0000h FD_FAFF_FFFFh Reserved Page Tables
801  * FD_FB00_0000h FD_FBFF_FFFFh Address Translation
802  * FD_FC00_0000h FD_FDFF_FFFFh I/O Space
803  * FD_FE00_0000h FD_FFFF_FFFFh Configuration
804  * FE_0000_0000h FE_1FFF_FFFFh Extended Configuration/Device Messages
805  * FE_2000_0000h FF_FFFF_FFFFh Reserved
806  *
807  * See AMD IOMMU spec, section 2.1.2 "IOMMU Logical Topology",
808  * Table 3: Special Address Controls (GPA) for more information.
809  */
810 #define AMD_HT_START         0xfd00000000UL
811 #define AMD_HT_END           0xffffffffffUL
812 #define AMD_ABOVE_1TB_START  (AMD_HT_END + 1)
813 #define AMD_HT_SIZE          (AMD_ABOVE_1TB_START - AMD_HT_START)
814 
pc_memory_init(PCMachineState * pcms,MemoryRegion * system_memory,MemoryRegion * rom_memory,uint64_t pci_hole64_size)815 void pc_memory_init(PCMachineState *pcms,
816                     MemoryRegion *system_memory,
817                     MemoryRegion *rom_memory,
818                     uint64_t pci_hole64_size)
819 {
820     int linux_boot, i;
821     MemoryRegion *option_rom_mr;
822     MemoryRegion *ram_below_4g, *ram_above_4g;
823     FWCfgState *fw_cfg;
824     MachineState *machine = MACHINE(pcms);
825     MachineClass *mc = MACHINE_GET_CLASS(machine);
826     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
827     X86MachineState *x86ms = X86_MACHINE(pcms);
828     hwaddr maxphysaddr, maxusedaddr;
829     hwaddr cxl_base, cxl_resv_end = 0;
830     X86CPU *cpu = X86_CPU(first_cpu);
831 
832     assert(machine->ram_size == x86ms->below_4g_mem_size +
833                                 x86ms->above_4g_mem_size);
834 
835     linux_boot = (machine->kernel_filename != NULL);
836 
837     /*
838      * The HyperTransport range close to the 1T boundary is unique to AMD
839      * hosts with IOMMUs enabled. Restrict the ram-above-4g relocation
840      * to above 1T to AMD vCPUs only. @enforce_amd_1tb_hole is only false in
841      * older machine types (<= 7.0) for compatibility purposes.
842      */
843     if (IS_AMD_CPU(&cpu->env) && pcmc->enforce_amd_1tb_hole) {
844         /* Bail out if max possible address does not cross HT range */
845         if (pc_max_used_gpa(pcms, pci_hole64_size) >= AMD_HT_START) {
846             x86ms->above_4g_mem_start = AMD_ABOVE_1TB_START;
847         }
848 
849         /*
850          * Advertise the HT region if address space covers the reserved
851          * region or if we relocate.
852          */
853         if (cpu->phys_bits >= 40) {
854             e820_add_entry(AMD_HT_START, AMD_HT_SIZE, E820_RESERVED);
855         }
856     }
857 
858     /*
859      * phys-bits is required to be appropriately configured
860      * to make sure max used GPA is reachable.
861      */
862     maxusedaddr = pc_max_used_gpa(pcms, pci_hole64_size);
863     maxphysaddr = ((hwaddr)1 << cpu->phys_bits) - 1;
864     if (maxphysaddr < maxusedaddr) {
865         error_report("Address space limit 0x%"PRIx64" < 0x%"PRIx64
866                      " phys-bits too low (%u)",
867                      maxphysaddr, maxusedaddr, cpu->phys_bits);
868         exit(EXIT_FAILURE);
869     }
870 
871     /*
872      * Split single memory region and use aliases to address portions of it,
873      * done for backwards compatibility with older qemus.
874      */
875     ram_below_4g = g_malloc(sizeof(*ram_below_4g));
876     memory_region_init_alias(ram_below_4g, NULL, "ram-below-4g", machine->ram,
877                              0, x86ms->below_4g_mem_size);
878     memory_region_add_subregion(system_memory, 0, ram_below_4g);
879     e820_add_entry(0, x86ms->below_4g_mem_size, E820_RAM);
880     if (x86ms->above_4g_mem_size > 0) {
881         ram_above_4g = g_malloc(sizeof(*ram_above_4g));
882         memory_region_init_alias(ram_above_4g, NULL, "ram-above-4g",
883                                  machine->ram,
884                                  x86ms->below_4g_mem_size,
885                                  x86ms->above_4g_mem_size);
886         memory_region_add_subregion(system_memory, x86ms->above_4g_mem_start,
887                                     ram_above_4g);
888         e820_add_entry(x86ms->above_4g_mem_start, x86ms->above_4g_mem_size,
889                        E820_RAM);
890     }
891 
892     if (pcms->sgx_epc.size != 0) {
893         e820_add_entry(pcms->sgx_epc.base, pcms->sgx_epc.size, E820_RESERVED);
894     }
895 
896     if (!pcmc->has_reserved_memory &&
897         (machine->ram_slots ||
898          (machine->maxram_size > machine->ram_size))) {
899 
900         error_report("\"-memory 'slots|maxmem'\" is not supported by: %s",
901                      mc->name);
902         exit(EXIT_FAILURE);
903     }
904 
905     /* initialize device memory address space */
906     if (pcmc->has_reserved_memory &&
907         (machine->ram_size < machine->maxram_size)) {
908         ram_addr_t device_mem_size;
909         hwaddr device_mem_base;
910 
911         if (machine->ram_slots > ACPI_MAX_RAM_SLOTS) {
912             error_report("unsupported amount of memory slots: %"PRIu64,
913                          machine->ram_slots);
914             exit(EXIT_FAILURE);
915         }
916 
917         if (QEMU_ALIGN_UP(machine->maxram_size,
918                           TARGET_PAGE_SIZE) != machine->maxram_size) {
919             error_report("maximum memory size must by aligned to multiple of "
920                          "%d bytes", TARGET_PAGE_SIZE);
921             exit(EXIT_FAILURE);
922         }
923 
924         pc_get_device_memory_range(pcms, &device_mem_base, &device_mem_size);
925 
926         if (device_mem_base + device_mem_size < device_mem_size) {
927             error_report("unsupported amount of maximum memory: " RAM_ADDR_FMT,
928                          machine->maxram_size);
929             exit(EXIT_FAILURE);
930         }
931         machine_memory_devices_init(machine, device_mem_base, device_mem_size);
932     }
933 
934     if (pcms->cxl_devices_state.is_enabled) {
935         MemoryRegion *mr = &pcms->cxl_devices_state.host_mr;
936         hwaddr cxl_size = MiB;
937 
938         cxl_base = pc_get_cxl_range_start(pcms);
939         memory_region_init(mr, OBJECT(machine), "cxl_host_reg", cxl_size);
940         memory_region_add_subregion(system_memory, cxl_base, mr);
941         cxl_resv_end = cxl_base + cxl_size;
942         if (pcms->cxl_devices_state.fixed_windows) {
943             hwaddr cxl_fmw_base;
944             GList *it;
945 
946             cxl_fmw_base = ROUND_UP(cxl_base + cxl_size, 256 * MiB);
947             for (it = pcms->cxl_devices_state.fixed_windows; it; it = it->next) {
948                 CXLFixedWindow *fw = it->data;
949 
950                 fw->base = cxl_fmw_base;
951                 memory_region_init_io(&fw->mr, OBJECT(machine), &cfmws_ops, fw,
952                                       "cxl-fixed-memory-region", fw->size);
953                 memory_region_add_subregion(system_memory, fw->base, &fw->mr);
954                 cxl_fmw_base += fw->size;
955                 cxl_resv_end = cxl_fmw_base;
956             }
957         }
958     }
959 
960     /* Initialize PC system firmware */
961     pc_system_firmware_init(pcms, rom_memory);
962 
963     option_rom_mr = g_malloc(sizeof(*option_rom_mr));
964     if (machine_require_guest_memfd(machine)) {
965         memory_region_init_ram_guest_memfd(option_rom_mr, NULL, "pc.rom",
966                                            PC_ROM_SIZE, &error_fatal);
967     } else {
968         memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE,
969                                &error_fatal);
970         if (pcmc->pci_enabled) {
971             memory_region_set_readonly(option_rom_mr, true);
972         }
973     }
974     memory_region_add_subregion_overlap(rom_memory,
975                                         PC_ROM_MIN_VGA,
976                                         option_rom_mr,
977                                         1);
978 
979     fw_cfg = fw_cfg_arch_create(machine,
980                                 x86ms->boot_cpus, x86ms->apic_id_limit);
981 
982     rom_set_fw(fw_cfg);
983 
984     if (machine->device_memory) {
985         uint64_t *val = g_malloc(sizeof(*val));
986         uint64_t res_mem_end = machine->device_memory->base;
987 
988         if (!pcmc->broken_reserved_end) {
989             res_mem_end += memory_region_size(&machine->device_memory->mr);
990         }
991 
992         if (pcms->cxl_devices_state.is_enabled) {
993             res_mem_end = cxl_resv_end;
994         }
995         *val = cpu_to_le64(ROUND_UP(res_mem_end, 1 * GiB));
996         fw_cfg_add_file(fw_cfg, "etc/reserved-memory-end", val, sizeof(*val));
997     }
998 
999     if (linux_boot) {
1000         x86_load_linux(x86ms, fw_cfg, PC_FW_DATA, pcmc->pvh_enabled);
1001     }
1002 
1003     for (i = 0; i < nb_option_roms; i++) {
1004         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1005     }
1006     x86ms->fw_cfg = fw_cfg;
1007 
1008     /* Init default IOAPIC address space */
1009     x86ms->ioapic_as = &address_space_memory;
1010 
1011     /* Init ACPI memory hotplug IO base address */
1012     pcms->memhp_io_base = ACPI_MEMORY_HOTPLUG_BASE;
1013 }
1014 
1015 /*
1016  * The 64bit pci hole starts after "above 4G RAM" and
1017  * potentially the space reserved for memory hotplug.
1018  */
pc_pci_hole64_start(void)1019 uint64_t pc_pci_hole64_start(void)
1020 {
1021     PCMachineState *pcms = PC_MACHINE(qdev_get_machine());
1022     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1023     MachineState *ms = MACHINE(pcms);
1024     uint64_t hole64_start = 0;
1025     ram_addr_t size = 0;
1026 
1027     if (pcms->cxl_devices_state.is_enabled) {
1028         hole64_start = pc_get_cxl_range_end(pcms);
1029     } else if (pcmc->has_reserved_memory && (ms->ram_size < ms->maxram_size)) {
1030         pc_get_device_memory_range(pcms, &hole64_start, &size);
1031         if (!pcmc->broken_reserved_end) {
1032             hole64_start += size;
1033         }
1034     } else {
1035         hole64_start = pc_above_4g_end(pcms);
1036     }
1037 
1038     return ROUND_UP(hole64_start, 1 * GiB);
1039 }
1040 
pc_vga_init(ISABus * isa_bus,PCIBus * pci_bus)1041 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1042 {
1043     DeviceState *dev = NULL;
1044 
1045     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_VGA);
1046     if (pci_bus) {
1047         PCIDevice *pcidev = pci_vga_init(pci_bus);
1048         dev = pcidev ? &pcidev->qdev : NULL;
1049     } else if (isa_bus) {
1050         ISADevice *isadev = isa_vga_init(isa_bus);
1051         dev = isadev ? DEVICE(isadev) : NULL;
1052     }
1053     rom_reset_order_override();
1054     return dev;
1055 }
1056 
1057 static const MemoryRegionOps ioport80_io_ops = {
1058     .write = ioport80_write,
1059     .read = ioport80_read,
1060     .endianness = DEVICE_NATIVE_ENDIAN,
1061     .impl = {
1062         .min_access_size = 1,
1063         .max_access_size = 1,
1064     },
1065 };
1066 
1067 static const MemoryRegionOps ioportF0_io_ops = {
1068     .write = ioportF0_write,
1069     .read = ioportF0_read,
1070     .endianness = DEVICE_NATIVE_ENDIAN,
1071     .impl = {
1072         .min_access_size = 1,
1073         .max_access_size = 1,
1074     },
1075 };
1076 
pc_superio_init(ISABus * isa_bus,bool create_fdctrl,bool create_i8042,bool no_vmport,Error ** errp)1077 static void pc_superio_init(ISABus *isa_bus, bool create_fdctrl,
1078                             bool create_i8042, bool no_vmport, Error **errp)
1079 {
1080     int i;
1081     DriveInfo *fd[MAX_FD];
1082     qemu_irq *a20_line;
1083     ISADevice *i8042, *port92, *vmmouse;
1084 
1085     serial_hds_isa_init(isa_bus, 0, MAX_ISA_SERIAL_PORTS);
1086     parallel_hds_isa_init(isa_bus, MAX_PARALLEL_PORTS);
1087 
1088     for (i = 0; i < MAX_FD; i++) {
1089         fd[i] = drive_get(IF_FLOPPY, 0, i);
1090         create_fdctrl |= !!fd[i];
1091     }
1092     if (create_fdctrl) {
1093 #ifdef CONFIG_FDC_ISA
1094         ISADevice *fdc = isa_new(TYPE_ISA_FDC);
1095         if (fdc) {
1096             isa_realize_and_unref(fdc, isa_bus, &error_fatal);
1097             isa_fdc_init_drives(fdc, fd);
1098         }
1099 #endif
1100     }
1101 
1102     if (!create_i8042) {
1103         if (!no_vmport) {
1104             error_setg(errp,
1105                        "vmport requires the i8042 controller to be enabled");
1106         }
1107         return;
1108     }
1109 
1110     i8042 = isa_create_simple(isa_bus, TYPE_I8042);
1111     if (!no_vmport) {
1112         isa_create_simple(isa_bus, TYPE_VMPORT);
1113         vmmouse = isa_try_new("vmmouse");
1114     } else {
1115         vmmouse = NULL;
1116     }
1117     if (vmmouse) {
1118         object_property_set_link(OBJECT(vmmouse), TYPE_I8042, OBJECT(i8042),
1119                                  &error_abort);
1120         isa_realize_and_unref(vmmouse, isa_bus, &error_fatal);
1121     }
1122     port92 = isa_create_simple(isa_bus, TYPE_PORT92);
1123 
1124     a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1125     qdev_connect_gpio_out_named(DEVICE(i8042),
1126                                 I8042_A20_LINE, 0, a20_line[0]);
1127     qdev_connect_gpio_out_named(DEVICE(port92),
1128                                 PORT92_A20_LINE, 0, a20_line[1]);
1129     g_free(a20_line);
1130 }
1131 
pc_basic_device_init(struct PCMachineState * pcms,ISABus * isa_bus,qemu_irq * gsi,ISADevice * rtc_state,bool create_fdctrl,uint32_t hpet_irqs)1132 void pc_basic_device_init(struct PCMachineState *pcms,
1133                           ISABus *isa_bus, qemu_irq *gsi,
1134                           ISADevice *rtc_state,
1135                           bool create_fdctrl,
1136                           uint32_t hpet_irqs)
1137 {
1138     int i;
1139     DeviceState *hpet = NULL;
1140     int pit_isa_irq = 0;
1141     qemu_irq pit_alt_irq = NULL;
1142     ISADevice *pit = NULL;
1143     MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1144     MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
1145     X86MachineState *x86ms = X86_MACHINE(pcms);
1146 
1147     memory_region_init_io(ioport80_io, NULL, &ioport80_io_ops, NULL, "ioport80", 1);
1148     memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
1149 
1150     memory_region_init_io(ioportF0_io, NULL, &ioportF0_io_ops, NULL, "ioportF0", 1);
1151     memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
1152 
1153     /*
1154      * Check if an HPET shall be created.
1155      */
1156     if (pcms->hpet_enabled) {
1157         qemu_irq rtc_irq;
1158 
1159         hpet = qdev_try_new(TYPE_HPET);
1160         if (!hpet) {
1161             error_report("couldn't create HPET device");
1162             exit(1);
1163         }
1164         /*
1165          * For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-*,
1166          * use IRQ16~23, IRQ8 and IRQ2.  If the user has already set
1167          * the property, use whatever mask they specified.
1168          */
1169         uint8_t compat = object_property_get_uint(OBJECT(hpet),
1170                 HPET_INTCAP, NULL);
1171         if (!compat) {
1172             qdev_prop_set_uint32(hpet, HPET_INTCAP, hpet_irqs);
1173         }
1174         sysbus_realize_and_unref(SYS_BUS_DEVICE(hpet), &error_fatal);
1175         sysbus_mmio_map(SYS_BUS_DEVICE(hpet), 0, HPET_BASE);
1176 
1177         for (i = 0; i < IOAPIC_NUM_PINS; i++) {
1178             sysbus_connect_irq(SYS_BUS_DEVICE(hpet), i, gsi[i]);
1179         }
1180         pit_isa_irq = -1;
1181         pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1182         rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1183 
1184         /* overwrite connection created by south bridge */
1185         qdev_connect_gpio_out(DEVICE(rtc_state), 0, rtc_irq);
1186     }
1187 
1188     object_property_add_alias(OBJECT(pcms), "rtc-time", OBJECT(rtc_state),
1189                               "date");
1190 
1191 #ifdef CONFIG_XEN_EMU
1192     if (xen_mode == XEN_EMULATE) {
1193         xen_overlay_create();
1194         xen_evtchn_create(IOAPIC_NUM_PINS, gsi);
1195         xen_gnttab_create();
1196         xen_xenstore_create();
1197         if (pcms->pcibus) {
1198             pci_create_simple(pcms->pcibus, -1, "xen-platform");
1199         }
1200         xen_bus_init();
1201     }
1202 #endif
1203 
1204     qemu_register_boot_set(pc_boot_set, pcms);
1205     set_boot_dev(pcms, MC146818_RTC(rtc_state),
1206                  MACHINE(pcms)->boot_config.order, &error_fatal);
1207 
1208     if (!xen_enabled() &&
1209         (x86ms->pit == ON_OFF_AUTO_AUTO || x86ms->pit == ON_OFF_AUTO_ON)) {
1210         if (kvm_pit_in_kernel()) {
1211             pit = kvm_pit_init(isa_bus, 0x40);
1212         } else {
1213             pit = i8254_pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1214         }
1215         if (hpet) {
1216             /* connect PIT to output control line of the HPET */
1217             qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(DEVICE(pit), 0));
1218         }
1219         object_property_set_link(OBJECT(pcms->pcspk), "pit",
1220                                  OBJECT(pit), &error_fatal);
1221         isa_realize_and_unref(pcms->pcspk, isa_bus, &error_fatal);
1222     }
1223 
1224     assert(pcms->vmport >= 0 && pcms->vmport < ON_OFF_AUTO__MAX);
1225     if (pcms->vmport == ON_OFF_AUTO_AUTO) {
1226         pcms->vmport = (xen_enabled() || !pcms->i8042_enabled)
1227             ? ON_OFF_AUTO_OFF : ON_OFF_AUTO_ON;
1228     }
1229 
1230     /* Super I/O */
1231     pc_superio_init(isa_bus, create_fdctrl, pcms->i8042_enabled,
1232                     pcms->vmport != ON_OFF_AUTO_ON, &error_fatal);
1233 }
1234 
pc_nic_init(PCMachineClass * pcmc,ISABus * isa_bus,PCIBus * pci_bus)1235 void pc_nic_init(PCMachineClass *pcmc, ISABus *isa_bus, PCIBus *pci_bus)
1236 {
1237     MachineClass *mc = MACHINE_CLASS(pcmc);
1238     bool default_is_ne2k = g_str_equal(mc->default_nic, TYPE_ISA_NE2000);
1239     NICInfo *nd;
1240 
1241     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_NIC);
1242 
1243     while ((nd = qemu_find_nic_info(TYPE_ISA_NE2000, default_is_ne2k, NULL))) {
1244         pc_init_ne2k_isa(isa_bus, nd, &error_fatal);
1245     }
1246 
1247     /* Anything remaining should be a PCI NIC */
1248     pci_init_nic_devices(pci_bus, mc->default_nic);
1249 
1250     rom_reset_order_override();
1251 }
1252 
pc_i8259_create(ISABus * isa_bus,qemu_irq * i8259_irqs)1253 void pc_i8259_create(ISABus *isa_bus, qemu_irq *i8259_irqs)
1254 {
1255     qemu_irq *i8259;
1256 
1257     if (kvm_pic_in_kernel()) {
1258         i8259 = kvm_i8259_init(isa_bus);
1259     } else if (xen_enabled()) {
1260         i8259 = xen_interrupt_controller_init();
1261     } else {
1262         i8259 = i8259_init(isa_bus, x86_allocate_cpu_irq());
1263     }
1264 
1265     for (size_t i = 0; i < ISA_NUM_IRQS; i++) {
1266         i8259_irqs[i] = i8259[i];
1267     }
1268 
1269     g_free(i8259);
1270 }
1271 
pc_memory_pre_plug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1272 static void pc_memory_pre_plug(HotplugHandler *hotplug_dev, DeviceState *dev,
1273                                Error **errp)
1274 {
1275     const X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1276     const MachineState *ms = MACHINE(hotplug_dev);
1277     const bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1278     Error *local_err = NULL;
1279 
1280     /*
1281      * When "acpi=off" is used with the Q35 machine type, no ACPI is built,
1282      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1283      * addition to cover this case.
1284      */
1285     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1286         error_setg(errp,
1287                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1288         return;
1289     }
1290 
1291     if (is_nvdimm && !ms->nvdimms_state->is_enabled) {
1292         error_setg(errp, "nvdimm is not enabled: missing 'nvdimm' in '-M'");
1293         return;
1294     }
1295 
1296     hotplug_handler_pre_plug(x86ms->acpi_dev, dev, &local_err);
1297     if (local_err) {
1298         error_propagate(errp, local_err);
1299         return;
1300     }
1301 
1302     pc_dimm_pre_plug(PC_DIMM(dev), MACHINE(hotplug_dev), errp);
1303 }
1304 
pc_memory_plug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1305 static void pc_memory_plug(HotplugHandler *hotplug_dev,
1306                            DeviceState *dev, Error **errp)
1307 {
1308     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1309     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1310     MachineState *ms = MACHINE(hotplug_dev);
1311     bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1312 
1313     pc_dimm_plug(PC_DIMM(dev), MACHINE(pcms));
1314 
1315     if (is_nvdimm) {
1316         nvdimm_plug(ms->nvdimms_state);
1317     }
1318 
1319     hotplug_handler_plug(x86ms->acpi_dev, dev, &error_abort);
1320 }
1321 
pc_memory_unplug_request(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1322 static void pc_memory_unplug_request(HotplugHandler *hotplug_dev,
1323                                      DeviceState *dev, Error **errp)
1324 {
1325     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1326 
1327     /*
1328      * When "acpi=off" is used with the Q35 machine type, no ACPI is built,
1329      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1330      * addition to cover this case.
1331      */
1332     if (!x86ms->acpi_dev || !x86_machine_is_acpi_enabled(x86ms)) {
1333         error_setg(errp,
1334                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1335         return;
1336     }
1337 
1338     if (object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM)) {
1339         error_setg(errp, "nvdimm device hot unplug is not supported yet.");
1340         return;
1341     }
1342 
1343     hotplug_handler_unplug_request(x86ms->acpi_dev, dev,
1344                                    errp);
1345 }
1346 
pc_memory_unplug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1347 static void pc_memory_unplug(HotplugHandler *hotplug_dev,
1348                              DeviceState *dev, Error **errp)
1349 {
1350     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1351     X86MachineState *x86ms = X86_MACHINE(hotplug_dev);
1352     Error *local_err = NULL;
1353 
1354     hotplug_handler_unplug(x86ms->acpi_dev, dev, &local_err);
1355     if (local_err) {
1356         goto out;
1357     }
1358 
1359     pc_dimm_unplug(PC_DIMM(dev), MACHINE(pcms));
1360     qdev_unrealize(dev);
1361  out:
1362     error_propagate(errp, local_err);
1363 }
1364 
pc_hv_balloon_pre_plug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1365 static void pc_hv_balloon_pre_plug(HotplugHandler *hotplug_dev,
1366                                    DeviceState *dev, Error **errp)
1367 {
1368     /* The vmbus handler has no hotplug handler; we should never end up here. */
1369     g_assert(!dev->hotplugged);
1370     memory_device_pre_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev), errp);
1371 }
1372 
pc_hv_balloon_plug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1373 static void pc_hv_balloon_plug(HotplugHandler *hotplug_dev,
1374                                DeviceState *dev, Error **errp)
1375 {
1376     memory_device_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev));
1377 }
1378 
pc_machine_device_pre_plug_cb(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1379 static void pc_machine_device_pre_plug_cb(HotplugHandler *hotplug_dev,
1380                                           DeviceState *dev, Error **errp)
1381 {
1382     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1383         pc_memory_pre_plug(hotplug_dev, dev, errp);
1384     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1385         x86_cpu_pre_plug(hotplug_dev, dev, errp);
1386     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1387         virtio_md_pci_pre_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1388     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1389         /* Declare the APIC range as the reserved MSI region */
1390         char *resv_prop_str = g_strdup_printf("0xfee00000:0xfeefffff:%d",
1391                                               VIRTIO_IOMMU_RESV_MEM_T_MSI);
1392         QList *reserved_regions = qlist_new();
1393 
1394         qlist_append_str(reserved_regions, resv_prop_str);
1395         qdev_prop_set_array(dev, "reserved-regions", reserved_regions);
1396 
1397         g_free(resv_prop_str);
1398     }
1399 
1400     if (object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE) ||
1401         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI)) {
1402         PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1403 
1404         if (pcms->iommu) {
1405             error_setg(errp, "QEMU does not support multiple vIOMMUs "
1406                        "for x86 yet.");
1407             return;
1408         }
1409         pcms->iommu = dev;
1410     } else if (object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON)) {
1411         pc_hv_balloon_pre_plug(hotplug_dev, dev, errp);
1412     }
1413 }
1414 
pc_machine_device_plug_cb(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1415 static void pc_machine_device_plug_cb(HotplugHandler *hotplug_dev,
1416                                       DeviceState *dev, Error **errp)
1417 {
1418     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1419         pc_memory_plug(hotplug_dev, dev, errp);
1420     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1421         x86_cpu_plug(hotplug_dev, dev, errp);
1422     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1423         virtio_md_pci_plug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1424     } else if (object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON)) {
1425         pc_hv_balloon_plug(hotplug_dev, dev, errp);
1426     }
1427 }
1428 
pc_machine_device_unplug_request_cb(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1429 static void pc_machine_device_unplug_request_cb(HotplugHandler *hotplug_dev,
1430                                                 DeviceState *dev, Error **errp)
1431 {
1432     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1433         pc_memory_unplug_request(hotplug_dev, dev, errp);
1434     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1435         x86_cpu_unplug_request_cb(hotplug_dev, dev, errp);
1436     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1437         virtio_md_pci_unplug_request(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev),
1438                                      errp);
1439     } else {
1440         error_setg(errp, "acpi: device unplug request for not supported device"
1441                    " type: %s", object_get_typename(OBJECT(dev)));
1442     }
1443 }
1444 
pc_machine_device_unplug_cb(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1445 static void pc_machine_device_unplug_cb(HotplugHandler *hotplug_dev,
1446                                         DeviceState *dev, Error **errp)
1447 {
1448     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1449         pc_memory_unplug(hotplug_dev, dev, errp);
1450     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1451         x86_cpu_unplug_cb(hotplug_dev, dev, errp);
1452     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
1453         virtio_md_pci_unplug(VIRTIO_MD_PCI(dev), MACHINE(hotplug_dev), errp);
1454     } else {
1455         error_setg(errp, "acpi: device unplug for not supported device"
1456                    " type: %s", object_get_typename(OBJECT(dev)));
1457     }
1458 }
1459 
pc_get_hotplug_handler(MachineState * machine,DeviceState * dev)1460 static HotplugHandler *pc_get_hotplug_handler(MachineState *machine,
1461                                              DeviceState *dev)
1462 {
1463     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM) ||
1464         object_dynamic_cast(OBJECT(dev), TYPE_CPU) ||
1465         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI) ||
1466         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_IOMMU_PCI) ||
1467         object_dynamic_cast(OBJECT(dev), TYPE_HV_BALLOON) ||
1468         object_dynamic_cast(OBJECT(dev), TYPE_X86_IOMMU_DEVICE)) {
1469         return HOTPLUG_HANDLER(machine);
1470     }
1471 
1472     return NULL;
1473 }
1474 
pc_machine_get_vmport(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1475 static void pc_machine_get_vmport(Object *obj, Visitor *v, const char *name,
1476                                   void *opaque, Error **errp)
1477 {
1478     PCMachineState *pcms = PC_MACHINE(obj);
1479     OnOffAuto vmport = pcms->vmport;
1480 
1481     visit_type_OnOffAuto(v, name, &vmport, errp);
1482 }
1483 
pc_machine_set_vmport(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1484 static void pc_machine_set_vmport(Object *obj, Visitor *v, const char *name,
1485                                   void *opaque, Error **errp)
1486 {
1487     PCMachineState *pcms = PC_MACHINE(obj);
1488 
1489     visit_type_OnOffAuto(v, name, &pcms->vmport, errp);
1490 }
1491 
pc_machine_get_fd_bootchk(Object * obj,Error ** errp)1492 static bool pc_machine_get_fd_bootchk(Object *obj, Error **errp)
1493 {
1494     PCMachineState *pcms = PC_MACHINE(obj);
1495 
1496     return pcms->fd_bootchk;
1497 }
1498 
pc_machine_set_fd_bootchk(Object * obj,bool value,Error ** errp)1499 static void pc_machine_set_fd_bootchk(Object *obj, bool value, Error **errp)
1500 {
1501     PCMachineState *pcms = PC_MACHINE(obj);
1502 
1503     pcms->fd_bootchk = value;
1504 }
1505 
pc_machine_get_smbus(Object * obj,Error ** errp)1506 static bool pc_machine_get_smbus(Object *obj, Error **errp)
1507 {
1508     PCMachineState *pcms = PC_MACHINE(obj);
1509 
1510     return pcms->smbus_enabled;
1511 }
1512 
pc_machine_set_smbus(Object * obj,bool value,Error ** errp)1513 static void pc_machine_set_smbus(Object *obj, bool value, Error **errp)
1514 {
1515     PCMachineState *pcms = PC_MACHINE(obj);
1516 
1517     pcms->smbus_enabled = value;
1518 }
1519 
pc_machine_get_sata(Object * obj,Error ** errp)1520 static bool pc_machine_get_sata(Object *obj, Error **errp)
1521 {
1522     PCMachineState *pcms = PC_MACHINE(obj);
1523 
1524     return pcms->sata_enabled;
1525 }
1526 
pc_machine_set_sata(Object * obj,bool value,Error ** errp)1527 static void pc_machine_set_sata(Object *obj, bool value, Error **errp)
1528 {
1529     PCMachineState *pcms = PC_MACHINE(obj);
1530 
1531     pcms->sata_enabled = value;
1532 }
1533 
pc_machine_get_hpet(Object * obj,Error ** errp)1534 static bool pc_machine_get_hpet(Object *obj, Error **errp)
1535 {
1536     PCMachineState *pcms = PC_MACHINE(obj);
1537 
1538     return pcms->hpet_enabled;
1539 }
1540 
pc_machine_set_hpet(Object * obj,bool value,Error ** errp)1541 static void pc_machine_set_hpet(Object *obj, bool value, Error **errp)
1542 {
1543     PCMachineState *pcms = PC_MACHINE(obj);
1544 
1545     pcms->hpet_enabled = value;
1546 }
1547 
pc_machine_get_i8042(Object * obj,Error ** errp)1548 static bool pc_machine_get_i8042(Object *obj, Error **errp)
1549 {
1550     PCMachineState *pcms = PC_MACHINE(obj);
1551 
1552     return pcms->i8042_enabled;
1553 }
1554 
pc_machine_set_i8042(Object * obj,bool value,Error ** errp)1555 static void pc_machine_set_i8042(Object *obj, bool value, Error **errp)
1556 {
1557     PCMachineState *pcms = PC_MACHINE(obj);
1558 
1559     pcms->i8042_enabled = value;
1560 }
1561 
pc_machine_get_default_bus_bypass_iommu(Object * obj,Error ** errp)1562 static bool pc_machine_get_default_bus_bypass_iommu(Object *obj, Error **errp)
1563 {
1564     PCMachineState *pcms = PC_MACHINE(obj);
1565 
1566     return pcms->default_bus_bypass_iommu;
1567 }
1568 
pc_machine_set_default_bus_bypass_iommu(Object * obj,bool value,Error ** errp)1569 static void pc_machine_set_default_bus_bypass_iommu(Object *obj, bool value,
1570                                                     Error **errp)
1571 {
1572     PCMachineState *pcms = PC_MACHINE(obj);
1573 
1574     pcms->default_bus_bypass_iommu = value;
1575 }
1576 
pc_machine_get_smbios_ep(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1577 static void pc_machine_get_smbios_ep(Object *obj, Visitor *v, const char *name,
1578                                      void *opaque, Error **errp)
1579 {
1580     PCMachineState *pcms = PC_MACHINE(obj);
1581     SmbiosEntryPointType smbios_entry_point_type = pcms->smbios_entry_point_type;
1582 
1583     visit_type_SmbiosEntryPointType(v, name, &smbios_entry_point_type, errp);
1584 }
1585 
pc_machine_set_smbios_ep(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1586 static void pc_machine_set_smbios_ep(Object *obj, Visitor *v, const char *name,
1587                                      void *opaque, Error **errp)
1588 {
1589     PCMachineState *pcms = PC_MACHINE(obj);
1590 
1591     visit_type_SmbiosEntryPointType(v, name, &pcms->smbios_entry_point_type, errp);
1592 }
1593 
pc_machine_get_max_ram_below_4g(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1594 static void pc_machine_get_max_ram_below_4g(Object *obj, Visitor *v,
1595                                             const char *name, void *opaque,
1596                                             Error **errp)
1597 {
1598     PCMachineState *pcms = PC_MACHINE(obj);
1599     uint64_t value = pcms->max_ram_below_4g;
1600 
1601     visit_type_size(v, name, &value, errp);
1602 }
1603 
pc_machine_set_max_ram_below_4g(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1604 static void pc_machine_set_max_ram_below_4g(Object *obj, Visitor *v,
1605                                             const char *name, void *opaque,
1606                                             Error **errp)
1607 {
1608     PCMachineState *pcms = PC_MACHINE(obj);
1609     uint64_t value;
1610 
1611     if (!visit_type_size(v, name, &value, errp)) {
1612         return;
1613     }
1614     if (value > 4 * GiB) {
1615         error_setg(errp,
1616                    "Machine option 'max-ram-below-4g=%"PRIu64
1617                    "' expects size less than or equal to 4G", value);
1618         return;
1619     }
1620 
1621     if (value < 1 * MiB) {
1622         warn_report("Only %" PRIu64 " bytes of RAM below the 4GiB boundary,"
1623                     "BIOS may not work with less than 1MiB", value);
1624     }
1625 
1626     pcms->max_ram_below_4g = value;
1627 }
1628 
pc_machine_get_max_fw_size(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1629 static void pc_machine_get_max_fw_size(Object *obj, Visitor *v,
1630                                        const char *name, void *opaque,
1631                                        Error **errp)
1632 {
1633     PCMachineState *pcms = PC_MACHINE(obj);
1634     uint64_t value = pcms->max_fw_size;
1635 
1636     visit_type_size(v, name, &value, errp);
1637 }
1638 
pc_machine_set_max_fw_size(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1639 static void pc_machine_set_max_fw_size(Object *obj, Visitor *v,
1640                                        const char *name, void *opaque,
1641                                        Error **errp)
1642 {
1643     PCMachineState *pcms = PC_MACHINE(obj);
1644     uint64_t value;
1645 
1646     if (!visit_type_size(v, name, &value, errp)) {
1647         return;
1648     }
1649 
1650     /*
1651      * We don't have a theoretically justifiable exact lower bound on the base
1652      * address of any flash mapping. In practice, the IO-APIC MMIO range is
1653      * [0xFEE00000..0xFEE01000] -- see IO_APIC_DEFAULT_ADDRESS --, leaving free
1654      * only 18MiB-4KiB below 4GiB. For now, restrict the cumulative mapping to
1655      * 16MiB in size.
1656      */
1657     if (value > 16 * MiB) {
1658         error_setg(errp,
1659                    "User specified max allowed firmware size %" PRIu64 " is "
1660                    "greater than 16MiB. If combined firmware size exceeds "
1661                    "16MiB the system may not boot, or experience intermittent"
1662                    "stability issues.",
1663                    value);
1664         return;
1665     }
1666 
1667     pcms->max_fw_size = value;
1668 }
1669 
1670 
pc_machine_initfn(Object * obj)1671 static void pc_machine_initfn(Object *obj)
1672 {
1673     PCMachineState *pcms = PC_MACHINE(obj);
1674     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1675 
1676 #ifdef CONFIG_VMPORT
1677     pcms->vmport = ON_OFF_AUTO_AUTO;
1678 #else
1679     pcms->vmport = ON_OFF_AUTO_OFF;
1680 #endif /* CONFIG_VMPORT */
1681     pcms->max_ram_below_4g = 0; /* use default */
1682     pcms->smbios_entry_point_type = pcmc->default_smbios_ep_type;
1683     pcms->south_bridge = pcmc->default_south_bridge;
1684 
1685     /* acpi build is enabled by default if machine supports it */
1686     pcms->acpi_build_enabled = pcmc->has_acpi_build;
1687     pcms->smbus_enabled = true;
1688     pcms->sata_enabled = true;
1689     pcms->i8042_enabled = true;
1690     pcms->max_fw_size = 8 * MiB;
1691 #ifdef CONFIG_HPET
1692     pcms->hpet_enabled = true;
1693 #endif
1694     pcms->fd_bootchk = true;
1695     pcms->default_bus_bypass_iommu = false;
1696 
1697     pc_system_flash_create(pcms);
1698     pcms->pcspk = isa_new(TYPE_PC_SPEAKER);
1699     object_property_add_alias(OBJECT(pcms), "pcspk-audiodev",
1700                               OBJECT(pcms->pcspk), "audiodev");
1701     if (pcmc->pci_enabled) {
1702         cxl_machine_init(obj, &pcms->cxl_devices_state);
1703     }
1704 
1705     pcms->machine_done.notify = pc_machine_done;
1706     qemu_add_machine_init_done_notifier(&pcms->machine_done);
1707 }
1708 
pc_machine_reset(MachineState * machine,ShutdownCause reason)1709 static void pc_machine_reset(MachineState *machine, ShutdownCause reason)
1710 {
1711     CPUState *cs;
1712     X86CPU *cpu;
1713 
1714     qemu_devices_reset(reason);
1715 
1716     /* Reset APIC after devices have been reset to cancel
1717      * any changes that qemu_devices_reset() might have done.
1718      */
1719     CPU_FOREACH(cs) {
1720         cpu = X86_CPU(cs);
1721 
1722         x86_cpu_after_reset(cpu);
1723     }
1724 }
1725 
pc_machine_wakeup(MachineState * machine)1726 static void pc_machine_wakeup(MachineState *machine)
1727 {
1728     cpu_synchronize_all_states();
1729     pc_machine_reset(machine, SHUTDOWN_CAUSE_NONE);
1730     cpu_synchronize_all_post_reset();
1731 }
1732 
pc_hotplug_allowed(MachineState * ms,DeviceState * dev,Error ** errp)1733 static bool pc_hotplug_allowed(MachineState *ms, DeviceState *dev, Error **errp)
1734 {
1735     X86IOMMUState *iommu = x86_iommu_get_default();
1736     IntelIOMMUState *intel_iommu;
1737 
1738     if (iommu &&
1739         object_dynamic_cast((Object *)iommu, TYPE_INTEL_IOMMU_DEVICE) &&
1740         object_dynamic_cast((Object *)dev, "vfio-pci")) {
1741         intel_iommu = INTEL_IOMMU_DEVICE(iommu);
1742         if (!intel_iommu->caching_mode) {
1743             error_setg(errp, "Device assignment is not allowed without "
1744                        "enabling caching-mode=on for Intel IOMMU.");
1745             return false;
1746         }
1747     }
1748 
1749     return true;
1750 }
1751 
pc_machine_class_init(ObjectClass * oc,void * data)1752 static void pc_machine_class_init(ObjectClass *oc, void *data)
1753 {
1754     MachineClass *mc = MACHINE_CLASS(oc);
1755     X86MachineClass *x86mc = X86_MACHINE_CLASS(oc);
1756     PCMachineClass *pcmc = PC_MACHINE_CLASS(oc);
1757     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
1758 
1759     pcmc->pci_enabled = true;
1760     pcmc->has_acpi_build = true;
1761     pcmc->smbios_defaults = true;
1762     pcmc->gigabyte_align = true;
1763     pcmc->has_reserved_memory = true;
1764     pcmc->enforce_amd_1tb_hole = true;
1765     pcmc->isa_bios_alias = true;
1766     pcmc->pvh_enabled = true;
1767     pcmc->kvmclock_create_always = true;
1768     x86mc->apic_xrupt_override = true;
1769     assert(!mc->get_hotplug_handler);
1770     mc->get_hotplug_handler = pc_get_hotplug_handler;
1771     mc->hotplug_allowed = pc_hotplug_allowed;
1772     mc->auto_enable_numa_with_memhp = true;
1773     mc->auto_enable_numa_with_memdev = true;
1774     mc->has_hotpluggable_cpus = true;
1775     mc->default_boot_order = "cad";
1776     mc->block_default_type = IF_IDE;
1777     mc->max_cpus = 255;
1778     mc->reset = pc_machine_reset;
1779     mc->wakeup = pc_machine_wakeup;
1780     hc->pre_plug = pc_machine_device_pre_plug_cb;
1781     hc->plug = pc_machine_device_plug_cb;
1782     hc->unplug_request = pc_machine_device_unplug_request_cb;
1783     hc->unplug = pc_machine_device_unplug_cb;
1784     mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE;
1785     mc->nvdimm_supported = true;
1786     mc->smp_props.dies_supported = true;
1787     mc->smp_props.modules_supported = true;
1788     mc->default_ram_id = "pc.ram";
1789     pcmc->default_smbios_ep_type = SMBIOS_ENTRY_POINT_TYPE_AUTO;
1790 
1791     object_class_property_add(oc, PC_MACHINE_MAX_RAM_BELOW_4G, "size",
1792         pc_machine_get_max_ram_below_4g, pc_machine_set_max_ram_below_4g,
1793         NULL, NULL);
1794     object_class_property_set_description(oc, PC_MACHINE_MAX_RAM_BELOW_4G,
1795         "Maximum ram below the 4G boundary (32bit boundary)");
1796 
1797     object_class_property_add(oc, PC_MACHINE_VMPORT, "OnOffAuto",
1798         pc_machine_get_vmport, pc_machine_set_vmport,
1799         NULL, NULL);
1800     object_class_property_set_description(oc, PC_MACHINE_VMPORT,
1801         "Enable vmport (pc & q35)");
1802 
1803     object_class_property_add_bool(oc, PC_MACHINE_SMBUS,
1804         pc_machine_get_smbus, pc_machine_set_smbus);
1805     object_class_property_set_description(oc, PC_MACHINE_SMBUS,
1806         "Enable/disable system management bus");
1807 
1808     object_class_property_add_bool(oc, PC_MACHINE_SATA,
1809         pc_machine_get_sata, pc_machine_set_sata);
1810     object_class_property_set_description(oc, PC_MACHINE_SATA,
1811         "Enable/disable Serial ATA bus");
1812 
1813     object_class_property_add_bool(oc, "hpet",
1814         pc_machine_get_hpet, pc_machine_set_hpet);
1815     object_class_property_set_description(oc, "hpet",
1816         "Enable/disable high precision event timer emulation");
1817 
1818     object_class_property_add_bool(oc, PC_MACHINE_I8042,
1819         pc_machine_get_i8042, pc_machine_set_i8042);
1820 
1821     object_class_property_add_bool(oc, "default-bus-bypass-iommu",
1822         pc_machine_get_default_bus_bypass_iommu,
1823         pc_machine_set_default_bus_bypass_iommu);
1824 
1825     object_class_property_add(oc, PC_MACHINE_MAX_FW_SIZE, "size",
1826         pc_machine_get_max_fw_size, pc_machine_set_max_fw_size,
1827         NULL, NULL);
1828     object_class_property_set_description(oc, PC_MACHINE_MAX_FW_SIZE,
1829         "Maximum combined firmware size");
1830 
1831     object_class_property_add(oc, PC_MACHINE_SMBIOS_EP, "str",
1832         pc_machine_get_smbios_ep, pc_machine_set_smbios_ep,
1833         NULL, NULL);
1834     object_class_property_set_description(oc, PC_MACHINE_SMBIOS_EP,
1835         "SMBIOS Entry Point type [32, 64]");
1836 
1837     object_class_property_add_bool(oc, "fd-bootchk",
1838         pc_machine_get_fd_bootchk,
1839         pc_machine_set_fd_bootchk);
1840 }
1841 
1842 static const TypeInfo pc_machine_info = {
1843     .name = TYPE_PC_MACHINE,
1844     .parent = TYPE_X86_MACHINE,
1845     .abstract = true,
1846     .instance_size = sizeof(PCMachineState),
1847     .instance_init = pc_machine_initfn,
1848     .class_size = sizeof(PCMachineClass),
1849     .class_init = pc_machine_class_init,
1850     .interfaces = (InterfaceInfo[]) {
1851          { TYPE_HOTPLUG_HANDLER },
1852          { }
1853     },
1854 };
1855 
pc_machine_register_types(void)1856 static void pc_machine_register_types(void)
1857 {
1858     type_register_static(&pc_machine_info);
1859 }
1860 
1861 type_init(pc_machine_register_types)
1862