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/x86.h"
28 #include "hw/i386/pc.h"
29 #include "hw/char/serial.h"
30 #include "hw/char/parallel.h"
31 #include "hw/i386/apic.h"
32 #include "hw/i386/topology.h"
33 #include "hw/i386/fw_cfg.h"
34 #include "sysemu/cpus.h"
35 #include "hw/block/fdc.h"
36 #include "hw/ide.h"
37 #include "hw/pci/pci.h"
38 #include "hw/pci/pci_bus.h"
39 #include "hw/nvram/fw_cfg.h"
40 #include "hw/timer/hpet.h"
41 #include "hw/firmware/smbios.h"
42 #include "hw/loader.h"
43 #include "elf.h"
44 #include "migration/vmstate.h"
45 #include "multiboot.h"
46 #include "hw/rtc/mc146818rtc.h"
47 #include "hw/intc/i8259.h"
48 #include "hw/dma/i8257.h"
49 #include "hw/timer/i8254.h"
50 #include "hw/input/i8042.h"
51 #include "hw/irq.h"
52 #include "hw/audio/pcspk.h"
53 #include "hw/pci/msi.h"
54 #include "hw/sysbus.h"
55 #include "sysemu/sysemu.h"
56 #include "sysemu/tcg.h"
57 #include "sysemu/numa.h"
58 #include "sysemu/kvm.h"
59 #include "sysemu/qtest.h"
60 #include "sysemu/reset.h"
61 #include "sysemu/runstate.h"
62 #include "kvm_i386.h"
63 #include "hw/xen/xen.h"
64 #include "hw/xen/start_info.h"
65 #include "ui/qemu-spice.h"
66 #include "exec/memory.h"
67 #include "exec/address-spaces.h"
68 #include "sysemu/arch_init.h"
69 #include "qemu/bitmap.h"
70 #include "qemu/config-file.h"
71 #include "qemu/error-report.h"
72 #include "qemu/option.h"
73 #include "qemu/cutils.h"
74 #include "hw/acpi/acpi.h"
75 #include "hw/acpi/cpu_hotplug.h"
76 #include "hw/boards.h"
77 #include "acpi-build.h"
78 #include "hw/mem/pc-dimm.h"
79 #include "hw/mem/nvdimm.h"
80 #include "qapi/error.h"
81 #include "qapi/qapi-visit-common.h"
82 #include "qapi/visitor.h"
83 #include "hw/core/cpu.h"
84 #include "hw/usb.h"
85 #include "hw/i386/intel_iommu.h"
86 #include "hw/net/ne2000-isa.h"
87 #include "standard-headers/asm-x86/bootparam.h"
88 #include "hw/virtio/virtio-pmem-pci.h"
89 #include "hw/mem/memory-device.h"
90 #include "sysemu/replay.h"
91 #include "qapi/qmp/qerror.h"
92 #include "config-devices.h"
93 #include "e820_memory_layout.h"
94 #include "fw_cfg.h"
95 #include "trace.h"
96 
97 GlobalProperty pc_compat_4_2[] = {
98     { "mch", "smbase-smram", "off" },
99 };
100 const size_t pc_compat_4_2_len = G_N_ELEMENTS(pc_compat_4_2);
101 
102 GlobalProperty pc_compat_4_1[] = {};
103 const size_t pc_compat_4_1_len = G_N_ELEMENTS(pc_compat_4_1);
104 
105 GlobalProperty pc_compat_4_0[] = {};
106 const size_t pc_compat_4_0_len = G_N_ELEMENTS(pc_compat_4_0);
107 
108 GlobalProperty pc_compat_3_1[] = {
109     { "intel-iommu", "dma-drain", "off" },
110     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "off" },
111     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "off" },
112     { "Opteron_G4" "-" TYPE_X86_CPU, "npt", "off" },
113     { "Opteron_G4" "-" TYPE_X86_CPU, "nrip-save", "off" },
114     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "off" },
115     { "Opteron_G5" "-" TYPE_X86_CPU, "npt", "off" },
116     { "Opteron_G5" "-" TYPE_X86_CPU, "nrip-save", "off" },
117     { "EPYC" "-" TYPE_X86_CPU, "npt", "off" },
118     { "EPYC" "-" TYPE_X86_CPU, "nrip-save", "off" },
119     { "EPYC-IBPB" "-" TYPE_X86_CPU, "npt", "off" },
120     { "EPYC-IBPB" "-" TYPE_X86_CPU, "nrip-save", "off" },
121     { "Skylake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
122     { "Skylake-Client-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
123     { "Skylake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
124     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "mpx", "on" },
125     { "Cascadelake-Server" "-" TYPE_X86_CPU,  "mpx", "on" },
126     { "Icelake-Client" "-" TYPE_X86_CPU,      "mpx", "on" },
127     { "Icelake-Server" "-" TYPE_X86_CPU,      "mpx", "on" },
128     { "Cascadelake-Server" "-" TYPE_X86_CPU, "stepping", "5" },
129     { TYPE_X86_CPU, "x-intel-pt-auto-level", "off" },
130 };
131 const size_t pc_compat_3_1_len = G_N_ELEMENTS(pc_compat_3_1);
132 
133 GlobalProperty pc_compat_3_0[] = {
134     { TYPE_X86_CPU, "x-hv-synic-kvm-only", "on" },
135     { "Skylake-Server" "-" TYPE_X86_CPU, "pku", "off" },
136     { "Skylake-Server-IBRS" "-" TYPE_X86_CPU, "pku", "off" },
137 };
138 const size_t pc_compat_3_0_len = G_N_ELEMENTS(pc_compat_3_0);
139 
140 GlobalProperty pc_compat_2_12[] = {
141     { TYPE_X86_CPU, "legacy-cache", "on" },
142     { TYPE_X86_CPU, "topoext", "off" },
143     { "EPYC-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
144     { "EPYC-IBPB-" TYPE_X86_CPU, "xlevel", "0x8000000a" },
145 };
146 const size_t pc_compat_2_12_len = G_N_ELEMENTS(pc_compat_2_12);
147 
148 GlobalProperty pc_compat_2_11[] = {
149     { TYPE_X86_CPU, "x-migrate-smi-count", "off" },
150     { "Skylake-Server" "-" TYPE_X86_CPU, "clflushopt", "off" },
151 };
152 const size_t pc_compat_2_11_len = G_N_ELEMENTS(pc_compat_2_11);
153 
154 GlobalProperty pc_compat_2_10[] = {
155     { TYPE_X86_CPU, "x-hv-max-vps", "0x40" },
156     { "i440FX-pcihost", "x-pci-hole64-fix", "off" },
157     { "q35-pcihost", "x-pci-hole64-fix", "off" },
158 };
159 const size_t pc_compat_2_10_len = G_N_ELEMENTS(pc_compat_2_10);
160 
161 GlobalProperty pc_compat_2_9[] = {
162     { "mch", "extended-tseg-mbytes", "0" },
163 };
164 const size_t pc_compat_2_9_len = G_N_ELEMENTS(pc_compat_2_9);
165 
166 GlobalProperty pc_compat_2_8[] = {
167     { TYPE_X86_CPU, "tcg-cpuid", "off" },
168     { "kvmclock", "x-mach-use-reliable-get-clock", "off" },
169     { "ICH9-LPC", "x-smi-broadcast", "off" },
170     { TYPE_X86_CPU, "vmware-cpuid-freq", "off" },
171     { "Haswell-" TYPE_X86_CPU, "stepping", "1" },
172 };
173 const size_t pc_compat_2_8_len = G_N_ELEMENTS(pc_compat_2_8);
174 
175 GlobalProperty pc_compat_2_7[] = {
176     { TYPE_X86_CPU, "l3-cache", "off" },
177     { TYPE_X86_CPU, "full-cpuid-auto-level", "off" },
178     { "Opteron_G3" "-" TYPE_X86_CPU, "family", "15" },
179     { "Opteron_G3" "-" TYPE_X86_CPU, "model", "6" },
180     { "Opteron_G3" "-" TYPE_X86_CPU, "stepping", "1" },
181     { "isa-pcspk", "migrate", "off" },
182 };
183 const size_t pc_compat_2_7_len = G_N_ELEMENTS(pc_compat_2_7);
184 
185 GlobalProperty pc_compat_2_6[] = {
186     { TYPE_X86_CPU, "cpuid-0xb", "off" },
187     { "vmxnet3", "romfile", "" },
188     { TYPE_X86_CPU, "fill-mtrr-mask", "off" },
189     { "apic-common", "legacy-instance-id", "on", }
190 };
191 const size_t pc_compat_2_6_len = G_N_ELEMENTS(pc_compat_2_6);
192 
193 GlobalProperty pc_compat_2_5[] = {};
194 const size_t pc_compat_2_5_len = G_N_ELEMENTS(pc_compat_2_5);
195 
196 GlobalProperty pc_compat_2_4[] = {
197     PC_CPU_MODEL_IDS("2.4.0")
198     { "Haswell-" TYPE_X86_CPU, "abm", "off" },
199     { "Haswell-noTSX-" TYPE_X86_CPU, "abm", "off" },
200     { "Broadwell-" TYPE_X86_CPU, "abm", "off" },
201     { "Broadwell-noTSX-" TYPE_X86_CPU, "abm", "off" },
202     { "host" "-" TYPE_X86_CPU, "host-cache-info", "on" },
203     { TYPE_X86_CPU, "check", "off" },
204     { "qemu64" "-" TYPE_X86_CPU, "sse4a", "on" },
205     { "qemu64" "-" TYPE_X86_CPU, "abm", "on" },
206     { "qemu64" "-" TYPE_X86_CPU, "popcnt", "on" },
207     { "qemu32" "-" TYPE_X86_CPU, "popcnt", "on" },
208     { "Opteron_G2" "-" TYPE_X86_CPU, "rdtscp", "on" },
209     { "Opteron_G3" "-" TYPE_X86_CPU, "rdtscp", "on" },
210     { "Opteron_G4" "-" TYPE_X86_CPU, "rdtscp", "on" },
211     { "Opteron_G5" "-" TYPE_X86_CPU, "rdtscp", "on", }
212 };
213 const size_t pc_compat_2_4_len = G_N_ELEMENTS(pc_compat_2_4);
214 
215 GlobalProperty pc_compat_2_3[] = {
216     PC_CPU_MODEL_IDS("2.3.0")
217     { TYPE_X86_CPU, "arat", "off" },
218     { "qemu64" "-" TYPE_X86_CPU, "min-level", "4" },
219     { "kvm64" "-" TYPE_X86_CPU, "min-level", "5" },
220     { "pentium3" "-" TYPE_X86_CPU, "min-level", "2" },
221     { "n270" "-" TYPE_X86_CPU, "min-level", "5" },
222     { "Conroe" "-" TYPE_X86_CPU, "min-level", "4" },
223     { "Penryn" "-" TYPE_X86_CPU, "min-level", "4" },
224     { "Nehalem" "-" TYPE_X86_CPU, "min-level", "4" },
225     { "n270" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
226     { "Penryn" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
227     { "Conroe" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
228     { "Nehalem" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
229     { "Westmere" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
230     { "SandyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
231     { "IvyBridge" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
232     { "Haswell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
233     { "Haswell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
234     { "Broadwell" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
235     { "Broadwell-noTSX" "-" TYPE_X86_CPU, "min-xlevel", "0x8000000a" },
236     { TYPE_X86_CPU, "kvm-no-smi-migration", "on" },
237 };
238 const size_t pc_compat_2_3_len = G_N_ELEMENTS(pc_compat_2_3);
239 
240 GlobalProperty pc_compat_2_2[] = {
241     PC_CPU_MODEL_IDS("2.2.0")
242     { "kvm64" "-" TYPE_X86_CPU, "vme", "off" },
243     { "kvm32" "-" TYPE_X86_CPU, "vme", "off" },
244     { "Conroe" "-" TYPE_X86_CPU, "vme", "off" },
245     { "Penryn" "-" TYPE_X86_CPU, "vme", "off" },
246     { "Nehalem" "-" TYPE_X86_CPU, "vme", "off" },
247     { "Westmere" "-" TYPE_X86_CPU, "vme", "off" },
248     { "SandyBridge" "-" TYPE_X86_CPU, "vme", "off" },
249     { "Haswell" "-" TYPE_X86_CPU, "vme", "off" },
250     { "Broadwell" "-" TYPE_X86_CPU, "vme", "off" },
251     { "Opteron_G1" "-" TYPE_X86_CPU, "vme", "off" },
252     { "Opteron_G2" "-" TYPE_X86_CPU, "vme", "off" },
253     { "Opteron_G3" "-" TYPE_X86_CPU, "vme", "off" },
254     { "Opteron_G4" "-" TYPE_X86_CPU, "vme", "off" },
255     { "Opteron_G5" "-" TYPE_X86_CPU, "vme", "off" },
256     { "Haswell" "-" TYPE_X86_CPU, "f16c", "off" },
257     { "Haswell" "-" TYPE_X86_CPU, "rdrand", "off" },
258     { "Broadwell" "-" TYPE_X86_CPU, "f16c", "off" },
259     { "Broadwell" "-" TYPE_X86_CPU, "rdrand", "off" },
260 };
261 const size_t pc_compat_2_2_len = G_N_ELEMENTS(pc_compat_2_2);
262 
263 GlobalProperty pc_compat_2_1[] = {
264     PC_CPU_MODEL_IDS("2.1.0")
265     { "coreduo" "-" TYPE_X86_CPU, "vmx", "on" },
266     { "core2duo" "-" TYPE_X86_CPU, "vmx", "on" },
267 };
268 const size_t pc_compat_2_1_len = G_N_ELEMENTS(pc_compat_2_1);
269 
270 GlobalProperty pc_compat_2_0[] = {
271     PC_CPU_MODEL_IDS("2.0.0")
272     { "virtio-scsi-pci", "any_layout", "off" },
273     { "PIIX4_PM", "memory-hotplug-support", "off" },
274     { "apic", "version", "0x11" },
275     { "nec-usb-xhci", "superspeed-ports-first", "off" },
276     { "nec-usb-xhci", "force-pcie-endcap", "on" },
277     { "pci-serial", "prog_if", "0" },
278     { "pci-serial-2x", "prog_if", "0" },
279     { "pci-serial-4x", "prog_if", "0" },
280     { "virtio-net-pci", "guest_announce", "off" },
281     { "ICH9-LPC", "memory-hotplug-support", "off" },
282     { "xio3130-downstream", COMPAT_PROP_PCP, "off" },
283     { "ioh3420", COMPAT_PROP_PCP, "off" },
284 };
285 const size_t pc_compat_2_0_len = G_N_ELEMENTS(pc_compat_2_0);
286 
287 GlobalProperty pc_compat_1_7[] = {
288     PC_CPU_MODEL_IDS("1.7.0")
289     { TYPE_USB_DEVICE, "msos-desc", "no" },
290     { "PIIX4_PM", "acpi-pci-hotplug-with-bridge-support", "off" },
291     { "hpet", HPET_INTCAP, "4" },
292 };
293 const size_t pc_compat_1_7_len = G_N_ELEMENTS(pc_compat_1_7);
294 
295 GlobalProperty pc_compat_1_6[] = {
296     PC_CPU_MODEL_IDS("1.6.0")
297     { "e1000", "mitigation", "off" },
298     { "qemu64-" TYPE_X86_CPU, "model", "2" },
299     { "qemu32-" TYPE_X86_CPU, "model", "3" },
300     { "i440FX-pcihost", "short_root_bus", "1" },
301     { "q35-pcihost", "short_root_bus", "1" },
302 };
303 const size_t pc_compat_1_6_len = G_N_ELEMENTS(pc_compat_1_6);
304 
305 GlobalProperty pc_compat_1_5[] = {
306     PC_CPU_MODEL_IDS("1.5.0")
307     { "Conroe-" TYPE_X86_CPU, "model", "2" },
308     { "Conroe-" TYPE_X86_CPU, "min-level", "2" },
309     { "Penryn-" TYPE_X86_CPU, "model", "2" },
310     { "Penryn-" TYPE_X86_CPU, "min-level", "2" },
311     { "Nehalem-" TYPE_X86_CPU, "model", "2" },
312     { "Nehalem-" TYPE_X86_CPU, "min-level", "2" },
313     { "virtio-net-pci", "any_layout", "off" },
314     { TYPE_X86_CPU, "pmu", "on" },
315     { "i440FX-pcihost", "short_root_bus", "0" },
316     { "q35-pcihost", "short_root_bus", "0" },
317 };
318 const size_t pc_compat_1_5_len = G_N_ELEMENTS(pc_compat_1_5);
319 
320 GlobalProperty pc_compat_1_4[] = {
321     PC_CPU_MODEL_IDS("1.4.0")
322     { "scsi-hd", "discard_granularity", "0" },
323     { "scsi-cd", "discard_granularity", "0" },
324     { "scsi-disk", "discard_granularity", "0" },
325     { "ide-hd", "discard_granularity", "0" },
326     { "ide-cd", "discard_granularity", "0" },
327     { "ide-drive", "discard_granularity", "0" },
328     { "virtio-blk-pci", "discard_granularity", "0" },
329     /* DEV_NVECTORS_UNSPECIFIED as a uint32_t string: */
330     { "virtio-serial-pci", "vectors", "0xFFFFFFFF" },
331     { "virtio-net-pci", "ctrl_guest_offloads", "off" },
332     { "e1000", "romfile", "pxe-e1000.rom" },
333     { "ne2k_pci", "romfile", "pxe-ne2k_pci.rom" },
334     { "pcnet", "romfile", "pxe-pcnet.rom" },
335     { "rtl8139", "romfile", "pxe-rtl8139.rom" },
336     { "virtio-net-pci", "romfile", "pxe-virtio.rom" },
337     { "486-" TYPE_X86_CPU, "model", "0" },
338     { "n270" "-" TYPE_X86_CPU, "movbe", "off" },
339     { "Westmere" "-" TYPE_X86_CPU, "pclmulqdq", "off" },
340 };
341 const size_t pc_compat_1_4_len = G_N_ELEMENTS(pc_compat_1_4);
342 
pc_gsi_create(qemu_irq ** irqs,bool pci_enabled)343 GSIState *pc_gsi_create(qemu_irq **irqs, bool pci_enabled)
344 {
345     GSIState *s;
346 
347     s = g_new0(GSIState, 1);
348     if (kvm_ioapic_in_kernel()) {
349         kvm_pc_setup_irq_routing(pci_enabled);
350     }
351     *irqs = qemu_allocate_irqs(gsi_handler, s, GSI_NUM_PINS);
352 
353     return s;
354 }
355 
ioport80_write(void * opaque,hwaddr addr,uint64_t data,unsigned size)356 static void ioport80_write(void *opaque, hwaddr addr, uint64_t data,
357                            unsigned size)
358 {
359 }
360 
ioport80_read(void * opaque,hwaddr addr,unsigned size)361 static uint64_t ioport80_read(void *opaque, hwaddr addr, unsigned size)
362 {
363     return 0xffffffffffffffffULL;
364 }
365 
366 /* MSDOS compatibility mode FPU exception support */
ioportF0_write(void * opaque,hwaddr addr,uint64_t data,unsigned size)367 static void ioportF0_write(void *opaque, hwaddr addr, uint64_t data,
368                            unsigned size)
369 {
370     if (tcg_enabled()) {
371         cpu_set_ignne();
372     }
373 }
374 
ioportF0_read(void * opaque,hwaddr addr,unsigned size)375 static uint64_t ioportF0_read(void *opaque, hwaddr addr, unsigned size)
376 {
377     return 0xffffffffffffffffULL;
378 }
379 
380 /* PC cmos mappings */
381 
382 #define REG_EQUIPMENT_BYTE          0x14
383 
cmos_get_fd_drive_type(FloppyDriveType fd0)384 int cmos_get_fd_drive_type(FloppyDriveType fd0)
385 {
386     int val;
387 
388     switch (fd0) {
389     case FLOPPY_DRIVE_TYPE_144:
390         /* 1.44 Mb 3"5 drive */
391         val = 4;
392         break;
393     case FLOPPY_DRIVE_TYPE_288:
394         /* 2.88 Mb 3"5 drive */
395         val = 5;
396         break;
397     case FLOPPY_DRIVE_TYPE_120:
398         /* 1.2 Mb 5"5 drive */
399         val = 2;
400         break;
401     case FLOPPY_DRIVE_TYPE_NONE:
402     default:
403         val = 0;
404         break;
405     }
406     return val;
407 }
408 
cmos_init_hd(ISADevice * s,int type_ofs,int info_ofs,int16_t cylinders,int8_t heads,int8_t sectors)409 static void cmos_init_hd(ISADevice *s, int type_ofs, int info_ofs,
410                          int16_t cylinders, int8_t heads, int8_t sectors)
411 {
412     rtc_set_memory(s, type_ofs, 47);
413     rtc_set_memory(s, info_ofs, cylinders);
414     rtc_set_memory(s, info_ofs + 1, cylinders >> 8);
415     rtc_set_memory(s, info_ofs + 2, heads);
416     rtc_set_memory(s, info_ofs + 3, 0xff);
417     rtc_set_memory(s, info_ofs + 4, 0xff);
418     rtc_set_memory(s, info_ofs + 5, 0xc0 | ((heads > 8) << 3));
419     rtc_set_memory(s, info_ofs + 6, cylinders);
420     rtc_set_memory(s, info_ofs + 7, cylinders >> 8);
421     rtc_set_memory(s, info_ofs + 8, sectors);
422 }
423 
424 /* convert boot_device letter to something recognizable by the bios */
boot_device2nibble(char boot_device)425 static int boot_device2nibble(char boot_device)
426 {
427     switch(boot_device) {
428     case 'a':
429     case 'b':
430         return 0x01; /* floppy boot */
431     case 'c':
432         return 0x02; /* hard drive boot */
433     case 'd':
434         return 0x03; /* CD-ROM boot */
435     case 'n':
436         return 0x04; /* Network boot */
437     }
438     return 0;
439 }
440 
set_boot_dev(ISADevice * s,const char * boot_device,Error ** errp)441 static void set_boot_dev(ISADevice *s, const char *boot_device, Error **errp)
442 {
443 #define PC_MAX_BOOT_DEVICES 3
444     int nbds, bds[3] = { 0, };
445     int i;
446 
447     nbds = strlen(boot_device);
448     if (nbds > PC_MAX_BOOT_DEVICES) {
449         error_setg(errp, "Too many boot devices for PC");
450         return;
451     }
452     for (i = 0; i < nbds; i++) {
453         bds[i] = boot_device2nibble(boot_device[i]);
454         if (bds[i] == 0) {
455             error_setg(errp, "Invalid boot device for PC: '%c'",
456                        boot_device[i]);
457             return;
458         }
459     }
460     rtc_set_memory(s, 0x3d, (bds[1] << 4) | bds[0]);
461     rtc_set_memory(s, 0x38, (bds[2] << 4) | (fd_bootchk ? 0x0 : 0x1));
462 }
463 
pc_boot_set(void * opaque,const char * boot_device,Error ** errp)464 static void pc_boot_set(void *opaque, const char *boot_device, Error **errp)
465 {
466     set_boot_dev(opaque, boot_device, errp);
467 }
468 
pc_cmos_init_floppy(ISADevice * rtc_state,ISADevice * floppy)469 static void pc_cmos_init_floppy(ISADevice *rtc_state, ISADevice *floppy)
470 {
471     int val, nb, i;
472     FloppyDriveType fd_type[2] = { FLOPPY_DRIVE_TYPE_NONE,
473                                    FLOPPY_DRIVE_TYPE_NONE };
474 
475     /* floppy type */
476     if (floppy) {
477         for (i = 0; i < 2; i++) {
478             fd_type[i] = isa_fdc_get_drive_type(floppy, i);
479         }
480     }
481     val = (cmos_get_fd_drive_type(fd_type[0]) << 4) |
482         cmos_get_fd_drive_type(fd_type[1]);
483     rtc_set_memory(rtc_state, 0x10, val);
484 
485     val = rtc_get_memory(rtc_state, REG_EQUIPMENT_BYTE);
486     nb = 0;
487     if (fd_type[0] != FLOPPY_DRIVE_TYPE_NONE) {
488         nb++;
489     }
490     if (fd_type[1] != FLOPPY_DRIVE_TYPE_NONE) {
491         nb++;
492     }
493     switch (nb) {
494     case 0:
495         break;
496     case 1:
497         val |= 0x01; /* 1 drive, ready for boot */
498         break;
499     case 2:
500         val |= 0x41; /* 2 drives, ready for boot */
501         break;
502     }
503     rtc_set_memory(rtc_state, REG_EQUIPMENT_BYTE, val);
504 }
505 
506 typedef struct pc_cmos_init_late_arg {
507     ISADevice *rtc_state;
508     BusState *idebus[2];
509 } pc_cmos_init_late_arg;
510 
511 typedef struct check_fdc_state {
512     ISADevice *floppy;
513     bool multiple;
514 } CheckFdcState;
515 
check_fdc(Object * obj,void * opaque)516 static int check_fdc(Object *obj, void *opaque)
517 {
518     CheckFdcState *state = opaque;
519     Object *fdc;
520     uint32_t iobase;
521     Error *local_err = NULL;
522 
523     fdc = object_dynamic_cast(obj, TYPE_ISA_FDC);
524     if (!fdc) {
525         return 0;
526     }
527 
528     iobase = object_property_get_uint(obj, "iobase", &local_err);
529     if (local_err || iobase != 0x3f0) {
530         error_free(local_err);
531         return 0;
532     }
533 
534     if (state->floppy) {
535         state->multiple = true;
536     } else {
537         state->floppy = ISA_DEVICE(obj);
538     }
539     return 0;
540 }
541 
542 static const char * const fdc_container_path[] = {
543     "/unattached", "/peripheral", "/peripheral-anon"
544 };
545 
546 /*
547  * Locate the FDC at IO address 0x3f0, in order to configure the CMOS registers
548  * and ACPI objects.
549  */
pc_find_fdc0(void)550 ISADevice *pc_find_fdc0(void)
551 {
552     int i;
553     Object *container;
554     CheckFdcState state = { 0 };
555 
556     for (i = 0; i < ARRAY_SIZE(fdc_container_path); i++) {
557         container = container_get(qdev_get_machine(), fdc_container_path[i]);
558         object_child_foreach(container, check_fdc, &state);
559     }
560 
561     if (state.multiple) {
562         warn_report("multiple floppy disk controllers with "
563                     "iobase=0x3f0 have been found");
564         error_printf("the one being picked for CMOS setup might not reflect "
565                      "your intent");
566     }
567 
568     return state.floppy;
569 }
570 
pc_cmos_init_late(void * opaque)571 static void pc_cmos_init_late(void *opaque)
572 {
573     pc_cmos_init_late_arg *arg = opaque;
574     ISADevice *s = arg->rtc_state;
575     int16_t cylinders;
576     int8_t heads, sectors;
577     int val;
578     int i, trans;
579 
580     val = 0;
581     if (arg->idebus[0] && ide_get_geometry(arg->idebus[0], 0,
582                                            &cylinders, &heads, &sectors) >= 0) {
583         cmos_init_hd(s, 0x19, 0x1b, cylinders, heads, sectors);
584         val |= 0xf0;
585     }
586     if (arg->idebus[0] && ide_get_geometry(arg->idebus[0], 1,
587                                            &cylinders, &heads, &sectors) >= 0) {
588         cmos_init_hd(s, 0x1a, 0x24, cylinders, heads, sectors);
589         val |= 0x0f;
590     }
591     rtc_set_memory(s, 0x12, val);
592 
593     val = 0;
594     for (i = 0; i < 4; i++) {
595         /* NOTE: ide_get_geometry() returns the physical
596            geometry.  It is always such that: 1 <= sects <= 63, 1
597            <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
598            geometry can be different if a translation is done. */
599         if (arg->idebus[i / 2] &&
600             ide_get_geometry(arg->idebus[i / 2], i % 2,
601                              &cylinders, &heads, &sectors) >= 0) {
602             trans = ide_get_bios_chs_trans(arg->idebus[i / 2], i % 2) - 1;
603             assert((trans & ~3) == 0);
604             val |= trans << (i * 2);
605         }
606     }
607     rtc_set_memory(s, 0x39, val);
608 
609     pc_cmos_init_floppy(s, pc_find_fdc0());
610 
611     qemu_unregister_reset(pc_cmos_init_late, opaque);
612 }
613 
pc_cmos_init(PCMachineState * pcms,BusState * idebus0,BusState * idebus1,ISADevice * s)614 void pc_cmos_init(PCMachineState *pcms,
615                   BusState *idebus0, BusState *idebus1,
616                   ISADevice *s)
617 {
618     int val;
619     static pc_cmos_init_late_arg arg;
620     X86MachineState *x86ms = X86_MACHINE(pcms);
621 
622     /* various important CMOS locations needed by PC/Bochs bios */
623 
624     /* memory size */
625     /* base memory (first MiB) */
626     val = MIN(x86ms->below_4g_mem_size / KiB, 640);
627     rtc_set_memory(s, 0x15, val);
628     rtc_set_memory(s, 0x16, val >> 8);
629     /* extended memory (next 64MiB) */
630     if (x86ms->below_4g_mem_size > 1 * MiB) {
631         val = (x86ms->below_4g_mem_size - 1 * MiB) / KiB;
632     } else {
633         val = 0;
634     }
635     if (val > 65535)
636         val = 65535;
637     rtc_set_memory(s, 0x17, val);
638     rtc_set_memory(s, 0x18, val >> 8);
639     rtc_set_memory(s, 0x30, val);
640     rtc_set_memory(s, 0x31, val >> 8);
641     /* memory between 16MiB and 4GiB */
642     if (x86ms->below_4g_mem_size > 16 * MiB) {
643         val = (x86ms->below_4g_mem_size - 16 * MiB) / (64 * KiB);
644     } else {
645         val = 0;
646     }
647     if (val > 65535)
648         val = 65535;
649     rtc_set_memory(s, 0x34, val);
650     rtc_set_memory(s, 0x35, val >> 8);
651     /* memory above 4GiB */
652     val = x86ms->above_4g_mem_size / 65536;
653     rtc_set_memory(s, 0x5b, val);
654     rtc_set_memory(s, 0x5c, val >> 8);
655     rtc_set_memory(s, 0x5d, val >> 16);
656 
657     object_property_add_link(OBJECT(pcms), "rtc_state",
658                              TYPE_ISA_DEVICE,
659                              (Object **)&x86ms->rtc,
660                              object_property_allow_set_link,
661                              OBJ_PROP_LINK_STRONG, &error_abort);
662     object_property_set_link(OBJECT(pcms), OBJECT(s),
663                              "rtc_state", &error_abort);
664 
665     set_boot_dev(s, MACHINE(pcms)->boot_order, &error_fatal);
666 
667     val = 0;
668     val |= 0x02; /* FPU is there */
669     val |= 0x04; /* PS/2 mouse installed */
670     rtc_set_memory(s, REG_EQUIPMENT_BYTE, val);
671 
672     /* hard drives and FDC */
673     arg.rtc_state = s;
674     arg.idebus[0] = idebus0;
675     arg.idebus[1] = idebus1;
676     qemu_register_reset(pc_cmos_init_late, &arg);
677 }
678 
handle_a20_line_change(void * opaque,int irq,int level)679 static void handle_a20_line_change(void *opaque, int irq, int level)
680 {
681     X86CPU *cpu = opaque;
682 
683     /* XXX: send to all CPUs ? */
684     /* XXX: add logic to handle multiple A20 line sources */
685     x86_cpu_set_a20(cpu, level);
686 }
687 
688 #define NE2000_NB_MAX 6
689 
690 static const int ne2000_io[NE2000_NB_MAX] = { 0x300, 0x320, 0x340, 0x360,
691                                               0x280, 0x380 };
692 static const int ne2000_irq[NE2000_NB_MAX] = { 9, 10, 11, 3, 4, 5 };
693 
pc_init_ne2k_isa(ISABus * bus,NICInfo * nd)694 void pc_init_ne2k_isa(ISABus *bus, NICInfo *nd)
695 {
696     static int nb_ne2k = 0;
697 
698     if (nb_ne2k == NE2000_NB_MAX)
699         return;
700     isa_ne2000_init(bus, ne2000_io[nb_ne2k],
701                     ne2000_irq[nb_ne2k], nd);
702     nb_ne2k++;
703 }
704 
pc_acpi_smi_interrupt(void * opaque,int irq,int level)705 void pc_acpi_smi_interrupt(void *opaque, int irq, int level)
706 {
707     X86CPU *cpu = opaque;
708 
709     if (level) {
710         cpu_interrupt(CPU(cpu), CPU_INTERRUPT_SMI);
711     }
712 }
713 
714 /*
715  * This function is very similar to smp_parse()
716  * in hw/core/machine.c but includes CPU die support.
717  */
pc_smp_parse(MachineState * ms,QemuOpts * opts)718 void pc_smp_parse(MachineState *ms, QemuOpts *opts)
719 {
720     X86MachineState *x86ms = X86_MACHINE(ms);
721 
722     if (opts) {
723         unsigned cpus    = qemu_opt_get_number(opts, "cpus", 0);
724         unsigned sockets = qemu_opt_get_number(opts, "sockets", 0);
725         unsigned dies = qemu_opt_get_number(opts, "dies", 1);
726         unsigned cores   = qemu_opt_get_number(opts, "cores", 0);
727         unsigned threads = qemu_opt_get_number(opts, "threads", 0);
728 
729         /* compute missing values, prefer sockets over cores over threads */
730         if (cpus == 0 || sockets == 0) {
731             cores = cores > 0 ? cores : 1;
732             threads = threads > 0 ? threads : 1;
733             if (cpus == 0) {
734                 sockets = sockets > 0 ? sockets : 1;
735                 cpus = cores * threads * dies * sockets;
736             } else {
737                 ms->smp.max_cpus =
738                         qemu_opt_get_number(opts, "maxcpus", cpus);
739                 sockets = ms->smp.max_cpus / (cores * threads * dies);
740             }
741         } else if (cores == 0) {
742             threads = threads > 0 ? threads : 1;
743             cores = cpus / (sockets * dies * threads);
744             cores = cores > 0 ? cores : 1;
745         } else if (threads == 0) {
746             threads = cpus / (cores * dies * sockets);
747             threads = threads > 0 ? threads : 1;
748         } else if (sockets * dies * cores * threads < cpus) {
749             error_report("cpu topology: "
750                          "sockets (%u) * dies (%u) * cores (%u) * threads (%u) < "
751                          "smp_cpus (%u)",
752                          sockets, dies, cores, threads, cpus);
753             exit(1);
754         }
755 
756         ms->smp.max_cpus =
757                 qemu_opt_get_number(opts, "maxcpus", cpus);
758 
759         if (ms->smp.max_cpus < cpus) {
760             error_report("maxcpus must be equal to or greater than smp");
761             exit(1);
762         }
763 
764         if (sockets * dies * cores * threads > ms->smp.max_cpus) {
765             error_report("cpu topology: "
766                          "sockets (%u) * dies (%u) * cores (%u) * threads (%u) > "
767                          "maxcpus (%u)",
768                          sockets, dies, cores, threads,
769                          ms->smp.max_cpus);
770             exit(1);
771         }
772 
773         if (sockets * dies * cores * threads != ms->smp.max_cpus) {
774             warn_report("Invalid CPU topology deprecated: "
775                         "sockets (%u) * dies (%u) * cores (%u) * threads (%u) "
776                         "!= maxcpus (%u)",
777                         sockets, dies, cores, threads,
778                         ms->smp.max_cpus);
779         }
780 
781         ms->smp.cpus = cpus;
782         ms->smp.cores = cores;
783         ms->smp.threads = threads;
784         ms->smp.sockets = sockets;
785         x86ms->smp_dies = dies;
786     }
787 
788     if (ms->smp.cpus > 1) {
789         Error *blocker = NULL;
790         error_setg(&blocker, QERR_REPLAY_NOT_SUPPORTED, "smp");
791         replay_add_blocker(blocker);
792     }
793 }
794 
pc_hot_add_cpu(MachineState * ms,const int64_t id,Error ** errp)795 void pc_hot_add_cpu(MachineState *ms, const int64_t id, Error **errp)
796 {
797     X86MachineState *x86ms = X86_MACHINE(ms);
798     int64_t apic_id = x86_cpu_apic_id_from_index(x86ms, id);
799     Error *local_err = NULL;
800 
801     if (id < 0) {
802         error_setg(errp, "Invalid CPU id: %" PRIi64, id);
803         return;
804     }
805 
806     if (apic_id >= ACPI_CPU_HOTPLUG_ID_LIMIT) {
807         error_setg(errp, "Unable to add CPU: %" PRIi64
808                    ", resulting APIC ID (%" PRIi64 ") is too large",
809                    id, apic_id);
810         return;
811     }
812 
813 
814     x86_cpu_new(X86_MACHINE(ms), apic_id, &local_err);
815     if (local_err) {
816         error_propagate(errp, local_err);
817         return;
818     }
819 }
820 
rtc_set_cpus_count(ISADevice * rtc,uint16_t cpus_count)821 static void rtc_set_cpus_count(ISADevice *rtc, uint16_t cpus_count)
822 {
823     if (cpus_count > 0xff) {
824         /* If the number of CPUs can't be represented in 8 bits, the
825          * BIOS must use "FW_CFG_NB_CPUS". Set RTC field to 0 just
826          * to make old BIOSes fail more predictably.
827          */
828         rtc_set_memory(rtc, 0x5f, 0);
829     } else {
830         rtc_set_memory(rtc, 0x5f, cpus_count - 1);
831     }
832 }
833 
834 static
pc_machine_done(Notifier * notifier,void * data)835 void pc_machine_done(Notifier *notifier, void *data)
836 {
837     PCMachineState *pcms = container_of(notifier,
838                                         PCMachineState, machine_done);
839     X86MachineState *x86ms = X86_MACHINE(pcms);
840     PCIBus *bus = pcms->bus;
841 
842     /* set the number of CPUs */
843     rtc_set_cpus_count(x86ms->rtc, x86ms->boot_cpus);
844 
845     if (bus) {
846         int extra_hosts = 0;
847 
848         QLIST_FOREACH(bus, &bus->child, sibling) {
849             /* look for expander root buses */
850             if (pci_bus_is_root(bus)) {
851                 extra_hosts++;
852             }
853         }
854         if (extra_hosts && x86ms->fw_cfg) {
855             uint64_t *val = g_malloc(sizeof(*val));
856             *val = cpu_to_le64(extra_hosts);
857             fw_cfg_add_file(x86ms->fw_cfg,
858                     "etc/extra-pci-roots", val, sizeof(*val));
859         }
860     }
861 
862     acpi_setup();
863     if (x86ms->fw_cfg) {
864         fw_cfg_build_smbios(MACHINE(pcms), x86ms->fw_cfg);
865         fw_cfg_build_feature_control(MACHINE(pcms), x86ms->fw_cfg);
866         /* update FW_CFG_NB_CPUS to account for -device added CPUs */
867         fw_cfg_modify_i16(x86ms->fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
868     }
869 
870     if (x86ms->apic_id_limit > 255 && !xen_enabled()) {
871         IntelIOMMUState *iommu = INTEL_IOMMU_DEVICE(x86_iommu_get_default());
872 
873         if (!iommu || !x86_iommu_ir_supported(X86_IOMMU_DEVICE(iommu)) ||
874             iommu->intr_eim != ON_OFF_AUTO_ON) {
875             error_report("current -smp configuration requires "
876                          "Extended Interrupt Mode enabled. "
877                          "You can add an IOMMU using: "
878                          "-device intel-iommu,intremap=on,eim=on");
879             exit(EXIT_FAILURE);
880         }
881     }
882 }
883 
pc_guest_info_init(PCMachineState * pcms)884 void pc_guest_info_init(PCMachineState *pcms)
885 {
886     int i;
887     MachineState *ms = MACHINE(pcms);
888     X86MachineState *x86ms = X86_MACHINE(pcms);
889 
890     x86ms->apic_xrupt_override = kvm_allows_irq0_override();
891     pcms->numa_nodes = ms->numa_state->num_nodes;
892     pcms->node_mem = g_malloc0(pcms->numa_nodes *
893                                     sizeof *pcms->node_mem);
894     for (i = 0; i < ms->numa_state->num_nodes; i++) {
895         pcms->node_mem[i] = ms->numa_state->nodes[i].node_mem;
896     }
897 
898     pcms->machine_done.notify = pc_machine_done;
899     qemu_add_machine_init_done_notifier(&pcms->machine_done);
900 }
901 
902 /* setup pci memory address space mapping into system address space */
pc_pci_as_mapping_init(Object * owner,MemoryRegion * system_memory,MemoryRegion * pci_address_space)903 void pc_pci_as_mapping_init(Object *owner, MemoryRegion *system_memory,
904                             MemoryRegion *pci_address_space)
905 {
906     /* Set to lower priority than RAM */
907     memory_region_add_subregion_overlap(system_memory, 0x0,
908                                         pci_address_space, -1);
909 }
910 
xen_load_linux(PCMachineState * pcms)911 void xen_load_linux(PCMachineState *pcms)
912 {
913     int i;
914     FWCfgState *fw_cfg;
915     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
916     X86MachineState *x86ms = X86_MACHINE(pcms);
917 
918     assert(MACHINE(pcms)->kernel_filename != NULL);
919 
920     fw_cfg = fw_cfg_init_io(FW_CFG_IO_BASE);
921     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
922     rom_set_fw(fw_cfg);
923 
924     x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
925                    pcmc->pvh_enabled, pcmc->linuxboot_dma_enabled);
926     for (i = 0; i < nb_option_roms; i++) {
927         assert(!strcmp(option_rom[i].name, "linuxboot.bin") ||
928                !strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
929                !strcmp(option_rom[i].name, "pvh.bin") ||
930                !strcmp(option_rom[i].name, "multiboot.bin"));
931         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
932     }
933     x86ms->fw_cfg = fw_cfg;
934 }
935 
pc_memory_init(PCMachineState * pcms,MemoryRegion * system_memory,MemoryRegion * rom_memory,MemoryRegion ** ram_memory)936 void pc_memory_init(PCMachineState *pcms,
937                     MemoryRegion *system_memory,
938                     MemoryRegion *rom_memory,
939                     MemoryRegion **ram_memory)
940 {
941     int linux_boot, i;
942     MemoryRegion *option_rom_mr;
943     MemoryRegion *ram_below_4g, *ram_above_4g;
944     FWCfgState *fw_cfg;
945     MachineState *machine = MACHINE(pcms);
946     MachineClass *mc = MACHINE_GET_CLASS(machine);
947     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
948     X86MachineState *x86ms = X86_MACHINE(pcms);
949 
950     assert(machine->ram_size == x86ms->below_4g_mem_size +
951                                 x86ms->above_4g_mem_size);
952 
953     linux_boot = (machine->kernel_filename != NULL);
954 
955     /*
956      * Split single memory region and use aliases to address portions of it,
957      * done for backwards compatibility with older qemus.
958      */
959     *ram_memory = machine->ram;
960     ram_below_4g = g_malloc(sizeof(*ram_below_4g));
961     memory_region_init_alias(ram_below_4g, NULL, "ram-below-4g", machine->ram,
962                              0, x86ms->below_4g_mem_size);
963     memory_region_add_subregion(system_memory, 0, ram_below_4g);
964     e820_add_entry(0, x86ms->below_4g_mem_size, E820_RAM);
965     if (x86ms->above_4g_mem_size > 0) {
966         ram_above_4g = g_malloc(sizeof(*ram_above_4g));
967         memory_region_init_alias(ram_above_4g, NULL, "ram-above-4g",
968                                  machine->ram,
969                                  x86ms->below_4g_mem_size,
970                                  x86ms->above_4g_mem_size);
971         memory_region_add_subregion(system_memory, 0x100000000ULL,
972                                     ram_above_4g);
973         e820_add_entry(0x100000000ULL, x86ms->above_4g_mem_size, E820_RAM);
974     }
975 
976     if (!pcmc->has_reserved_memory &&
977         (machine->ram_slots ||
978          (machine->maxram_size > machine->ram_size))) {
979 
980         error_report("\"-memory 'slots|maxmem'\" is not supported by: %s",
981                      mc->name);
982         exit(EXIT_FAILURE);
983     }
984 
985     /* always allocate the device memory information */
986     machine->device_memory = g_malloc0(sizeof(*machine->device_memory));
987 
988     /* initialize device memory address space */
989     if (pcmc->has_reserved_memory &&
990         (machine->ram_size < machine->maxram_size)) {
991         ram_addr_t device_mem_size = machine->maxram_size - machine->ram_size;
992 
993         if (machine->ram_slots > ACPI_MAX_RAM_SLOTS) {
994             error_report("unsupported amount of memory slots: %"PRIu64,
995                          machine->ram_slots);
996             exit(EXIT_FAILURE);
997         }
998 
999         if (QEMU_ALIGN_UP(machine->maxram_size,
1000                           TARGET_PAGE_SIZE) != machine->maxram_size) {
1001             error_report("maximum memory size must by aligned to multiple of "
1002                          "%d bytes", TARGET_PAGE_SIZE);
1003             exit(EXIT_FAILURE);
1004         }
1005 
1006         machine->device_memory->base =
1007             ROUND_UP(0x100000000ULL + x86ms->above_4g_mem_size, 1 * GiB);
1008 
1009         if (pcmc->enforce_aligned_dimm) {
1010             /* size device region assuming 1G page max alignment per slot */
1011             device_mem_size += (1 * GiB) * machine->ram_slots;
1012         }
1013 
1014         if ((machine->device_memory->base + device_mem_size) <
1015             device_mem_size) {
1016             error_report("unsupported amount of maximum memory: " RAM_ADDR_FMT,
1017                          machine->maxram_size);
1018             exit(EXIT_FAILURE);
1019         }
1020 
1021         memory_region_init(&machine->device_memory->mr, OBJECT(pcms),
1022                            "device-memory", device_mem_size);
1023         memory_region_add_subregion(system_memory, machine->device_memory->base,
1024                                     &machine->device_memory->mr);
1025     }
1026 
1027     /* Initialize PC system firmware */
1028     pc_system_firmware_init(pcms, rom_memory);
1029 
1030     option_rom_mr = g_malloc(sizeof(*option_rom_mr));
1031     memory_region_init_ram(option_rom_mr, NULL, "pc.rom", PC_ROM_SIZE,
1032                            &error_fatal);
1033     if (pcmc->pci_enabled) {
1034         memory_region_set_readonly(option_rom_mr, true);
1035     }
1036     memory_region_add_subregion_overlap(rom_memory,
1037                                         PC_ROM_MIN_VGA,
1038                                         option_rom_mr,
1039                                         1);
1040 
1041     fw_cfg = fw_cfg_arch_create(machine,
1042                                 x86ms->boot_cpus, x86ms->apic_id_limit);
1043 
1044     rom_set_fw(fw_cfg);
1045 
1046     if (pcmc->has_reserved_memory && machine->device_memory->base) {
1047         uint64_t *val = g_malloc(sizeof(*val));
1048         PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1049         uint64_t res_mem_end = machine->device_memory->base;
1050 
1051         if (!pcmc->broken_reserved_end) {
1052             res_mem_end += memory_region_size(&machine->device_memory->mr);
1053         }
1054         *val = cpu_to_le64(ROUND_UP(res_mem_end, 1 * GiB));
1055         fw_cfg_add_file(fw_cfg, "etc/reserved-memory-end", val, sizeof(*val));
1056     }
1057 
1058     if (linux_boot) {
1059         x86_load_linux(x86ms, fw_cfg, pcmc->acpi_data_size,
1060                        pcmc->pvh_enabled, pcmc->linuxboot_dma_enabled);
1061     }
1062 
1063     for (i = 0; i < nb_option_roms; i++) {
1064         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
1065     }
1066     x86ms->fw_cfg = fw_cfg;
1067 
1068     /* Init default IOAPIC address space */
1069     x86ms->ioapic_as = &address_space_memory;
1070 
1071     /* Init ACPI memory hotplug IO base address */
1072     pcms->memhp_io_base = ACPI_MEMORY_HOTPLUG_BASE;
1073 }
1074 
1075 /*
1076  * The 64bit pci hole starts after "above 4G RAM" and
1077  * potentially the space reserved for memory hotplug.
1078  */
pc_pci_hole64_start(void)1079 uint64_t pc_pci_hole64_start(void)
1080 {
1081     PCMachineState *pcms = PC_MACHINE(qdev_get_machine());
1082     PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1083     MachineState *ms = MACHINE(pcms);
1084     X86MachineState *x86ms = X86_MACHINE(pcms);
1085     uint64_t hole64_start = 0;
1086 
1087     if (pcmc->has_reserved_memory && ms->device_memory->base) {
1088         hole64_start = ms->device_memory->base;
1089         if (!pcmc->broken_reserved_end) {
1090             hole64_start += memory_region_size(&ms->device_memory->mr);
1091         }
1092     } else {
1093         hole64_start = 0x100000000ULL + x86ms->above_4g_mem_size;
1094     }
1095 
1096     return ROUND_UP(hole64_start, 1 * GiB);
1097 }
1098 
pc_vga_init(ISABus * isa_bus,PCIBus * pci_bus)1099 DeviceState *pc_vga_init(ISABus *isa_bus, PCIBus *pci_bus)
1100 {
1101     DeviceState *dev = NULL;
1102 
1103     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_VGA);
1104     if (pci_bus) {
1105         PCIDevice *pcidev = pci_vga_init(pci_bus);
1106         dev = pcidev ? &pcidev->qdev : NULL;
1107     } else if (isa_bus) {
1108         ISADevice *isadev = isa_vga_init(isa_bus);
1109         dev = isadev ? DEVICE(isadev) : NULL;
1110     }
1111     rom_reset_order_override();
1112     return dev;
1113 }
1114 
1115 static const MemoryRegionOps ioport80_io_ops = {
1116     .write = ioport80_write,
1117     .read = ioport80_read,
1118     .endianness = DEVICE_NATIVE_ENDIAN,
1119     .impl = {
1120         .min_access_size = 1,
1121         .max_access_size = 1,
1122     },
1123 };
1124 
1125 static const MemoryRegionOps ioportF0_io_ops = {
1126     .write = ioportF0_write,
1127     .read = ioportF0_read,
1128     .endianness = DEVICE_NATIVE_ENDIAN,
1129     .impl = {
1130         .min_access_size = 1,
1131         .max_access_size = 1,
1132     },
1133 };
1134 
pc_superio_init(ISABus * isa_bus,bool create_fdctrl,bool no_vmport)1135 static void pc_superio_init(ISABus *isa_bus, bool create_fdctrl, bool no_vmport)
1136 {
1137     int i;
1138     DriveInfo *fd[MAX_FD];
1139     qemu_irq *a20_line;
1140     ISADevice *i8042, *port92, *vmmouse;
1141 
1142     serial_hds_isa_init(isa_bus, 0, MAX_ISA_SERIAL_PORTS);
1143     parallel_hds_isa_init(isa_bus, MAX_PARALLEL_PORTS);
1144 
1145     for (i = 0; i < MAX_FD; i++) {
1146         fd[i] = drive_get(IF_FLOPPY, 0, i);
1147         create_fdctrl |= !!fd[i];
1148     }
1149     if (create_fdctrl) {
1150         fdctrl_init_isa(isa_bus, fd);
1151     }
1152 
1153     i8042 = isa_create_simple(isa_bus, "i8042");
1154     if (!no_vmport) {
1155         vmport_init(isa_bus);
1156         vmmouse = isa_try_create(isa_bus, "vmmouse");
1157     } else {
1158         vmmouse = NULL;
1159     }
1160     if (vmmouse) {
1161         object_property_set_link(OBJECT(vmmouse), OBJECT(i8042),
1162                                  "i8042", &error_abort);
1163         qdev_init_nofail(DEVICE(vmmouse));
1164     }
1165     port92 = isa_create_simple(isa_bus, TYPE_PORT92);
1166 
1167     a20_line = qemu_allocate_irqs(handle_a20_line_change, first_cpu, 2);
1168     i8042_setup_a20_line(i8042, a20_line[0]);
1169     qdev_connect_gpio_out_named(DEVICE(port92),
1170                                 PORT92_A20_LINE, 0, a20_line[1]);
1171     g_free(a20_line);
1172 }
1173 
pc_basic_device_init(ISABus * isa_bus,qemu_irq * gsi,ISADevice ** rtc_state,bool create_fdctrl,bool no_vmport,bool has_pit,uint32_t hpet_irqs)1174 void pc_basic_device_init(ISABus *isa_bus, qemu_irq *gsi,
1175                           ISADevice **rtc_state,
1176                           bool create_fdctrl,
1177                           bool no_vmport,
1178                           bool has_pit,
1179                           uint32_t hpet_irqs)
1180 {
1181     int i;
1182     DeviceState *hpet = NULL;
1183     int pit_isa_irq = 0;
1184     qemu_irq pit_alt_irq = NULL;
1185     qemu_irq rtc_irq = NULL;
1186     ISADevice *pit = NULL;
1187     MemoryRegion *ioport80_io = g_new(MemoryRegion, 1);
1188     MemoryRegion *ioportF0_io = g_new(MemoryRegion, 1);
1189 
1190     memory_region_init_io(ioport80_io, NULL, &ioport80_io_ops, NULL, "ioport80", 1);
1191     memory_region_add_subregion(isa_bus->address_space_io, 0x80, ioport80_io);
1192 
1193     memory_region_init_io(ioportF0_io, NULL, &ioportF0_io_ops, NULL, "ioportF0", 1);
1194     memory_region_add_subregion(isa_bus->address_space_io, 0xf0, ioportF0_io);
1195 
1196     /*
1197      * Check if an HPET shall be created.
1198      *
1199      * Without KVM_CAP_PIT_STATE2, we cannot switch off the in-kernel PIT
1200      * when the HPET wants to take over. Thus we have to disable the latter.
1201      */
1202     if (!no_hpet && (!kvm_irqchip_in_kernel() || kvm_has_pit_state2())) {
1203         hpet = qdev_try_create(NULL, TYPE_HPET);
1204         if (hpet) {
1205             /* For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-1.7
1206              * and earlier, use IRQ2 for compat. Otherwise, use IRQ16~23,
1207              * IRQ8 and IRQ2.
1208              */
1209             uint8_t compat = object_property_get_uint(OBJECT(hpet),
1210                     HPET_INTCAP, NULL);
1211             if (!compat) {
1212                 qdev_prop_set_uint32(hpet, HPET_INTCAP, hpet_irqs);
1213             }
1214             qdev_init_nofail(hpet);
1215             sysbus_mmio_map(SYS_BUS_DEVICE(hpet), 0, HPET_BASE);
1216 
1217             for (i = 0; i < GSI_NUM_PINS; i++) {
1218                 sysbus_connect_irq(SYS_BUS_DEVICE(hpet), i, gsi[i]);
1219             }
1220             pit_isa_irq = -1;
1221             pit_alt_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_PIT_INT);
1222             rtc_irq = qdev_get_gpio_in(hpet, HPET_LEGACY_RTC_INT);
1223         }
1224     }
1225     *rtc_state = mc146818_rtc_init(isa_bus, 2000, rtc_irq);
1226 
1227     qemu_register_boot_set(pc_boot_set, *rtc_state);
1228 
1229     if (!xen_enabled() && has_pit) {
1230         if (kvm_pit_in_kernel()) {
1231             pit = kvm_pit_init(isa_bus, 0x40);
1232         } else {
1233             pit = i8254_pit_init(isa_bus, 0x40, pit_isa_irq, pit_alt_irq);
1234         }
1235         if (hpet) {
1236             /* connect PIT to output control line of the HPET */
1237             qdev_connect_gpio_out(hpet, 0, qdev_get_gpio_in(DEVICE(pit), 0));
1238         }
1239         pcspk_init(isa_bus, pit);
1240     }
1241 
1242     i8257_dma_init(isa_bus, 0);
1243 
1244     /* Super I/O */
1245     pc_superio_init(isa_bus, create_fdctrl, no_vmport);
1246 }
1247 
pc_nic_init(PCMachineClass * pcmc,ISABus * isa_bus,PCIBus * pci_bus)1248 void pc_nic_init(PCMachineClass *pcmc, ISABus *isa_bus, PCIBus *pci_bus)
1249 {
1250     int i;
1251 
1252     rom_set_order_override(FW_CFG_ORDER_OVERRIDE_NIC);
1253     for (i = 0; i < nb_nics; i++) {
1254         NICInfo *nd = &nd_table[i];
1255         const char *model = nd->model ? nd->model : pcmc->default_nic_model;
1256 
1257         if (g_str_equal(model, "ne2k_isa")) {
1258             pc_init_ne2k_isa(isa_bus, nd);
1259         } else {
1260             pci_nic_init_nofail(nd, pci_bus, model, NULL);
1261         }
1262     }
1263     rom_reset_order_override();
1264 }
1265 
pc_i8259_create(ISABus * isa_bus,qemu_irq * i8259_irqs)1266 void pc_i8259_create(ISABus *isa_bus, qemu_irq *i8259_irqs)
1267 {
1268     qemu_irq *i8259;
1269 
1270     if (kvm_pic_in_kernel()) {
1271         i8259 = kvm_i8259_init(isa_bus);
1272     } else if (xen_enabled()) {
1273         i8259 = xen_interrupt_controller_init();
1274     } else {
1275         i8259 = i8259_init(isa_bus, x86_allocate_cpu_irq());
1276     }
1277 
1278     for (size_t i = 0; i < ISA_NUM_IRQS; i++) {
1279         i8259_irqs[i] = i8259[i];
1280     }
1281 
1282     g_free(i8259);
1283 }
1284 
pc_memory_pre_plug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1285 static void pc_memory_pre_plug(HotplugHandler *hotplug_dev, DeviceState *dev,
1286                                Error **errp)
1287 {
1288     const PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1289     const PCMachineClass *pcmc = PC_MACHINE_GET_CLASS(pcms);
1290     const MachineState *ms = MACHINE(hotplug_dev);
1291     const bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1292     const uint64_t legacy_align = TARGET_PAGE_SIZE;
1293     Error *local_err = NULL;
1294 
1295     /*
1296      * When -no-acpi is used with Q35 machine type, no ACPI is built,
1297      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1298      * addition to cover this case.
1299      */
1300     if (!pcms->acpi_dev || !acpi_enabled) {
1301         error_setg(errp,
1302                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1303         return;
1304     }
1305 
1306     if (is_nvdimm && !ms->nvdimms_state->is_enabled) {
1307         error_setg(errp, "nvdimm is not enabled: missing 'nvdimm' in '-M'");
1308         return;
1309     }
1310 
1311     hotplug_handler_pre_plug(pcms->acpi_dev, dev, &local_err);
1312     if (local_err) {
1313         error_propagate(errp, local_err);
1314         return;
1315     }
1316 
1317     pc_dimm_pre_plug(PC_DIMM(dev), MACHINE(hotplug_dev),
1318                      pcmc->enforce_aligned_dimm ? NULL : &legacy_align, errp);
1319 }
1320 
pc_memory_plug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1321 static void pc_memory_plug(HotplugHandler *hotplug_dev,
1322                            DeviceState *dev, Error **errp)
1323 {
1324     Error *local_err = NULL;
1325     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1326     MachineState *ms = MACHINE(hotplug_dev);
1327     bool is_nvdimm = object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM);
1328 
1329     pc_dimm_plug(PC_DIMM(dev), MACHINE(pcms), &local_err);
1330     if (local_err) {
1331         goto out;
1332     }
1333 
1334     if (is_nvdimm) {
1335         nvdimm_plug(ms->nvdimms_state);
1336     }
1337 
1338     hotplug_handler_plug(HOTPLUG_HANDLER(pcms->acpi_dev), dev, &error_abort);
1339 out:
1340     error_propagate(errp, local_err);
1341 }
1342 
pc_memory_unplug_request(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1343 static void pc_memory_unplug_request(HotplugHandler *hotplug_dev,
1344                                      DeviceState *dev, Error **errp)
1345 {
1346     Error *local_err = NULL;
1347     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1348 
1349     /*
1350      * When -no-acpi is used with Q35 machine type, no ACPI is built,
1351      * but pcms->acpi_dev is still created. Check !acpi_enabled in
1352      * addition to cover this case.
1353      */
1354     if (!pcms->acpi_dev || !acpi_enabled) {
1355         error_setg(&local_err,
1356                    "memory hotplug is not enabled: missing acpi device or acpi disabled");
1357         goto out;
1358     }
1359 
1360     if (object_dynamic_cast(OBJECT(dev), TYPE_NVDIMM)) {
1361         error_setg(&local_err,
1362                    "nvdimm device hot unplug is not supported yet.");
1363         goto out;
1364     }
1365 
1366     hotplug_handler_unplug_request(HOTPLUG_HANDLER(pcms->acpi_dev), dev,
1367                                    &local_err);
1368 out:
1369     error_propagate(errp, local_err);
1370 }
1371 
pc_memory_unplug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1372 static void pc_memory_unplug(HotplugHandler *hotplug_dev,
1373                              DeviceState *dev, Error **errp)
1374 {
1375     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1376     Error *local_err = NULL;
1377 
1378     hotplug_handler_unplug(HOTPLUG_HANDLER(pcms->acpi_dev), dev, &local_err);
1379     if (local_err) {
1380         goto out;
1381     }
1382 
1383     pc_dimm_unplug(PC_DIMM(dev), MACHINE(pcms));
1384     object_property_set_bool(OBJECT(dev), false, "realized", NULL);
1385  out:
1386     error_propagate(errp, local_err);
1387 }
1388 
pc_apic_cmp(const void * a,const void * b)1389 static int pc_apic_cmp(const void *a, const void *b)
1390 {
1391    CPUArchId *apic_a = (CPUArchId *)a;
1392    CPUArchId *apic_b = (CPUArchId *)b;
1393 
1394    return apic_a->arch_id - apic_b->arch_id;
1395 }
1396 
1397 /* returns pointer to CPUArchId descriptor that matches CPU's apic_id
1398  * in ms->possible_cpus->cpus, if ms->possible_cpus->cpus has no
1399  * entry corresponding to CPU's apic_id returns NULL.
1400  */
pc_find_cpu_slot(MachineState * ms,uint32_t id,int * idx)1401 static CPUArchId *pc_find_cpu_slot(MachineState *ms, uint32_t id, int *idx)
1402 {
1403     CPUArchId apic_id, *found_cpu;
1404 
1405     apic_id.arch_id = id;
1406     found_cpu = bsearch(&apic_id, ms->possible_cpus->cpus,
1407         ms->possible_cpus->len, sizeof(*ms->possible_cpus->cpus),
1408         pc_apic_cmp);
1409     if (found_cpu && idx) {
1410         *idx = found_cpu - ms->possible_cpus->cpus;
1411     }
1412     return found_cpu;
1413 }
1414 
pc_cpu_plug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1415 static void pc_cpu_plug(HotplugHandler *hotplug_dev,
1416                         DeviceState *dev, Error **errp)
1417 {
1418     CPUArchId *found_cpu;
1419     Error *local_err = NULL;
1420     X86CPU *cpu = X86_CPU(dev);
1421     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1422     X86MachineState *x86ms = X86_MACHINE(pcms);
1423 
1424     if (pcms->acpi_dev) {
1425         hotplug_handler_plug(HOTPLUG_HANDLER(pcms->acpi_dev), dev, &local_err);
1426         if (local_err) {
1427             goto out;
1428         }
1429     }
1430 
1431     /* increment the number of CPUs */
1432     x86ms->boot_cpus++;
1433     if (x86ms->rtc) {
1434         rtc_set_cpus_count(x86ms->rtc, x86ms->boot_cpus);
1435     }
1436     if (x86ms->fw_cfg) {
1437         fw_cfg_modify_i16(x86ms->fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
1438     }
1439 
1440     found_cpu = pc_find_cpu_slot(MACHINE(pcms), cpu->apic_id, NULL);
1441     found_cpu->cpu = OBJECT(dev);
1442 out:
1443     error_propagate(errp, local_err);
1444 }
pc_cpu_unplug_request_cb(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1445 static void pc_cpu_unplug_request_cb(HotplugHandler *hotplug_dev,
1446                                      DeviceState *dev, Error **errp)
1447 {
1448     int idx = -1;
1449     Error *local_err = NULL;
1450     X86CPU *cpu = X86_CPU(dev);
1451     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1452 
1453     if (!pcms->acpi_dev) {
1454         error_setg(&local_err, "CPU hot unplug not supported without ACPI");
1455         goto out;
1456     }
1457 
1458     pc_find_cpu_slot(MACHINE(pcms), cpu->apic_id, &idx);
1459     assert(idx != -1);
1460     if (idx == 0) {
1461         error_setg(&local_err, "Boot CPU is unpluggable");
1462         goto out;
1463     }
1464 
1465     hotplug_handler_unplug_request(HOTPLUG_HANDLER(pcms->acpi_dev), dev,
1466                                    &local_err);
1467     if (local_err) {
1468         goto out;
1469     }
1470 
1471  out:
1472     error_propagate(errp, local_err);
1473 
1474 }
1475 
pc_cpu_unplug_cb(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1476 static void pc_cpu_unplug_cb(HotplugHandler *hotplug_dev,
1477                              DeviceState *dev, Error **errp)
1478 {
1479     CPUArchId *found_cpu;
1480     Error *local_err = NULL;
1481     X86CPU *cpu = X86_CPU(dev);
1482     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1483     X86MachineState *x86ms = X86_MACHINE(pcms);
1484 
1485     hotplug_handler_unplug(HOTPLUG_HANDLER(pcms->acpi_dev), dev, &local_err);
1486     if (local_err) {
1487         goto out;
1488     }
1489 
1490     found_cpu = pc_find_cpu_slot(MACHINE(pcms), cpu->apic_id, NULL);
1491     found_cpu->cpu = NULL;
1492     object_property_set_bool(OBJECT(dev), false, "realized", NULL);
1493 
1494     /* decrement the number of CPUs */
1495     x86ms->boot_cpus--;
1496     /* Update the number of CPUs in CMOS */
1497     rtc_set_cpus_count(x86ms->rtc, x86ms->boot_cpus);
1498     fw_cfg_modify_i16(x86ms->fw_cfg, FW_CFG_NB_CPUS, x86ms->boot_cpus);
1499  out:
1500     error_propagate(errp, local_err);
1501 }
1502 
pc_cpu_pre_plug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1503 static void pc_cpu_pre_plug(HotplugHandler *hotplug_dev,
1504                             DeviceState *dev, Error **errp)
1505 {
1506     int idx;
1507     CPUState *cs;
1508     CPUArchId *cpu_slot;
1509     X86CPUTopoIDs topo_ids;
1510     X86CPU *cpu = X86_CPU(dev);
1511     CPUX86State *env = &cpu->env;
1512     MachineState *ms = MACHINE(hotplug_dev);
1513     PCMachineState *pcms = PC_MACHINE(hotplug_dev);
1514     X86MachineState *x86ms = X86_MACHINE(pcms);
1515     unsigned int smp_cores = ms->smp.cores;
1516     unsigned int smp_threads = ms->smp.threads;
1517     X86CPUTopoInfo topo_info;
1518 
1519     if(!object_dynamic_cast(OBJECT(cpu), ms->cpu_type)) {
1520         error_setg(errp, "Invalid CPU type, expected cpu type: '%s'",
1521                    ms->cpu_type);
1522         return;
1523     }
1524 
1525     init_topo_info(&topo_info, x86ms);
1526 
1527     env->nr_dies = x86ms->smp_dies;
1528     env->nr_nodes = topo_info.nodes_per_pkg;
1529 
1530     /*
1531      * If APIC ID is not set,
1532      * set it based on socket/die/core/thread properties.
1533      */
1534     if (cpu->apic_id == UNASSIGNED_APIC_ID) {
1535         int max_socket = (ms->smp.max_cpus - 1) /
1536                                 smp_threads / smp_cores / x86ms->smp_dies;
1537 
1538         /*
1539          * die-id was optional in QEMU 4.0 and older, so keep it optional
1540          * if there's only one die per socket.
1541          */
1542         if (cpu->die_id < 0 && x86ms->smp_dies == 1) {
1543             cpu->die_id = 0;
1544         }
1545 
1546         if (cpu->socket_id < 0) {
1547             error_setg(errp, "CPU socket-id is not set");
1548             return;
1549         } else if (cpu->socket_id > max_socket) {
1550             error_setg(errp, "Invalid CPU socket-id: %u must be in range 0:%u",
1551                        cpu->socket_id, max_socket);
1552             return;
1553         }
1554         if (cpu->die_id < 0) {
1555             error_setg(errp, "CPU die-id is not set");
1556             return;
1557         } else if (cpu->die_id > x86ms->smp_dies - 1) {
1558             error_setg(errp, "Invalid CPU die-id: %u must be in range 0:%u",
1559                        cpu->die_id, x86ms->smp_dies - 1);
1560             return;
1561         }
1562         if (cpu->core_id < 0) {
1563             error_setg(errp, "CPU core-id is not set");
1564             return;
1565         } else if (cpu->core_id > (smp_cores - 1)) {
1566             error_setg(errp, "Invalid CPU core-id: %u must be in range 0:%u",
1567                        cpu->core_id, smp_cores - 1);
1568             return;
1569         }
1570         if (cpu->thread_id < 0) {
1571             error_setg(errp, "CPU thread-id is not set");
1572             return;
1573         } else if (cpu->thread_id > (smp_threads - 1)) {
1574             error_setg(errp, "Invalid CPU thread-id: %u must be in range 0:%u",
1575                        cpu->thread_id, smp_threads - 1);
1576             return;
1577         }
1578 
1579         topo_ids.pkg_id = cpu->socket_id;
1580         topo_ids.die_id = cpu->die_id;
1581         topo_ids.core_id = cpu->core_id;
1582         topo_ids.smt_id = cpu->thread_id;
1583         cpu->apic_id = x86_apicid_from_topo_ids(&topo_info, &topo_ids);
1584     }
1585 
1586     cpu_slot = pc_find_cpu_slot(MACHINE(pcms), cpu->apic_id, &idx);
1587     if (!cpu_slot) {
1588         MachineState *ms = MACHINE(pcms);
1589 
1590         x86_topo_ids_from_apicid(cpu->apic_id, &topo_info, &topo_ids);
1591         error_setg(errp,
1592             "Invalid CPU [socket: %u, die: %u, core: %u, thread: %u] with"
1593             " APIC ID %" PRIu32 ", valid index range 0:%d",
1594             topo_ids.pkg_id, topo_ids.die_id, topo_ids.core_id, topo_ids.smt_id,
1595             cpu->apic_id, ms->possible_cpus->len - 1);
1596         return;
1597     }
1598 
1599     if (cpu_slot->cpu) {
1600         error_setg(errp, "CPU[%d] with APIC ID %" PRIu32 " exists",
1601                    idx, cpu->apic_id);
1602         return;
1603     }
1604 
1605     /* if 'address' properties socket-id/core-id/thread-id are not set, set them
1606      * so that machine_query_hotpluggable_cpus would show correct values
1607      */
1608     /* TODO: move socket_id/core_id/thread_id checks into x86_cpu_realizefn()
1609      * once -smp refactoring is complete and there will be CPU private
1610      * CPUState::nr_cores and CPUState::nr_threads fields instead of globals */
1611     x86_topo_ids_from_apicid(cpu->apic_id, &topo_info, &topo_ids);
1612     if (cpu->socket_id != -1 && cpu->socket_id != topo_ids.pkg_id) {
1613         error_setg(errp, "property socket-id: %u doesn't match set apic-id:"
1614             " 0x%x (socket-id: %u)", cpu->socket_id, cpu->apic_id,
1615             topo_ids.pkg_id);
1616         return;
1617     }
1618     cpu->socket_id = topo_ids.pkg_id;
1619 
1620     if (cpu->die_id != -1 && cpu->die_id != topo_ids.die_id) {
1621         error_setg(errp, "property die-id: %u doesn't match set apic-id:"
1622             " 0x%x (die-id: %u)", cpu->die_id, cpu->apic_id, topo_ids.die_id);
1623         return;
1624     }
1625     cpu->die_id = topo_ids.die_id;
1626 
1627     if (cpu->core_id != -1 && cpu->core_id != topo_ids.core_id) {
1628         error_setg(errp, "property core-id: %u doesn't match set apic-id:"
1629             " 0x%x (core-id: %u)", cpu->core_id, cpu->apic_id,
1630             topo_ids.core_id);
1631         return;
1632     }
1633     cpu->core_id = topo_ids.core_id;
1634 
1635     if (cpu->thread_id != -1 && cpu->thread_id != topo_ids.smt_id) {
1636         error_setg(errp, "property thread-id: %u doesn't match set apic-id:"
1637             " 0x%x (thread-id: %u)", cpu->thread_id, cpu->apic_id,
1638             topo_ids.smt_id);
1639         return;
1640     }
1641     cpu->thread_id = topo_ids.smt_id;
1642 
1643     if (hyperv_feat_enabled(cpu, HYPERV_FEAT_VPINDEX) &&
1644         !kvm_hv_vpindex_settable()) {
1645         error_setg(errp, "kernel doesn't allow setting HyperV VP_INDEX");
1646         return;
1647     }
1648 
1649     cs = CPU(cpu);
1650     cs->cpu_index = idx;
1651 
1652     numa_cpu_pre_plug(cpu_slot, dev, errp);
1653 }
1654 
pc_virtio_pmem_pci_pre_plug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1655 static void pc_virtio_pmem_pci_pre_plug(HotplugHandler *hotplug_dev,
1656                                         DeviceState *dev, Error **errp)
1657 {
1658     HotplugHandler *hotplug_dev2 = qdev_get_bus_hotplug_handler(dev);
1659     Error *local_err = NULL;
1660 
1661     if (!hotplug_dev2) {
1662         /*
1663          * Without a bus hotplug handler, we cannot control the plug/unplug
1664          * order. This should never be the case on x86, however better add
1665          * a safety net.
1666          */
1667         error_setg(errp, "virtio-pmem-pci not supported on this bus.");
1668         return;
1669     }
1670     /*
1671      * First, see if we can plug this memory device at all. If that
1672      * succeeds, branch of to the actual hotplug handler.
1673      */
1674     memory_device_pre_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev), NULL,
1675                            &local_err);
1676     if (!local_err) {
1677         hotplug_handler_pre_plug(hotplug_dev2, dev, &local_err);
1678     }
1679     error_propagate(errp, local_err);
1680 }
1681 
pc_virtio_pmem_pci_plug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1682 static void pc_virtio_pmem_pci_plug(HotplugHandler *hotplug_dev,
1683                                     DeviceState *dev, Error **errp)
1684 {
1685     HotplugHandler *hotplug_dev2 = qdev_get_bus_hotplug_handler(dev);
1686     Error *local_err = NULL;
1687 
1688     /*
1689      * Plug the memory device first and then branch off to the actual
1690      * hotplug handler. If that one fails, we can easily undo the memory
1691      * device bits.
1692      */
1693     memory_device_plug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev));
1694     hotplug_handler_plug(hotplug_dev2, dev, &local_err);
1695     if (local_err) {
1696         memory_device_unplug(MEMORY_DEVICE(dev), MACHINE(hotplug_dev));
1697     }
1698     error_propagate(errp, local_err);
1699 }
1700 
pc_virtio_pmem_pci_unplug_request(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1701 static void pc_virtio_pmem_pci_unplug_request(HotplugHandler *hotplug_dev,
1702                                               DeviceState *dev, Error **errp)
1703 {
1704     /* We don't support virtio pmem hot unplug */
1705     error_setg(errp, "virtio pmem device unplug not supported.");
1706 }
1707 
pc_virtio_pmem_pci_unplug(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1708 static void pc_virtio_pmem_pci_unplug(HotplugHandler *hotplug_dev,
1709                                       DeviceState *dev, Error **errp)
1710 {
1711     /* We don't support virtio pmem hot unplug */
1712 }
1713 
pc_machine_device_pre_plug_cb(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1714 static void pc_machine_device_pre_plug_cb(HotplugHandler *hotplug_dev,
1715                                           DeviceState *dev, Error **errp)
1716 {
1717     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1718         pc_memory_pre_plug(hotplug_dev, dev, errp);
1719     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1720         pc_cpu_pre_plug(hotplug_dev, dev, errp);
1721     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI)) {
1722         pc_virtio_pmem_pci_pre_plug(hotplug_dev, dev, errp);
1723     }
1724 }
1725 
pc_machine_device_plug_cb(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1726 static void pc_machine_device_plug_cb(HotplugHandler *hotplug_dev,
1727                                       DeviceState *dev, Error **errp)
1728 {
1729     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1730         pc_memory_plug(hotplug_dev, dev, errp);
1731     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1732         pc_cpu_plug(hotplug_dev, dev, errp);
1733     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI)) {
1734         pc_virtio_pmem_pci_plug(hotplug_dev, dev, errp);
1735     }
1736 }
1737 
pc_machine_device_unplug_request_cb(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1738 static void pc_machine_device_unplug_request_cb(HotplugHandler *hotplug_dev,
1739                                                 DeviceState *dev, Error **errp)
1740 {
1741     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1742         pc_memory_unplug_request(hotplug_dev, dev, errp);
1743     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1744         pc_cpu_unplug_request_cb(hotplug_dev, dev, errp);
1745     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI)) {
1746         pc_virtio_pmem_pci_unplug_request(hotplug_dev, dev, errp);
1747     } else {
1748         error_setg(errp, "acpi: device unplug request for not supported device"
1749                    " type: %s", object_get_typename(OBJECT(dev)));
1750     }
1751 }
1752 
pc_machine_device_unplug_cb(HotplugHandler * hotplug_dev,DeviceState * dev,Error ** errp)1753 static void pc_machine_device_unplug_cb(HotplugHandler *hotplug_dev,
1754                                         DeviceState *dev, Error **errp)
1755 {
1756     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM)) {
1757         pc_memory_unplug(hotplug_dev, dev, errp);
1758     } else if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
1759         pc_cpu_unplug_cb(hotplug_dev, dev, errp);
1760     } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI)) {
1761         pc_virtio_pmem_pci_unplug(hotplug_dev, dev, errp);
1762     } else {
1763         error_setg(errp, "acpi: device unplug for not supported device"
1764                    " type: %s", object_get_typename(OBJECT(dev)));
1765     }
1766 }
1767 
pc_get_hotplug_handler(MachineState * machine,DeviceState * dev)1768 static HotplugHandler *pc_get_hotplug_handler(MachineState *machine,
1769                                              DeviceState *dev)
1770 {
1771     if (object_dynamic_cast(OBJECT(dev), TYPE_PC_DIMM) ||
1772         object_dynamic_cast(OBJECT(dev), TYPE_CPU) ||
1773         object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_PMEM_PCI)) {
1774         return HOTPLUG_HANDLER(machine);
1775     }
1776 
1777     return NULL;
1778 }
1779 
1780 static void
pc_machine_get_device_memory_region_size(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1781 pc_machine_get_device_memory_region_size(Object *obj, Visitor *v,
1782                                          const char *name, void *opaque,
1783                                          Error **errp)
1784 {
1785     MachineState *ms = MACHINE(obj);
1786     int64_t value = 0;
1787 
1788     if (ms->device_memory) {
1789         value = memory_region_size(&ms->device_memory->mr);
1790     }
1791 
1792     visit_type_int(v, name, &value, errp);
1793 }
1794 
pc_machine_get_vmport(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1795 static void pc_machine_get_vmport(Object *obj, Visitor *v, const char *name,
1796                                   void *opaque, Error **errp)
1797 {
1798     PCMachineState *pcms = PC_MACHINE(obj);
1799     OnOffAuto vmport = pcms->vmport;
1800 
1801     visit_type_OnOffAuto(v, name, &vmport, errp);
1802 }
1803 
pc_machine_set_vmport(Object * obj,Visitor * v,const char * name,void * opaque,Error ** errp)1804 static void pc_machine_set_vmport(Object *obj, Visitor *v, const char *name,
1805                                   void *opaque, Error **errp)
1806 {
1807     PCMachineState *pcms = PC_MACHINE(obj);
1808 
1809     visit_type_OnOffAuto(v, name, &pcms->vmport, errp);
1810 }
1811 
pc_machine_get_smbus(Object * obj,Error ** errp)1812 static bool pc_machine_get_smbus(Object *obj, Error **errp)
1813 {
1814     PCMachineState *pcms = PC_MACHINE(obj);
1815 
1816     return pcms->smbus_enabled;
1817 }
1818 
pc_machine_set_smbus(Object * obj,bool value,Error ** errp)1819 static void pc_machine_set_smbus(Object *obj, bool value, Error **errp)
1820 {
1821     PCMachineState *pcms = PC_MACHINE(obj);
1822 
1823     pcms->smbus_enabled = value;
1824 }
1825 
pc_machine_get_sata(Object * obj,Error ** errp)1826 static bool pc_machine_get_sata(Object *obj, Error **errp)
1827 {
1828     PCMachineState *pcms = PC_MACHINE(obj);
1829 
1830     return pcms->sata_enabled;
1831 }
1832 
pc_machine_set_sata(Object * obj,bool value,Error ** errp)1833 static void pc_machine_set_sata(Object *obj, bool value, Error **errp)
1834 {
1835     PCMachineState *pcms = PC_MACHINE(obj);
1836 
1837     pcms->sata_enabled = value;
1838 }
1839 
pc_machine_get_pit(Object * obj,Error ** errp)1840 static bool pc_machine_get_pit(Object *obj, Error **errp)
1841 {
1842     PCMachineState *pcms = PC_MACHINE(obj);
1843 
1844     return pcms->pit_enabled;
1845 }
1846 
pc_machine_set_pit(Object * obj,bool value,Error ** errp)1847 static void pc_machine_set_pit(Object *obj, bool value, Error **errp)
1848 {
1849     PCMachineState *pcms = PC_MACHINE(obj);
1850 
1851     pcms->pit_enabled = value;
1852 }
1853 
pc_machine_initfn(Object * obj)1854 static void pc_machine_initfn(Object *obj)
1855 {
1856     PCMachineState *pcms = PC_MACHINE(obj);
1857 
1858 #ifdef CONFIG_VMPORT
1859     pcms->vmport = ON_OFF_AUTO_AUTO;
1860 #else
1861     pcms->vmport = ON_OFF_AUTO_OFF;
1862 #endif /* CONFIG_VMPORT */
1863     /* acpi build is enabled by default if machine supports it */
1864     pcms->acpi_build_enabled = PC_MACHINE_GET_CLASS(pcms)->has_acpi_build;
1865     pcms->smbus_enabled = true;
1866     pcms->sata_enabled = true;
1867     pcms->pit_enabled = true;
1868 
1869     pc_system_flash_create(pcms);
1870 }
1871 
pc_machine_reset(MachineState * machine)1872 static void pc_machine_reset(MachineState *machine)
1873 {
1874     CPUState *cs;
1875     X86CPU *cpu;
1876 
1877     qemu_devices_reset();
1878 
1879     /* Reset APIC after devices have been reset to cancel
1880      * any changes that qemu_devices_reset() might have done.
1881      */
1882     CPU_FOREACH(cs) {
1883         cpu = X86_CPU(cs);
1884 
1885         if (cpu->apic_state) {
1886             device_legacy_reset(cpu->apic_state);
1887         }
1888     }
1889 }
1890 
pc_machine_wakeup(MachineState * machine)1891 static void pc_machine_wakeup(MachineState *machine)
1892 {
1893     cpu_synchronize_all_states();
1894     pc_machine_reset(machine);
1895     cpu_synchronize_all_post_reset();
1896 }
1897 
pc_hotplug_allowed(MachineState * ms,DeviceState * dev,Error ** errp)1898 static bool pc_hotplug_allowed(MachineState *ms, DeviceState *dev, Error **errp)
1899 {
1900     X86IOMMUState *iommu = x86_iommu_get_default();
1901     IntelIOMMUState *intel_iommu;
1902 
1903     if (iommu &&
1904         object_dynamic_cast((Object *)iommu, TYPE_INTEL_IOMMU_DEVICE) &&
1905         object_dynamic_cast((Object *)dev, "vfio-pci")) {
1906         intel_iommu = INTEL_IOMMU_DEVICE(iommu);
1907         if (!intel_iommu->caching_mode) {
1908             error_setg(errp, "Device assignment is not allowed without "
1909                        "enabling caching-mode=on for Intel IOMMU.");
1910             return false;
1911         }
1912     }
1913 
1914     return true;
1915 }
1916 
pc_machine_class_init(ObjectClass * oc,void * data)1917 static void pc_machine_class_init(ObjectClass *oc, void *data)
1918 {
1919     MachineClass *mc = MACHINE_CLASS(oc);
1920     PCMachineClass *pcmc = PC_MACHINE_CLASS(oc);
1921     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
1922 
1923     pcmc->pci_enabled = true;
1924     pcmc->has_acpi_build = true;
1925     pcmc->rsdp_in_ram = true;
1926     pcmc->smbios_defaults = true;
1927     pcmc->smbios_uuid_encoded = true;
1928     pcmc->gigabyte_align = true;
1929     pcmc->has_reserved_memory = true;
1930     pcmc->kvmclock_enabled = true;
1931     pcmc->enforce_aligned_dimm = true;
1932     /* BIOS ACPI tables: 128K. Other BIOS datastructures: less than 4K reported
1933      * to be used at the moment, 32K should be enough for a while.  */
1934     pcmc->acpi_data_size = 0x20000 + 0x8000;
1935     pcmc->linuxboot_dma_enabled = true;
1936     pcmc->pvh_enabled = true;
1937     assert(!mc->get_hotplug_handler);
1938     mc->get_hotplug_handler = pc_get_hotplug_handler;
1939     mc->hotplug_allowed = pc_hotplug_allowed;
1940     mc->cpu_index_to_instance_props = x86_cpu_index_to_props;
1941     mc->get_default_cpu_node_id = x86_get_default_cpu_node_id;
1942     mc->possible_cpu_arch_ids = x86_possible_cpu_arch_ids;
1943     mc->auto_enable_numa_with_memhp = true;
1944     mc->has_hotpluggable_cpus = true;
1945     mc->default_boot_order = "cad";
1946     mc->hot_add_cpu = pc_hot_add_cpu;
1947     mc->smp_parse = pc_smp_parse;
1948     mc->block_default_type = IF_IDE;
1949     mc->max_cpus = 255;
1950     mc->reset = pc_machine_reset;
1951     mc->wakeup = pc_machine_wakeup;
1952     hc->pre_plug = pc_machine_device_pre_plug_cb;
1953     hc->plug = pc_machine_device_plug_cb;
1954     hc->unplug_request = pc_machine_device_unplug_request_cb;
1955     hc->unplug = pc_machine_device_unplug_cb;
1956     mc->default_cpu_type = TARGET_DEFAULT_CPU_TYPE;
1957     mc->nvdimm_supported = true;
1958     mc->numa_mem_supported = true;
1959     mc->default_ram_id = "pc.ram";
1960 
1961     object_class_property_add(oc, PC_MACHINE_DEVMEM_REGION_SIZE, "int",
1962         pc_machine_get_device_memory_region_size, NULL,
1963         NULL, NULL, &error_abort);
1964 
1965     object_class_property_add(oc, PC_MACHINE_VMPORT, "OnOffAuto",
1966         pc_machine_get_vmport, pc_machine_set_vmport,
1967         NULL, NULL, &error_abort);
1968     object_class_property_set_description(oc, PC_MACHINE_VMPORT,
1969         "Enable vmport (pc & q35)", &error_abort);
1970 
1971     object_class_property_add_bool(oc, PC_MACHINE_SMBUS,
1972         pc_machine_get_smbus, pc_machine_set_smbus, &error_abort);
1973 
1974     object_class_property_add_bool(oc, PC_MACHINE_SATA,
1975         pc_machine_get_sata, pc_machine_set_sata, &error_abort);
1976 
1977     object_class_property_add_bool(oc, PC_MACHINE_PIT,
1978         pc_machine_get_pit, pc_machine_set_pit, &error_abort);
1979 }
1980 
1981 static const TypeInfo pc_machine_info = {
1982     .name = TYPE_PC_MACHINE,
1983     .parent = TYPE_X86_MACHINE,
1984     .abstract = true,
1985     .instance_size = sizeof(PCMachineState),
1986     .instance_init = pc_machine_initfn,
1987     .class_size = sizeof(PCMachineClass),
1988     .class_init = pc_machine_class_init,
1989     .interfaces = (InterfaceInfo[]) {
1990          { TYPE_HOTPLUG_HANDLER },
1991          { }
1992     },
1993 };
1994 
pc_machine_register_types(void)1995 static void pc_machine_register_types(void)
1996 {
1997     type_register_static(&pc_machine_info);
1998 }
1999 
2000 type_init(pc_machine_register_types)
2001