2 * QEMU PC System Emulator
4 * Copyright (c) 2003-2004 Fabrice Bellard
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:
13 * The above copyright notice and this permission notice shall be included in
14 * all copies or substantial portions of the Software.
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
25 #include "qemu/osdep.h"
26 #include "qemu/units.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 "sysemu/cpus.h"
34 #include "hw/block/fdc.h"
36 #include "hw/pci/pci.h"
37 #include "hw/pci/pci_bus.h"
38 #include "hw/nvram/fw_cfg.h"
39 #include "hw/timer/hpet.h"
40 #include "hw/firmware/smbios.h"
41 #include "hw/loader.h"
43 #include "multiboot.h"
44 #include "hw/timer/mc146818rtc.h"
45 #include "hw/dma/i8257.h"
46 #include "hw/timer/i8254.h"
47 #include "hw/input/i8042.h"
48 #include "hw/audio/pcspk.h"
49 #include "hw/pci/msi.h"
50 #include "hw/sysbus.h"
51 #include "sysemu/sysemu.h"
52 #include "sysemu/numa.h"
53 #include "sysemu/kvm.h"
54 #include "sysemu/qtest.h"
56 #include "hw/xen/xen.h"
57 #include "hw/xen/start_info.h"
58 #include "ui/qemu-spice.h"
59 #include "exec/memory.h"
60 #include "exec/address-spaces.h"
61 #include "sysemu/arch_init.h"
62 #include "qemu/bitmap.h"
63 #include "qemu/config-file.h"
64 #include "qemu/error-report.h"
65 #include "qemu/option.h"
66 #include "hw/acpi/acpi.h"
67 #include "hw/acpi/cpu_hotplug.h"
68 #include "hw/boards.h"
69 #include "acpi-build.h"
70 #include "hw/mem/pc-dimm.h"
71 #include "qapi/error.h"
72 #include "qapi/qapi-visit-common.h"
73 #include "qapi/visitor.h"
77 #include "hw/i386/intel_iommu.h"
78 #include "hw/net/ne2000-isa.h"
80 /* debug PC/ISA interrupts */
84 #define DPRINTF(fmt, ...) \
85 do { printf("CPUIRQ: " fmt , ## __VA_ARGS__); } while (0)
87 #define DPRINTF(fmt, ...)
90 #define FW_CFG_ACPI_TABLES (FW_CFG_ARCH_LOCAL + 0)
91 #define FW_CFG_SMBIOS_ENTRIES (FW_CFG_ARCH_LOCAL + 1)
92 #define FW_CFG_IRQ0_OVERRIDE (FW_CFG_ARCH_LOCAL + 2)
93 #define FW_CFG_E820_TABLE (FW_CFG_ARCH_LOCAL + 3)
94 #define FW_CFG_HPET (FW_CFG_ARCH_LOCAL + 4)
96 #define E820_NR_ENTRIES 16
102 } QEMU_PACKED
__attribute((__aligned__(4)));
106 struct e820_entry entry
[E820_NR_ENTRIES
];
107 } QEMU_PACKED
__attribute((__aligned__(4)));
109 static struct e820_table e820_reserve
;
110 static struct e820_entry
*e820_table
;
111 static unsigned e820_entries
;
112 struct hpet_fw_config hpet_cfg
= {.count
= UINT8_MAX
};
114 /* Physical Address of PVH entry point read from kernel ELF NOTE */
115 static size_t pvh_start_addr
;
117 GlobalProperty pc_compat_3_1
[] = {
118 { "intel-iommu", "dma-drain", "off" },
119 { "Opteron_G3" "-" TYPE_X86_CPU
, "rdtscp", "off" },
120 { "Opteron_G4" "-" TYPE_X86_CPU
, "rdtscp", "off" },
121 { "Opteron_G4" "-" TYPE_X86_CPU
, "npt", "off" },
122 { "Opteron_G4" "-" TYPE_X86_CPU
, "nrip-save", "off" },
123 { "Opteron_G5" "-" TYPE_X86_CPU
, "rdtscp", "off" },
124 { "Opteron_G5" "-" TYPE_X86_CPU
, "npt", "off" },
125 { "Opteron_G5" "-" TYPE_X86_CPU
, "nrip-save", "off" },
126 { "EPYC" "-" TYPE_X86_CPU
, "npt", "off" },
127 { "EPYC" "-" TYPE_X86_CPU
, "nrip-save", "off" },
128 { "EPYC-IBPB" "-" TYPE_X86_CPU
, "npt", "off" },
129 { "EPYC-IBPB" "-" TYPE_X86_CPU
, "nrip-save", "off" },
130 { "Skylake-Client" "-" TYPE_X86_CPU
, "mpx", "on" },
131 { "Skylake-Client-IBRS" "-" TYPE_X86_CPU
, "mpx", "on" },
132 { "Skylake-Server" "-" TYPE_X86_CPU
, "mpx", "on" },
133 { "Skylake-Server-IBRS" "-" TYPE_X86_CPU
, "mpx", "on" },
134 { "Cascadelake-Server" "-" TYPE_X86_CPU
, "mpx", "on" },
135 { "Icelake-Client" "-" TYPE_X86_CPU
, "mpx", "on" },
136 { "Icelake-Server" "-" TYPE_X86_CPU
, "mpx", "on" },
137 { "Cascadelake-Server" "-" TYPE_X86_CPU
, "stepping", "5" },
139 const size_t pc_compat_3_1_len
= G_N_ELEMENTS(pc_compat_3_1
);
141 GlobalProperty pc_compat_3_0
[] = {
142 { TYPE_X86_CPU
, "x-hv-synic-kvm-only", "on" },
143 { "Skylake-Server" "-" TYPE_X86_CPU
, "pku", "off" },
144 { "Skylake-Server-IBRS" "-" TYPE_X86_CPU
, "pku", "off" },
146 const size_t pc_compat_3_0_len
= G_N_ELEMENTS(pc_compat_3_0
);
148 GlobalProperty pc_compat_2_12
[] = {
149 { TYPE_X86_CPU
, "legacy-cache", "on" },
150 { TYPE_X86_CPU
, "topoext", "off" },
151 { "EPYC-" TYPE_X86_CPU
, "xlevel", "0x8000000a" },
152 { "EPYC-IBPB-" TYPE_X86_CPU
, "xlevel", "0x8000000a" },
154 const size_t pc_compat_2_12_len
= G_N_ELEMENTS(pc_compat_2_12
);
156 GlobalProperty pc_compat_2_11
[] = {
157 { TYPE_X86_CPU
, "x-migrate-smi-count", "off" },
158 { "Skylake-Server" "-" TYPE_X86_CPU
, "clflushopt", "off" },
160 const size_t pc_compat_2_11_len
= G_N_ELEMENTS(pc_compat_2_11
);
162 GlobalProperty pc_compat_2_10
[] = {
163 { TYPE_X86_CPU
, "x-hv-max-vps", "0x40" },
164 { "i440FX-pcihost", "x-pci-hole64-fix", "off" },
165 { "q35-pcihost", "x-pci-hole64-fix", "off" },
167 const size_t pc_compat_2_10_len
= G_N_ELEMENTS(pc_compat_2_10
);
169 GlobalProperty pc_compat_2_9
[] = {
170 { "mch", "extended-tseg-mbytes", "0" },
172 const size_t pc_compat_2_9_len
= G_N_ELEMENTS(pc_compat_2_9
);
174 GlobalProperty pc_compat_2_8
[] = {
175 { TYPE_X86_CPU
, "tcg-cpuid", "off" },
176 { "kvmclock", "x-mach-use-reliable-get-clock", "off" },
177 { "ICH9-LPC", "x-smi-broadcast", "off" },
178 { TYPE_X86_CPU
, "vmware-cpuid-freq", "off" },
179 { "Haswell-" TYPE_X86_CPU
, "stepping", "1" },
181 const size_t pc_compat_2_8_len
= G_N_ELEMENTS(pc_compat_2_8
);
183 GlobalProperty pc_compat_2_7
[] = {
184 { TYPE_X86_CPU
, "l3-cache", "off" },
185 { TYPE_X86_CPU
, "full-cpuid-auto-level", "off" },
186 { "Opteron_G3" "-" TYPE_X86_CPU
, "family", "15" },
187 { "Opteron_G3" "-" TYPE_X86_CPU
, "model", "6" },
188 { "Opteron_G3" "-" TYPE_X86_CPU
, "stepping", "1" },
189 { "isa-pcspk", "migrate", "off" },
191 const size_t pc_compat_2_7_len
= G_N_ELEMENTS(pc_compat_2_7
);
193 GlobalProperty pc_compat_2_6
[] = {
194 { TYPE_X86_CPU
, "cpuid-0xb", "off" },
195 { "vmxnet3", "romfile", "" },
196 { TYPE_X86_CPU
, "fill-mtrr-mask", "off" },
197 { "apic-common", "legacy-instance-id", "on", }
199 const size_t pc_compat_2_6_len
= G_N_ELEMENTS(pc_compat_2_6
);
201 GlobalProperty pc_compat_2_5
[] = {};
202 const size_t pc_compat_2_5_len
= G_N_ELEMENTS(pc_compat_2_5
);
204 GlobalProperty pc_compat_2_4
[] = {
205 PC_CPU_MODEL_IDS("2.4.0")
206 { "Haswell-" TYPE_X86_CPU
, "abm", "off" },
207 { "Haswell-noTSX-" TYPE_X86_CPU
, "abm", "off" },
208 { "Broadwell-" TYPE_X86_CPU
, "abm", "off" },
209 { "Broadwell-noTSX-" TYPE_X86_CPU
, "abm", "off" },
210 { "host" "-" TYPE_X86_CPU
, "host-cache-info", "on" },
211 { TYPE_X86_CPU
, "check", "off" },
212 { "qemu64" "-" TYPE_X86_CPU
, "sse4a", "on" },
213 { "qemu64" "-" TYPE_X86_CPU
, "abm", "on" },
214 { "qemu64" "-" TYPE_X86_CPU
, "popcnt", "on" },
215 { "qemu32" "-" TYPE_X86_CPU
, "popcnt", "on" },
216 { "Opteron_G2" "-" TYPE_X86_CPU
, "rdtscp", "on" },
217 { "Opteron_G3" "-" TYPE_X86_CPU
, "rdtscp", "on" },
218 { "Opteron_G4" "-" TYPE_X86_CPU
, "rdtscp", "on" },
219 { "Opteron_G5" "-" TYPE_X86_CPU
, "rdtscp", "on", }
221 const size_t pc_compat_2_4_len
= G_N_ELEMENTS(pc_compat_2_4
);
223 GlobalProperty pc_compat_2_3
[] = {
224 PC_CPU_MODEL_IDS("2.3.0")
225 { TYPE_X86_CPU
, "arat", "off" },
226 { "qemu64" "-" TYPE_X86_CPU
, "min-level", "4" },
227 { "kvm64" "-" TYPE_X86_CPU
, "min-level", "5" },
228 { "pentium3" "-" TYPE_X86_CPU
, "min-level", "2" },
229 { "n270" "-" TYPE_X86_CPU
, "min-level", "5" },
230 { "Conroe" "-" TYPE_X86_CPU
, "min-level", "4" },
231 { "Penryn" "-" TYPE_X86_CPU
, "min-level", "4" },
232 { "Nehalem" "-" TYPE_X86_CPU
, "min-level", "4" },
233 { "n270" "-" TYPE_X86_CPU
, "min-xlevel", "0x8000000a" },
234 { "Penryn" "-" TYPE_X86_CPU
, "min-xlevel", "0x8000000a" },
235 { "Conroe" "-" TYPE_X86_CPU
, "min-xlevel", "0x8000000a" },
236 { "Nehalem" "-" TYPE_X86_CPU
, "min-xlevel", "0x8000000a" },
237 { "Westmere" "-" TYPE_X86_CPU
, "min-xlevel", "0x8000000a" },
238 { "SandyBridge" "-" TYPE_X86_CPU
, "min-xlevel", "0x8000000a" },
239 { "IvyBridge" "-" TYPE_X86_CPU
, "min-xlevel", "0x8000000a" },
240 { "Haswell" "-" TYPE_X86_CPU
, "min-xlevel", "0x8000000a" },
241 { "Haswell-noTSX" "-" TYPE_X86_CPU
, "min-xlevel", "0x8000000a" },
242 { "Broadwell" "-" TYPE_X86_CPU
, "min-xlevel", "0x8000000a" },
243 { "Broadwell-noTSX" "-" TYPE_X86_CPU
, "min-xlevel", "0x8000000a" },
244 { TYPE_X86_CPU
, "kvm-no-smi-migration", "on" },
246 const size_t pc_compat_2_3_len
= G_N_ELEMENTS(pc_compat_2_3
);
248 GlobalProperty pc_compat_2_2
[] = {
249 PC_CPU_MODEL_IDS("2.2.0")
250 { "kvm64" "-" TYPE_X86_CPU
, "vme", "off" },
251 { "kvm32" "-" TYPE_X86_CPU
, "vme", "off" },
252 { "Conroe" "-" TYPE_X86_CPU
, "vme", "off" },
253 { "Penryn" "-" TYPE_X86_CPU
, "vme", "off" },
254 { "Nehalem" "-" TYPE_X86_CPU
, "vme", "off" },
255 { "Westmere" "-" TYPE_X86_CPU
, "vme", "off" },
256 { "SandyBridge" "-" TYPE_X86_CPU
, "vme", "off" },
257 { "Haswell" "-" TYPE_X86_CPU
, "vme", "off" },
258 { "Broadwell" "-" TYPE_X86_CPU
, "vme", "off" },
259 { "Opteron_G1" "-" TYPE_X86_CPU
, "vme", "off" },
260 { "Opteron_G2" "-" TYPE_X86_CPU
, "vme", "off" },
261 { "Opteron_G3" "-" TYPE_X86_CPU
, "vme", "off" },
262 { "Opteron_G4" "-" TYPE_X86_CPU
, "vme", "off" },
263 { "Opteron_G5" "-" TYPE_X86_CPU
, "vme", "off" },
264 { "Haswell" "-" TYPE_X86_CPU
, "f16c", "off" },
265 { "Haswell" "-" TYPE_X86_CPU
, "rdrand", "off" },
266 { "Broadwell" "-" TYPE_X86_CPU
, "f16c", "off" },
267 { "Broadwell" "-" TYPE_X86_CPU
, "rdrand", "off" },
269 const size_t pc_compat_2_2_len
= G_N_ELEMENTS(pc_compat_2_2
);
271 GlobalProperty pc_compat_2_1
[] = {
272 PC_CPU_MODEL_IDS("2.1.0")
273 { "coreduo" "-" TYPE_X86_CPU
, "vmx", "on" },
274 { "core2duo" "-" TYPE_X86_CPU
, "vmx", "on" },
276 const size_t pc_compat_2_1_len
= G_N_ELEMENTS(pc_compat_2_1
);
278 GlobalProperty pc_compat_2_0
[] = {
279 PC_CPU_MODEL_IDS("2.0.0")
280 { "virtio-scsi-pci", "any_layout", "off" },
281 { "PIIX4_PM", "memory-hotplug-support", "off" },
282 { "apic", "version", "0x11" },
283 { "nec-usb-xhci", "superspeed-ports-first", "off" },
284 { "nec-usb-xhci", "force-pcie-endcap", "on" },
285 { "pci-serial", "prog_if", "0" },
286 { "pci-serial-2x", "prog_if", "0" },
287 { "pci-serial-4x", "prog_if", "0" },
288 { "virtio-net-pci", "guest_announce", "off" },
289 { "ICH9-LPC", "memory-hotplug-support", "off" },
290 { "xio3130-downstream", COMPAT_PROP_PCP
, "off" },
291 { "ioh3420", COMPAT_PROP_PCP
, "off" },
293 const size_t pc_compat_2_0_len
= G_N_ELEMENTS(pc_compat_2_0
);
295 GlobalProperty pc_compat_1_7
[] = {
296 PC_CPU_MODEL_IDS("1.7.0")
297 { TYPE_USB_DEVICE
, "msos-desc", "no" },
298 { "PIIX4_PM", "acpi-pci-hotplug-with-bridge-support", "off" },
299 { "hpet", HPET_INTCAP
, "4" },
301 const size_t pc_compat_1_7_len
= G_N_ELEMENTS(pc_compat_1_7
);
303 GlobalProperty pc_compat_1_6
[] = {
304 PC_CPU_MODEL_IDS("1.6.0")
305 { "e1000", "mitigation", "off" },
306 { "qemu64-" TYPE_X86_CPU
, "model", "2" },
307 { "qemu32-" TYPE_X86_CPU
, "model", "3" },
308 { "i440FX-pcihost", "short_root_bus", "1" },
309 { "q35-pcihost", "short_root_bus", "1" },
311 const size_t pc_compat_1_6_len
= G_N_ELEMENTS(pc_compat_1_6
);
313 GlobalProperty pc_compat_1_5
[] = {
314 PC_CPU_MODEL_IDS("1.5.0")
315 { "Conroe-" TYPE_X86_CPU
, "model", "2" },
316 { "Conroe-" TYPE_X86_CPU
, "min-level", "2" },
317 { "Penryn-" TYPE_X86_CPU
, "model", "2" },
318 { "Penryn-" TYPE_X86_CPU
, "min-level", "2" },
319 { "Nehalem-" TYPE_X86_CPU
, "model", "2" },
320 { "Nehalem-" TYPE_X86_CPU
, "min-level", "2" },
321 { "virtio-net-pci", "any_layout", "off" },
322 { TYPE_X86_CPU
, "pmu", "on" },
323 { "i440FX-pcihost", "short_root_bus", "0" },
324 { "q35-pcihost", "short_root_bus", "0" },
326 const size_t pc_compat_1_5_len
= G_N_ELEMENTS(pc_compat_1_5
);
328 GlobalProperty pc_compat_1_4
[] = {
329 PC_CPU_MODEL_IDS("1.4.0")
330 { "scsi-hd", "discard_granularity", "0" },
331 { "scsi-cd", "discard_granularity", "0" },
332 { "scsi-disk", "discard_granularity", "0" },
333 { "ide-hd", "discard_granularity", "0" },
334 { "ide-cd", "discard_granularity", "0" },
335 { "ide-drive", "discard_granularity", "0" },
336 { "virtio-blk-pci", "discard_granularity", "0" },
337 /* DEV_NVECTORS_UNSPECIFIED as a uint32_t string: */
338 { "virtio-serial-pci", "vectors", "0xFFFFFFFF" },
339 { "virtio-net-pci", "ctrl_guest_offloads", "off" },
340 { "e1000", "romfile", "pxe-e1000.rom" },
341 { "ne2k_pci", "romfile", "pxe-ne2k_pci.rom" },
342 { "pcnet", "romfile", "pxe-pcnet.rom" },
343 { "rtl8139", "romfile", "pxe-rtl8139.rom" },
344 { "virtio-net-pci", "romfile", "pxe-virtio.rom" },
345 { "486-" TYPE_X86_CPU
, "model", "0" },
346 { "n270" "-" TYPE_X86_CPU
, "movbe", "off" },
347 { "Westmere" "-" TYPE_X86_CPU
, "pclmulqdq", "off" },
349 const size_t pc_compat_1_4_len
= G_N_ELEMENTS(pc_compat_1_4
);
351 void gsi_handler(void *opaque
, int n
, int level
)
353 GSIState
*s
= opaque
;
355 DPRINTF("pc: %s GSI %d\n", level
? "raising" : "lowering", n
);
356 if (n
< ISA_NUM_IRQS
) {
357 qemu_set_irq(s
->i8259_irq
[n
], level
);
359 qemu_set_irq(s
->ioapic_irq
[n
], level
);
362 static void ioport80_write(void *opaque
, hwaddr addr
, uint64_t data
,
367 static uint64_t ioport80_read(void *opaque
, hwaddr addr
, unsigned size
)
369 return 0xffffffffffffffffULL
;
372 /* MSDOS compatibility mode FPU exception support */
373 static qemu_irq ferr_irq
;
375 void pc_register_ferr_irq(qemu_irq irq
)
380 /* XXX: add IGNNE support */
381 void cpu_set_ferr(CPUX86State
*s
)
383 qemu_irq_raise(ferr_irq
);
386 static void ioportF0_write(void *opaque
, hwaddr addr
, uint64_t data
,
389 qemu_irq_lower(ferr_irq
);
392 static uint64_t ioportF0_read(void *opaque
, hwaddr addr
, unsigned size
)
394 return 0xffffffffffffffffULL
;
398 uint64_t cpu_get_tsc(CPUX86State
*env
)
400 return cpu_get_ticks();
404 int cpu_get_pic_interrupt(CPUX86State
*env
)
406 X86CPU
*cpu
= x86_env_get_cpu(env
);
409 if (!kvm_irqchip_in_kernel()) {
410 intno
= apic_get_interrupt(cpu
->apic_state
);
414 /* read the irq from the PIC */
415 if (!apic_accept_pic_intr(cpu
->apic_state
)) {
420 intno
= pic_read_irq(isa_pic
);
424 static void pic_irq_request(void *opaque
, int irq
, int level
)
426 CPUState
*cs
= first_cpu
;
427 X86CPU
*cpu
= X86_CPU(cs
);
429 DPRINTF("pic_irqs: %s irq %d\n", level
? "raise" : "lower", irq
);
430 if (cpu
->apic_state
&& !kvm_irqchip_in_kernel()) {
433 if (apic_accept_pic_intr(cpu
->apic_state
)) {
434 apic_deliver_pic_intr(cpu
->apic_state
, level
);
439 cpu_interrupt(cs
, CPU_INTERRUPT_HARD
);
441 cpu_reset_interrupt(cs
, CPU_INTERRUPT_HARD
);
446 /* PC cmos mappings */
448 #define REG_EQUIPMENT_BYTE 0x14
450 int cmos_get_fd_drive_type(FloppyDriveType fd0
)
455 case FLOPPY_DRIVE_TYPE_144
:
456 /* 1.44 Mb 3"5 drive */
459 case FLOPPY_DRIVE_TYPE_288
:
460 /* 2.88 Mb 3"5 drive */
463 case FLOPPY_DRIVE_TYPE_120
:
464 /* 1.2 Mb 5"5 drive */
467 case FLOPPY_DRIVE_TYPE_NONE
:
475 static void cmos_init_hd(ISADevice
*s
, int type_ofs
, int info_ofs
,
476 int16_t cylinders
, int8_t heads
, int8_t sectors
)
478 rtc_set_memory(s
, type_ofs
, 47);
479 rtc_set_memory(s
, info_ofs
, cylinders
);
480 rtc_set_memory(s
, info_ofs
+ 1, cylinders
>> 8);
481 rtc_set_memory(s
, info_ofs
+ 2, heads
);
482 rtc_set_memory(s
, info_ofs
+ 3, 0xff);
483 rtc_set_memory(s
, info_ofs
+ 4, 0xff);
484 rtc_set_memory(s
, info_ofs
+ 5, 0xc0 | ((heads
> 8) << 3));
485 rtc_set_memory(s
, info_ofs
+ 6, cylinders
);
486 rtc_set_memory(s
, info_ofs
+ 7, cylinders
>> 8);
487 rtc_set_memory(s
, info_ofs
+ 8, sectors
);
490 /* convert boot_device letter to something recognizable by the bios */
491 static int boot_device2nibble(char boot_device
)
493 switch(boot_device
) {
496 return 0x01; /* floppy boot */
498 return 0x02; /* hard drive boot */
500 return 0x03; /* CD-ROM boot */
502 return 0x04; /* Network boot */
507 static void set_boot_dev(ISADevice
*s
, const char *boot_device
, Error
**errp
)
509 #define PC_MAX_BOOT_DEVICES 3
510 int nbds
, bds
[3] = { 0, };
513 nbds
= strlen(boot_device
);
514 if (nbds
> PC_MAX_BOOT_DEVICES
) {
515 error_setg(errp
, "Too many boot devices for PC");
518 for (i
= 0; i
< nbds
; i
++) {
519 bds
[i
] = boot_device2nibble(boot_device
[i
]);
521 error_setg(errp
, "Invalid boot device for PC: '%c'",
526 rtc_set_memory(s
, 0x3d, (bds
[1] << 4) | bds
[0]);
527 rtc_set_memory(s
, 0x38, (bds
[2] << 4) | (fd_bootchk
? 0x0 : 0x1));
530 static void pc_boot_set(void *opaque
, const char *boot_device
, Error
**errp
)
532 set_boot_dev(opaque
, boot_device
, errp
);
535 static void pc_cmos_init_floppy(ISADevice
*rtc_state
, ISADevice
*floppy
)
538 FloppyDriveType fd_type
[2] = { FLOPPY_DRIVE_TYPE_NONE
,
539 FLOPPY_DRIVE_TYPE_NONE
};
543 for (i
= 0; i
< 2; i
++) {
544 fd_type
[i
] = isa_fdc_get_drive_type(floppy
, i
);
547 val
= (cmos_get_fd_drive_type(fd_type
[0]) << 4) |
548 cmos_get_fd_drive_type(fd_type
[1]);
549 rtc_set_memory(rtc_state
, 0x10, val
);
551 val
= rtc_get_memory(rtc_state
, REG_EQUIPMENT_BYTE
);
553 if (fd_type
[0] != FLOPPY_DRIVE_TYPE_NONE
) {
556 if (fd_type
[1] != FLOPPY_DRIVE_TYPE_NONE
) {
563 val
|= 0x01; /* 1 drive, ready for boot */
566 val
|= 0x41; /* 2 drives, ready for boot */
569 rtc_set_memory(rtc_state
, REG_EQUIPMENT_BYTE
, val
);
572 typedef struct pc_cmos_init_late_arg
{
573 ISADevice
*rtc_state
;
575 } pc_cmos_init_late_arg
;
577 typedef struct check_fdc_state
{
582 static int check_fdc(Object
*obj
, void *opaque
)
584 CheckFdcState
*state
= opaque
;
587 Error
*local_err
= NULL
;
589 fdc
= object_dynamic_cast(obj
, TYPE_ISA_FDC
);
594 iobase
= object_property_get_uint(obj
, "iobase", &local_err
);
595 if (local_err
|| iobase
!= 0x3f0) {
596 error_free(local_err
);
601 state
->multiple
= true;
603 state
->floppy
= ISA_DEVICE(obj
);
608 static const char * const fdc_container_path
[] = {
609 "/unattached", "/peripheral", "/peripheral-anon"
613 * Locate the FDC at IO address 0x3f0, in order to configure the CMOS registers
616 ISADevice
*pc_find_fdc0(void)
620 CheckFdcState state
= { 0 };
622 for (i
= 0; i
< ARRAY_SIZE(fdc_container_path
); i
++) {
623 container
= container_get(qdev_get_machine(), fdc_container_path
[i
]);
624 object_child_foreach(container
, check_fdc
, &state
);
627 if (state
.multiple
) {
628 warn_report("multiple floppy disk controllers with "
629 "iobase=0x3f0 have been found");
630 error_printf("the one being picked for CMOS setup might not reflect "
637 static void pc_cmos_init_late(void *opaque
)
639 pc_cmos_init_late_arg
*arg
= opaque
;
640 ISADevice
*s
= arg
->rtc_state
;
642 int8_t heads
, sectors
;
647 if (arg
->idebus
[0] && ide_get_geometry(arg
->idebus
[0], 0,
648 &cylinders
, &heads
, §ors
) >= 0) {
649 cmos_init_hd(s
, 0x19, 0x1b, cylinders
, heads
, sectors
);
652 if (arg
->idebus
[0] && ide_get_geometry(arg
->idebus
[0], 1,
653 &cylinders
, &heads
, §ors
) >= 0) {
654 cmos_init_hd(s
, 0x1a, 0x24, cylinders
, heads
, sectors
);
657 rtc_set_memory(s
, 0x12, val
);
660 for (i
= 0; i
< 4; i
++) {
661 /* NOTE: ide_get_geometry() returns the physical
662 geometry. It is always such that: 1 <= sects <= 63, 1
663 <= heads <= 16, 1 <= cylinders <= 16383. The BIOS
664 geometry can be different if a translation is done. */
665 if (arg
->idebus
[i
/ 2] &&
666 ide_get_geometry(arg
->idebus
[i
/ 2], i
% 2,
667 &cylinders
, &heads
, §ors
) >= 0) {
668 trans
= ide_get_bios_chs_trans(arg
->idebus
[i
/ 2], i
% 2) - 1;
669 assert((trans
& ~3) == 0);
670 val
|= trans
<< (i
* 2);
673 rtc_set_memory(s
, 0x39, val
);
675 pc_cmos_init_floppy(s
, pc_find_fdc0());
677 qemu_unregister_reset(pc_cmos_init_late
, opaque
);
680 void pc_cmos_init(PCMachineState
*pcms
,
681 BusState
*idebus0
, BusState
*idebus1
,
685 static pc_cmos_init_late_arg arg
;
687 /* various important CMOS locations needed by PC/Bochs bios */
690 /* base memory (first MiB) */
691 val
= MIN(pcms
->below_4g_mem_size
/ KiB
, 640);
692 rtc_set_memory(s
, 0x15, val
);
693 rtc_set_memory(s
, 0x16, val
>> 8);
694 /* extended memory (next 64MiB) */
695 if (pcms
->below_4g_mem_size
> 1 * MiB
) {
696 val
= (pcms
->below_4g_mem_size
- 1 * MiB
) / KiB
;
702 rtc_set_memory(s
, 0x17, val
);
703 rtc_set_memory(s
, 0x18, val
>> 8);
704 rtc_set_memory(s
, 0x30, val
);
705 rtc_set_memory(s
, 0x31, val
>> 8);
706 /* memory between 16MiB and 4GiB */
707 if (pcms
->below_4g_mem_size
> 16 * MiB
) {
708 val
= (pcms
->below_4g_mem_size
- 16 * MiB
) / (64 * KiB
);
714 rtc_set_memory(s
, 0x34, val
);
715 rtc_set_memory(s
, 0x35, val
>> 8);
716 /* memory above 4GiB */
717 val
= pcms
->above_4g_mem_size
/ 65536;
718 rtc_set_memory(s
, 0x5b, val
);
719 rtc_set_memory(s
, 0x5c, val
>> 8);
720 rtc_set_memory(s
, 0x5d, val
>> 16);
722 object_property_add_link(OBJECT(pcms
), "rtc_state",
724 (Object
**)&pcms
->rtc
,
725 object_property_allow_set_link
,
726 OBJ_PROP_LINK_STRONG
, &error_abort
);
727 object_property_set_link(OBJECT(pcms
), OBJECT(s
),
728 "rtc_state", &error_abort
);
730 set_boot_dev(s
, MACHINE(pcms
)->boot_order
, &error_fatal
);
733 val
|= 0x02; /* FPU is there */
734 val
|= 0x04; /* PS/2 mouse installed */
735 rtc_set_memory(s
, REG_EQUIPMENT_BYTE
, val
);
737 /* hard drives and FDC */
739 arg
.idebus
[0] = idebus0
;
740 arg
.idebus
[1] = idebus1
;
741 qemu_register_reset(pc_cmos_init_late
, &arg
);
744 #define TYPE_PORT92 "port92"
745 #define PORT92(obj) OBJECT_CHECK(Port92State, (obj), TYPE_PORT92)
747 /* port 92 stuff: could be split off */
748 typedef struct Port92State
{
749 ISADevice parent_obj
;
756 static void port92_write(void *opaque
, hwaddr addr
, uint64_t val
,
759 Port92State
*s
= opaque
;
760 int oldval
= s
->outport
;
762 DPRINTF("port92: write 0x%02" PRIx64
"\n", val
);
764 qemu_set_irq(s
->a20_out
, (val
>> 1) & 1);
765 if ((val
& 1) && !(oldval
& 1)) {
766 qemu_system_reset_request(SHUTDOWN_CAUSE_GUEST_RESET
);
770 static uint64_t port92_read(void *opaque
, hwaddr addr
,
773 Port92State
*s
= opaque
;
777 DPRINTF("port92: read 0x%02x\n", ret
);
781 static void port92_init(ISADevice
*dev
, qemu_irq a20_out
)
783 qdev_connect_gpio_out_named(DEVICE(dev
), PORT92_A20_LINE
, 0, a20_out
);
786 static const VMStateDescription vmstate_port92_isa
= {
789 .minimum_version_id
= 1,
790 .fields
= (VMStateField
[]) {
791 VMSTATE_UINT8(outport
, Port92State
),
792 VMSTATE_END_OF_LIST()
796 static void port92_reset(DeviceState
*d
)
798 Port92State
*s
= PORT92(d
);
803 static const MemoryRegionOps port92_ops
= {
805 .write
= port92_write
,
807 .min_access_size
= 1,
808 .max_access_size
= 1,
810 .endianness
= DEVICE_LITTLE_ENDIAN
,
813 static void port92_initfn(Object
*obj
)
815 Port92State
*s
= PORT92(obj
);
817 memory_region_init_io(&s
->io
, OBJECT(s
), &port92_ops
, s
, "port92", 1);
821 qdev_init_gpio_out_named(DEVICE(obj
), &s
->a20_out
, PORT92_A20_LINE
, 1);
824 static void port92_realizefn(DeviceState
*dev
, Error
**errp
)
826 ISADevice
*isadev
= ISA_DEVICE(dev
);
827 Port92State
*s
= PORT92(dev
);
829 isa_register_ioport(isadev
, &s
->io
, 0x92);
832 static void port92_class_initfn(ObjectClass
*klass
, void *data
)
834 DeviceClass
*dc
= DEVICE_CLASS(klass
);
836 dc
->realize
= port92_realizefn
;
837 dc
->reset
= port92_reset
;
838 dc
->vmsd
= &vmstate_port92_isa
;
840 * Reason: unlike ordinary ISA devices, this one needs additional
841 * wiring: its A20 output line needs to be wired up by
844 dc
->user_creatable
= false;
847 static const TypeInfo port92_info
= {
849 .parent
= TYPE_ISA_DEVICE
,
850 .instance_size
= sizeof(Port92State
),
851 .instance_init
= port92_initfn
,
852 .class_init
= port92_class_initfn
,
855 static void port92_register_types(void)
857 type_register_static(&port92_info
);
860 type_init(port92_register_types
)
862 static void handle_a20_line_change(void *opaque
, int irq
, int level
)
864 X86CPU
*cpu
= opaque
;
866 /* XXX: send to all CPUs ? */
867 /* XXX: add logic to handle multiple A20 line sources */
868 x86_cpu_set_a20(cpu
, level
);
871 int e820_add_entry(uint64_t address
, uint64_t length
, uint32_t type
)
873 int index
= le32_to_cpu(e820_reserve
.count
);
874 struct e820_entry
*entry
;
876 if (type
!= E820_RAM
) {
877 /* old FW_CFG_E820_TABLE entry -- reservations only */
878 if (index
>= E820_NR_ENTRIES
) {
881 entry
= &e820_reserve
.entry
[index
++];
883 entry
->address
= cpu_to_le64(address
);
884 entry
->length
= cpu_to_le64(length
);
885 entry
->type
= cpu_to_le32(type
);
887 e820_reserve
.count
= cpu_to_le32(index
);
890 /* new "etc/e820" file -- include ram too */
891 e820_table
= g_renew(struct e820_entry
, e820_table
, e820_entries
+ 1);
892 e820_table
[e820_entries
].address
= cpu_to_le64(address
);
893 e820_table
[e820_entries
].length
= cpu_to_le64(length
);
894 e820_table
[e820_entries
].type
= cpu_to_le32(type
);
900 int e820_get_num_entries(void)
905 bool e820_get_entry(int idx
, uint32_t type
, uint64_t *address
, uint64_t *length
)
907 if (idx
< e820_entries
&& e820_table
[idx
].type
== cpu_to_le32(type
)) {
908 *address
= le64_to_cpu(e820_table
[idx
].address
);
909 *length
= le64_to_cpu(e820_table
[idx
].length
);
915 /* Enables contiguous-apic-ID mode, for compatibility */
916 static bool compat_apic_id_mode
;
918 void enable_compat_apic_id_mode(void)
920 compat_apic_id_mode
= true;
923 /* Calculates initial APIC ID for a specific CPU index
925 * Currently we need to be able to calculate the APIC ID from the CPU index
926 * alone (without requiring a CPU object), as the QEMU<->Seabios interfaces have
927 * no concept of "CPU index", and the NUMA tables on fw_cfg need the APIC ID of
928 * all CPUs up to max_cpus.
930 static uint32_t x86_cpu_apic_id_from_index(unsigned int cpu_index
)
935 correct_id
= x86_apicid_from_cpu_idx(smp_cores
, smp_threads
, cpu_index
);
936 if (compat_apic_id_mode
) {
937 if (cpu_index
!= correct_id
&& !warned
&& !qtest_enabled()) {
938 error_report("APIC IDs set in compatibility mode, "
939 "CPU topology won't match the configuration");
948 static void pc_build_smbios(PCMachineState
*pcms
)
950 uint8_t *smbios_tables
, *smbios_anchor
;
951 size_t smbios_tables_len
, smbios_anchor_len
;
952 struct smbios_phys_mem_area
*mem_array
;
953 unsigned i
, array_count
;
954 MachineState
*ms
= MACHINE(pcms
);
955 X86CPU
*cpu
= X86_CPU(ms
->possible_cpus
->cpus
[0].cpu
);
957 /* tell smbios about cpuid version and features */
958 smbios_set_cpuid(cpu
->env
.cpuid_version
, cpu
->env
.features
[FEAT_1_EDX
]);
960 smbios_tables
= smbios_get_table_legacy(&smbios_tables_len
);
962 fw_cfg_add_bytes(pcms
->fw_cfg
, FW_CFG_SMBIOS_ENTRIES
,
963 smbios_tables
, smbios_tables_len
);
966 /* build the array of physical mem area from e820 table */
967 mem_array
= g_malloc0(sizeof(*mem_array
) * e820_get_num_entries());
968 for (i
= 0, array_count
= 0; i
< e820_get_num_entries(); i
++) {
971 if (e820_get_entry(i
, E820_RAM
, &addr
, &len
)) {
972 mem_array
[array_count
].address
= addr
;
973 mem_array
[array_count
].length
= len
;
977 smbios_get_tables(mem_array
, array_count
,
978 &smbios_tables
, &smbios_tables_len
,
979 &smbios_anchor
, &smbios_anchor_len
);
983 fw_cfg_add_file(pcms
->fw_cfg
, "etc/smbios/smbios-tables",
984 smbios_tables
, smbios_tables_len
);
985 fw_cfg_add_file(pcms
->fw_cfg
, "etc/smbios/smbios-anchor",
986 smbios_anchor
, smbios_anchor_len
);
990 static FWCfgState
*bochs_bios_init(AddressSpace
*as
, PCMachineState
*pcms
)
993 uint64_t *numa_fw_cfg
;
995 const CPUArchIdList
*cpus
;
996 MachineClass
*mc
= MACHINE_GET_CLASS(pcms
);
998 fw_cfg
= fw_cfg_init_io_dma(FW_CFG_IO_BASE
, FW_CFG_IO_BASE
+ 4, as
);
999 fw_cfg_add_i16(fw_cfg
, FW_CFG_NB_CPUS
, pcms
->boot_cpus
);
1001 /* FW_CFG_MAX_CPUS is a bit confusing/problematic on x86:
1003 * For machine types prior to 1.8, SeaBIOS needs FW_CFG_MAX_CPUS for
1004 * building MPTable, ACPI MADT, ACPI CPU hotplug and ACPI SRAT table,
1005 * that tables are based on xAPIC ID and QEMU<->SeaBIOS interface
1006 * for CPU hotplug also uses APIC ID and not "CPU index".
1007 * This means that FW_CFG_MAX_CPUS is not the "maximum number of CPUs",
1008 * but the "limit to the APIC ID values SeaBIOS may see".
1010 * So for compatibility reasons with old BIOSes we are stuck with
1011 * "etc/max-cpus" actually being apic_id_limit
1013 fw_cfg_add_i16(fw_cfg
, FW_CFG_MAX_CPUS
, (uint16_t)pcms
->apic_id_limit
);
1014 fw_cfg_add_i64(fw_cfg
, FW_CFG_RAM_SIZE
, (uint64_t)ram_size
);
1015 fw_cfg_add_bytes(fw_cfg
, FW_CFG_ACPI_TABLES
,
1016 acpi_tables
, acpi_tables_len
);
1017 fw_cfg_add_i32(fw_cfg
, FW_CFG_IRQ0_OVERRIDE
, kvm_allows_irq0_override());
1019 fw_cfg_add_bytes(fw_cfg
, FW_CFG_E820_TABLE
,
1020 &e820_reserve
, sizeof(e820_reserve
));
1021 fw_cfg_add_file(fw_cfg
, "etc/e820", e820_table
,
1022 sizeof(struct e820_entry
) * e820_entries
);
1024 fw_cfg_add_bytes(fw_cfg
, FW_CFG_HPET
, &hpet_cfg
, sizeof(hpet_cfg
));
1025 /* allocate memory for the NUMA channel: one (64bit) word for the number
1026 * of nodes, one word for each VCPU->node and one word for each node to
1027 * hold the amount of memory.
1029 numa_fw_cfg
= g_new0(uint64_t, 1 + pcms
->apic_id_limit
+ nb_numa_nodes
);
1030 numa_fw_cfg
[0] = cpu_to_le64(nb_numa_nodes
);
1031 cpus
= mc
->possible_cpu_arch_ids(MACHINE(pcms
));
1032 for (i
= 0; i
< cpus
->len
; i
++) {
1033 unsigned int apic_id
= cpus
->cpus
[i
].arch_id
;
1034 assert(apic_id
< pcms
->apic_id_limit
);
1035 numa_fw_cfg
[apic_id
+ 1] = cpu_to_le64(cpus
->cpus
[i
].props
.node_id
);
1037 for (i
= 0; i
< nb_numa_nodes
; i
++) {
1038 numa_fw_cfg
[pcms
->apic_id_limit
+ 1 + i
] =
1039 cpu_to_le64(numa_info
[i
].node_mem
);
1041 fw_cfg_add_bytes(fw_cfg
, FW_CFG_NUMA
, numa_fw_cfg
,
1042 (1 + pcms
->apic_id_limit
+ nb_numa_nodes
) *
1043 sizeof(*numa_fw_cfg
));
1048 static long get_file_size(FILE *f
)
1052 /* XXX: on Unix systems, using fstat() probably makes more sense */
1055 fseek(f
, 0, SEEK_END
);
1057 fseek(f
, where
, SEEK_SET
);
1062 /* setup_data types */
1063 #define SETUP_NONE 0
1064 #define SETUP_E820_EXT 1
1074 } __attribute__((packed
));
1078 * The entry point into the kernel for PVH boot is different from
1079 * the native entry point. The PVH entry is defined by the x86/HVM
1080 * direct boot ABI and is available in an ELFNOTE in the kernel binary.
1082 * This function is passed to load_elf() when it is called from
1083 * load_elfboot() which then additionally checks for an ELF Note of
1084 * type XEN_ELFNOTE_PHYS32_ENTRY and passes it to this function to
1085 * parse the PVH entry address from the ELF Note.
1087 * Due to trickery in elf_opts.h, load_elf() is actually available as
1088 * load_elf32() or load_elf64() and this routine needs to be able
1089 * to deal with being called as 32 or 64 bit.
1091 * The address of the PVH entry point is saved to the 'pvh_start_addr'
1092 * global variable. (although the entry point is 32-bit, the kernel
1093 * binary can be either 32-bit or 64-bit).
1095 static uint64_t read_pvh_start_addr(void *arg1
, void *arg2
, bool is64
)
1097 size_t *elf_note_data_addr
;
1099 /* Check if ELF Note header passed in is valid */
1105 struct elf64_note
*nhdr64
= (struct elf64_note
*)arg1
;
1106 uint64_t nhdr_size64
= sizeof(struct elf64_note
);
1107 uint64_t phdr_align
= *(uint64_t *)arg2
;
1108 uint64_t nhdr_namesz
= nhdr64
->n_namesz
;
1110 elf_note_data_addr
=
1111 ((void *)nhdr64
) + nhdr_size64
+
1112 QEMU_ALIGN_UP(nhdr_namesz
, phdr_align
);
1114 struct elf32_note
*nhdr32
= (struct elf32_note
*)arg1
;
1115 uint32_t nhdr_size32
= sizeof(struct elf32_note
);
1116 uint32_t phdr_align
= *(uint32_t *)arg2
;
1117 uint32_t nhdr_namesz
= nhdr32
->n_namesz
;
1119 elf_note_data_addr
=
1120 ((void *)nhdr32
) + nhdr_size32
+
1121 QEMU_ALIGN_UP(nhdr_namesz
, phdr_align
);
1124 pvh_start_addr
= *elf_note_data_addr
;
1126 return pvh_start_addr
;
1129 static bool load_elfboot(const char *kernel_filename
,
1130 int kernel_file_size
,
1132 size_t pvh_xen_start_addr
,
1136 uint32_t mh_load_addr
= 0;
1137 uint32_t elf_kernel_size
= 0;
1139 uint64_t elf_low
, elf_high
;
1142 if (ldl_p(header
) != 0x464c457f) {
1143 return false; /* no elfboot */
1146 bool elf_is64
= header
[EI_CLASS
] == ELFCLASS64
;
1148 ((Elf64_Ehdr
*)header
)->e_flags
: ((Elf32_Ehdr
*)header
)->e_flags
;
1150 if (flags
& 0x00010004) { /* LOAD_ELF_HEADER_HAS_ADDR */
1151 error_report("elfboot unsupported flags = %x", flags
);
1155 uint64_t elf_note_type
= XEN_ELFNOTE_PHYS32_ENTRY
;
1156 kernel_size
= load_elf(kernel_filename
, read_pvh_start_addr
,
1157 NULL
, &elf_note_type
, &elf_entry
,
1158 &elf_low
, &elf_high
, 0, I386_ELF_MACHINE
,
1161 if (kernel_size
< 0) {
1162 error_report("Error while loading elf kernel");
1165 mh_load_addr
= elf_low
;
1166 elf_kernel_size
= elf_high
- elf_low
;
1168 if (pvh_start_addr
== 0) {
1169 error_report("Error loading uncompressed kernel without PVH ELF Note");
1172 fw_cfg_add_i32(fw_cfg
, FW_CFG_KERNEL_ENTRY
, pvh_start_addr
);
1173 fw_cfg_add_i32(fw_cfg
, FW_CFG_KERNEL_ADDR
, mh_load_addr
);
1174 fw_cfg_add_i32(fw_cfg
, FW_CFG_KERNEL_SIZE
, elf_kernel_size
);
1179 static void load_linux(PCMachineState
*pcms
,
1183 int setup_size
, kernel_size
, cmdline_size
;
1184 int dtb_size
, setup_data_offset
;
1185 uint32_t initrd_max
;
1186 uint8_t header
[8192], *setup
, *kernel
;
1187 hwaddr real_addr
, prot_addr
, cmdline_addr
, initrd_addr
= 0;
1190 MachineState
*machine
= MACHINE(pcms
);
1191 PCMachineClass
*pcmc
= PC_MACHINE_GET_CLASS(pcms
);
1192 struct setup_data
*setup_data
;
1193 const char *kernel_filename
= machine
->kernel_filename
;
1194 const char *initrd_filename
= machine
->initrd_filename
;
1195 const char *dtb_filename
= machine
->dtb
;
1196 const char *kernel_cmdline
= machine
->kernel_cmdline
;
1198 /* Align to 16 bytes as a paranoia measure */
1199 cmdline_size
= (strlen(kernel_cmdline
)+16) & ~15;
1201 /* load the kernel header */
1202 f
= fopen(kernel_filename
, "rb");
1203 if (!f
|| !(kernel_size
= get_file_size(f
)) ||
1204 fread(header
, 1, MIN(ARRAY_SIZE(header
), kernel_size
), f
) !=
1205 MIN(ARRAY_SIZE(header
), kernel_size
)) {
1206 fprintf(stderr
, "qemu: could not load kernel '%s': %s\n",
1207 kernel_filename
, strerror(errno
));
1211 /* kernel protocol version */
1213 fprintf(stderr
, "header magic: %#x\n", ldl_p(header
+0x202));
1215 if (ldl_p(header
+0x202) == 0x53726448) {
1216 protocol
= lduw_p(header
+0x206);
1219 * Check if the file is an uncompressed kernel file (ELF) and load it,
1220 * saving the PVH entry point used by the x86/HVM direct boot ABI.
1221 * If load_elfboot() is successful, populate the fw_cfg info.
1223 if (load_elfboot(kernel_filename
, kernel_size
,
1224 header
, pvh_start_addr
, fw_cfg
)) {
1225 struct hvm_modlist_entry ramdisk_mod
= { 0 };
1229 fw_cfg_add_i32(fw_cfg
, FW_CFG_CMDLINE_SIZE
,
1230 strlen(kernel_cmdline
) + 1);
1231 fw_cfg_add_string(fw_cfg
, FW_CFG_CMDLINE_DATA
, kernel_cmdline
);
1233 assert(machine
->device_memory
!= NULL
);
1234 ramdisk_mod
.paddr
= machine
->device_memory
->base
;
1236 memory_region_size(&machine
->device_memory
->mr
);
1238 fw_cfg_add_bytes(fw_cfg
, FW_CFG_KERNEL_DATA
, &ramdisk_mod
,
1239 sizeof(ramdisk_mod
));
1240 fw_cfg_add_i32(fw_cfg
, FW_CFG_SETUP_SIZE
, sizeof(header
));
1241 fw_cfg_add_bytes(fw_cfg
, FW_CFG_SETUP_DATA
,
1242 header
, sizeof(header
));
1246 /* This looks like a multiboot kernel. If it is, let's stop
1247 treating it like a Linux kernel. */
1248 if (load_multiboot(fw_cfg
, f
, kernel_filename
, initrd_filename
,
1249 kernel_cmdline
, kernel_size
, header
)) {
1255 if (protocol
< 0x200 || !(header
[0x211] & 0x01)) {
1257 real_addr
= 0x90000;
1258 cmdline_addr
= 0x9a000 - cmdline_size
;
1259 prot_addr
= 0x10000;
1260 } else if (protocol
< 0x202) {
1261 /* High but ancient kernel */
1262 real_addr
= 0x90000;
1263 cmdline_addr
= 0x9a000 - cmdline_size
;
1264 prot_addr
= 0x100000;
1266 /* High and recent kernel */
1267 real_addr
= 0x10000;
1268 cmdline_addr
= 0x20000;
1269 prot_addr
= 0x100000;
1274 "qemu: real_addr = 0x" TARGET_FMT_plx
"\n"
1275 "qemu: cmdline_addr = 0x" TARGET_FMT_plx
"\n"
1276 "qemu: prot_addr = 0x" TARGET_FMT_plx
"\n",
1282 /* highest address for loading the initrd */
1283 if (protocol
>= 0x203) {
1284 initrd_max
= ldl_p(header
+0x22c);
1286 initrd_max
= 0x37ffffff;
1289 if (initrd_max
>= pcms
->below_4g_mem_size
- pcmc
->acpi_data_size
) {
1290 initrd_max
= pcms
->below_4g_mem_size
- pcmc
->acpi_data_size
- 1;
1293 fw_cfg_add_i32(fw_cfg
, FW_CFG_CMDLINE_ADDR
, cmdline_addr
);
1294 fw_cfg_add_i32(fw_cfg
, FW_CFG_CMDLINE_SIZE
, strlen(kernel_cmdline
)+1);
1295 fw_cfg_add_string(fw_cfg
, FW_CFG_CMDLINE_DATA
, kernel_cmdline
);
1297 if (protocol
>= 0x202) {
1298 stl_p(header
+0x228, cmdline_addr
);
1300 stw_p(header
+0x20, 0xA33F);
1301 stw_p(header
+0x22, cmdline_addr
-real_addr
);
1304 /* handle vga= parameter */
1305 vmode
= strstr(kernel_cmdline
, "vga=");
1307 unsigned int video_mode
;
1310 if (!strncmp(vmode
, "normal", 6)) {
1311 video_mode
= 0xffff;
1312 } else if (!strncmp(vmode
, "ext", 3)) {
1313 video_mode
= 0xfffe;
1314 } else if (!strncmp(vmode
, "ask", 3)) {
1315 video_mode
= 0xfffd;
1317 video_mode
= strtol(vmode
, NULL
, 0);
1319 stw_p(header
+0x1fa, video_mode
);
1323 /* High nybble = B reserved for QEMU; low nybble is revision number.
1324 If this code is substantially changed, you may want to consider
1325 incrementing the revision. */
1326 if (protocol
>= 0x200) {
1327 header
[0x210] = 0xB0;
1330 if (protocol
>= 0x201) {
1331 header
[0x211] |= 0x80; /* CAN_USE_HEAP */
1332 stw_p(header
+0x224, cmdline_addr
-real_addr
-0x200);
1336 if (initrd_filename
) {
1339 GError
*gerr
= NULL
;
1341 if (protocol
< 0x200) {
1342 fprintf(stderr
, "qemu: linux kernel too old to load a ram disk\n");
1346 if (!g_file_get_contents(initrd_filename
, &initrd_data
,
1347 &initrd_size
, &gerr
)) {
1348 fprintf(stderr
, "qemu: error reading initrd %s: %s\n",
1349 initrd_filename
, gerr
->message
);
1352 if (initrd_size
>= initrd_max
) {
1353 fprintf(stderr
, "qemu: initrd is too large, cannot support."
1354 "(max: %"PRIu32
", need %"PRId64
")\n",
1355 initrd_max
, (uint64_t)initrd_size
);
1359 initrd_addr
= (initrd_max
-initrd_size
) & ~4095;
1361 fw_cfg_add_i32(fw_cfg
, FW_CFG_INITRD_ADDR
, initrd_addr
);
1362 fw_cfg_add_i32(fw_cfg
, FW_CFG_INITRD_SIZE
, initrd_size
);
1363 fw_cfg_add_bytes(fw_cfg
, FW_CFG_INITRD_DATA
, initrd_data
, initrd_size
);
1365 stl_p(header
+0x218, initrd_addr
);
1366 stl_p(header
+0x21c, initrd_size
);
1369 /* load kernel and setup */
1370 setup_size
= header
[0x1f1];
1371 if (setup_size
== 0) {
1374 setup_size
= (setup_size
+1)*512;
1375 if (setup_size
> kernel_size
) {
1376 fprintf(stderr
, "qemu: invalid kernel header\n");
1379 kernel_size
-= setup_size
;
1381 setup
= g_malloc(setup_size
);
1382 kernel
= g_malloc(kernel_size
);
1383 fseek(f
, 0, SEEK_SET
);
1384 if (fread(setup
, 1, setup_size
, f
) != setup_size
) {
1385 fprintf(stderr
, "fread() failed\n");
1388 if (fread(kernel
, 1, kernel_size
, f
) != kernel_size
) {
1389 fprintf(stderr
, "fread() failed\n");
1394 /* append dtb to kernel */
1396 if (protocol
< 0x209) {
1397 fprintf(stderr
, "qemu: Linux kernel too old to load a dtb\n");
1401 dtb_size
= get_image_size(dtb_filename
);
1402 if (dtb_size
<= 0) {
1403 fprintf(stderr
, "qemu: error reading dtb %s: %s\n",
1404 dtb_filename
, strerror(errno
));
1408 setup_data_offset
= QEMU_ALIGN_UP(kernel_size
, 16);
1409 kernel_size
= setup_data_offset
+ sizeof(struct setup_data
) + dtb_size
;
1410 kernel
= g_realloc(kernel
, kernel_size
);
1412 stq_p(header
+0x250, prot_addr
+ setup_data_offset
);
1414 setup_data
= (struct setup_data
*)(kernel
+ setup_data_offset
);
1415 setup_data
->next
= 0;
1416 setup_data
->type
= cpu_to_le32(SETUP_DTB
);
1417 setup_data
->len
= cpu_to_le32(dtb_size
);
1419 load_image_size(dtb_filename
, setup_data
->data
, dtb_size
);
1422 memcpy(setup
, header
, MIN(sizeof(header
), setup_size
));
1424 fw_cfg_add_i32(fw_cfg
, FW_CFG_KERNEL_ADDR
, prot_addr
);
1425 fw_cfg_add_i32(fw_cfg
, FW_CFG_KERNEL_SIZE
, kernel_size
);
1426 fw_cfg_add_bytes(fw_cfg
, FW_CFG_KERNEL_DATA
, kernel
, kernel_size
);
1428 fw_cfg_add_i32(fw_cfg
, FW_CFG_SETUP_ADDR
, real_addr
);
1429 fw_cfg_add_i32(fw_cfg
, FW_CFG_SETUP_SIZE
, setup_size
);
1430 fw_cfg_add_bytes(fw_cfg
, FW_CFG_SETUP_DATA
, setup
, setup_size
);
1432 option_rom
[nb_option_roms
].bootindex
= 0;
1433 option_rom
[nb_option_roms
].name
= "linuxboot.bin";
1434 if (pcmc
->linuxboot_dma_enabled
&& fw_cfg_dma_enabled(fw_cfg
)) {
1435 option_rom
[nb_option_roms
].name
= "linuxboot_dma.bin";
1440 #define NE2000_NB_MAX 6
1442 static const int ne2000_io
[NE2000_NB_MAX
] = { 0x300, 0x320, 0x340, 0x360,
1444 static const int ne2000_irq
[NE2000_NB_MAX
] = { 9, 10, 11, 3, 4, 5 };
1446 void pc_init_ne2k_isa(ISABus
*bus
, NICInfo
*nd
)
1448 static int nb_ne2k
= 0;
1450 if (nb_ne2k
== NE2000_NB_MAX
)
1452 isa_ne2000_init(bus
, ne2000_io
[nb_ne2k
],
1453 ne2000_irq
[nb_ne2k
], nd
);
1457 DeviceState
*cpu_get_current_apic(void)
1460 X86CPU
*cpu
= X86_CPU(current_cpu
);
1461 return cpu
->apic_state
;
1467 void pc_acpi_smi_interrupt(void *opaque
, int irq
, int level
)
1469 X86CPU
*cpu
= opaque
;
1472 cpu_interrupt(CPU(cpu
), CPU_INTERRUPT_SMI
);
1476 static void pc_new_cpu(const char *typename
, int64_t apic_id
, Error
**errp
)
1479 Error
*local_err
= NULL
;
1481 cpu
= object_new(typename
);
1483 object_property_set_uint(cpu
, apic_id
, "apic-id", &local_err
);
1484 object_property_set_bool(cpu
, true, "realized", &local_err
);
1487 error_propagate(errp
, local_err
);
1490 void pc_hot_add_cpu(const int64_t id
, Error
**errp
)
1492 MachineState
*ms
= MACHINE(qdev_get_machine());
1493 int64_t apic_id
= x86_cpu_apic_id_from_index(id
);
1494 Error
*local_err
= NULL
;
1497 error_setg(errp
, "Invalid CPU id: %" PRIi64
, id
);
1501 if (apic_id
>= ACPI_CPU_HOTPLUG_ID_LIMIT
) {
1502 error_setg(errp
, "Unable to add CPU: %" PRIi64
1503 ", resulting APIC ID (%" PRIi64
") is too large",
1508 pc_new_cpu(ms
->cpu_type
, apic_id
, &local_err
);
1510 error_propagate(errp
, local_err
);
1515 void pc_cpus_init(PCMachineState
*pcms
)
1518 const CPUArchIdList
*possible_cpus
;
1519 MachineState
*ms
= MACHINE(pcms
);
1520 MachineClass
*mc
= MACHINE_GET_CLASS(pcms
);
1522 /* Calculates the limit to CPU APIC ID values
1524 * Limit for the APIC ID value, so that all
1525 * CPU APIC IDs are < pcms->apic_id_limit.
1527 * This is used for FW_CFG_MAX_CPUS. See comments on bochs_bios_init().
1529 pcms
->apic_id_limit
= x86_cpu_apic_id_from_index(max_cpus
- 1) + 1;
1530 possible_cpus
= mc
->possible_cpu_arch_ids(ms
);
1531 for (i
= 0; i
< smp_cpus
; i
++) {
1532 pc_new_cpu(possible_cpus
->cpus
[i
].type
, possible_cpus
->cpus
[i
].arch_id
,
1537 static void pc_build_feature_control_file(PCMachineState
*pcms
)
1539 MachineState
*ms
= MACHINE(pcms
);
1540 X86CPU
*cpu
= X86_CPU(ms
->possible_cpus
->cpus
[0].cpu
);
1541 CPUX86State
*env
= &cpu
->env
;
1542 uint32_t unused
, ecx
, edx
;
1543 uint64_t feature_control_bits
= 0;
1546 cpu_x86_cpuid(env
, 1, 0, &unused
, &unused
, &ecx
, &edx
);
1547 if (ecx
& CPUID_EXT_VMX
) {
1548 feature_control_bits
|= FEATURE_CONTROL_VMXON_ENABLED_OUTSIDE_SMX
;
1551 if ((edx
& (CPUID_EXT2_MCE
| CPUID_EXT2_MCA
)) ==
1552 (CPUID_EXT2_MCE
| CPUID_EXT2_MCA
) &&
1553 (env
->mcg_cap
& MCG_LMCE_P
)) {
1554 feature_control_bits
|= FEATURE_CONTROL_LMCE
;
1557 if (!feature_control_bits
) {
1561 val
= g_malloc(sizeof(*val
));
1562 *val
= cpu_to_le64(feature_control_bits
| FEATURE_CONTROL_LOCKED
);
1563 fw_cfg_add_file(pcms
->fw_cfg
, "etc/msr_feature_control", val
, sizeof(*val
));
1566 static void rtc_set_cpus_count(ISADevice
*rtc
, uint16_t cpus_count
)
1568 if (cpus_count
> 0xff) {
1569 /* If the number of CPUs can't be represented in 8 bits, the
1570 * BIOS must use "FW_CFG_NB_CPUS". Set RTC field to 0 just
1571 * to make old BIOSes fail more predictably.
1573 rtc_set_memory(rtc
, 0x5f, 0);
1575 rtc_set_memory(rtc
, 0x5f, cpus_count
- 1);
1580 void pc_machine_done(Notifier
*notifier
, void *data
)
1582 PCMachineState
*pcms
= container_of(notifier
,
1583 PCMachineState
, machine_done
);
1584 PCIBus
*bus
= pcms
->bus
;
1586 /* set the number of CPUs */
1587 rtc_set_cpus_count(pcms
->rtc
, pcms
->boot_cpus
);
1590 int extra_hosts
= 0;
1592 QLIST_FOREACH(bus
, &bus
->child
, sibling
) {
1593 /* look for expander root buses */
1594 if (pci_bus_is_root(bus
)) {
1598 if (extra_hosts
&& pcms
->fw_cfg
) {
1599 uint64_t *val
= g_malloc(sizeof(*val
));
1600 *val
= cpu_to_le64(extra_hosts
);
1601 fw_cfg_add_file(pcms
->fw_cfg
,
1602 "etc/extra-pci-roots", val
, sizeof(*val
));
1608 pc_build_smbios(pcms
);
1609 pc_build_feature_control_file(pcms
);
1610 /* update FW_CFG_NB_CPUS to account for -device added CPUs */
1611 fw_cfg_modify_i16(pcms
->fw_cfg
, FW_CFG_NB_CPUS
, pcms
->boot_cpus
);
1614 if (pcms
->apic_id_limit
> 255 && !xen_enabled()) {
1615 IntelIOMMUState
*iommu
= INTEL_IOMMU_DEVICE(x86_iommu_get_default());
1617 if (!iommu
|| !x86_iommu_ir_supported(X86_IOMMU_DEVICE(iommu
)) ||
1618 iommu
->intr_eim
!= ON_OFF_AUTO_ON
) {
1619 error_report("current -smp configuration requires "
1620 "Extended Interrupt Mode enabled. "
1621 "You can add an IOMMU using: "
1622 "-device intel-iommu,intremap=on,eim=on");
1628 void pc_guest_info_init(PCMachineState
*pcms
)
1632 pcms
->apic_xrupt_override
= kvm_allows_irq0_override();
1633 pcms
->numa_nodes
= nb_numa_nodes
;
1634 pcms
->node_mem
= g_malloc0(pcms
->numa_nodes
*
1635 sizeof *pcms
->node_mem
);
1636 for (i
= 0; i
< nb_numa_nodes
; i
++) {
1637 pcms
->node_mem
[i
] = numa_info
[i
].node_mem
;
1640 pcms
->machine_done
.notify
= pc_machine_done
;
1641 qemu_add_machine_init_done_notifier(&pcms
->machine_done
);
1644 /* setup pci memory address space mapping into system address space */
1645 void pc_pci_as_mapping_init(Object
*owner
, MemoryRegion
*system_memory
,
1646 MemoryRegion
*pci_address_space
)
1648 /* Set to lower priority than RAM */
1649 memory_region_add_subregion_overlap(system_memory
, 0x0,
1650 pci_address_space
, -1);
1653 void pc_acpi_init(const char *default_dsdt
)
1657 if (acpi_tables
!= NULL
) {
1658 /* manually set via -acpitable, leave it alone */
1662 filename
= qemu_find_file(QEMU_FILE_TYPE_BIOS
, default_dsdt
);
1663 if (filename
== NULL
) {
1664 warn_report("failed to find %s", default_dsdt
);
1666 QemuOpts
*opts
= qemu_opts_create(qemu_find_opts("acpi"), NULL
, 0,
1670 qemu_opt_set(opts
, "file", filename
, &error_abort
);
1672 acpi_table_add_builtin(opts
, &err
);
1674 warn_reportf_err(err
, "failed to load %s: ", filename
);
1680 void xen_load_linux(PCMachineState
*pcms
)
1685 assert(MACHINE(pcms
)->kernel_filename
!= NULL
);
1687 fw_cfg
= fw_cfg_init_io(FW_CFG_IO_BASE
);
1688 fw_cfg_add_i16(fw_cfg
, FW_CFG_NB_CPUS
, pcms
->boot_cpus
);
1691 load_linux(pcms
, fw_cfg
);
1692 for (i
= 0; i
< nb_option_roms
; i
++) {
1693 assert(!strcmp(option_rom
[i
].name
, "linuxboot.bin") ||
1694 !strcmp(option_rom
[i
].name
, "linuxboot_dma.bin") ||
1695 !strcmp(option_rom
[i
].name
, "multiboot.bin"));
1696 rom_add_option(option_rom
[i
].name
, option_rom
[i
].bootindex
);
1698 pcms
->fw_cfg
= fw_cfg
;
1701 void pc_memory_init(PCMachineState
*pcms
,
1702 MemoryRegion
*system_memory
,
1703 MemoryRegion
*rom_memory
,
1704 MemoryRegion
**ram_memory
)
1707 MemoryRegion
*ram
, *option_rom_mr
;
1708 MemoryRegion
*ram_below_4g
, *ram_above_4g
;
1710 MachineState
*machine
= MACHINE(pcms
);
1711 PCMachineClass
*pcmc
= PC_MACHINE_GET_CLASS(pcms
);
1713 assert(machine
->ram_size
== pcms
->below_4g_mem_size
+
1714 pcms
->above_4g_mem_size
);
1716 linux_boot
= (machine
->kernel_filename
!= NULL
);
1718 /* Allocate RAM. We allocate it as a single memory region and use
1719 * aliases to address portions of it, mostly for backwards compatibility
1720 * with older qemus that used qemu_ram_alloc().
1722 ram
= g_malloc(sizeof(*ram
));
1723 memory_region_allocate_system_memory(ram
, NULL
, "pc.ram",
1726 ram_below_4g
= g_malloc(sizeof(*ram_below_4g
));
1727 memory_region_init_alias(ram_below_4g
, NULL
, "ram-below-4g", ram
,
1728 0, pcms
->below_4g_mem_size
);
1729 memory_region_add_subregion(system_memory
, 0, ram_below_4g
);
1730 e820_add_entry(0, pcms
->below_4g_mem_size
, E820_RAM
);
1731 if (pcms
->above_4g_mem_size
> 0) {
1732 ram_above_4g
= g_malloc(sizeof(*ram_above_4g
));
1733 memory_region_init_alias(ram_above_4g
, NULL
, "ram-above-4g", ram
,
1734 pcms
->below_4g_mem_size
,
1735 pcms
->above_4g_mem_size
);
1736 memory_region_add_subregion(system_memory
, 0x100000000ULL
,
1738 e820_add_entry(0x100000000ULL
, pcms
->above_4g_mem_size
, E820_RAM
);
1741 if (!pcmc
->has_reserved_memory
&&
1742 (machine
->ram_slots
||
1743 (machine
->maxram_size
> machine
->ram_size
))) {
1744 MachineClass
*mc
= MACHINE_GET_CLASS(machine
);
1746 error_report("\"-memory 'slots|maxmem'\" is not supported by: %s",
1751 /* always allocate the device memory information */
1752 machine
->device_memory
= g_malloc0(sizeof(*machine
->device_memory
));
1754 /* initialize device memory address space */
1755 if (pcmc
->has_reserved_memory
&&
1756 (machine
->ram_size
< machine
->maxram_size
)) {
1757 ram_addr_t device_mem_size
= machine
->maxram_size
- machine
->ram_size
;
1759 if (machine
->ram_slots
> ACPI_MAX_RAM_SLOTS
) {
1760 error_report("unsupported amount of memory slots: %"PRIu64
,
1761 machine
->ram_slots
);
1765 if (QEMU_ALIGN_UP(machine
->maxram_size
,
1766 TARGET_PAGE_SIZE
) != machine
->maxram_size
) {
1767 error_report("maximum memory size must by aligned to multiple of "
1768 "%d bytes", TARGET_PAGE_SIZE
);
1772 machine
->device_memory
->base
=
1773 ROUND_UP(0x100000000ULL
+ pcms
->above_4g_mem_size
, 1 * GiB
);
1775 if (pcmc
->enforce_aligned_dimm
) {
1776 /* size device region assuming 1G page max alignment per slot */
1777 device_mem_size
+= (1 * GiB
) * machine
->ram_slots
;
1780 if ((machine
->device_memory
->base
+ device_mem_size
) <
1782 error_report("unsupported amount of maximum memory: " RAM_ADDR_FMT
,
1783 machine
->maxram_size
);
1787 memory_region_init(&machine
->device_memory
->mr
, OBJECT(pcms
),
1788 "device-memory", device_mem_size
);
1789 memory_region_add_subregion(system_memory
, machine
->device_memory
->base
,
1790 &machine
->device_memory
->mr
);
1793 /* Initialize PC system firmware */
1794 pc_system_firmware_init(rom_memory
, !pcmc
->pci_enabled
);
1796 option_rom_mr
= g_malloc(sizeof(*option_rom_mr
));
1797 memory_region_init_ram(option_rom_mr
, NULL
, "pc.rom", PC_ROM_SIZE
,
1799 if (pcmc
->pci_enabled
) {
1800 memory_region_set_readonly(option_rom_mr
, true);
1802 memory_region_add_subregion_overlap(rom_memory
,
1807 fw_cfg
= bochs_bios_init(&address_space_memory
, pcms
);
1811 if (pcmc
->has_reserved_memory
&& machine
->device_memory
->base
) {
1812 uint64_t *val
= g_malloc(sizeof(*val
));
1813 PCMachineClass
*pcmc
= PC_MACHINE_GET_CLASS(pcms
);
1814 uint64_t res_mem_end
= machine
->device_memory
->base
;
1816 if (!pcmc
->broken_reserved_end
) {
1817 res_mem_end
+= memory_region_size(&machine
->device_memory
->mr
);
1819 *val
= cpu_to_le64(ROUND_UP(res_mem_end
, 1 * GiB
));
1820 fw_cfg_add_file(fw_cfg
, "etc/reserved-memory-end", val
, sizeof(*val
));
1824 load_linux(pcms
, fw_cfg
);
1827 for (i
= 0; i
< nb_option_roms
; i
++) {
1828 rom_add_option(option_rom
[i
].name
, option_rom
[i
].bootindex
);
1830 pcms
->fw_cfg
= fw_cfg
;
1832 /* Init default IOAPIC address space */
1833 pcms
->ioapic_as
= &address_space_memory
;
1837 * The 64bit pci hole starts after "above 4G RAM" and
1838 * potentially the space reserved for memory hotplug.
1840 uint64_t pc_pci_hole64_start(void)
1842 PCMachineState
*pcms
= PC_MACHINE(qdev_get_machine());
1843 PCMachineClass
*pcmc
= PC_MACHINE_GET_CLASS(pcms
);
1844 MachineState
*ms
= MACHINE(pcms
);
1845 uint64_t hole64_start
= 0;
1847 if (pcmc
->has_reserved_memory
&& ms
->device_memory
->base
) {
1848 hole64_start
= ms
->device_memory
->base
;
1849 if (!pcmc
->broken_reserved_end
) {
1850 hole64_start
+= memory_region_size(&ms
->device_memory
->mr
);
1853 hole64_start
= 0x100000000ULL
+ pcms
->above_4g_mem_size
;
1856 return ROUND_UP(hole64_start
, 1 * GiB
);
1859 qemu_irq
pc_allocate_cpu_irq(void)
1861 return qemu_allocate_irq(pic_irq_request
, NULL
, 0);
1864 DeviceState
*pc_vga_init(ISABus
*isa_bus
, PCIBus
*pci_bus
)
1866 DeviceState
*dev
= NULL
;
1868 rom_set_order_override(FW_CFG_ORDER_OVERRIDE_VGA
);
1870 PCIDevice
*pcidev
= pci_vga_init(pci_bus
);
1871 dev
= pcidev
? &pcidev
->qdev
: NULL
;
1872 } else if (isa_bus
) {
1873 ISADevice
*isadev
= isa_vga_init(isa_bus
);
1874 dev
= isadev
? DEVICE(isadev
) : NULL
;
1876 rom_reset_order_override();
1880 static const MemoryRegionOps ioport80_io_ops
= {
1881 .write
= ioport80_write
,
1882 .read
= ioport80_read
,
1883 .endianness
= DEVICE_NATIVE_ENDIAN
,
1885 .min_access_size
= 1,
1886 .max_access_size
= 1,
1890 static const MemoryRegionOps ioportF0_io_ops
= {
1891 .write
= ioportF0_write
,
1892 .read
= ioportF0_read
,
1893 .endianness
= DEVICE_NATIVE_ENDIAN
,
1895 .min_access_size
= 1,
1896 .max_access_size
= 1,
1900 static void pc_superio_init(ISABus
*isa_bus
, bool create_fdctrl
, bool no_vmport
)
1903 DriveInfo
*fd
[MAX_FD
];
1905 ISADevice
*i8042
, *port92
, *vmmouse
;
1907 serial_hds_isa_init(isa_bus
, 0, MAX_ISA_SERIAL_PORTS
);
1908 parallel_hds_isa_init(isa_bus
, MAX_PARALLEL_PORTS
);
1910 for (i
= 0; i
< MAX_FD
; i
++) {
1911 fd
[i
] = drive_get(IF_FLOPPY
, 0, i
);
1912 create_fdctrl
|= !!fd
[i
];
1914 if (create_fdctrl
) {
1915 fdctrl_init_isa(isa_bus
, fd
);
1918 i8042
= isa_create_simple(isa_bus
, "i8042");
1920 vmport_init(isa_bus
);
1921 vmmouse
= isa_try_create(isa_bus
, "vmmouse");
1926 DeviceState
*dev
= DEVICE(vmmouse
);
1927 qdev_prop_set_ptr(dev
, "ps2_mouse", i8042
);
1928 qdev_init_nofail(dev
);
1930 port92
= isa_create_simple(isa_bus
, "port92");
1932 a20_line
= qemu_allocate_irqs(handle_a20_line_change
, first_cpu
, 2);
1933 i8042_setup_a20_line(i8042
, a20_line
[0]);
1934 port92_init(port92
, a20_line
[1]);
1938 void pc_basic_device_init(ISABus
*isa_bus
, qemu_irq
*gsi
,
1939 ISADevice
**rtc_state
,
1946 DeviceState
*hpet
= NULL
;
1947 int pit_isa_irq
= 0;
1948 qemu_irq pit_alt_irq
= NULL
;
1949 qemu_irq rtc_irq
= NULL
;
1950 ISADevice
*pit
= NULL
;
1951 MemoryRegion
*ioport80_io
= g_new(MemoryRegion
, 1);
1952 MemoryRegion
*ioportF0_io
= g_new(MemoryRegion
, 1);
1954 memory_region_init_io(ioport80_io
, NULL
, &ioport80_io_ops
, NULL
, "ioport80", 1);
1955 memory_region_add_subregion(isa_bus
->address_space_io
, 0x80, ioport80_io
);
1957 memory_region_init_io(ioportF0_io
, NULL
, &ioportF0_io_ops
, NULL
, "ioportF0", 1);
1958 memory_region_add_subregion(isa_bus
->address_space_io
, 0xf0, ioportF0_io
);
1961 * Check if an HPET shall be created.
1963 * Without KVM_CAP_PIT_STATE2, we cannot switch off the in-kernel PIT
1964 * when the HPET wants to take over. Thus we have to disable the latter.
1966 if (!no_hpet
&& (!kvm_irqchip_in_kernel() || kvm_has_pit_state2())) {
1967 /* In order to set property, here not using sysbus_try_create_simple */
1968 hpet
= qdev_try_create(NULL
, TYPE_HPET
);
1970 /* For pc-piix-*, hpet's intcap is always IRQ2. For pc-q35-1.7
1971 * and earlier, use IRQ2 for compat. Otherwise, use IRQ16~23,
1974 uint8_t compat
= object_property_get_uint(OBJECT(hpet
),
1977 qdev_prop_set_uint32(hpet
, HPET_INTCAP
, hpet_irqs
);
1979 qdev_init_nofail(hpet
);
1980 sysbus_mmio_map(SYS_BUS_DEVICE(hpet
), 0, HPET_BASE
);
1982 for (i
= 0; i
< GSI_NUM_PINS
; i
++) {
1983 sysbus_connect_irq(SYS_BUS_DEVICE(hpet
), i
, gsi
[i
]);
1986 pit_alt_irq
= qdev_get_gpio_in(hpet
, HPET_LEGACY_PIT_INT
);
1987 rtc_irq
= qdev_get_gpio_in(hpet
, HPET_LEGACY_RTC_INT
);
1990 *rtc_state
= mc146818_rtc_init(isa_bus
, 2000, rtc_irq
);
1992 qemu_register_boot_set(pc_boot_set
, *rtc_state
);
1994 if (!xen_enabled() && has_pit
) {
1995 if (kvm_pit_in_kernel()) {
1996 pit
= kvm_pit_init(isa_bus
, 0x40);
1998 pit
= i8254_pit_init(isa_bus
, 0x40, pit_isa_irq
, pit_alt_irq
);
2001 /* connect PIT to output control line of the HPET */
2002 qdev_connect_gpio_out(hpet
, 0, qdev_get_gpio_in(DEVICE(pit
), 0));
2004 pcspk_init(isa_bus
, pit
);
2007 i8257_dma_init(isa_bus
, 0);
2010 pc_superio_init(isa_bus
, create_fdctrl
, no_vmport
);
2013 void pc_nic_init(PCMachineClass
*pcmc
, ISABus
*isa_bus
, PCIBus
*pci_bus
)
2017 rom_set_order_override(FW_CFG_ORDER_OVERRIDE_NIC
);
2018 for (i
= 0; i
< nb_nics
; i
++) {
2019 NICInfo
*nd
= &nd_table
[i
];
2020 const char *model
= nd
->model
? nd
->model
: pcmc
->default_nic_model
;
2022 if (g_str_equal(model
, "ne2k_isa")) {
2023 pc_init_ne2k_isa(isa_bus
, nd
);
2025 pci_nic_init_nofail(nd
, pci_bus
, model
, NULL
);
2028 rom_reset_order_override();
2031 void ioapic_init_gsi(GSIState
*gsi_state
, const char *parent_name
)
2037 if (kvm_ioapic_in_kernel()) {
2038 dev
= qdev_create(NULL
, TYPE_KVM_IOAPIC
);
2040 dev
= qdev_create(NULL
, TYPE_IOAPIC
);
2043 object_property_add_child(object_resolve_path(parent_name
, NULL
),
2044 "ioapic", OBJECT(dev
), NULL
);
2046 qdev_init_nofail(dev
);
2047 d
= SYS_BUS_DEVICE(dev
);
2048 sysbus_mmio_map(d
, 0, IO_APIC_DEFAULT_ADDRESS
);
2050 for (i
= 0; i
< IOAPIC_NUM_PINS
; i
++) {
2051 gsi_state
->ioapic_irq
[i
] = qdev_get_gpio_in(dev
, i
);
2055 static void pc_memory_pre_plug(HotplugHandler
*hotplug_dev
, DeviceState
*dev
,
2058 const PCMachineState
*pcms
= PC_MACHINE(hotplug_dev
);
2059 const PCMachineClass
*pcmc
= PC_MACHINE_GET_CLASS(pcms
);
2060 const bool is_nvdimm
= object_dynamic_cast(OBJECT(dev
), TYPE_NVDIMM
);
2061 const uint64_t legacy_align
= TARGET_PAGE_SIZE
;
2064 * When -no-acpi is used with Q35 machine type, no ACPI is built,
2065 * but pcms->acpi_dev is still created. Check !acpi_enabled in
2066 * addition to cover this case.
2068 if (!pcms
->acpi_dev
|| !acpi_enabled
) {
2070 "memory hotplug is not enabled: missing acpi device or acpi disabled");
2074 if (is_nvdimm
&& !pcms
->acpi_nvdimm_state
.is_enabled
) {
2075 error_setg(errp
, "nvdimm is not enabled: missing 'nvdimm' in '-M'");
2079 pc_dimm_pre_plug(PC_DIMM(dev
), MACHINE(hotplug_dev
),
2080 pcmc
->enforce_aligned_dimm
? NULL
: &legacy_align
, errp
);
2083 static void pc_memory_plug(HotplugHandler
*hotplug_dev
,
2084 DeviceState
*dev
, Error
**errp
)
2086 HotplugHandlerClass
*hhc
;
2087 Error
*local_err
= NULL
;
2088 PCMachineState
*pcms
= PC_MACHINE(hotplug_dev
);
2089 bool is_nvdimm
= object_dynamic_cast(OBJECT(dev
), TYPE_NVDIMM
);
2091 pc_dimm_plug(PC_DIMM(dev
), MACHINE(pcms
), &local_err
);
2097 nvdimm_plug(&pcms
->acpi_nvdimm_state
);
2100 hhc
= HOTPLUG_HANDLER_GET_CLASS(pcms
->acpi_dev
);
2101 hhc
->plug(HOTPLUG_HANDLER(pcms
->acpi_dev
), dev
, &error_abort
);
2103 error_propagate(errp
, local_err
);
2106 static void pc_memory_unplug_request(HotplugHandler
*hotplug_dev
,
2107 DeviceState
*dev
, Error
**errp
)
2109 HotplugHandlerClass
*hhc
;
2110 Error
*local_err
= NULL
;
2111 PCMachineState
*pcms
= PC_MACHINE(hotplug_dev
);
2114 * When -no-acpi is used with Q35 machine type, no ACPI is built,
2115 * but pcms->acpi_dev is still created. Check !acpi_enabled in
2116 * addition to cover this case.
2118 if (!pcms
->acpi_dev
|| !acpi_enabled
) {
2119 error_setg(&local_err
,
2120 "memory hotplug is not enabled: missing acpi device or acpi disabled");
2124 if (object_dynamic_cast(OBJECT(dev
), TYPE_NVDIMM
)) {
2125 error_setg(&local_err
,
2126 "nvdimm device hot unplug is not supported yet.");
2130 hhc
= HOTPLUG_HANDLER_GET_CLASS(pcms
->acpi_dev
);
2131 hhc
->unplug_request(HOTPLUG_HANDLER(pcms
->acpi_dev
), dev
, &local_err
);
2134 error_propagate(errp
, local_err
);
2137 static void pc_memory_unplug(HotplugHandler
*hotplug_dev
,
2138 DeviceState
*dev
, Error
**errp
)
2140 PCMachineState
*pcms
= PC_MACHINE(hotplug_dev
);
2141 HotplugHandlerClass
*hhc
;
2142 Error
*local_err
= NULL
;
2144 hhc
= HOTPLUG_HANDLER_GET_CLASS(pcms
->acpi_dev
);
2145 hhc
->unplug(HOTPLUG_HANDLER(pcms
->acpi_dev
), dev
, &local_err
);
2151 pc_dimm_unplug(PC_DIMM(dev
), MACHINE(pcms
));
2152 object_unparent(OBJECT(dev
));
2155 error_propagate(errp
, local_err
);
2158 static int pc_apic_cmp(const void *a
, const void *b
)
2160 CPUArchId
*apic_a
= (CPUArchId
*)a
;
2161 CPUArchId
*apic_b
= (CPUArchId
*)b
;
2163 return apic_a
->arch_id
- apic_b
->arch_id
;
2166 /* returns pointer to CPUArchId descriptor that matches CPU's apic_id
2167 * in ms->possible_cpus->cpus, if ms->possible_cpus->cpus has no
2168 * entry corresponding to CPU's apic_id returns NULL.
2170 static CPUArchId
*pc_find_cpu_slot(MachineState
*ms
, uint32_t id
, int *idx
)
2172 CPUArchId apic_id
, *found_cpu
;
2174 apic_id
.arch_id
= id
;
2175 found_cpu
= bsearch(&apic_id
, ms
->possible_cpus
->cpus
,
2176 ms
->possible_cpus
->len
, sizeof(*ms
->possible_cpus
->cpus
),
2178 if (found_cpu
&& idx
) {
2179 *idx
= found_cpu
- ms
->possible_cpus
->cpus
;
2184 static void pc_cpu_plug(HotplugHandler
*hotplug_dev
,
2185 DeviceState
*dev
, Error
**errp
)
2187 CPUArchId
*found_cpu
;
2188 HotplugHandlerClass
*hhc
;
2189 Error
*local_err
= NULL
;
2190 X86CPU
*cpu
= X86_CPU(dev
);
2191 PCMachineState
*pcms
= PC_MACHINE(hotplug_dev
);
2193 if (pcms
->acpi_dev
) {
2194 hhc
= HOTPLUG_HANDLER_GET_CLASS(pcms
->acpi_dev
);
2195 hhc
->plug(HOTPLUG_HANDLER(pcms
->acpi_dev
), dev
, &local_err
);
2201 /* increment the number of CPUs */
2204 rtc_set_cpus_count(pcms
->rtc
, pcms
->boot_cpus
);
2207 fw_cfg_modify_i16(pcms
->fw_cfg
, FW_CFG_NB_CPUS
, pcms
->boot_cpus
);
2210 found_cpu
= pc_find_cpu_slot(MACHINE(pcms
), cpu
->apic_id
, NULL
);
2211 found_cpu
->cpu
= OBJECT(dev
);
2213 error_propagate(errp
, local_err
);
2215 static void pc_cpu_unplug_request_cb(HotplugHandler
*hotplug_dev
,
2216 DeviceState
*dev
, Error
**errp
)
2219 HotplugHandlerClass
*hhc
;
2220 Error
*local_err
= NULL
;
2221 X86CPU
*cpu
= X86_CPU(dev
);
2222 PCMachineState
*pcms
= PC_MACHINE(hotplug_dev
);
2224 if (!pcms
->acpi_dev
) {
2225 error_setg(&local_err
, "CPU hot unplug not supported without ACPI");
2229 pc_find_cpu_slot(MACHINE(pcms
), cpu
->apic_id
, &idx
);
2232 error_setg(&local_err
, "Boot CPU is unpluggable");
2236 hhc
= HOTPLUG_HANDLER_GET_CLASS(pcms
->acpi_dev
);
2237 hhc
->unplug_request(HOTPLUG_HANDLER(pcms
->acpi_dev
), dev
, &local_err
);
2244 error_propagate(errp
, local_err
);
2248 static void pc_cpu_unplug_cb(HotplugHandler
*hotplug_dev
,
2249 DeviceState
*dev
, Error
**errp
)
2251 CPUArchId
*found_cpu
;
2252 HotplugHandlerClass
*hhc
;
2253 Error
*local_err
= NULL
;
2254 X86CPU
*cpu
= X86_CPU(dev
);
2255 PCMachineState
*pcms
= PC_MACHINE(hotplug_dev
);
2257 hhc
= HOTPLUG_HANDLER_GET_CLASS(pcms
->acpi_dev
);
2258 hhc
->unplug(HOTPLUG_HANDLER(pcms
->acpi_dev
), dev
, &local_err
);
2264 found_cpu
= pc_find_cpu_slot(MACHINE(pcms
), cpu
->apic_id
, NULL
);
2265 found_cpu
->cpu
= NULL
;
2266 object_unparent(OBJECT(dev
));
2268 /* decrement the number of CPUs */
2270 /* Update the number of CPUs in CMOS */
2271 rtc_set_cpus_count(pcms
->rtc
, pcms
->boot_cpus
);
2272 fw_cfg_modify_i16(pcms
->fw_cfg
, FW_CFG_NB_CPUS
, pcms
->boot_cpus
);
2274 error_propagate(errp
, local_err
);
2277 static void pc_cpu_pre_plug(HotplugHandler
*hotplug_dev
,
2278 DeviceState
*dev
, Error
**errp
)
2282 CPUArchId
*cpu_slot
;
2283 X86CPUTopoInfo topo
;
2284 X86CPU
*cpu
= X86_CPU(dev
);
2285 MachineState
*ms
= MACHINE(hotplug_dev
);
2286 PCMachineState
*pcms
= PC_MACHINE(hotplug_dev
);
2288 if(!object_dynamic_cast(OBJECT(cpu
), ms
->cpu_type
)) {
2289 error_setg(errp
, "Invalid CPU type, expected cpu type: '%s'",
2294 /* if APIC ID is not set, set it based on socket/core/thread properties */
2295 if (cpu
->apic_id
== UNASSIGNED_APIC_ID
) {
2296 int max_socket
= (max_cpus
- 1) / smp_threads
/ smp_cores
;
2298 if (cpu
->socket_id
< 0) {
2299 error_setg(errp
, "CPU socket-id is not set");
2301 } else if (cpu
->socket_id
> max_socket
) {
2302 error_setg(errp
, "Invalid CPU socket-id: %u must be in range 0:%u",
2303 cpu
->socket_id
, max_socket
);
2306 if (cpu
->core_id
< 0) {
2307 error_setg(errp
, "CPU core-id is not set");
2309 } else if (cpu
->core_id
> (smp_cores
- 1)) {
2310 error_setg(errp
, "Invalid CPU core-id: %u must be in range 0:%u",
2311 cpu
->core_id
, smp_cores
- 1);
2314 if (cpu
->thread_id
< 0) {
2315 error_setg(errp
, "CPU thread-id is not set");
2317 } else if (cpu
->thread_id
> (smp_threads
- 1)) {
2318 error_setg(errp
, "Invalid CPU thread-id: %u must be in range 0:%u",
2319 cpu
->thread_id
, smp_threads
- 1);
2323 topo
.pkg_id
= cpu
->socket_id
;
2324 topo
.core_id
= cpu
->core_id
;
2325 topo
.smt_id
= cpu
->thread_id
;
2326 cpu
->apic_id
= apicid_from_topo_ids(smp_cores
, smp_threads
, &topo
);
2329 cpu_slot
= pc_find_cpu_slot(MACHINE(pcms
), cpu
->apic_id
, &idx
);
2331 MachineState
*ms
= MACHINE(pcms
);
2333 x86_topo_ids_from_apicid(cpu
->apic_id
, smp_cores
, smp_threads
, &topo
);
2334 error_setg(errp
, "Invalid CPU [socket: %u, core: %u, thread: %u] with"
2335 " APIC ID %" PRIu32
", valid index range 0:%d",
2336 topo
.pkg_id
, topo
.core_id
, topo
.smt_id
, cpu
->apic_id
,
2337 ms
->possible_cpus
->len
- 1);
2341 if (cpu_slot
->cpu
) {
2342 error_setg(errp
, "CPU[%d] with APIC ID %" PRIu32
" exists",
2347 /* if 'address' properties socket-id/core-id/thread-id are not set, set them
2348 * so that machine_query_hotpluggable_cpus would show correct values
2350 /* TODO: move socket_id/core_id/thread_id checks into x86_cpu_realizefn()
2351 * once -smp refactoring is complete and there will be CPU private
2352 * CPUState::nr_cores and CPUState::nr_threads fields instead of globals */
2353 x86_topo_ids_from_apicid(cpu
->apic_id
, smp_cores
, smp_threads
, &topo
);
2354 if (cpu
->socket_id
!= -1 && cpu
->socket_id
!= topo
.pkg_id
) {
2355 error_setg(errp
, "property socket-id: %u doesn't match set apic-id:"
2356 " 0x%x (socket-id: %u)", cpu
->socket_id
, cpu
->apic_id
, topo
.pkg_id
);
2359 cpu
->socket_id
= topo
.pkg_id
;
2361 if (cpu
->core_id
!= -1 && cpu
->core_id
!= topo
.core_id
) {
2362 error_setg(errp
, "property core-id: %u doesn't match set apic-id:"
2363 " 0x%x (core-id: %u)", cpu
->core_id
, cpu
->apic_id
, topo
.core_id
);
2366 cpu
->core_id
= topo
.core_id
;
2368 if (cpu
->thread_id
!= -1 && cpu
->thread_id
!= topo
.smt_id
) {
2369 error_setg(errp
, "property thread-id: %u doesn't match set apic-id:"
2370 " 0x%x (thread-id: %u)", cpu
->thread_id
, cpu
->apic_id
, topo
.smt_id
);
2373 cpu
->thread_id
= topo
.smt_id
;
2375 if (cpu
->hyperv_vpindex
&& !kvm_hv_vpindex_settable()) {
2376 error_setg(errp
, "kernel doesn't allow setting HyperV VP_INDEX");
2381 cs
->cpu_index
= idx
;
2383 numa_cpu_pre_plug(cpu_slot
, dev
, errp
);
2386 static void pc_machine_device_pre_plug_cb(HotplugHandler
*hotplug_dev
,
2387 DeviceState
*dev
, Error
**errp
)
2389 if (object_dynamic_cast(OBJECT(dev
), TYPE_PC_DIMM
)) {
2390 pc_memory_pre_plug(hotplug_dev
, dev
, errp
);
2391 } else if (object_dynamic_cast(OBJECT(dev
), TYPE_CPU
)) {
2392 pc_cpu_pre_plug(hotplug_dev
, dev
, errp
);
2396 static void pc_machine_device_plug_cb(HotplugHandler
*hotplug_dev
,
2397 DeviceState
*dev
, Error
**errp
)
2399 if (object_dynamic_cast(OBJECT(dev
), TYPE_PC_DIMM
)) {
2400 pc_memory_plug(hotplug_dev
, dev
, errp
);
2401 } else if (object_dynamic_cast(OBJECT(dev
), TYPE_CPU
)) {
2402 pc_cpu_plug(hotplug_dev
, dev
, errp
);
2406 static void pc_machine_device_unplug_request_cb(HotplugHandler
*hotplug_dev
,
2407 DeviceState
*dev
, Error
**errp
)
2409 if (object_dynamic_cast(OBJECT(dev
), TYPE_PC_DIMM
)) {
2410 pc_memory_unplug_request(hotplug_dev
, dev
, errp
);
2411 } else if (object_dynamic_cast(OBJECT(dev
), TYPE_CPU
)) {
2412 pc_cpu_unplug_request_cb(hotplug_dev
, dev
, errp
);
2414 error_setg(errp
, "acpi: device unplug request for not supported device"
2415 " type: %s", object_get_typename(OBJECT(dev
)));
2419 static void pc_machine_device_unplug_cb(HotplugHandler
*hotplug_dev
,
2420 DeviceState
*dev
, Error
**errp
)
2422 if (object_dynamic_cast(OBJECT(dev
), TYPE_PC_DIMM
)) {
2423 pc_memory_unplug(hotplug_dev
, dev
, errp
);
2424 } else if (object_dynamic_cast(OBJECT(dev
), TYPE_CPU
)) {
2425 pc_cpu_unplug_cb(hotplug_dev
, dev
, errp
);
2427 error_setg(errp
, "acpi: device unplug for not supported device"
2428 " type: %s", object_get_typename(OBJECT(dev
)));
2432 static HotplugHandler
*pc_get_hotplug_handler(MachineState
*machine
,
2435 if (object_dynamic_cast(OBJECT(dev
), TYPE_PC_DIMM
) ||
2436 object_dynamic_cast(OBJECT(dev
), TYPE_CPU
)) {
2437 return HOTPLUG_HANDLER(machine
);
2444 pc_machine_get_device_memory_region_size(Object
*obj
, Visitor
*v
,
2445 const char *name
, void *opaque
,
2448 MachineState
*ms
= MACHINE(obj
);
2449 int64_t value
= memory_region_size(&ms
->device_memory
->mr
);
2451 visit_type_int(v
, name
, &value
, errp
);
2454 static void pc_machine_get_max_ram_below_4g(Object
*obj
, Visitor
*v
,
2455 const char *name
, void *opaque
,
2458 PCMachineState
*pcms
= PC_MACHINE(obj
);
2459 uint64_t value
= pcms
->max_ram_below_4g
;
2461 visit_type_size(v
, name
, &value
, errp
);
2464 static void pc_machine_set_max_ram_below_4g(Object
*obj
, Visitor
*v
,
2465 const char *name
, void *opaque
,
2468 PCMachineState
*pcms
= PC_MACHINE(obj
);
2469 Error
*error
= NULL
;
2472 visit_type_size(v
, name
, &value
, &error
);
2474 error_propagate(errp
, error
);
2477 if (value
> 4 * GiB
) {
2479 "Machine option 'max-ram-below-4g=%"PRIu64
2480 "' expects size less than or equal to 4G", value
);
2481 error_propagate(errp
, error
);
2485 if (value
< 1 * MiB
) {
2486 warn_report("Only %" PRIu64
" bytes of RAM below the 4GiB boundary,"
2487 "BIOS may not work with less than 1MiB", value
);
2490 pcms
->max_ram_below_4g
= value
;
2493 static void pc_machine_get_vmport(Object
*obj
, Visitor
*v
, const char *name
,
2494 void *opaque
, Error
**errp
)
2496 PCMachineState
*pcms
= PC_MACHINE(obj
);
2497 OnOffAuto vmport
= pcms
->vmport
;
2499 visit_type_OnOffAuto(v
, name
, &vmport
, errp
);
2502 static void pc_machine_set_vmport(Object
*obj
, Visitor
*v
, const char *name
,
2503 void *opaque
, Error
**errp
)
2505 PCMachineState
*pcms
= PC_MACHINE(obj
);
2507 visit_type_OnOffAuto(v
, name
, &pcms
->vmport
, errp
);
2510 bool pc_machine_is_smm_enabled(PCMachineState
*pcms
)
2512 bool smm_available
= false;
2514 if (pcms
->smm
== ON_OFF_AUTO_OFF
) {
2518 if (tcg_enabled() || qtest_enabled()) {
2519 smm_available
= true;
2520 } else if (kvm_enabled()) {
2521 smm_available
= kvm_has_smm();
2524 if (smm_available
) {
2528 if (pcms
->smm
== ON_OFF_AUTO_ON
) {
2529 error_report("System Management Mode not supported by this hypervisor.");
2535 static void pc_machine_get_smm(Object
*obj
, Visitor
*v
, const char *name
,
2536 void *opaque
, Error
**errp
)
2538 PCMachineState
*pcms
= PC_MACHINE(obj
);
2539 OnOffAuto smm
= pcms
->smm
;
2541 visit_type_OnOffAuto(v
, name
, &smm
, errp
);
2544 static void pc_machine_set_smm(Object
*obj
, Visitor
*v
, const char *name
,
2545 void *opaque
, Error
**errp
)
2547 PCMachineState
*pcms
= PC_MACHINE(obj
);
2549 visit_type_OnOffAuto(v
, name
, &pcms
->smm
, errp
);
2552 static bool pc_machine_get_nvdimm(Object
*obj
, Error
**errp
)
2554 PCMachineState
*pcms
= PC_MACHINE(obj
);
2556 return pcms
->acpi_nvdimm_state
.is_enabled
;
2559 static void pc_machine_set_nvdimm(Object
*obj
, bool value
, Error
**errp
)
2561 PCMachineState
*pcms
= PC_MACHINE(obj
);
2563 pcms
->acpi_nvdimm_state
.is_enabled
= value
;
2566 static char *pc_machine_get_nvdimm_persistence(Object
*obj
, Error
**errp
)
2568 PCMachineState
*pcms
= PC_MACHINE(obj
);
2570 return g_strdup(pcms
->acpi_nvdimm_state
.persistence_string
);
2573 static void pc_machine_set_nvdimm_persistence(Object
*obj
, const char *value
,
2576 PCMachineState
*pcms
= PC_MACHINE(obj
);
2577 AcpiNVDIMMState
*nvdimm_state
= &pcms
->acpi_nvdimm_state
;
2579 if (strcmp(value
, "cpu") == 0)
2580 nvdimm_state
->persistence
= 3;
2581 else if (strcmp(value
, "mem-ctrl") == 0)
2582 nvdimm_state
->persistence
= 2;
2584 error_setg(errp
, "-machine nvdimm-persistence=%s: unsupported option",
2589 g_free(nvdimm_state
->persistence_string
);
2590 nvdimm_state
->persistence_string
= g_strdup(value
);
2593 static bool pc_machine_get_smbus(Object
*obj
, Error
**errp
)
2595 PCMachineState
*pcms
= PC_MACHINE(obj
);
2597 return pcms
->smbus_enabled
;
2600 static void pc_machine_set_smbus(Object
*obj
, bool value
, Error
**errp
)
2602 PCMachineState
*pcms
= PC_MACHINE(obj
);
2604 pcms
->smbus_enabled
= value
;
2607 static bool pc_machine_get_sata(Object
*obj
, Error
**errp
)
2609 PCMachineState
*pcms
= PC_MACHINE(obj
);
2611 return pcms
->sata_enabled
;
2614 static void pc_machine_set_sata(Object
*obj
, bool value
, Error
**errp
)
2616 PCMachineState
*pcms
= PC_MACHINE(obj
);
2618 pcms
->sata_enabled
= value
;
2621 static bool pc_machine_get_pit(Object
*obj
, Error
**errp
)
2623 PCMachineState
*pcms
= PC_MACHINE(obj
);
2625 return pcms
->pit_enabled
;
2628 static void pc_machine_set_pit(Object
*obj
, bool value
, Error
**errp
)
2630 PCMachineState
*pcms
= PC_MACHINE(obj
);
2632 pcms
->pit_enabled
= value
;
2635 static void pc_machine_initfn(Object
*obj
)
2637 PCMachineState
*pcms
= PC_MACHINE(obj
);
2639 pcms
->max_ram_below_4g
= 0; /* use default */
2640 pcms
->smm
= ON_OFF_AUTO_AUTO
;
2641 pcms
->vmport
= ON_OFF_AUTO_AUTO
;
2642 /* nvdimm is disabled on default. */
2643 pcms
->acpi_nvdimm_state
.is_enabled
= false;
2644 /* acpi build is enabled by default if machine supports it */
2645 pcms
->acpi_build_enabled
= PC_MACHINE_GET_CLASS(pcms
)->has_acpi_build
;
2646 pcms
->smbus_enabled
= true;
2647 pcms
->sata_enabled
= true;
2648 pcms
->pit_enabled
= true;
2651 static void pc_machine_reset(void)
2656 qemu_devices_reset();
2658 /* Reset APIC after devices have been reset to cancel
2659 * any changes that qemu_devices_reset() might have done.
2664 if (cpu
->apic_state
) {
2665 device_reset(cpu
->apic_state
);
2670 static CpuInstanceProperties
2671 pc_cpu_index_to_props(MachineState
*ms
, unsigned cpu_index
)
2673 MachineClass
*mc
= MACHINE_GET_CLASS(ms
);
2674 const CPUArchIdList
*possible_cpus
= mc
->possible_cpu_arch_ids(ms
);
2676 assert(cpu_index
< possible_cpus
->len
);
2677 return possible_cpus
->cpus
[cpu_index
].props
;
2680 static int64_t pc_get_default_cpu_node_id(const MachineState
*ms
, int idx
)
2682 X86CPUTopoInfo topo
;
2684 assert(idx
< ms
->possible_cpus
->len
);
2685 x86_topo_ids_from_apicid(ms
->possible_cpus
->cpus
[idx
].arch_id
,
2686 smp_cores
, smp_threads
, &topo
);
2687 return topo
.pkg_id
% nb_numa_nodes
;
2690 static const CPUArchIdList
*pc_possible_cpu_arch_ids(MachineState
*ms
)
2694 if (ms
->possible_cpus
) {
2696 * make sure that max_cpus hasn't changed since the first use, i.e.
2697 * -smp hasn't been parsed after it
2699 assert(ms
->possible_cpus
->len
== max_cpus
);
2700 return ms
->possible_cpus
;
2703 ms
->possible_cpus
= g_malloc0(sizeof(CPUArchIdList
) +
2704 sizeof(CPUArchId
) * max_cpus
);
2705 ms
->possible_cpus
->len
= max_cpus
;
2706 for (i
= 0; i
< ms
->possible_cpus
->len
; i
++) {
2707 X86CPUTopoInfo topo
;
2709 ms
->possible_cpus
->cpus
[i
].type
= ms
->cpu_type
;
2710 ms
->possible_cpus
->cpus
[i
].vcpus_count
= 1;
2711 ms
->possible_cpus
->cpus
[i
].arch_id
= x86_cpu_apic_id_from_index(i
);
2712 x86_topo_ids_from_apicid(ms
->possible_cpus
->cpus
[i
].arch_id
,
2713 smp_cores
, smp_threads
, &topo
);
2714 ms
->possible_cpus
->cpus
[i
].props
.has_socket_id
= true;
2715 ms
->possible_cpus
->cpus
[i
].props
.socket_id
= topo
.pkg_id
;
2716 ms
->possible_cpus
->cpus
[i
].props
.has_core_id
= true;
2717 ms
->possible_cpus
->cpus
[i
].props
.core_id
= topo
.core_id
;
2718 ms
->possible_cpus
->cpus
[i
].props
.has_thread_id
= true;
2719 ms
->possible_cpus
->cpus
[i
].props
.thread_id
= topo
.smt_id
;
2721 return ms
->possible_cpus
;
2724 static void x86_nmi(NMIState
*n
, int cpu_index
, Error
**errp
)
2726 /* cpu index isn't used */
2730 X86CPU
*cpu
= X86_CPU(cs
);
2732 if (!cpu
->apic_state
) {
2733 cpu_interrupt(cs
, CPU_INTERRUPT_NMI
);
2735 apic_deliver_nmi(cpu
->apic_state
);
2740 static void pc_machine_class_init(ObjectClass
*oc
, void *data
)
2742 MachineClass
*mc
= MACHINE_CLASS(oc
);
2743 PCMachineClass
*pcmc
= PC_MACHINE_CLASS(oc
);
2744 HotplugHandlerClass
*hc
= HOTPLUG_HANDLER_CLASS(oc
);
2745 NMIClass
*nc
= NMI_CLASS(oc
);
2747 pcmc
->pci_enabled
= true;
2748 pcmc
->has_acpi_build
= true;
2749 pcmc
->rsdp_in_ram
= true;
2750 pcmc
->smbios_defaults
= true;
2751 pcmc
->smbios_uuid_encoded
= true;
2752 pcmc
->gigabyte_align
= true;
2753 pcmc
->has_reserved_memory
= true;
2754 pcmc
->kvmclock_enabled
= true;
2755 pcmc
->enforce_aligned_dimm
= true;
2756 /* BIOS ACPI tables: 128K. Other BIOS datastructures: less than 4K reported
2757 * to be used at the moment, 32K should be enough for a while. */
2758 pcmc
->acpi_data_size
= 0x20000 + 0x8000;
2759 pcmc
->save_tsc_khz
= true;
2760 pcmc
->linuxboot_dma_enabled
= true;
2761 assert(!mc
->get_hotplug_handler
);
2762 mc
->get_hotplug_handler
= pc_get_hotplug_handler
;
2763 mc
->cpu_index_to_instance_props
= pc_cpu_index_to_props
;
2764 mc
->get_default_cpu_node_id
= pc_get_default_cpu_node_id
;
2765 mc
->possible_cpu_arch_ids
= pc_possible_cpu_arch_ids
;
2766 mc
->auto_enable_numa_with_memhp
= true;
2767 mc
->has_hotpluggable_cpus
= true;
2768 mc
->default_boot_order
= "cad";
2769 mc
->hot_add_cpu
= pc_hot_add_cpu
;
2770 mc
->block_default_type
= IF_IDE
;
2772 mc
->reset
= pc_machine_reset
;
2773 hc
->pre_plug
= pc_machine_device_pre_plug_cb
;
2774 hc
->plug
= pc_machine_device_plug_cb
;
2775 hc
->unplug_request
= pc_machine_device_unplug_request_cb
;
2776 hc
->unplug
= pc_machine_device_unplug_cb
;
2777 nc
->nmi_monitor_handler
= x86_nmi
;
2778 mc
->default_cpu_type
= TARGET_DEFAULT_CPU_TYPE
;
2780 object_class_property_add(oc
, PC_MACHINE_DEVMEM_REGION_SIZE
, "int",
2781 pc_machine_get_device_memory_region_size
, NULL
,
2782 NULL
, NULL
, &error_abort
);
2784 object_class_property_add(oc
, PC_MACHINE_MAX_RAM_BELOW_4G
, "size",
2785 pc_machine_get_max_ram_below_4g
, pc_machine_set_max_ram_below_4g
,
2786 NULL
, NULL
, &error_abort
);
2788 object_class_property_set_description(oc
, PC_MACHINE_MAX_RAM_BELOW_4G
,
2789 "Maximum ram below the 4G boundary (32bit boundary)", &error_abort
);
2791 object_class_property_add(oc
, PC_MACHINE_SMM
, "OnOffAuto",
2792 pc_machine_get_smm
, pc_machine_set_smm
,
2793 NULL
, NULL
, &error_abort
);
2794 object_class_property_set_description(oc
, PC_MACHINE_SMM
,
2795 "Enable SMM (pc & q35)", &error_abort
);
2797 object_class_property_add(oc
, PC_MACHINE_VMPORT
, "OnOffAuto",
2798 pc_machine_get_vmport
, pc_machine_set_vmport
,
2799 NULL
, NULL
, &error_abort
);
2800 object_class_property_set_description(oc
, PC_MACHINE_VMPORT
,
2801 "Enable vmport (pc & q35)", &error_abort
);
2803 object_class_property_add_bool(oc
, PC_MACHINE_NVDIMM
,
2804 pc_machine_get_nvdimm
, pc_machine_set_nvdimm
, &error_abort
);
2806 object_class_property_add_str(oc
, PC_MACHINE_NVDIMM_PERSIST
,
2807 pc_machine_get_nvdimm_persistence
,
2808 pc_machine_set_nvdimm_persistence
, &error_abort
);
2810 object_class_property_add_bool(oc
, PC_MACHINE_SMBUS
,
2811 pc_machine_get_smbus
, pc_machine_set_smbus
, &error_abort
);
2813 object_class_property_add_bool(oc
, PC_MACHINE_SATA
,
2814 pc_machine_get_sata
, pc_machine_set_sata
, &error_abort
);
2816 object_class_property_add_bool(oc
, PC_MACHINE_PIT
,
2817 pc_machine_get_pit
, pc_machine_set_pit
, &error_abort
);
2820 static const TypeInfo pc_machine_info
= {
2821 .name
= TYPE_PC_MACHINE
,
2822 .parent
= TYPE_MACHINE
,
2824 .instance_size
= sizeof(PCMachineState
),
2825 .instance_init
= pc_machine_initfn
,
2826 .class_size
= sizeof(PCMachineClass
),
2827 .class_init
= pc_machine_class_init
,
2828 .interfaces
= (InterfaceInfo
[]) {
2829 { TYPE_HOTPLUG_HANDLER
},
2835 static void pc_machine_register_types(void)
2837 type_register_static(&pc_machine_info
);
2840 type_init(pc_machine_register_types
)