2 * QEMU PowerPC pSeries Logical Partition (aka sPAPR) hardware System Emulator
4 * Copyright (c) 2004-2007 Fabrice Bellard
5 * Copyright (c) 2007 Jocelyn Mayer
6 * Copyright (c) 2010 David Gibson, IBM Corporation.
8 * Permission is hereby granted, free of charge, to any person obtaining a copy
9 * of this software and associated documentation files (the "Software"), to deal
10 * in the Software without restriction, including without limitation the rights
11 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
12 * copies of the Software, and to permit persons to whom the Software is
13 * furnished to do so, subject to the following conditions:
15 * The above copyright notice and this permission notice shall be included in
16 * all copies or substantial portions of the Software.
18 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
19 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
20 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
21 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
22 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
23 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
27 #include "sysemu/sysemu.h"
31 #include "sysemu/blockdev.h"
32 #include "sysemu/cpus.h"
33 #include "sysemu/kvm.h"
36 #include "hw/boards.h"
38 #include "hw/loader.h"
41 #include "hw/spapr_vio.h"
42 #include "hw/spapr_pci.h"
44 #include "hw/pci/msi.h"
46 #include "sysemu/kvm.h"
50 #include "exec/address-spaces.h"
52 #include "qemu/config-file.h"
56 /* SLOF memory layout:
58 * SLOF raw image loaded at 0, copies its romfs right below the flat
59 * device-tree, then position SLOF itself 31M below that
61 * So we set FW_OVERHEAD to 40MB which should account for all of that
64 * We load our kernel at 4M, leaving space for SLOF initial image
66 #define FDT_MAX_SIZE 0x10000
67 #define RTAS_MAX_SIZE 0x10000
68 #define FW_MAX_SIZE 0x400000
69 #define FW_FILE_NAME "slof.bin"
70 #define FW_OVERHEAD 0x2800000
71 #define KERNEL_LOAD_ADDR FW_MAX_SIZE
73 #define MIN_RMA_SLOF 128UL
75 #define TIMEBASE_FREQ 512000000ULL
78 #define XICS_IRQS 1024
80 #define SPAPR_PCI_BUID 0x800000020000001ULL
81 #define SPAPR_PCI_MEM_WIN_ADDR (0x10000000000ULL + 0xA0000000)
82 #define SPAPR_PCI_MEM_WIN_SIZE 0x20000000
83 #define SPAPR_PCI_IO_WIN_ADDR (0x10000000000ULL + 0x80000000)
84 #define SPAPR_PCI_MSI_WIN_ADDR (0x10000000000ULL + 0x90000000)
86 #define PHANDLE_XICP 0x00001111
88 #define HTAB_SIZE(spapr) (1ULL << ((spapr)->htab_shift))
90 sPAPREnvironment
*spapr
;
92 int spapr_allocate_irq(int hint
, bool lsi
)
98 /* FIXME: we should probably check for collisions somehow */
100 irq
= spapr
->next_irq
++;
103 /* Configure irq type */
104 if (!xics_get_qirq(spapr
->icp
, irq
)) {
108 xics_set_irq_type(spapr
->icp
, irq
, lsi
);
113 /* Allocate block of consequtive IRQs, returns a number of the first */
114 int spapr_allocate_irq_block(int num
, bool lsi
)
119 for (i
= 0; i
< num
; ++i
) {
122 irq
= spapr_allocate_irq(0, lsi
);
131 /* If the above doesn't create a consecutive block then that's
133 assert(irq
== (first
+ i
));
139 static int spapr_fixup_cpu_dt(void *fdt
, sPAPREnvironment
*spapr
)
145 int smt
= kvmppc_smt_threads();
146 uint32_t pft_size_prop
[] = {0, cpu_to_be32(spapr
->htab_shift
)};
148 assert(spapr
->cpu_model
);
150 for (env
= first_cpu
; env
!= NULL
; env
= env
->next_cpu
) {
151 cpu
= ENV_GET_CPU(env
);
152 uint32_t associativity
[] = {cpu_to_be32(0x5),
156 cpu_to_be32(cpu
->numa_node
),
157 cpu_to_be32(env
->cpu_index
)};
159 if ((env
->cpu_index
% smt
) != 0) {
163 snprintf(cpu_model
, 32, "/cpus/%s@%x", spapr
->cpu_model
,
166 offset
= fdt_path_offset(fdt
, cpu_model
);
171 if (nb_numa_nodes
> 1) {
172 ret
= fdt_setprop(fdt
, offset
, "ibm,associativity", associativity
,
173 sizeof(associativity
));
179 ret
= fdt_setprop(fdt
, offset
, "ibm,pft-size",
180 pft_size_prop
, sizeof(pft_size_prop
));
189 static size_t create_page_sizes_prop(CPUPPCState
*env
, uint32_t *prop
,
192 size_t maxcells
= maxsize
/ sizeof(uint32_t);
196 for (i
= 0; i
< PPC_PAGE_SIZES_MAX_SZ
; i
++) {
197 struct ppc_one_seg_page_size
*sps
= &env
->sps
.sps
[i
];
199 if (!sps
->page_shift
) {
202 for (count
= 0; count
< PPC_PAGE_SIZES_MAX_SZ
; count
++) {
203 if (sps
->enc
[count
].page_shift
== 0) {
207 if ((p
- prop
) >= (maxcells
- 3 - count
* 2)) {
210 *(p
++) = cpu_to_be32(sps
->page_shift
);
211 *(p
++) = cpu_to_be32(sps
->slb_enc
);
212 *(p
++) = cpu_to_be32(count
);
213 for (j
= 0; j
< count
; j
++) {
214 *(p
++) = cpu_to_be32(sps
->enc
[j
].page_shift
);
215 *(p
++) = cpu_to_be32(sps
->enc
[j
].pte_enc
);
219 return (p
- prop
) * sizeof(uint32_t);
226 fprintf(stderr, "qemu: error creating device tree: %s: %s\n", \
227 #exp, fdt_strerror(ret)); \
233 static void *spapr_create_fdt_skel(const char *cpu_model
,
237 const char *boot_device
,
238 const char *kernel_cmdline
,
243 uint32_t start_prop
= cpu_to_be32(initrd_base
);
244 uint32_t end_prop
= cpu_to_be32(initrd_base
+ initrd_size
);
245 char hypertas_prop
[] = "hcall-pft\0hcall-term\0hcall-dabr\0hcall-interrupt"
246 "\0hcall-tce\0hcall-vio\0hcall-splpar\0hcall-bulk";
247 char qemu_hypertas_prop
[] = "hcall-memop1";
248 uint32_t refpoints
[] = {cpu_to_be32(0x4), cpu_to_be32(0x4)};
249 uint32_t interrupt_server_ranges_prop
[] = {0, cpu_to_be32(smp_cpus
)};
251 int i
, smt
= kvmppc_smt_threads();
252 unsigned char vec5
[] = {0x0, 0x0, 0x0, 0x0, 0x0, 0x80};
254 fdt
= g_malloc0(FDT_MAX_SIZE
);
255 _FDT((fdt_create(fdt
, FDT_MAX_SIZE
)));
258 _FDT((fdt_add_reservemap_entry(fdt
, KERNEL_LOAD_ADDR
, kernel_size
)));
261 _FDT((fdt_add_reservemap_entry(fdt
, initrd_base
, initrd_size
)));
263 _FDT((fdt_finish_reservemap(fdt
)));
266 _FDT((fdt_begin_node(fdt
, "")));
267 _FDT((fdt_property_string(fdt
, "device_type", "chrp")));
268 _FDT((fdt_property_string(fdt
, "model", "IBM pSeries (emulated by qemu)")));
270 _FDT((fdt_property_cell(fdt
, "#address-cells", 0x2)));
271 _FDT((fdt_property_cell(fdt
, "#size-cells", 0x2)));
274 _FDT((fdt_begin_node(fdt
, "chosen")));
276 /* Set Form1_affinity */
277 _FDT((fdt_property(fdt
, "ibm,architecture-vec-5", vec5
, sizeof(vec5
))));
279 _FDT((fdt_property_string(fdt
, "bootargs", kernel_cmdline
)));
280 _FDT((fdt_property(fdt
, "linux,initrd-start",
281 &start_prop
, sizeof(start_prop
))));
282 _FDT((fdt_property(fdt
, "linux,initrd-end",
283 &end_prop
, sizeof(end_prop
))));
285 uint64_t kprop
[2] = { cpu_to_be64(KERNEL_LOAD_ADDR
),
286 cpu_to_be64(kernel_size
) };
288 _FDT((fdt_property(fdt
, "qemu,boot-kernel", &kprop
, sizeof(kprop
))));
290 _FDT((fdt_property_string(fdt
, "qemu,boot-device", boot_device
)));
291 _FDT((fdt_property_cell(fdt
, "qemu,graphic-width", graphic_width
)));
292 _FDT((fdt_property_cell(fdt
, "qemu,graphic-height", graphic_height
)));
293 _FDT((fdt_property_cell(fdt
, "qemu,graphic-depth", graphic_depth
)));
295 _FDT((fdt_end_node(fdt
)));
298 _FDT((fdt_begin_node(fdt
, "cpus")));
300 _FDT((fdt_property_cell(fdt
, "#address-cells", 0x1)));
301 _FDT((fdt_property_cell(fdt
, "#size-cells", 0x0)));
303 modelname
= g_strdup(cpu_model
);
305 for (i
= 0; i
< strlen(modelname
); i
++) {
306 modelname
[i
] = toupper(modelname
[i
]);
309 /* This is needed during FDT finalization */
310 spapr
->cpu_model
= g_strdup(modelname
);
312 for (env
= first_cpu
; env
!= NULL
; env
= env
->next_cpu
) {
313 int index
= env
->cpu_index
;
314 uint32_t servers_prop
[smp_threads
];
315 uint32_t gservers_prop
[smp_threads
* 2];
317 uint32_t segs
[] = {cpu_to_be32(28), cpu_to_be32(40),
318 0xffffffff, 0xffffffff};
319 uint32_t tbfreq
= kvm_enabled() ? kvmppc_get_tbfreq() : TIMEBASE_FREQ
;
320 uint32_t cpufreq
= kvm_enabled() ? kvmppc_get_clockfreq() : 1000000000;
321 uint32_t page_sizes_prop
[64];
322 size_t page_sizes_prop_size
;
324 if ((index
% smt
) != 0) {
328 if (asprintf(&nodename
, "%s@%x", modelname
, index
) < 0) {
329 fprintf(stderr
, "Allocation failure\n");
333 _FDT((fdt_begin_node(fdt
, nodename
)));
337 _FDT((fdt_property_cell(fdt
, "reg", index
)));
338 _FDT((fdt_property_string(fdt
, "device_type", "cpu")));
340 _FDT((fdt_property_cell(fdt
, "cpu-version", env
->spr
[SPR_PVR
])));
341 _FDT((fdt_property_cell(fdt
, "dcache-block-size",
342 env
->dcache_line_size
)));
343 _FDT((fdt_property_cell(fdt
, "icache-block-size",
344 env
->icache_line_size
)));
345 _FDT((fdt_property_cell(fdt
, "timebase-frequency", tbfreq
)));
346 _FDT((fdt_property_cell(fdt
, "clock-frequency", cpufreq
)));
347 _FDT((fdt_property_cell(fdt
, "ibm,slb-size", env
->slb_nr
)));
348 _FDT((fdt_property_string(fdt
, "status", "okay")));
349 _FDT((fdt_property(fdt
, "64-bit", NULL
, 0)));
351 /* Build interrupt servers and gservers properties */
352 for (i
= 0; i
< smp_threads
; i
++) {
353 servers_prop
[i
] = cpu_to_be32(index
+ i
);
354 /* Hack, direct the group queues back to cpu 0 */
355 gservers_prop
[i
*2] = cpu_to_be32(index
+ i
);
356 gservers_prop
[i
*2 + 1] = 0;
358 _FDT((fdt_property(fdt
, "ibm,ppc-interrupt-server#s",
359 servers_prop
, sizeof(servers_prop
))));
360 _FDT((fdt_property(fdt
, "ibm,ppc-interrupt-gserver#s",
361 gservers_prop
, sizeof(gservers_prop
))));
363 if (env
->mmu_model
& POWERPC_MMU_1TSEG
) {
364 _FDT((fdt_property(fdt
, "ibm,processor-segment-sizes",
365 segs
, sizeof(segs
))));
368 /* Advertise VMX/VSX (vector extensions) if available
369 * 0 / no property == no vector extensions
370 * 1 == VMX / Altivec available
371 * 2 == VSX available */
372 if (env
->insns_flags
& PPC_ALTIVEC
) {
373 uint32_t vmx
= (env
->insns_flags2
& PPC2_VSX
) ? 2 : 1;
375 _FDT((fdt_property_cell(fdt
, "ibm,vmx", vmx
)));
378 /* Advertise DFP (Decimal Floating Point) if available
379 * 0 / no property == no DFP
380 * 1 == DFP available */
381 if (env
->insns_flags2
& PPC2_DFP
) {
382 _FDT((fdt_property_cell(fdt
, "ibm,dfp", 1)));
385 page_sizes_prop_size
= create_page_sizes_prop(env
, page_sizes_prop
,
386 sizeof(page_sizes_prop
));
387 if (page_sizes_prop_size
) {
388 _FDT((fdt_property(fdt
, "ibm,segment-page-sizes",
389 page_sizes_prop
, page_sizes_prop_size
)));
392 _FDT((fdt_end_node(fdt
)));
397 _FDT((fdt_end_node(fdt
)));
400 _FDT((fdt_begin_node(fdt
, "rtas")));
402 _FDT((fdt_property(fdt
, "ibm,hypertas-functions", hypertas_prop
,
403 sizeof(hypertas_prop
))));
404 _FDT((fdt_property(fdt
, "qemu,hypertas-functions", qemu_hypertas_prop
,
405 sizeof(qemu_hypertas_prop
))));
407 _FDT((fdt_property(fdt
, "ibm,associativity-reference-points",
408 refpoints
, sizeof(refpoints
))));
410 _FDT((fdt_property_cell(fdt
, "rtas-error-log-max", RTAS_ERROR_LOG_MAX
)));
412 _FDT((fdt_end_node(fdt
)));
414 /* interrupt controller */
415 _FDT((fdt_begin_node(fdt
, "interrupt-controller")));
417 _FDT((fdt_property_string(fdt
, "device_type",
418 "PowerPC-External-Interrupt-Presentation")));
419 _FDT((fdt_property_string(fdt
, "compatible", "IBM,ppc-xicp")));
420 _FDT((fdt_property(fdt
, "interrupt-controller", NULL
, 0)));
421 _FDT((fdt_property(fdt
, "ibm,interrupt-server-ranges",
422 interrupt_server_ranges_prop
,
423 sizeof(interrupt_server_ranges_prop
))));
424 _FDT((fdt_property_cell(fdt
, "#interrupt-cells", 2)));
425 _FDT((fdt_property_cell(fdt
, "linux,phandle", PHANDLE_XICP
)));
426 _FDT((fdt_property_cell(fdt
, "phandle", PHANDLE_XICP
)));
428 _FDT((fdt_end_node(fdt
)));
431 _FDT((fdt_begin_node(fdt
, "vdevice")));
433 _FDT((fdt_property_string(fdt
, "device_type", "vdevice")));
434 _FDT((fdt_property_string(fdt
, "compatible", "IBM,vdevice")));
435 _FDT((fdt_property_cell(fdt
, "#address-cells", 0x1)));
436 _FDT((fdt_property_cell(fdt
, "#size-cells", 0x0)));
437 _FDT((fdt_property_cell(fdt
, "#interrupt-cells", 0x2)));
438 _FDT((fdt_property(fdt
, "interrupt-controller", NULL
, 0)));
440 _FDT((fdt_end_node(fdt
)));
443 spapr_events_fdt_skel(fdt
, epow_irq
);
445 _FDT((fdt_end_node(fdt
))); /* close root node */
446 _FDT((fdt_finish(fdt
)));
451 static int spapr_populate_memory(sPAPREnvironment
*spapr
, void *fdt
)
453 uint32_t associativity
[] = {cpu_to_be32(0x4), cpu_to_be32(0x0),
454 cpu_to_be32(0x0), cpu_to_be32(0x0),
457 hwaddr node0_size
, mem_start
;
458 uint64_t mem_reg_property
[2];
462 node0_size
= (nb_numa_nodes
> 1) ? node_mem
[0] : ram_size
;
463 if (spapr
->rma_size
> node0_size
) {
464 spapr
->rma_size
= node0_size
;
468 mem_reg_property
[0] = 0;
469 mem_reg_property
[1] = cpu_to_be64(spapr
->rma_size
);
470 off
= fdt_add_subnode(fdt
, 0, "memory@0");
472 _FDT((fdt_setprop_string(fdt
, off
, "device_type", "memory")));
473 _FDT((fdt_setprop(fdt
, off
, "reg", mem_reg_property
,
474 sizeof(mem_reg_property
))));
475 _FDT((fdt_setprop(fdt
, off
, "ibm,associativity", associativity
,
476 sizeof(associativity
))));
479 if (node0_size
> spapr
->rma_size
) {
480 mem_reg_property
[0] = cpu_to_be64(spapr
->rma_size
);
481 mem_reg_property
[1] = cpu_to_be64(node0_size
- spapr
->rma_size
);
483 sprintf(mem_name
, "memory@" TARGET_FMT_lx
, spapr
->rma_size
);
484 off
= fdt_add_subnode(fdt
, 0, mem_name
);
486 _FDT((fdt_setprop_string(fdt
, off
, "device_type", "memory")));
487 _FDT((fdt_setprop(fdt
, off
, "reg", mem_reg_property
,
488 sizeof(mem_reg_property
))));
489 _FDT((fdt_setprop(fdt
, off
, "ibm,associativity", associativity
,
490 sizeof(associativity
))));
493 /* RAM: Node 1 and beyond */
494 mem_start
= node0_size
;
495 for (i
= 1; i
< nb_numa_nodes
; i
++) {
496 mem_reg_property
[0] = cpu_to_be64(mem_start
);
497 mem_reg_property
[1] = cpu_to_be64(node_mem
[i
]);
498 associativity
[3] = associativity
[4] = cpu_to_be32(i
);
499 sprintf(mem_name
, "memory@" TARGET_FMT_lx
, mem_start
);
500 off
= fdt_add_subnode(fdt
, 0, mem_name
);
502 _FDT((fdt_setprop_string(fdt
, off
, "device_type", "memory")));
503 _FDT((fdt_setprop(fdt
, off
, "reg", mem_reg_property
,
504 sizeof(mem_reg_property
))));
505 _FDT((fdt_setprop(fdt
, off
, "ibm,associativity", associativity
,
506 sizeof(associativity
))));
507 mem_start
+= node_mem
[i
];
513 static void spapr_finalize_fdt(sPAPREnvironment
*spapr
,
522 fdt
= g_malloc(FDT_MAX_SIZE
);
524 /* open out the base tree into a temp buffer for the final tweaks */
525 _FDT((fdt_open_into(spapr
->fdt_skel
, fdt
, FDT_MAX_SIZE
)));
527 ret
= spapr_populate_memory(spapr
, fdt
);
529 fprintf(stderr
, "couldn't setup memory nodes in fdt\n");
533 ret
= spapr_populate_vdevice(spapr
->vio_bus
, fdt
);
535 fprintf(stderr
, "couldn't setup vio devices in fdt\n");
539 QLIST_FOREACH(phb
, &spapr
->phbs
, list
) {
540 ret
= spapr_populate_pci_dt(phb
, PHANDLE_XICP
, fdt
);
544 fprintf(stderr
, "couldn't setup PCI devices in fdt\n");
549 ret
= spapr_rtas_device_tree_setup(fdt
, rtas_addr
, rtas_size
);
551 fprintf(stderr
, "Couldn't set up RTAS device tree properties\n");
554 /* Advertise NUMA via ibm,associativity */
555 ret
= spapr_fixup_cpu_dt(fdt
, spapr
);
557 fprintf(stderr
, "Couldn't finalize CPU device tree properties\n");
560 if (!spapr
->has_graphics
) {
561 spapr_populate_chosen_stdout(fdt
, spapr
->vio_bus
);
564 _FDT((fdt_pack(fdt
)));
566 if (fdt_totalsize(fdt
) > FDT_MAX_SIZE
) {
567 hw_error("FDT too big ! 0x%x bytes (max is 0x%x)\n",
568 fdt_totalsize(fdt
), FDT_MAX_SIZE
);
572 cpu_physical_memory_write(fdt_addr
, fdt
, fdt_totalsize(fdt
));
577 static uint64_t translate_kernel_address(void *opaque
, uint64_t addr
)
579 return (addr
& 0x0fffffff) + KERNEL_LOAD_ADDR
;
582 static void emulate_spapr_hypercall(PowerPCCPU
*cpu
)
584 CPUPPCState
*env
= &cpu
->env
;
587 hcall_dprintf("Hypercall made with MSR[PR]=1\n");
588 env
->gpr
[3] = H_PRIVILEGE
;
590 env
->gpr
[3] = spapr_hypercall(cpu
, env
->gpr
[3], &env
->gpr
[4]);
594 static void spapr_reset_htab(sPAPREnvironment
*spapr
)
598 /* allocate hash page table. For now we always make this 16mb,
599 * later we should probably make it scale to the size of guest
602 shift
= kvmppc_reset_htab(spapr
->htab_shift
);
605 /* Kernel handles htab, we don't need to allocate one */
606 spapr
->htab_shift
= shift
;
609 /* Allocate an htab if we don't yet have one */
610 spapr
->htab
= qemu_memalign(HTAB_SIZE(spapr
), HTAB_SIZE(spapr
));
614 memset(spapr
->htab
, 0, HTAB_SIZE(spapr
));
617 /* Update the RMA size if necessary */
618 if (spapr
->vrma_adjust
) {
619 spapr
->rma_size
= kvmppc_rma_size(ram_size
, spapr
->htab_shift
);
623 static void ppc_spapr_reset(void)
625 /* Reset the hash table & recalc the RMA */
626 spapr_reset_htab(spapr
);
628 qemu_devices_reset();
631 spapr_finalize_fdt(spapr
, spapr
->fdt_addr
, spapr
->rtas_addr
,
634 /* Set up the entry state */
635 first_cpu
->gpr
[3] = spapr
->fdt_addr
;
636 first_cpu
->gpr
[5] = 0;
637 first_cpu
->halted
= 0;
638 first_cpu
->nip
= spapr
->entry_point
;
642 static void spapr_cpu_reset(void *opaque
)
644 PowerPCCPU
*cpu
= opaque
;
645 CPUPPCState
*env
= &cpu
->env
;
649 /* All CPUs start halted. CPU0 is unhalted from the machine level
650 * reset code and the rest are explicitly started up by the guest
651 * using an RTAS call */
654 env
->spr
[SPR_HIOR
] = 0;
656 env
->external_htab
= spapr
->htab
;
658 env
->htab_mask
= HTAB_SIZE(spapr
) - 1;
659 env
->spr
[SPR_SDR1
] = (unsigned long)spapr
->htab
|
660 (spapr
->htab_shift
- 18);
663 static void spapr_create_nvram(sPAPREnvironment
*spapr
)
665 QemuOpts
*machine_opts
;
668 dev
= qdev_create(&spapr
->vio_bus
->bus
, "spapr-nvram");
670 machine_opts
= qemu_opts_find(qemu_find_opts("machine"), 0);
672 const char *drivename
;
674 drivename
= qemu_opt_get(machine_opts
, "nvram");
676 BlockDriverState
*bs
;
678 bs
= bdrv_find(drivename
);
680 fprintf(stderr
, "No such block device \"%s\" for nvram\n",
684 qdev_prop_set_drive_nofail(dev
, "drive", bs
);
688 qdev_init_nofail(dev
);
690 spapr
->nvram
= (struct sPAPRNVRAM
*)dev
;
693 /* Returns whether we want to use VGA or not */
694 static int spapr_vga_init(PCIBus
*pci_bus
)
696 switch (vga_interface_type
) {
699 return pci_vga_init(pci_bus
) != NULL
;
701 fprintf(stderr
, "This vga model is not supported,"
702 "currently it only supports -vga std\n");
708 /* pSeries LPAR / sPAPR hardware init */
709 static void ppc_spapr_init(QEMUMachineInitArgs
*args
)
711 ram_addr_t ram_size
= args
->ram_size
;
712 const char *cpu_model
= args
->cpu_model
;
713 const char *kernel_filename
= args
->kernel_filename
;
714 const char *kernel_cmdline
= args
->kernel_cmdline
;
715 const char *initrd_filename
= args
->initrd_filename
;
716 const char *boot_device
= args
->boot_device
;
721 MemoryRegion
*sysmem
= get_system_memory();
722 MemoryRegion
*ram
= g_new(MemoryRegion
, 1);
723 hwaddr rma_alloc_size
;
724 uint32_t initrd_base
= 0;
725 long kernel_size
= 0, initrd_size
= 0;
726 long load_limit
, rtas_limit
, fw_size
;
729 msi_supported
= true;
731 spapr
= g_malloc0(sizeof(*spapr
));
732 QLIST_INIT(&spapr
->phbs
);
734 cpu_ppc_hypercall
= emulate_spapr_hypercall
;
736 /* Allocate RMA if necessary */
737 rma_alloc_size
= kvmppc_alloc_rma("ppc_spapr.rma", sysmem
);
739 if (rma_alloc_size
== -1) {
740 hw_error("qemu: Unable to create RMA\n");
744 if (rma_alloc_size
&& (rma_alloc_size
< ram_size
)) {
745 spapr
->rma_size
= rma_alloc_size
;
747 spapr
->rma_size
= ram_size
;
749 /* With KVM, we don't actually know whether KVM supports an
750 * unbounded RMA (PR KVM) or is limited by the hash table size
751 * (HV KVM using VRMA), so we always assume the latter
753 * In that case, we also limit the initial allocations for RTAS
754 * etc... to 256M since we have no way to know what the VRMA size
755 * is going to be as it depends on the size of the hash table
756 * isn't determined yet.
759 spapr
->vrma_adjust
= 1;
760 spapr
->rma_size
= MIN(spapr
->rma_size
, 0x10000000);
764 /* We place the device tree and RTAS just below either the top of the RMA,
765 * or just below 2GB, whichever is lowere, so that it can be
766 * processed with 32-bit real mode code if necessary */
767 rtas_limit
= MIN(spapr
->rma_size
, 0x80000000);
768 spapr
->rtas_addr
= rtas_limit
- RTAS_MAX_SIZE
;
769 spapr
->fdt_addr
= spapr
->rtas_addr
- FDT_MAX_SIZE
;
770 load_limit
= spapr
->fdt_addr
- FW_OVERHEAD
;
772 /* We aim for a hash table of size 1/128 the size of RAM. The
773 * normal rule of thumb is 1/64 the size of RAM, but that's much
774 * more than needed for the Linux guests we support. */
775 spapr
->htab_shift
= 18; /* Minimum architected size */
776 while (spapr
->htab_shift
<= 46) {
777 if ((1ULL << (spapr
->htab_shift
+ 7)) >= ram_size
) {
784 if (cpu_model
== NULL
) {
785 cpu_model
= kvm_enabled() ? "host" : "POWER7";
787 for (i
= 0; i
< smp_cpus
; i
++) {
788 cpu
= cpu_ppc_init(cpu_model
);
790 fprintf(stderr
, "Unable to find PowerPC CPU definition\n");
795 /* Set time-base frequency to 512 MHz */
796 cpu_ppc_tb_init(env
, TIMEBASE_FREQ
);
798 /* PAPR always has exception vectors in RAM not ROM */
799 env
->hreset_excp_prefix
= 0;
801 /* Tell KVM that we're in PAPR mode */
803 kvmppc_set_papr(cpu
);
806 qemu_register_reset(spapr_cpu_reset
, cpu
);
810 spapr
->ram_limit
= ram_size
;
811 if (spapr
->ram_limit
> rma_alloc_size
) {
812 ram_addr_t nonrma_base
= rma_alloc_size
;
813 ram_addr_t nonrma_size
= spapr
->ram_limit
- rma_alloc_size
;
815 memory_region_init_ram(ram
, "ppc_spapr.ram", nonrma_size
);
816 vmstate_register_ram_global(ram
);
817 memory_region_add_subregion(sysmem
, nonrma_base
, ram
);
820 filename
= qemu_find_file(QEMU_FILE_TYPE_BIOS
, "spapr-rtas.bin");
821 spapr
->rtas_size
= load_image_targphys(filename
, spapr
->rtas_addr
,
822 rtas_limit
- spapr
->rtas_addr
);
823 if (spapr
->rtas_size
< 0) {
824 hw_error("qemu: could not load LPAR rtas '%s'\n", filename
);
827 if (spapr
->rtas_size
> RTAS_MAX_SIZE
) {
828 hw_error("RTAS too big ! 0x%lx bytes (max is 0x%x)\n",
829 spapr
->rtas_size
, RTAS_MAX_SIZE
);
835 /* Set up Interrupt Controller */
836 spapr
->icp
= xics_system_init(XICS_IRQS
);
837 spapr
->next_irq
= XICS_IRQ_BASE
;
839 /* Set up EPOW events infrastructure */
840 spapr_events_init(spapr
);
846 spapr
->vio_bus
= spapr_vio_bus_init();
848 for (i
= 0; i
< MAX_SERIAL_PORTS
; i
++) {
850 spapr_vty_create(spapr
->vio_bus
, serial_hds
[i
]);
854 /* We always have at least the nvram device on VIO */
855 spapr_create_nvram(spapr
);
858 spapr_pci_rtas_init();
860 spapr_create_phb(spapr
, "pci", SPAPR_PCI_BUID
,
861 SPAPR_PCI_MEM_WIN_ADDR
,
862 SPAPR_PCI_MEM_WIN_SIZE
,
863 SPAPR_PCI_IO_WIN_ADDR
,
864 SPAPR_PCI_MSI_WIN_ADDR
);
865 phb
= PCI_HOST_BRIDGE(QLIST_FIRST(&spapr
->phbs
));
867 for (i
= 0; i
< nb_nics
; i
++) {
868 NICInfo
*nd
= &nd_table
[i
];
871 nd
->model
= g_strdup("ibmveth");
874 if (strcmp(nd
->model
, "ibmveth") == 0) {
875 spapr_vlan_create(spapr
->vio_bus
, nd
);
877 pci_nic_init_nofail(&nd_table
[i
], nd
->model
, NULL
);
881 for (i
= 0; i
<= drive_get_max_bus(IF_SCSI
); i
++) {
882 spapr_vscsi_create(spapr
->vio_bus
);
886 if (spapr_vga_init(phb
->bus
)) {
887 spapr
->has_graphics
= true;
890 if (usb_enabled(spapr
->has_graphics
)) {
891 pci_create_simple(phb
->bus
, -1, "pci-ohci");
892 if (spapr
->has_graphics
) {
893 usbdevice_create("keyboard");
894 usbdevice_create("mouse");
898 if (spapr
->rma_size
< (MIN_RMA_SLOF
<< 20)) {
899 fprintf(stderr
, "qemu: pSeries SLOF firmware requires >= "
900 "%ldM guest RMA (Real Mode Area memory)\n", MIN_RMA_SLOF
);
904 if (kernel_filename
) {
905 uint64_t lowaddr
= 0;
907 kernel_size
= load_elf(kernel_filename
, translate_kernel_address
, NULL
,
908 NULL
, &lowaddr
, NULL
, 1, ELF_MACHINE
, 0);
909 if (kernel_size
< 0) {
910 kernel_size
= load_image_targphys(kernel_filename
,
912 load_limit
- KERNEL_LOAD_ADDR
);
914 if (kernel_size
< 0) {
915 fprintf(stderr
, "qemu: could not load kernel '%s'\n",
921 if (initrd_filename
) {
922 /* Try to locate the initrd in the gap between the kernel
923 * and the firmware. Add a bit of space just in case
925 initrd_base
= (KERNEL_LOAD_ADDR
+ kernel_size
+ 0x1ffff) & ~0xffff;
926 initrd_size
= load_image_targphys(initrd_filename
, initrd_base
,
927 load_limit
- initrd_base
);
928 if (initrd_size
< 0) {
929 fprintf(stderr
, "qemu: could not load initial ram disk '%s'\n",
939 filename
= qemu_find_file(QEMU_FILE_TYPE_BIOS
, FW_FILE_NAME
);
940 fw_size
= load_image_targphys(filename
, 0, FW_MAX_SIZE
);
942 hw_error("qemu: could not load LPAR rtas '%s'\n", filename
);
947 spapr
->entry_point
= 0x100;
949 /* Prepare the device tree */
950 spapr
->fdt_skel
= spapr_create_fdt_skel(cpu_model
,
951 initrd_base
, initrd_size
,
953 boot_device
, kernel_cmdline
,
955 assert(spapr
->fdt_skel
!= NULL
);
958 static QEMUMachine spapr_machine
= {
960 .desc
= "pSeries Logical Partition (PAPR compliant)",
961 .init
= ppc_spapr_init
,
962 .reset
= ppc_spapr_reset
,
963 .block_default_type
= IF_SCSI
,
964 .max_cpus
= MAX_CPUS
,
968 static void spapr_machine_init(void)
970 qemu_register_machine(&spapr_machine
);
973 machine_init(spapr_machine_init
);