|
|
1.1 root 1: /*
2: * QEMU PowerPC pSeries Logical Partition (aka sPAPR) hardware System Emulator
3: *
4: * Copyright (c) 2004-2007 Fabrice Bellard
5: * Copyright (c) 2007 Jocelyn Mayer
6: * Copyright (c) 2010 David Gibson, IBM Corporation.
7: *
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:
14: *
15: * The above copyright notice and this permission notice shall be included in
16: * all copies or substantial portions of the Software.
17: *
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
24: * THE SOFTWARE.
25: *
26: */
27: #include "sysemu.h"
28: #include "hw.h"
29: #include "elf.h"
30: #include "net.h"
31: #include "blockdev.h"
32:
33: #include "hw/boards.h"
34: #include "hw/ppc.h"
35: #include "hw/loader.h"
36:
37: #include "hw/spapr.h"
38: #include "hw/spapr_vio.h"
39: #include "hw/xics.h"
40:
41: #include <libfdt.h>
42:
43: #define KERNEL_LOAD_ADDR 0x00000000
44: #define INITRD_LOAD_ADDR 0x02800000
45: #define FDT_MAX_SIZE 0x10000
46: #define RTAS_MAX_SIZE 0x10000
47: #define FW_MAX_SIZE 0x400000
48: #define FW_FILE_NAME "slof.bin"
49:
50: #define MIN_RAM_SLOF 512UL
51:
52: #define TIMEBASE_FREQ 512000000ULL
53:
54: #define MAX_CPUS 256
55: #define XICS_IRQS 1024
56:
57: sPAPREnvironment *spapr;
58:
59: static void *spapr_create_fdt_skel(const char *cpu_model,
60: target_phys_addr_t initrd_base,
61: target_phys_addr_t initrd_size,
62: const char *boot_device,
63: const char *kernel_cmdline,
64: long hash_shift)
65: {
66: void *fdt;
67: CPUState *env;
68: uint64_t mem_reg_property[] = { 0, cpu_to_be64(ram_size) };
69: uint32_t start_prop = cpu_to_be32(initrd_base);
70: uint32_t end_prop = cpu_to_be32(initrd_base + initrd_size);
71: uint32_t pft_size_prop[] = {0, cpu_to_be32(hash_shift)};
72: char hypertas_prop[] = "hcall-pft\0hcall-term\0hcall-dabr\0hcall-interrupt"
73: "\0hcall-tce\0hcall-vio\0hcall-splpar";
74: uint32_t interrupt_server_ranges_prop[] = {0, cpu_to_be32(smp_cpus)};
75: int i;
76: char *modelname;
77:
78: #define _FDT(exp) \
79: do { \
80: int ret = (exp); \
81: if (ret < 0) { \
82: fprintf(stderr, "qemu: error creating device tree: %s: %s\n", \
83: #exp, fdt_strerror(ret)); \
84: exit(1); \
85: } \
86: } while (0)
87:
88: fdt = qemu_mallocz(FDT_MAX_SIZE);
89: _FDT((fdt_create(fdt, FDT_MAX_SIZE)));
90:
91: _FDT((fdt_finish_reservemap(fdt)));
92:
93: /* Root node */
94: _FDT((fdt_begin_node(fdt, "")));
95: _FDT((fdt_property_string(fdt, "device_type", "chrp")));
96: _FDT((fdt_property_string(fdt, "model", "IBM pSeries (emulated by qemu)")));
97:
98: _FDT((fdt_property_cell(fdt, "#address-cells", 0x2)));
99: _FDT((fdt_property_cell(fdt, "#size-cells", 0x2)));
100:
101: /* /chosen */
102: _FDT((fdt_begin_node(fdt, "chosen")));
103:
104: _FDT((fdt_property_string(fdt, "bootargs", kernel_cmdline)));
105: _FDT((fdt_property(fdt, "linux,initrd-start",
106: &start_prop, sizeof(start_prop))));
107: _FDT((fdt_property(fdt, "linux,initrd-end",
108: &end_prop, sizeof(end_prop))));
109: _FDT((fdt_property_string(fdt, "qemu,boot-device", boot_device)));
110:
111: _FDT((fdt_end_node(fdt)));
112:
113: /* memory node */
114: _FDT((fdt_begin_node(fdt, "memory@0")));
115:
116: _FDT((fdt_property_string(fdt, "device_type", "memory")));
117: _FDT((fdt_property(fdt, "reg",
118: mem_reg_property, sizeof(mem_reg_property))));
119:
120: _FDT((fdt_end_node(fdt)));
121:
122: /* cpus */
123: _FDT((fdt_begin_node(fdt, "cpus")));
124:
125: _FDT((fdt_property_cell(fdt, "#address-cells", 0x1)));
126: _FDT((fdt_property_cell(fdt, "#size-cells", 0x0)));
127:
128: modelname = qemu_strdup(cpu_model);
129:
130: for (i = 0; i < strlen(modelname); i++) {
131: modelname[i] = toupper(modelname[i]);
132: }
133:
134: for (env = first_cpu; env != NULL; env = env->next_cpu) {
135: int index = env->cpu_index;
136: uint32_t gserver_prop[] = {cpu_to_be32(index), 0}; /* HACK! */
137: char *nodename;
138: uint32_t segs[] = {cpu_to_be32(28), cpu_to_be32(40),
139: 0xffffffff, 0xffffffff};
140:
141: if (asprintf(&nodename, "%s@%x", modelname, index) < 0) {
142: fprintf(stderr, "Allocation failure\n");
143: exit(1);
144: }
145:
146: _FDT((fdt_begin_node(fdt, nodename)));
147:
148: free(nodename);
149:
150: _FDT((fdt_property_cell(fdt, "reg", index)));
151: _FDT((fdt_property_string(fdt, "device_type", "cpu")));
152:
153: _FDT((fdt_property_cell(fdt, "cpu-version", env->spr[SPR_PVR])));
154: _FDT((fdt_property_cell(fdt, "dcache-block-size",
155: env->dcache_line_size)));
156: _FDT((fdt_property_cell(fdt, "icache-block-size",
157: env->icache_line_size)));
158: _FDT((fdt_property_cell(fdt, "timebase-frequency", TIMEBASE_FREQ)));
159: /* Hardcode CPU frequency for now. It's kind of arbitrary on
160: * full emu, for kvm we should copy it from the host */
161: _FDT((fdt_property_cell(fdt, "clock-frequency", 1000000000)));
162: _FDT((fdt_property_cell(fdt, "ibm,slb-size", env->slb_nr)));
163: _FDT((fdt_property(fdt, "ibm,pft-size",
164: pft_size_prop, sizeof(pft_size_prop))));
165: _FDT((fdt_property_string(fdt, "status", "okay")));
166: _FDT((fdt_property(fdt, "64-bit", NULL, 0)));
167: _FDT((fdt_property_cell(fdt, "ibm,ppc-interrupt-server#s", index)));
168: _FDT((fdt_property(fdt, "ibm,ppc-interrupt-gserver#s",
169: gserver_prop, sizeof(gserver_prop))));
170:
171: if (env->mmu_model & POWERPC_MMU_1TSEG) {
172: _FDT((fdt_property(fdt, "ibm,processor-segment-sizes",
173: segs, sizeof(segs))));
174: }
175:
176: _FDT((fdt_end_node(fdt)));
177: }
178:
179: qemu_free(modelname);
180:
181: _FDT((fdt_end_node(fdt)));
182:
183: /* RTAS */
184: _FDT((fdt_begin_node(fdt, "rtas")));
185:
186: _FDT((fdt_property(fdt, "ibm,hypertas-functions", hypertas_prop,
187: sizeof(hypertas_prop))));
188:
189: _FDT((fdt_end_node(fdt)));
190:
191: /* interrupt controller */
192: _FDT((fdt_begin_node(fdt, "interrupt-controller@0")));
193:
194: _FDT((fdt_property_string(fdt, "device_type",
195: "PowerPC-External-Interrupt-Presentation")));
196: _FDT((fdt_property_string(fdt, "compatible", "IBM,ppc-xicp")));
197: _FDT((fdt_property_cell(fdt, "reg", 0)));
198: _FDT((fdt_property(fdt, "interrupt-controller", NULL, 0)));
199: _FDT((fdt_property(fdt, "ibm,interrupt-server-ranges",
200: interrupt_server_ranges_prop,
201: sizeof(interrupt_server_ranges_prop))));
202:
203: _FDT((fdt_end_node(fdt)));
204:
205: /* vdevice */
206: _FDT((fdt_begin_node(fdt, "vdevice")));
207:
208: _FDT((fdt_property_string(fdt, "device_type", "vdevice")));
209: _FDT((fdt_property_string(fdt, "compatible", "IBM,vdevice")));
210: _FDT((fdt_property_cell(fdt, "#address-cells", 0x1)));
211: _FDT((fdt_property_cell(fdt, "#size-cells", 0x0)));
212: _FDT((fdt_property_cell(fdt, "#interrupt-cells", 0x2)));
213: _FDT((fdt_property(fdt, "interrupt-controller", NULL, 0)));
214:
215: _FDT((fdt_end_node(fdt)));
216:
217: _FDT((fdt_end_node(fdt))); /* close root node */
218: _FDT((fdt_finish(fdt)));
219:
220: return fdt;
221: }
222:
223: static void spapr_finalize_fdt(sPAPREnvironment *spapr,
224: target_phys_addr_t fdt_addr,
225: target_phys_addr_t rtas_addr,
226: target_phys_addr_t rtas_size)
227: {
228: int ret;
229: void *fdt;
230:
231: fdt = qemu_malloc(FDT_MAX_SIZE);
232:
233: /* open out the base tree into a temp buffer for the final tweaks */
234: _FDT((fdt_open_into(spapr->fdt_skel, fdt, FDT_MAX_SIZE)));
235:
236: ret = spapr_populate_vdevice(spapr->vio_bus, fdt);
237: if (ret < 0) {
238: fprintf(stderr, "couldn't setup vio devices in fdt\n");
239: exit(1);
240: }
241:
242: /* RTAS */
243: ret = spapr_rtas_device_tree_setup(fdt, rtas_addr, rtas_size);
244: if (ret < 0) {
245: fprintf(stderr, "Couldn't set up RTAS device tree properties\n");
246: }
247:
248: _FDT((fdt_pack(fdt)));
249:
250: cpu_physical_memory_write(fdt_addr, fdt, fdt_totalsize(fdt));
251:
252: qemu_free(fdt);
253: }
254:
255: static uint64_t translate_kernel_address(void *opaque, uint64_t addr)
256: {
257: return (addr & 0x0fffffff) + KERNEL_LOAD_ADDR;
258: }
259:
260: static void emulate_spapr_hypercall(CPUState *env)
261: {
262: env->gpr[3] = spapr_hypercall(env, env->gpr[3], &env->gpr[4]);
263: }
264:
265: static void spapr_reset(void *opaque)
266: {
267: sPAPREnvironment *spapr = (sPAPREnvironment *)opaque;
268:
269: fprintf(stderr, "sPAPR reset\n");
270:
271: /* flush out the hash table */
272: memset(spapr->htab, 0, spapr->htab_size);
273:
274: /* Load the fdt */
275: spapr_finalize_fdt(spapr, spapr->fdt_addr, spapr->rtas_addr,
276: spapr->rtas_size);
277:
278: /* Set up the entry state */
279: first_cpu->gpr[3] = spapr->fdt_addr;
280: first_cpu->gpr[5] = 0;
281: first_cpu->halted = 0;
282: first_cpu->nip = spapr->entry_point;
283:
284: }
285:
286: /* pSeries LPAR / sPAPR hardware init */
287: static void ppc_spapr_init(ram_addr_t ram_size,
288: const char *boot_device,
289: const char *kernel_filename,
290: const char *kernel_cmdline,
291: const char *initrd_filename,
292: const char *cpu_model)
293: {
294: CPUState *env;
295: int i;
296: ram_addr_t ram_offset;
297: uint32_t initrd_base;
298: long kernel_size, initrd_size, fw_size;
299: long pteg_shift = 17;
300: char *filename;
301: int irq = 16;
302:
303: spapr = qemu_malloc(sizeof(*spapr));
304: cpu_ppc_hypercall = emulate_spapr_hypercall;
305:
306: /* We place the device tree just below either the top of RAM, or
307: * 2GB, so that it can be processed with 32-bit code if
308: * necessary */
309: spapr->fdt_addr = MIN(ram_size, 0x80000000) - FDT_MAX_SIZE;
310: spapr->rtas_addr = spapr->fdt_addr - RTAS_MAX_SIZE;
311:
312: /* init CPUs */
313: if (cpu_model == NULL) {
314: cpu_model = "POWER7";
315: }
316: for (i = 0; i < smp_cpus; i++) {
317: env = cpu_init(cpu_model);
318:
319: if (!env) {
320: fprintf(stderr, "Unable to find PowerPC CPU definition\n");
321: exit(1);
322: }
323: /* Set time-base frequency to 512 MHz */
324: cpu_ppc_tb_init(env, TIMEBASE_FREQ);
325: qemu_register_reset((QEMUResetHandler *)&cpu_reset, env);
326:
327: env->hreset_vector = 0x60;
328: env->hreset_excp_prefix = 0;
329: env->gpr[3] = env->cpu_index;
330: }
331:
332: /* allocate RAM */
333: ram_offset = qemu_ram_alloc(NULL, "ppc_spapr.ram", ram_size);
334: cpu_register_physical_memory(0, ram_size, ram_offset);
335:
336: /* allocate hash page table. For now we always make this 16mb,
337: * later we should probably make it scale to the size of guest
338: * RAM */
339: spapr->htab_size = 1ULL << (pteg_shift + 7);
340: spapr->htab = qemu_malloc(spapr->htab_size);
341:
342: for (env = first_cpu; env != NULL; env = env->next_cpu) {
343: env->external_htab = spapr->htab;
344: env->htab_base = -1;
345: env->htab_mask = spapr->htab_size - 1;
346: }
347:
348: filename = qemu_find_file(QEMU_FILE_TYPE_BIOS, "spapr-rtas.bin");
349: spapr->rtas_size = load_image_targphys(filename, spapr->rtas_addr,
350: ram_size - spapr->rtas_addr);
351: if (spapr->rtas_size < 0) {
352: hw_error("qemu: could not load LPAR rtas '%s'\n", filename);
353: exit(1);
354: }
355: qemu_free(filename);
356:
357: /* Set up Interrupt Controller */
358: spapr->icp = xics_system_init(XICS_IRQS);
359:
360: /* Set up VIO bus */
361: spapr->vio_bus = spapr_vio_bus_init();
362:
363: for (i = 0; i < MAX_SERIAL_PORTS; i++, irq++) {
364: if (serial_hds[i]) {
365: spapr_vty_create(spapr->vio_bus, SPAPR_VTY_BASE_ADDRESS + i,
366: serial_hds[i], xics_find_qirq(spapr->icp, irq),
367: irq);
368: }
369: }
370:
371: for (i = 0; i < nb_nics; i++, irq++) {
372: NICInfo *nd = &nd_table[i];
373:
374: if (!nd->model) {
375: nd->model = qemu_strdup("ibmveth");
376: }
377:
378: if (strcmp(nd->model, "ibmveth") == 0) {
379: spapr_vlan_create(spapr->vio_bus, 0x1000 + i, nd,
380: xics_find_qirq(spapr->icp, irq), irq);
381: } else {
382: fprintf(stderr, "pSeries (sPAPR) platform does not support "
383: "NIC model '%s' (only ibmveth is supported)\n",
384: nd->model);
385: exit(1);
386: }
387: }
388:
389: for (i = 0; i <= drive_get_max_bus(IF_SCSI); i++) {
390: spapr_vscsi_create(spapr->vio_bus, 0x2000 + i,
391: xics_find_qirq(spapr->icp, irq), irq);
392: irq++;
393: }
394:
395: if (kernel_filename) {
396: uint64_t lowaddr = 0;
397:
398: kernel_size = load_elf(kernel_filename, translate_kernel_address, NULL,
399: NULL, &lowaddr, NULL, 1, ELF_MACHINE, 0);
400: if (kernel_size < 0) {
401: kernel_size = load_image_targphys(kernel_filename,
402: KERNEL_LOAD_ADDR,
403: ram_size - KERNEL_LOAD_ADDR);
404: }
405: if (kernel_size < 0) {
406: fprintf(stderr, "qemu: could not load kernel '%s'\n",
407: kernel_filename);
408: exit(1);
409: }
410:
411: /* load initrd */
412: if (initrd_filename) {
413: initrd_base = INITRD_LOAD_ADDR;
414: initrd_size = load_image_targphys(initrd_filename, initrd_base,
415: ram_size - initrd_base);
416: if (initrd_size < 0) {
417: fprintf(stderr, "qemu: could not load initial ram disk '%s'\n",
418: initrd_filename);
419: exit(1);
420: }
421: } else {
422: initrd_base = 0;
423: initrd_size = 0;
424: }
425:
426: spapr->entry_point = KERNEL_LOAD_ADDR;
427: } else {
428: if (ram_size < (MIN_RAM_SLOF << 20)) {
429: fprintf(stderr, "qemu: pSeries SLOF firmware requires >= "
430: "%ldM guest RAM\n", MIN_RAM_SLOF);
431: exit(1);
432: }
433: filename = qemu_find_file(QEMU_FILE_TYPE_BIOS, "slof.bin");
434: fw_size = load_image_targphys(filename, 0, FW_MAX_SIZE);
435: if (fw_size < 0) {
436: hw_error("qemu: could not load LPAR rtas '%s'\n", filename);
437: exit(1);
438: }
439: qemu_free(filename);
440: spapr->entry_point = 0x100;
441: initrd_base = 0;
442: initrd_size = 0;
443:
444: /* SLOF will startup the secondary CPUs using RTAS,
445: rather than expecting a kexec() style entry */
446: for (env = first_cpu; env != NULL; env = env->next_cpu) {
447: env->halted = 1;
448: }
449: }
450:
451: /* Prepare the device tree */
452: spapr->fdt_skel = spapr_create_fdt_skel(cpu_model,
453: initrd_base, initrd_size,
454: boot_device, kernel_cmdline,
455: pteg_shift + 7);
456: assert(spapr->fdt_skel != NULL);
457:
458: qemu_register_reset(spapr_reset, spapr);
459: }
460:
461: static QEMUMachine spapr_machine = {
462: .name = "pseries",
463: .desc = "pSeries Logical Partition (PAPR compliant)",
464: .init = ppc_spapr_init,
465: .max_cpus = MAX_CPUS,
466: .no_vga = 1,
467: .no_parallel = 1,
468: .use_scsi = 1,
469: };
470:
471: static void spapr_machine_init(void)
472: {
473: qemu_register_machine(&spapr_machine);
474: }
475:
476: machine_init(spapr_machine_init);
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.