setup.c 25.1 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 *  S390 version
3
 *    Copyright IBM Corp. 1999, 2012
L
Linus Torvalds 已提交
4 5 6 7 8 9 10 11 12 13 14
 *    Author(s): Hartmut Penner (hp@de.ibm.com),
 *               Martin Schwidefsky (schwidefsky@de.ibm.com)
 *
 *  Derived from "arch/i386/kernel/setup.c"
 *    Copyright (C) 1995, Linus Torvalds
 */

/*
 * This file handles the architecture-dependent parts of initialization
 */

15 16 17
#define KMSG_COMPONENT "setup"
#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt

L
Linus Torvalds 已提交
18
#include <linux/errno.h>
H
Heiko Carstens 已提交
19
#include <linux/export.h>
L
Linus Torvalds 已提交
20 21
#include <linux/sched.h>
#include <linux/kernel.h>
T
Tejun Heo 已提交
22
#include <linux/memblock.h>
L
Linus Torvalds 已提交
23 24 25 26 27 28 29 30 31 32 33 34 35 36
#include <linux/mm.h>
#include <linux/stddef.h>
#include <linux/unistd.h>
#include <linux/ptrace.h>
#include <linux/user.h>
#include <linux/tty.h>
#include <linux/ioport.h>
#include <linux/delay.h>
#include <linux/init.h>
#include <linux/initrd.h>
#include <linux/bootmem.h>
#include <linux/root_dev.h>
#include <linux/console.h>
#include <linux/kernel_stat.h>
37
#include <linux/device.h>
38
#include <linux/notifier.h>
39
#include <linux/pfn.h>
H
Hongjie Yang 已提交
40
#include <linux/ctype.h>
41
#include <linux/reboot.h>
42
#include <linux/topology.h>
43
#include <linux/ftrace.h>
M
Michael Holzheu 已提交
44 45 46
#include <linux/kexec.h>
#include <linux/crash_dump.h>
#include <linux/memory.h>
47
#include <linux/compat.h>
L
Linus Torvalds 已提交
48

M
Michael Holzheu 已提交
49
#include <asm/ipl.h>
50
#include <asm/facility.h>
L
Linus Torvalds 已提交
51 52 53 54 55
#include <asm/smp.h>
#include <asm/mmu_context.h>
#include <asm/cpcmd.h>
#include <asm/lowcore.h>
#include <asm/irq.h>
56 57
#include <asm/page.h>
#include <asm/ptrace.h>
58
#include <asm/sections.h>
H
Hongjie Yang 已提交
59
#include <asm/ebcdic.h>
60
#include <asm/kvm_virtio.h>
M
Michael Holzheu 已提交
61
#include <asm/diag.h>
62
#include <asm/os_info.h>
63
#include <asm/sclp.h>
M
Martin Schwidefsky 已提交
64
#include "entry.h"
G
Gerald Schaefer 已提交
65

L
Linus Torvalds 已提交
66 67 68 69
/*
 * Machine setup..
 */
unsigned int console_mode = 0;
70 71
EXPORT_SYMBOL(console_mode);

L
Linus Torvalds 已提交
72
unsigned int console_devno = -1;
73 74
EXPORT_SYMBOL(console_devno);

L
Linus Torvalds 已提交
75
unsigned int console_irq = -1;
76 77
EXPORT_SYMBOL(console_irq);

78 79
unsigned long elf_hwcap = 0;
char elf_platform[ELF_PLATFORM_SIZE];
80

81 82
int __initdata memory_end_set;
unsigned long __initdata memory_end;
83
unsigned long __initdata max_physmem_end;
L
Linus Torvalds 已提交
84

85 86 87 88 89 90 91 92 93
unsigned long VMALLOC_START;
EXPORT_SYMBOL(VMALLOC_START);

unsigned long VMALLOC_END;
EXPORT_SYMBOL(VMALLOC_END);

struct page *vmemmap;
EXPORT_SYMBOL(vmemmap);

94 95 96 97 98
#ifdef CONFIG_64BIT
unsigned long MODULES_VADDR;
unsigned long MODULES_END;
#endif

F
Frank Munzert 已提交
99 100 101 102
/* An array with a pointer to the lowcore of every CPU. */
struct _lowcore *lowcore_ptr[NR_CPUS];
EXPORT_SYMBOL(lowcore_ptr);

L
Linus Torvalds 已提交
103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128
/*
 * This is set up by the setup-routine at boot-time
 * for S390 need to find out, what we have to setup
 * using address 0x10400 ...
 */

#include <asm/setup.h>

/*
 * condev= and conmode= setup parameter.
 */

static int __init condev_setup(char *str)
{
	int vdev;

	vdev = simple_strtoul(str, &str, 0);
	if (vdev >= 0 && vdev < 65536) {
		console_devno = vdev;
		console_irq = -1;
	}
	return 1;
}

__setup("condev=", condev_setup);

129 130
static void __init set_preferred_console(void)
{
131 132 133 134 135 136 137 138
	if (MACHINE_IS_KVM) {
		if (sclp_has_vt220())
			add_preferred_console("ttyS", 1, NULL);
		else if (sclp_has_linemode())
			add_preferred_console("ttyS", 0, NULL);
		else
			add_preferred_console("hvc", 0, NULL);
	} else if (CONSOLE_IS_3215 || CONSOLE_IS_SCLP)
139
		add_preferred_console("ttyS", 0, NULL);
140
	else if (CONSOLE_IS_3270)
141 142 143
		add_preferred_console("tty3270", 0, NULL);
}

L
Linus Torvalds 已提交
144 145
static int __init conmode_setup(char *str)
{
146
#if defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
L
Linus Torvalds 已提交
147 148 149 150 151 152 153 154 155 156 157
	if (strncmp(str, "hwc", 4) == 0 || strncmp(str, "sclp", 5) == 0)
                SET_CONSOLE_SCLP;
#endif
#if defined(CONFIG_TN3215_CONSOLE)
	if (strncmp(str, "3215", 5) == 0)
		SET_CONSOLE_3215;
#endif
#if defined(CONFIG_TN3270_CONSOLE)
	if (strncmp(str, "3270", 5) == 0)
		SET_CONSOLE_3270;
#endif
158
	set_preferred_console();
L
Linus Torvalds 已提交
159 160 161 162 163 164 165 166 167 168 169
        return 1;
}

__setup("conmode=", conmode_setup);

static void __init conmode_default(void)
{
	char query_buffer[1024];
	char *ptr;

        if (MACHINE_IS_VM) {
170
		cpcmd("QUERY CONSOLE", query_buffer, 1024, NULL);
L
Linus Torvalds 已提交
171 172 173
		console_devno = simple_strtoul(query_buffer + 5, NULL, 16);
		ptr = strstr(query_buffer, "SUBCHANNEL =");
		console_irq = simple_strtoul(ptr + 13, NULL, 16);
174
		cpcmd("QUERY TERM", query_buffer, 1024, NULL);
L
Linus Torvalds 已提交
175 176 177 178 179 180 181 182
		ptr = strstr(query_buffer, "CONMODE");
		/*
		 * Set the conmode to 3215 so that the device recognition 
		 * will set the cu_type of the console to 3215. If the
		 * conmode is 3270 and we don't set it back then both
		 * 3215 and the 3270 driver will try to access the console
		 * device (3215 as console and 3270 as normal tty).
		 */
183
		cpcmd("TERM CONMODE 3215", NULL, 0, NULL);
L
Linus Torvalds 已提交
184
		if (ptr == NULL) {
185
#if defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
L
Linus Torvalds 已提交
186 187 188 189 190 191 192 193 194
			SET_CONSOLE_SCLP;
#endif
			return;
		}
		if (strncmp(ptr + 8, "3270", 4) == 0) {
#if defined(CONFIG_TN3270_CONSOLE)
			SET_CONSOLE_3270;
#elif defined(CONFIG_TN3215_CONSOLE)
			SET_CONSOLE_3215;
195
#elif defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
L
Linus Torvalds 已提交
196 197 198 199 200 201 202
			SET_CONSOLE_SCLP;
#endif
		} else if (strncmp(ptr + 8, "3215", 4) == 0) {
#if defined(CONFIG_TN3215_CONSOLE)
			SET_CONSOLE_3215;
#elif defined(CONFIG_TN3270_CONSOLE)
			SET_CONSOLE_3270;
203
#elif defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
L
Linus Torvalds 已提交
204 205 206 207
			SET_CONSOLE_SCLP;
#endif
		}
	} else {
208
#if defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
L
Linus Torvalds 已提交
209 210 211 212 213
		SET_CONSOLE_SCLP;
#endif
	}
}

214
#ifdef CONFIG_CRASH_DUMP
215
static void __init setup_zfcpdump(void)
M
Michael Holzheu 已提交
216 217 218
{
	if (ipl_info.type != IPL_TYPE_FCP_DUMP)
		return;
219 220
	if (OLDMEM_BASE)
		return;
221
	strcat(boot_command_line, " cio_ignore=all,!ipldev,!condev");
M
Michael Holzheu 已提交
222 223 224
	console_loglevel = 2;
}
#else
225
static inline void setup_zfcpdump(void) {}
226
#endif /* CONFIG_CRASH_DUMP */
M
Michael Holzheu 已提交
227

L
Linus Torvalds 已提交
228 229 230 231 232 233 234
 /*
 * Reboot, halt and power_off stubs. They just call _machine_restart,
 * _machine_halt or _machine_power_off. 
 */

void machine_restart(char *command)
{
235
	if ((!in_interrupt() && !in_atomic()) || oops_in_progress)
236 237 238 239 240
		/*
		 * Only unblank the console if we are called in enabled
		 * context or a bust_spinlocks cleared the way for us.
		 */
		console_unblank();
L
Linus Torvalds 已提交
241 242 243 244 245
	_machine_restart(command);
}

void machine_halt(void)
{
246 247 248 249 250 251
	if (!in_interrupt() || oops_in_progress)
		/*
		 * Only unblank the console if we are called in enabled
		 * context or a bust_spinlocks cleared the way for us.
		 */
		console_unblank();
L
Linus Torvalds 已提交
252 253 254 255 256
	_machine_halt();
}

void machine_power_off(void)
{
257 258 259 260 261 262
	if (!in_interrupt() || oops_in_progress)
		/*
		 * Only unblank the console if we are called in enabled
		 * context or a bust_spinlocks cleared the way for us.
		 */
		console_unblank();
L
Linus Torvalds 已提交
263 264 265
	_machine_power_off();
}

266 267 268 269
/*
 * Dummy power off function.
 */
void (*pm_power_off)(void) = machine_power_off;
H
Heiko Carstens 已提交
270
EXPORT_SYMBOL_GPL(pm_power_off);
271

272 273 274
static int __init early_parse_mem(char *p)
{
	memory_end = memparse(p, &p);
275
	memory_end &= PAGE_MASK;
276
	memory_end_set = 1;
277 278 279 280
	return 0;
}
early_param("mem", early_parse_mem);

281 282 283 284 285 286 287 288 289
static int __init parse_vmalloc(char *arg)
{
	if (!arg)
		return -EINVAL;
	VMALLOC_END = (memparse(arg, &arg) + PAGE_SIZE - 1) & PAGE_MASK;
	return 0;
}
early_param("vmalloc", parse_vmalloc);

M
Martin Schwidefsky 已提交
290 291 292
void *restart_stack __attribute__((__section__(".data")));

static void __init setup_lowcore(void)
293 294 295 296 297 298
{
	struct _lowcore *lc;

	/*
	 * Setup lowcore for boot cpu
	 */
299
	BUILD_BUG_ON(sizeof(struct _lowcore) != LC_PAGES * 4096);
300
	lc = __alloc_bootmem_low(LC_PAGES * PAGE_SIZE, LC_PAGES * PAGE_SIZE, 0);
301
	lc->restart_psw.mask = PSW_KERNEL_BITS;
302
	lc->restart_psw.addr =
M
Martin Schwidefsky 已提交
303
		PSW_ADDR_AMODE | (unsigned long) restart_int_handler;
304
	lc->external_new_psw.mask = PSW_KERNEL_BITS |
305
		PSW_MASK_DAT | PSW_MASK_MCHECK;
306 307
	lc->external_new_psw.addr =
		PSW_ADDR_AMODE | (unsigned long) ext_int_handler;
308
	lc->svc_new_psw.mask = PSW_KERNEL_BITS |
309
		PSW_MASK_DAT | PSW_MASK_IO | PSW_MASK_EXT | PSW_MASK_MCHECK;
310
	lc->svc_new_psw.addr = PSW_ADDR_AMODE | (unsigned long) system_call;
311
	lc->program_new_psw.mask = PSW_KERNEL_BITS |
312
		PSW_MASK_DAT | PSW_MASK_MCHECK;
313
	lc->program_new_psw.addr =
314
		PSW_ADDR_AMODE | (unsigned long) pgm_check_handler;
315
	lc->mcck_new_psw.mask = PSW_KERNEL_BITS;
316 317
	lc->mcck_new_psw.addr =
		PSW_ADDR_AMODE | (unsigned long) mcck_int_handler;
318
	lc->io_new_psw.mask = PSW_KERNEL_BITS |
319
		PSW_MASK_DAT | PSW_MASK_MCHECK;
320
	lc->io_new_psw.addr = PSW_ADDR_AMODE | (unsigned long) io_int_handler;
321
	lc->clock_comparator = -1ULL;
322 323
	lc->kernel_stack = ((unsigned long) &init_thread_union)
		+ THREAD_SIZE - STACK_FRAME_OVERHEAD - sizeof(struct pt_regs);
324
	lc->async_stack = (unsigned long)
325 326
		__alloc_bootmem(ASYNC_SIZE, ASYNC_SIZE, 0)
		+ ASYNC_SIZE - STACK_FRAME_OVERHEAD - sizeof(struct pt_regs);
327
	lc->panic_stack = (unsigned long)
328 329
		__alloc_bootmem(PAGE_SIZE, PAGE_SIZE, 0)
		+ PAGE_SIZE - STACK_FRAME_OVERHEAD - sizeof(struct pt_regs);
330 331
	lc->current_task = (unsigned long) init_thread_union.thread_info.task;
	lc->thread_info = (unsigned long) &init_thread_union;
332
	lc->machine_flags = S390_lowcore.machine_flags;
333 334 335
	lc->stfl_fac_list = S390_lowcore.stfl_fac_list;
	memcpy(lc->stfle_fac_list, S390_lowcore.stfle_fac_list,
	       MAX_FACILITY_BIT/8);
336
#ifndef CONFIG_64BIT
337 338
	if (MACHINE_HAS_IEEE) {
		lc->extended_save_area_addr = (__u32)
339
			__alloc_bootmem_low(PAGE_SIZE, PAGE_SIZE, 0);
340
		/* enable extended save area */
H
Heiko Carstens 已提交
341
		__ctl_set_bit(14, 29);
342
	}
343 344
#else
	lc->vdso_per_cpu_data = (unsigned long) &lc->paste[0];
345
#endif
346 347 348 349 350 351 352 353
	lc->sync_enter_timer = S390_lowcore.sync_enter_timer;
	lc->async_enter_timer = S390_lowcore.async_enter_timer;
	lc->exit_timer = S390_lowcore.exit_timer;
	lc->user_timer = S390_lowcore.user_timer;
	lc->system_timer = S390_lowcore.system_timer;
	lc->steal_timer = S390_lowcore.steal_timer;
	lc->last_update_timer = S390_lowcore.last_update_timer;
	lc->last_update_clock = S390_lowcore.last_update_clock;
354
	lc->ftrace_func = S390_lowcore.ftrace_func;
M
Martin Schwidefsky 已提交
355 356 357 358 359 360

	restart_stack = __alloc_bootmem(ASYNC_SIZE, ASYNC_SIZE, 0);
	restart_stack += ASYNC_SIZE;

	/*
	 * Set up PSW restart to call ipl.c:do_restart(). Copy the relevant
361
	 * restart data to the absolute zero lowcore. This is necessary if
M
Martin Schwidefsky 已提交
362 363 364 365 366 367
	 * PSW restart is done on an offline CPU that has lowcore zero.
	 */
	lc->restart_stack = (unsigned long) restart_stack;
	lc->restart_fn = (unsigned long) do_restart;
	lc->restart_data = 0;
	lc->restart_source = -1UL;
368 369

	/* Setup absolute zero lowcore */
370 371 372 373 374
	mem_assign_absolute(S390_lowcore.restart_stack, lc->restart_stack);
	mem_assign_absolute(S390_lowcore.restart_fn, lc->restart_fn);
	mem_assign_absolute(S390_lowcore.restart_data, lc->restart_data);
	mem_assign_absolute(S390_lowcore.restart_source, lc->restart_source);
	mem_assign_absolute(S390_lowcore.restart_psw, lc->restart_psw);
M
Martin Schwidefsky 已提交
375

376 377 378 379
#ifdef CONFIG_SMP
	lc->spinlock_lockval = arch_spin_lockval(0);
#endif

380
	set_prefix((u32)(unsigned long) lc);
F
Frank Munzert 已提交
381
	lowcore_ptr[0] = lc;
382 383
}

384 385 386 387 388 389 390 391 392 393
static struct resource code_resource = {
	.name  = "Kernel code",
	.flags = IORESOURCE_BUSY | IORESOURCE_MEM,
};

static struct resource data_resource = {
	.name = "Kernel data",
	.flags = IORESOURCE_BUSY | IORESOURCE_MEM,
};

394 395 396 397 398
static struct resource bss_resource = {
	.name = "Kernel bss",
	.flags = IORESOURCE_BUSY | IORESOURCE_MEM,
};

399 400 401
static struct resource __initdata *standard_resources[] = {
	&code_resource,
	&data_resource,
402
	&bss_resource,
403 404 405
};

static void __init setup_resources(void)
406
{
407
	struct resource *res, *std_res, *sub_res;
408 409
	struct memblock_region *reg;
	int j;
410

411 412 413 414
	code_resource.start = (unsigned long) &_text;
	code_resource.end = (unsigned long) &_etext - 1;
	data_resource.start = (unsigned long) &_etext;
	data_resource.end = (unsigned long) &_edata - 1;
415 416
	bss_resource.start = (unsigned long) &__bss_start;
	bss_resource.end = (unsigned long) &__bss_stop - 1;
417

418
	for_each_memblock(memory, reg) {
419
		res = alloc_bootmem_low(sizeof(*res));
420
		res->flags = IORESOURCE_BUSY | IORESOURCE_MEM;
421 422 423 424

		res->name = "System RAM";
		res->start = reg->base;
		res->end = reg->base + reg->size - 1;
425
		request_resource(&iomem_resource, res);
H
Hongjie Yang 已提交
426

427 428 429 430 431 432 433 434 435 436 437 438 439 440
		for (j = 0; j < ARRAY_SIZE(standard_resources); j++) {
			std_res = standard_resources[j];
			if (std_res->start < res->start ||
			    std_res->start > res->end)
				continue;
			if (std_res->end > res->end) {
				sub_res = alloc_bootmem_low(sizeof(*sub_res));
				*sub_res = *std_res;
				sub_res->end = res->end;
				std_res->start = res->end + 1;
				request_resource(res, sub_res);
			} else {
				request_resource(res, std_res);
			}
H
Hongjie Yang 已提交
441
		}
442 443 444
	}
}

H
Heiko Carstens 已提交
445 446
static void __init setup_memory_end(void)
{
447
	unsigned long vmax, vmalloc_size, tmp;
448

449 450
	/* Choose kernel address space layout: 2, 3, or 4 levels. */
#ifdef CONFIG_64BIT
451
	vmalloc_size = VMALLOC_END ?: (128UL << 30) - MODULES_LEN;
452
	tmp = (memory_end ?: max_physmem_end) / PAGE_SIZE;
453 454 455 456 457
	tmp = tmp * (sizeof(struct page) + PAGE_SIZE) + vmalloc_size;
	if (tmp <= (1UL << 42))
		vmax = 1UL << 42;	/* 3-level kernel page table */
	else
		vmax = 1UL << 53;	/* 4-level kernel page table */
458 459 460 461
	/* module area is at the end of the kernel address space. */
	MODULES_END = vmax;
	MODULES_VADDR = MODULES_END - MODULES_LEN;
	VMALLOC_END = MODULES_VADDR;
462 463 464 465 466
#else
	vmalloc_size = VMALLOC_END ?: 96UL << 20;
	vmax = 1UL << 31;		/* 2-level kernel page table */
	/* vmalloc area is at the end of the kernel address space. */
	VMALLOC_END = vmax;
467
#endif
468 469 470 471
	VMALLOC_START = vmax - vmalloc_size;

	/* Split remaining virtual space between 1:1 mapping & vmemmap array */
	tmp = VMALLOC_START / (PAGE_SIZE + sizeof(struct page));
472 473
	/* vmemmap contains a multiple of PAGES_PER_SECTION struct pages */
	tmp = SECTION_ALIGN_UP(tmp);
474 475 476 477 478 479
	tmp = VMALLOC_START - tmp * sizeof(struct page);
	tmp &= ~((vmax >> 11) - 1);	/* align to page table level */
	tmp = min(tmp, 1UL << MAX_PHYSMEM_BITS);
	vmemmap = (struct page *) tmp;

	/* Take care that memory_end is set and <= vmemmap */
480 481 482
	memory_end = min(memory_end ?: max_physmem_end, tmp);
	max_pfn = max_low_pfn = PFN_DOWN(memory_end);
	memblock_remove(memory_end, ULONG_MAX);
H
Heiko Carstens 已提交
483

484
	pr_notice("Max memory size: %luMB\n", memory_end >> 20);
H
Heiko Carstens 已提交
485 486
}

M
Michael Holzheu 已提交
487 488
static void __init setup_vmcoreinfo(void)
{
489
	mem_assign_absolute(S390_lowcore.vmcore_info, paddr_vmcoreinfo_note());
M
Michael Holzheu 已提交
490 491
}

M
Michael Holzheu 已提交
492 493 494 495 496 497 498 499 500 501 502 503
#ifdef CONFIG_CRASH_DUMP

/*
 * When kdump is enabled, we have to ensure that no memory from
 * the area [0 - crashkernel memory size] and
 * [crashk_res.start - crashk_res.end] is set offline.
 */
static int kdump_mem_notifier(struct notifier_block *nb,
			      unsigned long action, void *data)
{
	struct memory_notify *arg = data;

504 505
	if (action != MEM_GOING_OFFLINE)
		return NOTIFY_OK;
M
Michael Holzheu 已提交
506 507 508 509 510 511 512 513 514 515 516 517 518 519 520
	if (arg->start_pfn < PFN_DOWN(resource_size(&crashk_res)))
		return NOTIFY_BAD;
	if (arg->start_pfn > PFN_DOWN(crashk_res.end))
		return NOTIFY_OK;
	if (arg->start_pfn + arg->nr_pages - 1 < PFN_DOWN(crashk_res.start))
		return NOTIFY_OK;
	return NOTIFY_BAD;
}

static struct notifier_block kdump_mem_nb = {
	.notifier_call = kdump_mem_notifier,
};

#endif

521 522 523 524 525
/*
 * Make sure that the area behind memory_end is protected
 */
static void reserve_memory_end(void)
{
526
#ifdef CONFIG_CRASH_DUMP
527 528 529 530 531 532 533 534 535 536 537 538
	if (ipl_info.type == IPL_TYPE_FCP_DUMP &&
	    !OLDMEM_BASE && sclp_get_hsa_size()) {
		memory_end = sclp_get_hsa_size();
		memory_end &= PAGE_MASK;
		memory_end_set = 1;
	}
#endif
	if (!memory_end_set)
		return;
	memblock_reserve(memory_end, ULONG_MAX);
}

M
Michael Holzheu 已提交
539 540 541 542 543 544
/*
 * Make sure that oldmem, where the dump is stored, is protected
 */
static void reserve_oldmem(void)
{
#ifdef CONFIG_CRASH_DUMP
545 546 547 548 549
	if (OLDMEM_BASE)
		/* Forget all memory above the running kdump system */
		memblock_reserve(OLDMEM_SIZE, (phys_addr_t)ULONG_MAX);
#endif
}
M
Michael Holzheu 已提交
550

551 552 553 554 555 556 557 558 559
/*
 * Make sure that oldmem, where the dump is stored, is protected
 */
static void remove_oldmem(void)
{
#ifdef CONFIG_CRASH_DUMP
	if (OLDMEM_BASE)
		/* Forget all memory above the running kdump system */
		memblock_remove(OLDMEM_SIZE, (phys_addr_t)ULONG_MAX);
M
Michael Holzheu 已提交
560 561 562 563 564 565 566 567 568 569
#endif
}

/*
 * Reserve memory for kdump kernel to be loaded with kexec
 */
static void __init reserve_crashkernel(void)
{
#ifdef CONFIG_CRASH_DUMP
	unsigned long long crash_base, crash_size;
570
	phys_addr_t low, high;
M
Michael Holzheu 已提交
571 572 573 574
	int rc;

	rc = parse_crashkernel(boot_command_line, memory_end, &crash_size,
			       &crash_base);
575

576 577
	crash_base = ALIGN(crash_base, KEXEC_CRASH_MEM_ALIGN);
	crash_size = ALIGN(crash_size, KEXEC_CRASH_MEM_ALIGN);
578
	if (rc || crash_size == 0)
M
Michael Holzheu 已提交
579
		return;
580 581 582 583

	if (memblock.memory.regions[0].size < crash_size) {
		pr_info("crashkernel reservation failed: %s\n",
			"first memory chunk must be at least crashkernel size");
M
Michael Holzheu 已提交
584 585
		return;
	}
586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609

	low = crash_base ?: OLDMEM_BASE;
	high = low + crash_size;
	if (low >= OLDMEM_BASE && high <= OLDMEM_BASE + OLDMEM_SIZE) {
		/* The crashkernel fits into OLDMEM, reuse OLDMEM */
		crash_base = low;
	} else {
		/* Find suitable area in free memory */
		low = max_t(unsigned long, crash_size, sclp_get_hsa_size());
		high = crash_base ? crash_base + crash_size : ULONG_MAX;

		if (crash_base && crash_base < low) {
			pr_info("crashkernel reservation failed: %s\n",
				"crash_base too low");
			return;
		}
		low = crash_base ?: low;
		crash_base = memblock_find_in_range(low, high, crash_size,
						    KEXEC_CRASH_MEM_ALIGN);
	}

	if (!crash_base) {
		pr_info("crashkernel reservation failed: %s\n",
			"no suitable area found");
M
Michael Holzheu 已提交
610 611
		return;
	}
612 613 614 615

	if (register_memory_notifier(&kdump_mem_nb))
		return;

M
Michael Holzheu 已提交
616 617 618 619 620
	if (!OLDMEM_BASE && MACHINE_IS_VM)
		diag10_range(PFN_DOWN(crash_base), PFN_DOWN(crash_size));
	crashk_res.start = crash_base;
	crashk_res.end = crash_base + crash_size - 1;
	insert_resource(&iomem_resource, &crashk_res);
621
	memblock_remove(crash_base, crash_size);
M
Michael Holzheu 已提交
622 623
	pr_info("Reserving %lluMB of memory at %lluMB "
		"for crashkernel (System RAM: %luMB)\n",
624 625
		crash_size >> 20, crash_base >> 20,
		(unsigned long)memblock.memory.total_size >> 20);
626
	os_info_crashkernel_add(crash_base, crash_size);
M
Michael Holzheu 已提交
627 628 629
#endif
}

630 631 632 633
/*
 * Reserve the initrd from being used by memblock
 */
static void __init reserve_initrd(void)
634
{
635 636 637 638 639 640
#ifdef CONFIG_BLK_DEV_INITRD
	initrd_start = INITRD_START;
	initrd_end = initrd_start + INITRD_SIZE;
	memblock_reserve(INITRD_START, INITRD_SIZE);
#endif
}
L
Linus Torvalds 已提交
641

642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662
/*
 * Check for initrd being in usable memory
 */
static void __init check_initrd(void)
{
#ifdef CONFIG_BLK_DEV_INITRD
	if (INITRD_START && INITRD_SIZE &&
	    !memblock_is_region_memory(INITRD_START, INITRD_SIZE)) {
		pr_err("initrd does not fit memory.\n");
		memblock_free(INITRD_START, INITRD_SIZE);
		initrd_start = initrd_end = 0;
	}
#endif
}

/*
 * Reserve all kernel text
 */
static void __init reserve_kernel(void)
{
	unsigned long start_pfn;
663
	start_pfn = PFN_UP(__pa(&_end));
L
Linus Torvalds 已提交
664

665
	/*
666
	 * Reserve memory used for lowcore/command line/kernel image.
667
	 */
668 669 670 671
	memblock_reserve(0, (unsigned long)_ehead);
	memblock_reserve((unsigned long)_stext, PFN_PHYS(start_pfn)
			 - (unsigned long)_stext);
}
672

673 674
static void __init reserve_elfcorehdr(void)
{
M
Michael Holzheu 已提交
675
#ifdef CONFIG_CRASH_DUMP
676 677 678
	if (is_kdump_kernel())
		memblock_reserve(elfcorehdr_addr - OLDMEM_BASE,
				 PAGE_ALIGN(elfcorehdr_size));
679
#endif
680
}
681

682 683 684
static void __init setup_memory(void)
{
	struct memblock_region *reg;
L
Linus Torvalds 已提交
685 686

	/*
687
	 * Init storage key for present memory
L
Linus Torvalds 已提交
688
	 */
689 690
	for_each_memblock(memory, reg) {
		storage_key_init_range(reg->base, reg->base + reg->size);
L
Linus Torvalds 已提交
691
	}
692 693
	psw_set_key(PAGE_DEFAULT_KEY);

694 695
	/* Only cosmetics */
	memblock_enforce_memory_limit(memblock_end_of_DRAM());
696
}
L
Linus Torvalds 已提交
697

698 699 700 701 702 703
/*
 * Setup hardware capabilities.
 */
static void __init setup_hwcaps(void)
{
	static const int stfl_bits[6] = { 0, 2, 7, 17, 19, 21 };
704
	struct cpuid cpu_id;
705 706 707 708 709 710 711 712 713 714 715 716
	int i;

	/*
	 * The store facility list bits numbers as found in the principles
	 * of operation are numbered with bit 1UL<<31 as number 0 to
	 * bit 1UL<<0 as number 31.
	 *   Bit 0: instructions named N3, "backported" to esa-mode
	 *   Bit 2: z/Architecture mode is active
	 *   Bit 7: the store-facility-list-extended facility is installed
	 *   Bit 17: the message-security assist is installed
	 *   Bit 19: the long-displacement facility is installed
	 *   Bit 21: the extended-immediate facility is installed
717 718
	 *   Bit 22: extended-translation facility 3 is installed
	 *   Bit 30: extended-translation facility 3 enhancement facility
719 720 721
	 * These get translated to:
	 *   HWCAP_S390_ESAN3 bit 0, HWCAP_S390_ZARCH bit 1,
	 *   HWCAP_S390_STFLE bit 2, HWCAP_S390_MSA bit 3,
722 723
	 *   HWCAP_S390_LDISP bit 4, HWCAP_S390_EIMM bit 5 and
	 *   HWCAP_S390_ETF3EH bit 8 (22 && 30).
724 725
	 */
	for (i = 0; i < 6; i++)
726
		if (test_facility(stfl_bits[i]))
727 728
			elf_hwcap |= 1UL << i;

729
	if (test_facility(22) && test_facility(30))
730
		elf_hwcap |= HWCAP_S390_ETF3EH;
731

732 733 734 735 736 737
	/*
	 * Check for additional facilities with store-facility-list-extended.
	 * stfle stores doublewords (8 byte) with bit 1ULL<<63 as bit 0
	 * and 1ULL<<0 as bit 63. Bits 0-31 contain the same information
	 * as stored by stfl, bits 32-xxx contain additional facilities.
	 * How many facility words are stored depends on the number of
L
Lucas De Marchi 已提交
738
	 * doublewords passed to the instruction. The additional facilities
739
	 * are:
740 741
	 *   Bit 42: decimal floating point facility is installed
	 *   Bit 44: perform floating point operation facility is installed
742
	 * translated to:
743
	 *   HWCAP_S390_DFP bit 6 (42 && 44).
744
	 */
745 746
	if ((elf_hwcap & (1UL << 2)) && test_facility(42) && test_facility(44))
		elf_hwcap |= HWCAP_S390_DFP;
747

748 749 750
	/*
	 * Huge page support HWCAP_S390_HPAGE is bit 7.
	 */
751
	if (MACHINE_HAS_HPAGE)
752 753
		elf_hwcap |= HWCAP_S390_HPAGE;

754
#if defined(CONFIG_64BIT)
755 756 757 758 759
	/*
	 * 64-bit register support for 31-bit processes
	 * HWCAP_S390_HIGH_GPRS is bit 9.
	 */
	elf_hwcap |= HWCAP_S390_HIGH_GPRS;
760 761 762 763 764 765

	/*
	 * Transactional execution support HWCAP_S390_TE is bit 10.
	 */
	if (test_facility(50) && test_facility(73))
		elf_hwcap |= HWCAP_S390_TE;
766
#endif
767

768 769
	get_cpu_id(&cpu_id);
	switch (cpu_id.machine) {
770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787
	case 0x9672:
#if !defined(CONFIG_64BIT)
	default:	/* Use "g5" as default for 31 bit kernels. */
#endif
		strcpy(elf_platform, "g5");
		break;
	case 0x2064:
	case 0x2066:
#if defined(CONFIG_64BIT)
	default:	/* Use "z900" as default for 64 bit kernels. */
#endif
		strcpy(elf_platform, "z900");
		break;
	case 0x2084:
	case 0x2086:
		strcpy(elf_platform, "z990");
		break;
	case 0x2094:
788
	case 0x2096:
789 790
		strcpy(elf_platform, "z9-109");
		break;
791 792 793 794
	case 0x2097:
	case 0x2098:
		strcpy(elf_platform, "z10");
		break;
795
	case 0x2817:
796
	case 0x2818:
797 798
		strcpy(elf_platform, "z196");
		break;
799
	case 0x2827:
800
	case 0x2828:
801 802
		strcpy(elf_platform, "zEC12");
		break;
803 804 805
	}
}

806 807 808 809
/*
 * Setup function called from init/main.c just after the banner
 * was printed.
 */
L
Linus Torvalds 已提交
810

M
Martin Schwidefsky 已提交
811
void __init setup_arch(char **cmdline_p)
812
{
L
Linus Torvalds 已提交
813
        /*
814
         * print what head.S has found out about the machine
L
Linus Torvalds 已提交
815
         */
816
#ifndef CONFIG_64BIT
817 818 819
	if (MACHINE_IS_VM)
		pr_info("Linux is running as a z/VM "
			"guest operating system in 31-bit mode\n");
820
	else if (MACHINE_IS_LPAR)
821 822 823 824 825 826 827
		pr_info("Linux is running natively in 31-bit mode\n");
	if (MACHINE_HAS_IEEE)
		pr_info("The hardware system has IEEE compatible "
			"floating point units\n");
	else
		pr_info("The hardware system has no IEEE compatible "
			"floating point units\n");
828
#else /* CONFIG_64BIT */
829
	if (MACHINE_IS_VM)
830 831
		pr_info("Linux is running as a z/VM "
			"guest operating system in 64-bit mode\n");
832
	else if (MACHINE_IS_KVM)
833
		pr_info("Linux is running under KVM in 64-bit mode\n");
834
	else if (MACHINE_IS_LPAR)
835
		pr_info("Linux is running natively in 64-bit mode\n");
836
#endif /* CONFIG_64BIT */
837

838 839 840
	/* Have one command line that is parsed and saved in /proc/cmdline */
	/* boot_command_line has been already set up in early.c */
	*cmdline_p = boot_command_line;
841

842
        ROOT_DEV = Root_RAM0;
843

844
	/* Is init_mm really needed? */
845 846 847 848 849 850
	init_mm.start_code = PAGE_OFFSET;
	init_mm.end_code = (unsigned long) &_etext;
	init_mm.end_data = (unsigned long) &_edata;
	init_mm.brk = (unsigned long) &_end;

	parse_early_param();
851
	os_info_init();
852
	setup_ipl();
853 854 855

	/* Do some memory reservations *before* memory is added to memblock */
	reserve_memory_end();
856
	reserve_oldmem();
857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874
	reserve_kernel();
	reserve_initrd();
	reserve_elfcorehdr();
	memblock_allow_resize();

	/* Get information about *all* installed memory */
	detect_memory_memblock();

	remove_oldmem();

	/*
	 * Make sure all chunks are MAX_ORDER aligned so we don't need the
	 * extra checks that HOLES_IN_ZONE would require.
	 *
	 * Is this still required?
	 */
	memblock_trim_memory(1UL << (MAX_ORDER - 1 + PAGE_SHIFT));

H
Heiko Carstens 已提交
875
	setup_memory_end();
876
	setup_memory();
877 878 879 880

	check_initrd();
	reserve_crashkernel();

881
	setup_resources();
M
Michael Holzheu 已提交
882
	setup_vmcoreinfo();
883
	setup_lowcore();
884
	smp_fill_possible_mask();
L
Linus Torvalds 已提交
885
        cpu_init();
886
	s390_init_cpu_topology();
L
Linus Torvalds 已提交
887

888 889 890 891 892
	/*
	 * Setup capabilities (ELF_HWCAP & ELF_PLATFORM).
	 */
	setup_hwcaps();

L
Linus Torvalds 已提交
893 894 895 896 897 898 899
	/*
	 * Create kernel page tables and switch to virtual addressing.
	 */
        paging_init();

        /* Setup default console */
	conmode_default();
900
	set_preferred_console();
M
Michael Holzheu 已提交
901 902

	/* Setup zfcpdump support */
903
	setup_zfcpdump();
L
Linus Torvalds 已提交
904
}
905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936

#ifdef CONFIG_32BIT
static int no_removal_warning __initdata;

static int __init parse_no_removal_warning(char *str)
{
	no_removal_warning = 1;
	return 0;
}
__setup("no_removal_warning", parse_no_removal_warning);

static int __init removal_warning(void)
{
	if (no_removal_warning)
		return 0;
	printk(KERN_ALERT "\n\n");
	printk(KERN_CONT "Warning - you are using a 31 bit kernel!\n\n");
	printk(KERN_CONT "We plan to remove 31 bit kernel support from the kernel sources in March 2015.\n");
	printk(KERN_CONT "Currently we assume that nobody is using the 31 bit kernel on old 31 bit\n");
	printk(KERN_CONT "hardware anymore. If you think that the code should not be removed and also\n");
	printk(KERN_CONT "future versions of the Linux kernel should be able to run in 31 bit mode\n");
	printk(KERN_CONT "please let us know. Please write to:\n");
	printk(KERN_CONT "linux390@de.ibm.com (mail address) and/or\n");
	printk(KERN_CONT "linux-s390@vger.kernel.org (mailing list).\n\n");
	printk(KERN_CONT "Thank you!\n\n");
	printk(KERN_CONT "If this kernel runs on a 64 bit machine you may consider using a 64 bit kernel.\n");
	printk(KERN_CONT "This message can be disabled with the \"no_removal_warning\" kernel parameter.\n");
	schedule_timeout_uninterruptible(300 * HZ);
	return 0;
}
early_initcall(removal_warning);
#endif