Linux Audio

Check our new training course

Loading...
v3.5.6
 
   1/*
   2 *  arch/s390/kernel/setup.c
   3 *
   4 *  S390 version
   5 *    Copyright (C) IBM Corp. 1999,2012
   6 *    Author(s): Hartmut Penner (hp@de.ibm.com),
   7 *               Martin Schwidefsky (schwidefsky@de.ibm.com)
   8 *
   9 *  Derived from "arch/i386/kernel/setup.c"
  10 *    Copyright (C) 1995, Linus Torvalds
  11 */
  12
  13/*
  14 * This file handles the architecture-dependent parts of initialization
  15 */
  16
  17#define KMSG_COMPONENT "setup"
  18#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
  19
  20#include <linux/errno.h>
  21#include <linux/module.h>
  22#include <linux/sched.h>
 
 
  23#include <linux/kernel.h>
  24#include <linux/memblock.h>
  25#include <linux/mm.h>
  26#include <linux/stddef.h>
  27#include <linux/unistd.h>
  28#include <linux/ptrace.h>
 
  29#include <linux/user.h>
  30#include <linux/tty.h>
  31#include <linux/ioport.h>
  32#include <linux/delay.h>
  33#include <linux/init.h>
  34#include <linux/initrd.h>
  35#include <linux/bootmem.h>
  36#include <linux/root_dev.h>
  37#include <linux/console.h>
  38#include <linux/kernel_stat.h>
 
  39#include <linux/device.h>
  40#include <linux/notifier.h>
  41#include <linux/pfn.h>
  42#include <linux/ctype.h>
  43#include <linux/reboot.h>
  44#include <linux/topology.h>
  45#include <linux/ftrace.h>
  46#include <linux/kexec.h>
  47#include <linux/crash_dump.h>
  48#include <linux/memory.h>
  49#include <linux/compat.h>
 
 
 
  50
 
  51#include <asm/ipl.h>
  52#include <asm/uaccess.h>
  53#include <asm/facility.h>
  54#include <asm/smp.h>
  55#include <asm/mmu_context.h>
  56#include <asm/cpcmd.h>
  57#include <asm/lowcore.h>
 
  58#include <asm/irq.h>
  59#include <asm/page.h>
  60#include <asm/ptrace.h>
  61#include <asm/sections.h>
  62#include <asm/ebcdic.h>
  63#include <asm/kvm_virtio.h>
  64#include <asm/diag.h>
  65#include <asm/os_info.h>
 
 
 
 
 
 
 
 
 
  66#include "entry.h"
  67
  68long psw_kernel_bits	= PSW_DEFAULT_KEY | PSW_MASK_BASE | PSW_ASC_PRIMARY |
  69			  PSW_MASK_EA | PSW_MASK_BA;
  70long psw_user_bits	= PSW_MASK_DAT | PSW_MASK_IO | PSW_MASK_EXT |
  71			  PSW_DEFAULT_KEY | PSW_MASK_BASE | PSW_MASK_MCHECK |
  72			  PSW_MASK_PSTATE | PSW_ASC_HOME;
  73
  74/*
  75 * User copy operations.
  76 */
  77struct uaccess_ops uaccess;
  78EXPORT_SYMBOL(uaccess);
  79
  80/*
  81 * Machine setup..
  82 */
  83unsigned int console_mode = 0;
  84EXPORT_SYMBOL(console_mode);
  85
  86unsigned int console_devno = -1;
  87EXPORT_SYMBOL(console_devno);
  88
  89unsigned int console_irq = -1;
  90EXPORT_SYMBOL(console_irq);
  91
  92unsigned long elf_hwcap = 0;
  93char elf_platform[ELF_PLATFORM_SIZE];
  94
  95struct mem_chunk __initdata memory_chunk[MEMORY_CHUNKS];
  96
  97int __initdata memory_end_set;
  98unsigned long __initdata memory_end;
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  99
 100unsigned long VMALLOC_START;
 101EXPORT_SYMBOL(VMALLOC_START);
 102
 103unsigned long VMALLOC_END;
 104EXPORT_SYMBOL(VMALLOC_END);
 105
 106struct page *vmemmap;
 107EXPORT_SYMBOL(vmemmap);
 
 
 
 
 108
 109/* An array with a pointer to the lowcore of every CPU. */
 110struct _lowcore *lowcore_ptr[NR_CPUS];
 111EXPORT_SYMBOL(lowcore_ptr);
 112
 113/*
 
 
 
 
 
 
 114 * This is set up by the setup-routine at boot-time
 115 * for S390 need to find out, what we have to setup
 116 * using address 0x10400 ...
 117 */
 118
 119#include <asm/setup.h>
 120
 121/*
 122 * condev= and conmode= setup parameter.
 123 */
 124
 125static int __init condev_setup(char *str)
 126{
 127	int vdev;
 128
 129	vdev = simple_strtoul(str, &str, 0);
 130	if (vdev >= 0 && vdev < 65536) {
 131		console_devno = vdev;
 132		console_irq = -1;
 133	}
 134	return 1;
 135}
 136
 137__setup("condev=", condev_setup);
 138
 139static void __init set_preferred_console(void)
 140{
 141	if (MACHINE_IS_KVM)
 142		add_preferred_console("hvc", 0, NULL);
 143	else if (CONSOLE_IS_3215 || CONSOLE_IS_SCLP)
 144		add_preferred_console("ttyS", 0, NULL);
 145	else if (CONSOLE_IS_3270)
 146		add_preferred_console("tty3270", 0, NULL);
 
 
 
 
 147}
 148
 149static int __init conmode_setup(char *str)
 150{
 151#if defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
 152	if (strncmp(str, "hwc", 4) == 0 || strncmp(str, "sclp", 5) == 0)
 153                SET_CONSOLE_SCLP;
 154#endif
 155#if defined(CONFIG_TN3215_CONSOLE)
 156	if (strncmp(str, "3215", 5) == 0)
 157		SET_CONSOLE_3215;
 158#endif
 159#if defined(CONFIG_TN3270_CONSOLE)
 160	if (strncmp(str, "3270", 5) == 0)
 161		SET_CONSOLE_3270;
 162#endif
 163	set_preferred_console();
 164        return 1;
 165}
 166
 167__setup("conmode=", conmode_setup);
 168
 169static void __init conmode_default(void)
 170{
 171	char query_buffer[1024];
 172	char *ptr;
 173
 174        if (MACHINE_IS_VM) {
 175		cpcmd("QUERY CONSOLE", query_buffer, 1024, NULL);
 176		console_devno = simple_strtoul(query_buffer + 5, NULL, 16);
 177		ptr = strstr(query_buffer, "SUBCHANNEL =");
 178		console_irq = simple_strtoul(ptr + 13, NULL, 16);
 179		cpcmd("QUERY TERM", query_buffer, 1024, NULL);
 180		ptr = strstr(query_buffer, "CONMODE");
 181		/*
 182		 * Set the conmode to 3215 so that the device recognition 
 183		 * will set the cu_type of the console to 3215. If the
 184		 * conmode is 3270 and we don't set it back then both
 185		 * 3215 and the 3270 driver will try to access the console
 186		 * device (3215 as console and 3270 as normal tty).
 187		 */
 188		cpcmd("TERM CONMODE 3215", NULL, 0, NULL);
 189		if (ptr == NULL) {
 190#if defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
 191			SET_CONSOLE_SCLP;
 192#endif
 193			return;
 194		}
 195		if (strncmp(ptr + 8, "3270", 4) == 0) {
 196#if defined(CONFIG_TN3270_CONSOLE)
 197			SET_CONSOLE_3270;
 198#elif defined(CONFIG_TN3215_CONSOLE)
 199			SET_CONSOLE_3215;
 200#elif defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
 201			SET_CONSOLE_SCLP;
 202#endif
 203		} else if (strncmp(ptr + 8, "3215", 4) == 0) {
 204#if defined(CONFIG_TN3215_CONSOLE)
 205			SET_CONSOLE_3215;
 206#elif defined(CONFIG_TN3270_CONSOLE)
 207			SET_CONSOLE_3270;
 208#elif defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
 209			SET_CONSOLE_SCLP;
 210#endif
 211		}
 
 
 
 
 
 
 
 212	} else {
 213#if defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
 214		SET_CONSOLE_SCLP;
 215#endif
 216	}
 217}
 218
 219#ifdef CONFIG_ZFCPDUMP
 220static void __init setup_zfcpdump(unsigned int console_devno)
 221{
 222	static char str[41];
 223
 224	if (ipl_info.type != IPL_TYPE_FCP_DUMP)
 225		return;
 226	if (OLDMEM_BASE)
 227		return;
 228	if (console_devno != -1)
 229		sprintf(str, " cio_ignore=all,!0.0.%04x,!0.0.%04x",
 230			ipl_info.data.fcp.dev_id.devno, console_devno);
 231	else
 232		sprintf(str, " cio_ignore=all,!0.0.%04x",
 233			ipl_info.data.fcp.dev_id.devno);
 234	strcat(boot_command_line, str);
 235	console_loglevel = 2;
 236}
 237#else
 238static inline void setup_zfcpdump(unsigned int console_devno) {}
 239#endif /* CONFIG_ZFCPDUMP */
 240
 241 /*
 242 * Reboot, halt and power_off stubs. They just call _machine_restart,
 243 * _machine_halt or _machine_power_off. 
 244 */
 245
 246void machine_restart(char *command)
 247{
 248	if ((!in_interrupt() && !in_atomic()) || oops_in_progress)
 249		/*
 250		 * Only unblank the console if we are called in enabled
 251		 * context or a bust_spinlocks cleared the way for us.
 252		 */
 253		console_unblank();
 254	_machine_restart(command);
 255}
 256
 257void machine_halt(void)
 258{
 259	if (!in_interrupt() || oops_in_progress)
 260		/*
 261		 * Only unblank the console if we are called in enabled
 262		 * context or a bust_spinlocks cleared the way for us.
 263		 */
 264		console_unblank();
 265	_machine_halt();
 266}
 267
 268void machine_power_off(void)
 269{
 270	if (!in_interrupt() || oops_in_progress)
 271		/*
 272		 * Only unblank the console if we are called in enabled
 273		 * context or a bust_spinlocks cleared the way for us.
 274		 */
 275		console_unblank();
 276	_machine_power_off();
 277}
 278
 279/*
 280 * Dummy power off function.
 281 */
 282void (*pm_power_off)(void) = machine_power_off;
 
 283
 284static int __init early_parse_mem(char *p)
 285{
 286	memory_end = memparse(p, &p);
 287	memory_end_set = 1;
 288	return 0;
 289}
 290early_param("mem", early_parse_mem);
 291
 292static int __init parse_vmalloc(char *arg)
 293{
 294	if (!arg)
 295		return -EINVAL;
 296	VMALLOC_END = (memparse(arg, &arg) + PAGE_SIZE - 1) & PAGE_MASK;
 297	return 0;
 298}
 299early_param("vmalloc", parse_vmalloc);
 300
 301unsigned int user_mode = HOME_SPACE_MODE;
 302EXPORT_SYMBOL_GPL(user_mode);
 303
 304static int set_amode_primary(void)
 305{
 306	psw_kernel_bits = (psw_kernel_bits & ~PSW_MASK_ASC) | PSW_ASC_HOME;
 307	psw_user_bits = (psw_user_bits & ~PSW_MASK_ASC) | PSW_ASC_PRIMARY;
 308#ifdef CONFIG_COMPAT
 309	psw32_user_bits =
 310		(psw32_user_bits & ~PSW32_MASK_ASC) | PSW32_ASC_PRIMARY;
 311#endif
 312
 313	if (MACHINE_HAS_MVCOS) {
 314		memcpy(&uaccess, &uaccess_mvcos_switch, sizeof(uaccess));
 315		return 1;
 316	} else {
 317		memcpy(&uaccess, &uaccess_pt, sizeof(uaccess));
 318		return 0;
 319	}
 320}
 321
 322/*
 323 * Switch kernel/user addressing modes?
 324 */
 325static int __init early_parse_switch_amode(char *p)
 326{
 327	user_mode = PRIMARY_SPACE_MODE;
 328	return 0;
 
 
 
 329}
 330early_param("switch_amode", early_parse_switch_amode);
 331
 332static int __init early_parse_user_mode(char *p)
 333{
 334	if (p && strcmp(p, "primary") == 0)
 335		user_mode = PRIMARY_SPACE_MODE;
 336	else if (!p || strcmp(p, "home") == 0)
 337		user_mode = HOME_SPACE_MODE;
 338	else
 339		return 1;
 340	return 0;
 341}
 342early_param("user_mode", early_parse_user_mode);
 343
 344static void setup_addressing_mode(void)
 345{
 346	if (user_mode == PRIMARY_SPACE_MODE) {
 347		if (set_amode_primary())
 348			pr_info("Address spaces switched, "
 349				"mvcos available\n");
 350		else
 351			pr_info("Address spaces switched, "
 352				"mvcos not available\n");
 353	}
 354}
 355
 356void *restart_stack __attribute__((__section__(".data")));
 
 
 
 
 
 
 
 
 
 
 
 357
 358static void __init setup_lowcore(void)
 359{
 360	struct _lowcore *lc;
 
 
 
 
 
 361
 362	/*
 363	 * Setup lowcore for boot cpu
 364	 */
 365	BUILD_BUG_ON(sizeof(struct _lowcore) != LC_PAGES * 4096);
 366	lc = __alloc_bootmem_low(LC_PAGES * PAGE_SIZE, LC_PAGES * PAGE_SIZE, 0);
 367	lc->restart_psw.mask = psw_kernel_bits;
 368	lc->restart_psw.addr =
 369		PSW_ADDR_AMODE | (unsigned long) restart_int_handler;
 370	lc->external_new_psw.mask = psw_kernel_bits |
 371		PSW_MASK_DAT | PSW_MASK_MCHECK;
 372	lc->external_new_psw.addr =
 373		PSW_ADDR_AMODE | (unsigned long) ext_int_handler;
 374	lc->svc_new_psw.mask = psw_kernel_bits |
 375		PSW_MASK_DAT | PSW_MASK_IO | PSW_MASK_EXT | PSW_MASK_MCHECK;
 376	lc->svc_new_psw.addr = PSW_ADDR_AMODE | (unsigned long) system_call;
 377	lc->program_new_psw.mask = psw_kernel_bits |
 378		PSW_MASK_DAT | PSW_MASK_MCHECK;
 379	lc->program_new_psw.addr =
 380		PSW_ADDR_AMODE | (unsigned long) pgm_check_handler;
 381	lc->mcck_new_psw.mask = psw_kernel_bits;
 382	lc->mcck_new_psw.addr =
 383		PSW_ADDR_AMODE | (unsigned long) mcck_int_handler;
 384	lc->io_new_psw.mask = psw_kernel_bits |
 385		PSW_MASK_DAT | PSW_MASK_MCHECK;
 386	lc->io_new_psw.addr = PSW_ADDR_AMODE | (unsigned long) io_int_handler;
 387	lc->clock_comparator = -1ULL;
 388	lc->kernel_stack = ((unsigned long) &init_thread_union) + THREAD_SIZE;
 389	lc->async_stack = (unsigned long)
 390		__alloc_bootmem(ASYNC_SIZE, ASYNC_SIZE, 0) + ASYNC_SIZE;
 391	lc->panic_stack = (unsigned long)
 392		__alloc_bootmem(PAGE_SIZE, PAGE_SIZE, 0) + PAGE_SIZE;
 393	lc->current_task = (unsigned long) init_thread_union.thread_info.task;
 394	lc->thread_info = (unsigned long) &init_thread_union;
 395	lc->machine_flags = S390_lowcore.machine_flags;
 396	lc->stfl_fac_list = S390_lowcore.stfl_fac_list;
 397	memcpy(lc->stfle_fac_list, S390_lowcore.stfle_fac_list,
 398	       MAX_FACILITY_BIT/8);
 399#ifndef CONFIG_64BIT
 400	if (MACHINE_HAS_IEEE) {
 401		lc->extended_save_area_addr = (__u32)
 402			__alloc_bootmem_low(PAGE_SIZE, PAGE_SIZE, 0);
 403		/* enable extended save area */
 404		__ctl_set_bit(14, 29);
 405	}
 406#else
 407	lc->vdso_per_cpu_data = (unsigned long) &lc->paste[0];
 408#endif
 409	lc->sync_enter_timer = S390_lowcore.sync_enter_timer;
 410	lc->async_enter_timer = S390_lowcore.async_enter_timer;
 411	lc->exit_timer = S390_lowcore.exit_timer;
 412	lc->user_timer = S390_lowcore.user_timer;
 413	lc->system_timer = S390_lowcore.system_timer;
 414	lc->steal_timer = S390_lowcore.steal_timer;
 415	lc->last_update_timer = S390_lowcore.last_update_timer;
 416	lc->last_update_clock = S390_lowcore.last_update_clock;
 417	lc->ftrace_func = S390_lowcore.ftrace_func;
 418
 419	restart_stack = __alloc_bootmem(ASYNC_SIZE, ASYNC_SIZE, 0);
 420	restart_stack += ASYNC_SIZE;
 
 
 
 
 
 
 
 421
 422	/*
 423	 * Set up PSW restart to call ipl.c:do_restart(). Copy the relevant
 424	 * restart data to the absolute zero lowcore. This is necesary if
 425	 * PSW restart is done on an offline CPU that has lowcore zero.
 426	 */
 427	lc->restart_stack = (unsigned long) restart_stack;
 428	lc->restart_fn = (unsigned long) do_restart;
 429	lc->restart_data = 0;
 430	lc->restart_source = -1UL;
 
 
 
 
 
 
 431
 432	/* Setup absolute zero lowcore */
 433	memcpy_absolute(&S390_lowcore.restart_stack, &lc->restart_stack,
 434			4 * sizeof(unsigned long));
 435	memcpy_absolute(&S390_lowcore.restart_psw, &lc->restart_psw,
 436			sizeof(lc->restart_psw));
 
 
 
 
 
 
 
 
 
 437
 438	set_prefix((u32)(unsigned long) lc);
 439	lowcore_ptr[0] = lc;
 440}
 441
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 442static struct resource code_resource = {
 443	.name  = "Kernel code",
 444	.flags = IORESOURCE_BUSY | IORESOURCE_MEM,
 445};
 446
 447static struct resource data_resource = {
 448	.name = "Kernel data",
 449	.flags = IORESOURCE_BUSY | IORESOURCE_MEM,
 450};
 451
 452static struct resource bss_resource = {
 453	.name = "Kernel bss",
 454	.flags = IORESOURCE_BUSY | IORESOURCE_MEM,
 455};
 456
 457static struct resource __initdata *standard_resources[] = {
 458	&code_resource,
 459	&data_resource,
 460	&bss_resource,
 461};
 462
 463static void __init setup_resources(void)
 464{
 465	struct resource *res, *std_res, *sub_res;
 466	int i, j;
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 467
 468	code_resource.start = (unsigned long) &_text;
 469	code_resource.end = (unsigned long) &_etext - 1;
 470	data_resource.start = (unsigned long) &_etext;
 471	data_resource.end = (unsigned long) &_edata - 1;
 472	bss_resource.start = (unsigned long) &__bss_start;
 473	bss_resource.end = (unsigned long) &__bss_stop - 1;
 474
 475	for (i = 0; i < MEMORY_CHUNKS; i++) {
 476		if (!memory_chunk[i].size)
 477			continue;
 478		if (memory_chunk[i].type == CHUNK_OLDMEM ||
 479		    memory_chunk[i].type == CHUNK_CRASHK)
 480			continue;
 481		res = alloc_bootmem_low(sizeof(*res));
 482		res->flags = IORESOURCE_BUSY | IORESOURCE_MEM;
 483		switch (memory_chunk[i].type) {
 484		case CHUNK_READ_WRITE:
 485		case CHUNK_CRASHK:
 486			res->name = "System RAM";
 487			break;
 488		case CHUNK_READ_ONLY:
 489			res->name = "System ROM";
 490			res->flags |= IORESOURCE_READONLY;
 491			break;
 492		default:
 493			res->name = "reserved";
 494		}
 495		res->start = memory_chunk[i].addr;
 496		res->end = res->start + memory_chunk[i].size - 1;
 497		request_resource(&iomem_resource, res);
 498
 499		for (j = 0; j < ARRAY_SIZE(standard_resources); j++) {
 500			std_res = standard_resources[j];
 501			if (std_res->start < res->start ||
 502			    std_res->start > res->end)
 503				continue;
 504			if (std_res->end > res->end) {
 505				sub_res = alloc_bootmem_low(sizeof(*sub_res));
 
 
 
 506				*sub_res = *std_res;
 507				sub_res->end = res->end;
 508				std_res->start = res->end + 1;
 509				request_resource(res, sub_res);
 510			} else {
 511				request_resource(res, std_res);
 512			}
 513		}
 514	}
 515}
 516
 517unsigned long real_memory_size;
 518EXPORT_SYMBOL_GPL(real_memory_size);
 519
 520static void __init setup_memory_end(void)
 521{
 522	unsigned long vmax, vmalloc_size, tmp;
 523	int i;
 524
 525
 526#ifdef CONFIG_ZFCPDUMP
 527	if (ipl_info.type == IPL_TYPE_FCP_DUMP && !OLDMEM_BASE) {
 528		memory_end = ZFCPDUMP_HSA_SIZE;
 529		memory_end_set = 1;
 530	}
 531#endif
 532	real_memory_size = 0;
 533	memory_end &= PAGE_MASK;
 534
 535	/*
 536	 * Make sure all chunks are MAX_ORDER aligned so we don't need the
 537	 * extra checks that HOLES_IN_ZONE would require.
 
 
 
 538	 */
 539	for (i = 0; i < MEMORY_CHUNKS; i++) {
 540		unsigned long start, end;
 541		struct mem_chunk *chunk;
 542		unsigned long align;
 543
 544		chunk = &memory_chunk[i];
 545		align = 1UL << (MAX_ORDER + PAGE_SHIFT - 1);
 546		start = (chunk->addr + align - 1) & ~(align - 1);
 547		end = (chunk->addr + chunk->size) & ~(align - 1);
 548		if (start >= end)
 549			memset(chunk, 0, sizeof(*chunk));
 550		else {
 551			chunk->addr = start;
 552			chunk->size = end - start;
 553		}
 554		real_memory_size = max(real_memory_size,
 555				       chunk->addr + chunk->size);
 556	}
 557
 558	/* Choose kernel address space layout: 2, 3, or 4 levels. */
 559#ifdef CONFIG_64BIT
 560	vmalloc_size = VMALLOC_END ?: 128UL << 30;
 561	tmp = (memory_end ?: real_memory_size) / PAGE_SIZE;
 562	tmp = tmp * (sizeof(struct page) + PAGE_SIZE) + vmalloc_size;
 563	if (tmp <= (1UL << 42))
 564		vmax = 1UL << 42;	/* 3-level kernel page table */
 565	else
 566		vmax = 1UL << 53;	/* 4-level kernel page table */
 567#else
 568	vmalloc_size = VMALLOC_END ?: 96UL << 20;
 569	vmax = 1UL << 31;		/* 2-level kernel page table */
 570#endif
 571	/* vmalloc area is at the end of the kernel address space. */
 572	VMALLOC_END = vmax;
 573	VMALLOC_START = vmax - vmalloc_size;
 574
 575	/* Split remaining virtual space between 1:1 mapping & vmemmap array */
 576	tmp = VMALLOC_START / (PAGE_SIZE + sizeof(struct page));
 577	tmp = VMALLOC_START - tmp * sizeof(struct page);
 578	tmp &= ~((vmax >> 11) - 1);	/* align to page table level */
 579	tmp = min(tmp, 1UL << MAX_PHYSMEM_BITS);
 580	vmemmap = (struct page *) tmp;
 581
 582	/* Take care that memory_end is set and <= vmemmap */
 583	memory_end = min(memory_end ?: real_memory_size, tmp);
 584
 585	/* Fixup memory chunk array to fit into 0..memory_end */
 586	for (i = 0; i < MEMORY_CHUNKS; i++) {
 587		struct mem_chunk *chunk = &memory_chunk[i];
 588
 589		if (chunk->addr >= memory_end) {
 590			memset(chunk, 0, sizeof(*chunk));
 591			continue;
 592		}
 593		if (chunk->addr + chunk->size > memory_end)
 594			chunk->size = memory_end - chunk->addr;
 595	}
 596}
 597
 598static void __init setup_vmcoreinfo(void)
 599{
 600#ifdef CONFIG_KEXEC
 601	unsigned long ptr = paddr_vmcoreinfo_note();
 602
 603	memcpy_absolute(&S390_lowcore.vmcore_info, &ptr, sizeof(ptr));
 604#endif
 605}
 606
 607#ifdef CONFIG_CRASH_DUMP
 608
 609/*
 610 * Find suitable location for crashkernel memory
 611 */
 612static unsigned long __init find_crash_base(unsigned long crash_size,
 613					    char **msg)
 614{
 615	unsigned long crash_base;
 616	struct mem_chunk *chunk;
 617	int i;
 618
 619	if (memory_chunk[0].size < crash_size) {
 620		*msg = "first memory chunk must be at least crashkernel size";
 621		return 0;
 622	}
 623	if (OLDMEM_BASE && crash_size == OLDMEM_SIZE)
 624		return OLDMEM_BASE;
 625
 626	for (i = MEMORY_CHUNKS - 1; i >= 0; i--) {
 627		chunk = &memory_chunk[i];
 628		if (chunk->size == 0)
 629			continue;
 630		if (chunk->type != CHUNK_READ_WRITE)
 631			continue;
 632		if (chunk->size < crash_size)
 633			continue;
 634		crash_base = (chunk->addr + chunk->size) - crash_size;
 635		if (crash_base < crash_size)
 636			continue;
 637		if (crash_base < ZFCPDUMP_HSA_SIZE_MAX)
 638			continue;
 639		if (crash_base < (unsigned long) INITRD_START + INITRD_SIZE)
 640			continue;
 641		return crash_base;
 642	}
 643	*msg = "no suitable area found";
 644	return 0;
 645}
 646
 647/*
 648 * Check if crash_base and crash_size is valid
 649 */
 650static int __init verify_crash_base(unsigned long crash_base,
 651				    unsigned long crash_size,
 652				    char **msg)
 653{
 654	struct mem_chunk *chunk;
 655	int i;
 656
 657	/*
 658	 * Because we do the swap to zero, we must have at least 'crash_size'
 659	 * bytes free space before crash_base
 660	 */
 661	if (crash_size > crash_base) {
 662		*msg = "crashkernel offset must be greater than size";
 663		return -EINVAL;
 664	}
 665
 666	/* First memory chunk must be at least crash_size */
 667	if (memory_chunk[0].size < crash_size) {
 668		*msg = "first memory chunk must be at least crashkernel size";
 669		return -EINVAL;
 670	}
 671	/* Check if we fit into the respective memory chunk */
 672	for (i = 0; i < MEMORY_CHUNKS; i++) {
 673		chunk = &memory_chunk[i];
 674		if (chunk->size == 0)
 675			continue;
 676		if (crash_base < chunk->addr)
 677			continue;
 678		if (crash_base >= chunk->addr + chunk->size)
 679			continue;
 680		/* we have found the memory chunk */
 681		if (crash_base + crash_size > chunk->addr + chunk->size) {
 682			*msg = "selected memory chunk is too small for "
 683				"crashkernel memory";
 684			return -EINVAL;
 685		}
 686		return 0;
 687	}
 688	*msg = "invalid memory range specified";
 689	return -EINVAL;
 690}
 691
 692/*
 693 * Reserve kdump memory by creating a memory hole in the mem_chunk array
 694 */
 695static void __init reserve_kdump_bootmem(unsigned long addr, unsigned long size,
 696					 int type)
 697{
 698	create_mem_hole(memory_chunk, addr, size, type);
 699}
 700
 701/*
 702 * When kdump is enabled, we have to ensure that no memory from
 703 * the area [0 - crashkernel memory size] and
 704 * [crashk_res.start - crashk_res.end] is set offline.
 705 */
 706static int kdump_mem_notifier(struct notifier_block *nb,
 707			      unsigned long action, void *data)
 708{
 709	struct memory_notify *arg = data;
 710
 
 
 711	if (arg->start_pfn < PFN_DOWN(resource_size(&crashk_res)))
 712		return NOTIFY_BAD;
 713	if (arg->start_pfn > PFN_DOWN(crashk_res.end))
 714		return NOTIFY_OK;
 715	if (arg->start_pfn + arg->nr_pages - 1 < PFN_DOWN(crashk_res.start))
 716		return NOTIFY_OK;
 717	return NOTIFY_BAD;
 718}
 719
 720static struct notifier_block kdump_mem_nb = {
 721	.notifier_call = kdump_mem_notifier,
 722};
 723
 724#endif
 725
 726/*
 727 * Make sure that oldmem, where the dump is stored, is protected
 728 */
 729static void reserve_oldmem(void)
 730{
 731#ifdef CONFIG_CRASH_DUMP
 732	if (!OLDMEM_BASE)
 733		return;
 734
 735	reserve_kdump_bootmem(OLDMEM_BASE, OLDMEM_SIZE, CHUNK_OLDMEM);
 736	reserve_kdump_bootmem(OLDMEM_SIZE, memory_end - OLDMEM_SIZE,
 737			      CHUNK_OLDMEM);
 738	if (OLDMEM_BASE + OLDMEM_SIZE == real_memory_size)
 739		saved_max_pfn = PFN_DOWN(OLDMEM_BASE) - 1;
 740	else
 741		saved_max_pfn = PFN_DOWN(real_memory_size) - 1;
 742#endif
 743}
 744
 745/*
 746 * Reserve memory for kdump kernel to be loaded with kexec
 747 */
 748static void __init reserve_crashkernel(void)
 749{
 750#ifdef CONFIG_CRASH_DUMP
 751	unsigned long long crash_base, crash_size;
 752	char *msg = NULL;
 753	int rc;
 754
 755	rc = parse_crashkernel(boot_command_line, memory_end, &crash_size,
 756			       &crash_base);
 757	if (rc || crash_size == 0)
 758		return;
 759	crash_base = ALIGN(crash_base, KEXEC_CRASH_MEM_ALIGN);
 760	crash_size = ALIGN(crash_size, KEXEC_CRASH_MEM_ALIGN);
 761	if (register_memory_notifier(&kdump_mem_nb))
 762		return;
 763	if (!crash_base)
 764		crash_base = find_crash_base(crash_size, &msg);
 765	if (!crash_base) {
 766		pr_info("crashkernel reservation failed: %s\n", msg);
 767		unregister_memory_notifier(&kdump_mem_nb);
 768		return;
 769	}
 770	if (verify_crash_base(crash_base, crash_size, &msg)) {
 771		pr_info("crashkernel reservation failed: %s\n", msg);
 772		unregister_memory_notifier(&kdump_mem_nb);
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 773		return;
 774	}
 
 
 
 
 775	if (!OLDMEM_BASE && MACHINE_IS_VM)
 776		diag10_range(PFN_DOWN(crash_base), PFN_DOWN(crash_size));
 777	crashk_res.start = crash_base;
 778	crashk_res.end = crash_base + crash_size - 1;
 779	insert_resource(&iomem_resource, &crashk_res);
 780	reserve_kdump_bootmem(crash_base, crash_size, CHUNK_CRASHK);
 781	pr_info("Reserving %lluMB of memory at %lluMB "
 782		"for crashkernel (System RAM: %luMB)\n",
 783		crash_size >> 20, crash_base >> 20, memory_end >> 20);
 
 784	os_info_crashkernel_add(crash_base, crash_size);
 785#endif
 786}
 787
 788static void __init setup_memory(void)
 
 
 
 789{
 790        unsigned long bootmap_size;
 791	unsigned long start_pfn, end_pfn;
 792	int i;
 
 
 
 
 
 793
 794	/*
 795	 * partially used pages are not usable - thus
 796	 * we are rounding upwards:
 797	 */
 798	start_pfn = PFN_UP(__pa(&_end));
 799	end_pfn = max_pfn = PFN_DOWN(memory_end);
 
 
 800
 801#ifdef CONFIG_BLK_DEV_INITRD
 802	/*
 803	 * Move the initrd in case the bitmap of the bootmem allocater
 804	 * would overwrite it.
 805	 */
 806
 807	if (INITRD_START && INITRD_SIZE) {
 808		unsigned long bmap_size;
 809		unsigned long start;
 
 810
 811		bmap_size = bootmem_bootmap_pages(end_pfn - start_pfn + 1);
 812		bmap_size = PFN_PHYS(bmap_size);
 
 813
 814		if (PFN_PHYS(start_pfn) + bmap_size > INITRD_START) {
 815			start = PFN_PHYS(start_pfn) + bmap_size + PAGE_SIZE;
 
 
 816
 817#ifdef CONFIG_CRASH_DUMP
 818			if (OLDMEM_BASE) {
 819				/* Move initrd behind kdump oldmem */
 820				if (start + INITRD_SIZE > OLDMEM_BASE &&
 821				    start < OLDMEM_BASE + OLDMEM_SIZE)
 822					start = OLDMEM_BASE + OLDMEM_SIZE;
 823			}
 824#endif
 825			if (start + INITRD_SIZE > memory_end) {
 826				pr_err("initrd extends beyond end of "
 827				       "memory (0x%08lx > 0x%08lx) "
 828				       "disabling initrd\n",
 829				       start + INITRD_SIZE, memory_end);
 830				INITRD_START = INITRD_SIZE = 0;
 831			} else {
 832				pr_info("Moving initrd (0x%08lx -> "
 833					"0x%08lx, size: %ld)\n",
 834					INITRD_START, start, INITRD_SIZE);
 835				memmove((void *) start, (void *) INITRD_START,
 836					INITRD_SIZE);
 837				INITRD_START = start;
 838			}
 839		}
 840	}
 841#endif
 842
 843	/*
 844	 * Initialize the boot-time allocator
 845	 */
 846	bootmap_size = init_bootmem(start_pfn, end_pfn);
 847
 848	/*
 849	 * Register RAM areas with the bootmem allocator.
 850	 */
 
 851
 852	for (i = 0; i < MEMORY_CHUNKS && memory_chunk[i].size > 0; i++) {
 853		unsigned long start_chunk, end_chunk, pfn;
 
 
 
 
 
 
 
 
 
 
 854
 855		if (memory_chunk[i].type != CHUNK_READ_WRITE &&
 856		    memory_chunk[i].type != CHUNK_CRASHK)
 857			continue;
 858		start_chunk = PFN_DOWN(memory_chunk[i].addr);
 859		end_chunk = start_chunk + PFN_DOWN(memory_chunk[i].size);
 860		end_chunk = min(end_chunk, end_pfn);
 861		if (start_chunk >= end_chunk)
 862			continue;
 863		memblock_add_node(PFN_PHYS(start_chunk),
 864				  PFN_PHYS(end_chunk - start_chunk), 0);
 865		pfn = max(start_chunk, start_pfn);
 866		for (; pfn < end_chunk; pfn++)
 867			page_set_storage_key(PFN_PHYS(pfn),
 868					     PAGE_DEFAULT_KEY, 0);
 869	}
 
 
 870
 871	psw_set_key(PAGE_DEFAULT_KEY);
 
 
 
 
 
 872
 873	free_bootmem_with_active_regions(0, max_pfn);
 
 
 
 
 
 
 
 
 
 874
 875	/*
 876	 * Reserve memory used for lowcore/command line/kernel image.
 877	 */
 878	reserve_bootmem(0, (unsigned long)_ehead, BOOTMEM_DEFAULT);
 879	reserve_bootmem((unsigned long)_stext,
 880			PFN_PHYS(start_pfn) - (unsigned long)_stext,
 881			BOOTMEM_DEFAULT);
 882	/*
 883	 * Reserve the bootmem bitmap itself as well. We do this in two
 884	 * steps (first step was init_bootmem()) because this catches
 885	 * the (very unlikely) case of us accidentally initializing the
 886	 * bootmem allocator with an invalid RAM area.
 887	 */
 888	reserve_bootmem(start_pfn << PAGE_SHIFT, bootmap_size,
 889			BOOTMEM_DEFAULT);
 890
 891#ifdef CONFIG_CRASH_DUMP
 892	if (crashk_res.start)
 893		reserve_bootmem(crashk_res.start,
 894				crashk_res.end - crashk_res.start + 1,
 895				BOOTMEM_DEFAULT);
 896	if (is_kdump_kernel())
 897		reserve_bootmem(elfcorehdr_addr - OLDMEM_BASE,
 898				PAGE_ALIGN(elfcorehdr_size), BOOTMEM_DEFAULT);
 899#endif
 900#ifdef CONFIG_BLK_DEV_INITRD
 901	if (INITRD_START && INITRD_SIZE) {
 902		if (INITRD_START + INITRD_SIZE <= memory_end) {
 903			reserve_bootmem(INITRD_START, INITRD_SIZE,
 904					BOOTMEM_DEFAULT);
 905			initrd_start = INITRD_START;
 906			initrd_end = initrd_start + INITRD_SIZE;
 907		} else {
 908			pr_err("initrd extends beyond end of "
 909			       "memory (0x%08lx > 0x%08lx) "
 910			       "disabling initrd\n",
 911			       initrd_start + INITRD_SIZE, memory_end);
 912			initrd_start = initrd_end = 0;
 913		}
 914	}
 915#endif
 916}
 917
 918/*
 919 * Setup hardware capabilities.
 920 */
 921static void __init setup_hwcaps(void)
 922{
 923	static const int stfl_bits[6] = { 0, 2, 7, 17, 19, 21 };
 924	struct cpuid cpu_id;
 925	int i;
 926
 927	/*
 928	 * The store facility list bits numbers as found in the principles
 929	 * of operation are numbered with bit 1UL<<31 as number 0 to
 930	 * bit 1UL<<0 as number 31.
 931	 *   Bit 0: instructions named N3, "backported" to esa-mode
 932	 *   Bit 2: z/Architecture mode is active
 933	 *   Bit 7: the store-facility-list-extended facility is installed
 934	 *   Bit 17: the message-security assist is installed
 935	 *   Bit 19: the long-displacement facility is installed
 936	 *   Bit 21: the extended-immediate facility is installed
 937	 *   Bit 22: extended-translation facility 3 is installed
 938	 *   Bit 30: extended-translation facility 3 enhancement facility
 939	 * These get translated to:
 940	 *   HWCAP_S390_ESAN3 bit 0, HWCAP_S390_ZARCH bit 1,
 941	 *   HWCAP_S390_STFLE bit 2, HWCAP_S390_MSA bit 3,
 942	 *   HWCAP_S390_LDISP bit 4, HWCAP_S390_EIMM bit 5 and
 943	 *   HWCAP_S390_ETF3EH bit 8 (22 && 30).
 944	 */
 945	for (i = 0; i < 6; i++)
 946		if (test_facility(stfl_bits[i]))
 947			elf_hwcap |= 1UL << i;
 948
 949	if (test_facility(22) && test_facility(30))
 950		elf_hwcap |= HWCAP_S390_ETF3EH;
 951
 952	/*
 953	 * Check for additional facilities with store-facility-list-extended.
 954	 * stfle stores doublewords (8 byte) with bit 1ULL<<63 as bit 0
 955	 * and 1ULL<<0 as bit 63. Bits 0-31 contain the same information
 956	 * as stored by stfl, bits 32-xxx contain additional facilities.
 957	 * How many facility words are stored depends on the number of
 958	 * doublewords passed to the instruction. The additional facilities
 959	 * are:
 960	 *   Bit 42: decimal floating point facility is installed
 961	 *   Bit 44: perform floating point operation facility is installed
 962	 * translated to:
 963	 *   HWCAP_S390_DFP bit 6 (42 && 44).
 964	 */
 965	if ((elf_hwcap & (1UL << 2)) && test_facility(42) && test_facility(44))
 966		elf_hwcap |= HWCAP_S390_DFP;
 967
 968	/*
 969	 * Huge page support HWCAP_S390_HPAGE is bit 7.
 970	 */
 971	if (MACHINE_HAS_HPAGE)
 972		elf_hwcap |= HWCAP_S390_HPAGE;
 973
 974	/*
 975	 * 64-bit register support for 31-bit processes
 976	 * HWCAP_S390_HIGH_GPRS is bit 9.
 977	 */
 978	elf_hwcap |= HWCAP_S390_HIGH_GPRS;
 979
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 980	get_cpu_id(&cpu_id);
 
 981	switch (cpu_id.machine) {
 982	case 0x9672:
 983#if !defined(CONFIG_64BIT)
 984	default:	/* Use "g5" as default for 31 bit kernels. */
 985#endif
 986		strcpy(elf_platform, "g5");
 987		break;
 988	case 0x2064:
 989	case 0x2066:
 990#if defined(CONFIG_64BIT)
 991	default:	/* Use "z900" as default for 64 bit kernels. */
 992#endif
 993		strcpy(elf_platform, "z900");
 994		break;
 995	case 0x2084:
 996	case 0x2086:
 997		strcpy(elf_platform, "z990");
 998		break;
 999	case 0x2094:
1000	case 0x2096:
1001		strcpy(elf_platform, "z9-109");
1002		break;
1003	case 0x2097:
1004	case 0x2098:
1005		strcpy(elf_platform, "z10");
1006		break;
1007	case 0x2817:
1008	case 0x2818:
1009		strcpy(elf_platform, "z196");
1010		break;
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1011	}
1012}
1013
1014/*
1015 * Setup function called from init/main.c just after the banner
1016 * was printed.
1017 */
1018
1019void __init setup_arch(char **cmdline_p)
1020{
1021        /*
1022         * print what head.S has found out about the machine
1023         */
1024#ifndef CONFIG_64BIT
1025	if (MACHINE_IS_VM)
1026		pr_info("Linux is running as a z/VM "
1027			"guest operating system in 31-bit mode\n");
1028	else if (MACHINE_IS_LPAR)
1029		pr_info("Linux is running natively in 31-bit mode\n");
1030	if (MACHINE_HAS_IEEE)
1031		pr_info("The hardware system has IEEE compatible "
1032			"floating point units\n");
1033	else
1034		pr_info("The hardware system has no IEEE compatible "
1035			"floating point units\n");
1036#else /* CONFIG_64BIT */
1037	if (MACHINE_IS_VM)
1038		pr_info("Linux is running as a z/VM "
1039			"guest operating system in 64-bit mode\n");
1040	else if (MACHINE_IS_KVM)
1041		pr_info("Linux is running under KVM in 64-bit mode\n");
1042	else if (MACHINE_IS_LPAR)
1043		pr_info("Linux is running natively in 64-bit mode\n");
1044#endif /* CONFIG_64BIT */
 
 
 
1045
1046	/* Have one command line that is parsed and saved in /proc/cmdline */
1047	/* boot_command_line has been already set up in early.c */
1048	*cmdline_p = boot_command_line;
1049
1050        ROOT_DEV = Root_RAM0;
1051
1052	init_mm.start_code = PAGE_OFFSET;
1053	init_mm.end_code = (unsigned long) &_etext;
1054	init_mm.end_data = (unsigned long) &_edata;
1055	init_mm.brk = (unsigned long) &_end;
1056
1057	if (MACHINE_HAS_MVCOS)
1058		memcpy(&uaccess, &uaccess_mvcos, sizeof(uaccess));
1059	else
1060		memcpy(&uaccess, &uaccess_std, sizeof(uaccess));
1061
 
1062	parse_early_param();
 
 
 
 
1063
1064	os_info_init();
1065	setup_ipl();
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1066	setup_memory_end();
1067	setup_addressing_mode();
1068	reserve_oldmem();
1069	reserve_crashkernel();
1070	setup_memory();
1071	setup_resources();
1072	setup_vmcoreinfo();
1073	setup_lowcore();
1074
1075        cpu_init();
1076	s390_init_cpu_topology();
1077
 
 
 
1078	/*
1079	 * Setup capabilities (ELF_HWCAP & ELF_PLATFORM).
 
1080	 */
1081	setup_hwcaps();
 
 
 
 
 
 
 
 
 
 
1082
1083	/*
1084	 * Create kernel page tables and switch to virtual addressing.
1085	 */
1086        paging_init();
1087
 
 
 
 
 
 
1088        /* Setup default console */
1089	conmode_default();
1090	set_preferred_console();
1091
1092	/* Setup zfcpdump support */
1093	setup_zfcpdump(console_devno);
 
 
 
 
 
 
 
1094}
v5.14.15
   1// SPDX-License-Identifier: GPL-2.0
   2/*
 
 
   3 *  S390 version
   4 *    Copyright IBM Corp. 1999, 2012
   5 *    Author(s): Hartmut Penner (hp@de.ibm.com),
   6 *               Martin Schwidefsky (schwidefsky@de.ibm.com)
   7 *
   8 *  Derived from "arch/i386/kernel/setup.c"
   9 *    Copyright (C) 1995, Linus Torvalds
  10 */
  11
  12/*
  13 * This file handles the architecture-dependent parts of initialization
  14 */
  15
  16#define KMSG_COMPONENT "setup"
  17#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
  18
  19#include <linux/errno.h>
  20#include <linux/export.h>
  21#include <linux/sched.h>
  22#include <linux/sched/task.h>
  23#include <linux/cpu.h>
  24#include <linux/kernel.h>
  25#include <linux/memblock.h>
  26#include <linux/mm.h>
  27#include <linux/stddef.h>
  28#include <linux/unistd.h>
  29#include <linux/ptrace.h>
  30#include <linux/random.h>
  31#include <linux/user.h>
  32#include <linux/tty.h>
  33#include <linux/ioport.h>
  34#include <linux/delay.h>
  35#include <linux/init.h>
  36#include <linux/initrd.h>
 
  37#include <linux/root_dev.h>
  38#include <linux/console.h>
  39#include <linux/kernel_stat.h>
  40#include <linux/dma-map-ops.h>
  41#include <linux/device.h>
  42#include <linux/notifier.h>
  43#include <linux/pfn.h>
  44#include <linux/ctype.h>
  45#include <linux/reboot.h>
  46#include <linux/topology.h>
 
  47#include <linux/kexec.h>
  48#include <linux/crash_dump.h>
  49#include <linux/memory.h>
  50#include <linux/compat.h>
  51#include <linux/start_kernel.h>
  52#include <linux/hugetlb.h>
  53#include <linux/kmemleak.h>
  54
  55#include <asm/boot_data.h>
  56#include <asm/ipl.h>
 
  57#include <asm/facility.h>
  58#include <asm/smp.h>
  59#include <asm/mmu_context.h>
  60#include <asm/cpcmd.h>
  61#include <asm/lowcore.h>
  62#include <asm/nmi.h>
  63#include <asm/irq.h>
  64#include <asm/page.h>
  65#include <asm/ptrace.h>
  66#include <asm/sections.h>
  67#include <asm/ebcdic.h>
 
  68#include <asm/diag.h>
  69#include <asm/os_info.h>
  70#include <asm/sclp.h>
  71#include <asm/stacktrace.h>
  72#include <asm/sysinfo.h>
  73#include <asm/numa.h>
  74#include <asm/alternative.h>
  75#include <asm/nospec-branch.h>
  76#include <asm/mem_detect.h>
  77#include <asm/uv.h>
  78#include <asm/asm-offsets.h>
  79#include "entry.h"
  80
 
 
 
 
 
 
 
 
 
 
 
 
  81/*
  82 * Machine setup..
  83 */
  84unsigned int console_mode = 0;
  85EXPORT_SYMBOL(console_mode);
  86
  87unsigned int console_devno = -1;
  88EXPORT_SYMBOL(console_devno);
  89
  90unsigned int console_irq = -1;
  91EXPORT_SYMBOL(console_irq);
  92
  93unsigned long elf_hwcap __read_mostly = 0;
  94char elf_platform[ELF_PLATFORM_SIZE];
  95
  96unsigned long int_hwcap = 0;
  97
  98int __bootdata(noexec_disabled);
  99unsigned long __bootdata(ident_map_size);
 100struct mem_detect_info __bootdata(mem_detect);
 101
 102struct exception_table_entry *__bootdata_preserved(__start_dma_ex_table);
 103struct exception_table_entry *__bootdata_preserved(__stop_dma_ex_table);
 104unsigned long __bootdata_preserved(__stext_dma);
 105unsigned long __bootdata_preserved(__etext_dma);
 106unsigned long __bootdata_preserved(__sdma);
 107unsigned long __bootdata_preserved(__edma);
 108unsigned long __bootdata_preserved(__kaslr_offset);
 109unsigned int __bootdata_preserved(zlib_dfltcc_support);
 110EXPORT_SYMBOL(zlib_dfltcc_support);
 111u64 __bootdata_preserved(stfle_fac_list[16]);
 112EXPORT_SYMBOL(stfle_fac_list);
 113u64 __bootdata_preserved(alt_stfle_fac_list[16]);
 114
 115unsigned long VMALLOC_START;
 116EXPORT_SYMBOL(VMALLOC_START);
 117
 118unsigned long VMALLOC_END;
 119EXPORT_SYMBOL(VMALLOC_END);
 120
 121struct page *vmemmap;
 122EXPORT_SYMBOL(vmemmap);
 123unsigned long vmemmap_size;
 124
 125unsigned long MODULES_VADDR;
 126unsigned long MODULES_END;
 127
 128/* An array with a pointer to the lowcore of every CPU. */
 129struct lowcore *lowcore_ptr[NR_CPUS];
 130EXPORT_SYMBOL(lowcore_ptr);
 131
 132/*
 133 * The Write Back bit position in the physaddr is given by the SLPC PCI.
 134 * Leaving the mask zero always uses write through which is safe
 135 */
 136unsigned long mio_wb_bit_mask __ro_after_init;
 137
 138/*
 139 * This is set up by the setup-routine at boot-time
 140 * for S390 need to find out, what we have to setup
 141 * using address 0x10400 ...
 142 */
 143
 144#include <asm/setup.h>
 145
 146/*
 147 * condev= and conmode= setup parameter.
 148 */
 149
 150static int __init condev_setup(char *str)
 151{
 152	int vdev;
 153
 154	vdev = simple_strtoul(str, &str, 0);
 155	if (vdev >= 0 && vdev < 65536) {
 156		console_devno = vdev;
 157		console_irq = -1;
 158	}
 159	return 1;
 160}
 161
 162__setup("condev=", condev_setup);
 163
 164static void __init set_preferred_console(void)
 165{
 166	if (CONSOLE_IS_3215 || CONSOLE_IS_SCLP)
 
 
 167		add_preferred_console("ttyS", 0, NULL);
 168	else if (CONSOLE_IS_3270)
 169		add_preferred_console("tty3270", 0, NULL);
 170	else if (CONSOLE_IS_VT220)
 171		add_preferred_console("ttysclp", 0, NULL);
 172	else if (CONSOLE_IS_HVC)
 173		add_preferred_console("hvc", 0, NULL);
 174}
 175
 176static int __init conmode_setup(char *str)
 177{
 178#if defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
 179	if (!strcmp(str, "hwc") || !strcmp(str, "sclp"))
 180                SET_CONSOLE_SCLP;
 181#endif
 182#if defined(CONFIG_TN3215_CONSOLE)
 183	if (!strcmp(str, "3215"))
 184		SET_CONSOLE_3215;
 185#endif
 186#if defined(CONFIG_TN3270_CONSOLE)
 187	if (!strcmp(str, "3270"))
 188		SET_CONSOLE_3270;
 189#endif
 190	set_preferred_console();
 191        return 1;
 192}
 193
 194__setup("conmode=", conmode_setup);
 195
 196static void __init conmode_default(void)
 197{
 198	char query_buffer[1024];
 199	char *ptr;
 200
 201        if (MACHINE_IS_VM) {
 202		cpcmd("QUERY CONSOLE", query_buffer, 1024, NULL);
 203		console_devno = simple_strtoul(query_buffer + 5, NULL, 16);
 204		ptr = strstr(query_buffer, "SUBCHANNEL =");
 205		console_irq = simple_strtoul(ptr + 13, NULL, 16);
 206		cpcmd("QUERY TERM", query_buffer, 1024, NULL);
 207		ptr = strstr(query_buffer, "CONMODE");
 208		/*
 209		 * Set the conmode to 3215 so that the device recognition 
 210		 * will set the cu_type of the console to 3215. If the
 211		 * conmode is 3270 and we don't set it back then both
 212		 * 3215 and the 3270 driver will try to access the console
 213		 * device (3215 as console and 3270 as normal tty).
 214		 */
 215		cpcmd("TERM CONMODE 3215", NULL, 0, NULL);
 216		if (ptr == NULL) {
 217#if defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
 218			SET_CONSOLE_SCLP;
 219#endif
 220			return;
 221		}
 222		if (str_has_prefix(ptr + 8, "3270")) {
 223#if defined(CONFIG_TN3270_CONSOLE)
 224			SET_CONSOLE_3270;
 225#elif defined(CONFIG_TN3215_CONSOLE)
 226			SET_CONSOLE_3215;
 227#elif defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
 228			SET_CONSOLE_SCLP;
 229#endif
 230		} else if (str_has_prefix(ptr + 8, "3215")) {
 231#if defined(CONFIG_TN3215_CONSOLE)
 232			SET_CONSOLE_3215;
 233#elif defined(CONFIG_TN3270_CONSOLE)
 234			SET_CONSOLE_3270;
 235#elif defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
 236			SET_CONSOLE_SCLP;
 237#endif
 238		}
 239	} else if (MACHINE_IS_KVM) {
 240		if (sclp.has_vt220 && IS_ENABLED(CONFIG_SCLP_VT220_CONSOLE))
 241			SET_CONSOLE_VT220;
 242		else if (sclp.has_linemode && IS_ENABLED(CONFIG_SCLP_CONSOLE))
 243			SET_CONSOLE_SCLP;
 244		else
 245			SET_CONSOLE_HVC;
 246	} else {
 247#if defined(CONFIG_SCLP_CONSOLE) || defined(CONFIG_SCLP_VT220_CONSOLE)
 248		SET_CONSOLE_SCLP;
 249#endif
 250	}
 251}
 252
 253#ifdef CONFIG_CRASH_DUMP
 254static void __init setup_zfcpdump(void)
 255{
 256	if (!is_ipl_type_dump())
 
 
 257		return;
 258	if (OLDMEM_BASE)
 259		return;
 260	strcat(boot_command_line, " cio_ignore=all,!ipldev,!condev");
 
 
 
 
 
 
 261	console_loglevel = 2;
 262}
 263#else
 264static inline void setup_zfcpdump(void) {}
 265#endif /* CONFIG_CRASH_DUMP */
 266
 267 /*
 268 * Reboot, halt and power_off stubs. They just call _machine_restart,
 269 * _machine_halt or _machine_power_off. 
 270 */
 271
 272void machine_restart(char *command)
 273{
 274	if ((!in_interrupt() && !in_atomic()) || oops_in_progress)
 275		/*
 276		 * Only unblank the console if we are called in enabled
 277		 * context or a bust_spinlocks cleared the way for us.
 278		 */
 279		console_unblank();
 280	_machine_restart(command);
 281}
 282
 283void machine_halt(void)
 284{
 285	if (!in_interrupt() || oops_in_progress)
 286		/*
 287		 * Only unblank the console if we are called in enabled
 288		 * context or a bust_spinlocks cleared the way for us.
 289		 */
 290		console_unblank();
 291	_machine_halt();
 292}
 293
 294void machine_power_off(void)
 295{
 296	if (!in_interrupt() || oops_in_progress)
 297		/*
 298		 * Only unblank the console if we are called in enabled
 299		 * context or a bust_spinlocks cleared the way for us.
 300		 */
 301		console_unblank();
 302	_machine_power_off();
 303}
 304
 305/*
 306 * Dummy power off function.
 307 */
 308void (*pm_power_off)(void) = machine_power_off;
 309EXPORT_SYMBOL_GPL(pm_power_off);
 310
 311void *restart_stack;
 
 
 
 
 
 
 312
 313unsigned long stack_alloc(void)
 314{
 315#ifdef CONFIG_VMAP_STACK
 316	void *ret;
 
 
 
 
 317
 318	ret = __vmalloc_node(THREAD_SIZE, THREAD_SIZE, THREADINFO_GFP,
 319			     NUMA_NO_NODE, __builtin_return_address(0));
 320	kmemleak_not_leak(ret);
 321	return (unsigned long)ret;
 322#else
 323	return __get_free_pages(GFP_KERNEL, THREAD_SIZE_ORDER);
 
 
 
 
 324#endif
 
 
 
 
 
 
 
 
 325}
 326
 327void stack_free(unsigned long stack)
 
 
 
 328{
 329#ifdef CONFIG_VMAP_STACK
 330	vfree((void *) stack);
 331#else
 332	free_pages(stack, THREAD_SIZE_ORDER);
 333#endif
 334}
 
 335
 336int __init arch_early_irq_init(void)
 337{
 338	unsigned long stack;
 339
 340	stack = __get_free_pages(GFP_KERNEL, THREAD_SIZE_ORDER);
 341	if (!stack)
 342		panic("Couldn't allocate async stack");
 343	S390_lowcore.async_stack = stack + STACK_INIT_OFFSET;
 344	return 0;
 345}
 
 346
 347void __init arch_call_rest_init(void)
 348{
 349	unsigned long stack;
 
 
 
 
 
 
 
 
 350
 351	stack = stack_alloc();
 352	if (!stack)
 353		panic("Couldn't allocate kernel stack");
 354	current->stack = (void *) stack;
 355#ifdef CONFIG_VMAP_STACK
 356	current->stack_vm_area = (void *) stack;
 357#endif
 358	set_task_stack_end_magic(current);
 359	stack += STACK_INIT_OFFSET;
 360	S390_lowcore.kernel_stack = stack;
 361	call_on_stack_noreturn(rest_init, stack);
 362}
 363
 364static void __init setup_lowcore_dat_off(void)
 365{
 366	unsigned long int_psw_mask = PSW_KERNEL_BITS;
 367	unsigned long mcck_stack;
 368	struct lowcore *lc;
 369
 370	if (IS_ENABLED(CONFIG_KASAN))
 371		int_psw_mask |= PSW_MASK_DAT;
 372
 373	/*
 374	 * Setup lowcore for boot cpu
 375	 */
 376	BUILD_BUG_ON(sizeof(struct lowcore) != LC_PAGES * PAGE_SIZE);
 377	lc = memblock_alloc_low(sizeof(*lc), sizeof(*lc));
 378	if (!lc)
 379		panic("%s: Failed to allocate %zu bytes align=%zx\n",
 380		      __func__, sizeof(*lc), sizeof(*lc));
 381
 382	lc->restart_psw.mask = PSW_KERNEL_BITS;
 383	lc->restart_psw.addr = (unsigned long) restart_int_handler;
 384	lc->external_new_psw.mask = int_psw_mask | PSW_MASK_MCHECK;
 385	lc->external_new_psw.addr = (unsigned long) ext_int_handler;
 386	lc->svc_new_psw.mask = int_psw_mask | PSW_MASK_MCHECK;
 387	lc->svc_new_psw.addr = (unsigned long) system_call;
 388	lc->program_new_psw.mask = int_psw_mask | PSW_MASK_MCHECK;
 389	lc->program_new_psw.addr = (unsigned long) pgm_check_handler;
 390	lc->mcck_new_psw.mask = PSW_KERNEL_BITS;
 391	lc->mcck_new_psw.addr = (unsigned long) mcck_int_handler;
 392	lc->io_new_psw.mask = int_psw_mask | PSW_MASK_MCHECK;
 393	lc->io_new_psw.addr = (unsigned long) io_int_handler;
 394	lc->clock_comparator = clock_comparator_max;
 395	lc->nodat_stack = ((unsigned long) &init_thread_union)
 396		+ THREAD_SIZE - STACK_FRAME_OVERHEAD - sizeof(struct pt_regs);
 397	lc->current_task = (unsigned long)&init_task;
 398	lc->lpp = LPP_MAGIC;
 
 
 
 
 
 
 
 399	lc->machine_flags = S390_lowcore.machine_flags;
 400	lc->preempt_count = S390_lowcore.preempt_count;
 401	nmi_alloc_boot_cpu(lc);
 402	lc->sys_enter_timer = S390_lowcore.sys_enter_timer;
 
 
 
 
 
 
 
 
 
 
 
 
 403	lc->exit_timer = S390_lowcore.exit_timer;
 404	lc->user_timer = S390_lowcore.user_timer;
 405	lc->system_timer = S390_lowcore.system_timer;
 406	lc->steal_timer = S390_lowcore.steal_timer;
 407	lc->last_update_timer = S390_lowcore.last_update_timer;
 408	lc->last_update_clock = S390_lowcore.last_update_clock;
 
 409
 410	/*
 411	 * Allocate the global restart stack which is the same for
 412	 * all CPUs in cast *one* of them does a PSW restart.
 413	 */
 414	restart_stack = memblock_alloc(THREAD_SIZE, THREAD_SIZE);
 415	if (!restart_stack)
 416		panic("%s: Failed to allocate %lu bytes align=0x%lx\n",
 417		      __func__, THREAD_SIZE, THREAD_SIZE);
 418	restart_stack += STACK_INIT_OFFSET;
 419
 420	/*
 421	 * Set up PSW restart to call ipl.c:do_restart(). Copy the relevant
 422	 * restart data to the absolute zero lowcore. This is necessary if
 423	 * PSW restart is done on an offline CPU that has lowcore zero.
 424	 */
 425	lc->restart_stack = (unsigned long) restart_stack;
 426	lc->restart_fn = (unsigned long) do_restart;
 427	lc->restart_data = 0;
 428	lc->restart_source = -1U;
 429
 430	mcck_stack = (unsigned long)memblock_alloc(THREAD_SIZE, THREAD_SIZE);
 431	if (!mcck_stack)
 432		panic("%s: Failed to allocate %lu bytes align=0x%lx\n",
 433		      __func__, THREAD_SIZE, THREAD_SIZE);
 434	lc->mcck_stack = mcck_stack + STACK_INIT_OFFSET;
 435
 436	/* Setup absolute zero lowcore */
 437	mem_assign_absolute(S390_lowcore.restart_stack, lc->restart_stack);
 438	mem_assign_absolute(S390_lowcore.restart_fn, lc->restart_fn);
 439	mem_assign_absolute(S390_lowcore.restart_data, lc->restart_data);
 440	mem_assign_absolute(S390_lowcore.restart_source, lc->restart_source);
 441	mem_assign_absolute(S390_lowcore.restart_psw, lc->restart_psw);
 442
 443	lc->spinlock_lockval = arch_spin_lockval(0);
 444	lc->spinlock_index = 0;
 445	arch_spin_lock_setup(0);
 446	lc->br_r1_trampoline = 0x07f1;	/* br %r1 */
 447	lc->return_lpswe = gen_lpswe(__LC_RETURN_PSW);
 448	lc->return_mcck_lpswe = gen_lpswe(__LC_RETURN_MCCK_PSW);
 449	lc->preempt_count = PREEMPT_DISABLED;
 450
 451	set_prefix((u32)(unsigned long) lc);
 452	lowcore_ptr[0] = lc;
 453}
 454
 455static void __init setup_lowcore_dat_on(void)
 456{
 457	struct lowcore *lc = lowcore_ptr[0];
 458
 459	__ctl_clear_bit(0, 28);
 460	S390_lowcore.external_new_psw.mask |= PSW_MASK_DAT;
 461	S390_lowcore.svc_new_psw.mask |= PSW_MASK_DAT;
 462	S390_lowcore.program_new_psw.mask |= PSW_MASK_DAT;
 463	S390_lowcore.io_new_psw.mask |= PSW_MASK_DAT;
 464	__ctl_store(S390_lowcore.cregs_save_area, 0, 15);
 465	__ctl_set_bit(0, 28);
 466	mem_assign_absolute(S390_lowcore.restart_flags, RESTART_FLAG_CTLREGS);
 467	mem_assign_absolute(S390_lowcore.program_new_psw, lc->program_new_psw);
 468	memcpy_absolute(&S390_lowcore.cregs_save_area, lc->cregs_save_area,
 469			sizeof(S390_lowcore.cregs_save_area));
 470}
 471
 472static struct resource code_resource = {
 473	.name  = "Kernel code",
 474	.flags = IORESOURCE_BUSY | IORESOURCE_SYSTEM_RAM,
 475};
 476
 477static struct resource data_resource = {
 478	.name = "Kernel data",
 479	.flags = IORESOURCE_BUSY | IORESOURCE_SYSTEM_RAM,
 480};
 481
 482static struct resource bss_resource = {
 483	.name = "Kernel bss",
 484	.flags = IORESOURCE_BUSY | IORESOURCE_SYSTEM_RAM,
 485};
 486
 487static struct resource __initdata *standard_resources[] = {
 488	&code_resource,
 489	&data_resource,
 490	&bss_resource,
 491};
 492
 493static void __init setup_resources(void)
 494{
 495	struct resource *res, *std_res, *sub_res;
 496	phys_addr_t start, end;
 497	int j;
 498	u64 i;
 499
 500	code_resource.start = (unsigned long) _text;
 501	code_resource.end = (unsigned long) _etext - 1;
 502	data_resource.start = (unsigned long) _etext;
 503	data_resource.end = (unsigned long) _edata - 1;
 504	bss_resource.start = (unsigned long) __bss_start;
 505	bss_resource.end = (unsigned long) __bss_stop - 1;
 506
 507	for_each_mem_range(i, &start, &end) {
 508		res = memblock_alloc(sizeof(*res), 8);
 509		if (!res)
 510			panic("%s: Failed to allocate %zu bytes align=0x%x\n",
 511			      __func__, sizeof(*res), 8);
 512		res->flags = IORESOURCE_BUSY | IORESOURCE_SYSTEM_RAM;
 513
 514		res->name = "System RAM";
 515		res->start = start;
 516		/*
 517		 * In memblock, end points to the first byte after the
 518		 * range while in resourses, end points to the last byte in
 519		 * the range.
 520		 */
 521		res->end = end - 1;
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 522		request_resource(&iomem_resource, res);
 523
 524		for (j = 0; j < ARRAY_SIZE(standard_resources); j++) {
 525			std_res = standard_resources[j];
 526			if (std_res->start < res->start ||
 527			    std_res->start > res->end)
 528				continue;
 529			if (std_res->end > res->end) {
 530				sub_res = memblock_alloc(sizeof(*sub_res), 8);
 531				if (!sub_res)
 532					panic("%s: Failed to allocate %zu bytes align=0x%x\n",
 533					      __func__, sizeof(*sub_res), 8);
 534				*sub_res = *std_res;
 535				sub_res->end = res->end;
 536				std_res->start = res->end + 1;
 537				request_resource(res, sub_res);
 538			} else {
 539				request_resource(res, std_res);
 540			}
 541		}
 542	}
 543#ifdef CONFIG_CRASH_DUMP
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 544	/*
 545	 * Re-add removed crash kernel memory as reserved memory. This makes
 546	 * sure it will be mapped with the identity mapping and struct pages
 547	 * will be created, so it can be resized later on.
 548	 * However add it later since the crash kernel resource should not be
 549	 * part of the System RAM resource.
 550	 */
 551	if (crashk_res.end) {
 552		memblock_add_node(crashk_res.start, resource_size(&crashk_res), 0);
 553		memblock_reserve(crashk_res.start, resource_size(&crashk_res));
 554		insert_resource(&iomem_resource, &crashk_res);
 
 
 
 
 
 
 
 
 
 
 
 
 
 555	}
 
 
 
 
 
 
 
 
 
 
 
 
 
 556#endif
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 557}
 558
 559static void __init setup_memory_end(void)
 560{
 561	memblock_remove(ident_map_size, ULONG_MAX);
 562	max_pfn = max_low_pfn = PFN_DOWN(ident_map_size);
 563	pr_notice("The maximum memory size is %luMB\n", ident_map_size >> 20);
 
 
 564}
 565
 566#ifdef CONFIG_CRASH_DUMP
 567
 568/*
 569 * When kdump is enabled, we have to ensure that no memory from the area
 570 * [0 - crashkernel memory size] is set offline - it will be exchanged with
 571 * the crashkernel memory region when kdump is triggered. The crashkernel
 572 * memory region can never get offlined (pages are unmovable).
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 573 */
 574static int kdump_mem_notifier(struct notifier_block *nb,
 575			      unsigned long action, void *data)
 576{
 577	struct memory_notify *arg = data;
 578
 579	if (action != MEM_GOING_OFFLINE)
 580		return NOTIFY_OK;
 581	if (arg->start_pfn < PFN_DOWN(resource_size(&crashk_res)))
 582		return NOTIFY_BAD;
 583	return NOTIFY_OK;
 
 
 
 
 584}
 585
 586static struct notifier_block kdump_mem_nb = {
 587	.notifier_call = kdump_mem_notifier,
 588};
 589
 590#endif
 591
 592/*
 593 * Make sure that the area above identity mapping is protected
 594 */
 595static void __init reserve_above_ident_map(void)
 596{
 597	memblock_reserve(ident_map_size, ULONG_MAX);
 
 
 
 
 
 
 
 
 
 
 
 598}
 599
 600/*
 601 * Reserve memory for kdump kernel to be loaded with kexec
 602 */
 603static void __init reserve_crashkernel(void)
 604{
 605#ifdef CONFIG_CRASH_DUMP
 606	unsigned long long crash_base, crash_size;
 607	phys_addr_t low, high;
 608	int rc;
 609
 610	rc = parse_crashkernel(boot_command_line, ident_map_size, &crash_size,
 611			       &crash_base);
 612
 
 613	crash_base = ALIGN(crash_base, KEXEC_CRASH_MEM_ALIGN);
 614	crash_size = ALIGN(crash_size, KEXEC_CRASH_MEM_ALIGN);
 615	if (rc || crash_size == 0)
 616		return;
 617
 618	if (memblock.memory.regions[0].size < crash_size) {
 619		pr_info("crashkernel reservation failed: %s\n",
 620			"first memory chunk must be at least crashkernel size");
 
 621		return;
 622	}
 623
 624	low = crash_base ?: OLDMEM_BASE;
 625	high = low + crash_size;
 626	if (low >= OLDMEM_BASE && high <= OLDMEM_BASE + OLDMEM_SIZE) {
 627		/* The crashkernel fits into OLDMEM, reuse OLDMEM */
 628		crash_base = low;
 629	} else {
 630		/* Find suitable area in free memory */
 631		low = max_t(unsigned long, crash_size, sclp.hsa_size);
 632		high = crash_base ? crash_base + crash_size : ULONG_MAX;
 633
 634		if (crash_base && crash_base < low) {
 635			pr_info("crashkernel reservation failed: %s\n",
 636				"crash_base too low");
 637			return;
 638		}
 639		low = crash_base ?: low;
 640		crash_base = memblock_find_in_range(low, high, crash_size,
 641						    KEXEC_CRASH_MEM_ALIGN);
 642	}
 643
 644	if (!crash_base) {
 645		pr_info("crashkernel reservation failed: %s\n",
 646			"no suitable area found");
 647		return;
 648	}
 649
 650	if (register_memory_notifier(&kdump_mem_nb))
 651		return;
 652
 653	if (!OLDMEM_BASE && MACHINE_IS_VM)
 654		diag10_range(PFN_DOWN(crash_base), PFN_DOWN(crash_size));
 655	crashk_res.start = crash_base;
 656	crashk_res.end = crash_base + crash_size - 1;
 657	memblock_remove(crash_base, crash_size);
 
 658	pr_info("Reserving %lluMB of memory at %lluMB "
 659		"for crashkernel (System RAM: %luMB)\n",
 660		crash_size >> 20, crash_base >> 20,
 661		(unsigned long)memblock.memory.total_size >> 20);
 662	os_info_crashkernel_add(crash_base, crash_size);
 663#endif
 664}
 665
 666/*
 667 * Reserve the initrd from being used by memblock
 668 */
 669static void __init reserve_initrd(void)
 670{
 671#ifdef CONFIG_BLK_DEV_INITRD
 672	if (!INITRD_START || !INITRD_SIZE)
 673		return;
 674	initrd_start = INITRD_START;
 675	initrd_end = initrd_start + INITRD_SIZE;
 676	memblock_reserve(INITRD_START, INITRD_SIZE);
 677#endif
 678}
 679
 680/*
 681 * Reserve the memory area used to pass the certificate lists
 682 */
 683static void __init reserve_certificate_list(void)
 684{
 685	if (ipl_cert_list_addr)
 686		memblock_reserve(ipl_cert_list_addr, ipl_cert_list_size);
 687}
 688
 689static void __init reserve_mem_detect_info(void)
 690{
 691	unsigned long start, size;
 
 
 692
 693	get_mem_detect_reserved(&start, &size);
 694	if (size)
 695		memblock_reserve(start, size);
 696}
 697
 698static void __init free_mem_detect_info(void)
 699{
 700	unsigned long start, size;
 701
 702	get_mem_detect_reserved(&start, &size);
 703	if (size)
 704		memblock_free(start, size);
 705}
 706
 707static const char * __init get_mem_info_source(void)
 708{
 709	switch (mem_detect.info_source) {
 710	case MEM_DETECT_SCLP_STOR_INFO:
 711		return "sclp storage info";
 712	case MEM_DETECT_DIAG260:
 713		return "diag260";
 714	case MEM_DETECT_SCLP_READ_INFO:
 715		return "sclp read info";
 716	case MEM_DETECT_BIN_SEARCH:
 717		return "binary search";
 
 
 
 
 
 
 
 
 
 
 
 
 718	}
 719	return "none";
 720}
 
 
 
 
 721
 722static void __init memblock_add_mem_detect_info(void)
 723{
 724	unsigned long start, end;
 725	int i;
 726
 727	pr_debug("physmem info source: %s (%hhd)\n",
 728		 get_mem_info_source(), mem_detect.info_source);
 729	/* keep memblock lists close to the kernel */
 730	memblock_set_bottom_up(true);
 731	for_each_mem_detect_block(i, &start, &end) {
 732		memblock_add(start, end - start);
 733		memblock_physmem_add(start, end - start);
 734	}
 735	memblock_set_bottom_up(false);
 736	memblock_set_node(0, ULONG_MAX, &memblock.memory, 0);
 737	memblock_dump_all();
 738}
 739
 740/*
 741 * Check for initrd being in usable memory
 742 */
 743static void __init check_initrd(void)
 744{
 745#ifdef CONFIG_BLK_DEV_INITRD
 746	if (INITRD_START && INITRD_SIZE &&
 747	    !memblock_is_region_memory(INITRD_START, INITRD_SIZE)) {
 748		pr_err("The initial RAM disk does not fit into the memory\n");
 749		memblock_free(INITRD_START, INITRD_SIZE);
 750		initrd_start = initrd_end = 0;
 
 
 
 751	}
 752#endif
 753}
 754
 755/*
 756 * Reserve memory used for lowcore/command line/kernel image.
 757 */
 758static void __init reserve_kernel(void)
 759{
 760	unsigned long start_pfn = PFN_UP(__pa(_end));
 761
 762	memblock_reserve(0, HEAD_END);
 763	memblock_reserve((unsigned long)_stext, PFN_PHYS(start_pfn)
 764			 - (unsigned long)_stext);
 765	memblock_reserve(__sdma, __edma - __sdma);
 766}
 767
 768static void __init setup_memory(void)
 769{
 770	phys_addr_t start, end;
 771	u64 i;
 772
 773	/*
 774	 * Init storage key for present memory
 
 
 
 
 
 
 
 
 
 
 775	 */
 776	for_each_mem_range(i, &start, &end)
 777		storage_key_init_range(start, end);
 778
 779	psw_set_key(PAGE_DEFAULT_KEY);
 780
 781	/* Only cosmetics */
 782	memblock_enforce_memory_limit(memblock_end_of_DRAM());
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 783}
 784
 785/*
 786 * Setup hardware capabilities.
 787 */
 788static int __init setup_hwcaps(void)
 789{
 790	static const int stfl_bits[6] = { 0, 2, 7, 17, 19, 21 };
 791	struct cpuid cpu_id;
 792	int i;
 793
 794	/*
 795	 * The store facility list bits numbers as found in the principles
 796	 * of operation are numbered with bit 1UL<<31 as number 0 to
 797	 * bit 1UL<<0 as number 31.
 798	 *   Bit 0: instructions named N3, "backported" to esa-mode
 799	 *   Bit 2: z/Architecture mode is active
 800	 *   Bit 7: the store-facility-list-extended facility is installed
 801	 *   Bit 17: the message-security assist is installed
 802	 *   Bit 19: the long-displacement facility is installed
 803	 *   Bit 21: the extended-immediate facility is installed
 804	 *   Bit 22: extended-translation facility 3 is installed
 805	 *   Bit 30: extended-translation facility 3 enhancement facility
 806	 * These get translated to:
 807	 *   HWCAP_S390_ESAN3 bit 0, HWCAP_S390_ZARCH bit 1,
 808	 *   HWCAP_S390_STFLE bit 2, HWCAP_S390_MSA bit 3,
 809	 *   HWCAP_S390_LDISP bit 4, HWCAP_S390_EIMM bit 5 and
 810	 *   HWCAP_S390_ETF3EH bit 8 (22 && 30).
 811	 */
 812	for (i = 0; i < 6; i++)
 813		if (test_facility(stfl_bits[i]))
 814			elf_hwcap |= 1UL << i;
 815
 816	if (test_facility(22) && test_facility(30))
 817		elf_hwcap |= HWCAP_S390_ETF3EH;
 818
 819	/*
 820	 * Check for additional facilities with store-facility-list-extended.
 821	 * stfle stores doublewords (8 byte) with bit 1ULL<<63 as bit 0
 822	 * and 1ULL<<0 as bit 63. Bits 0-31 contain the same information
 823	 * as stored by stfl, bits 32-xxx contain additional facilities.
 824	 * How many facility words are stored depends on the number of
 825	 * doublewords passed to the instruction. The additional facilities
 826	 * are:
 827	 *   Bit 42: decimal floating point facility is installed
 828	 *   Bit 44: perform floating point operation facility is installed
 829	 * translated to:
 830	 *   HWCAP_S390_DFP bit 6 (42 && 44).
 831	 */
 832	if ((elf_hwcap & (1UL << 2)) && test_facility(42) && test_facility(44))
 833		elf_hwcap |= HWCAP_S390_DFP;
 834
 835	/*
 836	 * Huge page support HWCAP_S390_HPAGE is bit 7.
 837	 */
 838	if (MACHINE_HAS_EDAT1)
 839		elf_hwcap |= HWCAP_S390_HPAGE;
 840
 841	/*
 842	 * 64-bit register support for 31-bit processes
 843	 * HWCAP_S390_HIGH_GPRS is bit 9.
 844	 */
 845	elf_hwcap |= HWCAP_S390_HIGH_GPRS;
 846
 847	/*
 848	 * Transactional execution support HWCAP_S390_TE is bit 10.
 849	 */
 850	if (MACHINE_HAS_TE)
 851		elf_hwcap |= HWCAP_S390_TE;
 852
 853	/*
 854	 * Vector extension HWCAP_S390_VXRS is bit 11. The Vector extension
 855	 * can be disabled with the "novx" parameter. Use MACHINE_HAS_VX
 856	 * instead of facility bit 129.
 857	 */
 858	if (MACHINE_HAS_VX) {
 859		elf_hwcap |= HWCAP_S390_VXRS;
 860		if (test_facility(134))
 861			elf_hwcap |= HWCAP_S390_VXRS_BCD;
 862		if (test_facility(135))
 863			elf_hwcap |= HWCAP_S390_VXRS_EXT;
 864		if (test_facility(148))
 865			elf_hwcap |= HWCAP_S390_VXRS_EXT2;
 866		if (test_facility(152))
 867			elf_hwcap |= HWCAP_S390_VXRS_PDE;
 868	}
 869	if (test_facility(150))
 870		elf_hwcap |= HWCAP_S390_SORT;
 871	if (test_facility(151))
 872		elf_hwcap |= HWCAP_S390_DFLT;
 873
 874	/*
 875	 * Guarded storage support HWCAP_S390_GS is bit 12.
 876	 */
 877	if (MACHINE_HAS_GS)
 878		elf_hwcap |= HWCAP_S390_GS;
 879
 880	get_cpu_id(&cpu_id);
 881	add_device_randomness(&cpu_id, sizeof(cpu_id));
 882	switch (cpu_id.machine) {
 
 
 
 
 
 
 883	case 0x2064:
 884	case 0x2066:
 
 885	default:	/* Use "z900" as default for 64 bit kernels. */
 
 886		strcpy(elf_platform, "z900");
 887		break;
 888	case 0x2084:
 889	case 0x2086:
 890		strcpy(elf_platform, "z990");
 891		break;
 892	case 0x2094:
 893	case 0x2096:
 894		strcpy(elf_platform, "z9-109");
 895		break;
 896	case 0x2097:
 897	case 0x2098:
 898		strcpy(elf_platform, "z10");
 899		break;
 900	case 0x2817:
 901	case 0x2818:
 902		strcpy(elf_platform, "z196");
 903		break;
 904	case 0x2827:
 905	case 0x2828:
 906		strcpy(elf_platform, "zEC12");
 907		break;
 908	case 0x2964:
 909	case 0x2965:
 910		strcpy(elf_platform, "z13");
 911		break;
 912	case 0x3906:
 913	case 0x3907:
 914		strcpy(elf_platform, "z14");
 915		break;
 916	case 0x8561:
 917	case 0x8562:
 918		strcpy(elf_platform, "z15");
 919		break;
 920	}
 921
 922	/*
 923	 * Virtualization support HWCAP_INT_SIE is bit 0.
 924	 */
 925	if (sclp.has_sief2)
 926		int_hwcap |= HWCAP_INT_SIE;
 927
 928	return 0;
 929}
 930arch_initcall(setup_hwcaps);
 931
 932/*
 933 * Add system information as device randomness
 934 */
 935static void __init setup_randomness(void)
 936{
 937	struct sysinfo_3_2_2 *vmms;
 938
 939	vmms = (struct sysinfo_3_2_2 *) memblock_phys_alloc(PAGE_SIZE,
 940							    PAGE_SIZE);
 941	if (!vmms)
 942		panic("Failed to allocate memory for sysinfo structure\n");
 943
 944	if (stsi(vmms, 3, 2, 2) == 0 && vmms->count)
 945		add_device_randomness(&vmms->vm, sizeof(vmms->vm[0]) * vmms->count);
 946	memblock_free((unsigned long) vmms, PAGE_SIZE);
 947}
 948
 949/*
 950 * Find the correct size for the task_struct. This depends on
 951 * the size of the struct fpu at the end of the thread_struct
 952 * which is embedded in the task_struct.
 953 */
 954static void __init setup_task_size(void)
 955{
 956	int task_size = sizeof(struct task_struct);
 957
 958	if (!MACHINE_HAS_VX) {
 959		task_size -= sizeof(__vector128) * __NUM_VXRS;
 960		task_size += sizeof(freg_t) * __NUM_FPRS;
 961	}
 962	arch_task_struct_size = task_size;
 963}
 964
 965/*
 966 * Issue diagnose 318 to set the control program name and
 967 * version codes.
 968 */
 969static void __init setup_control_program_code(void)
 970{
 971	union diag318_info diag318_info = {
 972		.cpnc = CPNC_LINUX,
 973		.cpvc = 0,
 974	};
 975
 976	if (!sclp.has_diag318)
 977		return;
 978
 979	diag_stat_inc(DIAG_STAT_X318);
 980	asm volatile("diag %0,0,0x318\n" : : "d" (diag318_info.val));
 981}
 982
 983/*
 984 * Print the component list from the IPL report
 985 */
 986static void __init log_component_list(void)
 987{
 988	struct ipl_rb_component_entry *ptr, *end;
 989	char *str;
 990
 991	if (!early_ipl_comp_list_addr)
 992		return;
 993	if (ipl_block.hdr.flags & IPL_PL_FLAG_SIPL)
 994		pr_info("Linux is running with Secure-IPL enabled\n");
 995	else
 996		pr_info("Linux is running with Secure-IPL disabled\n");
 997	ptr = (void *) early_ipl_comp_list_addr;
 998	end = (void *) ptr + early_ipl_comp_list_size;
 999	pr_info("The IPL report contains the following components:\n");
1000	while (ptr < end) {
1001		if (ptr->flags & IPL_RB_COMPONENT_FLAG_SIGNED) {
1002			if (ptr->flags & IPL_RB_COMPONENT_FLAG_VERIFIED)
1003				str = "signed, verified";
1004			else
1005				str = "signed, verification failed";
1006		} else {
1007			str = "not signed";
1008		}
1009		pr_info("%016llx - %016llx (%s)\n",
1010			ptr->addr, ptr->addr + ptr->len, str);
1011		ptr++;
1012	}
1013}
1014
1015/*
1016 * Setup function called from init/main.c just after the banner
1017 * was printed.
1018 */
1019
1020void __init setup_arch(char **cmdline_p)
1021{
1022        /*
1023         * print what head.S has found out about the machine
1024         */
 
 
 
 
 
 
 
 
 
 
 
 
 
1025	if (MACHINE_IS_VM)
1026		pr_info("Linux is running as a z/VM "
1027			"guest operating system in 64-bit mode\n");
1028	else if (MACHINE_IS_KVM)
1029		pr_info("Linux is running under KVM in 64-bit mode\n");
1030	else if (MACHINE_IS_LPAR)
1031		pr_info("Linux is running natively in 64-bit mode\n");
1032	else
1033		pr_info("Linux is running as a guest in 64-bit mode\n");
1034
1035	log_component_list();
1036
1037	/* Have one command line that is parsed and saved in /proc/cmdline */
1038	/* boot_command_line has been already set up in early.c */
1039	*cmdline_p = boot_command_line;
1040
1041        ROOT_DEV = Root_RAM0;
1042
1043	setup_initial_init_mm(_text, _etext, _edata, _end);
 
 
 
1044
1045	if (IS_ENABLED(CONFIG_EXPOLINE_AUTO))
1046		nospec_auto_detect();
 
 
1047
1048	jump_label_init();
1049	parse_early_param();
1050#ifdef CONFIG_CRASH_DUMP
1051	/* Deactivate elfcorehdr= kernel parameter */
1052	elfcorehdr_addr = ELFCORE_ADDR_MAX;
1053#endif
1054
1055	os_info_init();
1056	setup_ipl();
1057	setup_task_size();
1058	setup_control_program_code();
1059
1060	/* Do some memory reservations *before* memory is added to memblock */
1061	reserve_above_ident_map();
1062	reserve_kernel();
1063	reserve_initrd();
1064	reserve_certificate_list();
1065	reserve_mem_detect_info();
1066	memblock_allow_resize();
1067
1068	/* Get information about *all* installed memory */
1069	memblock_add_mem_detect_info();
1070
1071	free_mem_detect_info();
1072
1073	setup_uv();
1074	setup_memory_end();
 
 
 
1075	setup_memory();
1076	dma_contiguous_reserve(ident_map_size);
1077	vmcp_cma_reserve();
1078	if (MACHINE_HAS_EDAT2)
1079		hugetlb_cma_reserve(PUD_SHIFT - PAGE_SHIFT);
 
 
1080
1081	check_initrd();
1082	reserve_crashkernel();
1083#ifdef CONFIG_CRASH_DUMP
1084	/*
1085	 * Be aware that smp_save_dump_cpus() triggers a system reset.
1086	 * Therefore CPU and device initialization should be done afterwards.
1087	 */
1088	smp_save_dump_cpus();
1089#endif
1090
1091	setup_resources();
1092	setup_lowcore_dat_off();
1093	smp_fill_possible_mask();
1094	cpu_detect_mhz_feature();
1095        cpu_init();
1096	numa_setup();
1097	smp_detect_cpus();
1098	topology_init_early();
1099
1100	/*
1101	 * Create kernel page tables and switch to virtual addressing.
1102	 */
1103        paging_init();
1104
1105	/*
1106	 * After paging_init created the kernel page table, the new PSWs
1107	 * in lowcore can now run with DAT enabled.
1108	 */
1109	setup_lowcore_dat_on();
1110
1111        /* Setup default console */
1112	conmode_default();
1113	set_preferred_console();
1114
1115	apply_alternative_instructions();
1116	if (IS_ENABLED(CONFIG_EXPOLINE))
1117		nospec_init_branches();
1118
1119	/* Setup zfcp/nvme dump support */
1120	setup_zfcpdump();
1121
1122	/* Add system specific data to the random pool */
1123	setup_randomness();
1124}