diff options
Diffstat (limited to 'arch/s390/kernel/setup.c')
| -rw-r--r-- | arch/s390/kernel/setup.c | 851 |
1 files changed, 454 insertions, 397 deletions
diff --git a/arch/s390/kernel/setup.c b/arch/s390/kernel/setup.c index 0663287fa1b..1e2264b46e4 100644 --- a/arch/s390/kernel/setup.c +++ b/arch/s390/kernel/setup.c @@ -1,8 +1,6 @@ /* - * arch/s390/kernel/setup.c - * * S390 version - * Copyright (C) 1999,2000 IBM Deutschland Entwicklung GmbH, IBM Corporation + * Copyright IBM Corp. 1999, 2012 * Author(s): Hartmut Penner (hp@de.ibm.com), * Martin Schwidefsky (schwidefsky@de.ibm.com) * @@ -18,14 +16,14 @@ #define pr_fmt(fmt) KMSG_COMPONENT ": " fmt #include <linux/errno.h> -#include <linux/module.h> +#include <linux/export.h> #include <linux/sched.h> #include <linux/kernel.h> +#include <linux/memblock.h> #include <linux/mm.h> #include <linux/stddef.h> #include <linux/unistd.h> #include <linux/ptrace.h> -#include <linux/slab.h> #include <linux/user.h> #include <linux/tty.h> #include <linux/ioport.h> @@ -43,10 +41,13 @@ #include <linux/reboot.h> #include <linux/topology.h> #include <linux/ftrace.h> +#include <linux/kexec.h> +#include <linux/crash_dump.h> +#include <linux/memory.h> +#include <linux/compat.h> #include <asm/ipl.h> -#include <asm/uaccess.h> -#include <asm/system.h> +#include <asm/facility.h> #include <asm/smp.h> #include <asm/mmu_context.h> #include <asm/cpcmd.h> @@ -56,20 +57,11 @@ #include <asm/ptrace.h> #include <asm/sections.h> #include <asm/ebcdic.h> -#include <asm/compat.h> #include <asm/kvm_virtio.h> - -long psw_kernel_bits = (PSW_BASE_BITS | PSW_MASK_DAT | PSW_ASC_PRIMARY | - PSW_MASK_MCHECK | PSW_DEFAULT_KEY); -long psw_user_bits = (PSW_BASE_BITS | PSW_MASK_DAT | PSW_ASC_HOME | - PSW_MASK_IO | PSW_MASK_EXT | PSW_MASK_MCHECK | - PSW_MASK_PSTATE | PSW_DEFAULT_KEY); - -/* - * User copy operations. - */ -struct uaccess_ops uaccess; -EXPORT_SYMBOL(uaccess); +#include <asm/diag.h> +#include <asm/os_info.h> +#include <asm/sclp.h> +#include "entry.h" /* * Machine setup.. @@ -86,11 +78,23 @@ EXPORT_SYMBOL(console_irq); unsigned long elf_hwcap = 0; char elf_platform[ELF_PLATFORM_SIZE]; -struct mem_chunk __initdata memory_chunk[MEMORY_CHUNKS]; -volatile int __cpu_logical_map[NR_CPUS]; /* logical cpu to cpu address */ - int __initdata memory_end_set; unsigned long __initdata memory_end; +unsigned long __initdata max_physmem_end; + +unsigned long VMALLOC_START; +EXPORT_SYMBOL(VMALLOC_START); + +unsigned long VMALLOC_END; +EXPORT_SYMBOL(VMALLOC_END); + +struct page *vmemmap; +EXPORT_SYMBOL(vmemmap); + +#ifdef CONFIG_64BIT +unsigned long MODULES_VADDR; +unsigned long MODULES_END; +#endif /* An array with a pointer to the lowcore of every CPU. */ struct _lowcore *lowcore_ptr[NR_CPUS]; @@ -104,38 +108,6 @@ EXPORT_SYMBOL(lowcore_ptr); #include <asm/setup.h> -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, -}; - -/* - * cpu_init() initializes state that is per-CPU. - */ -void __cpuinit cpu_init(void) -{ - /* - * Store processor id in lowcore (used e.g. in timer_interrupt) - */ - get_cpu_id(&S390_lowcore.cpu_id); - - /* - * Force FPU initialization: - */ - clear_thread_flag(TIF_USEDFPU); - clear_used_math(); - - atomic_inc(&init_mm.mm_count); - current->active_mm = &init_mm; - BUG_ON(current->mm); - enter_lazy_tlb(&init_mm, current); -} - /* * condev= and conmode= setup parameter. */ @@ -156,9 +128,14 @@ __setup("condev=", condev_setup); static void __init set_preferred_console(void) { - if (MACHINE_IS_KVM) - add_preferred_console("hvc", 0, NULL); - else if (CONSOLE_IS_3215 || CONSOLE_IS_SCLP) + 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) add_preferred_console("ttyS", 0, NULL); else if (CONSOLE_IS_3270) add_preferred_console("tty3270", 0, NULL); @@ -234,25 +211,19 @@ static void __init conmode_default(void) } } -#ifdef CONFIG_ZFCPDUMP -static void __init setup_zfcpdump(unsigned int console_devno) +#ifdef CONFIG_CRASH_DUMP +static void __init setup_zfcpdump(void) { - static char str[41]; - if (ipl_info.type != IPL_TYPE_FCP_DUMP) return; - if (console_devno != -1) - sprintf(str, " cio_ignore=all,!0.0.%04x,!0.0.%04x", - ipl_info.data.fcp.dev_id.devno, console_devno); - else - sprintf(str, " cio_ignore=all,!0.0.%04x", - ipl_info.data.fcp.dev_id.devno); - strcat(boot_command_line, str); + if (OLDMEM_BASE) + return; + strcat(boot_command_line, " cio_ignore=all,!ipldev,!condev"); console_loglevel = 2; } #else -static inline void setup_zfcpdump(unsigned int console_devno) {} -#endif /* CONFIG_ZFCPDUMP */ +static inline void setup_zfcpdump(void) {} +#endif /* CONFIG_CRASH_DUMP */ /* * Reboot, halt and power_off stubs. They just call _machine_restart, @@ -296,154 +267,76 @@ void machine_power_off(void) * Dummy power off function. */ void (*pm_power_off)(void) = machine_power_off; +EXPORT_SYMBOL_GPL(pm_power_off); static int __init early_parse_mem(char *p) { memory_end = memparse(p, &p); + memory_end &= PAGE_MASK; memory_end_set = 1; return 0; } early_param("mem", early_parse_mem); -unsigned int user_mode = HOME_SPACE_MODE; -EXPORT_SYMBOL_GPL(user_mode); - -static int set_amode_and_uaccess(unsigned long user_amode, - unsigned long user32_amode) -{ - psw_user_bits = PSW_BASE_BITS | PSW_MASK_DAT | user_amode | - PSW_MASK_IO | PSW_MASK_EXT | PSW_MASK_MCHECK | - PSW_MASK_PSTATE | PSW_DEFAULT_KEY; -#ifdef CONFIG_COMPAT - psw_user32_bits = PSW_BASE32_BITS | PSW_MASK_DAT | user_amode | - PSW_MASK_IO | PSW_MASK_EXT | PSW_MASK_MCHECK | - PSW_MASK_PSTATE | PSW_DEFAULT_KEY; - psw32_user_bits = PSW32_BASE_BITS | PSW32_MASK_DAT | user32_amode | - PSW32_MASK_IO | PSW32_MASK_EXT | PSW32_MASK_MCHECK | - PSW32_MASK_PSTATE; -#endif - psw_kernel_bits = PSW_BASE_BITS | PSW_MASK_DAT | PSW_ASC_HOME | - PSW_MASK_MCHECK | PSW_DEFAULT_KEY; - - if (MACHINE_HAS_MVCOS) { - memcpy(&uaccess, &uaccess_mvcos_switch, sizeof(uaccess)); - return 1; - } else { - memcpy(&uaccess, &uaccess_pt, sizeof(uaccess)); - return 0; - } -} - -/* - * Switch kernel/user addressing modes? - */ -static int __init early_parse_switch_amode(char *p) -{ - if (user_mode != SECONDARY_SPACE_MODE) - user_mode = PRIMARY_SPACE_MODE; - return 0; -} -early_param("switch_amode", early_parse_switch_amode); - -static int __init early_parse_user_mode(char *p) -{ - if (p && strcmp(p, "primary") == 0) - user_mode = PRIMARY_SPACE_MODE; -#ifdef CONFIG_S390_EXEC_PROTECT - else if (p && strcmp(p, "secondary") == 0) - user_mode = SECONDARY_SPACE_MODE; -#endif - else if (!p || strcmp(p, "home") == 0) - user_mode = HOME_SPACE_MODE; - else - return 1; - return 0; -} -early_param("user_mode", early_parse_user_mode); - -#ifdef CONFIG_S390_EXEC_PROTECT -/* - * Enable execute protection? - */ -static int __init early_parse_noexec(char *p) +static int __init parse_vmalloc(char *arg) { - if (!strncmp(p, "off", 3)) - return 0; - user_mode = SECONDARY_SPACE_MODE; + if (!arg) + return -EINVAL; + VMALLOC_END = (memparse(arg, &arg) + PAGE_SIZE - 1) & PAGE_MASK; return 0; } -early_param("noexec", early_parse_noexec); -#endif /* CONFIG_S390_EXEC_PROTECT */ +early_param("vmalloc", parse_vmalloc); -static void setup_addressing_mode(void) -{ - if (user_mode == SECONDARY_SPACE_MODE) { - if (set_amode_and_uaccess(PSW_ASC_SECONDARY, - PSW32_ASC_SECONDARY)) - pr_info("Execute protection active, " - "mvcos available\n"); - else - pr_info("Execute protection active, " - "mvcos not available\n"); - } else if (user_mode == PRIMARY_SPACE_MODE) { - if (set_amode_and_uaccess(PSW_ASC_PRIMARY, PSW32_ASC_PRIMARY)) - pr_info("Address spaces switched, " - "mvcos available\n"); - else - pr_info("Address spaces switched, " - "mvcos not available\n"); - } -#ifdef CONFIG_TRACE_IRQFLAGS - sysc_restore_trace_psw.mask = psw_kernel_bits & ~PSW_MASK_MCHECK; - io_restore_trace_psw.mask = psw_kernel_bits & ~PSW_MASK_MCHECK; -#endif -} +void *restart_stack __attribute__((__section__(".data"))); -static void __init -setup_lowcore(void) +static void __init setup_lowcore(void) { struct _lowcore *lc; - int lc_pages; /* * Setup lowcore for boot cpu */ - lc_pages = sizeof(void *) == 8 ? 2 : 1; - lc = (struct _lowcore *) - __alloc_bootmem(lc_pages * PAGE_SIZE, lc_pages * PAGE_SIZE, 0); - memset(lc, 0, lc_pages * PAGE_SIZE); - lc->restart_psw.mask = PSW_BASE_BITS | PSW_DEFAULT_KEY; + BUILD_BUG_ON(sizeof(struct _lowcore) != LC_PAGES * 4096); + lc = __alloc_bootmem_low(LC_PAGES * PAGE_SIZE, LC_PAGES * PAGE_SIZE, 0); + lc->restart_psw.mask = PSW_KERNEL_BITS; lc->restart_psw.addr = PSW_ADDR_AMODE | (unsigned long) restart_int_handler; - if (user_mode != HOME_SPACE_MODE) - lc->restart_psw.mask |= PSW_ASC_HOME; - lc->external_new_psw.mask = psw_kernel_bits; + lc->external_new_psw.mask = PSW_KERNEL_BITS | + PSW_MASK_DAT | PSW_MASK_MCHECK; lc->external_new_psw.addr = PSW_ADDR_AMODE | (unsigned long) ext_int_handler; - lc->svc_new_psw.mask = psw_kernel_bits | PSW_MASK_IO | PSW_MASK_EXT; + lc->svc_new_psw.mask = PSW_KERNEL_BITS | + PSW_MASK_DAT | PSW_MASK_IO | PSW_MASK_EXT | PSW_MASK_MCHECK; lc->svc_new_psw.addr = PSW_ADDR_AMODE | (unsigned long) system_call; - lc->program_new_psw.mask = psw_kernel_bits; + lc->program_new_psw.mask = PSW_KERNEL_BITS | + PSW_MASK_DAT | PSW_MASK_MCHECK; lc->program_new_psw.addr = - PSW_ADDR_AMODE | (unsigned long)pgm_check_handler; - lc->mcck_new_psw.mask = - psw_kernel_bits & ~PSW_MASK_MCHECK & ~PSW_MASK_DAT; + PSW_ADDR_AMODE | (unsigned long) pgm_check_handler; + lc->mcck_new_psw.mask = PSW_KERNEL_BITS; lc->mcck_new_psw.addr = PSW_ADDR_AMODE | (unsigned long) mcck_int_handler; - lc->io_new_psw.mask = psw_kernel_bits; + lc->io_new_psw.mask = PSW_KERNEL_BITS | + PSW_MASK_DAT | PSW_MASK_MCHECK; lc->io_new_psw.addr = PSW_ADDR_AMODE | (unsigned long) io_int_handler; lc->clock_comparator = -1ULL; - lc->kernel_stack = ((unsigned long) &init_thread_union) + THREAD_SIZE; + lc->kernel_stack = ((unsigned long) &init_thread_union) + + THREAD_SIZE - STACK_FRAME_OVERHEAD - sizeof(struct pt_regs); lc->async_stack = (unsigned long) - __alloc_bootmem(ASYNC_SIZE, ASYNC_SIZE, 0) + ASYNC_SIZE; + __alloc_bootmem(ASYNC_SIZE, ASYNC_SIZE, 0) + + ASYNC_SIZE - STACK_FRAME_OVERHEAD - sizeof(struct pt_regs); lc->panic_stack = (unsigned long) - __alloc_bootmem(PAGE_SIZE, PAGE_SIZE, 0) + PAGE_SIZE; + __alloc_bootmem(PAGE_SIZE, PAGE_SIZE, 0) + + PAGE_SIZE - STACK_FRAME_OVERHEAD - sizeof(struct pt_regs); lc->current_task = (unsigned long) init_thread_union.thread_info.task; lc->thread_info = (unsigned long) &init_thread_union; lc->machine_flags = S390_lowcore.machine_flags; + lc->stfl_fac_list = S390_lowcore.stfl_fac_list; + memcpy(lc->stfle_fac_list, S390_lowcore.stfle_fac_list, + MAX_FACILITY_BIT/8); #ifndef CONFIG_64BIT if (MACHINE_HAS_IEEE) { lc->extended_save_area_addr = (__u32) - __alloc_bootmem(PAGE_SIZE, PAGE_SIZE, 0); + __alloc_bootmem_low(PAGE_SIZE, PAGE_SIZE, 0); /* enable extended save area */ __ctl_set_bit(14, 29); } @@ -459,243 +352,345 @@ setup_lowcore(void) lc->last_update_timer = S390_lowcore.last_update_timer; lc->last_update_clock = S390_lowcore.last_update_clock; lc->ftrace_func = S390_lowcore.ftrace_func; + + 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 + * restart data to the absolute zero lowcore. This is necessary if + * 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; + + /* Setup absolute zero lowcore */ + 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); + +#ifdef CONFIG_SMP + lc->spinlock_lockval = arch_spin_lockval(0); +#endif + set_prefix((u32)(unsigned long) lc); lowcore_ptr[0] = lc; } -static void __init -setup_resources(void) +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, +}; + +static struct resource bss_resource = { + .name = "Kernel bss", + .flags = IORESOURCE_BUSY | IORESOURCE_MEM, +}; + +static struct resource __initdata *standard_resources[] = { + &code_resource, + &data_resource, + &bss_resource, +}; + +static void __init setup_resources(void) { - struct resource *res, *sub_res; - int i; + struct resource *res, *std_res, *sub_res; + struct memblock_region *reg; + int j; 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; + bss_resource.start = (unsigned long) &__bss_start; + bss_resource.end = (unsigned long) &__bss_stop - 1; - for (i = 0; i < MEMORY_CHUNKS; i++) { - if (!memory_chunk[i].size) - continue; - res = alloc_bootmem_low(sizeof(struct resource)); + for_each_memblock(memory, reg) { + res = alloc_bootmem_low(sizeof(*res)); res->flags = IORESOURCE_BUSY | IORESOURCE_MEM; - switch (memory_chunk[i].type) { - case CHUNK_READ_WRITE: - res->name = "System RAM"; - break; - case CHUNK_READ_ONLY: - res->name = "System ROM"; - res->flags |= IORESOURCE_READONLY; - break; - default: - res->name = "reserved"; - } - res->start = memory_chunk[i].addr; - res->end = memory_chunk[i].addr + memory_chunk[i].size - 1; - request_resource(&iomem_resource, res); - if (code_resource.start >= res->start && - code_resource.start <= res->end && - code_resource.end > res->end) { - sub_res = alloc_bootmem_low(sizeof(struct resource)); - memcpy(sub_res, &code_resource, - sizeof(struct resource)); - sub_res->end = res->end; - code_resource.start = res->end + 1; - request_resource(res, sub_res); - } + res->name = "System RAM"; + res->start = reg->base; + res->end = reg->base + reg->size - 1; + request_resource(&iomem_resource, res); - if (code_resource.start >= res->start && - code_resource.start <= res->end && - code_resource.end <= res->end) - request_resource(res, &code_resource); - - if (data_resource.start >= res->start && - data_resource.start <= res->end && - data_resource.end > res->end) { - sub_res = alloc_bootmem_low(sizeof(struct resource)); - memcpy(sub_res, &data_resource, - sizeof(struct resource)); - sub_res->end = res->end; - data_resource.start = res->end + 1; - request_resource(res, sub_res); + 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); + } } - - if (data_resource.start >= res->start && - data_resource.start <= res->end && - data_resource.end <= res->end) - request_resource(res, &data_resource); } } -unsigned long real_memory_size; -EXPORT_SYMBOL_GPL(real_memory_size); - static void __init setup_memory_end(void) { - unsigned long memory_size; - unsigned long max_mem; - int i; - -#ifdef CONFIG_ZFCPDUMP - if (ipl_info.type == IPL_TYPE_FCP_DUMP) { - memory_end = ZFCPDUMP_HSA_SIZE; - memory_end_set = 1; - } + unsigned long vmax, vmalloc_size, tmp; + + /* Choose kernel address space layout: 2, 3, or 4 levels. */ +#ifdef CONFIG_64BIT + vmalloc_size = VMALLOC_END ?: (128UL << 30) - MODULES_LEN; + tmp = (memory_end ?: max_physmem_end) / PAGE_SIZE; + 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 */ + /* module area is at the end of the kernel address space. */ + MODULES_END = vmax; + MODULES_VADDR = MODULES_END - MODULES_LEN; + VMALLOC_END = MODULES_VADDR; +#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; #endif - memory_size = 0; - memory_end &= PAGE_MASK; + VMALLOC_START = vmax - vmalloc_size; + + /* Split remaining virtual space between 1:1 mapping & vmemmap array */ + tmp = VMALLOC_START / (PAGE_SIZE + sizeof(struct page)); + /* vmemmap contains a multiple of PAGES_PER_SECTION struct pages */ + tmp = SECTION_ALIGN_UP(tmp); + 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 */ + memory_end = min(memory_end ?: max_physmem_end, tmp); + max_pfn = max_low_pfn = PFN_DOWN(memory_end); + memblock_remove(memory_end, ULONG_MAX); + + pr_notice("Max memory size: %luMB\n", memory_end >> 20); +} - max_mem = memory_end ? min(VMEM_MAX_PHYS, memory_end) : VMEM_MAX_PHYS; - memory_end = min(max_mem, memory_end); +static void __init setup_vmcoreinfo(void) +{ + mem_assign_absolute(S390_lowcore.vmcore_info, paddr_vmcoreinfo_note()); +} - /* - * Make sure all chunks are MAX_ORDER aligned so we don't need the - * extra checks that HOLES_IN_ZONE would require. - */ - for (i = 0; i < MEMORY_CHUNKS; i++) { - unsigned long start, end; - struct mem_chunk *chunk; - unsigned long align; - - chunk = &memory_chunk[i]; - align = 1UL << (MAX_ORDER + PAGE_SHIFT - 1); - start = (chunk->addr + align - 1) & ~(align - 1); - end = (chunk->addr + chunk->size) & ~(align - 1); - if (start >= end) - memset(chunk, 0, sizeof(*chunk)); - else { - chunk->addr = start; - chunk->size = end - start; - } - } +#ifdef CONFIG_CRASH_DUMP - for (i = 0; i < MEMORY_CHUNKS; i++) { - struct mem_chunk *chunk = &memory_chunk[i]; +/* + * 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; + + 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; +} - real_memory_size = max(real_memory_size, - chunk->addr + chunk->size); - if (chunk->addr >= max_mem) { - memset(chunk, 0, sizeof(*chunk)); - continue; - } - if (chunk->addr + chunk->size > max_mem) - chunk->size = max_mem - chunk->addr; - memory_size = max(memory_size, chunk->addr + chunk->size); +static struct notifier_block kdump_mem_nb = { + .notifier_call = kdump_mem_notifier, +}; + +#endif + +/* + * Make sure that the area behind memory_end is protected + */ +static void reserve_memory_end(void) +{ +#ifdef CONFIG_CRASH_DUMP + 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; } - if (!memory_end) - memory_end = memory_size; +#endif + if (!memory_end_set) + return; + memblock_reserve(memory_end, ULONG_MAX); } -static void __init -setup_memory(void) +/* + * Make sure that oldmem, where the dump is stored, is protected + */ +static void reserve_oldmem(void) { - unsigned long bootmap_size; - unsigned long start_pfn, end_pfn; - int i; +#ifdef CONFIG_CRASH_DUMP + if (OLDMEM_BASE) + /* Forget all memory above the running kdump system */ + memblock_reserve(OLDMEM_SIZE, (phys_addr_t)ULONG_MAX); +#endif +} - /* - * partially used pages are not usable - thus - * we are rounding upwards: - */ - start_pfn = PFN_UP(__pa(&_end)); - end_pfn = max_pfn = PFN_DOWN(memory_end); +/* + * 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); +#endif +} -#ifdef CONFIG_BLK_DEV_INITRD - /* - * Move the initrd in case the bitmap of the bootmem allocater - * would overwrite it. - */ +/* + * 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; + phys_addr_t low, high; + int rc; - if (INITRD_START && INITRD_SIZE) { - unsigned long bmap_size; - unsigned long start; + rc = parse_crashkernel(boot_command_line, memory_end, &crash_size, + &crash_base); - bmap_size = bootmem_bootmap_pages(end_pfn - start_pfn + 1); - bmap_size = PFN_PHYS(bmap_size); + crash_base = ALIGN(crash_base, KEXEC_CRASH_MEM_ALIGN); + crash_size = ALIGN(crash_size, KEXEC_CRASH_MEM_ALIGN); + if (rc || crash_size == 0) + return; - if (PFN_PHYS(start_pfn) + bmap_size > INITRD_START) { - start = PFN_PHYS(start_pfn) + bmap_size + PAGE_SIZE; + if (memblock.memory.regions[0].size < crash_size) { + pr_info("crashkernel reservation failed: %s\n", + "first memory chunk must be at least crashkernel size"); + return; + } - if (start + INITRD_SIZE > memory_end) { - pr_err("initrd extends beyond end of " - "memory (0x%08lx > 0x%08lx) " - "disabling initrd\n", - start + INITRD_SIZE, memory_end); - INITRD_START = INITRD_SIZE = 0; - } else { - pr_info("Moving initrd (0x%08lx -> " - "0x%08lx, size: %ld)\n", - INITRD_START, start, INITRD_SIZE); - memmove((void *) start, (void *) INITRD_START, - INITRD_SIZE); - INITRD_START = start; - } + 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); } -#endif - /* - * Initialize the boot-time allocator - */ - bootmap_size = init_bootmem(start_pfn, end_pfn); + if (!crash_base) { + pr_info("crashkernel reservation failed: %s\n", + "no suitable area found"); + return; + } - /* - * Register RAM areas with the bootmem allocator. - */ + if (register_memory_notifier(&kdump_mem_nb)) + return; - for (i = 0; i < MEMORY_CHUNKS && memory_chunk[i].size > 0; i++) { - unsigned long start_chunk, end_chunk, pfn; - - if (memory_chunk[i].type != CHUNK_READ_WRITE) - continue; - start_chunk = PFN_DOWN(memory_chunk[i].addr); - end_chunk = start_chunk + PFN_DOWN(memory_chunk[i].size); - end_chunk = min(end_chunk, end_pfn); - if (start_chunk >= end_chunk) - continue; - add_active_range(0, start_chunk, end_chunk); - pfn = max(start_chunk, start_pfn); - for (; pfn < end_chunk; pfn++) - page_set_storage_key(PFN_PHYS(pfn), PAGE_DEFAULT_KEY); - } + 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); + memblock_remove(crash_base, crash_size); + pr_info("Reserving %lluMB of memory at %lluMB " + "for crashkernel (System RAM: %luMB)\n", + crash_size >> 20, crash_base >> 20, + (unsigned long)memblock.memory.total_size >> 20); + os_info_crashkernel_add(crash_base, crash_size); +#endif +} - psw_set_key(PAGE_DEFAULT_KEY); +/* + * Reserve the initrd from being used by memblock + */ +static void __init reserve_initrd(void) +{ +#ifdef CONFIG_BLK_DEV_INITRD + initrd_start = INITRD_START; + initrd_end = initrd_start + INITRD_SIZE; + memblock_reserve(INITRD_START, INITRD_SIZE); +#endif +} - free_bootmem_with_active_regions(0, max_pfn); +/* + * 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; + start_pfn = PFN_UP(__pa(&_end)); /* * Reserve memory used for lowcore/command line/kernel image. */ - reserve_bootmem(0, (unsigned long)_ehead, BOOTMEM_DEFAULT); - reserve_bootmem((unsigned long)_stext, - PFN_PHYS(start_pfn) - (unsigned long)_stext, - BOOTMEM_DEFAULT); + memblock_reserve(0, (unsigned long)_ehead); + memblock_reserve((unsigned long)_stext, PFN_PHYS(start_pfn) + - (unsigned long)_stext); +} + +static void __init reserve_elfcorehdr(void) +{ +#ifdef CONFIG_CRASH_DUMP + if (is_kdump_kernel()) + memblock_reserve(elfcorehdr_addr - OLDMEM_BASE, + PAGE_ALIGN(elfcorehdr_size)); +#endif +} + +static void __init setup_memory(void) +{ + struct memblock_region *reg; + /* - * Reserve the bootmem bitmap itself as well. We do this in two - * steps (first step was init_bootmem()) because this catches - * the (very unlikely) case of us accidentally initializing the - * bootmem allocator with an invalid RAM area. + * Init storage key for present memory */ - reserve_bootmem(start_pfn << PAGE_SHIFT, bootmap_size, - BOOTMEM_DEFAULT); - -#ifdef CONFIG_BLK_DEV_INITRD - if (INITRD_START && INITRD_SIZE) { - if (INITRD_START + INITRD_SIZE <= memory_end) { - reserve_bootmem(INITRD_START, INITRD_SIZE, - BOOTMEM_DEFAULT); - initrd_start = INITRD_START; - initrd_end = initrd_start + INITRD_SIZE; - } else { - pr_err("initrd extends beyond end of " - "memory (0x%08lx > 0x%08lx) " - "disabling initrd\n", - initrd_start + INITRD_SIZE, memory_end); - initrd_start = initrd_end = 0; - } + for_each_memblock(memory, reg) { + storage_key_init_range(reg->base, reg->base + reg->size); } -#endif + psw_set_key(PAGE_DEFAULT_KEY); + + /* Only cosmetics */ + memblock_enforce_memory_limit(memblock_end_of_DRAM()); } /* @@ -704,11 +699,9 @@ setup_memory(void) static void __init setup_hwcaps(void) { static const int stfl_bits[6] = { 0, 2, 7, 17, 19, 21 }; - unsigned long long facility_list_extended; - unsigned int facility_list; + struct cpuid cpu_id; int i; - facility_list = stfl(); /* * The store facility list bits numbers as found in the principles * of operation are numbered with bit 1UL<<31 as number 0 to @@ -728,11 +721,10 @@ static void __init setup_hwcaps(void) * HWCAP_S390_ETF3EH bit 8 (22 && 30). */ for (i = 0; i < 6; i++) - if (facility_list & (1UL << (31 - stfl_bits[i]))) + if (test_facility(stfl_bits[i])) elf_hwcap |= 1UL << i; - if ((facility_list & (1UL << (31 - 22))) - && (facility_list & (1UL << (31 - 30)))) + if (test_facility(22) && test_facility(30)) elf_hwcap |= HWCAP_S390_ETF3EH; /* @@ -741,19 +733,15 @@ static void __init setup_hwcaps(void) * 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 - * doublewords passed to the instruction. The additional facilites + * doublewords passed to the instruction. The additional facilities * are: * Bit 42: decimal floating point facility is installed * Bit 44: perform floating point operation facility is installed * translated to: * HWCAP_S390_DFP bit 6 (42 && 44). */ - if ((elf_hwcap & (1UL << 2)) && - __stfle(&facility_list_extended, 1) > 0) { - if ((facility_list_extended & (1ULL << (63 - 42))) - && (facility_list_extended & (1ULL << (63 - 44)))) - elf_hwcap |= HWCAP_S390_DFP; - } + if ((elf_hwcap & (1UL << 2)) && test_facility(42) && test_facility(44)) + elf_hwcap |= HWCAP_S390_DFP; /* * Huge page support HWCAP_S390_HPAGE is bit 7. @@ -761,13 +749,22 @@ static void __init setup_hwcaps(void) if (MACHINE_HAS_HPAGE) elf_hwcap |= HWCAP_S390_HPAGE; +#if defined(CONFIG_64BIT) /* * 64-bit register support for 31-bit processes * HWCAP_S390_HIGH_GPRS is bit 9. */ elf_hwcap |= HWCAP_S390_HIGH_GPRS; - switch (S390_lowcore.cpu_id.machine) { + /* + * Transactional execution support HWCAP_S390_TE is bit 10. + */ + if (test_facility(50) && test_facility(73)) + elf_hwcap |= HWCAP_S390_TE; +#endif + + get_cpu_id(&cpu_id); + switch (cpu_id.machine) { case 0x9672: #if !defined(CONFIG_64BIT) default: /* Use "g5" as default for 31 bit kernels. */ @@ -793,6 +790,14 @@ static void __init setup_hwcaps(void) case 0x2098: strcpy(elf_platform, "z10"); break; + case 0x2817: + case 0x2818: + strcpy(elf_platform, "z196"); + break; + case 0x2827: + case 0x2828: + strcpy(elf_platform, "zEC12"); + break; } } @@ -801,8 +806,7 @@ static void __init setup_hwcaps(void) * was printed. */ -void __init -setup_arch(char **cmdline_p) +void __init setup_arch(char **cmdline_p) { /* * print what head.S has found out about the machine @@ -811,7 +815,7 @@ setup_arch(char **cmdline_p) if (MACHINE_IS_VM) pr_info("Linux is running as a z/VM " "guest operating system in 31-bit mode\n"); - else + else if (MACHINE_IS_LPAR) pr_info("Linux is running natively in 31-bit mode\n"); if (MACHINE_HAS_IEEE) pr_info("The hardware system has IEEE compatible " @@ -825,7 +829,7 @@ setup_arch(char **cmdline_p) "guest operating system in 64-bit mode\n"); else if (MACHINE_IS_KVM) pr_info("Linux is running under KVM in 64-bit mode\n"); - else + else if (MACHINE_IS_LPAR) pr_info("Linux is running natively in 64-bit mode\n"); #endif /* CONFIG_64BIT */ @@ -835,27 +839,48 @@ setup_arch(char **cmdline_p) ROOT_DEV = Root_RAM0; + /* Is init_mm really needed? */ 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; - if (MACHINE_HAS_MVCOS) - memcpy(&uaccess, &uaccess_mvcos, sizeof(uaccess)); - else - memcpy(&uaccess, &uaccess_std, sizeof(uaccess)); - parse_early_param(); - + os_info_init(); setup_ipl(); + + /* Do some memory reservations *before* memory is added to memblock */ + reserve_memory_end(); + reserve_oldmem(); + 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)); + setup_memory_end(); - setup_addressing_mode(); setup_memory(); + + check_initrd(); + reserve_crashkernel(); + setup_resources(); + setup_vmcoreinfo(); setup_lowcore(); - + smp_fill_possible_mask(); cpu_init(); - __cpu_logical_map[0] = stap(); s390_init_cpu_topology(); /* @@ -873,5 +898,37 @@ setup_arch(char **cmdline_p) set_preferred_console(); /* Setup zfcpdump support */ - setup_zfcpdump(console_devno); + setup_zfcpdump(); +} + +#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 |