diff options
Diffstat (limited to 'arch/s390/kernel')
29 files changed, 2222 insertions, 1196 deletions
diff --git a/arch/s390/kernel/Makefile b/arch/s390/kernel/Makefile index 7d9ec924e7e..884b18afc86 100644 --- a/arch/s390/kernel/Makefile +++ b/arch/s390/kernel/Makefile @@ -23,7 +23,7 @@ CFLAGS_sysinfo.o += -Iinclude/math-emu -Iarch/s390/math-emu -w obj-y := bitmap.o traps.o time.o process.o base.o early.o setup.o vtime.o \ processor.o sys_s390.o ptrace.o signal.o cpcmd.o ebcdic.o nmi.o \ debug.o irq.o ipl.o dis.o diag.o mem_detect.o sclp.o vdso.o \ - sysinfo.o jump_label.o + sysinfo.o jump_label.o lgr.o os_info.o obj-y += $(if $(CONFIG_64BIT),entry64.o,entry.o) obj-y += $(if $(CONFIG_64BIT),reipl64.o,reipl.o) @@ -34,8 +34,6 @@ extra-y += $(if $(CONFIG_64BIT),head64.o,head31.o) obj-$(CONFIG_MODULES) += s390_ksyms.o module.o obj-$(CONFIG_SMP) += smp.o obj-$(CONFIG_SCHED_BOOK) += topology.o -obj-$(CONFIG_SMP) += $(if $(CONFIG_64BIT),switch_cpu64.o, \ - switch_cpu.o) obj-$(CONFIG_HIBERNATION) += suspend.o swsusp_asm64.o obj-$(CONFIG_AUDIT) += audit.o compat-obj-$(CONFIG_AUDIT) += compat_audit.o @@ -50,6 +48,7 @@ obj-$(CONFIG_DYNAMIC_FTRACE) += ftrace.o obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o obj-$(CONFIG_FTRACE_SYSCALLS) += ftrace.o obj-$(CONFIG_CRASH_DUMP) += crash_dump.o +obj-$(CONFIG_PERF_EVENTS) += perf_event.o perf_cpum_cf.o # Kexec part S390_KEXEC_OBJS := machine_kexec.o crash.o diff --git a/arch/s390/kernel/asm-offsets.c b/arch/s390/kernel/asm-offsets.c index 6e6a72e66d6..ed8c913db79 100644 --- a/arch/s390/kernel/asm-offsets.c +++ b/arch/s390/kernel/asm-offsets.c @@ -8,9 +8,11 @@ #include <linux/kbuild.h> #include <linux/sched.h> +#include <asm/cputime.h> +#include <asm/timer.h> #include <asm/vdso.h> -#include <asm/sigp.h> #include <asm/pgtable.h> +#include <asm/system.h> /* * Make sure that the compiler is new enough. We want a compiler that @@ -70,15 +72,15 @@ int main(void) DEFINE(__CLOCK_MONOTONIC, CLOCK_MONOTONIC); DEFINE(__CLOCK_REALTIME_RES, MONOTONIC_RES_NSEC); BLANK(); - /* constants for SIGP */ - DEFINE(__SIGP_STOP, sigp_stop); - DEFINE(__SIGP_RESTART, sigp_restart); - DEFINE(__SIGP_SENSE, sigp_sense); - DEFINE(__SIGP_INITIAL_CPU_RESET, sigp_initial_cpu_reset); - BLANK(); + /* idle data offsets */ + DEFINE(__IDLE_ENTER, offsetof(struct s390_idle_data, idle_enter)); + DEFINE(__IDLE_EXIT, offsetof(struct s390_idle_data, idle_exit)); + /* vtimer queue offsets */ + DEFINE(__VQ_IDLE_ENTER, offsetof(struct vtimer_queue, idle_enter)); + DEFINE(__VQ_IDLE_EXIT, offsetof(struct vtimer_queue, idle_exit)); /* lowcore offsets */ DEFINE(__LC_EXT_PARAMS, offsetof(struct _lowcore, ext_params)); - DEFINE(__LC_CPU_ADDRESS, offsetof(struct _lowcore, cpu_addr)); + DEFINE(__LC_EXT_CPU_ADDR, offsetof(struct _lowcore, ext_cpu_addr)); DEFINE(__LC_EXT_INT_CODE, offsetof(struct _lowcore, ext_int_code)); DEFINE(__LC_SVC_ILC, offsetof(struct _lowcore, svc_ilc)); DEFINE(__LC_SVC_INT_CODE, offsetof(struct _lowcore, svc_code)); @@ -95,20 +97,19 @@ int main(void) DEFINE(__LC_IO_INT_WORD, offsetof(struct _lowcore, io_int_word)); DEFINE(__LC_STFL_FAC_LIST, offsetof(struct _lowcore, stfl_fac_list)); DEFINE(__LC_MCCK_CODE, offsetof(struct _lowcore, mcck_interruption_code)); - DEFINE(__LC_DUMP_REIPL, offsetof(struct _lowcore, ipib)); - BLANK(); - DEFINE(__LC_RST_NEW_PSW, offsetof(struct _lowcore, restart_psw)); DEFINE(__LC_RST_OLD_PSW, offsetof(struct _lowcore, restart_old_psw)); DEFINE(__LC_EXT_OLD_PSW, offsetof(struct _lowcore, external_old_psw)); DEFINE(__LC_SVC_OLD_PSW, offsetof(struct _lowcore, svc_old_psw)); DEFINE(__LC_PGM_OLD_PSW, offsetof(struct _lowcore, program_old_psw)); DEFINE(__LC_MCK_OLD_PSW, offsetof(struct _lowcore, mcck_old_psw)); DEFINE(__LC_IO_OLD_PSW, offsetof(struct _lowcore, io_old_psw)); + DEFINE(__LC_RST_NEW_PSW, offsetof(struct _lowcore, restart_psw)); DEFINE(__LC_EXT_NEW_PSW, offsetof(struct _lowcore, external_new_psw)); DEFINE(__LC_SVC_NEW_PSW, offsetof(struct _lowcore, svc_new_psw)); DEFINE(__LC_PGM_NEW_PSW, offsetof(struct _lowcore, program_new_psw)); DEFINE(__LC_MCK_NEW_PSW, offsetof(struct _lowcore, mcck_new_psw)); DEFINE(__LC_IO_NEW_PSW, offsetof(struct _lowcore, io_new_psw)); + BLANK(); DEFINE(__LC_SAVE_AREA_SYNC, offsetof(struct _lowcore, save_area_sync)); DEFINE(__LC_SAVE_AREA_ASYNC, offsetof(struct _lowcore, save_area_async)); DEFINE(__LC_SAVE_AREA_RESTART, offsetof(struct _lowcore, save_area_restart)); @@ -129,12 +130,16 @@ int main(void) DEFINE(__LC_KERNEL_STACK, offsetof(struct _lowcore, kernel_stack)); DEFINE(__LC_ASYNC_STACK, offsetof(struct _lowcore, async_stack)); DEFINE(__LC_PANIC_STACK, offsetof(struct _lowcore, panic_stack)); + DEFINE(__LC_RESTART_STACK, offsetof(struct _lowcore, restart_stack)); + DEFINE(__LC_RESTART_FN, offsetof(struct _lowcore, restart_fn)); DEFINE(__LC_USER_ASCE, offsetof(struct _lowcore, user_asce)); DEFINE(__LC_INT_CLOCK, offsetof(struct _lowcore, int_clock)); DEFINE(__LC_MCCK_CLOCK, offsetof(struct _lowcore, mcck_clock)); DEFINE(__LC_MACHINE_FLAGS, offsetof(struct _lowcore, machine_flags)); DEFINE(__LC_FTRACE_FUNC, offsetof(struct _lowcore, ftrace_func)); DEFINE(__LC_IRB, offsetof(struct _lowcore, irb)); + DEFINE(__LC_DUMP_REIPL, offsetof(struct _lowcore, ipib)); + BLANK(); DEFINE(__LC_CPU_TIMER_SAVE_AREA, offsetof(struct _lowcore, cpu_timer_save_area)); DEFINE(__LC_CLOCK_COMP_SAVE_AREA, offsetof(struct _lowcore, clock_comp_save_area)); DEFINE(__LC_PSW_SAVE_AREA, offsetof(struct _lowcore, psw_save_area)); diff --git a/arch/s390/kernel/compat_signal.c b/arch/s390/kernel/compat_signal.c index 6fe78c2f95d..53a82c8d50e 100644 --- a/arch/s390/kernel/compat_signal.c +++ b/arch/s390/kernel/compat_signal.c @@ -581,7 +581,6 @@ give_sigsegv: int handle_signal32(unsigned long sig, struct k_sigaction *ka, siginfo_t *info, sigset_t *oldset, struct pt_regs *regs) { - sigset_t blocked; int ret; /* Set up the stack frame */ @@ -591,10 +590,7 @@ int handle_signal32(unsigned long sig, struct k_sigaction *ka, ret = setup_frame32(sig, ka, oldset, regs); if (ret) return ret; - sigorsets(&blocked, ¤t->blocked, &ka->sa.sa_mask); - if (!(ka->sa.sa_flags & SA_NODEFER)) - sigaddset(&blocked, sig); - set_current_blocked(&blocked); + block_sigmask(ka, sig); return 0; } diff --git a/arch/s390/kernel/crash_dump.c b/arch/s390/kernel/crash_dump.c index c383ce440d9..cc1172b2687 100644 --- a/arch/s390/kernel/crash_dump.c +++ b/arch/s390/kernel/crash_dump.c @@ -14,6 +14,7 @@ #include <linux/bootmem.h> #include <linux/elf.h> #include <asm/ipl.h> +#include <asm/os_info.h> #define PTR_ADD(x, y) (((char *) (x)) + ((unsigned long) (y))) #define PTR_SUB(x, y) (((char *) (x)) - ((unsigned long) (y))) @@ -51,7 +52,7 @@ ssize_t copy_oldmem_page(unsigned long pfn, char *buf, /* * Copy memory from old kernel */ -static int copy_from_oldmem(void *dest, void *src, size_t count) +int copy_from_oldmem(void *dest, void *src, size_t count) { unsigned long copied = 0; int rc; @@ -224,28 +225,44 @@ static void *nt_prpsinfo(void *ptr) } /* - * Initialize vmcoreinfo note (new kernel) + * Get vmcoreinfo using lowcore->vmcore_info (new kernel) */ -static void *nt_vmcoreinfo(void *ptr) +static void *get_vmcoreinfo_old(unsigned long *size) { char nt_name[11], *vmcoreinfo; Elf64_Nhdr note; void *addr; if (copy_from_oldmem(&addr, &S390_lowcore.vmcore_info, sizeof(addr))) - return ptr; + return NULL; memset(nt_name, 0, sizeof(nt_name)); if (copy_from_oldmem(¬e, addr, sizeof(note))) - return ptr; + return NULL; if (copy_from_oldmem(nt_name, addr + sizeof(note), sizeof(nt_name) - 1)) - return ptr; + return NULL; if (strcmp(nt_name, "VMCOREINFO") != 0) - return ptr; - vmcoreinfo = kzalloc_panic(note.n_descsz + 1); + return NULL; + vmcoreinfo = kzalloc_panic(note.n_descsz); if (copy_from_oldmem(vmcoreinfo, addr + 24, note.n_descsz)) + return NULL; + *size = note.n_descsz; + return vmcoreinfo; +} + +/* + * Initialize vmcoreinfo note (new kernel) + */ +static void *nt_vmcoreinfo(void *ptr) +{ + unsigned long size; + void *vmcoreinfo; + + vmcoreinfo = os_info_old_entry(OS_INFO_VMCOREINFO, &size); + if (!vmcoreinfo) + vmcoreinfo = get_vmcoreinfo_old(&size); + if (!vmcoreinfo) return ptr; - vmcoreinfo[note.n_descsz + 1] = 0; - return nt_init(ptr, 0, vmcoreinfo, note.n_descsz, "VMCOREINFO"); + return nt_init(ptr, 0, vmcoreinfo, size, "VMCOREINFO"); } /* diff --git a/arch/s390/kernel/debug.c b/arch/s390/kernel/debug.c index 6848828b962..19e5e9eba54 100644 --- a/arch/s390/kernel/debug.c +++ b/arch/s390/kernel/debug.c @@ -2,8 +2,8 @@ * arch/s390/kernel/debug.c * S/390 debug facility * - * Copyright (C) 1999, 2000 IBM Deutschland Entwicklung GmbH, - * IBM Corporation + * Copyright IBM Corp. 1999, 2012 + * * Author(s): Michael Holzheu (holzheu@de.ibm.com), * Holger Smolinski (Holger.Smolinski@de.ibm.com) * @@ -167,6 +167,7 @@ static debug_info_t *debug_area_last = NULL; static DEFINE_MUTEX(debug_mutex); static int initialized; +static int debug_critical; static const struct file_operations debug_file_ops = { .owner = THIS_MODULE, @@ -932,6 +933,11 @@ debug_stop_all(void) } +void debug_set_critical(void) +{ + debug_critical = 1; +} + /* * debug_event_common: * - write debug entry with given size @@ -945,7 +951,11 @@ debug_event_common(debug_info_t * id, int level, const void *buf, int len) if (!debug_active || !id->areas) return NULL; - spin_lock_irqsave(&id->lock, flags); + if (debug_critical) { + if (!spin_trylock_irqsave(&id->lock, flags)) + return NULL; + } else + spin_lock_irqsave(&id->lock, flags); active = get_active_entry(id); memset(DEBUG_DATA(active), 0, id->buf_size); memcpy(DEBUG_DATA(active), buf, min(len, id->buf_size)); @@ -968,7 +978,11 @@ debug_entry_t if (!debug_active || !id->areas) return NULL; - spin_lock_irqsave(&id->lock, flags); + if (debug_critical) { + if (!spin_trylock_irqsave(&id->lock, flags)) + return NULL; + } else + spin_lock_irqsave(&id->lock, flags); active = get_active_entry(id); memset(DEBUG_DATA(active), 0, id->buf_size); memcpy(DEBUG_DATA(active), buf, min(len, id->buf_size)); @@ -1013,7 +1027,11 @@ debug_sprintf_event(debug_info_t* id, int level,char *string,...) return NULL; numargs=debug_count_numargs(string); - spin_lock_irqsave(&id->lock, flags); + if (debug_critical) { + if (!spin_trylock_irqsave(&id->lock, flags)) + return NULL; + } else + spin_lock_irqsave(&id->lock, flags); active = get_active_entry(id); curr_event=(debug_sprintf_entry_t *) DEBUG_DATA(active); va_start(ap,string); @@ -1047,7 +1065,11 @@ debug_sprintf_exception(debug_info_t* id, int level,char *string,...) numargs=debug_count_numargs(string); - spin_lock_irqsave(&id->lock, flags); + if (debug_critical) { + if (!spin_trylock_irqsave(&id->lock, flags)) + return NULL; + } else + spin_lock_irqsave(&id->lock, flags); active = get_active_entry(id); curr_event=(debug_sprintf_entry_t *)DEBUG_DATA(active); va_start(ap,string); @@ -1428,10 +1450,10 @@ debug_hex_ascii_format_fn(debug_info_t * id, struct debug_view *view, rc += sprintf(out_buf + rc, "| "); for (i = 0; i < id->buf_size; i++) { unsigned char c = in_buf[i]; - if (!isprint(c)) - rc += sprintf(out_buf + rc, "."); - else + if (isascii(c) && isprint(c)) rc += sprintf(out_buf + rc, "%c", c); + else + rc += sprintf(out_buf + rc, "."); } rc += sprintf(out_buf + rc, "\n"); return rc; diff --git a/arch/s390/kernel/early.c b/arch/s390/kernel/early.c index 52098d6dfaa..578eb4e6d15 100644 --- a/arch/s390/kernel/early.c +++ b/arch/s390/kernel/early.c @@ -29,6 +29,7 @@ #include <asm/sysinfo.h> #include <asm/cpcmd.h> #include <asm/sclp.h> +#include <asm/system.h> #include "entry.h" /* @@ -262,25 +263,8 @@ static noinline __init void setup_lowcore_early(void) static noinline __init void setup_facility_list(void) { - unsigned long nr; - - S390_lowcore.stfl_fac_list = 0; - asm volatile( - " .insn s,0xb2b10000,0(0)\n" /* stfl */ - "0:\n" - EX_TABLE(0b,0b) : "=m" (S390_lowcore.stfl_fac_list)); - memcpy(&S390_lowcore.stfle_fac_list, &S390_lowcore.stfl_fac_list, 4); - nr = 4; /* # bytes stored by stfl */ - if (test_facility(7)) { - /* More facility bits available with stfle */ - register unsigned long reg0 asm("0") = MAX_FACILITY_BIT/64 - 1; - asm volatile(".insn s,0xb2b00000,%0" /* stfle */ - : "=m" (S390_lowcore.stfle_fac_list), "+d" (reg0) - : : "cc"); - nr = (reg0 + 1) * 8; /* # bytes stored by stfle */ - } - memset((char *) S390_lowcore.stfle_fac_list + nr, 0, - MAX_FACILITY_BIT/8 - nr); + stfle(S390_lowcore.stfle_fac_list, + ARRAY_SIZE(S390_lowcore.stfle_fac_list)); } static noinline __init void setup_hpage(void) diff --git a/arch/s390/kernel/entry.S b/arch/s390/kernel/entry.S index 3705700ed37..74ee563fe62 100644 --- a/arch/s390/kernel/entry.S +++ b/arch/s390/kernel/entry.S @@ -2,7 +2,7 @@ * arch/s390/kernel/entry.S * S390 low-level entry points. * - * Copyright (C) IBM Corp. 1999,2006 + * Copyright (C) IBM Corp. 1999,2012 * Author(s): Martin Schwidefsky (schwidefsky@de.ibm.com), * Hartmut Penner (hp@de.ibm.com), * Denis Joseph Barrow (djbarrow@de.ibm.com,barrow_dj@yahoo.com), @@ -105,14 +105,14 @@ STACK_SIZE = 1 << STACK_SHIFT .macro ADD64 high,low,timer al \high,\timer - al \low,\timer+4 + al \low,4+\timer brc 12,.+8 ahi \high,1 .endm .macro SUB64 high,low,timer sl \high,\timer - sl \low,\timer+4 + sl \low,4+\timer brc 3,.+8 ahi \high,-1 .endm @@ -471,7 +471,6 @@ io_tif: jnz io_work # there is work to do (signals etc.) io_restore: mvc __LC_RETURN_PSW(8),__PT_PSW(%r11) - ni __LC_RETURN_PSW+1,0xfd # clean wait state bit stpt __LC_EXIT_TIMER lm %r0,%r15,__PT_R0(%r11) lpsw __LC_RETURN_PSW @@ -606,12 +605,32 @@ ext_skip: stm %r8,%r9,__PT_PSW(%r11) TRACE_IRQS_OFF lr %r2,%r11 # pass pointer to pt_regs - l %r3,__LC_CPU_ADDRESS # get cpu address + interruption code + l %r3,__LC_EXT_CPU_ADDR # get cpu address + interruption code l %r4,__LC_EXT_PARAMS # get external parameters l %r1,BASED(.Ldo_extint) basr %r14,%r1 # call do_extint j io_return +/* + * Load idle PSW. The second "half" of this function is in cleanup_idle. + */ +ENTRY(psw_idle) + st %r4,__SF_EMPTY(%r15) + basr %r1,0 + la %r1,psw_idle_lpsw+4-.(%r1) + st %r1,__SF_EMPTY+4(%r15) + oi __SF_EMPTY+4(%r15),0x80 + la %r1,.Lvtimer_max-psw_idle_lpsw-4(%r1) + stck __IDLE_ENTER(%r2) + ltr %r5,%r5 + stpt __VQ_IDLE_ENTER(%r3) + jz psw_idle_lpsw + spt 0(%r1) +psw_idle_lpsw: + lpsw __SF_EMPTY(%r15) + br %r14 +psw_idle_end: + __critical_end: /* @@ -673,7 +692,6 @@ mcck_skip: TRACE_IRQS_ON mcck_return: mvc __LC_RETURN_MCCK_PSW(8),__PT_PSW(%r11) # move return PSW - ni __LC_RETURN_MCCK_PSW+1,0xfd # clear wait state bit tm __LC_RETURN_MCCK_PSW+1,0x01 # returning to user ? jno 0f lm %r0,%r15,__PT_R0(%r11) @@ -691,77 +709,30 @@ mcck_panic: 0: ahi %r15,-(STACK_FRAME_OVERHEAD + __PT_SIZE) j mcck_skip -/* - * Restart interruption handler, kick starter for additional CPUs - */ -#ifdef CONFIG_SMP - __CPUINIT -ENTRY(restart_int_handler) - basr %r1,0 -restart_base: - spt restart_vtime-restart_base(%r1) - stck __LC_LAST_UPDATE_CLOCK - mvc __LC_LAST_UPDATE_TIMER(8),restart_vtime-restart_base(%r1) - mvc __LC_EXIT_TIMER(8),restart_vtime-restart_base(%r1) - l %r15,__LC_GPREGS_SAVE_AREA+60 # load ksp - lctl %c0,%c15,__LC_CREGS_SAVE_AREA # get new ctl regs - lam %a0,%a15,__LC_AREGS_SAVE_AREA - lm %r6,%r15,__SF_GPRS(%r15)# load registers from clone - l %r1,__LC_THREAD_INFO - mvc __LC_USER_TIMER(8),__TI_user_timer(%r1) - mvc __LC_SYSTEM_TIMER(8),__TI_system_timer(%r1) - xc __LC_STEAL_TIMER(8),__LC_STEAL_TIMER - ssm __LC_PGM_NEW_PSW # turn dat on, keep irqs off - basr %r14,0 - l %r14,restart_addr-.(%r14) - basr %r14,%r14 # call start_secondary -restart_addr: - .long start_secondary - .align 8 -restart_vtime: - .long 0x7fffffff,0xffffffff - .previous -#else -/* - * If we do not run with SMP enabled, let the new CPU crash ... - */ -ENTRY(restart_int_handler) - basr %r1,0 -restart_base: - lpsw restart_crash-restart_base(%r1) - .align 8 -restart_crash: - .long 0x000a0000,0x00000000 -restart_go: -#endif - # # PSW restart interrupt handler # -ENTRY(psw_restart_int_handler) +ENTRY(restart_int_handler) st %r15,__LC_SAVE_AREA_RESTART - basr %r15,0 -0: l %r15,.Lrestart_stack-0b(%r15) # load restart stack - l %r15,0(%r15) + l %r15,__LC_RESTART_STACK ahi %r15,-__PT_SIZE # create pt_regs on stack + xc 0(__PT_SIZE,%r15),0(%r15) stm %r0,%r14,__PT_R0(%r15) mvc __PT_R15(4,%r15),__LC_SAVE_AREA_RESTART mvc __PT_PSW(8,%r15),__LC_RST_OLD_PSW # store restart old psw - ahi %r15,-STACK_FRAME_OVERHEAD - xc __SF_BACKCHAIN(4,%r15),__SF_BACKCHAIN(%r15) - basr %r14,0 -1: l %r14,.Ldo_restart-1b(%r14) - basr %r14,%r14 - basr %r14,0 # load disabled wait PSW if -2: lpsw restart_psw_crash-2b(%r14) # do_restart returns - .align 4 -.Ldo_restart: - .long do_restart -.Lrestart_stack: - .long restart_stack - .align 8 -restart_psw_crash: - .long 0x000a0000,0x00000000 + restart_psw_crash + ahi %r15,-STACK_FRAME_OVERHEAD # create stack frame on stack + xc 0(STACK_FRAME_OVERHEAD,%r15),0(%r15) + lm %r1,%r3,__LC_RESTART_FN # load fn, parm & source cpu + ltr %r3,%r3 # test source cpu address + jm 1f # negative -> skip source stop +0: sigp %r4,%r3,1 # sigp sense to source cpu + brc 10,0b # wait for status stored +1: basr %r14,%r1 # call function + stap __SF_EMPTY(%r15) # store cpu address + lh %r3,__SF_EMPTY(%r15) +2: sigp %r4,%r3,5 # sigp stop to current cpu + brc 2,2b +3: j 3b .section .kprobes.text, "ax" @@ -795,6 +766,8 @@ cleanup_table: .long io_tif + 0x80000000 .long io_restore + 0x80000000 .long io_done + 0x80000000 + .long psw_idle + 0x80000000 + .long psw_idle_end + 0x80000000 cleanup_critical: cl %r9,BASED(cleanup_table) # system_call @@ -813,6 +786,10 @@ cleanup_critical: jl cleanup_io_tif cl %r9,BASED(cleanup_table+28) # io_done jl cleanup_io_restore + cl %r9,BASED(cleanup_table+32) # psw_idle + jl 0f + cl %r9,BASED(cleanup_table+36) # psw_idle_end + jl cleanup_idle 0: br %r14 cleanup_system_call: @@ -896,7 +873,6 @@ cleanup_io_restore: jhe 0f l %r9,12(%r11) # get saved r11 pointer to pt_regs mvc __LC_RETURN_PSW(8),__PT_PSW(%r9) - ni __LC_RETURN_PSW+1,0xfd # clear wait state bit mvc 0(32,%r11),__PT_R8(%r9) lm %r0,%r7,__PT_R0(%r9) 0: lm %r8,%r9,__LC_RETURN_PSW @@ -904,11 +880,52 @@ cleanup_io_restore: cleanup_io_restore_insn: .long io_done - 4 + 0x80000000 +cleanup_idle: + # copy interrupt clock & cpu timer + mvc __IDLE_EXIT(8,%r2),__LC_INT_CLOCK + mvc __VQ_IDLE_EXIT(8,%r3),__LC_ASYNC_ENTER_TIMER + chi %r11,__LC_SAVE_AREA_ASYNC + je 0f + mvc __IDLE_EXIT(8,%r2),__LC_MCCK_CLOCK + mvc __VQ_IDLE_EXIT(8,%r3),__LC_MCCK_ENTER_TIMER +0: # check if stck has been executed + cl %r9,BASED(cleanup_idle_insn) + jhe 1f + mvc __IDLE_ENTER(8,%r2),__IDLE_EXIT(%r2) + mvc __VQ_IDLE_ENTER(8,%r3),__VQ_IDLE_EXIT(%r3) + j 2f +1: # check if the cpu timer has been reprogrammed + ltr %r5,%r5 + jz 2f + spt __VQ_IDLE_ENTER(%r3) +2: # account system time going idle + lm %r9,%r10,__LC_STEAL_TIMER + ADD64 %r9,%r10,__IDLE_ENTER(%r2) + SUB64 %r9,%r10,__LC_LAST_UPDATE_CLOCK + stm %r9,%r10,__LC_STEAL_TIMER + mvc __LC_LAST_UPDATE_CLOCK(8),__IDLE_EXIT(%r2) + lm %r9,%r10,__LC_SYSTEM_TIMER + ADD64 %r9,%r10,__LC_LAST_UPDATE_TIMER + SUB64 %r9,%r10,__VQ_IDLE_ENTER(%r3) + stm %r9,%r10,__LC_SYSTEM_TIMER + mvc __LC_LAST_UPDATE_TIMER(8),__VQ_IDLE_EXIT(%r3) + # prepare return psw + n %r8,BASED(cleanup_idle_wait) # clear wait state bit + l %r9,24(%r11) # return from psw_idle + br %r14 +cleanup_idle_insn: + .long psw_idle_lpsw + 0x80000000 +cleanup_idle_wait: + .long 0xfffdffff + /* * Integer constants */ .align 4 -.Lnr_syscalls: .long NR_syscalls +.Lnr_syscalls: + .long NR_syscalls +.Lvtimer_max: + .quad 0x7fffffffffffffff /* * Symbol constants diff --git a/arch/s390/kernel/entry.h b/arch/s390/kernel/entry.h index bf538aaf407..6cdddac93a2 100644 --- a/arch/s390/kernel/entry.h +++ b/arch/s390/kernel/entry.h @@ -4,11 +4,22 @@ #include <linux/types.h> #include <linux/signal.h> #include <asm/ptrace.h> - +#include <asm/cputime.h> +#include <asm/timer.h> extern void (*pgm_check_table[128])(struct pt_regs *); extern void *restart_stack; +void system_call(void); +void pgm_check_handler(void); +void ext_int_handler(void); +void io_int_handler(void); +void mcck_int_handler(void); +void restart_int_handler(void); +void restart_call_handler(void); +void psw_idle(struct s390_idle_data *, struct vtimer_queue *, + unsigned long, int); + asmlinkage long do_syscall_trace_enter(struct pt_regs *regs); asmlinkage void do_syscall_trace_exit(struct pt_regs *regs); @@ -24,9 +35,9 @@ int handle_signal32(unsigned long sig, struct k_sigaction *ka, siginfo_t *info, sigset_t *oldset, struct pt_regs *regs); void do_notify_resume(struct pt_regs *regs); -void do_extint(struct pt_regs *regs, unsigned int, unsigned int, unsigned long); +struct ext_code; +void do_extint(struct pt_regs *regs, struct ext_code, unsigned int, unsigned long); void do_restart(void); -int __cpuinit start_secondary(void *cpuvoid); void __init startup_init(void); void die(struct pt_regs *regs, const char *str); diff --git a/arch/s390/kernel/entry64.S b/arch/s390/kernel/entry64.S index 412a7b8783d..4e1c292fa7e 100644 --- a/arch/s390/kernel/entry64.S +++ b/arch/s390/kernel/entry64.S @@ -2,7 +2,7 @@ * arch/s390/kernel/entry64.S * S390 low-level entry points. * - * Copyright (C) IBM Corp. 1999,2010 + * Copyright (C) IBM Corp. 1999,2012 * Author(s): Martin Schwidefsky (schwidefsky@de.ibm.com), * Hartmut Penner (hp@de.ibm.com), * Denis Joseph Barrow (djbarrow@de.ibm.com,barrow_dj@yahoo.com), @@ -489,7 +489,6 @@ io_restore: lg %r14,__LC_VDSO_PER_CPU lmg %r0,%r10,__PT_R0(%r11) mvc __LC_RETURN_PSW(16),__PT_PSW(%r11) - ni __LC_RETURN_PSW+1,0xfd # clear wait state bit stpt __LC_EXIT_TIMER mvc __VDSO_ECTG_BASE(16,%r14),__LC_EXIT_TIMER lmg %r11,%r15,__PT_R11(%r11) @@ -625,12 +624,30 @@ ext_skip: TRACE_IRQS_OFF lghi %r1,4096 lgr %r2,%r11 # pass pointer to pt_regs - llgf %r3,__LC_CPU_ADDRESS # get cpu address + interruption code + llgf %r3,__LC_EXT_CPU_ADDR # get cpu address + interruption code llgf %r4,__LC_EXT_PARAMS # get external parameter lg %r5,__LC_EXT_PARAMS2-4096(%r1) # get 64 bit external parameter brasl %r14,do_extint j io_return +/* + * Load idle PSW. The second "half" of this function is in cleanup_idle. + */ +ENTRY(psw_idle) + stg %r4,__SF_EMPTY(%r15) + larl %r1,psw_idle_lpsw+4 + stg %r1,__SF_EMPTY+8(%r15) + larl %r1,.Lvtimer_max + stck __IDLE_ENTER(%r2) + ltr %r5,%r5 + stpt __VQ_IDLE_ENTER(%r3) + jz psw_idle_lpsw + spt 0(%r1) +psw_idle_lpsw: + lpswe __SF_EMPTY(%r15) + br %r14 +psw_idle_end: + __critical_end: /* @@ -696,7 +713,6 @@ mcck_return: lg %r14,__LC_VDSO_PER_CPU lmg %r0,%r10,__PT_R0(%r11) mvc __LC_RETURN_MCCK_PSW(16),__PT_PSW(%r11) # move return PSW - ni __LC_RETURN_MCCK_PSW+1,0xfd # clear wait state bit tm __LC_RETURN_MCCK_PSW+1,0x01 # returning to user ? jno 0f stpt __LC_EXIT_TIMER @@ -713,68 +729,30 @@ mcck_panic: 0: aghi %r15,-(STACK_FRAME_OVERHEAD + __PT_SIZE) j mcck_skip -/* - * Restart interruption handler, kick starter for additional CPUs - */ -#ifdef CONFIG_SMP - __CPUINIT -ENTRY(restart_int_handler) - basr %r1,0 -restart_base: - spt restart_vtime-restart_base(%r1) - stck __LC_LAST_UPDATE_CLOCK - mvc __LC_LAST_UPDATE_TIMER(8),restart_vtime-restart_base(%r1) - mvc __LC_EXIT_TIMER(8),restart_vtime-restart_base(%r1) - lghi %r10,__LC_GPREGS_SAVE_AREA - lg %r15,120(%r10) # load ksp - lghi %r10,__LC_CREGS_SAVE_AREA - lctlg %c0,%c15,0(%r10) # get new ctl regs - lghi %r10,__LC_AREGS_SAVE_AREA - lam %a0,%a15,0(%r10) - lmg %r6,%r15,__SF_GPRS(%r15)# load registers from clone - lg %r1,__LC_THREAD_INFO - mvc __LC_USER_TIMER(8),__TI_user_timer(%r1) - mvc __LC_SYSTEM_TIMER(8),__TI_system_timer(%r1) - xc __LC_STEAL_TIMER(8),__LC_STEAL_TIMER - ssm __LC_PGM_NEW_PSW # turn dat on, keep irqs off - brasl %r14,start_secondary - .align 8 -restart_vtime: - .long 0x7fffffff,0xffffffff - .previous -#else -/* - * If we do not run with SMP enabled, let the new CPU crash ... - */ -ENTRY(restart_int_handler) - basr %r1,0 -restart_base: - lpswe restart_crash-restart_base(%r1) - .align 8 -restart_crash: - .long 0x000a0000,0x00000000,0x00000000,0x00000000 -restart_go: -#endif - # # PSW restart interrupt handler # -ENTRY(psw_restart_int_handler) +ENTRY(restart_int_handler) stg %r15,__LC_SAVE_AREA_RESTART - larl %r15,restart_stack # load restart stack - lg %r15,0(%r15) + lg %r15,__LC_RESTART_STACK aghi %r15,-__PT_SIZE # create pt_regs on stack + xc 0(__PT_SIZE,%r15),0(%r15) stmg %r0,%r14,__PT_R0(%r15) mvc __PT_R15(8,%r15),__LC_SAVE_AREA_RESTART mvc __PT_PSW(16,%r15),__LC_RST_OLD_PSW # store restart old psw - aghi %r15,-STACK_FRAME_OVERHEAD - xc __SF_BACKCHAIN(8,%r15),__SF_BACKCHAIN(%r15) - brasl %r14,do_restart - larl %r14,restart_psw_crash # load disabled wait PSW if - lpswe 0(%r14) # do_restart returns - .align 8 -restart_psw_crash: - .quad 0x0002000080000000,0x0000000000000000 + restart_psw_crash + aghi %r15,-STACK_FRAME_OVERHEAD # create stack frame on stack + xc 0(STACK_FRAME_OVERHEAD,%r15),0(%r15) + lmg %r1,%r3,__LC_RESTART_FN # load fn, parm & source cpu + ltgr %r3,%r3 # test source cpu address + jm 1f # negative -> skip source stop +0: sigp %r4,%r3,1 # sigp sense to source cpu + brc 10,0b # wait for status stored +1: basr %r14,%r1 # call function + stap __SF_EMPTY(%r15) # store cpu address + llgh %r3,__SF_EMPTY(%r15) +2: sigp %r4,%r3,5 # sigp stop to current cpu + brc 2,2b +3: j 3b .section .kprobes.text, "ax" @@ -808,6 +786,8 @@ cleanup_table: .quad io_tif .quad io_restore .quad io_done + .quad psw_idle + .quad psw_idle_end cleanup_critical: clg %r9,BASED(cleanup_table) # system_call @@ -826,6 +806,10 @@ cleanup_critical: jl cleanup_io_tif clg %r9,BASED(cleanup_table+56) # io_done jl cleanup_io_restore + clg %r9,BASED(cleanup_table+64) # psw_idle + jl 0f + clg %r9,BASED(cleanup_table+72) # psw_idle_end + jl cleanup_idle 0: br %r14 @@ -915,7 +899,6 @@ cleanup_io_restore: je 0f lg %r9,24(%r11) # get saved r11 pointer to pt_regs mvc __LC_RETURN_PSW(16),__PT_PSW(%r9) - ni __LC_RETURN_PSW+1,0xfd # clear wait state bit mvc 0(64,%r11),__PT_R8(%r9) lmg %r0,%r7,__PT_R0(%r9) 0: lmg %r8,%r9,__LC_RETURN_PSW @@ -923,6 +906,42 @@ cleanup_io_restore: cleanup_io_restore_insn: .quad io_done - 4 +cleanup_idle: + # copy interrupt clock & cpu timer + mvc __IDLE_EXIT(8,%r2),__LC_INT_CLOCK + mvc __VQ_IDLE_EXIT(8,%r3),__LC_ASYNC_ENTER_TIMER + cghi %r11,__LC_SAVE_AREA_ASYNC + je 0f + mvc __IDLE_EXIT(8,%r2),__LC_MCCK_CLOCK + mvc __VQ_IDLE_EXIT(8,%r3),__LC_MCCK_ENTER_TIMER +0: # check if stck & stpt have been executed + clg %r9,BASED(cleanup_idle_insn) + jhe 1f + mvc __IDLE_ENTER(8,%r2),__IDLE_EXIT(%r2) + mvc __VQ_IDLE_ENTER(8,%r3),__VQ_IDLE_EXIT(%r3) + j 2f +1: # check if the cpu timer has been reprogrammed + ltr %r5,%r5 + jz 2f + spt __VQ_IDLE_ENTER(%r3) +2: # account system time going idle + lg %r9,__LC_STEAL_TIMER + alg %r9,__IDLE_ENTER(%r2) + slg %r9,__LC_LAST_UPDATE_CLOCK + stg %r9,__LC_STEAL_TIMER + mvc __LC_LAST_UPDATE_CLOCK(8),__IDLE_EXIT(%r2) + lg %r9,__LC_SYSTEM_TIMER + alg %r9,__LC_LAST_UPDATE_TIMER + slg %r9,__VQ_IDLE_ENTER(%r3) + stg %r9,__LC_SYSTEM_TIMER + mvc __LC_LAST_UPDATE_TIMER(8),__VQ_IDLE_EXIT(%r3) + # prepare return psw + nihh %r8,0xfffd # clear wait state bit + lg %r9,48(%r11) # return from psw_idle + br %r14 +cleanup_idle_insn: + .quad psw_idle_lpsw + /* * Integer constants */ @@ -931,6 +950,8 @@ cleanup_io_restore_insn: .quad __critical_start .Lcritical_length: .quad __critical_end - __critical_start +.Lvtimer_max: + .quad 0x7fffffffffffffff #if defined(CONFIG_KVM) || defined(CONFIG_KVM_MODULE) diff --git a/arch/s390/kernel/ipl.c b/arch/s390/kernel/ipl.c index affa8e68124..8342e65a140 100644 --- a/arch/s390/kernel/ipl.c +++ b/arch/s390/kernel/ipl.c @@ -2,7 +2,7 @@ * arch/s390/kernel/ipl.c * ipl/reipl/dump support for Linux on s390. * - * Copyright IBM Corp. 2005,2007 + * Copyright IBM Corp. 2005,2012 * Author(s): Michael Holzheu <holzheu@de.ibm.com> * Heiko Carstens <heiko.carstens@de.ibm.com> * Volker Sameske <sameske@de.ibm.com> @@ -17,6 +17,7 @@ #include <linux/fs.h> #include <linux/gfp.h> #include <linux/crash_dump.h> +#include <linux/debug_locks.h> #include <asm/ipl.h> #include <asm/smp.h> #include <asm/setup.h> @@ -25,8 +26,9 @@ #include <asm/ebcdic.h> #include <asm/reset.h> #include <asm/sclp.h> -#include <asm/sigp.h> #include <asm/checksum.h> +#include <asm/debug.h> +#include <asm/os_info.h> #include "entry.h" #define IPL_PARM_BLOCK_VERSION 0 @@ -571,7 +573,7 @@ static void __ipl_run(void *unused) static void ipl_run(struct shutdown_trigger *trigger) { - smp_switch_to_ipl_cpu(__ipl_run, NULL); + smp_call_ipl_cpu(__ipl_run, NULL); } static int __init ipl_init(void) @@ -950,6 +952,13 @@ static struct attribute_group reipl_nss_attr_group = { .attrs = reipl_nss_attrs, }; +static void set_reipl_block_actual(struct ipl_parameter_block *reipl_block) +{ + reipl_block_actual = reipl_block; + os_info_entry_add(OS_INFO_REIPL_BLOCK, reipl_block_actual, + reipl_block->hdr.len); +} + /* reipl type */ static int reipl_set_type(enum ipl_type type) @@ -965,7 +974,7 @@ static int reipl_set_type(enum ipl_type type) reipl_method = REIPL_METHOD_CCW_VM; else reipl_method = REIPL_METHOD_CCW_CIO; - reipl_block_actual = reipl_block_ccw; + set_reipl_block_actual(reipl_block_ccw); break; case IPL_TYPE_FCP: if (diag308_set_works) @@ -974,7 +983,7 @@ static int reipl_set_type(enum ipl_type type) reipl_method = REIPL_METHOD_FCP_RO_VM; else reipl_method = REIPL_METHOD_FCP_RO_DIAG; - reipl_block_actual = reipl_block_fcp; + set_reipl_block_actual(reipl_block_fcp); break; case IPL_TYPE_FCP_DUMP: reipl_method = REIPL_METHOD_FCP_DUMP; @@ -984,7 +993,7 @@ static int reipl_set_type(enum ipl_type type) reipl_method = REIPL_METHOD_NSS_DIAG; else reipl_method = REIPL_METHOD_NSS; - reipl_block_actual = reipl_block_nss; + set_reipl_block_actual(reipl_block_nss); break; case IPL_TYPE_UNKNOWN: reipl_method = REIPL_METHOD_DEFAULT; @@ -1101,7 +1110,7 @@ static void __reipl_run(void *unused) static void reipl_run(struct shutdown_trigger *trigger) { - smp_switch_to_ipl_cpu(__reipl_run, NULL); + smp_call_ipl_cpu(__reipl_run, NULL); } static void reipl_block_ccw_init(struct ipl_parameter_block *ipb) @@ -1256,6 +1265,29 @@ static int __init reipl_fcp_init(void) return 0; } +static int __init reipl_type_init(void) +{ + enum ipl_type reipl_type = ipl_info.type; + struct ipl_parameter_block *reipl_block; + unsigned long size; + + reipl_block = os_info_old_entry(OS_INFO_REIPL_BLOCK, &size); + if (!reipl_block) + goto out; + /* + * If we have an OS info reipl block, this will be used + */ + if (reipl_block->hdr.pbt == DIAG308_IPL_TYPE_FCP) { + memcpy(reipl_block_fcp, reipl_block, size); + reipl_type = IPL_TYPE_FCP; + } else if (reipl_block->hdr.pbt == DIAG308_IPL_TYPE_CCW) { + memcpy(reipl_block_ccw, reipl_block, size); + reipl_type = IPL_TYPE_CCW; + } +out: + return reipl_set_type(reipl_type); +} + static int __init reipl_init(void) { int rc; @@ -1277,10 +1309,7 @@ static int __init reipl_init(void) rc = reipl_nss_init(); if (rc) return rc; - rc = reipl_set_type(ipl_info.type); - if (rc) - return rc; - return 0; + return reipl_type_init(); } static struct shutdown_action __refdata reipl_action = { @@ -1421,7 +1450,7 @@ static void dump_run(struct shutdown_trigger *trigger) if (dump_method == DUMP_METHOD_NONE) return; smp_send_stop(); - smp_switch_to_ipl_cpu(__dump_run, NULL); + smp_call_ipl_cpu(__dump_run, NULL); } static int __init dump_ccw_init(void) @@ -1499,30 +1528,12 @@ static struct shutdown_action __refdata dump_action = { static void dump_reipl_run(struct shutdown_trigger *trigger) { - preempt_disable(); - /* - * Bypass dynamic address translation (DAT) when storing IPL parameter - * information block address and checksum into the prefix area - * (corresponding to absolute addresses 0-8191). - * When enhanced DAT applies and the STE format control in one, - * the absolute address is formed without prefixing. In this case a - * normal store (stg/st) into the prefix area would no more match to - * absolute addresses 0-8191. - */ -#ifdef CONFIG_64BIT - asm volatile("sturg %0,%1" - :: "a" ((unsigned long) reipl_block_actual), - "a" (&lowcore_ptr[smp_processor_id()]->ipib)); -#else - asm volatile("stura %0,%1" - :: "a" ((unsigned long) reipl_block_actual), - "a" (&lowcore_ptr[smp_processor_id()]->ipib)); -#endif - asm volatile("stura %0,%1" - :: "a" (csum_partial(reipl_block_actual, - reipl_block_actual->hdr.len, 0)), - "a" (&lowcore_ptr[smp_processor_id()]->ipib_checksum)); - preempt_enable(); + u32 csum; + + csum = csum_partial(reipl_block_actual, reipl_block_actual->hdr.len, 0); + copy_to_absolute_zero(&S390_lowcore.ipib_checksum, &csum, sizeof(csum)); + copy_to_absolute_zero(&S390_lowcore.ipib, &reipl_block_actual, + sizeof(reipl_block_actual)); dump_run(trigger); } @@ -1623,9 +1634,7 @@ static void stop_run(struct shutdown_trigger *trigger) if (strcmp(trigger->name, ON_PANIC_STR) == 0 || strcmp(trigger->name, ON_RESTART_STR) == 0) disabled_wait((unsigned long) __builtin_return_address(0)); - while (sigp(smp_processor_id(), sigp_stop) == sigp_busy) - cpu_relax(); - for (;;); + smp_stop_cpu(); } static struct shutdown_action stop_action = {SHUTDOWN_ACTION_STOP_STR, @@ -1713,6 +1722,7 @@ static struct kobj_attribute on_panic_attr = static void do_panic(void) { + lgr_info_log(); on_panic_trigger.action->fn(&on_panic_trigger); stop_run(&on_panic_trigger); } @@ -1738,9 +1748,8 @@ static ssize_t on_restart_store(struct kobject *kobj, static struct kobj_attribute on_restart_attr = __ATTR(on_restart, 0644, on_restart_show, on_restart_store); -void do_restart(void) +static void __do_restart(void *ignore) { - smp_restart_with_online_cpu(); smp_send_stop(); #ifdef CONFIG_CRASH_DUMP crash_kexec(NULL); @@ -1749,6 +1758,14 @@ void do_restart(void) stop_run(&on_restart_trigger); } +void do_restart(void) +{ + tracing_off(); + debug_locks_off(); + lgr_info_log(); + smp_call_online_cpu(__do_restart, NULL); +} + /* on halt */ static struct shutdown_trigger on_halt_trigger = {ON_HALT_STR, &stop_action}; diff --git a/arch/s390/kernel/irq.c b/arch/s390/kernel/irq.c index b9a7fdd9c81..1c2cdd59ccd 100644 --- a/arch/s390/kernel/irq.c +++ b/arch/s390/kernel/irq.c @@ -165,13 +165,6 @@ static inline int ext_hash(u16 code) return (code + (code >> 9)) & 0xff; } -static void ext_int_hash_update(struct rcu_head *head) -{ - struct ext_int_info *p = container_of(head, struct ext_int_info, rcu); - - kfree(p); -} - int register_external_interrupt(u16 code, ext_int_handler_t handler) { struct ext_int_info *p; @@ -202,38 +195,34 @@ int unregister_external_interrupt(u16 code, ext_int_handler_t handler) list_for_each_entry_rcu(p, &ext_int_hash[index], entry) if (p->code == code && p->handler == handler) { list_del_rcu(&p->entry); - call_rcu(&p->rcu, ext_int_hash_update); + kfree_rcu(p, rcu); } spin_unlock_irqrestore(&ext_int_hash_lock, flags); return 0; } EXPORT_SYMBOL(unregister_external_interrupt); -void __irq_entry do_extint(struct pt_regs *regs, unsigned int ext_int_code, +void __irq_entry do_extint(struct pt_regs *regs, struct ext_code ext_code, unsigned int param32, unsigned long param64) { struct pt_regs *old_regs; - unsigned short code; struct ext_int_info *p; int index; - code = (unsigned short) ext_int_code; old_regs = set_irq_regs(regs); - s390_idle_check(regs, S390_lowcore.int_clock, - S390_lowcore.async_enter_timer); irq_enter(); if (S390_lowcore.int_clock >= S390_lowcore.clock_comparator) /* Serve timer interrupts first. */ clock_comparator_work(); kstat_cpu(smp_processor_id()).irqs[EXTERNAL_INTERRUPT]++; - if (code != 0x1004) + if (ext_code.code != 0x1004) __get_cpu_var(s390_idle).nohz_delay = 1; - index = ext_hash(code); + index = ext_hash(ext_code.code); rcu_read_lock(); list_for_each_entry_rcu(p, &ext_int_hash[index], entry) - if (likely(p->code == code)) - p->handler(ext_int_code, param32, param64); + if (likely(p->code == ext_code.code)) + p->handler(ext_code, param32, param64); rcu_read_unlock(); irq_exit(); set_irq_regs(old_regs); @@ -266,3 +255,26 @@ void service_subclass_irq_unregister(void) spin_unlock(&sc_irq_lock); } EXPORT_SYMBOL(service_subclass_irq_unregister); + +static DEFINE_SPINLOCK(ma_subclass_lock); +static int ma_subclass_refcount; + +void measurement_alert_subclass_register(void) +{ + spin_lock(&ma_subclass_lock); + if (!ma_subclass_refcount) + ctl_set_bit(0, 5); + ma_subclass_refcount++; + spin_unlock(&ma_subclass_lock); +} +EXPORT_SYMBOL(measurement_alert_subclass_register); + +void measurement_alert_subclass_unregister(void) +{ + spin_lock(&ma_subclass_lock); + ma_subclass_refcount--; + if (!ma_subclass_refcount) + ctl_clear_bit(0, 5); + spin_unlock(&ma_subclass_lock); +} +EXPORT_SYMBOL(measurement_alert_subclass_unregister); diff --git a/arch/s390/kernel/lgr.c b/arch/s390/kernel/lgr.c new file mode 100644 index 00000000000..8431b92ca3a --- /dev/null +++ b/arch/s390/kernel/lgr.c @@ -0,0 +1,200 @@ +/* + * Linux Guest Relocation (LGR) detection + * + * Copyright IBM Corp. 2012 + * Author(s): Michael Holzheu <holzheu@linux.vnet.ibm.com> + */ + +#include <linux/module.h> +#include <linux/timer.h> +#include <linux/slab.h> +#include <asm/sysinfo.h> +#include <asm/ebcdic.h> +#include <asm/system.h> +#include <asm/debug.h> +#include <asm/ipl.h> + +#define LGR_TIMER_INTERVAL_SECS (30 * 60) +#define VM_LEVEL_MAX 2 /* Maximum is 8, but we only record two levels */ + +/* + * LGR info: Contains stfle and stsi data + */ +struct lgr_info { + /* Bit field with facility information: 4 DWORDs are stored */ + u64 stfle_fac_list[4]; + /* Level of system (1 = CEC, 2 = LPAR, 3 = z/VM */ + u32 level; + /* Level 1: CEC info (stsi 1.1.1) */ + char manufacturer[16]; + char type[4]; + char sequence[16]; + char plant[4]; + char model[16]; + /* Level 2: LPAR info (stsi 2.2.2) */ + u16 lpar_number; + char name[8]; + /* Level 3: VM info (stsi 3.2.2) */ + u8 vm_count; + struct { + char name[8]; + char cpi[16]; + } vm[VM_LEVEL_MAX]; +} __packed __aligned(8); + +/* + * LGR globals + */ +static void *lgr_page; +static struct lgr_info lgr_info_last; +static struct lgr_info lgr_info_cur; +static struct debug_info *lgr_dbf; + +/* + * Return number of valid stsi levels + */ +static inline int stsi_0(void) +{ + int rc = stsi(NULL, 0, 0, 0); + + return rc == -ENOSYS ? rc : (((unsigned int) rc) >> 28); +} + +/* + * Copy buffer and then convert it to ASCII + */ +static void cpascii(char *dst, char *src, int size) +{ + memcpy(dst, src, size); + EBCASC(dst, size); +} + +/* + * Fill LGR info with 1.1.1 stsi data + */ +static void lgr_stsi_1_1_1(struct lgr_info *lgr_info) +{ + struct sysinfo_1_1_1 *si = lgr_page; + + if (stsi(si, 1, 1, 1) == -ENOSYS) + return; + cpascii(lgr_info->manufacturer, si->manufacturer, + sizeof(si->manufacturer)); + cpascii(lgr_info->type, si->type, sizeof(si->type)); + cpascii(lgr_info->model, si->model, sizeof(si->model)); + cpascii(lgr_info->sequence, si->sequence, sizeof(si->sequence)); + cpascii(lgr_info->plant, si->plant, sizeof(si->plant)); +} + +/* + * Fill LGR info with 2.2.2 stsi data + */ +static void lgr_stsi_2_2_2(struct lgr_info *lgr_info) +{ + struct sysinfo_2_2_2 *si = lgr_page; + + if (stsi(si, 2, 2, 2) == -ENOSYS) + return; + cpascii(lgr_info->name, si->name, sizeof(si->name)); + memcpy(&lgr_info->lpar_number, &si->lpar_number, + sizeof(lgr_info->lpar_number)); +} + +/* + * Fill LGR info with 3.2.2 stsi data + */ +static void lgr_stsi_3_2_2(struct lgr_info *lgr_info) +{ + struct sysinfo_3_2_2 *si = lgr_page; + int i; + + if (stsi(si, 3, 2, 2) == -ENOSYS) + return; + for (i = 0; i < min_t(u8, si->count, VM_LEVEL_MAX); i++) { + cpascii(lgr_info->vm[i].name, si->vm[i].name, + sizeof(si->vm[i].name)); + cpascii(lgr_info->vm[i].cpi, si->vm[i].cpi, + sizeof(si->vm[i].cpi)); + } + lgr_info->vm_count = si->count; +} + +/* + * Fill LGR info with current data + */ +static void lgr_info_get(struct lgr_info *lgr_info) +{ + memset(lgr_info, 0, sizeof(*lgr_info)); + stfle(lgr_info->stfle_fac_list, ARRAY_SIZE(lgr_info->stfle_fac_list)); + lgr_info->level = stsi_0(); + if (lgr_info->level == -ENOSYS) + return; + if (lgr_info->level >= 1) + lgr_stsi_1_1_1(lgr_info); + if (lgr_info->level >= 2) + lgr_stsi_2_2_2(lgr_info); + if (lgr_info->level >= 3) + lgr_stsi_3_2_2(lgr_info); +} + +/* + * Check if LGR info has changed and if yes log new LGR info to s390dbf + */ +void lgr_info_log(void) +{ + static DEFINE_SPINLOCK(lgr_info_lock); + unsigned long flags; + + if (!spin_trylock_irqsave(&lgr_info_lock, flags)) + return; + lgr_info_get(&lgr_info_cur); + if (memcmp(&lgr_info_last, &lgr_info_cur, sizeof(lgr_info_cur)) != 0) { + debug_event(lgr_dbf, 1, &lgr_info_cur, sizeof(lgr_info_cur)); + lgr_info_last = lgr_info_cur; + } + spin_unlock_irqrestore(&lgr_info_lock, flags); +} +EXPORT_SYMBOL_GPL(lgr_info_log); + +static void lgr_timer_set(void); + +/* + * LGR timer callback + */ +static void lgr_timer_fn(unsigned long ignored) +{ + lgr_info_log(); + lgr_timer_set(); +} + +static struct timer_list lgr_timer = + TIMER_DEFERRED_INITIALIZER(lgr_timer_fn, 0, 0); + +/* + * Setup next LGR timer + */ +static void lgr_timer_set(void) +{ + mod_timer(&lgr_timer, jiffies + LGR_TIMER_INTERVAL_SECS * HZ); +} + +/* + * Initialize LGR: Add s390dbf, write initial lgr_info and setup timer + */ +static int __init lgr_init(void) +{ + lgr_page = (void *) __get_free_pages(GFP_KERNEL, 0); + if (!lgr_page) + return -ENOMEM; + lgr_dbf = debug_register("lgr", 1, 1, sizeof(struct lgr_info)); + if (!lgr_dbf) { + free_page((unsigned long) lgr_page); + return -ENOMEM; + } + debug_register_view(lgr_dbf, &debug_hex_ascii_view); + lgr_info_get(&lgr_info_last); + debug_event(lgr_dbf, 1, &lgr_info_last, sizeof(lgr_info_last)); + lgr_timer_set(); + return 0; +} +module_init(lgr_init); diff --git a/arch/s390/kernel/machine_kexec.c b/arch/s390/kernel/machine_kexec.c index 47b168fb29c..0f8cdf1268d 100644 --- a/arch/s390/kernel/machine_kexec.c +++ b/arch/s390/kernel/machine_kexec.c @@ -14,6 +14,7 @@ #include <linux/delay.h> #include <linux/reboot.h> #include <linux/ftrace.h> +#include <linux/debug_locks.h> #include <asm/cio.h> #include <asm/setup.h> #include <asm/pgtable.h> @@ -49,50 +50,21 @@ static void add_elf_notes(int cpu) } /* - * Store status of next available physical CPU - */ -static int store_status_next(int start_cpu, int this_cpu) -{ - struct save_area *sa = (void *) 4608 + store_prefix(); - int cpu, rc; - - for (cpu = start_cpu; cpu < 65536; cpu++) { - if (cpu == this_cpu) - continue; - do { - rc = raw_sigp(cpu, sigp_stop_and_store_status); - } while (rc == sigp_busy); - if (rc != sigp_order_code_accepted) - continue; - if (sa->pref_reg) - return cpu; - } - return -1; -} - -/* * Initialize CPU ELF notes */ void setup_regs(void) { unsigned long sa = S390_lowcore.prefixreg_save_area + SAVE_AREA_BASE; - int cpu, this_cpu, phys_cpu = 0, first = 1; + int cpu, this_cpu; - this_cpu = stap(); - - if (!S390_lowcore.prefixreg_save_area) - first = 0; + this_cpu = smp_find_processor_id(stap()); + add_elf_notes(this_cpu); for_each_online_cpu(cpu) { - if (first) { - add_elf_notes(cpu); - first = 0; + if (cpu == this_cpu) + continue; + if (smp_store_status(cpu)) continue; - } - phys_cpu = store_status_next(phys_cpu, this_cpu); - if (phys_cpu == -1) - break; add_elf_notes(cpu); - phys_cpu++; } /* Copy dump CPU store status info to absolute zero */ memcpy((void *) SAVE_AREA_BASE, (void *) sa, sizeof(struct save_area)); @@ -238,10 +210,14 @@ static void __machine_kexec(void *data) struct kimage *image = data; pfault_fini(); - if (image->type == KEXEC_TYPE_CRASH) + tracing_off(); + debug_locks_off(); + if (image->type == KEXEC_TYPE_CRASH) { + lgr_info_log(); s390_reset_system(__do_machine_kdump, data); - else + } else { s390_reset_system(__do_machine_kexec, data); + } disabled_wait((unsigned long) __builtin_return_address(0)); } @@ -255,5 +231,5 @@ void machine_kexec(struct kimage *image) return; tracer_disable(); smp_send_stop(); - smp_switch_to_ipl_cpu(__machine_kexec, image); + smp_call_ipl_cpu(__machine_kexec, image); } diff --git a/arch/s390/kernel/nmi.c b/arch/s390/kernel/nmi.c index 0fd2e863e11..8c372ca6135 100644 --- a/arch/s390/kernel/nmi.c +++ b/arch/s390/kernel/nmi.c @@ -254,8 +254,6 @@ void notrace s390_do_machine_check(struct pt_regs *regs) int umode; nmi_enter(); - s390_idle_check(regs, S390_lowcore.mcck_clock, - S390_lowcore.mcck_enter_timer); kstat_cpu(smp_processor_id()).irqs[NMI_NMI]++; mci = (struct mci *) &S390_lowcore.mcck_interruption_code; mcck = &__get_cpu_var(cpu_mcck); diff --git a/arch/s390/kernel/os_info.c b/arch/s390/kernel/os_info.c new file mode 100644 index 00000000000..bbe522672e0 --- /dev/null +++ b/arch/s390/kernel/os_info.c @@ -0,0 +1,169 @@ +/* + * OS info memory interface + * + * Copyright IBM Corp. 2012 + * Author(s): Michael Holzheu <holzheu@linux.vnet.ibm.com> + */ + +#define KMSG_COMPONENT "os_info" +#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt + +#include <linux/crash_dump.h> +#include <linux/kernel.h> +#include <asm/checksum.h> +#include <asm/lowcore.h> +#include <asm/system.h> +#include <asm/os_info.h> + +/* + * OS info structure has to be page aligned + */ +static struct os_info os_info __page_aligned_data; + +/* + * Compute checksum over OS info structure + */ +u32 os_info_csum(struct os_info *os_info) +{ + int size = sizeof(*os_info) - offsetof(struct os_info, version_major); + return csum_partial(&os_info->version_major, size, 0); +} + +/* + * Add crashkernel info to OS info and update checksum + */ +void os_info_crashkernel_add(unsigned long base, unsigned long size) +{ + os_info.crashkernel_addr = (u64)(unsigned long)base; + os_info.crashkernel_size = (u64)(unsigned long)size; + os_info.csum = os_info_csum(&os_info); +} + +/* + * Add OS info entry and update checksum + */ +void os_info_entry_add(int nr, void *ptr, u64 size) +{ + os_info.entry[nr].addr = (u64)(unsigned long)ptr; + os_info.entry[nr].size = size; + os_info.entry[nr].csum = csum_partial(ptr, size, 0); + os_info.csum = os_info_csum(&os_info); +} + +/* + * Initialize OS info struture and set lowcore pointer + */ +void __init os_info_init(void) +{ + void *ptr = &os_info; + + os_info.version_major = OS_INFO_VERSION_MAJOR; + os_info.version_minor = OS_INFO_VERSION_MINOR; + os_info.magic = OS_INFO_MAGIC; + os_info.csum = os_info_csum(&os_info); + copy_to_absolute_zero(&S390_lowcore.os_info, &ptr, sizeof(ptr)); +} + +#ifdef CONFIG_CRASH_DUMP + +static struct os_info *os_info_old; + +/* + * Allocate and copy OS info entry from oldmem + */ +static void os_info_old_alloc(int nr, int align) +{ + unsigned long addr, size = 0; + char *buf, *buf_align, *msg; + u32 csum; + + addr = os_info_old->entry[nr].addr; + if (!addr) { + msg = "not available"; + goto fail; + } + size = os_info_old->entry[nr].size; + buf = kmalloc(size + align - 1, GFP_KERNEL); + if (!buf) { + msg = "alloc failed"; + goto fail; + } + buf_align = PTR_ALIGN(buf, align); + if (copy_from_oldmem(buf_align, (void *) addr, size)) { + msg = "copy failed"; + goto fail_free; + } + csum = csum_partial(buf_align, size, 0); + if (csum != os_info_old->entry[nr].csum) { + msg = "checksum failed"; + goto fail_free; + } + os_info_old->entry[nr].addr = (u64)(unsigned long)buf_align; + msg = "copied"; + goto out; +fail_free: + kfree(buf); +fail: + os_info_old->entry[nr].addr = 0; +out: + pr_info("entry %i: %s (addr=0x%lx size=%lu)\n", + nr, msg, addr, size); +} + +/* + * Initialize os info and os info entries from oldmem + */ +static void os_info_old_init(void) +{ + static int os_info_init; + unsigned long addr; + + if (os_info_init) + return; + if (!OLDMEM_BASE) + goto fail; + if (copy_from_oldmem(&addr, &S390_lowcore.os_info, sizeof(addr))) + goto fail; + if (addr == 0 || addr % PAGE_SIZE) + goto fail; + os_info_old = kzalloc(sizeof(*os_info_old), GFP_KERNEL); + if (!os_info_old) + goto fail; + if (copy_from_oldmem(os_info_old, (void *) addr, sizeof(*os_info_old))) + goto fail_free; + if (os_info_old->magic != OS_INFO_MAGIC) + goto fail_free; + if (os_info_old->csum != os_info_csum(os_info_old)) + goto fail_free; + if (os_info_old->version_major > OS_INFO_VERSION_MAJOR) + goto fail_free; + os_info_old_alloc(OS_INFO_VMCOREINFO, 1); + os_info_old_alloc(OS_INFO_REIPL_BLOCK, 1); + os_info_old_alloc(OS_INFO_INIT_FN, PAGE_SIZE); + pr_info("crashkernel: addr=0x%lx size=%lu\n", + (unsigned long) os_info_old->crashkernel_addr, + (unsigned long) os_info_old->crashkernel_size); + os_info_init = 1; + return; +fail_free: + kfree(os_info_old); +fail: + os_info_init = 1; + os_info_old = NULL; +} + +/* + * Return pointer to os infor entry and its size + */ +void *os_info_old_entry(int nr, unsigned long *size) +{ + os_info_old_init(); + + if (!os_info_old) + return NULL; + if (!os_info_old->entry[nr].addr) + return NULL; + *size = (unsigned long) os_info_old->entry[nr].size; + return (void *)(unsigned long)os_info_old->entry[nr].addr; +} +#endif diff --git a/arch/s390/kernel/perf_cpum_cf.c b/arch/s390/kernel/perf_cpum_cf.c new file mode 100644 index 00000000000..8481ecf2ad7 --- /dev/null +++ b/arch/s390/kernel/perf_cpum_cf.c @@ -0,0 +1,690 @@ +/* + * Performance event support for s390x - CPU-measurement Counter Facility + * + * Copyright IBM Corp. 2012 + * Author(s): Hendrik Brueckner <brueckner@linux.vnet.ibm.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License (version 2 only) + * as published by the Free Software Foundation. + */ +#define KMSG_COMPONENT "cpum_cf" +#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt + +#include <linux/kernel.h> +#include <linux/kernel_stat.h> +#include <linux/perf_event.h> +#include <linux/percpu.h> +#include <linux/notifier.h> +#include <linux/init.h> +#include <linux/export.h> +#include <asm/system.h> +#include <asm/irq.h> +#include <asm/cpu_mf.h> + +/* CPU-measurement counter facility supports these CPU counter sets: + * For CPU counter sets: + * Basic counter set: 0-31 + * Problem-state counter set: 32-63 + * Crypto-activity counter set: 64-127 + * Extented counter set: 128-159 + */ +enum cpumf_ctr_set { + /* CPU counter sets */ + CPUMF_CTR_SET_BASIC = 0, + CPUMF_CTR_SET_USER = 1, + CPUMF_CTR_SET_CRYPTO = 2, + CPUMF_CTR_SET_EXT = 3, + + /* Maximum number of counter sets */ + CPUMF_CTR_SET_MAX, +}; + +#define CPUMF_LCCTL_ENABLE_SHIFT 16 +#define CPUMF_LCCTL_ACTCTL_SHIFT 0 +static const u64 cpumf_state_ctl[CPUMF_CTR_SET_MAX] = { + [CPUMF_CTR_SET_BASIC] = 0x02, + [CPUMF_CTR_SET_USER] = 0x04, + [CPUMF_CTR_SET_CRYPTO] = 0x08, + [CPUMF_CTR_SET_EXT] = 0x01, +}; + +static void ctr_set_enable(u64 *state, int ctr_set) +{ + *state |= cpumf_state_ctl[ctr_set] << CPUMF_LCCTL_ENABLE_SHIFT; +} +static void ctr_set_disable(u64 *state, int ctr_set) +{ + *state &= ~(cpumf_state_ctl[ctr_set] << CPUMF_LCCTL_ENABLE_SHIFT); +} +static void ctr_set_start(u64 *state, int ctr_set) +{ + *state |= cpumf_state_ctl[ctr_set] << CPUMF_LCCTL_ACTCTL_SHIFT; +} +static void ctr_set_stop(u64 *state, int ctr_set) +{ + *state &= ~(cpumf_state_ctl[ctr_set] << CPUMF_LCCTL_ACTCTL_SHIFT); +} + +/* Local CPUMF event structure */ +struct cpu_hw_events { + struct cpumf_ctr_info info; + atomic_t ctr_set[CPUMF_CTR_SET_MAX]; + u64 state, tx_state; + unsigned int flags; +}; +static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = { + .ctr_set = { + [CPUMF_CTR_SET_BASIC] = ATOMIC_INIT(0), + [CPUMF_CTR_SET_USER] = ATOMIC_INIT(0), + [CPUMF_CTR_SET_CRYPTO] = ATOMIC_INIT(0), + [CPUMF_CTR_SET_EXT] = ATOMIC_INIT(0), + }, + .state = 0, + .flags = 0, +}; + +static int get_counter_set(u64 event) +{ + int set = -1; + + if (event < 32) + set = CPUMF_CTR_SET_BASIC; + else if (event < 64) + set = CPUMF_CTR_SET_USER; + else if (event < 128) + set = CPUMF_CTR_SET_CRYPTO; + else if (event < 160) + set = CPUMF_CTR_SET_EXT; + + return set; +} + +static int validate_event(const struct hw_perf_event *hwc) +{ + switch (hwc->config_base) { + case CPUMF_CTR_SET_BASIC: + case CPUMF_CTR_SET_USER: + case CPUMF_CTR_SET_CRYPTO: + case CPUMF_CTR_SET_EXT: + /* check for reserved counters */ + if ((hwc->config >= 6 && hwc->config <= 31) || + (hwc->config >= 38 && hwc->config <= 63) || + (hwc->config >= 80 && hwc->config <= 127)) + return -EOPNOTSUPP; + break; + default: + return -EINVAL; + } + + return 0; +} + +static int validate_ctr_version(const struct hw_perf_event *hwc) +{ + struct cpu_hw_events *cpuhw; + int err = 0; + + cpuhw = &get_cpu_var(cpu_hw_events); + + /* check required version for counter sets */ + switch (hwc->config_base) { + case CPUMF_CTR_SET_BASIC: + case CPUMF_CTR_SET_USER: + if (cpuhw->info.cfvn < 1) + err = -EOPNOTSUPP; + break; + case CPUMF_CTR_SET_CRYPTO: + case CPUMF_CTR_SET_EXT: + if (cpuhw->info.csvn < 1) + err = -EOPNOTSUPP; + break; + } + + put_cpu_var(cpu_hw_events); + return err; +} + +static int validate_ctr_auth(const struct hw_perf_event *hwc) +{ + struct cpu_hw_events *cpuhw; + u64 ctrs_state; + int err = 0; + + cpuhw = &get_cpu_var(cpu_hw_events); + + /* check authorization for cpu counter sets */ + ctrs_state = cpumf_state_ctl[hwc->config_base]; + if (!(ctrs_state & cpuhw->info.auth_ctl)) + err = -EPERM; + + put_cpu_var(cpu_hw_events); + return err; +} + +/* + * Change the CPUMF state to active. + * Enable and activate the CPU-counter sets according + * to the per-cpu control state. + */ +static void cpumf_pmu_enable(struct pmu *pmu) +{ + struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events); + int err; + + if (cpuhw->flags & PMU_F_ENABLED) + return; + + err = lcctl(cpuhw->state); + if (err) { + pr_err("Enabling the performance measuring unit " + "failed with rc=%lx\n", err); + return; + } + + cpuhw->flags |= PMU_F_ENABLED; +} + +/* + * Change the CPUMF state to inactive. + * Disable and enable (inactive) the CPU-counter sets according + * to the per-cpu control state. + */ +static void cpumf_pmu_disable(struct pmu *pmu) +{ + struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events); + int err; + u64 inactive; + + if (!(cpuhw->flags & PMU_F_ENABLED)) + return; + + inactive = cpuhw->state & ~((1 << CPUMF_LCCTL_ENABLE_SHIFT) - 1); + err = lcctl(inactive); + if (err) { + pr_err("Disabling the performance measuring unit " + "failed with rc=%lx\n", err); + return; + } + + cpuhw->flags &= ~PMU_F_ENABLED; +} + + +/* Number of perf events counting hardware events */ +static atomic_t num_events = ATOMIC_INIT(0); +/* Used to avoid races in calling reserve/release_cpumf_hardware */ +static DEFINE_MUTEX(pmc_reserve_mutex); + +/* CPU-measurement alerts for the counter facility */ +static void cpumf_measurement_alert(struct ext_code ext_code, + unsigned int alert, unsigned long unused) +{ + struct cpu_hw_events *cpuhw; + + if (!(alert & CPU_MF_INT_CF_MASK)) + return; + + kstat_cpu(smp_processor_id()).irqs[EXTINT_CPM]++; + cpuhw = &__get_cpu_var(cpu_hw_events); + + /* Measurement alerts are shared and might happen when the PMU + * is not reserved. Ignore these alerts in this case. */ + if (!(cpuhw->flags & PMU_F_RESERVED)) + return; + + /* counter authorization change alert */ + if (alert & CPU_MF_INT_CF_CACA) + qctri(&cpuhw->info); + + /* loss of counter data alert */ + if (alert & CPU_MF_INT_CF_LCDA) + pr_err("CPU[%i] Counter data was lost\n", smp_processor_id()); +} + +#define PMC_INIT 0 +#define PMC_RELEASE 1 +static void setup_pmc_cpu(void *flags) +{ + struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events); + + switch (*((int *) flags)) { + case PMC_INIT: + memset(&cpuhw->info, 0, sizeof(cpuhw->info)); + qctri(&cpuhw->info); + cpuhw->flags |= PMU_F_RESERVED; + break; + + case PMC_RELEASE: + cpuhw->flags &= ~PMU_F_RESERVED; + break; + } + + /* Disable CPU counter sets */ + lcctl(0); +} + +/* Initialize the CPU-measurement facility */ +static int reserve_pmc_hardware(void) +{ + int flags = PMC_INIT; + + on_each_cpu(setup_pmc_cpu, &flags, 1); + measurement_alert_subclass_register(); + + return 0; +} + +/* Release the CPU-measurement facility */ +static void release_pmc_hardware(void) +{ + int flags = PMC_RELEASE; + + on_each_cpu(setup_pmc_cpu, &flags, 1); + measurement_alert_subclass_unregister(); +} + +/* Release the PMU if event is the last perf event */ +static void hw_perf_event_destroy(struct perf_event *event) +{ + if (!atomic_add_unless(&num_events, -1, 1)) { + mutex_lock(&pmc_reserve_mutex); + if (atomic_dec_return(&num_events) == 0) + release_pmc_hardware(); + mutex_unlock(&pmc_reserve_mutex); + } +} + +/* CPUMF <-> perf event mappings for kernel+userspace (basic set) */ +static const int cpumf_generic_events_basic[] = { + [PERF_COUNT_HW_CPU_CYCLES] = 0, + [PERF_COUNT_HW_INSTRUCTIONS] = 1, + [PERF_COUNT_HW_CACHE_REFERENCES] = -1, + [PERF_COUNT_HW_CACHE_MISSES] = -1, + [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = -1, + [PERF_COUNT_HW_BRANCH_MISSES] = -1, + [PERF_COUNT_HW_BUS_CYCLES] = -1, +}; +/* CPUMF <-> perf event mappings for userspace (problem-state set) */ +static const int cpumf_generic_events_user[] = { + [PERF_COUNT_HW_CPU_CYCLES] = 32, + [PERF_COUNT_HW_INSTRUCTIONS] = 33, + [PERF_COUNT_HW_CACHE_REFERENCES] = -1, + [PERF_COUNT_HW_CACHE_MISSES] = -1, + [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = -1, + [PERF_COUNT_HW_BRANCH_MISSES] = -1, + [PERF_COUNT_HW_BUS_CYCLES] = -1, +}; + +static int __hw_perf_event_init(struct perf_event *event) +{ + struct perf_event_attr *attr = &event->attr; + struct hw_perf_event *hwc = &event->hw; + int err; + u64 ev; + + switch (attr->type) { + case PERF_TYPE_RAW: + /* Raw events are used to access counters directly, + * hence do not permit excludes */ + if (attr->exclude_kernel || attr->exclude_user || + attr->exclude_hv) + return -EOPNOTSUPP; + ev = attr->config; + break; + + case PERF_TYPE_HARDWARE: + ev = attr->config; + /* Count user space (problem-state) only */ + if (!attr->exclude_user && attr->exclude_kernel) { + if (ev >= ARRAY_SIZE(cpumf_generic_events_user)) + return -EOPNOTSUPP; + ev = cpumf_generic_events_user[ev]; + + /* No support for kernel space counters only */ + } else if (!attr->exclude_kernel && attr->exclude_user) { + return -EOPNOTSUPP; + + /* Count user and kernel space */ + } else { + if (ev >= ARRAY_SIZE(cpumf_generic_events_basic)) + return -EOPNOTSUPP; + ev = cpumf_generic_events_basic[ev]; + } + break; + + default: + return -ENOENT; + } + + if (ev == -1) + return -ENOENT; + + if (ev >= PERF_CPUM_CF_MAX_CTR) + return -EINVAL; + + /* The CPU measurement counter facility does not have any interrupts + * to do sampling. Sampling must be provided by external means, + * for example, by timers. + */ + if (hwc->sample_period) + return -EINVAL; + + /* Use the hardware perf event structure to store the counter number + * in 'config' member and the counter set to which the counter belongs + * in the 'config_base'. The counter set (config_base) is then used + * to enable/disable the counters. + */ + hwc->config = ev; + hwc->config_base = get_counter_set(ev); + + /* Validate the counter that is assigned to this event. + * Because the counter facility can use numerous counters at the + * same time without constraints, it is not necessary to explicity + * validate event groups (event->group_leader != event). + */ + err = validate_event(hwc); + if (err) + return err; + + /* Initialize for using the CPU-measurement counter facility */ + if (!atomic_inc_not_zero(&num_events)) { + mutex_lock(&pmc_reserve_mutex); + if (atomic_read(&num_events) == 0 && reserve_pmc_hardware()) + err = -EBUSY; + else + atomic_inc(&num_events); + mutex_unlock(&pmc_reserve_mutex); + } + event->destroy = hw_perf_event_destroy; + + /* Finally, validate version and authorization of the counter set */ + err = validate_ctr_auth(hwc); + if (!err) + err = validate_ctr_version(hwc); + + return err; +} + +static int cpumf_pmu_event_init(struct perf_event *event) +{ + int err; + + switch (event->attr.type) { + case PERF_TYPE_HARDWARE: + case PERF_TYPE_HW_CACHE: + case PERF_TYPE_RAW: + err = __hw_perf_event_init(event); + break; + default: + return -ENOENT; + } + + if (unlikely(err) && event->destroy) + event->destroy(event); + + return err; +} + +static int hw_perf_event_reset(struct perf_event *event) +{ + u64 prev, new; + int err; + + do { + prev = local64_read(&event->hw.prev_count); + err = ecctr(event->hw.config, &new); + if (err) { + if (err != 3) + break; + /* The counter is not (yet) available. This + * might happen if the counter set to which + * this counter belongs is in the disabled + * state. + */ + new = 0; + } + } while (local64_cmpxchg(&event->hw.prev_count, prev, new) != prev); + + return err; +} + +static int hw_perf_event_update(struct perf_event *event) +{ + u64 prev, new, delta; + int err; + + do { + prev = local64_read(&event->hw.prev_count); + err = ecctr(event->hw.config, &new); + if (err) + goto out; + } while (local64_cmpxchg(&event->hw.prev_count, prev, new) != prev); + + delta = (prev <= new) ? new - prev + : (-1ULL - prev) + new + 1; /* overflow */ + local64_add(delta, &event->count); +out: + return err; +} + +static void cpumf_pmu_read(struct perf_event *event) +{ + if (event->hw.state & PERF_HES_STOPPED) + return; + + hw_perf_event_update(event); +} + +static void cpumf_pmu_start(struct perf_event *event, int flags) +{ + struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events); + struct hw_perf_event *hwc = &event->hw; + + if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED))) + return; + + if (WARN_ON_ONCE(hwc->config == -1)) + return; + + if (flags & PERF_EF_RELOAD) + WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE)); + + hwc->state = 0; + + /* (Re-)enable and activate the counter set */ + ctr_set_enable(&cpuhw->state, hwc->config_base); + ctr_set_start(&cpuhw->state, hwc->config_base); + + /* The counter set to which this counter belongs can be already active. + * Because all counters in a set are active, the event->hw.prev_count + * needs to be synchronized. At this point, the counter set can be in + * the inactive or disabled state. + */ + hw_perf_event_reset(event); + + /* increment refcount for this counter set */ + atomic_inc(&cpuhw->ctr_set[hwc->config_base]); +} + +static void cpumf_pmu_stop(struct perf_event *event, int flags) +{ + struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events); + struct hw_perf_event *hwc = &event->hw; + + if (!(hwc->state & PERF_HES_STOPPED)) { + /* Decrement reference count for this counter set and if this + * is the last used counter in the set, clear activation + * control and set the counter set state to inactive. + */ + if (!atomic_dec_return(&cpuhw->ctr_set[hwc->config_base])) + ctr_set_stop(&cpuhw->state, hwc->config_base); + event->hw.state |= PERF_HES_STOPPED; + } + + if ((flags & PERF_EF_UPDATE) && !(hwc->state & PERF_HES_UPTODATE)) { + hw_perf_event_update(event); + event->hw.state |= PERF_HES_UPTODATE; + } +} + +static int cpumf_pmu_add(struct perf_event *event, int flags) +{ + struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events); + + /* Check authorization for the counter set to which this + * counter belongs. + * For group events transaction, the authorization check is + * done in cpumf_pmu_commit_txn(). + */ + if (!(cpuhw->flags & PERF_EVENT_TXN)) + if (validate_ctr_auth(&event->hw)) + return -EPERM; + + ctr_set_enable(&cpuhw->state, event->hw.config_base); + event->hw.state = PERF_HES_UPTODATE | PERF_HES_STOPPED; + + if (flags & PERF_EF_START) + cpumf_pmu_start(event, PERF_EF_RELOAD); + + perf_event_update_userpage(event); + + return 0; +} + +static void cpumf_pmu_del(struct perf_event *event, int flags) +{ + struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events); + + cpumf_pmu_stop(event, PERF_EF_UPDATE); + + /* Check if any counter in the counter set is still used. If not used, + * change the counter set to the disabled state. This also clears the + * content of all counters in the set. + * + * When a new perf event has been added but not yet started, this can + * clear enable control and resets all counters in a set. Therefore, + * cpumf_pmu_start() always has to reenable a counter set. + */ + if (!atomic_read(&cpuhw->ctr_set[event->hw.config_base])) + ctr_set_disable(&cpuhw->state, event->hw.config_base); + + perf_event_update_userpage(event); +} + +/* + * Start group events scheduling transaction. + * Set flags to perform a single test at commit time. + */ +static void cpumf_pmu_start_txn(struct pmu *pmu) +{ + struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events); + + perf_pmu_disable(pmu); + cpuhw->flags |= PERF_EVENT_TXN; + cpuhw->tx_state = cpuhw->state; +} + +/* + * Stop and cancel a group events scheduling tranctions. + * Assumes cpumf_pmu_del() is called for each successful added + * cpumf_pmu_add() during the transaction. + */ +static void cpumf_pmu_cancel_txn(struct pmu *pmu) +{ + struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events); + + WARN_ON(cpuhw->tx_state != cpuhw->state); + + cpuhw->flags &= ~PERF_EVENT_TXN; + perf_pmu_enable(pmu); +} + +/* + * Commit the group events scheduling transaction. On success, the + * transaction is closed. On error, the transaction is kept open + * until cpumf_pmu_cancel_txn() is called. + */ +static int cpumf_pmu_commit_txn(struct pmu *pmu) +{ + struct cpu_hw_events *cpuhw = &__get_cpu_var(cpu_hw_events); + u64 state; + + /* check if the updated state can be scheduled */ + state = cpuhw->state & ~((1 << CPUMF_LCCTL_ENABLE_SHIFT) - 1); + state >>= CPUMF_LCCTL_ENABLE_SHIFT; + if ((state & cpuhw->info.auth_ctl) != state) + return -EPERM; + + cpuhw->flags &= ~PERF_EVENT_TXN; + perf_pmu_enable(pmu); + return 0; +} + +/* Performance monitoring unit for s390x */ +static struct pmu cpumf_pmu = { + .pmu_enable = cpumf_pmu_enable, + .pmu_disable = cpumf_pmu_disable, + .event_init = cpumf_pmu_event_init, + .add = cpumf_pmu_add, + .del = cpumf_pmu_del, + .start = cpumf_pmu_start, + .stop = cpumf_pmu_stop, + .read = cpumf_pmu_read, + .start_txn = cpumf_pmu_start_txn, + .commit_txn = cpumf_pmu_commit_txn, + .cancel_txn = cpumf_pmu_cancel_txn, +}; + +static int __cpuinit cpumf_pmu_notifier(struct notifier_block *self, + unsigned long action, void *hcpu) +{ + unsigned int cpu = (long) hcpu; + int flags; + + switch (action & ~CPU_TASKS_FROZEN) { + case CPU_ONLINE: + flags = PMC_INIT; + smp_call_function_single(cpu, setup_pmc_cpu, &flags, 1); + break; + case CPU_DOWN_PREPARE: + flags = PMC_RELEASE; + smp_call_function_single(cpu, setup_pmc_cpu, &flags, 1); + break; + default: + break; + } + + return NOTIFY_OK; +} + +static int __init cpumf_pmu_init(void) +{ + int rc; + + if (!cpum_cf_avail()) + return -ENODEV; + + /* clear bit 15 of cr0 to unauthorize problem-state to + * extract measurement counters */ + ctl_clear_bit(0, 48); + + /* register handler for measurement-alert interruptions */ + rc = register_external_interrupt(0x1407, cpumf_measurement_alert); + if (rc) { + pr_err("Registering for CPU-measurement alerts " + "failed with rc=%i\n", rc); + goto out; + } + + rc = perf_pmu_register(&cpumf_pmu, "cpum_cf", PERF_TYPE_RAW); + if (rc) { + pr_err("Registering the cpum_cf PMU failed with rc=%i\n", rc); + unregister_external_interrupt(0x1407, cpumf_measurement_alert); + goto out; + } + perf_cpu_notifier(cpumf_pmu_notifier); +out: + return rc; +} +early_initcall(cpumf_pmu_init); diff --git a/arch/s390/kernel/perf_event.c b/arch/s390/kernel/perf_event.c new file mode 100644 index 00000000000..609f985198c --- /dev/null +++ b/arch/s390/kernel/perf_event.c @@ -0,0 +1,125 @@ +/* + * Performance event support for s390x + * + * Copyright IBM Corp. 2012 + * Author(s): Hendrik Brueckner <brueckner@linux.vnet.ibm.com> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License (version 2 only) + * as published by the Free Software Foundation. + */ +#define KMSG_COMPONENT "perf" +#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt + +#include <linux/kernel.h> +#include <linux/perf_event.h> +#include <linux/percpu.h> +#include <linux/export.h> +#include <asm/system.h> +#include <asm/irq.h> +#include <asm/cpu_mf.h> +#include <asm/lowcore.h> +#include <asm/processor.h> + +const char *perf_pmu_name(void) +{ + if (cpum_cf_avail() || cpum_sf_avail()) + return "CPU-measurement facilities (CPUMF)"; + return "pmu"; +} +EXPORT_SYMBOL(perf_pmu_name); + +int perf_num_counters(void) +{ + int num = 0; + + if (cpum_cf_avail()) + num += PERF_CPUM_CF_MAX_CTR; + + return num; +} +EXPORT_SYMBOL(perf_num_counters); + +void perf_event_print_debug(void) +{ + struct cpumf_ctr_info cf_info; + unsigned long flags; + int cpu; + + if (!cpum_cf_avail()) + return; + + local_irq_save(flags); + + cpu = smp_processor_id(); + memset(&cf_info, 0, sizeof(cf_info)); + if (!qctri(&cf_info)) { + pr_info("CPU[%i] CPUM_CF: ver=%u.%u A=%04x E=%04x C=%04x\n", + cpu, cf_info.cfvn, cf_info.csvn, + cf_info.auth_ctl, cf_info.enable_ctl, cf_info.act_ctl); + print_hex_dump_bytes("CPUMF Query: ", DUMP_PREFIX_OFFSET, + &cf_info, sizeof(cf_info)); + } + + local_irq_restore(flags); +} + +/* See also arch/s390/kernel/traps.c */ +static unsigned long __store_trace(struct perf_callchain_entry *entry, + unsigned long sp, + unsigned long low, unsigned long high) +{ + struct stack_frame *sf; + struct pt_regs *regs; + + while (1) { + sp = sp & PSW_ADDR_INSN; + if (sp < low || sp > high - sizeof(*sf)) + return sp; + sf = (struct stack_frame *) sp; + perf_callchain_store(entry, sf->gprs[8] & PSW_ADDR_INSN); + /* Follow the backchain. */ + while (1) { + low = sp; + sp = sf->back_chain & PSW_ADDR_INSN; + if (!sp) + break; + if (sp <= low || sp > high - sizeof(*sf)) + return sp; + sf = (struct stack_frame *) sp; + perf_callchain_store(entry, + sf->gprs[8] & PSW_ADDR_INSN); + } + /* Zero backchain detected, check for interrupt frame. */ + sp = (unsigned long) (sf + 1); + if (sp <= low || sp > high - sizeof(*regs)) + return sp; + regs = (struct pt_regs *) sp; + perf_callchain_store(entry, sf->gprs[8] & PSW_ADDR_INSN); + low = sp; + sp = regs->gprs[15]; + } +} + +void perf_callchain_kernel(struct perf_callchain_entry *entry, + struct pt_regs *regs) +{ + unsigned long head; + struct stack_frame *head_sf; + + if (user_mode(regs)) + return; + + head = regs->gprs[15]; + head_sf = (struct stack_frame *) head; + + if (!head_sf || !head_sf->back_chain) + return; + + head = head_sf->back_chain; + head = __store_trace(entry, head, S390_lowcore.async_stack - ASYNC_SIZE, + S390_lowcore.async_stack); + + __store_trace(entry, head, S390_lowcore.thread_info, + S390_lowcore.thread_info + THREAD_SIZE); +} diff --git a/arch/s390/kernel/process.c b/arch/s390/kernel/process.c index e795933eb2c..3732e4c09cb 100644 --- a/arch/s390/kernel/process.c +++ b/arch/s390/kernel/process.c @@ -77,13 +77,8 @@ static void default_idle(void) local_irq_enable(); return; } - trace_hardirqs_on(); - /* Don't trace preempt off for idle. */ - stop_critical_timings(); - /* Stop virtual timer and halt the cpu. */ + /* Halt the cpu and keep track of cpu time accounting. */ vtime_stop_cpu(); - /* Reenable preemption tracer. */ - start_critical_timings(); } void cpu_idle(void) @@ -97,9 +92,7 @@ void cpu_idle(void) tick_nohz_idle_exit(); if (test_thread_flag(TIF_MCCK_PENDING)) s390_handle_mcck(); - preempt_enable_no_resched(); - schedule(); - preempt_disable(); + schedule_preempt_disabled(); } } diff --git a/arch/s390/kernel/setup.c b/arch/s390/kernel/setup.c index 3b2efc81f34..38e751278bf 100644 --- a/arch/s390/kernel/setup.c +++ b/arch/s390/kernel/setup.c @@ -2,7 +2,7 @@ * arch/s390/kernel/setup.c * * S390 version - * Copyright (C) IBM Corp. 1999,2010 + * Copyright (C) IBM Corp. 1999,2012 * Author(s): Hartmut Penner (hp@de.ibm.com), * Martin Schwidefsky (schwidefsky@de.ibm.com) * @@ -62,6 +62,8 @@ #include <asm/ebcdic.h> #include <asm/kvm_virtio.h> #include <asm/diag.h> +#include <asm/os_info.h> +#include "entry.h" long psw_kernel_bits = PSW_DEFAULT_KEY | PSW_MASK_BASE | PSW_ASC_PRIMARY | PSW_MASK_EA | PSW_MASK_BA; @@ -351,8 +353,9 @@ static void setup_addressing_mode(void) } } -static void __init -setup_lowcore(void) +void *restart_stack __attribute__((__section__(".data"))); + +static void __init setup_lowcore(void) { struct _lowcore *lc; @@ -363,7 +366,7 @@ setup_lowcore(void) 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) psw_restart_int_handler; + PSW_ADDR_AMODE | (unsigned long) restart_int_handler; lc->external_new_psw.mask = psw_kernel_bits | PSW_MASK_DAT | PSW_MASK_MCHECK; lc->external_new_psw.addr = @@ -412,6 +415,24 @@ 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 necesary 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; + memcpy(&S390_lowcore.restart_stack, &lc->restart_stack, + 4*sizeof(unsigned long)); + copy_to_absolute_zero(&S390_lowcore.restart_psw, + &lc->restart_psw, sizeof(psw_t)); + set_prefix((u32)(unsigned long) lc); lowcore_ptr[0] = lc; } @@ -572,27 +593,6 @@ static void __init setup_memory_end(void) } } -void *restart_stack __attribute__((__section__(".data"))); - -/* - * Setup new PSW and allocate stack for PSW restart interrupt - */ -static void __init setup_restart_psw(void) -{ - psw_t psw; - - restart_stack = __alloc_bootmem(ASYNC_SIZE, ASYNC_SIZE, 0); - restart_stack += ASYNC_SIZE; - - /* - * Setup restart PSW for absolute zero lowcore. This is necesary - * if PSW restart is done on an offline CPU that has lowcore zero - */ - psw.mask = PSW_DEFAULT_KEY | PSW_MASK_BASE | PSW_MASK_EA | PSW_MASK_BA; - psw.addr = PSW_ADDR_AMODE | (unsigned long) psw_restart_int_handler; - copy_to_absolute_zero(&S390_lowcore.restart_psw, &psw, sizeof(psw)); -} - static void __init setup_vmcoreinfo(void) { #ifdef CONFIG_KEXEC @@ -747,7 +747,7 @@ static void __init reserve_crashkernel(void) { #ifdef CONFIG_CRASH_DUMP unsigned long long crash_base, crash_size; - char *msg; + char *msg = NULL; int rc; rc = parse_crashkernel(boot_command_line, memory_end, &crash_size, @@ -779,11 +779,11 @@ static void __init reserve_crashkernel(void) pr_info("Reserving %lluMB of memory at %lluMB " "for crashkernel (System RAM: %luMB)\n", crash_size >> 20, crash_base >> 20, memory_end >> 20); + os_info_crashkernel_add(crash_base, crash_size); #endif } -static void __init -setup_memory(void) +static void __init setup_memory(void) { unsigned long bootmap_size; unsigned long start_pfn, end_pfn; @@ -1014,8 +1014,7 @@ 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 @@ -1060,6 +1059,7 @@ setup_arch(char **cmdline_p) parse_early_param(); + os_info_init(); setup_ipl(); setup_memory_end(); setup_addressing_mode(); @@ -1068,7 +1068,6 @@ setup_arch(char **cmdline_p) setup_memory(); setup_resources(); setup_vmcoreinfo(); - setup_restart_psw(); setup_lowcore(); cpu_init(); diff --git a/arch/s390/kernel/signal.c b/arch/s390/kernel/signal.c index 2d421d90fad..f29f5ef400e 100644 --- a/arch/s390/kernel/signal.c +++ b/arch/s390/kernel/signal.c @@ -384,7 +384,6 @@ static int handle_signal(unsigned long sig, struct k_sigaction *ka, siginfo_t *info, sigset_t *oldset, struct pt_regs *regs) { - sigset_t blocked; int ret; /* Set up the stack frame */ @@ -394,10 +393,7 @@ static int handle_signal(unsigned long sig, struct k_sigaction *ka, ret = setup_frame(sig, ka, oldset, regs); if (ret) return ret; - sigorsets(&blocked, ¤t->blocked, &ka->sa.sa_mask); - if (!(ka->sa.sa_flags & SA_NODEFER)) - sigaddset(&blocked, sig); - set_current_blocked(&blocked); + block_sigmask(ka, sig); return 0; } diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c index 2398ce6b15a..a8bf9994b08 100644 --- a/arch/s390/kernel/smp.c +++ b/arch/s390/kernel/smp.c @@ -1,23 +1,18 @@ /* - * arch/s390/kernel/smp.c + * SMP related functions * - * Copyright IBM Corp. 1999, 2009 - * Author(s): Denis Joseph Barrow (djbarrow@de.ibm.com,barrow_dj@yahoo.com), - * Martin Schwidefsky (schwidefsky@de.ibm.com) - * Heiko Carstens (heiko.carstens@de.ibm.com) + * Copyright IBM Corp. 1999,2012 + * Author(s): Denis Joseph Barrow, + * Martin Schwidefsky <schwidefsky@de.ibm.com>, + * Heiko Carstens <heiko.carstens@de.ibm.com>, * * based on other smp stuff by * (c) 1995 Alan Cox, CymruNET Ltd <alan@cymru.net> * (c) 1998 Ingo Molnar * - * We work with logical cpu numbering everywhere we can. The only - * functions using the real cpu address (got from STAP) are the sigp - * functions. For all other functions we use the identity mapping. - * That means that cpu_number_map[i] == i for every cpu. cpu_number_map is - * used e.g. to find the idle task belonging to a logical cpu. Every array - * in the kernel is sorted by the logical cpu number and not by the physical - * one which is causing all the confusion with __cpu_logical_map and - * cpu_number_map in other architectures. + * The code outside of smp.c uses logical cpu numbers, only smp.c does + * the translation of logical to physical cpu ids. All new code that + * operates on physical cpu numbers needs to go into smp.c. */ #define KMSG_COMPONENT "cpu" @@ -31,198 +26,433 @@ #include <linux/spinlock.h> #include <linux/kernel_stat.h> #include <linux/delay.h> -#include <linux/cache.h> #include <linux/interrupt.h> #include <linux/irqflags.h> #include <linux/cpu.h> -#include <linux/timex.h> -#include <linux/bootmem.h> #include <linux/slab.h> #include <linux/crash_dump.h> #include <asm/asm-offsets.h> #include <asm/ipl.h> #include <asm/setup.h> -#include <asm/sigp.h> -#include <asm/pgalloc.h> #include <asm/irq.h> -#include <asm/cpcmd.h> #include <asm/tlbflush.h> #include <asm/timer.h> #include <asm/lowcore.h> #include <asm/sclp.h> -#include <asm/cputime.h> #include <asm/vdso.h> -#include <asm/cpu.h> +#include <asm/debug.h> +#include <asm/os_info.h> #include "entry.h" -/* logical cpu to cpu address */ -unsigned short __cpu_logical_map[NR_CPUS]; +enum { + sigp_sense = 1, + sigp_external_call = 2, + sigp_emergency_signal = 3, + sigp_start = 4, + sigp_stop = 5, + sigp_restart = 6, + sigp_stop_and_store_status = 9, + sigp_initial_cpu_reset = 11, + sigp_cpu_reset = 12, + sigp_set_prefix = 13, + sigp_store_status_at_address = 14, + sigp_store_extended_status_at_address = 15, + sigp_set_architecture = 18, + sigp_conditional_emergency_signal = 19, + sigp_sense_running = 21, +}; -static struct task_struct *current_set[NR_CPUS]; +enum { + sigp_order_code_accepted = 0, + sigp_status_stored = 1, + sigp_busy = 2, + sigp_not_operational = 3, +}; -static u8 smp_cpu_type; -static int smp_use_sigp_detection; +enum { + ec_schedule = 0, + ec_call_function, + ec_call_function_single, + ec_stop_cpu, +}; -enum s390_cpu_state { +enum { CPU_STATE_STANDBY, CPU_STATE_CONFIGURED, }; +struct pcpu { + struct cpu cpu; + struct task_struct *idle; /* idle process for the cpu */ + struct _lowcore *lowcore; /* lowcore page(s) for the cpu */ + unsigned long async_stack; /* async stack for the cpu */ + unsigned long panic_stack; /* panic stack for the cpu */ + unsigned long ec_mask; /* bit mask for ec_xxx functions */ + int state; /* physical cpu state */ + u32 status; /* last status received via sigp */ + u16 address; /* physical cpu address */ +}; + +static u8 boot_cpu_type; +static u16 boot_cpu_address; +static struct pcpu pcpu_devices[NR_CPUS]; + DEFINE_MUTEX(smp_cpu_state_mutex); -static int smp_cpu_state[NR_CPUS]; -static DEFINE_PER_CPU(struct cpu, cpu_devices); +/* + * Signal processor helper functions. + */ +static inline int __pcpu_sigp(u16 addr, u8 order, u32 parm, u32 *status) +{ + register unsigned int reg1 asm ("1") = parm; + int cc; -static void smp_ext_bitcall(int, int); + asm volatile( + " sigp %1,%2,0(%3)\n" + " ipm %0\n" + " srl %0,28\n" + : "=d" (cc), "+d" (reg1) : "d" (addr), "a" (order) : "cc"); + if (status && cc == 1) + *status = reg1; + return cc; +} -static int raw_cpu_stopped(int cpu) +static inline int __pcpu_sigp_relax(u16 addr, u8 order, u32 parm, u32 *status) { - u32 status; + int cc; - switch (raw_sigp_ps(&status, 0, cpu, sigp_sense)) { - case sigp_status_stored: - /* Check for stopped and check stop state */ - if (status & 0x50) - return 1; - break; - default: - break; + while (1) { + cc = __pcpu_sigp(addr, order, parm, status); + if (cc != sigp_busy) + return cc; + cpu_relax(); } - return 0; } -static inline int cpu_stopped(int cpu) +static int pcpu_sigp_retry(struct pcpu *pcpu, u8 order, u32 parm) { - return raw_cpu_stopped(cpu_logical_map(cpu)); + int cc, retry; + + for (retry = 0; ; retry++) { + cc = __pcpu_sigp(pcpu->address, order, parm, &pcpu->status); + if (cc != sigp_busy) + break; + if (retry >= 3) + udelay(10); + } + return cc; +} + +static inline int pcpu_stopped(struct pcpu *pcpu) +{ + if (__pcpu_sigp(pcpu->address, sigp_sense, + 0, &pcpu->status) != sigp_status_stored) + return 0; + /* Check for stopped and check stop state */ + return !!(pcpu->status & 0x50); +} + +static inline int pcpu_running(struct pcpu *pcpu) +{ + if (__pcpu_sigp(pcpu->address, sigp_sense_running, + 0, &pcpu->status) != sigp_status_stored) + return 1; + /* Check for running status */ + return !(pcpu->status & 0x400); } /* - * Ensure that PSW restart is done on an online CPU + * Find struct pcpu by cpu address. */ -void smp_restart_with_online_cpu(void) +static struct pcpu *pcpu_find_address(const struct cpumask *mask, int address) { int cpu; - for_each_online_cpu(cpu) { - if (stap() == __cpu_logical_map[cpu]) { - /* We are online: Enable DAT again and return */ - __load_psw_mask(psw_kernel_bits | PSW_MASK_DAT); - return; - } + for_each_cpu(cpu, mask) + if (pcpu_devices[cpu].address == address) + return pcpu_devices + cpu; + return NULL; +} + +static void pcpu_ec_call(struct pcpu *pcpu, int ec_bit) +{ + int order; + + set_bit(ec_bit, &pcpu->ec_mask); + order = pcpu_running(pcpu) ? + sigp_external_call : sigp_emergency_signal; + pcpu_sigp_retry(pcpu, order, 0); +} + +static int __cpuinit pcpu_alloc_lowcore(struct pcpu *pcpu, int cpu) +{ + struct _lowcore *lc; + + if (pcpu != &pcpu_devices[0]) { + pcpu->lowcore = (struct _lowcore *) + __get_free_pages(GFP_KERNEL | GFP_DMA, LC_ORDER); + pcpu->async_stack = __get_free_pages(GFP_KERNEL, ASYNC_ORDER); + pcpu->panic_stack = __get_free_page(GFP_KERNEL); + if (!pcpu->lowcore || !pcpu->panic_stack || !pcpu->async_stack) + goto out; } - /* We are not online: Do PSW restart on an online CPU */ - while (sigp(cpu, sigp_restart) == sigp_busy) - cpu_relax(); - /* And stop ourself */ - while (raw_sigp(stap(), sigp_stop) == sigp_busy) - cpu_relax(); - for (;;); + lc = pcpu->lowcore; + memcpy(lc, &S390_lowcore, 512); + memset((char *) lc + 512, 0, sizeof(*lc) - 512); + lc->async_stack = pcpu->async_stack + ASYNC_SIZE; + lc->panic_stack = pcpu->panic_stack + PAGE_SIZE; + lc->cpu_nr = cpu; +#ifndef CONFIG_64BIT + if (MACHINE_HAS_IEEE) { + lc->extended_save_area_addr = get_zeroed_page(GFP_KERNEL); + if (!lc->extended_save_area_addr) + goto out; + } +#else + if (vdso_alloc_per_cpu(lc)) + goto out; +#endif + lowcore_ptr[cpu] = lc; + pcpu_sigp_retry(pcpu, sigp_set_prefix, (u32)(unsigned long) lc); + return 0; +out: + if (pcpu != &pcpu_devices[0]) { + free_page(pcpu->panic_stack); + free_pages(pcpu->async_stack, ASYNC_ORDER); + free_pages((unsigned long) pcpu->lowcore, LC_ORDER); + } + return -ENOMEM; } -void smp_switch_to_ipl_cpu(void (*func)(void *), void *data) +static void pcpu_free_lowcore(struct pcpu *pcpu) { - struct _lowcore *lc, *current_lc; - struct stack_frame *sf; - struct pt_regs *regs; - unsigned long sp; - - if (smp_processor_id() == 0) - func(data); - __load_psw_mask(PSW_DEFAULT_KEY | PSW_MASK_BASE | - PSW_MASK_EA | PSW_MASK_BA); - /* Disable lowcore protection */ - __ctl_clear_bit(0, 28); - current_lc = lowcore_ptr[smp_processor_id()]; - lc = lowcore_ptr[0]; - if (!lc) - lc = current_lc; - lc->restart_psw.mask = - PSW_DEFAULT_KEY | PSW_MASK_BASE | PSW_MASK_EA | PSW_MASK_BA; - lc->restart_psw.addr = PSW_ADDR_AMODE | (unsigned long) smp_restart_cpu; - if (!cpu_online(0)) - smp_switch_to_cpu(func, data, 0, stap(), __cpu_logical_map[0]); - while (sigp(0, sigp_stop_and_store_status) == sigp_busy) - cpu_relax(); - sp = lc->panic_stack; - sp -= sizeof(struct pt_regs); - regs = (struct pt_regs *) sp; - memcpy(®s->gprs, ¤t_lc->gpregs_save_area, sizeof(regs->gprs)); - regs->psw = current_lc->psw_save_area; - sp -= STACK_FRAME_OVERHEAD; - sf = (struct stack_frame *) sp; - sf->back_chain = 0; - smp_switch_to_cpu(func, data, sp, stap(), __cpu_logical_map[0]); + pcpu_sigp_retry(pcpu, sigp_set_prefix, 0); + lowcore_ptr[pcpu - pcpu_devices] = NULL; +#ifndef CONFIG_64BIT + if (MACHINE_HAS_IEEE) { + struct _lowcore *lc = pcpu->lowcore; + + free_page((unsigned long) lc->extended_save_area_addr); + lc->extended_save_area_addr = 0; + } +#else + vdso_free_per_cpu(pcpu->lowcore); +#endif + if (pcpu != &pcpu_devices[0]) { + free_page(pcpu->panic_stack); + free_pages(pcpu->async_stack, ASYNC_ORDER); + free_pages((unsigned long) pcpu->lowcore, LC_ORDER); + } +} + +static void pcpu_prepare_secondary(struct pcpu *pcpu, int cpu) +{ + struct _lowcore *lc = pcpu->lowcore; + + atomic_inc(&init_mm.context.attach_count); + lc->cpu_nr = cpu; + lc->percpu_offset = __per_cpu_offset[cpu]; + lc->kernel_asce = S390_lowcore.kernel_asce; + lc->machine_flags = S390_lowcore.machine_flags; + lc->ftrace_func = S390_lowcore.ftrace_func; + lc->user_timer = lc->system_timer = lc->steal_timer = 0; + __ctl_store(lc->cregs_save_area, 0, 15); + save_access_regs((unsigned int *) lc->access_regs_save_area); + memcpy(lc->stfle_fac_list, S390_lowcore.stfle_fac_list, + MAX_FACILITY_BIT/8); +} + +static void pcpu_attach_task(struct pcpu *pcpu, struct task_struct *tsk) +{ + struct _lowcore *lc = pcpu->lowcore; + struct thread_info *ti = task_thread_info(tsk); + + lc->kernel_stack = (unsigned long) task_stack_page(tsk) + THREAD_SIZE; + lc->thread_info = (unsigned long) task_thread_info(tsk); + lc->current_task = (unsigned long) tsk; + lc->user_timer = ti->user_timer; + lc->system_timer = ti->system_timer; + lc->steal_timer = 0; +} + +static void pcpu_start_fn(struct pcpu *pcpu, void (*func)(void *), void *data) +{ + struct _lowcore *lc = pcpu->lowcore; + + lc->restart_stack = lc->kernel_stack; + lc->restart_fn = (unsigned long) func; + lc->restart_data = (unsigned long) data; + lc->restart_source = -1UL; + pcpu_sigp_retry(pcpu, sigp_restart, 0); +} + +/* + * Call function via PSW restart on pcpu and stop the current cpu. + */ +static void pcpu_delegate(struct pcpu *pcpu, void (*func)(void *), + void *data, unsigned long stack) +{ + struct _lowcore *lc = pcpu->lowcore; + unsigned short this_cpu; + + __load_psw_mask(psw_kernel_bits); + this_cpu = stap(); + if (pcpu->address == this_cpu) + func(data); /* should not return */ + /* Stop target cpu (if func returns this stops the current cpu). */ + pcpu_sigp_retry(pcpu, sigp_stop, 0); + /* Restart func on the target cpu and stop the current cpu. */ + lc->restart_stack = stack; + lc->restart_fn = (unsigned long) func; + lc->restart_data = (unsigned long) data; + lc->restart_source = (unsigned long) this_cpu; + asm volatile( + "0: sigp 0,%0,6 # sigp restart to target cpu\n" + " brc 2,0b # busy, try again\n" + "1: sigp 0,%1,5 # sigp stop to current cpu\n" + " brc 2,1b # busy, try again\n" + : : "d" (pcpu->address), "d" (this_cpu) : "0", "1", "cc"); + for (;;) ; +} + +/* + * Call function on an online CPU. + */ +void smp_call_online_cpu(void (*func)(void *), void *data) +{ + struct pcpu *pcpu; + + /* Use the current cpu if it is online. */ + pcpu = pcpu_find_address(cpu_online_mask, stap()); + if (!pcpu) + /* Use the first online cpu. */ + pcpu = pcpu_devices + cpumask_first(cpu_online_mask); + pcpu_delegate(pcpu, func, data, (unsigned long) restart_stack); } -static void smp_stop_cpu(void) +/* + * Call function on the ipl CPU. + */ +void smp_call_ipl_cpu(void (*func)(void *), void *data) { - while (sigp(smp_processor_id(), sigp_stop) == sigp_busy) + pcpu_delegate(&pcpu_devices[0], func, data, + pcpu_devices->panic_stack + PAGE_SIZE); +} + +int smp_find_processor_id(u16 address) +{ + int cpu; + + for_each_present_cpu(cpu) + if (pcpu_devices[cpu].address == address) + return cpu; + return -1; +} + +int smp_vcpu_scheduled(int cpu) +{ + return pcpu_running(pcpu_devices + cpu); +} + +void smp_yield(void) +{ + if (MACHINE_HAS_DIAG44) + asm volatile("diag 0,0,0x44"); +} + +void smp_yield_cpu(int cpu) +{ + if (MACHINE_HAS_DIAG9C) + asm volatile("diag %0,0,0x9c" + : : "d" (pcpu_devices[cpu].address)); + else if (MACHINE_HAS_DIAG44) + asm volatile("diag 0,0,0x44"); +} + +/* + * Send cpus emergency shutdown signal. This gives the cpus the + * opportunity to complete outstanding interrupts. + */ +void smp_emergency_stop(cpumask_t *cpumask) +{ + u64 end; + int cpu; + + end = get_clock() + (1000000UL << 12); + for_each_cpu(cpu, cpumask) { + struct pcpu *pcpu = pcpu_devices + cpu; + set_bit(ec_stop_cpu, &pcpu->ec_mask); + while (__pcpu_sigp(pcpu->address, sigp_emergency_signal, + 0, NULL) == sigp_busy && + get_clock() < end) + cpu_relax(); + } + while (get_clock() < end) { + for_each_cpu(cpu, cpumask) + if (pcpu_stopped(pcpu_devices + cpu)) + cpumask_clear_cpu(cpu, cpumask); + if (cpumask_empty(cpumask)) + break; cpu_relax(); + } } +/* + * Stop all cpus but the current one. + */ void smp_send_stop(void) { cpumask_t cpumask; int cpu; - u64 end; /* Disable all interrupts/machine checks */ __load_psw_mask(psw_kernel_bits | PSW_MASK_DAT); trace_hardirqs_off(); + debug_set_critical(); cpumask_copy(&cpumask, cpu_online_mask); cpumask_clear_cpu(smp_processor_id(), &cpumask); - if (oops_in_progress) { - /* - * Give the other cpus the opportunity to complete - * outstanding interrupts before stopping them. - */ - end = get_clock() + (1000000UL << 12); - for_each_cpu(cpu, &cpumask) { - set_bit(ec_stop_cpu, (unsigned long *) - &lowcore_ptr[cpu]->ext_call_fast); - while (sigp(cpu, sigp_emergency_signal) == sigp_busy && - get_clock() < end) - cpu_relax(); - } - while (get_clock() < end) { - for_each_cpu(cpu, &cpumask) - if (cpu_stopped(cpu)) - cpumask_clear_cpu(cpu, &cpumask); - if (cpumask_empty(&cpumask)) - break; - cpu_relax(); - } - } + if (oops_in_progress) + smp_emergency_stop(&cpumask); /* stop all processors */ for_each_cpu(cpu, &cpumask) { - while (sigp(cpu, sigp_stop) == sigp_busy) - cpu_relax(); - while (!cpu_stopped(cpu)) + struct pcpu *pcpu = pcpu_devices + cpu; + pcpu_sigp_retry(pcpu, sigp_stop, 0); + while (!pcpu_stopped(pcpu)) cpu_relax(); } } /* + * Stop the current cpu. + */ +void smp_stop_cpu(void) +{ + pcpu_sigp_retry(pcpu_devices + smp_processor_id(), sigp_stop, 0); + for (;;) ; +} + +/* * This is the main routine where commands issued by other * cpus are handled. */ - -static void do_ext_call_interrupt(unsigned int ext_int_code, +static void do_ext_call_interrupt(struct ext_code ext_code, unsigned int param32, unsigned long param64) { unsigned long bits; + int cpu; - if ((ext_int_code & 0xffff) == 0x1202) - kstat_cpu(smp_processor_id()).irqs[EXTINT_EXC]++; + cpu = smp_processor_id(); + if (ext_code.code == 0x1202) + kstat_cpu(cpu).irqs[EXTINT_EXC]++; else - kstat_cpu(smp_processor_id()).irqs[EXTINT_EMS]++; + kstat_cpu(cpu).irqs[EXTINT_EMS]++; /* * handle bit signal external calls */ - bits = xchg(&S390_lowcore.ext_call_fast, 0); + bits = xchg(&pcpu_devices[cpu].ec_mask, 0); if (test_bit(ec_stop_cpu, &bits)) smp_stop_cpu(); @@ -238,38 +468,17 @@ static void do_ext_call_interrupt(unsigned int ext_int_code, } -/* - * Send an external call sigp to another cpu and return without waiting - * for its completion. - */ -static void smp_ext_bitcall(int cpu, int sig) -{ - int order; - - /* - * Set signaling bit in lowcore of target cpu and kick it - */ - set_bit(sig, (unsigned long *) &lowcore_ptr[cpu]->ext_call_fast); - while (1) { - order = smp_vcpu_scheduled(cpu) ? - sigp_external_call : sigp_emergency_signal; - if (sigp(cpu, order) != sigp_busy) - break; - udelay(10); - } -} - void arch_send_call_function_ipi_mask(const struct cpumask *mask) { int cpu; for_each_cpu(cpu, mask) - smp_ext_bitcall(cpu, ec_call_function); + pcpu_ec_call(pcpu_devices + cpu, ec_call_function); } void arch_send_call_function_single_ipi(int cpu) { - smp_ext_bitcall(cpu, ec_call_function_single); + pcpu_ec_call(pcpu_devices + cpu, ec_call_function_single); } #ifndef CONFIG_64BIT @@ -295,15 +504,16 @@ EXPORT_SYMBOL(smp_ptlb_all); */ void smp_send_reschedule(int cpu) { - smp_ext_bitcall(cpu, ec_schedule); + pcpu_ec_call(pcpu_devices + cpu, ec_schedule); } /* * parameter area for the set/clear control bit callbacks */ struct ec_creg_mask_parms { - unsigned long orvals[16]; - unsigned long andvals[16]; + unsigned long orval; + unsigned long andval; + int cr; }; /* @@ -313,11 +523,9 @@ static void smp_ctl_bit_callback(void *info) { struct ec_creg_mask_parms *pp = info; unsigned long cregs[16]; - int i; __ctl_store(cregs, 0, 15); - for (i = 0; i <= 15; i++) - cregs[i] = (cregs[i] & pp->andvals[i]) | pp->orvals[i]; + cregs[pp->cr] = (cregs[pp->cr] & pp->andval) | pp->orval; __ctl_load(cregs, 0, 15); } @@ -326,11 +534,8 @@ static void smp_ctl_bit_callback(void *info) */ void smp_ctl_set_bit(int cr, int bit) { - struct ec_creg_mask_parms parms; + struct ec_creg_mask_parms parms = { 1UL << bit, -1UL, cr }; - memset(&parms.orvals, 0, sizeof(parms.orvals)); - memset(&parms.andvals, 0xff, sizeof(parms.andvals)); - parms.orvals[cr] = 1UL << bit; on_each_cpu(smp_ctl_bit_callback, &parms, 1); } EXPORT_SYMBOL(smp_ctl_set_bit); @@ -340,226 +545,178 @@ EXPORT_SYMBOL(smp_ctl_set_bit); */ void smp_ctl_clear_bit(int cr, int bit) { - struct ec_creg_mask_parms parms; + struct ec_creg_mask_parms parms = { 0, ~(1UL << bit), cr }; - memset(&parms.orvals, 0, sizeof(parms.orvals)); - memset(&parms.andvals, 0xff, sizeof(parms.andvals)); - parms.andvals[cr] = ~(1UL << bit); on_each_cpu(smp_ctl_bit_callback, &parms, 1); } EXPORT_SYMBOL(smp_ctl_clear_bit); #if defined(CONFIG_ZFCPDUMP) || defined(CONFIG_CRASH_DUMP) -static void __init smp_get_save_area(unsigned int cpu, unsigned int phy_cpu) +struct save_area *zfcpdump_save_areas[NR_CPUS + 1]; +EXPORT_SYMBOL_GPL(zfcpdump_save_areas); + +static void __init smp_get_save_area(int cpu, u16 address) { - if (ipl_info.type != IPL_TYPE_FCP_DUMP && !OLDMEM_BASE) - return; + void *lc = pcpu_devices[0].lowcore; + struct save_area *save_area; + if (is_kdump_kernel()) return; + if (!OLDMEM_BASE && (address == boot_cpu_address || + ipl_info.type != IPL_TYPE_FCP_DUMP)) + return; if (cpu >= NR_CPUS) { - pr_warning("CPU %i exceeds the maximum %i and is excluded from " - "the dump\n", cpu, NR_CPUS - 1); + pr_warning("CPU %i exceeds the maximum %i and is excluded " + "from the dump\n", cpu, NR_CPUS - 1); return; } - zfcpdump_save_areas[cpu] = kmalloc(sizeof(struct save_area), GFP_KERNEL); - while (raw_sigp(phy_cpu, sigp_stop_and_store_status) == sigp_busy) - cpu_relax(); - memcpy_real(zfcpdump_save_areas[cpu], - (void *)(unsigned long) store_prefix() + SAVE_AREA_BASE, - sizeof(struct save_area)); + save_area = kmalloc(sizeof(struct save_area), GFP_KERNEL); + if (!save_area) + panic("could not allocate memory for save area\n"); + zfcpdump_save_areas[cpu] = save_area; +#ifdef CONFIG_CRASH_DUMP + if (address == boot_cpu_address) { + /* Copy the registers of the boot cpu. */ + copy_oldmem_page(1, (void *) save_area, sizeof(*save_area), + SAVE_AREA_BASE - PAGE_SIZE, 0); + return; + } +#endif + /* Get the registers of a non-boot cpu. */ + __pcpu_sigp_relax(address, sigp_stop_and_store_status, 0, NULL); + memcpy_real(save_area, lc + SAVE_AREA_BASE, sizeof(*save_area)); } -struct save_area *zfcpdump_save_areas[NR_CPUS + 1]; -EXPORT_SYMBOL_GPL(zfcpdump_save_areas); - -#else - -static inline void smp_get_save_area(unsigned int cpu, unsigned int phy_cpu) { } - -#endif /* CONFIG_ZFCPDUMP */ - -static int cpu_known(int cpu_id) +int smp_store_status(int cpu) { - int cpu; + struct pcpu *pcpu; - for_each_present_cpu(cpu) { - if (__cpu_logical_map[cpu] == cpu_id) - return 1; - } + pcpu = pcpu_devices + cpu; + if (__pcpu_sigp_relax(pcpu->address, sigp_stop_and_store_status, + 0, NULL) != sigp_order_code_accepted) + return -EIO; return 0; } -static int smp_rescan_cpus_sigp(cpumask_t avail) -{ - int cpu_id, logical_cpu; +#else /* CONFIG_ZFCPDUMP || CONFIG_CRASH_DUMP */ - logical_cpu = cpumask_first(&avail); - if (logical_cpu >= nr_cpu_ids) - return 0; - for (cpu_id = 0; cpu_id <= MAX_CPU_ADDRESS; cpu_id++) { - if (cpu_known(cpu_id)) - continue; - __cpu_logical_map[logical_cpu] = cpu_id; - cpu_set_polarization(logical_cpu, POLARIZATION_UNKNOWN); - if (!cpu_stopped(logical_cpu)) - continue; - set_cpu_present(logical_cpu, true); - smp_cpu_state[logical_cpu] = CPU_STATE_CONFIGURED; - logical_cpu = cpumask_next(logical_cpu, &avail); - if (logical_cpu >= nr_cpu_ids) - break; - } - return 0; -} +static inline void smp_get_save_area(int cpu, u16 address) { } + +#endif /* CONFIG_ZFCPDUMP || CONFIG_CRASH_DUMP */ -static int smp_rescan_cpus_sclp(cpumask_t avail) +static struct sclp_cpu_info *smp_get_cpu_info(void) { + static int use_sigp_detection; struct sclp_cpu_info *info; - int cpu_id, logical_cpu, cpu; - int rc; - - logical_cpu = cpumask_first(&avail); - if (logical_cpu >= nr_cpu_ids) - return 0; - info = kmalloc(sizeof(*info), GFP_KERNEL); - if (!info) - return -ENOMEM; - rc = sclp_get_cpu_info(info); - if (rc) - goto out; - for (cpu = 0; cpu < info->combined; cpu++) { - if (info->has_cpu_type && info->cpu[cpu].type != smp_cpu_type) - continue; - cpu_id = info->cpu[cpu].address; - if (cpu_known(cpu_id)) - continue; - __cpu_logical_map[logical_cpu] = cpu_id; - cpu_set_polarization(logical_cpu, POLARIZATION_UNKNOWN); - set_cpu_present(logical_cpu, true); - if (cpu >= info->configured) - smp_cpu_state[logical_cpu] = CPU_STATE_STANDBY; - else - smp_cpu_state[logical_cpu] = CPU_STATE_CONFIGURED; - logical_cpu = cpumask_next(logical_cpu, &avail); - if (logical_cpu >= nr_cpu_ids) - break; + int address; + + info = kzalloc(sizeof(*info), GFP_KERNEL); + if (info && (use_sigp_detection || sclp_get_cpu_info(info))) { + use_sigp_detection = 1; + for (address = 0; address <= MAX_CPU_ADDRESS; address++) { + if (__pcpu_sigp_relax(address, sigp_sense, 0, NULL) == + sigp_not_operational) + continue; + info->cpu[info->configured].address = address; + info->configured++; + } + info->combined = info->configured; } -out: - kfree(info); - return rc; + return info; } -static int __smp_rescan_cpus(void) +static int __devinit smp_add_present_cpu(int cpu); + +static int __devinit __smp_rescan_cpus(struct sclp_cpu_info *info, + int sysfs_add) { + struct pcpu *pcpu; cpumask_t avail; + int cpu, nr, i; + nr = 0; cpumask_xor(&avail, cpu_possible_mask, cpu_present_mask); - if (smp_use_sigp_detection) - return smp_rescan_cpus_sigp(avail); - else - return smp_rescan_cpus_sclp(avail); + cpu = cpumask_first(&avail); + for (i = 0; (i < info->combined) && (cpu < nr_cpu_ids); i++) { + if (info->has_cpu_type && info->cpu[i].type != boot_cpu_type) + continue; + if (pcpu_find_address(cpu_present_mask, info->cpu[i].address)) + continue; + pcpu = pcpu_devices + cpu; + pcpu->address = info->cpu[i].address; + pcpu->state = (cpu >= info->configured) ? + CPU_STATE_STANDBY : CPU_STATE_CONFIGURED; + cpu_set_polarization(cpu, POLARIZATION_UNKNOWN); + set_cpu_present(cpu, true); + if (sysfs_add && smp_add_present_cpu(cpu) != 0) + set_cpu_present(cpu, false); + else + nr++; + cpu = cpumask_next(cpu, &avail); + } + return nr; } static void __init smp_detect_cpus(void) { unsigned int cpu, c_cpus, s_cpus; struct sclp_cpu_info *info; - u16 boot_cpu_addr, cpu_addr; - c_cpus = 1; - s_cpus = 0; - boot_cpu_addr = __cpu_logical_map[0]; - info = kmalloc(sizeof(*info), GFP_KERNEL); + info = smp_get_cpu_info(); if (!info) panic("smp_detect_cpus failed to allocate memory\n"); -#ifdef CONFIG_CRASH_DUMP - if (OLDMEM_BASE && !is_kdump_kernel()) { - struct save_area *save_area; - - save_area = kmalloc(sizeof(*save_area), GFP_KERNEL); - if (!save_area) - panic("could not allocate memory for save area\n"); - copy_oldmem_page(1, (void *) save_area, sizeof(*save_area), - 0x200, 0); - zfcpdump_save_areas[0] = save_area; - } -#endif - /* Use sigp detection algorithm if sclp doesn't work. */ - if (sclp_get_cpu_info(info)) { - smp_use_sigp_detection = 1; - for (cpu = 0; cpu <= MAX_CPU_ADDRESS; cpu++) { - if (cpu == boot_cpu_addr) - continue; - if (!raw_cpu_stopped(cpu)) - continue; - smp_get_save_area(c_cpus, cpu); - c_cpus++; - } - goto out; - } - if (info->has_cpu_type) { for (cpu = 0; cpu < info->combined; cpu++) { - if (info->cpu[cpu].address == boot_cpu_addr) { - smp_cpu_type = info->cpu[cpu].type; - break; - } + if (info->cpu[cpu].address != boot_cpu_address) + continue; + /* The boot cpu dictates the cpu type. */ + boot_cpu_type = info->cpu[cpu].type; + break; } } - + c_cpus = s_cpus = 0; for (cpu = 0; cpu < info->combined; cpu++) { - if (info->has_cpu_type && info->cpu[cpu].type != smp_cpu_type) - continue; - cpu_addr = info->cpu[cpu].address; - if (cpu_addr == boot_cpu_addr) + if (info->has_cpu_type && info->cpu[cpu].type != boot_cpu_type) continue; - if (!raw_cpu_stopped(cpu_addr)) { + if (cpu < info->configured) { + smp_get_save_area(c_cpus, info->cpu[cpu].address); + c_cpus++; + } else s_cpus++; - continue; - } - smp_get_save_area(c_cpus, cpu_addr); - c_cpus++; } -out: - kfree(info); pr_info("%d configured CPUs, %d standby CPUs\n", c_cpus, s_cpus); get_online_cpus(); - __smp_rescan_cpus(); + __smp_rescan_cpus(info, 0); put_online_cpus(); + kfree(info); } /* * Activate a secondary processor. */ -int __cpuinit start_secondary(void *cpuvoid) +static void __cpuinit smp_start_secondary(void *cpuvoid) { + S390_lowcore.last_update_clock = get_clock(); + S390_lowcore.restart_stack = (unsigned long) restart_stack; + S390_lowcore.restart_fn = (unsigned long) do_restart; + S390_lowcore.restart_data = 0; + S390_lowcore.restart_source = -1UL; + restore_access_regs(S390_lowcore.access_regs_save_area); + __ctl_load(S390_lowcore.cregs_save_area, 0, 15); + __load_psw_mask(psw_kernel_bits | PSW_MASK_DAT); cpu_init(); preempt_disable(); init_cpu_timer(); init_cpu_vtimer(); pfault_init(); - notify_cpu_starting(smp_processor_id()); ipi_call_lock(); set_cpu_online(smp_processor_id(), true); ipi_call_unlock(); - __ctl_clear_bit(0, 28); /* Disable lowcore protection */ - S390_lowcore.restart_psw.mask = - PSW_DEFAULT_KEY | PSW_MASK_BASE | PSW_MASK_EA | PSW_MASK_BA; - S390_lowcore.restart_psw.addr = - PSW_ADDR_AMODE | (unsigned long) psw_restart_int_handler; - __ctl_set_bit(0, 28); /* Enable lowcore protection */ - /* - * Wait until the cpu which brought this one up marked it - * active before enabling interrupts. - */ - while (!cpumask_test_cpu(smp_processor_id(), cpu_active_mask)) - cpu_relax(); local_irq_enable(); /* cpu_idle will call schedule for us */ cpu_idle(); - return 0; } struct create_idle { @@ -578,82 +735,20 @@ static void __cpuinit smp_fork_idle(struct work_struct *work) complete(&c_idle->done); } -static int __cpuinit smp_alloc_lowcore(int cpu) -{ - unsigned long async_stack, panic_stack; - struct _lowcore *lowcore; - - lowcore = (void *) __get_free_pages(GFP_KERNEL | GFP_DMA, LC_ORDER); - if (!lowcore) - return -ENOMEM; - async_stack = __get_free_pages(GFP_KERNEL, ASYNC_ORDER); - panic_stack = __get_free_page(GFP_KERNEL); - if (!panic_stack || !async_stack) - goto out; - memcpy(lowcore, &S390_lowcore, 512); - memset((char *)lowcore + 512, 0, sizeof(*lowcore) - 512); - lowcore->async_stack = async_stack + ASYNC_SIZE; - lowcore->panic_stack = panic_stack + PAGE_SIZE; - lowcore->restart_psw.mask = - PSW_DEFAULT_KEY | PSW_MASK_BASE | PSW_MASK_EA | PSW_MASK_BA; - lowcore->restart_psw.addr = - PSW_ADDR_AMODE | (unsigned long) restart_int_handler; - if (user_mode != HOME_SPACE_MODE) - lowcore->restart_psw.mask |= PSW_ASC_HOME; -#ifndef CONFIG_64BIT - if (MACHINE_HAS_IEEE) { - unsigned long save_area; - - save_area = get_zeroed_page(GFP_KERNEL); - if (!save_area) - goto out; - lowcore->extended_save_area_addr = (u32) save_area; - } -#else - if (vdso_alloc_per_cpu(cpu, lowcore)) - goto out; -#endif - lowcore_ptr[cpu] = lowcore; - return 0; - -out: - free_page(panic_stack); - free_pages(async_stack, ASYNC_ORDER); - free_pages((unsigned long) lowcore, LC_ORDER); - return -ENOMEM; -} - -static void smp_free_lowcore(int cpu) -{ - struct _lowcore *lowcore; - - lowcore = lowcore_ptr[cpu]; -#ifndef CONFIG_64BIT - if (MACHINE_HAS_IEEE) - free_page((unsigned long) lowcore->extended_save_area_addr); -#else - vdso_free_per_cpu(cpu, lowcore); -#endif - free_page(lowcore->panic_stack - PAGE_SIZE); - free_pages(lowcore->async_stack - ASYNC_SIZE, ASYNC_ORDER); - free_pages((unsigned long) lowcore, LC_ORDER); - lowcore_ptr[cpu] = NULL; -} - /* Upping and downing of CPUs */ int __cpuinit __cpu_up(unsigned int cpu) { - struct _lowcore *cpu_lowcore; struct create_idle c_idle; - struct task_struct *idle; - struct stack_frame *sf; - u32 lowcore; - int ccode; + struct pcpu *pcpu; + int rc; - if (smp_cpu_state[cpu] != CPU_STATE_CONFIGURED) + pcpu = pcpu_devices + cpu; + if (pcpu->state != CPU_STATE_CONFIGURED) + return -EIO; + if (pcpu_sigp_retry(pcpu, sigp_initial_cpu_reset, 0) != + sigp_order_code_accepted) return -EIO; - idle = current_set[cpu]; - if (!idle) { + if (!pcpu->idle) { c_idle.done = COMPLETION_INITIALIZER_ONSTACK(c_idle.done); INIT_WORK_ONSTACK(&c_idle.work, smp_fork_idle); c_idle.cpu = cpu; @@ -661,68 +756,28 @@ int __cpuinit __cpu_up(unsigned int cpu) wait_for_completion(&c_idle.done); if (IS_ERR(c_idle.idle)) return PTR_ERR(c_idle.idle); - idle = c_idle.idle; - current_set[cpu] = c_idle.idle; + pcpu->idle = c_idle.idle; } - init_idle(idle, cpu); - if (smp_alloc_lowcore(cpu)) - return -ENOMEM; - do { - ccode = sigp(cpu, sigp_initial_cpu_reset); - if (ccode == sigp_busy) - udelay(10); - if (ccode == sigp_not_operational) - goto err_out; - } while (ccode == sigp_busy); - - lowcore = (u32)(unsigned long)lowcore_ptr[cpu]; - while (sigp_p(lowcore, cpu, sigp_set_prefix) == sigp_busy) - udelay(10); - - cpu_lowcore = lowcore_ptr[cpu]; - cpu_lowcore->kernel_stack = (unsigned long) - task_stack_page(idle) + THREAD_SIZE; - cpu_lowcore->thread_info = (unsigned long) task_thread_info(idle); - sf = (struct stack_frame *) (cpu_lowcore->kernel_stack - - sizeof(struct pt_regs) - - sizeof(struct stack_frame)); - memset(sf, 0, sizeof(struct stack_frame)); - sf->gprs[9] = (unsigned long) sf; - cpu_lowcore->gpregs_save_area[15] = (unsigned long) sf; - __ctl_store(cpu_lowcore->cregs_save_area, 0, 15); - atomic_inc(&init_mm.context.attach_count); - asm volatile( - " stam 0,15,0(%0)" - : : "a" (&cpu_lowcore->access_regs_save_area) : "memory"); - cpu_lowcore->percpu_offset = __per_cpu_offset[cpu]; - cpu_lowcore->current_task = (unsigned long) idle; - cpu_lowcore->cpu_nr = cpu; - cpu_lowcore->kernel_asce = S390_lowcore.kernel_asce; - cpu_lowcore->machine_flags = S390_lowcore.machine_flags; - cpu_lowcore->ftrace_func = S390_lowcore.ftrace_func; - memcpy(cpu_lowcore->stfle_fac_list, S390_lowcore.stfle_fac_list, - MAX_FACILITY_BIT/8); - eieio(); - - while (sigp(cpu, sigp_restart) == sigp_busy) - udelay(10); - + init_idle(pcpu->idle, cpu); + rc = pcpu_alloc_lowcore(pcpu, cpu); + if (rc) + return rc; + pcpu_prepare_secondary(pcpu, cpu); + pcpu_attach_task(pcpu, pcpu->idle); + pcpu_start_fn(pcpu, smp_start_secondary, NULL); while (!cpu_online(cpu)) cpu_relax(); return 0; - -err_out: - smp_free_lowcore(cpu); - return -EIO; } static int __init setup_possible_cpus(char *s) { - int pcpus, cpu; + int max, cpu; - pcpus = simple_strtoul(s, NULL, 0); + if (kstrtoint(s, 0, &max) < 0) + return 0; init_cpu_possible(cpumask_of(0)); - for (cpu = 1; cpu < pcpus && cpu < nr_cpu_ids; cpu++) + for (cpu = 1; cpu < max && cpu < nr_cpu_ids; cpu++) set_cpu_possible(cpu, true); return 0; } @@ -732,113 +787,79 @@ early_param("possible_cpus", setup_possible_cpus); int __cpu_disable(void) { - struct ec_creg_mask_parms cr_parms; - int cpu = smp_processor_id(); - - set_cpu_online(cpu, false); + unsigned long cregs[16]; - /* Disable pfault pseudo page faults on this cpu. */ + set_cpu_online(smp_processor_id(), false); + /* Disable pseudo page faults on this cpu. */ pfault_fini(); - - memset(&cr_parms.orvals, 0, sizeof(cr_parms.orvals)); - memset(&cr_parms.andvals, 0xff, sizeof(cr_parms.andvals)); - - /* disable all external interrupts */ - cr_parms.orvals[0] = 0; - cr_parms.andvals[0] = ~(1 << 15 | 1 << 14 | 1 << 13 | 1 << 11 | - 1 << 10 | 1 << 9 | 1 << 6 | 1 << 5 | - 1 << 4); - /* disable all I/O interrupts */ - cr_parms.orvals[6] = 0; - cr_parms.andvals[6] = ~(1 << 31 | 1 << 30 | 1 << 29 | 1 << 28 | - 1 << 27 | 1 << 26 | 1 << 25 | 1 << 24); - /* disable most machine checks */ - cr_parms.orvals[14] = 0; - cr_parms.andvals[14] = ~(1 << 28 | 1 << 27 | 1 << 26 | - 1 << 25 | 1 << 24); - - smp_ctl_bit_callback(&cr_parms); - + /* Disable interrupt sources via control register. */ + __ctl_store(cregs, 0, 15); + cregs[0] &= ~0x0000ee70UL; /* disable all external interrupts */ + cregs[6] &= ~0xff000000UL; /* disable all I/O interrupts */ + cregs[14] &= ~0x1f000000UL; /* disable most machine checks */ + __ctl_load(cregs, 0, 15); return 0; } void __cpu_die(unsigned int cpu) { + struct pcpu *pcpu; + /* Wait until target cpu is down */ - while (!cpu_stopped(cpu)) + pcpu = pcpu_devices + cpu; + while (!pcpu_stopped(pcpu)) cpu_relax(); - while (sigp_p(0, cpu, sigp_set_prefix) == sigp_busy) - udelay(10); - smp_free_lowcore(cpu); + pcpu_free_lowcore(pcpu); atomic_dec(&init_mm.context.attach_count); } void __noreturn cpu_die(void) { idle_task_exit(); - while (sigp(smp_processor_id(), sigp_stop) == sigp_busy) - cpu_relax(); - for (;;); + pcpu_sigp_retry(pcpu_devices + smp_processor_id(), sigp_stop, 0); + for (;;) ; } #endif /* CONFIG_HOTPLUG_CPU */ -void __init smp_prepare_cpus(unsigned int max_cpus) +static void smp_call_os_info_init_fn(void) { -#ifndef CONFIG_64BIT - unsigned long save_area = 0; -#endif - unsigned long async_stack, panic_stack; - struct _lowcore *lowcore; + int (*init_fn)(void); + unsigned long size; - smp_detect_cpus(); + init_fn = os_info_old_entry(OS_INFO_INIT_FN, &size); + if (!init_fn) + return; + init_fn(); +} +void __init smp_prepare_cpus(unsigned int max_cpus) +{ /* request the 0x1201 emergency signal external interrupt */ if (register_external_interrupt(0x1201, do_ext_call_interrupt) != 0) panic("Couldn't request external interrupt 0x1201"); /* request the 0x1202 external call external interrupt */ if (register_external_interrupt(0x1202, do_ext_call_interrupt) != 0) panic("Couldn't request external interrupt 0x1202"); - - /* Reallocate current lowcore, but keep its contents. */ - lowcore = (void *) __get_free_pages(GFP_KERNEL | GFP_DMA, LC_ORDER); - panic_stack = __get_free_page(GFP_KERNEL); - async_stack = __get_free_pages(GFP_KERNEL, ASYNC_ORDER); - BUG_ON(!lowcore || !panic_stack || !async_stack); -#ifndef CONFIG_64BIT - if (MACHINE_HAS_IEEE) - save_area = get_zeroed_page(GFP_KERNEL); -#endif - local_irq_disable(); - local_mcck_disable(); - lowcore_ptr[smp_processor_id()] = lowcore; - *lowcore = S390_lowcore; - lowcore->panic_stack = panic_stack + PAGE_SIZE; - lowcore->async_stack = async_stack + ASYNC_SIZE; -#ifndef CONFIG_64BIT - if (MACHINE_HAS_IEEE) - lowcore->extended_save_area_addr = (u32) save_area; -#endif - set_prefix((u32)(unsigned long) lowcore); - local_mcck_enable(); - local_irq_enable(); -#ifdef CONFIG_64BIT - if (vdso_alloc_per_cpu(smp_processor_id(), &S390_lowcore)) - BUG(); -#endif + smp_call_os_info_init_fn(); + smp_detect_cpus(); } void __init smp_prepare_boot_cpu(void) { - BUG_ON(smp_processor_id() != 0); - - current_thread_info()->cpu = 0; - set_cpu_present(0, true); - set_cpu_online(0, true); + struct pcpu *pcpu = pcpu_devices; + + boot_cpu_address = stap(); + pcpu->idle = current; + pcpu->state = CPU_STATE_CONFIGURED; + pcpu->address = boot_cpu_address; + pcpu->lowcore = (struct _lowcore *)(unsigned long) store_prefix(); + pcpu->async_stack = S390_lowcore.async_stack - ASYNC_SIZE; + pcpu->panic_stack = S390_lowcore.panic_stack - PAGE_SIZE; S390_lowcore.percpu_offset = __per_cpu_offset[0]; - current_set[0] = current; - smp_cpu_state[0] = CPU_STATE_CONFIGURED; cpu_set_polarization(0, POLARIZATION_UNKNOWN); + set_cpu_present(0, true); + set_cpu_online(0, true); } void __init smp_cpus_done(unsigned int max_cpus) @@ -848,7 +869,6 @@ void __init smp_cpus_done(unsigned int max_cpus) void __init smp_setup_processor_id(void) { S390_lowcore.cpu_nr = 0; - __cpu_logical_map[0] = stap(); } /* @@ -864,56 +884,57 @@ int setup_profiling_timer(unsigned int multiplier) #ifdef CONFIG_HOTPLUG_CPU static ssize_t cpu_configure_show(struct device *dev, - struct device_attribute *attr, char *buf) + struct device_attribute *attr, char *buf) { ssize_t count; mutex_lock(&smp_cpu_state_mutex); - count = sprintf(buf, "%d\n", smp_cpu_state[dev->id]); + count = sprintf(buf, "%d\n", pcpu_devices[dev->id].state); mutex_unlock(&smp_cpu_state_mutex); return count; } static ssize_t cpu_configure_store(struct device *dev, - struct device_attribute *attr, - const char *buf, size_t count) + struct device_attribute *attr, + const char *buf, size_t count) { - int cpu = dev->id; - int val, rc; + struct pcpu *pcpu; + int cpu, val, rc; char delim; if (sscanf(buf, "%d %c", &val, &delim) != 1) return -EINVAL; if (val != 0 && val != 1) return -EINVAL; - get_online_cpus(); mutex_lock(&smp_cpu_state_mutex); rc = -EBUSY; /* disallow configuration changes of online cpus and cpu 0 */ + cpu = dev->id; if (cpu_online(cpu) || cpu == 0) goto out; + pcpu = pcpu_devices + cpu; rc = 0; switch (val) { case 0: - if (smp_cpu_state[cpu] == CPU_STATE_CONFIGURED) { - rc = sclp_cpu_deconfigure(__cpu_logical_map[cpu]); - if (!rc) { - smp_cpu_state[cpu] = CPU_STATE_STANDBY; - cpu_set_polarization(cpu, POLARIZATION_UNKNOWN); - topology_expect_change(); - } - } + if (pcpu->state != CPU_STATE_CONFIGURED) + break; + rc = sclp_cpu_deconfigure(pcpu->address); + if (rc) + break; + pcpu->state = CPU_STATE_STANDBY; + cpu_set_polarization(cpu, POLARIZATION_UNKNOWN); + topology_expect_change(); break; case 1: - if (smp_cpu_state[cpu] == CPU_STATE_STANDBY) { - rc = sclp_cpu_configure(__cpu_logical_map[cpu]); - if (!rc) { - smp_cpu_state[cpu] = CPU_STATE_CONFIGURED; - cpu_set_polarization(cpu, POLARIZATION_UNKNOWN); - topology_expect_change(); - } - } + if (pcpu->state != CPU_STATE_STANDBY) + break; + rc = sclp_cpu_configure(pcpu->address); + if (rc) + break; + pcpu->state = CPU_STATE_CONFIGURED; + cpu_set_polarization(cpu, POLARIZATION_UNKNOWN); + topology_expect_change(); break; default: break; @@ -929,7 +950,7 @@ static DEVICE_ATTR(configure, 0644, cpu_configure_show, cpu_configure_store); static ssize_t show_cpu_address(struct device *dev, struct device_attribute *attr, char *buf) { - return sprintf(buf, "%d\n", __cpu_logical_map[dev->id]); + return sprintf(buf, "%d\n", pcpu_devices[dev->id].address); } static DEVICE_ATTR(address, 0444, show_cpu_address, NULL); @@ -961,22 +982,16 @@ static DEVICE_ATTR(capability, 0444, show_capability, NULL); static ssize_t show_idle_count(struct device *dev, struct device_attribute *attr, char *buf) { - struct s390_idle_data *idle; + struct s390_idle_data *idle = &per_cpu(s390_idle, dev->id); unsigned long long idle_count; unsigned int sequence; - idle = &per_cpu(s390_idle, dev->id); -repeat: - sequence = idle->sequence; - smp_rmb(); - if (sequence & 1) - goto repeat; - idle_count = idle->idle_count; - if (idle->idle_enter) - idle_count++; - smp_rmb(); - if (idle->sequence != sequence) - goto repeat; + do { + sequence = ACCESS_ONCE(idle->sequence); + idle_count = ACCESS_ONCE(idle->idle_count); + if (ACCESS_ONCE(idle->idle_enter)) + idle_count++; + } while ((sequence & 1) || (idle->sequence != sequence)); return sprintf(buf, "%llu\n", idle_count); } static DEVICE_ATTR(idle_count, 0444, show_idle_count, NULL); @@ -984,24 +999,18 @@ static DEVICE_ATTR(idle_count, 0444, show_idle_count, NULL); static ssize_t show_idle_time(struct device *dev, struct device_attribute *attr, char *buf) { - struct s390_idle_data *idle; - unsigned long long now, idle_time, idle_enter; + struct s390_idle_data *idle = &per_cpu(s390_idle, dev->id); + unsigned long long now, idle_time, idle_enter, idle_exit; unsigned int sequence; - idle = &per_cpu(s390_idle, dev->id); - now = get_clock(); -repeat: - sequence = idle->sequence; - smp_rmb(); - if (sequence & 1) - goto repeat; - idle_time = idle->idle_time; - idle_enter = idle->idle_enter; - if (idle_enter != 0ULL && idle_enter < now) - idle_time += now - idle_enter; - smp_rmb(); - if (idle->sequence != sequence) - goto repeat; + do { + now = get_clock(); + sequence = ACCESS_ONCE(idle->sequence); + idle_time = ACCESS_ONCE(idle->idle_time); + idle_enter = ACCESS_ONCE(idle->idle_enter); + idle_exit = ACCESS_ONCE(idle->idle_exit); + } while ((sequence & 1) || (idle->sequence != sequence)); + idle_time += idle_enter ? ((idle_exit ? : now) - idle_enter) : 0; return sprintf(buf, "%llu\n", idle_time >> 12); } static DEVICE_ATTR(idle_time_us, 0444, show_idle_time, NULL); @@ -1021,7 +1030,7 @@ static int __cpuinit smp_cpu_notify(struct notifier_block *self, unsigned long action, void *hcpu) { unsigned int cpu = (unsigned int)(long)hcpu; - struct cpu *c = &per_cpu(cpu_devices, cpu); + struct cpu *c = &pcpu_devices[cpu].cpu; struct device *s = &c->dev; struct s390_idle_data *idle; int err = 0; @@ -1047,7 +1056,7 @@ static struct notifier_block __cpuinitdata smp_cpu_nb = { static int __devinit smp_add_present_cpu(int cpu) { - struct cpu *c = &per_cpu(cpu_devices, cpu); + struct cpu *c = &pcpu_devices[cpu].cpu; struct device *s = &c->dev; int rc; @@ -1085,29 +1094,21 @@ out: int __ref smp_rescan_cpus(void) { - cpumask_t newcpus; - int cpu; - int rc; + struct sclp_cpu_info *info; + int nr; + info = smp_get_cpu_info(); + if (!info) + return -ENOMEM; get_online_cpus(); mutex_lock(&smp_cpu_state_mutex); - cpumask_copy(&newcpus, cpu_present_mask); - rc = __smp_rescan_cpus(); - if (rc) - goto out; - cpumask_andnot(&newcpus, cpu_present_mask, &newcpus); - for_each_cpu(cpu, &newcpus) { - rc = smp_add_present_cpu(cpu); - if (rc) - set_cpu_present(cpu, false); - } - rc = 0; -out: + nr = __smp_rescan_cpus(info, 1); mutex_unlock(&smp_cpu_state_mutex); put_online_cpus(); - if (!cpumask_empty(&newcpus)) + kfree(info); + if (nr) topology_schedule_update(); - return rc; + return 0; } static ssize_t __ref rescan_store(struct device *dev, diff --git a/arch/s390/kernel/switch_cpu.S b/arch/s390/kernel/switch_cpu.S deleted file mode 100644 index bfe070bc765..00000000000 --- a/arch/s390/kernel/switch_cpu.S +++ /dev/null @@ -1,58 +0,0 @@ -/* - * 31-bit switch cpu code - * - * Copyright IBM Corp. 2009 - * - */ - -#include <linux/linkage.h> -#include <asm/asm-offsets.h> -#include <asm/ptrace.h> - -# smp_switch_to_cpu switches to destination cpu and executes the passed function -# Parameter: %r2 - function to call -# %r3 - function parameter -# %r4 - stack poiner -# %r5 - current cpu -# %r6 - destination cpu - - .section .text -ENTRY(smp_switch_to_cpu) - stm %r6,%r15,__SF_GPRS(%r15) - lr %r1,%r15 - ahi %r15,-STACK_FRAME_OVERHEAD - st %r1,__SF_BACKCHAIN(%r15) - basr %r13,0 -0: la %r1,.gprregs_addr-0b(%r13) - l %r1,0(%r1) - stm %r0,%r15,0(%r1) -1: sigp %r0,%r6,__SIGP_RESTART /* start destination CPU */ - brc 2,1b /* busy, try again */ -2: sigp %r0,%r5,__SIGP_STOP /* stop current CPU */ - brc 2,2b /* busy, try again */ -3: j 3b - -ENTRY(smp_restart_cpu) - basr %r13,0 -0: la %r1,.gprregs_addr-0b(%r13) - l %r1,0(%r1) - lm %r0,%r15,0(%r1) -1: sigp %r0,%r5,__SIGP_SENSE /* Wait for calling CPU */ - brc 10,1b /* busy, accepted (status 0), running */ - tmll %r0,0x40 /* Test if calling CPU is stopped */ - jz 1b - ltr %r4,%r4 /* New stack ? */ - jz 1f - lr %r15,%r4 -1: lr %r14,%r2 /* r14: Function to call */ - lr %r2,%r3 /* r2 : Parameter for function*/ - basr %r14,%r14 /* Call function */ - -.gprregs_addr: - .long .gprregs - - .section .data,"aw",@progbits -.gprregs: - .rept 16 - .long 0 - .endr diff --git a/arch/s390/kernel/switch_cpu64.S b/arch/s390/kernel/switch_cpu64.S deleted file mode 100644 index fcc42d799e4..00000000000 --- a/arch/s390/kernel/switch_cpu64.S +++ /dev/null @@ -1,51 +0,0 @@ -/* - * 64-bit switch cpu code - * - * Copyright IBM Corp. 2009 - * - */ - -#include <linux/linkage.h> -#include <asm/asm-offsets.h> -#include <asm/ptrace.h> - -# smp_switch_to_cpu switches to destination cpu and executes the passed function -# Parameter: %r2 - function to call -# %r3 - function parameter -# %r4 - stack poiner -# %r5 - current cpu -# %r6 - destination cpu - - .section .text -ENTRY(smp_switch_to_cpu) - stmg %r6,%r15,__SF_GPRS(%r15) - lgr %r1,%r15 - aghi %r15,-STACK_FRAME_OVERHEAD - stg %r1,__SF_BACKCHAIN(%r15) - larl %r1,.gprregs - stmg %r0,%r15,0(%r1) -1: sigp %r0,%r6,__SIGP_RESTART /* start destination CPU */ - brc 2,1b /* busy, try again */ -2: sigp %r0,%r5,__SIGP_STOP /* stop current CPU */ - brc 2,2b /* busy, try again */ -3: j 3b - -ENTRY(smp_restart_cpu) - larl %r1,.gprregs - lmg %r0,%r15,0(%r1) -1: sigp %r0,%r5,__SIGP_SENSE /* Wait for calling CPU */ - brc 10,1b /* busy, accepted (status 0), running */ - tmll %r0,0x40 /* Test if calling CPU is stopped */ - jz 1b - ltgr %r4,%r4 /* New stack ? */ - jz 1f - lgr %r15,%r4 -1: lgr %r14,%r2 /* r14: Function to call */ - lgr %r2,%r3 /* r2 : Parameter for function*/ - basr %r14,%r14 /* Call function */ - - .section .data,"aw",@progbits -.gprregs: - .rept 16 - .quad 0 - .endr diff --git a/arch/s390/kernel/swsusp_asm64.S b/arch/s390/kernel/swsusp_asm64.S index acb78cdee89..dd70ef04605 100644 --- a/arch/s390/kernel/swsusp_asm64.S +++ b/arch/s390/kernel/swsusp_asm64.S @@ -42,7 +42,7 @@ ENTRY(swsusp_arch_suspend) lghi %r1,0x1000 /* Save CPU address */ - stap __LC_CPU_ADDRESS(%r0) + stap __LC_EXT_CPU_ADDR(%r0) /* Store registers */ mvc 0x318(4,%r1),__SF_EMPTY(%r15) /* move prefix to lowcore */ @@ -173,15 +173,15 @@ pgm_check_entry: larl %r1,.Lresume_cpu /* Resume CPU address: r2 */ stap 0(%r1) llgh %r2,0(%r1) - llgh %r1,__LC_CPU_ADDRESS(%r0) /* Suspend CPU address: r1 */ + llgh %r1,__LC_EXT_CPU_ADDR(%r0) /* Suspend CPU address: r1 */ cgr %r1,%r2 je restore_registers /* r1 = r2 -> nothing to do */ larl %r4,.Lrestart_suspend_psw /* Set new restart PSW */ mvc __LC_RST_NEW_PSW(16,%r0),0(%r4) 3: - sigp %r9,%r1,__SIGP_INITIAL_CPU_RESET - brc 8,4f /* accepted */ - brc 2,3b /* busy, try again */ + sigp %r9,%r1,11 /* sigp initial cpu reset */ + brc 8,4f /* accepted */ + brc 2,3b /* busy, try again */ /* Suspend CPU not available -> panic */ larl %r15,init_thread_union @@ -196,10 +196,10 @@ pgm_check_entry: lpsw 0(%r3) 4: /* Switch to suspend CPU */ - sigp %r9,%r1,__SIGP_RESTART /* start suspend CPU */ + sigp %r9,%r1,6 /* sigp restart to suspend CPU */ brc 2,4b /* busy, try again */ 5: - sigp %r9,%r2,__SIGP_STOP /* stop resume (current) CPU */ + sigp %r9,%r2,5 /* sigp stop to current resume CPU */ brc 2,5b /* busy, try again */ 6: j 6b @@ -207,7 +207,7 @@ restart_suspend: larl %r1,.Lresume_cpu llgh %r2,0(%r1) 7: - sigp %r9,%r2,__SIGP_SENSE /* Wait for resume CPU */ + sigp %r9,%r2,1 /* sigp sense, wait for resume CPU */ brc 8,7b /* accepted, status 0, still running */ brc 2,7b /* busy, try again */ tmll %r9,0x40 /* Test if resume CPU is stopped */ @@ -257,6 +257,9 @@ restore_registers: lghi %r2,0 brasl %r14,arch_set_page_states + /* Log potential guest relocation */ + brasl %r14,lgr_info_log + /* Reinitialize the channel subsystem */ brasl %r14,channel_subsystem_reinit diff --git a/arch/s390/kernel/time.c b/arch/s390/kernel/time.c index 14da278febb..d4e1cb1dbcd 100644 --- a/arch/s390/kernel/time.c +++ b/arch/s390/kernel/time.c @@ -165,7 +165,7 @@ void init_cpu_timer(void) __ctl_set_bit(0, 4); } -static void clock_comparator_interrupt(unsigned int ext_int_code, +static void clock_comparator_interrupt(struct ext_code ext_code, unsigned int param32, unsigned long param64) { @@ -177,7 +177,7 @@ static void clock_comparator_interrupt(unsigned int ext_int_code, static void etr_timing_alert(struct etr_irq_parm *); static void stp_timing_alert(struct stp_irq_parm *); -static void timing_alert_interrupt(unsigned int ext_int_code, +static void timing_alert_interrupt(struct ext_code ext_code, unsigned int param32, unsigned long param64) { kstat_cpu(smp_processor_id()).irqs[EXTINT_TLA]++; diff --git a/arch/s390/kernel/topology.c b/arch/s390/kernel/topology.c index 7370a41948c..4f8dc942257 100644 --- a/arch/s390/kernel/topology.c +++ b/arch/s390/kernel/topology.c @@ -79,12 +79,12 @@ static struct mask_info *add_cpus_to_mask(struct topology_cpu *tl_cpu, cpu < TOPOLOGY_CPU_BITS; cpu = find_next_bit(&tl_cpu->mask[0], TOPOLOGY_CPU_BITS, cpu + 1)) { - unsigned int rcpu, lcpu; + unsigned int rcpu; + int lcpu; rcpu = TOPOLOGY_CPU_BITS - 1 - cpu + tl_cpu->origin; - for_each_present_cpu(lcpu) { - if (cpu_logical_map(lcpu) != rcpu) - continue; + lcpu = smp_find_processor_id(rcpu); + if (lcpu >= 0) { cpumask_set_cpu(lcpu, &book->mask); cpu_book_id[lcpu] = book->id; cpumask_set_cpu(lcpu, &core->mask); diff --git a/arch/s390/kernel/traps.c b/arch/s390/kernel/traps.c index 5ce3750b181..cd6ebe12c48 100644 --- a/arch/s390/kernel/traps.c +++ b/arch/s390/kernel/traps.c @@ -41,6 +41,7 @@ #include <asm/cpcmd.h> #include <asm/lowcore.h> #include <asm/debug.h> +#include <asm/ipl.h> #include "entry.h" void (*pgm_check_table[128])(struct pt_regs *regs); @@ -144,8 +145,8 @@ void show_stack(struct task_struct *task, unsigned long *sp) for (i = 0; i < kstack_depth_to_print; i++) { if (((addr_t) stack & (THREAD_SIZE-1)) == 0) break; - if (i && ((i * sizeof (long) % 32) == 0)) - printk("\n "); + if ((i * sizeof(long) % 32) == 0) + printk("%s ", i == 0 ? "" : "\n"); printk(LONG, *stack++); } printk("\n"); @@ -239,6 +240,7 @@ void die(struct pt_regs *regs, const char *str) static int die_counter; oops_enter(); + lgr_info_log(); debug_stop_all(); console_verbose(); spin_lock_irq(&die_lock); diff --git a/arch/s390/kernel/vdso.c b/arch/s390/kernel/vdso.c index d73630b4fe1..9c80138206b 100644 --- a/arch/s390/kernel/vdso.c +++ b/arch/s390/kernel/vdso.c @@ -89,18 +89,11 @@ static void vdso_init_data(struct vdso_data *vd) #ifdef CONFIG_64BIT /* - * Setup per cpu vdso data page. - */ -static void vdso_init_per_cpu_data(int cpu, struct vdso_per_cpu_data *vpcd) -{ -} - -/* * Allocate/free per cpu vdso data. */ #define SEGMENT_ORDER 2 -int vdso_alloc_per_cpu(int cpu, struct _lowcore *lowcore) +int vdso_alloc_per_cpu(struct _lowcore *lowcore) { unsigned long segment_table, page_table, page_frame; u32 *psal, *aste; @@ -139,7 +132,6 @@ int vdso_alloc_per_cpu(int cpu, struct _lowcore *lowcore) aste[4] = (u32)(addr_t) psal; lowcore->vdso_per_cpu_data = page_frame; - vdso_init_per_cpu_data(cpu, (struct vdso_per_cpu_data *) page_frame); return 0; out: @@ -149,7 +141,7 @@ out: return -ENOMEM; } -void vdso_free_per_cpu(int cpu, struct _lowcore *lowcore) +void vdso_free_per_cpu(struct _lowcore *lowcore) { unsigned long segment_table, page_table, page_frame; u32 *psal, *aste; @@ -168,19 +160,15 @@ void vdso_free_per_cpu(int cpu, struct _lowcore *lowcore) free_pages(segment_table, SEGMENT_ORDER); } -static void __vdso_init_cr5(void *dummy) +static void vdso_init_cr5(void) { unsigned long cr5; + if (user_mode == HOME_SPACE_MODE || !vdso_enabled) + return; cr5 = offsetof(struct _lowcore, paste); __ctl_load(cr5, 5, 5); } - -static void vdso_init_cr5(void) -{ - if (user_mode != HOME_SPACE_MODE && vdso_enabled) - on_each_cpu(__vdso_init_cr5, NULL, 1); -} #endif /* CONFIG_64BIT */ /* @@ -253,17 +241,11 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp) * on the "data" page of the vDSO or you'll stop getting kernel * updates and your nice userland gettimeofday will be totally dead. * It's fine to use that for setting breakpoints in the vDSO code - * pages though - * - * Make sure the vDSO gets into every core dump. - * Dumping its contents makes post-mortem fully interpretable later - * without matching up the same kernel and hardware config to see - * what PC values meant. + * pages though. */ rc = install_special_mapping(mm, vdso_base, vdso_pages << PAGE_SHIFT, VM_READ|VM_EXEC| - VM_MAYREAD|VM_MAYWRITE|VM_MAYEXEC| - VM_ALWAYSDUMP, + VM_MAYREAD|VM_MAYWRITE|VM_MAYEXEC, vdso_pagelist); if (rc) current->mm->context.vdso_base = 0; @@ -322,10 +304,8 @@ static int __init vdso_init(void) } vdso64_pagelist[vdso64_pages - 1] = virt_to_page(vdso_data); vdso64_pagelist[vdso64_pages] = NULL; -#ifndef CONFIG_SMP - if (vdso_alloc_per_cpu(0, &S390_lowcore)) + if (vdso_alloc_per_cpu(&S390_lowcore)) BUG(); -#endif vdso_init_cr5(); #endif /* CONFIG_64BIT */ @@ -335,7 +315,7 @@ static int __init vdso_init(void) return 0; } -arch_initcall(vdso_init); +early_initcall(vdso_init); int in_gate_area_no_mm(unsigned long addr) { diff --git a/arch/s390/kernel/vtime.c b/arch/s390/kernel/vtime.c index bb48977f546..39ebff50694 100644 --- a/arch/s390/kernel/vtime.c +++ b/arch/s390/kernel/vtime.c @@ -26,6 +26,7 @@ #include <asm/irq_regs.h> #include <asm/cputime.h> #include <asm/irq.h> +#include "entry.h" static DEFINE_PER_CPU(struct vtimer_queue, virt_cpu_timer); @@ -123,153 +124,53 @@ void account_system_vtime(struct task_struct *tsk) } EXPORT_SYMBOL_GPL(account_system_vtime); -void __kprobes vtime_start_cpu(__u64 int_clock, __u64 enter_timer) +void __kprobes vtime_stop_cpu(void) { struct s390_idle_data *idle = &__get_cpu_var(s390_idle); struct vtimer_queue *vq = &__get_cpu_var(virt_cpu_timer); - __u64 idle_time, expires; + unsigned long long idle_time; + unsigned long psw_mask; - if (idle->idle_enter == 0ULL) - return; + trace_hardirqs_on(); + /* Don't trace preempt off for idle. */ + stop_critical_timings(); - /* Account time spent with enabled wait psw loaded as idle time. */ - idle_time = int_clock - idle->idle_enter; - account_idle_time(idle_time); - S390_lowcore.steal_timer += - idle->idle_enter - S390_lowcore.last_update_clock; - S390_lowcore.last_update_clock = int_clock; - - /* Account system time spent going idle. */ - S390_lowcore.system_timer += S390_lowcore.last_update_timer - vq->idle; - S390_lowcore.last_update_timer = enter_timer; - - /* Restart vtime CPU timer */ - if (vq->do_spt) { - /* Program old expire value but first save progress. */ - expires = vq->idle - enter_timer; - expires += get_vtimer(); - set_vtimer(expires); - } else { - /* Don't account the CPU timer delta while the cpu was idle. */ - vq->elapsed -= vq->idle - enter_timer; - } + /* Wait for external, I/O or machine check interrupt. */ + psw_mask = psw_kernel_bits | PSW_MASK_WAIT | PSW_MASK_DAT | + PSW_MASK_IO | PSW_MASK_EXT | PSW_MASK_MCHECK; + idle->nohz_delay = 0; + /* Call the assembler magic in entry.S */ + psw_idle(idle, vq, psw_mask, !list_empty(&vq->list)); + + /* Reenable preemption tracer. */ + start_critical_timings(); + + /* Account time spent with enabled wait psw loaded as idle time. */ idle->sequence++; smp_wmb(); + idle_time = idle->idle_exit - idle->idle_enter; idle->idle_time += idle_time; - idle->idle_enter = 0ULL; + idle->idle_enter = idle->idle_exit = 0ULL; idle->idle_count++; + account_idle_time(idle_time); smp_wmb(); idle->sequence++; } -void __kprobes vtime_stop_cpu(void) -{ - struct s390_idle_data *idle = &__get_cpu_var(s390_idle); - struct vtimer_queue *vq = &__get_cpu_var(virt_cpu_timer); - psw_t psw; - - /* Wait for external, I/O or machine check interrupt. */ - psw.mask = psw_kernel_bits | PSW_MASK_WAIT | - PSW_MASK_DAT | PSW_MASK_IO | PSW_MASK_EXT | PSW_MASK_MCHECK; - - idle->nohz_delay = 0; - - /* Check if the CPU timer needs to be reprogrammed. */ - if (vq->do_spt) { - __u64 vmax = VTIMER_MAX_SLICE; - /* - * The inline assembly is equivalent to - * vq->idle = get_cpu_timer(); - * set_cpu_timer(VTIMER_MAX_SLICE); - * idle->idle_enter = get_clock(); - * __load_psw_mask(psw_kernel_bits | PSW_MASK_WAIT | - * PSW_MASK_DAT | PSW_MASK_IO | - * PSW_MASK_EXT | PSW_MASK_MCHECK); - * The difference is that the inline assembly makes sure that - * the last three instruction are stpt, stck and lpsw in that - * order. This is done to increase the precision. - */ - asm volatile( -#ifndef CONFIG_64BIT - " basr 1,0\n" - "0: ahi 1,1f-0b\n" - " st 1,4(%2)\n" -#else /* CONFIG_64BIT */ - " larl 1,1f\n" - " stg 1,8(%2)\n" -#endif /* CONFIG_64BIT */ - " stpt 0(%4)\n" - " spt 0(%5)\n" - " stck 0(%3)\n" -#ifndef CONFIG_64BIT - " lpsw 0(%2)\n" -#else /* CONFIG_64BIT */ - " lpswe 0(%2)\n" -#endif /* CONFIG_64BIT */ - "1:" - : "=m" (idle->idle_enter), "=m" (vq->idle) - : "a" (&psw), "a" (&idle->idle_enter), - "a" (&vq->idle), "a" (&vmax), "m" (vmax), "m" (psw) - : "memory", "cc", "1"); - } else { - /* - * The inline assembly is equivalent to - * vq->idle = get_cpu_timer(); - * idle->idle_enter = get_clock(); - * __load_psw_mask(psw_kernel_bits | PSW_MASK_WAIT | - * PSW_MASK_DAT | PSW_MASK_IO | - * PSW_MASK_EXT | PSW_MASK_MCHECK); - * The difference is that the inline assembly makes sure that - * the last three instruction are stpt, stck and lpsw in that - * order. This is done to increase the precision. - */ - asm volatile( -#ifndef CONFIG_64BIT - " basr 1,0\n" - "0: ahi 1,1f-0b\n" - " st 1,4(%2)\n" -#else /* CONFIG_64BIT */ - " larl 1,1f\n" - " stg 1,8(%2)\n" -#endif /* CONFIG_64BIT */ - " stpt 0(%4)\n" - " stck 0(%3)\n" -#ifndef CONFIG_64BIT - " lpsw 0(%2)\n" -#else /* CONFIG_64BIT */ - " lpswe 0(%2)\n" -#endif /* CONFIG_64BIT */ - "1:" - : "=m" (idle->idle_enter), "=m" (vq->idle) - : "a" (&psw), "a" (&idle->idle_enter), - "a" (&vq->idle), "m" (psw) - : "memory", "cc", "1"); - } -} - cputime64_t s390_get_idle_time(int cpu) { - struct s390_idle_data *idle; - unsigned long long now, idle_time, idle_enter; + struct s390_idle_data *idle = &per_cpu(s390_idle, cpu); + unsigned long long now, idle_enter, idle_exit; unsigned int sequence; - idle = &per_cpu(s390_idle, cpu); - - now = get_clock(); -repeat: - sequence = idle->sequence; - smp_rmb(); - if (sequence & 1) - goto repeat; - idle_time = 0; - idle_enter = idle->idle_enter; - if (idle_enter != 0ULL && idle_enter < now) - idle_time = now - idle_enter; - smp_rmb(); - if (idle->sequence != sequence) - goto repeat; - return idle_time; + do { + now = get_clock(); + sequence = ACCESS_ONCE(idle->sequence); + idle_enter = ACCESS_ONCE(idle->idle_enter); + idle_exit = ACCESS_ONCE(idle->idle_exit); + } while ((sequence & 1) || (idle->sequence != sequence)); + return idle_enter ? ((idle_exit ? : now) - idle_enter) : 0; } /* @@ -319,7 +220,7 @@ static void do_callbacks(struct list_head *cb_list) /* * Handler for the virtual CPU timer. */ -static void do_cpu_timer_interrupt(unsigned int ext_int_code, +static void do_cpu_timer_interrupt(struct ext_code ext_code, unsigned int param32, unsigned long param64) { struct vtimer_queue *vq; @@ -346,7 +247,6 @@ static void do_cpu_timer_interrupt(unsigned int ext_int_code, } spin_unlock(&vq->lock); - vq->do_spt = list_empty(&cb_list); do_callbacks(&cb_list); /* next event is first in list */ @@ -355,8 +255,7 @@ static void do_cpu_timer_interrupt(unsigned int ext_int_code, if (!list_empty(&vq->list)) { event = list_first_entry(&vq->list, struct vtimer_list, entry); next = event->expires; - } else - vq->do_spt = 0; + } spin_unlock(&vq->lock); /* * To improve precision add the time spent by the @@ -570,6 +469,9 @@ void init_cpu_vtimer(void) /* enable cpu timer interrupts */ __ctl_set_bit(0,10); + + /* set initial cpu timer */ + set_vtimer(0x7fffffffffffffffULL); } static int __cpuinit s390_nohz_notify(struct notifier_block *self, |