diff options
Diffstat (limited to 'arch/sh')
32 files changed, 1149 insertions, 672 deletions
diff --git a/arch/sh/Kconfig b/arch/sh/Kconfig index ae6c7368903..4eaf3401a15 100644 --- a/arch/sh/Kconfig +++ b/arch/sh/Kconfig @@ -42,6 +42,8 @@ config SUPERH32 select HAVE_FTRACE_NMI_ENTER if DYNAMIC_FTRACE select HAVE_FUNCTION_GRAPH_TRACER select HAVE_ARCH_KGDB + select HAVE_HW_BREAKPOINT + select PERF_EVENTS if HAVE_HW_BREAKPOINT select ARCH_HIBERNATION_POSSIBLE if MMU config SUPERH64 diff --git a/arch/sh/include/asm/Kbuild b/arch/sh/include/asm/Kbuild index e121c30f797..46cb93477bc 100644 --- a/arch/sh/include/asm/Kbuild +++ b/arch/sh/include/asm/Kbuild @@ -1,6 +1,8 @@ include include/asm-generic/Kbuild.asm -header-y += cachectl.h cpu-features.h +header-y += cachectl.h +header-y += cpu-features.h +header-y += hw_breakpoint.h unifdef-y += unistd_32.h unifdef-y += unistd_64.h diff --git a/arch/sh/include/asm/fpu.h b/arch/sh/include/asm/fpu.h index fb6bbb9b1cc..06c4281aab6 100644 --- a/arch/sh/include/asm/fpu.h +++ b/arch/sh/include/asm/fpu.h @@ -2,8 +2,8 @@ #define __ASM_SH_FPU_H #ifndef __ASSEMBLY__ -#include <linux/preempt.h> -#include <asm/ptrace.h> + +struct task_struct; #ifdef CONFIG_SH_FPU static inline void release_fpu(struct pt_regs *regs) @@ -16,22 +16,23 @@ static inline void grab_fpu(struct pt_regs *regs) regs->sr &= ~SR_FD; } -struct task_struct; - extern void save_fpu(struct task_struct *__tsk); -void fpu_state_restore(struct pt_regs *regs); +extern void restore_fpu(struct task_struct *__tsk); +extern void fpu_state_restore(struct pt_regs *regs); +extern void __fpu_state_restore(void); #else - -#define save_fpu(tsk) do { } while (0) -#define release_fpu(regs) do { } while (0) -#define grab_fpu(regs) do { } while (0) -#define fpu_state_restore(regs) do { } while (0) - +#define save_fpu(tsk) do { } while (0) +#define restore_fpu(tsk) do { } while (0) +#define release_fpu(regs) do { } while (0) +#define grab_fpu(regs) do { } while (0) +#define fpu_state_restore(regs) do { } while (0) +#define __fpu_state_restore(regs) do { } while (0) #endif struct user_regset; extern int do_fpu_inst(unsigned short, struct pt_regs *); +extern int init_fpu(struct task_struct *); extern int fpregs_get(struct task_struct *target, const struct user_regset *regset, @@ -65,18 +66,6 @@ static inline void clear_fpu(struct task_struct *tsk, struct pt_regs *regs) preempt_enable(); } -static inline int init_fpu(struct task_struct *tsk) -{ - if (tsk_used_math(tsk)) { - if ((boot_cpu_data.flags & CPU_HAS_FPU) && tsk == current) - unlazy_fpu(tsk, task_pt_regs(tsk)); - return 0; - } - - set_stopped_child_used_math(tsk); - return 0; -} - #endif /* __ASSEMBLY__ */ #endif /* __ASM_SH_FPU_H */ diff --git a/arch/sh/include/asm/hw_breakpoint.h b/arch/sh/include/asm/hw_breakpoint.h new file mode 100644 index 00000000000..7295d629024 --- /dev/null +++ b/arch/sh/include/asm/hw_breakpoint.h @@ -0,0 +1,67 @@ +#ifndef __ASM_SH_HW_BREAKPOINT_H +#define __ASM_SH_HW_BREAKPOINT_H + +#include <linux/kdebug.h> +#include <linux/types.h> + +#ifdef __KERNEL__ +#define __ARCH_HW_BREAKPOINT_H + +struct arch_hw_breakpoint { + char *name; /* Contains name of the symbol to set bkpt */ + unsigned long address; + u16 len; + u16 type; +}; + +enum { + SH_BREAKPOINT_READ = (1 << 1), + SH_BREAKPOINT_WRITE = (1 << 2), + SH_BREAKPOINT_RW = SH_BREAKPOINT_READ | SH_BREAKPOINT_WRITE, + + SH_BREAKPOINT_LEN_1 = (1 << 12), + SH_BREAKPOINT_LEN_2 = (1 << 13), + SH_BREAKPOINT_LEN_4 = SH_BREAKPOINT_LEN_1 | SH_BREAKPOINT_LEN_2, + SH_BREAKPOINT_LEN_8 = (1 << 14), +}; + +struct sh_ubc { + const char *name; + unsigned int num_events; + unsigned int trap_nr; + void (*enable)(struct arch_hw_breakpoint *, int); + void (*disable)(struct arch_hw_breakpoint *, int); + void (*enable_all)(unsigned long); + void (*disable_all)(void); + unsigned long (*active_mask)(void); + unsigned long (*triggered_mask)(void); + void (*clear_triggered_mask)(unsigned long); + struct clk *clk; /* optional interface clock / MSTP bit */ +}; + +struct perf_event; +struct task_struct; +struct pmu; + +/* Maximum number of UBC channels */ +#define HBP_NUM 2 + +/* arch/sh/kernel/hw_breakpoint.c */ +extern int arch_check_va_in_userspace(unsigned long va, u16 hbp_len); +extern int arch_validate_hwbkpt_settings(struct perf_event *bp, + struct task_struct *tsk); +extern int hw_breakpoint_exceptions_notify(struct notifier_block *unused, + unsigned long val, void *data); + +int arch_install_hw_breakpoint(struct perf_event *bp); +void arch_uninstall_hw_breakpoint(struct perf_event *bp); +void hw_breakpoint_pmu_read(struct perf_event *bp); +void hw_breakpoint_pmu_unthrottle(struct perf_event *bp); + +extern void arch_fill_perf_breakpoint(struct perf_event *bp); +extern int register_sh_ubc(struct sh_ubc *); + +extern struct pmu perf_ops_bp; + +#endif /* __KERNEL__ */ +#endif /* __ASM_SH_HW_BREAKPOINT_H */ diff --git a/arch/sh/include/asm/kdebug.h b/arch/sh/include/asm/kdebug.h index 985219f9759..5f6d2e9ccb7 100644 --- a/arch/sh/include/asm/kdebug.h +++ b/arch/sh/include/asm/kdebug.h @@ -6,6 +6,8 @@ enum die_val { DIE_TRAP, DIE_NMI, DIE_OOPS, + DIE_BREAKPOINT, + DIE_SSTEP, }; #endif /* __ASM_SH_KDEBUG_H */ diff --git a/arch/sh/include/asm/processor_32.h b/arch/sh/include/asm/processor_32.h index 1f3d6fab660..5fd83125fb8 100644 --- a/arch/sh/include/asm/processor_32.h +++ b/arch/sh/include/asm/processor_32.h @@ -14,6 +14,7 @@ #include <asm/page.h> #include <asm/types.h> #include <asm/ptrace.h> +#include <asm/hw_breakpoint.h> /* * Default implementation of macro that returns current @@ -90,48 +91,40 @@ struct sh_fpu_soft_struct { unsigned long entry_pc; }; -union sh_fpu_union { - struct sh_fpu_hard_struct hard; - struct sh_fpu_soft_struct soft; +union thread_xstate { + struct sh_fpu_hard_struct hardfpu; + struct sh_fpu_soft_struct softfpu; }; +extern unsigned int xstate_size; +extern void free_thread_xstate(struct task_struct *); +extern struct kmem_cache *task_xstate_cachep; + struct thread_struct { /* Saved registers when thread is descheduled */ unsigned long sp; unsigned long pc; - /* Hardware debugging registers */ - unsigned long ubc_pc; - - /* floating point info */ - union sh_fpu_union fpu; + /* Save middle states of ptrace breakpoints */ + struct perf_event *ptrace_bps[HBP_NUM]; #ifdef CONFIG_SH_DSP /* Dsp status information */ struct sh_dsp_struct dsp_status; #endif -}; -/* Count of active tasks with UBC settings */ -extern int ubc_usercnt; + /* Extended processor state */ + union thread_xstate *xstate; +}; #define INIT_THREAD { \ .sp = sizeof(init_stack) + (long) &init_stack, \ } -/* - * Do necessary setup to start up a newly executed thread. - */ -#define start_thread(_regs, new_pc, new_sp) \ - set_fs(USER_DS); \ - _regs->pr = 0; \ - _regs->sr = SR_FD; /* User mode. */ \ - _regs->pc = new_pc; \ - _regs->regs[15] = new_sp - /* Forward declaration, a strange C thing */ struct task_struct; -struct mm_struct; + +extern void start_thread(struct pt_regs *regs, unsigned long new_pc, unsigned long new_sp); /* Free all resources held by a thread. */ extern void release_thread(struct task_struct *); diff --git a/arch/sh/include/asm/ptrace.h b/arch/sh/include/asm/ptrace.h index 1dc12cb44a2..201d11ef211 100644 --- a/arch/sh/include/asm/ptrace.h +++ b/arch/sh/include/asm/ptrace.h @@ -124,6 +124,12 @@ struct task_struct; extern void user_enable_single_step(struct task_struct *); extern void user_disable_single_step(struct task_struct *); +struct perf_event; +struct perf_sample_data; + +extern void ptrace_triggered(struct perf_event *bp, int nmi, + struct perf_sample_data *data, struct pt_regs *regs); + #define task_pt_regs(task) \ ((struct pt_regs *) (task_stack_page(task) + THREAD_SIZE) - 1) diff --git a/arch/sh/include/asm/system.h b/arch/sh/include/asm/system.h index b74d2532cc2..62e4fc1e440 100644 --- a/arch/sh/include/asm/system.h +++ b/arch/sh/include/asm/system.h @@ -144,8 +144,6 @@ void per_cpu_trap_init(void); void default_idle(void); void cpu_idle_wait(void); -asmlinkage void break_point_trap(void); - #ifdef CONFIG_SUPERH32 #define BUILD_TRAP_HANDLER(name) \ asmlinkage void name##_trap_handler(unsigned long r4, unsigned long r5, \ diff --git a/arch/sh/include/asm/thread_info.h b/arch/sh/include/asm/thread_info.h index 1f3d927e226..55a36fef687 100644 --- a/arch/sh/include/asm/thread_info.h +++ b/arch/sh/include/asm/thread_info.h @@ -93,14 +93,16 @@ static inline struct thread_info *current_thread_info(void) #define THREAD_SIZE_ORDER (THREAD_SHIFT - PAGE_SHIFT) -#else /* THREAD_SHIFT < PAGE_SHIFT */ - -#define __HAVE_ARCH_THREAD_INFO_ALLOCATOR +#endif extern struct thread_info *alloc_thread_info(struct task_struct *tsk); extern void free_thread_info(struct thread_info *ti); +extern void arch_task_cache_init(void); +#define arch_task_cache_init arch_task_cache_init +extern int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src); +extern void init_thread_xstate(void); -#endif /* THREAD_SHIFT < PAGE_SHIFT */ +#define __HAVE_ARCH_THREAD_INFO_ALLOCATOR #endif /* __ASSEMBLY__ */ diff --git a/arch/sh/include/asm/ubc.h b/arch/sh/include/asm/ubc.h deleted file mode 100644 index 9bf96168443..00000000000 --- a/arch/sh/include/asm/ubc.h +++ /dev/null @@ -1,64 +0,0 @@ -/* - * include/asm-sh/ubc.h - * - * Copyright (C) 1999 Niibe Yutaka - * Copyright (C) 2002, 2003 Paul Mundt - * - * This file is subject to the terms and conditions of the GNU General Public - * License. See the file "COPYING" in the main directory of this archive - * for more details. - */ -#ifndef __ASM_SH_UBC_H -#define __ASM_SH_UBC_H -#ifdef __KERNEL__ - -#include <cpu/ubc.h> - -/* User Break Controller */ -#if defined(CONFIG_CPU_SUBTYPE_SH7707) || defined(CONFIG_CPU_SUBTYPE_SH7709) -#define UBC_TYPE_SH7729 (current_cpu_data.type == CPU_SH7729) -#else -#define UBC_TYPE_SH7729 0 -#endif - -#define BAMR_ASID (1 << 2) -#define BAMR_NONE 0 -#define BAMR_10 0x1 -#define BAMR_12 0x2 -#define BAMR_ALL 0x3 -#define BAMR_16 0x8 -#define BAMR_20 0x9 - -#define BBR_INST (1 << 4) -#define BBR_DATA (2 << 4) -#define BBR_READ (1 << 2) -#define BBR_WRITE (2 << 2) -#define BBR_BYTE 0x1 -#define BBR_HALF 0x2 -#define BBR_LONG 0x3 -#define BBR_QUAD (1 << 6) /* SH7750 */ -#define BBR_CPU (1 << 6) /* SH7709A,SH7729 */ -#define BBR_DMA (2 << 6) /* SH7709A,SH7729 */ - -#define BRCR_CMFA (1 << 15) -#define BRCR_CMFB (1 << 14) - -#if defined CONFIG_CPU_SH2A -#define BRCR_CMFCA (1 << 15) -#define BRCR_CMFCB (1 << 14) -#define BRCR_CMFDA (1 << 13) -#define BRCR_CMFDB (1 << 12) -#define BRCR_PCBB (1 << 6) /* 1: after execution */ -#define BRCR_PCBA (1 << 5) /* 1: after execution */ -#define BRCR_PCTE 0 -#else -#define BRCR_PCTE (1 << 11) -#define BRCR_PCBA (1 << 10) /* 1: after execution */ -#define BRCR_DBEB (1 << 7) -#define BRCR_PCBB (1 << 6) -#define BRCR_SEQ (1 << 3) -#define BRCR_UBDE (1 << 0) -#endif - -#endif /* __KERNEL__ */ -#endif /* __ASM_SH_UBC_H */ diff --git a/arch/sh/include/cpu-sh2/cpu/ubc.h b/arch/sh/include/cpu-sh2/cpu/ubc.h deleted file mode 100644 index ba0e87f19c7..00000000000 --- a/arch/sh/include/cpu-sh2/cpu/ubc.h +++ /dev/null @@ -1,32 +0,0 @@ -/* - * include/asm-sh/cpu-sh2/ubc.h - * - * Copyright (C) 2003 Paul Mundt - * - * This file is subject to the terms and conditions of the GNU General Public - * License. See the file "COPYING" in the main directory of this archive - * for more details. - */ -#ifndef __ASM_CPU_SH2_UBC_H -#define __ASM_CPU_SH2_UBC_H - -#define UBC_BARA 0xffffff40 -#define UBC_BAMRA 0xffffff44 -#define UBC_BBRA 0xffffff48 -#define UBC_BARB 0xffffff60 -#define UBC_BAMRB 0xffffff64 -#define UBC_BBRB 0xffffff68 -#define UBC_BDRB 0xffffff70 -#define UBC_BDMRB 0xffffff74 -#define UBC_BRCR 0xffffff78 - -/* - * We don't have any ASID changes to make in the UBC on the SH-2. - * - * Make these purposely invalid to track misuse. - */ -#define UBC_BASRA 0x00000000 -#define UBC_BASRB 0x00000000 - -#endif /* __ASM_CPU_SH2_UBC_H */ - diff --git a/arch/sh/include/cpu-sh3/cpu/ubc.h b/arch/sh/include/cpu-sh3/cpu/ubc.h deleted file mode 100644 index 4e6381d5ff7..00000000000 --- a/arch/sh/include/cpu-sh3/cpu/ubc.h +++ /dev/null @@ -1,42 +0,0 @@ -/* - * include/asm-sh/cpu-sh3/ubc.h - * - * Copyright (C) 1999 Niibe Yutaka - * Copyright (C) 2003 Paul Mundt - * - * This file is subject to the terms and conditions of the GNU General Public - * License. See the file "COPYING" in the main directory of this archive - * for more details. - */ -#ifndef __ASM_CPU_SH3_UBC_H -#define __ASM_CPU_SH3_UBC_H - -#if defined(CONFIG_CPU_SUBTYPE_SH7710) || \ - defined(CONFIG_CPU_SUBTYPE_SH7720) || \ - defined(CONFIG_CPU_SUBTYPE_SH7721) -#define UBC_BARA 0xa4ffffb0 -#define UBC_BAMRA 0xa4ffffb4 -#define UBC_BBRA 0xa4ffffb8 -#define UBC_BASRA 0xffffffe4 -#define UBC_BARB 0xa4ffffa0 -#define UBC_BAMRB 0xa4ffffa4 -#define UBC_BBRB 0xa4ffffa8 -#define UBC_BASRB 0xffffffe8 -#define UBC_BDRB 0xa4ffff90 -#define UBC_BDMRB 0xa4ffff94 -#define UBC_BRCR 0xa4ffff98 -#else -#define UBC_BARA 0xffffffb0 -#define UBC_BAMRA 0xffffffb4 -#define UBC_BBRA 0xffffffb8 -#define UBC_BASRA 0xffffffe4 -#define UBC_BARB 0xffffffa0 -#define UBC_BAMRB 0xffffffa4 -#define UBC_BBRB 0xffffffa8 -#define UBC_BASRB 0xffffffe8 -#define UBC_BDRB 0xffffff90 -#define UBC_BDMRB 0xffffff94 -#define UBC_BRCR 0xffffff98 -#endif - -#endif /* __ASM_CPU_SH3_UBC_H */ diff --git a/arch/sh/include/cpu-sh4/cpu/ubc.h b/arch/sh/include/cpu-sh4/cpu/ubc.h deleted file mode 100644 index c86e1705093..00000000000 --- a/arch/sh/include/cpu-sh4/cpu/ubc.h +++ /dev/null @@ -1,64 +0,0 @@ -/* - * include/asm-sh/cpu-sh4/ubc.h - * - * Copyright (C) 1999 Niibe Yutaka - * Copyright (C) 2003 Paul Mundt - * Copyright (C) 2006 Lineo Solutions Inc. support SH4A UBC - * - * This file is subject to the terms and conditions of the GNU General Public - * License. See the file "COPYING" in the main directory of this archive - * for more details. - */ -#ifndef __ASM_CPU_SH4_UBC_H -#define __ASM_CPU_SH4_UBC_H - -#if defined(CONFIG_CPU_SH4A) -#define UBC_CBR0 0xff200000 -#define UBC_CRR0 0xff200004 -#define UBC_CAR0 0xff200008 -#define UBC_CAMR0 0xff20000c -#define UBC_CBR1 0xff200020 -#define UBC_CRR1 0xff200024 -#define UBC_CAR1 0xff200028 -#define UBC_CAMR1 0xff20002c -#define UBC_CDR1 0xff200030 -#define UBC_CDMR1 0xff200034 -#define UBC_CETR1 0xff200038 -#define UBC_CCMFR 0xff200600 -#define UBC_CBCR 0xff200620 - -/* CBR */ -#define UBC_CBR_AIE (0x01<<30) -#define UBC_CBR_ID_INST (0x01<<4) -#define UBC_CBR_RW_READ (0x01<<1) -#define UBC_CBR_CE (0x01) - -#define UBC_CBR_AIV_MASK (0x00FF0000) -#define UBC_CBR_AIV_SHIFT (16) -#define UBC_CBR_AIV_SET(asid) (((asid)<<UBC_CBR_AIV_SHIFT) & UBC_CBR_AIV_MASK) - -#define UBC_CBR_INIT 0x20000000 - -/* CRR */ -#define UBC_CRR_RES (0x01<<13) -#define UBC_CRR_PCB (0x01<<1) -#define UBC_CRR_BIE (0x01) - -#define UBC_CRR_INIT 0x00002000 - -#else /* CONFIG_CPU_SH4 */ -#define UBC_BARA 0xff200000 -#define UBC_BAMRA 0xff200004 -#define UBC_BBRA 0xff200008 -#define UBC_BASRA 0xff000014 -#define UBC_BARB 0xff20000c -#define UBC_BAMRB 0xff200010 -#define UBC_BBRB 0xff200014 -#define UBC_BASRB 0xff000018 -#define UBC_BDRB 0xff200018 -#define UBC_BDMRB 0xff20001c -#define UBC_BRCR 0xff200020 -#endif /* CONFIG_CPU_SH4 */ - -#endif /* __ASM_CPU_SH4_UBC_H */ - diff --git a/arch/sh/kernel/Makefile b/arch/sh/kernel/Makefile index 5bec10c8bd7..56704a6d723 100644 --- a/arch/sh/kernel/Makefile +++ b/arch/sh/kernel/Makefile @@ -13,8 +13,8 @@ CFLAGS_REMOVE_return_address.o = -pg obj-y := debugtraps.o dma-nommu.o dumpstack.o \ idle.o io.o io_generic.o irq.o \ - irq_$(BITS).o machvec.o nmi_debug.o process_$(BITS).o \ - ptrace_$(BITS).o return_address.o \ + irq_$(BITS).o machvec.o nmi_debug.o process.o \ + process_$(BITS).o ptrace_$(BITS).o return_address.o \ setup.o signal_$(BITS).o sys_sh.o sys_sh$(BITS).o \ syscalls_$(BITS).o time.o topology.o traps.o \ traps_$(BITS).o unwinder.o @@ -39,6 +39,7 @@ obj-$(CONFIG_HIBERNATION) += swsusp.o obj-$(CONFIG_DWARF_UNWINDER) += dwarf.o obj-$(CONFIG_PERF_EVENTS) += perf_event.o perf_callchain.o +obj-$(CONFIG_HAVE_HW_BREAKPOINT) += hw_breakpoint.o obj-$(CONFIG_GENERIC_CLOCKEVENTS_BROADCAST) += localtimer.o EXTRA_CFLAGS += -Werror diff --git a/arch/sh/kernel/cpu/Makefile b/arch/sh/kernel/cpu/Makefile index d97c803719e..0e48bc61c27 100644 --- a/arch/sh/kernel/cpu/Makefile +++ b/arch/sh/kernel/cpu/Makefile @@ -17,5 +17,7 @@ obj-$(CONFIG_ARCH_SHMOBILE) += shmobile/ obj-$(CONFIG_SH_ADC) += adc.o obj-$(CONFIG_SH_CLK_CPG) += clock-cpg.o +obj-$(CONFIG_SH_FPU) += fpu.o +obj-$(CONFIG_SH_FPU_EMU) += fpu.o obj-y += irq/ init.o clock.o hwblk.o diff --git a/arch/sh/kernel/cpu/fpu.c b/arch/sh/kernel/cpu/fpu.c new file mode 100644 index 00000000000..c23e6727002 --- /dev/null +++ b/arch/sh/kernel/cpu/fpu.c @@ -0,0 +1,82 @@ +#include <linux/sched.h> +#include <asm/processor.h> +#include <asm/fpu.h> + +int init_fpu(struct task_struct *tsk) +{ + if (tsk_used_math(tsk)) { + if ((boot_cpu_data.flags & CPU_HAS_FPU) && tsk == current) + unlazy_fpu(tsk, task_pt_regs(tsk)); + return 0; + } + + /* + * Memory allocation at the first usage of the FPU and other state. + */ + if (!tsk->thread.xstate) { + tsk->thread.xstate = kmem_cache_alloc(task_xstate_cachep, + GFP_KERNEL); + if (!tsk->thread.xstate) + return -ENOMEM; + } + + if (boot_cpu_data.flags & CPU_HAS_FPU) { + struct sh_fpu_hard_struct *fp = &tsk->thread.xstate->hardfpu; + memset(fp, 0, xstate_size); + fp->fpscr = FPSCR_INIT; + } else { + struct sh_fpu_soft_struct *fp = &tsk->thread.xstate->softfpu; + memset(fp, 0, xstate_size); + fp->fpscr = FPSCR_INIT; + } + + set_stopped_child_used_math(tsk); + return 0; +} + +#ifdef CONFIG_SH_FPU +void __fpu_state_restore(void) +{ + struct task_struct *tsk = current; + + restore_fpu(tsk); + + task_thread_info(tsk)->status |= TS_USEDFPU; + tsk->fpu_counter++; +} + +void fpu_state_restore(struct pt_regs *regs) +{ + struct task_struct *tsk = current; + + if (unlikely(!user_mode(regs))) { + printk(KERN_ERR "BUG: FPU is used in kernel mode.\n"); + BUG(); + return; + } + + if (!tsk_used_math(tsk)) { + /* + * does a slab alloc which can sleep + */ + if (init_fpu(tsk)) { + /* + * ran out of memory! + */ + do_group_exit(SIGKILL); + return; + } + } + + grab_fpu(regs); + + __fpu_state_restore(); +} + +BUILD_TRAP_HANDLER(fpu_state_restore) +{ + TRAP_HANDLER_DECL; + + fpu_state_restore(regs); +} +#endif /* CONFIG_SH_FPU */ diff --git a/arch/sh/kernel/cpu/init.c b/arch/sh/kernel/cpu/init.c index 89b4b76c0d7..a5bb0550bbf 100644 --- a/arch/sh/kernel/cpu/init.c +++ b/arch/sh/kernel/cpu/init.c @@ -24,22 +24,31 @@ #include <asm/elf.h> #include <asm/io.h> #include <asm/smp.h> -#ifdef CONFIG_SUPERH32 -#include <asm/ubc.h> + +#ifdef CONFIG_SH_FPU +#define cpu_has_fpu 1 +#else +#define cpu_has_fpu 0 +#endif + +#ifdef CONFIG_SH_DSP +#define cpu_has_dsp 1 +#else +#define cpu_has_dsp 0 #endif /* * Generic wrapper for command line arguments to disable on-chip * peripherals (nofpu, nodsp, and so forth). */ -#define onchip_setup(x) \ -static int x##_disabled __initdata = 0; \ - \ -static int __init x##_setup(char *opts) \ -{ \ - x##_disabled = 1; \ - return 1; \ -} \ +#define onchip_setup(x) \ +static int x##_disabled __initdata = !cpu_has_##x; \ + \ +static int __init x##_setup(char *opts) \ +{ \ + x##_disabled = 1; \ + return 1; \ +} \ __setup("no" __stringify(x), x##_setup); onchip_setup(fpu); @@ -207,6 +216,18 @@ static void detect_cache_shape(void) l2_cache_shape = -1; /* No S-cache */ } +static void __init fpu_init(void) +{ + /* Disable the FPU */ + if (fpu_disabled && (current_cpu_data.flags & CPU_HAS_FPU)) { + printk("FPU Disabled\n"); + current_cpu_data.flags &= ~CPU_HAS_FPU; + } + + disable_fpu(); + clear_used_math(); +} + #ifdef CONFIG_SH_DSP static void __init release_dsp(void) { @@ -244,28 +265,35 @@ static void __init dsp_init(void) if (sr & SR_DSP) current_cpu_data.flags |= CPU_HAS_DSP; + /* Disable the DSP */ + if (dsp_disabled && (current_cpu_data.flags & CPU_HAS_DSP)) { + printk("DSP Disabled\n"); + current_cpu_data.flags &= ~CPU_HAS_DSP; + } + /* Now that we've determined the DSP status, clear the DSP bit. */ release_dsp(); } +#else +static inline void __init dsp_init(void) { } #endif /* CONFIG_SH_DSP */ /** * sh_cpu_init * - * This is our initial entry point for each CPU, and is invoked on the boot - * CPU prior to calling start_kernel(). For SMP, a combination of this and - * start_secondary() will bring up each processor to a ready state prior - * to hand forking the idle loop. + * This is our initial entry point for each CPU, and is invoked on the + * boot CPU prior to calling start_kernel(). For SMP, a combination of + * this and start_secondary() will bring up each processor to a ready + * state prior to hand forking the idle loop. * - * We do all of the basic processor init here, including setting up the - * caches, FPU, DSP, kicking the UBC, etc. By the time start_kernel() is - * hit (and subsequently platform_setup()) things like determining the - * CPU subtype and initial configuration will all be done. + * We do all of the basic processor init here, including setting up + * the caches, FPU, DSP, etc. By the time start_kernel() is hit (and + * subsequently platform_setup()) things like determining the CPU + * subtype and initial configuration will all be done. * * Each processor family is still responsible for doing its own probing * and cache configuration in detect_cpu_and_cache_system(). */ - asmlinkage void __init sh_cpu_init(void) { current_thread_info()->cpu = hard_smp_processor_id(); @@ -302,18 +330,8 @@ asmlinkage void __init sh_cpu_init(void) detect_cache_shape(); } - /* Disable the FPU */ - if (fpu_disabled) { - printk("FPU Disabled\n"); - current_cpu_data.flags &= ~CPU_HAS_FPU; - } - - /* FPU initialization */ - disable_fpu(); - if ((current_cpu_data.flags & CPU_HAS_FPU)) { - current_thread_info()->status &= ~TS_USEDFPU; - clear_used_math(); - } + fpu_init(); + dsp_init(); /* * Initialize the per-CPU ASID cache very early, since the @@ -321,18 +339,12 @@ asmlinkage void __init sh_cpu_init(void) */ current_cpu_data.asid_cache = NO_CONTEXT; -#ifdef CONFIG_SH_DSP - /* Probe for DSP */ - dsp_init(); - - /* Disable the DSP */ - if (dsp_disabled) { - printk("DSP Disabled\n"); - current_cpu_data.flags &= ~CPU_HAS_DSP; - release_dsp(); - } -#endif - speculative_execution_init(); expmask_init(); + + /* + * Boot processor to setup the FP and extended state context info. + */ + if (raw_smp_processor_id() == 0) + init_thread_xstate(); } diff --git a/arch/sh/kernel/cpu/sh2a/fpu.c b/arch/sh/kernel/cpu/sh2a/fpu.c index d395ce5740e..488d24e0cdf 100644 --- a/arch/sh/kernel/cpu/sh2a/fpu.c +++ b/arch/sh/kernel/cpu/sh2a/fpu.c @@ -26,8 +26,7 @@ /* * Save FPU registers onto task structure. */ -void -save_fpu(struct task_struct *tsk) +void save_fpu(struct task_struct *tsk) { unsigned long dummy; @@ -52,7 +51,7 @@ save_fpu(struct task_struct *tsk) "fmov.s fr0, @-%0\n\t" "lds %3, fpscr\n\t" : "=r" (dummy) - : "0" ((char *)(&tsk->thread.fpu.hard.status)), + : "0" ((char *)(&tsk->thread.xstate->hardfpu.status)), "r" (FPSCR_RCHG), "r" (FPSCR_INIT) : "memory"); @@ -60,8 +59,7 @@ save_fpu(struct task_struct *tsk) disable_fpu(); } -static void -restore_fpu(struct task_struct *tsk) +void restore_fpu(struct task_struct *tsk) { unsigned long dummy; @@ -85,45 +83,12 @@ restore_fpu(struct task_struct *tsk) "lds.l @%0+, fpscr\n\t" "lds.l @%0+, fpul\n\t" : "=r" (dummy) - : "0" (&tsk->thread.fpu), "r" (FPSCR_RCHG) + : "0" (tsk->thread.xstate), "r" (FPSCR_RCHG) : "memory"); disable_fpu(); } /* - * Load the FPU with signalling NANS. This bit pattern we're using - * has the property that no matter wether considered as single or as - * double precission represents signaling NANS. - */ - -static void -fpu_init(void) -{ - enable_fpu(); - asm volatile("lds %0, fpul\n\t" - "fsts fpul, fr0\n\t" - "fsts fpul, fr1\n\t" - "fsts fpul, fr2\n\t" - "fsts fpul, fr3\n\t" - "fsts fpul, fr4\n\t" - "fsts fpul, fr5\n\t" - "fsts fpul, fr6\n\t" - "fsts fpul, fr7\n\t" - "fsts fpul, fr8\n\t" - "fsts fpul, fr9\n\t" - "fsts fpul, fr10\n\t" - "fsts fpul, fr11\n\t" - "fsts fpul, fr12\n\t" - "fsts fpul, fr13\n\t" - "fsts fpul, fr14\n\t" - "fsts fpul, fr15\n\t" - "lds %2, fpscr\n\t" - : /* no output */ - : "r" (0), "r" (FPSCR_RCHG), "r" (FPSCR_INIT)); - disable_fpu(); -} - -/* * Emulate arithmetic ops on denormalized number for some FPU insns. */ @@ -490,9 +455,9 @@ ieee_fpe_handler (struct pt_regs *regs) if ((finsn & 0xf1ff) == 0xf0ad) { /* fcnvsd */ struct task_struct *tsk = current; - if ((tsk->thread.fpu.hard.fpscr & FPSCR_FPU_ERROR)) { + if ((tsk->thread.xstate->hardfpu.fpscr & FPSCR_FPU_ERROR)) { /* FPU error */ - denormal_to_double (&tsk->thread.fpu.hard, + denormal_to_double (&tsk->thread.xstate->hardfpu, (finsn >> 8) & 0xf); } else return 0; @@ -507,9 +472,9 @@ ieee_fpe_handler (struct pt_regs *regs) n = (finsn >> 8) & 0xf; m = (finsn >> 4) & 0xf; - hx = tsk->thread.fpu.hard.fp_regs[n]; - hy = tsk->thread.fpu.hard.fp_regs[m]; - fpscr = tsk->thread.fpu.hard.fpscr; + hx = tsk->thread.xstate->hardfpu.fp_regs[n]; + hy = tsk->thread.xstate->hardfpu.fp_regs[m]; + fpscr = tsk->thread.xstate->hardfpu.fpscr; prec = fpscr & (1 << 19); if ((fpscr & FPSCR_FPU_ERROR) @@ -519,15 +484,15 @@ ieee_fpe_handler (struct pt_regs *regs) /* FPU error because of denormal */ llx = ((long long) hx << 32) - | tsk->thread.fpu.hard.fp_regs[n+1]; + | tsk->thread.xstate->hardfpu.fp_regs[n+1]; lly = ((long long) hy << 32) - | tsk->thread.fpu.hard.fp_regs[m+1]; + | tsk->thread.xstate->hardfpu.fp_regs[m+1]; if ((hx & 0x7fffffff) >= 0x00100000) llx = denormal_muld(lly, llx); else llx = denormal_muld(llx, lly); - tsk->thread.fpu.hard.fp_regs[n] = llx >> 32; - tsk->thread.fpu.hard.fp_regs[n+1] = llx & 0xffffffff; + tsk->thread.xstate->hardfpu.fp_regs[n] = llx >> 32; + tsk->thread.xstate->hardfpu.fp_regs[n+1] = llx & 0xffffffff; } else if ((fpscr & FPSCR_FPU_ERROR) && (!prec && ((hx & 0x7fffffff) < 0x00800000 || (hy & 0x7fffffff) < 0x00800000))) { @@ -536,7 +501,7 @@ ieee_fpe_handler (struct pt_regs *regs) hx = denormal_mulf(hy, hx); else hx = denormal_mulf(hx, hy); - tsk->thread.fpu.hard.fp_regs[n] = hx; + tsk->thread.xstate->hardfpu.fp_regs[n] = hx; } else return 0; @@ -550,9 +515,9 @@ ieee_fpe_handler (struct pt_regs *regs) n = (finsn >> 8) & 0xf; m = (finsn >> 4) & 0xf; - hx = tsk->thread.fpu.hard.fp_regs[n]; - hy = tsk->thread.fpu.hard.fp_regs[m]; - fpscr = tsk->thread.fpu.hard.fpscr; + hx = tsk->thread.xstate->hardfpu.fp_regs[n]; + hy = tsk->thread.xstate->hardfpu.fp_regs[m]; + fpscr = tsk->thread.xstate->hardfpu.fpscr; prec = fpscr & (1 << 19); if ((fpscr & FPSCR_FPU_ERROR) @@ -562,15 +527,15 @@ ieee_fpe_handler (struct pt_regs *regs) /* FPU error because of denormal */ llx = ((long long) hx << 32) - | tsk->thread.fpu.hard.fp_regs[n+1]; + | tsk->thread.xstate->hardfpu.fp_regs[n+1]; lly = ((long long) hy << 32) - | tsk->thread.fpu.hard.fp_regs[m+1]; + | tsk->thread.xstate->hardfpu.fp_regs[m+1]; if ((finsn & 0xf00f) == 0xf000) llx = denormal_addd(llx, lly); else llx = denormal_addd(llx, lly ^ (1LL << 63)); - tsk->thread.fpu.hard.fp_regs[n] = llx >> 32; - tsk->thread.fpu.hard.fp_regs[n+1] = llx & 0xffffffff; + tsk->thread.xstate->hardfpu.fp_regs[n] = llx >> 32; + tsk->thread.xstate->hardfpu.fp_regs[n+1] = llx & 0xffffffff; } else if ((fpscr & FPSCR_FPU_ERROR) && (!prec && ((hx & 0x7fffffff) < 0x00800000 || (hy & 0x7fffffff) < 0x00800000))) { @@ -579,7 +544,7 @@ ieee_fpe_handler (struct pt_regs *regs) hx = denormal_addf(hx, hy); else hx = denormal_addf(hx, hy ^ 0x80000000); - tsk->thread.fpu.hard.fp_regs[n] = hx; + tsk->thread.xstate->hardfpu.fp_regs[n] = hx; } else return 0; @@ -597,7 +562,7 @@ BUILD_TRAP_HANDLER(fpu_error) __unlazy_fpu(tsk, regs); if (ieee_fpe_handler(regs)) { - tsk->thread.fpu.hard.fpscr &= + tsk->thread.xstate->hardfpu.fpscr &= ~(FPSCR_CAUSE_MASK | FPSCR_FLAG_MASK); grab_fpu(regs); restore_fpu(tsk); @@ -607,33 +572,3 @@ BUILD_TRAP_HANDLER(fpu_error) force_sig(SIGFPE, tsk); } - -void fpu_state_restore(struct pt_regs *regs) -{ - struct task_struct *tsk = current; - - grab_fpu(regs); - if (unlikely(!user_mode(regs))) { - printk(KERN_ERR "BUG: FPU is used in kernel mode.\n"); - BUG(); - return; - } - - if (likely(used_math())) { - /* Using the FPU again. */ - restore_fpu(tsk); - } else { - /* First time FPU user. */ - fpu_init(); - set_used_math(); - } - task_thread_info(tsk)->status |= TS_USEDFPU; - tsk->fpu_counter++; -} - -BUILD_TRAP_HANDLER(fpu_state_restore) -{ - TRAP_HANDLER_DECL; - - fpu_state_restore(regs); -} diff --git a/arch/sh/kernel/cpu/sh3/ex.S b/arch/sh/kernel/cpu/sh3/ex.S index 46610c35c23..99b4d020179 100644 --- a/arch/sh/kernel/cpu/sh3/ex.S +++ b/arch/sh/kernel/cpu/sh3/ex.S @@ -49,7 +49,7 @@ ENTRY(exception_handling_table) .long exception_error ! reserved_instruction (filled by trap_init) /* 180 */ .long exception_error ! illegal_slot_instruction (filled by trap_init) /*1A0*/ .long nmi_trap_handler /* 1C0 */ ! Allow trap to debugger - .long break_point_trap /* 1E0 */ + .long breakpoint_trap_handler /* 1E0 */ /* * Pad the remainder of the table out, exceptions residing in far diff --git a/arch/sh/kernel/cpu/sh4/fpu.c b/arch/sh/kernel/cpu/sh4/fpu.c index e97857aec8a..447482d7f65 100644 --- a/arch/sh/kernel/cpu/sh4/fpu.c +++ b/arch/sh/kernel/cpu/sh4/fpu.c @@ -85,14 +85,14 @@ void save_fpu(struct task_struct *tsk) "fmov.s fr1, @-%0\n\t" "fmov.s fr0, @-%0\n\t" "lds %3, fpscr\n\t":"=r" (dummy) - :"0"((char *)(&tsk->thread.fpu.hard.status)), + :"0"((char *)(&tsk->thread.xstate->hardfpu.status)), "r"(FPSCR_RCHG), "r"(FPSCR_INIT) :"memory"); disable_fpu(); } -static void restore_fpu(struct task_struct *tsk) +void restore_fpu(struct task_struct *tsk) { unsigned long dummy; @@ -135,62 +135,11 @@ static void restore_fpu(struct task_struct *tsk) "lds.l @%0+, fpscr\n\t" "lds.l @%0+, fpul\n\t" :"=r" (dummy) - :"0"(&tsk->thread.fpu), "r"(FPSCR_RCHG) + :"0" (tsk->thread.xstate), "r" (FPSCR_RCHG) :"memory"); disable_fpu(); } -/* - * Load the FPU with signalling NANS. This bit pattern we're using - * has the property that no matter wether considered as single or as - * double precision represents signaling NANS. - */ - -static void fpu_init(void) -{ - enable_fpu(); - asm volatile ( "lds %0, fpul\n\t" - "lds %1, fpscr\n\t" - "fsts fpul, fr0\n\t" - "fsts fpul, fr1\n\t" - "fsts fpul, fr2\n\t" - "fsts fpul, fr3\n\t" - "fsts fpul, fr4\n\t" - "fsts fpul, fr5\n\t" - "fsts fpul, fr6\n\t" - "fsts fpul, fr7\n\t" - "fsts fpul, fr8\n\t" - "fsts fpul, fr9\n\t" - "fsts fpul, fr10\n\t" - "fsts fpul, fr11\n\t" - "fsts fpul, fr12\n\t" - "fsts fpul, fr13\n\t" - "fsts fpul, fr14\n\t" - "fsts fpul, fr15\n\t" - "frchg\n\t" - "fsts fpul, fr0\n\t" - "fsts fpul, fr1\n\t" - "fsts fpul, fr2\n\t" - "fsts fpul, fr3\n\t" - "fsts fpul, fr4\n\t" - "fsts fpul, fr5\n\t" - "fsts fpul, fr6\n\t" - "fsts fpul, fr7\n\t" - "fsts fpul, fr8\n\t" - "fsts fpul, fr9\n\t" - "fsts fpul, fr10\n\t" - "fsts fpul, fr11\n\t" - "fsts fpul, fr12\n\t" - "fsts fpul, fr13\n\t" - "fsts fpul, fr14\n\t" - "fsts fpul, fr15\n\t" - "frchg\n\t" - "lds %2, fpscr\n\t" - : /* no output */ - :"r" (0), "r"(FPSCR_RCHG), "r"(FPSCR_INIT)); - disable_fpu(); -} - /** * denormal_to_double - Given denormalized float number, * store double float @@ -282,9 +231,9 @@ static int ieee_fpe_handler(struct pt_regs *regs) /* fcnvsd */ struct task_struct *tsk = current; - if ((tsk->thread.fpu.hard.fpscr & FPSCR_CAUSE_ERROR)) + if ((tsk->thread.xstate->hardfpu.fpscr & FPSCR_CAUSE_ERROR)) /* FPU error */ - denormal_to_double(&tsk->thread.fpu.hard, + denormal_to_double(&tsk->thread.xstate->hardfpu, (finsn >> 8) & 0xf); else return 0; @@ -300,9 +249,9 @@ static int ieee_fpe_handler(struct pt_regs *regs) n = (finsn >> 8) & 0xf; m = (finsn >> 4) & 0xf; - hx = tsk->thread.fpu.hard.fp_regs[n]; - hy = tsk->thread.fpu.hard.fp_regs[m]; - fpscr = tsk->thread.fpu.hard.fpscr; + hx = tsk->thread.xstate->hardfpu.fp_regs[n]; + hy = tsk->thread.xstate->hardfpu.fp_regs[m]; + fpscr = tsk->thread.xstate->hardfpu.fpscr; prec = fpscr & FPSCR_DBL_PRECISION; if ((fpscr & FPSCR_CAUSE_ERROR) @@ -312,18 +261,18 @@ static int ieee_fpe_handler(struct pt_regs *regs) /* FPU error because of denormal (doubles) */ llx = ((long long)hx << 32) - | tsk->thread.fpu.hard.fp_regs[n + 1]; + | tsk->thread.xstate->hardfpu.fp_regs[n + 1]; lly = ((long long)hy << 32) - | tsk->thread.fpu.hard.fp_regs[m + 1]; + | tsk->thread.xstate->hardfpu.fp_regs[m + 1]; llx = float64_mul(llx, lly); - tsk->thread.fpu.hard.fp_regs[n] = llx >> 32; - tsk->thread.fpu.hard.fp_regs[n + 1] = llx & 0xffffffff; + tsk->thread.xstate->hardfpu.fp_regs[n] = llx >> 32; + tsk->thread.xstate->hardfpu.fp_regs[n + 1] = llx & 0xffffffff; } else if ((fpscr & FPSCR_CAUSE_ERROR) && (!prec && ((hx & 0x7fffffff) < 0x00800000 || (hy & 0x7fffffff) < 0x00800000))) { /* FPU error because of denormal (floats) */ hx = float32_mul(hx, hy); - tsk->thread.fpu.hard.fp_regs[n] = hx; + tsk->thread.xstate->hardfpu.fp_regs[n] = hx; } else return 0; @@ -338,9 +287,9 @@ static int ieee_fpe_handler(struct pt_regs *regs) n = (finsn >> 8) & 0xf; m = (finsn >> 4) & 0xf; - hx = tsk->thread.fpu.hard.fp_regs[n]; - hy = tsk->thread.fpu.hard.fp_regs[m]; - fpscr = tsk->thread.fpu.hard.fpscr; + hx = tsk->thread.xstate->hardfpu.fp_regs[n]; + hy = tsk->thread.xstate->hardfpu.fp_regs[m]; + fpscr = tsk->thread.xstate->hardfpu.fpscr; prec = fpscr & FPSCR_DBL_PRECISION; if ((fpscr & FPSCR_CAUSE_ERROR) @@ -350,15 +299,15 @@ static int ieee_fpe_handler(struct pt_regs *regs) /* FPU error because of denormal (doubles) */ llx = ((long long)hx << 32) - | tsk->thread.fpu.hard.fp_regs[n + 1]; + | tsk->thread.xstate->hardfpu.fp_regs[n + 1]; lly = ((long long)hy << 32) - | tsk->thread.fpu.hard.fp_regs[m + 1]; + | tsk->thread.xstate->hardfpu.fp_regs[m + 1]; if ((finsn & 0xf00f) == 0xf000) llx = float64_add(llx, lly); else llx = float64_sub(llx, lly); - tsk->thread.fpu.hard.fp_regs[n] = llx >> 32; - tsk->thread.fpu.hard.fp_regs[n + 1] = llx & 0xffffffff; + tsk->thread.xstate->hardfpu.fp_regs[n] = llx >> 32; + tsk->thread.xstate->hardfpu.fp_regs[n + 1] = llx & 0xffffffff; } else if ((fpscr & FPSCR_CAUSE_ERROR) && (!prec && ((hx & 0x7fffffff) < 0x00800000 || (hy & 0x7fffffff) < 0x00800000))) { @@ -367,7 +316,7 @@ static int ieee_fpe_handler(struct pt_regs *regs) hx = float32_add(hx, hy); else hx = float32_sub(hx, hy); - tsk->thread.fpu.hard.fp_regs[n] = hx; + tsk->thread.xstate->hardfpu.fp_regs[n] = hx; } else return 0; @@ -382,9 +331,9 @@ static int ieee_fpe_handler(struct pt_regs *regs) n = (finsn >> 8) & 0xf; m = (finsn >> 4) & 0xf; - hx = tsk->thread.fpu.hard.fp_regs[n]; - hy = tsk->thread.fpu.hard.fp_regs[m]; - fpscr = tsk->thread.fpu.hard.fpscr; + hx = tsk->thread.xstate->hardfpu.fp_regs[n]; + hy = tsk->thread.xstate->hardfpu.fp_regs[m]; + fpscr = tsk->thread.xstate->hardfpu.fpscr; prec = fpscr & FPSCR_DBL_PRECISION; if ((fpscr & FPSCR_CAUSE_ERROR) @@ -394,20 +343,20 @@ static int ieee_fpe_handler(struct pt_regs *regs) /* FPU error because of denormal (doubles) */ llx = ((long long)hx << 32) - | tsk->thread.fpu.hard.fp_regs[n + 1]; + | tsk->thread.xstate->hardfpu.fp_regs[n + 1]; lly = ((long long)hy << 32) - | tsk->thread.fpu.hard.fp_regs[m + 1]; + | tsk->thread.xstate->hardfpu.fp_regs[m + 1]; llx = float64_div(llx, lly); - tsk->thread.fpu.hard.fp_regs[n] = llx >> 32; - tsk->thread.fpu.hard.fp_regs[n + 1] = llx & 0xffffffff; + tsk->thread.xstate->hardfpu.fp_regs[n] = llx >> 32; + tsk->thread.xstate->hardfpu.fp_regs[n + 1] = llx & 0xffffffff; } else if ((fpscr & FPSCR_CAUSE_ERROR) && (!prec && ((hx & 0x7fffffff) < 0x00800000 || (hy & 0x7fffffff) < 0x00800000))) { /* FPU error because of denormal (floats) */ hx = float32_div(hx, hy); - tsk->thread.fpu.hard.fp_regs[n] = hx; + tsk->thread.xstate->hardfpu.fp_regs[n] = hx; } else return 0; @@ -420,17 +369,17 @@ static int ieee_fpe_handler(struct pt_regs *regs) unsigned int hx; m = (finsn >> 8) & 0x7; - hx = tsk->thread.fpu.hard.fp_regs[m]; + hx = tsk->thread.xstate->hardfpu.fp_regs[m]; - if ((tsk->thread.fpu.hard.fpscr & FPSCR_CAUSE_ERROR) + if ((tsk->thread.xstate->hardfpu.fpscr & FPSCR_CAUSE_ERROR) && ((hx & 0x7fffffff) < 0x00100000)) { /* subnormal double to float conversion */ long long llx; - llx = ((long long)tsk->thread.fpu.hard.fp_regs[m] << 32) - | tsk->thread.fpu.hard.fp_regs[m + 1]; + llx = ((long long)tsk->thread.xstate->hardfpu.fp_regs[m] << 32) + | tsk->thread.xstate->hardfpu.fp_regs[m + 1]; - tsk->thread.fpu.hard.fpul = float64_to_float32(llx); + tsk->thread.xstate->hardfpu.fpul = float64_to_float32(llx); } else return 0; @@ -449,7 +398,7 @@ void float_raise(unsigned int flags) int float_rounding_mode(void) { struct task_struct *tsk = current; - int roundingMode = FPSCR_ROUNDING_MODE(tsk->thread.fpu.hard.fpscr); + int roundingMode = FPSCR_ROUNDING_MODE(tsk->thread.xstate->hardfpu.fpscr); return roundingMode; } @@ -461,16 +410,16 @@ BUILD_TRAP_HANDLER(fpu_error) __unlazy_fpu(tsk, regs); fpu_exception_flags = 0; if (ieee_fpe_handler(regs)) { - tsk->thread.fpu.hard.fpscr &= + tsk->thread.xstate->hardfpu.fpscr &= ~(FPSCR_CAUSE_MASK | FPSCR_FLAG_MASK); - tsk->thread.fpu.hard.fpscr |= fpu_exception_flags; + tsk->thread.xstate->hardfpu.fpscr |= fpu_exception_flags; /* Set the FPSCR flag as well as cause bits - simply * replicate the cause */ - tsk->thread.fpu.hard.fpscr |= (fpu_exception_flags >> 10); + tsk->thread.xstate->hardfpu.fpscr |= (fpu_exception_flags >> 10); grab_fpu(regs); restore_fpu(tsk); task_thread_info(tsk)->status |= TS_USEDFPU; - if ((((tsk->thread.fpu.hard.fpscr & FPSCR_ENABLE_MASK) >> 7) & + if ((((tsk->thread.xstate->hardfpu.fpscr & FPSCR_ENABLE_MASK) >> 7) & (fpu_exception_flags >> 2)) == 0) { return; } @@ -478,33 +427,3 @@ BUILD_TRAP_HANDLER(fpu_error) force_sig(SIGFPE, tsk); } - -void fpu_state_restore(struct pt_regs *regs) -{ - struct task_struct *tsk = current; - - grab_fpu(regs); - if (unlikely(!user_mode(regs))) { - printk(KERN_ERR "BUG: FPU is used in kernel mode.\n"); - BUG(); - return; - } - - if (likely(used_math())) { - /* Using the FPU again. */ - restore_fpu(tsk); - } else { - /* First time FPU user. */ - fpu_init(); - set_used_math(); - } - task_thread_info(tsk)->status |= TS_USEDFPU; - tsk->fpu_counter++; -} - -BUILD_TRAP_HANDLER(fpu_state_restore) -{ - TRAP_HANDLER_DECL; - - fpu_state_restore(regs); -} diff --git a/arch/sh/kernel/cpu/sh4a/Makefile b/arch/sh/kernel/cpu/sh4a/Makefile index 33bab477d2e..b144e8af89d 100644 --- a/arch/sh/kernel/cpu/sh4a/Makefile +++ b/arch/sh/kernel/cpu/sh4a/Makefile @@ -41,7 +41,8 @@ pinmux-$(CONFIG_CPU_SUBTYPE_SH7757) := pinmux-sh7757.o pinmux-$(CONFIG_CPU_SUBTYPE_SH7785) := pinmux-sh7785.o pinmux-$(CONFIG_CPU_SUBTYPE_SH7786) := pinmux-sh7786.o -obj-y += $(clock-y) -obj-$(CONFIG_SMP) += $(smp-y) -obj-$(CONFIG_GENERIC_GPIO) += $(pinmux-y) -obj-$(CONFIG_PERF_EVENTS) += perf_event.o +obj-y += $(clock-y) +obj-$(CONFIG_SMP) += $(smp-y) +obj-$(CONFIG_GENERIC_GPIO) += $(pinmux-y) +obj-$(CONFIG_PERF_EVENTS) += perf_event.o +obj-$(CONFIG_HAVE_HW_BREAKPOINT) += ubc.o diff --git a/arch/sh/kernel/cpu/sh4a/ubc.c b/arch/sh/kernel/cpu/sh4a/ubc.c new file mode 100644 index 00000000000..efb2745bcb3 --- /dev/null +++ b/arch/sh/kernel/cpu/sh4a/ubc.c @@ -0,0 +1,133 @@ +/* + * arch/sh/kernel/cpu/sh4a/ubc.c + * + * On-chip UBC support for SH-4A CPUs. + * + * Copyright (C) 2009 - 2010 Paul Mundt + * + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file "COPYING" in the main directory of this archive + * for more details. + */ +#include <linux/init.h> +#include <linux/err.h> +#include <linux/clk.h> +#include <linux/io.h> +#include <asm/hw_breakpoint.h> + +#define UBC_CBR(idx) (0xff200000 + (0x20 * idx)) +#define UBC_CRR(idx) (0xff200004 + (0x20 * idx)) +#define UBC_CAR(idx) (0xff200008 + (0x20 * idx)) +#define UBC_CAMR(idx) (0xff20000c + (0x20 * idx)) + +#define UBC_CCMFR 0xff200600 +#define UBC_CBCR 0xff200620 + +/* CRR */ +#define UBC_CRR_PCB (1 << 1) +#define UBC_CRR_BIE (1 << 0) + +/* CBR */ +#define UBC_CBR_CE (1 << 0) + +static struct sh_ubc sh4a_ubc; + +static void sh4a_ubc_enable(struct arch_hw_breakpoint *info, int idx) +{ + __raw_writel(UBC_CBR_CE | info->len | info->type, UBC_CBR(idx)); + __raw_writel(info->address, UBC_CAR(idx)); +} + +static void sh4a_ubc_disable(struct arch_hw_breakpoint *info, int idx) +{ + __raw_writel(0, UBC_CBR(idx)); + __raw_writel(0, UBC_CAR(idx)); +} + +static void sh4a_ubc_enable_all(unsigned long mask) +{ + int i; + + for (i = 0; i < sh4a_ubc.num_events; i++) + if (mask & (1 << i)) + __raw_writel(__raw_readl(UBC_CBR(i)) | UBC_CBR_CE, + UBC_CBR(i)); +} + +static void sh4a_ubc_disable_all(void) +{ + int i; + + for (i = 0; i < sh4a_ubc.num_events; i++) + __raw_writel(__raw_readl(UBC_CBR(i)) & ~UBC_CBR_CE, + UBC_CBR(i)); +} + +static unsigned long sh4a_ubc_active_mask(void) +{ + unsigned long active = 0; + int i; + + for (i = 0; i < sh4a_ubc.num_events; i++) + if (__raw_readl(UBC_CBR(i)) & UBC_CBR_CE) + active |= (1 << i); + + return active; +} + +static unsigned long sh4a_ubc_triggered_mask(void) +{ + return __raw_readl(UBC_CCMFR); +} + +static void sh4a_ubc_clear_triggered_mask(unsigned long mask) +{ + __raw_writel(__raw_readl(UBC_CCMFR) & ~mask, UBC_CCMFR); +} + +static struct sh_ubc sh4a_ubc = { + .name = "SH-4A", + .num_events = 2, + .trap_nr = 0x1e0, + .enable = sh4a_ubc_enable, + .disable = sh4a_ubc_disable, + .enable_all = sh4a_ubc_enable_all, + .disable_all = sh4a_ubc_disable_all, + .active_mask = sh4a_ubc_active_mask, + .triggered_mask = sh4a_ubc_triggered_mask, + .clear_triggered_mask = sh4a_ubc_clear_triggered_mask, +}; + +static int __init sh4a_ubc_init(void) +{ + struct clk *ubc_iclk = clk_get(NULL, "ubc0"); + int i; + + /* + * The UBC MSTP bit is optional, as not all platforms will have + * it. Just ignore it if we can't find it. + */ + if (IS_ERR(ubc_iclk)) + ubc_iclk = NULL; + + clk_enable(ubc_iclk); + + __raw_writel(0, UBC_CBCR); + + for (i = 0; i < sh4a_ubc.num_events; i++) { + __raw_writel(0, UBC_CAMR(i)); + __raw_writel(0, UBC_CBR(i)); + + __raw_writel(UBC_CRR_BIE | UBC_CRR_PCB, UBC_CRR(i)); + + /* dummy read for write posting */ + (void)__raw_readl(UBC_CRR(i)); + } + + clk_disable(ubc_iclk); + + sh4a_ubc.clk = ubc_iclk; + + return register_sh_ubc(&sh4a_ubc); +} +arch_initcall(sh4a_ubc_init); diff --git a/arch/sh/kernel/debugtraps.S b/arch/sh/kernel/debugtraps.S index 591741383ee..7a1b46fec0f 100644 --- a/arch/sh/kernel/debugtraps.S +++ b/arch/sh/kernel/debugtraps.S @@ -13,7 +13,6 @@ #include <linux/linkage.h> #if !defined(CONFIG_KGDB) -#define breakpoint_trap_handler debug_trap_handler #define singlestep_trap_handler debug_trap_handler #endif diff --git a/arch/sh/kernel/hw_breakpoint.c b/arch/sh/kernel/hw_breakpoint.c new file mode 100644 index 00000000000..e2f1753d275 --- /dev/null +++ b/arch/sh/kernel/hw_breakpoint.c @@ -0,0 +1,463 @@ +/* + * arch/sh/kernel/hw_breakpoint.c + * + * Unified kernel/user-space hardware breakpoint facility for the on-chip UBC. + * + * Copyright (C) 2009 - 2010 Paul Mundt + * + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file "COPYING" in the main directory of this archive + * for more details. + */ +#include <linux/init.h> +#include <linux/perf_event.h> +#include <linux/hw_breakpoint.h> +#include <linux/percpu.h> +#include <linux/kallsyms.h> +#include <linux/notifier.h> +#include <linux/kprobes.h> +#include <linux/kdebug.h> +#include <linux/io.h> +#include <linux/clk.h> +#include <asm/hw_breakpoint.h> +#include <asm/mmu_context.h> +#include <asm/ptrace.h> + +/* + * Stores the breakpoints currently in use on each breakpoint address + * register for each cpus + */ +static DEFINE_PER_CPU(struct perf_event *, bp_per_reg[HBP_NUM]); + +/* + * A dummy placeholder for early accesses until the CPUs get a chance to + * register their UBCs later in the boot process. + */ +static struct sh_ubc ubc_dummy = { .num_events = 0 }; + +static struct sh_ubc *sh_ubc __read_mostly = &ubc_dummy; + +/* + * Install a perf counter breakpoint. + * + * We seek a free UBC channel and use it for this breakpoint. + * + * Atomic: we hold the counter->ctx->lock and we only handle variables + * and registers local to this cpu. + */ +int arch_install_hw_breakpoint(struct perf_event *bp) +{ + struct arch_hw_breakpoint *info = counter_arch_bp(bp); + int i; + + for (i = 0; i < sh_ubc->num_events; i++) { + struct perf_event **slot = &__get_cpu_var(bp_per_reg[i]); + + if (!*slot) { + *slot = bp; + break; + } + } + + if (WARN_ONCE(i == sh_ubc->num_events, "Can't find any breakpoint slot")) + return -EBUSY; + + clk_enable(sh_ubc->clk); + sh_ubc->enable(info, i); + + return 0; +} + +/* + * Uninstall the breakpoint contained in the given counter. + * + * First we search the debug address register it uses and then we disable + * it. + * + * Atomic: we hold the counter->ctx->lock and we only handle variables + * and registers local to this cpu. + */ +void arch_uninstall_hw_breakpoint(struct perf_event *bp) +{ + struct arch_hw_breakpoint *info = counter_arch_bp(bp); + int i; + + for (i = 0; i < sh_ubc->num_events; i++) { + struct perf_event **slot = &__get_cpu_var(bp_per_reg[i]); + + if (*slot == bp) { + *slot = NULL; + break; + } + } + + if (WARN_ONCE(i == sh_ubc->num_events, "Can't find any breakpoint slot")) + return; + + sh_ubc->disable(info, i); + clk_disable(sh_ubc->clk); +} + +static int get_hbp_len(u16 hbp_len) +{ + unsigned int len_in_bytes = 0; + + switch (hbp_len) { + case SH_BREAKPOINT_LEN_1: + len_in_bytes = 1; + break; + case SH_BREAKPOINT_LEN_2: + len_in_bytes = 2; + break; + case SH_BREAKPOINT_LEN_4: + len_in_bytes = 4; + break; + case SH_BREAKPOINT_LEN_8: + len_in_bytes = 8; + break; + } + return len_in_bytes; +} + +/* + * Check for virtual address in user space. + */ +int arch_check_va_in_userspace(unsigned long va, u16 hbp_len) +{ + unsigned int len; + + len = get_hbp_len(hbp_len); + + return (va <= TASK_SIZE - len); +} + +/* + * Check for virtual address in kernel space. + */ +static int arch_check_va_in_kernelspace(unsigned long va, u8 hbp_len) +{ + unsigned int len; + + len = get_hbp_len(hbp_len); + + return (va >= TASK_SIZE) && ((va + len - 1) >= TASK_SIZE); +} + +/* + * Store a breakpoint's encoded address, length, and type. + */ +static int arch_store_info(struct perf_event *bp) +{ + struct arch_hw_breakpoint *info = counter_arch_bp(bp); + + /* + * User-space requests will always have the address field populated + * For kernel-addresses, either the address or symbol name can be + * specified. + */ + if (info->name) + info->address = (unsigned long)kallsyms_lookup_name(info->name); + if (info->address) + return 0; + + return -EINVAL; +} + +int arch_bp_generic_fields(int sh_len, int sh_type, + int *gen_len, int *gen_type) +{ + /* Len */ + switch (sh_len) { + case SH_BREAKPOINT_LEN_1: + *gen_len = HW_BREAKPOINT_LEN_1; + break; + case SH_BREAKPOINT_LEN_2: + *gen_len = HW_BREAKPOINT_LEN_2; + break; + case SH_BREAKPOINT_LEN_4: + *gen_len = HW_BREAKPOINT_LEN_4; + break; + case SH_BREAKPOINT_LEN_8: + *gen_len = HW_BREAKPOINT_LEN_8; + break; + default: + return -EINVAL; + } + + /* Type */ + switch (sh_type) { + case SH_BREAKPOINT_READ: + *gen_type = HW_BREAKPOINT_R; + case SH_BREAKPOINT_WRITE: + *gen_type = HW_BREAKPOINT_W; + break; + case SH_BREAKPOINT_RW: + *gen_type = HW_BREAKPOINT_W | HW_BREAKPOINT_R; + break; + default: + return -EINVAL; + } + + return 0; +} + +static int arch_build_bp_info(struct perf_event *bp) +{ + struct arch_hw_breakpoint *info = counter_arch_bp(bp); + + info->address = bp->attr.bp_addr; + + /* Len */ + switch (bp->attr.bp_len) { + case HW_BREAKPOINT_LEN_1: + info->len = SH_BREAKPOINT_LEN_1; + break; + case HW_BREAKPOINT_LEN_2: + info->len = SH_BREAKPOINT_LEN_2; + break; + case HW_BREAKPOINT_LEN_4: + info->len = SH_BREAKPOINT_LEN_4; + break; + case HW_BREAKPOINT_LEN_8: + info->len = SH_BREAKPOINT_LEN_8; + break; + default: + return -EINVAL; + } + + /* Type */ + switch (bp->attr.bp_type) { + case HW_BREAKPOINT_R: + info->type = SH_BREAKPOINT_READ; + break; + case HW_BREAKPOINT_W: + info->type = SH_BREAKPOINT_WRITE; + break; + case HW_BREAKPOINT_W | HW_BREAKPOINT_R: + info->type = SH_BREAKPOINT_RW; + break; + default: + return -EINVAL; + } + + return 0; +} + +/* + * Validate the arch-specific HW Breakpoint register settings + */ +int arch_validate_hwbkpt_settings(struct perf_event *bp, + struct task_struct *tsk) +{ + struct arch_hw_breakpoint *info = counter_arch_bp(bp); + unsigned int align; + int ret; + + ret = arch_build_bp_info(bp); + if (ret) + return ret; + + ret = -EINVAL; + + switch (info->len) { + case SH_BREAKPOINT_LEN_1: + align = 0; + break; + case SH_BREAKPOINT_LEN_2: + align = 1; + break; + case SH_BREAKPOINT_LEN_4: + align = 3; + break; + case SH_BREAKPOINT_LEN_8: + align = 7; + break; + default: + return ret; + } + + ret = arch_store_info(bp); + + if (ret < 0) + return ret; + + /* + * Check that the low-order bits of the address are appropriate + * for the alignment implied by len. + */ + if (info->address & align) + return -EINVAL; + + /* Check that the virtual address is in the proper range */ + if (tsk) { + if (!arch_check_va_in_userspace(info->address, info->len)) + return -EFAULT; + } else { + if (!arch_check_va_in_kernelspace(info->address, info->len)) + return -EFAULT; + } + + return 0; +} + +/* + * Release the user breakpoints used by ptrace + */ +void flush_ptrace_hw_breakpoint(struct task_struct *tsk) +{ + int i; + struct thread_struct *t = &tsk->thread; + + for (i = 0; i < sh_ubc->num_events; i++) { + unregister_hw_breakpoint(t->ptrace_bps[i]); + t->ptrace_bps[i] = NULL; + } +} + +static int __kprobes hw_breakpoint_handler(struct die_args *args) +{ + int cpu, i, rc = NOTIFY_STOP; + struct perf_event *bp; + unsigned int cmf, resume_mask; + + /* + * Do an early return if none of the channels triggered. + */ + cmf = sh_ubc->triggered_mask(); + if (unlikely(!cmf)) + return NOTIFY_DONE; + + /* + * By default, resume all of the active channels. + */ + resume_mask = sh_ubc->active_mask(); + + /* + * Disable breakpoints during exception handling. + */ + sh_ubc->disable_all(); + + cpu = get_cpu(); + for (i = 0; i < sh_ubc->num_events; i++) { + unsigned long event_mask = (1 << i); + + if (likely(!(cmf & event_mask))) + continue; + + /* + * The counter may be concurrently released but that can only + * occur from a call_rcu() path. We can then safely fetch + * the breakpoint, use its callback, touch its counter + * while we are in an rcu_read_lock() path. + */ + rcu_read_lock(); + + bp = per_cpu(bp_per_reg[i], cpu); + if (bp) + rc = NOTIFY_DONE; + + /* + * Reset the condition match flag to denote completion of + * exception handling. + */ + sh_ubc->clear_triggered_mask(event_mask); + + /* + * bp can be NULL due to concurrent perf counter + * removing. + */ + if (!bp) { + rcu_read_unlock(); + break; + } + + /* + * Don't restore the channel if the breakpoint is from + * ptrace, as it always operates in one-shot mode. + */ + if (bp->overflow_handler == ptrace_triggered) + resume_mask &= ~(1 << i); + + perf_bp_event(bp, args->regs); + + /* Deliver the signal to userspace */ + if (arch_check_va_in_userspace(bp->attr.bp_addr, + bp->attr.bp_len)) { + siginfo_t info; + + info.si_signo = args->signr; + info.si_errno = notifier_to_errno(rc); + info.si_code = TRAP_HWBKPT; + + force_sig_info(args->signr, &info, current); + } + + rcu_read_unlock(); + } + + if (cmf == 0) + rc = NOTIFY_DONE; + + sh_ubc->enable_all(resume_mask); + + put_cpu(); + + return rc; +} + +BUILD_TRAP_HANDLER(breakpoint) +{ + unsigned long ex = lookup_exception_vector(); + TRAP_HANDLER_DECL; + + notify_die(DIE_BREAKPOINT, "breakpoint", regs, 0, ex, SIGTRAP); +} + +/* + * Handle debug exception notifications. + */ +int __kprobes hw_breakpoint_exceptions_notify(struct notifier_block *unused, + unsigned long val, void *data) +{ + struct die_args *args = data; + + if (val != DIE_BREAKPOINT) + return NOTIFY_DONE; + + /* + * If the breakpoint hasn't been triggered by the UBC, it's + * probably from a debugger, so don't do anything more here. + * + * This also permits the UBC interface clock to remain off for + * non-UBC breakpoints, as we don't need to check the triggered + * or active channel masks. + */ + if (args->trapnr != sh_ubc->trap_nr) + return NOTIFY_DONE; + + return hw_breakpoint_handler(data); +} + +void hw_breakpoint_pmu_read(struct perf_event *bp) +{ + /* TODO */ +} + +void hw_breakpoint_pmu_unthrottle(struct perf_event *bp) +{ + /* TODO */ +} + +int register_sh_ubc(struct sh_ubc *ubc) +{ + /* Bail if it's already assigned */ + if (sh_ubc != &ubc_dummy) + return -EBUSY; + sh_ubc = ubc; + + pr_info("HW Breakpoints: %s UBC support registered\n", ubc->name); + + WARN_ON(ubc->num_events > HBP_NUM); + + return 0; +} diff --git a/arch/sh/kernel/kgdb.c b/arch/sh/kernel/kgdb.c index 3e532d0d4a5..70c69659b84 100644 --- a/arch/sh/kernel/kgdb.c +++ b/arch/sh/kernel/kgdb.c @@ -1,7 +1,7 @@ /* * SuperH KGDB support * - * Copyright (C) 2008 Paul Mundt + * Copyright (C) 2008 - 2009 Paul Mundt * * Single stepping taken from the old stub by Henry Bell and Jeremy Siegel. * @@ -251,24 +251,60 @@ BUILD_TRAP_HANDLER(singlestep) local_irq_restore(flags); } +static int __kgdb_notify(struct die_args *args, unsigned long cmd) +{ + int ret; + + switch (cmd) { + case DIE_BREAKPOINT: + /* + * This means a user thread is single stepping + * a system call which should be ignored + */ + if (test_thread_flag(TIF_SINGLESTEP)) + return NOTIFY_DONE; + + ret = kgdb_handle_exception(args->trapnr & 0xff, args->signr, + args->err, args->regs); + if (ret) + return NOTIFY_DONE; + + break; + } -BUILD_TRAP_HANDLER(breakpoint) + return NOTIFY_STOP; +} + +static int +kgdb_notify(struct notifier_block *self, unsigned long cmd, void *ptr) { unsigned long flags; - TRAP_HANDLER_DECL; + int ret; local_irq_save(flags); - kgdb_handle_exception(vec >> 2, SIGTRAP, 0, regs); + ret = __kgdb_notify(ptr, cmd); local_irq_restore(flags); + + return ret; } +static struct notifier_block kgdb_notifier = { + .notifier_call = kgdb_notify, + + /* + * Lowest-prio notifier priority, we want to be notified last: + */ + .priority = -INT_MAX, +}; + int kgdb_arch_init(void) { - return 0; + return register_die_notifier(&kgdb_notifier); } void kgdb_arch_exit(void) { + unregister_die_notifier(&kgdb_notifier); } struct kgdb_arch arch_kgdb_ops = { diff --git a/arch/sh/kernel/process.c b/arch/sh/kernel/process.c new file mode 100644 index 00000000000..81add9b9ea6 --- /dev/null +++ b/arch/sh/kernel/process.c @@ -0,0 +1,100 @@ +#include <linux/mm.h> +#include <linux/kernel.h> +#include <linux/sched.h> + +struct kmem_cache *task_xstate_cachep = NULL; +unsigned int xstate_size; + +int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src) +{ + *dst = *src; + + if (src->thread.xstate) { + dst->thread.xstate = kmem_cache_alloc(task_xstate_cachep, + GFP_KERNEL); + if (!dst->thread.xstate) + return -ENOMEM; + memcpy(dst->thread.xstate, src->thread.xstate, xstate_size); + } + + return 0; +} + +void free_thread_xstate(struct task_struct *tsk) +{ + if (tsk->thread.xstate) { + kmem_cache_free(task_xstate_cachep, tsk->thread.xstate); + tsk->thread.xstate = NULL; + } +} + +#if THREAD_SHIFT < PAGE_SHIFT +static struct kmem_cache *thread_info_cache; + +struct thread_info *alloc_thread_info(struct task_struct *tsk) +{ + struct thread_info *ti; + + ti = kmem_cache_alloc(thread_info_cache, GFP_KERNEL); + if (unlikely(ti == NULL)) + return NULL; +#ifdef CONFIG_DEBUG_STACK_USAGE + memset(ti, 0, THREAD_SIZE); +#endif + return ti; +} + +void free_thread_info(struct thread_info *ti) +{ + free_thread_xstate(ti->task); + kmem_cache_free(thread_info_cache, ti); +} + +void thread_info_cache_init(void) +{ + thread_info_cache = kmem_cache_create("thread_info", THREAD_SIZE, + THREAD_SIZE, SLAB_PANIC, NULL); +} +#else +struct thread_info *alloc_thread_info(struct task_struct *tsk) +{ +#ifdef CONFIG_DEBUG_STACK_USAGE + gfp_t mask = GFP_KERNEL | __GFP_ZERO; +#else + gfp_t mask = GFP_KERNEL; +#endif + return (struct thread_info *)__get_free_pages(mask, THREAD_SIZE_ORDER); +} + +void free_thread_info(struct thread_info *ti) +{ + free_thread_xstate(ti->task); + free_pages((unsigned long)ti, THREAD_SIZE_ORDER); +} +#endif /* THREAD_SHIFT < PAGE_SHIFT */ + +void arch_task_cache_init(void) +{ + if (!xstate_size) + return; + + task_xstate_cachep = kmem_cache_create("task_xstate", xstate_size, + __alignof__(union thread_xstate), + SLAB_PANIC | SLAB_NOTRACK, NULL); +} + +#ifdef CONFIG_SH_FPU_EMU +# define HAVE_SOFTFP 1 +#else +# define HAVE_SOFTFP 0 +#endif + +void init_thread_xstate(void) +{ + if (boot_cpu_data.flags & CPU_HAS_FPU) + xstate_size = sizeof(struct sh_fpu_hard_struct); + else if (HAVE_SOFTFP) + xstate_size = sizeof(struct sh_fpu_soft_struct); + else + xstate_size = 0; +} diff --git a/arch/sh/kernel/process_32.c b/arch/sh/kernel/process_32.c index d8af889366a..856010f9ebc 100644 --- a/arch/sh/kernel/process_32.c +++ b/arch/sh/kernel/process_32.c @@ -25,17 +25,15 @@ #include <linux/fs.h> #include <linux/ftrace.h> #include <linux/preempt.h> +#include <linux/hw_breakpoint.h> #include <asm/uaccess.h> #include <asm/mmu_context.h> #include <asm/pgalloc.h> #include <asm/system.h> -#include <asm/ubc.h> #include <asm/fpu.h> #include <asm/syscalls.h> #include <asm/watchdog.h> -int ubc_usercnt = 0; - #ifdef CONFIG_32BIT static void watchdog_trigger_immediate(void) { @@ -147,21 +145,34 @@ int kernel_thread(int (*fn)(void *), void * arg, unsigned long flags) } EXPORT_SYMBOL(kernel_thread); +void start_thread(struct pt_regs *regs, unsigned long new_pc, + unsigned long new_sp) +{ + set_fs(USER_DS); + + regs->pr = 0; + regs->sr = SR_FD; + regs->pc = new_pc; + regs->regs[15] = new_sp; + + free_thread_xstate(current); +} +EXPORT_SYMBOL(start_thread); + /* * Free current thread data structures etc.. */ void exit_thread(void) { - if (current->thread.ubc_pc) { - current->thread.ubc_pc = 0; - ubc_usercnt -= 1; - } } void flush_thread(void) { -#if defined(CONFIG_SH_FPU) struct task_struct *tsk = current; + + flush_ptrace_hw_breakpoint(tsk); + +#if defined(CONFIG_SH_FPU) /* Forget lazy FPU state */ clear_fpu(tsk, task_pt_regs(tsk)); clear_used_math(); @@ -209,11 +220,10 @@ int copy_thread(unsigned long clone_flags, unsigned long usp, { struct thread_info *ti = task_thread_info(p); struct pt_regs *childregs; + #if defined(CONFIG_SH_DSP) struct task_struct *tsk = current; -#endif -#if defined(CONFIG_SH_DSP) if (is_dsp_enabled(tsk)) { /* We can use the __save_dsp or just copy the struct: * __save_dsp(p); @@ -244,53 +254,11 @@ int copy_thread(unsigned long clone_flags, unsigned long usp, p->thread.sp = (unsigned long) childregs; p->thread.pc = (unsigned long) ret_from_fork; - p->thread.ubc_pc = 0; + memset(p->thread.ptrace_bps, 0, sizeof(p->thread.ptrace_bps)); return 0; } -/* Tracing by user break controller. */ -static void ubc_set_tracing(int asid, unsigned long pc) -{ -#if defined(CONFIG_CPU_SH4A) - unsigned long val; - - val = (UBC_CBR_ID_INST | UBC_CBR_RW_READ | UBC_CBR_CE); - val |= (UBC_CBR_AIE | UBC_CBR_AIV_SET(asid)); - - ctrl_outl(val, UBC_CBR0); - ctrl_outl(pc, UBC_CAR0); - ctrl_outl(0x0, UBC_CAMR0); - ctrl_outl(0x0, UBC_CBCR); - - val = (UBC_CRR_RES | UBC_CRR_PCB | UBC_CRR_BIE); - ctrl_outl(val, UBC_CRR0); - - /* Read UBC register that we wrote last, for checking update */ - val = ctrl_inl(UBC_CRR0); - -#else /* CONFIG_CPU_SH4A */ - ctrl_outl(pc, UBC_BARA); - -#ifdef CONFIG_MMU - ctrl_outb(asid, UBC_BASRA); -#endif - - ctrl_outl(0, UBC_BAMRA); - - if (current_cpu_data.type == CPU_SH7729 || - current_cpu_data.type == CPU_SH7710 || - current_cpu_data.type == CPU_SH7712 || - current_cpu_data.type == CPU_SH7203){ - ctrl_outw(BBR_INST | BBR_READ | BBR_CPU, UBC_BBRA); - ctrl_outl(BRCR_PCBA | BRCR_PCTE, UBC_BRCR); - } else { - ctrl_outw(BBR_INST | BBR_READ, UBC_BBRA); - ctrl_outw(BRCR_PCBA, UBC_BRCR); - } -#endif /* CONFIG_CPU_SH4A */ -} - /* * switch_to(x,y) should switch tasks from x to y. * @@ -304,7 +272,7 @@ __switch_to(struct task_struct *prev, struct task_struct *next) /* we're going to use this soon, after a few expensive things */ if (next->fpu_counter > 5) - prefetch(&next_t->fpu.hard); + prefetch(next_t->xstate); #ifdef CONFIG_MMU /* @@ -316,32 +284,13 @@ __switch_to(struct task_struct *prev, struct task_struct *next) : "r" (task_thread_info(next))); #endif - /* If no tasks are using the UBC, we're done */ - if (ubc_usercnt == 0) - /* If no tasks are using the UBC, we're done */; - else if (next->thread.ubc_pc && next->mm) { - int asid = 0; -#ifdef CONFIG_MMU - asid |= cpu_asid(smp_processor_id(), next->mm); -#endif - ubc_set_tracing(asid, next->thread.ubc_pc); - } else { -#if defined(CONFIG_CPU_SH4A) - ctrl_outl(UBC_CBR_INIT, UBC_CBR0); - ctrl_outl(UBC_CRR_INIT, UBC_CRR0); -#else - ctrl_outw(0, UBC_BBRA); - ctrl_outw(0, UBC_BBRB); -#endif - } - /* * If the task has used fpu the last 5 timeslices, just do a full * restore of the math state immediately to avoid the trap; the * chances of needing FPU soon are obviously high now */ if (next->fpu_counter > 5) - fpu_state_restore(task_pt_regs(next)); + __fpu_state_restore(); return prev; } @@ -434,20 +383,3 @@ unsigned long get_wchan(struct task_struct *p) return pc; } - -asmlinkage void break_point_trap(void) -{ - /* Clear tracing. */ -#if defined(CONFIG_CPU_SH4A) - ctrl_outl(UBC_CBR_INIT, UBC_CBR0); - ctrl_outl(UBC_CRR_INIT, UBC_CRR0); -#else - ctrl_outw(0, UBC_BBRA); - ctrl_outw(0, UBC_BBRB); - ctrl_outl(0, UBC_BRCR); -#endif - current->thread.ubc_pc = 0; - ubc_usercnt -= 1; - - force_sig(SIGTRAP, current); -} diff --git a/arch/sh/kernel/ptrace_32.c b/arch/sh/kernel/ptrace_32.c index 9be35f34809..c625cdab76d 100644 --- a/arch/sh/kernel/ptrace_32.c +++ b/arch/sh/kernel/ptrace_32.c @@ -2,7 +2,7 @@ * SuperH process tracing * * Copyright (C) 1999, 2000 Kaz Kojima & Niibe Yutaka - * Copyright (C) 2002 - 2008 Paul Mundt + * Copyright (C) 2002 - 2009 Paul Mundt * * Audit support by Yuichi Nakamura <ynakam@hitachisoft.jp> * @@ -26,6 +26,7 @@ #include <linux/tracehook.h> #include <linux/elf.h> #include <linux/regset.h> +#include <linux/hw_breakpoint.h> #include <asm/uaccess.h> #include <asm/pgtable.h> #include <asm/system.h> @@ -63,33 +64,64 @@ static inline int put_stack_long(struct task_struct *task, int offset, return 0; } -void user_enable_single_step(struct task_struct *child) +void ptrace_triggered(struct perf_event *bp, int nmi, + struct perf_sample_data *data, struct pt_regs *regs) { - /* Next scheduling will set up UBC */ - if (child->thread.ubc_pc == 0) - ubc_usercnt += 1; + struct perf_event_attr attr; + + /* + * Disable the breakpoint request here since ptrace has defined a + * one-shot behaviour for breakpoint exceptions. + */ + attr = bp->attr; + attr.disabled = true; + modify_user_hw_breakpoint(bp, &attr); +} + +static int set_single_step(struct task_struct *tsk, unsigned long addr) +{ + struct thread_struct *thread = &tsk->thread; + struct perf_event *bp; + struct perf_event_attr attr; + + bp = thread->ptrace_bps[0]; + if (!bp) { + hw_breakpoint_init(&attr); + + attr.bp_addr = addr; + attr.bp_len = HW_BREAKPOINT_LEN_2; + attr.bp_type = HW_BREAKPOINT_R; + + bp = register_user_hw_breakpoint(&attr, ptrace_triggered, tsk); + if (IS_ERR(bp)) + return PTR_ERR(bp); + + thread->ptrace_bps[0] = bp; + } else { + int err; + + attr = bp->attr; + attr.bp_addr = addr; + err = modify_user_hw_breakpoint(bp, &attr); + if (unlikely(err)) + return err; + } + + return 0; +} - child->thread.ubc_pc = get_stack_long(child, - offsetof(struct pt_regs, pc)); +void user_enable_single_step(struct task_struct *child) +{ + unsigned long pc = get_stack_long(child, offsetof(struct pt_regs, pc)); set_tsk_thread_flag(child, TIF_SINGLESTEP); + + set_single_step(child, pc); } void user_disable_single_step(struct task_struct *child) { clear_tsk_thread_flag(child, TIF_SINGLESTEP); - - /* - * Ensure the UBC is not programmed at the next context switch. - * - * Normally this is not needed but there are sequences such as - * singlestep, signal delivery, and continue that leave the - * ubc_pc non-zero leading to spurious SIGTRAPs. - */ - if (child->thread.ubc_pc != 0) { - ubc_usercnt -= 1; - child->thread.ubc_pc = 0; - } } /* @@ -163,10 +195,10 @@ int fpregs_get(struct task_struct *target, if ((boot_cpu_data.flags & CPU_HAS_FPU)) return user_regset_copyout(&pos, &count, &kbuf, &ubuf, - &target->thread.fpu.hard, 0, -1); + &target->thread.xstate->hardfpu, 0, -1); return user_regset_copyout(&pos, &count, &kbuf, &ubuf, - &target->thread.fpu.soft, 0, -1); + &target->thread.xstate->softfpu, 0, -1); } static int fpregs_set(struct task_struct *target, @@ -184,10 +216,10 @@ static int fpregs_set(struct task_struct *target, if ((boot_cpu_data.flags & CPU_HAS_FPU)) return user_regset_copyin(&pos, &count, &kbuf, &ubuf, - &target->thread.fpu.hard, 0, -1); + &target->thread.xstate->hardfpu, 0, -1); return user_regset_copyin(&pos, &count, &kbuf, &ubuf, - &target->thread.fpu.soft, 0, -1); + &target->thread.xstate->softfpu, 0, -1); } static int fpregs_active(struct task_struct *target, @@ -333,7 +365,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data) else tmp = 0; } else - tmp = ((long *)&child->thread.fpu) + tmp = ((long *)child->thread.xstate) [(addr - (long)&dummy->fpu) >> 2]; } else if (addr == (long) &dummy->u_fpvalid) tmp = !!tsk_used_math(child); @@ -362,7 +394,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data) else if (addr >= (long) &dummy->fpu && addr < (long) &dummy->u_fpvalid) { set_stopped_child_used_math(child); - ((long *)&child->thread.fpu) + ((long *)child->thread.xstate) [(addr - (long)&dummy->fpu) >> 2] = data; ret = 0; } else if (addr == (long) &dummy->u_fpvalid) { diff --git a/arch/sh/kernel/signal_32.c b/arch/sh/kernel/signal_32.c index 12815ce01ec..6a7cce79eb4 100644 --- a/arch/sh/kernel/signal_32.c +++ b/arch/sh/kernel/signal_32.c @@ -150,7 +150,7 @@ static inline int restore_sigcontext_fpu(struct sigcontext __user *sc) return 0; set_used_math(); - return __copy_from_user(&tsk->thread.fpu.hard, &sc->sc_fpregs[0], + return __copy_from_user(&tsk->thread.xstate->hardfpu, &sc->sc_fpregs[0], sizeof(long)*(16*2+2)); } @@ -175,7 +175,7 @@ static inline int save_sigcontext_fpu(struct sigcontext __user *sc, clear_used_math(); unlazy_fpu(tsk, regs); - return __copy_to_user(&sc->sc_fpregs[0], &tsk->thread.fpu.hard, + return __copy_to_user(&sc->sc_fpregs[0], &tsk->thread.xstate->hardfpu, sizeof(long)*(16*2+2)); } #endif /* CONFIG_SH_FPU */ diff --git a/arch/sh/kernel/traps_32.c b/arch/sh/kernel/traps_32.c index 204def6ecb6..9c090cb6887 100644 --- a/arch/sh/kernel/traps_32.c +++ b/arch/sh/kernel/traps_32.c @@ -846,7 +846,7 @@ void __init trap_init(void) #endif #ifdef TRAP_UBC - set_exception_table_vec(TRAP_UBC, break_point_trap); + set_exception_table_vec(TRAP_UBC, breakpoint_trap_handler); #endif /* Save off the BIOS VBR, if there is one */ diff --git a/arch/sh/math-emu/math.c b/arch/sh/math-emu/math.c index d6c15cae091..1fcdb122097 100644 --- a/arch/sh/math-emu/math.c +++ b/arch/sh/math-emu/math.c @@ -471,10 +471,10 @@ static int fpu_emulate(u16 code, struct sh_fpu_soft_struct *fregs, struct pt_reg * denormal_to_double - Given denormalized float number, * store double float * - * @fpu: Pointer to sh_fpu_hard structure + * @fpu: Pointer to sh_fpu_soft structure * @n: Index to FP register */ -static void denormal_to_double(struct sh_fpu_hard_struct *fpu, int n) +static void denormal_to_double(struct sh_fpu_soft_struct *fpu, int n) { unsigned long du, dl; unsigned long x = fpu->fpul; @@ -552,11 +552,11 @@ static int ieee_fpe_handler(struct pt_regs *regs) if ((finsn & 0xf1ff) == 0xf0ad) { /* fcnvsd */ struct task_struct *tsk = current; - if ((tsk->thread.fpu.hard.fpscr & (1 << 17))) { + if ((tsk->thread.xstate->softfpu.fpscr & (1 << 17))) { /* FPU error */ - denormal_to_double (&tsk->thread.fpu.hard, + denormal_to_double (&tsk->thread.xstate->softfpu, (finsn >> 8) & 0xf); - tsk->thread.fpu.hard.fpscr &= + tsk->thread.xstate->softfpu.fpscr &= ~(FPSCR_CAUSE_MASK | FPSCR_FLAG_MASK); task_thread_info(tsk)->status |= TS_USEDFPU; } else { @@ -617,7 +617,7 @@ static void fpu_init(struct sh_fpu_soft_struct *fpu) int do_fpu_inst(unsigned short inst, struct pt_regs *regs) { struct task_struct *tsk = current; - struct sh_fpu_soft_struct *fpu = &(tsk->thread.fpu.soft); + struct sh_fpu_soft_struct *fpu = &(tsk->thread.xstate->softfpu); if (!(task_thread_info(tsk)->status & TS_USEDFPU)) { /* initialize once. */ diff --git a/arch/sh/mm/init.c b/arch/sh/mm/init.c index 761910d142f..d5fb014279a 100644 --- a/arch/sh/mm/init.c +++ b/arch/sh/mm/init.c @@ -283,35 +283,6 @@ void free_initrd_mem(unsigned long start, unsigned long end) } #endif -#if THREAD_SHIFT < PAGE_SHIFT -static struct kmem_cache *thread_info_cache; - -struct thread_info *alloc_thread_info(struct task_struct *tsk) -{ - struct thread_info *ti; - - ti = kmem_cache_alloc(thread_info_cache, GFP_KERNEL); - if (unlikely(ti == NULL)) - return NULL; -#ifdef CONFIG_DEBUG_STACK_USAGE - memset(ti, 0, THREAD_SIZE); -#endif - return ti; -} - -void free_thread_info(struct thread_info *ti) -{ - kmem_cache_free(thread_info_cache, ti); -} - -void thread_info_cache_init(void) -{ - thread_info_cache = kmem_cache_create("thread_info", THREAD_SIZE, - THREAD_SIZE, 0, NULL); - BUG_ON(thread_info_cache == NULL); -} -#endif /* THREAD_SHIFT < PAGE_SHIFT */ - #ifdef CONFIG_MEMORY_HOTPLUG int arch_add_memory(int nid, u64 start, u64 size) { |