diff options
Diffstat (limited to 'arch/sh/kernel')
38 files changed, 1790 insertions, 941 deletions
diff --git a/arch/sh/kernel/Makefile b/arch/sh/kernel/Makefile index 0d587da1ef12..02fd3ae8b0ee 100644 --- a/arch/sh/kernel/Makefile +++ b/arch/sh/kernel/Makefile @@ -13,8 +13,9 @@ 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 \ + reboot.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 @@ -22,7 +23,7 @@ obj-y := debugtraps.o dma-nommu.o dumpstack.o \ obj-y += cpu/ obj-$(CONFIG_VSYSCALL) += vsyscall/ obj-$(CONFIG_SMP) += smp.o -obj-$(CONFIG_SH_STANDARD_BIOS) += sh_bios.o early_printk.o +obj-$(CONFIG_SH_STANDARD_BIOS) += sh_bios.o obj-$(CONFIG_KGDB) += kgdb.o obj-$(CONFIG_SH_CPU_FREQ) += cpufreq.o obj-$(CONFIG_MODULES) += sh_ksyms_$(BITS).o module.o @@ -39,6 +40,7 @@ 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-$(CONFIG_GENERIC_CLOCKEVENTS_BROADCAST) += localtimer.o EXTRA_CFLAGS += -Werror diff --git a/arch/sh/kernel/cpu/Makefile b/arch/sh/kernel/cpu/Makefile index d97c803719ec..0e48bc61c272 100644 --- a/arch/sh/kernel/cpu/Makefile +++ b/arch/sh/kernel/cpu/Makefile @@ -17,5 +17,7 @@ obj-$(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/clock-cpg.c b/arch/sh/kernel/cpu/clock-cpg.c index 6dfe2cced3fc..2827abb5d2ab 100644 --- a/arch/sh/kernel/cpu/clock-cpg.c +++ b/arch/sh/kernel/cpu/clock-cpg.c @@ -160,13 +160,81 @@ static unsigned long sh_clk_div4_recalc(struct clk *clk) return clk->freq_table[idx].frequency; } +static int sh_clk_div4_set_parent(struct clk *clk, struct clk *parent) +{ + struct clk_div_mult_table *table = clk->priv; + u32 value; + int ret; + + if (!strcmp("pll_clk", parent->name)) + value = __raw_readl(clk->enable_reg) & ~(1 << 7); + else + value = __raw_readl(clk->enable_reg) | (1 << 7); + + ret = clk_reparent(clk, parent); + if (ret < 0) + return ret; + + __raw_writel(value, clk->enable_reg); + + /* Rebiuld the frequency table */ + clk_rate_table_build(clk, clk->freq_table, table->nr_divisors, + table, &clk->arch_flags); + + return 0; +} + +static int sh_clk_div4_set_rate(struct clk *clk, unsigned long rate, int algo_id) +{ + unsigned long value; + int idx = clk_rate_table_find(clk, clk->freq_table, rate); + if (idx < 0) + return idx; + + value = __raw_readl(clk->enable_reg); + value &= ~0xf; + value |= idx; + __raw_writel(value, clk->enable_reg); + + return 0; +} + +static int sh_clk_div4_enable(struct clk *clk) +{ + __raw_writel(__raw_readl(clk->enable_reg) & ~(1 << 8), clk->enable_reg); + return 0; +} + +static void sh_clk_div4_disable(struct clk *clk) +{ + __raw_writel(__raw_readl(clk->enable_reg) | (1 << 8), clk->enable_reg); +} + static struct clk_ops sh_clk_div4_clk_ops = { .recalc = sh_clk_div4_recalc, + .set_rate = sh_clk_div4_set_rate, .round_rate = sh_clk_div_round_rate, }; -int __init sh_clk_div4_register(struct clk *clks, int nr, - struct clk_div_mult_table *table) +static struct clk_ops sh_clk_div4_enable_clk_ops = { + .recalc = sh_clk_div4_recalc, + .set_rate = sh_clk_div4_set_rate, + .round_rate = sh_clk_div_round_rate, + .enable = sh_clk_div4_enable, + .disable = sh_clk_div4_disable, +}; + +static struct clk_ops sh_clk_div4_reparent_clk_ops = { + .recalc = sh_clk_div4_recalc, + .set_rate = sh_clk_div4_set_rate, + .round_rate = sh_clk_div_round_rate, + .enable = sh_clk_div4_enable, + .disable = sh_clk_div4_disable, + .set_parent = sh_clk_div4_set_parent, +}; + +static int __init sh_clk_div4_register_ops(struct clk *clks, int nr, + struct clk_div_mult_table *table, struct clk_ops *ops) { struct clk *clkp; void *freq_table; @@ -185,7 +253,7 @@ int __init sh_clk_div4_register(struct clk *clks, int nr, for (k = 0; !ret && (k < nr); k++) { clkp = clks + k; - clkp->ops = &sh_clk_div4_clk_ops; + clkp->ops = ops; clkp->id = -1; clkp->priv = table; @@ -198,6 +266,26 @@ int __init sh_clk_div4_register(struct clk *clks, int nr, return ret; } +int __init sh_clk_div4_register(struct clk *clks, int nr, + struct clk_div_mult_table *table) +{ + return sh_clk_div4_register_ops(clks, nr, table, &sh_clk_div4_clk_ops); +} + +int __init sh_clk_div4_enable_register(struct clk *clks, int nr, + struct clk_div_mult_table *table) +{ + return sh_clk_div4_register_ops(clks, nr, table, + &sh_clk_div4_enable_clk_ops); +} + +int __init sh_clk_div4_reparent_register(struct clk *clks, int nr, + struct clk_div_mult_table *table) +{ + return sh_clk_div4_register_ops(clks, nr, table, + &sh_clk_div4_reparent_clk_ops); +} + #ifdef CONFIG_SH_CLK_CPG_LEGACY static struct clk master_clk = { .name = "master_clk", diff --git a/arch/sh/kernel/cpu/fpu.c b/arch/sh/kernel/cpu/fpu.c new file mode 100644 index 000000000000..f059ed62cf57 --- /dev/null +++ b/arch/sh/kernel/cpu/fpu.c @@ -0,0 +1,84 @@ +#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)) { + local_irq_enable(); + /* + * does a slab alloc which can sleep + */ + if (init_fpu(tsk)) { + /* + * ran out of memory! + */ + do_group_exit(SIGKILL); + return; + } + local_irq_disable(); + } + + 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 89b4b76c0d76..a5bb0550bbf3 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 d395ce5740e7..488d24e0cdf0 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 46610c35c232..99b4d020179a 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 e97857aec8a0..447482d7f65e 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/sh4/probe.c b/arch/sh/kernel/cpu/sh4/probe.c index d36f0c45f55f..cc02b3145cca 100644 --- a/arch/sh/kernel/cpu/sh4/probe.c +++ b/arch/sh/kernel/cpu/sh4/probe.c @@ -71,11 +71,11 @@ int __init detect_cpu_and_cache_system(void) boot_cpu_data.dcache.ways = 4; } else { /* And some SH-4 defaults.. */ - boot_cpu_data.flags |= CPU_HAS_PTEA; + boot_cpu_data.flags |= CPU_HAS_PTEA | CPU_HAS_FPU; boot_cpu_data.family = CPU_FAMILY_SH4; } - /* FPU detection works for everyone */ + /* FPU detection works for almost everyone */ if ((cvr & 0x20000000)) boot_cpu_data.flags |= CPU_HAS_FPU; @@ -124,6 +124,7 @@ int __init detect_cpu_and_cache_system(void) boot_cpu_data.type = CPU_SH7785; break; case 0x4004: + case 0x4005: boot_cpu_data.type = CPU_SH7786; boot_cpu_data.flags |= CPU_HAS_PTEAEX | CPU_HAS_L2_CACHE; break; @@ -160,6 +161,7 @@ int __init detect_cpu_and_cache_system(void) break; case 0x700: boot_cpu_data.type = CPU_SH4_501; + boot_cpu_data.flags &= ~CPU_HAS_FPU; boot_cpu_data.icache.ways = 2; boot_cpu_data.dcache.ways = 2; break; @@ -227,7 +229,7 @@ int __init detect_cpu_and_cache_system(void) * Size calculation is much more sensible * than it is for the L1. * - * Sizes are 128KB, 258KB, 512KB, and 1MB. + * Sizes are 128KB, 256KB, 512KB, and 1MB. */ size = (cvr & 0xf) << 17; diff --git a/arch/sh/kernel/cpu/sh4a/Makefile b/arch/sh/kernel/cpu/sh4a/Makefile index 33bab477d2e2..b144e8af89dc 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/clock-sh7722.c b/arch/sh/kernel/cpu/sh4a/clock-sh7722.c index ea38b554dc05..860ee2bf4bf0 100644 --- a/arch/sh/kernel/cpu/sh4a/clock-sh7722.c +++ b/arch/sh/kernel/cpu/sh4a/clock-sh7722.c @@ -117,12 +117,11 @@ static struct clk_div_mult_table div4_table = { .nr_multipliers = ARRAY_SIZE(multipliers), }; -enum { DIV4_I, DIV4_U, DIV4_SH, DIV4_B, DIV4_B3, DIV4_P, - DIV4_SIUA, DIV4_SIUB, DIV4_IRDA, DIV4_NR }; - #define DIV4(_str, _reg, _bit, _mask, _flags) \ SH_CLK_DIV4(_str, &pll_clk, _reg, _bit, _mask, _flags) +enum { DIV4_I, DIV4_U, DIV4_SH, DIV4_B, DIV4_B3, DIV4_P, DIV4_NR }; + struct clk div4_clks[DIV4_NR] = { [DIV4_I] = DIV4("cpu_clk", FRQCR, 20, 0x1fef, CLK_ENABLE_ON_INIT), [DIV4_U] = DIV4("umem_clk", FRQCR, 16, 0x1fff, CLK_ENABLE_ON_INIT), @@ -130,9 +129,19 @@ struct clk div4_clks[DIV4_NR] = { [DIV4_B] = DIV4("bus_clk", FRQCR, 8, 0x1fff, CLK_ENABLE_ON_INIT), [DIV4_B3] = DIV4("b3_clk", FRQCR, 4, 0x1fff, CLK_ENABLE_ON_INIT), [DIV4_P] = DIV4("peripheral_clk", FRQCR, 0, 0x1fff, 0), +}; + +enum { DIV4_IRDA, DIV4_ENABLE_NR }; + +struct clk div4_enable_clks[DIV4_ENABLE_NR] = { + [DIV4_IRDA] = DIV4("irda_clk", IRDACLKCR, 0, 0x1fff, 0), +}; + +enum { DIV4_SIUA, DIV4_SIUB, DIV4_REPARENT_NR }; + +struct clk div4_reparent_clks[DIV4_REPARENT_NR] = { [DIV4_SIUA] = DIV4("siua_clk", SCLKACR, 0, 0x1fff, 0), [DIV4_SIUB] = DIV4("siub_clk", SCLKBCR, 0, 0x1fff, 0), - [DIV4_IRDA] = DIV4("irda_clk", IRDACLKCR, 0, 0x1fff, 0), }; struct clk div6_clks[] = { @@ -189,6 +198,14 @@ int __init arch_clk_init(void) ret = sh_clk_div4_register(div4_clks, DIV4_NR, &div4_table); if (!ret) + ret = sh_clk_div4_enable_register(div4_enable_clks, + DIV4_ENABLE_NR, &div4_table); + + if (!ret) + ret = sh_clk_div4_reparent_register(div4_reparent_clks, + DIV4_REPARENT_NR, &div4_table); + + if (!ret) ret = sh_clk_div6_register(div6_clks, ARRAY_SIZE(div6_clks)); if (!ret) diff --git a/arch/sh/kernel/cpu/sh4a/clock-sh7786.c b/arch/sh/kernel/cpu/sh4a/clock-sh7786.c index a0e8869071ac..494c636012bb 100644 --- a/arch/sh/kernel/cpu/sh4a/clock-sh7786.c +++ b/arch/sh/kernel/cpu/sh4a/clock-sh7786.c @@ -3,11 +3,7 @@ * * SH7786 support for the clock framework * - * Copyright (C) 2008, 2009 Renesas Solutions Corp. - * Kuninori Morimoto <morimoto.kuninori@renesas.com> - * - * Based on SH7785 - * Copyright (C) 2007 Paul Mundt + * Copyright (C) 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 @@ -15,127 +11,123 @@ */ #include <linux/init.h> #include <linux/kernel.h> +#include <linux/clk.h> +#include <linux/io.h> #include <asm/clock.h> #include <asm/freq.h> -#include <asm/io.h> - -static int ifc_divisors[] = { 1, 2, 4, 1 }; -static int sfc_divisors[] = { 1, 1, 4, 1 }; -static int bfc_divisors[] = { 1, 1, 1, 1, 1, 12, 16, 1, - 24, 32, 1, 1, 1, 1, 1, 1 }; -static int mfc_divisors[] = { 1, 1, 4, 1 }; -static int pfc_divisors[] = { 1, 1, 1, 1, 1, 1, 16, 1, - 24, 32, 1, 48, 1, 1, 1, 1 }; - -static void master_clk_init(struct clk *clk) -{ - clk->rate *= pfc_divisors[ctrl_inl(FRQMR1) & 0x000f]; -} -static struct clk_ops sh7786_master_clk_ops = { - .init = master_clk_init, +/* + * Default rate for the root input clock, reset this with clk_set_rate() + * from the platform code. + */ +static struct clk extal_clk = { + .name = "extal", + .id = -1, + .rate = 33333333, }; -static unsigned long module_clk_recalc(struct clk *clk) +static unsigned long pll_recalc(struct clk *clk) { - int idx = (ctrl_inl(FRQMR1) & 0x000f); - return clk->parent->rate / pfc_divisors[idx]; -} + int multiplier; -static struct clk_ops sh7786_module_clk_ops = { - .recalc = module_clk_recalc, -}; + /* + * Clock modes 0, 1, and 2 use an x64 multiplier against PLL1, + * while modes 3, 4, and 5 use an x32. + */ + multiplier = (sh_mv.mv_mode_pins() & 0xf) < 3 ? 64 : 32; -static unsigned long bus_clk_recalc(struct clk *clk) -{ - int idx = ((ctrl_inl(FRQMR1) >> 16) & 0x000f); - return clk->parent->rate / bfc_divisors[idx]; + return clk->parent->rate * multiplier; } -static struct clk_ops sh7786_bus_clk_ops = { - .recalc = bus_clk_recalc, +static struct clk_ops pll_clk_ops = { + .recalc = pll_recalc, }; -static unsigned long cpu_clk_recalc(struct clk *clk) -{ - int idx = ((ctrl_inl(FRQMR1) >> 28) & 0x0003); - return clk->parent->rate / ifc_divisors[idx]; -} - -static struct clk_ops sh7786_cpu_clk_ops = { - .recalc = cpu_clk_recalc, +static struct clk pll_clk = { + .name = "pll_clk", + .id = -1, + .ops = &pll_clk_ops, + .parent = &extal_clk, + .flags = CLK_ENABLE_ON_INIT, }; -static struct clk_ops *sh7786_clk_ops[] = { - &sh7786_master_clk_ops, - &sh7786_module_clk_ops, - &sh7786_bus_clk_ops, - &sh7786_cpu_clk_ops, +static struct clk *clks[] = { + &extal_clk, + &pll_clk, }; -void __init arch_init_clk_ops(struct clk_ops **ops, int idx) -{ - if (idx < ARRAY_SIZE(sh7786_clk_ops)) - *ops = sh7786_clk_ops[idx]; -} - -static unsigned long shyway_clk_recalc(struct clk *clk) -{ - int idx = ((ctrl_inl(FRQMR1) >> 20) & 0x0003); - return clk->parent->rate / sfc_divisors[idx]; -} +static unsigned int div2[] = { 1, 2, 4, 6, 8, 12, 16, 18, + 24, 32, 36, 48 }; -static struct clk_ops sh7786_shyway_clk_ops = { - .recalc = shyway_clk_recalc, +static struct clk_div_mult_table div4_table = { + .divisors = div2, + .nr_divisors = ARRAY_SIZE(div2), }; -static struct clk sh7786_shyway_clk = { - .name = "shyway_clk", - .flags = CLK_ENABLE_ON_INIT, - .ops = &sh7786_shyway_clk_ops, -}; +enum { DIV4_I, DIV4_SH, DIV4_B, DIV4_DDR, DIV4_DU, DIV4_P, DIV4_NR }; -static unsigned long ddr_clk_recalc(struct clk *clk) -{ - int idx = ((ctrl_inl(FRQMR1) >> 12) & 0x0003); - return clk->parent->rate / mfc_divisors[idx]; -} +#define DIV4(_str, _bit, _mask, _flags) \ + SH_CLK_DIV4(_str, &pll_clk, FRQMR1, _bit, _mask, _flags) -static struct clk_ops sh7786_ddr_clk_ops = { - .recalc = ddr_clk_recalc, +struct clk div4_clks[DIV4_NR] = { + [DIV4_P] = DIV4("peripheral_clk", 0, 0x0b40, 0), + [DIV4_DU] = DIV4("du_clk", 4, 0x0010, 0), + [DIV4_DDR] = DIV4("ddr_clk", 12, 0x0002, CLK_ENABLE_ON_INIT), + [DIV4_B] = DIV4("bus_clk", 16, 0x0360, CLK_ENABLE_ON_INIT), + [DIV4_SH] = DIV4("shyway_clk", 20, 0x0002, CLK_ENABLE_ON_INIT), + [DIV4_I] = DIV4("cpu_clk", 28, 0x0006, CLK_ENABLE_ON_INIT), }; -static struct clk sh7786_ddr_clk = { - .name = "ddr_clk", - .flags = CLK_ENABLE_ON_INIT, - .ops = &sh7786_ddr_clk_ops, -}; - -/* - * Additional SH7786-specific on-chip clocks that aren't already part of the - * clock framework - */ -static struct clk *sh7786_onchip_clocks[] = { - &sh7786_shyway_clk, - &sh7786_ddr_clk, +#define MSTPCR0 0xffc40030 +#define MSTPCR1 0xffc40034 + +static struct clk mstp_clks[] = { + /* MSTPCR0 */ + SH_CLK_MSTP32("scif_fck", 5, &div4_clks[DIV4_P], MSTPCR0, 29, 0), + SH_CLK_MSTP32("scif_fck", 4, &div4_clks[DIV4_P], MSTPCR0, 28, 0), + SH_CLK_MSTP32("scif_fck", 3, &div4_clks[DIV4_P], MSTPCR0, 27, 0), + SH_CLK_MSTP32("scif_fck", 2, &div4_clks[DIV4_P], MSTPCR0, 26, 0), + SH_CLK_MSTP32("scif_fck", 1, &div4_clks[DIV4_P], MSTPCR0, 25, 0), + SH_CLK_MSTP32("scif_fck", 0, &div4_clks[DIV4_P], MSTPCR0, 24, 0), + SH_CLK_MSTP32("ssi_fck", 3, &div4_clks[DIV4_P], MSTPCR0, 23, 0), + SH_CLK_MSTP32("ssi_fck", 2, &div4_clks[DIV4_P], MSTPCR0, 22, 0), + SH_CLK_MSTP32("ssi_fck", 1, &div4_clks[DIV4_P], MSTPCR0, 21, 0), + SH_CLK_MSTP32("ssi_fck", 0, &div4_clks[DIV4_P], MSTPCR0, 20, 0), + SH_CLK_MSTP32("hac_fck", 1, &div4_clks[DIV4_P], MSTPCR0, 17, 0), + SH_CLK_MSTP32("hac_fck", 0, &div4_clks[DIV4_P], MSTPCR0, 16, 0), + SH_CLK_MSTP32("i2c_fck", 1, &div4_clks[DIV4_P], MSTPCR0, 15, 0), + SH_CLK_MSTP32("i2c_fck", 0, &div4_clks[DIV4_P], MSTPCR0, 14, 0), + SH_CLK_MSTP32("tmu9_11_fck", -1, &div4_clks[DIV4_P], MSTPCR0, 11, 0), + SH_CLK_MSTP32("tmu678_fck", -1, &div4_clks[DIV4_P], MSTPCR0, 10, 0), + SH_CLK_MSTP32("tmu345_fck", -1, &div4_clks[DIV4_P], MSTPCR0, 9, 0), + SH_CLK_MSTP32("tmu012_fck", -1, &div4_clks[DIV4_P], MSTPCR0, 8, 0), + SH_CLK_MSTP32("sdif_fck", 1, &div4_clks[DIV4_P], MSTPCR0, 5, 0), + SH_CLK_MSTP32("sdif_fck", 0, &div4_clks[DIV4_P], MSTPCR0, 4, 0), + SH_CLK_MSTP32("hspi_fck", -1, &div4_clks[DIV4_P], MSTPCR0, 2, 0), + + /* MSTPCR1 */ + SH_CLK_MSTP32("usb_fck", -1, NULL, MSTPCR1, 12, 0), + SH_CLK_MSTP32("pcie_fck", 2, NULL, MSTPCR1, 10, 0), + SH_CLK_MSTP32("pcie_fck", 1, NULL, MSTPCR1, 9, 0), + SH_CLK_MSTP32("pcie_fck", 0, NULL, MSTPCR1, 8, 0), + SH_CLK_MSTP32("dmac_11_6_fck", -1, NULL, MSTPCR1, 5, 0), + SH_CLK_MSTP32("dmac_5_0_fck", -1, NULL, MSTPCR1, 4, 0), + SH_CLK_MSTP32("du_fck", -1, NULL, MSTPCR1, 3, 0), + SH_CLK_MSTP32("ether_fck", -1, NULL, MSTPCR1, 2, 0), }; int __init arch_clk_init(void) { - struct clk *clk; int i, ret = 0; - cpg_clk_init(); - - clk = clk_get(NULL, "master_clk"); - for (i = 0; i < ARRAY_SIZE(sh7786_onchip_clocks); i++) { - struct clk *clkp = sh7786_onchip_clocks[i]; - - clkp->parent = clk; - ret |= clk_register(clkp); - } + for (i = 0; i < ARRAY_SIZE(clks); i++) + ret |= clk_register(clks[i]); - clk_put(clk); + if (!ret) + ret = sh_clk_div4_register(div4_clks, ARRAY_SIZE(div4_clks), + &div4_table); + if (!ret) + ret = sh_clk_mstp32_register(mstp_clks, ARRAY_SIZE(mstp_clks)); return ret; } diff --git a/arch/sh/kernel/cpu/sh4a/smp-shx3.c b/arch/sh/kernel/cpu/sh4a/smp-shx3.c index 5863e0c4d02f..11bf4c1e25c0 100644 --- a/arch/sh/kernel/cpu/sh4a/smp-shx3.c +++ b/arch/sh/kernel/cpu/sh4a/smp-shx3.c @@ -78,7 +78,10 @@ void __init plat_prepare_cpus(unsigned int max_cpus) void plat_start_cpu(unsigned int cpu, unsigned long entry_point) { - __raw_writel(entry_point, RESET_REG(cpu)); + if (__in_29bit_mode()) + __raw_writel(entry_point, RESET_REG(cpu)); + else + __raw_writel(virt_to_phys(entry_point), RESET_REG(cpu)); if (!(__raw_readl(STBCR_REG(cpu)) & STBCR_MSTP)) __raw_writel(STBCR_MSTP, STBCR_REG(cpu)); diff --git a/arch/sh/kernel/cpu/sh4a/ubc.c b/arch/sh/kernel/cpu/sh4a/ubc.c new file mode 100644 index 000000000000..efb2745bcb36 --- /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/cpu/sh5/entry.S b/arch/sh/kernel/cpu/sh5/entry.S index 8f13f73cb2cb..6b80295dd7a4 100644 --- a/arch/sh/kernel/cpu/sh5/entry.S +++ b/arch/sh/kernel/cpu/sh5/entry.S @@ -187,7 +187,7 @@ trap_jtable: .rept 6 .long do_exception_error /* 0x880 - 0x920 */ .endr - .long do_software_break_point /* 0x940 */ + .long breakpoint_trap_handler /* 0x940 */ .long do_exception_error /* 0x960 */ .long do_single_step /* 0x980 */ @@ -1124,7 +1124,7 @@ fpu_error_or_IRQA: pta its_IRQ, tr0 beqi/l r4, EVENT_INTERRUPT, tr0 #ifdef CONFIG_SH_FPU - movi do_fpu_state_restore, r6 + movi fpu_state_restore_trap_handler, r6 #else movi do_exception_error, r6 #endif @@ -1135,7 +1135,7 @@ fpu_error_or_IRQB: pta its_IRQ, tr0 beqi/l r4, EVENT_INTERRUPT, tr0 #ifdef CONFIG_SH_FPU - movi do_fpu_state_restore, r6 + movi fpu_state_restore_trap_handler, r6 #else movi do_exception_error, r6 #endif diff --git a/arch/sh/kernel/cpu/sh5/fpu.c b/arch/sh/kernel/cpu/sh5/fpu.c index 4648ccee6c4d..4b3bb35e99f3 100644 --- a/arch/sh/kernel/cpu/sh5/fpu.c +++ b/arch/sh/kernel/cpu/sh5/fpu.c @@ -15,24 +15,6 @@ #include <linux/sched.h> #include <linux/signal.h> #include <asm/processor.h> -#include <asm/user.h> -#include <asm/io.h> -#include <asm/fpu.h> - -/* - * Initially load the FPU with signalling NANS. This bit pattern - * has the property that no matter whether considered as single or as - * double precision, it still represents a signalling NAN. - */ -#define sNAN64 0xFFFFFFFFFFFFFFFFULL -#define sNAN32 0xFFFFFFFFUL - -static union sh_fpu_union init_fpuregs = { - .hard = { - .fp_regs = { [0 ... 63] = sNAN32 }, - .fpscr = FPSCR_INIT - } -}; void save_fpu(struct task_struct *tsk) { @@ -72,12 +54,11 @@ void save_fpu(struct task_struct *tsk) "fgetscr fr63\n\t" "fst.s %0, (32*8), fr63\n\t" : /* no output */ - : "r" (&tsk->thread.fpu.hard) + : "r" (&tsk->thread.xstate->hardfpu) : "memory"); } -static inline void -fpload(struct sh_fpu_hard_struct *fpregs) +void restore_fpu(struct task_struct *tsk) { asm volatile("fld.p %0, (0*8), fp0\n\t" "fld.p %0, (1*8), fp2\n\t" @@ -116,16 +97,11 @@ fpload(struct sh_fpu_hard_struct *fpregs) "fld.p %0, (31*8), fp62\n\t" : /* no output */ - : "r" (fpregs) ); -} - -void fpinit(struct sh_fpu_hard_struct *fpregs) -{ - *fpregs = init_fpuregs.hard; + : "r" (&tsk->thread.xstate->hardfpu) + : "memory"); } -asmlinkage void -do_fpu_error(unsigned long ex, struct pt_regs *regs) +asmlinkage void do_fpu_error(unsigned long ex, struct pt_regs *regs) { struct task_struct *tsk = current; @@ -133,35 +109,6 @@ do_fpu_error(unsigned long ex, struct pt_regs *regs) tsk->thread.trap_no = 11; tsk->thread.error_code = 0; - force_sig(SIGFPE, tsk); -} - - -asmlinkage void -do_fpu_state_restore(unsigned long ex, struct pt_regs *regs) -{ - void die(const char *str, struct pt_regs *regs, long err); - - if (! user_mode(regs)) - die("FPU used in kernel", regs, ex); - regs->sr &= ~SR_FD; - - if (last_task_used_math == current) - return; - - enable_fpu(); - if (last_task_used_math != NULL) - /* Other processes fpu state, save away */ - save_fpu(last_task_used_math); - - last_task_used_math = current; - if (used_math()) { - fpload(¤t->thread.fpu.hard); - } else { - /* First time FPU user. */ - fpload(&init_fpuregs.hard); - set_used_math(); - } - disable_fpu(); + force_sig(SIGFPE, tsk); } diff --git a/arch/sh/kernel/debugtraps.S b/arch/sh/kernel/debugtraps.S index 591741383ee6..7a1b46fec0f4 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/early_printk.c b/arch/sh/kernel/early_printk.c deleted file mode 100644 index f8bb50c6e050..000000000000 --- a/arch/sh/kernel/early_printk.c +++ /dev/null @@ -1,85 +0,0 @@ -/* - * arch/sh/kernel/early_printk.c - * - * Copyright (C) 1999, 2000 Niibe Yutaka - * Copyright (C) 2002 M. R. Brown - * Copyright (C) 2004 - 2007 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/console.h> -#include <linux/tty.h> -#include <linux/init.h> -#include <linux/io.h> -#include <linux/delay.h> - -#include <asm/sh_bios.h> - -/* - * Print a string through the BIOS - */ -static void sh_console_write(struct console *co, const char *s, - unsigned count) -{ - sh_bios_console_write(s, count); -} - -/* - * Setup initial baud/bits/parity. We do two things here: - * - construct a cflag setting for the first rs_open() - * - initialize the serial port - * Return non-zero if we didn't find a serial port. - */ -static int __init sh_console_setup(struct console *co, char *options) -{ - int cflag = CREAD | HUPCL | CLOCAL; - - /* - * Now construct a cflag setting. - * TODO: this is a totally bogus cflag, as we have - * no idea what serial settings the BIOS is using, or - * even if its using the serial port at all. - */ - cflag |= B115200 | CS8 | /*no parity*/0; - - co->cflag = cflag; - - return 0; -} - -static struct console bios_console = { - .name = "bios", - .write = sh_console_write, - .setup = sh_console_setup, - .flags = CON_PRINTBUFFER, - .index = -1, -}; - -static struct console *early_console; - -static int __init setup_early_printk(char *buf) -{ - int keep_early = 0; - - if (!buf) - return 0; - - if (strstr(buf, "keep")) - keep_early = 1; - - if (!strncmp(buf, "bios", 4)) - early_console = &bios_console; - - if (likely(early_console)) { - if (keep_early) - early_console->flags &= ~CON_BOOT; - else - early_console->flags |= CON_BOOT; - register_console(early_console); - } - - return 0; -} -early_param("earlyprintk", setup_early_printk); diff --git a/arch/sh/kernel/head_32.S b/arch/sh/kernel/head_32.S index 1151ecdffa71..8ee31a0b973e 100644 --- a/arch/sh/kernel/head_32.S +++ b/arch/sh/kernel/head_32.S @@ -3,6 +3,7 @@ * arch/sh/kernel/head.S * * Copyright (C) 1999, 2000 Niibe Yutaka & Kaz Kojima + * Copyright (C) 2010 Matt Fleming * * 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 @@ -13,6 +14,8 @@ #include <linux/init.h> #include <linux/linkage.h> #include <asm/thread_info.h> +#include <asm/mmu.h> +#include <cpu/mmu_context.h> #ifdef CONFIG_CPU_SH4A #define SYNCO() synco @@ -33,7 +36,7 @@ ENTRY(empty_zero_page) .long 1 /* LOADER_TYPE */ .long 0x00000000 /* INITRD_START */ .long 0x00000000 /* INITRD_SIZE */ -#if defined(CONFIG_32BIT) && defined(CONFIG_PMB_FIXED) +#ifdef CONFIG_32BIT .long 0x53453f00 + 32 /* "SE?" = 32 bit */ #else .long 0x53453f00 + 29 /* "SE?" = 29 bit */ @@ -82,6 +85,236 @@ ENTRY(_stext) ldc r0, r7_bank ! ... and initial thread_info #endif +#if defined(CONFIG_PMB) && !defined(CONFIG_PMB_LEGACY) + /* + * Reconfigure the initial PMB mappings setup by the hardware. + * + * When we boot in 32-bit MMU mode there are 2 PMB entries already + * setup for us. + * + * Entry VPN PPN V SZ C UB WT + * --------------------------------------------------------------- + * 0 0x80000000 0x00000000 1 512MB 1 0 1 + * 1 0xA0000000 0x00000000 1 512MB 0 0 0 + * + * But we reprogram them here because we want complete control over + * our address space and the initial mappings may not map PAGE_OFFSET + * to __MEMORY_START (or even map all of our RAM). + * + * Once we've setup cached and uncached mappings for all of RAM we + * clear the rest of the PMB entries. + * + * This clearing also deals with the fact that PMB entries can persist + * across reboots. The PMB could have been left in any state when the + * reboot occurred, so to be safe we clear all entries and start with + * with a clean slate. + */ + + mov.l .LMMUCR, r1 /* Flush the TLB */ + mov.l @r1, r0 + or #MMUCR_TI, r0 + mov.l r0, @r1 + + mov.l .LMEMORY_SIZE, r5 + mov r5, r7 + + mov #PMB_E_SHIFT, r0 + mov #0x1, r4 + shld r0, r4 + + mov.l .LFIRST_DATA_ENTRY, r0 + mov.l .LPMB_DATA, r1 + mov.l .LFIRST_ADDR_ENTRY, r2 + mov.l .LPMB_ADDR, r3 + + mov #0, r10 + + /* + * r0 = PMB_DATA data field + * r1 = PMB_DATA address field + * r2 = PMB_ADDR data field + * r3 = PMB_ADDR address field + * r4 = PMB_E_SHIFT + * r5 = remaining amount of RAM to map + * r6 = PMB mapping size we're trying to use + * r7 = cached_to_uncached + * r8 = scratch register + * r9 = scratch register + * r10 = number of PMB entries we've setup + */ +.L512: + mov #(512 >> 4), r6 + shll16 r6 + shll8 r6 + + cmp/hi r5, r6 + bt .L128 + + mov #(PMB_SZ_512M >> 2), r9 + shll2 r9 + + /* + * Cached mapping + */ + mov #PMB_C, r8 + or r0, r8 + or r9, r8 + mov.l r8, @r1 + mov.l r2, @r3 + + add r4, r1 /* Increment to the next PMB_DATA entry */ + add r4, r3 /* Increment to the next PMB_ADDR entry */ + + add #1, r10 /* Increment number of PMB entries */ + + /* + * Uncached mapping + */ + mov #(PMB_UB >> 8), r8 + shll8 r8 + + or r0, r8 + or r9, r8 + mov.l r8, @r1 + mov r2, r8 + add r7, r8 + mov.l r8, @r3 + + add r4, r1 /* Increment to the next PMB_DATA entry */ + add r4, r3 /* Increment to the next PMB_ADDR entry */ + + add #1, r10 /* Increment number of PMB entries */ + + sub r6, r5 + add r6, r0 + add r6, r2 + + bra .L512 + +.L128: + mov #(128 >> 4), r6 + shll16 r6 + shll8 r6 + + cmp/hi r5, r6 + bt .L64 + + mov #(PMB_SZ_128M >> 2), r9 + shll2 r9 + + /* + * Cached mapping + */ + mov #PMB_C, r8 + or r0, r8 + or r9, r8 + mov.l r8, @r1 + mov.l r2, @r3 + + add r4, r1 /* Increment to the next PMB_DATA entry */ + add r4, r3 /* Increment to the next PMB_ADDR entry */ + + add #1, r10 /* Increment number of PMB entries */ + + /* + * Uncached mapping + */ + mov #(PMB_UB >> 8), r8 + shll8 r8 + + or r0, r8 + or r9, r8 + mov.l r8, @r1 + mov r2, r8 + add r7, r8 + mov.l r8, @r3 + + add r4, r1 /* Increment to the next PMB_DATA entry */ + add r4, r3 /* Increment to the next PMB_ADDR entry */ + + add #1, r10 /* Increment number of PMB entries */ + + sub r6, r5 + add r6, r0 + add r6, r2 + + bra .L128 + +.L64: + mov #(64 >> 4), r6 + shll16 r6 + shll8 r6 + + cmp/hi r5, r6 + bt .Ldone + + mov #(PMB_SZ_64M >> 2), r9 + shll2 r9 + + /* + * Cached mapping + */ + mov #PMB_C, r8 + or r0, r8 + or r9, r8 + mov.l r8, @r1 + mov.l r2, @r3 + + add r4, r1 /* Increment to the next PMB_DATA entry */ + add r4, r3 /* Increment to the next PMB_ADDR entry */ + + add #1, r10 /* Increment number of PMB entries */ + + /* + * Uncached mapping + */ + mov #(PMB_UB >> 8), r8 + shll8 r8 + + or r0, r8 + or r9, r8 + mov.l r8, @r1 + mov r2, r8 + add r7, r8 + mov.l r8, @r3 + + add r4, r1 /* Increment to the next PMB_DATA entry */ + add r4, r3 /* Increment to the next PMB_ADDR entry */ + + add #1, r10 /* Increment number of PMB entries */ + + sub r6, r5 + add r6, r0 + add r6, r2 + + bra .L64 + +.Ldone: + /* Update cached_to_uncached */ + mov.l .Lcached_to_uncached, r0 + mov.l r7, @r0 + + /* + * Clear the remaining PMB entries. + * + * r3 = entry to begin clearing from + * r10 = number of entries we've setup so far + */ + mov #0, r1 + mov #PMB_ENTRY_MAX, r0 + +.Lagain: + mov.l r1, @r3 /* Clear PMB_ADDR entry */ + add #1, r10 /* Increment the loop counter */ + cmp/eq r0, r10 + bf/s .Lagain + add r4, r3 /* Increment to the next PMB_ADDR entry */ + + mov.l 6f, r0 + icbi @r0 + +#endif /* !CONFIG_PMB_LEGACY */ + #ifndef CONFIG_SH_NO_BSS_INIT /* * Don't clear BSS if running on slow platforms such as an RTL simulation, @@ -131,3 +364,13 @@ ENTRY(stack_start) 5: .long start_kernel 6: .long sh_cpu_init 7: .long init_thread_union + +#if defined(CONFIG_PMB) && !defined(CONFIG_PMB_LEGACY) +.LPMB_ADDR: .long PMB_ADDR +.LPMB_DATA: .long PMB_DATA +.LFIRST_ADDR_ENTRY: .long PAGE_OFFSET | PMB_V +.LFIRST_DATA_ENTRY: .long __MEMORY_START | PMB_V +.LMMUCR: .long MMUCR +.Lcached_to_uncached: .long cached_to_uncached +.LMEMORY_SIZE: .long __MEMORY_SIZE +#endif diff --git a/arch/sh/kernel/head_64.S b/arch/sh/kernel/head_64.S index 3ea765844c74..defd851abefa 100644 --- a/arch/sh/kernel/head_64.S +++ b/arch/sh/kernel/head_64.S @@ -220,7 +220,6 @@ clear_DTLB: add.l r22, r63, r22 /* Sign extend */ putcfg r21, 0, r22 /* Set MMUDR[0].PTEH */ -#ifdef CONFIG_EARLY_PRINTK /* * Setup a DTLB translation for SCIF phys. */ @@ -231,7 +230,6 @@ clear_DTLB: movi 0xfa03, r22 /* 0xfa030000, fixed SCIF virt */ shori 0x0003, r22 putcfg r21, 0, r22 /* PTEH last */ -#endif /* * Set cache behaviours. diff --git a/arch/sh/kernel/hw_breakpoint.c b/arch/sh/kernel/hw_breakpoint.c new file mode 100644 index 000000000000..e2f1753d275c --- /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/idle.c b/arch/sh/kernel/idle.c index 6b3d706deac1..0fd7b41f0a22 100644 --- a/arch/sh/kernel/idle.c +++ b/arch/sh/kernel/idle.c @@ -20,10 +20,9 @@ #include <asm/system.h> #include <asm/atomic.h> -static int hlt_counter; void (*pm_idle)(void) = NULL; -void (*pm_power_off)(void); -EXPORT_SYMBOL(pm_power_off); + +static int hlt_counter; static int __init nohlt_setup(char *__unused) { @@ -131,6 +130,15 @@ static void do_nothing(void *unused) { } +void stop_this_cpu(void *unused) +{ + local_irq_disable(); + cpu_clear(smp_processor_id(), cpu_online_map); + + for (;;) + cpu_sleep(); +} + /* * cpu_idle_wait - Used to ensure that all the CPUs discard old value of * pm_idle and update to new pm_idle value. Required while changing pm_idle diff --git a/arch/sh/kernel/kgdb.c b/arch/sh/kernel/kgdb.c index 3e532d0d4a5c..70c69659b846 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/machine_kexec.c b/arch/sh/kernel/machine_kexec.c index 76f280223ebd..7672141c841b 100644 --- a/arch/sh/kernel/machine_kexec.c +++ b/arch/sh/kernel/machine_kexec.c @@ -21,6 +21,8 @@ #include <asm/mmu_context.h> #include <asm/io.h> #include <asm/cacheflush.h> +#include <asm/sh_bios.h> +#include <asm/reboot.h> typedef void (*relocate_new_kernel_t)(unsigned long indirection_page, unsigned long reboot_code_buffer, @@ -28,15 +30,11 @@ typedef void (*relocate_new_kernel_t)(unsigned long indirection_page, extern const unsigned char relocate_new_kernel[]; extern const unsigned int relocate_new_kernel_size; -extern void *gdb_vbr_vector; extern void *vbr_base; -void machine_shutdown(void) -{ -} - -void machine_crash_shutdown(struct pt_regs *regs) +void native_machine_crash_shutdown(struct pt_regs *regs) { + /* Nothing to do for UP, but definitely broken for SMP.. */ } /* @@ -117,11 +115,7 @@ void machine_kexec(struct kimage *image) kexec_info(image); flush_cache_all(); -#if defined(CONFIG_SH_STANDARD_BIOS) - asm volatile("ldc %0, vbr" : - : "r" (((unsigned long) gdb_vbr_vector) - 0x100) - : "memory"); -#endif + sh_bios_vbr_reload(); /* now call it */ rnk = (relocate_new_kernel_t) reboot_code_buffer; diff --git a/arch/sh/kernel/process.c b/arch/sh/kernel/process.c new file mode 100644 index 000000000000..81add9b9ea6e --- /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 d8af889366a4..b6f43f0ea743 100644 --- a/arch/sh/kernel/process_32.c +++ b/arch/sh/kernel/process_32.c @@ -16,65 +16,15 @@ #include <linux/module.h> #include <linux/mm.h> #include <linux/elfcore.h> -#include <linux/pm.h> #include <linux/kallsyms.h> -#include <linux/kexec.h> -#include <linux/kdebug.h> -#include <linux/tick.h> -#include <linux/reboot.h> #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) -{ - sh_wdt_write_cnt(0xFF); - sh_wdt_write_csr(0xC2); -} - -void machine_restart(char * __unused) -{ - local_irq_disable(); - - /* Use watchdog timer to trigger reset */ - watchdog_trigger_immediate(); - - while (1) - cpu_sleep(); -} -#else -void machine_restart(char * __unused) -{ - /* SR.BL=1 and invoke address error to let CPU reset (manual reset) */ - asm volatile("ldc %0, sr\n\t" - "mov.l @%1, %0" : : "r" (0x10000000), "r" (0x80000001)); -} -#endif - -void machine_halt(void) -{ - local_irq_disable(); - - while (1) - cpu_sleep(); -} - -void machine_power_off(void) -{ - if (pm_power_off) - pm_power_off(); -} void show_regs(struct pt_regs * regs) { @@ -147,21 +97,34 @@ 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 +172,10 @@ 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 +206,11 @@ 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 +224,7 @@ __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 +236,13 @@ __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 +335,3 @@ 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/process_64.c b/arch/sh/kernel/process_64.c index 31f80c61b031..6ae5987230f8 100644 --- a/arch/sh/kernel/process_64.c +++ b/arch/sh/kernel/process_64.c @@ -32,30 +32,7 @@ struct task_struct *last_task_used_math = NULL; -void machine_restart(char * __unused) -{ - extern void phys_stext(void); - - phys_stext(); -} - -void machine_halt(void) -{ - for (;;); -} - -void machine_power_off(void) -{ - __asm__ __volatile__ ( - "sleep\n\t" - "synci\n\t" - "nop;nop;nop;nop\n\t" - ); - - panic("Unexpected wakeup!\n"); -} - -void show_regs(struct pt_regs * regs) +void show_regs(struct pt_regs *regs) { unsigned long long ah, al, bh, bl, ch, cl; @@ -410,7 +387,7 @@ int dump_fpu(struct pt_regs *regs, elf_fpregset_t *fpu) regs->sr |= SR_FD; } - memcpy(fpu, &tsk->thread.fpu.hard, sizeof(*fpu)); + memcpy(fpu, &tsk->thread.xstate->hardfpu, sizeof(*fpu)); } return fpvalid; diff --git a/arch/sh/kernel/ptrace_32.c b/arch/sh/kernel/ptrace_32.c index 9be35f348093..c625cdab76dd 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/ptrace_64.c b/arch/sh/kernel/ptrace_64.c index 873ebdc4f98e..2f6140e95ec5 100644 --- a/arch/sh/kernel/ptrace_64.c +++ b/arch/sh/kernel/ptrace_64.c @@ -88,7 +88,7 @@ get_fpu_long(struct task_struct *task, unsigned long addr) regs->sr |= SR_FD; } - tmp = ((long *)&task->thread.fpu)[addr / sizeof(unsigned long)]; + tmp = ((long *)task->thread.xstate)[addr / sizeof(unsigned long)]; return tmp; } @@ -114,8 +114,7 @@ put_fpu_long(struct task_struct *task, unsigned long addr, unsigned long data) regs = (struct pt_regs*)((unsigned char *)task + THREAD_SIZE) - 1; if (!tsk_used_math(task)) { - fpinit(&task->thread.fpu.hard); - set_stopped_child_used_math(task); + init_fpu(task); } else if (last_task_used_math == task) { enable_fpu(); save_fpu(task); @@ -124,7 +123,7 @@ put_fpu_long(struct task_struct *task, unsigned long addr, unsigned long data) regs->sr |= SR_FD; } - ((long *)&task->thread.fpu)[addr / sizeof(unsigned long)] = data; + ((long *)task->thread.xstate)[addr / sizeof(unsigned long)] = data; return 0; } @@ -222,7 +221,7 @@ int fpregs_get(struct task_struct *target, return ret; return user_regset_copyout(&pos, &count, &kbuf, &ubuf, - &target->thread.fpu.hard, 0, -1); + &target->thread.xstate->hardfpu, 0, -1); } static int fpregs_set(struct task_struct *target, @@ -239,7 +238,7 @@ static int fpregs_set(struct task_struct *target, set_stopped_child_used_math(target); return user_regset_copyin(&pos, &count, &kbuf, &ubuf, - &target->thread.fpu.hard, 0, -1); + &target->thread.xstate->hardfpu, 0, -1); } static int fpregs_active(struct task_struct *target, @@ -479,9 +478,10 @@ asmlinkage void do_single_step(unsigned long long vec, struct pt_regs *regs) } /* Called with interrupts disabled */ -asmlinkage void do_software_break_point(unsigned long long vec, - struct pt_regs *regs) +BUILD_TRAP_HANDLER(breakpoint) { + TRAP_HANDLER_DECL; + /* We need to forward step the PC, to counteract the backstep done in signal.c. */ local_irq_enable(); diff --git a/arch/sh/kernel/reboot.c b/arch/sh/kernel/reboot.c new file mode 100644 index 000000000000..b1fca66bb92e --- /dev/null +++ b/arch/sh/kernel/reboot.c @@ -0,0 +1,98 @@ +#include <linux/pm.h> +#include <linux/kexec.h> +#include <linux/kernel.h> +#include <linux/reboot.h> +#include <linux/module.h> +#ifdef CONFIG_SUPERH32 +#include <asm/watchdog.h> +#endif +#include <asm/addrspace.h> +#include <asm/reboot.h> +#include <asm/system.h> + +void (*pm_power_off)(void); +EXPORT_SYMBOL(pm_power_off); + +#ifdef CONFIG_SUPERH32 +static void watchdog_trigger_immediate(void) +{ + sh_wdt_write_cnt(0xFF); + sh_wdt_write_csr(0xC2); +} +#endif + +static void native_machine_restart(char * __unused) +{ + local_irq_disable(); + + /* Address error with SR.BL=1 first. */ + trigger_address_error(); + +#ifdef CONFIG_SUPERH32 + /* If that fails or is unsupported, go for the watchdog next. */ + watchdog_trigger_immediate(); +#endif + + /* + * Give up and sleep. + */ + while (1) + cpu_sleep(); +} + +static void native_machine_shutdown(void) +{ + smp_send_stop(); +} + +static void native_machine_power_off(void) +{ + if (pm_power_off) + pm_power_off(); +} + +static void native_machine_halt(void) +{ + /* stop other cpus */ + machine_shutdown(); + + /* stop this cpu */ + stop_this_cpu(NULL); +} + +struct machine_ops machine_ops = { + .power_off = native_machine_power_off, + .shutdown = native_machine_shutdown, + .restart = native_machine_restart, + .halt = native_machine_halt, +#ifdef CONFIG_KEXEC + .crash_shutdown = native_machine_crash_shutdown, +#endif +}; + +void machine_power_off(void) +{ + machine_ops.power_off(); +} + +void machine_shutdown(void) +{ + machine_ops.shutdown(); +} + +void machine_restart(char *cmd) +{ + machine_ops.restart(cmd); +} + +void machine_halt(void) +{ + machine_ops.halt(); +} + +#ifdef CONFIG_KEXEC +void machine_crash_shutdown(struct pt_regs *regs) +{ + machine_ops.crash_shutdown(regs); +} +#endif diff --git a/arch/sh/kernel/setup.c b/arch/sh/kernel/setup.c index 8b0e69792cf4..e187750dd319 100644 --- a/arch/sh/kernel/setup.c +++ b/arch/sh/kernel/setup.c @@ -449,17 +449,15 @@ void __init setup_arch(char **cmdline_p) #ifdef CONFIG_DUMMY_CONSOLE conswitchp = &dummy_con; #endif + paging_init(); + pmb_init(); + + ioremap_fixed_init(); /* Perform the machine specific initialisation */ if (likely(sh_mv.mv_setup)) sh_mv.mv_setup(cmdline_p); - paging_init(); - -#ifdef CONFIG_PMB_ENABLE - pmb_init(); -#endif - #ifdef CONFIG_SMP plat_smp_setup(); #endif diff --git a/arch/sh/kernel/sh_bios.c b/arch/sh/kernel/sh_bios.c index c852f7805728..47475cca068a 100644 --- a/arch/sh/kernel/sh_bios.c +++ b/arch/sh/kernel/sh_bios.c @@ -1,19 +1,30 @@ /* - * linux/arch/sh/kernel/sh_bios.c * C interface for trapping into the standard LinuxSH BIOS. * * Copyright (C) 2000 Greg Banks, Mitch Davis + * Copyright (C) 1999, 2000 Niibe Yutaka + * Copyright (C) 2002 M. R. Brown + * Copyright (C) 2004 - 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/module.h> +#include <linux/console.h> +#include <linux/tty.h> +#include <linux/init.h> +#include <linux/io.h> +#include <linux/delay.h> #include <asm/sh_bios.h> #define BIOS_CALL_CONSOLE_WRITE 0 #define BIOS_CALL_ETH_NODE_ADDR 10 #define BIOS_CALL_SHUTDOWN 11 -#define BIOS_CALL_CHAR_OUT 0x1f /* TODO: hack */ #define BIOS_CALL_GDB_DETACH 0xff +void *gdb_vbr_vector = NULL; + static inline long sh_bios_call(long func, long arg0, long arg1, long arg2, long arg3) { @@ -23,6 +34,9 @@ static inline long sh_bios_call(long func, long arg0, long arg1, long arg2, register long r6 __asm__("r6") = arg2; register long r7 __asm__("r7") = arg3; + if (!gdb_vbr_vector) + return -ENOSYS; + __asm__ __volatile__("trapa #0x3f":"=z"(r0) :"0"(r0), "r"(r4), "r"(r5), "r"(r6), "r"(r7) :"memory"); @@ -34,11 +48,6 @@ void sh_bios_console_write(const char *buf, unsigned int len) sh_bios_call(BIOS_CALL_CONSOLE_WRITE, (long)buf, (long)len, 0, 0); } -void sh_bios_char_out(char ch) -{ - sh_bios_call(BIOS_CALL_CHAR_OUT, ch, 0, 0, 0); -} - void sh_bios_gdb_detach(void) { sh_bios_call(BIOS_CALL_GDB_DETACH, 0, 0, 0, 0); @@ -55,3 +64,109 @@ void sh_bios_shutdown(unsigned int how) { sh_bios_call(BIOS_CALL_SHUTDOWN, how, 0, 0, 0); } + +/* + * Read the old value of the VBR register to initialise the vector + * through which debug and BIOS traps are delegated by the Linux trap + * handler. + */ +void sh_bios_vbr_init(void) +{ + unsigned long vbr; + + if (unlikely(gdb_vbr_vector)) + return; + + __asm__ __volatile__ ("stc vbr, %0" : "=r" (vbr)); + + if (vbr) { + gdb_vbr_vector = (void *)(vbr + 0x100); + printk(KERN_NOTICE "Setting GDB trap vector to %p\n", + gdb_vbr_vector); + } else + printk(KERN_NOTICE "SH-BIOS not detected\n"); +} + +/** + * sh_bios_vbr_reload - Re-load the system VBR from the BIOS vector. + * + * This can be used by save/restore code to reinitialize the system VBR + * from the fixed BIOS VBR. A no-op if no BIOS VBR is known. + */ +void sh_bios_vbr_reload(void) +{ + if (gdb_vbr_vector) + __asm__ __volatile__ ( + "ldc %0, vbr" + : + : "r" (((unsigned long) gdb_vbr_vector) - 0x100) + : "memory" + ); +} + +/* + * Print a string through the BIOS + */ +static void sh_console_write(struct console *co, const char *s, + unsigned count) +{ + sh_bios_console_write(s, count); +} + +/* + * Setup initial baud/bits/parity. We do two things here: + * - construct a cflag setting for the first rs_open() + * - initialize the serial port + * Return non-zero if we didn't find a serial port. + */ +static int __init sh_console_setup(struct console *co, char *options) +{ + int cflag = CREAD | HUPCL | CLOCAL; + + /* + * Now construct a cflag setting. + * TODO: this is a totally bogus cflag, as we have + * no idea what serial settings the BIOS is using, or + * even if its using the serial port at all. + */ + cflag |= B115200 | CS8 | /*no parity*/0; + + co->cflag = cflag; + + return 0; +} + +static struct console bios_console = { + .name = "bios", + .write = sh_console_write, + .setup = sh_console_setup, + .flags = CON_PRINTBUFFER, + .index = -1, +}; + +static struct console *early_console; + +static int __init setup_early_printk(char *buf) +{ + int keep_early = 0; + + if (!buf) + return 0; + + if (strstr(buf, "keep")) + keep_early = 1; + + if (!strncmp(buf, "bios", 4)) + early_console = &bios_console; + + if (likely(early_console)) { + if (keep_early) + early_console->flags &= ~CON_BOOT; + else + early_console->flags |= CON_BOOT; + register_console(early_console); + } + + return 0; +} +early_param("earlyprintk", setup_early_printk); diff --git a/arch/sh/kernel/signal_32.c b/arch/sh/kernel/signal_32.c index 12815ce01ecd..6a7cce79eb4e 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/signal_64.c b/arch/sh/kernel/signal_64.c index ce76dbdef294..4733bfc59d39 100644 --- a/arch/sh/kernel/signal_64.c +++ b/arch/sh/kernel/signal_64.c @@ -295,7 +295,7 @@ restore_sigcontext_fpu(struct pt_regs *regs, struct sigcontext __user *sc) regs->sr |= SR_FD; } - err |= __copy_from_user(¤t->thread.fpu.hard, &sc->sc_fpregs[0], + err |= __copy_from_user(¤t->thread.xstate->hardfpu, &sc->sc_fpregs[0], (sizeof(long long) * 32) + (sizeof(int) * 1)); return err; @@ -320,7 +320,7 @@ setup_sigcontext_fpu(struct pt_regs *regs, struct sigcontext __user *sc) regs->sr |= SR_FD; } - err |= __copy_to_user(&sc->sc_fpregs[0], ¤t->thread.fpu.hard, + err |= __copy_to_user(&sc->sc_fpregs[0], ¤t->thread.xstate->hardfpu, (sizeof(long long) * 32) + (sizeof(int) * 1)); clear_used_math(); diff --git a/arch/sh/kernel/smp.c b/arch/sh/kernel/smp.c index 983e0792d5f3..e124cf7008df 100644 --- a/arch/sh/kernel/smp.c +++ b/arch/sh/kernel/smp.c @@ -161,15 +161,6 @@ void smp_send_reschedule(int cpu) plat_send_ipi(cpu, SMP_MSG_RESCHEDULE); } -static void stop_this_cpu(void *unused) -{ - cpu_clear(smp_processor_id(), cpu_online_map); - local_irq_disable(); - - for (;;) - cpu_relax(); -} - void smp_send_stop(void) { smp_call_function(stop_this_cpu, 0, 0); diff --git a/arch/sh/kernel/traps_32.c b/arch/sh/kernel/traps_32.c index 86639beac3a2..9c090cb68878 100644 --- a/arch/sh/kernel/traps_32.c +++ b/arch/sh/kernel/traps_32.c @@ -24,13 +24,13 @@ #include <linux/kdebug.h> #include <linux/kexec.h> #include <linux/limits.h> -#include <linux/proc_fs.h> -#include <linux/seq_file.h> #include <linux/sysfs.h> +#include <linux/uaccess.h> #include <asm/system.h> -#include <asm/uaccess.h> +#include <asm/alignment.h> #include <asm/fpu.h> #include <asm/kprobes.h> +#include <asm/sh_bios.h> #ifdef CONFIG_CPU_SH2 # define TRAP_RESERVED_INST 4 @@ -47,73 +47,6 @@ #define TRAP_ILLEGAL_SLOT_INST 13 #endif -static unsigned long se_user; -static unsigned long se_sys; -static unsigned long se_half; -static unsigned long se_word; -static unsigned long se_dword; -static unsigned long se_multi; -/* bitfield: 1: warn 2: fixup 4: signal -> combinations 2|4 && 1|2|4 are not - valid! */ -static int se_usermode = 3; -/* 0: no warning 1: print a warning message, disabled by default */ -static int se_kernmode_warn; - -#ifdef CONFIG_PROC_FS -static const char *se_usermode_action[] = { - "ignored", - "warn", - "fixup", - "fixup+warn", - "signal", - "signal+warn" -}; - -static int alignment_proc_show(struct seq_file *m, void *v) -{ - seq_printf(m, "User:\t\t%lu\n", se_user); - seq_printf(m, "System:\t\t%lu\n", se_sys); - seq_printf(m, "Half:\t\t%lu\n", se_half); - seq_printf(m, "Word:\t\t%lu\n", se_word); - seq_printf(m, "DWord:\t\t%lu\n", se_dword); - seq_printf(m, "Multi:\t\t%lu\n", se_multi); - seq_printf(m, "User faults:\t%i (%s)\n", se_usermode, - se_usermode_action[se_usermode]); - seq_printf(m, "Kernel faults:\t%i (fixup%s)\n", se_kernmode_warn, - se_kernmode_warn ? "+warn" : ""); - return 0; -} - -static int alignment_proc_open(struct inode *inode, struct file *file) -{ - return single_open(file, alignment_proc_show, NULL); -} - -static ssize_t alignment_proc_write(struct file *file, - const char __user *buffer, size_t count, loff_t *pos) -{ - int *data = PDE(file->f_path.dentry->d_inode)->data; - char mode; - - if (count > 0) { - if (get_user(mode, buffer)) - return -EFAULT; - if (mode >= '0' && mode <= '5') - *data = mode - '0'; - } - return count; -} - -static const struct file_operations alignment_proc_fops = { - .owner = THIS_MODULE, - .open = alignment_proc_open, - .read = seq_read, - .llseek = seq_lseek, - .release = single_release, - .write = alignment_proc_write, -}; -#endif - static void dump_mem(const char *str, unsigned long bottom, unsigned long top) { unsigned long p; @@ -265,10 +198,10 @@ static int handle_unaligned_ins(insn_size_t instruction, struct pt_regs *regs, count = 1<<(instruction&3); switch (count) { - case 1: se_half += 1; break; - case 2: se_word += 1; break; - case 4: se_dword += 1; break; - case 8: se_multi += 1; break; /* ??? */ + case 1: inc_unaligned_byte_access(); break; + case 2: inc_unaligned_word_access(); break; + case 4: inc_unaligned_dword_access(); break; + case 8: inc_unaligned_multi_access(); break; } ret = -EFAULT; @@ -452,18 +385,8 @@ int handle_unaligned_access(insn_size_t instruction, struct pt_regs *regs, rm = regs->regs[index]; /* shout about fixups */ - if (!expected) { - if (user_mode(regs) && (se_usermode & 1) && printk_ratelimit()) - pr_notice("Fixing up unaligned userspace access " - "in \"%s\" pid=%d pc=0x%p ins=0x%04hx\n", - current->comm, task_pid_nr(current), - (void *)regs->pc, instruction); - else if (se_kernmode_warn && printk_ratelimit()) - pr_notice("Fixing up unaligned kernel access " - "in \"%s\" pid=%d pc=0x%p ins=0x%04hx\n", - current->comm, task_pid_nr(current), - (void *)regs->pc, instruction); - } + if (!expected) + unaligned_fixups_notify(current, instruction, regs); ret = -EFAULT; switch (instruction&0xF000) { @@ -616,10 +539,10 @@ asmlinkage void do_address_error(struct pt_regs *regs, if (user_mode(regs)) { int si_code = BUS_ADRERR; + unsigned int user_action; local_irq_enable(); - - se_user += 1; + inc_unaligned_user_access(); set_fs(USER_DS); if (copy_from_user(&instruction, (insn_size_t *)(regs->pc & ~1), @@ -630,16 +553,12 @@ asmlinkage void do_address_error(struct pt_regs *regs, set_fs(oldfs); /* shout about userspace fixups */ - if (se_usermode & 1) - printk(KERN_NOTICE "Unaligned userspace access " - "in \"%s\" pid=%d pc=0x%p ins=0x%04hx\n", - current->comm, current->pid, (void *)regs->pc, - instruction); + unaligned_fixups_notify(current, instruction, regs); - if (se_usermode & 2) + user_action = unaligned_user_action(); + if (user_action & UM_FIXUP) goto fixup; - - if (se_usermode & 4) + if (user_action & UM_SIGNAL) goto uspace_segv; else { /* ignore */ @@ -659,7 +578,7 @@ fixup: &user_mem_access, 0); set_fs(oldfs); - if (tmp==0) + if (tmp == 0) return; /* sorted */ uspace_segv: printk(KERN_NOTICE "Sending SIGBUS to \"%s\" due to unaligned " @@ -672,7 +591,7 @@ uspace_segv: info.si_addr = (void __user *)address; force_sig_info(SIGBUS, &info, current); } else { - se_sys += 1; + inc_unaligned_kernel_access(); if (regs->pc & 1) die("unaligned program counter", regs, error_code); @@ -687,11 +606,7 @@ uspace_segv: die("insn faulting in do_address_error", regs, 0); } - if (se_kernmode_warn) - printk(KERN_NOTICE "Unaligned kernel access " - "on behalf of \"%s\" pid=%d pc=0x%p ins=0x%04hx\n", - current->comm, current->pid, (void *)regs->pc, - instruction); + unaligned_fixups_notify(current, instruction, regs); handle_unaligned_access(instruction, regs, &user_mem_access, 0); @@ -876,35 +791,10 @@ asmlinkage void do_exception_error(unsigned long r4, unsigned long r5, die_if_kernel("exception", regs, ex); } -#if defined(CONFIG_SH_STANDARD_BIOS) -void *gdb_vbr_vector; - -static inline void __init gdb_vbr_init(void) -{ - register unsigned long vbr; - - /* - * Read the old value of the VBR register to initialise - * the vector through which debug and BIOS traps are - * delegated by the Linux trap handler. - */ - asm volatile("stc vbr, %0" : "=r" (vbr)); - - gdb_vbr_vector = (void *)(vbr + 0x100); - printk("Setting GDB trap vector to 0x%08lx\n", - (unsigned long)gdb_vbr_vector); -} -#endif - void __cpuinit per_cpu_trap_init(void) { extern void *vbr_base; -#ifdef CONFIG_SH_STANDARD_BIOS - if (raw_smp_processor_id() == 0) - gdb_vbr_init(); -#endif - /* NOTE: The VBR value should be at P1 (or P2, virtural "fixed" address space). It's definitely should not in physical address. */ @@ -956,9 +846,12 @@ 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 */ + sh_bios_vbr_init(); + /* Setup VBR for boot cpu */ per_cpu_trap_init(); } @@ -985,34 +878,3 @@ void dump_stack(void) show_stack(NULL, NULL); } EXPORT_SYMBOL(dump_stack); - -#ifdef CONFIG_PROC_FS -/* - * This needs to be done after sysctl_init, otherwise sys/ will be - * overwritten. Actually, this shouldn't be in sys/ at all since - * it isn't a sysctl, and it doesn't contain sysctl information. - * We now locate it in /proc/cpu/alignment instead. - */ -static int __init alignment_init(void) -{ - struct proc_dir_entry *dir, *res; - - dir = proc_mkdir("cpu", NULL); - if (!dir) - return -ENOMEM; - - res = proc_create_data("alignment", S_IWUSR | S_IRUGO, dir, - &alignment_proc_fops, &se_usermode); - if (!res) - return -ENOMEM; - - res = proc_create_data("kernel_alignment", S_IWUSR | S_IRUGO, dir, - &alignment_proc_fops, &se_kernmode_warn); - if (!res) - return -ENOMEM; - - return 0; -} - -fs_initcall(alignment_init); -#endif diff --git a/arch/sh/kernel/traps_64.c b/arch/sh/kernel/traps_64.c index d86f5315a0c1..e3f92eb05ffd 100644 --- a/arch/sh/kernel/traps_64.c +++ b/arch/sh/kernel/traps_64.c @@ -611,19 +611,19 @@ static int misaligned_fpu_load(struct pt_regs *regs, switch (width_shift) { case 2: - current->thread.fpu.hard.fp_regs[destreg] = buflo; + current->thread.xstate->hardfpu.fp_regs[destreg] = buflo; break; case 3: if (do_paired_load) { - current->thread.fpu.hard.fp_regs[destreg] = buflo; - current->thread.fpu.hard.fp_regs[destreg+1] = bufhi; + current->thread.xstate->hardfpu.fp_regs[destreg] = buflo; + current->thread.xstate->hardfpu.fp_regs[destreg+1] = bufhi; } else { #if defined(CONFIG_CPU_LITTLE_ENDIAN) - current->thread.fpu.hard.fp_regs[destreg] = bufhi; - current->thread.fpu.hard.fp_regs[destreg+1] = buflo; + current->thread.xstate->hardfpu.fp_regs[destreg] = bufhi; + current->thread.xstate->hardfpu.fp_regs[destreg+1] = buflo; #else - current->thread.fpu.hard.fp_regs[destreg] = buflo; - current->thread.fpu.hard.fp_regs[destreg+1] = bufhi; + current->thread.xstate->hardfpu.fp_regs[destreg] = buflo; + current->thread.xstate->hardfpu.fp_regs[destreg+1] = bufhi; #endif } break; @@ -681,19 +681,19 @@ static int misaligned_fpu_store(struct pt_regs *regs, switch (width_shift) { case 2: - buflo = current->thread.fpu.hard.fp_regs[srcreg]; + buflo = current->thread.xstate->hardfpu.fp_regs[srcreg]; break; case 3: if (do_paired_load) { - buflo = current->thread.fpu.hard.fp_regs[srcreg]; - bufhi = current->thread.fpu.hard.fp_regs[srcreg+1]; + buflo = current->thread.xstate->hardfpu.fp_regs[srcreg]; + bufhi = current->thread.xstate->hardfpu.fp_regs[srcreg+1]; } else { #if defined(CONFIG_CPU_LITTLE_ENDIAN) - bufhi = current->thread.fpu.hard.fp_regs[srcreg]; - buflo = current->thread.fpu.hard.fp_regs[srcreg+1]; + bufhi = current->thread.xstate->hardfpu.fp_regs[srcreg]; + buflo = current->thread.xstate->hardfpu.fp_regs[srcreg+1]; #else - buflo = current->thread.fpu.hard.fp_regs[srcreg]; - bufhi = current->thread.fpu.hard.fp_regs[srcreg+1]; + buflo = current->thread.xstate->hardfpu.fp_regs[srcreg]; + bufhi = current->thread.xstate->hardfpu.fp_regs[srcreg+1]; #endif } break; diff --git a/arch/sh/kernel/vmlinux.lds.S b/arch/sh/kernel/vmlinux.lds.S index a1e4ec24f1f5..9e5a5878eeae 100644 --- a/arch/sh/kernel/vmlinux.lds.S +++ b/arch/sh/kernel/vmlinux.lds.S @@ -14,17 +14,16 @@ OUTPUT_ARCH(sh) #include <asm/cache.h> #include <asm/vmlinux.lds.h> +#if defined(CONFIG_32BIT) && !defined(CONFIG_PMB_LEGACY) +#define MEMORY_OFFSET 0 +#else +#define MEMORY_OFFSET (CONFIG_MEMORY_START & 0x1fffffff) +#endif + ENTRY(_start) SECTIONS { -#ifdef CONFIG_PMB_FIXED - . = CONFIG_PAGE_OFFSET + (CONFIG_MEMORY_START & 0x1fffffff) + - CONFIG_ZERO_PAGE_OFFSET; -#elif defined(CONFIG_32BIT) - . = CONFIG_PAGE_OFFSET + CONFIG_ZERO_PAGE_OFFSET; -#else - . = CONFIG_PAGE_OFFSET + CONFIG_MEMORY_START + CONFIG_ZERO_PAGE_OFFSET; -#endif + . = CONFIG_PAGE_OFFSET + MEMORY_OFFSET + CONFIG_ZERO_PAGE_OFFSET; _text = .; /* Text and read-only data */ |