1 /* 2 * Copyright (C) 1995 Linus Torvalds 3 * 4 * Pentium III FXSR, SSE support 5 * Gareth Hughes <gareth@valinux.com>, May 2000 6 */ 7 8 /* 9 * This file handles the architecture-dependent parts of process handling.. 10 */ 11 12 #include <linux/stackprotector.h> 13 #include <linux/cpu.h> 14 #include <linux/errno.h> 15 #include <linux/sched.h> 16 #include <linux/fs.h> 17 #include <linux/kernel.h> 18 #include <linux/mm.h> 19 #include <linux/elfcore.h> 20 #include <linux/smp.h> 21 #include <linux/stddef.h> 22 #include <linux/slab.h> 23 #include <linux/vmalloc.h> 24 #include <linux/user.h> 25 #include <linux/interrupt.h> 26 #include <linux/delay.h> 27 #include <linux/reboot.h> 28 #include <linux/init.h> 29 #include <linux/mc146818rtc.h> 30 #include <linux/module.h> 31 #include <linux/kallsyms.h> 32 #include <linux/ptrace.h> 33 #include <linux/personality.h> 34 #include <linux/tick.h> 35 #include <linux/percpu.h> 36 #include <linux/prctl.h> 37 #include <linux/ftrace.h> 38 #include <linux/uaccess.h> 39 #include <linux/io.h> 40 #include <linux/kdebug.h> 41 #include <linux/cpuidle.h> 42 43 #include <asm/pgtable.h> 44 #include <asm/system.h> 45 #include <asm/ldt.h> 46 #include <asm/processor.h> 47 #include <asm/i387.h> 48 #include <asm/desc.h> 49 #ifdef CONFIG_MATH_EMULATION 50 #include <asm/math_emu.h> 51 #endif 52 53 #include <linux/err.h> 54 55 #include <asm/tlbflush.h> 56 #include <asm/cpu.h> 57 #include <asm/idle.h> 58 #include <asm/syscalls.h> 59 #include <asm/debugreg.h> 60 #include <asm/nmi.h> 61 62 asmlinkage void ret_from_fork(void) __asm__("ret_from_fork"); 63 64 /* 65 * Return saved PC of a blocked thread. 66 */ 67 unsigned long thread_saved_pc(struct task_struct *tsk) 68 { 69 return ((unsigned long *)tsk->thread.sp)[3]; 70 } 71 72 #ifndef CONFIG_SMP 73 static inline void play_dead(void) 74 { 75 BUG(); 76 } 77 #endif 78 79 /* 80 * The idle thread. There's no useful work to be 81 * done, so just try to conserve power and have a 82 * low exit latency (ie sit in a loop waiting for 83 * somebody to say that they'd like to reschedule) 84 */ 85 void cpu_idle(void) 86 { 87 int cpu = smp_processor_id(); 88 89 /* 90 * If we're the non-boot CPU, nothing set the stack canary up 91 * for us. CPU0 already has it initialized but no harm in 92 * doing it again. This is a good place for updating it, as 93 * we wont ever return from this function (so the invalid 94 * canaries already on the stack wont ever trigger). 95 */ 96 boot_init_stack_canary(); 97 98 current_thread_info()->status |= TS_POLLING; 99 100 /* endless idle loop with no priority at all */ 101 while (1) { 102 tick_nohz_idle_enter(); 103 rcu_idle_enter(); 104 while (!need_resched()) { 105 106 check_pgt_cache(); 107 rmb(); 108 109 if (cpu_is_offline(cpu)) 110 play_dead(); 111 112 local_touch_nmi(); 113 local_irq_disable(); 114 /* Don't trace irqs off for idle */ 115 stop_critical_timings(); 116 if (cpuidle_idle_call()) 117 pm_idle(); 118 start_critical_timings(); 119 } 120 rcu_idle_exit(); 121 tick_nohz_idle_exit(); 122 preempt_enable_no_resched(); 123 schedule(); 124 preempt_disable(); 125 } 126 } 127 128 void __show_regs(struct pt_regs *regs, int all) 129 { 130 unsigned long cr0 = 0L, cr2 = 0L, cr3 = 0L, cr4 = 0L; 131 unsigned long d0, d1, d2, d3, d6, d7; 132 unsigned long sp; 133 unsigned short ss, gs; 134 135 if (user_mode_vm(regs)) { 136 sp = regs->sp; 137 ss = regs->ss & 0xffff; 138 gs = get_user_gs(regs); 139 } else { 140 sp = kernel_stack_pointer(regs); 141 savesegment(ss, ss); 142 savesegment(gs, gs); 143 } 144 145 show_regs_common(); 146 147 printk(KERN_DEFAULT "EIP: %04x:[<%08lx>] EFLAGS: %08lx CPU: %d\n", 148 (u16)regs->cs, regs->ip, regs->flags, 149 smp_processor_id()); 150 print_symbol("EIP is at %s\n", regs->ip); 151 152 printk(KERN_DEFAULT "EAX: %08lx EBX: %08lx ECX: %08lx EDX: %08lx\n", 153 regs->ax, regs->bx, regs->cx, regs->dx); 154 printk(KERN_DEFAULT "ESI: %08lx EDI: %08lx EBP: %08lx ESP: %08lx\n", 155 regs->si, regs->di, regs->bp, sp); 156 printk(KERN_DEFAULT " DS: %04x ES: %04x FS: %04x GS: %04x SS: %04x\n", 157 (u16)regs->ds, (u16)regs->es, (u16)regs->fs, gs, ss); 158 159 if (!all) 160 return; 161 162 cr0 = read_cr0(); 163 cr2 = read_cr2(); 164 cr3 = read_cr3(); 165 cr4 = read_cr4_safe(); 166 printk(KERN_DEFAULT "CR0: %08lx CR2: %08lx CR3: %08lx CR4: %08lx\n", 167 cr0, cr2, cr3, cr4); 168 169 get_debugreg(d0, 0); 170 get_debugreg(d1, 1); 171 get_debugreg(d2, 2); 172 get_debugreg(d3, 3); 173 printk(KERN_DEFAULT "DR0: %08lx DR1: %08lx DR2: %08lx DR3: %08lx\n", 174 d0, d1, d2, d3); 175 176 get_debugreg(d6, 6); 177 get_debugreg(d7, 7); 178 printk(KERN_DEFAULT "DR6: %08lx DR7: %08lx\n", 179 d6, d7); 180 } 181 182 void release_thread(struct task_struct *dead_task) 183 { 184 BUG_ON(dead_task->mm); 185 release_vm86_irqs(dead_task); 186 } 187 188 /* 189 * This gets called before we allocate a new thread and copy 190 * the current task into it. 191 */ 192 void prepare_to_copy(struct task_struct *tsk) 193 { 194 unlazy_fpu(tsk); 195 } 196 197 int copy_thread(unsigned long clone_flags, unsigned long sp, 198 unsigned long unused, 199 struct task_struct *p, struct pt_regs *regs) 200 { 201 struct pt_regs *childregs; 202 struct task_struct *tsk; 203 int err; 204 205 childregs = task_pt_regs(p); 206 *childregs = *regs; 207 childregs->ax = 0; 208 childregs->sp = sp; 209 210 p->thread.sp = (unsigned long) childregs; 211 p->thread.sp0 = (unsigned long) (childregs+1); 212 213 p->thread.ip = (unsigned long) ret_from_fork; 214 215 task_user_gs(p) = get_user_gs(regs); 216 217 p->fpu_counter = 0; 218 p->thread.io_bitmap_ptr = NULL; 219 tsk = current; 220 err = -ENOMEM; 221 222 memset(p->thread.ptrace_bps, 0, sizeof(p->thread.ptrace_bps)); 223 224 if (unlikely(test_tsk_thread_flag(tsk, TIF_IO_BITMAP))) { 225 p->thread.io_bitmap_ptr = kmemdup(tsk->thread.io_bitmap_ptr, 226 IO_BITMAP_BYTES, GFP_KERNEL); 227 if (!p->thread.io_bitmap_ptr) { 228 p->thread.io_bitmap_max = 0; 229 return -ENOMEM; 230 } 231 set_tsk_thread_flag(p, TIF_IO_BITMAP); 232 } 233 234 err = 0; 235 236 /* 237 * Set a new TLS for the child thread? 238 */ 239 if (clone_flags & CLONE_SETTLS) 240 err = do_set_thread_area(p, -1, 241 (struct user_desc __user *)childregs->si, 0); 242 243 if (err && p->thread.io_bitmap_ptr) { 244 kfree(p->thread.io_bitmap_ptr); 245 p->thread.io_bitmap_max = 0; 246 } 247 return err; 248 } 249 250 void 251 start_thread(struct pt_regs *regs, unsigned long new_ip, unsigned long new_sp) 252 { 253 set_user_gs(regs, 0); 254 regs->fs = 0; 255 regs->ds = __USER_DS; 256 regs->es = __USER_DS; 257 regs->ss = __USER_DS; 258 regs->cs = __USER_CS; 259 regs->ip = new_ip; 260 regs->sp = new_sp; 261 /* 262 * Free the old FP and other extended state 263 */ 264 free_thread_xstate(current); 265 } 266 EXPORT_SYMBOL_GPL(start_thread); 267 268 269 /* 270 * switch_to(x,y) should switch tasks from x to y. 271 * 272 * We fsave/fwait so that an exception goes off at the right time 273 * (as a call from the fsave or fwait in effect) rather than to 274 * the wrong process. Lazy FP saving no longer makes any sense 275 * with modern CPU's, and this simplifies a lot of things (SMP 276 * and UP become the same). 277 * 278 * NOTE! We used to use the x86 hardware context switching. The 279 * reason for not using it any more becomes apparent when you 280 * try to recover gracefully from saved state that is no longer 281 * valid (stale segment register values in particular). With the 282 * hardware task-switch, there is no way to fix up bad state in 283 * a reasonable manner. 284 * 285 * The fact that Intel documents the hardware task-switching to 286 * be slow is a fairly red herring - this code is not noticeably 287 * faster. However, there _is_ some room for improvement here, 288 * so the performance issues may eventually be a valid point. 289 * More important, however, is the fact that this allows us much 290 * more flexibility. 291 * 292 * The return value (in %ax) will be the "prev" task after 293 * the task-switch, and shows up in ret_from_fork in entry.S, 294 * for example. 295 */ 296 __notrace_funcgraph struct task_struct * 297 __switch_to(struct task_struct *prev_p, struct task_struct *next_p) 298 { 299 struct thread_struct *prev = &prev_p->thread, 300 *next = &next_p->thread; 301 int cpu = smp_processor_id(); 302 struct tss_struct *tss = &per_cpu(init_tss, cpu); 303 fpu_switch_t fpu; 304 305 /* never put a printk in __switch_to... printk() calls wake_up*() indirectly */ 306 307 fpu = switch_fpu_prepare(prev_p, next_p, cpu); 308 309 /* 310 * Reload esp0. 311 */ 312 load_sp0(tss, next); 313 314 /* 315 * Save away %gs. No need to save %fs, as it was saved on the 316 * stack on entry. No need to save %es and %ds, as those are 317 * always kernel segments while inside the kernel. Doing this 318 * before setting the new TLS descriptors avoids the situation 319 * where we temporarily have non-reloadable segments in %fs 320 * and %gs. This could be an issue if the NMI handler ever 321 * used %fs or %gs (it does not today), or if the kernel is 322 * running inside of a hypervisor layer. 323 */ 324 lazy_save_gs(prev->gs); 325 326 /* 327 * Load the per-thread Thread-Local Storage descriptor. 328 */ 329 load_TLS(next, cpu); 330 331 /* 332 * Restore IOPL if needed. In normal use, the flags restore 333 * in the switch assembly will handle this. But if the kernel 334 * is running virtualized at a non-zero CPL, the popf will 335 * not restore flags, so it must be done in a separate step. 336 */ 337 if (get_kernel_rpl() && unlikely(prev->iopl != next->iopl)) 338 set_iopl_mask(next->iopl); 339 340 /* 341 * Now maybe handle debug registers and/or IO bitmaps 342 */ 343 if (unlikely(task_thread_info(prev_p)->flags & _TIF_WORK_CTXSW_PREV || 344 task_thread_info(next_p)->flags & _TIF_WORK_CTXSW_NEXT)) 345 __switch_to_xtra(prev_p, next_p, tss); 346 347 /* 348 * Leave lazy mode, flushing any hypercalls made here. 349 * This must be done before restoring TLS segments so 350 * the GDT and LDT are properly updated, and must be 351 * done before math_state_restore, so the TS bit is up 352 * to date. 353 */ 354 arch_end_context_switch(next_p); 355 356 /* 357 * Restore %gs if needed (which is common) 358 */ 359 if (prev->gs | next->gs) 360 lazy_load_gs(next->gs); 361 362 switch_fpu_finish(next_p, fpu); 363 364 percpu_write(current_task, next_p); 365 366 return prev_p; 367 } 368 369 #define top_esp (THREAD_SIZE - sizeof(unsigned long)) 370 #define top_ebp (THREAD_SIZE - 2*sizeof(unsigned long)) 371 372 unsigned long get_wchan(struct task_struct *p) 373 { 374 unsigned long bp, sp, ip; 375 unsigned long stack_page; 376 int count = 0; 377 if (!p || p == current || p->state == TASK_RUNNING) 378 return 0; 379 stack_page = (unsigned long)task_stack_page(p); 380 sp = p->thread.sp; 381 if (!stack_page || sp < stack_page || sp > top_esp+stack_page) 382 return 0; 383 /* include/asm-i386/system.h:switch_to() pushes bp last. */ 384 bp = *(unsigned long *) sp; 385 do { 386 if (bp < stack_page || bp > top_ebp+stack_page) 387 return 0; 388 ip = *(unsigned long *) (bp+4); 389 if (!in_sched_functions(ip)) 390 return ip; 391 bp = *(unsigned long *) bp; 392 } while (count++ < 16); 393 return 0; 394 } 395 396