process.c 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481
  1. /*
  2. * Based on arch/arm/kernel/process.c
  3. *
  4. * Original Copyright (C) 1995 Linus Torvalds
  5. * Copyright (C) 1996-2000 Russell King - Converted to ARM.
  6. * Copyright (C) 2012 ARM Ltd.
  7. *
  8. * This program is free software; you can redistribute it and/or modify
  9. * it under the terms of the GNU General Public License version 2 as
  10. * published by the Free Software Foundation.
  11. *
  12. * This program is distributed in the hope that it will be useful,
  13. * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  15. * GNU General Public License for more details.
  16. *
  17. * You should have received a copy of the GNU General Public License
  18. * along with this program. If not, see <http://www.gnu.org/licenses/>.
  19. */
  20. #include <stdarg.h>
  21. #include <linux/compat.h>
  22. #include <linux/efi.h>
  23. #include <linux/export.h>
  24. #include <linux/sched.h>
  25. #include <linux/sched/debug.h>
  26. #include <linux/sched/task.h>
  27. #include <linux/sched/task_stack.h>
  28. #include <linux/kernel.h>
  29. #include <linux/mm.h>
  30. #include <linux/stddef.h>
  31. #include <linux/unistd.h>
  32. #include <linux/user.h>
  33. #include <linux/delay.h>
  34. #include <linux/reboot.h>
  35. #include <linux/interrupt.h>
  36. #include <linux/init.h>
  37. #include <linux/cpu.h>
  38. #include <linux/elfcore.h>
  39. #include <linux/pm.h>
  40. #include <linux/tick.h>
  41. #include <linux/utsname.h>
  42. #include <linux/uaccess.h>
  43. #include <linux/random.h>
  44. #include <linux/hw_breakpoint.h>
  45. #include <linux/personality.h>
  46. #include <linux/notifier.h>
  47. #include <trace/events/power.h>
  48. #include <linux/percpu.h>
  49. #include <linux/thread_info.h>
  50. #include <asm/alternative.h>
  51. #include <asm/compat.h>
  52. #include <asm/cacheflush.h>
  53. #include <asm/exec.h>
  54. #include <asm/fpsimd.h>
  55. #include <asm/mmu_context.h>
  56. #include <asm/processor.h>
  57. #include <asm/stacktrace.h>
  58. #ifdef CONFIG_CC_STACKPROTECTOR
  59. #include <linux/stackprotector.h>
  60. unsigned long __stack_chk_guard __read_mostly;
  61. EXPORT_SYMBOL(__stack_chk_guard);
  62. #endif
  63. /*
  64. * Function pointers to optional machine specific functions
  65. */
  66. void (*pm_power_off)(void);
  67. EXPORT_SYMBOL_GPL(pm_power_off);
  68. void (*arm_pm_restart)(enum reboot_mode reboot_mode, const char *cmd);
  69. /*
  70. * This is our default idle handler.
  71. */
  72. void arch_cpu_idle(void)
  73. {
  74. /*
  75. * This should do all the clock switching and wait for interrupt
  76. * tricks
  77. */
  78. trace_cpu_idle_rcuidle(1, smp_processor_id());
  79. cpu_do_idle();
  80. local_irq_enable();
  81. trace_cpu_idle_rcuidle(PWR_EVENT_EXIT, smp_processor_id());
  82. }
  83. #ifdef CONFIG_HOTPLUG_CPU
  84. void arch_cpu_idle_dead(void)
  85. {
  86. cpu_die();
  87. }
  88. #endif
  89. /*
  90. * Called by kexec, immediately prior to machine_kexec().
  91. *
  92. * This must completely disable all secondary CPUs; simply causing those CPUs
  93. * to execute e.g. a RAM-based pin loop is not sufficient. This allows the
  94. * kexec'd kernel to use any and all RAM as it sees fit, without having to
  95. * avoid any code or data used by any SW CPU pin loop. The CPU hotplug
  96. * functionality embodied in disable_nonboot_cpus() to achieve this.
  97. */
  98. void machine_shutdown(void)
  99. {
  100. disable_nonboot_cpus();
  101. }
  102. /*
  103. * Halting simply requires that the secondary CPUs stop performing any
  104. * activity (executing tasks, handling interrupts). smp_send_stop()
  105. * achieves this.
  106. */
  107. void machine_halt(void)
  108. {
  109. local_irq_disable();
  110. smp_send_stop();
  111. while (1);
  112. }
  113. /*
  114. * Power-off simply requires that the secondary CPUs stop performing any
  115. * activity (executing tasks, handling interrupts). smp_send_stop()
  116. * achieves this. When the system power is turned off, it will take all CPUs
  117. * with it.
  118. */
  119. void machine_power_off(void)
  120. {
  121. local_irq_disable();
  122. smp_send_stop();
  123. if (pm_power_off)
  124. pm_power_off();
  125. }
  126. /*
  127. * Restart requires that the secondary CPUs stop performing any activity
  128. * while the primary CPU resets the system. Systems with multiple CPUs must
  129. * provide a HW restart implementation, to ensure that all CPUs reset at once.
  130. * This is required so that any code running after reset on the primary CPU
  131. * doesn't have to co-ordinate with other CPUs to ensure they aren't still
  132. * executing pre-reset code, and using RAM that the primary CPU's code wishes
  133. * to use. Implementing such co-ordination would be essentially impossible.
  134. */
  135. void machine_restart(char *cmd)
  136. {
  137. /* Disable interrupts first */
  138. local_irq_disable();
  139. smp_send_stop();
  140. /*
  141. * UpdateCapsule() depends on the system being reset via
  142. * ResetSystem().
  143. */
  144. if (efi_enabled(EFI_RUNTIME_SERVICES))
  145. efi_reboot(reboot_mode, NULL);
  146. /* Now call the architecture specific reboot code. */
  147. if (arm_pm_restart)
  148. arm_pm_restart(reboot_mode, cmd);
  149. else
  150. do_kernel_restart(cmd);
  151. /*
  152. * Whoops - the architecture was unable to reboot.
  153. */
  154. printk("Reboot failed -- System halted\n");
  155. while (1);
  156. }
  157. static void print_pstate(struct pt_regs *regs)
  158. {
  159. u64 pstate = regs->pstate;
  160. if (compat_user_mode(regs)) {
  161. printk("pstate: %08llx (%c%c%c%c %c %s %s %c%c%c)\n",
  162. pstate,
  163. pstate & COMPAT_PSR_N_BIT ? 'N' : 'n',
  164. pstate & COMPAT_PSR_Z_BIT ? 'Z' : 'z',
  165. pstate & COMPAT_PSR_C_BIT ? 'C' : 'c',
  166. pstate & COMPAT_PSR_V_BIT ? 'V' : 'v',
  167. pstate & COMPAT_PSR_Q_BIT ? 'Q' : 'q',
  168. pstate & COMPAT_PSR_T_BIT ? "T32" : "A32",
  169. pstate & COMPAT_PSR_E_BIT ? "BE" : "LE",
  170. pstate & COMPAT_PSR_A_BIT ? 'A' : 'a',
  171. pstate & COMPAT_PSR_I_BIT ? 'I' : 'i',
  172. pstate & COMPAT_PSR_F_BIT ? 'F' : 'f');
  173. } else {
  174. printk("pstate: %08llx (%c%c%c%c %c%c%c%c %cPAN %cUAO)\n",
  175. pstate,
  176. pstate & PSR_N_BIT ? 'N' : 'n',
  177. pstate & PSR_Z_BIT ? 'Z' : 'z',
  178. pstate & PSR_C_BIT ? 'C' : 'c',
  179. pstate & PSR_V_BIT ? 'V' : 'v',
  180. pstate & PSR_D_BIT ? 'D' : 'd',
  181. pstate & PSR_A_BIT ? 'A' : 'a',
  182. pstate & PSR_I_BIT ? 'I' : 'i',
  183. pstate & PSR_F_BIT ? 'F' : 'f',
  184. pstate & PSR_PAN_BIT ? '+' : '-',
  185. pstate & PSR_UAO_BIT ? '+' : '-');
  186. }
  187. }
  188. void __show_regs(struct pt_regs *regs)
  189. {
  190. int i, top_reg;
  191. u64 lr, sp;
  192. if (compat_user_mode(regs)) {
  193. lr = regs->compat_lr;
  194. sp = regs->compat_sp;
  195. top_reg = 12;
  196. } else {
  197. lr = regs->regs[30];
  198. sp = regs->sp;
  199. top_reg = 29;
  200. }
  201. show_regs_print_info(KERN_DEFAULT);
  202. print_pstate(regs);
  203. printk("pc : %pS\n", (void *)regs->pc);
  204. printk("lr : %pS\n", (void *)lr);
  205. printk("sp : %016llx\n", sp);
  206. i = top_reg;
  207. while (i >= 0) {
  208. printk("x%-2d: %016llx ", i, regs->regs[i]);
  209. i--;
  210. if (i % 2 == 0) {
  211. pr_cont("x%-2d: %016llx ", i, regs->regs[i]);
  212. i--;
  213. }
  214. pr_cont("\n");
  215. }
  216. }
  217. void show_regs(struct pt_regs * regs)
  218. {
  219. __show_regs(regs);
  220. dump_backtrace(regs, NULL);
  221. }
  222. static void tls_thread_flush(void)
  223. {
  224. write_sysreg(0, tpidr_el0);
  225. if (is_compat_task()) {
  226. current->thread.tp_value = 0;
  227. /*
  228. * We need to ensure ordering between the shadow state and the
  229. * hardware state, so that we don't corrupt the hardware state
  230. * with a stale shadow state during context switch.
  231. */
  232. barrier();
  233. write_sysreg(0, tpidrro_el0);
  234. }
  235. }
  236. void flush_thread(void)
  237. {
  238. fpsimd_flush_thread();
  239. tls_thread_flush();
  240. flush_ptrace_hw_breakpoint(current);
  241. }
  242. void release_thread(struct task_struct *dead_task)
  243. {
  244. }
  245. void arch_release_task_struct(struct task_struct *tsk)
  246. {
  247. fpsimd_release_task(tsk);
  248. }
  249. /*
  250. * src and dst may temporarily have aliased sve_state after task_struct
  251. * is copied. We cannot fix this properly here, because src may have
  252. * live SVE state and dst's thread_info may not exist yet, so tweaking
  253. * either src's or dst's TIF_SVE is not safe.
  254. *
  255. * The unaliasing is done in copy_thread() instead. This works because
  256. * dst is not schedulable or traceable until both of these functions
  257. * have been called.
  258. */
  259. int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
  260. {
  261. if (current->mm)
  262. fpsimd_preserve_current_state();
  263. *dst = *src;
  264. return 0;
  265. }
  266. asmlinkage void ret_from_fork(void) asm("ret_from_fork");
  267. int copy_thread(unsigned long clone_flags, unsigned long stack_start,
  268. unsigned long stk_sz, struct task_struct *p)
  269. {
  270. struct pt_regs *childregs = task_pt_regs(p);
  271. memset(&p->thread.cpu_context, 0, sizeof(struct cpu_context));
  272. /*
  273. * Unalias p->thread.sve_state (if any) from the parent task
  274. * and disable discard SVE state for p:
  275. */
  276. clear_tsk_thread_flag(p, TIF_SVE);
  277. p->thread.sve_state = NULL;
  278. if (likely(!(p->flags & PF_KTHREAD))) {
  279. *childregs = *current_pt_regs();
  280. childregs->regs[0] = 0;
  281. /*
  282. * Read the current TLS pointer from tpidr_el0 as it may be
  283. * out-of-sync with the saved value.
  284. */
  285. *task_user_tls(p) = read_sysreg(tpidr_el0);
  286. if (stack_start) {
  287. if (is_compat_thread(task_thread_info(p)))
  288. childregs->compat_sp = stack_start;
  289. else
  290. childregs->sp = stack_start;
  291. }
  292. /*
  293. * If a TLS pointer was passed to clone (4th argument), use it
  294. * for the new thread.
  295. */
  296. if (clone_flags & CLONE_SETTLS)
  297. p->thread.tp_value = childregs->regs[3];
  298. } else {
  299. memset(childregs, 0, sizeof(struct pt_regs));
  300. childregs->pstate = PSR_MODE_EL1h;
  301. if (IS_ENABLED(CONFIG_ARM64_UAO) &&
  302. cpus_have_const_cap(ARM64_HAS_UAO))
  303. childregs->pstate |= PSR_UAO_BIT;
  304. p->thread.cpu_context.x19 = stack_start;
  305. p->thread.cpu_context.x20 = stk_sz;
  306. }
  307. p->thread.cpu_context.pc = (unsigned long)ret_from_fork;
  308. p->thread.cpu_context.sp = (unsigned long)childregs;
  309. ptrace_hw_copy_thread(p);
  310. return 0;
  311. }
  312. void tls_preserve_current_state(void)
  313. {
  314. *task_user_tls(current) = read_sysreg(tpidr_el0);
  315. }
  316. static void tls_thread_switch(struct task_struct *next)
  317. {
  318. unsigned long tpidr, tpidrro;
  319. tls_preserve_current_state();
  320. tpidr = *task_user_tls(next);
  321. tpidrro = is_compat_thread(task_thread_info(next)) ?
  322. next->thread.tp_value : 0;
  323. write_sysreg(tpidr, tpidr_el0);
  324. write_sysreg(tpidrro, tpidrro_el0);
  325. }
  326. /* Restore the UAO state depending on next's addr_limit */
  327. void uao_thread_switch(struct task_struct *next)
  328. {
  329. if (IS_ENABLED(CONFIG_ARM64_UAO)) {
  330. if (task_thread_info(next)->addr_limit == KERNEL_DS)
  331. asm(ALTERNATIVE("nop", SET_PSTATE_UAO(1), ARM64_HAS_UAO));
  332. else
  333. asm(ALTERNATIVE("nop", SET_PSTATE_UAO(0), ARM64_HAS_UAO));
  334. }
  335. }
  336. /*
  337. * We store our current task in sp_el0, which is clobbered by userspace. Keep a
  338. * shadow copy so that we can restore this upon entry from userspace.
  339. *
  340. * This is *only* for exception entry from EL0, and is not valid until we
  341. * __switch_to() a user task.
  342. */
  343. DEFINE_PER_CPU(struct task_struct *, __entry_task);
  344. static void entry_task_switch(struct task_struct *next)
  345. {
  346. __this_cpu_write(__entry_task, next);
  347. }
  348. /*
  349. * Thread switching.
  350. */
  351. __notrace_funcgraph struct task_struct *__switch_to(struct task_struct *prev,
  352. struct task_struct *next)
  353. {
  354. struct task_struct *last;
  355. fpsimd_thread_switch(next);
  356. tls_thread_switch(next);
  357. hw_breakpoint_thread_switch(next);
  358. contextidr_thread_switch(next);
  359. entry_task_switch(next);
  360. uao_thread_switch(next);
  361. /*
  362. * Complete any pending TLB or cache maintenance on this CPU in case
  363. * the thread migrates to a different CPU.
  364. * This full barrier is also required by the membarrier system
  365. * call.
  366. */
  367. dsb(ish);
  368. /* the actual thread switch */
  369. last = cpu_switch_to(prev, next);
  370. return last;
  371. }
  372. unsigned long get_wchan(struct task_struct *p)
  373. {
  374. struct stackframe frame;
  375. unsigned long stack_page, ret = 0;
  376. int count = 0;
  377. if (!p || p == current || p->state == TASK_RUNNING)
  378. return 0;
  379. stack_page = (unsigned long)try_get_task_stack(p);
  380. if (!stack_page)
  381. return 0;
  382. frame.fp = thread_saved_fp(p);
  383. frame.pc = thread_saved_pc(p);
  384. #ifdef CONFIG_FUNCTION_GRAPH_TRACER
  385. frame.graph = p->curr_ret_stack;
  386. #endif
  387. do {
  388. if (unwind_frame(p, &frame))
  389. goto out;
  390. if (!in_sched_functions(frame.pc)) {
  391. ret = frame.pc;
  392. goto out;
  393. }
  394. } while (count ++ < 16);
  395. out:
  396. put_task_stack(p);
  397. return ret;
  398. }
  399. unsigned long arch_align_stack(unsigned long sp)
  400. {
  401. if (!(current->personality & ADDR_NO_RANDOMIZE) && randomize_va_space)
  402. sp -= get_random_int() & ~PAGE_MASK;
  403. return sp & ~0xf;
  404. }
  405. unsigned long arch_randomize_brk(struct mm_struct *mm)
  406. {
  407. if (is_compat_task())
  408. return randomize_page(mm->brk, SZ_32M);
  409. else
  410. return randomize_page(mm->brk, SZ_1G);
  411. }
  412. /*
  413. * Called from setup_new_exec() after (COMPAT_)SET_PERSONALITY.
  414. */
  415. void arch_setup_new_exec(void)
  416. {
  417. current->mm->context.flags = is_compat_task() ? MMCF_AARCH32 : 0;
  418. }