[kbuild-all] [tglx-devel:x86/entry 8/21] arch/x86/entry/common.c:434:3: error: implicit declaration of function 'irq_enter_from_user_mode'; did you mean

tree: https://git.kernel.org/pub/scm/linux/kernel/git/tglx/devel.git x86/entry head: 5c8d5a67cdc4a03d2c383cb4e70375c2da1eb5eb commit: 20049a8a124513a9de53d614179c434ade144242 [8/21] x86/entry: Use generic syscall entry function config: i386-tinyconfig (attached as .config) compiler: gcc-9 (Debian 9.3.0-14) 9.3.0 reproduce (this is a W=1 build): git checkout 20049a8a124513a9de53d614179c434ade144242 # save the attached .config to linux build tree make W=1 ARCH=i386 If you fix the issue, kindly add following tag as appropriate Reported-by: kernel test robot <lkp(a)intel.com&gt; Note: the tglx-devel/x86/entry HEAD 5c8d5a67cdc4a03d2c383cb4e70375c2da1eb5eb builds fine. It only hurts bisectibility. All errors (new ones prefixed by >>): arch/x86/entry/common.c:222:24: warning: no previous prototype for 'syscall_return_slowpath' [-Wmissing-prototypes] 222 | __visible noinstr void syscall_return_slowpath(struct pt_regs *regs) | ^~~~~~~~~~~~~~~~~~~~~~~ arch/x86/entry/common.c:282:24: warning: no previous prototype for 'do_int80_syscall_32' [-Wmissing-prototypes] 282 | __visible noinstr void do_int80_syscall_32(struct pt_regs *regs) | ^~~~~~~~~~~~~~~~~~~ arch/x86/entry/common.c:324:24: warning: no previous prototype for 'do_fast_syscall_32' [-Wmissing-prototypes] 324 | __visible noinstr long do_fast_syscall_32(struct pt_regs *regs) | ^~~~~~~~~~~~~~~~~~ arch/x86/entry/common.c:376:24: warning: no previous prototype for 'do_SYSENTER_32' [-Wmissing-prototypes] 376 | __visible noinstr long do_SYSENTER_32(struct pt_regs *regs) | ^~~~~~~~~~~~~~ arch/x86/entry/common.c: In function 'idtentry_enter': 434 | irq_enter_from_user_mode(regs); | ^~~~~~~~~~~~~~~~~~~~~~~~ | irqentry_enter_from_user_mode cc1: some warnings being treated as errors vim +434 arch/x86/entry/common.c 392 393 /** 394 * idtentry_enter - Handle state tracking on ordinary idtentries 395 * @regs: Pointer to pt_regs of interrupted context 396 * 397 * Invokes: 398 * - lockdep irqflag state tracking as low level ASM entry disabled 399 * interrupts. 400 * 401 * - Context tracking if the exception hit user mode. 402 * 403 * - The hardirq tracer to keep the state consistent as low level ASM 404 * entry disabled interrupts. 405 * 406 * As a precondition, this requires that the entry came from user mode, 407 * idle, or a kernel context in which RCU is watching. 408 * 409 * For kernel mode entries RCU handling is done conditional. If RCU is 410 * watching then the only RCU requirement is to check whether the tick has 411 * to be restarted. If RCU is not watching then rcu_irq_enter() has to be 412 * invoked on entry and rcu_irq_exit() on exit. 413 * 414 * Avoiding the rcu_irq_enter/exit() calls is an optimization but also 415 * solves the problem of kernel mode pagefaults which can schedule, which 416 * is not possible after invoking rcu_irq_enter() without undoing it. 417 * 418 * For user mode entries irq_enter_from_user_mode() must be invoked to 419 * establish the proper context for NOHZ_FULL. Otherwise scheduling on exit 420 * would not be possible. 421 * 422 * Returns: An opaque object that must be passed to idtentry_exit() 423 * 424 * The return value must be fed into the state argument of 425 * idtentry_exit(). 426 */ 427 idtentry_state_t noinstr idtentry_enter(struct pt_regs *regs) 428 { 429 idtentry_state_t ret = { 430 .exit_rcu = false, 431 }; 432 433 if (user_mode(regs)) { 435 return ret; 436 } 437 438 /* 439 * If this entry hit the idle task invoke rcu_irq_enter() whether 440 * RCU is watching or not. 441 * 442 * Interupts can nest when the first interrupt invokes softirq 443 * processing on return which enables interrupts. 444 * 445 * Scheduler ticks in the idle task can mark quiescent state and 446 * terminate a grace period, if and only if the timer interrupt is 447 * not nested into another interrupt. 448 * 449 * Checking for __rcu_is_watching() here would prevent the nesting 450 * interrupt to invoke rcu_irq_enter(). If that nested interrupt is 451 * the tick then rcu_flavor_sched_clock_irq() would wrongfully 452 * assume that it is the first interupt and eventually claim 453 * quiescient state and end grace periods prematurely. 454 * 455 * Unconditionally invoke rcu_irq_enter() so RCU state stays 456 * consistent. 457 * 458 * TINY_RCU does not support EQS, so let the compiler eliminate 459 * this part when enabled. 460 */ 461 if (!IS_ENABLED(CONFIG_TINY_RCU) && is_idle_task(current)) { 462 /* 463 * If RCU is not watching then the same careful 464 * sequence vs. lockdep and tracing is required 465 * as in irq_enter_from_user_mode(). 466 */ 467 lockdep_hardirqs_off(CALLER_ADDR0); 468 rcu_irq_enter(); 469 instrumentation_begin(); 470 trace_hardirqs_off_finish(); 471 instrumentation_end(); 472 473 ret.exit_rcu = true; 474 return ret; 475 } 476 477 /* 478 * If RCU is watching then RCU only wants to check whether it needs 479 * to restart the tick in NOHZ mode. rcu_irq_enter_check_tick() 480 * already contains a warning when RCU is not watching, so no point 481 * in having another one here. 482 */ 483 instrumentation_begin(); 484 rcu_irq_enter_check_tick(); 485 /* Use the combo lockdep/tracing function */ 486 trace_hardirqs_off(); 487 instrumentation_end(); 488 489 return ret; 490 } 491 --- 0-DAY CI Kernel Test Service, Intel Corporation https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org