1 2/* $NetBSD: trap.c,v 1.309 2023/10/05 19:41:04 ad Exp $ */ 3 4/*- 5 * Copyright (c) 1998, 2000, 2005, 2006, 2007, 2008 The NetBSD Foundation, Inc. 6 * All rights reserved. 7 * 8 * This code is derived from software contributed to The NetBSD Foundation 9 * by Charles M. Hannum. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 21 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 22 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 23 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 24 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 25 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 26 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 27 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 28 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 29 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 30 * POSSIBILITY OF SUCH DAMAGE. 31 */ 32 33/*- 34 * Copyright (c) 1990 The Regents of the University of California. 35 * All rights reserved. 36 * 37 * This code is derived from software contributed to Berkeley by 38 * the University of Utah, and William Jolitz. 39 * 40 * Redistribution and use in source and binary forms, with or without 41 * modification, are permitted provided that the following conditions 42 * are met: 43 * 1. Redistributions of source code must retain the above copyright 44 * notice, this list of conditions and the following disclaimer. 45 * 2. Redistributions in binary form must reproduce the above copyright 46 * notice, this list of conditions and the following disclaimer in the 47 * documentation and/or other materials provided with the distribution. 48 * 3. Neither the name of the University nor the names of its contributors 49 * may be used to endorse or promote products derived from this software 50 * without specific prior written permission. 51 * 52 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 53 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 54 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 55 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 56 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 57 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 58 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 59 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 60 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 61 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 62 * SUCH DAMAGE. 63 * 64 * @(#)trap.c 7.4 (Berkeley) 5/13/91 65 */ 66 67/* 68 * 386 Trap and System call handling 69 */ 70 71#include <sys/cdefs.h> 72__KERNEL_RCSID(0, "$NetBSD: trap.c,v 1.309 2023/10/05 19:41:04 ad Exp $"); 73 74#include "opt_ddb.h" 75#include "opt_kgdb.h" 76#include "opt_lockdebug.h" 77#include "opt_multiprocessor.h" 78#include "opt_xen.h" 79#include "opt_dtrace.h" 80#include "opt_compat_netbsd.h" 81 82#include <sys/param.h> 83#include <sys/systm.h> 84#include <sys/proc.h> 85#include <sys/acct.h> 86#include <sys/kauth.h> 87#include <sys/kernel.h> 88#include <sys/kmem.h> 89#include <sys/ras.h> 90#include <sys/signal.h> 91#include <sys/syscall.h> 92#include <sys/cpu.h> 93#include <sys/ucontext.h> 94 95#include <uvm/uvm_extern.h> 96 97#include <machine/cpufunc.h> 98#include <machine/psl.h> 99#include <machine/reg.h> 100#include <machine/trap.h> 101#include <machine/userret.h> 102#include <machine/db_machdep.h> 103#include <machine/pmap_private.h> 104 105#include "mca.h" 106#if NMCA > 0 107#include <machine/mca_machdep.h> 108#endif 109 110#include <x86/dbregs.h> 111#include <x86/nmi.h> 112 113#include "isa.h" 114 115#include <sys/kgdb.h> 116 117#ifdef KDTRACE_HOOKS 118#include <sys/dtrace_bsd.h> 119 120/* 121 * This is a hook which is initialized by the dtrace module 122 * to handle traps which might occur during DTrace probe 123 * execution. 124 */ 125dtrace_trap_func_t dtrace_trap_func = NULL; 126 127dtrace_doubletrap_func_t dtrace_doubletrap_func = NULL; 128#endif 129 130void trap(struct trapframe *); 131void trap_tss(struct i386tss *, int, int); 132void trap_return_fault_return(struct trapframe *) __dead; 133#ifndef XENPV 134int ss_shadow(struct trapframe *tf); 135#endif 136 137const char * const trap_type[] = { 138 "privileged instruction fault", /* 0 T_PRIVINFLT */ 139 "breakpoint trap", /* 1 T_BPTFLT */ 140 "arithmetic trap", /* 2 T_ARITHTRAP */ 141 "asynchronous system trap", /* 3 T_ASTFLT */ 142 "protection fault", /* 4 T_PROTFLT */ 143 "trace trap", /* 5 T_TRCTRAP */ 144 "page fault", /* 6 T_PAGEFLT */ 145 "alignment fault", /* 7 T_ALIGNFLT */ 146 "integer divide fault", /* 8 T_DIVIDE */ 147 "non-maskable interrupt", /* 9 T_NMI */ 148 "overflow trap", /* 10 T_OFLOW */ 149 "bounds check fault", /* 11 T_BOUND */ 150 "FPU not available fault", /* 12 T_DNA */ 151 "double fault", /* 13 T_DOUBLEFLT */ 152 "FPU operand fetch fault", /* 14 T_FPOPFLT */ 153 "invalid TSS fault", /* 15 T_TSSFLT */ 154 "segment not present fault", /* 16 T_SEGNPFLT */ 155 "stack fault", /* 17 T_STKFLT */ 156 "machine check fault", /* 18 T_MCA */ 157 "SSE FP exception", /* 19 T_XMM */ 158 "reserved trap", /* 20 T_RESERVED */ 159}; 160int trap_types = __arraycount(trap_type); 161 162#ifdef DEBUG 163int trapdebug = 0; 164#endif 165 166#define IDTVEC(name) __CONCAT(X, name) 167 168#ifdef TRAP_SIGDEBUG 169static void sigdebug(const struct trapframe *, const ksiginfo_t *, int); 170#define SIGDEBUG(a, b, c) sigdebug(a, b, c) 171#else 172#define SIGDEBUG(a, b, c) 173#endif 174 175void 176trap_tss(struct i386tss *tss, int trapno, int code) 177{ 178 struct trapframe tf; 179 180 tf.tf_gs = tss->tss_gs; 181 tf.tf_fs = tss->tss_fs; 182 tf.tf_es = tss->__tss_es; 183 tf.tf_ds = tss->__tss_ds; 184 tf.tf_edi = tss->__tss_edi; 185 tf.tf_esi = tss->__tss_esi; 186 tf.tf_ebp = tss->tss_ebp; 187 tf.tf_ebx = tss->__tss_ebx; 188 tf.tf_edx = tss->__tss_edx; 189 tf.tf_ecx = tss->__tss_ecx; 190 tf.tf_eax = tss->__tss_eax; 191 tf.tf_trapno = trapno; 192 tf.tf_err = code | TC_TSS; 193 tf.tf_eip = tss->__tss_eip; 194 tf.tf_cs = tss->__tss_cs; 195 tf.tf_eflags = tss->__tss_eflags; 196 tf.tf_esp = tss->tss_esp; 197 tf.tf_ss = tss->__tss_ss; 198 trap(&tf); 199} 200 201static void * 202onfault_handler(const struct pcb *pcb, const struct trapframe *tf) 203{ 204 struct onfault_table { 205 uintptr_t start; 206 uintptr_t end; 207 void *handler; 208 }; 209 extern const struct onfault_table onfault_table[]; 210 const struct onfault_table *p; 211 uintptr_t pc; 212 213 if (pcb->pcb_onfault != NULL) { 214 return pcb->pcb_onfault; 215 } 216 217 pc = tf->tf_eip; 218 for (p = onfault_table; p->start; p++) { 219 if (p->start <= pc && pc < p->end) { 220 return p->handler; 221 } 222 } 223 return NULL; 224} 225 226static void 227trap_print(const struct trapframe *frame, const lwp_t *l) 228{ 229 const int type = frame->tf_trapno; 230 231 if (frame->tf_trapno < trap_types) { 232 printf("fatal %s", trap_type[type]); 233 } else { 234 printf("unknown trap %d", type); 235 } 236 printf(" in %s mode\n", (type & T_USER) ? "user" : "supervisor"); 237 238 printf("trap type %d code %#x eip %#x cs %#x eflags %#x cr2 %#lx " 239 "ilevel %#x esp %#x\n", 240 type, frame->tf_err, frame->tf_eip, frame->tf_cs, frame->tf_eflags, 241 (long)rcr2(), curcpu()->ci_ilevel, frame->tf_esp); 242 243 printf("curlwp %p pid %d lid %d lowest kstack %p\n", 244 l, l->l_proc->p_pid, l->l_lid, KSTACK_LOWEST_ADDR(l)); 245} 246 247#ifndef XENPV 248int 249ss_shadow(struct trapframe *tf) 250{ 251 struct gate_descriptor *gd; 252 struct cpu_info *ci; 253 struct idt_vec *iv; 254 idt_descriptor_t *idt; 255 uintptr_t eip, func; 256 size_t i; 257 258 eip = tf->tf_eip; 259 ci = curcpu(); 260 iv = idt_vec_ref(&ci->ci_idtvec); 261 idt = iv->iv_idt; 262 263 for (i = 0; i < 256; i++) { 264 gd = &idt[i]; 265 func = (gd->gd_hioffset << 16) | gd->gd_looffset; 266 if (eip == func) 267 return 1; 268 } 269 270 return 0; 271} 272#endif 273 274/* 275 * trap(frame): exception, fault, and trap interface to BSD kernel. 276 * 277 * This common code is called from assembly language IDT gate entry routines 278 * that prepare a suitable stack frame, and restore this frame after the 279 * exception has been processed. Note that the effect is as if the arguments 280 * were passed call by reference. 281 */ 282void 283trap(struct trapframe *frame) 284{ 285 struct lwp *l = curlwp; 286 struct proc *p; 287 struct pcb *pcb; 288 extern char kcopy_fault[], return_address_fault[]; 289 struct trapframe *vframe; 290 ksiginfo_t ksi; 291 void *onfault; 292 int type, error = 0; 293 uint32_t cr2; 294 bool pfail; 295 296 if (__predict_true(l != NULL)) { 297 pcb = lwp_getpcb(l); 298 p = l->l_proc; 299 } else { 300 /* 301 * this can happen eg. on break points in early on boot. 302 */ 303 pcb = NULL; 304 p = NULL; 305 } 306 type = frame->tf_trapno; 307 308#ifdef DEBUG 309 if (trapdebug) { 310 trap_print(frame, l); 311 } 312#endif 313 if (type != T_NMI && !KERNELMODE(frame->tf_cs)) { 314 type |= T_USER; 315 l->l_md.md_regs = frame; 316 pcb->pcb_cr2 = 0; 317 } 318 319#ifdef KDTRACE_HOOKS 320 /* 321 * A trap can occur while DTrace executes a probe. Before 322 * executing the probe, DTrace blocks re-scheduling and sets 323 * a flag in its per-cpu flags to indicate that it doesn't 324 * want to fault. On returning from the probe, the no-fault 325 * flag is cleared and finally re-scheduling is enabled. 326 * 327 * If the DTrace kernel module has registered a trap handler, 328 * call it and if it returns non-zero, assume that it has 329 * handled the trap and modified the trap frame so that this 330 * function can return normally. 331 */ 332 if ((type == T_PROTFLT || type == T_PAGEFLT) && 333 dtrace_trap_func != NULL) { 334 if ((*dtrace_trap_func)(frame, type)) { 335 return; 336 } 337 } 338#endif 339 340 switch (type) { 341 342 default: 343 we_re_toast: 344 trap_print(frame, l); 345 346 if (kdb_trap(type, 0, frame)) 347 return; 348 if (kgdb_trap(type, frame)) 349 return; 350 /* 351 * If this is a breakpoint, don't panic if we're not connected. 352 */ 353 if (type == T_BPTFLT && kgdb_disconnected()) { 354 printf("kgdb: ignored %s\n", trap_type[type]); 355 return; 356 } 357 panic("trap"); 358 /*NOTREACHED*/ 359 360 case T_PROTFLT: 361 case T_SEGNPFLT: 362 case T_ALIGNFLT: 363 case T_STKFLT: 364 case T_TSSFLT: 365 if (p == NULL) 366 goto we_re_toast; 367 /* Check for copyin/copyout fault. */ 368 onfault = onfault_handler(pcb, frame); 369 if (onfault != NULL) { 370copyefault: 371 error = EFAULT; 372copyfault: 373 frame->tf_eip = (uintptr_t)onfault; 374 frame->tf_eax = error; 375 return; 376 } 377 378 /* 379 * Check for failure during return to user mode. 380 * This can happen loading invalid values into the segment 381 * registers, or during the 'iret' itself. 382 * 383 * We do this by looking at the instruction we faulted on. 384 * The specific instructions we recognize only happen when 385 * returning from a trap, syscall, or interrupt. 386 */ 387 388kernelfault: 389 KSI_INIT_TRAP(&ksi); 390 ksi.ksi_signo = SIGSEGV; 391 ksi.ksi_code = SEGV_ACCERR; 392 ksi.ksi_trap = type; 393 394 switch (*(u_char *)frame->tf_eip) { 395 case 0xcf: /* iret */ 396 /* 397 * The 'iret' instruction faulted, so we have the 398 * 'user' registers saved after the kernel %eip:%cs:%fl 399 * of the 'iret' and below that the user %eip:%cs:%fl 400 * the 'iret' was processing. 401 * We must delete the 3 words of kernel return address 402 * from the stack to generate a normal stack frame 403 * (eg for sending a SIGSEGV). 404 */ 405 vframe = (void *)((int *)frame + 3); 406 if (KERNELMODE(vframe->tf_cs)) 407 goto we_re_toast; 408 memmove(vframe, frame, 409 offsetof(struct trapframe, tf_eip)); 410 /* Set the faulting address to the user %eip */ 411 ksi.ksi_addr = (void *)vframe->tf_eip; 412 break; 413 case 0x8e: 414 switch (*(uint32_t *)frame->tf_eip) { 415 case 0x8e242c8e: /* mov (%esp,%gs), then */ 416 case 0x0424648e: /* mov 0x4(%esp),%fs */ 417 case 0x0824448e: /* mov 0x8(%esp),%es */ 418 case 0x0c245c8e: /* mov 0xc(%esp),%ds */ 419 break; 420 default: 421 goto we_re_toast; 422 } 423 /* 424 * We faulted loading one of the user segment registers. 425 * The stack frame containing the user registers is 426 * still valid and is just below the %eip:%cs:%fl of 427 * the kernel fault frame. 428 */ 429 vframe = (void *)(&frame->tf_eflags + 1); 430 if (KERNELMODE(vframe->tf_cs)) 431 goto we_re_toast; 432 /* There is no valid address for the fault */ 433 break; 434 default: 435 goto we_re_toast; 436 } 437 /* 438 * We might have faulted trying to execute the 439 * trampoline for a local (nested) signal handler. 440 * Only generate SIGSEGV if the user %cs isn't changed. 441 * (This is only strictly necessary in the 'iret' case.) 442 */ 443 if (!pmap_exec_fixup(&p->p_vmspace->vm_map, vframe, pcb)) { 444 /* Save outer frame for any signal return */ 445 l->l_md.md_regs = vframe; 446 SIGDEBUG(vframe, &ksi, error); 447 (*p->p_emul->e_trapsignal)(l, &ksi); 448 } 449 /* Return to user by reloading the user frame */ 450 trap_return_fault_return(vframe); 451 /* NOTREACHED */ 452 453 case T_PROTFLT|T_USER: /* protection fault */ 454#if defined(COMPAT_10) || defined(COMPAT_NOMID) 455 { 456#define LCALLSZ 7 457 /* Check for the osyscall lcall instruction. */ 458 if (frame->tf_eip < VM_MAXUSER_ADDRESS - LCALLSZ && 459 x86_cpu_is_lcall((const void *)frame->tf_eip) == 0) { 460 461 /* Advance past the lcall. */ 462 frame->tf_eip += LCALLSZ; 463 464 /* Do the syscall. */ 465 p->p_md.md_syscall(frame); 466 goto out; 467 } 468 } 469#endif 470 /* FALLTHROUGH */ 471 case T_TSSFLT|T_USER: 472 case T_SEGNPFLT|T_USER: 473 case T_STKFLT|T_USER: 474 case T_ALIGNFLT|T_USER: 475 KSI_INIT_TRAP(&ksi); 476 477 ksi.ksi_addr = (void *)rcr2(); 478 switch (type) { 479 case T_SEGNPFLT|T_USER: 480 case T_STKFLT|T_USER: 481 ksi.ksi_signo = SIGBUS; 482 ksi.ksi_code = BUS_ADRERR; 483 break; 484 case T_TSSFLT|T_USER: 485 ksi.ksi_signo = SIGBUS; 486 ksi.ksi_code = BUS_OBJERR; 487 break; 488 case T_ALIGNFLT|T_USER: 489 ksi.ksi_signo = SIGBUS; 490 ksi.ksi_code = BUS_ADRALN; 491 break; 492 case T_PROTFLT|T_USER: 493 /* 494 * If pmap_exec_fixup does something, 495 * let's retry the trap. 496 */ 497 if (pmap_exec_fixup(&p->p_vmspace->vm_map, frame, pcb)){ 498 goto out; 499 } 500 ksi.ksi_signo = SIGSEGV; 501 ksi.ksi_code = SEGV_ACCERR; 502 break; 503 default: 504 KASSERT(0); 505 break; 506 } 507 goto trapsignal; 508 509 case T_PRIVINFLT|T_USER: /* privileged instruction fault */ 510 case T_FPOPFLT|T_USER: /* coprocessor operand fault */ 511 KSI_INIT_TRAP(&ksi); 512 ksi.ksi_signo = SIGILL; 513 ksi.ksi_addr = (void *) frame->tf_eip; 514 switch (type) { 515 case T_PRIVINFLT|T_USER: 516 ksi.ksi_code = ILL_PRVOPC; 517 break; 518 case T_FPOPFLT|T_USER: 519 ksi.ksi_code = ILL_COPROC; 520 break; 521 default: 522 ksi.ksi_code = 0; 523 break; 524 } 525 goto trapsignal; 526 527 case T_ASTFLT|T_USER: 528 /* Allow process switch. */ 529 //curcpu()->ci_data.cpu_nast++; 530 if (l->l_pflag & LP_OWEUPC) { 531 l->l_pflag &= ~LP_OWEUPC; 532 ADDUPROF(l); 533 } 534 goto out; 535 536 case T_BOUND|T_USER: 537 case T_OFLOW|T_USER: 538 case T_DIVIDE|T_USER: 539 KSI_INIT_TRAP(&ksi); 540 ksi.ksi_signo = SIGFPE; 541 ksi.ksi_addr = (void *)frame->tf_eip; 542 switch (type) { 543 case T_BOUND|T_USER: 544 ksi.ksi_code = FPE_FLTSUB; 545 break; 546 case T_OFLOW|T_USER: 547 ksi.ksi_code = FPE_INTOVF; 548 break; 549 case T_DIVIDE|T_USER: 550 ksi.ksi_code = FPE_INTDIV; 551 break; 552 default: 553 ksi.ksi_code = 0; 554 break; 555 } 556 goto trapsignal; 557 558 case T_PAGEFLT: 559 /* Allow page faults in kernel mode. */ 560 if (__predict_false(l == NULL)) 561 goto we_re_toast; 562 563 onfault = pcb->pcb_onfault; 564 if (onfault == return_address_fault) { 565 goto copyefault; 566 } 567 if (cpu_intr_p() || (l->l_pflag & LP_INTR) != 0) { 568 goto we_re_toast; 569 } 570 571 cr2 = rcr2(); 572 573 if (frame->tf_err & PGEX_I) { 574 /* SMEP might have brought us here */ 575 if (cr2 > VM_MIN_ADDRESS && cr2 <= VM_MAXUSER_ADDRESS) { 576 printf("prevented execution of %p (SMEP)\n", 577 (void *)cr2); 578 goto we_re_toast; 579 } 580 } 581 582 if ((frame->tf_err & PGEX_P) && 583 cr2 < VM_MAXUSER_ADDRESS) { 584 /* SMAP might have brought us here */ 585 if (onfault_handler(pcb, frame) == NULL) { 586 printf("prevented access to %p (SMAP)\n", 587 (void *)cr2); 588 goto we_re_toast; 589 } 590 } 591 592 goto faultcommon; 593 594 case T_PAGEFLT|T_USER: { /* page fault */ 595 register vaddr_t va; 596 register struct vmspace *vm; 597 register struct vm_map *map; 598 vm_prot_t ftype; 599 extern struct vm_map *kernel_map; 600 601 cr2 = rcr2(); 602faultcommon: 603 vm = p->p_vmspace; 604 if (__predict_false(vm == NULL)) { 605 goto we_re_toast; 606 } 607 pcb->pcb_cr2 = cr2; 608 va = trunc_page((vaddr_t)cr2); 609 /* 610 * It is only a kernel address space fault iff: 611 * 1. (type & T_USER) == 0 and 612 * 2. pcb_onfault not set or 613 * 3. pcb_onfault set but supervisor space fault 614 * The last can occur during an exec() copyin where the 615 * argument space is lazy-allocated. 616 */ 617 if (type == T_PAGEFLT && va >= VM_MIN_KERNEL_ADDRESS) 618 map = kernel_map; 619 else 620 map = &vm->vm_map; 621 if (frame->tf_err & PGEX_W) 622 ftype = VM_PROT_WRITE; 623 else if (frame->tf_err & PGEX_I) 624 ftype = VM_PROT_EXECUTE; 625 else 626 ftype = VM_PROT_READ; 627 628#ifdef DIAGNOSTIC 629 if (map == kernel_map && va == 0) { 630 printf("trap: bad kernel access at %lx\n", va); 631 goto we_re_toast; 632 } 633#endif 634 /* Fault the original page in. */ 635 onfault = pcb->pcb_onfault; 636 pcb->pcb_onfault = NULL; 637 error = uvm_fault(map, va, ftype); 638 pcb->pcb_onfault = onfault; 639 if (error == 0) { 640 if (map != kernel_map && (void *)va >= vm->vm_maxsaddr) 641 uvm_grow(p, va); 642 643 pfail = false; 644 while (type == T_PAGEFLT) { 645 /* 646 * we need to switch pmap now if we're in 647 * the middle of copyin/out. 648 * 649 * but we don't need to do so for kcopy as 650 * it never touch userspace. 651 */ 652 kpreempt_disable(); 653 if (curcpu()->ci_want_pmapload) { 654 onfault = onfault_handler(pcb, frame); 655 if (onfault != kcopy_fault) { 656 pmap_load(); 657 } 658 } 659 /* 660 * We need to keep the pmap loaded and 661 * so avoid being preempted until back 662 * into the copy functions. Disable 663 * interrupts at the hardware level before 664 * re-enabling preemption. Interrupts 665 * will be re-enabled by 'iret' when 666 * returning back out of the trap stub. 667 * They'll only be re-enabled when the 668 * program counter is once again in 669 * the copy functions, and so visible 670 * to cpu_kpreempt_exit(). 671 */ 672#ifndef XENPV 673 x86_disable_intr(); 674#endif 675 l->l_nopreempt--; 676 if (l->l_nopreempt > 0 || !l->l_dopreempt || 677 pfail) { 678 return; 679 } 680#ifndef XENPV 681 x86_enable_intr(); 682#endif 683 /* 684 * If preemption fails for some reason, 685 * don't retry it. The conditions won't 686 * change under our nose. 687 */ 688 pfail = kpreempt(0); 689 } 690 goto out; 691 } 692 693 if (type == T_PAGEFLT) { 694 onfault = onfault_handler(pcb, frame); 695 if (onfault != NULL) 696 goto copyfault; 697 printf("uvm_fault(%p, %#lx, %d) -> %#x\n", 698 map, va, ftype, error); 699 goto kernelfault; 700 } 701 702 KSI_INIT_TRAP(&ksi); 703 ksi.ksi_trap = type & ~T_USER; 704 ksi.ksi_addr = (void *)cr2; 705 switch (error) { 706 case EINVAL: 707 ksi.ksi_signo = SIGBUS; 708 ksi.ksi_code = BUS_ADRERR; 709 break; 710 case EACCES: 711 ksi.ksi_signo = SIGSEGV; 712 ksi.ksi_code = SEGV_ACCERR; 713 error = EFAULT; 714 break; 715 case ENOMEM: 716 ksi.ksi_signo = SIGKILL; 717 printf("UVM: pid %d.%d (%s), uid %d killed: " 718 "out of swap\n", p->p_pid, l->l_lid, p->p_comm, 719 l->l_cred ? kauth_cred_geteuid(l->l_cred) : -1); 720 break; 721 default: 722 ksi.ksi_signo = SIGSEGV; 723 ksi.ksi_code = SEGV_MAPERR; 724 break; 725 } 726 727 SIGDEBUG(frame, &ksi, error); 728 (*p->p_emul->e_trapsignal)(l, &ksi); 729 break; 730 } 731 732 case T_TRCTRAP: 733 /* 734 * Ignore debug register trace traps due to 735 * accesses in the user's address space, which 736 * can happen under several conditions such as 737 * if a user sets a watchpoint on a buffer and 738 * then passes that buffer to a system call. 739 * We still want to get TRCTRAPS for addresses 740 * in kernel space because that is useful when 741 * debugging the kernel. 742 */ 743 if (x86_dbregs_user_trap()) 744 break; 745 746 goto we_re_toast; 747 748 case T_BPTFLT|T_USER: /* bpt instruction fault */ 749 case T_TRCTRAP|T_USER: /* trace trap */ 750 /* 751 * Don't go single-stepping into a RAS. 752 */ 753 if (p->p_raslist == NULL || 754 (ras_lookup(p, (void *)frame->tf_eip) == (void *)-1)) { 755 KSI_INIT_TRAP(&ksi); 756 ksi.ksi_signo = SIGTRAP; 757 ksi.ksi_trap = type & ~T_USER; 758 if (x86_dbregs_user_trap()) { 759 x86_dbregs_store_dr6(l); 760 ksi.ksi_code = TRAP_DBREG; 761 } else if (type == (T_BPTFLT|T_USER)) 762 ksi.ksi_code = TRAP_BRKPT; 763 else 764 ksi.ksi_code = TRAP_TRACE; 765 ksi.ksi_addr = (void *)frame->tf_eip; 766 SIGDEBUG(frame, &ksi, error); 767 (*p->p_emul->e_trapsignal)(l, &ksi); 768 } 769 break; 770 771 case T_NMI: 772 if (nmi_dispatch(frame)) 773 return; 774 /* NMI can be hooked up to a pushbutton for debugging */ 775 if (kgdb_trap(type, frame)) 776 return; 777 if (kdb_trap(type, 0, frame)) 778 return; 779 /* machine/parity/power fail/"kitchen sink" faults */ 780#if NMCA > 0 781 mca_nmi(); 782#endif 783 x86_nmi(); 784 } 785 786 if ((type & T_USER) == 0) 787 return; 788out: 789 userret(l); 790 return; 791trapsignal: 792 ksi.ksi_trap = type & ~T_USER; 793 SIGDEBUG(frame, &ksi, error); 794 (*p->p_emul->e_trapsignal)(l, &ksi); 795 userret(l); 796} 797 798/* 799 * startlwp: start of a new LWP. 800 */ 801void 802startlwp(void *arg) 803{ 804 ucontext_t *uc = arg; 805 lwp_t *l = curlwp; 806 int error __diagused; 807 808 error = cpu_setmcontext(l, &uc->uc_mcontext, uc->uc_flags); 809 KASSERT(error == 0); 810 811 kmem_free(uc, sizeof(ucontext_t)); 812 userret(l); 813} 814 815#ifdef TRAP_SIGDEBUG 816static void 817frame_dump(const struct trapframe *tf, const struct pcb *pcb) 818{ 819 uint64_t fsd, gsd; 820 821 printf("trapframe %p\n", tf); 822 printf("eip 0x%08x esp 0x%08x efl 0x%08x\n", 823 tf->tf_eip, tf->tf_esp, tf->tf_eflags); 824 printf("edi 0x%08x esi 0x%08x edx 0x%08x\n", 825 tf->tf_edi, tf->tf_esi, tf->tf_edx); 826 printf("ecx 0x%08x\n", 827 tf->tf_ecx); 828 printf("ebp 0x%08x ebx 0x%08x eax 0x%08x\n", 829 tf->tf_ebp, tf->tf_ebx, tf->tf_eax); 830 printf("cs 0x%04x ds 0x%04x es 0x%04x " 831 "fs 0x%04x gs 0x%04x ss 0x%04x\n", 832 tf->tf_cs & 0xffff, tf->tf_ds & 0xffff, tf->tf_es & 0xffff, 833 tf->tf_fs & 0xffff, tf->tf_gs & 0xffff, tf->tf_ss & 0xffff); 834 memcpy(&fsd, &pcb->pcb_fsd, sizeof(fsd)); 835 memcpy(&gsd, &pcb->pcb_gsd, sizeof(gsd)); 836 printf("fsbase 0x%016llx gsbase 0x%016llx\n", fsd, gsd); 837 printf("\n"); 838 hexdump(printf, "Stack dump", tf, 256); 839} 840 841static void 842sigdebug(const struct trapframe *tf, const ksiginfo_t *ksi, int e) 843{ 844 struct lwp *l = curlwp; 845 struct proc *p = l->l_proc; 846 847 printf("pid %d.%d (%s): signal %d code=%d (trap %x) " 848 "@eip %#x addr %#x error=%d\n", 849 p->p_pid, l->l_lid, p->p_comm, ksi->ksi_signo, ksi->ksi_code, 850 tf->tf_trapno, tf->tf_eip, rcr2(), e); 851 frame_dump(tf, lwp_getpcb(l)); 852} 853#endif 854