HardenedBSD/sys/kern/subr_kdb.c

676 lines
14 KiB
C
Raw Normal View History

/*-
* Copyright (c) 2004 The FreeBSD Project
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHORS ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_kdb.h"
#include "opt_stack.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/cons.h>
#include <sys/kdb.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/pcpu.h>
#include <sys/proc.h>
#include <sys/sbuf.h>
#include <sys/smp.h>
#include <sys/stack.h>
#include <sys/sysctl.h>
#include <machine/kdb.h>
#include <machine/pcb.h>
#ifdef SMP
#include <machine/smp.h>
#endif
int kdb_active = 0;
static void *kdb_jmpbufp = NULL;
struct kdb_dbbe *kdb_dbbe = NULL;
static struct pcb kdb_pcb;
struct pcb *kdb_thrctx = NULL;
struct thread *kdb_thread = NULL;
struct trapframe *kdb_frame = NULL;
#ifdef BREAK_TO_DEBUGGER
#define KDB_BREAK_TO_DEBUGGER 1
#else
#define KDB_BREAK_TO_DEBUGGER 0
#endif
#ifdef ALT_BREAK_TO_DEBUGGER
#define KDB_ALT_BREAK_TO_DEBUGGER 1
#else
#define KDB_ALT_BREAK_TO_DEBUGGER 0
#endif
static int kdb_break_to_debugger = KDB_BREAK_TO_DEBUGGER;
static int kdb_alt_break_to_debugger = KDB_ALT_BREAK_TO_DEBUGGER;
KDB_BACKEND(null, NULL, NULL, NULL, NULL);
SET_DECLARE(kdb_dbbe_set, struct kdb_dbbe);
static int kdb_sysctl_available(SYSCTL_HANDLER_ARGS);
static int kdb_sysctl_current(SYSCTL_HANDLER_ARGS);
static int kdb_sysctl_enter(SYSCTL_HANDLER_ARGS);
static int kdb_sysctl_panic(SYSCTL_HANDLER_ARGS);
static int kdb_sysctl_trap(SYSCTL_HANDLER_ARGS);
static int kdb_sysctl_trap_code(SYSCTL_HANDLER_ARGS);
static SYSCTL_NODE(_debug, OID_AUTO, kdb, CTLFLAG_RW, NULL, "KDB nodes");
SYSCTL_PROC(_debug_kdb, OID_AUTO, available, CTLTYPE_STRING | CTLFLAG_RD, NULL,
0, kdb_sysctl_available, "A", "list of available KDB backends");
SYSCTL_PROC(_debug_kdb, OID_AUTO, current, CTLTYPE_STRING | CTLFLAG_RW, NULL,
0, kdb_sysctl_current, "A", "currently selected KDB backend");
SYSCTL_PROC(_debug_kdb, OID_AUTO, enter,
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE, NULL, 0,
kdb_sysctl_enter, "I", "set to enter the debugger");
SYSCTL_PROC(_debug_kdb, OID_AUTO, panic,
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE, NULL, 0,
kdb_sysctl_panic, "I", "set to panic the kernel");
SYSCTL_PROC(_debug_kdb, OID_AUTO, trap,
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE, NULL, 0,
kdb_sysctl_trap, "I", "set to cause a page fault via data access");
SYSCTL_PROC(_debug_kdb, OID_AUTO, trap_code,
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE, NULL, 0,
kdb_sysctl_trap_code, "I", "set to cause a page fault via code access");
SYSCTL_INT(_debug_kdb, OID_AUTO, break_to_debugger,
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_TUN | CTLFLAG_SECURE,
&kdb_break_to_debugger, 0, "Enable break to debugger");
TUNABLE_INT("debug.kdb.break_to_debugger", &kdb_break_to_debugger);
SYSCTL_INT(_debug_kdb, OID_AUTO, alt_break_to_debugger,
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_TUN | CTLFLAG_SECURE,
&kdb_alt_break_to_debugger, 0, "Enable alternative break to debugger");
TUNABLE_INT("debug.kdb.alt_break_to_debugger", &kdb_alt_break_to_debugger);
/*
* Flag to indicate to debuggers why the debugger was entered.
*/
const char * volatile kdb_why = KDB_WHY_UNSET;
static int
kdb_sysctl_available(SYSCTL_HANDLER_ARGS)
{
struct kdb_dbbe **iter;
struct sbuf sbuf;
int error;
sbuf_new_for_sysctl(&sbuf, NULL, 64, req);
SET_FOREACH(iter, kdb_dbbe_set) {
if ((*iter)->dbbe_active == 0)
sbuf_printf(&sbuf, "%s ", (*iter)->dbbe_name);
}
error = sbuf_finish(&sbuf);
sbuf_delete(&sbuf);
return (error);
}
static int
kdb_sysctl_current(SYSCTL_HANDLER_ARGS)
{
char buf[16];
int error;
if (kdb_dbbe != NULL)
strlcpy(buf, kdb_dbbe->dbbe_name, sizeof(buf));
else
*buf = '\0';
error = sysctl_handle_string(oidp, buf, sizeof(buf), req);
if (error != 0 || req->newptr == NULL)
return (error);
if (kdb_active)
return (EBUSY);
return (kdb_dbbe_select(buf));
}
static int
kdb_sysctl_enter(SYSCTL_HANDLER_ARGS)
{
int error, i;
error = sysctl_wire_old_buffer(req, sizeof(int));
if (error == 0) {
i = 0;
error = sysctl_handle_int(oidp, &i, 0, req);
}
if (error != 0 || req->newptr == NULL)
return (error);
if (kdb_active)
return (EBUSY);
kdb_enter(KDB_WHY_SYSCTL, "sysctl debug.kdb.enter");
return (0);
}
static int
kdb_sysctl_panic(SYSCTL_HANDLER_ARGS)
{
int error, i;
error = sysctl_wire_old_buffer(req, sizeof(int));
if (error == 0) {
i = 0;
error = sysctl_handle_int(oidp, &i, 0, req);
}
if (error != 0 || req->newptr == NULL)
return (error);
panic("kdb_sysctl_panic");
return (0);
}
static int
kdb_sysctl_trap(SYSCTL_HANDLER_ARGS)
{
int error, i;
int *addr = (int *)0x10;
error = sysctl_wire_old_buffer(req, sizeof(int));
if (error == 0) {
i = 0;
error = sysctl_handle_int(oidp, &i, 0, req);
}
if (error != 0 || req->newptr == NULL)
return (error);
return (*addr);
}
static int
kdb_sysctl_trap_code(SYSCTL_HANDLER_ARGS)
{
int error, i;
void (*fp)(u_int, u_int, u_int) = (void *)0xdeadc0de;
error = sysctl_wire_old_buffer(req, sizeof(int));
if (error == 0) {
i = 0;
error = sysctl_handle_int(oidp, &i, 0, req);
}
if (error != 0 || req->newptr == NULL)
return (error);
(*fp)(0x11111111, 0x22222222, 0x33333333);
return (0);
}
void
kdb_panic(const char *msg)
{
printf("KDB: panic\n");
panic("%s", msg);
}
void
kdb_reboot(void)
{
printf("KDB: reboot requested\n");
shutdown_nice(0);
}
/*
* Solaris implements a new BREAK which is initiated by a character sequence
* CR ~ ^b which is similar to a familiar pattern used on Sun servers by the
* Remote Console.
*
* Note that this function may be called from almost anywhere, with interrupts
* disabled and with unknown locks held, so it must not access data other than
* its arguments. Its up to the caller to ensure that the state variable is
* consistent.
*/
#define KEY_CR 13 /* CR '\r' */
#define KEY_TILDE 126 /* ~ */
#define KEY_CRTLB 2 /* ^B */
#define KEY_CRTLP 16 /* ^P */
#define KEY_CRTLR 18 /* ^R */
2011-06-15 09:20:22 +02:00
/* States of th KDB "alternate break sequence" detecting state machine. */
enum {
KDB_ALT_BREAK_SEEN_NONE,
KDB_ALT_BREAK_SEEN_CR,
KDB_ALT_BREAK_SEEN_CR_TILDE,
};
int
kdb_break(void)
{
if (!kdb_break_to_debugger)
return (0);
kdb_enter(KDB_WHY_BREAK, "Break to debugger");
return (KDB_REQ_DEBUGGER);
}
static int
kdb_alt_break_state(int key, int *state)
{
int brk;
2011-06-15 09:20:22 +02:00
/* All states transition to KDB_ALT_BREAK_SEEN_CR on a CR. */
if (key == KEY_CR) {
*state = KDB_ALT_BREAK_SEEN_CR;
return (0);
}
brk = 0;
switch (*state) {
2011-06-15 09:20:22 +02:00
case KDB_ALT_BREAK_SEEN_CR:
*state = KDB_ALT_BREAK_SEEN_NONE;
if (key == KEY_TILDE)
2011-06-15 09:20:22 +02:00
*state = KDB_ALT_BREAK_SEEN_CR_TILDE;
break;
2011-06-15 09:20:22 +02:00
case KDB_ALT_BREAK_SEEN_CR_TILDE:
*state = KDB_ALT_BREAK_SEEN_NONE;
if (key == KEY_CRTLB)
brk = KDB_REQ_DEBUGGER;
else if (key == KEY_CRTLP)
brk = KDB_REQ_PANIC;
else if (key == KEY_CRTLR)
brk = KDB_REQ_REBOOT;
2011-06-15 09:20:22 +02:00
break;
case KDB_ALT_BREAK_SEEN_NONE:
default:
*state = KDB_ALT_BREAK_SEEN_NONE;
break;
}
return (brk);
}
static int
kdb_alt_break_internal(int key, int *state, int force_gdb)
{
int brk;
if (!kdb_alt_break_to_debugger)
return (0);
brk = kdb_alt_break_state(key, state);
switch (brk) {
case KDB_REQ_DEBUGGER:
if (force_gdb)
kdb_dbbe_select("gdb");
kdb_enter(KDB_WHY_BREAK, "Break to debugger");
break;
case KDB_REQ_PANIC:
if (force_gdb)
kdb_dbbe_select("gdb");
kdb_panic("Panic sequence on console");
break;
case KDB_REQ_REBOOT:
kdb_reboot();
break;
}
return (0);
}
int
kdb_alt_break(int key, int *state)
{
return (kdb_alt_break_internal(key, state, 0));
}
/*
* This variation on kdb_alt_break() is used only by dcons, which has its own
* configuration flag to force GDB use regardless of the global KDB
* configuration.
*/
int
kdb_alt_break_gdb(int key, int *state)
{
return (kdb_alt_break_internal(key, state, 1));
}
/*
* Print a backtrace of the calling thread. The backtrace is generated by
* the selected debugger, provided it supports backtraces. If no debugger
* is selected or the current debugger does not support backtraces, this
* function silently returns.
*/
void
kdb_backtrace(void)
{
if (kdb_dbbe != NULL && kdb_dbbe->dbbe_trace != NULL) {
printf("KDB: stack backtrace:\n");
kdb_dbbe->dbbe_trace();
}
#ifdef STACK
else {
struct stack st;
printf("KDB: stack backtrace:\n");
stack_zero(&st);
stack_save(&st);
stack_print_ddb(&st);
}
#endif
}
/*
* Similar to kdb_backtrace() except that it prints a backtrace of an
* arbitrary thread rather than the calling thread.
*/
void
kdb_backtrace_thread(struct thread *td)
{
if (kdb_dbbe != NULL && kdb_dbbe->dbbe_trace_thread != NULL) {
printf("KDB: stack backtrace of thread %d:\n", td->td_tid);
kdb_dbbe->dbbe_trace_thread(td);
}
#ifdef STACK
else {
struct stack st;
printf("KDB: stack backtrace of thread %d:\n", td->td_tid);
stack_zero(&st);
stack_save_td(&st, td);
stack_print_ddb(&st);
}
#endif
}
/*
* Set/change the current backend.
*/
int
kdb_dbbe_select(const char *name)
{
struct kdb_dbbe *be, **iter;
SET_FOREACH(iter, kdb_dbbe_set) {
be = *iter;
if (be->dbbe_active == 0 && strcmp(be->dbbe_name, name) == 0) {
kdb_dbbe = be;
return (0);
}
}
return (EINVAL);
}
/*
* Enter the currently selected debugger. If a message has been provided,
* it is printed first. If the debugger does not support the enter method,
* it is entered by using breakpoint(), which enters the debugger through
* kdb_trap(). The 'why' argument will contain a more mechanically usable
* string than 'msg', and is relied upon by DDB scripting to identify the
* reason for entering the debugger so that the right script can be run.
*/
void
kdb_enter(const char *why, const char *msg)
{
if (kdb_dbbe != NULL && kdb_active == 0) {
if (msg != NULL)
printf("KDB: enter: %s\n", msg);
kdb_why = why;
breakpoint();
kdb_why = KDB_WHY_UNSET;
}
}
/*
* Initialize the kernel debugger interface.
*/
void
kdb_init(void)
{
struct kdb_dbbe *be, **iter;
int cur_pri, pri;
kdb_active = 0;
kdb_dbbe = NULL;
cur_pri = -1;
SET_FOREACH(iter, kdb_dbbe_set) {
be = *iter;
pri = (be->dbbe_init != NULL) ? be->dbbe_init() : -1;
be->dbbe_active = (pri >= 0) ? 0 : -1;
if (pri > cur_pri) {
cur_pri = pri;
kdb_dbbe = be;
}
}
if (kdb_dbbe != NULL) {
printf("KDB: debugger backends:");
SET_FOREACH(iter, kdb_dbbe_set) {
be = *iter;
if (be->dbbe_active == 0)
printf(" %s", be->dbbe_name);
}
printf("\n");
printf("KDB: current backend: %s\n",
kdb_dbbe->dbbe_name);
}
}
/*
* Handle contexts.
*/
void *
kdb_jmpbuf(jmp_buf new)
{
void *old;
old = kdb_jmpbufp;
kdb_jmpbufp = new;
return (old);
}
void
kdb_reenter(void)
{
if (!kdb_active || kdb_jmpbufp == NULL)
return;
longjmp(kdb_jmpbufp, 1);
/* NOTREACHED */
}
/*
* Thread related support functions.
*/
struct pcb *
kdb_thr_ctx(struct thread *thr)
{
#if defined(SMP) && defined(KDB_STOPPEDPCB)
struct pcpu *pc;
#endif
if (thr == curthread)
return (&kdb_pcb);
#if defined(SMP) && defined(KDB_STOPPEDPCB)
STAILQ_FOREACH(pc, &cpuhead, pc_allcpu) {
Commit the support for removing cpumask_t and replacing it directly with cpuset_t objects. That is going to offer the underlying support for a simple bump of MAXCPU and then support for number of cpus > 32 (as it is today). Right now, cpumask_t is an int, 32 bits on all our supported architecture. cpumask_t on the other side is implemented as an array of longs, and easilly extendible by definition. The architectures touched by this commit are the following: - amd64 - i386 - pc98 - arm - ia64 - XEN while the others are still missing. Userland is believed to be fully converted with the changes contained here. Some technical notes: - This commit may be considered an ABI nop for all the architectures different from amd64 and ia64 (and sparc64 in the future) - per-cpu members, which are now converted to cpuset_t, needs to be accessed avoiding migration, because the size of cpuset_t should be considered unknown - size of cpuset_t objects is different from kernel and userland (this is primirally done in order to leave some more space in userland to cope with KBI extensions). If you need to access kernel cpuset_t from the userland please refer to example in this patch on how to do that correctly (kgdb may be a good source, for example). - Support for other architectures is going to be added soon - Only MAXCPU for amd64 is bumped now The patch has been tested by sbruno and Nicholas Esborn on opteron 4 x 12 pack CPUs. More testing on big SMP is expected to came soon. pluknet tested the patch with his 8-ways on both amd64 and i386. Tested by: pluknet, sbruno, gianni, Nicholas Esborn Reviewed by: jeff, jhb, sbruno
2011-05-05 16:39:14 +02:00
if (pc->pc_curthread == thr &&
CPU_ISSET(pc->pc_cpuid, &stopped_cpus))
return (KDB_STOPPEDPCB(pc));
}
#endif
return (thr->td_pcb);
}
struct thread *
kdb_thr_first(void)
{
struct proc *p;
struct thread *thr;
p = LIST_FIRST(&allproc);
while (p != NULL) {
if (p->p_flag & P_INMEM) {
thr = FIRST_THREAD_IN_PROC(p);
if (thr != NULL)
return (thr);
}
p = LIST_NEXT(p, p_list);
}
return (NULL);
}
struct thread *
kdb_thr_from_pid(pid_t pid)
{
struct proc *p;
p = LIST_FIRST(&allproc);
while (p != NULL) {
if (p->p_flag & P_INMEM && p->p_pid == pid)
return (FIRST_THREAD_IN_PROC(p));
p = LIST_NEXT(p, p_list);
}
return (NULL);
}
struct thread *
kdb_thr_lookup(lwpid_t tid)
{
struct thread *thr;
thr = kdb_thr_first();
while (thr != NULL && thr->td_tid != tid)
thr = kdb_thr_next(thr);
return (thr);
}
struct thread *
kdb_thr_next(struct thread *thr)
{
struct proc *p;
p = thr->td_proc;
thr = TAILQ_NEXT(thr, td_plist);
do {
if (thr != NULL)
return (thr);
p = LIST_NEXT(p, p_list);
if (p != NULL && (p->p_flag & P_INMEM))
thr = FIRST_THREAD_IN_PROC(p);
} while (p != NULL);
return (NULL);
}
int
kdb_thr_select(struct thread *thr)
{
if (thr == NULL)
return (EINVAL);
kdb_thread = thr;
kdb_thrctx = kdb_thr_ctx(thr);
return (0);
}
/*
* Enter the debugger due to a trap.
*/
int
kdb_trap(int type, int code, struct trapframe *tf)
{
#ifdef SMP
cpuset_t other_cpus;
#endif
2011-06-27 15:02:23 +02:00
struct kdb_dbbe *be;
register_t intr;
int handled;
panic: add a switch and infrastructure for stopping other CPUs in SMP case Historical behavior of letting other CPUs merily go on is a default for time being. The new behavior can be switched on via kern.stop_scheduler_on_panic tunable and sysctl. Stopping of the CPUs has (at least) the following benefits: - more of the system state at panic time is preserved intact - threads and interrupts do not interfere with dumping of the system state Only one thread runs uninterrupted after panic if stop_scheduler_on_panic is set. That thread might call code that is also used in normal context and that code might use locks to prevent concurrent execution of certain parts. Those locks might be held by the stopped threads and would never be released. To work around this issue, it was decided that instead of explicit checks for panic context, we would rather put those checks inside the locking primitives. This change has substantial portions written and re-written by attilio and kib at various times. Other changes are heavily based on the ideas and patches submitted by jhb and mdf. bde has provided many insights into the details and history of the current code. The new behavior may cause problems for systems that use a USB keyboard for interfacing with system console. This is because of some unusual locking patterns in the ukbd code which have to be used because on one hand ukbd is below syscons, but on the other hand it has to interface with other usb code that uses regular mutexes/Giant for its concurrency protection. Dumping to USB-connected disks may also be affected. PR: amd64/139614 (at least) In cooperation with: attilio, jhb, kib, mdf Discussed with: arch@, bde Tested by: Eugene Grosbein <eugen@grosbein.net>, gnn, Steven Hartland <killing@multiplay.co.uk>, glebius, Andrew Boyer <aboyer@averesystems.com> (various versions of the patch) MFC after: 3 months (or never)
2011-12-11 22:02:01 +01:00
#ifdef SMP
int did_stop_cpus;
#endif
be = kdb_dbbe;
if (be == NULL || be->dbbe_trap == NULL)
return (0);
/* We reenter the debugger through kdb_reenter(). */
if (kdb_active)
return (0);
intr = intr_disable();
#ifdef SMP
panic: add a switch and infrastructure for stopping other CPUs in SMP case Historical behavior of letting other CPUs merily go on is a default for time being. The new behavior can be switched on via kern.stop_scheduler_on_panic tunable and sysctl. Stopping of the CPUs has (at least) the following benefits: - more of the system state at panic time is preserved intact - threads and interrupts do not interfere with dumping of the system state Only one thread runs uninterrupted after panic if stop_scheduler_on_panic is set. That thread might call code that is also used in normal context and that code might use locks to prevent concurrent execution of certain parts. Those locks might be held by the stopped threads and would never be released. To work around this issue, it was decided that instead of explicit checks for panic context, we would rather put those checks inside the locking primitives. This change has substantial portions written and re-written by attilio and kib at various times. Other changes are heavily based on the ideas and patches submitted by jhb and mdf. bde has provided many insights into the details and history of the current code. The new behavior may cause problems for systems that use a USB keyboard for interfacing with system console. This is because of some unusual locking patterns in the ukbd code which have to be used because on one hand ukbd is below syscons, but on the other hand it has to interface with other usb code that uses regular mutexes/Giant for its concurrency protection. Dumping to USB-connected disks may also be affected. PR: amd64/139614 (at least) In cooperation with: attilio, jhb, kib, mdf Discussed with: arch@, bde Tested by: Eugene Grosbein <eugen@grosbein.net>, gnn, Steven Hartland <killing@multiplay.co.uk>, glebius, Andrew Boyer <aboyer@averesystems.com> (various versions of the patch) MFC after: 3 months (or never)
2011-12-11 22:02:01 +01:00
if (!SCHEDULER_STOPPED()) {
other_cpus = all_cpus;
CPU_CLR(PCPU_GET(cpuid), &other_cpus);
stop_cpus_hard(other_cpus);
did_stop_cpus = 1;
} else
did_stop_cpus = 0;
#endif
kdb_active++;
kdb_frame = tf;
/* Let MD code do its thing first... */
kdb_cpu_trap(type, code);
makectx(tf, &kdb_pcb);
kdb_thr_select(curthread);
cngrab();
for (;;) {
handled = be->dbbe_trap(type, code);
if (be == kdb_dbbe)
break;
be = kdb_dbbe;
if (be == NULL || be->dbbe_trap == NULL)
break;
printf("Switching to %s back-end\n", be->dbbe_name);
}
cnungrab();
kdb_active--;
#ifdef SMP
panic: add a switch and infrastructure for stopping other CPUs in SMP case Historical behavior of letting other CPUs merily go on is a default for time being. The new behavior can be switched on via kern.stop_scheduler_on_panic tunable and sysctl. Stopping of the CPUs has (at least) the following benefits: - more of the system state at panic time is preserved intact - threads and interrupts do not interfere with dumping of the system state Only one thread runs uninterrupted after panic if stop_scheduler_on_panic is set. That thread might call code that is also used in normal context and that code might use locks to prevent concurrent execution of certain parts. Those locks might be held by the stopped threads and would never be released. To work around this issue, it was decided that instead of explicit checks for panic context, we would rather put those checks inside the locking primitives. This change has substantial portions written and re-written by attilio and kib at various times. Other changes are heavily based on the ideas and patches submitted by jhb and mdf. bde has provided many insights into the details and history of the current code. The new behavior may cause problems for systems that use a USB keyboard for interfacing with system console. This is because of some unusual locking patterns in the ukbd code which have to be used because on one hand ukbd is below syscons, but on the other hand it has to interface with other usb code that uses regular mutexes/Giant for its concurrency protection. Dumping to USB-connected disks may also be affected. PR: amd64/139614 (at least) In cooperation with: attilio, jhb, kib, mdf Discussed with: arch@, bde Tested by: Eugene Grosbein <eugen@grosbein.net>, gnn, Steven Hartland <killing@multiplay.co.uk>, glebius, Andrew Boyer <aboyer@averesystems.com> (various versions of the patch) MFC after: 3 months (or never)
2011-12-11 22:02:01 +01:00
if (did_stop_cpus)
restart_cpus(stopped_cpus);
#endif
intr_restore(intr);
return (handled);
}