fedora core 6 1.2949 + vserver 2.2.0
[linux-2.6.git] / arch / x86_64 / kernel / traps-xen.c
index 13c6d3d..adc678c 100644 (file)
 #include <linux/delay.h>
 #include <linux/spinlock.h>
 #include <linux/interrupt.h>
+#include <linux/kallsyms.h>
 #include <linux/module.h>
 #include <linux/moduleparam.h>
 #include <linux/nmi.h>
 #include <linux/kprobes.h>
 #include <linux/kexec.h>
 #include <linux/unwind.h>
+#include <linux/uaccess.h>
+#include <linux/bug.h>
 
 #include <asm/system.h>
-#include <asm/uaccess.h>
 #include <asm/io.h>
 #include <asm/atomic.h>
 #include <asm/debugreg.h>
@@ -70,6 +72,7 @@ asmlinkage void machine_check(void);
 asmlinkage void spurious_interrupt_bug(void);
 
 ATOMIC_NOTIFIER_HEAD(die_chain);
+EXPORT_SYMBOL(die_chain);
 
 extern char last_sysfs_file[];
 
@@ -108,15 +111,9 @@ static inline void preempt_conditional_cli(struct pt_regs *regs)
        preempt_enable_no_resched();
 }
 
-static int kstack_depth_to_print = 10;
-#ifdef CONFIG_STACK_UNWIND
-static int call_trace = 1;
-#else
-#define call_trace (-1)
-#endif
+int kstack_depth_to_print = 12;
 
 #ifdef CONFIG_KALLSYMS
-# include <linux/kallsyms.h>
 void printk_address(unsigned long address)
 {
        unsigned long offset = 0, symsize;
@@ -164,26 +161,7 @@ static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack,
         * 'stack' is in one of them:
         */
        for (k = 0; k < N_EXCEPTION_STACKS; k++) {
-               unsigned long end;
-
-               /*
-                * set 'end' to the end of the exception stack.
-                */
-               switch (k + 1) {
-               /*
-                * TODO: this block is not needed i think, because
-                * setup64.c:cpu_init() sets up t->ist[DEBUG_STACK]
-                * properly too.
-                */
-#if DEBUG_STKSZ > EXCEPTION_STKSZ
-               case DEBUG_STACK:
-                       end = cpu_pda(cpu)->debugstack + DEBUG_STKSZ;
-                       break;
-#endif
-               default:
-                       end = per_cpu(orig_ist, cpu).ist[k];
-                       break;
-               }
+               unsigned long end = per_cpu(orig_ist, cpu).ist[k];
                /*
                 * Is 'stack' above this exception frame's end?
                 * If yes then skip to the next frame.
@@ -238,24 +216,7 @@ static unsigned long *in_exception_stack(unsigned cpu, unsigned long stack,
        return NULL;
 }
 
-struct ops_and_data {
-       struct stacktrace_ops *ops;
-       void *data;
-};
-
-static int dump_trace_unwind(struct unwind_frame_info *info, void *context)
-{
-       struct ops_and_data *oad = (struct ops_and_data *)context;
-       int n = 0;
-
-       while (unwind(info) == 0 && UNW_PC(info)) {
-               if (arch_unw_user_mode(info))
-                       break;
-               n++;
-               oad->ops->address(oad->data, UNW_PC(info));
-       }
-       return n;
-}
+#define MSG(txt) ops->warning(data, txt)
 
 /*
  * x86-64 can have upto three kernel stacks: 
@@ -264,46 +225,24 @@ static int dump_trace_unwind(struct unwind_frame_info *info, void *context)
  * severe exception (double fault, nmi, stack fault, debug, mce) hardware stack
  */
 
-void dump_trace(struct task_struct *tsk, struct pt_regs *regs, unsigned long * stack,
-       struct stacktrace_ops *ops, void *data)
+static inline int valid_stack_ptr(struct thread_info *tinfo, void *p)
+{
+       void *t = (void *)tinfo;
+        return p > t && p < t + THREAD_SIZE - 3;
+}
+
+void dump_trace(struct task_struct *tsk, struct pt_regs *regs,
+               unsigned long *stack,
+               struct stacktrace_ops *ops, void *data)
 {
-       const unsigned cpu = safe_smp_processor_id();
-       unsigned long *irqstack_end = (unsigned long *)cpu_pda(cpu)->irqstackptr;
+       const unsigned cpu = get_cpu();
+       unsigned long *irqstack_end = (unsigned long*)cpu_pda(cpu)->irqstackptr;
        unsigned used = 0;
+       struct thread_info *tinfo;
 
        if (!tsk)
                tsk = current;
 
-       if (call_trace >= 0) {
-               int unw_ret = 0;
-               struct unwind_frame_info info;
-               struct ops_and_data oad = { .ops = ops, .data = data };
-
-               if (regs) {
-                       if (unwind_init_frame_info(&info, tsk, regs) == 0)
-                               unw_ret = dump_trace_unwind(&info, &oad);
-               } else if (tsk == current)
-                       unw_ret = unwind_init_running(&info, dump_trace_unwind, &oad);
-               else {
-                       if (unwind_init_blocked(&info, tsk) == 0)
-                               unw_ret = dump_trace_unwind(&info, &oad);
-               }
-               if (unw_ret > 0) {
-                       if (call_trace == 1 && !arch_unw_user_mode(&info)) {
-                               ops->warning_symbol(data, "DWARF2 unwinder stuck at %s\n",
-                                            UNW_PC(&info));
-                               if ((long)UNW_SP(&info) < 0) {
-                                       ops->warning(data, "Leftover inexact backtrace:\n");
-                                       stack = (unsigned long *)UNW_SP(&info);
-                               } else
-                                       ops->warning(data, "Full inexact backtrace again:\n");
-                       } else if (call_trace >= 1)
-                               return;
-                       else
-                               ops->warning(data, "Full inexact backtrace again:\n");
-               } else
-                       ops->warning(data, "Inexact backtrace:\n");
-       }
        if (!stack) {
                unsigned long dummy;
                stack = &dummy;
@@ -311,15 +250,17 @@ void dump_trace(struct task_struct *tsk, struct pt_regs *regs, unsigned long * s
                        stack = (unsigned long *)tsk->thread.rsp;
        }
 
-       /*
-        * Print function call entries within a stack. 'cond' is the
-        * "end of stackframe" condition, that the 'stack++'
-        * iteration will eventually trigger.
-        */
+       /*
+        * Print function call entries within a stack. 'cond' is the
+        * "end of stackframe" condition, that the 'stack++'
+        * iteration will eventually trigger.
+        */
 #define HANDLE_STACK(cond) \
        do while (cond) { \
                unsigned long addr = *stack++; \
-               if (kernel_text_address(addr)) { \
+               /* Use unlocked access here because except for NMIs     \
+                  we should be already protected against module unloads */ \
+               if (__kernel_text_address(addr)) { \
                        /* \
                         * If the address is either in the text segment of the \
                         * kernel, or in the region which contains vmalloc'ed \
@@ -328,7 +269,7 @@ void dump_trace(struct task_struct *tsk, struct pt_regs *regs, unsigned long * s
                         * down the cause of the crash will be able to figure \
                         * out the call path that was taken. \
                         */ \
-                       ops->address(data, addr); \
+                       ops->address(data, addr);   \
                } \
        } while (0)
 
@@ -345,7 +286,7 @@ void dump_trace(struct task_struct *tsk, struct pt_regs *regs, unsigned long * s
 
                if (estack_end) {
                        if (ops->stack(data, id) < 0)
-                                break;
+                               break;
                        HANDLE_STACK (stack < estack_end);
                        ops->stack(data, "<EOE>");
                        /*
@@ -382,8 +323,10 @@ void dump_trace(struct task_struct *tsk, struct pt_regs *regs, unsigned long * s
        /*
         * This handles the process stack:
         */
-       HANDLE_STACK (((long) stack & (THREAD_SIZE-1)) != 0);
+       tinfo = task_thread_info(tsk);
+       HANDLE_STACK (valid_stack_ptr(tinfo, stack));
 #undef HANDLE_STACK
+       put_cpu();
 }
 EXPORT_SYMBOL(dump_trace);
 
@@ -430,7 +373,7 @@ _show_stack(struct task_struct *tsk, struct pt_regs *regs, unsigned long *rsp)
 {
        unsigned long *stack;
        int i;
-       const int cpu = safe_smp_processor_id();
+       const int cpu = smp_processor_id();
        unsigned long *irqstack_end = (unsigned long *) (cpu_pda(cpu)->irqstackptr);
        unsigned long *irqstack = (unsigned long *) (cpu_pda(cpu)->irqstackptr - IRQSTACKSIZE);
 
@@ -457,7 +400,7 @@ _show_stack(struct task_struct *tsk, struct pt_regs *regs, unsigned long *rsp)
                }
                if (i && ((i % 4) == 0))
                        printk("\n");
-               printk(" %016lx ", *stack++);
+               printk(" %016lx", *stack++);
                touch_nmi_watchdog();
        }
        show_trace(tsk, regs, rsp);
@@ -484,7 +427,7 @@ void show_registers(struct pt_regs *regs)
        int i;
        int in_kernel = !user_mode(regs);
        unsigned long rsp;
-       const int cpu = safe_smp_processor_id(); 
+       const int cpu = smp_processor_id();
        struct task_struct *cur = cpu_pda(cpu)->pcurrent;
 
                rsp = regs->rsp;
@@ -521,36 +464,22 @@ bad:
        printk("\n");
 }      
 
-void handle_BUG(struct pt_regs *regs)
-{ 
-       struct bug_frame f;
-       long len;
-       const char *prefix = "";
+int is_valid_bugaddr(unsigned long rip)
+{
+       unsigned short ud2;
 
-       if (user_mode(regs))
-               return; 
-       if (__copy_from_user(&f, (const void __user *) regs->rip,
-                            sizeof(struct bug_frame)))
-               return; 
-       if (f.filename >= 0 ||
-           f.ud2[0] != 0x0f || f.ud2[1] != 0x0b) 
-               return;
-       len = __strnlen_user((char *)(long)f.filename, PATH_MAX) - 1;
-       if (len < 0 || len >= PATH_MAX)
-               f.filename = (int)(long)"unmapped filename";
-       else if (len > 50) {
-               f.filename += len - 50;
-               prefix = "...";
-       }
-       printk("----------- [cut here ] --------- [please bite here ] ---------\n");
-       printk(KERN_ALERT "Kernel BUG at %s%.50s:%d\n", prefix, (char *)(long)f.filename, f.line);
-} 
+       if (__copy_from_user(&ud2, (const void __user *) rip, sizeof(ud2)))
+               return 0;
+
+       return ud2 == 0x0b0f;
+}
 
 #ifdef CONFIG_BUG
 void out_of_line_bug(void)
 { 
        BUG(); 
 } 
+EXPORT_SYMBOL(out_of_line_bug);
 #endif
 
 static DEFINE_SPINLOCK(die_lock);
@@ -559,9 +488,11 @@ static unsigned int die_nest_count;
 
 unsigned __kprobes long oops_begin(void)
 {
-       int cpu = safe_smp_processor_id();
+       int cpu = smp_processor_id();
        unsigned long flags;
 
+       oops_enter();
+
        /* racy, but better than risking deadlock. */
        local_irq_save(flags);
        if (!spin_trylock(&die_lock)) { 
@@ -590,6 +521,7 @@ void __kprobes oops_end(unsigned long flags)
                spin_unlock_irqrestore(&die_lock, flags);
        if (panic_on_oops)
                panic("Fatal exception");
+       oops_exit();
 }
 
 void __kprobes __die(const char * str, struct pt_regs * regs, long err)
@@ -623,14 +555,15 @@ void die(const char * str, struct pt_regs * regs, long err)
 {
        unsigned long flags = oops_begin();
 
-       handle_BUG(regs);
+       if (!user_mode(regs))
+               report_bug(regs->rip);
+
        __die(str, regs, err);
        oops_end(flags);
        do_exit(SIGSEGV); 
 }
 
-#ifdef CONFIG_X86_LOCAL_APIC
-void __kprobes die_nmi(char *str, struct pt_regs *regs)
+void __kprobes die_nmi(char *str, struct pt_regs *regs, int do_panic)
 {
        unsigned long flags = oops_begin();
 
@@ -638,19 +571,17 @@ void __kprobes die_nmi(char *str, struct pt_regs *regs)
         * We are in trouble anyway, lets at least try
         * to get a message out.
         */
-       printk(str, safe_smp_processor_id());
+       printk(str, smp_processor_id());
        show_registers(regs);
        if (kexec_should_crash(current))
                crash_kexec(regs);
-       if (panic_on_timeout || panic_on_oops)
-               panic("nmi watchdog");
-       printk("console shuts up ...\n");
+       if (do_panic || panic_on_oops)
+               panic("Non maskable interrupt");
        oops_end(flags);
        nmi_exit();
        local_irq_enable();
        do_exit(SIGSEGV);
 }
-#endif
 
 static void __kprobes do_trap(int trapnr, int signr, char *str,
                              struct pt_regs * regs, long error_code,
@@ -664,8 +595,8 @@ static void __kprobes do_trap(int trapnr, int signr, char *str,
        if (user_mode(regs)) {
                if (exception_trace && unhandled_signal(tsk, signr))
                        printk(KERN_INFO
-                              "%s[%d] trap %s rip:%lx rsp:%lx error:%lx\n",
-                              tsk->comm, tsk->pid, str,
+                              "%s[%d:#%u] trap %s rip:%lx rsp:%lx error:%lx\n",
+                              tsk->comm, tsk->pid, tsk->xid, str,
                               regs->rip, regs->rsp, error_code); 
 
                if (info)
@@ -765,8 +696,8 @@ asmlinkage void __kprobes do_general_protection(struct pt_regs * regs,
        if (user_mode(regs)) {
                if (exception_trace && unhandled_signal(tsk, SIGSEGV))
                        printk(KERN_INFO
-                      "%s[%d] general protection rip:%lx rsp:%lx error:%lx\n",
-                              tsk->comm, tsk->pid,
+                      "%s[%d:#%u] general protection rip:%lx rsp:%lx error:%lx\n",
+                              tsk->comm, tsk->pid, tsk->xid,
                               regs->rip, regs->rsp, error_code); 
 
                force_sig(SIGSEGV, tsk);
@@ -791,8 +722,14 @@ asmlinkage void __kprobes do_general_protection(struct pt_regs * regs,
 static __kprobes void
 mem_parity_error(unsigned char reason, struct pt_regs * regs)
 {
-       printk("Uhhuh. NMI received. Dazed and confused, but trying to continue\n");
-       printk("You probably have a hardware problem with your RAM chips\n");
+       printk(KERN_EMERG "Uhhuh. NMI received for unknown reason %02x.\n",
+               reason);
+       printk(KERN_EMERG "You have some hardware problem, likely on the PCI bus.\n");
+
+       if (panic_on_unrecovered_nmi)
+               panic("NMI: Not continuing");
+
+       printk(KERN_EMERG "Dazed and confused, but trying to continue\n");
 
 #if 0 /* XEN */
        /* Clear and disable the memory parity error line. */
@@ -819,9 +756,15 @@ io_check_error(unsigned char reason, struct pt_regs * regs)
 
 static __kprobes void
 unknown_nmi_error(unsigned char reason, struct pt_regs * regs)
-{      printk("Uhhuh. NMI received for unknown reason %02x.\n", reason);
-       printk("Dazed and confused, but trying to continue\n");
-       printk("Do you have a strange power saving mode enabled?\n");
+{
+       printk(KERN_EMERG "Uhhuh. NMI received for unknown reason %02x.\n",
+               reason);
+       printk(KERN_EMERG "Do you have a strange power saving mode enabled?\n");
+
+       if (panic_on_unrecovered_nmi)
+               panic("NMI: Not continuing");
+
+       printk(KERN_EMERG "Dazed and confused, but trying to continue\n");
 }
 
 /* Runs on IST stack. This code must keep interrupts off all the time.
@@ -841,17 +784,15 @@ asmlinkage __kprobes void default_do_nmi(struct pt_regs *regs)
                if (notify_die(DIE_NMI_IPI, "nmi_ipi", regs, reason, 2, SIGINT)
                                                                == NOTIFY_STOP)
                        return;
-#ifdef CONFIG_X86_LOCAL_APIC
                /*
                 * Ok, so this is none of the documented NMI sources,
                 * so it must be the NMI watchdog.
                 */
-               if (nmi_watchdog > 0) {
-                       nmi_watchdog_tick(regs,reason);
+               if (nmi_watchdog_tick(regs,reason))
                        return;
-               }
-#endif
-               unknown_nmi_error(reason, regs);
+               if (!do_nmi_callback(regs,cpu))
+                       unknown_nmi_error(reason, regs);
+
                return;
        }
        if (notify_die(DIE_NMI, "nmi", regs, reason, 2, SIGINT) == NOTIFY_STOP)
@@ -935,14 +876,6 @@ asmlinkage void __kprobes do_debug(struct pt_regs * regs,
                 */
                 if (!user_mode(regs))
                        goto clear_TF_reenable;
-               /*
-                * Was the TF flag set by a debugger? If so, clear it now,
-                * so that register information is correct.
-                */
-               if (tsk->ptrace & PT_DTRACE) {
-                       regs->eflags &= ~TF_MASK;
-                       tsk->ptrace &= ~PT_DTRACE;
-               }
        }
 
        /* Ok, finally something we can handle */
@@ -1138,6 +1071,7 @@ asmlinkage void math_state_restore(void)
                init_fpu(me);
        restore_fpu_checking(&me->thread.i387.fxsave);
        task_thread_info(me)->status |= TS_USEDFPU;
+       me->fpu_counter++;
 }
 
 
@@ -1200,33 +1134,21 @@ void smp_trap_init(trap_info_t *trap_ctxt)
 }
 
 
-/* Actual parsing is done early in setup.c. */
-static int __init oops_dummy(char *s)
+static int __init oops_setup(char *s)
 { 
-       panic_on_oops = 1;
-       return 1;
+       if (!s)
+               return -EINVAL;
+       if (!strcmp(s, "panic"))
+               panic_on_oops = 1;
+       return 0;
 } 
-__setup("oops=", oops_dummy); 
+early_param("oops", oops_setup);
 
 static int __init kstack_setup(char *s)
 {
+       if (!s)
+               return -EINVAL;
        kstack_depth_to_print = simple_strtoul(s,NULL,0);
-       return 1;
-}
-__setup("kstack=", kstack_setup);
-
-#ifdef CONFIG_STACK_UNWIND
-static int __init call_trace_setup(char *s)
-{
-       if (strcmp(s, "old") == 0)
-               call_trace = -1;
-       else if (strcmp(s, "both") == 0)
-               call_trace = 0;
-       else if (strcmp(s, "newfallback") == 0)
-               call_trace = 1;
-       else if (strcmp(s, "new") == 0)
-               call_trace = 2;
-       return 1;
+       return 0;
 }
-__setup("call_trace=", call_trace_setup);
-#endif
+early_param("kstack", kstack_setup);