sched: use show_regs() to improve __schedule_bug() output
authorSatyam Sharma <satyam@infradead.org>
Wed, 24 Oct 2007 16:23:50 +0000 (18:23 +0200)
committerIngo Molnar <mingo@elte.hu>
Wed, 24 Oct 2007 16:23:50 +0000 (18:23 +0200)
A full register dump along with stack backtrace would make the
"scheduling while atomic" message more helpful. Use show_regs() instead
of dump_stack() for this. We already know we're atomic in here (that is
why this function was called) so show_regs()'s atomicity expectations
are guaranteed.

Also, modify the output of the "BUG: scheduling while atomic:" header a
bit to keep task->comm and task->pid together and preempt_count() after
them.

Signed-off-by: Satyam Sharma <satyam@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
kernel/sched.c

index af02a4d..d1e6663 100644 (file)
@@ -66,6 +66,7 @@
 #include <linux/pagemap.h>
 
 #include <asm/tlb.h>
+#include <asm/irq_regs.h>
 
 /*
  * Scheduler clock - returns current time in nanosec units.
@@ -3507,12 +3508,19 @@ EXPORT_SYMBOL(sub_preempt_count);
  */
 static noinline void __schedule_bug(struct task_struct *prev)
 {
-       printk(KERN_ERR "BUG: scheduling while atomic: %s/0x%08x/%d\n",
-               prev->comm, preempt_count(), task_pid_nr(prev));
+       struct pt_regs *regs = get_irq_regs();
+
+       printk(KERN_ERR "BUG: scheduling while atomic: %s/%d/0x%08x\n",
+               prev->comm, prev->pid, preempt_count());
+
        debug_show_held_locks(prev);
        if (irqs_disabled())
                print_irqtrace_events(prev);
-       dump_stack();
+
+       if (regs)
+               show_regs(regs);
+       else
+               dump_stack();
 }
 
 /*