Commit 28a0ce7f authored Mar 04, 2009 by Johannes Weiner Committed by Chris Zankel Apr 02, 2009

xtensa: use correct stack pointer for stack traces



Right now, the xtensa stacktrace code reads the _current_ kernel stack
pointer if nothing is supplied.  With debugging facilities like sysrq
this means that the backtrace of the sysrq-handler is printed instead
of a trace of the given task's stack.

When no stack pointer is specified in show_trace() and show_stack(),
use the stack pointer that comes with the handed in task descriptor to
make stack traces more useful.

Signed-off-by: Johannes Weiner <jw@emlix.com>
Signed-off-by: Chris Zankel <chris@zankel.net>

parent 35f9cd08

arch/xtensa/kernel/traps.c

+6 −8

Original line number	Diff line number	Diff line
		@@ -372,11 +372,10 @@ void show_trace(struct task_struct task, unsigned long sp)
		unsigned long a0, a1, pc;
		unsigned long sp_start, sp_end;

		if (sp)
		a1 = (unsigned long)sp;

		if (a1 == 0)
		__asm__ __volatile__ ("mov %0, a1\n" : "=a"(a1));

		else
		a1 = task->thread.sp;

		sp_start = a1 & ~(THREAD_SIZE-1);
		sp_end = sp_start + THREAD_SIZE;
		@@ -418,9 +417,8 @@ void show_stack(struct task_struct task, unsigned long sp)
		int i = 0;
		unsigned long *stack;

		if (sp == 0)
		__asm__ __volatile__ ("mov %0, a1\n" : "=a"(sp));

		if (!sp)
		sp = (unsigned long *)task->thread.sp;
		stack = sp;

		printk("\nStack: ");