With lockdep we got the following trace after a panic:
Badness at /home/autobuild/BUILD/linux-2.6.28-
20090204/kernel/lockdep.c:2878
[...]
Call Trace:
[<
0000000000176334>] lock_acquire+0x54/0xbc
[<
000000000050b4fe>] __atomic_notifier_call_chain+0x6e/0xdc
[<
000000000050b59c>] atomic_notifier_call_chain+0x30/0x44
[<
0000000000504274>] panic+0xd0/0x1e8
[...]
INFO: lockdep is turned off.
Last Breaking-Event-Address:
[<
0000000000170e62>] check_flags+0xae/0x15c
possible reason: unannotated irqs-off.
lockdep is right. We missed a trace_hardirq_off in our smp_send_stop
function and smp_send_stop is called before the panic call chain.
Reported-by: Mijo <Safradin mijo@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
#include <linux/delay.h>
#include <linux/cache.h>
#include <linux/interrupt.h>
+#include <linux/irqflags.h>
#include <linux/cpu.h>
#include <linux/timex.h>
#include <linux/bootmem.h>
/* Disable all interrupts/machine checks */
__load_psw_mask(psw_kernel_bits & ~PSW_MASK_MCHECK);
+ trace_hardirqs_off();
/* stop all processors */
for_each_online_cpu(cpu) {