powerpc,kgdb: Introduce low level trap catching

The only way the debugger can handle a trap in inside rcu_lock,
notify_die, or atomic_notifier_call_chain without a recursive fault is
to allow the kernel debugger to handle the exception first in
program_check_exception().

The other change here is to make sure that kgdb_handle_exception() is
called with correct parameters when catching an oops, because kdb
needs to know if the entry was an oops, single step, or breakpoint
exception.

[benh@kernel.crashing.org: move debugger_bpt instead of #ifdef]

CC: Paul Mackerras <paulus@samba.org>
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

1 files changed, 5 insertions, 2 deletions

diff --git a/arch/powerpc/kernel/traps.c b/arch/powerpc/kernel/traps.c
index 29d128eb6c43..b6859aade9c2 100644
--- a/arch/powerpc/kernel/traps.c
+++ b/arch/powerpc/kernel/traps.c
@@ -815,12 +815,15 @@ void __kprobes program_check_exception(struct pt_regs *regs)
                 return;
         }
         if (reason & REASON_TRAP) {
+                /* Debugger is first in line to stop recursive faults in
+                 * rcu_lock, notify_die, or atomic_notifier_call_chain */
+                if (debugger_bpt(regs))
+                        return;
+
                 /* trap exception */
                 if (notify_die(DIE_BPT, "breakpoint", regs, 5, 5, SIGTRAP)
                                 == NOTIFY_STOP)
                         return;
-                if (debugger_bpt(regs))
-                        return;
 
                 if (!(regs->msr & MSR_PR) &&  /* not user-mode */
                     report_bug(regs->nip, regs) == BUG_TRAP_TYPE_WARN) {