From: Sage Weil <sage@redhat.com>
Date: Tue, 10 Jan 2017 03:52:02 +0000 (-0600)
Subject: global/signal_handler: behave if g_ceph_context or _log are null
X-Git-Tag: v12.0.0~170^2~1
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=d1cc9e4836b7d360ccb3dcd129c4799508e41e51;p=ceph.git

global/signal_handler: behave if g_ceph_context or _log are null

This can happen if we trigger an assertion while g_ceph_context is being
torn down.

Signed-off-by: Sage Weil <sage@redhat.com>
---

diff --git a/src/global/signal_handler.cc b/src/global/signal_handler.cc
index 7d5069828c14..e8f21c1610a1 100644
--- a/src/global/signal_handler.cc
+++ b/src/global/signal_handler.cc
@@ -113,7 +113,9 @@ static void handle_fatal_signal(int signum)
 
   // avoid recursion back into logging code if that is where
   // we got the SEGV.
-  if (!g_ceph_context->_log->is_inside_log_lock()) {
+  if (g_ceph_context &&
+      g_ceph_context->_log &&
+      !g_ceph_context->_log->is_inside_log_lock()) {
     // TODO: don't use an ostringstream here. It could call malloc(), which we
     // don't want inside a signal handler.
     // Also fix the backtrace code not to allocate memory.